diff --git "a/trainer_state.json" "b/trainer_state.json"
deleted file mode 100644--- "a/trainer_state.json"
+++ /dev/null
@@ -1,350101 +0,0 @@
-{
-  "best_metric": null,
-  "best_model_checkpoint": null,
-  "epoch": 26.07561929595828,
-  "eval_steps": 5000,
-  "global_step": 50000,
-  "is_hyper_param_search": false,
-  "is_local_process_zero": true,
-  "is_world_process_zero": true,
-  "log_history": [
-    {
-      "epoch": 0.0005215123859191656,
-      "grad_norm": Infinity,
-      "learning_rate": 0.0,
-      "loss": 9.3221,
-      "step": 1
-    },
-    {
-      "epoch": 0.0010430247718383311,
-      "grad_norm": 18.024211883544922,
-      "learning_rate": 2.0000000000000002e-07,
-      "loss": 9.0362,
-      "step": 2
-    },
-    {
-      "epoch": 0.0015645371577574967,
-      "grad_norm": 18.082225799560547,
-      "learning_rate": 4.0000000000000003e-07,
-      "loss": 9.3207,
-      "step": 3
-    },
-    {
-      "epoch": 0.0020860495436766623,
-      "grad_norm": 17.8996639251709,
-      "learning_rate": 6.000000000000001e-07,
-      "loss": 9.3106,
-      "step": 4
-    },
-    {
-      "epoch": 0.002607561929595828,
-      "grad_norm": 18.508499145507812,
-      "learning_rate": 8.000000000000001e-07,
-      "loss": 9.3007,
-      "step": 5
-    },
-    {
-      "epoch": 0.0031290743155149934,
-      "grad_norm": 19.56254768371582,
-      "learning_rate": 1.0000000000000002e-06,
-      "loss": 9.3288,
-      "step": 6
-    },
-    {
-      "epoch": 0.003650586701434159,
-      "grad_norm": 17.659278869628906,
-      "learning_rate": 1.2000000000000002e-06,
-      "loss": 9.2988,
-      "step": 7
-    },
-    {
-      "epoch": 0.0041720990873533245,
-      "grad_norm": 15.949824333190918,
-      "learning_rate": 1.4000000000000001e-06,
-      "loss": 9.3166,
-      "step": 8
-    },
-    {
-      "epoch": 0.00469361147327249,
-      "grad_norm": 14.140177726745605,
-      "learning_rate": 1.6000000000000001e-06,
-      "loss": 9.0064,
-      "step": 9
-    },
-    {
-      "epoch": 0.005215123859191656,
-      "grad_norm": 13.076690673828125,
-      "learning_rate": 1.8e-06,
-      "loss": 9.0732,
-      "step": 10
-    },
-    {
-      "epoch": 0.005736636245110821,
-      "grad_norm": 10.833678245544434,
-      "learning_rate": 2.0000000000000003e-06,
-      "loss": 9.1203,
-      "step": 11
-    },
-    {
-      "epoch": 0.006258148631029987,
-      "grad_norm": 9.970856666564941,
-      "learning_rate": 2.2e-06,
-      "loss": 8.9484,
-      "step": 12
-    },
-    {
-      "epoch": 0.006779661016949152,
-      "grad_norm": 9.647157669067383,
-      "learning_rate": 2.4000000000000003e-06,
-      "loss": 9.1161,
-      "step": 13
-    },
-    {
-      "epoch": 0.007301173402868318,
-      "grad_norm": 8.991828918457031,
-      "learning_rate": 2.6e-06,
-      "loss": 9.0114,
-      "step": 14
-    },
-    {
-      "epoch": 0.007822685788787484,
-      "grad_norm": 7.713730335235596,
-      "learning_rate": 2.8000000000000003e-06,
-      "loss": 8.9205,
-      "step": 15
-    },
-    {
-      "epoch": 0.008344198174706649,
-      "grad_norm": 7.970811367034912,
-      "learning_rate": 3e-06,
-      "loss": 8.9524,
-      "step": 16
-    },
-    {
-      "epoch": 0.008865710560625815,
-      "grad_norm": 7.163078784942627,
-      "learning_rate": 3.2000000000000003e-06,
-      "loss": 8.8868,
-      "step": 17
-    },
-    {
-      "epoch": 0.00938722294654498,
-      "grad_norm": 5.800992965698242,
-      "learning_rate": 3.4000000000000005e-06,
-      "loss": 8.8257,
-      "step": 18
-    },
-    {
-      "epoch": 0.009908735332464147,
-      "grad_norm": 5.590092658996582,
-      "learning_rate": 3.6e-06,
-      "loss": 8.821,
-      "step": 19
-    },
-    {
-      "epoch": 0.010430247718383311,
-      "grad_norm": 5.476081371307373,
-      "learning_rate": 3.8e-06,
-      "loss": 8.7248,
-      "step": 20
-    },
-    {
-      "epoch": 0.010951760104302478,
-      "grad_norm": 4.819453239440918,
-      "learning_rate": 4.000000000000001e-06,
-      "loss": 8.7742,
-      "step": 21
-    },
-    {
-      "epoch": 0.011473272490221642,
-      "grad_norm": 4.722670078277588,
-      "learning_rate": 4.2000000000000004e-06,
-      "loss": 8.7207,
-      "step": 22
-    },
-    {
-      "epoch": 0.011994784876140809,
-      "grad_norm": 4.683105945587158,
-      "learning_rate": 4.4e-06,
-      "loss": 8.6636,
-      "step": 23
-    },
-    {
-      "epoch": 0.012516297262059974,
-      "grad_norm": 4.900444507598877,
-      "learning_rate": 4.6e-06,
-      "loss": 8.6013,
-      "step": 24
-    },
-    {
-      "epoch": 0.01303780964797914,
-      "grad_norm": 4.241089820861816,
-      "learning_rate": 4.800000000000001e-06,
-      "loss": 8.5963,
-      "step": 25
-    },
-    {
-      "epoch": 0.013559322033898305,
-      "grad_norm": 4.1339874267578125,
-      "learning_rate": 5e-06,
-      "loss": 8.627,
-      "step": 26
-    },
-    {
-      "epoch": 0.014080834419817471,
-      "grad_norm": 4.197762489318848,
-      "learning_rate": 5.2e-06,
-      "loss": 8.5626,
-      "step": 27
-    },
-    {
-      "epoch": 0.014602346805736636,
-      "grad_norm": 4.505918979644775,
-      "learning_rate": 5.4e-06,
-      "loss": 8.4761,
-      "step": 28
-    },
-    {
-      "epoch": 0.015123859191655802,
-      "grad_norm": 4.118350982666016,
-      "learning_rate": 5.600000000000001e-06,
-      "loss": 8.4802,
-      "step": 29
-    },
-    {
-      "epoch": 0.01564537157757497,
-      "grad_norm": 4.4556660652160645,
-      "learning_rate": 5.8e-06,
-      "loss": 8.5576,
-      "step": 30
-    },
-    {
-      "epoch": 0.01616688396349413,
-      "grad_norm": 3.6120073795318604,
-      "learning_rate": 6e-06,
-      "loss": 8.4474,
-      "step": 31
-    },
-    {
-      "epoch": 0.016688396349413298,
-      "grad_norm": 5.583066940307617,
-      "learning_rate": 6.2e-06,
-      "loss": 7.4272,
-      "step": 32
-    },
-    {
-      "epoch": 0.017209908735332465,
-      "grad_norm": 5.695413589477539,
-      "learning_rate": 6.4000000000000006e-06,
-      "loss": 8.3017,
-      "step": 33
-    },
-    {
-      "epoch": 0.01773142112125163,
-      "grad_norm": 4.050053119659424,
-      "learning_rate": 6.6e-06,
-      "loss": 8.4141,
-      "step": 34
-    },
-    {
-      "epoch": 0.018252933507170794,
-      "grad_norm": 3.966257333755493,
-      "learning_rate": 6.800000000000001e-06,
-      "loss": 8.4491,
-      "step": 35
-    },
-    {
-      "epoch": 0.01877444589308996,
-      "grad_norm": 3.792633056640625,
-      "learning_rate": 7.000000000000001e-06,
-      "loss": 8.3827,
-      "step": 36
-    },
-    {
-      "epoch": 0.019295958279009127,
-      "grad_norm": 4.522724628448486,
-      "learning_rate": 7.2e-06,
-      "loss": 8.3003,
-      "step": 37
-    },
-    {
-      "epoch": 0.019817470664928293,
-      "grad_norm": 3.689426898956299,
-      "learning_rate": 7.4e-06,
-      "loss": 8.3769,
-      "step": 38
-    },
-    {
-      "epoch": 0.020338983050847456,
-      "grad_norm": 3.616961717605591,
-      "learning_rate": 7.6e-06,
-      "loss": 8.3962,
-      "step": 39
-    },
-    {
-      "epoch": 0.020860495436766623,
-      "grad_norm": 3.489570379257202,
-      "learning_rate": 7.8e-06,
-      "loss": 8.3578,
-      "step": 40
-    },
-    {
-      "epoch": 0.02138200782268579,
-      "grad_norm": 3.925410509109497,
-      "learning_rate": 8.000000000000001e-06,
-      "loss": 8.3282,
-      "step": 41
-    },
-    {
-      "epoch": 0.021903520208604955,
-      "grad_norm": 3.7634716033935547,
-      "learning_rate": 8.200000000000001e-06,
-      "loss": 8.2869,
-      "step": 42
-    },
-    {
-      "epoch": 0.02242503259452412,
-      "grad_norm": 3.7333362102508545,
-      "learning_rate": 8.400000000000001e-06,
-      "loss": 8.2405,
-      "step": 43
-    },
-    {
-      "epoch": 0.022946544980443285,
-      "grad_norm": 3.0992345809936523,
-      "learning_rate": 8.599999999999999e-06,
-      "loss": 8.2989,
-      "step": 44
-    },
-    {
-      "epoch": 0.02346805736636245,
-      "grad_norm": 6.670464515686035,
-      "learning_rate": 8.8e-06,
-      "loss": 8.003,
-      "step": 45
-    },
-    {
-      "epoch": 0.023989569752281618,
-      "grad_norm": 5.897899627685547,
-      "learning_rate": 9e-06,
-      "loss": 7.8498,
-      "step": 46
-    },
-    {
-      "epoch": 0.02451108213820078,
-      "grad_norm": 4.171645641326904,
-      "learning_rate": 9.2e-06,
-      "loss": 8.2926,
-      "step": 47
-    },
-    {
-      "epoch": 0.025032594524119947,
-      "grad_norm": 4.116384029388428,
-      "learning_rate": 9.4e-06,
-      "loss": 8.2173,
-      "step": 48
-    },
-    {
-      "epoch": 0.025554106910039114,
-      "grad_norm": 4.061080455780029,
-      "learning_rate": 9.600000000000001e-06,
-      "loss": 8.1813,
-      "step": 49
-    },
-    {
-      "epoch": 0.02607561929595828,
-      "grad_norm": 3.8002374172210693,
-      "learning_rate": 9.800000000000001e-06,
-      "loss": 8.162,
-      "step": 50
-    },
-    {
-      "epoch": 0.026597131681877443,
-      "grad_norm": 3.3633811473846436,
-      "learning_rate": 1e-05,
-      "loss": 8.1976,
-      "step": 51
-    },
-    {
-      "epoch": 0.02711864406779661,
-      "grad_norm": 3.8181605339050293,
-      "learning_rate": 1.02e-05,
-      "loss": 8.1911,
-      "step": 52
-    },
-    {
-      "epoch": 0.027640156453715776,
-      "grad_norm": 3.7845466136932373,
-      "learning_rate": 1.04e-05,
-      "loss": 8.0857,
-      "step": 53
-    },
-    {
-      "epoch": 0.028161668839634942,
-      "grad_norm": 3.9014339447021484,
-      "learning_rate": 1.06e-05,
-      "loss": 8.0618,
-      "step": 54
-    },
-    {
-      "epoch": 0.028683181225554105,
-      "grad_norm": 4.385609149932861,
-      "learning_rate": 1.08e-05,
-      "loss": 8.0047,
-      "step": 55
-    },
-    {
-      "epoch": 0.02920469361147327,
-      "grad_norm": 4.386277198791504,
-      "learning_rate": 1.1000000000000001e-05,
-      "loss": 7.9167,
-      "step": 56
-    },
-    {
-      "epoch": 0.029726205997392438,
-      "grad_norm": 5.018571853637695,
-      "learning_rate": 1.1200000000000001e-05,
-      "loss": 7.8501,
-      "step": 57
-    },
-    {
-      "epoch": 0.030247718383311605,
-      "grad_norm": 4.002060413360596,
-      "learning_rate": 1.1400000000000001e-05,
-      "loss": 8.0016,
-      "step": 58
-    },
-    {
-      "epoch": 0.03076923076923077,
-      "grad_norm": 3.611271381378174,
-      "learning_rate": 1.16e-05,
-      "loss": 8.0786,
-      "step": 59
-    },
-    {
-      "epoch": 0.03129074315514994,
-      "grad_norm": 4.341294288635254,
-      "learning_rate": 1.18e-05,
-      "loss": 7.9634,
-      "step": 60
-    },
-    {
-      "epoch": 0.031812255541069104,
-      "grad_norm": 4.467768669128418,
-      "learning_rate": 1.2e-05,
-      "loss": 7.9299,
-      "step": 61
-    },
-    {
-      "epoch": 0.03233376792698826,
-      "grad_norm": 3.67276668548584,
-      "learning_rate": 1.22e-05,
-      "loss": 7.969,
-      "step": 62
-    },
-    {
-      "epoch": 0.03285528031290743,
-      "grad_norm": 4.076155662536621,
-      "learning_rate": 1.24e-05,
-      "loss": 7.9558,
-      "step": 63
-    },
-    {
-      "epoch": 0.033376792698826596,
-      "grad_norm": 4.484344005584717,
-      "learning_rate": 1.2600000000000001e-05,
-      "loss": 7.9411,
-      "step": 64
-    },
-    {
-      "epoch": 0.03389830508474576,
-      "grad_norm": 4.032654285430908,
-      "learning_rate": 1.2800000000000001e-05,
-      "loss": 7.892,
-      "step": 65
-    },
-    {
-      "epoch": 0.03441981747066493,
-      "grad_norm": 4.794155120849609,
-      "learning_rate": 1.3000000000000001e-05,
-      "loss": 7.6581,
-      "step": 66
-    },
-    {
-      "epoch": 0.034941329856584095,
-      "grad_norm": 3.885824680328369,
-      "learning_rate": 1.32e-05,
-      "loss": 7.8953,
-      "step": 67
-    },
-    {
-      "epoch": 0.03546284224250326,
-      "grad_norm": 4.162301063537598,
-      "learning_rate": 1.3400000000000002e-05,
-      "loss": 7.797,
-      "step": 68
-    },
-    {
-      "epoch": 0.03598435462842243,
-      "grad_norm": 4.53833532333374,
-      "learning_rate": 1.3600000000000002e-05,
-      "loss": 7.8279,
-      "step": 69
-    },
-    {
-      "epoch": 0.03650586701434159,
-      "grad_norm": 3.926084041595459,
-      "learning_rate": 1.3800000000000002e-05,
-      "loss": 7.9131,
-      "step": 70
-    },
-    {
-      "epoch": 0.037027379400260754,
-      "grad_norm": 3.81913161277771,
-      "learning_rate": 1.4000000000000001e-05,
-      "loss": 7.8192,
-      "step": 71
-    },
-    {
-      "epoch": 0.03754889178617992,
-      "grad_norm": 3.9308505058288574,
-      "learning_rate": 1.42e-05,
-      "loss": 7.8109,
-      "step": 72
-    },
-    {
-      "epoch": 0.03807040417209909,
-      "grad_norm": 4.004209041595459,
-      "learning_rate": 1.44e-05,
-      "loss": 7.7685,
-      "step": 73
-    },
-    {
-      "epoch": 0.038591916558018254,
-      "grad_norm": 5.752688884735107,
-      "learning_rate": 1.4599999999999999e-05,
-      "loss": 7.0768,
-      "step": 74
-    },
-    {
-      "epoch": 0.03911342894393742,
-      "grad_norm": 5.2515788078308105,
-      "learning_rate": 1.48e-05,
-      "loss": 7.4493,
-      "step": 75
-    },
-    {
-      "epoch": 0.039634941329856586,
-      "grad_norm": 3.6931424140930176,
-      "learning_rate": 1.5e-05,
-      "loss": 7.7851,
-      "step": 76
-    },
-    {
-      "epoch": 0.04015645371577575,
-      "grad_norm": 4.673013687133789,
-      "learning_rate": 1.52e-05,
-      "loss": 7.2595,
-      "step": 77
-    },
-    {
-      "epoch": 0.04067796610169491,
-      "grad_norm": 3.719449043273926,
-      "learning_rate": 1.54e-05,
-      "loss": 7.7229,
-      "step": 78
-    },
-    {
-      "epoch": 0.04119947848761408,
-      "grad_norm": 4.387474060058594,
-      "learning_rate": 1.56e-05,
-      "loss": 7.5895,
-      "step": 79
-    },
-    {
-      "epoch": 0.041720990873533245,
-      "grad_norm": 3.777567148208618,
-      "learning_rate": 1.58e-05,
-      "loss": 7.6765,
-      "step": 80
-    },
-    {
-      "epoch": 0.04224250325945241,
-      "grad_norm": 4.529924392700195,
-      "learning_rate": 1.6000000000000003e-05,
-      "loss": 7.4633,
-      "step": 81
-    },
-    {
-      "epoch": 0.04276401564537158,
-      "grad_norm": 4.4846272468566895,
-      "learning_rate": 1.62e-05,
-      "loss": 7.596,
-      "step": 82
-    },
-    {
-      "epoch": 0.043285528031290744,
-      "grad_norm": 4.144165992736816,
-      "learning_rate": 1.6400000000000002e-05,
-      "loss": 7.5337,
-      "step": 83
-    },
-    {
-      "epoch": 0.04380704041720991,
-      "grad_norm": 4.42596435546875,
-      "learning_rate": 1.66e-05,
-      "loss": 7.4749,
-      "step": 84
-    },
-    {
-      "epoch": 0.04432855280312908,
-      "grad_norm": 4.529561519622803,
-      "learning_rate": 1.6800000000000002e-05,
-      "loss": 7.7078,
-      "step": 85
-    },
-    {
-      "epoch": 0.04485006518904824,
-      "grad_norm": 4.734699249267578,
-      "learning_rate": 1.7000000000000003e-05,
-      "loss": 7.4827,
-      "step": 86
-    },
-    {
-      "epoch": 0.0453715775749674,
-      "grad_norm": 3.5461740493774414,
-      "learning_rate": 1.7199999999999998e-05,
-      "loss": 7.6987,
-      "step": 87
-    },
-    {
-      "epoch": 0.04589308996088657,
-      "grad_norm": 4.5075812339782715,
-      "learning_rate": 1.74e-05,
-      "loss": 7.6847,
-      "step": 88
-    },
-    {
-      "epoch": 0.046414602346805736,
-      "grad_norm": 3.5770719051361084,
-      "learning_rate": 1.76e-05,
-      "loss": 7.6693,
-      "step": 89
-    },
-    {
-      "epoch": 0.0469361147327249,
-      "grad_norm": 3.6447737216949463,
-      "learning_rate": 1.78e-05,
-      "loss": 7.6698,
-      "step": 90
-    },
-    {
-      "epoch": 0.04745762711864407,
-      "grad_norm": 4.477330684661865,
-      "learning_rate": 1.8e-05,
-      "loss": 7.2415,
-      "step": 91
-    },
-    {
-      "epoch": 0.047979139504563235,
-      "grad_norm": 4.59464693069458,
-      "learning_rate": 1.8200000000000002e-05,
-      "loss": 7.5612,
-      "step": 92
-    },
-    {
-      "epoch": 0.0485006518904824,
-      "grad_norm": 4.561006546020508,
-      "learning_rate": 1.84e-05,
-      "loss": 7.3928,
-      "step": 93
-    },
-    {
-      "epoch": 0.04902216427640156,
-      "grad_norm": 3.7959651947021484,
-      "learning_rate": 1.86e-05,
-      "loss": 7.6621,
-      "step": 94
-    },
-    {
-      "epoch": 0.04954367666232073,
-      "grad_norm": 3.2018020153045654,
-      "learning_rate": 1.88e-05,
-      "loss": 7.461,
-      "step": 95
-    },
-    {
-      "epoch": 0.050065189048239894,
-      "grad_norm": 3.7467145919799805,
-      "learning_rate": 1.9e-05,
-      "loss": 7.5199,
-      "step": 96
-    },
-    {
-      "epoch": 0.05058670143415906,
-      "grad_norm": 3.6669843196868896,
-      "learning_rate": 1.9200000000000003e-05,
-      "loss": 7.4672,
-      "step": 97
-    },
-    {
-      "epoch": 0.05110821382007823,
-      "grad_norm": 3.8501787185668945,
-      "learning_rate": 1.94e-05,
-      "loss": 7.6553,
-      "step": 98
-    },
-    {
-      "epoch": 0.051629726205997394,
-      "grad_norm": 3.7197256088256836,
-      "learning_rate": 1.9600000000000002e-05,
-      "loss": 7.503,
-      "step": 99
-    },
-    {
-      "epoch": 0.05215123859191656,
-      "grad_norm": 3.496748924255371,
-      "learning_rate": 1.9800000000000004e-05,
-      "loss": 7.5369,
-      "step": 100
-    },
-    {
-      "epoch": 0.052672750977835726,
-      "grad_norm": 3.6881065368652344,
-      "learning_rate": 2e-05,
-      "loss": 7.6187,
-      "step": 101
-    },
-    {
-      "epoch": 0.053194263363754886,
-      "grad_norm": 3.723168134689331,
-      "learning_rate": 2.0200000000000003e-05,
-      "loss": 7.6025,
-      "step": 102
-    },
-    {
-      "epoch": 0.05371577574967405,
-      "grad_norm": 3.4807724952697754,
-      "learning_rate": 2.04e-05,
-      "loss": 7.5207,
-      "step": 103
-    },
-    {
-      "epoch": 0.05423728813559322,
-      "grad_norm": 5.443449974060059,
-      "learning_rate": 2.06e-05,
-      "loss": 7.6484,
-      "step": 104
-    },
-    {
-      "epoch": 0.054758800521512385,
-      "grad_norm": 4.682538986206055,
-      "learning_rate": 2.08e-05,
-      "loss": 7.466,
-      "step": 105
-    },
-    {
-      "epoch": 0.05528031290743155,
-      "grad_norm": 3.433757781982422,
-      "learning_rate": 2.1e-05,
-      "loss": 7.6806,
-      "step": 106
-    },
-    {
-      "epoch": 0.05580182529335072,
-      "grad_norm": 3.9352850914001465,
-      "learning_rate": 2.12e-05,
-      "loss": 7.4971,
-      "step": 107
-    },
-    {
-      "epoch": 0.056323337679269884,
-      "grad_norm": 4.322215557098389,
-      "learning_rate": 2.1400000000000002e-05,
-      "loss": 7.3326,
-      "step": 108
-    },
-    {
-      "epoch": 0.05684485006518905,
-      "grad_norm": 3.155344247817993,
-      "learning_rate": 2.16e-05,
-      "loss": 7.6482,
-      "step": 109
-    },
-    {
-      "epoch": 0.05736636245110821,
-      "grad_norm": 3.5959250926971436,
-      "learning_rate": 2.18e-05,
-      "loss": 7.4452,
-      "step": 110
-    },
-    {
-      "epoch": 0.05788787483702738,
-      "grad_norm": 3.504971742630005,
-      "learning_rate": 2.2000000000000003e-05,
-      "loss": 7.5226,
-      "step": 111
-    },
-    {
-      "epoch": 0.05840938722294654,
-      "grad_norm": 4.071589946746826,
-      "learning_rate": 2.22e-05,
-      "loss": 7.5775,
-      "step": 112
-    },
-    {
-      "epoch": 0.05893089960886571,
-      "grad_norm": 4.240148544311523,
-      "learning_rate": 2.2400000000000002e-05,
-      "loss": 7.3025,
-      "step": 113
-    },
-    {
-      "epoch": 0.059452411994784876,
-      "grad_norm": 2.76248836517334,
-      "learning_rate": 2.26e-05,
-      "loss": 7.6323,
-      "step": 114
-    },
-    {
-      "epoch": 0.05997392438070404,
-      "grad_norm": 2.944532632827759,
-      "learning_rate": 2.2800000000000002e-05,
-      "loss": 7.5755,
-      "step": 115
-    },
-    {
-      "epoch": 0.06049543676662321,
-      "grad_norm": 3.8600049018859863,
-      "learning_rate": 2.3000000000000003e-05,
-      "loss": 7.5862,
-      "step": 116
-    },
-    {
-      "epoch": 0.061016949152542375,
-      "grad_norm": 3.798712730407715,
-      "learning_rate": 2.32e-05,
-      "loss": 7.2015,
-      "step": 117
-    },
-    {
-      "epoch": 0.06153846153846154,
-      "grad_norm": 4.453895092010498,
-      "learning_rate": 2.3400000000000003e-05,
-      "loss": 7.4686,
-      "step": 118
-    },
-    {
-      "epoch": 0.0620599739243807,
-      "grad_norm": 3.261066198348999,
-      "learning_rate": 2.36e-05,
-      "loss": 7.4823,
-      "step": 119
-    },
-    {
-      "epoch": 0.06258148631029987,
-      "grad_norm": 4.026912689208984,
-      "learning_rate": 2.38e-05,
-      "loss": 6.8815,
-      "step": 120
-    },
-    {
-      "epoch": 0.06310299869621903,
-      "grad_norm": 5.84824800491333,
-      "learning_rate": 2.4e-05,
-      "loss": 6.57,
-      "step": 121
-    },
-    {
-      "epoch": 0.06362451108213821,
-      "grad_norm": 3.3494112491607666,
-      "learning_rate": 2.4200000000000002e-05,
-      "loss": 7.5444,
-      "step": 122
-    },
-    {
-      "epoch": 0.06414602346805737,
-      "grad_norm": 3.446805477142334,
-      "learning_rate": 2.44e-05,
-      "loss": 7.4487,
-      "step": 123
-    },
-    {
-      "epoch": 0.06466753585397653,
-      "grad_norm": 3.2302680015563965,
-      "learning_rate": 2.46e-05,
-      "loss": 7.5182,
-      "step": 124
-    },
-    {
-      "epoch": 0.0651890482398957,
-      "grad_norm": 3.4136993885040283,
-      "learning_rate": 2.48e-05,
-      "loss": 7.4691,
-      "step": 125
-    },
-    {
-      "epoch": 0.06571056062581486,
-      "grad_norm": 3.1575052738189697,
-      "learning_rate": 2.5e-05,
-      "loss": 7.587,
-      "step": 126
-    },
-    {
-      "epoch": 0.06623207301173403,
-      "grad_norm": 3.0641024112701416,
-      "learning_rate": 2.5200000000000003e-05,
-      "loss": 7.4983,
-      "step": 127
-    },
-    {
-      "epoch": 0.06675358539765319,
-      "grad_norm": 3.3988382816314697,
-      "learning_rate": 2.54e-05,
-      "loss": 7.4704,
-      "step": 128
-    },
-    {
-      "epoch": 0.06727509778357237,
-      "grad_norm": 3.5170507431030273,
-      "learning_rate": 2.5600000000000002e-05,
-      "loss": 7.4556,
-      "step": 129
-    },
-    {
-      "epoch": 0.06779661016949153,
-      "grad_norm": 3.233266830444336,
-      "learning_rate": 2.58e-05,
-      "loss": 7.5896,
-      "step": 130
-    },
-    {
-      "epoch": 0.06831812255541068,
-      "grad_norm": 5.319973468780518,
-      "learning_rate": 2.6000000000000002e-05,
-      "loss": 7.062,
-      "step": 131
-    },
-    {
-      "epoch": 0.06883963494132986,
-      "grad_norm": 4.150874137878418,
-      "learning_rate": 2.6200000000000003e-05,
-      "loss": 7.3406,
-      "step": 132
-    },
-    {
-      "epoch": 0.06936114732724902,
-      "grad_norm": 3.942434787750244,
-      "learning_rate": 2.64e-05,
-      "loss": 7.3488,
-      "step": 133
-    },
-    {
-      "epoch": 0.06988265971316819,
-      "grad_norm": 6.038638114929199,
-      "learning_rate": 2.6600000000000003e-05,
-      "loss": 7.2262,
-      "step": 134
-    },
-    {
-      "epoch": 0.07040417209908735,
-      "grad_norm": 4.91567325592041,
-      "learning_rate": 2.6800000000000004e-05,
-      "loss": 7.3306,
-      "step": 135
-    },
-    {
-      "epoch": 0.07092568448500652,
-      "grad_norm": 3.807650327682495,
-      "learning_rate": 2.7000000000000002e-05,
-      "loss": 7.233,
-      "step": 136
-    },
-    {
-      "epoch": 0.07144719687092568,
-      "grad_norm": 3.7115111351013184,
-      "learning_rate": 2.7200000000000004e-05,
-      "loss": 7.4093,
-      "step": 137
-    },
-    {
-      "epoch": 0.07196870925684486,
-      "grad_norm": 4.007230758666992,
-      "learning_rate": 2.7400000000000002e-05,
-      "loss": 7.3865,
-      "step": 138
-    },
-    {
-      "epoch": 0.07249022164276402,
-      "grad_norm": 4.70639705657959,
-      "learning_rate": 2.7600000000000003e-05,
-      "loss": 7.1557,
-      "step": 139
-    },
-    {
-      "epoch": 0.07301173402868318,
-      "grad_norm": 4.678992748260498,
-      "learning_rate": 2.7800000000000005e-05,
-      "loss": 7.5268,
-      "step": 140
-    },
-    {
-      "epoch": 0.07353324641460235,
-      "grad_norm": 5.117120742797852,
-      "learning_rate": 2.8000000000000003e-05,
-      "loss": 7.3199,
-      "step": 141
-    },
-    {
-      "epoch": 0.07405475880052151,
-      "grad_norm": 4.096368312835693,
-      "learning_rate": 2.8199999999999998e-05,
-      "loss": 7.322,
-      "step": 142
-    },
-    {
-      "epoch": 0.07457627118644068,
-      "grad_norm": 3.6049387454986572,
-      "learning_rate": 2.84e-05,
-      "loss": 7.4737,
-      "step": 143
-    },
-    {
-      "epoch": 0.07509778357235984,
-      "grad_norm": 3.5590572357177734,
-      "learning_rate": 2.86e-05,
-      "loss": 7.5124,
-      "step": 144
-    },
-    {
-      "epoch": 0.07561929595827901,
-      "grad_norm": 3.052095890045166,
-      "learning_rate": 2.88e-05,
-      "loss": 7.4555,
-      "step": 145
-    },
-    {
-      "epoch": 0.07614080834419817,
-      "grad_norm": 3.6913254261016846,
-      "learning_rate": 2.9e-05,
-      "loss": 7.5133,
-      "step": 146
-    },
-    {
-      "epoch": 0.07666232073011733,
-      "grad_norm": 2.9980180263519287,
-      "learning_rate": 2.9199999999999998e-05,
-      "loss": 7.5124,
-      "step": 147
-    },
-    {
-      "epoch": 0.07718383311603651,
-      "grad_norm": 3.3327064514160156,
-      "learning_rate": 2.94e-05,
-      "loss": 7.3965,
-      "step": 148
-    },
-    {
-      "epoch": 0.07770534550195567,
-      "grad_norm": 3.835602045059204,
-      "learning_rate": 2.96e-05,
-      "loss": 7.3371,
-      "step": 149
-    },
-    {
-      "epoch": 0.07822685788787484,
-      "grad_norm": 3.212507486343384,
-      "learning_rate": 2.98e-05,
-      "loss": 7.3495,
-      "step": 150
-    },
-    {
-      "epoch": 0.078748370273794,
-      "grad_norm": 3.987884759902954,
-      "learning_rate": 3e-05,
-      "loss": 6.7947,
-      "step": 151
-    },
-    {
-      "epoch": 0.07926988265971317,
-      "grad_norm": 3.093217611312866,
-      "learning_rate": 3.02e-05,
-      "loss": 7.2575,
-      "step": 152
-    },
-    {
-      "epoch": 0.07979139504563233,
-      "grad_norm": 3.0359363555908203,
-      "learning_rate": 3.04e-05,
-      "loss": 7.5516,
-      "step": 153
-    },
-    {
-      "epoch": 0.0803129074315515,
-      "grad_norm": 3.396735906600952,
-      "learning_rate": 3.06e-05,
-      "loss": 7.2979,
-      "step": 154
-    },
-    {
-      "epoch": 0.08083441981747067,
-      "grad_norm": 3.282459259033203,
-      "learning_rate": 3.08e-05,
-      "loss": 7.2913,
-      "step": 155
-    },
-    {
-      "epoch": 0.08135593220338982,
-      "grad_norm": 3.2505970001220703,
-      "learning_rate": 3.1e-05,
-      "loss": 7.6147,
-      "step": 156
-    },
-    {
-      "epoch": 0.081877444589309,
-      "grad_norm": 3.0416345596313477,
-      "learning_rate": 3.12e-05,
-      "loss": 7.5666,
-      "step": 157
-    },
-    {
-      "epoch": 0.08239895697522816,
-      "grad_norm": 3.6219711303710938,
-      "learning_rate": 3.1400000000000004e-05,
-      "loss": 7.4667,
-      "step": 158
-    },
-    {
-      "epoch": 0.08292046936114733,
-      "grad_norm": 2.634970188140869,
-      "learning_rate": 3.16e-05,
-      "loss": 7.5679,
-      "step": 159
-    },
-    {
-      "epoch": 0.08344198174706649,
-      "grad_norm": 3.3388631343841553,
-      "learning_rate": 3.18e-05,
-      "loss": 7.5941,
-      "step": 160
-    },
-    {
-      "epoch": 0.08396349413298566,
-      "grad_norm": 3.418192148208618,
-      "learning_rate": 3.2000000000000005e-05,
-      "loss": 7.3521,
-      "step": 161
-    },
-    {
-      "epoch": 0.08448500651890482,
-      "grad_norm": 4.8227105140686035,
-      "learning_rate": 3.2200000000000003e-05,
-      "loss": 6.8505,
-      "step": 162
-    },
-    {
-      "epoch": 0.08500651890482398,
-      "grad_norm": 3.985224723815918,
-      "learning_rate": 3.24e-05,
-      "loss": 7.463,
-      "step": 163
-    },
-    {
-      "epoch": 0.08552803129074316,
-      "grad_norm": 3.390418767929077,
-      "learning_rate": 3.26e-05,
-      "loss": 7.3933,
-      "step": 164
-    },
-    {
-      "epoch": 0.08604954367666232,
-      "grad_norm": 3.4502861499786377,
-      "learning_rate": 3.2800000000000004e-05,
-      "loss": 7.2525,
-      "step": 165
-    },
-    {
-      "epoch": 0.08657105606258149,
-      "grad_norm": 3.928596258163452,
-      "learning_rate": 3.3e-05,
-      "loss": 7.4648,
-      "step": 166
-    },
-    {
-      "epoch": 0.08709256844850065,
-      "grad_norm": 2.900792121887207,
-      "learning_rate": 3.32e-05,
-      "loss": 7.4054,
-      "step": 167
-    },
-    {
-      "epoch": 0.08761408083441982,
-      "grad_norm": 3.5969791412353516,
-      "learning_rate": 3.3400000000000005e-05,
-      "loss": 7.1412,
-      "step": 168
-    },
-    {
-      "epoch": 0.08813559322033898,
-      "grad_norm": 2.8989689350128174,
-      "learning_rate": 3.3600000000000004e-05,
-      "loss": 7.4241,
-      "step": 169
-    },
-    {
-      "epoch": 0.08865710560625815,
-      "grad_norm": 2.927574872970581,
-      "learning_rate": 3.38e-05,
-      "loss": 7.3529,
-      "step": 170
-    },
-    {
-      "epoch": 0.08917861799217731,
-      "grad_norm": 3.080883264541626,
-      "learning_rate": 3.4000000000000007e-05,
-      "loss": 7.3821,
-      "step": 171
-    },
-    {
-      "epoch": 0.08970013037809647,
-      "grad_norm": 3.5870022773742676,
-      "learning_rate": 3.4200000000000005e-05,
-      "loss": 7.0631,
-      "step": 172
-    },
-    {
-      "epoch": 0.09022164276401565,
-      "grad_norm": 3.4165422916412354,
-      "learning_rate": 3.4399999999999996e-05,
-      "loss": 7.4823,
-      "step": 173
-    },
-    {
-      "epoch": 0.0907431551499348,
-      "grad_norm": 3.157397985458374,
-      "learning_rate": 3.46e-05,
-      "loss": 7.3439,
-      "step": 174
-    },
-    {
-      "epoch": 0.09126466753585398,
-      "grad_norm": 3.3898069858551025,
-      "learning_rate": 3.48e-05,
-      "loss": 7.2797,
-      "step": 175
-    },
-    {
-      "epoch": 0.09178617992177314,
-      "grad_norm": 3.834183931350708,
-      "learning_rate": 3.5e-05,
-      "loss": 7.2583,
-      "step": 176
-    },
-    {
-      "epoch": 0.09230769230769231,
-      "grad_norm": 2.7050435543060303,
-      "learning_rate": 3.52e-05,
-      "loss": 7.5331,
-      "step": 177
-    },
-    {
-      "epoch": 0.09282920469361147,
-      "grad_norm": 2.892559766769409,
-      "learning_rate": 3.54e-05,
-      "loss": 7.5713,
-      "step": 178
-    },
-    {
-      "epoch": 0.09335071707953065,
-      "grad_norm": 3.3834376335144043,
-      "learning_rate": 3.56e-05,
-      "loss": 7.3731,
-      "step": 179
-    },
-    {
-      "epoch": 0.0938722294654498,
-      "grad_norm": 2.4844014644622803,
-      "learning_rate": 3.58e-05,
-      "loss": 7.576,
-      "step": 180
-    },
-    {
-      "epoch": 0.09439374185136896,
-      "grad_norm": 3.313626766204834,
-      "learning_rate": 3.6e-05,
-      "loss": 7.4822,
-      "step": 181
-    },
-    {
-      "epoch": 0.09491525423728814,
-      "grad_norm": 4.537646770477295,
-      "learning_rate": 3.62e-05,
-      "loss": 7.2434,
-      "step": 182
-    },
-    {
-      "epoch": 0.0954367666232073,
-      "grad_norm": 3.511963367462158,
-      "learning_rate": 3.6400000000000004e-05,
-      "loss": 7.4671,
-      "step": 183
-    },
-    {
-      "epoch": 0.09595827900912647,
-      "grad_norm": 3.1389901638031006,
-      "learning_rate": 3.66e-05,
-      "loss": 7.2465,
-      "step": 184
-    },
-    {
-      "epoch": 0.09647979139504563,
-      "grad_norm": 3.084373712539673,
-      "learning_rate": 3.68e-05,
-      "loss": 7.4383,
-      "step": 185
-    },
-    {
-      "epoch": 0.0970013037809648,
-      "grad_norm": 3.6729354858398438,
-      "learning_rate": 3.7e-05,
-      "loss": 7.1553,
-      "step": 186
-    },
-    {
-      "epoch": 0.09752281616688396,
-      "grad_norm": 3.3839831352233887,
-      "learning_rate": 3.72e-05,
-      "loss": 7.3268,
-      "step": 187
-    },
-    {
-      "epoch": 0.09804432855280312,
-      "grad_norm": 4.4720892906188965,
-      "learning_rate": 3.74e-05,
-      "loss": 7.4081,
-      "step": 188
-    },
-    {
-      "epoch": 0.0985658409387223,
-      "grad_norm": 3.6028249263763428,
-      "learning_rate": 3.76e-05,
-      "loss": 7.0773,
-      "step": 189
-    },
-    {
-      "epoch": 0.09908735332464146,
-      "grad_norm": 3.095641613006592,
-      "learning_rate": 3.7800000000000004e-05,
-      "loss": 7.5315,
-      "step": 190
-    },
-    {
-      "epoch": 0.09960886571056063,
-      "grad_norm": 2.9480347633361816,
-      "learning_rate": 3.8e-05,
-      "loss": 7.4514,
-      "step": 191
-    },
-    {
-      "epoch": 0.10013037809647979,
-      "grad_norm": 2.6473734378814697,
-      "learning_rate": 3.82e-05,
-      "loss": 7.5321,
-      "step": 192
-    },
-    {
-      "epoch": 0.10065189048239896,
-      "grad_norm": 4.760178089141846,
-      "learning_rate": 3.8400000000000005e-05,
-      "loss": 7.3236,
-      "step": 193
-    },
-    {
-      "epoch": 0.10117340286831812,
-      "grad_norm": 2.964203357696533,
-      "learning_rate": 3.86e-05,
-      "loss": 7.5497,
-      "step": 194
-    },
-    {
-      "epoch": 0.1016949152542373,
-      "grad_norm": 3.2622687816619873,
-      "learning_rate": 3.88e-05,
-      "loss": 7.2481,
-      "step": 195
-    },
-    {
-      "epoch": 0.10221642764015645,
-      "grad_norm": 3.5849318504333496,
-      "learning_rate": 3.9000000000000006e-05,
-      "loss": 6.9631,
-      "step": 196
-    },
-    {
-      "epoch": 0.10273794002607561,
-      "grad_norm": 2.8486852645874023,
-      "learning_rate": 3.9200000000000004e-05,
-      "loss": 7.3895,
-      "step": 197
-    },
-    {
-      "epoch": 0.10325945241199479,
-      "grad_norm": 4.567874431610107,
-      "learning_rate": 3.94e-05,
-      "loss": 7.0744,
-      "step": 198
-    },
-    {
-      "epoch": 0.10378096479791395,
-      "grad_norm": 3.6248340606689453,
-      "learning_rate": 3.960000000000001e-05,
-      "loss": 7.0411,
-      "step": 199
-    },
-    {
-      "epoch": 0.10430247718383312,
-      "grad_norm": 2.7067694664001465,
-      "learning_rate": 3.9800000000000005e-05,
-      "loss": 7.3924,
-      "step": 200
-    },
-    {
-      "epoch": 0.10482398956975228,
-      "grad_norm": 3.14599871635437,
-      "learning_rate": 4e-05,
-      "loss": 7.3167,
-      "step": 201
-    },
-    {
-      "epoch": 0.10534550195567145,
-      "grad_norm": 2.6617326736450195,
-      "learning_rate": 4.02e-05,
-      "loss": 7.5365,
-      "step": 202
-    },
-    {
-      "epoch": 0.10586701434159061,
-      "grad_norm": 4.169859886169434,
-      "learning_rate": 4.0400000000000006e-05,
-      "loss": 7.0596,
-      "step": 203
-    },
-    {
-      "epoch": 0.10638852672750977,
-      "grad_norm": 2.913005828857422,
-      "learning_rate": 4.0600000000000004e-05,
-      "loss": 7.475,
-      "step": 204
-    },
-    {
-      "epoch": 0.10691003911342895,
-      "grad_norm": 3.477138042449951,
-      "learning_rate": 4.08e-05,
-      "loss": 7.2434,
-      "step": 205
-    },
-    {
-      "epoch": 0.1074315514993481,
-      "grad_norm": 2.9127392768859863,
-      "learning_rate": 4.1e-05,
-      "loss": 7.4914,
-      "step": 206
-    },
-    {
-      "epoch": 0.10795306388526728,
-      "grad_norm": 3.4160966873168945,
-      "learning_rate": 4.12e-05,
-      "loss": 7.2338,
-      "step": 207
-    },
-    {
-      "epoch": 0.10847457627118644,
-      "grad_norm": 3.0882041454315186,
-      "learning_rate": 4.14e-05,
-      "loss": 7.2559,
-      "step": 208
-    },
-    {
-      "epoch": 0.10899608865710561,
-      "grad_norm": 3.879185438156128,
-      "learning_rate": 4.16e-05,
-      "loss": 7.0722,
-      "step": 209
-    },
-    {
-      "epoch": 0.10951760104302477,
-      "grad_norm": 2.9452905654907227,
-      "learning_rate": 4.18e-05,
-      "loss": 7.4291,
-      "step": 210
-    },
-    {
-      "epoch": 0.11003911342894394,
-      "grad_norm": 4.876804351806641,
-      "learning_rate": 4.2e-05,
-      "loss": 7.4369,
-      "step": 211
-    },
-    {
-      "epoch": 0.1105606258148631,
-      "grad_norm": 2.561800956726074,
-      "learning_rate": 4.22e-05,
-      "loss": 7.4774,
-      "step": 212
-    },
-    {
-      "epoch": 0.11108213820078226,
-      "grad_norm": 4.161262035369873,
-      "learning_rate": 4.24e-05,
-      "loss": 6.5584,
-      "step": 213
-    },
-    {
-      "epoch": 0.11160365058670144,
-      "grad_norm": 3.3121731281280518,
-      "learning_rate": 4.26e-05,
-      "loss": 7.2167,
-      "step": 214
-    },
-    {
-      "epoch": 0.1121251629726206,
-      "grad_norm": 3.4215073585510254,
-      "learning_rate": 4.2800000000000004e-05,
-      "loss": 7.0574,
-      "step": 215
-    },
-    {
-      "epoch": 0.11264667535853977,
-      "grad_norm": 2.8985512256622314,
-      "learning_rate": 4.3e-05,
-      "loss": 7.3702,
-      "step": 216
-    },
-    {
-      "epoch": 0.11316818774445893,
-      "grad_norm": 3.5382654666900635,
-      "learning_rate": 4.32e-05,
-      "loss": 7.128,
-      "step": 217
-    },
-    {
-      "epoch": 0.1136897001303781,
-      "grad_norm": 2.850515127182007,
-      "learning_rate": 4.3400000000000005e-05,
-      "loss": 7.4392,
-      "step": 218
-    },
-    {
-      "epoch": 0.11421121251629726,
-      "grad_norm": 3.423903703689575,
-      "learning_rate": 4.36e-05,
-      "loss": 7.3433,
-      "step": 219
-    },
-    {
-      "epoch": 0.11473272490221642,
-      "grad_norm": 5.7699360847473145,
-      "learning_rate": 4.38e-05,
-      "loss": 7.2612,
-      "step": 220
-    },
-    {
-      "epoch": 0.1152542372881356,
-      "grad_norm": 4.314525127410889,
-      "learning_rate": 4.4000000000000006e-05,
-      "loss": 7.3813,
-      "step": 221
-    },
-    {
-      "epoch": 0.11577574967405475,
-      "grad_norm": 4.324763774871826,
-      "learning_rate": 4.4200000000000004e-05,
-      "loss": 6.7292,
-      "step": 222
-    },
-    {
-      "epoch": 0.11629726205997393,
-      "grad_norm": 7.078699111938477,
-      "learning_rate": 4.44e-05,
-      "loss": 7.0438,
-      "step": 223
-    },
-    {
-      "epoch": 0.11681877444589309,
-      "grad_norm": 5.300456523895264,
-      "learning_rate": 4.46e-05,
-      "loss": 7.1135,
-      "step": 224
-    },
-    {
-      "epoch": 0.11734028683181226,
-      "grad_norm": 3.453906536102295,
-      "learning_rate": 4.4800000000000005e-05,
-      "loss": 7.3285,
-      "step": 225
-    },
-    {
-      "epoch": 0.11786179921773142,
-      "grad_norm": 5.6908183097839355,
-      "learning_rate": 4.5e-05,
-      "loss": 7.0077,
-      "step": 226
-    },
-    {
-      "epoch": 0.11838331160365059,
-      "grad_norm": 3.9178898334503174,
-      "learning_rate": 4.52e-05,
-      "loss": 7.3279,
-      "step": 227
-    },
-    {
-      "epoch": 0.11890482398956975,
-      "grad_norm": 3.508704900741577,
-      "learning_rate": 4.5400000000000006e-05,
-      "loss": 7.144,
-      "step": 228
-    },
-    {
-      "epoch": 0.11942633637548891,
-      "grad_norm": 4.181432247161865,
-      "learning_rate": 4.5600000000000004e-05,
-      "loss": 7.241,
-      "step": 229
-    },
-    {
-      "epoch": 0.11994784876140809,
-      "grad_norm": 2.5863866806030273,
-      "learning_rate": 4.58e-05,
-      "loss": 7.4811,
-      "step": 230
-    },
-    {
-      "epoch": 0.12046936114732724,
-      "grad_norm": 8.508010864257812,
-      "learning_rate": 4.600000000000001e-05,
-      "loss": 6.8705,
-      "step": 231
-    },
-    {
-      "epoch": 0.12099087353324642,
-      "grad_norm": 3.1672141551971436,
-      "learning_rate": 4.6200000000000005e-05,
-      "loss": 7.3065,
-      "step": 232
-    },
-    {
-      "epoch": 0.12151238591916558,
-      "grad_norm": 3.6299948692321777,
-      "learning_rate": 4.64e-05,
-      "loss": 7.1479,
-      "step": 233
-    },
-    {
-      "epoch": 0.12203389830508475,
-      "grad_norm": 4.699325084686279,
-      "learning_rate": 4.660000000000001e-05,
-      "loss": 6.7042,
-      "step": 234
-    },
-    {
-      "epoch": 0.12255541069100391,
-      "grad_norm": 3.7025651931762695,
-      "learning_rate": 4.6800000000000006e-05,
-      "loss": 7.3001,
-      "step": 235
-    },
-    {
-      "epoch": 0.12307692307692308,
-      "grad_norm": 3.0916712284088135,
-      "learning_rate": 4.7e-05,
-      "loss": 7.235,
-      "step": 236
-    },
-    {
-      "epoch": 0.12359843546284224,
-      "grad_norm": 4.043112277984619,
-      "learning_rate": 4.72e-05,
-      "loss": 6.7526,
-      "step": 237
-    },
-    {
-      "epoch": 0.1241199478487614,
-      "grad_norm": 3.4944348335266113,
-      "learning_rate": 4.74e-05,
-      "loss": 7.3302,
-      "step": 238
-    },
-    {
-      "epoch": 0.12464146023468058,
-      "grad_norm": 2.8404836654663086,
-      "learning_rate": 4.76e-05,
-      "loss": 7.398,
-      "step": 239
-    },
-    {
-      "epoch": 0.12516297262059975,
-      "grad_norm": 4.0313801765441895,
-      "learning_rate": 4.78e-05,
-      "loss": 6.9056,
-      "step": 240
-    },
-    {
-      "epoch": 0.1256844850065189,
-      "grad_norm": 3.906691789627075,
-      "learning_rate": 4.8e-05,
-      "loss": 6.644,
-      "step": 241
-    },
-    {
-      "epoch": 0.12620599739243807,
-      "grad_norm": 3.601623773574829,
-      "learning_rate": 4.82e-05,
-      "loss": 7.0927,
-      "step": 242
-    },
-    {
-      "epoch": 0.12672750977835723,
-      "grad_norm": 2.7283427715301514,
-      "learning_rate": 4.8400000000000004e-05,
-      "loss": 7.4099,
-      "step": 243
-    },
-    {
-      "epoch": 0.12724902216427642,
-      "grad_norm": 3.994126796722412,
-      "learning_rate": 4.86e-05,
-      "loss": 6.9068,
-      "step": 244
-    },
-    {
-      "epoch": 0.12777053455019557,
-      "grad_norm": 3.048358917236328,
-      "learning_rate": 4.88e-05,
-      "loss": 7.3069,
-      "step": 245
-    },
-    {
-      "epoch": 0.12829204693611473,
-      "grad_norm": 3.4312479496002197,
-      "learning_rate": 4.9e-05,
-      "loss": 7.2706,
-      "step": 246
-    },
-    {
-      "epoch": 0.1288135593220339,
-      "grad_norm": 3.605164051055908,
-      "learning_rate": 4.92e-05,
-      "loss": 7.1076,
-      "step": 247
-    },
-    {
-      "epoch": 0.12933507170795305,
-      "grad_norm": 5.564464569091797,
-      "learning_rate": 4.94e-05,
-      "loss": 6.2989,
-      "step": 248
-    },
-    {
-      "epoch": 0.12985658409387224,
-      "grad_norm": 3.8566784858703613,
-      "learning_rate": 4.96e-05,
-      "loss": 7.0423,
-      "step": 249
-    },
-    {
-      "epoch": 0.1303780964797914,
-      "grad_norm": 3.1338601112365723,
-      "learning_rate": 4.9800000000000004e-05,
-      "loss": 7.0444,
-      "step": 250
-    },
-    {
-      "epoch": 0.13089960886571056,
-      "grad_norm": 2.8090527057647705,
-      "learning_rate": 5e-05,
-      "loss": 7.4891,
-      "step": 251
-    },
-    {
-      "epoch": 0.13142112125162972,
-      "grad_norm": 7.6368632316589355,
-      "learning_rate": 5.02e-05,
-      "loss": 6.0822,
-      "step": 252
-    },
-    {
-      "epoch": 0.13194263363754888,
-      "grad_norm": 4.872903823852539,
-      "learning_rate": 5.0400000000000005e-05,
-      "loss": 7.0497,
-      "step": 253
-    },
-    {
-      "epoch": 0.13246414602346807,
-      "grad_norm": 5.52940559387207,
-      "learning_rate": 5.0600000000000003e-05,
-      "loss": 6.6472,
-      "step": 254
-    },
-    {
-      "epoch": 0.13298565840938723,
-      "grad_norm": 2.805499792098999,
-      "learning_rate": 5.08e-05,
-      "loss": 7.3001,
-      "step": 255
-    },
-    {
-      "epoch": 0.13350717079530638,
-      "grad_norm": 4.028730869293213,
-      "learning_rate": 5.1000000000000006e-05,
-      "loss": 6.8972,
-      "step": 256
-    },
-    {
-      "epoch": 0.13402868318122554,
-      "grad_norm": 4.018136024475098,
-      "learning_rate": 5.1200000000000004e-05,
-      "loss": 7.3112,
-      "step": 257
-    },
-    {
-      "epoch": 0.13455019556714473,
-      "grad_norm": 3.7548141479492188,
-      "learning_rate": 5.14e-05,
-      "loss": 7.1665,
-      "step": 258
-    },
-    {
-      "epoch": 0.1350717079530639,
-      "grad_norm": 3.9899773597717285,
-      "learning_rate": 5.16e-05,
-      "loss": 6.7029,
-      "step": 259
-    },
-    {
-      "epoch": 0.13559322033898305,
-      "grad_norm": 5.135863304138184,
-      "learning_rate": 5.1800000000000005e-05,
-      "loss": 7.2739,
-      "step": 260
-    },
-    {
-      "epoch": 0.1361147327249022,
-      "grad_norm": 2.649033784866333,
-      "learning_rate": 5.2000000000000004e-05,
-      "loss": 7.3011,
-      "step": 261
-    },
-    {
-      "epoch": 0.13663624511082137,
-      "grad_norm": 3.3404154777526855,
-      "learning_rate": 5.22e-05,
-      "loss": 7.3246,
-      "step": 262
-    },
-    {
-      "epoch": 0.13715775749674056,
-      "grad_norm": 3.8582091331481934,
-      "learning_rate": 5.2400000000000007e-05,
-      "loss": 7.0718,
-      "step": 263
-    },
-    {
-      "epoch": 0.13767926988265972,
-      "grad_norm": 2.9569761753082275,
-      "learning_rate": 5.2600000000000005e-05,
-      "loss": 7.1488,
-      "step": 264
-    },
-    {
-      "epoch": 0.13820078226857888,
-      "grad_norm": 2.694443702697754,
-      "learning_rate": 5.28e-05,
-      "loss": 7.366,
-      "step": 265
-    },
-    {
-      "epoch": 0.13872229465449804,
-      "grad_norm": 5.168875694274902,
-      "learning_rate": 5.300000000000001e-05,
-      "loss": 7.0269,
-      "step": 266
-    },
-    {
-      "epoch": 0.13924380704041722,
-      "grad_norm": 4.086500644683838,
-      "learning_rate": 5.3200000000000006e-05,
-      "loss": 7.2118,
-      "step": 267
-    },
-    {
-      "epoch": 0.13976531942633638,
-      "grad_norm": 4.048635959625244,
-      "learning_rate": 5.3400000000000004e-05,
-      "loss": 7.21,
-      "step": 268
-    },
-    {
-      "epoch": 0.14028683181225554,
-      "grad_norm": 4.643430233001709,
-      "learning_rate": 5.360000000000001e-05,
-      "loss": 6.8379,
-      "step": 269
-    },
-    {
-      "epoch": 0.1408083441981747,
-      "grad_norm": 4.895519733428955,
-      "learning_rate": 5.380000000000001e-05,
-      "loss": 6.992,
-      "step": 270
-    },
-    {
-      "epoch": 0.14132985658409386,
-      "grad_norm": 3.6653215885162354,
-      "learning_rate": 5.4000000000000005e-05,
-      "loss": 6.9136,
-      "step": 271
-    },
-    {
-      "epoch": 0.14185136897001305,
-      "grad_norm": 4.5786662101745605,
-      "learning_rate": 5.420000000000001e-05,
-      "loss": 6.9135,
-      "step": 272
-    },
-    {
-      "epoch": 0.1423728813559322,
-      "grad_norm": 4.77662992477417,
-      "learning_rate": 5.440000000000001e-05,
-      "loss": 6.5562,
-      "step": 273
-    },
-    {
-      "epoch": 0.14289439374185137,
-      "grad_norm": 3.748788595199585,
-      "learning_rate": 5.4600000000000006e-05,
-      "loss": 7.1313,
-      "step": 274
-    },
-    {
-      "epoch": 0.14341590612777053,
-      "grad_norm": 2.907221555709839,
-      "learning_rate": 5.4800000000000004e-05,
-      "loss": 7.2938,
-      "step": 275
-    },
-    {
-      "epoch": 0.1439374185136897,
-      "grad_norm": 4.719839096069336,
-      "learning_rate": 5.500000000000001e-05,
-      "loss": 7.179,
-      "step": 276
-    },
-    {
-      "epoch": 0.14445893089960887,
-      "grad_norm": 4.704406261444092,
-      "learning_rate": 5.520000000000001e-05,
-      "loss": 6.7314,
-      "step": 277
-    },
-    {
-      "epoch": 0.14498044328552803,
-      "grad_norm": 3.7895584106445312,
-      "learning_rate": 5.5400000000000005e-05,
-      "loss": 7.1991,
-      "step": 278
-    },
-    {
-      "epoch": 0.1455019556714472,
-      "grad_norm": 3.7524566650390625,
-      "learning_rate": 5.560000000000001e-05,
-      "loss": 7.1896,
-      "step": 279
-    },
-    {
-      "epoch": 0.14602346805736635,
-      "grad_norm": 3.232865810394287,
-      "learning_rate": 5.580000000000001e-05,
-      "loss": 7.4014,
-      "step": 280
-    },
-    {
-      "epoch": 0.14654498044328554,
-      "grad_norm": 2.797893762588501,
-      "learning_rate": 5.6000000000000006e-05,
-      "loss": 7.1049,
-      "step": 281
-    },
-    {
-      "epoch": 0.1470664928292047,
-      "grad_norm": 3.1756818294525146,
-      "learning_rate": 5.620000000000001e-05,
-      "loss": 6.8909,
-      "step": 282
-    },
-    {
-      "epoch": 0.14758800521512386,
-      "grad_norm": 3.7766144275665283,
-      "learning_rate": 5.6399999999999995e-05,
-      "loss": 6.999,
-      "step": 283
-    },
-    {
-      "epoch": 0.14810951760104302,
-      "grad_norm": 3.2954635620117188,
-      "learning_rate": 5.66e-05,
-      "loss": 6.9622,
-      "step": 284
-    },
-    {
-      "epoch": 0.14863102998696218,
-      "grad_norm": 2.8703413009643555,
-      "learning_rate": 5.68e-05,
-      "loss": 7.0317,
-      "step": 285
-    },
-    {
-      "epoch": 0.14915254237288136,
-      "grad_norm": 3.632871150970459,
-      "learning_rate": 5.6999999999999996e-05,
-      "loss": 7.4122,
-      "step": 286
-    },
-    {
-      "epoch": 0.14967405475880052,
-      "grad_norm": 3.128606081008911,
-      "learning_rate": 5.72e-05,
-      "loss": 7.2959,
-      "step": 287
-    },
-    {
-      "epoch": 0.15019556714471968,
-      "grad_norm": 3.604712724685669,
-      "learning_rate": 5.74e-05,
-      "loss": 6.9134,
-      "step": 288
-    },
-    {
-      "epoch": 0.15071707953063884,
-      "grad_norm": 3.8073525428771973,
-      "learning_rate": 5.76e-05,
-      "loss": 7.05,
-      "step": 289
-    },
-    {
-      "epoch": 0.15123859191655803,
-      "grad_norm": 3.9711482524871826,
-      "learning_rate": 5.7799999999999995e-05,
-      "loss": 7.0299,
-      "step": 290
-    },
-    {
-      "epoch": 0.1517601043024772,
-      "grad_norm": 3.2127535343170166,
-      "learning_rate": 5.8e-05,
-      "loss": 6.9698,
-      "step": 291
-    },
-    {
-      "epoch": 0.15228161668839635,
-      "grad_norm": 4.119180679321289,
-      "learning_rate": 5.82e-05,
-      "loss": 6.839,
-      "step": 292
-    },
-    {
-      "epoch": 0.1528031290743155,
-      "grad_norm": 2.627946615219116,
-      "learning_rate": 5.8399999999999997e-05,
-      "loss": 7.3496,
-      "step": 293
-    },
-    {
-      "epoch": 0.15332464146023467,
-      "grad_norm": 3.819568157196045,
-      "learning_rate": 5.86e-05,
-      "loss": 6.5435,
-      "step": 294
-    },
-    {
-      "epoch": 0.15384615384615385,
-      "grad_norm": 2.9470767974853516,
-      "learning_rate": 5.88e-05,
-      "loss": 7.1781,
-      "step": 295
-    },
-    {
-      "epoch": 0.15436766623207301,
-      "grad_norm": 2.706519603729248,
-      "learning_rate": 5.9e-05,
-      "loss": 7.0022,
-      "step": 296
-    },
-    {
-      "epoch": 0.15488917861799217,
-      "grad_norm": 2.6944267749786377,
-      "learning_rate": 5.92e-05,
-      "loss": 6.7678,
-      "step": 297
-    },
-    {
-      "epoch": 0.15541069100391133,
-      "grad_norm": 2.8793187141418457,
-      "learning_rate": 5.94e-05,
-      "loss": 6.9549,
-      "step": 298
-    },
-    {
-      "epoch": 0.15593220338983052,
-      "grad_norm": 3.094952344894409,
-      "learning_rate": 5.96e-05,
-      "loss": 6.7861,
-      "step": 299
-    },
-    {
-      "epoch": 0.15645371577574968,
-      "grad_norm": 2.9189517498016357,
-      "learning_rate": 5.9800000000000003e-05,
-      "loss": 7.1257,
-      "step": 300
-    },
-    {
-      "epoch": 0.15697522816166884,
-      "grad_norm": 3.8729801177978516,
-      "learning_rate": 6e-05,
-      "loss": 7.1658,
-      "step": 301
-    },
-    {
-      "epoch": 0.157496740547588,
-      "grad_norm": 3.2805135250091553,
-      "learning_rate": 6.02e-05,
-      "loss": 6.7815,
-      "step": 302
-    },
-    {
-      "epoch": 0.15801825293350716,
-      "grad_norm": 3.4292726516723633,
-      "learning_rate": 6.04e-05,
-      "loss": 7.0912,
-      "step": 303
-    },
-    {
-      "epoch": 0.15853976531942635,
-      "grad_norm": 5.145147800445557,
-      "learning_rate": 6.06e-05,
-      "loss": 6.5012,
-      "step": 304
-    },
-    {
-      "epoch": 0.1590612777053455,
-      "grad_norm": 2.6266815662384033,
-      "learning_rate": 6.08e-05,
-      "loss": 6.8954,
-      "step": 305
-    },
-    {
-      "epoch": 0.15958279009126466,
-      "grad_norm": 2.963740825653076,
-      "learning_rate": 6.1e-05,
-      "loss": 6.9509,
-      "step": 306
-    },
-    {
-      "epoch": 0.16010430247718382,
-      "grad_norm": 3.066110372543335,
-      "learning_rate": 6.12e-05,
-      "loss": 6.9598,
-      "step": 307
-    },
-    {
-      "epoch": 0.160625814863103,
-      "grad_norm": 3.2593555450439453,
-      "learning_rate": 6.14e-05,
-      "loss": 6.8141,
-      "step": 308
-    },
-    {
-      "epoch": 0.16114732724902217,
-      "grad_norm": 3.2543344497680664,
-      "learning_rate": 6.16e-05,
-      "loss": 6.9847,
-      "step": 309
-    },
-    {
-      "epoch": 0.16166883963494133,
-      "grad_norm": 3.4616024494171143,
-      "learning_rate": 6.18e-05,
-      "loss": 6.6924,
-      "step": 310
-    },
-    {
-      "epoch": 0.1621903520208605,
-      "grad_norm": 3.570981979370117,
-      "learning_rate": 6.2e-05,
-      "loss": 6.7445,
-      "step": 311
-    },
-    {
-      "epoch": 0.16271186440677965,
-      "grad_norm": 4.955770969390869,
-      "learning_rate": 6.220000000000001e-05,
-      "loss": 6.3881,
-      "step": 312
-    },
-    {
-      "epoch": 0.16323337679269884,
-      "grad_norm": 3.3917860984802246,
-      "learning_rate": 6.24e-05,
-      "loss": 6.5997,
-      "step": 313
-    },
-    {
-      "epoch": 0.163754889178618,
-      "grad_norm": 3.502500295639038,
-      "learning_rate": 6.26e-05,
-      "loss": 6.6838,
-      "step": 314
-    },
-    {
-      "epoch": 0.16427640156453716,
-      "grad_norm": 4.28438663482666,
-      "learning_rate": 6.280000000000001e-05,
-      "loss": 7.0345,
-      "step": 315
-    },
-    {
-      "epoch": 0.16479791395045632,
-      "grad_norm": 2.617302179336548,
-      "learning_rate": 6.3e-05,
-      "loss": 7.3267,
-      "step": 316
-    },
-    {
-      "epoch": 0.1653194263363755,
-      "grad_norm": 5.027987480163574,
-      "learning_rate": 6.32e-05,
-      "loss": 6.7456,
-      "step": 317
-    },
-    {
-      "epoch": 0.16584093872229466,
-      "grad_norm": 3.9923369884490967,
-      "learning_rate": 6.340000000000001e-05,
-      "loss": 6.9154,
-      "step": 318
-    },
-    {
-      "epoch": 0.16636245110821382,
-      "grad_norm": 3.381068229675293,
-      "learning_rate": 6.36e-05,
-      "loss": 6.7825,
-      "step": 319
-    },
-    {
-      "epoch": 0.16688396349413298,
-      "grad_norm": 3.726672410964966,
-      "learning_rate": 6.38e-05,
-      "loss": 7.0749,
-      "step": 320
-    },
-    {
-      "epoch": 0.16740547588005214,
-      "grad_norm": 3.0546324253082275,
-      "learning_rate": 6.400000000000001e-05,
-      "loss": 6.9388,
-      "step": 321
-    },
-    {
-      "epoch": 0.16792698826597133,
-      "grad_norm": 2.4811761379241943,
-      "learning_rate": 6.42e-05,
-      "loss": 7.3094,
-      "step": 322
-    },
-    {
-      "epoch": 0.1684485006518905,
-      "grad_norm": 4.7711710929870605,
-      "learning_rate": 6.440000000000001e-05,
-      "loss": 6.7317,
-      "step": 323
-    },
-    {
-      "epoch": 0.16897001303780965,
-      "grad_norm": 3.3538639545440674,
-      "learning_rate": 6.460000000000001e-05,
-      "loss": 7.0458,
-      "step": 324
-    },
-    {
-      "epoch": 0.1694915254237288,
-      "grad_norm": 3.4860734939575195,
-      "learning_rate": 6.48e-05,
-      "loss": 6.8537,
-      "step": 325
-    },
-    {
-      "epoch": 0.17001303780964797,
-      "grad_norm": 3.4439752101898193,
-      "learning_rate": 6.500000000000001e-05,
-      "loss": 6.6793,
-      "step": 326
-    },
-    {
-      "epoch": 0.17053455019556715,
-      "grad_norm": 2.9555447101593018,
-      "learning_rate": 6.52e-05,
-      "loss": 6.9347,
-      "step": 327
-    },
-    {
-      "epoch": 0.1710560625814863,
-      "grad_norm": 4.9550461769104,
-      "learning_rate": 6.54e-05,
-      "loss": 6.1947,
-      "step": 328
-    },
-    {
-      "epoch": 0.17157757496740547,
-      "grad_norm": 3.706544876098633,
-      "learning_rate": 6.560000000000001e-05,
-      "loss": 7.0336,
-      "step": 329
-    },
-    {
-      "epoch": 0.17209908735332463,
-      "grad_norm": 2.4361014366149902,
-      "learning_rate": 6.58e-05,
-      "loss": 7.2412,
-      "step": 330
-    },
-    {
-      "epoch": 0.17262059973924382,
-      "grad_norm": 4.105840682983398,
-      "learning_rate": 6.6e-05,
-      "loss": 7.0146,
-      "step": 331
-    },
-    {
-      "epoch": 0.17314211212516298,
-      "grad_norm": 2.911102294921875,
-      "learning_rate": 6.620000000000001e-05,
-      "loss": 7.1755,
-      "step": 332
-    },
-    {
-      "epoch": 0.17366362451108214,
-      "grad_norm": 4.427680015563965,
-      "learning_rate": 6.64e-05,
-      "loss": 7.1427,
-      "step": 333
-    },
-    {
-      "epoch": 0.1741851368970013,
-      "grad_norm": 3.9352667331695557,
-      "learning_rate": 6.66e-05,
-      "loss": 7.0795,
-      "step": 334
-    },
-    {
-      "epoch": 0.17470664928292046,
-      "grad_norm": 3.0695817470550537,
-      "learning_rate": 6.680000000000001e-05,
-      "loss": 6.519,
-      "step": 335
-    },
-    {
-      "epoch": 0.17522816166883964,
-      "grad_norm": 5.045691013336182,
-      "learning_rate": 6.7e-05,
-      "loss": 6.8514,
-      "step": 336
-    },
-    {
-      "epoch": 0.1757496740547588,
-      "grad_norm": 3.0756583213806152,
-      "learning_rate": 6.720000000000001e-05,
-      "loss": 6.8241,
-      "step": 337
-    },
-    {
-      "epoch": 0.17627118644067796,
-      "grad_norm": 2.9580092430114746,
-      "learning_rate": 6.740000000000001e-05,
-      "loss": 6.8762,
-      "step": 338
-    },
-    {
-      "epoch": 0.17679269882659712,
-      "grad_norm": 3.5179967880249023,
-      "learning_rate": 6.76e-05,
-      "loss": 6.3788,
-      "step": 339
-    },
-    {
-      "epoch": 0.1773142112125163,
-      "grad_norm": 2.7665321826934814,
-      "learning_rate": 6.780000000000001e-05,
-      "loss": 6.7854,
-      "step": 340
-    },
-    {
-      "epoch": 0.17783572359843547,
-      "grad_norm": 3.25581693649292,
-      "learning_rate": 6.800000000000001e-05,
-      "loss": 6.9394,
-      "step": 341
-    },
-    {
-      "epoch": 0.17835723598435463,
-      "grad_norm": 3.7985944747924805,
-      "learning_rate": 6.82e-05,
-      "loss": 6.7423,
-      "step": 342
-    },
-    {
-      "epoch": 0.1788787483702738,
-      "grad_norm": 2.765812635421753,
-      "learning_rate": 6.840000000000001e-05,
-      "loss": 7.1229,
-      "step": 343
-    },
-    {
-      "epoch": 0.17940026075619295,
-      "grad_norm": 3.301640748977661,
-      "learning_rate": 6.860000000000001e-05,
-      "loss": 6.761,
-      "step": 344
-    },
-    {
-      "epoch": 0.17992177314211213,
-      "grad_norm": 3.5700604915618896,
-      "learning_rate": 6.879999999999999e-05,
-      "loss": 7.0995,
-      "step": 345
-    },
-    {
-      "epoch": 0.1804432855280313,
-      "grad_norm": 3.4261698722839355,
-      "learning_rate": 6.9e-05,
-      "loss": 6.6845,
-      "step": 346
-    },
-    {
-      "epoch": 0.18096479791395045,
-      "grad_norm": 2.867846965789795,
-      "learning_rate": 6.92e-05,
-      "loss": 7.1234,
-      "step": 347
-    },
-    {
-      "epoch": 0.1814863102998696,
-      "grad_norm": 4.3554182052612305,
-      "learning_rate": 6.939999999999999e-05,
-      "loss": 6.3443,
-      "step": 348
-    },
-    {
-      "epoch": 0.1820078226857888,
-      "grad_norm": 4.192983627319336,
-      "learning_rate": 6.96e-05,
-      "loss": 6.6085,
-      "step": 349
-    },
-    {
-      "epoch": 0.18252933507170796,
-      "grad_norm": 3.7375059127807617,
-      "learning_rate": 6.98e-05,
-      "loss": 6.9625,
-      "step": 350
-    },
-    {
-      "epoch": 0.18305084745762712,
-      "grad_norm": 3.2049381732940674,
-      "learning_rate": 7e-05,
-      "loss": 6.6007,
-      "step": 351
-    },
-    {
-      "epoch": 0.18357235984354628,
-      "grad_norm": 3.154276132583618,
-      "learning_rate": 7.02e-05,
-      "loss": 6.1258,
-      "step": 352
-    },
-    {
-      "epoch": 0.18409387222946544,
-      "grad_norm": 3.293564796447754,
-      "learning_rate": 7.04e-05,
-      "loss": 6.8757,
-      "step": 353
-    },
-    {
-      "epoch": 0.18461538461538463,
-      "grad_norm": 3.0282490253448486,
-      "learning_rate": 7.06e-05,
-      "loss": 6.8665,
-      "step": 354
-    },
-    {
-      "epoch": 0.18513689700130379,
-      "grad_norm": 3.028792142868042,
-      "learning_rate": 7.08e-05,
-      "loss": 6.9648,
-      "step": 355
-    },
-    {
-      "epoch": 0.18565840938722294,
-      "grad_norm": 3.3097164630889893,
-      "learning_rate": 7.1e-05,
-      "loss": 6.6744,
-      "step": 356
-    },
-    {
-      "epoch": 0.1861799217731421,
-      "grad_norm": 2.978569746017456,
-      "learning_rate": 7.12e-05,
-      "loss": 6.827,
-      "step": 357
-    },
-    {
-      "epoch": 0.1867014341590613,
-      "grad_norm": 2.8361692428588867,
-      "learning_rate": 7.14e-05,
-      "loss": 7.1999,
-      "step": 358
-    },
-    {
-      "epoch": 0.18722294654498045,
-      "grad_norm": 2.434253454208374,
-      "learning_rate": 7.16e-05,
-      "loss": 7.1673,
-      "step": 359
-    },
-    {
-      "epoch": 0.1877444589308996,
-      "grad_norm": 3.8715171813964844,
-      "learning_rate": 7.18e-05,
-      "loss": 6.5985,
-      "step": 360
-    },
-    {
-      "epoch": 0.18826597131681877,
-      "grad_norm": 2.6759724617004395,
-      "learning_rate": 7.2e-05,
-      "loss": 6.9218,
-      "step": 361
-    },
-    {
-      "epoch": 0.18878748370273793,
-      "grad_norm": 3.4225521087646484,
-      "learning_rate": 7.22e-05,
-      "loss": 6.2147,
-      "step": 362
-    },
-    {
-      "epoch": 0.18930899608865712,
-      "grad_norm": 3.0106003284454346,
-      "learning_rate": 7.24e-05,
-      "loss": 6.539,
-      "step": 363
-    },
-    {
-      "epoch": 0.18983050847457628,
-      "grad_norm": 2.9981253147125244,
-      "learning_rate": 7.26e-05,
-      "loss": 6.8481,
-      "step": 364
-    },
-    {
-      "epoch": 0.19035202086049544,
-      "grad_norm": 2.7975802421569824,
-      "learning_rate": 7.280000000000001e-05,
-      "loss": 6.7999,
-      "step": 365
-    },
-    {
-      "epoch": 0.1908735332464146,
-      "grad_norm": 2.4574432373046875,
-      "learning_rate": 7.3e-05,
-      "loss": 7.1035,
-      "step": 366
-    },
-    {
-      "epoch": 0.19139504563233375,
-      "grad_norm": 4.905591011047363,
-      "learning_rate": 7.32e-05,
-      "loss": 6.0549,
-      "step": 367
-    },
-    {
-      "epoch": 0.19191655801825294,
-      "grad_norm": 3.8138203620910645,
-      "learning_rate": 7.340000000000001e-05,
-      "loss": 6.8213,
-      "step": 368
-    },
-    {
-      "epoch": 0.1924380704041721,
-      "grad_norm": 3.0640335083007812,
-      "learning_rate": 7.36e-05,
-      "loss": 6.7799,
-      "step": 369
-    },
-    {
-      "epoch": 0.19295958279009126,
-      "grad_norm": 3.3823201656341553,
-      "learning_rate": 7.38e-05,
-      "loss": 6.9555,
-      "step": 370
-    },
-    {
-      "epoch": 0.19348109517601042,
-      "grad_norm": 2.893047571182251,
-      "learning_rate": 7.4e-05,
-      "loss": 6.9372,
-      "step": 371
-    },
-    {
-      "epoch": 0.1940026075619296,
-      "grad_norm": 2.6517035961151123,
-      "learning_rate": 7.42e-05,
-      "loss": 7.2126,
-      "step": 372
-    },
-    {
-      "epoch": 0.19452411994784877,
-      "grad_norm": 2.4646477699279785,
-      "learning_rate": 7.44e-05,
-      "loss": 7.114,
-      "step": 373
-    },
-    {
-      "epoch": 0.19504563233376793,
-      "grad_norm": 4.219038009643555,
-      "learning_rate": 7.46e-05,
-      "loss": 6.3893,
-      "step": 374
-    },
-    {
-      "epoch": 0.19556714471968709,
-      "grad_norm": 2.6734843254089355,
-      "learning_rate": 7.48e-05,
-      "loss": 7.0357,
-      "step": 375
-    },
-    {
-      "epoch": 0.19608865710560625,
-      "grad_norm": 3.205026626586914,
-      "learning_rate": 7.500000000000001e-05,
-      "loss": 6.4775,
-      "step": 376
-    },
-    {
-      "epoch": 0.19661016949152543,
-      "grad_norm": 2.869208812713623,
-      "learning_rate": 7.52e-05,
-      "loss": 7.0149,
-      "step": 377
-    },
-    {
-      "epoch": 0.1971316818774446,
-      "grad_norm": 3.115974187850952,
-      "learning_rate": 7.54e-05,
-      "loss": 6.8281,
-      "step": 378
-    },
-    {
-      "epoch": 0.19765319426336375,
-      "grad_norm": 2.7429001331329346,
-      "learning_rate": 7.560000000000001e-05,
-      "loss": 7.1209,
-      "step": 379
-    },
-    {
-      "epoch": 0.1981747066492829,
-      "grad_norm": 2.84698748588562,
-      "learning_rate": 7.58e-05,
-      "loss": 6.5012,
-      "step": 380
-    },
-    {
-      "epoch": 0.1986962190352021,
-      "grad_norm": 3.101663112640381,
-      "learning_rate": 7.6e-05,
-      "loss": 6.759,
-      "step": 381
-    },
-    {
-      "epoch": 0.19921773142112126,
-      "grad_norm": 2.9939825534820557,
-      "learning_rate": 7.620000000000001e-05,
-      "loss": 6.9159,
-      "step": 382
-    },
-    {
-      "epoch": 0.19973924380704042,
-      "grad_norm": 2.900312662124634,
-      "learning_rate": 7.64e-05,
-      "loss": 7.1468,
-      "step": 383
-    },
-    {
-      "epoch": 0.20026075619295958,
-      "grad_norm": 3.3308403491973877,
-      "learning_rate": 7.66e-05,
-      "loss": 6.5847,
-      "step": 384
-    },
-    {
-      "epoch": 0.20078226857887874,
-      "grad_norm": 2.6935532093048096,
-      "learning_rate": 7.680000000000001e-05,
-      "loss": 7.1005,
-      "step": 385
-    },
-    {
-      "epoch": 0.20130378096479792,
-      "grad_norm": 3.0341670513153076,
-      "learning_rate": 7.7e-05,
-      "loss": 6.5444,
-      "step": 386
-    },
-    {
-      "epoch": 0.20182529335071708,
-      "grad_norm": 2.6228854656219482,
-      "learning_rate": 7.72e-05,
-      "loss": 6.927,
-      "step": 387
-    },
-    {
-      "epoch": 0.20234680573663624,
-      "grad_norm": 2.779094696044922,
-      "learning_rate": 7.740000000000001e-05,
-      "loss": 7.0569,
-      "step": 388
-    },
-    {
-      "epoch": 0.2028683181225554,
-      "grad_norm": 3.2601938247680664,
-      "learning_rate": 7.76e-05,
-      "loss": 6.7258,
-      "step": 389
-    },
-    {
-      "epoch": 0.2033898305084746,
-      "grad_norm": 3.3600635528564453,
-      "learning_rate": 7.780000000000001e-05,
-      "loss": 6.4443,
-      "step": 390
-    },
-    {
-      "epoch": 0.20391134289439375,
-      "grad_norm": 2.924651861190796,
-      "learning_rate": 7.800000000000001e-05,
-      "loss": 7.1191,
-      "step": 391
-    },
-    {
-      "epoch": 0.2044328552803129,
-      "grad_norm": 2.5779831409454346,
-      "learning_rate": 7.82e-05,
-      "loss": 7.0077,
-      "step": 392
-    },
-    {
-      "epoch": 0.20495436766623207,
-      "grad_norm": 3.5986835956573486,
-      "learning_rate": 7.840000000000001e-05,
-      "loss": 6.9599,
-      "step": 393
-    },
-    {
-      "epoch": 0.20547588005215123,
-      "grad_norm": 3.574509859085083,
-      "learning_rate": 7.860000000000001e-05,
-      "loss": 7.0468,
-      "step": 394
-    },
-    {
-      "epoch": 0.20599739243807041,
-      "grad_norm": 3.1655123233795166,
-      "learning_rate": 7.88e-05,
-      "loss": 6.8709,
-      "step": 395
-    },
-    {
-      "epoch": 0.20651890482398957,
-      "grad_norm": 3.0388529300689697,
-      "learning_rate": 7.900000000000001e-05,
-      "loss": 7.0886,
-      "step": 396
-    },
-    {
-      "epoch": 0.20704041720990873,
-      "grad_norm": 4.3477020263671875,
-      "learning_rate": 7.920000000000001e-05,
-      "loss": 6.6419,
-      "step": 397
-    },
-    {
-      "epoch": 0.2075619295958279,
-      "grad_norm": 2.8662290573120117,
-      "learning_rate": 7.94e-05,
-      "loss": 6.9051,
-      "step": 398
-    },
-    {
-      "epoch": 0.20808344198174705,
-      "grad_norm": 4.2950758934021,
-      "learning_rate": 7.960000000000001e-05,
-      "loss": 5.8443,
-      "step": 399
-    },
-    {
-      "epoch": 0.20860495436766624,
-      "grad_norm": 3.039278745651245,
-      "learning_rate": 7.98e-05,
-      "loss": 6.596,
-      "step": 400
-    },
-    {
-      "epoch": 0.2091264667535854,
-      "grad_norm": 3.4841458797454834,
-      "learning_rate": 8e-05,
-      "loss": 5.7305,
-      "step": 401
-    },
-    {
-      "epoch": 0.20964797913950456,
-      "grad_norm": 2.7502031326293945,
-      "learning_rate": 8.020000000000001e-05,
-      "loss": 6.7217,
-      "step": 402
-    },
-    {
-      "epoch": 0.21016949152542372,
-      "grad_norm": 3.4197518825531006,
-      "learning_rate": 8.04e-05,
-      "loss": 7.1718,
-      "step": 403
-    },
-    {
-      "epoch": 0.2106910039113429,
-      "grad_norm": 3.0966851711273193,
-      "learning_rate": 8.060000000000001e-05,
-      "loss": 6.3327,
-      "step": 404
-    },
-    {
-      "epoch": 0.21121251629726207,
-      "grad_norm": 2.639968156814575,
-      "learning_rate": 8.080000000000001e-05,
-      "loss": 6.8873,
-      "step": 405
-    },
-    {
-      "epoch": 0.21173402868318122,
-      "grad_norm": 3.080145835876465,
-      "learning_rate": 8.1e-05,
-      "loss": 6.34,
-      "step": 406
-    },
-    {
-      "epoch": 0.21225554106910038,
-      "grad_norm": 3.6168289184570312,
-      "learning_rate": 8.120000000000001e-05,
-      "loss": 6.4417,
-      "step": 407
-    },
-    {
-      "epoch": 0.21277705345501954,
-      "grad_norm": 3.0011940002441406,
-      "learning_rate": 8.14e-05,
-      "loss": 6.564,
-      "step": 408
-    },
-    {
-      "epoch": 0.21329856584093873,
-      "grad_norm": 3.090268135070801,
-      "learning_rate": 8.16e-05,
-      "loss": 6.8972,
-      "step": 409
-    },
-    {
-      "epoch": 0.2138200782268579,
-      "grad_norm": 2.9102587699890137,
-      "learning_rate": 8.18e-05,
-      "loss": 6.9113,
-      "step": 410
-    },
-    {
-      "epoch": 0.21434159061277705,
-      "grad_norm": 3.6984519958496094,
-      "learning_rate": 8.2e-05,
-      "loss": 6.0702,
-      "step": 411
-    },
-    {
-      "epoch": 0.2148631029986962,
-      "grad_norm": 2.714181661605835,
-      "learning_rate": 8.22e-05,
-      "loss": 6.946,
-      "step": 412
-    },
-    {
-      "epoch": 0.2153846153846154,
-      "grad_norm": 2.511662721633911,
-      "learning_rate": 8.24e-05,
-      "loss": 7.1757,
-      "step": 413
-    },
-    {
-      "epoch": 0.21590612777053456,
-      "grad_norm": 3.234900712966919,
-      "learning_rate": 8.26e-05,
-      "loss": 6.817,
-      "step": 414
-    },
-    {
-      "epoch": 0.21642764015645372,
-      "grad_norm": 3.1684656143188477,
-      "learning_rate": 8.28e-05,
-      "loss": 6.9316,
-      "step": 415
-    },
-    {
-      "epoch": 0.21694915254237288,
-      "grad_norm": 2.9769480228424072,
-      "learning_rate": 8.3e-05,
-      "loss": 6.6332,
-      "step": 416
-    },
-    {
-      "epoch": 0.21747066492829203,
-      "grad_norm": 3.721759080886841,
-      "learning_rate": 8.32e-05,
-      "loss": 6.3789,
-      "step": 417
-    },
-    {
-      "epoch": 0.21799217731421122,
-      "grad_norm": 3.035891056060791,
-      "learning_rate": 8.34e-05,
-      "loss": 7.0388,
-      "step": 418
-    },
-    {
-      "epoch": 0.21851368970013038,
-      "grad_norm": 2.604626178741455,
-      "learning_rate": 8.36e-05,
-      "loss": 7.0204,
-      "step": 419
-    },
-    {
-      "epoch": 0.21903520208604954,
-      "grad_norm": 3.0238819122314453,
-      "learning_rate": 8.38e-05,
-      "loss": 6.7262,
-      "step": 420
-    },
-    {
-      "epoch": 0.2195567144719687,
-      "grad_norm": 2.8538432121276855,
-      "learning_rate": 8.4e-05,
-      "loss": 6.625,
-      "step": 421
-    },
-    {
-      "epoch": 0.2200782268578879,
-      "grad_norm": 3.035665273666382,
-      "learning_rate": 8.42e-05,
-      "loss": 6.7138,
-      "step": 422
-    },
-    {
-      "epoch": 0.22059973924380705,
-      "grad_norm": 2.5486512184143066,
-      "learning_rate": 8.44e-05,
-      "loss": 6.8959,
-      "step": 423
-    },
-    {
-      "epoch": 0.2211212516297262,
-      "grad_norm": 3.3886094093322754,
-      "learning_rate": 8.46e-05,
-      "loss": 6.3696,
-      "step": 424
-    },
-    {
-      "epoch": 0.22164276401564537,
-      "grad_norm": 3.423654556274414,
-      "learning_rate": 8.48e-05,
-      "loss": 7.0734,
-      "step": 425
-    },
-    {
-      "epoch": 0.22216427640156453,
-      "grad_norm": 2.735738515853882,
-      "learning_rate": 8.5e-05,
-      "loss": 6.7616,
-      "step": 426
-    },
-    {
-      "epoch": 0.2226857887874837,
-      "grad_norm": 3.713524580001831,
-      "learning_rate": 8.52e-05,
-      "loss": 6.2687,
-      "step": 427
-    },
-    {
-      "epoch": 0.22320730117340287,
-      "grad_norm": 2.945415735244751,
-      "learning_rate": 8.54e-05,
-      "loss": 6.5513,
-      "step": 428
-    },
-    {
-      "epoch": 0.22372881355932203,
-      "grad_norm": 2.5077168941497803,
-      "learning_rate": 8.560000000000001e-05,
-      "loss": 6.7392,
-      "step": 429
-    },
-    {
-      "epoch": 0.2242503259452412,
-      "grad_norm": 2.3934662342071533,
-      "learning_rate": 8.58e-05,
-      "loss": 7.0249,
-      "step": 430
-    },
-    {
-      "epoch": 0.22477183833116038,
-      "grad_norm": 2.6413142681121826,
-      "learning_rate": 8.6e-05,
-      "loss": 6.8845,
-      "step": 431
-    },
-    {
-      "epoch": 0.22529335071707954,
-      "grad_norm": 2.520956516265869,
-      "learning_rate": 8.620000000000001e-05,
-      "loss": 6.8159,
-      "step": 432
-    },
-    {
-      "epoch": 0.2258148631029987,
-      "grad_norm": 2.6290929317474365,
-      "learning_rate": 8.64e-05,
-      "loss": 7.0407,
-      "step": 433
-    },
-    {
-      "epoch": 0.22633637548891786,
-      "grad_norm": 2.630131721496582,
-      "learning_rate": 8.66e-05,
-      "loss": 6.8931,
-      "step": 434
-    },
-    {
-      "epoch": 0.22685788787483702,
-      "grad_norm": 2.8762598037719727,
-      "learning_rate": 8.680000000000001e-05,
-      "loss": 6.8662,
-      "step": 435
-    },
-    {
-      "epoch": 0.2273794002607562,
-      "grad_norm": 2.833972930908203,
-      "learning_rate": 8.7e-05,
-      "loss": 6.8259,
-      "step": 436
-    },
-    {
-      "epoch": 0.22790091264667536,
-      "grad_norm": 2.7063074111938477,
-      "learning_rate": 8.72e-05,
-      "loss": 6.5592,
-      "step": 437
-    },
-    {
-      "epoch": 0.22842242503259452,
-      "grad_norm": 2.695197820663452,
-      "learning_rate": 8.740000000000001e-05,
-      "loss": 6.7381,
-      "step": 438
-    },
-    {
-      "epoch": 0.22894393741851368,
-      "grad_norm": 3.287299394607544,
-      "learning_rate": 8.76e-05,
-      "loss": 6.3144,
-      "step": 439
-    },
-    {
-      "epoch": 0.22946544980443284,
-      "grad_norm": 2.936960220336914,
-      "learning_rate": 8.78e-05,
-      "loss": 6.7655,
-      "step": 440
-    },
-    {
-      "epoch": 0.22998696219035203,
-      "grad_norm": 3.0835437774658203,
-      "learning_rate": 8.800000000000001e-05,
-      "loss": 6.7237,
-      "step": 441
-    },
-    {
-      "epoch": 0.2305084745762712,
-      "grad_norm": 2.654698371887207,
-      "learning_rate": 8.82e-05,
-      "loss": 7.0268,
-      "step": 442
-    },
-    {
-      "epoch": 0.23102998696219035,
-      "grad_norm": 15.455745697021484,
-      "learning_rate": 8.840000000000001e-05,
-      "loss": 5.8494,
-      "step": 443
-    },
-    {
-      "epoch": 0.2315514993481095,
-      "grad_norm": 3.2970926761627197,
-      "learning_rate": 8.86e-05,
-      "loss": 6.8435,
-      "step": 444
-    },
-    {
-      "epoch": 0.2320730117340287,
-      "grad_norm": 2.7584433555603027,
-      "learning_rate": 8.88e-05,
-      "loss": 6.4046,
-      "step": 445
-    },
-    {
-      "epoch": 0.23259452411994785,
-      "grad_norm": 2.5174078941345215,
-      "learning_rate": 8.900000000000001e-05,
-      "loss": 7.0402,
-      "step": 446
-    },
-    {
-      "epoch": 0.233116036505867,
-      "grad_norm": 2.509023904800415,
-      "learning_rate": 8.92e-05,
-      "loss": 6.7754,
-      "step": 447
-    },
-    {
-      "epoch": 0.23363754889178617,
-      "grad_norm": 4.349696636199951,
-      "learning_rate": 8.94e-05,
-      "loss": 6.8015,
-      "step": 448
-    },
-    {
-      "epoch": 0.23415906127770533,
-      "grad_norm": 2.6458146572113037,
-      "learning_rate": 8.960000000000001e-05,
-      "loss": 6.7895,
-      "step": 449
-    },
-    {
-      "epoch": 0.23468057366362452,
-      "grad_norm": 2.8139069080352783,
-      "learning_rate": 8.98e-05,
-      "loss": 6.9038,
-      "step": 450
-    },
-    {
-      "epoch": 0.23520208604954368,
-      "grad_norm": 3.1928818225860596,
-      "learning_rate": 9e-05,
-      "loss": 6.5367,
-      "step": 451
-    },
-    {
-      "epoch": 0.23572359843546284,
-      "grad_norm": 4.430820941925049,
-      "learning_rate": 9.020000000000001e-05,
-      "loss": 6.6013,
-      "step": 452
-    },
-    {
-      "epoch": 0.236245110821382,
-      "grad_norm": 4.031113624572754,
-      "learning_rate": 9.04e-05,
-      "loss": 6.9254,
-      "step": 453
-    },
-    {
-      "epoch": 0.23676662320730119,
-      "grad_norm": 3.2106876373291016,
-      "learning_rate": 9.06e-05,
-      "loss": 6.5961,
-      "step": 454
-    },
-    {
-      "epoch": 0.23728813559322035,
-      "grad_norm": 4.746469497680664,
-      "learning_rate": 9.080000000000001e-05,
-      "loss": 6.6948,
-      "step": 455
-    },
-    {
-      "epoch": 0.2378096479791395,
-      "grad_norm": 3.455411672592163,
-      "learning_rate": 9.1e-05,
-      "loss": 5.6938,
-      "step": 456
-    },
-    {
-      "epoch": 0.23833116036505866,
-      "grad_norm": 3.9724597930908203,
-      "learning_rate": 9.120000000000001e-05,
-      "loss": 6.8219,
-      "step": 457
-    },
-    {
-      "epoch": 0.23885267275097782,
-      "grad_norm": 2.8081107139587402,
-      "learning_rate": 9.140000000000001e-05,
-      "loss": 6.5813,
-      "step": 458
-    },
-    {
-      "epoch": 0.239374185136897,
-      "grad_norm": 3.2344539165496826,
-      "learning_rate": 9.16e-05,
-      "loss": 6.6187,
-      "step": 459
-    },
-    {
-      "epoch": 0.23989569752281617,
-      "grad_norm": 2.544550895690918,
-      "learning_rate": 9.180000000000001e-05,
-      "loss": 6.8158,
-      "step": 460
-    },
-    {
-      "epoch": 0.24041720990873533,
-      "grad_norm": 2.41996693611145,
-      "learning_rate": 9.200000000000001e-05,
-      "loss": 6.9909,
-      "step": 461
-    },
-    {
-      "epoch": 0.2409387222946545,
-      "grad_norm": 2.891789674758911,
-      "learning_rate": 9.22e-05,
-      "loss": 6.9594,
-      "step": 462
-    },
-    {
-      "epoch": 0.24146023468057368,
-      "grad_norm": 2.933687448501587,
-      "learning_rate": 9.240000000000001e-05,
-      "loss": 6.7971,
-      "step": 463
-    },
-    {
-      "epoch": 0.24198174706649284,
-      "grad_norm": 3.0920803546905518,
-      "learning_rate": 9.260000000000001e-05,
-      "loss": 6.1003,
-      "step": 464
-    },
-    {
-      "epoch": 0.242503259452412,
-      "grad_norm": 2.788957118988037,
-      "learning_rate": 9.28e-05,
-      "loss": 6.9173,
-      "step": 465
-    },
-    {
-      "epoch": 0.24302477183833116,
-      "grad_norm": 4.545304775238037,
-      "learning_rate": 9.300000000000001e-05,
-      "loss": 6.9782,
-      "step": 466
-    },
-    {
-      "epoch": 0.24354628422425031,
-      "grad_norm": 2.7203845977783203,
-      "learning_rate": 9.320000000000002e-05,
-      "loss": 6.703,
-      "step": 467
-    },
-    {
-      "epoch": 0.2440677966101695,
-      "grad_norm": 5.148874282836914,
-      "learning_rate": 9.340000000000001e-05,
-      "loss": 6.2721,
-      "step": 468
-    },
-    {
-      "epoch": 0.24458930899608866,
-      "grad_norm": 2.7999868392944336,
-      "learning_rate": 9.360000000000001e-05,
-      "loss": 6.975,
-      "step": 469
-    },
-    {
-      "epoch": 0.24511082138200782,
-      "grad_norm": 2.686619281768799,
-      "learning_rate": 9.38e-05,
-      "loss": 6.4811,
-      "step": 470
-    },
-    {
-      "epoch": 0.24563233376792698,
-      "grad_norm": 3.537447929382324,
-      "learning_rate": 9.4e-05,
-      "loss": 6.8074,
-      "step": 471
-    },
-    {
-      "epoch": 0.24615384615384617,
-      "grad_norm": 2.357633352279663,
-      "learning_rate": 9.42e-05,
-      "loss": 6.9754,
-      "step": 472
-    },
-    {
-      "epoch": 0.24667535853976533,
-      "grad_norm": 2.735482931137085,
-      "learning_rate": 9.44e-05,
-      "loss": 6.7566,
-      "step": 473
-    },
-    {
-      "epoch": 0.2471968709256845,
-      "grad_norm": 2.3248891830444336,
-      "learning_rate": 9.46e-05,
-      "loss": 6.822,
-      "step": 474
-    },
-    {
-      "epoch": 0.24771838331160365,
-      "grad_norm": 2.6541171073913574,
-      "learning_rate": 9.48e-05,
-      "loss": 6.7979,
-      "step": 475
-    },
-    {
-      "epoch": 0.2482398956975228,
-      "grad_norm": 2.36301851272583,
-      "learning_rate": 9.5e-05,
-      "loss": 6.9476,
-      "step": 476
-    },
-    {
-      "epoch": 0.248761408083442,
-      "grad_norm": 2.5697407722473145,
-      "learning_rate": 9.52e-05,
-      "loss": 6.8588,
-      "step": 477
-    },
-    {
-      "epoch": 0.24928292046936115,
-      "grad_norm": 2.7867579460144043,
-      "learning_rate": 9.54e-05,
-      "loss": 6.6956,
-      "step": 478
-    },
-    {
-      "epoch": 0.2498044328552803,
-      "grad_norm": 2.825334072113037,
-      "learning_rate": 9.56e-05,
-      "loss": 6.7513,
-      "step": 479
-    },
-    {
-      "epoch": 0.2503259452411995,
-      "grad_norm": 3.315460681915283,
-      "learning_rate": 9.58e-05,
-      "loss": 7.0593,
-      "step": 480
-    },
-    {
-      "epoch": 0.25084745762711863,
-      "grad_norm": 2.340892791748047,
-      "learning_rate": 9.6e-05,
-      "loss": 6.6995,
-      "step": 481
-    },
-    {
-      "epoch": 0.2513689700130378,
-      "grad_norm": 3.2200984954833984,
-      "learning_rate": 9.620000000000001e-05,
-      "loss": 6.5835,
-      "step": 482
-    },
-    {
-      "epoch": 0.25189048239895695,
-      "grad_norm": 3.7848949432373047,
-      "learning_rate": 9.64e-05,
-      "loss": 6.7126,
-      "step": 483
-    },
-    {
-      "epoch": 0.25241199478487614,
-      "grad_norm": 3.250565528869629,
-      "learning_rate": 9.66e-05,
-      "loss": 6.9432,
-      "step": 484
-    },
-    {
-      "epoch": 0.2529335071707953,
-      "grad_norm": 5.082596302032471,
-      "learning_rate": 9.680000000000001e-05,
-      "loss": 5.4362,
-      "step": 485
-    },
-    {
-      "epoch": 0.25345501955671446,
-      "grad_norm": 2.3779993057250977,
-      "learning_rate": 9.7e-05,
-      "loss": 6.9548,
-      "step": 486
-    },
-    {
-      "epoch": 0.25397653194263364,
-      "grad_norm": 2.4856936931610107,
-      "learning_rate": 9.72e-05,
-      "loss": 6.8634,
-      "step": 487
-    },
-    {
-      "epoch": 0.25449804432855283,
-      "grad_norm": 3.136725425720215,
-      "learning_rate": 9.74e-05,
-      "loss": 6.5332,
-      "step": 488
-    },
-    {
-      "epoch": 0.25501955671447196,
-      "grad_norm": 3.2251977920532227,
-      "learning_rate": 9.76e-05,
-      "loss": 7.0258,
-      "step": 489
-    },
-    {
-      "epoch": 0.25554106910039115,
-      "grad_norm": 4.4574875831604,
-      "learning_rate": 9.78e-05,
-      "loss": 6.5353,
-      "step": 490
-    },
-    {
-      "epoch": 0.2560625814863103,
-      "grad_norm": 4.009222507476807,
-      "learning_rate": 9.8e-05,
-      "loss": 6.259,
-      "step": 491
-    },
-    {
-      "epoch": 0.25658409387222947,
-      "grad_norm": 3.212877035140991,
-      "learning_rate": 9.82e-05,
-      "loss": 6.6794,
-      "step": 492
-    },
-    {
-      "epoch": 0.25710560625814866,
-      "grad_norm": 2.506984233856201,
-      "learning_rate": 9.84e-05,
-      "loss": 6.8881,
-      "step": 493
-    },
-    {
-      "epoch": 0.2576271186440678,
-      "grad_norm": 3.386939287185669,
-      "learning_rate": 9.86e-05,
-      "loss": 6.4301,
-      "step": 494
-    },
-    {
-      "epoch": 0.258148631029987,
-      "grad_norm": 3.279521942138672,
-      "learning_rate": 9.88e-05,
-      "loss": 6.2215,
-      "step": 495
-    },
-    {
-      "epoch": 0.2586701434159061,
-      "grad_norm": 3.4294681549072266,
-      "learning_rate": 9.900000000000001e-05,
-      "loss": 6.8048,
-      "step": 496
-    },
-    {
-      "epoch": 0.2591916558018253,
-      "grad_norm": 2.5611634254455566,
-      "learning_rate": 9.92e-05,
-      "loss": 6.566,
-      "step": 497
-    },
-    {
-      "epoch": 0.2597131681877445,
-      "grad_norm": 2.3411877155303955,
-      "learning_rate": 9.94e-05,
-      "loss": 7.0717,
-      "step": 498
-    },
-    {
-      "epoch": 0.2602346805736636,
-      "grad_norm": 2.4214110374450684,
-      "learning_rate": 9.960000000000001e-05,
-      "loss": 6.8023,
-      "step": 499
-    },
-    {
-      "epoch": 0.2607561929595828,
-      "grad_norm": 2.475004196166992,
-      "learning_rate": 9.98e-05,
-      "loss": 6.5057,
-      "step": 500
-    },
-    {
-      "epoch": 0.26127770534550193,
-      "grad_norm": 2.73582124710083,
-      "learning_rate": 0.0001,
-      "loss": 6.7666,
-      "step": 501
-    },
-    {
-      "epoch": 0.2617992177314211,
-      "grad_norm": 2.2623517513275146,
-      "learning_rate": 9.999899497487437e-05,
-      "loss": 6.8927,
-      "step": 502
-    },
-    {
-      "epoch": 0.2623207301173403,
-      "grad_norm": 3.2750511169433594,
-      "learning_rate": 9.999798994974875e-05,
-      "loss": 6.3848,
-      "step": 503
-    },
-    {
-      "epoch": 0.26284224250325944,
-      "grad_norm": 3.193834066390991,
-      "learning_rate": 9.999698492462311e-05,
-      "loss": 6.6813,
-      "step": 504
-    },
-    {
-      "epoch": 0.2633637548891786,
-      "grad_norm": 2.648184061050415,
-      "learning_rate": 9.999597989949749e-05,
-      "loss": 6.9224,
-      "step": 505
-    },
-    {
-      "epoch": 0.26388526727509776,
-      "grad_norm": 3.534034013748169,
-      "learning_rate": 9.999497487437187e-05,
-      "loss": 6.9134,
-      "step": 506
-    },
-    {
-      "epoch": 0.26440677966101694,
-      "grad_norm": 2.204660415649414,
-      "learning_rate": 9.999396984924624e-05,
-      "loss": 7.1112,
-      "step": 507
-    },
-    {
-      "epoch": 0.26492829204693613,
-      "grad_norm": 2.4047603607177734,
-      "learning_rate": 9.999296482412061e-05,
-      "loss": 6.8362,
-      "step": 508
-    },
-    {
-      "epoch": 0.26544980443285526,
-      "grad_norm": 2.6058859825134277,
-      "learning_rate": 9.999195979899499e-05,
-      "loss": 6.9067,
-      "step": 509
-    },
-    {
-      "epoch": 0.26597131681877445,
-      "grad_norm": 2.683767795562744,
-      "learning_rate": 9.999095477386935e-05,
-      "loss": 6.5132,
-      "step": 510
-    },
-    {
-      "epoch": 0.26649282920469364,
-      "grad_norm": 2.2093727588653564,
-      "learning_rate": 9.998994974874373e-05,
-      "loss": 6.7814,
-      "step": 511
-    },
-    {
-      "epoch": 0.26701434159061277,
-      "grad_norm": 2.3152880668640137,
-      "learning_rate": 9.998894472361809e-05,
-      "loss": 6.5789,
-      "step": 512
-    },
-    {
-      "epoch": 0.26753585397653196,
-      "grad_norm": 2.3612897396087646,
-      "learning_rate": 9.998793969849247e-05,
-      "loss": 6.7686,
-      "step": 513
-    },
-    {
-      "epoch": 0.2680573663624511,
-      "grad_norm": 2.7442212104797363,
-      "learning_rate": 9.998693467336684e-05,
-      "loss": 6.708,
-      "step": 514
-    },
-    {
-      "epoch": 0.2685788787483703,
-      "grad_norm": 3.004387140274048,
-      "learning_rate": 9.99859296482412e-05,
-      "loss": 6.1397,
-      "step": 515
-    },
-    {
-      "epoch": 0.26910039113428946,
-      "grad_norm": 2.567108631134033,
-      "learning_rate": 9.998492462311558e-05,
-      "loss": 6.7375,
-      "step": 516
-    },
-    {
-      "epoch": 0.2696219035202086,
-      "grad_norm": 2.468379497528076,
-      "learning_rate": 9.998391959798996e-05,
-      "loss": 6.7837,
-      "step": 517
-    },
-    {
-      "epoch": 0.2701434159061278,
-      "grad_norm": 3.2258684635162354,
-      "learning_rate": 9.998291457286433e-05,
-      "loss": 6.682,
-      "step": 518
-    },
-    {
-      "epoch": 0.2706649282920469,
-      "grad_norm": 2.224414587020874,
-      "learning_rate": 9.99819095477387e-05,
-      "loss": 6.912,
-      "step": 519
-    },
-    {
-      "epoch": 0.2711864406779661,
-      "grad_norm": 2.5131306648254395,
-      "learning_rate": 9.998090452261308e-05,
-      "loss": 7.0706,
-      "step": 520
-    },
-    {
-      "epoch": 0.2717079530638853,
-      "grad_norm": 2.430908679962158,
-      "learning_rate": 9.997989949748744e-05,
-      "loss": 6.9863,
-      "step": 521
-    },
-    {
-      "epoch": 0.2722294654498044,
-      "grad_norm": 2.695842742919922,
-      "learning_rate": 9.997889447236182e-05,
-      "loss": 6.9889,
-      "step": 522
-    },
-    {
-      "epoch": 0.2727509778357236,
-      "grad_norm": 3.565156936645508,
-      "learning_rate": 9.997788944723618e-05,
-      "loss": 6.2835,
-      "step": 523
-    },
-    {
-      "epoch": 0.27327249022164274,
-      "grad_norm": 2.9504036903381348,
-      "learning_rate": 9.997688442211056e-05,
-      "loss": 6.8601,
-      "step": 524
-    },
-    {
-      "epoch": 0.2737940026075619,
-      "grad_norm": 2.996612548828125,
-      "learning_rate": 9.997587939698492e-05,
-      "loss": 6.6386,
-      "step": 525
-    },
-    {
-      "epoch": 0.2743155149934811,
-      "grad_norm": 2.6341590881347656,
-      "learning_rate": 9.99748743718593e-05,
-      "loss": 6.7187,
-      "step": 526
-    },
-    {
-      "epoch": 0.27483702737940024,
-      "grad_norm": 2.4681437015533447,
-      "learning_rate": 9.997386934673368e-05,
-      "loss": 6.9114,
-      "step": 527
-    },
-    {
-      "epoch": 0.27535853976531943,
-      "grad_norm": 4.2253947257995605,
-      "learning_rate": 9.997286432160804e-05,
-      "loss": 6.4639,
-      "step": 528
-    },
-    {
-      "epoch": 0.2758800521512386,
-      "grad_norm": 4.804105758666992,
-      "learning_rate": 9.997185929648242e-05,
-      "loss": 6.5992,
-      "step": 529
-    },
-    {
-      "epoch": 0.27640156453715775,
-      "grad_norm": 3.748114824295044,
-      "learning_rate": 9.997085427135679e-05,
-      "loss": 6.7112,
-      "step": 530
-    },
-    {
-      "epoch": 0.27692307692307694,
-      "grad_norm": 3.581264019012451,
-      "learning_rate": 9.996984924623116e-05,
-      "loss": 6.478,
-      "step": 531
-    },
-    {
-      "epoch": 0.27744458930899607,
-      "grad_norm": 2.9176101684570312,
-      "learning_rate": 9.996884422110553e-05,
-      "loss": 6.5075,
-      "step": 532
-    },
-    {
-      "epoch": 0.27796610169491526,
-      "grad_norm": 4.499645233154297,
-      "learning_rate": 9.99678391959799e-05,
-      "loss": 6.7193,
-      "step": 533
-    },
-    {
-      "epoch": 0.27848761408083444,
-      "grad_norm": 3.435164451599121,
-      "learning_rate": 9.996683417085427e-05,
-      "loss": 6.6621,
-      "step": 534
-    },
-    {
-      "epoch": 0.2790091264667536,
-      "grad_norm": 2.5283384323120117,
-      "learning_rate": 9.996582914572865e-05,
-      "loss": 6.8493,
-      "step": 535
-    },
-    {
-      "epoch": 0.27953063885267276,
-      "grad_norm": 2.3399932384490967,
-      "learning_rate": 9.996482412060301e-05,
-      "loss": 6.5532,
-      "step": 536
-    },
-    {
-      "epoch": 0.2800521512385919,
-      "grad_norm": 3.006356716156006,
-      "learning_rate": 9.996381909547739e-05,
-      "loss": 6.7371,
-      "step": 537
-    },
-    {
-      "epoch": 0.2805736636245111,
-      "grad_norm": 2.1820175647735596,
-      "learning_rate": 9.996281407035177e-05,
-      "loss": 6.9289,
-      "step": 538
-    },
-    {
-      "epoch": 0.28109517601043027,
-      "grad_norm": 2.4511280059814453,
-      "learning_rate": 9.996180904522615e-05,
-      "loss": 6.4717,
-      "step": 539
-    },
-    {
-      "epoch": 0.2816166883963494,
-      "grad_norm": 2.5437686443328857,
-      "learning_rate": 9.996080402010051e-05,
-      "loss": 6.5271,
-      "step": 540
-    },
-    {
-      "epoch": 0.2821382007822686,
-      "grad_norm": 2.9577114582061768,
-      "learning_rate": 9.995979899497487e-05,
-      "loss": 6.0008,
-      "step": 541
-    },
-    {
-      "epoch": 0.2826597131681877,
-      "grad_norm": 3.3158369064331055,
-      "learning_rate": 9.995879396984925e-05,
-      "loss": 6.7803,
-      "step": 542
-    },
-    {
-      "epoch": 0.2831812255541069,
-      "grad_norm": 2.789043426513672,
-      "learning_rate": 9.995778894472362e-05,
-      "loss": 6.9305,
-      "step": 543
-    },
-    {
-      "epoch": 0.2837027379400261,
-      "grad_norm": 2.564784049987793,
-      "learning_rate": 9.9956783919598e-05,
-      "loss": 6.4814,
-      "step": 544
-    },
-    {
-      "epoch": 0.2842242503259452,
-      "grad_norm": 2.3991527557373047,
-      "learning_rate": 9.995577889447236e-05,
-      "loss": 6.6684,
-      "step": 545
-    },
-    {
-      "epoch": 0.2847457627118644,
-      "grad_norm": 2.5177085399627686,
-      "learning_rate": 9.995477386934674e-05,
-      "loss": 6.8058,
-      "step": 546
-    },
-    {
-      "epoch": 0.28526727509778355,
-      "grad_norm": 2.5932178497314453,
-      "learning_rate": 9.995376884422111e-05,
-      "loss": 6.4201,
-      "step": 547
-    },
-    {
-      "epoch": 0.28578878748370273,
-      "grad_norm": 2.2884035110473633,
-      "learning_rate": 9.995276381909549e-05,
-      "loss": 6.6815,
-      "step": 548
-    },
-    {
-      "epoch": 0.2863102998696219,
-      "grad_norm": 2.2769322395324707,
-      "learning_rate": 9.995175879396986e-05,
-      "loss": 6.7661,
-      "step": 549
-    },
-    {
-      "epoch": 0.28683181225554105,
-      "grad_norm": 2.198941230773926,
-      "learning_rate": 9.995075376884423e-05,
-      "loss": 6.8029,
-      "step": 550
-    },
-    {
-      "epoch": 0.28735332464146024,
-      "grad_norm": 2.370363235473633,
-      "learning_rate": 9.99497487437186e-05,
-      "loss": 6.5167,
-      "step": 551
-    },
-    {
-      "epoch": 0.2878748370273794,
-      "grad_norm": 2.798598289489746,
-      "learning_rate": 9.994874371859298e-05,
-      "loss": 6.6494,
-      "step": 552
-    },
-    {
-      "epoch": 0.28839634941329856,
-      "grad_norm": 2.337928056716919,
-      "learning_rate": 9.994773869346734e-05,
-      "loss": 6.9835,
-      "step": 553
-    },
-    {
-      "epoch": 0.28891786179921775,
-      "grad_norm": 2.35713791847229,
-      "learning_rate": 9.994673366834172e-05,
-      "loss": 6.9039,
-      "step": 554
-    },
-    {
-      "epoch": 0.2894393741851369,
-      "grad_norm": 2.443599224090576,
-      "learning_rate": 9.994572864321608e-05,
-      "loss": 6.6804,
-      "step": 555
-    },
-    {
-      "epoch": 0.28996088657105606,
-      "grad_norm": 2.2196035385131836,
-      "learning_rate": 9.994472361809045e-05,
-      "loss": 6.9186,
-      "step": 556
-    },
-    {
-      "epoch": 0.29048239895697525,
-      "grad_norm": 2.3514981269836426,
-      "learning_rate": 9.994371859296482e-05,
-      "loss": 6.8633,
-      "step": 557
-    },
-    {
-      "epoch": 0.2910039113428944,
-      "grad_norm": 2.501081943511963,
-      "learning_rate": 9.99427135678392e-05,
-      "loss": 6.8908,
-      "step": 558
-    },
-    {
-      "epoch": 0.29152542372881357,
-      "grad_norm": 2.3073713779449463,
-      "learning_rate": 9.994170854271358e-05,
-      "loss": 6.554,
-      "step": 559
-    },
-    {
-      "epoch": 0.2920469361147327,
-      "grad_norm": 2.7027711868286133,
-      "learning_rate": 9.994070351758794e-05,
-      "loss": 6.5762,
-      "step": 560
-    },
-    {
-      "epoch": 0.2925684485006519,
-      "grad_norm": 2.681204319000244,
-      "learning_rate": 9.993969849246232e-05,
-      "loss": 6.3981,
-      "step": 561
-    },
-    {
-      "epoch": 0.2930899608865711,
-      "grad_norm": 3.6177210807800293,
-      "learning_rate": 9.993869346733669e-05,
-      "loss": 6.3199,
-      "step": 562
-    },
-    {
-      "epoch": 0.2936114732724902,
-      "grad_norm": 2.3057701587677,
-      "learning_rate": 9.993768844221106e-05,
-      "loss": 6.8121,
-      "step": 563
-    },
-    {
-      "epoch": 0.2941329856584094,
-      "grad_norm": 2.4482412338256836,
-      "learning_rate": 9.993668341708543e-05,
-      "loss": 6.8837,
-      "step": 564
-    },
-    {
-      "epoch": 0.29465449804432853,
-      "grad_norm": 2.292982816696167,
-      "learning_rate": 9.99356783919598e-05,
-      "loss": 6.5851,
-      "step": 565
-    },
-    {
-      "epoch": 0.2951760104302477,
-      "grad_norm": 2.7785377502441406,
-      "learning_rate": 9.993467336683417e-05,
-      "loss": 6.4553,
-      "step": 566
-    },
-    {
-      "epoch": 0.2956975228161669,
-      "grad_norm": 3.0232276916503906,
-      "learning_rate": 9.993366834170855e-05,
-      "loss": 6.3535,
-      "step": 567
-    },
-    {
-      "epoch": 0.29621903520208603,
-      "grad_norm": 2.3795006275177,
-      "learning_rate": 9.993266331658293e-05,
-      "loss": 6.906,
-      "step": 568
-    },
-    {
-      "epoch": 0.2967405475880052,
-      "grad_norm": 2.426237106323242,
-      "learning_rate": 9.993165829145729e-05,
-      "loss": 5.973,
-      "step": 569
-    },
-    {
-      "epoch": 0.29726205997392435,
-      "grad_norm": 2.931170701980591,
-      "learning_rate": 9.993065326633167e-05,
-      "loss": 6.9815,
-      "step": 570
-    },
-    {
-      "epoch": 0.29778357235984354,
-      "grad_norm": 2.250025987625122,
-      "learning_rate": 9.992964824120603e-05,
-      "loss": 6.7264,
-      "step": 571
-    },
-    {
-      "epoch": 0.2983050847457627,
-      "grad_norm": 2.563656806945801,
-      "learning_rate": 9.992864321608041e-05,
-      "loss": 6.875,
-      "step": 572
-    },
-    {
-      "epoch": 0.29882659713168186,
-      "grad_norm": 2.0841214656829834,
-      "learning_rate": 9.992763819095477e-05,
-      "loss": 6.9514,
-      "step": 573
-    },
-    {
-      "epoch": 0.29934810951760105,
-      "grad_norm": 2.6887941360473633,
-      "learning_rate": 9.992663316582915e-05,
-      "loss": 6.3593,
-      "step": 574
-    },
-    {
-      "epoch": 0.29986962190352023,
-      "grad_norm": 3.577542543411255,
-      "learning_rate": 9.992562814070352e-05,
-      "loss": 6.1976,
-      "step": 575
-    },
-    {
-      "epoch": 0.30039113428943937,
-      "grad_norm": 2.6426141262054443,
-      "learning_rate": 9.99246231155779e-05,
-      "loss": 6.8532,
-      "step": 576
-    },
-    {
-      "epoch": 0.30091264667535855,
-      "grad_norm": 2.703559637069702,
-      "learning_rate": 9.992361809045226e-05,
-      "loss": 6.4574,
-      "step": 577
-    },
-    {
-      "epoch": 0.3014341590612777,
-      "grad_norm": 3.2348217964172363,
-      "learning_rate": 9.992261306532664e-05,
-      "loss": 6.8975,
-      "step": 578
-    },
-    {
-      "epoch": 0.30195567144719687,
-      "grad_norm": 3.617314338684082,
-      "learning_rate": 9.992160804020101e-05,
-      "loss": 6.2764,
-      "step": 579
-    },
-    {
-      "epoch": 0.30247718383311606,
-      "grad_norm": 2.5697176456451416,
-      "learning_rate": 9.992060301507539e-05,
-      "loss": 6.4337,
-      "step": 580
-    },
-    {
-      "epoch": 0.3029986962190352,
-      "grad_norm": 2.1568777561187744,
-      "learning_rate": 9.991959798994976e-05,
-      "loss": 7.0039,
-      "step": 581
-    },
-    {
-      "epoch": 0.3035202086049544,
-      "grad_norm": 2.116344690322876,
-      "learning_rate": 9.991859296482412e-05,
-      "loss": 6.7774,
-      "step": 582
-    },
-    {
-      "epoch": 0.3040417209908735,
-      "grad_norm": 2.5044496059417725,
-      "learning_rate": 9.99175879396985e-05,
-      "loss": 6.659,
-      "step": 583
-    },
-    {
-      "epoch": 0.3045632333767927,
-      "grad_norm": 3.5272934436798096,
-      "learning_rate": 9.991658291457286e-05,
-      "loss": 5.9951,
-      "step": 584
-    },
-    {
-      "epoch": 0.3050847457627119,
-      "grad_norm": 4.261949062347412,
-      "learning_rate": 9.991557788944724e-05,
-      "loss": 5.7512,
-      "step": 585
-    },
-    {
-      "epoch": 0.305606258148631,
-      "grad_norm": 2.383537769317627,
-      "learning_rate": 9.99145728643216e-05,
-      "loss": 6.6171,
-      "step": 586
-    },
-    {
-      "epoch": 0.3061277705345502,
-      "grad_norm": 2.4009881019592285,
-      "learning_rate": 9.991356783919598e-05,
-      "loss": 6.6186,
-      "step": 587
-    },
-    {
-      "epoch": 0.30664928292046933,
-      "grad_norm": 2.8905534744262695,
-      "learning_rate": 9.991256281407035e-05,
-      "loss": 6.5925,
-      "step": 588
-    },
-    {
-      "epoch": 0.3071707953063885,
-      "grad_norm": 2.1318790912628174,
-      "learning_rate": 9.991155778894473e-05,
-      "loss": 6.8829,
-      "step": 589
-    },
-    {
-      "epoch": 0.3076923076923077,
-      "grad_norm": 3.37378191947937,
-      "learning_rate": 9.99105527638191e-05,
-      "loss": 6.314,
-      "step": 590
-    },
-    {
-      "epoch": 0.30821382007822684,
-      "grad_norm": 2.8562064170837402,
-      "learning_rate": 9.990954773869348e-05,
-      "loss": 6.285,
-      "step": 591
-    },
-    {
-      "epoch": 0.30873533246414603,
-      "grad_norm": 2.5292999744415283,
-      "learning_rate": 9.990854271356785e-05,
-      "loss": 6.816,
-      "step": 592
-    },
-    {
-      "epoch": 0.3092568448500652,
-      "grad_norm": 2.5964102745056152,
-      "learning_rate": 9.990753768844222e-05,
-      "loss": 6.6583,
-      "step": 593
-    },
-    {
-      "epoch": 0.30977835723598435,
-      "grad_norm": 2.4368226528167725,
-      "learning_rate": 9.990653266331659e-05,
-      "loss": 6.8602,
-      "step": 594
-    },
-    {
-      "epoch": 0.31029986962190353,
-      "grad_norm": 2.9984607696533203,
-      "learning_rate": 9.990552763819095e-05,
-      "loss": 6.5601,
-      "step": 595
-    },
-    {
-      "epoch": 0.31082138200782267,
-      "grad_norm": 2.356990098953247,
-      "learning_rate": 9.990452261306533e-05,
-      "loss": 6.5403,
-      "step": 596
-    },
-    {
-      "epoch": 0.31134289439374185,
-      "grad_norm": 2.809706687927246,
-      "learning_rate": 9.99035175879397e-05,
-      "loss": 6.7957,
-      "step": 597
-    },
-    {
-      "epoch": 0.31186440677966104,
-      "grad_norm": 2.398435354232788,
-      "learning_rate": 9.990251256281407e-05,
-      "loss": 6.5643,
-      "step": 598
-    },
-    {
-      "epoch": 0.3123859191655802,
-      "grad_norm": 2.753350257873535,
-      "learning_rate": 9.990150753768845e-05,
-      "loss": 6.5292,
-      "step": 599
-    },
-    {
-      "epoch": 0.31290743155149936,
-      "grad_norm": 2.7954092025756836,
-      "learning_rate": 9.990050251256283e-05,
-      "loss": 6.7809,
-      "step": 600
-    },
-    {
-      "epoch": 0.3134289439374185,
-      "grad_norm": 2.3736412525177,
-      "learning_rate": 9.989949748743719e-05,
-      "loss": 6.6319,
-      "step": 601
-    },
-    {
-      "epoch": 0.3139504563233377,
-      "grad_norm": 2.091228723526001,
-      "learning_rate": 9.989849246231157e-05,
-      "loss": 6.7365,
-      "step": 602
-    },
-    {
-      "epoch": 0.31447196870925687,
-      "grad_norm": 2.360786199569702,
-      "learning_rate": 9.989748743718593e-05,
-      "loss": 6.5564,
-      "step": 603
-    },
-    {
-      "epoch": 0.314993481095176,
-      "grad_norm": 2.7194275856018066,
-      "learning_rate": 9.989648241206031e-05,
-      "loss": 6.5311,
-      "step": 604
-    },
-    {
-      "epoch": 0.3155149934810952,
-      "grad_norm": 2.5315980911254883,
-      "learning_rate": 9.989547738693468e-05,
-      "loss": 6.8755,
-      "step": 605
-    },
-    {
-      "epoch": 0.3160365058670143,
-      "grad_norm": 2.4782521724700928,
-      "learning_rate": 9.989447236180905e-05,
-      "loss": 6.6772,
-      "step": 606
-    },
-    {
-      "epoch": 0.3165580182529335,
-      "grad_norm": 2.5762054920196533,
-      "learning_rate": 9.989346733668342e-05,
-      "loss": 6.8217,
-      "step": 607
-    },
-    {
-      "epoch": 0.3170795306388527,
-      "grad_norm": 2.8427186012268066,
-      "learning_rate": 9.989246231155778e-05,
-      "loss": 6.1232,
-      "step": 608
-    },
-    {
-      "epoch": 0.3176010430247718,
-      "grad_norm": 2.958988666534424,
-      "learning_rate": 9.989145728643216e-05,
-      "loss": 6.7649,
-      "step": 609
-    },
-    {
-      "epoch": 0.318122555410691,
-      "grad_norm": 3.5217721462249756,
-      "learning_rate": 9.989045226130654e-05,
-      "loss": 6.7988,
-      "step": 610
-    },
-    {
-      "epoch": 0.31864406779661014,
-      "grad_norm": 2.9620628356933594,
-      "learning_rate": 9.988944723618092e-05,
-      "loss": 6.6146,
-      "step": 611
-    },
-    {
-      "epoch": 0.31916558018252933,
-      "grad_norm": 2.283308267593384,
-      "learning_rate": 9.988844221105528e-05,
-      "loss": 6.7742,
-      "step": 612
-    },
-    {
-      "epoch": 0.3196870925684485,
-      "grad_norm": 2.426337957382202,
-      "learning_rate": 9.988743718592966e-05,
-      "loss": 6.5583,
-      "step": 613
-    },
-    {
-      "epoch": 0.32020860495436765,
-      "grad_norm": 2.8968324661254883,
-      "learning_rate": 9.988643216080402e-05,
-      "loss": 6.9948,
-      "step": 614
-    },
-    {
-      "epoch": 0.32073011734028684,
-      "grad_norm": 2.178039073944092,
-      "learning_rate": 9.98854271356784e-05,
-      "loss": 6.5209,
-      "step": 615
-    },
-    {
-      "epoch": 0.321251629726206,
-      "grad_norm": 3.11246919631958,
-      "learning_rate": 9.988442211055276e-05,
-      "loss": 6.6784,
-      "step": 616
-    },
-    {
-      "epoch": 0.32177314211212515,
-      "grad_norm": 2.2160632610321045,
-      "learning_rate": 9.988341708542714e-05,
-      "loss": 6.8533,
-      "step": 617
-    },
-    {
-      "epoch": 0.32229465449804434,
-      "grad_norm": 2.8915889263153076,
-      "learning_rate": 9.98824120603015e-05,
-      "loss": 6.2564,
-      "step": 618
-    },
-    {
-      "epoch": 0.3228161668839635,
-      "grad_norm": 2.2910103797912598,
-      "learning_rate": 9.988140703517588e-05,
-      "loss": 6.8245,
-      "step": 619
-    },
-    {
-      "epoch": 0.32333767926988266,
-      "grad_norm": 5.047464370727539,
-      "learning_rate": 9.988040201005026e-05,
-      "loss": 5.2356,
-      "step": 620
-    },
-    {
-      "epoch": 0.32385919165580185,
-      "grad_norm": 3.101825475692749,
-      "learning_rate": 9.987939698492463e-05,
-      "loss": 6.2499,
-      "step": 621
-    },
-    {
-      "epoch": 0.324380704041721,
-      "grad_norm": 2.597949981689453,
-      "learning_rate": 9.9878391959799e-05,
-      "loss": 6.8516,
-      "step": 622
-    },
-    {
-      "epoch": 0.32490221642764017,
-      "grad_norm": 2.2827794551849365,
-      "learning_rate": 9.987738693467337e-05,
-      "loss": 6.699,
-      "step": 623
-    },
-    {
-      "epoch": 0.3254237288135593,
-      "grad_norm": 2.6016316413879395,
-      "learning_rate": 9.987638190954775e-05,
-      "loss": 6.6628,
-      "step": 624
-    },
-    {
-      "epoch": 0.3259452411994785,
-      "grad_norm": 2.5033156871795654,
-      "learning_rate": 9.987537688442211e-05,
-      "loss": 6.3799,
-      "step": 625
-    },
-    {
-      "epoch": 0.3264667535853977,
-      "grad_norm": 2.48293399810791,
-      "learning_rate": 9.987437185929649e-05,
-      "loss": 6.6465,
-      "step": 626
-    },
-    {
-      "epoch": 0.3269882659713168,
-      "grad_norm": 3.0237913131713867,
-      "learning_rate": 9.987336683417085e-05,
-      "loss": 6.106,
-      "step": 627
-    },
-    {
-      "epoch": 0.327509778357236,
-      "grad_norm": 2.6413562297821045,
-      "learning_rate": 9.987236180904523e-05,
-      "loss": 6.5199,
-      "step": 628
-    },
-    {
-      "epoch": 0.3280312907431551,
-      "grad_norm": 2.531050682067871,
-      "learning_rate": 9.98713567839196e-05,
-      "loss": 6.4822,
-      "step": 629
-    },
-    {
-      "epoch": 0.3285528031290743,
-      "grad_norm": 2.211895704269409,
-      "learning_rate": 9.987035175879397e-05,
-      "loss": 6.417,
-      "step": 630
-    },
-    {
-      "epoch": 0.3290743155149935,
-      "grad_norm": 2.857788562774658,
-      "learning_rate": 9.986934673366835e-05,
-      "loss": 6.3971,
-      "step": 631
-    },
-    {
-      "epoch": 0.32959582790091263,
-      "grad_norm": 2.409496784210205,
-      "learning_rate": 9.986834170854273e-05,
-      "loss": 6.6121,
-      "step": 632
-    },
-    {
-      "epoch": 0.3301173402868318,
-      "grad_norm": 3.8872246742248535,
-      "learning_rate": 9.986733668341709e-05,
-      "loss": 5.9356,
-      "step": 633
-    },
-    {
-      "epoch": 0.330638852672751,
-      "grad_norm": 3.0166659355163574,
-      "learning_rate": 9.986633165829146e-05,
-      "loss": 6.4228,
-      "step": 634
-    },
-    {
-      "epoch": 0.33116036505867014,
-      "grad_norm": 2.2669386863708496,
-      "learning_rate": 9.986532663316583e-05,
-      "loss": 6.3234,
-      "step": 635
-    },
-    {
-      "epoch": 0.3316818774445893,
-      "grad_norm": 2.7591681480407715,
-      "learning_rate": 9.98643216080402e-05,
-      "loss": 6.3055,
-      "step": 636
-    },
-    {
-      "epoch": 0.33220338983050846,
-      "grad_norm": 2.8966269493103027,
-      "learning_rate": 9.986331658291458e-05,
-      "loss": 6.2917,
-      "step": 637
-    },
-    {
-      "epoch": 0.33272490221642764,
-      "grad_norm": 2.310699462890625,
-      "learning_rate": 9.986231155778894e-05,
-      "loss": 6.4535,
-      "step": 638
-    },
-    {
-      "epoch": 0.33324641460234683,
-      "grad_norm": 2.1760547161102295,
-      "learning_rate": 9.986130653266332e-05,
-      "loss": 6.4803,
-      "step": 639
-    },
-    {
-      "epoch": 0.33376792698826596,
-      "grad_norm": 2.1403820514678955,
-      "learning_rate": 9.98603015075377e-05,
-      "loss": 6.9452,
-      "step": 640
-    },
-    {
-      "epoch": 0.33428943937418515,
-      "grad_norm": 2.1203603744506836,
-      "learning_rate": 9.985929648241207e-05,
-      "loss": 6.944,
-      "step": 641
-    },
-    {
-      "epoch": 0.3348109517601043,
-      "grad_norm": 2.634228467941284,
-      "learning_rate": 9.985829145728644e-05,
-      "loss": 6.2601,
-      "step": 642
-    },
-    {
-      "epoch": 0.33533246414602347,
-      "grad_norm": 2.673799753189087,
-      "learning_rate": 9.985728643216082e-05,
-      "loss": 6.4082,
-      "step": 643
-    },
-    {
-      "epoch": 0.33585397653194266,
-      "grad_norm": 2.0828588008880615,
-      "learning_rate": 9.985628140703518e-05,
-      "loss": 6.593,
-      "step": 644
-    },
-    {
-      "epoch": 0.3363754889178618,
-      "grad_norm": 2.9630815982818604,
-      "learning_rate": 9.985527638190956e-05,
-      "loss": 6.3006,
-      "step": 645
-    },
-    {
-      "epoch": 0.336897001303781,
-      "grad_norm": 2.163931369781494,
-      "learning_rate": 9.985427135678392e-05,
-      "loss": 6.8087,
-      "step": 646
-    },
-    {
-      "epoch": 0.3374185136897001,
-      "grad_norm": 2.118283987045288,
-      "learning_rate": 9.98532663316583e-05,
-      "loss": 6.6582,
-      "step": 647
-    },
-    {
-      "epoch": 0.3379400260756193,
-      "grad_norm": 2.1185507774353027,
-      "learning_rate": 9.985226130653266e-05,
-      "loss": 6.5589,
-      "step": 648
-    },
-    {
-      "epoch": 0.3384615384615385,
-      "grad_norm": 2.769584894180298,
-      "learning_rate": 9.985125628140703e-05,
-      "loss": 6.2612,
-      "step": 649
-    },
-    {
-      "epoch": 0.3389830508474576,
-      "grad_norm": 2.097851276397705,
-      "learning_rate": 9.98502512562814e-05,
-      "loss": 6.5514,
-      "step": 650
-    },
-    {
-      "epoch": 0.3395045632333768,
-      "grad_norm": 3.0042083263397217,
-      "learning_rate": 9.984924623115578e-05,
-      "loss": 6.6406,
-      "step": 651
-    },
-    {
-      "epoch": 0.34002607561929593,
-      "grad_norm": 2.238783359527588,
-      "learning_rate": 9.984824120603016e-05,
-      "loss": 6.4638,
-      "step": 652
-    },
-    {
-      "epoch": 0.3405475880052151,
-      "grad_norm": 2.270066261291504,
-      "learning_rate": 9.984723618090453e-05,
-      "loss": 6.2751,
-      "step": 653
-    },
-    {
-      "epoch": 0.3410691003911343,
-      "grad_norm": 2.1098685264587402,
-      "learning_rate": 9.98462311557789e-05,
-      "loss": 6.5967,
-      "step": 654
-    },
-    {
-      "epoch": 0.34159061277705344,
-      "grad_norm": 2.4665894508361816,
-      "learning_rate": 9.984522613065327e-05,
-      "loss": 5.9501,
-      "step": 655
-    },
-    {
-      "epoch": 0.3421121251629726,
-      "grad_norm": 2.433864116668701,
-      "learning_rate": 9.984422110552765e-05,
-      "loss": 6.4734,
-      "step": 656
-    },
-    {
-      "epoch": 0.3426336375488918,
-      "grad_norm": 2.2921559810638428,
-      "learning_rate": 9.984321608040201e-05,
-      "loss": 6.5201,
-      "step": 657
-    },
-    {
-      "epoch": 0.34315514993481094,
-      "grad_norm": 1.889911413192749,
-      "learning_rate": 9.984221105527639e-05,
-      "loss": 6.8002,
-      "step": 658
-    },
-    {
-      "epoch": 0.34367666232073013,
-      "grad_norm": 3.1400258541107178,
-      "learning_rate": 9.984120603015075e-05,
-      "loss": 6.0453,
-      "step": 659
-    },
-    {
-      "epoch": 0.34419817470664926,
-      "grad_norm": 2.0831573009490967,
-      "learning_rate": 9.984020100502513e-05,
-      "loss": 6.637,
-      "step": 660
-    },
-    {
-      "epoch": 0.34471968709256845,
-      "grad_norm": 2.1338939666748047,
-      "learning_rate": 9.983919597989951e-05,
-      "loss": 6.1151,
-      "step": 661
-    },
-    {
-      "epoch": 0.34524119947848764,
-      "grad_norm": 2.0769996643066406,
-      "learning_rate": 9.983819095477387e-05,
-      "loss": 6.8921,
-      "step": 662
-    },
-    {
-      "epoch": 0.34576271186440677,
-      "grad_norm": 2.5557219982147217,
-      "learning_rate": 9.983718592964825e-05,
-      "loss": 6.5197,
-      "step": 663
-    },
-    {
-      "epoch": 0.34628422425032596,
-      "grad_norm": 3.2359161376953125,
-      "learning_rate": 9.983618090452261e-05,
-      "loss": 6.3139,
-      "step": 664
-    },
-    {
-      "epoch": 0.3468057366362451,
-      "grad_norm": 2.768599510192871,
-      "learning_rate": 9.983517587939699e-05,
-      "loss": 5.7752,
-      "step": 665
-    },
-    {
-      "epoch": 0.3473272490221643,
-      "grad_norm": 2.582446575164795,
-      "learning_rate": 9.983417085427136e-05,
-      "loss": 6.3778,
-      "step": 666
-    },
-    {
-      "epoch": 0.34784876140808346,
-      "grad_norm": 2.22094988822937,
-      "learning_rate": 9.983316582914573e-05,
-      "loss": 6.5065,
-      "step": 667
-    },
-    {
-      "epoch": 0.3483702737940026,
-      "grad_norm": 2.9147276878356934,
-      "learning_rate": 9.98321608040201e-05,
-      "loss": 6.1764,
-      "step": 668
-    },
-    {
-      "epoch": 0.3488917861799218,
-      "grad_norm": 2.245464563369751,
-      "learning_rate": 9.983115577889448e-05,
-      "loss": 6.3914,
-      "step": 669
-    },
-    {
-      "epoch": 0.3494132985658409,
-      "grad_norm": 2.8328449726104736,
-      "learning_rate": 9.983015075376884e-05,
-      "loss": 5.925,
-      "step": 670
-    },
-    {
-      "epoch": 0.3499348109517601,
-      "grad_norm": 2.521390676498413,
-      "learning_rate": 9.982914572864322e-05,
-      "loss": 6.3792,
-      "step": 671
-    },
-    {
-      "epoch": 0.3504563233376793,
-      "grad_norm": 2.3779420852661133,
-      "learning_rate": 9.98281407035176e-05,
-      "loss": 6.1934,
-      "step": 672
-    },
-    {
-      "epoch": 0.3509778357235984,
-      "grad_norm": 2.066786289215088,
-      "learning_rate": 9.982713567839197e-05,
-      "loss": 6.9087,
-      "step": 673
-    },
-    {
-      "epoch": 0.3514993481095176,
-      "grad_norm": 2.0983738899230957,
-      "learning_rate": 9.982613065326634e-05,
-      "loss": 6.6262,
-      "step": 674
-    },
-    {
-      "epoch": 0.3520208604954368,
-      "grad_norm": 1.9429088830947876,
-      "learning_rate": 9.98251256281407e-05,
-      "loss": 6.483,
-      "step": 675
-    },
-    {
-      "epoch": 0.3525423728813559,
-      "grad_norm": 2.257582426071167,
-      "learning_rate": 9.982412060301508e-05,
-      "loss": 6.5038,
-      "step": 676
-    },
-    {
-      "epoch": 0.3530638852672751,
-      "grad_norm": 2.220712423324585,
-      "learning_rate": 9.982311557788945e-05,
-      "loss": 6.8907,
-      "step": 677
-    },
-    {
-      "epoch": 0.35358539765319424,
-      "grad_norm": 2.099017381668091,
-      "learning_rate": 9.982211055276382e-05,
-      "loss": 6.6702,
-      "step": 678
-    },
-    {
-      "epoch": 0.35410691003911343,
-      "grad_norm": 2.3172242641448975,
-      "learning_rate": 9.982110552763819e-05,
-      "loss": 6.9045,
-      "step": 679
-    },
-    {
-      "epoch": 0.3546284224250326,
-      "grad_norm": 2.212599277496338,
-      "learning_rate": 9.982010050251257e-05,
-      "loss": 6.4172,
-      "step": 680
-    },
-    {
-      "epoch": 0.35514993481095175,
-      "grad_norm": 2.87436580657959,
-      "learning_rate": 9.981909547738694e-05,
-      "loss": 6.7151,
-      "step": 681
-    },
-    {
-      "epoch": 0.35567144719687094,
-      "grad_norm": 2.53831148147583,
-      "learning_rate": 9.981809045226132e-05,
-      "loss": 6.4728,
-      "step": 682
-    },
-    {
-      "epoch": 0.35619295958279007,
-      "grad_norm": 2.370199203491211,
-      "learning_rate": 9.981708542713569e-05,
-      "loss": 6.7124,
-      "step": 683
-    },
-    {
-      "epoch": 0.35671447196870926,
-      "grad_norm": 2.186441421508789,
-      "learning_rate": 9.981608040201006e-05,
-      "loss": 6.4603,
-      "step": 684
-    },
-    {
-      "epoch": 0.35723598435462844,
-      "grad_norm": 2.382138252258301,
-      "learning_rate": 9.981507537688443e-05,
-      "loss": 6.3521,
-      "step": 685
-    },
-    {
-      "epoch": 0.3577574967405476,
-      "grad_norm": 2.411463737487793,
-      "learning_rate": 9.98140703517588e-05,
-      "loss": 6.6357,
-      "step": 686
-    },
-    {
-      "epoch": 0.35827900912646676,
-      "grad_norm": 2.4420485496520996,
-      "learning_rate": 9.981306532663317e-05,
-      "loss": 6.5282,
-      "step": 687
-    },
-    {
-      "epoch": 0.3588005215123859,
-      "grad_norm": 1.8494997024536133,
-      "learning_rate": 9.981206030150753e-05,
-      "loss": 6.8825,
-      "step": 688
-    },
-    {
-      "epoch": 0.3593220338983051,
-      "grad_norm": 1.9439504146575928,
-      "learning_rate": 9.981105527638191e-05,
-      "loss": 6.3554,
-      "step": 689
-    },
-    {
-      "epoch": 0.35984354628422427,
-      "grad_norm": 3.761709690093994,
-      "learning_rate": 9.981005025125628e-05,
-      "loss": 6.1133,
-      "step": 690
-    },
-    {
-      "epoch": 0.3603650586701434,
-      "grad_norm": 2.7955455780029297,
-      "learning_rate": 9.980904522613065e-05,
-      "loss": 6.2528,
-      "step": 691
-    },
-    {
-      "epoch": 0.3608865710560626,
-      "grad_norm": 3.0165700912475586,
-      "learning_rate": 9.980804020100503e-05,
-      "loss": 6.4799,
-      "step": 692
-    },
-    {
-      "epoch": 0.3614080834419817,
-      "grad_norm": 3.0634753704071045,
-      "learning_rate": 9.980703517587941e-05,
-      "loss": 6.5751,
-      "step": 693
-    },
-    {
-      "epoch": 0.3619295958279009,
-      "grad_norm": 2.1592202186584473,
-      "learning_rate": 9.980603015075377e-05,
-      "loss": 6.7329,
-      "step": 694
-    },
-    {
-      "epoch": 0.3624511082138201,
-      "grad_norm": 2.629218101501465,
-      "learning_rate": 9.980502512562815e-05,
-      "loss": 6.6906,
-      "step": 695
-    },
-    {
-      "epoch": 0.3629726205997392,
-      "grad_norm": 2.9310076236724854,
-      "learning_rate": 9.980402010050252e-05,
-      "loss": 6.6852,
-      "step": 696
-    },
-    {
-      "epoch": 0.3634941329856584,
-      "grad_norm": 2.3942654132843018,
-      "learning_rate": 9.98030150753769e-05,
-      "loss": 6.7677,
-      "step": 697
-    },
-    {
-      "epoch": 0.3640156453715776,
-      "grad_norm": 2.7797329425811768,
-      "learning_rate": 9.980201005025126e-05,
-      "loss": 6.842,
-      "step": 698
-    },
-    {
-      "epoch": 0.36453715775749673,
-      "grad_norm": 2.6364879608154297,
-      "learning_rate": 9.980100502512564e-05,
-      "loss": 6.7427,
-      "step": 699
-    },
-    {
-      "epoch": 0.3650586701434159,
-      "grad_norm": 2.9886577129364014,
-      "learning_rate": 9.98e-05,
-      "loss": 6.1343,
-      "step": 700
-    },
-    {
-      "epoch": 0.36558018252933505,
-      "grad_norm": 2.7793991565704346,
-      "learning_rate": 9.979899497487438e-05,
-      "loss": 6.3457,
-      "step": 701
-    },
-    {
-      "epoch": 0.36610169491525424,
-      "grad_norm": 2.1975605487823486,
-      "learning_rate": 9.979798994974876e-05,
-      "loss": 6.2639,
-      "step": 702
-    },
-    {
-      "epoch": 0.3666232073011734,
-      "grad_norm": 2.404585838317871,
-      "learning_rate": 9.979698492462312e-05,
-      "loss": 6.3425,
-      "step": 703
-    },
-    {
-      "epoch": 0.36714471968709256,
-      "grad_norm": 2.1477532386779785,
-      "learning_rate": 9.97959798994975e-05,
-      "loss": 6.8087,
-      "step": 704
-    },
-    {
-      "epoch": 0.36766623207301175,
-      "grad_norm": 1.9459115266799927,
-      "learning_rate": 9.979497487437186e-05,
-      "loss": 6.6319,
-      "step": 705
-    },
-    {
-      "epoch": 0.3681877444589309,
-      "grad_norm": 2.8141002655029297,
-      "learning_rate": 9.979396984924624e-05,
-      "loss": 6.683,
-      "step": 706
-    },
-    {
-      "epoch": 0.36870925684485006,
-      "grad_norm": 2.0927507877349854,
-      "learning_rate": 9.97929648241206e-05,
-      "loss": 6.3955,
-      "step": 707
-    },
-    {
-      "epoch": 0.36923076923076925,
-      "grad_norm": 2.4862911701202393,
-      "learning_rate": 9.979195979899498e-05,
-      "loss": 6.1181,
-      "step": 708
-    },
-    {
-      "epoch": 0.3697522816166884,
-      "grad_norm": 2.4091734886169434,
-      "learning_rate": 9.979095477386935e-05,
-      "loss": 6.4157,
-      "step": 709
-    },
-    {
-      "epoch": 0.37027379400260757,
-      "grad_norm": 2.641993761062622,
-      "learning_rate": 9.978994974874372e-05,
-      "loss": 5.8782,
-      "step": 710
-    },
-    {
-      "epoch": 0.3707953063885267,
-      "grad_norm": 2.108973264694214,
-      "learning_rate": 9.978894472361809e-05,
-      "loss": 6.7079,
-      "step": 711
-    },
-    {
-      "epoch": 0.3713168187744459,
-      "grad_norm": 2.3215339183807373,
-      "learning_rate": 9.978793969849247e-05,
-      "loss": 6.3592,
-      "step": 712
-    },
-    {
-      "epoch": 0.3718383311603651,
-      "grad_norm": 2.543102502822876,
-      "learning_rate": 9.978693467336684e-05,
-      "loss": 6.1239,
-      "step": 713
-    },
-    {
-      "epoch": 0.3723598435462842,
-      "grad_norm": 2.2113935947418213,
-      "learning_rate": 9.978592964824121e-05,
-      "loss": 6.6394,
-      "step": 714
-    },
-    {
-      "epoch": 0.3728813559322034,
-      "grad_norm": 2.406519651412964,
-      "learning_rate": 9.978492462311559e-05,
-      "loss": 6.7071,
-      "step": 715
-    },
-    {
-      "epoch": 0.3734028683181226,
-      "grad_norm": 2.3539154529571533,
-      "learning_rate": 9.978391959798995e-05,
-      "loss": 6.1004,
-      "step": 716
-    },
-    {
-      "epoch": 0.3739243807040417,
-      "grad_norm": 2.1021957397460938,
-      "learning_rate": 9.978291457286433e-05,
-      "loss": 6.8711,
-      "step": 717
-    },
-    {
-      "epoch": 0.3744458930899609,
-      "grad_norm": 2.0883538722991943,
-      "learning_rate": 9.978190954773869e-05,
-      "loss": 6.8728,
-      "step": 718
-    },
-    {
-      "epoch": 0.37496740547588003,
-      "grad_norm": 2.173269748687744,
-      "learning_rate": 9.978090452261307e-05,
-      "loss": 6.5409,
-      "step": 719
-    },
-    {
-      "epoch": 0.3754889178617992,
-      "grad_norm": 2.509430408477783,
-      "learning_rate": 9.977989949748743e-05,
-      "loss": 6.9534,
-      "step": 720
-    },
-    {
-      "epoch": 0.3760104302477184,
-      "grad_norm": 2.534581184387207,
-      "learning_rate": 9.977889447236181e-05,
-      "loss": 6.1659,
-      "step": 721
-    },
-    {
-      "epoch": 0.37653194263363754,
-      "grad_norm": 2.591604709625244,
-      "learning_rate": 9.977788944723619e-05,
-      "loss": 6.6487,
-      "step": 722
-    },
-    {
-      "epoch": 0.3770534550195567,
-      "grad_norm": 2.5615108013153076,
-      "learning_rate": 9.977688442211057e-05,
-      "loss": 6.3818,
-      "step": 723
-    },
-    {
-      "epoch": 0.37757496740547586,
-      "grad_norm": 2.3022994995117188,
-      "learning_rate": 9.977587939698493e-05,
-      "loss": 6.3038,
-      "step": 724
-    },
-    {
-      "epoch": 0.37809647979139505,
-      "grad_norm": 1.9913159608840942,
-      "learning_rate": 9.977487437185931e-05,
-      "loss": 6.8417,
-      "step": 725
-    },
-    {
-      "epoch": 0.37861799217731423,
-      "grad_norm": 2.830613136291504,
-      "learning_rate": 9.977386934673367e-05,
-      "loss": 6.3994,
-      "step": 726
-    },
-    {
-      "epoch": 0.37913950456323336,
-      "grad_norm": 2.2367968559265137,
-      "learning_rate": 9.977286432160805e-05,
-      "loss": 5.9519,
-      "step": 727
-    },
-    {
-      "epoch": 0.37966101694915255,
-      "grad_norm": 2.3432517051696777,
-      "learning_rate": 9.977185929648242e-05,
-      "loss": 6.4948,
-      "step": 728
-    },
-    {
-      "epoch": 0.3801825293350717,
-      "grad_norm": 2.349743127822876,
-      "learning_rate": 9.977085427135678e-05,
-      "loss": 6.4056,
-      "step": 729
-    },
-    {
-      "epoch": 0.38070404172099087,
-      "grad_norm": 1.8101089000701904,
-      "learning_rate": 9.976984924623116e-05,
-      "loss": 6.5163,
-      "step": 730
-    },
-    {
-      "epoch": 0.38122555410691006,
-      "grad_norm": 1.9853712320327759,
-      "learning_rate": 9.976884422110552e-05,
-      "loss": 6.7933,
-      "step": 731
-    },
-    {
-      "epoch": 0.3817470664928292,
-      "grad_norm": 2.36020827293396,
-      "learning_rate": 9.97678391959799e-05,
-      "loss": 6.0693,
-      "step": 732
-    },
-    {
-      "epoch": 0.3822685788787484,
-      "grad_norm": 2.329319477081299,
-      "learning_rate": 9.976683417085428e-05,
-      "loss": 6.4357,
-      "step": 733
-    },
-    {
-      "epoch": 0.3827900912646675,
-      "grad_norm": 2.8170721530914307,
-      "learning_rate": 9.976582914572866e-05,
-      "loss": 6.4844,
-      "step": 734
-    },
-    {
-      "epoch": 0.3833116036505867,
-      "grad_norm": 2.363882303237915,
-      "learning_rate": 9.976482412060302e-05,
-      "loss": 6.8889,
-      "step": 735
-    },
-    {
-      "epoch": 0.3838331160365059,
-      "grad_norm": 2.246117115020752,
-      "learning_rate": 9.97638190954774e-05,
-      "loss": 6.4265,
-      "step": 736
-    },
-    {
-      "epoch": 0.384354628422425,
-      "grad_norm": 2.067037343978882,
-      "learning_rate": 9.976281407035176e-05,
-      "loss": 6.5094,
-      "step": 737
-    },
-    {
-      "epoch": 0.3848761408083442,
-      "grad_norm": 2.13342547416687,
-      "learning_rate": 9.976180904522614e-05,
-      "loss": 6.6682,
-      "step": 738
-    },
-    {
-      "epoch": 0.3853976531942634,
-      "grad_norm": 2.1345455646514893,
-      "learning_rate": 9.97608040201005e-05,
-      "loss": 6.6607,
-      "step": 739
-    },
-    {
-      "epoch": 0.3859191655801825,
-      "grad_norm": 2.126084089279175,
-      "learning_rate": 9.975979899497488e-05,
-      "loss": 6.5625,
-      "step": 740
-    },
-    {
-      "epoch": 0.3864406779661017,
-      "grad_norm": 2.372014284133911,
-      "learning_rate": 9.975879396984925e-05,
-      "loss": 6.2901,
-      "step": 741
-    },
-    {
-      "epoch": 0.38696219035202084,
-      "grad_norm": 2.119297742843628,
-      "learning_rate": 9.975778894472362e-05,
-      "loss": 6.8048,
-      "step": 742
-    },
-    {
-      "epoch": 0.38748370273794003,
-      "grad_norm": 2.1009063720703125,
-      "learning_rate": 9.9756783919598e-05,
-      "loss": 6.3845,
-      "step": 743
-    },
-    {
-      "epoch": 0.3880052151238592,
-      "grad_norm": 2.2610244750976562,
-      "learning_rate": 9.975577889447237e-05,
-      "loss": 6.6487,
-      "step": 744
-    },
-    {
-      "epoch": 0.38852672750977835,
-      "grad_norm": 2.0792105197906494,
-      "learning_rate": 9.975477386934674e-05,
-      "loss": 6.3372,
-      "step": 745
-    },
-    {
-      "epoch": 0.38904823989569753,
-      "grad_norm": 3.173232316970825,
-      "learning_rate": 9.975376884422111e-05,
-      "loss": 6.1376,
-      "step": 746
-    },
-    {
-      "epoch": 0.38956975228161667,
-      "grad_norm": 2.279728651046753,
-      "learning_rate": 9.975276381909549e-05,
-      "loss": 6.2877,
-      "step": 747
-    },
-    {
-      "epoch": 0.39009126466753585,
-      "grad_norm": 2.0728516578674316,
-      "learning_rate": 9.975175879396985e-05,
-      "loss": 6.6308,
-      "step": 748
-    },
-    {
-      "epoch": 0.39061277705345504,
-      "grad_norm": 2.2767817974090576,
-      "learning_rate": 9.975075376884423e-05,
-      "loss": 6.4936,
-      "step": 749
-    },
-    {
-      "epoch": 0.39113428943937417,
-      "grad_norm": 2.400374412536621,
-      "learning_rate": 9.974974874371859e-05,
-      "loss": 6.607,
-      "step": 750
-    },
-    {
-      "epoch": 0.39165580182529336,
-      "grad_norm": 2.1324334144592285,
-      "learning_rate": 9.974874371859297e-05,
-      "loss": 6.1019,
-      "step": 751
-    },
-    {
-      "epoch": 0.3921773142112125,
-      "grad_norm": 2.673335075378418,
-      "learning_rate": 9.974773869346734e-05,
-      "loss": 6.3559,
-      "step": 752
-    },
-    {
-      "epoch": 0.3926988265971317,
-      "grad_norm": 2.1924960613250732,
-      "learning_rate": 9.974673366834171e-05,
-      "loss": 6.1355,
-      "step": 753
-    },
-    {
-      "epoch": 0.39322033898305087,
-      "grad_norm": 3.365873098373413,
-      "learning_rate": 9.974572864321609e-05,
-      "loss": 5.9735,
-      "step": 754
-    },
-    {
-      "epoch": 0.39374185136897,
-      "grad_norm": 2.3410911560058594,
-      "learning_rate": 9.974472361809046e-05,
-      "loss": 6.4903,
-      "step": 755
-    },
-    {
-      "epoch": 0.3942633637548892,
-      "grad_norm": 2.600425958633423,
-      "learning_rate": 9.974371859296483e-05,
-      "loss": 6.5502,
-      "step": 756
-    },
-    {
-      "epoch": 0.39478487614080837,
-      "grad_norm": 2.579627513885498,
-      "learning_rate": 9.97427135678392e-05,
-      "loss": 6.8163,
-      "step": 757
-    },
-    {
-      "epoch": 0.3953063885267275,
-      "grad_norm": 2.5041720867156982,
-      "learning_rate": 9.974170854271358e-05,
-      "loss": 6.0078,
-      "step": 758
-    },
-    {
-      "epoch": 0.3958279009126467,
-      "grad_norm": 2.5352189540863037,
-      "learning_rate": 9.974070351758794e-05,
-      "loss": 6.7737,
-      "step": 759
-    },
-    {
-      "epoch": 0.3963494132985658,
-      "grad_norm": 2.391979455947876,
-      "learning_rate": 9.973969849246232e-05,
-      "loss": 6.0156,
-      "step": 760
-    },
-    {
-      "epoch": 0.396870925684485,
-      "grad_norm": 2.4471120834350586,
-      "learning_rate": 9.973869346733668e-05,
-      "loss": 6.498,
-      "step": 761
-    },
-    {
-      "epoch": 0.3973924380704042,
-      "grad_norm": 2.259601593017578,
-      "learning_rate": 9.973768844221106e-05,
-      "loss": 6.6708,
-      "step": 762
-    },
-    {
-      "epoch": 0.39791395045632333,
-      "grad_norm": 2.0752549171447754,
-      "learning_rate": 9.973668341708542e-05,
-      "loss": 6.6007,
-      "step": 763
-    },
-    {
-      "epoch": 0.3984354628422425,
-      "grad_norm": 1.9034416675567627,
-      "learning_rate": 9.97356783919598e-05,
-      "loss": 6.7455,
-      "step": 764
-    },
-    {
-      "epoch": 0.39895697522816165,
-      "grad_norm": 2.1480019092559814,
-      "learning_rate": 9.973467336683418e-05,
-      "loss": 6.7664,
-      "step": 765
-    },
-    {
-      "epoch": 0.39947848761408083,
-      "grad_norm": 3.482856512069702,
-      "learning_rate": 9.973366834170856e-05,
-      "loss": 6.4067,
-      "step": 766
-    },
-    {
-      "epoch": 0.4,
-      "grad_norm": 4.285748481750488,
-      "learning_rate": 9.973266331658292e-05,
-      "loss": 6.2284,
-      "step": 767
-    },
-    {
-      "epoch": 0.40052151238591915,
-      "grad_norm": 3.017019033432007,
-      "learning_rate": 9.973165829145729e-05,
-      "loss": 6.0039,
-      "step": 768
-    },
-    {
-      "epoch": 0.40104302477183834,
-      "grad_norm": 2.2138030529022217,
-      "learning_rate": 9.973065326633166e-05,
-      "loss": 6.7996,
-      "step": 769
-    },
-    {
-      "epoch": 0.4015645371577575,
-      "grad_norm": 2.1760737895965576,
-      "learning_rate": 9.972964824120603e-05,
-      "loss": 5.9331,
-      "step": 770
-    },
-    {
-      "epoch": 0.40208604954367666,
-      "grad_norm": 2.1760194301605225,
-      "learning_rate": 9.97286432160804e-05,
-      "loss": 6.5415,
-      "step": 771
-    },
-    {
-      "epoch": 0.40260756192959585,
-      "grad_norm": 2.3232998847961426,
-      "learning_rate": 9.972763819095477e-05,
-      "loss": 6.1442,
-      "step": 772
-    },
-    {
-      "epoch": 0.403129074315515,
-      "grad_norm": 2.1887319087982178,
-      "learning_rate": 9.972663316582915e-05,
-      "loss": 6.4638,
-      "step": 773
-    },
-    {
-      "epoch": 0.40365058670143417,
-      "grad_norm": 2.012146234512329,
-      "learning_rate": 9.972562814070353e-05,
-      "loss": 6.5892,
-      "step": 774
-    },
-    {
-      "epoch": 0.4041720990873533,
-      "grad_norm": 2.053077220916748,
-      "learning_rate": 9.97246231155779e-05,
-      "loss": 6.6057,
-      "step": 775
-    },
-    {
-      "epoch": 0.4046936114732725,
-      "grad_norm": 2.199997901916504,
-      "learning_rate": 9.972361809045227e-05,
-      "loss": 6.4031,
-      "step": 776
-    },
-    {
-      "epoch": 0.4052151238591917,
-      "grad_norm": 2.076964855194092,
-      "learning_rate": 9.972261306532665e-05,
-      "loss": 6.6246,
-      "step": 777
-    },
-    {
-      "epoch": 0.4057366362451108,
-      "grad_norm": 2.625805377960205,
-      "learning_rate": 9.972160804020101e-05,
-      "loss": 5.8286,
-      "step": 778
-    },
-    {
-      "epoch": 0.40625814863103,
-      "grad_norm": 2.041363477706909,
-      "learning_rate": 9.972060301507539e-05,
-      "loss": 6.7248,
-      "step": 779
-    },
-    {
-      "epoch": 0.4067796610169492,
-      "grad_norm": 2.3348333835601807,
-      "learning_rate": 9.971959798994975e-05,
-      "loss": 6.7389,
-      "step": 780
-    },
-    {
-      "epoch": 0.4073011734028683,
-      "grad_norm": 2.599583625793457,
-      "learning_rate": 9.971859296482412e-05,
-      "loss": 6.1583,
-      "step": 781
-    },
-    {
-      "epoch": 0.4078226857887875,
-      "grad_norm": 2.1481375694274902,
-      "learning_rate": 9.97175879396985e-05,
-      "loss": 6.4425,
-      "step": 782
-    },
-    {
-      "epoch": 0.40834419817470663,
-      "grad_norm": 2.2883105278015137,
-      "learning_rate": 9.971658291457286e-05,
-      "loss": 6.2921,
-      "step": 783
-    },
-    {
-      "epoch": 0.4088657105606258,
-      "grad_norm": 2.9295544624328613,
-      "learning_rate": 9.971557788944724e-05,
-      "loss": 6.6697,
-      "step": 784
-    },
-    {
-      "epoch": 0.409387222946545,
-      "grad_norm": 3.041402816772461,
-      "learning_rate": 9.971457286432161e-05,
-      "loss": 6.5913,
-      "step": 785
-    },
-    {
-      "epoch": 0.40990873533246414,
-      "grad_norm": 2.110576868057251,
-      "learning_rate": 9.971356783919599e-05,
-      "loss": 6.7166,
-      "step": 786
-    },
-    {
-      "epoch": 0.4104302477183833,
-      "grad_norm": 2.453145980834961,
-      "learning_rate": 9.971256281407036e-05,
-      "loss": 6.4385,
-      "step": 787
-    },
-    {
-      "epoch": 0.41095176010430245,
-      "grad_norm": 2.2735378742218018,
-      "learning_rate": 9.971155778894473e-05,
-      "loss": 6.2105,
-      "step": 788
-    },
-    {
-      "epoch": 0.41147327249022164,
-      "grad_norm": 2.40952205657959,
-      "learning_rate": 9.97105527638191e-05,
-      "loss": 6.975,
-      "step": 789
-    },
-    {
-      "epoch": 0.41199478487614083,
-      "grad_norm": 2.142829418182373,
-      "learning_rate": 9.970954773869348e-05,
-      "loss": 6.33,
-      "step": 790
-    },
-    {
-      "epoch": 0.41251629726205996,
-      "grad_norm": 2.208801031112671,
-      "learning_rate": 9.970854271356784e-05,
-      "loss": 6.5763,
-      "step": 791
-    },
-    {
-      "epoch": 0.41303780964797915,
-      "grad_norm": 2.2279272079467773,
-      "learning_rate": 9.970753768844222e-05,
-      "loss": 6.7193,
-      "step": 792
-    },
-    {
-      "epoch": 0.4135593220338983,
-      "grad_norm": 2.445939064025879,
-      "learning_rate": 9.970653266331658e-05,
-      "loss": 5.6927,
-      "step": 793
-    },
-    {
-      "epoch": 0.41408083441981747,
-      "grad_norm": 2.1655194759368896,
-      "learning_rate": 9.970552763819096e-05,
-      "loss": 6.53,
-      "step": 794
-    },
-    {
-      "epoch": 0.41460234680573665,
-      "grad_norm": 2.0326120853424072,
-      "learning_rate": 9.970452261306534e-05,
-      "loss": 6.6162,
-      "step": 795
-    },
-    {
-      "epoch": 0.4151238591916558,
-      "grad_norm": 2.154041290283203,
-      "learning_rate": 9.97035175879397e-05,
-      "loss": 6.1554,
-      "step": 796
-    },
-    {
-      "epoch": 0.415645371577575,
-      "grad_norm": 2.786252021789551,
-      "learning_rate": 9.970251256281408e-05,
-      "loss": 6.5729,
-      "step": 797
-    },
-    {
-      "epoch": 0.4161668839634941,
-      "grad_norm": 2.426708936691284,
-      "learning_rate": 9.970150753768844e-05,
-      "loss": 6.3655,
-      "step": 798
-    },
-    {
-      "epoch": 0.4166883963494133,
-      "grad_norm": 2.152690887451172,
-      "learning_rate": 9.970050251256282e-05,
-      "loss": 6.4706,
-      "step": 799
-    },
-    {
-      "epoch": 0.4172099087353325,
-      "grad_norm": 2.748511791229248,
-      "learning_rate": 9.969949748743719e-05,
-      "loss": 5.6839,
-      "step": 800
-    },
-    {
-      "epoch": 0.4177314211212516,
-      "grad_norm": 2.1645658016204834,
-      "learning_rate": 9.969849246231156e-05,
-      "loss": 6.258,
-      "step": 801
-    },
-    {
-      "epoch": 0.4182529335071708,
-      "grad_norm": 1.7470781803131104,
-      "learning_rate": 9.969748743718593e-05,
-      "loss": 6.6934,
-      "step": 802
-    },
-    {
-      "epoch": 0.41877444589309,
-      "grad_norm": 2.1566150188446045,
-      "learning_rate": 9.96964824120603e-05,
-      "loss": 6.4163,
-      "step": 803
-    },
-    {
-      "epoch": 0.4192959582790091,
-      "grad_norm": 2.085911750793457,
-      "learning_rate": 9.969547738693467e-05,
-      "loss": 6.4241,
-      "step": 804
-    },
-    {
-      "epoch": 0.4198174706649283,
-      "grad_norm": 2.668271780014038,
-      "learning_rate": 9.969447236180905e-05,
-      "loss": 6.0993,
-      "step": 805
-    },
-    {
-      "epoch": 0.42033898305084744,
-      "grad_norm": 2.191254138946533,
-      "learning_rate": 9.969346733668343e-05,
-      "loss": 6.7284,
-      "step": 806
-    },
-    {
-      "epoch": 0.4208604954367666,
-      "grad_norm": 2.0428121089935303,
-      "learning_rate": 9.969246231155779e-05,
-      "loss": 6.4521,
-      "step": 807
-    },
-    {
-      "epoch": 0.4213820078226858,
-      "grad_norm": 2.4796793460845947,
-      "learning_rate": 9.969145728643217e-05,
-      "loss": 6.4149,
-      "step": 808
-    },
-    {
-      "epoch": 0.42190352020860494,
-      "grad_norm": 1.9883676767349243,
-      "learning_rate": 9.969045226130653e-05,
-      "loss": 6.7675,
-      "step": 809
-    },
-    {
-      "epoch": 0.42242503259452413,
-      "grad_norm": 2.1834592819213867,
-      "learning_rate": 9.968944723618091e-05,
-      "loss": 6.7581,
-      "step": 810
-    },
-    {
-      "epoch": 0.42294654498044326,
-      "grad_norm": 2.4649016857147217,
-      "learning_rate": 9.968844221105527e-05,
-      "loss": 6.3115,
-      "step": 811
-    },
-    {
-      "epoch": 0.42346805736636245,
-      "grad_norm": 2.1063711643218994,
-      "learning_rate": 9.968743718592965e-05,
-      "loss": 6.4236,
-      "step": 812
-    },
-    {
-      "epoch": 0.42398956975228164,
-      "grad_norm": 2.2991364002227783,
-      "learning_rate": 9.968643216080402e-05,
-      "loss": 6.5763,
-      "step": 813
-    },
-    {
-      "epoch": 0.42451108213820077,
-      "grad_norm": 2.2292230129241943,
-      "learning_rate": 9.96854271356784e-05,
-      "loss": 5.647,
-      "step": 814
-    },
-    {
-      "epoch": 0.42503259452411996,
-      "grad_norm": 2.248950481414795,
-      "learning_rate": 9.968442211055277e-05,
-      "loss": 6.0973,
-      "step": 815
-    },
-    {
-      "epoch": 0.4255541069100391,
-      "grad_norm": 2.0685508251190186,
-      "learning_rate": 9.968341708542715e-05,
-      "loss": 6.6272,
-      "step": 816
-    },
-    {
-      "epoch": 0.4260756192959583,
-      "grad_norm": 2.3320984840393066,
-      "learning_rate": 9.968241206030151e-05,
-      "loss": 6.0772,
-      "step": 817
-    },
-    {
-      "epoch": 0.42659713168187746,
-      "grad_norm": 2.918611526489258,
-      "learning_rate": 9.968140703517589e-05,
-      "loss": 6.1137,
-      "step": 818
-    },
-    {
-      "epoch": 0.4271186440677966,
-      "grad_norm": 2.216487169265747,
-      "learning_rate": 9.968040201005026e-05,
-      "loss": 6.4165,
-      "step": 819
-    },
-    {
-      "epoch": 0.4276401564537158,
-      "grad_norm": 2.4474472999572754,
-      "learning_rate": 9.967939698492463e-05,
-      "loss": 6.4032,
-      "step": 820
-    },
-    {
-      "epoch": 0.42816166883963497,
-      "grad_norm": 2.467963218688965,
-      "learning_rate": 9.9678391959799e-05,
-      "loss": 6.5711,
-      "step": 821
-    },
-    {
-      "epoch": 0.4286831812255541,
-      "grad_norm": 2.1296226978302,
-      "learning_rate": 9.967738693467336e-05,
-      "loss": 6.7074,
-      "step": 822
-    },
-    {
-      "epoch": 0.4292046936114733,
-      "grad_norm": 1.93131685256958,
-      "learning_rate": 9.967638190954774e-05,
-      "loss": 6.6004,
-      "step": 823
-    },
-    {
-      "epoch": 0.4297262059973924,
-      "grad_norm": 2.0198779106140137,
-      "learning_rate": 9.96753768844221e-05,
-      "loss": 6.3437,
-      "step": 824
-    },
-    {
-      "epoch": 0.4302477183833116,
-      "grad_norm": 2.828876495361328,
-      "learning_rate": 9.967437185929648e-05,
-      "loss": 6.3619,
-      "step": 825
-    },
-    {
-      "epoch": 0.4307692307692308,
-      "grad_norm": 2.0243163108825684,
-      "learning_rate": 9.967336683417086e-05,
-      "loss": 6.6018,
-      "step": 826
-    },
-    {
-      "epoch": 0.4312907431551499,
-      "grad_norm": 2.898697853088379,
-      "learning_rate": 9.967236180904524e-05,
-      "loss": 6.3946,
-      "step": 827
-    },
-    {
-      "epoch": 0.4318122555410691,
-      "grad_norm": 2.850396156311035,
-      "learning_rate": 9.96713567839196e-05,
-      "loss": 6.2984,
-      "step": 828
-    },
-    {
-      "epoch": 0.43233376792698824,
-      "grad_norm": 2.146773099899292,
-      "learning_rate": 9.967035175879398e-05,
-      "loss": 6.1838,
-      "step": 829
-    },
-    {
-      "epoch": 0.43285528031290743,
-      "grad_norm": 2.1175990104675293,
-      "learning_rate": 9.966934673366835e-05,
-      "loss": 6.2523,
-      "step": 830
-    },
-    {
-      "epoch": 0.4333767926988266,
-      "grad_norm": 2.2733535766601562,
-      "learning_rate": 9.966834170854272e-05,
-      "loss": 6.0775,
-      "step": 831
-    },
-    {
-      "epoch": 0.43389830508474575,
-      "grad_norm": 2.080181360244751,
-      "learning_rate": 9.966733668341709e-05,
-      "loss": 6.4696,
-      "step": 832
-    },
-    {
-      "epoch": 0.43441981747066494,
-      "grad_norm": 2.3921234607696533,
-      "learning_rate": 9.966633165829147e-05,
-      "loss": 6.4152,
-      "step": 833
-    },
-    {
-      "epoch": 0.43494132985658407,
-      "grad_norm": 2.218611478805542,
-      "learning_rate": 9.966532663316583e-05,
-      "loss": 6.0767,
-      "step": 834
-    },
-    {
-      "epoch": 0.43546284224250326,
-      "grad_norm": 2.674525737762451,
-      "learning_rate": 9.966432160804021e-05,
-      "loss": 6.377,
-      "step": 835
-    },
-    {
-      "epoch": 0.43598435462842244,
-      "grad_norm": 2.659274101257324,
-      "learning_rate": 9.966331658291458e-05,
-      "loss": 6.121,
-      "step": 836
-    },
-    {
-      "epoch": 0.4365058670143416,
-      "grad_norm": 2.6279726028442383,
-      "learning_rate": 9.966231155778895e-05,
-      "loss": 6.1335,
-      "step": 837
-    },
-    {
-      "epoch": 0.43702737940026076,
-      "grad_norm": 2.000652551651001,
-      "learning_rate": 9.966130653266333e-05,
-      "loss": 6.4938,
-      "step": 838
-    },
-    {
-      "epoch": 0.4375488917861799,
-      "grad_norm": 1.9366097450256348,
-      "learning_rate": 9.966030150753769e-05,
-      "loss": 6.5206,
-      "step": 839
-    },
-    {
-      "epoch": 0.4380704041720991,
-      "grad_norm": 2.2027428150177,
-      "learning_rate": 9.965929648241207e-05,
-      "loss": 6.7759,
-      "step": 840
-    },
-    {
-      "epoch": 0.43859191655801827,
-      "grad_norm": 2.574753522872925,
-      "learning_rate": 9.965829145728643e-05,
-      "loss": 6.3747,
-      "step": 841
-    },
-    {
-      "epoch": 0.4391134289439374,
-      "grad_norm": 1.8134504556655884,
-      "learning_rate": 9.965728643216081e-05,
-      "loss": 6.6625,
-      "step": 842
-    },
-    {
-      "epoch": 0.4396349413298566,
-      "grad_norm": 2.208388328552246,
-      "learning_rate": 9.965628140703518e-05,
-      "loss": 6.28,
-      "step": 843
-    },
-    {
-      "epoch": 0.4401564537157758,
-      "grad_norm": 2.0692834854125977,
-      "learning_rate": 9.965527638190955e-05,
-      "loss": 6.2188,
-      "step": 844
-    },
-    {
-      "epoch": 0.4406779661016949,
-      "grad_norm": 2.3833465576171875,
-      "learning_rate": 9.965427135678392e-05,
-      "loss": 6.6833,
-      "step": 845
-    },
-    {
-      "epoch": 0.4411994784876141,
-      "grad_norm": 2.536780834197998,
-      "learning_rate": 9.96532663316583e-05,
-      "loss": 6.4912,
-      "step": 846
-    },
-    {
-      "epoch": 0.4417209908735332,
-      "grad_norm": 2.005906343460083,
-      "learning_rate": 9.965226130653267e-05,
-      "loss": 6.3401,
-      "step": 847
-    },
-    {
-      "epoch": 0.4422425032594524,
-      "grad_norm": 2.355052947998047,
-      "learning_rate": 9.965125628140704e-05,
-      "loss": 6.5562,
-      "step": 848
-    },
-    {
-      "epoch": 0.4427640156453716,
-      "grad_norm": 2.3538742065429688,
-      "learning_rate": 9.965025125628142e-05,
-      "loss": 6.5453,
-      "step": 849
-    },
-    {
-      "epoch": 0.44328552803129073,
-      "grad_norm": 1.905880093574524,
-      "learning_rate": 9.964924623115578e-05,
-      "loss": 6.6314,
-      "step": 850
-    },
-    {
-      "epoch": 0.4438070404172099,
-      "grad_norm": 2.212345600128174,
-      "learning_rate": 9.964824120603016e-05,
-      "loss": 5.8242,
-      "step": 851
-    },
-    {
-      "epoch": 0.44432855280312905,
-      "grad_norm": 2.20920991897583,
-      "learning_rate": 9.964723618090452e-05,
-      "loss": 6.6775,
-      "step": 852
-    },
-    {
-      "epoch": 0.44485006518904824,
-      "grad_norm": 2.1129469871520996,
-      "learning_rate": 9.96462311557789e-05,
-      "loss": 6.4668,
-      "step": 853
-    },
-    {
-      "epoch": 0.4453715775749674,
-      "grad_norm": 1.9672009944915771,
-      "learning_rate": 9.964522613065326e-05,
-      "loss": 6.4218,
-      "step": 854
-    },
-    {
-      "epoch": 0.44589308996088656,
-      "grad_norm": 2.4036476612091064,
-      "learning_rate": 9.964422110552764e-05,
-      "loss": 6.1991,
-      "step": 855
-    },
-    {
-      "epoch": 0.44641460234680574,
-      "grad_norm": 2.628465414047241,
-      "learning_rate": 9.964321608040202e-05,
-      "loss": 6.1609,
-      "step": 856
-    },
-    {
-      "epoch": 0.4469361147327249,
-      "grad_norm": 2.142796039581299,
-      "learning_rate": 9.96422110552764e-05,
-      "loss": 6.3854,
-      "step": 857
-    },
-    {
-      "epoch": 0.44745762711864406,
-      "grad_norm": 2.1347718238830566,
-      "learning_rate": 9.964120603015076e-05,
-      "loss": 6.6483,
-      "step": 858
-    },
-    {
-      "epoch": 0.44797913950456325,
-      "grad_norm": 1.9581797122955322,
-      "learning_rate": 9.964020100502514e-05,
-      "loss": 6.5637,
-      "step": 859
-    },
-    {
-      "epoch": 0.4485006518904824,
-      "grad_norm": 2.314955472946167,
-      "learning_rate": 9.96391959798995e-05,
-      "loss": 5.8771,
-      "step": 860
-    },
-    {
-      "epoch": 0.44902216427640157,
-      "grad_norm": 2.3754324913024902,
-      "learning_rate": 9.963819095477387e-05,
-      "loss": 6.4736,
-      "step": 861
-    },
-    {
-      "epoch": 0.44954367666232076,
-      "grad_norm": 2.0141539573669434,
-      "learning_rate": 9.963718592964825e-05,
-      "loss": 5.9831,
-      "step": 862
-    },
-    {
-      "epoch": 0.4500651890482399,
-      "grad_norm": 2.220392942428589,
-      "learning_rate": 9.963618090452261e-05,
-      "loss": 6.4879,
-      "step": 863
-    },
-    {
-      "epoch": 0.4505867014341591,
-      "grad_norm": 2.6271748542785645,
-      "learning_rate": 9.963517587939699e-05,
-      "loss": 6.0724,
-      "step": 864
-    },
-    {
-      "epoch": 0.4511082138200782,
-      "grad_norm": 2.093454360961914,
-      "learning_rate": 9.963417085427135e-05,
-      "loss": 6.3358,
-      "step": 865
-    },
-    {
-      "epoch": 0.4516297262059974,
-      "grad_norm": 2.137228012084961,
-      "learning_rate": 9.963316582914573e-05,
-      "loss": 6.5176,
-      "step": 866
-    },
-    {
-      "epoch": 0.4521512385919166,
-      "grad_norm": 2.132906436920166,
-      "learning_rate": 9.963216080402011e-05,
-      "loss": 6.3885,
-      "step": 867
-    },
-    {
-      "epoch": 0.4526727509778357,
-      "grad_norm": 1.9319450855255127,
-      "learning_rate": 9.963115577889449e-05,
-      "loss": 6.6291,
-      "step": 868
-    },
-    {
-      "epoch": 0.4531942633637549,
-      "grad_norm": 2.303853750228882,
-      "learning_rate": 9.963015075376885e-05,
-      "loss": 6.2823,
-      "step": 869
-    },
-    {
-      "epoch": 0.45371577574967403,
-      "grad_norm": 2.088036298751831,
-      "learning_rate": 9.962914572864323e-05,
-      "loss": 6.4187,
-      "step": 870
-    },
-    {
-      "epoch": 0.4542372881355932,
-      "grad_norm": 1.9422059059143066,
-      "learning_rate": 9.962814070351759e-05,
-      "loss": 6.2571,
-      "step": 871
-    },
-    {
-      "epoch": 0.4547588005215124,
-      "grad_norm": 2.202895164489746,
-      "learning_rate": 9.962713567839197e-05,
-      "loss": 6.3058,
-      "step": 872
-    },
-    {
-      "epoch": 0.45528031290743154,
-      "grad_norm": 3.0498602390289307,
-      "learning_rate": 9.962613065326633e-05,
-      "loss": 6.3952,
-      "step": 873
-    },
-    {
-      "epoch": 0.4558018252933507,
-      "grad_norm": 2.319120168685913,
-      "learning_rate": 9.96251256281407e-05,
-      "loss": 6.1164,
-      "step": 874
-    },
-    {
-      "epoch": 0.45632333767926986,
-      "grad_norm": 2.3087847232818604,
-      "learning_rate": 9.962412060301508e-05,
-      "loss": 5.9293,
-      "step": 875
-    },
-    {
-      "epoch": 0.45684485006518905,
-      "grad_norm": 2.0732786655426025,
-      "learning_rate": 9.962311557788945e-05,
-      "loss": 6.2632,
-      "step": 876
-    },
-    {
-      "epoch": 0.45736636245110823,
-      "grad_norm": 2.2241642475128174,
-      "learning_rate": 9.962211055276383e-05,
-      "loss": 6.4525,
-      "step": 877
-    },
-    {
-      "epoch": 0.45788787483702736,
-      "grad_norm": 1.9518303871154785,
-      "learning_rate": 9.96211055276382e-05,
-      "loss": 5.958,
-      "step": 878
-    },
-    {
-      "epoch": 0.45840938722294655,
-      "grad_norm": 2.0389230251312256,
-      "learning_rate": 9.962010050251257e-05,
-      "loss": 6.6528,
-      "step": 879
-    },
-    {
-      "epoch": 0.4589308996088657,
-      "grad_norm": 2.852085590362549,
-      "learning_rate": 9.961909547738694e-05,
-      "loss": 5.5762,
-      "step": 880
-    },
-    {
-      "epoch": 0.45945241199478487,
-      "grad_norm": 2.5660274028778076,
-      "learning_rate": 9.961809045226132e-05,
-      "loss": 6.5803,
-      "step": 881
-    },
-    {
-      "epoch": 0.45997392438070406,
-      "grad_norm": 2.298530340194702,
-      "learning_rate": 9.961708542713568e-05,
-      "loss": 6.46,
-      "step": 882
-    },
-    {
-      "epoch": 0.4604954367666232,
-      "grad_norm": 2.127445697784424,
-      "learning_rate": 9.961608040201006e-05,
-      "loss": 6.4486,
-      "step": 883
-    },
-    {
-      "epoch": 0.4610169491525424,
-      "grad_norm": 1.897956132888794,
-      "learning_rate": 9.961507537688442e-05,
-      "loss": 6.7523,
-      "step": 884
-    },
-    {
-      "epoch": 0.46153846153846156,
-      "grad_norm": 4.019052028656006,
-      "learning_rate": 9.96140703517588e-05,
-      "loss": 6.1086,
-      "step": 885
-    },
-    {
-      "epoch": 0.4620599739243807,
-      "grad_norm": 1.951799988746643,
-      "learning_rate": 9.961306532663316e-05,
-      "loss": 6.7259,
-      "step": 886
-    },
-    {
-      "epoch": 0.4625814863102999,
-      "grad_norm": 1.9468040466308594,
-      "learning_rate": 9.961206030150754e-05,
-      "loss": 6.6451,
-      "step": 887
-    },
-    {
-      "epoch": 0.463102998696219,
-      "grad_norm": 2.5095787048339844,
-      "learning_rate": 9.961105527638192e-05,
-      "loss": 5.9871,
-      "step": 888
-    },
-    {
-      "epoch": 0.4636245110821382,
-      "grad_norm": 2.0893378257751465,
-      "learning_rate": 9.961005025125628e-05,
-      "loss": 6.3916,
-      "step": 889
-    },
-    {
-      "epoch": 0.4641460234680574,
-      "grad_norm": 2.000333070755005,
-      "learning_rate": 9.960904522613066e-05,
-      "loss": 6.5248,
-      "step": 890
-    },
-    {
-      "epoch": 0.4646675358539765,
-      "grad_norm": 2.454456090927124,
-      "learning_rate": 9.960804020100503e-05,
-      "loss": 6.4069,
-      "step": 891
-    },
-    {
-      "epoch": 0.4651890482398957,
-      "grad_norm": 2.1235005855560303,
-      "learning_rate": 9.96070351758794e-05,
-      "loss": 6.4154,
-      "step": 892
-    },
-    {
-      "epoch": 0.46571056062581484,
-      "grad_norm": 2.0405213832855225,
-      "learning_rate": 9.960603015075377e-05,
-      "loss": 6.624,
-      "step": 893
-    },
-    {
-      "epoch": 0.466232073011734,
-      "grad_norm": 2.1271770000457764,
-      "learning_rate": 9.960502512562815e-05,
-      "loss": 6.5836,
-      "step": 894
-    },
-    {
-      "epoch": 0.4667535853976532,
-      "grad_norm": 2.0283665657043457,
-      "learning_rate": 9.960402010050251e-05,
-      "loss": 6.6333,
-      "step": 895
-    },
-    {
-      "epoch": 0.46727509778357235,
-      "grad_norm": 3.0277304649353027,
-      "learning_rate": 9.960301507537689e-05,
-      "loss": 6.4652,
-      "step": 896
-    },
-    {
-      "epoch": 0.46779661016949153,
-      "grad_norm": 3.6672751903533936,
-      "learning_rate": 9.960201005025127e-05,
-      "loss": 6.3166,
-      "step": 897
-    },
-    {
-      "epoch": 0.46831812255541067,
-      "grad_norm": 2.3754241466522217,
-      "learning_rate": 9.960100502512564e-05,
-      "loss": 6.6083,
-      "step": 898
-    },
-    {
-      "epoch": 0.46883963494132985,
-      "grad_norm": 2.811269998550415,
-      "learning_rate": 9.960000000000001e-05,
-      "loss": 6.2006,
-      "step": 899
-    },
-    {
-      "epoch": 0.46936114732724904,
-      "grad_norm": 1.9116339683532715,
-      "learning_rate": 9.959899497487437e-05,
-      "loss": 6.7957,
-      "step": 900
-    },
-    {
-      "epoch": 0.46988265971316817,
-      "grad_norm": 1.9986357688903809,
-      "learning_rate": 9.959798994974875e-05,
-      "loss": 6.3889,
-      "step": 901
-    },
-    {
-      "epoch": 0.47040417209908736,
-      "grad_norm": 1.9240593910217285,
-      "learning_rate": 9.959698492462312e-05,
-      "loss": 6.5054,
-      "step": 902
-    },
-    {
-      "epoch": 0.47092568448500655,
-      "grad_norm": 1.8461600542068481,
-      "learning_rate": 9.959597989949749e-05,
-      "loss": 6.5647,
-      "step": 903
-    },
-    {
-      "epoch": 0.4714471968709257,
-      "grad_norm": 1.726055383682251,
-      "learning_rate": 9.959497487437186e-05,
-      "loss": 6.3581,
-      "step": 904
-    },
-    {
-      "epoch": 0.47196870925684486,
-      "grad_norm": 1.9957642555236816,
-      "learning_rate": 9.959396984924623e-05,
-      "loss": 6.472,
-      "step": 905
-    },
-    {
-      "epoch": 0.472490221642764,
-      "grad_norm": 1.817854404449463,
-      "learning_rate": 9.95929648241206e-05,
-      "loss": 6.4174,
-      "step": 906
-    },
-    {
-      "epoch": 0.4730117340286832,
-      "grad_norm": 2.026216506958008,
-      "learning_rate": 9.959195979899498e-05,
-      "loss": 6.4387,
-      "step": 907
-    },
-    {
-      "epoch": 0.47353324641460237,
-      "grad_norm": 1.8881945610046387,
-      "learning_rate": 9.959095477386935e-05,
-      "loss": 6.5739,
-      "step": 908
-    },
-    {
-      "epoch": 0.4740547588005215,
-      "grad_norm": 2.0653135776519775,
-      "learning_rate": 9.958994974874373e-05,
-      "loss": 6.6318,
-      "step": 909
-    },
-    {
-      "epoch": 0.4745762711864407,
-      "grad_norm": 2.2507307529449463,
-      "learning_rate": 9.95889447236181e-05,
-      "loss": 6.3731,
-      "step": 910
-    },
-    {
-      "epoch": 0.4750977835723598,
-      "grad_norm": 2.0260534286499023,
-      "learning_rate": 9.958793969849247e-05,
-      "loss": 6.1886,
-      "step": 911
-    },
-    {
-      "epoch": 0.475619295958279,
-      "grad_norm": 2.2879021167755127,
-      "learning_rate": 9.958693467336684e-05,
-      "loss": 5.9656,
-      "step": 912
-    },
-    {
-      "epoch": 0.4761408083441982,
-      "grad_norm": 1.8898496627807617,
-      "learning_rate": 9.958592964824122e-05,
-      "loss": 6.4944,
-      "step": 913
-    },
-    {
-      "epoch": 0.47666232073011733,
-      "grad_norm": 1.9814594984054565,
-      "learning_rate": 9.958492462311558e-05,
-      "loss": 6.361,
-      "step": 914
-    },
-    {
-      "epoch": 0.4771838331160365,
-      "grad_norm": 2.2697174549102783,
-      "learning_rate": 9.958391959798995e-05,
-      "loss": 6.4851,
-      "step": 915
-    },
-    {
-      "epoch": 0.47770534550195565,
-      "grad_norm": 2.321835517883301,
-      "learning_rate": 9.958291457286432e-05,
-      "loss": 5.9252,
-      "step": 916
-    },
-    {
-      "epoch": 0.47822685788787483,
-      "grad_norm": 2.1413791179656982,
-      "learning_rate": 9.958190954773869e-05,
-      "loss": 6.5192,
-      "step": 917
-    },
-    {
-      "epoch": 0.478748370273794,
-      "grad_norm": 2.2743680477142334,
-      "learning_rate": 9.958090452261307e-05,
-      "loss": 6.183,
-      "step": 918
-    },
-    {
-      "epoch": 0.47926988265971315,
-      "grad_norm": 2.1204566955566406,
-      "learning_rate": 9.957989949748744e-05,
-      "loss": 6.2142,
-      "step": 919
-    },
-    {
-      "epoch": 0.47979139504563234,
-      "grad_norm": 2.070068836212158,
-      "learning_rate": 9.957889447236182e-05,
-      "loss": 6.7194,
-      "step": 920
-    },
-    {
-      "epoch": 0.4803129074315515,
-      "grad_norm": 2.5954298973083496,
-      "learning_rate": 9.957788944723619e-05,
-      "loss": 5.7381,
-      "step": 921
-    },
-    {
-      "epoch": 0.48083441981747066,
-      "grad_norm": 1.9493515491485596,
-      "learning_rate": 9.957688442211056e-05,
-      "loss": 6.4862,
-      "step": 922
-    },
-    {
-      "epoch": 0.48135593220338985,
-      "grad_norm": 2.2104249000549316,
-      "learning_rate": 9.957587939698493e-05,
-      "loss": 6.1622,
-      "step": 923
-    },
-    {
-      "epoch": 0.481877444589309,
-      "grad_norm": 2.772524118423462,
-      "learning_rate": 9.95748743718593e-05,
-      "loss": 5.6473,
-      "step": 924
-    },
-    {
-      "epoch": 0.48239895697522817,
-      "grad_norm": 2.035775661468506,
-      "learning_rate": 9.957386934673367e-05,
-      "loss": 6.0676,
-      "step": 925
-    },
-    {
-      "epoch": 0.48292046936114735,
-      "grad_norm": 2.280177354812622,
-      "learning_rate": 9.957286432160805e-05,
-      "loss": 6.2864,
-      "step": 926
-    },
-    {
-      "epoch": 0.4834419817470665,
-      "grad_norm": 1.8228402137756348,
-      "learning_rate": 9.957185929648241e-05,
-      "loss": 6.6061,
-      "step": 927
-    },
-    {
-      "epoch": 0.48396349413298567,
-      "grad_norm": 2.4340789318084717,
-      "learning_rate": 9.957085427135679e-05,
-      "loss": 6.4215,
-      "step": 928
-    },
-    {
-      "epoch": 0.4844850065189048,
-      "grad_norm": 1.9269063472747803,
-      "learning_rate": 9.956984924623117e-05,
-      "loss": 6.2029,
-      "step": 929
-    },
-    {
-      "epoch": 0.485006518904824,
-      "grad_norm": 2.2546534538269043,
-      "learning_rate": 9.956884422110553e-05,
-      "loss": 6.2769,
-      "step": 930
-    },
-    {
-      "epoch": 0.4855280312907432,
-      "grad_norm": 1.8607932329177856,
-      "learning_rate": 9.956783919597991e-05,
-      "loss": 6.6443,
-      "step": 931
-    },
-    {
-      "epoch": 0.4860495436766623,
-      "grad_norm": 1.9565026760101318,
-      "learning_rate": 9.956683417085427e-05,
-      "loss": 6.5523,
-      "step": 932
-    },
-    {
-      "epoch": 0.4865710560625815,
-      "grad_norm": 1.8473814725875854,
-      "learning_rate": 9.956582914572865e-05,
-      "loss": 6.6484,
-      "step": 933
-    },
-    {
-      "epoch": 0.48709256844850063,
-      "grad_norm": 2.0442159175872803,
-      "learning_rate": 9.956482412060302e-05,
-      "loss": 6.6033,
-      "step": 934
-    },
-    {
-      "epoch": 0.4876140808344198,
-      "grad_norm": 1.9316771030426025,
-      "learning_rate": 9.95638190954774e-05,
-      "loss": 6.2141,
-      "step": 935
-    },
-    {
-      "epoch": 0.488135593220339,
-      "grad_norm": 1.9396896362304688,
-      "learning_rate": 9.956281407035176e-05,
-      "loss": 6.6897,
-      "step": 936
-    },
-    {
-      "epoch": 0.48865710560625814,
-      "grad_norm": 1.804550290107727,
-      "learning_rate": 9.956180904522614e-05,
-      "loss": 6.5316,
-      "step": 937
-    },
-    {
-      "epoch": 0.4891786179921773,
-      "grad_norm": 2.3102989196777344,
-      "learning_rate": 9.95608040201005e-05,
-      "loss": 6.467,
-      "step": 938
-    },
-    {
-      "epoch": 0.48970013037809645,
-      "grad_norm": 1.9410513639450073,
-      "learning_rate": 9.955979899497488e-05,
-      "loss": 6.698,
-      "step": 939
-    },
-    {
-      "epoch": 0.49022164276401564,
-      "grad_norm": 1.9564518928527832,
-      "learning_rate": 9.955879396984926e-05,
-      "loss": 6.4822,
-      "step": 940
-    },
-    {
-      "epoch": 0.49074315514993483,
-      "grad_norm": 1.8969272375106812,
-      "learning_rate": 9.955778894472362e-05,
-      "loss": 6.2348,
-      "step": 941
-    },
-    {
-      "epoch": 0.49126466753585396,
-      "grad_norm": 2.329587697982788,
-      "learning_rate": 9.9556783919598e-05,
-      "loss": 6.0454,
-      "step": 942
-    },
-    {
-      "epoch": 0.49178617992177315,
-      "grad_norm": 1.7472556829452515,
-      "learning_rate": 9.955577889447236e-05,
-      "loss": 6.4282,
-      "step": 943
-    },
-    {
-      "epoch": 0.49230769230769234,
-      "grad_norm": 1.8456318378448486,
-      "learning_rate": 9.955477386934674e-05,
-      "loss": 6.3436,
-      "step": 944
-    },
-    {
-      "epoch": 0.49282920469361147,
-      "grad_norm": 2.0255699157714844,
-      "learning_rate": 9.95537688442211e-05,
-      "loss": 6.1632,
-      "step": 945
-    },
-    {
-      "epoch": 0.49335071707953065,
-      "grad_norm": 2.161907911300659,
-      "learning_rate": 9.955276381909548e-05,
-      "loss": 6.2036,
-      "step": 946
-    },
-    {
-      "epoch": 0.4938722294654498,
-      "grad_norm": 2.113365411758423,
-      "learning_rate": 9.955175879396985e-05,
-      "loss": 6.1591,
-      "step": 947
-    },
-    {
-      "epoch": 0.494393741851369,
-      "grad_norm": 2.2957074642181396,
-      "learning_rate": 9.955075376884422e-05,
-      "loss": 6.2229,
-      "step": 948
-    },
-    {
-      "epoch": 0.49491525423728816,
-      "grad_norm": 2.9717864990234375,
-      "learning_rate": 9.95497487437186e-05,
-      "loss": 6.6651,
-      "step": 949
-    },
-    {
-      "epoch": 0.4954367666232073,
-      "grad_norm": 1.9626245498657227,
-      "learning_rate": 9.954874371859298e-05,
-      "loss": 6.6075,
-      "step": 950
-    },
-    {
-      "epoch": 0.4959582790091265,
-      "grad_norm": 3.1651618480682373,
-      "learning_rate": 9.954773869346734e-05,
-      "loss": 6.1353,
-      "step": 951
-    },
-    {
-      "epoch": 0.4964797913950456,
-      "grad_norm": 2.3175432682037354,
-      "learning_rate": 9.954673366834172e-05,
-      "loss": 6.5026,
-      "step": 952
-    },
-    {
-      "epoch": 0.4970013037809648,
-      "grad_norm": 2.5663411617279053,
-      "learning_rate": 9.954572864321609e-05,
-      "loss": 6.6599,
-      "step": 953
-    },
-    {
-      "epoch": 0.497522816166884,
-      "grad_norm": 2.2984795570373535,
-      "learning_rate": 9.954472361809045e-05,
-      "loss": 6.636,
-      "step": 954
-    },
-    {
-      "epoch": 0.4980443285528031,
-      "grad_norm": 2.301086664199829,
-      "learning_rate": 9.954371859296483e-05,
-      "loss": 6.476,
-      "step": 955
-    },
-    {
-      "epoch": 0.4985658409387223,
-      "grad_norm": 2.7066967487335205,
-      "learning_rate": 9.954271356783919e-05,
-      "loss": 6.0005,
-      "step": 956
-    },
-    {
-      "epoch": 0.49908735332464144,
-      "grad_norm": 2.74013352394104,
-      "learning_rate": 9.954170854271357e-05,
-      "loss": 6.6643,
-      "step": 957
-    },
-    {
-      "epoch": 0.4996088657105606,
-      "grad_norm": 1.9838446378707886,
-      "learning_rate": 9.954070351758793e-05,
-      "loss": 6.4186,
-      "step": 958
-    },
-    {
-      "epoch": 0.5001303780964798,
-      "grad_norm": 2.4673357009887695,
-      "learning_rate": 9.953969849246231e-05,
-      "loss": 6.5595,
-      "step": 959
-    },
-    {
-      "epoch": 0.500651890482399,
-      "grad_norm": 2.108759880065918,
-      "learning_rate": 9.953869346733669e-05,
-      "loss": 6.5914,
-      "step": 960
-    },
-    {
-      "epoch": 0.5011734028683181,
-      "grad_norm": 2.391068696975708,
-      "learning_rate": 9.953768844221107e-05,
-      "loss": 6.661,
-      "step": 961
-    },
-    {
-      "epoch": 0.5016949152542373,
-      "grad_norm": 2.1729373931884766,
-      "learning_rate": 9.953668341708543e-05,
-      "loss": 6.1647,
-      "step": 962
-    },
-    {
-      "epoch": 0.5022164276401565,
-      "grad_norm": 2.26346755027771,
-      "learning_rate": 9.953567839195981e-05,
-      "loss": 6.261,
-      "step": 963
-    },
-    {
-      "epoch": 0.5027379400260756,
-      "grad_norm": 2.5807383060455322,
-      "learning_rate": 9.953467336683417e-05,
-      "loss": 6.1459,
-      "step": 964
-    },
-    {
-      "epoch": 0.5032594524119948,
-      "grad_norm": 2.2451083660125732,
-      "learning_rate": 9.953366834170855e-05,
-      "loss": 6.489,
-      "step": 965
-    },
-    {
-      "epoch": 0.5037809647979139,
-      "grad_norm": 2.171003580093384,
-      "learning_rate": 9.953266331658292e-05,
-      "loss": 6.6078,
-      "step": 966
-    },
-    {
-      "epoch": 0.5043024771838331,
-      "grad_norm": 3.1965880393981934,
-      "learning_rate": 9.953165829145728e-05,
-      "loss": 5.5818,
-      "step": 967
-    },
-    {
-      "epoch": 0.5048239895697523,
-      "grad_norm": 2.145333766937256,
-      "learning_rate": 9.953065326633166e-05,
-      "loss": 6.2235,
-      "step": 968
-    },
-    {
-      "epoch": 0.5053455019556714,
-      "grad_norm": 2.572251558303833,
-      "learning_rate": 9.952964824120604e-05,
-      "loss": 6.2302,
-      "step": 969
-    },
-    {
-      "epoch": 0.5058670143415906,
-      "grad_norm": 2.1392300128936768,
-      "learning_rate": 9.952864321608041e-05,
-      "loss": 6.6791,
-      "step": 970
-    },
-    {
-      "epoch": 0.5063885267275098,
-      "grad_norm": 2.185953140258789,
-      "learning_rate": 9.952763819095478e-05,
-      "loss": 6.5173,
-      "step": 971
-    },
-    {
-      "epoch": 0.5069100391134289,
-      "grad_norm": 2.472808361053467,
-      "learning_rate": 9.952663316582916e-05,
-      "loss": 6.738,
-      "step": 972
-    },
-    {
-      "epoch": 0.5074315514993482,
-      "grad_norm": 2.662168025970459,
-      "learning_rate": 9.952562814070352e-05,
-      "loss": 6.2035,
-      "step": 973
-    },
-    {
-      "epoch": 0.5079530638852673,
-      "grad_norm": 2.2262117862701416,
-      "learning_rate": 9.95246231155779e-05,
-      "loss": 6.2722,
-      "step": 974
-    },
-    {
-      "epoch": 0.5084745762711864,
-      "grad_norm": 3.2270820140838623,
-      "learning_rate": 9.952361809045226e-05,
-      "loss": 6.3604,
-      "step": 975
-    },
-    {
-      "epoch": 0.5089960886571057,
-      "grad_norm": 2.8973543643951416,
-      "learning_rate": 9.952261306532664e-05,
-      "loss": 6.4082,
-      "step": 976
-    },
-    {
-      "epoch": 0.5095176010430248,
-      "grad_norm": 2.3042736053466797,
-      "learning_rate": 9.9521608040201e-05,
-      "loss": 6.3213,
-      "step": 977
-    },
-    {
-      "epoch": 0.5100391134289439,
-      "grad_norm": 2.7366580963134766,
-      "learning_rate": 9.952060301507538e-05,
-      "loss": 5.7565,
-      "step": 978
-    },
-    {
-      "epoch": 0.5105606258148631,
-      "grad_norm": 2.349928140640259,
-      "learning_rate": 9.951959798994975e-05,
-      "loss": 5.8961,
-      "step": 979
-    },
-    {
-      "epoch": 0.5110821382007823,
-      "grad_norm": 2.2525947093963623,
-      "learning_rate": 9.951859296482412e-05,
-      "loss": 6.388,
-      "step": 980
-    },
-    {
-      "epoch": 0.5116036505867014,
-      "grad_norm": 2.0856735706329346,
-      "learning_rate": 9.95175879396985e-05,
-      "loss": 6.6325,
-      "step": 981
-    },
-    {
-      "epoch": 0.5121251629726206,
-      "grad_norm": 2.2722132205963135,
-      "learning_rate": 9.951658291457287e-05,
-      "loss": 5.8759,
-      "step": 982
-    },
-    {
-      "epoch": 0.5126466753585398,
-      "grad_norm": 2.305325746536255,
-      "learning_rate": 9.951557788944724e-05,
-      "loss": 6.5283,
-      "step": 983
-    },
-    {
-      "epoch": 0.5131681877444589,
-      "grad_norm": 2.8892693519592285,
-      "learning_rate": 9.951457286432161e-05,
-      "loss": 5.6446,
-      "step": 984
-    },
-    {
-      "epoch": 0.5136897001303781,
-      "grad_norm": 2.2443737983703613,
-      "learning_rate": 9.951356783919599e-05,
-      "loss": 6.1867,
-      "step": 985
-    },
-    {
-      "epoch": 0.5142112125162973,
-      "grad_norm": 2.326246500015259,
-      "learning_rate": 9.951256281407035e-05,
-      "loss": 6.4661,
-      "step": 986
-    },
-    {
-      "epoch": 0.5147327249022164,
-      "grad_norm": 2.3732025623321533,
-      "learning_rate": 9.951155778894473e-05,
-      "loss": 6.0932,
-      "step": 987
-    },
-    {
-      "epoch": 0.5152542372881356,
-      "grad_norm": 2.115880250930786,
-      "learning_rate": 9.951055276381909e-05,
-      "loss": 6.2948,
-      "step": 988
-    },
-    {
-      "epoch": 0.5157757496740547,
-      "grad_norm": 2.152890682220459,
-      "learning_rate": 9.950954773869347e-05,
-      "loss": 6.5644,
-      "step": 989
-    },
-    {
-      "epoch": 0.516297262059974,
-      "grad_norm": 2.4016923904418945,
-      "learning_rate": 9.950854271356785e-05,
-      "loss": 6.6785,
-      "step": 990
-    },
-    {
-      "epoch": 0.5168187744458931,
-      "grad_norm": 2.1402158737182617,
-      "learning_rate": 9.950753768844223e-05,
-      "loss": 6.4978,
-      "step": 991
-    },
-    {
-      "epoch": 0.5173402868318122,
-      "grad_norm": 2.263880729675293,
-      "learning_rate": 9.950653266331659e-05,
-      "loss": 6.4149,
-      "step": 992
-    },
-    {
-      "epoch": 0.5178617992177315,
-      "grad_norm": 2.540372848510742,
-      "learning_rate": 9.950552763819096e-05,
-      "loss": 6.4855,
-      "step": 993
-    },
-    {
-      "epoch": 0.5183833116036506,
-      "grad_norm": 2.118459463119507,
-      "learning_rate": 9.950452261306533e-05,
-      "loss": 6.001,
-      "step": 994
-    },
-    {
-      "epoch": 0.5189048239895697,
-      "grad_norm": 1.845294713973999,
-      "learning_rate": 9.95035175879397e-05,
-      "loss": 6.5425,
-      "step": 995
-    },
-    {
-      "epoch": 0.519426336375489,
-      "grad_norm": 2.0339488983154297,
-      "learning_rate": 9.950251256281408e-05,
-      "loss": 6.5648,
-      "step": 996
-    },
-    {
-      "epoch": 0.5199478487614081,
-      "grad_norm": 2.080930233001709,
-      "learning_rate": 9.950150753768844e-05,
-      "loss": 5.9018,
-      "step": 997
-    },
-    {
-      "epoch": 0.5204693611473272,
-      "grad_norm": 2.14528489112854,
-      "learning_rate": 9.950050251256282e-05,
-      "loss": 6.103,
-      "step": 998
-    },
-    {
-      "epoch": 0.5209908735332465,
-      "grad_norm": 2.3449432849884033,
-      "learning_rate": 9.949949748743718e-05,
-      "loss": 6.5891,
-      "step": 999
-    },
-    {
-      "epoch": 0.5215123859191656,
-      "grad_norm": 1.9489444494247437,
-      "learning_rate": 9.949849246231156e-05,
-      "loss": 6.4501,
-      "step": 1000
-    },
-    {
-      "epoch": 0.5220338983050847,
-      "grad_norm": 2.270794153213501,
-      "learning_rate": 9.949748743718594e-05,
-      "loss": 6.4565,
-      "step": 1001
-    },
-    {
-      "epoch": 0.5225554106910039,
-      "grad_norm": 1.788223147392273,
-      "learning_rate": 9.949648241206032e-05,
-      "loss": 6.5355,
-      "step": 1002
-    },
-    {
-      "epoch": 0.5230769230769231,
-      "grad_norm": 1.8515621423721313,
-      "learning_rate": 9.949547738693468e-05,
-      "loss": 6.565,
-      "step": 1003
-    },
-    {
-      "epoch": 0.5235984354628422,
-      "grad_norm": 2.0171244144439697,
-      "learning_rate": 9.949447236180906e-05,
-      "loss": 6.5739,
-      "step": 1004
-    },
-    {
-      "epoch": 0.5241199478487614,
-      "grad_norm": 1.7724168300628662,
-      "learning_rate": 9.949346733668342e-05,
-      "loss": 6.2755,
-      "step": 1005
-    },
-    {
-      "epoch": 0.5246414602346806,
-      "grad_norm": 2.0184807777404785,
-      "learning_rate": 9.94924623115578e-05,
-      "loss": 6.457,
-      "step": 1006
-    },
-    {
-      "epoch": 0.5251629726205997,
-      "grad_norm": 2.0908501148223877,
-      "learning_rate": 9.949145728643216e-05,
-      "loss": 6.297,
-      "step": 1007
-    },
-    {
-      "epoch": 0.5256844850065189,
-      "grad_norm": 1.9920762777328491,
-      "learning_rate": 9.949045226130653e-05,
-      "loss": 6.3217,
-      "step": 1008
-    },
-    {
-      "epoch": 0.5262059973924381,
-      "grad_norm": 2.2184832096099854,
-      "learning_rate": 9.94894472361809e-05,
-      "loss": 5.7467,
-      "step": 1009
-    },
-    {
-      "epoch": 0.5267275097783573,
-      "grad_norm": 1.9815866947174072,
-      "learning_rate": 9.948844221105528e-05,
-      "loss": 6.1595,
-      "step": 1010
-    },
-    {
-      "epoch": 0.5272490221642764,
-      "grad_norm": 3.5510051250457764,
-      "learning_rate": 9.948743718592966e-05,
-      "loss": 5.9881,
-      "step": 1011
-    },
-    {
-      "epoch": 0.5277705345501955,
-      "grad_norm": 2.057077169418335,
-      "learning_rate": 9.948643216080403e-05,
-      "loss": 6.2812,
-      "step": 1012
-    },
-    {
-      "epoch": 0.5282920469361148,
-      "grad_norm": 2.4590835571289062,
-      "learning_rate": 9.94854271356784e-05,
-      "loss": 6.6273,
-      "step": 1013
-    },
-    {
-      "epoch": 0.5288135593220339,
-      "grad_norm": 1.7845070362091064,
-      "learning_rate": 9.948442211055277e-05,
-      "loss": 6.6754,
-      "step": 1014
-    },
-    {
-      "epoch": 0.529335071707953,
-      "grad_norm": 1.9927841424942017,
-      "learning_rate": 9.948341708542715e-05,
-      "loss": 6.1964,
-      "step": 1015
-    },
-    {
-      "epoch": 0.5298565840938723,
-      "grad_norm": 1.9606233835220337,
-      "learning_rate": 9.948241206030151e-05,
-      "loss": 5.9889,
-      "step": 1016
-    },
-    {
-      "epoch": 0.5303780964797914,
-      "grad_norm": 2.0790512561798096,
-      "learning_rate": 9.948140703517589e-05,
-      "loss": 6.4329,
-      "step": 1017
-    },
-    {
-      "epoch": 0.5308996088657105,
-      "grad_norm": 1.935746431350708,
-      "learning_rate": 9.948040201005025e-05,
-      "loss": 6.4808,
-      "step": 1018
-    },
-    {
-      "epoch": 0.5314211212516298,
-      "grad_norm": 1.5664710998535156,
-      "learning_rate": 9.947939698492463e-05,
-      "loss": 5.8561,
-      "step": 1019
-    },
-    {
-      "epoch": 0.5319426336375489,
-      "grad_norm": 2.1359148025512695,
-      "learning_rate": 9.9478391959799e-05,
-      "loss": 6.5126,
-      "step": 1020
-    },
-    {
-      "epoch": 0.532464146023468,
-      "grad_norm": 2.975498676300049,
-      "learning_rate": 9.947738693467337e-05,
-      "loss": 5.5307,
-      "step": 1021
-    },
-    {
-      "epoch": 0.5329856584093873,
-      "grad_norm": 2.526907444000244,
-      "learning_rate": 9.947638190954775e-05,
-      "loss": 6.3619,
-      "step": 1022
-    },
-    {
-      "epoch": 0.5335071707953064,
-      "grad_norm": 1.9787168502807617,
-      "learning_rate": 9.947537688442211e-05,
-      "loss": 6.4997,
-      "step": 1023
-    },
-    {
-      "epoch": 0.5340286831812255,
-      "grad_norm": 2.2634403705596924,
-      "learning_rate": 9.947437185929649e-05,
-      "loss": 6.4142,
-      "step": 1024
-    },
-    {
-      "epoch": 0.5345501955671447,
-      "grad_norm": 2.2375221252441406,
-      "learning_rate": 9.947336683417086e-05,
-      "loss": 6.6987,
-      "step": 1025
-    },
-    {
-      "epoch": 0.5350717079530639,
-      "grad_norm": 1.8052479028701782,
-      "learning_rate": 9.947236180904523e-05,
-      "loss": 6.7399,
-      "step": 1026
-    },
-    {
-      "epoch": 0.535593220338983,
-      "grad_norm": 2.8728318214416504,
-      "learning_rate": 9.94713567839196e-05,
-      "loss": 6.3754,
-      "step": 1027
-    },
-    {
-      "epoch": 0.5361147327249022,
-      "grad_norm": 2.69226336479187,
-      "learning_rate": 9.947035175879398e-05,
-      "loss": 6.2582,
-      "step": 1028
-    },
-    {
-      "epoch": 0.5366362451108214,
-      "grad_norm": 2.250656843185425,
-      "learning_rate": 9.946934673366834e-05,
-      "loss": 5.9589,
-      "step": 1029
-    },
-    {
-      "epoch": 0.5371577574967406,
-      "grad_norm": 2.127631425857544,
-      "learning_rate": 9.946834170854272e-05,
-      "loss": 5.9818,
-      "step": 1030
-    },
-    {
-      "epoch": 0.5376792698826597,
-      "grad_norm": 1.8615456819534302,
-      "learning_rate": 9.94673366834171e-05,
-      "loss": 6.4924,
-      "step": 1031
-    },
-    {
-      "epoch": 0.5382007822685789,
-      "grad_norm": 2.1967973709106445,
-      "learning_rate": 9.946633165829147e-05,
-      "loss": 6.3876,
-      "step": 1032
-    },
-    {
-      "epoch": 0.5387222946544981,
-      "grad_norm": 1.914030909538269,
-      "learning_rate": 9.946532663316584e-05,
-      "loss": 6.4738,
-      "step": 1033
-    },
-    {
-      "epoch": 0.5392438070404172,
-      "grad_norm": 2.0085256099700928,
-      "learning_rate": 9.94643216080402e-05,
-      "loss": 6.4261,
-      "step": 1034
-    },
-    {
-      "epoch": 0.5397653194263363,
-      "grad_norm": 1.7855981588363647,
-      "learning_rate": 9.946331658291458e-05,
-      "loss": 6.4943,
-      "step": 1035
-    },
-    {
-      "epoch": 0.5402868318122556,
-      "grad_norm": 1.8241933584213257,
-      "learning_rate": 9.946231155778894e-05,
-      "loss": 6.7336,
-      "step": 1036
-    },
-    {
-      "epoch": 0.5408083441981747,
-      "grad_norm": 2.0525999069213867,
-      "learning_rate": 9.946130653266332e-05,
-      "loss": 6.3461,
-      "step": 1037
-    },
-    {
-      "epoch": 0.5413298565840938,
-      "grad_norm": 2.5363781452178955,
-      "learning_rate": 9.946030150753769e-05,
-      "loss": 6.5937,
-      "step": 1038
-    },
-    {
-      "epoch": 0.5418513689700131,
-      "grad_norm": 2.140744209289551,
-      "learning_rate": 9.945929648241206e-05,
-      "loss": 6.5601,
-      "step": 1039
-    },
-    {
-      "epoch": 0.5423728813559322,
-      "grad_norm": 1.9791829586029053,
-      "learning_rate": 9.945829145728643e-05,
-      "loss": 5.9308,
-      "step": 1040
-    },
-    {
-      "epoch": 0.5428943937418513,
-      "grad_norm": 2.283832550048828,
-      "learning_rate": 9.94572864321608e-05,
-      "loss": 6.6413,
-      "step": 1041
-    },
-    {
-      "epoch": 0.5434159061277706,
-      "grad_norm": 1.9554219245910645,
-      "learning_rate": 9.945628140703518e-05,
-      "loss": 6.1505,
-      "step": 1042
-    },
-    {
-      "epoch": 0.5439374185136897,
-      "grad_norm": 2.2553982734680176,
-      "learning_rate": 9.945527638190956e-05,
-      "loss": 6.7015,
-      "step": 1043
-    },
-    {
-      "epoch": 0.5444589308996088,
-      "grad_norm": 2.226177453994751,
-      "learning_rate": 9.945427135678393e-05,
-      "loss": 6.4264,
-      "step": 1044
-    },
-    {
-      "epoch": 0.5449804432855281,
-      "grad_norm": 2.015780210494995,
-      "learning_rate": 9.94532663316583e-05,
-      "loss": 6.4164,
-      "step": 1045
-    },
-    {
-      "epoch": 0.5455019556714472,
-      "grad_norm": 2.360095977783203,
-      "learning_rate": 9.945226130653267e-05,
-      "loss": 6.4349,
-      "step": 1046
-    },
-    {
-      "epoch": 0.5460234680573663,
-      "grad_norm": 2.0980846881866455,
-      "learning_rate": 9.945125628140703e-05,
-      "loss": 6.1571,
-      "step": 1047
-    },
-    {
-      "epoch": 0.5465449804432855,
-      "grad_norm": 2.199629306793213,
-      "learning_rate": 9.945025125628141e-05,
-      "loss": 6.1847,
-      "step": 1048
-    },
-    {
-      "epoch": 0.5470664928292047,
-      "grad_norm": 2.2445294857025146,
-      "learning_rate": 9.944924623115577e-05,
-      "loss": 6.1854,
-      "step": 1049
-    },
-    {
-      "epoch": 0.5475880052151239,
-      "grad_norm": 2.2078707218170166,
-      "learning_rate": 9.944824120603015e-05,
-      "loss": 6.4354,
-      "step": 1050
-    },
-    {
-      "epoch": 0.548109517601043,
-      "grad_norm": 2.116070032119751,
-      "learning_rate": 9.944723618090453e-05,
-      "loss": 6.4002,
-      "step": 1051
-    },
-    {
-      "epoch": 0.5486310299869622,
-      "grad_norm": 2.0001003742218018,
-      "learning_rate": 9.944623115577891e-05,
-      "loss": 6.0909,
-      "step": 1052
-    },
-    {
-      "epoch": 0.5491525423728814,
-      "grad_norm": 1.7960060834884644,
-      "learning_rate": 9.944522613065327e-05,
-      "loss": 6.7664,
-      "step": 1053
-    },
-    {
-      "epoch": 0.5496740547588005,
-      "grad_norm": 1.8234988451004028,
-      "learning_rate": 9.944422110552765e-05,
-      "loss": 6.6887,
-      "step": 1054
-    },
-    {
-      "epoch": 0.5501955671447197,
-      "grad_norm": 2.5000674724578857,
-      "learning_rate": 9.944321608040201e-05,
-      "loss": 6.3386,
-      "step": 1055
-    },
-    {
-      "epoch": 0.5507170795306389,
-      "grad_norm": 2.168358087539673,
-      "learning_rate": 9.944221105527639e-05,
-      "loss": 6.6375,
-      "step": 1056
-    },
-    {
-      "epoch": 0.551238591916558,
-      "grad_norm": 1.9939749240875244,
-      "learning_rate": 9.944120603015076e-05,
-      "loss": 6.0951,
-      "step": 1057
-    },
-    {
-      "epoch": 0.5517601043024772,
-      "grad_norm": 2.2767982482910156,
-      "learning_rate": 9.944020100502513e-05,
-      "loss": 6.1074,
-      "step": 1058
-    },
-    {
-      "epoch": 0.5522816166883964,
-      "grad_norm": 2.473562002182007,
-      "learning_rate": 9.94391959798995e-05,
-      "loss": 6.2916,
-      "step": 1059
-    },
-    {
-      "epoch": 0.5528031290743155,
-      "grad_norm": 2.0120818614959717,
-      "learning_rate": 9.943819095477386e-05,
-      "loss": 5.6506,
-      "step": 1060
-    },
-    {
-      "epoch": 0.5533246414602346,
-      "grad_norm": 2.1312575340270996,
-      "learning_rate": 9.943718592964824e-05,
-      "loss": 6.1445,
-      "step": 1061
-    },
-    {
-      "epoch": 0.5538461538461539,
-      "grad_norm": 2.3592288494110107,
-      "learning_rate": 9.943618090452262e-05,
-      "loss": 6.2612,
-      "step": 1062
-    },
-    {
-      "epoch": 0.554367666232073,
-      "grad_norm": 1.9979387521743774,
-      "learning_rate": 9.9435175879397e-05,
-      "loss": 6.4108,
-      "step": 1063
-    },
-    {
-      "epoch": 0.5548891786179921,
-      "grad_norm": 2.145378828048706,
-      "learning_rate": 9.943417085427136e-05,
-      "loss": 5.9514,
-      "step": 1064
-    },
-    {
-      "epoch": 0.5554106910039114,
-      "grad_norm": 1.7572509050369263,
-      "learning_rate": 9.943316582914574e-05,
-      "loss": 6.4152,
-      "step": 1065
-    },
-    {
-      "epoch": 0.5559322033898305,
-      "grad_norm": 1.8876161575317383,
-      "learning_rate": 9.94321608040201e-05,
-      "loss": 6.4293,
-      "step": 1066
-    },
-    {
-      "epoch": 0.5564537157757496,
-      "grad_norm": 1.8606067895889282,
-      "learning_rate": 9.943115577889448e-05,
-      "loss": 6.6877,
-      "step": 1067
-    },
-    {
-      "epoch": 0.5569752281616689,
-      "grad_norm": 2.1855692863464355,
-      "learning_rate": 9.943015075376885e-05,
-      "loss": 6.1459,
-      "step": 1068
-    },
-    {
-      "epoch": 0.557496740547588,
-      "grad_norm": 2.0761561393737793,
-      "learning_rate": 9.942914572864322e-05,
-      "loss": 6.502,
-      "step": 1069
-    },
-    {
-      "epoch": 0.5580182529335072,
-      "grad_norm": 1.9480054378509521,
-      "learning_rate": 9.942814070351759e-05,
-      "loss": 6.5445,
-      "step": 1070
-    },
-    {
-      "epoch": 0.5585397653194263,
-      "grad_norm": 2.7435081005096436,
-      "learning_rate": 9.942713567839197e-05,
-      "loss": 5.4653,
-      "step": 1071
-    },
-    {
-      "epoch": 0.5590612777053455,
-      "grad_norm": 2.2142210006713867,
-      "learning_rate": 9.942613065326634e-05,
-      "loss": 5.6591,
-      "step": 1072
-    },
-    {
-      "epoch": 0.5595827900912647,
-      "grad_norm": 1.8588271141052246,
-      "learning_rate": 9.942512562814071e-05,
-      "loss": 6.497,
-      "step": 1073
-    },
-    {
-      "epoch": 0.5601043024771838,
-      "grad_norm": 2.1879067420959473,
-      "learning_rate": 9.942412060301508e-05,
-      "loss": 5.8484,
-      "step": 1074
-    },
-    {
-      "epoch": 0.560625814863103,
-      "grad_norm": 2.1053307056427,
-      "learning_rate": 9.942311557788945e-05,
-      "loss": 5.9358,
-      "step": 1075
-    },
-    {
-      "epoch": 0.5611473272490222,
-      "grad_norm": 2.069387197494507,
-      "learning_rate": 9.942211055276383e-05,
-      "loss": 6.5896,
-      "step": 1076
-    },
-    {
-      "epoch": 0.5616688396349413,
-      "grad_norm": 2.250141143798828,
-      "learning_rate": 9.942110552763819e-05,
-      "loss": 6.5784,
-      "step": 1077
-    },
-    {
-      "epoch": 0.5621903520208605,
-      "grad_norm": 1.9212855100631714,
-      "learning_rate": 9.942010050251257e-05,
-      "loss": 6.7726,
-      "step": 1078
-    },
-    {
-      "epoch": 0.5627118644067797,
-      "grad_norm": 1.8462995290756226,
-      "learning_rate": 9.941909547738693e-05,
-      "loss": 6.5867,
-      "step": 1079
-    },
-    {
-      "epoch": 0.5632333767926988,
-      "grad_norm": 2.0089070796966553,
-      "learning_rate": 9.941809045226131e-05,
-      "loss": 5.8652,
-      "step": 1080
-    },
-    {
-      "epoch": 0.563754889178618,
-      "grad_norm": 1.8268921375274658,
-      "learning_rate": 9.941708542713568e-05,
-      "loss": 6.4779,
-      "step": 1081
-    },
-    {
-      "epoch": 0.5642764015645372,
-      "grad_norm": 1.631845474243164,
-      "learning_rate": 9.941608040201005e-05,
-      "loss": 6.6743,
-      "step": 1082
-    },
-    {
-      "epoch": 0.5647979139504563,
-      "grad_norm": 1.6352670192718506,
-      "learning_rate": 9.941507537688443e-05,
-      "loss": 6.3974,
-      "step": 1083
-    },
-    {
-      "epoch": 0.5653194263363754,
-      "grad_norm": 1.7636839151382446,
-      "learning_rate": 9.941407035175881e-05,
-      "loss": 6.6397,
-      "step": 1084
-    },
-    {
-      "epoch": 0.5658409387222947,
-      "grad_norm": 2.027069330215454,
-      "learning_rate": 9.941306532663317e-05,
-      "loss": 6.1071,
-      "step": 1085
-    },
-    {
-      "epoch": 0.5663624511082138,
-      "grad_norm": 1.8907263278961182,
-      "learning_rate": 9.941206030150754e-05,
-      "loss": 6.5588,
-      "step": 1086
-    },
-    {
-      "epoch": 0.566883963494133,
-      "grad_norm": 1.6413065195083618,
-      "learning_rate": 9.941105527638192e-05,
-      "loss": 6.7847,
-      "step": 1087
-    },
-    {
-      "epoch": 0.5674054758800522,
-      "grad_norm": 1.6439162492752075,
-      "learning_rate": 9.941005025125628e-05,
-      "loss": 6.4677,
-      "step": 1088
-    },
-    {
-      "epoch": 0.5679269882659713,
-      "grad_norm": 1.8425122499465942,
-      "learning_rate": 9.940904522613066e-05,
-      "loss": 6.5117,
-      "step": 1089
-    },
-    {
-      "epoch": 0.5684485006518905,
-      "grad_norm": 2.7872750759124756,
-      "learning_rate": 9.940804020100502e-05,
-      "loss": 5.5311,
-      "step": 1090
-    },
-    {
-      "epoch": 0.5689700130378097,
-      "grad_norm": 2.0331318378448486,
-      "learning_rate": 9.94070351758794e-05,
-      "loss": 6.3681,
-      "step": 1091
-    },
-    {
-      "epoch": 0.5694915254237288,
-      "grad_norm": 2.093341588973999,
-      "learning_rate": 9.940603015075376e-05,
-      "loss": 6.4774,
-      "step": 1092
-    },
-    {
-      "epoch": 0.570013037809648,
-      "grad_norm": 2.029984712600708,
-      "learning_rate": 9.940502512562814e-05,
-      "loss": 6.2532,
-      "step": 1093
-    },
-    {
-      "epoch": 0.5705345501955671,
-      "grad_norm": 2.1496570110321045,
-      "learning_rate": 9.940402010050252e-05,
-      "loss": 6.373,
-      "step": 1094
-    },
-    {
-      "epoch": 0.5710560625814863,
-      "grad_norm": 2.257427453994751,
-      "learning_rate": 9.94030150753769e-05,
-      "loss": 5.5611,
-      "step": 1095
-    },
-    {
-      "epoch": 0.5715775749674055,
-      "grad_norm": 2.026304244995117,
-      "learning_rate": 9.940201005025126e-05,
-      "loss": 6.5535,
-      "step": 1096
-    },
-    {
-      "epoch": 0.5720990873533246,
-      "grad_norm": 2.959139823913574,
-      "learning_rate": 9.940100502512564e-05,
-      "loss": 6.1954,
-      "step": 1097
-    },
-    {
-      "epoch": 0.5726205997392438,
-      "grad_norm": 2.200242757797241,
-      "learning_rate": 9.94e-05,
-      "loss": 6.2894,
-      "step": 1098
-    },
-    {
-      "epoch": 0.573142112125163,
-      "grad_norm": 2.430525064468384,
-      "learning_rate": 9.939899497487438e-05,
-      "loss": 5.8289,
-      "step": 1099
-    },
-    {
-      "epoch": 0.5736636245110821,
-      "grad_norm": 1.819159984588623,
-      "learning_rate": 9.939798994974875e-05,
-      "loss": 6.5941,
-      "step": 1100
-    },
-    {
-      "epoch": 0.5741851368970013,
-      "grad_norm": 2.1125643253326416,
-      "learning_rate": 9.939698492462311e-05,
-      "loss": 6.5426,
-      "step": 1101
-    },
-    {
-      "epoch": 0.5747066492829205,
-      "grad_norm": 1.901065707206726,
-      "learning_rate": 9.939597989949749e-05,
-      "loss": 6.5261,
-      "step": 1102
-    },
-    {
-      "epoch": 0.5752281616688396,
-      "grad_norm": 2.6235954761505127,
-      "learning_rate": 9.939497487437187e-05,
-      "loss": 5.2949,
-      "step": 1103
-    },
-    {
-      "epoch": 0.5757496740547589,
-      "grad_norm": 2.304426670074463,
-      "learning_rate": 9.939396984924624e-05,
-      "loss": 6.2389,
-      "step": 1104
-    },
-    {
-      "epoch": 0.576271186440678,
-      "grad_norm": 1.906935453414917,
-      "learning_rate": 9.939296482412061e-05,
-      "loss": 6.1475,
-      "step": 1105
-    },
-    {
-      "epoch": 0.5767926988265971,
-      "grad_norm": 1.9593114852905273,
-      "learning_rate": 9.939195979899499e-05,
-      "loss": 6.3794,
-      "step": 1106
-    },
-    {
-      "epoch": 0.5773142112125162,
-      "grad_norm": 1.981544852256775,
-      "learning_rate": 9.939095477386935e-05,
-      "loss": 6.2827,
-      "step": 1107
-    },
-    {
-      "epoch": 0.5778357235984355,
-      "grad_norm": 1.8649400472640991,
-      "learning_rate": 9.938994974874373e-05,
-      "loss": 6.3705,
-      "step": 1108
-    },
-    {
-      "epoch": 0.5783572359843546,
-      "grad_norm": 1.8049644231796265,
-      "learning_rate": 9.938894472361809e-05,
-      "loss": 6.3085,
-      "step": 1109
-    },
-    {
-      "epoch": 0.5788787483702738,
-      "grad_norm": 2.325085163116455,
-      "learning_rate": 9.938793969849247e-05,
-      "loss": 6.0134,
-      "step": 1110
-    },
-    {
-      "epoch": 0.579400260756193,
-      "grad_norm": 1.9665300846099854,
-      "learning_rate": 9.938693467336683e-05,
-      "loss": 6.4956,
-      "step": 1111
-    },
-    {
-      "epoch": 0.5799217731421121,
-      "grad_norm": 2.0801262855529785,
-      "learning_rate": 9.938592964824121e-05,
-      "loss": 6.7524,
-      "step": 1112
-    },
-    {
-      "epoch": 0.5804432855280313,
-      "grad_norm": 1.720389485359192,
-      "learning_rate": 9.938492462311558e-05,
-      "loss": 6.6196,
-      "step": 1113
-    },
-    {
-      "epoch": 0.5809647979139505,
-      "grad_norm": 2.0624289512634277,
-      "learning_rate": 9.938391959798995e-05,
-      "loss": 6.532,
-      "step": 1114
-    },
-    {
-      "epoch": 0.5814863102998696,
-      "grad_norm": 2.2252767086029053,
-      "learning_rate": 9.938291457286433e-05,
-      "loss": 5.5149,
-      "step": 1115
-    },
-    {
-      "epoch": 0.5820078226857888,
-      "grad_norm": 1.7989685535430908,
-      "learning_rate": 9.93819095477387e-05,
-      "loss": 6.6338,
-      "step": 1116
-    },
-    {
-      "epoch": 0.5825293350717079,
-      "grad_norm": 2.1111786365509033,
-      "learning_rate": 9.938090452261307e-05,
-      "loss": 6.3048,
-      "step": 1117
-    },
-    {
-      "epoch": 0.5830508474576271,
-      "grad_norm": 1.7639018297195435,
-      "learning_rate": 9.937989949748744e-05,
-      "loss": 6.4724,
-      "step": 1118
-    },
-    {
-      "epoch": 0.5835723598435463,
-      "grad_norm": 1.664967656135559,
-      "learning_rate": 9.937889447236182e-05,
-      "loss": 6.7072,
-      "step": 1119
-    },
-    {
-      "epoch": 0.5840938722294654,
-      "grad_norm": 2.8242104053497314,
-      "learning_rate": 9.937788944723618e-05,
-      "loss": 5.9272,
-      "step": 1120
-    },
-    {
-      "epoch": 0.5846153846153846,
-      "grad_norm": 1.8767706155776978,
-      "learning_rate": 9.937688442211056e-05,
-      "loss": 6.6593,
-      "step": 1121
-    },
-    {
-      "epoch": 0.5851368970013038,
-      "grad_norm": 2.1955630779266357,
-      "learning_rate": 9.937587939698492e-05,
-      "loss": 6.3052,
-      "step": 1122
-    },
-    {
-      "epoch": 0.5856584093872229,
-      "grad_norm": 2.174100875854492,
-      "learning_rate": 9.93748743718593e-05,
-      "loss": 6.4872,
-      "step": 1123
-    },
-    {
-      "epoch": 0.5861799217731422,
-      "grad_norm": 2.197890043258667,
-      "learning_rate": 9.937386934673368e-05,
-      "loss": 6.3364,
-      "step": 1124
-    },
-    {
-      "epoch": 0.5867014341590613,
-      "grad_norm": 1.782468318939209,
-      "learning_rate": 9.937286432160806e-05,
-      "loss": 6.3847,
-      "step": 1125
-    },
-    {
-      "epoch": 0.5872229465449804,
-      "grad_norm": 2.293619394302368,
-      "learning_rate": 9.937185929648242e-05,
-      "loss": 6.2027,
-      "step": 1126
-    },
-    {
-      "epoch": 0.5877444589308997,
-      "grad_norm": 1.9395240545272827,
-      "learning_rate": 9.937085427135678e-05,
-      "loss": 6.1196,
-      "step": 1127
-    },
-    {
-      "epoch": 0.5882659713168188,
-      "grad_norm": 1.8216606378555298,
-      "learning_rate": 9.936984924623116e-05,
-      "loss": 6.3899,
-      "step": 1128
-    },
-    {
-      "epoch": 0.5887874837027379,
-      "grad_norm": 2.6699256896972656,
-      "learning_rate": 9.936884422110553e-05,
-      "loss": 5.9487,
-      "step": 1129
-    },
-    {
-      "epoch": 0.5893089960886571,
-      "grad_norm": 1.7982321977615356,
-      "learning_rate": 9.93678391959799e-05,
-      "loss": 6.0509,
-      "step": 1130
-    },
-    {
-      "epoch": 0.5898305084745763,
-      "grad_norm": 2.2850613594055176,
-      "learning_rate": 9.936683417085427e-05,
-      "loss": 6.0642,
-      "step": 1131
-    },
-    {
-      "epoch": 0.5903520208604954,
-      "grad_norm": 1.7728103399276733,
-      "learning_rate": 9.936582914572865e-05,
-      "loss": 6.3612,
-      "step": 1132
-    },
-    {
-      "epoch": 0.5908735332464146,
-      "grad_norm": 1.8440884351730347,
-      "learning_rate": 9.936482412060301e-05,
-      "loss": 6.4031,
-      "step": 1133
-    },
-    {
-      "epoch": 0.5913950456323338,
-      "grad_norm": 1.899545431137085,
-      "learning_rate": 9.936381909547739e-05,
-      "loss": 6.047,
-      "step": 1134
-    },
-    {
-      "epoch": 0.5919165580182529,
-      "grad_norm": 1.9492884874343872,
-      "learning_rate": 9.936281407035177e-05,
-      "loss": 6.229,
-      "step": 1135
-    },
-    {
-      "epoch": 0.5924380704041721,
-      "grad_norm": 2.0436065196990967,
-      "learning_rate": 9.936180904522614e-05,
-      "loss": 6.541,
-      "step": 1136
-    },
-    {
-      "epoch": 0.5929595827900913,
-      "grad_norm": 1.8518743515014648,
-      "learning_rate": 9.936080402010051e-05,
-      "loss": 6.4035,
-      "step": 1137
-    },
-    {
-      "epoch": 0.5934810951760104,
-      "grad_norm": 2.591897487640381,
-      "learning_rate": 9.935979899497489e-05,
-      "loss": 5.8126,
-      "step": 1138
-    },
-    {
-      "epoch": 0.5940026075619296,
-      "grad_norm": 1.6692888736724854,
-      "learning_rate": 9.935879396984925e-05,
-      "loss": 6.6656,
-      "step": 1139
-    },
-    {
-      "epoch": 0.5945241199478487,
-      "grad_norm": 1.7847602367401123,
-      "learning_rate": 9.935778894472362e-05,
-      "loss": 6.1819,
-      "step": 1140
-    },
-    {
-      "epoch": 0.595045632333768,
-      "grad_norm": 2.053975820541382,
-      "learning_rate": 9.935678391959799e-05,
-      "loss": 6.2015,
-      "step": 1141
-    },
-    {
-      "epoch": 0.5955671447196871,
-      "grad_norm": 2.16721510887146,
-      "learning_rate": 9.935577889447236e-05,
-      "loss": 6.322,
-      "step": 1142
-    },
-    {
-      "epoch": 0.5960886571056062,
-      "grad_norm": 2.0767982006073,
-      "learning_rate": 9.935477386934673e-05,
-      "loss": 5.7837,
-      "step": 1143
-    },
-    {
-      "epoch": 0.5966101694915255,
-      "grad_norm": 2.1614692211151123,
-      "learning_rate": 9.935376884422111e-05,
-      "loss": 6.2995,
-      "step": 1144
-    },
-    {
-      "epoch": 0.5971316818774446,
-      "grad_norm": 1.9652820825576782,
-      "learning_rate": 9.935276381909549e-05,
-      "loss": 6.2763,
-      "step": 1145
-    },
-    {
-      "epoch": 0.5976531942633637,
-      "grad_norm": 1.8606866598129272,
-      "learning_rate": 9.935175879396985e-05,
-      "loss": 6.3984,
-      "step": 1146
-    },
-    {
-      "epoch": 0.598174706649283,
-      "grad_norm": 3.1722075939178467,
-      "learning_rate": 9.935075376884423e-05,
-      "loss": 5.8785,
-      "step": 1147
-    },
-    {
-      "epoch": 0.5986962190352021,
-      "grad_norm": 1.7217590808868408,
-      "learning_rate": 9.93497487437186e-05,
-      "loss": 6.613,
-      "step": 1148
-    },
-    {
-      "epoch": 0.5992177314211212,
-      "grad_norm": 2.163100481033325,
-      "learning_rate": 9.934874371859297e-05,
-      "loss": 6.4155,
-      "step": 1149
-    },
-    {
-      "epoch": 0.5997392438070405,
-      "grad_norm": 1.902317762374878,
-      "learning_rate": 9.934773869346734e-05,
-      "loss": 6.3051,
-      "step": 1150
-    },
-    {
-      "epoch": 0.6002607561929596,
-      "grad_norm": 1.8997035026550293,
-      "learning_rate": 9.934673366834172e-05,
-      "loss": 6.5693,
-      "step": 1151
-    },
-    {
-      "epoch": 0.6007822685788787,
-      "grad_norm": 1.8708505630493164,
-      "learning_rate": 9.934572864321608e-05,
-      "loss": 6.4053,
-      "step": 1152
-    },
-    {
-      "epoch": 0.6013037809647979,
-      "grad_norm": 1.93083655834198,
-      "learning_rate": 9.934472361809045e-05,
-      "loss": 6.1654,
-      "step": 1153
-    },
-    {
-      "epoch": 0.6018252933507171,
-      "grad_norm": 1.918582558631897,
-      "learning_rate": 9.934371859296482e-05,
-      "loss": 5.8484,
-      "step": 1154
-    },
-    {
-      "epoch": 0.6023468057366362,
-      "grad_norm": 1.9105820655822754,
-      "learning_rate": 9.93427135678392e-05,
-      "loss": 6.4204,
-      "step": 1155
-    },
-    {
-      "epoch": 0.6028683181225554,
-      "grad_norm": 2.1229636669158936,
-      "learning_rate": 9.934170854271358e-05,
-      "loss": 6.3691,
-      "step": 1156
-    },
-    {
-      "epoch": 0.6033898305084746,
-      "grad_norm": 2.0528697967529297,
-      "learning_rate": 9.934070351758794e-05,
-      "loss": 6.735,
-      "step": 1157
-    },
-    {
-      "epoch": 0.6039113428943937,
-      "grad_norm": 2.188812732696533,
-      "learning_rate": 9.933969849246232e-05,
-      "loss": 6.3616,
-      "step": 1158
-    },
-    {
-      "epoch": 0.6044328552803129,
-      "grad_norm": 2.288559913635254,
-      "learning_rate": 9.933869346733669e-05,
-      "loss": 6.547,
-      "step": 1159
-    },
-    {
-      "epoch": 0.6049543676662321,
-      "grad_norm": 1.7310538291931152,
-      "learning_rate": 9.933768844221106e-05,
-      "loss": 6.6614,
-      "step": 1160
-    },
-    {
-      "epoch": 0.6054758800521512,
-      "grad_norm": 2.2639284133911133,
-      "learning_rate": 9.933668341708543e-05,
-      "loss": 6.3984,
-      "step": 1161
-    },
-    {
-      "epoch": 0.6059973924380704,
-      "grad_norm": 1.7019494771957397,
-      "learning_rate": 9.93356783919598e-05,
-      "loss": 6.5026,
-      "step": 1162
-    },
-    {
-      "epoch": 0.6065189048239896,
-      "grad_norm": 1.9687600135803223,
-      "learning_rate": 9.933467336683417e-05,
-      "loss": 6.4789,
-      "step": 1163
-    },
-    {
-      "epoch": 0.6070404172099088,
-      "grad_norm": 1.8984894752502441,
-      "learning_rate": 9.933366834170855e-05,
-      "loss": 6.4769,
-      "step": 1164
-    },
-    {
-      "epoch": 0.6075619295958279,
-      "grad_norm": 2.1345672607421875,
-      "learning_rate": 9.933266331658293e-05,
-      "loss": 6.1621,
-      "step": 1165
-    },
-    {
-      "epoch": 0.608083441981747,
-      "grad_norm": 2.157698392868042,
-      "learning_rate": 9.933165829145729e-05,
-      "loss": 6.2369,
-      "step": 1166
-    },
-    {
-      "epoch": 0.6086049543676663,
-      "grad_norm": 2.080808639526367,
-      "learning_rate": 9.933065326633167e-05,
-      "loss": 6.2962,
-      "step": 1167
-    },
-    {
-      "epoch": 0.6091264667535854,
-      "grad_norm": 1.8234134912490845,
-      "learning_rate": 9.932964824120603e-05,
-      "loss": 6.3768,
-      "step": 1168
-    },
-    {
-      "epoch": 0.6096479791395045,
-      "grad_norm": 1.9063458442687988,
-      "learning_rate": 9.932864321608041e-05,
-      "loss": 6.4957,
-      "step": 1169
-    },
-    {
-      "epoch": 0.6101694915254238,
-      "grad_norm": 1.779676914215088,
-      "learning_rate": 9.932763819095477e-05,
-      "loss": 6.2882,
-      "step": 1170
-    },
-    {
-      "epoch": 0.6106910039113429,
-      "grad_norm": 1.87655508518219,
-      "learning_rate": 9.932663316582915e-05,
-      "loss": 6.5468,
-      "step": 1171
-    },
-    {
-      "epoch": 0.611212516297262,
-      "grad_norm": 1.9955928325653076,
-      "learning_rate": 9.932562814070352e-05,
-      "loss": 5.9997,
-      "step": 1172
-    },
-    {
-      "epoch": 0.6117340286831813,
-      "grad_norm": 3.1170473098754883,
-      "learning_rate": 9.93246231155779e-05,
-      "loss": 6.168,
-      "step": 1173
-    },
-    {
-      "epoch": 0.6122555410691004,
-      "grad_norm": 2.2072954177856445,
-      "learning_rate": 9.932361809045226e-05,
-      "loss": 6.4796,
-      "step": 1174
-    },
-    {
-      "epoch": 0.6127770534550195,
-      "grad_norm": 2.2578601837158203,
-      "learning_rate": 9.932261306532664e-05,
-      "loss": 6.3757,
-      "step": 1175
-    },
-    {
-      "epoch": 0.6132985658409387,
-      "grad_norm": 2.2653486728668213,
-      "learning_rate": 9.932160804020101e-05,
-      "loss": 5.8629,
-      "step": 1176
-    },
-    {
-      "epoch": 0.6138200782268579,
-      "grad_norm": 2.1006174087524414,
-      "learning_rate": 9.932060301507539e-05,
-      "loss": 6.3708,
-      "step": 1177
-    },
-    {
-      "epoch": 0.614341590612777,
-      "grad_norm": 2.440850019454956,
-      "learning_rate": 9.931959798994976e-05,
-      "loss": 5.9661,
-      "step": 1178
-    },
-    {
-      "epoch": 0.6148631029986962,
-      "grad_norm": 1.9377793073654175,
-      "learning_rate": 9.931859296482413e-05,
-      "loss": 6.5853,
-      "step": 1179
-    },
-    {
-      "epoch": 0.6153846153846154,
-      "grad_norm": 1.788891077041626,
-      "learning_rate": 9.93175879396985e-05,
-      "loss": 6.4273,
-      "step": 1180
-    },
-    {
-      "epoch": 0.6159061277705346,
-      "grad_norm": 2.273021697998047,
-      "learning_rate": 9.931658291457286e-05,
-      "loss": 6.5507,
-      "step": 1181
-    },
-    {
-      "epoch": 0.6164276401564537,
-      "grad_norm": 1.8314498662948608,
-      "learning_rate": 9.931557788944724e-05,
-      "loss": 6.4072,
-      "step": 1182
-    },
-    {
-      "epoch": 0.6169491525423729,
-      "grad_norm": 1.9090126752853394,
-      "learning_rate": 9.93145728643216e-05,
-      "loss": 6.455,
-      "step": 1183
-    },
-    {
-      "epoch": 0.6174706649282921,
-      "grad_norm": 1.825119137763977,
-      "learning_rate": 9.931356783919598e-05,
-      "loss": 6.7681,
-      "step": 1184
-    },
-    {
-      "epoch": 0.6179921773142112,
-      "grad_norm": 1.8277357816696167,
-      "learning_rate": 9.931256281407036e-05,
-      "loss": 6.5272,
-      "step": 1185
-    },
-    {
-      "epoch": 0.6185136897001304,
-      "grad_norm": 2.247853994369507,
-      "learning_rate": 9.931155778894474e-05,
-      "loss": 5.9878,
-      "step": 1186
-    },
-    {
-      "epoch": 0.6190352020860496,
-      "grad_norm": 1.789474368095398,
-      "learning_rate": 9.93105527638191e-05,
-      "loss": 6.3125,
-      "step": 1187
-    },
-    {
-      "epoch": 0.6195567144719687,
-      "grad_norm": 2.32488751411438,
-      "learning_rate": 9.930954773869348e-05,
-      "loss": 5.7663,
-      "step": 1188
-    },
-    {
-      "epoch": 0.6200782268578878,
-      "grad_norm": 2.320885181427002,
-      "learning_rate": 9.930854271356784e-05,
-      "loss": 5.5406,
-      "step": 1189
-    },
-    {
-      "epoch": 0.6205997392438071,
-      "grad_norm": 1.9397251605987549,
-      "learning_rate": 9.930753768844222e-05,
-      "loss": 6.3136,
-      "step": 1190
-    },
-    {
-      "epoch": 0.6211212516297262,
-      "grad_norm": 1.7406009435653687,
-      "learning_rate": 9.930653266331659e-05,
-      "loss": 6.4718,
-      "step": 1191
-    },
-    {
-      "epoch": 0.6216427640156453,
-      "grad_norm": 2.0325028896331787,
-      "learning_rate": 9.930552763819096e-05,
-      "loss": 5.8915,
-      "step": 1192
-    },
-    {
-      "epoch": 0.6221642764015646,
-      "grad_norm": 2.1372628211975098,
-      "learning_rate": 9.930452261306533e-05,
-      "loss": 5.5932,
-      "step": 1193
-    },
-    {
-      "epoch": 0.6226857887874837,
-      "grad_norm": 2.163978099822998,
-      "learning_rate": 9.930351758793969e-05,
-      "loss": 5.8113,
-      "step": 1194
-    },
-    {
-      "epoch": 0.6232073011734028,
-      "grad_norm": 2.0743110179901123,
-      "learning_rate": 9.930251256281407e-05,
-      "loss": 6.4113,
-      "step": 1195
-    },
-    {
-      "epoch": 0.6237288135593221,
-      "grad_norm": 2.2374720573425293,
-      "learning_rate": 9.930150753768845e-05,
-      "loss": 6.1799,
-      "step": 1196
-    },
-    {
-      "epoch": 0.6242503259452412,
-      "grad_norm": 1.8585360050201416,
-      "learning_rate": 9.930050251256283e-05,
-      "loss": 6.4773,
-      "step": 1197
-    },
-    {
-      "epoch": 0.6247718383311603,
-      "grad_norm": 1.854195475578308,
-      "learning_rate": 9.929949748743719e-05,
-      "loss": 6.5368,
-      "step": 1198
-    },
-    {
-      "epoch": 0.6252933507170795,
-      "grad_norm": 1.9343338012695312,
-      "learning_rate": 9.929849246231157e-05,
-      "loss": 6.5308,
-      "step": 1199
-    },
-    {
-      "epoch": 0.6258148631029987,
-      "grad_norm": 1.9238635301589966,
-      "learning_rate": 9.929748743718593e-05,
-      "loss": 6.1491,
-      "step": 1200
-    },
-    {
-      "epoch": 0.6263363754889179,
-      "grad_norm": 2.309617280960083,
-      "learning_rate": 9.929648241206031e-05,
-      "loss": 6.5636,
-      "step": 1201
-    },
-    {
-      "epoch": 0.626857887874837,
-      "grad_norm": 1.8611011505126953,
-      "learning_rate": 9.929547738693467e-05,
-      "loss": 6.3989,
-      "step": 1202
-    },
-    {
-      "epoch": 0.6273794002607562,
-      "grad_norm": 2.3958685398101807,
-      "learning_rate": 9.929447236180905e-05,
-      "loss": 6.0467,
-      "step": 1203
-    },
-    {
-      "epoch": 0.6279009126466754,
-      "grad_norm": 2.3826165199279785,
-      "learning_rate": 9.929346733668342e-05,
-      "loss": 6.0121,
-      "step": 1204
-    },
-    {
-      "epoch": 0.6284224250325945,
-      "grad_norm": 2.1858012676239014,
-      "learning_rate": 9.92924623115578e-05,
-      "loss": 6.5982,
-      "step": 1205
-    },
-    {
-      "epoch": 0.6289439374185137,
-      "grad_norm": 1.8589776754379272,
-      "learning_rate": 9.929145728643217e-05,
-      "loss": 6.6164,
-      "step": 1206
-    },
-    {
-      "epoch": 0.6294654498044329,
-      "grad_norm": 1.6967905759811401,
-      "learning_rate": 9.929045226130654e-05,
-      "loss": 6.0849,
-      "step": 1207
-    },
-    {
-      "epoch": 0.629986962190352,
-      "grad_norm": 1.9245147705078125,
-      "learning_rate": 9.928944723618091e-05,
-      "loss": 6.126,
-      "step": 1208
-    },
-    {
-      "epoch": 0.6305084745762712,
-      "grad_norm": 1.9393119812011719,
-      "learning_rate": 9.928844221105528e-05,
-      "loss": 6.4905,
-      "step": 1209
-    },
-    {
-      "epoch": 0.6310299869621904,
-      "grad_norm": 1.8244280815124512,
-      "learning_rate": 9.928743718592966e-05,
-      "loss": 6.3695,
-      "step": 1210
-    },
-    {
-      "epoch": 0.6315514993481095,
-      "grad_norm": 2.2182328701019287,
-      "learning_rate": 9.928643216080402e-05,
-      "loss": 5.9904,
-      "step": 1211
-    },
-    {
-      "epoch": 0.6320730117340286,
-      "grad_norm": 1.8193329572677612,
-      "learning_rate": 9.92854271356784e-05,
-      "loss": 6.4533,
-      "step": 1212
-    },
-    {
-      "epoch": 0.6325945241199479,
-      "grad_norm": 1.874064326286316,
-      "learning_rate": 9.928442211055276e-05,
-      "loss": 6.4634,
-      "step": 1213
-    },
-    {
-      "epoch": 0.633116036505867,
-      "grad_norm": 2.1150290966033936,
-      "learning_rate": 9.928341708542714e-05,
-      "loss": 6.2769,
-      "step": 1214
-    },
-    {
-      "epoch": 0.6336375488917861,
-      "grad_norm": 1.9426897764205933,
-      "learning_rate": 9.92824120603015e-05,
-      "loss": 6.3897,
-      "step": 1215
-    },
-    {
-      "epoch": 0.6341590612777054,
-      "grad_norm": 1.9069862365722656,
-      "learning_rate": 9.928140703517588e-05,
-      "loss": 6.0429,
-      "step": 1216
-    },
-    {
-      "epoch": 0.6346805736636245,
-      "grad_norm": 1.8784018754959106,
-      "learning_rate": 9.928040201005026e-05,
-      "loss": 6.5544,
-      "step": 1217
-    },
-    {
-      "epoch": 0.6352020860495436,
-      "grad_norm": 1.7192256450653076,
-      "learning_rate": 9.927939698492464e-05,
-      "loss": 6.4387,
-      "step": 1218
-    },
-    {
-      "epoch": 0.6357235984354629,
-      "grad_norm": 1.593495488166809,
-      "learning_rate": 9.9278391959799e-05,
-      "loss": 6.5836,
-      "step": 1219
-    },
-    {
-      "epoch": 0.636245110821382,
-      "grad_norm": 1.7476557493209839,
-      "learning_rate": 9.927738693467337e-05,
-      "loss": 6.2185,
-      "step": 1220
-    },
-    {
-      "epoch": 0.6367666232073012,
-      "grad_norm": 1.6952474117279053,
-      "learning_rate": 9.927638190954774e-05,
-      "loss": 6.4145,
-      "step": 1221
-    },
-    {
-      "epoch": 0.6372881355932203,
-      "grad_norm": 1.852982521057129,
-      "learning_rate": 9.927537688442211e-05,
-      "loss": 6.4145,
-      "step": 1222
-    },
-    {
-      "epoch": 0.6378096479791395,
-      "grad_norm": 1.8660837411880493,
-      "learning_rate": 9.927437185929649e-05,
-      "loss": 6.0361,
-      "step": 1223
-    },
-    {
-      "epoch": 0.6383311603650587,
-      "grad_norm": 1.7795859575271606,
-      "learning_rate": 9.927336683417085e-05,
-      "loss": 6.2594,
-      "step": 1224
-    },
-    {
-      "epoch": 0.6388526727509778,
-      "grad_norm": 1.7390978336334229,
-      "learning_rate": 9.927236180904523e-05,
-      "loss": 6.2544,
-      "step": 1225
-    },
-    {
-      "epoch": 0.639374185136897,
-      "grad_norm": 1.9701423645019531,
-      "learning_rate": 9.927135678391961e-05,
-      "loss": 6.1094,
-      "step": 1226
-    },
-    {
-      "epoch": 0.6398956975228162,
-      "grad_norm": 2.1514768600463867,
-      "learning_rate": 9.927035175879398e-05,
-      "loss": 6.4946,
-      "step": 1227
-    },
-    {
-      "epoch": 0.6404172099087353,
-      "grad_norm": 1.6748924255371094,
-      "learning_rate": 9.926934673366835e-05,
-      "loss": 6.6379,
-      "step": 1228
-    },
-    {
-      "epoch": 0.6409387222946545,
-      "grad_norm": 2.049246311187744,
-      "learning_rate": 9.926834170854273e-05,
-      "loss": 6.0481,
-      "step": 1229
-    },
-    {
-      "epoch": 0.6414602346805737,
-      "grad_norm": 2.0611886978149414,
-      "learning_rate": 9.926733668341709e-05,
-      "loss": 6.5221,
-      "step": 1230
-    },
-    {
-      "epoch": 0.6419817470664928,
-      "grad_norm": 2.0248711109161377,
-      "learning_rate": 9.926633165829147e-05,
-      "loss": 6.4768,
-      "step": 1231
-    },
-    {
-      "epoch": 0.642503259452412,
-      "grad_norm": 1.9739201068878174,
-      "learning_rate": 9.926532663316583e-05,
-      "loss": 6.5902,
-      "step": 1232
-    },
-    {
-      "epoch": 0.6430247718383312,
-      "grad_norm": 2.2955453395843506,
-      "learning_rate": 9.92643216080402e-05,
-      "loss": 6.2011,
-      "step": 1233
-    },
-    {
-      "epoch": 0.6435462842242503,
-      "grad_norm": 2.43315052986145,
-      "learning_rate": 9.926331658291458e-05,
-      "loss": 6.4694,
-      "step": 1234
-    },
-    {
-      "epoch": 0.6440677966101694,
-      "grad_norm": 2.2616562843322754,
-      "learning_rate": 9.926231155778894e-05,
-      "loss": 6.0659,
-      "step": 1235
-    },
-    {
-      "epoch": 0.6445893089960887,
-      "grad_norm": 1.7571403980255127,
-      "learning_rate": 9.926130653266332e-05,
-      "loss": 6.574,
-      "step": 1236
-    },
-    {
-      "epoch": 0.6451108213820078,
-      "grad_norm": 2.2987818717956543,
-      "learning_rate": 9.92603015075377e-05,
-      "loss": 5.557,
-      "step": 1237
-    },
-    {
-      "epoch": 0.645632333767927,
-      "grad_norm": 1.8230011463165283,
-      "learning_rate": 9.925929648241207e-05,
-      "loss": 6.251,
-      "step": 1238
-    },
-    {
-      "epoch": 0.6461538461538462,
-      "grad_norm": 2.1173017024993896,
-      "learning_rate": 9.925829145728644e-05,
-      "loss": 6.4815,
-      "step": 1239
-    },
-    {
-      "epoch": 0.6466753585397653,
-      "grad_norm": 1.8613569736480713,
-      "learning_rate": 9.925728643216082e-05,
-      "loss": 6.6348,
-      "step": 1240
-    },
-    {
-      "epoch": 0.6471968709256845,
-      "grad_norm": 1.6213207244873047,
-      "learning_rate": 9.925628140703518e-05,
-      "loss": 6.5309,
-      "step": 1241
-    },
-    {
-      "epoch": 0.6477183833116037,
-      "grad_norm": 2.0249369144439697,
-      "learning_rate": 9.925527638190956e-05,
-      "loss": 6.5142,
-      "step": 1242
-    },
-    {
-      "epoch": 0.6482398956975228,
-      "grad_norm": 2.0371780395507812,
-      "learning_rate": 9.925427135678392e-05,
-      "loss": 6.3588,
-      "step": 1243
-    },
-    {
-      "epoch": 0.648761408083442,
-      "grad_norm": 2.447199583053589,
-      "learning_rate": 9.92532663316583e-05,
-      "loss": 6.3764,
-      "step": 1244
-    },
-    {
-      "epoch": 0.6492829204693612,
-      "grad_norm": 1.746205449104309,
-      "learning_rate": 9.925226130653266e-05,
-      "loss": 6.3675,
-      "step": 1245
-    },
-    {
-      "epoch": 0.6498044328552803,
-      "grad_norm": 2.0738072395324707,
-      "learning_rate": 9.925125628140703e-05,
-      "loss": 5.9435,
-      "step": 1246
-    },
-    {
-      "epoch": 0.6503259452411995,
-      "grad_norm": 2.0077099800109863,
-      "learning_rate": 9.92502512562814e-05,
-      "loss": 6.0831,
-      "step": 1247
-    },
-    {
-      "epoch": 0.6508474576271186,
-      "grad_norm": 1.974311351776123,
-      "learning_rate": 9.924924623115578e-05,
-      "loss": 6.3185,
-      "step": 1248
-    },
-    {
-      "epoch": 0.6513689700130378,
-      "grad_norm": 1.8912620544433594,
-      "learning_rate": 9.924824120603016e-05,
-      "loss": 6.7106,
-      "step": 1249
-    },
-    {
-      "epoch": 0.651890482398957,
-      "grad_norm": 1.8888311386108398,
-      "learning_rate": 9.924723618090453e-05,
-      "loss": 6.4595,
-      "step": 1250
-    },
-    {
-      "epoch": 0.6524119947848761,
-      "grad_norm": 1.9805779457092285,
-      "learning_rate": 9.92462311557789e-05,
-      "loss": 6.5483,
-      "step": 1251
-    },
-    {
-      "epoch": 0.6529335071707953,
-      "grad_norm": 2.192660093307495,
-      "learning_rate": 9.924522613065327e-05,
-      "loss": 6.4706,
-      "step": 1252
-    },
-    {
-      "epoch": 0.6534550195567145,
-      "grad_norm": 1.837544560432434,
-      "learning_rate": 9.924422110552765e-05,
-      "loss": 6.4827,
-      "step": 1253
-    },
-    {
-      "epoch": 0.6539765319426336,
-      "grad_norm": 1.8572523593902588,
-      "learning_rate": 9.924321608040201e-05,
-      "loss": 6.4135,
-      "step": 1254
-    },
-    {
-      "epoch": 0.6544980443285529,
-      "grad_norm": 2.07061767578125,
-      "learning_rate": 9.924221105527639e-05,
-      "loss": 6.0722,
-      "step": 1255
-    },
-    {
-      "epoch": 0.655019556714472,
-      "grad_norm": 2.2605385780334473,
-      "learning_rate": 9.924120603015075e-05,
-      "loss": 5.9388,
-      "step": 1256
-    },
-    {
-      "epoch": 0.6555410691003911,
-      "grad_norm": 2.2526824474334717,
-      "learning_rate": 9.924020100502513e-05,
-      "loss": 5.918,
-      "step": 1257
-    },
-    {
-      "epoch": 0.6560625814863102,
-      "grad_norm": 1.8854278326034546,
-      "learning_rate": 9.923919597989951e-05,
-      "loss": 6.0172,
-      "step": 1258
-    },
-    {
-      "epoch": 0.6565840938722295,
-      "grad_norm": 1.7376154661178589,
-      "learning_rate": 9.923819095477387e-05,
-      "loss": 5.8039,
-      "step": 1259
-    },
-    {
-      "epoch": 0.6571056062581486,
-      "grad_norm": 1.967217206954956,
-      "learning_rate": 9.923718592964825e-05,
-      "loss": 5.795,
-      "step": 1260
-    },
-    {
-      "epoch": 0.6576271186440678,
-      "grad_norm": 1.7940503358840942,
-      "learning_rate": 9.923618090452261e-05,
-      "loss": 6.5282,
-      "step": 1261
-    },
-    {
-      "epoch": 0.658148631029987,
-      "grad_norm": 1.6767634153366089,
-      "learning_rate": 9.923517587939699e-05,
-      "loss": 6.6692,
-      "step": 1262
-    },
-    {
-      "epoch": 0.6586701434159061,
-      "grad_norm": 2.1176092624664307,
-      "learning_rate": 9.923417085427136e-05,
-      "loss": 6.3115,
-      "step": 1263
-    },
-    {
-      "epoch": 0.6591916558018253,
-      "grad_norm": 2.351166248321533,
-      "learning_rate": 9.923316582914573e-05,
-      "loss": 6.158,
-      "step": 1264
-    },
-    {
-      "epoch": 0.6597131681877445,
-      "grad_norm": 2.062457323074341,
-      "learning_rate": 9.92321608040201e-05,
-      "loss": 4.877,
-      "step": 1265
-    },
-    {
-      "epoch": 0.6602346805736636,
-      "grad_norm": 2.187218427658081,
-      "learning_rate": 9.923115577889448e-05,
-      "loss": 6.1125,
-      "step": 1266
-    },
-    {
-      "epoch": 0.6607561929595828,
-      "grad_norm": 2.3859918117523193,
-      "learning_rate": 9.923015075376884e-05,
-      "loss": 5.9177,
-      "step": 1267
-    },
-    {
-      "epoch": 0.661277705345502,
-      "grad_norm": 2.109041213989258,
-      "learning_rate": 9.922914572864322e-05,
-      "loss": 6.5515,
-      "step": 1268
-    },
-    {
-      "epoch": 0.6617992177314211,
-      "grad_norm": 2.451289415359497,
-      "learning_rate": 9.92281407035176e-05,
-      "loss": 6.0999,
-      "step": 1269
-    },
-    {
-      "epoch": 0.6623207301173403,
-      "grad_norm": 1.7792937755584717,
-      "learning_rate": 9.922713567839197e-05,
-      "loss": 6.3901,
-      "step": 1270
-    },
-    {
-      "epoch": 0.6628422425032594,
-      "grad_norm": 2.1809239387512207,
-      "learning_rate": 9.922613065326634e-05,
-      "loss": 5.8402,
-      "step": 1271
-    },
-    {
-      "epoch": 0.6633637548891786,
-      "grad_norm": 1.8536064624786377,
-      "learning_rate": 9.922512562814072e-05,
-      "loss": 6.301,
-      "step": 1272
-    },
-    {
-      "epoch": 0.6638852672750978,
-      "grad_norm": 1.9039344787597656,
-      "learning_rate": 9.922412060301508e-05,
-      "loss": 6.2659,
-      "step": 1273
-    },
-    {
-      "epoch": 0.6644067796610169,
-      "grad_norm": 2.198302984237671,
-      "learning_rate": 9.922311557788944e-05,
-      "loss": 5.824,
-      "step": 1274
-    },
-    {
-      "epoch": 0.6649282920469362,
-      "grad_norm": 1.8464465141296387,
-      "learning_rate": 9.922211055276382e-05,
-      "loss": 6.401,
-      "step": 1275
-    },
-    {
-      "epoch": 0.6654498044328553,
-      "grad_norm": 1.6977571249008179,
-      "learning_rate": 9.922110552763819e-05,
-      "loss": 6.2889,
-      "step": 1276
-    },
-    {
-      "epoch": 0.6659713168187744,
-      "grad_norm": 1.9438045024871826,
-      "learning_rate": 9.922010050251256e-05,
-      "loss": 6.6902,
-      "step": 1277
-    },
-    {
-      "epoch": 0.6664928292046937,
-      "grad_norm": 1.8544456958770752,
-      "learning_rate": 9.921909547738694e-05,
-      "loss": 6.3197,
-      "step": 1278
-    },
-    {
-      "epoch": 0.6670143415906128,
-      "grad_norm": 2.1300861835479736,
-      "learning_rate": 9.921809045226132e-05,
-      "loss": 6.284,
-      "step": 1279
-    },
-    {
-      "epoch": 0.6675358539765319,
-      "grad_norm": 2.162370204925537,
-      "learning_rate": 9.921708542713568e-05,
-      "loss": 6.545,
-      "step": 1280
-    },
-    {
-      "epoch": 0.668057366362451,
-      "grad_norm": 2.0180537700653076,
-      "learning_rate": 9.921608040201006e-05,
-      "loss": 6.4824,
-      "step": 1281
-    },
-    {
-      "epoch": 0.6685788787483703,
-      "grad_norm": 1.8317381143569946,
-      "learning_rate": 9.921507537688443e-05,
-      "loss": 6.3651,
-      "step": 1282
-    },
-    {
-      "epoch": 0.6691003911342894,
-      "grad_norm": 1.8794056177139282,
-      "learning_rate": 9.92140703517588e-05,
-      "loss": 6.5026,
-      "step": 1283
-    },
-    {
-      "epoch": 0.6696219035202086,
-      "grad_norm": 1.8696671724319458,
-      "learning_rate": 9.921306532663317e-05,
-      "loss": 6.3876,
-      "step": 1284
-    },
-    {
-      "epoch": 0.6701434159061278,
-      "grad_norm": 1.8482470512390137,
-      "learning_rate": 9.921206030150755e-05,
-      "loss": 6.389,
-      "step": 1285
-    },
-    {
-      "epoch": 0.6706649282920469,
-      "grad_norm": 1.7337751388549805,
-      "learning_rate": 9.921105527638191e-05,
-      "loss": 6.4841,
-      "step": 1286
-    },
-    {
-      "epoch": 0.6711864406779661,
-      "grad_norm": 2.004901647567749,
-      "learning_rate": 9.921005025125627e-05,
-      "loss": 5.9165,
-      "step": 1287
-    },
-    {
-      "epoch": 0.6717079530638853,
-      "grad_norm": 2.050809383392334,
-      "learning_rate": 9.920904522613065e-05,
-      "loss": 6.1752,
-      "step": 1288
-    },
-    {
-      "epoch": 0.6722294654498044,
-      "grad_norm": 1.867122769355774,
-      "learning_rate": 9.920804020100503e-05,
-      "loss": 6.0666,
-      "step": 1289
-    },
-    {
-      "epoch": 0.6727509778357236,
-      "grad_norm": 1.7209299802780151,
-      "learning_rate": 9.920703517587941e-05,
-      "loss": 6.2589,
-      "step": 1290
-    },
-    {
-      "epoch": 0.6732724902216428,
-      "grad_norm": 1.833649754524231,
-      "learning_rate": 9.920603015075377e-05,
-      "loss": 6.3143,
-      "step": 1291
-    },
-    {
-      "epoch": 0.673794002607562,
-      "grad_norm": 1.895180344581604,
-      "learning_rate": 9.920502512562815e-05,
-      "loss": 6.3483,
-      "step": 1292
-    },
-    {
-      "epoch": 0.6743155149934811,
-      "grad_norm": 2.510172128677368,
-      "learning_rate": 9.920402010050251e-05,
-      "loss": 6.0177,
-      "step": 1293
-    },
-    {
-      "epoch": 0.6748370273794002,
-      "grad_norm": 1.880961298942566,
-      "learning_rate": 9.920301507537689e-05,
-      "loss": 6.184,
-      "step": 1294
-    },
-    {
-      "epoch": 0.6753585397653195,
-      "grad_norm": 1.997738242149353,
-      "learning_rate": 9.920201005025126e-05,
-      "loss": 6.4835,
-      "step": 1295
-    },
-    {
-      "epoch": 0.6758800521512386,
-      "grad_norm": 1.7524503469467163,
-      "learning_rate": 9.920100502512563e-05,
-      "loss": 6.5082,
-      "step": 1296
-    },
-    {
-      "epoch": 0.6764015645371577,
-      "grad_norm": 1.8719860315322876,
-      "learning_rate": 9.92e-05,
-      "loss": 6.1785,
-      "step": 1297
-    },
-    {
-      "epoch": 0.676923076923077,
-      "grad_norm": 2.0365631580352783,
-      "learning_rate": 9.919899497487438e-05,
-      "loss": 6.6612,
-      "step": 1298
-    },
-    {
-      "epoch": 0.6774445893089961,
-      "grad_norm": 1.6351118087768555,
-      "learning_rate": 9.919798994974875e-05,
-      "loss": 6.6928,
-      "step": 1299
-    },
-    {
-      "epoch": 0.6779661016949152,
-      "grad_norm": 2.222693920135498,
-      "learning_rate": 9.919698492462312e-05,
-      "loss": 6.0948,
-      "step": 1300
-    },
-    {
-      "epoch": 0.6784876140808345,
-      "grad_norm": 2.037600517272949,
-      "learning_rate": 9.91959798994975e-05,
-      "loss": 5.8376,
-      "step": 1301
-    },
-    {
-      "epoch": 0.6790091264667536,
-      "grad_norm": 1.7934489250183105,
-      "learning_rate": 9.919497487437186e-05,
-      "loss": 6.1895,
-      "step": 1302
-    },
-    {
-      "epoch": 0.6795306388526727,
-      "grad_norm": 1.9172230958938599,
-      "learning_rate": 9.919396984924624e-05,
-      "loss": 5.7558,
-      "step": 1303
-    },
-    {
-      "epoch": 0.6800521512385919,
-      "grad_norm": 2.754615068435669,
-      "learning_rate": 9.91929648241206e-05,
-      "loss": 5.9009,
-      "step": 1304
-    },
-    {
-      "epoch": 0.6805736636245111,
-      "grad_norm": 2.217742443084717,
-      "learning_rate": 9.919195979899498e-05,
-      "loss": 6.3688,
-      "step": 1305
-    },
-    {
-      "epoch": 0.6810951760104302,
-      "grad_norm": 1.9266620874404907,
-      "learning_rate": 9.919095477386935e-05,
-      "loss": 6.1365,
-      "step": 1306
-    },
-    {
-      "epoch": 0.6816166883963494,
-      "grad_norm": 2.358166217803955,
-      "learning_rate": 9.918994974874372e-05,
-      "loss": 6.5645,
-      "step": 1307
-    },
-    {
-      "epoch": 0.6821382007822686,
-      "grad_norm": 2.411017417907715,
-      "learning_rate": 9.918894472361809e-05,
-      "loss": 6.2203,
-      "step": 1308
-    },
-    {
-      "epoch": 0.6826597131681877,
-      "grad_norm": 1.6456941366195679,
-      "learning_rate": 9.918793969849247e-05,
-      "loss": 6.2156,
-      "step": 1309
-    },
-    {
-      "epoch": 0.6831812255541069,
-      "grad_norm": 2.295139789581299,
-      "learning_rate": 9.918693467336684e-05,
-      "loss": 6.4303,
-      "step": 1310
-    },
-    {
-      "epoch": 0.6837027379400261,
-      "grad_norm": 2.089681625366211,
-      "learning_rate": 9.918592964824122e-05,
-      "loss": 5.9316,
-      "step": 1311
-    },
-    {
-      "epoch": 0.6842242503259452,
-      "grad_norm": 2.998058319091797,
-      "learning_rate": 9.918492462311559e-05,
-      "loss": 6.3628,
-      "step": 1312
-    },
-    {
-      "epoch": 0.6847457627118644,
-      "grad_norm": 2.1547327041625977,
-      "learning_rate": 9.918391959798995e-05,
-      "loss": 5.8071,
-      "step": 1313
-    },
-    {
-      "epoch": 0.6852672750977836,
-      "grad_norm": 1.840842366218567,
-      "learning_rate": 9.918291457286433e-05,
-      "loss": 6.4315,
-      "step": 1314
-    },
-    {
-      "epoch": 0.6857887874837028,
-      "grad_norm": 2.0396242141723633,
-      "learning_rate": 9.918190954773869e-05,
-      "loss": 5.8021,
-      "step": 1315
-    },
-    {
-      "epoch": 0.6863102998696219,
-      "grad_norm": 1.7985059022903442,
-      "learning_rate": 9.918090452261307e-05,
-      "loss": 6.1555,
-      "step": 1316
-    },
-    {
-      "epoch": 0.686831812255541,
-      "grad_norm": 1.8200675249099731,
-      "learning_rate": 9.917989949748743e-05,
-      "loss": 6.2221,
-      "step": 1317
-    },
-    {
-      "epoch": 0.6873533246414603,
-      "grad_norm": 1.65859055519104,
-      "learning_rate": 9.917889447236181e-05,
-      "loss": 6.3274,
-      "step": 1318
-    },
-    {
-      "epoch": 0.6878748370273794,
-      "grad_norm": 2.2409865856170654,
-      "learning_rate": 9.917788944723619e-05,
-      "loss": 6.2698,
-      "step": 1319
-    },
-    {
-      "epoch": 0.6883963494132985,
-      "grad_norm": 1.8096379041671753,
-      "learning_rate": 9.917688442211057e-05,
-      "loss": 6.1951,
-      "step": 1320
-    },
-    {
-      "epoch": 0.6889178617992178,
-      "grad_norm": 1.733154058456421,
-      "learning_rate": 9.917587939698493e-05,
-      "loss": 6.5585,
-      "step": 1321
-    },
-    {
-      "epoch": 0.6894393741851369,
-      "grad_norm": 1.6778079271316528,
-      "learning_rate": 9.917487437185931e-05,
-      "loss": 6.3903,
-      "step": 1322
-    },
-    {
-      "epoch": 0.689960886571056,
-      "grad_norm": 1.8648881912231445,
-      "learning_rate": 9.917386934673367e-05,
-      "loss": 6.0461,
-      "step": 1323
-    },
-    {
-      "epoch": 0.6904823989569753,
-      "grad_norm": 1.7817151546478271,
-      "learning_rate": 9.917286432160805e-05,
-      "loss": 5.8143,
-      "step": 1324
-    },
-    {
-      "epoch": 0.6910039113428944,
-      "grad_norm": 1.9449816942214966,
-      "learning_rate": 9.917185929648242e-05,
-      "loss": 6.1508,
-      "step": 1325
-    },
-    {
-      "epoch": 0.6915254237288135,
-      "grad_norm": 1.79619562625885,
-      "learning_rate": 9.917085427135678e-05,
-      "loss": 6.3163,
-      "step": 1326
-    },
-    {
-      "epoch": 0.6920469361147328,
-      "grad_norm": 1.4900660514831543,
-      "learning_rate": 9.916984924623116e-05,
-      "loss": 6.5095,
-      "step": 1327
-    },
-    {
-      "epoch": 0.6925684485006519,
-      "grad_norm": 1.7384716272354126,
-      "learning_rate": 9.916884422110552e-05,
-      "loss": 6.5707,
-      "step": 1328
-    },
-    {
-      "epoch": 0.693089960886571,
-      "grad_norm": 1.6734282970428467,
-      "learning_rate": 9.91678391959799e-05,
-      "loss": 6.5893,
-      "step": 1329
-    },
-    {
-      "epoch": 0.6936114732724902,
-      "grad_norm": 1.81753671169281,
-      "learning_rate": 9.916683417085428e-05,
-      "loss": 5.9292,
-      "step": 1330
-    },
-    {
-      "epoch": 0.6941329856584094,
-      "grad_norm": 2.5046372413635254,
-      "learning_rate": 9.916582914572866e-05,
-      "loss": 5.8917,
-      "step": 1331
-    },
-    {
-      "epoch": 0.6946544980443285,
-      "grad_norm": 1.9453530311584473,
-      "learning_rate": 9.916482412060302e-05,
-      "loss": 6.4001,
-      "step": 1332
-    },
-    {
-      "epoch": 0.6951760104302477,
-      "grad_norm": 2.032595634460449,
-      "learning_rate": 9.91638190954774e-05,
-      "loss": 6.527,
-      "step": 1333
-    },
-    {
-      "epoch": 0.6956975228161669,
-      "grad_norm": 2.470912218093872,
-      "learning_rate": 9.916281407035176e-05,
-      "loss": 5.1709,
-      "step": 1334
-    },
-    {
-      "epoch": 0.6962190352020861,
-      "grad_norm": 2.646174430847168,
-      "learning_rate": 9.916180904522614e-05,
-      "loss": 6.122,
-      "step": 1335
-    },
-    {
-      "epoch": 0.6967405475880052,
-      "grad_norm": 2.1192989349365234,
-      "learning_rate": 9.91608040201005e-05,
-      "loss": 6.1945,
-      "step": 1336
-    },
-    {
-      "epoch": 0.6972620599739244,
-      "grad_norm": 2.0626115798950195,
-      "learning_rate": 9.915979899497488e-05,
-      "loss": 5.863,
-      "step": 1337
-    },
-    {
-      "epoch": 0.6977835723598436,
-      "grad_norm": 1.9882649183273315,
-      "learning_rate": 9.915879396984925e-05,
-      "loss": 6.5519,
-      "step": 1338
-    },
-    {
-      "epoch": 0.6983050847457627,
-      "grad_norm": 1.8993260860443115,
-      "learning_rate": 9.915778894472362e-05,
-      "loss": 6.4309,
-      "step": 1339
-    },
-    {
-      "epoch": 0.6988265971316818,
-      "grad_norm": 1.897606611251831,
-      "learning_rate": 9.9156783919598e-05,
-      "loss": 6.1618,
-      "step": 1340
-    },
-    {
-      "epoch": 0.6993481095176011,
-      "grad_norm": 1.8617669343948364,
-      "learning_rate": 9.915577889447237e-05,
-      "loss": 6.468,
-      "step": 1341
-    },
-    {
-      "epoch": 0.6998696219035202,
-      "grad_norm": 1.8829879760742188,
-      "learning_rate": 9.915477386934674e-05,
-      "loss": 6.144,
-      "step": 1342
-    },
-    {
-      "epoch": 0.7003911342894393,
-      "grad_norm": 2.290776491165161,
-      "learning_rate": 9.915376884422111e-05,
-      "loss": 6.3441,
-      "step": 1343
-    },
-    {
-      "epoch": 0.7009126466753586,
-      "grad_norm": 1.8587820529937744,
-      "learning_rate": 9.915276381909549e-05,
-      "loss": 6.2323,
-      "step": 1344
-    },
-    {
-      "epoch": 0.7014341590612777,
-      "grad_norm": 2.2007639408111572,
-      "learning_rate": 9.915175879396985e-05,
-      "loss": 6.229,
-      "step": 1345
-    },
-    {
-      "epoch": 0.7019556714471968,
-      "grad_norm": 1.9643776416778564,
-      "learning_rate": 9.915075376884423e-05,
-      "loss": 6.2255,
-      "step": 1346
-    },
-    {
-      "epoch": 0.7024771838331161,
-      "grad_norm": 1.7954367399215698,
-      "learning_rate": 9.914974874371859e-05,
-      "loss": 6.4048,
-      "step": 1347
-    },
-    {
-      "epoch": 0.7029986962190352,
-      "grad_norm": 1.8727362155914307,
-      "learning_rate": 9.914874371859297e-05,
-      "loss": 6.0571,
-      "step": 1348
-    },
-    {
-      "epoch": 0.7035202086049543,
-      "grad_norm": 1.7820097208023071,
-      "learning_rate": 9.914773869346733e-05,
-      "loss": 6.4511,
-      "step": 1349
-    },
-    {
-      "epoch": 0.7040417209908736,
-      "grad_norm": 1.9630566835403442,
-      "learning_rate": 9.914673366834171e-05,
-      "loss": 6.0497,
-      "step": 1350
-    },
-    {
-      "epoch": 0.7045632333767927,
-      "grad_norm": 2.033202648162842,
-      "learning_rate": 9.914572864321609e-05,
-      "loss": 5.5071,
-      "step": 1351
-    },
-    {
-      "epoch": 0.7050847457627119,
-      "grad_norm": 2.129666566848755,
-      "learning_rate": 9.914472361809045e-05,
-      "loss": 6.2073,
-      "step": 1352
-    },
-    {
-      "epoch": 0.705606258148631,
-      "grad_norm": 2.2263948917388916,
-      "learning_rate": 9.914371859296483e-05,
-      "loss": 6.3499,
-      "step": 1353
-    },
-    {
-      "epoch": 0.7061277705345502,
-      "grad_norm": 1.6847009658813477,
-      "learning_rate": 9.91427135678392e-05,
-      "loss": 6.4733,
-      "step": 1354
-    },
-    {
-      "epoch": 0.7066492829204694,
-      "grad_norm": 2.293405771255493,
-      "learning_rate": 9.914170854271357e-05,
-      "loss": 5.9952,
-      "step": 1355
-    },
-    {
-      "epoch": 0.7071707953063885,
-      "grad_norm": 1.9283884763717651,
-      "learning_rate": 9.914070351758794e-05,
-      "loss": 6.4086,
-      "step": 1356
-    },
-    {
-      "epoch": 0.7076923076923077,
-      "grad_norm": 2.0034866333007812,
-      "learning_rate": 9.913969849246232e-05,
-      "loss": 6.5055,
-      "step": 1357
-    },
-    {
-      "epoch": 0.7082138200782269,
-      "grad_norm": 1.8390480279922485,
-      "learning_rate": 9.913869346733668e-05,
-      "loss": 6.4887,
-      "step": 1358
-    },
-    {
-      "epoch": 0.708735332464146,
-      "grad_norm": 1.684147596359253,
-      "learning_rate": 9.913768844221106e-05,
-      "loss": 6.2046,
-      "step": 1359
-    },
-    {
-      "epoch": 0.7092568448500652,
-      "grad_norm": 2.04543399810791,
-      "learning_rate": 9.913668341708544e-05,
-      "loss": 6.0661,
-      "step": 1360
-    },
-    {
-      "epoch": 0.7097783572359844,
-      "grad_norm": 1.8137935400009155,
-      "learning_rate": 9.913567839195981e-05,
-      "loss": 6.0942,
-      "step": 1361
-    },
-    {
-      "epoch": 0.7102998696219035,
-      "grad_norm": 1.6837714910507202,
-      "learning_rate": 9.913467336683418e-05,
-      "loss": 6.2989,
-      "step": 1362
-    },
-    {
-      "epoch": 0.7108213820078226,
-      "grad_norm": 2.3682711124420166,
-      "learning_rate": 9.913366834170856e-05,
-      "loss": 6.2318,
-      "step": 1363
-    },
-    {
-      "epoch": 0.7113428943937419,
-      "grad_norm": 1.941892385482788,
-      "learning_rate": 9.913266331658292e-05,
-      "loss": 6.0932,
-      "step": 1364
-    },
-    {
-      "epoch": 0.711864406779661,
-      "grad_norm": 2.045935869216919,
-      "learning_rate": 9.91316582914573e-05,
-      "loss": 5.7107,
-      "step": 1365
-    },
-    {
-      "epoch": 0.7123859191655801,
-      "grad_norm": 1.923986792564392,
-      "learning_rate": 9.913065326633166e-05,
-      "loss": 6.3409,
-      "step": 1366
-    },
-    {
-      "epoch": 0.7129074315514994,
-      "grad_norm": 1.878115177154541,
-      "learning_rate": 9.912964824120603e-05,
-      "loss": 6.1724,
-      "step": 1367
-    },
-    {
-      "epoch": 0.7134289439374185,
-      "grad_norm": 1.9086147546768188,
-      "learning_rate": 9.91286432160804e-05,
-      "loss": 6.2958,
-      "step": 1368
-    },
-    {
-      "epoch": 0.7139504563233376,
-      "grad_norm": 1.8487077951431274,
-      "learning_rate": 9.912763819095477e-05,
-      "loss": 6.2885,
-      "step": 1369
-    },
-    {
-      "epoch": 0.7144719687092569,
-      "grad_norm": 1.9221127033233643,
-      "learning_rate": 9.912663316582915e-05,
-      "loss": 6.079,
-      "step": 1370
-    },
-    {
-      "epoch": 0.714993481095176,
-      "grad_norm": 1.7881758213043213,
-      "learning_rate": 9.912562814070352e-05,
-      "loss": 6.4793,
-      "step": 1371
-    },
-    {
-      "epoch": 0.7155149934810952,
-      "grad_norm": 2.078024387359619,
-      "learning_rate": 9.91246231155779e-05,
-      "loss": 6.5469,
-      "step": 1372
-    },
-    {
-      "epoch": 0.7160365058670144,
-      "grad_norm": 2.1184799671173096,
-      "learning_rate": 9.912361809045227e-05,
-      "loss": 6.1762,
-      "step": 1373
-    },
-    {
-      "epoch": 0.7165580182529335,
-      "grad_norm": 1.7050516605377197,
-      "learning_rate": 9.912261306532664e-05,
-      "loss": 6.5793,
-      "step": 1374
-    },
-    {
-      "epoch": 0.7170795306388527,
-      "grad_norm": 1.6978082656860352,
-      "learning_rate": 9.912160804020101e-05,
-      "loss": 6.4262,
-      "step": 1375
-    },
-    {
-      "epoch": 0.7176010430247718,
-      "grad_norm": 1.8031973838806152,
-      "learning_rate": 9.912060301507539e-05,
-      "loss": 6.4686,
-      "step": 1376
-    },
-    {
-      "epoch": 0.718122555410691,
-      "grad_norm": 2.0421676635742188,
-      "learning_rate": 9.911959798994975e-05,
-      "loss": 6.0667,
-      "step": 1377
-    },
-    {
-      "epoch": 0.7186440677966102,
-      "grad_norm": 1.668738842010498,
-      "learning_rate": 9.911859296482413e-05,
-      "loss": 6.6257,
-      "step": 1378
-    },
-    {
-      "epoch": 0.7191655801825293,
-      "grad_norm": 1.8086986541748047,
-      "learning_rate": 9.911758793969849e-05,
-      "loss": 6.1899,
-      "step": 1379
-    },
-    {
-      "epoch": 0.7196870925684485,
-      "grad_norm": 1.7863272428512573,
-      "learning_rate": 9.911658291457287e-05,
-      "loss": 6.4031,
-      "step": 1380
-    },
-    {
-      "epoch": 0.7202086049543677,
-      "grad_norm": 1.7395962476730347,
-      "learning_rate": 9.911557788944725e-05,
-      "loss": 6.5891,
-      "step": 1381
-    },
-    {
-      "epoch": 0.7207301173402868,
-      "grad_norm": 2.206829309463501,
-      "learning_rate": 9.911457286432161e-05,
-      "loss": 5.8696,
-      "step": 1382
-    },
-    {
-      "epoch": 0.721251629726206,
-      "grad_norm": 1.7793954610824585,
-      "learning_rate": 9.911356783919599e-05,
-      "loss": 6.4146,
-      "step": 1383
-    },
-    {
-      "epoch": 0.7217731421121252,
-      "grad_norm": 1.937822699546814,
-      "learning_rate": 9.911256281407035e-05,
-      "loss": 5.8145,
-      "step": 1384
-    },
-    {
-      "epoch": 0.7222946544980443,
-      "grad_norm": 1.963646650314331,
-      "learning_rate": 9.911155778894473e-05,
-      "loss": 6.1204,
-      "step": 1385
-    },
-    {
-      "epoch": 0.7228161668839634,
-      "grad_norm": 2.0328638553619385,
-      "learning_rate": 9.91105527638191e-05,
-      "loss": 6.1133,
-      "step": 1386
-    },
-    {
-      "epoch": 0.7233376792698827,
-      "grad_norm": 2.0667145252227783,
-      "learning_rate": 9.910954773869347e-05,
-      "loss": 6.0303,
-      "step": 1387
-    },
-    {
-      "epoch": 0.7238591916558018,
-      "grad_norm": 1.812839150428772,
-      "learning_rate": 9.910854271356784e-05,
-      "loss": 6.547,
-      "step": 1388
-    },
-    {
-      "epoch": 0.724380704041721,
-      "grad_norm": 1.7212921380996704,
-      "learning_rate": 9.910753768844222e-05,
-      "loss": 6.521,
-      "step": 1389
-    },
-    {
-      "epoch": 0.7249022164276402,
-      "grad_norm": 1.8122483491897583,
-      "learning_rate": 9.910653266331658e-05,
-      "loss": 6.5369,
-      "step": 1390
-    },
-    {
-      "epoch": 0.7254237288135593,
-      "grad_norm": 1.9334791898727417,
-      "learning_rate": 9.910552763819096e-05,
-      "loss": 5.997,
-      "step": 1391
-    },
-    {
-      "epoch": 0.7259452411994785,
-      "grad_norm": 1.8224554061889648,
-      "learning_rate": 9.910452261306534e-05,
-      "loss": 6.3573,
-      "step": 1392
-    },
-    {
-      "epoch": 0.7264667535853977,
-      "grad_norm": 1.8500584363937378,
-      "learning_rate": 9.91035175879397e-05,
-      "loss": 6.3884,
-      "step": 1393
-    },
-    {
-      "epoch": 0.7269882659713168,
-      "grad_norm": 2.17646861076355,
-      "learning_rate": 9.910251256281408e-05,
-      "loss": 6.4312,
-      "step": 1394
-    },
-    {
-      "epoch": 0.727509778357236,
-      "grad_norm": 2.113665819168091,
-      "learning_rate": 9.910150753768844e-05,
-      "loss": 6.0552,
-      "step": 1395
-    },
-    {
-      "epoch": 0.7280312907431552,
-      "grad_norm": 1.659449815750122,
-      "learning_rate": 9.910050251256282e-05,
-      "loss": 6.3511,
-      "step": 1396
-    },
-    {
-      "epoch": 0.7285528031290743,
-      "grad_norm": 1.7127867937088013,
-      "learning_rate": 9.909949748743719e-05,
-      "loss": 6.3485,
-      "step": 1397
-    },
-    {
-      "epoch": 0.7290743155149935,
-      "grad_norm": 1.88431715965271,
-      "learning_rate": 9.909849246231156e-05,
-      "loss": 6.5242,
-      "step": 1398
-    },
-    {
-      "epoch": 0.7295958279009126,
-      "grad_norm": 1.7146869897842407,
-      "learning_rate": 9.909748743718593e-05,
-      "loss": 6.4066,
-      "step": 1399
-    },
-    {
-      "epoch": 0.7301173402868318,
-      "grad_norm": 1.7314929962158203,
-      "learning_rate": 9.90964824120603e-05,
-      "loss": 6.2814,
-      "step": 1400
-    },
-    {
-      "epoch": 0.730638852672751,
-      "grad_norm": 1.812191367149353,
-      "learning_rate": 9.909547738693468e-05,
-      "loss": 6.0934,
-      "step": 1401
-    },
-    {
-      "epoch": 0.7311603650586701,
-      "grad_norm": 1.746126651763916,
-      "learning_rate": 9.909447236180906e-05,
-      "loss": 6.4359,
-      "step": 1402
-    },
-    {
-      "epoch": 0.7316818774445893,
-      "grad_norm": 2.3535757064819336,
-      "learning_rate": 9.909346733668343e-05,
-      "loss": 6.5909,
-      "step": 1403
-    },
-    {
-      "epoch": 0.7322033898305085,
-      "grad_norm": 1.9074891805648804,
-      "learning_rate": 9.90924623115578e-05,
-      "loss": 6.0014,
-      "step": 1404
-    },
-    {
-      "epoch": 0.7327249022164276,
-      "grad_norm": 1.8047109842300415,
-      "learning_rate": 9.909145728643217e-05,
-      "loss": 6.3232,
-      "step": 1405
-    },
-    {
-      "epoch": 0.7332464146023469,
-      "grad_norm": 1.673736572265625,
-      "learning_rate": 9.909045226130653e-05,
-      "loss": 6.6315,
-      "step": 1406
-    },
-    {
-      "epoch": 0.733767926988266,
-      "grad_norm": 1.802794337272644,
-      "learning_rate": 9.908944723618091e-05,
-      "loss": 5.9461,
-      "step": 1407
-    },
-    {
-      "epoch": 0.7342894393741851,
-      "grad_norm": 2.095456123352051,
-      "learning_rate": 9.908844221105527e-05,
-      "loss": 5.9153,
-      "step": 1408
-    },
-    {
-      "epoch": 0.7348109517601042,
-      "grad_norm": 2.0065979957580566,
-      "learning_rate": 9.908743718592965e-05,
-      "loss": 6.2348,
-      "step": 1409
-    },
-    {
-      "epoch": 0.7353324641460235,
-      "grad_norm": 1.992843747138977,
-      "learning_rate": 9.908643216080402e-05,
-      "loss": 6.4291,
-      "step": 1410
-    },
-    {
-      "epoch": 0.7358539765319426,
-      "grad_norm": 1.914299726486206,
-      "learning_rate": 9.90854271356784e-05,
-      "loss": 6.351,
-      "step": 1411
-    },
-    {
-      "epoch": 0.7363754889178618,
-      "grad_norm": 1.7779158353805542,
-      "learning_rate": 9.908442211055277e-05,
-      "loss": 6.261,
-      "step": 1412
-    },
-    {
-      "epoch": 0.736897001303781,
-      "grad_norm": 1.99246346950531,
-      "learning_rate": 9.908341708542715e-05,
-      "loss": 6.1437,
-      "step": 1413
-    },
-    {
-      "epoch": 0.7374185136897001,
-      "grad_norm": 2.0438947677612305,
-      "learning_rate": 9.908241206030151e-05,
-      "loss": 6.3093,
-      "step": 1414
-    },
-    {
-      "epoch": 0.7379400260756193,
-      "grad_norm": 2.0654006004333496,
-      "learning_rate": 9.908140703517589e-05,
-      "loss": 6.209,
-      "step": 1415
-    },
-    {
-      "epoch": 0.7384615384615385,
-      "grad_norm": 1.7370989322662354,
-      "learning_rate": 9.908040201005026e-05,
-      "loss": 6.3714,
-      "step": 1416
-    },
-    {
-      "epoch": 0.7389830508474576,
-      "grad_norm": 1.9075545072555542,
-      "learning_rate": 9.907939698492463e-05,
-      "loss": 6.3272,
-      "step": 1417
-    },
-    {
-      "epoch": 0.7395045632333768,
-      "grad_norm": 1.9900084733963013,
-      "learning_rate": 9.9078391959799e-05,
-      "loss": 6.4492,
-      "step": 1418
-    },
-    {
-      "epoch": 0.740026075619296,
-      "grad_norm": 2.02169132232666,
-      "learning_rate": 9.907738693467336e-05,
-      "loss": 6.4734,
-      "step": 1419
-    },
-    {
-      "epoch": 0.7405475880052151,
-      "grad_norm": 2.2172529697418213,
-      "learning_rate": 9.907638190954774e-05,
-      "loss": 6.0227,
-      "step": 1420
-    },
-    {
-      "epoch": 0.7410691003911343,
-      "grad_norm": 2.0385639667510986,
-      "learning_rate": 9.90753768844221e-05,
-      "loss": 6.2609,
-      "step": 1421
-    },
-    {
-      "epoch": 0.7415906127770534,
-      "grad_norm": 1.70989990234375,
-      "learning_rate": 9.907437185929648e-05,
-      "loss": 6.5667,
-      "step": 1422
-    },
-    {
-      "epoch": 0.7421121251629726,
-      "grad_norm": 1.5108897686004639,
-      "learning_rate": 9.907336683417086e-05,
-      "loss": 6.5336,
-      "step": 1423
-    },
-    {
-      "epoch": 0.7426336375488918,
-      "grad_norm": 1.7108489274978638,
-      "learning_rate": 9.907236180904524e-05,
-      "loss": 5.6998,
-      "step": 1424
-    },
-    {
-      "epoch": 0.7431551499348109,
-      "grad_norm": 2.171581745147705,
-      "learning_rate": 9.90713567839196e-05,
-      "loss": 5.648,
-      "step": 1425
-    },
-    {
-      "epoch": 0.7436766623207302,
-      "grad_norm": 1.7710875272750854,
-      "learning_rate": 9.907035175879398e-05,
-      "loss": 6.0819,
-      "step": 1426
-    },
-    {
-      "epoch": 0.7441981747066493,
-      "grad_norm": 1.806067705154419,
-      "learning_rate": 9.906934673366834e-05,
-      "loss": 6.199,
-      "step": 1427
-    },
-    {
-      "epoch": 0.7447196870925684,
-      "grad_norm": 1.7951360940933228,
-      "learning_rate": 9.906834170854272e-05,
-      "loss": 5.808,
-      "step": 1428
-    },
-    {
-      "epoch": 0.7452411994784877,
-      "grad_norm": 1.8740195035934448,
-      "learning_rate": 9.906733668341709e-05,
-      "loss": 6.062,
-      "step": 1429
-    },
-    {
-      "epoch": 0.7457627118644068,
-      "grad_norm": 2.0066068172454834,
-      "learning_rate": 9.906633165829146e-05,
-      "loss": 6.2449,
-      "step": 1430
-    },
-    {
-      "epoch": 0.7462842242503259,
-      "grad_norm": 1.889897346496582,
-      "learning_rate": 9.906532663316583e-05,
-      "loss": 6.3622,
-      "step": 1431
-    },
-    {
-      "epoch": 0.7468057366362452,
-      "grad_norm": 1.6075999736785889,
-      "learning_rate": 9.90643216080402e-05,
-      "loss": 6.4702,
-      "step": 1432
-    },
-    {
-      "epoch": 0.7473272490221643,
-      "grad_norm": 2.4462218284606934,
-      "learning_rate": 9.906331658291458e-05,
-      "loss": 5.7908,
-      "step": 1433
-    },
-    {
-      "epoch": 0.7478487614080834,
-      "grad_norm": 1.9184459447860718,
-      "learning_rate": 9.906231155778895e-05,
-      "loss": 6.1585,
-      "step": 1434
-    },
-    {
-      "epoch": 0.7483702737940026,
-      "grad_norm": 2.1779956817626953,
-      "learning_rate": 9.906130653266333e-05,
-      "loss": 6.1793,
-      "step": 1435
-    },
-    {
-      "epoch": 0.7488917861799218,
-      "grad_norm": 1.7527437210083008,
-      "learning_rate": 9.906030150753769e-05,
-      "loss": 6.449,
-      "step": 1436
-    },
-    {
-      "epoch": 0.7494132985658409,
-      "grad_norm": 1.6552367210388184,
-      "learning_rate": 9.905929648241207e-05,
-      "loss": 6.5836,
-      "step": 1437
-    },
-    {
-      "epoch": 0.7499348109517601,
-      "grad_norm": 1.90924072265625,
-      "learning_rate": 9.905829145728643e-05,
-      "loss": 5.9668,
-      "step": 1438
-    },
-    {
-      "epoch": 0.7504563233376793,
-      "grad_norm": 2.0400619506835938,
-      "learning_rate": 9.905728643216081e-05,
-      "loss": 6.2105,
-      "step": 1439
-    },
-    {
-      "epoch": 0.7509778357235984,
-      "grad_norm": 1.8455311059951782,
-      "learning_rate": 9.905628140703517e-05,
-      "loss": 6.3227,
-      "step": 1440
-    },
-    {
-      "epoch": 0.7514993481095176,
-      "grad_norm": 2.2235541343688965,
-      "learning_rate": 9.905527638190955e-05,
-      "loss": 6.1011,
-      "step": 1441
-    },
-    {
-      "epoch": 0.7520208604954368,
-      "grad_norm": 1.8395979404449463,
-      "learning_rate": 9.905427135678392e-05,
-      "loss": 6.1377,
-      "step": 1442
-    },
-    {
-      "epoch": 0.752542372881356,
-      "grad_norm": 1.9957846403121948,
-      "learning_rate": 9.90532663316583e-05,
-      "loss": 6.0979,
-      "step": 1443
-    },
-    {
-      "epoch": 0.7530638852672751,
-      "grad_norm": 1.9281573295593262,
-      "learning_rate": 9.905226130653267e-05,
-      "loss": 6.4314,
-      "step": 1444
-    },
-    {
-      "epoch": 0.7535853976531942,
-      "grad_norm": 1.8461557626724243,
-      "learning_rate": 9.905125628140704e-05,
-      "loss": 6.603,
-      "step": 1445
-    },
-    {
-      "epoch": 0.7541069100391135,
-      "grad_norm": 1.6056270599365234,
-      "learning_rate": 9.905025125628141e-05,
-      "loss": 6.6123,
-      "step": 1446
-    },
-    {
-      "epoch": 0.7546284224250326,
-      "grad_norm": 1.977894902229309,
-      "learning_rate": 9.904924623115578e-05,
-      "loss": 6.2871,
-      "step": 1447
-    },
-    {
-      "epoch": 0.7551499348109517,
-      "grad_norm": 1.59738290309906,
-      "learning_rate": 9.904824120603016e-05,
-      "loss": 6.4567,
-      "step": 1448
-    },
-    {
-      "epoch": 0.755671447196871,
-      "grad_norm": 1.7254303693771362,
-      "learning_rate": 9.904723618090452e-05,
-      "loss": 6.3123,
-      "step": 1449
-    },
-    {
-      "epoch": 0.7561929595827901,
-      "grad_norm": 2.0217103958129883,
-      "learning_rate": 9.90462311557789e-05,
-      "loss": 6.0582,
-      "step": 1450
-    },
-    {
-      "epoch": 0.7567144719687092,
-      "grad_norm": 1.7805087566375732,
-      "learning_rate": 9.904522613065326e-05,
-      "loss": 6.3635,
-      "step": 1451
-    },
-    {
-      "epoch": 0.7572359843546285,
-      "grad_norm": 1.9256607294082642,
-      "learning_rate": 9.904422110552764e-05,
-      "loss": 5.7504,
-      "step": 1452
-    },
-    {
-      "epoch": 0.7577574967405476,
-      "grad_norm": 1.9249347448349,
-      "learning_rate": 9.904321608040202e-05,
-      "loss": 6.099,
-      "step": 1453
-    },
-    {
-      "epoch": 0.7582790091264667,
-      "grad_norm": 2.34395694732666,
-      "learning_rate": 9.90422110552764e-05,
-      "loss": 6.1244,
-      "step": 1454
-    },
-    {
-      "epoch": 0.758800521512386,
-      "grad_norm": 2.0598037242889404,
-      "learning_rate": 9.904120603015076e-05,
-      "loss": 6.4953,
-      "step": 1455
-    },
-    {
-      "epoch": 0.7593220338983051,
-      "grad_norm": 1.7266490459442139,
-      "learning_rate": 9.904020100502514e-05,
-      "loss": 6.4625,
-      "step": 1456
-    },
-    {
-      "epoch": 0.7598435462842242,
-      "grad_norm": 1.6677218675613403,
-      "learning_rate": 9.90391959798995e-05,
-      "loss": 6.4217,
-      "step": 1457
-    },
-    {
-      "epoch": 0.7603650586701434,
-      "grad_norm": 1.7407574653625488,
-      "learning_rate": 9.903819095477388e-05,
-      "loss": 6.395,
-      "step": 1458
-    },
-    {
-      "epoch": 0.7608865710560626,
-      "grad_norm": 1.788590431213379,
-      "learning_rate": 9.903718592964824e-05,
-      "loss": 6.2999,
-      "step": 1459
-    },
-    {
-      "epoch": 0.7614080834419817,
-      "grad_norm": 1.6275875568389893,
-      "learning_rate": 9.903618090452261e-05,
-      "loss": 6.4773,
-      "step": 1460
-    },
-    {
-      "epoch": 0.7619295958279009,
-      "grad_norm": 1.848962664604187,
-      "learning_rate": 9.903517587939699e-05,
-      "loss": 6.1702,
-      "step": 1461
-    },
-    {
-      "epoch": 0.7624511082138201,
-      "grad_norm": 1.6216939687728882,
-      "learning_rate": 9.903417085427135e-05,
-      "loss": 6.4403,
-      "step": 1462
-    },
-    {
-      "epoch": 0.7629726205997392,
-      "grad_norm": 1.8353079557418823,
-      "learning_rate": 9.903316582914573e-05,
-      "loss": 6.4074,
-      "step": 1463
-    },
-    {
-      "epoch": 0.7634941329856584,
-      "grad_norm": 1.9631184339523315,
-      "learning_rate": 9.903216080402011e-05,
-      "loss": 6.3446,
-      "step": 1464
-    },
-    {
-      "epoch": 0.7640156453715776,
-      "grad_norm": 1.9883472919464111,
-      "learning_rate": 9.903115577889448e-05,
-      "loss": 6.1008,
-      "step": 1465
-    },
-    {
-      "epoch": 0.7645371577574968,
-      "grad_norm": 1.910007119178772,
-      "learning_rate": 9.903015075376885e-05,
-      "loss": 6.2482,
-      "step": 1466
-    },
-    {
-      "epoch": 0.7650586701434159,
-      "grad_norm": 2.2316036224365234,
-      "learning_rate": 9.902914572864323e-05,
-      "loss": 6.1602,
-      "step": 1467
-    },
-    {
-      "epoch": 0.765580182529335,
-      "grad_norm": 1.9672383069992065,
-      "learning_rate": 9.902814070351759e-05,
-      "loss": 6.3586,
-      "step": 1468
-    },
-    {
-      "epoch": 0.7661016949152543,
-      "grad_norm": 2.0362443923950195,
-      "learning_rate": 9.902713567839197e-05,
-      "loss": 5.8586,
-      "step": 1469
-    },
-    {
-      "epoch": 0.7666232073011734,
-      "grad_norm": 2.102182388305664,
-      "learning_rate": 9.902613065326633e-05,
-      "loss": 5.5886,
-      "step": 1470
-    },
-    {
-      "epoch": 0.7671447196870925,
-      "grad_norm": 1.752789855003357,
-      "learning_rate": 9.902512562814071e-05,
-      "loss": 6.1534,
-      "step": 1471
-    },
-    {
-      "epoch": 0.7676662320730118,
-      "grad_norm": 1.9371451139450073,
-      "learning_rate": 9.902412060301508e-05,
-      "loss": 6.203,
-      "step": 1472
-    },
-    {
-      "epoch": 0.7681877444589309,
-      "grad_norm": 1.8908559083938599,
-      "learning_rate": 9.902311557788945e-05,
-      "loss": 6.0585,
-      "step": 1473
-    },
-    {
-      "epoch": 0.76870925684485,
-      "grad_norm": 1.9075958728790283,
-      "learning_rate": 9.902211055276383e-05,
-      "loss": 5.9371,
-      "step": 1474
-    },
-    {
-      "epoch": 0.7692307692307693,
-      "grad_norm": 1.8455158472061157,
-      "learning_rate": 9.90211055276382e-05,
-      "loss": 6.4795,
-      "step": 1475
-    },
-    {
-      "epoch": 0.7697522816166884,
-      "grad_norm": 1.840221881866455,
-      "learning_rate": 9.902010050251257e-05,
-      "loss": 6.209,
-      "step": 1476
-    },
-    {
-      "epoch": 0.7702737940026075,
-      "grad_norm": 1.9002190828323364,
-      "learning_rate": 9.901909547738694e-05,
-      "loss": 6.4243,
-      "step": 1477
-    },
-    {
-      "epoch": 0.7707953063885268,
-      "grad_norm": 1.634083867073059,
-      "learning_rate": 9.901809045226132e-05,
-      "loss": 6.5637,
-      "step": 1478
-    },
-    {
-      "epoch": 0.7713168187744459,
-      "grad_norm": 2.265779972076416,
-      "learning_rate": 9.901708542713568e-05,
-      "loss": 6.3149,
-      "step": 1479
-    },
-    {
-      "epoch": 0.771838331160365,
-      "grad_norm": 2.1377058029174805,
-      "learning_rate": 9.901608040201006e-05,
-      "loss": 6.0884,
-      "step": 1480
-    },
-    {
-      "epoch": 0.7723598435462842,
-      "grad_norm": 1.5956249237060547,
-      "learning_rate": 9.901507537688442e-05,
-      "loss": 6.5332,
-      "step": 1481
-    },
-    {
-      "epoch": 0.7728813559322034,
-      "grad_norm": 1.7575913667678833,
-      "learning_rate": 9.90140703517588e-05,
-      "loss": 6.3354,
-      "step": 1482
-    },
-    {
-      "epoch": 0.7734028683181225,
-      "grad_norm": 1.7829210758209229,
-      "learning_rate": 9.901306532663316e-05,
-      "loss": 6.4074,
-      "step": 1483
-    },
-    {
-      "epoch": 0.7739243807040417,
-      "grad_norm": 1.781745195388794,
-      "learning_rate": 9.901206030150754e-05,
-      "loss": 6.3292,
-      "step": 1484
-    },
-    {
-      "epoch": 0.7744458930899609,
-      "grad_norm": 1.8603249788284302,
-      "learning_rate": 9.901105527638192e-05,
-      "loss": 6.3827,
-      "step": 1485
-    },
-    {
-      "epoch": 0.7749674054758801,
-      "grad_norm": 1.7212809324264526,
-      "learning_rate": 9.901005025125628e-05,
-      "loss": 6.3597,
-      "step": 1486
-    },
-    {
-      "epoch": 0.7754889178617992,
-      "grad_norm": 2.077568531036377,
-      "learning_rate": 9.900904522613066e-05,
-      "loss": 6.5001,
-      "step": 1487
-    },
-    {
-      "epoch": 0.7760104302477184,
-      "grad_norm": 1.6344174146652222,
-      "learning_rate": 9.900804020100503e-05,
-      "loss": 6.5129,
-      "step": 1488
-    },
-    {
-      "epoch": 0.7765319426336376,
-      "grad_norm": 1.760164499282837,
-      "learning_rate": 9.90070351758794e-05,
-      "loss": 6.2177,
-      "step": 1489
-    },
-    {
-      "epoch": 0.7770534550195567,
-      "grad_norm": 1.7045994997024536,
-      "learning_rate": 9.900603015075377e-05,
-      "loss": 6.3082,
-      "step": 1490
-    },
-    {
-      "epoch": 0.7775749674054758,
-      "grad_norm": 1.8701021671295166,
-      "learning_rate": 9.900502512562815e-05,
-      "loss": 6.5448,
-      "step": 1491
-    },
-    {
-      "epoch": 0.7780964797913951,
-      "grad_norm": 1.9008667469024658,
-      "learning_rate": 9.900402010050251e-05,
-      "loss": 6.1047,
-      "step": 1492
-    },
-    {
-      "epoch": 0.7786179921773142,
-      "grad_norm": 2.518509864807129,
-      "learning_rate": 9.900301507537689e-05,
-      "loss": 5.6636,
-      "step": 1493
-    },
-    {
-      "epoch": 0.7791395045632333,
-      "grad_norm": 2.122746467590332,
-      "learning_rate": 9.900201005025127e-05,
-      "loss": 6.311,
-      "step": 1494
-    },
-    {
-      "epoch": 0.7796610169491526,
-      "grad_norm": 1.750341534614563,
-      "learning_rate": 9.900100502512564e-05,
-      "loss": 6.4305,
-      "step": 1495
-    },
-    {
-      "epoch": 0.7801825293350717,
-      "grad_norm": 2.263638496398926,
-      "learning_rate": 9.900000000000001e-05,
-      "loss": 6.0462,
-      "step": 1496
-    },
-    {
-      "epoch": 0.7807040417209908,
-      "grad_norm": 1.6788009405136108,
-      "learning_rate": 9.899899497487439e-05,
-      "loss": 6.3432,
-      "step": 1497
-    },
-    {
-      "epoch": 0.7812255541069101,
-      "grad_norm": 2.6085753440856934,
-      "learning_rate": 9.899798994974875e-05,
-      "loss": 5.5262,
-      "step": 1498
-    },
-    {
-      "epoch": 0.7817470664928292,
-      "grad_norm": 2.0676863193511963,
-      "learning_rate": 9.899698492462311e-05,
-      "loss": 5.8053,
-      "step": 1499
-    },
-    {
-      "epoch": 0.7822685788787483,
-      "grad_norm": 2.021592140197754,
-      "learning_rate": 9.899597989949749e-05,
-      "loss": 6.2348,
-      "step": 1500
-    },
-    {
-      "epoch": 0.7827900912646676,
-      "grad_norm": 2.2871415615081787,
-      "learning_rate": 9.899497487437186e-05,
-      "loss": 6.4151,
-      "step": 1501
-    },
-    {
-      "epoch": 0.7833116036505867,
-      "grad_norm": 2.0189568996429443,
-      "learning_rate": 9.899396984924623e-05,
-      "loss": 6.3597,
-      "step": 1502
-    },
-    {
-      "epoch": 0.7838331160365059,
-      "grad_norm": 1.8244130611419678,
-      "learning_rate": 9.89929648241206e-05,
-      "loss": 6.258,
-      "step": 1503
-    },
-    {
-      "epoch": 0.784354628422425,
-      "grad_norm": 1.9418656826019287,
-      "learning_rate": 9.899195979899498e-05,
-      "loss": 5.831,
-      "step": 1504
-    },
-    {
-      "epoch": 0.7848761408083442,
-      "grad_norm": 1.8899316787719727,
-      "learning_rate": 9.899095477386935e-05,
-      "loss": 6.2857,
-      "step": 1505
-    },
-    {
-      "epoch": 0.7853976531942634,
-      "grad_norm": 2.0739128589630127,
-      "learning_rate": 9.898994974874373e-05,
-      "loss": 5.6871,
-      "step": 1506
-    },
-    {
-      "epoch": 0.7859191655801825,
-      "grad_norm": 1.8384994268417358,
-      "learning_rate": 9.89889447236181e-05,
-      "loss": 6.3722,
-      "step": 1507
-    },
-    {
-      "epoch": 0.7864406779661017,
-      "grad_norm": 2.0411200523376465,
-      "learning_rate": 9.898793969849247e-05,
-      "loss": 5.9428,
-      "step": 1508
-    },
-    {
-      "epoch": 0.7869621903520209,
-      "grad_norm": 2.0504939556121826,
-      "learning_rate": 9.898693467336684e-05,
-      "loss": 6.3496,
-      "step": 1509
-    },
-    {
-      "epoch": 0.78748370273794,
-      "grad_norm": 1.8186893463134766,
-      "learning_rate": 9.898592964824122e-05,
-      "loss": 6.5377,
-      "step": 1510
-    },
-    {
-      "epoch": 0.7880052151238592,
-      "grad_norm": 1.8714625835418701,
-      "learning_rate": 9.898492462311558e-05,
-      "loss": 6.3488,
-      "step": 1511
-    },
-    {
-      "epoch": 0.7885267275097784,
-      "grad_norm": 2.114046096801758,
-      "learning_rate": 9.898391959798994e-05,
-      "loss": 6.2311,
-      "step": 1512
-    },
-    {
-      "epoch": 0.7890482398956975,
-      "grad_norm": 1.7104909420013428,
-      "learning_rate": 9.898291457286432e-05,
-      "loss": 6.2372,
-      "step": 1513
-    },
-    {
-      "epoch": 0.7895697522816167,
-      "grad_norm": 2.1042661666870117,
-      "learning_rate": 9.89819095477387e-05,
-      "loss": 6.2205,
-      "step": 1514
-    },
-    {
-      "epoch": 0.7900912646675359,
-      "grad_norm": 1.9569180011749268,
-      "learning_rate": 9.898090452261308e-05,
-      "loss": 5.8999,
-      "step": 1515
-    },
-    {
-      "epoch": 0.790612777053455,
-      "grad_norm": 1.9890902042388916,
-      "learning_rate": 9.897989949748744e-05,
-      "loss": 6.4819,
-      "step": 1516
-    },
-    {
-      "epoch": 0.7911342894393741,
-      "grad_norm": 1.888201117515564,
-      "learning_rate": 9.897889447236182e-05,
-      "loss": 6.1938,
-      "step": 1517
-    },
-    {
-      "epoch": 0.7916558018252934,
-      "grad_norm": 1.9150131940841675,
-      "learning_rate": 9.897788944723618e-05,
-      "loss": 6.1976,
-      "step": 1518
-    },
-    {
-      "epoch": 0.7921773142112125,
-      "grad_norm": 1.835471272468567,
-      "learning_rate": 9.897688442211056e-05,
-      "loss": 6.3683,
-      "step": 1519
-    },
-    {
-      "epoch": 0.7926988265971316,
-      "grad_norm": 1.6649370193481445,
-      "learning_rate": 9.897587939698493e-05,
-      "loss": 6.2805,
-      "step": 1520
-    },
-    {
-      "epoch": 0.7932203389830509,
-      "grad_norm": 2.5710160732269287,
-      "learning_rate": 9.89748743718593e-05,
-      "loss": 5.6337,
-      "step": 1521
-    },
-    {
-      "epoch": 0.79374185136897,
-      "grad_norm": 1.8277206420898438,
-      "learning_rate": 9.897386934673367e-05,
-      "loss": 6.1163,
-      "step": 1522
-    },
-    {
-      "epoch": 0.7942633637548892,
-      "grad_norm": 1.8245021104812622,
-      "learning_rate": 9.897286432160805e-05,
-      "loss": 6.3545,
-      "step": 1523
-    },
-    {
-      "epoch": 0.7947848761408084,
-      "grad_norm": 1.7888435125350952,
-      "learning_rate": 9.897185929648241e-05,
-      "loss": 6.1207,
-      "step": 1524
-    },
-    {
-      "epoch": 0.7953063885267275,
-      "grad_norm": 1.6552172899246216,
-      "learning_rate": 9.897085427135679e-05,
-      "loss": 6.1833,
-      "step": 1525
-    },
-    {
-      "epoch": 0.7958279009126467,
-      "grad_norm": 2.000274658203125,
-      "learning_rate": 9.896984924623117e-05,
-      "loss": 6.3528,
-      "step": 1526
-    },
-    {
-      "epoch": 0.7963494132985658,
-      "grad_norm": 2.166152238845825,
-      "learning_rate": 9.896884422110553e-05,
-      "loss": 6.5508,
-      "step": 1527
-    },
-    {
-      "epoch": 0.796870925684485,
-      "grad_norm": 2.298640012741089,
-      "learning_rate": 9.896783919597991e-05,
-      "loss": 6.0207,
-      "step": 1528
-    },
-    {
-      "epoch": 0.7973924380704042,
-      "grad_norm": 2.129709005355835,
-      "learning_rate": 9.896683417085427e-05,
-      "loss": 6.1335,
-      "step": 1529
-    },
-    {
-      "epoch": 0.7979139504563233,
-      "grad_norm": 2.158891201019287,
-      "learning_rate": 9.896582914572865e-05,
-      "loss": 5.933,
-      "step": 1530
-    },
-    {
-      "epoch": 0.7984354628422425,
-      "grad_norm": 2.039499044418335,
-      "learning_rate": 9.896482412060301e-05,
-      "loss": 6.2648,
-      "step": 1531
-    },
-    {
-      "epoch": 0.7989569752281617,
-      "grad_norm": 1.9083620309829712,
-      "learning_rate": 9.896381909547739e-05,
-      "loss": 6.5334,
-      "step": 1532
-    },
-    {
-      "epoch": 0.7994784876140808,
-      "grad_norm": 1.8761316537857056,
-      "learning_rate": 9.896281407035176e-05,
-      "loss": 6.2449,
-      "step": 1533
-    },
-    {
-      "epoch": 0.8,
-      "grad_norm": 1.6660499572753906,
-      "learning_rate": 9.896180904522613e-05,
-      "loss": 6.3593,
-      "step": 1534
-    },
-    {
-      "epoch": 0.8005215123859192,
-      "grad_norm": 1.6793335676193237,
-      "learning_rate": 9.896080402010051e-05,
-      "loss": 6.3603,
-      "step": 1535
-    },
-    {
-      "epoch": 0.8010430247718383,
-      "grad_norm": 1.666662573814392,
-      "learning_rate": 9.895979899497489e-05,
-      "loss": 5.7815,
-      "step": 1536
-    },
-    {
-      "epoch": 0.8015645371577576,
-      "grad_norm": 1.8911926746368408,
-      "learning_rate": 9.895879396984925e-05,
-      "loss": 6.3802,
-      "step": 1537
-    },
-    {
-      "epoch": 0.8020860495436767,
-      "grad_norm": 1.7046828269958496,
-      "learning_rate": 9.895778894472363e-05,
-      "loss": 6.2818,
-      "step": 1538
-    },
-    {
-      "epoch": 0.8026075619295958,
-      "grad_norm": 1.654525637626648,
-      "learning_rate": 9.8956783919598e-05,
-      "loss": 6.3808,
-      "step": 1539
-    },
-    {
-      "epoch": 0.803129074315515,
-      "grad_norm": 1.7619636058807373,
-      "learning_rate": 9.895577889447236e-05,
-      "loss": 6.4523,
-      "step": 1540
-    },
-    {
-      "epoch": 0.8036505867014342,
-      "grad_norm": 1.7219228744506836,
-      "learning_rate": 9.895477386934674e-05,
-      "loss": 6.2557,
-      "step": 1541
-    },
-    {
-      "epoch": 0.8041720990873533,
-      "grad_norm": 1.713586449623108,
-      "learning_rate": 9.89537688442211e-05,
-      "loss": 6.2463,
-      "step": 1542
-    },
-    {
-      "epoch": 0.8046936114732725,
-      "grad_norm": 1.813874363899231,
-      "learning_rate": 9.895276381909548e-05,
-      "loss": 6.1161,
-      "step": 1543
-    },
-    {
-      "epoch": 0.8052151238591917,
-      "grad_norm": 1.9660282135009766,
-      "learning_rate": 9.895175879396985e-05,
-      "loss": 6.1216,
-      "step": 1544
-    },
-    {
-      "epoch": 0.8057366362451108,
-      "grad_norm": 1.6842600107192993,
-      "learning_rate": 9.895075376884422e-05,
-      "loss": 6.4891,
-      "step": 1545
-    },
-    {
-      "epoch": 0.80625814863103,
-      "grad_norm": 2.0831360816955566,
-      "learning_rate": 9.89497487437186e-05,
-      "loss": 6.5017,
-      "step": 1546
-    },
-    {
-      "epoch": 0.8067796610169492,
-      "grad_norm": 1.654707908630371,
-      "learning_rate": 9.894874371859298e-05,
-      "loss": 6.5324,
-      "step": 1547
-    },
-    {
-      "epoch": 0.8073011734028683,
-      "grad_norm": 1.885933756828308,
-      "learning_rate": 9.894773869346734e-05,
-      "loss": 5.9692,
-      "step": 1548
-    },
-    {
-      "epoch": 0.8078226857887875,
-      "grad_norm": 1.806748390197754,
-      "learning_rate": 9.894673366834172e-05,
-      "loss": 6.3875,
-      "step": 1549
-    },
-    {
-      "epoch": 0.8083441981747066,
-      "grad_norm": 1.5926307439804077,
-      "learning_rate": 9.894572864321609e-05,
-      "loss": 6.3195,
-      "step": 1550
-    },
-    {
-      "epoch": 0.8088657105606258,
-      "grad_norm": 1.7504496574401855,
-      "learning_rate": 9.894472361809046e-05,
-      "loss": 6.2834,
-      "step": 1551
-    },
-    {
-      "epoch": 0.809387222946545,
-      "grad_norm": 1.9168084859848022,
-      "learning_rate": 9.894371859296483e-05,
-      "loss": 6.1883,
-      "step": 1552
-    },
-    {
-      "epoch": 0.8099087353324641,
-      "grad_norm": 1.6733273267745972,
-      "learning_rate": 9.894271356783919e-05,
-      "loss": 6.5441,
-      "step": 1553
-    },
-    {
-      "epoch": 0.8104302477183833,
-      "grad_norm": 2.0163559913635254,
-      "learning_rate": 9.894170854271357e-05,
-      "loss": 6.3247,
-      "step": 1554
-    },
-    {
-      "epoch": 0.8109517601043025,
-      "grad_norm": 3.170121431350708,
-      "learning_rate": 9.894070351758795e-05,
-      "loss": 6.4189,
-      "step": 1555
-    },
-    {
-      "epoch": 0.8114732724902216,
-      "grad_norm": 1.7728058099746704,
-      "learning_rate": 9.893969849246232e-05,
-      "loss": 6.1033,
-      "step": 1556
-    },
-    {
-      "epoch": 0.8119947848761409,
-      "grad_norm": 1.6014310121536255,
-      "learning_rate": 9.893869346733669e-05,
-      "loss": 6.4606,
-      "step": 1557
-    },
-    {
-      "epoch": 0.81251629726206,
-      "grad_norm": 1.60207998752594,
-      "learning_rate": 9.893768844221107e-05,
-      "loss": 6.6052,
-      "step": 1558
-    },
-    {
-      "epoch": 0.8130378096479791,
-      "grad_norm": 1.8946715593338013,
-      "learning_rate": 9.893668341708543e-05,
-      "loss": 5.8825,
-      "step": 1559
-    },
-    {
-      "epoch": 0.8135593220338984,
-      "grad_norm": 1.6949716806411743,
-      "learning_rate": 9.893567839195981e-05,
-      "loss": 6.4048,
-      "step": 1560
-    },
-    {
-      "epoch": 0.8140808344198175,
-      "grad_norm": 1.7373712062835693,
-      "learning_rate": 9.893467336683417e-05,
-      "loss": 6.2614,
-      "step": 1561
-    },
-    {
-      "epoch": 0.8146023468057366,
-      "grad_norm": 2.1420915126800537,
-      "learning_rate": 9.893366834170855e-05,
-      "loss": 5.641,
-      "step": 1562
-    },
-    {
-      "epoch": 0.8151238591916558,
-      "grad_norm": 1.9140840768814087,
-      "learning_rate": 9.893266331658292e-05,
-      "loss": 6.2816,
-      "step": 1563
-    },
-    {
-      "epoch": 0.815645371577575,
-      "grad_norm": 1.9483697414398193,
-      "learning_rate": 9.89316582914573e-05,
-      "loss": 6.436,
-      "step": 1564
-    },
-    {
-      "epoch": 0.8161668839634941,
-      "grad_norm": 2.1768035888671875,
-      "learning_rate": 9.893065326633166e-05,
-      "loss": 5.3937,
-      "step": 1565
-    },
-    {
-      "epoch": 0.8166883963494133,
-      "grad_norm": 1.8576078414916992,
-      "learning_rate": 9.892964824120604e-05,
-      "loss": 6.343,
-      "step": 1566
-    },
-    {
-      "epoch": 0.8172099087353325,
-      "grad_norm": 1.9725664854049683,
-      "learning_rate": 9.892864321608041e-05,
-      "loss": 6.306,
-      "step": 1567
-    },
-    {
-      "epoch": 0.8177314211212516,
-      "grad_norm": 1.7392650842666626,
-      "learning_rate": 9.892763819095478e-05,
-      "loss": 6.2763,
-      "step": 1568
-    },
-    {
-      "epoch": 0.8182529335071708,
-      "grad_norm": 1.9523593187332153,
-      "learning_rate": 9.892663316582916e-05,
-      "loss": 5.9679,
-      "step": 1569
-    },
-    {
-      "epoch": 0.81877444589309,
-      "grad_norm": 2.391331195831299,
-      "learning_rate": 9.892562814070352e-05,
-      "loss": 6.0673,
-      "step": 1570
-    },
-    {
-      "epoch": 0.8192959582790091,
-      "grad_norm": 1.7008408308029175,
-      "learning_rate": 9.89246231155779e-05,
-      "loss": 6.2428,
-      "step": 1571
-    },
-    {
-      "epoch": 0.8198174706649283,
-      "grad_norm": 1.6180990934371948,
-      "learning_rate": 9.892361809045226e-05,
-      "loss": 6.3696,
-      "step": 1572
-    },
-    {
-      "epoch": 0.8203389830508474,
-      "grad_norm": 1.9037576913833618,
-      "learning_rate": 9.892261306532664e-05,
-      "loss": 6.2464,
-      "step": 1573
-    },
-    {
-      "epoch": 0.8208604954367666,
-      "grad_norm": 1.7359576225280762,
-      "learning_rate": 9.8921608040201e-05,
-      "loss": 6.2016,
-      "step": 1574
-    },
-    {
-      "epoch": 0.8213820078226858,
-      "grad_norm": 2.0656278133392334,
-      "learning_rate": 9.892060301507538e-05,
-      "loss": 6.0986,
-      "step": 1575
-    },
-    {
-      "epoch": 0.8219035202086049,
-      "grad_norm": 1.735777497291565,
-      "learning_rate": 9.891959798994975e-05,
-      "loss": 6.4396,
-      "step": 1576
-    },
-    {
-      "epoch": 0.8224250325945242,
-      "grad_norm": 1.8029255867004395,
-      "learning_rate": 9.891859296482412e-05,
-      "loss": 6.2857,
-      "step": 1577
-    },
-    {
-      "epoch": 0.8229465449804433,
-      "grad_norm": 1.8384455442428589,
-      "learning_rate": 9.89175879396985e-05,
-      "loss": 6.0151,
-      "step": 1578
-    },
-    {
-      "epoch": 0.8234680573663624,
-      "grad_norm": 1.951324701309204,
-      "learning_rate": 9.891658291457287e-05,
-      "loss": 5.8272,
-      "step": 1579
-    },
-    {
-      "epoch": 0.8239895697522817,
-      "grad_norm": 1.8120818138122559,
-      "learning_rate": 9.891557788944724e-05,
-      "loss": 6.0491,
-      "step": 1580
-    },
-    {
-      "epoch": 0.8245110821382008,
-      "grad_norm": 1.7207704782485962,
-      "learning_rate": 9.891457286432161e-05,
-      "loss": 6.3431,
-      "step": 1581
-    },
-    {
-      "epoch": 0.8250325945241199,
-      "grad_norm": 1.9370810985565186,
-      "learning_rate": 9.891356783919599e-05,
-      "loss": 5.8802,
-      "step": 1582
-    },
-    {
-      "epoch": 0.8255541069100392,
-      "grad_norm": 1.954039454460144,
-      "learning_rate": 9.891256281407035e-05,
-      "loss": 6.5588,
-      "step": 1583
-    },
-    {
-      "epoch": 0.8260756192959583,
-      "grad_norm": 1.8196903467178345,
-      "learning_rate": 9.891155778894473e-05,
-      "loss": 5.7277,
-      "step": 1584
-    },
-    {
-      "epoch": 0.8265971316818774,
-      "grad_norm": 1.8717683553695679,
-      "learning_rate": 9.891055276381909e-05,
-      "loss": 6.379,
-      "step": 1585
-    },
-    {
-      "epoch": 0.8271186440677966,
-      "grad_norm": 1.6302410364151,
-      "learning_rate": 9.890954773869347e-05,
-      "loss": 6.3934,
-      "step": 1586
-    },
-    {
-      "epoch": 0.8276401564537158,
-      "grad_norm": 2.919593572616577,
-      "learning_rate": 9.890854271356785e-05,
-      "loss": 5.6854,
-      "step": 1587
-    },
-    {
-      "epoch": 0.8281616688396349,
-      "grad_norm": 1.7433992624282837,
-      "learning_rate": 9.890753768844223e-05,
-      "loss": 6.1599,
-      "step": 1588
-    },
-    {
-      "epoch": 0.8286831812255541,
-      "grad_norm": 1.7605253458023071,
-      "learning_rate": 9.890653266331659e-05,
-      "loss": 6.1212,
-      "step": 1589
-    },
-    {
-      "epoch": 0.8292046936114733,
-      "grad_norm": 1.6788815259933472,
-      "learning_rate": 9.890552763819097e-05,
-      "loss": 6.406,
-      "step": 1590
-    },
-    {
-      "epoch": 0.8297262059973924,
-      "grad_norm": 1.7766562700271606,
-      "learning_rate": 9.890452261306533e-05,
-      "loss": 6.4508,
-      "step": 1591
-    },
-    {
-      "epoch": 0.8302477183833116,
-      "grad_norm": 1.7444632053375244,
-      "learning_rate": 9.89035175879397e-05,
-      "loss": 6.055,
-      "step": 1592
-    },
-    {
-      "epoch": 0.8307692307692308,
-      "grad_norm": 2.1119027137756348,
-      "learning_rate": 9.890251256281407e-05,
-      "loss": 6.2266,
-      "step": 1593
-    },
-    {
-      "epoch": 0.83129074315515,
-      "grad_norm": 1.8859652280807495,
-      "learning_rate": 9.890150753768844e-05,
-      "loss": 6.3608,
-      "step": 1594
-    },
-    {
-      "epoch": 0.8318122555410691,
-      "grad_norm": 1.8703123331069946,
-      "learning_rate": 9.890050251256282e-05,
-      "loss": 5.9862,
-      "step": 1595
-    },
-    {
-      "epoch": 0.8323337679269882,
-      "grad_norm": 1.7631269693374634,
-      "learning_rate": 9.889949748743718e-05,
-      "loss": 6.1231,
-      "step": 1596
-    },
-    {
-      "epoch": 0.8328552803129075,
-      "grad_norm": 1.702431321144104,
-      "learning_rate": 9.889849246231156e-05,
-      "loss": 6.218,
-      "step": 1597
-    },
-    {
-      "epoch": 0.8333767926988266,
-      "grad_norm": 1.5598454475402832,
-      "learning_rate": 9.889748743718594e-05,
-      "loss": 6.4459,
-      "step": 1598
-    },
-    {
-      "epoch": 0.8338983050847457,
-      "grad_norm": 3.122119903564453,
-      "learning_rate": 9.889648241206031e-05,
-      "loss": 5.9532,
-      "step": 1599
-    },
-    {
-      "epoch": 0.834419817470665,
-      "grad_norm": 1.9566999673843384,
-      "learning_rate": 9.889547738693468e-05,
-      "loss": 6.1945,
-      "step": 1600
-    },
-    {
-      "epoch": 0.8349413298565841,
-      "grad_norm": 2.015418291091919,
-      "learning_rate": 9.889447236180906e-05,
-      "loss": 6.0165,
-      "step": 1601
-    },
-    {
-      "epoch": 0.8354628422425032,
-      "grad_norm": 1.7766563892364502,
-      "learning_rate": 9.889346733668342e-05,
-      "loss": 6.2291,
-      "step": 1602
-    },
-    {
-      "epoch": 0.8359843546284225,
-      "grad_norm": 2.0525100231170654,
-      "learning_rate": 9.88924623115578e-05,
-      "loss": 6.2915,
-      "step": 1603
-    },
-    {
-      "epoch": 0.8365058670143416,
-      "grad_norm": 1.9141136407852173,
-      "learning_rate": 9.889145728643216e-05,
-      "loss": 6.4239,
-      "step": 1604
-    },
-    {
-      "epoch": 0.8370273794002607,
-      "grad_norm": 2.252941370010376,
-      "learning_rate": 9.889045226130653e-05,
-      "loss": 5.5331,
-      "step": 1605
-    },
-    {
-      "epoch": 0.83754889178618,
-      "grad_norm": 2.044296979904175,
-      "learning_rate": 9.88894472361809e-05,
-      "loss": 5.9919,
-      "step": 1606
-    },
-    {
-      "epoch": 0.8380704041720991,
-      "grad_norm": 1.8609423637390137,
-      "learning_rate": 9.888844221105528e-05,
-      "loss": 6.5144,
-      "step": 1607
-    },
-    {
-      "epoch": 0.8385919165580182,
-      "grad_norm": 1.6780518293380737,
-      "learning_rate": 9.888743718592966e-05,
-      "loss": 6.1254,
-      "step": 1608
-    },
-    {
-      "epoch": 0.8391134289439374,
-      "grad_norm": 1.7610026597976685,
-      "learning_rate": 9.888643216080402e-05,
-      "loss": 6.1459,
-      "step": 1609
-    },
-    {
-      "epoch": 0.8396349413298566,
-      "grad_norm": 2.063675880432129,
-      "learning_rate": 9.88854271356784e-05,
-      "loss": 5.8401,
-      "step": 1610
-    },
-    {
-      "epoch": 0.8401564537157757,
-      "grad_norm": 1.9090136289596558,
-      "learning_rate": 9.888442211055277e-05,
-      "loss": 6.1468,
-      "step": 1611
-    },
-    {
-      "epoch": 0.8406779661016949,
-      "grad_norm": 1.9683161973953247,
-      "learning_rate": 9.888341708542714e-05,
-      "loss": 6.0548,
-      "step": 1612
-    },
-    {
-      "epoch": 0.8411994784876141,
-      "grad_norm": 1.8758752346038818,
-      "learning_rate": 9.888241206030151e-05,
-      "loss": 6.0022,
-      "step": 1613
-    },
-    {
-      "epoch": 0.8417209908735332,
-      "grad_norm": 1.9879589080810547,
-      "learning_rate": 9.888140703517589e-05,
-      "loss": 6.4184,
-      "step": 1614
-    },
-    {
-      "epoch": 0.8422425032594524,
-      "grad_norm": 1.6938636302947998,
-      "learning_rate": 9.888040201005025e-05,
-      "loss": 6.3096,
-      "step": 1615
-    },
-    {
-      "epoch": 0.8427640156453716,
-      "grad_norm": 1.819214940071106,
-      "learning_rate": 9.887939698492463e-05,
-      "loss": 6.1078,
-      "step": 1616
-    },
-    {
-      "epoch": 0.8432855280312908,
-      "grad_norm": 2.409513235092163,
-      "learning_rate": 9.887839195979899e-05,
-      "loss": 6.4294,
-      "step": 1617
-    },
-    {
-      "epoch": 0.8438070404172099,
-      "grad_norm": 2.883535623550415,
-      "learning_rate": 9.887738693467337e-05,
-      "loss": 6.4456,
-      "step": 1618
-    },
-    {
-      "epoch": 0.8443285528031291,
-      "grad_norm": 2.0457053184509277,
-      "learning_rate": 9.887638190954775e-05,
-      "loss": 6.2745,
-      "step": 1619
-    },
-    {
-      "epoch": 0.8448500651890483,
-      "grad_norm": 1.9166165590286255,
-      "learning_rate": 9.887537688442211e-05,
-      "loss": 6.5612,
-      "step": 1620
-    },
-    {
-      "epoch": 0.8453715775749674,
-      "grad_norm": 2.0632543563842773,
-      "learning_rate": 9.887437185929649e-05,
-      "loss": 5.371,
-      "step": 1621
-    },
-    {
-      "epoch": 0.8458930899608865,
-      "grad_norm": 1.76840078830719,
-      "learning_rate": 9.887336683417085e-05,
-      "loss": 6.5503,
-      "step": 1622
-    },
-    {
-      "epoch": 0.8464146023468058,
-      "grad_norm": 1.7861356735229492,
-      "learning_rate": 9.887236180904523e-05,
-      "loss": 5.9609,
-      "step": 1623
-    },
-    {
-      "epoch": 0.8469361147327249,
-      "grad_norm": 1.7596338987350464,
-      "learning_rate": 9.88713567839196e-05,
-      "loss": 5.9131,
-      "step": 1624
-    },
-    {
-      "epoch": 0.847457627118644,
-      "grad_norm": 1.798568606376648,
-      "learning_rate": 9.887035175879397e-05,
-      "loss": 6.1214,
-      "step": 1625
-    },
-    {
-      "epoch": 0.8479791395045633,
-      "grad_norm": 1.8737748861312866,
-      "learning_rate": 9.886934673366834e-05,
-      "loss": 6.2013,
-      "step": 1626
-    },
-    {
-      "epoch": 0.8485006518904824,
-      "grad_norm": 1.8132468461990356,
-      "learning_rate": 9.886834170854272e-05,
-      "loss": 5.842,
-      "step": 1627
-    },
-    {
-      "epoch": 0.8490221642764015,
-      "grad_norm": 1.8197094202041626,
-      "learning_rate": 9.88673366834171e-05,
-      "loss": 6.5088,
-      "step": 1628
-    },
-    {
-      "epoch": 0.8495436766623208,
-      "grad_norm": 1.6862690448760986,
-      "learning_rate": 9.886633165829147e-05,
-      "loss": 6.409,
-      "step": 1629
-    },
-    {
-      "epoch": 0.8500651890482399,
-      "grad_norm": 1.6131439208984375,
-      "learning_rate": 9.886532663316584e-05,
-      "loss": 6.3657,
-      "step": 1630
-    },
-    {
-      "epoch": 0.850586701434159,
-      "grad_norm": 1.8240110874176025,
-      "learning_rate": 9.886432160804021e-05,
-      "loss": 6.0665,
-      "step": 1631
-    },
-    {
-      "epoch": 0.8511082138200782,
-      "grad_norm": 1.997011423110962,
-      "learning_rate": 9.886331658291458e-05,
-      "loss": 6.0481,
-      "step": 1632
-    },
-    {
-      "epoch": 0.8516297262059974,
-      "grad_norm": 1.7994962930679321,
-      "learning_rate": 9.886231155778894e-05,
-      "loss": 5.8835,
-      "step": 1633
-    },
-    {
-      "epoch": 0.8521512385919165,
-      "grad_norm": 1.7089227437973022,
-      "learning_rate": 9.886130653266332e-05,
-      "loss": 6.0958,
-      "step": 1634
-    },
-    {
-      "epoch": 0.8526727509778357,
-      "grad_norm": 1.5072299242019653,
-      "learning_rate": 9.886030150753769e-05,
-      "loss": 6.4401,
-      "step": 1635
-    },
-    {
-      "epoch": 0.8531942633637549,
-      "grad_norm": 1.7127587795257568,
-      "learning_rate": 9.885929648241206e-05,
-      "loss": 6.4359,
-      "step": 1636
-    },
-    {
-      "epoch": 0.853715775749674,
-      "grad_norm": 1.9178413152694702,
-      "learning_rate": 9.885829145728643e-05,
-      "loss": 5.9645,
-      "step": 1637
-    },
-    {
-      "epoch": 0.8542372881355932,
-      "grad_norm": 1.8929202556610107,
-      "learning_rate": 9.88572864321608e-05,
-      "loss": 6.2902,
-      "step": 1638
-    },
-    {
-      "epoch": 0.8547588005215124,
-      "grad_norm": 2.042419910430908,
-      "learning_rate": 9.885628140703518e-05,
-      "loss": 6.3096,
-      "step": 1639
-    },
-    {
-      "epoch": 0.8552803129074316,
-      "grad_norm": 1.7103806734085083,
-      "learning_rate": 9.885527638190956e-05,
-      "loss": 6.3807,
-      "step": 1640
-    },
-    {
-      "epoch": 0.8558018252933507,
-      "grad_norm": 1.9331402778625488,
-      "learning_rate": 9.885427135678393e-05,
-      "loss": 6.4526,
-      "step": 1641
-    },
-    {
-      "epoch": 0.8563233376792699,
-      "grad_norm": 1.9394752979278564,
-      "learning_rate": 9.88532663316583e-05,
-      "loss": 6.305,
-      "step": 1642
-    },
-    {
-      "epoch": 0.8568448500651891,
-      "grad_norm": 2.5800442695617676,
-      "learning_rate": 9.885226130653267e-05,
-      "loss": 5.2474,
-      "step": 1643
-    },
-    {
-      "epoch": 0.8573663624511082,
-      "grad_norm": 1.7617117166519165,
-      "learning_rate": 9.885125628140705e-05,
-      "loss": 6.1225,
-      "step": 1644
-    },
-    {
-      "epoch": 0.8578878748370273,
-      "grad_norm": 2.0699551105499268,
-      "learning_rate": 9.885025125628141e-05,
-      "loss": 6.1364,
-      "step": 1645
-    },
-    {
-      "epoch": 0.8584093872229466,
-      "grad_norm": 2.011145830154419,
-      "learning_rate": 9.884924623115577e-05,
-      "loss": 6.5612,
-      "step": 1646
-    },
-    {
-      "epoch": 0.8589308996088657,
-      "grad_norm": 2.022282600402832,
-      "learning_rate": 9.884824120603015e-05,
-      "loss": 6.0344,
-      "step": 1647
-    },
-    {
-      "epoch": 0.8594524119947848,
-      "grad_norm": 2.03702449798584,
-      "learning_rate": 9.884723618090453e-05,
-      "loss": 6.2099,
-      "step": 1648
-    },
-    {
-      "epoch": 0.8599739243807041,
-      "grad_norm": 1.7508972883224487,
-      "learning_rate": 9.884623115577891e-05,
-      "loss": 6.2943,
-      "step": 1649
-    },
-    {
-      "epoch": 0.8604954367666232,
-      "grad_norm": 1.8152028322219849,
-      "learning_rate": 9.884522613065327e-05,
-      "loss": 6.1205,
-      "step": 1650
-    },
-    {
-      "epoch": 0.8610169491525423,
-      "grad_norm": 1.9179080724716187,
-      "learning_rate": 9.884422110552765e-05,
-      "loss": 6.2267,
-      "step": 1651
-    },
-    {
-      "epoch": 0.8615384615384616,
-      "grad_norm": 1.5328307151794434,
-      "learning_rate": 9.884321608040201e-05,
-      "loss": 6.3627,
-      "step": 1652
-    },
-    {
-      "epoch": 0.8620599739243807,
-      "grad_norm": 1.8169989585876465,
-      "learning_rate": 9.884221105527639e-05,
-      "loss": 6.1228,
-      "step": 1653
-    },
-    {
-      "epoch": 0.8625814863102998,
-      "grad_norm": 2.0403926372528076,
-      "learning_rate": 9.884120603015076e-05,
-      "loss": 6.3445,
-      "step": 1654
-    },
-    {
-      "epoch": 0.863102998696219,
-      "grad_norm": 1.8902539014816284,
-      "learning_rate": 9.884020100502513e-05,
-      "loss": 6.2774,
-      "step": 1655
-    },
-    {
-      "epoch": 0.8636245110821382,
-      "grad_norm": 2.031467914581299,
-      "learning_rate": 9.88391959798995e-05,
-      "loss": 5.5188,
-      "step": 1656
-    },
-    {
-      "epoch": 0.8641460234680574,
-      "grad_norm": 1.8601702451705933,
-      "learning_rate": 9.883819095477388e-05,
-      "loss": 6.4218,
-      "step": 1657
-    },
-    {
-      "epoch": 0.8646675358539765,
-      "grad_norm": 1.8415976762771606,
-      "learning_rate": 9.883718592964824e-05,
-      "loss": 5.8598,
-      "step": 1658
-    },
-    {
-      "epoch": 0.8651890482398957,
-      "grad_norm": 1.6482434272766113,
-      "learning_rate": 9.883618090452262e-05,
-      "loss": 6.5083,
-      "step": 1659
-    },
-    {
-      "epoch": 0.8657105606258149,
-      "grad_norm": 1.7652984857559204,
-      "learning_rate": 9.8835175879397e-05,
-      "loss": 6.1357,
-      "step": 1660
-    },
-    {
-      "epoch": 0.866232073011734,
-      "grad_norm": 2.12864089012146,
-      "learning_rate": 9.883417085427136e-05,
-      "loss": 5.5445,
-      "step": 1661
-    },
-    {
-      "epoch": 0.8667535853976532,
-      "grad_norm": 2.5631587505340576,
-      "learning_rate": 9.883316582914574e-05,
-      "loss": 6.0426,
-      "step": 1662
-    },
-    {
-      "epoch": 0.8672750977835724,
-      "grad_norm": 1.9603437185287476,
-      "learning_rate": 9.88321608040201e-05,
-      "loss": 6.3445,
-      "step": 1663
-    },
-    {
-      "epoch": 0.8677966101694915,
-      "grad_norm": 1.6258248090744019,
-      "learning_rate": 9.883115577889448e-05,
-      "loss": 6.1017,
-      "step": 1664
-    },
-    {
-      "epoch": 0.8683181225554107,
-      "grad_norm": 1.8574259281158447,
-      "learning_rate": 9.883015075376884e-05,
-      "loss": 6.4319,
-      "step": 1665
-    },
-    {
-      "epoch": 0.8688396349413299,
-      "grad_norm": 2.0191900730133057,
-      "learning_rate": 9.882914572864322e-05,
-      "loss": 4.9855,
-      "step": 1666
-    },
-    {
-      "epoch": 0.869361147327249,
-      "grad_norm": 1.8582489490509033,
-      "learning_rate": 9.882814070351759e-05,
-      "loss": 6.4653,
-      "step": 1667
-    },
-    {
-      "epoch": 0.8698826597131681,
-      "grad_norm": 1.9659878015518188,
-      "learning_rate": 9.882713567839196e-05,
-      "loss": 6.531,
-      "step": 1668
-    },
-    {
-      "epoch": 0.8704041720990874,
-      "grad_norm": 1.6383056640625,
-      "learning_rate": 9.882613065326634e-05,
-      "loss": 6.5757,
-      "step": 1669
-    },
-    {
-      "epoch": 0.8709256844850065,
-      "grad_norm": 2.2755446434020996,
-      "learning_rate": 9.882512562814072e-05,
-      "loss": 5.7911,
-      "step": 1670
-    },
-    {
-      "epoch": 0.8714471968709256,
-      "grad_norm": 1.9824647903442383,
-      "learning_rate": 9.882412060301508e-05,
-      "loss": 6.3458,
-      "step": 1671
-    },
-    {
-      "epoch": 0.8719687092568449,
-      "grad_norm": 1.9701985120773315,
-      "learning_rate": 9.882311557788945e-05,
-      "loss": 6.5847,
-      "step": 1672
-    },
-    {
-      "epoch": 0.872490221642764,
-      "grad_norm": 2.0365712642669678,
-      "learning_rate": 9.882211055276383e-05,
-      "loss": 6.2689,
-      "step": 1673
-    },
-    {
-      "epoch": 0.8730117340286832,
-      "grad_norm": 2.18672776222229,
-      "learning_rate": 9.882110552763819e-05,
-      "loss": 5.9488,
-      "step": 1674
-    },
-    {
-      "epoch": 0.8735332464146024,
-      "grad_norm": 1.8200421333312988,
-      "learning_rate": 9.882010050251257e-05,
-      "loss": 5.9885,
-      "step": 1675
-    },
-    {
-      "epoch": 0.8740547588005215,
-      "grad_norm": 2.351130247116089,
-      "learning_rate": 9.881909547738693e-05,
-      "loss": 4.6779,
-      "step": 1676
-    },
-    {
-      "epoch": 0.8745762711864407,
-      "grad_norm": 2.151280641555786,
-      "learning_rate": 9.881809045226131e-05,
-      "loss": 5.6456,
-      "step": 1677
-    },
-    {
-      "epoch": 0.8750977835723598,
-      "grad_norm": 1.717529296875,
-      "learning_rate": 9.881708542713567e-05,
-      "loss": 5.7936,
-      "step": 1678
-    },
-    {
-      "epoch": 0.875619295958279,
-      "grad_norm": 2.2586164474487305,
-      "learning_rate": 9.881608040201005e-05,
-      "loss": 5.4351,
-      "step": 1679
-    },
-    {
-      "epoch": 0.8761408083441982,
-      "grad_norm": 2.082540512084961,
-      "learning_rate": 9.881507537688443e-05,
-      "loss": 6.1644,
-      "step": 1680
-    },
-    {
-      "epoch": 0.8766623207301173,
-      "grad_norm": 1.6716521978378296,
-      "learning_rate": 9.881407035175881e-05,
-      "loss": 6.2942,
-      "step": 1681
-    },
-    {
-      "epoch": 0.8771838331160365,
-      "grad_norm": 1.7701725959777832,
-      "learning_rate": 9.881306532663317e-05,
-      "loss": 5.9054,
-      "step": 1682
-    },
-    {
-      "epoch": 0.8777053455019557,
-      "grad_norm": 2.3273019790649414,
-      "learning_rate": 9.881206030150755e-05,
-      "loss": 5.3712,
-      "step": 1683
-    },
-    {
-      "epoch": 0.8782268578878748,
-      "grad_norm": 1.9791053533554077,
-      "learning_rate": 9.881105527638191e-05,
-      "loss": 6.2034,
-      "step": 1684
-    },
-    {
-      "epoch": 0.878748370273794,
-      "grad_norm": 1.683030605316162,
-      "learning_rate": 9.881005025125628e-05,
-      "loss": 6.5014,
-      "step": 1685
-    },
-    {
-      "epoch": 0.8792698826597132,
-      "grad_norm": 1.854114294052124,
-      "learning_rate": 9.880904522613066e-05,
-      "loss": 6.2767,
-      "step": 1686
-    },
-    {
-      "epoch": 0.8797913950456323,
-      "grad_norm": 1.6722493171691895,
-      "learning_rate": 9.880804020100502e-05,
-      "loss": 6.2782,
-      "step": 1687
-    },
-    {
-      "epoch": 0.8803129074315516,
-      "grad_norm": 1.731178879737854,
-      "learning_rate": 9.88070351758794e-05,
-      "loss": 6.1457,
-      "step": 1688
-    },
-    {
-      "epoch": 0.8808344198174707,
-      "grad_norm": 1.559341311454773,
-      "learning_rate": 9.880603015075378e-05,
-      "loss": 6.5441,
-      "step": 1689
-    },
-    {
-      "epoch": 0.8813559322033898,
-      "grad_norm": 1.5850564241409302,
-      "learning_rate": 9.880502512562815e-05,
-      "loss": 6.2866,
-      "step": 1690
-    },
-    {
-      "epoch": 0.881877444589309,
-      "grad_norm": 1.7074015140533447,
-      "learning_rate": 9.880402010050252e-05,
-      "loss": 6.5135,
-      "step": 1691
-    },
-    {
-      "epoch": 0.8823989569752282,
-      "grad_norm": 1.5369197130203247,
-      "learning_rate": 9.88030150753769e-05,
-      "loss": 6.221,
-      "step": 1692
-    },
-    {
-      "epoch": 0.8829204693611473,
-      "grad_norm": 1.7250388860702515,
-      "learning_rate": 9.880201005025126e-05,
-      "loss": 6.2168,
-      "step": 1693
-    },
-    {
-      "epoch": 0.8834419817470665,
-      "grad_norm": 1.8233838081359863,
-      "learning_rate": 9.880100502512564e-05,
-      "loss": 6.174,
-      "step": 1694
-    },
-    {
-      "epoch": 0.8839634941329857,
-      "grad_norm": 1.7374942302703857,
-      "learning_rate": 9.88e-05,
-      "loss": 5.7856,
-      "step": 1695
-    },
-    {
-      "epoch": 0.8844850065189048,
-      "grad_norm": 1.9016749858856201,
-      "learning_rate": 9.879899497487438e-05,
-      "loss": 6.3599,
-      "step": 1696
-    },
-    {
-      "epoch": 0.885006518904824,
-      "grad_norm": 1.8541254997253418,
-      "learning_rate": 9.879798994974874e-05,
-      "loss": 5.1571,
-      "step": 1697
-    },
-    {
-      "epoch": 0.8855280312907432,
-      "grad_norm": 1.592788577079773,
-      "learning_rate": 9.879698492462311e-05,
-      "loss": 6.4222,
-      "step": 1698
-    },
-    {
-      "epoch": 0.8860495436766623,
-      "grad_norm": 1.5975663661956787,
-      "learning_rate": 9.879597989949749e-05,
-      "loss": 6.2935,
-      "step": 1699
-    },
-    {
-      "epoch": 0.8865710560625815,
-      "grad_norm": 1.7215317487716675,
-      "learning_rate": 9.879497487437186e-05,
-      "loss": 6.3248,
-      "step": 1700
-    },
-    {
-      "epoch": 0.8870925684485007,
-      "grad_norm": 1.6796884536743164,
-      "learning_rate": 9.879396984924624e-05,
-      "loss": 6.1983,
-      "step": 1701
-    },
-    {
-      "epoch": 0.8876140808344198,
-      "grad_norm": 1.7115135192871094,
-      "learning_rate": 9.879296482412061e-05,
-      "loss": 6.421,
-      "step": 1702
-    },
-    {
-      "epoch": 0.888135593220339,
-      "grad_norm": 1.6824716329574585,
-      "learning_rate": 9.879195979899498e-05,
-      "loss": 6.3495,
-      "step": 1703
-    },
-    {
-      "epoch": 0.8886571056062581,
-      "grad_norm": 1.7323739528656006,
-      "learning_rate": 9.879095477386935e-05,
-      "loss": 6.4326,
-      "step": 1704
-    },
-    {
-      "epoch": 0.8891786179921773,
-      "grad_norm": 1.7057945728302002,
-      "learning_rate": 9.878994974874373e-05,
-      "loss": 6.3177,
-      "step": 1705
-    },
-    {
-      "epoch": 0.8897001303780965,
-      "grad_norm": 1.6935622692108154,
-      "learning_rate": 9.878894472361809e-05,
-      "loss": 6.4874,
-      "step": 1706
-    },
-    {
-      "epoch": 0.8902216427640156,
-      "grad_norm": 1.9630391597747803,
-      "learning_rate": 9.878793969849247e-05,
-      "loss": 6.5104,
-      "step": 1707
-    },
-    {
-      "epoch": 0.8907431551499349,
-      "grad_norm": 2.1287498474121094,
-      "learning_rate": 9.878693467336683e-05,
-      "loss": 6.3868,
-      "step": 1708
-    },
-    {
-      "epoch": 0.891264667535854,
-      "grad_norm": 1.9277119636535645,
-      "learning_rate": 9.878592964824121e-05,
-      "loss": 5.7459,
-      "step": 1709
-    },
-    {
-      "epoch": 0.8917861799217731,
-      "grad_norm": 1.9207557439804077,
-      "learning_rate": 9.878492462311559e-05,
-      "loss": 6.427,
-      "step": 1710
-    },
-    {
-      "epoch": 0.8923076923076924,
-      "grad_norm": 2.0179965496063232,
-      "learning_rate": 9.878391959798995e-05,
-      "loss": 6.3432,
-      "step": 1711
-    },
-    {
-      "epoch": 0.8928292046936115,
-      "grad_norm": 1.8834829330444336,
-      "learning_rate": 9.878291457286433e-05,
-      "loss": 6.0137,
-      "step": 1712
-    },
-    {
-      "epoch": 0.8933507170795306,
-      "grad_norm": 1.7018119096755981,
-      "learning_rate": 9.87819095477387e-05,
-      "loss": 6.4548,
-      "step": 1713
-    },
-    {
-      "epoch": 0.8938722294654498,
-      "grad_norm": 1.7061896324157715,
-      "learning_rate": 9.878090452261307e-05,
-      "loss": 5.9995,
-      "step": 1714
-    },
-    {
-      "epoch": 0.894393741851369,
-      "grad_norm": 1.7229599952697754,
-      "learning_rate": 9.877989949748744e-05,
-      "loss": 6.0517,
-      "step": 1715
-    },
-    {
-      "epoch": 0.8949152542372881,
-      "grad_norm": 1.5875067710876465,
-      "learning_rate": 9.877889447236182e-05,
-      "loss": 6.6317,
-      "step": 1716
-    },
-    {
-      "epoch": 0.8954367666232073,
-      "grad_norm": 1.7189629077911377,
-      "learning_rate": 9.877788944723618e-05,
-      "loss": 5.836,
-      "step": 1717
-    },
-    {
-      "epoch": 0.8959582790091265,
-      "grad_norm": 2.8782334327697754,
-      "learning_rate": 9.877688442211056e-05,
-      "loss": 6.144,
-      "step": 1718
-    },
-    {
-      "epoch": 0.8964797913950456,
-      "grad_norm": 1.9008127450942993,
-      "learning_rate": 9.877587939698492e-05,
-      "loss": 6.4061,
-      "step": 1719
-    },
-    {
-      "epoch": 0.8970013037809648,
-      "grad_norm": 2.296552896499634,
-      "learning_rate": 9.87748743718593e-05,
-      "loss": 5.9501,
-      "step": 1720
-    },
-    {
-      "epoch": 0.897522816166884,
-      "grad_norm": 1.9610203504562378,
-      "learning_rate": 9.877386934673368e-05,
-      "loss": 6.2823,
-      "step": 1721
-    },
-    {
-      "epoch": 0.8980443285528031,
-      "grad_norm": 1.9161206483840942,
-      "learning_rate": 9.877286432160806e-05,
-      "loss": 6.2341,
-      "step": 1722
-    },
-    {
-      "epoch": 0.8985658409387223,
-      "grad_norm": 1.5760002136230469,
-      "learning_rate": 9.877185929648242e-05,
-      "loss": 6.6321,
-      "step": 1723
-    },
-    {
-      "epoch": 0.8990873533246415,
-      "grad_norm": 1.7403117418289185,
-      "learning_rate": 9.87708542713568e-05,
-      "loss": 6.3808,
-      "step": 1724
-    },
-    {
-      "epoch": 0.8996088657105606,
-      "grad_norm": 1.792900800704956,
-      "learning_rate": 9.876984924623116e-05,
-      "loss": 6.1505,
-      "step": 1725
-    },
-    {
-      "epoch": 0.9001303780964798,
-      "grad_norm": 1.8007700443267822,
-      "learning_rate": 9.876884422110553e-05,
-      "loss": 6.4719,
-      "step": 1726
-    },
-    {
-      "epoch": 0.9006518904823989,
-      "grad_norm": 1.6678389310836792,
-      "learning_rate": 9.87678391959799e-05,
-      "loss": 6.3762,
-      "step": 1727
-    },
-    {
-      "epoch": 0.9011734028683182,
-      "grad_norm": 1.6664470434188843,
-      "learning_rate": 9.876683417085427e-05,
-      "loss": 6.3374,
-      "step": 1728
-    },
-    {
-      "epoch": 0.9016949152542373,
-      "grad_norm": 1.8322288990020752,
-      "learning_rate": 9.876582914572865e-05,
-      "loss": 6.1869,
-      "step": 1729
-    },
-    {
-      "epoch": 0.9022164276401564,
-      "grad_norm": 1.7103643417358398,
-      "learning_rate": 9.876482412060302e-05,
-      "loss": 6.0917,
-      "step": 1730
-    },
-    {
-      "epoch": 0.9027379400260757,
-      "grad_norm": 1.6477649211883545,
-      "learning_rate": 9.87638190954774e-05,
-      "loss": 5.822,
-      "step": 1731
-    },
-    {
-      "epoch": 0.9032594524119948,
-      "grad_norm": 1.7735110521316528,
-      "learning_rate": 9.876281407035177e-05,
-      "loss": 6.5537,
-      "step": 1732
-    },
-    {
-      "epoch": 0.9037809647979139,
-      "grad_norm": 1.7689518928527832,
-      "learning_rate": 9.876180904522614e-05,
-      "loss": 6.1209,
-      "step": 1733
-    },
-    {
-      "epoch": 0.9043024771838332,
-      "grad_norm": 1.736576795578003,
-      "learning_rate": 9.876080402010051e-05,
-      "loss": 6.1883,
-      "step": 1734
-    },
-    {
-      "epoch": 0.9048239895697523,
-      "grad_norm": 2.2251172065734863,
-      "learning_rate": 9.875979899497489e-05,
-      "loss": 6.0837,
-      "step": 1735
-    },
-    {
-      "epoch": 0.9053455019556714,
-      "grad_norm": 1.9843007326126099,
-      "learning_rate": 9.875879396984925e-05,
-      "loss": 6.1071,
-      "step": 1736
-    },
-    {
-      "epoch": 0.9058670143415906,
-      "grad_norm": 2.2417023181915283,
-      "learning_rate": 9.875778894472363e-05,
-      "loss": 5.9756,
-      "step": 1737
-    },
-    {
-      "epoch": 0.9063885267275098,
-      "grad_norm": 1.7956897020339966,
-      "learning_rate": 9.875678391959799e-05,
-      "loss": 6.0028,
-      "step": 1738
-    },
-    {
-      "epoch": 0.9069100391134289,
-      "grad_norm": 1.6691679954528809,
-      "learning_rate": 9.875577889447236e-05,
-      "loss": 6.2581,
-      "step": 1739
-    },
-    {
-      "epoch": 0.9074315514993481,
-      "grad_norm": 1.5963270664215088,
-      "learning_rate": 9.875477386934673e-05,
-      "loss": 6.0376,
-      "step": 1740
-    },
-    {
-      "epoch": 0.9079530638852673,
-      "grad_norm": 1.7119053602218628,
-      "learning_rate": 9.875376884422111e-05,
-      "loss": 6.2609,
-      "step": 1741
-    },
-    {
-      "epoch": 0.9084745762711864,
-      "grad_norm": 1.8534127473831177,
-      "learning_rate": 9.875276381909549e-05,
-      "loss": 6.2267,
-      "step": 1742
-    },
-    {
-      "epoch": 0.9089960886571056,
-      "grad_norm": 1.8750230073928833,
-      "learning_rate": 9.875175879396985e-05,
-      "loss": 6.3913,
-      "step": 1743
-    },
-    {
-      "epoch": 0.9095176010430248,
-      "grad_norm": 1.5363609790802002,
-      "learning_rate": 9.875075376884423e-05,
-      "loss": 6.5447,
-      "step": 1744
-    },
-    {
-      "epoch": 0.910039113428944,
-      "grad_norm": 2.0098559856414795,
-      "learning_rate": 9.87497487437186e-05,
-      "loss": 6.062,
-      "step": 1745
-    },
-    {
-      "epoch": 0.9105606258148631,
-      "grad_norm": 2.1519622802734375,
-      "learning_rate": 9.874874371859297e-05,
-      "loss": 5.6723,
-      "step": 1746
-    },
-    {
-      "epoch": 0.9110821382007823,
-      "grad_norm": 1.838989019393921,
-      "learning_rate": 9.874773869346734e-05,
-      "loss": 6.4871,
-      "step": 1747
-    },
-    {
-      "epoch": 0.9116036505867015,
-      "grad_norm": 1.7751922607421875,
-      "learning_rate": 9.874673366834172e-05,
-      "loss": 6.2561,
-      "step": 1748
-    },
-    {
-      "epoch": 0.9121251629726206,
-      "grad_norm": 1.711822271347046,
-      "learning_rate": 9.874572864321608e-05,
-      "loss": 6.4721,
-      "step": 1749
-    },
-    {
-      "epoch": 0.9126466753585397,
-      "grad_norm": 1.937978744506836,
-      "learning_rate": 9.874472361809046e-05,
-      "loss": 6.0697,
-      "step": 1750
-    },
-    {
-      "epoch": 0.913168187744459,
-      "grad_norm": 1.9958463907241821,
-      "learning_rate": 9.874371859296482e-05,
-      "loss": 6.0823,
-      "step": 1751
-    },
-    {
-      "epoch": 0.9136897001303781,
-      "grad_norm": 1.7063099145889282,
-      "learning_rate": 9.87427135678392e-05,
-      "loss": 6.4678,
-      "step": 1752
-    },
-    {
-      "epoch": 0.9142112125162972,
-      "grad_norm": 1.628139853477478,
-      "learning_rate": 9.874170854271358e-05,
-      "loss": 6.5647,
-      "step": 1753
-    },
-    {
-      "epoch": 0.9147327249022165,
-      "grad_norm": 1.9758220911026,
-      "learning_rate": 9.874070351758794e-05,
-      "loss": 6.2301,
-      "step": 1754
-    },
-    {
-      "epoch": 0.9152542372881356,
-      "grad_norm": 2.1318085193634033,
-      "learning_rate": 9.873969849246232e-05,
-      "loss": 5.5774,
-      "step": 1755
-    },
-    {
-      "epoch": 0.9157757496740547,
-      "grad_norm": 1.7579708099365234,
-      "learning_rate": 9.873869346733668e-05,
-      "loss": 5.7774,
-      "step": 1756
-    },
-    {
-      "epoch": 0.916297262059974,
-      "grad_norm": 1.7938499450683594,
-      "learning_rate": 9.873768844221106e-05,
-      "loss": 6.1348,
-      "step": 1757
-    },
-    {
-      "epoch": 0.9168187744458931,
-      "grad_norm": 1.8468101024627686,
-      "learning_rate": 9.873668341708543e-05,
-      "loss": 6.5261,
-      "step": 1758
-    },
-    {
-      "epoch": 0.9173402868318122,
-      "grad_norm": 1.8074986934661865,
-      "learning_rate": 9.87356783919598e-05,
-      "loss": 6.4841,
-      "step": 1759
-    },
-    {
-      "epoch": 0.9178617992177314,
-      "grad_norm": 1.7092223167419434,
-      "learning_rate": 9.873467336683417e-05,
-      "loss": 6.3841,
-      "step": 1760
-    },
-    {
-      "epoch": 0.9183833116036506,
-      "grad_norm": 1.7861287593841553,
-      "learning_rate": 9.873366834170855e-05,
-      "loss": 6.355,
-      "step": 1761
-    },
-    {
-      "epoch": 0.9189048239895697,
-      "grad_norm": 1.6930010318756104,
-      "learning_rate": 9.873266331658292e-05,
-      "loss": 6.3363,
-      "step": 1762
-    },
-    {
-      "epoch": 0.9194263363754889,
-      "grad_norm": 1.8475645780563354,
-      "learning_rate": 9.87316582914573e-05,
-      "loss": 5.8782,
-      "step": 1763
-    },
-    {
-      "epoch": 0.9199478487614081,
-      "grad_norm": 1.5209002494812012,
-      "learning_rate": 9.873065326633167e-05,
-      "loss": 6.0526,
-      "step": 1764
-    },
-    {
-      "epoch": 0.9204693611473272,
-      "grad_norm": 1.700175166130066,
-      "learning_rate": 9.872964824120603e-05,
-      "loss": 6.5621,
-      "step": 1765
-    },
-    {
-      "epoch": 0.9209908735332464,
-      "grad_norm": 1.5412355661392212,
-      "learning_rate": 9.872864321608041e-05,
-      "loss": 6.4583,
-      "step": 1766
-    },
-    {
-      "epoch": 0.9215123859191656,
-      "grad_norm": 1.6577399969100952,
-      "learning_rate": 9.872763819095477e-05,
-      "loss": 6.0804,
-      "step": 1767
-    },
-    {
-      "epoch": 0.9220338983050848,
-      "grad_norm": 1.7236900329589844,
-      "learning_rate": 9.872663316582915e-05,
-      "loss": 6.3388,
-      "step": 1768
-    },
-    {
-      "epoch": 0.9225554106910039,
-      "grad_norm": 1.93367338180542,
-      "learning_rate": 9.872562814070351e-05,
-      "loss": 5.9426,
-      "step": 1769
-    },
-    {
-      "epoch": 0.9230769230769231,
-      "grad_norm": 1.6736849546432495,
-      "learning_rate": 9.872462311557789e-05,
-      "loss": 6.1373,
-      "step": 1770
-    },
-    {
-      "epoch": 0.9235984354628423,
-      "grad_norm": 1.5711864233016968,
-      "learning_rate": 9.872361809045226e-05,
-      "loss": 6.2082,
-      "step": 1771
-    },
-    {
-      "epoch": 0.9241199478487614,
-      "grad_norm": 1.7657711505889893,
-      "learning_rate": 9.872261306532663e-05,
-      "loss": 5.958,
-      "step": 1772
-    },
-    {
-      "epoch": 0.9246414602346805,
-      "grad_norm": 1.7789695262908936,
-      "learning_rate": 9.872160804020101e-05,
-      "loss": 6.1682,
-      "step": 1773
-    },
-    {
-      "epoch": 0.9251629726205998,
-      "grad_norm": 1.6570029258728027,
-      "learning_rate": 9.872060301507539e-05,
-      "loss": 6.2757,
-      "step": 1774
-    },
-    {
-      "epoch": 0.9256844850065189,
-      "grad_norm": 1.5490208864212036,
-      "learning_rate": 9.871959798994975e-05,
-      "loss": 6.4814,
-      "step": 1775
-    },
-    {
-      "epoch": 0.926205997392438,
-      "grad_norm": 1.5855228900909424,
-      "learning_rate": 9.871859296482413e-05,
-      "loss": 6.3562,
-      "step": 1776
-    },
-    {
-      "epoch": 0.9267275097783573,
-      "grad_norm": 1.965667963027954,
-      "learning_rate": 9.87175879396985e-05,
-      "loss": 5.9576,
-      "step": 1777
-    },
-    {
-      "epoch": 0.9272490221642764,
-      "grad_norm": 1.893012285232544,
-      "learning_rate": 9.871658291457286e-05,
-      "loss": 5.8361,
-      "step": 1778
-    },
-    {
-      "epoch": 0.9277705345501955,
-      "grad_norm": 1.981179118156433,
-      "learning_rate": 9.871557788944724e-05,
-      "loss": 5.4058,
-      "step": 1779
-    },
-    {
-      "epoch": 0.9282920469361148,
-      "grad_norm": 2.0810556411743164,
-      "learning_rate": 9.87145728643216e-05,
-      "loss": 6.0086,
-      "step": 1780
-    },
-    {
-      "epoch": 0.9288135593220339,
-      "grad_norm": 1.7562342882156372,
-      "learning_rate": 9.871356783919598e-05,
-      "loss": 6.1404,
-      "step": 1781
-    },
-    {
-      "epoch": 0.929335071707953,
-      "grad_norm": 1.414299726486206,
-      "learning_rate": 9.871256281407036e-05,
-      "loss": 6.5221,
-      "step": 1782
-    },
-    {
-      "epoch": 0.9298565840938723,
-      "grad_norm": 2.073514461517334,
-      "learning_rate": 9.871155778894474e-05,
-      "loss": 5.9284,
-      "step": 1783
-    },
-    {
-      "epoch": 0.9303780964797914,
-      "grad_norm": 1.5229054689407349,
-      "learning_rate": 9.87105527638191e-05,
-      "loss": 6.6462,
-      "step": 1784
-    },
-    {
-      "epoch": 0.9308996088657105,
-      "grad_norm": 1.5980912446975708,
-      "learning_rate": 9.870954773869348e-05,
-      "loss": 6.0822,
-      "step": 1785
-    },
-    {
-      "epoch": 0.9314211212516297,
-      "grad_norm": 1.7462660074234009,
-      "learning_rate": 9.870854271356784e-05,
-      "loss": 6.36,
-      "step": 1786
-    },
-    {
-      "epoch": 0.9319426336375489,
-      "grad_norm": 2.215930700302124,
-      "learning_rate": 9.870753768844222e-05,
-      "loss": 5.7601,
-      "step": 1787
-    },
-    {
-      "epoch": 0.932464146023468,
-      "grad_norm": 1.588579773902893,
-      "learning_rate": 9.870653266331659e-05,
-      "loss": 6.3966,
-      "step": 1788
-    },
-    {
-      "epoch": 0.9329856584093872,
-      "grad_norm": 1.9071831703186035,
-      "learning_rate": 9.870552763819096e-05,
-      "loss": 6.2341,
-      "step": 1789
-    },
-    {
-      "epoch": 0.9335071707953064,
-      "grad_norm": 1.9192125797271729,
-      "learning_rate": 9.870452261306533e-05,
-      "loss": 6.0921,
-      "step": 1790
-    },
-    {
-      "epoch": 0.9340286831812256,
-      "grad_norm": 1.903688669204712,
-      "learning_rate": 9.870351758793969e-05,
-      "loss": 6.0152,
-      "step": 1791
-    },
-    {
-      "epoch": 0.9345501955671447,
-      "grad_norm": 2.2227704524993896,
-      "learning_rate": 9.870251256281407e-05,
-      "loss": 5.5872,
-      "step": 1792
-    },
-    {
-      "epoch": 0.9350717079530639,
-      "grad_norm": 2.658254623413086,
-      "learning_rate": 9.870150753768845e-05,
-      "loss": 6.2801,
-      "step": 1793
-    },
-    {
-      "epoch": 0.9355932203389831,
-      "grad_norm": 1.718497633934021,
-      "learning_rate": 9.870050251256282e-05,
-      "loss": 6.3553,
-      "step": 1794
-    },
-    {
-      "epoch": 0.9361147327249022,
-      "grad_norm": 1.6464396715164185,
-      "learning_rate": 9.869949748743719e-05,
-      "loss": 6.3048,
-      "step": 1795
-    },
-    {
-      "epoch": 0.9366362451108213,
-      "grad_norm": 1.729474425315857,
-      "learning_rate": 9.869849246231157e-05,
-      "loss": 6.1583,
-      "step": 1796
-    },
-    {
-      "epoch": 0.9371577574967406,
-      "grad_norm": 1.7387269735336304,
-      "learning_rate": 9.869748743718593e-05,
-      "loss": 6.2634,
-      "step": 1797
-    },
-    {
-      "epoch": 0.9376792698826597,
-      "grad_norm": 1.7279126644134521,
-      "learning_rate": 9.869648241206031e-05,
-      "loss": 6.073,
-      "step": 1798
-    },
-    {
-      "epoch": 0.9382007822685788,
-      "grad_norm": 1.6976224184036255,
-      "learning_rate": 9.869547738693467e-05,
-      "loss": 6.6056,
-      "step": 1799
-    },
-    {
-      "epoch": 0.9387222946544981,
-      "grad_norm": 1.962341547012329,
-      "learning_rate": 9.869447236180905e-05,
-      "loss": 6.3039,
-      "step": 1800
-    },
-    {
-      "epoch": 0.9392438070404172,
-      "grad_norm": 1.8708937168121338,
-      "learning_rate": 9.869346733668342e-05,
-      "loss": 6.4098,
-      "step": 1801
-    },
-    {
-      "epoch": 0.9397653194263363,
-      "grad_norm": 1.9174119234085083,
-      "learning_rate": 9.86924623115578e-05,
-      "loss": 6.2019,
-      "step": 1802
-    },
-    {
-      "epoch": 0.9402868318122556,
-      "grad_norm": 1.9023312330245972,
-      "learning_rate": 9.869145728643217e-05,
-      "loss": 6.0684,
-      "step": 1803
-    },
-    {
-      "epoch": 0.9408083441981747,
-      "grad_norm": 1.9731063842773438,
-      "learning_rate": 9.869045226130654e-05,
-      "loss": 6.2027,
-      "step": 1804
-    },
-    {
-      "epoch": 0.9413298565840938,
-      "grad_norm": 1.8011584281921387,
-      "learning_rate": 9.868944723618091e-05,
-      "loss": 6.3769,
-      "step": 1805
-    },
-    {
-      "epoch": 0.9418513689700131,
-      "grad_norm": 1.8728646039962769,
-      "learning_rate": 9.868844221105528e-05,
-      "loss": 5.9545,
-      "step": 1806
-    },
-    {
-      "epoch": 0.9423728813559322,
-      "grad_norm": 1.7673097848892212,
-      "learning_rate": 9.868743718592966e-05,
-      "loss": 6.1181,
-      "step": 1807
-    },
-    {
-      "epoch": 0.9428943937418514,
-      "grad_norm": 1.7257661819458008,
-      "learning_rate": 9.868643216080402e-05,
-      "loss": 6.4183,
-      "step": 1808
-    },
-    {
-      "epoch": 0.9434159061277705,
-      "grad_norm": 1.774137020111084,
-      "learning_rate": 9.86854271356784e-05,
-      "loss": 6.1167,
-      "step": 1809
-    },
-    {
-      "epoch": 0.9439374185136897,
-      "grad_norm": 1.786062479019165,
-      "learning_rate": 9.868442211055276e-05,
-      "loss": 5.9211,
-      "step": 1810
-    },
-    {
-      "epoch": 0.9444589308996089,
-      "grad_norm": 3.52620530128479,
-      "learning_rate": 9.868341708542714e-05,
-      "loss": 6.0918,
-      "step": 1811
-    },
-    {
-      "epoch": 0.944980443285528,
-      "grad_norm": 1.9542983770370483,
-      "learning_rate": 9.86824120603015e-05,
-      "loss": 6.0449,
-      "step": 1812
-    },
-    {
-      "epoch": 0.9455019556714472,
-      "grad_norm": 1.9136114120483398,
-      "learning_rate": 9.868140703517588e-05,
-      "loss": 6.4076,
-      "step": 1813
-    },
-    {
-      "epoch": 0.9460234680573664,
-      "grad_norm": 1.745273232460022,
-      "learning_rate": 9.868040201005026e-05,
-      "loss": 6.477,
-      "step": 1814
-    },
-    {
-      "epoch": 0.9465449804432855,
-      "grad_norm": 1.7708137035369873,
-      "learning_rate": 9.867939698492464e-05,
-      "loss": 6.3215,
-      "step": 1815
-    },
-    {
-      "epoch": 0.9470664928292047,
-      "grad_norm": 1.7978767156600952,
-      "learning_rate": 9.8678391959799e-05,
-      "loss": 6.2015,
-      "step": 1816
-    },
-    {
-      "epoch": 0.9475880052151239,
-      "grad_norm": 1.886091947555542,
-      "learning_rate": 9.867738693467338e-05,
-      "loss": 5.5708,
-      "step": 1817
-    },
-    {
-      "epoch": 0.948109517601043,
-      "grad_norm": 2.059328079223633,
-      "learning_rate": 9.867638190954774e-05,
-      "loss": 6.1685,
-      "step": 1818
-    },
-    {
-      "epoch": 0.9486310299869621,
-      "grad_norm": 1.7448359727859497,
-      "learning_rate": 9.867537688442211e-05,
-      "loss": 6.3786,
-      "step": 1819
-    },
-    {
-      "epoch": 0.9491525423728814,
-      "grad_norm": 1.4995218515396118,
-      "learning_rate": 9.867437185929649e-05,
-      "loss": 5.0233,
-      "step": 1820
-    },
-    {
-      "epoch": 0.9496740547588005,
-      "grad_norm": 1.9038867950439453,
-      "learning_rate": 9.867336683417085e-05,
-      "loss": 6.2677,
-      "step": 1821
-    },
-    {
-      "epoch": 0.9501955671447196,
-      "grad_norm": 1.964781403541565,
-      "learning_rate": 9.867236180904523e-05,
-      "loss": 6.0163,
-      "step": 1822
-    },
-    {
-      "epoch": 0.9507170795306389,
-      "grad_norm": 1.859845757484436,
-      "learning_rate": 9.86713567839196e-05,
-      "loss": 6.0896,
-      "step": 1823
-    },
-    {
-      "epoch": 0.951238591916558,
-      "grad_norm": 1.9396536350250244,
-      "learning_rate": 9.867035175879398e-05,
-      "loss": 6.2767,
-      "step": 1824
-    },
-    {
-      "epoch": 0.9517601043024772,
-      "grad_norm": 1.6548652648925781,
-      "learning_rate": 9.866934673366835e-05,
-      "loss": 6.3923,
-      "step": 1825
-    },
-    {
-      "epoch": 0.9522816166883964,
-      "grad_norm": 1.84656822681427,
-      "learning_rate": 9.866834170854273e-05,
-      "loss": 5.658,
-      "step": 1826
-    },
-    {
-      "epoch": 0.9528031290743155,
-      "grad_norm": 1.8334338665008545,
-      "learning_rate": 9.866733668341709e-05,
-      "loss": 6.1049,
-      "step": 1827
-    },
-    {
-      "epoch": 0.9533246414602347,
-      "grad_norm": 1.528545618057251,
-      "learning_rate": 9.866633165829147e-05,
-      "loss": 6.3846,
-      "step": 1828
-    },
-    {
-      "epoch": 0.9538461538461539,
-      "grad_norm": 1.863281488418579,
-      "learning_rate": 9.866532663316583e-05,
-      "loss": 6.0952,
-      "step": 1829
-    },
-    {
-      "epoch": 0.954367666232073,
-      "grad_norm": 1.7164326906204224,
-      "learning_rate": 9.866432160804021e-05,
-      "loss": 6.4763,
-      "step": 1830
-    },
-    {
-      "epoch": 0.9548891786179922,
-      "grad_norm": 1.736375331878662,
-      "learning_rate": 9.866331658291457e-05,
-      "loss": 6.2349,
-      "step": 1831
-    },
-    {
-      "epoch": 0.9554106910039113,
-      "grad_norm": 1.8414313793182373,
-      "learning_rate": 9.866231155778894e-05,
-      "loss": 6.3186,
-      "step": 1832
-    },
-    {
-      "epoch": 0.9559322033898305,
-      "grad_norm": 1.798867106437683,
-      "learning_rate": 9.866130653266332e-05,
-      "loss": 5.9766,
-      "step": 1833
-    },
-    {
-      "epoch": 0.9564537157757497,
-      "grad_norm": 1.835066318511963,
-      "learning_rate": 9.86603015075377e-05,
-      "loss": 6.1279,
-      "step": 1834
-    },
-    {
-      "epoch": 0.9569752281616688,
-      "grad_norm": 1.9193572998046875,
-      "learning_rate": 9.865929648241207e-05,
-      "loss": 6.0124,
-      "step": 1835
-    },
-    {
-      "epoch": 0.957496740547588,
-      "grad_norm": 1.6664605140686035,
-      "learning_rate": 9.865829145728644e-05,
-      "loss": 6.1305,
-      "step": 1836
-    },
-    {
-      "epoch": 0.9580182529335072,
-      "grad_norm": 2.0482826232910156,
-      "learning_rate": 9.865728643216081e-05,
-      "loss": 6.1755,
-      "step": 1837
-    },
-    {
-      "epoch": 0.9585397653194263,
-      "grad_norm": 1.9236359596252441,
-      "learning_rate": 9.865628140703518e-05,
-      "loss": 5.6478,
-      "step": 1838
-    },
-    {
-      "epoch": 0.9590612777053455,
-      "grad_norm": 1.6822118759155273,
-      "learning_rate": 9.865527638190956e-05,
-      "loss": 6.1067,
-      "step": 1839
-    },
-    {
-      "epoch": 0.9595827900912647,
-      "grad_norm": 1.6255333423614502,
-      "learning_rate": 9.865427135678392e-05,
-      "loss": 6.2461,
-      "step": 1840
-    },
-    {
-      "epoch": 0.9601043024771838,
-      "grad_norm": 1.6045290231704712,
-      "learning_rate": 9.86532663316583e-05,
-      "loss": 6.3111,
-      "step": 1841
-    },
-    {
-      "epoch": 0.960625814863103,
-      "grad_norm": 1.4453548192977905,
-      "learning_rate": 9.865226130653266e-05,
-      "loss": 6.453,
-      "step": 1842
-    },
-    {
-      "epoch": 0.9611473272490222,
-      "grad_norm": 1.9731504917144775,
-      "learning_rate": 9.865125628140704e-05,
-      "loss": 6.0372,
-      "step": 1843
-    },
-    {
-      "epoch": 0.9616688396349413,
-      "grad_norm": 1.813751220703125,
-      "learning_rate": 9.865025125628142e-05,
-      "loss": 6.0699,
-      "step": 1844
-    },
-    {
-      "epoch": 0.9621903520208605,
-      "grad_norm": 1.7470742464065552,
-      "learning_rate": 9.864924623115578e-05,
-      "loss": 6.2216,
-      "step": 1845
-    },
-    {
-      "epoch": 0.9627118644067797,
-      "grad_norm": 1.889413595199585,
-      "learning_rate": 9.864824120603016e-05,
-      "loss": 6.3146,
-      "step": 1846
-    },
-    {
-      "epoch": 0.9632333767926988,
-      "grad_norm": 1.9028747081756592,
-      "learning_rate": 9.864723618090452e-05,
-      "loss": 6.115,
-      "step": 1847
-    },
-    {
-      "epoch": 0.963754889178618,
-      "grad_norm": 1.8532224893569946,
-      "learning_rate": 9.86462311557789e-05,
-      "loss": 6.1386,
-      "step": 1848
-    },
-    {
-      "epoch": 0.9642764015645372,
-      "grad_norm": 1.599428415298462,
-      "learning_rate": 9.864522613065327e-05,
-      "loss": 6.4411,
-      "step": 1849
-    },
-    {
-      "epoch": 0.9647979139504563,
-      "grad_norm": 1.677685260772705,
-      "learning_rate": 9.864422110552764e-05,
-      "loss": 6.4012,
-      "step": 1850
-    },
-    {
-      "epoch": 0.9653194263363755,
-      "grad_norm": 1.7733136415481567,
-      "learning_rate": 9.864321608040201e-05,
-      "loss": 5.9693,
-      "step": 1851
-    },
-    {
-      "epoch": 0.9658409387222947,
-      "grad_norm": 1.55734121799469,
-      "learning_rate": 9.864221105527639e-05,
-      "loss": 6.2712,
-      "step": 1852
-    },
-    {
-      "epoch": 0.9663624511082138,
-      "grad_norm": 1.5608091354370117,
-      "learning_rate": 9.864120603015075e-05,
-      "loss": 6.07,
-      "step": 1853
-    },
-    {
-      "epoch": 0.966883963494133,
-      "grad_norm": 1.8081550598144531,
-      "learning_rate": 9.864020100502513e-05,
-      "loss": 6.1857,
-      "step": 1854
-    },
-    {
-      "epoch": 0.9674054758800521,
-      "grad_norm": 1.903599739074707,
-      "learning_rate": 9.86391959798995e-05,
-      "loss": 5.9202,
-      "step": 1855
-    },
-    {
-      "epoch": 0.9679269882659713,
-      "grad_norm": 2.073634624481201,
-      "learning_rate": 9.863819095477388e-05,
-      "loss": 6.4165,
-      "step": 1856
-    },
-    {
-      "epoch": 0.9684485006518905,
-      "grad_norm": 1.8562136888504028,
-      "learning_rate": 9.863718592964825e-05,
-      "loss": 5.9238,
-      "step": 1857
-    },
-    {
-      "epoch": 0.9689700130378096,
-      "grad_norm": 1.9968205690383911,
-      "learning_rate": 9.863618090452261e-05,
-      "loss": 5.905,
-      "step": 1858
-    },
-    {
-      "epoch": 0.9694915254237289,
-      "grad_norm": 1.8912436962127686,
-      "learning_rate": 9.863517587939699e-05,
-      "loss": 5.9879,
-      "step": 1859
-    },
-    {
-      "epoch": 0.970013037809648,
-      "grad_norm": 1.8201597929000854,
-      "learning_rate": 9.863417085427136e-05,
-      "loss": 6.3433,
-      "step": 1860
-    },
-    {
-      "epoch": 0.9705345501955671,
-      "grad_norm": 1.501621127128601,
-      "learning_rate": 9.863316582914573e-05,
-      "loss": 6.2233,
-      "step": 1861
-    },
-    {
-      "epoch": 0.9710560625814864,
-      "grad_norm": 1.7901490926742554,
-      "learning_rate": 9.86321608040201e-05,
-      "loss": 6.4165,
-      "step": 1862
-    },
-    {
-      "epoch": 0.9715775749674055,
-      "grad_norm": 1.8126524686813354,
-      "learning_rate": 9.863115577889447e-05,
-      "loss": 6.2119,
-      "step": 1863
-    },
-    {
-      "epoch": 0.9720990873533246,
-      "grad_norm": 1.8406113386154175,
-      "learning_rate": 9.863015075376885e-05,
-      "loss": 6.1828,
-      "step": 1864
-    },
-    {
-      "epoch": 0.9726205997392438,
-      "grad_norm": 1.9171116352081299,
-      "learning_rate": 9.862914572864323e-05,
-      "loss": 6.1778,
-      "step": 1865
-    },
-    {
-      "epoch": 0.973142112125163,
-      "grad_norm": 1.883215069770813,
-      "learning_rate": 9.86281407035176e-05,
-      "loss": 6.0915,
-      "step": 1866
-    },
-    {
-      "epoch": 0.9736636245110821,
-      "grad_norm": 1.798251986503601,
-      "learning_rate": 9.862713567839197e-05,
-      "loss": 6.4367,
-      "step": 1867
-    },
-    {
-      "epoch": 0.9741851368970013,
-      "grad_norm": 1.9052170515060425,
-      "learning_rate": 9.862613065326634e-05,
-      "loss": 5.5588,
-      "step": 1868
-    },
-    {
-      "epoch": 0.9747066492829205,
-      "grad_norm": 1.575830340385437,
-      "learning_rate": 9.862512562814071e-05,
-      "loss": 6.0571,
-      "step": 1869
-    },
-    {
-      "epoch": 0.9752281616688396,
-      "grad_norm": 1.853869915008545,
-      "learning_rate": 9.862412060301508e-05,
-      "loss": 6.0275,
-      "step": 1870
-    },
-    {
-      "epoch": 0.9757496740547588,
-      "grad_norm": 1.7869341373443604,
-      "learning_rate": 9.862311557788944e-05,
-      "loss": 6.3859,
-      "step": 1871
-    },
-    {
-      "epoch": 0.976271186440678,
-      "grad_norm": 1.8462800979614258,
-      "learning_rate": 9.862211055276382e-05,
-      "loss": 5.3611,
-      "step": 1872
-    },
-    {
-      "epoch": 0.9767926988265971,
-      "grad_norm": 1.9830440282821655,
-      "learning_rate": 9.862110552763819e-05,
-      "loss": 6.1347,
-      "step": 1873
-    },
-    {
-      "epoch": 0.9773142112125163,
-      "grad_norm": 1.848233938217163,
-      "learning_rate": 9.862010050251256e-05,
-      "loss": 5.7121,
-      "step": 1874
-    },
-    {
-      "epoch": 0.9778357235984355,
-      "grad_norm": 1.6460009813308716,
-      "learning_rate": 9.861909547738694e-05,
-      "loss": 6.2635,
-      "step": 1875
-    },
-    {
-      "epoch": 0.9783572359843546,
-      "grad_norm": 1.6736358404159546,
-      "learning_rate": 9.861809045226132e-05,
-      "loss": 6.3444,
-      "step": 1876
-    },
-    {
-      "epoch": 0.9788787483702738,
-      "grad_norm": 2.030003547668457,
-      "learning_rate": 9.861708542713568e-05,
-      "loss": 5.6073,
-      "step": 1877
-    },
-    {
-      "epoch": 0.9794002607561929,
-      "grad_norm": 1.8256900310516357,
-      "learning_rate": 9.861608040201006e-05,
-      "loss": 5.9125,
-      "step": 1878
-    },
-    {
-      "epoch": 0.9799217731421122,
-      "grad_norm": 1.9384441375732422,
-      "learning_rate": 9.861507537688443e-05,
-      "loss": 6.2115,
-      "step": 1879
-    },
-    {
-      "epoch": 0.9804432855280313,
-      "grad_norm": 1.758963942527771,
-      "learning_rate": 9.86140703517588e-05,
-      "loss": 6.1118,
-      "step": 1880
-    },
-    {
-      "epoch": 0.9809647979139504,
-      "grad_norm": 2.266641855239868,
-      "learning_rate": 9.861306532663317e-05,
-      "loss": 5.9316,
-      "step": 1881
-    },
-    {
-      "epoch": 0.9814863102998697,
-      "grad_norm": 2.2349464893341064,
-      "learning_rate": 9.861206030150755e-05,
-      "loss": 6.0164,
-      "step": 1882
-    },
-    {
-      "epoch": 0.9820078226857888,
-      "grad_norm": 2.1468758583068848,
-      "learning_rate": 9.861105527638191e-05,
-      "loss": 5.8991,
-      "step": 1883
-    },
-    {
-      "epoch": 0.9825293350717079,
-      "grad_norm": 2.0775907039642334,
-      "learning_rate": 9.861005025125629e-05,
-      "loss": 6.0994,
-      "step": 1884
-    },
-    {
-      "epoch": 0.9830508474576272,
-      "grad_norm": 1.7565312385559082,
-      "learning_rate": 9.860904522613067e-05,
-      "loss": 6.1407,
-      "step": 1885
-    },
-    {
-      "epoch": 0.9835723598435463,
-      "grad_norm": 1.8905107975006104,
-      "learning_rate": 9.860804020100503e-05,
-      "loss": 6.3397,
-      "step": 1886
-    },
-    {
-      "epoch": 0.9840938722294654,
-      "grad_norm": 1.8091247081756592,
-      "learning_rate": 9.860703517587941e-05,
-      "loss": 6.4454,
-      "step": 1887
-    },
-    {
-      "epoch": 0.9846153846153847,
-      "grad_norm": 1.7579909563064575,
-      "learning_rate": 9.860603015075377e-05,
-      "loss": 6.4286,
-      "step": 1888
-    },
-    {
-      "epoch": 0.9851368970013038,
-      "grad_norm": 1.901070237159729,
-      "learning_rate": 9.860502512562815e-05,
-      "loss": 6.3575,
-      "step": 1889
-    },
-    {
-      "epoch": 0.9856584093872229,
-      "grad_norm": 2.136735200881958,
-      "learning_rate": 9.860402010050251e-05,
-      "loss": 5.8651,
-      "step": 1890
-    },
-    {
-      "epoch": 0.9861799217731421,
-      "grad_norm": 1.7569983005523682,
-      "learning_rate": 9.860301507537689e-05,
-      "loss": 6.2668,
-      "step": 1891
-    },
-    {
-      "epoch": 0.9867014341590613,
-      "grad_norm": 1.9299886226654053,
-      "learning_rate": 9.860201005025126e-05,
-      "loss": 6.1899,
-      "step": 1892
-    },
-    {
-      "epoch": 0.9872229465449804,
-      "grad_norm": 1.8312844038009644,
-      "learning_rate": 9.860100502512563e-05,
-      "loss": 5.8749,
-      "step": 1893
-    },
-    {
-      "epoch": 0.9877444589308996,
-      "grad_norm": 1.7669068574905396,
-      "learning_rate": 9.86e-05,
-      "loss": 5.3984,
-      "step": 1894
-    },
-    {
-      "epoch": 0.9882659713168188,
-      "grad_norm": 2.1258013248443604,
-      "learning_rate": 9.859899497487438e-05,
-      "loss": 5.7357,
-      "step": 1895
-    },
-    {
-      "epoch": 0.988787483702738,
-      "grad_norm": 1.752524971961975,
-      "learning_rate": 9.859798994974875e-05,
-      "loss": 5.7311,
-      "step": 1896
-    },
-    {
-      "epoch": 0.9893089960886571,
-      "grad_norm": 1.7557425498962402,
-      "learning_rate": 9.859698492462312e-05,
-      "loss": 5.766,
-      "step": 1897
-    },
-    {
-      "epoch": 0.9898305084745763,
-      "grad_norm": 1.8874051570892334,
-      "learning_rate": 9.85959798994975e-05,
-      "loss": 6.0503,
-      "step": 1898
-    },
-    {
-      "epoch": 0.9903520208604955,
-      "grad_norm": 1.5621223449707031,
-      "learning_rate": 9.859497487437186e-05,
-      "loss": 6.5691,
-      "step": 1899
-    },
-    {
-      "epoch": 0.9908735332464146,
-      "grad_norm": 1.6794296503067017,
-      "learning_rate": 9.859396984924624e-05,
-      "loss": 6.4146,
-      "step": 1900
-    },
-    {
-      "epoch": 0.9913950456323337,
-      "grad_norm": 1.704735279083252,
-      "learning_rate": 9.85929648241206e-05,
-      "loss": 6.3028,
-      "step": 1901
-    },
-    {
-      "epoch": 0.991916558018253,
-      "grad_norm": 1.6560001373291016,
-      "learning_rate": 9.859195979899498e-05,
-      "loss": 6.0321,
-      "step": 1902
-    },
-    {
-      "epoch": 0.9924380704041721,
-      "grad_norm": 1.8004179000854492,
-      "learning_rate": 9.859095477386934e-05,
-      "loss": 5.9194,
-      "step": 1903
-    },
-    {
-      "epoch": 0.9929595827900912,
-      "grad_norm": 1.9367026090621948,
-      "learning_rate": 9.858994974874372e-05,
-      "loss": 6.1279,
-      "step": 1904
-    },
-    {
-      "epoch": 0.9934810951760105,
-      "grad_norm": 1.9425865411758423,
-      "learning_rate": 9.858894472361809e-05,
-      "loss": 6.5871,
-      "step": 1905
-    },
-    {
-      "epoch": 0.9940026075619296,
-      "grad_norm": 1.9378076791763306,
-      "learning_rate": 9.858793969849246e-05,
-      "loss": 6.3338,
-      "step": 1906
-    },
-    {
-      "epoch": 0.9945241199478487,
-      "grad_norm": 2.6838061809539795,
-      "learning_rate": 9.858693467336684e-05,
-      "loss": 5.4407,
-      "step": 1907
-    },
-    {
-      "epoch": 0.995045632333768,
-      "grad_norm": 2.263880729675293,
-      "learning_rate": 9.858592964824122e-05,
-      "loss": 5.8836,
-      "step": 1908
-    },
-    {
-      "epoch": 0.9955671447196871,
-      "grad_norm": 1.8455873727798462,
-      "learning_rate": 9.858492462311558e-05,
-      "loss": 6.3885,
-      "step": 1909
-    },
-    {
-      "epoch": 0.9960886571056062,
-      "grad_norm": 1.9014992713928223,
-      "learning_rate": 9.858391959798996e-05,
-      "loss": 6.3106,
-      "step": 1910
-    },
-    {
-      "epoch": 0.9966101694915255,
-      "grad_norm": 1.7918483018875122,
-      "learning_rate": 9.858291457286433e-05,
-      "loss": 6.2332,
-      "step": 1911
-    },
-    {
-      "epoch": 0.9971316818774446,
-      "grad_norm": 1.7935258150100708,
-      "learning_rate": 9.858190954773869e-05,
-      "loss": 6.0331,
-      "step": 1912
-    },
-    {
-      "epoch": 0.9976531942633637,
-      "grad_norm": 1.972747564315796,
-      "learning_rate": 9.858090452261307e-05,
-      "loss": 6.0963,
-      "step": 1913
-    },
-    {
-      "epoch": 0.9981747066492829,
-      "grad_norm": 1.479304552078247,
-      "learning_rate": 9.857989949748743e-05,
-      "loss": 6.2654,
-      "step": 1914
-    },
-    {
-      "epoch": 0.9986962190352021,
-      "grad_norm": 2.468888282775879,
-      "learning_rate": 9.857889447236181e-05,
-      "loss": 5.4459,
-      "step": 1915
-    },
-    {
-      "epoch": 0.9992177314211212,
-      "grad_norm": 2.081630229949951,
-      "learning_rate": 9.857788944723619e-05,
-      "loss": 6.3112,
-      "step": 1916
-    },
-    {
-      "epoch": 0.9997392438070404,
-      "grad_norm": 2.034188747406006,
-      "learning_rate": 9.857688442211057e-05,
-      "loss": 6.0643,
-      "step": 1917
-    },
-    {
-      "epoch": 1.0002607561929595,
-      "grad_norm": 1.9129623174667358,
-      "learning_rate": 9.857587939698493e-05,
-      "loss": 6.4628,
-      "step": 1918
-    },
-    {
-      "epoch": 1.0007822685788788,
-      "grad_norm": 1.8343724012374878,
-      "learning_rate": 9.857487437185931e-05,
-      "loss": 6.4646,
-      "step": 1919
-    },
-    {
-      "epoch": 1.001303780964798,
-      "grad_norm": 1.682604193687439,
-      "learning_rate": 9.857386934673367e-05,
-      "loss": 6.3218,
-      "step": 1920
-    },
-    {
-      "epoch": 1.001825293350717,
-      "grad_norm": 1.7998607158660889,
-      "learning_rate": 9.857286432160805e-05,
-      "loss": 6.4673,
-      "step": 1921
-    },
-    {
-      "epoch": 1.0023468057366363,
-      "grad_norm": 1.6766268014907837,
-      "learning_rate": 9.857185929648241e-05,
-      "loss": 6.4773,
-      "step": 1922
-    },
-    {
-      "epoch": 1.0028683181225555,
-      "grad_norm": 1.8383983373641968,
-      "learning_rate": 9.857085427135679e-05,
-      "loss": 6.0813,
-      "step": 1923
-    },
-    {
-      "epoch": 1.0033898305084745,
-      "grad_norm": 1.5827420949935913,
-      "learning_rate": 9.856984924623116e-05,
-      "loss": 6.298,
-      "step": 1924
-    },
-    {
-      "epoch": 1.0039113428943938,
-      "grad_norm": 1.828986644744873,
-      "learning_rate": 9.856884422110552e-05,
-      "loss": 6.0962,
-      "step": 1925
-    },
-    {
-      "epoch": 1.004432855280313,
-      "grad_norm": 1.7401952743530273,
-      "learning_rate": 9.85678391959799e-05,
-      "loss": 6.1504,
-      "step": 1926
-    },
-    {
-      "epoch": 1.004954367666232,
-      "grad_norm": 1.6379196643829346,
-      "learning_rate": 9.856683417085428e-05,
-      "loss": 6.0205,
-      "step": 1927
-    },
-    {
-      "epoch": 1.0054758800521513,
-      "grad_norm": 2.1414053440093994,
-      "learning_rate": 9.856582914572865e-05,
-      "loss": 5.9953,
-      "step": 1928
-    },
-    {
-      "epoch": 1.0059973924380705,
-      "grad_norm": 2.381213903427124,
-      "learning_rate": 9.856482412060302e-05,
-      "loss": 5.7121,
-      "step": 1929
-    },
-    {
-      "epoch": 1.0065189048239895,
-      "grad_norm": 1.7482136487960815,
-      "learning_rate": 9.85638190954774e-05,
-      "loss": 6.1058,
-      "step": 1930
-    },
-    {
-      "epoch": 1.0070404172099088,
-      "grad_norm": 1.5391474962234497,
-      "learning_rate": 9.856281407035176e-05,
-      "loss": 6.4011,
-      "step": 1931
-    },
-    {
-      "epoch": 1.0075619295958278,
-      "grad_norm": 1.7834244966506958,
-      "learning_rate": 9.856180904522614e-05,
-      "loss": 6.237,
-      "step": 1932
-    },
-    {
-      "epoch": 1.008083441981747,
-      "grad_norm": 1.662527084350586,
-      "learning_rate": 9.85608040201005e-05,
-      "loss": 6.2767,
-      "step": 1933
-    },
-    {
-      "epoch": 1.0086049543676663,
-      "grad_norm": 1.704943299293518,
-      "learning_rate": 9.855979899497488e-05,
-      "loss": 6.2224,
-      "step": 1934
-    },
-    {
-      "epoch": 1.0091264667535853,
-      "grad_norm": 1.9073781967163086,
-      "learning_rate": 9.855879396984924e-05,
-      "loss": 6.1934,
-      "step": 1935
-    },
-    {
-      "epoch": 1.0096479791395045,
-      "grad_norm": 2.1955313682556152,
-      "learning_rate": 9.855778894472362e-05,
-      "loss": 5.7247,
-      "step": 1936
-    },
-    {
-      "epoch": 1.0101694915254238,
-      "grad_norm": 1.8141535520553589,
-      "learning_rate": 9.8556783919598e-05,
-      "loss": 5.604,
-      "step": 1937
-    },
-    {
-      "epoch": 1.0106910039113428,
-      "grad_norm": 1.701280951499939,
-      "learning_rate": 9.855577889447236e-05,
-      "loss": 6.3704,
-      "step": 1938
-    },
-    {
-      "epoch": 1.011212516297262,
-      "grad_norm": 1.7259966135025024,
-      "learning_rate": 9.855477386934674e-05,
-      "loss": 6.2628,
-      "step": 1939
-    },
-    {
-      "epoch": 1.0117340286831813,
-      "grad_norm": 2.0503947734832764,
-      "learning_rate": 9.855376884422111e-05,
-      "loss": 5.9671,
-      "step": 1940
-    },
-    {
-      "epoch": 1.0122555410691003,
-      "grad_norm": 2.145129919052124,
-      "learning_rate": 9.855276381909548e-05,
-      "loss": 6.0575,
-      "step": 1941
-    },
-    {
-      "epoch": 1.0127770534550196,
-      "grad_norm": 1.6872786283493042,
-      "learning_rate": 9.855175879396985e-05,
-      "loss": 6.2608,
-      "step": 1942
-    },
-    {
-      "epoch": 1.0132985658409388,
-      "grad_norm": 1.89679753780365,
-      "learning_rate": 9.855075376884423e-05,
-      "loss": 6.0626,
-      "step": 1943
-    },
-    {
-      "epoch": 1.0138200782268578,
-      "grad_norm": 1.792041301727295,
-      "learning_rate": 9.854974874371859e-05,
-      "loss": 6.1741,
-      "step": 1944
-    },
-    {
-      "epoch": 1.014341590612777,
-      "grad_norm": 1.738544225692749,
-      "learning_rate": 9.854874371859297e-05,
-      "loss": 5.8878,
-      "step": 1945
-    },
-    {
-      "epoch": 1.0148631029986963,
-      "grad_norm": 1.7884182929992676,
-      "learning_rate": 9.854773869346733e-05,
-      "loss": 6.0474,
-      "step": 1946
-    },
-    {
-      "epoch": 1.0153846153846153,
-      "grad_norm": 1.6626036167144775,
-      "learning_rate": 9.854673366834171e-05,
-      "loss": 6.3142,
-      "step": 1947
-    },
-    {
-      "epoch": 1.0159061277705346,
-      "grad_norm": 1.596867561340332,
-      "learning_rate": 9.854572864321609e-05,
-      "loss": 6.4722,
-      "step": 1948
-    },
-    {
-      "epoch": 1.0164276401564538,
-      "grad_norm": 1.572496771812439,
-      "learning_rate": 9.854472361809047e-05,
-      "loss": 6.4497,
-      "step": 1949
-    },
-    {
-      "epoch": 1.0169491525423728,
-      "grad_norm": 2.112011432647705,
-      "learning_rate": 9.854371859296483e-05,
-      "loss": 5.6047,
-      "step": 1950
-    },
-    {
-      "epoch": 1.017470664928292,
-      "grad_norm": 1.6629061698913574,
-      "learning_rate": 9.85427135678392e-05,
-      "loss": 6.1408,
-      "step": 1951
-    },
-    {
-      "epoch": 1.0179921773142113,
-      "grad_norm": 2.0328640937805176,
-      "learning_rate": 9.854170854271357e-05,
-      "loss": 5.6916,
-      "step": 1952
-    },
-    {
-      "epoch": 1.0185136897001303,
-      "grad_norm": 1.8895622491836548,
-      "learning_rate": 9.854070351758794e-05,
-      "loss": 6.0915,
-      "step": 1953
-    },
-    {
-      "epoch": 1.0190352020860496,
-      "grad_norm": 1.845371127128601,
-      "learning_rate": 9.853969849246232e-05,
-      "loss": 6.0751,
-      "step": 1954
-    },
-    {
-      "epoch": 1.0195567144719686,
-      "grad_norm": 1.6103541851043701,
-      "learning_rate": 9.853869346733668e-05,
-      "loss": 6.2483,
-      "step": 1955
-    },
-    {
-      "epoch": 1.0200782268578878,
-      "grad_norm": 1.5887151956558228,
-      "learning_rate": 9.853768844221106e-05,
-      "loss": 6.1339,
-      "step": 1956
-    },
-    {
-      "epoch": 1.020599739243807,
-      "grad_norm": 1.7419342994689941,
-      "learning_rate": 9.853668341708544e-05,
-      "loss": 5.8927,
-      "step": 1957
-    },
-    {
-      "epoch": 1.0211212516297261,
-      "grad_norm": 1.63241708278656,
-      "learning_rate": 9.853567839195981e-05,
-      "loss": 5.7099,
-      "step": 1958
-    },
-    {
-      "epoch": 1.0216427640156454,
-      "grad_norm": 1.866214632987976,
-      "learning_rate": 9.853467336683418e-05,
-      "loss": 6.2121,
-      "step": 1959
-    },
-    {
-      "epoch": 1.0221642764015646,
-      "grad_norm": 2.0659101009368896,
-      "learning_rate": 9.853366834170856e-05,
-      "loss": 5.9859,
-      "step": 1960
-    },
-    {
-      "epoch": 1.0226857887874836,
-      "grad_norm": 1.8834388256072998,
-      "learning_rate": 9.853266331658292e-05,
-      "loss": 5.9066,
-      "step": 1961
-    },
-    {
-      "epoch": 1.0232073011734029,
-      "grad_norm": 2.102869749069214,
-      "learning_rate": 9.85316582914573e-05,
-      "loss": 6.0997,
-      "step": 1962
-    },
-    {
-      "epoch": 1.023728813559322,
-      "grad_norm": 1.770065188407898,
-      "learning_rate": 9.853065326633166e-05,
-      "loss": 6.3884,
-      "step": 1963
-    },
-    {
-      "epoch": 1.0242503259452411,
-      "grad_norm": 1.7065154314041138,
-      "learning_rate": 9.852964824120603e-05,
-      "loss": 6.0609,
-      "step": 1964
-    },
-    {
-      "epoch": 1.0247718383311604,
-      "grad_norm": 1.9834550619125366,
-      "learning_rate": 9.85286432160804e-05,
-      "loss": 6.1505,
-      "step": 1965
-    },
-    {
-      "epoch": 1.0252933507170796,
-      "grad_norm": 1.844692349433899,
-      "learning_rate": 9.852763819095477e-05,
-      "loss": 5.6711,
-      "step": 1966
-    },
-    {
-      "epoch": 1.0258148631029986,
-      "grad_norm": 1.8565480709075928,
-      "learning_rate": 9.852663316582915e-05,
-      "loss": 5.6849,
-      "step": 1967
-    },
-    {
-      "epoch": 1.0263363754889179,
-      "grad_norm": 2.0983030796051025,
-      "learning_rate": 9.852562814070352e-05,
-      "loss": 5.4211,
-      "step": 1968
-    },
-    {
-      "epoch": 1.0268578878748371,
-      "grad_norm": 1.8495198488235474,
-      "learning_rate": 9.85246231155779e-05,
-      "loss": 5.9789,
-      "step": 1969
-    },
-    {
-      "epoch": 1.0273794002607561,
-      "grad_norm": 2.0760555267333984,
-      "learning_rate": 9.852361809045227e-05,
-      "loss": 6.0637,
-      "step": 1970
-    },
-    {
-      "epoch": 1.0279009126466754,
-      "grad_norm": 1.6629621982574463,
-      "learning_rate": 9.852261306532664e-05,
-      "loss": 5.9882,
-      "step": 1971
-    },
-    {
-      "epoch": 1.0284224250325946,
-      "grad_norm": 1.7916992902755737,
-      "learning_rate": 9.852160804020101e-05,
-      "loss": 6.0562,
-      "step": 1972
-    },
-    {
-      "epoch": 1.0289439374185136,
-      "grad_norm": 1.7200196981430054,
-      "learning_rate": 9.852060301507539e-05,
-      "loss": 6.4633,
-      "step": 1973
-    },
-    {
-      "epoch": 1.0294654498044329,
-      "grad_norm": 1.4872539043426514,
-      "learning_rate": 9.851959798994975e-05,
-      "loss": 6.4146,
-      "step": 1974
-    },
-    {
-      "epoch": 1.0299869621903521,
-      "grad_norm": 1.9057978391647339,
-      "learning_rate": 9.851859296482413e-05,
-      "loss": 5.738,
-      "step": 1975
-    },
-    {
-      "epoch": 1.0305084745762711,
-      "grad_norm": 1.5664172172546387,
-      "learning_rate": 9.851758793969849e-05,
-      "loss": 6.2933,
-      "step": 1976
-    },
-    {
-      "epoch": 1.0310299869621904,
-      "grad_norm": 1.8218095302581787,
-      "learning_rate": 9.851658291457287e-05,
-      "loss": 6.2938,
-      "step": 1977
-    },
-    {
-      "epoch": 1.0315514993481094,
-      "grad_norm": 1.6508524417877197,
-      "learning_rate": 9.851557788944725e-05,
-      "loss": 6.0992,
-      "step": 1978
-    },
-    {
-      "epoch": 1.0320730117340287,
-      "grad_norm": 1.6507478952407837,
-      "learning_rate": 9.851457286432161e-05,
-      "loss": 5.8736,
-      "step": 1979
-    },
-    {
-      "epoch": 1.032594524119948,
-      "grad_norm": 1.7887682914733887,
-      "learning_rate": 9.851356783919599e-05,
-      "loss": 6.2563,
-      "step": 1980
-    },
-    {
-      "epoch": 1.033116036505867,
-      "grad_norm": 1.625353217124939,
-      "learning_rate": 9.851256281407035e-05,
-      "loss": 5.8274,
-      "step": 1981
-    },
-    {
-      "epoch": 1.0336375488917862,
-      "grad_norm": 1.6090606451034546,
-      "learning_rate": 9.851155778894473e-05,
-      "loss": 6.2335,
-      "step": 1982
-    },
-    {
-      "epoch": 1.0341590612777054,
-      "grad_norm": 1.6227985620498657,
-      "learning_rate": 9.85105527638191e-05,
-      "loss": 6.192,
-      "step": 1983
-    },
-    {
-      "epoch": 1.0346805736636244,
-      "grad_norm": 1.8747210502624512,
-      "learning_rate": 9.850954773869347e-05,
-      "loss": 5.7952,
-      "step": 1984
-    },
-    {
-      "epoch": 1.0352020860495437,
-      "grad_norm": 1.5406162738800049,
-      "learning_rate": 9.850854271356784e-05,
-      "loss": 6.3294,
-      "step": 1985
-    },
-    {
-      "epoch": 1.035723598435463,
-      "grad_norm": 1.755257248878479,
-      "learning_rate": 9.850753768844222e-05,
-      "loss": 5.9776,
-      "step": 1986
-    },
-    {
-      "epoch": 1.036245110821382,
-      "grad_norm": 1.7343206405639648,
-      "learning_rate": 9.850653266331658e-05,
-      "loss": 6.2135,
-      "step": 1987
-    },
-    {
-      "epoch": 1.0367666232073012,
-      "grad_norm": 1.4382730722427368,
-      "learning_rate": 9.850552763819096e-05,
-      "loss": 6.3149,
-      "step": 1988
-    },
-    {
-      "epoch": 1.0372881355932204,
-      "grad_norm": 1.5399811267852783,
-      "learning_rate": 9.850452261306534e-05,
-      "loss": 6.2842,
-      "step": 1989
-    },
-    {
-      "epoch": 1.0378096479791394,
-      "grad_norm": 1.7079685926437378,
-      "learning_rate": 9.850351758793971e-05,
-      "loss": 6.1115,
-      "step": 1990
-    },
-    {
-      "epoch": 1.0383311603650587,
-      "grad_norm": 2.1128201484680176,
-      "learning_rate": 9.850251256281408e-05,
-      "loss": 5.342,
-      "step": 1991
-    },
-    {
-      "epoch": 1.038852672750978,
-      "grad_norm": 1.9430609941482544,
-      "learning_rate": 9.850150753768844e-05,
-      "loss": 5.3026,
-      "step": 1992
-    },
-    {
-      "epoch": 1.039374185136897,
-      "grad_norm": 2.039355754852295,
-      "learning_rate": 9.850050251256282e-05,
-      "loss": 5.6604,
-      "step": 1993
-    },
-    {
-      "epoch": 1.0398956975228162,
-      "grad_norm": 1.564090609550476,
-      "learning_rate": 9.849949748743718e-05,
-      "loss": 6.3321,
-      "step": 1994
-    },
-    {
-      "epoch": 1.0404172099087354,
-      "grad_norm": 1.8027979135513306,
-      "learning_rate": 9.849849246231156e-05,
-      "loss": 5.8961,
-      "step": 1995
-    },
-    {
-      "epoch": 1.0409387222946545,
-      "grad_norm": 1.6895747184753418,
-      "learning_rate": 9.849748743718593e-05,
-      "loss": 6.015,
-      "step": 1996
-    },
-    {
-      "epoch": 1.0414602346805737,
-      "grad_norm": 1.9850399494171143,
-      "learning_rate": 9.84964824120603e-05,
-      "loss": 5.5114,
-      "step": 1997
-    },
-    {
-      "epoch": 1.041981747066493,
-      "grad_norm": 2.297036647796631,
-      "learning_rate": 9.849547738693468e-05,
-      "loss": 5.8444,
-      "step": 1998
-    },
-    {
-      "epoch": 1.042503259452412,
-      "grad_norm": 1.6363486051559448,
-      "learning_rate": 9.849447236180906e-05,
-      "loss": 6.3536,
-      "step": 1999
-    },
-    {
-      "epoch": 1.0430247718383312,
-      "grad_norm": 1.6470552682876587,
-      "learning_rate": 9.849346733668342e-05,
-      "loss": 6.5008,
-      "step": 2000
-    },
-    {
-      "epoch": 1.0435462842242502,
-      "grad_norm": 1.8797277212142944,
-      "learning_rate": 9.84924623115578e-05,
-      "loss": 5.9143,
-      "step": 2001
-    },
-    {
-      "epoch": 1.0440677966101695,
-      "grad_norm": 1.6308670043945312,
-      "learning_rate": 9.849145728643217e-05,
-      "loss": 6.1703,
-      "step": 2002
-    },
-    {
-      "epoch": 1.0445893089960887,
-      "grad_norm": 1.7901939153671265,
-      "learning_rate": 9.849045226130654e-05,
-      "loss": 6.0045,
-      "step": 2003
-    },
-    {
-      "epoch": 1.0451108213820077,
-      "grad_norm": 1.9173728227615356,
-      "learning_rate": 9.848944723618091e-05,
-      "loss": 5.8901,
-      "step": 2004
-    },
-    {
-      "epoch": 1.045632333767927,
-      "grad_norm": 2.012906551361084,
-      "learning_rate": 9.848844221105527e-05,
-      "loss": 5.8192,
-      "step": 2005
-    },
-    {
-      "epoch": 1.0461538461538462,
-      "grad_norm": 1.6332141160964966,
-      "learning_rate": 9.848743718592965e-05,
-      "loss": 6.1516,
-      "step": 2006
-    },
-    {
-      "epoch": 1.0466753585397652,
-      "grad_norm": 1.6701242923736572,
-      "learning_rate": 9.848643216080401e-05,
-      "loss": 6.4004,
-      "step": 2007
-    },
-    {
-      "epoch": 1.0471968709256845,
-      "grad_norm": 2.0260472297668457,
-      "learning_rate": 9.848542713567839e-05,
-      "loss": 5.5856,
-      "step": 2008
-    },
-    {
-      "epoch": 1.0477183833116037,
-      "grad_norm": 1.9220818281173706,
-      "learning_rate": 9.848442211055277e-05,
-      "loss": 5.6215,
-      "step": 2009
-    },
-    {
-      "epoch": 1.0482398956975227,
-      "grad_norm": 1.9030778408050537,
-      "learning_rate": 9.848341708542715e-05,
-      "loss": 6.1956,
-      "step": 2010
-    },
-    {
-      "epoch": 1.048761408083442,
-      "grad_norm": 1.7378175258636475,
-      "learning_rate": 9.848241206030151e-05,
-      "loss": 6.4719,
-      "step": 2011
-    },
-    {
-      "epoch": 1.0492829204693612,
-      "grad_norm": 1.8204840421676636,
-      "learning_rate": 9.848140703517589e-05,
-      "loss": 5.8105,
-      "step": 2012
-    },
-    {
-      "epoch": 1.0498044328552802,
-      "grad_norm": 1.8509920835494995,
-      "learning_rate": 9.848040201005025e-05,
-      "loss": 6.1854,
-      "step": 2013
-    },
-    {
-      "epoch": 1.0503259452411995,
-      "grad_norm": 1.8737764358520508,
-      "learning_rate": 9.847939698492463e-05,
-      "loss": 5.6766,
-      "step": 2014
-    },
-    {
-      "epoch": 1.0508474576271187,
-      "grad_norm": 1.9872692823410034,
-      "learning_rate": 9.8478391959799e-05,
-      "loss": 5.9299,
-      "step": 2015
-    },
-    {
-      "epoch": 1.0513689700130378,
-      "grad_norm": 1.6695271730422974,
-      "learning_rate": 9.847738693467337e-05,
-      "loss": 6.208,
-      "step": 2016
-    },
-    {
-      "epoch": 1.051890482398957,
-      "grad_norm": 1.674023151397705,
-      "learning_rate": 9.847638190954774e-05,
-      "loss": 6.0535,
-      "step": 2017
-    },
-    {
-      "epoch": 1.0524119947848762,
-      "grad_norm": 1.775860071182251,
-      "learning_rate": 9.847537688442212e-05,
-      "loss": 5.9867,
-      "step": 2018
-    },
-    {
-      "epoch": 1.0529335071707953,
-      "grad_norm": 1.8346490859985352,
-      "learning_rate": 9.84743718592965e-05,
-      "loss": 6.1779,
-      "step": 2019
-    },
-    {
-      "epoch": 1.0534550195567145,
-      "grad_norm": 1.7631299495697021,
-      "learning_rate": 9.847336683417086e-05,
-      "loss": 5.8033,
-      "step": 2020
-    },
-    {
-      "epoch": 1.0539765319426337,
-      "grad_norm": 1.8419535160064697,
-      "learning_rate": 9.847236180904524e-05,
-      "loss": 6.1076,
-      "step": 2021
-    },
-    {
-      "epoch": 1.0544980443285528,
-      "grad_norm": 1.8955037593841553,
-      "learning_rate": 9.84713567839196e-05,
-      "loss": 6.1954,
-      "step": 2022
-    },
-    {
-      "epoch": 1.055019556714472,
-      "grad_norm": 1.6946598291397095,
-      "learning_rate": 9.847035175879398e-05,
-      "loss": 6.133,
-      "step": 2023
-    },
-    {
-      "epoch": 1.055541069100391,
-      "grad_norm": 1.8254743814468384,
-      "learning_rate": 9.846934673366834e-05,
-      "loss": 5.9449,
-      "step": 2024
-    },
-    {
-      "epoch": 1.0560625814863103,
-      "grad_norm": 1.554898738861084,
-      "learning_rate": 9.846834170854272e-05,
-      "loss": 6.2816,
-      "step": 2025
-    },
-    {
-      "epoch": 1.0565840938722295,
-      "grad_norm": 1.5383116006851196,
-      "learning_rate": 9.846733668341709e-05,
-      "loss": 6.2263,
-      "step": 2026
-    },
-    {
-      "epoch": 1.0571056062581485,
-      "grad_norm": 1.6168657541275024,
-      "learning_rate": 9.846633165829146e-05,
-      "loss": 5.9276,
-      "step": 2027
-    },
-    {
-      "epoch": 1.0576271186440678,
-      "grad_norm": 1.5930066108703613,
-      "learning_rate": 9.846532663316583e-05,
-      "loss": 5.8445,
-      "step": 2028
-    },
-    {
-      "epoch": 1.058148631029987,
-      "grad_norm": 1.5730817317962646,
-      "learning_rate": 9.84643216080402e-05,
-      "loss": 5.9498,
-      "step": 2029
-    },
-    {
-      "epoch": 1.058670143415906,
-      "grad_norm": 1.4855321645736694,
-      "learning_rate": 9.846331658291458e-05,
-      "loss": 6.1839,
-      "step": 2030
-    },
-    {
-      "epoch": 1.0591916558018253,
-      "grad_norm": 1.5094448328018188,
-      "learning_rate": 9.846231155778895e-05,
-      "loss": 6.2985,
-      "step": 2031
-    },
-    {
-      "epoch": 1.0597131681877445,
-      "grad_norm": 1.7278530597686768,
-      "learning_rate": 9.846130653266332e-05,
-      "loss": 6.2531,
-      "step": 2032
-    },
-    {
-      "epoch": 1.0602346805736635,
-      "grad_norm": 1.5366487503051758,
-      "learning_rate": 9.846030150753769e-05,
-      "loss": 6.0652,
-      "step": 2033
-    },
-    {
-      "epoch": 1.0607561929595828,
-      "grad_norm": 1.62075674533844,
-      "learning_rate": 9.845929648241207e-05,
-      "loss": 6.0717,
-      "step": 2034
-    },
-    {
-      "epoch": 1.061277705345502,
-      "grad_norm": 1.766707181930542,
-      "learning_rate": 9.845829145728643e-05,
-      "loss": 6.0662,
-      "step": 2035
-    },
-    {
-      "epoch": 1.061799217731421,
-      "grad_norm": 1.8812587261199951,
-      "learning_rate": 9.845728643216081e-05,
-      "loss": 5.5153,
-      "step": 2036
-    },
-    {
-      "epoch": 1.0623207301173403,
-      "grad_norm": 1.743231177330017,
-      "learning_rate": 9.845628140703517e-05,
-      "loss": 5.9511,
-      "step": 2037
-    },
-    {
-      "epoch": 1.0628422425032595,
-      "grad_norm": 1.5841785669326782,
-      "learning_rate": 9.845527638190955e-05,
-      "loss": 6.2901,
-      "step": 2038
-    },
-    {
-      "epoch": 1.0633637548891786,
-      "grad_norm": 1.7085474729537964,
-      "learning_rate": 9.845427135678393e-05,
-      "loss": 6.2665,
-      "step": 2039
-    },
-    {
-      "epoch": 1.0638852672750978,
-      "grad_norm": 1.6257435083389282,
-      "learning_rate": 9.845326633165831e-05,
-      "loss": 6.323,
-      "step": 2040
-    },
-    {
-      "epoch": 1.064406779661017,
-      "grad_norm": 1.8219819068908691,
-      "learning_rate": 9.845226130653267e-05,
-      "loss": 6.0413,
-      "step": 2041
-    },
-    {
-      "epoch": 1.064928292046936,
-      "grad_norm": 1.6297686100006104,
-      "learning_rate": 9.845125628140705e-05,
-      "loss": 6.3621,
-      "step": 2042
-    },
-    {
-      "epoch": 1.0654498044328553,
-      "grad_norm": 1.575216293334961,
-      "learning_rate": 9.845025125628141e-05,
-      "loss": 6.237,
-      "step": 2043
-    },
-    {
-      "epoch": 1.0659713168187746,
-      "grad_norm": 1.653867244720459,
-      "learning_rate": 9.844924623115578e-05,
-      "loss": 6.2894,
-      "step": 2044
-    },
-    {
-      "epoch": 1.0664928292046936,
-      "grad_norm": 1.923910140991211,
-      "learning_rate": 9.844824120603016e-05,
-      "loss": 5.6098,
-      "step": 2045
-    },
-    {
-      "epoch": 1.0670143415906128,
-      "grad_norm": 1.6443042755126953,
-      "learning_rate": 9.844723618090452e-05,
-      "loss": 6.1054,
-      "step": 2046
-    },
-    {
-      "epoch": 1.067535853976532,
-      "grad_norm": 1.9147251844406128,
-      "learning_rate": 9.84462311557789e-05,
-      "loss": 5.5705,
-      "step": 2047
-    },
-    {
-      "epoch": 1.068057366362451,
-      "grad_norm": 1.740979790687561,
-      "learning_rate": 9.844522613065326e-05,
-      "loss": 6.2621,
-      "step": 2048
-    },
-    {
-      "epoch": 1.0685788787483703,
-      "grad_norm": 2.3516361713409424,
-      "learning_rate": 9.844422110552764e-05,
-      "loss": 6.0933,
-      "step": 2049
-    },
-    {
-      "epoch": 1.0691003911342893,
-      "grad_norm": 1.9072270393371582,
-      "learning_rate": 9.844321608040202e-05,
-      "loss": 6.2036,
-      "step": 2050
-    },
-    {
-      "epoch": 1.0696219035202086,
-      "grad_norm": 1.9166215658187866,
-      "learning_rate": 9.84422110552764e-05,
-      "loss": 5.9267,
-      "step": 2051
-    },
-    {
-      "epoch": 1.0701434159061278,
-      "grad_norm": 1.8280320167541504,
-      "learning_rate": 9.844120603015076e-05,
-      "loss": 5.9086,
-      "step": 2052
-    },
-    {
-      "epoch": 1.0706649282920468,
-      "grad_norm": 1.831407904624939,
-      "learning_rate": 9.844020100502514e-05,
-      "loss": 6.0674,
-      "step": 2053
-    },
-    {
-      "epoch": 1.071186440677966,
-      "grad_norm": 1.6701538562774658,
-      "learning_rate": 9.84391959798995e-05,
-      "loss": 6.2811,
-      "step": 2054
-    },
-    {
-      "epoch": 1.0717079530638853,
-      "grad_norm": 1.583001732826233,
-      "learning_rate": 9.843819095477388e-05,
-      "loss": 6.468,
-      "step": 2055
-    },
-    {
-      "epoch": 1.0722294654498044,
-      "grad_norm": 2.1178581714630127,
-      "learning_rate": 9.843718592964824e-05,
-      "loss": 5.3708,
-      "step": 2056
-    },
-    {
-      "epoch": 1.0727509778357236,
-      "grad_norm": 1.7624542713165283,
-      "learning_rate": 9.843618090452261e-05,
-      "loss": 5.998,
-      "step": 2057
-    },
-    {
-      "epoch": 1.0732724902216428,
-      "grad_norm": 1.7489049434661865,
-      "learning_rate": 9.843517587939699e-05,
-      "loss": 6.3201,
-      "step": 2058
-    },
-    {
-      "epoch": 1.0737940026075619,
-      "grad_norm": 1.6883853673934937,
-      "learning_rate": 9.843417085427136e-05,
-      "loss": 6.5115,
-      "step": 2059
-    },
-    {
-      "epoch": 1.074315514993481,
-      "grad_norm": 1.4730063676834106,
-      "learning_rate": 9.843316582914574e-05,
-      "loss": 6.0829,
-      "step": 2060
-    },
-    {
-      "epoch": 1.0748370273794003,
-      "grad_norm": 1.5787452459335327,
-      "learning_rate": 9.84321608040201e-05,
-      "loss": 6.1248,
-      "step": 2061
-    },
-    {
-      "epoch": 1.0753585397653194,
-      "grad_norm": 1.7233240604400635,
-      "learning_rate": 9.843115577889448e-05,
-      "loss": 6.2,
-      "step": 2062
-    },
-    {
-      "epoch": 1.0758800521512386,
-      "grad_norm": 1.7375123500823975,
-      "learning_rate": 9.843015075376885e-05,
-      "loss": 6.1556,
-      "step": 2063
-    },
-    {
-      "epoch": 1.0764015645371579,
-      "grad_norm": 1.6828275918960571,
-      "learning_rate": 9.842914572864323e-05,
-      "loss": 5.9103,
-      "step": 2064
-    },
-    {
-      "epoch": 1.0769230769230769,
-      "grad_norm": 1.6976250410079956,
-      "learning_rate": 9.842814070351759e-05,
-      "loss": 6.2481,
-      "step": 2065
-    },
-    {
-      "epoch": 1.0774445893089961,
-      "grad_norm": 1.8267648220062256,
-      "learning_rate": 9.842713567839197e-05,
-      "loss": 5.7245,
-      "step": 2066
-    },
-    {
-      "epoch": 1.0779661016949154,
-      "grad_norm": 1.7747575044631958,
-      "learning_rate": 9.842613065326633e-05,
-      "loss": 6.2221,
-      "step": 2067
-    },
-    {
-      "epoch": 1.0784876140808344,
-      "grad_norm": 1.6890941858291626,
-      "learning_rate": 9.842512562814071e-05,
-      "loss": 6.2296,
-      "step": 2068
-    },
-    {
-      "epoch": 1.0790091264667536,
-      "grad_norm": 2.056994676589966,
-      "learning_rate": 9.842412060301507e-05,
-      "loss": 6.1521,
-      "step": 2069
-    },
-    {
-      "epoch": 1.0795306388526726,
-      "grad_norm": 1.6816191673278809,
-      "learning_rate": 9.842311557788945e-05,
-      "loss": 5.9184,
-      "step": 2070
-    },
-    {
-      "epoch": 1.0800521512385919,
-      "grad_norm": 1.544165849685669,
-      "learning_rate": 9.842211055276383e-05,
-      "loss": 6.4803,
-      "step": 2071
-    },
-    {
-      "epoch": 1.0805736636245111,
-      "grad_norm": 1.6133559942245483,
-      "learning_rate": 9.84211055276382e-05,
-      "loss": 5.9038,
-      "step": 2072
-    },
-    {
-      "epoch": 1.0810951760104301,
-      "grad_norm": 1.7555313110351562,
-      "learning_rate": 9.842010050251257e-05,
-      "loss": 6.1311,
-      "step": 2073
-    },
-    {
-      "epoch": 1.0816166883963494,
-      "grad_norm": 1.6976627111434937,
-      "learning_rate": 9.841909547738694e-05,
-      "loss": 5.9387,
-      "step": 2074
-    },
-    {
-      "epoch": 1.0821382007822686,
-      "grad_norm": 1.5983151197433472,
-      "learning_rate": 9.841809045226131e-05,
-      "loss": 6.0487,
-      "step": 2075
-    },
-    {
-      "epoch": 1.0826597131681877,
-      "grad_norm": 1.5447509288787842,
-      "learning_rate": 9.841708542713568e-05,
-      "loss": 6.2851,
-      "step": 2076
-    },
-    {
-      "epoch": 1.083181225554107,
-      "grad_norm": 1.6700820922851562,
-      "learning_rate": 9.841608040201006e-05,
-      "loss": 6.3008,
-      "step": 2077
-    },
-    {
-      "epoch": 1.0837027379400261,
-      "grad_norm": 1.7985095977783203,
-      "learning_rate": 9.841507537688442e-05,
-      "loss": 6.0742,
-      "step": 2078
-    },
-    {
-      "epoch": 1.0842242503259452,
-      "grad_norm": 1.6672126054763794,
-      "learning_rate": 9.84140703517588e-05,
-      "loss": 5.897,
-      "step": 2079
-    },
-    {
-      "epoch": 1.0847457627118644,
-      "grad_norm": 1.5944784879684448,
-      "learning_rate": 9.841306532663316e-05,
-      "loss": 6.1446,
-      "step": 2080
-    },
-    {
-      "epoch": 1.0852672750977836,
-      "grad_norm": 2.028343915939331,
-      "learning_rate": 9.841206030150754e-05,
-      "loss": 5.6221,
-      "step": 2081
-    },
-    {
-      "epoch": 1.0857887874837027,
-      "grad_norm": 2.0334553718566895,
-      "learning_rate": 9.841105527638192e-05,
-      "loss": 5.7647,
-      "step": 2082
-    },
-    {
-      "epoch": 1.086310299869622,
-      "grad_norm": 1.9639347791671753,
-      "learning_rate": 9.84100502512563e-05,
-      "loss": 5.8936,
-      "step": 2083
-    },
-    {
-      "epoch": 1.0868318122555412,
-      "grad_norm": 1.745819330215454,
-      "learning_rate": 9.840904522613066e-05,
-      "loss": 6.4195,
-      "step": 2084
-    },
-    {
-      "epoch": 1.0873533246414602,
-      "grad_norm": 1.7406219244003296,
-      "learning_rate": 9.840804020100502e-05,
-      "loss": 6.1403,
-      "step": 2085
-    },
-    {
-      "epoch": 1.0878748370273794,
-      "grad_norm": 1.6571893692016602,
-      "learning_rate": 9.84070351758794e-05,
-      "loss": 6.3012,
-      "step": 2086
-    },
-    {
-      "epoch": 1.0883963494132987,
-      "grad_norm": 1.6557159423828125,
-      "learning_rate": 9.840603015075377e-05,
-      "loss": 6.3509,
-      "step": 2087
-    },
-    {
-      "epoch": 1.0889178617992177,
-      "grad_norm": 1.6120842695236206,
-      "learning_rate": 9.840502512562814e-05,
-      "loss": 6.4066,
-      "step": 2088
-    },
-    {
-      "epoch": 1.089439374185137,
-      "grad_norm": 1.6781662702560425,
-      "learning_rate": 9.840402010050251e-05,
-      "loss": 6.2072,
-      "step": 2089
-    },
-    {
-      "epoch": 1.0899608865710562,
-      "grad_norm": 1.6951066255569458,
-      "learning_rate": 9.840301507537689e-05,
-      "loss": 6.0058,
-      "step": 2090
-    },
-    {
-      "epoch": 1.0904823989569752,
-      "grad_norm": 1.7157716751098633,
-      "learning_rate": 9.840201005025126e-05,
-      "loss": 5.9738,
-      "step": 2091
-    },
-    {
-      "epoch": 1.0910039113428944,
-      "grad_norm": 1.5716623067855835,
-      "learning_rate": 9.840100502512564e-05,
-      "loss": 6.0578,
-      "step": 2092
-    },
-    {
-      "epoch": 1.0915254237288137,
-      "grad_norm": 1.7220311164855957,
-      "learning_rate": 9.84e-05,
-      "loss": 6.465,
-      "step": 2093
-    },
-    {
-      "epoch": 1.0920469361147327,
-      "grad_norm": 1.594516396522522,
-      "learning_rate": 9.839899497487438e-05,
-      "loss": 6.207,
-      "step": 2094
-    },
-    {
-      "epoch": 1.092568448500652,
-      "grad_norm": 1.5586062669754028,
-      "learning_rate": 9.839798994974875e-05,
-      "loss": 6.2873,
-      "step": 2095
-    },
-    {
-      "epoch": 1.093089960886571,
-      "grad_norm": 1.5276882648468018,
-      "learning_rate": 9.839698492462313e-05,
-      "loss": 6.3951,
-      "step": 2096
-    },
-    {
-      "epoch": 1.0936114732724902,
-      "grad_norm": 1.5596115589141846,
-      "learning_rate": 9.839597989949749e-05,
-      "loss": 6.3728,
-      "step": 2097
-    },
-    {
-      "epoch": 1.0941329856584094,
-      "grad_norm": 1.6365646123886108,
-      "learning_rate": 9.839497487437186e-05,
-      "loss": 6.323,
-      "step": 2098
-    },
-    {
-      "epoch": 1.0946544980443285,
-      "grad_norm": 1.506773591041565,
-      "learning_rate": 9.839396984924623e-05,
-      "loss": 6.0078,
-      "step": 2099
-    },
-    {
-      "epoch": 1.0951760104302477,
-      "grad_norm": 1.5022132396697998,
-      "learning_rate": 9.83929648241206e-05,
-      "loss": 6.2774,
-      "step": 2100
-    },
-    {
-      "epoch": 1.095697522816167,
-      "grad_norm": 1.5496234893798828,
-      "learning_rate": 9.839195979899497e-05,
-      "loss": 6.3331,
-      "step": 2101
-    },
-    {
-      "epoch": 1.096219035202086,
-      "grad_norm": 1.4517372846603394,
-      "learning_rate": 9.839095477386935e-05,
-      "loss": 6.4021,
-      "step": 2102
-    },
-    {
-      "epoch": 1.0967405475880052,
-      "grad_norm": 1.9331247806549072,
-      "learning_rate": 9.838994974874373e-05,
-      "loss": 5.8071,
-      "step": 2103
-    },
-    {
-      "epoch": 1.0972620599739245,
-      "grad_norm": 1.5102336406707764,
-      "learning_rate": 9.83889447236181e-05,
-      "loss": 6.1295,
-      "step": 2104
-    },
-    {
-      "epoch": 1.0977835723598435,
-      "grad_norm": 1.818704605102539,
-      "learning_rate": 9.838793969849247e-05,
-      "loss": 6.1369,
-      "step": 2105
-    },
-    {
-      "epoch": 1.0983050847457627,
-      "grad_norm": 2.0119788646698,
-      "learning_rate": 9.838693467336684e-05,
-      "loss": 6.287,
-      "step": 2106
-    },
-    {
-      "epoch": 1.098826597131682,
-      "grad_norm": 1.5393317937850952,
-      "learning_rate": 9.838592964824121e-05,
-      "loss": 6.3275,
-      "step": 2107
-    },
-    {
-      "epoch": 1.099348109517601,
-      "grad_norm": 1.4994674921035767,
-      "learning_rate": 9.838492462311558e-05,
-      "loss": 6.4564,
-      "step": 2108
-    },
-    {
-      "epoch": 1.0998696219035202,
-      "grad_norm": 1.8602712154388428,
-      "learning_rate": 9.838391959798996e-05,
-      "loss": 6.0075,
-      "step": 2109
-    },
-    {
-      "epoch": 1.1003911342894395,
-      "grad_norm": 1.671539545059204,
-      "learning_rate": 9.838291457286432e-05,
-      "loss": 5.6595,
-      "step": 2110
-    },
-    {
-      "epoch": 1.1009126466753585,
-      "grad_norm": 1.6117165088653564,
-      "learning_rate": 9.83819095477387e-05,
-      "loss": 6.0615,
-      "step": 2111
-    },
-    {
-      "epoch": 1.1014341590612777,
-      "grad_norm": 1.7274749279022217,
-      "learning_rate": 9.838090452261308e-05,
-      "loss": 5.3156,
-      "step": 2112
-    },
-    {
-      "epoch": 1.101955671447197,
-      "grad_norm": 1.6751445531845093,
-      "learning_rate": 9.837989949748744e-05,
-      "loss": 5.8146,
-      "step": 2113
-    },
-    {
-      "epoch": 1.102477183833116,
-      "grad_norm": 1.924877405166626,
-      "learning_rate": 9.837889447236182e-05,
-      "loss": 5.9772,
-      "step": 2114
-    },
-    {
-      "epoch": 1.1029986962190352,
-      "grad_norm": 1.8831536769866943,
-      "learning_rate": 9.837788944723618e-05,
-      "loss": 5.9623,
-      "step": 2115
-    },
-    {
-      "epoch": 1.1035202086049543,
-      "grad_norm": 1.9220843315124512,
-      "learning_rate": 9.837688442211056e-05,
-      "loss": 6.2552,
-      "step": 2116
-    },
-    {
-      "epoch": 1.1040417209908735,
-      "grad_norm": 1.7072629928588867,
-      "learning_rate": 9.837587939698493e-05,
-      "loss": 5.7738,
-      "step": 2117
-    },
-    {
-      "epoch": 1.1045632333767927,
-      "grad_norm": 1.7481555938720703,
-      "learning_rate": 9.83748743718593e-05,
-      "loss": 5.8876,
-      "step": 2118
-    },
-    {
-      "epoch": 1.1050847457627118,
-      "grad_norm": 1.7137829065322876,
-      "learning_rate": 9.837386934673367e-05,
-      "loss": 6.2407,
-      "step": 2119
-    },
-    {
-      "epoch": 1.105606258148631,
-      "grad_norm": 1.6986284255981445,
-      "learning_rate": 9.837286432160805e-05,
-      "loss": 5.7636,
-      "step": 2120
-    },
-    {
-      "epoch": 1.1061277705345502,
-      "grad_norm": 1.6503149271011353,
-      "learning_rate": 9.837185929648241e-05,
-      "loss": 6.4542,
-      "step": 2121
-    },
-    {
-      "epoch": 1.1066492829204693,
-      "grad_norm": 1.5618839263916016,
-      "learning_rate": 9.837085427135679e-05,
-      "loss": 6.0379,
-      "step": 2122
-    },
-    {
-      "epoch": 1.1071707953063885,
-      "grad_norm": 1.5384618043899536,
-      "learning_rate": 9.836984924623117e-05,
-      "loss": 6.4178,
-      "step": 2123
-    },
-    {
-      "epoch": 1.1076923076923078,
-      "grad_norm": 1.634857177734375,
-      "learning_rate": 9.836884422110553e-05,
-      "loss": 6.5396,
-      "step": 2124
-    },
-    {
-      "epoch": 1.1082138200782268,
-      "grad_norm": 1.5410327911376953,
-      "learning_rate": 9.836783919597991e-05,
-      "loss": 6.1464,
-      "step": 2125
-    },
-    {
-      "epoch": 1.108735332464146,
-      "grad_norm": 1.5957677364349365,
-      "learning_rate": 9.836683417085427e-05,
-      "loss": 6.3153,
-      "step": 2126
-    },
-    {
-      "epoch": 1.1092568448500653,
-      "grad_norm": 1.618376612663269,
-      "learning_rate": 9.836582914572865e-05,
-      "loss": 6.134,
-      "step": 2127
-    },
-    {
-      "epoch": 1.1097783572359843,
-      "grad_norm": 1.6335352659225464,
-      "learning_rate": 9.836482412060301e-05,
-      "loss": 6.1493,
-      "step": 2128
-    },
-    {
-      "epoch": 1.1102998696219035,
-      "grad_norm": 1.6532281637191772,
-      "learning_rate": 9.836381909547739e-05,
-      "loss": 6.1335,
-      "step": 2129
-    },
-    {
-      "epoch": 1.1108213820078228,
-      "grad_norm": 1.6897399425506592,
-      "learning_rate": 9.836281407035176e-05,
-      "loss": 6.0623,
-      "step": 2130
-    },
-    {
-      "epoch": 1.1113428943937418,
-      "grad_norm": 1.7886675596237183,
-      "learning_rate": 9.836180904522613e-05,
-      "loss": 5.7708,
-      "step": 2131
-    },
-    {
-      "epoch": 1.111864406779661,
-      "grad_norm": 1.5723931789398193,
-      "learning_rate": 9.836080402010051e-05,
-      "loss": 5.7262,
-      "step": 2132
-    },
-    {
-      "epoch": 1.1123859191655803,
-      "grad_norm": 1.8316543102264404,
-      "learning_rate": 9.835979899497489e-05,
-      "loss": 5.7699,
-      "step": 2133
-    },
-    {
-      "epoch": 1.1129074315514993,
-      "grad_norm": 1.6696882247924805,
-      "learning_rate": 9.835879396984925e-05,
-      "loss": 6.0161,
-      "step": 2134
-    },
-    {
-      "epoch": 1.1134289439374185,
-      "grad_norm": 1.565625786781311,
-      "learning_rate": 9.835778894472363e-05,
-      "loss": 6.2272,
-      "step": 2135
-    },
-    {
-      "epoch": 1.1139504563233378,
-      "grad_norm": 1.6374708414077759,
-      "learning_rate": 9.8356783919598e-05,
-      "loss": 6.0726,
-      "step": 2136
-    },
-    {
-      "epoch": 1.1144719687092568,
-      "grad_norm": 1.6318912506103516,
-      "learning_rate": 9.835577889447236e-05,
-      "loss": 6.0006,
-      "step": 2137
-    },
-    {
-      "epoch": 1.114993481095176,
-      "grad_norm": 1.750747561454773,
-      "learning_rate": 9.835477386934674e-05,
-      "loss": 6.0683,
-      "step": 2138
-    },
-    {
-      "epoch": 1.1155149934810953,
-      "grad_norm": 1.8892841339111328,
-      "learning_rate": 9.83537688442211e-05,
-      "loss": 6.0745,
-      "step": 2139
-    },
-    {
-      "epoch": 1.1160365058670143,
-      "grad_norm": 1.969397783279419,
-      "learning_rate": 9.835276381909548e-05,
-      "loss": 5.6781,
-      "step": 2140
-    },
-    {
-      "epoch": 1.1165580182529335,
-      "grad_norm": 1.6905620098114014,
-      "learning_rate": 9.835175879396984e-05,
-      "loss": 5.5594,
-      "step": 2141
-    },
-    {
-      "epoch": 1.1170795306388526,
-      "grad_norm": 1.5232009887695312,
-      "learning_rate": 9.835075376884422e-05,
-      "loss": 6.4654,
-      "step": 2142
-    },
-    {
-      "epoch": 1.1176010430247718,
-      "grad_norm": 1.677577257156372,
-      "learning_rate": 9.83497487437186e-05,
-      "loss": 5.988,
-      "step": 2143
-    },
-    {
-      "epoch": 1.118122555410691,
-      "grad_norm": 1.5536683797836304,
-      "learning_rate": 9.834874371859298e-05,
-      "loss": 6.2869,
-      "step": 2144
-    },
-    {
-      "epoch": 1.11864406779661,
-      "grad_norm": 1.7479469776153564,
-      "learning_rate": 9.834773869346734e-05,
-      "loss": 6.1856,
-      "step": 2145
-    },
-    {
-      "epoch": 1.1191655801825293,
-      "grad_norm": 1.7326782941818237,
-      "learning_rate": 9.834673366834172e-05,
-      "loss": 6.4232,
-      "step": 2146
-    },
-    {
-      "epoch": 1.1196870925684486,
-      "grad_norm": 1.7394683361053467,
-      "learning_rate": 9.834572864321608e-05,
-      "loss": 6.1438,
-      "step": 2147
-    },
-    {
-      "epoch": 1.1202086049543676,
-      "grad_norm": 1.8489806652069092,
-      "learning_rate": 9.834472361809046e-05,
-      "loss": 5.7594,
-      "step": 2148
-    },
-    {
-      "epoch": 1.1207301173402868,
-      "grad_norm": 1.8415873050689697,
-      "learning_rate": 9.834371859296483e-05,
-      "loss": 6.0798,
-      "step": 2149
-    },
-    {
-      "epoch": 1.121251629726206,
-      "grad_norm": 1.6918288469314575,
-      "learning_rate": 9.834271356783919e-05,
-      "loss": 6.1011,
-      "step": 2150
-    },
-    {
-      "epoch": 1.121773142112125,
-      "grad_norm": 2.09799861907959,
-      "learning_rate": 9.834170854271357e-05,
-      "loss": 6.0128,
-      "step": 2151
-    },
-    {
-      "epoch": 1.1222946544980443,
-      "grad_norm": 1.820911169052124,
-      "learning_rate": 9.834070351758795e-05,
-      "loss": 6.3198,
-      "step": 2152
-    },
-    {
-      "epoch": 1.1228161668839636,
-      "grad_norm": 1.6595430374145508,
-      "learning_rate": 9.833969849246232e-05,
-      "loss": 6.1062,
-      "step": 2153
-    },
-    {
-      "epoch": 1.1233376792698826,
-      "grad_norm": 1.874584436416626,
-      "learning_rate": 9.833869346733669e-05,
-      "loss": 5.7289,
-      "step": 2154
-    },
-    {
-      "epoch": 1.1238591916558018,
-      "grad_norm": 1.949552059173584,
-      "learning_rate": 9.833768844221107e-05,
-      "loss": 5.9235,
-      "step": 2155
-    },
-    {
-      "epoch": 1.124380704041721,
-      "grad_norm": 1.8642287254333496,
-      "learning_rate": 9.833668341708543e-05,
-      "loss": 5.7948,
-      "step": 2156
-    },
-    {
-      "epoch": 1.12490221642764,
-      "grad_norm": 1.6870375871658325,
-      "learning_rate": 9.833567839195981e-05,
-      "loss": 5.8126,
-      "step": 2157
-    },
-    {
-      "epoch": 1.1254237288135593,
-      "grad_norm": 1.5605230331420898,
-      "learning_rate": 9.833467336683417e-05,
-      "loss": 6.2091,
-      "step": 2158
-    },
-    {
-      "epoch": 1.1259452411994784,
-      "grad_norm": 1.9269726276397705,
-      "learning_rate": 9.833366834170855e-05,
-      "loss": 6.1305,
-      "step": 2159
-    },
-    {
-      "epoch": 1.1264667535853976,
-      "grad_norm": 1.8271476030349731,
-      "learning_rate": 9.833266331658291e-05,
-      "loss": 5.6783,
-      "step": 2160
-    },
-    {
-      "epoch": 1.1269882659713168,
-      "grad_norm": 1.9713389873504639,
-      "learning_rate": 9.833165829145729e-05,
-      "loss": 5.9957,
-      "step": 2161
-    },
-    {
-      "epoch": 1.1275097783572359,
-      "grad_norm": 1.7722227573394775,
-      "learning_rate": 9.833065326633166e-05,
-      "loss": 5.904,
-      "step": 2162
-    },
-    {
-      "epoch": 1.1280312907431551,
-      "grad_norm": 1.75780189037323,
-      "learning_rate": 9.832964824120603e-05,
-      "loss": 5.881,
-      "step": 2163
-    },
-    {
-      "epoch": 1.1285528031290744,
-      "grad_norm": 1.7344143390655518,
-      "learning_rate": 9.832864321608041e-05,
-      "loss": 5.6713,
-      "step": 2164
-    },
-    {
-      "epoch": 1.1290743155149934,
-      "grad_norm": 1.8423362970352173,
-      "learning_rate": 9.832763819095478e-05,
-      "loss": 6.0549,
-      "step": 2165
-    },
-    {
-      "epoch": 1.1295958279009126,
-      "grad_norm": 1.6502448320388794,
-      "learning_rate": 9.832663316582915e-05,
-      "loss": 6.2476,
-      "step": 2166
-    },
-    {
-      "epoch": 1.1301173402868319,
-      "grad_norm": 1.7386568784713745,
-      "learning_rate": 9.832562814070352e-05,
-      "loss": 5.9287,
-      "step": 2167
-    },
-    {
-      "epoch": 1.1306388526727509,
-      "grad_norm": 1.5258994102478027,
-      "learning_rate": 9.83246231155779e-05,
-      "loss": 6.2101,
-      "step": 2168
-    },
-    {
-      "epoch": 1.1311603650586701,
-      "grad_norm": 1.5508569478988647,
-      "learning_rate": 9.832361809045226e-05,
-      "loss": 6.3921,
-      "step": 2169
-    },
-    {
-      "epoch": 1.1316818774445894,
-      "grad_norm": 1.5998291969299316,
-      "learning_rate": 9.832261306532664e-05,
-      "loss": 5.9298,
-      "step": 2170
-    },
-    {
-      "epoch": 1.1322033898305084,
-      "grad_norm": 1.7097409963607788,
-      "learning_rate": 9.8321608040201e-05,
-      "loss": 5.8778,
-      "step": 2171
-    },
-    {
-      "epoch": 1.1327249022164276,
-      "grad_norm": 1.778180480003357,
-      "learning_rate": 9.832060301507538e-05,
-      "loss": 5.8236,
-      "step": 2172
-    },
-    {
-      "epoch": 1.1332464146023469,
-      "grad_norm": 1.5483494997024536,
-      "learning_rate": 9.831959798994976e-05,
-      "loss": 6.2395,
-      "step": 2173
-    },
-    {
-      "epoch": 1.133767926988266,
-      "grad_norm": 1.9308470487594604,
-      "learning_rate": 9.831859296482414e-05,
-      "loss": 6.0561,
-      "step": 2174
-    },
-    {
-      "epoch": 1.1342894393741851,
-      "grad_norm": 1.6938871145248413,
-      "learning_rate": 9.83175879396985e-05,
-      "loss": 6.0515,
-      "step": 2175
-    },
-    {
-      "epoch": 1.1348109517601044,
-      "grad_norm": 1.7431280612945557,
-      "learning_rate": 9.831658291457288e-05,
-      "loss": 6.0175,
-      "step": 2176
-    },
-    {
-      "epoch": 1.1353324641460234,
-      "grad_norm": 1.915162444114685,
-      "learning_rate": 9.831557788944724e-05,
-      "loss": 5.7712,
-      "step": 2177
-    },
-    {
-      "epoch": 1.1358539765319426,
-      "grad_norm": 1.7051831483840942,
-      "learning_rate": 9.831457286432161e-05,
-      "loss": 6.3093,
-      "step": 2178
-    },
-    {
-      "epoch": 1.1363754889178619,
-      "grad_norm": 1.555277705192566,
-      "learning_rate": 9.831356783919598e-05,
-      "loss": 6.2157,
-      "step": 2179
-    },
-    {
-      "epoch": 1.136897001303781,
-      "grad_norm": 1.62564218044281,
-      "learning_rate": 9.831256281407035e-05,
-      "loss": 6.2933,
-      "step": 2180
-    },
-    {
-      "epoch": 1.1374185136897002,
-      "grad_norm": 1.742859125137329,
-      "learning_rate": 9.831155778894473e-05,
-      "loss": 5.3673,
-      "step": 2181
-    },
-    {
-      "epoch": 1.1379400260756194,
-      "grad_norm": 1.5772731304168701,
-      "learning_rate": 9.831055276381909e-05,
-      "loss": 6.1485,
-      "step": 2182
-    },
-    {
-      "epoch": 1.1384615384615384,
-      "grad_norm": 1.573353886604309,
-      "learning_rate": 9.830954773869347e-05,
-      "loss": 6.462,
-      "step": 2183
-    },
-    {
-      "epoch": 1.1389830508474577,
-      "grad_norm": 1.5485851764678955,
-      "learning_rate": 9.830854271356785e-05,
-      "loss": 6.2419,
-      "step": 2184
-    },
-    {
-      "epoch": 1.139504563233377,
-      "grad_norm": 1.6806432008743286,
-      "learning_rate": 9.830753768844222e-05,
-      "loss": 5.8934,
-      "step": 2185
-    },
-    {
-      "epoch": 1.140026075619296,
-      "grad_norm": 1.5984195470809937,
-      "learning_rate": 9.830653266331659e-05,
-      "loss": 6.374,
-      "step": 2186
-    },
-    {
-      "epoch": 1.1405475880052152,
-      "grad_norm": 2.130748748779297,
-      "learning_rate": 9.830552763819097e-05,
-      "loss": 5.4443,
-      "step": 2187
-    },
-    {
-      "epoch": 1.1410691003911344,
-      "grad_norm": 1.83894944190979,
-      "learning_rate": 9.830452261306533e-05,
-      "loss": 6.4328,
-      "step": 2188
-    },
-    {
-      "epoch": 1.1415906127770534,
-      "grad_norm": 2.2462737560272217,
-      "learning_rate": 9.830351758793971e-05,
-      "loss": 5.9004,
-      "step": 2189
-    },
-    {
-      "epoch": 1.1421121251629727,
-      "grad_norm": 1.5884034633636475,
-      "learning_rate": 9.830251256281407e-05,
-      "loss": 6.3034,
-      "step": 2190
-    },
-    {
-      "epoch": 1.1426336375488917,
-      "grad_norm": 1.7714293003082275,
-      "learning_rate": 9.830150753768844e-05,
-      "loss": 6.036,
-      "step": 2191
-    },
-    {
-      "epoch": 1.143155149934811,
-      "grad_norm": 1.5858405828475952,
-      "learning_rate": 9.830050251256282e-05,
-      "loss": 6.1191,
-      "step": 2192
-    },
-    {
-      "epoch": 1.1436766623207302,
-      "grad_norm": 1.5486083030700684,
-      "learning_rate": 9.829949748743719e-05,
-      "loss": 6.6528,
-      "step": 2193
-    },
-    {
-      "epoch": 1.1441981747066492,
-      "grad_norm": 1.5546194314956665,
-      "learning_rate": 9.829849246231157e-05,
-      "loss": 6.1703,
-      "step": 2194
-    },
-    {
-      "epoch": 1.1447196870925684,
-      "grad_norm": 1.4341551065444946,
-      "learning_rate": 9.829748743718594e-05,
-      "loss": 6.1345,
-      "step": 2195
-    },
-    {
-      "epoch": 1.1452411994784877,
-      "grad_norm": 1.745427131652832,
-      "learning_rate": 9.829648241206031e-05,
-      "loss": 5.8823,
-      "step": 2196
-    },
-    {
-      "epoch": 1.1457627118644067,
-      "grad_norm": 1.6985986232757568,
-      "learning_rate": 9.829547738693468e-05,
-      "loss": 5.9667,
-      "step": 2197
-    },
-    {
-      "epoch": 1.146284224250326,
-      "grad_norm": 1.8772536516189575,
-      "learning_rate": 9.829447236180906e-05,
-      "loss": 6.0903,
-      "step": 2198
-    },
-    {
-      "epoch": 1.1468057366362452,
-      "grad_norm": 1.6111915111541748,
-      "learning_rate": 9.829346733668342e-05,
-      "loss": 6.3498,
-      "step": 2199
-    },
-    {
-      "epoch": 1.1473272490221642,
-      "grad_norm": 1.5966535806655884,
-      "learning_rate": 9.82924623115578e-05,
-      "loss": 6.2569,
-      "step": 2200
-    },
-    {
-      "epoch": 1.1478487614080835,
-      "grad_norm": 1.5252742767333984,
-      "learning_rate": 9.829145728643216e-05,
-      "loss": 6.2356,
-      "step": 2201
-    },
-    {
-      "epoch": 1.1483702737940027,
-      "grad_norm": 1.8468446731567383,
-      "learning_rate": 9.829045226130654e-05,
-      "loss": 6.1507,
-      "step": 2202
-    },
-    {
-      "epoch": 1.1488917861799217,
-      "grad_norm": 1.9033254384994507,
-      "learning_rate": 9.82894472361809e-05,
-      "loss": 6.0196,
-      "step": 2203
-    },
-    {
-      "epoch": 1.149413298565841,
-      "grad_norm": 1.7099493741989136,
-      "learning_rate": 9.828844221105528e-05,
-      "loss": 6.2645,
-      "step": 2204
-    },
-    {
-      "epoch": 1.1499348109517602,
-      "grad_norm": 1.6157017946243286,
-      "learning_rate": 9.828743718592966e-05,
-      "loss": 6.0945,
-      "step": 2205
-    },
-    {
-      "epoch": 1.1504563233376792,
-      "grad_norm": 1.555907130241394,
-      "learning_rate": 9.828643216080402e-05,
-      "loss": 6.2589,
-      "step": 2206
-    },
-    {
-      "epoch": 1.1509778357235985,
-      "grad_norm": 1.6805888414382935,
-      "learning_rate": 9.82854271356784e-05,
-      "loss": 6.2222,
-      "step": 2207
-    },
-    {
-      "epoch": 1.1514993481095175,
-      "grad_norm": 1.697525978088379,
-      "learning_rate": 9.828442211055277e-05,
-      "loss": 5.9605,
-      "step": 2208
-    },
-    {
-      "epoch": 1.1520208604954367,
-      "grad_norm": 1.5336872339248657,
-      "learning_rate": 9.828341708542714e-05,
-      "loss": 6.3101,
-      "step": 2209
-    },
-    {
-      "epoch": 1.152542372881356,
-      "grad_norm": 1.678194284439087,
-      "learning_rate": 9.828241206030151e-05,
-      "loss": 6.0563,
-      "step": 2210
-    },
-    {
-      "epoch": 1.153063885267275,
-      "grad_norm": 1.6272337436676025,
-      "learning_rate": 9.828140703517589e-05,
-      "loss": 6.054,
-      "step": 2211
-    },
-    {
-      "epoch": 1.1535853976531942,
-      "grad_norm": 1.6122015714645386,
-      "learning_rate": 9.828040201005025e-05,
-      "loss": 6.2147,
-      "step": 2212
-    },
-    {
-      "epoch": 1.1541069100391135,
-      "grad_norm": 1.390834093093872,
-      "learning_rate": 9.827939698492463e-05,
-      "loss": 6.5119,
-      "step": 2213
-    },
-    {
-      "epoch": 1.1546284224250325,
-      "grad_norm": 2.453835964202881,
-      "learning_rate": 9.8278391959799e-05,
-      "loss": 5.8751,
-      "step": 2214
-    },
-    {
-      "epoch": 1.1551499348109517,
-      "grad_norm": 1.8139581680297852,
-      "learning_rate": 9.827738693467338e-05,
-      "loss": 6.4075,
-      "step": 2215
-    },
-    {
-      "epoch": 1.155671447196871,
-      "grad_norm": 1.6349258422851562,
-      "learning_rate": 9.827638190954775e-05,
-      "loss": 6.074,
-      "step": 2216
-    },
-    {
-      "epoch": 1.15619295958279,
-      "grad_norm": 1.4726719856262207,
-      "learning_rate": 9.827537688442211e-05,
-      "loss": 6.5689,
-      "step": 2217
-    },
-    {
-      "epoch": 1.1567144719687092,
-      "grad_norm": 1.6071897745132446,
-      "learning_rate": 9.827437185929649e-05,
-      "loss": 6.4558,
-      "step": 2218
-    },
-    {
-      "epoch": 1.1572359843546285,
-      "grad_norm": 1.5450910329818726,
-      "learning_rate": 9.827336683417085e-05,
-      "loss": 6.2967,
-      "step": 2219
-    },
-    {
-      "epoch": 1.1577574967405475,
-      "grad_norm": 1.5636141300201416,
-      "learning_rate": 9.827236180904523e-05,
-      "loss": 6.094,
-      "step": 2220
-    },
-    {
-      "epoch": 1.1582790091264668,
-      "grad_norm": 1.7181307077407837,
-      "learning_rate": 9.82713567839196e-05,
-      "loss": 5.8776,
-      "step": 2221
-    },
-    {
-      "epoch": 1.158800521512386,
-      "grad_norm": 1.8026118278503418,
-      "learning_rate": 9.827035175879397e-05,
-      "loss": 5.842,
-      "step": 2222
-    },
-    {
-      "epoch": 1.159322033898305,
-      "grad_norm": 1.553421974182129,
-      "learning_rate": 9.826934673366834e-05,
-      "loss": 6.4097,
-      "step": 2223
-    },
-    {
-      "epoch": 1.1598435462842243,
-      "grad_norm": 1.5685358047485352,
-      "learning_rate": 9.826834170854272e-05,
-      "loss": 6.1269,
-      "step": 2224
-    },
-    {
-      "epoch": 1.1603650586701435,
-      "grad_norm": 1.5738389492034912,
-      "learning_rate": 9.82673366834171e-05,
-      "loss": 6.2047,
-      "step": 2225
-    },
-    {
-      "epoch": 1.1608865710560625,
-      "grad_norm": 1.6662055253982544,
-      "learning_rate": 9.826633165829147e-05,
-      "loss": 5.367,
-      "step": 2226
-    },
-    {
-      "epoch": 1.1614080834419818,
-      "grad_norm": 1.5968204736709595,
-      "learning_rate": 9.826532663316584e-05,
-      "loss": 6.1703,
-      "step": 2227
-    },
-    {
-      "epoch": 1.161929595827901,
-      "grad_norm": 1.5658279657363892,
-      "learning_rate": 9.826432160804021e-05,
-      "loss": 6.1592,
-      "step": 2228
-    },
-    {
-      "epoch": 1.16245110821382,
-      "grad_norm": 1.8439615964889526,
-      "learning_rate": 9.826331658291458e-05,
-      "loss": 5.7523,
-      "step": 2229
-    },
-    {
-      "epoch": 1.1629726205997393,
-      "grad_norm": 1.8611191511154175,
-      "learning_rate": 9.826231155778894e-05,
-      "loss": 6.0335,
-      "step": 2230
-    },
-    {
-      "epoch": 1.1634941329856585,
-      "grad_norm": 1.8219220638275146,
-      "learning_rate": 9.826130653266332e-05,
-      "loss": 5.6515,
-      "step": 2231
-    },
-    {
-      "epoch": 1.1640156453715775,
-      "grad_norm": 1.6783543825149536,
-      "learning_rate": 9.826030150753768e-05,
-      "loss": 6.0214,
-      "step": 2232
-    },
-    {
-      "epoch": 1.1645371577574968,
-      "grad_norm": 1.8528460264205933,
-      "learning_rate": 9.825929648241206e-05,
-      "loss": 5.8591,
-      "step": 2233
-    },
-    {
-      "epoch": 1.165058670143416,
-      "grad_norm": 1.5973658561706543,
-      "learning_rate": 9.825829145728643e-05,
-      "loss": 6.1476,
-      "step": 2234
-    },
-    {
-      "epoch": 1.165580182529335,
-      "grad_norm": 1.5532174110412598,
-      "learning_rate": 9.82572864321608e-05,
-      "loss": 5.6784,
-      "step": 2235
-    },
-    {
-      "epoch": 1.1661016949152543,
-      "grad_norm": 1.471919059753418,
-      "learning_rate": 9.825628140703518e-05,
-      "loss": 6.4345,
-      "step": 2236
-    },
-    {
-      "epoch": 1.1666232073011735,
-      "grad_norm": 1.7118595838546753,
-      "learning_rate": 9.825527638190956e-05,
-      "loss": 5.7913,
-      "step": 2237
-    },
-    {
-      "epoch": 1.1671447196870925,
-      "grad_norm": 1.6788328886032104,
-      "learning_rate": 9.825427135678392e-05,
-      "loss": 5.7096,
-      "step": 2238
-    },
-    {
-      "epoch": 1.1676662320730118,
-      "grad_norm": 1.8063212633132935,
-      "learning_rate": 9.82532663316583e-05,
-      "loss": 5.7509,
-      "step": 2239
-    },
-    {
-      "epoch": 1.1681877444589308,
-      "grad_norm": 1.606229305267334,
-      "learning_rate": 9.825226130653267e-05,
-      "loss": 6.0547,
-      "step": 2240
-    },
-    {
-      "epoch": 1.16870925684485,
-      "grad_norm": 1.6397027969360352,
-      "learning_rate": 9.825125628140704e-05,
-      "loss": 6.3072,
-      "step": 2241
-    },
-    {
-      "epoch": 1.1692307692307693,
-      "grad_norm": 1.6101173162460327,
-      "learning_rate": 9.825025125628141e-05,
-      "loss": 5.8815,
-      "step": 2242
-    },
-    {
-      "epoch": 1.1697522816166883,
-      "grad_norm": 1.7230368852615356,
-      "learning_rate": 9.824924623115577e-05,
-      "loss": 6.0834,
-      "step": 2243
-    },
-    {
-      "epoch": 1.1702737940026076,
-      "grad_norm": 1.6329448223114014,
-      "learning_rate": 9.824824120603015e-05,
-      "loss": 6.1478,
-      "step": 2244
-    },
-    {
-      "epoch": 1.1707953063885268,
-      "grad_norm": 1.643519401550293,
-      "learning_rate": 9.824723618090453e-05,
-      "loss": 6.2166,
-      "step": 2245
-    },
-    {
-      "epoch": 1.1713168187744458,
-      "grad_norm": 1.6363450288772583,
-      "learning_rate": 9.82462311557789e-05,
-      "loss": 6.4417,
-      "step": 2246
-    },
-    {
-      "epoch": 1.171838331160365,
-      "grad_norm": 1.8582649230957031,
-      "learning_rate": 9.824522613065327e-05,
-      "loss": 5.889,
-      "step": 2247
-    },
-    {
-      "epoch": 1.1723598435462843,
-      "grad_norm": 1.55232834815979,
-      "learning_rate": 9.824422110552765e-05,
-      "loss": 6.1526,
-      "step": 2248
-    },
-    {
-      "epoch": 1.1728813559322033,
-      "grad_norm": 1.8015625476837158,
-      "learning_rate": 9.824321608040201e-05,
-      "loss": 6.0543,
-      "step": 2249
-    },
-    {
-      "epoch": 1.1734028683181226,
-      "grad_norm": 1.7950050830841064,
-      "learning_rate": 9.824221105527639e-05,
-      "loss": 6.2373,
-      "step": 2250
-    },
-    {
-      "epoch": 1.1739243807040418,
-      "grad_norm": 1.7826024293899536,
-      "learning_rate": 9.824120603015075e-05,
-      "loss": 6.3744,
-      "step": 2251
-    },
-    {
-      "epoch": 1.1744458930899608,
-      "grad_norm": 1.687927484512329,
-      "learning_rate": 9.824020100502513e-05,
-      "loss": 5.8684,
-      "step": 2252
-    },
-    {
-      "epoch": 1.17496740547588,
-      "grad_norm": 2.5282304286956787,
-      "learning_rate": 9.82391959798995e-05,
-      "loss": 5.8711,
-      "step": 2253
-    },
-    {
-      "epoch": 1.175488917861799,
-      "grad_norm": 1.7576898336410522,
-      "learning_rate": 9.823819095477387e-05,
-      "loss": 6.3262,
-      "step": 2254
-    },
-    {
-      "epoch": 1.1760104302477183,
-      "grad_norm": 1.9101850986480713,
-      "learning_rate": 9.823718592964824e-05,
-      "loss": 6.0912,
-      "step": 2255
-    },
-    {
-      "epoch": 1.1765319426336376,
-      "grad_norm": 1.9231549501419067,
-      "learning_rate": 9.823618090452262e-05,
-      "loss": 5.0419,
-      "step": 2256
-    },
-    {
-      "epoch": 1.1770534550195566,
-      "grad_norm": 1.8675200939178467,
-      "learning_rate": 9.8235175879397e-05,
-      "loss": 6.3504,
-      "step": 2257
-    },
-    {
-      "epoch": 1.1775749674054758,
-      "grad_norm": 1.917932152748108,
-      "learning_rate": 9.823417085427136e-05,
-      "loss": 5.8987,
-      "step": 2258
-    },
-    {
-      "epoch": 1.178096479791395,
-      "grad_norm": 1.615600824356079,
-      "learning_rate": 9.823316582914574e-05,
-      "loss": 6.1797,
-      "step": 2259
-    },
-    {
-      "epoch": 1.1786179921773141,
-      "grad_norm": 1.7991294860839844,
-      "learning_rate": 9.82321608040201e-05,
-      "loss": 6.0582,
-      "step": 2260
-    },
-    {
-      "epoch": 1.1791395045632334,
-      "grad_norm": 1.6269344091415405,
-      "learning_rate": 9.823115577889448e-05,
-      "loss": 5.973,
-      "step": 2261
-    },
-    {
-      "epoch": 1.1796610169491526,
-      "grad_norm": 1.6376138925552368,
-      "learning_rate": 9.823015075376884e-05,
-      "loss": 6.1668,
-      "step": 2262
-    },
-    {
-      "epoch": 1.1801825293350716,
-      "grad_norm": 2.362882614135742,
-      "learning_rate": 9.822914572864322e-05,
-      "loss": 5.8695,
-      "step": 2263
-    },
-    {
-      "epoch": 1.1807040417209909,
-      "grad_norm": 2.2332370281219482,
-      "learning_rate": 9.822814070351759e-05,
-      "loss": 5.3695,
-      "step": 2264
-    },
-    {
-      "epoch": 1.18122555410691,
-      "grad_norm": 2.100804090499878,
-      "learning_rate": 9.822713567839196e-05,
-      "loss": 5.5639,
-      "step": 2265
-    },
-    {
-      "epoch": 1.1817470664928291,
-      "grad_norm": 1.9866795539855957,
-      "learning_rate": 9.822613065326634e-05,
-      "loss": 5.975,
-      "step": 2266
-    },
-    {
-      "epoch": 1.1822685788787484,
-      "grad_norm": 1.7337366342544556,
-      "learning_rate": 9.822512562814072e-05,
-      "loss": 5.9724,
-      "step": 2267
-    },
-    {
-      "epoch": 1.1827900912646676,
-      "grad_norm": 1.8712918758392334,
-      "learning_rate": 9.822412060301508e-05,
-      "loss": 5.5507,
-      "step": 2268
-    },
-    {
-      "epoch": 1.1833116036505866,
-      "grad_norm": 1.8291009664535522,
-      "learning_rate": 9.822311557788946e-05,
-      "loss": 5.9644,
-      "step": 2269
-    },
-    {
-      "epoch": 1.1838331160365059,
-      "grad_norm": 1.9850845336914062,
-      "learning_rate": 9.822211055276383e-05,
-      "loss": 6.1363,
-      "step": 2270
-    },
-    {
-      "epoch": 1.1843546284224251,
-      "grad_norm": 1.5565494298934937,
-      "learning_rate": 9.822110552763819e-05,
-      "loss": 6.2219,
-      "step": 2271
-    },
-    {
-      "epoch": 1.1848761408083441,
-      "grad_norm": 1.5349458456039429,
-      "learning_rate": 9.822010050251257e-05,
-      "loss": 6.291,
-      "step": 2272
-    },
-    {
-      "epoch": 1.1853976531942634,
-      "grad_norm": 1.640566110610962,
-      "learning_rate": 9.821909547738693e-05,
-      "loss": 6.1211,
-      "step": 2273
-    },
-    {
-      "epoch": 1.1859191655801826,
-      "grad_norm": 1.5279864072799683,
-      "learning_rate": 9.821809045226131e-05,
-      "loss": 6.3587,
-      "step": 2274
-    },
-    {
-      "epoch": 1.1864406779661016,
-      "grad_norm": 1.7557909488677979,
-      "learning_rate": 9.821708542713567e-05,
-      "loss": 5.8717,
-      "step": 2275
-    },
-    {
-      "epoch": 1.1869621903520209,
-      "grad_norm": 1.6266311407089233,
-      "learning_rate": 9.821608040201005e-05,
-      "loss": 6.334,
-      "step": 2276
-    },
-    {
-      "epoch": 1.1874837027379401,
-      "grad_norm": 2.151165246963501,
-      "learning_rate": 9.821507537688443e-05,
-      "loss": 6.0189,
-      "step": 2277
-    },
-    {
-      "epoch": 1.1880052151238591,
-      "grad_norm": 1.840430736541748,
-      "learning_rate": 9.821407035175881e-05,
-      "loss": 6.2953,
-      "step": 2278
-    },
-    {
-      "epoch": 1.1885267275097784,
-      "grad_norm": 1.5451712608337402,
-      "learning_rate": 9.821306532663317e-05,
-      "loss": 6.0759,
-      "step": 2279
-    },
-    {
-      "epoch": 1.1890482398956976,
-      "grad_norm": 1.6075654029846191,
-      "learning_rate": 9.821206030150755e-05,
-      "loss": 5.9847,
-      "step": 2280
-    },
-    {
-      "epoch": 1.1895697522816167,
-      "grad_norm": 1.6984021663665771,
-      "learning_rate": 9.821105527638191e-05,
-      "loss": 6.363,
-      "step": 2281
-    },
-    {
-      "epoch": 1.190091264667536,
-      "grad_norm": 1.7678803205490112,
-      "learning_rate": 9.821005025125629e-05,
-      "loss": 5.2538,
-      "step": 2282
-    },
-    {
-      "epoch": 1.1906127770534551,
-      "grad_norm": 1.713875412940979,
-      "learning_rate": 9.820904522613066e-05,
-      "loss": 5.8226,
-      "step": 2283
-    },
-    {
-      "epoch": 1.1911342894393742,
-      "grad_norm": 1.529335379600525,
-      "learning_rate": 9.820804020100502e-05,
-      "loss": 6.3554,
-      "step": 2284
-    },
-    {
-      "epoch": 1.1916558018252934,
-      "grad_norm": 1.929689884185791,
-      "learning_rate": 9.82070351758794e-05,
-      "loss": 5.81,
-      "step": 2285
-    },
-    {
-      "epoch": 1.1921773142112124,
-      "grad_norm": 1.424947738647461,
-      "learning_rate": 9.820603015075378e-05,
-      "loss": 6.2951,
-      "step": 2286
-    },
-    {
-      "epoch": 1.1926988265971317,
-      "grad_norm": 1.5082155466079712,
-      "learning_rate": 9.820502512562815e-05,
-      "loss": 6.3561,
-      "step": 2287
-    },
-    {
-      "epoch": 1.193220338983051,
-      "grad_norm": 1.5439660549163818,
-      "learning_rate": 9.820402010050252e-05,
-      "loss": 6.1057,
-      "step": 2288
-    },
-    {
-      "epoch": 1.19374185136897,
-      "grad_norm": 1.7151106595993042,
-      "learning_rate": 9.82030150753769e-05,
-      "loss": 6.0657,
-      "step": 2289
-    },
-    {
-      "epoch": 1.1942633637548892,
-      "grad_norm": 1.5136423110961914,
-      "learning_rate": 9.820201005025126e-05,
-      "loss": 6.3777,
-      "step": 2290
-    },
-    {
-      "epoch": 1.1947848761408084,
-      "grad_norm": 1.7190476655960083,
-      "learning_rate": 9.820100502512564e-05,
-      "loss": 6.0312,
-      "step": 2291
-    },
-    {
-      "epoch": 1.1953063885267274,
-      "grad_norm": 1.7018376588821411,
-      "learning_rate": 9.82e-05,
-      "loss": 5.7159,
-      "step": 2292
-    },
-    {
-      "epoch": 1.1958279009126467,
-      "grad_norm": 1.499851942062378,
-      "learning_rate": 9.819899497487438e-05,
-      "loss": 6.5096,
-      "step": 2293
-    },
-    {
-      "epoch": 1.196349413298566,
-      "grad_norm": 2.2462544441223145,
-      "learning_rate": 9.819798994974874e-05,
-      "loss": 5.727,
-      "step": 2294
-    },
-    {
-      "epoch": 1.196870925684485,
-      "grad_norm": 1.4919116497039795,
-      "learning_rate": 9.819698492462312e-05,
-      "loss": 6.4189,
-      "step": 2295
-    },
-    {
-      "epoch": 1.1973924380704042,
-      "grad_norm": 1.8870748281478882,
-      "learning_rate": 9.819597989949749e-05,
-      "loss": 5.6393,
-      "step": 2296
-    },
-    {
-      "epoch": 1.1979139504563234,
-      "grad_norm": 2.049379587173462,
-      "learning_rate": 9.819497487437186e-05,
-      "loss": 5.533,
-      "step": 2297
-    },
-    {
-      "epoch": 1.1984354628422424,
-      "grad_norm": 1.7870759963989258,
-      "learning_rate": 9.819396984924624e-05,
-      "loss": 5.9648,
-      "step": 2298
-    },
-    {
-      "epoch": 1.1989569752281617,
-      "grad_norm": 2.0743887424468994,
-      "learning_rate": 9.81929648241206e-05,
-      "loss": 6.079,
-      "step": 2299
-    },
-    {
-      "epoch": 1.1994784876140807,
-      "grad_norm": 1.709718108177185,
-      "learning_rate": 9.819195979899498e-05,
-      "loss": 5.927,
-      "step": 2300
-    },
-    {
-      "epoch": 1.2,
-      "grad_norm": 1.67744779586792,
-      "learning_rate": 9.819095477386935e-05,
-      "loss": 5.7817,
-      "step": 2301
-    },
-    {
-      "epoch": 1.2005215123859192,
-      "grad_norm": 1.8184869289398193,
-      "learning_rate": 9.818994974874373e-05,
-      "loss": 5.8901,
-      "step": 2302
-    },
-    {
-      "epoch": 1.2010430247718382,
-      "grad_norm": 1.5468182563781738,
-      "learning_rate": 9.818894472361809e-05,
-      "loss": 5.96,
-      "step": 2303
-    },
-    {
-      "epoch": 1.2015645371577575,
-      "grad_norm": 2.268282651901245,
-      "learning_rate": 9.818793969849247e-05,
-      "loss": 5.69,
-      "step": 2304
-    },
-    {
-      "epoch": 1.2020860495436767,
-      "grad_norm": 1.4362901449203491,
-      "learning_rate": 9.818693467336683e-05,
-      "loss": 6.3722,
-      "step": 2305
-    },
-    {
-      "epoch": 1.2026075619295957,
-      "grad_norm": 1.658185601234436,
-      "learning_rate": 9.818592964824121e-05,
-      "loss": 5.6416,
-      "step": 2306
-    },
-    {
-      "epoch": 1.203129074315515,
-      "grad_norm": 1.4815093278884888,
-      "learning_rate": 9.818492462311559e-05,
-      "loss": 6.0633,
-      "step": 2307
-    },
-    {
-      "epoch": 1.2036505867014342,
-      "grad_norm": 1.7652782201766968,
-      "learning_rate": 9.818391959798997e-05,
-      "loss": 6.3984,
-      "step": 2308
-    },
-    {
-      "epoch": 1.2041720990873532,
-      "grad_norm": 1.8604273796081543,
-      "learning_rate": 9.818291457286433e-05,
-      "loss": 5.2822,
-      "step": 2309
-    },
-    {
-      "epoch": 1.2046936114732725,
-      "grad_norm": 1.6468194723129272,
-      "learning_rate": 9.81819095477387e-05,
-      "loss": 5.9139,
-      "step": 2310
-    },
-    {
-      "epoch": 1.2052151238591917,
-      "grad_norm": 1.622148036956787,
-      "learning_rate": 9.818090452261307e-05,
-      "loss": 6.2985,
-      "step": 2311
-    },
-    {
-      "epoch": 1.2057366362451107,
-      "grad_norm": 1.872686743736267,
-      "learning_rate": 9.817989949748744e-05,
-      "loss": 6.0084,
-      "step": 2312
-    },
-    {
-      "epoch": 1.20625814863103,
-      "grad_norm": 1.625709891319275,
-      "learning_rate": 9.817889447236181e-05,
-      "loss": 6.1685,
-      "step": 2313
-    },
-    {
-      "epoch": 1.2067796610169492,
-      "grad_norm": 1.7757817506790161,
-      "learning_rate": 9.817788944723618e-05,
-      "loss": 5.9205,
-      "step": 2314
-    },
-    {
-      "epoch": 1.2073011734028682,
-      "grad_norm": 1.7141445875167847,
-      "learning_rate": 9.817688442211056e-05,
-      "loss": 5.8054,
-      "step": 2315
-    },
-    {
-      "epoch": 1.2078226857887875,
-      "grad_norm": 1.5678503513336182,
-      "learning_rate": 9.817587939698492e-05,
-      "loss": 6.079,
-      "step": 2316
-    },
-    {
-      "epoch": 1.2083441981747067,
-      "grad_norm": 1.7246036529541016,
-      "learning_rate": 9.81748743718593e-05,
-      "loss": 6.3806,
-      "step": 2317
-    },
-    {
-      "epoch": 1.2088657105606258,
-      "grad_norm": 2.032944440841675,
-      "learning_rate": 9.817386934673368e-05,
-      "loss": 6.0508,
-      "step": 2318
-    },
-    {
-      "epoch": 1.209387222946545,
-      "grad_norm": 1.5565643310546875,
-      "learning_rate": 9.817286432160805e-05,
-      "loss": 6.3019,
-      "step": 2319
-    },
-    {
-      "epoch": 1.2099087353324642,
-      "grad_norm": 2.0052428245544434,
-      "learning_rate": 9.817185929648242e-05,
-      "loss": 6.4018,
-      "step": 2320
-    },
-    {
-      "epoch": 1.2104302477183833,
-      "grad_norm": 1.7622815370559692,
-      "learning_rate": 9.81708542713568e-05,
-      "loss": 5.9461,
-      "step": 2321
-    },
-    {
-      "epoch": 1.2109517601043025,
-      "grad_norm": 1.7361843585968018,
-      "learning_rate": 9.816984924623116e-05,
-      "loss": 6.2536,
-      "step": 2322
-    },
-    {
-      "epoch": 1.2114732724902217,
-      "grad_norm": 1.6813410520553589,
-      "learning_rate": 9.816884422110552e-05,
-      "loss": 5.9063,
-      "step": 2323
-    },
-    {
-      "epoch": 1.2119947848761408,
-      "grad_norm": 1.498180866241455,
-      "learning_rate": 9.81678391959799e-05,
-      "loss": 6.2711,
-      "step": 2324
-    },
-    {
-      "epoch": 1.21251629726206,
-      "grad_norm": 1.8266832828521729,
-      "learning_rate": 9.816683417085427e-05,
-      "loss": 5.0716,
-      "step": 2325
-    },
-    {
-      "epoch": 1.2130378096479792,
-      "grad_norm": 1.962540864944458,
-      "learning_rate": 9.816582914572864e-05,
-      "loss": 6.0388,
-      "step": 2326
-    },
-    {
-      "epoch": 1.2135593220338983,
-      "grad_norm": 2.35485577583313,
-      "learning_rate": 9.816482412060302e-05,
-      "loss": 5.7462,
-      "step": 2327
-    },
-    {
-      "epoch": 1.2140808344198175,
-      "grad_norm": 2.106419801712036,
-      "learning_rate": 9.81638190954774e-05,
-      "loss": 5.8976,
-      "step": 2328
-    },
-    {
-      "epoch": 1.2146023468057368,
-      "grad_norm": 1.767941951751709,
-      "learning_rate": 9.816281407035176e-05,
-      "loss": 5.8703,
-      "step": 2329
-    },
-    {
-      "epoch": 1.2151238591916558,
-      "grad_norm": 1.8493388891220093,
-      "learning_rate": 9.816180904522614e-05,
-      "loss": 5.8767,
-      "step": 2330
-    },
-    {
-      "epoch": 1.215645371577575,
-      "grad_norm": 1.4994268417358398,
-      "learning_rate": 9.81608040201005e-05,
-      "loss": 6.0914,
-      "step": 2331
-    },
-    {
-      "epoch": 1.216166883963494,
-      "grad_norm": 1.5960980653762817,
-      "learning_rate": 9.815979899497488e-05,
-      "loss": 5.801,
-      "step": 2332
-    },
-    {
-      "epoch": 1.2166883963494133,
-      "grad_norm": 1.6593005657196045,
-      "learning_rate": 9.815879396984925e-05,
-      "loss": 6.13,
-      "step": 2333
-    },
-    {
-      "epoch": 1.2172099087353325,
-      "grad_norm": 1.76396906375885,
-      "learning_rate": 9.815778894472363e-05,
-      "loss": 6.12,
-      "step": 2334
-    },
-    {
-      "epoch": 1.2177314211212515,
-      "grad_norm": 1.8069300651550293,
-      "learning_rate": 9.815678391959799e-05,
-      "loss": 5.8941,
-      "step": 2335
-    },
-    {
-      "epoch": 1.2182529335071708,
-      "grad_norm": 1.8025990724563599,
-      "learning_rate": 9.815577889447236e-05,
-      "loss": 5.9834,
-      "step": 2336
-    },
-    {
-      "epoch": 1.21877444589309,
-      "grad_norm": 1.6424545049667358,
-      "learning_rate": 9.815477386934673e-05,
-      "loss": 5.8012,
-      "step": 2337
-    },
-    {
-      "epoch": 1.219295958279009,
-      "grad_norm": 1.5494656562805176,
-      "learning_rate": 9.815376884422111e-05,
-      "loss": 6.2522,
-      "step": 2338
-    },
-    {
-      "epoch": 1.2198174706649283,
-      "grad_norm": 1.567967414855957,
-      "learning_rate": 9.815276381909549e-05,
-      "loss": 6.3534,
-      "step": 2339
-    },
-    {
-      "epoch": 1.2203389830508475,
-      "grad_norm": 1.6258360147476196,
-      "learning_rate": 9.815175879396985e-05,
-      "loss": 6.4877,
-      "step": 2340
-    },
-    {
-      "epoch": 1.2208604954367666,
-      "grad_norm": 1.8733817338943481,
-      "learning_rate": 9.815075376884423e-05,
-      "loss": 5.8444,
-      "step": 2341
-    },
-    {
-      "epoch": 1.2213820078226858,
-      "grad_norm": 1.7575840950012207,
-      "learning_rate": 9.81497487437186e-05,
-      "loss": 5.7722,
-      "step": 2342
-    },
-    {
-      "epoch": 1.221903520208605,
-      "grad_norm": 2.2345924377441406,
-      "learning_rate": 9.814874371859297e-05,
-      "loss": 5.6587,
-      "step": 2343
-    },
-    {
-      "epoch": 1.222425032594524,
-      "grad_norm": 2.177804946899414,
-      "learning_rate": 9.814773869346734e-05,
-      "loss": 5.6759,
-      "step": 2344
-    },
-    {
-      "epoch": 1.2229465449804433,
-      "grad_norm": 1.8077000379562378,
-      "learning_rate": 9.814673366834171e-05,
-      "loss": 5.6903,
-      "step": 2345
-    },
-    {
-      "epoch": 1.2234680573663623,
-      "grad_norm": 1.6206762790679932,
-      "learning_rate": 9.814572864321608e-05,
-      "loss": 5.9817,
-      "step": 2346
-    },
-    {
-      "epoch": 1.2239895697522816,
-      "grad_norm": 1.6954774856567383,
-      "learning_rate": 9.814472361809046e-05,
-      "loss": 6.5568,
-      "step": 2347
-    },
-    {
-      "epoch": 1.2245110821382008,
-      "grad_norm": 1.815686821937561,
-      "learning_rate": 9.814371859296483e-05,
-      "loss": 6.1389,
-      "step": 2348
-    },
-    {
-      "epoch": 1.2250325945241198,
-      "grad_norm": 1.7051798105239868,
-      "learning_rate": 9.814271356783921e-05,
-      "loss": 6.1791,
-      "step": 2349
-    },
-    {
-      "epoch": 1.225554106910039,
-      "grad_norm": 1.7045036554336548,
-      "learning_rate": 9.814170854271358e-05,
-      "loss": 6.0514,
-      "step": 2350
-    },
-    {
-      "epoch": 1.2260756192959583,
-      "grad_norm": 1.4944511651992798,
-      "learning_rate": 9.814070351758794e-05,
-      "loss": 6.2937,
-      "step": 2351
-    },
-    {
-      "epoch": 1.2265971316818773,
-      "grad_norm": 1.5316293239593506,
-      "learning_rate": 9.813969849246232e-05,
-      "loss": 6.1102,
-      "step": 2352
-    },
-    {
-      "epoch": 1.2271186440677966,
-      "grad_norm": 1.761322021484375,
-      "learning_rate": 9.813869346733668e-05,
-      "loss": 6.0906,
-      "step": 2353
-    },
-    {
-      "epoch": 1.2276401564537158,
-      "grad_norm": 1.932059407234192,
-      "learning_rate": 9.813768844221106e-05,
-      "loss": 5.2294,
-      "step": 2354
-    },
-    {
-      "epoch": 1.2281616688396348,
-      "grad_norm": 1.5344874858856201,
-      "learning_rate": 9.813668341708543e-05,
-      "loss": 6.3834,
-      "step": 2355
-    },
-    {
-      "epoch": 1.228683181225554,
-      "grad_norm": 1.8602535724639893,
-      "learning_rate": 9.81356783919598e-05,
-      "loss": 6.1281,
-      "step": 2356
-    },
-    {
-      "epoch": 1.2292046936114733,
-      "grad_norm": 2.0995049476623535,
-      "learning_rate": 9.813467336683417e-05,
-      "loss": 5.8005,
-      "step": 2357
-    },
-    {
-      "epoch": 1.2297262059973924,
-      "grad_norm": 1.6905689239501953,
-      "learning_rate": 9.813366834170855e-05,
-      "loss": 5.8638,
-      "step": 2358
-    },
-    {
-      "epoch": 1.2302477183833116,
-      "grad_norm": 1.4666197299957275,
-      "learning_rate": 9.813266331658292e-05,
-      "loss": 5.8753,
-      "step": 2359
-    },
-    {
-      "epoch": 1.2307692307692308,
-      "grad_norm": 1.5940220355987549,
-      "learning_rate": 9.81316582914573e-05,
-      "loss": 6.2041,
-      "step": 2360
-    },
-    {
-      "epoch": 1.2312907431551499,
-      "grad_norm": 1.532310128211975,
-      "learning_rate": 9.813065326633167e-05,
-      "loss": 5.5097,
-      "step": 2361
-    },
-    {
-      "epoch": 1.231812255541069,
-      "grad_norm": 1.5657151937484741,
-      "learning_rate": 9.812964824120604e-05,
-      "loss": 6.1737,
-      "step": 2362
-    },
-    {
-      "epoch": 1.2323337679269883,
-      "grad_norm": 1.6417806148529053,
-      "learning_rate": 9.812864321608041e-05,
-      "loss": 6.1769,
-      "step": 2363
-    },
-    {
-      "epoch": 1.2328552803129074,
-      "grad_norm": 2.9208431243896484,
-      "learning_rate": 9.812763819095477e-05,
-      "loss": 6.0494,
-      "step": 2364
-    },
-    {
-      "epoch": 1.2333767926988266,
-      "grad_norm": 1.9619948863983154,
-      "learning_rate": 9.812663316582915e-05,
-      "loss": 5.8946,
-      "step": 2365
-    },
-    {
-      "epoch": 1.2338983050847459,
-      "grad_norm": 1.982285976409912,
-      "learning_rate": 9.812562814070351e-05,
-      "loss": 5.5907,
-      "step": 2366
-    },
-    {
-      "epoch": 1.2344198174706649,
-      "grad_norm": 1.9110360145568848,
-      "learning_rate": 9.812462311557789e-05,
-      "loss": 6.056,
-      "step": 2367
-    },
-    {
-      "epoch": 1.2349413298565841,
-      "grad_norm": 1.975943684577942,
-      "learning_rate": 9.812361809045227e-05,
-      "loss": 6.2207,
-      "step": 2368
-    },
-    {
-      "epoch": 1.2354628422425034,
-      "grad_norm": 1.454530954360962,
-      "learning_rate": 9.812261306532665e-05,
-      "loss": 5.6655,
-      "step": 2369
-    },
-    {
-      "epoch": 1.2359843546284224,
-      "grad_norm": 1.6475328207015991,
-      "learning_rate": 9.812160804020101e-05,
-      "loss": 6.2606,
-      "step": 2370
-    },
-    {
-      "epoch": 1.2365058670143416,
-      "grad_norm": 1.6005642414093018,
-      "learning_rate": 9.812060301507539e-05,
-      "loss": 6.033,
-      "step": 2371
-    },
-    {
-      "epoch": 1.2370273794002609,
-      "grad_norm": 1.6820613145828247,
-      "learning_rate": 9.811959798994975e-05,
-      "loss": 6.1884,
-      "step": 2372
-    },
-    {
-      "epoch": 1.2375488917861799,
-      "grad_norm": 1.7590827941894531,
-      "learning_rate": 9.811859296482413e-05,
-      "loss": 6.0102,
-      "step": 2373
-    },
-    {
-      "epoch": 1.2380704041720991,
-      "grad_norm": 1.5519553422927856,
-      "learning_rate": 9.81175879396985e-05,
-      "loss": 6.2487,
-      "step": 2374
-    },
-    {
-      "epoch": 1.2385919165580184,
-      "grad_norm": 1.5279432535171509,
-      "learning_rate": 9.811658291457287e-05,
-      "loss": 5.5374,
-      "step": 2375
-    },
-    {
-      "epoch": 1.2391134289439374,
-      "grad_norm": 1.8449634313583374,
-      "learning_rate": 9.811557788944724e-05,
-      "loss": 5.9623,
-      "step": 2376
-    },
-    {
-      "epoch": 1.2396349413298566,
-      "grad_norm": 2.171844720840454,
-      "learning_rate": 9.81145728643216e-05,
-      "loss": 5.9813,
-      "step": 2377
-    },
-    {
-      "epoch": 1.2401564537157757,
-      "grad_norm": 1.9095181226730347,
-      "learning_rate": 9.811356783919598e-05,
-      "loss": 5.5188,
-      "step": 2378
-    },
-    {
-      "epoch": 1.240677966101695,
-      "grad_norm": 2.523716688156128,
-      "learning_rate": 9.811256281407036e-05,
-      "loss": 5.7852,
-      "step": 2379
-    },
-    {
-      "epoch": 1.2411994784876141,
-      "grad_norm": 2.035066843032837,
-      "learning_rate": 9.811155778894474e-05,
-      "loss": 5.7163,
-      "step": 2380
-    },
-    {
-      "epoch": 1.2417209908735332,
-      "grad_norm": 1.975915789604187,
-      "learning_rate": 9.81105527638191e-05,
-      "loss": 6.0897,
-      "step": 2381
-    },
-    {
-      "epoch": 1.2422425032594524,
-      "grad_norm": 1.7738310098648071,
-      "learning_rate": 9.810954773869348e-05,
-      "loss": 6.2117,
-      "step": 2382
-    },
-    {
-      "epoch": 1.2427640156453716,
-      "grad_norm": 1.6086384057998657,
-      "learning_rate": 9.810854271356784e-05,
-      "loss": 6.1091,
-      "step": 2383
-    },
-    {
-      "epoch": 1.2432855280312907,
-      "grad_norm": 2.0317318439483643,
-      "learning_rate": 9.810753768844222e-05,
-      "loss": 5.345,
-      "step": 2384
-    },
-    {
-      "epoch": 1.24380704041721,
-      "grad_norm": 1.6118898391723633,
-      "learning_rate": 9.810653266331658e-05,
-      "loss": 6.2362,
-      "step": 2385
-    },
-    {
-      "epoch": 1.2443285528031292,
-      "grad_norm": 1.6175588369369507,
-      "learning_rate": 9.810552763819096e-05,
-      "loss": 6.144,
-      "step": 2386
-    },
-    {
-      "epoch": 1.2448500651890482,
-      "grad_norm": 1.6560720205307007,
-      "learning_rate": 9.810452261306533e-05,
-      "loss": 6.0237,
-      "step": 2387
-    },
-    {
-      "epoch": 1.2453715775749674,
-      "grad_norm": 1.6471009254455566,
-      "learning_rate": 9.81035175879397e-05,
-      "loss": 5.5321,
-      "step": 2388
-    },
-    {
-      "epoch": 1.2458930899608867,
-      "grad_norm": 2.081552505493164,
-      "learning_rate": 9.810251256281408e-05,
-      "loss": 5.8745,
-      "step": 2389
-    },
-    {
-      "epoch": 1.2464146023468057,
-      "grad_norm": 1.6420671939849854,
-      "learning_rate": 9.810150753768845e-05,
-      "loss": 6.3909,
-      "step": 2390
-    },
-    {
-      "epoch": 1.246936114732725,
-      "grad_norm": 1.7266751527786255,
-      "learning_rate": 9.810050251256282e-05,
-      "loss": 6.219,
-      "step": 2391
-    },
-    {
-      "epoch": 1.2474576271186442,
-      "grad_norm": 1.7193596363067627,
-      "learning_rate": 9.809949748743719e-05,
-      "loss": 6.1863,
-      "step": 2392
-    },
-    {
-      "epoch": 1.2479791395045632,
-      "grad_norm": 1.615049958229065,
-      "learning_rate": 9.809849246231157e-05,
-      "loss": 6.1865,
-      "step": 2393
-    },
-    {
-      "epoch": 1.2485006518904824,
-      "grad_norm": 1.4636505842208862,
-      "learning_rate": 9.809748743718593e-05,
-      "loss": 6.1721,
-      "step": 2394
-    },
-    {
-      "epoch": 1.2490221642764014,
-      "grad_norm": 1.4940389394760132,
-      "learning_rate": 9.809648241206031e-05,
-      "loss": 6.2341,
-      "step": 2395
-    },
-    {
-      "epoch": 1.2495436766623207,
-      "grad_norm": 1.5442814826965332,
-      "learning_rate": 9.809547738693467e-05,
-      "loss": 6.3287,
-      "step": 2396
-    },
-    {
-      "epoch": 1.25006518904824,
-      "grad_norm": 1.662725567817688,
-      "learning_rate": 9.809447236180905e-05,
-      "loss": 6.0666,
-      "step": 2397
-    },
-    {
-      "epoch": 1.250586701434159,
-      "grad_norm": 1.5695070028305054,
-      "learning_rate": 9.809346733668341e-05,
-      "loss": 5.967,
-      "step": 2398
-    },
-    {
-      "epoch": 1.2511082138200782,
-      "grad_norm": 1.7494882345199585,
-      "learning_rate": 9.809246231155779e-05,
-      "loss": 5.7404,
-      "step": 2399
-    },
-    {
-      "epoch": 1.2516297262059974,
-      "grad_norm": 1.7852306365966797,
-      "learning_rate": 9.809145728643217e-05,
-      "loss": 5.748,
-      "step": 2400
-    },
-    {
-      "epoch": 1.2521512385919165,
-      "grad_norm": 1.620023488998413,
-      "learning_rate": 9.809045226130655e-05,
-      "loss": 5.7975,
-      "step": 2401
-    },
-    {
-      "epoch": 1.2526727509778357,
-      "grad_norm": 1.6267998218536377,
-      "learning_rate": 9.808944723618091e-05,
-      "loss": 6.3073,
-      "step": 2402
-    },
-    {
-      "epoch": 1.253194263363755,
-      "grad_norm": 1.6103161573410034,
-      "learning_rate": 9.808844221105528e-05,
-      "loss": 6.3882,
-      "step": 2403
-    },
-    {
-      "epoch": 1.253715775749674,
-      "grad_norm": 1.8769530057907104,
-      "learning_rate": 9.808743718592965e-05,
-      "loss": 6.2069,
-      "step": 2404
-    },
-    {
-      "epoch": 1.2542372881355932,
-      "grad_norm": 1.8078742027282715,
-      "learning_rate": 9.808643216080402e-05,
-      "loss": 6.1111,
-      "step": 2405
-    },
-    {
-      "epoch": 1.2547588005215125,
-      "grad_norm": 1.5848397016525269,
-      "learning_rate": 9.80854271356784e-05,
-      "loss": 6.0275,
-      "step": 2406
-    },
-    {
-      "epoch": 1.2552803129074315,
-      "grad_norm": 1.5705652236938477,
-      "learning_rate": 9.808442211055276e-05,
-      "loss": 6.0868,
-      "step": 2407
-    },
-    {
-      "epoch": 1.2558018252933507,
-      "grad_norm": 1.5766164064407349,
-      "learning_rate": 9.808341708542714e-05,
-      "loss": 6.3224,
-      "step": 2408
-    },
-    {
-      "epoch": 1.25632333767927,
-      "grad_norm": 1.7578343152999878,
-      "learning_rate": 9.80824120603015e-05,
-      "loss": 6.0232,
-      "step": 2409
-    },
-    {
-      "epoch": 1.256844850065189,
-      "grad_norm": 1.6132627725601196,
-      "learning_rate": 9.808140703517588e-05,
-      "loss": 6.3296,
-      "step": 2410
-    },
-    {
-      "epoch": 1.2573663624511082,
-      "grad_norm": 1.480827808380127,
-      "learning_rate": 9.808040201005026e-05,
-      "loss": 5.3716,
-      "step": 2411
-    },
-    {
-      "epoch": 1.2578878748370275,
-      "grad_norm": 1.8271872997283936,
-      "learning_rate": 9.807939698492464e-05,
-      "loss": 6.3604,
-      "step": 2412
-    },
-    {
-      "epoch": 1.2584093872229465,
-      "grad_norm": 1.5612602233886719,
-      "learning_rate": 9.8078391959799e-05,
-      "loss": 6.0017,
-      "step": 2413
-    },
-    {
-      "epoch": 1.2589308996088657,
-      "grad_norm": 1.9115878343582153,
-      "learning_rate": 9.807738693467338e-05,
-      "loss": 6.1742,
-      "step": 2414
-    },
-    {
-      "epoch": 1.259452411994785,
-      "grad_norm": 2.0935845375061035,
-      "learning_rate": 9.807638190954774e-05,
-      "loss": 6.3859,
-      "step": 2415
-    },
-    {
-      "epoch": 1.259973924380704,
-      "grad_norm": 1.5691924095153809,
-      "learning_rate": 9.807537688442211e-05,
-      "loss": 6.0657,
-      "step": 2416
-    },
-    {
-      "epoch": 1.2604954367666232,
-      "grad_norm": 1.652215838432312,
-      "learning_rate": 9.807437185929648e-05,
-      "loss": 5.5103,
-      "step": 2417
-    },
-    {
-      "epoch": 1.2610169491525425,
-      "grad_norm": 1.9450608491897583,
-      "learning_rate": 9.807336683417085e-05,
-      "loss": 6.2519,
-      "step": 2418
-    },
-    {
-      "epoch": 1.2615384615384615,
-      "grad_norm": 1.692731499671936,
-      "learning_rate": 9.807236180904523e-05,
-      "loss": 6.0318,
-      "step": 2419
-    },
-    {
-      "epoch": 1.2620599739243807,
-      "grad_norm": 1.6469907760620117,
-      "learning_rate": 9.80713567839196e-05,
-      "loss": 6.1992,
-      "step": 2420
-    },
-    {
-      "epoch": 1.2625814863103,
-      "grad_norm": 1.776899814605713,
-      "learning_rate": 9.807035175879398e-05,
-      "loss": 6.1549,
-      "step": 2421
-    },
-    {
-      "epoch": 1.263102998696219,
-      "grad_norm": 1.7042206525802612,
-      "learning_rate": 9.806934673366835e-05,
-      "loss": 6.1194,
-      "step": 2422
-    },
-    {
-      "epoch": 1.2636245110821382,
-      "grad_norm": 1.4940481185913086,
-      "learning_rate": 9.806834170854272e-05,
-      "loss": 5.7737,
-      "step": 2423
-    },
-    {
-      "epoch": 1.2641460234680575,
-      "grad_norm": 1.8041486740112305,
-      "learning_rate": 9.806733668341709e-05,
-      "loss": 6.1321,
-      "step": 2424
-    },
-    {
-      "epoch": 1.2646675358539765,
-      "grad_norm": 1.6374828815460205,
-      "learning_rate": 9.806633165829147e-05,
-      "loss": 6.1351,
-      "step": 2425
-    },
-    {
-      "epoch": 1.2651890482398958,
-      "grad_norm": 1.774246096611023,
-      "learning_rate": 9.806532663316583e-05,
-      "loss": 6.0356,
-      "step": 2426
-    },
-    {
-      "epoch": 1.265710560625815,
-      "grad_norm": 2.0529778003692627,
-      "learning_rate": 9.806432160804021e-05,
-      "loss": 5.5112,
-      "step": 2427
-    },
-    {
-      "epoch": 1.266232073011734,
-      "grad_norm": 1.7884122133255005,
-      "learning_rate": 9.806331658291457e-05,
-      "loss": 5.346,
-      "step": 2428
-    },
-    {
-      "epoch": 1.2667535853976533,
-      "grad_norm": 1.6786189079284668,
-      "learning_rate": 9.806231155778894e-05,
-      "loss": 5.8765,
-      "step": 2429
-    },
-    {
-      "epoch": 1.2672750977835723,
-      "grad_norm": 1.947405457496643,
-      "learning_rate": 9.806130653266332e-05,
-      "loss": 6.323,
-      "step": 2430
-    },
-    {
-      "epoch": 1.2677966101694915,
-      "grad_norm": 1.6861083507537842,
-      "learning_rate": 9.806030150753769e-05,
-      "loss": 6.3398,
-      "step": 2431
-    },
-    {
-      "epoch": 1.2683181225554108,
-      "grad_norm": 1.694516897201538,
-      "learning_rate": 9.805929648241207e-05,
-      "loss": 6.252,
-      "step": 2432
-    },
-    {
-      "epoch": 1.2688396349413298,
-      "grad_norm": 1.7161301374435425,
-      "learning_rate": 9.805829145728644e-05,
-      "loss": 6.4414,
-      "step": 2433
-    },
-    {
-      "epoch": 1.269361147327249,
-      "grad_norm": 1.750016689300537,
-      "learning_rate": 9.805728643216081e-05,
-      "loss": 6.0129,
-      "step": 2434
-    },
-    {
-      "epoch": 1.269882659713168,
-      "grad_norm": 1.5770639181137085,
-      "learning_rate": 9.805628140703518e-05,
-      "loss": 6.1028,
-      "step": 2435
-    },
-    {
-      "epoch": 1.2704041720990873,
-      "grad_norm": 1.5977411270141602,
-      "learning_rate": 9.805527638190956e-05,
-      "loss": 6.2965,
-      "step": 2436
-    },
-    {
-      "epoch": 1.2709256844850065,
-      "grad_norm": 1.6670105457305908,
-      "learning_rate": 9.805427135678392e-05,
-      "loss": 6.159,
-      "step": 2437
-    },
-    {
-      "epoch": 1.2714471968709256,
-      "grad_norm": 2.2631120681762695,
-      "learning_rate": 9.80532663316583e-05,
-      "loss": 5.8192,
-      "step": 2438
-    },
-    {
-      "epoch": 1.2719687092568448,
-      "grad_norm": 1.9512810707092285,
-      "learning_rate": 9.805226130653266e-05,
-      "loss": 5.774,
-      "step": 2439
-    },
-    {
-      "epoch": 1.272490221642764,
-      "grad_norm": 1.7135931253433228,
-      "learning_rate": 9.805125628140704e-05,
-      "loss": 6.0227,
-      "step": 2440
-    },
-    {
-      "epoch": 1.273011734028683,
-      "grad_norm": 1.7273602485656738,
-      "learning_rate": 9.805025125628142e-05,
-      "loss": 5.9689,
-      "step": 2441
-    },
-    {
-      "epoch": 1.2735332464146023,
-      "grad_norm": 1.6565227508544922,
-      "learning_rate": 9.80492462311558e-05,
-      "loss": 6.0811,
-      "step": 2442
-    },
-    {
-      "epoch": 1.2740547588005215,
-      "grad_norm": 1.4586776494979858,
-      "learning_rate": 9.804824120603016e-05,
-      "loss": 6.3572,
-      "step": 2443
-    },
-    {
-      "epoch": 1.2745762711864406,
-      "grad_norm": 1.7837969064712524,
-      "learning_rate": 9.804723618090452e-05,
-      "loss": 5.8148,
-      "step": 2444
-    },
-    {
-      "epoch": 1.2750977835723598,
-      "grad_norm": 2.64665150642395,
-      "learning_rate": 9.80462311557789e-05,
-      "loss": 6.1503,
-      "step": 2445
-    },
-    {
-      "epoch": 1.275619295958279,
-      "grad_norm": 1.5442789793014526,
-      "learning_rate": 9.804522613065327e-05,
-      "loss": 6.0142,
-      "step": 2446
-    },
-    {
-      "epoch": 1.276140808344198,
-      "grad_norm": 1.8197027444839478,
-      "learning_rate": 9.804422110552764e-05,
-      "loss": 6.1549,
-      "step": 2447
-    },
-    {
-      "epoch": 1.2766623207301173,
-      "grad_norm": 1.8374695777893066,
-      "learning_rate": 9.804321608040201e-05,
-      "loss": 6.1212,
-      "step": 2448
-    },
-    {
-      "epoch": 1.2771838331160366,
-      "grad_norm": 1.674656867980957,
-      "learning_rate": 9.804221105527639e-05,
-      "loss": 6.2472,
-      "step": 2449
-    },
-    {
-      "epoch": 1.2777053455019556,
-      "grad_norm": 1.585371971130371,
-      "learning_rate": 9.804120603015075e-05,
-      "loss": 6.1531,
-      "step": 2450
-    },
-    {
-      "epoch": 1.2782268578878748,
-      "grad_norm": 1.6216912269592285,
-      "learning_rate": 9.804020100502513e-05,
-      "loss": 6.1949,
-      "step": 2451
-    },
-    {
-      "epoch": 1.278748370273794,
-      "grad_norm": 1.6730854511260986,
-      "learning_rate": 9.80391959798995e-05,
-      "loss": 5.6518,
-      "step": 2452
-    },
-    {
-      "epoch": 1.279269882659713,
-      "grad_norm": 1.676228404045105,
-      "learning_rate": 9.803819095477388e-05,
-      "loss": 6.2649,
-      "step": 2453
-    },
-    {
-      "epoch": 1.2797913950456323,
-      "grad_norm": 1.7399259805679321,
-      "learning_rate": 9.803718592964825e-05,
-      "loss": 5.9254,
-      "step": 2454
-    },
-    {
-      "epoch": 1.2803129074315516,
-      "grad_norm": 1.9783879518508911,
-      "learning_rate": 9.803618090452263e-05,
-      "loss": 5.6008,
-      "step": 2455
-    },
-    {
-      "epoch": 1.2808344198174706,
-      "grad_norm": 1.4143816232681274,
-      "learning_rate": 9.803517587939699e-05,
-      "loss": 6.3869,
-      "step": 2456
-    },
-    {
-      "epoch": 1.2813559322033898,
-      "grad_norm": 1.5272953510284424,
-      "learning_rate": 9.803417085427135e-05,
-      "loss": 6.5396,
-      "step": 2457
-    },
-    {
-      "epoch": 1.281877444589309,
-      "grad_norm": 1.6473231315612793,
-      "learning_rate": 9.803316582914573e-05,
-      "loss": 6.0603,
-      "step": 2458
-    },
-    {
-      "epoch": 1.282398956975228,
-      "grad_norm": 1.557220697402954,
-      "learning_rate": 9.80321608040201e-05,
-      "loss": 6.1595,
-      "step": 2459
-    },
-    {
-      "epoch": 1.2829204693611473,
-      "grad_norm": 1.5079889297485352,
-      "learning_rate": 9.803115577889447e-05,
-      "loss": 6.1603,
-      "step": 2460
-    },
-    {
-      "epoch": 1.2834419817470666,
-      "grad_norm": 1.6709891557693481,
-      "learning_rate": 9.803015075376885e-05,
-      "loss": 5.3529,
-      "step": 2461
-    },
-    {
-      "epoch": 1.2839634941329856,
-      "grad_norm": 1.5749216079711914,
-      "learning_rate": 9.802914572864323e-05,
-      "loss": 6.0981,
-      "step": 2462
-    },
-    {
-      "epoch": 1.2844850065189048,
-      "grad_norm": 1.7351951599121094,
-      "learning_rate": 9.80281407035176e-05,
-      "loss": 6.0681,
-      "step": 2463
-    },
-    {
-      "epoch": 1.285006518904824,
-      "grad_norm": 1.6309770345687866,
-      "learning_rate": 9.802713567839197e-05,
-      "loss": 6.3394,
-      "step": 2464
-    },
-    {
-      "epoch": 1.2855280312907431,
-      "grad_norm": 1.4928302764892578,
-      "learning_rate": 9.802613065326634e-05,
-      "loss": 6.3404,
-      "step": 2465
-    },
-    {
-      "epoch": 1.2860495436766624,
-      "grad_norm": 1.9041593074798584,
-      "learning_rate": 9.802512562814071e-05,
-      "loss": 5.887,
-      "step": 2466
-    },
-    {
-      "epoch": 1.2865710560625816,
-      "grad_norm": 1.6199983358383179,
-      "learning_rate": 9.802412060301508e-05,
-      "loss": 6.254,
-      "step": 2467
-    },
-    {
-      "epoch": 1.2870925684485006,
-      "grad_norm": 1.7301759719848633,
-      "learning_rate": 9.802311557788946e-05,
-      "loss": 5.9548,
-      "step": 2468
-    },
-    {
-      "epoch": 1.2876140808344199,
-      "grad_norm": 1.7285408973693848,
-      "learning_rate": 9.802211055276382e-05,
-      "loss": 5.7616,
-      "step": 2469
-    },
-    {
-      "epoch": 1.288135593220339,
-      "grad_norm": 1.7891991138458252,
-      "learning_rate": 9.802110552763818e-05,
-      "loss": 6.2355,
-      "step": 2470
-    },
-    {
-      "epoch": 1.2886571056062581,
-      "grad_norm": 1.5435711145401,
-      "learning_rate": 9.802010050251256e-05,
-      "loss": 6.3816,
-      "step": 2471
-    },
-    {
-      "epoch": 1.2891786179921774,
-      "grad_norm": 1.8156284093856812,
-      "learning_rate": 9.801909547738694e-05,
-      "loss": 5.5133,
-      "step": 2472
-    },
-    {
-      "epoch": 1.2897001303780966,
-      "grad_norm": 1.7439326047897339,
-      "learning_rate": 9.801809045226132e-05,
-      "loss": 6.2033,
-      "step": 2473
-    },
-    {
-      "epoch": 1.2902216427640156,
-      "grad_norm": 1.7019009590148926,
-      "learning_rate": 9.801708542713568e-05,
-      "loss": 5.5285,
-      "step": 2474
-    },
-    {
-      "epoch": 1.2907431551499349,
-      "grad_norm": 1.8198002576828003,
-      "learning_rate": 9.801608040201006e-05,
-      "loss": 5.458,
-      "step": 2475
-    },
-    {
-      "epoch": 1.291264667535854,
-      "grad_norm": 1.6428302526474,
-      "learning_rate": 9.801507537688442e-05,
-      "loss": 6.1403,
-      "step": 2476
-    },
-    {
-      "epoch": 1.2917861799217731,
-      "grad_norm": 1.5422180891036987,
-      "learning_rate": 9.80140703517588e-05,
-      "loss": 6.2856,
-      "step": 2477
-    },
-    {
-      "epoch": 1.2923076923076924,
-      "grad_norm": 1.499881625175476,
-      "learning_rate": 9.801306532663317e-05,
-      "loss": 6.1848,
-      "step": 2478
-    },
-    {
-      "epoch": 1.2928292046936114,
-      "grad_norm": 1.6463180780410767,
-      "learning_rate": 9.801206030150754e-05,
-      "loss": 5.4377,
-      "step": 2479
-    },
-    {
-      "epoch": 1.2933507170795306,
-      "grad_norm": 1.7342177629470825,
-      "learning_rate": 9.801105527638191e-05,
-      "loss": 5.959,
-      "step": 2480
-    },
-    {
-      "epoch": 1.2938722294654499,
-      "grad_norm": 1.9241501092910767,
-      "learning_rate": 9.801005025125629e-05,
-      "loss": 5.8093,
-      "step": 2481
-    },
-    {
-      "epoch": 1.294393741851369,
-      "grad_norm": 1.6937975883483887,
-      "learning_rate": 9.800904522613066e-05,
-      "loss": 5.5399,
-      "step": 2482
-    },
-    {
-      "epoch": 1.2949152542372881,
-      "grad_norm": 1.646341323852539,
-      "learning_rate": 9.800804020100503e-05,
-      "loss": 6.2984,
-      "step": 2483
-    },
-    {
-      "epoch": 1.2954367666232072,
-      "grad_norm": 1.5861973762512207,
-      "learning_rate": 9.80070351758794e-05,
-      "loss": 6.2272,
-      "step": 2484
-    },
-    {
-      "epoch": 1.2959582790091264,
-      "grad_norm": 1.4318344593048096,
-      "learning_rate": 9.800603015075377e-05,
-      "loss": 6.023,
-      "step": 2485
-    },
-    {
-      "epoch": 1.2964797913950457,
-      "grad_norm": 1.5936084985733032,
-      "learning_rate": 9.800502512562815e-05,
-      "loss": 6.2481,
-      "step": 2486
-    },
-    {
-      "epoch": 1.2970013037809647,
-      "grad_norm": 1.5022445917129517,
-      "learning_rate": 9.800402010050251e-05,
-      "loss": 6.2545,
-      "step": 2487
-    },
-    {
-      "epoch": 1.297522816166884,
-      "grad_norm": 1.5663511753082275,
-      "learning_rate": 9.800301507537689e-05,
-      "loss": 5.8146,
-      "step": 2488
-    },
-    {
-      "epoch": 1.2980443285528032,
-      "grad_norm": 1.574946641921997,
-      "learning_rate": 9.800201005025125e-05,
-      "loss": 5.9538,
-      "step": 2489
-    },
-    {
-      "epoch": 1.2985658409387222,
-      "grad_norm": 1.575602650642395,
-      "learning_rate": 9.800100502512563e-05,
-      "loss": 5.959,
-      "step": 2490
-    },
-    {
-      "epoch": 1.2990873533246414,
-      "grad_norm": 1.4218430519104004,
-      "learning_rate": 9.8e-05,
-      "loss": 6.1358,
-      "step": 2491
-    },
-    {
-      "epoch": 1.2996088657105607,
-      "grad_norm": 1.6269416809082031,
-      "learning_rate": 9.799899497487437e-05,
-      "loss": 5.571,
-      "step": 2492
-    },
-    {
-      "epoch": 1.3001303780964797,
-      "grad_norm": 1.6003565788269043,
-      "learning_rate": 9.799798994974875e-05,
-      "loss": 5.9888,
-      "step": 2493
-    },
-    {
-      "epoch": 1.300651890482399,
-      "grad_norm": 2.0037682056427,
-      "learning_rate": 9.799698492462313e-05,
-      "loss": 6.3128,
-      "step": 2494
-    },
-    {
-      "epoch": 1.3011734028683182,
-      "grad_norm": 1.8058274984359741,
-      "learning_rate": 9.79959798994975e-05,
-      "loss": 5.7687,
-      "step": 2495
-    },
-    {
-      "epoch": 1.3016949152542372,
-      "grad_norm": 1.7981421947479248,
-      "learning_rate": 9.799497487437186e-05,
-      "loss": 6.1458,
-      "step": 2496
-    },
-    {
-      "epoch": 1.3022164276401564,
-      "grad_norm": 1.8295743465423584,
-      "learning_rate": 9.799396984924624e-05,
-      "loss": 5.6928,
-      "step": 2497
-    },
-    {
-      "epoch": 1.3027379400260757,
-      "grad_norm": 1.759970784187317,
-      "learning_rate": 9.79929648241206e-05,
-      "loss": 5.9358,
-      "step": 2498
-    },
-    {
-      "epoch": 1.3032594524119947,
-      "grad_norm": 1.6202988624572754,
-      "learning_rate": 9.799195979899498e-05,
-      "loss": 6.4561,
-      "step": 2499
-    },
-    {
-      "epoch": 1.303780964797914,
-      "grad_norm": 1.6213043928146362,
-      "learning_rate": 9.799095477386934e-05,
-      "loss": 6.0728,
-      "step": 2500
-    },
-    {
-      "epoch": 1.3043024771838332,
-      "grad_norm": 1.6311516761779785,
-      "learning_rate": 9.798994974874372e-05,
-      "loss": 6.0394,
-      "step": 2501
-    },
-    {
-      "epoch": 1.3048239895697522,
-      "grad_norm": 1.4928587675094604,
-      "learning_rate": 9.79889447236181e-05,
-      "loss": 6.1494,
-      "step": 2502
-    },
-    {
-      "epoch": 1.3053455019556715,
-      "grad_norm": 1.6309846639633179,
-      "learning_rate": 9.798793969849248e-05,
-      "loss": 6.1334,
-      "step": 2503
-    },
-    {
-      "epoch": 1.3058670143415907,
-      "grad_norm": 1.5226510763168335,
-      "learning_rate": 9.798693467336684e-05,
-      "loss": 6.2621,
-      "step": 2504
-    },
-    {
-      "epoch": 1.3063885267275097,
-      "grad_norm": 1.6723437309265137,
-      "learning_rate": 9.798592964824122e-05,
-      "loss": 6.0395,
-      "step": 2505
-    },
-    {
-      "epoch": 1.306910039113429,
-      "grad_norm": 1.5453710556030273,
-      "learning_rate": 9.798492462311558e-05,
-      "loss": 6.2832,
-      "step": 2506
-    },
-    {
-      "epoch": 1.3074315514993482,
-      "grad_norm": 1.5942221879959106,
-      "learning_rate": 9.798391959798996e-05,
-      "loss": 5.8928,
-      "step": 2507
-    },
-    {
-      "epoch": 1.3079530638852672,
-      "grad_norm": 1.722939133644104,
-      "learning_rate": 9.798291457286433e-05,
-      "loss": 5.9817,
-      "step": 2508
-    },
-    {
-      "epoch": 1.3084745762711865,
-      "grad_norm": 1.8619356155395508,
-      "learning_rate": 9.798190954773869e-05,
-      "loss": 5.6537,
-      "step": 2509
-    },
-    {
-      "epoch": 1.3089960886571057,
-      "grad_norm": 1.6537522077560425,
-      "learning_rate": 9.798090452261307e-05,
-      "loss": 6.0491,
-      "step": 2510
-    },
-    {
-      "epoch": 1.3095176010430247,
-      "grad_norm": 1.5271246433258057,
-      "learning_rate": 9.797989949748743e-05,
-      "loss": 6.2063,
-      "step": 2511
-    },
-    {
-      "epoch": 1.310039113428944,
-      "grad_norm": 1.718353509902954,
-      "learning_rate": 9.797889447236181e-05,
-      "loss": 5.6789,
-      "step": 2512
-    },
-    {
-      "epoch": 1.3105606258148632,
-      "grad_norm": 1.6641216278076172,
-      "learning_rate": 9.797788944723619e-05,
-      "loss": 5.8507,
-      "step": 2513
-    },
-    {
-      "epoch": 1.3110821382007822,
-      "grad_norm": 1.6516773700714111,
-      "learning_rate": 9.797688442211056e-05,
-      "loss": 5.7553,
-      "step": 2514
-    },
-    {
-      "epoch": 1.3116036505867015,
-      "grad_norm": 1.5930336713790894,
-      "learning_rate": 9.797587939698493e-05,
-      "loss": 6.3542,
-      "step": 2515
-    },
-    {
-      "epoch": 1.3121251629726207,
-      "grad_norm": 1.6110321283340454,
-      "learning_rate": 9.797487437185931e-05,
-      "loss": 6.1247,
-      "step": 2516
-    },
-    {
-      "epoch": 1.3126466753585397,
-      "grad_norm": 1.4728533029556274,
-      "learning_rate": 9.797386934673367e-05,
-      "loss": 6.1848,
-      "step": 2517
-    },
-    {
-      "epoch": 1.313168187744459,
-      "grad_norm": 1.437720537185669,
-      "learning_rate": 9.797286432160805e-05,
-      "loss": 6.1445,
-      "step": 2518
-    },
-    {
-      "epoch": 1.3136897001303782,
-      "grad_norm": 1.5480064153671265,
-      "learning_rate": 9.797185929648241e-05,
-      "loss": 6.1675,
-      "step": 2519
-    },
-    {
-      "epoch": 1.3142112125162972,
-      "grad_norm": 1.5881060361862183,
-      "learning_rate": 9.797085427135679e-05,
-      "loss": 6.1074,
-      "step": 2520
-    },
-    {
-      "epoch": 1.3147327249022165,
-      "grad_norm": 1.677389144897461,
-      "learning_rate": 9.796984924623116e-05,
-      "loss": 6.2863,
-      "step": 2521
-    },
-    {
-      "epoch": 1.3152542372881357,
-      "grad_norm": 1.5233243703842163,
-      "learning_rate": 9.796884422110553e-05,
-      "loss": 5.4793,
-      "step": 2522
-    },
-    {
-      "epoch": 1.3157757496740548,
-      "grad_norm": 1.6342347860336304,
-      "learning_rate": 9.796783919597991e-05,
-      "loss": 6.1585,
-      "step": 2523
-    },
-    {
-      "epoch": 1.316297262059974,
-      "grad_norm": 1.5781776905059814,
-      "learning_rate": 9.796683417085428e-05,
-      "loss": 6.2741,
-      "step": 2524
-    },
-    {
-      "epoch": 1.316818774445893,
-      "grad_norm": 1.4817887544631958,
-      "learning_rate": 9.796582914572865e-05,
-      "loss": 6.3891,
-      "step": 2525
-    },
-    {
-      "epoch": 1.3173402868318123,
-      "grad_norm": 1.7219842672348022,
-      "learning_rate": 9.796482412060302e-05,
-      "loss": 5.3118,
-      "step": 2526
-    },
-    {
-      "epoch": 1.3178617992177315,
-      "grad_norm": 1.8795174360275269,
-      "learning_rate": 9.79638190954774e-05,
-      "loss": 5.8285,
-      "step": 2527
-    },
-    {
-      "epoch": 1.3183833116036505,
-      "grad_norm": 1.6331959962844849,
-      "learning_rate": 9.796281407035176e-05,
-      "loss": 6.4864,
-      "step": 2528
-    },
-    {
-      "epoch": 1.3189048239895698,
-      "grad_norm": 1.7642707824707031,
-      "learning_rate": 9.796180904522614e-05,
-      "loss": 5.4986,
-      "step": 2529
-    },
-    {
-      "epoch": 1.3194263363754888,
-      "grad_norm": 2.0098085403442383,
-      "learning_rate": 9.79608040201005e-05,
-      "loss": 5.6884,
-      "step": 2530
-    },
-    {
-      "epoch": 1.319947848761408,
-      "grad_norm": 1.6574918031692505,
-      "learning_rate": 9.795979899497488e-05,
-      "loss": 6.1445,
-      "step": 2531
-    },
-    {
-      "epoch": 1.3204693611473273,
-      "grad_norm": 1.724198341369629,
-      "learning_rate": 9.795879396984924e-05,
-      "loss": 5.874,
-      "step": 2532
-    },
-    {
-      "epoch": 1.3209908735332463,
-      "grad_norm": 1.5275824069976807,
-      "learning_rate": 9.795778894472362e-05,
-      "loss": 6.3019,
-      "step": 2533
-    },
-    {
-      "epoch": 1.3215123859191655,
-      "grad_norm": 1.919723391532898,
-      "learning_rate": 9.7956783919598e-05,
-      "loss": 5.8738,
-      "step": 2534
-    },
-    {
-      "epoch": 1.3220338983050848,
-      "grad_norm": 1.7140527963638306,
-      "learning_rate": 9.795577889447238e-05,
-      "loss": 5.602,
-      "step": 2535
-    },
-    {
-      "epoch": 1.3225554106910038,
-      "grad_norm": 1.700982928276062,
-      "learning_rate": 9.795477386934674e-05,
-      "loss": 6.0537,
-      "step": 2536
-    },
-    {
-      "epoch": 1.323076923076923,
-      "grad_norm": 1.9550904035568237,
-      "learning_rate": 9.79537688442211e-05,
-      "loss": 5.993,
-      "step": 2537
-    },
-    {
-      "epoch": 1.3235984354628423,
-      "grad_norm": 1.9019474983215332,
-      "learning_rate": 9.795276381909548e-05,
-      "loss": 5.8358,
-      "step": 2538
-    },
-    {
-      "epoch": 1.3241199478487613,
-      "grad_norm": 1.6109199523925781,
-      "learning_rate": 9.795175879396985e-05,
-      "loss": 6.1591,
-      "step": 2539
-    },
-    {
-      "epoch": 1.3246414602346805,
-      "grad_norm": 1.4759520292282104,
-      "learning_rate": 9.795075376884423e-05,
-      "loss": 6.2137,
-      "step": 2540
-    },
-    {
-      "epoch": 1.3251629726205998,
-      "grad_norm": 1.5282145738601685,
-      "learning_rate": 9.794974874371859e-05,
-      "loss": 6.0001,
-      "step": 2541
-    },
-    {
-      "epoch": 1.3256844850065188,
-      "grad_norm": 1.5583438873291016,
-      "learning_rate": 9.794874371859297e-05,
-      "loss": 6.2771,
-      "step": 2542
-    },
-    {
-      "epoch": 1.326205997392438,
-      "grad_norm": 1.5829055309295654,
-      "learning_rate": 9.794773869346735e-05,
-      "loss": 6.5383,
-      "step": 2543
-    },
-    {
-      "epoch": 1.3267275097783573,
-      "grad_norm": 1.4666547775268555,
-      "learning_rate": 9.794673366834172e-05,
-      "loss": 5.9638,
-      "step": 2544
-    },
-    {
-      "epoch": 1.3272490221642763,
-      "grad_norm": 1.8580710887908936,
-      "learning_rate": 9.794572864321609e-05,
-      "loss": 5.5906,
-      "step": 2545
-    },
-    {
-      "epoch": 1.3277705345501956,
-      "grad_norm": 1.7471016645431519,
-      "learning_rate": 9.794472361809047e-05,
-      "loss": 5.8475,
-      "step": 2546
-    },
-    {
-      "epoch": 1.3282920469361148,
-      "grad_norm": 1.5614352226257324,
-      "learning_rate": 9.794371859296483e-05,
-      "loss": 5.8226,
-      "step": 2547
-    },
-    {
-      "epoch": 1.3288135593220338,
-      "grad_norm": 1.8175265789031982,
-      "learning_rate": 9.794271356783921e-05,
-      "loss": 5.9907,
-      "step": 2548
-    },
-    {
-      "epoch": 1.329335071707953,
-      "grad_norm": 1.7280012369155884,
-      "learning_rate": 9.794170854271357e-05,
-      "loss": 6.093,
-      "step": 2549
-    },
-    {
-      "epoch": 1.3298565840938723,
-      "grad_norm": 1.6838841438293457,
-      "learning_rate": 9.794070351758794e-05,
-      "loss": 6.0929,
-      "step": 2550
-    },
-    {
-      "epoch": 1.3303780964797913,
-      "grad_norm": 1.689023494720459,
-      "learning_rate": 9.793969849246231e-05,
-      "loss": 6.2114,
-      "step": 2551
-    },
-    {
-      "epoch": 1.3308996088657106,
-      "grad_norm": 1.5648472309112549,
-      "learning_rate": 9.793869346733668e-05,
-      "loss": 6.1307,
-      "step": 2552
-    },
-    {
-      "epoch": 1.3314211212516298,
-      "grad_norm": 1.5117080211639404,
-      "learning_rate": 9.793768844221106e-05,
-      "loss": 6.0646,
-      "step": 2553
-    },
-    {
-      "epoch": 1.3319426336375488,
-      "grad_norm": 1.6181011199951172,
-      "learning_rate": 9.793668341708543e-05,
-      "loss": 6.2321,
-      "step": 2554
-    },
-    {
-      "epoch": 1.332464146023468,
-      "grad_norm": 1.7066125869750977,
-      "learning_rate": 9.793567839195981e-05,
-      "loss": 5.7869,
-      "step": 2555
-    },
-    {
-      "epoch": 1.3329856584093873,
-      "grad_norm": 1.660563349723816,
-      "learning_rate": 9.793467336683418e-05,
-      "loss": 5.5365,
-      "step": 2556
-    },
-    {
-      "epoch": 1.3335071707953063,
-      "grad_norm": 1.5499671697616577,
-      "learning_rate": 9.793366834170855e-05,
-      "loss": 6.2316,
-      "step": 2557
-    },
-    {
-      "epoch": 1.3340286831812256,
-      "grad_norm": 1.5472174882888794,
-      "learning_rate": 9.793266331658292e-05,
-      "loss": 6.2175,
-      "step": 2558
-    },
-    {
-      "epoch": 1.3345501955671448,
-      "grad_norm": 1.8386262655258179,
-      "learning_rate": 9.79316582914573e-05,
-      "loss": 5.3997,
-      "step": 2559
-    },
-    {
-      "epoch": 1.3350717079530638,
-      "grad_norm": 1.4272487163543701,
-      "learning_rate": 9.793065326633166e-05,
-      "loss": 6.3611,
-      "step": 2560
-    },
-    {
-      "epoch": 1.335593220338983,
-      "grad_norm": 1.4717625379562378,
-      "learning_rate": 9.792964824120604e-05,
-      "loss": 6.0026,
-      "step": 2561
-    },
-    {
-      "epoch": 1.3361147327249023,
-      "grad_norm": 1.599033236503601,
-      "learning_rate": 9.79286432160804e-05,
-      "loss": 6.1974,
-      "step": 2562
-    },
-    {
-      "epoch": 1.3366362451108214,
-      "grad_norm": 1.5725860595703125,
-      "learning_rate": 9.792763819095477e-05,
-      "loss": 6.3259,
-      "step": 2563
-    },
-    {
-      "epoch": 1.3371577574967406,
-      "grad_norm": 2.6072471141815186,
-      "learning_rate": 9.792663316582914e-05,
-      "loss": 4.7133,
-      "step": 2564
-    },
-    {
-      "epoch": 1.3376792698826598,
-      "grad_norm": 2.007253408432007,
-      "learning_rate": 9.792562814070352e-05,
-      "loss": 5.3914,
-      "step": 2565
-    },
-    {
-      "epoch": 1.3382007822685789,
-      "grad_norm": 1.5861899852752686,
-      "learning_rate": 9.79246231155779e-05,
-      "loss": 6.2114,
-      "step": 2566
-    },
-    {
-      "epoch": 1.338722294654498,
-      "grad_norm": 1.6070928573608398,
-      "learning_rate": 9.792361809045226e-05,
-      "loss": 6.3889,
-      "step": 2567
-    },
-    {
-      "epoch": 1.3392438070404173,
-      "grad_norm": 1.71126127243042,
-      "learning_rate": 9.792261306532664e-05,
-      "loss": 6.2217,
-      "step": 2568
-    },
-    {
-      "epoch": 1.3397653194263364,
-      "grad_norm": 1.8321564197540283,
-      "learning_rate": 9.7921608040201e-05,
-      "loss": 5.0663,
-      "step": 2569
-    },
-    {
-      "epoch": 1.3402868318122556,
-      "grad_norm": 1.8267775774002075,
-      "learning_rate": 9.792060301507538e-05,
-      "loss": 5.8484,
-      "step": 2570
-    },
-    {
-      "epoch": 1.3408083441981746,
-      "grad_norm": 1.884848713874817,
-      "learning_rate": 9.791959798994975e-05,
-      "loss": 6.1498,
-      "step": 2571
-    },
-    {
-      "epoch": 1.3413298565840939,
-      "grad_norm": 1.9376534223556519,
-      "learning_rate": 9.791859296482413e-05,
-      "loss": 5.8715,
-      "step": 2572
-    },
-    {
-      "epoch": 1.3418513689700131,
-      "grad_norm": 1.5505872964859009,
-      "learning_rate": 9.791758793969849e-05,
-      "loss": 5.8914,
-      "step": 2573
-    },
-    {
-      "epoch": 1.3423728813559321,
-      "grad_norm": 1.549850583076477,
-      "learning_rate": 9.791658291457287e-05,
-      "loss": 5.8212,
-      "step": 2574
-    },
-    {
-      "epoch": 1.3428943937418514,
-      "grad_norm": 1.6205334663391113,
-      "learning_rate": 9.791557788944725e-05,
-      "loss": 6.0149,
-      "step": 2575
-    },
-    {
-      "epoch": 1.3434159061277704,
-      "grad_norm": 1.5993939638137817,
-      "learning_rate": 9.791457286432161e-05,
-      "loss": 5.9101,
-      "step": 2576
-    },
-    {
-      "epoch": 1.3439374185136896,
-      "grad_norm": 1.6862393617630005,
-      "learning_rate": 9.791356783919599e-05,
-      "loss": 5.8717,
-      "step": 2577
-    },
-    {
-      "epoch": 1.3444589308996089,
-      "grad_norm": 1.8221981525421143,
-      "learning_rate": 9.791256281407035e-05,
-      "loss": 5.9455,
-      "step": 2578
-    },
-    {
-      "epoch": 1.344980443285528,
-      "grad_norm": 1.7727500200271606,
-      "learning_rate": 9.791155778894473e-05,
-      "loss": 6.0823,
-      "step": 2579
-    },
-    {
-      "epoch": 1.3455019556714471,
-      "grad_norm": 1.5663912296295166,
-      "learning_rate": 9.79105527638191e-05,
-      "loss": 5.9271,
-      "step": 2580
-    },
-    {
-      "epoch": 1.3460234680573664,
-      "grad_norm": 1.5444380044937134,
-      "learning_rate": 9.790954773869347e-05,
-      "loss": 6.2839,
-      "step": 2581
-    },
-    {
-      "epoch": 1.3465449804432854,
-      "grad_norm": 1.6440356969833374,
-      "learning_rate": 9.790854271356784e-05,
-      "loss": 5.8643,
-      "step": 2582
-    },
-    {
-      "epoch": 1.3470664928292047,
-      "grad_norm": 1.5112228393554688,
-      "learning_rate": 9.790753768844221e-05,
-      "loss": 6.2394,
-      "step": 2583
-    },
-    {
-      "epoch": 1.347588005215124,
-      "grad_norm": 1.6379401683807373,
-      "learning_rate": 9.790653266331658e-05,
-      "loss": 6.1044,
-      "step": 2584
-    },
-    {
-      "epoch": 1.348109517601043,
-      "grad_norm": 2.836538791656494,
-      "learning_rate": 9.790552763819096e-05,
-      "loss": 5.6336,
-      "step": 2585
-    },
-    {
-      "epoch": 1.3486310299869622,
-      "grad_norm": 1.7374294996261597,
-      "learning_rate": 9.790452261306533e-05,
-      "loss": 5.8256,
-      "step": 2586
-    },
-    {
-      "epoch": 1.3491525423728814,
-      "grad_norm": 1.745182991027832,
-      "learning_rate": 9.790351758793971e-05,
-      "loss": 6.1143,
-      "step": 2587
-    },
-    {
-      "epoch": 1.3496740547588004,
-      "grad_norm": 1.5580577850341797,
-      "learning_rate": 9.790251256281408e-05,
-      "loss": 5.5485,
-      "step": 2588
-    },
-    {
-      "epoch": 1.3501955671447197,
-      "grad_norm": 1.682558298110962,
-      "learning_rate": 9.790150753768844e-05,
-      "loss": 5.9472,
-      "step": 2589
-    },
-    {
-      "epoch": 1.350717079530639,
-      "grad_norm": 1.5831084251403809,
-      "learning_rate": 9.790050251256282e-05,
-      "loss": 5.9733,
-      "step": 2590
-    },
-    {
-      "epoch": 1.351238591916558,
-      "grad_norm": 1.820956826210022,
-      "learning_rate": 9.789949748743718e-05,
-      "loss": 5.5316,
-      "step": 2591
-    },
-    {
-      "epoch": 1.3517601043024772,
-      "grad_norm": 1.6184927225112915,
-      "learning_rate": 9.789849246231156e-05,
-      "loss": 5.986,
-      "step": 2592
-    },
-    {
-      "epoch": 1.3522816166883964,
-      "grad_norm": 1.6540985107421875,
-      "learning_rate": 9.789748743718593e-05,
-      "loss": 6.3191,
-      "step": 2593
-    },
-    {
-      "epoch": 1.3528031290743154,
-      "grad_norm": 1.6854184865951538,
-      "learning_rate": 9.78964824120603e-05,
-      "loss": 6.4295,
-      "step": 2594
-    },
-    {
-      "epoch": 1.3533246414602347,
-      "grad_norm": 1.6246769428253174,
-      "learning_rate": 9.789547738693468e-05,
-      "loss": 6.2036,
-      "step": 2595
-    },
-    {
-      "epoch": 1.353846153846154,
-      "grad_norm": 1.6471805572509766,
-      "learning_rate": 9.789447236180906e-05,
-      "loss": 6.0365,
-      "step": 2596
-    },
-    {
-      "epoch": 1.354367666232073,
-      "grad_norm": 1.5774614810943604,
-      "learning_rate": 9.789346733668342e-05,
-      "loss": 5.7276,
-      "step": 2597
-    },
-    {
-      "epoch": 1.3548891786179922,
-      "grad_norm": 1.7998982667922974,
-      "learning_rate": 9.78924623115578e-05,
-      "loss": 5.8563,
-      "step": 2598
-    },
-    {
-      "epoch": 1.3554106910039114,
-      "grad_norm": 1.6175079345703125,
-      "learning_rate": 9.789145728643217e-05,
-      "loss": 6.2784,
-      "step": 2599
-    },
-    {
-      "epoch": 1.3559322033898304,
-      "grad_norm": 1.7844960689544678,
-      "learning_rate": 9.789045226130654e-05,
-      "loss": 5.6735,
-      "step": 2600
-    },
-    {
-      "epoch": 1.3564537157757497,
-      "grad_norm": 1.676839828491211,
-      "learning_rate": 9.788944723618091e-05,
-      "loss": 5.9944,
-      "step": 2601
-    },
-    {
-      "epoch": 1.356975228161669,
-      "grad_norm": 1.8011897802352905,
-      "learning_rate": 9.788844221105527e-05,
-      "loss": 6.2693,
-      "step": 2602
-    },
-    {
-      "epoch": 1.357496740547588,
-      "grad_norm": 1.7827959060668945,
-      "learning_rate": 9.788743718592965e-05,
-      "loss": 4.96,
-      "step": 2603
-    },
-    {
-      "epoch": 1.3580182529335072,
-      "grad_norm": 1.8489748239517212,
-      "learning_rate": 9.788643216080401e-05,
-      "loss": 5.6311,
-      "step": 2604
-    },
-    {
-      "epoch": 1.3585397653194264,
-      "grad_norm": 1.7378653287887573,
-      "learning_rate": 9.788542713567839e-05,
-      "loss": 6.0367,
-      "step": 2605
-    },
-    {
-      "epoch": 1.3590612777053455,
-      "grad_norm": 1.6756761074066162,
-      "learning_rate": 9.788442211055277e-05,
-      "loss": 5.8726,
-      "step": 2606
-    },
-    {
-      "epoch": 1.3595827900912647,
-      "grad_norm": 1.737670660018921,
-      "learning_rate": 9.788341708542715e-05,
-      "loss": 5.6778,
-      "step": 2607
-    },
-    {
-      "epoch": 1.360104302477184,
-      "grad_norm": 1.5372390747070312,
-      "learning_rate": 9.788241206030151e-05,
-      "loss": 6.3004,
-      "step": 2608
-    },
-    {
-      "epoch": 1.360625814863103,
-      "grad_norm": 1.6686307191848755,
-      "learning_rate": 9.788140703517589e-05,
-      "loss": 5.7527,
-      "step": 2609
-    },
-    {
-      "epoch": 1.3611473272490222,
-      "grad_norm": 1.5270382165908813,
-      "learning_rate": 9.788040201005025e-05,
-      "loss": 6.2194,
-      "step": 2610
-    },
-    {
-      "epoch": 1.3616688396349415,
-      "grad_norm": 1.7912288904190063,
-      "learning_rate": 9.787939698492463e-05,
-      "loss": 5.5666,
-      "step": 2611
-    },
-    {
-      "epoch": 1.3621903520208605,
-      "grad_norm": 1.9645494222640991,
-      "learning_rate": 9.7878391959799e-05,
-      "loss": 5.7711,
-      "step": 2612
-    },
-    {
-      "epoch": 1.3627118644067797,
-      "grad_norm": 2.0565121173858643,
-      "learning_rate": 9.787738693467337e-05,
-      "loss": 5.6659,
-      "step": 2613
-    },
-    {
-      "epoch": 1.363233376792699,
-      "grad_norm": 2.1548197269439697,
-      "learning_rate": 9.787638190954774e-05,
-      "loss": 5.1618,
-      "step": 2614
-    },
-    {
-      "epoch": 1.363754889178618,
-      "grad_norm": 2.06243634223938,
-      "learning_rate": 9.787537688442212e-05,
-      "loss": 6.034,
-      "step": 2615
-    },
-    {
-      "epoch": 1.3642764015645372,
-      "grad_norm": 1.6652909517288208,
-      "learning_rate": 9.78743718592965e-05,
-      "loss": 6.0894,
-      "step": 2616
-    },
-    {
-      "epoch": 1.3647979139504562,
-      "grad_norm": 1.676997184753418,
-      "learning_rate": 9.787336683417086e-05,
-      "loss": 6.19,
-      "step": 2617
-    },
-    {
-      "epoch": 1.3653194263363755,
-      "grad_norm": 2.0292418003082275,
-      "learning_rate": 9.787236180904524e-05,
-      "loss": 5.9899,
-      "step": 2618
-    },
-    {
-      "epoch": 1.3658409387222947,
-      "grad_norm": 1.7349538803100586,
-      "learning_rate": 9.78713567839196e-05,
-      "loss": 5.9425,
-      "step": 2619
-    },
-    {
-      "epoch": 1.3663624511082137,
-      "grad_norm": 1.5623798370361328,
-      "learning_rate": 9.787035175879398e-05,
-      "loss": 6.1844,
-      "step": 2620
-    },
-    {
-      "epoch": 1.366883963494133,
-      "grad_norm": 1.6435014009475708,
-      "learning_rate": 9.786934673366834e-05,
-      "loss": 6.3674,
-      "step": 2621
-    },
-    {
-      "epoch": 1.367405475880052,
-      "grad_norm": 1.6038990020751953,
-      "learning_rate": 9.786834170854272e-05,
-      "loss": 5.898,
-      "step": 2622
-    },
-    {
-      "epoch": 1.3679269882659713,
-      "grad_norm": 1.5299803018569946,
-      "learning_rate": 9.786733668341708e-05,
-      "loss": 6.1502,
-      "step": 2623
-    },
-    {
-      "epoch": 1.3684485006518905,
-      "grad_norm": 1.4204024076461792,
-      "learning_rate": 9.786633165829146e-05,
-      "loss": 6.3264,
-      "step": 2624
-    },
-    {
-      "epoch": 1.3689700130378095,
-      "grad_norm": 1.6097935438156128,
-      "learning_rate": 9.786532663316583e-05,
-      "loss": 6.056,
-      "step": 2625
-    },
-    {
-      "epoch": 1.3694915254237288,
-      "grad_norm": 1.7671841382980347,
-      "learning_rate": 9.78643216080402e-05,
-      "loss": 6.0276,
-      "step": 2626
-    },
-    {
-      "epoch": 1.370013037809648,
-      "grad_norm": 1.5279979705810547,
-      "learning_rate": 9.786331658291458e-05,
-      "loss": 6.2113,
-      "step": 2627
-    },
-    {
-      "epoch": 1.370534550195567,
-      "grad_norm": 1.5320380926132202,
-      "learning_rate": 9.786231155778896e-05,
-      "loss": 5.7059,
-      "step": 2628
-    },
-    {
-      "epoch": 1.3710560625814863,
-      "grad_norm": 1.603962779045105,
-      "learning_rate": 9.786130653266332e-05,
-      "loss": 6.2179,
-      "step": 2629
-    },
-    {
-      "epoch": 1.3715775749674055,
-      "grad_norm": 1.5197679996490479,
-      "learning_rate": 9.786030150753769e-05,
-      "loss": 6.0437,
-      "step": 2630
-    },
-    {
-      "epoch": 1.3720990873533245,
-      "grad_norm": 1.4926389455795288,
-      "learning_rate": 9.785929648241207e-05,
-      "loss": 6.2209,
-      "step": 2631
-    },
-    {
-      "epoch": 1.3726205997392438,
-      "grad_norm": 1.773118019104004,
-      "learning_rate": 9.785829145728643e-05,
-      "loss": 5.7632,
-      "step": 2632
-    },
-    {
-      "epoch": 1.373142112125163,
-      "grad_norm": 1.9592387676239014,
-      "learning_rate": 9.785728643216081e-05,
-      "loss": 6.2943,
-      "step": 2633
-    },
-    {
-      "epoch": 1.373663624511082,
-      "grad_norm": 1.56898832321167,
-      "learning_rate": 9.785628140703517e-05,
-      "loss": 6.4302,
-      "step": 2634
-    },
-    {
-      "epoch": 1.3741851368970013,
-      "grad_norm": 1.6073887348175049,
-      "learning_rate": 9.785527638190955e-05,
-      "loss": 5.8032,
-      "step": 2635
-    },
-    {
-      "epoch": 1.3747066492829205,
-      "grad_norm": 1.637276291847229,
-      "learning_rate": 9.785427135678393e-05,
-      "loss": 6.316,
-      "step": 2636
-    },
-    {
-      "epoch": 1.3752281616688395,
-      "grad_norm": 1.6882108449935913,
-      "learning_rate": 9.78532663316583e-05,
-      "loss": 5.9569,
-      "step": 2637
-    },
-    {
-      "epoch": 1.3757496740547588,
-      "grad_norm": 1.5448179244995117,
-      "learning_rate": 9.785226130653267e-05,
-      "loss": 6.1033,
-      "step": 2638
-    },
-    {
-      "epoch": 1.376271186440678,
-      "grad_norm": 1.5533088445663452,
-      "learning_rate": 9.785125628140705e-05,
-      "loss": 6.2964,
-      "step": 2639
-    },
-    {
-      "epoch": 1.376792698826597,
-      "grad_norm": 1.6889455318450928,
-      "learning_rate": 9.785025125628141e-05,
-      "loss": 6.0242,
-      "step": 2640
-    },
-    {
-      "epoch": 1.3773142112125163,
-      "grad_norm": 1.469394564628601,
-      "learning_rate": 9.784924623115579e-05,
-      "loss": 6.2216,
-      "step": 2641
-    },
-    {
-      "epoch": 1.3778357235984355,
-      "grad_norm": 1.7211103439331055,
-      "learning_rate": 9.784824120603015e-05,
-      "loss": 6.2173,
-      "step": 2642
-    },
-    {
-      "epoch": 1.3783572359843546,
-      "grad_norm": 1.776984691619873,
-      "learning_rate": 9.784723618090452e-05,
-      "loss": 5.8262,
-      "step": 2643
-    },
-    {
-      "epoch": 1.3788787483702738,
-      "grad_norm": 1.743001103401184,
-      "learning_rate": 9.78462311557789e-05,
-      "loss": 5.6768,
-      "step": 2644
-    },
-    {
-      "epoch": 1.379400260756193,
-      "grad_norm": 1.719679832458496,
-      "learning_rate": 9.784522613065326e-05,
-      "loss": 5.7834,
-      "step": 2645
-    },
-    {
-      "epoch": 1.379921773142112,
-      "grad_norm": 1.7682908773422241,
-      "learning_rate": 9.784422110552764e-05,
-      "loss": 5.7643,
-      "step": 2646
-    },
-    {
-      "epoch": 1.3804432855280313,
-      "grad_norm": 1.6111764907836914,
-      "learning_rate": 9.784321608040202e-05,
-      "loss": 5.8891,
-      "step": 2647
-    },
-    {
-      "epoch": 1.3809647979139505,
-      "grad_norm": 1.6279090642929077,
-      "learning_rate": 9.78422110552764e-05,
-      "loss": 6.5447,
-      "step": 2648
-    },
-    {
-      "epoch": 1.3814863102998696,
-      "grad_norm": 1.7054316997528076,
-      "learning_rate": 9.784120603015076e-05,
-      "loss": 5.9229,
-      "step": 2649
-    },
-    {
-      "epoch": 1.3820078226857888,
-      "grad_norm": 1.66355562210083,
-      "learning_rate": 9.784020100502514e-05,
-      "loss": 6.3505,
-      "step": 2650
-    },
-    {
-      "epoch": 1.382529335071708,
-      "grad_norm": 1.6712533235549927,
-      "learning_rate": 9.78391959798995e-05,
-      "loss": 5.9786,
-      "step": 2651
-    },
-    {
-      "epoch": 1.383050847457627,
-      "grad_norm": 1.773200511932373,
-      "learning_rate": 9.783819095477388e-05,
-      "loss": 5.9672,
-      "step": 2652
-    },
-    {
-      "epoch": 1.3835723598435463,
-      "grad_norm": 1.87320077419281,
-      "learning_rate": 9.783718592964824e-05,
-      "loss": 6.2531,
-      "step": 2653
-    },
-    {
-      "epoch": 1.3840938722294656,
-      "grad_norm": 1.7074990272521973,
-      "learning_rate": 9.783618090452262e-05,
-      "loss": 6.1667,
-      "step": 2654
-    },
-    {
-      "epoch": 1.3846153846153846,
-      "grad_norm": 1.6075319051742554,
-      "learning_rate": 9.783517587939698e-05,
-      "loss": 6.4496,
-      "step": 2655
-    },
-    {
-      "epoch": 1.3851368970013038,
-      "grad_norm": 1.4664288759231567,
-      "learning_rate": 9.783417085427136e-05,
-      "loss": 6.3227,
-      "step": 2656
-    },
-    {
-      "epoch": 1.385658409387223,
-      "grad_norm": 1.835418701171875,
-      "learning_rate": 9.783316582914574e-05,
-      "loss": 5.9629,
-      "step": 2657
-    },
-    {
-      "epoch": 1.386179921773142,
-      "grad_norm": 1.5901494026184082,
-      "learning_rate": 9.78321608040201e-05,
-      "loss": 6.282,
-      "step": 2658
-    },
-    {
-      "epoch": 1.3867014341590613,
-      "grad_norm": 1.5904797315597534,
-      "learning_rate": 9.783115577889448e-05,
-      "loss": 6.1755,
-      "step": 2659
-    },
-    {
-      "epoch": 1.3872229465449806,
-      "grad_norm": 1.4861235618591309,
-      "learning_rate": 9.783015075376885e-05,
-      "loss": 6.2742,
-      "step": 2660
-    },
-    {
-      "epoch": 1.3877444589308996,
-      "grad_norm": 1.5851759910583496,
-      "learning_rate": 9.782914572864322e-05,
-      "loss": 5.6079,
-      "step": 2661
-    },
-    {
-      "epoch": 1.3882659713168188,
-      "grad_norm": 1.512555480003357,
-      "learning_rate": 9.782814070351759e-05,
-      "loss": 5.8674,
-      "step": 2662
-    },
-    {
-      "epoch": 1.3887874837027379,
-      "grad_norm": 1.62095046043396,
-      "learning_rate": 9.782713567839197e-05,
-      "loss": 6.1174,
-      "step": 2663
-    },
-    {
-      "epoch": 1.389308996088657,
-      "grad_norm": 1.9584076404571533,
-      "learning_rate": 9.782613065326633e-05,
-      "loss": 6.2369,
-      "step": 2664
-    },
-    {
-      "epoch": 1.3898305084745763,
-      "grad_norm": 1.4809750318527222,
-      "learning_rate": 9.782512562814071e-05,
-      "loss": 6.4352,
-      "step": 2665
-    },
-    {
-      "epoch": 1.3903520208604954,
-      "grad_norm": 1.7495768070220947,
-      "learning_rate": 9.782412060301507e-05,
-      "loss": 6.1479,
-      "step": 2666
-    },
-    {
-      "epoch": 1.3908735332464146,
-      "grad_norm": 1.6603760719299316,
-      "learning_rate": 9.782311557788945e-05,
-      "loss": 6.1712,
-      "step": 2667
-    },
-    {
-      "epoch": 1.3913950456323338,
-      "grad_norm": 1.6336594820022583,
-      "learning_rate": 9.782211055276383e-05,
-      "loss": 6.0849,
-      "step": 2668
-    },
-    {
-      "epoch": 1.3919165580182529,
-      "grad_norm": 1.7293792963027954,
-      "learning_rate": 9.782110552763819e-05,
-      "loss": 6.3651,
-      "step": 2669
-    },
-    {
-      "epoch": 1.3924380704041721,
-      "grad_norm": 1.815760850906372,
-      "learning_rate": 9.782010050251257e-05,
-      "loss": 5.8891,
-      "step": 2670
-    },
-    {
-      "epoch": 1.3929595827900911,
-      "grad_norm": 2.0131959915161133,
-      "learning_rate": 9.781909547738694e-05,
-      "loss": 5.8572,
-      "step": 2671
-    },
-    {
-      "epoch": 1.3934810951760104,
-      "grad_norm": 1.7193174362182617,
-      "learning_rate": 9.781809045226131e-05,
-      "loss": 6.1702,
-      "step": 2672
-    },
-    {
-      "epoch": 1.3940026075619296,
-      "grad_norm": 2.0554697513580322,
-      "learning_rate": 9.781708542713568e-05,
-      "loss": 5.1767,
-      "step": 2673
-    },
-    {
-      "epoch": 1.3945241199478486,
-      "grad_norm": 1.6696985960006714,
-      "learning_rate": 9.781608040201006e-05,
-      "loss": 5.9303,
-      "step": 2674
-    },
-    {
-      "epoch": 1.3950456323337679,
-      "grad_norm": 1.638900876045227,
-      "learning_rate": 9.781507537688442e-05,
-      "loss": 6.1467,
-      "step": 2675
-    },
-    {
-      "epoch": 1.3955671447196871,
-      "grad_norm": 1.4648538827896118,
-      "learning_rate": 9.78140703517588e-05,
-      "loss": 6.1998,
-      "step": 2676
-    },
-    {
-      "epoch": 1.3960886571056061,
-      "grad_norm": 1.5157066583633423,
-      "learning_rate": 9.781306532663318e-05,
-      "loss": 5.7231,
-      "step": 2677
-    },
-    {
-      "epoch": 1.3966101694915254,
-      "grad_norm": 1.7398536205291748,
-      "learning_rate": 9.781206030150755e-05,
-      "loss": 5.7624,
-      "step": 2678
-    },
-    {
-      "epoch": 1.3971316818774446,
-      "grad_norm": 1.8303221464157104,
-      "learning_rate": 9.781105527638192e-05,
-      "loss": 6.1247,
-      "step": 2679
-    },
-    {
-      "epoch": 1.3976531942633637,
-      "grad_norm": 2.1140778064727783,
-      "learning_rate": 9.78100502512563e-05,
-      "loss": 5.4852,
-      "step": 2680
-    },
-    {
-      "epoch": 1.398174706649283,
-      "grad_norm": 2.1840293407440186,
-      "learning_rate": 9.780904522613066e-05,
-      "loss": 5.8905,
-      "step": 2681
-    },
-    {
-      "epoch": 1.3986962190352021,
-      "grad_norm": 1.8187013864517212,
-      "learning_rate": 9.780804020100502e-05,
-      "loss": 6.2925,
-      "step": 2682
-    },
-    {
-      "epoch": 1.3992177314211212,
-      "grad_norm": 1.5236451625823975,
-      "learning_rate": 9.78070351758794e-05,
-      "loss": 6.1928,
-      "step": 2683
-    },
-    {
-      "epoch": 1.3997392438070404,
-      "grad_norm": 1.9740772247314453,
-      "learning_rate": 9.780603015075377e-05,
-      "loss": 5.9138,
-      "step": 2684
-    },
-    {
-      "epoch": 1.4002607561929596,
-      "grad_norm": 1.7422348260879517,
-      "learning_rate": 9.780502512562814e-05,
-      "loss": 5.9113,
-      "step": 2685
-    },
-    {
-      "epoch": 1.4007822685788787,
-      "grad_norm": 1.651078462600708,
-      "learning_rate": 9.780402010050251e-05,
-      "loss": 6.2694,
-      "step": 2686
-    },
-    {
-      "epoch": 1.401303780964798,
-      "grad_norm": 1.6408063173294067,
-      "learning_rate": 9.780301507537689e-05,
-      "loss": 5.7764,
-      "step": 2687
-    },
-    {
-      "epoch": 1.4018252933507171,
-      "grad_norm": 1.7678227424621582,
-      "learning_rate": 9.780201005025126e-05,
-      "loss": 6.1006,
-      "step": 2688
-    },
-    {
-      "epoch": 1.4023468057366362,
-      "grad_norm": 1.836049199104309,
-      "learning_rate": 9.780100502512564e-05,
-      "loss": 5.6566,
-      "step": 2689
-    },
-    {
-      "epoch": 1.4028683181225554,
-      "grad_norm": 1.9151856899261475,
-      "learning_rate": 9.78e-05,
-      "loss": 6.0176,
-      "step": 2690
-    },
-    {
-      "epoch": 1.4033898305084747,
-      "grad_norm": 1.669245719909668,
-      "learning_rate": 9.779899497487438e-05,
-      "loss": 5.9382,
-      "step": 2691
-    },
-    {
-      "epoch": 1.4039113428943937,
-      "grad_norm": 1.576112985610962,
-      "learning_rate": 9.779798994974875e-05,
-      "loss": 6.1099,
-      "step": 2692
-    },
-    {
-      "epoch": 1.404432855280313,
-      "grad_norm": 1.7071670293807983,
-      "learning_rate": 9.779698492462313e-05,
-      "loss": 6.1505,
-      "step": 2693
-    },
-    {
-      "epoch": 1.4049543676662322,
-      "grad_norm": 1.6029512882232666,
-      "learning_rate": 9.779597989949749e-05,
-      "loss": 6.0897,
-      "step": 2694
-    },
-    {
-      "epoch": 1.4054758800521512,
-      "grad_norm": 1.475574016571045,
-      "learning_rate": 9.779497487437185e-05,
-      "loss": 6.3107,
-      "step": 2695
-    },
-    {
-      "epoch": 1.4059973924380704,
-      "grad_norm": 1.6027474403381348,
-      "learning_rate": 9.779396984924623e-05,
-      "loss": 6.0719,
-      "step": 2696
-    },
-    {
-      "epoch": 1.4065189048239897,
-      "grad_norm": 1.8850823640823364,
-      "learning_rate": 9.779296482412061e-05,
-      "loss": 5.3675,
-      "step": 2697
-    },
-    {
-      "epoch": 1.4070404172099087,
-      "grad_norm": 1.5701134204864502,
-      "learning_rate": 9.779195979899499e-05,
-      "loss": 5.8423,
-      "step": 2698
-    },
-    {
-      "epoch": 1.407561929595828,
-      "grad_norm": 1.754403829574585,
-      "learning_rate": 9.779095477386935e-05,
-      "loss": 5.6602,
-      "step": 2699
-    },
-    {
-      "epoch": 1.4080834419817472,
-      "grad_norm": 1.9073467254638672,
-      "learning_rate": 9.778994974874373e-05,
-      "loss": 5.7155,
-      "step": 2700
-    },
-    {
-      "epoch": 1.4086049543676662,
-      "grad_norm": 1.5590382814407349,
-      "learning_rate": 9.77889447236181e-05,
-      "loss": 6.246,
-      "step": 2701
-    },
-    {
-      "epoch": 1.4091264667535854,
-      "grad_norm": 1.701776385307312,
-      "learning_rate": 9.778793969849247e-05,
-      "loss": 5.85,
-      "step": 2702
-    },
-    {
-      "epoch": 1.4096479791395047,
-      "grad_norm": 1.4830223321914673,
-      "learning_rate": 9.778693467336684e-05,
-      "loss": 6.0624,
-      "step": 2703
-    },
-    {
-      "epoch": 1.4101694915254237,
-      "grad_norm": 1.811763048171997,
-      "learning_rate": 9.778592964824121e-05,
-      "loss": 5.9242,
-      "step": 2704
-    },
-    {
-      "epoch": 1.410691003911343,
-      "grad_norm": 1.4055840969085693,
-      "learning_rate": 9.778492462311558e-05,
-      "loss": 5.6296,
-      "step": 2705
-    },
-    {
-      "epoch": 1.4112125162972622,
-      "grad_norm": 1.5805575847625732,
-      "learning_rate": 9.778391959798996e-05,
-      "loss": 5.8183,
-      "step": 2706
-    },
-    {
-      "epoch": 1.4117340286831812,
-      "grad_norm": 1.620363473892212,
-      "learning_rate": 9.778291457286432e-05,
-      "loss": 6.1067,
-      "step": 2707
-    },
-    {
-      "epoch": 1.4122555410691005,
-      "grad_norm": 1.528887391090393,
-      "learning_rate": 9.77819095477387e-05,
-      "loss": 6.1093,
-      "step": 2708
-    },
-    {
-      "epoch": 1.4127770534550197,
-      "grad_norm": 1.5594453811645508,
-      "learning_rate": 9.778090452261308e-05,
-      "loss": 6.4691,
-      "step": 2709
-    },
-    {
-      "epoch": 1.4132985658409387,
-      "grad_norm": 1.5446897745132446,
-      "learning_rate": 9.777989949748744e-05,
-      "loss": 6.1942,
-      "step": 2710
-    },
-    {
-      "epoch": 1.413820078226858,
-      "grad_norm": 1.3707178831100464,
-      "learning_rate": 9.777889447236182e-05,
-      "loss": 6.1318,
-      "step": 2711
-    },
-    {
-      "epoch": 1.414341590612777,
-      "grad_norm": 1.3916586637496948,
-      "learning_rate": 9.777788944723618e-05,
-      "loss": 5.9678,
-      "step": 2712
-    },
-    {
-      "epoch": 1.4148631029986962,
-      "grad_norm": 1.5228615999221802,
-      "learning_rate": 9.777688442211056e-05,
-      "loss": 6.245,
-      "step": 2713
-    },
-    {
-      "epoch": 1.4153846153846155,
-      "grad_norm": 1.6578292846679688,
-      "learning_rate": 9.777587939698492e-05,
-      "loss": 6.0219,
-      "step": 2714
-    },
-    {
-      "epoch": 1.4159061277705345,
-      "grad_norm": 1.4831353425979614,
-      "learning_rate": 9.77748743718593e-05,
-      "loss": 6.0208,
-      "step": 2715
-    },
-    {
-      "epoch": 1.4164276401564537,
-      "grad_norm": 1.5010249614715576,
-      "learning_rate": 9.777386934673367e-05,
-      "loss": 5.989,
-      "step": 2716
-    },
-    {
-      "epoch": 1.4169491525423727,
-      "grad_norm": 1.4538003206253052,
-      "learning_rate": 9.777286432160804e-05,
-      "loss": 6.2988,
-      "step": 2717
-    },
-    {
-      "epoch": 1.417470664928292,
-      "grad_norm": 1.6703176498413086,
-      "learning_rate": 9.777185929648242e-05,
-      "loss": 5.8872,
-      "step": 2718
-    },
-    {
-      "epoch": 1.4179921773142112,
-      "grad_norm": 1.7749847173690796,
-      "learning_rate": 9.77708542713568e-05,
-      "loss": 5.6526,
-      "step": 2719
-    },
-    {
-      "epoch": 1.4185136897001303,
-      "grad_norm": 1.5599079132080078,
-      "learning_rate": 9.776984924623116e-05,
-      "loss": 6.2287,
-      "step": 2720
-    },
-    {
-      "epoch": 1.4190352020860495,
-      "grad_norm": 1.6760694980621338,
-      "learning_rate": 9.776884422110554e-05,
-      "loss": 5.886,
-      "step": 2721
-    },
-    {
-      "epoch": 1.4195567144719687,
-      "grad_norm": 1.5350369215011597,
-      "learning_rate": 9.77678391959799e-05,
-      "loss": 6.3132,
-      "step": 2722
-    },
-    {
-      "epoch": 1.4200782268578878,
-      "grad_norm": 1.721165418624878,
-      "learning_rate": 9.776683417085427e-05,
-      "loss": 5.8893,
-      "step": 2723
-    },
-    {
-      "epoch": 1.420599739243807,
-      "grad_norm": 1.448067307472229,
-      "learning_rate": 9.776582914572865e-05,
-      "loss": 5.8933,
-      "step": 2724
-    },
-    {
-      "epoch": 1.4211212516297262,
-      "grad_norm": 2.8793842792510986,
-      "learning_rate": 9.776482412060301e-05,
-      "loss": 5.7259,
-      "step": 2725
-    },
-    {
-      "epoch": 1.4216427640156453,
-      "grad_norm": 1.7226959466934204,
-      "learning_rate": 9.776381909547739e-05,
-      "loss": 6.0488,
-      "step": 2726
-    },
-    {
-      "epoch": 1.4221642764015645,
-      "grad_norm": 1.6190671920776367,
-      "learning_rate": 9.776281407035175e-05,
-      "loss": 5.774,
-      "step": 2727
-    },
-    {
-      "epoch": 1.4226857887874838,
-      "grad_norm": 1.7586593627929688,
-      "learning_rate": 9.776180904522613e-05,
-      "loss": 5.5945,
-      "step": 2728
-    },
-    {
-      "epoch": 1.4232073011734028,
-      "grad_norm": 1.689913272857666,
-      "learning_rate": 9.776080402010051e-05,
-      "loss": 6.0817,
-      "step": 2729
-    },
-    {
-      "epoch": 1.423728813559322,
-      "grad_norm": 1.766281247138977,
-      "learning_rate": 9.775979899497489e-05,
-      "loss": 5.992,
-      "step": 2730
-    },
-    {
-      "epoch": 1.4242503259452413,
-      "grad_norm": 1.5891350507736206,
-      "learning_rate": 9.775879396984925e-05,
-      "loss": 6.1043,
-      "step": 2731
-    },
-    {
-      "epoch": 1.4247718383311603,
-      "grad_norm": 1.8605815172195435,
-      "learning_rate": 9.775778894472363e-05,
-      "loss": 5.7457,
-      "step": 2732
-    },
-    {
-      "epoch": 1.4252933507170795,
-      "grad_norm": 1.4682244062423706,
-      "learning_rate": 9.7756783919598e-05,
-      "loss": 6.2523,
-      "step": 2733
-    },
-    {
-      "epoch": 1.4258148631029988,
-      "grad_norm": 1.5217187404632568,
-      "learning_rate": 9.775577889447237e-05,
-      "loss": 5.8937,
-      "step": 2734
-    },
-    {
-      "epoch": 1.4263363754889178,
-      "grad_norm": 1.4801461696624756,
-      "learning_rate": 9.775477386934674e-05,
-      "loss": 6.2666,
-      "step": 2735
-    },
-    {
-      "epoch": 1.426857887874837,
-      "grad_norm": 1.696235179901123,
-      "learning_rate": 9.77537688442211e-05,
-      "loss": 6.2227,
-      "step": 2736
-    },
-    {
-      "epoch": 1.4273794002607563,
-      "grad_norm": 1.5992404222488403,
-      "learning_rate": 9.775276381909548e-05,
-      "loss": 6.2008,
-      "step": 2737
-    },
-    {
-      "epoch": 1.4279009126466753,
-      "grad_norm": 1.5044103860855103,
-      "learning_rate": 9.775175879396984e-05,
-      "loss": 6.2463,
-      "step": 2738
-    },
-    {
-      "epoch": 1.4284224250325945,
-      "grad_norm": 1.8731878995895386,
-      "learning_rate": 9.775075376884422e-05,
-      "loss": 5.9088,
-      "step": 2739
-    },
-    {
-      "epoch": 1.4289439374185138,
-      "grad_norm": 1.772975206375122,
-      "learning_rate": 9.77497487437186e-05,
-      "loss": 5.6106,
-      "step": 2740
-    },
-    {
-      "epoch": 1.4294654498044328,
-      "grad_norm": 1.5398154258728027,
-      "learning_rate": 9.774874371859298e-05,
-      "loss": 5.8087,
-      "step": 2741
-    },
-    {
-      "epoch": 1.429986962190352,
-      "grad_norm": 1.6439192295074463,
-      "learning_rate": 9.774773869346734e-05,
-      "loss": 6.1478,
-      "step": 2742
-    },
-    {
-      "epoch": 1.4305084745762713,
-      "grad_norm": 1.5353777408599854,
-      "learning_rate": 9.774673366834172e-05,
-      "loss": 6.1206,
-      "step": 2743
-    },
-    {
-      "epoch": 1.4310299869621903,
-      "grad_norm": 1.945823311805725,
-      "learning_rate": 9.774572864321608e-05,
-      "loss": 5.619,
-      "step": 2744
-    },
-    {
-      "epoch": 1.4315514993481095,
-      "grad_norm": 1.5973284244537354,
-      "learning_rate": 9.774472361809046e-05,
-      "loss": 5.9759,
-      "step": 2745
-    },
-    {
-      "epoch": 1.4320730117340288,
-      "grad_norm": 1.4522231817245483,
-      "learning_rate": 9.774371859296483e-05,
-      "loss": 6.3957,
-      "step": 2746
-    },
-    {
-      "epoch": 1.4325945241199478,
-      "grad_norm": 1.6614530086517334,
-      "learning_rate": 9.77427135678392e-05,
-      "loss": 6.0201,
-      "step": 2747
-    },
-    {
-      "epoch": 1.433116036505867,
-      "grad_norm": 1.624030351638794,
-      "learning_rate": 9.774170854271357e-05,
-      "loss": 6.4179,
-      "step": 2748
-    },
-    {
-      "epoch": 1.4336375488917863,
-      "grad_norm": 1.7130664587020874,
-      "learning_rate": 9.774070351758795e-05,
-      "loss": 6.2297,
-      "step": 2749
-    },
-    {
-      "epoch": 1.4341590612777053,
-      "grad_norm": 1.7549505233764648,
-      "learning_rate": 9.773969849246232e-05,
-      "loss": 6.1637,
-      "step": 2750
-    },
-    {
-      "epoch": 1.4346805736636246,
-      "grad_norm": 1.5874719619750977,
-      "learning_rate": 9.773869346733669e-05,
-      "loss": 5.8404,
-      "step": 2751
-    },
-    {
-      "epoch": 1.4352020860495438,
-      "grad_norm": 1.5968170166015625,
-      "learning_rate": 9.773768844221106e-05,
-      "loss": 6.0682,
-      "step": 2752
-    },
-    {
-      "epoch": 1.4357235984354628,
-      "grad_norm": 1.7284387350082397,
-      "learning_rate": 9.773668341708543e-05,
-      "loss": 6.0219,
-      "step": 2753
-    },
-    {
-      "epoch": 1.436245110821382,
-      "grad_norm": 1.857333779335022,
-      "learning_rate": 9.773567839195981e-05,
-      "loss": 5.5317,
-      "step": 2754
-    },
-    {
-      "epoch": 1.4367666232073013,
-      "grad_norm": 1.6999911069869995,
-      "learning_rate": 9.773467336683417e-05,
-      "loss": 6.1317,
-      "step": 2755
-    },
-    {
-      "epoch": 1.4372881355932203,
-      "grad_norm": 1.6325955390930176,
-      "learning_rate": 9.773366834170855e-05,
-      "loss": 6.3398,
-      "step": 2756
-    },
-    {
-      "epoch": 1.4378096479791396,
-      "grad_norm": 1.7335489988327026,
-      "learning_rate": 9.773266331658291e-05,
-      "loss": 5.9226,
-      "step": 2757
-    },
-    {
-      "epoch": 1.4383311603650586,
-      "grad_norm": 1.9692493677139282,
-      "learning_rate": 9.773165829145729e-05,
-      "loss": 6.0559,
-      "step": 2758
-    },
-    {
-      "epoch": 1.4388526727509778,
-      "grad_norm": 2.2112321853637695,
-      "learning_rate": 9.773065326633166e-05,
-      "loss": 5.7823,
-      "step": 2759
-    },
-    {
-      "epoch": 1.439374185136897,
-      "grad_norm": 1.7404147386550903,
-      "learning_rate": 9.772964824120603e-05,
-      "loss": 5.2362,
-      "step": 2760
-    },
-    {
-      "epoch": 1.439895697522816,
-      "grad_norm": 1.7193166017532349,
-      "learning_rate": 9.772864321608041e-05,
-      "loss": 6.2647,
-      "step": 2761
-    },
-    {
-      "epoch": 1.4404172099087353,
-      "grad_norm": 1.6993381977081299,
-      "learning_rate": 9.772763819095478e-05,
-      "loss": 5.7926,
-      "step": 2762
-    },
-    {
-      "epoch": 1.4409387222946544,
-      "grad_norm": 1.695243000984192,
-      "learning_rate": 9.772663316582915e-05,
-      "loss": 6.0883,
-      "step": 2763
-    },
-    {
-      "epoch": 1.4414602346805736,
-      "grad_norm": 1.5530554056167603,
-      "learning_rate": 9.772562814070352e-05,
-      "loss": 6.1312,
-      "step": 2764
-    },
-    {
-      "epoch": 1.4419817470664928,
-      "grad_norm": 1.697091817855835,
-      "learning_rate": 9.77246231155779e-05,
-      "loss": 6.063,
-      "step": 2765
-    },
-    {
-      "epoch": 1.4425032594524119,
-      "grad_norm": 1.5669046640396118,
-      "learning_rate": 9.772361809045226e-05,
-      "loss": 5.9498,
-      "step": 2766
-    },
-    {
-      "epoch": 1.443024771838331,
-      "grad_norm": 1.5745625495910645,
-      "learning_rate": 9.772261306532664e-05,
-      "loss": 6.099,
-      "step": 2767
-    },
-    {
-      "epoch": 1.4435462842242504,
-      "grad_norm": 1.4733214378356934,
-      "learning_rate": 9.7721608040201e-05,
-      "loss": 6.3993,
-      "step": 2768
-    },
-    {
-      "epoch": 1.4440677966101694,
-      "grad_norm": 1.5762883424758911,
-      "learning_rate": 9.772060301507538e-05,
-      "loss": 6.1406,
-      "step": 2769
-    },
-    {
-      "epoch": 1.4445893089960886,
-      "grad_norm": 1.6982038021087646,
-      "learning_rate": 9.771959798994976e-05,
-      "loss": 6.309,
-      "step": 2770
-    },
-    {
-      "epoch": 1.4451108213820079,
-      "grad_norm": 1.6006107330322266,
-      "learning_rate": 9.771859296482414e-05,
-      "loss": 6.1712,
-      "step": 2771
-    },
-    {
-      "epoch": 1.4456323337679269,
-      "grad_norm": 1.9190771579742432,
-      "learning_rate": 9.77175879396985e-05,
-      "loss": 5.6892,
-      "step": 2772
-    },
-    {
-      "epoch": 1.4461538461538461,
-      "grad_norm": 1.6955370903015137,
-      "learning_rate": 9.771658291457288e-05,
-      "loss": 6.2863,
-      "step": 2773
-    },
-    {
-      "epoch": 1.4466753585397654,
-      "grad_norm": 1.7755554914474487,
-      "learning_rate": 9.771557788944724e-05,
-      "loss": 5.2662,
-      "step": 2774
-    },
-    {
-      "epoch": 1.4471968709256844,
-      "grad_norm": 1.620536208152771,
-      "learning_rate": 9.77145728643216e-05,
-      "loss": 6.131,
-      "step": 2775
-    },
-    {
-      "epoch": 1.4477183833116036,
-      "grad_norm": 1.549432396888733,
-      "learning_rate": 9.771356783919598e-05,
-      "loss": 5.8983,
-      "step": 2776
-    },
-    {
-      "epoch": 1.4482398956975229,
-      "grad_norm": 1.6327165365219116,
-      "learning_rate": 9.771256281407035e-05,
-      "loss": 6.4203,
-      "step": 2777
-    },
-    {
-      "epoch": 1.448761408083442,
-      "grad_norm": 1.7144237756729126,
-      "learning_rate": 9.771155778894473e-05,
-      "loss": 5.8586,
-      "step": 2778
-    },
-    {
-      "epoch": 1.4492829204693611,
-      "grad_norm": 1.787179708480835,
-      "learning_rate": 9.771055276381909e-05,
-      "loss": 6.3244,
-      "step": 2779
-    },
-    {
-      "epoch": 1.4498044328552804,
-      "grad_norm": 1.541170597076416,
-      "learning_rate": 9.770954773869347e-05,
-      "loss": 6.2233,
-      "step": 2780
-    },
-    {
-      "epoch": 1.4503259452411994,
-      "grad_norm": 1.7627884149551392,
-      "learning_rate": 9.770854271356785e-05,
-      "loss": 6.0454,
-      "step": 2781
-    },
-    {
-      "epoch": 1.4508474576271186,
-      "grad_norm": 1.892922043800354,
-      "learning_rate": 9.770753768844222e-05,
-      "loss": 5.5029,
-      "step": 2782
-    },
-    {
-      "epoch": 1.4513689700130379,
-      "grad_norm": 1.7500706911087036,
-      "learning_rate": 9.770653266331659e-05,
-      "loss": 6.0293,
-      "step": 2783
-    },
-    {
-      "epoch": 1.451890482398957,
-      "grad_norm": 1.5444124937057495,
-      "learning_rate": 9.770552763819097e-05,
-      "loss": 6.1176,
-      "step": 2784
-    },
-    {
-      "epoch": 1.4524119947848761,
-      "grad_norm": 1.8232090473175049,
-      "learning_rate": 9.770452261306533e-05,
-      "loss": 5.5497,
-      "step": 2785
-    },
-    {
-      "epoch": 1.4529335071707954,
-      "grad_norm": 2.1292386054992676,
-      "learning_rate": 9.770351758793971e-05,
-      "loss": 5.7434,
-      "step": 2786
-    },
-    {
-      "epoch": 1.4534550195567144,
-      "grad_norm": 1.7782337665557861,
-      "learning_rate": 9.770251256281407e-05,
-      "loss": 6.1511,
-      "step": 2787
-    },
-    {
-      "epoch": 1.4539765319426337,
-      "grad_norm": 1.6669893264770508,
-      "learning_rate": 9.770150753768844e-05,
-      "loss": 5.9931,
-      "step": 2788
-    },
-    {
-      "epoch": 1.454498044328553,
-      "grad_norm": 1.544895052909851,
-      "learning_rate": 9.770050251256281e-05,
-      "loss": 5.963,
-      "step": 2789
-    },
-    {
-      "epoch": 1.455019556714472,
-      "grad_norm": 1.584064245223999,
-      "learning_rate": 9.769949748743719e-05,
-      "loss": 6.0779,
-      "step": 2790
-    },
-    {
-      "epoch": 1.4555410691003912,
-      "grad_norm": 1.780301809310913,
-      "learning_rate": 9.769849246231157e-05,
-      "loss": 6.0508,
-      "step": 2791
-    },
-    {
-      "epoch": 1.4560625814863104,
-      "grad_norm": 1.5685304403305054,
-      "learning_rate": 9.769748743718593e-05,
-      "loss": 5.5051,
-      "step": 2792
-    },
-    {
-      "epoch": 1.4565840938722294,
-      "grad_norm": 1.5297799110412598,
-      "learning_rate": 9.769648241206031e-05,
-      "loss": 6.2241,
-      "step": 2793
-    },
-    {
-      "epoch": 1.4571056062581487,
-      "grad_norm": 1.5676274299621582,
-      "learning_rate": 9.769547738693468e-05,
-      "loss": 5.9569,
-      "step": 2794
-    },
-    {
-      "epoch": 1.457627118644068,
-      "grad_norm": 1.5621898174285889,
-      "learning_rate": 9.769447236180905e-05,
-      "loss": 6.048,
-      "step": 2795
-    },
-    {
-      "epoch": 1.458148631029987,
-      "grad_norm": 1.7106941938400269,
-      "learning_rate": 9.769346733668342e-05,
-      "loss": 5.8046,
-      "step": 2796
-    },
-    {
-      "epoch": 1.4586701434159062,
-      "grad_norm": 1.6666656732559204,
-      "learning_rate": 9.76924623115578e-05,
-      "loss": 5.7507,
-      "step": 2797
-    },
-    {
-      "epoch": 1.4591916558018254,
-      "grad_norm": 1.7873709201812744,
-      "learning_rate": 9.769145728643216e-05,
-      "loss": 5.8235,
-      "step": 2798
-    },
-    {
-      "epoch": 1.4597131681877444,
-      "grad_norm": 1.6920788288116455,
-      "learning_rate": 9.769045226130654e-05,
-      "loss": 6.1177,
-      "step": 2799
-    },
-    {
-      "epoch": 1.4602346805736637,
-      "grad_norm": 1.6523910760879517,
-      "learning_rate": 9.76894472361809e-05,
-      "loss": 6.1985,
-      "step": 2800
-    },
-    {
-      "epoch": 1.460756192959583,
-      "grad_norm": 1.450553297996521,
-      "learning_rate": 9.768844221105528e-05,
-      "loss": 6.2034,
-      "step": 2801
-    },
-    {
-      "epoch": 1.461277705345502,
-      "grad_norm": 1.5712532997131348,
-      "learning_rate": 9.768743718592966e-05,
-      "loss": 5.3465,
-      "step": 2802
-    },
-    {
-      "epoch": 1.4617992177314212,
-      "grad_norm": 1.7631434202194214,
-      "learning_rate": 9.768643216080402e-05,
-      "loss": 5.8792,
-      "step": 2803
-    },
-    {
-      "epoch": 1.4623207301173402,
-      "grad_norm": 1.5004957914352417,
-      "learning_rate": 9.76854271356784e-05,
-      "loss": 6.1505,
-      "step": 2804
-    },
-    {
-      "epoch": 1.4628422425032594,
-      "grad_norm": 1.6706433296203613,
-      "learning_rate": 9.768442211055276e-05,
-      "loss": 6.3158,
-      "step": 2805
-    },
-    {
-      "epoch": 1.4633637548891787,
-      "grad_norm": 1.5519438982009888,
-      "learning_rate": 9.768341708542714e-05,
-      "loss": 6.0472,
-      "step": 2806
-    },
-    {
-      "epoch": 1.4638852672750977,
-      "grad_norm": 1.572006344795227,
-      "learning_rate": 9.76824120603015e-05,
-      "loss": 6.3187,
-      "step": 2807
-    },
-    {
-      "epoch": 1.464406779661017,
-      "grad_norm": 1.6571171283721924,
-      "learning_rate": 9.768140703517588e-05,
-      "loss": 6.2947,
-      "step": 2808
-    },
-    {
-      "epoch": 1.4649282920469362,
-      "grad_norm": 1.80170476436615,
-      "learning_rate": 9.768040201005025e-05,
-      "loss": 5.6707,
-      "step": 2809
-    },
-    {
-      "epoch": 1.4654498044328552,
-      "grad_norm": 1.6589888334274292,
-      "learning_rate": 9.767939698492463e-05,
-      "loss": 5.9271,
-      "step": 2810
-    },
-    {
-      "epoch": 1.4659713168187745,
-      "grad_norm": 1.5281791687011719,
-      "learning_rate": 9.7678391959799e-05,
-      "loss": 6.1576,
-      "step": 2811
-    },
-    {
-      "epoch": 1.4664928292046935,
-      "grad_norm": 1.7295739650726318,
-      "learning_rate": 9.767738693467338e-05,
-      "loss": 6.1883,
-      "step": 2812
-    },
-    {
-      "epoch": 1.4670143415906127,
-      "grad_norm": 1.55623459815979,
-      "learning_rate": 9.767638190954775e-05,
-      "loss": 6.1236,
-      "step": 2813
-    },
-    {
-      "epoch": 1.467535853976532,
-      "grad_norm": 1.4653511047363281,
-      "learning_rate": 9.767537688442212e-05,
-      "loss": 5.6758,
-      "step": 2814
-    },
-    {
-      "epoch": 1.468057366362451,
-      "grad_norm": 1.4706426858901978,
-      "learning_rate": 9.767437185929649e-05,
-      "loss": 6.071,
-      "step": 2815
-    },
-    {
-      "epoch": 1.4685788787483702,
-      "grad_norm": 1.8940315246582031,
-      "learning_rate": 9.767336683417085e-05,
-      "loss": 5.4638,
-      "step": 2816
-    },
-    {
-      "epoch": 1.4691003911342895,
-      "grad_norm": 1.5454559326171875,
-      "learning_rate": 9.767236180904523e-05,
-      "loss": 6.3419,
-      "step": 2817
-    },
-    {
-      "epoch": 1.4696219035202085,
-      "grad_norm": 1.528501033782959,
-      "learning_rate": 9.76713567839196e-05,
-      "loss": 6.2238,
-      "step": 2818
-    },
-    {
-      "epoch": 1.4701434159061277,
-      "grad_norm": 1.6060926914215088,
-      "learning_rate": 9.767035175879397e-05,
-      "loss": 6.1678,
-      "step": 2819
-    },
-    {
-      "epoch": 1.470664928292047,
-      "grad_norm": 1.8323644399642944,
-      "learning_rate": 9.766934673366834e-05,
-      "loss": 5.7754,
-      "step": 2820
-    },
-    {
-      "epoch": 1.471186440677966,
-      "grad_norm": 1.4659833908081055,
-      "learning_rate": 9.766834170854271e-05,
-      "loss": 6.2661,
-      "step": 2821
-    },
-    {
-      "epoch": 1.4717079530638852,
-      "grad_norm": 1.4832403659820557,
-      "learning_rate": 9.766733668341709e-05,
-      "loss": 6.2244,
-      "step": 2822
-    },
-    {
-      "epoch": 1.4722294654498045,
-      "grad_norm": 1.5771267414093018,
-      "learning_rate": 9.766633165829147e-05,
-      "loss": 6.141,
-      "step": 2823
-    },
-    {
-      "epoch": 1.4727509778357235,
-      "grad_norm": 1.6609662771224976,
-      "learning_rate": 9.766532663316583e-05,
-      "loss": 6.0953,
-      "step": 2824
-    },
-    {
-      "epoch": 1.4732724902216427,
-      "grad_norm": 1.6120318174362183,
-      "learning_rate": 9.766432160804021e-05,
-      "loss": 5.3603,
-      "step": 2825
-    },
-    {
-      "epoch": 1.473794002607562,
-      "grad_norm": 1.627646565437317,
-      "learning_rate": 9.766331658291458e-05,
-      "loss": 5.9943,
-      "step": 2826
-    },
-    {
-      "epoch": 1.474315514993481,
-      "grad_norm": 1.608627200126648,
-      "learning_rate": 9.766231155778895e-05,
-      "loss": 5.4972,
-      "step": 2827
-    },
-    {
-      "epoch": 1.4748370273794003,
-      "grad_norm": 1.4947052001953125,
-      "learning_rate": 9.766130653266332e-05,
-      "loss": 6.052,
-      "step": 2828
-    },
-    {
-      "epoch": 1.4753585397653195,
-      "grad_norm": 1.654911756515503,
-      "learning_rate": 9.766030150753768e-05,
-      "loss": 5.8259,
-      "step": 2829
-    },
-    {
-      "epoch": 1.4758800521512385,
-      "grad_norm": 1.4821324348449707,
-      "learning_rate": 9.765929648241206e-05,
-      "loss": 6.0947,
-      "step": 2830
-    },
-    {
-      "epoch": 1.4764015645371578,
-      "grad_norm": 1.5970264673233032,
-      "learning_rate": 9.765829145728644e-05,
-      "loss": 5.8676,
-      "step": 2831
-    },
-    {
-      "epoch": 1.476923076923077,
-      "grad_norm": 1.470488429069519,
-      "learning_rate": 9.765728643216082e-05,
-      "loss": 6.0662,
-      "step": 2832
-    },
-    {
-      "epoch": 1.477444589308996,
-      "grad_norm": 1.6056243181228638,
-      "learning_rate": 9.765628140703518e-05,
-      "loss": 6.0034,
-      "step": 2833
-    },
-    {
-      "epoch": 1.4779661016949153,
-      "grad_norm": 1.5965206623077393,
-      "learning_rate": 9.765527638190956e-05,
-      "loss": 6.1443,
-      "step": 2834
-    },
-    {
-      "epoch": 1.4784876140808345,
-      "grad_norm": 1.5480287075042725,
-      "learning_rate": 9.765427135678392e-05,
-      "loss": 6.2511,
-      "step": 2835
-    },
-    {
-      "epoch": 1.4790091264667535,
-      "grad_norm": 1.6252895593643188,
-      "learning_rate": 9.76532663316583e-05,
-      "loss": 5.8563,
-      "step": 2836
-    },
-    {
-      "epoch": 1.4795306388526728,
-      "grad_norm": 1.6706434488296509,
-      "learning_rate": 9.765226130653267e-05,
-      "loss": 5.8372,
-      "step": 2837
-    },
-    {
-      "epoch": 1.480052151238592,
-      "grad_norm": 1.8390557765960693,
-      "learning_rate": 9.765125628140704e-05,
-      "loss": 6.1495,
-      "step": 2838
-    },
-    {
-      "epoch": 1.480573663624511,
-      "grad_norm": 1.649429202079773,
-      "learning_rate": 9.765025125628141e-05,
-      "loss": 6.1639,
-      "step": 2839
-    },
-    {
-      "epoch": 1.4810951760104303,
-      "grad_norm": 1.5332372188568115,
-      "learning_rate": 9.764924623115579e-05,
-      "loss": 6.1565,
-      "step": 2840
-    },
-    {
-      "epoch": 1.4816166883963495,
-      "grad_norm": 1.7449291944503784,
-      "learning_rate": 9.764824120603015e-05,
-      "loss": 5.7117,
-      "step": 2841
-    },
-    {
-      "epoch": 1.4821382007822685,
-      "grad_norm": 1.5824756622314453,
-      "learning_rate": 9.764723618090453e-05,
-      "loss": 6.0968,
-      "step": 2842
-    },
-    {
-      "epoch": 1.4826597131681878,
-      "grad_norm": 1.5848503112792969,
-      "learning_rate": 9.76462311557789e-05,
-      "loss": 6.3909,
-      "step": 2843
-    },
-    {
-      "epoch": 1.483181225554107,
-      "grad_norm": 1.5621473789215088,
-      "learning_rate": 9.764522613065327e-05,
-      "loss": 6.2478,
-      "step": 2844
-    },
-    {
-      "epoch": 1.483702737940026,
-      "grad_norm": 2.170722007751465,
-      "learning_rate": 9.764422110552765e-05,
-      "loss": 5.9168,
-      "step": 2845
-    },
-    {
-      "epoch": 1.4842242503259453,
-      "grad_norm": 1.6070153713226318,
-      "learning_rate": 9.764321608040201e-05,
-      "loss": 6.374,
-      "step": 2846
-    },
-    {
-      "epoch": 1.4847457627118645,
-      "grad_norm": 1.587161898612976,
-      "learning_rate": 9.764221105527639e-05,
-      "loss": 6.1393,
-      "step": 2847
-    },
-    {
-      "epoch": 1.4852672750977836,
-      "grad_norm": 1.6479389667510986,
-      "learning_rate": 9.764120603015075e-05,
-      "loss": 6.1762,
-      "step": 2848
-    },
-    {
-      "epoch": 1.4857887874837028,
-      "grad_norm": 1.7014250755310059,
-      "learning_rate": 9.764020100502513e-05,
-      "loss": 5.653,
-      "step": 2849
-    },
-    {
-      "epoch": 1.4863102998696218,
-      "grad_norm": 2.1229970455169678,
-      "learning_rate": 9.76391959798995e-05,
-      "loss": 5.3146,
-      "step": 2850
-    },
-    {
-      "epoch": 1.486831812255541,
-      "grad_norm": 1.7179540395736694,
-      "learning_rate": 9.763819095477387e-05,
-      "loss": 6.4392,
-      "step": 2851
-    },
-    {
-      "epoch": 1.4873533246414603,
-      "grad_norm": 1.461183786392212,
-      "learning_rate": 9.763718592964825e-05,
-      "loss": 6.4847,
-      "step": 2852
-    },
-    {
-      "epoch": 1.4878748370273793,
-      "grad_norm": 1.5658162832260132,
-      "learning_rate": 9.763618090452263e-05,
-      "loss": 6.114,
-      "step": 2853
-    },
-    {
-      "epoch": 1.4883963494132986,
-      "grad_norm": 2.1612656116485596,
-      "learning_rate": 9.7635175879397e-05,
-      "loss": 5.9358,
-      "step": 2854
-    },
-    {
-      "epoch": 1.4889178617992178,
-      "grad_norm": 1.7783305644989014,
-      "learning_rate": 9.763417085427136e-05,
-      "loss": 5.9594,
-      "step": 2855
-    },
-    {
-      "epoch": 1.4894393741851368,
-      "grad_norm": 2.0530593395233154,
-      "learning_rate": 9.763316582914574e-05,
-      "loss": 5.9313,
-      "step": 2856
-    },
-    {
-      "epoch": 1.489960886571056,
-      "grad_norm": 1.7588746547698975,
-      "learning_rate": 9.76321608040201e-05,
-      "loss": 5.927,
-      "step": 2857
-    },
-    {
-      "epoch": 1.490482398956975,
-      "grad_norm": 1.7404680252075195,
-      "learning_rate": 9.763115577889448e-05,
-      "loss": 6.4275,
-      "step": 2858
-    },
-    {
-      "epoch": 1.4910039113428943,
-      "grad_norm": 1.607073187828064,
-      "learning_rate": 9.763015075376884e-05,
-      "loss": 6.0119,
-      "step": 2859
-    },
-    {
-      "epoch": 1.4915254237288136,
-      "grad_norm": 1.7869770526885986,
-      "learning_rate": 9.762914572864322e-05,
-      "loss": 5.8664,
-      "step": 2860
-    },
-    {
-      "epoch": 1.4920469361147326,
-      "grad_norm": 1.5147517919540405,
-      "learning_rate": 9.762814070351758e-05,
-      "loss": 6.2189,
-      "step": 2861
-    },
-    {
-      "epoch": 1.4925684485006518,
-      "grad_norm": 1.593854546546936,
-      "learning_rate": 9.762713567839196e-05,
-      "loss": 5.8206,
-      "step": 2862
-    },
-    {
-      "epoch": 1.493089960886571,
-      "grad_norm": 1.5516552925109863,
-      "learning_rate": 9.762613065326634e-05,
-      "loss": 5.8553,
-      "step": 2863
-    },
-    {
-      "epoch": 1.49361147327249,
-      "grad_norm": 1.632185935974121,
-      "learning_rate": 9.762512562814072e-05,
-      "loss": 5.8157,
-      "step": 2864
-    },
-    {
-      "epoch": 1.4941329856584094,
-      "grad_norm": 1.693359136581421,
-      "learning_rate": 9.762412060301508e-05,
-      "loss": 5.7293,
-      "step": 2865
-    },
-    {
-      "epoch": 1.4946544980443286,
-      "grad_norm": 1.6333162784576416,
-      "learning_rate": 9.762311557788946e-05,
-      "loss": 5.8918,
-      "step": 2866
-    },
-    {
-      "epoch": 1.4951760104302476,
-      "grad_norm": 1.72529137134552,
-      "learning_rate": 9.762211055276382e-05,
-      "loss": 5.5389,
-      "step": 2867
-    },
-    {
-      "epoch": 1.4956975228161669,
-      "grad_norm": 1.593548059463501,
-      "learning_rate": 9.762110552763819e-05,
-      "loss": 6.1077,
-      "step": 2868
-    },
-    {
-      "epoch": 1.496219035202086,
-      "grad_norm": 1.6674294471740723,
-      "learning_rate": 9.762010050251257e-05,
-      "loss": 6.1928,
-      "step": 2869
-    },
-    {
-      "epoch": 1.4967405475880051,
-      "grad_norm": 2.0850799083709717,
-      "learning_rate": 9.761909547738693e-05,
-      "loss": 5.5981,
-      "step": 2870
-    },
-    {
-      "epoch": 1.4972620599739244,
-      "grad_norm": 1.980785608291626,
-      "learning_rate": 9.761809045226131e-05,
-      "loss": 5.4568,
-      "step": 2871
-    },
-    {
-      "epoch": 1.4977835723598436,
-      "grad_norm": 1.558742642402649,
-      "learning_rate": 9.761708542713569e-05,
-      "loss": 6.1087,
-      "step": 2872
-    },
-    {
-      "epoch": 1.4983050847457626,
-      "grad_norm": 1.6420050859451294,
-      "learning_rate": 9.761608040201006e-05,
-      "loss": 5.937,
-      "step": 2873
-    },
-    {
-      "epoch": 1.4988265971316819,
-      "grad_norm": 1.5581105947494507,
-      "learning_rate": 9.761507537688443e-05,
-      "loss": 5.9875,
-      "step": 2874
-    },
-    {
-      "epoch": 1.4993481095176011,
-      "grad_norm": 1.6250531673431396,
-      "learning_rate": 9.76140703517588e-05,
-      "loss": 6.2148,
-      "step": 2875
-    },
-    {
-      "epoch": 1.4998696219035201,
-      "grad_norm": 1.7030606269836426,
-      "learning_rate": 9.761306532663317e-05,
-      "loss": 5.8734,
-      "step": 2876
-    },
-    {
-      "epoch": 1.5003911342894394,
-      "grad_norm": 1.5861607789993286,
-      "learning_rate": 9.761206030150755e-05,
-      "loss": 6.0202,
-      "step": 2877
-    },
-    {
-      "epoch": 1.5009126466753586,
-      "grad_norm": 1.554383397102356,
-      "learning_rate": 9.761105527638191e-05,
-      "loss": 6.1231,
-      "step": 2878
-    },
-    {
-      "epoch": 1.5014341590612776,
-      "grad_norm": 1.6653318405151367,
-      "learning_rate": 9.761005025125629e-05,
-      "loss": 5.8554,
-      "step": 2879
-    },
-    {
-      "epoch": 1.5019556714471969,
-      "grad_norm": 1.762291669845581,
-      "learning_rate": 9.760904522613065e-05,
-      "loss": 6.0767,
-      "step": 2880
-    },
-    {
-      "epoch": 1.5024771838331161,
-      "grad_norm": 1.4581154584884644,
-      "learning_rate": 9.760804020100502e-05,
-      "loss": 6.263,
-      "step": 2881
-    },
-    {
-      "epoch": 1.5029986962190351,
-      "grad_norm": 1.7813777923583984,
-      "learning_rate": 9.76070351758794e-05,
-      "loss": 5.5883,
-      "step": 2882
-    },
-    {
-      "epoch": 1.5035202086049544,
-      "grad_norm": 1.5640543699264526,
-      "learning_rate": 9.760603015075377e-05,
-      "loss": 5.9936,
-      "step": 2883
-    },
-    {
-      "epoch": 1.5040417209908736,
-      "grad_norm": 1.7181663513183594,
-      "learning_rate": 9.760502512562815e-05,
-      "loss": 5.5832,
-      "step": 2884
-    },
-    {
-      "epoch": 1.5045632333767927,
-      "grad_norm": 1.6281917095184326,
-      "learning_rate": 9.760402010050252e-05,
-      "loss": 5.8035,
-      "step": 2885
-    },
-    {
-      "epoch": 1.505084745762712,
-      "grad_norm": 1.650791883468628,
-      "learning_rate": 9.76030150753769e-05,
-      "loss": 5.8989,
-      "step": 2886
-    },
-    {
-      "epoch": 1.5056062581486311,
-      "grad_norm": 1.5169929265975952,
-      "learning_rate": 9.760201005025126e-05,
-      "loss": 6.0039,
-      "step": 2887
-    },
-    {
-      "epoch": 1.5061277705345502,
-      "grad_norm": 1.5851532220840454,
-      "learning_rate": 9.760100502512564e-05,
-      "loss": 6.2984,
-      "step": 2888
-    },
-    {
-      "epoch": 1.5066492829204694,
-      "grad_norm": 1.801254153251648,
-      "learning_rate": 9.76e-05,
-      "loss": 5.8432,
-      "step": 2889
-    },
-    {
-      "epoch": 1.5071707953063886,
-      "grad_norm": 2.037017583847046,
-      "learning_rate": 9.759899497487438e-05,
-      "loss": 5.7578,
-      "step": 2890
-    },
-    {
-      "epoch": 1.5076923076923077,
-      "grad_norm": 1.8728010654449463,
-      "learning_rate": 9.759798994974874e-05,
-      "loss": 5.7716,
-      "step": 2891
-    },
-    {
-      "epoch": 1.508213820078227,
-      "grad_norm": 1.9070972204208374,
-      "learning_rate": 9.759698492462312e-05,
-      "loss": 6.0053,
-      "step": 2892
-    },
-    {
-      "epoch": 1.5087353324641462,
-      "grad_norm": 1.8258485794067383,
-      "learning_rate": 9.75959798994975e-05,
-      "loss": 5.7379,
-      "step": 2893
-    },
-    {
-      "epoch": 1.5092568448500652,
-      "grad_norm": 1.6042457818984985,
-      "learning_rate": 9.759497487437188e-05,
-      "loss": 6.3639,
-      "step": 2894
-    },
-    {
-      "epoch": 1.5097783572359842,
-      "grad_norm": 1.659169316291809,
-      "learning_rate": 9.759396984924624e-05,
-      "loss": 5.841,
-      "step": 2895
-    },
-    {
-      "epoch": 1.5102998696219037,
-      "grad_norm": 1.4794549942016602,
-      "learning_rate": 9.75929648241206e-05,
-      "loss": 6.0871,
-      "step": 2896
-    },
-    {
-      "epoch": 1.5108213820078227,
-      "grad_norm": 1.6386024951934814,
-      "learning_rate": 9.759195979899498e-05,
-      "loss": 6.0702,
-      "step": 2897
-    },
-    {
-      "epoch": 1.5113428943937417,
-      "grad_norm": 2.325228214263916,
-      "learning_rate": 9.759095477386935e-05,
-      "loss": 5.6941,
-      "step": 2898
-    },
-    {
-      "epoch": 1.5118644067796612,
-      "grad_norm": 1.6784334182739258,
-      "learning_rate": 9.758994974874372e-05,
-      "loss": 5.9596,
-      "step": 2899
-    },
-    {
-      "epoch": 1.5123859191655802,
-      "grad_norm": 1.5968056917190552,
-      "learning_rate": 9.758894472361809e-05,
-      "loss": 5.8579,
-      "step": 2900
-    },
-    {
-      "epoch": 1.5129074315514992,
-      "grad_norm": 1.658576488494873,
-      "learning_rate": 9.758793969849247e-05,
-      "loss": 6.2242,
-      "step": 2901
-    },
-    {
-      "epoch": 1.5134289439374187,
-      "grad_norm": 1.8618217706680298,
-      "learning_rate": 9.758693467336683e-05,
-      "loss": 5.6366,
-      "step": 2902
-    },
-    {
-      "epoch": 1.5139504563233377,
-      "grad_norm": 1.681307315826416,
-      "learning_rate": 9.758592964824121e-05,
-      "loss": 6.0446,
-      "step": 2903
-    },
-    {
-      "epoch": 1.5144719687092567,
-      "grad_norm": 1.5531169176101685,
-      "learning_rate": 9.758492462311559e-05,
-      "loss": 6.3236,
-      "step": 2904
-    },
-    {
-      "epoch": 1.5149934810951762,
-      "grad_norm": 1.6814061403274536,
-      "learning_rate": 9.758391959798996e-05,
-      "loss": 5.9854,
-      "step": 2905
-    },
-    {
-      "epoch": 1.5155149934810952,
-      "grad_norm": 1.6177403926849365,
-      "learning_rate": 9.758291457286433e-05,
-      "loss": 5.8279,
-      "step": 2906
-    },
-    {
-      "epoch": 1.5160365058670142,
-      "grad_norm": 1.6217435598373413,
-      "learning_rate": 9.75819095477387e-05,
-      "loss": 5.7753,
-      "step": 2907
-    },
-    {
-      "epoch": 1.5165580182529335,
-      "grad_norm": 1.6294753551483154,
-      "learning_rate": 9.758090452261307e-05,
-      "loss": 5.8529,
-      "step": 2908
-    },
-    {
-      "epoch": 1.5170795306388527,
-      "grad_norm": 1.7994980812072754,
-      "learning_rate": 9.757989949748744e-05,
-      "loss": 6.033,
-      "step": 2909
-    },
-    {
-      "epoch": 1.5176010430247717,
-      "grad_norm": 1.6275047063827515,
-      "learning_rate": 9.757889447236181e-05,
-      "loss": 6.2808,
-      "step": 2910
-    },
-    {
-      "epoch": 1.518122555410691,
-      "grad_norm": 1.6710172891616821,
-      "learning_rate": 9.757788944723618e-05,
-      "loss": 5.7843,
-      "step": 2911
-    },
-    {
-      "epoch": 1.5186440677966102,
-      "grad_norm": 1.6221249103546143,
-      "learning_rate": 9.757688442211056e-05,
-      "loss": 6.0414,
-      "step": 2912
-    },
-    {
-      "epoch": 1.5191655801825292,
-      "grad_norm": 1.5152809619903564,
-      "learning_rate": 9.757587939698492e-05,
-      "loss": 6.0427,
-      "step": 2913
-    },
-    {
-      "epoch": 1.5196870925684485,
-      "grad_norm": 1.7017035484313965,
-      "learning_rate": 9.75748743718593e-05,
-      "loss": 5.9066,
-      "step": 2914
-    },
-    {
-      "epoch": 1.5202086049543677,
-      "grad_norm": 1.6341040134429932,
-      "learning_rate": 9.757386934673368e-05,
-      "loss": 5.5702,
-      "step": 2915
-    },
-    {
-      "epoch": 1.5207301173402867,
-      "grad_norm": 1.7104073762893677,
-      "learning_rate": 9.757286432160805e-05,
-      "loss": 5.8711,
-      "step": 2916
-    },
-    {
-      "epoch": 1.521251629726206,
-      "grad_norm": 1.750266194343567,
-      "learning_rate": 9.757185929648242e-05,
-      "loss": 5.5838,
-      "step": 2917
-    },
-    {
-      "epoch": 1.5217731421121252,
-      "grad_norm": 1.659591555595398,
-      "learning_rate": 9.75708542713568e-05,
-      "loss": 5.9391,
-      "step": 2918
-    },
-    {
-      "epoch": 1.5222946544980442,
-      "grad_norm": 1.646873116493225,
-      "learning_rate": 9.756984924623116e-05,
-      "loss": 5.9978,
-      "step": 2919
-    },
-    {
-      "epoch": 1.5228161668839635,
-      "grad_norm": 1.7272626161575317,
-      "learning_rate": 9.756884422110554e-05,
-      "loss": 5.5819,
-      "step": 2920
-    },
-    {
-      "epoch": 1.5233376792698827,
-      "grad_norm": 1.7887064218521118,
-      "learning_rate": 9.75678391959799e-05,
-      "loss": 5.846,
-      "step": 2921
-    },
-    {
-      "epoch": 1.5238591916558017,
-      "grad_norm": 1.9516392946243286,
-      "learning_rate": 9.756683417085427e-05,
-      "loss": 6.3186,
-      "step": 2922
-    },
-    {
-      "epoch": 1.524380704041721,
-      "grad_norm": 2.01957368850708,
-      "learning_rate": 9.756582914572864e-05,
-      "loss": 5.7755,
-      "step": 2923
-    },
-    {
-      "epoch": 1.5249022164276402,
-      "grad_norm": 1.7032673358917236,
-      "learning_rate": 9.756482412060302e-05,
-      "loss": 5.8643,
-      "step": 2924
-    },
-    {
-      "epoch": 1.5254237288135593,
-      "grad_norm": 1.404292106628418,
-      "learning_rate": 9.75638190954774e-05,
-      "loss": 6.3486,
-      "step": 2925
-    },
-    {
-      "epoch": 1.5259452411994785,
-      "grad_norm": 1.519274115562439,
-      "learning_rate": 9.756281407035176e-05,
-      "loss": 6.1759,
-      "step": 2926
-    },
-    {
-      "epoch": 1.5264667535853977,
-      "grad_norm": 1.6258536577224731,
-      "learning_rate": 9.756180904522614e-05,
-      "loss": 6.0833,
-      "step": 2927
-    },
-    {
-      "epoch": 1.5269882659713168,
-      "grad_norm": 1.7093464136123657,
-      "learning_rate": 9.75608040201005e-05,
-      "loss": 6.1114,
-      "step": 2928
-    },
-    {
-      "epoch": 1.527509778357236,
-      "grad_norm": 1.6579821109771729,
-      "learning_rate": 9.755979899497488e-05,
-      "loss": 5.83,
-      "step": 2929
-    },
-    {
-      "epoch": 1.5280312907431552,
-      "grad_norm": 1.6382776498794556,
-      "learning_rate": 9.755879396984925e-05,
-      "loss": 5.8508,
-      "step": 2930
-    },
-    {
-      "epoch": 1.5285528031290743,
-      "grad_norm": 1.6001276969909668,
-      "learning_rate": 9.755778894472363e-05,
-      "loss": 5.9308,
-      "step": 2931
-    },
-    {
-      "epoch": 1.5290743155149935,
-      "grad_norm": 1.5809142589569092,
-      "learning_rate": 9.755678391959799e-05,
-      "loss": 6.1609,
-      "step": 2932
-    },
-    {
-      "epoch": 1.5295958279009128,
-      "grad_norm": 1.8534437417984009,
-      "learning_rate": 9.755577889447237e-05,
-      "loss": 5.8663,
-      "step": 2933
-    },
-    {
-      "epoch": 1.5301173402868318,
-      "grad_norm": 1.6129528284072876,
-      "learning_rate": 9.755477386934673e-05,
-      "loss": 6.2429,
-      "step": 2934
-    },
-    {
-      "epoch": 1.530638852672751,
-      "grad_norm": 1.6421903371810913,
-      "learning_rate": 9.755376884422111e-05,
-      "loss": 5.9604,
-      "step": 2935
-    },
-    {
-      "epoch": 1.5311603650586703,
-      "grad_norm": 1.5983402729034424,
-      "learning_rate": 9.755276381909549e-05,
-      "loss": 5.666,
-      "step": 2936
-    },
-    {
-      "epoch": 1.5316818774445893,
-      "grad_norm": 1.7107360363006592,
-      "learning_rate": 9.755175879396985e-05,
-      "loss": 5.7312,
-      "step": 2937
-    },
-    {
-      "epoch": 1.5322033898305085,
-      "grad_norm": 1.8987083435058594,
-      "learning_rate": 9.755075376884423e-05,
-      "loss": 5.5693,
-      "step": 2938
-    },
-    {
-      "epoch": 1.5327249022164278,
-      "grad_norm": 1.514696478843689,
-      "learning_rate": 9.75497487437186e-05,
-      "loss": 6.2017,
-      "step": 2939
-    },
-    {
-      "epoch": 1.5332464146023468,
-      "grad_norm": 1.960425615310669,
-      "learning_rate": 9.754874371859297e-05,
-      "loss": 6.2485,
-      "step": 2940
-    },
-    {
-      "epoch": 1.5337679269882658,
-      "grad_norm": 1.7381919622421265,
-      "learning_rate": 9.754773869346734e-05,
-      "loss": 6.1301,
-      "step": 2941
-    },
-    {
-      "epoch": 1.5342894393741853,
-      "grad_norm": 1.4492655992507935,
-      "learning_rate": 9.754673366834171e-05,
-      "loss": 6.1159,
-      "step": 2942
-    },
-    {
-      "epoch": 1.5348109517601043,
-      "grad_norm": 1.6336946487426758,
-      "learning_rate": 9.754572864321608e-05,
-      "loss": 6.1576,
-      "step": 2943
-    },
-    {
-      "epoch": 1.5353324641460233,
-      "grad_norm": 1.7284420728683472,
-      "learning_rate": 9.754472361809046e-05,
-      "loss": 6.1785,
-      "step": 2944
-    },
-    {
-      "epoch": 1.5358539765319428,
-      "grad_norm": 1.6475152969360352,
-      "learning_rate": 9.754371859296483e-05,
-      "loss": 6.2428,
-      "step": 2945
-    },
-    {
-      "epoch": 1.5363754889178618,
-      "grad_norm": 1.6339117288589478,
-      "learning_rate": 9.754271356783921e-05,
-      "loss": 6.153,
-      "step": 2946
-    },
-    {
-      "epoch": 1.5368970013037808,
-      "grad_norm": 1.715817928314209,
-      "learning_rate": 9.754170854271358e-05,
-      "loss": 5.9706,
-      "step": 2947
-    },
-    {
-      "epoch": 1.5374185136897003,
-      "grad_norm": 1.686881184577942,
-      "learning_rate": 9.754070351758794e-05,
-      "loss": 6.1089,
-      "step": 2948
-    },
-    {
-      "epoch": 1.5379400260756193,
-      "grad_norm": 1.6033015251159668,
-      "learning_rate": 9.753969849246232e-05,
-      "loss": 6.0584,
-      "step": 2949
-    },
-    {
-      "epoch": 1.5384615384615383,
-      "grad_norm": 1.640824556350708,
-      "learning_rate": 9.753869346733668e-05,
-      "loss": 6.1831,
-      "step": 2950
-    },
-    {
-      "epoch": 1.5389830508474578,
-      "grad_norm": 1.4297370910644531,
-      "learning_rate": 9.753768844221106e-05,
-      "loss": 5.9394,
-      "step": 2951
-    },
-    {
-      "epoch": 1.5395045632333768,
-      "grad_norm": 1.4852330684661865,
-      "learning_rate": 9.753668341708542e-05,
-      "loss": 6.3877,
-      "step": 2952
-    },
-    {
-      "epoch": 1.5400260756192958,
-      "grad_norm": 1.996718168258667,
-      "learning_rate": 9.75356783919598e-05,
-      "loss": 5.4326,
-      "step": 2953
-    },
-    {
-      "epoch": 1.540547588005215,
-      "grad_norm": 1.5923837423324585,
-      "learning_rate": 9.753467336683417e-05,
-      "loss": 6.1394,
-      "step": 2954
-    },
-    {
-      "epoch": 1.5410691003911343,
-      "grad_norm": 1.5062754154205322,
-      "learning_rate": 9.753366834170854e-05,
-      "loss": 6.1537,
-      "step": 2955
-    },
-    {
-      "epoch": 1.5415906127770533,
-      "grad_norm": 1.5277647972106934,
-      "learning_rate": 9.753266331658292e-05,
-      "loss": 5.9143,
-      "step": 2956
-    },
-    {
-      "epoch": 1.5421121251629726,
-      "grad_norm": 1.3999603986740112,
-      "learning_rate": 9.75316582914573e-05,
-      "loss": 6.2244,
-      "step": 2957
-    },
-    {
-      "epoch": 1.5426336375488918,
-      "grad_norm": 1.6692534685134888,
-      "learning_rate": 9.753065326633166e-05,
-      "loss": 6.0189,
-      "step": 2958
-    },
-    {
-      "epoch": 1.5431551499348108,
-      "grad_norm": 1.6099843978881836,
-      "learning_rate": 9.752964824120604e-05,
-      "loss": 6.044,
-      "step": 2959
-    },
-    {
-      "epoch": 1.54367666232073,
-      "grad_norm": 1.4609355926513672,
-      "learning_rate": 9.75286432160804e-05,
-      "loss": 5.9263,
-      "step": 2960
-    },
-    {
-      "epoch": 1.5441981747066493,
-      "grad_norm": 1.9037601947784424,
-      "learning_rate": 9.752763819095477e-05,
-      "loss": 5.5976,
-      "step": 2961
-    },
-    {
-      "epoch": 1.5447196870925683,
-      "grad_norm": 1.6415376663208008,
-      "learning_rate": 9.752663316582915e-05,
-      "loss": 5.9765,
-      "step": 2962
-    },
-    {
-      "epoch": 1.5452411994784876,
-      "grad_norm": 1.714272141456604,
-      "learning_rate": 9.752562814070351e-05,
-      "loss": 5.4352,
-      "step": 2963
-    },
-    {
-      "epoch": 1.5457627118644068,
-      "grad_norm": 1.7286732196807861,
-      "learning_rate": 9.752462311557789e-05,
-      "loss": 5.9192,
-      "step": 2964
-    },
-    {
-      "epoch": 1.5462842242503259,
-      "grad_norm": 1.4834184646606445,
-      "learning_rate": 9.752361809045227e-05,
-      "loss": 6.2592,
-      "step": 2965
-    },
-    {
-      "epoch": 1.546805736636245,
-      "grad_norm": 1.61598539352417,
-      "learning_rate": 9.752261306532665e-05,
-      "loss": 5.8179,
-      "step": 2966
-    },
-    {
-      "epoch": 1.5473272490221643,
-      "grad_norm": 1.7281643152236938,
-      "learning_rate": 9.752160804020101e-05,
-      "loss": 6.0512,
-      "step": 2967
-    },
-    {
-      "epoch": 1.5478487614080834,
-      "grad_norm": 1.6342471837997437,
-      "learning_rate": 9.752060301507539e-05,
-      "loss": 6.0184,
-      "step": 2968
-    },
-    {
-      "epoch": 1.5483702737940026,
-      "grad_norm": 1.7903777360916138,
-      "learning_rate": 9.751959798994975e-05,
-      "loss": 5.3036,
-      "step": 2969
-    },
-    {
-      "epoch": 1.5488917861799218,
-      "grad_norm": 1.9066871404647827,
-      "learning_rate": 9.751859296482413e-05,
-      "loss": 5.4797,
-      "step": 2970
-    },
-    {
-      "epoch": 1.5494132985658409,
-      "grad_norm": 1.6245708465576172,
-      "learning_rate": 9.75175879396985e-05,
-      "loss": 6.0497,
-      "step": 2971
-    },
-    {
-      "epoch": 1.5499348109517601,
-      "grad_norm": 1.5795633792877197,
-      "learning_rate": 9.751658291457287e-05,
-      "loss": 5.8559,
-      "step": 2972
-    },
-    {
-      "epoch": 1.5504563233376794,
-      "grad_norm": 1.4975231885910034,
-      "learning_rate": 9.751557788944724e-05,
-      "loss": 6.3377,
-      "step": 2973
-    },
-    {
-      "epoch": 1.5509778357235984,
-      "grad_norm": 1.8085834980010986,
-      "learning_rate": 9.75145728643216e-05,
-      "loss": 5.9806,
-      "step": 2974
-    },
-    {
-      "epoch": 1.5514993481095176,
-      "grad_norm": 1.5248533487319946,
-      "learning_rate": 9.751356783919598e-05,
-      "loss": 6.3091,
-      "step": 2975
-    },
-    {
-      "epoch": 1.5520208604954369,
-      "grad_norm": 2.4740777015686035,
-      "learning_rate": 9.751256281407036e-05,
-      "loss": 5.5885,
-      "step": 2976
-    },
-    {
-      "epoch": 1.5525423728813559,
-      "grad_norm": 1.3745386600494385,
-      "learning_rate": 9.751155778894473e-05,
-      "loss": 6.216,
-      "step": 2977
-    },
-    {
-      "epoch": 1.5530638852672751,
-      "grad_norm": 1.5964089632034302,
-      "learning_rate": 9.75105527638191e-05,
-      "loss": 5.419,
-      "step": 2978
-    },
-    {
-      "epoch": 1.5535853976531944,
-      "grad_norm": 1.5235254764556885,
-      "learning_rate": 9.750954773869348e-05,
-      "loss": 6.1601,
-      "step": 2979
-    },
-    {
-      "epoch": 1.5541069100391134,
-      "grad_norm": 1.5818287134170532,
-      "learning_rate": 9.750854271356784e-05,
-      "loss": 5.849,
-      "step": 2980
-    },
-    {
-      "epoch": 1.5546284224250326,
-      "grad_norm": 1.5845973491668701,
-      "learning_rate": 9.750753768844222e-05,
-      "loss": 6.0963,
-      "step": 2981
-    },
-    {
-      "epoch": 1.5551499348109519,
-      "grad_norm": 1.655224084854126,
-      "learning_rate": 9.750653266331658e-05,
-      "loss": 5.596,
-      "step": 2982
-    },
-    {
-      "epoch": 1.555671447196871,
-      "grad_norm": 1.7107782363891602,
-      "learning_rate": 9.750552763819096e-05,
-      "loss": 5.8844,
-      "step": 2983
-    },
-    {
-      "epoch": 1.5561929595827901,
-      "grad_norm": 1.5279433727264404,
-      "learning_rate": 9.750452261306533e-05,
-      "loss": 5.7067,
-      "step": 2984
-    },
-    {
-      "epoch": 1.5567144719687094,
-      "grad_norm": 1.6591229438781738,
-      "learning_rate": 9.75035175879397e-05,
-      "loss": 6.0565,
-      "step": 2985
-    },
-    {
-      "epoch": 1.5572359843546284,
-      "grad_norm": 1.5764596462249756,
-      "learning_rate": 9.750251256281408e-05,
-      "loss": 5.9206,
-      "step": 2986
-    },
-    {
-      "epoch": 1.5577574967405476,
-      "grad_norm": 1.5098923444747925,
-      "learning_rate": 9.750150753768846e-05,
-      "loss": 6.392,
-      "step": 2987
-    },
-    {
-      "epoch": 1.5582790091264669,
-      "grad_norm": 1.9125800132751465,
-      "learning_rate": 9.750050251256282e-05,
-      "loss": 5.6939,
-      "step": 2988
-    },
-    {
-      "epoch": 1.558800521512386,
-      "grad_norm": 1.614362120628357,
-      "learning_rate": 9.749949748743719e-05,
-      "loss": 5.2786,
-      "step": 2989
-    },
-    {
-      "epoch": 1.559322033898305,
-      "grad_norm": 1.8829795122146606,
-      "learning_rate": 9.749849246231156e-05,
-      "loss": 5.7584,
-      "step": 2990
-    },
-    {
-      "epoch": 1.5598435462842244,
-      "grad_norm": 1.995069146156311,
-      "learning_rate": 9.749748743718593e-05,
-      "loss": 5.7102,
-      "step": 2991
-    },
-    {
-      "epoch": 1.5603650586701434,
-      "grad_norm": 1.6811987161636353,
-      "learning_rate": 9.749648241206031e-05,
-      "loss": 6.2956,
-      "step": 2992
-    },
-    {
-      "epoch": 1.5608865710560624,
-      "grad_norm": 1.592775583267212,
-      "learning_rate": 9.749547738693467e-05,
-      "loss": 6.1885,
-      "step": 2993
-    },
-    {
-      "epoch": 1.561408083441982,
-      "grad_norm": 2.03885555267334,
-      "learning_rate": 9.749447236180905e-05,
-      "loss": 5.5371,
-      "step": 2994
-    },
-    {
-      "epoch": 1.561929595827901,
-      "grad_norm": 1.621597409248352,
-      "learning_rate": 9.749346733668341e-05,
-      "loss": 5.6533,
-      "step": 2995
-    },
-    {
-      "epoch": 1.56245110821382,
-      "grad_norm": 1.7323951721191406,
-      "learning_rate": 9.749246231155779e-05,
-      "loss": 6.096,
-      "step": 2996
-    },
-    {
-      "epoch": 1.5629726205997394,
-      "grad_norm": 1.7036476135253906,
-      "learning_rate": 9.749145728643217e-05,
-      "loss": 5.7614,
-      "step": 2997
-    },
-    {
-      "epoch": 1.5634941329856584,
-      "grad_norm": 1.639918565750122,
-      "learning_rate": 9.749045226130655e-05,
-      "loss": 5.8986,
-      "step": 2998
-    },
-    {
-      "epoch": 1.5640156453715774,
-      "grad_norm": 1.728251338005066,
-      "learning_rate": 9.748944723618091e-05,
-      "loss": 6.0599,
-      "step": 2999
-    },
-    {
-      "epoch": 1.5645371577574967,
-      "grad_norm": 1.6250648498535156,
-      "learning_rate": 9.748844221105529e-05,
-      "loss": 6.2399,
-      "step": 3000
-    },
-    {
-      "epoch": 1.565058670143416,
-      "grad_norm": 2.3469226360321045,
-      "learning_rate": 9.748743718592965e-05,
-      "loss": 5.2829,
-      "step": 3001
-    },
-    {
-      "epoch": 1.565580182529335,
-      "grad_norm": 1.505710482597351,
-      "learning_rate": 9.748643216080402e-05,
-      "loss": 6.2257,
-      "step": 3002
-    },
-    {
-      "epoch": 1.5661016949152542,
-      "grad_norm": 1.5264219045639038,
-      "learning_rate": 9.74854271356784e-05,
-      "loss": 5.8644,
-      "step": 3003
-    },
-    {
-      "epoch": 1.5666232073011734,
-      "grad_norm": 1.6579408645629883,
-      "learning_rate": 9.748442211055276e-05,
-      "loss": 5.711,
-      "step": 3004
-    },
-    {
-      "epoch": 1.5671447196870925,
-      "grad_norm": 1.549682855606079,
-      "learning_rate": 9.748341708542714e-05,
-      "loss": 6.0775,
-      "step": 3005
-    },
-    {
-      "epoch": 1.5676662320730117,
-      "grad_norm": 1.4245432615280151,
-      "learning_rate": 9.748241206030152e-05,
-      "loss": 6.2478,
-      "step": 3006
-    },
-    {
-      "epoch": 1.568187744458931,
-      "grad_norm": 1.4779332876205444,
-      "learning_rate": 9.748140703517589e-05,
-      "loss": 6.167,
-      "step": 3007
-    },
-    {
-      "epoch": 1.56870925684485,
-      "grad_norm": 1.5079220533370972,
-      "learning_rate": 9.748040201005026e-05,
-      "loss": 6.2643,
-      "step": 3008
-    },
-    {
-      "epoch": 1.5692307692307692,
-      "grad_norm": 1.641533613204956,
-      "learning_rate": 9.747939698492464e-05,
-      "loss": 6.1743,
-      "step": 3009
-    },
-    {
-      "epoch": 1.5697522816166884,
-      "grad_norm": 1.50382661819458,
-      "learning_rate": 9.7478391959799e-05,
-      "loss": 6.3332,
-      "step": 3010
-    },
-    {
-      "epoch": 1.5702737940026075,
-      "grad_norm": 1.461124062538147,
-      "learning_rate": 9.747738693467338e-05,
-      "loss": 6.0997,
-      "step": 3011
-    },
-    {
-      "epoch": 1.5707953063885267,
-      "grad_norm": 1.540845274925232,
-      "learning_rate": 9.747638190954774e-05,
-      "loss": 6.2184,
-      "step": 3012
-    },
-    {
-      "epoch": 1.571316818774446,
-      "grad_norm": 1.6550092697143555,
-      "learning_rate": 9.747537688442212e-05,
-      "loss": 6.1339,
-      "step": 3013
-    },
-    {
-      "epoch": 1.571838331160365,
-      "grad_norm": 1.4923017024993896,
-      "learning_rate": 9.747437185929648e-05,
-      "loss": 6.4135,
-      "step": 3014
-    },
-    {
-      "epoch": 1.5723598435462842,
-      "grad_norm": 1.6147205829620361,
-      "learning_rate": 9.747336683417085e-05,
-      "loss": 6.1525,
-      "step": 3015
-    },
-    {
-      "epoch": 1.5728813559322035,
-      "grad_norm": 1.5925735235214233,
-      "learning_rate": 9.747236180904523e-05,
-      "loss": 6.1855,
-      "step": 3016
-    },
-    {
-      "epoch": 1.5734028683181225,
-      "grad_norm": 1.4535584449768066,
-      "learning_rate": 9.74713567839196e-05,
-      "loss": 6.3462,
-      "step": 3017
-    },
-    {
-      "epoch": 1.5739243807040417,
-      "grad_norm": 1.6119768619537354,
-      "learning_rate": 9.747035175879398e-05,
-      "loss": 6.2605,
-      "step": 3018
-    },
-    {
-      "epoch": 1.574445893089961,
-      "grad_norm": 1.7026129961013794,
-      "learning_rate": 9.746934673366835e-05,
-      "loss": 5.8512,
-      "step": 3019
-    },
-    {
-      "epoch": 1.57496740547588,
-      "grad_norm": 1.547124981880188,
-      "learning_rate": 9.746834170854272e-05,
-      "loss": 5.981,
-      "step": 3020
-    },
-    {
-      "epoch": 1.5754889178617992,
-      "grad_norm": 1.8274706602096558,
-      "learning_rate": 9.746733668341709e-05,
-      "loss": 5.6709,
-      "step": 3021
-    },
-    {
-      "epoch": 1.5760104302477185,
-      "grad_norm": 1.559336543083191,
-      "learning_rate": 9.746633165829147e-05,
-      "loss": 6.1293,
-      "step": 3022
-    },
-    {
-      "epoch": 1.5765319426336375,
-      "grad_norm": 1.7963043451309204,
-      "learning_rate": 9.746532663316583e-05,
-      "loss": 6.0766,
-      "step": 3023
-    },
-    {
-      "epoch": 1.5770534550195567,
-      "grad_norm": 1.5406216382980347,
-      "learning_rate": 9.746432160804021e-05,
-      "loss": 5.9616,
-      "step": 3024
-    },
-    {
-      "epoch": 1.577574967405476,
-      "grad_norm": 1.4781417846679688,
-      "learning_rate": 9.746331658291457e-05,
-      "loss": 5.9903,
-      "step": 3025
-    },
-    {
-      "epoch": 1.578096479791395,
-      "grad_norm": 1.6290030479431152,
-      "learning_rate": 9.746231155778895e-05,
-      "loss": 6.0626,
-      "step": 3026
-    },
-    {
-      "epoch": 1.5786179921773142,
-      "grad_norm": 1.7072821855545044,
-      "learning_rate": 9.746130653266333e-05,
-      "loss": 6.2445,
-      "step": 3027
-    },
-    {
-      "epoch": 1.5791395045632335,
-      "grad_norm": 1.7702466249465942,
-      "learning_rate": 9.746030150753769e-05,
-      "loss": 5.8532,
-      "step": 3028
-    },
-    {
-      "epoch": 1.5796610169491525,
-      "grad_norm": 1.4710116386413574,
-      "learning_rate": 9.745929648241207e-05,
-      "loss": 5.9021,
-      "step": 3029
-    },
-    {
-      "epoch": 1.5801825293350718,
-      "grad_norm": 1.680172324180603,
-      "learning_rate": 9.745829145728643e-05,
-      "loss": 5.7276,
-      "step": 3030
-    },
-    {
-      "epoch": 1.580704041720991,
-      "grad_norm": 1.5179733037948608,
-      "learning_rate": 9.745728643216081e-05,
-      "loss": 5.9933,
-      "step": 3031
-    },
-    {
-      "epoch": 1.58122555410691,
-      "grad_norm": 1.4295916557312012,
-      "learning_rate": 9.745628140703518e-05,
-      "loss": 6.1653,
-      "step": 3032
-    },
-    {
-      "epoch": 1.5817470664928293,
-      "grad_norm": 2.4275026321411133,
-      "learning_rate": 9.745527638190955e-05,
-      "loss": 6.0781,
-      "step": 3033
-    },
-    {
-      "epoch": 1.5822685788787485,
-      "grad_norm": 1.754798173904419,
-      "learning_rate": 9.745427135678392e-05,
-      "loss": 5.9422,
-      "step": 3034
-    },
-    {
-      "epoch": 1.5827900912646675,
-      "grad_norm": 1.4458229541778564,
-      "learning_rate": 9.74532663316583e-05,
-      "loss": 6.2068,
-      "step": 3035
-    },
-    {
-      "epoch": 1.5833116036505865,
-      "grad_norm": 1.5155539512634277,
-      "learning_rate": 9.745226130653266e-05,
-      "loss": 6.0428,
-      "step": 3036
-    },
-    {
-      "epoch": 1.583833116036506,
-      "grad_norm": 1.578027606010437,
-      "learning_rate": 9.745125628140704e-05,
-      "loss": 6.3008,
-      "step": 3037
-    },
-    {
-      "epoch": 1.584354628422425,
-      "grad_norm": 1.6520020961761475,
-      "learning_rate": 9.745025125628142e-05,
-      "loss": 5.924,
-      "step": 3038
-    },
-    {
-      "epoch": 1.584876140808344,
-      "grad_norm": 1.5327036380767822,
-      "learning_rate": 9.74492462311558e-05,
-      "loss": 6.187,
-      "step": 3039
-    },
-    {
-      "epoch": 1.5853976531942635,
-      "grad_norm": 1.4840681552886963,
-      "learning_rate": 9.744824120603016e-05,
-      "loss": 6.0787,
-      "step": 3040
-    },
-    {
-      "epoch": 1.5859191655801825,
-      "grad_norm": 1.6682302951812744,
-      "learning_rate": 9.744723618090452e-05,
-      "loss": 5.9859,
-      "step": 3041
-    },
-    {
-      "epoch": 1.5864406779661016,
-      "grad_norm": 1.7628681659698486,
-      "learning_rate": 9.74462311557789e-05,
-      "loss": 5.0342,
-      "step": 3042
-    },
-    {
-      "epoch": 1.586962190352021,
-      "grad_norm": 1.3723901510238647,
-      "learning_rate": 9.744522613065326e-05,
-      "loss": 6.2808,
-      "step": 3043
-    },
-    {
-      "epoch": 1.58748370273794,
-      "grad_norm": 1.7502182722091675,
-      "learning_rate": 9.744422110552764e-05,
-      "loss": 5.5828,
-      "step": 3044
-    },
-    {
-      "epoch": 1.588005215123859,
-      "grad_norm": 1.4252878427505493,
-      "learning_rate": 9.7443216080402e-05,
-      "loss": 5.8077,
-      "step": 3045
-    },
-    {
-      "epoch": 1.5885267275097783,
-      "grad_norm": 1.544830083847046,
-      "learning_rate": 9.744221105527638e-05,
-      "loss": 5.8496,
-      "step": 3046
-    },
-    {
-      "epoch": 1.5890482398956975,
-      "grad_norm": 1.733964443206787,
-      "learning_rate": 9.744120603015076e-05,
-      "loss": 5.1959,
-      "step": 3047
-    },
-    {
-      "epoch": 1.5895697522816166,
-      "grad_norm": 2.145663022994995,
-      "learning_rate": 9.744020100502514e-05,
-      "loss": 5.6511,
-      "step": 3048
-    },
-    {
-      "epoch": 1.5900912646675358,
-      "grad_norm": 1.7752289772033691,
-      "learning_rate": 9.74391959798995e-05,
-      "loss": 5.4873,
-      "step": 3049
-    },
-    {
-      "epoch": 1.590612777053455,
-      "grad_norm": 1.7373502254486084,
-      "learning_rate": 9.743819095477388e-05,
-      "loss": 5.6868,
-      "step": 3050
-    },
-    {
-      "epoch": 1.591134289439374,
-      "grad_norm": 1.5320210456848145,
-      "learning_rate": 9.743718592964825e-05,
-      "loss": 6.3112,
-      "step": 3051
-    },
-    {
-      "epoch": 1.5916558018252933,
-      "grad_norm": 1.4615966081619263,
-      "learning_rate": 9.743618090452262e-05,
-      "loss": 6.4131,
-      "step": 3052
-    },
-    {
-      "epoch": 1.5921773142112126,
-      "grad_norm": 1.70128333568573,
-      "learning_rate": 9.743517587939699e-05,
-      "loss": 5.7769,
-      "step": 3053
-    },
-    {
-      "epoch": 1.5926988265971316,
-      "grad_norm": 1.4361833333969116,
-      "learning_rate": 9.743417085427135e-05,
-      "loss": 6.1996,
-      "step": 3054
-    },
-    {
-      "epoch": 1.5932203389830508,
-      "grad_norm": 1.9747556447982788,
-      "learning_rate": 9.743316582914573e-05,
-      "loss": 5.4913,
-      "step": 3055
-    },
-    {
-      "epoch": 1.59374185136897,
-      "grad_norm": 1.5286611318588257,
-      "learning_rate": 9.74321608040201e-05,
-      "loss": 5.9351,
-      "step": 3056
-    },
-    {
-      "epoch": 1.594263363754889,
-      "grad_norm": 1.548088550567627,
-      "learning_rate": 9.743115577889447e-05,
-      "loss": 5.9806,
-      "step": 3057
-    },
-    {
-      "epoch": 1.5947848761408083,
-      "grad_norm": 1.812968134880066,
-      "learning_rate": 9.743015075376885e-05,
-      "loss": 6.0099,
-      "step": 3058
-    },
-    {
-      "epoch": 1.5953063885267276,
-      "grad_norm": 1.438568353652954,
-      "learning_rate": 9.742914572864323e-05,
-      "loss": 6.2415,
-      "step": 3059
-    },
-    {
-      "epoch": 1.5958279009126466,
-      "grad_norm": 1.4927949905395508,
-      "learning_rate": 9.742814070351759e-05,
-      "loss": 5.6877,
-      "step": 3060
-    },
-    {
-      "epoch": 1.5963494132985658,
-      "grad_norm": 1.5806691646575928,
-      "learning_rate": 9.742713567839197e-05,
-      "loss": 6.2952,
-      "step": 3061
-    },
-    {
-      "epoch": 1.596870925684485,
-      "grad_norm": 1.5213185548782349,
-      "learning_rate": 9.742613065326633e-05,
-      "loss": 5.9267,
-      "step": 3062
-    },
-    {
-      "epoch": 1.597392438070404,
-      "grad_norm": 1.5299967527389526,
-      "learning_rate": 9.742512562814071e-05,
-      "loss": 5.8624,
-      "step": 3063
-    },
-    {
-      "epoch": 1.5979139504563233,
-      "grad_norm": 2.186304807662964,
-      "learning_rate": 9.742412060301508e-05,
-      "loss": 5.6364,
-      "step": 3064
-    },
-    {
-      "epoch": 1.5984354628422426,
-      "grad_norm": 1.8593779802322388,
-      "learning_rate": 9.742311557788945e-05,
-      "loss": 5.7762,
-      "step": 3065
-    },
-    {
-      "epoch": 1.5989569752281616,
-      "grad_norm": 1.5214787721633911,
-      "learning_rate": 9.742211055276382e-05,
-      "loss": 5.6702,
-      "step": 3066
-    },
-    {
-      "epoch": 1.5994784876140808,
-      "grad_norm": 1.5965590476989746,
-      "learning_rate": 9.742110552763818e-05,
-      "loss": 6.1784,
-      "step": 3067
-    },
-    {
-      "epoch": 1.6,
-      "grad_norm": 2.020270586013794,
-      "learning_rate": 9.742010050251256e-05,
-      "loss": 6.126,
-      "step": 3068
-    },
-    {
-      "epoch": 1.600521512385919,
-      "grad_norm": 1.525932788848877,
-      "learning_rate": 9.741909547738694e-05,
-      "loss": 6.2303,
-      "step": 3069
-    },
-    {
-      "epoch": 1.6010430247718384,
-      "grad_norm": 1.7978794574737549,
-      "learning_rate": 9.741809045226132e-05,
-      "loss": 5.7706,
-      "step": 3070
-    },
-    {
-      "epoch": 1.6015645371577576,
-      "grad_norm": 1.65778648853302,
-      "learning_rate": 9.741708542713568e-05,
-      "loss": 5.6908,
-      "step": 3071
-    },
-    {
-      "epoch": 1.6020860495436766,
-      "grad_norm": 1.375840663909912,
-      "learning_rate": 9.741608040201006e-05,
-      "loss": 6.1165,
-      "step": 3072
-    },
-    {
-      "epoch": 1.6026075619295959,
-      "grad_norm": 1.5219671726226807,
-      "learning_rate": 9.741507537688442e-05,
-      "loss": 6.0988,
-      "step": 3073
-    },
-    {
-      "epoch": 1.603129074315515,
-      "grad_norm": 1.7993241548538208,
-      "learning_rate": 9.74140703517588e-05,
-      "loss": 5.8465,
-      "step": 3074
-    },
-    {
-      "epoch": 1.6036505867014341,
-      "grad_norm": 1.4790480136871338,
-      "learning_rate": 9.741306532663317e-05,
-      "loss": 6.1904,
-      "step": 3075
-    },
-    {
-      "epoch": 1.6041720990873534,
-      "grad_norm": 1.670688509941101,
-      "learning_rate": 9.741206030150754e-05,
-      "loss": 5.8278,
-      "step": 3076
-    },
-    {
-      "epoch": 1.6046936114732726,
-      "grad_norm": 1.8832173347473145,
-      "learning_rate": 9.741105527638191e-05,
-      "loss": 5.718,
-      "step": 3077
-    },
-    {
-      "epoch": 1.6052151238591916,
-      "grad_norm": 1.6235421895980835,
-      "learning_rate": 9.741005025125629e-05,
-      "loss": 6.0965,
-      "step": 3078
-    },
-    {
-      "epoch": 1.6057366362451109,
-      "grad_norm": 1.6298588514328003,
-      "learning_rate": 9.740904522613066e-05,
-      "loss": 6.049,
-      "step": 3079
-    },
-    {
-      "epoch": 1.6062581486310301,
-      "grad_norm": 1.5675996541976929,
-      "learning_rate": 9.740804020100504e-05,
-      "loss": 5.7409,
-      "step": 3080
-    },
-    {
-      "epoch": 1.6067796610169491,
-      "grad_norm": 1.7733455896377563,
-      "learning_rate": 9.74070351758794e-05,
-      "loss": 5.7855,
-      "step": 3081
-    },
-    {
-      "epoch": 1.6073011734028682,
-      "grad_norm": 1.679229974746704,
-      "learning_rate": 9.740603015075377e-05,
-      "loss": 6.0233,
-      "step": 3082
-    },
-    {
-      "epoch": 1.6078226857887876,
-      "grad_norm": 1.6783993244171143,
-      "learning_rate": 9.740502512562815e-05,
-      "loss": 5.5203,
-      "step": 3083
-    },
-    {
-      "epoch": 1.6083441981747066,
-      "grad_norm": 1.510166049003601,
-      "learning_rate": 9.740402010050251e-05,
-      "loss": 6.3932,
-      "step": 3084
-    },
-    {
-      "epoch": 1.6088657105606257,
-      "grad_norm": 1.5554659366607666,
-      "learning_rate": 9.740301507537689e-05,
-      "loss": 6.0891,
-      "step": 3085
-    },
-    {
-      "epoch": 1.6093872229465451,
-      "grad_norm": 1.4308596849441528,
-      "learning_rate": 9.740201005025125e-05,
-      "loss": 6.2292,
-      "step": 3086
-    },
-    {
-      "epoch": 1.6099087353324641,
-      "grad_norm": 1.6161266565322876,
-      "learning_rate": 9.740100502512563e-05,
-      "loss": 5.7356,
-      "step": 3087
-    },
-    {
-      "epoch": 1.6104302477183832,
-      "grad_norm": 1.5871121883392334,
-      "learning_rate": 9.74e-05,
-      "loss": 5.7743,
-      "step": 3088
-    },
-    {
-      "epoch": 1.6109517601043026,
-      "grad_norm": 1.6464309692382812,
-      "learning_rate": 9.739899497487437e-05,
-      "loss": 6.06,
-      "step": 3089
-    },
-    {
-      "epoch": 1.6114732724902217,
-      "grad_norm": 1.6074179410934448,
-      "learning_rate": 9.739798994974875e-05,
-      "loss": 6.275,
-      "step": 3090
-    },
-    {
-      "epoch": 1.6119947848761407,
-      "grad_norm": 1.4984294176101685,
-      "learning_rate": 9.739698492462313e-05,
-      "loss": 6.1576,
-      "step": 3091
-    },
-    {
-      "epoch": 1.6125162972620601,
-      "grad_norm": 1.8707150220870972,
-      "learning_rate": 9.73959798994975e-05,
-      "loss": 5.5464,
-      "step": 3092
-    },
-    {
-      "epoch": 1.6130378096479792,
-      "grad_norm": 1.5556368827819824,
-      "learning_rate": 9.739497487437187e-05,
-      "loss": 5.8181,
-      "step": 3093
-    },
-    {
-      "epoch": 1.6135593220338982,
-      "grad_norm": 1.747147560119629,
-      "learning_rate": 9.739396984924624e-05,
-      "loss": 5.8047,
-      "step": 3094
-    },
-    {
-      "epoch": 1.6140808344198174,
-      "grad_norm": 1.6061432361602783,
-      "learning_rate": 9.73929648241206e-05,
-      "loss": 6.1276,
-      "step": 3095
-    },
-    {
-      "epoch": 1.6146023468057367,
-      "grad_norm": 1.7293413877487183,
-      "learning_rate": 9.739195979899498e-05,
-      "loss": 5.3286,
-      "step": 3096
-    },
-    {
-      "epoch": 1.6151238591916557,
-      "grad_norm": 1.447916030883789,
-      "learning_rate": 9.739095477386934e-05,
-      "loss": 5.6049,
-      "step": 3097
-    },
-    {
-      "epoch": 1.615645371577575,
-      "grad_norm": 1.6178771257400513,
-      "learning_rate": 9.738994974874372e-05,
-      "loss": 5.8783,
-      "step": 3098
-    },
-    {
-      "epoch": 1.6161668839634942,
-      "grad_norm": 1.8009456396102905,
-      "learning_rate": 9.73889447236181e-05,
-      "loss": 5.924,
-      "step": 3099
-    },
-    {
-      "epoch": 1.6166883963494132,
-      "grad_norm": 1.552306056022644,
-      "learning_rate": 9.738793969849248e-05,
-      "loss": 6.1086,
-      "step": 3100
-    },
-    {
-      "epoch": 1.6172099087353324,
-      "grad_norm": 1.544073224067688,
-      "learning_rate": 9.738693467336684e-05,
-      "loss": 5.2937,
-      "step": 3101
-    },
-    {
-      "epoch": 1.6177314211212517,
-      "grad_norm": 1.6039420366287231,
-      "learning_rate": 9.738592964824122e-05,
-      "loss": 6.1915,
-      "step": 3102
-    },
-    {
-      "epoch": 1.6182529335071707,
-      "grad_norm": 1.5217640399932861,
-      "learning_rate": 9.738492462311558e-05,
-      "loss": 6.0091,
-      "step": 3103
-    },
-    {
-      "epoch": 1.61877444589309,
-      "grad_norm": 1.600091814994812,
-      "learning_rate": 9.738391959798996e-05,
-      "loss": 6.2108,
-      "step": 3104
-    },
-    {
-      "epoch": 1.6192959582790092,
-      "grad_norm": 1.5660648345947266,
-      "learning_rate": 9.738291457286432e-05,
-      "loss": 5.9596,
-      "step": 3105
-    },
-    {
-      "epoch": 1.6198174706649282,
-      "grad_norm": 1.6846814155578613,
-      "learning_rate": 9.73819095477387e-05,
-      "loss": 5.7896,
-      "step": 3106
-    },
-    {
-      "epoch": 1.6203389830508474,
-      "grad_norm": 1.575038194656372,
-      "learning_rate": 9.738090452261307e-05,
-      "loss": 6.0597,
-      "step": 3107
-    },
-    {
-      "epoch": 1.6208604954367667,
-      "grad_norm": 1.5629881620407104,
-      "learning_rate": 9.737989949748743e-05,
-      "loss": 6.0841,
-      "step": 3108
-    },
-    {
-      "epoch": 1.6213820078226857,
-      "grad_norm": 1.6237661838531494,
-      "learning_rate": 9.737889447236181e-05,
-      "loss": 5.8951,
-      "step": 3109
-    },
-    {
-      "epoch": 1.621903520208605,
-      "grad_norm": 1.5605125427246094,
-      "learning_rate": 9.737788944723619e-05,
-      "loss": 6.056,
-      "step": 3110
-    },
-    {
-      "epoch": 1.6224250325945242,
-      "grad_norm": 1.6115822792053223,
-      "learning_rate": 9.737688442211056e-05,
-      "loss": 5.8353,
-      "step": 3111
-    },
-    {
-      "epoch": 1.6229465449804432,
-      "grad_norm": 1.6092780828475952,
-      "learning_rate": 9.737587939698493e-05,
-      "loss": 6.2188,
-      "step": 3112
-    },
-    {
-      "epoch": 1.6234680573663625,
-      "grad_norm": 1.494054913520813,
-      "learning_rate": 9.73748743718593e-05,
-      "loss": 6.2143,
-      "step": 3113
-    },
-    {
-      "epoch": 1.6239895697522817,
-      "grad_norm": 1.4322123527526855,
-      "learning_rate": 9.737386934673367e-05,
-      "loss": 6.3,
-      "step": 3114
-    },
-    {
-      "epoch": 1.6245110821382007,
-      "grad_norm": 1.6598525047302246,
-      "learning_rate": 9.737286432160805e-05,
-      "loss": 5.5408,
-      "step": 3115
-    },
-    {
-      "epoch": 1.62503259452412,
-      "grad_norm": 1.647043228149414,
-      "learning_rate": 9.737185929648241e-05,
-      "loss": 6.0035,
-      "step": 3116
-    },
-    {
-      "epoch": 1.6255541069100392,
-      "grad_norm": 1.543968677520752,
-      "learning_rate": 9.737085427135679e-05,
-      "loss": 6.0252,
-      "step": 3117
-    },
-    {
-      "epoch": 1.6260756192959582,
-      "grad_norm": 1.6839088201522827,
-      "learning_rate": 9.736984924623115e-05,
-      "loss": 6.2788,
-      "step": 3118
-    },
-    {
-      "epoch": 1.6265971316818775,
-      "grad_norm": 1.6307122707366943,
-      "learning_rate": 9.736884422110553e-05,
-      "loss": 6.0629,
-      "step": 3119
-    },
-    {
-      "epoch": 1.6271186440677967,
-      "grad_norm": 1.541347861289978,
-      "learning_rate": 9.736783919597991e-05,
-      "loss": 6.0879,
-      "step": 3120
-    },
-    {
-      "epoch": 1.6276401564537157,
-      "grad_norm": 1.6512030363082886,
-      "learning_rate": 9.736683417085427e-05,
-      "loss": 5.8473,
-      "step": 3121
-    },
-    {
-      "epoch": 1.628161668839635,
-      "grad_norm": 1.576765775680542,
-      "learning_rate": 9.736582914572865e-05,
-      "loss": 6.1983,
-      "step": 3122
-    },
-    {
-      "epoch": 1.6286831812255542,
-      "grad_norm": 1.4205610752105713,
-      "learning_rate": 9.736482412060302e-05,
-      "loss": 6.1726,
-      "step": 3123
-    },
-    {
-      "epoch": 1.6292046936114732,
-      "grad_norm": 1.5198034048080444,
-      "learning_rate": 9.73638190954774e-05,
-      "loss": 5.8932,
-      "step": 3124
-    },
-    {
-      "epoch": 1.6297262059973925,
-      "grad_norm": 1.707255244255066,
-      "learning_rate": 9.736281407035176e-05,
-      "loss": 5.5705,
-      "step": 3125
-    },
-    {
-      "epoch": 1.6302477183833117,
-      "grad_norm": 1.4681293964385986,
-      "learning_rate": 9.736180904522614e-05,
-      "loss": 6.2079,
-      "step": 3126
-    },
-    {
-      "epoch": 1.6307692307692307,
-      "grad_norm": 1.739350438117981,
-      "learning_rate": 9.73608040201005e-05,
-      "loss": 5.7466,
-      "step": 3127
-    },
-    {
-      "epoch": 1.6312907431551498,
-      "grad_norm": 1.8174374103546143,
-      "learning_rate": 9.735979899497488e-05,
-      "loss": 6.0534,
-      "step": 3128
-    },
-    {
-      "epoch": 1.6318122555410692,
-      "grad_norm": 1.691854476928711,
-      "learning_rate": 9.735879396984924e-05,
-      "loss": 5.692,
-      "step": 3129
-    },
-    {
-      "epoch": 1.6323337679269883,
-      "grad_norm": 1.603854775428772,
-      "learning_rate": 9.735778894472362e-05,
-      "loss": 6.09,
-      "step": 3130
-    },
-    {
-      "epoch": 1.6328552803129073,
-      "grad_norm": 1.5197350978851318,
-      "learning_rate": 9.7356783919598e-05,
-      "loss": 5.7362,
-      "step": 3131
-    },
-    {
-      "epoch": 1.6333767926988267,
-      "grad_norm": 1.5920674800872803,
-      "learning_rate": 9.735577889447238e-05,
-      "loss": 5.8172,
-      "step": 3132
-    },
-    {
-      "epoch": 1.6338983050847458,
-      "grad_norm": 1.6657129526138306,
-      "learning_rate": 9.735477386934674e-05,
-      "loss": 5.9969,
-      "step": 3133
-    },
-    {
-      "epoch": 1.6344198174706648,
-      "grad_norm": 1.5910208225250244,
-      "learning_rate": 9.73537688442211e-05,
-      "loss": 6.0633,
-      "step": 3134
-    },
-    {
-      "epoch": 1.6349413298565842,
-      "grad_norm": 1.6786820888519287,
-      "learning_rate": 9.735276381909548e-05,
-      "loss": 5.8129,
-      "step": 3135
-    },
-    {
-      "epoch": 1.6354628422425033,
-      "grad_norm": 1.702570915222168,
-      "learning_rate": 9.735175879396985e-05,
-      "loss": 6.4482,
-      "step": 3136
-    },
-    {
-      "epoch": 1.6359843546284223,
-      "grad_norm": 1.7444863319396973,
-      "learning_rate": 9.735075376884422e-05,
-      "loss": 5.9886,
-      "step": 3137
-    },
-    {
-      "epoch": 1.6365058670143418,
-      "grad_norm": 1.555273175239563,
-      "learning_rate": 9.734974874371859e-05,
-      "loss": 6.1227,
-      "step": 3138
-    },
-    {
-      "epoch": 1.6370273794002608,
-      "grad_norm": 1.495219111442566,
-      "learning_rate": 9.734874371859297e-05,
-      "loss": 6.3125,
-      "step": 3139
-    },
-    {
-      "epoch": 1.6375488917861798,
-      "grad_norm": 1.6466037034988403,
-      "learning_rate": 9.734773869346734e-05,
-      "loss": 6.2784,
-      "step": 3140
-    },
-    {
-      "epoch": 1.638070404172099,
-      "grad_norm": 1.6692683696746826,
-      "learning_rate": 9.734673366834172e-05,
-      "loss": 6.3513,
-      "step": 3141
-    },
-    {
-      "epoch": 1.6385919165580183,
-      "grad_norm": 1.5703704357147217,
-      "learning_rate": 9.734572864321609e-05,
-      "loss": 6.1411,
-      "step": 3142
-    },
-    {
-      "epoch": 1.6391134289439373,
-      "grad_norm": 1.8088418245315552,
-      "learning_rate": 9.734472361809046e-05,
-      "loss": 5.4836,
-      "step": 3143
-    },
-    {
-      "epoch": 1.6396349413298565,
-      "grad_norm": 1.7659289836883545,
-      "learning_rate": 9.734371859296483e-05,
-      "loss": 5.6414,
-      "step": 3144
-    },
-    {
-      "epoch": 1.6401564537157758,
-      "grad_norm": 1.4511542320251465,
-      "learning_rate": 9.73427135678392e-05,
-      "loss": 5.3162,
-      "step": 3145
-    },
-    {
-      "epoch": 1.6406779661016948,
-      "grad_norm": 1.4907194375991821,
-      "learning_rate": 9.734170854271357e-05,
-      "loss": 5.9429,
-      "step": 3146
-    },
-    {
-      "epoch": 1.641199478487614,
-      "grad_norm": 1.6705886125564575,
-      "learning_rate": 9.734070351758794e-05,
-      "loss": 5.7635,
-      "step": 3147
-    },
-    {
-      "epoch": 1.6417209908735333,
-      "grad_norm": 1.6852425336837769,
-      "learning_rate": 9.733969849246231e-05,
-      "loss": 6.226,
-      "step": 3148
-    },
-    {
-      "epoch": 1.6422425032594523,
-      "grad_norm": 1.6095467805862427,
-      "learning_rate": 9.733869346733668e-05,
-      "loss": 6.2492,
-      "step": 3149
-    },
-    {
-      "epoch": 1.6427640156453716,
-      "grad_norm": 1.592104434967041,
-      "learning_rate": 9.733768844221106e-05,
-      "loss": 5.9898,
-      "step": 3150
-    },
-    {
-      "epoch": 1.6432855280312908,
-      "grad_norm": 1.6531606912612915,
-      "learning_rate": 9.733668341708543e-05,
-      "loss": 6.0379,
-      "step": 3151
-    },
-    {
-      "epoch": 1.6438070404172098,
-      "grad_norm": 2.132061004638672,
-      "learning_rate": 9.733567839195981e-05,
-      "loss": 6.0662,
-      "step": 3152
-    },
-    {
-      "epoch": 1.644328552803129,
-      "grad_norm": 1.8207018375396729,
-      "learning_rate": 9.733467336683418e-05,
-      "loss": 5.7647,
-      "step": 3153
-    },
-    {
-      "epoch": 1.6448500651890483,
-      "grad_norm": 1.7841594219207764,
-      "learning_rate": 9.733366834170855e-05,
-      "loss": 6.0311,
-      "step": 3154
-    },
-    {
-      "epoch": 1.6453715775749673,
-      "grad_norm": 2.0698485374450684,
-      "learning_rate": 9.733266331658292e-05,
-      "loss": 6.168,
-      "step": 3155
-    },
-    {
-      "epoch": 1.6458930899608866,
-      "grad_norm": 1.9563928842544556,
-      "learning_rate": 9.73316582914573e-05,
-      "loss": 5.2311,
-      "step": 3156
-    },
-    {
-      "epoch": 1.6464146023468058,
-      "grad_norm": 1.6322273015975952,
-      "learning_rate": 9.733065326633166e-05,
-      "loss": 6.0019,
-      "step": 3157
-    },
-    {
-      "epoch": 1.6469361147327248,
-      "grad_norm": 1.7840666770935059,
-      "learning_rate": 9.732964824120604e-05,
-      "loss": 5.8147,
-      "step": 3158
-    },
-    {
-      "epoch": 1.647457627118644,
-      "grad_norm": 1.6237585544586182,
-      "learning_rate": 9.73286432160804e-05,
-      "loss": 6.1855,
-      "step": 3159
-    },
-    {
-      "epoch": 1.6479791395045633,
-      "grad_norm": 1.6893723011016846,
-      "learning_rate": 9.732763819095478e-05,
-      "loss": 5.4287,
-      "step": 3160
-    },
-    {
-      "epoch": 1.6485006518904823,
-      "grad_norm": 1.9960300922393799,
-      "learning_rate": 9.732663316582916e-05,
-      "loss": 6.0214,
-      "step": 3161
-    },
-    {
-      "epoch": 1.6490221642764016,
-      "grad_norm": 1.9659291505813599,
-      "learning_rate": 9.732562814070352e-05,
-      "loss": 6.0264,
-      "step": 3162
-    },
-    {
-      "epoch": 1.6495436766623208,
-      "grad_norm": 1.6435083150863647,
-      "learning_rate": 9.73246231155779e-05,
-      "loss": 6.1167,
-      "step": 3163
-    },
-    {
-      "epoch": 1.6500651890482398,
-      "grad_norm": 1.759752869606018,
-      "learning_rate": 9.732361809045226e-05,
-      "loss": 5.6,
-      "step": 3164
-    },
-    {
-      "epoch": 1.650586701434159,
-      "grad_norm": 1.8139715194702148,
-      "learning_rate": 9.732261306532664e-05,
-      "loss": 6.154,
-      "step": 3165
-    },
-    {
-      "epoch": 1.6511082138200783,
-      "grad_norm": 1.6587743759155273,
-      "learning_rate": 9.7321608040201e-05,
-      "loss": 6.0956,
-      "step": 3166
-    },
-    {
-      "epoch": 1.6516297262059974,
-      "grad_norm": 1.5413472652435303,
-      "learning_rate": 9.732060301507538e-05,
-      "loss": 6.356,
-      "step": 3167
-    },
-    {
-      "epoch": 1.6521512385919166,
-      "grad_norm": 2.0291459560394287,
-      "learning_rate": 9.731959798994975e-05,
-      "loss": 5.2967,
-      "step": 3168
-    },
-    {
-      "epoch": 1.6526727509778358,
-      "grad_norm": 1.5828510522842407,
-      "learning_rate": 9.731859296482413e-05,
-      "loss": 5.9587,
-      "step": 3169
-    },
-    {
-      "epoch": 1.6531942633637549,
-      "grad_norm": 1.633156180381775,
-      "learning_rate": 9.731758793969849e-05,
-      "loss": 5.6907,
-      "step": 3170
-    },
-    {
-      "epoch": 1.653715775749674,
-      "grad_norm": 1.5092873573303223,
-      "learning_rate": 9.731658291457287e-05,
-      "loss": 5.9621,
-      "step": 3171
-    },
-    {
-      "epoch": 1.6542372881355933,
-      "grad_norm": 1.314452052116394,
-      "learning_rate": 9.731557788944725e-05,
-      "loss": 6.321,
-      "step": 3172
-    },
-    {
-      "epoch": 1.6547588005215124,
-      "grad_norm": 1.517844319343567,
-      "learning_rate": 9.731457286432162e-05,
-      "loss": 5.93,
-      "step": 3173
-    },
-    {
-      "epoch": 1.6552803129074316,
-      "grad_norm": 1.4646230936050415,
-      "learning_rate": 9.731356783919599e-05,
-      "loss": 6.2153,
-      "step": 3174
-    },
-    {
-      "epoch": 1.6558018252933508,
-      "grad_norm": 1.441266417503357,
-      "learning_rate": 9.731256281407035e-05,
-      "loss": 6.0711,
-      "step": 3175
-    },
-    {
-      "epoch": 1.6563233376792699,
-      "grad_norm": 1.4708000421524048,
-      "learning_rate": 9.731155778894473e-05,
-      "loss": 5.9965,
-      "step": 3176
-    },
-    {
-      "epoch": 1.656844850065189,
-      "grad_norm": 1.5106927156448364,
-      "learning_rate": 9.73105527638191e-05,
-      "loss": 6.0369,
-      "step": 3177
-    },
-    {
-      "epoch": 1.6573663624511084,
-      "grad_norm": 1.526909589767456,
-      "learning_rate": 9.730954773869347e-05,
-      "loss": 5.6303,
-      "step": 3178
-    },
-    {
-      "epoch": 1.6578878748370274,
-      "grad_norm": 1.5698353052139282,
-      "learning_rate": 9.730854271356784e-05,
-      "loss": 5.858,
-      "step": 3179
-    },
-    {
-      "epoch": 1.6584093872229464,
-      "grad_norm": 1.653381586074829,
-      "learning_rate": 9.730753768844221e-05,
-      "loss": 5.6504,
-      "step": 3180
-    },
-    {
-      "epoch": 1.6589308996088659,
-      "grad_norm": 1.536844253540039,
-      "learning_rate": 9.730653266331659e-05,
-      "loss": 5.8652,
-      "step": 3181
-    },
-    {
-      "epoch": 1.6594524119947849,
-      "grad_norm": 1.779517650604248,
-      "learning_rate": 9.730552763819097e-05,
-      "loss": 5.9035,
-      "step": 3182
-    },
-    {
-      "epoch": 1.659973924380704,
-      "grad_norm": 1.669467806816101,
-      "learning_rate": 9.730452261306533e-05,
-      "loss": 5.2017,
-      "step": 3183
-    },
-    {
-      "epoch": 1.6604954367666234,
-      "grad_norm": 1.5142687559127808,
-      "learning_rate": 9.730351758793971e-05,
-      "loss": 6.1466,
-      "step": 3184
-    },
-    {
-      "epoch": 1.6610169491525424,
-      "grad_norm": 1.5393387079238892,
-      "learning_rate": 9.730251256281408e-05,
-      "loss": 6.1589,
-      "step": 3185
-    },
-    {
-      "epoch": 1.6615384615384614,
-      "grad_norm": 1.948114037513733,
-      "learning_rate": 9.730150753768845e-05,
-      "loss": 5.5209,
-      "step": 3186
-    },
-    {
-      "epoch": 1.6620599739243807,
-      "grad_norm": 1.6124712228775024,
-      "learning_rate": 9.730050251256282e-05,
-      "loss": 5.79,
-      "step": 3187
-    },
-    {
-      "epoch": 1.6625814863103,
-      "grad_norm": 1.585602879524231,
-      "learning_rate": 9.729949748743718e-05,
-      "loss": 6.0741,
-      "step": 3188
-    },
-    {
-      "epoch": 1.663102998696219,
-      "grad_norm": 1.609656572341919,
-      "learning_rate": 9.729849246231156e-05,
-      "loss": 5.7526,
-      "step": 3189
-    },
-    {
-      "epoch": 1.6636245110821382,
-      "grad_norm": 1.617051362991333,
-      "learning_rate": 9.729748743718592e-05,
-      "loss": 6.0574,
-      "step": 3190
-    },
-    {
-      "epoch": 1.6641460234680574,
-      "grad_norm": 1.704528570175171,
-      "learning_rate": 9.72964824120603e-05,
-      "loss": 5.5729,
-      "step": 3191
-    },
-    {
-      "epoch": 1.6646675358539764,
-      "grad_norm": 1.463706374168396,
-      "learning_rate": 9.729547738693468e-05,
-      "loss": 5.8071,
-      "step": 3192
-    },
-    {
-      "epoch": 1.6651890482398957,
-      "grad_norm": 1.5594342947006226,
-      "learning_rate": 9.729447236180906e-05,
-      "loss": 6.0381,
-      "step": 3193
-    },
-    {
-      "epoch": 1.665710560625815,
-      "grad_norm": 1.6204968690872192,
-      "learning_rate": 9.729346733668342e-05,
-      "loss": 5.994,
-      "step": 3194
-    },
-    {
-      "epoch": 1.666232073011734,
-      "grad_norm": 1.850070834159851,
-      "learning_rate": 9.72924623115578e-05,
-      "loss": 6.234,
-      "step": 3195
-    },
-    {
-      "epoch": 1.6667535853976532,
-      "grad_norm": 1.6641219854354858,
-      "learning_rate": 9.729145728643216e-05,
-      "loss": 5.5622,
-      "step": 3196
-    },
-    {
-      "epoch": 1.6672750977835724,
-      "grad_norm": 1.4465763568878174,
-      "learning_rate": 9.729045226130654e-05,
-      "loss": 5.7803,
-      "step": 3197
-    },
-    {
-      "epoch": 1.6677966101694914,
-      "grad_norm": 1.6652206182479858,
-      "learning_rate": 9.72894472361809e-05,
-      "loss": 5.7948,
-      "step": 3198
-    },
-    {
-      "epoch": 1.6683181225554107,
-      "grad_norm": 1.6243849992752075,
-      "learning_rate": 9.728844221105528e-05,
-      "loss": 6.0146,
-      "step": 3199
-    },
-    {
-      "epoch": 1.66883963494133,
-      "grad_norm": 1.4728500843048096,
-      "learning_rate": 9.728743718592965e-05,
-      "loss": 5.8717,
-      "step": 3200
-    },
-    {
-      "epoch": 1.669361147327249,
-      "grad_norm": 1.6355260610580444,
-      "learning_rate": 9.728643216080403e-05,
-      "loss": 6.2207,
-      "step": 3201
-    },
-    {
-      "epoch": 1.6698826597131682,
-      "grad_norm": 1.5135046243667603,
-      "learning_rate": 9.72854271356784e-05,
-      "loss": 6.2645,
-      "step": 3202
-    },
-    {
-      "epoch": 1.6704041720990874,
-      "grad_norm": 1.6245781183242798,
-      "learning_rate": 9.728442211055277e-05,
-      "loss": 5.8798,
-      "step": 3203
-    },
-    {
-      "epoch": 1.6709256844850064,
-      "grad_norm": 1.7407125234603882,
-      "learning_rate": 9.728341708542715e-05,
-      "loss": 5.881,
-      "step": 3204
-    },
-    {
-      "epoch": 1.6714471968709257,
-      "grad_norm": 2.1430344581604004,
-      "learning_rate": 9.728241206030151e-05,
-      "loss": 5.3271,
-      "step": 3205
-    },
-    {
-      "epoch": 1.671968709256845,
-      "grad_norm": 1.6417948007583618,
-      "learning_rate": 9.728140703517589e-05,
-      "loss": 5.8358,
-      "step": 3206
-    },
-    {
-      "epoch": 1.672490221642764,
-      "grad_norm": 1.7306737899780273,
-      "learning_rate": 9.728040201005025e-05,
-      "loss": 6.0933,
-      "step": 3207
-    },
-    {
-      "epoch": 1.6730117340286832,
-      "grad_norm": 1.6175457239151,
-      "learning_rate": 9.727939698492463e-05,
-      "loss": 6.1129,
-      "step": 3208
-    },
-    {
-      "epoch": 1.6735332464146024,
-      "grad_norm": 1.6006898880004883,
-      "learning_rate": 9.7278391959799e-05,
-      "loss": 5.7058,
-      "step": 3209
-    },
-    {
-      "epoch": 1.6740547588005215,
-      "grad_norm": 1.5916205644607544,
-      "learning_rate": 9.727738693467337e-05,
-      "loss": 5.6739,
-      "step": 3210
-    },
-    {
-      "epoch": 1.6745762711864407,
-      "grad_norm": 1.5880261659622192,
-      "learning_rate": 9.727638190954774e-05,
-      "loss": 6.168,
-      "step": 3211
-    },
-    {
-      "epoch": 1.67509778357236,
-      "grad_norm": 1.5823211669921875,
-      "learning_rate": 9.727537688442211e-05,
-      "loss": 5.5296,
-      "step": 3212
-    },
-    {
-      "epoch": 1.675619295958279,
-      "grad_norm": 1.749687671661377,
-      "learning_rate": 9.727437185929649e-05,
-      "loss": 6.1563,
-      "step": 3213
-    },
-    {
-      "epoch": 1.6761408083441982,
-      "grad_norm": 1.472648024559021,
-      "learning_rate": 9.727336683417086e-05,
-      "loss": 6.03,
-      "step": 3214
-    },
-    {
-      "epoch": 1.6766623207301175,
-      "grad_norm": 1.5022220611572266,
-      "learning_rate": 9.727236180904523e-05,
-      "loss": 6.3691,
-      "step": 3215
-    },
-    {
-      "epoch": 1.6771838331160365,
-      "grad_norm": 1.4343161582946777,
-      "learning_rate": 9.72713567839196e-05,
-      "loss": 6.2584,
-      "step": 3216
-    },
-    {
-      "epoch": 1.6777053455019557,
-      "grad_norm": 1.6226125955581665,
-      "learning_rate": 9.727035175879398e-05,
-      "loss": 5.626,
-      "step": 3217
-    },
-    {
-      "epoch": 1.678226857887875,
-      "grad_norm": 1.6637554168701172,
-      "learning_rate": 9.726934673366834e-05,
-      "loss": 6.1724,
-      "step": 3218
-    },
-    {
-      "epoch": 1.678748370273794,
-      "grad_norm": 1.526485800743103,
-      "learning_rate": 9.726834170854272e-05,
-      "loss": 5.9241,
-      "step": 3219
-    },
-    {
-      "epoch": 1.6792698826597132,
-      "grad_norm": 1.4458792209625244,
-      "learning_rate": 9.726733668341708e-05,
-      "loss": 6.0839,
-      "step": 3220
-    },
-    {
-      "epoch": 1.6797913950456325,
-      "grad_norm": 1.759611964225769,
-      "learning_rate": 9.726633165829146e-05,
-      "loss": 6.1272,
-      "step": 3221
-    },
-    {
-      "epoch": 1.6803129074315515,
-      "grad_norm": 1.697718620300293,
-      "learning_rate": 9.726532663316584e-05,
-      "loss": 5.2536,
-      "step": 3222
-    },
-    {
-      "epoch": 1.6808344198174705,
-      "grad_norm": 1.6743053197860718,
-      "learning_rate": 9.726432160804022e-05,
-      "loss": 5.4143,
-      "step": 3223
-    },
-    {
-      "epoch": 1.68135593220339,
-      "grad_norm": 1.784546971321106,
-      "learning_rate": 9.726331658291458e-05,
-      "loss": 6.0877,
-      "step": 3224
-    },
-    {
-      "epoch": 1.681877444589309,
-      "grad_norm": 1.6950150728225708,
-      "learning_rate": 9.726231155778896e-05,
-      "loss": 5.778,
-      "step": 3225
-    },
-    {
-      "epoch": 1.682398956975228,
-      "grad_norm": 1.5330315828323364,
-      "learning_rate": 9.726130653266332e-05,
-      "loss": 6.0452,
-      "step": 3226
-    },
-    {
-      "epoch": 1.6829204693611475,
-      "grad_norm": 1.6510361433029175,
-      "learning_rate": 9.726030150753769e-05,
-      "loss": 6.0363,
-      "step": 3227
-    },
-    {
-      "epoch": 1.6834419817470665,
-      "grad_norm": 1.49409818649292,
-      "learning_rate": 9.725929648241207e-05,
-      "loss": 5.5592,
-      "step": 3228
-    },
-    {
-      "epoch": 1.6839634941329855,
-      "grad_norm": 1.4729665517807007,
-      "learning_rate": 9.725829145728643e-05,
-      "loss": 6.1572,
-      "step": 3229
-    },
-    {
-      "epoch": 1.684485006518905,
-      "grad_norm": 1.8135627508163452,
-      "learning_rate": 9.725728643216081e-05,
-      "loss": 5.8689,
-      "step": 3230
-    },
-    {
-      "epoch": 1.685006518904824,
-      "grad_norm": 1.5554484128952026,
-      "learning_rate": 9.725628140703517e-05,
-      "loss": 6.1711,
-      "step": 3231
-    },
-    {
-      "epoch": 1.685528031290743,
-      "grad_norm": 1.8536498546600342,
-      "learning_rate": 9.725527638190955e-05,
-      "loss": 5.6095,
-      "step": 3232
-    },
-    {
-      "epoch": 1.6860495436766623,
-      "grad_norm": 1.6792150735855103,
-      "learning_rate": 9.725427135678393e-05,
-      "loss": 6.177,
-      "step": 3233
-    },
-    {
-      "epoch": 1.6865710560625815,
-      "grad_norm": 1.4636952877044678,
-      "learning_rate": 9.72532663316583e-05,
-      "loss": 5.9558,
-      "step": 3234
-    },
-    {
-      "epoch": 1.6870925684485005,
-      "grad_norm": 1.5282926559448242,
-      "learning_rate": 9.725226130653267e-05,
-      "loss": 6.1731,
-      "step": 3235
-    },
-    {
-      "epoch": 1.6876140808344198,
-      "grad_norm": 1.7123357057571411,
-      "learning_rate": 9.725125628140705e-05,
-      "loss": 5.9655,
-      "step": 3236
-    },
-    {
-      "epoch": 1.688135593220339,
-      "grad_norm": 1.8475371599197388,
-      "learning_rate": 9.725025125628141e-05,
-      "loss": 6.1073,
-      "step": 3237
-    },
-    {
-      "epoch": 1.688657105606258,
-      "grad_norm": 1.6527007818222046,
-      "learning_rate": 9.724924623115579e-05,
-      "loss": 5.6579,
-      "step": 3238
-    },
-    {
-      "epoch": 1.6891786179921773,
-      "grad_norm": 1.5052242279052734,
-      "learning_rate": 9.724824120603015e-05,
-      "loss": 6.1953,
-      "step": 3239
-    },
-    {
-      "epoch": 1.6897001303780965,
-      "grad_norm": 1.5291155576705933,
-      "learning_rate": 9.724723618090452e-05,
-      "loss": 6.3444,
-      "step": 3240
-    },
-    {
-      "epoch": 1.6902216427640155,
-      "grad_norm": 2.2314364910125732,
-      "learning_rate": 9.72462311557789e-05,
-      "loss": 4.9503,
-      "step": 3241
-    },
-    {
-      "epoch": 1.6907431551499348,
-      "grad_norm": 1.610428810119629,
-      "learning_rate": 9.724522613065326e-05,
-      "loss": 6.2835,
-      "step": 3242
-    },
-    {
-      "epoch": 1.691264667535854,
-      "grad_norm": 1.706899642944336,
-      "learning_rate": 9.724422110552764e-05,
-      "loss": 5.3784,
-      "step": 3243
-    },
-    {
-      "epoch": 1.691786179921773,
-      "grad_norm": 1.4577761888504028,
-      "learning_rate": 9.724321608040202e-05,
-      "loss": 5.8671,
-      "step": 3244
-    },
-    {
-      "epoch": 1.6923076923076923,
-      "grad_norm": 1.4243916273117065,
-      "learning_rate": 9.724221105527639e-05,
-      "loss": 6.0417,
-      "step": 3245
-    },
-    {
-      "epoch": 1.6928292046936115,
-      "grad_norm": 1.9628136157989502,
-      "learning_rate": 9.724120603015076e-05,
-      "loss": 5.5775,
-      "step": 3246
-    },
-    {
-      "epoch": 1.6933507170795306,
-      "grad_norm": 2.1769423484802246,
-      "learning_rate": 9.724020100502514e-05,
-      "loss": 5.0881,
-      "step": 3247
-    },
-    {
-      "epoch": 1.6938722294654498,
-      "grad_norm": 1.6167597770690918,
-      "learning_rate": 9.72391959798995e-05,
-      "loss": 6.1707,
-      "step": 3248
-    },
-    {
-      "epoch": 1.694393741851369,
-      "grad_norm": 1.4802266359329224,
-      "learning_rate": 9.723819095477388e-05,
-      "loss": 6.2948,
-      "step": 3249
-    },
-    {
-      "epoch": 1.694915254237288,
-      "grad_norm": 1.6667572259902954,
-      "learning_rate": 9.723718592964824e-05,
-      "loss": 5.7585,
-      "step": 3250
-    },
-    {
-      "epoch": 1.6954367666232073,
-      "grad_norm": 1.5481176376342773,
-      "learning_rate": 9.723618090452262e-05,
-      "loss": 5.9769,
-      "step": 3251
-    },
-    {
-      "epoch": 1.6959582790091265,
-      "grad_norm": 1.5769729614257812,
-      "learning_rate": 9.723517587939698e-05,
-      "loss": 6.1965,
-      "step": 3252
-    },
-    {
-      "epoch": 1.6964797913950456,
-      "grad_norm": 1.7354010343551636,
-      "learning_rate": 9.723417085427136e-05,
-      "loss": 5.8371,
-      "step": 3253
-    },
-    {
-      "epoch": 1.6970013037809648,
-      "grad_norm": 1.6017951965332031,
-      "learning_rate": 9.723316582914574e-05,
-      "loss": 5.5837,
-      "step": 3254
-    },
-    {
-      "epoch": 1.697522816166884,
-      "grad_norm": 1.2437846660614014,
-      "learning_rate": 9.72321608040201e-05,
-      "loss": 6.2058,
-      "step": 3255
-    },
-    {
-      "epoch": 1.698044328552803,
-      "grad_norm": 1.50468909740448,
-      "learning_rate": 9.723115577889448e-05,
-      "loss": 6.2282,
-      "step": 3256
-    },
-    {
-      "epoch": 1.6985658409387223,
-      "grad_norm": 1.3210327625274658,
-      "learning_rate": 9.723015075376885e-05,
-      "loss": 6.1736,
-      "step": 3257
-    },
-    {
-      "epoch": 1.6990873533246416,
-      "grad_norm": 1.5534708499908447,
-      "learning_rate": 9.722914572864322e-05,
-      "loss": 5.6591,
-      "step": 3258
-    },
-    {
-      "epoch": 1.6996088657105606,
-      "grad_norm": 1.5405446290969849,
-      "learning_rate": 9.722814070351759e-05,
-      "loss": 5.7173,
-      "step": 3259
-    },
-    {
-      "epoch": 1.7001303780964798,
-      "grad_norm": 1.3668922185897827,
-      "learning_rate": 9.722713567839197e-05,
-      "loss": 6.2671,
-      "step": 3260
-    },
-    {
-      "epoch": 1.700651890482399,
-      "grad_norm": 1.8384426832199097,
-      "learning_rate": 9.722613065326633e-05,
-      "loss": 6.1787,
-      "step": 3261
-    },
-    {
-      "epoch": 1.701173402868318,
-      "grad_norm": 1.519514799118042,
-      "learning_rate": 9.722512562814071e-05,
-      "loss": 6.1819,
-      "step": 3262
-    },
-    {
-      "epoch": 1.7016949152542373,
-      "grad_norm": 1.448196530342102,
-      "learning_rate": 9.722412060301507e-05,
-      "loss": 5.9922,
-      "step": 3263
-    },
-    {
-      "epoch": 1.7022164276401566,
-      "grad_norm": 1.866915225982666,
-      "learning_rate": 9.722311557788945e-05,
-      "loss": 5.6248,
-      "step": 3264
-    },
-    {
-      "epoch": 1.7027379400260756,
-      "grad_norm": 1.9794881343841553,
-      "learning_rate": 9.722211055276383e-05,
-      "loss": 6.1928,
-      "step": 3265
-    },
-    {
-      "epoch": 1.7032594524119948,
-      "grad_norm": 1.7719507217407227,
-      "learning_rate": 9.72211055276382e-05,
-      "loss": 5.8398,
-      "step": 3266
-    },
-    {
-      "epoch": 1.703780964797914,
-      "grad_norm": 1.4759310483932495,
-      "learning_rate": 9.722010050251257e-05,
-      "loss": 6.2204,
-      "step": 3267
-    },
-    {
-      "epoch": 1.704302477183833,
-      "grad_norm": 1.6018708944320679,
-      "learning_rate": 9.721909547738693e-05,
-      "loss": 5.3853,
-      "step": 3268
-    },
-    {
-      "epoch": 1.7048239895697521,
-      "grad_norm": 1.46084725856781,
-      "learning_rate": 9.721809045226131e-05,
-      "loss": 6.1681,
-      "step": 3269
-    },
-    {
-      "epoch": 1.7053455019556716,
-      "grad_norm": 1.3885011672973633,
-      "learning_rate": 9.721708542713568e-05,
-      "loss": 6.2089,
-      "step": 3270
-    },
-    {
-      "epoch": 1.7058670143415906,
-      "grad_norm": 1.4984090328216553,
-      "learning_rate": 9.721608040201005e-05,
-      "loss": 6.1931,
-      "step": 3271
-    },
-    {
-      "epoch": 1.7063885267275096,
-      "grad_norm": 1.3669795989990234,
-      "learning_rate": 9.721507537688442e-05,
-      "loss": 6.1169,
-      "step": 3272
-    },
-    {
-      "epoch": 1.706910039113429,
-      "grad_norm": 1.523040533065796,
-      "learning_rate": 9.72140703517588e-05,
-      "loss": 5.8951,
-      "step": 3273
-    },
-    {
-      "epoch": 1.707431551499348,
-      "grad_norm": 1.7961832284927368,
-      "learning_rate": 9.721306532663317e-05,
-      "loss": 5.721,
-      "step": 3274
-    },
-    {
-      "epoch": 1.7079530638852671,
-      "grad_norm": 1.6705384254455566,
-      "learning_rate": 9.721206030150755e-05,
-      "loss": 5.9702,
-      "step": 3275
-    },
-    {
-      "epoch": 1.7084745762711866,
-      "grad_norm": 1.9442307949066162,
-      "learning_rate": 9.721105527638192e-05,
-      "loss": 5.9098,
-      "step": 3276
-    },
-    {
-      "epoch": 1.7089960886571056,
-      "grad_norm": 1.4218543767929077,
-      "learning_rate": 9.72100502512563e-05,
-      "loss": 6.2064,
-      "step": 3277
-    },
-    {
-      "epoch": 1.7095176010430246,
-      "grad_norm": 1.6957820653915405,
-      "learning_rate": 9.720904522613066e-05,
-      "loss": 5.8875,
-      "step": 3278
-    },
-    {
-      "epoch": 1.710039113428944,
-      "grad_norm": 1.9108747243881226,
-      "learning_rate": 9.720804020100504e-05,
-      "loss": 5.4027,
-      "step": 3279
-    },
-    {
-      "epoch": 1.7105606258148631,
-      "grad_norm": 1.7203845977783203,
-      "learning_rate": 9.72070351758794e-05,
-      "loss": 5.8904,
-      "step": 3280
-    },
-    {
-      "epoch": 1.7110821382007821,
-      "grad_norm": 1.6585044860839844,
-      "learning_rate": 9.720603015075376e-05,
-      "loss": 6.0254,
-      "step": 3281
-    },
-    {
-      "epoch": 1.7116036505867014,
-      "grad_norm": 1.764907717704773,
-      "learning_rate": 9.720502512562814e-05,
-      "loss": 5.9133,
-      "step": 3282
-    },
-    {
-      "epoch": 1.7121251629726206,
-      "grad_norm": 1.5757007598876953,
-      "learning_rate": 9.72040201005025e-05,
-      "loss": 5.8941,
-      "step": 3283
-    },
-    {
-      "epoch": 1.7126466753585396,
-      "grad_norm": 1.5602794885635376,
-      "learning_rate": 9.720301507537688e-05,
-      "loss": 5.9031,
-      "step": 3284
-    },
-    {
-      "epoch": 1.713168187744459,
-      "grad_norm": 1.8277525901794434,
-      "learning_rate": 9.720201005025126e-05,
-      "loss": 5.226,
-      "step": 3285
-    },
-    {
-      "epoch": 1.7136897001303781,
-      "grad_norm": 1.4694033861160278,
-      "learning_rate": 9.720100502512564e-05,
-      "loss": 6.3207,
-      "step": 3286
-    },
-    {
-      "epoch": 1.7142112125162972,
-      "grad_norm": 1.516701579093933,
-      "learning_rate": 9.72e-05,
-      "loss": 6.1813,
-      "step": 3287
-    },
-    {
-      "epoch": 1.7147327249022164,
-      "grad_norm": 1.7726422548294067,
-      "learning_rate": 9.719899497487438e-05,
-      "loss": 6.0361,
-      "step": 3288
-    },
-    {
-      "epoch": 1.7152542372881356,
-      "grad_norm": 1.670492172241211,
-      "learning_rate": 9.719798994974875e-05,
-      "loss": 5.9087,
-      "step": 3289
-    },
-    {
-      "epoch": 1.7157757496740547,
-      "grad_norm": 1.8022445440292358,
-      "learning_rate": 9.719698492462312e-05,
-      "loss": 5.8934,
-      "step": 3290
-    },
-    {
-      "epoch": 1.716297262059974,
-      "grad_norm": 1.7031400203704834,
-      "learning_rate": 9.719597989949749e-05,
-      "loss": 6.0479,
-      "step": 3291
-    },
-    {
-      "epoch": 1.7168187744458931,
-      "grad_norm": 1.783332109451294,
-      "learning_rate": 9.719497487437187e-05,
-      "loss": 5.9079,
-      "step": 3292
-    },
-    {
-      "epoch": 1.7173402868318122,
-      "grad_norm": 1.495786190032959,
-      "learning_rate": 9.719396984924623e-05,
-      "loss": 6.118,
-      "step": 3293
-    },
-    {
-      "epoch": 1.7178617992177314,
-      "grad_norm": 1.5326268672943115,
-      "learning_rate": 9.719296482412061e-05,
-      "loss": 6.096,
-      "step": 3294
-    },
-    {
-      "epoch": 1.7183833116036507,
-      "grad_norm": 1.570668339729309,
-      "learning_rate": 9.719195979899499e-05,
-      "loss": 6.2606,
-      "step": 3295
-    },
-    {
-      "epoch": 1.7189048239895697,
-      "grad_norm": 1.4823070764541626,
-      "learning_rate": 9.719095477386935e-05,
-      "loss": 6.1616,
-      "step": 3296
-    },
-    {
-      "epoch": 1.719426336375489,
-      "grad_norm": 1.6309150457382202,
-      "learning_rate": 9.718994974874373e-05,
-      "loss": 5.8717,
-      "step": 3297
-    },
-    {
-      "epoch": 1.7199478487614082,
-      "grad_norm": 1.5561985969543457,
-      "learning_rate": 9.718894472361809e-05,
-      "loss": 6.2413,
-      "step": 3298
-    },
-    {
-      "epoch": 1.7204693611473272,
-      "grad_norm": 1.670965552330017,
-      "learning_rate": 9.718793969849247e-05,
-      "loss": 6.2541,
-      "step": 3299
-    },
-    {
-      "epoch": 1.7209908735332464,
-      "grad_norm": 1.5011696815490723,
-      "learning_rate": 9.718693467336683e-05,
-      "loss": 5.779,
-      "step": 3300
-    },
-    {
-      "epoch": 1.7215123859191657,
-      "grad_norm": 1.6187682151794434,
-      "learning_rate": 9.718592964824121e-05,
-      "loss": 5.8724,
-      "step": 3301
-    },
-    {
-      "epoch": 1.7220338983050847,
-      "grad_norm": 1.737080454826355,
-      "learning_rate": 9.718492462311558e-05,
-      "loss": 6.0139,
-      "step": 3302
-    },
-    {
-      "epoch": 1.722555410691004,
-      "grad_norm": 3.3389365673065186,
-      "learning_rate": 9.718391959798995e-05,
-      "loss": 5.8996,
-      "step": 3303
-    },
-    {
-      "epoch": 1.7230769230769232,
-      "grad_norm": 1.7126438617706299,
-      "learning_rate": 9.718291457286432e-05,
-      "loss": 6.0762,
-      "step": 3304
-    },
-    {
-      "epoch": 1.7235984354628422,
-      "grad_norm": 1.518081784248352,
-      "learning_rate": 9.71819095477387e-05,
-      "loss": 6.3283,
-      "step": 3305
-    },
-    {
-      "epoch": 1.7241199478487614,
-      "grad_norm": 1.5217375755310059,
-      "learning_rate": 9.718090452261307e-05,
-      "loss": 5.7146,
-      "step": 3306
-    },
-    {
-      "epoch": 1.7246414602346807,
-      "grad_norm": 1.6588184833526611,
-      "learning_rate": 9.717989949748744e-05,
-      "loss": 4.9533,
-      "step": 3307
-    },
-    {
-      "epoch": 1.7251629726205997,
-      "grad_norm": 1.5553386211395264,
-      "learning_rate": 9.717889447236182e-05,
-      "loss": 6.3043,
-      "step": 3308
-    },
-    {
-      "epoch": 1.725684485006519,
-      "grad_norm": 1.6457551717758179,
-      "learning_rate": 9.717788944723618e-05,
-      "loss": 5.8313,
-      "step": 3309
-    },
-    {
-      "epoch": 1.7262059973924382,
-      "grad_norm": 2.029294729232788,
-      "learning_rate": 9.717688442211056e-05,
-      "loss": 5.214,
-      "step": 3310
-    },
-    {
-      "epoch": 1.7267275097783572,
-      "grad_norm": 1.8718783855438232,
-      "learning_rate": 9.717587939698492e-05,
-      "loss": 6.2013,
-      "step": 3311
-    },
-    {
-      "epoch": 1.7272490221642764,
-      "grad_norm": 1.6630576848983765,
-      "learning_rate": 9.71748743718593e-05,
-      "loss": 6.0053,
-      "step": 3312
-    },
-    {
-      "epoch": 1.7277705345501957,
-      "grad_norm": 1.5276439189910889,
-      "learning_rate": 9.717386934673367e-05,
-      "loss": 6.0402,
-      "step": 3313
-    },
-    {
-      "epoch": 1.7282920469361147,
-      "grad_norm": 1.8371270895004272,
-      "learning_rate": 9.717286432160804e-05,
-      "loss": 6.0884,
-      "step": 3314
-    },
-    {
-      "epoch": 1.7288135593220337,
-      "grad_norm": 1.7210265398025513,
-      "learning_rate": 9.717185929648242e-05,
-      "loss": 5.9401,
-      "step": 3315
-    },
-    {
-      "epoch": 1.7293350717079532,
-      "grad_norm": 2.0925912857055664,
-      "learning_rate": 9.71708542713568e-05,
-      "loss": 5.7091,
-      "step": 3316
-    },
-    {
-      "epoch": 1.7298565840938722,
-      "grad_norm": 1.5949950218200684,
-      "learning_rate": 9.716984924623116e-05,
-      "loss": 5.8663,
-      "step": 3317
-    },
-    {
-      "epoch": 1.7303780964797912,
-      "grad_norm": 1.8782790899276733,
-      "learning_rate": 9.716884422110554e-05,
-      "loss": 5.7639,
-      "step": 3318
-    },
-    {
-      "epoch": 1.7308996088657107,
-      "grad_norm": 1.514481782913208,
-      "learning_rate": 9.71678391959799e-05,
-      "loss": 6.0857,
-      "step": 3319
-    },
-    {
-      "epoch": 1.7314211212516297,
-      "grad_norm": 1.7982304096221924,
-      "learning_rate": 9.716683417085427e-05,
-      "loss": 5.6281,
-      "step": 3320
-    },
-    {
-      "epoch": 1.7319426336375487,
-      "grad_norm": 1.679223895072937,
-      "learning_rate": 9.716582914572865e-05,
-      "loss": 5.1745,
-      "step": 3321
-    },
-    {
-      "epoch": 1.7324641460234682,
-      "grad_norm": 2.737560272216797,
-      "learning_rate": 9.716482412060301e-05,
-      "loss": 5.4417,
-      "step": 3322
-    },
-    {
-      "epoch": 1.7329856584093872,
-      "grad_norm": 1.8556020259857178,
-      "learning_rate": 9.716381909547739e-05,
-      "loss": 5.3947,
-      "step": 3323
-    },
-    {
-      "epoch": 1.7335071707953063,
-      "grad_norm": 1.6604238748550415,
-      "learning_rate": 9.716281407035175e-05,
-      "loss": 6.0392,
-      "step": 3324
-    },
-    {
-      "epoch": 1.7340286831812257,
-      "grad_norm": 1.582079529762268,
-      "learning_rate": 9.716180904522613e-05,
-      "loss": 4.9405,
-      "step": 3325
-    },
-    {
-      "epoch": 1.7345501955671447,
-      "grad_norm": 1.5796748399734497,
-      "learning_rate": 9.716080402010051e-05,
-      "loss": 6.1195,
-      "step": 3326
-    },
-    {
-      "epoch": 1.7350717079530638,
-      "grad_norm": 1.5330551862716675,
-      "learning_rate": 9.715979899497489e-05,
-      "loss": 6.4154,
-      "step": 3327
-    },
-    {
-      "epoch": 1.735593220338983,
-      "grad_norm": 1.6535720825195312,
-      "learning_rate": 9.715879396984925e-05,
-      "loss": 6.2119,
-      "step": 3328
-    },
-    {
-      "epoch": 1.7361147327249022,
-      "grad_norm": 1.4664764404296875,
-      "learning_rate": 9.715778894472363e-05,
-      "loss": 6.0851,
-      "step": 3329
-    },
-    {
-      "epoch": 1.7366362451108213,
-      "grad_norm": 1.472206950187683,
-      "learning_rate": 9.7156783919598e-05,
-      "loss": 5.9087,
-      "step": 3330
-    },
-    {
-      "epoch": 1.7371577574967405,
-      "grad_norm": 1.4665530920028687,
-      "learning_rate": 9.715577889447237e-05,
-      "loss": 6.3229,
-      "step": 3331
-    },
-    {
-      "epoch": 1.7376792698826597,
-      "grad_norm": 1.3584777116775513,
-      "learning_rate": 9.715477386934674e-05,
-      "loss": 5.3922,
-      "step": 3332
-    },
-    {
-      "epoch": 1.7382007822685788,
-      "grad_norm": 1.5922040939331055,
-      "learning_rate": 9.71537688442211e-05,
-      "loss": 5.9226,
-      "step": 3333
-    },
-    {
-      "epoch": 1.738722294654498,
-      "grad_norm": 1.6438482999801636,
-      "learning_rate": 9.715276381909548e-05,
-      "loss": 6.1637,
-      "step": 3334
-    },
-    {
-      "epoch": 1.7392438070404173,
-      "grad_norm": 1.4272111654281616,
-      "learning_rate": 9.715175879396986e-05,
-      "loss": 6.209,
-      "step": 3335
-    },
-    {
-      "epoch": 1.7397653194263363,
-      "grad_norm": 1.4767323732376099,
-      "learning_rate": 9.715075376884423e-05,
-      "loss": 5.3347,
-      "step": 3336
-    },
-    {
-      "epoch": 1.7402868318122555,
-      "grad_norm": 1.3619269132614136,
-      "learning_rate": 9.71497487437186e-05,
-      "loss": 6.0317,
-      "step": 3337
-    },
-    {
-      "epoch": 1.7408083441981748,
-      "grad_norm": 1.4023441076278687,
-      "learning_rate": 9.714874371859298e-05,
-      "loss": 6.1441,
-      "step": 3338
-    },
-    {
-      "epoch": 1.7413298565840938,
-      "grad_norm": 1.679480791091919,
-      "learning_rate": 9.714773869346734e-05,
-      "loss": 5.7566,
-      "step": 3339
-    },
-    {
-      "epoch": 1.741851368970013,
-      "grad_norm": 1.4111641645431519,
-      "learning_rate": 9.714673366834172e-05,
-      "loss": 6.1269,
-      "step": 3340
-    },
-    {
-      "epoch": 1.7423728813559323,
-      "grad_norm": 1.4637426137924194,
-      "learning_rate": 9.714572864321608e-05,
-      "loss": 6.3632,
-      "step": 3341
-    },
-    {
-      "epoch": 1.7428943937418513,
-      "grad_norm": 1.6454591751098633,
-      "learning_rate": 9.714472361809046e-05,
-      "loss": 5.5887,
-      "step": 3342
-    },
-    {
-      "epoch": 1.7434159061277705,
-      "grad_norm": 1.7154889106750488,
-      "learning_rate": 9.714371859296482e-05,
-      "loss": 6.1023,
-      "step": 3343
-    },
-    {
-      "epoch": 1.7439374185136898,
-      "grad_norm": 1.5314412117004395,
-      "learning_rate": 9.71427135678392e-05,
-      "loss": 6.0471,
-      "step": 3344
-    },
-    {
-      "epoch": 1.7444589308996088,
-      "grad_norm": 1.543838381767273,
-      "learning_rate": 9.714170854271357e-05,
-      "loss": 6.1968,
-      "step": 3345
-    },
-    {
-      "epoch": 1.744980443285528,
-      "grad_norm": 1.5923969745635986,
-      "learning_rate": 9.714070351758794e-05,
-      "loss": 5.5729,
-      "step": 3346
-    },
-    {
-      "epoch": 1.7455019556714473,
-      "grad_norm": 1.6715855598449707,
-      "learning_rate": 9.713969849246232e-05,
-      "loss": 6.4064,
-      "step": 3347
-    },
-    {
-      "epoch": 1.7460234680573663,
-      "grad_norm": 1.6406844854354858,
-      "learning_rate": 9.713869346733669e-05,
-      "loss": 5.9657,
-      "step": 3348
-    },
-    {
-      "epoch": 1.7465449804432855,
-      "grad_norm": 1.6807109117507935,
-      "learning_rate": 9.713768844221106e-05,
-      "loss": 5.4293,
-      "step": 3349
-    },
-    {
-      "epoch": 1.7470664928292048,
-      "grad_norm": 1.5913708209991455,
-      "learning_rate": 9.713668341708543e-05,
-      "loss": 6.1809,
-      "step": 3350
-    },
-    {
-      "epoch": 1.7475880052151238,
-      "grad_norm": 1.7646077871322632,
-      "learning_rate": 9.71356783919598e-05,
-      "loss": 5.509,
-      "step": 3351
-    },
-    {
-      "epoch": 1.748109517601043,
-      "grad_norm": 1.5574208498001099,
-      "learning_rate": 9.713467336683417e-05,
-      "loss": 5.5986,
-      "step": 3352
-    },
-    {
-      "epoch": 1.7486310299869623,
-      "grad_norm": 1.732813835144043,
-      "learning_rate": 9.713366834170855e-05,
-      "loss": 5.7303,
-      "step": 3353
-    },
-    {
-      "epoch": 1.7491525423728813,
-      "grad_norm": 1.608209252357483,
-      "learning_rate": 9.713266331658291e-05,
-      "loss": 6.1821,
-      "step": 3354
-    },
-    {
-      "epoch": 1.7496740547588006,
-      "grad_norm": 1.464309811592102,
-      "learning_rate": 9.713165829145729e-05,
-      "loss": 6.1374,
-      "step": 3355
-    },
-    {
-      "epoch": 1.7501955671447198,
-      "grad_norm": 1.570180892944336,
-      "learning_rate": 9.713065326633167e-05,
-      "loss": 6.1028,
-      "step": 3356
-    },
-    {
-      "epoch": 1.7507170795306388,
-      "grad_norm": 1.7300324440002441,
-      "learning_rate": 9.712964824120605e-05,
-      "loss": 5.7946,
-      "step": 3357
-    },
-    {
-      "epoch": 1.751238591916558,
-      "grad_norm": 1.4357713460922241,
-      "learning_rate": 9.712864321608041e-05,
-      "loss": 6.3007,
-      "step": 3358
-    },
-    {
-      "epoch": 1.7517601043024773,
-      "grad_norm": 1.6414669752120972,
-      "learning_rate": 9.712763819095479e-05,
-      "loss": 5.4064,
-      "step": 3359
-    },
-    {
-      "epoch": 1.7522816166883963,
-      "grad_norm": 1.6113923788070679,
-      "learning_rate": 9.712663316582915e-05,
-      "loss": 5.6724,
-      "step": 3360
-    },
-    {
-      "epoch": 1.7528031290743156,
-      "grad_norm": 1.573392391204834,
-      "learning_rate": 9.712562814070352e-05,
-      "loss": 6.0677,
-      "step": 3361
-    },
-    {
-      "epoch": 1.7533246414602348,
-      "grad_norm": 1.553194522857666,
-      "learning_rate": 9.71246231155779e-05,
-      "loss": 5.9444,
-      "step": 3362
-    },
-    {
-      "epoch": 1.7538461538461538,
-      "grad_norm": 1.4644821882247925,
-      "learning_rate": 9.712361809045226e-05,
-      "loss": 6.3711,
-      "step": 3363
-    },
-    {
-      "epoch": 1.7543676662320729,
-      "grad_norm": 1.4744619131088257,
-      "learning_rate": 9.712261306532664e-05,
-      "loss": 6.0969,
-      "step": 3364
-    },
-    {
-      "epoch": 1.7548891786179923,
-      "grad_norm": 1.6132123470306396,
-      "learning_rate": 9.7121608040201e-05,
-      "loss": 6.2089,
-      "step": 3365
-    },
-    {
-      "epoch": 1.7554106910039113,
-      "grad_norm": 1.6961557865142822,
-      "learning_rate": 9.712060301507538e-05,
-      "loss": 5.9151,
-      "step": 3366
-    },
-    {
-      "epoch": 1.7559322033898304,
-      "grad_norm": 1.5807000398635864,
-      "learning_rate": 9.711959798994976e-05,
-      "loss": 5.8969,
-      "step": 3367
-    },
-    {
-      "epoch": 1.7564537157757498,
-      "grad_norm": 1.5537190437316895,
-      "learning_rate": 9.711859296482413e-05,
-      "loss": 6.2752,
-      "step": 3368
-    },
-    {
-      "epoch": 1.7569752281616688,
-      "grad_norm": 1.529908299446106,
-      "learning_rate": 9.71175879396985e-05,
-      "loss": 6.1472,
-      "step": 3369
-    },
-    {
-      "epoch": 1.7574967405475879,
-      "grad_norm": 1.5507049560546875,
-      "learning_rate": 9.711658291457288e-05,
-      "loss": 6.1645,
-      "step": 3370
-    },
-    {
-      "epoch": 1.7580182529335073,
-      "grad_norm": 1.8250657320022583,
-      "learning_rate": 9.711557788944724e-05,
-      "loss": 6.0594,
-      "step": 3371
-    },
-    {
-      "epoch": 1.7585397653194264,
-      "grad_norm": 1.6674892902374268,
-      "learning_rate": 9.711457286432162e-05,
-      "loss": 5.9769,
-      "step": 3372
-    },
-    {
-      "epoch": 1.7590612777053454,
-      "grad_norm": 1.6310733556747437,
-      "learning_rate": 9.711356783919598e-05,
-      "loss": 5.8109,
-      "step": 3373
-    },
-    {
-      "epoch": 1.7595827900912646,
-      "grad_norm": 1.7524957656860352,
-      "learning_rate": 9.711256281407035e-05,
-      "loss": 6.1912,
-      "step": 3374
-    },
-    {
-      "epoch": 1.7601043024771839,
-      "grad_norm": 1.4561899900436401,
-      "learning_rate": 9.711155778894472e-05,
-      "loss": 6.1365,
-      "step": 3375
-    },
-    {
-      "epoch": 1.7606258148631029,
-      "grad_norm": 1.5613276958465576,
-      "learning_rate": 9.71105527638191e-05,
-      "loss": 5.9976,
-      "step": 3376
-    },
-    {
-      "epoch": 1.7611473272490221,
-      "grad_norm": 1.4331763982772827,
-      "learning_rate": 9.710954773869348e-05,
-      "loss": 6.2593,
-      "step": 3377
-    },
-    {
-      "epoch": 1.7616688396349414,
-      "grad_norm": 2.2194039821624756,
-      "learning_rate": 9.710854271356784e-05,
-      "loss": 5.8111,
-      "step": 3378
-    },
-    {
-      "epoch": 1.7621903520208604,
-      "grad_norm": 1.4964829683303833,
-      "learning_rate": 9.710753768844222e-05,
-      "loss": 6.0591,
-      "step": 3379
-    },
-    {
-      "epoch": 1.7627118644067796,
-      "grad_norm": 1.4492355585098267,
-      "learning_rate": 9.710653266331659e-05,
-      "loss": 6.2314,
-      "step": 3380
-    },
-    {
-      "epoch": 1.7632333767926989,
-      "grad_norm": 1.5991970300674438,
-      "learning_rate": 9.710552763819096e-05,
-      "loss": 6.2326,
-      "step": 3381
-    },
-    {
-      "epoch": 1.763754889178618,
-      "grad_norm": 1.5764954090118408,
-      "learning_rate": 9.710452261306533e-05,
-      "loss": 6.1551,
-      "step": 3382
-    },
-    {
-      "epoch": 1.7642764015645371,
-      "grad_norm": 1.803238034248352,
-      "learning_rate": 9.71035175879397e-05,
-      "loss": 5.6768,
-      "step": 3383
-    },
-    {
-      "epoch": 1.7647979139504564,
-      "grad_norm": 1.6471593379974365,
-      "learning_rate": 9.710251256281407e-05,
-      "loss": 5.4339,
-      "step": 3384
-    },
-    {
-      "epoch": 1.7653194263363754,
-      "grad_norm": 1.7341952323913574,
-      "learning_rate": 9.710150753768845e-05,
-      "loss": 6.0701,
-      "step": 3385
-    },
-    {
-      "epoch": 1.7658409387222946,
-      "grad_norm": 1.4746848344802856,
-      "learning_rate": 9.710050251256281e-05,
-      "loss": 6.2508,
-      "step": 3386
-    },
-    {
-      "epoch": 1.7663624511082139,
-      "grad_norm": 1.5259249210357666,
-      "learning_rate": 9.709949748743719e-05,
-      "loss": 5.7826,
-      "step": 3387
-    },
-    {
-      "epoch": 1.766883963494133,
-      "grad_norm": 1.4762948751449585,
-      "learning_rate": 9.709849246231157e-05,
-      "loss": 5.9028,
-      "step": 3388
-    },
-    {
-      "epoch": 1.7674054758800521,
-      "grad_norm": 1.5306518077850342,
-      "learning_rate": 9.709748743718593e-05,
-      "loss": 5.9101,
-      "step": 3389
-    },
-    {
-      "epoch": 1.7679269882659714,
-      "grad_norm": 1.5504504442214966,
-      "learning_rate": 9.709648241206031e-05,
-      "loss": 6.2105,
-      "step": 3390
-    },
-    {
-      "epoch": 1.7684485006518904,
-      "grad_norm": 1.6335573196411133,
-      "learning_rate": 9.709547738693468e-05,
-      "loss": 6.0706,
-      "step": 3391
-    },
-    {
-      "epoch": 1.7689700130378097,
-      "grad_norm": 1.774306058883667,
-      "learning_rate": 9.709447236180905e-05,
-      "loss": 5.4662,
-      "step": 3392
-    },
-    {
-      "epoch": 1.769491525423729,
-      "grad_norm": 1.3517709970474243,
-      "learning_rate": 9.709346733668342e-05,
-      "loss": 6.2963,
-      "step": 3393
-    },
-    {
-      "epoch": 1.770013037809648,
-      "grad_norm": 1.4020494222640991,
-      "learning_rate": 9.70924623115578e-05,
-      "loss": 6.194,
-      "step": 3394
-    },
-    {
-      "epoch": 1.7705345501955672,
-      "grad_norm": 1.624859094619751,
-      "learning_rate": 9.709145728643216e-05,
-      "loss": 5.4284,
-      "step": 3395
-    },
-    {
-      "epoch": 1.7710560625814864,
-      "grad_norm": 1.7723333835601807,
-      "learning_rate": 9.709045226130654e-05,
-      "loss": 6.0049,
-      "step": 3396
-    },
-    {
-      "epoch": 1.7715775749674054,
-      "grad_norm": 1.8757661581039429,
-      "learning_rate": 9.70894472361809e-05,
-      "loss": 6.0504,
-      "step": 3397
-    },
-    {
-      "epoch": 1.7720990873533247,
-      "grad_norm": 1.5301741361618042,
-      "learning_rate": 9.708844221105528e-05,
-      "loss": 6.1439,
-      "step": 3398
-    },
-    {
-      "epoch": 1.772620599739244,
-      "grad_norm": 1.6773927211761475,
-      "learning_rate": 9.708743718592966e-05,
-      "loss": 5.8625,
-      "step": 3399
-    },
-    {
-      "epoch": 1.773142112125163,
-      "grad_norm": 1.6877857446670532,
-      "learning_rate": 9.708643216080402e-05,
-      "loss": 5.5765,
-      "step": 3400
-    },
-    {
-      "epoch": 1.7736636245110822,
-      "grad_norm": 1.3745710849761963,
-      "learning_rate": 9.70854271356784e-05,
-      "loss": 6.0789,
-      "step": 3401
-    },
-    {
-      "epoch": 1.7741851368970014,
-      "grad_norm": 1.8772443532943726,
-      "learning_rate": 9.708442211055276e-05,
-      "loss": 5.6929,
-      "step": 3402
-    },
-    {
-      "epoch": 1.7747066492829204,
-      "grad_norm": 1.653212308883667,
-      "learning_rate": 9.708341708542714e-05,
-      "loss": 6.0236,
-      "step": 3403
-    },
-    {
-      "epoch": 1.7752281616688397,
-      "grad_norm": 1.6786621809005737,
-      "learning_rate": 9.70824120603015e-05,
-      "loss": 5.7432,
-      "step": 3404
-    },
-    {
-      "epoch": 1.775749674054759,
-      "grad_norm": 1.7271103858947754,
-      "learning_rate": 9.708140703517588e-05,
-      "loss": 5.6483,
-      "step": 3405
-    },
-    {
-      "epoch": 1.776271186440678,
-      "grad_norm": 1.6076699495315552,
-      "learning_rate": 9.708040201005025e-05,
-      "loss": 5.8815,
-      "step": 3406
-    },
-    {
-      "epoch": 1.7767926988265972,
-      "grad_norm": 1.5660747289657593,
-      "learning_rate": 9.707939698492463e-05,
-      "loss": 5.8855,
-      "step": 3407
-    },
-    {
-      "epoch": 1.7773142112125164,
-      "grad_norm": 1.5821980237960815,
-      "learning_rate": 9.7078391959799e-05,
-      "loss": 6.2526,
-      "step": 3408
-    },
-    {
-      "epoch": 1.7778357235984354,
-      "grad_norm": 1.4485617876052856,
-      "learning_rate": 9.707738693467338e-05,
-      "loss": 6.0431,
-      "step": 3409
-    },
-    {
-      "epoch": 1.7783572359843545,
-      "grad_norm": 1.438031792640686,
-      "learning_rate": 9.707638190954775e-05,
-      "loss": 6.095,
-      "step": 3410
-    },
-    {
-      "epoch": 1.778878748370274,
-      "grad_norm": 1.4515784978866577,
-      "learning_rate": 9.707537688442212e-05,
-      "loss": 6.0939,
-      "step": 3411
-    },
-    {
-      "epoch": 1.779400260756193,
-      "grad_norm": 1.772674560546875,
-      "learning_rate": 9.707437185929649e-05,
-      "loss": 5.6474,
-      "step": 3412
-    },
-    {
-      "epoch": 1.779921773142112,
-      "grad_norm": 1.5566948652267456,
-      "learning_rate": 9.707336683417085e-05,
-      "loss": 6.1361,
-      "step": 3413
-    },
-    {
-      "epoch": 1.7804432855280314,
-      "grad_norm": 1.7735984325408936,
-      "learning_rate": 9.707236180904523e-05,
-      "loss": 5.5231,
-      "step": 3414
-    },
-    {
-      "epoch": 1.7809647979139505,
-      "grad_norm": 1.4642142057418823,
-      "learning_rate": 9.70713567839196e-05,
-      "loss": 6.0834,
-      "step": 3415
-    },
-    {
-      "epoch": 1.7814863102998695,
-      "grad_norm": 1.6048023700714111,
-      "learning_rate": 9.707035175879397e-05,
-      "loss": 5.869,
-      "step": 3416
-    },
-    {
-      "epoch": 1.782007822685789,
-      "grad_norm": 1.6047550439834595,
-      "learning_rate": 9.706934673366834e-05,
-      "loss": 6.0868,
-      "step": 3417
-    },
-    {
-      "epoch": 1.782529335071708,
-      "grad_norm": 1.5889705419540405,
-      "learning_rate": 9.706834170854271e-05,
-      "loss": 6.1874,
-      "step": 3418
-    },
-    {
-      "epoch": 1.783050847457627,
-      "grad_norm": 1.7397993803024292,
-      "learning_rate": 9.706733668341709e-05,
-      "loss": 5.8478,
-      "step": 3419
-    },
-    {
-      "epoch": 1.7835723598435462,
-      "grad_norm": 1.524584412574768,
-      "learning_rate": 9.706633165829147e-05,
-      "loss": 6.2041,
-      "step": 3420
-    },
-    {
-      "epoch": 1.7840938722294655,
-      "grad_norm": 1.886851191520691,
-      "learning_rate": 9.706532663316583e-05,
-      "loss": 6.0316,
-      "step": 3421
-    },
-    {
-      "epoch": 1.7846153846153845,
-      "grad_norm": 1.659754991531372,
-      "learning_rate": 9.706432160804021e-05,
-      "loss": 6.0272,
-      "step": 3422
-    },
-    {
-      "epoch": 1.7851368970013037,
-      "grad_norm": 1.5530073642730713,
-      "learning_rate": 9.706331658291458e-05,
-      "loss": 6.065,
-      "step": 3423
-    },
-    {
-      "epoch": 1.785658409387223,
-      "grad_norm": 1.7824958562850952,
-      "learning_rate": 9.706231155778895e-05,
-      "loss": 5.8887,
-      "step": 3424
-    },
-    {
-      "epoch": 1.786179921773142,
-      "grad_norm": 1.7361865043640137,
-      "learning_rate": 9.706130653266332e-05,
-      "loss": 5.7271,
-      "step": 3425
-    },
-    {
-      "epoch": 1.7867014341590612,
-      "grad_norm": 1.4935905933380127,
-      "learning_rate": 9.706030150753768e-05,
-      "loss": 6.2631,
-      "step": 3426
-    },
-    {
-      "epoch": 1.7872229465449805,
-      "grad_norm": 1.4938560724258423,
-      "learning_rate": 9.705929648241206e-05,
-      "loss": 6.0801,
-      "step": 3427
-    },
-    {
-      "epoch": 1.7877444589308995,
-      "grad_norm": 1.551841139793396,
-      "learning_rate": 9.705829145728644e-05,
-      "loss": 5.8064,
-      "step": 3428
-    },
-    {
-      "epoch": 1.7882659713168187,
-      "grad_norm": 1.4372341632843018,
-      "learning_rate": 9.705728643216082e-05,
-      "loss": 6.0318,
-      "step": 3429
-    },
-    {
-      "epoch": 1.788787483702738,
-      "grad_norm": 1.7153056859970093,
-      "learning_rate": 9.705628140703518e-05,
-      "loss": 5.5594,
-      "step": 3430
-    },
-    {
-      "epoch": 1.789308996088657,
-      "grad_norm": 1.4397196769714355,
-      "learning_rate": 9.705527638190956e-05,
-      "loss": 5.8875,
-      "step": 3431
-    },
-    {
-      "epoch": 1.7898305084745763,
-      "grad_norm": 1.4717869758605957,
-      "learning_rate": 9.705427135678392e-05,
-      "loss": 6.0481,
-      "step": 3432
-    },
-    {
-      "epoch": 1.7903520208604955,
-      "grad_norm": 1.4425760507583618,
-      "learning_rate": 9.70532663316583e-05,
-      "loss": 6.4561,
-      "step": 3433
-    },
-    {
-      "epoch": 1.7908735332464145,
-      "grad_norm": 1.4299579858779907,
-      "learning_rate": 9.705226130653266e-05,
-      "loss": 5.8588,
-      "step": 3434
-    },
-    {
-      "epoch": 1.7913950456323338,
-      "grad_norm": 1.4884129762649536,
-      "learning_rate": 9.705125628140704e-05,
-      "loss": 5.8936,
-      "step": 3435
-    },
-    {
-      "epoch": 1.791916558018253,
-      "grad_norm": 1.5652891397476196,
-      "learning_rate": 9.70502512562814e-05,
-      "loss": 5.9654,
-      "step": 3436
-    },
-    {
-      "epoch": 1.792438070404172,
-      "grad_norm": 1.4190223217010498,
-      "learning_rate": 9.704924623115578e-05,
-      "loss": 6.109,
-      "step": 3437
-    },
-    {
-      "epoch": 1.7929595827900913,
-      "grad_norm": 1.406278371810913,
-      "learning_rate": 9.704824120603015e-05,
-      "loss": 6.2161,
-      "step": 3438
-    },
-    {
-      "epoch": 1.7934810951760105,
-      "grad_norm": 1.5271124839782715,
-      "learning_rate": 9.704723618090453e-05,
-      "loss": 6.2097,
-      "step": 3439
-    },
-    {
-      "epoch": 1.7940026075619295,
-      "grad_norm": 1.7708873748779297,
-      "learning_rate": 9.70462311557789e-05,
-      "loss": 5.065,
-      "step": 3440
-    },
-    {
-      "epoch": 1.7945241199478488,
-      "grad_norm": 1.5382492542266846,
-      "learning_rate": 9.704522613065327e-05,
-      "loss": 6.0505,
-      "step": 3441
-    },
-    {
-      "epoch": 1.795045632333768,
-      "grad_norm": 1.623565673828125,
-      "learning_rate": 9.704422110552765e-05,
-      "loss": 5.8352,
-      "step": 3442
-    },
-    {
-      "epoch": 1.795567144719687,
-      "grad_norm": 1.801328182220459,
-      "learning_rate": 9.704321608040201e-05,
-      "loss": 4.7086,
-      "step": 3443
-    },
-    {
-      "epoch": 1.7960886571056063,
-      "grad_norm": 1.8106749057769775,
-      "learning_rate": 9.704221105527639e-05,
-      "loss": 5.8866,
-      "step": 3444
-    },
-    {
-      "epoch": 1.7966101694915255,
-      "grad_norm": 1.5402570962905884,
-      "learning_rate": 9.704120603015075e-05,
-      "loss": 5.9812,
-      "step": 3445
-    },
-    {
-      "epoch": 1.7971316818774445,
-      "grad_norm": 1.6236382722854614,
-      "learning_rate": 9.704020100502513e-05,
-      "loss": 5.5292,
-      "step": 3446
-    },
-    {
-      "epoch": 1.7976531942633638,
-      "grad_norm": 1.6535191535949707,
-      "learning_rate": 9.70391959798995e-05,
-      "loss": 6.0806,
-      "step": 3447
-    },
-    {
-      "epoch": 1.798174706649283,
-      "grad_norm": 1.5745586156845093,
-      "learning_rate": 9.703819095477387e-05,
-      "loss": 5.8841,
-      "step": 3448
-    },
-    {
-      "epoch": 1.798696219035202,
-      "grad_norm": 1.4843578338623047,
-      "learning_rate": 9.703718592964825e-05,
-      "loss": 6.2798,
-      "step": 3449
-    },
-    {
-      "epoch": 1.7992177314211213,
-      "grad_norm": 1.9440903663635254,
-      "learning_rate": 9.703618090452263e-05,
-      "loss": 5.1926,
-      "step": 3450
-    },
-    {
-      "epoch": 1.7997392438070405,
-      "grad_norm": 1.854526162147522,
-      "learning_rate": 9.703517587939699e-05,
-      "loss": 5.785,
-      "step": 3451
-    },
-    {
-      "epoch": 1.8002607561929596,
-      "grad_norm": 1.6898843050003052,
-      "learning_rate": 9.703417085427137e-05,
-      "loss": 5.9249,
-      "step": 3452
-    },
-    {
-      "epoch": 1.8007822685788788,
-      "grad_norm": 1.592730164527893,
-      "learning_rate": 9.703316582914573e-05,
-      "loss": 5.89,
-      "step": 3453
-    },
-    {
-      "epoch": 1.801303780964798,
-      "grad_norm": 1.7539761066436768,
-      "learning_rate": 9.70321608040201e-05,
-      "loss": 6.1321,
-      "step": 3454
-    },
-    {
-      "epoch": 1.801825293350717,
-      "grad_norm": 1.6388393640518188,
-      "learning_rate": 9.703115577889448e-05,
-      "loss": 5.9846,
-      "step": 3455
-    },
-    {
-      "epoch": 1.802346805736636,
-      "grad_norm": 1.582400918006897,
-      "learning_rate": 9.703015075376884e-05,
-      "loss": 6.0675,
-      "step": 3456
-    },
-    {
-      "epoch": 1.8028683181225555,
-      "grad_norm": 2.0983474254608154,
-      "learning_rate": 9.702914572864322e-05,
-      "loss": 5.7575,
-      "step": 3457
-    },
-    {
-      "epoch": 1.8033898305084746,
-      "grad_norm": 1.803412675857544,
-      "learning_rate": 9.702814070351758e-05,
-      "loss": 6.1308,
-      "step": 3458
-    },
-    {
-      "epoch": 1.8039113428943936,
-      "grad_norm": 1.8067283630371094,
-      "learning_rate": 9.702713567839196e-05,
-      "loss": 5.3885,
-      "step": 3459
-    },
-    {
-      "epoch": 1.804432855280313,
-      "grad_norm": 1.5002249479293823,
-      "learning_rate": 9.702613065326634e-05,
-      "loss": 6.2755,
-      "step": 3460
-    },
-    {
-      "epoch": 1.804954367666232,
-      "grad_norm": 1.6799871921539307,
-      "learning_rate": 9.702512562814072e-05,
-      "loss": 5.9691,
-      "step": 3461
-    },
-    {
-      "epoch": 1.805475880052151,
-      "grad_norm": 1.5397225618362427,
-      "learning_rate": 9.702412060301508e-05,
-      "loss": 6.2171,
-      "step": 3462
-    },
-    {
-      "epoch": 1.8059973924380706,
-      "grad_norm": 1.6402848958969116,
-      "learning_rate": 9.702311557788946e-05,
-      "loss": 5.8595,
-      "step": 3463
-    },
-    {
-      "epoch": 1.8065189048239896,
-      "grad_norm": 2.529421806335449,
-      "learning_rate": 9.702211055276382e-05,
-      "loss": 5.6316,
-      "step": 3464
-    },
-    {
-      "epoch": 1.8070404172099086,
-      "grad_norm": 1.891155481338501,
-      "learning_rate": 9.70211055276382e-05,
-      "loss": 6.0724,
-      "step": 3465
-    },
-    {
-      "epoch": 1.807561929595828,
-      "grad_norm": 1.5303215980529785,
-      "learning_rate": 9.702010050251257e-05,
-      "loss": 6.0312,
-      "step": 3466
-    },
-    {
-      "epoch": 1.808083441981747,
-      "grad_norm": 1.5245964527130127,
-      "learning_rate": 9.701909547738693e-05,
-      "loss": 6.1743,
-      "step": 3467
-    },
-    {
-      "epoch": 1.808604954367666,
-      "grad_norm": 1.6625903844833374,
-      "learning_rate": 9.701809045226131e-05,
-      "loss": 5.9876,
-      "step": 3468
-    },
-    {
-      "epoch": 1.8091264667535853,
-      "grad_norm": 1.609916090965271,
-      "learning_rate": 9.701708542713568e-05,
-      "loss": 5.8721,
-      "step": 3469
-    },
-    {
-      "epoch": 1.8096479791395046,
-      "grad_norm": 1.5482451915740967,
-      "learning_rate": 9.701608040201006e-05,
-      "loss": 5.9456,
-      "step": 3470
-    },
-    {
-      "epoch": 1.8101694915254236,
-      "grad_norm": 1.290881872177124,
-      "learning_rate": 9.701507537688443e-05,
-      "loss": 5.8151,
-      "step": 3471
-    },
-    {
-      "epoch": 1.8106910039113429,
-      "grad_norm": 1.5847201347351074,
-      "learning_rate": 9.70140703517588e-05,
-      "loss": 5.8954,
-      "step": 3472
-    },
-    {
-      "epoch": 1.811212516297262,
-      "grad_norm": 1.5597525835037231,
-      "learning_rate": 9.701306532663317e-05,
-      "loss": 5.3822,
-      "step": 3473
-    },
-    {
-      "epoch": 1.8117340286831811,
-      "grad_norm": 1.6361896991729736,
-      "learning_rate": 9.701206030150755e-05,
-      "loss": 5.9887,
-      "step": 3474
-    },
-    {
-      "epoch": 1.8122555410691004,
-      "grad_norm": 1.5049463510513306,
-      "learning_rate": 9.701105527638191e-05,
-      "loss": 6.4319,
-      "step": 3475
-    },
-    {
-      "epoch": 1.8127770534550196,
-      "grad_norm": 1.4886294603347778,
-      "learning_rate": 9.701005025125629e-05,
-      "loss": 6.2015,
-      "step": 3476
-    },
-    {
-      "epoch": 1.8132985658409386,
-      "grad_norm": 1.7461503744125366,
-      "learning_rate": 9.700904522613065e-05,
-      "loss": 5.7579,
-      "step": 3477
-    },
-    {
-      "epoch": 1.8138200782268579,
-      "grad_norm": 1.7426531314849854,
-      "learning_rate": 9.700804020100503e-05,
-      "loss": 6.0545,
-      "step": 3478
-    },
-    {
-      "epoch": 1.814341590612777,
-      "grad_norm": 1.6803030967712402,
-      "learning_rate": 9.70070351758794e-05,
-      "loss": 6.2379,
-      "step": 3479
-    },
-    {
-      "epoch": 1.8148631029986961,
-      "grad_norm": 1.469314455986023,
-      "learning_rate": 9.700603015075377e-05,
-      "loss": 6.2643,
-      "step": 3480
-    },
-    {
-      "epoch": 1.8153846153846154,
-      "grad_norm": 1.507590413093567,
-      "learning_rate": 9.700502512562815e-05,
-      "loss": 6.1819,
-      "step": 3481
-    },
-    {
-      "epoch": 1.8159061277705346,
-      "grad_norm": 2.1014201641082764,
-      "learning_rate": 9.700402010050252e-05,
-      "loss": 5.8218,
-      "step": 3482
-    },
-    {
-      "epoch": 1.8164276401564536,
-      "grad_norm": 2.0744240283966064,
-      "learning_rate": 9.70030150753769e-05,
-      "loss": 5.8149,
-      "step": 3483
-    },
-    {
-      "epoch": 1.8169491525423729,
-      "grad_norm": 1.7443263530731201,
-      "learning_rate": 9.700201005025126e-05,
-      "loss": 6.0395,
-      "step": 3484
-    },
-    {
-      "epoch": 1.8174706649282921,
-      "grad_norm": 1.690757393836975,
-      "learning_rate": 9.700100502512564e-05,
-      "loss": 6.3058,
-      "step": 3485
-    },
-    {
-      "epoch": 1.8179921773142111,
-      "grad_norm": 1.668442964553833,
-      "learning_rate": 9.7e-05,
-      "loss": 5.9292,
-      "step": 3486
-    },
-    {
-      "epoch": 1.8185136897001304,
-      "grad_norm": 1.6220622062683105,
-      "learning_rate": 9.699899497487438e-05,
-      "loss": 5.8296,
-      "step": 3487
-    },
-    {
-      "epoch": 1.8190352020860496,
-      "grad_norm": 1.5136754512786865,
-      "learning_rate": 9.699798994974874e-05,
-      "loss": 6.0959,
-      "step": 3488
-    },
-    {
-      "epoch": 1.8195567144719687,
-      "grad_norm": 1.5258347988128662,
-      "learning_rate": 9.699698492462312e-05,
-      "loss": 6.4227,
-      "step": 3489
-    },
-    {
-      "epoch": 1.820078226857888,
-      "grad_norm": 1.4603675603866577,
-      "learning_rate": 9.69959798994975e-05,
-      "loss": 6.0721,
-      "step": 3490
-    },
-    {
-      "epoch": 1.8205997392438071,
-      "grad_norm": 1.6585214138031006,
-      "learning_rate": 9.699497487437188e-05,
-      "loss": 5.8413,
-      "step": 3491
-    },
-    {
-      "epoch": 1.8211212516297262,
-      "grad_norm": 1.7849349975585938,
-      "learning_rate": 9.699396984924624e-05,
-      "loss": 5.9434,
-      "step": 3492
-    },
-    {
-      "epoch": 1.8216427640156454,
-      "grad_norm": 1.6531637907028198,
-      "learning_rate": 9.69929648241206e-05,
-      "loss": 6.0698,
-      "step": 3493
-    },
-    {
-      "epoch": 1.8221642764015646,
-      "grad_norm": 1.6902382373809814,
-      "learning_rate": 9.699195979899498e-05,
-      "loss": 5.7329,
-      "step": 3494
-    },
-    {
-      "epoch": 1.8226857887874837,
-      "grad_norm": 1.61663818359375,
-      "learning_rate": 9.699095477386935e-05,
-      "loss": 5.8391,
-      "step": 3495
-    },
-    {
-      "epoch": 1.823207301173403,
-      "grad_norm": 1.6554762125015259,
-      "learning_rate": 9.698994974874372e-05,
-      "loss": 6.0434,
-      "step": 3496
-    },
-    {
-      "epoch": 1.8237288135593221,
-      "grad_norm": 1.7411292791366577,
-      "learning_rate": 9.698894472361809e-05,
-      "loss": 5.5667,
-      "step": 3497
-    },
-    {
-      "epoch": 1.8242503259452412,
-      "grad_norm": 1.6766211986541748,
-      "learning_rate": 9.698793969849247e-05,
-      "loss": 6.034,
-      "step": 3498
-    },
-    {
-      "epoch": 1.8247718383311604,
-      "grad_norm": 1.5605521202087402,
-      "learning_rate": 9.698693467336683e-05,
-      "loss": 6.2319,
-      "step": 3499
-    },
-    {
-      "epoch": 1.8252933507170797,
-      "grad_norm": 1.7649636268615723,
-      "learning_rate": 9.698592964824121e-05,
-      "loss": 5.566,
-      "step": 3500
-    },
-    {
-      "epoch": 1.8258148631029987,
-      "grad_norm": 1.5264796018600464,
-      "learning_rate": 9.698492462311559e-05,
-      "loss": 5.5563,
-      "step": 3501
-    },
-    {
-      "epoch": 1.8263363754889177,
-      "grad_norm": 1.58991539478302,
-      "learning_rate": 9.698391959798996e-05,
-      "loss": 5.8224,
-      "step": 3502
-    },
-    {
-      "epoch": 1.8268578878748372,
-      "grad_norm": 1.978438138961792,
-      "learning_rate": 9.698291457286433e-05,
-      "loss": 5.8882,
-      "step": 3503
-    },
-    {
-      "epoch": 1.8273794002607562,
-      "grad_norm": 1.487432599067688,
-      "learning_rate": 9.69819095477387e-05,
-      "loss": 6.0081,
-      "step": 3504
-    },
-    {
-      "epoch": 1.8279009126466752,
-      "grad_norm": 1.5642988681793213,
-      "learning_rate": 9.698090452261307e-05,
-      "loss": 6.0111,
-      "step": 3505
-    },
-    {
-      "epoch": 1.8284224250325947,
-      "grad_norm": 1.683190107345581,
-      "learning_rate": 9.697989949748743e-05,
-      "loss": 6.2369,
-      "step": 3506
-    },
-    {
-      "epoch": 1.8289439374185137,
-      "grad_norm": 1.4596670866012573,
-      "learning_rate": 9.697889447236181e-05,
-      "loss": 6.2156,
-      "step": 3507
-    },
-    {
-      "epoch": 1.8294654498044327,
-      "grad_norm": 1.307924509048462,
-      "learning_rate": 9.697788944723618e-05,
-      "loss": 5.8237,
-      "step": 3508
-    },
-    {
-      "epoch": 1.8299869621903522,
-      "grad_norm": 1.692699909210205,
-      "learning_rate": 9.697688442211055e-05,
-      "loss": 5.6566,
-      "step": 3509
-    },
-    {
-      "epoch": 1.8305084745762712,
-      "grad_norm": 1.6478583812713623,
-      "learning_rate": 9.697587939698493e-05,
-      "loss": 5.3311,
-      "step": 3510
-    },
-    {
-      "epoch": 1.8310299869621902,
-      "grad_norm": 1.4691568613052368,
-      "learning_rate": 9.697487437185931e-05,
-      "loss": 6.2814,
-      "step": 3511
-    },
-    {
-      "epoch": 1.8315514993481097,
-      "grad_norm": 1.7584285736083984,
-      "learning_rate": 9.697386934673367e-05,
-      "loss": 6.0498,
-      "step": 3512
-    },
-    {
-      "epoch": 1.8320730117340287,
-      "grad_norm": 1.5221638679504395,
-      "learning_rate": 9.697286432160805e-05,
-      "loss": 6.0866,
-      "step": 3513
-    },
-    {
-      "epoch": 1.8325945241199477,
-      "grad_norm": 1.6671223640441895,
-      "learning_rate": 9.697185929648242e-05,
-      "loss": 5.5608,
-      "step": 3514
-    },
-    {
-      "epoch": 1.833116036505867,
-      "grad_norm": 1.5549525022506714,
-      "learning_rate": 9.69708542713568e-05,
-      "loss": 6.2136,
-      "step": 3515
-    },
-    {
-      "epoch": 1.8336375488917862,
-      "grad_norm": 1.5825947523117065,
-      "learning_rate": 9.696984924623116e-05,
-      "loss": 5.8405,
-      "step": 3516
-    },
-    {
-      "epoch": 1.8341590612777052,
-      "grad_norm": 1.693906545639038,
-      "learning_rate": 9.696884422110554e-05,
-      "loss": 5.5286,
-      "step": 3517
-    },
-    {
-      "epoch": 1.8346805736636245,
-      "grad_norm": 1.7293037176132202,
-      "learning_rate": 9.69678391959799e-05,
-      "loss": 6.0295,
-      "step": 3518
-    },
-    {
-      "epoch": 1.8352020860495437,
-      "grad_norm": 1.5592045783996582,
-      "learning_rate": 9.696683417085426e-05,
-      "loss": 6.2459,
-      "step": 3519
-    },
-    {
-      "epoch": 1.8357235984354627,
-      "grad_norm": 1.4750192165374756,
-      "learning_rate": 9.696582914572864e-05,
-      "loss": 5.811,
-      "step": 3520
-    },
-    {
-      "epoch": 1.836245110821382,
-      "grad_norm": 1.6404991149902344,
-      "learning_rate": 9.696482412060302e-05,
-      "loss": 5.3284,
-      "step": 3521
-    },
-    {
-      "epoch": 1.8367666232073012,
-      "grad_norm": 1.6890718936920166,
-      "learning_rate": 9.69638190954774e-05,
-      "loss": 6.0738,
-      "step": 3522
-    },
-    {
-      "epoch": 1.8372881355932202,
-      "grad_norm": 1.5716408491134644,
-      "learning_rate": 9.696281407035176e-05,
-      "loss": 5.7744,
-      "step": 3523
-    },
-    {
-      "epoch": 1.8378096479791395,
-      "grad_norm": 1.584784984588623,
-      "learning_rate": 9.696180904522614e-05,
-      "loss": 6.1147,
-      "step": 3524
-    },
-    {
-      "epoch": 1.8383311603650587,
-      "grad_norm": 1.6987148523330688,
-      "learning_rate": 9.69608040201005e-05,
-      "loss": 5.9771,
-      "step": 3525
-    },
-    {
-      "epoch": 1.8388526727509777,
-      "grad_norm": 1.8942301273345947,
-      "learning_rate": 9.695979899497488e-05,
-      "loss": 5.7769,
-      "step": 3526
-    },
-    {
-      "epoch": 1.839374185136897,
-      "grad_norm": 1.4818071126937866,
-      "learning_rate": 9.695879396984925e-05,
-      "loss": 6.1555,
-      "step": 3527
-    },
-    {
-      "epoch": 1.8398956975228162,
-      "grad_norm": 1.7362920045852661,
-      "learning_rate": 9.695778894472362e-05,
-      "loss": 5.8292,
-      "step": 3528
-    },
-    {
-      "epoch": 1.8404172099087353,
-      "grad_norm": 1.4890836477279663,
-      "learning_rate": 9.695678391959799e-05,
-      "loss": 5.929,
-      "step": 3529
-    },
-    {
-      "epoch": 1.8409387222946545,
-      "grad_norm": 1.589530348777771,
-      "learning_rate": 9.695577889447237e-05,
-      "loss": 5.9247,
-      "step": 3530
-    },
-    {
-      "epoch": 1.8414602346805737,
-      "grad_norm": 1.4957821369171143,
-      "learning_rate": 9.695477386934674e-05,
-      "loss": 6.0141,
-      "step": 3531
-    },
-    {
-      "epoch": 1.8419817470664928,
-      "grad_norm": 1.649237036705017,
-      "learning_rate": 9.695376884422112e-05,
-      "loss": 6.0488,
-      "step": 3532
-    },
-    {
-      "epoch": 1.842503259452412,
-      "grad_norm": 1.5398844480514526,
-      "learning_rate": 9.695276381909549e-05,
-      "loss": 5.5527,
-      "step": 3533
-    },
-    {
-      "epoch": 1.8430247718383312,
-      "grad_norm": 1.643724799156189,
-      "learning_rate": 9.695175879396985e-05,
-      "loss": 6.1538,
-      "step": 3534
-    },
-    {
-      "epoch": 1.8435462842242503,
-      "grad_norm": 1.4018816947937012,
-      "learning_rate": 9.695075376884423e-05,
-      "loss": 6.3606,
-      "step": 3535
-    },
-    {
-      "epoch": 1.8440677966101695,
-      "grad_norm": 1.5994594097137451,
-      "learning_rate": 9.694974874371859e-05,
-      "loss": 5.717,
-      "step": 3536
-    },
-    {
-      "epoch": 1.8445893089960888,
-      "grad_norm": 1.5260415077209473,
-      "learning_rate": 9.694874371859297e-05,
-      "loss": 6.0545,
-      "step": 3537
-    },
-    {
-      "epoch": 1.8451108213820078,
-      "grad_norm": 1.6104403734207153,
-      "learning_rate": 9.694773869346733e-05,
-      "loss": 5.9096,
-      "step": 3538
-    },
-    {
-      "epoch": 1.845632333767927,
-      "grad_norm": 1.954505443572998,
-      "learning_rate": 9.694673366834171e-05,
-      "loss": 5.7639,
-      "step": 3539
-    },
-    {
-      "epoch": 1.8461538461538463,
-      "grad_norm": 1.5488784313201904,
-      "learning_rate": 9.694572864321608e-05,
-      "loss": 5.9992,
-      "step": 3540
-    },
-    {
-      "epoch": 1.8466753585397653,
-      "grad_norm": 1.4249992370605469,
-      "learning_rate": 9.694472361809045e-05,
-      "loss": 6.205,
-      "step": 3541
-    },
-    {
-      "epoch": 1.8471968709256845,
-      "grad_norm": 1.4550275802612305,
-      "learning_rate": 9.694371859296483e-05,
-      "loss": 6.0722,
-      "step": 3542
-    },
-    {
-      "epoch": 1.8477183833116038,
-      "grad_norm": 1.8217707872390747,
-      "learning_rate": 9.694271356783921e-05,
-      "loss": 5.5045,
-      "step": 3543
-    },
-    {
-      "epoch": 1.8482398956975228,
-      "grad_norm": 1.561005711555481,
-      "learning_rate": 9.694170854271357e-05,
-      "loss": 6.2676,
-      "step": 3544
-    },
-    {
-      "epoch": 1.848761408083442,
-      "grad_norm": 1.5794198513031006,
-      "learning_rate": 9.694070351758795e-05,
-      "loss": 5.897,
-      "step": 3545
-    },
-    {
-      "epoch": 1.8492829204693613,
-      "grad_norm": 1.5415103435516357,
-      "learning_rate": 9.693969849246232e-05,
-      "loss": 5.985,
-      "step": 3546
-    },
-    {
-      "epoch": 1.8498044328552803,
-      "grad_norm": 1.5353405475616455,
-      "learning_rate": 9.693869346733668e-05,
-      "loss": 6.0127,
-      "step": 3547
-    },
-    {
-      "epoch": 1.8503259452411995,
-      "grad_norm": 1.559869647026062,
-      "learning_rate": 9.693768844221106e-05,
-      "loss": 6.1704,
-      "step": 3548
-    },
-    {
-      "epoch": 1.8508474576271188,
-      "grad_norm": 1.6556882858276367,
-      "learning_rate": 9.693668341708542e-05,
-      "loss": 5.8365,
-      "step": 3549
-    },
-    {
-      "epoch": 1.8513689700130378,
-      "grad_norm": 1.6140103340148926,
-      "learning_rate": 9.69356783919598e-05,
-      "loss": 5.6238,
-      "step": 3550
-    },
-    {
-      "epoch": 1.8518904823989568,
-      "grad_norm": 1.4674066305160522,
-      "learning_rate": 9.693467336683418e-05,
-      "loss": 6.0331,
-      "step": 3551
-    },
-    {
-      "epoch": 1.8524119947848763,
-      "grad_norm": 1.599793791770935,
-      "learning_rate": 9.693366834170856e-05,
-      "loss": 5.2199,
-      "step": 3552
-    },
-    {
-      "epoch": 1.8529335071707953,
-      "grad_norm": 1.4060035943984985,
-      "learning_rate": 9.693266331658292e-05,
-      "loss": 5.7664,
-      "step": 3553
-    },
-    {
-      "epoch": 1.8534550195567143,
-      "grad_norm": 1.574790358543396,
-      "learning_rate": 9.69316582914573e-05,
-      "loss": 6.0178,
-      "step": 3554
-    },
-    {
-      "epoch": 1.8539765319426338,
-      "grad_norm": 1.4803346395492554,
-      "learning_rate": 9.693065326633166e-05,
-      "loss": 6.2984,
-      "step": 3555
-    },
-    {
-      "epoch": 1.8544980443285528,
-      "grad_norm": 1.8222908973693848,
-      "learning_rate": 9.692964824120604e-05,
-      "loss": 5.6655,
-      "step": 3556
-    },
-    {
-      "epoch": 1.8550195567144718,
-      "grad_norm": 1.6952451467514038,
-      "learning_rate": 9.69286432160804e-05,
-      "loss": 5.886,
-      "step": 3557
-    },
-    {
-      "epoch": 1.8555410691003913,
-      "grad_norm": 1.5570813417434692,
-      "learning_rate": 9.692763819095478e-05,
-      "loss": 6.1473,
-      "step": 3558
-    },
-    {
-      "epoch": 1.8560625814863103,
-      "grad_norm": 1.4632219076156616,
-      "learning_rate": 9.692663316582915e-05,
-      "loss": 5.9331,
-      "step": 3559
-    },
-    {
-      "epoch": 1.8565840938722293,
-      "grad_norm": 1.6619189977645874,
-      "learning_rate": 9.692562814070351e-05,
-      "loss": 5.6498,
-      "step": 3560
-    },
-    {
-      "epoch": 1.8571056062581486,
-      "grad_norm": 1.6818832159042358,
-      "learning_rate": 9.692462311557789e-05,
-      "loss": 5.9141,
-      "step": 3561
-    },
-    {
-      "epoch": 1.8576271186440678,
-      "grad_norm": 1.857456922531128,
-      "learning_rate": 9.692361809045227e-05,
-      "loss": 5.522,
-      "step": 3562
-    },
-    {
-      "epoch": 1.8581486310299868,
-      "grad_norm": 1.5724027156829834,
-      "learning_rate": 9.692261306532665e-05,
-      "loss": 5.8487,
-      "step": 3563
-    },
-    {
-      "epoch": 1.858670143415906,
-      "grad_norm": 1.577483057975769,
-      "learning_rate": 9.692160804020101e-05,
-      "loss": 6.0312,
-      "step": 3564
-    },
-    {
-      "epoch": 1.8591916558018253,
-      "grad_norm": 1.5000864267349243,
-      "learning_rate": 9.692060301507539e-05,
-      "loss": 6.1475,
-      "step": 3565
-    },
-    {
-      "epoch": 1.8597131681877443,
-      "grad_norm": 1.5401833057403564,
-      "learning_rate": 9.691959798994975e-05,
-      "loss": 5.7778,
-      "step": 3566
-    },
-    {
-      "epoch": 1.8602346805736636,
-      "grad_norm": 1.595280408859253,
-      "learning_rate": 9.691859296482413e-05,
-      "loss": 5.6764,
-      "step": 3567
-    },
-    {
-      "epoch": 1.8607561929595828,
-      "grad_norm": 1.5504183769226074,
-      "learning_rate": 9.69175879396985e-05,
-      "loss": 6.2006,
-      "step": 3568
-    },
-    {
-      "epoch": 1.8612777053455019,
-      "grad_norm": 1.5815843343734741,
-      "learning_rate": 9.691658291457287e-05,
-      "loss": 6.287,
-      "step": 3569
-    },
-    {
-      "epoch": 1.861799217731421,
-      "grad_norm": 1.5669087171554565,
-      "learning_rate": 9.691557788944724e-05,
-      "loss": 6.1947,
-      "step": 3570
-    },
-    {
-      "epoch": 1.8623207301173403,
-      "grad_norm": 1.7047145366668701,
-      "learning_rate": 9.691457286432161e-05,
-      "loss": 5.807,
-      "step": 3571
-    },
-    {
-      "epoch": 1.8628422425032594,
-      "grad_norm": 1.723549246788025,
-      "learning_rate": 9.691356783919598e-05,
-      "loss": 5.7011,
-      "step": 3572
-    },
-    {
-      "epoch": 1.8633637548891786,
-      "grad_norm": 1.532957673072815,
-      "learning_rate": 9.691256281407036e-05,
-      "loss": 6.1912,
-      "step": 3573
-    },
-    {
-      "epoch": 1.8638852672750978,
-      "grad_norm": 1.662353754043579,
-      "learning_rate": 9.691155778894473e-05,
-      "loss": 6.0034,
-      "step": 3574
-    },
-    {
-      "epoch": 1.8644067796610169,
-      "grad_norm": 1.577138066291809,
-      "learning_rate": 9.69105527638191e-05,
-      "loss": 6.0462,
-      "step": 3575
-    },
-    {
-      "epoch": 1.864928292046936,
-      "grad_norm": 1.9192328453063965,
-      "learning_rate": 9.690954773869348e-05,
-      "loss": 5.8168,
-      "step": 3576
-    },
-    {
-      "epoch": 1.8654498044328554,
-      "grad_norm": 2.001338481903076,
-      "learning_rate": 9.690854271356784e-05,
-      "loss": 5.673,
-      "step": 3577
-    },
-    {
-      "epoch": 1.8659713168187744,
-      "grad_norm": 1.696831464767456,
-      "learning_rate": 9.690753768844222e-05,
-      "loss": 5.8345,
-      "step": 3578
-    },
-    {
-      "epoch": 1.8664928292046936,
-      "grad_norm": 1.3572980165481567,
-      "learning_rate": 9.690653266331658e-05,
-      "loss": 6.1702,
-      "step": 3579
-    },
-    {
-      "epoch": 1.8670143415906129,
-      "grad_norm": 1.3938509225845337,
-      "learning_rate": 9.690552763819096e-05,
-      "loss": 6.0875,
-      "step": 3580
-    },
-    {
-      "epoch": 1.8675358539765319,
-      "grad_norm": 1.474871277809143,
-      "learning_rate": 9.690452261306532e-05,
-      "loss": 5.8459,
-      "step": 3581
-    },
-    {
-      "epoch": 1.8680573663624511,
-      "grad_norm": 1.4626671075820923,
-      "learning_rate": 9.69035175879397e-05,
-      "loss": 6.2527,
-      "step": 3582
-    },
-    {
-      "epoch": 1.8685788787483704,
-      "grad_norm": 1.4968199729919434,
-      "learning_rate": 9.690251256281408e-05,
-      "loss": 5.7092,
-      "step": 3583
-    },
-    {
-      "epoch": 1.8691003911342894,
-      "grad_norm": 1.5720598697662354,
-      "learning_rate": 9.690150753768846e-05,
-      "loss": 5.9111,
-      "step": 3584
-    },
-    {
-      "epoch": 1.8696219035202086,
-      "grad_norm": 1.4595754146575928,
-      "learning_rate": 9.690050251256282e-05,
-      "loss": 6.2048,
-      "step": 3585
-    },
-    {
-      "epoch": 1.8701434159061279,
-      "grad_norm": 1.3937532901763916,
-      "learning_rate": 9.689949748743719e-05,
-      "loss": 6.111,
-      "step": 3586
-    },
-    {
-      "epoch": 1.870664928292047,
-      "grad_norm": 1.5596139430999756,
-      "learning_rate": 9.689849246231156e-05,
-      "loss": 5.6974,
-      "step": 3587
-    },
-    {
-      "epoch": 1.8711864406779661,
-      "grad_norm": 1.8474706411361694,
-      "learning_rate": 9.689748743718593e-05,
-      "loss": 5.6772,
-      "step": 3588
-    },
-    {
-      "epoch": 1.8717079530638854,
-      "grad_norm": 1.7659916877746582,
-      "learning_rate": 9.68964824120603e-05,
-      "loss": 6.1995,
-      "step": 3589
-    },
-    {
-      "epoch": 1.8722294654498044,
-      "grad_norm": 1.4868820905685425,
-      "learning_rate": 9.689547738693467e-05,
-      "loss": 6.3916,
-      "step": 3590
-    },
-    {
-      "epoch": 1.8727509778357236,
-      "grad_norm": 1.5156104564666748,
-      "learning_rate": 9.689447236180905e-05,
-      "loss": 5.9658,
-      "step": 3591
-    },
-    {
-      "epoch": 1.8732724902216429,
-      "grad_norm": 1.9636346101760864,
-      "learning_rate": 9.689346733668341e-05,
-      "loss": 5.6547,
-      "step": 3592
-    },
-    {
-      "epoch": 1.873794002607562,
-      "grad_norm": 1.7381705045700073,
-      "learning_rate": 9.689246231155779e-05,
-      "loss": 6.1177,
-      "step": 3593
-    },
-    {
-      "epoch": 1.8743155149934811,
-      "grad_norm": 1.7234363555908203,
-      "learning_rate": 9.689145728643217e-05,
-      "loss": 5.7103,
-      "step": 3594
-    },
-    {
-      "epoch": 1.8748370273794004,
-      "grad_norm": 1.4314123392105103,
-      "learning_rate": 9.689045226130655e-05,
-      "loss": 6.0758,
-      "step": 3595
-    },
-    {
-      "epoch": 1.8753585397653194,
-      "grad_norm": 1.7011404037475586,
-      "learning_rate": 9.688944723618091e-05,
-      "loss": 6.1152,
-      "step": 3596
-    },
-    {
-      "epoch": 1.8758800521512384,
-      "grad_norm": 1.7060736417770386,
-      "learning_rate": 9.688844221105529e-05,
-      "loss": 6.0837,
-      "step": 3597
-    },
-    {
-      "epoch": 1.876401564537158,
-      "grad_norm": 1.57675302028656,
-      "learning_rate": 9.688743718592965e-05,
-      "loss": 6.4183,
-      "step": 3598
-    },
-    {
-      "epoch": 1.876923076923077,
-      "grad_norm": 1.592848539352417,
-      "learning_rate": 9.688643216080402e-05,
-      "loss": 5.8429,
-      "step": 3599
-    },
-    {
-      "epoch": 1.877444589308996,
-      "grad_norm": 2.520181894302368,
-      "learning_rate": 9.68854271356784e-05,
-      "loss": 5.3106,
-      "step": 3600
-    },
-    {
-      "epoch": 1.8779661016949154,
-      "grad_norm": 1.861150860786438,
-      "learning_rate": 9.688442211055276e-05,
-      "loss": 5.5475,
-      "step": 3601
-    },
-    {
-      "epoch": 1.8784876140808344,
-      "grad_norm": 1.7574419975280762,
-      "learning_rate": 9.688341708542714e-05,
-      "loss": 5.9741,
-      "step": 3602
-    },
-    {
-      "epoch": 1.8790091264667534,
-      "grad_norm": 1.8051307201385498,
-      "learning_rate": 9.688241206030151e-05,
-      "loss": 5.9724,
-      "step": 3603
-    },
-    {
-      "epoch": 1.879530638852673,
-      "grad_norm": 1.9332389831542969,
-      "learning_rate": 9.688140703517589e-05,
-      "loss": 5.9075,
-      "step": 3604
-    },
-    {
-      "epoch": 1.880052151238592,
-      "grad_norm": 1.6104470491409302,
-      "learning_rate": 9.688040201005026e-05,
-      "loss": 5.9914,
-      "step": 3605
-    },
-    {
-      "epoch": 1.880573663624511,
-      "grad_norm": 1.6659700870513916,
-      "learning_rate": 9.687939698492463e-05,
-      "loss": 6.073,
-      "step": 3606
-    },
-    {
-      "epoch": 1.8810951760104304,
-      "grad_norm": 1.621062994003296,
-      "learning_rate": 9.6878391959799e-05,
-      "loss": 5.5826,
-      "step": 3607
-    },
-    {
-      "epoch": 1.8816166883963494,
-      "grad_norm": 1.5182889699935913,
-      "learning_rate": 9.687738693467338e-05,
-      "loss": 6.1019,
-      "step": 3608
-    },
-    {
-      "epoch": 1.8821382007822685,
-      "grad_norm": 1.5832126140594482,
-      "learning_rate": 9.687638190954774e-05,
-      "loss": 6.1679,
-      "step": 3609
-    },
-    {
-      "epoch": 1.8826597131681877,
-      "grad_norm": 1.609300136566162,
-      "learning_rate": 9.687537688442212e-05,
-      "loss": 6.16,
-      "step": 3610
-    },
-    {
-      "epoch": 1.883181225554107,
-      "grad_norm": 1.4409281015396118,
-      "learning_rate": 9.687437185929648e-05,
-      "loss": 6.409,
-      "step": 3611
-    },
-    {
-      "epoch": 1.883702737940026,
-      "grad_norm": 1.9251627922058105,
-      "learning_rate": 9.687336683417086e-05,
-      "loss": 5.3217,
-      "step": 3612
-    },
-    {
-      "epoch": 1.8842242503259452,
-      "grad_norm": 1.660509467124939,
-      "learning_rate": 9.687236180904522e-05,
-      "loss": 5.8955,
-      "step": 3613
-    },
-    {
-      "epoch": 1.8847457627118644,
-      "grad_norm": 1.8304067850112915,
-      "learning_rate": 9.68713567839196e-05,
-      "loss": 5.4229,
-      "step": 3614
-    },
-    {
-      "epoch": 1.8852672750977835,
-      "grad_norm": 1.6469128131866455,
-      "learning_rate": 9.687035175879398e-05,
-      "loss": 6.1424,
-      "step": 3615
-    },
-    {
-      "epoch": 1.8857887874837027,
-      "grad_norm": 1.35207998752594,
-      "learning_rate": 9.686934673366834e-05,
-      "loss": 5.8327,
-      "step": 3616
-    },
-    {
-      "epoch": 1.886310299869622,
-      "grad_norm": 1.7950124740600586,
-      "learning_rate": 9.686834170854272e-05,
-      "loss": 5.6104,
-      "step": 3617
-    },
-    {
-      "epoch": 1.886831812255541,
-      "grad_norm": 1.408860445022583,
-      "learning_rate": 9.686733668341709e-05,
-      "loss": 6.1429,
-      "step": 3618
-    },
-    {
-      "epoch": 1.8873533246414602,
-      "grad_norm": 1.6640617847442627,
-      "learning_rate": 9.686633165829146e-05,
-      "loss": 5.9708,
-      "step": 3619
-    },
-    {
-      "epoch": 1.8878748370273795,
-      "grad_norm": 1.5460660457611084,
-      "learning_rate": 9.686532663316583e-05,
-      "loss": 6.082,
-      "step": 3620
-    },
-    {
-      "epoch": 1.8883963494132985,
-      "grad_norm": 1.579760193824768,
-      "learning_rate": 9.686432160804021e-05,
-      "loss": 6.1989,
-      "step": 3621
-    },
-    {
-      "epoch": 1.8889178617992177,
-      "grad_norm": 1.460542917251587,
-      "learning_rate": 9.686331658291457e-05,
-      "loss": 6.0314,
-      "step": 3622
-    },
-    {
-      "epoch": 1.889439374185137,
-      "grad_norm": 1.4485174417495728,
-      "learning_rate": 9.686231155778895e-05,
-      "loss": 6.1999,
-      "step": 3623
-    },
-    {
-      "epoch": 1.889960886571056,
-      "grad_norm": 1.6901365518569946,
-      "learning_rate": 9.686130653266333e-05,
-      "loss": 5.3317,
-      "step": 3624
-    },
-    {
-      "epoch": 1.8904823989569752,
-      "grad_norm": 1.723552942276001,
-      "learning_rate": 9.68603015075377e-05,
-      "loss": 5.6563,
-      "step": 3625
-    },
-    {
-      "epoch": 1.8910039113428945,
-      "grad_norm": 2.3357629776000977,
-      "learning_rate": 9.685929648241207e-05,
-      "loss": 5.672,
-      "step": 3626
-    },
-    {
-      "epoch": 1.8915254237288135,
-      "grad_norm": 1.4890269041061401,
-      "learning_rate": 9.685829145728643e-05,
-      "loss": 6.1165,
-      "step": 3627
-    },
-    {
-      "epoch": 1.8920469361147327,
-      "grad_norm": 2.029008388519287,
-      "learning_rate": 9.685728643216081e-05,
-      "loss": 5.5383,
-      "step": 3628
-    },
-    {
-      "epoch": 1.892568448500652,
-      "grad_norm": 1.7153542041778564,
-      "learning_rate": 9.685628140703518e-05,
-      "loss": 5.7703,
-      "step": 3629
-    },
-    {
-      "epoch": 1.893089960886571,
-      "grad_norm": 1.4662227630615234,
-      "learning_rate": 9.685527638190955e-05,
-      "loss": 6.0931,
-      "step": 3630
-    },
-    {
-      "epoch": 1.8936114732724902,
-      "grad_norm": 1.8050633668899536,
-      "learning_rate": 9.685427135678392e-05,
-      "loss": 5.6545,
-      "step": 3631
-    },
-    {
-      "epoch": 1.8941329856584095,
-      "grad_norm": 1.5558801889419556,
-      "learning_rate": 9.68532663316583e-05,
-      "loss": 5.9254,
-      "step": 3632
-    },
-    {
-      "epoch": 1.8946544980443285,
-      "grad_norm": 1.4596816301345825,
-      "learning_rate": 9.685226130653266e-05,
-      "loss": 6.1951,
-      "step": 3633
-    },
-    {
-      "epoch": 1.8951760104302477,
-      "grad_norm": 1.4725292921066284,
-      "learning_rate": 9.685125628140704e-05,
-      "loss": 6.156,
-      "step": 3634
-    },
-    {
-      "epoch": 1.895697522816167,
-      "grad_norm": 1.5823264122009277,
-      "learning_rate": 9.685025125628142e-05,
-      "loss": 6.1368,
-      "step": 3635
-    },
-    {
-      "epoch": 1.896219035202086,
-      "grad_norm": 1.4274561405181885,
-      "learning_rate": 9.684924623115579e-05,
-      "loss": 6.1916,
-      "step": 3636
-    },
-    {
-      "epoch": 1.8967405475880053,
-      "grad_norm": 1.5348559617996216,
-      "learning_rate": 9.684824120603016e-05,
-      "loss": 6.1261,
-      "step": 3637
-    },
-    {
-      "epoch": 1.8972620599739245,
-      "grad_norm": 1.5813374519348145,
-      "learning_rate": 9.684723618090454e-05,
-      "loss": 6.1783,
-      "step": 3638
-    },
-    {
-      "epoch": 1.8977835723598435,
-      "grad_norm": 1.9786345958709717,
-      "learning_rate": 9.68462311557789e-05,
-      "loss": 5.6011,
-      "step": 3639
-    },
-    {
-      "epoch": 1.8983050847457628,
-      "grad_norm": 1.7165980339050293,
-      "learning_rate": 9.684522613065326e-05,
-      "loss": 5.8182,
-      "step": 3640
-    },
-    {
-      "epoch": 1.898826597131682,
-      "grad_norm": 1.8394607305526733,
-      "learning_rate": 9.684422110552764e-05,
-      "loss": 5.8259,
-      "step": 3641
-    },
-    {
-      "epoch": 1.899348109517601,
-      "grad_norm": 1.7221581935882568,
-      "learning_rate": 9.6843216080402e-05,
-      "loss": 5.7388,
-      "step": 3642
-    },
-    {
-      "epoch": 1.89986962190352,
-      "grad_norm": 1.5938208103179932,
-      "learning_rate": 9.684221105527638e-05,
-      "loss": 5.613,
-      "step": 3643
-    },
-    {
-      "epoch": 1.9003911342894395,
-      "grad_norm": 1.4545577764511108,
-      "learning_rate": 9.684120603015076e-05,
-      "loss": 5.9846,
-      "step": 3644
-    },
-    {
-      "epoch": 1.9009126466753585,
-      "grad_norm": 1.4908982515335083,
-      "learning_rate": 9.684020100502514e-05,
-      "loss": 5.8721,
-      "step": 3645
-    },
-    {
-      "epoch": 1.9014341590612776,
-      "grad_norm": 1.6030458211898804,
-      "learning_rate": 9.68391959798995e-05,
-      "loss": 6.164,
-      "step": 3646
-    },
-    {
-      "epoch": 1.901955671447197,
-      "grad_norm": 1.5814929008483887,
-      "learning_rate": 9.683819095477388e-05,
-      "loss": 5.4399,
-      "step": 3647
-    },
-    {
-      "epoch": 1.902477183833116,
-      "grad_norm": 1.5326217412948608,
-      "learning_rate": 9.683718592964825e-05,
-      "loss": 6.2265,
-      "step": 3648
-    },
-    {
-      "epoch": 1.902998696219035,
-      "grad_norm": 1.4068875312805176,
-      "learning_rate": 9.683618090452262e-05,
-      "loss": 6.3311,
-      "step": 3649
-    },
-    {
-      "epoch": 1.9035202086049545,
-      "grad_norm": 1.43031644821167,
-      "learning_rate": 9.683517587939699e-05,
-      "loss": 6.3475,
-      "step": 3650
-    },
-    {
-      "epoch": 1.9040417209908735,
-      "grad_norm": 1.4738000631332397,
-      "learning_rate": 9.683417085427137e-05,
-      "loss": 5.8946,
-      "step": 3651
-    },
-    {
-      "epoch": 1.9045632333767926,
-      "grad_norm": 1.7692428827285767,
-      "learning_rate": 9.683316582914573e-05,
-      "loss": 5.363,
-      "step": 3652
-    },
-    {
-      "epoch": 1.905084745762712,
-      "grad_norm": 1.4844504594802856,
-      "learning_rate": 9.68321608040201e-05,
-      "loss": 5.8269,
-      "step": 3653
-    },
-    {
-      "epoch": 1.905606258148631,
-      "grad_norm": 1.7501931190490723,
-      "learning_rate": 9.683115577889447e-05,
-      "loss": 6.0528,
-      "step": 3654
-    },
-    {
-      "epoch": 1.90612777053455,
-      "grad_norm": 1.7964050769805908,
-      "learning_rate": 9.683015075376885e-05,
-      "loss": 5.849,
-      "step": 3655
-    },
-    {
-      "epoch": 1.9066492829204693,
-      "grad_norm": 1.3265413045883179,
-      "learning_rate": 9.682914572864323e-05,
-      "loss": 6.1906,
-      "step": 3656
-    },
-    {
-      "epoch": 1.9071707953063886,
-      "grad_norm": 1.7927899360656738,
-      "learning_rate": 9.682814070351759e-05,
-      "loss": 5.7423,
-      "step": 3657
-    },
-    {
-      "epoch": 1.9076923076923076,
-      "grad_norm": 1.301763653755188,
-      "learning_rate": 9.682713567839197e-05,
-      "loss": 6.3257,
-      "step": 3658
-    },
-    {
-      "epoch": 1.9082138200782268,
-      "grad_norm": 1.6177808046340942,
-      "learning_rate": 9.682613065326633e-05,
-      "loss": 5.9454,
-      "step": 3659
-    },
-    {
-      "epoch": 1.908735332464146,
-      "grad_norm": 2.0057876110076904,
-      "learning_rate": 9.682512562814071e-05,
-      "loss": 5.9168,
-      "step": 3660
-    },
-    {
-      "epoch": 1.909256844850065,
-      "grad_norm": 1.649571180343628,
-      "learning_rate": 9.682412060301508e-05,
-      "loss": 5.9585,
-      "step": 3661
-    },
-    {
-      "epoch": 1.9097783572359843,
-      "grad_norm": 1.9685922861099243,
-      "learning_rate": 9.682311557788945e-05,
-      "loss": 5.5481,
-      "step": 3662
-    },
-    {
-      "epoch": 1.9102998696219036,
-      "grad_norm": 1.5866222381591797,
-      "learning_rate": 9.682211055276382e-05,
-      "loss": 5.9588,
-      "step": 3663
-    },
-    {
-      "epoch": 1.9108213820078226,
-      "grad_norm": 1.67631196975708,
-      "learning_rate": 9.68211055276382e-05,
-      "loss": 6.1244,
-      "step": 3664
-    },
-    {
-      "epoch": 1.9113428943937418,
-      "grad_norm": 1.4832837581634521,
-      "learning_rate": 9.682010050251257e-05,
-      "loss": 5.9704,
-      "step": 3665
-    },
-    {
-      "epoch": 1.911864406779661,
-      "grad_norm": 1.8317663669586182,
-      "learning_rate": 9.681909547738694e-05,
-      "loss": 5.3972,
-      "step": 3666
-    },
-    {
-      "epoch": 1.91238591916558,
-      "grad_norm": 1.7771217823028564,
-      "learning_rate": 9.681809045226132e-05,
-      "loss": 6.0876,
-      "step": 3667
-    },
-    {
-      "epoch": 1.9129074315514993,
-      "grad_norm": 1.3874270915985107,
-      "learning_rate": 9.681708542713568e-05,
-      "loss": 5.7121,
-      "step": 3668
-    },
-    {
-      "epoch": 1.9134289439374186,
-      "grad_norm": 1.506945252418518,
-      "learning_rate": 9.681608040201006e-05,
-      "loss": 6.0447,
-      "step": 3669
-    },
-    {
-      "epoch": 1.9139504563233376,
-      "grad_norm": 1.4367058277130127,
-      "learning_rate": 9.681507537688442e-05,
-      "loss": 5.656,
-      "step": 3670
-    },
-    {
-      "epoch": 1.9144719687092568,
-      "grad_norm": 1.5422313213348389,
-      "learning_rate": 9.68140703517588e-05,
-      "loss": 5.9431,
-      "step": 3671
-    },
-    {
-      "epoch": 1.914993481095176,
-      "grad_norm": 1.6148920059204102,
-      "learning_rate": 9.681306532663316e-05,
-      "loss": 5.8026,
-      "step": 3672
-    },
-    {
-      "epoch": 1.915514993481095,
-      "grad_norm": 1.6064120531082153,
-      "learning_rate": 9.681206030150754e-05,
-      "loss": 5.8293,
-      "step": 3673
-    },
-    {
-      "epoch": 1.9160365058670144,
-      "grad_norm": 1.6061739921569824,
-      "learning_rate": 9.68110552763819e-05,
-      "loss": 5.6975,
-      "step": 3674
-    },
-    {
-      "epoch": 1.9165580182529336,
-      "grad_norm": 1.921130657196045,
-      "learning_rate": 9.681005025125628e-05,
-      "loss": 5.4473,
-      "step": 3675
-    },
-    {
-      "epoch": 1.9170795306388526,
-      "grad_norm": 1.4987750053405762,
-      "learning_rate": 9.680904522613066e-05,
-      "loss": 6.0276,
-      "step": 3676
-    },
-    {
-      "epoch": 1.9176010430247719,
-      "grad_norm": 1.5462708473205566,
-      "learning_rate": 9.680804020100504e-05,
-      "loss": 6.0424,
-      "step": 3677
-    },
-    {
-      "epoch": 1.918122555410691,
-      "grad_norm": 1.6817291975021362,
-      "learning_rate": 9.68070351758794e-05,
-      "loss": 5.3902,
-      "step": 3678
-    },
-    {
-      "epoch": 1.9186440677966101,
-      "grad_norm": 1.459525227546692,
-      "learning_rate": 9.680603015075377e-05,
-      "loss": 5.698,
-      "step": 3679
-    },
-    {
-      "epoch": 1.9191655801825294,
-      "grad_norm": 1.5248008966445923,
-      "learning_rate": 9.680502512562815e-05,
-      "loss": 5.7205,
-      "step": 3680
-    },
-    {
-      "epoch": 1.9196870925684486,
-      "grad_norm": 1.4505418539047241,
-      "learning_rate": 9.680402010050251e-05,
-      "loss": 5.5123,
-      "step": 3681
-    },
-    {
-      "epoch": 1.9202086049543676,
-      "grad_norm": 1.4754718542099,
-      "learning_rate": 9.680301507537689e-05,
-      "loss": 6.1092,
-      "step": 3682
-    },
-    {
-      "epoch": 1.9207301173402869,
-      "grad_norm": 1.613889455795288,
-      "learning_rate": 9.680201005025125e-05,
-      "loss": 6.1246,
-      "step": 3683
-    },
-    {
-      "epoch": 1.9212516297262061,
-      "grad_norm": 1.6242589950561523,
-      "learning_rate": 9.680100502512563e-05,
-      "loss": 5.9992,
-      "step": 3684
-    },
-    {
-      "epoch": 1.9217731421121251,
-      "grad_norm": 1.5692026615142822,
-      "learning_rate": 9.680000000000001e-05,
-      "loss": 5.8122,
-      "step": 3685
-    },
-    {
-      "epoch": 1.9222946544980444,
-      "grad_norm": 1.7350645065307617,
-      "learning_rate": 9.679899497487439e-05,
-      "loss": 5.7234,
-      "step": 3686
-    },
-    {
-      "epoch": 1.9228161668839636,
-      "grad_norm": 1.5124965906143188,
-      "learning_rate": 9.679798994974875e-05,
-      "loss": 5.7415,
-      "step": 3687
-    },
-    {
-      "epoch": 1.9233376792698826,
-      "grad_norm": 1.4327722787857056,
-      "learning_rate": 9.679698492462313e-05,
-      "loss": 6.4429,
-      "step": 3688
-    },
-    {
-      "epoch": 1.9238591916558017,
-      "grad_norm": 1.4378658533096313,
-      "learning_rate": 9.679597989949749e-05,
-      "loss": 6.2302,
-      "step": 3689
-    },
-    {
-      "epoch": 1.9243807040417211,
-      "grad_norm": 1.5136994123458862,
-      "learning_rate": 9.679497487437187e-05,
-      "loss": 5.8949,
-      "step": 3690
-    },
-    {
-      "epoch": 1.9249022164276401,
-      "grad_norm": 1.6525558233261108,
-      "learning_rate": 9.679396984924623e-05,
-      "loss": 5.7527,
-      "step": 3691
-    },
-    {
-      "epoch": 1.9254237288135592,
-      "grad_norm": 1.4742815494537354,
-      "learning_rate": 9.67929648241206e-05,
-      "loss": 5.6675,
-      "step": 3692
-    },
-    {
-      "epoch": 1.9259452411994786,
-      "grad_norm": 1.4292960166931152,
-      "learning_rate": 9.679195979899498e-05,
-      "loss": 6.0268,
-      "step": 3693
-    },
-    {
-      "epoch": 1.9264667535853977,
-      "grad_norm": 1.3192503452301025,
-      "learning_rate": 9.679095477386934e-05,
-      "loss": 6.151,
-      "step": 3694
-    },
-    {
-      "epoch": 1.9269882659713167,
-      "grad_norm": 1.5001747608184814,
-      "learning_rate": 9.678994974874372e-05,
-      "loss": 6.0082,
-      "step": 3695
-    },
-    {
-      "epoch": 1.9275097783572361,
-      "grad_norm": 1.9304540157318115,
-      "learning_rate": 9.67889447236181e-05,
-      "loss": 5.8408,
-      "step": 3696
-    },
-    {
-      "epoch": 1.9280312907431552,
-      "grad_norm": 1.487881064414978,
-      "learning_rate": 9.678793969849247e-05,
-      "loss": 6.2902,
-      "step": 3697
-    },
-    {
-      "epoch": 1.9285528031290742,
-      "grad_norm": 1.7343789339065552,
-      "learning_rate": 9.678693467336684e-05,
-      "loss": 6.0957,
-      "step": 3698
-    },
-    {
-      "epoch": 1.9290743155149936,
-      "grad_norm": 1.4530911445617676,
-      "learning_rate": 9.678592964824122e-05,
-      "loss": 6.4411,
-      "step": 3699
-    },
-    {
-      "epoch": 1.9295958279009127,
-      "grad_norm": 1.397850751876831,
-      "learning_rate": 9.678492462311558e-05,
-      "loss": 5.3955,
-      "step": 3700
-    },
-    {
-      "epoch": 1.9301173402868317,
-      "grad_norm": 1.5376310348510742,
-      "learning_rate": 9.678391959798996e-05,
-      "loss": 6.1986,
-      "step": 3701
-    },
-    {
-      "epoch": 1.930638852672751,
-      "grad_norm": 1.6636452674865723,
-      "learning_rate": 9.678291457286432e-05,
-      "loss": 5.9068,
-      "step": 3702
-    },
-    {
-      "epoch": 1.9311603650586702,
-      "grad_norm": 1.582450270652771,
-      "learning_rate": 9.67819095477387e-05,
-      "loss": 5.9694,
-      "step": 3703
-    },
-    {
-      "epoch": 1.9316818774445892,
-      "grad_norm": 1.6411999464035034,
-      "learning_rate": 9.678090452261307e-05,
-      "loss": 6.144,
-      "step": 3704
-    },
-    {
-      "epoch": 1.9322033898305084,
-      "grad_norm": 1.5312954187393188,
-      "learning_rate": 9.677989949748744e-05,
-      "loss": 6.1232,
-      "step": 3705
-    },
-    {
-      "epoch": 1.9327249022164277,
-      "grad_norm": 1.5187140703201294,
-      "learning_rate": 9.677889447236182e-05,
-      "loss": 6.0082,
-      "step": 3706
-    },
-    {
-      "epoch": 1.9332464146023467,
-      "grad_norm": 1.7073029279708862,
-      "learning_rate": 9.677788944723619e-05,
-      "loss": 6.1666,
-      "step": 3707
-    },
-    {
-      "epoch": 1.933767926988266,
-      "grad_norm": 1.5173527002334595,
-      "learning_rate": 9.677688442211056e-05,
-      "loss": 6.3999,
-      "step": 3708
-    },
-    {
-      "epoch": 1.9342894393741852,
-      "grad_norm": 1.805742859840393,
-      "learning_rate": 9.677587939698493e-05,
-      "loss": 5.6408,
-      "step": 3709
-    },
-    {
-      "epoch": 1.9348109517601042,
-      "grad_norm": 1.513910174369812,
-      "learning_rate": 9.67748743718593e-05,
-      "loss": 6.1759,
-      "step": 3710
-    },
-    {
-      "epoch": 1.9353324641460234,
-      "grad_norm": 1.7959589958190918,
-      "learning_rate": 9.677386934673367e-05,
-      "loss": 5.3493,
-      "step": 3711
-    },
-    {
-      "epoch": 1.9358539765319427,
-      "grad_norm": 1.4577901363372803,
-      "learning_rate": 9.677286432160805e-05,
-      "loss": 6.3691,
-      "step": 3712
-    },
-    {
-      "epoch": 1.9363754889178617,
-      "grad_norm": 1.5365177392959595,
-      "learning_rate": 9.677185929648241e-05,
-      "loss": 5.9217,
-      "step": 3713
-    },
-    {
-      "epoch": 1.936897001303781,
-      "grad_norm": 1.5541095733642578,
-      "learning_rate": 9.677085427135679e-05,
-      "loss": 5.951,
-      "step": 3714
-    },
-    {
-      "epoch": 1.9374185136897002,
-      "grad_norm": 1.780725121498108,
-      "learning_rate": 9.676984924623115e-05,
-      "loss": 5.321,
-      "step": 3715
-    },
-    {
-      "epoch": 1.9379400260756192,
-      "grad_norm": 1.5229734182357788,
-      "learning_rate": 9.676884422110553e-05,
-      "loss": 5.98,
-      "step": 3716
-    },
-    {
-      "epoch": 1.9384615384615385,
-      "grad_norm": 1.438604712486267,
-      "learning_rate": 9.676783919597991e-05,
-      "loss": 6.3953,
-      "step": 3717
-    },
-    {
-      "epoch": 1.9389830508474577,
-      "grad_norm": 1.3150635957717896,
-      "learning_rate": 9.676683417085429e-05,
-      "loss": 6.2537,
-      "step": 3718
-    },
-    {
-      "epoch": 1.9395045632333767,
-      "grad_norm": 1.6807818412780762,
-      "learning_rate": 9.676582914572865e-05,
-      "loss": 5.6899,
-      "step": 3719
-    },
-    {
-      "epoch": 1.940026075619296,
-      "grad_norm": 1.4965217113494873,
-      "learning_rate": 9.676482412060302e-05,
-      "loss": 6.1998,
-      "step": 3720
-    },
-    {
-      "epoch": 1.9405475880052152,
-      "grad_norm": 1.3748282194137573,
-      "learning_rate": 9.67638190954774e-05,
-      "loss": 6.2531,
-      "step": 3721
-    },
-    {
-      "epoch": 1.9410691003911342,
-      "grad_norm": 1.4825294017791748,
-      "learning_rate": 9.676281407035176e-05,
-      "loss": 6.174,
-      "step": 3722
-    },
-    {
-      "epoch": 1.9415906127770535,
-      "grad_norm": 1.6296523809432983,
-      "learning_rate": 9.676180904522614e-05,
-      "loss": 5.9069,
-      "step": 3723
-    },
-    {
-      "epoch": 1.9421121251629727,
-      "grad_norm": 1.5608360767364502,
-      "learning_rate": 9.67608040201005e-05,
-      "loss": 5.8187,
-      "step": 3724
-    },
-    {
-      "epoch": 1.9426336375488917,
-      "grad_norm": 1.426539659500122,
-      "learning_rate": 9.675979899497488e-05,
-      "loss": 6.1302,
-      "step": 3725
-    },
-    {
-      "epoch": 1.943155149934811,
-      "grad_norm": 1.6397687196731567,
-      "learning_rate": 9.675879396984924e-05,
-      "loss": 5.624,
-      "step": 3726
-    },
-    {
-      "epoch": 1.9436766623207302,
-      "grad_norm": 1.4884954690933228,
-      "learning_rate": 9.675778894472362e-05,
-      "loss": 6.1793,
-      "step": 3727
-    },
-    {
-      "epoch": 1.9441981747066492,
-      "grad_norm": 1.4461029767990112,
-      "learning_rate": 9.6756783919598e-05,
-      "loss": 6.4219,
-      "step": 3728
-    },
-    {
-      "epoch": 1.9447196870925685,
-      "grad_norm": 1.827540397644043,
-      "learning_rate": 9.675577889447238e-05,
-      "loss": 5.2261,
-      "step": 3729
-    },
-    {
-      "epoch": 1.9452411994784877,
-      "grad_norm": 1.523616075515747,
-      "learning_rate": 9.675477386934674e-05,
-      "loss": 5.3159,
-      "step": 3730
-    },
-    {
-      "epoch": 1.9457627118644067,
-      "grad_norm": 1.957961916923523,
-      "learning_rate": 9.675376884422112e-05,
-      "loss": 5.4794,
-      "step": 3731
-    },
-    {
-      "epoch": 1.946284224250326,
-      "grad_norm": 1.5601928234100342,
-      "learning_rate": 9.675276381909548e-05,
-      "loss": 6.1306,
-      "step": 3732
-    },
-    {
-      "epoch": 1.9468057366362452,
-      "grad_norm": 1.561591386795044,
-      "learning_rate": 9.675175879396985e-05,
-      "loss": 5.7421,
-      "step": 3733
-    },
-    {
-      "epoch": 1.9473272490221643,
-      "grad_norm": 1.6185028553009033,
-      "learning_rate": 9.675075376884422e-05,
-      "loss": 5.677,
-      "step": 3734
-    },
-    {
-      "epoch": 1.9478487614080835,
-      "grad_norm": 1.4494253396987915,
-      "learning_rate": 9.674974874371859e-05,
-      "loss": 6.225,
-      "step": 3735
-    },
-    {
-      "epoch": 1.9483702737940027,
-      "grad_norm": 1.5567940473556519,
-      "learning_rate": 9.674874371859297e-05,
-      "loss": 6.0748,
-      "step": 3736
-    },
-    {
-      "epoch": 1.9488917861799218,
-      "grad_norm": 1.3867876529693604,
-      "learning_rate": 9.674773869346734e-05,
-      "loss": 6.2342,
-      "step": 3737
-    },
-    {
-      "epoch": 1.9494132985658408,
-      "grad_norm": 1.466679334640503,
-      "learning_rate": 9.674673366834172e-05,
-      "loss": 6.1619,
-      "step": 3738
-    },
-    {
-      "epoch": 1.9499348109517602,
-      "grad_norm": 1.555573582649231,
-      "learning_rate": 9.674572864321609e-05,
-      "loss": 5.7609,
-      "step": 3739
-    },
-    {
-      "epoch": 1.9504563233376793,
-      "grad_norm": 1.586954116821289,
-      "learning_rate": 9.674472361809046e-05,
-      "loss": 5.9281,
-      "step": 3740
-    },
-    {
-      "epoch": 1.9509778357235983,
-      "grad_norm": 1.5040411949157715,
-      "learning_rate": 9.674371859296483e-05,
-      "loss": 5.9583,
-      "step": 3741
-    },
-    {
-      "epoch": 1.9514993481095178,
-      "grad_norm": 1.2743133306503296,
-      "learning_rate": 9.67427135678392e-05,
-      "loss": 6.1137,
-      "step": 3742
-    },
-    {
-      "epoch": 1.9520208604954368,
-      "grad_norm": 1.4357311725616455,
-      "learning_rate": 9.674170854271357e-05,
-      "loss": 6.1235,
-      "step": 3743
-    },
-    {
-      "epoch": 1.9525423728813558,
-      "grad_norm": 1.805456519126892,
-      "learning_rate": 9.674070351758795e-05,
-      "loss": 5.7003,
-      "step": 3744
-    },
-    {
-      "epoch": 1.9530638852672753,
-      "grad_norm": 1.77345609664917,
-      "learning_rate": 9.673969849246231e-05,
-      "loss": 6.123,
-      "step": 3745
-    },
-    {
-      "epoch": 1.9535853976531943,
-      "grad_norm": 1.6671754121780396,
-      "learning_rate": 9.673869346733668e-05,
-      "loss": 5.6419,
-      "step": 3746
-    },
-    {
-      "epoch": 1.9541069100391133,
-      "grad_norm": 1.8957421779632568,
-      "learning_rate": 9.673768844221105e-05,
-      "loss": 5.7793,
-      "step": 3747
-    },
-    {
-      "epoch": 1.9546284224250325,
-      "grad_norm": 1.832861065864563,
-      "learning_rate": 9.673668341708543e-05,
-      "loss": 5.7934,
-      "step": 3748
-    },
-    {
-      "epoch": 1.9551499348109518,
-      "grad_norm": 1.5750383138656616,
-      "learning_rate": 9.673567839195981e-05,
-      "loss": 5.48,
-      "step": 3749
-    },
-    {
-      "epoch": 1.9556714471968708,
-      "grad_norm": 1.452635645866394,
-      "learning_rate": 9.673467336683417e-05,
-      "loss": 5.8993,
-      "step": 3750
-    },
-    {
-      "epoch": 1.95619295958279,
-      "grad_norm": 1.7086045742034912,
-      "learning_rate": 9.673366834170855e-05,
-      "loss": 5.4557,
-      "step": 3751
-    },
-    {
-      "epoch": 1.9567144719687093,
-      "grad_norm": 1.454647183418274,
-      "learning_rate": 9.673266331658292e-05,
-      "loss": 6.1394,
-      "step": 3752
-    },
-    {
-      "epoch": 1.9572359843546283,
-      "grad_norm": 1.5458574295043945,
-      "learning_rate": 9.67316582914573e-05,
-      "loss": 6.1411,
-      "step": 3753
-    },
-    {
-      "epoch": 1.9577574967405476,
-      "grad_norm": 1.579566478729248,
-      "learning_rate": 9.673065326633166e-05,
-      "loss": 5.6519,
-      "step": 3754
-    },
-    {
-      "epoch": 1.9582790091264668,
-      "grad_norm": 1.4239953756332397,
-      "learning_rate": 9.672964824120604e-05,
-      "loss": 6.176,
-      "step": 3755
-    },
-    {
-      "epoch": 1.9588005215123858,
-      "grad_norm": 1.6183122396469116,
-      "learning_rate": 9.67286432160804e-05,
-      "loss": 5.2611,
-      "step": 3756
-    },
-    {
-      "epoch": 1.959322033898305,
-      "grad_norm": 1.4526921510696411,
-      "learning_rate": 9.672763819095478e-05,
-      "loss": 6.4183,
-      "step": 3757
-    },
-    {
-      "epoch": 1.9598435462842243,
-      "grad_norm": 1.457470178604126,
-      "learning_rate": 9.672663316582916e-05,
-      "loss": 6.1146,
-      "step": 3758
-    },
-    {
-      "epoch": 1.9603650586701433,
-      "grad_norm": 1.4826791286468506,
-      "learning_rate": 9.672562814070352e-05,
-      "loss": 6.2853,
-      "step": 3759
-    },
-    {
-      "epoch": 1.9608865710560626,
-      "grad_norm": 1.5994478464126587,
-      "learning_rate": 9.67246231155779e-05,
-      "loss": 5.5664,
-      "step": 3760
-    },
-    {
-      "epoch": 1.9614080834419818,
-      "grad_norm": 1.586102843284607,
-      "learning_rate": 9.672361809045226e-05,
-      "loss": 5.9162,
-      "step": 3761
-    },
-    {
-      "epoch": 1.9619295958279008,
-      "grad_norm": 1.4627211093902588,
-      "learning_rate": 9.672261306532664e-05,
-      "loss": 5.5841,
-      "step": 3762
-    },
-    {
-      "epoch": 1.96245110821382,
-      "grad_norm": 1.7927237749099731,
-      "learning_rate": 9.6721608040201e-05,
-      "loss": 5.6951,
-      "step": 3763
-    },
-    {
-      "epoch": 1.9629726205997393,
-      "grad_norm": 1.4786993265151978,
-      "learning_rate": 9.672060301507538e-05,
-      "loss": 6.142,
-      "step": 3764
-    },
-    {
-      "epoch": 1.9634941329856583,
-      "grad_norm": 1.4124550819396973,
-      "learning_rate": 9.671959798994975e-05,
-      "loss": 6.111,
-      "step": 3765
-    },
-    {
-      "epoch": 1.9640156453715776,
-      "grad_norm": 1.4885871410369873,
-      "learning_rate": 9.671859296482412e-05,
-      "loss": 6.1683,
-      "step": 3766
-    },
-    {
-      "epoch": 1.9645371577574968,
-      "grad_norm": 1.4775676727294922,
-      "learning_rate": 9.671758793969849e-05,
-      "loss": 5.9439,
-      "step": 3767
-    },
-    {
-      "epoch": 1.9650586701434158,
-      "grad_norm": 1.5198376178741455,
-      "learning_rate": 9.671658291457287e-05,
-      "loss": 6.0779,
-      "step": 3768
-    },
-    {
-      "epoch": 1.965580182529335,
-      "grad_norm": 1.6087232828140259,
-      "learning_rate": 9.671557788944724e-05,
-      "loss": 6.021,
-      "step": 3769
-    },
-    {
-      "epoch": 1.9661016949152543,
-      "grad_norm": 1.5046133995056152,
-      "learning_rate": 9.671457286432162e-05,
-      "loss": 6.2484,
-      "step": 3770
-    },
-    {
-      "epoch": 1.9666232073011733,
-      "grad_norm": 1.5241562128067017,
-      "learning_rate": 9.671356783919599e-05,
-      "loss": 6.2958,
-      "step": 3771
-    },
-    {
-      "epoch": 1.9671447196870926,
-      "grad_norm": 1.6915507316589355,
-      "learning_rate": 9.671256281407035e-05,
-      "loss": 5.7114,
-      "step": 3772
-    },
-    {
-      "epoch": 1.9676662320730118,
-      "grad_norm": 1.9184249639511108,
-      "learning_rate": 9.671155778894473e-05,
-      "loss": 6.0257,
-      "step": 3773
-    },
-    {
-      "epoch": 1.9681877444589309,
-      "grad_norm": 1.6501777172088623,
-      "learning_rate": 9.671055276381909e-05,
-      "loss": 5.6487,
-      "step": 3774
-    },
-    {
-      "epoch": 1.96870925684485,
-      "grad_norm": 1.6688206195831299,
-      "learning_rate": 9.670954773869347e-05,
-      "loss": 6.1452,
-      "step": 3775
-    },
-    {
-      "epoch": 1.9692307692307693,
-      "grad_norm": 1.5813205242156982,
-      "learning_rate": 9.670854271356784e-05,
-      "loss": 6.1194,
-      "step": 3776
-    },
-    {
-      "epoch": 1.9697522816166884,
-      "grad_norm": 1.46859610080719,
-      "learning_rate": 9.670753768844221e-05,
-      "loss": 6.1389,
-      "step": 3777
-    },
-    {
-      "epoch": 1.9702737940026076,
-      "grad_norm": 1.5424631834030151,
-      "learning_rate": 9.670653266331659e-05,
-      "loss": 6.0538,
-      "step": 3778
-    },
-    {
-      "epoch": 1.9707953063885268,
-      "grad_norm": 1.7152600288391113,
-      "learning_rate": 9.670552763819097e-05,
-      "loss": 5.8784,
-      "step": 3779
-    },
-    {
-      "epoch": 1.9713168187744459,
-      "grad_norm": 1.6198248863220215,
-      "learning_rate": 9.670452261306533e-05,
-      "loss": 6.0991,
-      "step": 3780
-    },
-    {
-      "epoch": 1.971838331160365,
-      "grad_norm": 1.5231430530548096,
-      "learning_rate": 9.670351758793971e-05,
-      "loss": 6.2892,
-      "step": 3781
-    },
-    {
-      "epoch": 1.9723598435462844,
-      "grad_norm": 1.3140794038772583,
-      "learning_rate": 9.670251256281407e-05,
-      "loss": 6.2536,
-      "step": 3782
-    },
-    {
-      "epoch": 1.9728813559322034,
-      "grad_norm": 1.7115626335144043,
-      "learning_rate": 9.670150753768845e-05,
-      "loss": 5.8252,
-      "step": 3783
-    },
-    {
-      "epoch": 1.9734028683181224,
-      "grad_norm": 1.4999061822891235,
-      "learning_rate": 9.670050251256282e-05,
-      "loss": 6.0298,
-      "step": 3784
-    },
-    {
-      "epoch": 1.9739243807040419,
-      "grad_norm": 1.4897143840789795,
-      "learning_rate": 9.669949748743718e-05,
-      "loss": 5.8075,
-      "step": 3785
-    },
-    {
-      "epoch": 1.9744458930899609,
-      "grad_norm": 1.5358606576919556,
-      "learning_rate": 9.669849246231156e-05,
-      "loss": 5.7169,
-      "step": 3786
-    },
-    {
-      "epoch": 1.97496740547588,
-      "grad_norm": 1.6340683698654175,
-      "learning_rate": 9.669748743718592e-05,
-      "loss": 5.4938,
-      "step": 3787
-    },
-    {
-      "epoch": 1.9754889178617994,
-      "grad_norm": 1.6434447765350342,
-      "learning_rate": 9.66964824120603e-05,
-      "loss": 5.9882,
-      "step": 3788
-    },
-    {
-      "epoch": 1.9760104302477184,
-      "grad_norm": 1.490289330482483,
-      "learning_rate": 9.669547738693468e-05,
-      "loss": 5.9047,
-      "step": 3789
-    },
-    {
-      "epoch": 1.9765319426336374,
-      "grad_norm": 1.490796446800232,
-      "learning_rate": 9.669447236180906e-05,
-      "loss": 5.6832,
-      "step": 3790
-    },
-    {
-      "epoch": 1.9770534550195569,
-      "grad_norm": 1.7789455652236938,
-      "learning_rate": 9.669346733668342e-05,
-      "loss": 5.7787,
-      "step": 3791
-    },
-    {
-      "epoch": 1.977574967405476,
-      "grad_norm": 1.4790477752685547,
-      "learning_rate": 9.66924623115578e-05,
-      "loss": 5.9383,
-      "step": 3792
-    },
-    {
-      "epoch": 1.978096479791395,
-      "grad_norm": 1.644132375717163,
-      "learning_rate": 9.669145728643216e-05,
-      "loss": 6.1026,
-      "step": 3793
-    },
-    {
-      "epoch": 1.9786179921773144,
-      "grad_norm": 1.5769768953323364,
-      "learning_rate": 9.669045226130654e-05,
-      "loss": 5.7081,
-      "step": 3794
-    },
-    {
-      "epoch": 1.9791395045632334,
-      "grad_norm": 1.4699238538742065,
-      "learning_rate": 9.66894472361809e-05,
-      "loss": 5.7658,
-      "step": 3795
-    },
-    {
-      "epoch": 1.9796610169491524,
-      "grad_norm": 1.4182850122451782,
-      "learning_rate": 9.668844221105528e-05,
-      "loss": 5.9603,
-      "step": 3796
-    },
-    {
-      "epoch": 1.9801825293350717,
-      "grad_norm": 1.4898478984832764,
-      "learning_rate": 9.668743718592965e-05,
-      "loss": 6.0428,
-      "step": 3797
-    },
-    {
-      "epoch": 1.980704041720991,
-      "grad_norm": 1.700779914855957,
-      "learning_rate": 9.668643216080403e-05,
-      "loss": 5.9131,
-      "step": 3798
-    },
-    {
-      "epoch": 1.98122555410691,
-      "grad_norm": 1.4077318906784058,
-      "learning_rate": 9.66854271356784e-05,
-      "loss": 5.9479,
-      "step": 3799
-    },
-    {
-      "epoch": 1.9817470664928292,
-      "grad_norm": 1.3818029165267944,
-      "learning_rate": 9.668442211055277e-05,
-      "loss": 6.0141,
-      "step": 3800
-    },
-    {
-      "epoch": 1.9822685788787484,
-      "grad_norm": 1.4354159832000732,
-      "learning_rate": 9.668341708542715e-05,
-      "loss": 5.8762,
-      "step": 3801
-    },
-    {
-      "epoch": 1.9827900912646674,
-      "grad_norm": 1.5172733068466187,
-      "learning_rate": 9.668241206030151e-05,
-      "loss": 5.8266,
-      "step": 3802
-    },
-    {
-      "epoch": 1.9833116036505867,
-      "grad_norm": 1.4415496587753296,
-      "learning_rate": 9.668140703517589e-05,
-      "loss": 6.0618,
-      "step": 3803
-    },
-    {
-      "epoch": 1.983833116036506,
-      "grad_norm": 1.5004628896713257,
-      "learning_rate": 9.668040201005025e-05,
-      "loss": 5.4733,
-      "step": 3804
-    },
-    {
-      "epoch": 1.984354628422425,
-      "grad_norm": 1.3964412212371826,
-      "learning_rate": 9.667939698492463e-05,
-      "loss": 5.9146,
-      "step": 3805
-    },
-    {
-      "epoch": 1.9848761408083442,
-      "grad_norm": 1.3824177980422974,
-      "learning_rate": 9.6678391959799e-05,
-      "loss": 6.2999,
-      "step": 3806
-    },
-    {
-      "epoch": 1.9853976531942634,
-      "grad_norm": 1.4919296503067017,
-      "learning_rate": 9.667738693467337e-05,
-      "loss": 6.042,
-      "step": 3807
-    },
-    {
-      "epoch": 1.9859191655801824,
-      "grad_norm": 1.5292043685913086,
-      "learning_rate": 9.667638190954774e-05,
-      "loss": 6.1426,
-      "step": 3808
-    },
-    {
-      "epoch": 1.9864406779661017,
-      "grad_norm": 1.6136348247528076,
-      "learning_rate": 9.667537688442211e-05,
-      "loss": 5.7518,
-      "step": 3809
-    },
-    {
-      "epoch": 1.986962190352021,
-      "grad_norm": 1.4757779836654663,
-      "learning_rate": 9.667437185929649e-05,
-      "loss": 6.2082,
-      "step": 3810
-    },
-    {
-      "epoch": 1.98748370273794,
-      "grad_norm": 1.518336296081543,
-      "learning_rate": 9.667336683417087e-05,
-      "loss": 5.8855,
-      "step": 3811
-    },
-    {
-      "epoch": 1.9880052151238592,
-      "grad_norm": 1.4585412740707397,
-      "learning_rate": 9.667236180904523e-05,
-      "loss": 5.9526,
-      "step": 3812
-    },
-    {
-      "epoch": 1.9885267275097784,
-      "grad_norm": 1.7243646383285522,
-      "learning_rate": 9.66713567839196e-05,
-      "loss": 5.7422,
-      "step": 3813
-    },
-    {
-      "epoch": 1.9890482398956975,
-      "grad_norm": 1.3863049745559692,
-      "learning_rate": 9.667035175879398e-05,
-      "loss": 6.2211,
-      "step": 3814
-    },
-    {
-      "epoch": 1.9895697522816167,
-      "grad_norm": 1.8290050029754639,
-      "learning_rate": 9.666934673366834e-05,
-      "loss": 5.6805,
-      "step": 3815
-    },
-    {
-      "epoch": 1.990091264667536,
-      "grad_norm": 1.4061254262924194,
-      "learning_rate": 9.666834170854272e-05,
-      "loss": 6.1466,
-      "step": 3816
-    },
-    {
-      "epoch": 1.990612777053455,
-      "grad_norm": 1.6117753982543945,
-      "learning_rate": 9.666733668341708e-05,
-      "loss": 5.9243,
-      "step": 3817
-    },
-    {
-      "epoch": 1.9911342894393742,
-      "grad_norm": 1.3681560754776,
-      "learning_rate": 9.666633165829146e-05,
-      "loss": 6.0134,
-      "step": 3818
-    },
-    {
-      "epoch": 1.9916558018252934,
-      "grad_norm": 1.6231834888458252,
-      "learning_rate": 9.666532663316584e-05,
-      "loss": 5.595,
-      "step": 3819
-    },
-    {
-      "epoch": 1.9921773142112125,
-      "grad_norm": 1.4839259386062622,
-      "learning_rate": 9.666432160804022e-05,
-      "loss": 6.0455,
-      "step": 3820
-    },
-    {
-      "epoch": 1.9926988265971317,
-      "grad_norm": 1.4152849912643433,
-      "learning_rate": 9.666331658291458e-05,
-      "loss": 5.9413,
-      "step": 3821
-    },
-    {
-      "epoch": 1.993220338983051,
-      "grad_norm": 1.4076203107833862,
-      "learning_rate": 9.666231155778896e-05,
-      "loss": 6.0805,
-      "step": 3822
-    },
-    {
-      "epoch": 1.99374185136897,
-      "grad_norm": 1.4839211702346802,
-      "learning_rate": 9.666130653266332e-05,
-      "loss": 6.1253,
-      "step": 3823
-    },
-    {
-      "epoch": 1.9942633637548892,
-      "grad_norm": 1.455725073814392,
-      "learning_rate": 9.66603015075377e-05,
-      "loss": 6.0291,
-      "step": 3824
-    },
-    {
-      "epoch": 1.9947848761408085,
-      "grad_norm": 1.6449155807495117,
-      "learning_rate": 9.665929648241206e-05,
-      "loss": 6.0575,
-      "step": 3825
-    },
-    {
-      "epoch": 1.9953063885267275,
-      "grad_norm": 1.462751865386963,
-      "learning_rate": 9.665829145728643e-05,
-      "loss": 6.1895,
-      "step": 3826
-    },
-    {
-      "epoch": 1.9958279009126467,
-      "grad_norm": 1.4507899284362793,
-      "learning_rate": 9.66572864321608e-05,
-      "loss": 6.1622,
-      "step": 3827
-    },
-    {
-      "epoch": 1.996349413298566,
-      "grad_norm": 1.6966131925582886,
-      "learning_rate": 9.665628140703517e-05,
-      "loss": 4.9905,
-      "step": 3828
-    },
-    {
-      "epoch": 1.996870925684485,
-      "grad_norm": 1.7638691663742065,
-      "learning_rate": 9.665527638190955e-05,
-      "loss": 5.4357,
-      "step": 3829
-    },
-    {
-      "epoch": 1.997392438070404,
-      "grad_norm": 1.645431637763977,
-      "learning_rate": 9.665427135678393e-05,
-      "loss": 5.9473,
-      "step": 3830
-    },
-    {
-      "epoch": 1.9979139504563235,
-      "grad_norm": 1.3116331100463867,
-      "learning_rate": 9.66532663316583e-05,
-      "loss": 6.1634,
-      "step": 3831
-    },
-    {
-      "epoch": 1.9984354628422425,
-      "grad_norm": 1.5586097240447998,
-      "learning_rate": 9.665226130653267e-05,
-      "loss": 5.2557,
-      "step": 3832
-    },
-    {
-      "epoch": 1.9989569752281615,
-      "grad_norm": 1.672777771949768,
-      "learning_rate": 9.665125628140705e-05,
-      "loss": 5.6665,
-      "step": 3833
-    },
-    {
-      "epoch": 1.999478487614081,
-      "grad_norm": 1.5142594575881958,
-      "learning_rate": 9.665025125628141e-05,
-      "loss": 5.9705,
-      "step": 3834
-    },
-    {
-      "epoch": 2.0,
-      "grad_norm": 1.6594988107681274,
-      "learning_rate": 9.664924623115579e-05,
-      "loss": 5.7251,
-      "step": 3835
-    },
-    {
-      "epoch": 2.000521512385919,
-      "grad_norm": 1.591359257698059,
-      "learning_rate": 9.664824120603015e-05,
-      "loss": 6.2197,
-      "step": 3836
-    },
-    {
-      "epoch": 2.0010430247718385,
-      "grad_norm": 1.522283673286438,
-      "learning_rate": 9.664723618090453e-05,
-      "loss": 5.9446,
-      "step": 3837
-    },
-    {
-      "epoch": 2.0015645371577575,
-      "grad_norm": 1.6052857637405396,
-      "learning_rate": 9.66462311557789e-05,
-      "loss": 5.6827,
-      "step": 3838
-    },
-    {
-      "epoch": 2.0020860495436765,
-      "grad_norm": 1.4782124757766724,
-      "learning_rate": 9.664522613065327e-05,
-      "loss": 6.2934,
-      "step": 3839
-    },
-    {
-      "epoch": 2.002607561929596,
-      "grad_norm": 1.678267240524292,
-      "learning_rate": 9.664422110552765e-05,
-      "loss": 5.896,
-      "step": 3840
-    },
-    {
-      "epoch": 2.003129074315515,
-      "grad_norm": 1.5414811372756958,
-      "learning_rate": 9.664321608040201e-05,
-      "loss": 6.1196,
-      "step": 3841
-    },
-    {
-      "epoch": 2.003650586701434,
-      "grad_norm": 1.3931604623794556,
-      "learning_rate": 9.664221105527639e-05,
-      "loss": 6.2276,
-      "step": 3842
-    },
-    {
-      "epoch": 2.0041720990873535,
-      "grad_norm": 1.4502414464950562,
-      "learning_rate": 9.664120603015076e-05,
-      "loss": 6.0468,
-      "step": 3843
-    },
-    {
-      "epoch": 2.0046936114732725,
-      "grad_norm": 1.5150848627090454,
-      "learning_rate": 9.664020100502513e-05,
-      "loss": 6.2001,
-      "step": 3844
-    },
-    {
-      "epoch": 2.0052151238591915,
-      "grad_norm": 1.566394329071045,
-      "learning_rate": 9.66391959798995e-05,
-      "loss": 5.3592,
-      "step": 3845
-    },
-    {
-      "epoch": 2.005736636245111,
-      "grad_norm": 1.4968899488449097,
-      "learning_rate": 9.663819095477388e-05,
-      "loss": 5.6579,
-      "step": 3846
-    },
-    {
-      "epoch": 2.00625814863103,
-      "grad_norm": 1.525282621383667,
-      "learning_rate": 9.663718592964824e-05,
-      "loss": 5.6269,
-      "step": 3847
-    },
-    {
-      "epoch": 2.006779661016949,
-      "grad_norm": 1.497970461845398,
-      "learning_rate": 9.663618090452262e-05,
-      "loss": 6.0927,
-      "step": 3848
-    },
-    {
-      "epoch": 2.0073011734028685,
-      "grad_norm": 1.5611066818237305,
-      "learning_rate": 9.663517587939698e-05,
-      "loss": 6.1371,
-      "step": 3849
-    },
-    {
-      "epoch": 2.0078226857887875,
-      "grad_norm": 1.592439889907837,
-      "learning_rate": 9.663417085427136e-05,
-      "loss": 5.9127,
-      "step": 3850
-    },
-    {
-      "epoch": 2.0083441981747066,
-      "grad_norm": 1.4829678535461426,
-      "learning_rate": 9.663316582914574e-05,
-      "loss": 5.91,
-      "step": 3851
-    },
-    {
-      "epoch": 2.008865710560626,
-      "grad_norm": 1.4874320030212402,
-      "learning_rate": 9.66321608040201e-05,
-      "loss": 5.9968,
-      "step": 3852
-    },
-    {
-      "epoch": 2.009387222946545,
-      "grad_norm": 1.4022375345230103,
-      "learning_rate": 9.663115577889448e-05,
-      "loss": 5.5431,
-      "step": 3853
-    },
-    {
-      "epoch": 2.009908735332464,
-      "grad_norm": 1.426586627960205,
-      "learning_rate": 9.663015075376884e-05,
-      "loss": 5.9981,
-      "step": 3854
-    },
-    {
-      "epoch": 2.0104302477183835,
-      "grad_norm": 1.6040619611740112,
-      "learning_rate": 9.662914572864322e-05,
-      "loss": 5.7153,
-      "step": 3855
-    },
-    {
-      "epoch": 2.0109517601043025,
-      "grad_norm": 1.6716164350509644,
-      "learning_rate": 9.662814070351759e-05,
-      "loss": 6.1254,
-      "step": 3856
-    },
-    {
-      "epoch": 2.0114732724902216,
-      "grad_norm": 1.7159833908081055,
-      "learning_rate": 9.662713567839196e-05,
-      "loss": 5.5416,
-      "step": 3857
-    },
-    {
-      "epoch": 2.011994784876141,
-      "grad_norm": 1.431359887123108,
-      "learning_rate": 9.662613065326633e-05,
-      "loss": 6.0572,
-      "step": 3858
-    },
-    {
-      "epoch": 2.01251629726206,
-      "grad_norm": 1.6079773902893066,
-      "learning_rate": 9.662512562814071e-05,
-      "loss": 6.0283,
-      "step": 3859
-    },
-    {
-      "epoch": 2.013037809647979,
-      "grad_norm": 1.5487617254257202,
-      "learning_rate": 9.662412060301508e-05,
-      "loss": 6.2086,
-      "step": 3860
-    },
-    {
-      "epoch": 2.013559322033898,
-      "grad_norm": 1.469484806060791,
-      "learning_rate": 9.662311557788946e-05,
-      "loss": 5.9104,
-      "step": 3861
-    },
-    {
-      "epoch": 2.0140808344198176,
-      "grad_norm": 1.5899367332458496,
-      "learning_rate": 9.662211055276383e-05,
-      "loss": 5.9402,
-      "step": 3862
-    },
-    {
-      "epoch": 2.0146023468057366,
-      "grad_norm": 1.7232006788253784,
-      "learning_rate": 9.66211055276382e-05,
-      "loss": 5.7653,
-      "step": 3863
-    },
-    {
-      "epoch": 2.0151238591916556,
-      "grad_norm": 1.868619441986084,
-      "learning_rate": 9.662010050251257e-05,
-      "loss": 5.4457,
-      "step": 3864
-    },
-    {
-      "epoch": 2.015645371577575,
-      "grad_norm": 1.5389204025268555,
-      "learning_rate": 9.661909547738693e-05,
-      "loss": 5.9007,
-      "step": 3865
-    },
-    {
-      "epoch": 2.016166883963494,
-      "grad_norm": 1.610851764678955,
-      "learning_rate": 9.661809045226131e-05,
-      "loss": 5.8552,
-      "step": 3866
-    },
-    {
-      "epoch": 2.016688396349413,
-      "grad_norm": 1.612640619277954,
-      "learning_rate": 9.661708542713568e-05,
-      "loss": 6.0428,
-      "step": 3867
-    },
-    {
-      "epoch": 2.0172099087353326,
-      "grad_norm": 1.695522427558899,
-      "learning_rate": 9.661608040201005e-05,
-      "loss": 5.5925,
-      "step": 3868
-    },
-    {
-      "epoch": 2.0177314211212516,
-      "grad_norm": 1.6532551050186157,
-      "learning_rate": 9.661507537688442e-05,
-      "loss": 5.7311,
-      "step": 3869
-    },
-    {
-      "epoch": 2.0182529335071706,
-      "grad_norm": 1.60237455368042,
-      "learning_rate": 9.66140703517588e-05,
-      "loss": 5.9292,
-      "step": 3870
-    },
-    {
-      "epoch": 2.01877444589309,
-      "grad_norm": 1.5588821172714233,
-      "learning_rate": 9.661306532663317e-05,
-      "loss": 6.1017,
-      "step": 3871
-    },
-    {
-      "epoch": 2.019295958279009,
-      "grad_norm": 1.529531478881836,
-      "learning_rate": 9.661206030150755e-05,
-      "loss": 5.6163,
-      "step": 3872
-    },
-    {
-      "epoch": 2.019817470664928,
-      "grad_norm": 1.6026877164840698,
-      "learning_rate": 9.661105527638192e-05,
-      "loss": 6.1441,
-      "step": 3873
-    },
-    {
-      "epoch": 2.0203389830508476,
-      "grad_norm": 1.5312128067016602,
-      "learning_rate": 9.661005025125629e-05,
-      "loss": 5.8343,
-      "step": 3874
-    },
-    {
-      "epoch": 2.0208604954367666,
-      "grad_norm": 1.5682915449142456,
-      "learning_rate": 9.660904522613066e-05,
-      "loss": 5.9322,
-      "step": 3875
-    },
-    {
-      "epoch": 2.0213820078226856,
-      "grad_norm": 1.4330214262008667,
-      "learning_rate": 9.660804020100504e-05,
-      "loss": 5.642,
-      "step": 3876
-    },
-    {
-      "epoch": 2.021903520208605,
-      "grad_norm": 1.4143586158752441,
-      "learning_rate": 9.66070351758794e-05,
-      "loss": 5.9079,
-      "step": 3877
-    },
-    {
-      "epoch": 2.022425032594524,
-      "grad_norm": 1.4480289220809937,
-      "learning_rate": 9.660603015075376e-05,
-      "loss": 6.0554,
-      "step": 3878
-    },
-    {
-      "epoch": 2.022946544980443,
-      "grad_norm": 1.5762571096420288,
-      "learning_rate": 9.660502512562814e-05,
-      "loss": 5.8765,
-      "step": 3879
-    },
-    {
-      "epoch": 2.0234680573663626,
-      "grad_norm": 1.5856318473815918,
-      "learning_rate": 9.660402010050252e-05,
-      "loss": 5.3811,
-      "step": 3880
-    },
-    {
-      "epoch": 2.0239895697522816,
-      "grad_norm": 1.4558180570602417,
-      "learning_rate": 9.66030150753769e-05,
-      "loss": 5.811,
-      "step": 3881
-    },
-    {
-      "epoch": 2.0245110821382006,
-      "grad_norm": 1.688279151916504,
-      "learning_rate": 9.660201005025126e-05,
-      "loss": 5.6806,
-      "step": 3882
-    },
-    {
-      "epoch": 2.02503259452412,
-      "grad_norm": 1.5003595352172852,
-      "learning_rate": 9.660100502512564e-05,
-      "loss": 5.928,
-      "step": 3883
-    },
-    {
-      "epoch": 2.025554106910039,
-      "grad_norm": 1.7635873556137085,
-      "learning_rate": 9.66e-05,
-      "loss": 5.3602,
-      "step": 3884
-    },
-    {
-      "epoch": 2.026075619295958,
-      "grad_norm": 1.4176634550094604,
-      "learning_rate": 9.659899497487438e-05,
-      "loss": 6.0984,
-      "step": 3885
-    },
-    {
-      "epoch": 2.0265971316818776,
-      "grad_norm": 1.5006334781646729,
-      "learning_rate": 9.659798994974875e-05,
-      "loss": 5.7193,
-      "step": 3886
-    },
-    {
-      "epoch": 2.0271186440677966,
-      "grad_norm": 1.4504597187042236,
-      "learning_rate": 9.659698492462312e-05,
-      "loss": 5.9575,
-      "step": 3887
-    },
-    {
-      "epoch": 2.0276401564537156,
-      "grad_norm": 1.6058359146118164,
-      "learning_rate": 9.659597989949749e-05,
-      "loss": 6.0705,
-      "step": 3888
-    },
-    {
-      "epoch": 2.028161668839635,
-      "grad_norm": 1.6142563819885254,
-      "learning_rate": 9.659497487437187e-05,
-      "loss": 5.5714,
-      "step": 3889
-    },
-    {
-      "epoch": 2.028683181225554,
-      "grad_norm": 1.6901997327804565,
-      "learning_rate": 9.659396984924623e-05,
-      "loss": 5.81,
-      "step": 3890
-    },
-    {
-      "epoch": 2.029204693611473,
-      "grad_norm": 1.5869756937026978,
-      "learning_rate": 9.659296482412061e-05,
-      "loss": 5.9748,
-      "step": 3891
-    },
-    {
-      "epoch": 2.0297262059973926,
-      "grad_norm": 1.6090866327285767,
-      "learning_rate": 9.659195979899499e-05,
-      "loss": 6.1625,
-      "step": 3892
-    },
-    {
-      "epoch": 2.0302477183833116,
-      "grad_norm": 1.4449563026428223,
-      "learning_rate": 9.659095477386935e-05,
-      "loss": 5.9832,
-      "step": 3893
-    },
-    {
-      "epoch": 2.0307692307692307,
-      "grad_norm": 1.5957400798797607,
-      "learning_rate": 9.658994974874373e-05,
-      "loss": 6.0209,
-      "step": 3894
-    },
-    {
-      "epoch": 2.03129074315515,
-      "grad_norm": 1.4640491008758545,
-      "learning_rate": 9.658894472361809e-05,
-      "loss": 5.9661,
-      "step": 3895
-    },
-    {
-      "epoch": 2.031812255541069,
-      "grad_norm": 1.554290771484375,
-      "learning_rate": 9.658793969849247e-05,
-      "loss": 6.0164,
-      "step": 3896
-    },
-    {
-      "epoch": 2.032333767926988,
-      "grad_norm": 1.6988391876220703,
-      "learning_rate": 9.658693467336683e-05,
-      "loss": 5.5334,
-      "step": 3897
-    },
-    {
-      "epoch": 2.0328552803129076,
-      "grad_norm": 1.529518723487854,
-      "learning_rate": 9.658592964824121e-05,
-      "loss": 6.1945,
-      "step": 3898
-    },
-    {
-      "epoch": 2.0333767926988267,
-      "grad_norm": 1.6130419969558716,
-      "learning_rate": 9.658492462311558e-05,
-      "loss": 5.6871,
-      "step": 3899
-    },
-    {
-      "epoch": 2.0338983050847457,
-      "grad_norm": 1.4580706357955933,
-      "learning_rate": 9.658391959798995e-05,
-      "loss": 6.0258,
-      "step": 3900
-    },
-    {
-      "epoch": 2.034419817470665,
-      "grad_norm": 1.4608551263809204,
-      "learning_rate": 9.658291457286432e-05,
-      "loss": 6.0448,
-      "step": 3901
-    },
-    {
-      "epoch": 2.034941329856584,
-      "grad_norm": 1.5851224660873413,
-      "learning_rate": 9.65819095477387e-05,
-      "loss": 6.0628,
-      "step": 3902
-    },
-    {
-      "epoch": 2.035462842242503,
-      "grad_norm": 1.6293730735778809,
-      "learning_rate": 9.658090452261307e-05,
-      "loss": 5.6049,
-      "step": 3903
-    },
-    {
-      "epoch": 2.0359843546284226,
-      "grad_norm": 1.6573991775512695,
-      "learning_rate": 9.657989949748745e-05,
-      "loss": 5.5879,
-      "step": 3904
-    },
-    {
-      "epoch": 2.0365058670143417,
-      "grad_norm": 1.5741170644760132,
-      "learning_rate": 9.657889447236182e-05,
-      "loss": 5.6968,
-      "step": 3905
-    },
-    {
-      "epoch": 2.0370273794002607,
-      "grad_norm": 1.6803165674209595,
-      "learning_rate": 9.657788944723618e-05,
-      "loss": 5.8301,
-      "step": 3906
-    },
-    {
-      "epoch": 2.0375488917861797,
-      "grad_norm": 1.5366028547286987,
-      "learning_rate": 9.657688442211056e-05,
-      "loss": 5.3497,
-      "step": 3907
-    },
-    {
-      "epoch": 2.038070404172099,
-      "grad_norm": 1.7429250478744507,
-      "learning_rate": 9.657587939698492e-05,
-      "loss": 5.9486,
-      "step": 3908
-    },
-    {
-      "epoch": 2.038591916558018,
-      "grad_norm": 1.690645694732666,
-      "learning_rate": 9.65748743718593e-05,
-      "loss": 6.1225,
-      "step": 3909
-    },
-    {
-      "epoch": 2.039113428943937,
-      "grad_norm": 1.5855175256729126,
-      "learning_rate": 9.657386934673366e-05,
-      "loss": 4.9894,
-      "step": 3910
-    },
-    {
-      "epoch": 2.0396349413298567,
-      "grad_norm": 1.6470962762832642,
-      "learning_rate": 9.657286432160804e-05,
-      "loss": 6.0837,
-      "step": 3911
-    },
-    {
-      "epoch": 2.0401564537157757,
-      "grad_norm": 1.7859163284301758,
-      "learning_rate": 9.657185929648242e-05,
-      "loss": 5.9401,
-      "step": 3912
-    },
-    {
-      "epoch": 2.0406779661016947,
-      "grad_norm": 1.4241307973861694,
-      "learning_rate": 9.65708542713568e-05,
-      "loss": 6.3431,
-      "step": 3913
-    },
-    {
-      "epoch": 2.041199478487614,
-      "grad_norm": 1.505110263824463,
-      "learning_rate": 9.656984924623116e-05,
-      "loss": 5.9135,
-      "step": 3914
-    },
-    {
-      "epoch": 2.041720990873533,
-      "grad_norm": 1.7363044023513794,
-      "learning_rate": 9.656884422110554e-05,
-      "loss": 6.1511,
-      "step": 3915
-    },
-    {
-      "epoch": 2.0422425032594522,
-      "grad_norm": 1.6521581411361694,
-      "learning_rate": 9.65678391959799e-05,
-      "loss": 5.6167,
-      "step": 3916
-    },
-    {
-      "epoch": 2.0427640156453717,
-      "grad_norm": 1.510424017906189,
-      "learning_rate": 9.656683417085428e-05,
-      "loss": 5.6457,
-      "step": 3917
-    },
-    {
-      "epoch": 2.0432855280312907,
-      "grad_norm": 1.8391637802124023,
-      "learning_rate": 9.656582914572865e-05,
-      "loss": 6.0547,
-      "step": 3918
-    },
-    {
-      "epoch": 2.0438070404172097,
-      "grad_norm": 1.5372259616851807,
-      "learning_rate": 9.656482412060301e-05,
-      "loss": 5.4644,
-      "step": 3919
-    },
-    {
-      "epoch": 2.044328552803129,
-      "grad_norm": 1.5180379152297974,
-      "learning_rate": 9.656381909547739e-05,
-      "loss": 5.8849,
-      "step": 3920
-    },
-    {
-      "epoch": 2.044850065189048,
-      "grad_norm": 1.5068578720092773,
-      "learning_rate": 9.656281407035175e-05,
-      "loss": 5.9186,
-      "step": 3921
-    },
-    {
-      "epoch": 2.0453715775749672,
-      "grad_norm": 1.4703480005264282,
-      "learning_rate": 9.656180904522613e-05,
-      "loss": 6.3898,
-      "step": 3922
-    },
-    {
-      "epoch": 2.0458930899608867,
-      "grad_norm": 1.5425384044647217,
-      "learning_rate": 9.656080402010051e-05,
-      "loss": 5.5468,
-      "step": 3923
-    },
-    {
-      "epoch": 2.0464146023468057,
-      "grad_norm": 1.9228389263153076,
-      "learning_rate": 9.655979899497489e-05,
-      "loss": 5.6935,
-      "step": 3924
-    },
-    {
-      "epoch": 2.0469361147327247,
-      "grad_norm": 1.5889586210250854,
-      "learning_rate": 9.655879396984925e-05,
-      "loss": 5.598,
-      "step": 3925
-    },
-    {
-      "epoch": 2.047457627118644,
-      "grad_norm": 1.8198912143707275,
-      "learning_rate": 9.655778894472363e-05,
-      "loss": 5.7975,
-      "step": 3926
-    },
-    {
-      "epoch": 2.0479791395045632,
-      "grad_norm": 1.7293827533721924,
-      "learning_rate": 9.655678391959799e-05,
-      "loss": 5.581,
-      "step": 3927
-    },
-    {
-      "epoch": 2.0485006518904822,
-      "grad_norm": 1.5769946575164795,
-      "learning_rate": 9.655577889447237e-05,
-      "loss": 6.0248,
-      "step": 3928
-    },
-    {
-      "epoch": 2.0490221642764017,
-      "grad_norm": 1.509264588356018,
-      "learning_rate": 9.655477386934673e-05,
-      "loss": 5.905,
-      "step": 3929
-    },
-    {
-      "epoch": 2.0495436766623207,
-      "grad_norm": 1.5452791452407837,
-      "learning_rate": 9.655376884422111e-05,
-      "loss": 5.8895,
-      "step": 3930
-    },
-    {
-      "epoch": 2.0500651890482398,
-      "grad_norm": 1.654995083808899,
-      "learning_rate": 9.655276381909548e-05,
-      "loss": 5.5559,
-      "step": 3931
-    },
-    {
-      "epoch": 2.050586701434159,
-      "grad_norm": 1.5105726718902588,
-      "learning_rate": 9.655175879396985e-05,
-      "loss": 5.8949,
-      "step": 3932
-    },
-    {
-      "epoch": 2.0511082138200782,
-      "grad_norm": 1.3797494173049927,
-      "learning_rate": 9.655075376884423e-05,
-      "loss": 6.4024,
-      "step": 3933
-    },
-    {
-      "epoch": 2.0516297262059973,
-      "grad_norm": 1.5611279010772705,
-      "learning_rate": 9.65497487437186e-05,
-      "loss": 5.719,
-      "step": 3934
-    },
-    {
-      "epoch": 2.0521512385919167,
-      "grad_norm": 1.4523956775665283,
-      "learning_rate": 9.654874371859297e-05,
-      "loss": 6.3847,
-      "step": 3935
-    },
-    {
-      "epoch": 2.0526727509778357,
-      "grad_norm": 1.584633231163025,
-      "learning_rate": 9.654773869346734e-05,
-      "loss": 5.9134,
-      "step": 3936
-    },
-    {
-      "epoch": 2.0531942633637548,
-      "grad_norm": 1.4611663818359375,
-      "learning_rate": 9.654673366834172e-05,
-      "loss": 5.838,
-      "step": 3937
-    },
-    {
-      "epoch": 2.0537157757496742,
-      "grad_norm": 1.6152817010879517,
-      "learning_rate": 9.654572864321608e-05,
-      "loss": 6.1164,
-      "step": 3938
-    },
-    {
-      "epoch": 2.0542372881355933,
-      "grad_norm": 1.4726665019989014,
-      "learning_rate": 9.654472361809046e-05,
-      "loss": 5.9445,
-      "step": 3939
-    },
-    {
-      "epoch": 2.0547588005215123,
-      "grad_norm": 1.4976749420166016,
-      "learning_rate": 9.654371859296482e-05,
-      "loss": 5.5959,
-      "step": 3940
-    },
-    {
-      "epoch": 2.0552803129074317,
-      "grad_norm": 1.5572130680084229,
-      "learning_rate": 9.65427135678392e-05,
-      "loss": 5.8448,
-      "step": 3941
-    },
-    {
-      "epoch": 2.0558018252933508,
-      "grad_norm": 1.5943560600280762,
-      "learning_rate": 9.654170854271357e-05,
-      "loss": 6.1084,
-      "step": 3942
-    },
-    {
-      "epoch": 2.05632333767927,
-      "grad_norm": 1.4644074440002441,
-      "learning_rate": 9.654070351758794e-05,
-      "loss": 6.2114,
-      "step": 3943
-    },
-    {
-      "epoch": 2.0568448500651892,
-      "grad_norm": 1.5441384315490723,
-      "learning_rate": 9.653969849246232e-05,
-      "loss": 5.6302,
-      "step": 3944
-    },
-    {
-      "epoch": 2.0573663624511083,
-      "grad_norm": 1.536192536354065,
-      "learning_rate": 9.653869346733669e-05,
-      "loss": 6.0086,
-      "step": 3945
-    },
-    {
-      "epoch": 2.0578878748370273,
-      "grad_norm": 1.895548701286316,
-      "learning_rate": 9.653768844221106e-05,
-      "loss": 5.5938,
-      "step": 3946
-    },
-    {
-      "epoch": 2.0584093872229468,
-      "grad_norm": 1.5269862413406372,
-      "learning_rate": 9.653668341708543e-05,
-      "loss": 5.9658,
-      "step": 3947
-    },
-    {
-      "epoch": 2.0589308996088658,
-      "grad_norm": 1.6709858179092407,
-      "learning_rate": 9.65356783919598e-05,
-      "loss": 5.8162,
-      "step": 3948
-    },
-    {
-      "epoch": 2.059452411994785,
-      "grad_norm": 1.7584911584854126,
-      "learning_rate": 9.653467336683417e-05,
-      "loss": 5.8389,
-      "step": 3949
-    },
-    {
-      "epoch": 2.0599739243807043,
-      "grad_norm": 1.6268733739852905,
-      "learning_rate": 9.653366834170855e-05,
-      "loss": 5.7966,
-      "step": 3950
-    },
-    {
-      "epoch": 2.0604954367666233,
-      "grad_norm": 1.7307806015014648,
-      "learning_rate": 9.653266331658291e-05,
-      "loss": 5.8234,
-      "step": 3951
-    },
-    {
-      "epoch": 2.0610169491525423,
-      "grad_norm": 1.8314112424850464,
-      "learning_rate": 9.653165829145729e-05,
-      "loss": 5.7399,
-      "step": 3952
-    },
-    {
-      "epoch": 2.0615384615384613,
-      "grad_norm": 1.503750205039978,
-      "learning_rate": 9.653065326633167e-05,
-      "loss": 5.6829,
-      "step": 3953
-    },
-    {
-      "epoch": 2.062059973924381,
-      "grad_norm": 1.414094090461731,
-      "learning_rate": 9.652964824120604e-05,
-      "loss": 6.158,
-      "step": 3954
-    },
-    {
-      "epoch": 2.0625814863103,
-      "grad_norm": 1.7355551719665527,
-      "learning_rate": 9.652864321608041e-05,
-      "loss": 5.5072,
-      "step": 3955
-    },
-    {
-      "epoch": 2.063102998696219,
-      "grad_norm": 1.463829517364502,
-      "learning_rate": 9.652763819095479e-05,
-      "loss": 6.0604,
-      "step": 3956
-    },
-    {
-      "epoch": 2.0636245110821383,
-      "grad_norm": 1.6242479085922241,
-      "learning_rate": 9.652663316582915e-05,
-      "loss": 6.0493,
-      "step": 3957
-    },
-    {
-      "epoch": 2.0641460234680573,
-      "grad_norm": 1.633217692375183,
-      "learning_rate": 9.652562814070352e-05,
-      "loss": 5.8793,
-      "step": 3958
-    },
-    {
-      "epoch": 2.0646675358539763,
-      "grad_norm": 1.720479965209961,
-      "learning_rate": 9.65246231155779e-05,
-      "loss": 6.0504,
-      "step": 3959
-    },
-    {
-      "epoch": 2.065189048239896,
-      "grad_norm": 1.557276725769043,
-      "learning_rate": 9.652361809045226e-05,
-      "loss": 5.9358,
-      "step": 3960
-    },
-    {
-      "epoch": 2.065710560625815,
-      "grad_norm": 1.6484123468399048,
-      "learning_rate": 9.652261306532664e-05,
-      "loss": 5.8154,
-      "step": 3961
-    },
-    {
-      "epoch": 2.066232073011734,
-      "grad_norm": 1.497696876525879,
-      "learning_rate": 9.6521608040201e-05,
-      "loss": 6.0611,
-      "step": 3962
-    },
-    {
-      "epoch": 2.0667535853976533,
-      "grad_norm": 1.5143600702285767,
-      "learning_rate": 9.652060301507538e-05,
-      "loss": 5.915,
-      "step": 3963
-    },
-    {
-      "epoch": 2.0672750977835723,
-      "grad_norm": 1.6158775091171265,
-      "learning_rate": 9.651959798994976e-05,
-      "loss": 5.9119,
-      "step": 3964
-    },
-    {
-      "epoch": 2.0677966101694913,
-      "grad_norm": 1.7102049589157104,
-      "learning_rate": 9.651859296482413e-05,
-      "loss": 5.7317,
-      "step": 3965
-    },
-    {
-      "epoch": 2.068318122555411,
-      "grad_norm": 1.4505635499954224,
-      "learning_rate": 9.65175879396985e-05,
-      "loss": 6.0799,
-      "step": 3966
-    },
-    {
-      "epoch": 2.06883963494133,
-      "grad_norm": 1.8569592237472534,
-      "learning_rate": 9.651658291457288e-05,
-      "loss": 5.7805,
-      "step": 3967
-    },
-    {
-      "epoch": 2.069361147327249,
-      "grad_norm": 1.5475157499313354,
-      "learning_rate": 9.651557788944724e-05,
-      "loss": 6.171,
-      "step": 3968
-    },
-    {
-      "epoch": 2.0698826597131683,
-      "grad_norm": 1.5666013956069946,
-      "learning_rate": 9.651457286432162e-05,
-      "loss": 5.9443,
-      "step": 3969
-    },
-    {
-      "epoch": 2.0704041720990873,
-      "grad_norm": 1.365666151046753,
-      "learning_rate": 9.651356783919598e-05,
-      "loss": 6.1164,
-      "step": 3970
-    },
-    {
-      "epoch": 2.0709256844850064,
-      "grad_norm": 1.5494128465652466,
-      "learning_rate": 9.651256281407036e-05,
-      "loss": 5.8838,
-      "step": 3971
-    },
-    {
-      "epoch": 2.071447196870926,
-      "grad_norm": 1.8812227249145508,
-      "learning_rate": 9.651155778894472e-05,
-      "loss": 5.8664,
-      "step": 3972
-    },
-    {
-      "epoch": 2.071968709256845,
-      "grad_norm": 1.7095237970352173,
-      "learning_rate": 9.65105527638191e-05,
-      "loss": 5.9277,
-      "step": 3973
-    },
-    {
-      "epoch": 2.072490221642764,
-      "grad_norm": 1.5555728673934937,
-      "learning_rate": 9.650954773869348e-05,
-      "loss": 6.0149,
-      "step": 3974
-    },
-    {
-      "epoch": 2.0730117340286833,
-      "grad_norm": 1.5653584003448486,
-      "learning_rate": 9.650854271356784e-05,
-      "loss": 5.8689,
-      "step": 3975
-    },
-    {
-      "epoch": 2.0735332464146023,
-      "grad_norm": 1.4866209030151367,
-      "learning_rate": 9.650753768844222e-05,
-      "loss": 5.9425,
-      "step": 3976
-    },
-    {
-      "epoch": 2.0740547588005214,
-      "grad_norm": 2.0871634483337402,
-      "learning_rate": 9.650653266331659e-05,
-      "loss": 5.2804,
-      "step": 3977
-    },
-    {
-      "epoch": 2.074576271186441,
-      "grad_norm": 1.7012065649032593,
-      "learning_rate": 9.650552763819096e-05,
-      "loss": 5.672,
-      "step": 3978
-    },
-    {
-      "epoch": 2.07509778357236,
-      "grad_norm": 1.4521323442459106,
-      "learning_rate": 9.650452261306533e-05,
-      "loss": 6.3439,
-      "step": 3979
-    },
-    {
-      "epoch": 2.075619295958279,
-      "grad_norm": 1.5571131706237793,
-      "learning_rate": 9.65035175879397e-05,
-      "loss": 6.0636,
-      "step": 3980
-    },
-    {
-      "epoch": 2.0761408083441983,
-      "grad_norm": 1.4965920448303223,
-      "learning_rate": 9.650251256281407e-05,
-      "loss": 5.7312,
-      "step": 3981
-    },
-    {
-      "epoch": 2.0766623207301174,
-      "grad_norm": 1.4877055883407593,
-      "learning_rate": 9.650150753768845e-05,
-      "loss": 5.7582,
-      "step": 3982
-    },
-    {
-      "epoch": 2.0771838331160364,
-      "grad_norm": 1.5904436111450195,
-      "learning_rate": 9.650050251256281e-05,
-      "loss": 5.2658,
-      "step": 3983
-    },
-    {
-      "epoch": 2.077705345501956,
-      "grad_norm": 1.57761549949646,
-      "learning_rate": 9.649949748743719e-05,
-      "loss": 6.02,
-      "step": 3984
-    },
-    {
-      "epoch": 2.078226857887875,
-      "grad_norm": 1.5657042264938354,
-      "learning_rate": 9.649849246231157e-05,
-      "loss": 5.9347,
-      "step": 3985
-    },
-    {
-      "epoch": 2.078748370273794,
-      "grad_norm": 1.5054913759231567,
-      "learning_rate": 9.649748743718593e-05,
-      "loss": 6.1691,
-      "step": 3986
-    },
-    {
-      "epoch": 2.0792698826597134,
-      "grad_norm": 1.5268501043319702,
-      "learning_rate": 9.649648241206031e-05,
-      "loss": 5.8295,
-      "step": 3987
-    },
-    {
-      "epoch": 2.0797913950456324,
-      "grad_norm": 1.723880410194397,
-      "learning_rate": 9.649547738693467e-05,
-      "loss": 5.8641,
-      "step": 3988
-    },
-    {
-      "epoch": 2.0803129074315514,
-      "grad_norm": 1.9011268615722656,
-      "learning_rate": 9.649447236180905e-05,
-      "loss": 6.0203,
-      "step": 3989
-    },
-    {
-      "epoch": 2.080834419817471,
-      "grad_norm": 1.5875879526138306,
-      "learning_rate": 9.649346733668342e-05,
-      "loss": 5.5395,
-      "step": 3990
-    },
-    {
-      "epoch": 2.08135593220339,
-      "grad_norm": 1.588598370552063,
-      "learning_rate": 9.64924623115578e-05,
-      "loss": 5.9627,
-      "step": 3991
-    },
-    {
-      "epoch": 2.081877444589309,
-      "grad_norm": 1.5612637996673584,
-      "learning_rate": 9.649145728643216e-05,
-      "loss": 5.4503,
-      "step": 3992
-    },
-    {
-      "epoch": 2.0823989569752284,
-      "grad_norm": 1.4996964931488037,
-      "learning_rate": 9.649045226130654e-05,
-      "loss": 5.8918,
-      "step": 3993
-    },
-    {
-      "epoch": 2.0829204693611474,
-      "grad_norm": 1.5235801935195923,
-      "learning_rate": 9.648944723618091e-05,
-      "loss": 6.0025,
-      "step": 3994
-    },
-    {
-      "epoch": 2.0834419817470664,
-      "grad_norm": 1.4258363246917725,
-      "learning_rate": 9.648844221105529e-05,
-      "loss": 5.8052,
-      "step": 3995
-    },
-    {
-      "epoch": 2.083963494132986,
-      "grad_norm": 1.5788785219192505,
-      "learning_rate": 9.648743718592966e-05,
-      "loss": 5.9388,
-      "step": 3996
-    },
-    {
-      "epoch": 2.084485006518905,
-      "grad_norm": 1.5934689044952393,
-      "learning_rate": 9.648643216080403e-05,
-      "loss": 5.2253,
-      "step": 3997
-    },
-    {
-      "epoch": 2.085006518904824,
-      "grad_norm": 1.4850819110870361,
-      "learning_rate": 9.64854271356784e-05,
-      "loss": 5.9429,
-      "step": 3998
-    },
-    {
-      "epoch": 2.0855280312907434,
-      "grad_norm": 1.6332536935806274,
-      "learning_rate": 9.648442211055276e-05,
-      "loss": 6.106,
-      "step": 3999
-    },
-    {
-      "epoch": 2.0860495436766624,
-      "grad_norm": 1.7181698083877563,
-      "learning_rate": 9.648341708542714e-05,
-      "loss": 5.8514,
-      "step": 4000
-    },
-    {
-      "epoch": 2.0865710560625814,
-      "grad_norm": 1.4431015253067017,
-      "learning_rate": 9.64824120603015e-05,
-      "loss": 6.1348,
-      "step": 4001
-    },
-    {
-      "epoch": 2.0870925684485004,
-      "grad_norm": 1.458450198173523,
-      "learning_rate": 9.648140703517588e-05,
-      "loss": 5.9369,
-      "step": 4002
-    },
-    {
-      "epoch": 2.08761408083442,
-      "grad_norm": 1.7926479578018188,
-      "learning_rate": 9.648040201005025e-05,
-      "loss": 5.5524,
-      "step": 4003
-    },
-    {
-      "epoch": 2.088135593220339,
-      "grad_norm": 1.5437053442001343,
-      "learning_rate": 9.647939698492462e-05,
-      "loss": 5.9147,
-      "step": 4004
-    },
-    {
-      "epoch": 2.088657105606258,
-      "grad_norm": 1.5385366678237915,
-      "learning_rate": 9.6478391959799e-05,
-      "loss": 5.8543,
-      "step": 4005
-    },
-    {
-      "epoch": 2.0891786179921774,
-      "grad_norm": 1.580931305885315,
-      "learning_rate": 9.647738693467338e-05,
-      "loss": 5.8669,
-      "step": 4006
-    },
-    {
-      "epoch": 2.0897001303780964,
-      "grad_norm": 1.5132478475570679,
-      "learning_rate": 9.647638190954774e-05,
-      "loss": 5.99,
-      "step": 4007
-    },
-    {
-      "epoch": 2.0902216427640155,
-      "grad_norm": 1.4598722457885742,
-      "learning_rate": 9.647537688442212e-05,
-      "loss": 6.3002,
-      "step": 4008
-    },
-    {
-      "epoch": 2.090743155149935,
-      "grad_norm": 1.5921202898025513,
-      "learning_rate": 9.647437185929649e-05,
-      "loss": 5.7844,
-      "step": 4009
-    },
-    {
-      "epoch": 2.091264667535854,
-      "grad_norm": 2.1312170028686523,
-      "learning_rate": 9.647336683417086e-05,
-      "loss": 5.0035,
-      "step": 4010
-    },
-    {
-      "epoch": 2.091786179921773,
-      "grad_norm": 1.5497820377349854,
-      "learning_rate": 9.647236180904523e-05,
-      "loss": 6.1362,
-      "step": 4011
-    },
-    {
-      "epoch": 2.0923076923076924,
-      "grad_norm": 1.6374157667160034,
-      "learning_rate": 9.647135678391959e-05,
-      "loss": 5.308,
-      "step": 4012
-    },
-    {
-      "epoch": 2.0928292046936114,
-      "grad_norm": 1.6109650135040283,
-      "learning_rate": 9.647035175879397e-05,
-      "loss": 5.9434,
-      "step": 4013
-    },
-    {
-      "epoch": 2.0933507170795305,
-      "grad_norm": 1.5867388248443604,
-      "learning_rate": 9.646934673366835e-05,
-      "loss": 6.306,
-      "step": 4014
-    },
-    {
-      "epoch": 2.09387222946545,
-      "grad_norm": 1.6429471969604492,
-      "learning_rate": 9.646834170854273e-05,
-      "loss": 5.8896,
-      "step": 4015
-    },
-    {
-      "epoch": 2.094393741851369,
-      "grad_norm": 1.5101276636123657,
-      "learning_rate": 9.646733668341709e-05,
-      "loss": 6.1561,
-      "step": 4016
-    },
-    {
-      "epoch": 2.094915254237288,
-      "grad_norm": 1.5615636110305786,
-      "learning_rate": 9.646633165829147e-05,
-      "loss": 5.803,
-      "step": 4017
-    },
-    {
-      "epoch": 2.0954367666232074,
-      "grad_norm": 1.7508108615875244,
-      "learning_rate": 9.646532663316583e-05,
-      "loss": 6.0263,
-      "step": 4018
-    },
-    {
-      "epoch": 2.0959582790091265,
-      "grad_norm": 1.6000175476074219,
-      "learning_rate": 9.646432160804021e-05,
-      "loss": 5.9665,
-      "step": 4019
-    },
-    {
-      "epoch": 2.0964797913950455,
-      "grad_norm": 1.7278801202774048,
-      "learning_rate": 9.646331658291457e-05,
-      "loss": 5.7844,
-      "step": 4020
-    },
-    {
-      "epoch": 2.097001303780965,
-      "grad_norm": 1.519051432609558,
-      "learning_rate": 9.646231155778895e-05,
-      "loss": 6.0921,
-      "step": 4021
-    },
-    {
-      "epoch": 2.097522816166884,
-      "grad_norm": 1.4523789882659912,
-      "learning_rate": 9.646130653266332e-05,
-      "loss": 6.0326,
-      "step": 4022
-    },
-    {
-      "epoch": 2.098044328552803,
-      "grad_norm": 1.4682538509368896,
-      "learning_rate": 9.64603015075377e-05,
-      "loss": 5.7331,
-      "step": 4023
-    },
-    {
-      "epoch": 2.0985658409387224,
-      "grad_norm": 1.566469669342041,
-      "learning_rate": 9.645929648241206e-05,
-      "loss": 6.1406,
-      "step": 4024
-    },
-    {
-      "epoch": 2.0990873533246415,
-      "grad_norm": 1.5545958280563354,
-      "learning_rate": 9.645829145728644e-05,
-      "loss": 5.5258,
-      "step": 4025
-    },
-    {
-      "epoch": 2.0996088657105605,
-      "grad_norm": 1.524930477142334,
-      "learning_rate": 9.645728643216081e-05,
-      "loss": 5.9815,
-      "step": 4026
-    },
-    {
-      "epoch": 2.10013037809648,
-      "grad_norm": 1.4868464469909668,
-      "learning_rate": 9.645628140703518e-05,
-      "loss": 6.1474,
-      "step": 4027
-    },
-    {
-      "epoch": 2.100651890482399,
-      "grad_norm": 1.7345813512802124,
-      "learning_rate": 9.645527638190956e-05,
-      "loss": 5.9141,
-      "step": 4028
-    },
-    {
-      "epoch": 2.101173402868318,
-      "grad_norm": 1.4309501647949219,
-      "learning_rate": 9.645427135678392e-05,
-      "loss": 6.205,
-      "step": 4029
-    },
-    {
-      "epoch": 2.1016949152542375,
-      "grad_norm": 1.3851560354232788,
-      "learning_rate": 9.64532663316583e-05,
-      "loss": 6.2263,
-      "step": 4030
-    },
-    {
-      "epoch": 2.1022164276401565,
-      "grad_norm": 1.5849758386611938,
-      "learning_rate": 9.645226130653266e-05,
-      "loss": 5.774,
-      "step": 4031
-    },
-    {
-      "epoch": 2.1027379400260755,
-      "grad_norm": 1.5967957973480225,
-      "learning_rate": 9.645125628140704e-05,
-      "loss": 5.7253,
-      "step": 4032
-    },
-    {
-      "epoch": 2.103259452411995,
-      "grad_norm": 1.5128581523895264,
-      "learning_rate": 9.64502512562814e-05,
-      "loss": 5.7215,
-      "step": 4033
-    },
-    {
-      "epoch": 2.103780964797914,
-      "grad_norm": 1.4798904657363892,
-      "learning_rate": 9.644924623115578e-05,
-      "loss": 6.0232,
-      "step": 4034
-    },
-    {
-      "epoch": 2.104302477183833,
-      "grad_norm": 1.59605073928833,
-      "learning_rate": 9.644824120603016e-05,
-      "loss": 5.4373,
-      "step": 4035
-    },
-    {
-      "epoch": 2.1048239895697525,
-      "grad_norm": 1.5852055549621582,
-      "learning_rate": 9.644723618090454e-05,
-      "loss": 5.9024,
-      "step": 4036
-    },
-    {
-      "epoch": 2.1053455019556715,
-      "grad_norm": 1.6397496461868286,
-      "learning_rate": 9.64462311557789e-05,
-      "loss": 5.8841,
-      "step": 4037
-    },
-    {
-      "epoch": 2.1058670143415905,
-      "grad_norm": 1.6135516166687012,
-      "learning_rate": 9.644522613065327e-05,
-      "loss": 5.7623,
-      "step": 4038
-    },
-    {
-      "epoch": 2.10638852672751,
-      "grad_norm": 1.404086947441101,
-      "learning_rate": 9.644422110552765e-05,
-      "loss": 6.382,
-      "step": 4039
-    },
-    {
-      "epoch": 2.106910039113429,
-      "grad_norm": 1.554582118988037,
-      "learning_rate": 9.644321608040201e-05,
-      "loss": 5.9852,
-      "step": 4040
-    },
-    {
-      "epoch": 2.107431551499348,
-      "grad_norm": 1.5648540258407593,
-      "learning_rate": 9.644221105527639e-05,
-      "loss": 6.0648,
-      "step": 4041
-    },
-    {
-      "epoch": 2.1079530638852675,
-      "grad_norm": 1.491792917251587,
-      "learning_rate": 9.644120603015075e-05,
-      "loss": 6.0059,
-      "step": 4042
-    },
-    {
-      "epoch": 2.1084745762711865,
-      "grad_norm": 1.5058050155639648,
-      "learning_rate": 9.644020100502513e-05,
-      "loss": 6.0285,
-      "step": 4043
-    },
-    {
-      "epoch": 2.1089960886571055,
-      "grad_norm": 1.7225233316421509,
-      "learning_rate": 9.64391959798995e-05,
-      "loss": 5.6989,
-      "step": 4044
-    },
-    {
-      "epoch": 2.1095176010430245,
-      "grad_norm": 1.6390562057495117,
-      "learning_rate": 9.643819095477387e-05,
-      "loss": 5.8308,
-      "step": 4045
-    },
-    {
-      "epoch": 2.110039113428944,
-      "grad_norm": 1.5344575643539429,
-      "learning_rate": 9.643718592964825e-05,
-      "loss": 5.9846,
-      "step": 4046
-    },
-    {
-      "epoch": 2.110560625814863,
-      "grad_norm": 1.4064332246780396,
-      "learning_rate": 9.643618090452263e-05,
-      "loss": 6.068,
-      "step": 4047
-    },
-    {
-      "epoch": 2.111082138200782,
-      "grad_norm": 1.471551537513733,
-      "learning_rate": 9.643517587939699e-05,
-      "loss": 5.8043,
-      "step": 4048
-    },
-    {
-      "epoch": 2.1116036505867015,
-      "grad_norm": 1.5230891704559326,
-      "learning_rate": 9.643417085427137e-05,
-      "loss": 5.992,
-      "step": 4049
-    },
-    {
-      "epoch": 2.1121251629726205,
-      "grad_norm": 1.6296758651733398,
-      "learning_rate": 9.643316582914573e-05,
-      "loss": 5.6869,
-      "step": 4050
-    },
-    {
-      "epoch": 2.1126466753585396,
-      "grad_norm": 1.5253201723098755,
-      "learning_rate": 9.64321608040201e-05,
-      "loss": 5.9923,
-      "step": 4051
-    },
-    {
-      "epoch": 2.113168187744459,
-      "grad_norm": 1.433785319328308,
-      "learning_rate": 9.643115577889448e-05,
-      "loss": 5.8441,
-      "step": 4052
-    },
-    {
-      "epoch": 2.113689700130378,
-      "grad_norm": 1.6340131759643555,
-      "learning_rate": 9.643015075376884e-05,
-      "loss": 5.7386,
-      "step": 4053
-    },
-    {
-      "epoch": 2.114211212516297,
-      "grad_norm": 1.5894780158996582,
-      "learning_rate": 9.642914572864322e-05,
-      "loss": 6.0266,
-      "step": 4054
-    },
-    {
-      "epoch": 2.1147327249022165,
-      "grad_norm": 1.4413528442382812,
-      "learning_rate": 9.642814070351758e-05,
-      "loss": 6.1539,
-      "step": 4055
-    },
-    {
-      "epoch": 2.1152542372881356,
-      "grad_norm": 1.5714190006256104,
-      "learning_rate": 9.642713567839196e-05,
-      "loss": 5.7521,
-      "step": 4056
-    },
-    {
-      "epoch": 2.1157757496740546,
-      "grad_norm": 1.5080009698867798,
-      "learning_rate": 9.642613065326634e-05,
-      "loss": 6.0576,
-      "step": 4057
-    },
-    {
-      "epoch": 2.116297262059974,
-      "grad_norm": 1.5810869932174683,
-      "learning_rate": 9.642512562814072e-05,
-      "loss": 5.8681,
-      "step": 4058
-    },
-    {
-      "epoch": 2.116818774445893,
-      "grad_norm": 1.5400056838989258,
-      "learning_rate": 9.642412060301508e-05,
-      "loss": 5.7973,
-      "step": 4059
-    },
-    {
-      "epoch": 2.117340286831812,
-      "grad_norm": 1.499199390411377,
-      "learning_rate": 9.642311557788946e-05,
-      "loss": 5.8028,
-      "step": 4060
-    },
-    {
-      "epoch": 2.1178617992177315,
-      "grad_norm": 1.5160057544708252,
-      "learning_rate": 9.642211055276382e-05,
-      "loss": 5.9625,
-      "step": 4061
-    },
-    {
-      "epoch": 2.1183833116036506,
-      "grad_norm": 1.5690933465957642,
-      "learning_rate": 9.64211055276382e-05,
-      "loss": 5.8115,
-      "step": 4062
-    },
-    {
-      "epoch": 2.1189048239895696,
-      "grad_norm": 1.6263052225112915,
-      "learning_rate": 9.642010050251256e-05,
-      "loss": 5.7829,
-      "step": 4063
-    },
-    {
-      "epoch": 2.119426336375489,
-      "grad_norm": 1.5997531414031982,
-      "learning_rate": 9.641909547738694e-05,
-      "loss": 6.0098,
-      "step": 4064
-    },
-    {
-      "epoch": 2.119947848761408,
-      "grad_norm": 1.6260979175567627,
-      "learning_rate": 9.64180904522613e-05,
-      "loss": 5.7525,
-      "step": 4065
-    },
-    {
-      "epoch": 2.120469361147327,
-      "grad_norm": 1.4114818572998047,
-      "learning_rate": 9.641708542713568e-05,
-      "loss": 6.1736,
-      "step": 4066
-    },
-    {
-      "epoch": 2.1209908735332466,
-      "grad_norm": 1.5938032865524292,
-      "learning_rate": 9.641608040201006e-05,
-      "loss": 5.7228,
-      "step": 4067
-    },
-    {
-      "epoch": 2.1215123859191656,
-      "grad_norm": 1.5444624423980713,
-      "learning_rate": 9.641507537688443e-05,
-      "loss": 6.1097,
-      "step": 4068
-    },
-    {
-      "epoch": 2.1220338983050846,
-      "grad_norm": 1.4992380142211914,
-      "learning_rate": 9.64140703517588e-05,
-      "loss": 5.7934,
-      "step": 4069
-    },
-    {
-      "epoch": 2.122555410691004,
-      "grad_norm": 1.4691975116729736,
-      "learning_rate": 9.641306532663317e-05,
-      "loss": 6.1119,
-      "step": 4070
-    },
-    {
-      "epoch": 2.123076923076923,
-      "grad_norm": 1.4008792638778687,
-      "learning_rate": 9.641206030150755e-05,
-      "loss": 6.3395,
-      "step": 4071
-    },
-    {
-      "epoch": 2.123598435462842,
-      "grad_norm": 1.6199954748153687,
-      "learning_rate": 9.641105527638191e-05,
-      "loss": 6.1734,
-      "step": 4072
-    },
-    {
-      "epoch": 2.1241199478487616,
-      "grad_norm": 1.6932867765426636,
-      "learning_rate": 9.641005025125629e-05,
-      "loss": 5.2122,
-      "step": 4073
-    },
-    {
-      "epoch": 2.1246414602346806,
-      "grad_norm": 1.4030659198760986,
-      "learning_rate": 9.640904522613065e-05,
-      "loss": 5.9789,
-      "step": 4074
-    },
-    {
-      "epoch": 2.1251629726205996,
-      "grad_norm": 1.6059961318969727,
-      "learning_rate": 9.640804020100503e-05,
-      "loss": 6.0803,
-      "step": 4075
-    },
-    {
-      "epoch": 2.125684485006519,
-      "grad_norm": 1.414086937904358,
-      "learning_rate": 9.64070351758794e-05,
-      "loss": 6.0654,
-      "step": 4076
-    },
-    {
-      "epoch": 2.126205997392438,
-      "grad_norm": 1.471876859664917,
-      "learning_rate": 9.640603015075377e-05,
-      "loss": 6.2065,
-      "step": 4077
-    },
-    {
-      "epoch": 2.126727509778357,
-      "grad_norm": 1.6456488370895386,
-      "learning_rate": 9.640502512562815e-05,
-      "loss": 5.7527,
-      "step": 4078
-    },
-    {
-      "epoch": 2.1272490221642766,
-      "grad_norm": 1.4757113456726074,
-      "learning_rate": 9.640402010050251e-05,
-      "loss": 5.732,
-      "step": 4079
-    },
-    {
-      "epoch": 2.1277705345501956,
-      "grad_norm": 1.602286458015442,
-      "learning_rate": 9.640301507537689e-05,
-      "loss": 6.1571,
-      "step": 4080
-    },
-    {
-      "epoch": 2.1282920469361146,
-      "grad_norm": 1.5600515604019165,
-      "learning_rate": 9.640201005025126e-05,
-      "loss": 5.5636,
-      "step": 4081
-    },
-    {
-      "epoch": 2.128813559322034,
-      "grad_norm": 1.5856138467788696,
-      "learning_rate": 9.640100502512563e-05,
-      "loss": 5.9468,
-      "step": 4082
-    },
-    {
-      "epoch": 2.129335071707953,
-      "grad_norm": 1.6452568769454956,
-      "learning_rate": 9.64e-05,
-      "loss": 6.0919,
-      "step": 4083
-    },
-    {
-      "epoch": 2.129856584093872,
-      "grad_norm": 1.4803102016448975,
-      "learning_rate": 9.639899497487438e-05,
-      "loss": 5.9781,
-      "step": 4084
-    },
-    {
-      "epoch": 2.1303780964797916,
-      "grad_norm": 1.5239051580429077,
-      "learning_rate": 9.639798994974874e-05,
-      "loss": 5.4678,
-      "step": 4085
-    },
-    {
-      "epoch": 2.1308996088657106,
-      "grad_norm": 1.4939051866531372,
-      "learning_rate": 9.639698492462312e-05,
-      "loss": 6.0916,
-      "step": 4086
-    },
-    {
-      "epoch": 2.1314211212516296,
-      "grad_norm": 1.4965680837631226,
-      "learning_rate": 9.63959798994975e-05,
-      "loss": 5.982,
-      "step": 4087
-    },
-    {
-      "epoch": 2.131942633637549,
-      "grad_norm": 1.720604658126831,
-      "learning_rate": 9.639497487437187e-05,
-      "loss": 5.6354,
-      "step": 4088
-    },
-    {
-      "epoch": 2.132464146023468,
-      "grad_norm": 1.5248960256576538,
-      "learning_rate": 9.639396984924624e-05,
-      "loss": 6.2344,
-      "step": 4089
-    },
-    {
-      "epoch": 2.132985658409387,
-      "grad_norm": 1.582960844039917,
-      "learning_rate": 9.639296482412062e-05,
-      "loss": 5.7565,
-      "step": 4090
-    },
-    {
-      "epoch": 2.1335071707953066,
-      "grad_norm": 1.611987590789795,
-      "learning_rate": 9.639195979899498e-05,
-      "loss": 5.0651,
-      "step": 4091
-    },
-    {
-      "epoch": 2.1340286831812256,
-      "grad_norm": 1.4891964197158813,
-      "learning_rate": 9.639095477386934e-05,
-      "loss": 5.9972,
-      "step": 4092
-    },
-    {
-      "epoch": 2.1345501955671446,
-      "grad_norm": 1.5517021417617798,
-      "learning_rate": 9.638994974874372e-05,
-      "loss": 6.1375,
-      "step": 4093
-    },
-    {
-      "epoch": 2.135071707953064,
-      "grad_norm": 2.0423271656036377,
-      "learning_rate": 9.638894472361809e-05,
-      "loss": 5.8264,
-      "step": 4094
-    },
-    {
-      "epoch": 2.135593220338983,
-      "grad_norm": 1.6766407489776611,
-      "learning_rate": 9.638793969849246e-05,
-      "loss": 5.5975,
-      "step": 4095
-    },
-    {
-      "epoch": 2.136114732724902,
-      "grad_norm": 1.561306118965149,
-      "learning_rate": 9.638693467336683e-05,
-      "loss": 5.8673,
-      "step": 4096
-    },
-    {
-      "epoch": 2.136636245110821,
-      "grad_norm": 1.6276477575302124,
-      "learning_rate": 9.638592964824121e-05,
-      "loss": 5.8063,
-      "step": 4097
-    },
-    {
-      "epoch": 2.1371577574967406,
-      "grad_norm": 1.4295010566711426,
-      "learning_rate": 9.638492462311558e-05,
-      "loss": 6.0659,
-      "step": 4098
-    },
-    {
-      "epoch": 2.1376792698826597,
-      "grad_norm": 1.5658425092697144,
-      "learning_rate": 9.638391959798996e-05,
-      "loss": 5.5665,
-      "step": 4099
-    },
-    {
-      "epoch": 2.1382007822685787,
-      "grad_norm": 1.420542597770691,
-      "learning_rate": 9.638291457286433e-05,
-      "loss": 6.2492,
-      "step": 4100
-    },
-    {
-      "epoch": 2.138722294654498,
-      "grad_norm": 1.4702461957931519,
-      "learning_rate": 9.63819095477387e-05,
-      "loss": 5.9868,
-      "step": 4101
-    },
-    {
-      "epoch": 2.139243807040417,
-      "grad_norm": 1.813607931137085,
-      "learning_rate": 9.638090452261307e-05,
-      "loss": 5.7942,
-      "step": 4102
-    },
-    {
-      "epoch": 2.139765319426336,
-      "grad_norm": 1.6354705095291138,
-      "learning_rate": 9.637989949748745e-05,
-      "loss": 5.7793,
-      "step": 4103
-    },
-    {
-      "epoch": 2.1402868318122557,
-      "grad_norm": 1.837010145187378,
-      "learning_rate": 9.637889447236181e-05,
-      "loss": 5.8024,
-      "step": 4104
-    },
-    {
-      "epoch": 2.1408083441981747,
-      "grad_norm": 1.3408119678497314,
-      "learning_rate": 9.637788944723618e-05,
-      "loss": 5.8835,
-      "step": 4105
-    },
-    {
-      "epoch": 2.1413298565840937,
-      "grad_norm": 1.3725417852401733,
-      "learning_rate": 9.637688442211055e-05,
-      "loss": 6.3886,
-      "step": 4106
-    },
-    {
-      "epoch": 2.141851368970013,
-      "grad_norm": 1.642970085144043,
-      "learning_rate": 9.637587939698493e-05,
-      "loss": 5.8507,
-      "step": 4107
-    },
-    {
-      "epoch": 2.142372881355932,
-      "grad_norm": 1.8076266050338745,
-      "learning_rate": 9.637487437185931e-05,
-      "loss": 5.1372,
-      "step": 4108
-    },
-    {
-      "epoch": 2.142894393741851,
-      "grad_norm": 1.6173713207244873,
-      "learning_rate": 9.637386934673367e-05,
-      "loss": 5.9407,
-      "step": 4109
-    },
-    {
-      "epoch": 2.1434159061277707,
-      "grad_norm": 1.4310003519058228,
-      "learning_rate": 9.637286432160805e-05,
-      "loss": 5.8436,
-      "step": 4110
-    },
-    {
-      "epoch": 2.1439374185136897,
-      "grad_norm": 1.4432495832443237,
-      "learning_rate": 9.637185929648242e-05,
-      "loss": 5.8959,
-      "step": 4111
-    },
-    {
-      "epoch": 2.1444589308996087,
-      "grad_norm": 1.4538654088974,
-      "learning_rate": 9.637085427135679e-05,
-      "loss": 5.8117,
-      "step": 4112
-    },
-    {
-      "epoch": 2.144980443285528,
-      "grad_norm": 1.5540939569473267,
-      "learning_rate": 9.636984924623116e-05,
-      "loss": 5.7308,
-      "step": 4113
-    },
-    {
-      "epoch": 2.145501955671447,
-      "grad_norm": 1.3418340682983398,
-      "learning_rate": 9.636884422110554e-05,
-      "loss": 5.6499,
-      "step": 4114
-    },
-    {
-      "epoch": 2.146023468057366,
-      "grad_norm": 1.488295555114746,
-      "learning_rate": 9.63678391959799e-05,
-      "loss": 5.4413,
-      "step": 4115
-    },
-    {
-      "epoch": 2.1465449804432857,
-      "grad_norm": 1.5624216794967651,
-      "learning_rate": 9.636683417085428e-05,
-      "loss": 5.6711,
-      "step": 4116
-    },
-    {
-      "epoch": 2.1470664928292047,
-      "grad_norm": 2.7317700386047363,
-      "learning_rate": 9.636582914572864e-05,
-      "loss": 4.9592,
-      "step": 4117
-    },
-    {
-      "epoch": 2.1475880052151237,
-      "grad_norm": 1.4838392734527588,
-      "learning_rate": 9.636482412060302e-05,
-      "loss": 5.8295,
-      "step": 4118
-    },
-    {
-      "epoch": 2.148109517601043,
-      "grad_norm": 1.5660083293914795,
-      "learning_rate": 9.63638190954774e-05,
-      "loss": 5.8831,
-      "step": 4119
-    },
-    {
-      "epoch": 2.148631029986962,
-      "grad_norm": 1.6533023118972778,
-      "learning_rate": 9.636281407035176e-05,
-      "loss": 5.7993,
-      "step": 4120
-    },
-    {
-      "epoch": 2.1491525423728812,
-      "grad_norm": 1.318649411201477,
-      "learning_rate": 9.636180904522614e-05,
-      "loss": 5.2913,
-      "step": 4121
-    },
-    {
-      "epoch": 2.1496740547588007,
-      "grad_norm": 1.4923579692840576,
-      "learning_rate": 9.63608040201005e-05,
-      "loss": 5.6565,
-      "step": 4122
-    },
-    {
-      "epoch": 2.1501955671447197,
-      "grad_norm": 1.674963355064392,
-      "learning_rate": 9.635979899497488e-05,
-      "loss": 5.713,
-      "step": 4123
-    },
-    {
-      "epoch": 2.1507170795306387,
-      "grad_norm": 1.5670394897460938,
-      "learning_rate": 9.635879396984925e-05,
-      "loss": 6.189,
-      "step": 4124
-    },
-    {
-      "epoch": 2.151238591916558,
-      "grad_norm": 1.4565386772155762,
-      "learning_rate": 9.635778894472362e-05,
-      "loss": 5.8786,
-      "step": 4125
-    },
-    {
-      "epoch": 2.151760104302477,
-      "grad_norm": 1.587632179260254,
-      "learning_rate": 9.635678391959799e-05,
-      "loss": 5.4922,
-      "step": 4126
-    },
-    {
-      "epoch": 2.1522816166883962,
-      "grad_norm": 1.6793723106384277,
-      "learning_rate": 9.635577889447237e-05,
-      "loss": 5.9624,
-      "step": 4127
-    },
-    {
-      "epoch": 2.1528031290743157,
-      "grad_norm": 1.5574718713760376,
-      "learning_rate": 9.635477386934674e-05,
-      "loss": 5.9757,
-      "step": 4128
-    },
-    {
-      "epoch": 2.1533246414602347,
-      "grad_norm": 1.498686671257019,
-      "learning_rate": 9.635376884422112e-05,
-      "loss": 6.2249,
-      "step": 4129
-    },
-    {
-      "epoch": 2.1538461538461537,
-      "grad_norm": 1.4474456310272217,
-      "learning_rate": 9.635276381909549e-05,
-      "loss": 6.1448,
-      "step": 4130
-    },
-    {
-      "epoch": 2.154367666232073,
-      "grad_norm": 1.4918137788772583,
-      "learning_rate": 9.635175879396985e-05,
-      "loss": 5.8211,
-      "step": 4131
-    },
-    {
-      "epoch": 2.1548891786179922,
-      "grad_norm": 1.6043626070022583,
-      "learning_rate": 9.635075376884423e-05,
-      "loss": 6.2438,
-      "step": 4132
-    },
-    {
-      "epoch": 2.1554106910039112,
-      "grad_norm": 1.5281509160995483,
-      "learning_rate": 9.634974874371859e-05,
-      "loss": 5.9261,
-      "step": 4133
-    },
-    {
-      "epoch": 2.1559322033898307,
-      "grad_norm": 1.6100736856460571,
-      "learning_rate": 9.634874371859297e-05,
-      "loss": 5.8531,
-      "step": 4134
-    },
-    {
-      "epoch": 2.1564537157757497,
-      "grad_norm": 1.5253342390060425,
-      "learning_rate": 9.634773869346733e-05,
-      "loss": 5.9399,
-      "step": 4135
-    },
-    {
-      "epoch": 2.1569752281616688,
-      "grad_norm": 1.6915301084518433,
-      "learning_rate": 9.634673366834171e-05,
-      "loss": 5.7574,
-      "step": 4136
-    },
-    {
-      "epoch": 2.1574967405475878,
-      "grad_norm": 1.4801241159439087,
-      "learning_rate": 9.634572864321608e-05,
-      "loss": 5.8196,
-      "step": 4137
-    },
-    {
-      "epoch": 2.1580182529335072,
-      "grad_norm": 1.3558902740478516,
-      "learning_rate": 9.634472361809045e-05,
-      "loss": 6.2963,
-      "step": 4138
-    },
-    {
-      "epoch": 2.1585397653194263,
-      "grad_norm": 1.705980658531189,
-      "learning_rate": 9.634371859296483e-05,
-      "loss": 5.8074,
-      "step": 4139
-    },
-    {
-      "epoch": 2.1590612777053453,
-      "grad_norm": 1.4758377075195312,
-      "learning_rate": 9.634271356783921e-05,
-      "loss": 5.9884,
-      "step": 4140
-    },
-    {
-      "epoch": 2.1595827900912647,
-      "grad_norm": 1.509399175643921,
-      "learning_rate": 9.634170854271357e-05,
-      "loss": 5.996,
-      "step": 4141
-    },
-    {
-      "epoch": 2.1601043024771838,
-      "grad_norm": 1.4701310396194458,
-      "learning_rate": 9.634070351758795e-05,
-      "loss": 6.2279,
-      "step": 4142
-    },
-    {
-      "epoch": 2.160625814863103,
-      "grad_norm": 1.4827938079833984,
-      "learning_rate": 9.633969849246232e-05,
-      "loss": 5.6108,
-      "step": 4143
-    },
-    {
-      "epoch": 2.1611473272490223,
-      "grad_norm": 1.4241420030593872,
-      "learning_rate": 9.633869346733668e-05,
-      "loss": 6.2766,
-      "step": 4144
-    },
-    {
-      "epoch": 2.1616688396349413,
-      "grad_norm": 1.5790462493896484,
-      "learning_rate": 9.633768844221106e-05,
-      "loss": 5.6752,
-      "step": 4145
-    },
-    {
-      "epoch": 2.1621903520208603,
-      "grad_norm": 1.5406605005264282,
-      "learning_rate": 9.633668341708542e-05,
-      "loss": 5.8458,
-      "step": 4146
-    },
-    {
-      "epoch": 2.1627118644067798,
-      "grad_norm": 1.4286932945251465,
-      "learning_rate": 9.63356783919598e-05,
-      "loss": 5.7916,
-      "step": 4147
-    },
-    {
-      "epoch": 2.163233376792699,
-      "grad_norm": 1.7935925722122192,
-      "learning_rate": 9.633467336683418e-05,
-      "loss": 5.4178,
-      "step": 4148
-    },
-    {
-      "epoch": 2.163754889178618,
-      "grad_norm": 1.5130184888839722,
-      "learning_rate": 9.633366834170856e-05,
-      "loss": 6.0685,
-      "step": 4149
-    },
-    {
-      "epoch": 2.1642764015645373,
-      "grad_norm": 1.5140042304992676,
-      "learning_rate": 9.633266331658292e-05,
-      "loss": 5.8385,
-      "step": 4150
-    },
-    {
-      "epoch": 2.1647979139504563,
-      "grad_norm": 1.3754451274871826,
-      "learning_rate": 9.63316582914573e-05,
-      "loss": 6.1534,
-      "step": 4151
-    },
-    {
-      "epoch": 2.1653194263363753,
-      "grad_norm": 1.4569108486175537,
-      "learning_rate": 9.633065326633166e-05,
-      "loss": 5.857,
-      "step": 4152
-    },
-    {
-      "epoch": 2.1658409387222948,
-      "grad_norm": 1.430336356163025,
-      "learning_rate": 9.632964824120604e-05,
-      "loss": 6.0599,
-      "step": 4153
-    },
-    {
-      "epoch": 2.166362451108214,
-      "grad_norm": 1.5880956649780273,
-      "learning_rate": 9.63286432160804e-05,
-      "loss": 5.6173,
-      "step": 4154
-    },
-    {
-      "epoch": 2.166883963494133,
-      "grad_norm": 1.4856092929840088,
-      "learning_rate": 9.632763819095478e-05,
-      "loss": 5.8909,
-      "step": 4155
-    },
-    {
-      "epoch": 2.1674054758800523,
-      "grad_norm": 1.473040223121643,
-      "learning_rate": 9.632663316582915e-05,
-      "loss": 5.9461,
-      "step": 4156
-    },
-    {
-      "epoch": 2.1679269882659713,
-      "grad_norm": 1.711106300354004,
-      "learning_rate": 9.632562814070352e-05,
-      "loss": 5.6823,
-      "step": 4157
-    },
-    {
-      "epoch": 2.1684485006518903,
-      "grad_norm": 1.7813776731491089,
-      "learning_rate": 9.632462311557789e-05,
-      "loss": 5.4334,
-      "step": 4158
-    },
-    {
-      "epoch": 2.16897001303781,
-      "grad_norm": 1.412376880645752,
-      "learning_rate": 9.632361809045227e-05,
-      "loss": 5.9774,
-      "step": 4159
-    },
-    {
-      "epoch": 2.169491525423729,
-      "grad_norm": 1.4302338361740112,
-      "learning_rate": 9.632261306532664e-05,
-      "loss": 5.8441,
-      "step": 4160
-    },
-    {
-      "epoch": 2.170013037809648,
-      "grad_norm": 1.491858959197998,
-      "learning_rate": 9.632160804020101e-05,
-      "loss": 6.0293,
-      "step": 4161
-    },
-    {
-      "epoch": 2.1705345501955673,
-      "grad_norm": 1.7326774597167969,
-      "learning_rate": 9.632060301507539e-05,
-      "loss": 5.5771,
-      "step": 4162
-    },
-    {
-      "epoch": 2.1710560625814863,
-      "grad_norm": 1.4889442920684814,
-      "learning_rate": 9.631959798994975e-05,
-      "loss": 5.6667,
-      "step": 4163
-    },
-    {
-      "epoch": 2.1715775749674053,
-      "grad_norm": 1.4325683116912842,
-      "learning_rate": 9.631859296482413e-05,
-      "loss": 6.184,
-      "step": 4164
-    },
-    {
-      "epoch": 2.172099087353325,
-      "grad_norm": 1.5356519222259521,
-      "learning_rate": 9.631758793969849e-05,
-      "loss": 5.3529,
-      "step": 4165
-    },
-    {
-      "epoch": 2.172620599739244,
-      "grad_norm": 1.4379349946975708,
-      "learning_rate": 9.631658291457287e-05,
-      "loss": 6.2292,
-      "step": 4166
-    },
-    {
-      "epoch": 2.173142112125163,
-      "grad_norm": 1.4488633871078491,
-      "learning_rate": 9.631557788944723e-05,
-      "loss": 6.2015,
-      "step": 4167
-    },
-    {
-      "epoch": 2.1736636245110823,
-      "grad_norm": 1.5173065662384033,
-      "learning_rate": 9.631457286432161e-05,
-      "loss": 6.22,
-      "step": 4168
-    },
-    {
-      "epoch": 2.1741851368970013,
-      "grad_norm": 1.4114947319030762,
-      "learning_rate": 9.631356783919599e-05,
-      "loss": 6.0998,
-      "step": 4169
-    },
-    {
-      "epoch": 2.1747066492829203,
-      "grad_norm": 1.4875038862228394,
-      "learning_rate": 9.631256281407037e-05,
-      "loss": 5.9362,
-      "step": 4170
-    },
-    {
-      "epoch": 2.17522816166884,
-      "grad_norm": 1.5078508853912354,
-      "learning_rate": 9.631155778894473e-05,
-      "loss": 5.7843,
-      "step": 4171
-    },
-    {
-      "epoch": 2.175749674054759,
-      "grad_norm": 1.5399279594421387,
-      "learning_rate": 9.63105527638191e-05,
-      "loss": 6.0644,
-      "step": 4172
-    },
-    {
-      "epoch": 2.176271186440678,
-      "grad_norm": 1.6153684854507446,
-      "learning_rate": 9.630954773869347e-05,
-      "loss": 5.9724,
-      "step": 4173
-    },
-    {
-      "epoch": 2.1767926988265973,
-      "grad_norm": 1.5140578746795654,
-      "learning_rate": 9.630854271356784e-05,
-      "loss": 5.2287,
-      "step": 4174
-    },
-    {
-      "epoch": 2.1773142112125163,
-      "grad_norm": 1.5339590311050415,
-      "learning_rate": 9.630753768844222e-05,
-      "loss": 5.7597,
-      "step": 4175
-    },
-    {
-      "epoch": 2.1778357235984354,
-      "grad_norm": 1.5767537355422974,
-      "learning_rate": 9.630653266331658e-05,
-      "loss": 5.9243,
-      "step": 4176
-    },
-    {
-      "epoch": 2.178357235984355,
-      "grad_norm": 1.3952763080596924,
-      "learning_rate": 9.630552763819096e-05,
-      "loss": 6.1219,
-      "step": 4177
-    },
-    {
-      "epoch": 2.178878748370274,
-      "grad_norm": 1.3526992797851562,
-      "learning_rate": 9.630452261306532e-05,
-      "loss": 6.0543,
-      "step": 4178
-    },
-    {
-      "epoch": 2.179400260756193,
-      "grad_norm": 1.4731237888336182,
-      "learning_rate": 9.63035175879397e-05,
-      "loss": 5.0264,
-      "step": 4179
-    },
-    {
-      "epoch": 2.1799217731421123,
-      "grad_norm": 1.5704865455627441,
-      "learning_rate": 9.630251256281408e-05,
-      "loss": 6.0201,
-      "step": 4180
-    },
-    {
-      "epoch": 2.1804432855280313,
-      "grad_norm": 1.5120298862457275,
-      "learning_rate": 9.630150753768846e-05,
-      "loss": 5.8322,
-      "step": 4181
-    },
-    {
-      "epoch": 2.1809647979139504,
-      "grad_norm": 1.4749548435211182,
-      "learning_rate": 9.630050251256282e-05,
-      "loss": 5.7964,
-      "step": 4182
-    },
-    {
-      "epoch": 2.18148631029987,
-      "grad_norm": 1.5404901504516602,
-      "learning_rate": 9.62994974874372e-05,
-      "loss": 6.0657,
-      "step": 4183
-    },
-    {
-      "epoch": 2.182007822685789,
-      "grad_norm": 1.4624485969543457,
-      "learning_rate": 9.629849246231156e-05,
-      "loss": 5.8291,
-      "step": 4184
-    },
-    {
-      "epoch": 2.182529335071708,
-      "grad_norm": 1.398445963859558,
-      "learning_rate": 9.629748743718593e-05,
-      "loss": 5.9797,
-      "step": 4185
-    },
-    {
-      "epoch": 2.1830508474576273,
-      "grad_norm": 1.6162575483322144,
-      "learning_rate": 9.62964824120603e-05,
-      "loss": 5.6593,
-      "step": 4186
-    },
-    {
-      "epoch": 2.1835723598435464,
-      "grad_norm": 1.8853237628936768,
-      "learning_rate": 9.629547738693467e-05,
-      "loss": 5.6815,
-      "step": 4187
-    },
-    {
-      "epoch": 2.1840938722294654,
-      "grad_norm": 1.5584781169891357,
-      "learning_rate": 9.629447236180905e-05,
-      "loss": 5.3786,
-      "step": 4188
-    },
-    {
-      "epoch": 2.184615384615385,
-      "grad_norm": 1.5942392349243164,
-      "learning_rate": 9.629346733668342e-05,
-      "loss": 5.5381,
-      "step": 4189
-    },
-    {
-      "epoch": 2.185136897001304,
-      "grad_norm": 1.6615869998931885,
-      "learning_rate": 9.62924623115578e-05,
-      "loss": 5.9774,
-      "step": 4190
-    },
-    {
-      "epoch": 2.185658409387223,
-      "grad_norm": 1.5335358381271362,
-      "learning_rate": 9.629145728643217e-05,
-      "loss": 5.5768,
-      "step": 4191
-    },
-    {
-      "epoch": 2.186179921773142,
-      "grad_norm": 1.5381380319595337,
-      "learning_rate": 9.629045226130654e-05,
-      "loss": 6.1425,
-      "step": 4192
-    },
-    {
-      "epoch": 2.1867014341590614,
-      "grad_norm": 1.5397645235061646,
-      "learning_rate": 9.628944723618091e-05,
-      "loss": 6.2133,
-      "step": 4193
-    },
-    {
-      "epoch": 2.1872229465449804,
-      "grad_norm": 1.8237745761871338,
-      "learning_rate": 9.628844221105529e-05,
-      "loss": 5.9163,
-      "step": 4194
-    },
-    {
-      "epoch": 2.1877444589308994,
-      "grad_norm": 1.562219262123108,
-      "learning_rate": 9.628743718592965e-05,
-      "loss": 5.9384,
-      "step": 4195
-    },
-    {
-      "epoch": 2.188265971316819,
-      "grad_norm": 1.6798456907272339,
-      "learning_rate": 9.628643216080403e-05,
-      "loss": 5.6887,
-      "step": 4196
-    },
-    {
-      "epoch": 2.188787483702738,
-      "grad_norm": 1.5610941648483276,
-      "learning_rate": 9.62854271356784e-05,
-      "loss": 5.8478,
-      "step": 4197
-    },
-    {
-      "epoch": 2.189308996088657,
-      "grad_norm": 1.6944063901901245,
-      "learning_rate": 9.628442211055276e-05,
-      "loss": 5.792,
-      "step": 4198
-    },
-    {
-      "epoch": 2.1898305084745764,
-      "grad_norm": 1.4909008741378784,
-      "learning_rate": 9.628341708542714e-05,
-      "loss": 6.2984,
-      "step": 4199
-    },
-    {
-      "epoch": 2.1903520208604954,
-      "grad_norm": 1.4918161630630493,
-      "learning_rate": 9.628241206030151e-05,
-      "loss": 6.1658,
-      "step": 4200
-    },
-    {
-      "epoch": 2.1908735332464144,
-      "grad_norm": 1.5655443668365479,
-      "learning_rate": 9.628140703517589e-05,
-      "loss": 5.8382,
-      "step": 4201
-    },
-    {
-      "epoch": 2.191395045632334,
-      "grad_norm": 1.6505444049835205,
-      "learning_rate": 9.628040201005026e-05,
-      "loss": 5.6874,
-      "step": 4202
-    },
-    {
-      "epoch": 2.191916558018253,
-      "grad_norm": 1.5654683113098145,
-      "learning_rate": 9.627939698492463e-05,
-      "loss": 6.027,
-      "step": 4203
-    },
-    {
-      "epoch": 2.192438070404172,
-      "grad_norm": 1.4687938690185547,
-      "learning_rate": 9.6278391959799e-05,
-      "loss": 6.0516,
-      "step": 4204
-    },
-    {
-      "epoch": 2.1929595827900914,
-      "grad_norm": 1.531295657157898,
-      "learning_rate": 9.627738693467338e-05,
-      "loss": 6.0373,
-      "step": 4205
-    },
-    {
-      "epoch": 2.1934810951760104,
-      "grad_norm": 1.4420169591903687,
-      "learning_rate": 9.627638190954774e-05,
-      "loss": 6.207,
-      "step": 4206
-    },
-    {
-      "epoch": 2.1940026075619294,
-      "grad_norm": 1.6116113662719727,
-      "learning_rate": 9.627537688442212e-05,
-      "loss": 5.8803,
-      "step": 4207
-    },
-    {
-      "epoch": 2.194524119947849,
-      "grad_norm": 1.4752699136734009,
-      "learning_rate": 9.627437185929648e-05,
-      "loss": 5.9457,
-      "step": 4208
-    },
-    {
-      "epoch": 2.195045632333768,
-      "grad_norm": 1.4827733039855957,
-      "learning_rate": 9.627336683417086e-05,
-      "loss": 6.2149,
-      "step": 4209
-    },
-    {
-      "epoch": 2.195567144719687,
-      "grad_norm": 1.7995855808258057,
-      "learning_rate": 9.627236180904524e-05,
-      "loss": 5.9222,
-      "step": 4210
-    },
-    {
-      "epoch": 2.1960886571056064,
-      "grad_norm": 1.5062789916992188,
-      "learning_rate": 9.62713567839196e-05,
-      "loss": 6.1665,
-      "step": 4211
-    },
-    {
-      "epoch": 2.1966101694915254,
-      "grad_norm": 1.385849952697754,
-      "learning_rate": 9.627035175879398e-05,
-      "loss": 5.8802,
-      "step": 4212
-    },
-    {
-      "epoch": 2.1971316818774445,
-      "grad_norm": 1.4672178030014038,
-      "learning_rate": 9.626934673366834e-05,
-      "loss": 6.105,
-      "step": 4213
-    },
-    {
-      "epoch": 2.197653194263364,
-      "grad_norm": 1.5439850091934204,
-      "learning_rate": 9.626834170854272e-05,
-      "loss": 5.5583,
-      "step": 4214
-    },
-    {
-      "epoch": 2.198174706649283,
-      "grad_norm": 1.5459290742874146,
-      "learning_rate": 9.626733668341709e-05,
-      "loss": 6.0693,
-      "step": 4215
-    },
-    {
-      "epoch": 2.198696219035202,
-      "grad_norm": 1.5792269706726074,
-      "learning_rate": 9.626633165829146e-05,
-      "loss": 5.6666,
-      "step": 4216
-    },
-    {
-      "epoch": 2.1992177314211214,
-      "grad_norm": 1.4074305295944214,
-      "learning_rate": 9.626532663316583e-05,
-      "loss": 6.1492,
-      "step": 4217
-    },
-    {
-      "epoch": 2.1997392438070404,
-      "grad_norm": 1.566886067390442,
-      "learning_rate": 9.62643216080402e-05,
-      "loss": 6.0673,
-      "step": 4218
-    },
-    {
-      "epoch": 2.2002607561929595,
-      "grad_norm": 1.395219087600708,
-      "learning_rate": 9.626331658291457e-05,
-      "loss": 5.7235,
-      "step": 4219
-    },
-    {
-      "epoch": 2.200782268578879,
-      "grad_norm": 1.5499751567840576,
-      "learning_rate": 9.626231155778895e-05,
-      "loss": 5.6857,
-      "step": 4220
-    },
-    {
-      "epoch": 2.201303780964798,
-      "grad_norm": 1.4168339967727661,
-      "learning_rate": 9.626130653266333e-05,
-      "loss": 6.2124,
-      "step": 4221
-    },
-    {
-      "epoch": 2.201825293350717,
-      "grad_norm": 1.5315395593643188,
-      "learning_rate": 9.62603015075377e-05,
-      "loss": 5.4852,
-      "step": 4222
-    },
-    {
-      "epoch": 2.2023468057366364,
-      "grad_norm": 1.511814832687378,
-      "learning_rate": 9.625929648241207e-05,
-      "loss": 6.0523,
-      "step": 4223
-    },
-    {
-      "epoch": 2.2028683181225555,
-      "grad_norm": 1.5781643390655518,
-      "learning_rate": 9.625829145728643e-05,
-      "loss": 5.4806,
-      "step": 4224
-    },
-    {
-      "epoch": 2.2033898305084745,
-      "grad_norm": 1.5573172569274902,
-      "learning_rate": 9.625728643216081e-05,
-      "loss": 5.9037,
-      "step": 4225
-    },
-    {
-      "epoch": 2.203911342894394,
-      "grad_norm": 1.4940944910049438,
-      "learning_rate": 9.625628140703517e-05,
-      "loss": 5.7336,
-      "step": 4226
-    },
-    {
-      "epoch": 2.204432855280313,
-      "grad_norm": 1.5248242616653442,
-      "learning_rate": 9.625527638190955e-05,
-      "loss": 5.8876,
-      "step": 4227
-    },
-    {
-      "epoch": 2.204954367666232,
-      "grad_norm": 1.4782854318618774,
-      "learning_rate": 9.625427135678392e-05,
-      "loss": 5.993,
-      "step": 4228
-    },
-    {
-      "epoch": 2.205475880052151,
-      "grad_norm": 1.724940538406372,
-      "learning_rate": 9.62532663316583e-05,
-      "loss": 5.3682,
-      "step": 4229
-    },
-    {
-      "epoch": 2.2059973924380705,
-      "grad_norm": 1.558679461479187,
-      "learning_rate": 9.625226130653266e-05,
-      "loss": 5.9558,
-      "step": 4230
-    },
-    {
-      "epoch": 2.2065189048239895,
-      "grad_norm": 1.523055076599121,
-      "learning_rate": 9.625125628140704e-05,
-      "loss": 6.1006,
-      "step": 4231
-    },
-    {
-      "epoch": 2.2070404172099085,
-      "grad_norm": 1.6131079196929932,
-      "learning_rate": 9.625025125628141e-05,
-      "loss": 5.4843,
-      "step": 4232
-    },
-    {
-      "epoch": 2.207561929595828,
-      "grad_norm": 1.5662521123886108,
-      "learning_rate": 9.624924623115579e-05,
-      "loss": 5.4823,
-      "step": 4233
-    },
-    {
-      "epoch": 2.208083441981747,
-      "grad_norm": 1.6755551099777222,
-      "learning_rate": 9.624824120603016e-05,
-      "loss": 6.1826,
-      "step": 4234
-    },
-    {
-      "epoch": 2.208604954367666,
-      "grad_norm": 1.504688024520874,
-      "learning_rate": 9.624723618090453e-05,
-      "loss": 6.2576,
-      "step": 4235
-    },
-    {
-      "epoch": 2.2091264667535855,
-      "grad_norm": 1.5661379098892212,
-      "learning_rate": 9.62462311557789e-05,
-      "loss": 5.929,
-      "step": 4236
-    },
-    {
-      "epoch": 2.2096479791395045,
-      "grad_norm": 1.617279052734375,
-      "learning_rate": 9.624522613065326e-05,
-      "loss": 6.0677,
-      "step": 4237
-    },
-    {
-      "epoch": 2.2101694915254235,
-      "grad_norm": 1.54302978515625,
-      "learning_rate": 9.624422110552764e-05,
-      "loss": 6.288,
-      "step": 4238
-    },
-    {
-      "epoch": 2.210691003911343,
-      "grad_norm": 1.4098293781280518,
-      "learning_rate": 9.6243216080402e-05,
-      "loss": 5.7342,
-      "step": 4239
-    },
-    {
-      "epoch": 2.211212516297262,
-      "grad_norm": 1.533058524131775,
-      "learning_rate": 9.624221105527638e-05,
-      "loss": 6.125,
-      "step": 4240
-    },
-    {
-      "epoch": 2.211734028683181,
-      "grad_norm": 1.825698971748352,
-      "learning_rate": 9.624120603015076e-05,
-      "loss": 5.6156,
-      "step": 4241
-    },
-    {
-      "epoch": 2.2122555410691005,
-      "grad_norm": 1.5709443092346191,
-      "learning_rate": 9.624020100502514e-05,
-      "loss": 5.8662,
-      "step": 4242
-    },
-    {
-      "epoch": 2.2127770534550195,
-      "grad_norm": 1.6004518270492554,
-      "learning_rate": 9.62391959798995e-05,
-      "loss": 5.9142,
-      "step": 4243
-    },
-    {
-      "epoch": 2.2132985658409385,
-      "grad_norm": 1.69973623752594,
-      "learning_rate": 9.623819095477388e-05,
-      "loss": 5.9341,
-      "step": 4244
-    },
-    {
-      "epoch": 2.213820078226858,
-      "grad_norm": 1.8616737127304077,
-      "learning_rate": 9.623718592964824e-05,
-      "loss": 5.1441,
-      "step": 4245
-    },
-    {
-      "epoch": 2.214341590612777,
-      "grad_norm": 1.5581775903701782,
-      "learning_rate": 9.623618090452262e-05,
-      "loss": 5.7673,
-      "step": 4246
-    },
-    {
-      "epoch": 2.214863102998696,
-      "grad_norm": 1.606339454650879,
-      "learning_rate": 9.623517587939699e-05,
-      "loss": 6.0073,
-      "step": 4247
-    },
-    {
-      "epoch": 2.2153846153846155,
-      "grad_norm": 1.6371214389801025,
-      "learning_rate": 9.623417085427136e-05,
-      "loss": 5.9918,
-      "step": 4248
-    },
-    {
-      "epoch": 2.2159061277705345,
-      "grad_norm": 1.6329894065856934,
-      "learning_rate": 9.623316582914573e-05,
-      "loss": 5.7093,
-      "step": 4249
-    },
-    {
-      "epoch": 2.2164276401564535,
-      "grad_norm": 1.369614601135254,
-      "learning_rate": 9.62321608040201e-05,
-      "loss": 6.1167,
-      "step": 4250
-    },
-    {
-      "epoch": 2.216949152542373,
-      "grad_norm": 1.6419516801834106,
-      "learning_rate": 9.623115577889447e-05,
-      "loss": 6.0394,
-      "step": 4251
-    },
-    {
-      "epoch": 2.217470664928292,
-      "grad_norm": 1.577561855316162,
-      "learning_rate": 9.623015075376885e-05,
-      "loss": 5.7569,
-      "step": 4252
-    },
-    {
-      "epoch": 2.217992177314211,
-      "grad_norm": 1.5859085321426392,
-      "learning_rate": 9.622914572864323e-05,
-      "loss": 5.5057,
-      "step": 4253
-    },
-    {
-      "epoch": 2.2185136897001305,
-      "grad_norm": 1.4830831289291382,
-      "learning_rate": 9.622814070351759e-05,
-      "loss": 5.5024,
-      "step": 4254
-    },
-    {
-      "epoch": 2.2190352020860495,
-      "grad_norm": 1.8068671226501465,
-      "learning_rate": 9.622713567839197e-05,
-      "loss": 6.0393,
-      "step": 4255
-    },
-    {
-      "epoch": 2.2195567144719686,
-      "grad_norm": 1.6043729782104492,
-      "learning_rate": 9.622613065326633e-05,
-      "loss": 5.5409,
-      "step": 4256
-    },
-    {
-      "epoch": 2.220078226857888,
-      "grad_norm": 1.5886725187301636,
-      "learning_rate": 9.622512562814071e-05,
-      "loss": 5.6257,
-      "step": 4257
-    },
-    {
-      "epoch": 2.220599739243807,
-      "grad_norm": 1.414002776145935,
-      "learning_rate": 9.622412060301507e-05,
-      "loss": 6.2903,
-      "step": 4258
-    },
-    {
-      "epoch": 2.221121251629726,
-      "grad_norm": 1.5920710563659668,
-      "learning_rate": 9.622311557788945e-05,
-      "loss": 5.7007,
-      "step": 4259
-    },
-    {
-      "epoch": 2.2216427640156455,
-      "grad_norm": 1.6425975561141968,
-      "learning_rate": 9.622211055276382e-05,
-      "loss": 6.0939,
-      "step": 4260
-    },
-    {
-      "epoch": 2.2221642764015646,
-      "grad_norm": 1.7067420482635498,
-      "learning_rate": 9.62211055276382e-05,
-      "loss": 5.2148,
-      "step": 4261
-    },
-    {
-      "epoch": 2.2226857887874836,
-      "grad_norm": 1.5603572130203247,
-      "learning_rate": 9.622010050251257e-05,
-      "loss": 5.9083,
-      "step": 4262
-    },
-    {
-      "epoch": 2.223207301173403,
-      "grad_norm": 1.7151682376861572,
-      "learning_rate": 9.621909547738695e-05,
-      "loss": 5.5316,
-      "step": 4263
-    },
-    {
-      "epoch": 2.223728813559322,
-      "grad_norm": 1.7428044080734253,
-      "learning_rate": 9.621809045226131e-05,
-      "loss": 5.6826,
-      "step": 4264
-    },
-    {
-      "epoch": 2.224250325945241,
-      "grad_norm": 1.6141570806503296,
-      "learning_rate": 9.621708542713568e-05,
-      "loss": 5.972,
-      "step": 4265
-    },
-    {
-      "epoch": 2.2247718383311605,
-      "grad_norm": 1.4812819957733154,
-      "learning_rate": 9.621608040201006e-05,
-      "loss": 6.1273,
-      "step": 4266
-    },
-    {
-      "epoch": 2.2252933507170796,
-      "grad_norm": 1.4849430322647095,
-      "learning_rate": 9.621507537688442e-05,
-      "loss": 5.8146,
-      "step": 4267
-    },
-    {
-      "epoch": 2.2258148631029986,
-      "grad_norm": 1.6830472946166992,
-      "learning_rate": 9.62140703517588e-05,
-      "loss": 6.216,
-      "step": 4268
-    },
-    {
-      "epoch": 2.226336375488918,
-      "grad_norm": 1.4764950275421143,
-      "learning_rate": 9.621306532663316e-05,
-      "loss": 5.8217,
-      "step": 4269
-    },
-    {
-      "epoch": 2.226857887874837,
-      "grad_norm": 1.5452719926834106,
-      "learning_rate": 9.621206030150754e-05,
-      "loss": 5.3149,
-      "step": 4270
-    },
-    {
-      "epoch": 2.227379400260756,
-      "grad_norm": 1.4844051599502563,
-      "learning_rate": 9.62110552763819e-05,
-      "loss": 5.7902,
-      "step": 4271
-    },
-    {
-      "epoch": 2.2279009126466756,
-      "grad_norm": 2.1199533939361572,
-      "learning_rate": 9.621005025125628e-05,
-      "loss": 5.6594,
-      "step": 4272
-    },
-    {
-      "epoch": 2.2284224250325946,
-      "grad_norm": 1.4286423921585083,
-      "learning_rate": 9.620904522613066e-05,
-      "loss": 5.6212,
-      "step": 4273
-    },
-    {
-      "epoch": 2.2289439374185136,
-      "grad_norm": 1.4109450578689575,
-      "learning_rate": 9.620804020100504e-05,
-      "loss": 6.1969,
-      "step": 4274
-    },
-    {
-      "epoch": 2.229465449804433,
-      "grad_norm": 1.4942718744277954,
-      "learning_rate": 9.62070351758794e-05,
-      "loss": 6.1276,
-      "step": 4275
-    },
-    {
-      "epoch": 2.229986962190352,
-      "grad_norm": 1.536807656288147,
-      "learning_rate": 9.620603015075378e-05,
-      "loss": 5.9792,
-      "step": 4276
-    },
-    {
-      "epoch": 2.230508474576271,
-      "grad_norm": 1.5497015714645386,
-      "learning_rate": 9.620502512562815e-05,
-      "loss": 5.8802,
-      "step": 4277
-    },
-    {
-      "epoch": 2.2310299869621906,
-      "grad_norm": 1.4245916604995728,
-      "learning_rate": 9.620402010050251e-05,
-      "loss": 5.9729,
-      "step": 4278
-    },
-    {
-      "epoch": 2.2315514993481096,
-      "grad_norm": 1.392893671989441,
-      "learning_rate": 9.620301507537689e-05,
-      "loss": 6.044,
-      "step": 4279
-    },
-    {
-      "epoch": 2.2320730117340286,
-      "grad_norm": 1.6448756456375122,
-      "learning_rate": 9.620201005025125e-05,
-      "loss": 5.9053,
-      "step": 4280
-    },
-    {
-      "epoch": 2.232594524119948,
-      "grad_norm": 1.5301905870437622,
-      "learning_rate": 9.620100502512563e-05,
-      "loss": 5.609,
-      "step": 4281
-    },
-    {
-      "epoch": 2.233116036505867,
-      "grad_norm": 1.4638302326202393,
-      "learning_rate": 9.620000000000001e-05,
-      "loss": 5.8859,
-      "step": 4282
-    },
-    {
-      "epoch": 2.233637548891786,
-      "grad_norm": 1.5584371089935303,
-      "learning_rate": 9.619899497487439e-05,
-      "loss": 5.6482,
-      "step": 4283
-    },
-    {
-      "epoch": 2.234159061277705,
-      "grad_norm": 1.4065990447998047,
-      "learning_rate": 9.619798994974875e-05,
-      "loss": 6.1765,
-      "step": 4284
-    },
-    {
-      "epoch": 2.2346805736636246,
-      "grad_norm": 1.478596568107605,
-      "learning_rate": 9.619698492462313e-05,
-      "loss": 6.3193,
-      "step": 4285
-    },
-    {
-      "epoch": 2.2352020860495436,
-      "grad_norm": 1.3523467779159546,
-      "learning_rate": 9.619597989949749e-05,
-      "loss": 6.1597,
-      "step": 4286
-    },
-    {
-      "epoch": 2.2357235984354626,
-      "grad_norm": 1.549220323562622,
-      "learning_rate": 9.619497487437187e-05,
-      "loss": 5.9618,
-      "step": 4287
-    },
-    {
-      "epoch": 2.236245110821382,
-      "grad_norm": 1.5357836484909058,
-      "learning_rate": 9.619396984924623e-05,
-      "loss": 5.9269,
-      "step": 4288
-    },
-    {
-      "epoch": 2.236766623207301,
-      "grad_norm": 1.3849581480026245,
-      "learning_rate": 9.619296482412061e-05,
-      "loss": 6.1708,
-      "step": 4289
-    },
-    {
-      "epoch": 2.23728813559322,
-      "grad_norm": 1.5183906555175781,
-      "learning_rate": 9.619195979899498e-05,
-      "loss": 5.8385,
-      "step": 4290
-    },
-    {
-      "epoch": 2.2378096479791396,
-      "grad_norm": 1.7050729990005493,
-      "learning_rate": 9.619095477386934e-05,
-      "loss": 5.7082,
-      "step": 4291
-    },
-    {
-      "epoch": 2.2383311603650586,
-      "grad_norm": 1.8123122453689575,
-      "learning_rate": 9.618994974874372e-05,
-      "loss": 5.8841,
-      "step": 4292
-    },
-    {
-      "epoch": 2.2388526727509777,
-      "grad_norm": 1.5362670421600342,
-      "learning_rate": 9.61889447236181e-05,
-      "loss": 6.1838,
-      "step": 4293
-    },
-    {
-      "epoch": 2.239374185136897,
-      "grad_norm": 1.494167685508728,
-      "learning_rate": 9.618793969849247e-05,
-      "loss": 5.8561,
-      "step": 4294
-    },
-    {
-      "epoch": 2.239895697522816,
-      "grad_norm": 1.6800209283828735,
-      "learning_rate": 9.618693467336684e-05,
-      "loss": 5.9858,
-      "step": 4295
-    },
-    {
-      "epoch": 2.240417209908735,
-      "grad_norm": 1.3043183088302612,
-      "learning_rate": 9.618592964824122e-05,
-      "loss": 6.1153,
-      "step": 4296
-    },
-    {
-      "epoch": 2.2409387222946546,
-      "grad_norm": 1.5613501071929932,
-      "learning_rate": 9.618492462311558e-05,
-      "loss": 5.8189,
-      "step": 4297
-    },
-    {
-      "epoch": 2.2414602346805736,
-      "grad_norm": 1.4424045085906982,
-      "learning_rate": 9.618391959798996e-05,
-      "loss": 6.1198,
-      "step": 4298
-    },
-    {
-      "epoch": 2.2419817470664927,
-      "grad_norm": 1.499439001083374,
-      "learning_rate": 9.618291457286432e-05,
-      "loss": 5.7946,
-      "step": 4299
-    },
-    {
-      "epoch": 2.242503259452412,
-      "grad_norm": 1.6948657035827637,
-      "learning_rate": 9.61819095477387e-05,
-      "loss": 6.1627,
-      "step": 4300
-    },
-    {
-      "epoch": 2.243024771838331,
-      "grad_norm": 1.5387322902679443,
-      "learning_rate": 9.618090452261306e-05,
-      "loss": 6.0627,
-      "step": 4301
-    },
-    {
-      "epoch": 2.24354628422425,
-      "grad_norm": 1.4642891883850098,
-      "learning_rate": 9.617989949748744e-05,
-      "loss": 6.2121,
-      "step": 4302
-    },
-    {
-      "epoch": 2.2440677966101696,
-      "grad_norm": 1.7481859922409058,
-      "learning_rate": 9.617889447236182e-05,
-      "loss": 5.4606,
-      "step": 4303
-    },
-    {
-      "epoch": 2.2445893089960887,
-      "grad_norm": 1.5753322839736938,
-      "learning_rate": 9.617788944723618e-05,
-      "loss": 6.0436,
-      "step": 4304
-    },
-    {
-      "epoch": 2.2451108213820077,
-      "grad_norm": 1.5423458814620972,
-      "learning_rate": 9.617688442211056e-05,
-      "loss": 6.1791,
-      "step": 4305
-    },
-    {
-      "epoch": 2.245632333767927,
-      "grad_norm": 1.5293922424316406,
-      "learning_rate": 9.617587939698493e-05,
-      "loss": 6.0013,
-      "step": 4306
-    },
-    {
-      "epoch": 2.246153846153846,
-      "grad_norm": 1.3917522430419922,
-      "learning_rate": 9.61748743718593e-05,
-      "loss": 5.7268,
-      "step": 4307
-    },
-    {
-      "epoch": 2.246675358539765,
-      "grad_norm": 1.5378646850585938,
-      "learning_rate": 9.617386934673367e-05,
-      "loss": 5.9108,
-      "step": 4308
-    },
-    {
-      "epoch": 2.2471968709256847,
-      "grad_norm": 1.5497835874557495,
-      "learning_rate": 9.617286432160805e-05,
-      "loss": 6.2001,
-      "step": 4309
-    },
-    {
-      "epoch": 2.2477183833116037,
-      "grad_norm": 1.397845983505249,
-      "learning_rate": 9.617185929648241e-05,
-      "loss": 6.1975,
-      "step": 4310
-    },
-    {
-      "epoch": 2.2482398956975227,
-      "grad_norm": 1.3602240085601807,
-      "learning_rate": 9.617085427135679e-05,
-      "loss": 5.8275,
-      "step": 4311
-    },
-    {
-      "epoch": 2.248761408083442,
-      "grad_norm": 1.4670957326889038,
-      "learning_rate": 9.616984924623115e-05,
-      "loss": 5.7987,
-      "step": 4312
-    },
-    {
-      "epoch": 2.249282920469361,
-      "grad_norm": 1.549096703529358,
-      "learning_rate": 9.616884422110553e-05,
-      "loss": 5.7638,
-      "step": 4313
-    },
-    {
-      "epoch": 2.24980443285528,
-      "grad_norm": 1.3964394330978394,
-      "learning_rate": 9.616783919597991e-05,
-      "loss": 6.2001,
-      "step": 4314
-    },
-    {
-      "epoch": 2.2503259452411997,
-      "grad_norm": 1.4519095420837402,
-      "learning_rate": 9.616683417085429e-05,
-      "loss": 5.9808,
-      "step": 4315
-    },
-    {
-      "epoch": 2.2508474576271187,
-      "grad_norm": 1.6788465976715088,
-      "learning_rate": 9.616582914572865e-05,
-      "loss": 5.3371,
-      "step": 4316
-    },
-    {
-      "epoch": 2.2513689700130377,
-      "grad_norm": 1.4285285472869873,
-      "learning_rate": 9.616482412060301e-05,
-      "loss": 5.7738,
-      "step": 4317
-    },
-    {
-      "epoch": 2.2518904823989567,
-      "grad_norm": 1.3309590816497803,
-      "learning_rate": 9.616381909547739e-05,
-      "loss": 6.2485,
-      "step": 4318
-    },
-    {
-      "epoch": 2.252411994784876,
-      "grad_norm": 1.6147722005844116,
-      "learning_rate": 9.616281407035176e-05,
-      "loss": 5.5363,
-      "step": 4319
-    },
-    {
-      "epoch": 2.252933507170795,
-      "grad_norm": 1.5292545557022095,
-      "learning_rate": 9.616180904522613e-05,
-      "loss": 5.9387,
-      "step": 4320
-    },
-    {
-      "epoch": 2.2534550195567142,
-      "grad_norm": 1.735137939453125,
-      "learning_rate": 9.61608040201005e-05,
-      "loss": 5.7481,
-      "step": 4321
-    },
-    {
-      "epoch": 2.2539765319426337,
-      "grad_norm": 1.6055974960327148,
-      "learning_rate": 9.615979899497488e-05,
-      "loss": 5.6314,
-      "step": 4322
-    },
-    {
-      "epoch": 2.2544980443285527,
-      "grad_norm": 1.7140007019042969,
-      "learning_rate": 9.615879396984925e-05,
-      "loss": 5.6946,
-      "step": 4323
-    },
-    {
-      "epoch": 2.2550195567144717,
-      "grad_norm": 1.3877720832824707,
-      "learning_rate": 9.615778894472363e-05,
-      "loss": 6.141,
-      "step": 4324
-    },
-    {
-      "epoch": 2.255541069100391,
-      "grad_norm": 1.5285241603851318,
-      "learning_rate": 9.6156783919598e-05,
-      "loss": 5.9295,
-      "step": 4325
-    },
-    {
-      "epoch": 2.2560625814863102,
-      "grad_norm": 1.7423820495605469,
-      "learning_rate": 9.615577889447237e-05,
-      "loss": 5.8851,
-      "step": 4326
-    },
-    {
-      "epoch": 2.2565840938722292,
-      "grad_norm": 1.3713252544403076,
-      "learning_rate": 9.615477386934674e-05,
-      "loss": 5.5109,
-      "step": 4327
-    },
-    {
-      "epoch": 2.2571056062581487,
-      "grad_norm": 1.3748435974121094,
-      "learning_rate": 9.615376884422112e-05,
-      "loss": 6.1063,
-      "step": 4328
-    },
-    {
-      "epoch": 2.2576271186440677,
-      "grad_norm": 1.3678829669952393,
-      "learning_rate": 9.615276381909548e-05,
-      "loss": 5.9995,
-      "step": 4329
-    },
-    {
-      "epoch": 2.2581486310299868,
-      "grad_norm": 1.4961903095245361,
-      "learning_rate": 9.615175879396984e-05,
-      "loss": 5.7581,
-      "step": 4330
-    },
-    {
-      "epoch": 2.258670143415906,
-      "grad_norm": 1.9454693794250488,
-      "learning_rate": 9.615075376884422e-05,
-      "loss": 4.9097,
-      "step": 4331
-    },
-    {
-      "epoch": 2.2591916558018252,
-      "grad_norm": 1.4550645351409912,
-      "learning_rate": 9.614974874371859e-05,
-      "loss": 6.2048,
-      "step": 4332
-    },
-    {
-      "epoch": 2.2597131681877443,
-      "grad_norm": 1.4141511917114258,
-      "learning_rate": 9.614874371859296e-05,
-      "loss": 5.699,
-      "step": 4333
-    },
-    {
-      "epoch": 2.2602346805736637,
-      "grad_norm": 1.5122029781341553,
-      "learning_rate": 9.614773869346734e-05,
-      "loss": 5.9592,
-      "step": 4334
-    },
-    {
-      "epoch": 2.2607561929595827,
-      "grad_norm": 1.4960200786590576,
-      "learning_rate": 9.614673366834172e-05,
-      "loss": 5.8384,
-      "step": 4335
-    },
-    {
-      "epoch": 2.2612777053455018,
-      "grad_norm": 1.6694279909133911,
-      "learning_rate": 9.614572864321608e-05,
-      "loss": 5.6613,
-      "step": 4336
-    },
-    {
-      "epoch": 2.2617992177314212,
-      "grad_norm": 1.4061635732650757,
-      "learning_rate": 9.614472361809046e-05,
-      "loss": 6.1611,
-      "step": 4337
-    },
-    {
-      "epoch": 2.2623207301173403,
-      "grad_norm": 1.5312711000442505,
-      "learning_rate": 9.614371859296483e-05,
-      "loss": 5.5652,
-      "step": 4338
-    },
-    {
-      "epoch": 2.2628422425032593,
-      "grad_norm": 1.5969191789627075,
-      "learning_rate": 9.61427135678392e-05,
-      "loss": 5.5144,
-      "step": 4339
-    },
-    {
-      "epoch": 2.2633637548891787,
-      "grad_norm": 1.5830482244491577,
-      "learning_rate": 9.614170854271357e-05,
-      "loss": 6.021,
-      "step": 4340
-    },
-    {
-      "epoch": 2.2638852672750978,
-      "grad_norm": 1.474297046661377,
-      "learning_rate": 9.614070351758795e-05,
-      "loss": 5.7039,
-      "step": 4341
-    },
-    {
-      "epoch": 2.2644067796610168,
-      "grad_norm": 1.695778250694275,
-      "learning_rate": 9.613969849246231e-05,
-      "loss": 5.4556,
-      "step": 4342
-    },
-    {
-      "epoch": 2.2649282920469362,
-      "grad_norm": 1.3813631534576416,
-      "learning_rate": 9.613869346733669e-05,
-      "loss": 6.0123,
-      "step": 4343
-    },
-    {
-      "epoch": 2.2654498044328553,
-      "grad_norm": 1.7414000034332275,
-      "learning_rate": 9.613768844221107e-05,
-      "loss": 5.8064,
-      "step": 4344
-    },
-    {
-      "epoch": 2.2659713168187743,
-      "grad_norm": 1.7448735237121582,
-      "learning_rate": 9.613668341708543e-05,
-      "loss": 5.5019,
-      "step": 4345
-    },
-    {
-      "epoch": 2.2664928292046937,
-      "grad_norm": 1.7771391868591309,
-      "learning_rate": 9.613567839195981e-05,
-      "loss": 5.6437,
-      "step": 4346
-    },
-    {
-      "epoch": 2.2670143415906128,
-      "grad_norm": 1.4635871648788452,
-      "learning_rate": 9.613467336683417e-05,
-      "loss": 5.8773,
-      "step": 4347
-    },
-    {
-      "epoch": 2.267535853976532,
-      "grad_norm": 1.4158730506896973,
-      "learning_rate": 9.613366834170855e-05,
-      "loss": 6.1698,
-      "step": 4348
-    },
-    {
-      "epoch": 2.2680573663624513,
-      "grad_norm": 1.696439504623413,
-      "learning_rate": 9.613266331658292e-05,
-      "loss": 5.4449,
-      "step": 4349
-    },
-    {
-      "epoch": 2.2685788787483703,
-      "grad_norm": 1.5143623352050781,
-      "learning_rate": 9.613165829145729e-05,
-      "loss": 5.5649,
-      "step": 4350
-    },
-    {
-      "epoch": 2.2691003911342893,
-      "grad_norm": 1.6823991537094116,
-      "learning_rate": 9.613065326633166e-05,
-      "loss": 5.7486,
-      "step": 4351
-    },
-    {
-      "epoch": 2.2696219035202088,
-      "grad_norm": 1.483682632446289,
-      "learning_rate": 9.612964824120604e-05,
-      "loss": 5.7702,
-      "step": 4352
-    },
-    {
-      "epoch": 2.270143415906128,
-      "grad_norm": 1.4481534957885742,
-      "learning_rate": 9.61286432160804e-05,
-      "loss": 6.1517,
-      "step": 4353
-    },
-    {
-      "epoch": 2.270664928292047,
-      "grad_norm": 1.407548189163208,
-      "learning_rate": 9.612763819095478e-05,
-      "loss": 5.9806,
-      "step": 4354
-    },
-    {
-      "epoch": 2.2711864406779663,
-      "grad_norm": 1.3390120267868042,
-      "learning_rate": 9.612663316582916e-05,
-      "loss": 5.8685,
-      "step": 4355
-    },
-    {
-      "epoch": 2.2717079530638853,
-      "grad_norm": 1.4609589576721191,
-      "learning_rate": 9.612562814070353e-05,
-      "loss": 5.9152,
-      "step": 4356
-    },
-    {
-      "epoch": 2.2722294654498043,
-      "grad_norm": 1.6170445680618286,
-      "learning_rate": 9.61246231155779e-05,
-      "loss": 5.884,
-      "step": 4357
-    },
-    {
-      "epoch": 2.2727509778357238,
-      "grad_norm": 1.502043604850769,
-      "learning_rate": 9.612361809045226e-05,
-      "loss": 5.8997,
-      "step": 4358
-    },
-    {
-      "epoch": 2.273272490221643,
-      "grad_norm": 1.4751697778701782,
-      "learning_rate": 9.612261306532664e-05,
-      "loss": 5.9774,
-      "step": 4359
-    },
-    {
-      "epoch": 2.273794002607562,
-      "grad_norm": 1.401495337486267,
-      "learning_rate": 9.6121608040201e-05,
-      "loss": 6.1311,
-      "step": 4360
-    },
-    {
-      "epoch": 2.2743155149934813,
-      "grad_norm": 1.4665179252624512,
-      "learning_rate": 9.612060301507538e-05,
-      "loss": 5.8177,
-      "step": 4361
-    },
-    {
-      "epoch": 2.2748370273794003,
-      "grad_norm": 1.6854355335235596,
-      "learning_rate": 9.611959798994975e-05,
-      "loss": 5.8978,
-      "step": 4362
-    },
-    {
-      "epoch": 2.2753585397653193,
-      "grad_norm": 1.5078668594360352,
-      "learning_rate": 9.611859296482412e-05,
-      "loss": 6.052,
-      "step": 4363
-    },
-    {
-      "epoch": 2.275880052151239,
-      "grad_norm": 1.5174840688705444,
-      "learning_rate": 9.61175879396985e-05,
-      "loss": 5.8474,
-      "step": 4364
-    },
-    {
-      "epoch": 2.276401564537158,
-      "grad_norm": 1.5173968076705933,
-      "learning_rate": 9.611658291457288e-05,
-      "loss": 5.7233,
-      "step": 4365
-    },
-    {
-      "epoch": 2.276923076923077,
-      "grad_norm": 1.9840943813323975,
-      "learning_rate": 9.611557788944724e-05,
-      "loss": 5.6917,
-      "step": 4366
-    },
-    {
-      "epoch": 2.2774445893089963,
-      "grad_norm": 1.3654333353042603,
-      "learning_rate": 9.611457286432162e-05,
-      "loss": 6.3043,
-      "step": 4367
-    },
-    {
-      "epoch": 2.2779661016949153,
-      "grad_norm": 1.4341760873794556,
-      "learning_rate": 9.611356783919599e-05,
-      "loss": 6.2549,
-      "step": 4368
-    },
-    {
-      "epoch": 2.2784876140808343,
-      "grad_norm": 1.5898112058639526,
-      "learning_rate": 9.611256281407036e-05,
-      "loss": 5.7252,
-      "step": 4369
-    },
-    {
-      "epoch": 2.279009126466754,
-      "grad_norm": 1.5204517841339111,
-      "learning_rate": 9.611155778894473e-05,
-      "loss": 5.709,
-      "step": 4370
-    },
-    {
-      "epoch": 2.279530638852673,
-      "grad_norm": 1.7262276411056519,
-      "learning_rate": 9.611055276381909e-05,
-      "loss": 6.2853,
-      "step": 4371
-    },
-    {
-      "epoch": 2.280052151238592,
-      "grad_norm": 1.6239138841629028,
-      "learning_rate": 9.610954773869347e-05,
-      "loss": 6.1104,
-      "step": 4372
-    },
-    {
-      "epoch": 2.2805736636245113,
-      "grad_norm": 1.4525542259216309,
-      "learning_rate": 9.610854271356783e-05,
-      "loss": 5.9999,
-      "step": 4373
-    },
-    {
-      "epoch": 2.2810951760104303,
-      "grad_norm": 1.5398870706558228,
-      "learning_rate": 9.610753768844221e-05,
-      "loss": 5.8398,
-      "step": 4374
-    },
-    {
-      "epoch": 2.2816166883963493,
-      "grad_norm": 1.5358282327651978,
-      "learning_rate": 9.610653266331659e-05,
-      "loss": 5.3313,
-      "step": 4375
-    },
-    {
-      "epoch": 2.282138200782269,
-      "grad_norm": 1.573818325996399,
-      "learning_rate": 9.610552763819097e-05,
-      "loss": 5.7179,
-      "step": 4376
-    },
-    {
-      "epoch": 2.282659713168188,
-      "grad_norm": 1.6951656341552734,
-      "learning_rate": 9.610452261306533e-05,
-      "loss": 5.0575,
-      "step": 4377
-    },
-    {
-      "epoch": 2.283181225554107,
-      "grad_norm": 1.4231294393539429,
-      "learning_rate": 9.610351758793971e-05,
-      "loss": 5.9981,
-      "step": 4378
-    },
-    {
-      "epoch": 2.2837027379400263,
-      "grad_norm": 1.4839390516281128,
-      "learning_rate": 9.610251256281407e-05,
-      "loss": 5.2833,
-      "step": 4379
-    },
-    {
-      "epoch": 2.2842242503259453,
-      "grad_norm": 1.5798044204711914,
-      "learning_rate": 9.610150753768845e-05,
-      "loss": 5.6201,
-      "step": 4380
-    },
-    {
-      "epoch": 2.2847457627118644,
-      "grad_norm": 1.5019915103912354,
-      "learning_rate": 9.610050251256282e-05,
-      "loss": 5.5449,
-      "step": 4381
-    },
-    {
-      "epoch": 2.2852672750977834,
-      "grad_norm": 1.5175471305847168,
-      "learning_rate": 9.60994974874372e-05,
-      "loss": 6.1084,
-      "step": 4382
-    },
-    {
-      "epoch": 2.285788787483703,
-      "grad_norm": 1.4830224514007568,
-      "learning_rate": 9.609849246231156e-05,
-      "loss": 6.0644,
-      "step": 4383
-    },
-    {
-      "epoch": 2.286310299869622,
-      "grad_norm": 1.5157499313354492,
-      "learning_rate": 9.609748743718592e-05,
-      "loss": 6.0146,
-      "step": 4384
-    },
-    {
-      "epoch": 2.286831812255541,
-      "grad_norm": 1.4522380828857422,
-      "learning_rate": 9.60964824120603e-05,
-      "loss": 6.2253,
-      "step": 4385
-    },
-    {
-      "epoch": 2.2873533246414604,
-      "grad_norm": 1.4442965984344482,
-      "learning_rate": 9.609547738693468e-05,
-      "loss": 5.7933,
-      "step": 4386
-    },
-    {
-      "epoch": 2.2878748370273794,
-      "grad_norm": 1.7577446699142456,
-      "learning_rate": 9.609447236180906e-05,
-      "loss": 5.9946,
-      "step": 4387
-    },
-    {
-      "epoch": 2.2883963494132984,
-      "grad_norm": 1.5554211139678955,
-      "learning_rate": 9.609346733668342e-05,
-      "loss": 5.7703,
-      "step": 4388
-    },
-    {
-      "epoch": 2.288917861799218,
-      "grad_norm": 1.5655378103256226,
-      "learning_rate": 9.60924623115578e-05,
-      "loss": 5.6236,
-      "step": 4389
-    },
-    {
-      "epoch": 2.289439374185137,
-      "grad_norm": 1.6425355672836304,
-      "learning_rate": 9.609145728643216e-05,
-      "loss": 5.6981,
-      "step": 4390
-    },
-    {
-      "epoch": 2.289960886571056,
-      "grad_norm": 2.027571201324463,
-      "learning_rate": 9.609045226130654e-05,
-      "loss": 4.8238,
-      "step": 4391
-    },
-    {
-      "epoch": 2.2904823989569754,
-      "grad_norm": 1.6873457431793213,
-      "learning_rate": 9.60894472361809e-05,
-      "loss": 5.9091,
-      "step": 4392
-    },
-    {
-      "epoch": 2.2910039113428944,
-      "grad_norm": 1.572218894958496,
-      "learning_rate": 9.608844221105528e-05,
-      "loss": 5.819,
-      "step": 4393
-    },
-    {
-      "epoch": 2.2915254237288134,
-      "grad_norm": 1.587289571762085,
-      "learning_rate": 9.608743718592965e-05,
-      "loss": 5.7624,
-      "step": 4394
-    },
-    {
-      "epoch": 2.292046936114733,
-      "grad_norm": 1.6663979291915894,
-      "learning_rate": 9.608643216080402e-05,
-      "loss": 5.7209,
-      "step": 4395
-    },
-    {
-      "epoch": 2.292568448500652,
-      "grad_norm": 1.6945639848709106,
-      "learning_rate": 9.60854271356784e-05,
-      "loss": 5.5485,
-      "step": 4396
-    },
-    {
-      "epoch": 2.293089960886571,
-      "grad_norm": 1.6336361169815063,
-      "learning_rate": 9.608442211055277e-05,
-      "loss": 6.0035,
-      "step": 4397
-    },
-    {
-      "epoch": 2.2936114732724904,
-      "grad_norm": 1.7970013618469238,
-      "learning_rate": 9.608341708542714e-05,
-      "loss": 6.0355,
-      "step": 4398
-    },
-    {
-      "epoch": 2.2941329856584094,
-      "grad_norm": 1.6688491106033325,
-      "learning_rate": 9.608241206030151e-05,
-      "loss": 5.8533,
-      "step": 4399
-    },
-    {
-      "epoch": 2.2946544980443284,
-      "grad_norm": 2.2948222160339355,
-      "learning_rate": 9.608140703517589e-05,
-      "loss": 5.6316,
-      "step": 4400
-    },
-    {
-      "epoch": 2.295176010430248,
-      "grad_norm": 1.535705804824829,
-      "learning_rate": 9.608040201005025e-05,
-      "loss": 6.1406,
-      "step": 4401
-    },
-    {
-      "epoch": 2.295697522816167,
-      "grad_norm": 1.7565983533859253,
-      "learning_rate": 9.607939698492463e-05,
-      "loss": 5.6555,
-      "step": 4402
-    },
-    {
-      "epoch": 2.296219035202086,
-      "grad_norm": 1.6478540897369385,
-      "learning_rate": 9.607839195979899e-05,
-      "loss": 5.8054,
-      "step": 4403
-    },
-    {
-      "epoch": 2.2967405475880054,
-      "grad_norm": 1.502612829208374,
-      "learning_rate": 9.607738693467337e-05,
-      "loss": 5.9685,
-      "step": 4404
-    },
-    {
-      "epoch": 2.2972620599739244,
-      "grad_norm": 1.5802887678146362,
-      "learning_rate": 9.607638190954773e-05,
-      "loss": 6.1117,
-      "step": 4405
-    },
-    {
-      "epoch": 2.2977835723598434,
-      "grad_norm": 1.6156667470932007,
-      "learning_rate": 9.607537688442211e-05,
-      "loss": 5.6882,
-      "step": 4406
-    },
-    {
-      "epoch": 2.298305084745763,
-      "grad_norm": 1.5162800550460815,
-      "learning_rate": 9.607437185929649e-05,
-      "loss": 5.8482,
-      "step": 4407
-    },
-    {
-      "epoch": 2.298826597131682,
-      "grad_norm": 1.5501666069030762,
-      "learning_rate": 9.607336683417087e-05,
-      "loss": 6.0215,
-      "step": 4408
-    },
-    {
-      "epoch": 2.299348109517601,
-      "grad_norm": 1.4644646644592285,
-      "learning_rate": 9.607236180904523e-05,
-      "loss": 6.3952,
-      "step": 4409
-    },
-    {
-      "epoch": 2.2998696219035204,
-      "grad_norm": 1.480393886566162,
-      "learning_rate": 9.60713567839196e-05,
-      "loss": 5.9113,
-      "step": 4410
-    },
-    {
-      "epoch": 2.3003911342894394,
-      "grad_norm": 1.51812744140625,
-      "learning_rate": 9.607035175879397e-05,
-      "loss": 5.7259,
-      "step": 4411
-    },
-    {
-      "epoch": 2.3009126466753584,
-      "grad_norm": 1.4744949340820312,
-      "learning_rate": 9.606934673366834e-05,
-      "loss": 6.2802,
-      "step": 4412
-    },
-    {
-      "epoch": 2.3014341590612775,
-      "grad_norm": 1.4951553344726562,
-      "learning_rate": 9.606834170854272e-05,
-      "loss": 5.7906,
-      "step": 4413
-    },
-    {
-      "epoch": 2.301955671447197,
-      "grad_norm": 1.4387564659118652,
-      "learning_rate": 9.606733668341708e-05,
-      "loss": 6.2377,
-      "step": 4414
-    },
-    {
-      "epoch": 2.302477183833116,
-      "grad_norm": 1.440932273864746,
-      "learning_rate": 9.606633165829146e-05,
-      "loss": 5.518,
-      "step": 4415
-    },
-    {
-      "epoch": 2.302998696219035,
-      "grad_norm": 1.5527325868606567,
-      "learning_rate": 9.606532663316584e-05,
-      "loss": 5.9647,
-      "step": 4416
-    },
-    {
-      "epoch": 2.3035202086049544,
-      "grad_norm": 1.4493411779403687,
-      "learning_rate": 9.606432160804021e-05,
-      "loss": 6.1405,
-      "step": 4417
-    },
-    {
-      "epoch": 2.3040417209908735,
-      "grad_norm": 1.4464961290359497,
-      "learning_rate": 9.606331658291458e-05,
-      "loss": 5.8679,
-      "step": 4418
-    },
-    {
-      "epoch": 2.3045632333767925,
-      "grad_norm": 1.521118402481079,
-      "learning_rate": 9.606231155778896e-05,
-      "loss": 5.7272,
-      "step": 4419
-    },
-    {
-      "epoch": 2.305084745762712,
-      "grad_norm": 1.5399384498596191,
-      "learning_rate": 9.606130653266332e-05,
-      "loss": 6.2709,
-      "step": 4420
-    },
-    {
-      "epoch": 2.305606258148631,
-      "grad_norm": 1.576090693473816,
-      "learning_rate": 9.60603015075377e-05,
-      "loss": 5.8097,
-      "step": 4421
-    },
-    {
-      "epoch": 2.30612777053455,
-      "grad_norm": 1.4556787014007568,
-      "learning_rate": 9.605929648241206e-05,
-      "loss": 6.0951,
-      "step": 4422
-    },
-    {
-      "epoch": 2.3066492829204694,
-      "grad_norm": 1.5314087867736816,
-      "learning_rate": 9.605829145728644e-05,
-      "loss": 5.9041,
-      "step": 4423
-    },
-    {
-      "epoch": 2.3071707953063885,
-      "grad_norm": 1.5983115434646606,
-      "learning_rate": 9.60572864321608e-05,
-      "loss": 5.8668,
-      "step": 4424
-    },
-    {
-      "epoch": 2.3076923076923075,
-      "grad_norm": 1.3969697952270508,
-      "learning_rate": 9.605628140703517e-05,
-      "loss": 5.8842,
-      "step": 4425
-    },
-    {
-      "epoch": 2.308213820078227,
-      "grad_norm": 1.4701638221740723,
-      "learning_rate": 9.605527638190955e-05,
-      "loss": 5.8686,
-      "step": 4426
-    },
-    {
-      "epoch": 2.308735332464146,
-      "grad_norm": 1.4844608306884766,
-      "learning_rate": 9.605427135678392e-05,
-      "loss": 6.0574,
-      "step": 4427
-    },
-    {
-      "epoch": 2.309256844850065,
-      "grad_norm": 1.4334423542022705,
-      "learning_rate": 9.60532663316583e-05,
-      "loss": 6.0693,
-      "step": 4428
-    },
-    {
-      "epoch": 2.3097783572359845,
-      "grad_norm": 1.4853509664535522,
-      "learning_rate": 9.605226130653267e-05,
-      "loss": 5.8437,
-      "step": 4429
-    },
-    {
-      "epoch": 2.3102998696219035,
-      "grad_norm": 1.6010940074920654,
-      "learning_rate": 9.605125628140704e-05,
-      "loss": 5.6738,
-      "step": 4430
-    },
-    {
-      "epoch": 2.3108213820078225,
-      "grad_norm": 1.3954604864120483,
-      "learning_rate": 9.605025125628141e-05,
-      "loss": 6.2042,
-      "step": 4431
-    },
-    {
-      "epoch": 2.311342894393742,
-      "grad_norm": 1.596882939338684,
-      "learning_rate": 9.604924623115579e-05,
-      "loss": 5.6366,
-      "step": 4432
-    },
-    {
-      "epoch": 2.311864406779661,
-      "grad_norm": 1.5305641889572144,
-      "learning_rate": 9.604824120603015e-05,
-      "loss": 5.9767,
-      "step": 4433
-    },
-    {
-      "epoch": 2.31238591916558,
-      "grad_norm": 1.6144206523895264,
-      "learning_rate": 9.604723618090453e-05,
-      "loss": 5.8965,
-      "step": 4434
-    },
-    {
-      "epoch": 2.3129074315514995,
-      "grad_norm": 1.5516972541809082,
-      "learning_rate": 9.60462311557789e-05,
-      "loss": 5.8671,
-      "step": 4435
-    },
-    {
-      "epoch": 2.3134289439374185,
-      "grad_norm": 1.5360769033432007,
-      "learning_rate": 9.604522613065327e-05,
-      "loss": 5.6933,
-      "step": 4436
-    },
-    {
-      "epoch": 2.3139504563233375,
-      "grad_norm": 1.4740028381347656,
-      "learning_rate": 9.604422110552765e-05,
-      "loss": 6.2965,
-      "step": 4437
-    },
-    {
-      "epoch": 2.314471968709257,
-      "grad_norm": 1.4330319166183472,
-      "learning_rate": 9.604321608040201e-05,
-      "loss": 5.9253,
-      "step": 4438
-    },
-    {
-      "epoch": 2.314993481095176,
-      "grad_norm": 1.765086054801941,
-      "learning_rate": 9.604221105527639e-05,
-      "loss": 5.5736,
-      "step": 4439
-    },
-    {
-      "epoch": 2.315514993481095,
-      "grad_norm": 1.4596048593521118,
-      "learning_rate": 9.604120603015076e-05,
-      "loss": 5.88,
-      "step": 4440
-    },
-    {
-      "epoch": 2.3160365058670145,
-      "grad_norm": 1.415076494216919,
-      "learning_rate": 9.604020100502513e-05,
-      "loss": 6.061,
-      "step": 4441
-    },
-    {
-      "epoch": 2.3165580182529335,
-      "grad_norm": 1.5891660451889038,
-      "learning_rate": 9.60391959798995e-05,
-      "loss": 6.1773,
-      "step": 4442
-    },
-    {
-      "epoch": 2.3170795306388525,
-      "grad_norm": 1.5754941701889038,
-      "learning_rate": 9.603819095477388e-05,
-      "loss": 5.922,
-      "step": 4443
-    },
-    {
-      "epoch": 2.317601043024772,
-      "grad_norm": 1.4644241333007812,
-      "learning_rate": 9.603718592964824e-05,
-      "loss": 5.897,
-      "step": 4444
-    },
-    {
-      "epoch": 2.318122555410691,
-      "grad_norm": 1.592297077178955,
-      "learning_rate": 9.603618090452262e-05,
-      "loss": 5.7542,
-      "step": 4445
-    },
-    {
-      "epoch": 2.31864406779661,
-      "grad_norm": 1.6834914684295654,
-      "learning_rate": 9.603517587939698e-05,
-      "loss": 5.8045,
-      "step": 4446
-    },
-    {
-      "epoch": 2.3191655801825295,
-      "grad_norm": 1.720765233039856,
-      "learning_rate": 9.603417085427136e-05,
-      "loss": 5.8312,
-      "step": 4447
-    },
-    {
-      "epoch": 2.3196870925684485,
-      "grad_norm": 1.7956619262695312,
-      "learning_rate": 9.603316582914574e-05,
-      "loss": 5.4393,
-      "step": 4448
-    },
-    {
-      "epoch": 2.3202086049543675,
-      "grad_norm": 1.8300572633743286,
-      "learning_rate": 9.603216080402012e-05,
-      "loss": 5.7789,
-      "step": 4449
-    },
-    {
-      "epoch": 2.320730117340287,
-      "grad_norm": 1.712662935256958,
-      "learning_rate": 9.603115577889448e-05,
-      "loss": 5.8468,
-      "step": 4450
-    },
-    {
-      "epoch": 2.321251629726206,
-      "grad_norm": 1.781042218208313,
-      "learning_rate": 9.603015075376884e-05,
-      "loss": 5.2116,
-      "step": 4451
-    },
-    {
-      "epoch": 2.321773142112125,
-      "grad_norm": 1.5941672325134277,
-      "learning_rate": 9.602914572864322e-05,
-      "loss": 5.477,
-      "step": 4452
-    },
-    {
-      "epoch": 2.3222946544980445,
-      "grad_norm": 2.090144395828247,
-      "learning_rate": 9.602814070351759e-05,
-      "loss": 4.8318,
-      "step": 4453
-    },
-    {
-      "epoch": 2.3228161668839635,
-      "grad_norm": 1.7761808633804321,
-      "learning_rate": 9.602713567839196e-05,
-      "loss": 5.8825,
-      "step": 4454
-    },
-    {
-      "epoch": 2.3233376792698825,
-      "grad_norm": 1.5097930431365967,
-      "learning_rate": 9.602613065326633e-05,
-      "loss": 5.6314,
-      "step": 4455
-    },
-    {
-      "epoch": 2.323859191655802,
-      "grad_norm": 1.5698878765106201,
-      "learning_rate": 9.60251256281407e-05,
-      "loss": 5.8268,
-      "step": 4456
-    },
-    {
-      "epoch": 2.324380704041721,
-      "grad_norm": 1.5804098844528198,
-      "learning_rate": 9.602412060301508e-05,
-      "loss": 6.1351,
-      "step": 4457
-    },
-    {
-      "epoch": 2.32490221642764,
-      "grad_norm": 2.0465869903564453,
-      "learning_rate": 9.602311557788946e-05,
-      "loss": 5.3564,
-      "step": 4458
-    },
-    {
-      "epoch": 2.3254237288135595,
-      "grad_norm": 1.5301759243011475,
-      "learning_rate": 9.602211055276383e-05,
-      "loss": 5.8345,
-      "step": 4459
-    },
-    {
-      "epoch": 2.3259452411994785,
-      "grad_norm": 1.6353880167007446,
-      "learning_rate": 9.60211055276382e-05,
-      "loss": 5.7697,
-      "step": 4460
-    },
-    {
-      "epoch": 2.3264667535853976,
-      "grad_norm": 1.5556309223175049,
-      "learning_rate": 9.602010050251257e-05,
-      "loss": 5.9247,
-      "step": 4461
-    },
-    {
-      "epoch": 2.326988265971317,
-      "grad_norm": 1.3763997554779053,
-      "learning_rate": 9.601909547738695e-05,
-      "loss": 6.1411,
-      "step": 4462
-    },
-    {
-      "epoch": 2.327509778357236,
-      "grad_norm": 1.5687496662139893,
-      "learning_rate": 9.601809045226131e-05,
-      "loss": 5.6664,
-      "step": 4463
-    },
-    {
-      "epoch": 2.328031290743155,
-      "grad_norm": 1.8230608701705933,
-      "learning_rate": 9.601708542713567e-05,
-      "loss": 5.7319,
-      "step": 4464
-    },
-    {
-      "epoch": 2.3285528031290745,
-      "grad_norm": 1.5991131067276,
-      "learning_rate": 9.601608040201005e-05,
-      "loss": 5.59,
-      "step": 4465
-    },
-    {
-      "epoch": 2.3290743155149936,
-      "grad_norm": 1.5195105075836182,
-      "learning_rate": 9.601507537688442e-05,
-      "loss": 5.876,
-      "step": 4466
-    },
-    {
-      "epoch": 2.3295958279009126,
-      "grad_norm": 1.5950733423233032,
-      "learning_rate": 9.60140703517588e-05,
-      "loss": 6.0194,
-      "step": 4467
-    },
-    {
-      "epoch": 2.330117340286832,
-      "grad_norm": 1.636050820350647,
-      "learning_rate": 9.601306532663317e-05,
-      "loss": 5.7971,
-      "step": 4468
-    },
-    {
-      "epoch": 2.330638852672751,
-      "grad_norm": 1.6218888759613037,
-      "learning_rate": 9.601206030150755e-05,
-      "loss": 6.1563,
-      "step": 4469
-    },
-    {
-      "epoch": 2.33116036505867,
-      "grad_norm": 1.6371983289718628,
-      "learning_rate": 9.601105527638191e-05,
-      "loss": 6.0322,
-      "step": 4470
-    },
-    {
-      "epoch": 2.3316818774445895,
-      "grad_norm": 1.8263182640075684,
-      "learning_rate": 9.601005025125629e-05,
-      "loss": 5.9884,
-      "step": 4471
-    },
-    {
-      "epoch": 2.3322033898305086,
-      "grad_norm": 1.5905038118362427,
-      "learning_rate": 9.600904522613066e-05,
-      "loss": 5.7703,
-      "step": 4472
-    },
-    {
-      "epoch": 2.3327249022164276,
-      "grad_norm": 1.7052695751190186,
-      "learning_rate": 9.600804020100503e-05,
-      "loss": 5.8047,
-      "step": 4473
-    },
-    {
-      "epoch": 2.333246414602347,
-      "grad_norm": 1.6596468687057495,
-      "learning_rate": 9.60070351758794e-05,
-      "loss": 5.9574,
-      "step": 4474
-    },
-    {
-      "epoch": 2.333767926988266,
-      "grad_norm": 1.6254647970199585,
-      "learning_rate": 9.600603015075378e-05,
-      "loss": 5.6522,
-      "step": 4475
-    },
-    {
-      "epoch": 2.334289439374185,
-      "grad_norm": 2.4358134269714355,
-      "learning_rate": 9.600502512562814e-05,
-      "loss": 5.3955,
-      "step": 4476
-    },
-    {
-      "epoch": 2.334810951760104,
-      "grad_norm": 1.777006983757019,
-      "learning_rate": 9.600402010050252e-05,
-      "loss": 5.9254,
-      "step": 4477
-    },
-    {
-      "epoch": 2.3353324641460236,
-      "grad_norm": 2.2574398517608643,
-      "learning_rate": 9.60030150753769e-05,
-      "loss": 5.7396,
-      "step": 4478
-    },
-    {
-      "epoch": 2.3358539765319426,
-      "grad_norm": 1.642750859260559,
-      "learning_rate": 9.600201005025126e-05,
-      "loss": 6.1626,
-      "step": 4479
-    },
-    {
-      "epoch": 2.3363754889178616,
-      "grad_norm": 1.462148666381836,
-      "learning_rate": 9.600100502512564e-05,
-      "loss": 5.7837,
-      "step": 4480
-    },
-    {
-      "epoch": 2.336897001303781,
-      "grad_norm": 1.365599274635315,
-      "learning_rate": 9.6e-05,
-      "loss": 6.0614,
-      "step": 4481
-    },
-    {
-      "epoch": 2.3374185136897,
-      "grad_norm": 1.2903046607971191,
-      "learning_rate": 9.599899497487438e-05,
-      "loss": 5.2908,
-      "step": 4482
-    },
-    {
-      "epoch": 2.337940026075619,
-      "grad_norm": 1.455536127090454,
-      "learning_rate": 9.599798994974874e-05,
-      "loss": 6.0224,
-      "step": 4483
-    },
-    {
-      "epoch": 2.3384615384615386,
-      "grad_norm": 1.3664742708206177,
-      "learning_rate": 9.599698492462312e-05,
-      "loss": 6.1121,
-      "step": 4484
-    },
-    {
-      "epoch": 2.3389830508474576,
-      "grad_norm": 1.528079867362976,
-      "learning_rate": 9.599597989949749e-05,
-      "loss": 5.9836,
-      "step": 4485
-    },
-    {
-      "epoch": 2.3395045632333766,
-      "grad_norm": 1.520763874053955,
-      "learning_rate": 9.599497487437186e-05,
-      "loss": 5.2677,
-      "step": 4486
-    },
-    {
-      "epoch": 2.340026075619296,
-      "grad_norm": 1.4886879920959473,
-      "learning_rate": 9.599396984924623e-05,
-      "loss": 6.1658,
-      "step": 4487
-    },
-    {
-      "epoch": 2.340547588005215,
-      "grad_norm": 1.5882923603057861,
-      "learning_rate": 9.59929648241206e-05,
-      "loss": 5.6774,
-      "step": 4488
-    },
-    {
-      "epoch": 2.341069100391134,
-      "grad_norm": 1.6901179552078247,
-      "learning_rate": 9.599195979899498e-05,
-      "loss": 4.8573,
-      "step": 4489
-    },
-    {
-      "epoch": 2.3415906127770536,
-      "grad_norm": 1.4726617336273193,
-      "learning_rate": 9.599095477386935e-05,
-      "loss": 6.0376,
-      "step": 4490
-    },
-    {
-      "epoch": 2.3421121251629726,
-      "grad_norm": 1.5287299156188965,
-      "learning_rate": 9.598994974874373e-05,
-      "loss": 5.7189,
-      "step": 4491
-    },
-    {
-      "epoch": 2.3426336375488916,
-      "grad_norm": 2.0952367782592773,
-      "learning_rate": 9.598894472361809e-05,
-      "loss": 4.8662,
-      "step": 4492
-    },
-    {
-      "epoch": 2.343155149934811,
-      "grad_norm": 1.6680235862731934,
-      "learning_rate": 9.598793969849247e-05,
-      "loss": 5.654,
-      "step": 4493
-    },
-    {
-      "epoch": 2.34367666232073,
-      "grad_norm": 1.5571452379226685,
-      "learning_rate": 9.598693467336683e-05,
-      "loss": 5.8913,
-      "step": 4494
-    },
-    {
-      "epoch": 2.344198174706649,
-      "grad_norm": 1.5031296014785767,
-      "learning_rate": 9.598592964824121e-05,
-      "loss": 5.9042,
-      "step": 4495
-    },
-    {
-      "epoch": 2.3447196870925686,
-      "grad_norm": 1.675705075263977,
-      "learning_rate": 9.598492462311557e-05,
-      "loss": 6.2285,
-      "step": 4496
-    },
-    {
-      "epoch": 2.3452411994784876,
-      "grad_norm": 1.730527400970459,
-      "learning_rate": 9.598391959798995e-05,
-      "loss": 6.115,
-      "step": 4497
-    },
-    {
-      "epoch": 2.3457627118644067,
-      "grad_norm": 1.8623813390731812,
-      "learning_rate": 9.598291457286433e-05,
-      "loss": 5.7115,
-      "step": 4498
-    },
-    {
-      "epoch": 2.346284224250326,
-      "grad_norm": 1.4876582622528076,
-      "learning_rate": 9.598190954773871e-05,
-      "loss": 5.8698,
-      "step": 4499
-    },
-    {
-      "epoch": 2.346805736636245,
-      "grad_norm": 1.4324908256530762,
-      "learning_rate": 9.598090452261307e-05,
-      "loss": 5.9155,
-      "step": 4500
-    },
-    {
-      "epoch": 2.347327249022164,
-      "grad_norm": 1.574563980102539,
-      "learning_rate": 9.597989949748745e-05,
-      "loss": 4.9874,
-      "step": 4501
-    },
-    {
-      "epoch": 2.3478487614080836,
-      "grad_norm": 1.4200713634490967,
-      "learning_rate": 9.597889447236181e-05,
-      "loss": 6.0569,
-      "step": 4502
-    },
-    {
-      "epoch": 2.3483702737940026,
-      "grad_norm": 1.4853901863098145,
-      "learning_rate": 9.597788944723618e-05,
-      "loss": 5.8588,
-      "step": 4503
-    },
-    {
-      "epoch": 2.3488917861799217,
-      "grad_norm": 1.4420915842056274,
-      "learning_rate": 9.597688442211056e-05,
-      "loss": 5.7611,
-      "step": 4504
-    },
-    {
-      "epoch": 2.3494132985658407,
-      "grad_norm": 1.532476544380188,
-      "learning_rate": 9.597587939698492e-05,
-      "loss": 5.1777,
-      "step": 4505
-    },
-    {
-      "epoch": 2.34993481095176,
-      "grad_norm": 1.428236722946167,
-      "learning_rate": 9.59748743718593e-05,
-      "loss": 6.1347,
-      "step": 4506
-    },
-    {
-      "epoch": 2.350456323337679,
-      "grad_norm": 1.5343890190124512,
-      "learning_rate": 9.597386934673366e-05,
-      "loss": 6.329,
-      "step": 4507
-    },
-    {
-      "epoch": 2.350977835723598,
-      "grad_norm": 1.4280807971954346,
-      "learning_rate": 9.597286432160804e-05,
-      "loss": 5.8936,
-      "step": 4508
-    },
-    {
-      "epoch": 2.3514993481095177,
-      "grad_norm": 1.4378780126571655,
-      "learning_rate": 9.597185929648242e-05,
-      "loss": 5.7711,
-      "step": 4509
-    },
-    {
-      "epoch": 2.3520208604954367,
-      "grad_norm": 1.4348413944244385,
-      "learning_rate": 9.59708542713568e-05,
-      "loss": 6.2966,
-      "step": 4510
-    },
-    {
-      "epoch": 2.3525423728813557,
-      "grad_norm": 1.4434428215026855,
-      "learning_rate": 9.596984924623116e-05,
-      "loss": 5.8932,
-      "step": 4511
-    },
-    {
-      "epoch": 2.353063885267275,
-      "grad_norm": 1.5675967931747437,
-      "learning_rate": 9.596884422110554e-05,
-      "loss": 5.9126,
-      "step": 4512
-    },
-    {
-      "epoch": 2.353585397653194,
-      "grad_norm": 1.4478498697280884,
-      "learning_rate": 9.59678391959799e-05,
-      "loss": 5.7973,
-      "step": 4513
-    },
-    {
-      "epoch": 2.354106910039113,
-      "grad_norm": 1.6586133241653442,
-      "learning_rate": 9.596683417085428e-05,
-      "loss": 5.6017,
-      "step": 4514
-    },
-    {
-      "epoch": 2.3546284224250327,
-      "grad_norm": 1.5454996824264526,
-      "learning_rate": 9.596582914572865e-05,
-      "loss": 5.6388,
-      "step": 4515
-    },
-    {
-      "epoch": 2.3551499348109517,
-      "grad_norm": 1.5169801712036133,
-      "learning_rate": 9.596482412060302e-05,
-      "loss": 6.0144,
-      "step": 4516
-    },
-    {
-      "epoch": 2.3556714471968707,
-      "grad_norm": 1.511386513710022,
-      "learning_rate": 9.596381909547739e-05,
-      "loss": 6.0205,
-      "step": 4517
-    },
-    {
-      "epoch": 2.35619295958279,
-      "grad_norm": 1.4136213064193726,
-      "learning_rate": 9.596281407035177e-05,
-      "loss": 6.3008,
-      "step": 4518
-    },
-    {
-      "epoch": 2.356714471968709,
-      "grad_norm": 1.461074948310852,
-      "learning_rate": 9.596180904522614e-05,
-      "loss": 6.0524,
-      "step": 4519
-    },
-    {
-      "epoch": 2.3572359843546282,
-      "grad_norm": 1.5305243730545044,
-      "learning_rate": 9.596080402010051e-05,
-      "loss": 5.3605,
-      "step": 4520
-    },
-    {
-      "epoch": 2.3577574967405477,
-      "grad_norm": 1.663451075553894,
-      "learning_rate": 9.595979899497489e-05,
-      "loss": 5.4035,
-      "step": 4521
-    },
-    {
-      "epoch": 2.3582790091264667,
-      "grad_norm": 1.4723806381225586,
-      "learning_rate": 9.595879396984925e-05,
-      "loss": 6.159,
-      "step": 4522
-    },
-    {
-      "epoch": 2.3588005215123857,
-      "grad_norm": 1.403035283088684,
-      "learning_rate": 9.595778894472363e-05,
-      "loss": 6.0607,
-      "step": 4523
-    },
-    {
-      "epoch": 2.359322033898305,
-      "grad_norm": 1.5772836208343506,
-      "learning_rate": 9.595678391959799e-05,
-      "loss": 6.0094,
-      "step": 4524
-    },
-    {
-      "epoch": 2.359843546284224,
-      "grad_norm": 1.7120325565338135,
-      "learning_rate": 9.595577889447237e-05,
-      "loss": 5.5456,
-      "step": 4525
-    },
-    {
-      "epoch": 2.3603650586701432,
-      "grad_norm": 1.5275436639785767,
-      "learning_rate": 9.595477386934673e-05,
-      "loss": 5.7637,
-      "step": 4526
-    },
-    {
-      "epoch": 2.3608865710560627,
-      "grad_norm": 1.5996558666229248,
-      "learning_rate": 9.595376884422111e-05,
-      "loss": 5.6699,
-      "step": 4527
-    },
-    {
-      "epoch": 2.3614080834419817,
-      "grad_norm": 2.269113302230835,
-      "learning_rate": 9.595276381909548e-05,
-      "loss": 6.0552,
-      "step": 4528
-    },
-    {
-      "epoch": 2.3619295958279007,
-      "grad_norm": 1.7253103256225586,
-      "learning_rate": 9.595175879396985e-05,
-      "loss": 5.8906,
-      "step": 4529
-    },
-    {
-      "epoch": 2.36245110821382,
-      "grad_norm": 1.7174782752990723,
-      "learning_rate": 9.595075376884423e-05,
-      "loss": 6.0944,
-      "step": 4530
-    },
-    {
-      "epoch": 2.3629726205997392,
-      "grad_norm": 1.4801464080810547,
-      "learning_rate": 9.59497487437186e-05,
-      "loss": 6.0059,
-      "step": 4531
-    },
-    {
-      "epoch": 2.3634941329856582,
-      "grad_norm": 1.729068636894226,
-      "learning_rate": 9.594874371859297e-05,
-      "loss": 5.7179,
-      "step": 4532
-    },
-    {
-      "epoch": 2.3640156453715777,
-      "grad_norm": 1.4728573560714722,
-      "learning_rate": 9.594773869346734e-05,
-      "loss": 5.7603,
-      "step": 4533
-    },
-    {
-      "epoch": 2.3645371577574967,
-      "grad_norm": 1.4392963647842407,
-      "learning_rate": 9.594673366834172e-05,
-      "loss": 6.1734,
-      "step": 4534
-    },
-    {
-      "epoch": 2.3650586701434158,
-      "grad_norm": 1.7286458015441895,
-      "learning_rate": 9.594572864321608e-05,
-      "loss": 5.1894,
-      "step": 4535
-    },
-    {
-      "epoch": 2.365580182529335,
-      "grad_norm": 1.550196886062622,
-      "learning_rate": 9.594472361809046e-05,
-      "loss": 5.9731,
-      "step": 4536
-    },
-    {
-      "epoch": 2.3661016949152542,
-      "grad_norm": 1.5125144720077515,
-      "learning_rate": 9.594371859296482e-05,
-      "loss": 6.0515,
-      "step": 4537
-    },
-    {
-      "epoch": 2.3666232073011733,
-      "grad_norm": 1.3853062391281128,
-      "learning_rate": 9.59427135678392e-05,
-      "loss": 6.0473,
-      "step": 4538
-    },
-    {
-      "epoch": 2.3671447196870927,
-      "grad_norm": 1.6001782417297363,
-      "learning_rate": 9.594170854271358e-05,
-      "loss": 5.7747,
-      "step": 4539
-    },
-    {
-      "epoch": 2.3676662320730117,
-      "grad_norm": 1.567945122718811,
-      "learning_rate": 9.594070351758796e-05,
-      "loss": 5.8247,
-      "step": 4540
-    },
-    {
-      "epoch": 2.3681877444589308,
-      "grad_norm": 1.7327954769134521,
-      "learning_rate": 9.593969849246232e-05,
-      "loss": 4.9337,
-      "step": 4541
-    },
-    {
-      "epoch": 2.3687092568448502,
-      "grad_norm": 1.6749775409698486,
-      "learning_rate": 9.59386934673367e-05,
-      "loss": 5.8816,
-      "step": 4542
-    },
-    {
-      "epoch": 2.3692307692307693,
-      "grad_norm": 1.703731656074524,
-      "learning_rate": 9.593768844221106e-05,
-      "loss": 5.3507,
-      "step": 4543
-    },
-    {
-      "epoch": 2.3697522816166883,
-      "grad_norm": 1.559857964515686,
-      "learning_rate": 9.593668341708543e-05,
-      "loss": 5.9921,
-      "step": 4544
-    },
-    {
-      "epoch": 2.3702737940026077,
-      "grad_norm": 1.3558025360107422,
-      "learning_rate": 9.59356783919598e-05,
-      "loss": 6.0457,
-      "step": 4545
-    },
-    {
-      "epoch": 2.3707953063885268,
-      "grad_norm": 1.5474129915237427,
-      "learning_rate": 9.593467336683417e-05,
-      "loss": 5.7703,
-      "step": 4546
-    },
-    {
-      "epoch": 2.371316818774446,
-      "grad_norm": 1.417605996131897,
-      "learning_rate": 9.593366834170855e-05,
-      "loss": 5.9026,
-      "step": 4547
-    },
-    {
-      "epoch": 2.3718383311603652,
-      "grad_norm": 1.5283540487289429,
-      "learning_rate": 9.593266331658291e-05,
-      "loss": 5.7226,
-      "step": 4548
-    },
-    {
-      "epoch": 2.3723598435462843,
-      "grad_norm": 1.4524253606796265,
-      "learning_rate": 9.593165829145729e-05,
-      "loss": 5.824,
-      "step": 4549
-    },
-    {
-      "epoch": 2.3728813559322033,
-      "grad_norm": 1.6348868608474731,
-      "learning_rate": 9.593065326633167e-05,
-      "loss": 5.349,
-      "step": 4550
-    },
-    {
-      "epoch": 2.3734028683181227,
-      "grad_norm": 1.6062861680984497,
-      "learning_rate": 9.592964824120604e-05,
-      "loss": 5.9257,
-      "step": 4551
-    },
-    {
-      "epoch": 2.3739243807040418,
-      "grad_norm": 1.508410930633545,
-      "learning_rate": 9.592864321608041e-05,
-      "loss": 5.6424,
-      "step": 4552
-    },
-    {
-      "epoch": 2.374445893089961,
-      "grad_norm": 1.4030510187149048,
-      "learning_rate": 9.592763819095479e-05,
-      "loss": 6.3982,
-      "step": 4553
-    },
-    {
-      "epoch": 2.3749674054758803,
-      "grad_norm": 1.8900052309036255,
-      "learning_rate": 9.592663316582915e-05,
-      "loss": 5.3418,
-      "step": 4554
-    },
-    {
-      "epoch": 2.3754889178617993,
-      "grad_norm": 1.3713457584381104,
-      "learning_rate": 9.592562814070353e-05,
-      "loss": 5.8274,
-      "step": 4555
-    },
-    {
-      "epoch": 2.3760104302477183,
-      "grad_norm": 1.3878411054611206,
-      "learning_rate": 9.592462311557789e-05,
-      "loss": 6.1804,
-      "step": 4556
-    },
-    {
-      "epoch": 2.3765319426336378,
-      "grad_norm": 1.6658408641815186,
-      "learning_rate": 9.592361809045226e-05,
-      "loss": 5.4819,
-      "step": 4557
-    },
-    {
-      "epoch": 2.377053455019557,
-      "grad_norm": 1.5359734296798706,
-      "learning_rate": 9.592261306532663e-05,
-      "loss": 5.8124,
-      "step": 4558
-    },
-    {
-      "epoch": 2.377574967405476,
-      "grad_norm": 1.6980619430541992,
-      "learning_rate": 9.5921608040201e-05,
-      "loss": 6.0929,
-      "step": 4559
-    },
-    {
-      "epoch": 2.3780964797913953,
-      "grad_norm": 1.5610415935516357,
-      "learning_rate": 9.592060301507538e-05,
-      "loss": 5.8385,
-      "step": 4560
-    },
-    {
-      "epoch": 2.3786179921773143,
-      "grad_norm": 1.5844590663909912,
-      "learning_rate": 9.591959798994975e-05,
-      "loss": 5.5091,
-      "step": 4561
-    },
-    {
-      "epoch": 2.3791395045632333,
-      "grad_norm": 1.5414739847183228,
-      "learning_rate": 9.591859296482413e-05,
-      "loss": 5.7191,
-      "step": 4562
-    },
-    {
-      "epoch": 2.3796610169491528,
-      "grad_norm": 1.387691617012024,
-      "learning_rate": 9.59175879396985e-05,
-      "loss": 6.0045,
-      "step": 4563
-    },
-    {
-      "epoch": 2.380182529335072,
-      "grad_norm": 1.513169527053833,
-      "learning_rate": 9.591658291457287e-05,
-      "loss": 5.7804,
-      "step": 4564
-    },
-    {
-      "epoch": 2.380704041720991,
-      "grad_norm": 1.8554631471633911,
-      "learning_rate": 9.591557788944724e-05,
-      "loss": 5.3349,
-      "step": 4565
-    },
-    {
-      "epoch": 2.3812255541069103,
-      "grad_norm": 1.8047826290130615,
-      "learning_rate": 9.591457286432162e-05,
-      "loss": 5.9193,
-      "step": 4566
-    },
-    {
-      "epoch": 2.3817470664928293,
-      "grad_norm": 1.5484076738357544,
-      "learning_rate": 9.591356783919598e-05,
-      "loss": 5.8028,
-      "step": 4567
-    },
-    {
-      "epoch": 2.3822685788787483,
-      "grad_norm": 1.5238128900527954,
-      "learning_rate": 9.591256281407036e-05,
-      "loss": 6.213,
-      "step": 4568
-    },
-    {
-      "epoch": 2.3827900912646673,
-      "grad_norm": 1.4808193445205688,
-      "learning_rate": 9.591155778894472e-05,
-      "loss": 5.8315,
-      "step": 4569
-    },
-    {
-      "epoch": 2.383311603650587,
-      "grad_norm": 1.717157244682312,
-      "learning_rate": 9.59105527638191e-05,
-      "loss": 5.318,
-      "step": 4570
-    },
-    {
-      "epoch": 2.383833116036506,
-      "grad_norm": 1.5507549047470093,
-      "learning_rate": 9.590954773869348e-05,
-      "loss": 5.8774,
-      "step": 4571
-    },
-    {
-      "epoch": 2.384354628422425,
-      "grad_norm": 1.53330659866333,
-      "learning_rate": 9.590854271356784e-05,
-      "loss": 5.9961,
-      "step": 4572
-    },
-    {
-      "epoch": 2.3848761408083443,
-      "grad_norm": 1.7568445205688477,
-      "learning_rate": 9.590753768844222e-05,
-      "loss": 6.0764,
-      "step": 4573
-    },
-    {
-      "epoch": 2.3853976531942633,
-      "grad_norm": 1.6111587285995483,
-      "learning_rate": 9.590653266331658e-05,
-      "loss": 5.6564,
-      "step": 4574
-    },
-    {
-      "epoch": 2.3859191655801824,
-      "grad_norm": 1.4940314292907715,
-      "learning_rate": 9.590552763819096e-05,
-      "loss": 5.5372,
-      "step": 4575
-    },
-    {
-      "epoch": 2.386440677966102,
-      "grad_norm": 1.6053948402404785,
-      "learning_rate": 9.590452261306533e-05,
-      "loss": 6.1279,
-      "step": 4576
-    },
-    {
-      "epoch": 2.386962190352021,
-      "grad_norm": 1.5130175352096558,
-      "learning_rate": 9.59035175879397e-05,
-      "loss": 5.9668,
-      "step": 4577
-    },
-    {
-      "epoch": 2.38748370273794,
-      "grad_norm": 1.5389124155044556,
-      "learning_rate": 9.590251256281407e-05,
-      "loss": 5.811,
-      "step": 4578
-    },
-    {
-      "epoch": 2.3880052151238593,
-      "grad_norm": 1.3652188777923584,
-      "learning_rate": 9.590150753768845e-05,
-      "loss": 5.9825,
-      "step": 4579
-    },
-    {
-      "epoch": 2.3885267275097783,
-      "grad_norm": 1.3439152240753174,
-      "learning_rate": 9.590050251256281e-05,
-      "loss": 6.0908,
-      "step": 4580
-    },
-    {
-      "epoch": 2.3890482398956974,
-      "grad_norm": 1.574779748916626,
-      "learning_rate": 9.589949748743719e-05,
-      "loss": 5.7468,
-      "step": 4581
-    },
-    {
-      "epoch": 2.389569752281617,
-      "grad_norm": 1.4520387649536133,
-      "learning_rate": 9.589849246231157e-05,
-      "loss": 6.0826,
-      "step": 4582
-    },
-    {
-      "epoch": 2.390091264667536,
-      "grad_norm": 1.4827810525894165,
-      "learning_rate": 9.589748743718593e-05,
-      "loss": 6.0136,
-      "step": 4583
-    },
-    {
-      "epoch": 2.390612777053455,
-      "grad_norm": 1.5205166339874268,
-      "learning_rate": 9.589648241206031e-05,
-      "loss": 5.6181,
-      "step": 4584
-    },
-    {
-      "epoch": 2.3911342894393743,
-      "grad_norm": 2.024791955947876,
-      "learning_rate": 9.589547738693467e-05,
-      "loss": 6.0385,
-      "step": 4585
-    },
-    {
-      "epoch": 2.3916558018252934,
-      "grad_norm": 1.4049091339111328,
-      "learning_rate": 9.589447236180905e-05,
-      "loss": 6.0032,
-      "step": 4586
-    },
-    {
-      "epoch": 2.3921773142112124,
-      "grad_norm": 1.4850876331329346,
-      "learning_rate": 9.589346733668342e-05,
-      "loss": 5.8961,
-      "step": 4587
-    },
-    {
-      "epoch": 2.392698826597132,
-      "grad_norm": 1.6333667039871216,
-      "learning_rate": 9.589246231155779e-05,
-      "loss": 5.7758,
-      "step": 4588
-    },
-    {
-      "epoch": 2.393220338983051,
-      "grad_norm": 1.673466444015503,
-      "learning_rate": 9.589145728643216e-05,
-      "loss": 5.5469,
-      "step": 4589
-    },
-    {
-      "epoch": 2.39374185136897,
-      "grad_norm": 1.4635727405548096,
-      "learning_rate": 9.589045226130654e-05,
-      "loss": 6.0237,
-      "step": 4590
-    },
-    {
-      "epoch": 2.3942633637548894,
-      "grad_norm": 1.6333473920822144,
-      "learning_rate": 9.588944723618091e-05,
-      "loss": 5.1516,
-      "step": 4591
-    },
-    {
-      "epoch": 2.3947848761408084,
-      "grad_norm": 1.5557090044021606,
-      "learning_rate": 9.588844221105529e-05,
-      "loss": 5.5709,
-      "step": 4592
-    },
-    {
-      "epoch": 2.3953063885267274,
-      "grad_norm": 1.6182422637939453,
-      "learning_rate": 9.588743718592966e-05,
-      "loss": 5.4256,
-      "step": 4593
-    },
-    {
-      "epoch": 2.395827900912647,
-      "grad_norm": 1.4957574605941772,
-      "learning_rate": 9.588643216080403e-05,
-      "loss": 6.0476,
-      "step": 4594
-    },
-    {
-      "epoch": 2.396349413298566,
-      "grad_norm": 1.7536898851394653,
-      "learning_rate": 9.58854271356784e-05,
-      "loss": 5.1066,
-      "step": 4595
-    },
-    {
-      "epoch": 2.396870925684485,
-      "grad_norm": 1.4505441188812256,
-      "learning_rate": 9.588442211055276e-05,
-      "loss": 5.4728,
-      "step": 4596
-    },
-    {
-      "epoch": 2.3973924380704044,
-      "grad_norm": 1.4545371532440186,
-      "learning_rate": 9.588341708542714e-05,
-      "loss": 6.0833,
-      "step": 4597
-    },
-    {
-      "epoch": 2.3979139504563234,
-      "grad_norm": 1.5019114017486572,
-      "learning_rate": 9.58824120603015e-05,
-      "loss": 5.6848,
-      "step": 4598
-    },
-    {
-      "epoch": 2.3984354628422424,
-      "grad_norm": 1.6857526302337646,
-      "learning_rate": 9.588140703517588e-05,
-      "loss": 5.5029,
-      "step": 4599
-    },
-    {
-      "epoch": 2.3989569752281614,
-      "grad_norm": 2.2362821102142334,
-      "learning_rate": 9.588040201005025e-05,
-      "loss": 5.8717,
-      "step": 4600
-    },
-    {
-      "epoch": 2.399478487614081,
-      "grad_norm": 1.6894989013671875,
-      "learning_rate": 9.587939698492462e-05,
-      "loss": 5.8724,
-      "step": 4601
-    },
-    {
-      "epoch": 2.4,
-      "grad_norm": 1.631055474281311,
-      "learning_rate": 9.5878391959799e-05,
-      "loss": 5.8487,
-      "step": 4602
-    },
-    {
-      "epoch": 2.400521512385919,
-      "grad_norm": 1.458709716796875,
-      "learning_rate": 9.587738693467338e-05,
-      "loss": 6.1201,
-      "step": 4603
-    },
-    {
-      "epoch": 2.4010430247718384,
-      "grad_norm": 1.5102814435958862,
-      "learning_rate": 9.587638190954774e-05,
-      "loss": 5.9145,
-      "step": 4604
-    },
-    {
-      "epoch": 2.4015645371577574,
-      "grad_norm": 1.3990081548690796,
-      "learning_rate": 9.587537688442212e-05,
-      "loss": 6.0381,
-      "step": 4605
-    },
-    {
-      "epoch": 2.4020860495436764,
-      "grad_norm": 1.5177686214447021,
-      "learning_rate": 9.587437185929649e-05,
-      "loss": 5.8868,
-      "step": 4606
-    },
-    {
-      "epoch": 2.402607561929596,
-      "grad_norm": 1.540640950202942,
-      "learning_rate": 9.587336683417086e-05,
-      "loss": 5.7346,
-      "step": 4607
-    },
-    {
-      "epoch": 2.403129074315515,
-      "grad_norm": 1.5567206144332886,
-      "learning_rate": 9.587236180904523e-05,
-      "loss": 5.985,
-      "step": 4608
-    },
-    {
-      "epoch": 2.403650586701434,
-      "grad_norm": 1.5360157489776611,
-      "learning_rate": 9.58713567839196e-05,
-      "loss": 5.4928,
-      "step": 4609
-    },
-    {
-      "epoch": 2.4041720990873534,
-      "grad_norm": 1.5946108102798462,
-      "learning_rate": 9.587035175879397e-05,
-      "loss": 5.6713,
-      "step": 4610
-    },
-    {
-      "epoch": 2.4046936114732724,
-      "grad_norm": 1.4823250770568848,
-      "learning_rate": 9.586934673366835e-05,
-      "loss": 6.1183,
-      "step": 4611
-    },
-    {
-      "epoch": 2.4052151238591915,
-      "grad_norm": 1.450477123260498,
-      "learning_rate": 9.586834170854273e-05,
-      "loss": 6.335,
-      "step": 4612
-    },
-    {
-      "epoch": 2.405736636245111,
-      "grad_norm": 1.6137751340866089,
-      "learning_rate": 9.586733668341709e-05,
-      "loss": 5.6883,
-      "step": 4613
-    },
-    {
-      "epoch": 2.40625814863103,
-      "grad_norm": 1.3729891777038574,
-      "learning_rate": 9.586633165829147e-05,
-      "loss": 6.294,
-      "step": 4614
-    },
-    {
-      "epoch": 2.406779661016949,
-      "grad_norm": 1.6082178354263306,
-      "learning_rate": 9.586532663316583e-05,
-      "loss": 5.831,
-      "step": 4615
-    },
-    {
-      "epoch": 2.4073011734028684,
-      "grad_norm": 1.5930973291397095,
-      "learning_rate": 9.586432160804021e-05,
-      "loss": 5.4537,
-      "step": 4616
-    },
-    {
-      "epoch": 2.4078226857887874,
-      "grad_norm": 1.6211504936218262,
-      "learning_rate": 9.586331658291457e-05,
-      "loss": 5.7793,
-      "step": 4617
-    },
-    {
-      "epoch": 2.4083441981747065,
-      "grad_norm": 1.5887620449066162,
-      "learning_rate": 9.586231155778895e-05,
-      "loss": 5.4626,
-      "step": 4618
-    },
-    {
-      "epoch": 2.408865710560626,
-      "grad_norm": 1.5088756084442139,
-      "learning_rate": 9.586130653266332e-05,
-      "loss": 5.4662,
-      "step": 4619
-    },
-    {
-      "epoch": 2.409387222946545,
-      "grad_norm": 1.4432193040847778,
-      "learning_rate": 9.58603015075377e-05,
-      "loss": 6.064,
-      "step": 4620
-    },
-    {
-      "epoch": 2.409908735332464,
-      "grad_norm": 1.4517239332199097,
-      "learning_rate": 9.585929648241206e-05,
-      "loss": 6.1191,
-      "step": 4621
-    },
-    {
-      "epoch": 2.4104302477183834,
-      "grad_norm": 1.4885787963867188,
-      "learning_rate": 9.585829145728644e-05,
-      "loss": 5.9381,
-      "step": 4622
-    },
-    {
-      "epoch": 2.4109517601043025,
-      "grad_norm": 1.346200942993164,
-      "learning_rate": 9.585728643216081e-05,
-      "loss": 6.2227,
-      "step": 4623
-    },
-    {
-      "epoch": 2.4114732724902215,
-      "grad_norm": 1.4736666679382324,
-      "learning_rate": 9.585628140703518e-05,
-      "loss": 5.7237,
-      "step": 4624
-    },
-    {
-      "epoch": 2.411994784876141,
-      "grad_norm": 1.4049171209335327,
-      "learning_rate": 9.585527638190956e-05,
-      "loss": 6.0698,
-      "step": 4625
-    },
-    {
-      "epoch": 2.41251629726206,
-      "grad_norm": 1.551905870437622,
-      "learning_rate": 9.585427135678392e-05,
-      "loss": 5.675,
-      "step": 4626
-    },
-    {
-      "epoch": 2.413037809647979,
-      "grad_norm": 1.4387238025665283,
-      "learning_rate": 9.58532663316583e-05,
-      "loss": 6.113,
-      "step": 4627
-    },
-    {
-      "epoch": 2.4135593220338984,
-      "grad_norm": 1.519805908203125,
-      "learning_rate": 9.585226130653266e-05,
-      "loss": 5.8956,
-      "step": 4628
-    },
-    {
-      "epoch": 2.4140808344198175,
-      "grad_norm": 1.4321504831314087,
-      "learning_rate": 9.585125628140704e-05,
-      "loss": 5.9626,
-      "step": 4629
-    },
-    {
-      "epoch": 2.4146023468057365,
-      "grad_norm": 1.4967197179794312,
-      "learning_rate": 9.58502512562814e-05,
-      "loss": 6.1177,
-      "step": 4630
-    },
-    {
-      "epoch": 2.415123859191656,
-      "grad_norm": 1.3577544689178467,
-      "learning_rate": 9.584924623115578e-05,
-      "loss": 5.5475,
-      "step": 4631
-    },
-    {
-      "epoch": 2.415645371577575,
-      "grad_norm": 1.4306024312973022,
-      "learning_rate": 9.584824120603016e-05,
-      "loss": 5.9575,
-      "step": 4632
-    },
-    {
-      "epoch": 2.416166883963494,
-      "grad_norm": 1.3611029386520386,
-      "learning_rate": 9.584723618090454e-05,
-      "loss": 6.1253,
-      "step": 4633
-    },
-    {
-      "epoch": 2.4166883963494135,
-      "grad_norm": 1.4228655099868774,
-      "learning_rate": 9.58462311557789e-05,
-      "loss": 6.0514,
-      "step": 4634
-    },
-    {
-      "epoch": 2.4172099087353325,
-      "grad_norm": 1.5801129341125488,
-      "learning_rate": 9.584522613065328e-05,
-      "loss": 5.8469,
-      "step": 4635
-    },
-    {
-      "epoch": 2.4177314211212515,
-      "grad_norm": 1.451980471611023,
-      "learning_rate": 9.584422110552764e-05,
-      "loss": 6.1922,
-      "step": 4636
-    },
-    {
-      "epoch": 2.418252933507171,
-      "grad_norm": 1.3929085731506348,
-      "learning_rate": 9.584321608040201e-05,
-      "loss": 6.1539,
-      "step": 4637
-    },
-    {
-      "epoch": 2.41877444589309,
-      "grad_norm": 1.604539394378662,
-      "learning_rate": 9.584221105527639e-05,
-      "loss": 5.7634,
-      "step": 4638
-    },
-    {
-      "epoch": 2.419295958279009,
-      "grad_norm": 1.4325387477874756,
-      "learning_rate": 9.584120603015075e-05,
-      "loss": 6.032,
-      "step": 4639
-    },
-    {
-      "epoch": 2.4198174706649285,
-      "grad_norm": 1.3736181259155273,
-      "learning_rate": 9.584020100502513e-05,
-      "loss": 6.0197,
-      "step": 4640
-    },
-    {
-      "epoch": 2.4203389830508475,
-      "grad_norm": 1.39095938205719,
-      "learning_rate": 9.583919597989949e-05,
-      "loss": 5.8531,
-      "step": 4641
-    },
-    {
-      "epoch": 2.4208604954367665,
-      "grad_norm": 1.4647318124771118,
-      "learning_rate": 9.583819095477387e-05,
-      "loss": 6.0885,
-      "step": 4642
-    },
-    {
-      "epoch": 2.421382007822686,
-      "grad_norm": 1.495478868484497,
-      "learning_rate": 9.583718592964825e-05,
-      "loss": 5.6628,
-      "step": 4643
-    },
-    {
-      "epoch": 2.421903520208605,
-      "grad_norm": 1.4674917459487915,
-      "learning_rate": 9.583618090452263e-05,
-      "loss": 5.8017,
-      "step": 4644
-    },
-    {
-      "epoch": 2.422425032594524,
-      "grad_norm": 1.6027084589004517,
-      "learning_rate": 9.583517587939699e-05,
-      "loss": 5.8724,
-      "step": 4645
-    },
-    {
-      "epoch": 2.4229465449804435,
-      "grad_norm": 1.400303840637207,
-      "learning_rate": 9.583417085427137e-05,
-      "loss": 6.215,
-      "step": 4646
-    },
-    {
-      "epoch": 2.4234680573663625,
-      "grad_norm": 1.4617226123809814,
-      "learning_rate": 9.583316582914573e-05,
-      "loss": 6.1428,
-      "step": 4647
-    },
-    {
-      "epoch": 2.4239895697522815,
-      "grad_norm": 1.8257728815078735,
-      "learning_rate": 9.583216080402011e-05,
-      "loss": 5.3695,
-      "step": 4648
-    },
-    {
-      "epoch": 2.424511082138201,
-      "grad_norm": 1.6658849716186523,
-      "learning_rate": 9.583115577889447e-05,
-      "loss": 5.6682,
-      "step": 4649
-    },
-    {
-      "epoch": 2.42503259452412,
-      "grad_norm": 1.6101568937301636,
-      "learning_rate": 9.583015075376884e-05,
-      "loss": 6.2043,
-      "step": 4650
-    },
-    {
-      "epoch": 2.425554106910039,
-      "grad_norm": 1.4413847923278809,
-      "learning_rate": 9.582914572864322e-05,
-      "loss": 5.9258,
-      "step": 4651
-    },
-    {
-      "epoch": 2.4260756192959585,
-      "grad_norm": 1.541852355003357,
-      "learning_rate": 9.58281407035176e-05,
-      "loss": 5.7365,
-      "step": 4652
-    },
-    {
-      "epoch": 2.4265971316818775,
-      "grad_norm": 1.359946608543396,
-      "learning_rate": 9.582713567839197e-05,
-      "loss": 6.0591,
-      "step": 4653
-    },
-    {
-      "epoch": 2.4271186440677965,
-      "grad_norm": 1.4947724342346191,
-      "learning_rate": 9.582613065326634e-05,
-      "loss": 6.0119,
-      "step": 4654
-    },
-    {
-      "epoch": 2.427640156453716,
-      "grad_norm": 1.4753408432006836,
-      "learning_rate": 9.582512562814071e-05,
-      "loss": 5.9772,
-      "step": 4655
-    },
-    {
-      "epoch": 2.428161668839635,
-      "grad_norm": 1.4500113725662231,
-      "learning_rate": 9.582412060301508e-05,
-      "loss": 6.1391,
-      "step": 4656
-    },
-    {
-      "epoch": 2.428683181225554,
-      "grad_norm": 1.4572314023971558,
-      "learning_rate": 9.582311557788946e-05,
-      "loss": 5.8599,
-      "step": 4657
-    },
-    {
-      "epoch": 2.4292046936114735,
-      "grad_norm": 1.4415271282196045,
-      "learning_rate": 9.582211055276382e-05,
-      "loss": 6.0784,
-      "step": 4658
-    },
-    {
-      "epoch": 2.4297262059973925,
-      "grad_norm": 1.4185330867767334,
-      "learning_rate": 9.58211055276382e-05,
-      "loss": 6.1459,
-      "step": 4659
-    },
-    {
-      "epoch": 2.4302477183833116,
-      "grad_norm": 1.424500584602356,
-      "learning_rate": 9.582010050251256e-05,
-      "loss": 5.7663,
-      "step": 4660
-    },
-    {
-      "epoch": 2.430769230769231,
-      "grad_norm": 1.5055536031723022,
-      "learning_rate": 9.581909547738694e-05,
-      "loss": 5.8789,
-      "step": 4661
-    },
-    {
-      "epoch": 2.43129074315515,
-      "grad_norm": 1.6713930368423462,
-      "learning_rate": 9.58180904522613e-05,
-      "loss": 5.4247,
-      "step": 4662
-    },
-    {
-      "epoch": 2.431812255541069,
-      "grad_norm": 1.8298808336257935,
-      "learning_rate": 9.581708542713568e-05,
-      "loss": 5.7908,
-      "step": 4663
-    },
-    {
-      "epoch": 2.432333767926988,
-      "grad_norm": 1.4650450944900513,
-      "learning_rate": 9.581608040201006e-05,
-      "loss": 5.9735,
-      "step": 4664
-    },
-    {
-      "epoch": 2.4328552803129075,
-      "grad_norm": 1.2734521627426147,
-      "learning_rate": 9.581507537688443e-05,
-      "loss": 6.0537,
-      "step": 4665
-    },
-    {
-      "epoch": 2.4333767926988266,
-      "grad_norm": 1.4770925045013428,
-      "learning_rate": 9.58140703517588e-05,
-      "loss": 5.8323,
-      "step": 4666
-    },
-    {
-      "epoch": 2.4338983050847456,
-      "grad_norm": 1.4586842060089111,
-      "learning_rate": 9.581306532663317e-05,
-      "loss": 6.236,
-      "step": 4667
-    },
-    {
-      "epoch": 2.434419817470665,
-      "grad_norm": 1.4851281642913818,
-      "learning_rate": 9.581206030150754e-05,
-      "loss": 5.8193,
-      "step": 4668
-    },
-    {
-      "epoch": 2.434941329856584,
-      "grad_norm": 1.426017165184021,
-      "learning_rate": 9.581105527638191e-05,
-      "loss": 5.9804,
-      "step": 4669
-    },
-    {
-      "epoch": 2.435462842242503,
-      "grad_norm": 1.4702129364013672,
-      "learning_rate": 9.581005025125629e-05,
-      "loss": 5.8185,
-      "step": 4670
-    },
-    {
-      "epoch": 2.4359843546284226,
-      "grad_norm": 1.3530999422073364,
-      "learning_rate": 9.580904522613065e-05,
-      "loss": 6.1128,
-      "step": 4671
-    },
-    {
-      "epoch": 2.4365058670143416,
-      "grad_norm": 1.3138591051101685,
-      "learning_rate": 9.580804020100503e-05,
-      "loss": 6.0128,
-      "step": 4672
-    },
-    {
-      "epoch": 2.4370273794002606,
-      "grad_norm": 1.565187931060791,
-      "learning_rate": 9.580703517587941e-05,
-      "loss": 5.8332,
-      "step": 4673
-    },
-    {
-      "epoch": 2.43754889178618,
-      "grad_norm": 1.5213370323181152,
-      "learning_rate": 9.580603015075378e-05,
-      "loss": 5.6769,
-      "step": 4674
-    },
-    {
-      "epoch": 2.438070404172099,
-      "grad_norm": 1.5922952890396118,
-      "learning_rate": 9.580502512562815e-05,
-      "loss": 4.7799,
-      "step": 4675
-    },
-    {
-      "epoch": 2.438591916558018,
-      "grad_norm": 1.5340605974197388,
-      "learning_rate": 9.580402010050251e-05,
-      "loss": 6.0069,
-      "step": 4676
-    },
-    {
-      "epoch": 2.4391134289439376,
-      "grad_norm": 1.5610226392745972,
-      "learning_rate": 9.580301507537689e-05,
-      "loss": 5.7036,
-      "step": 4677
-    },
-    {
-      "epoch": 2.4396349413298566,
-      "grad_norm": 1.5556902885437012,
-      "learning_rate": 9.580201005025126e-05,
-      "loss": 5.7152,
-      "step": 4678
-    },
-    {
-      "epoch": 2.4401564537157756,
-      "grad_norm": 1.5280637741088867,
-      "learning_rate": 9.580100502512563e-05,
-      "loss": 5.395,
-      "step": 4679
-    },
-    {
-      "epoch": 2.440677966101695,
-      "grad_norm": 1.4267419576644897,
-      "learning_rate": 9.58e-05,
-      "loss": 6.193,
-      "step": 4680
-    },
-    {
-      "epoch": 2.441199478487614,
-      "grad_norm": 1.59563410282135,
-      "learning_rate": 9.579899497487438e-05,
-      "loss": 5.9021,
-      "step": 4681
-    },
-    {
-      "epoch": 2.441720990873533,
-      "grad_norm": 1.6640350818634033,
-      "learning_rate": 9.579798994974874e-05,
-      "loss": 5.3433,
-      "step": 4682
-    },
-    {
-      "epoch": 2.4422425032594526,
-      "grad_norm": 1.5032236576080322,
-      "learning_rate": 9.579698492462312e-05,
-      "loss": 5.204,
-      "step": 4683
-    },
-    {
-      "epoch": 2.4427640156453716,
-      "grad_norm": 1.8647037744522095,
-      "learning_rate": 9.57959798994975e-05,
-      "loss": 5.2493,
-      "step": 4684
-    },
-    {
-      "epoch": 2.4432855280312906,
-      "grad_norm": 1.6246224641799927,
-      "learning_rate": 9.579497487437187e-05,
-      "loss": 5.9873,
-      "step": 4685
-    },
-    {
-      "epoch": 2.44380704041721,
-      "grad_norm": 1.489141583442688,
-      "learning_rate": 9.579396984924624e-05,
-      "loss": 5.2149,
-      "step": 4686
-    },
-    {
-      "epoch": 2.444328552803129,
-      "grad_norm": 1.4809342622756958,
-      "learning_rate": 9.579296482412062e-05,
-      "loss": 5.49,
-      "step": 4687
-    },
-    {
-      "epoch": 2.444850065189048,
-      "grad_norm": 1.4862064123153687,
-      "learning_rate": 9.579195979899498e-05,
-      "loss": 5.7145,
-      "step": 4688
-    },
-    {
-      "epoch": 2.4453715775749676,
-      "grad_norm": 1.4272385835647583,
-      "learning_rate": 9.579095477386934e-05,
-      "loss": 6.1677,
-      "step": 4689
-    },
-    {
-      "epoch": 2.4458930899608866,
-      "grad_norm": 1.5657410621643066,
-      "learning_rate": 9.578994974874372e-05,
-      "loss": 5.5093,
-      "step": 4690
-    },
-    {
-      "epoch": 2.4464146023468056,
-      "grad_norm": 1.6081621646881104,
-      "learning_rate": 9.578894472361809e-05,
-      "loss": 4.6397,
-      "step": 4691
-    },
-    {
-      "epoch": 2.4469361147327247,
-      "grad_norm": 1.640946388244629,
-      "learning_rate": 9.578793969849246e-05,
-      "loss": 5.8015,
-      "step": 4692
-    },
-    {
-      "epoch": 2.447457627118644,
-      "grad_norm": 1.390912652015686,
-      "learning_rate": 9.578693467336684e-05,
-      "loss": 6.375,
-      "step": 4693
-    },
-    {
-      "epoch": 2.447979139504563,
-      "grad_norm": 1.6535742282867432,
-      "learning_rate": 9.578592964824122e-05,
-      "loss": 5.1161,
-      "step": 4694
-    },
-    {
-      "epoch": 2.448500651890482,
-      "grad_norm": 1.5865446329116821,
-      "learning_rate": 9.578492462311558e-05,
-      "loss": 5.1061,
-      "step": 4695
-    },
-    {
-      "epoch": 2.4490221642764016,
-      "grad_norm": 1.513089895248413,
-      "learning_rate": 9.578391959798996e-05,
-      "loss": 5.5305,
-      "step": 4696
-    },
-    {
-      "epoch": 2.4495436766623206,
-      "grad_norm": 1.5821477174758911,
-      "learning_rate": 9.578291457286433e-05,
-      "loss": 5.9291,
-      "step": 4697
-    },
-    {
-      "epoch": 2.4500651890482397,
-      "grad_norm": 1.3279614448547363,
-      "learning_rate": 9.57819095477387e-05,
-      "loss": 6.1858,
-      "step": 4698
-    },
-    {
-      "epoch": 2.450586701434159,
-      "grad_norm": 1.6804585456848145,
-      "learning_rate": 9.578090452261307e-05,
-      "loss": 5.3348,
-      "step": 4699
-    },
-    {
-      "epoch": 2.451108213820078,
-      "grad_norm": 1.386252522468567,
-      "learning_rate": 9.577989949748745e-05,
-      "loss": 6.0966,
-      "step": 4700
-    },
-    {
-      "epoch": 2.451629726205997,
-      "grad_norm": 1.5297173261642456,
-      "learning_rate": 9.577889447236181e-05,
-      "loss": 6.1776,
-      "step": 4701
-    },
-    {
-      "epoch": 2.4521512385919166,
-      "grad_norm": 1.6720232963562012,
-      "learning_rate": 9.577788944723619e-05,
-      "loss": 5.7404,
-      "step": 4702
-    },
-    {
-      "epoch": 2.4526727509778357,
-      "grad_norm": 1.7850764989852905,
-      "learning_rate": 9.577688442211055e-05,
-      "loss": 5.8397,
-      "step": 4703
-    },
-    {
-      "epoch": 2.4531942633637547,
-      "grad_norm": 1.8239599466323853,
-      "learning_rate": 9.577587939698493e-05,
-      "loss": 5.5024,
-      "step": 4704
-    },
-    {
-      "epoch": 2.453715775749674,
-      "grad_norm": 1.4998780488967896,
-      "learning_rate": 9.577487437185931e-05,
-      "loss": 6.064,
-      "step": 4705
-    },
-    {
-      "epoch": 2.454237288135593,
-      "grad_norm": 1.599571704864502,
-      "learning_rate": 9.577386934673367e-05,
-      "loss": 5.8617,
-      "step": 4706
-    },
-    {
-      "epoch": 2.454758800521512,
-      "grad_norm": 1.620847463607788,
-      "learning_rate": 9.577286432160805e-05,
-      "loss": 5.819,
-      "step": 4707
-    },
-    {
-      "epoch": 2.4552803129074317,
-      "grad_norm": 1.6738375425338745,
-      "learning_rate": 9.577185929648241e-05,
-      "loss": 5.762,
-      "step": 4708
-    },
-    {
-      "epoch": 2.4558018252933507,
-      "grad_norm": 1.5590448379516602,
-      "learning_rate": 9.577085427135679e-05,
-      "loss": 6.0509,
-      "step": 4709
-    },
-    {
-      "epoch": 2.4563233376792697,
-      "grad_norm": 1.493691325187683,
-      "learning_rate": 9.576984924623116e-05,
-      "loss": 6.1952,
-      "step": 4710
-    },
-    {
-      "epoch": 2.456844850065189,
-      "grad_norm": 1.7104767560958862,
-      "learning_rate": 9.576884422110553e-05,
-      "loss": 5.5211,
-      "step": 4711
-    },
-    {
-      "epoch": 2.457366362451108,
-      "grad_norm": 1.7929563522338867,
-      "learning_rate": 9.57678391959799e-05,
-      "loss": 5.7398,
-      "step": 4712
-    },
-    {
-      "epoch": 2.457887874837027,
-      "grad_norm": 1.6599271297454834,
-      "learning_rate": 9.576683417085428e-05,
-      "loss": 5.7055,
-      "step": 4713
-    },
-    {
-      "epoch": 2.4584093872229467,
-      "grad_norm": 1.6074079275131226,
-      "learning_rate": 9.576582914572864e-05,
-      "loss": 6.1165,
-      "step": 4714
-    },
-    {
-      "epoch": 2.4589308996088657,
-      "grad_norm": 1.6088043451309204,
-      "learning_rate": 9.576482412060302e-05,
-      "loss": 5.8875,
-      "step": 4715
-    },
-    {
-      "epoch": 2.4594524119947847,
-      "grad_norm": 1.594504952430725,
-      "learning_rate": 9.57638190954774e-05,
-      "loss": 6.0006,
-      "step": 4716
-    },
-    {
-      "epoch": 2.459973924380704,
-      "grad_norm": 1.6713759899139404,
-      "learning_rate": 9.576281407035176e-05,
-      "loss": 5.3991,
-      "step": 4717
-    },
-    {
-      "epoch": 2.460495436766623,
-      "grad_norm": 2.031285285949707,
-      "learning_rate": 9.576180904522614e-05,
-      "loss": 5.3336,
-      "step": 4718
-    },
-    {
-      "epoch": 2.461016949152542,
-      "grad_norm": 1.549748182296753,
-      "learning_rate": 9.57608040201005e-05,
-      "loss": 5.8627,
-      "step": 4719
-    },
-    {
-      "epoch": 2.4615384615384617,
-      "grad_norm": 1.3644412755966187,
-      "learning_rate": 9.575979899497488e-05,
-      "loss": 5.841,
-      "step": 4720
-    },
-    {
-      "epoch": 2.4620599739243807,
-      "grad_norm": 1.56881844997406,
-      "learning_rate": 9.575879396984924e-05,
-      "loss": 6.0316,
-      "step": 4721
-    },
-    {
-      "epoch": 2.4625814863102997,
-      "grad_norm": 1.541034460067749,
-      "learning_rate": 9.575778894472362e-05,
-      "loss": 5.7192,
-      "step": 4722
-    },
-    {
-      "epoch": 2.463102998696219,
-      "grad_norm": 1.4114419221878052,
-      "learning_rate": 9.575678391959799e-05,
-      "loss": 6.1718,
-      "step": 4723
-    },
-    {
-      "epoch": 2.463624511082138,
-      "grad_norm": 1.4744092226028442,
-      "learning_rate": 9.575577889447236e-05,
-      "loss": 5.459,
-      "step": 4724
-    },
-    {
-      "epoch": 2.4641460234680572,
-      "grad_norm": 1.394659161567688,
-      "learning_rate": 9.575477386934674e-05,
-      "loss": 6.0779,
-      "step": 4725
-    },
-    {
-      "epoch": 2.4646675358539767,
-      "grad_norm": 1.498979091644287,
-      "learning_rate": 9.575376884422112e-05,
-      "loss": 5.7417,
-      "step": 4726
-    },
-    {
-      "epoch": 2.4651890482398957,
-      "grad_norm": 1.8554108142852783,
-      "learning_rate": 9.575276381909548e-05,
-      "loss": 5.2206,
-      "step": 4727
-    },
-    {
-      "epoch": 2.4657105606258147,
-      "grad_norm": 1.4611144065856934,
-      "learning_rate": 9.575175879396986e-05,
-      "loss": 5.6197,
-      "step": 4728
-    },
-    {
-      "epoch": 2.466232073011734,
-      "grad_norm": 1.5486135482788086,
-      "learning_rate": 9.575075376884423e-05,
-      "loss": 6.0438,
-      "step": 4729
-    },
-    {
-      "epoch": 2.466753585397653,
-      "grad_norm": 1.507443904876709,
-      "learning_rate": 9.574974874371859e-05,
-      "loss": 5.8684,
-      "step": 4730
-    },
-    {
-      "epoch": 2.4672750977835722,
-      "grad_norm": 1.5569655895233154,
-      "learning_rate": 9.574874371859297e-05,
-      "loss": 5.813,
-      "step": 4731
-    },
-    {
-      "epoch": 2.4677966101694917,
-      "grad_norm": 1.544553518295288,
-      "learning_rate": 9.574773869346733e-05,
-      "loss": 5.7935,
-      "step": 4732
-    },
-    {
-      "epoch": 2.4683181225554107,
-      "grad_norm": 1.3460643291473389,
-      "learning_rate": 9.574673366834171e-05,
-      "loss": 6.0375,
-      "step": 4733
-    },
-    {
-      "epoch": 2.4688396349413297,
-      "grad_norm": 1.3455497026443481,
-      "learning_rate": 9.574572864321608e-05,
-      "loss": 5.7537,
-      "step": 4734
-    },
-    {
-      "epoch": 2.469361147327249,
-      "grad_norm": 1.5775635242462158,
-      "learning_rate": 9.574472361809045e-05,
-      "loss": 5.754,
-      "step": 4735
-    },
-    {
-      "epoch": 2.4698826597131682,
-      "grad_norm": 1.4673882722854614,
-      "learning_rate": 9.574371859296483e-05,
-      "loss": 6.143,
-      "step": 4736
-    },
-    {
-      "epoch": 2.4704041720990872,
-      "grad_norm": 1.4998196363449097,
-      "learning_rate": 9.574271356783921e-05,
-      "loss": 5.946,
-      "step": 4737
-    },
-    {
-      "epoch": 2.4709256844850067,
-      "grad_norm": 1.546087384223938,
-      "learning_rate": 9.574170854271357e-05,
-      "loss": 5.8402,
-      "step": 4738
-    },
-    {
-      "epoch": 2.4714471968709257,
-      "grad_norm": 1.560832142829895,
-      "learning_rate": 9.574070351758795e-05,
-      "loss": 5.5857,
-      "step": 4739
-    },
-    {
-      "epoch": 2.4719687092568448,
-      "grad_norm": 1.7449475526809692,
-      "learning_rate": 9.573969849246231e-05,
-      "loss": 5.7353,
-      "step": 4740
-    },
-    {
-      "epoch": 2.472490221642764,
-      "grad_norm": 1.5870628356933594,
-      "learning_rate": 9.573869346733669e-05,
-      "loss": 5.5628,
-      "step": 4741
-    },
-    {
-      "epoch": 2.4730117340286832,
-      "grad_norm": 1.463445782661438,
-      "learning_rate": 9.573768844221106e-05,
-      "loss": 5.5606,
-      "step": 4742
-    },
-    {
-      "epoch": 2.4735332464146023,
-      "grad_norm": 1.6449066400527954,
-      "learning_rate": 9.573668341708542e-05,
-      "loss": 5.1979,
-      "step": 4743
-    },
-    {
-      "epoch": 2.4740547588005217,
-      "grad_norm": 1.6277861595153809,
-      "learning_rate": 9.57356783919598e-05,
-      "loss": 5.5052,
-      "step": 4744
-    },
-    {
-      "epoch": 2.4745762711864407,
-      "grad_norm": 1.63543701171875,
-      "learning_rate": 9.573467336683418e-05,
-      "loss": 5.4885,
-      "step": 4745
-    },
-    {
-      "epoch": 2.4750977835723598,
-      "grad_norm": 1.6254291534423828,
-      "learning_rate": 9.573366834170855e-05,
-      "loss": 5.6125,
-      "step": 4746
-    },
-    {
-      "epoch": 2.4756192959582792,
-      "grad_norm": 1.6313624382019043,
-      "learning_rate": 9.573266331658292e-05,
-      "loss": 5.7128,
-      "step": 4747
-    },
-    {
-      "epoch": 2.4761408083441983,
-      "grad_norm": 1.4559804201126099,
-      "learning_rate": 9.57316582914573e-05,
-      "loss": 5.7288,
-      "step": 4748
-    },
-    {
-      "epoch": 2.4766623207301173,
-      "grad_norm": 2.1019771099090576,
-      "learning_rate": 9.573065326633166e-05,
-      "loss": 5.9452,
-      "step": 4749
-    },
-    {
-      "epoch": 2.4771838331160367,
-      "grad_norm": 1.559616208076477,
-      "learning_rate": 9.572964824120604e-05,
-      "loss": 5.5625,
-      "step": 4750
-    },
-    {
-      "epoch": 2.4777053455019558,
-      "grad_norm": 1.6559454202651978,
-      "learning_rate": 9.57286432160804e-05,
-      "loss": 5.727,
-      "step": 4751
-    },
-    {
-      "epoch": 2.478226857887875,
-      "grad_norm": 1.516391634941101,
-      "learning_rate": 9.572763819095478e-05,
-      "loss": 5.9819,
-      "step": 4752
-    },
-    {
-      "epoch": 2.4787483702737942,
-      "grad_norm": 1.4842084646224976,
-      "learning_rate": 9.572663316582915e-05,
-      "loss": 5.7383,
-      "step": 4753
-    },
-    {
-      "epoch": 2.4792698826597133,
-      "grad_norm": 1.40224289894104,
-      "learning_rate": 9.572562814070352e-05,
-      "loss": 6.0331,
-      "step": 4754
-    },
-    {
-      "epoch": 2.4797913950456323,
-      "grad_norm": 1.5375807285308838,
-      "learning_rate": 9.572462311557789e-05,
-      "loss": 5.6231,
-      "step": 4755
-    },
-    {
-      "epoch": 2.4803129074315513,
-      "grad_norm": 1.5442537069320679,
-      "learning_rate": 9.572361809045227e-05,
-      "loss": 5.702,
-      "step": 4756
-    },
-    {
-      "epoch": 2.4808344198174708,
-      "grad_norm": 1.5787838697433472,
-      "learning_rate": 9.572261306532664e-05,
-      "loss": 5.6647,
-      "step": 4757
-    },
-    {
-      "epoch": 2.48135593220339,
-      "grad_norm": 1.7905805110931396,
-      "learning_rate": 9.572160804020101e-05,
-      "loss": 5.2894,
-      "step": 4758
-    },
-    {
-      "epoch": 2.481877444589309,
-      "grad_norm": 1.6314866542816162,
-      "learning_rate": 9.572060301507539e-05,
-      "loss": 5.9956,
-      "step": 4759
-    },
-    {
-      "epoch": 2.4823989569752283,
-      "grad_norm": 1.91982901096344,
-      "learning_rate": 9.571959798994975e-05,
-      "loss": 5.6837,
-      "step": 4760
-    },
-    {
-      "epoch": 2.4829204693611473,
-      "grad_norm": 1.7811248302459717,
-      "learning_rate": 9.571859296482413e-05,
-      "loss": 6.1067,
-      "step": 4761
-    },
-    {
-      "epoch": 2.4834419817470663,
-      "grad_norm": 1.5440597534179688,
-      "learning_rate": 9.571758793969849e-05,
-      "loss": 5.9178,
-      "step": 4762
-    },
-    {
-      "epoch": 2.483963494132986,
-      "grad_norm": 1.6911349296569824,
-      "learning_rate": 9.571658291457287e-05,
-      "loss": 5.6568,
-      "step": 4763
-    },
-    {
-      "epoch": 2.484485006518905,
-      "grad_norm": 1.6182178258895874,
-      "learning_rate": 9.571557788944723e-05,
-      "loss": 5.8058,
-      "step": 4764
-    },
-    {
-      "epoch": 2.485006518904824,
-      "grad_norm": 1.519123911857605,
-      "learning_rate": 9.571457286432161e-05,
-      "loss": 5.7598,
-      "step": 4765
-    },
-    {
-      "epoch": 2.4855280312907433,
-      "grad_norm": 1.5415984392166138,
-      "learning_rate": 9.571356783919599e-05,
-      "loss": 5.8972,
-      "step": 4766
-    },
-    {
-      "epoch": 2.4860495436766623,
-      "grad_norm": 1.5686134099960327,
-      "learning_rate": 9.571256281407037e-05,
-      "loss": 5.7374,
-      "step": 4767
-    },
-    {
-      "epoch": 2.4865710560625813,
-      "grad_norm": 2.31042742729187,
-      "learning_rate": 9.571155778894473e-05,
-      "loss": 5.3739,
-      "step": 4768
-    },
-    {
-      "epoch": 2.487092568448501,
-      "grad_norm": 1.6043740510940552,
-      "learning_rate": 9.57105527638191e-05,
-      "loss": 5.6459,
-      "step": 4769
-    },
-    {
-      "epoch": 2.48761408083442,
-      "grad_norm": 1.3261629343032837,
-      "learning_rate": 9.570954773869347e-05,
-      "loss": 5.8653,
-      "step": 4770
-    },
-    {
-      "epoch": 2.488135593220339,
-      "grad_norm": 1.441320776939392,
-      "learning_rate": 9.570854271356784e-05,
-      "loss": 5.873,
-      "step": 4771
-    },
-    {
-      "epoch": 2.4886571056062583,
-      "grad_norm": 1.447891354560852,
-      "learning_rate": 9.570753768844222e-05,
-      "loss": 6.0027,
-      "step": 4772
-    },
-    {
-      "epoch": 2.4891786179921773,
-      "grad_norm": 1.6774041652679443,
-      "learning_rate": 9.570653266331658e-05,
-      "loss": 5.1079,
-      "step": 4773
-    },
-    {
-      "epoch": 2.4897001303780963,
-      "grad_norm": 1.5251694917678833,
-      "learning_rate": 9.570552763819096e-05,
-      "loss": 5.8963,
-      "step": 4774
-    },
-    {
-      "epoch": 2.490221642764016,
-      "grad_norm": 1.3824256658554077,
-      "learning_rate": 9.570452261306532e-05,
-      "loss": 6.0796,
-      "step": 4775
-    },
-    {
-      "epoch": 2.490743155149935,
-      "grad_norm": 1.3858067989349365,
-      "learning_rate": 9.57035175879397e-05,
-      "loss": 5.8237,
-      "step": 4776
-    },
-    {
-      "epoch": 2.491264667535854,
-      "grad_norm": 1.51829195022583,
-      "learning_rate": 9.570251256281408e-05,
-      "loss": 5.5509,
-      "step": 4777
-    },
-    {
-      "epoch": 2.4917861799217733,
-      "grad_norm": 1.477586030960083,
-      "learning_rate": 9.570150753768846e-05,
-      "loss": 5.6836,
-      "step": 4778
-    },
-    {
-      "epoch": 2.4923076923076923,
-      "grad_norm": 1.3985092639923096,
-      "learning_rate": 9.570050251256282e-05,
-      "loss": 6.1285,
-      "step": 4779
-    },
-    {
-      "epoch": 2.4928292046936114,
-      "grad_norm": 1.4266798496246338,
-      "learning_rate": 9.56994974874372e-05,
-      "loss": 6.2102,
-      "step": 4780
-    },
-    {
-      "epoch": 2.493350717079531,
-      "grad_norm": 1.603654384613037,
-      "learning_rate": 9.569849246231156e-05,
-      "loss": 5.817,
-      "step": 4781
-    },
-    {
-      "epoch": 2.49387222946545,
-      "grad_norm": 1.3765369653701782,
-      "learning_rate": 9.569748743718593e-05,
-      "loss": 6.139,
-      "step": 4782
-    },
-    {
-      "epoch": 2.494393741851369,
-      "grad_norm": 1.5857133865356445,
-      "learning_rate": 9.56964824120603e-05,
-      "loss": 5.8695,
-      "step": 4783
-    },
-    {
-      "epoch": 2.4949152542372883,
-      "grad_norm": 1.5431194305419922,
-      "learning_rate": 9.569547738693467e-05,
-      "loss": 6.276,
-      "step": 4784
-    },
-    {
-      "epoch": 2.4954367666232073,
-      "grad_norm": 1.583506464958191,
-      "learning_rate": 9.569447236180905e-05,
-      "loss": 5.5877,
-      "step": 4785
-    },
-    {
-      "epoch": 2.4959582790091264,
-      "grad_norm": 1.578717589378357,
-      "learning_rate": 9.569346733668342e-05,
-      "loss": 5.7058,
-      "step": 4786
-    },
-    {
-      "epoch": 2.4964797913950454,
-      "grad_norm": 1.6609982252120972,
-      "learning_rate": 9.56924623115578e-05,
-      "loss": 5.4426,
-      "step": 4787
-    },
-    {
-      "epoch": 2.497001303780965,
-      "grad_norm": 1.5315263271331787,
-      "learning_rate": 9.569145728643217e-05,
-      "loss": 5.5813,
-      "step": 4788
-    },
-    {
-      "epoch": 2.497522816166884,
-      "grad_norm": 1.8720115423202515,
-      "learning_rate": 9.569045226130654e-05,
-      "loss": 5.769,
-      "step": 4789
-    },
-    {
-      "epoch": 2.498044328552803,
-      "grad_norm": 1.5439645051956177,
-      "learning_rate": 9.568944723618091e-05,
-      "loss": 6.0571,
-      "step": 4790
-    },
-    {
-      "epoch": 2.4985658409387224,
-      "grad_norm": 1.4756160974502563,
-      "learning_rate": 9.568844221105529e-05,
-      "loss": 6.154,
-      "step": 4791
-    },
-    {
-      "epoch": 2.4990873533246414,
-      "grad_norm": 1.6719236373901367,
-      "learning_rate": 9.568743718592965e-05,
-      "loss": 5.434,
-      "step": 4792
-    },
-    {
-      "epoch": 2.4996088657105604,
-      "grad_norm": 2.0609192848205566,
-      "learning_rate": 9.568643216080403e-05,
-      "loss": 5.8785,
-      "step": 4793
-    },
-    {
-      "epoch": 2.50013037809648,
-      "grad_norm": 1.601980447769165,
-      "learning_rate": 9.568542713567839e-05,
-      "loss": 5.8929,
-      "step": 4794
-    },
-    {
-      "epoch": 2.500651890482399,
-      "grad_norm": 1.697913408279419,
-      "learning_rate": 9.568442211055277e-05,
-      "loss": 5.5296,
-      "step": 4795
-    },
-    {
-      "epoch": 2.501173402868318,
-      "grad_norm": 1.605042815208435,
-      "learning_rate": 9.568341708542713e-05,
-      "loss": 5.5867,
-      "step": 4796
-    },
-    {
-      "epoch": 2.5016949152542374,
-      "grad_norm": 1.446962833404541,
-      "learning_rate": 9.568241206030151e-05,
-      "loss": 6.0315,
-      "step": 4797
-    },
-    {
-      "epoch": 2.5022164276401564,
-      "grad_norm": 1.5430487394332886,
-      "learning_rate": 9.568140703517589e-05,
-      "loss": 5.7864,
-      "step": 4798
-    },
-    {
-      "epoch": 2.5027379400260754,
-      "grad_norm": 1.6912599802017212,
-      "learning_rate": 9.568040201005025e-05,
-      "loss": 5.8151,
-      "step": 4799
-    },
-    {
-      "epoch": 2.503259452411995,
-      "grad_norm": 1.7032774686813354,
-      "learning_rate": 9.567939698492463e-05,
-      "loss": 5.856,
-      "step": 4800
-    },
-    {
-      "epoch": 2.503780964797914,
-      "grad_norm": 2.432194948196411,
-      "learning_rate": 9.5678391959799e-05,
-      "loss": 5.803,
-      "step": 4801
-    },
-    {
-      "epoch": 2.504302477183833,
-      "grad_norm": 1.4741883277893066,
-      "learning_rate": 9.567738693467337e-05,
-      "loss": 5.8889,
-      "step": 4802
-    },
-    {
-      "epoch": 2.5048239895697524,
-      "grad_norm": 1.4171416759490967,
-      "learning_rate": 9.567638190954774e-05,
-      "loss": 5.7609,
-      "step": 4803
-    },
-    {
-      "epoch": 2.5053455019556714,
-      "grad_norm": 1.400147557258606,
-      "learning_rate": 9.567537688442212e-05,
-      "loss": 5.9252,
-      "step": 4804
-    },
-    {
-      "epoch": 2.5058670143415904,
-      "grad_norm": 1.4549416303634644,
-      "learning_rate": 9.567437185929648e-05,
-      "loss": 5.7882,
-      "step": 4805
-    },
-    {
-      "epoch": 2.50638852672751,
-      "grad_norm": 1.640956163406372,
-      "learning_rate": 9.567336683417086e-05,
-      "loss": 5.6103,
-      "step": 4806
-    },
-    {
-      "epoch": 2.506910039113429,
-      "grad_norm": 1.5338428020477295,
-      "learning_rate": 9.567236180904524e-05,
-      "loss": 5.8763,
-      "step": 4807
-    },
-    {
-      "epoch": 2.507431551499348,
-      "grad_norm": 1.457040786743164,
-      "learning_rate": 9.567135678391961e-05,
-      "loss": 6.2281,
-      "step": 4808
-    },
-    {
-      "epoch": 2.5079530638852674,
-      "grad_norm": 1.5344401597976685,
-      "learning_rate": 9.567035175879398e-05,
-      "loss": 5.9402,
-      "step": 4809
-    },
-    {
-      "epoch": 2.5084745762711864,
-      "grad_norm": 1.7055158615112305,
-      "learning_rate": 9.566934673366834e-05,
-      "loss": 5.8292,
-      "step": 4810
-    },
-    {
-      "epoch": 2.5089960886571054,
-      "grad_norm": 1.4953925609588623,
-      "learning_rate": 9.566834170854272e-05,
-      "loss": 6.1041,
-      "step": 4811
-    },
-    {
-      "epoch": 2.509517601043025,
-      "grad_norm": 1.5839004516601562,
-      "learning_rate": 9.566733668341708e-05,
-      "loss": 5.8289,
-      "step": 4812
-    },
-    {
-      "epoch": 2.510039113428944,
-      "grad_norm": 1.5163946151733398,
-      "learning_rate": 9.566633165829146e-05,
-      "loss": 5.2557,
-      "step": 4813
-    },
-    {
-      "epoch": 2.510560625814863,
-      "grad_norm": 1.5256576538085938,
-      "learning_rate": 9.566532663316583e-05,
-      "loss": 6.2139,
-      "step": 4814
-    },
-    {
-      "epoch": 2.5110821382007824,
-      "grad_norm": 1.6333038806915283,
-      "learning_rate": 9.56643216080402e-05,
-      "loss": 5.7649,
-      "step": 4815
-    },
-    {
-      "epoch": 2.5116036505867014,
-      "grad_norm": 1.4789555072784424,
-      "learning_rate": 9.566331658291457e-05,
-      "loss": 5.9014,
-      "step": 4816
-    },
-    {
-      "epoch": 2.5121251629726205,
-      "grad_norm": 1.6760848760604858,
-      "learning_rate": 9.566231155778895e-05,
-      "loss": 5.7538,
-      "step": 4817
-    },
-    {
-      "epoch": 2.51264667535854,
-      "grad_norm": 1.4620388746261597,
-      "learning_rate": 9.566130653266332e-05,
-      "loss": 5.9263,
-      "step": 4818
-    },
-    {
-      "epoch": 2.513168187744459,
-      "grad_norm": 1.491621971130371,
-      "learning_rate": 9.56603015075377e-05,
-      "loss": 6.107,
-      "step": 4819
-    },
-    {
-      "epoch": 2.513689700130378,
-      "grad_norm": 1.4770426750183105,
-      "learning_rate": 9.565929648241207e-05,
-      "loss": 5.9551,
-      "step": 4820
-    },
-    {
-      "epoch": 2.5142112125162974,
-      "grad_norm": 1.3915717601776123,
-      "learning_rate": 9.565829145728644e-05,
-      "loss": 5.9464,
-      "step": 4821
-    },
-    {
-      "epoch": 2.5147327249022164,
-      "grad_norm": 1.5212647914886475,
-      "learning_rate": 9.565728643216081e-05,
-      "loss": 5.9409,
-      "step": 4822
-    },
-    {
-      "epoch": 2.5152542372881355,
-      "grad_norm": 1.5616729259490967,
-      "learning_rate": 9.565628140703517e-05,
-      "loss": 5.487,
-      "step": 4823
-    },
-    {
-      "epoch": 2.515775749674055,
-      "grad_norm": 1.6819775104522705,
-      "learning_rate": 9.565527638190955e-05,
-      "loss": 5.4623,
-      "step": 4824
-    },
-    {
-      "epoch": 2.516297262059974,
-      "grad_norm": 1.412816047668457,
-      "learning_rate": 9.565427135678392e-05,
-      "loss": 5.7539,
-      "step": 4825
-    },
-    {
-      "epoch": 2.516818774445893,
-      "grad_norm": 1.454261064529419,
-      "learning_rate": 9.565326633165829e-05,
-      "loss": 6.1526,
-      "step": 4826
-    },
-    {
-      "epoch": 2.5173402868318124,
-      "grad_norm": 1.3960659503936768,
-      "learning_rate": 9.565226130653267e-05,
-      "loss": 6.3605,
-      "step": 4827
-    },
-    {
-      "epoch": 2.5178617992177315,
-      "grad_norm": 1.6657761335372925,
-      "learning_rate": 9.565125628140705e-05,
-      "loss": 4.9502,
-      "step": 4828
-    },
-    {
-      "epoch": 2.5183833116036505,
-      "grad_norm": 1.5701698064804077,
-      "learning_rate": 9.565025125628141e-05,
-      "loss": 6.1012,
-      "step": 4829
-    },
-    {
-      "epoch": 2.51890482398957,
-      "grad_norm": 1.5692607164382935,
-      "learning_rate": 9.564924623115579e-05,
-      "loss": 5.4425,
-      "step": 4830
-    },
-    {
-      "epoch": 2.519426336375489,
-      "grad_norm": 1.587478756904602,
-      "learning_rate": 9.564824120603016e-05,
-      "loss": 5.3714,
-      "step": 4831
-    },
-    {
-      "epoch": 2.519947848761408,
-      "grad_norm": 1.453775405883789,
-      "learning_rate": 9.564723618090453e-05,
-      "loss": 6.014,
-      "step": 4832
-    },
-    {
-      "epoch": 2.5204693611473274,
-      "grad_norm": 1.4537262916564941,
-      "learning_rate": 9.56462311557789e-05,
-      "loss": 5.9701,
-      "step": 4833
-    },
-    {
-      "epoch": 2.5209908735332465,
-      "grad_norm": 1.4299412965774536,
-      "learning_rate": 9.564522613065328e-05,
-      "loss": 6.1377,
-      "step": 4834
-    },
-    {
-      "epoch": 2.5215123859191655,
-      "grad_norm": 1.6620330810546875,
-      "learning_rate": 9.564422110552764e-05,
-      "loss": 5.7877,
-      "step": 4835
-    },
-    {
-      "epoch": 2.522033898305085,
-      "grad_norm": 1.4737210273742676,
-      "learning_rate": 9.5643216080402e-05,
-      "loss": 5.9387,
-      "step": 4836
-    },
-    {
-      "epoch": 2.522555410691004,
-      "grad_norm": 1.4558378458023071,
-      "learning_rate": 9.564221105527638e-05,
-      "loss": 5.8647,
-      "step": 4837
-    },
-    {
-      "epoch": 2.523076923076923,
-      "grad_norm": 1.3782732486724854,
-      "learning_rate": 9.564120603015076e-05,
-      "loss": 5.9465,
-      "step": 4838
-    },
-    {
-      "epoch": 2.5235984354628425,
-      "grad_norm": 1.4266347885131836,
-      "learning_rate": 9.564020100502514e-05,
-      "loss": 6.1785,
-      "step": 4839
-    },
-    {
-      "epoch": 2.5241199478487615,
-      "grad_norm": 1.575951337814331,
-      "learning_rate": 9.56391959798995e-05,
-      "loss": 5.886,
-      "step": 4840
-    },
-    {
-      "epoch": 2.5246414602346805,
-      "grad_norm": 1.402287244796753,
-      "learning_rate": 9.563819095477388e-05,
-      "loss": 6.1753,
-      "step": 4841
-    },
-    {
-      "epoch": 2.5251629726206,
-      "grad_norm": 1.5526705980300903,
-      "learning_rate": 9.563718592964824e-05,
-      "loss": 5.5445,
-      "step": 4842
-    },
-    {
-      "epoch": 2.525684485006519,
-      "grad_norm": 1.9571220874786377,
-      "learning_rate": 9.563618090452262e-05,
-      "loss": 5.5883,
-      "step": 4843
-    },
-    {
-      "epoch": 2.526205997392438,
-      "grad_norm": 1.5122853517532349,
-      "learning_rate": 9.563517587939699e-05,
-      "loss": 5.6845,
-      "step": 4844
-    },
-    {
-      "epoch": 2.5267275097783575,
-      "grad_norm": 1.51768159866333,
-      "learning_rate": 9.563417085427136e-05,
-      "loss": 5.5043,
-      "step": 4845
-    },
-    {
-      "epoch": 2.5272490221642765,
-      "grad_norm": 1.6747007369995117,
-      "learning_rate": 9.563316582914573e-05,
-      "loss": 5.6175,
-      "step": 4846
-    },
-    {
-      "epoch": 2.5277705345501955,
-      "grad_norm": 1.3852779865264893,
-      "learning_rate": 9.56321608040201e-05,
-      "loss": 5.893,
-      "step": 4847
-    },
-    {
-      "epoch": 2.528292046936115,
-      "grad_norm": 1.5290780067443848,
-      "learning_rate": 9.563115577889448e-05,
-      "loss": 5.6784,
-      "step": 4848
-    },
-    {
-      "epoch": 2.528813559322034,
-      "grad_norm": 1.4161065816879272,
-      "learning_rate": 9.563015075376885e-05,
-      "loss": 5.8204,
-      "step": 4849
-    },
-    {
-      "epoch": 2.529335071707953,
-      "grad_norm": 1.4041119813919067,
-      "learning_rate": 9.562914572864323e-05,
-      "loss": 5.5141,
-      "step": 4850
-    },
-    {
-      "epoch": 2.5298565840938725,
-      "grad_norm": 1.652336597442627,
-      "learning_rate": 9.562814070351759e-05,
-      "loss": 5.8498,
-      "step": 4851
-    },
-    {
-      "epoch": 2.5303780964797915,
-      "grad_norm": 1.5477467775344849,
-      "learning_rate": 9.562713567839197e-05,
-      "loss": 5.3674,
-      "step": 4852
-    },
-    {
-      "epoch": 2.5308996088657105,
-      "grad_norm": 1.5873663425445557,
-      "learning_rate": 9.562613065326633e-05,
-      "loss": 5.9149,
-      "step": 4853
-    },
-    {
-      "epoch": 2.53142112125163,
-      "grad_norm": 1.652782917022705,
-      "learning_rate": 9.562512562814071e-05,
-      "loss": 6.099,
-      "step": 4854
-    },
-    {
-      "epoch": 2.531942633637549,
-      "grad_norm": 1.3967654705047607,
-      "learning_rate": 9.562412060301507e-05,
-      "loss": 6.0186,
-      "step": 4855
-    },
-    {
-      "epoch": 2.532464146023468,
-      "grad_norm": 1.523650050163269,
-      "learning_rate": 9.562311557788945e-05,
-      "loss": 5.6946,
-      "step": 4856
-    },
-    {
-      "epoch": 2.5329856584093875,
-      "grad_norm": 1.7817187309265137,
-      "learning_rate": 9.562211055276382e-05,
-      "loss": 5.7264,
-      "step": 4857
-    },
-    {
-      "epoch": 2.5335071707953065,
-      "grad_norm": 1.386020541191101,
-      "learning_rate": 9.56211055276382e-05,
-      "loss": 6.0123,
-      "step": 4858
-    },
-    {
-      "epoch": 2.5340286831812255,
-      "grad_norm": 1.3775678873062134,
-      "learning_rate": 9.562010050251257e-05,
-      "loss": 6.342,
-      "step": 4859
-    },
-    {
-      "epoch": 2.5345501955671446,
-      "grad_norm": 1.7185759544372559,
-      "learning_rate": 9.561909547738695e-05,
-      "loss": 5.9197,
-      "step": 4860
-    },
-    {
-      "epoch": 2.535071707953064,
-      "grad_norm": 1.3208074569702148,
-      "learning_rate": 9.561809045226131e-05,
-      "loss": 6.4258,
-      "step": 4861
-    },
-    {
-      "epoch": 2.535593220338983,
-      "grad_norm": 1.3878352642059326,
-      "learning_rate": 9.561708542713568e-05,
-      "loss": 5.7574,
-      "step": 4862
-    },
-    {
-      "epoch": 2.536114732724902,
-      "grad_norm": 1.3933541774749756,
-      "learning_rate": 9.561608040201006e-05,
-      "loss": 5.9771,
-      "step": 4863
-    },
-    {
-      "epoch": 2.5366362451108215,
-      "grad_norm": 1.628386378288269,
-      "learning_rate": 9.561507537688442e-05,
-      "loss": 5.9234,
-      "step": 4864
-    },
-    {
-      "epoch": 2.5371577574967406,
-      "grad_norm": 1.4690829515457153,
-      "learning_rate": 9.56140703517588e-05,
-      "loss": 5.7361,
-      "step": 4865
-    },
-    {
-      "epoch": 2.5376792698826596,
-      "grad_norm": 1.6475598812103271,
-      "learning_rate": 9.561306532663316e-05,
-      "loss": 5.643,
-      "step": 4866
-    },
-    {
-      "epoch": 2.538200782268579,
-      "grad_norm": 1.423583745956421,
-      "learning_rate": 9.561206030150754e-05,
-      "loss": 5.9888,
-      "step": 4867
-    },
-    {
-      "epoch": 2.538722294654498,
-      "grad_norm": 1.5557703971862793,
-      "learning_rate": 9.561105527638192e-05,
-      "loss": 5.5081,
-      "step": 4868
-    },
-    {
-      "epoch": 2.539243807040417,
-      "grad_norm": 1.4079763889312744,
-      "learning_rate": 9.56100502512563e-05,
-      "loss": 5.9362,
-      "step": 4869
-    },
-    {
-      "epoch": 2.539765319426336,
-      "grad_norm": 1.3977445363998413,
-      "learning_rate": 9.560904522613066e-05,
-      "loss": 5.9808,
-      "step": 4870
-    },
-    {
-      "epoch": 2.5402868318122556,
-      "grad_norm": 1.5023306608200073,
-      "learning_rate": 9.560804020100504e-05,
-      "loss": 5.756,
-      "step": 4871
-    },
-    {
-      "epoch": 2.5408083441981746,
-      "grad_norm": 1.4479445219039917,
-      "learning_rate": 9.56070351758794e-05,
-      "loss": 5.8998,
-      "step": 4872
-    },
-    {
-      "epoch": 2.5413298565840936,
-      "grad_norm": 1.3944220542907715,
-      "learning_rate": 9.560603015075378e-05,
-      "loss": 5.8181,
-      "step": 4873
-    },
-    {
-      "epoch": 2.541851368970013,
-      "grad_norm": 1.6817361116409302,
-      "learning_rate": 9.560502512562814e-05,
-      "loss": 5.5604,
-      "step": 4874
-    },
-    {
-      "epoch": 2.542372881355932,
-      "grad_norm": 1.4943729639053345,
-      "learning_rate": 9.560402010050252e-05,
-      "loss": 6.3862,
-      "step": 4875
-    },
-    {
-      "epoch": 2.542894393741851,
-      "grad_norm": 1.6542384624481201,
-      "learning_rate": 9.560301507537689e-05,
-      "loss": 5.9109,
-      "step": 4876
-    },
-    {
-      "epoch": 2.5434159061277706,
-      "grad_norm": 1.7097663879394531,
-      "learning_rate": 9.560201005025125e-05,
-      "loss": 5.5542,
-      "step": 4877
-    },
-    {
-      "epoch": 2.5439374185136896,
-      "grad_norm": 1.5016015768051147,
-      "learning_rate": 9.560100502512563e-05,
-      "loss": 5.5871,
-      "step": 4878
-    },
-    {
-      "epoch": 2.5444589308996086,
-      "grad_norm": 1.5243148803710938,
-      "learning_rate": 9.56e-05,
-      "loss": 6.1649,
-      "step": 4879
-    },
-    {
-      "epoch": 2.544980443285528,
-      "grad_norm": 1.3934766054153442,
-      "learning_rate": 9.559899497487438e-05,
-      "loss": 6.0201,
-      "step": 4880
-    },
-    {
-      "epoch": 2.545501955671447,
-      "grad_norm": 1.353205919265747,
-      "learning_rate": 9.559798994974875e-05,
-      "loss": 6.1229,
-      "step": 4881
-    },
-    {
-      "epoch": 2.546023468057366,
-      "grad_norm": 1.8142156600952148,
-      "learning_rate": 9.559698492462313e-05,
-      "loss": 5.572,
-      "step": 4882
-    },
-    {
-      "epoch": 2.5465449804432856,
-      "grad_norm": 1.532049298286438,
-      "learning_rate": 9.559597989949749e-05,
-      "loss": 5.893,
-      "step": 4883
-    },
-    {
-      "epoch": 2.5470664928292046,
-      "grad_norm": 1.498363733291626,
-      "learning_rate": 9.559497487437187e-05,
-      "loss": 5.809,
-      "step": 4884
-    },
-    {
-      "epoch": 2.5475880052151236,
-      "grad_norm": 1.379697322845459,
-      "learning_rate": 9.559396984924623e-05,
-      "loss": 5.6439,
-      "step": 4885
-    },
-    {
-      "epoch": 2.548109517601043,
-      "grad_norm": 1.4061715602874756,
-      "learning_rate": 9.559296482412061e-05,
-      "loss": 5.8238,
-      "step": 4886
-    },
-    {
-      "epoch": 2.548631029986962,
-      "grad_norm": 1.4150760173797607,
-      "learning_rate": 9.559195979899497e-05,
-      "loss": 5.9641,
-      "step": 4887
-    },
-    {
-      "epoch": 2.549152542372881,
-      "grad_norm": 1.4395184516906738,
-      "learning_rate": 9.559095477386935e-05,
-      "loss": 5.2822,
-      "step": 4888
-    },
-    {
-      "epoch": 2.5496740547588006,
-      "grad_norm": 1.5797990560531616,
-      "learning_rate": 9.558994974874372e-05,
-      "loss": 5.9153,
-      "step": 4889
-    },
-    {
-      "epoch": 2.5501955671447196,
-      "grad_norm": 1.3622783422470093,
-      "learning_rate": 9.55889447236181e-05,
-      "loss": 6.1547,
-      "step": 4890
-    },
-    {
-      "epoch": 2.5507170795306386,
-      "grad_norm": 1.7607394456863403,
-      "learning_rate": 9.558793969849247e-05,
-      "loss": 5.386,
-      "step": 4891
-    },
-    {
-      "epoch": 2.551238591916558,
-      "grad_norm": 1.645982027053833,
-      "learning_rate": 9.558693467336684e-05,
-      "loss": 5.7626,
-      "step": 4892
-    },
-    {
-      "epoch": 2.551760104302477,
-      "grad_norm": 1.4103668928146362,
-      "learning_rate": 9.558592964824121e-05,
-      "loss": 5.9749,
-      "step": 4893
-    },
-    {
-      "epoch": 2.552281616688396,
-      "grad_norm": 1.505454659461975,
-      "learning_rate": 9.558492462311558e-05,
-      "loss": 5.9183,
-      "step": 4894
-    },
-    {
-      "epoch": 2.5528031290743156,
-      "grad_norm": 1.3817424774169922,
-      "learning_rate": 9.558391959798996e-05,
-      "loss": 6.1864,
-      "step": 4895
-    },
-    {
-      "epoch": 2.5533246414602346,
-      "grad_norm": 1.4669145345687866,
-      "learning_rate": 9.558291457286432e-05,
-      "loss": 6.0256,
-      "step": 4896
-    },
-    {
-      "epoch": 2.5538461538461537,
-      "grad_norm": 1.3954702615737915,
-      "learning_rate": 9.55819095477387e-05,
-      "loss": 6.2333,
-      "step": 4897
-    },
-    {
-      "epoch": 2.554367666232073,
-      "grad_norm": 1.648632526397705,
-      "learning_rate": 9.558090452261306e-05,
-      "loss": 5.8438,
-      "step": 4898
-    },
-    {
-      "epoch": 2.554889178617992,
-      "grad_norm": 1.800253987312317,
-      "learning_rate": 9.557989949748744e-05,
-      "loss": 5.2236,
-      "step": 4899
-    },
-    {
-      "epoch": 2.555410691003911,
-      "grad_norm": 1.7411351203918457,
-      "learning_rate": 9.557889447236182e-05,
-      "loss": 5.938,
-      "step": 4900
-    },
-    {
-      "epoch": 2.5559322033898306,
-      "grad_norm": 1.5227670669555664,
-      "learning_rate": 9.55778894472362e-05,
-      "loss": 5.787,
-      "step": 4901
-    },
-    {
-      "epoch": 2.5564537157757496,
-      "grad_norm": 1.5491819381713867,
-      "learning_rate": 9.557688442211056e-05,
-      "loss": 5.7506,
-      "step": 4902
-    },
-    {
-      "epoch": 2.5569752281616687,
-      "grad_norm": 1.4868522882461548,
-      "learning_rate": 9.557587939698493e-05,
-      "loss": 5.6488,
-      "step": 4903
-    },
-    {
-      "epoch": 2.557496740547588,
-      "grad_norm": 1.8262728452682495,
-      "learning_rate": 9.55748743718593e-05,
-      "loss": 5.7859,
-      "step": 4904
-    },
-    {
-      "epoch": 2.558018252933507,
-      "grad_norm": 1.544114112854004,
-      "learning_rate": 9.557386934673367e-05,
-      "loss": 5.5315,
-      "step": 4905
-    },
-    {
-      "epoch": 2.558539765319426,
-      "grad_norm": 1.5082577466964722,
-      "learning_rate": 9.557286432160804e-05,
-      "loss": 5.6623,
-      "step": 4906
-    },
-    {
-      "epoch": 2.5590612777053456,
-      "grad_norm": 1.4401352405548096,
-      "learning_rate": 9.557185929648241e-05,
-      "loss": 6.0024,
-      "step": 4907
-    },
-    {
-      "epoch": 2.5595827900912647,
-      "grad_norm": 1.5661818981170654,
-      "learning_rate": 9.557085427135679e-05,
-      "loss": 5.4126,
-      "step": 4908
-    },
-    {
-      "epoch": 2.5601043024771837,
-      "grad_norm": 1.4719387292861938,
-      "learning_rate": 9.556984924623115e-05,
-      "loss": 5.7271,
-      "step": 4909
-    },
-    {
-      "epoch": 2.560625814863103,
-      "grad_norm": 1.5360444784164429,
-      "learning_rate": 9.556884422110553e-05,
-      "loss": 5.9958,
-      "step": 4910
-    },
-    {
-      "epoch": 2.561147327249022,
-      "grad_norm": 1.5458017587661743,
-      "learning_rate": 9.556783919597991e-05,
-      "loss": 5.9146,
-      "step": 4911
-    },
-    {
-      "epoch": 2.561668839634941,
-      "grad_norm": 1.454142689704895,
-      "learning_rate": 9.556683417085428e-05,
-      "loss": 5.6069,
-      "step": 4912
-    },
-    {
-      "epoch": 2.5621903520208607,
-      "grad_norm": 1.5150444507598877,
-      "learning_rate": 9.556582914572865e-05,
-      "loss": 6.0435,
-      "step": 4913
-    },
-    {
-      "epoch": 2.5627118644067797,
-      "grad_norm": 1.492445707321167,
-      "learning_rate": 9.556482412060303e-05,
-      "loss": 5.8739,
-      "step": 4914
-    },
-    {
-      "epoch": 2.5632333767926987,
-      "grad_norm": 1.5032119750976562,
-      "learning_rate": 9.556381909547739e-05,
-      "loss": 5.6988,
-      "step": 4915
-    },
-    {
-      "epoch": 2.563754889178618,
-      "grad_norm": 1.543714165687561,
-      "learning_rate": 9.556281407035176e-05,
-      "loss": 5.8963,
-      "step": 4916
-    },
-    {
-      "epoch": 2.564276401564537,
-      "grad_norm": 1.3285914659500122,
-      "learning_rate": 9.556180904522613e-05,
-      "loss": 6.3071,
-      "step": 4917
-    },
-    {
-      "epoch": 2.564797913950456,
-      "grad_norm": 1.9831669330596924,
-      "learning_rate": 9.55608040201005e-05,
-      "loss": 4.7935,
-      "step": 4918
-    },
-    {
-      "epoch": 2.5653194263363757,
-      "grad_norm": 1.4202433824539185,
-      "learning_rate": 9.555979899497488e-05,
-      "loss": 6.1302,
-      "step": 4919
-    },
-    {
-      "epoch": 2.5658409387222947,
-      "grad_norm": 1.6701037883758545,
-      "learning_rate": 9.555879396984925e-05,
-      "loss": 5.6922,
-      "step": 4920
-    },
-    {
-      "epoch": 2.5663624511082137,
-      "grad_norm": 1.4507468938827515,
-      "learning_rate": 9.555778894472363e-05,
-      "loss": 6.1946,
-      "step": 4921
-    },
-    {
-      "epoch": 2.566883963494133,
-      "grad_norm": 1.585044026374817,
-      "learning_rate": 9.5556783919598e-05,
-      "loss": 5.6405,
-      "step": 4922
-    },
-    {
-      "epoch": 2.567405475880052,
-      "grad_norm": 1.5023295879364014,
-      "learning_rate": 9.555577889447237e-05,
-      "loss": 5.7222,
-      "step": 4923
-    },
-    {
-      "epoch": 2.567926988265971,
-      "grad_norm": 1.506818413734436,
-      "learning_rate": 9.555477386934674e-05,
-      "loss": 6.2653,
-      "step": 4924
-    },
-    {
-      "epoch": 2.5684485006518907,
-      "grad_norm": 1.4110885858535767,
-      "learning_rate": 9.555376884422112e-05,
-      "loss": 6.1549,
-      "step": 4925
-    },
-    {
-      "epoch": 2.5689700130378097,
-      "grad_norm": 1.5171003341674805,
-      "learning_rate": 9.555276381909548e-05,
-      "loss": 6.171,
-      "step": 4926
-    },
-    {
-      "epoch": 2.5694915254237287,
-      "grad_norm": 1.512742042541504,
-      "learning_rate": 9.555175879396986e-05,
-      "loss": 5.8396,
-      "step": 4927
-    },
-    {
-      "epoch": 2.570013037809648,
-      "grad_norm": 1.3136080503463745,
-      "learning_rate": 9.555075376884422e-05,
-      "loss": 6.2355,
-      "step": 4928
-    },
-    {
-      "epoch": 2.570534550195567,
-      "grad_norm": 1.3885431289672852,
-      "learning_rate": 9.554974874371859e-05,
-      "loss": 6.1975,
-      "step": 4929
-    },
-    {
-      "epoch": 2.5710560625814862,
-      "grad_norm": 1.5574791431427002,
-      "learning_rate": 9.554874371859296e-05,
-      "loss": 5.9415,
-      "step": 4930
-    },
-    {
-      "epoch": 2.5715775749674057,
-      "grad_norm": 1.5767273902893066,
-      "learning_rate": 9.554773869346734e-05,
-      "loss": 5.2698,
-      "step": 4931
-    },
-    {
-      "epoch": 2.5720990873533247,
-      "grad_norm": 1.4684633016586304,
-      "learning_rate": 9.554673366834172e-05,
-      "loss": 5.976,
-      "step": 4932
-    },
-    {
-      "epoch": 2.5726205997392437,
-      "grad_norm": 1.5518721342086792,
-      "learning_rate": 9.554572864321608e-05,
-      "loss": 5.7763,
-      "step": 4933
-    },
-    {
-      "epoch": 2.573142112125163,
-      "grad_norm": 1.42588472366333,
-      "learning_rate": 9.554472361809046e-05,
-      "loss": 6.045,
-      "step": 4934
-    },
-    {
-      "epoch": 2.573663624511082,
-      "grad_norm": 1.3854042291641235,
-      "learning_rate": 9.554371859296483e-05,
-      "loss": 6.1132,
-      "step": 4935
-    },
-    {
-      "epoch": 2.5741851368970012,
-      "grad_norm": 1.5302084684371948,
-      "learning_rate": 9.55427135678392e-05,
-      "loss": 5.8721,
-      "step": 4936
-    },
-    {
-      "epoch": 2.5747066492829207,
-      "grad_norm": 1.9314842224121094,
-      "learning_rate": 9.554170854271357e-05,
-      "loss": 5.1593,
-      "step": 4937
-    },
-    {
-      "epoch": 2.5752281616688397,
-      "grad_norm": 1.5143386125564575,
-      "learning_rate": 9.554070351758795e-05,
-      "loss": 6.1635,
-      "step": 4938
-    },
-    {
-      "epoch": 2.5757496740547587,
-      "grad_norm": 1.7665318250656128,
-      "learning_rate": 9.553969849246231e-05,
-      "loss": 5.8289,
-      "step": 4939
-    },
-    {
-      "epoch": 2.576271186440678,
-      "grad_norm": 1.4895192384719849,
-      "learning_rate": 9.553869346733669e-05,
-      "loss": 5.7266,
-      "step": 4940
-    },
-    {
-      "epoch": 2.5767926988265972,
-      "grad_norm": 1.491462230682373,
-      "learning_rate": 9.553768844221107e-05,
-      "loss": 6.2341,
-      "step": 4941
-    },
-    {
-      "epoch": 2.5773142112125162,
-      "grad_norm": 1.443203091621399,
-      "learning_rate": 9.553668341708543e-05,
-      "loss": 6.0857,
-      "step": 4942
-    },
-    {
-      "epoch": 2.5778357235984357,
-      "grad_norm": 1.5826008319854736,
-      "learning_rate": 9.553567839195981e-05,
-      "loss": 5.712,
-      "step": 4943
-    },
-    {
-      "epoch": 2.5783572359843547,
-      "grad_norm": 1.6070513725280762,
-      "learning_rate": 9.553467336683417e-05,
-      "loss": 5.5146,
-      "step": 4944
-    },
-    {
-      "epoch": 2.5788787483702738,
-      "grad_norm": 1.7267638444900513,
-      "learning_rate": 9.553366834170855e-05,
-      "loss": 5.3383,
-      "step": 4945
-    },
-    {
-      "epoch": 2.579400260756193,
-      "grad_norm": 1.65509831905365,
-      "learning_rate": 9.553266331658291e-05,
-      "loss": 5.8969,
-      "step": 4946
-    },
-    {
-      "epoch": 2.5799217731421122,
-      "grad_norm": 1.5705424547195435,
-      "learning_rate": 9.553165829145729e-05,
-      "loss": 6.2231,
-      "step": 4947
-    },
-    {
-      "epoch": 2.5804432855280313,
-      "grad_norm": 1.3533644676208496,
-      "learning_rate": 9.553065326633166e-05,
-      "loss": 6.1414,
-      "step": 4948
-    },
-    {
-      "epoch": 2.5809647979139507,
-      "grad_norm": 1.3411778211593628,
-      "learning_rate": 9.552964824120603e-05,
-      "loss": 5.8285,
-      "step": 4949
-    },
-    {
-      "epoch": 2.5814863102998697,
-      "grad_norm": 1.6119413375854492,
-      "learning_rate": 9.55286432160804e-05,
-      "loss": 5.6827,
-      "step": 4950
-    },
-    {
-      "epoch": 2.5820078226857888,
-      "grad_norm": 1.5210593938827515,
-      "learning_rate": 9.552763819095478e-05,
-      "loss": 5.9937,
-      "step": 4951
-    },
-    {
-      "epoch": 2.582529335071708,
-      "grad_norm": 1.3873943090438843,
-      "learning_rate": 9.552663316582915e-05,
-      "loss": 6.1973,
-      "step": 4952
-    },
-    {
-      "epoch": 2.5830508474576273,
-      "grad_norm": 1.7004578113555908,
-      "learning_rate": 9.552562814070353e-05,
-      "loss": 5.7198,
-      "step": 4953
-    },
-    {
-      "epoch": 2.5835723598435463,
-      "grad_norm": 1.4841238260269165,
-      "learning_rate": 9.55246231155779e-05,
-      "loss": 5.9674,
-      "step": 4954
-    },
-    {
-      "epoch": 2.5840938722294653,
-      "grad_norm": 1.6378461122512817,
-      "learning_rate": 9.552361809045226e-05,
-      "loss": 5.6279,
-      "step": 4955
-    },
-    {
-      "epoch": 2.5846153846153848,
-      "grad_norm": 1.4548827409744263,
-      "learning_rate": 9.552261306532664e-05,
-      "loss": 5.8073,
-      "step": 4956
-    },
-    {
-      "epoch": 2.585136897001304,
-      "grad_norm": 1.4955155849456787,
-      "learning_rate": 9.5521608040201e-05,
-      "loss": 6.1286,
-      "step": 4957
-    },
-    {
-      "epoch": 2.585658409387223,
-      "grad_norm": 1.456142783164978,
-      "learning_rate": 9.552060301507538e-05,
-      "loss": 6.2438,
-      "step": 4958
-    },
-    {
-      "epoch": 2.5861799217731423,
-      "grad_norm": 1.8174327611923218,
-      "learning_rate": 9.551959798994974e-05,
-      "loss": 5.1813,
-      "step": 4959
-    },
-    {
-      "epoch": 2.5867014341590613,
-      "grad_norm": 1.562380075454712,
-      "learning_rate": 9.551859296482412e-05,
-      "loss": 5.9697,
-      "step": 4960
-    },
-    {
-      "epoch": 2.5872229465449803,
-      "grad_norm": 1.456675410270691,
-      "learning_rate": 9.55175879396985e-05,
-      "loss": 5.9005,
-      "step": 4961
-    },
-    {
-      "epoch": 2.5877444589308998,
-      "grad_norm": 1.533780574798584,
-      "learning_rate": 9.551658291457288e-05,
-      "loss": 5.5308,
-      "step": 4962
-    },
-    {
-      "epoch": 2.588265971316819,
-      "grad_norm": 1.4614955186843872,
-      "learning_rate": 9.551557788944724e-05,
-      "loss": 6.1012,
-      "step": 4963
-    },
-    {
-      "epoch": 2.588787483702738,
-      "grad_norm": 1.6414686441421509,
-      "learning_rate": 9.551457286432162e-05,
-      "loss": 5.7823,
-      "step": 4964
-    },
-    {
-      "epoch": 2.589308996088657,
-      "grad_norm": 1.6913644075393677,
-      "learning_rate": 9.551356783919598e-05,
-      "loss": 6.14,
-      "step": 4965
-    },
-    {
-      "epoch": 2.5898305084745763,
-      "grad_norm": 1.6518419981002808,
-      "learning_rate": 9.551256281407036e-05,
-      "loss": 5.7146,
-      "step": 4966
-    },
-    {
-      "epoch": 2.5903520208604953,
-      "grad_norm": 1.5552659034729004,
-      "learning_rate": 9.551155778894473e-05,
-      "loss": 6.2022,
-      "step": 4967
-    },
-    {
-      "epoch": 2.5908735332464143,
-      "grad_norm": 1.525368571281433,
-      "learning_rate": 9.55105527638191e-05,
-      "loss": 5.6289,
-      "step": 4968
-    },
-    {
-      "epoch": 2.591395045632334,
-      "grad_norm": 1.6114280223846436,
-      "learning_rate": 9.550954773869347e-05,
-      "loss": 5.5026,
-      "step": 4969
-    },
-    {
-      "epoch": 2.591916558018253,
-      "grad_norm": 1.5574963092803955,
-      "learning_rate": 9.550854271356783e-05,
-      "loss": 5.8758,
-      "step": 4970
-    },
-    {
-      "epoch": 2.592438070404172,
-      "grad_norm": 1.4775002002716064,
-      "learning_rate": 9.550753768844221e-05,
-      "loss": 5.7687,
-      "step": 4971
-    },
-    {
-      "epoch": 2.5929595827900913,
-      "grad_norm": 1.3690000772476196,
-      "learning_rate": 9.550653266331659e-05,
-      "loss": 6.1546,
-      "step": 4972
-    },
-    {
-      "epoch": 2.5934810951760103,
-      "grad_norm": 1.43038010597229,
-      "learning_rate": 9.550552763819097e-05,
-      "loss": 6.0218,
-      "step": 4973
-    },
-    {
-      "epoch": 2.5940026075619294,
-      "grad_norm": 1.4997000694274902,
-      "learning_rate": 9.550452261306533e-05,
-      "loss": 5.9948,
-      "step": 4974
-    },
-    {
-      "epoch": 2.594524119947849,
-      "grad_norm": 1.4385737180709839,
-      "learning_rate": 9.550351758793971e-05,
-      "loss": 6.0313,
-      "step": 4975
-    },
-    {
-      "epoch": 2.595045632333768,
-      "grad_norm": 1.5429884195327759,
-      "learning_rate": 9.550251256281407e-05,
-      "loss": 6.1105,
-      "step": 4976
-    },
-    {
-      "epoch": 2.595567144719687,
-      "grad_norm": 1.461922526359558,
-      "learning_rate": 9.550150753768845e-05,
-      "loss": 6.0955,
-      "step": 4977
-    },
-    {
-      "epoch": 2.5960886571056063,
-      "grad_norm": 1.4797965288162231,
-      "learning_rate": 9.550050251256281e-05,
-      "loss": 5.7342,
-      "step": 4978
-    },
-    {
-      "epoch": 2.5966101694915253,
-      "grad_norm": 1.434741497039795,
-      "learning_rate": 9.549949748743719e-05,
-      "loss": 6.1408,
-      "step": 4979
-    },
-    {
-      "epoch": 2.5971316818774444,
-      "grad_norm": 1.3131860494613647,
-      "learning_rate": 9.549849246231156e-05,
-      "loss": 6.1472,
-      "step": 4980
-    },
-    {
-      "epoch": 2.597653194263364,
-      "grad_norm": 1.3303923606872559,
-      "learning_rate": 9.549748743718593e-05,
-      "loss": 5.994,
-      "step": 4981
-    },
-    {
-      "epoch": 2.598174706649283,
-      "grad_norm": 1.5733730792999268,
-      "learning_rate": 9.549648241206031e-05,
-      "loss": 5.9427,
-      "step": 4982
-    },
-    {
-      "epoch": 2.598696219035202,
-      "grad_norm": 1.4695848226547241,
-      "learning_rate": 9.549547738693468e-05,
-      "loss": 5.7732,
-      "step": 4983
-    },
-    {
-      "epoch": 2.5992177314211213,
-      "grad_norm": 1.4286550283432007,
-      "learning_rate": 9.549447236180905e-05,
-      "loss": 5.9407,
-      "step": 4984
-    },
-    {
-      "epoch": 2.5997392438070404,
-      "grad_norm": 1.4792206287384033,
-      "learning_rate": 9.549346733668342e-05,
-      "loss": 5.8254,
-      "step": 4985
-    },
-    {
-      "epoch": 2.6002607561929594,
-      "grad_norm": 1.4887049198150635,
-      "learning_rate": 9.54924623115578e-05,
-      "loss": 6.0146,
-      "step": 4986
-    },
-    {
-      "epoch": 2.600782268578879,
-      "grad_norm": 1.3689192533493042,
-      "learning_rate": 9.549145728643216e-05,
-      "loss": 5.6423,
-      "step": 4987
-    },
-    {
-      "epoch": 2.601303780964798,
-      "grad_norm": 1.4413626194000244,
-      "learning_rate": 9.549045226130654e-05,
-      "loss": 5.9723,
-      "step": 4988
-    },
-    {
-      "epoch": 2.601825293350717,
-      "grad_norm": 1.3952202796936035,
-      "learning_rate": 9.54894472361809e-05,
-      "loss": 5.8967,
-      "step": 4989
-    },
-    {
-      "epoch": 2.6023468057366363,
-      "grad_norm": 1.548154354095459,
-      "learning_rate": 9.548844221105528e-05,
-      "loss": 5.8004,
-      "step": 4990
-    },
-    {
-      "epoch": 2.6028683181225554,
-      "grad_norm": 1.4952946901321411,
-      "learning_rate": 9.548743718592965e-05,
-      "loss": 5.6286,
-      "step": 4991
-    },
-    {
-      "epoch": 2.6033898305084744,
-      "grad_norm": 1.5337224006652832,
-      "learning_rate": 9.548643216080402e-05,
-      "loss": 5.627,
-      "step": 4992
-    },
-    {
-      "epoch": 2.603911342894394,
-      "grad_norm": 1.517376184463501,
-      "learning_rate": 9.54854271356784e-05,
-      "loss": 6.2875,
-      "step": 4993
-    },
-    {
-      "epoch": 2.604432855280313,
-      "grad_norm": 1.521404504776001,
-      "learning_rate": 9.548442211055278e-05,
-      "loss": 6.0356,
-      "step": 4994
-    },
-    {
-      "epoch": 2.604954367666232,
-      "grad_norm": 1.956773042678833,
-      "learning_rate": 9.548341708542714e-05,
-      "loss": 5.4372,
-      "step": 4995
-    },
-    {
-      "epoch": 2.6054758800521514,
-      "grad_norm": 1.4439547061920166,
-      "learning_rate": 9.548241206030151e-05,
-      "loss": 6.24,
-      "step": 4996
-    },
-    {
-      "epoch": 2.6059973924380704,
-      "grad_norm": 1.5614346265792847,
-      "learning_rate": 9.548140703517589e-05,
-      "loss": 5.7138,
-      "step": 4997
-    },
-    {
-      "epoch": 2.6065189048239894,
-      "grad_norm": 1.5995956659317017,
-      "learning_rate": 9.548040201005025e-05,
-      "loss": 5.7402,
-      "step": 4998
-    },
-    {
-      "epoch": 2.607040417209909,
-      "grad_norm": 1.6731740236282349,
-      "learning_rate": 9.547939698492463e-05,
-      "loss": 5.6629,
-      "step": 4999
-    },
-    {
-      "epoch": 2.607561929595828,
-      "grad_norm": 1.6680164337158203,
-      "learning_rate": 9.547839195979899e-05,
-      "loss": 5.5758,
-      "step": 5000
-    },
-    {
-      "epoch": 2.607561929595828,
-      "eval_loss": 5.870407581329346,
-      "eval_runtime": 42.6912,
-      "eval_samples_per_second": 28.718,
-      "eval_steps_per_second": 3.607,
-      "step": 5000
-    },
-    {
-      "epoch": 2.608083441981747,
-      "grad_norm": 1.472576379776001,
-      "learning_rate": 9.547738693467337e-05,
-      "loss": 5.9558,
-      "step": 5001
-    },
-    {
-      "epoch": 2.6086049543676664,
-      "grad_norm": 1.6486186981201172,
-      "learning_rate": 9.547638190954775e-05,
-      "loss": 5.9835,
-      "step": 5002
-    },
-    {
-      "epoch": 2.6091264667535854,
-      "grad_norm": 1.7186744213104248,
-      "learning_rate": 9.547537688442213e-05,
-      "loss": 5.8526,
-      "step": 5003
-    },
-    {
-      "epoch": 2.6096479791395044,
-      "grad_norm": 1.8182179927825928,
-      "learning_rate": 9.547437185929649e-05,
-      "loss": 5.8459,
-      "step": 5004
-    },
-    {
-      "epoch": 2.610169491525424,
-      "grad_norm": 1.5836211442947388,
-      "learning_rate": 9.547336683417087e-05,
-      "loss": 5.7555,
-      "step": 5005
-    },
-    {
-      "epoch": 2.610691003911343,
-      "grad_norm": 1.366097092628479,
-      "learning_rate": 9.547236180904523e-05,
-      "loss": 5.4636,
-      "step": 5006
-    },
-    {
-      "epoch": 2.611212516297262,
-      "grad_norm": 1.4980669021606445,
-      "learning_rate": 9.547135678391961e-05,
-      "loss": 5.9617,
-      "step": 5007
-    },
-    {
-      "epoch": 2.6117340286831814,
-      "grad_norm": 1.5618220567703247,
-      "learning_rate": 9.547035175879397e-05,
-      "loss": 5.6871,
-      "step": 5008
-    },
-    {
-      "epoch": 2.6122555410691004,
-      "grad_norm": 1.6605079174041748,
-      "learning_rate": 9.546934673366834e-05,
-      "loss": 5.9256,
-      "step": 5009
-    },
-    {
-      "epoch": 2.6127770534550194,
-      "grad_norm": 1.5125423669815063,
-      "learning_rate": 9.546834170854272e-05,
-      "loss": 6.1618,
-      "step": 5010
-    },
-    {
-      "epoch": 2.613298565840939,
-      "grad_norm": 1.5087411403656006,
-      "learning_rate": 9.546733668341708e-05,
-      "loss": 5.7258,
-      "step": 5011
-    },
-    {
-      "epoch": 2.613820078226858,
-      "grad_norm": 1.5925954580307007,
-      "learning_rate": 9.546633165829146e-05,
-      "loss": 6.0616,
-      "step": 5012
-    },
-    {
-      "epoch": 2.614341590612777,
-      "grad_norm": 1.599152684211731,
-      "learning_rate": 9.546532663316584e-05,
-      "loss": 5.9506,
-      "step": 5013
-    },
-    {
-      "epoch": 2.6148631029986964,
-      "grad_norm": 1.51369309425354,
-      "learning_rate": 9.546432160804021e-05,
-      "loss": 6.0551,
-      "step": 5014
-    },
-    {
-      "epoch": 2.6153846153846154,
-      "grad_norm": 1.5197774171829224,
-      "learning_rate": 9.546331658291458e-05,
-      "loss": 5.7489,
-      "step": 5015
-    },
-    {
-      "epoch": 2.6159061277705344,
-      "grad_norm": 1.4035998582839966,
-      "learning_rate": 9.546231155778896e-05,
-      "loss": 6.1557,
-      "step": 5016
-    },
-    {
-      "epoch": 2.616427640156454,
-      "grad_norm": 1.3745861053466797,
-      "learning_rate": 9.546130653266332e-05,
-      "loss": 5.994,
-      "step": 5017
-    },
-    {
-      "epoch": 2.616949152542373,
-      "grad_norm": 1.6477463245391846,
-      "learning_rate": 9.54603015075377e-05,
-      "loss": 5.7367,
-      "step": 5018
-    },
-    {
-      "epoch": 2.617470664928292,
-      "grad_norm": 1.6669809818267822,
-      "learning_rate": 9.545929648241206e-05,
-      "loss": 5.7648,
-      "step": 5019
-    },
-    {
-      "epoch": 2.6179921773142114,
-      "grad_norm": 1.3900141716003418,
-      "learning_rate": 9.545829145728644e-05,
-      "loss": 5.9756,
-      "step": 5020
-    },
-    {
-      "epoch": 2.6185136897001304,
-      "grad_norm": 1.6460955142974854,
-      "learning_rate": 9.54572864321608e-05,
-      "loss": 5.465,
-      "step": 5021
-    },
-    {
-      "epoch": 2.6190352020860495,
-      "grad_norm": 1.5359174013137817,
-      "learning_rate": 9.545628140703518e-05,
-      "loss": 5.9153,
-      "step": 5022
-    },
-    {
-      "epoch": 2.619556714471969,
-      "grad_norm": 1.3302761316299438,
-      "learning_rate": 9.545527638190956e-05,
-      "loss": 6.1481,
-      "step": 5023
-    },
-    {
-      "epoch": 2.620078226857888,
-      "grad_norm": 1.5097354650497437,
-      "learning_rate": 9.545427135678392e-05,
-      "loss": 5.9412,
-      "step": 5024
-    },
-    {
-      "epoch": 2.620599739243807,
-      "grad_norm": 1.4498836994171143,
-      "learning_rate": 9.54532663316583e-05,
-      "loss": 6.0881,
-      "step": 5025
-    },
-    {
-      "epoch": 2.6211212516297264,
-      "grad_norm": 1.3311947584152222,
-      "learning_rate": 9.545226130653267e-05,
-      "loss": 6.0251,
-      "step": 5026
-    },
-    {
-      "epoch": 2.6216427640156454,
-      "grad_norm": 1.3834559917449951,
-      "learning_rate": 9.545125628140704e-05,
-      "loss": 5.1921,
-      "step": 5027
-    },
-    {
-      "epoch": 2.6221642764015645,
-      "grad_norm": 1.5339542627334595,
-      "learning_rate": 9.545025125628141e-05,
-      "loss": 5.9142,
-      "step": 5028
-    },
-    {
-      "epoch": 2.622685788787484,
-      "grad_norm": 1.3102749586105347,
-      "learning_rate": 9.544924623115579e-05,
-      "loss": 6.201,
-      "step": 5029
-    },
-    {
-      "epoch": 2.623207301173403,
-      "grad_norm": 1.4846876859664917,
-      "learning_rate": 9.544824120603015e-05,
-      "loss": 5.789,
-      "step": 5030
-    },
-    {
-      "epoch": 2.623728813559322,
-      "grad_norm": 1.4042407274246216,
-      "learning_rate": 9.544723618090453e-05,
-      "loss": 6.146,
-      "step": 5031
-    },
-    {
-      "epoch": 2.6242503259452414,
-      "grad_norm": 1.5999616384506226,
-      "learning_rate": 9.544623115577889e-05,
-      "loss": 5.6465,
-      "step": 5032
-    },
-    {
-      "epoch": 2.6247718383311605,
-      "grad_norm": 1.5846381187438965,
-      "learning_rate": 9.544522613065327e-05,
-      "loss": 5.5475,
-      "step": 5033
-    },
-    {
-      "epoch": 2.6252933507170795,
-      "grad_norm": 1.3971928358078003,
-      "learning_rate": 9.544422110552765e-05,
-      "loss": 6.2333,
-      "step": 5034
-    },
-    {
-      "epoch": 2.625814863102999,
-      "grad_norm": 1.4666469097137451,
-      "learning_rate": 9.544321608040201e-05,
-      "loss": 5.4194,
-      "step": 5035
-    },
-    {
-      "epoch": 2.626336375488918,
-      "grad_norm": 1.5029730796813965,
-      "learning_rate": 9.544221105527639e-05,
-      "loss": 5.7225,
-      "step": 5036
-    },
-    {
-      "epoch": 2.626857887874837,
-      "grad_norm": 1.3892558813095093,
-      "learning_rate": 9.544120603015075e-05,
-      "loss": 6.0422,
-      "step": 5037
-    },
-    {
-      "epoch": 2.6273794002607564,
-      "grad_norm": 1.5366686582565308,
-      "learning_rate": 9.544020100502513e-05,
-      "loss": 5.9961,
-      "step": 5038
-    },
-    {
-      "epoch": 2.6279009126466755,
-      "grad_norm": 1.545335054397583,
-      "learning_rate": 9.54391959798995e-05,
-      "loss": 5.7841,
-      "step": 5039
-    },
-    {
-      "epoch": 2.6284224250325945,
-      "grad_norm": 1.6340769529342651,
-      "learning_rate": 9.543819095477387e-05,
-      "loss": 5.5915,
-      "step": 5040
-    },
-    {
-      "epoch": 2.628943937418514,
-      "grad_norm": 1.677231788635254,
-      "learning_rate": 9.543718592964824e-05,
-      "loss": 5.713,
-      "step": 5041
-    },
-    {
-      "epoch": 2.629465449804433,
-      "grad_norm": 1.5500590801239014,
-      "learning_rate": 9.543618090452262e-05,
-      "loss": 5.8206,
-      "step": 5042
-    },
-    {
-      "epoch": 2.629986962190352,
-      "grad_norm": 1.632042407989502,
-      "learning_rate": 9.543517587939698e-05,
-      "loss": 5.4749,
-      "step": 5043
-    },
-    {
-      "epoch": 2.6305084745762715,
-      "grad_norm": 1.4213882684707642,
-      "learning_rate": 9.543417085427136e-05,
-      "loss": 5.806,
-      "step": 5044
-    },
-    {
-      "epoch": 2.6310299869621905,
-      "grad_norm": 1.5072643756866455,
-      "learning_rate": 9.543316582914574e-05,
-      "loss": 5.9314,
-      "step": 5045
-    },
-    {
-      "epoch": 2.6315514993481095,
-      "grad_norm": 1.4365506172180176,
-      "learning_rate": 9.543216080402011e-05,
-      "loss": 5.9133,
-      "step": 5046
-    },
-    {
-      "epoch": 2.6320730117340285,
-      "grad_norm": 1.436258316040039,
-      "learning_rate": 9.543115577889448e-05,
-      "loss": 6.0337,
-      "step": 5047
-    },
-    {
-      "epoch": 2.632594524119948,
-      "grad_norm": 1.5551789999008179,
-      "learning_rate": 9.543015075376884e-05,
-      "loss": 5.9603,
-      "step": 5048
-    },
-    {
-      "epoch": 2.633116036505867,
-      "grad_norm": 1.539718747138977,
-      "learning_rate": 9.542914572864322e-05,
-      "loss": 5.6086,
-      "step": 5049
-    },
-    {
-      "epoch": 2.633637548891786,
-      "grad_norm": 1.736962080001831,
-      "learning_rate": 9.542814070351758e-05,
-      "loss": 5.7569,
-      "step": 5050
-    },
-    {
-      "epoch": 2.6341590612777055,
-      "grad_norm": 1.513782024383545,
-      "learning_rate": 9.542713567839196e-05,
-      "loss": 5.7692,
-      "step": 5051
-    },
-    {
-      "epoch": 2.6346805736636245,
-      "grad_norm": 1.8075569868087769,
-      "learning_rate": 9.542613065326633e-05,
-      "loss": 5.5231,
-      "step": 5052
-    },
-    {
-      "epoch": 2.6352020860495435,
-      "grad_norm": 1.522943377494812,
-      "learning_rate": 9.54251256281407e-05,
-      "loss": 5.656,
-      "step": 5053
-    },
-    {
-      "epoch": 2.635723598435463,
-      "grad_norm": 1.5258946418762207,
-      "learning_rate": 9.542412060301508e-05,
-      "loss": 5.6287,
-      "step": 5054
-    },
-    {
-      "epoch": 2.636245110821382,
-      "grad_norm": 1.4363160133361816,
-      "learning_rate": 9.542311557788946e-05,
-      "loss": 5.8832,
-      "step": 5055
-    },
-    {
-      "epoch": 2.636766623207301,
-      "grad_norm": 1.6591776609420776,
-      "learning_rate": 9.542211055276382e-05,
-      "loss": 5.7782,
-      "step": 5056
-    },
-    {
-      "epoch": 2.63728813559322,
-      "grad_norm": 1.5033172369003296,
-      "learning_rate": 9.54211055276382e-05,
-      "loss": 5.529,
-      "step": 5057
-    },
-    {
-      "epoch": 2.6378096479791395,
-      "grad_norm": 1.5052874088287354,
-      "learning_rate": 9.542010050251257e-05,
-      "loss": 6.0534,
-      "step": 5058
-    },
-    {
-      "epoch": 2.6383311603650585,
-      "grad_norm": 1.443816900253296,
-      "learning_rate": 9.541909547738694e-05,
-      "loss": 6.0019,
-      "step": 5059
-    },
-    {
-      "epoch": 2.6388526727509776,
-      "grad_norm": 1.3907922506332397,
-      "learning_rate": 9.541809045226131e-05,
-      "loss": 6.1866,
-      "step": 5060
-    },
-    {
-      "epoch": 2.639374185136897,
-      "grad_norm": 1.5504872798919678,
-      "learning_rate": 9.541708542713569e-05,
-      "loss": 6.1834,
-      "step": 5061
-    },
-    {
-      "epoch": 2.639895697522816,
-      "grad_norm": 1.9608761072158813,
-      "learning_rate": 9.541608040201005e-05,
-      "loss": 5.7587,
-      "step": 5062
-    },
-    {
-      "epoch": 2.640417209908735,
-      "grad_norm": 1.4263288974761963,
-      "learning_rate": 9.541507537688442e-05,
-      "loss": 6.0363,
-      "step": 5063
-    },
-    {
-      "epoch": 2.6409387222946545,
-      "grad_norm": 1.5398608446121216,
-      "learning_rate": 9.541407035175879e-05,
-      "loss": 5.8031,
-      "step": 5064
-    },
-    {
-      "epoch": 2.6414602346805736,
-      "grad_norm": 1.4484831094741821,
-      "learning_rate": 9.541306532663317e-05,
-      "loss": 6.164,
-      "step": 5065
-    },
-    {
-      "epoch": 2.6419817470664926,
-      "grad_norm": 1.34367835521698,
-      "learning_rate": 9.541206030150755e-05,
-      "loss": 6.0677,
-      "step": 5066
-    },
-    {
-      "epoch": 2.642503259452412,
-      "grad_norm": 1.5244886875152588,
-      "learning_rate": 9.541105527638191e-05,
-      "loss": 5.6891,
-      "step": 5067
-    },
-    {
-      "epoch": 2.643024771838331,
-      "grad_norm": 1.5506577491760254,
-      "learning_rate": 9.541005025125629e-05,
-      "loss": 5.5522,
-      "step": 5068
-    },
-    {
-      "epoch": 2.64354628422425,
-      "grad_norm": 1.6631300449371338,
-      "learning_rate": 9.540904522613066e-05,
-      "loss": 5.4891,
-      "step": 5069
-    },
-    {
-      "epoch": 2.6440677966101696,
-      "grad_norm": 1.4600266218185425,
-      "learning_rate": 9.540804020100503e-05,
-      "loss": 5.6986,
-      "step": 5070
-    },
-    {
-      "epoch": 2.6445893089960886,
-      "grad_norm": 1.6264759302139282,
-      "learning_rate": 9.54070351758794e-05,
-      "loss": 5.4695,
-      "step": 5071
-    },
-    {
-      "epoch": 2.6451108213820076,
-      "grad_norm": 1.4172817468643188,
-      "learning_rate": 9.540603015075378e-05,
-      "loss": 5.8627,
-      "step": 5072
-    },
-    {
-      "epoch": 2.645632333767927,
-      "grad_norm": 1.5001908540725708,
-      "learning_rate": 9.540502512562814e-05,
-      "loss": 5.7113,
-      "step": 5073
-    },
-    {
-      "epoch": 2.646153846153846,
-      "grad_norm": 1.6415660381317139,
-      "learning_rate": 9.540402010050252e-05,
-      "loss": 5.4965,
-      "step": 5074
-    },
-    {
-      "epoch": 2.646675358539765,
-      "grad_norm": 1.3839116096496582,
-      "learning_rate": 9.54030150753769e-05,
-      "loss": 5.9701,
-      "step": 5075
-    },
-    {
-      "epoch": 2.6471968709256846,
-      "grad_norm": 1.4515818357467651,
-      "learning_rate": 9.540201005025126e-05,
-      "loss": 6.1795,
-      "step": 5076
-    },
-    {
-      "epoch": 2.6477183833116036,
-      "grad_norm": 1.5052530765533447,
-      "learning_rate": 9.540100502512564e-05,
-      "loss": 5.8367,
-      "step": 5077
-    },
-    {
-      "epoch": 2.6482398956975226,
-      "grad_norm": 1.445517659187317,
-      "learning_rate": 9.54e-05,
-      "loss": 5.8987,
-      "step": 5078
-    },
-    {
-      "epoch": 2.648761408083442,
-      "grad_norm": 1.46699059009552,
-      "learning_rate": 9.539899497487438e-05,
-      "loss": 6.0236,
-      "step": 5079
-    },
-    {
-      "epoch": 2.649282920469361,
-      "grad_norm": 1.8584587574005127,
-      "learning_rate": 9.539798994974874e-05,
-      "loss": 5.1286,
-      "step": 5080
-    },
-    {
-      "epoch": 2.64980443285528,
-      "grad_norm": 1.3936880826950073,
-      "learning_rate": 9.539698492462312e-05,
-      "loss": 6.0714,
-      "step": 5081
-    },
-    {
-      "epoch": 2.6503259452411996,
-      "grad_norm": 2.976459264755249,
-      "learning_rate": 9.539597989949749e-05,
-      "loss": 4.9812,
-      "step": 5082
-    },
-    {
-      "epoch": 2.6508474576271186,
-      "grad_norm": 1.6329282522201538,
-      "learning_rate": 9.539497487437186e-05,
-      "loss": 5.9051,
-      "step": 5083
-    },
-    {
-      "epoch": 2.6513689700130376,
-      "grad_norm": 1.4490526914596558,
-      "learning_rate": 9.539396984924623e-05,
-      "loss": 5.8654,
-      "step": 5084
-    },
-    {
-      "epoch": 2.651890482398957,
-      "grad_norm": 1.683050274848938,
-      "learning_rate": 9.53929648241206e-05,
-      "loss": 5.5123,
-      "step": 5085
-    },
-    {
-      "epoch": 2.652411994784876,
-      "grad_norm": 1.479514241218567,
-      "learning_rate": 9.539195979899498e-05,
-      "loss": 6.0158,
-      "step": 5086
-    },
-    {
-      "epoch": 2.652933507170795,
-      "grad_norm": 1.6067718267440796,
-      "learning_rate": 9.539095477386936e-05,
-      "loss": 5.8476,
-      "step": 5087
-    },
-    {
-      "epoch": 2.6534550195567146,
-      "grad_norm": 1.6182746887207031,
-      "learning_rate": 9.538994974874373e-05,
-      "loss": 5.991,
-      "step": 5088
-    },
-    {
-      "epoch": 2.6539765319426336,
-      "grad_norm": 1.6602853536605835,
-      "learning_rate": 9.538894472361809e-05,
-      "loss": 5.8801,
-      "step": 5089
-    },
-    {
-      "epoch": 2.6544980443285526,
-      "grad_norm": 1.5242979526519775,
-      "learning_rate": 9.538793969849247e-05,
-      "loss": 6.0922,
-      "step": 5090
-    },
-    {
-      "epoch": 2.655019556714472,
-      "grad_norm": 1.442383885383606,
-      "learning_rate": 9.538693467336683e-05,
-      "loss": 5.8011,
-      "step": 5091
-    },
-    {
-      "epoch": 2.655541069100391,
-      "grad_norm": 1.4333420991897583,
-      "learning_rate": 9.538592964824121e-05,
-      "loss": 5.7155,
-      "step": 5092
-    },
-    {
-      "epoch": 2.65606258148631,
-      "grad_norm": 1.4986058473587036,
-      "learning_rate": 9.538492462311557e-05,
-      "loss": 5.5191,
-      "step": 5093
-    },
-    {
-      "epoch": 2.6565840938722296,
-      "grad_norm": 1.3976603746414185,
-      "learning_rate": 9.538391959798995e-05,
-      "loss": 6.0328,
-      "step": 5094
-    },
-    {
-      "epoch": 2.6571056062581486,
-      "grad_norm": 1.400056004524231,
-      "learning_rate": 9.538291457286433e-05,
-      "loss": 6.127,
-      "step": 5095
-    },
-    {
-      "epoch": 2.6576271186440676,
-      "grad_norm": 1.5347181558609009,
-      "learning_rate": 9.538190954773871e-05,
-      "loss": 5.5926,
-      "step": 5096
-    },
-    {
-      "epoch": 2.658148631029987,
-      "grad_norm": 1.5451070070266724,
-      "learning_rate": 9.538090452261307e-05,
-      "loss": 5.6362,
-      "step": 5097
-    },
-    {
-      "epoch": 2.658670143415906,
-      "grad_norm": 1.563220739364624,
-      "learning_rate": 9.537989949748745e-05,
-      "loss": 6.0911,
-      "step": 5098
-    },
-    {
-      "epoch": 2.659191655801825,
-      "grad_norm": 1.447637915611267,
-      "learning_rate": 9.537889447236181e-05,
-      "loss": 5.7648,
-      "step": 5099
-    },
-    {
-      "epoch": 2.6597131681877446,
-      "grad_norm": 1.5903068780899048,
-      "learning_rate": 9.537788944723619e-05,
-      "loss": 5.8409,
-      "step": 5100
-    },
-    {
-      "epoch": 2.6602346805736636,
-      "grad_norm": 1.5306520462036133,
-      "learning_rate": 9.537688442211056e-05,
-      "loss": 5.929,
-      "step": 5101
-    },
-    {
-      "epoch": 2.6607561929595827,
-      "grad_norm": 1.5478739738464355,
-      "learning_rate": 9.537587939698492e-05,
-      "loss": 5.774,
-      "step": 5102
-    },
-    {
-      "epoch": 2.661277705345502,
-      "grad_norm": 1.6153430938720703,
-      "learning_rate": 9.53748743718593e-05,
-      "loss": 5.6799,
-      "step": 5103
-    },
-    {
-      "epoch": 2.661799217731421,
-      "grad_norm": 1.8335466384887695,
-      "learning_rate": 9.537386934673366e-05,
-      "loss": 5.3785,
-      "step": 5104
-    },
-    {
-      "epoch": 2.66232073011734,
-      "grad_norm": 1.7194207906723022,
-      "learning_rate": 9.537286432160804e-05,
-      "loss": 5.7453,
-      "step": 5105
-    },
-    {
-      "epoch": 2.6628422425032596,
-      "grad_norm": 1.543682336807251,
-      "learning_rate": 9.537185929648242e-05,
-      "loss": 5.8407,
-      "step": 5106
-    },
-    {
-      "epoch": 2.6633637548891786,
-      "grad_norm": 1.4730786085128784,
-      "learning_rate": 9.53708542713568e-05,
-      "loss": 5.7341,
-      "step": 5107
-    },
-    {
-      "epoch": 2.6638852672750977,
-      "grad_norm": 1.3241785764694214,
-      "learning_rate": 9.536984924623116e-05,
-      "loss": 6.1994,
-      "step": 5108
-    },
-    {
-      "epoch": 2.664406779661017,
-      "grad_norm": 1.4168648719787598,
-      "learning_rate": 9.536884422110554e-05,
-      "loss": 5.6842,
-      "step": 5109
-    },
-    {
-      "epoch": 2.664928292046936,
-      "grad_norm": 1.5501210689544678,
-      "learning_rate": 9.53678391959799e-05,
-      "loss": 5.6424,
-      "step": 5110
-    },
-    {
-      "epoch": 2.665449804432855,
-      "grad_norm": 1.5539036989212036,
-      "learning_rate": 9.536683417085428e-05,
-      "loss": 5.6763,
-      "step": 5111
-    },
-    {
-      "epoch": 2.6659713168187746,
-      "grad_norm": 1.6299200057983398,
-      "learning_rate": 9.536582914572864e-05,
-      "loss": 5.732,
-      "step": 5112
-    },
-    {
-      "epoch": 2.6664928292046937,
-      "grad_norm": 1.5042723417282104,
-      "learning_rate": 9.536482412060302e-05,
-      "loss": 5.9362,
-      "step": 5113
-    },
-    {
-      "epoch": 2.6670143415906127,
-      "grad_norm": 1.5157870054244995,
-      "learning_rate": 9.536381909547739e-05,
-      "loss": 5.3122,
-      "step": 5114
-    },
-    {
-      "epoch": 2.667535853976532,
-      "grad_norm": 1.6283293962478638,
-      "learning_rate": 9.536281407035176e-05,
-      "loss": 5.8761,
-      "step": 5115
-    },
-    {
-      "epoch": 2.668057366362451,
-      "grad_norm": 1.4683616161346436,
-      "learning_rate": 9.536180904522614e-05,
-      "loss": 6.0856,
-      "step": 5116
-    },
-    {
-      "epoch": 2.66857887874837,
-      "grad_norm": 1.3522953987121582,
-      "learning_rate": 9.53608040201005e-05,
-      "loss": 6.1401,
-      "step": 5117
-    },
-    {
-      "epoch": 2.6691003911342897,
-      "grad_norm": 1.7956067323684692,
-      "learning_rate": 9.535979899497488e-05,
-      "loss": 5.5571,
-      "step": 5118
-    },
-    {
-      "epoch": 2.6696219035202087,
-      "grad_norm": 1.7063608169555664,
-      "learning_rate": 9.535879396984925e-05,
-      "loss": 5.7644,
-      "step": 5119
-    },
-    {
-      "epoch": 2.6701434159061277,
-      "grad_norm": 1.540897011756897,
-      "learning_rate": 9.535778894472363e-05,
-      "loss": 5.69,
-      "step": 5120
-    },
-    {
-      "epoch": 2.670664928292047,
-      "grad_norm": 1.49859619140625,
-      "learning_rate": 9.535678391959799e-05,
-      "loss": 5.5124,
-      "step": 5121
-    },
-    {
-      "epoch": 2.671186440677966,
-      "grad_norm": 1.5233063697814941,
-      "learning_rate": 9.535577889447237e-05,
-      "loss": 5.5869,
-      "step": 5122
-    },
-    {
-      "epoch": 2.671707953063885,
-      "grad_norm": 1.841001272201538,
-      "learning_rate": 9.535477386934673e-05,
-      "loss": 5.5394,
-      "step": 5123
-    },
-    {
-      "epoch": 2.6722294654498047,
-      "grad_norm": 1.5955227613449097,
-      "learning_rate": 9.535376884422111e-05,
-      "loss": 6.1203,
-      "step": 5124
-    },
-    {
-      "epoch": 2.6727509778357237,
-      "grad_norm": 1.506130576133728,
-      "learning_rate": 9.535276381909547e-05,
-      "loss": 6.0362,
-      "step": 5125
-    },
-    {
-      "epoch": 2.6732724902216427,
-      "grad_norm": 1.6731722354888916,
-      "learning_rate": 9.535175879396985e-05,
-      "loss": 5.1234,
-      "step": 5126
-    },
-    {
-      "epoch": 2.673794002607562,
-      "grad_norm": 1.5869596004486084,
-      "learning_rate": 9.535075376884423e-05,
-      "loss": 6.0809,
-      "step": 5127
-    },
-    {
-      "epoch": 2.674315514993481,
-      "grad_norm": 1.615325927734375,
-      "learning_rate": 9.53497487437186e-05,
-      "loss": 5.6983,
-      "step": 5128
-    },
-    {
-      "epoch": 2.6748370273794,
-      "grad_norm": 1.5391204357147217,
-      "learning_rate": 9.534874371859297e-05,
-      "loss": 5.827,
-      "step": 5129
-    },
-    {
-      "epoch": 2.6753585397653197,
-      "grad_norm": 1.4242204427719116,
-      "learning_rate": 9.534773869346734e-05,
-      "loss": 6.3313,
-      "step": 5130
-    },
-    {
-      "epoch": 2.6758800521512387,
-      "grad_norm": 1.5119777917861938,
-      "learning_rate": 9.534673366834171e-05,
-      "loss": 5.3284,
-      "step": 5131
-    },
-    {
-      "epoch": 2.6764015645371577,
-      "grad_norm": 1.461450219154358,
-      "learning_rate": 9.534572864321608e-05,
-      "loss": 6.1421,
-      "step": 5132
-    },
-    {
-      "epoch": 2.676923076923077,
-      "grad_norm": 1.3794804811477661,
-      "learning_rate": 9.534472361809046e-05,
-      "loss": 6.1087,
-      "step": 5133
-    },
-    {
-      "epoch": 2.677444589308996,
-      "grad_norm": 1.720292091369629,
-      "learning_rate": 9.534371859296482e-05,
-      "loss": 5.5028,
-      "step": 5134
-    },
-    {
-      "epoch": 2.6779661016949152,
-      "grad_norm": 1.5049498081207275,
-      "learning_rate": 9.53427135678392e-05,
-      "loss": 5.955,
-      "step": 5135
-    },
-    {
-      "epoch": 2.6784876140808347,
-      "grad_norm": 1.4400975704193115,
-      "learning_rate": 9.534170854271358e-05,
-      "loss": 5.8641,
-      "step": 5136
-    },
-    {
-      "epoch": 2.6790091264667537,
-      "grad_norm": 1.5083614587783813,
-      "learning_rate": 9.534070351758795e-05,
-      "loss": 5.3277,
-      "step": 5137
-    },
-    {
-      "epoch": 2.6795306388526727,
-      "grad_norm": 1.5147428512573242,
-      "learning_rate": 9.533969849246232e-05,
-      "loss": 5.3186,
-      "step": 5138
-    },
-    {
-      "epoch": 2.6800521512385918,
-      "grad_norm": 1.4857791662216187,
-      "learning_rate": 9.53386934673367e-05,
-      "loss": 5.739,
-      "step": 5139
-    },
-    {
-      "epoch": 2.680573663624511,
-      "grad_norm": 1.4180806875228882,
-      "learning_rate": 9.533768844221106e-05,
-      "loss": 5.8829,
-      "step": 5140
-    },
-    {
-      "epoch": 2.6810951760104302,
-      "grad_norm": 1.3846855163574219,
-      "learning_rate": 9.533668341708543e-05,
-      "loss": 5.905,
-      "step": 5141
-    },
-    {
-      "epoch": 2.6816166883963493,
-      "grad_norm": 1.390498161315918,
-      "learning_rate": 9.53356783919598e-05,
-      "loss": 5.9663,
-      "step": 5142
-    },
-    {
-      "epoch": 2.6821382007822687,
-      "grad_norm": 1.3757820129394531,
-      "learning_rate": 9.533467336683417e-05,
-      "loss": 5.9549,
-      "step": 5143
-    },
-    {
-      "epoch": 2.6826597131681877,
-      "grad_norm": 1.4589306116104126,
-      "learning_rate": 9.533366834170855e-05,
-      "loss": 5.8502,
-      "step": 5144
-    },
-    {
-      "epoch": 2.6831812255541068,
-      "grad_norm": 1.3386086225509644,
-      "learning_rate": 9.533266331658291e-05,
-      "loss": 5.9957,
-      "step": 5145
-    },
-    {
-      "epoch": 2.6837027379400262,
-      "grad_norm": 1.4587695598602295,
-      "learning_rate": 9.533165829145729e-05,
-      "loss": 5.8309,
-      "step": 5146
-    },
-    {
-      "epoch": 2.6842242503259452,
-      "grad_norm": 1.360872745513916,
-      "learning_rate": 9.533065326633166e-05,
-      "loss": 5.8577,
-      "step": 5147
-    },
-    {
-      "epoch": 2.6847457627118643,
-      "grad_norm": 1.595858097076416,
-      "learning_rate": 9.532964824120604e-05,
-      "loss": 5.6721,
-      "step": 5148
-    },
-    {
-      "epoch": 2.6852672750977837,
-      "grad_norm": 1.4469940662384033,
-      "learning_rate": 9.532864321608041e-05,
-      "loss": 6.0894,
-      "step": 5149
-    },
-    {
-      "epoch": 2.6857887874837028,
-      "grad_norm": 1.5330045223236084,
-      "learning_rate": 9.532763819095478e-05,
-      "loss": 5.9846,
-      "step": 5150
-    },
-    {
-      "epoch": 2.6863102998696218,
-      "grad_norm": 1.7494968175888062,
-      "learning_rate": 9.532663316582915e-05,
-      "loss": 5.6543,
-      "step": 5151
-    },
-    {
-      "epoch": 2.686831812255541,
-      "grad_norm": 1.5228216648101807,
-      "learning_rate": 9.532562814070353e-05,
-      "loss": 5.7057,
-      "step": 5152
-    },
-    {
-      "epoch": 2.6873533246414603,
-      "grad_norm": 1.5397599935531616,
-      "learning_rate": 9.532462311557789e-05,
-      "loss": 5.7234,
-      "step": 5153
-    },
-    {
-      "epoch": 2.6878748370273793,
-      "grad_norm": 1.4691836833953857,
-      "learning_rate": 9.532361809045227e-05,
-      "loss": 5.7489,
-      "step": 5154
-    },
-    {
-      "epoch": 2.6883963494132983,
-      "grad_norm": 1.2525509595870972,
-      "learning_rate": 9.532261306532663e-05,
-      "loss": 5.9964,
-      "step": 5155
-    },
-    {
-      "epoch": 2.6889178617992178,
-      "grad_norm": 1.3811547756195068,
-      "learning_rate": 9.532160804020101e-05,
-      "loss": 6.0054,
-      "step": 5156
-    },
-    {
-      "epoch": 2.689439374185137,
-      "grad_norm": 1.4274972677230835,
-      "learning_rate": 9.532060301507539e-05,
-      "loss": 5.8716,
-      "step": 5157
-    },
-    {
-      "epoch": 2.689960886571056,
-      "grad_norm": 1.4014118909835815,
-      "learning_rate": 9.531959798994975e-05,
-      "loss": 5.8567,
-      "step": 5158
-    },
-    {
-      "epoch": 2.6904823989569753,
-      "grad_norm": 1.4200385808944702,
-      "learning_rate": 9.531859296482413e-05,
-      "loss": 5.802,
-      "step": 5159
-    },
-    {
-      "epoch": 2.6910039113428943,
-      "grad_norm": 1.6902804374694824,
-      "learning_rate": 9.53175879396985e-05,
-      "loss": 5.4695,
-      "step": 5160
-    },
-    {
-      "epoch": 2.6915254237288133,
-      "grad_norm": 1.6169764995574951,
-      "learning_rate": 9.531658291457287e-05,
-      "loss": 5.5174,
-      "step": 5161
-    },
-    {
-      "epoch": 2.692046936114733,
-      "grad_norm": 1.5280276536941528,
-      "learning_rate": 9.531557788944724e-05,
-      "loss": 6.0423,
-      "step": 5162
-    },
-    {
-      "epoch": 2.692568448500652,
-      "grad_norm": 1.4429396390914917,
-      "learning_rate": 9.531457286432162e-05,
-      "loss": 6.2644,
-      "step": 5163
-    },
-    {
-      "epoch": 2.693089960886571,
-      "grad_norm": 1.4513996839523315,
-      "learning_rate": 9.531356783919598e-05,
-      "loss": 6.1521,
-      "step": 5164
-    },
-    {
-      "epoch": 2.6936114732724903,
-      "grad_norm": 1.4250222444534302,
-      "learning_rate": 9.531256281407036e-05,
-      "loss": 6.0255,
-      "step": 5165
-    },
-    {
-      "epoch": 2.6941329856584093,
-      "grad_norm": 1.502251386642456,
-      "learning_rate": 9.531155778894472e-05,
-      "loss": 5.8056,
-      "step": 5166
-    },
-    {
-      "epoch": 2.6946544980443283,
-      "grad_norm": 1.4401469230651855,
-      "learning_rate": 9.53105527638191e-05,
-      "loss": 5.7866,
-      "step": 5167
-    },
-    {
-      "epoch": 2.695176010430248,
-      "grad_norm": 1.6457639932632446,
-      "learning_rate": 9.530954773869348e-05,
-      "loss": 5.9127,
-      "step": 5168
-    },
-    {
-      "epoch": 2.695697522816167,
-      "grad_norm": 1.3917865753173828,
-      "learning_rate": 9.530854271356784e-05,
-      "loss": 6.0433,
-      "step": 5169
-    },
-    {
-      "epoch": 2.696219035202086,
-      "grad_norm": 1.559657096862793,
-      "learning_rate": 9.530753768844222e-05,
-      "loss": 5.6674,
-      "step": 5170
-    },
-    {
-      "epoch": 2.6967405475880053,
-      "grad_norm": 1.5140495300292969,
-      "learning_rate": 9.530653266331658e-05,
-      "loss": 6.1845,
-      "step": 5171
-    },
-    {
-      "epoch": 2.6972620599739243,
-      "grad_norm": 1.5789153575897217,
-      "learning_rate": 9.530552763819096e-05,
-      "loss": 5.4262,
-      "step": 5172
-    },
-    {
-      "epoch": 2.6977835723598433,
-      "grad_norm": 1.3890787363052368,
-      "learning_rate": 9.530452261306533e-05,
-      "loss": 6.1416,
-      "step": 5173
-    },
-    {
-      "epoch": 2.698305084745763,
-      "grad_norm": 1.504050612449646,
-      "learning_rate": 9.53035175879397e-05,
-      "loss": 6.0222,
-      "step": 5174
-    },
-    {
-      "epoch": 2.698826597131682,
-      "grad_norm": 1.6642489433288574,
-      "learning_rate": 9.530251256281407e-05,
-      "loss": 5.6307,
-      "step": 5175
-    },
-    {
-      "epoch": 2.699348109517601,
-      "grad_norm": 1.6988753080368042,
-      "learning_rate": 9.530150753768845e-05,
-      "loss": 5.6507,
-      "step": 5176
-    },
-    {
-      "epoch": 2.6998696219035203,
-      "grad_norm": 1.7431178092956543,
-      "learning_rate": 9.530050251256282e-05,
-      "loss": 5.2001,
-      "step": 5177
-    },
-    {
-      "epoch": 2.7003911342894393,
-      "grad_norm": 1.5542442798614502,
-      "learning_rate": 9.52994974874372e-05,
-      "loss": 5.9271,
-      "step": 5178
-    },
-    {
-      "epoch": 2.7009126466753584,
-      "grad_norm": 1.6295182704925537,
-      "learning_rate": 9.529849246231157e-05,
-      "loss": 5.9589,
-      "step": 5179
-    },
-    {
-      "epoch": 2.701434159061278,
-      "grad_norm": 1.5354334115982056,
-      "learning_rate": 9.529748743718594e-05,
-      "loss": 5.4012,
-      "step": 5180
-    },
-    {
-      "epoch": 2.701955671447197,
-      "grad_norm": 1.575677752494812,
-      "learning_rate": 9.529648241206031e-05,
-      "loss": 5.685,
-      "step": 5181
-    },
-    {
-      "epoch": 2.702477183833116,
-      "grad_norm": 1.5911293029785156,
-      "learning_rate": 9.529547738693467e-05,
-      "loss": 5.9932,
-      "step": 5182
-    },
-    {
-      "epoch": 2.7029986962190353,
-      "grad_norm": 1.590155005455017,
-      "learning_rate": 9.529447236180905e-05,
-      "loss": 5.5379,
-      "step": 5183
-    },
-    {
-      "epoch": 2.7035202086049543,
-      "grad_norm": 1.3691349029541016,
-      "learning_rate": 9.529346733668341e-05,
-      "loss": 5.8124,
-      "step": 5184
-    },
-    {
-      "epoch": 2.7040417209908734,
-      "grad_norm": 1.5409071445465088,
-      "learning_rate": 9.529246231155779e-05,
-      "loss": 6.0204,
-      "step": 5185
-    },
-    {
-      "epoch": 2.704563233376793,
-      "grad_norm": 1.450858473777771,
-      "learning_rate": 9.529145728643216e-05,
-      "loss": 5.7684,
-      "step": 5186
-    },
-    {
-      "epoch": 2.705084745762712,
-      "grad_norm": 1.5022752285003662,
-      "learning_rate": 9.529045226130653e-05,
-      "loss": 5.687,
-      "step": 5187
-    },
-    {
-      "epoch": 2.705606258148631,
-      "grad_norm": 1.5702344179153442,
-      "learning_rate": 9.528944723618091e-05,
-      "loss": 5.7013,
-      "step": 5188
-    },
-    {
-      "epoch": 2.7061277705345503,
-      "grad_norm": 1.4816606044769287,
-      "learning_rate": 9.528844221105529e-05,
-      "loss": 6.062,
-      "step": 5189
-    },
-    {
-      "epoch": 2.7066492829204694,
-      "grad_norm": 1.5153523683547974,
-      "learning_rate": 9.528743718592965e-05,
-      "loss": 5.7961,
-      "step": 5190
-    },
-    {
-      "epoch": 2.7071707953063884,
-      "grad_norm": 1.3507033586502075,
-      "learning_rate": 9.528643216080403e-05,
-      "loss": 5.9947,
-      "step": 5191
-    },
-    {
-      "epoch": 2.707692307692308,
-      "grad_norm": 1.6690802574157715,
-      "learning_rate": 9.52854271356784e-05,
-      "loss": 5.3276,
-      "step": 5192
-    },
-    {
-      "epoch": 2.708213820078227,
-      "grad_norm": 1.5505234003067017,
-      "learning_rate": 9.528442211055277e-05,
-      "loss": 5.9616,
-      "step": 5193
-    },
-    {
-      "epoch": 2.708735332464146,
-      "grad_norm": 1.3932942152023315,
-      "learning_rate": 9.528341708542714e-05,
-      "loss": 6.2604,
-      "step": 5194
-    },
-    {
-      "epoch": 2.7092568448500653,
-      "grad_norm": 1.5112119913101196,
-      "learning_rate": 9.52824120603015e-05,
-      "loss": 5.7811,
-      "step": 5195
-    },
-    {
-      "epoch": 2.7097783572359844,
-      "grad_norm": 1.6142544746398926,
-      "learning_rate": 9.528140703517588e-05,
-      "loss": 5.9023,
-      "step": 5196
-    },
-    {
-      "epoch": 2.7102998696219034,
-      "grad_norm": 1.3298146724700928,
-      "learning_rate": 9.528040201005026e-05,
-      "loss": 6.2009,
-      "step": 5197
-    },
-    {
-      "epoch": 2.710821382007823,
-      "grad_norm": 1.4620105028152466,
-      "learning_rate": 9.527939698492464e-05,
-      "loss": 5.9448,
-      "step": 5198
-    },
-    {
-      "epoch": 2.711342894393742,
-      "grad_norm": 1.51407790184021,
-      "learning_rate": 9.5278391959799e-05,
-      "loss": 5.7596,
-      "step": 5199
-    },
-    {
-      "epoch": 2.711864406779661,
-      "grad_norm": 1.4731444120407104,
-      "learning_rate": 9.527738693467338e-05,
-      "loss": 5.8057,
-      "step": 5200
-    },
-    {
-      "epoch": 2.7123859191655804,
-      "grad_norm": 1.6514036655426025,
-      "learning_rate": 9.527638190954774e-05,
-      "loss": 6.2216,
-      "step": 5201
-    },
-    {
-      "epoch": 2.7129074315514994,
-      "grad_norm": 1.6662817001342773,
-      "learning_rate": 9.527537688442212e-05,
-      "loss": 5.9857,
-      "step": 5202
-    },
-    {
-      "epoch": 2.7134289439374184,
-      "grad_norm": 1.5811858177185059,
-      "learning_rate": 9.527437185929648e-05,
-      "loss": 5.6259,
-      "step": 5203
-    },
-    {
-      "epoch": 2.713950456323338,
-      "grad_norm": 1.4241440296173096,
-      "learning_rate": 9.527336683417086e-05,
-      "loss": 5.5792,
-      "step": 5204
-    },
-    {
-      "epoch": 2.714471968709257,
-      "grad_norm": 1.6506435871124268,
-      "learning_rate": 9.527236180904523e-05,
-      "loss": 5.7458,
-      "step": 5205
-    },
-    {
-      "epoch": 2.714993481095176,
-      "grad_norm": 1.456315517425537,
-      "learning_rate": 9.52713567839196e-05,
-      "loss": 5.8962,
-      "step": 5206
-    },
-    {
-      "epoch": 2.7155149934810954,
-      "grad_norm": 1.5771361589431763,
-      "learning_rate": 9.527035175879397e-05,
-      "loss": 5.98,
-      "step": 5207
-    },
-    {
-      "epoch": 2.7160365058670144,
-      "grad_norm": 1.5656245946884155,
-      "learning_rate": 9.526934673366835e-05,
-      "loss": 6.1064,
-      "step": 5208
-    },
-    {
-      "epoch": 2.7165580182529334,
-      "grad_norm": 1.4260226488113403,
-      "learning_rate": 9.526834170854272e-05,
-      "loss": 5.9251,
-      "step": 5209
-    },
-    {
-      "epoch": 2.717079530638853,
-      "grad_norm": 1.6678003072738647,
-      "learning_rate": 9.526733668341709e-05,
-      "loss": 5.5837,
-      "step": 5210
-    },
-    {
-      "epoch": 2.717601043024772,
-      "grad_norm": 1.546895146369934,
-      "learning_rate": 9.526633165829147e-05,
-      "loss": 6.3094,
-      "step": 5211
-    },
-    {
-      "epoch": 2.718122555410691,
-      "grad_norm": 1.4008996486663818,
-      "learning_rate": 9.526532663316583e-05,
-      "loss": 6.0016,
-      "step": 5212
-    },
-    {
-      "epoch": 2.7186440677966104,
-      "grad_norm": 1.5887113809585571,
-      "learning_rate": 9.526432160804021e-05,
-      "loss": 5.7557,
-      "step": 5213
-    },
-    {
-      "epoch": 2.7191655801825294,
-      "grad_norm": 1.6973627805709839,
-      "learning_rate": 9.526331658291457e-05,
-      "loss": 5.9702,
-      "step": 5214
-    },
-    {
-      "epoch": 2.7196870925684484,
-      "grad_norm": 1.6317579746246338,
-      "learning_rate": 9.526231155778895e-05,
-      "loss": 5.6162,
-      "step": 5215
-    },
-    {
-      "epoch": 2.720208604954368,
-      "grad_norm": 1.635650396347046,
-      "learning_rate": 9.526130653266331e-05,
-      "loss": 6.1131,
-      "step": 5216
-    },
-    {
-      "epoch": 2.720730117340287,
-      "grad_norm": 1.7450494766235352,
-      "learning_rate": 9.526030150753769e-05,
-      "loss": 5.6754,
-      "step": 5217
-    },
-    {
-      "epoch": 2.721251629726206,
-      "grad_norm": 1.7048437595367432,
-      "learning_rate": 9.525929648241206e-05,
-      "loss": 5.5702,
-      "step": 5218
-    },
-    {
-      "epoch": 2.7217731421121254,
-      "grad_norm": 1.4270488023757935,
-      "learning_rate": 9.525829145728643e-05,
-      "loss": 5.9984,
-      "step": 5219
-    },
-    {
-      "epoch": 2.7222946544980444,
-      "grad_norm": 1.9475847482681274,
-      "learning_rate": 9.525728643216081e-05,
-      "loss": 5.6623,
-      "step": 5220
-    },
-    {
-      "epoch": 2.7228161668839634,
-      "grad_norm": 1.6363919973373413,
-      "learning_rate": 9.525628140703518e-05,
-      "loss": 5.2335,
-      "step": 5221
-    },
-    {
-      "epoch": 2.723337679269883,
-      "grad_norm": 1.4743471145629883,
-      "learning_rate": 9.525527638190955e-05,
-      "loss": 5.9159,
-      "step": 5222
-    },
-    {
-      "epoch": 2.723859191655802,
-      "grad_norm": 1.596564531326294,
-      "learning_rate": 9.525427135678392e-05,
-      "loss": 5.4452,
-      "step": 5223
-    },
-    {
-      "epoch": 2.724380704041721,
-      "grad_norm": 1.6014529466629028,
-      "learning_rate": 9.52532663316583e-05,
-      "loss": 5.9816,
-      "step": 5224
-    },
-    {
-      "epoch": 2.7249022164276404,
-      "grad_norm": 1.4191795587539673,
-      "learning_rate": 9.525226130653266e-05,
-      "loss": 5.5061,
-      "step": 5225
-    },
-    {
-      "epoch": 2.7254237288135594,
-      "grad_norm": 1.6846827268600464,
-      "learning_rate": 9.525125628140704e-05,
-      "loss": 5.4852,
-      "step": 5226
-    },
-    {
-      "epoch": 2.7259452411994785,
-      "grad_norm": 1.5207194089889526,
-      "learning_rate": 9.52502512562814e-05,
-      "loss": 5.8801,
-      "step": 5227
-    },
-    {
-      "epoch": 2.726466753585398,
-      "grad_norm": 1.5339123010635376,
-      "learning_rate": 9.524924623115578e-05,
-      "loss": 5.5618,
-      "step": 5228
-    },
-    {
-      "epoch": 2.726988265971317,
-      "grad_norm": 1.5552427768707275,
-      "learning_rate": 9.524824120603016e-05,
-      "loss": 5.8779,
-      "step": 5229
-    },
-    {
-      "epoch": 2.727509778357236,
-      "grad_norm": 1.5081803798675537,
-      "learning_rate": 9.524723618090454e-05,
-      "loss": 6.1867,
-      "step": 5230
-    },
-    {
-      "epoch": 2.7280312907431554,
-      "grad_norm": 1.4801026582717896,
-      "learning_rate": 9.52462311557789e-05,
-      "loss": 5.9867,
-      "step": 5231
-    },
-    {
-      "epoch": 2.7285528031290744,
-      "grad_norm": 1.6295626163482666,
-      "learning_rate": 9.524522613065328e-05,
-      "loss": 5.6305,
-      "step": 5232
-    },
-    {
-      "epoch": 2.7290743155149935,
-      "grad_norm": 1.6634424924850464,
-      "learning_rate": 9.524422110552764e-05,
-      "loss": 5.8667,
-      "step": 5233
-    },
-    {
-      "epoch": 2.7295958279009125,
-      "grad_norm": 1.5615614652633667,
-      "learning_rate": 9.524321608040202e-05,
-      "loss": 5.5773,
-      "step": 5234
-    },
-    {
-      "epoch": 2.730117340286832,
-      "grad_norm": 1.3656904697418213,
-      "learning_rate": 9.524221105527639e-05,
-      "loss": 5.8018,
-      "step": 5235
-    },
-    {
-      "epoch": 2.730638852672751,
-      "grad_norm": 1.6661819219589233,
-      "learning_rate": 9.524120603015075e-05,
-      "loss": 5.3977,
-      "step": 5236
-    },
-    {
-      "epoch": 2.73116036505867,
-      "grad_norm": 1.5089644193649292,
-      "learning_rate": 9.524020100502513e-05,
-      "loss": 6.0428,
-      "step": 5237
-    },
-    {
-      "epoch": 2.7316818774445895,
-      "grad_norm": 1.4739046096801758,
-      "learning_rate": 9.523919597989949e-05,
-      "loss": 5.6986,
-      "step": 5238
-    },
-    {
-      "epoch": 2.7322033898305085,
-      "grad_norm": 1.7612227201461792,
-      "learning_rate": 9.523819095477387e-05,
-      "loss": 5.9634,
-      "step": 5239
-    },
-    {
-      "epoch": 2.7327249022164275,
-      "grad_norm": 1.6065841913223267,
-      "learning_rate": 9.523718592964825e-05,
-      "loss": 5.8587,
-      "step": 5240
-    },
-    {
-      "epoch": 2.733246414602347,
-      "grad_norm": 1.4992141723632812,
-      "learning_rate": 9.523618090452263e-05,
-      "loss": 5.8403,
-      "step": 5241
-    },
-    {
-      "epoch": 2.733767926988266,
-      "grad_norm": 1.5832908153533936,
-      "learning_rate": 9.523517587939699e-05,
-      "loss": 5.7498,
-      "step": 5242
-    },
-    {
-      "epoch": 2.734289439374185,
-      "grad_norm": 1.5719740390777588,
-      "learning_rate": 9.523417085427137e-05,
-      "loss": 5.6622,
-      "step": 5243
-    },
-    {
-      "epoch": 2.734810951760104,
-      "grad_norm": 1.4680424928665161,
-      "learning_rate": 9.523316582914573e-05,
-      "loss": 6.0262,
-      "step": 5244
-    },
-    {
-      "epoch": 2.7353324641460235,
-      "grad_norm": 1.319040298461914,
-      "learning_rate": 9.523216080402011e-05,
-      "loss": 5.6795,
-      "step": 5245
-    },
-    {
-      "epoch": 2.7358539765319425,
-      "grad_norm": 1.4283701181411743,
-      "learning_rate": 9.523115577889447e-05,
-      "loss": 5.9524,
-      "step": 5246
-    },
-    {
-      "epoch": 2.7363754889178615,
-      "grad_norm": 1.646993637084961,
-      "learning_rate": 9.523015075376885e-05,
-      "loss": 5.7868,
-      "step": 5247
-    },
-    {
-      "epoch": 2.736897001303781,
-      "grad_norm": 1.6843148469924927,
-      "learning_rate": 9.522914572864322e-05,
-      "loss": 5.2376,
-      "step": 5248
-    },
-    {
-      "epoch": 2.7374185136897,
-      "grad_norm": 1.5444928407669067,
-      "learning_rate": 9.52281407035176e-05,
-      "loss": 5.7794,
-      "step": 5249
-    },
-    {
-      "epoch": 2.737940026075619,
-      "grad_norm": 1.584246277809143,
-      "learning_rate": 9.522713567839197e-05,
-      "loss": 5.7508,
-      "step": 5250
-    },
-    {
-      "epoch": 2.7384615384615385,
-      "grad_norm": 1.4963221549987793,
-      "learning_rate": 9.522613065326634e-05,
-      "loss": 5.7942,
-      "step": 5251
-    },
-    {
-      "epoch": 2.7389830508474575,
-      "grad_norm": 1.556229829788208,
-      "learning_rate": 9.522512562814071e-05,
-      "loss": 6.1317,
-      "step": 5252
-    },
-    {
-      "epoch": 2.7395045632333765,
-      "grad_norm": 1.2980328798294067,
-      "learning_rate": 9.522412060301508e-05,
-      "loss": 6.0471,
-      "step": 5253
-    },
-    {
-      "epoch": 2.740026075619296,
-      "grad_norm": 1.4990075826644897,
-      "learning_rate": 9.522311557788946e-05,
-      "loss": 5.8681,
-      "step": 5254
-    },
-    {
-      "epoch": 2.740547588005215,
-      "grad_norm": 1.4045356512069702,
-      "learning_rate": 9.522211055276382e-05,
-      "loss": 5.7768,
-      "step": 5255
-    },
-    {
-      "epoch": 2.741069100391134,
-      "grad_norm": 1.508265733718872,
-      "learning_rate": 9.52211055276382e-05,
-      "loss": 6.0307,
-      "step": 5256
-    },
-    {
-      "epoch": 2.7415906127770535,
-      "grad_norm": 1.4944521188735962,
-      "learning_rate": 9.522010050251256e-05,
-      "loss": 5.8939,
-      "step": 5257
-    },
-    {
-      "epoch": 2.7421121251629725,
-      "grad_norm": 1.5403271913528442,
-      "learning_rate": 9.521909547738694e-05,
-      "loss": 5.7276,
-      "step": 5258
-    },
-    {
-      "epoch": 2.7426336375488916,
-      "grad_norm": 1.3685622215270996,
-      "learning_rate": 9.52180904522613e-05,
-      "loss": 5.5307,
-      "step": 5259
-    },
-    {
-      "epoch": 2.743155149934811,
-      "grad_norm": 1.7211834192276,
-      "learning_rate": 9.521708542713568e-05,
-      "loss": 6.1457,
-      "step": 5260
-    },
-    {
-      "epoch": 2.74367666232073,
-      "grad_norm": 1.4081915616989136,
-      "learning_rate": 9.521608040201006e-05,
-      "loss": 6.0878,
-      "step": 5261
-    },
-    {
-      "epoch": 2.744198174706649,
-      "grad_norm": 1.3571991920471191,
-      "learning_rate": 9.521507537688442e-05,
-      "loss": 5.8913,
-      "step": 5262
-    },
-    {
-      "epoch": 2.7447196870925685,
-      "grad_norm": 1.3831044435501099,
-      "learning_rate": 9.52140703517588e-05,
-      "loss": 6.1563,
-      "step": 5263
-    },
-    {
-      "epoch": 2.7452411994784875,
-      "grad_norm": 1.5365378856658936,
-      "learning_rate": 9.521306532663317e-05,
-      "loss": 5.8124,
-      "step": 5264
-    },
-    {
-      "epoch": 2.7457627118644066,
-      "grad_norm": 1.4498540163040161,
-      "learning_rate": 9.521206030150754e-05,
-      "loss": 5.812,
-      "step": 5265
-    },
-    {
-      "epoch": 2.746284224250326,
-      "grad_norm": 1.5595202445983887,
-      "learning_rate": 9.521105527638191e-05,
-      "loss": 5.526,
-      "step": 5266
-    },
-    {
-      "epoch": 2.746805736636245,
-      "grad_norm": 1.626511573791504,
-      "learning_rate": 9.521005025125629e-05,
-      "loss": 5.2402,
-      "step": 5267
-    },
-    {
-      "epoch": 2.747327249022164,
-      "grad_norm": 1.680371642112732,
-      "learning_rate": 9.520904522613065e-05,
-      "loss": 5.7109,
-      "step": 5268
-    },
-    {
-      "epoch": 2.7478487614080835,
-      "grad_norm": 1.5864266157150269,
-      "learning_rate": 9.520804020100503e-05,
-      "loss": 5.8349,
-      "step": 5269
-    },
-    {
-      "epoch": 2.7483702737940026,
-      "grad_norm": 1.446987271308899,
-      "learning_rate": 9.52070351758794e-05,
-      "loss": 6.3372,
-      "step": 5270
-    },
-    {
-      "epoch": 2.7488917861799216,
-      "grad_norm": 1.6633707284927368,
-      "learning_rate": 9.520603015075378e-05,
-      "loss": 5.4647,
-      "step": 5271
-    },
-    {
-      "epoch": 2.749413298565841,
-      "grad_norm": 1.7349506616592407,
-      "learning_rate": 9.520502512562815e-05,
-      "loss": 5.6277,
-      "step": 5272
-    },
-    {
-      "epoch": 2.74993481095176,
-      "grad_norm": 1.840499758720398,
-      "learning_rate": 9.520402010050253e-05,
-      "loss": 5.8586,
-      "step": 5273
-    },
-    {
-      "epoch": 2.750456323337679,
-      "grad_norm": 1.4551581144332886,
-      "learning_rate": 9.520301507537689e-05,
-      "loss": 5.6955,
-      "step": 5274
-    },
-    {
-      "epoch": 2.7509778357235986,
-      "grad_norm": 1.585686445236206,
-      "learning_rate": 9.520201005025125e-05,
-      "loss": 6.0729,
-      "step": 5275
-    },
-    {
-      "epoch": 2.7514993481095176,
-      "grad_norm": 1.6242752075195312,
-      "learning_rate": 9.520100502512563e-05,
-      "loss": 6.153,
-      "step": 5276
-    },
-    {
-      "epoch": 2.7520208604954366,
-      "grad_norm": 1.4138590097427368,
-      "learning_rate": 9.52e-05,
-      "loss": 4.8482,
-      "step": 5277
-    },
-    {
-      "epoch": 2.752542372881356,
-      "grad_norm": 1.6448571681976318,
-      "learning_rate": 9.519899497487437e-05,
-      "loss": 5.9421,
-      "step": 5278
-    },
-    {
-      "epoch": 2.753063885267275,
-      "grad_norm": 1.4570629596710205,
-      "learning_rate": 9.519798994974874e-05,
-      "loss": 6.1594,
-      "step": 5279
-    },
-    {
-      "epoch": 2.753585397653194,
-      "grad_norm": 1.3304738998413086,
-      "learning_rate": 9.519698492462312e-05,
-      "loss": 6.165,
-      "step": 5280
-    },
-    {
-      "epoch": 2.7541069100391136,
-      "grad_norm": 1.4532541036605835,
-      "learning_rate": 9.51959798994975e-05,
-      "loss": 5.5049,
-      "step": 5281
-    },
-    {
-      "epoch": 2.7546284224250326,
-      "grad_norm": 1.843823790550232,
-      "learning_rate": 9.519497487437187e-05,
-      "loss": 5.5123,
-      "step": 5282
-    },
-    {
-      "epoch": 2.7551499348109516,
-      "grad_norm": 1.6437034606933594,
-      "learning_rate": 9.519396984924624e-05,
-      "loss": 5.7397,
-      "step": 5283
-    },
-    {
-      "epoch": 2.755671447196871,
-      "grad_norm": 1.440700888633728,
-      "learning_rate": 9.519296482412061e-05,
-      "loss": 5.8074,
-      "step": 5284
-    },
-    {
-      "epoch": 2.75619295958279,
-      "grad_norm": 1.4381103515625,
-      "learning_rate": 9.519195979899498e-05,
-      "loss": 5.9209,
-      "step": 5285
-    },
-    {
-      "epoch": 2.756714471968709,
-      "grad_norm": 1.539421796798706,
-      "learning_rate": 9.519095477386936e-05,
-      "loss": 6.116,
-      "step": 5286
-    },
-    {
-      "epoch": 2.7572359843546286,
-      "grad_norm": 1.4918252229690552,
-      "learning_rate": 9.518994974874372e-05,
-      "loss": 6.007,
-      "step": 5287
-    },
-    {
-      "epoch": 2.7577574967405476,
-      "grad_norm": 1.6424527168273926,
-      "learning_rate": 9.518894472361808e-05,
-      "loss": 5.4255,
-      "step": 5288
-    },
-    {
-      "epoch": 2.7582790091264666,
-      "grad_norm": 1.709058403968811,
-      "learning_rate": 9.518793969849246e-05,
-      "loss": 5.2981,
-      "step": 5289
-    },
-    {
-      "epoch": 2.758800521512386,
-      "grad_norm": 1.7231090068817139,
-      "learning_rate": 9.518693467336684e-05,
-      "loss": 6.138,
-      "step": 5290
-    },
-    {
-      "epoch": 2.759322033898305,
-      "grad_norm": 1.4244486093521118,
-      "learning_rate": 9.518592964824122e-05,
-      "loss": 6.224,
-      "step": 5291
-    },
-    {
-      "epoch": 2.759843546284224,
-      "grad_norm": 1.5862233638763428,
-      "learning_rate": 9.518492462311558e-05,
-      "loss": 5.9853,
-      "step": 5292
-    },
-    {
-      "epoch": 2.7603650586701436,
-      "grad_norm": 1.7865623235702515,
-      "learning_rate": 9.518391959798996e-05,
-      "loss": 5.917,
-      "step": 5293
-    },
-    {
-      "epoch": 2.7608865710560626,
-      "grad_norm": 1.4637805223464966,
-      "learning_rate": 9.518291457286432e-05,
-      "loss": 5.9428,
-      "step": 5294
-    },
-    {
-      "epoch": 2.7614080834419816,
-      "grad_norm": 1.5762534141540527,
-      "learning_rate": 9.51819095477387e-05,
-      "loss": 6.1152,
-      "step": 5295
-    },
-    {
-      "epoch": 2.761929595827901,
-      "grad_norm": 1.545445442199707,
-      "learning_rate": 9.518090452261307e-05,
-      "loss": 6.152,
-      "step": 5296
-    },
-    {
-      "epoch": 2.76245110821382,
-      "grad_norm": 1.57672119140625,
-      "learning_rate": 9.517989949748744e-05,
-      "loss": 5.6974,
-      "step": 5297
-    },
-    {
-      "epoch": 2.762972620599739,
-      "grad_norm": 1.5162392854690552,
-      "learning_rate": 9.517889447236181e-05,
-      "loss": 5.714,
-      "step": 5298
-    },
-    {
-      "epoch": 2.7634941329856586,
-      "grad_norm": 1.56229829788208,
-      "learning_rate": 9.517788944723619e-05,
-      "loss": 5.6277,
-      "step": 5299
-    },
-    {
-      "epoch": 2.7640156453715776,
-      "grad_norm": 1.5812842845916748,
-      "learning_rate": 9.517688442211055e-05,
-      "loss": 4.6416,
-      "step": 5300
-    },
-    {
-      "epoch": 2.7645371577574966,
-      "grad_norm": 1.3698291778564453,
-      "learning_rate": 9.517587939698493e-05,
-      "loss": 5.9934,
-      "step": 5301
-    },
-    {
-      "epoch": 2.765058670143416,
-      "grad_norm": 1.4683197736740112,
-      "learning_rate": 9.51748743718593e-05,
-      "loss": 5.9069,
-      "step": 5302
-    },
-    {
-      "epoch": 2.765580182529335,
-      "grad_norm": 1.49152410030365,
-      "learning_rate": 9.517386934673367e-05,
-      "loss": 5.5447,
-      "step": 5303
-    },
-    {
-      "epoch": 2.766101694915254,
-      "grad_norm": 1.404457688331604,
-      "learning_rate": 9.517286432160805e-05,
-      "loss": 5.9199,
-      "step": 5304
-    },
-    {
-      "epoch": 2.7666232073011736,
-      "grad_norm": 1.3284046649932861,
-      "learning_rate": 9.517185929648241e-05,
-      "loss": 6.0208,
-      "step": 5305
-    },
-    {
-      "epoch": 2.7671447196870926,
-      "grad_norm": 1.4136748313903809,
-      "learning_rate": 9.517085427135679e-05,
-      "loss": 5.6044,
-      "step": 5306
-    },
-    {
-      "epoch": 2.7676662320730117,
-      "grad_norm": 1.5955919027328491,
-      "learning_rate": 9.516984924623116e-05,
-      "loss": 5.7154,
-      "step": 5307
-    },
-    {
-      "epoch": 2.768187744458931,
-      "grad_norm": 1.4675759077072144,
-      "learning_rate": 9.516884422110553e-05,
-      "loss": 5.6239,
-      "step": 5308
-    },
-    {
-      "epoch": 2.76870925684485,
-      "grad_norm": 1.539570689201355,
-      "learning_rate": 9.51678391959799e-05,
-      "loss": 5.4812,
-      "step": 5309
-    },
-    {
-      "epoch": 2.769230769230769,
-      "grad_norm": 1.4700385332107544,
-      "learning_rate": 9.516683417085428e-05,
-      "loss": 5.9133,
-      "step": 5310
-    },
-    {
-      "epoch": 2.7697522816166886,
-      "grad_norm": 1.6473584175109863,
-      "learning_rate": 9.516582914572865e-05,
-      "loss": 5.2062,
-      "step": 5311
-    },
-    {
-      "epoch": 2.7702737940026076,
-      "grad_norm": 1.4371368885040283,
-      "learning_rate": 9.516482412060303e-05,
-      "loss": 5.9706,
-      "step": 5312
-    },
-    {
-      "epoch": 2.7707953063885267,
-      "grad_norm": 1.3796151876449585,
-      "learning_rate": 9.51638190954774e-05,
-      "loss": 5.8902,
-      "step": 5313
-    },
-    {
-      "epoch": 2.771316818774446,
-      "grad_norm": 1.6886078119277954,
-      "learning_rate": 9.516281407035176e-05,
-      "loss": 4.981,
-      "step": 5314
-    },
-    {
-      "epoch": 2.771838331160365,
-      "grad_norm": 1.6681435108184814,
-      "learning_rate": 9.516180904522614e-05,
-      "loss": 6.0854,
-      "step": 5315
-    },
-    {
-      "epoch": 2.772359843546284,
-      "grad_norm": 1.42898428440094,
-      "learning_rate": 9.51608040201005e-05,
-      "loss": 5.865,
-      "step": 5316
-    },
-    {
-      "epoch": 2.7728813559322036,
-      "grad_norm": 1.4149826765060425,
-      "learning_rate": 9.515979899497488e-05,
-      "loss": 6.04,
-      "step": 5317
-    },
-    {
-      "epoch": 2.7734028683181227,
-      "grad_norm": 1.6551878452301025,
-      "learning_rate": 9.515879396984924e-05,
-      "loss": 5.4065,
-      "step": 5318
-    },
-    {
-      "epoch": 2.7739243807040417,
-      "grad_norm": 1.6741372346878052,
-      "learning_rate": 9.515778894472362e-05,
-      "loss": 5.5303,
-      "step": 5319
-    },
-    {
-      "epoch": 2.774445893089961,
-      "grad_norm": 1.5373040437698364,
-      "learning_rate": 9.515678391959799e-05,
-      "loss": 5.3272,
-      "step": 5320
-    },
-    {
-      "epoch": 2.77496740547588,
-      "grad_norm": 1.3805116415023804,
-      "learning_rate": 9.515577889447236e-05,
-      "loss": 6.1417,
-      "step": 5321
-    },
-    {
-      "epoch": 2.775488917861799,
-      "grad_norm": 1.7865455150604248,
-      "learning_rate": 9.515477386934674e-05,
-      "loss": 5.8941,
-      "step": 5322
-    },
-    {
-      "epoch": 2.7760104302477187,
-      "grad_norm": 1.5107587575912476,
-      "learning_rate": 9.515376884422112e-05,
-      "loss": 5.6905,
-      "step": 5323
-    },
-    {
-      "epoch": 2.7765319426336377,
-      "grad_norm": 1.3821722269058228,
-      "learning_rate": 9.515276381909548e-05,
-      "loss": 5.8651,
-      "step": 5324
-    },
-    {
-      "epoch": 2.7770534550195567,
-      "grad_norm": 1.5068247318267822,
-      "learning_rate": 9.515175879396986e-05,
-      "loss": 5.7301,
-      "step": 5325
-    },
-    {
-      "epoch": 2.7775749674054757,
-      "grad_norm": 1.6178607940673828,
-      "learning_rate": 9.515075376884423e-05,
-      "loss": 5.8373,
-      "step": 5326
-    },
-    {
-      "epoch": 2.778096479791395,
-      "grad_norm": 1.5161018371582031,
-      "learning_rate": 9.51497487437186e-05,
-      "loss": 5.9352,
-      "step": 5327
-    },
-    {
-      "epoch": 2.778617992177314,
-      "grad_norm": 1.3542464971542358,
-      "learning_rate": 9.514874371859297e-05,
-      "loss": 6.1018,
-      "step": 5328
-    },
-    {
-      "epoch": 2.779139504563233,
-      "grad_norm": 1.4921947717666626,
-      "learning_rate": 9.514773869346733e-05,
-      "loss": 5.8118,
-      "step": 5329
-    },
-    {
-      "epoch": 2.7796610169491527,
-      "grad_norm": 1.4626189470291138,
-      "learning_rate": 9.514673366834171e-05,
-      "loss": 5.637,
-      "step": 5330
-    },
-    {
-      "epoch": 2.7801825293350717,
-      "grad_norm": 1.5856717824935913,
-      "learning_rate": 9.514572864321609e-05,
-      "loss": 5.7706,
-      "step": 5331
-    },
-    {
-      "epoch": 2.7807040417209907,
-      "grad_norm": 1.7111527919769287,
-      "learning_rate": 9.514472361809047e-05,
-      "loss": 5.4238,
-      "step": 5332
-    },
-    {
-      "epoch": 2.78122555410691,
-      "grad_norm": 1.4562779664993286,
-      "learning_rate": 9.514371859296483e-05,
-      "loss": 5.7839,
-      "step": 5333
-    },
-    {
-      "epoch": 2.781747066492829,
-      "grad_norm": 1.4199082851409912,
-      "learning_rate": 9.514271356783921e-05,
-      "loss": 6.2704,
-      "step": 5334
-    },
-    {
-      "epoch": 2.7822685788787482,
-      "grad_norm": 1.3992305994033813,
-      "learning_rate": 9.514170854271357e-05,
-      "loss": 5.6577,
-      "step": 5335
-    },
-    {
-      "epoch": 2.7827900912646677,
-      "grad_norm": 1.749971628189087,
-      "learning_rate": 9.514070351758795e-05,
-      "loss": 5.7725,
-      "step": 5336
-    },
-    {
-      "epoch": 2.7833116036505867,
-      "grad_norm": 1.4763962030410767,
-      "learning_rate": 9.513969849246231e-05,
-      "loss": 5.9232,
-      "step": 5337
-    },
-    {
-      "epoch": 2.7838331160365057,
-      "grad_norm": 1.4396134614944458,
-      "learning_rate": 9.513869346733669e-05,
-      "loss": 6.0913,
-      "step": 5338
-    },
-    {
-      "epoch": 2.7843546284224248,
-      "grad_norm": 1.7151845693588257,
-      "learning_rate": 9.513768844221106e-05,
-      "loss": 5.6339,
-      "step": 5339
-    },
-    {
-      "epoch": 2.7848761408083442,
-      "grad_norm": 1.5447626113891602,
-      "learning_rate": 9.513668341708543e-05,
-      "loss": 5.652,
-      "step": 5340
-    },
-    {
-      "epoch": 2.7853976531942632,
-      "grad_norm": 1.5760295391082764,
-      "learning_rate": 9.51356783919598e-05,
-      "loss": 5.9678,
-      "step": 5341
-    },
-    {
-      "epoch": 2.7859191655801823,
-      "grad_norm": 1.720670461654663,
-      "learning_rate": 9.513467336683418e-05,
-      "loss": 5.6917,
-      "step": 5342
-    },
-    {
-      "epoch": 2.7864406779661017,
-      "grad_norm": 1.5652952194213867,
-      "learning_rate": 9.513366834170855e-05,
-      "loss": 5.9148,
-      "step": 5343
-    },
-    {
-      "epoch": 2.7869621903520208,
-      "grad_norm": 1.4579381942749023,
-      "learning_rate": 9.513266331658292e-05,
-      "loss": 6.1761,
-      "step": 5344
-    },
-    {
-      "epoch": 2.7874837027379398,
-      "grad_norm": 1.4980690479278564,
-      "learning_rate": 9.51316582914573e-05,
-      "loss": 6.0517,
-      "step": 5345
-    },
-    {
-      "epoch": 2.7880052151238592,
-      "grad_norm": 2.142726182937622,
-      "learning_rate": 9.513065326633166e-05,
-      "loss": 5.4164,
-      "step": 5346
-    },
-    {
-      "epoch": 2.7885267275097783,
-      "grad_norm": 1.4630898237228394,
-      "learning_rate": 9.512964824120604e-05,
-      "loss": 5.9949,
-      "step": 5347
-    },
-    {
-      "epoch": 2.7890482398956973,
-      "grad_norm": 1.5615463256835938,
-      "learning_rate": 9.51286432160804e-05,
-      "loss": 5.5716,
-      "step": 5348
-    },
-    {
-      "epoch": 2.7895697522816167,
-      "grad_norm": 1.5081136226654053,
-      "learning_rate": 9.512763819095478e-05,
-      "loss": 5.9938,
-      "step": 5349
-    },
-    {
-      "epoch": 2.7900912646675358,
-      "grad_norm": 1.423330307006836,
-      "learning_rate": 9.512663316582914e-05,
-      "loss": 6.0489,
-      "step": 5350
-    },
-    {
-      "epoch": 2.790612777053455,
-      "grad_norm": 1.4629098176956177,
-      "learning_rate": 9.512562814070352e-05,
-      "loss": 5.9993,
-      "step": 5351
-    },
-    {
-      "epoch": 2.7911342894393742,
-      "grad_norm": 1.6566526889801025,
-      "learning_rate": 9.51246231155779e-05,
-      "loss": 5.8187,
-      "step": 5352
-    },
-    {
-      "epoch": 2.7916558018252933,
-      "grad_norm": 1.7587112188339233,
-      "learning_rate": 9.512361809045228e-05,
-      "loss": 5.4436,
-      "step": 5353
-    },
-    {
-      "epoch": 2.7921773142112123,
-      "grad_norm": 1.542237639427185,
-      "learning_rate": 9.512261306532664e-05,
-      "loss": 5.6578,
-      "step": 5354
-    },
-    {
-      "epoch": 2.7926988265971318,
-      "grad_norm": 1.5793741941452026,
-      "learning_rate": 9.5121608040201e-05,
-      "loss": 5.5994,
-      "step": 5355
-    },
-    {
-      "epoch": 2.7932203389830508,
-      "grad_norm": 1.7326124906539917,
-      "learning_rate": 9.512060301507538e-05,
-      "loss": 5.4092,
-      "step": 5356
-    },
-    {
-      "epoch": 2.79374185136897,
-      "grad_norm": 1.6850502490997314,
-      "learning_rate": 9.511959798994975e-05,
-      "loss": 5.8681,
-      "step": 5357
-    },
-    {
-      "epoch": 2.7942633637548893,
-      "grad_norm": 1.6342360973358154,
-      "learning_rate": 9.511859296482413e-05,
-      "loss": 5.7698,
-      "step": 5358
-    },
-    {
-      "epoch": 2.7947848761408083,
-      "grad_norm": 2.0453426837921143,
-      "learning_rate": 9.511758793969849e-05,
-      "loss": 5.513,
-      "step": 5359
-    },
-    {
-      "epoch": 2.7953063885267273,
-      "grad_norm": 1.3269829750061035,
-      "learning_rate": 9.511658291457287e-05,
-      "loss": 6.1498,
-      "step": 5360
-    },
-    {
-      "epoch": 2.7958279009126468,
-      "grad_norm": 1.7415722608566284,
-      "learning_rate": 9.511557788944723e-05,
-      "loss": 5.6446,
-      "step": 5361
-    },
-    {
-      "epoch": 2.796349413298566,
-      "grad_norm": 1.816104531288147,
-      "learning_rate": 9.511457286432161e-05,
-      "loss": 5.5064,
-      "step": 5362
-    },
-    {
-      "epoch": 2.796870925684485,
-      "grad_norm": 1.4518098831176758,
-      "learning_rate": 9.511356783919599e-05,
-      "loss": 6.2357,
-      "step": 5363
-    },
-    {
-      "epoch": 2.7973924380704043,
-      "grad_norm": 1.3511797189712524,
-      "learning_rate": 9.511256281407037e-05,
-      "loss": 6.2815,
-      "step": 5364
-    },
-    {
-      "epoch": 2.7979139504563233,
-      "grad_norm": 1.592790126800537,
-      "learning_rate": 9.511155778894473e-05,
-      "loss": 5.4512,
-      "step": 5365
-    },
-    {
-      "epoch": 2.7984354628422423,
-      "grad_norm": 1.5891486406326294,
-      "learning_rate": 9.511055276381911e-05,
-      "loss": 5.8074,
-      "step": 5366
-    },
-    {
-      "epoch": 2.798956975228162,
-      "grad_norm": 1.4719278812408447,
-      "learning_rate": 9.510954773869347e-05,
-      "loss": 6.0028,
-      "step": 5367
-    },
-    {
-      "epoch": 2.799478487614081,
-      "grad_norm": 1.3128036260604858,
-      "learning_rate": 9.510854271356784e-05,
-      "loss": 5.9743,
-      "step": 5368
-    },
-    {
-      "epoch": 2.8,
-      "grad_norm": 1.3965797424316406,
-      "learning_rate": 9.510753768844221e-05,
-      "loss": 5.5463,
-      "step": 5369
-    },
-    {
-      "epoch": 2.8005215123859193,
-      "grad_norm": 1.5687730312347412,
-      "learning_rate": 9.510653266331658e-05,
-      "loss": 5.6928,
-      "step": 5370
-    },
-    {
-      "epoch": 2.8010430247718383,
-      "grad_norm": 1.8856099843978882,
-      "learning_rate": 9.510552763819096e-05,
-      "loss": 5.4445,
-      "step": 5371
-    },
-    {
-      "epoch": 2.8015645371577573,
-      "grad_norm": 1.4559097290039062,
-      "learning_rate": 9.510452261306532e-05,
-      "loss": 5.9814,
-      "step": 5372
-    },
-    {
-      "epoch": 2.802086049543677,
-      "grad_norm": 1.4248132705688477,
-      "learning_rate": 9.51035175879397e-05,
-      "loss": 5.853,
-      "step": 5373
-    },
-    {
-      "epoch": 2.802607561929596,
-      "grad_norm": 1.5703685283660889,
-      "learning_rate": 9.510251256281408e-05,
-      "loss": 5.6143,
-      "step": 5374
-    },
-    {
-      "epoch": 2.803129074315515,
-      "grad_norm": 1.511752963066101,
-      "learning_rate": 9.510150753768845e-05,
-      "loss": 5.9094,
-      "step": 5375
-    },
-    {
-      "epoch": 2.8036505867014343,
-      "grad_norm": 1.4544997215270996,
-      "learning_rate": 9.510050251256282e-05,
-      "loss": 5.8296,
-      "step": 5376
-    },
-    {
-      "epoch": 2.8041720990873533,
-      "grad_norm": 1.4706684350967407,
-      "learning_rate": 9.50994974874372e-05,
-      "loss": 5.8256,
-      "step": 5377
-    },
-    {
-      "epoch": 2.8046936114732723,
-      "grad_norm": 1.5199040174484253,
-      "learning_rate": 9.509849246231156e-05,
-      "loss": 5.6294,
-      "step": 5378
-    },
-    {
-      "epoch": 2.805215123859192,
-      "grad_norm": 1.4112098217010498,
-      "learning_rate": 9.509748743718594e-05,
-      "loss": 5.9636,
-      "step": 5379
-    },
-    {
-      "epoch": 2.805736636245111,
-      "grad_norm": 1.7174955606460571,
-      "learning_rate": 9.50964824120603e-05,
-      "loss": 5.4974,
-      "step": 5380
-    },
-    {
-      "epoch": 2.80625814863103,
-      "grad_norm": 1.4473001956939697,
-      "learning_rate": 9.509547738693467e-05,
-      "loss": 5.9944,
-      "step": 5381
-    },
-    {
-      "epoch": 2.8067796610169493,
-      "grad_norm": 1.4157061576843262,
-      "learning_rate": 9.509447236180905e-05,
-      "loss": 6.0787,
-      "step": 5382
-    },
-    {
-      "epoch": 2.8073011734028683,
-      "grad_norm": 1.3892755508422852,
-      "learning_rate": 9.509346733668342e-05,
-      "loss": 6.0385,
-      "step": 5383
-    },
-    {
-      "epoch": 2.8078226857887874,
-      "grad_norm": 1.507362961769104,
-      "learning_rate": 9.50924623115578e-05,
-      "loss": 5.6126,
-      "step": 5384
-    },
-    {
-      "epoch": 2.808344198174707,
-      "grad_norm": 1.5747789144515991,
-      "learning_rate": 9.509145728643216e-05,
-      "loss": 5.4299,
-      "step": 5385
-    },
-    {
-      "epoch": 2.808865710560626,
-      "grad_norm": 1.4668720960617065,
-      "learning_rate": 9.509045226130654e-05,
-      "loss": 5.503,
-      "step": 5386
-    },
-    {
-      "epoch": 2.809387222946545,
-      "grad_norm": 1.4045443534851074,
-      "learning_rate": 9.508944723618091e-05,
-      "loss": 6.026,
-      "step": 5387
-    },
-    {
-      "epoch": 2.8099087353324643,
-      "grad_norm": 1.4384702444076538,
-      "learning_rate": 9.508844221105528e-05,
-      "loss": 5.8359,
-      "step": 5388
-    },
-    {
-      "epoch": 2.8104302477183833,
-      "grad_norm": 1.3221298456192017,
-      "learning_rate": 9.508743718592965e-05,
-      "loss": 6.1395,
-      "step": 5389
-    },
-    {
-      "epoch": 2.8109517601043024,
-      "grad_norm": 1.432021141052246,
-      "learning_rate": 9.508643216080403e-05,
-      "loss": 5.8207,
-      "step": 5390
-    },
-    {
-      "epoch": 2.811473272490222,
-      "grad_norm": 1.4086962938308716,
-      "learning_rate": 9.508542713567839e-05,
-      "loss": 5.8159,
-      "step": 5391
-    },
-    {
-      "epoch": 2.811994784876141,
-      "grad_norm": 1.6543505191802979,
-      "learning_rate": 9.508442211055277e-05,
-      "loss": 5.2449,
-      "step": 5392
-    },
-    {
-      "epoch": 2.81251629726206,
-      "grad_norm": 1.4830341339111328,
-      "learning_rate": 9.508341708542713e-05,
-      "loss": 5.7108,
-      "step": 5393
-    },
-    {
-      "epoch": 2.8130378096479793,
-      "grad_norm": 1.4738376140594482,
-      "learning_rate": 9.508241206030151e-05,
-      "loss": 5.8759,
-      "step": 5394
-    },
-    {
-      "epoch": 2.8135593220338984,
-      "grad_norm": 1.4974769353866577,
-      "learning_rate": 9.508140703517589e-05,
-      "loss": 5.9035,
-      "step": 5395
-    },
-    {
-      "epoch": 2.8140808344198174,
-      "grad_norm": 1.7222765684127808,
-      "learning_rate": 9.508040201005025e-05,
-      "loss": 5.5961,
-      "step": 5396
-    },
-    {
-      "epoch": 2.814602346805737,
-      "grad_norm": 1.488580346107483,
-      "learning_rate": 9.507939698492463e-05,
-      "loss": 6.1087,
-      "step": 5397
-    },
-    {
-      "epoch": 2.815123859191656,
-      "grad_norm": 1.7291523218154907,
-      "learning_rate": 9.5078391959799e-05,
-      "loss": 5.0922,
-      "step": 5398
-    },
-    {
-      "epoch": 2.815645371577575,
-      "grad_norm": 1.4147628545761108,
-      "learning_rate": 9.507738693467337e-05,
-      "loss": 5.8352,
-      "step": 5399
-    },
-    {
-      "epoch": 2.8161668839634943,
-      "grad_norm": 1.519407033920288,
-      "learning_rate": 9.507638190954774e-05,
-      "loss": 5.809,
-      "step": 5400
-    },
-    {
-      "epoch": 2.8166883963494134,
-      "grad_norm": 1.5176655054092407,
-      "learning_rate": 9.507537688442212e-05,
-      "loss": 5.843,
-      "step": 5401
-    },
-    {
-      "epoch": 2.8172099087353324,
-      "grad_norm": 1.625676155090332,
-      "learning_rate": 9.507437185929648e-05,
-      "loss": 5.8345,
-      "step": 5402
-    },
-    {
-      "epoch": 2.817731421121252,
-      "grad_norm": 1.5023150444030762,
-      "learning_rate": 9.507336683417086e-05,
-      "loss": 5.7397,
-      "step": 5403
-    },
-    {
-      "epoch": 2.818252933507171,
-      "grad_norm": 1.5248651504516602,
-      "learning_rate": 9.507236180904524e-05,
-      "loss": 5.8339,
-      "step": 5404
-    },
-    {
-      "epoch": 2.81877444589309,
-      "grad_norm": 1.570803165435791,
-      "learning_rate": 9.507135678391961e-05,
-      "loss": 5.94,
-      "step": 5405
-    },
-    {
-      "epoch": 2.8192959582790094,
-      "grad_norm": 1.7953946590423584,
-      "learning_rate": 9.507035175879398e-05,
-      "loss": 5.2396,
-      "step": 5406
-    },
-    {
-      "epoch": 2.8198174706649284,
-      "grad_norm": 1.4574475288391113,
-      "learning_rate": 9.506934673366834e-05,
-      "loss": 5.9126,
-      "step": 5407
-    },
-    {
-      "epoch": 2.8203389830508474,
-      "grad_norm": 1.7406460046768188,
-      "learning_rate": 9.506834170854272e-05,
-      "loss": 5.4409,
-      "step": 5408
-    },
-    {
-      "epoch": 2.820860495436767,
-      "grad_norm": 1.6594797372817993,
-      "learning_rate": 9.506733668341708e-05,
-      "loss": 5.1006,
-      "step": 5409
-    },
-    {
-      "epoch": 2.821382007822686,
-      "grad_norm": 1.917070984840393,
-      "learning_rate": 9.506633165829146e-05,
-      "loss": 5.7658,
-      "step": 5410
-    },
-    {
-      "epoch": 2.821903520208605,
-      "grad_norm": 1.6266204118728638,
-      "learning_rate": 9.506532663316583e-05,
-      "loss": 5.5152,
-      "step": 5411
-    },
-    {
-      "epoch": 2.8224250325945244,
-      "grad_norm": 1.4641432762145996,
-      "learning_rate": 9.50643216080402e-05,
-      "loss": 5.8899,
-      "step": 5412
-    },
-    {
-      "epoch": 2.8229465449804434,
-      "grad_norm": 1.5079774856567383,
-      "learning_rate": 9.506331658291457e-05,
-      "loss": 6.0276,
-      "step": 5413
-    },
-    {
-      "epoch": 2.8234680573663624,
-      "grad_norm": 1.505654215812683,
-      "learning_rate": 9.506231155778895e-05,
-      "loss": 6.0536,
-      "step": 5414
-    },
-    {
-      "epoch": 2.823989569752282,
-      "grad_norm": 1.4815384149551392,
-      "learning_rate": 9.506130653266332e-05,
-      "loss": 6.0923,
-      "step": 5415
-    },
-    {
-      "epoch": 2.824511082138201,
-      "grad_norm": 1.402385950088501,
-      "learning_rate": 9.50603015075377e-05,
-      "loss": 6.1574,
-      "step": 5416
-    },
-    {
-      "epoch": 2.82503259452412,
-      "grad_norm": 1.7535570859909058,
-      "learning_rate": 9.505929648241207e-05,
-      "loss": 5.5894,
-      "step": 5417
-    },
-    {
-      "epoch": 2.8255541069100394,
-      "grad_norm": 1.4100937843322754,
-      "learning_rate": 9.505829145728644e-05,
-      "loss": 5.819,
-      "step": 5418
-    },
-    {
-      "epoch": 2.8260756192959584,
-      "grad_norm": 1.7048600912094116,
-      "learning_rate": 9.505728643216081e-05,
-      "loss": 5.6877,
-      "step": 5419
-    },
-    {
-      "epoch": 2.8265971316818774,
-      "grad_norm": 1.4972076416015625,
-      "learning_rate": 9.505628140703519e-05,
-      "loss": 6.0138,
-      "step": 5420
-    },
-    {
-      "epoch": 2.8271186440677964,
-      "grad_norm": 1.4135085344314575,
-      "learning_rate": 9.505527638190955e-05,
-      "loss": 6.0942,
-      "step": 5421
-    },
-    {
-      "epoch": 2.827640156453716,
-      "grad_norm": 1.540353775024414,
-      "learning_rate": 9.505427135678391e-05,
-      "loss": 5.8088,
-      "step": 5422
-    },
-    {
-      "epoch": 2.828161668839635,
-      "grad_norm": 1.3191885948181152,
-      "learning_rate": 9.505326633165829e-05,
-      "loss": 5.9692,
-      "step": 5423
-    },
-    {
-      "epoch": 2.828683181225554,
-      "grad_norm": 1.4066498279571533,
-      "learning_rate": 9.505226130653267e-05,
-      "loss": 5.9971,
-      "step": 5424
-    },
-    {
-      "epoch": 2.8292046936114734,
-      "grad_norm": 1.4812179803848267,
-      "learning_rate": 9.505125628140705e-05,
-      "loss": 5.6833,
-      "step": 5425
-    },
-    {
-      "epoch": 2.8297262059973924,
-      "grad_norm": 1.6258361339569092,
-      "learning_rate": 9.505025125628141e-05,
-      "loss": 5.7476,
-      "step": 5426
-    },
-    {
-      "epoch": 2.8302477183833115,
-      "grad_norm": 1.734784722328186,
-      "learning_rate": 9.504924623115579e-05,
-      "loss": 5.3781,
-      "step": 5427
-    },
-    {
-      "epoch": 2.830769230769231,
-      "grad_norm": 1.6339596509933472,
-      "learning_rate": 9.504824120603015e-05,
-      "loss": 5.7235,
-      "step": 5428
-    },
-    {
-      "epoch": 2.83129074315515,
-      "grad_norm": 1.5125846862792969,
-      "learning_rate": 9.504723618090453e-05,
-      "loss": 5.7626,
-      "step": 5429
-    },
-    {
-      "epoch": 2.831812255541069,
-      "grad_norm": 1.3996344804763794,
-      "learning_rate": 9.50462311557789e-05,
-      "loss": 5.8532,
-      "step": 5430
-    },
-    {
-      "epoch": 2.832333767926988,
-      "grad_norm": 1.362195611000061,
-      "learning_rate": 9.504522613065327e-05,
-      "loss": 5.97,
-      "step": 5431
-    },
-    {
-      "epoch": 2.8328552803129075,
-      "grad_norm": 1.4344533681869507,
-      "learning_rate": 9.504422110552764e-05,
-      "loss": 5.9257,
-      "step": 5432
-    },
-    {
-      "epoch": 2.8333767926988265,
-      "grad_norm": 1.478940725326538,
-      "learning_rate": 9.504321608040202e-05,
-      "loss": 5.8004,
-      "step": 5433
-    },
-    {
-      "epoch": 2.8338983050847455,
-      "grad_norm": 1.429983377456665,
-      "learning_rate": 9.504221105527638e-05,
-      "loss": 5.5731,
-      "step": 5434
-    },
-    {
-      "epoch": 2.834419817470665,
-      "grad_norm": 1.4382768869400024,
-      "learning_rate": 9.504120603015076e-05,
-      "loss": 6.0552,
-      "step": 5435
-    },
-    {
-      "epoch": 2.834941329856584,
-      "grad_norm": 1.300857663154602,
-      "learning_rate": 9.504020100502514e-05,
-      "loss": 5.9213,
-      "step": 5436
-    },
-    {
-      "epoch": 2.835462842242503,
-      "grad_norm": 1.4121084213256836,
-      "learning_rate": 9.50391959798995e-05,
-      "loss": 5.699,
-      "step": 5437
-    },
-    {
-      "epoch": 2.8359843546284225,
-      "grad_norm": 1.455958366394043,
-      "learning_rate": 9.503819095477388e-05,
-      "loss": 5.7875,
-      "step": 5438
-    },
-    {
-      "epoch": 2.8365058670143415,
-      "grad_norm": 1.4203109741210938,
-      "learning_rate": 9.503718592964824e-05,
-      "loss": 5.9604,
-      "step": 5439
-    },
-    {
-      "epoch": 2.8370273794002605,
-      "grad_norm": 1.6025632619857788,
-      "learning_rate": 9.503618090452262e-05,
-      "loss": 5.7122,
-      "step": 5440
-    },
-    {
-      "epoch": 2.83754889178618,
-      "grad_norm": 1.4892053604125977,
-      "learning_rate": 9.503517587939698e-05,
-      "loss": 5.8097,
-      "step": 5441
-    },
-    {
-      "epoch": 2.838070404172099,
-      "grad_norm": 1.4848047494888306,
-      "learning_rate": 9.503417085427136e-05,
-      "loss": 5.4508,
-      "step": 5442
-    },
-    {
-      "epoch": 2.838591916558018,
-      "grad_norm": 1.587677240371704,
-      "learning_rate": 9.503316582914573e-05,
-      "loss": 5.6504,
-      "step": 5443
-    },
-    {
-      "epoch": 2.8391134289439375,
-      "grad_norm": 1.5530227422714233,
-      "learning_rate": 9.50321608040201e-05,
-      "loss": 6.0686,
-      "step": 5444
-    },
-    {
-      "epoch": 2.8396349413298565,
-      "grad_norm": 1.4392659664154053,
-      "learning_rate": 9.503115577889448e-05,
-      "loss": 5.484,
-      "step": 5445
-    },
-    {
-      "epoch": 2.8401564537157755,
-      "grad_norm": 1.4964516162872314,
-      "learning_rate": 9.503015075376886e-05,
-      "loss": 6.1956,
-      "step": 5446
-    },
-    {
-      "epoch": 2.840677966101695,
-      "grad_norm": 1.532198190689087,
-      "learning_rate": 9.502914572864322e-05,
-      "loss": 5.9125,
-      "step": 5447
-    },
-    {
-      "epoch": 2.841199478487614,
-      "grad_norm": 1.5567309856414795,
-      "learning_rate": 9.502814070351759e-05,
-      "loss": 5.6818,
-      "step": 5448
-    },
-    {
-      "epoch": 2.841720990873533,
-      "grad_norm": 1.6278194189071655,
-      "learning_rate": 9.502713567839197e-05,
-      "loss": 5.6605,
-      "step": 5449
-    },
-    {
-      "epoch": 2.8422425032594525,
-      "grad_norm": 1.5022722482681274,
-      "learning_rate": 9.502613065326633e-05,
-      "loss": 5.6544,
-      "step": 5450
-    },
-    {
-      "epoch": 2.8427640156453715,
-      "grad_norm": 1.595187783241272,
-      "learning_rate": 9.502512562814071e-05,
-      "loss": 5.3391,
-      "step": 5451
-    },
-    {
-      "epoch": 2.8432855280312905,
-      "grad_norm": 1.3358861207962036,
-      "learning_rate": 9.502412060301507e-05,
-      "loss": 5.6795,
-      "step": 5452
-    },
-    {
-      "epoch": 2.84380704041721,
-      "grad_norm": 1.5346157550811768,
-      "learning_rate": 9.502311557788945e-05,
-      "loss": 5.3749,
-      "step": 5453
-    },
-    {
-      "epoch": 2.844328552803129,
-      "grad_norm": 1.4484132528305054,
-      "learning_rate": 9.502211055276381e-05,
-      "loss": 5.903,
-      "step": 5454
-    },
-    {
-      "epoch": 2.844850065189048,
-      "grad_norm": 1.3442084789276123,
-      "learning_rate": 9.502110552763819e-05,
-      "loss": 6.1121,
-      "step": 5455
-    },
-    {
-      "epoch": 2.8453715775749675,
-      "grad_norm": 1.3326928615570068,
-      "learning_rate": 9.502010050251257e-05,
-      "loss": 4.9753,
-      "step": 5456
-    },
-    {
-      "epoch": 2.8458930899608865,
-      "grad_norm": 1.4835693836212158,
-      "learning_rate": 9.501909547738695e-05,
-      "loss": 5.9082,
-      "step": 5457
-    },
-    {
-      "epoch": 2.8464146023468055,
-      "grad_norm": 1.3572524785995483,
-      "learning_rate": 9.501809045226131e-05,
-      "loss": 5.2772,
-      "step": 5458
-    },
-    {
-      "epoch": 2.846936114732725,
-      "grad_norm": 1.4405840635299683,
-      "learning_rate": 9.501708542713569e-05,
-      "loss": 5.9282,
-      "step": 5459
-    },
-    {
-      "epoch": 2.847457627118644,
-      "grad_norm": 1.5230143070220947,
-      "learning_rate": 9.501608040201005e-05,
-      "loss": 5.8626,
-      "step": 5460
-    },
-    {
-      "epoch": 2.847979139504563,
-      "grad_norm": 1.3778576850891113,
-      "learning_rate": 9.501507537688442e-05,
-      "loss": 5.7872,
-      "step": 5461
-    },
-    {
-      "epoch": 2.8485006518904825,
-      "grad_norm": 1.5256186723709106,
-      "learning_rate": 9.50140703517588e-05,
-      "loss": 5.7536,
-      "step": 5462
-    },
-    {
-      "epoch": 2.8490221642764015,
-      "grad_norm": 1.4342402219772339,
-      "learning_rate": 9.501306532663316e-05,
-      "loss": 5.558,
-      "step": 5463
-    },
-    {
-      "epoch": 2.8495436766623206,
-      "grad_norm": 1.6662518978118896,
-      "learning_rate": 9.501206030150754e-05,
-      "loss": 5.4814,
-      "step": 5464
-    },
-    {
-      "epoch": 2.85006518904824,
-      "grad_norm": 1.5614107847213745,
-      "learning_rate": 9.501105527638192e-05,
-      "loss": 5.8797,
-      "step": 5465
-    },
-    {
-      "epoch": 2.850586701434159,
-      "grad_norm": 1.4563586711883545,
-      "learning_rate": 9.50100502512563e-05,
-      "loss": 5.6478,
-      "step": 5466
-    },
-    {
-      "epoch": 2.851108213820078,
-      "grad_norm": 1.4055339097976685,
-      "learning_rate": 9.500904522613066e-05,
-      "loss": 5.8571,
-      "step": 5467
-    },
-    {
-      "epoch": 2.8516297262059975,
-      "grad_norm": 1.4934945106506348,
-      "learning_rate": 9.500804020100504e-05,
-      "loss": 5.4422,
-      "step": 5468
-    },
-    {
-      "epoch": 2.8521512385919165,
-      "grad_norm": 1.5744796991348267,
-      "learning_rate": 9.50070351758794e-05,
-      "loss": 5.7252,
-      "step": 5469
-    },
-    {
-      "epoch": 2.8526727509778356,
-      "grad_norm": 1.7383615970611572,
-      "learning_rate": 9.500603015075378e-05,
-      "loss": 5.3504,
-      "step": 5470
-    },
-    {
-      "epoch": 2.853194263363755,
-      "grad_norm": 1.5227082967758179,
-      "learning_rate": 9.500502512562814e-05,
-      "loss": 5.8242,
-      "step": 5471
-    },
-    {
-      "epoch": 2.853715775749674,
-      "grad_norm": 1.5848947763442993,
-      "learning_rate": 9.500402010050252e-05,
-      "loss": 5.4656,
-      "step": 5472
-    },
-    {
-      "epoch": 2.854237288135593,
-      "grad_norm": 1.5149763822555542,
-      "learning_rate": 9.500301507537689e-05,
-      "loss": 5.7653,
-      "step": 5473
-    },
-    {
-      "epoch": 2.8547588005215125,
-      "grad_norm": 1.5359797477722168,
-      "learning_rate": 9.500201005025125e-05,
-      "loss": 5.9524,
-      "step": 5474
-    },
-    {
-      "epoch": 2.8552803129074316,
-      "grad_norm": 1.417772650718689,
-      "learning_rate": 9.500100502512563e-05,
-      "loss": 5.9466,
-      "step": 5475
-    },
-    {
-      "epoch": 2.8558018252933506,
-      "grad_norm": 1.5595239400863647,
-      "learning_rate": 9.5e-05,
-      "loss": 5.5856,
-      "step": 5476
-    },
-    {
-      "epoch": 2.85632333767927,
-      "grad_norm": 1.4452238082885742,
-      "learning_rate": 9.499899497487438e-05,
-      "loss": 6.2531,
-      "step": 5477
-    },
-    {
-      "epoch": 2.856844850065189,
-      "grad_norm": 1.556312084197998,
-      "learning_rate": 9.499798994974875e-05,
-      "loss": 5.6933,
-      "step": 5478
-    },
-    {
-      "epoch": 2.857366362451108,
-      "grad_norm": 1.6568973064422607,
-      "learning_rate": 9.499698492462313e-05,
-      "loss": 5.1634,
-      "step": 5479
-    },
-    {
-      "epoch": 2.8578878748370276,
-      "grad_norm": 1.5206397771835327,
-      "learning_rate": 9.499597989949749e-05,
-      "loss": 6.1232,
-      "step": 5480
-    },
-    {
-      "epoch": 2.8584093872229466,
-      "grad_norm": 1.4250537157058716,
-      "learning_rate": 9.499497487437187e-05,
-      "loss": 5.292,
-      "step": 5481
-    },
-    {
-      "epoch": 2.8589308996088656,
-      "grad_norm": 1.371863603591919,
-      "learning_rate": 9.499396984924623e-05,
-      "loss": 5.7788,
-      "step": 5482
-    },
-    {
-      "epoch": 2.859452411994785,
-      "grad_norm": 1.4204965829849243,
-      "learning_rate": 9.499296482412061e-05,
-      "loss": 5.8184,
-      "step": 5483
-    },
-    {
-      "epoch": 2.859973924380704,
-      "grad_norm": 1.381773591041565,
-      "learning_rate": 9.499195979899497e-05,
-      "loss": 6.055,
-      "step": 5484
-    },
-    {
-      "epoch": 2.860495436766623,
-      "grad_norm": 1.3773912191390991,
-      "learning_rate": 9.499095477386935e-05,
-      "loss": 5.5454,
-      "step": 5485
-    },
-    {
-      "epoch": 2.8610169491525426,
-      "grad_norm": 1.4980450868606567,
-      "learning_rate": 9.498994974874373e-05,
-      "loss": 6.307,
-      "step": 5486
-    },
-    {
-      "epoch": 2.8615384615384616,
-      "grad_norm": 1.544684648513794,
-      "learning_rate": 9.49889447236181e-05,
-      "loss": 5.4023,
-      "step": 5487
-    },
-    {
-      "epoch": 2.8620599739243806,
-      "grad_norm": 1.5524007081985474,
-      "learning_rate": 9.498793969849247e-05,
-      "loss": 6.016,
-      "step": 5488
-    },
-    {
-      "epoch": 2.8625814863103,
-      "grad_norm": 1.5435796976089478,
-      "learning_rate": 9.498693467336684e-05,
-      "loss": 6.1719,
-      "step": 5489
-    },
-    {
-      "epoch": 2.863102998696219,
-      "grad_norm": 1.579563856124878,
-      "learning_rate": 9.498592964824121e-05,
-      "loss": 5.7286,
-      "step": 5490
-    },
-    {
-      "epoch": 2.863624511082138,
-      "grad_norm": 1.4664534330368042,
-      "learning_rate": 9.498492462311558e-05,
-      "loss": 6.0726,
-      "step": 5491
-    },
-    {
-      "epoch": 2.8641460234680576,
-      "grad_norm": 1.6674902439117432,
-      "learning_rate": 9.498391959798996e-05,
-      "loss": 5.5623,
-      "step": 5492
-    },
-    {
-      "epoch": 2.8646675358539766,
-      "grad_norm": 1.636983036994934,
-      "learning_rate": 9.498291457286432e-05,
-      "loss": 5.7435,
-      "step": 5493
-    },
-    {
-      "epoch": 2.8651890482398956,
-      "grad_norm": 1.6069456338882446,
-      "learning_rate": 9.49819095477387e-05,
-      "loss": 5.6692,
-      "step": 5494
-    },
-    {
-      "epoch": 2.865710560625815,
-      "grad_norm": 1.4391626119613647,
-      "learning_rate": 9.498090452261306e-05,
-      "loss": 6.0863,
-      "step": 5495
-    },
-    {
-      "epoch": 2.866232073011734,
-      "grad_norm": 1.5208591222763062,
-      "learning_rate": 9.497989949748744e-05,
-      "loss": 5.5298,
-      "step": 5496
-    },
-    {
-      "epoch": 2.866753585397653,
-      "grad_norm": 1.386375904083252,
-      "learning_rate": 9.497889447236182e-05,
-      "loss": 5.9272,
-      "step": 5497
-    },
-    {
-      "epoch": 2.8672750977835726,
-      "grad_norm": 1.4868290424346924,
-      "learning_rate": 9.49778894472362e-05,
-      "loss": 5.9604,
-      "step": 5498
-    },
-    {
-      "epoch": 2.8677966101694916,
-      "grad_norm": 1.542762279510498,
-      "learning_rate": 9.497688442211056e-05,
-      "loss": 5.7142,
-      "step": 5499
-    },
-    {
-      "epoch": 2.8683181225554106,
-      "grad_norm": 1.5957883596420288,
-      "learning_rate": 9.497587939698492e-05,
-      "loss": 5.3213,
-      "step": 5500
-    },
-    {
-      "epoch": 2.86883963494133,
-      "grad_norm": 1.4624228477478027,
-      "learning_rate": 9.49748743718593e-05,
-      "loss": 5.7974,
-      "step": 5501
-    },
-    {
-      "epoch": 2.869361147327249,
-      "grad_norm": 1.7716432809829712,
-      "learning_rate": 9.497386934673367e-05,
-      "loss": 5.3457,
-      "step": 5502
-    },
-    {
-      "epoch": 2.869882659713168,
-      "grad_norm": 1.6208593845367432,
-      "learning_rate": 9.497286432160804e-05,
-      "loss": 5.5484,
-      "step": 5503
-    },
-    {
-      "epoch": 2.8704041720990876,
-      "grad_norm": 1.4539518356323242,
-      "learning_rate": 9.497185929648241e-05,
-      "loss": 5.7758,
-      "step": 5504
-    },
-    {
-      "epoch": 2.8709256844850066,
-      "grad_norm": 1.8079545497894287,
-      "learning_rate": 9.497085427135679e-05,
-      "loss": 5.5796,
-      "step": 5505
-    },
-    {
-      "epoch": 2.8714471968709256,
-      "grad_norm": 1.576878309249878,
-      "learning_rate": 9.496984924623116e-05,
-      "loss": 5.7139,
-      "step": 5506
-    },
-    {
-      "epoch": 2.871968709256845,
-      "grad_norm": 1.630937933921814,
-      "learning_rate": 9.496884422110554e-05,
-      "loss": 5.637,
-      "step": 5507
-    },
-    {
-      "epoch": 2.872490221642764,
-      "grad_norm": 1.4767199754714966,
-      "learning_rate": 9.49678391959799e-05,
-      "loss": 6.0818,
-      "step": 5508
-    },
-    {
-      "epoch": 2.873011734028683,
-      "grad_norm": 1.413651466369629,
-      "learning_rate": 9.496683417085428e-05,
-      "loss": 5.5064,
-      "step": 5509
-    },
-    {
-      "epoch": 2.8735332464146026,
-      "grad_norm": 1.2960796356201172,
-      "learning_rate": 9.496582914572865e-05,
-      "loss": 5.967,
-      "step": 5510
-    },
-    {
-      "epoch": 2.8740547588005216,
-      "grad_norm": 1.406331181526184,
-      "learning_rate": 9.496482412060303e-05,
-      "loss": 6.0891,
-      "step": 5511
-    },
-    {
-      "epoch": 2.8745762711864407,
-      "grad_norm": 1.3793679475784302,
-      "learning_rate": 9.496381909547739e-05,
-      "loss": 6.0884,
-      "step": 5512
-    },
-    {
-      "epoch": 2.8750977835723597,
-      "grad_norm": 1.4555490016937256,
-      "learning_rate": 9.496281407035177e-05,
-      "loss": 5.9107,
-      "step": 5513
-    },
-    {
-      "epoch": 2.875619295958279,
-      "grad_norm": 1.5267013311386108,
-      "learning_rate": 9.496180904522613e-05,
-      "loss": 5.5727,
-      "step": 5514
-    },
-    {
-      "epoch": 2.876140808344198,
-      "grad_norm": 1.5769462585449219,
-      "learning_rate": 9.49608040201005e-05,
-      "loss": 5.3566,
-      "step": 5515
-    },
-    {
-      "epoch": 2.876662320730117,
-      "grad_norm": 1.6439229249954224,
-      "learning_rate": 9.495979899497487e-05,
-      "loss": 5.8007,
-      "step": 5516
-    },
-    {
-      "epoch": 2.8771838331160366,
-      "grad_norm": 1.8192572593688965,
-      "learning_rate": 9.495879396984925e-05,
-      "loss": 5.8796,
-      "step": 5517
-    },
-    {
-      "epoch": 2.8777053455019557,
-      "grad_norm": 1.9187562465667725,
-      "learning_rate": 9.495778894472363e-05,
-      "loss": 4.946,
-      "step": 5518
-    },
-    {
-      "epoch": 2.8782268578878747,
-      "grad_norm": 1.513122320175171,
-      "learning_rate": 9.4956783919598e-05,
-      "loss": 5.9961,
-      "step": 5519
-    },
-    {
-      "epoch": 2.878748370273794,
-      "grad_norm": 1.560675024986267,
-      "learning_rate": 9.495577889447237e-05,
-      "loss": 6.0901,
-      "step": 5520
-    },
-    {
-      "epoch": 2.879269882659713,
-      "grad_norm": 1.5983387231826782,
-      "learning_rate": 9.495477386934674e-05,
-      "loss": 5.7216,
-      "step": 5521
-    },
-    {
-      "epoch": 2.879791395045632,
-      "grad_norm": 1.3903577327728271,
-      "learning_rate": 9.495376884422111e-05,
-      "loss": 5.9275,
-      "step": 5522
-    },
-    {
-      "epoch": 2.8803129074315517,
-      "grad_norm": 1.4374443292617798,
-      "learning_rate": 9.495276381909548e-05,
-      "loss": 6.0819,
-      "step": 5523
-    },
-    {
-      "epoch": 2.8808344198174707,
-      "grad_norm": 1.6240648031234741,
-      "learning_rate": 9.495175879396986e-05,
-      "loss": 5.5084,
-      "step": 5524
-    },
-    {
-      "epoch": 2.8813559322033897,
-      "grad_norm": 1.4976316690444946,
-      "learning_rate": 9.495075376884422e-05,
-      "loss": 6.0516,
-      "step": 5525
-    },
-    {
-      "epoch": 2.8818774445893087,
-      "grad_norm": 1.41958487033844,
-      "learning_rate": 9.49497487437186e-05,
-      "loss": 6.0718,
-      "step": 5526
-    },
-    {
-      "epoch": 2.882398956975228,
-      "grad_norm": 1.4065748453140259,
-      "learning_rate": 9.494874371859298e-05,
-      "loss": 6.1582,
-      "step": 5527
-    },
-    {
-      "epoch": 2.882920469361147,
-      "grad_norm": 1.6554980278015137,
-      "learning_rate": 9.494773869346734e-05,
-      "loss": 4.9616,
-      "step": 5528
-    },
-    {
-      "epoch": 2.8834419817470662,
-      "grad_norm": 1.546533226966858,
-      "learning_rate": 9.494673366834172e-05,
-      "loss": 5.572,
-      "step": 5529
-    },
-    {
-      "epoch": 2.8839634941329857,
-      "grad_norm": 1.5250715017318726,
-      "learning_rate": 9.494572864321608e-05,
-      "loss": 5.7914,
-      "step": 5530
-    },
-    {
-      "epoch": 2.8844850065189047,
-      "grad_norm": 1.5076898336410522,
-      "learning_rate": 9.494472361809046e-05,
-      "loss": 5.8171,
-      "step": 5531
-    },
-    {
-      "epoch": 2.8850065189048237,
-      "grad_norm": 1.508210301399231,
-      "learning_rate": 9.494371859296482e-05,
-      "loss": 6.1137,
-      "step": 5532
-    },
-    {
-      "epoch": 2.885528031290743,
-      "grad_norm": 1.399817943572998,
-      "learning_rate": 9.49427135678392e-05,
-      "loss": 5.9738,
-      "step": 5533
-    },
-    {
-      "epoch": 2.886049543676662,
-      "grad_norm": 1.5446008443832397,
-      "learning_rate": 9.494170854271357e-05,
-      "loss": 5.9479,
-      "step": 5534
-    },
-    {
-      "epoch": 2.8865710560625812,
-      "grad_norm": 1.7366011142730713,
-      "learning_rate": 9.494070351758794e-05,
-      "loss": 5.7081,
-      "step": 5535
-    },
-    {
-      "epoch": 2.8870925684485007,
-      "grad_norm": 1.6573882102966309,
-      "learning_rate": 9.493969849246231e-05,
-      "loss": 5.2053,
-      "step": 5536
-    },
-    {
-      "epoch": 2.8876140808344197,
-      "grad_norm": 1.385514497756958,
-      "learning_rate": 9.493869346733669e-05,
-      "loss": 5.9661,
-      "step": 5537
-    },
-    {
-      "epoch": 2.8881355932203387,
-      "grad_norm": 1.395749568939209,
-      "learning_rate": 9.493768844221106e-05,
-      "loss": 5.8266,
-      "step": 5538
-    },
-    {
-      "epoch": 2.888657105606258,
-      "grad_norm": 1.4438681602478027,
-      "learning_rate": 9.493668341708544e-05,
-      "loss": 5.8758,
-      "step": 5539
-    },
-    {
-      "epoch": 2.8891786179921772,
-      "grad_norm": 1.4077194929122925,
-      "learning_rate": 9.49356783919598e-05,
-      "loss": 6.2478,
-      "step": 5540
-    },
-    {
-      "epoch": 2.8897001303780963,
-      "grad_norm": 1.8659459352493286,
-      "learning_rate": 9.493467336683417e-05,
-      "loss": 4.9709,
-      "step": 5541
-    },
-    {
-      "epoch": 2.8902216427640157,
-      "grad_norm": 1.4451112747192383,
-      "learning_rate": 9.493366834170855e-05,
-      "loss": 6.0871,
-      "step": 5542
-    },
-    {
-      "epoch": 2.8907431551499347,
-      "grad_norm": 1.5694302320480347,
-      "learning_rate": 9.493266331658291e-05,
-      "loss": 5.3811,
-      "step": 5543
-    },
-    {
-      "epoch": 2.8912646675358538,
-      "grad_norm": 1.551498293876648,
-      "learning_rate": 9.493165829145729e-05,
-      "loss": 5.3843,
-      "step": 5544
-    },
-    {
-      "epoch": 2.8917861799217732,
-      "grad_norm": 1.49081289768219,
-      "learning_rate": 9.493065326633166e-05,
-      "loss": 6.0154,
-      "step": 5545
-    },
-    {
-      "epoch": 2.8923076923076922,
-      "grad_norm": 1.5730396509170532,
-      "learning_rate": 9.492964824120603e-05,
-      "loss": 5.5861,
-      "step": 5546
-    },
-    {
-      "epoch": 2.8928292046936113,
-      "grad_norm": 1.5991287231445312,
-      "learning_rate": 9.49286432160804e-05,
-      "loss": 6.1216,
-      "step": 5547
-    },
-    {
-      "epoch": 2.8933507170795307,
-      "grad_norm": 1.5304365158081055,
-      "learning_rate": 9.492763819095478e-05,
-      "loss": 6.0717,
-      "step": 5548
-    },
-    {
-      "epoch": 2.8938722294654498,
-      "grad_norm": 1.4365475177764893,
-      "learning_rate": 9.492663316582915e-05,
-      "loss": 5.9597,
-      "step": 5549
-    },
-    {
-      "epoch": 2.8943937418513688,
-      "grad_norm": 1.5242412090301514,
-      "learning_rate": 9.492562814070353e-05,
-      "loss": 5.8683,
-      "step": 5550
-    },
-    {
-      "epoch": 2.8949152542372882,
-      "grad_norm": 1.6503796577453613,
-      "learning_rate": 9.49246231155779e-05,
-      "loss": 5.5276,
-      "step": 5551
-    },
-    {
-      "epoch": 2.8954367666232073,
-      "grad_norm": 1.4438531398773193,
-      "learning_rate": 9.492361809045227e-05,
-      "loss": 5.6723,
-      "step": 5552
-    },
-    {
-      "epoch": 2.8959582790091263,
-      "grad_norm": 1.5143316984176636,
-      "learning_rate": 9.492261306532664e-05,
-      "loss": 5.2545,
-      "step": 5553
-    },
-    {
-      "epoch": 2.8964797913950457,
-      "grad_norm": 1.7184367179870605,
-      "learning_rate": 9.4921608040201e-05,
-      "loss": 5.6888,
-      "step": 5554
-    },
-    {
-      "epoch": 2.8970013037809648,
-      "grad_norm": 1.4547812938690186,
-      "learning_rate": 9.492060301507538e-05,
-      "loss": 6.1858,
-      "step": 5555
-    },
-    {
-      "epoch": 2.897522816166884,
-      "grad_norm": 1.58124577999115,
-      "learning_rate": 9.491959798994974e-05,
-      "loss": 5.7511,
-      "step": 5556
-    },
-    {
-      "epoch": 2.8980443285528033,
-      "grad_norm": 1.4802014827728271,
-      "learning_rate": 9.491859296482412e-05,
-      "loss": 6.0066,
-      "step": 5557
-    },
-    {
-      "epoch": 2.8985658409387223,
-      "grad_norm": 1.438165545463562,
-      "learning_rate": 9.49175879396985e-05,
-      "loss": 5.8155,
-      "step": 5558
-    },
-    {
-      "epoch": 2.8990873533246413,
-      "grad_norm": 1.6287916898727417,
-      "learning_rate": 9.491658291457288e-05,
-      "loss": 5.7862,
-      "step": 5559
-    },
-    {
-      "epoch": 2.8996088657105608,
-      "grad_norm": 1.4050424098968506,
-      "learning_rate": 9.491557788944724e-05,
-      "loss": 5.8399,
-      "step": 5560
-    },
-    {
-      "epoch": 2.9001303780964798,
-      "grad_norm": 1.3505676984786987,
-      "learning_rate": 9.491457286432162e-05,
-      "loss": 6.1767,
-      "step": 5561
-    },
-    {
-      "epoch": 2.900651890482399,
-      "grad_norm": 1.541554570198059,
-      "learning_rate": 9.491356783919598e-05,
-      "loss": 5.3691,
-      "step": 5562
-    },
-    {
-      "epoch": 2.9011734028683183,
-      "grad_norm": 1.5259424448013306,
-      "learning_rate": 9.491256281407036e-05,
-      "loss": 4.9509,
-      "step": 5563
-    },
-    {
-      "epoch": 2.9016949152542373,
-      "grad_norm": 1.597011685371399,
-      "learning_rate": 9.491155778894473e-05,
-      "loss": 5.4731,
-      "step": 5564
-    },
-    {
-      "epoch": 2.9022164276401563,
-      "grad_norm": 1.4970884323120117,
-      "learning_rate": 9.49105527638191e-05,
-      "loss": 5.7372,
-      "step": 5565
-    },
-    {
-      "epoch": 2.9027379400260758,
-      "grad_norm": 1.469681739807129,
-      "learning_rate": 9.490954773869347e-05,
-      "loss": 5.6845,
-      "step": 5566
-    },
-    {
-      "epoch": 2.903259452411995,
-      "grad_norm": 1.468544602394104,
-      "learning_rate": 9.490854271356783e-05,
-      "loss": 5.9744,
-      "step": 5567
-    },
-    {
-      "epoch": 2.903780964797914,
-      "grad_norm": 1.6005274057388306,
-      "learning_rate": 9.490753768844221e-05,
-      "loss": 5.7703,
-      "step": 5568
-    },
-    {
-      "epoch": 2.9043024771838333,
-      "grad_norm": 1.3922381401062012,
-      "learning_rate": 9.490653266331659e-05,
-      "loss": 6.1833,
-      "step": 5569
-    },
-    {
-      "epoch": 2.9048239895697523,
-      "grad_norm": 1.4666475057601929,
-      "learning_rate": 9.490552763819097e-05,
-      "loss": 5.9374,
-      "step": 5570
-    },
-    {
-      "epoch": 2.9053455019556713,
-      "grad_norm": 1.5384091138839722,
-      "learning_rate": 9.490452261306533e-05,
-      "loss": 6.1484,
-      "step": 5571
-    },
-    {
-      "epoch": 2.905867014341591,
-      "grad_norm": 1.4271762371063232,
-      "learning_rate": 9.490351758793971e-05,
-      "loss": 5.9283,
-      "step": 5572
-    },
-    {
-      "epoch": 2.90638852672751,
-      "grad_norm": 1.4915387630462646,
-      "learning_rate": 9.490251256281407e-05,
-      "loss": 5.9849,
-      "step": 5573
-    },
-    {
-      "epoch": 2.906910039113429,
-      "grad_norm": 1.3377097845077515,
-      "learning_rate": 9.490150753768845e-05,
-      "loss": 6.1545,
-      "step": 5574
-    },
-    {
-      "epoch": 2.9074315514993483,
-      "grad_norm": 1.5475226640701294,
-      "learning_rate": 9.490050251256281e-05,
-      "loss": 5.633,
-      "step": 5575
-    },
-    {
-      "epoch": 2.9079530638852673,
-      "grad_norm": 1.4059722423553467,
-      "learning_rate": 9.489949748743719e-05,
-      "loss": 5.7741,
-      "step": 5576
-    },
-    {
-      "epoch": 2.9084745762711863,
-      "grad_norm": 1.4943660497665405,
-      "learning_rate": 9.489849246231156e-05,
-      "loss": 5.6764,
-      "step": 5577
-    },
-    {
-      "epoch": 2.908996088657106,
-      "grad_norm": 1.7238649129867554,
-      "learning_rate": 9.489748743718593e-05,
-      "loss": 5.6739,
-      "step": 5578
-    },
-    {
-      "epoch": 2.909517601043025,
-      "grad_norm": 1.4704395532608032,
-      "learning_rate": 9.489648241206031e-05,
-      "loss": 6.1874,
-      "step": 5579
-    },
-    {
-      "epoch": 2.910039113428944,
-      "grad_norm": 1.4585376977920532,
-      "learning_rate": 9.489547738693468e-05,
-      "loss": 5.777,
-      "step": 5580
-    },
-    {
-      "epoch": 2.9105606258148633,
-      "grad_norm": 1.5897916555404663,
-      "learning_rate": 9.489447236180905e-05,
-      "loss": 5.9803,
-      "step": 5581
-    },
-    {
-      "epoch": 2.9110821382007823,
-      "grad_norm": 1.5488941669464111,
-      "learning_rate": 9.489346733668342e-05,
-      "loss": 5.4327,
-      "step": 5582
-    },
-    {
-      "epoch": 2.9116036505867013,
-      "grad_norm": 1.34518563747406,
-      "learning_rate": 9.48924623115578e-05,
-      "loss": 6.1353,
-      "step": 5583
-    },
-    {
-      "epoch": 2.912125162972621,
-      "grad_norm": 1.4807626008987427,
-      "learning_rate": 9.489145728643216e-05,
-      "loss": 6.1081,
-      "step": 5584
-    },
-    {
-      "epoch": 2.91264667535854,
-      "grad_norm": 1.587143063545227,
-      "learning_rate": 9.489045226130654e-05,
-      "loss": 5.3597,
-      "step": 5585
-    },
-    {
-      "epoch": 2.913168187744459,
-      "grad_norm": 1.497709035873413,
-      "learning_rate": 9.48894472361809e-05,
-      "loss": 6.0506,
-      "step": 5586
-    },
-    {
-      "epoch": 2.9136897001303783,
-      "grad_norm": 1.4687743186950684,
-      "learning_rate": 9.488844221105528e-05,
-      "loss": 5.8789,
-      "step": 5587
-    },
-    {
-      "epoch": 2.9142112125162973,
-      "grad_norm": 1.4672162532806396,
-      "learning_rate": 9.488743718592964e-05,
-      "loss": 6.0882,
-      "step": 5588
-    },
-    {
-      "epoch": 2.9147327249022164,
-      "grad_norm": 1.664939284324646,
-      "learning_rate": 9.488643216080402e-05,
-      "loss": 5.747,
-      "step": 5589
-    },
-    {
-      "epoch": 2.915254237288136,
-      "grad_norm": 1.4529097080230713,
-      "learning_rate": 9.48854271356784e-05,
-      "loss": 5.9817,
-      "step": 5590
-    },
-    {
-      "epoch": 2.915775749674055,
-      "grad_norm": 1.486401081085205,
-      "learning_rate": 9.488442211055278e-05,
-      "loss": 6.0475,
-      "step": 5591
-    },
-    {
-      "epoch": 2.916297262059974,
-      "grad_norm": 1.4367727041244507,
-      "learning_rate": 9.488341708542714e-05,
-      "loss": 5.6162,
-      "step": 5592
-    },
-    {
-      "epoch": 2.9168187744458933,
-      "grad_norm": 1.50130033493042,
-      "learning_rate": 9.48824120603015e-05,
-      "loss": 5.8907,
-      "step": 5593
-    },
-    {
-      "epoch": 2.9173402868318123,
-      "grad_norm": 1.5267356634140015,
-      "learning_rate": 9.488140703517588e-05,
-      "loss": 5.7765,
-      "step": 5594
-    },
-    {
-      "epoch": 2.9178617992177314,
-      "grad_norm": 1.402603030204773,
-      "learning_rate": 9.488040201005025e-05,
-      "loss": 5.7163,
-      "step": 5595
-    },
-    {
-      "epoch": 2.918383311603651,
-      "grad_norm": 1.6174359321594238,
-      "learning_rate": 9.487939698492463e-05,
-      "loss": 5.3703,
-      "step": 5596
-    },
-    {
-      "epoch": 2.91890482398957,
-      "grad_norm": 1.472778081893921,
-      "learning_rate": 9.487839195979899e-05,
-      "loss": 5.5394,
-      "step": 5597
-    },
-    {
-      "epoch": 2.919426336375489,
-      "grad_norm": 1.5636790990829468,
-      "learning_rate": 9.487738693467337e-05,
-      "loss": 5.8308,
-      "step": 5598
-    },
-    {
-      "epoch": 2.9199478487614083,
-      "grad_norm": 1.8219919204711914,
-      "learning_rate": 9.487638190954775e-05,
-      "loss": 5.464,
-      "step": 5599
-    },
-    {
-      "epoch": 2.9204693611473274,
-      "grad_norm": 1.4276323318481445,
-      "learning_rate": 9.487537688442212e-05,
-      "loss": 5.9445,
-      "step": 5600
-    },
-    {
-      "epoch": 2.9209908735332464,
-      "grad_norm": 1.461767554283142,
-      "learning_rate": 9.487437185929649e-05,
-      "loss": 5.9794,
-      "step": 5601
-    },
-    {
-      "epoch": 2.921512385919166,
-      "grad_norm": 1.6382936239242554,
-      "learning_rate": 9.487336683417087e-05,
-      "loss": 5.6286,
-      "step": 5602
-    },
-    {
-      "epoch": 2.922033898305085,
-      "grad_norm": 1.7635223865509033,
-      "learning_rate": 9.487236180904523e-05,
-      "loss": 5.6945,
-      "step": 5603
-    },
-    {
-      "epoch": 2.922555410691004,
-      "grad_norm": 1.4411667585372925,
-      "learning_rate": 9.487135678391961e-05,
-      "loss": 5.9232,
-      "step": 5604
-    },
-    {
-      "epoch": 2.9230769230769234,
-      "grad_norm": 1.3225001096725464,
-      "learning_rate": 9.487035175879397e-05,
-      "loss": 6.1514,
-      "step": 5605
-    },
-    {
-      "epoch": 2.9235984354628424,
-      "grad_norm": 1.552101492881775,
-      "learning_rate": 9.486934673366835e-05,
-      "loss": 5.9468,
-      "step": 5606
-    },
-    {
-      "epoch": 2.9241199478487614,
-      "grad_norm": 1.7488456964492798,
-      "learning_rate": 9.486834170854271e-05,
-      "loss": 5.2925,
-      "step": 5607
-    },
-    {
-      "epoch": 2.9246414602346804,
-      "grad_norm": 1.5159887075424194,
-      "learning_rate": 9.486733668341708e-05,
-      "loss": 5.9201,
-      "step": 5608
-    },
-    {
-      "epoch": 2.9251629726206,
-      "grad_norm": 1.5896421670913696,
-      "learning_rate": 9.486633165829146e-05,
-      "loss": 5.7753,
-      "step": 5609
-    },
-    {
-      "epoch": 2.925684485006519,
-      "grad_norm": 1.5869134664535522,
-      "learning_rate": 9.486532663316583e-05,
-      "loss": 5.5884,
-      "step": 5610
-    },
-    {
-      "epoch": 2.926205997392438,
-      "grad_norm": 1.4332365989685059,
-      "learning_rate": 9.486432160804021e-05,
-      "loss": 5.6749,
-      "step": 5611
-    },
-    {
-      "epoch": 2.9267275097783574,
-      "grad_norm": 1.5491622686386108,
-      "learning_rate": 9.486331658291458e-05,
-      "loss": 5.7502,
-      "step": 5612
-    },
-    {
-      "epoch": 2.9272490221642764,
-      "grad_norm": 1.4510512351989746,
-      "learning_rate": 9.486231155778895e-05,
-      "loss": 5.7301,
-      "step": 5613
-    },
-    {
-      "epoch": 2.9277705345501954,
-      "grad_norm": 1.3961373567581177,
-      "learning_rate": 9.486130653266332e-05,
-      "loss": 5.9144,
-      "step": 5614
-    },
-    {
-      "epoch": 2.928292046936115,
-      "grad_norm": 1.7326092720031738,
-      "learning_rate": 9.48603015075377e-05,
-      "loss": 4.9302,
-      "step": 5615
-    },
-    {
-      "epoch": 2.928813559322034,
-      "grad_norm": 1.5977157354354858,
-      "learning_rate": 9.485929648241206e-05,
-      "loss": 5.8305,
-      "step": 5616
-    },
-    {
-      "epoch": 2.929335071707953,
-      "grad_norm": 1.4200847148895264,
-      "learning_rate": 9.485829145728644e-05,
-      "loss": 6.0305,
-      "step": 5617
-    },
-    {
-      "epoch": 2.9298565840938724,
-      "grad_norm": 1.375240445137024,
-      "learning_rate": 9.48572864321608e-05,
-      "loss": 5.9016,
-      "step": 5618
-    },
-    {
-      "epoch": 2.9303780964797914,
-      "grad_norm": 1.414204478263855,
-      "learning_rate": 9.485628140703518e-05,
-      "loss": 6.1293,
-      "step": 5619
-    },
-    {
-      "epoch": 2.9308996088657104,
-      "grad_norm": 1.5879254341125488,
-      "learning_rate": 9.485527638190956e-05,
-      "loss": 5.8432,
-      "step": 5620
-    },
-    {
-      "epoch": 2.9314211212516295,
-      "grad_norm": 1.3899239301681519,
-      "learning_rate": 9.485427135678392e-05,
-      "loss": 5.6386,
-      "step": 5621
-    },
-    {
-      "epoch": 2.931942633637549,
-      "grad_norm": 1.6143795251846313,
-      "learning_rate": 9.48532663316583e-05,
-      "loss": 6.0487,
-      "step": 5622
-    },
-    {
-      "epoch": 2.932464146023468,
-      "grad_norm": 1.525539517402649,
-      "learning_rate": 9.485226130653267e-05,
-      "loss": 5.9726,
-      "step": 5623
-    },
-    {
-      "epoch": 2.932985658409387,
-      "grad_norm": 1.4849265813827515,
-      "learning_rate": 9.485125628140704e-05,
-      "loss": 5.2215,
-      "step": 5624
-    },
-    {
-      "epoch": 2.9335071707953064,
-      "grad_norm": 1.5697646141052246,
-      "learning_rate": 9.485025125628141e-05,
-      "loss": 5.8088,
-      "step": 5625
-    },
-    {
-      "epoch": 2.9340286831812254,
-      "grad_norm": 1.5876095294952393,
-      "learning_rate": 9.484924623115578e-05,
-      "loss": 6.2564,
-      "step": 5626
-    },
-    {
-      "epoch": 2.9345501955671445,
-      "grad_norm": 1.7351067066192627,
-      "learning_rate": 9.484824120603015e-05,
-      "loss": 6.0734,
-      "step": 5627
-    },
-    {
-      "epoch": 2.935071707953064,
-      "grad_norm": 1.461938738822937,
-      "learning_rate": 9.484723618090453e-05,
-      "loss": 5.7064,
-      "step": 5628
-    },
-    {
-      "epoch": 2.935593220338983,
-      "grad_norm": 1.3551006317138672,
-      "learning_rate": 9.484623115577889e-05,
-      "loss": 6.125,
-      "step": 5629
-    },
-    {
-      "epoch": 2.936114732724902,
-      "grad_norm": 1.4544097185134888,
-      "learning_rate": 9.484522613065327e-05,
-      "loss": 5.6953,
-      "step": 5630
-    },
-    {
-      "epoch": 2.9366362451108214,
-      "grad_norm": 1.7247930765151978,
-      "learning_rate": 9.484422110552765e-05,
-      "loss": 5.3649,
-      "step": 5631
-    },
-    {
-      "epoch": 2.9371577574967405,
-      "grad_norm": 1.7665141820907593,
-      "learning_rate": 9.484321608040202e-05,
-      "loss": 5.5419,
-      "step": 5632
-    },
-    {
-      "epoch": 2.9376792698826595,
-      "grad_norm": 1.472756266593933,
-      "learning_rate": 9.484221105527639e-05,
-      "loss": 5.3947,
-      "step": 5633
-    },
-    {
-      "epoch": 2.938200782268579,
-      "grad_norm": 1.4072251319885254,
-      "learning_rate": 9.484120603015075e-05,
-      "loss": 5.791,
-      "step": 5634
-    },
-    {
-      "epoch": 2.938722294654498,
-      "grad_norm": 1.3676085472106934,
-      "learning_rate": 9.484020100502513e-05,
-      "loss": 5.3628,
-      "step": 5635
-    },
-    {
-      "epoch": 2.939243807040417,
-      "grad_norm": 1.455832839012146,
-      "learning_rate": 9.48391959798995e-05,
-      "loss": 6.0159,
-      "step": 5636
-    },
-    {
-      "epoch": 2.9397653194263365,
-      "grad_norm": 1.473459243774414,
-      "learning_rate": 9.483819095477387e-05,
-      "loss": 5.7373,
-      "step": 5637
-    },
-    {
-      "epoch": 2.9402868318122555,
-      "grad_norm": 1.543699026107788,
-      "learning_rate": 9.483718592964824e-05,
-      "loss": 5.8528,
-      "step": 5638
-    },
-    {
-      "epoch": 2.9408083441981745,
-      "grad_norm": 1.4179481267929077,
-      "learning_rate": 9.483618090452262e-05,
-      "loss": 6.2253,
-      "step": 5639
-    },
-    {
-      "epoch": 2.941329856584094,
-      "grad_norm": 1.4675509929656982,
-      "learning_rate": 9.4835175879397e-05,
-      "loss": 5.6815,
-      "step": 5640
-    },
-    {
-      "epoch": 2.941851368970013,
-      "grad_norm": 1.6443326473236084,
-      "learning_rate": 9.483417085427137e-05,
-      "loss": 5.5112,
-      "step": 5641
-    },
-    {
-      "epoch": 2.942372881355932,
-      "grad_norm": 1.6505610942840576,
-      "learning_rate": 9.483316582914574e-05,
-      "loss": 5.8538,
-      "step": 5642
-    },
-    {
-      "epoch": 2.9428943937418515,
-      "grad_norm": 1.365644097328186,
-      "learning_rate": 9.483216080402011e-05,
-      "loss": 5.9482,
-      "step": 5643
-    },
-    {
-      "epoch": 2.9434159061277705,
-      "grad_norm": 1.3721506595611572,
-      "learning_rate": 9.483115577889448e-05,
-      "loss": 5.7317,
-      "step": 5644
-    },
-    {
-      "epoch": 2.9439374185136895,
-      "grad_norm": 1.5284404754638672,
-      "learning_rate": 9.483015075376886e-05,
-      "loss": 5.5124,
-      "step": 5645
-    },
-    {
-      "epoch": 2.944458930899609,
-      "grad_norm": 1.693989634513855,
-      "learning_rate": 9.482914572864322e-05,
-      "loss": 5.5377,
-      "step": 5646
-    },
-    {
-      "epoch": 2.944980443285528,
-      "grad_norm": 1.6544488668441772,
-      "learning_rate": 9.482814070351758e-05,
-      "loss": 5.9952,
-      "step": 5647
-    },
-    {
-      "epoch": 2.945501955671447,
-      "grad_norm": 1.486353874206543,
-      "learning_rate": 9.482713567839196e-05,
-      "loss": 6.2196,
-      "step": 5648
-    },
-    {
-      "epoch": 2.9460234680573665,
-      "grad_norm": 1.6713837385177612,
-      "learning_rate": 9.482613065326633e-05,
-      "loss": 5.5003,
-      "step": 5649
-    },
-    {
-      "epoch": 2.9465449804432855,
-      "grad_norm": 1.5086448192596436,
-      "learning_rate": 9.48251256281407e-05,
-      "loss": 5.8956,
-      "step": 5650
-    },
-    {
-      "epoch": 2.9470664928292045,
-      "grad_norm": 1.4716354608535767,
-      "learning_rate": 9.482412060301508e-05,
-      "loss": 5.8852,
-      "step": 5651
-    },
-    {
-      "epoch": 2.947588005215124,
-      "grad_norm": 1.5243581533432007,
-      "learning_rate": 9.482311557788946e-05,
-      "loss": 5.7754,
-      "step": 5652
-    },
-    {
-      "epoch": 2.948109517601043,
-      "grad_norm": 1.5727814435958862,
-      "learning_rate": 9.482211055276382e-05,
-      "loss": 5.9325,
-      "step": 5653
-    },
-    {
-      "epoch": 2.948631029986962,
-      "grad_norm": 1.5930489301681519,
-      "learning_rate": 9.48211055276382e-05,
-      "loss": 5.5824,
-      "step": 5654
-    },
-    {
-      "epoch": 2.9491525423728815,
-      "grad_norm": 1.499829888343811,
-      "learning_rate": 9.482010050251257e-05,
-      "loss": 5.6296,
-      "step": 5655
-    },
-    {
-      "epoch": 2.9496740547588005,
-      "grad_norm": 1.6006007194519043,
-      "learning_rate": 9.481909547738694e-05,
-      "loss": 5.3989,
-      "step": 5656
-    },
-    {
-      "epoch": 2.9501955671447195,
-      "grad_norm": 1.5014506578445435,
-      "learning_rate": 9.481809045226131e-05,
-      "loss": 5.7784,
-      "step": 5657
-    },
-    {
-      "epoch": 2.950717079530639,
-      "grad_norm": 1.5451650619506836,
-      "learning_rate": 9.481708542713569e-05,
-      "loss": 5.7072,
-      "step": 5658
-    },
-    {
-      "epoch": 2.951238591916558,
-      "grad_norm": 1.3398109674453735,
-      "learning_rate": 9.481608040201005e-05,
-      "loss": 5.937,
-      "step": 5659
-    },
-    {
-      "epoch": 2.951760104302477,
-      "grad_norm": 1.3988779783248901,
-      "learning_rate": 9.481507537688443e-05,
-      "loss": 5.9181,
-      "step": 5660
-    },
-    {
-      "epoch": 2.9522816166883965,
-      "grad_norm": 1.4654664993286133,
-      "learning_rate": 9.48140703517588e-05,
-      "loss": 6.2438,
-      "step": 5661
-    },
-    {
-      "epoch": 2.9528031290743155,
-      "grad_norm": 1.4928780794143677,
-      "learning_rate": 9.481306532663317e-05,
-      "loss": 5.0629,
-      "step": 5662
-    },
-    {
-      "epoch": 2.9533246414602345,
-      "grad_norm": 1.5862302780151367,
-      "learning_rate": 9.481206030150755e-05,
-      "loss": 5.6133,
-      "step": 5663
-    },
-    {
-      "epoch": 2.953846153846154,
-      "grad_norm": 1.4297378063201904,
-      "learning_rate": 9.481105527638191e-05,
-      "loss": 5.9868,
-      "step": 5664
-    },
-    {
-      "epoch": 2.954367666232073,
-      "grad_norm": 1.5494303703308105,
-      "learning_rate": 9.481005025125629e-05,
-      "loss": 5.7933,
-      "step": 5665
-    },
-    {
-      "epoch": 2.954889178617992,
-      "grad_norm": 1.5773588418960571,
-      "learning_rate": 9.480904522613065e-05,
-      "loss": 5.092,
-      "step": 5666
-    },
-    {
-      "epoch": 2.9554106910039115,
-      "grad_norm": 1.5671412944793701,
-      "learning_rate": 9.480804020100503e-05,
-      "loss": 5.5462,
-      "step": 5667
-    },
-    {
-      "epoch": 2.9559322033898305,
-      "grad_norm": 1.572758674621582,
-      "learning_rate": 9.48070351758794e-05,
-      "loss": 6.1114,
-      "step": 5668
-    },
-    {
-      "epoch": 2.9564537157757496,
-      "grad_norm": 1.4552154541015625,
-      "learning_rate": 9.480603015075377e-05,
-      "loss": 5.6481,
-      "step": 5669
-    },
-    {
-      "epoch": 2.956975228161669,
-      "grad_norm": 1.5403410196304321,
-      "learning_rate": 9.480502512562814e-05,
-      "loss": 5.3032,
-      "step": 5670
-    },
-    {
-      "epoch": 2.957496740547588,
-      "grad_norm": 1.6212360858917236,
-      "learning_rate": 9.480402010050252e-05,
-      "loss": 5.7797,
-      "step": 5671
-    },
-    {
-      "epoch": 2.958018252933507,
-      "grad_norm": 1.4734113216400146,
-      "learning_rate": 9.48030150753769e-05,
-      "loss": 5.4639,
-      "step": 5672
-    },
-    {
-      "epoch": 2.9585397653194265,
-      "grad_norm": 1.2675044536590576,
-      "learning_rate": 9.480201005025126e-05,
-      "loss": 6.1917,
-      "step": 5673
-    },
-    {
-      "epoch": 2.9590612777053455,
-      "grad_norm": 1.4518061876296997,
-      "learning_rate": 9.480100502512564e-05,
-      "loss": 5.8391,
-      "step": 5674
-    },
-    {
-      "epoch": 2.9595827900912646,
-      "grad_norm": 1.5727113485336304,
-      "learning_rate": 9.48e-05,
-      "loss": 5.9255,
-      "step": 5675
-    },
-    {
-      "epoch": 2.960104302477184,
-      "grad_norm": 1.3488913774490356,
-      "learning_rate": 9.479899497487438e-05,
-      "loss": 5.9383,
-      "step": 5676
-    },
-    {
-      "epoch": 2.960625814863103,
-      "grad_norm": 1.4377706050872803,
-      "learning_rate": 9.479798994974874e-05,
-      "loss": 5.9509,
-      "step": 5677
-    },
-    {
-      "epoch": 2.961147327249022,
-      "grad_norm": 1.599881649017334,
-      "learning_rate": 9.479698492462312e-05,
-      "loss": 5.8221,
-      "step": 5678
-    },
-    {
-      "epoch": 2.9616688396349415,
-      "grad_norm": 1.5068464279174805,
-      "learning_rate": 9.479597989949748e-05,
-      "loss": 5.9585,
-      "step": 5679
-    },
-    {
-      "epoch": 2.9621903520208606,
-      "grad_norm": 1.545324444770813,
-      "learning_rate": 9.479497487437186e-05,
-      "loss": 5.7056,
-      "step": 5680
-    },
-    {
-      "epoch": 2.9627118644067796,
-      "grad_norm": 1.5406014919281006,
-      "learning_rate": 9.479396984924624e-05,
-      "loss": 5.1505,
-      "step": 5681
-    },
-    {
-      "epoch": 2.963233376792699,
-      "grad_norm": 1.4956676959991455,
-      "learning_rate": 9.479296482412062e-05,
-      "loss": 5.7456,
-      "step": 5682
-    },
-    {
-      "epoch": 2.963754889178618,
-      "grad_norm": 1.6768978834152222,
-      "learning_rate": 9.479195979899498e-05,
-      "loss": 5.4777,
-      "step": 5683
-    },
-    {
-      "epoch": 2.964276401564537,
-      "grad_norm": 1.5714713335037231,
-      "learning_rate": 9.479095477386936e-05,
-      "loss": 5.7941,
-      "step": 5684
-    },
-    {
-      "epoch": 2.9647979139504566,
-      "grad_norm": 1.4647243022918701,
-      "learning_rate": 9.478994974874372e-05,
-      "loss": 5.8484,
-      "step": 5685
-    },
-    {
-      "epoch": 2.9653194263363756,
-      "grad_norm": 1.5708014965057373,
-      "learning_rate": 9.47889447236181e-05,
-      "loss": 5.4629,
-      "step": 5686
-    },
-    {
-      "epoch": 2.9658409387222946,
-      "grad_norm": 1.4093737602233887,
-      "learning_rate": 9.478793969849247e-05,
-      "loss": 6.0379,
-      "step": 5687
-    },
-    {
-      "epoch": 2.966362451108214,
-      "grad_norm": 1.462388038635254,
-      "learning_rate": 9.478693467336683e-05,
-      "loss": 5.891,
-      "step": 5688
-    },
-    {
-      "epoch": 2.966883963494133,
-      "grad_norm": 1.395670771598816,
-      "learning_rate": 9.478592964824121e-05,
-      "loss": 5.9153,
-      "step": 5689
-    },
-    {
-      "epoch": 2.967405475880052,
-      "grad_norm": 1.5872045755386353,
-      "learning_rate": 9.478492462311557e-05,
-      "loss": 5.8976,
-      "step": 5690
-    },
-    {
-      "epoch": 2.9679269882659716,
-      "grad_norm": 1.495665192604065,
-      "learning_rate": 9.478391959798995e-05,
-      "loss": 5.6124,
-      "step": 5691
-    },
-    {
-      "epoch": 2.9684485006518906,
-      "grad_norm": 1.4160020351409912,
-      "learning_rate": 9.478291457286433e-05,
-      "loss": 6.3358,
-      "step": 5692
-    },
-    {
-      "epoch": 2.9689700130378096,
-      "grad_norm": 1.5315049886703491,
-      "learning_rate": 9.47819095477387e-05,
-      "loss": 5.8767,
-      "step": 5693
-    },
-    {
-      "epoch": 2.969491525423729,
-      "grad_norm": 1.5887882709503174,
-      "learning_rate": 9.478090452261307e-05,
-      "loss": 5.7797,
-      "step": 5694
-    },
-    {
-      "epoch": 2.970013037809648,
-      "grad_norm": 1.4905885457992554,
-      "learning_rate": 9.477989949748745e-05,
-      "loss": 5.9486,
-      "step": 5695
-    },
-    {
-      "epoch": 2.970534550195567,
-      "grad_norm": 1.5904028415679932,
-      "learning_rate": 9.477889447236181e-05,
-      "loss": 5.572,
-      "step": 5696
-    },
-    {
-      "epoch": 2.9710560625814866,
-      "grad_norm": 1.4948248863220215,
-      "learning_rate": 9.477788944723619e-05,
-      "loss": 5.6951,
-      "step": 5697
-    },
-    {
-      "epoch": 2.9715775749674056,
-      "grad_norm": 1.6443378925323486,
-      "learning_rate": 9.477688442211055e-05,
-      "loss": 5.5549,
-      "step": 5698
-    },
-    {
-      "epoch": 2.9720990873533246,
-      "grad_norm": 1.3996539115905762,
-      "learning_rate": 9.477587939698493e-05,
-      "loss": 5.9622,
-      "step": 5699
-    },
-    {
-      "epoch": 2.9726205997392436,
-      "grad_norm": 1.5761632919311523,
-      "learning_rate": 9.47748743718593e-05,
-      "loss": 5.3111,
-      "step": 5700
-    },
-    {
-      "epoch": 2.973142112125163,
-      "grad_norm": 1.4086540937423706,
-      "learning_rate": 9.477386934673366e-05,
-      "loss": 6.1915,
-      "step": 5701
-    },
-    {
-      "epoch": 2.973663624511082,
-      "grad_norm": 1.6071149110794067,
-      "learning_rate": 9.477286432160804e-05,
-      "loss": 5.6636,
-      "step": 5702
-    },
-    {
-      "epoch": 2.974185136897001,
-      "grad_norm": 1.5928254127502441,
-      "learning_rate": 9.477185929648242e-05,
-      "loss": 5.4637,
-      "step": 5703
-    },
-    {
-      "epoch": 2.9747066492829206,
-      "grad_norm": 1.5876655578613281,
-      "learning_rate": 9.47708542713568e-05,
-      "loss": 5.4165,
-      "step": 5704
-    },
-    {
-      "epoch": 2.9752281616688396,
-      "grad_norm": 1.4683613777160645,
-      "learning_rate": 9.476984924623116e-05,
-      "loss": 5.7055,
-      "step": 5705
-    },
-    {
-      "epoch": 2.9757496740547587,
-      "grad_norm": 1.523417353630066,
-      "learning_rate": 9.476884422110554e-05,
-      "loss": 5.5511,
-      "step": 5706
-    },
-    {
-      "epoch": 2.976271186440678,
-      "grad_norm": 1.6262173652648926,
-      "learning_rate": 9.47678391959799e-05,
-      "loss": 5.789,
-      "step": 5707
-    },
-    {
-      "epoch": 2.976792698826597,
-      "grad_norm": 1.5964269638061523,
-      "learning_rate": 9.476683417085428e-05,
-      "loss": 5.5677,
-      "step": 5708
-    },
-    {
-      "epoch": 2.977314211212516,
-      "grad_norm": 1.3546854257583618,
-      "learning_rate": 9.476582914572864e-05,
-      "loss": 5.9315,
-      "step": 5709
-    },
-    {
-      "epoch": 2.9778357235984356,
-      "grad_norm": 1.3683171272277832,
-      "learning_rate": 9.476482412060302e-05,
-      "loss": 5.9894,
-      "step": 5710
-    },
-    {
-      "epoch": 2.9783572359843546,
-      "grad_norm": 1.3763943910598755,
-      "learning_rate": 9.476381909547739e-05,
-      "loss": 6.0469,
-      "step": 5711
-    },
-    {
-      "epoch": 2.9788787483702737,
-      "grad_norm": 1.3965799808502197,
-      "learning_rate": 9.476281407035176e-05,
-      "loss": 6.0733,
-      "step": 5712
-    },
-    {
-      "epoch": 2.9794002607561927,
-      "grad_norm": 1.6557425260543823,
-      "learning_rate": 9.476180904522614e-05,
-      "loss": 5.1168,
-      "step": 5713
-    },
-    {
-      "epoch": 2.979921773142112,
-      "grad_norm": 1.378940224647522,
-      "learning_rate": 9.47608040201005e-05,
-      "loss": 5.8397,
-      "step": 5714
-    },
-    {
-      "epoch": 2.980443285528031,
-      "grad_norm": 1.3388956785202026,
-      "learning_rate": 9.475979899497488e-05,
-      "loss": 5.8605,
-      "step": 5715
-    },
-    {
-      "epoch": 2.98096479791395,
-      "grad_norm": 1.4287981986999512,
-      "learning_rate": 9.475879396984925e-05,
-      "loss": 6.075,
-      "step": 5716
-    },
-    {
-      "epoch": 2.9814863102998697,
-      "grad_norm": 1.383670449256897,
-      "learning_rate": 9.475778894472363e-05,
-      "loss": 5.9189,
-      "step": 5717
-    },
-    {
-      "epoch": 2.9820078226857887,
-      "grad_norm": 1.4722269773483276,
-      "learning_rate": 9.475678391959799e-05,
-      "loss": 5.6182,
-      "step": 5718
-    },
-    {
-      "epoch": 2.9825293350717077,
-      "grad_norm": 1.6591142416000366,
-      "learning_rate": 9.475577889447237e-05,
-      "loss": 5.6252,
-      "step": 5719
-    },
-    {
-      "epoch": 2.983050847457627,
-      "grad_norm": 1.6826410293579102,
-      "learning_rate": 9.475477386934673e-05,
-      "loss": 5.4618,
-      "step": 5720
-    },
-    {
-      "epoch": 2.983572359843546,
-      "grad_norm": 1.5898313522338867,
-      "learning_rate": 9.475376884422111e-05,
-      "loss": 5.5191,
-      "step": 5721
-    },
-    {
-      "epoch": 2.984093872229465,
-      "grad_norm": 1.4392000436782837,
-      "learning_rate": 9.475276381909547e-05,
-      "loss": 6.0886,
-      "step": 5722
-    },
-    {
-      "epoch": 2.9846153846153847,
-      "grad_norm": 1.5297683477401733,
-      "learning_rate": 9.475175879396985e-05,
-      "loss": 4.875,
-      "step": 5723
-    },
-    {
-      "epoch": 2.9851368970013037,
-      "grad_norm": 2.012697696685791,
-      "learning_rate": 9.475075376884423e-05,
-      "loss": 5.5471,
-      "step": 5724
-    },
-    {
-      "epoch": 2.9856584093872227,
-      "grad_norm": 1.5634838342666626,
-      "learning_rate": 9.474974874371861e-05,
-      "loss": 5.6011,
-      "step": 5725
-    },
-    {
-      "epoch": 2.986179921773142,
-      "grad_norm": 2.0749900341033936,
-      "learning_rate": 9.474874371859297e-05,
-      "loss": 4.8957,
-      "step": 5726
-    },
-    {
-      "epoch": 2.986701434159061,
-      "grad_norm": 1.7019398212432861,
-      "learning_rate": 9.474773869346734e-05,
-      "loss": 5.7448,
-      "step": 5727
-    },
-    {
-      "epoch": 2.98722294654498,
-      "grad_norm": 1.5106923580169678,
-      "learning_rate": 9.474673366834171e-05,
-      "loss": 5.8263,
-      "step": 5728
-    },
-    {
-      "epoch": 2.9877444589308997,
-      "grad_norm": 1.9888776540756226,
-      "learning_rate": 9.474572864321608e-05,
-      "loss": 5.0353,
-      "step": 5729
-    },
-    {
-      "epoch": 2.9882659713168187,
-      "grad_norm": 1.6606978178024292,
-      "learning_rate": 9.474472361809046e-05,
-      "loss": 5.5123,
-      "step": 5730
-    },
-    {
-      "epoch": 2.9887874837027377,
-      "grad_norm": 1.3773466348648071,
-      "learning_rate": 9.474371859296482e-05,
-      "loss": 5.9856,
-      "step": 5731
-    },
-    {
-      "epoch": 2.989308996088657,
-      "grad_norm": 1.4099878072738647,
-      "learning_rate": 9.47427135678392e-05,
-      "loss": 6.1027,
-      "step": 5732
-    },
-    {
-      "epoch": 2.989830508474576,
-      "grad_norm": 1.4960269927978516,
-      "learning_rate": 9.474170854271358e-05,
-      "loss": 5.488,
-      "step": 5733
-    },
-    {
-      "epoch": 2.9903520208604952,
-      "grad_norm": 1.3266113996505737,
-      "learning_rate": 9.474070351758795e-05,
-      "loss": 5.6943,
-      "step": 5734
-    },
-    {
-      "epoch": 2.9908735332464147,
-      "grad_norm": 1.333306908607483,
-      "learning_rate": 9.473969849246232e-05,
-      "loss": 5.8785,
-      "step": 5735
-    },
-    {
-      "epoch": 2.9913950456323337,
-      "grad_norm": 1.577088475227356,
-      "learning_rate": 9.47386934673367e-05,
-      "loss": 5.628,
-      "step": 5736
-    },
-    {
-      "epoch": 2.9919165580182527,
-      "grad_norm": 1.6364824771881104,
-      "learning_rate": 9.473768844221106e-05,
-      "loss": 5.3675,
-      "step": 5737
-    },
-    {
-      "epoch": 2.992438070404172,
-      "grad_norm": 1.5676016807556152,
-      "learning_rate": 9.473668341708544e-05,
-      "loss": 5.5322,
-      "step": 5738
-    },
-    {
-      "epoch": 2.992959582790091,
-      "grad_norm": 1.3868201971054077,
-      "learning_rate": 9.47356783919598e-05,
-      "loss": 5.7354,
-      "step": 5739
-    },
-    {
-      "epoch": 2.9934810951760102,
-      "grad_norm": 1.366782307624817,
-      "learning_rate": 9.473467336683417e-05,
-      "loss": 6.0451,
-      "step": 5740
-    },
-    {
-      "epoch": 2.9940026075619297,
-      "grad_norm": 1.390046238899231,
-      "learning_rate": 9.473366834170854e-05,
-      "loss": 6.0958,
-      "step": 5741
-    },
-    {
-      "epoch": 2.9945241199478487,
-      "grad_norm": 1.3986254930496216,
-      "learning_rate": 9.473266331658291e-05,
-      "loss": 6.0261,
-      "step": 5742
-    },
-    {
-      "epoch": 2.9950456323337677,
-      "grad_norm": 1.651505947113037,
-      "learning_rate": 9.473165829145729e-05,
-      "loss": 5.7223,
-      "step": 5743
-    },
-    {
-      "epoch": 2.995567144719687,
-      "grad_norm": 1.5541599988937378,
-      "learning_rate": 9.473065326633166e-05,
-      "loss": 5.4382,
-      "step": 5744
-    },
-    {
-      "epoch": 2.9960886571056062,
-      "grad_norm": 1.4342402219772339,
-      "learning_rate": 9.472964824120604e-05,
-      "loss": 6.1019,
-      "step": 5745
-    },
-    {
-      "epoch": 2.9966101694915253,
-      "grad_norm": 1.5408830642700195,
-      "learning_rate": 9.47286432160804e-05,
-      "loss": 5.6497,
-      "step": 5746
-    },
-    {
-      "epoch": 2.9971316818774447,
-      "grad_norm": 1.456231713294983,
-      "learning_rate": 9.472763819095478e-05,
-      "loss": 6.179,
-      "step": 5747
-    },
-    {
-      "epoch": 2.9976531942633637,
-      "grad_norm": 1.5744142532348633,
-      "learning_rate": 9.472663316582915e-05,
-      "loss": 5.7413,
-      "step": 5748
-    },
-    {
-      "epoch": 2.9981747066492828,
-      "grad_norm": 1.5043144226074219,
-      "learning_rate": 9.472562814070353e-05,
-      "loss": 5.3917,
-      "step": 5749
-    },
-    {
-      "epoch": 2.9986962190352022,
-      "grad_norm": 1.4089245796203613,
-      "learning_rate": 9.472462311557789e-05,
-      "loss": 6.1994,
-      "step": 5750
-    },
-    {
-      "epoch": 2.9992177314211212,
-      "grad_norm": 1.372663140296936,
-      "learning_rate": 9.472361809045227e-05,
-      "loss": 5.8687,
-      "step": 5751
-    },
-    {
-      "epoch": 2.9997392438070403,
-      "grad_norm": 1.5765117406845093,
-      "learning_rate": 9.472261306532663e-05,
-      "loss": 5.8483,
-      "step": 5752
-    },
-    {
-      "epoch": 3.0002607561929597,
-      "grad_norm": 1.4377204179763794,
-      "learning_rate": 9.472160804020101e-05,
-      "loss": 5.9295,
-      "step": 5753
-    },
-    {
-      "epoch": 3.0007822685788788,
-      "grad_norm": 1.3878682851791382,
-      "learning_rate": 9.472060301507539e-05,
-      "loss": 6.1257,
-      "step": 5754
-    },
-    {
-      "epoch": 3.0013037809647978,
-      "grad_norm": 1.4637115001678467,
-      "learning_rate": 9.471959798994975e-05,
-      "loss": 6.0069,
-      "step": 5755
-    },
-    {
-      "epoch": 3.0018252933507172,
-      "grad_norm": 1.5836331844329834,
-      "learning_rate": 9.471859296482413e-05,
-      "loss": 5.7355,
-      "step": 5756
-    },
-    {
-      "epoch": 3.0023468057366363,
-      "grad_norm": 1.6220051050186157,
-      "learning_rate": 9.47175879396985e-05,
-      "loss": 5.6817,
-      "step": 5757
-    },
-    {
-      "epoch": 3.0028683181225553,
-      "grad_norm": 1.4258496761322021,
-      "learning_rate": 9.471658291457287e-05,
-      "loss": 6.0942,
-      "step": 5758
-    },
-    {
-      "epoch": 3.0033898305084747,
-      "grad_norm": 1.5069963932037354,
-      "learning_rate": 9.471557788944724e-05,
-      "loss": 5.5966,
-      "step": 5759
-    },
-    {
-      "epoch": 3.0039113428943938,
-      "grad_norm": 1.6024360656738281,
-      "learning_rate": 9.471457286432161e-05,
-      "loss": 5.8251,
-      "step": 5760
-    },
-    {
-      "epoch": 3.004432855280313,
-      "grad_norm": 1.5853402614593506,
-      "learning_rate": 9.471356783919598e-05,
-      "loss": 5.7461,
-      "step": 5761
-    },
-    {
-      "epoch": 3.0049543676662323,
-      "grad_norm": 1.4520171880722046,
-      "learning_rate": 9.471256281407036e-05,
-      "loss": 6.1088,
-      "step": 5762
-    },
-    {
-      "epoch": 3.0054758800521513,
-      "grad_norm": 1.450812816619873,
-      "learning_rate": 9.471155778894472e-05,
-      "loss": 5.6208,
-      "step": 5763
-    },
-    {
-      "epoch": 3.0059973924380703,
-      "grad_norm": 1.7311595678329468,
-      "learning_rate": 9.47105527638191e-05,
-      "loss": 5.4351,
-      "step": 5764
-    },
-    {
-      "epoch": 3.0065189048239898,
-      "grad_norm": 1.3856539726257324,
-      "learning_rate": 9.470954773869348e-05,
-      "loss": 5.4222,
-      "step": 5765
-    },
-    {
-      "epoch": 3.007040417209909,
-      "grad_norm": 1.4523017406463623,
-      "learning_rate": 9.470854271356784e-05,
-      "loss": 5.6188,
-      "step": 5766
-    },
-    {
-      "epoch": 3.007561929595828,
-      "grad_norm": 1.5164885520935059,
-      "learning_rate": 9.470753768844222e-05,
-      "loss": 5.792,
-      "step": 5767
-    },
-    {
-      "epoch": 3.0080834419817473,
-      "grad_norm": 1.5523091554641724,
-      "learning_rate": 9.470653266331658e-05,
-      "loss": 5.5296,
-      "step": 5768
-    },
-    {
-      "epoch": 3.0086049543676663,
-      "grad_norm": 1.5385515689849854,
-      "learning_rate": 9.470552763819096e-05,
-      "loss": 5.3755,
-      "step": 5769
-    },
-    {
-      "epoch": 3.0091264667535853,
-      "grad_norm": 1.4706642627716064,
-      "learning_rate": 9.470452261306532e-05,
-      "loss": 5.9415,
-      "step": 5770
-    },
-    {
-      "epoch": 3.0096479791395048,
-      "grad_norm": 1.4606150388717651,
-      "learning_rate": 9.47035175879397e-05,
-      "loss": 5.9363,
-      "step": 5771
-    },
-    {
-      "epoch": 3.010169491525424,
-      "grad_norm": 1.360920786857605,
-      "learning_rate": 9.470251256281407e-05,
-      "loss": 5.6328,
-      "step": 5772
-    },
-    {
-      "epoch": 3.010691003911343,
-      "grad_norm": 1.5204272270202637,
-      "learning_rate": 9.470150753768844e-05,
-      "loss": 5.9155,
-      "step": 5773
-    },
-    {
-      "epoch": 3.0112125162972623,
-      "grad_norm": 1.4005473852157593,
-      "learning_rate": 9.470050251256282e-05,
-      "loss": 5.949,
-      "step": 5774
-    },
-    {
-      "epoch": 3.0117340286831813,
-      "grad_norm": 1.3721369504928589,
-      "learning_rate": 9.46994974874372e-05,
-      "loss": 6.1523,
-      "step": 5775
-    },
-    {
-      "epoch": 3.0122555410691003,
-      "grad_norm": 1.598720908164978,
-      "learning_rate": 9.469849246231156e-05,
-      "loss": 5.3843,
-      "step": 5776
-    },
-    {
-      "epoch": 3.0127770534550193,
-      "grad_norm": 1.5146353244781494,
-      "learning_rate": 9.469748743718594e-05,
-      "loss": 5.4694,
-      "step": 5777
-    },
-    {
-      "epoch": 3.013298565840939,
-      "grad_norm": 1.5398303270339966,
-      "learning_rate": 9.469648241206031e-05,
-      "loss": 5.889,
-      "step": 5778
-    },
-    {
-      "epoch": 3.013820078226858,
-      "grad_norm": 1.4430100917816162,
-      "learning_rate": 9.469547738693468e-05,
-      "loss": 5.8776,
-      "step": 5779
-    },
-    {
-      "epoch": 3.014341590612777,
-      "grad_norm": 1.5189950466156006,
-      "learning_rate": 9.469447236180905e-05,
-      "loss": 6.0229,
-      "step": 5780
-    },
-    {
-      "epoch": 3.0148631029986963,
-      "grad_norm": 1.3621580600738525,
-      "learning_rate": 9.469346733668341e-05,
-      "loss": 5.2735,
-      "step": 5781
-    },
-    {
-      "epoch": 3.0153846153846153,
-      "grad_norm": 1.4718878269195557,
-      "learning_rate": 9.469246231155779e-05,
-      "loss": 5.9845,
-      "step": 5782
-    },
-    {
-      "epoch": 3.0159061277705344,
-      "grad_norm": 1.358791708946228,
-      "learning_rate": 9.469145728643216e-05,
-      "loss": 5.2418,
-      "step": 5783
-    },
-    {
-      "epoch": 3.016427640156454,
-      "grad_norm": 1.4125661849975586,
-      "learning_rate": 9.469045226130653e-05,
-      "loss": 5.902,
-      "step": 5784
-    },
-    {
-      "epoch": 3.016949152542373,
-      "grad_norm": 1.5206825733184814,
-      "learning_rate": 9.468944723618091e-05,
-      "loss": 5.7103,
-      "step": 5785
-    },
-    {
-      "epoch": 3.017470664928292,
-      "grad_norm": 1.505163311958313,
-      "learning_rate": 9.468844221105529e-05,
-      "loss": 5.9654,
-      "step": 5786
-    },
-    {
-      "epoch": 3.0179921773142113,
-      "grad_norm": 1.4645029306411743,
-      "learning_rate": 9.468743718592965e-05,
-      "loss": 5.945,
-      "step": 5787
-    },
-    {
-      "epoch": 3.0185136897001303,
-      "grad_norm": 1.6592308282852173,
-      "learning_rate": 9.468643216080403e-05,
-      "loss": 5.6577,
-      "step": 5788
-    },
-    {
-      "epoch": 3.0190352020860494,
-      "grad_norm": 1.4215037822723389,
-      "learning_rate": 9.46854271356784e-05,
-      "loss": 6.1287,
-      "step": 5789
-    },
-    {
-      "epoch": 3.019556714471969,
-      "grad_norm": 1.3522984981536865,
-      "learning_rate": 9.468442211055277e-05,
-      "loss": 5.6766,
-      "step": 5790
-    },
-    {
-      "epoch": 3.020078226857888,
-      "grad_norm": 1.4199585914611816,
-      "learning_rate": 9.468341708542714e-05,
-      "loss": 5.8297,
-      "step": 5791
-    },
-    {
-      "epoch": 3.020599739243807,
-      "grad_norm": 1.475765585899353,
-      "learning_rate": 9.468241206030152e-05,
-      "loss": 5.9101,
-      "step": 5792
-    },
-    {
-      "epoch": 3.0211212516297263,
-      "grad_norm": 1.586753010749817,
-      "learning_rate": 9.468140703517588e-05,
-      "loss": 5.7375,
-      "step": 5793
-    },
-    {
-      "epoch": 3.0216427640156454,
-      "grad_norm": 1.5540598630905151,
-      "learning_rate": 9.468040201005026e-05,
-      "loss": 5.4643,
-      "step": 5794
-    },
-    {
-      "epoch": 3.0221642764015644,
-      "grad_norm": 1.4363815784454346,
-      "learning_rate": 9.467939698492463e-05,
-      "loss": 5.5748,
-      "step": 5795
-    },
-    {
-      "epoch": 3.022685788787484,
-      "grad_norm": 1.395122766494751,
-      "learning_rate": 9.4678391959799e-05,
-      "loss": 5.4364,
-      "step": 5796
-    },
-    {
-      "epoch": 3.023207301173403,
-      "grad_norm": 1.4867584705352783,
-      "learning_rate": 9.467738693467338e-05,
-      "loss": 6.1721,
-      "step": 5797
-    },
-    {
-      "epoch": 3.023728813559322,
-      "grad_norm": 1.4129140377044678,
-      "learning_rate": 9.467638190954774e-05,
-      "loss": 5.6208,
-      "step": 5798
-    },
-    {
-      "epoch": 3.0242503259452413,
-      "grad_norm": 1.4504239559173584,
-      "learning_rate": 9.467537688442212e-05,
-      "loss": 5.6813,
-      "step": 5799
-    },
-    {
-      "epoch": 3.0247718383311604,
-      "grad_norm": 1.2899487018585205,
-      "learning_rate": 9.467437185929648e-05,
-      "loss": 6.0924,
-      "step": 5800
-    },
-    {
-      "epoch": 3.0252933507170794,
-      "grad_norm": 1.5992485284805298,
-      "learning_rate": 9.467336683417086e-05,
-      "loss": 5.3625,
-      "step": 5801
-    },
-    {
-      "epoch": 3.025814863102999,
-      "grad_norm": 1.5796823501586914,
-      "learning_rate": 9.467236180904523e-05,
-      "loss": 5.3334,
-      "step": 5802
-    },
-    {
-      "epoch": 3.026336375488918,
-      "grad_norm": 1.6074366569519043,
-      "learning_rate": 9.46713567839196e-05,
-      "loss": 5.4262,
-      "step": 5803
-    },
-    {
-      "epoch": 3.026857887874837,
-      "grad_norm": 1.643280029296875,
-      "learning_rate": 9.467035175879397e-05,
-      "loss": 5.3232,
-      "step": 5804
-    },
-    {
-      "epoch": 3.0273794002607564,
-      "grad_norm": 1.5248814821243286,
-      "learning_rate": 9.466934673366835e-05,
-      "loss": 6.1436,
-      "step": 5805
-    },
-    {
-      "epoch": 3.0279009126466754,
-      "grad_norm": 1.3710942268371582,
-      "learning_rate": 9.466834170854272e-05,
-      "loss": 6.0537,
-      "step": 5806
-    },
-    {
-      "epoch": 3.0284224250325944,
-      "grad_norm": 1.504469394683838,
-      "learning_rate": 9.466733668341709e-05,
-      "loss": 5.7787,
-      "step": 5807
-    },
-    {
-      "epoch": 3.028943937418514,
-      "grad_norm": 1.4599465131759644,
-      "learning_rate": 9.466633165829147e-05,
-      "loss": 5.7793,
-      "step": 5808
-    },
-    {
-      "epoch": 3.029465449804433,
-      "grad_norm": 1.4530867338180542,
-      "learning_rate": 9.466532663316583e-05,
-      "loss": 5.881,
-      "step": 5809
-    },
-    {
-      "epoch": 3.029986962190352,
-      "grad_norm": 1.5725669860839844,
-      "learning_rate": 9.466432160804021e-05,
-      "loss": 5.5362,
-      "step": 5810
-    },
-    {
-      "epoch": 3.0305084745762714,
-      "grad_norm": 1.4958685636520386,
-      "learning_rate": 9.466331658291457e-05,
-      "loss": 5.4548,
-      "step": 5811
-    },
-    {
-      "epoch": 3.0310299869621904,
-      "grad_norm": 1.6312061548233032,
-      "learning_rate": 9.466231155778895e-05,
-      "loss": 5.6883,
-      "step": 5812
-    },
-    {
-      "epoch": 3.0315514993481094,
-      "grad_norm": 1.619693398475647,
-      "learning_rate": 9.466130653266331e-05,
-      "loss": 5.4717,
-      "step": 5813
-    },
-    {
-      "epoch": 3.032073011734029,
-      "grad_norm": 1.4062529802322388,
-      "learning_rate": 9.466030150753769e-05,
-      "loss": 5.8903,
-      "step": 5814
-    },
-    {
-      "epoch": 3.032594524119948,
-      "grad_norm": 1.5683705806732178,
-      "learning_rate": 9.465929648241207e-05,
-      "loss": 5.7117,
-      "step": 5815
-    },
-    {
-      "epoch": 3.033116036505867,
-      "grad_norm": 1.5052239894866943,
-      "learning_rate": 9.465829145728645e-05,
-      "loss": 5.6858,
-      "step": 5816
-    },
-    {
-      "epoch": 3.0336375488917864,
-      "grad_norm": 1.899996042251587,
-      "learning_rate": 9.465728643216081e-05,
-      "loss": 4.9304,
-      "step": 5817
-    },
-    {
-      "epoch": 3.0341590612777054,
-      "grad_norm": 1.5951212644577026,
-      "learning_rate": 9.465628140703519e-05,
-      "loss": 5.7052,
-      "step": 5818
-    },
-    {
-      "epoch": 3.0346805736636244,
-      "grad_norm": 1.5659523010253906,
-      "learning_rate": 9.465527638190955e-05,
-      "loss": 5.6977,
-      "step": 5819
-    },
-    {
-      "epoch": 3.035202086049544,
-      "grad_norm": 1.3789373636245728,
-      "learning_rate": 9.465427135678392e-05,
-      "loss": 6.1535,
-      "step": 5820
-    },
-    {
-      "epoch": 3.035723598435463,
-      "grad_norm": 1.4808989763259888,
-      "learning_rate": 9.46532663316583e-05,
-      "loss": 5.886,
-      "step": 5821
-    },
-    {
-      "epoch": 3.036245110821382,
-      "grad_norm": 1.3488909006118774,
-      "learning_rate": 9.465226130653266e-05,
-      "loss": 6.0475,
-      "step": 5822
-    },
-    {
-      "epoch": 3.036766623207301,
-      "grad_norm": 1.7176576852798462,
-      "learning_rate": 9.465125628140704e-05,
-      "loss": 5.4142,
-      "step": 5823
-    },
-    {
-      "epoch": 3.0372881355932204,
-      "grad_norm": 1.5322469472885132,
-      "learning_rate": 9.46502512562814e-05,
-      "loss": 5.9665,
-      "step": 5824
-    },
-    {
-      "epoch": 3.0378096479791394,
-      "grad_norm": 1.3634486198425293,
-      "learning_rate": 9.464924623115578e-05,
-      "loss": 5.8839,
-      "step": 5825
-    },
-    {
-      "epoch": 3.0383311603650585,
-      "grad_norm": 1.4671117067337036,
-      "learning_rate": 9.464824120603016e-05,
-      "loss": 5.7509,
-      "step": 5826
-    },
-    {
-      "epoch": 3.038852672750978,
-      "grad_norm": 1.3588237762451172,
-      "learning_rate": 9.464723618090454e-05,
-      "loss": 6.0299,
-      "step": 5827
-    },
-    {
-      "epoch": 3.039374185136897,
-      "grad_norm": 1.5312944650650024,
-      "learning_rate": 9.46462311557789e-05,
-      "loss": 5.9171,
-      "step": 5828
-    },
-    {
-      "epoch": 3.039895697522816,
-      "grad_norm": 1.436220645904541,
-      "learning_rate": 9.464522613065328e-05,
-      "loss": 6.023,
-      "step": 5829
-    },
-    {
-      "epoch": 3.0404172099087354,
-      "grad_norm": 1.474253535270691,
-      "learning_rate": 9.464422110552764e-05,
-      "loss": 5.8057,
-      "step": 5830
-    },
-    {
-      "epoch": 3.0409387222946545,
-      "grad_norm": 1.4670852422714233,
-      "learning_rate": 9.464321608040202e-05,
-      "loss": 5.8795,
-      "step": 5831
-    },
-    {
-      "epoch": 3.0414602346805735,
-      "grad_norm": 1.5747942924499512,
-      "learning_rate": 9.464221105527638e-05,
-      "loss": 6.0905,
-      "step": 5832
-    },
-    {
-      "epoch": 3.041981747066493,
-      "grad_norm": 1.484284520149231,
-      "learning_rate": 9.464120603015075e-05,
-      "loss": 5.7966,
-      "step": 5833
-    },
-    {
-      "epoch": 3.042503259452412,
-      "grad_norm": 1.5133543014526367,
-      "learning_rate": 9.464020100502513e-05,
-      "loss": 6.0068,
-      "step": 5834
-    },
-    {
-      "epoch": 3.043024771838331,
-      "grad_norm": 1.4207285642623901,
-      "learning_rate": 9.46391959798995e-05,
-      "loss": 5.8866,
-      "step": 5835
-    },
-    {
-      "epoch": 3.0435462842242504,
-      "grad_norm": 1.7068456411361694,
-      "learning_rate": 9.463819095477388e-05,
-      "loss": 5.5827,
-      "step": 5836
-    },
-    {
-      "epoch": 3.0440677966101695,
-      "grad_norm": 1.7409515380859375,
-      "learning_rate": 9.463718592964825e-05,
-      "loss": 5.5897,
-      "step": 5837
-    },
-    {
-      "epoch": 3.0445893089960885,
-      "grad_norm": 1.513808012008667,
-      "learning_rate": 9.463618090452262e-05,
-      "loss": 4.9028,
-      "step": 5838
-    },
-    {
-      "epoch": 3.045110821382008,
-      "grad_norm": 1.4568588733673096,
-      "learning_rate": 9.463517587939699e-05,
-      "loss": 5.8086,
-      "step": 5839
-    },
-    {
-      "epoch": 3.045632333767927,
-      "grad_norm": 1.5937316417694092,
-      "learning_rate": 9.463417085427137e-05,
-      "loss": 5.6933,
-      "step": 5840
-    },
-    {
-      "epoch": 3.046153846153846,
-      "grad_norm": 1.5717334747314453,
-      "learning_rate": 9.463316582914573e-05,
-      "loss": 5.3998,
-      "step": 5841
-    },
-    {
-      "epoch": 3.0466753585397655,
-      "grad_norm": 1.5751968622207642,
-      "learning_rate": 9.463216080402011e-05,
-      "loss": 5.6339,
-      "step": 5842
-    },
-    {
-      "epoch": 3.0471968709256845,
-      "grad_norm": 1.5425190925598145,
-      "learning_rate": 9.463115577889447e-05,
-      "loss": 6.1554,
-      "step": 5843
-    },
-    {
-      "epoch": 3.0477183833116035,
-      "grad_norm": 1.519472360610962,
-      "learning_rate": 9.463015075376885e-05,
-      "loss": 5.9995,
-      "step": 5844
-    },
-    {
-      "epoch": 3.048239895697523,
-      "grad_norm": 1.3662327527999878,
-      "learning_rate": 9.462914572864321e-05,
-      "loss": 5.3337,
-      "step": 5845
-    },
-    {
-      "epoch": 3.048761408083442,
-      "grad_norm": 1.5622271299362183,
-      "learning_rate": 9.462814070351759e-05,
-      "loss": 5.4109,
-      "step": 5846
-    },
-    {
-      "epoch": 3.049282920469361,
-      "grad_norm": 1.5849497318267822,
-      "learning_rate": 9.462713567839197e-05,
-      "loss": 6.1005,
-      "step": 5847
-    },
-    {
-      "epoch": 3.0498044328552805,
-      "grad_norm": 1.4504023790359497,
-      "learning_rate": 9.462613065326633e-05,
-      "loss": 5.6537,
-      "step": 5848
-    },
-    {
-      "epoch": 3.0503259452411995,
-      "grad_norm": 1.4177052974700928,
-      "learning_rate": 9.462512562814071e-05,
-      "loss": 5.8151,
-      "step": 5849
-    },
-    {
-      "epoch": 3.0508474576271185,
-      "grad_norm": 1.560486078262329,
-      "learning_rate": 9.462412060301508e-05,
-      "loss": 5.9267,
-      "step": 5850
-    },
-    {
-      "epoch": 3.051368970013038,
-      "grad_norm": 1.4382741451263428,
-      "learning_rate": 9.462311557788945e-05,
-      "loss": 5.7341,
-      "step": 5851
-    },
-    {
-      "epoch": 3.051890482398957,
-      "grad_norm": 1.557877779006958,
-      "learning_rate": 9.462211055276382e-05,
-      "loss": 5.5543,
-      "step": 5852
-    },
-    {
-      "epoch": 3.052411994784876,
-      "grad_norm": 1.4534897804260254,
-      "learning_rate": 9.46211055276382e-05,
-      "loss": 5.9318,
-      "step": 5853
-    },
-    {
-      "epoch": 3.0529335071707955,
-      "grad_norm": 1.612915277481079,
-      "learning_rate": 9.462010050251256e-05,
-      "loss": 5.547,
-      "step": 5854
-    },
-    {
-      "epoch": 3.0534550195567145,
-      "grad_norm": 1.7588146924972534,
-      "learning_rate": 9.461909547738694e-05,
-      "loss": 5.4807,
-      "step": 5855
-    },
-    {
-      "epoch": 3.0539765319426335,
-      "grad_norm": 1.558422565460205,
-      "learning_rate": 9.461809045226132e-05,
-      "loss": 5.4864,
-      "step": 5856
-    },
-    {
-      "epoch": 3.054498044328553,
-      "grad_norm": 1.5462599992752075,
-      "learning_rate": 9.46170854271357e-05,
-      "loss": 4.9951,
-      "step": 5857
-    },
-    {
-      "epoch": 3.055019556714472,
-      "grad_norm": 1.4371531009674072,
-      "learning_rate": 9.461608040201006e-05,
-      "loss": 5.763,
-      "step": 5858
-    },
-    {
-      "epoch": 3.055541069100391,
-      "grad_norm": 1.5271117687225342,
-      "learning_rate": 9.461507537688442e-05,
-      "loss": 5.6543,
-      "step": 5859
-    },
-    {
-      "epoch": 3.0560625814863105,
-      "grad_norm": 1.3888907432556152,
-      "learning_rate": 9.46140703517588e-05,
-      "loss": 5.9318,
-      "step": 5860
-    },
-    {
-      "epoch": 3.0565840938722295,
-      "grad_norm": 1.4006364345550537,
-      "learning_rate": 9.461306532663317e-05,
-      "loss": 5.802,
-      "step": 5861
-    },
-    {
-      "epoch": 3.0571056062581485,
-      "grad_norm": 1.4768800735473633,
-      "learning_rate": 9.461206030150754e-05,
-      "loss": 5.6076,
-      "step": 5862
-    },
-    {
-      "epoch": 3.057627118644068,
-      "grad_norm": 1.4391425848007202,
-      "learning_rate": 9.461105527638191e-05,
-      "loss": 5.9272,
-      "step": 5863
-    },
-    {
-      "epoch": 3.058148631029987,
-      "grad_norm": 1.5824787616729736,
-      "learning_rate": 9.461005025125628e-05,
-      "loss": 5.7183,
-      "step": 5864
-    },
-    {
-      "epoch": 3.058670143415906,
-      "grad_norm": 1.392802357673645,
-      "learning_rate": 9.460904522613065e-05,
-      "loss": 5.9929,
-      "step": 5865
-    },
-    {
-      "epoch": 3.0591916558018255,
-      "grad_norm": 1.3415416479110718,
-      "learning_rate": 9.460804020100503e-05,
-      "loss": 6.0666,
-      "step": 5866
-    },
-    {
-      "epoch": 3.0597131681877445,
-      "grad_norm": 1.5675480365753174,
-      "learning_rate": 9.46070351758794e-05,
-      "loss": 5.7511,
-      "step": 5867
-    },
-    {
-      "epoch": 3.0602346805736635,
-      "grad_norm": 1.5871272087097168,
-      "learning_rate": 9.460603015075378e-05,
-      "loss": 5.2832,
-      "step": 5868
-    },
-    {
-      "epoch": 3.060756192959583,
-      "grad_norm": 1.638749361038208,
-      "learning_rate": 9.460502512562815e-05,
-      "loss": 5.6168,
-      "step": 5869
-    },
-    {
-      "epoch": 3.061277705345502,
-      "grad_norm": 1.512995958328247,
-      "learning_rate": 9.460402010050252e-05,
-      "loss": 5.6994,
-      "step": 5870
-    },
-    {
-      "epoch": 3.061799217731421,
-      "grad_norm": 1.4366130828857422,
-      "learning_rate": 9.460301507537689e-05,
-      "loss": 5.8468,
-      "step": 5871
-    },
-    {
-      "epoch": 3.06232073011734,
-      "grad_norm": 1.6672816276550293,
-      "learning_rate": 9.460201005025127e-05,
-      "loss": 5.3897,
-      "step": 5872
-    },
-    {
-      "epoch": 3.0628422425032595,
-      "grad_norm": 1.3812741041183472,
-      "learning_rate": 9.460100502512563e-05,
-      "loss": 6.0461,
-      "step": 5873
-    },
-    {
-      "epoch": 3.0633637548891786,
-      "grad_norm": 1.6512980461120605,
-      "learning_rate": 9.46e-05,
-      "loss": 5.7461,
-      "step": 5874
-    },
-    {
-      "epoch": 3.0638852672750976,
-      "grad_norm": 1.5828953981399536,
-      "learning_rate": 9.459899497487437e-05,
-      "loss": 5.2683,
-      "step": 5875
-    },
-    {
-      "epoch": 3.064406779661017,
-      "grad_norm": 1.509559154510498,
-      "learning_rate": 9.459798994974874e-05,
-      "loss": 5.9163,
-      "step": 5876
-    },
-    {
-      "epoch": 3.064928292046936,
-      "grad_norm": 2.1858792304992676,
-      "learning_rate": 9.459698492462312e-05,
-      "loss": 5.6758,
-      "step": 5877
-    },
-    {
-      "epoch": 3.065449804432855,
-      "grad_norm": 1.5580583810806274,
-      "learning_rate": 9.45959798994975e-05,
-      "loss": 5.9811,
-      "step": 5878
-    },
-    {
-      "epoch": 3.0659713168187746,
-      "grad_norm": 1.4582642316818237,
-      "learning_rate": 9.459497487437187e-05,
-      "loss": 5.7829,
-      "step": 5879
-    },
-    {
-      "epoch": 3.0664928292046936,
-      "grad_norm": 1.4365782737731934,
-      "learning_rate": 9.459396984924624e-05,
-      "loss": 5.6398,
-      "step": 5880
-    },
-    {
-      "epoch": 3.0670143415906126,
-      "grad_norm": 2.1087427139282227,
-      "learning_rate": 9.459296482412061e-05,
-      "loss": 5.741,
-      "step": 5881
-    },
-    {
-      "epoch": 3.067535853976532,
-      "grad_norm": 1.5821714401245117,
-      "learning_rate": 9.459195979899498e-05,
-      "loss": 5.6478,
-      "step": 5882
-    },
-    {
-      "epoch": 3.068057366362451,
-      "grad_norm": 1.4344381093978882,
-      "learning_rate": 9.459095477386936e-05,
-      "loss": 6.1032,
-      "step": 5883
-    },
-    {
-      "epoch": 3.06857887874837,
-      "grad_norm": 1.5576742887496948,
-      "learning_rate": 9.458994974874372e-05,
-      "loss": 5.7646,
-      "step": 5884
-    },
-    {
-      "epoch": 3.0691003911342896,
-      "grad_norm": 1.5506314039230347,
-      "learning_rate": 9.45889447236181e-05,
-      "loss": 5.534,
-      "step": 5885
-    },
-    {
-      "epoch": 3.0696219035202086,
-      "grad_norm": 1.4927634000778198,
-      "learning_rate": 9.458793969849246e-05,
-      "loss": 6.0125,
-      "step": 5886
-    },
-    {
-      "epoch": 3.0701434159061276,
-      "grad_norm": 1.6473774909973145,
-      "learning_rate": 9.458693467336684e-05,
-      "loss": 5.9649,
-      "step": 5887
-    },
-    {
-      "epoch": 3.070664928292047,
-      "grad_norm": 1.560056209564209,
-      "learning_rate": 9.458592964824122e-05,
-      "loss": 5.3517,
-      "step": 5888
-    },
-    {
-      "epoch": 3.071186440677966,
-      "grad_norm": 1.6311829090118408,
-      "learning_rate": 9.458492462311558e-05,
-      "loss": 5.6599,
-      "step": 5889
-    },
-    {
-      "epoch": 3.071707953063885,
-      "grad_norm": 1.6032822132110596,
-      "learning_rate": 9.458391959798996e-05,
-      "loss": 5.4932,
-      "step": 5890
-    },
-    {
-      "epoch": 3.0722294654498046,
-      "grad_norm": 1.6072977781295776,
-      "learning_rate": 9.458291457286432e-05,
-      "loss": 5.2739,
-      "step": 5891
-    },
-    {
-      "epoch": 3.0727509778357236,
-      "grad_norm": 1.4746280908584595,
-      "learning_rate": 9.45819095477387e-05,
-      "loss": 5.7097,
-      "step": 5892
-    },
-    {
-      "epoch": 3.0732724902216426,
-      "grad_norm": 1.3764489889144897,
-      "learning_rate": 9.458090452261307e-05,
-      "loss": 5.8225,
-      "step": 5893
-    },
-    {
-      "epoch": 3.073794002607562,
-      "grad_norm": 1.7624118328094482,
-      "learning_rate": 9.457989949748744e-05,
-      "loss": 5.2923,
-      "step": 5894
-    },
-    {
-      "epoch": 3.074315514993481,
-      "grad_norm": 1.4760856628417969,
-      "learning_rate": 9.457889447236181e-05,
-      "loss": 5.7976,
-      "step": 5895
-    },
-    {
-      "epoch": 3.0748370273794,
-      "grad_norm": 1.5168676376342773,
-      "learning_rate": 9.457788944723619e-05,
-      "loss": 5.4996,
-      "step": 5896
-    },
-    {
-      "epoch": 3.0753585397653196,
-      "grad_norm": 1.5580554008483887,
-      "learning_rate": 9.457688442211055e-05,
-      "loss": 5.7098,
-      "step": 5897
-    },
-    {
-      "epoch": 3.0758800521512386,
-      "grad_norm": 1.4205609560012817,
-      "learning_rate": 9.457587939698493e-05,
-      "loss": 6.0651,
-      "step": 5898
-    },
-    {
-      "epoch": 3.0764015645371576,
-      "grad_norm": 1.4074150323867798,
-      "learning_rate": 9.45748743718593e-05,
-      "loss": 6.098,
-      "step": 5899
-    },
-    {
-      "epoch": 3.076923076923077,
-      "grad_norm": 1.58516263961792,
-      "learning_rate": 9.457386934673367e-05,
-      "loss": 5.526,
-      "step": 5900
-    },
-    {
-      "epoch": 3.077444589308996,
-      "grad_norm": 1.6538965702056885,
-      "learning_rate": 9.457286432160805e-05,
-      "loss": 5.2366,
-      "step": 5901
-    },
-    {
-      "epoch": 3.077966101694915,
-      "grad_norm": 1.427926778793335,
-      "learning_rate": 9.457185929648241e-05,
-      "loss": 5.9894,
-      "step": 5902
-    },
-    {
-      "epoch": 3.0784876140808346,
-      "grad_norm": 1.4214593172073364,
-      "learning_rate": 9.457085427135679e-05,
-      "loss": 5.6836,
-      "step": 5903
-    },
-    {
-      "epoch": 3.0790091264667536,
-      "grad_norm": 1.6015725135803223,
-      "learning_rate": 9.456984924623115e-05,
-      "loss": 5.4728,
-      "step": 5904
-    },
-    {
-      "epoch": 3.0795306388526726,
-      "grad_norm": 1.6532886028289795,
-      "learning_rate": 9.456884422110553e-05,
-      "loss": 5.4083,
-      "step": 5905
-    },
-    {
-      "epoch": 3.080052151238592,
-      "grad_norm": 1.7719734907150269,
-      "learning_rate": 9.45678391959799e-05,
-      "loss": 5.181,
-      "step": 5906
-    },
-    {
-      "epoch": 3.080573663624511,
-      "grad_norm": 1.5277974605560303,
-      "learning_rate": 9.456683417085427e-05,
-      "loss": 5.5575,
-      "step": 5907
-    },
-    {
-      "epoch": 3.08109517601043,
-      "grad_norm": 1.488882303237915,
-      "learning_rate": 9.456582914572865e-05,
-      "loss": 6.0537,
-      "step": 5908
-    },
-    {
-      "epoch": 3.0816166883963496,
-      "grad_norm": 1.6720529794692993,
-      "learning_rate": 9.456482412060303e-05,
-      "loss": 5.5987,
-      "step": 5909
-    },
-    {
-      "epoch": 3.0821382007822686,
-      "grad_norm": 1.5618151426315308,
-      "learning_rate": 9.45638190954774e-05,
-      "loss": 5.3473,
-      "step": 5910
-    },
-    {
-      "epoch": 3.0826597131681877,
-      "grad_norm": 1.6426540613174438,
-      "learning_rate": 9.456281407035177e-05,
-      "loss": 5.8116,
-      "step": 5911
-    },
-    {
-      "epoch": 3.083181225554107,
-      "grad_norm": 1.4284110069274902,
-      "learning_rate": 9.456180904522614e-05,
-      "loss": 5.8414,
-      "step": 5912
-    },
-    {
-      "epoch": 3.083702737940026,
-      "grad_norm": 1.4357808828353882,
-      "learning_rate": 9.45608040201005e-05,
-      "loss": 6.0882,
-      "step": 5913
-    },
-    {
-      "epoch": 3.084224250325945,
-      "grad_norm": 1.4815322160720825,
-      "learning_rate": 9.455979899497488e-05,
-      "loss": 5.6096,
-      "step": 5914
-    },
-    {
-      "epoch": 3.084745762711864,
-      "grad_norm": 1.3797491788864136,
-      "learning_rate": 9.455879396984924e-05,
-      "loss": 6.2326,
-      "step": 5915
-    },
-    {
-      "epoch": 3.0852672750977836,
-      "grad_norm": 1.5625903606414795,
-      "learning_rate": 9.455778894472362e-05,
-      "loss": 5.3565,
-      "step": 5916
-    },
-    {
-      "epoch": 3.0857887874837027,
-      "grad_norm": 1.6335209608078003,
-      "learning_rate": 9.455678391959798e-05,
-      "loss": 5.8099,
-      "step": 5917
-    },
-    {
-      "epoch": 3.0863102998696217,
-      "grad_norm": 1.396146535873413,
-      "learning_rate": 9.455577889447236e-05,
-      "loss": 6.2562,
-      "step": 5918
-    },
-    {
-      "epoch": 3.086831812255541,
-      "grad_norm": 1.334381103515625,
-      "learning_rate": 9.455477386934674e-05,
-      "loss": 6.1637,
-      "step": 5919
-    },
-    {
-      "epoch": 3.08735332464146,
-      "grad_norm": 1.219536304473877,
-      "learning_rate": 9.455376884422112e-05,
-      "loss": 6.2134,
-      "step": 5920
-    },
-    {
-      "epoch": 3.087874837027379,
-      "grad_norm": 1.2963420152664185,
-      "learning_rate": 9.455276381909548e-05,
-      "loss": 6.208,
-      "step": 5921
-    },
-    {
-      "epoch": 3.0883963494132987,
-      "grad_norm": 1.5001893043518066,
-      "learning_rate": 9.455175879396986e-05,
-      "loss": 5.8838,
-      "step": 5922
-    },
-    {
-      "epoch": 3.0889178617992177,
-      "grad_norm": 1.4513239860534668,
-      "learning_rate": 9.455075376884422e-05,
-      "loss": 5.2857,
-      "step": 5923
-    },
-    {
-      "epoch": 3.0894393741851367,
-      "grad_norm": 1.58176589012146,
-      "learning_rate": 9.45497487437186e-05,
-      "loss": 5.5655,
-      "step": 5924
-    },
-    {
-      "epoch": 3.089960886571056,
-      "grad_norm": 1.5114370584487915,
-      "learning_rate": 9.454874371859297e-05,
-      "loss": 5.9633,
-      "step": 5925
-    },
-    {
-      "epoch": 3.090482398956975,
-      "grad_norm": 1.4420485496520996,
-      "learning_rate": 9.454773869346733e-05,
-      "loss": 6.0799,
-      "step": 5926
-    },
-    {
-      "epoch": 3.091003911342894,
-      "grad_norm": 1.4343568086624146,
-      "learning_rate": 9.454673366834171e-05,
-      "loss": 5.9314,
-      "step": 5927
-    },
-    {
-      "epoch": 3.0915254237288137,
-      "grad_norm": 1.4300570487976074,
-      "learning_rate": 9.454572864321609e-05,
-      "loss": 5.5417,
-      "step": 5928
-    },
-    {
-      "epoch": 3.0920469361147327,
-      "grad_norm": 1.4604412317276,
-      "learning_rate": 9.454472361809046e-05,
-      "loss": 5.9906,
-      "step": 5929
-    },
-    {
-      "epoch": 3.0925684485006517,
-      "grad_norm": 1.4193544387817383,
-      "learning_rate": 9.454371859296483e-05,
-      "loss": 5.7321,
-      "step": 5930
-    },
-    {
-      "epoch": 3.093089960886571,
-      "grad_norm": 1.4256242513656616,
-      "learning_rate": 9.45427135678392e-05,
-      "loss": 6.0287,
-      "step": 5931
-    },
-    {
-      "epoch": 3.09361147327249,
-      "grad_norm": 1.4964123964309692,
-      "learning_rate": 9.454170854271357e-05,
-      "loss": 5.8263,
-      "step": 5932
-    },
-    {
-      "epoch": 3.094132985658409,
-      "grad_norm": 1.5592036247253418,
-      "learning_rate": 9.454070351758795e-05,
-      "loss": 5.6518,
-      "step": 5933
-    },
-    {
-      "epoch": 3.0946544980443287,
-      "grad_norm": 1.5268688201904297,
-      "learning_rate": 9.453969849246231e-05,
-      "loss": 5.8864,
-      "step": 5934
-    },
-    {
-      "epoch": 3.0951760104302477,
-      "grad_norm": 1.6359882354736328,
-      "learning_rate": 9.453869346733669e-05,
-      "loss": 5.8274,
-      "step": 5935
-    },
-    {
-      "epoch": 3.0956975228161667,
-      "grad_norm": 1.4909464120864868,
-      "learning_rate": 9.453768844221105e-05,
-      "loss": 5.9757,
-      "step": 5936
-    },
-    {
-      "epoch": 3.096219035202086,
-      "grad_norm": 1.5061111450195312,
-      "learning_rate": 9.453668341708543e-05,
-      "loss": 5.7473,
-      "step": 5937
-    },
-    {
-      "epoch": 3.096740547588005,
-      "grad_norm": 1.6529176235198975,
-      "learning_rate": 9.45356783919598e-05,
-      "loss": 5.7972,
-      "step": 5938
-    },
-    {
-      "epoch": 3.0972620599739242,
-      "grad_norm": 1.5472520589828491,
-      "learning_rate": 9.453467336683417e-05,
-      "loss": 5.8479,
-      "step": 5939
-    },
-    {
-      "epoch": 3.0977835723598437,
-      "grad_norm": 1.8916460275650024,
-      "learning_rate": 9.453366834170855e-05,
-      "loss": 5.0422,
-      "step": 5940
-    },
-    {
-      "epoch": 3.0983050847457627,
-      "grad_norm": 1.5509400367736816,
-      "learning_rate": 9.453266331658292e-05,
-      "loss": 6.0129,
-      "step": 5941
-    },
-    {
-      "epoch": 3.0988265971316817,
-      "grad_norm": 1.6247501373291016,
-      "learning_rate": 9.45316582914573e-05,
-      "loss": 5.5638,
-      "step": 5942
-    },
-    {
-      "epoch": 3.099348109517601,
-      "grad_norm": 1.4283233880996704,
-      "learning_rate": 9.453065326633166e-05,
-      "loss": 5.3822,
-      "step": 5943
-    },
-    {
-      "epoch": 3.0998696219035202,
-      "grad_norm": 1.5038272142410278,
-      "learning_rate": 9.452964824120604e-05,
-      "loss": 5.6856,
-      "step": 5944
-    },
-    {
-      "epoch": 3.1003911342894392,
-      "grad_norm": 1.495367169380188,
-      "learning_rate": 9.45286432160804e-05,
-      "loss": 6.0635,
-      "step": 5945
-    },
-    {
-      "epoch": 3.1009126466753587,
-      "grad_norm": 1.4465676546096802,
-      "learning_rate": 9.452763819095478e-05,
-      "loss": 5.7256,
-      "step": 5946
-    },
-    {
-      "epoch": 3.1014341590612777,
-      "grad_norm": 1.538765788078308,
-      "learning_rate": 9.452663316582914e-05,
-      "loss": 6.0468,
-      "step": 5947
-    },
-    {
-      "epoch": 3.1019556714471967,
-      "grad_norm": 1.4355207681655884,
-      "learning_rate": 9.452562814070352e-05,
-      "loss": 6.3044,
-      "step": 5948
-    },
-    {
-      "epoch": 3.102477183833116,
-      "grad_norm": 1.756052017211914,
-      "learning_rate": 9.45246231155779e-05,
-      "loss": 5.1981,
-      "step": 5949
-    },
-    {
-      "epoch": 3.1029986962190352,
-      "grad_norm": 1.6237757205963135,
-      "learning_rate": 9.452361809045228e-05,
-      "loss": 5.9434,
-      "step": 5950
-    },
-    {
-      "epoch": 3.1035202086049543,
-      "grad_norm": 1.4441238641738892,
-      "learning_rate": 9.452261306532664e-05,
-      "loss": 5.9907,
-      "step": 5951
-    },
-    {
-      "epoch": 3.1040417209908737,
-      "grad_norm": 1.4781140089035034,
-      "learning_rate": 9.4521608040201e-05,
-      "loss": 5.1045,
-      "step": 5952
-    },
-    {
-      "epoch": 3.1045632333767927,
-      "grad_norm": 1.5835574865341187,
-      "learning_rate": 9.452060301507538e-05,
-      "loss": 5.6959,
-      "step": 5953
-    },
-    {
-      "epoch": 3.1050847457627118,
-      "grad_norm": 1.4697521924972534,
-      "learning_rate": 9.451959798994975e-05,
-      "loss": 5.6298,
-      "step": 5954
-    },
-    {
-      "epoch": 3.1056062581486312,
-      "grad_norm": 1.340589165687561,
-      "learning_rate": 9.451859296482413e-05,
-      "loss": 5.8457,
-      "step": 5955
-    },
-    {
-      "epoch": 3.1061277705345502,
-      "grad_norm": 1.4671962261199951,
-      "learning_rate": 9.451758793969849e-05,
-      "loss": 5.5994,
-      "step": 5956
-    },
-    {
-      "epoch": 3.1066492829204693,
-      "grad_norm": 1.4438464641571045,
-      "learning_rate": 9.451658291457287e-05,
-      "loss": 6.0035,
-      "step": 5957
-    },
-    {
-      "epoch": 3.1071707953063887,
-      "grad_norm": 1.3971376419067383,
-      "learning_rate": 9.451557788944723e-05,
-      "loss": 5.5575,
-      "step": 5958
-    },
-    {
-      "epoch": 3.1076923076923078,
-      "grad_norm": 1.5090712308883667,
-      "learning_rate": 9.451457286432161e-05,
-      "loss": 5.684,
-      "step": 5959
-    },
-    {
-      "epoch": 3.1082138200782268,
-      "grad_norm": 1.3829509019851685,
-      "learning_rate": 9.451356783919599e-05,
-      "loss": 5.5098,
-      "step": 5960
-    },
-    {
-      "epoch": 3.1087353324641462,
-      "grad_norm": 1.4362118244171143,
-      "learning_rate": 9.451256281407037e-05,
-      "loss": 5.9396,
-      "step": 5961
-    },
-    {
-      "epoch": 3.1092568448500653,
-      "grad_norm": 1.3199585676193237,
-      "learning_rate": 9.451155778894473e-05,
-      "loss": 5.3113,
-      "step": 5962
-    },
-    {
-      "epoch": 3.1097783572359843,
-      "grad_norm": 1.462471604347229,
-      "learning_rate": 9.451055276381911e-05,
-      "loss": 5.4175,
-      "step": 5963
-    },
-    {
-      "epoch": 3.1102998696219037,
-      "grad_norm": 1.5190324783325195,
-      "learning_rate": 9.450954773869347e-05,
-      "loss": 5.6788,
-      "step": 5964
-    },
-    {
-      "epoch": 3.1108213820078228,
-      "grad_norm": 1.3350346088409424,
-      "learning_rate": 9.450854271356785e-05,
-      "loss": 5.9516,
-      "step": 5965
-    },
-    {
-      "epoch": 3.111342894393742,
-      "grad_norm": 1.3691679239273071,
-      "learning_rate": 9.450753768844221e-05,
-      "loss": 5.9994,
-      "step": 5966
-    },
-    {
-      "epoch": 3.111864406779661,
-      "grad_norm": 1.391628384590149,
-      "learning_rate": 9.450653266331658e-05,
-      "loss": 5.8559,
-      "step": 5967
-    },
-    {
-      "epoch": 3.1123859191655803,
-      "grad_norm": 1.37540864944458,
-      "learning_rate": 9.450552763819096e-05,
-      "loss": 6.054,
-      "step": 5968
-    },
-    {
-      "epoch": 3.1129074315514993,
-      "grad_norm": 1.538871169090271,
-      "learning_rate": 9.450452261306533e-05,
-      "loss": 5.0338,
-      "step": 5969
-    },
-    {
-      "epoch": 3.1134289439374183,
-      "grad_norm": 1.582134485244751,
-      "learning_rate": 9.450351758793971e-05,
-      "loss": 5.8991,
-      "step": 5970
-    },
-    {
-      "epoch": 3.113950456323338,
-      "grad_norm": 1.5040698051452637,
-      "learning_rate": 9.450251256281408e-05,
-      "loss": 5.785,
-      "step": 5971
-    },
-    {
-      "epoch": 3.114471968709257,
-      "grad_norm": 1.6494308710098267,
-      "learning_rate": 9.450150753768845e-05,
-      "loss": 5.7682,
-      "step": 5972
-    },
-    {
-      "epoch": 3.114993481095176,
-      "grad_norm": 1.397558569908142,
-      "learning_rate": 9.450050251256282e-05,
-      "loss": 6.094,
-      "step": 5973
-    },
-    {
-      "epoch": 3.1155149934810953,
-      "grad_norm": 1.4531022310256958,
-      "learning_rate": 9.44994974874372e-05,
-      "loss": 5.9844,
-      "step": 5974
-    },
-    {
-      "epoch": 3.1160365058670143,
-      "grad_norm": 1.4732410907745361,
-      "learning_rate": 9.449849246231156e-05,
-      "loss": 5.7983,
-      "step": 5975
-    },
-    {
-      "epoch": 3.1165580182529333,
-      "grad_norm": 1.818155288696289,
-      "learning_rate": 9.449748743718594e-05,
-      "loss": 5.2632,
-      "step": 5976
-    },
-    {
-      "epoch": 3.117079530638853,
-      "grad_norm": 1.5746917724609375,
-      "learning_rate": 9.44964824120603e-05,
-      "loss": 5.5873,
-      "step": 5977
-    },
-    {
-      "epoch": 3.117601043024772,
-      "grad_norm": 1.5462315082550049,
-      "learning_rate": 9.449547738693468e-05,
-      "loss": 5.7091,
-      "step": 5978
-    },
-    {
-      "epoch": 3.118122555410691,
-      "grad_norm": 1.8027981519699097,
-      "learning_rate": 9.449447236180904e-05,
-      "loss": 5.3698,
-      "step": 5979
-    },
-    {
-      "epoch": 3.1186440677966103,
-      "grad_norm": 1.6408286094665527,
-      "learning_rate": 9.449346733668342e-05,
-      "loss": 5.8103,
-      "step": 5980
-    },
-    {
-      "epoch": 3.1191655801825293,
-      "grad_norm": 1.5473012924194336,
-      "learning_rate": 9.44924623115578e-05,
-      "loss": 5.6519,
-      "step": 5981
-    },
-    {
-      "epoch": 3.1196870925684483,
-      "grad_norm": 1.5296783447265625,
-      "learning_rate": 9.449145728643216e-05,
-      "loss": 5.4603,
-      "step": 5982
-    },
-    {
-      "epoch": 3.120208604954368,
-      "grad_norm": 1.4732483625411987,
-      "learning_rate": 9.449045226130654e-05,
-      "loss": 6.0197,
-      "step": 5983
-    },
-    {
-      "epoch": 3.120730117340287,
-      "grad_norm": 1.3387699127197266,
-      "learning_rate": 9.44894472361809e-05,
-      "loss": 6.1192,
-      "step": 5984
-    },
-    {
-      "epoch": 3.121251629726206,
-      "grad_norm": 1.6230127811431885,
-      "learning_rate": 9.448844221105528e-05,
-      "loss": 5.1001,
-      "step": 5985
-    },
-    {
-      "epoch": 3.1217731421121253,
-      "grad_norm": 1.7012475728988647,
-      "learning_rate": 9.448743718592965e-05,
-      "loss": 5.387,
-      "step": 5986
-    },
-    {
-      "epoch": 3.1222946544980443,
-      "grad_norm": 1.5545032024383545,
-      "learning_rate": 9.448643216080403e-05,
-      "loss": 5.7839,
-      "step": 5987
-    },
-    {
-      "epoch": 3.1228161668839634,
-      "grad_norm": 1.654433250427246,
-      "learning_rate": 9.448542713567839e-05,
-      "loss": 5.4428,
-      "step": 5988
-    },
-    {
-      "epoch": 3.123337679269883,
-      "grad_norm": 1.5025675296783447,
-      "learning_rate": 9.448442211055277e-05,
-      "loss": 5.4167,
-      "step": 5989
-    },
-    {
-      "epoch": 3.123859191655802,
-      "grad_norm": 1.5476739406585693,
-      "learning_rate": 9.448341708542715e-05,
-      "loss": 5.8965,
-      "step": 5990
-    },
-    {
-      "epoch": 3.124380704041721,
-      "grad_norm": 1.4096719026565552,
-      "learning_rate": 9.448241206030152e-05,
-      "loss": 6.1156,
-      "step": 5991
-    },
-    {
-      "epoch": 3.1249022164276403,
-      "grad_norm": 1.653701663017273,
-      "learning_rate": 9.448140703517589e-05,
-      "loss": 5.9378,
-      "step": 5992
-    },
-    {
-      "epoch": 3.1254237288135593,
-      "grad_norm": 1.6070947647094727,
-      "learning_rate": 9.448040201005025e-05,
-      "loss": 5.999,
-      "step": 5993
-    },
-    {
-      "epoch": 3.1259452411994784,
-      "grad_norm": 1.4291785955429077,
-      "learning_rate": 9.447939698492463e-05,
-      "loss": 5.1386,
-      "step": 5994
-    },
-    {
-      "epoch": 3.126466753585398,
-      "grad_norm": 1.5960363149642944,
-      "learning_rate": 9.4478391959799e-05,
-      "loss": 5.6513,
-      "step": 5995
-    },
-    {
-      "epoch": 3.126988265971317,
-      "grad_norm": 1.8519086837768555,
-      "learning_rate": 9.447738693467337e-05,
-      "loss": 5.1645,
-      "step": 5996
-    },
-    {
-      "epoch": 3.127509778357236,
-      "grad_norm": 1.8942108154296875,
-      "learning_rate": 9.447638190954774e-05,
-      "loss": 5.519,
-      "step": 5997
-    },
-    {
-      "epoch": 3.1280312907431553,
-      "grad_norm": 1.4592946767807007,
-      "learning_rate": 9.447537688442211e-05,
-      "loss": 5.8364,
-      "step": 5998
-    },
-    {
-      "epoch": 3.1285528031290744,
-      "grad_norm": 1.4667688608169556,
-      "learning_rate": 9.447437185929648e-05,
-      "loss": 5.924,
-      "step": 5999
-    },
-    {
-      "epoch": 3.1290743155149934,
-      "grad_norm": 1.4714646339416504,
-      "learning_rate": 9.447336683417086e-05,
-      "loss": 5.7784,
-      "step": 6000
-    },
-    {
-      "epoch": 3.129595827900913,
-      "grad_norm": 1.8773066997528076,
-      "learning_rate": 9.447236180904523e-05,
-      "loss": 5.0069,
-      "step": 6001
-    },
-    {
-      "epoch": 3.130117340286832,
-      "grad_norm": 1.6906769275665283,
-      "learning_rate": 9.447135678391961e-05,
-      "loss": 5.9346,
-      "step": 6002
-    },
-    {
-      "epoch": 3.130638852672751,
-      "grad_norm": 1.8142781257629395,
-      "learning_rate": 9.447035175879398e-05,
-      "loss": 5.5597,
-      "step": 6003
-    },
-    {
-      "epoch": 3.1311603650586703,
-      "grad_norm": 1.3168134689331055,
-      "learning_rate": 9.446934673366835e-05,
-      "loss": 5.346,
-      "step": 6004
-    },
-    {
-      "epoch": 3.1316818774445894,
-      "grad_norm": 1.4408619403839111,
-      "learning_rate": 9.446834170854272e-05,
-      "loss": 5.6314,
-      "step": 6005
-    },
-    {
-      "epoch": 3.1322033898305084,
-      "grad_norm": 1.3345054388046265,
-      "learning_rate": 9.446733668341708e-05,
-      "loss": 6.0806,
-      "step": 6006
-    },
-    {
-      "epoch": 3.1327249022164274,
-      "grad_norm": 1.5057436227798462,
-      "learning_rate": 9.446633165829146e-05,
-      "loss": 5.7438,
-      "step": 6007
-    },
-    {
-      "epoch": 3.133246414602347,
-      "grad_norm": 1.3786718845367432,
-      "learning_rate": 9.446532663316582e-05,
-      "loss": 5.8939,
-      "step": 6008
-    },
-    {
-      "epoch": 3.133767926988266,
-      "grad_norm": 1.5443403720855713,
-      "learning_rate": 9.44643216080402e-05,
-      "loss": 5.6266,
-      "step": 6009
-    },
-    {
-      "epoch": 3.134289439374185,
-      "grad_norm": 1.5332121849060059,
-      "learning_rate": 9.446331658291458e-05,
-      "loss": 5.8174,
-      "step": 6010
-    },
-    {
-      "epoch": 3.1348109517601044,
-      "grad_norm": 1.6315749883651733,
-      "learning_rate": 9.446231155778896e-05,
-      "loss": 5.7356,
-      "step": 6011
-    },
-    {
-      "epoch": 3.1353324641460234,
-      "grad_norm": 1.5787394046783447,
-      "learning_rate": 9.446130653266332e-05,
-      "loss": 5.9696,
-      "step": 6012
-    },
-    {
-      "epoch": 3.1358539765319424,
-      "grad_norm": 1.541546106338501,
-      "learning_rate": 9.44603015075377e-05,
-      "loss": 5.5079,
-      "step": 6013
-    },
-    {
-      "epoch": 3.136375488917862,
-      "grad_norm": 1.458977460861206,
-      "learning_rate": 9.445929648241206e-05,
-      "loss": 6.0796,
-      "step": 6014
-    },
-    {
-      "epoch": 3.136897001303781,
-      "grad_norm": 1.4625805616378784,
-      "learning_rate": 9.445829145728644e-05,
-      "loss": 4.9626,
-      "step": 6015
-    },
-    {
-      "epoch": 3.1374185136897,
-      "grad_norm": 1.3486944437026978,
-      "learning_rate": 9.445728643216081e-05,
-      "loss": 6.0129,
-      "step": 6016
-    },
-    {
-      "epoch": 3.1379400260756194,
-      "grad_norm": 1.6166188716888428,
-      "learning_rate": 9.445628140703518e-05,
-      "loss": 5.2544,
-      "step": 6017
-    },
-    {
-      "epoch": 3.1384615384615384,
-      "grad_norm": 1.4961820840835571,
-      "learning_rate": 9.445527638190955e-05,
-      "loss": 5.3073,
-      "step": 6018
-    },
-    {
-      "epoch": 3.1389830508474574,
-      "grad_norm": 1.5515804290771484,
-      "learning_rate": 9.445427135678391e-05,
-      "loss": 5.8037,
-      "step": 6019
-    },
-    {
-      "epoch": 3.139504563233377,
-      "grad_norm": 1.4997084140777588,
-      "learning_rate": 9.445326633165829e-05,
-      "loss": 5.4583,
-      "step": 6020
-    },
-    {
-      "epoch": 3.140026075619296,
-      "grad_norm": 1.5628770589828491,
-      "learning_rate": 9.445226130653267e-05,
-      "loss": 5.7477,
-      "step": 6021
-    },
-    {
-      "epoch": 3.140547588005215,
-      "grad_norm": 1.3544843196868896,
-      "learning_rate": 9.445125628140705e-05,
-      "loss": 5.2025,
-      "step": 6022
-    },
-    {
-      "epoch": 3.1410691003911344,
-      "grad_norm": 1.4447132349014282,
-      "learning_rate": 9.445025125628141e-05,
-      "loss": 6.1132,
-      "step": 6023
-    },
-    {
-      "epoch": 3.1415906127770534,
-      "grad_norm": 1.4827256202697754,
-      "learning_rate": 9.444924623115579e-05,
-      "loss": 5.7322,
-      "step": 6024
-    },
-    {
-      "epoch": 3.1421121251629724,
-      "grad_norm": 1.6576439142227173,
-      "learning_rate": 9.444824120603015e-05,
-      "loss": 5.4234,
-      "step": 6025
-    },
-    {
-      "epoch": 3.142633637548892,
-      "grad_norm": 1.700350046157837,
-      "learning_rate": 9.444723618090453e-05,
-      "loss": 5.7305,
-      "step": 6026
-    },
-    {
-      "epoch": 3.143155149934811,
-      "grad_norm": 1.5395225286483765,
-      "learning_rate": 9.44462311557789e-05,
-      "loss": 5.45,
-      "step": 6027
-    },
-    {
-      "epoch": 3.14367666232073,
-      "grad_norm": 1.5126898288726807,
-      "learning_rate": 9.444522613065327e-05,
-      "loss": 5.8802,
-      "step": 6028
-    },
-    {
-      "epoch": 3.1441981747066494,
-      "grad_norm": 1.5107595920562744,
-      "learning_rate": 9.444422110552764e-05,
-      "loss": 5.9033,
-      "step": 6029
-    },
-    {
-      "epoch": 3.1447196870925684,
-      "grad_norm": 1.5929425954818726,
-      "learning_rate": 9.444321608040202e-05,
-      "loss": 5.9111,
-      "step": 6030
-    },
-    {
-      "epoch": 3.1452411994784875,
-      "grad_norm": 1.576727032661438,
-      "learning_rate": 9.444221105527638e-05,
-      "loss": 5.655,
-      "step": 6031
-    },
-    {
-      "epoch": 3.145762711864407,
-      "grad_norm": 1.4252485036849976,
-      "learning_rate": 9.444120603015076e-05,
-      "loss": 5.9926,
-      "step": 6032
-    },
-    {
-      "epoch": 3.146284224250326,
-      "grad_norm": 1.6902813911437988,
-      "learning_rate": 9.444020100502514e-05,
-      "loss": 5.5387,
-      "step": 6033
-    },
-    {
-      "epoch": 3.146805736636245,
-      "grad_norm": 1.4553967714309692,
-      "learning_rate": 9.44391959798995e-05,
-      "loss": 5.8227,
-      "step": 6034
-    },
-    {
-      "epoch": 3.1473272490221644,
-      "grad_norm": 1.5301074981689453,
-      "learning_rate": 9.443819095477388e-05,
-      "loss": 5.8581,
-      "step": 6035
-    },
-    {
-      "epoch": 3.1478487614080835,
-      "grad_norm": 1.3797202110290527,
-      "learning_rate": 9.443718592964824e-05,
-      "loss": 5.7593,
-      "step": 6036
-    },
-    {
-      "epoch": 3.1483702737940025,
-      "grad_norm": 1.5734614133834839,
-      "learning_rate": 9.443618090452262e-05,
-      "loss": 5.7256,
-      "step": 6037
-    },
-    {
-      "epoch": 3.148891786179922,
-      "grad_norm": 1.495334506034851,
-      "learning_rate": 9.443517587939698e-05,
-      "loss": 5.8223,
-      "step": 6038
-    },
-    {
-      "epoch": 3.149413298565841,
-      "grad_norm": 1.4399739503860474,
-      "learning_rate": 9.443417085427136e-05,
-      "loss": 5.8111,
-      "step": 6039
-    },
-    {
-      "epoch": 3.14993481095176,
-      "grad_norm": 1.624007225036621,
-      "learning_rate": 9.443316582914573e-05,
-      "loss": 4.9693,
-      "step": 6040
-    },
-    {
-      "epoch": 3.1504563233376794,
-      "grad_norm": 1.446977972984314,
-      "learning_rate": 9.44321608040201e-05,
-      "loss": 6.0927,
-      "step": 6041
-    },
-    {
-      "epoch": 3.1509778357235985,
-      "grad_norm": 1.4708313941955566,
-      "learning_rate": 9.443115577889448e-05,
-      "loss": 5.7453,
-      "step": 6042
-    },
-    {
-      "epoch": 3.1514993481095175,
-      "grad_norm": 1.644184947013855,
-      "learning_rate": 9.443015075376886e-05,
-      "loss": 5.6096,
-      "step": 6043
-    },
-    {
-      "epoch": 3.152020860495437,
-      "grad_norm": 1.5620064735412598,
-      "learning_rate": 9.442914572864322e-05,
-      "loss": 6.174,
-      "step": 6044
-    },
-    {
-      "epoch": 3.152542372881356,
-      "grad_norm": 1.5655866861343384,
-      "learning_rate": 9.44281407035176e-05,
-      "loss": 5.6261,
-      "step": 6045
-    },
-    {
-      "epoch": 3.153063885267275,
-      "grad_norm": 1.477390170097351,
-      "learning_rate": 9.442713567839197e-05,
-      "loss": 5.3152,
-      "step": 6046
-    },
-    {
-      "epoch": 3.1535853976531945,
-      "grad_norm": 1.8749686479568481,
-      "learning_rate": 9.442613065326633e-05,
-      "loss": 5.4449,
-      "step": 6047
-    },
-    {
-      "epoch": 3.1541069100391135,
-      "grad_norm": 1.433998942375183,
-      "learning_rate": 9.442512562814071e-05,
-      "loss": 5.8313,
-      "step": 6048
-    },
-    {
-      "epoch": 3.1546284224250325,
-      "grad_norm": 1.5486464500427246,
-      "learning_rate": 9.442412060301507e-05,
-      "loss": 5.7931,
-      "step": 6049
-    },
-    {
-      "epoch": 3.155149934810952,
-      "grad_norm": 1.4365578889846802,
-      "learning_rate": 9.442311557788945e-05,
-      "loss": 5.7501,
-      "step": 6050
-    },
-    {
-      "epoch": 3.155671447196871,
-      "grad_norm": 1.441725254058838,
-      "learning_rate": 9.442211055276381e-05,
-      "loss": 5.5554,
-      "step": 6051
-    },
-    {
-      "epoch": 3.15619295958279,
-      "grad_norm": 1.3874943256378174,
-      "learning_rate": 9.442110552763819e-05,
-      "loss": 5.989,
-      "step": 6052
-    },
-    {
-      "epoch": 3.1567144719687095,
-      "grad_norm": 1.451522946357727,
-      "learning_rate": 9.442010050251257e-05,
-      "loss": 5.662,
-      "step": 6053
-    },
-    {
-      "epoch": 3.1572359843546285,
-      "grad_norm": 1.3636561632156372,
-      "learning_rate": 9.441909547738695e-05,
-      "loss": 5.8379,
-      "step": 6054
-    },
-    {
-      "epoch": 3.1577574967405475,
-      "grad_norm": 1.2279136180877686,
-      "learning_rate": 9.441809045226131e-05,
-      "loss": 5.2987,
-      "step": 6055
-    },
-    {
-      "epoch": 3.158279009126467,
-      "grad_norm": 1.4659367799758911,
-      "learning_rate": 9.441708542713569e-05,
-      "loss": 5.5469,
-      "step": 6056
-    },
-    {
-      "epoch": 3.158800521512386,
-      "grad_norm": 1.4097001552581787,
-      "learning_rate": 9.441608040201005e-05,
-      "loss": 5.7817,
-      "step": 6057
-    },
-    {
-      "epoch": 3.159322033898305,
-      "grad_norm": 1.4194533824920654,
-      "learning_rate": 9.441507537688443e-05,
-      "loss": 5.8874,
-      "step": 6058
-    },
-    {
-      "epoch": 3.1598435462842245,
-      "grad_norm": 1.4247093200683594,
-      "learning_rate": 9.44140703517588e-05,
-      "loss": 5.6,
-      "step": 6059
-    },
-    {
-      "epoch": 3.1603650586701435,
-      "grad_norm": 1.4381343126296997,
-      "learning_rate": 9.441306532663316e-05,
-      "loss": 6.1708,
-      "step": 6060
-    },
-    {
-      "epoch": 3.1608865710560625,
-      "grad_norm": 1.387538194656372,
-      "learning_rate": 9.441206030150754e-05,
-      "loss": 5.8294,
-      "step": 6061
-    },
-    {
-      "epoch": 3.1614080834419815,
-      "grad_norm": 1.4847896099090576,
-      "learning_rate": 9.441105527638192e-05,
-      "loss": 5.7407,
-      "step": 6062
-    },
-    {
-      "epoch": 3.161929595827901,
-      "grad_norm": 1.4319195747375488,
-      "learning_rate": 9.44100502512563e-05,
-      "loss": 5.8631,
-      "step": 6063
-    },
-    {
-      "epoch": 3.16245110821382,
-      "grad_norm": 1.426947832107544,
-      "learning_rate": 9.440904522613066e-05,
-      "loss": 5.4207,
-      "step": 6064
-    },
-    {
-      "epoch": 3.162972620599739,
-      "grad_norm": 1.5201994180679321,
-      "learning_rate": 9.440804020100504e-05,
-      "loss": 5.586,
-      "step": 6065
-    },
-    {
-      "epoch": 3.1634941329856585,
-      "grad_norm": 1.363964319229126,
-      "learning_rate": 9.44070351758794e-05,
-      "loss": 5.9386,
-      "step": 6066
-    },
-    {
-      "epoch": 3.1640156453715775,
-      "grad_norm": 1.4356647729873657,
-      "learning_rate": 9.440603015075378e-05,
-      "loss": 5.8257,
-      "step": 6067
-    },
-    {
-      "epoch": 3.1645371577574966,
-      "grad_norm": 1.3996728658676147,
-      "learning_rate": 9.440502512562814e-05,
-      "loss": 5.8291,
-      "step": 6068
-    },
-    {
-      "epoch": 3.165058670143416,
-      "grad_norm": 1.3440876007080078,
-      "learning_rate": 9.440402010050252e-05,
-      "loss": 6.0764,
-      "step": 6069
-    },
-    {
-      "epoch": 3.165580182529335,
-      "grad_norm": 1.410819172859192,
-      "learning_rate": 9.440301507537688e-05,
-      "loss": 6.0274,
-      "step": 6070
-    },
-    {
-      "epoch": 3.166101694915254,
-      "grad_norm": 1.5255370140075684,
-      "learning_rate": 9.440201005025126e-05,
-      "loss": 5.768,
-      "step": 6071
-    },
-    {
-      "epoch": 3.1666232073011735,
-      "grad_norm": 1.7141236066818237,
-      "learning_rate": 9.440100502512563e-05,
-      "loss": 5.0864,
-      "step": 6072
-    },
-    {
-      "epoch": 3.1671447196870925,
-      "grad_norm": 1.5139796733856201,
-      "learning_rate": 9.44e-05,
-      "loss": 5.9771,
-      "step": 6073
-    },
-    {
-      "epoch": 3.1676662320730116,
-      "grad_norm": 1.5054020881652832,
-      "learning_rate": 9.439899497487438e-05,
-      "loss": 5.8732,
-      "step": 6074
-    },
-    {
-      "epoch": 3.168187744458931,
-      "grad_norm": 1.3833472728729248,
-      "learning_rate": 9.439798994974875e-05,
-      "loss": 6.0313,
-      "step": 6075
-    },
-    {
-      "epoch": 3.16870925684485,
-      "grad_norm": 1.567466378211975,
-      "learning_rate": 9.439698492462312e-05,
-      "loss": 5.4994,
-      "step": 6076
-    },
-    {
-      "epoch": 3.169230769230769,
-      "grad_norm": 1.5985236167907715,
-      "learning_rate": 9.439597989949749e-05,
-      "loss": 5.6879,
-      "step": 6077
-    },
-    {
-      "epoch": 3.1697522816166885,
-      "grad_norm": 1.7675803899765015,
-      "learning_rate": 9.439497487437187e-05,
-      "loss": 5.1523,
-      "step": 6078
-    },
-    {
-      "epoch": 3.1702737940026076,
-      "grad_norm": 1.955353021621704,
-      "learning_rate": 9.439396984924623e-05,
-      "loss": 6.0882,
-      "step": 6079
-    },
-    {
-      "epoch": 3.1707953063885266,
-      "grad_norm": 1.4026310443878174,
-      "learning_rate": 9.439296482412061e-05,
-      "loss": 5.9947,
-      "step": 6080
-    },
-    {
-      "epoch": 3.171316818774446,
-      "grad_norm": 1.45640230178833,
-      "learning_rate": 9.439195979899497e-05,
-      "loss": 5.7971,
-      "step": 6081
-    },
-    {
-      "epoch": 3.171838331160365,
-      "grad_norm": 1.6185758113861084,
-      "learning_rate": 9.439095477386935e-05,
-      "loss": 5.7397,
-      "step": 6082
-    },
-    {
-      "epoch": 3.172359843546284,
-      "grad_norm": 1.566951870918274,
-      "learning_rate": 9.438994974874373e-05,
-      "loss": 5.5345,
-      "step": 6083
-    },
-    {
-      "epoch": 3.1728813559322036,
-      "grad_norm": 1.386574387550354,
-      "learning_rate": 9.43889447236181e-05,
-      "loss": 6.0292,
-      "step": 6084
-    },
-    {
-      "epoch": 3.1734028683181226,
-      "grad_norm": 1.3778495788574219,
-      "learning_rate": 9.438793969849247e-05,
-      "loss": 6.2037,
-      "step": 6085
-    },
-    {
-      "epoch": 3.1739243807040416,
-      "grad_norm": 1.4927259683609009,
-      "learning_rate": 9.438693467336683e-05,
-      "loss": 5.8481,
-      "step": 6086
-    },
-    {
-      "epoch": 3.174445893089961,
-      "grad_norm": 1.4684858322143555,
-      "learning_rate": 9.438592964824121e-05,
-      "loss": 5.8527,
-      "step": 6087
-    },
-    {
-      "epoch": 3.17496740547588,
-      "grad_norm": 1.4262646436691284,
-      "learning_rate": 9.438492462311558e-05,
-      "loss": 5.9398,
-      "step": 6088
-    },
-    {
-      "epoch": 3.175488917861799,
-      "grad_norm": 1.5022201538085938,
-      "learning_rate": 9.438391959798995e-05,
-      "loss": 5.6911,
-      "step": 6089
-    },
-    {
-      "epoch": 3.1760104302477186,
-      "grad_norm": 1.6021486520767212,
-      "learning_rate": 9.438291457286432e-05,
-      "loss": 5.4866,
-      "step": 6090
-    },
-    {
-      "epoch": 3.1765319426336376,
-      "grad_norm": 1.419243574142456,
-      "learning_rate": 9.43819095477387e-05,
-      "loss": 5.9273,
-      "step": 6091
-    },
-    {
-      "epoch": 3.1770534550195566,
-      "grad_norm": 1.5401893854141235,
-      "learning_rate": 9.438090452261306e-05,
-      "loss": 5.6473,
-      "step": 6092
-    },
-    {
-      "epoch": 3.177574967405476,
-      "grad_norm": 1.4306854009628296,
-      "learning_rate": 9.437989949748744e-05,
-      "loss": 6.0221,
-      "step": 6093
-    },
-    {
-      "epoch": 3.178096479791395,
-      "grad_norm": 1.4795480966567993,
-      "learning_rate": 9.437889447236182e-05,
-      "loss": 5.8446,
-      "step": 6094
-    },
-    {
-      "epoch": 3.178617992177314,
-      "grad_norm": 1.435036063194275,
-      "learning_rate": 9.43778894472362e-05,
-      "loss": 6.0145,
-      "step": 6095
-    },
-    {
-      "epoch": 3.1791395045632336,
-      "grad_norm": 1.466265082359314,
-      "learning_rate": 9.437688442211056e-05,
-      "loss": 5.4105,
-      "step": 6096
-    },
-    {
-      "epoch": 3.1796610169491526,
-      "grad_norm": 1.4791849851608276,
-      "learning_rate": 9.437587939698494e-05,
-      "loss": 5.816,
-      "step": 6097
-    },
-    {
-      "epoch": 3.1801825293350716,
-      "grad_norm": 1.4403960704803467,
-      "learning_rate": 9.43748743718593e-05,
-      "loss": 5.8518,
-      "step": 6098
-    },
-    {
-      "epoch": 3.1807040417209906,
-      "grad_norm": 1.3632322549819946,
-      "learning_rate": 9.437386934673367e-05,
-      "loss": 6.1304,
-      "step": 6099
-    },
-    {
-      "epoch": 3.18122555410691,
-      "grad_norm": 1.5270798206329346,
-      "learning_rate": 9.437286432160804e-05,
-      "loss": 5.7269,
-      "step": 6100
-    },
-    {
-      "epoch": 3.181747066492829,
-      "grad_norm": 1.3566936254501343,
-      "learning_rate": 9.437185929648241e-05,
-      "loss": 5.7453,
-      "step": 6101
-    },
-    {
-      "epoch": 3.182268578878748,
-      "grad_norm": 1.4867922067642212,
-      "learning_rate": 9.437085427135679e-05,
-      "loss": 5.7366,
-      "step": 6102
-    },
-    {
-      "epoch": 3.1827900912646676,
-      "grad_norm": 1.4395008087158203,
-      "learning_rate": 9.436984924623116e-05,
-      "loss": 6.0992,
-      "step": 6103
-    },
-    {
-      "epoch": 3.1833116036505866,
-      "grad_norm": 1.4493725299835205,
-      "learning_rate": 9.436884422110554e-05,
-      "loss": 5.7234,
-      "step": 6104
-    },
-    {
-      "epoch": 3.1838331160365057,
-      "grad_norm": 1.5806353092193604,
-      "learning_rate": 9.43678391959799e-05,
-      "loss": 5.7826,
-      "step": 6105
-    },
-    {
-      "epoch": 3.184354628422425,
-      "grad_norm": 1.5170636177062988,
-      "learning_rate": 9.436683417085428e-05,
-      "loss": 5.876,
-      "step": 6106
-    },
-    {
-      "epoch": 3.184876140808344,
-      "grad_norm": 1.3262099027633667,
-      "learning_rate": 9.436582914572865e-05,
-      "loss": 5.4689,
-      "step": 6107
-    },
-    {
-      "epoch": 3.185397653194263,
-      "grad_norm": 1.426417589187622,
-      "learning_rate": 9.436482412060302e-05,
-      "loss": 5.9184,
-      "step": 6108
-    },
-    {
-      "epoch": 3.1859191655801826,
-      "grad_norm": 1.5642647743225098,
-      "learning_rate": 9.436381909547739e-05,
-      "loss": 5.6027,
-      "step": 6109
-    },
-    {
-      "epoch": 3.1864406779661016,
-      "grad_norm": 1.4244892597198486,
-      "learning_rate": 9.436281407035177e-05,
-      "loss": 5.8159,
-      "step": 6110
-    },
-    {
-      "epoch": 3.1869621903520207,
-      "grad_norm": 1.5422443151474,
-      "learning_rate": 9.436180904522613e-05,
-      "loss": 5.9187,
-      "step": 6111
-    },
-    {
-      "epoch": 3.18748370273794,
-      "grad_norm": 2.1564443111419678,
-      "learning_rate": 9.43608040201005e-05,
-      "loss": 5.8057,
-      "step": 6112
-    },
-    {
-      "epoch": 3.188005215123859,
-      "grad_norm": 1.6485815048217773,
-      "learning_rate": 9.435979899497487e-05,
-      "loss": 5.4979,
-      "step": 6113
-    },
-    {
-      "epoch": 3.188526727509778,
-      "grad_norm": 1.616163969039917,
-      "learning_rate": 9.435879396984925e-05,
-      "loss": 5.7881,
-      "step": 6114
-    },
-    {
-      "epoch": 3.1890482398956976,
-      "grad_norm": 1.4401814937591553,
-      "learning_rate": 9.435778894472363e-05,
-      "loss": 5.9127,
-      "step": 6115
-    },
-    {
-      "epoch": 3.1895697522816167,
-      "grad_norm": 1.4818191528320312,
-      "learning_rate": 9.4356783919598e-05,
-      "loss": 5.5766,
-      "step": 6116
-    },
-    {
-      "epoch": 3.1900912646675357,
-      "grad_norm": 1.3276259899139404,
-      "learning_rate": 9.435577889447237e-05,
-      "loss": 5.4775,
-      "step": 6117
-    },
-    {
-      "epoch": 3.190612777053455,
-      "grad_norm": 1.566565752029419,
-      "learning_rate": 9.435477386934674e-05,
-      "loss": 5.9043,
-      "step": 6118
-    },
-    {
-      "epoch": 3.191134289439374,
-      "grad_norm": 1.5047739744186401,
-      "learning_rate": 9.435376884422111e-05,
-      "loss": 6.1088,
-      "step": 6119
-    },
-    {
-      "epoch": 3.191655801825293,
-      "grad_norm": 1.364888310432434,
-      "learning_rate": 9.435276381909548e-05,
-      "loss": 5.96,
-      "step": 6120
-    },
-    {
-      "epoch": 3.1921773142112126,
-      "grad_norm": 1.382431983947754,
-      "learning_rate": 9.435175879396986e-05,
-      "loss": 6.1059,
-      "step": 6121
-    },
-    {
-      "epoch": 3.1926988265971317,
-      "grad_norm": 1.442171573638916,
-      "learning_rate": 9.435075376884422e-05,
-      "loss": 5.898,
-      "step": 6122
-    },
-    {
-      "epoch": 3.1932203389830507,
-      "grad_norm": 1.4232732057571411,
-      "learning_rate": 9.43497487437186e-05,
-      "loss": 5.4687,
-      "step": 6123
-    },
-    {
-      "epoch": 3.19374185136897,
-      "grad_norm": 1.4841036796569824,
-      "learning_rate": 9.434874371859298e-05,
-      "loss": 5.698,
-      "step": 6124
-    },
-    {
-      "epoch": 3.194263363754889,
-      "grad_norm": 1.4730920791625977,
-      "learning_rate": 9.434773869346734e-05,
-      "loss": 5.8187,
-      "step": 6125
-    },
-    {
-      "epoch": 3.194784876140808,
-      "grad_norm": 1.4859837293624878,
-      "learning_rate": 9.434673366834172e-05,
-      "loss": 5.7071,
-      "step": 6126
-    },
-    {
-      "epoch": 3.1953063885267277,
-      "grad_norm": 1.4210513830184937,
-      "learning_rate": 9.434572864321608e-05,
-      "loss": 6.1255,
-      "step": 6127
-    },
-    {
-      "epoch": 3.1958279009126467,
-      "grad_norm": 1.3236411809921265,
-      "learning_rate": 9.434472361809046e-05,
-      "loss": 6.0863,
-      "step": 6128
-    },
-    {
-      "epoch": 3.1963494132985657,
-      "grad_norm": 1.4877365827560425,
-      "learning_rate": 9.434371859296482e-05,
-      "loss": 5.6671,
-      "step": 6129
-    },
-    {
-      "epoch": 3.196870925684485,
-      "grad_norm": 1.759805679321289,
-      "learning_rate": 9.43427135678392e-05,
-      "loss": 5.6058,
-      "step": 6130
-    },
-    {
-      "epoch": 3.197392438070404,
-      "grad_norm": 1.6147934198379517,
-      "learning_rate": 9.434170854271357e-05,
-      "loss": 5.3863,
-      "step": 6131
-    },
-    {
-      "epoch": 3.197913950456323,
-      "grad_norm": 1.4364882707595825,
-      "learning_rate": 9.434070351758794e-05,
-      "loss": 5.9259,
-      "step": 6132
-    },
-    {
-      "epoch": 3.1984354628422427,
-      "grad_norm": 1.528192400932312,
-      "learning_rate": 9.433969849246231e-05,
-      "loss": 5.5135,
-      "step": 6133
-    },
-    {
-      "epoch": 3.1989569752281617,
-      "grad_norm": 1.563140869140625,
-      "learning_rate": 9.433869346733669e-05,
-      "loss": 5.6918,
-      "step": 6134
-    },
-    {
-      "epoch": 3.1994784876140807,
-      "grad_norm": 1.3805739879608154,
-      "learning_rate": 9.433768844221106e-05,
-      "loss": 6.0711,
-      "step": 6135
-    },
-    {
-      "epoch": 3.2,
-      "grad_norm": 1.4779897928237915,
-      "learning_rate": 9.433668341708544e-05,
-      "loss": 5.8135,
-      "step": 6136
-    },
-    {
-      "epoch": 3.200521512385919,
-      "grad_norm": 1.4872703552246094,
-      "learning_rate": 9.43356783919598e-05,
-      "loss": 5.7767,
-      "step": 6137
-    },
-    {
-      "epoch": 3.201043024771838,
-      "grad_norm": 1.5415042638778687,
-      "learning_rate": 9.433467336683418e-05,
-      "loss": 5.4083,
-      "step": 6138
-    },
-    {
-      "epoch": 3.2015645371577577,
-      "grad_norm": 1.3856995105743408,
-      "learning_rate": 9.433366834170855e-05,
-      "loss": 5.1743,
-      "step": 6139
-    },
-    {
-      "epoch": 3.2020860495436767,
-      "grad_norm": 1.4336004257202148,
-      "learning_rate": 9.433266331658291e-05,
-      "loss": 5.9174,
-      "step": 6140
-    },
-    {
-      "epoch": 3.2026075619295957,
-      "grad_norm": 1.4959810972213745,
-      "learning_rate": 9.433165829145729e-05,
-      "loss": 5.668,
-      "step": 6141
-    },
-    {
-      "epoch": 3.203129074315515,
-      "grad_norm": 1.725293517112732,
-      "learning_rate": 9.433065326633165e-05,
-      "loss": 5.4051,
-      "step": 6142
-    },
-    {
-      "epoch": 3.203650586701434,
-      "grad_norm": 1.5430099964141846,
-      "learning_rate": 9.432964824120603e-05,
-      "loss": 5.8292,
-      "step": 6143
-    },
-    {
-      "epoch": 3.2041720990873532,
-      "grad_norm": 1.504542350769043,
-      "learning_rate": 9.432864321608041e-05,
-      "loss": 5.9079,
-      "step": 6144
-    },
-    {
-      "epoch": 3.2046936114732727,
-      "grad_norm": 1.5091444253921509,
-      "learning_rate": 9.432763819095479e-05,
-      "loss": 5.9387,
-      "step": 6145
-    },
-    {
-      "epoch": 3.2052151238591917,
-      "grad_norm": 1.4000188112258911,
-      "learning_rate": 9.432663316582915e-05,
-      "loss": 6.0136,
-      "step": 6146
-    },
-    {
-      "epoch": 3.2057366362451107,
-      "grad_norm": 1.5463130474090576,
-      "learning_rate": 9.432562814070353e-05,
-      "loss": 6.0266,
-      "step": 6147
-    },
-    {
-      "epoch": 3.20625814863103,
-      "grad_norm": 1.504245638847351,
-      "learning_rate": 9.43246231155779e-05,
-      "loss": 6.0189,
-      "step": 6148
-    },
-    {
-      "epoch": 3.2067796610169492,
-      "grad_norm": 1.4624865055084229,
-      "learning_rate": 9.432361809045227e-05,
-      "loss": 5.9523,
-      "step": 6149
-    },
-    {
-      "epoch": 3.2073011734028682,
-      "grad_norm": 1.4019678831100464,
-      "learning_rate": 9.432261306532664e-05,
-      "loss": 5.9323,
-      "step": 6150
-    },
-    {
-      "epoch": 3.2078226857887877,
-      "grad_norm": 1.5969626903533936,
-      "learning_rate": 9.432160804020101e-05,
-      "loss": 5.4796,
-      "step": 6151
-    },
-    {
-      "epoch": 3.2083441981747067,
-      "grad_norm": 1.4079861640930176,
-      "learning_rate": 9.432060301507538e-05,
-      "loss": 5.9885,
-      "step": 6152
-    },
-    {
-      "epoch": 3.2088657105606258,
-      "grad_norm": 1.4934377670288086,
-      "learning_rate": 9.431959798994974e-05,
-      "loss": 5.8049,
-      "step": 6153
-    },
-    {
-      "epoch": 3.209387222946545,
-      "grad_norm": 1.526559591293335,
-      "learning_rate": 9.431859296482412e-05,
-      "loss": 5.6143,
-      "step": 6154
-    },
-    {
-      "epoch": 3.2099087353324642,
-      "grad_norm": Infinity,
-      "learning_rate": 9.431859296482412e-05,
-      "loss": 5.4821,
-      "step": 6155
-    },
-    {
-      "epoch": 3.2104302477183833,
-      "grad_norm": 1.4396079778671265,
-      "learning_rate": 9.43175879396985e-05,
-      "loss": 5.8192,
-      "step": 6156
-    },
-    {
-      "epoch": 3.2109517601043023,
-      "grad_norm": 1.6479132175445557,
-      "learning_rate": 9.431658291457288e-05,
-      "loss": 5.1891,
-      "step": 6157
-    },
-    {
-      "epoch": 3.2114732724902217,
-      "grad_norm": 1.5770562887191772,
-      "learning_rate": 9.431557788944724e-05,
-      "loss": 5.7889,
-      "step": 6158
-    },
-    {
-      "epoch": 3.2119947848761408,
-      "grad_norm": 1.4472016096115112,
-      "learning_rate": 9.431457286432162e-05,
-      "loss": 6.1605,
-      "step": 6159
-    },
-    {
-      "epoch": 3.21251629726206,
-      "grad_norm": 1.5895344018936157,
-      "learning_rate": 9.431356783919598e-05,
-      "loss": 5.8367,
-      "step": 6160
-    },
-    {
-      "epoch": 3.2130378096479792,
-      "grad_norm": 1.6122419834136963,
-      "learning_rate": 9.431256281407036e-05,
-      "loss": 5.8029,
-      "step": 6161
-    },
-    {
-      "epoch": 3.2135593220338983,
-      "grad_norm": 1.5912319421768188,
-      "learning_rate": 9.431155778894472e-05,
-      "loss": 5.8672,
-      "step": 6162
-    },
-    {
-      "epoch": 3.2140808344198173,
-      "grad_norm": 1.5507028102874756,
-      "learning_rate": 9.43105527638191e-05,
-      "loss": 5.7505,
-      "step": 6163
-    },
-    {
-      "epoch": 3.2146023468057368,
-      "grad_norm": 1.6034657955169678,
-      "learning_rate": 9.430954773869347e-05,
-      "loss": 5.8018,
-      "step": 6164
-    },
-    {
-      "epoch": 3.2151238591916558,
-      "grad_norm": 1.3653719425201416,
-      "learning_rate": 9.430854271356784e-05,
-      "loss": 5.7991,
-      "step": 6165
-    },
-    {
-      "epoch": 3.215645371577575,
-      "grad_norm": 1.4457855224609375,
-      "learning_rate": 9.430753768844222e-05,
-      "loss": 5.9308,
-      "step": 6166
-    },
-    {
-      "epoch": 3.2161668839634943,
-      "grad_norm": 1.4226210117340088,
-      "learning_rate": 9.430653266331659e-05,
-      "loss": 5.8975,
-      "step": 6167
-    },
-    {
-      "epoch": 3.2166883963494133,
-      "grad_norm": 1.3609356880187988,
-      "learning_rate": 9.430552763819096e-05,
-      "loss": 6.0341,
-      "step": 6168
-    },
-    {
-      "epoch": 3.2172099087353323,
-      "grad_norm": 1.5904911756515503,
-      "learning_rate": 9.430452261306533e-05,
-      "loss": 5.1053,
-      "step": 6169
-    },
-    {
-      "epoch": 3.2177314211212518,
-      "grad_norm": 1.5178179740905762,
-      "learning_rate": 9.43035175879397e-05,
-      "loss": 5.9833,
-      "step": 6170
-    },
-    {
-      "epoch": 3.218252933507171,
-      "grad_norm": 1.5483949184417725,
-      "learning_rate": 9.430251256281407e-05,
-      "loss": 5.8001,
-      "step": 6171
-    },
-    {
-      "epoch": 3.21877444589309,
-      "grad_norm": 1.507005214691162,
-      "learning_rate": 9.430150753768845e-05,
-      "loss": 5.8381,
-      "step": 6172
-    },
-    {
-      "epoch": 3.2192959582790093,
-      "grad_norm": 1.5757869482040405,
-      "learning_rate": 9.430050251256281e-05,
-      "loss": 5.9566,
-      "step": 6173
-    },
-    {
-      "epoch": 3.2198174706649283,
-      "grad_norm": 1.5093121528625488,
-      "learning_rate": 9.429949748743719e-05,
-      "loss": 5.8179,
-      "step": 6174
-    },
-    {
-      "epoch": 3.2203389830508473,
-      "grad_norm": 1.487396001815796,
-      "learning_rate": 9.429849246231155e-05,
-      "loss": 6.0699,
-      "step": 6175
-    },
-    {
-      "epoch": 3.220860495436767,
-      "grad_norm": 1.5610569715499878,
-      "learning_rate": 9.429748743718593e-05,
-      "loss": 5.8325,
-      "step": 6176
-    },
-    {
-      "epoch": 3.221382007822686,
-      "grad_norm": 1.5405446290969849,
-      "learning_rate": 9.429648241206031e-05,
-      "loss": 5.539,
-      "step": 6177
-    },
-    {
-      "epoch": 3.221903520208605,
-      "grad_norm": 1.6973882913589478,
-      "learning_rate": 9.429547738693469e-05,
-      "loss": 5.1859,
-      "step": 6178
-    },
-    {
-      "epoch": 3.2224250325945243,
-      "grad_norm": 1.4973224401474,
-      "learning_rate": 9.429447236180905e-05,
-      "loss": 6.0823,
-      "step": 6179
-    },
-    {
-      "epoch": 3.2229465449804433,
-      "grad_norm": 1.4997273683547974,
-      "learning_rate": 9.429346733668342e-05,
-      "loss": 5.795,
-      "step": 6180
-    },
-    {
-      "epoch": 3.2234680573663623,
-      "grad_norm": 1.463489294052124,
-      "learning_rate": 9.42924623115578e-05,
-      "loss": 5.5135,
-      "step": 6181
-    },
-    {
-      "epoch": 3.223989569752282,
-      "grad_norm": 1.7521713972091675,
-      "learning_rate": 9.429145728643216e-05,
-      "loss": 5.8986,
-      "step": 6182
-    },
-    {
-      "epoch": 3.224511082138201,
-      "grad_norm": 1.4429104328155518,
-      "learning_rate": 9.429045226130654e-05,
-      "loss": 6.2022,
-      "step": 6183
-    },
-    {
-      "epoch": 3.22503259452412,
-      "grad_norm": 1.5415095090866089,
-      "learning_rate": 9.42894472361809e-05,
-      "loss": 5.8418,
-      "step": 6184
-    },
-    {
-      "epoch": 3.2255541069100393,
-      "grad_norm": 1.4995697736740112,
-      "learning_rate": 9.428844221105528e-05,
-      "loss": 5.501,
-      "step": 6185
-    },
-    {
-      "epoch": 3.2260756192959583,
-      "grad_norm": 1.4442633390426636,
-      "learning_rate": 9.428743718592966e-05,
-      "loss": 5.9899,
-      "step": 6186
-    },
-    {
-      "epoch": 3.2265971316818773,
-      "grad_norm": 1.5890473127365112,
-      "learning_rate": 9.428643216080403e-05,
-      "loss": 5.6381,
-      "step": 6187
-    },
-    {
-      "epoch": 3.227118644067797,
-      "grad_norm": 1.4777387380599976,
-      "learning_rate": 9.42854271356784e-05,
-      "loss": 5.3432,
-      "step": 6188
-    },
-    {
-      "epoch": 3.227640156453716,
-      "grad_norm": 1.4187445640563965,
-      "learning_rate": 9.428442211055278e-05,
-      "loss": 5.7174,
-      "step": 6189
-    },
-    {
-      "epoch": 3.228161668839635,
-      "grad_norm": 1.4590586423873901,
-      "learning_rate": 9.428341708542714e-05,
-      "loss": 5.5079,
-      "step": 6190
-    },
-    {
-      "epoch": 3.2286831812255543,
-      "grad_norm": 1.3635677099227905,
-      "learning_rate": 9.428241206030152e-05,
-      "loss": 5.7444,
-      "step": 6191
-    },
-    {
-      "epoch": 3.2292046936114733,
-      "grad_norm": 1.4760468006134033,
-      "learning_rate": 9.428140703517588e-05,
-      "loss": 5.7087,
-      "step": 6192
-    },
-    {
-      "epoch": 3.2297262059973924,
-      "grad_norm": 1.5209574699401855,
-      "learning_rate": 9.428040201005025e-05,
-      "loss": 5.0851,
-      "step": 6193
-    },
-    {
-      "epoch": 3.2302477183833114,
-      "grad_norm": 1.878067970275879,
-      "learning_rate": 9.427939698492463e-05,
-      "loss": 5.6347,
-      "step": 6194
-    },
-    {
-      "epoch": 3.230769230769231,
-      "grad_norm": 1.3821834325790405,
-      "learning_rate": 9.427839195979899e-05,
-      "loss": 5.679,
-      "step": 6195
-    },
-    {
-      "epoch": 3.23129074315515,
-      "grad_norm": 1.4105803966522217,
-      "learning_rate": 9.427738693467337e-05,
-      "loss": 5.7259,
-      "step": 6196
-    },
-    {
-      "epoch": 3.231812255541069,
-      "grad_norm": 1.5469595193862915,
-      "learning_rate": 9.427638190954775e-05,
-      "loss": 5.6666,
-      "step": 6197
-    },
-    {
-      "epoch": 3.2323337679269883,
-      "grad_norm": 1.4934452772140503,
-      "learning_rate": 9.427537688442212e-05,
-      "loss": 5.738,
-      "step": 6198
-    },
-    {
-      "epoch": 3.2328552803129074,
-      "grad_norm": 1.5218204259872437,
-      "learning_rate": 9.427437185929649e-05,
-      "loss": 5.7836,
-      "step": 6199
-    },
-    {
-      "epoch": 3.2333767926988264,
-      "grad_norm": 1.5871773958206177,
-      "learning_rate": 9.427336683417087e-05,
-      "loss": 5.9141,
-      "step": 6200
-    },
-    {
-      "epoch": 3.233898305084746,
-      "grad_norm": 1.7273310422897339,
-      "learning_rate": 9.427236180904523e-05,
-      "loss": 5.7793,
-      "step": 6201
-    },
-    {
-      "epoch": 3.234419817470665,
-      "grad_norm": 1.4747353792190552,
-      "learning_rate": 9.427135678391961e-05,
-      "loss": 5.7821,
-      "step": 6202
-    },
-    {
-      "epoch": 3.234941329856584,
-      "grad_norm": 1.597123384475708,
-      "learning_rate": 9.427035175879397e-05,
-      "loss": 5.7171,
-      "step": 6203
-    },
-    {
-      "epoch": 3.2354628422425034,
-      "grad_norm": 1.7552379369735718,
-      "learning_rate": 9.426934673366835e-05,
-      "loss": 5.4173,
-      "step": 6204
-    },
-    {
-      "epoch": 3.2359843546284224,
-      "grad_norm": 1.599637508392334,
-      "learning_rate": 9.426834170854271e-05,
-      "loss": 5.8695,
-      "step": 6205
-    },
-    {
-      "epoch": 3.2365058670143414,
-      "grad_norm": 1.4572144746780396,
-      "learning_rate": 9.426733668341708e-05,
-      "loss": 5.8079,
-      "step": 6206
-    },
-    {
-      "epoch": 3.237027379400261,
-      "grad_norm": 1.5812299251556396,
-      "learning_rate": 9.426633165829146e-05,
-      "loss": 5.6675,
-      "step": 6207
-    },
-    {
-      "epoch": 3.23754889178618,
-      "grad_norm": 1.8448617458343506,
-      "learning_rate": 9.426532663316583e-05,
-      "loss": 5.1875,
-      "step": 6208
-    },
-    {
-      "epoch": 3.238070404172099,
-      "grad_norm": 1.4621716737747192,
-      "learning_rate": 9.426432160804021e-05,
-      "loss": 5.9924,
-      "step": 6209
-    },
-    {
-      "epoch": 3.2385919165580184,
-      "grad_norm": 1.5728709697723389,
-      "learning_rate": 9.426331658291458e-05,
-      "loss": 5.5001,
-      "step": 6210
-    },
-    {
-      "epoch": 3.2391134289439374,
-      "grad_norm": 1.4774038791656494,
-      "learning_rate": 9.426231155778895e-05,
-      "loss": 5.579,
-      "step": 6211
-    },
-    {
-      "epoch": 3.2396349413298564,
-      "grad_norm": 1.3596423864364624,
-      "learning_rate": 9.426130653266332e-05,
-      "loss": 5.9127,
-      "step": 6212
-    },
-    {
-      "epoch": 3.240156453715776,
-      "grad_norm": 1.4205951690673828,
-      "learning_rate": 9.42603015075377e-05,
-      "loss": 6.031,
-      "step": 6213
-    },
-    {
-      "epoch": 3.240677966101695,
-      "grad_norm": 1.4626179933547974,
-      "learning_rate": 9.425929648241206e-05,
-      "loss": 5.7473,
-      "step": 6214
-    },
-    {
-      "epoch": 3.241199478487614,
-      "grad_norm": 1.5989692211151123,
-      "learning_rate": 9.425829145728644e-05,
-      "loss": 6.1115,
-      "step": 6215
-    },
-    {
-      "epoch": 3.2417209908735334,
-      "grad_norm": 1.9289218187332153,
-      "learning_rate": 9.42572864321608e-05,
-      "loss": 5.3352,
-      "step": 6216
-    },
-    {
-      "epoch": 3.2422425032594524,
-      "grad_norm": 1.7610050439834595,
-      "learning_rate": 9.425628140703518e-05,
-      "loss": 5.5337,
-      "step": 6217
-    },
-    {
-      "epoch": 3.2427640156453714,
-      "grad_norm": 1.6252809762954712,
-      "learning_rate": 9.425527638190956e-05,
-      "loss": 5.7601,
-      "step": 6218
-    },
-    {
-      "epoch": 3.243285528031291,
-      "grad_norm": 2.0292553901672363,
-      "learning_rate": 9.425427135678392e-05,
-      "loss": 5.61,
-      "step": 6219
-    },
-    {
-      "epoch": 3.24380704041721,
-      "grad_norm": 1.5304805040359497,
-      "learning_rate": 9.42532663316583e-05,
-      "loss": 5.5402,
-      "step": 6220
-    },
-    {
-      "epoch": 3.244328552803129,
-      "grad_norm": 2.130779266357422,
-      "learning_rate": 9.425226130653266e-05,
-      "loss": 5.5874,
-      "step": 6221
-    },
-    {
-      "epoch": 3.2448500651890484,
-      "grad_norm": 1.6013003587722778,
-      "learning_rate": 9.425125628140704e-05,
-      "loss": 5.8521,
-      "step": 6222
-    },
-    {
-      "epoch": 3.2453715775749674,
-      "grad_norm": 1.3950397968292236,
-      "learning_rate": 9.42502512562814e-05,
-      "loss": 6.0946,
-      "step": 6223
-    },
-    {
-      "epoch": 3.2458930899608864,
-      "grad_norm": 1.4838836193084717,
-      "learning_rate": 9.424924623115578e-05,
-      "loss": 5.8705,
-      "step": 6224
-    },
-    {
-      "epoch": 3.246414602346806,
-      "grad_norm": 1.5574755668640137,
-      "learning_rate": 9.424824120603015e-05,
-      "loss": 5.2937,
-      "step": 6225
-    },
-    {
-      "epoch": 3.246936114732725,
-      "grad_norm": 1.6120214462280273,
-      "learning_rate": 9.424723618090453e-05,
-      "loss": 5.7608,
-      "step": 6226
-    },
-    {
-      "epoch": 3.247457627118644,
-      "grad_norm": 2.0517330169677734,
-      "learning_rate": 9.424623115577889e-05,
-      "loss": 5.3524,
-      "step": 6227
-    },
-    {
-      "epoch": 3.2479791395045634,
-      "grad_norm": 1.5514570474624634,
-      "learning_rate": 9.424522613065327e-05,
-      "loss": 5.5999,
-      "step": 6228
-    },
-    {
-      "epoch": 3.2485006518904824,
-      "grad_norm": 1.4709736108779907,
-      "learning_rate": 9.424422110552765e-05,
-      "loss": 5.0958,
-      "step": 6229
-    },
-    {
-      "epoch": 3.2490221642764014,
-      "grad_norm": 1.4478611946105957,
-      "learning_rate": 9.424321608040202e-05,
-      "loss": 5.348,
-      "step": 6230
-    },
-    {
-      "epoch": 3.249543676662321,
-      "grad_norm": 1.3686213493347168,
-      "learning_rate": 9.424221105527639e-05,
-      "loss": 5.5376,
-      "step": 6231
-    },
-    {
-      "epoch": 3.25006518904824,
-      "grad_norm": 1.3932489156723022,
-      "learning_rate": 9.424120603015077e-05,
-      "loss": 5.7222,
-      "step": 6232
-    },
-    {
-      "epoch": 3.250586701434159,
-      "grad_norm": 1.4102818965911865,
-      "learning_rate": 9.424020100502513e-05,
-      "loss": 5.9446,
-      "step": 6233
-    },
-    {
-      "epoch": 3.2511082138200784,
-      "grad_norm": 1.4870766401290894,
-      "learning_rate": 9.42391959798995e-05,
-      "loss": 5.6183,
-      "step": 6234
-    },
-    {
-      "epoch": 3.2516297262059974,
-      "grad_norm": 1.5106326341629028,
-      "learning_rate": 9.423819095477387e-05,
-      "loss": 5.3844,
-      "step": 6235
-    },
-    {
-      "epoch": 3.2521512385919165,
-      "grad_norm": 1.4735982418060303,
-      "learning_rate": 9.423718592964824e-05,
-      "loss": 5.4775,
-      "step": 6236
-    },
-    {
-      "epoch": 3.252672750977836,
-      "grad_norm": 1.5121105909347534,
-      "learning_rate": 9.423618090452261e-05,
-      "loss": 5.6267,
-      "step": 6237
-    },
-    {
-      "epoch": 3.253194263363755,
-      "grad_norm": 1.3464893102645874,
-      "learning_rate": 9.423517587939699e-05,
-      "loss": 5.8976,
-      "step": 6238
-    },
-    {
-      "epoch": 3.253715775749674,
-      "grad_norm": 1.4475207328796387,
-      "learning_rate": 9.423417085427137e-05,
-      "loss": 6.142,
-      "step": 6239
-    },
-    {
-      "epoch": 3.2542372881355934,
-      "grad_norm": 1.5315498113632202,
-      "learning_rate": 9.423316582914573e-05,
-      "loss": 5.5301,
-      "step": 6240
-    },
-    {
-      "epoch": 3.2547588005215125,
-      "grad_norm": 1.4776184558868408,
-      "learning_rate": 9.423216080402011e-05,
-      "loss": 6.0038,
-      "step": 6241
-    },
-    {
-      "epoch": 3.2552803129074315,
-      "grad_norm": 1.9797841310501099,
-      "learning_rate": 9.423115577889448e-05,
-      "loss": 5.6206,
-      "step": 6242
-    },
-    {
-      "epoch": 3.255801825293351,
-      "grad_norm": 1.600041151046753,
-      "learning_rate": 9.423015075376885e-05,
-      "loss": 5.4084,
-      "step": 6243
-    },
-    {
-      "epoch": 3.25632333767927,
-      "grad_norm": 1.8355685472488403,
-      "learning_rate": 9.422914572864322e-05,
-      "loss": 5.497,
-      "step": 6244
-    },
-    {
-      "epoch": 3.256844850065189,
-      "grad_norm": 1.7045159339904785,
-      "learning_rate": 9.42281407035176e-05,
-      "loss": 5.5723,
-      "step": 6245
-    },
-    {
-      "epoch": 3.2573663624511084,
-      "grad_norm": 1.6835495233535767,
-      "learning_rate": 9.422713567839196e-05,
-      "loss": 5.5444,
-      "step": 6246
-    },
-    {
-      "epoch": 3.2578878748370275,
-      "grad_norm": 1.5292305946350098,
-      "learning_rate": 9.422613065326632e-05,
-      "loss": 5.8151,
-      "step": 6247
-    },
-    {
-      "epoch": 3.2584093872229465,
-      "grad_norm": 1.321090817451477,
-      "learning_rate": 9.42251256281407e-05,
-      "loss": 5.5419,
-      "step": 6248
-    },
-    {
-      "epoch": 3.258930899608866,
-      "grad_norm": 1.6109527349472046,
-      "learning_rate": 9.422412060301508e-05,
-      "loss": 5.3523,
-      "step": 6249
-    },
-    {
-      "epoch": 3.259452411994785,
-      "grad_norm": 1.4249200820922852,
-      "learning_rate": 9.422311557788946e-05,
-      "loss": 5.9587,
-      "step": 6250
-    },
-    {
-      "epoch": 3.259973924380704,
-      "grad_norm": 1.5037729740142822,
-      "learning_rate": 9.422211055276382e-05,
-      "loss": 5.8155,
-      "step": 6251
-    },
-    {
-      "epoch": 3.2604954367666235,
-      "grad_norm": 1.536786437034607,
-      "learning_rate": 9.42211055276382e-05,
-      "loss": 5.7871,
-      "step": 6252
-    },
-    {
-      "epoch": 3.2610169491525425,
-      "grad_norm": 1.4061353206634521,
-      "learning_rate": 9.422010050251256e-05,
-      "loss": 5.838,
-      "step": 6253
-    },
-    {
-      "epoch": 3.2615384615384615,
-      "grad_norm": 1.4665380716323853,
-      "learning_rate": 9.421909547738694e-05,
-      "loss": 5.5961,
-      "step": 6254
-    },
-    {
-      "epoch": 3.2620599739243805,
-      "grad_norm": 1.559973120689392,
-      "learning_rate": 9.421809045226131e-05,
-      "loss": 5.863,
-      "step": 6255
-    },
-    {
-      "epoch": 3.2625814863103,
-      "grad_norm": 1.5399184226989746,
-      "learning_rate": 9.421708542713568e-05,
-      "loss": 5.9697,
-      "step": 6256
-    },
-    {
-      "epoch": 3.263102998696219,
-      "grad_norm": 1.605275273323059,
-      "learning_rate": 9.421608040201005e-05,
-      "loss": 5.6683,
-      "step": 6257
-    },
-    {
-      "epoch": 3.263624511082138,
-      "grad_norm": 1.561556100845337,
-      "learning_rate": 9.421507537688443e-05,
-      "loss": 5.604,
-      "step": 6258
-    },
-    {
-      "epoch": 3.2641460234680575,
-      "grad_norm": 1.568371057510376,
-      "learning_rate": 9.42140703517588e-05,
-      "loss": 5.4717,
-      "step": 6259
-    },
-    {
-      "epoch": 3.2646675358539765,
-      "grad_norm": 1.5287699699401855,
-      "learning_rate": 9.421306532663317e-05,
-      "loss": 5.5309,
-      "step": 6260
-    },
-    {
-      "epoch": 3.2651890482398955,
-      "grad_norm": 1.4565407037734985,
-      "learning_rate": 9.421206030150755e-05,
-      "loss": 5.991,
-      "step": 6261
-    },
-    {
-      "epoch": 3.265710560625815,
-      "grad_norm": 1.5254547595977783,
-      "learning_rate": 9.421105527638191e-05,
-      "loss": 5.6364,
-      "step": 6262
-    },
-    {
-      "epoch": 3.266232073011734,
-      "grad_norm": 1.5330129861831665,
-      "learning_rate": 9.421005025125629e-05,
-      "loss": 5.8375,
-      "step": 6263
-    },
-    {
-      "epoch": 3.266753585397653,
-      "grad_norm": 1.5813093185424805,
-      "learning_rate": 9.420904522613065e-05,
-      "loss": 5.5621,
-      "step": 6264
-    },
-    {
-      "epoch": 3.2672750977835725,
-      "grad_norm": 1.4711748361587524,
-      "learning_rate": 9.420804020100503e-05,
-      "loss": 5.9392,
-      "step": 6265
-    },
-    {
-      "epoch": 3.2677966101694915,
-      "grad_norm": 1.3868285417556763,
-      "learning_rate": 9.42070351758794e-05,
-      "loss": 6.0265,
-      "step": 6266
-    },
-    {
-      "epoch": 3.2683181225554105,
-      "grad_norm": 1.5267021656036377,
-      "learning_rate": 9.420603015075377e-05,
-      "loss": 5.3824,
-      "step": 6267
-    },
-    {
-      "epoch": 3.26883963494133,
-      "grad_norm": 1.511199951171875,
-      "learning_rate": 9.420502512562814e-05,
-      "loss": 5.5165,
-      "step": 6268
-    },
-    {
-      "epoch": 3.269361147327249,
-      "grad_norm": 1.5234041213989258,
-      "learning_rate": 9.420402010050252e-05,
-      "loss": 5.3421,
-      "step": 6269
-    },
-    {
-      "epoch": 3.269882659713168,
-      "grad_norm": 1.406374216079712,
-      "learning_rate": 9.420301507537689e-05,
-      "loss": 5.9479,
-      "step": 6270
-    },
-    {
-      "epoch": 3.2704041720990875,
-      "grad_norm": 1.6851351261138916,
-      "learning_rate": 9.420201005025127e-05,
-      "loss": 5.4759,
-      "step": 6271
-    },
-    {
-      "epoch": 3.2709256844850065,
-      "grad_norm": 1.625986099243164,
-      "learning_rate": 9.420100502512564e-05,
-      "loss": 5.9714,
-      "step": 6272
-    },
-    {
-      "epoch": 3.2714471968709256,
-      "grad_norm": 1.6122866868972778,
-      "learning_rate": 9.42e-05,
-      "loss": 5.7673,
-      "step": 6273
-    },
-    {
-      "epoch": 3.271968709256845,
-      "grad_norm": 1.575973391532898,
-      "learning_rate": 9.419899497487438e-05,
-      "loss": 5.3833,
-      "step": 6274
-    },
-    {
-      "epoch": 3.272490221642764,
-      "grad_norm": 1.5484036207199097,
-      "learning_rate": 9.419798994974874e-05,
-      "loss": 5.8943,
-      "step": 6275
-    },
-    {
-      "epoch": 3.273011734028683,
-      "grad_norm": 1.5021764039993286,
-      "learning_rate": 9.419698492462312e-05,
-      "loss": 5.7167,
-      "step": 6276
-    },
-    {
-      "epoch": 3.2735332464146025,
-      "grad_norm": 1.3298256397247314,
-      "learning_rate": 9.419597989949748e-05,
-      "loss": 5.8269,
-      "step": 6277
-    },
-    {
-      "epoch": 3.2740547588005215,
-      "grad_norm": 1.4415249824523926,
-      "learning_rate": 9.419497487437186e-05,
-      "loss": 5.8175,
-      "step": 6278
-    },
-    {
-      "epoch": 3.2745762711864406,
-      "grad_norm": 1.510022521018982,
-      "learning_rate": 9.419396984924624e-05,
-      "loss": 5.9399,
-      "step": 6279
-    },
-    {
-      "epoch": 3.27509778357236,
-      "grad_norm": 1.5080536603927612,
-      "learning_rate": 9.419296482412062e-05,
-      "loss": 6.111,
-      "step": 6280
-    },
-    {
-      "epoch": 3.275619295958279,
-      "grad_norm": 1.4757399559020996,
-      "learning_rate": 9.419195979899498e-05,
-      "loss": 5.955,
-      "step": 6281
-    },
-    {
-      "epoch": 3.276140808344198,
-      "grad_norm": 1.520505428314209,
-      "learning_rate": 9.419095477386936e-05,
-      "loss": 5.5962,
-      "step": 6282
-    },
-    {
-      "epoch": 3.276662320730117,
-      "grad_norm": 1.5172866582870483,
-      "learning_rate": 9.418994974874372e-05,
-      "loss": 5.8843,
-      "step": 6283
-    },
-    {
-      "epoch": 3.2771838331160366,
-      "grad_norm": 1.4995776414871216,
-      "learning_rate": 9.41889447236181e-05,
-      "loss": 5.85,
-      "step": 6284
-    },
-    {
-      "epoch": 3.2777053455019556,
-      "grad_norm": 1.474346399307251,
-      "learning_rate": 9.418793969849247e-05,
-      "loss": 5.6411,
-      "step": 6285
-    },
-    {
-      "epoch": 3.2782268578878746,
-      "grad_norm": 1.4377052783966064,
-      "learning_rate": 9.418693467336683e-05,
-      "loss": 6.004,
-      "step": 6286
-    },
-    {
-      "epoch": 3.278748370273794,
-      "grad_norm": 1.3709311485290527,
-      "learning_rate": 9.418592964824121e-05,
-      "loss": 6.1462,
-      "step": 6287
-    },
-    {
-      "epoch": 3.279269882659713,
-      "grad_norm": 1.5570496320724487,
-      "learning_rate": 9.418492462311557e-05,
-      "loss": 5.7198,
-      "step": 6288
-    },
-    {
-      "epoch": 3.279791395045632,
-      "grad_norm": 1.5457420349121094,
-      "learning_rate": 9.418391959798995e-05,
-      "loss": 5.7194,
-      "step": 6289
-    },
-    {
-      "epoch": 3.2803129074315516,
-      "grad_norm": 1.5750361680984497,
-      "learning_rate": 9.418291457286433e-05,
-      "loss": 5.8121,
-      "step": 6290
-    },
-    {
-      "epoch": 3.2808344198174706,
-      "grad_norm": 1.5564978122711182,
-      "learning_rate": 9.41819095477387e-05,
-      "loss": 6.1313,
-      "step": 6291
-    },
-    {
-      "epoch": 3.2813559322033896,
-      "grad_norm": 1.36670982837677,
-      "learning_rate": 9.418090452261307e-05,
-      "loss": 5.7456,
-      "step": 6292
-    },
-    {
-      "epoch": 3.281877444589309,
-      "grad_norm": 1.5152801275253296,
-      "learning_rate": 9.417989949748745e-05,
-      "loss": 5.859,
-      "step": 6293
-    },
-    {
-      "epoch": 3.282398956975228,
-      "grad_norm": 1.5515871047973633,
-      "learning_rate": 9.417889447236181e-05,
-      "loss": 5.7316,
-      "step": 6294
-    },
-    {
-      "epoch": 3.282920469361147,
-      "grad_norm": 1.3828035593032837,
-      "learning_rate": 9.417788944723619e-05,
-      "loss": 5.8897,
-      "step": 6295
-    },
-    {
-      "epoch": 3.2834419817470666,
-      "grad_norm": 1.3857958316802979,
-      "learning_rate": 9.417688442211055e-05,
-      "loss": 6.0004,
-      "step": 6296
-    },
-    {
-      "epoch": 3.2839634941329856,
-      "grad_norm": 1.5058804750442505,
-      "learning_rate": 9.417587939698493e-05,
-      "loss": 5.8873,
-      "step": 6297
-    },
-    {
-      "epoch": 3.2844850065189046,
-      "grad_norm": 1.6497896909713745,
-      "learning_rate": 9.41748743718593e-05,
-      "loss": 5.5469,
-      "step": 6298
-    },
-    {
-      "epoch": 3.285006518904824,
-      "grad_norm": 1.5003020763397217,
-      "learning_rate": 9.417386934673367e-05,
-      "loss": 5.9339,
-      "step": 6299
-    },
-    {
-      "epoch": 3.285528031290743,
-      "grad_norm": 1.633744239807129,
-      "learning_rate": 9.417286432160805e-05,
-      "loss": 5.5447,
-      "step": 6300
-    },
-    {
-      "epoch": 3.286049543676662,
-      "grad_norm": 1.451022744178772,
-      "learning_rate": 9.417185929648242e-05,
-      "loss": 6.1105,
-      "step": 6301
-    },
-    {
-      "epoch": 3.2865710560625816,
-      "grad_norm": 1.4520457983016968,
-      "learning_rate": 9.41708542713568e-05,
-      "loss": 5.7056,
-      "step": 6302
-    },
-    {
-      "epoch": 3.2870925684485006,
-      "grad_norm": 1.5921359062194824,
-      "learning_rate": 9.416984924623116e-05,
-      "loss": 5.5969,
-      "step": 6303
-    },
-    {
-      "epoch": 3.2876140808344196,
-      "grad_norm": 1.4548046588897705,
-      "learning_rate": 9.416884422110554e-05,
-      "loss": 5.8507,
-      "step": 6304
-    },
-    {
-      "epoch": 3.288135593220339,
-      "grad_norm": 1.4354314804077148,
-      "learning_rate": 9.41678391959799e-05,
-      "loss": 5.9224,
-      "step": 6305
-    },
-    {
-      "epoch": 3.288657105606258,
-      "grad_norm": 1.568544864654541,
-      "learning_rate": 9.416683417085428e-05,
-      "loss": 5.6933,
-      "step": 6306
-    },
-    {
-      "epoch": 3.289178617992177,
-      "grad_norm": 1.408516526222229,
-      "learning_rate": 9.416582914572864e-05,
-      "loss": 5.3399,
-      "step": 6307
-    },
-    {
-      "epoch": 3.2897001303780966,
-      "grad_norm": 1.516940951347351,
-      "learning_rate": 9.416482412060302e-05,
-      "loss": 6.087,
-      "step": 6308
-    },
-    {
-      "epoch": 3.2902216427640156,
-      "grad_norm": 1.546155333518982,
-      "learning_rate": 9.416381909547738e-05,
-      "loss": 5.8785,
-      "step": 6309
-    },
-    {
-      "epoch": 3.2907431551499347,
-      "grad_norm": 1.515830159187317,
-      "learning_rate": 9.416281407035176e-05,
-      "loss": 5.8744,
-      "step": 6310
-    },
-    {
-      "epoch": 3.291264667535854,
-      "grad_norm": 1.4667038917541504,
-      "learning_rate": 9.416180904522614e-05,
-      "loss": 5.7633,
-      "step": 6311
-    },
-    {
-      "epoch": 3.291786179921773,
-      "grad_norm": 1.4612380266189575,
-      "learning_rate": 9.41608040201005e-05,
-      "loss": 5.5035,
-      "step": 6312
-    },
-    {
-      "epoch": 3.292307692307692,
-      "grad_norm": 1.676938533782959,
-      "learning_rate": 9.415979899497488e-05,
-      "loss": 5.3423,
-      "step": 6313
-    },
-    {
-      "epoch": 3.2928292046936116,
-      "grad_norm": 1.6605854034423828,
-      "learning_rate": 9.415879396984925e-05,
-      "loss": 6.0312,
-      "step": 6314
-    },
-    {
-      "epoch": 3.2933507170795306,
-      "grad_norm": 1.5755916833877563,
-      "learning_rate": 9.415778894472362e-05,
-      "loss": 5.7056,
-      "step": 6315
-    },
-    {
-      "epoch": 3.2938722294654497,
-      "grad_norm": 1.5381625890731812,
-      "learning_rate": 9.415678391959799e-05,
-      "loss": 5.7779,
-      "step": 6316
-    },
-    {
-      "epoch": 3.294393741851369,
-      "grad_norm": 1.5037357807159424,
-      "learning_rate": 9.415577889447237e-05,
-      "loss": 5.9729,
-      "step": 6317
-    },
-    {
-      "epoch": 3.294915254237288,
-      "grad_norm": 1.6524542570114136,
-      "learning_rate": 9.415477386934673e-05,
-      "loss": 5.197,
-      "step": 6318
-    },
-    {
-      "epoch": 3.295436766623207,
-      "grad_norm": 1.7889353036880493,
-      "learning_rate": 9.415376884422111e-05,
-      "loss": 5.5201,
-      "step": 6319
-    },
-    {
-      "epoch": 3.2959582790091266,
-      "grad_norm": 1.4422242641448975,
-      "learning_rate": 9.415276381909549e-05,
-      "loss": 5.5314,
-      "step": 6320
-    },
-    {
-      "epoch": 3.2964797913950457,
-      "grad_norm": 1.4173004627227783,
-      "learning_rate": 9.415175879396986e-05,
-      "loss": 6.0459,
-      "step": 6321
-    },
-    {
-      "epoch": 3.2970013037809647,
-      "grad_norm": 1.5218665599822998,
-      "learning_rate": 9.415075376884423e-05,
-      "loss": 5.8927,
-      "step": 6322
-    },
-    {
-      "epoch": 3.297522816166884,
-      "grad_norm": 1.5410388708114624,
-      "learning_rate": 9.41497487437186e-05,
-      "loss": 5.8328,
-      "step": 6323
-    },
-    {
-      "epoch": 3.298044328552803,
-      "grad_norm": 1.4395819902420044,
-      "learning_rate": 9.414874371859297e-05,
-      "loss": 5.5142,
-      "step": 6324
-    },
-    {
-      "epoch": 3.298565840938722,
-      "grad_norm": 1.6833597421646118,
-      "learning_rate": 9.414773869346735e-05,
-      "loss": 5.1359,
-      "step": 6325
-    },
-    {
-      "epoch": 3.2990873533246416,
-      "grad_norm": 1.6448813676834106,
-      "learning_rate": 9.414673366834171e-05,
-      "loss": 5.4546,
-      "step": 6326
-    },
-    {
-      "epoch": 3.2996088657105607,
-      "grad_norm": 1.4420127868652344,
-      "learning_rate": 9.414572864321608e-05,
-      "loss": 5.8197,
-      "step": 6327
-    },
-    {
-      "epoch": 3.3001303780964797,
-      "grad_norm": 1.7373369932174683,
-      "learning_rate": 9.414472361809045e-05,
-      "loss": 5.5324,
-      "step": 6328
-    },
-    {
-      "epoch": 3.300651890482399,
-      "grad_norm": 1.632926106452942,
-      "learning_rate": 9.414371859296482e-05,
-      "loss": 4.6859,
-      "step": 6329
-    },
-    {
-      "epoch": 3.301173402868318,
-      "grad_norm": 1.5780954360961914,
-      "learning_rate": 9.41427135678392e-05,
-      "loss": 5.6283,
-      "step": 6330
-    },
-    {
-      "epoch": 3.301694915254237,
-      "grad_norm": 1.4100595712661743,
-      "learning_rate": 9.414170854271357e-05,
-      "loss": 5.762,
-      "step": 6331
-    },
-    {
-      "epoch": 3.3022164276401567,
-      "grad_norm": 1.4546133279800415,
-      "learning_rate": 9.414070351758795e-05,
-      "loss": 5.5743,
-      "step": 6332
-    },
-    {
-      "epoch": 3.3027379400260757,
-      "grad_norm": 1.309650182723999,
-      "learning_rate": 9.413969849246232e-05,
-      "loss": 5.936,
-      "step": 6333
-    },
-    {
-      "epoch": 3.3032594524119947,
-      "grad_norm": 1.7646243572235107,
-      "learning_rate": 9.41386934673367e-05,
-      "loss": 5.2422,
-      "step": 6334
-    },
-    {
-      "epoch": 3.303780964797914,
-      "grad_norm": 1.6041548252105713,
-      "learning_rate": 9.413768844221106e-05,
-      "loss": 5.9187,
-      "step": 6335
-    },
-    {
-      "epoch": 3.304302477183833,
-      "grad_norm": 1.5420531034469604,
-      "learning_rate": 9.413668341708544e-05,
-      "loss": 6.1333,
-      "step": 6336
-    },
-    {
-      "epoch": 3.304823989569752,
-      "grad_norm": 1.463961124420166,
-      "learning_rate": 9.41356783919598e-05,
-      "loss": 5.7786,
-      "step": 6337
-    },
-    {
-      "epoch": 3.3053455019556717,
-      "grad_norm": 1.6471130847930908,
-      "learning_rate": 9.413467336683418e-05,
-      "loss": 5.2185,
-      "step": 6338
-    },
-    {
-      "epoch": 3.3058670143415907,
-      "grad_norm": 1.5845367908477783,
-      "learning_rate": 9.413366834170854e-05,
-      "loss": 5.9985,
-      "step": 6339
-    },
-    {
-      "epoch": 3.3063885267275097,
-      "grad_norm": 1.623010277748108,
-      "learning_rate": 9.413266331658292e-05,
-      "loss": 5.1589,
-      "step": 6340
-    },
-    {
-      "epoch": 3.306910039113429,
-      "grad_norm": 1.464404582977295,
-      "learning_rate": 9.41316582914573e-05,
-      "loss": 5.8392,
-      "step": 6341
-    },
-    {
-      "epoch": 3.307431551499348,
-      "grad_norm": 1.41786789894104,
-      "learning_rate": 9.413065326633166e-05,
-      "loss": 6.0195,
-      "step": 6342
-    },
-    {
-      "epoch": 3.307953063885267,
-      "grad_norm": 1.5298436880111694,
-      "learning_rate": 9.412964824120604e-05,
-      "loss": 5.4,
-      "step": 6343
-    },
-    {
-      "epoch": 3.3084745762711867,
-      "grad_norm": 1.4281961917877197,
-      "learning_rate": 9.41286432160804e-05,
-      "loss": 5.7986,
-      "step": 6344
-    },
-    {
-      "epoch": 3.3089960886571057,
-      "grad_norm": 1.468389868736267,
-      "learning_rate": 9.412763819095478e-05,
-      "loss": 5.7825,
-      "step": 6345
-    },
-    {
-      "epoch": 3.3095176010430247,
-      "grad_norm": 1.503420114517212,
-      "learning_rate": 9.412663316582915e-05,
-      "loss": 5.8598,
-      "step": 6346
-    },
-    {
-      "epoch": 3.3100391134289437,
-      "grad_norm": 1.594232201576233,
-      "learning_rate": 9.412562814070352e-05,
-      "loss": 5.7053,
-      "step": 6347
-    },
-    {
-      "epoch": 3.310560625814863,
-      "grad_norm": 1.617573618888855,
-      "learning_rate": 9.412462311557789e-05,
-      "loss": 5.9103,
-      "step": 6348
-    },
-    {
-      "epoch": 3.3110821382007822,
-      "grad_norm": 1.515859603881836,
-      "learning_rate": 9.412361809045227e-05,
-      "loss": 5.7651,
-      "step": 6349
-    },
-    {
-      "epoch": 3.3116036505867013,
-      "grad_norm": 1.4639352560043335,
-      "learning_rate": 9.412261306532663e-05,
-      "loss": 6.1045,
-      "step": 6350
-    },
-    {
-      "epoch": 3.3121251629726207,
-      "grad_norm": 1.443021535873413,
-      "learning_rate": 9.412160804020101e-05,
-      "loss": 5.7979,
-      "step": 6351
-    },
-    {
-      "epoch": 3.3126466753585397,
-      "grad_norm": 1.6071858406066895,
-      "learning_rate": 9.412060301507539e-05,
-      "loss": 5.7652,
-      "step": 6352
-    },
-    {
-      "epoch": 3.3131681877444588,
-      "grad_norm": 1.4958491325378418,
-      "learning_rate": 9.411959798994975e-05,
-      "loss": 5.5006,
-      "step": 6353
-    },
-    {
-      "epoch": 3.3136897001303782,
-      "grad_norm": 1.4827463626861572,
-      "learning_rate": 9.411859296482413e-05,
-      "loss": 5.3885,
-      "step": 6354
-    },
-    {
-      "epoch": 3.3142112125162972,
-      "grad_norm": 1.5755757093429565,
-      "learning_rate": 9.41175879396985e-05,
-      "loss": 5.8804,
-      "step": 6355
-    },
-    {
-      "epoch": 3.3147327249022163,
-      "grad_norm": 1.6555750370025635,
-      "learning_rate": 9.411658291457287e-05,
-      "loss": 6.029,
-      "step": 6356
-    },
-    {
-      "epoch": 3.3152542372881357,
-      "grad_norm": 1.5130627155303955,
-      "learning_rate": 9.411557788944724e-05,
-      "loss": 6.1876,
-      "step": 6357
-    },
-    {
-      "epoch": 3.3157757496740548,
-      "grad_norm": 1.5127336978912354,
-      "learning_rate": 9.411457286432161e-05,
-      "loss": 5.6832,
-      "step": 6358
-    },
-    {
-      "epoch": 3.3162972620599738,
-      "grad_norm": 1.5517032146453857,
-      "learning_rate": 9.411356783919598e-05,
-      "loss": 5.6401,
-      "step": 6359
-    },
-    {
-      "epoch": 3.3168187744458932,
-      "grad_norm": 1.5533636808395386,
-      "learning_rate": 9.411256281407036e-05,
-      "loss": 5.794,
-      "step": 6360
-    },
-    {
-      "epoch": 3.3173402868318123,
-      "grad_norm": 1.5055230855941772,
-      "learning_rate": 9.411155778894472e-05,
-      "loss": 5.4846,
-      "step": 6361
-    },
-    {
-      "epoch": 3.3178617992177313,
-      "grad_norm": 1.5254874229431152,
-      "learning_rate": 9.41105527638191e-05,
-      "loss": 5.553,
-      "step": 6362
-    },
-    {
-      "epoch": 3.3183833116036507,
-      "grad_norm": 1.4541101455688477,
-      "learning_rate": 9.410954773869348e-05,
-      "loss": 5.512,
-      "step": 6363
-    },
-    {
-      "epoch": 3.3189048239895698,
-      "grad_norm": 1.4146724939346313,
-      "learning_rate": 9.410854271356785e-05,
-      "loss": 5.8886,
-      "step": 6364
-    },
-    {
-      "epoch": 3.319426336375489,
-      "grad_norm": 1.3644574880599976,
-      "learning_rate": 9.410753768844222e-05,
-      "loss": 6.21,
-      "step": 6365
-    },
-    {
-      "epoch": 3.3199478487614082,
-      "grad_norm": 1.454546570777893,
-      "learning_rate": 9.410653266331658e-05,
-      "loss": 6.128,
-      "step": 6366
-    },
-    {
-      "epoch": 3.3204693611473273,
-      "grad_norm": 1.4316980838775635,
-      "learning_rate": 9.410552763819096e-05,
-      "loss": 5.9424,
-      "step": 6367
-    },
-    {
-      "epoch": 3.3209908735332463,
-      "grad_norm": 1.4441211223602295,
-      "learning_rate": 9.410452261306532e-05,
-      "loss": 5.8854,
-      "step": 6368
-    },
-    {
-      "epoch": 3.3215123859191658,
-      "grad_norm": 1.5681601762771606,
-      "learning_rate": 9.41035175879397e-05,
-      "loss": 5.5148,
-      "step": 6369
-    },
-    {
-      "epoch": 3.3220338983050848,
-      "grad_norm": 1.6701209545135498,
-      "learning_rate": 9.410251256281407e-05,
-      "loss": 5.7655,
-      "step": 6370
-    },
-    {
-      "epoch": 3.322555410691004,
-      "grad_norm": 1.4533343315124512,
-      "learning_rate": 9.410150753768844e-05,
-      "loss": 6.0016,
-      "step": 6371
-    },
-    {
-      "epoch": 3.3230769230769233,
-      "grad_norm": 1.57914137840271,
-      "learning_rate": 9.410050251256282e-05,
-      "loss": 5.161,
-      "step": 6372
-    },
-    {
-      "epoch": 3.3235984354628423,
-      "grad_norm": 1.5068385601043701,
-      "learning_rate": 9.40994974874372e-05,
-      "loss": 5.0814,
-      "step": 6373
-    },
-    {
-      "epoch": 3.3241199478487613,
-      "grad_norm": 1.561515212059021,
-      "learning_rate": 9.409849246231156e-05,
-      "loss": 5.7566,
-      "step": 6374
-    },
-    {
-      "epoch": 3.3246414602346808,
-      "grad_norm": 1.3420336246490479,
-      "learning_rate": 9.409748743718594e-05,
-      "loss": 6.1102,
-      "step": 6375
-    },
-    {
-      "epoch": 3.3251629726206,
-      "grad_norm": 1.457801342010498,
-      "learning_rate": 9.40964824120603e-05,
-      "loss": 5.6199,
-      "step": 6376
-    },
-    {
-      "epoch": 3.325684485006519,
-      "grad_norm": 1.455648422241211,
-      "learning_rate": 9.409547738693468e-05,
-      "loss": 5.6653,
-      "step": 6377
-    },
-    {
-      "epoch": 3.326205997392438,
-      "grad_norm": 1.5172191858291626,
-      "learning_rate": 9.409447236180905e-05,
-      "loss": 5.2972,
-      "step": 6378
-    },
-    {
-      "epoch": 3.3267275097783573,
-      "grad_norm": 1.5300291776657104,
-      "learning_rate": 9.409346733668341e-05,
-      "loss": 5.4847,
-      "step": 6379
-    },
-    {
-      "epoch": 3.3272490221642763,
-      "grad_norm": 1.5404859781265259,
-      "learning_rate": 9.409246231155779e-05,
-      "loss": 5.604,
-      "step": 6380
-    },
-    {
-      "epoch": 3.3277705345501953,
-      "grad_norm": 1.6615705490112305,
-      "learning_rate": 9.409145728643215e-05,
-      "loss": 5.5327,
-      "step": 6381
-    },
-    {
-      "epoch": 3.328292046936115,
-      "grad_norm": 1.4334287643432617,
-      "learning_rate": 9.409045226130653e-05,
-      "loss": 5.7275,
-      "step": 6382
-    },
-    {
-      "epoch": 3.328813559322034,
-      "grad_norm": 1.4742642641067505,
-      "learning_rate": 9.408944723618091e-05,
-      "loss": 5.8876,
-      "step": 6383
-    },
-    {
-      "epoch": 3.329335071707953,
-      "grad_norm": 1.4157058000564575,
-      "learning_rate": 9.408844221105529e-05,
-      "loss": 5.9304,
-      "step": 6384
-    },
-    {
-      "epoch": 3.3298565840938723,
-      "grad_norm": 1.482864260673523,
-      "learning_rate": 9.408743718592965e-05,
-      "loss": 5.686,
-      "step": 6385
-    },
-    {
-      "epoch": 3.3303780964797913,
-      "grad_norm": 1.4490660429000854,
-      "learning_rate": 9.408643216080403e-05,
-      "loss": 5.7379,
-      "step": 6386
-    },
-    {
-      "epoch": 3.3308996088657103,
-      "grad_norm": 1.3994752168655396,
-      "learning_rate": 9.40854271356784e-05,
-      "loss": 6.064,
-      "step": 6387
-    },
-    {
-      "epoch": 3.33142112125163,
-      "grad_norm": 1.5044711828231812,
-      "learning_rate": 9.408442211055277e-05,
-      "loss": 5.9965,
-      "step": 6388
-    },
-    {
-      "epoch": 3.331942633637549,
-      "grad_norm": 1.583184003829956,
-      "learning_rate": 9.408341708542714e-05,
-      "loss": 5.3363,
-      "step": 6389
-    },
-    {
-      "epoch": 3.332464146023468,
-      "grad_norm": 1.439503788948059,
-      "learning_rate": 9.408241206030151e-05,
-      "loss": 5.6403,
-      "step": 6390
-    },
-    {
-      "epoch": 3.3329856584093873,
-      "grad_norm": 1.473559856414795,
-      "learning_rate": 9.408140703517588e-05,
-      "loss": 5.6616,
-      "step": 6391
-    },
-    {
-      "epoch": 3.3335071707953063,
-      "grad_norm": 1.3508695363998413,
-      "learning_rate": 9.408040201005026e-05,
-      "loss": 6.0295,
-      "step": 6392
-    },
-    {
-      "epoch": 3.3340286831812254,
-      "grad_norm": 1.5003570318222046,
-      "learning_rate": 9.407939698492463e-05,
-      "loss": 5.7976,
-      "step": 6393
-    },
-    {
-      "epoch": 3.334550195567145,
-      "grad_norm": 1.38614821434021,
-      "learning_rate": 9.4078391959799e-05,
-      "loss": 5.8254,
-      "step": 6394
-    },
-    {
-      "epoch": 3.335071707953064,
-      "grad_norm": 1.4649542570114136,
-      "learning_rate": 9.407738693467338e-05,
-      "loss": 5.6401,
-      "step": 6395
-    },
-    {
-      "epoch": 3.335593220338983,
-      "grad_norm": 1.518234372138977,
-      "learning_rate": 9.407638190954774e-05,
-      "loss": 5.7988,
-      "step": 6396
-    },
-    {
-      "epoch": 3.3361147327249023,
-      "grad_norm": 1.447389841079712,
-      "learning_rate": 9.407537688442212e-05,
-      "loss": 5.836,
-      "step": 6397
-    },
-    {
-      "epoch": 3.3366362451108214,
-      "grad_norm": 1.6171414852142334,
-      "learning_rate": 9.407437185929648e-05,
-      "loss": 5.657,
-      "step": 6398
-    },
-    {
-      "epoch": 3.3371577574967404,
-      "grad_norm": 1.524062156677246,
-      "learning_rate": 9.407336683417086e-05,
-      "loss": 5.655,
-      "step": 6399
-    },
-    {
-      "epoch": 3.33767926988266,
-      "grad_norm": 1.4842967987060547,
-      "learning_rate": 9.407236180904522e-05,
-      "loss": 5.5815,
-      "step": 6400
-    },
-    {
-      "epoch": 3.338200782268579,
-      "grad_norm": 1.468058705329895,
-      "learning_rate": 9.40713567839196e-05,
-      "loss": 5.9616,
-      "step": 6401
-    },
-    {
-      "epoch": 3.338722294654498,
-      "grad_norm": 1.4441412687301636,
-      "learning_rate": 9.407035175879397e-05,
-      "loss": 5.9398,
-      "step": 6402
-    },
-    {
-      "epoch": 3.3392438070404173,
-      "grad_norm": 1.589705467224121,
-      "learning_rate": 9.406934673366834e-05,
-      "loss": 5.4792,
-      "step": 6403
-    },
-    {
-      "epoch": 3.3397653194263364,
-      "grad_norm": 1.590893268585205,
-      "learning_rate": 9.406834170854272e-05,
-      "loss": 5.1545,
-      "step": 6404
-    },
-    {
-      "epoch": 3.3402868318122554,
-      "grad_norm": 1.415752649307251,
-      "learning_rate": 9.406733668341709e-05,
-      "loss": 5.5264,
-      "step": 6405
-    },
-    {
-      "epoch": 3.340808344198175,
-      "grad_norm": 1.5852222442626953,
-      "learning_rate": 9.406633165829146e-05,
-      "loss": 5.748,
-      "step": 6406
-    },
-    {
-      "epoch": 3.341329856584094,
-      "grad_norm": 1.490362524986267,
-      "learning_rate": 9.406532663316583e-05,
-      "loss": 5.8458,
-      "step": 6407
-    },
-    {
-      "epoch": 3.341851368970013,
-      "grad_norm": 1.492091178894043,
-      "learning_rate": 9.40643216080402e-05,
-      "loss": 5.3645,
-      "step": 6408
-    },
-    {
-      "epoch": 3.3423728813559324,
-      "grad_norm": 1.5137943029403687,
-      "learning_rate": 9.406331658291457e-05,
-      "loss": 5.9262,
-      "step": 6409
-    },
-    {
-      "epoch": 3.3428943937418514,
-      "grad_norm": 1.6387317180633545,
-      "learning_rate": 9.406231155778895e-05,
-      "loss": 5.7096,
-      "step": 6410
-    },
-    {
-      "epoch": 3.3434159061277704,
-      "grad_norm": 1.3785731792449951,
-      "learning_rate": 9.406130653266331e-05,
-      "loss": 5.525,
-      "step": 6411
-    },
-    {
-      "epoch": 3.34393741851369,
-      "grad_norm": 1.3889933824539185,
-      "learning_rate": 9.406030150753769e-05,
-      "loss": 5.8437,
-      "step": 6412
-    },
-    {
-      "epoch": 3.344458930899609,
-      "grad_norm": 1.3700517416000366,
-      "learning_rate": 9.405929648241207e-05,
-      "loss": 5.885,
-      "step": 6413
-    },
-    {
-      "epoch": 3.344980443285528,
-      "grad_norm": 1.4141086339950562,
-      "learning_rate": 9.405829145728645e-05,
-      "loss": 5.9811,
-      "step": 6414
-    },
-    {
-      "epoch": 3.3455019556714474,
-      "grad_norm": 1.4639480113983154,
-      "learning_rate": 9.405728643216081e-05,
-      "loss": 6.0593,
-      "step": 6415
-    },
-    {
-      "epoch": 3.3460234680573664,
-      "grad_norm": 1.5258677005767822,
-      "learning_rate": 9.405628140703519e-05,
-      "loss": 5.813,
-      "step": 6416
-    },
-    {
-      "epoch": 3.3465449804432854,
-      "grad_norm": 1.773355484008789,
-      "learning_rate": 9.405527638190955e-05,
-      "loss": 5.2558,
-      "step": 6417
-    },
-    {
-      "epoch": 3.347066492829205,
-      "grad_norm": 1.4043976068496704,
-      "learning_rate": 9.405427135678393e-05,
-      "loss": 6.0572,
-      "step": 6418
-    },
-    {
-      "epoch": 3.347588005215124,
-      "grad_norm": 1.3784087896347046,
-      "learning_rate": 9.40532663316583e-05,
-      "loss": 5.7787,
-      "step": 6419
-    },
-    {
-      "epoch": 3.348109517601043,
-      "grad_norm": 1.4332334995269775,
-      "learning_rate": 9.405226130653266e-05,
-      "loss": 5.7922,
-      "step": 6420
-    },
-    {
-      "epoch": 3.3486310299869624,
-      "grad_norm": 1.5401835441589355,
-      "learning_rate": 9.405125628140704e-05,
-      "loss": 5.1052,
-      "step": 6421
-    },
-    {
-      "epoch": 3.3491525423728814,
-      "grad_norm": 1.4492846727371216,
-      "learning_rate": 9.40502512562814e-05,
-      "loss": 5.6992,
-      "step": 6422
-    },
-    {
-      "epoch": 3.3496740547588004,
-      "grad_norm": 1.3719035387039185,
-      "learning_rate": 9.404924623115578e-05,
-      "loss": 5.9632,
-      "step": 6423
-    },
-    {
-      "epoch": 3.35019556714472,
-      "grad_norm": 1.5205992460250854,
-      "learning_rate": 9.404824120603016e-05,
-      "loss": 5.8245,
-      "step": 6424
-    },
-    {
-      "epoch": 3.350717079530639,
-      "grad_norm": 1.5881708860397339,
-      "learning_rate": 9.404723618090453e-05,
-      "loss": 5.6207,
-      "step": 6425
-    },
-    {
-      "epoch": 3.351238591916558,
-      "grad_norm": 1.5960251092910767,
-      "learning_rate": 9.40462311557789e-05,
-      "loss": 5.5833,
-      "step": 6426
-    },
-    {
-      "epoch": 3.3517601043024774,
-      "grad_norm": 1.4527392387390137,
-      "learning_rate": 9.404522613065328e-05,
-      "loss": 5.8865,
-      "step": 6427
-    },
-    {
-      "epoch": 3.3522816166883964,
-      "grad_norm": 1.6216917037963867,
-      "learning_rate": 9.404422110552764e-05,
-      "loss": 5.0898,
-      "step": 6428
-    },
-    {
-      "epoch": 3.3528031290743154,
-      "grad_norm": 1.5218192338943481,
-      "learning_rate": 9.404321608040202e-05,
-      "loss": 6.2305,
-      "step": 6429
-    },
-    {
-      "epoch": 3.353324641460235,
-      "grad_norm": 1.7974135875701904,
-      "learning_rate": 9.404221105527638e-05,
-      "loss": 5.3839,
-      "step": 6430
-    },
-    {
-      "epoch": 3.353846153846154,
-      "grad_norm": 1.4692200422286987,
-      "learning_rate": 9.404120603015076e-05,
-      "loss": 5.8771,
-      "step": 6431
-    },
-    {
-      "epoch": 3.354367666232073,
-      "grad_norm": 1.4745904207229614,
-      "learning_rate": 9.404020100502513e-05,
-      "loss": 5.6909,
-      "step": 6432
-    },
-    {
-      "epoch": 3.3548891786179924,
-      "grad_norm": 1.6213864088058472,
-      "learning_rate": 9.40391959798995e-05,
-      "loss": 5.9198,
-      "step": 6433
-    },
-    {
-      "epoch": 3.3554106910039114,
-      "grad_norm": 1.6370612382888794,
-      "learning_rate": 9.403819095477388e-05,
-      "loss": 5.6473,
-      "step": 6434
-    },
-    {
-      "epoch": 3.3559322033898304,
-      "grad_norm": 1.5853809118270874,
-      "learning_rate": 9.403718592964825e-05,
-      "loss": 5.4175,
-      "step": 6435
-    },
-    {
-      "epoch": 3.35645371577575,
-      "grad_norm": 1.519562840461731,
-      "learning_rate": 9.403618090452262e-05,
-      "loss": 5.8427,
-      "step": 6436
-    },
-    {
-      "epoch": 3.356975228161669,
-      "grad_norm": 1.610128402709961,
-      "learning_rate": 9.403517587939699e-05,
-      "loss": 5.8247,
-      "step": 6437
-    },
-    {
-      "epoch": 3.357496740547588,
-      "grad_norm": 1.520509123802185,
-      "learning_rate": 9.403417085427137e-05,
-      "loss": 5.4194,
-      "step": 6438
-    },
-    {
-      "epoch": 3.3580182529335074,
-      "grad_norm": 1.6838332414627075,
-      "learning_rate": 9.403316582914573e-05,
-      "loss": 5.5234,
-      "step": 6439
-    },
-    {
-      "epoch": 3.3585397653194264,
-      "grad_norm": 1.7237896919250488,
-      "learning_rate": 9.403216080402011e-05,
-      "loss": 5.2745,
-      "step": 6440
-    },
-    {
-      "epoch": 3.3590612777053455,
-      "grad_norm": 1.5761135816574097,
-      "learning_rate": 9.403115577889447e-05,
-      "loss": 5.6603,
-      "step": 6441
-    },
-    {
-      "epoch": 3.3595827900912645,
-      "grad_norm": 1.7206511497497559,
-      "learning_rate": 9.403015075376885e-05,
-      "loss": 5.4105,
-      "step": 6442
-    },
-    {
-      "epoch": 3.360104302477184,
-      "grad_norm": 1.7045494318008423,
-      "learning_rate": 9.402914572864321e-05,
-      "loss": 5.6755,
-      "step": 6443
-    },
-    {
-      "epoch": 3.360625814863103,
-      "grad_norm": 1.617280125617981,
-      "learning_rate": 9.402814070351759e-05,
-      "loss": 5.2217,
-      "step": 6444
-    },
-    {
-      "epoch": 3.361147327249022,
-      "grad_norm": 1.490423321723938,
-      "learning_rate": 9.402713567839197e-05,
-      "loss": 5.6791,
-      "step": 6445
-    },
-    {
-      "epoch": 3.3616688396349415,
-      "grad_norm": 1.5271817445755005,
-      "learning_rate": 9.402613065326633e-05,
-      "loss": 5.9129,
-      "step": 6446
-    },
-    {
-      "epoch": 3.3621903520208605,
-      "grad_norm": 1.3382233381271362,
-      "learning_rate": 9.402512562814071e-05,
-      "loss": 5.9247,
-      "step": 6447
-    },
-    {
-      "epoch": 3.3627118644067795,
-      "grad_norm": 1.5612832307815552,
-      "learning_rate": 9.402412060301508e-05,
-      "loss": 5.3969,
-      "step": 6448
-    },
-    {
-      "epoch": 3.363233376792699,
-      "grad_norm": 1.5798447132110596,
-      "learning_rate": 9.402311557788945e-05,
-      "loss": 5.9233,
-      "step": 6449
-    },
-    {
-      "epoch": 3.363754889178618,
-      "grad_norm": 1.377052664756775,
-      "learning_rate": 9.402211055276382e-05,
-      "loss": 6.0737,
-      "step": 6450
-    },
-    {
-      "epoch": 3.364276401564537,
-      "grad_norm": 1.3942419290542603,
-      "learning_rate": 9.40211055276382e-05,
-      "loss": 6.023,
-      "step": 6451
-    },
-    {
-      "epoch": 3.3647979139504565,
-      "grad_norm": 1.417812705039978,
-      "learning_rate": 9.402010050251256e-05,
-      "loss": 6.1282,
-      "step": 6452
-    },
-    {
-      "epoch": 3.3653194263363755,
-      "grad_norm": 1.3771721124649048,
-      "learning_rate": 9.401909547738694e-05,
-      "loss": 5.6269,
-      "step": 6453
-    },
-    {
-      "epoch": 3.3658409387222945,
-      "grad_norm": 1.456269383430481,
-      "learning_rate": 9.401809045226132e-05,
-      "loss": 6.0227,
-      "step": 6454
-    },
-    {
-      "epoch": 3.366362451108214,
-      "grad_norm": 1.4427556991577148,
-      "learning_rate": 9.40170854271357e-05,
-      "loss": 5.5068,
-      "step": 6455
-    },
-    {
-      "epoch": 3.366883963494133,
-      "grad_norm": 1.510710597038269,
-      "learning_rate": 9.401608040201006e-05,
-      "loss": 5.8937,
-      "step": 6456
-    },
-    {
-      "epoch": 3.367405475880052,
-      "grad_norm": 1.6357951164245605,
-      "learning_rate": 9.401507537688444e-05,
-      "loss": 5.8729,
-      "step": 6457
-    },
-    {
-      "epoch": 3.3679269882659715,
-      "grad_norm": 1.6245770454406738,
-      "learning_rate": 9.40140703517588e-05,
-      "loss": 5.7556,
-      "step": 6458
-    },
-    {
-      "epoch": 3.3684485006518905,
-      "grad_norm": 1.7717359066009521,
-      "learning_rate": 9.401306532663316e-05,
-      "loss": 5.8173,
-      "step": 6459
-    },
-    {
-      "epoch": 3.3689700130378095,
-      "grad_norm": 1.8022582530975342,
-      "learning_rate": 9.401206030150754e-05,
-      "loss": 4.9304,
-      "step": 6460
-    },
-    {
-      "epoch": 3.369491525423729,
-      "grad_norm": 1.3410917520523071,
-      "learning_rate": 9.40110552763819e-05,
-      "loss": 5.3652,
-      "step": 6461
-    },
-    {
-      "epoch": 3.370013037809648,
-      "grad_norm": 1.4655345678329468,
-      "learning_rate": 9.401005025125628e-05,
-      "loss": 6.0267,
-      "step": 6462
-    },
-    {
-      "epoch": 3.370534550195567,
-      "grad_norm": 1.4135409593582153,
-      "learning_rate": 9.400904522613065e-05,
-      "loss": 5.74,
-      "step": 6463
-    },
-    {
-      "epoch": 3.3710560625814865,
-      "grad_norm": 1.4556366205215454,
-      "learning_rate": 9.400804020100503e-05,
-      "loss": 5.7254,
-      "step": 6464
-    },
-    {
-      "epoch": 3.3715775749674055,
-      "grad_norm": 1.4365060329437256,
-      "learning_rate": 9.40070351758794e-05,
-      "loss": 5.8429,
-      "step": 6465
-    },
-    {
-      "epoch": 3.3720990873533245,
-      "grad_norm": 1.479904055595398,
-      "learning_rate": 9.400603015075378e-05,
-      "loss": 5.8722,
-      "step": 6466
-    },
-    {
-      "epoch": 3.372620599739244,
-      "grad_norm": 1.4162895679473877,
-      "learning_rate": 9.400502512562815e-05,
-      "loss": 6.2072,
-      "step": 6467
-    },
-    {
-      "epoch": 3.373142112125163,
-      "grad_norm": 1.381284236907959,
-      "learning_rate": 9.400402010050252e-05,
-      "loss": 5.2954,
-      "step": 6468
-    },
-    {
-      "epoch": 3.373663624511082,
-      "grad_norm": 1.4764337539672852,
-      "learning_rate": 9.400301507537689e-05,
-      "loss": 5.9313,
-      "step": 6469
-    },
-    {
-      "epoch": 3.374185136897001,
-      "grad_norm": 1.445488691329956,
-      "learning_rate": 9.400201005025127e-05,
-      "loss": 5.6215,
-      "step": 6470
-    },
-    {
-      "epoch": 3.3747066492829205,
-      "grad_norm": 1.3670355081558228,
-      "learning_rate": 9.400100502512563e-05,
-      "loss": 5.8369,
-      "step": 6471
-    },
-    {
-      "epoch": 3.3752281616688395,
-      "grad_norm": 1.4261252880096436,
-      "learning_rate": 9.4e-05,
-      "loss": 6.0461,
-      "step": 6472
-    },
-    {
-      "epoch": 3.3757496740547586,
-      "grad_norm": 1.5171678066253662,
-      "learning_rate": 9.399899497487437e-05,
-      "loss": 5.4054,
-      "step": 6473
-    },
-    {
-      "epoch": 3.376271186440678,
-      "grad_norm": 1.752041220664978,
-      "learning_rate": 9.399798994974875e-05,
-      "loss": 5.5749,
-      "step": 6474
-    },
-    {
-      "epoch": 3.376792698826597,
-      "grad_norm": 1.5243462324142456,
-      "learning_rate": 9.399698492462313e-05,
-      "loss": 5.3607,
-      "step": 6475
-    },
-    {
-      "epoch": 3.377314211212516,
-      "grad_norm": 1.493709921836853,
-      "learning_rate": 9.399597989949749e-05,
-      "loss": 5.7654,
-      "step": 6476
-    },
-    {
-      "epoch": 3.3778357235984355,
-      "grad_norm": 1.4140244722366333,
-      "learning_rate": 9.399497487437187e-05,
-      "loss": 5.4763,
-      "step": 6477
-    },
-    {
-      "epoch": 3.3783572359843546,
-      "grad_norm": 1.474185824394226,
-      "learning_rate": 9.399396984924623e-05,
-      "loss": 5.7317,
-      "step": 6478
-    },
-    {
-      "epoch": 3.3788787483702736,
-      "grad_norm": 1.5064306259155273,
-      "learning_rate": 9.399296482412061e-05,
-      "loss": 5.9905,
-      "step": 6479
-    },
-    {
-      "epoch": 3.379400260756193,
-      "grad_norm": 1.48843514919281,
-      "learning_rate": 9.399195979899498e-05,
-      "loss": 5.7787,
-      "step": 6480
-    },
-    {
-      "epoch": 3.379921773142112,
-      "grad_norm": 1.448557734489441,
-      "learning_rate": 9.399095477386935e-05,
-      "loss": 5.6171,
-      "step": 6481
-    },
-    {
-      "epoch": 3.380443285528031,
-      "grad_norm": 1.4944267272949219,
-      "learning_rate": 9.398994974874372e-05,
-      "loss": 5.8917,
-      "step": 6482
-    },
-    {
-      "epoch": 3.3809647979139505,
-      "grad_norm": 1.4344645738601685,
-      "learning_rate": 9.39889447236181e-05,
-      "loss": 5.8834,
-      "step": 6483
-    },
-    {
-      "epoch": 3.3814863102998696,
-      "grad_norm": 1.4446418285369873,
-      "learning_rate": 9.398793969849246e-05,
-      "loss": 5.7758,
-      "step": 6484
-    },
-    {
-      "epoch": 3.3820078226857886,
-      "grad_norm": 1.5224469900131226,
-      "learning_rate": 9.398693467336684e-05,
-      "loss": 6.1677,
-      "step": 6485
-    },
-    {
-      "epoch": 3.382529335071708,
-      "grad_norm": 1.6215016841888428,
-      "learning_rate": 9.398592964824122e-05,
-      "loss": 5.537,
-      "step": 6486
-    },
-    {
-      "epoch": 3.383050847457627,
-      "grad_norm": 1.5413540601730347,
-      "learning_rate": 9.398492462311558e-05,
-      "loss": 5.9245,
-      "step": 6487
-    },
-    {
-      "epoch": 3.383572359843546,
-      "grad_norm": 1.5161811113357544,
-      "learning_rate": 9.398391959798996e-05,
-      "loss": 5.6298,
-      "step": 6488
-    },
-    {
-      "epoch": 3.3840938722294656,
-      "grad_norm": 1.4435259103775024,
-      "learning_rate": 9.398291457286432e-05,
-      "loss": 6.1302,
-      "step": 6489
-    },
-    {
-      "epoch": 3.3846153846153846,
-      "grad_norm": 1.6425511837005615,
-      "learning_rate": 9.39819095477387e-05,
-      "loss": 5.3557,
-      "step": 6490
-    },
-    {
-      "epoch": 3.3851368970013036,
-      "grad_norm": 1.4682636260986328,
-      "learning_rate": 9.398090452261306e-05,
-      "loss": 5.6554,
-      "step": 6491
-    },
-    {
-      "epoch": 3.385658409387223,
-      "grad_norm": 1.4465373754501343,
-      "learning_rate": 9.397989949748744e-05,
-      "loss": 5.4763,
-      "step": 6492
-    },
-    {
-      "epoch": 3.386179921773142,
-      "grad_norm": 1.519806146621704,
-      "learning_rate": 9.397889447236181e-05,
-      "loss": 5.9476,
-      "step": 6493
-    },
-    {
-      "epoch": 3.386701434159061,
-      "grad_norm": 1.5260028839111328,
-      "learning_rate": 9.397788944723618e-05,
-      "loss": 5.5217,
-      "step": 6494
-    },
-    {
-      "epoch": 3.3872229465449806,
-      "grad_norm": 1.82406747341156,
-      "learning_rate": 9.397688442211056e-05,
-      "loss": 4.9978,
-      "step": 6495
-    },
-    {
-      "epoch": 3.3877444589308996,
-      "grad_norm": 1.5789798498153687,
-      "learning_rate": 9.397587939698494e-05,
-      "loss": 5.2265,
-      "step": 6496
-    },
-    {
-      "epoch": 3.3882659713168186,
-      "grad_norm": 1.605324625968933,
-      "learning_rate": 9.39748743718593e-05,
-      "loss": 5.3143,
-      "step": 6497
-    },
-    {
-      "epoch": 3.388787483702738,
-      "grad_norm": 1.4530377388000488,
-      "learning_rate": 9.397386934673368e-05,
-      "loss": 6.0444,
-      "step": 6498
-    },
-    {
-      "epoch": 3.389308996088657,
-      "grad_norm": 1.5502046346664429,
-      "learning_rate": 9.397286432160805e-05,
-      "loss": 5.7193,
-      "step": 6499
-    },
-    {
-      "epoch": 3.389830508474576,
-      "grad_norm": 1.5165114402770996,
-      "learning_rate": 9.397185929648241e-05,
-      "loss": 5.3546,
-      "step": 6500
-    },
-    {
-      "epoch": 3.3903520208604956,
-      "grad_norm": 1.553282380104065,
-      "learning_rate": 9.397085427135679e-05,
-      "loss": 6.0829,
-      "step": 6501
-    },
-    {
-      "epoch": 3.3908735332464146,
-      "grad_norm": 1.8076362609863281,
-      "learning_rate": 9.396984924623115e-05,
-      "loss": 5.6219,
-      "step": 6502
-    },
-    {
-      "epoch": 3.3913950456323336,
-      "grad_norm": 1.6001783609390259,
-      "learning_rate": 9.396884422110553e-05,
-      "loss": 5.7384,
-      "step": 6503
-    },
-    {
-      "epoch": 3.391916558018253,
-      "grad_norm": 1.4761345386505127,
-      "learning_rate": 9.39678391959799e-05,
-      "loss": 5.4711,
-      "step": 6504
-    },
-    {
-      "epoch": 3.392438070404172,
-      "grad_norm": 1.4391748905181885,
-      "learning_rate": 9.396683417085427e-05,
-      "loss": 5.7536,
-      "step": 6505
-    },
-    {
-      "epoch": 3.392959582790091,
-      "grad_norm": 1.533321738243103,
-      "learning_rate": 9.396582914572865e-05,
-      "loss": 5.9117,
-      "step": 6506
-    },
-    {
-      "epoch": 3.3934810951760106,
-      "grad_norm": 1.4328850507736206,
-      "learning_rate": 9.396482412060303e-05,
-      "loss": 5.7987,
-      "step": 6507
-    },
-    {
-      "epoch": 3.3940026075619296,
-      "grad_norm": 1.489643931388855,
-      "learning_rate": 9.396381909547739e-05,
-      "loss": 6.01,
-      "step": 6508
-    },
-    {
-      "epoch": 3.3945241199478486,
-      "grad_norm": 1.563175916671753,
-      "learning_rate": 9.396281407035177e-05,
-      "loss": 5.6983,
-      "step": 6509
-    },
-    {
-      "epoch": 3.395045632333768,
-      "grad_norm": 1.548081636428833,
-      "learning_rate": 9.396180904522614e-05,
-      "loss": 5.7492,
-      "step": 6510
-    },
-    {
-      "epoch": 3.395567144719687,
-      "grad_norm": 1.541178584098816,
-      "learning_rate": 9.396080402010051e-05,
-      "loss": 5.8637,
-      "step": 6511
-    },
-    {
-      "epoch": 3.396088657105606,
-      "grad_norm": 1.621647834777832,
-      "learning_rate": 9.395979899497488e-05,
-      "loss": 5.3276,
-      "step": 6512
-    },
-    {
-      "epoch": 3.3966101694915256,
-      "grad_norm": 1.3239518404006958,
-      "learning_rate": 9.395879396984924e-05,
-      "loss": 6.1462,
-      "step": 6513
-    },
-    {
-      "epoch": 3.3971316818774446,
-      "grad_norm": 1.3908641338348389,
-      "learning_rate": 9.395778894472362e-05,
-      "loss": 5.8648,
-      "step": 6514
-    },
-    {
-      "epoch": 3.3976531942633637,
-      "grad_norm": 1.498016595840454,
-      "learning_rate": 9.3956783919598e-05,
-      "loss": 5.8555,
-      "step": 6515
-    },
-    {
-      "epoch": 3.398174706649283,
-      "grad_norm": 1.5761386156082153,
-      "learning_rate": 9.395577889447237e-05,
-      "loss": 6.0078,
-      "step": 6516
-    },
-    {
-      "epoch": 3.398696219035202,
-      "grad_norm": 1.7212343215942383,
-      "learning_rate": 9.395477386934674e-05,
-      "loss": 5.2859,
-      "step": 6517
-    },
-    {
-      "epoch": 3.399217731421121,
-      "grad_norm": 1.5033856630325317,
-      "learning_rate": 9.395376884422112e-05,
-      "loss": 5.7599,
-      "step": 6518
-    },
-    {
-      "epoch": 3.3997392438070406,
-      "grad_norm": 1.6325711011886597,
-      "learning_rate": 9.395276381909548e-05,
-      "loss": 5.7241,
-      "step": 6519
-    },
-    {
-      "epoch": 3.4002607561929596,
-      "grad_norm": 1.7897790670394897,
-      "learning_rate": 9.395175879396986e-05,
-      "loss": 5.9737,
-      "step": 6520
-    },
-    {
-      "epoch": 3.4007822685788787,
-      "grad_norm": 1.4545655250549316,
-      "learning_rate": 9.395075376884422e-05,
-      "loss": 5.4988,
-      "step": 6521
-    },
-    {
-      "epoch": 3.401303780964798,
-      "grad_norm": 1.5968705415725708,
-      "learning_rate": 9.39497487437186e-05,
-      "loss": 5.6458,
-      "step": 6522
-    },
-    {
-      "epoch": 3.401825293350717,
-      "grad_norm": 1.4853588342666626,
-      "learning_rate": 9.394874371859297e-05,
-      "loss": 6.0466,
-      "step": 6523
-    },
-    {
-      "epoch": 3.402346805736636,
-      "grad_norm": 1.5399503707885742,
-      "learning_rate": 9.394773869346734e-05,
-      "loss": 5.9619,
-      "step": 6524
-    },
-    {
-      "epoch": 3.4028683181225556,
-      "grad_norm": 1.882503628730774,
-      "learning_rate": 9.394673366834171e-05,
-      "loss": 5.6956,
-      "step": 6525
-    },
-    {
-      "epoch": 3.4033898305084747,
-      "grad_norm": 1.6144382953643799,
-      "learning_rate": 9.394572864321609e-05,
-      "loss": 5.5976,
-      "step": 6526
-    },
-    {
-      "epoch": 3.4039113428943937,
-      "grad_norm": 1.5639595985412598,
-      "learning_rate": 9.394472361809046e-05,
-      "loss": 5.8781,
-      "step": 6527
-    },
-    {
-      "epoch": 3.404432855280313,
-      "grad_norm": 1.5439175367355347,
-      "learning_rate": 9.394371859296483e-05,
-      "loss": 5.8828,
-      "step": 6528
-    },
-    {
-      "epoch": 3.404954367666232,
-      "grad_norm": 1.5717159509658813,
-      "learning_rate": 9.39427135678392e-05,
-      "loss": 5.7316,
-      "step": 6529
-    },
-    {
-      "epoch": 3.405475880052151,
-      "grad_norm": 1.4159138202667236,
-      "learning_rate": 9.394170854271357e-05,
-      "loss": 6.0799,
-      "step": 6530
-    },
-    {
-      "epoch": 3.4059973924380706,
-      "grad_norm": 1.426571011543274,
-      "learning_rate": 9.394070351758795e-05,
-      "loss": 6.1232,
-      "step": 6531
-    },
-    {
-      "epoch": 3.4065189048239897,
-      "grad_norm": 1.424534797668457,
-      "learning_rate": 9.393969849246231e-05,
-      "loss": 5.9758,
-      "step": 6532
-    },
-    {
-      "epoch": 3.4070404172099087,
-      "grad_norm": 1.558273434638977,
-      "learning_rate": 9.393869346733669e-05,
-      "loss": 5.9933,
-      "step": 6533
-    },
-    {
-      "epoch": 3.4075619295958277,
-      "grad_norm": 1.4258747100830078,
-      "learning_rate": 9.393768844221105e-05,
-      "loss": 6.088,
-      "step": 6534
-    },
-    {
-      "epoch": 3.408083441981747,
-      "grad_norm": 1.4185106754302979,
-      "learning_rate": 9.393668341708543e-05,
-      "loss": 5.5963,
-      "step": 6535
-    },
-    {
-      "epoch": 3.408604954367666,
-      "grad_norm": 1.3699321746826172,
-      "learning_rate": 9.39356783919598e-05,
-      "loss": 5.5412,
-      "step": 6536
-    },
-    {
-      "epoch": 3.409126466753585,
-      "grad_norm": 1.414751648902893,
-      "learning_rate": 9.393467336683417e-05,
-      "loss": 5.8081,
-      "step": 6537
-    },
-    {
-      "epoch": 3.4096479791395047,
-      "grad_norm": 1.5143978595733643,
-      "learning_rate": 9.393366834170855e-05,
-      "loss": 5.9845,
-      "step": 6538
-    },
-    {
-      "epoch": 3.4101694915254237,
-      "grad_norm": 1.5359721183776855,
-      "learning_rate": 9.393266331658292e-05,
-      "loss": 5.924,
-      "step": 6539
-    },
-    {
-      "epoch": 3.4106910039113427,
-      "grad_norm": 1.9311178922653198,
-      "learning_rate": 9.39316582914573e-05,
-      "loss": 4.8777,
-      "step": 6540
-    },
-    {
-      "epoch": 3.411212516297262,
-      "grad_norm": 1.5462849140167236,
-      "learning_rate": 9.393065326633166e-05,
-      "loss": 5.5404,
-      "step": 6541
-    },
-    {
-      "epoch": 3.411734028683181,
-      "grad_norm": 1.6157095432281494,
-      "learning_rate": 9.392964824120604e-05,
-      "loss": 5.6673,
-      "step": 6542
-    },
-    {
-      "epoch": 3.4122555410691002,
-      "grad_norm": 1.5912237167358398,
-      "learning_rate": 9.39286432160804e-05,
-      "loss": 5.4115,
-      "step": 6543
-    },
-    {
-      "epoch": 3.4127770534550197,
-      "grad_norm": 1.451368808746338,
-      "learning_rate": 9.392763819095478e-05,
-      "loss": 5.9312,
-      "step": 6544
-    },
-    {
-      "epoch": 3.4132985658409387,
-      "grad_norm": 1.5446959733963013,
-      "learning_rate": 9.392663316582914e-05,
-      "loss": 5.761,
-      "step": 6545
-    },
-    {
-      "epoch": 3.4138200782268577,
-      "grad_norm": 1.6865609884262085,
-      "learning_rate": 9.392562814070352e-05,
-      "loss": 5.8478,
-      "step": 6546
-    },
-    {
-      "epoch": 3.414341590612777,
-      "grad_norm": 1.4885061979293823,
-      "learning_rate": 9.39246231155779e-05,
-      "loss": 6.0549,
-      "step": 6547
-    },
-    {
-      "epoch": 3.414863102998696,
-      "grad_norm": 1.610979437828064,
-      "learning_rate": 9.392361809045228e-05,
-      "loss": 5.7821,
-      "step": 6548
-    },
-    {
-      "epoch": 3.4153846153846152,
-      "grad_norm": 1.5412813425064087,
-      "learning_rate": 9.392261306532664e-05,
-      "loss": 5.5426,
-      "step": 6549
-    },
-    {
-      "epoch": 3.4159061277705347,
-      "grad_norm": 1.6531035900115967,
-      "learning_rate": 9.392160804020102e-05,
-      "loss": 5.5665,
-      "step": 6550
-    },
-    {
-      "epoch": 3.4164276401564537,
-      "grad_norm": 1.7348772287368774,
-      "learning_rate": 9.392060301507538e-05,
-      "loss": 5.4497,
-      "step": 6551
-    },
-    {
-      "epoch": 3.4169491525423727,
-      "grad_norm": 1.4983173608779907,
-      "learning_rate": 9.391959798994975e-05,
-      "loss": 5.9426,
-      "step": 6552
-    },
-    {
-      "epoch": 3.417470664928292,
-      "grad_norm": 1.392858862876892,
-      "learning_rate": 9.391859296482412e-05,
-      "loss": 5.7955,
-      "step": 6553
-    },
-    {
-      "epoch": 3.4179921773142112,
-      "grad_norm": 1.4265427589416504,
-      "learning_rate": 9.391758793969849e-05,
-      "loss": 5.5697,
-      "step": 6554
-    },
-    {
-      "epoch": 3.4185136897001303,
-      "grad_norm": 1.582919955253601,
-      "learning_rate": 9.391658291457287e-05,
-      "loss": 5.4613,
-      "step": 6555
-    },
-    {
-      "epoch": 3.4190352020860497,
-      "grad_norm": 1.8276000022888184,
-      "learning_rate": 9.391557788944723e-05,
-      "loss": 4.9726,
-      "step": 6556
-    },
-    {
-      "epoch": 3.4195567144719687,
-      "grad_norm": 1.6476658582687378,
-      "learning_rate": 9.391457286432161e-05,
-      "loss": 5.5401,
-      "step": 6557
-    },
-    {
-      "epoch": 3.4200782268578878,
-      "grad_norm": 1.3307180404663086,
-      "learning_rate": 9.391356783919599e-05,
-      "loss": 6.0084,
-      "step": 6558
-    },
-    {
-      "epoch": 3.4205997392438072,
-      "grad_norm": 1.3770465850830078,
-      "learning_rate": 9.391256281407036e-05,
-      "loss": 5.7313,
-      "step": 6559
-    },
-    {
-      "epoch": 3.4211212516297262,
-      "grad_norm": 1.4043294191360474,
-      "learning_rate": 9.391155778894473e-05,
-      "loss": 6.1037,
-      "step": 6560
-    },
-    {
-      "epoch": 3.4216427640156453,
-      "grad_norm": 1.3740513324737549,
-      "learning_rate": 9.39105527638191e-05,
-      "loss": 5.842,
-      "step": 6561
-    },
-    {
-      "epoch": 3.4221642764015647,
-      "grad_norm": 1.4339752197265625,
-      "learning_rate": 9.390954773869347e-05,
-      "loss": 5.5815,
-      "step": 6562
-    },
-    {
-      "epoch": 3.4226857887874838,
-      "grad_norm": 1.508982539176941,
-      "learning_rate": 9.390854271356785e-05,
-      "loss": 5.7548,
-      "step": 6563
-    },
-    {
-      "epoch": 3.4232073011734028,
-      "grad_norm": 1.3650470972061157,
-      "learning_rate": 9.390753768844221e-05,
-      "loss": 5.8276,
-      "step": 6564
-    },
-    {
-      "epoch": 3.423728813559322,
-      "grad_norm": 1.4773926734924316,
-      "learning_rate": 9.390653266331658e-05,
-      "loss": 5.7884,
-      "step": 6565
-    },
-    {
-      "epoch": 3.4242503259452413,
-      "grad_norm": 1.4572947025299072,
-      "learning_rate": 9.390552763819095e-05,
-      "loss": 5.658,
-      "step": 6566
-    },
-    {
-      "epoch": 3.4247718383311603,
-      "grad_norm": 1.6104774475097656,
-      "learning_rate": 9.390452261306533e-05,
-      "loss": 5.6263,
-      "step": 6567
-    },
-    {
-      "epoch": 3.4252933507170793,
-      "grad_norm": 1.599902629852295,
-      "learning_rate": 9.390351758793971e-05,
-      "loss": 5.3852,
-      "step": 6568
-    },
-    {
-      "epoch": 3.4258148631029988,
-      "grad_norm": 1.4869414567947388,
-      "learning_rate": 9.390251256281407e-05,
-      "loss": 5.8993,
-      "step": 6569
-    },
-    {
-      "epoch": 3.426336375488918,
-      "grad_norm": 1.6254209280014038,
-      "learning_rate": 9.390150753768845e-05,
-      "loss": 5.5851,
-      "step": 6570
-    },
-    {
-      "epoch": 3.426857887874837,
-      "grad_norm": 1.516729712486267,
-      "learning_rate": 9.390050251256282e-05,
-      "loss": 5.9488,
-      "step": 6571
-    },
-    {
-      "epoch": 3.4273794002607563,
-      "grad_norm": 1.5659940242767334,
-      "learning_rate": 9.38994974874372e-05,
-      "loss": 5.7598,
-      "step": 6572
-    },
-    {
-      "epoch": 3.4279009126466753,
-      "grad_norm": 1.5914667844772339,
-      "learning_rate": 9.389849246231156e-05,
-      "loss": 5.8017,
-      "step": 6573
-    },
-    {
-      "epoch": 3.4284224250325943,
-      "grad_norm": 1.6391860246658325,
-      "learning_rate": 9.389748743718594e-05,
-      "loss": 5.5948,
-      "step": 6574
-    },
-    {
-      "epoch": 3.4289439374185138,
-      "grad_norm": 1.673113226890564,
-      "learning_rate": 9.38964824120603e-05,
-      "loss": 5.4251,
-      "step": 6575
-    },
-    {
-      "epoch": 3.429465449804433,
-      "grad_norm": 1.6293518543243408,
-      "learning_rate": 9.389547738693468e-05,
-      "loss": 5.9745,
-      "step": 6576
-    },
-    {
-      "epoch": 3.429986962190352,
-      "grad_norm": 1.4809448719024658,
-      "learning_rate": 9.389447236180904e-05,
-      "loss": 6.0293,
-      "step": 6577
-    },
-    {
-      "epoch": 3.4305084745762713,
-      "grad_norm": 1.5601394176483154,
-      "learning_rate": 9.389346733668342e-05,
-      "loss": 5.4166,
-      "step": 6578
-    },
-    {
-      "epoch": 3.4310299869621903,
-      "grad_norm": 1.5302101373672485,
-      "learning_rate": 9.38924623115578e-05,
-      "loss": 5.4488,
-      "step": 6579
-    },
-    {
-      "epoch": 3.4315514993481093,
-      "grad_norm": 1.5051119327545166,
-      "learning_rate": 9.389145728643216e-05,
-      "loss": 5.895,
-      "step": 6580
-    },
-    {
-      "epoch": 3.432073011734029,
-      "grad_norm": 1.509310007095337,
-      "learning_rate": 9.389045226130654e-05,
-      "loss": 5.8249,
-      "step": 6581
-    },
-    {
-      "epoch": 3.432594524119948,
-      "grad_norm": 1.4073339700698853,
-      "learning_rate": 9.38894472361809e-05,
-      "loss": 5.7177,
-      "step": 6582
-    },
-    {
-      "epoch": 3.433116036505867,
-      "grad_norm": 1.4969583749771118,
-      "learning_rate": 9.388844221105528e-05,
-      "loss": 5.8456,
-      "step": 6583
-    },
-    {
-      "epoch": 3.4336375488917863,
-      "grad_norm": 1.3901371955871582,
-      "learning_rate": 9.388743718592965e-05,
-      "loss": 5.9751,
-      "step": 6584
-    },
-    {
-      "epoch": 3.4341590612777053,
-      "grad_norm": 1.7261689901351929,
-      "learning_rate": 9.388643216080402e-05,
-      "loss": 5.8478,
-      "step": 6585
-    },
-    {
-      "epoch": 3.4346805736636243,
-      "grad_norm": 1.5623786449432373,
-      "learning_rate": 9.388542713567839e-05,
-      "loss": 5.8553,
-      "step": 6586
-    },
-    {
-      "epoch": 3.435202086049544,
-      "grad_norm": 1.550999402999878,
-      "learning_rate": 9.388442211055277e-05,
-      "loss": 5.6507,
-      "step": 6587
-    },
-    {
-      "epoch": 3.435723598435463,
-      "grad_norm": 1.5739843845367432,
-      "learning_rate": 9.388341708542714e-05,
-      "loss": 5.953,
-      "step": 6588
-    },
-    {
-      "epoch": 3.436245110821382,
-      "grad_norm": 1.5290378332138062,
-      "learning_rate": 9.388241206030152e-05,
-      "loss": 5.6624,
-      "step": 6589
-    },
-    {
-      "epoch": 3.4367666232073013,
-      "grad_norm": 1.5524166822433472,
-      "learning_rate": 9.388140703517589e-05,
-      "loss": 5.5577,
-      "step": 6590
-    },
-    {
-      "epoch": 3.4372881355932203,
-      "grad_norm": 1.6305341720581055,
-      "learning_rate": 9.388040201005026e-05,
-      "loss": 5.3048,
-      "step": 6591
-    },
-    {
-      "epoch": 3.4378096479791393,
-      "grad_norm": 1.6362624168395996,
-      "learning_rate": 9.387939698492463e-05,
-      "loss": 5.5064,
-      "step": 6592
-    },
-    {
-      "epoch": 3.438331160365059,
-      "grad_norm": 1.5737831592559814,
-      "learning_rate": 9.3878391959799e-05,
-      "loss": 5.8229,
-      "step": 6593
-    },
-    {
-      "epoch": 3.438852672750978,
-      "grad_norm": 1.4959999322891235,
-      "learning_rate": 9.387738693467337e-05,
-      "loss": 5.936,
-      "step": 6594
-    },
-    {
-      "epoch": 3.439374185136897,
-      "grad_norm": 1.396419644355774,
-      "learning_rate": 9.387638190954774e-05,
-      "loss": 6.1069,
-      "step": 6595
-    },
-    {
-      "epoch": 3.4398956975228163,
-      "grad_norm": 1.880556344985962,
-      "learning_rate": 9.387537688442211e-05,
-      "loss": 5.3327,
-      "step": 6596
-    },
-    {
-      "epoch": 3.4404172099087353,
-      "grad_norm": 1.6038509607315063,
-      "learning_rate": 9.387437185929648e-05,
-      "loss": 5.8171,
-      "step": 6597
-    },
-    {
-      "epoch": 3.4409387222946544,
-      "grad_norm": 1.666276216506958,
-      "learning_rate": 9.387336683417086e-05,
-      "loss": 5.7179,
-      "step": 6598
-    },
-    {
-      "epoch": 3.441460234680574,
-      "grad_norm": 1.4037463665008545,
-      "learning_rate": 9.387236180904523e-05,
-      "loss": 5.7841,
-      "step": 6599
-    },
-    {
-      "epoch": 3.441981747066493,
-      "grad_norm": 1.4923604726791382,
-      "learning_rate": 9.387135678391961e-05,
-      "loss": 5.8195,
-      "step": 6600
-    },
-    {
-      "epoch": 3.442503259452412,
-      "grad_norm": 1.5877501964569092,
-      "learning_rate": 9.387035175879398e-05,
-      "loss": 6.0516,
-      "step": 6601
-    },
-    {
-      "epoch": 3.4430247718383313,
-      "grad_norm": 1.499282717704773,
-      "learning_rate": 9.386934673366835e-05,
-      "loss": 5.5857,
-      "step": 6602
-    },
-    {
-      "epoch": 3.4435462842242504,
-      "grad_norm": 1.5643787384033203,
-      "learning_rate": 9.386834170854272e-05,
-      "loss": 5.4177,
-      "step": 6603
-    },
-    {
-      "epoch": 3.4440677966101694,
-      "grad_norm": 1.4347288608551025,
-      "learning_rate": 9.38673366834171e-05,
-      "loss": 5.8908,
-      "step": 6604
-    },
-    {
-      "epoch": 3.444589308996089,
-      "grad_norm": 1.5068795680999756,
-      "learning_rate": 9.386633165829146e-05,
-      "loss": 5.4803,
-      "step": 6605
-    },
-    {
-      "epoch": 3.445110821382008,
-      "grad_norm": 1.438913345336914,
-      "learning_rate": 9.386532663316582e-05,
-      "loss": 5.8202,
-      "step": 6606
-    },
-    {
-      "epoch": 3.445632333767927,
-      "grad_norm": 1.4150618314743042,
-      "learning_rate": 9.38643216080402e-05,
-      "loss": 5.8957,
-      "step": 6607
-    },
-    {
-      "epoch": 3.4461538461538463,
-      "grad_norm": 1.514570713043213,
-      "learning_rate": 9.386331658291458e-05,
-      "loss": 5.7693,
-      "step": 6608
-    },
-    {
-      "epoch": 3.4466753585397654,
-      "grad_norm": 1.6335532665252686,
-      "learning_rate": 9.386231155778896e-05,
-      "loss": 5.4071,
-      "step": 6609
-    },
-    {
-      "epoch": 3.4471968709256844,
-      "grad_norm": 1.3357943296432495,
-      "learning_rate": 9.386130653266332e-05,
-      "loss": 5.7355,
-      "step": 6610
-    },
-    {
-      "epoch": 3.447718383311604,
-      "grad_norm": 1.7022035121917725,
-      "learning_rate": 9.38603015075377e-05,
-      "loss": 5.4866,
-      "step": 6611
-    },
-    {
-      "epoch": 3.448239895697523,
-      "grad_norm": 1.5016149282455444,
-      "learning_rate": 9.385929648241206e-05,
-      "loss": 5.7188,
-      "step": 6612
-    },
-    {
-      "epoch": 3.448761408083442,
-      "grad_norm": 1.4516148567199707,
-      "learning_rate": 9.385829145728644e-05,
-      "loss": 5.8095,
-      "step": 6613
-    },
-    {
-      "epoch": 3.4492829204693614,
-      "grad_norm": 1.4787834882736206,
-      "learning_rate": 9.38572864321608e-05,
-      "loss": 5.3873,
-      "step": 6614
-    },
-    {
-      "epoch": 3.4498044328552804,
-      "grad_norm": 1.458290696144104,
-      "learning_rate": 9.385628140703518e-05,
-      "loss": 5.9995,
-      "step": 6615
-    },
-    {
-      "epoch": 3.4503259452411994,
-      "grad_norm": 1.7118620872497559,
-      "learning_rate": 9.385527638190955e-05,
-      "loss": 6.0057,
-      "step": 6616
-    },
-    {
-      "epoch": 3.450847457627119,
-      "grad_norm": 1.298499345779419,
-      "learning_rate": 9.385427135678393e-05,
-      "loss": 6.2426,
-      "step": 6617
-    },
-    {
-      "epoch": 3.451368970013038,
-      "grad_norm": 1.4216827154159546,
-      "learning_rate": 9.385326633165829e-05,
-      "loss": 6.0461,
-      "step": 6618
-    },
-    {
-      "epoch": 3.451890482398957,
-      "grad_norm": 1.681530475616455,
-      "learning_rate": 9.385226130653267e-05,
-      "loss": 5.5226,
-      "step": 6619
-    },
-    {
-      "epoch": 3.4524119947848764,
-      "grad_norm": 1.4151504039764404,
-      "learning_rate": 9.385125628140705e-05,
-      "loss": 5.7454,
-      "step": 6620
-    },
-    {
-      "epoch": 3.4529335071707954,
-      "grad_norm": 1.3347724676132202,
-      "learning_rate": 9.385025125628141e-05,
-      "loss": 5.8242,
-      "step": 6621
-    },
-    {
-      "epoch": 3.4534550195567144,
-      "grad_norm": 1.4965983629226685,
-      "learning_rate": 9.384924623115579e-05,
-      "loss": 6.2195,
-      "step": 6622
-    },
-    {
-      "epoch": 3.453976531942634,
-      "grad_norm": 1.553650975227356,
-      "learning_rate": 9.384824120603015e-05,
-      "loss": 5.5846,
-      "step": 6623
-    },
-    {
-      "epoch": 3.454498044328553,
-      "grad_norm": 1.4847723245620728,
-      "learning_rate": 9.384723618090453e-05,
-      "loss": 5.9112,
-      "step": 6624
-    },
-    {
-      "epoch": 3.455019556714472,
-      "grad_norm": 1.516342043876648,
-      "learning_rate": 9.38462311557789e-05,
-      "loss": 5.7741,
-      "step": 6625
-    },
-    {
-      "epoch": 3.4555410691003914,
-      "grad_norm": 1.5443488359451294,
-      "learning_rate": 9.384522613065327e-05,
-      "loss": 5.3263,
-      "step": 6626
-    },
-    {
-      "epoch": 3.4560625814863104,
-      "grad_norm": 1.4456732273101807,
-      "learning_rate": 9.384422110552764e-05,
-      "loss": 6.1292,
-      "step": 6627
-    },
-    {
-      "epoch": 3.4565840938722294,
-      "grad_norm": 1.902841567993164,
-      "learning_rate": 9.384321608040201e-05,
-      "loss": 5.0048,
-      "step": 6628
-    },
-    {
-      "epoch": 3.4571056062581484,
-      "grad_norm": 1.8350127935409546,
-      "learning_rate": 9.384221105527639e-05,
-      "loss": 5.5224,
-      "step": 6629
-    },
-    {
-      "epoch": 3.457627118644068,
-      "grad_norm": 1.5191256999969482,
-      "learning_rate": 9.384120603015077e-05,
-      "loss": 6.043,
-      "step": 6630
-    },
-    {
-      "epoch": 3.458148631029987,
-      "grad_norm": 2.2479302883148193,
-      "learning_rate": 9.384020100502513e-05,
-      "loss": 5.6505,
-      "step": 6631
-    },
-    {
-      "epoch": 3.458670143415906,
-      "grad_norm": 1.5241508483886719,
-      "learning_rate": 9.38391959798995e-05,
-      "loss": 5.8403,
-      "step": 6632
-    },
-    {
-      "epoch": 3.4591916558018254,
-      "grad_norm": 1.5110018253326416,
-      "learning_rate": 9.383819095477388e-05,
-      "loss": 5.7992,
-      "step": 6633
-    },
-    {
-      "epoch": 3.4597131681877444,
-      "grad_norm": 1.4047322273254395,
-      "learning_rate": 9.383718592964824e-05,
-      "loss": 6.1065,
-      "step": 6634
-    },
-    {
-      "epoch": 3.4602346805736635,
-      "grad_norm": 1.509946346282959,
-      "learning_rate": 9.383618090452262e-05,
-      "loss": 5.6617,
-      "step": 6635
-    },
-    {
-      "epoch": 3.460756192959583,
-      "grad_norm": 1.503730058670044,
-      "learning_rate": 9.383517587939698e-05,
-      "loss": 6.0458,
-      "step": 6636
-    },
-    {
-      "epoch": 3.461277705345502,
-      "grad_norm": 1.4189656972885132,
-      "learning_rate": 9.383417085427136e-05,
-      "loss": 5.8626,
-      "step": 6637
-    },
-    {
-      "epoch": 3.461799217731421,
-      "grad_norm": 1.3749526739120483,
-      "learning_rate": 9.383316582914572e-05,
-      "loss": 5.6927,
-      "step": 6638
-    },
-    {
-      "epoch": 3.4623207301173404,
-      "grad_norm": 1.45967435836792,
-      "learning_rate": 9.38321608040201e-05,
-      "loss": 5.5792,
-      "step": 6639
-    },
-    {
-      "epoch": 3.4628422425032594,
-      "grad_norm": 1.6039050817489624,
-      "learning_rate": 9.383115577889448e-05,
-      "loss": 5.8064,
-      "step": 6640
-    },
-    {
-      "epoch": 3.4633637548891785,
-      "grad_norm": 1.4519388675689697,
-      "learning_rate": 9.383015075376886e-05,
-      "loss": 6.2917,
-      "step": 6641
-    },
-    {
-      "epoch": 3.463885267275098,
-      "grad_norm": 1.4016318321228027,
-      "learning_rate": 9.382914572864322e-05,
-      "loss": 5.9879,
-      "step": 6642
-    },
-    {
-      "epoch": 3.464406779661017,
-      "grad_norm": 1.621645450592041,
-      "learning_rate": 9.38281407035176e-05,
-      "loss": 5.5346,
-      "step": 6643
-    },
-    {
-      "epoch": 3.464928292046936,
-      "grad_norm": 1.639356017112732,
-      "learning_rate": 9.382713567839196e-05,
-      "loss": 5.3508,
-      "step": 6644
-    },
-    {
-      "epoch": 3.4654498044328554,
-      "grad_norm": 2.164665937423706,
-      "learning_rate": 9.382613065326633e-05,
-      "loss": 5.4644,
-      "step": 6645
-    },
-    {
-      "epoch": 3.4659713168187745,
-      "grad_norm": 1.4670137166976929,
-      "learning_rate": 9.38251256281407e-05,
-      "loss": 6.1142,
-      "step": 6646
-    },
-    {
-      "epoch": 3.4664928292046935,
-      "grad_norm": 1.4563056230545044,
-      "learning_rate": 9.382412060301507e-05,
-      "loss": 5.7108,
-      "step": 6647
-    },
-    {
-      "epoch": 3.467014341590613,
-      "grad_norm": 1.355540156364441,
-      "learning_rate": 9.382311557788945e-05,
-      "loss": 6.0788,
-      "step": 6648
-    },
-    {
-      "epoch": 3.467535853976532,
-      "grad_norm": 1.5267772674560547,
-      "learning_rate": 9.382211055276383e-05,
-      "loss": 5.3276,
-      "step": 6649
-    },
-    {
-      "epoch": 3.468057366362451,
-      "grad_norm": 1.5112308263778687,
-      "learning_rate": 9.38211055276382e-05,
-      "loss": 5.6088,
-      "step": 6650
-    },
-    {
-      "epoch": 3.4685788787483705,
-      "grad_norm": 1.426392674446106,
-      "learning_rate": 9.382010050251257e-05,
-      "loss": 6.0311,
-      "step": 6651
-    },
-    {
-      "epoch": 3.4691003911342895,
-      "grad_norm": 1.5667036771774292,
-      "learning_rate": 9.381909547738695e-05,
-      "loss": 5.7174,
-      "step": 6652
-    },
-    {
-      "epoch": 3.4696219035202085,
-      "grad_norm": 1.5902961492538452,
-      "learning_rate": 9.381809045226131e-05,
-      "loss": 5.4754,
-      "step": 6653
-    },
-    {
-      "epoch": 3.470143415906128,
-      "grad_norm": 1.343227505683899,
-      "learning_rate": 9.381708542713569e-05,
-      "loss": 6.0112,
-      "step": 6654
-    },
-    {
-      "epoch": 3.470664928292047,
-      "grad_norm": 1.4775993824005127,
-      "learning_rate": 9.381608040201005e-05,
-      "loss": 5.752,
-      "step": 6655
-    },
-    {
-      "epoch": 3.471186440677966,
-      "grad_norm": 1.455881953239441,
-      "learning_rate": 9.381507537688443e-05,
-      "loss": 5.8514,
-      "step": 6656
-    },
-    {
-      "epoch": 3.471707953063885,
-      "grad_norm": 1.5065813064575195,
-      "learning_rate": 9.38140703517588e-05,
-      "loss": 5.8573,
-      "step": 6657
-    },
-    {
-      "epoch": 3.4722294654498045,
-      "grad_norm": 1.4667167663574219,
-      "learning_rate": 9.381306532663316e-05,
-      "loss": 5.893,
-      "step": 6658
-    },
-    {
-      "epoch": 3.4727509778357235,
-      "grad_norm": 1.4245109558105469,
-      "learning_rate": 9.381206030150754e-05,
-      "loss": 5.7042,
-      "step": 6659
-    },
-    {
-      "epoch": 3.4732724902216425,
-      "grad_norm": 1.5085673332214355,
-      "learning_rate": 9.381105527638191e-05,
-      "loss": 5.576,
-      "step": 6660
-    },
-    {
-      "epoch": 3.473794002607562,
-      "grad_norm": 1.4945487976074219,
-      "learning_rate": 9.381005025125629e-05,
-      "loss": 5.9713,
-      "step": 6661
-    },
-    {
-      "epoch": 3.474315514993481,
-      "grad_norm": 1.5869314670562744,
-      "learning_rate": 9.380904522613066e-05,
-      "loss": 5.6719,
-      "step": 6662
-    },
-    {
-      "epoch": 3.4748370273794,
-      "grad_norm": 1.330153226852417,
-      "learning_rate": 9.380804020100503e-05,
-      "loss": 5.9293,
-      "step": 6663
-    },
-    {
-      "epoch": 3.4753585397653195,
-      "grad_norm": 1.4297603368759155,
-      "learning_rate": 9.38070351758794e-05,
-      "loss": 5.5187,
-      "step": 6664
-    },
-    {
-      "epoch": 3.4758800521512385,
-      "grad_norm": 1.3853750228881836,
-      "learning_rate": 9.380603015075378e-05,
-      "loss": 5.9606,
-      "step": 6665
-    },
-    {
-      "epoch": 3.4764015645371575,
-      "grad_norm": 1.3818566799163818,
-      "learning_rate": 9.380502512562814e-05,
-      "loss": 5.628,
-      "step": 6666
-    },
-    {
-      "epoch": 3.476923076923077,
-      "grad_norm": 1.4198203086853027,
-      "learning_rate": 9.380402010050252e-05,
-      "loss": 5.9571,
-      "step": 6667
-    },
-    {
-      "epoch": 3.477444589308996,
-      "grad_norm": 1.3652359247207642,
-      "learning_rate": 9.380301507537688e-05,
-      "loss": 5.789,
-      "step": 6668
-    },
-    {
-      "epoch": 3.477966101694915,
-      "grad_norm": 1.4546282291412354,
-      "learning_rate": 9.380201005025126e-05,
-      "loss": 6.0332,
-      "step": 6669
-    },
-    {
-      "epoch": 3.4784876140808345,
-      "grad_norm": 1.629219889640808,
-      "learning_rate": 9.380100502512564e-05,
-      "loss": 4.7585,
-      "step": 6670
-    },
-    {
-      "epoch": 3.4790091264667535,
-      "grad_norm": 1.3091163635253906,
-      "learning_rate": 9.38e-05,
-      "loss": 6.0605,
-      "step": 6671
-    },
-    {
-      "epoch": 3.4795306388526726,
-      "grad_norm": 1.4081790447235107,
-      "learning_rate": 9.379899497487438e-05,
-      "loss": 5.7456,
-      "step": 6672
-    },
-    {
-      "epoch": 3.480052151238592,
-      "grad_norm": 1.6523429155349731,
-      "learning_rate": 9.379798994974875e-05,
-      "loss": 5.0601,
-      "step": 6673
-    },
-    {
-      "epoch": 3.480573663624511,
-      "grad_norm": 1.5568499565124512,
-      "learning_rate": 9.379698492462312e-05,
-      "loss": 5.7539,
-      "step": 6674
-    },
-    {
-      "epoch": 3.48109517601043,
-      "grad_norm": 1.4979923963546753,
-      "learning_rate": 9.379597989949749e-05,
-      "loss": 5.6272,
-      "step": 6675
-    },
-    {
-      "epoch": 3.4816166883963495,
-      "grad_norm": 1.6634958982467651,
-      "learning_rate": 9.379497487437187e-05,
-      "loss": 5.1706,
-      "step": 6676
-    },
-    {
-      "epoch": 3.4821382007822685,
-      "grad_norm": 1.3872904777526855,
-      "learning_rate": 9.379396984924623e-05,
-      "loss": 5.744,
-      "step": 6677
-    },
-    {
-      "epoch": 3.4826597131681876,
-      "grad_norm": 1.6882271766662598,
-      "learning_rate": 9.379296482412061e-05,
-      "loss": 5.5906,
-      "step": 6678
-    },
-    {
-      "epoch": 3.483181225554107,
-      "grad_norm": 1.5967460870742798,
-      "learning_rate": 9.379195979899497e-05,
-      "loss": 5.219,
-      "step": 6679
-    },
-    {
-      "epoch": 3.483702737940026,
-      "grad_norm": 1.473443865776062,
-      "learning_rate": 9.379095477386935e-05,
-      "loss": 5.8302,
-      "step": 6680
-    },
-    {
-      "epoch": 3.484224250325945,
-      "grad_norm": 1.4015426635742188,
-      "learning_rate": 9.378994974874373e-05,
-      "loss": 5.5922,
-      "step": 6681
-    },
-    {
-      "epoch": 3.4847457627118645,
-      "grad_norm": 1.4231053590774536,
-      "learning_rate": 9.37889447236181e-05,
-      "loss": 6.0124,
-      "step": 6682
-    },
-    {
-      "epoch": 3.4852672750977836,
-      "grad_norm": 1.4743679761886597,
-      "learning_rate": 9.378793969849247e-05,
-      "loss": 5.414,
-      "step": 6683
-    },
-    {
-      "epoch": 3.4857887874837026,
-      "grad_norm": 1.5142967700958252,
-      "learning_rate": 9.378693467336685e-05,
-      "loss": 5.4161,
-      "step": 6684
-    },
-    {
-      "epoch": 3.486310299869622,
-      "grad_norm": 1.5752254724502563,
-      "learning_rate": 9.378592964824121e-05,
-      "loss": 5.767,
-      "step": 6685
-    },
-    {
-      "epoch": 3.486831812255541,
-      "grad_norm": 1.5675541162490845,
-      "learning_rate": 9.378492462311558e-05,
-      "loss": 5.7705,
-      "step": 6686
-    },
-    {
-      "epoch": 3.48735332464146,
-      "grad_norm": 2.1700732707977295,
-      "learning_rate": 9.378391959798995e-05,
-      "loss": 5.4788,
-      "step": 6687
-    },
-    {
-      "epoch": 3.4878748370273795,
-      "grad_norm": 1.623518705368042,
-      "learning_rate": 9.378291457286432e-05,
-      "loss": 6.2171,
-      "step": 6688
-    },
-    {
-      "epoch": 3.4883963494132986,
-      "grad_norm": 1.8324391841888428,
-      "learning_rate": 9.37819095477387e-05,
-      "loss": 5.7916,
-      "step": 6689
-    },
-    {
-      "epoch": 3.4889178617992176,
-      "grad_norm": 1.6197644472122192,
-      "learning_rate": 9.378090452261306e-05,
-      "loss": 5.7418,
-      "step": 6690
-    },
-    {
-      "epoch": 3.489439374185137,
-      "grad_norm": 1.7779282331466675,
-      "learning_rate": 9.377989949748744e-05,
-      "loss": 5.3195,
-      "step": 6691
-    },
-    {
-      "epoch": 3.489960886571056,
-      "grad_norm": 1.742152214050293,
-      "learning_rate": 9.377889447236182e-05,
-      "loss": 5.2932,
-      "step": 6692
-    },
-    {
-      "epoch": 3.490482398956975,
-      "grad_norm": 1.5503824949264526,
-      "learning_rate": 9.37778894472362e-05,
-      "loss": 5.6906,
-      "step": 6693
-    },
-    {
-      "epoch": 3.4910039113428946,
-      "grad_norm": 1.3012527227401733,
-      "learning_rate": 9.377688442211056e-05,
-      "loss": 6.0423,
-      "step": 6694
-    },
-    {
-      "epoch": 3.4915254237288136,
-      "grad_norm": 1.6402363777160645,
-      "learning_rate": 9.377587939698494e-05,
-      "loss": 5.1606,
-      "step": 6695
-    },
-    {
-      "epoch": 3.4920469361147326,
-      "grad_norm": 1.5132253170013428,
-      "learning_rate": 9.37748743718593e-05,
-      "loss": 5.9895,
-      "step": 6696
-    },
-    {
-      "epoch": 3.492568448500652,
-      "grad_norm": 1.6209053993225098,
-      "learning_rate": 9.377386934673368e-05,
-      "loss": 5.8754,
-      "step": 6697
-    },
-    {
-      "epoch": 3.493089960886571,
-      "grad_norm": 1.5189119577407837,
-      "learning_rate": 9.377286432160804e-05,
-      "loss": 5.8027,
-      "step": 6698
-    },
-    {
-      "epoch": 3.49361147327249,
-      "grad_norm": 1.549461007118225,
-      "learning_rate": 9.37718592964824e-05,
-      "loss": 5.6424,
-      "step": 6699
-    },
-    {
-      "epoch": 3.4941329856584096,
-      "grad_norm": 1.4738787412643433,
-      "learning_rate": 9.377085427135678e-05,
-      "loss": 5.7068,
-      "step": 6700
-    },
-    {
-      "epoch": 3.4946544980443286,
-      "grad_norm": 1.5362428426742554,
-      "learning_rate": 9.376984924623116e-05,
-      "loss": 5.8443,
-      "step": 6701
-    },
-    {
-      "epoch": 3.4951760104302476,
-      "grad_norm": 1.389529824256897,
-      "learning_rate": 9.376884422110554e-05,
-      "loss": 6.0809,
-      "step": 6702
-    },
-    {
-      "epoch": 3.495697522816167,
-      "grad_norm": 1.5592011213302612,
-      "learning_rate": 9.37678391959799e-05,
-      "loss": 5.4351,
-      "step": 6703
-    },
-    {
-      "epoch": 3.496219035202086,
-      "grad_norm": 1.6276932954788208,
-      "learning_rate": 9.376683417085428e-05,
-      "loss": 5.5809,
-      "step": 6704
-    },
-    {
-      "epoch": 3.496740547588005,
-      "grad_norm": 1.4372819662094116,
-      "learning_rate": 9.376582914572865e-05,
-      "loss": 6.0987,
-      "step": 6705
-    },
-    {
-      "epoch": 3.4972620599739246,
-      "grad_norm": 1.4826847314834595,
-      "learning_rate": 9.376482412060302e-05,
-      "loss": 5.9527,
-      "step": 6706
-    },
-    {
-      "epoch": 3.4977835723598436,
-      "grad_norm": 1.3536359071731567,
-      "learning_rate": 9.376381909547739e-05,
-      "loss": 5.9444,
-      "step": 6707
-    },
-    {
-      "epoch": 3.4983050847457626,
-      "grad_norm": 1.4969521760940552,
-      "learning_rate": 9.376281407035177e-05,
-      "loss": 6.0206,
-      "step": 6708
-    },
-    {
-      "epoch": 3.498826597131682,
-      "grad_norm": 1.5260578393936157,
-      "learning_rate": 9.376180904522613e-05,
-      "loss": 5.969,
-      "step": 6709
-    },
-    {
-      "epoch": 3.499348109517601,
-      "grad_norm": 1.7617967128753662,
-      "learning_rate": 9.376080402010051e-05,
-      "loss": 5.4568,
-      "step": 6710
-    },
-    {
-      "epoch": 3.49986962190352,
-      "grad_norm": 1.4503662586212158,
-      "learning_rate": 9.375979899497487e-05,
-      "loss": 5.8123,
-      "step": 6711
-    },
-    {
-      "epoch": 3.5003911342894396,
-      "grad_norm": 1.5299630165100098,
-      "learning_rate": 9.375879396984925e-05,
-      "loss": 5.6094,
-      "step": 6712
-    },
-    {
-      "epoch": 3.5009126466753586,
-      "grad_norm": 1.5533326864242554,
-      "learning_rate": 9.375778894472363e-05,
-      "loss": 5.8498,
-      "step": 6713
-    },
-    {
-      "epoch": 3.5014341590612776,
-      "grad_norm": 1.3899163007736206,
-      "learning_rate": 9.375678391959799e-05,
-      "loss": 6.0214,
-      "step": 6714
-    },
-    {
-      "epoch": 3.501955671447197,
-      "grad_norm": 1.3556277751922607,
-      "learning_rate": 9.375577889447237e-05,
-      "loss": 5.9884,
-      "step": 6715
-    },
-    {
-      "epoch": 3.502477183833116,
-      "grad_norm": 1.4719994068145752,
-      "learning_rate": 9.375477386934673e-05,
-      "loss": 5.4115,
-      "step": 6716
-    },
-    {
-      "epoch": 3.502998696219035,
-      "grad_norm": 1.6984734535217285,
-      "learning_rate": 9.375376884422111e-05,
-      "loss": 5.6738,
-      "step": 6717
-    },
-    {
-      "epoch": 3.5035202086049546,
-      "grad_norm": 1.3937760591506958,
-      "learning_rate": 9.375276381909548e-05,
-      "loss": 5.7454,
-      "step": 6718
-    },
-    {
-      "epoch": 3.5040417209908736,
-      "grad_norm": 1.308585286140442,
-      "learning_rate": 9.375175879396985e-05,
-      "loss": 5.9364,
-      "step": 6719
-    },
-    {
-      "epoch": 3.5045632333767927,
-      "grad_norm": 1.5324649810791016,
-      "learning_rate": 9.375075376884422e-05,
-      "loss": 5.8499,
-      "step": 6720
-    },
-    {
-      "epoch": 3.505084745762712,
-      "grad_norm": 1.3970049619674683,
-      "learning_rate": 9.37497487437186e-05,
-      "loss": 5.5446,
-      "step": 6721
-    },
-    {
-      "epoch": 3.505606258148631,
-      "grad_norm": 1.3376753330230713,
-      "learning_rate": 9.374874371859297e-05,
-      "loss": 6.193,
-      "step": 6722
-    },
-    {
-      "epoch": 3.50612777053455,
-      "grad_norm": 1.3524177074432373,
-      "learning_rate": 9.374773869346735e-05,
-      "loss": 6.0968,
-      "step": 6723
-    },
-    {
-      "epoch": 3.5066492829204696,
-      "grad_norm": 1.4086419343948364,
-      "learning_rate": 9.374673366834172e-05,
-      "loss": 6.0634,
-      "step": 6724
-    },
-    {
-      "epoch": 3.5071707953063886,
-      "grad_norm": 2.1302547454833984,
-      "learning_rate": 9.374572864321608e-05,
-      "loss": 5.4294,
-      "step": 6725
-    },
-    {
-      "epoch": 3.5076923076923077,
-      "grad_norm": 1.6257654428482056,
-      "learning_rate": 9.374472361809046e-05,
-      "loss": 5.8636,
-      "step": 6726
-    },
-    {
-      "epoch": 3.508213820078227,
-      "grad_norm": 1.6204512119293213,
-      "learning_rate": 9.374371859296482e-05,
-      "loss": 5.218,
-      "step": 6727
-    },
-    {
-      "epoch": 3.508735332464146,
-      "grad_norm": 1.6032049655914307,
-      "learning_rate": 9.37427135678392e-05,
-      "loss": 5.3196,
-      "step": 6728
-    },
-    {
-      "epoch": 3.509256844850065,
-      "grad_norm": 1.4963319301605225,
-      "learning_rate": 9.374170854271356e-05,
-      "loss": 5.8199,
-      "step": 6729
-    },
-    {
-      "epoch": 3.509778357235984,
-      "grad_norm": 1.4500224590301514,
-      "learning_rate": 9.374070351758794e-05,
-      "loss": 5.8819,
-      "step": 6730
-    },
-    {
-      "epoch": 3.5102998696219037,
-      "grad_norm": 1.35861074924469,
-      "learning_rate": 9.373969849246231e-05,
-      "loss": 6.1633,
-      "step": 6731
-    },
-    {
-      "epoch": 3.5108213820078227,
-      "grad_norm": 1.608127474784851,
-      "learning_rate": 9.373869346733668e-05,
-      "loss": 5.6522,
-      "step": 6732
-    },
-    {
-      "epoch": 3.5113428943937417,
-      "grad_norm": 1.5624979734420776,
-      "learning_rate": 9.373768844221106e-05,
-      "loss": 5.4789,
-      "step": 6733
-    },
-    {
-      "epoch": 3.511864406779661,
-      "grad_norm": 1.612146258354187,
-      "learning_rate": 9.373668341708544e-05,
-      "loss": 5.5371,
-      "step": 6734
-    },
-    {
-      "epoch": 3.51238591916558,
-      "grad_norm": 1.3103408813476562,
-      "learning_rate": 9.37356783919598e-05,
-      "loss": 6.1346,
-      "step": 6735
-    },
-    {
-      "epoch": 3.512907431551499,
-      "grad_norm": 2.056246280670166,
-      "learning_rate": 9.373467336683418e-05,
-      "loss": 5.5654,
-      "step": 6736
-    },
-    {
-      "epoch": 3.5134289439374187,
-      "grad_norm": 1.4977360963821411,
-      "learning_rate": 9.373366834170855e-05,
-      "loss": 5.7089,
-      "step": 6737
-    },
-    {
-      "epoch": 3.5139504563233377,
-      "grad_norm": 1.6363375186920166,
-      "learning_rate": 9.373266331658291e-05,
-      "loss": 5.7208,
-      "step": 6738
-    },
-    {
-      "epoch": 3.5144719687092567,
-      "grad_norm": 1.3955926895141602,
-      "learning_rate": 9.373165829145729e-05,
-      "loss": 6.0194,
-      "step": 6739
-    },
-    {
-      "epoch": 3.514993481095176,
-      "grad_norm": 1.450245976448059,
-      "learning_rate": 9.373065326633165e-05,
-      "loss": 5.9459,
-      "step": 6740
-    },
-    {
-      "epoch": 3.515514993481095,
-      "grad_norm": 1.554410457611084,
-      "learning_rate": 9.372964824120603e-05,
-      "loss": 5.4273,
-      "step": 6741
-    },
-    {
-      "epoch": 3.516036505867014,
-      "grad_norm": 1.5028961896896362,
-      "learning_rate": 9.372864321608041e-05,
-      "loss": 6.2075,
-      "step": 6742
-    },
-    {
-      "epoch": 3.5165580182529332,
-      "grad_norm": 1.414805293083191,
-      "learning_rate": 9.372763819095479e-05,
-      "loss": 5.9259,
-      "step": 6743
-    },
-    {
-      "epoch": 3.5170795306388527,
-      "grad_norm": 1.5399208068847656,
-      "learning_rate": 9.372663316582915e-05,
-      "loss": 5.663,
-      "step": 6744
-    },
-    {
-      "epoch": 3.5176010430247717,
-      "grad_norm": 1.3553940057754517,
-      "learning_rate": 9.372562814070353e-05,
-      "loss": 5.4974,
-      "step": 6745
-    },
-    {
-      "epoch": 3.5181225554106907,
-      "grad_norm": 1.5037813186645508,
-      "learning_rate": 9.372462311557789e-05,
-      "loss": 5.8822,
-      "step": 6746
-    },
-    {
-      "epoch": 3.51864406779661,
-      "grad_norm": 1.302720308303833,
-      "learning_rate": 9.372361809045227e-05,
-      "loss": 5.7823,
-      "step": 6747
-    },
-    {
-      "epoch": 3.5191655801825292,
-      "grad_norm": 1.6097068786621094,
-      "learning_rate": 9.372261306532664e-05,
-      "loss": 5.5633,
-      "step": 6748
-    },
-    {
-      "epoch": 3.5196870925684483,
-      "grad_norm": 1.4835314750671387,
-      "learning_rate": 9.372160804020101e-05,
-      "loss": 5.9501,
-      "step": 6749
-    },
-    {
-      "epoch": 3.5202086049543677,
-      "grad_norm": 1.4716057777404785,
-      "learning_rate": 9.372060301507538e-05,
-      "loss": 5.5612,
-      "step": 6750
-    },
-    {
-      "epoch": 3.5207301173402867,
-      "grad_norm": 1.5794897079467773,
-      "learning_rate": 9.371959798994974e-05,
-      "loss": 5.5504,
-      "step": 6751
-    },
-    {
-      "epoch": 3.5212516297262058,
-      "grad_norm": 1.5934181213378906,
-      "learning_rate": 9.371859296482412e-05,
-      "loss": 5.4545,
-      "step": 6752
-    },
-    {
-      "epoch": 3.521773142112125,
-      "grad_norm": 1.5200958251953125,
-      "learning_rate": 9.37175879396985e-05,
-      "loss": 5.7425,
-      "step": 6753
-    },
-    {
-      "epoch": 3.5222946544980442,
-      "grad_norm": 1.5170706510543823,
-      "learning_rate": 9.371658291457287e-05,
-      "loss": 5.7668,
-      "step": 6754
-    },
-    {
-      "epoch": 3.5228161668839633,
-      "grad_norm": 1.6036436557769775,
-      "learning_rate": 9.371557788944724e-05,
-      "loss": 4.8532,
-      "step": 6755
-    },
-    {
-      "epoch": 3.5233376792698827,
-      "grad_norm": 1.6023293733596802,
-      "learning_rate": 9.371457286432162e-05,
-      "loss": 5.407,
-      "step": 6756
-    },
-    {
-      "epoch": 3.5238591916558017,
-      "grad_norm": 1.491373896598816,
-      "learning_rate": 9.371356783919598e-05,
-      "loss": 5.9984,
-      "step": 6757
-    },
-    {
-      "epoch": 3.5243807040417208,
-      "grad_norm": 1.4562264680862427,
-      "learning_rate": 9.371256281407036e-05,
-      "loss": 5.5326,
-      "step": 6758
-    },
-    {
-      "epoch": 3.5249022164276402,
-      "grad_norm": 1.5026049613952637,
-      "learning_rate": 9.371155778894472e-05,
-      "loss": 5.6318,
-      "step": 6759
-    },
-    {
-      "epoch": 3.5254237288135593,
-      "grad_norm": 1.623980164527893,
-      "learning_rate": 9.37105527638191e-05,
-      "loss": 5.5478,
-      "step": 6760
-    },
-    {
-      "epoch": 3.5259452411994783,
-      "grad_norm": 1.494152307510376,
-      "learning_rate": 9.370954773869347e-05,
-      "loss": 5.8064,
-      "step": 6761
-    },
-    {
-      "epoch": 3.5264667535853977,
-      "grad_norm": 1.517565369606018,
-      "learning_rate": 9.370854271356784e-05,
-      "loss": 5.4861,
-      "step": 6762
-    },
-    {
-      "epoch": 3.5269882659713168,
-      "grad_norm": 1.523149847984314,
-      "learning_rate": 9.370753768844222e-05,
-      "loss": 5.8954,
-      "step": 6763
-    },
-    {
-      "epoch": 3.527509778357236,
-      "grad_norm": 1.4593862295150757,
-      "learning_rate": 9.370653266331659e-05,
-      "loss": 5.9466,
-      "step": 6764
-    },
-    {
-      "epoch": 3.5280312907431552,
-      "grad_norm": 1.3891209363937378,
-      "learning_rate": 9.370552763819096e-05,
-      "loss": 5.6893,
-      "step": 6765
-    },
-    {
-      "epoch": 3.5285528031290743,
-      "grad_norm": 1.4780783653259277,
-      "learning_rate": 9.370452261306533e-05,
-      "loss": 5.7885,
-      "step": 6766
-    },
-    {
-      "epoch": 3.5290743155149933,
-      "grad_norm": 1.550339698791504,
-      "learning_rate": 9.37035175879397e-05,
-      "loss": 5.8575,
-      "step": 6767
-    },
-    {
-      "epoch": 3.5295958279009128,
-      "grad_norm": 1.528484582901001,
-      "learning_rate": 9.370251256281407e-05,
-      "loss": 5.8926,
-      "step": 6768
-    },
-    {
-      "epoch": 3.5301173402868318,
-      "grad_norm": 1.460328221321106,
-      "learning_rate": 9.370150753768845e-05,
-      "loss": 5.6154,
-      "step": 6769
-    },
-    {
-      "epoch": 3.530638852672751,
-      "grad_norm": 1.518493890762329,
-      "learning_rate": 9.370050251256281e-05,
-      "loss": 5.7375,
-      "step": 6770
-    },
-    {
-      "epoch": 3.5311603650586703,
-      "grad_norm": 1.3918448686599731,
-      "learning_rate": 9.369949748743719e-05,
-      "loss": 6.1041,
-      "step": 6771
-    },
-    {
-      "epoch": 3.5316818774445893,
-      "grad_norm": 1.6287676095962524,
-      "learning_rate": 9.369849246231155e-05,
-      "loss": 5.8486,
-      "step": 6772
-    },
-    {
-      "epoch": 3.5322033898305083,
-      "grad_norm": 1.4579936265945435,
-      "learning_rate": 9.369748743718593e-05,
-      "loss": 6.1125,
-      "step": 6773
-    },
-    {
-      "epoch": 3.5327249022164278,
-      "grad_norm": 1.5070048570632935,
-      "learning_rate": 9.369648241206031e-05,
-      "loss": 5.9285,
-      "step": 6774
-    },
-    {
-      "epoch": 3.533246414602347,
-      "grad_norm": 1.5988515615463257,
-      "learning_rate": 9.369547738693469e-05,
-      "loss": 5.8352,
-      "step": 6775
-    },
-    {
-      "epoch": 3.533767926988266,
-      "grad_norm": 1.4927008152008057,
-      "learning_rate": 9.369447236180905e-05,
-      "loss": 5.5095,
-      "step": 6776
-    },
-    {
-      "epoch": 3.5342894393741853,
-      "grad_norm": 1.4838271141052246,
-      "learning_rate": 9.369346733668343e-05,
-      "loss": 5.7209,
-      "step": 6777
-    },
-    {
-      "epoch": 3.5348109517601043,
-      "grad_norm": 1.5043238401412964,
-      "learning_rate": 9.36924623115578e-05,
-      "loss": 4.9815,
-      "step": 6778
-    },
-    {
-      "epoch": 3.5353324641460233,
-      "grad_norm": 1.4178813695907593,
-      "learning_rate": 9.369145728643216e-05,
-      "loss": 5.7824,
-      "step": 6779
-    },
-    {
-      "epoch": 3.5358539765319428,
-      "grad_norm": 1.4742341041564941,
-      "learning_rate": 9.369045226130654e-05,
-      "loss": 5.7549,
-      "step": 6780
-    },
-    {
-      "epoch": 3.536375488917862,
-      "grad_norm": 1.5698914527893066,
-      "learning_rate": 9.36894472361809e-05,
-      "loss": 5.7693,
-      "step": 6781
-    },
-    {
-      "epoch": 3.536897001303781,
-      "grad_norm": 1.557430386543274,
-      "learning_rate": 9.368844221105528e-05,
-      "loss": 5.3739,
-      "step": 6782
-    },
-    {
-      "epoch": 3.5374185136897003,
-      "grad_norm": 1.4710289239883423,
-      "learning_rate": 9.368743718592966e-05,
-      "loss": 5.6676,
-      "step": 6783
-    },
-    {
-      "epoch": 3.5379400260756193,
-      "grad_norm": 1.4429258108139038,
-      "learning_rate": 9.368643216080403e-05,
-      "loss": 5.9515,
-      "step": 6784
-    },
-    {
-      "epoch": 3.5384615384615383,
-      "grad_norm": 1.5355881452560425,
-      "learning_rate": 9.36854271356784e-05,
-      "loss": 5.8846,
-      "step": 6785
-    },
-    {
-      "epoch": 3.538983050847458,
-      "grad_norm": 1.5289205312728882,
-      "learning_rate": 9.368442211055278e-05,
-      "loss": 6.0205,
-      "step": 6786
-    },
-    {
-      "epoch": 3.539504563233377,
-      "grad_norm": 1.4295034408569336,
-      "learning_rate": 9.368341708542714e-05,
-      "loss": 6.122,
-      "step": 6787
-    },
-    {
-      "epoch": 3.540026075619296,
-      "grad_norm": 1.6144932508468628,
-      "learning_rate": 9.368241206030152e-05,
-      "loss": 5.6577,
-      "step": 6788
-    },
-    {
-      "epoch": 3.5405475880052153,
-      "grad_norm": 1.5656191110610962,
-      "learning_rate": 9.368140703517588e-05,
-      "loss": 5.9539,
-      "step": 6789
-    },
-    {
-      "epoch": 3.5410691003911343,
-      "grad_norm": 1.4656474590301514,
-      "learning_rate": 9.368040201005026e-05,
-      "loss": 5.8874,
-      "step": 6790
-    },
-    {
-      "epoch": 3.5415906127770533,
-      "grad_norm": 1.5593163967132568,
-      "learning_rate": 9.367939698492462e-05,
-      "loss": 5.7532,
-      "step": 6791
-    },
-    {
-      "epoch": 3.542112125162973,
-      "grad_norm": 1.5778427124023438,
-      "learning_rate": 9.367839195979899e-05,
-      "loss": 6.2578,
-      "step": 6792
-    },
-    {
-      "epoch": 3.542633637548892,
-      "grad_norm": 1.5305495262145996,
-      "learning_rate": 9.367738693467337e-05,
-      "loss": 5.6411,
-      "step": 6793
-    },
-    {
-      "epoch": 3.543155149934811,
-      "grad_norm": 1.437207579612732,
-      "learning_rate": 9.367638190954774e-05,
-      "loss": 5.6023,
-      "step": 6794
-    },
-    {
-      "epoch": 3.5436766623207303,
-      "grad_norm": 1.6374677419662476,
-      "learning_rate": 9.367537688442212e-05,
-      "loss": 5.7315,
-      "step": 6795
-    },
-    {
-      "epoch": 3.5441981747066493,
-      "grad_norm": 1.4308102130889893,
-      "learning_rate": 9.367437185929649e-05,
-      "loss": 5.8646,
-      "step": 6796
-    },
-    {
-      "epoch": 3.5447196870925683,
-      "grad_norm": 1.4962522983551025,
-      "learning_rate": 9.367336683417086e-05,
-      "loss": 6.0144,
-      "step": 6797
-    },
-    {
-      "epoch": 3.545241199478488,
-      "grad_norm": 1.3831360340118408,
-      "learning_rate": 9.367236180904523e-05,
-      "loss": 5.9922,
-      "step": 6798
-    },
-    {
-      "epoch": 3.545762711864407,
-      "grad_norm": 1.5433286428451538,
-      "learning_rate": 9.36713567839196e-05,
-      "loss": 5.6331,
-      "step": 6799
-    },
-    {
-      "epoch": 3.546284224250326,
-      "grad_norm": 1.393429160118103,
-      "learning_rate": 9.367035175879397e-05,
-      "loss": 5.812,
-      "step": 6800
-    },
-    {
-      "epoch": 3.5468057366362453,
-      "grad_norm": 1.422887921333313,
-      "learning_rate": 9.366934673366835e-05,
-      "loss": 5.8153,
-      "step": 6801
-    },
-    {
-      "epoch": 3.5473272490221643,
-      "grad_norm": 1.4431641101837158,
-      "learning_rate": 9.366834170854271e-05,
-      "loss": 5.4828,
-      "step": 6802
-    },
-    {
-      "epoch": 3.5478487614080834,
-      "grad_norm": 1.4801768064498901,
-      "learning_rate": 9.366733668341709e-05,
-      "loss": 5.6167,
-      "step": 6803
-    },
-    {
-      "epoch": 3.548370273794003,
-      "grad_norm": 1.4948431253433228,
-      "learning_rate": 9.366633165829147e-05,
-      "loss": 5.7848,
-      "step": 6804
-    },
-    {
-      "epoch": 3.548891786179922,
-      "grad_norm": 1.540388822555542,
-      "learning_rate": 9.366532663316583e-05,
-      "loss": 5.5861,
-      "step": 6805
-    },
-    {
-      "epoch": 3.549413298565841,
-      "grad_norm": 1.3801501989364624,
-      "learning_rate": 9.366432160804021e-05,
-      "loss": 6.0132,
-      "step": 6806
-    },
-    {
-      "epoch": 3.5499348109517603,
-      "grad_norm": 1.4095205068588257,
-      "learning_rate": 9.366331658291457e-05,
-      "loss": 6.1008,
-      "step": 6807
-    },
-    {
-      "epoch": 3.5504563233376794,
-      "grad_norm": 1.5053099393844604,
-      "learning_rate": 9.366231155778895e-05,
-      "loss": 6.002,
-      "step": 6808
-    },
-    {
-      "epoch": 3.5509778357235984,
-      "grad_norm": 1.7069369554519653,
-      "learning_rate": 9.366130653266332e-05,
-      "loss": 5.2743,
-      "step": 6809
-    },
-    {
-      "epoch": 3.551499348109518,
-      "grad_norm": 1.5175182819366455,
-      "learning_rate": 9.36603015075377e-05,
-      "loss": 5.757,
-      "step": 6810
-    },
-    {
-      "epoch": 3.552020860495437,
-      "grad_norm": 1.5140146017074585,
-      "learning_rate": 9.365929648241206e-05,
-      "loss": 6.0598,
-      "step": 6811
-    },
-    {
-      "epoch": 3.552542372881356,
-      "grad_norm": 1.3476978540420532,
-      "learning_rate": 9.365829145728644e-05,
-      "loss": 5.9597,
-      "step": 6812
-    },
-    {
-      "epoch": 3.5530638852672753,
-      "grad_norm": 1.5037477016448975,
-      "learning_rate": 9.36572864321608e-05,
-      "loss": 5.5444,
-      "step": 6813
-    },
-    {
-      "epoch": 3.5535853976531944,
-      "grad_norm": 1.496700644493103,
-      "learning_rate": 9.365628140703518e-05,
-      "loss": 5.774,
-      "step": 6814
-    },
-    {
-      "epoch": 3.5541069100391134,
-      "grad_norm": 1.504744291305542,
-      "learning_rate": 9.365527638190956e-05,
-      "loss": 5.3697,
-      "step": 6815
-    },
-    {
-      "epoch": 3.554628422425033,
-      "grad_norm": 1.3509753942489624,
-      "learning_rate": 9.365427135678393e-05,
-      "loss": 5.8862,
-      "step": 6816
-    },
-    {
-      "epoch": 3.555149934810952,
-      "grad_norm": 1.4837826490402222,
-      "learning_rate": 9.36532663316583e-05,
-      "loss": 5.7488,
-      "step": 6817
-    },
-    {
-      "epoch": 3.555671447196871,
-      "grad_norm": 1.4449288845062256,
-      "learning_rate": 9.365226130653266e-05,
-      "loss": 5.8544,
-      "step": 6818
-    },
-    {
-      "epoch": 3.5561929595827904,
-      "grad_norm": 1.4951460361480713,
-      "learning_rate": 9.365125628140704e-05,
-      "loss": 6.1774,
-      "step": 6819
-    },
-    {
-      "epoch": 3.5567144719687094,
-      "grad_norm": 1.4116355180740356,
-      "learning_rate": 9.36502512562814e-05,
-      "loss": 6.0866,
-      "step": 6820
-    },
-    {
-      "epoch": 3.5572359843546284,
-      "grad_norm": 1.5097860097885132,
-      "learning_rate": 9.364924623115578e-05,
-      "loss": 5.7212,
-      "step": 6821
-    },
-    {
-      "epoch": 3.557757496740548,
-      "grad_norm": 1.4505469799041748,
-      "learning_rate": 9.364824120603015e-05,
-      "loss": 5.7694,
-      "step": 6822
-    },
-    {
-      "epoch": 3.558279009126467,
-      "grad_norm": 1.3850070238113403,
-      "learning_rate": 9.364723618090452e-05,
-      "loss": 6.0401,
-      "step": 6823
-    },
-    {
-      "epoch": 3.558800521512386,
-      "grad_norm": 1.5888214111328125,
-      "learning_rate": 9.36462311557789e-05,
-      "loss": 5.4822,
-      "step": 6824
-    },
-    {
-      "epoch": 3.559322033898305,
-      "grad_norm": 1.5971285104751587,
-      "learning_rate": 9.364522613065328e-05,
-      "loss": 5.9007,
-      "step": 6825
-    },
-    {
-      "epoch": 3.5598435462842244,
-      "grad_norm": 1.5771061182022095,
-      "learning_rate": 9.364422110552764e-05,
-      "loss": 5.8294,
-      "step": 6826
-    },
-    {
-      "epoch": 3.5603650586701434,
-      "grad_norm": 1.426080584526062,
-      "learning_rate": 9.364321608040202e-05,
-      "loss": 5.6197,
-      "step": 6827
-    },
-    {
-      "epoch": 3.5608865710560624,
-      "grad_norm": 1.4910591840744019,
-      "learning_rate": 9.364221105527639e-05,
-      "loss": 5.7713,
-      "step": 6828
-    },
-    {
-      "epoch": 3.561408083441982,
-      "grad_norm": 1.4725879430770874,
-      "learning_rate": 9.364120603015076e-05,
-      "loss": 5.4805,
-      "step": 6829
-    },
-    {
-      "epoch": 3.561929595827901,
-      "grad_norm": 1.406074047088623,
-      "learning_rate": 9.364020100502513e-05,
-      "loss": 5.8328,
-      "step": 6830
-    },
-    {
-      "epoch": 3.56245110821382,
-      "grad_norm": 1.671393632888794,
-      "learning_rate": 9.36391959798995e-05,
-      "loss": 5.7102,
-      "step": 6831
-    },
-    {
-      "epoch": 3.5629726205997394,
-      "grad_norm": 1.501727819442749,
-      "learning_rate": 9.363819095477387e-05,
-      "loss": 5.6989,
-      "step": 6832
-    },
-    {
-      "epoch": 3.5634941329856584,
-      "grad_norm": 1.5371509790420532,
-      "learning_rate": 9.363718592964824e-05,
-      "loss": 5.9144,
-      "step": 6833
-    },
-    {
-      "epoch": 3.5640156453715774,
-      "grad_norm": 1.4923715591430664,
-      "learning_rate": 9.363618090452261e-05,
-      "loss": 5.5012,
-      "step": 6834
-    },
-    {
-      "epoch": 3.5645371577574965,
-      "grad_norm": 1.4875831604003906,
-      "learning_rate": 9.363517587939699e-05,
-      "loss": 5.8962,
-      "step": 6835
-    },
-    {
-      "epoch": 3.565058670143416,
-      "grad_norm": 1.5725582838058472,
-      "learning_rate": 9.363417085427137e-05,
-      "loss": 5.7393,
-      "step": 6836
-    },
-    {
-      "epoch": 3.565580182529335,
-      "grad_norm": 1.466874599456787,
-      "learning_rate": 9.363316582914573e-05,
-      "loss": 5.6146,
-      "step": 6837
-    },
-    {
-      "epoch": 3.566101694915254,
-      "grad_norm": 1.498960018157959,
-      "learning_rate": 9.363216080402011e-05,
-      "loss": 5.607,
-      "step": 6838
-    },
-    {
-      "epoch": 3.5666232073011734,
-      "grad_norm": 1.6629704236984253,
-      "learning_rate": 9.363115577889448e-05,
-      "loss": 5.7833,
-      "step": 6839
-    },
-    {
-      "epoch": 3.5671447196870925,
-      "grad_norm": 1.5827151536941528,
-      "learning_rate": 9.363015075376885e-05,
-      "loss": 5.9138,
-      "step": 6840
-    },
-    {
-      "epoch": 3.5676662320730115,
-      "grad_norm": 1.405080795288086,
-      "learning_rate": 9.362914572864322e-05,
-      "loss": 6.0245,
-      "step": 6841
-    },
-    {
-      "epoch": 3.568187744458931,
-      "grad_norm": 1.703434944152832,
-      "learning_rate": 9.36281407035176e-05,
-      "loss": 5.7502,
-      "step": 6842
-    },
-    {
-      "epoch": 3.56870925684485,
-      "grad_norm": 1.3881704807281494,
-      "learning_rate": 9.362713567839196e-05,
-      "loss": 6.0196,
-      "step": 6843
-    },
-    {
-      "epoch": 3.569230769230769,
-      "grad_norm": 1.3772093057632446,
-      "learning_rate": 9.362613065326634e-05,
-      "loss": 5.6336,
-      "step": 6844
-    },
-    {
-      "epoch": 3.5697522816166884,
-      "grad_norm": 1.7843724489212036,
-      "learning_rate": 9.362512562814072e-05,
-      "loss": 5.5352,
-      "step": 6845
-    },
-    {
-      "epoch": 3.5702737940026075,
-      "grad_norm": 1.608591914176941,
-      "learning_rate": 9.362412060301508e-05,
-      "loss": 5.9506,
-      "step": 6846
-    },
-    {
-      "epoch": 3.5707953063885265,
-      "grad_norm": 1.453364610671997,
-      "learning_rate": 9.362311557788946e-05,
-      "loss": 5.8615,
-      "step": 6847
-    },
-    {
-      "epoch": 3.571316818774446,
-      "grad_norm": 1.5107251405715942,
-      "learning_rate": 9.362211055276382e-05,
-      "loss": 5.7107,
-      "step": 6848
-    },
-    {
-      "epoch": 3.571838331160365,
-      "grad_norm": 1.525380253791809,
-      "learning_rate": 9.36211055276382e-05,
-      "loss": 5.8206,
-      "step": 6849
-    },
-    {
-      "epoch": 3.572359843546284,
-      "grad_norm": 1.5033036470413208,
-      "learning_rate": 9.362010050251256e-05,
-      "loss": 5.3721,
-      "step": 6850
-    },
-    {
-      "epoch": 3.5728813559322035,
-      "grad_norm": 1.7732305526733398,
-      "learning_rate": 9.361909547738694e-05,
-      "loss": 5.4542,
-      "step": 6851
-    },
-    {
-      "epoch": 3.5734028683181225,
-      "grad_norm": 1.5028077363967896,
-      "learning_rate": 9.36180904522613e-05,
-      "loss": 4.9389,
-      "step": 6852
-    },
-    {
-      "epoch": 3.5739243807040415,
-      "grad_norm": 1.6341519355773926,
-      "learning_rate": 9.361708542713568e-05,
-      "loss": 5.7513,
-      "step": 6853
-    },
-    {
-      "epoch": 3.574445893089961,
-      "grad_norm": 1.5568238496780396,
-      "learning_rate": 9.361608040201005e-05,
-      "loss": 5.8738,
-      "step": 6854
-    },
-    {
-      "epoch": 3.57496740547588,
-      "grad_norm": 1.59091317653656,
-      "learning_rate": 9.361507537688443e-05,
-      "loss": 5.6566,
-      "step": 6855
-    },
-    {
-      "epoch": 3.575488917861799,
-      "grad_norm": 1.5449665784835815,
-      "learning_rate": 9.36140703517588e-05,
-      "loss": 5.7875,
-      "step": 6856
-    },
-    {
-      "epoch": 3.5760104302477185,
-      "grad_norm": 1.396132469177246,
-      "learning_rate": 9.361306532663318e-05,
-      "loss": 5.7099,
-      "step": 6857
-    },
-    {
-      "epoch": 3.5765319426336375,
-      "grad_norm": 1.68631112575531,
-      "learning_rate": 9.361206030150755e-05,
-      "loss": 5.9687,
-      "step": 6858
-    },
-    {
-      "epoch": 3.5770534550195565,
-      "grad_norm": 1.5919063091278076,
-      "learning_rate": 9.361105527638191e-05,
-      "loss": 5.6823,
-      "step": 6859
-    },
-    {
-      "epoch": 3.577574967405476,
-      "grad_norm": 1.4949712753295898,
-      "learning_rate": 9.361005025125629e-05,
-      "loss": 5.241,
-      "step": 6860
-    },
-    {
-      "epoch": 3.578096479791395,
-      "grad_norm": 1.6010483503341675,
-      "learning_rate": 9.360904522613065e-05,
-      "loss": 5.8532,
-      "step": 6861
-    },
-    {
-      "epoch": 3.578617992177314,
-      "grad_norm": 1.5974032878875732,
-      "learning_rate": 9.360804020100503e-05,
-      "loss": 6.2063,
-      "step": 6862
-    },
-    {
-      "epoch": 3.5791395045632335,
-      "grad_norm": 1.5263973474502563,
-      "learning_rate": 9.36070351758794e-05,
-      "loss": 5.7751,
-      "step": 6863
-    },
-    {
-      "epoch": 3.5796610169491525,
-      "grad_norm": 1.4749189615249634,
-      "learning_rate": 9.360603015075377e-05,
-      "loss": 5.9876,
-      "step": 6864
-    },
-    {
-      "epoch": 3.5801825293350715,
-      "grad_norm": 1.5189263820648193,
-      "learning_rate": 9.360502512562814e-05,
-      "loss": 6.0059,
-      "step": 6865
-    },
-    {
-      "epoch": 3.580704041720991,
-      "grad_norm": 1.3961609601974487,
-      "learning_rate": 9.360402010050251e-05,
-      "loss": 6.2034,
-      "step": 6866
-    },
-    {
-      "epoch": 3.58122555410691,
-      "grad_norm": 1.4077609777450562,
-      "learning_rate": 9.360301507537689e-05,
-      "loss": 5.9095,
-      "step": 6867
-    },
-    {
-      "epoch": 3.581747066492829,
-      "grad_norm": 1.5758384466171265,
-      "learning_rate": 9.360201005025127e-05,
-      "loss": 5.62,
-      "step": 6868
-    },
-    {
-      "epoch": 3.5822685788787485,
-      "grad_norm": 1.7166107892990112,
-      "learning_rate": 9.360100502512563e-05,
-      "loss": 5.2069,
-      "step": 6869
-    },
-    {
-      "epoch": 3.5827900912646675,
-      "grad_norm": 1.4898594617843628,
-      "learning_rate": 9.360000000000001e-05,
-      "loss": 5.6262,
-      "step": 6870
-    },
-    {
-      "epoch": 3.5833116036505865,
-      "grad_norm": 1.413983941078186,
-      "learning_rate": 9.359899497487438e-05,
-      "loss": 5.7946,
-      "step": 6871
-    },
-    {
-      "epoch": 3.583833116036506,
-      "grad_norm": 1.4677363634109497,
-      "learning_rate": 9.359798994974874e-05,
-      "loss": 5.7692,
-      "step": 6872
-    },
-    {
-      "epoch": 3.584354628422425,
-      "grad_norm": 1.4581385850906372,
-      "learning_rate": 9.359698492462312e-05,
-      "loss": 5.9509,
-      "step": 6873
-    },
-    {
-      "epoch": 3.584876140808344,
-      "grad_norm": 1.4012049436569214,
-      "learning_rate": 9.359597989949748e-05,
-      "loss": 5.1732,
-      "step": 6874
-    },
-    {
-      "epoch": 3.5853976531942635,
-      "grad_norm": 1.3268972635269165,
-      "learning_rate": 9.359497487437186e-05,
-      "loss": 5.7503,
-      "step": 6875
-    },
-    {
-      "epoch": 3.5859191655801825,
-      "grad_norm": 1.4763849973678589,
-      "learning_rate": 9.359396984924624e-05,
-      "loss": 6.001,
-      "step": 6876
-    },
-    {
-      "epoch": 3.5864406779661016,
-      "grad_norm": 1.4205474853515625,
-      "learning_rate": 9.359296482412062e-05,
-      "loss": 6.052,
-      "step": 6877
-    },
-    {
-      "epoch": 3.586962190352021,
-      "grad_norm": 1.562045693397522,
-      "learning_rate": 9.359195979899498e-05,
-      "loss": 5.8459,
-      "step": 6878
-    },
-    {
-      "epoch": 3.58748370273794,
-      "grad_norm": 1.4696348905563354,
-      "learning_rate": 9.359095477386936e-05,
-      "loss": 5.9648,
-      "step": 6879
-    },
-    {
-      "epoch": 3.588005215123859,
-      "grad_norm": 1.4018970727920532,
-      "learning_rate": 9.358994974874372e-05,
-      "loss": 5.7933,
-      "step": 6880
-    },
-    {
-      "epoch": 3.5885267275097785,
-      "grad_norm": 1.4879355430603027,
-      "learning_rate": 9.35889447236181e-05,
-      "loss": 5.6376,
-      "step": 6881
-    },
-    {
-      "epoch": 3.5890482398956975,
-      "grad_norm": 1.4292739629745483,
-      "learning_rate": 9.358793969849246e-05,
-      "loss": 6.1167,
-      "step": 6882
-    },
-    {
-      "epoch": 3.5895697522816166,
-      "grad_norm": 1.533405065536499,
-      "learning_rate": 9.358693467336684e-05,
-      "loss": 6.136,
-      "step": 6883
-    },
-    {
-      "epoch": 3.590091264667536,
-      "grad_norm": 1.3921878337860107,
-      "learning_rate": 9.35859296482412e-05,
-      "loss": 5.8444,
-      "step": 6884
-    },
-    {
-      "epoch": 3.590612777053455,
-      "grad_norm": 1.4075108766555786,
-      "learning_rate": 9.358492462311557e-05,
-      "loss": 5.9878,
-      "step": 6885
-    },
-    {
-      "epoch": 3.591134289439374,
-      "grad_norm": 1.324178695678711,
-      "learning_rate": 9.358391959798995e-05,
-      "loss": 6.1119,
-      "step": 6886
-    },
-    {
-      "epoch": 3.5916558018252935,
-      "grad_norm": 1.3948732614517212,
-      "learning_rate": 9.358291457286433e-05,
-      "loss": 5.9962,
-      "step": 6887
-    },
-    {
-      "epoch": 3.5921773142112126,
-      "grad_norm": 1.4392926692962646,
-      "learning_rate": 9.35819095477387e-05,
-      "loss": 5.9424,
-      "step": 6888
-    },
-    {
-      "epoch": 3.5926988265971316,
-      "grad_norm": 1.5435951948165894,
-      "learning_rate": 9.358090452261307e-05,
-      "loss": 5.6765,
-      "step": 6889
-    },
-    {
-      "epoch": 3.593220338983051,
-      "grad_norm": 1.6878023147583008,
-      "learning_rate": 9.357989949748745e-05,
-      "loss": 5.2532,
-      "step": 6890
-    },
-    {
-      "epoch": 3.59374185136897,
-      "grad_norm": 1.4812532663345337,
-      "learning_rate": 9.357889447236181e-05,
-      "loss": 5.6591,
-      "step": 6891
-    },
-    {
-      "epoch": 3.594263363754889,
-      "grad_norm": 1.420653223991394,
-      "learning_rate": 9.357788944723619e-05,
-      "loss": 5.7632,
-      "step": 6892
-    },
-    {
-      "epoch": 3.5947848761408085,
-      "grad_norm": 1.4115720987319946,
-      "learning_rate": 9.357688442211055e-05,
-      "loss": 5.798,
-      "step": 6893
-    },
-    {
-      "epoch": 3.5953063885267276,
-      "grad_norm": 1.5068799257278442,
-      "learning_rate": 9.357587939698493e-05,
-      "loss": 5.7412,
-      "step": 6894
-    },
-    {
-      "epoch": 3.5958279009126466,
-      "grad_norm": 1.433834195137024,
-      "learning_rate": 9.35748743718593e-05,
-      "loss": 6.0532,
-      "step": 6895
-    },
-    {
-      "epoch": 3.596349413298566,
-      "grad_norm": 1.4348320960998535,
-      "learning_rate": 9.357386934673367e-05,
-      "loss": 5.8764,
-      "step": 6896
-    },
-    {
-      "epoch": 3.596870925684485,
-      "grad_norm": 1.502423882484436,
-      "learning_rate": 9.357286432160805e-05,
-      "loss": 5.6252,
-      "step": 6897
-    },
-    {
-      "epoch": 3.597392438070404,
-      "grad_norm": 1.4934662580490112,
-      "learning_rate": 9.357185929648241e-05,
-      "loss": 5.7368,
-      "step": 6898
-    },
-    {
-      "epoch": 3.5979139504563236,
-      "grad_norm": 1.5547089576721191,
-      "learning_rate": 9.357085427135679e-05,
-      "loss": 5.0837,
-      "step": 6899
-    },
-    {
-      "epoch": 3.5984354628422426,
-      "grad_norm": 1.4599767923355103,
-      "learning_rate": 9.356984924623116e-05,
-      "loss": 5.6357,
-      "step": 6900
-    },
-    {
-      "epoch": 3.5989569752281616,
-      "grad_norm": 1.3981791734695435,
-      "learning_rate": 9.356884422110553e-05,
-      "loss": 5.6084,
-      "step": 6901
-    },
-    {
-      "epoch": 3.599478487614081,
-      "grad_norm": 1.4656355381011963,
-      "learning_rate": 9.35678391959799e-05,
-      "loss": 6.0239,
-      "step": 6902
-    },
-    {
-      "epoch": 3.6,
-      "grad_norm": 1.649858832359314,
-      "learning_rate": 9.356683417085428e-05,
-      "loss": 5.6213,
-      "step": 6903
-    },
-    {
-      "epoch": 3.600521512385919,
-      "grad_norm": 1.4889661073684692,
-      "learning_rate": 9.356582914572864e-05,
-      "loss": 5.7472,
-      "step": 6904
-    },
-    {
-      "epoch": 3.6010430247718386,
-      "grad_norm": 1.5171128511428833,
-      "learning_rate": 9.356482412060302e-05,
-      "loss": 6.0173,
-      "step": 6905
-    },
-    {
-      "epoch": 3.6015645371577576,
-      "grad_norm": 1.4820151329040527,
-      "learning_rate": 9.356381909547738e-05,
-      "loss": 5.6385,
-      "step": 6906
-    },
-    {
-      "epoch": 3.6020860495436766,
-      "grad_norm": 1.39638090133667,
-      "learning_rate": 9.356281407035176e-05,
-      "loss": 5.7745,
-      "step": 6907
-    },
-    {
-      "epoch": 3.602607561929596,
-      "grad_norm": 1.3798282146453857,
-      "learning_rate": 9.356180904522614e-05,
-      "loss": 5.9719,
-      "step": 6908
-    },
-    {
-      "epoch": 3.603129074315515,
-      "grad_norm": 1.4505019187927246,
-      "learning_rate": 9.356080402010052e-05,
-      "loss": 6.0753,
-      "step": 6909
-    },
-    {
-      "epoch": 3.603650586701434,
-      "grad_norm": 1.4945056438446045,
-      "learning_rate": 9.355979899497488e-05,
-      "loss": 5.6782,
-      "step": 6910
-    },
-    {
-      "epoch": 3.6041720990873536,
-      "grad_norm": 1.2827863693237305,
-      "learning_rate": 9.355879396984925e-05,
-      "loss": 6.128,
-      "step": 6911
-    },
-    {
-      "epoch": 3.6046936114732726,
-      "grad_norm": 1.4397509098052979,
-      "learning_rate": 9.355778894472362e-05,
-      "loss": 5.9843,
-      "step": 6912
-    },
-    {
-      "epoch": 3.6052151238591916,
-      "grad_norm": 1.3544766902923584,
-      "learning_rate": 9.355678391959799e-05,
-      "loss": 5.5331,
-      "step": 6913
-    },
-    {
-      "epoch": 3.605736636245111,
-      "grad_norm": 1.6406874656677246,
-      "learning_rate": 9.355577889447237e-05,
-      "loss": 5.7444,
-      "step": 6914
-    },
-    {
-      "epoch": 3.60625814863103,
-      "grad_norm": 1.4029700756072998,
-      "learning_rate": 9.355477386934673e-05,
-      "loss": 5.6262,
-      "step": 6915
-    },
-    {
-      "epoch": 3.606779661016949,
-      "grad_norm": 1.5530447959899902,
-      "learning_rate": 9.355376884422111e-05,
-      "loss": 6.0757,
-      "step": 6916
-    },
-    {
-      "epoch": 3.607301173402868,
-      "grad_norm": 1.4106826782226562,
-      "learning_rate": 9.355276381909549e-05,
-      "loss": 5.9752,
-      "step": 6917
-    },
-    {
-      "epoch": 3.6078226857887876,
-      "grad_norm": 1.4556238651275635,
-      "learning_rate": 9.355175879396986e-05,
-      "loss": 5.86,
-      "step": 6918
-    },
-    {
-      "epoch": 3.6083441981747066,
-      "grad_norm": 1.3775568008422852,
-      "learning_rate": 9.355075376884423e-05,
-      "loss": 5.8982,
-      "step": 6919
-    },
-    {
-      "epoch": 3.6088657105606257,
-      "grad_norm": 1.5369001626968384,
-      "learning_rate": 9.35497487437186e-05,
-      "loss": 5.4901,
-      "step": 6920
-    },
-    {
-      "epoch": 3.609387222946545,
-      "grad_norm": 1.3803353309631348,
-      "learning_rate": 9.354874371859297e-05,
-      "loss": 5.9655,
-      "step": 6921
-    },
-    {
-      "epoch": 3.609908735332464,
-      "grad_norm": 1.7061947584152222,
-      "learning_rate": 9.354773869346735e-05,
-      "loss": 5.7856,
-      "step": 6922
-    },
-    {
-      "epoch": 3.610430247718383,
-      "grad_norm": 1.4817575216293335,
-      "learning_rate": 9.354673366834171e-05,
-      "loss": 5.8383,
-      "step": 6923
-    },
-    {
-      "epoch": 3.6109517601043026,
-      "grad_norm": 1.4101464748382568,
-      "learning_rate": 9.354572864321608e-05,
-      "loss": 6.004,
-      "step": 6924
-    },
-    {
-      "epoch": 3.6114732724902217,
-      "grad_norm": 1.5287854671478271,
-      "learning_rate": 9.354472361809045e-05,
-      "loss": 6.1112,
-      "step": 6925
-    },
-    {
-      "epoch": 3.6119947848761407,
-      "grad_norm": 1.5323861837387085,
-      "learning_rate": 9.354371859296482e-05,
-      "loss": 5.6673,
-      "step": 6926
-    },
-    {
-      "epoch": 3.61251629726206,
-      "grad_norm": 1.5977439880371094,
-      "learning_rate": 9.35427135678392e-05,
-      "loss": 5.4685,
-      "step": 6927
-    },
-    {
-      "epoch": 3.613037809647979,
-      "grad_norm": 1.4773221015930176,
-      "learning_rate": 9.354170854271357e-05,
-      "loss": 5.8095,
-      "step": 6928
-    },
-    {
-      "epoch": 3.613559322033898,
-      "grad_norm": 1.5083141326904297,
-      "learning_rate": 9.354070351758795e-05,
-      "loss": 5.8477,
-      "step": 6929
-    },
-    {
-      "epoch": 3.614080834419817,
-      "grad_norm": 1.474170446395874,
-      "learning_rate": 9.353969849246232e-05,
-      "loss": 5.9531,
-      "step": 6930
-    },
-    {
-      "epoch": 3.6146023468057367,
-      "grad_norm": 1.3986929655075073,
-      "learning_rate": 9.35386934673367e-05,
-      "loss": 5.7439,
-      "step": 6931
-    },
-    {
-      "epoch": 3.6151238591916557,
-      "grad_norm": 1.4565824270248413,
-      "learning_rate": 9.353768844221106e-05,
-      "loss": 5.8178,
-      "step": 6932
-    },
-    {
-      "epoch": 3.6156453715775747,
-      "grad_norm": 1.601528286933899,
-      "learning_rate": 9.353668341708544e-05,
-      "loss": 5.5391,
-      "step": 6933
-    },
-    {
-      "epoch": 3.616166883963494,
-      "grad_norm": 1.6320865154266357,
-      "learning_rate": 9.35356783919598e-05,
-      "loss": 5.8251,
-      "step": 6934
-    },
-    {
-      "epoch": 3.616688396349413,
-      "grad_norm": 1.5504592657089233,
-      "learning_rate": 9.353467336683418e-05,
-      "loss": 5.7701,
-      "step": 6935
-    },
-    {
-      "epoch": 3.617209908735332,
-      "grad_norm": 1.602881908416748,
-      "learning_rate": 9.353366834170854e-05,
-      "loss": 5.8008,
-      "step": 6936
-    },
-    {
-      "epoch": 3.6177314211212517,
-      "grad_norm": 1.4159990549087524,
-      "learning_rate": 9.353266331658292e-05,
-      "loss": 6.053,
-      "step": 6937
-    },
-    {
-      "epoch": 3.6182529335071707,
-      "grad_norm": 1.5103563070297241,
-      "learning_rate": 9.35316582914573e-05,
-      "loss": 6.0597,
-      "step": 6938
-    },
-    {
-      "epoch": 3.6187744458930897,
-      "grad_norm": 1.4954168796539307,
-      "learning_rate": 9.353065326633166e-05,
-      "loss": 5.4636,
-      "step": 6939
-    },
-    {
-      "epoch": 3.619295958279009,
-      "grad_norm": 1.426872968673706,
-      "learning_rate": 9.352964824120604e-05,
-      "loss": 6.1467,
-      "step": 6940
-    },
-    {
-      "epoch": 3.619817470664928,
-      "grad_norm": 1.646280288696289,
-      "learning_rate": 9.35286432160804e-05,
-      "loss": 5.3514,
-      "step": 6941
-    },
-    {
-      "epoch": 3.6203389830508472,
-      "grad_norm": 1.4335969686508179,
-      "learning_rate": 9.352763819095478e-05,
-      "loss": 6.1869,
-      "step": 6942
-    },
-    {
-      "epoch": 3.6208604954367667,
-      "grad_norm": 1.4257240295410156,
-      "learning_rate": 9.352663316582915e-05,
-      "loss": 6.0428,
-      "step": 6943
-    },
-    {
-      "epoch": 3.6213820078226857,
-      "grad_norm": 1.5464763641357422,
-      "learning_rate": 9.352562814070352e-05,
-      "loss": 5.5642,
-      "step": 6944
-    },
-    {
-      "epoch": 3.6219035202086047,
-      "grad_norm": 1.3919329643249512,
-      "learning_rate": 9.352462311557789e-05,
-      "loss": 6.0843,
-      "step": 6945
-    },
-    {
-      "epoch": 3.622425032594524,
-      "grad_norm": 1.4617398977279663,
-      "learning_rate": 9.352361809045227e-05,
-      "loss": 5.7949,
-      "step": 6946
-    },
-    {
-      "epoch": 3.622946544980443,
-      "grad_norm": 1.5851168632507324,
-      "learning_rate": 9.352261306532663e-05,
-      "loss": 5.9043,
-      "step": 6947
-    },
-    {
-      "epoch": 3.6234680573663622,
-      "grad_norm": 1.373176097869873,
-      "learning_rate": 9.352160804020101e-05,
-      "loss": 5.7437,
-      "step": 6948
-    },
-    {
-      "epoch": 3.6239895697522817,
-      "grad_norm": 1.6997206211090088,
-      "learning_rate": 9.352060301507539e-05,
-      "loss": 5.6051,
-      "step": 6949
-    },
-    {
-      "epoch": 3.6245110821382007,
-      "grad_norm": 1.4387316703796387,
-      "learning_rate": 9.351959798994976e-05,
-      "loss": 6.0871,
-      "step": 6950
-    },
-    {
-      "epoch": 3.6250325945241197,
-      "grad_norm": 1.4237005710601807,
-      "learning_rate": 9.351859296482413e-05,
-      "loss": 6.025,
-      "step": 6951
-    },
-    {
-      "epoch": 3.625554106910039,
-      "grad_norm": 1.6108957529067993,
-      "learning_rate": 9.351758793969849e-05,
-      "loss": 5.8635,
-      "step": 6952
-    },
-    {
-      "epoch": 3.6260756192959582,
-      "grad_norm": 1.6192713975906372,
-      "learning_rate": 9.351658291457287e-05,
-      "loss": 5.932,
-      "step": 6953
-    },
-    {
-      "epoch": 3.6265971316818773,
-      "grad_norm": 1.6241966485977173,
-      "learning_rate": 9.351557788944723e-05,
-      "loss": 5.3448,
-      "step": 6954
-    },
-    {
-      "epoch": 3.6271186440677967,
-      "grad_norm": 1.5147408246994019,
-      "learning_rate": 9.351457286432161e-05,
-      "loss": 5.8407,
-      "step": 6955
-    },
-    {
-      "epoch": 3.6276401564537157,
-      "grad_norm": 1.494148850440979,
-      "learning_rate": 9.351356783919598e-05,
-      "loss": 5.9548,
-      "step": 6956
-    },
-    {
-      "epoch": 3.6281616688396348,
-      "grad_norm": 1.5650743246078491,
-      "learning_rate": 9.351256281407035e-05,
-      "loss": 5.8892,
-      "step": 6957
-    },
-    {
-      "epoch": 3.628683181225554,
-      "grad_norm": 1.5746864080429077,
-      "learning_rate": 9.351155778894473e-05,
-      "loss": 5.5157,
-      "step": 6958
-    },
-    {
-      "epoch": 3.6292046936114732,
-      "grad_norm": 1.6509934663772583,
-      "learning_rate": 9.351055276381911e-05,
-      "loss": 5.3898,
-      "step": 6959
-    },
-    {
-      "epoch": 3.6297262059973923,
-      "grad_norm": 1.4974578619003296,
-      "learning_rate": 9.350954773869347e-05,
-      "loss": 5.8562,
-      "step": 6960
-    },
-    {
-      "epoch": 3.6302477183833117,
-      "grad_norm": 1.46342933177948,
-      "learning_rate": 9.350854271356785e-05,
-      "loss": 5.8228,
-      "step": 6961
-    },
-    {
-      "epoch": 3.6307692307692307,
-      "grad_norm": 1.5128607749938965,
-      "learning_rate": 9.350753768844222e-05,
-      "loss": 5.7642,
-      "step": 6962
-    },
-    {
-      "epoch": 3.6312907431551498,
-      "grad_norm": 1.3214689493179321,
-      "learning_rate": 9.35065326633166e-05,
-      "loss": 5.9902,
-      "step": 6963
-    },
-    {
-      "epoch": 3.6318122555410692,
-      "grad_norm": 1.5082287788391113,
-      "learning_rate": 9.350552763819096e-05,
-      "loss": 5.4666,
-      "step": 6964
-    },
-    {
-      "epoch": 3.6323337679269883,
-      "grad_norm": 1.5935252904891968,
-      "learning_rate": 9.350452261306532e-05,
-      "loss": 5.7479,
-      "step": 6965
-    },
-    {
-      "epoch": 3.6328552803129073,
-      "grad_norm": 1.4112643003463745,
-      "learning_rate": 9.35035175879397e-05,
-      "loss": 5.9646,
-      "step": 6966
-    },
-    {
-      "epoch": 3.6333767926988267,
-      "grad_norm": 1.51386559009552,
-      "learning_rate": 9.350251256281406e-05,
-      "loss": 5.7332,
-      "step": 6967
-    },
-    {
-      "epoch": 3.6338983050847458,
-      "grad_norm": 1.3717272281646729,
-      "learning_rate": 9.350150753768844e-05,
-      "loss": 6.0705,
-      "step": 6968
-    },
-    {
-      "epoch": 3.634419817470665,
-      "grad_norm": 1.387502670288086,
-      "learning_rate": 9.350050251256282e-05,
-      "loss": 5.9247,
-      "step": 6969
-    },
-    {
-      "epoch": 3.6349413298565842,
-      "grad_norm": 1.6512454748153687,
-      "learning_rate": 9.34994974874372e-05,
-      "loss": 5.5091,
-      "step": 6970
-    },
-    {
-      "epoch": 3.6354628422425033,
-      "grad_norm": 1.5581622123718262,
-      "learning_rate": 9.349849246231156e-05,
-      "loss": 5.4997,
-      "step": 6971
-    },
-    {
-      "epoch": 3.6359843546284223,
-      "grad_norm": 1.5530822277069092,
-      "learning_rate": 9.349748743718594e-05,
-      "loss": 5.696,
-      "step": 6972
-    },
-    {
-      "epoch": 3.6365058670143418,
-      "grad_norm": 1.4497745037078857,
-      "learning_rate": 9.34964824120603e-05,
-      "loss": 5.6685,
-      "step": 6973
-    },
-    {
-      "epoch": 3.6370273794002608,
-      "grad_norm": 1.4639405012130737,
-      "learning_rate": 9.349547738693468e-05,
-      "loss": 5.6884,
-      "step": 6974
-    },
-    {
-      "epoch": 3.63754889178618,
-      "grad_norm": 1.4224026203155518,
-      "learning_rate": 9.349447236180905e-05,
-      "loss": 5.7743,
-      "step": 6975
-    },
-    {
-      "epoch": 3.6380704041720993,
-      "grad_norm": 1.544215202331543,
-      "learning_rate": 9.349346733668342e-05,
-      "loss": 5.6067,
-      "step": 6976
-    },
-    {
-      "epoch": 3.6385919165580183,
-      "grad_norm": 1.4786514043807983,
-      "learning_rate": 9.349246231155779e-05,
-      "loss": 5.7619,
-      "step": 6977
-    },
-    {
-      "epoch": 3.6391134289439373,
-      "grad_norm": 1.4969890117645264,
-      "learning_rate": 9.349145728643217e-05,
-      "loss": 6.003,
-      "step": 6978
-    },
-    {
-      "epoch": 3.6396349413298568,
-      "grad_norm": 1.5170743465423584,
-      "learning_rate": 9.349045226130654e-05,
-      "loss": 5.7452,
-      "step": 6979
-    },
-    {
-      "epoch": 3.640156453715776,
-      "grad_norm": 1.6823762655258179,
-      "learning_rate": 9.348944723618091e-05,
-      "loss": 5.653,
-      "step": 6980
-    },
-    {
-      "epoch": 3.640677966101695,
-      "grad_norm": 1.380423665046692,
-      "learning_rate": 9.348844221105529e-05,
-      "loss": 5.0421,
-      "step": 6981
-    },
-    {
-      "epoch": 3.6411994784876143,
-      "grad_norm": 1.5066813230514526,
-      "learning_rate": 9.348743718592965e-05,
-      "loss": 5.7063,
-      "step": 6982
-    },
-    {
-      "epoch": 3.6417209908735333,
-      "grad_norm": 1.5106076002120972,
-      "learning_rate": 9.348643216080403e-05,
-      "loss": 5.8842,
-      "step": 6983
-    },
-    {
-      "epoch": 3.6422425032594523,
-      "grad_norm": 1.4765019416809082,
-      "learning_rate": 9.348542713567839e-05,
-      "loss": 5.7175,
-      "step": 6984
-    },
-    {
-      "epoch": 3.6427640156453718,
-      "grad_norm": 1.6228536367416382,
-      "learning_rate": 9.348442211055277e-05,
-      "loss": 5.2125,
-      "step": 6985
-    },
-    {
-      "epoch": 3.643285528031291,
-      "grad_norm": 1.3125760555267334,
-      "learning_rate": 9.348341708542714e-05,
-      "loss": 6.1559,
-      "step": 6986
-    },
-    {
-      "epoch": 3.64380704041721,
-      "grad_norm": 1.5719709396362305,
-      "learning_rate": 9.348241206030151e-05,
-      "loss": 5.8974,
-      "step": 6987
-    },
-    {
-      "epoch": 3.6443285528031293,
-      "grad_norm": 1.5354093313217163,
-      "learning_rate": 9.348140703517588e-05,
-      "loss": 5.4638,
-      "step": 6988
-    },
-    {
-      "epoch": 3.6448500651890483,
-      "grad_norm": 1.613311529159546,
-      "learning_rate": 9.348040201005026e-05,
-      "loss": 5.4519,
-      "step": 6989
-    },
-    {
-      "epoch": 3.6453715775749673,
-      "grad_norm": 1.497916579246521,
-      "learning_rate": 9.347939698492463e-05,
-      "loss": 5.4487,
-      "step": 6990
-    },
-    {
-      "epoch": 3.645893089960887,
-      "grad_norm": 1.4887266159057617,
-      "learning_rate": 9.3478391959799e-05,
-      "loss": 5.6188,
-      "step": 6991
-    },
-    {
-      "epoch": 3.646414602346806,
-      "grad_norm": 1.5240713357925415,
-      "learning_rate": 9.347738693467338e-05,
-      "loss": 5.9876,
-      "step": 6992
-    },
-    {
-      "epoch": 3.646936114732725,
-      "grad_norm": 1.4290990829467773,
-      "learning_rate": 9.347638190954774e-05,
-      "loss": 5.8446,
-      "step": 6993
-    },
-    {
-      "epoch": 3.6474576271186443,
-      "grad_norm": 1.5863919258117676,
-      "learning_rate": 9.347537688442212e-05,
-      "loss": 5.6336,
-      "step": 6994
-    },
-    {
-      "epoch": 3.6479791395045633,
-      "grad_norm": 1.3621937036514282,
-      "learning_rate": 9.347437185929648e-05,
-      "loss": 6.2207,
-      "step": 6995
-    },
-    {
-      "epoch": 3.6485006518904823,
-      "grad_norm": 1.3958585262298584,
-      "learning_rate": 9.347336683417086e-05,
-      "loss": 6.0533,
-      "step": 6996
-    },
-    {
-      "epoch": 3.649022164276402,
-      "grad_norm": 1.386458158493042,
-      "learning_rate": 9.347236180904522e-05,
-      "loss": 5.9849,
-      "step": 6997
-    },
-    {
-      "epoch": 3.649543676662321,
-      "grad_norm": 1.4610226154327393,
-      "learning_rate": 9.34713567839196e-05,
-      "loss": 5.4203,
-      "step": 6998
-    },
-    {
-      "epoch": 3.65006518904824,
-      "grad_norm": 1.4660662412643433,
-      "learning_rate": 9.347035175879398e-05,
-      "loss": 5.7817,
-      "step": 6999
-    },
-    {
-      "epoch": 3.6505867014341593,
-      "grad_norm": 1.5710549354553223,
-      "learning_rate": 9.346934673366836e-05,
-      "loss": 5.1513,
-      "step": 7000
-    },
-    {
-      "epoch": 3.6511082138200783,
-      "grad_norm": 1.4879204034805298,
-      "learning_rate": 9.346834170854272e-05,
-      "loss": 5.897,
-      "step": 7001
-    },
-    {
-      "epoch": 3.6516297262059974,
-      "grad_norm": 1.4350230693817139,
-      "learning_rate": 9.34673366834171e-05,
-      "loss": 5.8321,
-      "step": 7002
-    },
-    {
-      "epoch": 3.652151238591917,
-      "grad_norm": 1.4425678253173828,
-      "learning_rate": 9.346633165829146e-05,
-      "loss": 5.7894,
-      "step": 7003
-    },
-    {
-      "epoch": 3.652672750977836,
-      "grad_norm": 1.3172986507415771,
-      "learning_rate": 9.346532663316583e-05,
-      "loss": 5.9681,
-      "step": 7004
-    },
-    {
-      "epoch": 3.653194263363755,
-      "grad_norm": 1.5918426513671875,
-      "learning_rate": 9.34643216080402e-05,
-      "loss": 5.0853,
-      "step": 7005
-    },
-    {
-      "epoch": 3.6537157757496743,
-      "grad_norm": 1.641069769859314,
-      "learning_rate": 9.346331658291457e-05,
-      "loss": 5.8079,
-      "step": 7006
-    },
-    {
-      "epoch": 3.6542372881355933,
-      "grad_norm": 1.4998801946640015,
-      "learning_rate": 9.346231155778895e-05,
-      "loss": 5.9092,
-      "step": 7007
-    },
-    {
-      "epoch": 3.6547588005215124,
-      "grad_norm": 1.6488269567489624,
-      "learning_rate": 9.346130653266331e-05,
-      "loss": 5.709,
-      "step": 7008
-    },
-    {
-      "epoch": 3.655280312907432,
-      "grad_norm": 1.4858925342559814,
-      "learning_rate": 9.346030150753769e-05,
-      "loss": 5.5682,
-      "step": 7009
-    },
-    {
-      "epoch": 3.655801825293351,
-      "grad_norm": 1.4368996620178223,
-      "learning_rate": 9.345929648241207e-05,
-      "loss": 5.9772,
-      "step": 7010
-    },
-    {
-      "epoch": 3.65632333767927,
-      "grad_norm": 1.4265004396438599,
-      "learning_rate": 9.345829145728645e-05,
-      "loss": 4.7863,
-      "step": 7011
-    },
-    {
-      "epoch": 3.656844850065189,
-      "grad_norm": 1.5192221403121948,
-      "learning_rate": 9.345728643216081e-05,
-      "loss": 5.8857,
-      "step": 7012
-    },
-    {
-      "epoch": 3.6573663624511084,
-      "grad_norm": 1.505833625793457,
-      "learning_rate": 9.345628140703519e-05,
-      "loss": 6.1111,
-      "step": 7013
-    },
-    {
-      "epoch": 3.6578878748370274,
-      "grad_norm": 1.5164717435836792,
-      "learning_rate": 9.345527638190955e-05,
-      "loss": 5.1709,
-      "step": 7014
-    },
-    {
-      "epoch": 3.6584093872229464,
-      "grad_norm": 1.4406355619430542,
-      "learning_rate": 9.345427135678393e-05,
-      "loss": 5.8385,
-      "step": 7015
-    },
-    {
-      "epoch": 3.658930899608866,
-      "grad_norm": 1.5299161672592163,
-      "learning_rate": 9.34532663316583e-05,
-      "loss": 5.6954,
-      "step": 7016
-    },
-    {
-      "epoch": 3.659452411994785,
-      "grad_norm": 1.488677740097046,
-      "learning_rate": 9.345226130653266e-05,
-      "loss": 5.8737,
-      "step": 7017
-    },
-    {
-      "epoch": 3.659973924380704,
-      "grad_norm": 1.5530858039855957,
-      "learning_rate": 9.345125628140704e-05,
-      "loss": 5.8156,
-      "step": 7018
-    },
-    {
-      "epoch": 3.6604954367666234,
-      "grad_norm": 1.3230679035186768,
-      "learning_rate": 9.345025125628141e-05,
-      "loss": 6.2401,
-      "step": 7019
-    },
-    {
-      "epoch": 3.6610169491525424,
-      "grad_norm": 1.3270398378372192,
-      "learning_rate": 9.344924623115579e-05,
-      "loss": 6.1277,
-      "step": 7020
-    },
-    {
-      "epoch": 3.6615384615384614,
-      "grad_norm": 1.384616732597351,
-      "learning_rate": 9.344824120603016e-05,
-      "loss": 5.9564,
-      "step": 7021
-    },
-    {
-      "epoch": 3.6620599739243804,
-      "grad_norm": 1.5136234760284424,
-      "learning_rate": 9.344723618090453e-05,
-      "loss": 5.829,
-      "step": 7022
-    },
-    {
-      "epoch": 3.6625814863103,
-      "grad_norm": 1.5446760654449463,
-      "learning_rate": 9.34462311557789e-05,
-      "loss": 5.339,
-      "step": 7023
-    },
-    {
-      "epoch": 3.663102998696219,
-      "grad_norm": 1.3896344900131226,
-      "learning_rate": 9.344522613065328e-05,
-      "loss": 6.0225,
-      "step": 7024
-    },
-    {
-      "epoch": 3.663624511082138,
-      "grad_norm": 1.495092749595642,
-      "learning_rate": 9.344422110552764e-05,
-      "loss": 5.5607,
-      "step": 7025
-    },
-    {
-      "epoch": 3.6641460234680574,
-      "grad_norm": 1.6290724277496338,
-      "learning_rate": 9.344321608040202e-05,
-      "loss": 5.3952,
-      "step": 7026
-    },
-    {
-      "epoch": 3.6646675358539764,
-      "grad_norm": 1.4316827058792114,
-      "learning_rate": 9.344221105527638e-05,
-      "loss": 5.9642,
-      "step": 7027
-    },
-    {
-      "epoch": 3.6651890482398954,
-      "grad_norm": 1.439576268196106,
-      "learning_rate": 9.344120603015076e-05,
-      "loss": 5.4803,
-      "step": 7028
-    },
-    {
-      "epoch": 3.665710560625815,
-      "grad_norm": 1.4314627647399902,
-      "learning_rate": 9.344020100502512e-05,
-      "loss": 5.7772,
-      "step": 7029
-    },
-    {
-      "epoch": 3.666232073011734,
-      "grad_norm": 1.5099279880523682,
-      "learning_rate": 9.34391959798995e-05,
-      "loss": 5.6698,
-      "step": 7030
-    },
-    {
-      "epoch": 3.666753585397653,
-      "grad_norm": 1.5364707708358765,
-      "learning_rate": 9.343819095477388e-05,
-      "loss": 5.5044,
-      "step": 7031
-    },
-    {
-      "epoch": 3.6672750977835724,
-      "grad_norm": 1.4501142501831055,
-      "learning_rate": 9.343718592964824e-05,
-      "loss": 5.6192,
-      "step": 7032
-    },
-    {
-      "epoch": 3.6677966101694914,
-      "grad_norm": 1.4350626468658447,
-      "learning_rate": 9.343618090452262e-05,
-      "loss": 5.3731,
-      "step": 7033
-    },
-    {
-      "epoch": 3.6683181225554105,
-      "grad_norm": 1.4982316493988037,
-      "learning_rate": 9.343517587939699e-05,
-      "loss": 5.8832,
-      "step": 7034
-    },
-    {
-      "epoch": 3.66883963494133,
-      "grad_norm": 1.384792685508728,
-      "learning_rate": 9.343417085427136e-05,
-      "loss": 5.831,
-      "step": 7035
-    },
-    {
-      "epoch": 3.669361147327249,
-      "grad_norm": 1.438024640083313,
-      "learning_rate": 9.343316582914573e-05,
-      "loss": 5.4215,
-      "step": 7036
-    },
-    {
-      "epoch": 3.669882659713168,
-      "grad_norm": 1.4174267053604126,
-      "learning_rate": 9.34321608040201e-05,
-      "loss": 5.9711,
-      "step": 7037
-    },
-    {
-      "epoch": 3.6704041720990874,
-      "grad_norm": 1.354045033454895,
-      "learning_rate": 9.343115577889447e-05,
-      "loss": 6.183,
-      "step": 7038
-    },
-    {
-      "epoch": 3.6709256844850064,
-      "grad_norm": 1.460918664932251,
-      "learning_rate": 9.343015075376885e-05,
-      "loss": 6.2018,
-      "step": 7039
-    },
-    {
-      "epoch": 3.6714471968709255,
-      "grad_norm": 1.8349027633666992,
-      "learning_rate": 9.342914572864321e-05,
-      "loss": 5.0873,
-      "step": 7040
-    },
-    {
-      "epoch": 3.671968709256845,
-      "grad_norm": 1.5097715854644775,
-      "learning_rate": 9.342814070351759e-05,
-      "loss": 6.0222,
-      "step": 7041
-    },
-    {
-      "epoch": 3.672490221642764,
-      "grad_norm": 1.5525531768798828,
-      "learning_rate": 9.342713567839197e-05,
-      "loss": 5.6614,
-      "step": 7042
-    },
-    {
-      "epoch": 3.673011734028683,
-      "grad_norm": 1.4713733196258545,
-      "learning_rate": 9.342613065326635e-05,
-      "loss": 5.3081,
-      "step": 7043
-    },
-    {
-      "epoch": 3.6735332464146024,
-      "grad_norm": 1.4880425930023193,
-      "learning_rate": 9.342512562814071e-05,
-      "loss": 5.7003,
-      "step": 7044
-    },
-    {
-      "epoch": 3.6740547588005215,
-      "grad_norm": 1.4369738101959229,
-      "learning_rate": 9.342412060301507e-05,
-      "loss": 5.4863,
-      "step": 7045
-    },
-    {
-      "epoch": 3.6745762711864405,
-      "grad_norm": 1.4913148880004883,
-      "learning_rate": 9.342311557788945e-05,
-      "loss": 5.5362,
-      "step": 7046
-    },
-    {
-      "epoch": 3.67509778357236,
-      "grad_norm": 1.344566822052002,
-      "learning_rate": 9.342211055276382e-05,
-      "loss": 6.1305,
-      "step": 7047
-    },
-    {
-      "epoch": 3.675619295958279,
-      "grad_norm": 1.4426178932189941,
-      "learning_rate": 9.34211055276382e-05,
-      "loss": 4.6851,
-      "step": 7048
-    },
-    {
-      "epoch": 3.676140808344198,
-      "grad_norm": 1.4049689769744873,
-      "learning_rate": 9.342010050251256e-05,
-      "loss": 5.5067,
-      "step": 7049
-    },
-    {
-      "epoch": 3.6766623207301175,
-      "grad_norm": 1.4476603269577026,
-      "learning_rate": 9.341909547738694e-05,
-      "loss": 5.8957,
-      "step": 7050
-    },
-    {
-      "epoch": 3.6771838331160365,
-      "grad_norm": 1.3590470552444458,
-      "learning_rate": 9.341809045226131e-05,
-      "loss": 6.0658,
-      "step": 7051
-    },
-    {
-      "epoch": 3.6777053455019555,
-      "grad_norm": 1.5423229932785034,
-      "learning_rate": 9.341708542713569e-05,
-      "loss": 5.4093,
-      "step": 7052
-    },
-    {
-      "epoch": 3.678226857887875,
-      "grad_norm": 1.639768362045288,
-      "learning_rate": 9.341608040201006e-05,
-      "loss": 5.525,
-      "step": 7053
-    },
-    {
-      "epoch": 3.678748370273794,
-      "grad_norm": 1.4362624883651733,
-      "learning_rate": 9.341507537688443e-05,
-      "loss": 5.879,
-      "step": 7054
-    },
-    {
-      "epoch": 3.679269882659713,
-      "grad_norm": 1.3900130987167358,
-      "learning_rate": 9.34140703517588e-05,
-      "loss": 5.7807,
-      "step": 7055
-    },
-    {
-      "epoch": 3.6797913950456325,
-      "grad_norm": 1.4527782201766968,
-      "learning_rate": 9.341306532663318e-05,
-      "loss": 4.9261,
-      "step": 7056
-    },
-    {
-      "epoch": 3.6803129074315515,
-      "grad_norm": 1.4593982696533203,
-      "learning_rate": 9.341206030150754e-05,
-      "loss": 5.9209,
-      "step": 7057
-    },
-    {
-      "epoch": 3.6808344198174705,
-      "grad_norm": 1.4559870958328247,
-      "learning_rate": 9.34110552763819e-05,
-      "loss": 5.8339,
-      "step": 7058
-    },
-    {
-      "epoch": 3.68135593220339,
-      "grad_norm": 1.3414433002471924,
-      "learning_rate": 9.341005025125628e-05,
-      "loss": 5.5587,
-      "step": 7059
-    },
-    {
-      "epoch": 3.681877444589309,
-      "grad_norm": 1.4395246505737305,
-      "learning_rate": 9.340904522613065e-05,
-      "loss": 5.7092,
-      "step": 7060
-    },
-    {
-      "epoch": 3.682398956975228,
-      "grad_norm": 1.5973732471466064,
-      "learning_rate": 9.340804020100503e-05,
-      "loss": 5.2823,
-      "step": 7061
-    },
-    {
-      "epoch": 3.6829204693611475,
-      "grad_norm": 1.5057066679000854,
-      "learning_rate": 9.34070351758794e-05,
-      "loss": 5.2951,
-      "step": 7062
-    },
-    {
-      "epoch": 3.6834419817470665,
-      "grad_norm": 1.3808677196502686,
-      "learning_rate": 9.340603015075378e-05,
-      "loss": 5.8972,
-      "step": 7063
-    },
-    {
-      "epoch": 3.6839634941329855,
-      "grad_norm": 1.3882790803909302,
-      "learning_rate": 9.340502512562814e-05,
-      "loss": 5.8086,
-      "step": 7064
-    },
-    {
-      "epoch": 3.684485006518905,
-      "grad_norm": 1.5697991847991943,
-      "learning_rate": 9.340402010050252e-05,
-      "loss": 5.7429,
-      "step": 7065
-    },
-    {
-      "epoch": 3.685006518904824,
-      "grad_norm": 1.5416051149368286,
-      "learning_rate": 9.340301507537689e-05,
-      "loss": 5.687,
-      "step": 7066
-    },
-    {
-      "epoch": 3.685528031290743,
-      "grad_norm": 1.4893791675567627,
-      "learning_rate": 9.340201005025126e-05,
-      "loss": 5.7826,
-      "step": 7067
-    },
-    {
-      "epoch": 3.6860495436766625,
-      "grad_norm": 1.418459177017212,
-      "learning_rate": 9.340100502512563e-05,
-      "loss": 5.9258,
-      "step": 7068
-    },
-    {
-      "epoch": 3.6865710560625815,
-      "grad_norm": 1.4895086288452148,
-      "learning_rate": 9.340000000000001e-05,
-      "loss": 5.4678,
-      "step": 7069
-    },
-    {
-      "epoch": 3.6870925684485005,
-      "grad_norm": 1.5759357213974,
-      "learning_rate": 9.339899497487437e-05,
-      "loss": 5.8896,
-      "step": 7070
-    },
-    {
-      "epoch": 3.68761408083442,
-      "grad_norm": 1.6505433320999146,
-      "learning_rate": 9.339798994974875e-05,
-      "loss": 5.4788,
-      "step": 7071
-    },
-    {
-      "epoch": 3.688135593220339,
-      "grad_norm": 1.3575966358184814,
-      "learning_rate": 9.339698492462313e-05,
-      "loss": 6.0762,
-      "step": 7072
-    },
-    {
-      "epoch": 3.688657105606258,
-      "grad_norm": 1.289697289466858,
-      "learning_rate": 9.339597989949749e-05,
-      "loss": 5.3123,
-      "step": 7073
-    },
-    {
-      "epoch": 3.6891786179921775,
-      "grad_norm": 1.7307019233703613,
-      "learning_rate": 9.339497487437187e-05,
-      "loss": 5.3798,
-      "step": 7074
-    },
-    {
-      "epoch": 3.6897001303780965,
-      "grad_norm": 1.558272123336792,
-      "learning_rate": 9.339396984924623e-05,
-      "loss": 5.5509,
-      "step": 7075
-    },
-    {
-      "epoch": 3.6902216427640155,
-      "grad_norm": 1.4052753448486328,
-      "learning_rate": 9.339296482412061e-05,
-      "loss": 5.7596,
-      "step": 7076
-    },
-    {
-      "epoch": 3.690743155149935,
-      "grad_norm": 1.4011518955230713,
-      "learning_rate": 9.339195979899498e-05,
-      "loss": 5.7918,
-      "step": 7077
-    },
-    {
-      "epoch": 3.691264667535854,
-      "grad_norm": 1.5099977254867554,
-      "learning_rate": 9.339095477386935e-05,
-      "loss": 5.8228,
-      "step": 7078
-    },
-    {
-      "epoch": 3.691786179921773,
-      "grad_norm": 1.541614294052124,
-      "learning_rate": 9.338994974874372e-05,
-      "loss": 5.3396,
-      "step": 7079
-    },
-    {
-      "epoch": 3.6923076923076925,
-      "grad_norm": 1.4532945156097412,
-      "learning_rate": 9.33889447236181e-05,
-      "loss": 5.6242,
-      "step": 7080
-    },
-    {
-      "epoch": 3.6928292046936115,
-      "grad_norm": 1.4778698682785034,
-      "learning_rate": 9.338793969849246e-05,
-      "loss": 5.9388,
-      "step": 7081
-    },
-    {
-      "epoch": 3.6933507170795306,
-      "grad_norm": 1.5695003271102905,
-      "learning_rate": 9.338693467336684e-05,
-      "loss": 5.2572,
-      "step": 7082
-    },
-    {
-      "epoch": 3.69387222946545,
-      "grad_norm": 1.472528338432312,
-      "learning_rate": 9.338592964824122e-05,
-      "loss": 5.444,
-      "step": 7083
-    },
-    {
-      "epoch": 3.694393741851369,
-      "grad_norm": 1.4267178773880005,
-      "learning_rate": 9.338492462311558e-05,
-      "loss": 6.0963,
-      "step": 7084
-    },
-    {
-      "epoch": 3.694915254237288,
-      "grad_norm": 1.4427791833877563,
-      "learning_rate": 9.338391959798996e-05,
-      "loss": 5.8147,
-      "step": 7085
-    },
-    {
-      "epoch": 3.6954367666232075,
-      "grad_norm": 1.495205044746399,
-      "learning_rate": 9.338291457286432e-05,
-      "loss": 5.8751,
-      "step": 7086
-    },
-    {
-      "epoch": 3.6959582790091265,
-      "grad_norm": 1.6204521656036377,
-      "learning_rate": 9.33819095477387e-05,
-      "loss": 5.5026,
-      "step": 7087
-    },
-    {
-      "epoch": 3.6964797913950456,
-      "grad_norm": 1.5209815502166748,
-      "learning_rate": 9.338090452261306e-05,
-      "loss": 5.3233,
-      "step": 7088
-    },
-    {
-      "epoch": 3.697001303780965,
-      "grad_norm": 1.6296061277389526,
-      "learning_rate": 9.337989949748744e-05,
-      "loss": 5.6748,
-      "step": 7089
-    },
-    {
-      "epoch": 3.697522816166884,
-      "grad_norm": 1.5483607053756714,
-      "learning_rate": 9.33788944723618e-05,
-      "loss": 5.5865,
-      "step": 7090
-    },
-    {
-      "epoch": 3.698044328552803,
-      "grad_norm": 1.5365935564041138,
-      "learning_rate": 9.337788944723618e-05,
-      "loss": 5.4778,
-      "step": 7091
-    },
-    {
-      "epoch": 3.6985658409387225,
-      "grad_norm": 1.5664517879486084,
-      "learning_rate": 9.337688442211056e-05,
-      "loss": 5.837,
-      "step": 7092
-    },
-    {
-      "epoch": 3.6990873533246416,
-      "grad_norm": 1.3794665336608887,
-      "learning_rate": 9.337587939698494e-05,
-      "loss": 5.8759,
-      "step": 7093
-    },
-    {
-      "epoch": 3.6996088657105606,
-      "grad_norm": 1.7324854135513306,
-      "learning_rate": 9.33748743718593e-05,
-      "loss": 5.4481,
-      "step": 7094
-    },
-    {
-      "epoch": 3.70013037809648,
-      "grad_norm": 1.5653431415557861,
-      "learning_rate": 9.337386934673368e-05,
-      "loss": 5.6578,
-      "step": 7095
-    },
-    {
-      "epoch": 3.700651890482399,
-      "grad_norm": 1.7156696319580078,
-      "learning_rate": 9.337286432160805e-05,
-      "loss": 5.1855,
-      "step": 7096
-    },
-    {
-      "epoch": 3.701173402868318,
-      "grad_norm": 1.4905757904052734,
-      "learning_rate": 9.337185929648241e-05,
-      "loss": 5.448,
-      "step": 7097
-    },
-    {
-      "epoch": 3.7016949152542376,
-      "grad_norm": 1.4102139472961426,
-      "learning_rate": 9.337085427135679e-05,
-      "loss": 5.8454,
-      "step": 7098
-    },
-    {
-      "epoch": 3.7022164276401566,
-      "grad_norm": 1.7421170473098755,
-      "learning_rate": 9.336984924623115e-05,
-      "loss": 5.1253,
-      "step": 7099
-    },
-    {
-      "epoch": 3.7027379400260756,
-      "grad_norm": 1.5678882598876953,
-      "learning_rate": 9.336884422110553e-05,
-      "loss": 5.998,
-      "step": 7100
-    },
-    {
-      "epoch": 3.703259452411995,
-      "grad_norm": 1.7802516222000122,
-      "learning_rate": 9.33678391959799e-05,
-      "loss": 5.5837,
-      "step": 7101
-    },
-    {
-      "epoch": 3.703780964797914,
-      "grad_norm": 1.3431756496429443,
-      "learning_rate": 9.336683417085427e-05,
-      "loss": 5.6267,
-      "step": 7102
-    },
-    {
-      "epoch": 3.704302477183833,
-      "grad_norm": 1.5176632404327393,
-      "learning_rate": 9.336582914572865e-05,
-      "loss": 5.7622,
-      "step": 7103
-    },
-    {
-      "epoch": 3.704823989569752,
-      "grad_norm": 1.433733582496643,
-      "learning_rate": 9.336482412060303e-05,
-      "loss": 5.8306,
-      "step": 7104
-    },
-    {
-      "epoch": 3.7053455019556716,
-      "grad_norm": 1.4294573068618774,
-      "learning_rate": 9.336381909547739e-05,
-      "loss": 6.1294,
-      "step": 7105
-    },
-    {
-      "epoch": 3.7058670143415906,
-      "grad_norm": 1.5673069953918457,
-      "learning_rate": 9.336281407035177e-05,
-      "loss": 5.3017,
-      "step": 7106
-    },
-    {
-      "epoch": 3.7063885267275096,
-      "grad_norm": 1.733873724937439,
-      "learning_rate": 9.336180904522613e-05,
-      "loss": 5.9482,
-      "step": 7107
-    },
-    {
-      "epoch": 3.706910039113429,
-      "grad_norm": 1.4529964923858643,
-      "learning_rate": 9.336080402010051e-05,
-      "loss": 5.7361,
-      "step": 7108
-    },
-    {
-      "epoch": 3.707431551499348,
-      "grad_norm": 1.461370587348938,
-      "learning_rate": 9.335979899497488e-05,
-      "loss": 5.9837,
-      "step": 7109
-    },
-    {
-      "epoch": 3.707953063885267,
-      "grad_norm": 1.5137066841125488,
-      "learning_rate": 9.335879396984924e-05,
-      "loss": 5.6109,
-      "step": 7110
-    },
-    {
-      "epoch": 3.7084745762711866,
-      "grad_norm": 1.4195433855056763,
-      "learning_rate": 9.335778894472362e-05,
-      "loss": 5.7316,
-      "step": 7111
-    },
-    {
-      "epoch": 3.7089960886571056,
-      "grad_norm": 1.3399107456207275,
-      "learning_rate": 9.3356783919598e-05,
-      "loss": 6.1522,
-      "step": 7112
-    },
-    {
-      "epoch": 3.7095176010430246,
-      "grad_norm": 1.4708726406097412,
-      "learning_rate": 9.335577889447237e-05,
-      "loss": 5.6083,
-      "step": 7113
-    },
-    {
-      "epoch": 3.710039113428944,
-      "grad_norm": 1.3465808629989624,
-      "learning_rate": 9.335477386934674e-05,
-      "loss": 5.6968,
-      "step": 7114
-    },
-    {
-      "epoch": 3.710560625814863,
-      "grad_norm": 1.5656622648239136,
-      "learning_rate": 9.335376884422112e-05,
-      "loss": 5.6475,
-      "step": 7115
-    },
-    {
-      "epoch": 3.711082138200782,
-      "grad_norm": 1.558059811592102,
-      "learning_rate": 9.335276381909548e-05,
-      "loss": 4.6389,
-      "step": 7116
-    },
-    {
-      "epoch": 3.711603650586701,
-      "grad_norm": 1.4750794172286987,
-      "learning_rate": 9.335175879396986e-05,
-      "loss": 5.6088,
-      "step": 7117
-    },
-    {
-      "epoch": 3.7121251629726206,
-      "grad_norm": 1.557699203491211,
-      "learning_rate": 9.335075376884422e-05,
-      "loss": 5.8315,
-      "step": 7118
-    },
-    {
-      "epoch": 3.7126466753585396,
-      "grad_norm": 1.6379343271255493,
-      "learning_rate": 9.33497487437186e-05,
-      "loss": 6.1365,
-      "step": 7119
-    },
-    {
-      "epoch": 3.7131681877444587,
-      "grad_norm": 1.4784435033798218,
-      "learning_rate": 9.334874371859296e-05,
-      "loss": 5.7336,
-      "step": 7120
-    },
-    {
-      "epoch": 3.713689700130378,
-      "grad_norm": 1.3685170412063599,
-      "learning_rate": 9.334773869346734e-05,
-      "loss": 6.0237,
-      "step": 7121
-    },
-    {
-      "epoch": 3.714211212516297,
-      "grad_norm": 1.3943976163864136,
-      "learning_rate": 9.33467336683417e-05,
-      "loss": 5.8507,
-      "step": 7122
-    },
-    {
-      "epoch": 3.714732724902216,
-      "grad_norm": 1.4288177490234375,
-      "learning_rate": 9.334572864321608e-05,
-      "loss": 5.337,
-      "step": 7123
-    },
-    {
-      "epoch": 3.7152542372881356,
-      "grad_norm": 1.4738540649414062,
-      "learning_rate": 9.334472361809046e-05,
-      "loss": 5.6156,
-      "step": 7124
-    },
-    {
-      "epoch": 3.7157757496740547,
-      "grad_norm": 1.4455350637435913,
-      "learning_rate": 9.334371859296483e-05,
-      "loss": 5.6818,
-      "step": 7125
-    },
-    {
-      "epoch": 3.7162972620599737,
-      "grad_norm": 1.4057294130325317,
-      "learning_rate": 9.33427135678392e-05,
-      "loss": 5.8931,
-      "step": 7126
-    },
-    {
-      "epoch": 3.716818774445893,
-      "grad_norm": 1.3574397563934326,
-      "learning_rate": 9.334170854271357e-05,
-      "loss": 5.7867,
-      "step": 7127
-    },
-    {
-      "epoch": 3.717340286831812,
-      "grad_norm": 1.3998303413391113,
-      "learning_rate": 9.334070351758795e-05,
-      "loss": 5.6385,
-      "step": 7128
-    },
-    {
-      "epoch": 3.717861799217731,
-      "grad_norm": 1.4585435390472412,
-      "learning_rate": 9.333969849246231e-05,
-      "loss": 5.9325,
-      "step": 7129
-    },
-    {
-      "epoch": 3.7183833116036507,
-      "grad_norm": 1.439855933189392,
-      "learning_rate": 9.333869346733669e-05,
-      "loss": 6.0231,
-      "step": 7130
-    },
-    {
-      "epoch": 3.7189048239895697,
-      "grad_norm": 1.503981351852417,
-      "learning_rate": 9.333768844221105e-05,
-      "loss": 5.5094,
-      "step": 7131
-    },
-    {
-      "epoch": 3.7194263363754887,
-      "grad_norm": 1.5188429355621338,
-      "learning_rate": 9.333668341708543e-05,
-      "loss": 5.9689,
-      "step": 7132
-    },
-    {
-      "epoch": 3.719947848761408,
-      "grad_norm": 1.443252444267273,
-      "learning_rate": 9.333567839195981e-05,
-      "loss": 5.6526,
-      "step": 7133
-    },
-    {
-      "epoch": 3.720469361147327,
-      "grad_norm": 1.598919153213501,
-      "learning_rate": 9.333467336683419e-05,
-      "loss": 5.6078,
-      "step": 7134
-    },
-    {
-      "epoch": 3.720990873533246,
-      "grad_norm": 1.4989668130874634,
-      "learning_rate": 9.333366834170855e-05,
-      "loss": 5.6883,
-      "step": 7135
-    },
-    {
-      "epoch": 3.7215123859191657,
-      "grad_norm": 1.5961953401565552,
-      "learning_rate": 9.333266331658293e-05,
-      "loss": 5.4955,
-      "step": 7136
-    },
-    {
-      "epoch": 3.7220338983050847,
-      "grad_norm": 1.37103271484375,
-      "learning_rate": 9.333165829145729e-05,
-      "loss": 5.6774,
-      "step": 7137
-    },
-    {
-      "epoch": 3.7225554106910037,
-      "grad_norm": 1.4646453857421875,
-      "learning_rate": 9.333065326633166e-05,
-      "loss": 5.9224,
-      "step": 7138
-    },
-    {
-      "epoch": 3.723076923076923,
-      "grad_norm": 1.4542596340179443,
-      "learning_rate": 9.332964824120603e-05,
-      "loss": 5.8611,
-      "step": 7139
-    },
-    {
-      "epoch": 3.723598435462842,
-      "grad_norm": 1.4691792726516724,
-      "learning_rate": 9.33286432160804e-05,
-      "loss": 5.7376,
-      "step": 7140
-    },
-    {
-      "epoch": 3.724119947848761,
-      "grad_norm": 1.433346152305603,
-      "learning_rate": 9.332763819095478e-05,
-      "loss": 5.8587,
-      "step": 7141
-    },
-    {
-      "epoch": 3.7246414602346807,
-      "grad_norm": 1.317259430885315,
-      "learning_rate": 9.332663316582914e-05,
-      "loss": 6.1521,
-      "step": 7142
-    },
-    {
-      "epoch": 3.7251629726205997,
-      "grad_norm": 1.4735013246536255,
-      "learning_rate": 9.332562814070352e-05,
-      "loss": 5.9624,
-      "step": 7143
-    },
-    {
-      "epoch": 3.7256844850065187,
-      "grad_norm": 1.5436524152755737,
-      "learning_rate": 9.33246231155779e-05,
-      "loss": 5.2357,
-      "step": 7144
-    },
-    {
-      "epoch": 3.726205997392438,
-      "grad_norm": 1.4885532855987549,
-      "learning_rate": 9.332361809045227e-05,
-      "loss": 5.7375,
-      "step": 7145
-    },
-    {
-      "epoch": 3.726727509778357,
-      "grad_norm": 1.477995753288269,
-      "learning_rate": 9.332261306532664e-05,
-      "loss": 5.643,
-      "step": 7146
-    },
-    {
-      "epoch": 3.7272490221642762,
-      "grad_norm": 1.4834468364715576,
-      "learning_rate": 9.332160804020102e-05,
-      "loss": 5.8137,
-      "step": 7147
-    },
-    {
-      "epoch": 3.7277705345501957,
-      "grad_norm": 1.504822015762329,
-      "learning_rate": 9.332060301507538e-05,
-      "loss": 5.8265,
-      "step": 7148
-    },
-    {
-      "epoch": 3.7282920469361147,
-      "grad_norm": 1.5279245376586914,
-      "learning_rate": 9.331959798994976e-05,
-      "loss": 5.7722,
-      "step": 7149
-    },
-    {
-      "epoch": 3.7288135593220337,
-      "grad_norm": 1.4921904802322388,
-      "learning_rate": 9.331859296482412e-05,
-      "loss": 5.8326,
-      "step": 7150
-    },
-    {
-      "epoch": 3.729335071707953,
-      "grad_norm": 1.4843816757202148,
-      "learning_rate": 9.331758793969849e-05,
-      "loss": 5.2011,
-      "step": 7151
-    },
-    {
-      "epoch": 3.729856584093872,
-      "grad_norm": 1.505672812461853,
-      "learning_rate": 9.331658291457287e-05,
-      "loss": 6.1325,
-      "step": 7152
-    },
-    {
-      "epoch": 3.7303780964797912,
-      "grad_norm": 1.4861228466033936,
-      "learning_rate": 9.331557788944724e-05,
-      "loss": 5.906,
-      "step": 7153
-    },
-    {
-      "epoch": 3.7308996088657107,
-      "grad_norm": 1.4283781051635742,
-      "learning_rate": 9.331457286432162e-05,
-      "loss": 5.8118,
-      "step": 7154
-    },
-    {
-      "epoch": 3.7314211212516297,
-      "grad_norm": 1.4170303344726562,
-      "learning_rate": 9.331356783919599e-05,
-      "loss": 6.1711,
-      "step": 7155
-    },
-    {
-      "epoch": 3.7319426336375487,
-      "grad_norm": 1.455032467842102,
-      "learning_rate": 9.331256281407036e-05,
-      "loss": 5.7403,
-      "step": 7156
-    },
-    {
-      "epoch": 3.732464146023468,
-      "grad_norm": 1.6105223894119263,
-      "learning_rate": 9.331155778894473e-05,
-      "loss": 5.7563,
-      "step": 7157
-    },
-    {
-      "epoch": 3.7329856584093872,
-      "grad_norm": 1.6175833940505981,
-      "learning_rate": 9.33105527638191e-05,
-      "loss": 5.7889,
-      "step": 7158
-    },
-    {
-      "epoch": 3.7335071707953063,
-      "grad_norm": 1.622033715248108,
-      "learning_rate": 9.330954773869347e-05,
-      "loss": 5.825,
-      "step": 7159
-    },
-    {
-      "epoch": 3.7340286831812257,
-      "grad_norm": 1.4206181764602661,
-      "learning_rate": 9.330854271356785e-05,
-      "loss": 5.7173,
-      "step": 7160
-    },
-    {
-      "epoch": 3.7345501955671447,
-      "grad_norm": 1.5119035243988037,
-      "learning_rate": 9.330753768844221e-05,
-      "loss": 5.3716,
-      "step": 7161
-    },
-    {
-      "epoch": 3.7350717079530638,
-      "grad_norm": 1.4509031772613525,
-      "learning_rate": 9.330653266331659e-05,
-      "loss": 6.0847,
-      "step": 7162
-    },
-    {
-      "epoch": 3.7355932203389832,
-      "grad_norm": 1.4662338495254517,
-      "learning_rate": 9.330552763819095e-05,
-      "loss": 6.0955,
-      "step": 7163
-    },
-    {
-      "epoch": 3.7361147327249022,
-      "grad_norm": 1.455979585647583,
-      "learning_rate": 9.330452261306533e-05,
-      "loss": 6.0037,
-      "step": 7164
-    },
-    {
-      "epoch": 3.7366362451108213,
-      "grad_norm": 1.4780399799346924,
-      "learning_rate": 9.330351758793971e-05,
-      "loss": 5.7048,
-      "step": 7165
-    },
-    {
-      "epoch": 3.7371577574967407,
-      "grad_norm": 1.5267775058746338,
-      "learning_rate": 9.330251256281407e-05,
-      "loss": 5.4724,
-      "step": 7166
-    },
-    {
-      "epoch": 3.7376792698826597,
-      "grad_norm": 1.3959031105041504,
-      "learning_rate": 9.330150753768845e-05,
-      "loss": 5.4785,
-      "step": 7167
-    },
-    {
-      "epoch": 3.7382007822685788,
-      "grad_norm": 1.5438158512115479,
-      "learning_rate": 9.330050251256282e-05,
-      "loss": 5.5544,
-      "step": 7168
-    },
-    {
-      "epoch": 3.7387222946544982,
-      "grad_norm": 1.5348466634750366,
-      "learning_rate": 9.32994974874372e-05,
-      "loss": 5.9541,
-      "step": 7169
-    },
-    {
-      "epoch": 3.7392438070404173,
-      "grad_norm": 1.3930391073226929,
-      "learning_rate": 9.329849246231156e-05,
-      "loss": 5.9569,
-      "step": 7170
-    },
-    {
-      "epoch": 3.7397653194263363,
-      "grad_norm": 1.515562653541565,
-      "learning_rate": 9.329748743718594e-05,
-      "loss": 5.8101,
-      "step": 7171
-    },
-    {
-      "epoch": 3.7402868318122557,
-      "grad_norm": 1.501370906829834,
-      "learning_rate": 9.32964824120603e-05,
-      "loss": 5.5987,
-      "step": 7172
-    },
-    {
-      "epoch": 3.7408083441981748,
-      "grad_norm": 1.5222851037979126,
-      "learning_rate": 9.329547738693468e-05,
-      "loss": 5.6979,
-      "step": 7173
-    },
-    {
-      "epoch": 3.741329856584094,
-      "grad_norm": 1.3771880865097046,
-      "learning_rate": 9.329447236180906e-05,
-      "loss": 5.9444,
-      "step": 7174
-    },
-    {
-      "epoch": 3.7418513689700132,
-      "grad_norm": 1.4558560848236084,
-      "learning_rate": 9.329346733668343e-05,
-      "loss": 5.8133,
-      "step": 7175
-    },
-    {
-      "epoch": 3.7423728813559323,
-      "grad_norm": 1.453021764755249,
-      "learning_rate": 9.32924623115578e-05,
-      "loss": 5.996,
-      "step": 7176
-    },
-    {
-      "epoch": 3.7428943937418513,
-      "grad_norm": 1.611220359802246,
-      "learning_rate": 9.329145728643216e-05,
-      "loss": 5.7464,
-      "step": 7177
-    },
-    {
-      "epoch": 3.7434159061277708,
-      "grad_norm": 1.5124820470809937,
-      "learning_rate": 9.329045226130654e-05,
-      "loss": 5.4252,
-      "step": 7178
-    },
-    {
-      "epoch": 3.7439374185136898,
-      "grad_norm": 1.552105188369751,
-      "learning_rate": 9.32894472361809e-05,
-      "loss": 6.0202,
-      "step": 7179
-    },
-    {
-      "epoch": 3.744458930899609,
-      "grad_norm": 1.4051285982131958,
-      "learning_rate": 9.328844221105528e-05,
-      "loss": 6.0398,
-      "step": 7180
-    },
-    {
-      "epoch": 3.7449804432855283,
-      "grad_norm": 1.3960286378860474,
-      "learning_rate": 9.328743718592965e-05,
-      "loss": 5.9059,
-      "step": 7181
-    },
-    {
-      "epoch": 3.7455019556714473,
-      "grad_norm": 1.5719902515411377,
-      "learning_rate": 9.328643216080402e-05,
-      "loss": 5.3359,
-      "step": 7182
-    },
-    {
-      "epoch": 3.7460234680573663,
-      "grad_norm": 1.4371936321258545,
-      "learning_rate": 9.328542713567839e-05,
-      "loss": 5.9243,
-      "step": 7183
-    },
-    {
-      "epoch": 3.7465449804432858,
-      "grad_norm": 1.4647128582000732,
-      "learning_rate": 9.328442211055277e-05,
-      "loss": 6.0369,
-      "step": 7184
-    },
-    {
-      "epoch": 3.747066492829205,
-      "grad_norm": 1.344878911972046,
-      "learning_rate": 9.328341708542714e-05,
-      "loss": 6.0636,
-      "step": 7185
-    },
-    {
-      "epoch": 3.747588005215124,
-      "grad_norm": 1.4798606634140015,
-      "learning_rate": 9.328241206030152e-05,
-      "loss": 5.7759,
-      "step": 7186
-    },
-    {
-      "epoch": 3.7481095176010433,
-      "grad_norm": 1.374454140663147,
-      "learning_rate": 9.328140703517589e-05,
-      "loss": 6.09,
-      "step": 7187
-    },
-    {
-      "epoch": 3.7486310299869623,
-      "grad_norm": 1.3482415676116943,
-      "learning_rate": 9.328040201005026e-05,
-      "loss": 5.8672,
-      "step": 7188
-    },
-    {
-      "epoch": 3.7491525423728813,
-      "grad_norm": 1.3266233205795288,
-      "learning_rate": 9.327939698492463e-05,
-      "loss": 5.8613,
-      "step": 7189
-    },
-    {
-      "epoch": 3.749674054758801,
-      "grad_norm": 1.8683648109436035,
-      "learning_rate": 9.327839195979899e-05,
-      "loss": 5.1099,
-      "step": 7190
-    },
-    {
-      "epoch": 3.75019556714472,
-      "grad_norm": 1.4043333530426025,
-      "learning_rate": 9.327738693467337e-05,
-      "loss": 5.8037,
-      "step": 7191
-    },
-    {
-      "epoch": 3.750717079530639,
-      "grad_norm": 1.4409458637237549,
-      "learning_rate": 9.327638190954773e-05,
-      "loss": 4.9895,
-      "step": 7192
-    },
-    {
-      "epoch": 3.7512385919165583,
-      "grad_norm": 1.4674501419067383,
-      "learning_rate": 9.327537688442211e-05,
-      "loss": 5.6206,
-      "step": 7193
-    },
-    {
-      "epoch": 3.7517601043024773,
-      "grad_norm": 1.5053274631500244,
-      "learning_rate": 9.327437185929648e-05,
-      "loss": 5.702,
-      "step": 7194
-    },
-    {
-      "epoch": 3.7522816166883963,
-      "grad_norm": 1.4777790307998657,
-      "learning_rate": 9.327336683417085e-05,
-      "loss": 5.8974,
-      "step": 7195
-    },
-    {
-      "epoch": 3.752803129074316,
-      "grad_norm": 1.8290059566497803,
-      "learning_rate": 9.327236180904523e-05,
-      "loss": 5.56,
-      "step": 7196
-    },
-    {
-      "epoch": 3.753324641460235,
-      "grad_norm": 1.438625693321228,
-      "learning_rate": 9.327135678391961e-05,
-      "loss": 5.5376,
-      "step": 7197
-    },
-    {
-      "epoch": 3.753846153846154,
-      "grad_norm": 1.3616231679916382,
-      "learning_rate": 9.327035175879397e-05,
-      "loss": 6.0821,
-      "step": 7198
-    },
-    {
-      "epoch": 3.754367666232073,
-      "grad_norm": 1.5707908868789673,
-      "learning_rate": 9.326934673366835e-05,
-      "loss": 5.2319,
-      "step": 7199
-    },
-    {
-      "epoch": 3.7548891786179923,
-      "grad_norm": 1.7347354888916016,
-      "learning_rate": 9.326834170854272e-05,
-      "loss": 5.4574,
-      "step": 7200
-    },
-    {
-      "epoch": 3.7554106910039113,
-      "grad_norm": 1.454602837562561,
-      "learning_rate": 9.32673366834171e-05,
-      "loss": 5.5007,
-      "step": 7201
-    },
-    {
-      "epoch": 3.7559322033898304,
-      "grad_norm": 1.3563610315322876,
-      "learning_rate": 9.326633165829146e-05,
-      "loss": 5.788,
-      "step": 7202
-    },
-    {
-      "epoch": 3.75645371577575,
-      "grad_norm": 1.3581563234329224,
-      "learning_rate": 9.326532663316582e-05,
-      "loss": 5.7897,
-      "step": 7203
-    },
-    {
-      "epoch": 3.756975228161669,
-      "grad_norm": 1.5297315120697021,
-      "learning_rate": 9.32643216080402e-05,
-      "loss": 5.8699,
-      "step": 7204
-    },
-    {
-      "epoch": 3.757496740547588,
-      "grad_norm": 1.4880070686340332,
-      "learning_rate": 9.326331658291458e-05,
-      "loss": 5.7665,
-      "step": 7205
-    },
-    {
-      "epoch": 3.7580182529335073,
-      "grad_norm": 1.4058812856674194,
-      "learning_rate": 9.326231155778896e-05,
-      "loss": 5.7026,
-      "step": 7206
-    },
-    {
-      "epoch": 3.7585397653194264,
-      "grad_norm": 1.4285526275634766,
-      "learning_rate": 9.326130653266332e-05,
-      "loss": 6.0716,
-      "step": 7207
-    },
-    {
-      "epoch": 3.7590612777053454,
-      "grad_norm": 1.393763542175293,
-      "learning_rate": 9.32603015075377e-05,
-      "loss": 5.93,
-      "step": 7208
-    },
-    {
-      "epoch": 3.7595827900912644,
-      "grad_norm": 1.5102362632751465,
-      "learning_rate": 9.325929648241206e-05,
-      "loss": 5.5759,
-      "step": 7209
-    },
-    {
-      "epoch": 3.760104302477184,
-      "grad_norm": 1.4420245885849,
-      "learning_rate": 9.325829145728644e-05,
-      "loss": 5.6187,
-      "step": 7210
-    },
-    {
-      "epoch": 3.760625814863103,
-      "grad_norm": 1.4810888767242432,
-      "learning_rate": 9.32572864321608e-05,
-      "loss": 5.6117,
-      "step": 7211
-    },
-    {
-      "epoch": 3.761147327249022,
-      "grad_norm": 1.4028464555740356,
-      "learning_rate": 9.325628140703518e-05,
-      "loss": 5.9744,
-      "step": 7212
-    },
-    {
-      "epoch": 3.7616688396349414,
-      "grad_norm": 1.429746150970459,
-      "learning_rate": 9.325527638190955e-05,
-      "loss": 5.5089,
-      "step": 7213
-    },
-    {
-      "epoch": 3.7621903520208604,
-      "grad_norm": 1.6163625717163086,
-      "learning_rate": 9.325427135678392e-05,
-      "loss": 5.779,
-      "step": 7214
-    },
-    {
-      "epoch": 3.7627118644067794,
-      "grad_norm": 1.8043291568756104,
-      "learning_rate": 9.325326633165829e-05,
-      "loss": 4.8775,
-      "step": 7215
-    },
-    {
-      "epoch": 3.763233376792699,
-      "grad_norm": 1.5143229961395264,
-      "learning_rate": 9.325226130653267e-05,
-      "loss": 5.9827,
-      "step": 7216
-    },
-    {
-      "epoch": 3.763754889178618,
-      "grad_norm": 1.4556866884231567,
-      "learning_rate": 9.325125628140704e-05,
-      "loss": 5.7868,
-      "step": 7217
-    },
-    {
-      "epoch": 3.764276401564537,
-      "grad_norm": 1.5184823274612427,
-      "learning_rate": 9.325025125628141e-05,
-      "loss": 5.5276,
-      "step": 7218
-    },
-    {
-      "epoch": 3.7647979139504564,
-      "grad_norm": 1.5395252704620361,
-      "learning_rate": 9.324924623115579e-05,
-      "loss": 5.4721,
-      "step": 7219
-    },
-    {
-      "epoch": 3.7653194263363754,
-      "grad_norm": 1.6493306159973145,
-      "learning_rate": 9.324824120603015e-05,
-      "loss": 5.561,
-      "step": 7220
-    },
-    {
-      "epoch": 3.7658409387222944,
-      "grad_norm": 1.4607335329055786,
-      "learning_rate": 9.324723618090453e-05,
-      "loss": 5.5343,
-      "step": 7221
-    },
-    {
-      "epoch": 3.766362451108214,
-      "grad_norm": 1.5821516513824463,
-      "learning_rate": 9.324623115577889e-05,
-      "loss": 5.3913,
-      "step": 7222
-    },
-    {
-      "epoch": 3.766883963494133,
-      "grad_norm": 1.551813006401062,
-      "learning_rate": 9.324522613065327e-05,
-      "loss": 5.892,
-      "step": 7223
-    },
-    {
-      "epoch": 3.767405475880052,
-      "grad_norm": 1.3982185125350952,
-      "learning_rate": 9.324422110552764e-05,
-      "loss": 5.5583,
-      "step": 7224
-    },
-    {
-      "epoch": 3.7679269882659714,
-      "grad_norm": 1.5022096633911133,
-      "learning_rate": 9.324321608040201e-05,
-      "loss": 5.9607,
-      "step": 7225
-    },
-    {
-      "epoch": 3.7684485006518904,
-      "grad_norm": 1.5564601421356201,
-      "learning_rate": 9.324221105527639e-05,
-      "loss": 5.6197,
-      "step": 7226
-    },
-    {
-      "epoch": 3.7689700130378094,
-      "grad_norm": 1.539781928062439,
-      "learning_rate": 9.324120603015077e-05,
-      "loss": 5.7089,
-      "step": 7227
-    },
-    {
-      "epoch": 3.769491525423729,
-      "grad_norm": 1.4592862129211426,
-      "learning_rate": 9.324020100502513e-05,
-      "loss": 5.7603,
-      "step": 7228
-    },
-    {
-      "epoch": 3.770013037809648,
-      "grad_norm": 1.4448012113571167,
-      "learning_rate": 9.323919597989951e-05,
-      "loss": 5.7562,
-      "step": 7229
-    },
-    {
-      "epoch": 3.770534550195567,
-      "grad_norm": 1.5434327125549316,
-      "learning_rate": 9.323819095477388e-05,
-      "loss": 5.8547,
-      "step": 7230
-    },
-    {
-      "epoch": 3.7710560625814864,
-      "grad_norm": 1.5035698413848877,
-      "learning_rate": 9.323718592964824e-05,
-      "loss": 5.2588,
-      "step": 7231
-    },
-    {
-      "epoch": 3.7715775749674054,
-      "grad_norm": 1.3654375076293945,
-      "learning_rate": 9.323618090452262e-05,
-      "loss": 6.1657,
-      "step": 7232
-    },
-    {
-      "epoch": 3.7720990873533244,
-      "grad_norm": 1.5737384557724,
-      "learning_rate": 9.323517587939698e-05,
-      "loss": 5.2929,
-      "step": 7233
-    },
-    {
-      "epoch": 3.772620599739244,
-      "grad_norm": 1.4058940410614014,
-      "learning_rate": 9.323417085427136e-05,
-      "loss": 4.9742,
-      "step": 7234
-    },
-    {
-      "epoch": 3.773142112125163,
-      "grad_norm": 1.484915018081665,
-      "learning_rate": 9.323316582914572e-05,
-      "loss": 5.8895,
-      "step": 7235
-    },
-    {
-      "epoch": 3.773663624511082,
-      "grad_norm": 1.4383556842803955,
-      "learning_rate": 9.32321608040201e-05,
-      "loss": 5.7523,
-      "step": 7236
-    },
-    {
-      "epoch": 3.7741851368970014,
-      "grad_norm": 1.45618736743927,
-      "learning_rate": 9.323115577889448e-05,
-      "loss": 5.9138,
-      "step": 7237
-    },
-    {
-      "epoch": 3.7747066492829204,
-      "grad_norm": 1.5671652555465698,
-      "learning_rate": 9.323015075376886e-05,
-      "loss": 5.4332,
-      "step": 7238
-    },
-    {
-      "epoch": 3.7752281616688395,
-      "grad_norm": 1.4395960569381714,
-      "learning_rate": 9.322914572864322e-05,
-      "loss": 5.4964,
-      "step": 7239
-    },
-    {
-      "epoch": 3.775749674054759,
-      "grad_norm": 1.4388784170150757,
-      "learning_rate": 9.32281407035176e-05,
-      "loss": 6.074,
-      "step": 7240
-    },
-    {
-      "epoch": 3.776271186440678,
-      "grad_norm": 1.5895932912826538,
-      "learning_rate": 9.322713567839196e-05,
-      "loss": 5.6353,
-      "step": 7241
-    },
-    {
-      "epoch": 3.776792698826597,
-      "grad_norm": 1.6872508525848389,
-      "learning_rate": 9.322613065326634e-05,
-      "loss": 5.3505,
-      "step": 7242
-    },
-    {
-      "epoch": 3.7773142112125164,
-      "grad_norm": 1.4466300010681152,
-      "learning_rate": 9.32251256281407e-05,
-      "loss": 5.841,
-      "step": 7243
-    },
-    {
-      "epoch": 3.7778357235984354,
-      "grad_norm": 1.572801947593689,
-      "learning_rate": 9.322412060301507e-05,
-      "loss": 5.8432,
-      "step": 7244
-    },
-    {
-      "epoch": 3.7783572359843545,
-      "grad_norm": 1.3997225761413574,
-      "learning_rate": 9.322311557788945e-05,
-      "loss": 5.8922,
-      "step": 7245
-    },
-    {
-      "epoch": 3.778878748370274,
-      "grad_norm": 1.5943304300308228,
-      "learning_rate": 9.322211055276383e-05,
-      "loss": 5.4997,
-      "step": 7246
-    },
-    {
-      "epoch": 3.779400260756193,
-      "grad_norm": 1.7823513746261597,
-      "learning_rate": 9.32211055276382e-05,
-      "loss": 5.6113,
-      "step": 7247
-    },
-    {
-      "epoch": 3.779921773142112,
-      "grad_norm": 1.5381051301956177,
-      "learning_rate": 9.322010050251257e-05,
-      "loss": 5.6708,
-      "step": 7248
-    },
-    {
-      "epoch": 3.7804432855280314,
-      "grad_norm": 1.616721510887146,
-      "learning_rate": 9.321909547738695e-05,
-      "loss": 5.4968,
-      "step": 7249
-    },
-    {
-      "epoch": 3.7809647979139505,
-      "grad_norm": 1.6024445295333862,
-      "learning_rate": 9.321809045226131e-05,
-      "loss": 5.5316,
-      "step": 7250
-    },
-    {
-      "epoch": 3.7814863102998695,
-      "grad_norm": 1.5536161661148071,
-      "learning_rate": 9.321708542713569e-05,
-      "loss": 5.4711,
-      "step": 7251
-    },
-    {
-      "epoch": 3.782007822685789,
-      "grad_norm": 1.4964324235916138,
-      "learning_rate": 9.321608040201005e-05,
-      "loss": 5.8638,
-      "step": 7252
-    },
-    {
-      "epoch": 3.782529335071708,
-      "grad_norm": 1.395397663116455,
-      "learning_rate": 9.321507537688443e-05,
-      "loss": 5.9046,
-      "step": 7253
-    },
-    {
-      "epoch": 3.783050847457627,
-      "grad_norm": 1.3754669427871704,
-      "learning_rate": 9.32140703517588e-05,
-      "loss": 5.8539,
-      "step": 7254
-    },
-    {
-      "epoch": 3.7835723598435465,
-      "grad_norm": 1.4803260564804077,
-      "learning_rate": 9.321306532663317e-05,
-      "loss": 5.8478,
-      "step": 7255
-    },
-    {
-      "epoch": 3.7840938722294655,
-      "grad_norm": 1.5437946319580078,
-      "learning_rate": 9.321206030150754e-05,
-      "loss": 5.6587,
-      "step": 7256
-    },
-    {
-      "epoch": 3.7846153846153845,
-      "grad_norm": 1.576158881187439,
-      "learning_rate": 9.321105527638191e-05,
-      "loss": 5.7187,
-      "step": 7257
-    },
-    {
-      "epoch": 3.785136897001304,
-      "grad_norm": 1.4382407665252686,
-      "learning_rate": 9.321005025125629e-05,
-      "loss": 5.7485,
-      "step": 7258
-    },
-    {
-      "epoch": 3.785658409387223,
-      "grad_norm": 1.462908387184143,
-      "learning_rate": 9.320904522613066e-05,
-      "loss": 5.7715,
-      "step": 7259
-    },
-    {
-      "epoch": 3.786179921773142,
-      "grad_norm": 1.4529008865356445,
-      "learning_rate": 9.320804020100503e-05,
-      "loss": 5.806,
-      "step": 7260
-    },
-    {
-      "epoch": 3.7867014341590615,
-      "grad_norm": 1.6221095323562622,
-      "learning_rate": 9.32070351758794e-05,
-      "loss": 5.1466,
-      "step": 7261
-    },
-    {
-      "epoch": 3.7872229465449805,
-      "grad_norm": 1.6239399909973145,
-      "learning_rate": 9.320603015075378e-05,
-      "loss": 5.6495,
-      "step": 7262
-    },
-    {
-      "epoch": 3.7877444589308995,
-      "grad_norm": 1.593031644821167,
-      "learning_rate": 9.320502512562814e-05,
-      "loss": 5.696,
-      "step": 7263
-    },
-    {
-      "epoch": 3.788265971316819,
-      "grad_norm": 1.4813724756240845,
-      "learning_rate": 9.320402010050252e-05,
-      "loss": 5.4802,
-      "step": 7264
-    },
-    {
-      "epoch": 3.788787483702738,
-      "grad_norm": 1.6157581806182861,
-      "learning_rate": 9.320301507537688e-05,
-      "loss": 5.2086,
-      "step": 7265
-    },
-    {
-      "epoch": 3.789308996088657,
-      "grad_norm": 1.5502675771713257,
-      "learning_rate": 9.320201005025126e-05,
-      "loss": 6.0485,
-      "step": 7266
-    },
-    {
-      "epoch": 3.7898305084745765,
-      "grad_norm": 1.2895877361297607,
-      "learning_rate": 9.320100502512564e-05,
-      "loss": 6.1049,
-      "step": 7267
-    },
-    {
-      "epoch": 3.7903520208604955,
-      "grad_norm": 1.4142566919326782,
-      "learning_rate": 9.320000000000002e-05,
-      "loss": 5.861,
-      "step": 7268
-    },
-    {
-      "epoch": 3.7908735332464145,
-      "grad_norm": 1.5140761137008667,
-      "learning_rate": 9.319899497487438e-05,
-      "loss": 5.7137,
-      "step": 7269
-    },
-    {
-      "epoch": 3.791395045632334,
-      "grad_norm": 1.4101840257644653,
-      "learning_rate": 9.319798994974874e-05,
-      "loss": 5.823,
-      "step": 7270
-    },
-    {
-      "epoch": 3.791916558018253,
-      "grad_norm": 1.2947514057159424,
-      "learning_rate": 9.319698492462312e-05,
-      "loss": 5.8618,
-      "step": 7271
-    },
-    {
-      "epoch": 3.792438070404172,
-      "grad_norm": 1.4608880281448364,
-      "learning_rate": 9.319597989949749e-05,
-      "loss": 5.5652,
-      "step": 7272
-    },
-    {
-      "epoch": 3.7929595827900915,
-      "grad_norm": 1.518391728401184,
-      "learning_rate": 9.319497487437186e-05,
-      "loss": 5.4611,
-      "step": 7273
-    },
-    {
-      "epoch": 3.7934810951760105,
-      "grad_norm": 1.664723515510559,
-      "learning_rate": 9.319396984924623e-05,
-      "loss": 5.4504,
-      "step": 7274
-    },
-    {
-      "epoch": 3.7940026075619295,
-      "grad_norm": 1.416684865951538,
-      "learning_rate": 9.31929648241206e-05,
-      "loss": 5.7277,
-      "step": 7275
-    },
-    {
-      "epoch": 3.794524119947849,
-      "grad_norm": 1.4352178573608398,
-      "learning_rate": 9.319195979899497e-05,
-      "loss": 5.91,
-      "step": 7276
-    },
-    {
-      "epoch": 3.795045632333768,
-      "grad_norm": 1.4008833169937134,
-      "learning_rate": 9.319095477386935e-05,
-      "loss": 5.6066,
-      "step": 7277
-    },
-    {
-      "epoch": 3.795567144719687,
-      "grad_norm": 1.3875792026519775,
-      "learning_rate": 9.318994974874373e-05,
-      "loss": 5.7657,
-      "step": 7278
-    },
-    {
-      "epoch": 3.7960886571056065,
-      "grad_norm": 1.8859283924102783,
-      "learning_rate": 9.31889447236181e-05,
-      "loss": 5.284,
-      "step": 7279
-    },
-    {
-      "epoch": 3.7966101694915255,
-      "grad_norm": 1.4276055097579956,
-      "learning_rate": 9.318793969849247e-05,
-      "loss": 5.8303,
-      "step": 7280
-    },
-    {
-      "epoch": 3.7971316818774445,
-      "grad_norm": 1.4568891525268555,
-      "learning_rate": 9.318693467336685e-05,
-      "loss": 5.6691,
-      "step": 7281
-    },
-    {
-      "epoch": 3.797653194263364,
-      "grad_norm": 1.5402214527130127,
-      "learning_rate": 9.318592964824121e-05,
-      "loss": 5.2943,
-      "step": 7282
-    },
-    {
-      "epoch": 3.798174706649283,
-      "grad_norm": 1.5098837614059448,
-      "learning_rate": 9.318492462311557e-05,
-      "loss": 5.7724,
-      "step": 7283
-    },
-    {
-      "epoch": 3.798696219035202,
-      "grad_norm": 1.5579032897949219,
-      "learning_rate": 9.318391959798995e-05,
-      "loss": 5.4571,
-      "step": 7284
-    },
-    {
-      "epoch": 3.7992177314211215,
-      "grad_norm": 1.5444303750991821,
-      "learning_rate": 9.318291457286432e-05,
-      "loss": 5.6682,
-      "step": 7285
-    },
-    {
-      "epoch": 3.7997392438070405,
-      "grad_norm": 1.6458990573883057,
-      "learning_rate": 9.31819095477387e-05,
-      "loss": 5.4333,
-      "step": 7286
-    },
-    {
-      "epoch": 3.8002607561929596,
-      "grad_norm": 1.695273995399475,
-      "learning_rate": 9.318090452261307e-05,
-      "loss": 5.3142,
-      "step": 7287
-    },
-    {
-      "epoch": 3.800782268578879,
-      "grad_norm": 1.6122409105300903,
-      "learning_rate": 9.317989949748745e-05,
-      "loss": 6.178,
-      "step": 7288
-    },
-    {
-      "epoch": 3.801303780964798,
-      "grad_norm": 1.466016173362732,
-      "learning_rate": 9.317889447236181e-05,
-      "loss": 5.9777,
-      "step": 7289
-    },
-    {
-      "epoch": 3.801825293350717,
-      "grad_norm": 1.5747724771499634,
-      "learning_rate": 9.317788944723619e-05,
-      "loss": 5.4203,
-      "step": 7290
-    },
-    {
-      "epoch": 3.802346805736636,
-      "grad_norm": 1.5811963081359863,
-      "learning_rate": 9.317688442211056e-05,
-      "loss": 5.5837,
-      "step": 7291
-    },
-    {
-      "epoch": 3.8028683181225555,
-      "grad_norm": 1.588461995124817,
-      "learning_rate": 9.317587939698493e-05,
-      "loss": 5.3391,
-      "step": 7292
-    },
-    {
-      "epoch": 3.8033898305084746,
-      "grad_norm": 1.4210489988327026,
-      "learning_rate": 9.31748743718593e-05,
-      "loss": 5.7186,
-      "step": 7293
-    },
-    {
-      "epoch": 3.8039113428943936,
-      "grad_norm": 1.428564190864563,
-      "learning_rate": 9.317386934673368e-05,
-      "loss": 5.7963,
-      "step": 7294
-    },
-    {
-      "epoch": 3.804432855280313,
-      "grad_norm": 1.6226942539215088,
-      "learning_rate": 9.317286432160804e-05,
-      "loss": 5.549,
-      "step": 7295
-    },
-    {
-      "epoch": 3.804954367666232,
-      "grad_norm": 1.4528485536575317,
-      "learning_rate": 9.31718592964824e-05,
-      "loss": 5.7482,
-      "step": 7296
-    },
-    {
-      "epoch": 3.805475880052151,
-      "grad_norm": 1.4543788433074951,
-      "learning_rate": 9.317085427135678e-05,
-      "loss": 5.8548,
-      "step": 7297
-    },
-    {
-      "epoch": 3.8059973924380706,
-      "grad_norm": 1.6189948320388794,
-      "learning_rate": 9.316984924623116e-05,
-      "loss": 5.3922,
-      "step": 7298
-    },
-    {
-      "epoch": 3.8065189048239896,
-      "grad_norm": 1.5571082830429077,
-      "learning_rate": 9.316884422110554e-05,
-      "loss": 5.8023,
-      "step": 7299
-    },
-    {
-      "epoch": 3.8070404172099086,
-      "grad_norm": 1.3986806869506836,
-      "learning_rate": 9.31678391959799e-05,
-      "loss": 6.1645,
-      "step": 7300
-    },
-    {
-      "epoch": 3.807561929595828,
-      "grad_norm": 1.3392630815505981,
-      "learning_rate": 9.316683417085428e-05,
-      "loss": 5.9051,
-      "step": 7301
-    },
-    {
-      "epoch": 3.808083441981747,
-      "grad_norm": 1.4621467590332031,
-      "learning_rate": 9.316582914572864e-05,
-      "loss": 5.8731,
-      "step": 7302
-    },
-    {
-      "epoch": 3.808604954367666,
-      "grad_norm": 1.4365191459655762,
-      "learning_rate": 9.316482412060302e-05,
-      "loss": 5.5736,
-      "step": 7303
-    },
-    {
-      "epoch": 3.809126466753585,
-      "grad_norm": 1.5669788122177124,
-      "learning_rate": 9.316381909547739e-05,
-      "loss": 5.4309,
-      "step": 7304
-    },
-    {
-      "epoch": 3.8096479791395046,
-      "grad_norm": 1.4692438840866089,
-      "learning_rate": 9.316281407035176e-05,
-      "loss": 5.6156,
-      "step": 7305
-    },
-    {
-      "epoch": 3.8101694915254236,
-      "grad_norm": 1.5946383476257324,
-      "learning_rate": 9.316180904522613e-05,
-      "loss": 5.6235,
-      "step": 7306
-    },
-    {
-      "epoch": 3.8106910039113426,
-      "grad_norm": 1.4411476850509644,
-      "learning_rate": 9.316080402010051e-05,
-      "loss": 5.474,
-      "step": 7307
-    },
-    {
-      "epoch": 3.811212516297262,
-      "grad_norm": 1.5943909883499146,
-      "learning_rate": 9.315979899497488e-05,
-      "loss": 5.7842,
-      "step": 7308
-    },
-    {
-      "epoch": 3.811734028683181,
-      "grad_norm": 1.4566446542739868,
-      "learning_rate": 9.315879396984926e-05,
-      "loss": 5.8226,
-      "step": 7309
-    },
-    {
-      "epoch": 3.8122555410691,
-      "grad_norm": 1.4428954124450684,
-      "learning_rate": 9.315778894472363e-05,
-      "loss": 6.093,
-      "step": 7310
-    },
-    {
-      "epoch": 3.8127770534550196,
-      "grad_norm": 1.6049491167068481,
-      "learning_rate": 9.315678391959799e-05,
-      "loss": 5.8609,
-      "step": 7311
-    },
-    {
-      "epoch": 3.8132985658409386,
-      "grad_norm": 1.4038115739822388,
-      "learning_rate": 9.315577889447237e-05,
-      "loss": 5.271,
-      "step": 7312
-    },
-    {
-      "epoch": 3.8138200782268576,
-      "grad_norm": 1.8441298007965088,
-      "learning_rate": 9.315477386934673e-05,
-      "loss": 5.5178,
-      "step": 7313
-    },
-    {
-      "epoch": 3.814341590612777,
-      "grad_norm": 1.3638814687728882,
-      "learning_rate": 9.315376884422111e-05,
-      "loss": 5.865,
-      "step": 7314
-    },
-    {
-      "epoch": 3.814863102998696,
-      "grad_norm": 1.3874976634979248,
-      "learning_rate": 9.315276381909548e-05,
-      "loss": 6.179,
-      "step": 7315
-    },
-    {
-      "epoch": 3.815384615384615,
-      "grad_norm": 1.5154038667678833,
-      "learning_rate": 9.315175879396985e-05,
-      "loss": 5.6758,
-      "step": 7316
-    },
-    {
-      "epoch": 3.8159061277705346,
-      "grad_norm": 1.4961614608764648,
-      "learning_rate": 9.315075376884422e-05,
-      "loss": 5.4213,
-      "step": 7317
-    },
-    {
-      "epoch": 3.8164276401564536,
-      "grad_norm": 1.4776335954666138,
-      "learning_rate": 9.31497487437186e-05,
-      "loss": 5.8478,
-      "step": 7318
-    },
-    {
-      "epoch": 3.8169491525423727,
-      "grad_norm": 1.537476658821106,
-      "learning_rate": 9.314874371859297e-05,
-      "loss": 5.9095,
-      "step": 7319
-    },
-    {
-      "epoch": 3.817470664928292,
-      "grad_norm": 1.3879420757293701,
-      "learning_rate": 9.314773869346735e-05,
-      "loss": 5.9753,
-      "step": 7320
-    },
-    {
-      "epoch": 3.817992177314211,
-      "grad_norm": 1.4604583978652954,
-      "learning_rate": 9.314673366834172e-05,
-      "loss": 5.3277,
-      "step": 7321
-    },
-    {
-      "epoch": 3.81851368970013,
-      "grad_norm": 1.5530931949615479,
-      "learning_rate": 9.314572864321609e-05,
-      "loss": 5.344,
-      "step": 7322
-    },
-    {
-      "epoch": 3.8190352020860496,
-      "grad_norm": 1.7422301769256592,
-      "learning_rate": 9.314472361809046e-05,
-      "loss": 5.529,
-      "step": 7323
-    },
-    {
-      "epoch": 3.8195567144719687,
-      "grad_norm": 1.6281100511550903,
-      "learning_rate": 9.314371859296482e-05,
-      "loss": 4.8211,
-      "step": 7324
-    },
-    {
-      "epoch": 3.8200782268578877,
-      "grad_norm": 1.9276742935180664,
-      "learning_rate": 9.31427135678392e-05,
-      "loss": 5.0726,
-      "step": 7325
-    },
-    {
-      "epoch": 3.820599739243807,
-      "grad_norm": 1.4983651638031006,
-      "learning_rate": 9.314170854271356e-05,
-      "loss": 5.4877,
-      "step": 7326
-    },
-    {
-      "epoch": 3.821121251629726,
-      "grad_norm": 1.5378259420394897,
-      "learning_rate": 9.314070351758794e-05,
-      "loss": 5.5339,
-      "step": 7327
-    },
-    {
-      "epoch": 3.821642764015645,
-      "grad_norm": 1.4841055870056152,
-      "learning_rate": 9.313969849246232e-05,
-      "loss": 5.7306,
-      "step": 7328
-    },
-    {
-      "epoch": 3.8221642764015646,
-      "grad_norm": 1.5960055589675903,
-      "learning_rate": 9.31386934673367e-05,
-      "loss": 5.159,
-      "step": 7329
-    },
-    {
-      "epoch": 3.8226857887874837,
-      "grad_norm": 1.4041827917099,
-      "learning_rate": 9.313768844221106e-05,
-      "loss": 5.6839,
-      "step": 7330
-    },
-    {
-      "epoch": 3.8232073011734027,
-      "grad_norm": 1.4106749296188354,
-      "learning_rate": 9.313668341708544e-05,
-      "loss": 5.8108,
-      "step": 7331
-    },
-    {
-      "epoch": 3.823728813559322,
-      "grad_norm": 1.524141788482666,
-      "learning_rate": 9.31356783919598e-05,
-      "loss": 5.0341,
-      "step": 7332
-    },
-    {
-      "epoch": 3.824250325945241,
-      "grad_norm": 1.4767173528671265,
-      "learning_rate": 9.313467336683418e-05,
-      "loss": 5.8565,
-      "step": 7333
-    },
-    {
-      "epoch": 3.82477183833116,
-      "grad_norm": 1.259976863861084,
-      "learning_rate": 9.313366834170855e-05,
-      "loss": 5.2375,
-      "step": 7334
-    },
-    {
-      "epoch": 3.8252933507170797,
-      "grad_norm": 1.5163867473602295,
-      "learning_rate": 9.313266331658292e-05,
-      "loss": 5.7012,
-      "step": 7335
-    },
-    {
-      "epoch": 3.8258148631029987,
-      "grad_norm": 1.5600740909576416,
-      "learning_rate": 9.313165829145729e-05,
-      "loss": 5.5869,
-      "step": 7336
-    },
-    {
-      "epoch": 3.8263363754889177,
-      "grad_norm": 1.5768271684646606,
-      "learning_rate": 9.313065326633165e-05,
-      "loss": 5.86,
-      "step": 7337
-    },
-    {
-      "epoch": 3.826857887874837,
-      "grad_norm": 1.4443084001541138,
-      "learning_rate": 9.312964824120603e-05,
-      "loss": 5.8948,
-      "step": 7338
-    },
-    {
-      "epoch": 3.827379400260756,
-      "grad_norm": 1.5020993947982788,
-      "learning_rate": 9.312864321608041e-05,
-      "loss": 5.8936,
-      "step": 7339
-    },
-    {
-      "epoch": 3.827900912646675,
-      "grad_norm": 1.5863524675369263,
-      "learning_rate": 9.312763819095479e-05,
-      "loss": 5.917,
-      "step": 7340
-    },
-    {
-      "epoch": 3.8284224250325947,
-      "grad_norm": 1.5403082370758057,
-      "learning_rate": 9.312663316582915e-05,
-      "loss": 5.2954,
-      "step": 7341
-    },
-    {
-      "epoch": 3.8289439374185137,
-      "grad_norm": 1.7329519987106323,
-      "learning_rate": 9.312562814070353e-05,
-      "loss": 5.965,
-      "step": 7342
-    },
-    {
-      "epoch": 3.8294654498044327,
-      "grad_norm": 1.5207000970840454,
-      "learning_rate": 9.312462311557789e-05,
-      "loss": 5.9517,
-      "step": 7343
-    },
-    {
-      "epoch": 3.829986962190352,
-      "grad_norm": 1.4660696983337402,
-      "learning_rate": 9.312361809045227e-05,
-      "loss": 5.368,
-      "step": 7344
-    },
-    {
-      "epoch": 3.830508474576271,
-      "grad_norm": 1.6758716106414795,
-      "learning_rate": 9.312261306532663e-05,
-      "loss": 5.7404,
-      "step": 7345
-    },
-    {
-      "epoch": 3.83102998696219,
-      "grad_norm": 1.622606873512268,
-      "learning_rate": 9.312160804020101e-05,
-      "loss": 6.0296,
-      "step": 7346
-    },
-    {
-      "epoch": 3.8315514993481097,
-      "grad_norm": 1.5950168371200562,
-      "learning_rate": 9.312060301507538e-05,
-      "loss": 5.636,
-      "step": 7347
-    },
-    {
-      "epoch": 3.8320730117340287,
-      "grad_norm": 1.4020154476165771,
-      "learning_rate": 9.311959798994975e-05,
-      "loss": 5.795,
-      "step": 7348
-    },
-    {
-      "epoch": 3.8325945241199477,
-      "grad_norm": 1.5346837043762207,
-      "learning_rate": 9.311859296482413e-05,
-      "loss": 5.9293,
-      "step": 7349
-    },
-    {
-      "epoch": 3.833116036505867,
-      "grad_norm": 1.5926657915115356,
-      "learning_rate": 9.31175879396985e-05,
-      "loss": 5.702,
-      "step": 7350
-    },
-    {
-      "epoch": 3.833637548891786,
-      "grad_norm": 1.3878870010375977,
-      "learning_rate": 9.311658291457287e-05,
-      "loss": 6.0429,
-      "step": 7351
-    },
-    {
-      "epoch": 3.8341590612777052,
-      "grad_norm": 1.4329559803009033,
-      "learning_rate": 9.311557788944724e-05,
-      "loss": 6.0858,
-      "step": 7352
-    },
-    {
-      "epoch": 3.8346805736636247,
-      "grad_norm": 1.5703593492507935,
-      "learning_rate": 9.311457286432162e-05,
-      "loss": 5.2387,
-      "step": 7353
-    },
-    {
-      "epoch": 3.8352020860495437,
-      "grad_norm": 1.588128924369812,
-      "learning_rate": 9.311356783919598e-05,
-      "loss": 5.7643,
-      "step": 7354
-    },
-    {
-      "epoch": 3.8357235984354627,
-      "grad_norm": 1.440496563911438,
-      "learning_rate": 9.311256281407036e-05,
-      "loss": 5.7707,
-      "step": 7355
-    },
-    {
-      "epoch": 3.836245110821382,
-      "grad_norm": 1.3987655639648438,
-      "learning_rate": 9.311155778894472e-05,
-      "loss": 5.7321,
-      "step": 7356
-    },
-    {
-      "epoch": 3.836766623207301,
-      "grad_norm": 1.4961631298065186,
-      "learning_rate": 9.31105527638191e-05,
-      "loss": 5.9595,
-      "step": 7357
-    },
-    {
-      "epoch": 3.8372881355932202,
-      "grad_norm": 1.5718039274215698,
-      "learning_rate": 9.310954773869346e-05,
-      "loss": 5.7678,
-      "step": 7358
-    },
-    {
-      "epoch": 3.8378096479791397,
-      "grad_norm": 1.4814274311065674,
-      "learning_rate": 9.310854271356784e-05,
-      "loss": 5.7824,
-      "step": 7359
-    },
-    {
-      "epoch": 3.8383311603650587,
-      "grad_norm": 1.5473933219909668,
-      "learning_rate": 9.310753768844222e-05,
-      "loss": 5.5492,
-      "step": 7360
-    },
-    {
-      "epoch": 3.8388526727509777,
-      "grad_norm": 1.5744168758392334,
-      "learning_rate": 9.31065326633166e-05,
-      "loss": 5.9439,
-      "step": 7361
-    },
-    {
-      "epoch": 3.839374185136897,
-      "grad_norm": 1.817595362663269,
-      "learning_rate": 9.310552763819096e-05,
-      "loss": 5.8264,
-      "step": 7362
-    },
-    {
-      "epoch": 3.8398956975228162,
-      "grad_norm": 1.365769863128662,
-      "learning_rate": 9.310452261306533e-05,
-      "loss": 5.8306,
-      "step": 7363
-    },
-    {
-      "epoch": 3.8404172099087353,
-      "grad_norm": 1.4149446487426758,
-      "learning_rate": 9.31035175879397e-05,
-      "loss": 5.696,
-      "step": 7364
-    },
-    {
-      "epoch": 3.8409387222946547,
-      "grad_norm": 1.5193815231323242,
-      "learning_rate": 9.310251256281407e-05,
-      "loss": 5.7487,
-      "step": 7365
-    },
-    {
-      "epoch": 3.8414602346805737,
-      "grad_norm": 1.3934236764907837,
-      "learning_rate": 9.310150753768845e-05,
-      "loss": 5.9013,
-      "step": 7366
-    },
-    {
-      "epoch": 3.8419817470664928,
-      "grad_norm": 1.521831750869751,
-      "learning_rate": 9.310050251256281e-05,
-      "loss": 5.6412,
-      "step": 7367
-    },
-    {
-      "epoch": 3.8425032594524122,
-      "grad_norm": 1.4701393842697144,
-      "learning_rate": 9.309949748743719e-05,
-      "loss": 6.2105,
-      "step": 7368
-    },
-    {
-      "epoch": 3.8430247718383312,
-      "grad_norm": 1.4762911796569824,
-      "learning_rate": 9.309849246231155e-05,
-      "loss": 5.265,
-      "step": 7369
-    },
-    {
-      "epoch": 3.8435462842242503,
-      "grad_norm": 1.4623039960861206,
-      "learning_rate": 9.309748743718593e-05,
-      "loss": 5.7053,
-      "step": 7370
-    },
-    {
-      "epoch": 3.8440677966101697,
-      "grad_norm": 1.442940354347229,
-      "learning_rate": 9.309648241206031e-05,
-      "loss": 5.6107,
-      "step": 7371
-    },
-    {
-      "epoch": 3.8445893089960888,
-      "grad_norm": 1.4975831508636475,
-      "learning_rate": 9.309547738693469e-05,
-      "loss": 5.7626,
-      "step": 7372
-    },
-    {
-      "epoch": 3.8451108213820078,
-      "grad_norm": 1.4760650396347046,
-      "learning_rate": 9.309447236180905e-05,
-      "loss": 5.8203,
-      "step": 7373
-    },
-    {
-      "epoch": 3.8456323337679272,
-      "grad_norm": 1.346336841583252,
-      "learning_rate": 9.309346733668343e-05,
-      "loss": 5.7178,
-      "step": 7374
-    },
-    {
-      "epoch": 3.8461538461538463,
-      "grad_norm": 1.4331563711166382,
-      "learning_rate": 9.309246231155779e-05,
-      "loss": 5.5786,
-      "step": 7375
-    },
-    {
-      "epoch": 3.8466753585397653,
-      "grad_norm": 1.5538933277130127,
-      "learning_rate": 9.309145728643216e-05,
-      "loss": 5.1849,
-      "step": 7376
-    },
-    {
-      "epoch": 3.8471968709256847,
-      "grad_norm": 1.509332299232483,
-      "learning_rate": 9.309045226130653e-05,
-      "loss": 5.6361,
-      "step": 7377
-    },
-    {
-      "epoch": 3.8477183833116038,
-      "grad_norm": 1.5013967752456665,
-      "learning_rate": 9.30894472361809e-05,
-      "loss": 5.8558,
-      "step": 7378
-    },
-    {
-      "epoch": 3.848239895697523,
-      "grad_norm": 1.5540711879730225,
-      "learning_rate": 9.308844221105528e-05,
-      "loss": 5.8808,
-      "step": 7379
-    },
-    {
-      "epoch": 3.8487614080834422,
-      "grad_norm": 1.3763415813446045,
-      "learning_rate": 9.308743718592965e-05,
-      "loss": 5.7128,
-      "step": 7380
-    },
-    {
-      "epoch": 3.8492829204693613,
-      "grad_norm": 1.5063140392303467,
-      "learning_rate": 9.308643216080403e-05,
-      "loss": 5.5197,
-      "step": 7381
-    },
-    {
-      "epoch": 3.8498044328552803,
-      "grad_norm": 1.6199727058410645,
-      "learning_rate": 9.30854271356784e-05,
-      "loss": 5.6432,
-      "step": 7382
-    },
-    {
-      "epoch": 3.8503259452411998,
-      "grad_norm": 1.4238388538360596,
-      "learning_rate": 9.308442211055277e-05,
-      "loss": 5.7256,
-      "step": 7383
-    },
-    {
-      "epoch": 3.8508474576271188,
-      "grad_norm": 1.488381266593933,
-      "learning_rate": 9.308341708542714e-05,
-      "loss": 5.7625,
-      "step": 7384
-    },
-    {
-      "epoch": 3.851368970013038,
-      "grad_norm": 1.540092945098877,
-      "learning_rate": 9.308241206030152e-05,
-      "loss": 5.6911,
-      "step": 7385
-    },
-    {
-      "epoch": 3.851890482398957,
-      "grad_norm": 1.4613125324249268,
-      "learning_rate": 9.308140703517588e-05,
-      "loss": 5.5411,
-      "step": 7386
-    },
-    {
-      "epoch": 3.8524119947848763,
-      "grad_norm": 1.5402295589447021,
-      "learning_rate": 9.308040201005026e-05,
-      "loss": 5.9203,
-      "step": 7387
-    },
-    {
-      "epoch": 3.8529335071707953,
-      "grad_norm": 1.3402388095855713,
-      "learning_rate": 9.307939698492462e-05,
-      "loss": 6.102,
-      "step": 7388
-    },
-    {
-      "epoch": 3.8534550195567143,
-      "grad_norm": 1.401375412940979,
-      "learning_rate": 9.307839195979899e-05,
-      "loss": 5.8722,
-      "step": 7389
-    },
-    {
-      "epoch": 3.853976531942634,
-      "grad_norm": 1.422985315322876,
-      "learning_rate": 9.307738693467337e-05,
-      "loss": 6.1178,
-      "step": 7390
-    },
-    {
-      "epoch": 3.854498044328553,
-      "grad_norm": 1.3574943542480469,
-      "learning_rate": 9.307638190954774e-05,
-      "loss": 5.8447,
-      "step": 7391
-    },
-    {
-      "epoch": 3.855019556714472,
-      "grad_norm": 1.3394254446029663,
-      "learning_rate": 9.307537688442212e-05,
-      "loss": 6.0167,
-      "step": 7392
-    },
-    {
-      "epoch": 3.8555410691003913,
-      "grad_norm": 1.533672571182251,
-      "learning_rate": 9.307437185929649e-05,
-      "loss": 5.7297,
-      "step": 7393
-    },
-    {
-      "epoch": 3.8560625814863103,
-      "grad_norm": 1.4482028484344482,
-      "learning_rate": 9.307336683417086e-05,
-      "loss": 5.8843,
-      "step": 7394
-    },
-    {
-      "epoch": 3.8565840938722293,
-      "grad_norm": 1.3827320337295532,
-      "learning_rate": 9.307236180904523e-05,
-      "loss": 6.0301,
-      "step": 7395
-    },
-    {
-      "epoch": 3.8571056062581484,
-      "grad_norm": 1.476450800895691,
-      "learning_rate": 9.30713567839196e-05,
-      "loss": 5.9115,
-      "step": 7396
-    },
-    {
-      "epoch": 3.857627118644068,
-      "grad_norm": 1.450947642326355,
-      "learning_rate": 9.307035175879397e-05,
-      "loss": 5.7652,
-      "step": 7397
-    },
-    {
-      "epoch": 3.858148631029987,
-      "grad_norm": 1.5619162321090698,
-      "learning_rate": 9.306934673366835e-05,
-      "loss": 5.8607,
-      "step": 7398
-    },
-    {
-      "epoch": 3.858670143415906,
-      "grad_norm": 1.583642840385437,
-      "learning_rate": 9.306834170854271e-05,
-      "loss": 5.5721,
-      "step": 7399
-    },
-    {
-      "epoch": 3.8591916558018253,
-      "grad_norm": 1.614371657371521,
-      "learning_rate": 9.306733668341709e-05,
-      "loss": 5.2191,
-      "step": 7400
-    },
-    {
-      "epoch": 3.8597131681877443,
-      "grad_norm": 1.3573259115219116,
-      "learning_rate": 9.306633165829147e-05,
-      "loss": 5.1005,
-      "step": 7401
-    },
-    {
-      "epoch": 3.8602346805736634,
-      "grad_norm": 1.3951905965805054,
-      "learning_rate": 9.306532663316585e-05,
-      "loss": 5.7928,
-      "step": 7402
-    },
-    {
-      "epoch": 3.860756192959583,
-      "grad_norm": 1.4096719026565552,
-      "learning_rate": 9.306432160804021e-05,
-      "loss": 5.4934,
-      "step": 7403
-    },
-    {
-      "epoch": 3.861277705345502,
-      "grad_norm": 1.4047895669937134,
-      "learning_rate": 9.306331658291457e-05,
-      "loss": 5.6583,
-      "step": 7404
-    },
-    {
-      "epoch": 3.861799217731421,
-      "grad_norm": 1.5388154983520508,
-      "learning_rate": 9.306231155778895e-05,
-      "loss": 5.2656,
-      "step": 7405
-    },
-    {
-      "epoch": 3.8623207301173403,
-      "grad_norm": 1.5309275388717651,
-      "learning_rate": 9.306130653266332e-05,
-      "loss": 5.6995,
-      "step": 7406
-    },
-    {
-      "epoch": 3.8628422425032594,
-      "grad_norm": 1.4648752212524414,
-      "learning_rate": 9.30603015075377e-05,
-      "loss": 5.8571,
-      "step": 7407
-    },
-    {
-      "epoch": 3.8633637548891784,
-      "grad_norm": 1.5527669191360474,
-      "learning_rate": 9.305929648241206e-05,
-      "loss": 5.7188,
-      "step": 7408
-    },
-    {
-      "epoch": 3.863885267275098,
-      "grad_norm": 1.4530038833618164,
-      "learning_rate": 9.305829145728644e-05,
-      "loss": 6.1491,
-      "step": 7409
-    },
-    {
-      "epoch": 3.864406779661017,
-      "grad_norm": 1.472064733505249,
-      "learning_rate": 9.30572864321608e-05,
-      "loss": 5.4047,
-      "step": 7410
-    },
-    {
-      "epoch": 3.864928292046936,
-      "grad_norm": 1.478812575340271,
-      "learning_rate": 9.305628140703518e-05,
-      "loss": 5.6873,
-      "step": 7411
-    },
-    {
-      "epoch": 3.8654498044328554,
-      "grad_norm": 1.4151957035064697,
-      "learning_rate": 9.305527638190956e-05,
-      "loss": 5.813,
-      "step": 7412
-    },
-    {
-      "epoch": 3.8659713168187744,
-      "grad_norm": 1.4414981603622437,
-      "learning_rate": 9.305427135678393e-05,
-      "loss": 5.7353,
-      "step": 7413
-    },
-    {
-      "epoch": 3.8664928292046934,
-      "grad_norm": 1.5408568382263184,
-      "learning_rate": 9.30532663316583e-05,
-      "loss": 6.1171,
-      "step": 7414
-    },
-    {
-      "epoch": 3.867014341590613,
-      "grad_norm": 1.4267795085906982,
-      "learning_rate": 9.305226130653268e-05,
-      "loss": 5.9389,
-      "step": 7415
-    },
-    {
-      "epoch": 3.867535853976532,
-      "grad_norm": 1.6250139474868774,
-      "learning_rate": 9.305125628140704e-05,
-      "loss": 5.6821,
-      "step": 7416
-    },
-    {
-      "epoch": 3.868057366362451,
-      "grad_norm": 1.6536126136779785,
-      "learning_rate": 9.30502512562814e-05,
-      "loss": 5.3526,
-      "step": 7417
-    },
-    {
-      "epoch": 3.8685788787483704,
-      "grad_norm": 1.5202834606170654,
-      "learning_rate": 9.304924623115578e-05,
-      "loss": 6.027,
-      "step": 7418
-    },
-    {
-      "epoch": 3.8691003911342894,
-      "grad_norm": 1.5742031335830688,
-      "learning_rate": 9.304824120603015e-05,
-      "loss": 5.4087,
-      "step": 7419
-    },
-    {
-      "epoch": 3.8696219035202084,
-      "grad_norm": 1.4938205480575562,
-      "learning_rate": 9.304723618090452e-05,
-      "loss": 5.4023,
-      "step": 7420
-    },
-    {
-      "epoch": 3.870143415906128,
-      "grad_norm": 1.5057107210159302,
-      "learning_rate": 9.30462311557789e-05,
-      "loss": 5.6376,
-      "step": 7421
-    },
-    {
-      "epoch": 3.870664928292047,
-      "grad_norm": 1.477750301361084,
-      "learning_rate": 9.304522613065328e-05,
-      "loss": 5.9479,
-      "step": 7422
-    },
-    {
-      "epoch": 3.871186440677966,
-      "grad_norm": 1.5440618991851807,
-      "learning_rate": 9.304422110552764e-05,
-      "loss": 5.5301,
-      "step": 7423
-    },
-    {
-      "epoch": 3.8717079530638854,
-      "grad_norm": 1.5467777252197266,
-      "learning_rate": 9.304321608040202e-05,
-      "loss": 5.9813,
-      "step": 7424
-    },
-    {
-      "epoch": 3.8722294654498044,
-      "grad_norm": 1.5010435581207275,
-      "learning_rate": 9.304221105527639e-05,
-      "loss": 5.5503,
-      "step": 7425
-    },
-    {
-      "epoch": 3.8727509778357234,
-      "grad_norm": 1.6053048372268677,
-      "learning_rate": 9.304120603015076e-05,
-      "loss": 5.39,
-      "step": 7426
-    },
-    {
-      "epoch": 3.873272490221643,
-      "grad_norm": 1.4336978197097778,
-      "learning_rate": 9.304020100502513e-05,
-      "loss": 5.6162,
-      "step": 7427
-    },
-    {
-      "epoch": 3.873794002607562,
-      "grad_norm": 1.4311177730560303,
-      "learning_rate": 9.30391959798995e-05,
-      "loss": 5.8641,
-      "step": 7428
-    },
-    {
-      "epoch": 3.874315514993481,
-      "grad_norm": 1.3721880912780762,
-      "learning_rate": 9.303819095477387e-05,
-      "loss": 5.9185,
-      "step": 7429
-    },
-    {
-      "epoch": 3.8748370273794004,
-      "grad_norm": 1.6035172939300537,
-      "learning_rate": 9.303718592964823e-05,
-      "loss": 5.5294,
-      "step": 7430
-    },
-    {
-      "epoch": 3.8753585397653194,
-      "grad_norm": 1.346137285232544,
-      "learning_rate": 9.303618090452261e-05,
-      "loss": 6.0711,
-      "step": 7431
-    },
-    {
-      "epoch": 3.8758800521512384,
-      "grad_norm": 1.4295613765716553,
-      "learning_rate": 9.303517587939699e-05,
-      "loss": 5.7064,
-      "step": 7432
-    },
-    {
-      "epoch": 3.876401564537158,
-      "grad_norm": 1.6339564323425293,
-      "learning_rate": 9.303417085427137e-05,
-      "loss": 5.0129,
-      "step": 7433
-    },
-    {
-      "epoch": 3.876923076923077,
-      "grad_norm": 1.5336120128631592,
-      "learning_rate": 9.303316582914573e-05,
-      "loss": 5.7114,
-      "step": 7434
-    },
-    {
-      "epoch": 3.877444589308996,
-      "grad_norm": 1.6509746313095093,
-      "learning_rate": 9.303216080402011e-05,
-      "loss": 5.0414,
-      "step": 7435
-    },
-    {
-      "epoch": 3.8779661016949154,
-      "grad_norm": 1.4861316680908203,
-      "learning_rate": 9.303115577889447e-05,
-      "loss": 5.6393,
-      "step": 7436
-    },
-    {
-      "epoch": 3.8784876140808344,
-      "grad_norm": 1.4018250703811646,
-      "learning_rate": 9.303015075376885e-05,
-      "loss": 6.0318,
-      "step": 7437
-    },
-    {
-      "epoch": 3.8790091264667534,
-      "grad_norm": 1.2611461877822876,
-      "learning_rate": 9.302914572864322e-05,
-      "loss": 6.0721,
-      "step": 7438
-    },
-    {
-      "epoch": 3.879530638852673,
-      "grad_norm": 1.6758310794830322,
-      "learning_rate": 9.30281407035176e-05,
-      "loss": 5.3176,
-      "step": 7439
-    },
-    {
-      "epoch": 3.880052151238592,
-      "grad_norm": 1.5434646606445312,
-      "learning_rate": 9.302713567839196e-05,
-      "loss": 5.3169,
-      "step": 7440
-    },
-    {
-      "epoch": 3.880573663624511,
-      "grad_norm": 1.5318260192871094,
-      "learning_rate": 9.302613065326634e-05,
-      "loss": 6.0378,
-      "step": 7441
-    },
-    {
-      "epoch": 3.8810951760104304,
-      "grad_norm": 1.4274393320083618,
-      "learning_rate": 9.302512562814071e-05,
-      "loss": 5.8074,
-      "step": 7442
-    },
-    {
-      "epoch": 3.8816166883963494,
-      "grad_norm": 1.5367181301116943,
-      "learning_rate": 9.302412060301508e-05,
-      "loss": 5.6289,
-      "step": 7443
-    },
-    {
-      "epoch": 3.8821382007822685,
-      "grad_norm": 1.608225703239441,
-      "learning_rate": 9.302311557788946e-05,
-      "loss": 5.5976,
-      "step": 7444
-    },
-    {
-      "epoch": 3.882659713168188,
-      "grad_norm": 1.3789702653884888,
-      "learning_rate": 9.302211055276382e-05,
-      "loss": 5.7309,
-      "step": 7445
-    },
-    {
-      "epoch": 3.883181225554107,
-      "grad_norm": 1.4663275480270386,
-      "learning_rate": 9.30211055276382e-05,
-      "loss": 5.7647,
-      "step": 7446
-    },
-    {
-      "epoch": 3.883702737940026,
-      "grad_norm": 1.5276672840118408,
-      "learning_rate": 9.302010050251256e-05,
-      "loss": 5.6749,
-      "step": 7447
-    },
-    {
-      "epoch": 3.8842242503259454,
-      "grad_norm": 1.5292439460754395,
-      "learning_rate": 9.301909547738694e-05,
-      "loss": 5.7349,
-      "step": 7448
-    },
-    {
-      "epoch": 3.8847457627118644,
-      "grad_norm": 1.4629054069519043,
-      "learning_rate": 9.30180904522613e-05,
-      "loss": 6.0572,
-      "step": 7449
-    },
-    {
-      "epoch": 3.8852672750977835,
-      "grad_norm": 1.4691002368927002,
-      "learning_rate": 9.301708542713568e-05,
-      "loss": 5.836,
-      "step": 7450
-    },
-    {
-      "epoch": 3.885788787483703,
-      "grad_norm": 1.5686439275741577,
-      "learning_rate": 9.301608040201005e-05,
-      "loss": 5.5289,
-      "step": 7451
-    },
-    {
-      "epoch": 3.886310299869622,
-      "grad_norm": 1.5806677341461182,
-      "learning_rate": 9.301507537688442e-05,
-      "loss": 5.5114,
-      "step": 7452
-    },
-    {
-      "epoch": 3.886831812255541,
-      "grad_norm": 1.3830044269561768,
-      "learning_rate": 9.30140703517588e-05,
-      "loss": 6.1422,
-      "step": 7453
-    },
-    {
-      "epoch": 3.8873533246414604,
-      "grad_norm": 1.4468969106674194,
-      "learning_rate": 9.301306532663318e-05,
-      "loss": 5.8428,
-      "step": 7454
-    },
-    {
-      "epoch": 3.8878748370273795,
-      "grad_norm": 1.5743839740753174,
-      "learning_rate": 9.301206030150754e-05,
-      "loss": 5.1823,
-      "step": 7455
-    },
-    {
-      "epoch": 3.8883963494132985,
-      "grad_norm": 1.662235975265503,
-      "learning_rate": 9.301105527638191e-05,
-      "loss": 5.1281,
-      "step": 7456
-    },
-    {
-      "epoch": 3.888917861799218,
-      "grad_norm": 1.4992676973342896,
-      "learning_rate": 9.301005025125629e-05,
-      "loss": 5.6663,
-      "step": 7457
-    },
-    {
-      "epoch": 3.889439374185137,
-      "grad_norm": 1.5631073713302612,
-      "learning_rate": 9.300904522613065e-05,
-      "loss": 5.6072,
-      "step": 7458
-    },
-    {
-      "epoch": 3.889960886571056,
-      "grad_norm": 1.5182746648788452,
-      "learning_rate": 9.300804020100503e-05,
-      "loss": 5.3214,
-      "step": 7459
-    },
-    {
-      "epoch": 3.8904823989569755,
-      "grad_norm": 1.403206706047058,
-      "learning_rate": 9.300703517587939e-05,
-      "loss": 5.8482,
-      "step": 7460
-    },
-    {
-      "epoch": 3.8910039113428945,
-      "grad_norm": 1.4329872131347656,
-      "learning_rate": 9.300603015075377e-05,
-      "loss": 5.6059,
-      "step": 7461
-    },
-    {
-      "epoch": 3.8915254237288135,
-      "grad_norm": 1.5377227067947388,
-      "learning_rate": 9.300502512562815e-05,
-      "loss": 5.9023,
-      "step": 7462
-    },
-    {
-      "epoch": 3.892046936114733,
-      "grad_norm": 1.6596980094909668,
-      "learning_rate": 9.300402010050253e-05,
-      "loss": 5.1913,
-      "step": 7463
-    },
-    {
-      "epoch": 3.892568448500652,
-      "grad_norm": 1.5277000665664673,
-      "learning_rate": 9.300301507537689e-05,
-      "loss": 5.6356,
-      "step": 7464
-    },
-    {
-      "epoch": 3.893089960886571,
-      "grad_norm": 1.583370327949524,
-      "learning_rate": 9.300201005025127e-05,
-      "loss": 5.6137,
-      "step": 7465
-    },
-    {
-      "epoch": 3.8936114732724905,
-      "grad_norm": 1.6872515678405762,
-      "learning_rate": 9.300100502512563e-05,
-      "loss": 5.2131,
-      "step": 7466
-    },
-    {
-      "epoch": 3.8941329856584095,
-      "grad_norm": 1.495287299156189,
-      "learning_rate": 9.300000000000001e-05,
-      "loss": 5.5891,
-      "step": 7467
-    },
-    {
-      "epoch": 3.8946544980443285,
-      "grad_norm": 1.4534456729888916,
-      "learning_rate": 9.299899497487438e-05,
-      "loss": 5.5515,
-      "step": 7468
-    },
-    {
-      "epoch": 3.895176010430248,
-      "grad_norm": 1.6854496002197266,
-      "learning_rate": 9.299798994974874e-05,
-      "loss": 5.3055,
-      "step": 7469
-    },
-    {
-      "epoch": 3.895697522816167,
-      "grad_norm": 1.9250746965408325,
-      "learning_rate": 9.299698492462312e-05,
-      "loss": 4.9859,
-      "step": 7470
-    },
-    {
-      "epoch": 3.896219035202086,
-      "grad_norm": 1.4997434616088867,
-      "learning_rate": 9.299597989949748e-05,
-      "loss": 5.4935,
-      "step": 7471
-    },
-    {
-      "epoch": 3.8967405475880055,
-      "grad_norm": 1.476922869682312,
-      "learning_rate": 9.299497487437186e-05,
-      "loss": 5.6764,
-      "step": 7472
-    },
-    {
-      "epoch": 3.8972620599739245,
-      "grad_norm": 1.4271098375320435,
-      "learning_rate": 9.299396984924624e-05,
-      "loss": 5.7544,
-      "step": 7473
-    },
-    {
-      "epoch": 3.8977835723598435,
-      "grad_norm": 1.4630951881408691,
-      "learning_rate": 9.299296482412061e-05,
-      "loss": 6.0254,
-      "step": 7474
-    },
-    {
-      "epoch": 3.898305084745763,
-      "grad_norm": 1.4139411449432373,
-      "learning_rate": 9.299195979899498e-05,
-      "loss": 6.1193,
-      "step": 7475
-    },
-    {
-      "epoch": 3.898826597131682,
-      "grad_norm": 1.588281273841858,
-      "learning_rate": 9.299095477386936e-05,
-      "loss": 5.5082,
-      "step": 7476
-    },
-    {
-      "epoch": 3.899348109517601,
-      "grad_norm": 2.37221097946167,
-      "learning_rate": 9.298994974874372e-05,
-      "loss": 5.4785,
-      "step": 7477
-    },
-    {
-      "epoch": 3.89986962190352,
-      "grad_norm": 1.4668819904327393,
-      "learning_rate": 9.29889447236181e-05,
-      "loss": 6.075,
-      "step": 7478
-    },
-    {
-      "epoch": 3.9003911342894395,
-      "grad_norm": 1.547419548034668,
-      "learning_rate": 9.298793969849246e-05,
-      "loss": 5.7685,
-      "step": 7479
-    },
-    {
-      "epoch": 3.9009126466753585,
-      "grad_norm": 1.425195336341858,
-      "learning_rate": 9.298693467336684e-05,
-      "loss": 5.6616,
-      "step": 7480
-    },
-    {
-      "epoch": 3.9014341590612776,
-      "grad_norm": 1.563821792602539,
-      "learning_rate": 9.29859296482412e-05,
-      "loss": 5.8288,
-      "step": 7481
-    },
-    {
-      "epoch": 3.901955671447197,
-      "grad_norm": 1.326955795288086,
-      "learning_rate": 9.298492462311558e-05,
-      "loss": 6.1041,
-      "step": 7482
-    },
-    {
-      "epoch": 3.902477183833116,
-      "grad_norm": 1.5521398782730103,
-      "learning_rate": 9.298391959798996e-05,
-      "loss": 5.8085,
-      "step": 7483
-    },
-    {
-      "epoch": 3.902998696219035,
-      "grad_norm": 1.4717357158660889,
-      "learning_rate": 9.298291457286433e-05,
-      "loss": 5.7656,
-      "step": 7484
-    },
-    {
-      "epoch": 3.9035202086049545,
-      "grad_norm": 1.5879474878311157,
-      "learning_rate": 9.29819095477387e-05,
-      "loss": 5.393,
-      "step": 7485
-    },
-    {
-      "epoch": 3.9040417209908735,
-      "grad_norm": 1.4940433502197266,
-      "learning_rate": 9.298090452261307e-05,
-      "loss": 5.7401,
-      "step": 7486
-    },
-    {
-      "epoch": 3.9045632333767926,
-      "grad_norm": 1.3808059692382812,
-      "learning_rate": 9.297989949748745e-05,
-      "loss": 6.0001,
-      "step": 7487
-    },
-    {
-      "epoch": 3.905084745762712,
-      "grad_norm": 1.3587883710861206,
-      "learning_rate": 9.297889447236181e-05,
-      "loss": 5.74,
-      "step": 7488
-    },
-    {
-      "epoch": 3.905606258148631,
-      "grad_norm": 1.4871399402618408,
-      "learning_rate": 9.297788944723619e-05,
-      "loss": 5.7021,
-      "step": 7489
-    },
-    {
-      "epoch": 3.90612777053455,
-      "grad_norm": 1.3692749738693237,
-      "learning_rate": 9.297688442211055e-05,
-      "loss": 5.6565,
-      "step": 7490
-    },
-    {
-      "epoch": 3.906649282920469,
-      "grad_norm": 1.6146240234375,
-      "learning_rate": 9.297587939698493e-05,
-      "loss": 5.4211,
-      "step": 7491
-    },
-    {
-      "epoch": 3.9071707953063886,
-      "grad_norm": 1.6459628343582153,
-      "learning_rate": 9.29748743718593e-05,
-      "loss": 6.0388,
-      "step": 7492
-    },
-    {
-      "epoch": 3.9076923076923076,
-      "grad_norm": 1.500594973564148,
-      "learning_rate": 9.297386934673367e-05,
-      "loss": 5.5474,
-      "step": 7493
-    },
-    {
-      "epoch": 3.9082138200782266,
-      "grad_norm": 1.5119051933288574,
-      "learning_rate": 9.297286432160805e-05,
-      "loss": 5.8912,
-      "step": 7494
-    },
-    {
-      "epoch": 3.908735332464146,
-      "grad_norm": 1.5113798379898071,
-      "learning_rate": 9.297185929648243e-05,
-      "loss": 5.7602,
-      "step": 7495
-    },
-    {
-      "epoch": 3.909256844850065,
-      "grad_norm": 1.6350038051605225,
-      "learning_rate": 9.297085427135679e-05,
-      "loss": 5.5792,
-      "step": 7496
-    },
-    {
-      "epoch": 3.909778357235984,
-      "grad_norm": 1.4959030151367188,
-      "learning_rate": 9.296984924623116e-05,
-      "loss": 5.6566,
-      "step": 7497
-    },
-    {
-      "epoch": 3.9102998696219036,
-      "grad_norm": 1.2636975049972534,
-      "learning_rate": 9.296884422110553e-05,
-      "loss": 6.0959,
-      "step": 7498
-    },
-    {
-      "epoch": 3.9108213820078226,
-      "grad_norm": 1.5478317737579346,
-      "learning_rate": 9.29678391959799e-05,
-      "loss": 5.6066,
-      "step": 7499
-    },
-    {
-      "epoch": 3.9113428943937416,
-      "grad_norm": 1.7196043729782104,
-      "learning_rate": 9.296683417085428e-05,
-      "loss": 5.6258,
-      "step": 7500
-    },
-    {
-      "epoch": 3.911864406779661,
-      "grad_norm": 1.3744937181472778,
-      "learning_rate": 9.296582914572864e-05,
-      "loss": 6.0598,
-      "step": 7501
-    },
-    {
-      "epoch": 3.91238591916558,
-      "grad_norm": 1.4504047632217407,
-      "learning_rate": 9.296482412060302e-05,
-      "loss": 5.9541,
-      "step": 7502
-    },
-    {
-      "epoch": 3.912907431551499,
-      "grad_norm": 1.5273666381835938,
-      "learning_rate": 9.29638190954774e-05,
-      "loss": 5.355,
-      "step": 7503
-    },
-    {
-      "epoch": 3.9134289439374186,
-      "grad_norm": 1.4829984903335571,
-      "learning_rate": 9.296281407035177e-05,
-      "loss": 5.9562,
-      "step": 7504
-    },
-    {
-      "epoch": 3.9139504563233376,
-      "grad_norm": 1.4446399211883545,
-      "learning_rate": 9.296180904522614e-05,
-      "loss": 5.2616,
-      "step": 7505
-    },
-    {
-      "epoch": 3.9144719687092566,
-      "grad_norm": 1.4765549898147583,
-      "learning_rate": 9.296080402010052e-05,
-      "loss": 6.0661,
-      "step": 7506
-    },
-    {
-      "epoch": 3.914993481095176,
-      "grad_norm": 1.4246501922607422,
-      "learning_rate": 9.295979899497488e-05,
-      "loss": 5.8992,
-      "step": 7507
-    },
-    {
-      "epoch": 3.915514993481095,
-      "grad_norm": 1.4726570844650269,
-      "learning_rate": 9.295879396984926e-05,
-      "loss": 5.7851,
-      "step": 7508
-    },
-    {
-      "epoch": 3.916036505867014,
-      "grad_norm": 1.4989628791809082,
-      "learning_rate": 9.295778894472362e-05,
-      "loss": 5.4837,
-      "step": 7509
-    },
-    {
-      "epoch": 3.9165580182529336,
-      "grad_norm": 1.3995634317398071,
-      "learning_rate": 9.295678391959799e-05,
-      "loss": 5.8205,
-      "step": 7510
-    },
-    {
-      "epoch": 3.9170795306388526,
-      "grad_norm": 1.5223127603530884,
-      "learning_rate": 9.295577889447236e-05,
-      "loss": 5.7295,
-      "step": 7511
-    },
-    {
-      "epoch": 3.9176010430247716,
-      "grad_norm": 1.5087918043136597,
-      "learning_rate": 9.295477386934673e-05,
-      "loss": 5.3821,
-      "step": 7512
-    },
-    {
-      "epoch": 3.918122555410691,
-      "grad_norm": 1.348781704902649,
-      "learning_rate": 9.29537688442211e-05,
-      "loss": 5.9685,
-      "step": 7513
-    },
-    {
-      "epoch": 3.91864406779661,
-      "grad_norm": 1.4479998350143433,
-      "learning_rate": 9.295276381909548e-05,
-      "loss": 5.7061,
-      "step": 7514
-    },
-    {
-      "epoch": 3.919165580182529,
-      "grad_norm": 1.9498530626296997,
-      "learning_rate": 9.295175879396986e-05,
-      "loss": 5.7486,
-      "step": 7515
-    },
-    {
-      "epoch": 3.9196870925684486,
-      "grad_norm": 1.5440491437911987,
-      "learning_rate": 9.295075376884423e-05,
-      "loss": 5.3503,
-      "step": 7516
-    },
-    {
-      "epoch": 3.9202086049543676,
-      "grad_norm": 1.404083013534546,
-      "learning_rate": 9.29497487437186e-05,
-      "loss": 5.7557,
-      "step": 7517
-    },
-    {
-      "epoch": 3.9207301173402866,
-      "grad_norm": 1.5156038999557495,
-      "learning_rate": 9.294874371859297e-05,
-      "loss": 5.9383,
-      "step": 7518
-    },
-    {
-      "epoch": 3.921251629726206,
-      "grad_norm": 1.5546176433563232,
-      "learning_rate": 9.294773869346735e-05,
-      "loss": 5.7311,
-      "step": 7519
-    },
-    {
-      "epoch": 3.921773142112125,
-      "grad_norm": 1.4764430522918701,
-      "learning_rate": 9.294673366834171e-05,
-      "loss": 5.631,
-      "step": 7520
-    },
-    {
-      "epoch": 3.922294654498044,
-      "grad_norm": 1.4786019325256348,
-      "learning_rate": 9.294572864321609e-05,
-      "loss": 5.515,
-      "step": 7521
-    },
-    {
-      "epoch": 3.9228161668839636,
-      "grad_norm": 1.3405482769012451,
-      "learning_rate": 9.294472361809045e-05,
-      "loss": 6.0299,
-      "step": 7522
-    },
-    {
-      "epoch": 3.9233376792698826,
-      "grad_norm": 1.5859111547470093,
-      "learning_rate": 9.294371859296482e-05,
-      "loss": 5.4544,
-      "step": 7523
-    },
-    {
-      "epoch": 3.9238591916558017,
-      "grad_norm": 1.3977775573730469,
-      "learning_rate": 9.29427135678392e-05,
-      "loss": 4.9013,
-      "step": 7524
-    },
-    {
-      "epoch": 3.924380704041721,
-      "grad_norm": 1.3908288478851318,
-      "learning_rate": 9.294170854271357e-05,
-      "loss": 5.9341,
-      "step": 7525
-    },
-    {
-      "epoch": 3.92490221642764,
-      "grad_norm": 1.5150518417358398,
-      "learning_rate": 9.294070351758795e-05,
-      "loss": 5.7293,
-      "step": 7526
-    },
-    {
-      "epoch": 3.925423728813559,
-      "grad_norm": 1.4363093376159668,
-      "learning_rate": 9.293969849246231e-05,
-      "loss": 5.6272,
-      "step": 7527
-    },
-    {
-      "epoch": 3.9259452411994786,
-      "grad_norm": 1.4544016122817993,
-      "learning_rate": 9.293869346733669e-05,
-      "loss": 5.9819,
-      "step": 7528
-    },
-    {
-      "epoch": 3.9264667535853977,
-      "grad_norm": 1.5386769771575928,
-      "learning_rate": 9.293768844221106e-05,
-      "loss": 5.8177,
-      "step": 7529
-    },
-    {
-      "epoch": 3.9269882659713167,
-      "grad_norm": 1.4453848600387573,
-      "learning_rate": 9.293668341708543e-05,
-      "loss": 5.665,
-      "step": 7530
-    },
-    {
-      "epoch": 3.927509778357236,
-      "grad_norm": 1.5423554182052612,
-      "learning_rate": 9.29356783919598e-05,
-      "loss": 5.5903,
-      "step": 7531
-    },
-    {
-      "epoch": 3.928031290743155,
-      "grad_norm": 1.6550586223602295,
-      "learning_rate": 9.293467336683418e-05,
-      "loss": 5.5713,
-      "step": 7532
-    },
-    {
-      "epoch": 3.928552803129074,
-      "grad_norm": 1.6271883249282837,
-      "learning_rate": 9.293366834170854e-05,
-      "loss": 5.5668,
-      "step": 7533
-    },
-    {
-      "epoch": 3.9290743155149936,
-      "grad_norm": 1.4324886798858643,
-      "learning_rate": 9.293266331658292e-05,
-      "loss": 5.4394,
-      "step": 7534
-    },
-    {
-      "epoch": 3.9295958279009127,
-      "grad_norm": 1.427464246749878,
-      "learning_rate": 9.29316582914573e-05,
-      "loss": 5.9379,
-      "step": 7535
-    },
-    {
-      "epoch": 3.9301173402868317,
-      "grad_norm": 1.3041108846664429,
-      "learning_rate": 9.293065326633166e-05,
-      "loss": 6.0153,
-      "step": 7536
-    },
-    {
-      "epoch": 3.930638852672751,
-      "grad_norm": 1.4490835666656494,
-      "learning_rate": 9.292964824120604e-05,
-      "loss": 5.9307,
-      "step": 7537
-    },
-    {
-      "epoch": 3.93116036505867,
-      "grad_norm": 1.6295241117477417,
-      "learning_rate": 9.29286432160804e-05,
-      "loss": 5.054,
-      "step": 7538
-    },
-    {
-      "epoch": 3.931681877444589,
-      "grad_norm": 1.4586610794067383,
-      "learning_rate": 9.292763819095478e-05,
-      "loss": 5.9426,
-      "step": 7539
-    },
-    {
-      "epoch": 3.9322033898305087,
-      "grad_norm": 1.8300080299377441,
-      "learning_rate": 9.292663316582915e-05,
-      "loss": 5.0026,
-      "step": 7540
-    },
-    {
-      "epoch": 3.9327249022164277,
-      "grad_norm": 1.4347162246704102,
-      "learning_rate": 9.292562814070352e-05,
-      "loss": 6.0806,
-      "step": 7541
-    },
-    {
-      "epoch": 3.9332464146023467,
-      "grad_norm": 1.4386295080184937,
-      "learning_rate": 9.292462311557789e-05,
-      "loss": 5.8059,
-      "step": 7542
-    },
-    {
-      "epoch": 3.933767926988266,
-      "grad_norm": 1.4731091260910034,
-      "learning_rate": 9.292361809045226e-05,
-      "loss": 5.6268,
-      "step": 7543
-    },
-    {
-      "epoch": 3.934289439374185,
-      "grad_norm": 1.5240801572799683,
-      "learning_rate": 9.292261306532663e-05,
-      "loss": 5.7642,
-      "step": 7544
-    },
-    {
-      "epoch": 3.934810951760104,
-      "grad_norm": 1.5005446672439575,
-      "learning_rate": 9.292160804020101e-05,
-      "loss": 5.4764,
-      "step": 7545
-    },
-    {
-      "epoch": 3.9353324641460237,
-      "grad_norm": 1.4473650455474854,
-      "learning_rate": 9.292060301507538e-05,
-      "loss": 5.8381,
-      "step": 7546
-    },
-    {
-      "epoch": 3.9358539765319427,
-      "grad_norm": 1.4751291275024414,
-      "learning_rate": 9.291959798994976e-05,
-      "loss": 5.7636,
-      "step": 7547
-    },
-    {
-      "epoch": 3.9363754889178617,
-      "grad_norm": 1.3723390102386475,
-      "learning_rate": 9.291859296482413e-05,
-      "loss": 5.882,
-      "step": 7548
-    },
-    {
-      "epoch": 3.936897001303781,
-      "grad_norm": 1.585329294204712,
-      "learning_rate": 9.291758793969849e-05,
-      "loss": 5.4485,
-      "step": 7549
-    },
-    {
-      "epoch": 3.9374185136897,
-      "grad_norm": 1.4724208116531372,
-      "learning_rate": 9.291658291457287e-05,
-      "loss": 5.9485,
-      "step": 7550
-    },
-    {
-      "epoch": 3.937940026075619,
-      "grad_norm": 1.5824873447418213,
-      "learning_rate": 9.291557788944723e-05,
-      "loss": 5.6632,
-      "step": 7551
-    },
-    {
-      "epoch": 3.9384615384615387,
-      "grad_norm": 1.585875391960144,
-      "learning_rate": 9.291457286432161e-05,
-      "loss": 5.5688,
-      "step": 7552
-    },
-    {
-      "epoch": 3.9389830508474577,
-      "grad_norm": 1.4935669898986816,
-      "learning_rate": 9.291356783919598e-05,
-      "loss": 5.7277,
-      "step": 7553
-    },
-    {
-      "epoch": 3.9395045632333767,
-      "grad_norm": 1.6523576974868774,
-      "learning_rate": 9.291256281407035e-05,
-      "loss": 6.1026,
-      "step": 7554
-    },
-    {
-      "epoch": 3.940026075619296,
-      "grad_norm": 1.4205392599105835,
-      "learning_rate": 9.291155778894473e-05,
-      "loss": 5.8633,
-      "step": 7555
-    },
-    {
-      "epoch": 3.940547588005215,
-      "grad_norm": 1.318536400794983,
-      "learning_rate": 9.291055276381911e-05,
-      "loss": 6.0398,
-      "step": 7556
-    },
-    {
-      "epoch": 3.9410691003911342,
-      "grad_norm": 1.6051002740859985,
-      "learning_rate": 9.290954773869347e-05,
-      "loss": 5.284,
-      "step": 7557
-    },
-    {
-      "epoch": 3.9415906127770537,
-      "grad_norm": 1.5022789239883423,
-      "learning_rate": 9.290854271356785e-05,
-      "loss": 5.8473,
-      "step": 7558
-    },
-    {
-      "epoch": 3.9421121251629727,
-      "grad_norm": 1.5657141208648682,
-      "learning_rate": 9.290753768844222e-05,
-      "loss": 5.5074,
-      "step": 7559
-    },
-    {
-      "epoch": 3.9426336375488917,
-      "grad_norm": 1.459274411201477,
-      "learning_rate": 9.290653266331659e-05,
-      "loss": 5.7827,
-      "step": 7560
-    },
-    {
-      "epoch": 3.943155149934811,
-      "grad_norm": 1.405693531036377,
-      "learning_rate": 9.290552763819096e-05,
-      "loss": 5.2736,
-      "step": 7561
-    },
-    {
-      "epoch": 3.94367666232073,
-      "grad_norm": 1.511597752571106,
-      "learning_rate": 9.290452261306532e-05,
-      "loss": 5.6205,
-      "step": 7562
-    },
-    {
-      "epoch": 3.9441981747066492,
-      "grad_norm": 1.6349049806594849,
-      "learning_rate": 9.29035175879397e-05,
-      "loss": 5.7479,
-      "step": 7563
-    },
-    {
-      "epoch": 3.9447196870925687,
-      "grad_norm": 1.4813382625579834,
-      "learning_rate": 9.290251256281406e-05,
-      "loss": 5.7309,
-      "step": 7564
-    },
-    {
-      "epoch": 3.9452411994784877,
-      "grad_norm": 1.4299134016036987,
-      "learning_rate": 9.290150753768844e-05,
-      "loss": 6.0001,
-      "step": 7565
-    },
-    {
-      "epoch": 3.9457627118644067,
-      "grad_norm": 1.4920083284378052,
-      "learning_rate": 9.290050251256282e-05,
-      "loss": 5.8295,
-      "step": 7566
-    },
-    {
-      "epoch": 3.946284224250326,
-      "grad_norm": 1.468429446220398,
-      "learning_rate": 9.28994974874372e-05,
-      "loss": 5.9551,
-      "step": 7567
-    },
-    {
-      "epoch": 3.9468057366362452,
-      "grad_norm": 1.422792673110962,
-      "learning_rate": 9.289849246231156e-05,
-      "loss": 5.9708,
-      "step": 7568
-    },
-    {
-      "epoch": 3.9473272490221643,
-      "grad_norm": 1.548006534576416,
-      "learning_rate": 9.289748743718594e-05,
-      "loss": 5.8878,
-      "step": 7569
-    },
-    {
-      "epoch": 3.9478487614080837,
-      "grad_norm": 1.5350459814071655,
-      "learning_rate": 9.28964824120603e-05,
-      "loss": 5.713,
-      "step": 7570
-    },
-    {
-      "epoch": 3.9483702737940027,
-      "grad_norm": 1.5185812711715698,
-      "learning_rate": 9.289547738693468e-05,
-      "loss": 6.029,
-      "step": 7571
-    },
-    {
-      "epoch": 3.9488917861799218,
-      "grad_norm": 1.4402681589126587,
-      "learning_rate": 9.289447236180905e-05,
-      "loss": 5.7709,
-      "step": 7572
-    },
-    {
-      "epoch": 3.949413298565841,
-      "grad_norm": 1.627800464630127,
-      "learning_rate": 9.289346733668342e-05,
-      "loss": 5.2502,
-      "step": 7573
-    },
-    {
-      "epoch": 3.9499348109517602,
-      "grad_norm": 1.446333885192871,
-      "learning_rate": 9.289246231155779e-05,
-      "loss": 6.0545,
-      "step": 7574
-    },
-    {
-      "epoch": 3.9504563233376793,
-      "grad_norm": 1.5081079006195068,
-      "learning_rate": 9.289145728643217e-05,
-      "loss": 5.4392,
-      "step": 7575
-    },
-    {
-      "epoch": 3.9509778357235983,
-      "grad_norm": 1.4688866138458252,
-      "learning_rate": 9.289045226130654e-05,
-      "loss": 6.1615,
-      "step": 7576
-    },
-    {
-      "epoch": 3.9514993481095178,
-      "grad_norm": 1.60893714427948,
-      "learning_rate": 9.288944723618091e-05,
-      "loss": 5.0862,
-      "step": 7577
-    },
-    {
-      "epoch": 3.9520208604954368,
-      "grad_norm": 1.5656142234802246,
-      "learning_rate": 9.288844221105529e-05,
-      "loss": 5.4297,
-      "step": 7578
-    },
-    {
-      "epoch": 3.952542372881356,
-      "grad_norm": 1.430444359779358,
-      "learning_rate": 9.288743718592965e-05,
-      "loss": 6.1117,
-      "step": 7579
-    },
-    {
-      "epoch": 3.9530638852672753,
-      "grad_norm": 1.396181583404541,
-      "learning_rate": 9.288643216080403e-05,
-      "loss": 6.0451,
-      "step": 7580
-    },
-    {
-      "epoch": 3.9535853976531943,
-      "grad_norm": 1.4671739339828491,
-      "learning_rate": 9.288542713567839e-05,
-      "loss": 5.6149,
-      "step": 7581
-    },
-    {
-      "epoch": 3.9541069100391133,
-      "grad_norm": 1.406620740890503,
-      "learning_rate": 9.288442211055277e-05,
-      "loss": 5.9762,
-      "step": 7582
-    },
-    {
-      "epoch": 3.9546284224250323,
-      "grad_norm": 1.5574852228164673,
-      "learning_rate": 9.288341708542713e-05,
-      "loss": 5.4051,
-      "step": 7583
-    },
-    {
-      "epoch": 3.955149934810952,
-      "grad_norm": 1.5554401874542236,
-      "learning_rate": 9.288241206030151e-05,
-      "loss": 5.6875,
-      "step": 7584
-    },
-    {
-      "epoch": 3.955671447196871,
-      "grad_norm": 1.449561595916748,
-      "learning_rate": 9.288140703517588e-05,
-      "loss": 6.2057,
-      "step": 7585
-    },
-    {
-      "epoch": 3.95619295958279,
-      "grad_norm": 1.4294153451919556,
-      "learning_rate": 9.288040201005025e-05,
-      "loss": 5.3888,
-      "step": 7586
-    },
-    {
-      "epoch": 3.9567144719687093,
-      "grad_norm": 1.441821575164795,
-      "learning_rate": 9.287939698492463e-05,
-      "loss": 5.9233,
-      "step": 7587
-    },
-    {
-      "epoch": 3.9572359843546283,
-      "grad_norm": 1.5153042078018188,
-      "learning_rate": 9.287839195979901e-05,
-      "loss": 5.5207,
-      "step": 7588
-    },
-    {
-      "epoch": 3.9577574967405473,
-      "grad_norm": 1.571147084236145,
-      "learning_rate": 9.287738693467337e-05,
-      "loss": 5.713,
-      "step": 7589
-    },
-    {
-      "epoch": 3.958279009126467,
-      "grad_norm": 1.419402837753296,
-      "learning_rate": 9.287638190954774e-05,
-      "loss": 5.6989,
-      "step": 7590
-    },
-    {
-      "epoch": 3.958800521512386,
-      "grad_norm": 1.5178238153457642,
-      "learning_rate": 9.287537688442212e-05,
-      "loss": 5.8249,
-      "step": 7591
-    },
-    {
-      "epoch": 3.959322033898305,
-      "grad_norm": 1.6613214015960693,
-      "learning_rate": 9.287437185929648e-05,
-      "loss": 5.9029,
-      "step": 7592
-    },
-    {
-      "epoch": 3.9598435462842243,
-      "grad_norm": 1.4186713695526123,
-      "learning_rate": 9.287336683417086e-05,
-      "loss": 5.6948,
-      "step": 7593
-    },
-    {
-      "epoch": 3.9603650586701433,
-      "grad_norm": 1.4362927675247192,
-      "learning_rate": 9.287236180904522e-05,
-      "loss": 4.9006,
-      "step": 7594
-    },
-    {
-      "epoch": 3.9608865710560623,
-      "grad_norm": 1.539212703704834,
-      "learning_rate": 9.28713567839196e-05,
-      "loss": 5.3297,
-      "step": 7595
-    },
-    {
-      "epoch": 3.961408083441982,
-      "grad_norm": 1.5027996301651,
-      "learning_rate": 9.287035175879398e-05,
-      "loss": 5.7344,
-      "step": 7596
-    },
-    {
-      "epoch": 3.961929595827901,
-      "grad_norm": 1.4980460405349731,
-      "learning_rate": 9.286934673366836e-05,
-      "loss": 5.6767,
-      "step": 7597
-    },
-    {
-      "epoch": 3.96245110821382,
-      "grad_norm": 1.6255090236663818,
-      "learning_rate": 9.286834170854272e-05,
-      "loss": 5.645,
-      "step": 7598
-    },
-    {
-      "epoch": 3.9629726205997393,
-      "grad_norm": 1.4851247072219849,
-      "learning_rate": 9.28673366834171e-05,
-      "loss": 5.7222,
-      "step": 7599
-    },
-    {
-      "epoch": 3.9634941329856583,
-      "grad_norm": 1.484117865562439,
-      "learning_rate": 9.286633165829146e-05,
-      "loss": 5.0496,
-      "step": 7600
-    },
-    {
-      "epoch": 3.9640156453715774,
-      "grad_norm": 1.484295129776001,
-      "learning_rate": 9.286532663316584e-05,
-      "loss": 5.9216,
-      "step": 7601
-    },
-    {
-      "epoch": 3.964537157757497,
-      "grad_norm": 1.51368248462677,
-      "learning_rate": 9.28643216080402e-05,
-      "loss": 6.0834,
-      "step": 7602
-    },
-    {
-      "epoch": 3.965058670143416,
-      "grad_norm": 1.5821075439453125,
-      "learning_rate": 9.286331658291457e-05,
-      "loss": 5.5843,
-      "step": 7603
-    },
-    {
-      "epoch": 3.965580182529335,
-      "grad_norm": 1.4412554502487183,
-      "learning_rate": 9.286231155778895e-05,
-      "loss": 5.9201,
-      "step": 7604
-    },
-    {
-      "epoch": 3.9661016949152543,
-      "grad_norm": 1.4975872039794922,
-      "learning_rate": 9.286130653266331e-05,
-      "loss": 5.4509,
-      "step": 7605
-    },
-    {
-      "epoch": 3.9666232073011733,
-      "grad_norm": 1.629420518875122,
-      "learning_rate": 9.286030150753769e-05,
-      "loss": 5.0407,
-      "step": 7606
-    },
-    {
-      "epoch": 3.9671447196870924,
-      "grad_norm": 1.5303471088409424,
-      "learning_rate": 9.285929648241207e-05,
-      "loss": 5.9024,
-      "step": 7607
-    },
-    {
-      "epoch": 3.967666232073012,
-      "grad_norm": 1.8433928489685059,
-      "learning_rate": 9.285829145728644e-05,
-      "loss": 5.3276,
-      "step": 7608
-    },
-    {
-      "epoch": 3.968187744458931,
-      "grad_norm": 1.5610548257827759,
-      "learning_rate": 9.285728643216081e-05,
-      "loss": 5.7488,
-      "step": 7609
-    },
-    {
-      "epoch": 3.96870925684485,
-      "grad_norm": 1.3613871335983276,
-      "learning_rate": 9.285628140703519e-05,
-      "loss": 5.9179,
-      "step": 7610
-    },
-    {
-      "epoch": 3.9692307692307693,
-      "grad_norm": 1.537189245223999,
-      "learning_rate": 9.285527638190955e-05,
-      "loss": 5.7852,
-      "step": 7611
-    },
-    {
-      "epoch": 3.9697522816166884,
-      "grad_norm": 1.3163654804229736,
-      "learning_rate": 9.285427135678393e-05,
-      "loss": 5.898,
-      "step": 7612
-    },
-    {
-      "epoch": 3.9702737940026074,
-      "grad_norm": 1.590118408203125,
-      "learning_rate": 9.285326633165829e-05,
-      "loss": 5.6266,
-      "step": 7613
-    },
-    {
-      "epoch": 3.970795306388527,
-      "grad_norm": 1.457840085029602,
-      "learning_rate": 9.285226130653267e-05,
-      "loss": 5.5584,
-      "step": 7614
-    },
-    {
-      "epoch": 3.971316818774446,
-      "grad_norm": 1.3191081285476685,
-      "learning_rate": 9.285125628140703e-05,
-      "loss": 6.1224,
-      "step": 7615
-    },
-    {
-      "epoch": 3.971838331160365,
-      "grad_norm": 1.3947561979293823,
-      "learning_rate": 9.285025125628141e-05,
-      "loss": 5.7126,
-      "step": 7616
-    },
-    {
-      "epoch": 3.9723598435462844,
-      "grad_norm": 1.405626893043518,
-      "learning_rate": 9.284924623115579e-05,
-      "loss": 5.4398,
-      "step": 7617
-    },
-    {
-      "epoch": 3.9728813559322034,
-      "grad_norm": 1.3092596530914307,
-      "learning_rate": 9.284824120603015e-05,
-      "loss": 6.0616,
-      "step": 7618
-    },
-    {
-      "epoch": 3.9734028683181224,
-      "grad_norm": 1.5253963470458984,
-      "learning_rate": 9.284723618090453e-05,
-      "loss": 5.5582,
-      "step": 7619
-    },
-    {
-      "epoch": 3.973924380704042,
-      "grad_norm": 1.3398442268371582,
-      "learning_rate": 9.28462311557789e-05,
-      "loss": 5.2515,
-      "step": 7620
-    },
-    {
-      "epoch": 3.974445893089961,
-      "grad_norm": 1.4413524866104126,
-      "learning_rate": 9.284522613065327e-05,
-      "loss": 5.5538,
-      "step": 7621
-    },
-    {
-      "epoch": 3.97496740547588,
-      "grad_norm": 1.470799207687378,
-      "learning_rate": 9.284422110552764e-05,
-      "loss": 5.2757,
-      "step": 7622
-    },
-    {
-      "epoch": 3.9754889178617994,
-      "grad_norm": 1.5957432985305786,
-      "learning_rate": 9.284321608040202e-05,
-      "loss": 5.6332,
-      "step": 7623
-    },
-    {
-      "epoch": 3.9760104302477184,
-      "grad_norm": 1.3365963697433472,
-      "learning_rate": 9.284221105527638e-05,
-      "loss": 5.8816,
-      "step": 7624
-    },
-    {
-      "epoch": 3.9765319426336374,
-      "grad_norm": 1.533018946647644,
-      "learning_rate": 9.284120603015076e-05,
-      "loss": 5.6524,
-      "step": 7625
-    },
-    {
-      "epoch": 3.977053455019557,
-      "grad_norm": 1.4953924417495728,
-      "learning_rate": 9.284020100502512e-05,
-      "loss": 6.1326,
-      "step": 7626
-    },
-    {
-      "epoch": 3.977574967405476,
-      "grad_norm": 1.4917718172073364,
-      "learning_rate": 9.28391959798995e-05,
-      "loss": 5.7034,
-      "step": 7627
-    },
-    {
-      "epoch": 3.978096479791395,
-      "grad_norm": 1.8134647607803345,
-      "learning_rate": 9.283819095477388e-05,
-      "loss": 5.4472,
-      "step": 7628
-    },
-    {
-      "epoch": 3.9786179921773144,
-      "grad_norm": 1.3935800790786743,
-      "learning_rate": 9.283718592964824e-05,
-      "loss": 5.8886,
-      "step": 7629
-    },
-    {
-      "epoch": 3.9791395045632334,
-      "grad_norm": 1.4648017883300781,
-      "learning_rate": 9.283618090452262e-05,
-      "loss": 5.8747,
-      "step": 7630
-    },
-    {
-      "epoch": 3.9796610169491524,
-      "grad_norm": 1.7654238939285278,
-      "learning_rate": 9.283517587939699e-05,
-      "loss": 6.0085,
-      "step": 7631
-    },
-    {
-      "epoch": 3.980182529335072,
-      "grad_norm": 1.5801446437835693,
-      "learning_rate": 9.283417085427136e-05,
-      "loss": 5.9736,
-      "step": 7632
-    },
-    {
-      "epoch": 3.980704041720991,
-      "grad_norm": 1.424108862876892,
-      "learning_rate": 9.283316582914573e-05,
-      "loss": 6.044,
-      "step": 7633
-    },
-    {
-      "epoch": 3.98122555410691,
-      "grad_norm": 1.3887425661087036,
-      "learning_rate": 9.28321608040201e-05,
-      "loss": 5.5715,
-      "step": 7634
-    },
-    {
-      "epoch": 3.9817470664928294,
-      "grad_norm": 1.8764564990997314,
-      "learning_rate": 9.283115577889447e-05,
-      "loss": 4.9603,
-      "step": 7635
-    },
-    {
-      "epoch": 3.9822685788787484,
-      "grad_norm": 1.5525306463241577,
-      "learning_rate": 9.283015075376885e-05,
-      "loss": 5.6231,
-      "step": 7636
-    },
-    {
-      "epoch": 3.9827900912646674,
-      "grad_norm": 1.5488239526748657,
-      "learning_rate": 9.282914572864323e-05,
-      "loss": 5.716,
-      "step": 7637
-    },
-    {
-      "epoch": 3.983311603650587,
-      "grad_norm": 1.4555163383483887,
-      "learning_rate": 9.28281407035176e-05,
-      "loss": 6.0467,
-      "step": 7638
-    },
-    {
-      "epoch": 3.983833116036506,
-      "grad_norm": 1.4187097549438477,
-      "learning_rate": 9.282713567839197e-05,
-      "loss": 5.8483,
-      "step": 7639
-    },
-    {
-      "epoch": 3.984354628422425,
-      "grad_norm": 1.5016056299209595,
-      "learning_rate": 9.282613065326635e-05,
-      "loss": 5.8125,
-      "step": 7640
-    },
-    {
-      "epoch": 3.9848761408083444,
-      "grad_norm": 1.5796687602996826,
-      "learning_rate": 9.282512562814071e-05,
-      "loss": 5.6155,
-      "step": 7641
-    },
-    {
-      "epoch": 3.9853976531942634,
-      "grad_norm": 1.368722915649414,
-      "learning_rate": 9.282412060301507e-05,
-      "loss": 5.9012,
-      "step": 7642
-    },
-    {
-      "epoch": 3.9859191655801824,
-      "grad_norm": 1.3489478826522827,
-      "learning_rate": 9.282311557788945e-05,
-      "loss": 6.0632,
-      "step": 7643
-    },
-    {
-      "epoch": 3.986440677966102,
-      "grad_norm": 1.5053138732910156,
-      "learning_rate": 9.282211055276382e-05,
-      "loss": 5.1006,
-      "step": 7644
-    },
-    {
-      "epoch": 3.986962190352021,
-      "grad_norm": 1.7075389623641968,
-      "learning_rate": 9.28211055276382e-05,
-      "loss": 5.4289,
-      "step": 7645
-    },
-    {
-      "epoch": 3.98748370273794,
-      "grad_norm": 1.3612264394760132,
-      "learning_rate": 9.282010050251256e-05,
-      "loss": 5.8387,
-      "step": 7646
-    },
-    {
-      "epoch": 3.9880052151238594,
-      "grad_norm": 1.3387603759765625,
-      "learning_rate": 9.281909547738694e-05,
-      "loss": 5.8833,
-      "step": 7647
-    },
-    {
-      "epoch": 3.9885267275097784,
-      "grad_norm": 1.4316917657852173,
-      "learning_rate": 9.281809045226131e-05,
-      "loss": 5.7647,
-      "step": 7648
-    },
-    {
-      "epoch": 3.9890482398956975,
-      "grad_norm": 1.4199106693267822,
-      "learning_rate": 9.281708542713569e-05,
-      "loss": 5.6353,
-      "step": 7649
-    },
-    {
-      "epoch": 3.989569752281617,
-      "grad_norm": 1.523295283317566,
-      "learning_rate": 9.281608040201006e-05,
-      "loss": 5.4034,
-      "step": 7650
-    },
-    {
-      "epoch": 3.990091264667536,
-      "grad_norm": 1.4094607830047607,
-      "learning_rate": 9.281507537688443e-05,
-      "loss": 5.8889,
-      "step": 7651
-    },
-    {
-      "epoch": 3.990612777053455,
-      "grad_norm": 1.4275928735733032,
-      "learning_rate": 9.28140703517588e-05,
-      "loss": 5.5642,
-      "step": 7652
-    },
-    {
-      "epoch": 3.9911342894393744,
-      "grad_norm": 1.4821397066116333,
-      "learning_rate": 9.281306532663318e-05,
-      "loss": 5.8514,
-      "step": 7653
-    },
-    {
-      "epoch": 3.9916558018252934,
-      "grad_norm": 1.406541109085083,
-      "learning_rate": 9.281206030150754e-05,
-      "loss": 5.6963,
-      "step": 7654
-    },
-    {
-      "epoch": 3.9921773142112125,
-      "grad_norm": 1.3825905323028564,
-      "learning_rate": 9.28110552763819e-05,
-      "loss": 5.9992,
-      "step": 7655
-    },
-    {
-      "epoch": 3.992698826597132,
-      "grad_norm": 1.6371179819107056,
-      "learning_rate": 9.281005025125628e-05,
-      "loss": 5.877,
-      "step": 7656
-    },
-    {
-      "epoch": 3.993220338983051,
-      "grad_norm": 1.6671850681304932,
-      "learning_rate": 9.280904522613066e-05,
-      "loss": 5.1487,
-      "step": 7657
-    },
-    {
-      "epoch": 3.99374185136897,
-      "grad_norm": 1.4873112440109253,
-      "learning_rate": 9.280804020100504e-05,
-      "loss": 5.6765,
-      "step": 7658
-    },
-    {
-      "epoch": 3.9942633637548894,
-      "grad_norm": 1.481482982635498,
-      "learning_rate": 9.28070351758794e-05,
-      "loss": 5.5725,
-      "step": 7659
-    },
-    {
-      "epoch": 3.9947848761408085,
-      "grad_norm": 1.6333673000335693,
-      "learning_rate": 9.280603015075378e-05,
-      "loss": 5.7679,
-      "step": 7660
-    },
-    {
-      "epoch": 3.9953063885267275,
-      "grad_norm": 1.4642921686172485,
-      "learning_rate": 9.280502512562814e-05,
-      "loss": 5.5233,
-      "step": 7661
-    },
-    {
-      "epoch": 3.995827900912647,
-      "grad_norm": 1.3596277236938477,
-      "learning_rate": 9.280402010050252e-05,
-      "loss": 5.7152,
-      "step": 7662
-    },
-    {
-      "epoch": 3.996349413298566,
-      "grad_norm": 1.4807305335998535,
-      "learning_rate": 9.280301507537689e-05,
-      "loss": 5.6184,
-      "step": 7663
-    },
-    {
-      "epoch": 3.996870925684485,
-      "grad_norm": 1.4810280799865723,
-      "learning_rate": 9.280201005025126e-05,
-      "loss": 5.3688,
-      "step": 7664
-    },
-    {
-      "epoch": 3.997392438070404,
-      "grad_norm": 1.5020126104354858,
-      "learning_rate": 9.280100502512563e-05,
-      "loss": 5.7442,
-      "step": 7665
-    },
-    {
-      "epoch": 3.9979139504563235,
-      "grad_norm": 1.5125086307525635,
-      "learning_rate": 9.28e-05,
-      "loss": 5.6178,
-      "step": 7666
-    },
-    {
-      "epoch": 3.9984354628422425,
-      "grad_norm": 1.5281710624694824,
-      "learning_rate": 9.279899497487437e-05,
-      "loss": 5.647,
-      "step": 7667
-    },
-    {
-      "epoch": 3.9989569752281615,
-      "grad_norm": 1.6058510541915894,
-      "learning_rate": 9.279798994974875e-05,
-      "loss": 5.0744,
-      "step": 7668
-    },
-    {
-      "epoch": 3.999478487614081,
-      "grad_norm": 1.623013973236084,
-      "learning_rate": 9.279698492462313e-05,
-      "loss": 5.2234,
-      "step": 7669
-    },
-    {
-      "epoch": 4.0,
-      "grad_norm": 1.7330044507980347,
-      "learning_rate": 9.279597989949749e-05,
-      "loss": 5.5358,
-      "step": 7670
-    },
-    {
-      "epoch": 4.0005215123859195,
-      "grad_norm": 1.608900785446167,
-      "learning_rate": 9.279497487437187e-05,
-      "loss": 5.4956,
-      "step": 7671
-    },
-    {
-      "epoch": 4.001043024771838,
-      "grad_norm": 1.503663182258606,
-      "learning_rate": 9.279396984924623e-05,
-      "loss": 5.6129,
-      "step": 7672
-    },
-    {
-      "epoch": 4.0015645371577575,
-      "grad_norm": 1.5316232442855835,
-      "learning_rate": 9.279296482412061e-05,
-      "loss": 5.455,
-      "step": 7673
-    },
-    {
-      "epoch": 4.002086049543677,
-      "grad_norm": 1.6079626083374023,
-      "learning_rate": 9.279195979899497e-05,
-      "loss": 5.8496,
-      "step": 7674
-    },
-    {
-      "epoch": 4.0026075619295955,
-      "grad_norm": 1.5660747289657593,
-      "learning_rate": 9.279095477386935e-05,
-      "loss": 5.5557,
-      "step": 7675
-    },
-    {
-      "epoch": 4.003129074315515,
-      "grad_norm": 1.5197049379348755,
-      "learning_rate": 9.278994974874372e-05,
-      "loss": 5.859,
-      "step": 7676
-    },
-    {
-      "epoch": 4.0036505867014345,
-      "grad_norm": 1.54172945022583,
-      "learning_rate": 9.27889447236181e-05,
-      "loss": 5.6918,
-      "step": 7677
-    },
-    {
-      "epoch": 4.004172099087353,
-      "grad_norm": 1.5502054691314697,
-      "learning_rate": 9.278793969849247e-05,
-      "loss": 5.5953,
-      "step": 7678
-    },
-    {
-      "epoch": 4.0046936114732725,
-      "grad_norm": 1.424559235572815,
-      "learning_rate": 9.278693467336685e-05,
-      "loss": 5.894,
-      "step": 7679
-    },
-    {
-      "epoch": 4.005215123859192,
-      "grad_norm": 1.5219298601150513,
-      "learning_rate": 9.278592964824121e-05,
-      "loss": 5.5015,
-      "step": 7680
-    },
-    {
-      "epoch": 4.005736636245111,
-      "grad_norm": 1.484984278678894,
-      "learning_rate": 9.278492462311559e-05,
-      "loss": 5.6447,
-      "step": 7681
-    },
-    {
-      "epoch": 4.00625814863103,
-      "grad_norm": 1.5557782649993896,
-      "learning_rate": 9.278391959798996e-05,
-      "loss": 5.4462,
-      "step": 7682
-    },
-    {
-      "epoch": 4.0067796610169495,
-      "grad_norm": 1.6625183820724487,
-      "learning_rate": 9.278291457286432e-05,
-      "loss": 5.2324,
-      "step": 7683
-    },
-    {
-      "epoch": 4.007301173402868,
-      "grad_norm": 1.4944870471954346,
-      "learning_rate": 9.27819095477387e-05,
-      "loss": 5.8019,
-      "step": 7684
-    },
-    {
-      "epoch": 4.0078226857887875,
-      "grad_norm": 1.5170711278915405,
-      "learning_rate": 9.278090452261306e-05,
-      "loss": 5.6584,
-      "step": 7685
-    },
-    {
-      "epoch": 4.008344198174707,
-      "grad_norm": 1.423714280128479,
-      "learning_rate": 9.277989949748744e-05,
-      "loss": 5.9605,
-      "step": 7686
-    },
-    {
-      "epoch": 4.008865710560626,
-      "grad_norm": 1.6840181350708008,
-      "learning_rate": 9.27788944723618e-05,
-      "loss": 5.4314,
-      "step": 7687
-    },
-    {
-      "epoch": 4.009387222946545,
-      "grad_norm": 1.5774157047271729,
-      "learning_rate": 9.277788944723618e-05,
-      "loss": 5.2665,
-      "step": 7688
-    },
-    {
-      "epoch": 4.0099087353324645,
-      "grad_norm": 1.5384485721588135,
-      "learning_rate": 9.277688442211056e-05,
-      "loss": 5.6497,
-      "step": 7689
-    },
-    {
-      "epoch": 4.010430247718383,
-      "grad_norm": 1.4617236852645874,
-      "learning_rate": 9.277587939698494e-05,
-      "loss": 5.6907,
-      "step": 7690
-    },
-    {
-      "epoch": 4.0109517601043025,
-      "grad_norm": 1.4371092319488525,
-      "learning_rate": 9.27748743718593e-05,
-      "loss": 5.6221,
-      "step": 7691
-    },
-    {
-      "epoch": 4.011473272490222,
-      "grad_norm": 1.3725258111953735,
-      "learning_rate": 9.277386934673368e-05,
-      "loss": 6.0872,
-      "step": 7692
-    },
-    {
-      "epoch": 4.011994784876141,
-      "grad_norm": 1.4017165899276733,
-      "learning_rate": 9.277286432160804e-05,
-      "loss": 5.6827,
-      "step": 7693
-    },
-    {
-      "epoch": 4.01251629726206,
-      "grad_norm": 1.4511051177978516,
-      "learning_rate": 9.277185929648242e-05,
-      "loss": 5.8644,
-      "step": 7694
-    },
-    {
-      "epoch": 4.0130378096479795,
-      "grad_norm": 1.7444595098495483,
-      "learning_rate": 9.277085427135679e-05,
-      "loss": 5.6013,
-      "step": 7695
-    },
-    {
-      "epoch": 4.013559322033898,
-      "grad_norm": 1.5286741256713867,
-      "learning_rate": 9.276984924623115e-05,
-      "loss": 5.2085,
-      "step": 7696
-    },
-    {
-      "epoch": 4.014080834419818,
-      "grad_norm": 1.4398947954177856,
-      "learning_rate": 9.276884422110553e-05,
-      "loss": 5.8112,
-      "step": 7697
-    },
-    {
-      "epoch": 4.014602346805737,
-      "grad_norm": 1.4774574041366577,
-      "learning_rate": 9.276783919597989e-05,
-      "loss": 5.8915,
-      "step": 7698
-    },
-    {
-      "epoch": 4.015123859191656,
-      "grad_norm": 1.4351156949996948,
-      "learning_rate": 9.276683417085427e-05,
-      "loss": 5.9567,
-      "step": 7699
-    },
-    {
-      "epoch": 4.015645371577575,
-      "grad_norm": 1.4024503231048584,
-      "learning_rate": 9.276582914572865e-05,
-      "loss": 5.6869,
-      "step": 7700
-    },
-    {
-      "epoch": 4.0161668839634945,
-      "grad_norm": 1.6496344804763794,
-      "learning_rate": 9.276482412060303e-05,
-      "loss": 5.6466,
-      "step": 7701
-    },
-    {
-      "epoch": 4.016688396349413,
-      "grad_norm": 1.4414211511611938,
-      "learning_rate": 9.276381909547739e-05,
-      "loss": 5.9198,
-      "step": 7702
-    },
-    {
-      "epoch": 4.017209908735333,
-      "grad_norm": 1.5342429876327515,
-      "learning_rate": 9.276281407035177e-05,
-      "loss": 5.4157,
-      "step": 7703
-    },
-    {
-      "epoch": 4.017731421121252,
-      "grad_norm": 1.4718905687332153,
-      "learning_rate": 9.276180904522613e-05,
-      "loss": 5.7798,
-      "step": 7704
-    },
-    {
-      "epoch": 4.018252933507171,
-      "grad_norm": 1.5217394828796387,
-      "learning_rate": 9.276080402010051e-05,
-      "loss": 5.8179,
-      "step": 7705
-    },
-    {
-      "epoch": 4.01877444589309,
-      "grad_norm": 1.467195987701416,
-      "learning_rate": 9.275979899497488e-05,
-      "loss": 5.5417,
-      "step": 7706
-    },
-    {
-      "epoch": 4.0192959582790095,
-      "grad_norm": 1.5121022462844849,
-      "learning_rate": 9.275879396984925e-05,
-      "loss": 5.8053,
-      "step": 7707
-    },
-    {
-      "epoch": 4.019817470664928,
-      "grad_norm": 1.4933853149414062,
-      "learning_rate": 9.275778894472362e-05,
-      "loss": 5.6759,
-      "step": 7708
-    },
-    {
-      "epoch": 4.020338983050848,
-      "grad_norm": 1.4149904251098633,
-      "learning_rate": 9.2756783919598e-05,
-      "loss": 5.8774,
-      "step": 7709
-    },
-    {
-      "epoch": 4.020860495436767,
-      "grad_norm": 1.48702073097229,
-      "learning_rate": 9.275577889447237e-05,
-      "loss": 5.6804,
-      "step": 7710
-    },
-    {
-      "epoch": 4.021382007822686,
-      "grad_norm": 1.4852831363677979,
-      "learning_rate": 9.275477386934674e-05,
-      "loss": 5.4904,
-      "step": 7711
-    },
-    {
-      "epoch": 4.021903520208605,
-      "grad_norm": 1.5853972434997559,
-      "learning_rate": 9.275376884422111e-05,
-      "loss": 5.4615,
-      "step": 7712
-    },
-    {
-      "epoch": 4.0224250325945246,
-      "grad_norm": 1.7149549722671509,
-      "learning_rate": 9.275276381909548e-05,
-      "loss": 5.9469,
-      "step": 7713
-    },
-    {
-      "epoch": 4.022946544980443,
-      "grad_norm": 1.577808141708374,
-      "learning_rate": 9.275175879396986e-05,
-      "loss": 5.6752,
-      "step": 7714
-    },
-    {
-      "epoch": 4.023468057366363,
-      "grad_norm": 1.4081953763961792,
-      "learning_rate": 9.275075376884422e-05,
-      "loss": 6.0338,
-      "step": 7715
-    },
-    {
-      "epoch": 4.023989569752282,
-      "grad_norm": 1.3134089708328247,
-      "learning_rate": 9.27497487437186e-05,
-      "loss": 6.0395,
-      "step": 7716
-    },
-    {
-      "epoch": 4.024511082138201,
-      "grad_norm": 1.3429710865020752,
-      "learning_rate": 9.274874371859296e-05,
-      "loss": 5.9513,
-      "step": 7717
-    },
-    {
-      "epoch": 4.02503259452412,
-      "grad_norm": 1.4886252880096436,
-      "learning_rate": 9.274773869346734e-05,
-      "loss": 5.949,
-      "step": 7718
-    },
-    {
-      "epoch": 4.025554106910039,
-      "grad_norm": 1.3194868564605713,
-      "learning_rate": 9.27467336683417e-05,
-      "loss": 6.03,
-      "step": 7719
-    },
-    {
-      "epoch": 4.026075619295958,
-      "grad_norm": 1.3274924755096436,
-      "learning_rate": 9.274572864321608e-05,
-      "loss": 6.1435,
-      "step": 7720
-    },
-    {
-      "epoch": 4.026597131681878,
-      "grad_norm": 1.4881117343902588,
-      "learning_rate": 9.274472361809046e-05,
-      "loss": 5.7653,
-      "step": 7721
-    },
-    {
-      "epoch": 4.027118644067796,
-      "grad_norm": 1.5620120763778687,
-      "learning_rate": 9.274371859296483e-05,
-      "loss": 5.6183,
-      "step": 7722
-    },
-    {
-      "epoch": 4.027640156453716,
-      "grad_norm": 1.4780586957931519,
-      "learning_rate": 9.27427135678392e-05,
-      "loss": 5.7696,
-      "step": 7723
-    },
-    {
-      "epoch": 4.028161668839635,
-      "grad_norm": 1.4265094995498657,
-      "learning_rate": 9.274170854271357e-05,
-      "loss": 5.4559,
-      "step": 7724
-    },
-    {
-      "epoch": 4.028683181225554,
-      "grad_norm": 1.5559715032577515,
-      "learning_rate": 9.274070351758795e-05,
-      "loss": 5.5493,
-      "step": 7725
-    },
-    {
-      "epoch": 4.029204693611473,
-      "grad_norm": 1.4153403043746948,
-      "learning_rate": 9.273969849246231e-05,
-      "loss": 5.916,
-      "step": 7726
-    },
-    {
-      "epoch": 4.029726205997393,
-      "grad_norm": 1.4770768880844116,
-      "learning_rate": 9.273869346733669e-05,
-      "loss": 5.3997,
-      "step": 7727
-    },
-    {
-      "epoch": 4.030247718383311,
-      "grad_norm": 1.426910400390625,
-      "learning_rate": 9.273768844221105e-05,
-      "loss": 5.9063,
-      "step": 7728
-    },
-    {
-      "epoch": 4.030769230769231,
-      "grad_norm": 1.796269178390503,
-      "learning_rate": 9.273668341708543e-05,
-      "loss": 5.6479,
-      "step": 7729
-    },
-    {
-      "epoch": 4.03129074315515,
-      "grad_norm": 1.398111343383789,
-      "learning_rate": 9.273567839195981e-05,
-      "loss": 5.7758,
-      "step": 7730
-    },
-    {
-      "epoch": 4.031812255541069,
-      "grad_norm": 1.3928663730621338,
-      "learning_rate": 9.273467336683419e-05,
-      "loss": 5.7595,
-      "step": 7731
-    },
-    {
-      "epoch": 4.032333767926988,
-      "grad_norm": 1.5666611194610596,
-      "learning_rate": 9.273366834170855e-05,
-      "loss": 5.4759,
-      "step": 7732
-    },
-    {
-      "epoch": 4.032855280312908,
-      "grad_norm": 1.4694808721542358,
-      "learning_rate": 9.273266331658293e-05,
-      "loss": 6.1685,
-      "step": 7733
-    },
-    {
-      "epoch": 4.033376792698826,
-      "grad_norm": 1.5558421611785889,
-      "learning_rate": 9.273165829145729e-05,
-      "loss": 5.801,
-      "step": 7734
-    },
-    {
-      "epoch": 4.033898305084746,
-      "grad_norm": 1.416417121887207,
-      "learning_rate": 9.273065326633166e-05,
-      "loss": 5.8836,
-      "step": 7735
-    },
-    {
-      "epoch": 4.034419817470665,
-      "grad_norm": 1.4595131874084473,
-      "learning_rate": 9.272964824120603e-05,
-      "loss": 5.9976,
-      "step": 7736
-    },
-    {
-      "epoch": 4.034941329856584,
-      "grad_norm": 1.4396376609802246,
-      "learning_rate": 9.27286432160804e-05,
-      "loss": 5.7767,
-      "step": 7737
-    },
-    {
-      "epoch": 4.035462842242503,
-      "grad_norm": 1.5201600790023804,
-      "learning_rate": 9.272763819095478e-05,
-      "loss": 5.4103,
-      "step": 7738
-    },
-    {
-      "epoch": 4.035984354628423,
-      "grad_norm": 1.4653089046478271,
-      "learning_rate": 9.272663316582914e-05,
-      "loss": 5.6577,
-      "step": 7739
-    },
-    {
-      "epoch": 4.036505867014341,
-      "grad_norm": 1.4416295289993286,
-      "learning_rate": 9.272562814070352e-05,
-      "loss": 5.8298,
-      "step": 7740
-    },
-    {
-      "epoch": 4.037027379400261,
-      "grad_norm": 1.5999308824539185,
-      "learning_rate": 9.27246231155779e-05,
-      "loss": 5.5366,
-      "step": 7741
-    },
-    {
-      "epoch": 4.03754889178618,
-      "grad_norm": 1.4435508251190186,
-      "learning_rate": 9.272361809045227e-05,
-      "loss": 5.8591,
-      "step": 7742
-    },
-    {
-      "epoch": 4.038070404172099,
-      "grad_norm": 1.6125752925872803,
-      "learning_rate": 9.272261306532664e-05,
-      "loss": 5.5196,
-      "step": 7743
-    },
-    {
-      "epoch": 4.038591916558018,
-      "grad_norm": 1.7814494371414185,
-      "learning_rate": 9.272160804020102e-05,
-      "loss": 5.7281,
-      "step": 7744
-    },
-    {
-      "epoch": 4.039113428943938,
-      "grad_norm": 1.517842411994934,
-      "learning_rate": 9.272060301507538e-05,
-      "loss": 5.4989,
-      "step": 7745
-    },
-    {
-      "epoch": 4.039634941329856,
-      "grad_norm": 1.4527209997177124,
-      "learning_rate": 9.271959798994976e-05,
-      "loss": 6.0198,
-      "step": 7746
-    },
-    {
-      "epoch": 4.040156453715776,
-      "grad_norm": 1.3927206993103027,
-      "learning_rate": 9.271859296482412e-05,
-      "loss": 5.6146,
-      "step": 7747
-    },
-    {
-      "epoch": 4.040677966101695,
-      "grad_norm": 1.4075472354888916,
-      "learning_rate": 9.271758793969849e-05,
-      "loss": 5.6879,
-      "step": 7748
-    },
-    {
-      "epoch": 4.041199478487614,
-      "grad_norm": 1.5470061302185059,
-      "learning_rate": 9.271658291457286e-05,
-      "loss": 5.637,
-      "step": 7749
-    },
-    {
-      "epoch": 4.041720990873533,
-      "grad_norm": 1.4664270877838135,
-      "learning_rate": 9.271557788944724e-05,
-      "loss": 5.5054,
-      "step": 7750
-    },
-    {
-      "epoch": 4.042242503259453,
-      "grad_norm": 1.5448638200759888,
-      "learning_rate": 9.271457286432162e-05,
-      "loss": 6.0242,
-      "step": 7751
-    },
-    {
-      "epoch": 4.042764015645371,
-      "grad_norm": 1.4784748554229736,
-      "learning_rate": 9.271356783919598e-05,
-      "loss": 6.1007,
-      "step": 7752
-    },
-    {
-      "epoch": 4.043285528031291,
-      "grad_norm": 1.4604638814926147,
-      "learning_rate": 9.271256281407036e-05,
-      "loss": 5.5787,
-      "step": 7753
-    },
-    {
-      "epoch": 4.04380704041721,
-      "grad_norm": 1.4200048446655273,
-      "learning_rate": 9.271155778894473e-05,
-      "loss": 5.7969,
-      "step": 7754
-    },
-    {
-      "epoch": 4.044328552803129,
-      "grad_norm": 1.2903281450271606,
-      "learning_rate": 9.27105527638191e-05,
-      "loss": 6.1428,
-      "step": 7755
-    },
-    {
-      "epoch": 4.044850065189048,
-      "grad_norm": 2.5602188110351562,
-      "learning_rate": 9.270954773869347e-05,
-      "loss": 5.1813,
-      "step": 7756
-    },
-    {
-      "epoch": 4.045371577574968,
-      "grad_norm": 1.551783800125122,
-      "learning_rate": 9.270854271356785e-05,
-      "loss": 5.324,
-      "step": 7757
-    },
-    {
-      "epoch": 4.045893089960886,
-      "grad_norm": 1.48075532913208,
-      "learning_rate": 9.270753768844221e-05,
-      "loss": 5.7773,
-      "step": 7758
-    },
-    {
-      "epoch": 4.046414602346806,
-      "grad_norm": 1.6551659107208252,
-      "learning_rate": 9.270653266331659e-05,
-      "loss": 5.6422,
-      "step": 7759
-    },
-    {
-      "epoch": 4.046936114732725,
-      "grad_norm": 1.4410001039505005,
-      "learning_rate": 9.270552763819095e-05,
-      "loss": 5.8324,
-      "step": 7760
-    },
-    {
-      "epoch": 4.047457627118644,
-      "grad_norm": 1.372829794883728,
-      "learning_rate": 9.270452261306533e-05,
-      "loss": 5.9677,
-      "step": 7761
-    },
-    {
-      "epoch": 4.047979139504563,
-      "grad_norm": 1.5349671840667725,
-      "learning_rate": 9.270351758793971e-05,
-      "loss": 5.3899,
-      "step": 7762
-    },
-    {
-      "epoch": 4.048500651890483,
-      "grad_norm": 1.4967706203460693,
-      "learning_rate": 9.270251256281407e-05,
-      "loss": 5.989,
-      "step": 7763
-    },
-    {
-      "epoch": 4.049022164276401,
-      "grad_norm": 1.472976565361023,
-      "learning_rate": 9.270150753768845e-05,
-      "loss": 5.8665,
-      "step": 7764
-    },
-    {
-      "epoch": 4.049543676662321,
-      "grad_norm": 1.645326852798462,
-      "learning_rate": 9.270050251256281e-05,
-      "loss": 5.4161,
-      "step": 7765
-    },
-    {
-      "epoch": 4.05006518904824,
-      "grad_norm": 1.3409143686294556,
-      "learning_rate": 9.269949748743719e-05,
-      "loss": 5.9341,
-      "step": 7766
-    },
-    {
-      "epoch": 4.050586701434159,
-      "grad_norm": 1.5240248441696167,
-      "learning_rate": 9.269849246231156e-05,
-      "loss": 5.8171,
-      "step": 7767
-    },
-    {
-      "epoch": 4.051108213820078,
-      "grad_norm": 1.3809316158294678,
-      "learning_rate": 9.269748743718593e-05,
-      "loss": 5.6442,
-      "step": 7768
-    },
-    {
-      "epoch": 4.051629726205998,
-      "grad_norm": 1.505065679550171,
-      "learning_rate": 9.26964824120603e-05,
-      "loss": 5.4101,
-      "step": 7769
-    },
-    {
-      "epoch": 4.052151238591916,
-      "grad_norm": 1.3311412334442139,
-      "learning_rate": 9.269547738693468e-05,
-      "loss": 5.9654,
-      "step": 7770
-    },
-    {
-      "epoch": 4.052672750977836,
-      "grad_norm": 1.7631900310516357,
-      "learning_rate": 9.269447236180905e-05,
-      "loss": 4.6755,
-      "step": 7771
-    },
-    {
-      "epoch": 4.053194263363755,
-      "grad_norm": 1.4635214805603027,
-      "learning_rate": 9.269346733668343e-05,
-      "loss": 5.8484,
-      "step": 7772
-    },
-    {
-      "epoch": 4.053715775749674,
-      "grad_norm": 1.4534518718719482,
-      "learning_rate": 9.26924623115578e-05,
-      "loss": 5.8568,
-      "step": 7773
-    },
-    {
-      "epoch": 4.054237288135593,
-      "grad_norm": 1.387290596961975,
-      "learning_rate": 9.269145728643217e-05,
-      "loss": 5.5539,
-      "step": 7774
-    },
-    {
-      "epoch": 4.054758800521513,
-      "grad_norm": 1.5100772380828857,
-      "learning_rate": 9.269045226130654e-05,
-      "loss": 5.4018,
-      "step": 7775
-    },
-    {
-      "epoch": 4.055280312907431,
-      "grad_norm": 1.3560349941253662,
-      "learning_rate": 9.26894472361809e-05,
-      "loss": 5.7,
-      "step": 7776
-    },
-    {
-      "epoch": 4.055801825293351,
-      "grad_norm": 1.522064208984375,
-      "learning_rate": 9.268844221105528e-05,
-      "loss": 5.534,
-      "step": 7777
-    },
-    {
-      "epoch": 4.05632333767927,
-      "grad_norm": 1.3884671926498413,
-      "learning_rate": 9.268743718592965e-05,
-      "loss": 5.9232,
-      "step": 7778
-    },
-    {
-      "epoch": 4.056844850065189,
-      "grad_norm": 1.495206356048584,
-      "learning_rate": 9.268643216080402e-05,
-      "loss": 5.8095,
-      "step": 7779
-    },
-    {
-      "epoch": 4.057366362451108,
-      "grad_norm": 1.2516480684280396,
-      "learning_rate": 9.268542713567839e-05,
-      "loss": 6.1777,
-      "step": 7780
-    },
-    {
-      "epoch": 4.057887874837028,
-      "grad_norm": 1.534379482269287,
-      "learning_rate": 9.268442211055276e-05,
-      "loss": 5.7913,
-      "step": 7781
-    },
-    {
-      "epoch": 4.058409387222946,
-      "grad_norm": 1.5259021520614624,
-      "learning_rate": 9.268341708542714e-05,
-      "loss": 6.1763,
-      "step": 7782
-    },
-    {
-      "epoch": 4.058930899608866,
-      "grad_norm": 1.428620457649231,
-      "learning_rate": 9.268241206030152e-05,
-      "loss": 5.8834,
-      "step": 7783
-    },
-    {
-      "epoch": 4.059452411994785,
-      "grad_norm": 1.426031231880188,
-      "learning_rate": 9.268140703517588e-05,
-      "loss": 5.4946,
-      "step": 7784
-    },
-    {
-      "epoch": 4.059973924380704,
-      "grad_norm": 1.348958134651184,
-      "learning_rate": 9.268040201005026e-05,
-      "loss": 5.8724,
-      "step": 7785
-    },
-    {
-      "epoch": 4.060495436766623,
-      "grad_norm": 1.5968679189682007,
-      "learning_rate": 9.267939698492463e-05,
-      "loss": 5.1261,
-      "step": 7786
-    },
-    {
-      "epoch": 4.061016949152543,
-      "grad_norm": 1.514798641204834,
-      "learning_rate": 9.2678391959799e-05,
-      "loss": 5.1686,
-      "step": 7787
-    },
-    {
-      "epoch": 4.061538461538461,
-      "grad_norm": 1.512370228767395,
-      "learning_rate": 9.267738693467337e-05,
-      "loss": 5.6299,
-      "step": 7788
-    },
-    {
-      "epoch": 4.062059973924381,
-      "grad_norm": 1.5720722675323486,
-      "learning_rate": 9.267638190954773e-05,
-      "loss": 5.8612,
-      "step": 7789
-    },
-    {
-      "epoch": 4.0625814863103,
-      "grad_norm": 1.3575310707092285,
-      "learning_rate": 9.267537688442211e-05,
-      "loss": 5.5367,
-      "step": 7790
-    },
-    {
-      "epoch": 4.063102998696219,
-      "grad_norm": 1.6446588039398193,
-      "learning_rate": 9.267437185929649e-05,
-      "loss": 5.6676,
-      "step": 7791
-    },
-    {
-      "epoch": 4.063624511082138,
-      "grad_norm": 1.4908177852630615,
-      "learning_rate": 9.267336683417087e-05,
-      "loss": 5.5288,
-      "step": 7792
-    },
-    {
-      "epoch": 4.064146023468058,
-      "grad_norm": 1.4198685884475708,
-      "learning_rate": 9.267236180904523e-05,
-      "loss": 6.0277,
-      "step": 7793
-    },
-    {
-      "epoch": 4.064667535853976,
-      "grad_norm": 1.4246426820755005,
-      "learning_rate": 9.267135678391961e-05,
-      "loss": 6.0894,
-      "step": 7794
-    },
-    {
-      "epoch": 4.065189048239896,
-      "grad_norm": 1.3967424631118774,
-      "learning_rate": 9.267035175879397e-05,
-      "loss": 5.9353,
-      "step": 7795
-    },
-    {
-      "epoch": 4.065710560625815,
-      "grad_norm": 1.4990264177322388,
-      "learning_rate": 9.266934673366835e-05,
-      "loss": 5.5904,
-      "step": 7796
-    },
-    {
-      "epoch": 4.066232073011734,
-      "grad_norm": 1.6803250312805176,
-      "learning_rate": 9.266834170854272e-05,
-      "loss": 5.3202,
-      "step": 7797
-    },
-    {
-      "epoch": 4.066753585397653,
-      "grad_norm": 1.5388721227645874,
-      "learning_rate": 9.266733668341709e-05,
-      "loss": 5.2679,
-      "step": 7798
-    },
-    {
-      "epoch": 4.067275097783573,
-      "grad_norm": 1.4957226514816284,
-      "learning_rate": 9.266633165829146e-05,
-      "loss": 5.7429,
-      "step": 7799
-    },
-    {
-      "epoch": 4.067796610169491,
-      "grad_norm": 1.5074998140335083,
-      "learning_rate": 9.266532663316584e-05,
-      "loss": 5.2797,
-      "step": 7800
-    },
-    {
-      "epoch": 4.068318122555411,
-      "grad_norm": 1.4297837018966675,
-      "learning_rate": 9.26643216080402e-05,
-      "loss": 6.0692,
-      "step": 7801
-    },
-    {
-      "epoch": 4.06883963494133,
-      "grad_norm": 1.4485712051391602,
-      "learning_rate": 9.266331658291458e-05,
-      "loss": 5.9786,
-      "step": 7802
-    },
-    {
-      "epoch": 4.069361147327249,
-      "grad_norm": 1.522100567817688,
-      "learning_rate": 9.266231155778896e-05,
-      "loss": 5.9051,
-      "step": 7803
-    },
-    {
-      "epoch": 4.069882659713168,
-      "grad_norm": 1.452163815498352,
-      "learning_rate": 9.266130653266332e-05,
-      "loss": 5.6852,
-      "step": 7804
-    },
-    {
-      "epoch": 4.070404172099088,
-      "grad_norm": 1.3876086473464966,
-      "learning_rate": 9.26603015075377e-05,
-      "loss": 5.5676,
-      "step": 7805
-    },
-    {
-      "epoch": 4.070925684485006,
-      "grad_norm": 1.4466336965560913,
-      "learning_rate": 9.265929648241206e-05,
-      "loss": 5.7301,
-      "step": 7806
-    },
-    {
-      "epoch": 4.071447196870926,
-      "grad_norm": 1.5619091987609863,
-      "learning_rate": 9.265829145728644e-05,
-      "loss": 5.4527,
-      "step": 7807
-    },
-    {
-      "epoch": 4.071968709256845,
-      "grad_norm": 1.4330257177352905,
-      "learning_rate": 9.26572864321608e-05,
-      "loss": 5.9768,
-      "step": 7808
-    },
-    {
-      "epoch": 4.072490221642764,
-      "grad_norm": 1.467200517654419,
-      "learning_rate": 9.265628140703518e-05,
-      "loss": 5.9427,
-      "step": 7809
-    },
-    {
-      "epoch": 4.073011734028683,
-      "grad_norm": 1.4852330684661865,
-      "learning_rate": 9.265527638190955e-05,
-      "loss": 5.6571,
-      "step": 7810
-    },
-    {
-      "epoch": 4.073533246414602,
-      "grad_norm": 1.4207968711853027,
-      "learning_rate": 9.265427135678392e-05,
-      "loss": 5.8103,
-      "step": 7811
-    },
-    {
-      "epoch": 4.074054758800521,
-      "grad_norm": 1.3855139017105103,
-      "learning_rate": 9.26532663316583e-05,
-      "loss": 5.9616,
-      "step": 7812
-    },
-    {
-      "epoch": 4.074576271186441,
-      "grad_norm": 1.5995575189590454,
-      "learning_rate": 9.265226130653268e-05,
-      "loss": 5.2768,
-      "step": 7813
-    },
-    {
-      "epoch": 4.075097783572359,
-      "grad_norm": 1.411941647529602,
-      "learning_rate": 9.265125628140704e-05,
-      "loss": 5.7762,
-      "step": 7814
-    },
-    {
-      "epoch": 4.075619295958279,
-      "grad_norm": 1.6782230138778687,
-      "learning_rate": 9.265025125628141e-05,
-      "loss": 5.3224,
-      "step": 7815
-    },
-    {
-      "epoch": 4.076140808344198,
-      "grad_norm": 1.4913586378097534,
-      "learning_rate": 9.264924623115579e-05,
-      "loss": 5.5944,
-      "step": 7816
-    },
-    {
-      "epoch": 4.076662320730117,
-      "grad_norm": 1.3571511507034302,
-      "learning_rate": 9.264824120603015e-05,
-      "loss": 5.5187,
-      "step": 7817
-    },
-    {
-      "epoch": 4.077183833116036,
-      "grad_norm": 1.529752492904663,
-      "learning_rate": 9.264723618090453e-05,
-      "loss": 5.3579,
-      "step": 7818
-    },
-    {
-      "epoch": 4.077705345501956,
-      "grad_norm": 1.3558462858200073,
-      "learning_rate": 9.264623115577889e-05,
-      "loss": 5.8429,
-      "step": 7819
-    },
-    {
-      "epoch": 4.078226857887874,
-      "grad_norm": 1.4224931001663208,
-      "learning_rate": 9.264522613065327e-05,
-      "loss": 5.6495,
-      "step": 7820
-    },
-    {
-      "epoch": 4.078748370273794,
-      "grad_norm": 1.3902943134307861,
-      "learning_rate": 9.264422110552763e-05,
-      "loss": 5.5789,
-      "step": 7821
-    },
-    {
-      "epoch": 4.079269882659713,
-      "grad_norm": 1.378652572631836,
-      "learning_rate": 9.264321608040201e-05,
-      "loss": 5.879,
-      "step": 7822
-    },
-    {
-      "epoch": 4.079791395045632,
-      "grad_norm": 1.3300321102142334,
-      "learning_rate": 9.264221105527639e-05,
-      "loss": 6.2782,
-      "step": 7823
-    },
-    {
-      "epoch": 4.080312907431551,
-      "grad_norm": 1.613664984703064,
-      "learning_rate": 9.264120603015077e-05,
-      "loss": 5.3662,
-      "step": 7824
-    },
-    {
-      "epoch": 4.080834419817471,
-      "grad_norm": 1.6215964555740356,
-      "learning_rate": 9.264020100502513e-05,
-      "loss": 5.6516,
-      "step": 7825
-    },
-    {
-      "epoch": 4.081355932203389,
-      "grad_norm": 1.4671005010604858,
-      "learning_rate": 9.263919597989951e-05,
-      "loss": 5.9505,
-      "step": 7826
-    },
-    {
-      "epoch": 4.081877444589309,
-      "grad_norm": 1.51253342628479,
-      "learning_rate": 9.263819095477387e-05,
-      "loss": 6.139,
-      "step": 7827
-    },
-    {
-      "epoch": 4.082398956975228,
-      "grad_norm": 1.5839118957519531,
-      "learning_rate": 9.263718592964824e-05,
-      "loss": 5.4195,
-      "step": 7828
-    },
-    {
-      "epoch": 4.082920469361147,
-      "grad_norm": 1.4242699146270752,
-      "learning_rate": 9.263618090452262e-05,
-      "loss": 5.6728,
-      "step": 7829
-    },
-    {
-      "epoch": 4.083441981747066,
-      "grad_norm": 1.3379266262054443,
-      "learning_rate": 9.263517587939698e-05,
-      "loss": 5.9512,
-      "step": 7830
-    },
-    {
-      "epoch": 4.083963494132986,
-      "grad_norm": 1.6825493574142456,
-      "learning_rate": 9.263417085427136e-05,
-      "loss": 5.5446,
-      "step": 7831
-    },
-    {
-      "epoch": 4.0844850065189044,
-      "grad_norm": 1.4088844060897827,
-      "learning_rate": 9.263316582914574e-05,
-      "loss": 6.1129,
-      "step": 7832
-    },
-    {
-      "epoch": 4.085006518904824,
-      "grad_norm": 1.4431238174438477,
-      "learning_rate": 9.263216080402011e-05,
-      "loss": 5.9501,
-      "step": 7833
-    },
-    {
-      "epoch": 4.085528031290743,
-      "grad_norm": 1.501042366027832,
-      "learning_rate": 9.263115577889448e-05,
-      "loss": 5.7203,
-      "step": 7834
-    },
-    {
-      "epoch": 4.086049543676662,
-      "grad_norm": 1.5105688571929932,
-      "learning_rate": 9.263015075376886e-05,
-      "loss": 5.2921,
-      "step": 7835
-    },
-    {
-      "epoch": 4.086571056062581,
-      "grad_norm": 1.484744906425476,
-      "learning_rate": 9.262914572864322e-05,
-      "loss": 5.3079,
-      "step": 7836
-    },
-    {
-      "epoch": 4.087092568448501,
-      "grad_norm": 1.466486930847168,
-      "learning_rate": 9.26281407035176e-05,
-      "loss": 5.6513,
-      "step": 7837
-    },
-    {
-      "epoch": 4.0876140808344195,
-      "grad_norm": 1.4407843351364136,
-      "learning_rate": 9.262713567839196e-05,
-      "loss": 5.8353,
-      "step": 7838
-    },
-    {
-      "epoch": 4.088135593220339,
-      "grad_norm": 1.3655436038970947,
-      "learning_rate": 9.262613065326634e-05,
-      "loss": 5.5426,
-      "step": 7839
-    },
-    {
-      "epoch": 4.088657105606258,
-      "grad_norm": 1.4706255197525024,
-      "learning_rate": 9.26251256281407e-05,
-      "loss": 5.5358,
-      "step": 7840
-    },
-    {
-      "epoch": 4.089178617992177,
-      "grad_norm": 1.5802680253982544,
-      "learning_rate": 9.262412060301507e-05,
-      "loss": 5.6086,
-      "step": 7841
-    },
-    {
-      "epoch": 4.089700130378096,
-      "grad_norm": 1.4216445684432983,
-      "learning_rate": 9.262311557788945e-05,
-      "loss": 5.7229,
-      "step": 7842
-    },
-    {
-      "epoch": 4.090221642764016,
-      "grad_norm": 1.4805076122283936,
-      "learning_rate": 9.262211055276382e-05,
-      "loss": 5.7629,
-      "step": 7843
-    },
-    {
-      "epoch": 4.0907431551499345,
-      "grad_norm": 1.5099854469299316,
-      "learning_rate": 9.26211055276382e-05,
-      "loss": 5.8572,
-      "step": 7844
-    },
-    {
-      "epoch": 4.091264667535854,
-      "grad_norm": 1.5518715381622314,
-      "learning_rate": 9.262010050251257e-05,
-      "loss": 5.8294,
-      "step": 7845
-    },
-    {
-      "epoch": 4.091786179921773,
-      "grad_norm": 1.448813557624817,
-      "learning_rate": 9.261909547738694e-05,
-      "loss": 5.8629,
-      "step": 7846
-    },
-    {
-      "epoch": 4.092307692307692,
-      "grad_norm": 1.4683127403259277,
-      "learning_rate": 9.261809045226131e-05,
-      "loss": 5.9436,
-      "step": 7847
-    },
-    {
-      "epoch": 4.0928292046936114,
-      "grad_norm": 1.5080376863479614,
-      "learning_rate": 9.261708542713569e-05,
-      "loss": 5.6822,
-      "step": 7848
-    },
-    {
-      "epoch": 4.093350717079531,
-      "grad_norm": 1.5682848691940308,
-      "learning_rate": 9.261608040201005e-05,
-      "loss": 5.3844,
-      "step": 7849
-    },
-    {
-      "epoch": 4.0938722294654495,
-      "grad_norm": 1.3002808094024658,
-      "learning_rate": 9.261507537688443e-05,
-      "loss": 5.9643,
-      "step": 7850
-    },
-    {
-      "epoch": 4.094393741851369,
-      "grad_norm": 1.480974793434143,
-      "learning_rate": 9.261407035175879e-05,
-      "loss": 5.379,
-      "step": 7851
-    },
-    {
-      "epoch": 4.094915254237288,
-      "grad_norm": 1.5606924295425415,
-      "learning_rate": 9.261306532663317e-05,
-      "loss": 6.0206,
-      "step": 7852
-    },
-    {
-      "epoch": 4.095436766623207,
-      "grad_norm": 1.7071213722229004,
-      "learning_rate": 9.261206030150753e-05,
-      "loss": 6.0218,
-      "step": 7853
-    },
-    {
-      "epoch": 4.0959582790091265,
-      "grad_norm": 1.4019232988357544,
-      "learning_rate": 9.261105527638191e-05,
-      "loss": 5.3954,
-      "step": 7854
-    },
-    {
-      "epoch": 4.096479791395046,
-      "grad_norm": 1.5293500423431396,
-      "learning_rate": 9.261005025125629e-05,
-      "loss": 5.7789,
-      "step": 7855
-    },
-    {
-      "epoch": 4.0970013037809645,
-      "grad_norm": 1.4657803773880005,
-      "learning_rate": 9.260904522613065e-05,
-      "loss": 5.5836,
-      "step": 7856
-    },
-    {
-      "epoch": 4.097522816166884,
-      "grad_norm": 1.630781888961792,
-      "learning_rate": 9.260804020100503e-05,
-      "loss": 5.8042,
-      "step": 7857
-    },
-    {
-      "epoch": 4.098044328552803,
-      "grad_norm": 1.4896132946014404,
-      "learning_rate": 9.26070351758794e-05,
-      "loss": 4.7895,
-      "step": 7858
-    },
-    {
-      "epoch": 4.098565840938722,
-      "grad_norm": 1.4963663816452026,
-      "learning_rate": 9.260603015075377e-05,
-      "loss": 5.6306,
-      "step": 7859
-    },
-    {
-      "epoch": 4.0990873533246415,
-      "grad_norm": 1.4638421535491943,
-      "learning_rate": 9.260502512562814e-05,
-      "loss": 5.8229,
-      "step": 7860
-    },
-    {
-      "epoch": 4.099608865710561,
-      "grad_norm": 1.4387476444244385,
-      "learning_rate": 9.260402010050252e-05,
-      "loss": 5.3654,
-      "step": 7861
-    },
-    {
-      "epoch": 4.1001303780964795,
-      "grad_norm": 1.4353965520858765,
-      "learning_rate": 9.260301507537688e-05,
-      "loss": 5.1393,
-      "step": 7862
-    },
-    {
-      "epoch": 4.100651890482399,
-      "grad_norm": 1.6442482471466064,
-      "learning_rate": 9.260201005025126e-05,
-      "loss": 4.5064,
-      "step": 7863
-    },
-    {
-      "epoch": 4.101173402868318,
-      "grad_norm": 1.5158424377441406,
-      "learning_rate": 9.260100502512564e-05,
-      "loss": 5.8922,
-      "step": 7864
-    },
-    {
-      "epoch": 4.101694915254237,
-      "grad_norm": 1.569647192955017,
-      "learning_rate": 9.260000000000001e-05,
-      "loss": 5.7434,
-      "step": 7865
-    },
-    {
-      "epoch": 4.1022164276401565,
-      "grad_norm": 1.4959419965744019,
-      "learning_rate": 9.259899497487438e-05,
-      "loss": 5.4061,
-      "step": 7866
-    },
-    {
-      "epoch": 4.102737940026076,
-      "grad_norm": 1.466566562652588,
-      "learning_rate": 9.259798994974876e-05,
-      "loss": 5.6213,
-      "step": 7867
-    },
-    {
-      "epoch": 4.1032594524119945,
-      "grad_norm": 1.5008877515792847,
-      "learning_rate": 9.259698492462312e-05,
-      "loss": 5.4114,
-      "step": 7868
-    },
-    {
-      "epoch": 4.103780964797914,
-      "grad_norm": 1.4613921642303467,
-      "learning_rate": 9.259597989949749e-05,
-      "loss": 5.6441,
-      "step": 7869
-    },
-    {
-      "epoch": 4.1043024771838335,
-      "grad_norm": 1.4618943929672241,
-      "learning_rate": 9.259497487437186e-05,
-      "loss": 5.6818,
-      "step": 7870
-    },
-    {
-      "epoch": 4.104823989569752,
-      "grad_norm": 1.4624210596084595,
-      "learning_rate": 9.259396984924623e-05,
-      "loss": 5.7114,
-      "step": 7871
-    },
-    {
-      "epoch": 4.1053455019556715,
-      "grad_norm": 1.4972069263458252,
-      "learning_rate": 9.25929648241206e-05,
-      "loss": 5.5877,
-      "step": 7872
-    },
-    {
-      "epoch": 4.105867014341591,
-      "grad_norm": 1.4567420482635498,
-      "learning_rate": 9.259195979899497e-05,
-      "loss": 6.0044,
-      "step": 7873
-    },
-    {
-      "epoch": 4.1063885267275095,
-      "grad_norm": 1.4059394598007202,
-      "learning_rate": 9.259095477386935e-05,
-      "loss": 5.9232,
-      "step": 7874
-    },
-    {
-      "epoch": 4.106910039113429,
-      "grad_norm": 1.5342375040054321,
-      "learning_rate": 9.258994974874373e-05,
-      "loss": 4.8226,
-      "step": 7875
-    },
-    {
-      "epoch": 4.1074315514993485,
-      "grad_norm": 1.4270868301391602,
-      "learning_rate": 9.25889447236181e-05,
-      "loss": 5.627,
-      "step": 7876
-    },
-    {
-      "epoch": 4.107953063885267,
-      "grad_norm": 1.4449658393859863,
-      "learning_rate": 9.258793969849247e-05,
-      "loss": 5.6322,
-      "step": 7877
-    },
-    {
-      "epoch": 4.1084745762711865,
-      "grad_norm": 1.4410638809204102,
-      "learning_rate": 9.258693467336685e-05,
-      "loss": 5.9928,
-      "step": 7878
-    },
-    {
-      "epoch": 4.108996088657106,
-      "grad_norm": 1.3757808208465576,
-      "learning_rate": 9.258592964824121e-05,
-      "loss": 5.8552,
-      "step": 7879
-    },
-    {
-      "epoch": 4.1095176010430245,
-      "grad_norm": 1.4711267948150635,
-      "learning_rate": 9.258492462311559e-05,
-      "loss": 5.2836,
-      "step": 7880
-    },
-    {
-      "epoch": 4.110039113428944,
-      "grad_norm": 1.507796049118042,
-      "learning_rate": 9.258391959798995e-05,
-      "loss": 5.7032,
-      "step": 7881
-    },
-    {
-      "epoch": 4.1105606258148635,
-      "grad_norm": 1.633172631263733,
-      "learning_rate": 9.258291457286432e-05,
-      "loss": 5.9098,
-      "step": 7882
-    },
-    {
-      "epoch": 4.111082138200782,
-      "grad_norm": 1.473927617073059,
-      "learning_rate": 9.25819095477387e-05,
-      "loss": 5.733,
-      "step": 7883
-    },
-    {
-      "epoch": 4.1116036505867015,
-      "grad_norm": 1.3980087041854858,
-      "learning_rate": 9.258090452261307e-05,
-      "loss": 5.7791,
-      "step": 7884
-    },
-    {
-      "epoch": 4.112125162972621,
-      "grad_norm": 1.584582805633545,
-      "learning_rate": 9.257989949748745e-05,
-      "loss": 5.4682,
-      "step": 7885
-    },
-    {
-      "epoch": 4.11264667535854,
-      "grad_norm": 1.5834003686904907,
-      "learning_rate": 9.257889447236181e-05,
-      "loss": 5.6921,
-      "step": 7886
-    },
-    {
-      "epoch": 4.113168187744459,
-      "grad_norm": 1.579816460609436,
-      "learning_rate": 9.257788944723619e-05,
-      "loss": 5.8401,
-      "step": 7887
-    },
-    {
-      "epoch": 4.1136897001303785,
-      "grad_norm": 1.3022929430007935,
-      "learning_rate": 9.257688442211056e-05,
-      "loss": 5.2846,
-      "step": 7888
-    },
-    {
-      "epoch": 4.114211212516297,
-      "grad_norm": 1.5924038887023926,
-      "learning_rate": 9.257587939698493e-05,
-      "loss": 5.3418,
-      "step": 7889
-    },
-    {
-      "epoch": 4.1147327249022165,
-      "grad_norm": 1.6345165967941284,
-      "learning_rate": 9.25748743718593e-05,
-      "loss": 5.3985,
-      "step": 7890
-    },
-    {
-      "epoch": 4.115254237288136,
-      "grad_norm": 1.5700654983520508,
-      "learning_rate": 9.257386934673368e-05,
-      "loss": 5.7724,
-      "step": 7891
-    },
-    {
-      "epoch": 4.115775749674055,
-      "grad_norm": 1.3302252292633057,
-      "learning_rate": 9.257286432160804e-05,
-      "loss": 5.8437,
-      "step": 7892
-    },
-    {
-      "epoch": 4.116297262059974,
-      "grad_norm": 1.578761100769043,
-      "learning_rate": 9.257185929648242e-05,
-      "loss": 5.3953,
-      "step": 7893
-    },
-    {
-      "epoch": 4.1168187744458935,
-      "grad_norm": 1.4199178218841553,
-      "learning_rate": 9.257085427135678e-05,
-      "loss": 5.8331,
-      "step": 7894
-    },
-    {
-      "epoch": 4.117340286831812,
-      "grad_norm": 1.4709476232528687,
-      "learning_rate": 9.256984924623116e-05,
-      "loss": 5.947,
-      "step": 7895
-    },
-    {
-      "epoch": 4.1178617992177315,
-      "grad_norm": 1.4902665615081787,
-      "learning_rate": 9.256884422110554e-05,
-      "loss": 5.5397,
-      "step": 7896
-    },
-    {
-      "epoch": 4.118383311603651,
-      "grad_norm": 1.3731474876403809,
-      "learning_rate": 9.25678391959799e-05,
-      "loss": 5.9759,
-      "step": 7897
-    },
-    {
-      "epoch": 4.11890482398957,
-      "grad_norm": 1.4054491519927979,
-      "learning_rate": 9.256683417085428e-05,
-      "loss": 5.5073,
-      "step": 7898
-    },
-    {
-      "epoch": 4.119426336375489,
-      "grad_norm": 1.5316388607025146,
-      "learning_rate": 9.256582914572864e-05,
-      "loss": 5.6306,
-      "step": 7899
-    },
-    {
-      "epoch": 4.1199478487614085,
-      "grad_norm": 1.4608274698257446,
-      "learning_rate": 9.256482412060302e-05,
-      "loss": 5.7544,
-      "step": 7900
-    },
-    {
-      "epoch": 4.120469361147327,
-      "grad_norm": 1.5278129577636719,
-      "learning_rate": 9.256381909547739e-05,
-      "loss": 5.2985,
-      "step": 7901
-    },
-    {
-      "epoch": 4.120990873533247,
-      "grad_norm": 1.323892593383789,
-      "learning_rate": 9.256281407035176e-05,
-      "loss": 6.077,
-      "step": 7902
-    },
-    {
-      "epoch": 4.121512385919166,
-      "grad_norm": 1.436211109161377,
-      "learning_rate": 9.256180904522613e-05,
-      "loss": 5.9673,
-      "step": 7903
-    },
-    {
-      "epoch": 4.122033898305085,
-      "grad_norm": 1.5019803047180176,
-      "learning_rate": 9.25608040201005e-05,
-      "loss": 5.4464,
-      "step": 7904
-    },
-    {
-      "epoch": 4.122555410691004,
-      "grad_norm": 1.529310941696167,
-      "learning_rate": 9.255979899497488e-05,
-      "loss": 5.8382,
-      "step": 7905
-    },
-    {
-      "epoch": 4.123076923076923,
-      "grad_norm": 1.551797866821289,
-      "learning_rate": 9.255879396984926e-05,
-      "loss": 5.8559,
-      "step": 7906
-    },
-    {
-      "epoch": 4.123598435462842,
-      "grad_norm": 1.4375746250152588,
-      "learning_rate": 9.255778894472363e-05,
-      "loss": 5.558,
-      "step": 7907
-    },
-    {
-      "epoch": 4.124119947848762,
-      "grad_norm": 1.5525685548782349,
-      "learning_rate": 9.255678391959799e-05,
-      "loss": 5.3522,
-      "step": 7908
-    },
-    {
-      "epoch": 4.12464146023468,
-      "grad_norm": 1.4690592288970947,
-      "learning_rate": 9.255577889447237e-05,
-      "loss": 5.5808,
-      "step": 7909
-    },
-    {
-      "epoch": 4.1251629726206,
-      "grad_norm": 1.3985209465026855,
-      "learning_rate": 9.255477386934673e-05,
-      "loss": 5.4361,
-      "step": 7910
-    },
-    {
-      "epoch": 4.125684485006519,
-      "grad_norm": 1.4502190351486206,
-      "learning_rate": 9.255376884422111e-05,
-      "loss": 6.0109,
-      "step": 7911
-    },
-    {
-      "epoch": 4.126205997392438,
-      "grad_norm": 1.4844197034835815,
-      "learning_rate": 9.255276381909547e-05,
-      "loss": 5.6406,
-      "step": 7912
-    },
-    {
-      "epoch": 4.126727509778357,
-      "grad_norm": 1.3644835948944092,
-      "learning_rate": 9.255175879396985e-05,
-      "loss": 6.0641,
-      "step": 7913
-    },
-    {
-      "epoch": 4.127249022164277,
-      "grad_norm": 1.5031625032424927,
-      "learning_rate": 9.255075376884422e-05,
-      "loss": 5.1412,
-      "step": 7914
-    },
-    {
-      "epoch": 4.127770534550195,
-      "grad_norm": 1.5173677206039429,
-      "learning_rate": 9.25497487437186e-05,
-      "loss": 5.8119,
-      "step": 7915
-    },
-    {
-      "epoch": 4.128292046936115,
-      "grad_norm": 1.4309849739074707,
-      "learning_rate": 9.254874371859297e-05,
-      "loss": 5.8947,
-      "step": 7916
-    },
-    {
-      "epoch": 4.128813559322034,
-      "grad_norm": 1.5204359292984009,
-      "learning_rate": 9.254773869346735e-05,
-      "loss": 5.3283,
-      "step": 7917
-    },
-    {
-      "epoch": 4.129335071707953,
-      "grad_norm": 1.494659185409546,
-      "learning_rate": 9.254673366834171e-05,
-      "loss": 5.0787,
-      "step": 7918
-    },
-    {
-      "epoch": 4.129856584093872,
-      "grad_norm": 1.6365699768066406,
-      "learning_rate": 9.254572864321609e-05,
-      "loss": 5.115,
-      "step": 7919
-    },
-    {
-      "epoch": 4.130378096479792,
-      "grad_norm": 1.482642650604248,
-      "learning_rate": 9.254472361809046e-05,
-      "loss": 5.1921,
-      "step": 7920
-    },
-    {
-      "epoch": 4.13089960886571,
-      "grad_norm": 1.5191080570220947,
-      "learning_rate": 9.254371859296482e-05,
-      "loss": 5.9552,
-      "step": 7921
-    },
-    {
-      "epoch": 4.13142112125163,
-      "grad_norm": 1.465709924697876,
-      "learning_rate": 9.25427135678392e-05,
-      "loss": 5.9255,
-      "step": 7922
-    },
-    {
-      "epoch": 4.131942633637549,
-      "grad_norm": 1.558511734008789,
-      "learning_rate": 9.254170854271356e-05,
-      "loss": 5.9157,
-      "step": 7923
-    },
-    {
-      "epoch": 4.132464146023468,
-      "grad_norm": 1.5937254428863525,
-      "learning_rate": 9.254070351758794e-05,
-      "loss": 5.5093,
-      "step": 7924
-    },
-    {
-      "epoch": 4.132985658409387,
-      "grad_norm": 1.5068422555923462,
-      "learning_rate": 9.253969849246232e-05,
-      "loss": 5.5133,
-      "step": 7925
-    },
-    {
-      "epoch": 4.133507170795307,
-      "grad_norm": 1.5828957557678223,
-      "learning_rate": 9.25386934673367e-05,
-      "loss": 5.9056,
-      "step": 7926
-    },
-    {
-      "epoch": 4.134028683181225,
-      "grad_norm": 1.4433095455169678,
-      "learning_rate": 9.253768844221106e-05,
-      "loss": 5.2279,
-      "step": 7927
-    },
-    {
-      "epoch": 4.134550195567145,
-      "grad_norm": 1.802423119544983,
-      "learning_rate": 9.253668341708544e-05,
-      "loss": 5.5105,
-      "step": 7928
-    },
-    {
-      "epoch": 4.135071707953064,
-      "grad_norm": 1.664573073387146,
-      "learning_rate": 9.25356783919598e-05,
-      "loss": 5.4618,
-      "step": 7929
-    },
-    {
-      "epoch": 4.135593220338983,
-      "grad_norm": 1.4259589910507202,
-      "learning_rate": 9.253467336683418e-05,
-      "loss": 5.81,
-      "step": 7930
-    },
-    {
-      "epoch": 4.136114732724902,
-      "grad_norm": 1.652590274810791,
-      "learning_rate": 9.253366834170854e-05,
-      "loss": 5.2233,
-      "step": 7931
-    },
-    {
-      "epoch": 4.136636245110822,
-      "grad_norm": 1.51335608959198,
-      "learning_rate": 9.253266331658292e-05,
-      "loss": 6.0791,
-      "step": 7932
-    },
-    {
-      "epoch": 4.13715775749674,
-      "grad_norm": 1.384605050086975,
-      "learning_rate": 9.253165829145729e-05,
-      "loss": 6.1452,
-      "step": 7933
-    },
-    {
-      "epoch": 4.13767926988266,
-      "grad_norm": 1.392144799232483,
-      "learning_rate": 9.253065326633165e-05,
-      "loss": 5.93,
-      "step": 7934
-    },
-    {
-      "epoch": 4.138200782268579,
-      "grad_norm": 1.40146005153656,
-      "learning_rate": 9.252964824120603e-05,
-      "loss": 5.8465,
-      "step": 7935
-    },
-    {
-      "epoch": 4.138722294654498,
-      "grad_norm": 1.5170398950576782,
-      "learning_rate": 9.25286432160804e-05,
-      "loss": 5.6845,
-      "step": 7936
-    },
-    {
-      "epoch": 4.139243807040417,
-      "grad_norm": 1.6018669605255127,
-      "learning_rate": 9.252763819095478e-05,
-      "loss": 5.7977,
-      "step": 7937
-    },
-    {
-      "epoch": 4.139765319426337,
-      "grad_norm": 1.4677003622055054,
-      "learning_rate": 9.252663316582915e-05,
-      "loss": 5.8254,
-      "step": 7938
-    },
-    {
-      "epoch": 4.140286831812255,
-      "grad_norm": 1.37051260471344,
-      "learning_rate": 9.252562814070353e-05,
-      "loss": 5.9204,
-      "step": 7939
-    },
-    {
-      "epoch": 4.140808344198175,
-      "grad_norm": 1.4085888862609863,
-      "learning_rate": 9.252462311557789e-05,
-      "loss": 5.9496,
-      "step": 7940
-    },
-    {
-      "epoch": 4.141329856584094,
-      "grad_norm": 1.4396485090255737,
-      "learning_rate": 9.252361809045227e-05,
-      "loss": 5.2377,
-      "step": 7941
-    },
-    {
-      "epoch": 4.141851368970013,
-      "grad_norm": 1.4618613719940186,
-      "learning_rate": 9.252261306532663e-05,
-      "loss": 5.6297,
-      "step": 7942
-    },
-    {
-      "epoch": 4.142372881355932,
-      "grad_norm": 1.4706356525421143,
-      "learning_rate": 9.252160804020101e-05,
-      "loss": 6.1027,
-      "step": 7943
-    },
-    {
-      "epoch": 4.142894393741852,
-      "grad_norm": 1.4448494911193848,
-      "learning_rate": 9.252060301507538e-05,
-      "loss": 5.1612,
-      "step": 7944
-    },
-    {
-      "epoch": 4.14341590612777,
-      "grad_norm": 1.3651525974273682,
-      "learning_rate": 9.251959798994975e-05,
-      "loss": 5.8663,
-      "step": 7945
-    },
-    {
-      "epoch": 4.14393741851369,
-      "grad_norm": 1.5383317470550537,
-      "learning_rate": 9.251859296482413e-05,
-      "loss": 5.5437,
-      "step": 7946
-    },
-    {
-      "epoch": 4.144458930899609,
-      "grad_norm": 1.5273597240447998,
-      "learning_rate": 9.251758793969851e-05,
-      "loss": 5.4389,
-      "step": 7947
-    },
-    {
-      "epoch": 4.144980443285528,
-      "grad_norm": 1.3910081386566162,
-      "learning_rate": 9.251658291457287e-05,
-      "loss": 5.9814,
-      "step": 7948
-    },
-    {
-      "epoch": 4.145501955671447,
-      "grad_norm": 1.5585614442825317,
-      "learning_rate": 9.251557788944724e-05,
-      "loss": 5.5628,
-      "step": 7949
-    },
-    {
-      "epoch": 4.146023468057367,
-      "grad_norm": 1.479105830192566,
-      "learning_rate": 9.251457286432162e-05,
-      "loss": 5.8745,
-      "step": 7950
-    },
-    {
-      "epoch": 4.146544980443285,
-      "grad_norm": 1.5561717748641968,
-      "learning_rate": 9.251356783919598e-05,
-      "loss": 5.1674,
-      "step": 7951
-    },
-    {
-      "epoch": 4.147066492829205,
-      "grad_norm": 1.5374099016189575,
-      "learning_rate": 9.251256281407036e-05,
-      "loss": 5.4857,
-      "step": 7952
-    },
-    {
-      "epoch": 4.147588005215124,
-      "grad_norm": 1.5257409811019897,
-      "learning_rate": 9.251155778894472e-05,
-      "loss": 5.5494,
-      "step": 7953
-    },
-    {
-      "epoch": 4.148109517601043,
-      "grad_norm": 1.6378060579299927,
-      "learning_rate": 9.25105527638191e-05,
-      "loss": 5.6747,
-      "step": 7954
-    },
-    {
-      "epoch": 4.148631029986962,
-      "grad_norm": 1.3717561960220337,
-      "learning_rate": 9.250954773869346e-05,
-      "loss": 5.891,
-      "step": 7955
-    },
-    {
-      "epoch": 4.149152542372882,
-      "grad_norm": 1.407088279724121,
-      "learning_rate": 9.250854271356784e-05,
-      "loss": 5.881,
-      "step": 7956
-    },
-    {
-      "epoch": 4.1496740547588,
-      "grad_norm": 1.562442421913147,
-      "learning_rate": 9.250753768844222e-05,
-      "loss": 5.8994,
-      "step": 7957
-    },
-    {
-      "epoch": 4.15019556714472,
-      "grad_norm": 1.3944790363311768,
-      "learning_rate": 9.25065326633166e-05,
-      "loss": 5.7611,
-      "step": 7958
-    },
-    {
-      "epoch": 4.150717079530639,
-      "grad_norm": 1.3896006345748901,
-      "learning_rate": 9.250552763819096e-05,
-      "loss": 6.0281,
-      "step": 7959
-    },
-    {
-      "epoch": 4.151238591916558,
-      "grad_norm": 1.4281668663024902,
-      "learning_rate": 9.250452261306534e-05,
-      "loss": 5.382,
-      "step": 7960
-    },
-    {
-      "epoch": 4.151760104302477,
-      "grad_norm": 1.5073614120483398,
-      "learning_rate": 9.25035175879397e-05,
-      "loss": 5.5142,
-      "step": 7961
-    },
-    {
-      "epoch": 4.152281616688397,
-      "grad_norm": 1.5959811210632324,
-      "learning_rate": 9.250251256281407e-05,
-      "loss": 4.9596,
-      "step": 7962
-    },
-    {
-      "epoch": 4.152803129074315,
-      "grad_norm": 1.3472485542297363,
-      "learning_rate": 9.250150753768845e-05,
-      "loss": 5.7849,
-      "step": 7963
-    },
-    {
-      "epoch": 4.153324641460235,
-      "grad_norm": 1.5522563457489014,
-      "learning_rate": 9.250050251256281e-05,
-      "loss": 5.6841,
-      "step": 7964
-    },
-    {
-      "epoch": 4.153846153846154,
-      "grad_norm": 1.5052525997161865,
-      "learning_rate": 9.249949748743719e-05,
-      "loss": 5.6975,
-      "step": 7965
-    },
-    {
-      "epoch": 4.154367666232073,
-      "grad_norm": 1.6255697011947632,
-      "learning_rate": 9.249849246231157e-05,
-      "loss": 5.8057,
-      "step": 7966
-    },
-    {
-      "epoch": 4.154889178617992,
-      "grad_norm": 1.364441156387329,
-      "learning_rate": 9.249748743718594e-05,
-      "loss": 5.9097,
-      "step": 7967
-    },
-    {
-      "epoch": 4.155410691003912,
-      "grad_norm": 1.4942996501922607,
-      "learning_rate": 9.249648241206031e-05,
-      "loss": 5.0428,
-      "step": 7968
-    },
-    {
-      "epoch": 4.15593220338983,
-      "grad_norm": 1.4311703443527222,
-      "learning_rate": 9.249547738693469e-05,
-      "loss": 5.575,
-      "step": 7969
-    },
-    {
-      "epoch": 4.15645371577575,
-      "grad_norm": 1.424970269203186,
-      "learning_rate": 9.249447236180905e-05,
-      "loss": 5.6306,
-      "step": 7970
-    },
-    {
-      "epoch": 4.156975228161669,
-      "grad_norm": 1.4862653017044067,
-      "learning_rate": 9.249346733668343e-05,
-      "loss": 5.15,
-      "step": 7971
-    },
-    {
-      "epoch": 4.157496740547588,
-      "grad_norm": 1.4854646921157837,
-      "learning_rate": 9.249246231155779e-05,
-      "loss": 5.7152,
-      "step": 7972
-    },
-    {
-      "epoch": 4.158018252933507,
-      "grad_norm": 1.49619722366333,
-      "learning_rate": 9.249145728643217e-05,
-      "loss": 5.9255,
-      "step": 7973
-    },
-    {
-      "epoch": 4.158539765319427,
-      "grad_norm": 1.3058419227600098,
-      "learning_rate": 9.249045226130653e-05,
-      "loss": 5.7663,
-      "step": 7974
-    },
-    {
-      "epoch": 4.159061277705345,
-      "grad_norm": 1.4758565425872803,
-      "learning_rate": 9.24894472361809e-05,
-      "loss": 5.8576,
-      "step": 7975
-    },
-    {
-      "epoch": 4.159582790091265,
-      "grad_norm": 1.5442945957183838,
-      "learning_rate": 9.248844221105528e-05,
-      "loss": 5.1784,
-      "step": 7976
-    },
-    {
-      "epoch": 4.160104302477184,
-      "grad_norm": 1.3577169179916382,
-      "learning_rate": 9.248743718592965e-05,
-      "loss": 5.7748,
-      "step": 7977
-    },
-    {
-      "epoch": 4.160625814863103,
-      "grad_norm": 1.4710315465927124,
-      "learning_rate": 9.248643216080403e-05,
-      "loss": 5.5217,
-      "step": 7978
-    },
-    {
-      "epoch": 4.161147327249022,
-      "grad_norm": 1.5290608406066895,
-      "learning_rate": 9.24854271356784e-05,
-      "loss": 5.7468,
-      "step": 7979
-    },
-    {
-      "epoch": 4.161668839634942,
-      "grad_norm": 1.4326730966567993,
-      "learning_rate": 9.248442211055277e-05,
-      "loss": 5.68,
-      "step": 7980
-    },
-    {
-      "epoch": 4.16219035202086,
-      "grad_norm": 1.4315330982208252,
-      "learning_rate": 9.248341708542714e-05,
-      "loss": 5.2461,
-      "step": 7981
-    },
-    {
-      "epoch": 4.16271186440678,
-      "grad_norm": 1.4383541345596313,
-      "learning_rate": 9.248241206030152e-05,
-      "loss": 5.766,
-      "step": 7982
-    },
-    {
-      "epoch": 4.163233376792699,
-      "grad_norm": 1.3988003730773926,
-      "learning_rate": 9.248140703517588e-05,
-      "loss": 5.7453,
-      "step": 7983
-    },
-    {
-      "epoch": 4.163754889178618,
-      "grad_norm": 1.5318245887756348,
-      "learning_rate": 9.248040201005026e-05,
-      "loss": 5.4455,
-      "step": 7984
-    },
-    {
-      "epoch": 4.164276401564537,
-      "grad_norm": 1.4302725791931152,
-      "learning_rate": 9.247939698492462e-05,
-      "loss": 5.8479,
-      "step": 7985
-    },
-    {
-      "epoch": 4.164797913950457,
-      "grad_norm": 1.4330464601516724,
-      "learning_rate": 9.2478391959799e-05,
-      "loss": 5.6143,
-      "step": 7986
-    },
-    {
-      "epoch": 4.165319426336375,
-      "grad_norm": 1.5396825075149536,
-      "learning_rate": 9.247738693467338e-05,
-      "loss": 5.7541,
-      "step": 7987
-    },
-    {
-      "epoch": 4.165840938722295,
-      "grad_norm": 1.403669834136963,
-      "learning_rate": 9.247638190954774e-05,
-      "loss": 5.8228,
-      "step": 7988
-    },
-    {
-      "epoch": 4.166362451108214,
-      "grad_norm": 1.4943293333053589,
-      "learning_rate": 9.247537688442212e-05,
-      "loss": 5.5104,
-      "step": 7989
-    },
-    {
-      "epoch": 4.166883963494133,
-      "grad_norm": 1.4229488372802734,
-      "learning_rate": 9.247437185929648e-05,
-      "loss": 5.8853,
-      "step": 7990
-    },
-    {
-      "epoch": 4.167405475880052,
-      "grad_norm": 1.4458130598068237,
-      "learning_rate": 9.247336683417086e-05,
-      "loss": 5.7799,
-      "step": 7991
-    },
-    {
-      "epoch": 4.167926988265972,
-      "grad_norm": 1.358916997909546,
-      "learning_rate": 9.247236180904523e-05,
-      "loss": 5.8871,
-      "step": 7992
-    },
-    {
-      "epoch": 4.16844850065189,
-      "grad_norm": 1.5322297811508179,
-      "learning_rate": 9.24713567839196e-05,
-      "loss": 5.6711,
-      "step": 7993
-    },
-    {
-      "epoch": 4.16897001303781,
-      "grad_norm": 1.5065218210220337,
-      "learning_rate": 9.247035175879397e-05,
-      "loss": 5.3637,
-      "step": 7994
-    },
-    {
-      "epoch": 4.169491525423728,
-      "grad_norm": 1.64816153049469,
-      "learning_rate": 9.246934673366835e-05,
-      "loss": 5.7693,
-      "step": 7995
-    },
-    {
-      "epoch": 4.170013037809648,
-      "grad_norm": 1.508472204208374,
-      "learning_rate": 9.246834170854271e-05,
-      "loss": 5.9295,
-      "step": 7996
-    },
-    {
-      "epoch": 4.170534550195567,
-      "grad_norm": 1.5284219980239868,
-      "learning_rate": 9.246733668341709e-05,
-      "loss": 5.7533,
-      "step": 7997
-    },
-    {
-      "epoch": 4.171056062581487,
-      "grad_norm": 1.4177913665771484,
-      "learning_rate": 9.246633165829147e-05,
-      "loss": 5.7313,
-      "step": 7998
-    },
-    {
-      "epoch": 4.171577574967405,
-      "grad_norm": 1.3781781196594238,
-      "learning_rate": 9.246532663316584e-05,
-      "loss": 5.5933,
-      "step": 7999
-    },
-    {
-      "epoch": 4.172099087353325,
-      "grad_norm": 1.5088404417037964,
-      "learning_rate": 9.246432160804021e-05,
-      "loss": 5.614,
-      "step": 8000
-    },
-    {
-      "epoch": 4.172620599739243,
-      "grad_norm": 1.5555975437164307,
-      "learning_rate": 9.246331658291457e-05,
-      "loss": 5.4365,
-      "step": 8001
-    },
-    {
-      "epoch": 4.173142112125163,
-      "grad_norm": 1.69544517993927,
-      "learning_rate": 9.246231155778895e-05,
-      "loss": 6.0654,
-      "step": 8002
-    },
-    {
-      "epoch": 4.173663624511082,
-      "grad_norm": 1.627017855644226,
-      "learning_rate": 9.246130653266331e-05,
-      "loss": 5.5741,
-      "step": 8003
-    },
-    {
-      "epoch": 4.174185136897001,
-      "grad_norm": 1.559555172920227,
-      "learning_rate": 9.246030150753769e-05,
-      "loss": 5.8388,
-      "step": 8004
-    },
-    {
-      "epoch": 4.17470664928292,
-      "grad_norm": 1.5767152309417725,
-      "learning_rate": 9.245929648241206e-05,
-      "loss": 5.0984,
-      "step": 8005
-    },
-    {
-      "epoch": 4.17522816166884,
-      "grad_norm": 1.6931376457214355,
-      "learning_rate": 9.245829145728643e-05,
-      "loss": 6.188,
-      "step": 8006
-    },
-    {
-      "epoch": 4.175749674054758,
-      "grad_norm": 1.7225605249404907,
-      "learning_rate": 9.245728643216081e-05,
-      "loss": 5.2533,
-      "step": 8007
-    },
-    {
-      "epoch": 4.176271186440678,
-      "grad_norm": 1.493012547492981,
-      "learning_rate": 9.245628140703519e-05,
-      "loss": 6.1509,
-      "step": 8008
-    },
-    {
-      "epoch": 4.176792698826597,
-      "grad_norm": 1.5385997295379639,
-      "learning_rate": 9.245527638190955e-05,
-      "loss": 5.5863,
-      "step": 8009
-    },
-    {
-      "epoch": 4.177314211212516,
-      "grad_norm": 1.5017783641815186,
-      "learning_rate": 9.245427135678393e-05,
-      "loss": 5.4661,
-      "step": 8010
-    },
-    {
-      "epoch": 4.177835723598435,
-      "grad_norm": 1.472206950187683,
-      "learning_rate": 9.24532663316583e-05,
-      "loss": 5.2551,
-      "step": 8011
-    },
-    {
-      "epoch": 4.178357235984355,
-      "grad_norm": 1.410810112953186,
-      "learning_rate": 9.245226130653267e-05,
-      "loss": 5.3972,
-      "step": 8012
-    },
-    {
-      "epoch": 4.178878748370273,
-      "grad_norm": 1.5154269933700562,
-      "learning_rate": 9.245125628140704e-05,
-      "loss": 5.6251,
-      "step": 8013
-    },
-    {
-      "epoch": 4.179400260756193,
-      "grad_norm": 1.6027485132217407,
-      "learning_rate": 9.24502512562814e-05,
-      "loss": 5.5104,
-      "step": 8014
-    },
-    {
-      "epoch": 4.179921773142112,
-      "grad_norm": 1.5796518325805664,
-      "learning_rate": 9.244924623115578e-05,
-      "loss": 5.6809,
-      "step": 8015
-    },
-    {
-      "epoch": 4.180443285528031,
-      "grad_norm": 1.4736182689666748,
-      "learning_rate": 9.244824120603015e-05,
-      "loss": 5.6487,
-      "step": 8016
-    },
-    {
-      "epoch": 4.18096479791395,
-      "grad_norm": 1.4706374406814575,
-      "learning_rate": 9.244723618090452e-05,
-      "loss": 5.7969,
-      "step": 8017
-    },
-    {
-      "epoch": 4.18148631029987,
-      "grad_norm": 1.2628108263015747,
-      "learning_rate": 9.24462311557789e-05,
-      "loss": 6.1244,
-      "step": 8018
-    },
-    {
-      "epoch": 4.182007822685788,
-      "grad_norm": 1.364096999168396,
-      "learning_rate": 9.244522613065328e-05,
-      "loss": 5.9715,
-      "step": 8019
-    },
-    {
-      "epoch": 4.182529335071708,
-      "grad_norm": 1.9296715259552002,
-      "learning_rate": 9.244422110552764e-05,
-      "loss": 5.1742,
-      "step": 8020
-    },
-    {
-      "epoch": 4.183050847457627,
-      "grad_norm": 1.4944796562194824,
-      "learning_rate": 9.244321608040202e-05,
-      "loss": 5.7869,
-      "step": 8021
-    },
-    {
-      "epoch": 4.183572359843546,
-      "grad_norm": 1.5324578285217285,
-      "learning_rate": 9.244221105527638e-05,
-      "loss": 5.2627,
-      "step": 8022
-    },
-    {
-      "epoch": 4.184093872229465,
-      "grad_norm": 1.5570334196090698,
-      "learning_rate": 9.244120603015076e-05,
-      "loss": 5.2615,
-      "step": 8023
-    },
-    {
-      "epoch": 4.184615384615385,
-      "grad_norm": 1.636074185371399,
-      "learning_rate": 9.244020100502513e-05,
-      "loss": 6.0295,
-      "step": 8024
-    },
-    {
-      "epoch": 4.185136897001303,
-      "grad_norm": 1.5826523303985596,
-      "learning_rate": 9.24391959798995e-05,
-      "loss": 5.3125,
-      "step": 8025
-    },
-    {
-      "epoch": 4.185658409387223,
-      "grad_norm": 1.5885404348373413,
-      "learning_rate": 9.243819095477387e-05,
-      "loss": 5.1305,
-      "step": 8026
-    },
-    {
-      "epoch": 4.186179921773142,
-      "grad_norm": 1.5010654926300049,
-      "learning_rate": 9.243718592964823e-05,
-      "loss": 5.6081,
-      "step": 8027
-    },
-    {
-      "epoch": 4.186701434159061,
-      "grad_norm": 1.4546791315078735,
-      "learning_rate": 9.243618090452261e-05,
-      "loss": 4.908,
-      "step": 8028
-    },
-    {
-      "epoch": 4.18722294654498,
-      "grad_norm": 1.3269590139389038,
-      "learning_rate": 9.243517587939699e-05,
-      "loss": 5.6341,
-      "step": 8029
-    },
-    {
-      "epoch": 4.1877444589309,
-      "grad_norm": 1.4500688314437866,
-      "learning_rate": 9.243417085427137e-05,
-      "loss": 5.6559,
-      "step": 8030
-    },
-    {
-      "epoch": 4.188265971316818,
-      "grad_norm": 1.6876097917556763,
-      "learning_rate": 9.243316582914573e-05,
-      "loss": 5.7327,
-      "step": 8031
-    },
-    {
-      "epoch": 4.188787483702738,
-      "grad_norm": 1.7296757698059082,
-      "learning_rate": 9.243216080402011e-05,
-      "loss": 5.6283,
-      "step": 8032
-    },
-    {
-      "epoch": 4.189308996088657,
-      "grad_norm": 1.5519893169403076,
-      "learning_rate": 9.243115577889447e-05,
-      "loss": 5.474,
-      "step": 8033
-    },
-    {
-      "epoch": 4.189830508474576,
-      "grad_norm": 1.4642395973205566,
-      "learning_rate": 9.243015075376885e-05,
-      "loss": 5.7745,
-      "step": 8034
-    },
-    {
-      "epoch": 4.190352020860495,
-      "grad_norm": 1.4915616512298584,
-      "learning_rate": 9.242914572864322e-05,
-      "loss": 5.6578,
-      "step": 8035
-    },
-    {
-      "epoch": 4.190873533246415,
-      "grad_norm": 1.5998424291610718,
-      "learning_rate": 9.24281407035176e-05,
-      "loss": 5.8621,
-      "step": 8036
-    },
-    {
-      "epoch": 4.1913950456323334,
-      "grad_norm": 1.5893317461013794,
-      "learning_rate": 9.242713567839196e-05,
-      "loss": 5.7653,
-      "step": 8037
-    },
-    {
-      "epoch": 4.191916558018253,
-      "grad_norm": 1.6256020069122314,
-      "learning_rate": 9.242613065326634e-05,
-      "loss": 5.5599,
-      "step": 8038
-    },
-    {
-      "epoch": 4.192438070404172,
-      "grad_norm": 1.5049679279327393,
-      "learning_rate": 9.242512562814071e-05,
-      "loss": 5.6968,
-      "step": 8039
-    },
-    {
-      "epoch": 4.192959582790091,
-      "grad_norm": 1.598910927772522,
-      "learning_rate": 9.242412060301509e-05,
-      "loss": 5.2,
-      "step": 8040
-    },
-    {
-      "epoch": 4.19348109517601,
-      "grad_norm": 1.5259355306625366,
-      "learning_rate": 9.242311557788946e-05,
-      "loss": 5.9417,
-      "step": 8041
-    },
-    {
-      "epoch": 4.19400260756193,
-      "grad_norm": 1.4717252254486084,
-      "learning_rate": 9.242211055276382e-05,
-      "loss": 5.6665,
-      "step": 8042
-    },
-    {
-      "epoch": 4.1945241199478485,
-      "grad_norm": 1.3602797985076904,
-      "learning_rate": 9.24211055276382e-05,
-      "loss": 5.8723,
-      "step": 8043
-    },
-    {
-      "epoch": 4.195045632333768,
-      "grad_norm": 1.5320335626602173,
-      "learning_rate": 9.242010050251256e-05,
-      "loss": 5.8053,
-      "step": 8044
-    },
-    {
-      "epoch": 4.195567144719687,
-      "grad_norm": 1.3722988367080688,
-      "learning_rate": 9.241909547738694e-05,
-      "loss": 5.7699,
-      "step": 8045
-    },
-    {
-      "epoch": 4.196088657105606,
-      "grad_norm": 1.485425353050232,
-      "learning_rate": 9.24180904522613e-05,
-      "loss": 5.5694,
-      "step": 8046
-    },
-    {
-      "epoch": 4.196610169491525,
-      "grad_norm": 1.444112777709961,
-      "learning_rate": 9.241708542713568e-05,
-      "loss": 5.9495,
-      "step": 8047
-    },
-    {
-      "epoch": 4.197131681877445,
-      "grad_norm": 1.403891682624817,
-      "learning_rate": 9.241608040201005e-05,
-      "loss": 5.7753,
-      "step": 8048
-    },
-    {
-      "epoch": 4.1976531942633635,
-      "grad_norm": 1.435088872909546,
-      "learning_rate": 9.241507537688442e-05,
-      "loss": 5.7373,
-      "step": 8049
-    },
-    {
-      "epoch": 4.198174706649283,
-      "grad_norm": 1.491249918937683,
-      "learning_rate": 9.24140703517588e-05,
-      "loss": 5.9124,
-      "step": 8050
-    },
-    {
-      "epoch": 4.198696219035202,
-      "grad_norm": 1.3308773040771484,
-      "learning_rate": 9.241306532663318e-05,
-      "loss": 6.0038,
-      "step": 8051
-    },
-    {
-      "epoch": 4.199217731421121,
-      "grad_norm": 1.4706188440322876,
-      "learning_rate": 9.241206030150754e-05,
-      "loss": 5.6725,
-      "step": 8052
-    },
-    {
-      "epoch": 4.1997392438070404,
-      "grad_norm": 1.4802765846252441,
-      "learning_rate": 9.241105527638192e-05,
-      "loss": 5.5719,
-      "step": 8053
-    },
-    {
-      "epoch": 4.20026075619296,
-      "grad_norm": 1.4592845439910889,
-      "learning_rate": 9.241005025125629e-05,
-      "loss": 5.7131,
-      "step": 8054
-    },
-    {
-      "epoch": 4.2007822685788785,
-      "grad_norm": 1.630377173423767,
-      "learning_rate": 9.240904522613065e-05,
-      "loss": 5.362,
-      "step": 8055
-    },
-    {
-      "epoch": 4.201303780964798,
-      "grad_norm": 1.4707138538360596,
-      "learning_rate": 9.240804020100503e-05,
-      "loss": 5.6395,
-      "step": 8056
-    },
-    {
-      "epoch": 4.201825293350717,
-      "grad_norm": 1.4473018646240234,
-      "learning_rate": 9.240703517587939e-05,
-      "loss": 5.2294,
-      "step": 8057
-    },
-    {
-      "epoch": 4.202346805736636,
-      "grad_norm": 1.4597961902618408,
-      "learning_rate": 9.240603015075377e-05,
-      "loss": 5.665,
-      "step": 8058
-    },
-    {
-      "epoch": 4.2028683181225555,
-      "grad_norm": 1.6309971809387207,
-      "learning_rate": 9.240502512562815e-05,
-      "loss": 5.4593,
-      "step": 8059
-    },
-    {
-      "epoch": 4.203389830508475,
-      "grad_norm": 1.3604625463485718,
-      "learning_rate": 9.240402010050253e-05,
-      "loss": 5.9838,
-      "step": 8060
-    },
-    {
-      "epoch": 4.2039113428943935,
-      "grad_norm": 1.3801305294036865,
-      "learning_rate": 9.240301507537689e-05,
-      "loss": 5.9604,
-      "step": 8061
-    },
-    {
-      "epoch": 4.204432855280313,
-      "grad_norm": 1.6933174133300781,
-      "learning_rate": 9.240201005025127e-05,
-      "loss": 5.5815,
-      "step": 8062
-    },
-    {
-      "epoch": 4.204954367666232,
-      "grad_norm": 1.545485019683838,
-      "learning_rate": 9.240100502512563e-05,
-      "loss": 5.7194,
-      "step": 8063
-    },
-    {
-      "epoch": 4.205475880052151,
-      "grad_norm": 1.4869271516799927,
-      "learning_rate": 9.240000000000001e-05,
-      "loss": 5.5889,
-      "step": 8064
-    },
-    {
-      "epoch": 4.2059973924380705,
-      "grad_norm": 1.4415230751037598,
-      "learning_rate": 9.239899497487437e-05,
-      "loss": 5.8783,
-      "step": 8065
-    },
-    {
-      "epoch": 4.20651890482399,
-      "grad_norm": 1.5789380073547363,
-      "learning_rate": 9.239798994974875e-05,
-      "loss": 5.5629,
-      "step": 8066
-    },
-    {
-      "epoch": 4.2070404172099085,
-      "grad_norm": 1.4147603511810303,
-      "learning_rate": 9.239698492462312e-05,
-      "loss": 5.7243,
-      "step": 8067
-    },
-    {
-      "epoch": 4.207561929595828,
-      "grad_norm": 1.3588975667953491,
-      "learning_rate": 9.239597989949748e-05,
-      "loss": 5.8473,
-      "step": 8068
-    },
-    {
-      "epoch": 4.208083441981747,
-      "grad_norm": 1.413963794708252,
-      "learning_rate": 9.239497487437186e-05,
-      "loss": 5.7056,
-      "step": 8069
-    },
-    {
-      "epoch": 4.208604954367666,
-      "grad_norm": 1.479322075843811,
-      "learning_rate": 9.239396984924624e-05,
-      "loss": 5.7815,
-      "step": 8070
-    },
-    {
-      "epoch": 4.2091264667535855,
-      "grad_norm": 1.493443489074707,
-      "learning_rate": 9.239296482412061e-05,
-      "loss": 5.6818,
-      "step": 8071
-    },
-    {
-      "epoch": 4.209647979139505,
-      "grad_norm": 1.341110110282898,
-      "learning_rate": 9.239195979899498e-05,
-      "loss": 5.9466,
-      "step": 8072
-    },
-    {
-      "epoch": 4.2101694915254235,
-      "grad_norm": 1.4842380285263062,
-      "learning_rate": 9.239095477386936e-05,
-      "loss": 5.8909,
-      "step": 8073
-    },
-    {
-      "epoch": 4.210691003911343,
-      "grad_norm": 1.4551920890808105,
-      "learning_rate": 9.238994974874372e-05,
-      "loss": 5.5545,
-      "step": 8074
-    },
-    {
-      "epoch": 4.2112125162972625,
-      "grad_norm": 1.4633193016052246,
-      "learning_rate": 9.23889447236181e-05,
-      "loss": 5.3482,
-      "step": 8075
-    },
-    {
-      "epoch": 4.211734028683181,
-      "grad_norm": 1.6214423179626465,
-      "learning_rate": 9.238793969849246e-05,
-      "loss": 5.0165,
-      "step": 8076
-    },
-    {
-      "epoch": 4.2122555410691005,
-      "grad_norm": 1.3305046558380127,
-      "learning_rate": 9.238693467336684e-05,
-      "loss": 6.0842,
-      "step": 8077
-    },
-    {
-      "epoch": 4.21277705345502,
-      "grad_norm": 1.534300446510315,
-      "learning_rate": 9.23859296482412e-05,
-      "loss": 5.573,
-      "step": 8078
-    },
-    {
-      "epoch": 4.2132985658409385,
-      "grad_norm": 1.4500904083251953,
-      "learning_rate": 9.238492462311558e-05,
-      "loss": 5.5678,
-      "step": 8079
-    },
-    {
-      "epoch": 4.213820078226858,
-      "grad_norm": 1.4248017072677612,
-      "learning_rate": 9.238391959798996e-05,
-      "loss": 5.6908,
-      "step": 8080
-    },
-    {
-      "epoch": 4.2143415906127775,
-      "grad_norm": 1.4029027223587036,
-      "learning_rate": 9.238291457286432e-05,
-      "loss": 5.1869,
-      "step": 8081
-    },
-    {
-      "epoch": 4.214863102998696,
-      "grad_norm": 1.471298336982727,
-      "learning_rate": 9.23819095477387e-05,
-      "loss": 5.8501,
-      "step": 8082
-    },
-    {
-      "epoch": 4.2153846153846155,
-      "grad_norm": 1.5383131504058838,
-      "learning_rate": 9.238090452261307e-05,
-      "loss": 5.3858,
-      "step": 8083
-    },
-    {
-      "epoch": 4.215906127770535,
-      "grad_norm": 1.4498381614685059,
-      "learning_rate": 9.237989949748744e-05,
-      "loss": 5.6163,
-      "step": 8084
-    },
-    {
-      "epoch": 4.2164276401564535,
-      "grad_norm": 1.5453029870986938,
-      "learning_rate": 9.237889447236181e-05,
-      "loss": 5.5652,
-      "step": 8085
-    },
-    {
-      "epoch": 4.216949152542373,
-      "grad_norm": 1.442050814628601,
-      "learning_rate": 9.237788944723619e-05,
-      "loss": 5.879,
-      "step": 8086
-    },
-    {
-      "epoch": 4.2174706649282925,
-      "grad_norm": 1.4322459697723389,
-      "learning_rate": 9.237688442211055e-05,
-      "loss": 6.0147,
-      "step": 8087
-    },
-    {
-      "epoch": 4.217992177314211,
-      "grad_norm": 1.515594720840454,
-      "learning_rate": 9.237587939698493e-05,
-      "loss": 5.2921,
-      "step": 8088
-    },
-    {
-      "epoch": 4.2185136897001305,
-      "grad_norm": 1.3660866022109985,
-      "learning_rate": 9.237487437185929e-05,
-      "loss": 5.762,
-      "step": 8089
-    },
-    {
-      "epoch": 4.219035202086049,
-      "grad_norm": 1.4891606569290161,
-      "learning_rate": 9.237386934673367e-05,
-      "loss": 4.9394,
-      "step": 8090
-    },
-    {
-      "epoch": 4.219556714471969,
-      "grad_norm": 1.5223805904388428,
-      "learning_rate": 9.237286432160805e-05,
-      "loss": 5.6508,
-      "step": 8091
-    },
-    {
-      "epoch": 4.220078226857888,
-      "grad_norm": 1.5101184844970703,
-      "learning_rate": 9.237185929648243e-05,
-      "loss": 5.4397,
-      "step": 8092
-    },
-    {
-      "epoch": 4.2205997392438075,
-      "grad_norm": 1.6051876544952393,
-      "learning_rate": 9.237085427135679e-05,
-      "loss": 5.8335,
-      "step": 8093
-    },
-    {
-      "epoch": 4.221121251629726,
-      "grad_norm": 1.9319525957107544,
-      "learning_rate": 9.236984924623115e-05,
-      "loss": 5.1592,
-      "step": 8094
-    },
-    {
-      "epoch": 4.2216427640156455,
-      "grad_norm": 1.6756622791290283,
-      "learning_rate": 9.236884422110553e-05,
-      "loss": 5.4938,
-      "step": 8095
-    },
-    {
-      "epoch": 4.222164276401564,
-      "grad_norm": 1.4761368036270142,
-      "learning_rate": 9.23678391959799e-05,
-      "loss": 5.5827,
-      "step": 8096
-    },
-    {
-      "epoch": 4.222685788787484,
-      "grad_norm": 1.4994632005691528,
-      "learning_rate": 9.236683417085427e-05,
-      "loss": 5.9181,
-      "step": 8097
-    },
-    {
-      "epoch": 4.223207301173403,
-      "grad_norm": 1.5741597414016724,
-      "learning_rate": 9.236582914572864e-05,
-      "loss": 5.3733,
-      "step": 8098
-    },
-    {
-      "epoch": 4.223728813559322,
-      "grad_norm": 1.4558801651000977,
-      "learning_rate": 9.236482412060302e-05,
-      "loss": 5.8769,
-      "step": 8099
-    },
-    {
-      "epoch": 4.224250325945241,
-      "grad_norm": 1.7334845066070557,
-      "learning_rate": 9.23638190954774e-05,
-      "loss": 5.1694,
-      "step": 8100
-    },
-    {
-      "epoch": 4.2247718383311605,
-      "grad_norm": 1.4909974336624146,
-      "learning_rate": 9.236281407035177e-05,
-      "loss": 4.9027,
-      "step": 8101
-    },
-    {
-      "epoch": 4.225293350717079,
-      "grad_norm": 1.4431517124176025,
-      "learning_rate": 9.236180904522614e-05,
-      "loss": 5.9496,
-      "step": 8102
-    },
-    {
-      "epoch": 4.225814863102999,
-      "grad_norm": 1.3642315864562988,
-      "learning_rate": 9.236080402010051e-05,
-      "loss": 5.9466,
-      "step": 8103
-    },
-    {
-      "epoch": 4.226336375488918,
-      "grad_norm": 1.3197346925735474,
-      "learning_rate": 9.235979899497488e-05,
-      "loss": 6.1465,
-      "step": 8104
-    },
-    {
-      "epoch": 4.226857887874837,
-      "grad_norm": 2.0636520385742188,
-      "learning_rate": 9.235879396984926e-05,
-      "loss": 3.8172,
-      "step": 8105
-    },
-    {
-      "epoch": 4.227379400260756,
-      "grad_norm": 1.411077857017517,
-      "learning_rate": 9.235778894472362e-05,
-      "loss": 6.0691,
-      "step": 8106
-    },
-    {
-      "epoch": 4.227900912646676,
-      "grad_norm": 1.543269395828247,
-      "learning_rate": 9.235678391959799e-05,
-      "loss": 5.7796,
-      "step": 8107
-    },
-    {
-      "epoch": 4.228422425032594,
-      "grad_norm": 1.3450006246566772,
-      "learning_rate": 9.235577889447236e-05,
-      "loss": 5.8724,
-      "step": 8108
-    },
-    {
-      "epoch": 4.228943937418514,
-      "grad_norm": 1.3910506963729858,
-      "learning_rate": 9.235477386934673e-05,
-      "loss": 5.9735,
-      "step": 8109
-    },
-    {
-      "epoch": 4.229465449804433,
-      "grad_norm": 1.4848816394805908,
-      "learning_rate": 9.23537688442211e-05,
-      "loss": 5.5918,
-      "step": 8110
-    },
-    {
-      "epoch": 4.229986962190352,
-      "grad_norm": 1.3522087335586548,
-      "learning_rate": 9.235276381909548e-05,
-      "loss": 5.9402,
-      "step": 8111
-    },
-    {
-      "epoch": 4.230508474576271,
-      "grad_norm": 1.4571064710617065,
-      "learning_rate": 9.235175879396986e-05,
-      "loss": 5.9054,
-      "step": 8112
-    },
-    {
-      "epoch": 4.231029986962191,
-      "grad_norm": 1.4212530851364136,
-      "learning_rate": 9.235075376884423e-05,
-      "loss": 6.0357,
-      "step": 8113
-    },
-    {
-      "epoch": 4.231551499348109,
-      "grad_norm": 1.387215495109558,
-      "learning_rate": 9.23497487437186e-05,
-      "loss": 5.8226,
-      "step": 8114
-    },
-    {
-      "epoch": 4.232073011734029,
-      "grad_norm": 1.4048011302947998,
-      "learning_rate": 9.234874371859297e-05,
-      "loss": 5.9157,
-      "step": 8115
-    },
-    {
-      "epoch": 4.232594524119948,
-      "grad_norm": 1.3140244483947754,
-      "learning_rate": 9.234773869346735e-05,
-      "loss": 5.9048,
-      "step": 8116
-    },
-    {
-      "epoch": 4.233116036505867,
-      "grad_norm": 1.5777630805969238,
-      "learning_rate": 9.234673366834171e-05,
-      "loss": 5.5979,
-      "step": 8117
-    },
-    {
-      "epoch": 4.233637548891786,
-      "grad_norm": 1.4291855096817017,
-      "learning_rate": 9.234572864321609e-05,
-      "loss": 5.9305,
-      "step": 8118
-    },
-    {
-      "epoch": 4.234159061277706,
-      "grad_norm": 1.5020042657852173,
-      "learning_rate": 9.234472361809045e-05,
-      "loss": 5.4507,
-      "step": 8119
-    },
-    {
-      "epoch": 4.234680573663624,
-      "grad_norm": 1.6197021007537842,
-      "learning_rate": 9.234371859296483e-05,
-      "loss": 5.8418,
-      "step": 8120
-    },
-    {
-      "epoch": 4.235202086049544,
-      "grad_norm": 1.4735828638076782,
-      "learning_rate": 9.234271356783921e-05,
-      "loss": 5.5402,
-      "step": 8121
-    },
-    {
-      "epoch": 4.235723598435463,
-      "grad_norm": 1.4384886026382446,
-      "learning_rate": 9.234170854271357e-05,
-      "loss": 5.7671,
-      "step": 8122
-    },
-    {
-      "epoch": 4.236245110821382,
-      "grad_norm": 1.551974892616272,
-      "learning_rate": 9.234070351758795e-05,
-      "loss": 5.6608,
-      "step": 8123
-    },
-    {
-      "epoch": 4.236766623207301,
-      "grad_norm": 1.574962854385376,
-      "learning_rate": 9.233969849246231e-05,
-      "loss": 5.7942,
-      "step": 8124
-    },
-    {
-      "epoch": 4.237288135593221,
-      "grad_norm": 1.3233717679977417,
-      "learning_rate": 9.233869346733669e-05,
-      "loss": 5.9314,
-      "step": 8125
-    },
-    {
-      "epoch": 4.237809647979139,
-      "grad_norm": 1.480355143547058,
-      "learning_rate": 9.233768844221106e-05,
-      "loss": 5.6188,
-      "step": 8126
-    },
-    {
-      "epoch": 4.238331160365059,
-      "grad_norm": 1.4255951642990112,
-      "learning_rate": 9.233668341708543e-05,
-      "loss": 5.8434,
-      "step": 8127
-    },
-    {
-      "epoch": 4.238852672750978,
-      "grad_norm": 1.6663564443588257,
-      "learning_rate": 9.23356783919598e-05,
-      "loss": 5.9194,
-      "step": 8128
-    },
-    {
-      "epoch": 4.239374185136897,
-      "grad_norm": 1.5636684894561768,
-      "learning_rate": 9.233467336683418e-05,
-      "loss": 5.7119,
-      "step": 8129
-    },
-    {
-      "epoch": 4.239895697522816,
-      "grad_norm": 1.4824167490005493,
-      "learning_rate": 9.233366834170854e-05,
-      "loss": 5.5152,
-      "step": 8130
-    },
-    {
-      "epoch": 4.240417209908736,
-      "grad_norm": 1.6236249208450317,
-      "learning_rate": 9.233266331658292e-05,
-      "loss": 5.6748,
-      "step": 8131
-    },
-    {
-      "epoch": 4.240938722294654,
-      "grad_norm": 1.6402028799057007,
-      "learning_rate": 9.23316582914573e-05,
-      "loss": 5.7157,
-      "step": 8132
-    },
-    {
-      "epoch": 4.241460234680574,
-      "grad_norm": 1.5611157417297363,
-      "learning_rate": 9.233065326633167e-05,
-      "loss": 5.6647,
-      "step": 8133
-    },
-    {
-      "epoch": 4.241981747066493,
-      "grad_norm": 1.4712896347045898,
-      "learning_rate": 9.232964824120604e-05,
-      "loss": 5.8031,
-      "step": 8134
-    },
-    {
-      "epoch": 4.242503259452412,
-      "grad_norm": 1.3820230960845947,
-      "learning_rate": 9.23286432160804e-05,
-      "loss": 6.2216,
-      "step": 8135
-    },
-    {
-      "epoch": 4.243024771838331,
-      "grad_norm": 1.7106844186782837,
-      "learning_rate": 9.232763819095478e-05,
-      "loss": 5.2614,
-      "step": 8136
-    },
-    {
-      "epoch": 4.243546284224251,
-      "grad_norm": 1.5530357360839844,
-      "learning_rate": 9.232663316582914e-05,
-      "loss": 5.8776,
-      "step": 8137
-    },
-    {
-      "epoch": 4.244067796610169,
-      "grad_norm": 1.416183352470398,
-      "learning_rate": 9.232562814070352e-05,
-      "loss": 5.216,
-      "step": 8138
-    },
-    {
-      "epoch": 4.244589308996089,
-      "grad_norm": 1.6258443593978882,
-      "learning_rate": 9.232462311557789e-05,
-      "loss": 5.6457,
-      "step": 8139
-    },
-    {
-      "epoch": 4.245110821382008,
-      "grad_norm": 1.585181474685669,
-      "learning_rate": 9.232361809045226e-05,
-      "loss": 5.2485,
-      "step": 8140
-    },
-    {
-      "epoch": 4.245632333767927,
-      "grad_norm": 1.6142514944076538,
-      "learning_rate": 9.232261306532664e-05,
-      "loss": 5.9728,
-      "step": 8141
-    },
-    {
-      "epoch": 4.246153846153846,
-      "grad_norm": 1.4848946332931519,
-      "learning_rate": 9.232160804020102e-05,
-      "loss": 5.6655,
-      "step": 8142
-    },
-    {
-      "epoch": 4.246675358539766,
-      "grad_norm": 1.508735179901123,
-      "learning_rate": 9.232060301507538e-05,
-      "loss": 5.5072,
-      "step": 8143
-    },
-    {
-      "epoch": 4.247196870925684,
-      "grad_norm": 1.488559365272522,
-      "learning_rate": 9.231959798994976e-05,
-      "loss": 5.1086,
-      "step": 8144
-    },
-    {
-      "epoch": 4.247718383311604,
-      "grad_norm": 1.5528466701507568,
-      "learning_rate": 9.231859296482413e-05,
-      "loss": 5.7921,
-      "step": 8145
-    },
-    {
-      "epoch": 4.248239895697523,
-      "grad_norm": 1.594647765159607,
-      "learning_rate": 9.23175879396985e-05,
-      "loss": 5.8262,
-      "step": 8146
-    },
-    {
-      "epoch": 4.248761408083442,
-      "grad_norm": 1.522204875946045,
-      "learning_rate": 9.231658291457287e-05,
-      "loss": 5.453,
-      "step": 8147
-    },
-    {
-      "epoch": 4.249282920469361,
-      "grad_norm": 1.4430971145629883,
-      "learning_rate": 9.231557788944723e-05,
-      "loss": 5.8618,
-      "step": 8148
-    },
-    {
-      "epoch": 4.249804432855281,
-      "grad_norm": 1.510711669921875,
-      "learning_rate": 9.231457286432161e-05,
-      "loss": 5.7433,
-      "step": 8149
-    },
-    {
-      "epoch": 4.250325945241199,
-      "grad_norm": 1.4264568090438843,
-      "learning_rate": 9.231356783919597e-05,
-      "loss": 6.0582,
-      "step": 8150
-    },
-    {
-      "epoch": 4.250847457627119,
-      "grad_norm": 1.5914751291275024,
-      "learning_rate": 9.231256281407035e-05,
-      "loss": 5.5834,
-      "step": 8151
-    },
-    {
-      "epoch": 4.251368970013038,
-      "grad_norm": 1.624890923500061,
-      "learning_rate": 9.231155778894473e-05,
-      "loss": 5.4993,
-      "step": 8152
-    },
-    {
-      "epoch": 4.251890482398957,
-      "grad_norm": 1.559191346168518,
-      "learning_rate": 9.231055276381911e-05,
-      "loss": 5.2716,
-      "step": 8153
-    },
-    {
-      "epoch": 4.252411994784876,
-      "grad_norm": 1.663712501525879,
-      "learning_rate": 9.230954773869347e-05,
-      "loss": 5.8165,
-      "step": 8154
-    },
-    {
-      "epoch": 4.252933507170796,
-      "grad_norm": 1.4753141403198242,
-      "learning_rate": 9.230854271356785e-05,
-      "loss": 5.3953,
-      "step": 8155
-    },
-    {
-      "epoch": 4.253455019556714,
-      "grad_norm": 1.5185589790344238,
-      "learning_rate": 9.230753768844221e-05,
-      "loss": 5.5593,
-      "step": 8156
-    },
-    {
-      "epoch": 4.253976531942634,
-      "grad_norm": 1.4911926984786987,
-      "learning_rate": 9.230653266331659e-05,
-      "loss": 5.7487,
-      "step": 8157
-    },
-    {
-      "epoch": 4.254498044328553,
-      "grad_norm": 1.5621613264083862,
-      "learning_rate": 9.230552763819096e-05,
-      "loss": 5.432,
-      "step": 8158
-    },
-    {
-      "epoch": 4.255019556714472,
-      "grad_norm": 1.4591282606124878,
-      "learning_rate": 9.230452261306533e-05,
-      "loss": 5.4393,
-      "step": 8159
-    },
-    {
-      "epoch": 4.255541069100391,
-      "grad_norm": 1.4202383756637573,
-      "learning_rate": 9.23035175879397e-05,
-      "loss": 5.5061,
-      "step": 8160
-    },
-    {
-      "epoch": 4.256062581486311,
-      "grad_norm": 1.8698112964630127,
-      "learning_rate": 9.230251256281408e-05,
-      "loss": 5.3721,
-      "step": 8161
-    },
-    {
-      "epoch": 4.256584093872229,
-      "grad_norm": 1.4089279174804688,
-      "learning_rate": 9.230150753768845e-05,
-      "loss": 5.9104,
-      "step": 8162
-    },
-    {
-      "epoch": 4.257105606258149,
-      "grad_norm": 1.3710532188415527,
-      "learning_rate": 9.230050251256282e-05,
-      "loss": 5.7923,
-      "step": 8163
-    },
-    {
-      "epoch": 4.257627118644068,
-      "grad_norm": 1.6177845001220703,
-      "learning_rate": 9.22994974874372e-05,
-      "loss": 5.5642,
-      "step": 8164
-    },
-    {
-      "epoch": 4.258148631029987,
-      "grad_norm": 1.7065272331237793,
-      "learning_rate": 9.229849246231156e-05,
-      "loss": 5.3446,
-      "step": 8165
-    },
-    {
-      "epoch": 4.258670143415906,
-      "grad_norm": 1.9377055168151855,
-      "learning_rate": 9.229748743718594e-05,
-      "loss": 5.4428,
-      "step": 8166
-    },
-    {
-      "epoch": 4.259191655801826,
-      "grad_norm": 1.6348875761032104,
-      "learning_rate": 9.22964824120603e-05,
-      "loss": 5.5372,
-      "step": 8167
-    },
-    {
-      "epoch": 4.259713168187744,
-      "grad_norm": 1.6998919248580933,
-      "learning_rate": 9.229547738693468e-05,
-      "loss": 5.7976,
-      "step": 8168
-    },
-    {
-      "epoch": 4.260234680573664,
-      "grad_norm": 1.843271255493164,
-      "learning_rate": 9.229447236180904e-05,
-      "loss": 5.8798,
-      "step": 8169
-    },
-    {
-      "epoch": 4.260756192959583,
-      "grad_norm": 1.5218173265457153,
-      "learning_rate": 9.229346733668342e-05,
-      "loss": 5.9107,
-      "step": 8170
-    },
-    {
-      "epoch": 4.261277705345502,
-      "grad_norm": 1.6195333003997803,
-      "learning_rate": 9.229246231155779e-05,
-      "loss": 5.3999,
-      "step": 8171
-    },
-    {
-      "epoch": 4.261799217731421,
-      "grad_norm": 1.7671188116073608,
-      "learning_rate": 9.229145728643216e-05,
-      "loss": 5.5851,
-      "step": 8172
-    },
-    {
-      "epoch": 4.262320730117341,
-      "grad_norm": 1.436708688735962,
-      "learning_rate": 9.229045226130654e-05,
-      "loss": 5.9291,
-      "step": 8173
-    },
-    {
-      "epoch": 4.262842242503259,
-      "grad_norm": 1.337443470954895,
-      "learning_rate": 9.228944723618091e-05,
-      "loss": 5.9242,
-      "step": 8174
-    },
-    {
-      "epoch": 4.263363754889179,
-      "grad_norm": 1.7119444608688354,
-      "learning_rate": 9.228844221105528e-05,
-      "loss": 5.3234,
-      "step": 8175
-    },
-    {
-      "epoch": 4.263885267275098,
-      "grad_norm": 1.4962674379348755,
-      "learning_rate": 9.228743718592965e-05,
-      "loss": 5.6466,
-      "step": 8176
-    },
-    {
-      "epoch": 4.264406779661017,
-      "grad_norm": 1.4163269996643066,
-      "learning_rate": 9.228643216080403e-05,
-      "loss": 5.6879,
-      "step": 8177
-    },
-    {
-      "epoch": 4.264928292046936,
-      "grad_norm": 1.393664836883545,
-      "learning_rate": 9.228542713567839e-05,
-      "loss": 5.5205,
-      "step": 8178
-    },
-    {
-      "epoch": 4.265449804432855,
-      "grad_norm": 1.4195810556411743,
-      "learning_rate": 9.228442211055277e-05,
-      "loss": 5.6404,
-      "step": 8179
-    },
-    {
-      "epoch": 4.265971316818774,
-      "grad_norm": 1.4232943058013916,
-      "learning_rate": 9.228341708542713e-05,
-      "loss": 5.925,
-      "step": 8180
-    },
-    {
-      "epoch": 4.266492829204694,
-      "grad_norm": 1.4507029056549072,
-      "learning_rate": 9.228241206030151e-05,
-      "loss": 5.5794,
-      "step": 8181
-    },
-    {
-      "epoch": 4.267014341590613,
-      "grad_norm": 1.7533091306686401,
-      "learning_rate": 9.228140703517588e-05,
-      "loss": 5.5335,
-      "step": 8182
-    },
-    {
-      "epoch": 4.267535853976532,
-      "grad_norm": 1.7871967554092407,
-      "learning_rate": 9.228040201005025e-05,
-      "loss": 5.5422,
-      "step": 8183
-    },
-    {
-      "epoch": 4.268057366362451,
-      "grad_norm": 1.6704190969467163,
-      "learning_rate": 9.227939698492463e-05,
-      "loss": 5.6682,
-      "step": 8184
-    },
-    {
-      "epoch": 4.26857887874837,
-      "grad_norm": 1.37361478805542,
-      "learning_rate": 9.227839195979901e-05,
-      "loss": 5.9673,
-      "step": 8185
-    },
-    {
-      "epoch": 4.269100391134289,
-      "grad_norm": 1.5022215843200684,
-      "learning_rate": 9.227738693467337e-05,
-      "loss": 6.0113,
-      "step": 8186
-    },
-    {
-      "epoch": 4.269621903520209,
-      "grad_norm": 1.3324862718582153,
-      "learning_rate": 9.227638190954774e-05,
-      "loss": 6.1827,
-      "step": 8187
-    },
-    {
-      "epoch": 4.270143415906128,
-      "grad_norm": 1.4685600996017456,
-      "learning_rate": 9.227537688442212e-05,
-      "loss": 5.808,
-      "step": 8188
-    },
-    {
-      "epoch": 4.270664928292047,
-      "grad_norm": 1.6173632144927979,
-      "learning_rate": 9.227437185929648e-05,
-      "loss": 5.374,
-      "step": 8189
-    },
-    {
-      "epoch": 4.271186440677966,
-      "grad_norm": 1.5838748216629028,
-      "learning_rate": 9.227336683417086e-05,
-      "loss": 5.4901,
-      "step": 8190
-    },
-    {
-      "epoch": 4.271707953063885,
-      "grad_norm": 1.6378674507141113,
-      "learning_rate": 9.227236180904522e-05,
-      "loss": 5.0881,
-      "step": 8191
-    },
-    {
-      "epoch": 4.272229465449804,
-      "grad_norm": 1.5104975700378418,
-      "learning_rate": 9.22713567839196e-05,
-      "loss": 5.7734,
-      "step": 8192
-    },
-    {
-      "epoch": 4.272750977835724,
-      "grad_norm": 1.5110844373703003,
-      "learning_rate": 9.227035175879398e-05,
-      "loss": 5.7168,
-      "step": 8193
-    },
-    {
-      "epoch": 4.273272490221642,
-      "grad_norm": 1.524345874786377,
-      "learning_rate": 9.226934673366835e-05,
-      "loss": 5.417,
-      "step": 8194
-    },
-    {
-      "epoch": 4.273794002607562,
-      "grad_norm": 1.5579891204833984,
-      "learning_rate": 9.226834170854272e-05,
-      "loss": 5.137,
-      "step": 8195
-    },
-    {
-      "epoch": 4.274315514993481,
-      "grad_norm": 1.3418807983398438,
-      "learning_rate": 9.22673366834171e-05,
-      "loss": 6.0523,
-      "step": 8196
-    },
-    {
-      "epoch": 4.2748370273794,
-      "grad_norm": 1.3959513902664185,
-      "learning_rate": 9.226633165829146e-05,
-      "loss": 5.7976,
-      "step": 8197
-    },
-    {
-      "epoch": 4.275358539765319,
-      "grad_norm": 1.454040288925171,
-      "learning_rate": 9.226532663316584e-05,
-      "loss": 5.9359,
-      "step": 8198
-    },
-    {
-      "epoch": 4.275880052151239,
-      "grad_norm": 1.4026768207550049,
-      "learning_rate": 9.22643216080402e-05,
-      "loss": 5.7754,
-      "step": 8199
-    },
-    {
-      "epoch": 4.276401564537157,
-      "grad_norm": 1.3072502613067627,
-      "learning_rate": 9.226331658291457e-05,
-      "loss": 5.8386,
-      "step": 8200
-    },
-    {
-      "epoch": 4.276923076923077,
-      "grad_norm": 1.4089983701705933,
-      "learning_rate": 9.226231155778895e-05,
-      "loss": 5.7708,
-      "step": 8201
-    },
-    {
-      "epoch": 4.277444589308996,
-      "grad_norm": 1.3349277973175049,
-      "learning_rate": 9.226130653266331e-05,
-      "loss": 5.6967,
-      "step": 8202
-    },
-    {
-      "epoch": 4.277966101694915,
-      "grad_norm": 1.459149718284607,
-      "learning_rate": 9.226030150753769e-05,
-      "loss": 5.6941,
-      "step": 8203
-    },
-    {
-      "epoch": 4.278487614080834,
-      "grad_norm": 1.5814669132232666,
-      "learning_rate": 9.225929648241207e-05,
-      "loss": 5.2754,
-      "step": 8204
-    },
-    {
-      "epoch": 4.279009126466754,
-      "grad_norm": 1.4944489002227783,
-      "learning_rate": 9.225829145728644e-05,
-      "loss": 5.6699,
-      "step": 8205
-    },
-    {
-      "epoch": 4.279530638852672,
-      "grad_norm": 1.438774585723877,
-      "learning_rate": 9.225728643216081e-05,
-      "loss": 5.5953,
-      "step": 8206
-    },
-    {
-      "epoch": 4.280052151238592,
-      "grad_norm": 1.5741498470306396,
-      "learning_rate": 9.225628140703519e-05,
-      "loss": 5.5386,
-      "step": 8207
-    },
-    {
-      "epoch": 4.280573663624511,
-      "grad_norm": 1.5372480154037476,
-      "learning_rate": 9.225527638190955e-05,
-      "loss": 5.795,
-      "step": 8208
-    },
-    {
-      "epoch": 4.28109517601043,
-      "grad_norm": 1.536525845527649,
-      "learning_rate": 9.225427135678393e-05,
-      "loss": 5.6796,
-      "step": 8209
-    },
-    {
-      "epoch": 4.281616688396349,
-      "grad_norm": 1.7071441411972046,
-      "learning_rate": 9.225326633165829e-05,
-      "loss": 5.4935,
-      "step": 8210
-    },
-    {
-      "epoch": 4.282138200782269,
-      "grad_norm": 1.4764519929885864,
-      "learning_rate": 9.225226130653267e-05,
-      "loss": 6.0142,
-      "step": 8211
-    },
-    {
-      "epoch": 4.282659713168187,
-      "grad_norm": 1.614942193031311,
-      "learning_rate": 9.225125628140703e-05,
-      "loss": 5.4503,
-      "step": 8212
-    },
-    {
-      "epoch": 4.283181225554107,
-      "grad_norm": 1.6166187524795532,
-      "learning_rate": 9.225025125628141e-05,
-      "loss": 5.2353,
-      "step": 8213
-    },
-    {
-      "epoch": 4.283702737940026,
-      "grad_norm": 1.451638102531433,
-      "learning_rate": 9.224924623115579e-05,
-      "loss": 5.6093,
-      "step": 8214
-    },
-    {
-      "epoch": 4.284224250325945,
-      "grad_norm": 1.3889650106430054,
-      "learning_rate": 9.224824120603015e-05,
-      "loss": 5.9482,
-      "step": 8215
-    },
-    {
-      "epoch": 4.284745762711864,
-      "grad_norm": 1.9577572345733643,
-      "learning_rate": 9.224723618090453e-05,
-      "loss": 5.2527,
-      "step": 8216
-    },
-    {
-      "epoch": 4.285267275097784,
-      "grad_norm": 1.489493727684021,
-      "learning_rate": 9.22462311557789e-05,
-      "loss": 5.8727,
-      "step": 8217
-    },
-    {
-      "epoch": 4.285788787483702,
-      "grad_norm": 1.4302802085876465,
-      "learning_rate": 9.224522613065327e-05,
-      "loss": 6.0055,
-      "step": 8218
-    },
-    {
-      "epoch": 4.286310299869622,
-      "grad_norm": 1.3200981616973877,
-      "learning_rate": 9.224422110552764e-05,
-      "loss": 5.8409,
-      "step": 8219
-    },
-    {
-      "epoch": 4.286831812255541,
-      "grad_norm": 1.5777288675308228,
-      "learning_rate": 9.224321608040202e-05,
-      "loss": 5.2896,
-      "step": 8220
-    },
-    {
-      "epoch": 4.28735332464146,
-      "grad_norm": 1.763900637626648,
-      "learning_rate": 9.224221105527638e-05,
-      "loss": 4.8608,
-      "step": 8221
-    },
-    {
-      "epoch": 4.287874837027379,
-      "grad_norm": 1.4336694478988647,
-      "learning_rate": 9.224120603015076e-05,
-      "loss": 5.8655,
-      "step": 8222
-    },
-    {
-      "epoch": 4.288396349413299,
-      "grad_norm": 1.3968309164047241,
-      "learning_rate": 9.224020100502512e-05,
-      "loss": 5.249,
-      "step": 8223
-    },
-    {
-      "epoch": 4.288917861799217,
-      "grad_norm": 1.4851211309432983,
-      "learning_rate": 9.22391959798995e-05,
-      "loss": 6.0764,
-      "step": 8224
-    },
-    {
-      "epoch": 4.289439374185137,
-      "grad_norm": 1.6122325658798218,
-      "learning_rate": 9.223819095477388e-05,
-      "loss": 5.6006,
-      "step": 8225
-    },
-    {
-      "epoch": 4.289960886571056,
-      "grad_norm": 1.5910156965255737,
-      "learning_rate": 9.223718592964826e-05,
-      "loss": 5.6395,
-      "step": 8226
-    },
-    {
-      "epoch": 4.290482398956975,
-      "grad_norm": 1.6631559133529663,
-      "learning_rate": 9.223618090452262e-05,
-      "loss": 5.5392,
-      "step": 8227
-    },
-    {
-      "epoch": 4.291003911342894,
-      "grad_norm": 1.510602355003357,
-      "learning_rate": 9.223517587939698e-05,
-      "loss": 5.9885,
-      "step": 8228
-    },
-    {
-      "epoch": 4.291525423728814,
-      "grad_norm": 1.6547300815582275,
-      "learning_rate": 9.223417085427136e-05,
-      "loss": 5.4395,
-      "step": 8229
-    },
-    {
-      "epoch": 4.292046936114732,
-      "grad_norm": 1.6233593225479126,
-      "learning_rate": 9.223316582914573e-05,
-      "loss": 5.3968,
-      "step": 8230
-    },
-    {
-      "epoch": 4.292568448500652,
-      "grad_norm": 1.5003654956817627,
-      "learning_rate": 9.22321608040201e-05,
-      "loss": 5.7329,
-      "step": 8231
-    },
-    {
-      "epoch": 4.293089960886571,
-      "grad_norm": 1.7074711322784424,
-      "learning_rate": 9.223115577889447e-05,
-      "loss": 5.562,
-      "step": 8232
-    },
-    {
-      "epoch": 4.29361147327249,
-      "grad_norm": 1.3998714685440063,
-      "learning_rate": 9.223015075376885e-05,
-      "loss": 5.7568,
-      "step": 8233
-    },
-    {
-      "epoch": 4.294132985658409,
-      "grad_norm": 1.3641412258148193,
-      "learning_rate": 9.222914572864322e-05,
-      "loss": 5.9578,
-      "step": 8234
-    },
-    {
-      "epoch": 4.294654498044329,
-      "grad_norm": 1.5097975730895996,
-      "learning_rate": 9.22281407035176e-05,
-      "loss": 5.4587,
-      "step": 8235
-    },
-    {
-      "epoch": 4.295176010430247,
-      "grad_norm": 1.4471375942230225,
-      "learning_rate": 9.222713567839197e-05,
-      "loss": 5.8648,
-      "step": 8236
-    },
-    {
-      "epoch": 4.295697522816167,
-      "grad_norm": 1.5024443864822388,
-      "learning_rate": 9.222613065326634e-05,
-      "loss": 5.4659,
-      "step": 8237
-    },
-    {
-      "epoch": 4.296219035202086,
-      "grad_norm": 1.5273675918579102,
-      "learning_rate": 9.222512562814071e-05,
-      "loss": 5.8314,
-      "step": 8238
-    },
-    {
-      "epoch": 4.296740547588005,
-      "grad_norm": 1.409576177597046,
-      "learning_rate": 9.222412060301509e-05,
-      "loss": 5.9524,
-      "step": 8239
-    },
-    {
-      "epoch": 4.297262059973924,
-      "grad_norm": 1.3998360633850098,
-      "learning_rate": 9.222311557788945e-05,
-      "loss": 5.7659,
-      "step": 8240
-    },
-    {
-      "epoch": 4.297783572359844,
-      "grad_norm": 1.4987726211547852,
-      "learning_rate": 9.222211055276381e-05,
-      "loss": 5.7813,
-      "step": 8241
-    },
-    {
-      "epoch": 4.2983050847457624,
-      "grad_norm": 1.4891552925109863,
-      "learning_rate": 9.222110552763819e-05,
-      "loss": 5.8661,
-      "step": 8242
-    },
-    {
-      "epoch": 4.298826597131682,
-      "grad_norm": 1.4616069793701172,
-      "learning_rate": 9.222010050251256e-05,
-      "loss": 5.3899,
-      "step": 8243
-    },
-    {
-      "epoch": 4.299348109517601,
-      "grad_norm": 1.7069586515426636,
-      "learning_rate": 9.221909547738693e-05,
-      "loss": 5.781,
-      "step": 8244
-    },
-    {
-      "epoch": 4.29986962190352,
-      "grad_norm": 1.4463789463043213,
-      "learning_rate": 9.221809045226131e-05,
-      "loss": 5.7591,
-      "step": 8245
-    },
-    {
-      "epoch": 4.300391134289439,
-      "grad_norm": 1.6459046602249146,
-      "learning_rate": 9.221708542713569e-05,
-      "loss": 5.2271,
-      "step": 8246
-    },
-    {
-      "epoch": 4.300912646675359,
-      "grad_norm": 1.41484797000885,
-      "learning_rate": 9.221608040201005e-05,
-      "loss": 6.163,
-      "step": 8247
-    },
-    {
-      "epoch": 4.3014341590612775,
-      "grad_norm": 1.3951733112335205,
-      "learning_rate": 9.221507537688443e-05,
-      "loss": 5.7739,
-      "step": 8248
-    },
-    {
-      "epoch": 4.301955671447197,
-      "grad_norm": 1.483382225036621,
-      "learning_rate": 9.22140703517588e-05,
-      "loss": 5.7216,
-      "step": 8249
-    },
-    {
-      "epoch": 4.302477183833116,
-      "grad_norm": 1.5166797637939453,
-      "learning_rate": 9.221306532663317e-05,
-      "loss": 5.828,
-      "step": 8250
-    },
-    {
-      "epoch": 4.302998696219035,
-      "grad_norm": 1.7322787046432495,
-      "learning_rate": 9.221206030150754e-05,
-      "loss": 5.1868,
-      "step": 8251
-    },
-    {
-      "epoch": 4.303520208604954,
-      "grad_norm": 1.4843647480010986,
-      "learning_rate": 9.221105527638192e-05,
-      "loss": 5.3207,
-      "step": 8252
-    },
-    {
-      "epoch": 4.304041720990874,
-      "grad_norm": 1.5308222770690918,
-      "learning_rate": 9.221005025125628e-05,
-      "loss": 5.4294,
-      "step": 8253
-    },
-    {
-      "epoch": 4.3045632333767925,
-      "grad_norm": 1.370050311088562,
-      "learning_rate": 9.220904522613066e-05,
-      "loss": 5.8593,
-      "step": 8254
-    },
-    {
-      "epoch": 4.305084745762712,
-      "grad_norm": 1.439728021621704,
-      "learning_rate": 9.220804020100504e-05,
-      "loss": 5.2714,
-      "step": 8255
-    },
-    {
-      "epoch": 4.305606258148631,
-      "grad_norm": 1.3228998184204102,
-      "learning_rate": 9.22070351758794e-05,
-      "loss": 5.8065,
-      "step": 8256
-    },
-    {
-      "epoch": 4.30612777053455,
-      "grad_norm": 1.4349212646484375,
-      "learning_rate": 9.220603015075378e-05,
-      "loss": 5.8353,
-      "step": 8257
-    },
-    {
-      "epoch": 4.3066492829204694,
-      "grad_norm": 1.3288393020629883,
-      "learning_rate": 9.220502512562814e-05,
-      "loss": 6.1164,
-      "step": 8258
-    },
-    {
-      "epoch": 4.307170795306389,
-      "grad_norm": 1.3884156942367554,
-      "learning_rate": 9.220402010050252e-05,
-      "loss": 5.8933,
-      "step": 8259
-    },
-    {
-      "epoch": 4.3076923076923075,
-      "grad_norm": 1.8922733068466187,
-      "learning_rate": 9.220301507537689e-05,
-      "loss": 5.9789,
-      "step": 8260
-    },
-    {
-      "epoch": 4.308213820078227,
-      "grad_norm": 1.3610353469848633,
-      "learning_rate": 9.220201005025126e-05,
-      "loss": 5.9777,
-      "step": 8261
-    },
-    {
-      "epoch": 4.308735332464146,
-      "grad_norm": 1.5024555921554565,
-      "learning_rate": 9.220100502512563e-05,
-      "loss": 5.9569,
-      "step": 8262
-    },
-    {
-      "epoch": 4.309256844850065,
-      "grad_norm": 1.4392033815383911,
-      "learning_rate": 9.22e-05,
-      "loss": 5.6902,
-      "step": 8263
-    },
-    {
-      "epoch": 4.3097783572359845,
-      "grad_norm": 1.544990062713623,
-      "learning_rate": 9.219899497487437e-05,
-      "loss": 5.5373,
-      "step": 8264
-    },
-    {
-      "epoch": 4.310299869621904,
-      "grad_norm": 1.5070922374725342,
-      "learning_rate": 9.219798994974875e-05,
-      "loss": 5.6636,
-      "step": 8265
-    },
-    {
-      "epoch": 4.3108213820078225,
-      "grad_norm": 1.4082627296447754,
-      "learning_rate": 9.219698492462312e-05,
-      "loss": 5.9458,
-      "step": 8266
-    },
-    {
-      "epoch": 4.311342894393742,
-      "grad_norm": 1.4262490272521973,
-      "learning_rate": 9.219597989949749e-05,
-      "loss": 5.9827,
-      "step": 8267
-    },
-    {
-      "epoch": 4.311864406779661,
-      "grad_norm": 1.4335191249847412,
-      "learning_rate": 9.219497487437187e-05,
-      "loss": 5.4422,
-      "step": 8268
-    },
-    {
-      "epoch": 4.31238591916558,
-      "grad_norm": 1.3269152641296387,
-      "learning_rate": 9.219396984924623e-05,
-      "loss": 5.6946,
-      "step": 8269
-    },
-    {
-      "epoch": 4.3129074315514995,
-      "grad_norm": 1.596197485923767,
-      "learning_rate": 9.219296482412061e-05,
-      "loss": 5.2704,
-      "step": 8270
-    },
-    {
-      "epoch": 4.313428943937419,
-      "grad_norm": 1.4878579378128052,
-      "learning_rate": 9.219195979899497e-05,
-      "loss": 5.6245,
-      "step": 8271
-    },
-    {
-      "epoch": 4.3139504563233375,
-      "grad_norm": 1.4377573728561401,
-      "learning_rate": 9.219095477386935e-05,
-      "loss": 5.5765,
-      "step": 8272
-    },
-    {
-      "epoch": 4.314471968709257,
-      "grad_norm": 1.7823678255081177,
-      "learning_rate": 9.218994974874372e-05,
-      "loss": 4.9428,
-      "step": 8273
-    },
-    {
-      "epoch": 4.3149934810951756,
-      "grad_norm": 1.48689603805542,
-      "learning_rate": 9.21889447236181e-05,
-      "loss": 5.6034,
-      "step": 8274
-    },
-    {
-      "epoch": 4.315514993481095,
-      "grad_norm": 1.5591906309127808,
-      "learning_rate": 9.218793969849247e-05,
-      "loss": 5.7031,
-      "step": 8275
-    },
-    {
-      "epoch": 4.3160365058670145,
-      "grad_norm": 1.5318689346313477,
-      "learning_rate": 9.218693467336685e-05,
-      "loss": 5.032,
-      "step": 8276
-    },
-    {
-      "epoch": 4.316558018252934,
-      "grad_norm": 1.5653507709503174,
-      "learning_rate": 9.218592964824121e-05,
-      "loss": 5.8399,
-      "step": 8277
-    },
-    {
-      "epoch": 4.3170795306388525,
-      "grad_norm": 1.4897016286849976,
-      "learning_rate": 9.218492462311559e-05,
-      "loss": 5.8772,
-      "step": 8278
-    },
-    {
-      "epoch": 4.317601043024772,
-      "grad_norm": 1.5363280773162842,
-      "learning_rate": 9.218391959798996e-05,
-      "loss": 5.0585,
-      "step": 8279
-    },
-    {
-      "epoch": 4.318122555410691,
-      "grad_norm": 1.4643422365188599,
-      "learning_rate": 9.218291457286432e-05,
-      "loss": 5.839,
-      "step": 8280
-    },
-    {
-      "epoch": 4.31864406779661,
-      "grad_norm": 1.350993037223816,
-      "learning_rate": 9.21819095477387e-05,
-      "loss": 5.4726,
-      "step": 8281
-    },
-    {
-      "epoch": 4.3191655801825295,
-      "grad_norm": 1.6100324392318726,
-      "learning_rate": 9.218090452261306e-05,
-      "loss": 5.5356,
-      "step": 8282
-    },
-    {
-      "epoch": 4.319687092568449,
-      "grad_norm": 1.4671727418899536,
-      "learning_rate": 9.217989949748744e-05,
-      "loss": 5.4441,
-      "step": 8283
-    },
-    {
-      "epoch": 4.3202086049543675,
-      "grad_norm": 1.7861403226852417,
-      "learning_rate": 9.21788944723618e-05,
-      "loss": 5.6657,
-      "step": 8284
-    },
-    {
-      "epoch": 4.320730117340287,
-      "grad_norm": 1.4524741172790527,
-      "learning_rate": 9.217788944723618e-05,
-      "loss": 5.9449,
-      "step": 8285
-    },
-    {
-      "epoch": 4.321251629726206,
-      "grad_norm": 1.5441502332687378,
-      "learning_rate": 9.217688442211056e-05,
-      "loss": 5.8028,
-      "step": 8286
-    },
-    {
-      "epoch": 4.321773142112125,
-      "grad_norm": 1.441234827041626,
-      "learning_rate": 9.217587939698494e-05,
-      "loss": 5.4962,
-      "step": 8287
-    },
-    {
-      "epoch": 4.3222946544980445,
-      "grad_norm": 1.4963382482528687,
-      "learning_rate": 9.21748743718593e-05,
-      "loss": 5.8134,
-      "step": 8288
-    },
-    {
-      "epoch": 4.322816166883963,
-      "grad_norm": 1.4531426429748535,
-      "learning_rate": 9.217386934673368e-05,
-      "loss": 5.5566,
-      "step": 8289
-    },
-    {
-      "epoch": 4.3233376792698825,
-      "grad_norm": 1.3265399932861328,
-      "learning_rate": 9.217286432160804e-05,
-      "loss": 5.8595,
-      "step": 8290
-    },
-    {
-      "epoch": 4.323859191655802,
-      "grad_norm": 1.561715841293335,
-      "learning_rate": 9.217185929648242e-05,
-      "loss": 5.337,
-      "step": 8291
-    },
-    {
-      "epoch": 4.324380704041721,
-      "grad_norm": 1.500207543373108,
-      "learning_rate": 9.217085427135679e-05,
-      "loss": 5.9507,
-      "step": 8292
-    },
-    {
-      "epoch": 4.32490221642764,
-      "grad_norm": 1.454492449760437,
-      "learning_rate": 9.216984924623115e-05,
-      "loss": 5.8808,
-      "step": 8293
-    },
-    {
-      "epoch": 4.3254237288135595,
-      "grad_norm": 1.433907389640808,
-      "learning_rate": 9.216884422110553e-05,
-      "loss": 5.4905,
-      "step": 8294
-    },
-    {
-      "epoch": 4.325945241199478,
-      "grad_norm": 1.5488888025283813,
-      "learning_rate": 9.21678391959799e-05,
-      "loss": 5.8872,
-      "step": 8295
-    },
-    {
-      "epoch": 4.326466753585398,
-      "grad_norm": 1.6049820184707642,
-      "learning_rate": 9.216683417085428e-05,
-      "loss": 5.4669,
-      "step": 8296
-    },
-    {
-      "epoch": 4.326988265971317,
-      "grad_norm": 1.405737042427063,
-      "learning_rate": 9.216582914572865e-05,
-      "loss": 5.8427,
-      "step": 8297
-    },
-    {
-      "epoch": 4.327509778357236,
-      "grad_norm": 1.5014399290084839,
-      "learning_rate": 9.216482412060303e-05,
-      "loss": 5.7054,
-      "step": 8298
-    },
-    {
-      "epoch": 4.328031290743155,
-      "grad_norm": 1.642096996307373,
-      "learning_rate": 9.216381909547739e-05,
-      "loss": 5.5729,
-      "step": 8299
-    },
-    {
-      "epoch": 4.3285528031290745,
-      "grad_norm": 1.4084992408752441,
-      "learning_rate": 9.216281407035177e-05,
-      "loss": 5.9822,
-      "step": 8300
-    },
-    {
-      "epoch": 4.329074315514993,
-      "grad_norm": 1.5618395805358887,
-      "learning_rate": 9.216180904522613e-05,
-      "loss": 5.9319,
-      "step": 8301
-    },
-    {
-      "epoch": 4.329595827900913,
-      "grad_norm": 1.4172029495239258,
-      "learning_rate": 9.216080402010051e-05,
-      "loss": 5.7945,
-      "step": 8302
-    },
-    {
-      "epoch": 4.330117340286832,
-      "grad_norm": 1.3057911396026611,
-      "learning_rate": 9.215979899497487e-05,
-      "loss": 6.0817,
-      "step": 8303
-    },
-    {
-      "epoch": 4.330638852672751,
-      "grad_norm": 1.4400438070297241,
-      "learning_rate": 9.215879396984925e-05,
-      "loss": 5.963,
-      "step": 8304
-    },
-    {
-      "epoch": 4.33116036505867,
-      "grad_norm": 1.3010063171386719,
-      "learning_rate": 9.215778894472362e-05,
-      "loss": 5.8811,
-      "step": 8305
-    },
-    {
-      "epoch": 4.3316818774445895,
-      "grad_norm": 1.3969155550003052,
-      "learning_rate": 9.2156783919598e-05,
-      "loss": 5.9768,
-      "step": 8306
-    },
-    {
-      "epoch": 4.332203389830508,
-      "grad_norm": 1.3720359802246094,
-      "learning_rate": 9.215577889447237e-05,
-      "loss": 5.6884,
-      "step": 8307
-    },
-    {
-      "epoch": 4.332724902216428,
-      "grad_norm": 1.526016116142273,
-      "learning_rate": 9.215477386934674e-05,
-      "loss": 5.4623,
-      "step": 8308
-    },
-    {
-      "epoch": 4.333246414602347,
-      "grad_norm": 1.5156179666519165,
-      "learning_rate": 9.215376884422111e-05,
-      "loss": 5.4928,
-      "step": 8309
-    },
-    {
-      "epoch": 4.333767926988266,
-      "grad_norm": 1.4007017612457275,
-      "learning_rate": 9.215276381909548e-05,
-      "loss": 5.6169,
-      "step": 8310
-    },
-    {
-      "epoch": 4.334289439374185,
-      "grad_norm": 1.4022386074066162,
-      "learning_rate": 9.215175879396986e-05,
-      "loss": 5.9981,
-      "step": 8311
-    },
-    {
-      "epoch": 4.334810951760105,
-      "grad_norm": 1.5535482168197632,
-      "learning_rate": 9.215075376884422e-05,
-      "loss": 5.5519,
-      "step": 8312
-    },
-    {
-      "epoch": 4.335332464146023,
-      "grad_norm": 1.3952038288116455,
-      "learning_rate": 9.21497487437186e-05,
-      "loss": 5.9426,
-      "step": 8313
-    },
-    {
-      "epoch": 4.335853976531943,
-      "grad_norm": 1.3793989419937134,
-      "learning_rate": 9.214874371859296e-05,
-      "loss": 5.999,
-      "step": 8314
-    },
-    {
-      "epoch": 4.336375488917862,
-      "grad_norm": 1.4024592638015747,
-      "learning_rate": 9.214773869346734e-05,
-      "loss": 6.0287,
-      "step": 8315
-    },
-    {
-      "epoch": 4.336897001303781,
-      "grad_norm": 1.5889145135879517,
-      "learning_rate": 9.214673366834172e-05,
-      "loss": 5.5329,
-      "step": 8316
-    },
-    {
-      "epoch": 4.3374185136897,
-      "grad_norm": 1.406420350074768,
-      "learning_rate": 9.21457286432161e-05,
-      "loss": 5.6124,
-      "step": 8317
-    },
-    {
-      "epoch": 4.33794002607562,
-      "grad_norm": 1.611305832862854,
-      "learning_rate": 9.214472361809046e-05,
-      "loss": 5.2362,
-      "step": 8318
-    },
-    {
-      "epoch": 4.338461538461538,
-      "grad_norm": 1.5082974433898926,
-      "learning_rate": 9.214371859296484e-05,
-      "loss": 5.7524,
-      "step": 8319
-    },
-    {
-      "epoch": 4.338983050847458,
-      "grad_norm": 1.4624544382095337,
-      "learning_rate": 9.21427135678392e-05,
-      "loss": 5.2791,
-      "step": 8320
-    },
-    {
-      "epoch": 4.339504563233377,
-      "grad_norm": 1.4520232677459717,
-      "learning_rate": 9.214170854271357e-05,
-      "loss": 5.7896,
-      "step": 8321
-    },
-    {
-      "epoch": 4.340026075619296,
-      "grad_norm": 1.5389471054077148,
-      "learning_rate": 9.214070351758794e-05,
-      "loss": 5.5861,
-      "step": 8322
-    },
-    {
-      "epoch": 4.340547588005215,
-      "grad_norm": 1.5276226997375488,
-      "learning_rate": 9.213969849246231e-05,
-      "loss": 5.0941,
-      "step": 8323
-    },
-    {
-      "epoch": 4.341069100391135,
-      "grad_norm": 1.5981365442276,
-      "learning_rate": 9.213869346733669e-05,
-      "loss": 5.8451,
-      "step": 8324
-    },
-    {
-      "epoch": 4.341590612777053,
-      "grad_norm": 1.3374897241592407,
-      "learning_rate": 9.213768844221105e-05,
-      "loss": 6.0184,
-      "step": 8325
-    },
-    {
-      "epoch": 4.342112125162973,
-      "grad_norm": 1.3429638147354126,
-      "learning_rate": 9.213668341708543e-05,
-      "loss": 5.8044,
-      "step": 8326
-    },
-    {
-      "epoch": 4.342633637548892,
-      "grad_norm": 1.4750995635986328,
-      "learning_rate": 9.21356783919598e-05,
-      "loss": 5.4614,
-      "step": 8327
-    },
-    {
-      "epoch": 4.343155149934811,
-      "grad_norm": 1.5890671014785767,
-      "learning_rate": 9.213467336683418e-05,
-      "loss": 5.8785,
-      "step": 8328
-    },
-    {
-      "epoch": 4.34367666232073,
-      "grad_norm": 1.6134897470474243,
-      "learning_rate": 9.213366834170855e-05,
-      "loss": 5.6074,
-      "step": 8329
-    },
-    {
-      "epoch": 4.34419817470665,
-      "grad_norm": 1.398579716682434,
-      "learning_rate": 9.213266331658293e-05,
-      "loss": 5.3171,
-      "step": 8330
-    },
-    {
-      "epoch": 4.344719687092568,
-      "grad_norm": 1.5976992845535278,
-      "learning_rate": 9.213165829145729e-05,
-      "loss": 5.8038,
-      "step": 8331
-    },
-    {
-      "epoch": 4.345241199478488,
-      "grad_norm": 1.424699068069458,
-      "learning_rate": 9.213065326633167e-05,
-      "loss": 5.6848,
-      "step": 8332
-    },
-    {
-      "epoch": 4.345762711864407,
-      "grad_norm": 1.4674835205078125,
-      "learning_rate": 9.212964824120603e-05,
-      "loss": 5.7585,
-      "step": 8333
-    },
-    {
-      "epoch": 4.346284224250326,
-      "grad_norm": 1.492355227470398,
-      "learning_rate": 9.21286432160804e-05,
-      "loss": 5.9154,
-      "step": 8334
-    },
-    {
-      "epoch": 4.346805736636245,
-      "grad_norm": 1.5886064767837524,
-      "learning_rate": 9.212763819095477e-05,
-      "loss": 5.5573,
-      "step": 8335
-    },
-    {
-      "epoch": 4.347327249022165,
-      "grad_norm": 1.3893243074417114,
-      "learning_rate": 9.212663316582915e-05,
-      "loss": 5.8334,
-      "step": 8336
-    },
-    {
-      "epoch": 4.347848761408083,
-      "grad_norm": 1.530548095703125,
-      "learning_rate": 9.212562814070353e-05,
-      "loss": 5.563,
-      "step": 8337
-    },
-    {
-      "epoch": 4.348370273794003,
-      "grad_norm": 1.5304397344589233,
-      "learning_rate": 9.21246231155779e-05,
-      "loss": 5.6397,
-      "step": 8338
-    },
-    {
-      "epoch": 4.348891786179922,
-      "grad_norm": 1.4951177835464478,
-      "learning_rate": 9.212361809045227e-05,
-      "loss": 5.7045,
-      "step": 8339
-    },
-    {
-      "epoch": 4.349413298565841,
-      "grad_norm": 1.5254300832748413,
-      "learning_rate": 9.212261306532664e-05,
-      "loss": 5.2991,
-      "step": 8340
-    },
-    {
-      "epoch": 4.34993481095176,
-      "grad_norm": 1.5525823831558228,
-      "learning_rate": 9.212160804020101e-05,
-      "loss": 5.7492,
-      "step": 8341
-    },
-    {
-      "epoch": 4.35045632333768,
-      "grad_norm": 1.5241031646728516,
-      "learning_rate": 9.212060301507538e-05,
-      "loss": 5.3788,
-      "step": 8342
-    },
-    {
-      "epoch": 4.350977835723598,
-      "grad_norm": 1.4791374206542969,
-      "learning_rate": 9.211959798994976e-05,
-      "loss": 5.7003,
-      "step": 8343
-    },
-    {
-      "epoch": 4.351499348109518,
-      "grad_norm": 1.3853644132614136,
-      "learning_rate": 9.211859296482412e-05,
-      "loss": 5.5897,
-      "step": 8344
-    },
-    {
-      "epoch": 4.352020860495437,
-      "grad_norm": 1.4254686832427979,
-      "learning_rate": 9.21175879396985e-05,
-      "loss": 5.8085,
-      "step": 8345
-    },
-    {
-      "epoch": 4.352542372881356,
-      "grad_norm": 1.519871711730957,
-      "learning_rate": 9.211658291457286e-05,
-      "loss": 5.6836,
-      "step": 8346
-    },
-    {
-      "epoch": 4.353063885267275,
-      "grad_norm": 1.4386110305786133,
-      "learning_rate": 9.211557788944724e-05,
-      "loss": 5.8079,
-      "step": 8347
-    },
-    {
-      "epoch": 4.353585397653195,
-      "grad_norm": 1.5834639072418213,
-      "learning_rate": 9.211457286432162e-05,
-      "loss": 5.5403,
-      "step": 8348
-    },
-    {
-      "epoch": 4.354106910039113,
-      "grad_norm": 1.3099068403244019,
-      "learning_rate": 9.211356783919598e-05,
-      "loss": 6.0781,
-      "step": 8349
-    },
-    {
-      "epoch": 4.354628422425033,
-      "grad_norm": 1.4098165035247803,
-      "learning_rate": 9.211256281407036e-05,
-      "loss": 5.5207,
-      "step": 8350
-    },
-    {
-      "epoch": 4.355149934810952,
-      "grad_norm": 1.5285649299621582,
-      "learning_rate": 9.211155778894473e-05,
-      "loss": 5.6153,
-      "step": 8351
-    },
-    {
-      "epoch": 4.355671447196871,
-      "grad_norm": 1.410719394683838,
-      "learning_rate": 9.21105527638191e-05,
-      "loss": 5.8071,
-      "step": 8352
-    },
-    {
-      "epoch": 4.35619295958279,
-      "grad_norm": 1.394924521446228,
-      "learning_rate": 9.210954773869347e-05,
-      "loss": 5.8622,
-      "step": 8353
-    },
-    {
-      "epoch": 4.35671447196871,
-      "grad_norm": 1.5606770515441895,
-      "learning_rate": 9.210854271356785e-05,
-      "loss": 5.902,
-      "step": 8354
-    },
-    {
-      "epoch": 4.357235984354628,
-      "grad_norm": 1.6630443334579468,
-      "learning_rate": 9.210753768844221e-05,
-      "loss": 5.5758,
-      "step": 8355
-    },
-    {
-      "epoch": 4.357757496740548,
-      "grad_norm": 1.7421988248825073,
-      "learning_rate": 9.210653266331659e-05,
-      "loss": 5.5141,
-      "step": 8356
-    },
-    {
-      "epoch": 4.358279009126467,
-      "grad_norm": 1.5668036937713623,
-      "learning_rate": 9.210552763819095e-05,
-      "loss": 5.7357,
-      "step": 8357
-    },
-    {
-      "epoch": 4.358800521512386,
-      "grad_norm": 1.4011017084121704,
-      "learning_rate": 9.210452261306533e-05,
-      "loss": 5.433,
-      "step": 8358
-    },
-    {
-      "epoch": 4.359322033898305,
-      "grad_norm": 1.5381667613983154,
-      "learning_rate": 9.210351758793971e-05,
-      "loss": 5.9227,
-      "step": 8359
-    },
-    {
-      "epoch": 4.359843546284225,
-      "grad_norm": 1.486783742904663,
-      "learning_rate": 9.210251256281407e-05,
-      "loss": 5.8329,
-      "step": 8360
-    },
-    {
-      "epoch": 4.360365058670143,
-      "grad_norm": 1.5890350341796875,
-      "learning_rate": 9.210150753768845e-05,
-      "loss": 5.393,
-      "step": 8361
-    },
-    {
-      "epoch": 4.360886571056063,
-      "grad_norm": 1.354018211364746,
-      "learning_rate": 9.210050251256281e-05,
-      "loss": 5.428,
-      "step": 8362
-    },
-    {
-      "epoch": 4.361408083441981,
-      "grad_norm": 1.4974853992462158,
-      "learning_rate": 9.209949748743719e-05,
-      "loss": 5.8411,
-      "step": 8363
-    },
-    {
-      "epoch": 4.361929595827901,
-      "grad_norm": 1.6185853481292725,
-      "learning_rate": 9.209849246231156e-05,
-      "loss": 4.7514,
-      "step": 8364
-    },
-    {
-      "epoch": 4.36245110821382,
-      "grad_norm": 1.6668237447738647,
-      "learning_rate": 9.209748743718593e-05,
-      "loss": 5.3646,
-      "step": 8365
-    },
-    {
-      "epoch": 4.36297262059974,
-      "grad_norm": 1.7495858669281006,
-      "learning_rate": 9.20964824120603e-05,
-      "loss": 5.6432,
-      "step": 8366
-    },
-    {
-      "epoch": 4.363494132985658,
-      "grad_norm": 1.4996405839920044,
-      "learning_rate": 9.209547738693468e-05,
-      "loss": 5.3552,
-      "step": 8367
-    },
-    {
-      "epoch": 4.364015645371578,
-      "grad_norm": 1.5243061780929565,
-      "learning_rate": 9.209447236180905e-05,
-      "loss": 5.5889,
-      "step": 8368
-    },
-    {
-      "epoch": 4.364537157757496,
-      "grad_norm": 1.3953418731689453,
-      "learning_rate": 9.209346733668343e-05,
-      "loss": 5.8574,
-      "step": 8369
-    },
-    {
-      "epoch": 4.365058670143416,
-      "grad_norm": 1.4753308296203613,
-      "learning_rate": 9.20924623115578e-05,
-      "loss": 5.7372,
-      "step": 8370
-    },
-    {
-      "epoch": 4.365580182529335,
-      "grad_norm": 1.351660966873169,
-      "learning_rate": 9.209145728643217e-05,
-      "loss": 4.9903,
-      "step": 8371
-    },
-    {
-      "epoch": 4.366101694915255,
-      "grad_norm": 1.3709269762039185,
-      "learning_rate": 9.209045226130654e-05,
-      "loss": 5.9865,
-      "step": 8372
-    },
-    {
-      "epoch": 4.366623207301173,
-      "grad_norm": 1.3364819288253784,
-      "learning_rate": 9.20894472361809e-05,
-      "loss": 5.7665,
-      "step": 8373
-    },
-    {
-      "epoch": 4.367144719687093,
-      "grad_norm": 1.4243733882904053,
-      "learning_rate": 9.208844221105528e-05,
-      "loss": 5.3551,
-      "step": 8374
-    },
-    {
-      "epoch": 4.367666232073011,
-      "grad_norm": 1.391890525817871,
-      "learning_rate": 9.208743718592964e-05,
-      "loss": 6.086,
-      "step": 8375
-    },
-    {
-      "epoch": 4.368187744458931,
-      "grad_norm": 1.4085190296173096,
-      "learning_rate": 9.208643216080402e-05,
-      "loss": 5.7305,
-      "step": 8376
-    },
-    {
-      "epoch": 4.36870925684485,
-      "grad_norm": 1.4323530197143555,
-      "learning_rate": 9.208542713567839e-05,
-      "loss": 5.4991,
-      "step": 8377
-    },
-    {
-      "epoch": 4.36923076923077,
-      "grad_norm": 1.4663281440734863,
-      "learning_rate": 9.208442211055276e-05,
-      "loss": 5.7581,
-      "step": 8378
-    },
-    {
-      "epoch": 4.369752281616688,
-      "grad_norm": 1.542172908782959,
-      "learning_rate": 9.208341708542714e-05,
-      "loss": 5.6001,
-      "step": 8379
-    },
-    {
-      "epoch": 4.370273794002608,
-      "grad_norm": 1.4254447221755981,
-      "learning_rate": 9.208241206030152e-05,
-      "loss": 6.0444,
-      "step": 8380
-    },
-    {
-      "epoch": 4.370795306388526,
-      "grad_norm": 1.5768224000930786,
-      "learning_rate": 9.208140703517588e-05,
-      "loss": 5.5014,
-      "step": 8381
-    },
-    {
-      "epoch": 4.371316818774446,
-      "grad_norm": 1.4610263109207153,
-      "learning_rate": 9.208040201005026e-05,
-      "loss": 5.3423,
-      "step": 8382
-    },
-    {
-      "epoch": 4.371838331160365,
-      "grad_norm": 1.477096438407898,
-      "learning_rate": 9.207939698492463e-05,
-      "loss": 5.5269,
-      "step": 8383
-    },
-    {
-      "epoch": 4.372359843546284,
-      "grad_norm": 1.5048654079437256,
-      "learning_rate": 9.2078391959799e-05,
-      "loss": 5.8143,
-      "step": 8384
-    },
-    {
-      "epoch": 4.372881355932203,
-      "grad_norm": 1.406623125076294,
-      "learning_rate": 9.207738693467337e-05,
-      "loss": 5.9251,
-      "step": 8385
-    },
-    {
-      "epoch": 4.373402868318123,
-      "grad_norm": 1.4663423299789429,
-      "learning_rate": 9.207638190954773e-05,
-      "loss": 5.3457,
-      "step": 8386
-    },
-    {
-      "epoch": 4.373924380704041,
-      "grad_norm": 1.526286005973816,
-      "learning_rate": 9.207537688442211e-05,
-      "loss": 5.7158,
-      "step": 8387
-    },
-    {
-      "epoch": 4.374445893089961,
-      "grad_norm": 1.4195431470870972,
-      "learning_rate": 9.207437185929649e-05,
-      "loss": 5.9976,
-      "step": 8388
-    },
-    {
-      "epoch": 4.37496740547588,
-      "grad_norm": 1.8005800247192383,
-      "learning_rate": 9.207336683417087e-05,
-      "loss": 5.5613,
-      "step": 8389
-    },
-    {
-      "epoch": 4.375488917861799,
-      "grad_norm": 1.5153237581253052,
-      "learning_rate": 9.207236180904523e-05,
-      "loss": 5.5277,
-      "step": 8390
-    },
-    {
-      "epoch": 4.376010430247718,
-      "grad_norm": 1.590010166168213,
-      "learning_rate": 9.207135678391961e-05,
-      "loss": 5.2405,
-      "step": 8391
-    },
-    {
-      "epoch": 4.376531942633638,
-      "grad_norm": 1.463494062423706,
-      "learning_rate": 9.207035175879397e-05,
-      "loss": 5.9051,
-      "step": 8392
-    },
-    {
-      "epoch": 4.377053455019556,
-      "grad_norm": 1.7580204010009766,
-      "learning_rate": 9.206934673366835e-05,
-      "loss": 5.3555,
-      "step": 8393
-    },
-    {
-      "epoch": 4.377574967405476,
-      "grad_norm": 1.3823187351226807,
-      "learning_rate": 9.206834170854271e-05,
-      "loss": 5.576,
-      "step": 8394
-    },
-    {
-      "epoch": 4.378096479791395,
-      "grad_norm": 1.4613856077194214,
-      "learning_rate": 9.206733668341709e-05,
-      "loss": 5.8373,
-      "step": 8395
-    },
-    {
-      "epoch": 4.378617992177314,
-      "grad_norm": 1.7275230884552002,
-      "learning_rate": 9.206633165829146e-05,
-      "loss": 5.4033,
-      "step": 8396
-    },
-    {
-      "epoch": 4.379139504563233,
-      "grad_norm": 1.3858929872512817,
-      "learning_rate": 9.206532663316583e-05,
-      "loss": 5.7299,
-      "step": 8397
-    },
-    {
-      "epoch": 4.379661016949153,
-      "grad_norm": 1.3780758380889893,
-      "learning_rate": 9.20643216080402e-05,
-      "loss": 5.643,
-      "step": 8398
-    },
-    {
-      "epoch": 4.380182529335071,
-      "grad_norm": 1.374350666999817,
-      "learning_rate": 9.206331658291458e-05,
-      "loss": 5.8481,
-      "step": 8399
-    },
-    {
-      "epoch": 4.380704041720991,
-      "grad_norm": 1.5050560235977173,
-      "learning_rate": 9.206231155778895e-05,
-      "loss": 5.2694,
-      "step": 8400
-    },
-    {
-      "epoch": 4.38122555410691,
-      "grad_norm": 1.3673675060272217,
-      "learning_rate": 9.206130653266332e-05,
-      "loss": 4.9404,
-      "step": 8401
-    },
-    {
-      "epoch": 4.381747066492829,
-      "grad_norm": 1.3309860229492188,
-      "learning_rate": 9.20603015075377e-05,
-      "loss": 5.8157,
-      "step": 8402
-    },
-    {
-      "epoch": 4.382268578878748,
-      "grad_norm": 1.5175151824951172,
-      "learning_rate": 9.205929648241206e-05,
-      "loss": 5.4714,
-      "step": 8403
-    },
-    {
-      "epoch": 4.382790091264668,
-      "grad_norm": 1.5734177827835083,
-      "learning_rate": 9.205829145728644e-05,
-      "loss": 5.0133,
-      "step": 8404
-    },
-    {
-      "epoch": 4.383311603650586,
-      "grad_norm": 1.4373915195465088,
-      "learning_rate": 9.20572864321608e-05,
-      "loss": 5.6947,
-      "step": 8405
-    },
-    {
-      "epoch": 4.383833116036506,
-      "grad_norm": 1.4614757299423218,
-      "learning_rate": 9.205628140703518e-05,
-      "loss": 5.6787,
-      "step": 8406
-    },
-    {
-      "epoch": 4.384354628422425,
-      "grad_norm": 1.3880873918533325,
-      "learning_rate": 9.205527638190954e-05,
-      "loss": 5.4356,
-      "step": 8407
-    },
-    {
-      "epoch": 4.384876140808344,
-      "grad_norm": 1.4495278596878052,
-      "learning_rate": 9.205427135678392e-05,
-      "loss": 6.1845,
-      "step": 8408
-    },
-    {
-      "epoch": 4.385397653194263,
-      "grad_norm": 1.4518636465072632,
-      "learning_rate": 9.20532663316583e-05,
-      "loss": 5.6366,
-      "step": 8409
-    },
-    {
-      "epoch": 4.385919165580183,
-      "grad_norm": 1.5414338111877441,
-      "learning_rate": 9.205226130653268e-05,
-      "loss": 5.6159,
-      "step": 8410
-    },
-    {
-      "epoch": 4.386440677966101,
-      "grad_norm": 1.560157299041748,
-      "learning_rate": 9.205125628140704e-05,
-      "loss": 5.6532,
-      "step": 8411
-    },
-    {
-      "epoch": 4.386962190352021,
-      "grad_norm": 1.695234775543213,
-      "learning_rate": 9.205025125628142e-05,
-      "loss": 5.7587,
-      "step": 8412
-    },
-    {
-      "epoch": 4.38748370273794,
-      "grad_norm": 1.6355156898498535,
-      "learning_rate": 9.204924623115578e-05,
-      "loss": 5.7145,
-      "step": 8413
-    },
-    {
-      "epoch": 4.388005215123859,
-      "grad_norm": 1.4882831573486328,
-      "learning_rate": 9.204824120603015e-05,
-      "loss": 5.1573,
-      "step": 8414
-    },
-    {
-      "epoch": 4.388526727509778,
-      "grad_norm": 1.3983027935028076,
-      "learning_rate": 9.204723618090453e-05,
-      "loss": 5.6067,
-      "step": 8415
-    },
-    {
-      "epoch": 4.389048239895698,
-      "grad_norm": 1.521182656288147,
-      "learning_rate": 9.204623115577889e-05,
-      "loss": 5.9549,
-      "step": 8416
-    },
-    {
-      "epoch": 4.389569752281616,
-      "grad_norm": 1.403138518333435,
-      "learning_rate": 9.204522613065327e-05,
-      "loss": 5.6383,
-      "step": 8417
-    },
-    {
-      "epoch": 4.390091264667536,
-      "grad_norm": 1.5549063682556152,
-      "learning_rate": 9.204422110552763e-05,
-      "loss": 5.5345,
-      "step": 8418
-    },
-    {
-      "epoch": 4.390612777053455,
-      "grad_norm": 1.3787462711334229,
-      "learning_rate": 9.204321608040201e-05,
-      "loss": 5.7699,
-      "step": 8419
-    },
-    {
-      "epoch": 4.391134289439374,
-      "grad_norm": 2.512211799621582,
-      "learning_rate": 9.204221105527639e-05,
-      "loss": 4.9544,
-      "step": 8420
-    },
-    {
-      "epoch": 4.391655801825293,
-      "grad_norm": 1.5605902671813965,
-      "learning_rate": 9.204120603015077e-05,
-      "loss": 5.5965,
-      "step": 8421
-    },
-    {
-      "epoch": 4.392177314211213,
-      "grad_norm": 1.5183457136154175,
-      "learning_rate": 9.204020100502513e-05,
-      "loss": 5.7996,
-      "step": 8422
-    },
-    {
-      "epoch": 4.392698826597131,
-      "grad_norm": 1.361090064048767,
-      "learning_rate": 9.203919597989951e-05,
-      "loss": 6.0559,
-      "step": 8423
-    },
-    {
-      "epoch": 4.393220338983051,
-      "grad_norm": 1.4565188884735107,
-      "learning_rate": 9.203819095477387e-05,
-      "loss": 5.434,
-      "step": 8424
-    },
-    {
-      "epoch": 4.39374185136897,
-      "grad_norm": 1.4279438257217407,
-      "learning_rate": 9.203718592964825e-05,
-      "loss": 5.9133,
-      "step": 8425
-    },
-    {
-      "epoch": 4.394263363754889,
-      "grad_norm": 1.3095670938491821,
-      "learning_rate": 9.203618090452262e-05,
-      "loss": 6.0918,
-      "step": 8426
-    },
-    {
-      "epoch": 4.394784876140808,
-      "grad_norm": 1.5468206405639648,
-      "learning_rate": 9.203517587939698e-05,
-      "loss": 5.5588,
-      "step": 8427
-    },
-    {
-      "epoch": 4.395306388526728,
-      "grad_norm": 1.5258855819702148,
-      "learning_rate": 9.203417085427136e-05,
-      "loss": 5.4129,
-      "step": 8428
-    },
-    {
-      "epoch": 4.395827900912646,
-      "grad_norm": 1.4535568952560425,
-      "learning_rate": 9.203316582914574e-05,
-      "loss": 5.5887,
-      "step": 8429
-    },
-    {
-      "epoch": 4.396349413298566,
-      "grad_norm": 1.3691370487213135,
-      "learning_rate": 9.203216080402011e-05,
-      "loss": 5.8739,
-      "step": 8430
-    },
-    {
-      "epoch": 4.396870925684485,
-      "grad_norm": 1.5456101894378662,
-      "learning_rate": 9.203115577889448e-05,
-      "loss": 5.6534,
-      "step": 8431
-    },
-    {
-      "epoch": 4.397392438070404,
-      "grad_norm": 1.592221736907959,
-      "learning_rate": 9.203015075376885e-05,
-      "loss": 4.8166,
-      "step": 8432
-    },
-    {
-      "epoch": 4.397913950456323,
-      "grad_norm": 1.545013189315796,
-      "learning_rate": 9.202914572864322e-05,
-      "loss": 5.7012,
-      "step": 8433
-    },
-    {
-      "epoch": 4.398435462842243,
-      "grad_norm": 1.4341459274291992,
-      "learning_rate": 9.20281407035176e-05,
-      "loss": 5.5957,
-      "step": 8434
-    },
-    {
-      "epoch": 4.398956975228161,
-      "grad_norm": 1.4897041320800781,
-      "learning_rate": 9.202713567839196e-05,
-      "loss": 5.5212,
-      "step": 8435
-    },
-    {
-      "epoch": 4.399478487614081,
-      "grad_norm": 1.4104868173599243,
-      "learning_rate": 9.202613065326634e-05,
-      "loss": 5.3544,
-      "step": 8436
-    },
-    {
-      "epoch": 4.4,
-      "grad_norm": 1.4893500804901123,
-      "learning_rate": 9.20251256281407e-05,
-      "loss": 5.6922,
-      "step": 8437
-    },
-    {
-      "epoch": 4.400521512385919,
-      "grad_norm": 1.5361658334732056,
-      "learning_rate": 9.202412060301508e-05,
-      "loss": 5.4917,
-      "step": 8438
-    },
-    {
-      "epoch": 4.401043024771838,
-      "grad_norm": 1.4369691610336304,
-      "learning_rate": 9.202311557788945e-05,
-      "loss": 5.7396,
-      "step": 8439
-    },
-    {
-      "epoch": 4.401564537157758,
-      "grad_norm": 1.6548476219177246,
-      "learning_rate": 9.202211055276382e-05,
-      "loss": 5.4319,
-      "step": 8440
-    },
-    {
-      "epoch": 4.402086049543676,
-      "grad_norm": 1.457098126411438,
-      "learning_rate": 9.20211055276382e-05,
-      "loss": 5.8626,
-      "step": 8441
-    },
-    {
-      "epoch": 4.402607561929596,
-      "grad_norm": 1.4279136657714844,
-      "learning_rate": 9.202010050251257e-05,
-      "loss": 5.7792,
-      "step": 8442
-    },
-    {
-      "epoch": 4.403129074315515,
-      "grad_norm": 1.5216660499572754,
-      "learning_rate": 9.201909547738694e-05,
-      "loss": 5.8591,
-      "step": 8443
-    },
-    {
-      "epoch": 4.403650586701434,
-      "grad_norm": 1.7144232988357544,
-      "learning_rate": 9.201809045226131e-05,
-      "loss": 4.9994,
-      "step": 8444
-    },
-    {
-      "epoch": 4.404172099087353,
-      "grad_norm": 1.5926251411437988,
-      "learning_rate": 9.201708542713569e-05,
-      "loss": 5.5904,
-      "step": 8445
-    },
-    {
-      "epoch": 4.404693611473273,
-      "grad_norm": 1.6200439929962158,
-      "learning_rate": 9.201608040201005e-05,
-      "loss": 5.2469,
-      "step": 8446
-    },
-    {
-      "epoch": 4.4052151238591915,
-      "grad_norm": 1.4537465572357178,
-      "learning_rate": 9.201507537688443e-05,
-      "loss": 5.9686,
-      "step": 8447
-    },
-    {
-      "epoch": 4.405736636245111,
-      "grad_norm": 1.345384955406189,
-      "learning_rate": 9.201407035175879e-05,
-      "loss": 6.07,
-      "step": 8448
-    },
-    {
-      "epoch": 4.40625814863103,
-      "grad_norm": 1.4377410411834717,
-      "learning_rate": 9.201306532663317e-05,
-      "loss": 5.953,
-      "step": 8449
-    },
-    {
-      "epoch": 4.406779661016949,
-      "grad_norm": 1.4386775493621826,
-      "learning_rate": 9.201206030150755e-05,
-      "loss": 5.8901,
-      "step": 8450
-    },
-    {
-      "epoch": 4.407301173402868,
-      "grad_norm": 1.4346686601638794,
-      "learning_rate": 9.201105527638193e-05,
-      "loss": 6.0064,
-      "step": 8451
-    },
-    {
-      "epoch": 4.407822685788788,
-      "grad_norm": 1.4705733060836792,
-      "learning_rate": 9.201005025125629e-05,
-      "loss": 5.8569,
-      "step": 8452
-    },
-    {
-      "epoch": 4.4083441981747065,
-      "grad_norm": 1.4149116277694702,
-      "learning_rate": 9.200904522613065e-05,
-      "loss": 5.9548,
-      "step": 8453
-    },
-    {
-      "epoch": 4.408865710560626,
-      "grad_norm": 1.5103158950805664,
-      "learning_rate": 9.200804020100503e-05,
-      "loss": 5.9015,
-      "step": 8454
-    },
-    {
-      "epoch": 4.409387222946545,
-      "grad_norm": 1.4623289108276367,
-      "learning_rate": 9.20070351758794e-05,
-      "loss": 5.5835,
-      "step": 8455
-    },
-    {
-      "epoch": 4.409908735332464,
-      "grad_norm": 1.57591712474823,
-      "learning_rate": 9.200603015075377e-05,
-      "loss": 5.148,
-      "step": 8456
-    },
-    {
-      "epoch": 4.410430247718383,
-      "grad_norm": 1.4319919347763062,
-      "learning_rate": 9.200502512562814e-05,
-      "loss": 5.8617,
-      "step": 8457
-    },
-    {
-      "epoch": 4.410951760104302,
-      "grad_norm": 1.5666691064834595,
-      "learning_rate": 9.200402010050252e-05,
-      "loss": 4.8042,
-      "step": 8458
-    },
-    {
-      "epoch": 4.4114732724902215,
-      "grad_norm": 1.3124778270721436,
-      "learning_rate": 9.200301507537688e-05,
-      "loss": 5.9753,
-      "step": 8459
-    },
-    {
-      "epoch": 4.411994784876141,
-      "grad_norm": 1.4362339973449707,
-      "learning_rate": 9.200201005025126e-05,
-      "loss": 5.7913,
-      "step": 8460
-    },
-    {
-      "epoch": 4.41251629726206,
-      "grad_norm": 1.4630024433135986,
-      "learning_rate": 9.200100502512564e-05,
-      "loss": 5.9437,
-      "step": 8461
-    },
-    {
-      "epoch": 4.413037809647979,
-      "grad_norm": 1.4020715951919556,
-      "learning_rate": 9.200000000000001e-05,
-      "loss": 5.4026,
-      "step": 8462
-    },
-    {
-      "epoch": 4.4135593220338984,
-      "grad_norm": 1.4005388021469116,
-      "learning_rate": 9.199899497487438e-05,
-      "loss": 5.8785,
-      "step": 8463
-    },
-    {
-      "epoch": 4.414080834419817,
-      "grad_norm": 1.3915948867797852,
-      "learning_rate": 9.199798994974876e-05,
-      "loss": 6.0384,
-      "step": 8464
-    },
-    {
-      "epoch": 4.4146023468057365,
-      "grad_norm": 1.4855458736419678,
-      "learning_rate": 9.199698492462312e-05,
-      "loss": 5.8079,
-      "step": 8465
-    },
-    {
-      "epoch": 4.415123859191656,
-      "grad_norm": 1.4312002658843994,
-      "learning_rate": 9.199597989949748e-05,
-      "loss": 5.6462,
-      "step": 8466
-    },
-    {
-      "epoch": 4.415645371577575,
-      "grad_norm": 1.4861952066421509,
-      "learning_rate": 9.199497487437186e-05,
-      "loss": 5.4427,
-      "step": 8467
-    },
-    {
-      "epoch": 4.416166883963494,
-      "grad_norm": 1.4727098941802979,
-      "learning_rate": 9.199396984924623e-05,
-      "loss": 5.9071,
-      "step": 8468
-    },
-    {
-      "epoch": 4.4166883963494135,
-      "grad_norm": 1.5043220520019531,
-      "learning_rate": 9.19929648241206e-05,
-      "loss": 5.9624,
-      "step": 8469
-    },
-    {
-      "epoch": 4.417209908735332,
-      "grad_norm": 1.3935840129852295,
-      "learning_rate": 9.199195979899498e-05,
-      "loss": 5.869,
-      "step": 8470
-    },
-    {
-      "epoch": 4.4177314211212515,
-      "grad_norm": 1.4584107398986816,
-      "learning_rate": 9.199095477386936e-05,
-      "loss": 5.9459,
-      "step": 8471
-    },
-    {
-      "epoch": 4.418252933507171,
-      "grad_norm": 1.551141619682312,
-      "learning_rate": 9.198994974874372e-05,
-      "loss": 5.5913,
-      "step": 8472
-    },
-    {
-      "epoch": 4.41877444589309,
-      "grad_norm": 1.510519027709961,
-      "learning_rate": 9.19889447236181e-05,
-      "loss": 5.4905,
-      "step": 8473
-    },
-    {
-      "epoch": 4.419295958279009,
-      "grad_norm": 1.5230745077133179,
-      "learning_rate": 9.198793969849247e-05,
-      "loss": 4.9994,
-      "step": 8474
-    },
-    {
-      "epoch": 4.4198174706649285,
-      "grad_norm": 1.6600494384765625,
-      "learning_rate": 9.198693467336684e-05,
-      "loss": 5.5652,
-      "step": 8475
-    },
-    {
-      "epoch": 4.420338983050847,
-      "grad_norm": 1.615227460861206,
-      "learning_rate": 9.198592964824121e-05,
-      "loss": 5.6311,
-      "step": 8476
-    },
-    {
-      "epoch": 4.4208604954367665,
-      "grad_norm": 1.4867579936981201,
-      "learning_rate": 9.198492462311559e-05,
-      "loss": 5.8333,
-      "step": 8477
-    },
-    {
-      "epoch": 4.421382007822686,
-      "grad_norm": 1.6071964502334595,
-      "learning_rate": 9.198391959798995e-05,
-      "loss": 5.53,
-      "step": 8478
-    },
-    {
-      "epoch": 4.4219035202086046,
-      "grad_norm": 1.3992449045181274,
-      "learning_rate": 9.198291457286433e-05,
-      "loss": 5.9556,
-      "step": 8479
-    },
-    {
-      "epoch": 4.422425032594524,
-      "grad_norm": 1.4068150520324707,
-      "learning_rate": 9.198190954773869e-05,
-      "loss": 5.9768,
-      "step": 8480
-    },
-    {
-      "epoch": 4.4229465449804435,
-      "grad_norm": 1.6571662425994873,
-      "learning_rate": 9.198090452261307e-05,
-      "loss": 5.2103,
-      "step": 8481
-    },
-    {
-      "epoch": 4.423468057366362,
-      "grad_norm": 1.4443618059158325,
-      "learning_rate": 9.197989949748745e-05,
-      "loss": 5.6709,
-      "step": 8482
-    },
-    {
-      "epoch": 4.4239895697522815,
-      "grad_norm": 1.5400941371917725,
-      "learning_rate": 9.197889447236181e-05,
-      "loss": 5.781,
-      "step": 8483
-    },
-    {
-      "epoch": 4.424511082138201,
-      "grad_norm": 1.4461978673934937,
-      "learning_rate": 9.197788944723619e-05,
-      "loss": 5.9138,
-      "step": 8484
-    },
-    {
-      "epoch": 4.42503259452412,
-      "grad_norm": 1.4809643030166626,
-      "learning_rate": 9.197688442211055e-05,
-      "loss": 5.9468,
-      "step": 8485
-    },
-    {
-      "epoch": 4.425554106910039,
-      "grad_norm": 1.3932218551635742,
-      "learning_rate": 9.197587939698493e-05,
-      "loss": 6.0351,
-      "step": 8486
-    },
-    {
-      "epoch": 4.4260756192959585,
-      "grad_norm": 1.3488879203796387,
-      "learning_rate": 9.19748743718593e-05,
-      "loss": 5.4122,
-      "step": 8487
-    },
-    {
-      "epoch": 4.426597131681877,
-      "grad_norm": 1.7205002307891846,
-      "learning_rate": 9.197386934673367e-05,
-      "loss": 5.3307,
-      "step": 8488
-    },
-    {
-      "epoch": 4.4271186440677965,
-      "grad_norm": 1.49896240234375,
-      "learning_rate": 9.197286432160804e-05,
-      "loss": 5.7515,
-      "step": 8489
-    },
-    {
-      "epoch": 4.427640156453716,
-      "grad_norm": 1.4991713762283325,
-      "learning_rate": 9.197185929648242e-05,
-      "loss": 5.8296,
-      "step": 8490
-    },
-    {
-      "epoch": 4.428161668839635,
-      "grad_norm": 1.3671789169311523,
-      "learning_rate": 9.19708542713568e-05,
-      "loss": 5.7061,
-      "step": 8491
-    },
-    {
-      "epoch": 4.428683181225554,
-      "grad_norm": 1.4467412233352661,
-      "learning_rate": 9.196984924623117e-05,
-      "loss": 5.9015,
-      "step": 8492
-    },
-    {
-      "epoch": 4.4292046936114735,
-      "grad_norm": 1.3666800260543823,
-      "learning_rate": 9.196884422110554e-05,
-      "loss": 6.086,
-      "step": 8493
-    },
-    {
-      "epoch": 4.429726205997392,
-      "grad_norm": 1.439660906791687,
-      "learning_rate": 9.19678391959799e-05,
-      "loss": 5.7588,
-      "step": 8494
-    },
-    {
-      "epoch": 4.4302477183833116,
-      "grad_norm": 1.4936786890029907,
-      "learning_rate": 9.196683417085428e-05,
-      "loss": 5.8316,
-      "step": 8495
-    },
-    {
-      "epoch": 4.430769230769231,
-      "grad_norm": 1.4797965288162231,
-      "learning_rate": 9.196582914572864e-05,
-      "loss": 5.9401,
-      "step": 8496
-    },
-    {
-      "epoch": 4.43129074315515,
-      "grad_norm": 1.413234829902649,
-      "learning_rate": 9.196482412060302e-05,
-      "loss": 5.7898,
-      "step": 8497
-    },
-    {
-      "epoch": 4.431812255541069,
-      "grad_norm": 1.5038443803787231,
-      "learning_rate": 9.196381909547739e-05,
-      "loss": 5.3227,
-      "step": 8498
-    },
-    {
-      "epoch": 4.4323337679269885,
-      "grad_norm": 1.6399686336517334,
-      "learning_rate": 9.196281407035176e-05,
-      "loss": 5.254,
-      "step": 8499
-    },
-    {
-      "epoch": 4.432855280312907,
-      "grad_norm": 1.483466386795044,
-      "learning_rate": 9.196180904522613e-05,
-      "loss": 5.5794,
-      "step": 8500
-    },
-    {
-      "epoch": 4.433376792698827,
-      "grad_norm": 1.499801754951477,
-      "learning_rate": 9.19608040201005e-05,
-      "loss": 5.3308,
-      "step": 8501
-    },
-    {
-      "epoch": 4.433898305084746,
-      "grad_norm": 1.4119614362716675,
-      "learning_rate": 9.195979899497488e-05,
-      "loss": 5.641,
-      "step": 8502
-    },
-    {
-      "epoch": 4.434419817470665,
-      "grad_norm": 1.4367282390594482,
-      "learning_rate": 9.195879396984926e-05,
-      "loss": 5.9951,
-      "step": 8503
-    },
-    {
-      "epoch": 4.434941329856584,
-      "grad_norm": 1.3815664052963257,
-      "learning_rate": 9.195778894472362e-05,
-      "loss": 5.8525,
-      "step": 8504
-    },
-    {
-      "epoch": 4.4354628422425035,
-      "grad_norm": 1.451997995376587,
-      "learning_rate": 9.1956783919598e-05,
-      "loss": 5.5217,
-      "step": 8505
-    },
-    {
-      "epoch": 4.435984354628422,
-      "grad_norm": 1.4958455562591553,
-      "learning_rate": 9.195577889447237e-05,
-      "loss": 5.5782,
-      "step": 8506
-    },
-    {
-      "epoch": 4.436505867014342,
-      "grad_norm": 1.344444990158081,
-      "learning_rate": 9.195477386934673e-05,
-      "loss": 5.9074,
-      "step": 8507
-    },
-    {
-      "epoch": 4.437027379400261,
-      "grad_norm": 1.3787497282028198,
-      "learning_rate": 9.195376884422111e-05,
-      "loss": 6.2664,
-      "step": 8508
-    },
-    {
-      "epoch": 4.43754889178618,
-      "grad_norm": 1.5069832801818848,
-      "learning_rate": 9.195276381909547e-05,
-      "loss": 5.8114,
-      "step": 8509
-    },
-    {
-      "epoch": 4.438070404172099,
-      "grad_norm": 1.4131133556365967,
-      "learning_rate": 9.195175879396985e-05,
-      "loss": 6.0574,
-      "step": 8510
-    },
-    {
-      "epoch": 4.4385919165580185,
-      "grad_norm": 1.4292198419570923,
-      "learning_rate": 9.195075376884422e-05,
-      "loss": 5.8268,
-      "step": 8511
-    },
-    {
-      "epoch": 4.439113428943937,
-      "grad_norm": 1.4499297142028809,
-      "learning_rate": 9.19497487437186e-05,
-      "loss": 5.9782,
-      "step": 8512
-    },
-    {
-      "epoch": 4.439634941329857,
-      "grad_norm": 1.5884841680526733,
-      "learning_rate": 9.194874371859297e-05,
-      "loss": 5.8748,
-      "step": 8513
-    },
-    {
-      "epoch": 4.440156453715776,
-      "grad_norm": 1.465819001197815,
-      "learning_rate": 9.194773869346735e-05,
-      "loss": 5.8577,
-      "step": 8514
-    },
-    {
-      "epoch": 4.440677966101695,
-      "grad_norm": 1.4402958154678345,
-      "learning_rate": 9.194673366834171e-05,
-      "loss": 5.5536,
-      "step": 8515
-    },
-    {
-      "epoch": 4.441199478487614,
-      "grad_norm": 1.5382643938064575,
-      "learning_rate": 9.194572864321609e-05,
-      "loss": 6.157,
-      "step": 8516
-    },
-    {
-      "epoch": 4.441720990873534,
-      "grad_norm": 1.4057233333587646,
-      "learning_rate": 9.194472361809046e-05,
-      "loss": 5.8058,
-      "step": 8517
-    },
-    {
-      "epoch": 4.442242503259452,
-      "grad_norm": 1.4820387363433838,
-      "learning_rate": 9.194371859296483e-05,
-      "loss": 5.8308,
-      "step": 8518
-    },
-    {
-      "epoch": 4.442764015645372,
-      "grad_norm": 1.5606887340545654,
-      "learning_rate": 9.19427135678392e-05,
-      "loss": 5.6488,
-      "step": 8519
-    },
-    {
-      "epoch": 4.443285528031291,
-      "grad_norm": 1.631502389907837,
-      "learning_rate": 9.194170854271356e-05,
-      "loss": 5.4746,
-      "step": 8520
-    },
-    {
-      "epoch": 4.44380704041721,
-      "grad_norm": 1.4100278615951538,
-      "learning_rate": 9.194070351758794e-05,
-      "loss": 5.7006,
-      "step": 8521
-    },
-    {
-      "epoch": 4.444328552803129,
-      "grad_norm": 1.665969967842102,
-      "learning_rate": 9.193969849246232e-05,
-      "loss": 5.7621,
-      "step": 8522
-    },
-    {
-      "epoch": 4.444850065189049,
-      "grad_norm": 1.6897965669631958,
-      "learning_rate": 9.19386934673367e-05,
-      "loss": 4.9844,
-      "step": 8523
-    },
-    {
-      "epoch": 4.445371577574967,
-      "grad_norm": 1.3030211925506592,
-      "learning_rate": 9.193768844221106e-05,
-      "loss": 5.2708,
-      "step": 8524
-    },
-    {
-      "epoch": 4.445893089960887,
-      "grad_norm": 1.5535478591918945,
-      "learning_rate": 9.193668341708544e-05,
-      "loss": 5.5621,
-      "step": 8525
-    },
-    {
-      "epoch": 4.446414602346806,
-      "grad_norm": 1.4582056999206543,
-      "learning_rate": 9.19356783919598e-05,
-      "loss": 5.7835,
-      "step": 8526
-    },
-    {
-      "epoch": 4.446936114732725,
-      "grad_norm": 1.5168739557266235,
-      "learning_rate": 9.193467336683418e-05,
-      "loss": 5.5721,
-      "step": 8527
-    },
-    {
-      "epoch": 4.447457627118644,
-      "grad_norm": 1.4553214311599731,
-      "learning_rate": 9.193366834170854e-05,
-      "loss": 5.6738,
-      "step": 8528
-    },
-    {
-      "epoch": 4.447979139504564,
-      "grad_norm": 1.5080841779708862,
-      "learning_rate": 9.193266331658292e-05,
-      "loss": 6.1544,
-      "step": 8529
-    },
-    {
-      "epoch": 4.448500651890482,
-      "grad_norm": 1.4700599908828735,
-      "learning_rate": 9.193165829145729e-05,
-      "loss": 6.0179,
-      "step": 8530
-    },
-    {
-      "epoch": 4.449022164276402,
-      "grad_norm": 1.6179734468460083,
-      "learning_rate": 9.193065326633166e-05,
-      "loss": 5.7113,
-      "step": 8531
-    },
-    {
-      "epoch": 4.449543676662321,
-      "grad_norm": 1.3442050218582153,
-      "learning_rate": 9.192964824120603e-05,
-      "loss": 6.1887,
-      "step": 8532
-    },
-    {
-      "epoch": 4.45006518904824,
-      "grad_norm": 1.412664532661438,
-      "learning_rate": 9.19286432160804e-05,
-      "loss": 5.4351,
-      "step": 8533
-    },
-    {
-      "epoch": 4.450586701434159,
-      "grad_norm": 1.536219596862793,
-      "learning_rate": 9.192763819095478e-05,
-      "loss": 5.2847,
-      "step": 8534
-    },
-    {
-      "epoch": 4.451108213820079,
-      "grad_norm": 1.3760039806365967,
-      "learning_rate": 9.192663316582915e-05,
-      "loss": 5.6688,
-      "step": 8535
-    },
-    {
-      "epoch": 4.451629726205997,
-      "grad_norm": 1.4286408424377441,
-      "learning_rate": 9.192562814070353e-05,
-      "loss": 5.7638,
-      "step": 8536
-    },
-    {
-      "epoch": 4.452151238591917,
-      "grad_norm": 1.4032690525054932,
-      "learning_rate": 9.192462311557789e-05,
-      "loss": 6.0695,
-      "step": 8537
-    },
-    {
-      "epoch": 4.452672750977836,
-      "grad_norm": 1.4611767530441284,
-      "learning_rate": 9.192361809045227e-05,
-      "loss": 5.6933,
-      "step": 8538
-    },
-    {
-      "epoch": 4.453194263363755,
-      "grad_norm": 2.585259199142456,
-      "learning_rate": 9.192261306532663e-05,
-      "loss": 5.4461,
-      "step": 8539
-    },
-    {
-      "epoch": 4.453715775749674,
-      "grad_norm": 1.4198665618896484,
-      "learning_rate": 9.192160804020101e-05,
-      "loss": 5.8548,
-      "step": 8540
-    },
-    {
-      "epoch": 4.454237288135594,
-      "grad_norm": 1.4937433004379272,
-      "learning_rate": 9.192060301507537e-05,
-      "loss": 5.1945,
-      "step": 8541
-    },
-    {
-      "epoch": 4.454758800521512,
-      "grad_norm": 1.4448587894439697,
-      "learning_rate": 9.191959798994975e-05,
-      "loss": 5.9158,
-      "step": 8542
-    },
-    {
-      "epoch": 4.455280312907432,
-      "grad_norm": 1.4473978281021118,
-      "learning_rate": 9.191859296482413e-05,
-      "loss": 5.8868,
-      "step": 8543
-    },
-    {
-      "epoch": 4.455801825293351,
-      "grad_norm": 1.4440726041793823,
-      "learning_rate": 9.191758793969851e-05,
-      "loss": 5.8285,
-      "step": 8544
-    },
-    {
-      "epoch": 4.45632333767927,
-      "grad_norm": 1.518088459968567,
-      "learning_rate": 9.191658291457287e-05,
-      "loss": 5.5903,
-      "step": 8545
-    },
-    {
-      "epoch": 4.456844850065189,
-      "grad_norm": 1.5184831619262695,
-      "learning_rate": 9.191557788944724e-05,
-      "loss": 5.8502,
-      "step": 8546
-    },
-    {
-      "epoch": 4.457366362451109,
-      "grad_norm": 1.522686243057251,
-      "learning_rate": 9.191457286432161e-05,
-      "loss": 5.3082,
-      "step": 8547
-    },
-    {
-      "epoch": 4.457887874837027,
-      "grad_norm": 1.6005076169967651,
-      "learning_rate": 9.191356783919598e-05,
-      "loss": 5.9446,
-      "step": 8548
-    },
-    {
-      "epoch": 4.458409387222947,
-      "grad_norm": 1.3821306228637695,
-      "learning_rate": 9.191256281407036e-05,
-      "loss": 5.7349,
-      "step": 8549
-    },
-    {
-      "epoch": 4.458930899608866,
-      "grad_norm": 1.351000428199768,
-      "learning_rate": 9.191155778894472e-05,
-      "loss": 5.8158,
-      "step": 8550
-    },
-    {
-      "epoch": 4.459452411994785,
-      "grad_norm": 1.5363116264343262,
-      "learning_rate": 9.19105527638191e-05,
-      "loss": 5.8607,
-      "step": 8551
-    },
-    {
-      "epoch": 4.459973924380704,
-      "grad_norm": 1.4856208562850952,
-      "learning_rate": 9.190954773869346e-05,
-      "loss": 5.8489,
-      "step": 8552
-    },
-    {
-      "epoch": 4.460495436766623,
-      "grad_norm": 1.3818703889846802,
-      "learning_rate": 9.190854271356784e-05,
-      "loss": 5.886,
-      "step": 8553
-    },
-    {
-      "epoch": 4.461016949152542,
-      "grad_norm": 1.5135165452957153,
-      "learning_rate": 9.190753768844222e-05,
-      "loss": 5.3485,
-      "step": 8554
-    },
-    {
-      "epoch": 4.461538461538462,
-      "grad_norm": 1.5400102138519287,
-      "learning_rate": 9.19065326633166e-05,
-      "loss": 5.3456,
-      "step": 8555
-    },
-    {
-      "epoch": 4.462059973924381,
-      "grad_norm": 1.5678253173828125,
-      "learning_rate": 9.190552763819096e-05,
-      "loss": 5.5728,
-      "step": 8556
-    },
-    {
-      "epoch": 4.4625814863103,
-      "grad_norm": 1.3783069849014282,
-      "learning_rate": 9.190452261306534e-05,
-      "loss": 6.0736,
-      "step": 8557
-    },
-    {
-      "epoch": 4.463102998696219,
-      "grad_norm": 1.3344589471817017,
-      "learning_rate": 9.19035175879397e-05,
-      "loss": 5.6555,
-      "step": 8558
-    },
-    {
-      "epoch": 4.463624511082138,
-      "grad_norm": 1.4527411460876465,
-      "learning_rate": 9.190251256281407e-05,
-      "loss": 5.8316,
-      "step": 8559
-    },
-    {
-      "epoch": 4.464146023468057,
-      "grad_norm": 1.5172758102416992,
-      "learning_rate": 9.190150753768844e-05,
-      "loss": 5.5564,
-      "step": 8560
-    },
-    {
-      "epoch": 4.464667535853977,
-      "grad_norm": 1.414687156677246,
-      "learning_rate": 9.190050251256281e-05,
-      "loss": 5.6541,
-      "step": 8561
-    },
-    {
-      "epoch": 4.465189048239896,
-      "grad_norm": 1.5021504163742065,
-      "learning_rate": 9.189949748743719e-05,
-      "loss": 5.6541,
-      "step": 8562
-    },
-    {
-      "epoch": 4.465710560625815,
-      "grad_norm": 1.6022124290466309,
-      "learning_rate": 9.189849246231156e-05,
-      "loss": 5.3845,
-      "step": 8563
-    },
-    {
-      "epoch": 4.466232073011734,
-      "grad_norm": 1.5220342874526978,
-      "learning_rate": 9.189748743718594e-05,
-      "loss": 5.9164,
-      "step": 8564
-    },
-    {
-      "epoch": 4.466753585397653,
-      "grad_norm": 1.5480796098709106,
-      "learning_rate": 9.18964824120603e-05,
-      "loss": 5.1833,
-      "step": 8565
-    },
-    {
-      "epoch": 4.467275097783572,
-      "grad_norm": 1.496398687362671,
-      "learning_rate": 9.189547738693468e-05,
-      "loss": 5.5376,
-      "step": 8566
-    },
-    {
-      "epoch": 4.467796610169492,
-      "grad_norm": 1.3889068365097046,
-      "learning_rate": 9.189447236180905e-05,
-      "loss": 6.2016,
-      "step": 8567
-    },
-    {
-      "epoch": 4.46831812255541,
-      "grad_norm": 1.4930278062820435,
-      "learning_rate": 9.189346733668343e-05,
-      "loss": 5.5242,
-      "step": 8568
-    },
-    {
-      "epoch": 4.46883963494133,
-      "grad_norm": 1.5703706741333008,
-      "learning_rate": 9.189246231155779e-05,
-      "loss": 5.8332,
-      "step": 8569
-    },
-    {
-      "epoch": 4.469361147327249,
-      "grad_norm": 1.769647240638733,
-      "learning_rate": 9.189145728643217e-05,
-      "loss": 5.1563,
-      "step": 8570
-    },
-    {
-      "epoch": 4.469882659713168,
-      "grad_norm": 1.5287221670150757,
-      "learning_rate": 9.189045226130653e-05,
-      "loss": 5.2833,
-      "step": 8571
-    },
-    {
-      "epoch": 4.470404172099087,
-      "grad_norm": 1.3657301664352417,
-      "learning_rate": 9.188944723618091e-05,
-      "loss": 5.9854,
-      "step": 8572
-    },
-    {
-      "epoch": 4.470925684485007,
-      "grad_norm": 1.5355093479156494,
-      "learning_rate": 9.188844221105527e-05,
-      "loss": 5.4672,
-      "step": 8573
-    },
-    {
-      "epoch": 4.471447196870925,
-      "grad_norm": 1.5090543031692505,
-      "learning_rate": 9.188743718592965e-05,
-      "loss": 5.6899,
-      "step": 8574
-    },
-    {
-      "epoch": 4.471968709256845,
-      "grad_norm": 1.5218592882156372,
-      "learning_rate": 9.188643216080403e-05,
-      "loss": 5.829,
-      "step": 8575
-    },
-    {
-      "epoch": 4.472490221642764,
-      "grad_norm": 1.5560450553894043,
-      "learning_rate": 9.18854271356784e-05,
-      "loss": 5.7415,
-      "step": 8576
-    },
-    {
-      "epoch": 4.473011734028683,
-      "grad_norm": 1.5394688844680786,
-      "learning_rate": 9.188442211055277e-05,
-      "loss": 5.7078,
-      "step": 8577
-    },
-    {
-      "epoch": 4.473533246414602,
-      "grad_norm": 1.5446885824203491,
-      "learning_rate": 9.188341708542714e-05,
-      "loss": 5.477,
-      "step": 8578
-    },
-    {
-      "epoch": 4.474054758800522,
-      "grad_norm": 1.4566049575805664,
-      "learning_rate": 9.188241206030151e-05,
-      "loss": 5.9237,
-      "step": 8579
-    },
-    {
-      "epoch": 4.47457627118644,
-      "grad_norm": 1.3754338026046753,
-      "learning_rate": 9.188140703517588e-05,
-      "loss": 6.1446,
-      "step": 8580
-    },
-    {
-      "epoch": 4.47509778357236,
-      "grad_norm": 1.3557640314102173,
-      "learning_rate": 9.188040201005026e-05,
-      "loss": 5.6307,
-      "step": 8581
-    },
-    {
-      "epoch": 4.475619295958279,
-      "grad_norm": 1.559217095375061,
-      "learning_rate": 9.187939698492462e-05,
-      "loss": 5.8377,
-      "step": 8582
-    },
-    {
-      "epoch": 4.476140808344198,
-      "grad_norm": 1.4263664484024048,
-      "learning_rate": 9.1878391959799e-05,
-      "loss": 5.8633,
-      "step": 8583
-    },
-    {
-      "epoch": 4.476662320730117,
-      "grad_norm": 1.6580474376678467,
-      "learning_rate": 9.187738693467338e-05,
-      "loss": 5.8852,
-      "step": 8584
-    },
-    {
-      "epoch": 4.477183833116037,
-      "grad_norm": 1.4045087099075317,
-      "learning_rate": 9.187638190954775e-05,
-      "loss": 5.8492,
-      "step": 8585
-    },
-    {
-      "epoch": 4.477705345501955,
-      "grad_norm": 1.6569414138793945,
-      "learning_rate": 9.187537688442212e-05,
-      "loss": 5.9031,
-      "step": 8586
-    },
-    {
-      "epoch": 4.478226857887875,
-      "grad_norm": 1.3638617992401123,
-      "learning_rate": 9.187437185929648e-05,
-      "loss": 5.8312,
-      "step": 8587
-    },
-    {
-      "epoch": 4.478748370273794,
-      "grad_norm": 1.5572504997253418,
-      "learning_rate": 9.187336683417086e-05,
-      "loss": 5.9045,
-      "step": 8588
-    },
-    {
-      "epoch": 4.479269882659713,
-      "grad_norm": 1.5944182872772217,
-      "learning_rate": 9.187236180904523e-05,
-      "loss": 5.9593,
-      "step": 8589
-    },
-    {
-      "epoch": 4.479791395045632,
-      "grad_norm": 1.414777159690857,
-      "learning_rate": 9.18713567839196e-05,
-      "loss": 5.6462,
-      "step": 8590
-    },
-    {
-      "epoch": 4.480312907431552,
-      "grad_norm": 1.508028268814087,
-      "learning_rate": 9.187035175879397e-05,
-      "loss": 5.641,
-      "step": 8591
-    },
-    {
-      "epoch": 4.48083441981747,
-      "grad_norm": 1.5503356456756592,
-      "learning_rate": 9.186934673366835e-05,
-      "loss": 5.8132,
-      "step": 8592
-    },
-    {
-      "epoch": 4.48135593220339,
-      "grad_norm": 1.3871147632598877,
-      "learning_rate": 9.186834170854271e-05,
-      "loss": 6.0496,
-      "step": 8593
-    },
-    {
-      "epoch": 4.481877444589309,
-      "grad_norm": 1.483432412147522,
-      "learning_rate": 9.186733668341709e-05,
-      "loss": 5.8166,
-      "step": 8594
-    },
-    {
-      "epoch": 4.482398956975228,
-      "grad_norm": 1.35496187210083,
-      "learning_rate": 9.186633165829147e-05,
-      "loss": 5.6819,
-      "step": 8595
-    },
-    {
-      "epoch": 4.482920469361147,
-      "grad_norm": 1.3960988521575928,
-      "learning_rate": 9.186532663316584e-05,
-      "loss": 5.728,
-      "step": 8596
-    },
-    {
-      "epoch": 4.483441981747067,
-      "grad_norm": 1.4727298021316528,
-      "learning_rate": 9.186432160804021e-05,
-      "loss": 5.817,
-      "step": 8597
-    },
-    {
-      "epoch": 4.483963494132985,
-      "grad_norm": 1.4193438291549683,
-      "learning_rate": 9.186331658291459e-05,
-      "loss": 5.9039,
-      "step": 8598
-    },
-    {
-      "epoch": 4.484485006518905,
-      "grad_norm": 1.3565367460250854,
-      "learning_rate": 9.186231155778895e-05,
-      "loss": 5.7083,
-      "step": 8599
-    },
-    {
-      "epoch": 4.485006518904824,
-      "grad_norm": 1.557286262512207,
-      "learning_rate": 9.186130653266331e-05,
-      "loss": 5.1771,
-      "step": 8600
-    },
-    {
-      "epoch": 4.485528031290743,
-      "grad_norm": 1.5794085264205933,
-      "learning_rate": 9.186030150753769e-05,
-      "loss": 5.7519,
-      "step": 8601
-    },
-    {
-      "epoch": 4.486049543676662,
-      "grad_norm": 1.3786311149597168,
-      "learning_rate": 9.185929648241206e-05,
-      "loss": 5.8461,
-      "step": 8602
-    },
-    {
-      "epoch": 4.486571056062582,
-      "grad_norm": 1.415019154548645,
-      "learning_rate": 9.185829145728643e-05,
-      "loss": 5.9663,
-      "step": 8603
-    },
-    {
-      "epoch": 4.4870925684485,
-      "grad_norm": 1.3741525411605835,
-      "learning_rate": 9.185728643216081e-05,
-      "loss": 6.0055,
-      "step": 8604
-    },
-    {
-      "epoch": 4.48761408083442,
-      "grad_norm": 1.4780945777893066,
-      "learning_rate": 9.185628140703519e-05,
-      "loss": 5.9083,
-      "step": 8605
-    },
-    {
-      "epoch": 4.488135593220339,
-      "grad_norm": 1.440409541130066,
-      "learning_rate": 9.185527638190955e-05,
-      "loss": 5.7522,
-      "step": 8606
-    },
-    {
-      "epoch": 4.488657105606258,
-      "grad_norm": 1.4269981384277344,
-      "learning_rate": 9.185427135678393e-05,
-      "loss": 5.6507,
-      "step": 8607
-    },
-    {
-      "epoch": 4.489178617992177,
-      "grad_norm": 1.4798200130462646,
-      "learning_rate": 9.18532663316583e-05,
-      "loss": 5.9282,
-      "step": 8608
-    },
-    {
-      "epoch": 4.489700130378097,
-      "grad_norm": 1.5554476976394653,
-      "learning_rate": 9.185226130653267e-05,
-      "loss": 5.68,
-      "step": 8609
-    },
-    {
-      "epoch": 4.490221642764015,
-      "grad_norm": 1.3552969694137573,
-      "learning_rate": 9.185125628140704e-05,
-      "loss": 5.8993,
-      "step": 8610
-    },
-    {
-      "epoch": 4.490743155149935,
-      "grad_norm": 1.3467867374420166,
-      "learning_rate": 9.185025125628142e-05,
-      "loss": 5.9724,
-      "step": 8611
-    },
-    {
-      "epoch": 4.491264667535854,
-      "grad_norm": 1.4372122287750244,
-      "learning_rate": 9.184924623115578e-05,
-      "loss": 5.7207,
-      "step": 8612
-    },
-    {
-      "epoch": 4.491786179921773,
-      "grad_norm": 1.3132340908050537,
-      "learning_rate": 9.184824120603014e-05,
-      "loss": 5.6983,
-      "step": 8613
-    },
-    {
-      "epoch": 4.492307692307692,
-      "grad_norm": 1.4207745790481567,
-      "learning_rate": 9.184723618090452e-05,
-      "loss": 5.7059,
-      "step": 8614
-    },
-    {
-      "epoch": 4.492829204693612,
-      "grad_norm": 1.5269960165023804,
-      "learning_rate": 9.18462311557789e-05,
-      "loss": 5.6554,
-      "step": 8615
-    },
-    {
-      "epoch": 4.49335071707953,
-      "grad_norm": 1.3541020154953003,
-      "learning_rate": 9.184522613065328e-05,
-      "loss": 5.7282,
-      "step": 8616
-    },
-    {
-      "epoch": 4.49387222946545,
-      "grad_norm": 1.4565259218215942,
-      "learning_rate": 9.184422110552764e-05,
-      "loss": 5.4086,
-      "step": 8617
-    },
-    {
-      "epoch": 4.494393741851369,
-      "grad_norm": 1.5182610750198364,
-      "learning_rate": 9.184321608040202e-05,
-      "loss": 5.6801,
-      "step": 8618
-    },
-    {
-      "epoch": 4.494915254237288,
-      "grad_norm": 1.4594647884368896,
-      "learning_rate": 9.184221105527638e-05,
-      "loss": 5.8581,
-      "step": 8619
-    },
-    {
-      "epoch": 4.495436766623207,
-      "grad_norm": 1.360872745513916,
-      "learning_rate": 9.184120603015076e-05,
-      "loss": 5.8351,
-      "step": 8620
-    },
-    {
-      "epoch": 4.495958279009127,
-      "grad_norm": 1.3282561302185059,
-      "learning_rate": 9.184020100502513e-05,
-      "loss": 5.9004,
-      "step": 8621
-    },
-    {
-      "epoch": 4.496479791395045,
-      "grad_norm": 1.5173332691192627,
-      "learning_rate": 9.18391959798995e-05,
-      "loss": 5.5809,
-      "step": 8622
-    },
-    {
-      "epoch": 4.497001303780965,
-      "grad_norm": 1.4153659343719482,
-      "learning_rate": 9.183819095477387e-05,
-      "loss": 5.9949,
-      "step": 8623
-    },
-    {
-      "epoch": 4.497522816166884,
-      "grad_norm": 1.3907524347305298,
-      "learning_rate": 9.183718592964825e-05,
-      "loss": 5.7586,
-      "step": 8624
-    },
-    {
-      "epoch": 4.498044328552803,
-      "grad_norm": 1.4530448913574219,
-      "learning_rate": 9.183618090452262e-05,
-      "loss": 5.7852,
-      "step": 8625
-    },
-    {
-      "epoch": 4.498565840938722,
-      "grad_norm": 1.3848886489868164,
-      "learning_rate": 9.183517587939699e-05,
-      "loss": 6.0236,
-      "step": 8626
-    },
-    {
-      "epoch": 4.499087353324642,
-      "grad_norm": 1.8457201719284058,
-      "learning_rate": 9.183417085427137e-05,
-      "loss": 5.8028,
-      "step": 8627
-    },
-    {
-      "epoch": 4.49960886571056,
-      "grad_norm": 1.6940728425979614,
-      "learning_rate": 9.183316582914573e-05,
-      "loss": 5.3094,
-      "step": 8628
-    },
-    {
-      "epoch": 4.50013037809648,
-      "grad_norm": 1.5547945499420166,
-      "learning_rate": 9.183216080402011e-05,
-      "loss": 5.1087,
-      "step": 8629
-    },
-    {
-      "epoch": 4.500651890482399,
-      "grad_norm": 1.7104084491729736,
-      "learning_rate": 9.183115577889447e-05,
-      "loss": 5.1528,
-      "step": 8630
-    },
-    {
-      "epoch": 4.501173402868318,
-      "grad_norm": 1.597961187362671,
-      "learning_rate": 9.183015075376885e-05,
-      "loss": 5.5687,
-      "step": 8631
-    },
-    {
-      "epoch": 4.501694915254237,
-      "grad_norm": 1.573063611984253,
-      "learning_rate": 9.182914572864321e-05,
-      "loss": 5.5501,
-      "step": 8632
-    },
-    {
-      "epoch": 4.502216427640157,
-      "grad_norm": 1.6524958610534668,
-      "learning_rate": 9.182814070351759e-05,
-      "loss": 5.6431,
-      "step": 8633
-    },
-    {
-      "epoch": 4.502737940026075,
-      "grad_norm": 1.6049894094467163,
-      "learning_rate": 9.182713567839196e-05,
-      "loss": 5.3403,
-      "step": 8634
-    },
-    {
-      "epoch": 4.503259452411995,
-      "grad_norm": 1.5342134237289429,
-      "learning_rate": 9.182613065326633e-05,
-      "loss": 5.6237,
-      "step": 8635
-    },
-    {
-      "epoch": 4.5037809647979135,
-      "grad_norm": 1.4730374813079834,
-      "learning_rate": 9.182512562814071e-05,
-      "loss": 5.7087,
-      "step": 8636
-    },
-    {
-      "epoch": 4.504302477183833,
-      "grad_norm": 1.5002256631851196,
-      "learning_rate": 9.182412060301509e-05,
-      "loss": 5.7181,
-      "step": 8637
-    },
-    {
-      "epoch": 4.504823989569752,
-      "grad_norm": 1.4677200317382812,
-      "learning_rate": 9.182311557788945e-05,
-      "loss": 5.7041,
-      "step": 8638
-    },
-    {
-      "epoch": 4.505345501955672,
-      "grad_norm": 1.510989785194397,
-      "learning_rate": 9.182211055276382e-05,
-      "loss": 5.608,
-      "step": 8639
-    },
-    {
-      "epoch": 4.50586701434159,
-      "grad_norm": 1.3488264083862305,
-      "learning_rate": 9.18211055276382e-05,
-      "loss": 5.9633,
-      "step": 8640
-    },
-    {
-      "epoch": 4.50638852672751,
-      "grad_norm": 1.6377688646316528,
-      "learning_rate": 9.182010050251256e-05,
-      "loss": 5.2273,
-      "step": 8641
-    },
-    {
-      "epoch": 4.5069100391134285,
-      "grad_norm": 1.4688682556152344,
-      "learning_rate": 9.181909547738694e-05,
-      "loss": 5.703,
-      "step": 8642
-    },
-    {
-      "epoch": 4.507431551499348,
-      "grad_norm": 1.4690124988555908,
-      "learning_rate": 9.18180904522613e-05,
-      "loss": 5.7019,
-      "step": 8643
-    },
-    {
-      "epoch": 4.507953063885267,
-      "grad_norm": 1.4913052320480347,
-      "learning_rate": 9.181708542713568e-05,
-      "loss": 5.3814,
-      "step": 8644
-    },
-    {
-      "epoch": 4.508474576271187,
-      "grad_norm": 1.340436339378357,
-      "learning_rate": 9.181608040201006e-05,
-      "loss": 5.9409,
-      "step": 8645
-    },
-    {
-      "epoch": 4.508996088657105,
-      "grad_norm": 1.4594584703445435,
-      "learning_rate": 9.181507537688444e-05,
-      "loss": 5.4165,
-      "step": 8646
-    },
-    {
-      "epoch": 4.509517601043025,
-      "grad_norm": 1.4875481128692627,
-      "learning_rate": 9.18140703517588e-05,
-      "loss": 5.7921,
-      "step": 8647
-    },
-    {
-      "epoch": 4.5100391134289435,
-      "grad_norm": 1.6275416612625122,
-      "learning_rate": 9.181306532663318e-05,
-      "loss": 5.1119,
-      "step": 8648
-    },
-    {
-      "epoch": 4.510560625814863,
-      "grad_norm": 1.4357553720474243,
-      "learning_rate": 9.181206030150754e-05,
-      "loss": 5.7113,
-      "step": 8649
-    },
-    {
-      "epoch": 4.511082138200782,
-      "grad_norm": 1.4437183141708374,
-      "learning_rate": 9.181105527638192e-05,
-      "loss": 5.6713,
-      "step": 8650
-    },
-    {
-      "epoch": 4.511603650586702,
-      "grad_norm": 1.3711045980453491,
-      "learning_rate": 9.181005025125628e-05,
-      "loss": 5.7976,
-      "step": 8651
-    },
-    {
-      "epoch": 4.5121251629726205,
-      "grad_norm": 1.3479400873184204,
-      "learning_rate": 9.180904522613065e-05,
-      "loss": 6.1029,
-      "step": 8652
-    },
-    {
-      "epoch": 4.51264667535854,
-      "grad_norm": 1.584433913230896,
-      "learning_rate": 9.180804020100503e-05,
-      "loss": 5.6639,
-      "step": 8653
-    },
-    {
-      "epoch": 4.5131681877444585,
-      "grad_norm": 1.4037836790084839,
-      "learning_rate": 9.180703517587939e-05,
-      "loss": 5.6903,
-      "step": 8654
-    },
-    {
-      "epoch": 4.513689700130378,
-      "grad_norm": 1.3228816986083984,
-      "learning_rate": 9.180603015075377e-05,
-      "loss": 4.6655,
-      "step": 8655
-    },
-    {
-      "epoch": 4.514211212516297,
-      "grad_norm": 1.471509575843811,
-      "learning_rate": 9.180502512562815e-05,
-      "loss": 5.8747,
-      "step": 8656
-    },
-    {
-      "epoch": 4.514732724902217,
-      "grad_norm": 1.4782936573028564,
-      "learning_rate": 9.180402010050252e-05,
-      "loss": 5.5354,
-      "step": 8657
-    },
-    {
-      "epoch": 4.5152542372881355,
-      "grad_norm": 1.483424425125122,
-      "learning_rate": 9.180301507537689e-05,
-      "loss": 5.7327,
-      "step": 8658
-    },
-    {
-      "epoch": 4.515775749674055,
-      "grad_norm": 1.429533839225769,
-      "learning_rate": 9.180201005025127e-05,
-      "loss": 5.4311,
-      "step": 8659
-    },
-    {
-      "epoch": 4.5162972620599735,
-      "grad_norm": 1.3572803735733032,
-      "learning_rate": 9.180100502512563e-05,
-      "loss": 5.7835,
-      "step": 8660
-    },
-    {
-      "epoch": 4.516818774445893,
-      "grad_norm": 1.533363938331604,
-      "learning_rate": 9.180000000000001e-05,
-      "loss": 5.5389,
-      "step": 8661
-    },
-    {
-      "epoch": 4.517340286831812,
-      "grad_norm": 1.574392557144165,
-      "learning_rate": 9.179899497487437e-05,
-      "loss": 5.4781,
-      "step": 8662
-    },
-    {
-      "epoch": 4.517861799217732,
-      "grad_norm": 1.4260330200195312,
-      "learning_rate": 9.179798994974875e-05,
-      "loss": 6.0582,
-      "step": 8663
-    },
-    {
-      "epoch": 4.5183833116036505,
-      "grad_norm": 1.4073082208633423,
-      "learning_rate": 9.179698492462312e-05,
-      "loss": 5.5068,
-      "step": 8664
-    },
-    {
-      "epoch": 4.51890482398957,
-      "grad_norm": 1.3972843885421753,
-      "learning_rate": 9.179597989949749e-05,
-      "loss": 5.8013,
-      "step": 8665
-    },
-    {
-      "epoch": 4.5194263363754885,
-      "grad_norm": 1.3590000867843628,
-      "learning_rate": 9.179497487437187e-05,
-      "loss": 5.8535,
-      "step": 8666
-    },
-    {
-      "epoch": 4.519947848761408,
-      "grad_norm": 1.4307937622070312,
-      "learning_rate": 9.179396984924624e-05,
-      "loss": 5.9192,
-      "step": 8667
-    },
-    {
-      "epoch": 4.5204693611473274,
-      "grad_norm": 1.2858229875564575,
-      "learning_rate": 9.179296482412061e-05,
-      "loss": 5.9523,
-      "step": 8668
-    },
-    {
-      "epoch": 4.520990873533247,
-      "grad_norm": 1.4741249084472656,
-      "learning_rate": 9.179195979899498e-05,
-      "loss": 5.7912,
-      "step": 8669
-    },
-    {
-      "epoch": 4.5215123859191655,
-      "grad_norm": 1.599625825881958,
-      "learning_rate": 9.179095477386936e-05,
-      "loss": 5.5054,
-      "step": 8670
-    },
-    {
-      "epoch": 4.522033898305085,
-      "grad_norm": 1.544257640838623,
-      "learning_rate": 9.178994974874372e-05,
-      "loss": 5.7665,
-      "step": 8671
-    },
-    {
-      "epoch": 4.5225554106910035,
-      "grad_norm": 1.4466798305511475,
-      "learning_rate": 9.17889447236181e-05,
-      "loss": 5.7608,
-      "step": 8672
-    },
-    {
-      "epoch": 4.523076923076923,
-      "grad_norm": 1.3976002931594849,
-      "learning_rate": 9.178793969849246e-05,
-      "loss": 5.5555,
-      "step": 8673
-    },
-    {
-      "epoch": 4.5235984354628425,
-      "grad_norm": 1.2706423997879028,
-      "learning_rate": 9.178693467336684e-05,
-      "loss": 6.121,
-      "step": 8674
-    },
-    {
-      "epoch": 4.524119947848761,
-      "grad_norm": 1.340128779411316,
-      "learning_rate": 9.17859296482412e-05,
-      "loss": 5.8943,
-      "step": 8675
-    },
-    {
-      "epoch": 4.5246414602346805,
-      "grad_norm": 1.560336947441101,
-      "learning_rate": 9.178492462311558e-05,
-      "loss": 5.7664,
-      "step": 8676
-    },
-    {
-      "epoch": 4.5251629726206,
-      "grad_norm": 1.4174737930297852,
-      "learning_rate": 9.178391959798996e-05,
-      "loss": 5.7851,
-      "step": 8677
-    },
-    {
-      "epoch": 4.5256844850065185,
-      "grad_norm": 1.4999704360961914,
-      "learning_rate": 9.178291457286434e-05,
-      "loss": 5.8814,
-      "step": 8678
-    },
-    {
-      "epoch": 4.526205997392438,
-      "grad_norm": 1.5226984024047852,
-      "learning_rate": 9.17819095477387e-05,
-      "loss": 5.5445,
-      "step": 8679
-    },
-    {
-      "epoch": 4.5267275097783575,
-      "grad_norm": 1.4151042699813843,
-      "learning_rate": 9.178090452261307e-05,
-      "loss": 5.4419,
-      "step": 8680
-    },
-    {
-      "epoch": 4.527249022164276,
-      "grad_norm": 1.3811421394348145,
-      "learning_rate": 9.177989949748744e-05,
-      "loss": 5.6137,
-      "step": 8681
-    },
-    {
-      "epoch": 4.5277705345501955,
-      "grad_norm": 1.4657293558120728,
-      "learning_rate": 9.177889447236181e-05,
-      "loss": 5.4847,
-      "step": 8682
-    },
-    {
-      "epoch": 4.528292046936115,
-      "grad_norm": 1.3787561655044556,
-      "learning_rate": 9.177788944723619e-05,
-      "loss": 5.6494,
-      "step": 8683
-    },
-    {
-      "epoch": 4.5288135593220336,
-      "grad_norm": 1.469744086265564,
-      "learning_rate": 9.177688442211055e-05,
-      "loss": 5.4992,
-      "step": 8684
-    },
-    {
-      "epoch": 4.529335071707953,
-      "grad_norm": 1.4011155366897583,
-      "learning_rate": 9.177587939698493e-05,
-      "loss": 5.9485,
-      "step": 8685
-    },
-    {
-      "epoch": 4.5298565840938725,
-      "grad_norm": 1.3122327327728271,
-      "learning_rate": 9.177487437185929e-05,
-      "loss": 5.9295,
-      "step": 8686
-    },
-    {
-      "epoch": 4.530378096479791,
-      "grad_norm": 1.4872660636901855,
-      "learning_rate": 9.177386934673367e-05,
-      "loss": 5.2667,
-      "step": 8687
-    },
-    {
-      "epoch": 4.5308996088657105,
-      "grad_norm": 1.4262043237686157,
-      "learning_rate": 9.177286432160805e-05,
-      "loss": 5.6177,
-      "step": 8688
-    },
-    {
-      "epoch": 4.53142112125163,
-      "grad_norm": 1.4216456413269043,
-      "learning_rate": 9.177185929648243e-05,
-      "loss": 5.7877,
-      "step": 8689
-    },
-    {
-      "epoch": 4.531942633637549,
-      "grad_norm": 1.4967772960662842,
-      "learning_rate": 9.177085427135679e-05,
-      "loss": 5.9375,
-      "step": 8690
-    },
-    {
-      "epoch": 4.532464146023468,
-      "grad_norm": 1.4880205392837524,
-      "learning_rate": 9.176984924623117e-05,
-      "loss": 5.7093,
-      "step": 8691
-    },
-    {
-      "epoch": 4.5329856584093875,
-      "grad_norm": 1.71427583694458,
-      "learning_rate": 9.176884422110553e-05,
-      "loss": 6.1252,
-      "step": 8692
-    },
-    {
-      "epoch": 4.533507170795306,
-      "grad_norm": 1.421834111213684,
-      "learning_rate": 9.17678391959799e-05,
-      "loss": 6.0578,
-      "step": 8693
-    },
-    {
-      "epoch": 4.5340286831812255,
-      "grad_norm": 1.4830957651138306,
-      "learning_rate": 9.176683417085427e-05,
-      "loss": 5.621,
-      "step": 8694
-    },
-    {
-      "epoch": 4.534550195567145,
-      "grad_norm": 1.423503041267395,
-      "learning_rate": 9.176582914572864e-05,
-      "loss": 6.1092,
-      "step": 8695
-    },
-    {
-      "epoch": 4.535071707953064,
-      "grad_norm": 1.677391529083252,
-      "learning_rate": 9.176482412060302e-05,
-      "loss": 5.8529,
-      "step": 8696
-    },
-    {
-      "epoch": 4.535593220338983,
-      "grad_norm": 1.4934273958206177,
-      "learning_rate": 9.17638190954774e-05,
-      "loss": 5.6132,
-      "step": 8697
-    },
-    {
-      "epoch": 4.5361147327249025,
-      "grad_norm": 1.444324254989624,
-      "learning_rate": 9.176281407035177e-05,
-      "loss": 5.8703,
-      "step": 8698
-    },
-    {
-      "epoch": 4.536636245110821,
-      "grad_norm": 1.494888186454773,
-      "learning_rate": 9.176180904522614e-05,
-      "loss": 5.6694,
-      "step": 8699
-    },
-    {
-      "epoch": 4.5371577574967406,
-      "grad_norm": 1.6088695526123047,
-      "learning_rate": 9.176080402010051e-05,
-      "loss": 5.4731,
-      "step": 8700
-    },
-    {
-      "epoch": 4.53767926988266,
-      "grad_norm": 1.5434436798095703,
-      "learning_rate": 9.175979899497488e-05,
-      "loss": 5.7504,
-      "step": 8701
-    },
-    {
-      "epoch": 4.538200782268579,
-      "grad_norm": 1.498025894165039,
-      "learning_rate": 9.175879396984926e-05,
-      "loss": 5.8716,
-      "step": 8702
-    },
-    {
-      "epoch": 4.538722294654498,
-      "grad_norm": 1.4497346878051758,
-      "learning_rate": 9.175778894472362e-05,
-      "loss": 6.0863,
-      "step": 8703
-    },
-    {
-      "epoch": 4.5392438070404175,
-      "grad_norm": 1.4767022132873535,
-      "learning_rate": 9.1756783919598e-05,
-      "loss": 5.2402,
-      "step": 8704
-    },
-    {
-      "epoch": 4.539765319426336,
-      "grad_norm": 1.3352497816085815,
-      "learning_rate": 9.175577889447236e-05,
-      "loss": 5.8142,
-      "step": 8705
-    },
-    {
-      "epoch": 4.540286831812256,
-      "grad_norm": 1.5846811532974243,
-      "learning_rate": 9.175477386934673e-05,
-      "loss": 5.515,
-      "step": 8706
-    },
-    {
-      "epoch": 4.540808344198175,
-      "grad_norm": 1.4087646007537842,
-      "learning_rate": 9.17537688442211e-05,
-      "loss": 5.7846,
-      "step": 8707
-    },
-    {
-      "epoch": 4.541329856584094,
-      "grad_norm": 1.5147112607955933,
-      "learning_rate": 9.175276381909548e-05,
-      "loss": 5.66,
-      "step": 8708
-    },
-    {
-      "epoch": 4.541851368970013,
-      "grad_norm": 1.4473127126693726,
-      "learning_rate": 9.175175879396986e-05,
-      "loss": 5.5291,
-      "step": 8709
-    },
-    {
-      "epoch": 4.5423728813559325,
-      "grad_norm": 1.4192097187042236,
-      "learning_rate": 9.175075376884422e-05,
-      "loss": 5.5304,
-      "step": 8710
-    },
-    {
-      "epoch": 4.542894393741851,
-      "grad_norm": 1.4624630212783813,
-      "learning_rate": 9.17497487437186e-05,
-      "loss": 5.9057,
-      "step": 8711
-    },
-    {
-      "epoch": 4.543415906127771,
-      "grad_norm": 1.585584282875061,
-      "learning_rate": 9.174874371859297e-05,
-      "loss": 5.3852,
-      "step": 8712
-    },
-    {
-      "epoch": 4.54393741851369,
-      "grad_norm": 1.5647541284561157,
-      "learning_rate": 9.174773869346734e-05,
-      "loss": 5.2074,
-      "step": 8713
-    },
-    {
-      "epoch": 4.544458930899609,
-      "grad_norm": 1.639654278755188,
-      "learning_rate": 9.174673366834171e-05,
-      "loss": 5.5947,
-      "step": 8714
-    },
-    {
-      "epoch": 4.544980443285528,
-      "grad_norm": 1.6701154708862305,
-      "learning_rate": 9.174572864321609e-05,
-      "loss": 5.25,
-      "step": 8715
-    },
-    {
-      "epoch": 4.5455019556714475,
-      "grad_norm": 1.5566840171813965,
-      "learning_rate": 9.174472361809045e-05,
-      "loss": 5.9526,
-      "step": 8716
-    },
-    {
-      "epoch": 4.546023468057366,
-      "grad_norm": 1.5511070489883423,
-      "learning_rate": 9.174371859296483e-05,
-      "loss": 5.9813,
-      "step": 8717
-    },
-    {
-      "epoch": 4.546544980443286,
-      "grad_norm": 1.4521782398223877,
-      "learning_rate": 9.17427135678392e-05,
-      "loss": 5.3153,
-      "step": 8718
-    },
-    {
-      "epoch": 4.547066492829205,
-      "grad_norm": 1.3967090845108032,
-      "learning_rate": 9.174170854271357e-05,
-      "loss": 5.7453,
-      "step": 8719
-    },
-    {
-      "epoch": 4.547588005215124,
-      "grad_norm": 1.680825114250183,
-      "learning_rate": 9.174070351758795e-05,
-      "loss": 5.264,
-      "step": 8720
-    },
-    {
-      "epoch": 4.548109517601043,
-      "grad_norm": 1.3267009258270264,
-      "learning_rate": 9.173969849246231e-05,
-      "loss": 6.1034,
-      "step": 8721
-    },
-    {
-      "epoch": 4.548631029986963,
-      "grad_norm": 1.4496004581451416,
-      "learning_rate": 9.173869346733669e-05,
-      "loss": 5.6644,
-      "step": 8722
-    },
-    {
-      "epoch": 4.549152542372881,
-      "grad_norm": 1.4133682250976562,
-      "learning_rate": 9.173768844221105e-05,
-      "loss": 6.1245,
-      "step": 8723
-    },
-    {
-      "epoch": 4.549674054758801,
-      "grad_norm": 1.5146615505218506,
-      "learning_rate": 9.173668341708543e-05,
-      "loss": 5.8531,
-      "step": 8724
-    },
-    {
-      "epoch": 4.55019556714472,
-      "grad_norm": 1.3808984756469727,
-      "learning_rate": 9.17356783919598e-05,
-      "loss": 5.6086,
-      "step": 8725
-    },
-    {
-      "epoch": 4.550717079530639,
-      "grad_norm": 1.46841299533844,
-      "learning_rate": 9.173467336683417e-05,
-      "loss": 5.4893,
-      "step": 8726
-    },
-    {
-      "epoch": 4.551238591916558,
-      "grad_norm": 1.4762296676635742,
-      "learning_rate": 9.173366834170854e-05,
-      "loss": 5.6428,
-      "step": 8727
-    },
-    {
-      "epoch": 4.551760104302478,
-      "grad_norm": 1.4300376176834106,
-      "learning_rate": 9.173266331658292e-05,
-      "loss": 5.5641,
-      "step": 8728
-    },
-    {
-      "epoch": 4.552281616688396,
-      "grad_norm": 1.4669941663742065,
-      "learning_rate": 9.17316582914573e-05,
-      "loss": 5.8773,
-      "step": 8729
-    },
-    {
-      "epoch": 4.552803129074316,
-      "grad_norm": 1.5359550714492798,
-      "learning_rate": 9.173065326633167e-05,
-      "loss": 5.8035,
-      "step": 8730
-    },
-    {
-      "epoch": 4.553324641460234,
-      "grad_norm": 1.5147991180419922,
-      "learning_rate": 9.172964824120604e-05,
-      "loss": 5.7443,
-      "step": 8731
-    },
-    {
-      "epoch": 4.553846153846154,
-      "grad_norm": 1.5537841320037842,
-      "learning_rate": 9.17286432160804e-05,
-      "loss": 5.622,
-      "step": 8732
-    },
-    {
-      "epoch": 4.554367666232073,
-      "grad_norm": 1.6170248985290527,
-      "learning_rate": 9.172763819095478e-05,
-      "loss": 5.7587,
-      "step": 8733
-    },
-    {
-      "epoch": 4.554889178617993,
-      "grad_norm": 1.4892398118972778,
-      "learning_rate": 9.172663316582914e-05,
-      "loss": 5.5878,
-      "step": 8734
-    },
-    {
-      "epoch": 4.555410691003911,
-      "grad_norm": 1.4726505279541016,
-      "learning_rate": 9.172562814070352e-05,
-      "loss": 5.0978,
-      "step": 8735
-    },
-    {
-      "epoch": 4.555932203389831,
-      "grad_norm": 1.3390599489212036,
-      "learning_rate": 9.172462311557789e-05,
-      "loss": 6.1221,
-      "step": 8736
-    },
-    {
-      "epoch": 4.556453715775749,
-      "grad_norm": 1.5746928453445435,
-      "learning_rate": 9.172361809045226e-05,
-      "loss": 5.3129,
-      "step": 8737
-    },
-    {
-      "epoch": 4.556975228161669,
-      "grad_norm": 1.510516881942749,
-      "learning_rate": 9.172261306532664e-05,
-      "loss": 5.6887,
-      "step": 8738
-    },
-    {
-      "epoch": 4.557496740547588,
-      "grad_norm": 1.4536935091018677,
-      "learning_rate": 9.172160804020102e-05,
-      "loss": 5.9217,
-      "step": 8739
-    },
-    {
-      "epoch": 4.558018252933508,
-      "grad_norm": 1.5286935567855835,
-      "learning_rate": 9.172060301507538e-05,
-      "loss": 5.4753,
-      "step": 8740
-    },
-    {
-      "epoch": 4.558539765319426,
-      "grad_norm": 1.454332947731018,
-      "learning_rate": 9.171959798994976e-05,
-      "loss": 5.8488,
-      "step": 8741
-    },
-    {
-      "epoch": 4.559061277705346,
-      "grad_norm": 1.4274616241455078,
-      "learning_rate": 9.171859296482412e-05,
-      "loss": 5.5225,
-      "step": 8742
-    },
-    {
-      "epoch": 4.559582790091264,
-      "grad_norm": 1.7226529121398926,
-      "learning_rate": 9.17175879396985e-05,
-      "loss": 5.6195,
-      "step": 8743
-    },
-    {
-      "epoch": 4.560104302477184,
-      "grad_norm": 1.4996707439422607,
-      "learning_rate": 9.171658291457287e-05,
-      "loss": 5.9606,
-      "step": 8744
-    },
-    {
-      "epoch": 4.560625814863103,
-      "grad_norm": 1.4412786960601807,
-      "learning_rate": 9.171557788944723e-05,
-      "loss": 5.7576,
-      "step": 8745
-    },
-    {
-      "epoch": 4.561147327249023,
-      "grad_norm": 1.3681682348251343,
-      "learning_rate": 9.171457286432161e-05,
-      "loss": 4.7421,
-      "step": 8746
-    },
-    {
-      "epoch": 4.561668839634941,
-      "grad_norm": 1.6577523946762085,
-      "learning_rate": 9.171356783919597e-05,
-      "loss": 5.7554,
-      "step": 8747
-    },
-    {
-      "epoch": 4.562190352020861,
-      "grad_norm": 1.4363926649093628,
-      "learning_rate": 9.171256281407035e-05,
-      "loss": 5.509,
-      "step": 8748
-    },
-    {
-      "epoch": 4.562711864406779,
-      "grad_norm": 1.6629704236984253,
-      "learning_rate": 9.171155778894473e-05,
-      "loss": 5.6517,
-      "step": 8749
-    },
-    {
-      "epoch": 4.563233376792699,
-      "grad_norm": 1.522254228591919,
-      "learning_rate": 9.171055276381911e-05,
-      "loss": 5.7574,
-      "step": 8750
-    },
-    {
-      "epoch": 4.563754889178618,
-      "grad_norm": 1.5916715860366821,
-      "learning_rate": 9.170954773869347e-05,
-      "loss": 5.2891,
-      "step": 8751
-    },
-    {
-      "epoch": 4.564276401564538,
-      "grad_norm": 1.6371300220489502,
-      "learning_rate": 9.170854271356785e-05,
-      "loss": 5.4513,
-      "step": 8752
-    },
-    {
-      "epoch": 4.564797913950456,
-      "grad_norm": 1.5032769441604614,
-      "learning_rate": 9.170753768844221e-05,
-      "loss": 5.8863,
-      "step": 8753
-    },
-    {
-      "epoch": 4.565319426336376,
-      "grad_norm": 1.6248977184295654,
-      "learning_rate": 9.170653266331659e-05,
-      "loss": 5.9064,
-      "step": 8754
-    },
-    {
-      "epoch": 4.565840938722294,
-      "grad_norm": 1.5138685703277588,
-      "learning_rate": 9.170552763819096e-05,
-      "loss": 5.5959,
-      "step": 8755
-    },
-    {
-      "epoch": 4.566362451108214,
-      "grad_norm": 1.5090042352676392,
-      "learning_rate": 9.170452261306533e-05,
-      "loss": 4.9782,
-      "step": 8756
-    },
-    {
-      "epoch": 4.566883963494133,
-      "grad_norm": 1.3818196058273315,
-      "learning_rate": 9.17035175879397e-05,
-      "loss": 6.0381,
-      "step": 8757
-    },
-    {
-      "epoch": 4.567405475880053,
-      "grad_norm": 1.4270144701004028,
-      "learning_rate": 9.170251256281408e-05,
-      "loss": 5.691,
-      "step": 8758
-    },
-    {
-      "epoch": 4.567926988265971,
-      "grad_norm": 1.3970242738723755,
-      "learning_rate": 9.170150753768845e-05,
-      "loss": 6.071,
-      "step": 8759
-    },
-    {
-      "epoch": 4.568448500651891,
-      "grad_norm": 1.5757447481155396,
-      "learning_rate": 9.170050251256282e-05,
-      "loss": 5.5438,
-      "step": 8760
-    },
-    {
-      "epoch": 4.568970013037809,
-      "grad_norm": 1.360434651374817,
-      "learning_rate": 9.16994974874372e-05,
-      "loss": 5.9421,
-      "step": 8761
-    },
-    {
-      "epoch": 4.569491525423729,
-      "grad_norm": 1.675575852394104,
-      "learning_rate": 9.169849246231156e-05,
-      "loss": 5.5551,
-      "step": 8762
-    },
-    {
-      "epoch": 4.570013037809648,
-      "grad_norm": 1.4929884672164917,
-      "learning_rate": 9.169748743718594e-05,
-      "loss": 5.7918,
-      "step": 8763
-    },
-    {
-      "epoch": 4.570534550195567,
-      "grad_norm": 1.4243988990783691,
-      "learning_rate": 9.16964824120603e-05,
-      "loss": 5.5457,
-      "step": 8764
-    },
-    {
-      "epoch": 4.571056062581486,
-      "grad_norm": 1.3796207904815674,
-      "learning_rate": 9.169547738693468e-05,
-      "loss": 5.7656,
-      "step": 8765
-    },
-    {
-      "epoch": 4.571577574967406,
-      "grad_norm": 1.4633604288101196,
-      "learning_rate": 9.169447236180904e-05,
-      "loss": 5.8621,
-      "step": 8766
-    },
-    {
-      "epoch": 4.572099087353324,
-      "grad_norm": 1.530441403388977,
-      "learning_rate": 9.169346733668342e-05,
-      "loss": 5.0478,
-      "step": 8767
-    },
-    {
-      "epoch": 4.572620599739244,
-      "grad_norm": 1.3494932651519775,
-      "learning_rate": 9.169246231155779e-05,
-      "loss": 6.0426,
-      "step": 8768
-    },
-    {
-      "epoch": 4.573142112125163,
-      "grad_norm": 1.4170658588409424,
-      "learning_rate": 9.169145728643216e-05,
-      "loss": 5.6762,
-      "step": 8769
-    },
-    {
-      "epoch": 4.573663624511082,
-      "grad_norm": 1.33936607837677,
-      "learning_rate": 9.169045226130654e-05,
-      "loss": 5.9197,
-      "step": 8770
-    },
-    {
-      "epoch": 4.574185136897001,
-      "grad_norm": 1.5455797910690308,
-      "learning_rate": 9.168944723618092e-05,
-      "loss": 5.2322,
-      "step": 8771
-    },
-    {
-      "epoch": 4.574706649282921,
-      "grad_norm": 1.478094458580017,
-      "learning_rate": 9.168844221105528e-05,
-      "loss": 5.6736,
-      "step": 8772
-    },
-    {
-      "epoch": 4.575228161668839,
-      "grad_norm": 1.526576042175293,
-      "learning_rate": 9.168743718592965e-05,
-      "loss": 4.5858,
-      "step": 8773
-    },
-    {
-      "epoch": 4.575749674054759,
-      "grad_norm": 1.5601142644882202,
-      "learning_rate": 9.168643216080403e-05,
-      "loss": 5.8187,
-      "step": 8774
-    },
-    {
-      "epoch": 4.576271186440678,
-      "grad_norm": 1.391241192817688,
-      "learning_rate": 9.168542713567839e-05,
-      "loss": 5.9999,
-      "step": 8775
-    },
-    {
-      "epoch": 4.576792698826597,
-      "grad_norm": 1.4582111835479736,
-      "learning_rate": 9.168442211055277e-05,
-      "loss": 5.5618,
-      "step": 8776
-    },
-    {
-      "epoch": 4.577314211212516,
-      "grad_norm": 1.4616389274597168,
-      "learning_rate": 9.168341708542713e-05,
-      "loss": 5.7089,
-      "step": 8777
-    },
-    {
-      "epoch": 4.577835723598436,
-      "grad_norm": 1.4980223178863525,
-      "learning_rate": 9.168241206030151e-05,
-      "loss": 5.3176,
-      "step": 8778
-    },
-    {
-      "epoch": 4.578357235984354,
-      "grad_norm": 1.4985878467559814,
-      "learning_rate": 9.168140703517589e-05,
-      "loss": 5.7584,
-      "step": 8779
-    },
-    {
-      "epoch": 4.578878748370274,
-      "grad_norm": 1.4382927417755127,
-      "learning_rate": 9.168040201005027e-05,
-      "loss": 5.9466,
-      "step": 8780
-    },
-    {
-      "epoch": 4.579400260756193,
-      "grad_norm": 1.5502241849899292,
-      "learning_rate": 9.167939698492463e-05,
-      "loss": 5.6022,
-      "step": 8781
-    },
-    {
-      "epoch": 4.579921773142112,
-      "grad_norm": 1.378084659576416,
-      "learning_rate": 9.167839195979901e-05,
-      "loss": 5.9144,
-      "step": 8782
-    },
-    {
-      "epoch": 4.580443285528031,
-      "grad_norm": 1.5262136459350586,
-      "learning_rate": 9.167738693467337e-05,
-      "loss": 5.6609,
-      "step": 8783
-    },
-    {
-      "epoch": 4.580964797913951,
-      "grad_norm": 1.5146045684814453,
-      "learning_rate": 9.167638190954775e-05,
-      "loss": 5.7288,
-      "step": 8784
-    },
-    {
-      "epoch": 4.581486310299869,
-      "grad_norm": 1.528562307357788,
-      "learning_rate": 9.167537688442211e-05,
-      "loss": 5.3503,
-      "step": 8785
-    },
-    {
-      "epoch": 4.582007822685789,
-      "grad_norm": 1.5825990438461304,
-      "learning_rate": 9.167437185929648e-05,
-      "loss": 5.439,
-      "step": 8786
-    },
-    {
-      "epoch": 4.582529335071708,
-      "grad_norm": 1.4602211713790894,
-      "learning_rate": 9.167336683417086e-05,
-      "loss": 5.4332,
-      "step": 8787
-    },
-    {
-      "epoch": 4.583050847457627,
-      "grad_norm": 1.5268741846084595,
-      "learning_rate": 9.167236180904522e-05,
-      "loss": 5.9975,
-      "step": 8788
-    },
-    {
-      "epoch": 4.583572359843546,
-      "grad_norm": 1.4936295747756958,
-      "learning_rate": 9.16713567839196e-05,
-      "loss": 5.3616,
-      "step": 8789
-    },
-    {
-      "epoch": 4.584093872229466,
-      "grad_norm": 1.466723084449768,
-      "learning_rate": 9.167035175879398e-05,
-      "loss": 5.3297,
-      "step": 8790
-    },
-    {
-      "epoch": 4.584615384615384,
-      "grad_norm": 1.4672462940216064,
-      "learning_rate": 9.166934673366835e-05,
-      "loss": 5.889,
-      "step": 8791
-    },
-    {
-      "epoch": 4.585136897001304,
-      "grad_norm": 1.6096230745315552,
-      "learning_rate": 9.166834170854272e-05,
-      "loss": 5.3103,
-      "step": 8792
-    },
-    {
-      "epoch": 4.585658409387223,
-      "grad_norm": 1.534489631652832,
-      "learning_rate": 9.16673366834171e-05,
-      "loss": 5.4349,
-      "step": 8793
-    },
-    {
-      "epoch": 4.586179921773142,
-      "grad_norm": 1.3953675031661987,
-      "learning_rate": 9.166633165829146e-05,
-      "loss": 5.828,
-      "step": 8794
-    },
-    {
-      "epoch": 4.586701434159061,
-      "grad_norm": 1.4405044317245483,
-      "learning_rate": 9.166532663316584e-05,
-      "loss": 5.6225,
-      "step": 8795
-    },
-    {
-      "epoch": 4.587222946544981,
-      "grad_norm": 1.6320164203643799,
-      "learning_rate": 9.16643216080402e-05,
-      "loss": 5.0623,
-      "step": 8796
-    },
-    {
-      "epoch": 4.587744458930899,
-      "grad_norm": 1.4186238050460815,
-      "learning_rate": 9.166331658291458e-05,
-      "loss": 5.7288,
-      "step": 8797
-    },
-    {
-      "epoch": 4.588265971316819,
-      "grad_norm": 1.3992280960083008,
-      "learning_rate": 9.166231155778894e-05,
-      "loss": 5.1396,
-      "step": 8798
-    },
-    {
-      "epoch": 4.588787483702738,
-      "grad_norm": 1.515467643737793,
-      "learning_rate": 9.166130653266332e-05,
-      "loss": 5.5447,
-      "step": 8799
-    },
-    {
-      "epoch": 4.589308996088657,
-      "grad_norm": 1.4290887117385864,
-      "learning_rate": 9.16603015075377e-05,
-      "loss": 5.8123,
-      "step": 8800
-    },
-    {
-      "epoch": 4.589830508474576,
-      "grad_norm": 1.3807953596115112,
-      "learning_rate": 9.165929648241206e-05,
-      "loss": 5.7783,
-      "step": 8801
-    },
-    {
-      "epoch": 4.590352020860496,
-      "grad_norm": 1.4325942993164062,
-      "learning_rate": 9.165829145728644e-05,
-      "loss": 4.8434,
-      "step": 8802
-    },
-    {
-      "epoch": 4.590873533246414,
-      "grad_norm": 1.4220232963562012,
-      "learning_rate": 9.16572864321608e-05,
-      "loss": 5.678,
-      "step": 8803
-    },
-    {
-      "epoch": 4.591395045632334,
-      "grad_norm": 1.4154174327850342,
-      "learning_rate": 9.165628140703518e-05,
-      "loss": 5.6142,
-      "step": 8804
-    },
-    {
-      "epoch": 4.591916558018253,
-      "grad_norm": 1.3722482919692993,
-      "learning_rate": 9.165527638190955e-05,
-      "loss": 5.5608,
-      "step": 8805
-    },
-    {
-      "epoch": 4.592438070404172,
-      "grad_norm": 1.2543498277664185,
-      "learning_rate": 9.165427135678393e-05,
-      "loss": 5.947,
-      "step": 8806
-    },
-    {
-      "epoch": 4.592959582790091,
-      "grad_norm": 1.3852460384368896,
-      "learning_rate": 9.165326633165829e-05,
-      "loss": 5.4052,
-      "step": 8807
-    },
-    {
-      "epoch": 4.593481095176011,
-      "grad_norm": 1.495202660560608,
-      "learning_rate": 9.165226130653267e-05,
-      "loss": 5.2904,
-      "step": 8808
-    },
-    {
-      "epoch": 4.594002607561929,
-      "grad_norm": 1.4949921369552612,
-      "learning_rate": 9.165125628140703e-05,
-      "loss": 5.9613,
-      "step": 8809
-    },
-    {
-      "epoch": 4.594524119947849,
-      "grad_norm": 1.5349218845367432,
-      "learning_rate": 9.165025125628141e-05,
-      "loss": 6.0211,
-      "step": 8810
-    },
-    {
-      "epoch": 4.595045632333768,
-      "grad_norm": 1.3602027893066406,
-      "learning_rate": 9.164924623115579e-05,
-      "loss": 5.7767,
-      "step": 8811
-    },
-    {
-      "epoch": 4.595567144719687,
-      "grad_norm": 1.56438410282135,
-      "learning_rate": 9.164824120603015e-05,
-      "loss": 5.3285,
-      "step": 8812
-    },
-    {
-      "epoch": 4.596088657105606,
-      "grad_norm": 1.4626444578170776,
-      "learning_rate": 9.164723618090453e-05,
-      "loss": 5.7047,
-      "step": 8813
-    },
-    {
-      "epoch": 4.596610169491526,
-      "grad_norm": 1.467127799987793,
-      "learning_rate": 9.16462311557789e-05,
-      "loss": 5.6033,
-      "step": 8814
-    },
-    {
-      "epoch": 4.597131681877444,
-      "grad_norm": 1.5471903085708618,
-      "learning_rate": 9.164522613065327e-05,
-      "loss": 5.9021,
-      "step": 8815
-    },
-    {
-      "epoch": 4.597653194263364,
-      "grad_norm": 1.5832263231277466,
-      "learning_rate": 9.164422110552764e-05,
-      "loss": 5.0782,
-      "step": 8816
-    },
-    {
-      "epoch": 4.598174706649283,
-      "grad_norm": 1.5199694633483887,
-      "learning_rate": 9.164321608040201e-05,
-      "loss": 5.5607,
-      "step": 8817
-    },
-    {
-      "epoch": 4.598696219035202,
-      "grad_norm": 1.6109683513641357,
-      "learning_rate": 9.164221105527638e-05,
-      "loss": 5.4597,
-      "step": 8818
-    },
-    {
-      "epoch": 4.599217731421121,
-      "grad_norm": 1.409184217453003,
-      "learning_rate": 9.164120603015076e-05,
-      "loss": 5.6409,
-      "step": 8819
-    },
-    {
-      "epoch": 4.599739243807041,
-      "grad_norm": 1.5530985593795776,
-      "learning_rate": 9.164020100502513e-05,
-      "loss": 5.8951,
-      "step": 8820
-    },
-    {
-      "epoch": 4.600260756192959,
-      "grad_norm": 1.4365237951278687,
-      "learning_rate": 9.163919597989951e-05,
-      "loss": 5.9634,
-      "step": 8821
-    },
-    {
-      "epoch": 4.600782268578879,
-      "grad_norm": 1.4714112281799316,
-      "learning_rate": 9.163819095477388e-05,
-      "loss": 5.97,
-      "step": 8822
-    },
-    {
-      "epoch": 4.601303780964798,
-      "grad_norm": 1.5788793563842773,
-      "learning_rate": 9.163718592964825e-05,
-      "loss": 5.389,
-      "step": 8823
-    },
-    {
-      "epoch": 4.601825293350717,
-      "grad_norm": 1.344529628753662,
-      "learning_rate": 9.163618090452262e-05,
-      "loss": 6.0297,
-      "step": 8824
-    },
-    {
-      "epoch": 4.602346805736636,
-      "grad_norm": 1.5211730003356934,
-      "learning_rate": 9.163517587939698e-05,
-      "loss": 5.393,
-      "step": 8825
-    },
-    {
-      "epoch": 4.602868318122555,
-      "grad_norm": 1.4628199338912964,
-      "learning_rate": 9.163417085427136e-05,
-      "loss": 5.5329,
-      "step": 8826
-    },
-    {
-      "epoch": 4.603389830508474,
-      "grad_norm": 1.4561302661895752,
-      "learning_rate": 9.163316582914573e-05,
-      "loss": 5.1572,
-      "step": 8827
-    },
-    {
-      "epoch": 4.603911342894394,
-      "grad_norm": 1.4034607410430908,
-      "learning_rate": 9.16321608040201e-05,
-      "loss": 5.8246,
-      "step": 8828
-    },
-    {
-      "epoch": 4.604432855280313,
-      "grad_norm": 1.6075559854507446,
-      "learning_rate": 9.163115577889447e-05,
-      "loss": 5.1842,
-      "step": 8829
-    },
-    {
-      "epoch": 4.604954367666232,
-      "grad_norm": 1.4692732095718384,
-      "learning_rate": 9.163015075376885e-05,
-      "loss": 5.186,
-      "step": 8830
-    },
-    {
-      "epoch": 4.605475880052151,
-      "grad_norm": 1.6216390132904053,
-      "learning_rate": 9.162914572864322e-05,
-      "loss": 5.104,
-      "step": 8831
-    },
-    {
-      "epoch": 4.60599739243807,
-      "grad_norm": 1.5397733449935913,
-      "learning_rate": 9.16281407035176e-05,
-      "loss": 5.4522,
-      "step": 8832
-    },
-    {
-      "epoch": 4.606518904823989,
-      "grad_norm": 1.5935832262039185,
-      "learning_rate": 9.162713567839197e-05,
-      "loss": 5.5685,
-      "step": 8833
-    },
-    {
-      "epoch": 4.607040417209909,
-      "grad_norm": 1.5095871686935425,
-      "learning_rate": 9.162613065326634e-05,
-      "loss": 5.6136,
-      "step": 8834
-    },
-    {
-      "epoch": 4.607561929595828,
-      "grad_norm": 1.358251929283142,
-      "learning_rate": 9.162512562814071e-05,
-      "loss": 5.7297,
-      "step": 8835
-    },
-    {
-      "epoch": 4.608083441981747,
-      "grad_norm": 1.468448281288147,
-      "learning_rate": 9.162412060301509e-05,
-      "loss": 5.2262,
-      "step": 8836
-    },
-    {
-      "epoch": 4.608604954367666,
-      "grad_norm": 1.5538736581802368,
-      "learning_rate": 9.162311557788945e-05,
-      "loss": 5.4248,
-      "step": 8837
-    },
-    {
-      "epoch": 4.609126466753585,
-      "grad_norm": 1.5137683153152466,
-      "learning_rate": 9.162211055276381e-05,
-      "loss": 5.505,
-      "step": 8838
-    },
-    {
-      "epoch": 4.609647979139504,
-      "grad_norm": 1.4768542051315308,
-      "learning_rate": 9.162110552763819e-05,
-      "loss": 5.9134,
-      "step": 8839
-    },
-    {
-      "epoch": 4.610169491525424,
-      "grad_norm": 1.5216604471206665,
-      "learning_rate": 9.162010050251256e-05,
-      "loss": 4.7665,
-      "step": 8840
-    },
-    {
-      "epoch": 4.610691003911343,
-      "grad_norm": 1.5291028022766113,
-      "learning_rate": 9.161909547738693e-05,
-      "loss": 5.8653,
-      "step": 8841
-    },
-    {
-      "epoch": 4.611212516297262,
-      "grad_norm": 1.5087006092071533,
-      "learning_rate": 9.161809045226131e-05,
-      "loss": 5.7902,
-      "step": 8842
-    },
-    {
-      "epoch": 4.611734028683181,
-      "grad_norm": 1.3554452657699585,
-      "learning_rate": 9.161708542713569e-05,
-      "loss": 5.9592,
-      "step": 8843
-    },
-    {
-      "epoch": 4.6122555410691,
-      "grad_norm": 1.570695161819458,
-      "learning_rate": 9.161608040201005e-05,
-      "loss": 5.5813,
-      "step": 8844
-    },
-    {
-      "epoch": 4.612777053455019,
-      "grad_norm": 1.385163426399231,
-      "learning_rate": 9.161507537688443e-05,
-      "loss": 5.9122,
-      "step": 8845
-    },
-    {
-      "epoch": 4.613298565840939,
-      "grad_norm": 1.5171250104904175,
-      "learning_rate": 9.16140703517588e-05,
-      "loss": 4.9443,
-      "step": 8846
-    },
-    {
-      "epoch": 4.613820078226858,
-      "grad_norm": 1.4526631832122803,
-      "learning_rate": 9.161306532663317e-05,
-      "loss": 5.8144,
-      "step": 8847
-    },
-    {
-      "epoch": 4.614341590612777,
-      "grad_norm": 1.7378019094467163,
-      "learning_rate": 9.161206030150754e-05,
-      "loss": 5.1773,
-      "step": 8848
-    },
-    {
-      "epoch": 4.614863102998696,
-      "grad_norm": 1.5640228986740112,
-      "learning_rate": 9.161105527638192e-05,
-      "loss": 4.9637,
-      "step": 8849
-    },
-    {
-      "epoch": 4.615384615384615,
-      "grad_norm": 1.5026824474334717,
-      "learning_rate": 9.161005025125628e-05,
-      "loss": 5.8132,
-      "step": 8850
-    },
-    {
-      "epoch": 4.615906127770534,
-      "grad_norm": 1.4358112812042236,
-      "learning_rate": 9.160904522613066e-05,
-      "loss": 5.5209,
-      "step": 8851
-    },
-    {
-      "epoch": 4.616427640156454,
-      "grad_norm": 1.3528293371200562,
-      "learning_rate": 9.160804020100504e-05,
-      "loss": 6.2289,
-      "step": 8852
-    },
-    {
-      "epoch": 4.616949152542373,
-      "grad_norm": 1.5077273845672607,
-      "learning_rate": 9.16070351758794e-05,
-      "loss": 5.4336,
-      "step": 8853
-    },
-    {
-      "epoch": 4.617470664928292,
-      "grad_norm": 1.4363735914230347,
-      "learning_rate": 9.160603015075378e-05,
-      "loss": 5.7049,
-      "step": 8854
-    },
-    {
-      "epoch": 4.617992177314211,
-      "grad_norm": 1.5072983503341675,
-      "learning_rate": 9.160502512562814e-05,
-      "loss": 5.6564,
-      "step": 8855
-    },
-    {
-      "epoch": 4.61851368970013,
-      "grad_norm": 1.529957890510559,
-      "learning_rate": 9.160402010050252e-05,
-      "loss": 5.6295,
-      "step": 8856
-    },
-    {
-      "epoch": 4.6190352020860495,
-      "grad_norm": 1.4247279167175293,
-      "learning_rate": 9.160301507537688e-05,
-      "loss": 4.949,
-      "step": 8857
-    },
-    {
-      "epoch": 4.619556714471969,
-      "grad_norm": 1.7308032512664795,
-      "learning_rate": 9.160201005025126e-05,
-      "loss": 5.3786,
-      "step": 8858
-    },
-    {
-      "epoch": 4.6200782268578875,
-      "grad_norm": 1.5831295251846313,
-      "learning_rate": 9.160100502512563e-05,
-      "loss": 5.139,
-      "step": 8859
-    },
-    {
-      "epoch": 4.620599739243807,
-      "grad_norm": 1.5297378301620483,
-      "learning_rate": 9.16e-05,
-      "loss": 5.9431,
-      "step": 8860
-    },
-    {
-      "epoch": 4.621121251629726,
-      "grad_norm": 1.801652193069458,
-      "learning_rate": 9.159899497487437e-05,
-      "loss": 5.2146,
-      "step": 8861
-    },
-    {
-      "epoch": 4.621642764015645,
-      "grad_norm": 1.4471423625946045,
-      "learning_rate": 9.159798994974875e-05,
-      "loss": 5.8936,
-      "step": 8862
-    },
-    {
-      "epoch": 4.6221642764015645,
-      "grad_norm": 1.5732618570327759,
-      "learning_rate": 9.159698492462312e-05,
-      "loss": 5.3459,
-      "step": 8863
-    },
-    {
-      "epoch": 4.622685788787484,
-      "grad_norm": 1.4443753957748413,
-      "learning_rate": 9.15959798994975e-05,
-      "loss": 5.8642,
-      "step": 8864
-    },
-    {
-      "epoch": 4.6232073011734025,
-      "grad_norm": 1.3780276775360107,
-      "learning_rate": 9.159497487437187e-05,
-      "loss": 6.192,
-      "step": 8865
-    },
-    {
-      "epoch": 4.623728813559322,
-      "grad_norm": 1.328078269958496,
-      "learning_rate": 9.159396984924623e-05,
-      "loss": 6.1086,
-      "step": 8866
-    },
-    {
-      "epoch": 4.624250325945241,
-      "grad_norm": 1.5539742708206177,
-      "learning_rate": 9.159296482412061e-05,
-      "loss": 5.7669,
-      "step": 8867
-    },
-    {
-      "epoch": 4.62477183833116,
-      "grad_norm": 1.398806095123291,
-      "learning_rate": 9.159195979899497e-05,
-      "loss": 6.0357,
-      "step": 8868
-    },
-    {
-      "epoch": 4.6252933507170795,
-      "grad_norm": 1.3689343929290771,
-      "learning_rate": 9.159095477386935e-05,
-      "loss": 5.6601,
-      "step": 8869
-    },
-    {
-      "epoch": 4.625814863102999,
-      "grad_norm": 1.3714078664779663,
-      "learning_rate": 9.158994974874371e-05,
-      "loss": 5.8136,
-      "step": 8870
-    },
-    {
-      "epoch": 4.6263363754889175,
-      "grad_norm": 1.5395281314849854,
-      "learning_rate": 9.158894472361809e-05,
-      "loss": 5.3022,
-      "step": 8871
-    },
-    {
-      "epoch": 4.626857887874837,
-      "grad_norm": 1.399531364440918,
-      "learning_rate": 9.158793969849247e-05,
-      "loss": 5.8124,
-      "step": 8872
-    },
-    {
-      "epoch": 4.6273794002607564,
-      "grad_norm": 1.7446609735488892,
-      "learning_rate": 9.158693467336685e-05,
-      "loss": 5.3629,
-      "step": 8873
-    },
-    {
-      "epoch": 4.627900912646675,
-      "grad_norm": 1.4858002662658691,
-      "learning_rate": 9.158592964824121e-05,
-      "loss": 5.3155,
-      "step": 8874
-    },
-    {
-      "epoch": 4.6284224250325945,
-      "grad_norm": 1.4753323793411255,
-      "learning_rate": 9.158492462311559e-05,
-      "loss": 5.7798,
-      "step": 8875
-    },
-    {
-      "epoch": 4.628943937418514,
-      "grad_norm": 2.341142416000366,
-      "learning_rate": 9.158391959798995e-05,
-      "loss": 5.4353,
-      "step": 8876
-    },
-    {
-      "epoch": 4.6294654498044325,
-      "grad_norm": 1.5102437734603882,
-      "learning_rate": 9.158291457286433e-05,
-      "loss": 5.674,
-      "step": 8877
-    },
-    {
-      "epoch": 4.629986962190352,
-      "grad_norm": 1.4230986833572388,
-      "learning_rate": 9.15819095477387e-05,
-      "loss": 5.5732,
-      "step": 8878
-    },
-    {
-      "epoch": 4.6305084745762715,
-      "grad_norm": 1.5111114978790283,
-      "learning_rate": 9.158090452261306e-05,
-      "loss": 5.5511,
-      "step": 8879
-    },
-    {
-      "epoch": 4.63102998696219,
-      "grad_norm": 1.5336503982543945,
-      "learning_rate": 9.157989949748744e-05,
-      "loss": 5.4364,
-      "step": 8880
-    },
-    {
-      "epoch": 4.6315514993481095,
-      "grad_norm": 1.5581824779510498,
-      "learning_rate": 9.15788944723618e-05,
-      "loss": 5.2983,
-      "step": 8881
-    },
-    {
-      "epoch": 4.632073011734029,
-      "grad_norm": 1.4625818729400635,
-      "learning_rate": 9.157788944723618e-05,
-      "loss": 5.6499,
-      "step": 8882
-    },
-    {
-      "epoch": 4.6325945241199475,
-      "grad_norm": 1.4452638626098633,
-      "learning_rate": 9.157688442211056e-05,
-      "loss": 5.5408,
-      "step": 8883
-    },
-    {
-      "epoch": 4.633116036505867,
-      "grad_norm": 1.665658950805664,
-      "learning_rate": 9.157587939698494e-05,
-      "loss": 5.1822,
-      "step": 8884
-    },
-    {
-      "epoch": 4.6336375488917865,
-      "grad_norm": 1.4525249004364014,
-      "learning_rate": 9.15748743718593e-05,
-      "loss": 5.7062,
-      "step": 8885
-    },
-    {
-      "epoch": 4.634159061277705,
-      "grad_norm": 1.691056251525879,
-      "learning_rate": 9.157386934673368e-05,
-      "loss": 5.7051,
-      "step": 8886
-    },
-    {
-      "epoch": 4.6346805736636245,
-      "grad_norm": 1.7112469673156738,
-      "learning_rate": 9.157286432160804e-05,
-      "loss": 5.4843,
-      "step": 8887
-    },
-    {
-      "epoch": 4.635202086049544,
-      "grad_norm": 1.4338489770889282,
-      "learning_rate": 9.157185929648242e-05,
-      "loss": 5.8355,
-      "step": 8888
-    },
-    {
-      "epoch": 4.6357235984354626,
-      "grad_norm": 1.7948569059371948,
-      "learning_rate": 9.157085427135678e-05,
-      "loss": 4.9228,
-      "step": 8889
-    },
-    {
-      "epoch": 4.636245110821382,
-      "grad_norm": 1.440433382987976,
-      "learning_rate": 9.156984924623116e-05,
-      "loss": 5.9302,
-      "step": 8890
-    },
-    {
-      "epoch": 4.6367666232073015,
-      "grad_norm": 1.4503041505813599,
-      "learning_rate": 9.156884422110553e-05,
-      "loss": 5.5507,
-      "step": 8891
-    },
-    {
-      "epoch": 4.63728813559322,
-      "grad_norm": 1.3866554498672485,
-      "learning_rate": 9.15678391959799e-05,
-      "loss": 5.7597,
-      "step": 8892
-    },
-    {
-      "epoch": 4.6378096479791395,
-      "grad_norm": 1.3843916654586792,
-      "learning_rate": 9.156683417085428e-05,
-      "loss": 6.0364,
-      "step": 8893
-    },
-    {
-      "epoch": 4.638331160365059,
-      "grad_norm": 1.2612018585205078,
-      "learning_rate": 9.156582914572865e-05,
-      "loss": 5.4888,
-      "step": 8894
-    },
-    {
-      "epoch": 4.638852672750978,
-      "grad_norm": 1.5215156078338623,
-      "learning_rate": 9.156482412060302e-05,
-      "loss": 5.6918,
-      "step": 8895
-    },
-    {
-      "epoch": 4.639374185136897,
-      "grad_norm": 1.3958278894424438,
-      "learning_rate": 9.156381909547739e-05,
-      "loss": 5.8621,
-      "step": 8896
-    },
-    {
-      "epoch": 4.6398956975228165,
-      "grad_norm": 1.4099003076553345,
-      "learning_rate": 9.156281407035177e-05,
-      "loss": 5.8216,
-      "step": 8897
-    },
-    {
-      "epoch": 4.640417209908735,
-      "grad_norm": 1.3947296142578125,
-      "learning_rate": 9.156180904522613e-05,
-      "loss": 5.4328,
-      "step": 8898
-    },
-    {
-      "epoch": 4.6409387222946545,
-      "grad_norm": 1.5159941911697388,
-      "learning_rate": 9.156080402010051e-05,
-      "loss": 5.9689,
-      "step": 8899
-    },
-    {
-      "epoch": 4.641460234680574,
-      "grad_norm": 1.307809829711914,
-      "learning_rate": 9.155979899497487e-05,
-      "loss": 4.9593,
-      "step": 8900
-    },
-    {
-      "epoch": 4.641981747066493,
-      "grad_norm": 1.4676752090454102,
-      "learning_rate": 9.155879396984925e-05,
-      "loss": 5.7988,
-      "step": 8901
-    },
-    {
-      "epoch": 4.642503259452412,
-      "grad_norm": 1.4932525157928467,
-      "learning_rate": 9.155778894472362e-05,
-      "loss": 5.8482,
-      "step": 8902
-    },
-    {
-      "epoch": 4.6430247718383315,
-      "grad_norm": 1.4859497547149658,
-      "learning_rate": 9.155678391959799e-05,
-      "loss": 5.5615,
-      "step": 8903
-    },
-    {
-      "epoch": 4.64354628422425,
-      "grad_norm": 1.3487786054611206,
-      "learning_rate": 9.155577889447237e-05,
-      "loss": 5.7663,
-      "step": 8904
-    },
-    {
-      "epoch": 4.6440677966101696,
-      "grad_norm": 1.3294057846069336,
-      "learning_rate": 9.155477386934674e-05,
-      "loss": 5.8078,
-      "step": 8905
-    },
-    {
-      "epoch": 4.644589308996089,
-      "grad_norm": 1.4576622247695923,
-      "learning_rate": 9.155376884422111e-05,
-      "loss": 5.7387,
-      "step": 8906
-    },
-    {
-      "epoch": 4.645110821382008,
-      "grad_norm": 1.5576660633087158,
-      "learning_rate": 9.155276381909548e-05,
-      "loss": 5.4381,
-      "step": 8907
-    },
-    {
-      "epoch": 4.645632333767927,
-      "grad_norm": 2.2306060791015625,
-      "learning_rate": 9.155175879396986e-05,
-      "loss": 5.074,
-      "step": 8908
-    },
-    {
-      "epoch": 4.6461538461538465,
-      "grad_norm": 1.3704370260238647,
-      "learning_rate": 9.155075376884422e-05,
-      "loss": 6.1558,
-      "step": 8909
-    },
-    {
-      "epoch": 4.646675358539765,
-      "grad_norm": 1.4179614782333374,
-      "learning_rate": 9.15497487437186e-05,
-      "loss": 5.7717,
-      "step": 8910
-    },
-    {
-      "epoch": 4.647196870925685,
-      "grad_norm": 1.6250797510147095,
-      "learning_rate": 9.154874371859296e-05,
-      "loss": 5.2835,
-      "step": 8911
-    },
-    {
-      "epoch": 4.647718383311604,
-      "grad_norm": 1.549228310585022,
-      "learning_rate": 9.154773869346734e-05,
-      "loss": 5.6355,
-      "step": 8912
-    },
-    {
-      "epoch": 4.648239895697523,
-      "grad_norm": 1.3285315036773682,
-      "learning_rate": 9.154673366834172e-05,
-      "loss": 5.9905,
-      "step": 8913
-    },
-    {
-      "epoch": 4.648761408083442,
-      "grad_norm": 1.4692710638046265,
-      "learning_rate": 9.15457286432161e-05,
-      "loss": 5.7967,
-      "step": 8914
-    },
-    {
-      "epoch": 4.6492829204693615,
-      "grad_norm": 1.5902204513549805,
-      "learning_rate": 9.154472361809046e-05,
-      "loss": 5.9225,
-      "step": 8915
-    },
-    {
-      "epoch": 4.64980443285528,
-      "grad_norm": 1.3674863576889038,
-      "learning_rate": 9.154371859296484e-05,
-      "loss": 5.8373,
-      "step": 8916
-    },
-    {
-      "epoch": 4.6503259452412,
-      "grad_norm": 1.371355414390564,
-      "learning_rate": 9.15427135678392e-05,
-      "loss": 5.7728,
-      "step": 8917
-    },
-    {
-      "epoch": 4.650847457627119,
-      "grad_norm": 1.4012854099273682,
-      "learning_rate": 9.154170854271357e-05,
-      "loss": 5.7148,
-      "step": 8918
-    },
-    {
-      "epoch": 4.651368970013038,
-      "grad_norm": 1.3569447994232178,
-      "learning_rate": 9.154070351758794e-05,
-      "loss": 4.7361,
-      "step": 8919
-    },
-    {
-      "epoch": 4.651890482398957,
-      "grad_norm": 1.3645433187484741,
-      "learning_rate": 9.153969849246231e-05,
-      "loss": 5.7414,
-      "step": 8920
-    },
-    {
-      "epoch": 4.652411994784876,
-      "grad_norm": 1.3499059677124023,
-      "learning_rate": 9.153869346733669e-05,
-      "loss": 5.7975,
-      "step": 8921
-    },
-    {
-      "epoch": 4.652933507170795,
-      "grad_norm": 1.4371075630187988,
-      "learning_rate": 9.153768844221105e-05,
-      "loss": 5.6486,
-      "step": 8922
-    },
-    {
-      "epoch": 4.653455019556715,
-      "grad_norm": 1.3871843814849854,
-      "learning_rate": 9.153668341708543e-05,
-      "loss": 5.5311,
-      "step": 8923
-    },
-    {
-      "epoch": 4.653976531942634,
-      "grad_norm": 1.6170580387115479,
-      "learning_rate": 9.15356783919598e-05,
-      "loss": 5.63,
-      "step": 8924
-    },
-    {
-      "epoch": 4.654498044328553,
-      "grad_norm": 1.6660054922103882,
-      "learning_rate": 9.153467336683418e-05,
-      "loss": 5.4734,
-      "step": 8925
-    },
-    {
-      "epoch": 4.655019556714472,
-      "grad_norm": 1.3537355661392212,
-      "learning_rate": 9.153366834170855e-05,
-      "loss": 5.8816,
-      "step": 8926
-    },
-    {
-      "epoch": 4.655541069100391,
-      "grad_norm": 1.499375581741333,
-      "learning_rate": 9.153266331658293e-05,
-      "loss": 5.3875,
-      "step": 8927
-    },
-    {
-      "epoch": 4.65606258148631,
-      "grad_norm": 1.3887629508972168,
-      "learning_rate": 9.153165829145729e-05,
-      "loss": 5.9481,
-      "step": 8928
-    },
-    {
-      "epoch": 4.65658409387223,
-      "grad_norm": 1.4425872564315796,
-      "learning_rate": 9.153065326633167e-05,
-      "loss": 5.814,
-      "step": 8929
-    },
-    {
-      "epoch": 4.657105606258149,
-      "grad_norm": 1.4644118547439575,
-      "learning_rate": 9.152964824120603e-05,
-      "loss": 5.7178,
-      "step": 8930
-    },
-    {
-      "epoch": 4.657627118644068,
-      "grad_norm": 1.4849790334701538,
-      "learning_rate": 9.152864321608041e-05,
-      "loss": 5.439,
-      "step": 8931
-    },
-    {
-      "epoch": 4.658148631029987,
-      "grad_norm": 1.6091728210449219,
-      "learning_rate": 9.152763819095477e-05,
-      "loss": 5.6283,
-      "step": 8932
-    },
-    {
-      "epoch": 4.658670143415906,
-      "grad_norm": 1.3358161449432373,
-      "learning_rate": 9.152663316582915e-05,
-      "loss": 5.3685,
-      "step": 8933
-    },
-    {
-      "epoch": 4.659191655801825,
-      "grad_norm": 1.5269153118133545,
-      "learning_rate": 9.152562814070353e-05,
-      "loss": 5.755,
-      "step": 8934
-    },
-    {
-      "epoch": 4.659713168187745,
-      "grad_norm": 1.4879975318908691,
-      "learning_rate": 9.15246231155779e-05,
-      "loss": 5.5222,
-      "step": 8935
-    },
-    {
-      "epoch": 4.660234680573664,
-      "grad_norm": 1.4070513248443604,
-      "learning_rate": 9.152361809045227e-05,
-      "loss": 5.5516,
-      "step": 8936
-    },
-    {
-      "epoch": 4.660756192959583,
-      "grad_norm": 1.6090805530548096,
-      "learning_rate": 9.152261306532664e-05,
-      "loss": 5.4754,
-      "step": 8937
-    },
-    {
-      "epoch": 4.661277705345502,
-      "grad_norm": 1.5029935836791992,
-      "learning_rate": 9.152160804020101e-05,
-      "loss": 5.9625,
-      "step": 8938
-    },
-    {
-      "epoch": 4.661799217731421,
-      "grad_norm": 1.3279627561569214,
-      "learning_rate": 9.152060301507538e-05,
-      "loss": 5.874,
-      "step": 8939
-    },
-    {
-      "epoch": 4.66232073011734,
-      "grad_norm": 1.3563531637191772,
-      "learning_rate": 9.151959798994976e-05,
-      "loss": 5.6903,
-      "step": 8940
-    },
-    {
-      "epoch": 4.66284224250326,
-      "grad_norm": 1.4384465217590332,
-      "learning_rate": 9.151859296482412e-05,
-      "loss": 5.433,
-      "step": 8941
-    },
-    {
-      "epoch": 4.663363754889179,
-      "grad_norm": 1.5777437686920166,
-      "learning_rate": 9.15175879396985e-05,
-      "loss": 4.9233,
-      "step": 8942
-    },
-    {
-      "epoch": 4.663885267275098,
-      "grad_norm": 1.427125334739685,
-      "learning_rate": 9.151658291457286e-05,
-      "loss": 5.771,
-      "step": 8943
-    },
-    {
-      "epoch": 4.664406779661017,
-      "grad_norm": 1.5229756832122803,
-      "learning_rate": 9.151557788944724e-05,
-      "loss": 5.6888,
-      "step": 8944
-    },
-    {
-      "epoch": 4.664928292046936,
-      "grad_norm": 1.436295986175537,
-      "learning_rate": 9.151457286432162e-05,
-      "loss": 5.5929,
-      "step": 8945
-    },
-    {
-      "epoch": 4.665449804432855,
-      "grad_norm": 1.527327060699463,
-      "learning_rate": 9.151356783919598e-05,
-      "loss": 5.4407,
-      "step": 8946
-    },
-    {
-      "epoch": 4.665971316818775,
-      "grad_norm": 1.5620461702346802,
-      "learning_rate": 9.151256281407036e-05,
-      "loss": 5.3203,
-      "step": 8947
-    },
-    {
-      "epoch": 4.666492829204694,
-      "grad_norm": 1.583417534828186,
-      "learning_rate": 9.151155778894472e-05,
-      "loss": 5.4058,
-      "step": 8948
-    },
-    {
-      "epoch": 4.667014341590613,
-      "grad_norm": 1.3752965927124023,
-      "learning_rate": 9.15105527638191e-05,
-      "loss": 5.8548,
-      "step": 8949
-    },
-    {
-      "epoch": 4.667535853976532,
-      "grad_norm": 1.4853248596191406,
-      "learning_rate": 9.150954773869347e-05,
-      "loss": 5.1245,
-      "step": 8950
-    },
-    {
-      "epoch": 4.668057366362451,
-      "grad_norm": 3.1839616298675537,
-      "learning_rate": 9.150854271356784e-05,
-      "loss": 5.4135,
-      "step": 8951
-    },
-    {
-      "epoch": 4.66857887874837,
-      "grad_norm": 1.5835926532745361,
-      "learning_rate": 9.150753768844221e-05,
-      "loss": 5.5596,
-      "step": 8952
-    },
-    {
-      "epoch": 4.66910039113429,
-      "grad_norm": 1.4251145124435425,
-      "learning_rate": 9.150653266331659e-05,
-      "loss": 5.7411,
-      "step": 8953
-    },
-    {
-      "epoch": 4.669621903520208,
-      "grad_norm": 1.4747676849365234,
-      "learning_rate": 9.150552763819096e-05,
-      "loss": 5.7814,
-      "step": 8954
-    },
-    {
-      "epoch": 4.670143415906128,
-      "grad_norm": 1.3662261962890625,
-      "learning_rate": 9.150452261306534e-05,
-      "loss": 5.9397,
-      "step": 8955
-    },
-    {
-      "epoch": 4.670664928292047,
-      "grad_norm": 1.4662872552871704,
-      "learning_rate": 9.15035175879397e-05,
-      "loss": 5.4231,
-      "step": 8956
-    },
-    {
-      "epoch": 4.671186440677966,
-      "grad_norm": 1.602906346321106,
-      "learning_rate": 9.150251256281408e-05,
-      "loss": 5.4193,
-      "step": 8957
-    },
-    {
-      "epoch": 4.671707953063885,
-      "grad_norm": 1.4468090534210205,
-      "learning_rate": 9.150150753768845e-05,
-      "loss": 5.835,
-      "step": 8958
-    },
-    {
-      "epoch": 4.672229465449805,
-      "grad_norm": 1.4304887056350708,
-      "learning_rate": 9.150050251256281e-05,
-      "loss": 5.8972,
-      "step": 8959
-    },
-    {
-      "epoch": 4.672750977835723,
-      "grad_norm": 1.3986389636993408,
-      "learning_rate": 9.149949748743719e-05,
-      "loss": 5.7505,
-      "step": 8960
-    },
-    {
-      "epoch": 4.673272490221643,
-      "grad_norm": 1.431880235671997,
-      "learning_rate": 9.149849246231155e-05,
-      "loss": 5.8482,
-      "step": 8961
-    },
-    {
-      "epoch": 4.673794002607562,
-      "grad_norm": 1.5084764957427979,
-      "learning_rate": 9.149748743718593e-05,
-      "loss": 5.967,
-      "step": 8962
-    },
-    {
-      "epoch": 4.674315514993481,
-      "grad_norm": 1.564209222793579,
-      "learning_rate": 9.14964824120603e-05,
-      "loss": 5.4923,
-      "step": 8963
-    },
-    {
-      "epoch": 4.6748370273794,
-      "grad_norm": 1.5230516195297241,
-      "learning_rate": 9.149547738693467e-05,
-      "loss": 5.8102,
-      "step": 8964
-    },
-    {
-      "epoch": 4.67535853976532,
-      "grad_norm": 1.5132273435592651,
-      "learning_rate": 9.149447236180905e-05,
-      "loss": 5.8582,
-      "step": 8965
-    },
-    {
-      "epoch": 4.675880052151238,
-      "grad_norm": 1.4988868236541748,
-      "learning_rate": 9.149346733668343e-05,
-      "loss": 5.5347,
-      "step": 8966
-    },
-    {
-      "epoch": 4.676401564537158,
-      "grad_norm": 1.3807342052459717,
-      "learning_rate": 9.14924623115578e-05,
-      "loss": 5.5683,
-      "step": 8967
-    },
-    {
-      "epoch": 4.676923076923077,
-      "grad_norm": 1.4003046751022339,
-      "learning_rate": 9.149145728643217e-05,
-      "loss": 5.6404,
-      "step": 8968
-    },
-    {
-      "epoch": 4.677444589308996,
-      "grad_norm": 1.3544319868087769,
-      "learning_rate": 9.149045226130654e-05,
-      "loss": 5.7221,
-      "step": 8969
-    },
-    {
-      "epoch": 4.677966101694915,
-      "grad_norm": 1.4137836694717407,
-      "learning_rate": 9.148944723618091e-05,
-      "loss": 5.3125,
-      "step": 8970
-    },
-    {
-      "epoch": 4.678487614080835,
-      "grad_norm": 1.5621602535247803,
-      "learning_rate": 9.148844221105528e-05,
-      "loss": 4.9161,
-      "step": 8971
-    },
-    {
-      "epoch": 4.679009126466753,
-      "grad_norm": 1.4041999578475952,
-      "learning_rate": 9.148743718592964e-05,
-      "loss": 5.8509,
-      "step": 8972
-    },
-    {
-      "epoch": 4.679530638852673,
-      "grad_norm": 1.4235516786575317,
-      "learning_rate": 9.148643216080402e-05,
-      "loss": 5.8531,
-      "step": 8973
-    },
-    {
-      "epoch": 4.680052151238592,
-      "grad_norm": 1.737185001373291,
-      "learning_rate": 9.14854271356784e-05,
-      "loss": 5.2012,
-      "step": 8974
-    },
-    {
-      "epoch": 4.680573663624511,
-      "grad_norm": 1.4435920715332031,
-      "learning_rate": 9.148442211055278e-05,
-      "loss": 5.5608,
-      "step": 8975
-    },
-    {
-      "epoch": 4.68109517601043,
-      "grad_norm": 1.4815598726272583,
-      "learning_rate": 9.148341708542714e-05,
-      "loss": 5.8218,
-      "step": 8976
-    },
-    {
-      "epoch": 4.68161668839635,
-      "grad_norm": 1.5097143650054932,
-      "learning_rate": 9.148241206030152e-05,
-      "loss": 5.7905,
-      "step": 8977
-    },
-    {
-      "epoch": 4.682138200782268,
-      "grad_norm": 1.5182691812515259,
-      "learning_rate": 9.148140703517588e-05,
-      "loss": 5.3923,
-      "step": 8978
-    },
-    {
-      "epoch": 4.682659713168188,
-      "grad_norm": 1.4539783000946045,
-      "learning_rate": 9.148040201005026e-05,
-      "loss": 5.9627,
-      "step": 8979
-    },
-    {
-      "epoch": 4.683181225554107,
-      "grad_norm": 1.6538606882095337,
-      "learning_rate": 9.147939698492462e-05,
-      "loss": 5.5894,
-      "step": 8980
-    },
-    {
-      "epoch": 4.683702737940026,
-      "grad_norm": 1.4775079488754272,
-      "learning_rate": 9.1478391959799e-05,
-      "loss": 5.809,
-      "step": 8981
-    },
-    {
-      "epoch": 4.684224250325945,
-      "grad_norm": 1.5564122200012207,
-      "learning_rate": 9.147738693467337e-05,
-      "loss": 5.5094,
-      "step": 8982
-    },
-    {
-      "epoch": 4.684745762711865,
-      "grad_norm": 1.4807054996490479,
-      "learning_rate": 9.147638190954774e-05,
-      "loss": 5.699,
-      "step": 8983
-    },
-    {
-      "epoch": 4.685267275097783,
-      "grad_norm": 1.3894282579421997,
-      "learning_rate": 9.147537688442211e-05,
-      "loss": 5.5643,
-      "step": 8984
-    },
-    {
-      "epoch": 4.685788787483703,
-      "grad_norm": 1.5078575611114502,
-      "learning_rate": 9.147437185929649e-05,
-      "loss": 5.6954,
-      "step": 8985
-    },
-    {
-      "epoch": 4.686310299869622,
-      "grad_norm": 1.5001779794692993,
-      "learning_rate": 9.147336683417086e-05,
-      "loss": 5.7835,
-      "step": 8986
-    },
-    {
-      "epoch": 4.686831812255541,
-      "grad_norm": 1.5259603261947632,
-      "learning_rate": 9.147236180904523e-05,
-      "loss": 5.5554,
-      "step": 8987
-    },
-    {
-      "epoch": 4.68735332464146,
-      "grad_norm": 1.6071741580963135,
-      "learning_rate": 9.147135678391961e-05,
-      "loss": 5.2273,
-      "step": 8988
-    },
-    {
-      "epoch": 4.68787483702738,
-      "grad_norm": 1.359210729598999,
-      "learning_rate": 9.147035175879397e-05,
-      "loss": 5.1423,
-      "step": 8989
-    },
-    {
-      "epoch": 4.688396349413298,
-      "grad_norm": 1.4546308517456055,
-      "learning_rate": 9.146934673366835e-05,
-      "loss": 5.9823,
-      "step": 8990
-    },
-    {
-      "epoch": 4.688917861799218,
-      "grad_norm": 1.4491236209869385,
-      "learning_rate": 9.146834170854271e-05,
-      "loss": 6.1331,
-      "step": 8991
-    },
-    {
-      "epoch": 4.689439374185137,
-      "grad_norm": 1.6614105701446533,
-      "learning_rate": 9.146733668341709e-05,
-      "loss": 5.7276,
-      "step": 8992
-    },
-    {
-      "epoch": 4.689960886571056,
-      "grad_norm": 1.532353401184082,
-      "learning_rate": 9.146633165829146e-05,
-      "loss": 5.66,
-      "step": 8993
-    },
-    {
-      "epoch": 4.690482398956975,
-      "grad_norm": 1.4076015949249268,
-      "learning_rate": 9.146532663316583e-05,
-      "loss": 5.6651,
-      "step": 8994
-    },
-    {
-      "epoch": 4.691003911342895,
-      "grad_norm": 1.4339251518249512,
-      "learning_rate": 9.146432160804021e-05,
-      "loss": 6.0726,
-      "step": 8995
-    },
-    {
-      "epoch": 4.691525423728813,
-      "grad_norm": 1.4163329601287842,
-      "learning_rate": 9.146331658291459e-05,
-      "loss": 5.8693,
-      "step": 8996
-    },
-    {
-      "epoch": 4.692046936114733,
-      "grad_norm": 1.3567012548446655,
-      "learning_rate": 9.146231155778895e-05,
-      "loss": 6.0816,
-      "step": 8997
-    },
-    {
-      "epoch": 4.692568448500652,
-      "grad_norm": 1.3494099378585815,
-      "learning_rate": 9.146130653266332e-05,
-      "loss": 5.7715,
-      "step": 8998
-    },
-    {
-      "epoch": 4.693089960886571,
-      "grad_norm": 1.460433840751648,
-      "learning_rate": 9.14603015075377e-05,
-      "loss": 5.6225,
-      "step": 8999
-    },
-    {
-      "epoch": 4.69361147327249,
-      "grad_norm": 1.6116814613342285,
-      "learning_rate": 9.145929648241206e-05,
-      "loss": 5.0279,
-      "step": 9000
-    },
-    {
-      "epoch": 4.69413298565841,
-      "grad_norm": 1.3569228649139404,
-      "learning_rate": 9.145829145728644e-05,
-      "loss": 5.827,
-      "step": 9001
-    },
-    {
-      "epoch": 4.694654498044328,
-      "grad_norm": 1.4466495513916016,
-      "learning_rate": 9.14572864321608e-05,
-      "loss": 5.5039,
-      "step": 9002
-    },
-    {
-      "epoch": 4.695176010430248,
-      "grad_norm": 1.4493533372879028,
-      "learning_rate": 9.145628140703518e-05,
-      "loss": 5.5406,
-      "step": 9003
-    },
-    {
-      "epoch": 4.695697522816167,
-      "grad_norm": 1.4221937656402588,
-      "learning_rate": 9.145527638190954e-05,
-      "loss": 6.0543,
-      "step": 9004
-    },
-    {
-      "epoch": 4.696219035202086,
-      "grad_norm": 1.4373024702072144,
-      "learning_rate": 9.145427135678392e-05,
-      "loss": 5.785,
-      "step": 9005
-    },
-    {
-      "epoch": 4.696740547588005,
-      "grad_norm": 1.7649263143539429,
-      "learning_rate": 9.14532663316583e-05,
-      "loss": 5.2041,
-      "step": 9006
-    },
-    {
-      "epoch": 4.697262059973925,
-      "grad_norm": 1.4852936267852783,
-      "learning_rate": 9.145226130653268e-05,
-      "loss": 5.6601,
-      "step": 9007
-    },
-    {
-      "epoch": 4.697783572359843,
-      "grad_norm": 1.399396538734436,
-      "learning_rate": 9.145125628140704e-05,
-      "loss": 5.5815,
-      "step": 9008
-    },
-    {
-      "epoch": 4.698305084745763,
-      "grad_norm": 1.3456224203109741,
-      "learning_rate": 9.145025125628142e-05,
-      "loss": 6.0985,
-      "step": 9009
-    },
-    {
-      "epoch": 4.698826597131681,
-      "grad_norm": 1.3736693859100342,
-      "learning_rate": 9.144924623115578e-05,
-      "loss": 5.9269,
-      "step": 9010
-    },
-    {
-      "epoch": 4.699348109517601,
-      "grad_norm": 1.3954074382781982,
-      "learning_rate": 9.144824120603015e-05,
-      "loss": 5.5458,
-      "step": 9011
-    },
-    {
-      "epoch": 4.69986962190352,
-      "grad_norm": 1.4896233081817627,
-      "learning_rate": 9.144723618090453e-05,
-      "loss": 5.7335,
-      "step": 9012
-    },
-    {
-      "epoch": 4.70039113428944,
-      "grad_norm": 1.7238097190856934,
-      "learning_rate": 9.144623115577889e-05,
-      "loss": 4.7773,
-      "step": 9013
-    },
-    {
-      "epoch": 4.700912646675358,
-      "grad_norm": 1.5320903062820435,
-      "learning_rate": 9.144522613065327e-05,
-      "loss": 5.4113,
-      "step": 9014
-    },
-    {
-      "epoch": 4.701434159061278,
-      "grad_norm": 1.4311516284942627,
-      "learning_rate": 9.144422110552763e-05,
-      "loss": 5.8728,
-      "step": 9015
-    },
-    {
-      "epoch": 4.701955671447196,
-      "grad_norm": 1.440740704536438,
-      "learning_rate": 9.144321608040201e-05,
-      "loss": 5.4243,
-      "step": 9016
-    },
-    {
-      "epoch": 4.702477183833116,
-      "grad_norm": 1.3550705909729004,
-      "learning_rate": 9.144221105527639e-05,
-      "loss": 5.9962,
-      "step": 9017
-    },
-    {
-      "epoch": 4.702998696219035,
-      "grad_norm": 1.640198826789856,
-      "learning_rate": 9.144120603015077e-05,
-      "loss": 5.8194,
-      "step": 9018
-    },
-    {
-      "epoch": 4.703520208604955,
-      "grad_norm": 1.4457640647888184,
-      "learning_rate": 9.144020100502513e-05,
-      "loss": 5.6678,
-      "step": 9019
-    },
-    {
-      "epoch": 4.704041720990873,
-      "grad_norm": 1.4302650690078735,
-      "learning_rate": 9.143919597989951e-05,
-      "loss": 5.6335,
-      "step": 9020
-    },
-    {
-      "epoch": 4.704563233376793,
-      "grad_norm": 1.5368174314498901,
-      "learning_rate": 9.143819095477387e-05,
-      "loss": 5.6934,
-      "step": 9021
-    },
-    {
-      "epoch": 4.705084745762711,
-      "grad_norm": 1.495213270187378,
-      "learning_rate": 9.143718592964825e-05,
-      "loss": 5.6112,
-      "step": 9022
-    },
-    {
-      "epoch": 4.705606258148631,
-      "grad_norm": 1.5291413068771362,
-      "learning_rate": 9.143618090452261e-05,
-      "loss": 5.541,
-      "step": 9023
-    },
-    {
-      "epoch": 4.70612777053455,
-      "grad_norm": 1.4981071949005127,
-      "learning_rate": 9.143517587939699e-05,
-      "loss": 5.6375,
-      "step": 9024
-    },
-    {
-      "epoch": 4.70664928292047,
-      "grad_norm": 1.413459062576294,
-      "learning_rate": 9.143417085427136e-05,
-      "loss": 5.6203,
-      "step": 9025
-    },
-    {
-      "epoch": 4.707170795306388,
-      "grad_norm": 1.4102391004562378,
-      "learning_rate": 9.143316582914573e-05,
-      "loss": 6.1645,
-      "step": 9026
-    },
-    {
-      "epoch": 4.707692307692308,
-      "grad_norm": 1.6332954168319702,
-      "learning_rate": 9.143216080402011e-05,
-      "loss": 5.2036,
-      "step": 9027
-    },
-    {
-      "epoch": 4.708213820078226,
-      "grad_norm": 1.3279831409454346,
-      "learning_rate": 9.143115577889448e-05,
-      "loss": 5.8759,
-      "step": 9028
-    },
-    {
-      "epoch": 4.708735332464146,
-      "grad_norm": 1.4790695905685425,
-      "learning_rate": 9.143015075376885e-05,
-      "loss": 5.8154,
-      "step": 9029
-    },
-    {
-      "epoch": 4.709256844850065,
-      "grad_norm": 1.5416569709777832,
-      "learning_rate": 9.142914572864322e-05,
-      "loss": 5.9505,
-      "step": 9030
-    },
-    {
-      "epoch": 4.709778357235985,
-      "grad_norm": 1.4662423133850098,
-      "learning_rate": 9.14281407035176e-05,
-      "loss": 5.7634,
-      "step": 9031
-    },
-    {
-      "epoch": 4.710299869621903,
-      "grad_norm": 1.4871242046356201,
-      "learning_rate": 9.142713567839196e-05,
-      "loss": 5.627,
-      "step": 9032
-    },
-    {
-      "epoch": 4.710821382007823,
-      "grad_norm": 2.0842692852020264,
-      "learning_rate": 9.142613065326634e-05,
-      "loss": 4.6802,
-      "step": 9033
-    },
-    {
-      "epoch": 4.711342894393741,
-      "grad_norm": 1.5658385753631592,
-      "learning_rate": 9.14251256281407e-05,
-      "loss": 5.7306,
-      "step": 9034
-    },
-    {
-      "epoch": 4.711864406779661,
-      "grad_norm": 1.3943983316421509,
-      "learning_rate": 9.142412060301508e-05,
-      "loss": 5.9888,
-      "step": 9035
-    },
-    {
-      "epoch": 4.71238591916558,
-      "grad_norm": 1.3971564769744873,
-      "learning_rate": 9.142311557788944e-05,
-      "loss": 5.5462,
-      "step": 9036
-    },
-    {
-      "epoch": 4.7129074315515,
-      "grad_norm": 1.4627493619918823,
-      "learning_rate": 9.142211055276382e-05,
-      "loss": 4.9882,
-      "step": 9037
-    },
-    {
-      "epoch": 4.713428943937418,
-      "grad_norm": 1.451422095298767,
-      "learning_rate": 9.14211055276382e-05,
-      "loss": 5.7891,
-      "step": 9038
-    },
-    {
-      "epoch": 4.713950456323338,
-      "grad_norm": 1.4828685522079468,
-      "learning_rate": 9.142010050251256e-05,
-      "loss": 6.0228,
-      "step": 9039
-    },
-    {
-      "epoch": 4.7144719687092564,
-      "grad_norm": 1.482391595840454,
-      "learning_rate": 9.141909547738694e-05,
-      "loss": 5.9859,
-      "step": 9040
-    },
-    {
-      "epoch": 4.714993481095176,
-      "grad_norm": 1.4323064088821411,
-      "learning_rate": 9.14180904522613e-05,
-      "loss": 6.0346,
-      "step": 9041
-    },
-    {
-      "epoch": 4.715514993481095,
-      "grad_norm": 1.3666317462921143,
-      "learning_rate": 9.141708542713568e-05,
-      "loss": 6.1018,
-      "step": 9042
-    },
-    {
-      "epoch": 4.716036505867015,
-      "grad_norm": 1.6140416860580444,
-      "learning_rate": 9.141608040201005e-05,
-      "loss": 5.1783,
-      "step": 9043
-    },
-    {
-      "epoch": 4.716558018252933,
-      "grad_norm": 1.3982648849487305,
-      "learning_rate": 9.141507537688443e-05,
-      "loss": 5.5526,
-      "step": 9044
-    },
-    {
-      "epoch": 4.717079530638853,
-      "grad_norm": 1.3734230995178223,
-      "learning_rate": 9.141407035175879e-05,
-      "loss": 5.9702,
-      "step": 9045
-    },
-    {
-      "epoch": 4.7176010430247715,
-      "grad_norm": 1.3606815338134766,
-      "learning_rate": 9.141306532663317e-05,
-      "loss": 5.7344,
-      "step": 9046
-    },
-    {
-      "epoch": 4.718122555410691,
-      "grad_norm": 1.4362610578536987,
-      "learning_rate": 9.141206030150755e-05,
-      "loss": 5.844,
-      "step": 9047
-    },
-    {
-      "epoch": 4.71864406779661,
-      "grad_norm": 1.369992971420288,
-      "learning_rate": 9.141105527638192e-05,
-      "loss": 6.0398,
-      "step": 9048
-    },
-    {
-      "epoch": 4.719165580182529,
-      "grad_norm": 1.417832374572754,
-      "learning_rate": 9.141005025125629e-05,
-      "loss": 5.8162,
-      "step": 9049
-    },
-    {
-      "epoch": 4.719687092568448,
-      "grad_norm": 1.4204243421554565,
-      "learning_rate": 9.140904522613067e-05,
-      "loss": 5.6441,
-      "step": 9050
-    },
-    {
-      "epoch": 4.720208604954368,
-      "grad_norm": 1.4764131307601929,
-      "learning_rate": 9.140804020100503e-05,
-      "loss": 5.8753,
-      "step": 9051
-    },
-    {
-      "epoch": 4.7207301173402865,
-      "grad_norm": 1.651686429977417,
-      "learning_rate": 9.14070351758794e-05,
-      "loss": 5.5999,
-      "step": 9052
-    },
-    {
-      "epoch": 4.721251629726206,
-      "grad_norm": 1.3869261741638184,
-      "learning_rate": 9.140603015075377e-05,
-      "loss": 5.748,
-      "step": 9053
-    },
-    {
-      "epoch": 4.721773142112125,
-      "grad_norm": 1.6321806907653809,
-      "learning_rate": 9.140502512562814e-05,
-      "loss": 5.4402,
-      "step": 9054
-    },
-    {
-      "epoch": 4.722294654498044,
-      "grad_norm": 1.4111485481262207,
-      "learning_rate": 9.140402010050251e-05,
-      "loss": 5.9797,
-      "step": 9055
-    },
-    {
-      "epoch": 4.722816166883963,
-      "grad_norm": 1.4663439989089966,
-      "learning_rate": 9.140301507537688e-05,
-      "loss": 5.4553,
-      "step": 9056
-    },
-    {
-      "epoch": 4.723337679269883,
-      "grad_norm": 1.4906105995178223,
-      "learning_rate": 9.140201005025126e-05,
-      "loss": 5.7151,
-      "step": 9057
-    },
-    {
-      "epoch": 4.7238591916558015,
-      "grad_norm": 1.602861762046814,
-      "learning_rate": 9.140100502512563e-05,
-      "loss": 5.2296,
-      "step": 9058
-    },
-    {
-      "epoch": 4.724380704041721,
-      "grad_norm": 1.5276763439178467,
-      "learning_rate": 9.140000000000001e-05,
-      "loss": 5.2685,
-      "step": 9059
-    },
-    {
-      "epoch": 4.72490221642764,
-      "grad_norm": 1.4650245904922485,
-      "learning_rate": 9.139899497487438e-05,
-      "loss": 5.307,
-      "step": 9060
-    },
-    {
-      "epoch": 4.725423728813559,
-      "grad_norm": 1.5386643409729004,
-      "learning_rate": 9.139798994974875e-05,
-      "loss": 5.2643,
-      "step": 9061
-    },
-    {
-      "epoch": 4.7259452411994785,
-      "grad_norm": 1.345570683479309,
-      "learning_rate": 9.139698492462312e-05,
-      "loss": 5.6409,
-      "step": 9062
-    },
-    {
-      "epoch": 4.726466753585398,
-      "grad_norm": 1.512941837310791,
-      "learning_rate": 9.13959798994975e-05,
-      "loss": 5.6289,
-      "step": 9063
-    },
-    {
-      "epoch": 4.7269882659713165,
-      "grad_norm": 1.60698664188385,
-      "learning_rate": 9.139497487437186e-05,
-      "loss": 5.1632,
-      "step": 9064
-    },
-    {
-      "epoch": 4.727509778357236,
-      "grad_norm": 1.3922162055969238,
-      "learning_rate": 9.139396984924623e-05,
-      "loss": 5.5216,
-      "step": 9065
-    },
-    {
-      "epoch": 4.728031290743155,
-      "grad_norm": 1.6042455434799194,
-      "learning_rate": 9.13929648241206e-05,
-      "loss": 5.5157,
-      "step": 9066
-    },
-    {
-      "epoch": 4.728552803129074,
-      "grad_norm": 1.4712001085281372,
-      "learning_rate": 9.139195979899498e-05,
-      "loss": 5.5428,
-      "step": 9067
-    },
-    {
-      "epoch": 4.7290743155149935,
-      "grad_norm": 1.8451874256134033,
-      "learning_rate": 9.139095477386936e-05,
-      "loss": 5.4963,
-      "step": 9068
-    },
-    {
-      "epoch": 4.729595827900913,
-      "grad_norm": 1.72721266746521,
-      "learning_rate": 9.138994974874372e-05,
-      "loss": 5.032,
-      "step": 9069
-    },
-    {
-      "epoch": 4.7301173402868315,
-      "grad_norm": 1.446954369544983,
-      "learning_rate": 9.13889447236181e-05,
-      "loss": 5.8181,
-      "step": 9070
-    },
-    {
-      "epoch": 4.730638852672751,
-      "grad_norm": 1.4077601432800293,
-      "learning_rate": 9.138793969849247e-05,
-      "loss": 5.7875,
-      "step": 9071
-    },
-    {
-      "epoch": 4.73116036505867,
-      "grad_norm": 1.346549153327942,
-      "learning_rate": 9.138693467336684e-05,
-      "loss": 5.9567,
-      "step": 9072
-    },
-    {
-      "epoch": 4.731681877444589,
-      "grad_norm": 1.4019945859909058,
-      "learning_rate": 9.138592964824121e-05,
-      "loss": 5.8842,
-      "step": 9073
-    },
-    {
-      "epoch": 4.7322033898305085,
-      "grad_norm": 1.4408315420150757,
-      "learning_rate": 9.138492462311559e-05,
-      "loss": 5.8469,
-      "step": 9074
-    },
-    {
-      "epoch": 4.732724902216428,
-      "grad_norm": 1.4403223991394043,
-      "learning_rate": 9.138391959798995e-05,
-      "loss": 5.9012,
-      "step": 9075
-    },
-    {
-      "epoch": 4.7332464146023465,
-      "grad_norm": 1.887406826019287,
-      "learning_rate": 9.138291457286433e-05,
-      "loss": 5.5895,
-      "step": 9076
-    },
-    {
-      "epoch": 4.733767926988266,
-      "grad_norm": 1.5437041521072388,
-      "learning_rate": 9.138190954773869e-05,
-      "loss": 5.5019,
-      "step": 9077
-    },
-    {
-      "epoch": 4.7342894393741854,
-      "grad_norm": 1.9618173837661743,
-      "learning_rate": 9.138090452261307e-05,
-      "loss": 5.4572,
-      "step": 9078
-    },
-    {
-      "epoch": 4.734810951760104,
-      "grad_norm": 1.5235689878463745,
-      "learning_rate": 9.137989949748745e-05,
-      "loss": 5.6323,
-      "step": 9079
-    },
-    {
-      "epoch": 4.7353324641460235,
-      "grad_norm": 1.4628865718841553,
-      "learning_rate": 9.137889447236181e-05,
-      "loss": 5.8498,
-      "step": 9080
-    },
-    {
-      "epoch": 4.735853976531943,
-      "grad_norm": 1.586213231086731,
-      "learning_rate": 9.137788944723619e-05,
-      "loss": 5.4729,
-      "step": 9081
-    },
-    {
-      "epoch": 4.7363754889178615,
-      "grad_norm": 1.5116747617721558,
-      "learning_rate": 9.137688442211055e-05,
-      "loss": 5.3241,
-      "step": 9082
-    },
-    {
-      "epoch": 4.736897001303781,
-      "grad_norm": 1.4192577600479126,
-      "learning_rate": 9.137587939698493e-05,
-      "loss": 5.716,
-      "step": 9083
-    },
-    {
-      "epoch": 4.7374185136897005,
-      "grad_norm": 1.6047323942184448,
-      "learning_rate": 9.13748743718593e-05,
-      "loss": 5.0733,
-      "step": 9084
-    },
-    {
-      "epoch": 4.737940026075619,
-      "grad_norm": 1.4939124584197998,
-      "learning_rate": 9.137386934673367e-05,
-      "loss": 5.2074,
-      "step": 9085
-    },
-    {
-      "epoch": 4.7384615384615385,
-      "grad_norm": 1.5527422428131104,
-      "learning_rate": 9.137286432160804e-05,
-      "loss": 5.4307,
-      "step": 9086
-    },
-    {
-      "epoch": 4.738983050847458,
-      "grad_norm": 1.457626461982727,
-      "learning_rate": 9.137185929648242e-05,
-      "loss": 5.5822,
-      "step": 9087
-    },
-    {
-      "epoch": 4.7395045632333765,
-      "grad_norm": 1.3833612203598022,
-      "learning_rate": 9.13708542713568e-05,
-      "loss": 5.5735,
-      "step": 9088
-    },
-    {
-      "epoch": 4.740026075619296,
-      "grad_norm": 1.473638653755188,
-      "learning_rate": 9.136984924623117e-05,
-      "loss": 5.451,
-      "step": 9089
-    },
-    {
-      "epoch": 4.7405475880052155,
-      "grad_norm": 1.5704928636550903,
-      "learning_rate": 9.136884422110554e-05,
-      "loss": 5.6413,
-      "step": 9090
-    },
-    {
-      "epoch": 4.741069100391134,
-      "grad_norm": 1.5840632915496826,
-      "learning_rate": 9.13678391959799e-05,
-      "loss": 5.9394,
-      "step": 9091
-    },
-    {
-      "epoch": 4.7415906127770535,
-      "grad_norm": 1.3526211977005005,
-      "learning_rate": 9.136683417085428e-05,
-      "loss": 5.6774,
-      "step": 9092
-    },
-    {
-      "epoch": 4.742112125162973,
-      "grad_norm": 1.466305136680603,
-      "learning_rate": 9.136582914572864e-05,
-      "loss": 5.4094,
-      "step": 9093
-    },
-    {
-      "epoch": 4.742633637548892,
-      "grad_norm": 1.5761760473251343,
-      "learning_rate": 9.136482412060302e-05,
-      "loss": 5.7431,
-      "step": 9094
-    },
-    {
-      "epoch": 4.743155149934811,
-      "grad_norm": 1.6857292652130127,
-      "learning_rate": 9.136381909547738e-05,
-      "loss": 5.1394,
-      "step": 9095
-    },
-    {
-      "epoch": 4.7436766623207305,
-      "grad_norm": 1.3879951238632202,
-      "learning_rate": 9.136281407035176e-05,
-      "loss": 5.8933,
-      "step": 9096
-    },
-    {
-      "epoch": 4.744198174706649,
-      "grad_norm": 1.4904520511627197,
-      "learning_rate": 9.136180904522613e-05,
-      "loss": 5.3952,
-      "step": 9097
-    },
-    {
-      "epoch": 4.7447196870925685,
-      "grad_norm": 1.4259960651397705,
-      "learning_rate": 9.13608040201005e-05,
-      "loss": 5.5144,
-      "step": 9098
-    },
-    {
-      "epoch": 4.745241199478488,
-      "grad_norm": 1.4510350227355957,
-      "learning_rate": 9.135979899497488e-05,
-      "loss": 5.2441,
-      "step": 9099
-    },
-    {
-      "epoch": 4.745762711864407,
-      "grad_norm": 1.461808443069458,
-      "learning_rate": 9.135879396984926e-05,
-      "loss": 5.5163,
-      "step": 9100
-    },
-    {
-      "epoch": 4.746284224250326,
-      "grad_norm": 1.4561563730239868,
-      "learning_rate": 9.135778894472362e-05,
-      "loss": 5.8927,
-      "step": 9101
-    },
-    {
-      "epoch": 4.7468057366362455,
-      "grad_norm": 1.5193779468536377,
-      "learning_rate": 9.1356783919598e-05,
-      "loss": 5.8097,
-      "step": 9102
-    },
-    {
-      "epoch": 4.747327249022164,
-      "grad_norm": 1.5146710872650146,
-      "learning_rate": 9.135577889447237e-05,
-      "loss": 5.5806,
-      "step": 9103
-    },
-    {
-      "epoch": 4.7478487614080835,
-      "grad_norm": 1.6041311025619507,
-      "learning_rate": 9.135477386934673e-05,
-      "loss": 5.0854,
-      "step": 9104
-    },
-    {
-      "epoch": 4.748370273794002,
-      "grad_norm": 1.437453269958496,
-      "learning_rate": 9.135376884422111e-05,
-      "loss": 5.7418,
-      "step": 9105
-    },
-    {
-      "epoch": 4.748891786179922,
-      "grad_norm": 1.6351330280303955,
-      "learning_rate": 9.135276381909547e-05,
-      "loss": 4.8252,
-      "step": 9106
-    },
-    {
-      "epoch": 4.749413298565841,
-      "grad_norm": 1.6741427183151245,
-      "learning_rate": 9.135175879396985e-05,
-      "loss": 5.8025,
-      "step": 9107
-    },
-    {
-      "epoch": 4.7499348109517605,
-      "grad_norm": 1.5103778839111328,
-      "learning_rate": 9.135075376884423e-05,
-      "loss": 5.4298,
-      "step": 9108
-    },
-    {
-      "epoch": 4.750456323337679,
-      "grad_norm": 1.3816335201263428,
-      "learning_rate": 9.13497487437186e-05,
-      "loss": 5.9762,
-      "step": 9109
-    },
-    {
-      "epoch": 4.7509778357235986,
-      "grad_norm": 1.5969825983047485,
-      "learning_rate": 9.134874371859297e-05,
-      "loss": 5.8125,
-      "step": 9110
-    },
-    {
-      "epoch": 4.751499348109517,
-      "grad_norm": 1.3938772678375244,
-      "learning_rate": 9.134773869346735e-05,
-      "loss": 5.7698,
-      "step": 9111
-    },
-    {
-      "epoch": 4.752020860495437,
-      "grad_norm": 1.5466505289077759,
-      "learning_rate": 9.134673366834171e-05,
-      "loss": 5.3483,
-      "step": 9112
-    },
-    {
-      "epoch": 4.752542372881356,
-      "grad_norm": 1.631811499595642,
-      "learning_rate": 9.134572864321609e-05,
-      "loss": 5.4662,
-      "step": 9113
-    },
-    {
-      "epoch": 4.7530638852672755,
-      "grad_norm": 1.4240127801895142,
-      "learning_rate": 9.134472361809045e-05,
-      "loss": 5.8259,
-      "step": 9114
-    },
-    {
-      "epoch": 4.753585397653194,
-      "grad_norm": 1.4549988508224487,
-      "learning_rate": 9.134371859296483e-05,
-      "loss": 5.8694,
-      "step": 9115
-    },
-    {
-      "epoch": 4.754106910039114,
-      "grad_norm": 1.4164403676986694,
-      "learning_rate": 9.13427135678392e-05,
-      "loss": 5.3762,
-      "step": 9116
-    },
-    {
-      "epoch": 4.754628422425032,
-      "grad_norm": 1.2941752672195435,
-      "learning_rate": 9.134170854271357e-05,
-      "loss": 6.1331,
-      "step": 9117
-    },
-    {
-      "epoch": 4.755149934810952,
-      "grad_norm": 1.4046121835708618,
-      "learning_rate": 9.134070351758794e-05,
-      "loss": 6.0018,
-      "step": 9118
-    },
-    {
-      "epoch": 4.755671447196871,
-      "grad_norm": 1.362645149230957,
-      "learning_rate": 9.133969849246232e-05,
-      "loss": 5.8689,
-      "step": 9119
-    },
-    {
-      "epoch": 4.7561929595827905,
-      "grad_norm": 1.3691561222076416,
-      "learning_rate": 9.13386934673367e-05,
-      "loss": 5.9532,
-      "step": 9120
-    },
-    {
-      "epoch": 4.756714471968709,
-      "grad_norm": 1.5510014295578003,
-      "learning_rate": 9.133768844221106e-05,
-      "loss": 5.2709,
-      "step": 9121
-    },
-    {
-      "epoch": 4.757235984354629,
-      "grad_norm": 1.4453661441802979,
-      "learning_rate": 9.133668341708544e-05,
-      "loss": 5.9252,
-      "step": 9122
-    },
-    {
-      "epoch": 4.757757496740547,
-      "grad_norm": 1.5274810791015625,
-      "learning_rate": 9.13356783919598e-05,
-      "loss": 5.6972,
-      "step": 9123
-    },
-    {
-      "epoch": 4.758279009126467,
-      "grad_norm": 1.6068421602249146,
-      "learning_rate": 9.133467336683418e-05,
-      "loss": 5.6297,
-      "step": 9124
-    },
-    {
-      "epoch": 4.758800521512386,
-      "grad_norm": 1.4487580060958862,
-      "learning_rate": 9.133366834170854e-05,
-      "loss": 5.8205,
-      "step": 9125
-    },
-    {
-      "epoch": 4.7593220338983055,
-      "grad_norm": 1.5535725355148315,
-      "learning_rate": 9.133266331658292e-05,
-      "loss": 5.4052,
-      "step": 9126
-    },
-    {
-      "epoch": 4.759843546284224,
-      "grad_norm": 1.5647845268249512,
-      "learning_rate": 9.133165829145728e-05,
-      "loss": 5.7759,
-      "step": 9127
-    },
-    {
-      "epoch": 4.760365058670144,
-      "grad_norm": 1.560448169708252,
-      "learning_rate": 9.133065326633166e-05,
-      "loss": 5.6621,
-      "step": 9128
-    },
-    {
-      "epoch": 4.760886571056062,
-      "grad_norm": 1.506197214126587,
-      "learning_rate": 9.132964824120604e-05,
-      "loss": 5.7506,
-      "step": 9129
-    },
-    {
-      "epoch": 4.761408083441982,
-      "grad_norm": 1.4386165142059326,
-      "learning_rate": 9.132864321608042e-05,
-      "loss": 5.5044,
-      "step": 9130
-    },
-    {
-      "epoch": 4.761929595827901,
-      "grad_norm": 1.301072359085083,
-      "learning_rate": 9.132763819095478e-05,
-      "loss": 6.0413,
-      "step": 9131
-    },
-    {
-      "epoch": 4.762451108213821,
-      "grad_norm": 1.5125846862792969,
-      "learning_rate": 9.132663316582915e-05,
-      "loss": 5.596,
-      "step": 9132
-    },
-    {
-      "epoch": 4.762972620599739,
-      "grad_norm": 1.4974647760391235,
-      "learning_rate": 9.132562814070352e-05,
-      "loss": 5.6211,
-      "step": 9133
-    },
-    {
-      "epoch": 4.763494132985659,
-      "grad_norm": 1.3536889553070068,
-      "learning_rate": 9.132462311557789e-05,
-      "loss": 5.8093,
-      "step": 9134
-    },
-    {
-      "epoch": 4.764015645371577,
-      "grad_norm": 1.540463924407959,
-      "learning_rate": 9.132361809045227e-05,
-      "loss": 5.6906,
-      "step": 9135
-    },
-    {
-      "epoch": 4.764537157757497,
-      "grad_norm": 1.395141839981079,
-      "learning_rate": 9.132261306532663e-05,
-      "loss": 5.6269,
-      "step": 9136
-    },
-    {
-      "epoch": 4.765058670143416,
-      "grad_norm": 1.4709663391113281,
-      "learning_rate": 9.132160804020101e-05,
-      "loss": 5.7312,
-      "step": 9137
-    },
-    {
-      "epoch": 4.765580182529335,
-      "grad_norm": 1.4944813251495361,
-      "learning_rate": 9.132060301507537e-05,
-      "loss": 5.2909,
-      "step": 9138
-    },
-    {
-      "epoch": 4.766101694915254,
-      "grad_norm": 1.3867850303649902,
-      "learning_rate": 9.131959798994975e-05,
-      "loss": 5.8935,
-      "step": 9139
-    },
-    {
-      "epoch": 4.766623207301174,
-      "grad_norm": 1.5259608030319214,
-      "learning_rate": 9.131859296482413e-05,
-      "loss": 5.8963,
-      "step": 9140
-    },
-    {
-      "epoch": 4.767144719687092,
-      "grad_norm": 1.3946045637130737,
-      "learning_rate": 9.13175879396985e-05,
-      "loss": 5.9652,
-      "step": 9141
-    },
-    {
-      "epoch": 4.767666232073012,
-      "grad_norm": 1.526745319366455,
-      "learning_rate": 9.131658291457287e-05,
-      "loss": 5.7687,
-      "step": 9142
-    },
-    {
-      "epoch": 4.768187744458931,
-      "grad_norm": 1.4344249963760376,
-      "learning_rate": 9.131557788944725e-05,
-      "loss": 5.5673,
-      "step": 9143
-    },
-    {
-      "epoch": 4.76870925684485,
-      "grad_norm": 1.6382249593734741,
-      "learning_rate": 9.131457286432161e-05,
-      "loss": 4.9605,
-      "step": 9144
-    },
-    {
-      "epoch": 4.769230769230769,
-      "grad_norm": 1.556731939315796,
-      "learning_rate": 9.131356783919598e-05,
-      "loss": 5.2825,
-      "step": 9145
-    },
-    {
-      "epoch": 4.769752281616689,
-      "grad_norm": 1.575476050376892,
-      "learning_rate": 9.131256281407036e-05,
-      "loss": 4.8136,
-      "step": 9146
-    },
-    {
-      "epoch": 4.770273794002607,
-      "grad_norm": 1.874936580657959,
-      "learning_rate": 9.131155778894472e-05,
-      "loss": 5.3173,
-      "step": 9147
-    },
-    {
-      "epoch": 4.770795306388527,
-      "grad_norm": 1.447564959526062,
-      "learning_rate": 9.13105527638191e-05,
-      "loss": 5.4181,
-      "step": 9148
-    },
-    {
-      "epoch": 4.771316818774446,
-      "grad_norm": 1.5838466882705688,
-      "learning_rate": 9.130954773869348e-05,
-      "loss": 5.7059,
-      "step": 9149
-    },
-    {
-      "epoch": 4.771838331160365,
-      "grad_norm": 1.4435768127441406,
-      "learning_rate": 9.130854271356785e-05,
-      "loss": 5.6171,
-      "step": 9150
-    },
-    {
-      "epoch": 4.772359843546284,
-      "grad_norm": 1.3939099311828613,
-      "learning_rate": 9.130753768844222e-05,
-      "loss": 5.7058,
-      "step": 9151
-    },
-    {
-      "epoch": 4.772881355932204,
-      "grad_norm": 1.5848551988601685,
-      "learning_rate": 9.13065326633166e-05,
-      "loss": 5.6698,
-      "step": 9152
-    },
-    {
-      "epoch": 4.773402868318122,
-      "grad_norm": 1.4105387926101685,
-      "learning_rate": 9.130552763819096e-05,
-      "loss": 5.2239,
-      "step": 9153
-    },
-    {
-      "epoch": 4.773924380704042,
-      "grad_norm": 1.502700924873352,
-      "learning_rate": 9.130452261306534e-05,
-      "loss": 5.6653,
-      "step": 9154
-    },
-    {
-      "epoch": 4.774445893089961,
-      "grad_norm": 1.4572415351867676,
-      "learning_rate": 9.13035175879397e-05,
-      "loss": 5.1227,
-      "step": 9155
-    },
-    {
-      "epoch": 4.77496740547588,
-      "grad_norm": 1.493349552154541,
-      "learning_rate": 9.130251256281408e-05,
-      "loss": 5.5554,
-      "step": 9156
-    },
-    {
-      "epoch": 4.775488917861799,
-      "grad_norm": 1.4434453248977661,
-      "learning_rate": 9.130150753768844e-05,
-      "loss": 5.5354,
-      "step": 9157
-    },
-    {
-      "epoch": 4.776010430247719,
-      "grad_norm": 1.3700617551803589,
-      "learning_rate": 9.130050251256281e-05,
-      "loss": 5.7642,
-      "step": 9158
-    },
-    {
-      "epoch": 4.776531942633637,
-      "grad_norm": 1.3702868223190308,
-      "learning_rate": 9.129949748743719e-05,
-      "loss": 5.802,
-      "step": 9159
-    },
-    {
-      "epoch": 4.777053455019557,
-      "grad_norm": 1.4306272268295288,
-      "learning_rate": 9.129849246231156e-05,
-      "loss": 5.5392,
-      "step": 9160
-    },
-    {
-      "epoch": 4.777574967405476,
-      "grad_norm": 1.3451359272003174,
-      "learning_rate": 9.129748743718594e-05,
-      "loss": 5.2147,
-      "step": 9161
-    },
-    {
-      "epoch": 4.778096479791395,
-      "grad_norm": 1.317811131477356,
-      "learning_rate": 9.12964824120603e-05,
-      "loss": 6.1418,
-      "step": 9162
-    },
-    {
-      "epoch": 4.778617992177314,
-      "grad_norm": 1.4160702228546143,
-      "learning_rate": 9.129547738693468e-05,
-      "loss": 5.3784,
-      "step": 9163
-    },
-    {
-      "epoch": 4.779139504563234,
-      "grad_norm": 1.4078781604766846,
-      "learning_rate": 9.129447236180905e-05,
-      "loss": 5.8647,
-      "step": 9164
-    },
-    {
-      "epoch": 4.779661016949152,
-      "grad_norm": 1.3747776746749878,
-      "learning_rate": 9.129346733668343e-05,
-      "loss": 5.6215,
-      "step": 9165
-    },
-    {
-      "epoch": 4.780182529335072,
-      "grad_norm": 1.5012469291687012,
-      "learning_rate": 9.129246231155779e-05,
-      "loss": 5.5002,
-      "step": 9166
-    },
-    {
-      "epoch": 4.780704041720991,
-      "grad_norm": 1.4056150913238525,
-      "learning_rate": 9.129145728643217e-05,
-      "loss": 5.6941,
-      "step": 9167
-    },
-    {
-      "epoch": 4.78122555410691,
-      "grad_norm": 1.4950807094573975,
-      "learning_rate": 9.129045226130653e-05,
-      "loss": 5.2233,
-      "step": 9168
-    },
-    {
-      "epoch": 4.781747066492829,
-      "grad_norm": 1.4580814838409424,
-      "learning_rate": 9.128944723618091e-05,
-      "loss": 5.281,
-      "step": 9169
-    },
-    {
-      "epoch": 4.782268578878749,
-      "grad_norm": 1.3682948350906372,
-      "learning_rate": 9.128844221105527e-05,
-      "loss": 6.0541,
-      "step": 9170
-    },
-    {
-      "epoch": 4.782790091264667,
-      "grad_norm": 1.407299280166626,
-      "learning_rate": 9.128743718592965e-05,
-      "loss": 5.7547,
-      "step": 9171
-    },
-    {
-      "epoch": 4.783311603650587,
-      "grad_norm": 1.4731935262680054,
-      "learning_rate": 9.128643216080403e-05,
-      "loss": 5.2749,
-      "step": 9172
-    },
-    {
-      "epoch": 4.783833116036506,
-      "grad_norm": 1.5174001455307007,
-      "learning_rate": 9.12854271356784e-05,
-      "loss": 5.5558,
-      "step": 9173
-    },
-    {
-      "epoch": 4.784354628422425,
-      "grad_norm": 1.4395952224731445,
-      "learning_rate": 9.128442211055277e-05,
-      "loss": 5.251,
-      "step": 9174
-    },
-    {
-      "epoch": 4.784876140808344,
-      "grad_norm": 1.7391411066055298,
-      "learning_rate": 9.128341708542714e-05,
-      "loss": 5.6005,
-      "step": 9175
-    },
-    {
-      "epoch": 4.785397653194264,
-      "grad_norm": 1.5495951175689697,
-      "learning_rate": 9.128241206030151e-05,
-      "loss": 5.3632,
-      "step": 9176
-    },
-    {
-      "epoch": 4.785919165580182,
-      "grad_norm": 1.6831517219543457,
-      "learning_rate": 9.128140703517588e-05,
-      "loss": 5.9957,
-      "step": 9177
-    },
-    {
-      "epoch": 4.786440677966102,
-      "grad_norm": 1.4336317777633667,
-      "learning_rate": 9.128040201005026e-05,
-      "loss": 5.7585,
-      "step": 9178
-    },
-    {
-      "epoch": 4.786962190352021,
-      "grad_norm": 1.5007797479629517,
-      "learning_rate": 9.127939698492462e-05,
-      "loss": 5.7118,
-      "step": 9179
-    },
-    {
-      "epoch": 4.78748370273794,
-      "grad_norm": 1.5858527421951294,
-      "learning_rate": 9.1278391959799e-05,
-      "loss": 5.7444,
-      "step": 9180
-    },
-    {
-      "epoch": 4.788005215123859,
-      "grad_norm": 1.4052417278289795,
-      "learning_rate": 9.127738693467338e-05,
-      "loss": 5.357,
-      "step": 9181
-    },
-    {
-      "epoch": 4.788526727509779,
-      "grad_norm": 1.5136873722076416,
-      "learning_rate": 9.127638190954775e-05,
-      "loss": 5.3587,
-      "step": 9182
-    },
-    {
-      "epoch": 4.789048239895697,
-      "grad_norm": 1.3230443000793457,
-      "learning_rate": 9.127537688442212e-05,
-      "loss": 6.1488,
-      "step": 9183
-    },
-    {
-      "epoch": 4.789569752281617,
-      "grad_norm": 1.5392048358917236,
-      "learning_rate": 9.127437185929648e-05,
-      "loss": 5.3886,
-      "step": 9184
-    },
-    {
-      "epoch": 4.790091264667536,
-      "grad_norm": 1.5167235136032104,
-      "learning_rate": 9.127336683417086e-05,
-      "loss": 5.8055,
-      "step": 9185
-    },
-    {
-      "epoch": 4.790612777053455,
-      "grad_norm": 1.4563608169555664,
-      "learning_rate": 9.127236180904522e-05,
-      "loss": 5.3622,
-      "step": 9186
-    },
-    {
-      "epoch": 4.791134289439374,
-      "grad_norm": 1.440362811088562,
-      "learning_rate": 9.12713567839196e-05,
-      "loss": 5.9062,
-      "step": 9187
-    },
-    {
-      "epoch": 4.791655801825294,
-      "grad_norm": 1.4581103324890137,
-      "learning_rate": 9.127035175879397e-05,
-      "loss": 5.3159,
-      "step": 9188
-    },
-    {
-      "epoch": 4.792177314211212,
-      "grad_norm": 1.2955031394958496,
-      "learning_rate": 9.126934673366834e-05,
-      "loss": 6.0017,
-      "step": 9189
-    },
-    {
-      "epoch": 4.792698826597132,
-      "grad_norm": 1.4146647453308105,
-      "learning_rate": 9.126834170854271e-05,
-      "loss": 5.8795,
-      "step": 9190
-    },
-    {
-      "epoch": 4.793220338983051,
-      "grad_norm": 1.3717771768569946,
-      "learning_rate": 9.126733668341709e-05,
-      "loss": 5.7957,
-      "step": 9191
-    },
-    {
-      "epoch": 4.79374185136897,
-      "grad_norm": 1.622470498085022,
-      "learning_rate": 9.126633165829146e-05,
-      "loss": 5.2551,
-      "step": 9192
-    },
-    {
-      "epoch": 4.794263363754889,
-      "grad_norm": 1.5048526525497437,
-      "learning_rate": 9.126532663316584e-05,
-      "loss": 5.5866,
-      "step": 9193
-    },
-    {
-      "epoch": 4.794784876140809,
-      "grad_norm": 1.4802837371826172,
-      "learning_rate": 9.12643216080402e-05,
-      "loss": 5.693,
-      "step": 9194
-    },
-    {
-      "epoch": 4.795306388526727,
-      "grad_norm": 1.7912784814834595,
-      "learning_rate": 9.126331658291458e-05,
-      "loss": 5.358,
-      "step": 9195
-    },
-    {
-      "epoch": 4.795827900912647,
-      "grad_norm": 1.4090813398361206,
-      "learning_rate": 9.126231155778895e-05,
-      "loss": 5.7942,
-      "step": 9196
-    },
-    {
-      "epoch": 4.796349413298566,
-      "grad_norm": 1.382284164428711,
-      "learning_rate": 9.126130653266331e-05,
-      "loss": 5.7545,
-      "step": 9197
-    },
-    {
-      "epoch": 4.796870925684485,
-      "grad_norm": 1.5322506427764893,
-      "learning_rate": 9.126030150753769e-05,
-      "loss": 5.8971,
-      "step": 9198
-    },
-    {
-      "epoch": 4.797392438070404,
-      "grad_norm": 1.4131627082824707,
-      "learning_rate": 9.125929648241205e-05,
-      "loss": 5.6542,
-      "step": 9199
-    },
-    {
-      "epoch": 4.797913950456323,
-      "grad_norm": 1.5863016843795776,
-      "learning_rate": 9.125829145728643e-05,
-      "loss": 5.6196,
-      "step": 9200
-    },
-    {
-      "epoch": 4.798435462842242,
-      "grad_norm": 1.4931647777557373,
-      "learning_rate": 9.125728643216081e-05,
-      "loss": 6.0294,
-      "step": 9201
-    },
-    {
-      "epoch": 4.798956975228162,
-      "grad_norm": 1.4778358936309814,
-      "learning_rate": 9.125628140703519e-05,
-      "loss": 5.7163,
-      "step": 9202
-    },
-    {
-      "epoch": 4.799478487614081,
-      "grad_norm": 1.4178093671798706,
-      "learning_rate": 9.125527638190955e-05,
-      "loss": 5.3801,
-      "step": 9203
-    },
-    {
-      "epoch": 4.8,
-      "grad_norm": 1.8072727918624878,
-      "learning_rate": 9.125427135678393e-05,
-      "loss": 5.4713,
-      "step": 9204
-    },
-    {
-      "epoch": 4.800521512385919,
-      "grad_norm": 1.5860620737075806,
-      "learning_rate": 9.12532663316583e-05,
-      "loss": 5.6139,
-      "step": 9205
-    },
-    {
-      "epoch": 4.801043024771838,
-      "grad_norm": 1.501033067703247,
-      "learning_rate": 9.125226130653267e-05,
-      "loss": 4.7683,
-      "step": 9206
-    },
-    {
-      "epoch": 4.801564537157757,
-      "grad_norm": 1.4287185668945312,
-      "learning_rate": 9.125125628140704e-05,
-      "loss": 5.8138,
-      "step": 9207
-    },
-    {
-      "epoch": 4.802086049543677,
-      "grad_norm": 1.7251499891281128,
-      "learning_rate": 9.125025125628141e-05,
-      "loss": 5.6245,
-      "step": 9208
-    },
-    {
-      "epoch": 4.802607561929596,
-      "grad_norm": 1.5832972526550293,
-      "learning_rate": 9.124924623115578e-05,
-      "loss": 5.4532,
-      "step": 9209
-    },
-    {
-      "epoch": 4.803129074315515,
-      "grad_norm": 1.5163676738739014,
-      "learning_rate": 9.124824120603016e-05,
-      "loss": 5.6448,
-      "step": 9210
-    },
-    {
-      "epoch": 4.803650586701434,
-      "grad_norm": 1.4867360591888428,
-      "learning_rate": 9.124723618090452e-05,
-      "loss": 5.5087,
-      "step": 9211
-    },
-    {
-      "epoch": 4.804172099087353,
-      "grad_norm": 1.4080156087875366,
-      "learning_rate": 9.12462311557789e-05,
-      "loss": 6.0164,
-      "step": 9212
-    },
-    {
-      "epoch": 4.804693611473272,
-      "grad_norm": 1.5526832342147827,
-      "learning_rate": 9.124522613065328e-05,
-      "loss": 5.2943,
-      "step": 9213
-    },
-    {
-      "epoch": 4.805215123859192,
-      "grad_norm": 1.5148519277572632,
-      "learning_rate": 9.124422110552764e-05,
-      "loss": 5.6564,
-      "step": 9214
-    },
-    {
-      "epoch": 4.805736636245111,
-      "grad_norm": 1.4474546909332275,
-      "learning_rate": 9.124321608040202e-05,
-      "loss": 5.8948,
-      "step": 9215
-    },
-    {
-      "epoch": 4.80625814863103,
-      "grad_norm": 1.3023163080215454,
-      "learning_rate": 9.124221105527638e-05,
-      "loss": 5.7305,
-      "step": 9216
-    },
-    {
-      "epoch": 4.806779661016949,
-      "grad_norm": 1.5272940397262573,
-      "learning_rate": 9.124120603015076e-05,
-      "loss": 5.7529,
-      "step": 9217
-    },
-    {
-      "epoch": 4.807301173402868,
-      "grad_norm": 1.4714819192886353,
-      "learning_rate": 9.124020100502513e-05,
-      "loss": 5.346,
-      "step": 9218
-    },
-    {
-      "epoch": 4.807822685788787,
-      "grad_norm": 1.4704169034957886,
-      "learning_rate": 9.12391959798995e-05,
-      "loss": 5.4741,
-      "step": 9219
-    },
-    {
-      "epoch": 4.808344198174707,
-      "grad_norm": 1.4541088342666626,
-      "learning_rate": 9.123819095477387e-05,
-      "loss": 5.8416,
-      "step": 9220
-    },
-    {
-      "epoch": 4.808865710560626,
-      "grad_norm": 1.5072027444839478,
-      "learning_rate": 9.123718592964824e-05,
-      "loss": 5.5403,
-      "step": 9221
-    },
-    {
-      "epoch": 4.809387222946545,
-      "grad_norm": 1.4763387441635132,
-      "learning_rate": 9.123618090452262e-05,
-      "loss": 5.4736,
-      "step": 9222
-    },
-    {
-      "epoch": 4.809908735332464,
-      "grad_norm": 1.4348726272583008,
-      "learning_rate": 9.1235175879397e-05,
-      "loss": 5.9259,
-      "step": 9223
-    },
-    {
-      "epoch": 4.810430247718383,
-      "grad_norm": 1.6023937463760376,
-      "learning_rate": 9.123417085427136e-05,
-      "loss": 4.8505,
-      "step": 9224
-    },
-    {
-      "epoch": 4.810951760104302,
-      "grad_norm": 1.8095829486846924,
-      "learning_rate": 9.123316582914573e-05,
-      "loss": 5.9335,
-      "step": 9225
-    },
-    {
-      "epoch": 4.811473272490222,
-      "grad_norm": 1.3951047658920288,
-      "learning_rate": 9.123216080402011e-05,
-      "loss": 5.8536,
-      "step": 9226
-    },
-    {
-      "epoch": 4.811994784876141,
-      "grad_norm": 1.4968910217285156,
-      "learning_rate": 9.123115577889447e-05,
-      "loss": 5.4357,
-      "step": 9227
-    },
-    {
-      "epoch": 4.81251629726206,
-      "grad_norm": 1.387775182723999,
-      "learning_rate": 9.123015075376885e-05,
-      "loss": 5.4818,
-      "step": 9228
-    },
-    {
-      "epoch": 4.813037809647979,
-      "grad_norm": 1.4894225597381592,
-      "learning_rate": 9.122914572864321e-05,
-      "loss": 5.8268,
-      "step": 9229
-    },
-    {
-      "epoch": 4.813559322033898,
-      "grad_norm": 1.4992468357086182,
-      "learning_rate": 9.122814070351759e-05,
-      "loss": 5.8371,
-      "step": 9230
-    },
-    {
-      "epoch": 4.814080834419817,
-      "grad_norm": 1.5869593620300293,
-      "learning_rate": 9.122713567839196e-05,
-      "loss": 4.9458,
-      "step": 9231
-    },
-    {
-      "epoch": 4.814602346805737,
-      "grad_norm": 1.4982163906097412,
-      "learning_rate": 9.122613065326633e-05,
-      "loss": 5.723,
-      "step": 9232
-    },
-    {
-      "epoch": 4.815123859191655,
-      "grad_norm": 1.4194111824035645,
-      "learning_rate": 9.122512562814071e-05,
-      "loss": 5.6377,
-      "step": 9233
-    },
-    {
-      "epoch": 4.815645371577575,
-      "grad_norm": 1.5596648454666138,
-      "learning_rate": 9.122412060301509e-05,
-      "loss": 5.6573,
-      "step": 9234
-    },
-    {
-      "epoch": 4.816166883963494,
-      "grad_norm": 1.322155475616455,
-      "learning_rate": 9.122311557788945e-05,
-      "loss": 5.9455,
-      "step": 9235
-    },
-    {
-      "epoch": 4.816688396349413,
-      "grad_norm": 1.3279646635055542,
-      "learning_rate": 9.122211055276383e-05,
-      "loss": 6.0966,
-      "step": 9236
-    },
-    {
-      "epoch": 4.817209908735332,
-      "grad_norm": 1.362428069114685,
-      "learning_rate": 9.12211055276382e-05,
-      "loss": 5.7814,
-      "step": 9237
-    },
-    {
-      "epoch": 4.817731421121252,
-      "grad_norm": 1.569911003112793,
-      "learning_rate": 9.122010050251256e-05,
-      "loss": 5.4444,
-      "step": 9238
-    },
-    {
-      "epoch": 4.81825293350717,
-      "grad_norm": 1.5388773679733276,
-      "learning_rate": 9.121909547738694e-05,
-      "loss": 5.5494,
-      "step": 9239
-    },
-    {
-      "epoch": 4.81877444589309,
-      "grad_norm": 1.5827172994613647,
-      "learning_rate": 9.12180904522613e-05,
-      "loss": 5.864,
-      "step": 9240
-    },
-    {
-      "epoch": 4.819295958279009,
-      "grad_norm": 1.4343013763427734,
-      "learning_rate": 9.121708542713568e-05,
-      "loss": 5.7891,
-      "step": 9241
-    },
-    {
-      "epoch": 4.819817470664928,
-      "grad_norm": 1.4862569570541382,
-      "learning_rate": 9.121608040201006e-05,
-      "loss": 5.3462,
-      "step": 9242
-    },
-    {
-      "epoch": 4.820338983050847,
-      "grad_norm": 1.3925949335098267,
-      "learning_rate": 9.121507537688444e-05,
-      "loss": 6.0967,
-      "step": 9243
-    },
-    {
-      "epoch": 4.820860495436767,
-      "grad_norm": 1.4368908405303955,
-      "learning_rate": 9.12140703517588e-05,
-      "loss": 5.8037,
-      "step": 9244
-    },
-    {
-      "epoch": 4.8213820078226854,
-      "grad_norm": 1.5739686489105225,
-      "learning_rate": 9.121306532663318e-05,
-      "loss": 5.8798,
-      "step": 9245
-    },
-    {
-      "epoch": 4.821903520208605,
-      "grad_norm": 1.6256200075149536,
-      "learning_rate": 9.121206030150754e-05,
-      "loss": 5.1229,
-      "step": 9246
-    },
-    {
-      "epoch": 4.822425032594524,
-      "grad_norm": 1.5344382524490356,
-      "learning_rate": 9.121105527638192e-05,
-      "loss": 5.7012,
-      "step": 9247
-    },
-    {
-      "epoch": 4.822946544980443,
-      "grad_norm": 1.5843276977539062,
-      "learning_rate": 9.121005025125628e-05,
-      "loss": 5.7181,
-      "step": 9248
-    },
-    {
-      "epoch": 4.823468057366362,
-      "grad_norm": 1.548142910003662,
-      "learning_rate": 9.120904522613066e-05,
-      "loss": 5.4549,
-      "step": 9249
-    },
-    {
-      "epoch": 4.823989569752282,
-      "grad_norm": 1.7478749752044678,
-      "learning_rate": 9.120804020100503e-05,
-      "loss": 5.6083,
-      "step": 9250
-    },
-    {
-      "epoch": 4.8245110821382005,
-      "grad_norm": 1.361341118812561,
-      "learning_rate": 9.120703517587939e-05,
-      "loss": 5.7902,
-      "step": 9251
-    },
-    {
-      "epoch": 4.82503259452412,
-      "grad_norm": 1.5743964910507202,
-      "learning_rate": 9.120603015075377e-05,
-      "loss": 5.2731,
-      "step": 9252
-    },
-    {
-      "epoch": 4.825554106910039,
-      "grad_norm": 1.4183433055877686,
-      "learning_rate": 9.120502512562815e-05,
-      "loss": 5.6402,
-      "step": 9253
-    },
-    {
-      "epoch": 4.826075619295958,
-      "grad_norm": 1.4950519800186157,
-      "learning_rate": 9.120402010050252e-05,
-      "loss": 5.582,
-      "step": 9254
-    },
-    {
-      "epoch": 4.826597131681877,
-      "grad_norm": 1.4537906646728516,
-      "learning_rate": 9.120301507537689e-05,
-      "loss": 5.5134,
-      "step": 9255
-    },
-    {
-      "epoch": 4.827118644067797,
-      "grad_norm": 1.5480533838272095,
-      "learning_rate": 9.120201005025127e-05,
-      "loss": 5.4403,
-      "step": 9256
-    },
-    {
-      "epoch": 4.8276401564537155,
-      "grad_norm": 1.5023926496505737,
-      "learning_rate": 9.120100502512563e-05,
-      "loss": 5.7625,
-      "step": 9257
-    },
-    {
-      "epoch": 4.828161668839635,
-      "grad_norm": 1.6865278482437134,
-      "learning_rate": 9.120000000000001e-05,
-      "loss": 5.5276,
-      "step": 9258
-    },
-    {
-      "epoch": 4.828683181225554,
-      "grad_norm": 1.6022428274154663,
-      "learning_rate": 9.119899497487437e-05,
-      "loss": 5.5368,
-      "step": 9259
-    },
-    {
-      "epoch": 4.829204693611473,
-      "grad_norm": 1.91963529586792,
-      "learning_rate": 9.119798994974875e-05,
-      "loss": 5.4684,
-      "step": 9260
-    },
-    {
-      "epoch": 4.829726205997392,
-      "grad_norm": 1.5416866540908813,
-      "learning_rate": 9.119698492462311e-05,
-      "loss": 5.3312,
-      "step": 9261
-    },
-    {
-      "epoch": 4.830247718383312,
-      "grad_norm": 1.6617872714996338,
-      "learning_rate": 9.119597989949749e-05,
-      "loss": 5.896,
-      "step": 9262
-    },
-    {
-      "epoch": 4.8307692307692305,
-      "grad_norm": 1.4679453372955322,
-      "learning_rate": 9.119497487437187e-05,
-      "loss": 5.7744,
-      "step": 9263
-    },
-    {
-      "epoch": 4.83129074315515,
-      "grad_norm": 1.3962866067886353,
-      "learning_rate": 9.119396984924623e-05,
-      "loss": 5.3435,
-      "step": 9264
-    },
-    {
-      "epoch": 4.831812255541069,
-      "grad_norm": 1.4566397666931152,
-      "learning_rate": 9.119296482412061e-05,
-      "loss": 5.7771,
-      "step": 9265
-    },
-    {
-      "epoch": 4.832333767926988,
-      "grad_norm": 1.4833298921585083,
-      "learning_rate": 9.119195979899498e-05,
-      "loss": 5.5102,
-      "step": 9266
-    },
-    {
-      "epoch": 4.8328552803129075,
-      "grad_norm": 1.5184600353240967,
-      "learning_rate": 9.119095477386935e-05,
-      "loss": 5.7137,
-      "step": 9267
-    },
-    {
-      "epoch": 4.833376792698827,
-      "grad_norm": 1.5187389850616455,
-      "learning_rate": 9.118994974874372e-05,
-      "loss": 5.3414,
-      "step": 9268
-    },
-    {
-      "epoch": 4.8338983050847455,
-      "grad_norm": 1.3666802644729614,
-      "learning_rate": 9.11889447236181e-05,
-      "loss": 5.8923,
-      "step": 9269
-    },
-    {
-      "epoch": 4.834419817470665,
-      "grad_norm": 1.3573850393295288,
-      "learning_rate": 9.118793969849246e-05,
-      "loss": 5.6331,
-      "step": 9270
-    },
-    {
-      "epoch": 4.834941329856584,
-      "grad_norm": 1.4911201000213623,
-      "learning_rate": 9.118693467336684e-05,
-      "loss": 5.5722,
-      "step": 9271
-    },
-    {
-      "epoch": 4.835462842242503,
-      "grad_norm": 1.3657877445220947,
-      "learning_rate": 9.11859296482412e-05,
-      "loss": 5.4478,
-      "step": 9272
-    },
-    {
-      "epoch": 4.8359843546284225,
-      "grad_norm": 1.3521032333374023,
-      "learning_rate": 9.118492462311558e-05,
-      "loss": 5.6861,
-      "step": 9273
-    },
-    {
-      "epoch": 4.836505867014342,
-      "grad_norm": 1.304948329925537,
-      "learning_rate": 9.118391959798996e-05,
-      "loss": 5.6432,
-      "step": 9274
-    },
-    {
-      "epoch": 4.8370273794002605,
-      "grad_norm": 1.4321249723434448,
-      "learning_rate": 9.118291457286434e-05,
-      "loss": 5.6818,
-      "step": 9275
-    },
-    {
-      "epoch": 4.83754889178618,
-      "grad_norm": 1.5321964025497437,
-      "learning_rate": 9.11819095477387e-05,
-      "loss": 5.8486,
-      "step": 9276
-    },
-    {
-      "epoch": 4.838070404172099,
-      "grad_norm": 1.4128707647323608,
-      "learning_rate": 9.118090452261306e-05,
-      "loss": 5.5959,
-      "step": 9277
-    },
-    {
-      "epoch": 4.838591916558018,
-      "grad_norm": 2.048426866531372,
-      "learning_rate": 9.117989949748744e-05,
-      "loss": 5.5242,
-      "step": 9278
-    },
-    {
-      "epoch": 4.8391134289439375,
-      "grad_norm": 1.550309658050537,
-      "learning_rate": 9.11788944723618e-05,
-      "loss": 5.7066,
-      "step": 9279
-    },
-    {
-      "epoch": 4.839634941329857,
-      "grad_norm": 1.4226003885269165,
-      "learning_rate": 9.117788944723618e-05,
-      "loss": 5.7614,
-      "step": 9280
-    },
-    {
-      "epoch": 4.8401564537157755,
-      "grad_norm": 1.741803765296936,
-      "learning_rate": 9.117688442211055e-05,
-      "loss": 5.2972,
-      "step": 9281
-    },
-    {
-      "epoch": 4.840677966101695,
-      "grad_norm": 1.4007216691970825,
-      "learning_rate": 9.117587939698493e-05,
-      "loss": 5.9674,
-      "step": 9282
-    },
-    {
-      "epoch": 4.8411994784876144,
-      "grad_norm": 1.3628923892974854,
-      "learning_rate": 9.11748743718593e-05,
-      "loss": 6.1243,
-      "step": 9283
-    },
-    {
-      "epoch": 4.841720990873533,
-      "grad_norm": 1.3534587621688843,
-      "learning_rate": 9.117386934673368e-05,
-      "loss": 5.9117,
-      "step": 9284
-    },
-    {
-      "epoch": 4.8422425032594525,
-      "grad_norm": 1.483657717704773,
-      "learning_rate": 9.117286432160805e-05,
-      "loss": 5.7305,
-      "step": 9285
-    },
-    {
-      "epoch": 4.842764015645372,
-      "grad_norm": 1.5553796291351318,
-      "learning_rate": 9.117185929648242e-05,
-      "loss": 5.4712,
-      "step": 9286
-    },
-    {
-      "epoch": 4.8432855280312905,
-      "grad_norm": 1.3317629098892212,
-      "learning_rate": 9.117085427135679e-05,
-      "loss": 5.9922,
-      "step": 9287
-    },
-    {
-      "epoch": 4.84380704041721,
-      "grad_norm": 1.3248112201690674,
-      "learning_rate": 9.116984924623117e-05,
-      "loss": 6.1447,
-      "step": 9288
-    },
-    {
-      "epoch": 4.8443285528031295,
-      "grad_norm": 1.4562768936157227,
-      "learning_rate": 9.116884422110553e-05,
-      "loss": 5.5182,
-      "step": 9289
-    },
-    {
-      "epoch": 4.844850065189048,
-      "grad_norm": 1.4521560668945312,
-      "learning_rate": 9.116783919597991e-05,
-      "loss": 5.3681,
-      "step": 9290
-    },
-    {
-      "epoch": 4.8453715775749675,
-      "grad_norm": 1.3736523389816284,
-      "learning_rate": 9.116683417085427e-05,
-      "loss": 5.6564,
-      "step": 9291
-    },
-    {
-      "epoch": 4.845893089960887,
-      "grad_norm": 1.459989070892334,
-      "learning_rate": 9.116582914572864e-05,
-      "loss": 5.9534,
-      "step": 9292
-    },
-    {
-      "epoch": 4.8464146023468055,
-      "grad_norm": 1.6139153242111206,
-      "learning_rate": 9.116482412060301e-05,
-      "loss": 5.7068,
-      "step": 9293
-    },
-    {
-      "epoch": 4.846936114732725,
-      "grad_norm": 1.4541608095169067,
-      "learning_rate": 9.116381909547739e-05,
-      "loss": 5.6516,
-      "step": 9294
-    },
-    {
-      "epoch": 4.847457627118644,
-      "grad_norm": 1.2638553380966187,
-      "learning_rate": 9.116281407035177e-05,
-      "loss": 6.0052,
-      "step": 9295
-    },
-    {
-      "epoch": 4.847979139504563,
-      "grad_norm": 1.4266360998153687,
-      "learning_rate": 9.116180904522613e-05,
-      "loss": 5.789,
-      "step": 9296
-    },
-    {
-      "epoch": 4.8485006518904825,
-      "grad_norm": 1.350271224975586,
-      "learning_rate": 9.116080402010051e-05,
-      "loss": 5.9609,
-      "step": 9297
-    },
-    {
-      "epoch": 4.849022164276402,
-      "grad_norm": 1.438230276107788,
-      "learning_rate": 9.115979899497488e-05,
-      "loss": 5.6213,
-      "step": 9298
-    },
-    {
-      "epoch": 4.849543676662321,
-      "grad_norm": 1.4094445705413818,
-      "learning_rate": 9.115879396984925e-05,
-      "loss": 5.4107,
-      "step": 9299
-    },
-    {
-      "epoch": 4.85006518904824,
-      "grad_norm": 1.4485507011413574,
-      "learning_rate": 9.115778894472362e-05,
-      "loss": 5.5516,
-      "step": 9300
-    },
-    {
-      "epoch": 4.850586701434159,
-      "grad_norm": 1.4261964559555054,
-      "learning_rate": 9.1156783919598e-05,
-      "loss": 6.0503,
-      "step": 9301
-    },
-    {
-      "epoch": 4.851108213820078,
-      "grad_norm": 1.504470944404602,
-      "learning_rate": 9.115577889447236e-05,
-      "loss": 5.4479,
-      "step": 9302
-    },
-    {
-      "epoch": 4.8516297262059975,
-      "grad_norm": 1.4866969585418701,
-      "learning_rate": 9.115477386934674e-05,
-      "loss": 5.2855,
-      "step": 9303
-    },
-    {
-      "epoch": 4.852151238591917,
-      "grad_norm": 1.4826127290725708,
-      "learning_rate": 9.115376884422112e-05,
-      "loss": 5.4432,
-      "step": 9304
-    },
-    {
-      "epoch": 4.852672750977836,
-      "grad_norm": 1.52995765209198,
-      "learning_rate": 9.115276381909548e-05,
-      "loss": 5.7464,
-      "step": 9305
-    },
-    {
-      "epoch": 4.853194263363755,
-      "grad_norm": 1.4683867692947388,
-      "learning_rate": 9.115175879396986e-05,
-      "loss": 5.8766,
-      "step": 9306
-    },
-    {
-      "epoch": 4.853715775749674,
-      "grad_norm": 1.3300960063934326,
-      "learning_rate": 9.115075376884422e-05,
-      "loss": 5.5447,
-      "step": 9307
-    },
-    {
-      "epoch": 4.854237288135593,
-      "grad_norm": 1.4404042959213257,
-      "learning_rate": 9.11497487437186e-05,
-      "loss": 5.1195,
-      "step": 9308
-    },
-    {
-      "epoch": 4.8547588005215125,
-      "grad_norm": 1.4807119369506836,
-      "learning_rate": 9.114874371859297e-05,
-      "loss": 5.5349,
-      "step": 9309
-    },
-    {
-      "epoch": 4.855280312907432,
-      "grad_norm": 1.3432385921478271,
-      "learning_rate": 9.114773869346734e-05,
-      "loss": 5.6125,
-      "step": 9310
-    },
-    {
-      "epoch": 4.855801825293351,
-      "grad_norm": 1.5868613719940186,
-      "learning_rate": 9.114673366834171e-05,
-      "loss": 5.7365,
-      "step": 9311
-    },
-    {
-      "epoch": 4.85632333767927,
-      "grad_norm": 1.4125149250030518,
-      "learning_rate": 9.114572864321609e-05,
-      "loss": 6.0113,
-      "step": 9312
-    },
-    {
-      "epoch": 4.856844850065189,
-      "grad_norm": 1.4138057231903076,
-      "learning_rate": 9.114472361809045e-05,
-      "loss": 5.6419,
-      "step": 9313
-    },
-    {
-      "epoch": 4.857366362451108,
-      "grad_norm": 1.447916865348816,
-      "learning_rate": 9.114371859296483e-05,
-      "loss": 5.9331,
-      "step": 9314
-    },
-    {
-      "epoch": 4.8578878748370276,
-      "grad_norm": 1.5571261644363403,
-      "learning_rate": 9.11427135678392e-05,
-      "loss": 5.2726,
-      "step": 9315
-    },
-    {
-      "epoch": 4.858409387222947,
-      "grad_norm": 1.4460302591323853,
-      "learning_rate": 9.114170854271358e-05,
-      "loss": 5.947,
-      "step": 9316
-    },
-    {
-      "epoch": 4.858930899608866,
-      "grad_norm": 1.3093674182891846,
-      "learning_rate": 9.114070351758795e-05,
-      "loss": 6.1393,
-      "step": 9317
-    },
-    {
-      "epoch": 4.859452411994785,
-      "grad_norm": 1.6715153455734253,
-      "learning_rate": 9.113969849246231e-05,
-      "loss": 5.3495,
-      "step": 9318
-    },
-    {
-      "epoch": 4.859973924380704,
-      "grad_norm": 1.5444271564483643,
-      "learning_rate": 9.113869346733669e-05,
-      "loss": 5.5903,
-      "step": 9319
-    },
-    {
-      "epoch": 4.860495436766623,
-      "grad_norm": 1.4306477308273315,
-      "learning_rate": 9.113768844221105e-05,
-      "loss": 5.575,
-      "step": 9320
-    },
-    {
-      "epoch": 4.861016949152543,
-      "grad_norm": 1.4778169393539429,
-      "learning_rate": 9.113668341708543e-05,
-      "loss": 5.5836,
-      "step": 9321
-    },
-    {
-      "epoch": 4.861538461538462,
-      "grad_norm": 1.461381435394287,
-      "learning_rate": 9.11356783919598e-05,
-      "loss": 5.5211,
-      "step": 9322
-    },
-    {
-      "epoch": 4.862059973924381,
-      "grad_norm": 1.4051614999771118,
-      "learning_rate": 9.113467336683417e-05,
-      "loss": 5.655,
-      "step": 9323
-    },
-    {
-      "epoch": 4.8625814863103,
-      "grad_norm": 1.4780243635177612,
-      "learning_rate": 9.113366834170855e-05,
-      "loss": 5.7677,
-      "step": 9324
-    },
-    {
-      "epoch": 4.863102998696219,
-      "grad_norm": 1.3828657865524292,
-      "learning_rate": 9.113266331658293e-05,
-      "loss": 5.7591,
-      "step": 9325
-    },
-    {
-      "epoch": 4.863624511082138,
-      "grad_norm": 1.5801982879638672,
-      "learning_rate": 9.11316582914573e-05,
-      "loss": 5.6065,
-      "step": 9326
-    },
-    {
-      "epoch": 4.864146023468058,
-      "grad_norm": 1.4674339294433594,
-      "learning_rate": 9.113065326633167e-05,
-      "loss": 5.8943,
-      "step": 9327
-    },
-    {
-      "epoch": 4.864667535853976,
-      "grad_norm": 1.3581868410110474,
-      "learning_rate": 9.112964824120604e-05,
-      "loss": 5.888,
-      "step": 9328
-    },
-    {
-      "epoch": 4.865189048239896,
-      "grad_norm": 1.3303418159484863,
-      "learning_rate": 9.112864321608041e-05,
-      "loss": 5.863,
-      "step": 9329
-    },
-    {
-      "epoch": 4.865710560625815,
-      "grad_norm": 1.5292084217071533,
-      "learning_rate": 9.112763819095478e-05,
-      "loss": 5.8049,
-      "step": 9330
-    },
-    {
-      "epoch": 4.866232073011734,
-      "grad_norm": 1.300917387008667,
-      "learning_rate": 9.112663316582914e-05,
-      "loss": 6.0319,
-      "step": 9331
-    },
-    {
-      "epoch": 4.866753585397653,
-      "grad_norm": 1.3341732025146484,
-      "learning_rate": 9.112562814070352e-05,
-      "loss": 6.0431,
-      "step": 9332
-    },
-    {
-      "epoch": 4.867275097783573,
-      "grad_norm": 1.3768621683120728,
-      "learning_rate": 9.112462311557788e-05,
-      "loss": 5.5139,
-      "step": 9333
-    },
-    {
-      "epoch": 4.867796610169491,
-      "grad_norm": 1.4998787641525269,
-      "learning_rate": 9.112361809045226e-05,
-      "loss": 6.0973,
-      "step": 9334
-    },
-    {
-      "epoch": 4.868318122555411,
-      "grad_norm": 1.4189060926437378,
-      "learning_rate": 9.112261306532664e-05,
-      "loss": 4.8085,
-      "step": 9335
-    },
-    {
-      "epoch": 4.86883963494133,
-      "grad_norm": 1.3150265216827393,
-      "learning_rate": 9.112160804020102e-05,
-      "loss": 5.7908,
-      "step": 9336
-    },
-    {
-      "epoch": 4.869361147327249,
-      "grad_norm": 1.3877546787261963,
-      "learning_rate": 9.112060301507538e-05,
-      "loss": 5.6236,
-      "step": 9337
-    },
-    {
-      "epoch": 4.869882659713168,
-      "grad_norm": 1.3660801649093628,
-      "learning_rate": 9.111959798994976e-05,
-      "loss": 5.6533,
-      "step": 9338
-    },
-    {
-      "epoch": 4.870404172099088,
-      "grad_norm": 1.298844337463379,
-      "learning_rate": 9.111859296482412e-05,
-      "loss": 5.686,
-      "step": 9339
-    },
-    {
-      "epoch": 4.870925684485006,
-      "grad_norm": 1.3915958404541016,
-      "learning_rate": 9.11175879396985e-05,
-      "loss": 5.6653,
-      "step": 9340
-    },
-    {
-      "epoch": 4.871447196870926,
-      "grad_norm": 1.4473007917404175,
-      "learning_rate": 9.111658291457287e-05,
-      "loss": 5.9343,
-      "step": 9341
-    },
-    {
-      "epoch": 4.871968709256845,
-      "grad_norm": 1.57075834274292,
-      "learning_rate": 9.111557788944724e-05,
-      "loss": 5.1206,
-      "step": 9342
-    },
-    {
-      "epoch": 4.872490221642764,
-      "grad_norm": 1.387607216835022,
-      "learning_rate": 9.111457286432161e-05,
-      "loss": 5.8418,
-      "step": 9343
-    },
-    {
-      "epoch": 4.873011734028683,
-      "grad_norm": 1.4739086627960205,
-      "learning_rate": 9.111356783919597e-05,
-      "loss": 5.7814,
-      "step": 9344
-    },
-    {
-      "epoch": 4.873533246414603,
-      "grad_norm": 1.4562593698501587,
-      "learning_rate": 9.111256281407035e-05,
-      "loss": 5.3762,
-      "step": 9345
-    },
-    {
-      "epoch": 4.874054758800521,
-      "grad_norm": 1.6553137302398682,
-      "learning_rate": 9.111155778894473e-05,
-      "loss": 4.976,
-      "step": 9346
-    },
-    {
-      "epoch": 4.874576271186441,
-      "grad_norm": 1.7254507541656494,
-      "learning_rate": 9.11105527638191e-05,
-      "loss": 5.4907,
-      "step": 9347
-    },
-    {
-      "epoch": 4.87509778357236,
-      "grad_norm": 1.476669192314148,
-      "learning_rate": 9.110954773869347e-05,
-      "loss": 5.6826,
-      "step": 9348
-    },
-    {
-      "epoch": 4.875619295958279,
-      "grad_norm": 1.497087001800537,
-      "learning_rate": 9.110854271356785e-05,
-      "loss": 5.2181,
-      "step": 9349
-    },
-    {
-      "epoch": 4.876140808344198,
-      "grad_norm": 1.39057457447052,
-      "learning_rate": 9.110753768844221e-05,
-      "loss": 5.6455,
-      "step": 9350
-    },
-    {
-      "epoch": 4.876662320730118,
-      "grad_norm": 1.5611858367919922,
-      "learning_rate": 9.110653266331659e-05,
-      "loss": 5.521,
-      "step": 9351
-    },
-    {
-      "epoch": 4.877183833116036,
-      "grad_norm": 1.4806443452835083,
-      "learning_rate": 9.110552763819095e-05,
-      "loss": 5.9352,
-      "step": 9352
-    },
-    {
-      "epoch": 4.877705345501956,
-      "grad_norm": 1.568584680557251,
-      "learning_rate": 9.110452261306533e-05,
-      "loss": 5.4091,
-      "step": 9353
-    },
-    {
-      "epoch": 4.878226857887875,
-      "grad_norm": 1.4509494304656982,
-      "learning_rate": 9.11035175879397e-05,
-      "loss": 6.0158,
-      "step": 9354
-    },
-    {
-      "epoch": 4.878748370273794,
-      "grad_norm": 1.5202879905700684,
-      "learning_rate": 9.110251256281407e-05,
-      "loss": 5.7209,
-      "step": 9355
-    },
-    {
-      "epoch": 4.879269882659713,
-      "grad_norm": 1.5131406784057617,
-      "learning_rate": 9.110150753768845e-05,
-      "loss": 5.6023,
-      "step": 9356
-    },
-    {
-      "epoch": 4.879791395045633,
-      "grad_norm": 1.4570741653442383,
-      "learning_rate": 9.110050251256282e-05,
-      "loss": 5.7224,
-      "step": 9357
-    },
-    {
-      "epoch": 4.880312907431551,
-      "grad_norm": 1.499408483505249,
-      "learning_rate": 9.10994974874372e-05,
-      "loss": 5.8401,
-      "step": 9358
-    },
-    {
-      "epoch": 4.880834419817471,
-      "grad_norm": 1.520756483078003,
-      "learning_rate": 9.109849246231156e-05,
-      "loss": 5.5546,
-      "step": 9359
-    },
-    {
-      "epoch": 4.88135593220339,
-      "grad_norm": 1.494465947151184,
-      "learning_rate": 9.109748743718594e-05,
-      "loss": 5.7074,
-      "step": 9360
-    },
-    {
-      "epoch": 4.881877444589309,
-      "grad_norm": 1.5333304405212402,
-      "learning_rate": 9.10964824120603e-05,
-      "loss": 4.9261,
-      "step": 9361
-    },
-    {
-      "epoch": 4.882398956975228,
-      "grad_norm": 1.4473456144332886,
-      "learning_rate": 9.109547738693468e-05,
-      "loss": 5.7631,
-      "step": 9362
-    },
-    {
-      "epoch": 4.882920469361148,
-      "grad_norm": 1.4831773042678833,
-      "learning_rate": 9.109447236180904e-05,
-      "loss": 5.7929,
-      "step": 9363
-    },
-    {
-      "epoch": 4.883441981747066,
-      "grad_norm": 1.5186314582824707,
-      "learning_rate": 9.109346733668342e-05,
-      "loss": 5.4476,
-      "step": 9364
-    },
-    {
-      "epoch": 4.883963494132986,
-      "grad_norm": 1.380068063735962,
-      "learning_rate": 9.109246231155778e-05,
-      "loss": 5.8617,
-      "step": 9365
-    },
-    {
-      "epoch": 4.884485006518905,
-      "grad_norm": 1.4257497787475586,
-      "learning_rate": 9.109145728643216e-05,
-      "loss": 5.7656,
-      "step": 9366
-    },
-    {
-      "epoch": 4.885006518904824,
-      "grad_norm": 1.354154348373413,
-      "learning_rate": 9.109045226130654e-05,
-      "loss": 6.0908,
-      "step": 9367
-    },
-    {
-      "epoch": 4.885528031290743,
-      "grad_norm": 1.3278284072875977,
-      "learning_rate": 9.108944723618092e-05,
-      "loss": 5.8533,
-      "step": 9368
-    },
-    {
-      "epoch": 4.886049543676663,
-      "grad_norm": 1.4890047311782837,
-      "learning_rate": 9.108844221105528e-05,
-      "loss": 5.4997,
-      "step": 9369
-    },
-    {
-      "epoch": 4.886571056062581,
-      "grad_norm": 1.5654428005218506,
-      "learning_rate": 9.108743718592965e-05,
-      "loss": 5.7629,
-      "step": 9370
-    },
-    {
-      "epoch": 4.887092568448501,
-      "grad_norm": 1.6418689489364624,
-      "learning_rate": 9.108643216080402e-05,
-      "loss": 5.1014,
-      "step": 9371
-    },
-    {
-      "epoch": 4.88761408083442,
-      "grad_norm": 1.7225359678268433,
-      "learning_rate": 9.108542713567839e-05,
-      "loss": 5.3718,
-      "step": 9372
-    },
-    {
-      "epoch": 4.888135593220339,
-      "grad_norm": 1.4604270458221436,
-      "learning_rate": 9.108442211055277e-05,
-      "loss": 5.8459,
-      "step": 9373
-    },
-    {
-      "epoch": 4.888657105606258,
-      "grad_norm": 1.4135218858718872,
-      "learning_rate": 9.108341708542713e-05,
-      "loss": 5.4729,
-      "step": 9374
-    },
-    {
-      "epoch": 4.889178617992178,
-      "grad_norm": 1.49873948097229,
-      "learning_rate": 9.108241206030151e-05,
-      "loss": 5.4259,
-      "step": 9375
-    },
-    {
-      "epoch": 4.889700130378096,
-      "grad_norm": 1.4721213579177856,
-      "learning_rate": 9.108140703517589e-05,
-      "loss": 5.2406,
-      "step": 9376
-    },
-    {
-      "epoch": 4.890221642764016,
-      "grad_norm": 1.4997867345809937,
-      "learning_rate": 9.108040201005026e-05,
-      "loss": 5.791,
-      "step": 9377
-    },
-    {
-      "epoch": 4.890743155149935,
-      "grad_norm": 1.54023277759552,
-      "learning_rate": 9.107939698492463e-05,
-      "loss": 5.5352,
-      "step": 9378
-    },
-    {
-      "epoch": 4.891264667535854,
-      "grad_norm": 1.4271931648254395,
-      "learning_rate": 9.1078391959799e-05,
-      "loss": 5.9129,
-      "step": 9379
-    },
-    {
-      "epoch": 4.891786179921773,
-      "grad_norm": 1.5283840894699097,
-      "learning_rate": 9.107738693467337e-05,
-      "loss": 5.6623,
-      "step": 9380
-    },
-    {
-      "epoch": 4.892307692307693,
-      "grad_norm": 1.4088325500488281,
-      "learning_rate": 9.107638190954775e-05,
-      "loss": 5.7835,
-      "step": 9381
-    },
-    {
-      "epoch": 4.892829204693611,
-      "grad_norm": 1.3421629667282104,
-      "learning_rate": 9.107537688442211e-05,
-      "loss": 5.8212,
-      "step": 9382
-    },
-    {
-      "epoch": 4.893350717079531,
-      "grad_norm": 1.4153391122817993,
-      "learning_rate": 9.107437185929649e-05,
-      "loss": 5.8393,
-      "step": 9383
-    },
-    {
-      "epoch": 4.893872229465449,
-      "grad_norm": 1.699793815612793,
-      "learning_rate": 9.107336683417086e-05,
-      "loss": 5.1491,
-      "step": 9384
-    },
-    {
-      "epoch": 4.894393741851369,
-      "grad_norm": 1.4392613172531128,
-      "learning_rate": 9.107236180904522e-05,
-      "loss": 5.8171,
-      "step": 9385
-    },
-    {
-      "epoch": 4.894915254237288,
-      "grad_norm": 1.4814019203186035,
-      "learning_rate": 9.10713567839196e-05,
-      "loss": 5.8058,
-      "step": 9386
-    },
-    {
-      "epoch": 4.895436766623208,
-      "grad_norm": 1.554839849472046,
-      "learning_rate": 9.107035175879398e-05,
-      "loss": 5.5459,
-      "step": 9387
-    },
-    {
-      "epoch": 4.895958279009126,
-      "grad_norm": 1.4767512083053589,
-      "learning_rate": 9.106934673366835e-05,
-      "loss": 5.5019,
-      "step": 9388
-    },
-    {
-      "epoch": 4.896479791395046,
-      "grad_norm": 1.5557408332824707,
-      "learning_rate": 9.106834170854272e-05,
-      "loss": 5.2855,
-      "step": 9389
-    },
-    {
-      "epoch": 4.897001303780964,
-      "grad_norm": 1.4546269178390503,
-      "learning_rate": 9.10673366834171e-05,
-      "loss": 5.6894,
-      "step": 9390
-    },
-    {
-      "epoch": 4.897522816166884,
-      "grad_norm": 1.7202287912368774,
-      "learning_rate": 9.106633165829146e-05,
-      "loss": 5.219,
-      "step": 9391
-    },
-    {
-      "epoch": 4.898044328552803,
-      "grad_norm": 1.3854893445968628,
-      "learning_rate": 9.106532663316584e-05,
-      "loss": 5.8198,
-      "step": 9392
-    },
-    {
-      "epoch": 4.898565840938723,
-      "grad_norm": 1.8713735342025757,
-      "learning_rate": 9.10643216080402e-05,
-      "loss": 5.4578,
-      "step": 9393
-    },
-    {
-      "epoch": 4.899087353324641,
-      "grad_norm": 1.438748836517334,
-      "learning_rate": 9.106331658291458e-05,
-      "loss": 5.9129,
-      "step": 9394
-    },
-    {
-      "epoch": 4.899608865710561,
-      "grad_norm": 1.5552420616149902,
-      "learning_rate": 9.106231155778894e-05,
-      "loss": 5.2891,
-      "step": 9395
-    },
-    {
-      "epoch": 4.900130378096479,
-      "grad_norm": 1.8511615991592407,
-      "learning_rate": 9.106130653266332e-05,
-      "loss": 5.3372,
-      "step": 9396
-    },
-    {
-      "epoch": 4.900651890482399,
-      "grad_norm": 1.5864999294281006,
-      "learning_rate": 9.10603015075377e-05,
-      "loss": 5.9007,
-      "step": 9397
-    },
-    {
-      "epoch": 4.901173402868318,
-      "grad_norm": 1.516449213027954,
-      "learning_rate": 9.105929648241206e-05,
-      "loss": 5.3411,
-      "step": 9398
-    },
-    {
-      "epoch": 4.901694915254238,
-      "grad_norm": 1.5337451696395874,
-      "learning_rate": 9.105829145728644e-05,
-      "loss": 5.6074,
-      "step": 9399
-    },
-    {
-      "epoch": 4.902216427640156,
-      "grad_norm": 1.4010370969772339,
-      "learning_rate": 9.10572864321608e-05,
-      "loss": 6.1563,
-      "step": 9400
-    },
-    {
-      "epoch": 4.902737940026076,
-      "grad_norm": 1.5485152006149292,
-      "learning_rate": 9.105628140703518e-05,
-      "loss": 6.0266,
-      "step": 9401
-    },
-    {
-      "epoch": 4.903259452411994,
-      "grad_norm": 1.4584873914718628,
-      "learning_rate": 9.105527638190955e-05,
-      "loss": 5.6044,
-      "step": 9402
-    },
-    {
-      "epoch": 4.903780964797914,
-      "grad_norm": 1.8672250509262085,
-      "learning_rate": 9.105427135678393e-05,
-      "loss": 5.9229,
-      "step": 9403
-    },
-    {
-      "epoch": 4.904302477183833,
-      "grad_norm": 1.9208084344863892,
-      "learning_rate": 9.105326633165829e-05,
-      "loss": 5.0152,
-      "step": 9404
-    },
-    {
-      "epoch": 4.904823989569753,
-      "grad_norm": 1.4721378087997437,
-      "learning_rate": 9.105226130653267e-05,
-      "loss": 6.0808,
-      "step": 9405
-    },
-    {
-      "epoch": 4.905345501955671,
-      "grad_norm": 1.4138704538345337,
-      "learning_rate": 9.105125628140703e-05,
-      "loss": 5.836,
-      "step": 9406
-    },
-    {
-      "epoch": 4.905867014341591,
-      "grad_norm": 1.4073163270950317,
-      "learning_rate": 9.105025125628141e-05,
-      "loss": 5.8614,
-      "step": 9407
-    },
-    {
-      "epoch": 4.906388526727509,
-      "grad_norm": 1.4413052797317505,
-      "learning_rate": 9.104924623115579e-05,
-      "loss": 5.9037,
-      "step": 9408
-    },
-    {
-      "epoch": 4.906910039113429,
-      "grad_norm": 1.6425275802612305,
-      "learning_rate": 9.104824120603017e-05,
-      "loss": 5.5354,
-      "step": 9409
-    },
-    {
-      "epoch": 4.907431551499348,
-      "grad_norm": 1.6388368606567383,
-      "learning_rate": 9.104723618090453e-05,
-      "loss": 5.7865,
-      "step": 9410
-    },
-    {
-      "epoch": 4.907953063885268,
-      "grad_norm": 1.4211171865463257,
-      "learning_rate": 9.10462311557789e-05,
-      "loss": 5.8618,
-      "step": 9411
-    },
-    {
-      "epoch": 4.908474576271186,
-      "grad_norm": 1.3258837461471558,
-      "learning_rate": 9.104522613065327e-05,
-      "loss": 5.8399,
-      "step": 9412
-    },
-    {
-      "epoch": 4.908996088657106,
-      "grad_norm": 1.31852388381958,
-      "learning_rate": 9.104422110552764e-05,
-      "loss": 6.004,
-      "step": 9413
-    },
-    {
-      "epoch": 4.909517601043024,
-      "grad_norm": 1.3715661764144897,
-      "learning_rate": 9.104321608040201e-05,
-      "loss": 5.4625,
-      "step": 9414
-    },
-    {
-      "epoch": 4.910039113428944,
-      "grad_norm": 1.4105144739151,
-      "learning_rate": 9.104221105527638e-05,
-      "loss": 6.1604,
-      "step": 9415
-    },
-    {
-      "epoch": 4.910560625814863,
-      "grad_norm": 1.50928795337677,
-      "learning_rate": 9.104120603015076e-05,
-      "loss": 5.9241,
-      "step": 9416
-    },
-    {
-      "epoch": 4.911082138200783,
-      "grad_norm": 1.6421071290969849,
-      "learning_rate": 9.104020100502513e-05,
-      "loss": 5.4649,
-      "step": 9417
-    },
-    {
-      "epoch": 4.911603650586701,
-      "grad_norm": 1.4532488584518433,
-      "learning_rate": 9.103919597989951e-05,
-      "loss": 5.4243,
-      "step": 9418
-    },
-    {
-      "epoch": 4.912125162972621,
-      "grad_norm": 1.6198644638061523,
-      "learning_rate": 9.103819095477388e-05,
-      "loss": 5.3546,
-      "step": 9419
-    },
-    {
-      "epoch": 4.912646675358539,
-      "grad_norm": 1.4275953769683838,
-      "learning_rate": 9.103718592964825e-05,
-      "loss": 5.7496,
-      "step": 9420
-    },
-    {
-      "epoch": 4.913168187744459,
-      "grad_norm": 1.4423511028289795,
-      "learning_rate": 9.103618090452262e-05,
-      "loss": 5.3867,
-      "step": 9421
-    },
-    {
-      "epoch": 4.913689700130378,
-      "grad_norm": 1.4166964292526245,
-      "learning_rate": 9.1035175879397e-05,
-      "loss": 5.6507,
-      "step": 9422
-    },
-    {
-      "epoch": 4.914211212516297,
-      "grad_norm": 1.3496978282928467,
-      "learning_rate": 9.103417085427136e-05,
-      "loss": 5.8646,
-      "step": 9423
-    },
-    {
-      "epoch": 4.914732724902216,
-      "grad_norm": 1.400513768196106,
-      "learning_rate": 9.103316582914572e-05,
-      "loss": 5.8356,
-      "step": 9424
-    },
-    {
-      "epoch": 4.915254237288136,
-      "grad_norm": 1.442557692527771,
-      "learning_rate": 9.10321608040201e-05,
-      "loss": 5.7954,
-      "step": 9425
-    },
-    {
-      "epoch": 4.915775749674054,
-      "grad_norm": 1.7103263139724731,
-      "learning_rate": 9.103115577889447e-05,
-      "loss": 5.3536,
-      "step": 9426
-    },
-    {
-      "epoch": 4.916297262059974,
-      "grad_norm": 1.374725341796875,
-      "learning_rate": 9.103015075376884e-05,
-      "loss": 5.7906,
-      "step": 9427
-    },
-    {
-      "epoch": 4.916818774445893,
-      "grad_norm": 1.5698847770690918,
-      "learning_rate": 9.102914572864322e-05,
-      "loss": 5.1926,
-      "step": 9428
-    },
-    {
-      "epoch": 4.917340286831812,
-      "grad_norm": 1.6266674995422363,
-      "learning_rate": 9.10281407035176e-05,
-      "loss": 5.6622,
-      "step": 9429
-    },
-    {
-      "epoch": 4.917861799217731,
-      "grad_norm": 1.3146287202835083,
-      "learning_rate": 9.102713567839196e-05,
-      "loss": 5.9735,
-      "step": 9430
-    },
-    {
-      "epoch": 4.918383311603651,
-      "grad_norm": 1.3534080982208252,
-      "learning_rate": 9.102613065326634e-05,
-      "loss": 5.8186,
-      "step": 9431
-    },
-    {
-      "epoch": 4.918904823989569,
-      "grad_norm": 1.4958523511886597,
-      "learning_rate": 9.10251256281407e-05,
-      "loss": 5.376,
-      "step": 9432
-    },
-    {
-      "epoch": 4.919426336375489,
-      "grad_norm": 1.7073709964752197,
-      "learning_rate": 9.102412060301508e-05,
-      "loss": 5.4069,
-      "step": 9433
-    },
-    {
-      "epoch": 4.919947848761408,
-      "grad_norm": 1.3544573783874512,
-      "learning_rate": 9.102311557788945e-05,
-      "loss": 5.695,
-      "step": 9434
-    },
-    {
-      "epoch": 4.920469361147327,
-      "grad_norm": 1.4455342292785645,
-      "learning_rate": 9.102211055276383e-05,
-      "loss": 5.833,
-      "step": 9435
-    },
-    {
-      "epoch": 4.920990873533246,
-      "grad_norm": 1.451461911201477,
-      "learning_rate": 9.102110552763819e-05,
-      "loss": 5.4009,
-      "step": 9436
-    },
-    {
-      "epoch": 4.921512385919166,
-      "grad_norm": 1.3886125087738037,
-      "learning_rate": 9.102010050251257e-05,
-      "loss": 5.5272,
-      "step": 9437
-    },
-    {
-      "epoch": 4.922033898305084,
-      "grad_norm": 1.506926417350769,
-      "learning_rate": 9.101909547738695e-05,
-      "loss": 5.6582,
-      "step": 9438
-    },
-    {
-      "epoch": 4.922555410691004,
-      "grad_norm": 1.4085625410079956,
-      "learning_rate": 9.101809045226131e-05,
-      "loss": 5.8507,
-      "step": 9439
-    },
-    {
-      "epoch": 4.923076923076923,
-      "grad_norm": 1.4849376678466797,
-      "learning_rate": 9.101708542713569e-05,
-      "loss": 5.5532,
-      "step": 9440
-    },
-    {
-      "epoch": 4.923598435462842,
-      "grad_norm": 1.3830053806304932,
-      "learning_rate": 9.101608040201005e-05,
-      "loss": 5.6157,
-      "step": 9441
-    },
-    {
-      "epoch": 4.924119947848761,
-      "grad_norm": 1.4043306112289429,
-      "learning_rate": 9.101507537688443e-05,
-      "loss": 5.5406,
-      "step": 9442
-    },
-    {
-      "epoch": 4.924641460234681,
-      "grad_norm": 1.4923397302627563,
-      "learning_rate": 9.10140703517588e-05,
-      "loss": 5.5923,
-      "step": 9443
-    },
-    {
-      "epoch": 4.925162972620599,
-      "grad_norm": 1.3725039958953857,
-      "learning_rate": 9.101306532663317e-05,
-      "loss": 5.2186,
-      "step": 9444
-    },
-    {
-      "epoch": 4.925684485006519,
-      "grad_norm": 1.4008874893188477,
-      "learning_rate": 9.101206030150754e-05,
-      "loss": 5.8808,
-      "step": 9445
-    },
-    {
-      "epoch": 4.926205997392438,
-      "grad_norm": 1.316645622253418,
-      "learning_rate": 9.101105527638191e-05,
-      "loss": 5.9233,
-      "step": 9446
-    },
-    {
-      "epoch": 4.926727509778357,
-      "grad_norm": 1.5887460708618164,
-      "learning_rate": 9.101005025125628e-05,
-      "loss": 4.9252,
-      "step": 9447
-    },
-    {
-      "epoch": 4.927249022164276,
-      "grad_norm": 1.5180329084396362,
-      "learning_rate": 9.100904522613066e-05,
-      "loss": 5.4201,
-      "step": 9448
-    },
-    {
-      "epoch": 4.927770534550196,
-      "grad_norm": 1.5920976400375366,
-      "learning_rate": 9.100804020100503e-05,
-      "loss": 5.3298,
-      "step": 9449
-    },
-    {
-      "epoch": 4.9282920469361144,
-      "grad_norm": 1.3984103202819824,
-      "learning_rate": 9.10070351758794e-05,
-      "loss": 5.848,
-      "step": 9450
-    },
-    {
-      "epoch": 4.928813559322034,
-      "grad_norm": 1.3771047592163086,
-      "learning_rate": 9.100603015075378e-05,
-      "loss": 5.9414,
-      "step": 9451
-    },
-    {
-      "epoch": 4.929335071707953,
-      "grad_norm": 1.4695042371749878,
-      "learning_rate": 9.100502512562814e-05,
-      "loss": 5.8318,
-      "step": 9452
-    },
-    {
-      "epoch": 4.929856584093872,
-      "grad_norm": 1.4828572273254395,
-      "learning_rate": 9.100402010050252e-05,
-      "loss": 4.7916,
-      "step": 9453
-    },
-    {
-      "epoch": 4.930378096479791,
-      "grad_norm": 1.3119230270385742,
-      "learning_rate": 9.100301507537688e-05,
-      "loss": 5.6552,
-      "step": 9454
-    },
-    {
-      "epoch": 4.930899608865711,
-      "grad_norm": 1.4734947681427002,
-      "learning_rate": 9.100201005025126e-05,
-      "loss": 4.8748,
-      "step": 9455
-    },
-    {
-      "epoch": 4.9314211212516295,
-      "grad_norm": 1.7135305404663086,
-      "learning_rate": 9.100100502512563e-05,
-      "loss": 5.272,
-      "step": 9456
-    },
-    {
-      "epoch": 4.931942633637549,
-      "grad_norm": 1.5139179229736328,
-      "learning_rate": 9.1e-05,
-      "loss": 5.4931,
-      "step": 9457
-    },
-    {
-      "epoch": 4.932464146023468,
-      "grad_norm": 1.7476977109909058,
-      "learning_rate": 9.099899497487438e-05,
-      "loss": 5.7799,
-      "step": 9458
-    },
-    {
-      "epoch": 4.932985658409387,
-      "grad_norm": 1.6605749130249023,
-      "learning_rate": 9.099798994974876e-05,
-      "loss": 5.8088,
-      "step": 9459
-    },
-    {
-      "epoch": 4.933507170795306,
-      "grad_norm": 1.5193214416503906,
-      "learning_rate": 9.099698492462312e-05,
-      "loss": 5.8406,
-      "step": 9460
-    },
-    {
-      "epoch": 4.934028683181226,
-      "grad_norm": 1.6792309284210205,
-      "learning_rate": 9.09959798994975e-05,
-      "loss": 5.7296,
-      "step": 9461
-    },
-    {
-      "epoch": 4.9345501955671445,
-      "grad_norm": 1.4395103454589844,
-      "learning_rate": 9.099497487437186e-05,
-      "loss": 5.9302,
-      "step": 9462
-    },
-    {
-      "epoch": 4.935071707953064,
-      "grad_norm": 1.6730462312698364,
-      "learning_rate": 9.099396984924623e-05,
-      "loss": 5.6989,
-      "step": 9463
-    },
-    {
-      "epoch": 4.935593220338983,
-      "grad_norm": 1.375931739807129,
-      "learning_rate": 9.099296482412061e-05,
-      "loss": 5.9066,
-      "step": 9464
-    },
-    {
-      "epoch": 4.936114732724902,
-      "grad_norm": 1.4723083972930908,
-      "learning_rate": 9.099195979899497e-05,
-      "loss": 5.8884,
-      "step": 9465
-    },
-    {
-      "epoch": 4.936636245110821,
-      "grad_norm": 1.536982774734497,
-      "learning_rate": 9.099095477386935e-05,
-      "loss": 5.4847,
-      "step": 9466
-    },
-    {
-      "epoch": 4.937157757496741,
-      "grad_norm": 1.4303357601165771,
-      "learning_rate": 9.098994974874371e-05,
-      "loss": 5.9326,
-      "step": 9467
-    },
-    {
-      "epoch": 4.9376792698826595,
-      "grad_norm": 1.5456264019012451,
-      "learning_rate": 9.098894472361809e-05,
-      "loss": 5.8387,
-      "step": 9468
-    },
-    {
-      "epoch": 4.938200782268579,
-      "grad_norm": 1.4804474115371704,
-      "learning_rate": 9.098793969849247e-05,
-      "loss": 5.7955,
-      "step": 9469
-    },
-    {
-      "epoch": 4.938722294654498,
-      "grad_norm": 1.4159609079360962,
-      "learning_rate": 9.098693467336685e-05,
-      "loss": 5.5897,
-      "step": 9470
-    },
-    {
-      "epoch": 4.939243807040417,
-      "grad_norm": 1.4278018474578857,
-      "learning_rate": 9.098592964824121e-05,
-      "loss": 5.6639,
-      "step": 9471
-    },
-    {
-      "epoch": 4.9397653194263365,
-      "grad_norm": 1.4861332178115845,
-      "learning_rate": 9.098492462311559e-05,
-      "loss": 5.6074,
-      "step": 9472
-    },
-    {
-      "epoch": 4.940286831812256,
-      "grad_norm": 1.5369443893432617,
-      "learning_rate": 9.098391959798995e-05,
-      "loss": 5.5954,
-      "step": 9473
-    },
-    {
-      "epoch": 4.9408083441981745,
-      "grad_norm": 1.4548251628875732,
-      "learning_rate": 9.098291457286433e-05,
-      "loss": 5.4274,
-      "step": 9474
-    },
-    {
-      "epoch": 4.941329856584094,
-      "grad_norm": 1.3125852346420288,
-      "learning_rate": 9.09819095477387e-05,
-      "loss": 6.105,
-      "step": 9475
-    },
-    {
-      "epoch": 4.941851368970013,
-      "grad_norm": 1.385012149810791,
-      "learning_rate": 9.098090452261307e-05,
-      "loss": 6.0171,
-      "step": 9476
-    },
-    {
-      "epoch": 4.942372881355932,
-      "grad_norm": 1.7182080745697021,
-      "learning_rate": 9.097989949748744e-05,
-      "loss": 5.7029,
-      "step": 9477
-    },
-    {
-      "epoch": 4.9428943937418515,
-      "grad_norm": 1.6209876537322998,
-      "learning_rate": 9.097889447236182e-05,
-      "loss": 5.6085,
-      "step": 9478
-    },
-    {
-      "epoch": 4.94341590612777,
-      "grad_norm": 1.587426781654358,
-      "learning_rate": 9.097788944723619e-05,
-      "loss": 5.5791,
-      "step": 9479
-    },
-    {
-      "epoch": 4.9439374185136895,
-      "grad_norm": 1.4025925397872925,
-      "learning_rate": 9.097688442211056e-05,
-      "loss": 5.6024,
-      "step": 9480
-    },
-    {
-      "epoch": 4.944458930899609,
-      "grad_norm": 1.5513529777526855,
-      "learning_rate": 9.097587939698494e-05,
-      "loss": 5.7844,
-      "step": 9481
-    },
-    {
-      "epoch": 4.944980443285528,
-      "grad_norm": 1.682541012763977,
-      "learning_rate": 9.09748743718593e-05,
-      "loss": 5.1822,
-      "step": 9482
-    },
-    {
-      "epoch": 4.945501955671447,
-      "grad_norm": 1.6363345384597778,
-      "learning_rate": 9.097386934673368e-05,
-      "loss": 5.6856,
-      "step": 9483
-    },
-    {
-      "epoch": 4.9460234680573665,
-      "grad_norm": 1.5173075199127197,
-      "learning_rate": 9.097286432160804e-05,
-      "loss": 5.8913,
-      "step": 9484
-    },
-    {
-      "epoch": 4.946544980443285,
-      "grad_norm": 1.7930773496627808,
-      "learning_rate": 9.097185929648242e-05,
-      "loss": 5.9791,
-      "step": 9485
-    },
-    {
-      "epoch": 4.9470664928292045,
-      "grad_norm": 1.4281975030899048,
-      "learning_rate": 9.097085427135678e-05,
-      "loss": 5.7972,
-      "step": 9486
-    },
-    {
-      "epoch": 4.947588005215124,
-      "grad_norm": 1.4980669021606445,
-      "learning_rate": 9.096984924623116e-05,
-      "loss": 5.8511,
-      "step": 9487
-    },
-    {
-      "epoch": 4.9481095176010434,
-      "grad_norm": 1.465827465057373,
-      "learning_rate": 9.096884422110553e-05,
-      "loss": 5.7684,
-      "step": 9488
-    },
-    {
-      "epoch": 4.948631029986962,
-      "grad_norm": 1.5295451879501343,
-      "learning_rate": 9.09678391959799e-05,
-      "loss": 5.3219,
-      "step": 9489
-    },
-    {
-      "epoch": 4.9491525423728815,
-      "grad_norm": 1.5520877838134766,
-      "learning_rate": 9.096683417085428e-05,
-      "loss": 6.1132,
-      "step": 9490
-    },
-    {
-      "epoch": 4.9496740547588,
-      "grad_norm": 1.6648917198181152,
-      "learning_rate": 9.096582914572865e-05,
-      "loss": 5.5262,
-      "step": 9491
-    },
-    {
-      "epoch": 4.9501955671447195,
-      "grad_norm": 1.3541868925094604,
-      "learning_rate": 9.096482412060302e-05,
-      "loss": 5.2264,
-      "step": 9492
-    },
-    {
-      "epoch": 4.950717079530639,
-      "grad_norm": 1.5014731884002686,
-      "learning_rate": 9.096381909547739e-05,
-      "loss": 6.012,
-      "step": 9493
-    },
-    {
-      "epoch": 4.9512385919165585,
-      "grad_norm": 1.4473848342895508,
-      "learning_rate": 9.096281407035177e-05,
-      "loss": 5.8941,
-      "step": 9494
-    },
-    {
-      "epoch": 4.951760104302477,
-      "grad_norm": 1.4296332597732544,
-      "learning_rate": 9.096180904522613e-05,
-      "loss": 6.1712,
-      "step": 9495
-    },
-    {
-      "epoch": 4.9522816166883965,
-      "grad_norm": 1.4028007984161377,
-      "learning_rate": 9.096080402010051e-05,
-      "loss": 5.7572,
-      "step": 9496
-    },
-    {
-      "epoch": 4.952803129074315,
-      "grad_norm": 1.364015817642212,
-      "learning_rate": 9.095979899497487e-05,
-      "loss": 4.98,
-      "step": 9497
-    },
-    {
-      "epoch": 4.9533246414602345,
-      "grad_norm": 1.5670381784439087,
-      "learning_rate": 9.095879396984925e-05,
-      "loss": 5.1124,
-      "step": 9498
-    },
-    {
-      "epoch": 4.953846153846154,
-      "grad_norm": 1.597085952758789,
-      "learning_rate": 9.095778894472361e-05,
-      "loss": 5.6751,
-      "step": 9499
-    },
-    {
-      "epoch": 4.9543676662320735,
-      "grad_norm": 1.5139598846435547,
-      "learning_rate": 9.095678391959799e-05,
-      "loss": 5.8707,
-      "step": 9500
-    },
-    {
-      "epoch": 4.954889178617992,
-      "grad_norm": 1.4488534927368164,
-      "learning_rate": 9.095577889447237e-05,
-      "loss": 5.6008,
-      "step": 9501
-    },
-    {
-      "epoch": 4.9554106910039115,
-      "grad_norm": 1.5516542196273804,
-      "learning_rate": 9.095477386934675e-05,
-      "loss": 5.5964,
-      "step": 9502
-    },
-    {
-      "epoch": 4.95593220338983,
-      "grad_norm": 1.4768667221069336,
-      "learning_rate": 9.095376884422111e-05,
-      "loss": 5.769,
-      "step": 9503
-    },
-    {
-      "epoch": 4.95645371577575,
-      "grad_norm": 1.4210253953933716,
-      "learning_rate": 9.095276381909548e-05,
-      "loss": 5.5513,
-      "step": 9504
-    },
-    {
-      "epoch": 4.956975228161669,
-      "grad_norm": 1.351924180984497,
-      "learning_rate": 9.095175879396985e-05,
-      "loss": 5.9911,
-      "step": 9505
-    },
-    {
-      "epoch": 4.9574967405475885,
-      "grad_norm": 1.5593440532684326,
-      "learning_rate": 9.095075376884422e-05,
-      "loss": 5.5771,
-      "step": 9506
-    },
-    {
-      "epoch": 4.958018252933507,
-      "grad_norm": 1.5490654706954956,
-      "learning_rate": 9.09497487437186e-05,
-      "loss": 5.3494,
-      "step": 9507
-    },
-    {
-      "epoch": 4.9585397653194265,
-      "grad_norm": 2.072201728820801,
-      "learning_rate": 9.094874371859296e-05,
-      "loss": 5.5346,
-      "step": 9508
-    },
-    {
-      "epoch": 4.959061277705345,
-      "grad_norm": 1.4470704793930054,
-      "learning_rate": 9.094773869346734e-05,
-      "loss": 5.3475,
-      "step": 9509
-    },
-    {
-      "epoch": 4.959582790091265,
-      "grad_norm": 1.5778701305389404,
-      "learning_rate": 9.094673366834172e-05,
-      "loss": 5.352,
-      "step": 9510
-    },
-    {
-      "epoch": 4.960104302477184,
-      "grad_norm": 1.4674078226089478,
-      "learning_rate": 9.09457286432161e-05,
-      "loss": 5.6995,
-      "step": 9511
-    },
-    {
-      "epoch": 4.960625814863103,
-      "grad_norm": 1.5061572790145874,
-      "learning_rate": 9.094472361809046e-05,
-      "loss": 5.7614,
-      "step": 9512
-    },
-    {
-      "epoch": 4.961147327249022,
-      "grad_norm": 1.729617953300476,
-      "learning_rate": 9.094371859296484e-05,
-      "loss": 5.4714,
-      "step": 9513
-    },
-    {
-      "epoch": 4.9616688396349415,
-      "grad_norm": 1.5308289527893066,
-      "learning_rate": 9.09427135678392e-05,
-      "loss": 5.9243,
-      "step": 9514
-    },
-    {
-      "epoch": 4.96219035202086,
-      "grad_norm": 1.4754334688186646,
-      "learning_rate": 9.094170854271358e-05,
-      "loss": 5.7547,
-      "step": 9515
-    },
-    {
-      "epoch": 4.96271186440678,
-      "grad_norm": 1.5298941135406494,
-      "learning_rate": 9.094070351758794e-05,
-      "loss": 5.5908,
-      "step": 9516
-    },
-    {
-      "epoch": 4.963233376792699,
-      "grad_norm": 1.6703627109527588,
-      "learning_rate": 9.09396984924623e-05,
-      "loss": 5.6519,
-      "step": 9517
-    },
-    {
-      "epoch": 4.963754889178618,
-      "grad_norm": 1.3146605491638184,
-      "learning_rate": 9.093869346733668e-05,
-      "loss": 6.0507,
-      "step": 9518
-    },
-    {
-      "epoch": 4.964276401564537,
-      "grad_norm": 1.3258230686187744,
-      "learning_rate": 9.093768844221105e-05,
-      "loss": 5.3924,
-      "step": 9519
-    },
-    {
-      "epoch": 4.9647979139504566,
-      "grad_norm": 1.32620370388031,
-      "learning_rate": 9.093668341708543e-05,
-      "loss": 6.0788,
-      "step": 9520
-    },
-    {
-      "epoch": 4.965319426336375,
-      "grad_norm": 1.4107927083969116,
-      "learning_rate": 9.09356783919598e-05,
-      "loss": 5.7069,
-      "step": 9521
-    },
-    {
-      "epoch": 4.965840938722295,
-      "grad_norm": 1.6384838819503784,
-      "learning_rate": 9.093467336683418e-05,
-      "loss": 5.2223,
-      "step": 9522
-    },
-    {
-      "epoch": 4.966362451108214,
-      "grad_norm": 1.527696132659912,
-      "learning_rate": 9.093366834170855e-05,
-      "loss": 5.45,
-      "step": 9523
-    },
-    {
-      "epoch": 4.966883963494133,
-      "grad_norm": 1.5236390829086304,
-      "learning_rate": 9.093266331658292e-05,
-      "loss": 5.4657,
-      "step": 9524
-    },
-    {
-      "epoch": 4.967405475880052,
-      "grad_norm": 1.3854538202285767,
-      "learning_rate": 9.093165829145729e-05,
-      "loss": 5.8235,
-      "step": 9525
-    },
-    {
-      "epoch": 4.967926988265972,
-      "grad_norm": 1.3617008924484253,
-      "learning_rate": 9.093065326633167e-05,
-      "loss": 5.95,
-      "step": 9526
-    },
-    {
-      "epoch": 4.96844850065189,
-      "grad_norm": 1.4061341285705566,
-      "learning_rate": 9.092964824120603e-05,
-      "loss": 5.7468,
-      "step": 9527
-    },
-    {
-      "epoch": 4.96897001303781,
-      "grad_norm": 1.4063454866409302,
-      "learning_rate": 9.092864321608041e-05,
-      "loss": 5.8414,
-      "step": 9528
-    },
-    {
-      "epoch": 4.969491525423729,
-      "grad_norm": 1.5409337282180786,
-      "learning_rate": 9.092763819095477e-05,
-      "loss": 5.7194,
-      "step": 9529
-    },
-    {
-      "epoch": 4.970013037809648,
-      "grad_norm": 1.5460985898971558,
-      "learning_rate": 9.092663316582915e-05,
-      "loss": 5.5166,
-      "step": 9530
-    },
-    {
-      "epoch": 4.970534550195567,
-      "grad_norm": 1.4051940441131592,
-      "learning_rate": 9.092562814070353e-05,
-      "loss": 5.8389,
-      "step": 9531
-    },
-    {
-      "epoch": 4.971056062581487,
-      "grad_norm": 1.391068458557129,
-      "learning_rate": 9.092462311557789e-05,
-      "loss": 6.0721,
-      "step": 9532
-    },
-    {
-      "epoch": 4.971577574967405,
-      "grad_norm": 1.4668761491775513,
-      "learning_rate": 9.092361809045227e-05,
-      "loss": 5.2495,
-      "step": 9533
-    },
-    {
-      "epoch": 4.972099087353325,
-      "grad_norm": 1.5273724794387817,
-      "learning_rate": 9.092261306532663e-05,
-      "loss": 5.2345,
-      "step": 9534
-    },
-    {
-      "epoch": 4.972620599739244,
-      "grad_norm": 1.4453911781311035,
-      "learning_rate": 9.092160804020101e-05,
-      "loss": 5.4535,
-      "step": 9535
-    },
-    {
-      "epoch": 4.973142112125163,
-      "grad_norm": 1.6361924409866333,
-      "learning_rate": 9.092060301507538e-05,
-      "loss": 5.562,
-      "step": 9536
-    },
-    {
-      "epoch": 4.973663624511082,
-      "grad_norm": 1.5112252235412598,
-      "learning_rate": 9.091959798994975e-05,
-      "loss": 5.5583,
-      "step": 9537
-    },
-    {
-      "epoch": 4.974185136897002,
-      "grad_norm": 1.4776006937026978,
-      "learning_rate": 9.091859296482412e-05,
-      "loss": 5.4313,
-      "step": 9538
-    },
-    {
-      "epoch": 4.97470664928292,
-      "grad_norm": 1.448299527168274,
-      "learning_rate": 9.09175879396985e-05,
-      "loss": 5.729,
-      "step": 9539
-    },
-    {
-      "epoch": 4.97522816166884,
-      "grad_norm": 1.4582806825637817,
-      "learning_rate": 9.091658291457286e-05,
-      "loss": 5.5718,
-      "step": 9540
-    },
-    {
-      "epoch": 4.975749674054759,
-      "grad_norm": 1.483436107635498,
-      "learning_rate": 9.091557788944724e-05,
-      "loss": 5.6077,
-      "step": 9541
-    },
-    {
-      "epoch": 4.976271186440678,
-      "grad_norm": 1.4505349397659302,
-      "learning_rate": 9.091457286432162e-05,
-      "loss": 5.2934,
-      "step": 9542
-    },
-    {
-      "epoch": 4.976792698826597,
-      "grad_norm": 1.4045169353485107,
-      "learning_rate": 9.091356783919598e-05,
-      "loss": 5.7949,
-      "step": 9543
-    },
-    {
-      "epoch": 4.977314211212517,
-      "grad_norm": 1.5387983322143555,
-      "learning_rate": 9.091256281407036e-05,
-      "loss": 5.6566,
-      "step": 9544
-    },
-    {
-      "epoch": 4.977835723598435,
-      "grad_norm": 1.427734613418579,
-      "learning_rate": 9.091155778894472e-05,
-      "loss": 5.7165,
-      "step": 9545
-    },
-    {
-      "epoch": 4.978357235984355,
-      "grad_norm": 1.4534152746200562,
-      "learning_rate": 9.09105527638191e-05,
-      "loss": 5.5994,
-      "step": 9546
-    },
-    {
-      "epoch": 4.978878748370274,
-      "grad_norm": 1.2882143259048462,
-      "learning_rate": 9.090954773869347e-05,
-      "loss": 6.0772,
-      "step": 9547
-    },
-    {
-      "epoch": 4.979400260756193,
-      "grad_norm": 1.518412470817566,
-      "learning_rate": 9.090854271356784e-05,
-      "loss": 5.5623,
-      "step": 9548
-    },
-    {
-      "epoch": 4.979921773142112,
-      "grad_norm": 1.4349064826965332,
-      "learning_rate": 9.090753768844221e-05,
-      "loss": 5.6247,
-      "step": 9549
-    },
-    {
-      "epoch": 4.980443285528032,
-      "grad_norm": 1.4152820110321045,
-      "learning_rate": 9.090653266331659e-05,
-      "loss": 5.7244,
-      "step": 9550
-    },
-    {
-      "epoch": 4.98096479791395,
-      "grad_norm": 1.3649566173553467,
-      "learning_rate": 9.090552763819096e-05,
-      "loss": 5.9428,
-      "step": 9551
-    },
-    {
-      "epoch": 4.98148631029987,
-      "grad_norm": 1.4020839929580688,
-      "learning_rate": 9.090452261306534e-05,
-      "loss": 6.0766,
-      "step": 9552
-    },
-    {
-      "epoch": 4.982007822685789,
-      "grad_norm": 1.5051881074905396,
-      "learning_rate": 9.09035175879397e-05,
-      "loss": 5.7072,
-      "step": 9553
-    },
-    {
-      "epoch": 4.982529335071708,
-      "grad_norm": 1.3744229078292847,
-      "learning_rate": 9.090251256281408e-05,
-      "loss": 5.6609,
-      "step": 9554
-    },
-    {
-      "epoch": 4.983050847457627,
-      "grad_norm": 1.5600794553756714,
-      "learning_rate": 9.090150753768845e-05,
-      "loss": 5.2753,
-      "step": 9555
-    },
-    {
-      "epoch": 4.983572359843547,
-      "grad_norm": 1.4500224590301514,
-      "learning_rate": 9.090050251256281e-05,
-      "loss": 5.8845,
-      "step": 9556
-    },
-    {
-      "epoch": 4.984093872229465,
-      "grad_norm": 1.5056763887405396,
-      "learning_rate": 9.089949748743719e-05,
-      "loss": 5.5953,
-      "step": 9557
-    },
-    {
-      "epoch": 4.984615384615385,
-      "grad_norm": 1.458542823791504,
-      "learning_rate": 9.089849246231155e-05,
-      "loss": 5.743,
-      "step": 9558
-    },
-    {
-      "epoch": 4.985136897001304,
-      "grad_norm": 1.4352842569351196,
-      "learning_rate": 9.089748743718593e-05,
-      "loss": 5.8879,
-      "step": 9559
-    },
-    {
-      "epoch": 4.985658409387223,
-      "grad_norm": 1.509617805480957,
-      "learning_rate": 9.08964824120603e-05,
-      "loss": 5.4957,
-      "step": 9560
-    },
-    {
-      "epoch": 4.986179921773142,
-      "grad_norm": 1.4734725952148438,
-      "learning_rate": 9.089547738693467e-05,
-      "loss": 5.909,
-      "step": 9561
-    },
-    {
-      "epoch": 4.986701434159062,
-      "grad_norm": 1.4185115098953247,
-      "learning_rate": 9.089447236180905e-05,
-      "loss": 5.3034,
-      "step": 9562
-    },
-    {
-      "epoch": 4.98722294654498,
-      "grad_norm": 1.3917583227157593,
-      "learning_rate": 9.089346733668343e-05,
-      "loss": 5.7737,
-      "step": 9563
-    },
-    {
-      "epoch": 4.9877444589309,
-      "grad_norm": 1.4229999780654907,
-      "learning_rate": 9.08924623115578e-05,
-      "loss": 5.5737,
-      "step": 9564
-    },
-    {
-      "epoch": 4.988265971316819,
-      "grad_norm": 1.3993674516677856,
-      "learning_rate": 9.089145728643217e-05,
-      "loss": 5.8591,
-      "step": 9565
-    },
-    {
-      "epoch": 4.988787483702738,
-      "grad_norm": 1.3302586078643799,
-      "learning_rate": 9.089045226130654e-05,
-      "loss": 5.4385,
-      "step": 9566
-    },
-    {
-      "epoch": 4.989308996088657,
-      "grad_norm": 1.453735113143921,
-      "learning_rate": 9.088944723618091e-05,
-      "loss": 5.4714,
-      "step": 9567
-    },
-    {
-      "epoch": 4.989830508474577,
-      "grad_norm": 1.37211275100708,
-      "learning_rate": 9.088844221105528e-05,
-      "loss": 5.0002,
-      "step": 9568
-    },
-    {
-      "epoch": 4.990352020860495,
-      "grad_norm": 1.3916655778884888,
-      "learning_rate": 9.088743718592966e-05,
-      "loss": 5.6153,
-      "step": 9569
-    },
-    {
-      "epoch": 4.990873533246415,
-      "grad_norm": 1.422653317451477,
-      "learning_rate": 9.088643216080402e-05,
-      "loss": 5.6719,
-      "step": 9570
-    },
-    {
-      "epoch": 4.991395045632334,
-      "grad_norm": 1.440453052520752,
-      "learning_rate": 9.08854271356784e-05,
-      "loss": 5.6978,
-      "step": 9571
-    },
-    {
-      "epoch": 4.991916558018253,
-      "grad_norm": 1.4746758937835693,
-      "learning_rate": 9.088442211055278e-05,
-      "loss": 5.7122,
-      "step": 9572
-    },
-    {
-      "epoch": 4.992438070404172,
-      "grad_norm": 1.535989761352539,
-      "learning_rate": 9.088341708542714e-05,
-      "loss": 5.5598,
-      "step": 9573
-    },
-    {
-      "epoch": 4.992959582790091,
-      "grad_norm": 1.5031657218933105,
-      "learning_rate": 9.088241206030152e-05,
-      "loss": 5.2815,
-      "step": 9574
-    },
-    {
-      "epoch": 4.99348109517601,
-      "grad_norm": 1.5244801044464111,
-      "learning_rate": 9.088140703517588e-05,
-      "loss": 5.7473,
-      "step": 9575
-    },
-    {
-      "epoch": 4.99400260756193,
-      "grad_norm": 1.3193918466567993,
-      "learning_rate": 9.088040201005026e-05,
-      "loss": 5.222,
-      "step": 9576
-    },
-    {
-      "epoch": 4.994524119947849,
-      "grad_norm": 1.6308892965316772,
-      "learning_rate": 9.087939698492462e-05,
-      "loss": 5.2045,
-      "step": 9577
-    },
-    {
-      "epoch": 4.995045632333768,
-      "grad_norm": 1.5379629135131836,
-      "learning_rate": 9.0878391959799e-05,
-      "loss": 5.3605,
-      "step": 9578
-    },
-    {
-      "epoch": 4.995567144719687,
-      "grad_norm": 1.3673256635665894,
-      "learning_rate": 9.087738693467337e-05,
-      "loss": 5.8343,
-      "step": 9579
-    },
-    {
-      "epoch": 4.996088657105606,
-      "grad_norm": 1.4191076755523682,
-      "learning_rate": 9.087638190954774e-05,
-      "loss": 5.5442,
-      "step": 9580
-    },
-    {
-      "epoch": 4.996610169491525,
-      "grad_norm": 1.3783507347106934,
-      "learning_rate": 9.087537688442211e-05,
-      "loss": 5.3831,
-      "step": 9581
-    },
-    {
-      "epoch": 4.997131681877445,
-      "grad_norm": 1.510187029838562,
-      "learning_rate": 9.087437185929649e-05,
-      "loss": 5.483,
-      "step": 9582
-    },
-    {
-      "epoch": 4.997653194263364,
-      "grad_norm": 1.507614016532898,
-      "learning_rate": 9.087336683417086e-05,
-      "loss": 5.5779,
-      "step": 9583
-    },
-    {
-      "epoch": 4.998174706649283,
-      "grad_norm": 1.4115822315216064,
-      "learning_rate": 9.087236180904523e-05,
-      "loss": 5.7898,
-      "step": 9584
-    },
-    {
-      "epoch": 4.998696219035202,
-      "grad_norm": 1.4444499015808105,
-      "learning_rate": 9.08713567839196e-05,
-      "loss": 5.7272,
-      "step": 9585
-    },
-    {
-      "epoch": 4.999217731421121,
-      "grad_norm": 1.549178123474121,
-      "learning_rate": 9.087035175879397e-05,
-      "loss": 5.4709,
-      "step": 9586
-    },
-    {
-      "epoch": 4.99973924380704,
-      "grad_norm": 1.4479438066482544,
-      "learning_rate": 9.086934673366835e-05,
-      "loss": 5.6143,
-      "step": 9587
-    },
-    {
-      "epoch": 5.00026075619296,
-      "grad_norm": 1.625806212425232,
-      "learning_rate": 9.086834170854271e-05,
-      "loss": 5.5754,
-      "step": 9588
-    },
-    {
-      "epoch": 5.000782268578878,
-      "grad_norm": 1.6331462860107422,
-      "learning_rate": 9.086733668341709e-05,
-      "loss": 4.9812,
-      "step": 9589
-    },
-    {
-      "epoch": 5.001303780964798,
-      "grad_norm": 1.6176091432571411,
-      "learning_rate": 9.086633165829145e-05,
-      "loss": 5.6923,
-      "step": 9590
-    },
-    {
-      "epoch": 5.001825293350717,
-      "grad_norm": 1.4320005178451538,
-      "learning_rate": 9.086532663316583e-05,
-      "loss": 6.015,
-      "step": 9591
-    },
-    {
-      "epoch": 5.002346805736636,
-      "grad_norm": 1.3827357292175293,
-      "learning_rate": 9.086432160804021e-05,
-      "loss": 5.6476,
-      "step": 9592
-    },
-    {
-      "epoch": 5.002868318122555,
-      "grad_norm": 1.474095106124878,
-      "learning_rate": 9.086331658291459e-05,
-      "loss": 5.8443,
-      "step": 9593
-    },
-    {
-      "epoch": 5.003389830508475,
-      "grad_norm": 1.5613101720809937,
-      "learning_rate": 9.086231155778895e-05,
-      "loss": 5.2433,
-      "step": 9594
-    },
-    {
-      "epoch": 5.003911342894393,
-      "grad_norm": 1.460621953010559,
-      "learning_rate": 9.086130653266333e-05,
-      "loss": 5.9803,
-      "step": 9595
-    },
-    {
-      "epoch": 5.004432855280313,
-      "grad_norm": 1.5051751136779785,
-      "learning_rate": 9.08603015075377e-05,
-      "loss": 5.7192,
-      "step": 9596
-    },
-    {
-      "epoch": 5.004954367666232,
-      "grad_norm": 1.4604213237762451,
-      "learning_rate": 9.085929648241206e-05,
-      "loss": 5.3661,
-      "step": 9597
-    },
-    {
-      "epoch": 5.005475880052151,
-      "grad_norm": 1.5485050678253174,
-      "learning_rate": 9.085829145728644e-05,
-      "loss": 5.7578,
-      "step": 9598
-    },
-    {
-      "epoch": 5.00599739243807,
-      "grad_norm": 1.643455147743225,
-      "learning_rate": 9.08572864321608e-05,
-      "loss": 5.9708,
-      "step": 9599
-    },
-    {
-      "epoch": 5.00651890482399,
-      "grad_norm": 1.5654058456420898,
-      "learning_rate": 9.085628140703518e-05,
-      "loss": 5.4986,
-      "step": 9600
-    },
-    {
-      "epoch": 5.007040417209908,
-      "grad_norm": 1.4612113237380981,
-      "learning_rate": 9.085527638190954e-05,
-      "loss": 5.5646,
-      "step": 9601
-    },
-    {
-      "epoch": 5.007561929595828,
-      "grad_norm": 1.461822509765625,
-      "learning_rate": 9.085427135678392e-05,
-      "loss": 5.4933,
-      "step": 9602
-    },
-    {
-      "epoch": 5.008083441981747,
-      "grad_norm": 1.7780969142913818,
-      "learning_rate": 9.08532663316583e-05,
-      "loss": 5.0113,
-      "step": 9603
-    },
-    {
-      "epoch": 5.008604954367666,
-      "grad_norm": 1.5177302360534668,
-      "learning_rate": 9.085226130653268e-05,
-      "loss": 5.2657,
-      "step": 9604
-    },
-    {
-      "epoch": 5.009126466753585,
-      "grad_norm": 1.5507392883300781,
-      "learning_rate": 9.085125628140704e-05,
-      "loss": 5.2775,
-      "step": 9605
-    },
-    {
-      "epoch": 5.009647979139505,
-      "grad_norm": 1.4160226583480835,
-      "learning_rate": 9.085025125628142e-05,
-      "loss": 5.8781,
-      "step": 9606
-    },
-    {
-      "epoch": 5.010169491525423,
-      "grad_norm": 1.4104321002960205,
-      "learning_rate": 9.084924623115578e-05,
-      "loss": 5.7939,
-      "step": 9607
-    },
-    {
-      "epoch": 5.010691003911343,
-      "grad_norm": 1.4358967542648315,
-      "learning_rate": 9.084824120603016e-05,
-      "loss": 5.9341,
-      "step": 9608
-    },
-    {
-      "epoch": 5.011212516297262,
-      "grad_norm": 1.4435081481933594,
-      "learning_rate": 9.084723618090452e-05,
-      "loss": 5.8186,
-      "step": 9609
-    },
-    {
-      "epoch": 5.011734028683181,
-      "grad_norm": 1.4456539154052734,
-      "learning_rate": 9.084623115577889e-05,
-      "loss": 5.2494,
-      "step": 9610
-    },
-    {
-      "epoch": 5.0122555410691,
-      "grad_norm": 1.7066278457641602,
-      "learning_rate": 9.084522613065327e-05,
-      "loss": 5.3539,
-      "step": 9611
-    },
-    {
-      "epoch": 5.01277705345502,
-      "grad_norm": 1.6565117835998535,
-      "learning_rate": 9.084422110552764e-05,
-      "loss": 5.1258,
-      "step": 9612
-    },
-    {
-      "epoch": 5.013298565840938,
-      "grad_norm": 1.626996397972107,
-      "learning_rate": 9.084321608040202e-05,
-      "loss": 5.3644,
-      "step": 9613
-    },
-    {
-      "epoch": 5.013820078226858,
-      "grad_norm": 1.5126397609710693,
-      "learning_rate": 9.084221105527639e-05,
-      "loss": 5.862,
-      "step": 9614
-    },
-    {
-      "epoch": 5.014341590612777,
-      "grad_norm": 1.359494924545288,
-      "learning_rate": 9.084120603015076e-05,
-      "loss": 5.7691,
-      "step": 9615
-    },
-    {
-      "epoch": 5.014863102998696,
-      "grad_norm": 1.4283158779144287,
-      "learning_rate": 9.084020100502513e-05,
-      "loss": 5.6098,
-      "step": 9616
-    },
-    {
-      "epoch": 5.015384615384615,
-      "grad_norm": 1.448838710784912,
-      "learning_rate": 9.08391959798995e-05,
-      "loss": 5.556,
-      "step": 9617
-    },
-    {
-      "epoch": 5.015906127770535,
-      "grad_norm": 1.3984805345535278,
-      "learning_rate": 9.083819095477387e-05,
-      "loss": 5.9086,
-      "step": 9618
-    },
-    {
-      "epoch": 5.016427640156453,
-      "grad_norm": 1.3539408445358276,
-      "learning_rate": 9.083718592964825e-05,
-      "loss": 5.9561,
-      "step": 9619
-    },
-    {
-      "epoch": 5.016949152542373,
-      "grad_norm": 1.601210117340088,
-      "learning_rate": 9.083618090452261e-05,
-      "loss": 5.0634,
-      "step": 9620
-    },
-    {
-      "epoch": 5.017470664928292,
-      "grad_norm": 1.4492470026016235,
-      "learning_rate": 9.083517587939699e-05,
-      "loss": 5.5131,
-      "step": 9621
-    },
-    {
-      "epoch": 5.017992177314211,
-      "grad_norm": 1.3575685024261475,
-      "learning_rate": 9.083417085427136e-05,
-      "loss": 6.0419,
-      "step": 9622
-    },
-    {
-      "epoch": 5.01851368970013,
-      "grad_norm": 1.416258454322815,
-      "learning_rate": 9.083316582914573e-05,
-      "loss": 5.6356,
-      "step": 9623
-    },
-    {
-      "epoch": 5.01903520208605,
-      "grad_norm": 1.453626275062561,
-      "learning_rate": 9.083216080402011e-05,
-      "loss": 5.7638,
-      "step": 9624
-    },
-    {
-      "epoch": 5.019556714471968,
-      "grad_norm": 1.5545746088027954,
-      "learning_rate": 9.083115577889448e-05,
-      "loss": 5.3387,
-      "step": 9625
-    },
-    {
-      "epoch": 5.020078226857888,
-      "grad_norm": 1.53034245967865,
-      "learning_rate": 9.083015075376885e-05,
-      "loss": 5.7037,
-      "step": 9626
-    },
-    {
-      "epoch": 5.020599739243807,
-      "grad_norm": 1.3486019372940063,
-      "learning_rate": 9.082914572864322e-05,
-      "loss": 6.0162,
-      "step": 9627
-    },
-    {
-      "epoch": 5.021121251629726,
-      "grad_norm": 1.3466379642486572,
-      "learning_rate": 9.08281407035176e-05,
-      "loss": 6.1135,
-      "step": 9628
-    },
-    {
-      "epoch": 5.021642764015645,
-      "grad_norm": 1.2867568731307983,
-      "learning_rate": 9.082713567839196e-05,
-      "loss": 6.1289,
-      "step": 9629
-    },
-    {
-      "epoch": 5.022164276401565,
-      "grad_norm": 1.6420936584472656,
-      "learning_rate": 9.082613065326634e-05,
-      "loss": 5.0427,
-      "step": 9630
-    },
-    {
-      "epoch": 5.022685788787483,
-      "grad_norm": 1.4582635164260864,
-      "learning_rate": 9.08251256281407e-05,
-      "loss": 5.682,
-      "step": 9631
-    },
-    {
-      "epoch": 5.023207301173403,
-      "grad_norm": 1.5015995502471924,
-      "learning_rate": 9.082412060301508e-05,
-      "loss": 5.6702,
-      "step": 9632
-    },
-    {
-      "epoch": 5.023728813559322,
-      "grad_norm": 1.4561374187469482,
-      "learning_rate": 9.082311557788946e-05,
-      "loss": 5.7117,
-      "step": 9633
-    },
-    {
-      "epoch": 5.024250325945241,
-      "grad_norm": 1.5734344720840454,
-      "learning_rate": 9.082211055276383e-05,
-      "loss": 5.655,
-      "step": 9634
-    },
-    {
-      "epoch": 5.02477183833116,
-      "grad_norm": 1.427914023399353,
-      "learning_rate": 9.08211055276382e-05,
-      "loss": 5.9042,
-      "step": 9635
-    },
-    {
-      "epoch": 5.02529335071708,
-      "grad_norm": 1.3230676651000977,
-      "learning_rate": 9.082010050251256e-05,
-      "loss": 5.9891,
-      "step": 9636
-    },
-    {
-      "epoch": 5.025814863102998,
-      "grad_norm": 1.329751968383789,
-      "learning_rate": 9.081909547738694e-05,
-      "loss": 5.9401,
-      "step": 9637
-    },
-    {
-      "epoch": 5.026336375488918,
-      "grad_norm": 1.3566443920135498,
-      "learning_rate": 9.08180904522613e-05,
-      "loss": 5.7345,
-      "step": 9638
-    },
-    {
-      "epoch": 5.026857887874837,
-      "grad_norm": 1.5137128829956055,
-      "learning_rate": 9.081708542713568e-05,
-      "loss": 5.4097,
-      "step": 9639
-    },
-    {
-      "epoch": 5.027379400260756,
-      "grad_norm": 1.303293228149414,
-      "learning_rate": 9.081608040201005e-05,
-      "loss": 5.763,
-      "step": 9640
-    },
-    {
-      "epoch": 5.027900912646675,
-      "grad_norm": 1.4488989114761353,
-      "learning_rate": 9.081507537688443e-05,
-      "loss": 5.6348,
-      "step": 9641
-    },
-    {
-      "epoch": 5.028422425032595,
-      "grad_norm": 1.337268590927124,
-      "learning_rate": 9.081407035175879e-05,
-      "loss": 5.702,
-      "step": 9642
-    },
-    {
-      "epoch": 5.028943937418513,
-      "grad_norm": 1.5853475332260132,
-      "learning_rate": 9.081306532663317e-05,
-      "loss": 5.4197,
-      "step": 9643
-    },
-    {
-      "epoch": 5.029465449804433,
-      "grad_norm": 1.4773191213607788,
-      "learning_rate": 9.081206030150755e-05,
-      "loss": 5.7977,
-      "step": 9644
-    },
-    {
-      "epoch": 5.029986962190352,
-      "grad_norm": 1.3736628293991089,
-      "learning_rate": 9.081105527638192e-05,
-      "loss": 5.4086,
-      "step": 9645
-    },
-    {
-      "epoch": 5.030508474576271,
-      "grad_norm": 1.5422024726867676,
-      "learning_rate": 9.081005025125629e-05,
-      "loss": 5.0358,
-      "step": 9646
-    },
-    {
-      "epoch": 5.03102998696219,
-      "grad_norm": 1.5655113458633423,
-      "learning_rate": 9.080904522613067e-05,
-      "loss": 5.2046,
-      "step": 9647
-    },
-    {
-      "epoch": 5.03155149934811,
-      "grad_norm": 1.4077508449554443,
-      "learning_rate": 9.080804020100503e-05,
-      "loss": 5.9684,
-      "step": 9648
-    },
-    {
-      "epoch": 5.032073011734028,
-      "grad_norm": 1.2865736484527588,
-      "learning_rate": 9.08070351758794e-05,
-      "loss": 6.0274,
-      "step": 9649
-    },
-    {
-      "epoch": 5.032594524119948,
-      "grad_norm": 1.5737452507019043,
-      "learning_rate": 9.080603015075377e-05,
-      "loss": 5.1977,
-      "step": 9650
-    },
-    {
-      "epoch": 5.033116036505867,
-      "grad_norm": 1.5545756816864014,
-      "learning_rate": 9.080502512562814e-05,
-      "loss": 5.6668,
-      "step": 9651
-    },
-    {
-      "epoch": 5.033637548891786,
-      "grad_norm": 1.4996898174285889,
-      "learning_rate": 9.080402010050251e-05,
-      "loss": 5.5639,
-      "step": 9652
-    },
-    {
-      "epoch": 5.034159061277705,
-      "grad_norm": 1.4492385387420654,
-      "learning_rate": 9.080301507537689e-05,
-      "loss": 5.7191,
-      "step": 9653
-    },
-    {
-      "epoch": 5.034680573663625,
-      "grad_norm": 1.407344937324524,
-      "learning_rate": 9.080201005025127e-05,
-      "loss": 5.937,
-      "step": 9654
-    },
-    {
-      "epoch": 5.0352020860495434,
-      "grad_norm": 1.4051048755645752,
-      "learning_rate": 9.080100502512563e-05,
-      "loss": 4.997,
-      "step": 9655
-    },
-    {
-      "epoch": 5.035723598435463,
-      "grad_norm": 1.5161521434783936,
-      "learning_rate": 9.080000000000001e-05,
-      "loss": 5.3697,
-      "step": 9656
-    },
-    {
-      "epoch": 5.036245110821382,
-      "grad_norm": 1.473875641822815,
-      "learning_rate": 9.079899497487438e-05,
-      "loss": 5.7287,
-      "step": 9657
-    },
-    {
-      "epoch": 5.036766623207301,
-      "grad_norm": 1.3484997749328613,
-      "learning_rate": 9.079798994974875e-05,
-      "loss": 5.7778,
-      "step": 9658
-    },
-    {
-      "epoch": 5.03728813559322,
-      "grad_norm": 1.5568821430206299,
-      "learning_rate": 9.079698492462312e-05,
-      "loss": 5.7683,
-      "step": 9659
-    },
-    {
-      "epoch": 5.03780964797914,
-      "grad_norm": 1.4554647207260132,
-      "learning_rate": 9.07959798994975e-05,
-      "loss": 5.3819,
-      "step": 9660
-    },
-    {
-      "epoch": 5.0383311603650585,
-      "grad_norm": 1.455867886543274,
-      "learning_rate": 9.079497487437186e-05,
-      "loss": 5.9223,
-      "step": 9661
-    },
-    {
-      "epoch": 5.038852672750978,
-      "grad_norm": 1.4219870567321777,
-      "learning_rate": 9.079396984924624e-05,
-      "loss": 5.7912,
-      "step": 9662
-    },
-    {
-      "epoch": 5.039374185136897,
-      "grad_norm": 1.6555315256118774,
-      "learning_rate": 9.07929648241206e-05,
-      "loss": 5.3327,
-      "step": 9663
-    },
-    {
-      "epoch": 5.039895697522816,
-      "grad_norm": 1.4760990142822266,
-      "learning_rate": 9.079195979899498e-05,
-      "loss": 5.9177,
-      "step": 9664
-    },
-    {
-      "epoch": 5.040417209908735,
-      "grad_norm": 1.7738091945648193,
-      "learning_rate": 9.079095477386936e-05,
-      "loss": 5.3492,
-      "step": 9665
-    },
-    {
-      "epoch": 5.040938722294655,
-      "grad_norm": 1.5841596126556396,
-      "learning_rate": 9.078994974874372e-05,
-      "loss": 5.8397,
-      "step": 9666
-    },
-    {
-      "epoch": 5.0414602346805735,
-      "grad_norm": 1.4651416540145874,
-      "learning_rate": 9.07889447236181e-05,
-      "loss": 5.5619,
-      "step": 9667
-    },
-    {
-      "epoch": 5.041981747066493,
-      "grad_norm": 1.428672194480896,
-      "learning_rate": 9.078793969849246e-05,
-      "loss": 5.8147,
-      "step": 9668
-    },
-    {
-      "epoch": 5.042503259452412,
-      "grad_norm": 1.4610378742218018,
-      "learning_rate": 9.078693467336684e-05,
-      "loss": 5.9564,
-      "step": 9669
-    },
-    {
-      "epoch": 5.043024771838331,
-      "grad_norm": 1.5630525350570679,
-      "learning_rate": 9.07859296482412e-05,
-      "loss": 5.7911,
-      "step": 9670
-    },
-    {
-      "epoch": 5.04354628422425,
-      "grad_norm": 1.3922629356384277,
-      "learning_rate": 9.078492462311558e-05,
-      "loss": 6.1054,
-      "step": 9671
-    },
-    {
-      "epoch": 5.04406779661017,
-      "grad_norm": 1.5127930641174316,
-      "learning_rate": 9.078391959798995e-05,
-      "loss": 5.782,
-      "step": 9672
-    },
-    {
-      "epoch": 5.0445893089960885,
-      "grad_norm": 1.3915724754333496,
-      "learning_rate": 9.078291457286433e-05,
-      "loss": 5.9943,
-      "step": 9673
-    },
-    {
-      "epoch": 5.045110821382008,
-      "grad_norm": 1.3761885166168213,
-      "learning_rate": 9.078190954773869e-05,
-      "loss": 5.9184,
-      "step": 9674
-    },
-    {
-      "epoch": 5.045632333767927,
-      "grad_norm": 1.6044615507125854,
-      "learning_rate": 9.078090452261307e-05,
-      "loss": 5.4293,
-      "step": 9675
-    },
-    {
-      "epoch": 5.046153846153846,
-      "grad_norm": 1.3893239498138428,
-      "learning_rate": 9.077989949748745e-05,
-      "loss": 5.4735,
-      "step": 9676
-    },
-    {
-      "epoch": 5.0466753585397655,
-      "grad_norm": 1.4942357540130615,
-      "learning_rate": 9.077889447236181e-05,
-      "loss": 5.646,
-      "step": 9677
-    },
-    {
-      "epoch": 5.047196870925685,
-      "grad_norm": 1.6065033674240112,
-      "learning_rate": 9.077788944723619e-05,
-      "loss": 5.101,
-      "step": 9678
-    },
-    {
-      "epoch": 5.0477183833116035,
-      "grad_norm": 1.5114279985427856,
-      "learning_rate": 9.077688442211055e-05,
-      "loss": 5.7689,
-      "step": 9679
-    },
-    {
-      "epoch": 5.048239895697523,
-      "grad_norm": 1.7003482580184937,
-      "learning_rate": 9.077587939698493e-05,
-      "loss": 4.7868,
-      "step": 9680
-    },
-    {
-      "epoch": 5.0487614080834415,
-      "grad_norm": 1.5474112033843994,
-      "learning_rate": 9.07748743718593e-05,
-      "loss": 5.74,
-      "step": 9681
-    },
-    {
-      "epoch": 5.049282920469361,
-      "grad_norm": 1.630046010017395,
-      "learning_rate": 9.077386934673367e-05,
-      "loss": 5.4271,
-      "step": 9682
-    },
-    {
-      "epoch": 5.0498044328552805,
-      "grad_norm": 1.3415035009384155,
-      "learning_rate": 9.077286432160804e-05,
-      "loss": 5.7669,
-      "step": 9683
-    },
-    {
-      "epoch": 5.050325945241199,
-      "grad_norm": 2.007796049118042,
-      "learning_rate": 9.077185929648241e-05,
-      "loss": 5.1961,
-      "step": 9684
-    },
-    {
-      "epoch": 5.0508474576271185,
-      "grad_norm": 1.484814167022705,
-      "learning_rate": 9.077085427135679e-05,
-      "loss": 5.5597,
-      "step": 9685
-    },
-    {
-      "epoch": 5.051368970013038,
-      "grad_norm": 1.4897054433822632,
-      "learning_rate": 9.076984924623117e-05,
-      "loss": 5.5614,
-      "step": 9686
-    },
-    {
-      "epoch": 5.0518904823989566,
-      "grad_norm": 1.5001835823059082,
-      "learning_rate": 9.076884422110553e-05,
-      "loss": 5.451,
-      "step": 9687
-    },
-    {
-      "epoch": 5.052411994784876,
-      "grad_norm": 1.3624945878982544,
-      "learning_rate": 9.076783919597991e-05,
-      "loss": 5.7762,
-      "step": 9688
-    },
-    {
-      "epoch": 5.0529335071707955,
-      "grad_norm": 1.6024837493896484,
-      "learning_rate": 9.076683417085428e-05,
-      "loss": 5.2915,
-      "step": 9689
-    },
-    {
-      "epoch": 5.053455019556714,
-      "grad_norm": 1.3378024101257324,
-      "learning_rate": 9.076582914572864e-05,
-      "loss": 5.8693,
-      "step": 9690
-    },
-    {
-      "epoch": 5.0539765319426335,
-      "grad_norm": 1.4738073348999023,
-      "learning_rate": 9.076482412060302e-05,
-      "loss": 5.6446,
-      "step": 9691
-    },
-    {
-      "epoch": 5.054498044328553,
-      "grad_norm": 1.3401093482971191,
-      "learning_rate": 9.076381909547738e-05,
-      "loss": 5.6551,
-      "step": 9692
-    },
-    {
-      "epoch": 5.055019556714472,
-      "grad_norm": 1.4309895038604736,
-      "learning_rate": 9.076281407035176e-05,
-      "loss": 6.0019,
-      "step": 9693
-    },
-    {
-      "epoch": 5.055541069100391,
-      "grad_norm": 1.6626337766647339,
-      "learning_rate": 9.076180904522613e-05,
-      "loss": 5.2261,
-      "step": 9694
-    },
-    {
-      "epoch": 5.0560625814863105,
-      "grad_norm": 1.3755441904067993,
-      "learning_rate": 9.07608040201005e-05,
-      "loss": 5.6063,
-      "step": 9695
-    },
-    {
-      "epoch": 5.056584093872229,
-      "grad_norm": 1.4033710956573486,
-      "learning_rate": 9.075979899497488e-05,
-      "loss": 5.8309,
-      "step": 9696
-    },
-    {
-      "epoch": 5.0571056062581485,
-      "grad_norm": 1.540905475616455,
-      "learning_rate": 9.075879396984926e-05,
-      "loss": 4.8308,
-      "step": 9697
-    },
-    {
-      "epoch": 5.057627118644068,
-      "grad_norm": 1.4502434730529785,
-      "learning_rate": 9.075778894472362e-05,
-      "loss": 5.1935,
-      "step": 9698
-    },
-    {
-      "epoch": 5.058148631029987,
-      "grad_norm": 1.5704296827316284,
-      "learning_rate": 9.0756783919598e-05,
-      "loss": 5.7042,
-      "step": 9699
-    },
-    {
-      "epoch": 5.058670143415906,
-      "grad_norm": 1.5186989307403564,
-      "learning_rate": 9.075577889447236e-05,
-      "loss": 5.3473,
-      "step": 9700
-    },
-    {
-      "epoch": 5.0591916558018255,
-      "grad_norm": 1.3873562812805176,
-      "learning_rate": 9.075477386934674e-05,
-      "loss": 5.969,
-      "step": 9701
-    },
-    {
-      "epoch": 5.059713168187744,
-      "grad_norm": 1.4044115543365479,
-      "learning_rate": 9.075376884422111e-05,
-      "loss": 6.0869,
-      "step": 9702
-    },
-    {
-      "epoch": 5.0602346805736635,
-      "grad_norm": 1.7077969312667847,
-      "learning_rate": 9.075276381909547e-05,
-      "loss": 5.6628,
-      "step": 9703
-    },
-    {
-      "epoch": 5.060756192959583,
-      "grad_norm": 1.5040206909179688,
-      "learning_rate": 9.075175879396985e-05,
-      "loss": 5.7937,
-      "step": 9704
-    },
-    {
-      "epoch": 5.061277705345502,
-      "grad_norm": 1.373559594154358,
-      "learning_rate": 9.075075376884423e-05,
-      "loss": 5.6147,
-      "step": 9705
-    },
-    {
-      "epoch": 5.061799217731421,
-      "grad_norm": 1.6548230648040771,
-      "learning_rate": 9.07497487437186e-05,
-      "loss": 4.9918,
-      "step": 9706
-    },
-    {
-      "epoch": 5.0623207301173405,
-      "grad_norm": 1.4347732067108154,
-      "learning_rate": 9.074874371859297e-05,
-      "loss": 5.6706,
-      "step": 9707
-    },
-    {
-      "epoch": 5.062842242503259,
-      "grad_norm": 1.3914793729782104,
-      "learning_rate": 9.074773869346735e-05,
-      "loss": 5.8049,
-      "step": 9708
-    },
-    {
-      "epoch": 5.063363754889179,
-      "grad_norm": 1.471257209777832,
-      "learning_rate": 9.074673366834171e-05,
-      "loss": 5.2838,
-      "step": 9709
-    },
-    {
-      "epoch": 5.063885267275098,
-      "grad_norm": 1.4206657409667969,
-      "learning_rate": 9.074572864321609e-05,
-      "loss": 5.6992,
-      "step": 9710
-    },
-    {
-      "epoch": 5.064406779661017,
-      "grad_norm": 1.539571762084961,
-      "learning_rate": 9.074472361809045e-05,
-      "loss": 5.1258,
-      "step": 9711
-    },
-    {
-      "epoch": 5.064928292046936,
-      "grad_norm": 1.5601158142089844,
-      "learning_rate": 9.074371859296483e-05,
-      "loss": 5.8603,
-      "step": 9712
-    },
-    {
-      "epoch": 5.0654498044328555,
-      "grad_norm": 1.6400681734085083,
-      "learning_rate": 9.07427135678392e-05,
-      "loss": 5.3704,
-      "step": 9713
-    },
-    {
-      "epoch": 5.065971316818774,
-      "grad_norm": 1.572854995727539,
-      "learning_rate": 9.074170854271357e-05,
-      "loss": 5.7176,
-      "step": 9714
-    },
-    {
-      "epoch": 5.066492829204694,
-      "grad_norm": 1.5074845552444458,
-      "learning_rate": 9.074070351758794e-05,
-      "loss": 4.9472,
-      "step": 9715
-    },
-    {
-      "epoch": 5.067014341590613,
-      "grad_norm": 1.4631696939468384,
-      "learning_rate": 9.073969849246232e-05,
-      "loss": 5.2816,
-      "step": 9716
-    },
-    {
-      "epoch": 5.067535853976532,
-      "grad_norm": 1.378720998764038,
-      "learning_rate": 9.073869346733669e-05,
-      "loss": 5.6105,
-      "step": 9717
-    },
-    {
-      "epoch": 5.068057366362451,
-      "grad_norm": 1.5094642639160156,
-      "learning_rate": 9.073768844221106e-05,
-      "loss": 5.4617,
-      "step": 9718
-    },
-    {
-      "epoch": 5.0685788787483705,
-      "grad_norm": 1.3430471420288086,
-      "learning_rate": 9.073668341708544e-05,
-      "loss": 4.9771,
-      "step": 9719
-    },
-    {
-      "epoch": 5.069100391134289,
-      "grad_norm": 1.631402611732483,
-      "learning_rate": 9.07356783919598e-05,
-      "loss": 5.3833,
-      "step": 9720
-    },
-    {
-      "epoch": 5.069621903520209,
-      "grad_norm": 1.3520491123199463,
-      "learning_rate": 9.073467336683418e-05,
-      "loss": 5.8105,
-      "step": 9721
-    },
-    {
-      "epoch": 5.070143415906128,
-      "grad_norm": 1.4429627656936646,
-      "learning_rate": 9.073366834170854e-05,
-      "loss": 5.8603,
-      "step": 9722
-    },
-    {
-      "epoch": 5.070664928292047,
-      "grad_norm": 1.3772015571594238,
-      "learning_rate": 9.073266331658292e-05,
-      "loss": 5.8779,
-      "step": 9723
-    },
-    {
-      "epoch": 5.071186440677966,
-      "grad_norm": 1.4569953680038452,
-      "learning_rate": 9.073165829145728e-05,
-      "loss": 5.7043,
-      "step": 9724
-    },
-    {
-      "epoch": 5.0717079530638856,
-      "grad_norm": 1.454325556755066,
-      "learning_rate": 9.073065326633166e-05,
-      "loss": 5.9363,
-      "step": 9725
-    },
-    {
-      "epoch": 5.072229465449804,
-      "grad_norm": 1.496927261352539,
-      "learning_rate": 9.072964824120604e-05,
-      "loss": 5.69,
-      "step": 9726
-    },
-    {
-      "epoch": 5.072750977835724,
-      "grad_norm": 1.3535856008529663,
-      "learning_rate": 9.072864321608042e-05,
-      "loss": 5.5874,
-      "step": 9727
-    },
-    {
-      "epoch": 5.073272490221643,
-      "grad_norm": 1.4858235120773315,
-      "learning_rate": 9.072763819095478e-05,
-      "loss": 5.4547,
-      "step": 9728
-    },
-    {
-      "epoch": 5.073794002607562,
-      "grad_norm": 1.5924283266067505,
-      "learning_rate": 9.072663316582915e-05,
-      "loss": 4.9759,
-      "step": 9729
-    },
-    {
-      "epoch": 5.074315514993481,
-      "grad_norm": 1.4746376276016235,
-      "learning_rate": 9.072562814070352e-05,
-      "loss": 5.9143,
-      "step": 9730
-    },
-    {
-      "epoch": 5.074837027379401,
-      "grad_norm": 1.6171820163726807,
-      "learning_rate": 9.072462311557789e-05,
-      "loss": 5.4089,
-      "step": 9731
-    },
-    {
-      "epoch": 5.075358539765319,
-      "grad_norm": 1.360903024673462,
-      "learning_rate": 9.072361809045227e-05,
-      "loss": 5.8715,
-      "step": 9732
-    },
-    {
-      "epoch": 5.075880052151239,
-      "grad_norm": 1.4499837160110474,
-      "learning_rate": 9.072261306532663e-05,
-      "loss": 5.8175,
-      "step": 9733
-    },
-    {
-      "epoch": 5.076401564537158,
-      "grad_norm": 1.4684085845947266,
-      "learning_rate": 9.072160804020101e-05,
-      "loss": 5.8982,
-      "step": 9734
-    },
-    {
-      "epoch": 5.076923076923077,
-      "grad_norm": 1.41493558883667,
-      "learning_rate": 9.072060301507537e-05,
-      "loss": 5.872,
-      "step": 9735
-    },
-    {
-      "epoch": 5.077444589308996,
-      "grad_norm": 1.5447040796279907,
-      "learning_rate": 9.071959798994975e-05,
-      "loss": 5.7936,
-      "step": 9736
-    },
-    {
-      "epoch": 5.077966101694916,
-      "grad_norm": 1.4709868431091309,
-      "learning_rate": 9.071859296482413e-05,
-      "loss": 5.6794,
-      "step": 9737
-    },
-    {
-      "epoch": 5.078487614080834,
-      "grad_norm": 1.5439459085464478,
-      "learning_rate": 9.07175879396985e-05,
-      "loss": 5.6377,
-      "step": 9738
-    },
-    {
-      "epoch": 5.079009126466754,
-      "grad_norm": 1.4479284286499023,
-      "learning_rate": 9.071658291457287e-05,
-      "loss": 5.931,
-      "step": 9739
-    },
-    {
-      "epoch": 5.079530638852673,
-      "grad_norm": 1.6863757371902466,
-      "learning_rate": 9.071557788944725e-05,
-      "loss": 4.592,
-      "step": 9740
-    },
-    {
-      "epoch": 5.080052151238592,
-      "grad_norm": 1.4134440422058105,
-      "learning_rate": 9.071457286432161e-05,
-      "loss": 5.6923,
-      "step": 9741
-    },
-    {
-      "epoch": 5.080573663624511,
-      "grad_norm": 1.5288927555084229,
-      "learning_rate": 9.071356783919599e-05,
-      "loss": 5.9839,
-      "step": 9742
-    },
-    {
-      "epoch": 5.081095176010431,
-      "grad_norm": 1.362660527229309,
-      "learning_rate": 9.071256281407035e-05,
-      "loss": 5.6391,
-      "step": 9743
-    },
-    {
-      "epoch": 5.081616688396349,
-      "grad_norm": 1.488044261932373,
-      "learning_rate": 9.071155778894472e-05,
-      "loss": 5.5249,
-      "step": 9744
-    },
-    {
-      "epoch": 5.082138200782269,
-      "grad_norm": 1.5648248195648193,
-      "learning_rate": 9.07105527638191e-05,
-      "loss": 5.6881,
-      "step": 9745
-    },
-    {
-      "epoch": 5.082659713168188,
-      "grad_norm": 1.4697654247283936,
-      "learning_rate": 9.070954773869347e-05,
-      "loss": 5.6773,
-      "step": 9746
-    },
-    {
-      "epoch": 5.083181225554107,
-      "grad_norm": 1.4515063762664795,
-      "learning_rate": 9.070854271356785e-05,
-      "loss": 5.7245,
-      "step": 9747
-    },
-    {
-      "epoch": 5.083702737940026,
-      "grad_norm": 1.3670313358306885,
-      "learning_rate": 9.070753768844222e-05,
-      "loss": 4.9729,
-      "step": 9748
-    },
-    {
-      "epoch": 5.084224250325946,
-      "grad_norm": 1.3423715829849243,
-      "learning_rate": 9.07065326633166e-05,
-      "loss": 5.6593,
-      "step": 9749
-    },
-    {
-      "epoch": 5.084745762711864,
-      "grad_norm": 1.4635305404663086,
-      "learning_rate": 9.070552763819096e-05,
-      "loss": 5.7544,
-      "step": 9750
-    },
-    {
-      "epoch": 5.085267275097784,
-      "grad_norm": 1.501207947731018,
-      "learning_rate": 9.070452261306534e-05,
-      "loss": 5.6287,
-      "step": 9751
-    },
-    {
-      "epoch": 5.085788787483703,
-      "grad_norm": 1.4701714515686035,
-      "learning_rate": 9.07035175879397e-05,
-      "loss": 5.6029,
-      "step": 9752
-    },
-    {
-      "epoch": 5.086310299869622,
-      "grad_norm": 1.4711005687713623,
-      "learning_rate": 9.070251256281408e-05,
-      "loss": 5.4826,
-      "step": 9753
-    },
-    {
-      "epoch": 5.086831812255541,
-      "grad_norm": 1.6107921600341797,
-      "learning_rate": 9.070150753768844e-05,
-      "loss": 5.3437,
-      "step": 9754
-    },
-    {
-      "epoch": 5.087353324641461,
-      "grad_norm": 1.395771861076355,
-      "learning_rate": 9.070050251256282e-05,
-      "loss": 4.7987,
-      "step": 9755
-    },
-    {
-      "epoch": 5.087874837027379,
-      "grad_norm": 1.4143706560134888,
-      "learning_rate": 9.069949748743718e-05,
-      "loss": 5.4825,
-      "step": 9756
-    },
-    {
-      "epoch": 5.088396349413299,
-      "grad_norm": 1.3993958234786987,
-      "learning_rate": 9.069849246231156e-05,
-      "loss": 6.0312,
-      "step": 9757
-    },
-    {
-      "epoch": 5.088917861799218,
-      "grad_norm": 1.4078961610794067,
-      "learning_rate": 9.069748743718594e-05,
-      "loss": 5.5264,
-      "step": 9758
-    },
-    {
-      "epoch": 5.089439374185137,
-      "grad_norm": 1.431488275527954,
-      "learning_rate": 9.06964824120603e-05,
-      "loss": 5.1128,
-      "step": 9759
-    },
-    {
-      "epoch": 5.089960886571056,
-      "grad_norm": 1.3280187845230103,
-      "learning_rate": 9.069547738693468e-05,
-      "loss": 6.0015,
-      "step": 9760
-    },
-    {
-      "epoch": 5.090482398956976,
-      "grad_norm": 1.5541555881500244,
-      "learning_rate": 9.069447236180905e-05,
-      "loss": 5.1856,
-      "step": 9761
-    },
-    {
-      "epoch": 5.091003911342894,
-      "grad_norm": 1.4489773511886597,
-      "learning_rate": 9.069346733668342e-05,
-      "loss": 6.1718,
-      "step": 9762
-    },
-    {
-      "epoch": 5.091525423728814,
-      "grad_norm": 1.4081870317459106,
-      "learning_rate": 9.069246231155779e-05,
-      "loss": 5.868,
-      "step": 9763
-    },
-    {
-      "epoch": 5.092046936114733,
-      "grad_norm": 1.4468175172805786,
-      "learning_rate": 9.069145728643217e-05,
-      "loss": 5.421,
-      "step": 9764
-    },
-    {
-      "epoch": 5.092568448500652,
-      "grad_norm": 1.471042513847351,
-      "learning_rate": 9.069045226130653e-05,
-      "loss": 5.3036,
-      "step": 9765
-    },
-    {
-      "epoch": 5.093089960886571,
-      "grad_norm": 1.7339308261871338,
-      "learning_rate": 9.068944723618091e-05,
-      "loss": 5.3725,
-      "step": 9766
-    },
-    {
-      "epoch": 5.093611473272491,
-      "grad_norm": 1.4780833721160889,
-      "learning_rate": 9.068844221105529e-05,
-      "loss": 5.8853,
-      "step": 9767
-    },
-    {
-      "epoch": 5.094132985658409,
-      "grad_norm": 1.4603228569030762,
-      "learning_rate": 9.068743718592966e-05,
-      "loss": 5.767,
-      "step": 9768
-    },
-    {
-      "epoch": 5.094654498044329,
-      "grad_norm": 1.62308669090271,
-      "learning_rate": 9.068643216080403e-05,
-      "loss": 5.4449,
-      "step": 9769
-    },
-    {
-      "epoch": 5.095176010430248,
-      "grad_norm": 1.5132569074630737,
-      "learning_rate": 9.068542713567839e-05,
-      "loss": 5.761,
-      "step": 9770
-    },
-    {
-      "epoch": 5.095697522816167,
-      "grad_norm": 1.355875849723816,
-      "learning_rate": 9.068442211055277e-05,
-      "loss": 6.0694,
-      "step": 9771
-    },
-    {
-      "epoch": 5.096219035202086,
-      "grad_norm": 1.4979161024093628,
-      "learning_rate": 9.068341708542713e-05,
-      "loss": 5.7338,
-      "step": 9772
-    },
-    {
-      "epoch": 5.096740547588006,
-      "grad_norm": 1.5663098096847534,
-      "learning_rate": 9.068241206030151e-05,
-      "loss": 4.9289,
-      "step": 9773
-    },
-    {
-      "epoch": 5.097262059973924,
-      "grad_norm": 1.4345723390579224,
-      "learning_rate": 9.068140703517588e-05,
-      "loss": 5.3931,
-      "step": 9774
-    },
-    {
-      "epoch": 5.097783572359844,
-      "grad_norm": 1.4108630418777466,
-      "learning_rate": 9.068040201005025e-05,
-      "loss": 5.8292,
-      "step": 9775
-    },
-    {
-      "epoch": 5.098305084745762,
-      "grad_norm": 1.6083283424377441,
-      "learning_rate": 9.067939698492462e-05,
-      "loss": 5.3571,
-      "step": 9776
-    },
-    {
-      "epoch": 5.098826597131682,
-      "grad_norm": 1.486512303352356,
-      "learning_rate": 9.0678391959799e-05,
-      "loss": 5.8291,
-      "step": 9777
-    },
-    {
-      "epoch": 5.099348109517601,
-      "grad_norm": 1.3981120586395264,
-      "learning_rate": 9.067738693467337e-05,
-      "loss": 5.107,
-      "step": 9778
-    },
-    {
-      "epoch": 5.09986962190352,
-      "grad_norm": 1.4322900772094727,
-      "learning_rate": 9.067638190954775e-05,
-      "loss": 5.511,
-      "step": 9779
-    },
-    {
-      "epoch": 5.100391134289439,
-      "grad_norm": 1.3672746419906616,
-      "learning_rate": 9.067537688442212e-05,
-      "loss": 5.6516,
-      "step": 9780
-    },
-    {
-      "epoch": 5.100912646675359,
-      "grad_norm": 1.4941037893295288,
-      "learning_rate": 9.06743718592965e-05,
-      "loss": 6.0601,
-      "step": 9781
-    },
-    {
-      "epoch": 5.101434159061277,
-      "grad_norm": 1.463607907295227,
-      "learning_rate": 9.067336683417086e-05,
-      "loss": 5.7807,
-      "step": 9782
-    },
-    {
-      "epoch": 5.101955671447197,
-      "grad_norm": 1.5489896535873413,
-      "learning_rate": 9.067236180904522e-05,
-      "loss": 5.5802,
-      "step": 9783
-    },
-    {
-      "epoch": 5.102477183833116,
-      "grad_norm": 1.5914478302001953,
-      "learning_rate": 9.06713567839196e-05,
-      "loss": 5.0058,
-      "step": 9784
-    },
-    {
-      "epoch": 5.102998696219035,
-      "grad_norm": 1.4818017482757568,
-      "learning_rate": 9.067035175879397e-05,
-      "loss": 5.8193,
-      "step": 9785
-    },
-    {
-      "epoch": 5.103520208604954,
-      "grad_norm": 1.3809045553207397,
-      "learning_rate": 9.066934673366834e-05,
-      "loss": 5.6353,
-      "step": 9786
-    },
-    {
-      "epoch": 5.104041720990874,
-      "grad_norm": 1.4551916122436523,
-      "learning_rate": 9.066834170854272e-05,
-      "loss": 5.9308,
-      "step": 9787
-    },
-    {
-      "epoch": 5.104563233376792,
-      "grad_norm": 1.4862685203552246,
-      "learning_rate": 9.06673366834171e-05,
-      "loss": 5.4249,
-      "step": 9788
-    },
-    {
-      "epoch": 5.105084745762712,
-      "grad_norm": 1.3899887800216675,
-      "learning_rate": 9.066633165829146e-05,
-      "loss": 5.8669,
-      "step": 9789
-    },
-    {
-      "epoch": 5.105606258148631,
-      "grad_norm": 1.5009490251541138,
-      "learning_rate": 9.066532663316584e-05,
-      "loss": 5.7767,
-      "step": 9790
-    },
-    {
-      "epoch": 5.10612777053455,
-      "grad_norm": 1.5084518194198608,
-      "learning_rate": 9.06643216080402e-05,
-      "loss": 5.665,
-      "step": 9791
-    },
-    {
-      "epoch": 5.106649282920469,
-      "grad_norm": 1.4969829320907593,
-      "learning_rate": 9.066331658291458e-05,
-      "loss": 5.2416,
-      "step": 9792
-    },
-    {
-      "epoch": 5.107170795306389,
-      "grad_norm": 1.8450356721878052,
-      "learning_rate": 9.066231155778895e-05,
-      "loss": 5.2188,
-      "step": 9793
-    },
-    {
-      "epoch": 5.107692307692307,
-      "grad_norm": 1.4782931804656982,
-      "learning_rate": 9.066130653266333e-05,
-      "loss": 5.8377,
-      "step": 9794
-    },
-    {
-      "epoch": 5.108213820078227,
-      "grad_norm": 1.5405880212783813,
-      "learning_rate": 9.066030150753769e-05,
-      "loss": 5.2095,
-      "step": 9795
-    },
-    {
-      "epoch": 5.108735332464146,
-      "grad_norm": 1.4334425926208496,
-      "learning_rate": 9.065929648241205e-05,
-      "loss": 5.8627,
-      "step": 9796
-    },
-    {
-      "epoch": 5.109256844850065,
-      "grad_norm": 1.375910758972168,
-      "learning_rate": 9.065829145728643e-05,
-      "loss": 5.6561,
-      "step": 9797
-    },
-    {
-      "epoch": 5.109778357235984,
-      "grad_norm": 1.5919585227966309,
-      "learning_rate": 9.065728643216081e-05,
-      "loss": 5.1581,
-      "step": 9798
-    },
-    {
-      "epoch": 5.110299869621904,
-      "grad_norm": 1.446583867073059,
-      "learning_rate": 9.065628140703519e-05,
-      "loss": 5.5144,
-      "step": 9799
-    },
-    {
-      "epoch": 5.110821382007822,
-      "grad_norm": 1.5610660314559937,
-      "learning_rate": 9.065527638190955e-05,
-      "loss": 5.0388,
-      "step": 9800
-    },
-    {
-      "epoch": 5.111342894393742,
-      "grad_norm": 1.4241214990615845,
-      "learning_rate": 9.065427135678393e-05,
-      "loss": 5.1163,
-      "step": 9801
-    },
-    {
-      "epoch": 5.111864406779661,
-      "grad_norm": 1.4626935720443726,
-      "learning_rate": 9.06532663316583e-05,
-      "loss": 5.9674,
-      "step": 9802
-    },
-    {
-      "epoch": 5.11238591916558,
-      "grad_norm": 1.5444343090057373,
-      "learning_rate": 9.065226130653267e-05,
-      "loss": 5.8328,
-      "step": 9803
-    },
-    {
-      "epoch": 5.112907431551499,
-      "grad_norm": 1.5496864318847656,
-      "learning_rate": 9.065125628140704e-05,
-      "loss": 5.9242,
-      "step": 9804
-    },
-    {
-      "epoch": 5.113428943937419,
-      "grad_norm": 1.4796339273452759,
-      "learning_rate": 9.065025125628141e-05,
-      "loss": 5.872,
-      "step": 9805
-    },
-    {
-      "epoch": 5.113950456323337,
-      "grad_norm": 1.5841716527938843,
-      "learning_rate": 9.064924623115578e-05,
-      "loss": 4.9304,
-      "step": 9806
-    },
-    {
-      "epoch": 5.114471968709257,
-      "grad_norm": 1.5000571012496948,
-      "learning_rate": 9.064824120603016e-05,
-      "loss": 5.5772,
-      "step": 9807
-    },
-    {
-      "epoch": 5.114993481095176,
-      "grad_norm": 1.4948301315307617,
-      "learning_rate": 9.064723618090453e-05,
-      "loss": 5.7241,
-      "step": 9808
-    },
-    {
-      "epoch": 5.115514993481095,
-      "grad_norm": 1.3835678100585938,
-      "learning_rate": 9.06462311557789e-05,
-      "loss": 6.0186,
-      "step": 9809
-    },
-    {
-      "epoch": 5.116036505867014,
-      "grad_norm": 1.462010145187378,
-      "learning_rate": 9.064522613065328e-05,
-      "loss": 5.5947,
-      "step": 9810
-    },
-    {
-      "epoch": 5.116558018252934,
-      "grad_norm": 1.33342707157135,
-      "learning_rate": 9.064422110552764e-05,
-      "loss": 5.9445,
-      "step": 9811
-    },
-    {
-      "epoch": 5.117079530638852,
-      "grad_norm": 1.3512498140335083,
-      "learning_rate": 9.064321608040202e-05,
-      "loss": 5.8685,
-      "step": 9812
-    },
-    {
-      "epoch": 5.117601043024772,
-      "grad_norm": 1.4778683185577393,
-      "learning_rate": 9.064221105527638e-05,
-      "loss": 5.7171,
-      "step": 9813
-    },
-    {
-      "epoch": 5.118122555410691,
-      "grad_norm": 1.4596741199493408,
-      "learning_rate": 9.064120603015076e-05,
-      "loss": 5.8663,
-      "step": 9814
-    },
-    {
-      "epoch": 5.11864406779661,
-      "grad_norm": 1.5302433967590332,
-      "learning_rate": 9.064020100502512e-05,
-      "loss": 5.544,
-      "step": 9815
-    },
-    {
-      "epoch": 5.119165580182529,
-      "grad_norm": 1.4922175407409668,
-      "learning_rate": 9.06391959798995e-05,
-      "loss": 5.5738,
-      "step": 9816
-    },
-    {
-      "epoch": 5.119687092568449,
-      "grad_norm": 1.572393536567688,
-      "learning_rate": 9.063819095477387e-05,
-      "loss": 5.6146,
-      "step": 9817
-    },
-    {
-      "epoch": 5.120208604954367,
-      "grad_norm": 1.3849321603775024,
-      "learning_rate": 9.063718592964824e-05,
-      "loss": 5.56,
-      "step": 9818
-    },
-    {
-      "epoch": 5.120730117340287,
-      "grad_norm": 1.4509801864624023,
-      "learning_rate": 9.063618090452262e-05,
-      "loss": 5.5305,
-      "step": 9819
-    },
-    {
-      "epoch": 5.121251629726206,
-      "grad_norm": 1.4161043167114258,
-      "learning_rate": 9.0635175879397e-05,
-      "loss": 5.737,
-      "step": 9820
-    },
-    {
-      "epoch": 5.121773142112125,
-      "grad_norm": 1.4041106700897217,
-      "learning_rate": 9.063417085427136e-05,
-      "loss": 5.5359,
-      "step": 9821
-    },
-    {
-      "epoch": 5.122294654498044,
-      "grad_norm": 1.411484718322754,
-      "learning_rate": 9.063316582914573e-05,
-      "loss": 5.5164,
-      "step": 9822
-    },
-    {
-      "epoch": 5.122816166883964,
-      "grad_norm": 1.4705920219421387,
-      "learning_rate": 9.06321608040201e-05,
-      "loss": 5.7782,
-      "step": 9823
-    },
-    {
-      "epoch": 5.123337679269882,
-      "grad_norm": 1.74918794631958,
-      "learning_rate": 9.063115577889447e-05,
-      "loss": 4.8537,
-      "step": 9824
-    },
-    {
-      "epoch": 5.123859191655802,
-      "grad_norm": 1.4819347858428955,
-      "learning_rate": 9.063015075376885e-05,
-      "loss": 5.8837,
-      "step": 9825
-    },
-    {
-      "epoch": 5.124380704041721,
-      "grad_norm": 1.3932563066482544,
-      "learning_rate": 9.062914572864321e-05,
-      "loss": 5.4595,
-      "step": 9826
-    },
-    {
-      "epoch": 5.12490221642764,
-      "grad_norm": 1.4674808979034424,
-      "learning_rate": 9.062814070351759e-05,
-      "loss": 5.9844,
-      "step": 9827
-    },
-    {
-      "epoch": 5.125423728813559,
-      "grad_norm": 1.5516093969345093,
-      "learning_rate": 9.062713567839195e-05,
-      "loss": 5.268,
-      "step": 9828
-    },
-    {
-      "epoch": 5.125945241199479,
-      "grad_norm": 1.5034351348876953,
-      "learning_rate": 9.062613065326633e-05,
-      "loss": 5.4152,
-      "step": 9829
-    },
-    {
-      "epoch": 5.126466753585397,
-      "grad_norm": 1.4039462804794312,
-      "learning_rate": 9.062512562814071e-05,
-      "loss": 5.773,
-      "step": 9830
-    },
-    {
-      "epoch": 5.126988265971317,
-      "grad_norm": 1.5145410299301147,
-      "learning_rate": 9.062412060301509e-05,
-      "loss": 5.9328,
-      "step": 9831
-    },
-    {
-      "epoch": 5.127509778357236,
-      "grad_norm": 1.4692150354385376,
-      "learning_rate": 9.062311557788945e-05,
-      "loss": 5.7311,
-      "step": 9832
-    },
-    {
-      "epoch": 5.128031290743155,
-      "grad_norm": 1.4548693895339966,
-      "learning_rate": 9.062211055276383e-05,
-      "loss": 5.6759,
-      "step": 9833
-    },
-    {
-      "epoch": 5.128552803129074,
-      "grad_norm": 1.3963068723678589,
-      "learning_rate": 9.06211055276382e-05,
-      "loss": 5.56,
-      "step": 9834
-    },
-    {
-      "epoch": 5.129074315514994,
-      "grad_norm": 1.5062861442565918,
-      "learning_rate": 9.062010050251257e-05,
-      "loss": 5.3152,
-      "step": 9835
-    },
-    {
-      "epoch": 5.129595827900912,
-      "grad_norm": 1.4415667057037354,
-      "learning_rate": 9.061909547738694e-05,
-      "loss": 5.9564,
-      "step": 9836
-    },
-    {
-      "epoch": 5.130117340286832,
-      "grad_norm": 1.4332231283187866,
-      "learning_rate": 9.06180904522613e-05,
-      "loss": 6.0575,
-      "step": 9837
-    },
-    {
-      "epoch": 5.130638852672751,
-      "grad_norm": 1.7090414762496948,
-      "learning_rate": 9.061708542713568e-05,
-      "loss": 4.7546,
-      "step": 9838
-    },
-    {
-      "epoch": 5.13116036505867,
-      "grad_norm": 1.496900200843811,
-      "learning_rate": 9.061608040201006e-05,
-      "loss": 5.7875,
-      "step": 9839
-    },
-    {
-      "epoch": 5.131681877444589,
-      "grad_norm": 1.3233397006988525,
-      "learning_rate": 9.061507537688443e-05,
-      "loss": 5.5907,
-      "step": 9840
-    },
-    {
-      "epoch": 5.132203389830509,
-      "grad_norm": 1.2502939701080322,
-      "learning_rate": 9.06140703517588e-05,
-      "loss": 6.0124,
-      "step": 9841
-    },
-    {
-      "epoch": 5.132724902216427,
-      "grad_norm": 1.5362752676010132,
-      "learning_rate": 9.061306532663318e-05,
-      "loss": 4.901,
-      "step": 9842
-    },
-    {
-      "epoch": 5.133246414602347,
-      "grad_norm": 1.4712611436843872,
-      "learning_rate": 9.061206030150754e-05,
-      "loss": 5.9767,
-      "step": 9843
-    },
-    {
-      "epoch": 5.133767926988266,
-      "grad_norm": 1.4584966897964478,
-      "learning_rate": 9.061105527638192e-05,
-      "loss": 6.128,
-      "step": 9844
-    },
-    {
-      "epoch": 5.134289439374185,
-      "grad_norm": 1.527664065361023,
-      "learning_rate": 9.061005025125628e-05,
-      "loss": 5.3689,
-      "step": 9845
-    },
-    {
-      "epoch": 5.134810951760104,
-      "grad_norm": 1.400612711906433,
-      "learning_rate": 9.060904522613066e-05,
-      "loss": 5.9846,
-      "step": 9846
-    },
-    {
-      "epoch": 5.135332464146024,
-      "grad_norm": 1.4903448820114136,
-      "learning_rate": 9.060804020100502e-05,
-      "loss": 5.9402,
-      "step": 9847
-    },
-    {
-      "epoch": 5.135853976531942,
-      "grad_norm": 1.4752806425094604,
-      "learning_rate": 9.06070351758794e-05,
-      "loss": 4.8726,
-      "step": 9848
-    },
-    {
-      "epoch": 5.136375488917862,
-      "grad_norm": 1.3786475658416748,
-      "learning_rate": 9.060603015075377e-05,
-      "loss": 5.896,
-      "step": 9849
-    },
-    {
-      "epoch": 5.136897001303781,
-      "grad_norm": 1.4106947183609009,
-      "learning_rate": 9.060502512562814e-05,
-      "loss": 6.1492,
-      "step": 9850
-    },
-    {
-      "epoch": 5.1374185136897,
-      "grad_norm": 1.6404285430908203,
-      "learning_rate": 9.060402010050252e-05,
-      "loss": 4.8631,
-      "step": 9851
-    },
-    {
-      "epoch": 5.137940026075619,
-      "grad_norm": 1.4154287576675415,
-      "learning_rate": 9.060301507537689e-05,
-      "loss": 6.0008,
-      "step": 9852
-    },
-    {
-      "epoch": 5.138461538461539,
-      "grad_norm": 1.669723391532898,
-      "learning_rate": 9.060201005025126e-05,
-      "loss": 5.3895,
-      "step": 9853
-    },
-    {
-      "epoch": 5.138983050847457,
-      "grad_norm": 1.3670129776000977,
-      "learning_rate": 9.060100502512563e-05,
-      "loss": 5.7036,
-      "step": 9854
-    },
-    {
-      "epoch": 5.139504563233377,
-      "grad_norm": 1.483460783958435,
-      "learning_rate": 9.06e-05,
-      "loss": 5.6754,
-      "step": 9855
-    },
-    {
-      "epoch": 5.140026075619296,
-      "grad_norm": 1.4509013891220093,
-      "learning_rate": 9.059899497487437e-05,
-      "loss": 5.7762,
-      "step": 9856
-    },
-    {
-      "epoch": 5.140547588005215,
-      "grad_norm": 1.6064743995666504,
-      "learning_rate": 9.059798994974875e-05,
-      "loss": 5.2685,
-      "step": 9857
-    },
-    {
-      "epoch": 5.141069100391134,
-      "grad_norm": 1.64907705783844,
-      "learning_rate": 9.059698492462311e-05,
-      "loss": 5.6139,
-      "step": 9858
-    },
-    {
-      "epoch": 5.141590612777054,
-      "grad_norm": 1.5877978801727295,
-      "learning_rate": 9.059597989949749e-05,
-      "loss": 5.6623,
-      "step": 9859
-    },
-    {
-      "epoch": 5.1421121251629724,
-      "grad_norm": 1.5105050802230835,
-      "learning_rate": 9.059497487437187e-05,
-      "loss": 5.3956,
-      "step": 9860
-    },
-    {
-      "epoch": 5.142633637548892,
-      "grad_norm": 1.4230408668518066,
-      "learning_rate": 9.059396984924625e-05,
-      "loss": 5.6681,
-      "step": 9861
-    },
-    {
-      "epoch": 5.143155149934811,
-      "grad_norm": 1.427879810333252,
-      "learning_rate": 9.059296482412061e-05,
-      "loss": 5.8953,
-      "step": 9862
-    },
-    {
-      "epoch": 5.14367666232073,
-      "grad_norm": 1.5457682609558105,
-      "learning_rate": 9.059195979899498e-05,
-      "loss": 5.3973,
-      "step": 9863
-    },
-    {
-      "epoch": 5.144198174706649,
-      "grad_norm": 1.5197532176971436,
-      "learning_rate": 9.059095477386935e-05,
-      "loss": 5.2124,
-      "step": 9864
-    },
-    {
-      "epoch": 5.144719687092568,
-      "grad_norm": 1.5479347705841064,
-      "learning_rate": 9.058994974874372e-05,
-      "loss": 5.8841,
-      "step": 9865
-    },
-    {
-      "epoch": 5.1452411994784875,
-      "grad_norm": 1.4521557092666626,
-      "learning_rate": 9.05889447236181e-05,
-      "loss": 5.8088,
-      "step": 9866
-    },
-    {
-      "epoch": 5.145762711864407,
-      "grad_norm": 1.4345005750656128,
-      "learning_rate": 9.058793969849246e-05,
-      "loss": 6.0466,
-      "step": 9867
-    },
-    {
-      "epoch": 5.146284224250326,
-      "grad_norm": 1.5639299154281616,
-      "learning_rate": 9.058693467336684e-05,
-      "loss": 4.9669,
-      "step": 9868
-    },
-    {
-      "epoch": 5.146805736636245,
-      "grad_norm": 1.5659247636795044,
-      "learning_rate": 9.05859296482412e-05,
-      "loss": 5.9409,
-      "step": 9869
-    },
-    {
-      "epoch": 5.147327249022164,
-      "grad_norm": 1.532589316368103,
-      "learning_rate": 9.058492462311558e-05,
-      "loss": 5.6503,
-      "step": 9870
-    },
-    {
-      "epoch": 5.147848761408083,
-      "grad_norm": 1.4843394756317139,
-      "learning_rate": 9.058391959798996e-05,
-      "loss": 5.1897,
-      "step": 9871
-    },
-    {
-      "epoch": 5.1483702737940025,
-      "grad_norm": 1.5393644571304321,
-      "learning_rate": 9.058291457286433e-05,
-      "loss": 5.5436,
-      "step": 9872
-    },
-    {
-      "epoch": 5.148891786179922,
-      "grad_norm": 1.4026321172714233,
-      "learning_rate": 9.05819095477387e-05,
-      "loss": 5.84,
-      "step": 9873
-    },
-    {
-      "epoch": 5.1494132985658405,
-      "grad_norm": 1.415183186531067,
-      "learning_rate": 9.058090452261308e-05,
-      "loss": 6.0467,
-      "step": 9874
-    },
-    {
-      "epoch": 5.14993481095176,
-      "grad_norm": 1.534190058708191,
-      "learning_rate": 9.057989949748744e-05,
-      "loss": 5.3389,
-      "step": 9875
-    },
-    {
-      "epoch": 5.150456323337679,
-      "grad_norm": 1.4060906171798706,
-      "learning_rate": 9.05788944723618e-05,
-      "loss": 5.6426,
-      "step": 9876
-    },
-    {
-      "epoch": 5.150977835723598,
-      "grad_norm": 1.5965155363082886,
-      "learning_rate": 9.057788944723618e-05,
-      "loss": 5.6661,
-      "step": 9877
-    },
-    {
-      "epoch": 5.1514993481095175,
-      "grad_norm": 1.67709481716156,
-      "learning_rate": 9.057688442211055e-05,
-      "loss": 5.5397,
-      "step": 9878
-    },
-    {
-      "epoch": 5.152020860495437,
-      "grad_norm": 1.57057785987854,
-      "learning_rate": 9.057587939698493e-05,
-      "loss": 5.6929,
-      "step": 9879
-    },
-    {
-      "epoch": 5.1525423728813555,
-      "grad_norm": 1.5763025283813477,
-      "learning_rate": 9.05748743718593e-05,
-      "loss": 5.6596,
-      "step": 9880
-    },
-    {
-      "epoch": 5.153063885267275,
-      "grad_norm": 1.4687844514846802,
-      "learning_rate": 9.057386934673368e-05,
-      "loss": 5.9008,
-      "step": 9881
-    },
-    {
-      "epoch": 5.1535853976531945,
-      "grad_norm": 1.6843392848968506,
-      "learning_rate": 9.057286432160805e-05,
-      "loss": 5.3926,
-      "step": 9882
-    },
-    {
-      "epoch": 5.154106910039113,
-      "grad_norm": 1.5013320446014404,
-      "learning_rate": 9.057185929648242e-05,
-      "loss": 5.5651,
-      "step": 9883
-    },
-    {
-      "epoch": 5.1546284224250325,
-      "grad_norm": 1.4606292247772217,
-      "learning_rate": 9.057085427135679e-05,
-      "loss": 5.5041,
-      "step": 9884
-    },
-    {
-      "epoch": 5.155149934810952,
-      "grad_norm": 1.3849786520004272,
-      "learning_rate": 9.056984924623117e-05,
-      "loss": 6.0217,
-      "step": 9885
-    },
-    {
-      "epoch": 5.1556714471968705,
-      "grad_norm": 1.420807123184204,
-      "learning_rate": 9.056884422110553e-05,
-      "loss": 5.3309,
-      "step": 9886
-    },
-    {
-      "epoch": 5.15619295958279,
-      "grad_norm": 1.4809677600860596,
-      "learning_rate": 9.056783919597991e-05,
-      "loss": 5.4858,
-      "step": 9887
-    },
-    {
-      "epoch": 5.1567144719687095,
-      "grad_norm": 1.3820477724075317,
-      "learning_rate": 9.056683417085427e-05,
-      "loss": 5.9631,
-      "step": 9888
-    },
-    {
-      "epoch": 5.157235984354628,
-      "grad_norm": 1.4236769676208496,
-      "learning_rate": 9.056582914572864e-05,
-      "loss": 5.6739,
-      "step": 9889
-    },
-    {
-      "epoch": 5.1577574967405475,
-      "grad_norm": 1.5265676975250244,
-      "learning_rate": 9.056482412060301e-05,
-      "loss": 5.0209,
-      "step": 9890
-    },
-    {
-      "epoch": 5.158279009126467,
-      "grad_norm": 1.4566926956176758,
-      "learning_rate": 9.056381909547739e-05,
-      "loss": 5.1486,
-      "step": 9891
-    },
-    {
-      "epoch": 5.1588005215123856,
-      "grad_norm": 1.5168931484222412,
-      "learning_rate": 9.056281407035177e-05,
-      "loss": 5.4519,
-      "step": 9892
-    },
-    {
-      "epoch": 5.159322033898305,
-      "grad_norm": 1.403723120689392,
-      "learning_rate": 9.056180904522613e-05,
-      "loss": 5.9349,
-      "step": 9893
-    },
-    {
-      "epoch": 5.1598435462842245,
-      "grad_norm": 1.4123072624206543,
-      "learning_rate": 9.056080402010051e-05,
-      "loss": 5.8165,
-      "step": 9894
-    },
-    {
-      "epoch": 5.160365058670143,
-      "grad_norm": 1.542697548866272,
-      "learning_rate": 9.055979899497488e-05,
-      "loss": 5.0845,
-      "step": 9895
-    },
-    {
-      "epoch": 5.1608865710560625,
-      "grad_norm": 1.4831340312957764,
-      "learning_rate": 9.055879396984925e-05,
-      "loss": 5.823,
-      "step": 9896
-    },
-    {
-      "epoch": 5.161408083441982,
-      "grad_norm": 1.4721499681472778,
-      "learning_rate": 9.055778894472362e-05,
-      "loss": 5.5354,
-      "step": 9897
-    },
-    {
-      "epoch": 5.161929595827901,
-      "grad_norm": 1.696143627166748,
-      "learning_rate": 9.0556783919598e-05,
-      "loss": 5.8719,
-      "step": 9898
-    },
-    {
-      "epoch": 5.16245110821382,
-      "grad_norm": 1.5072728395462036,
-      "learning_rate": 9.055577889447236e-05,
-      "loss": 5.6589,
-      "step": 9899
-    },
-    {
-      "epoch": 5.1629726205997395,
-      "grad_norm": 1.4187848567962646,
-      "learning_rate": 9.055477386934674e-05,
-      "loss": 5.8588,
-      "step": 9900
-    },
-    {
-      "epoch": 5.163494132985658,
-      "grad_norm": 1.558765172958374,
-      "learning_rate": 9.055376884422112e-05,
-      "loss": 5.344,
-      "step": 9901
-    },
-    {
-      "epoch": 5.1640156453715775,
-      "grad_norm": 1.5710194110870361,
-      "learning_rate": 9.055276381909548e-05,
-      "loss": 5.7589,
-      "step": 9902
-    },
-    {
-      "epoch": 5.164537157757497,
-      "grad_norm": 1.685858130455017,
-      "learning_rate": 9.055175879396986e-05,
-      "loss": 5.3133,
-      "step": 9903
-    },
-    {
-      "epoch": 5.165058670143416,
-      "grad_norm": 1.626330018043518,
-      "learning_rate": 9.055075376884422e-05,
-      "loss": 5.1475,
-      "step": 9904
-    },
-    {
-      "epoch": 5.165580182529335,
-      "grad_norm": 1.401828646659851,
-      "learning_rate": 9.05497487437186e-05,
-      "loss": 5.6071,
-      "step": 9905
-    },
-    {
-      "epoch": 5.1661016949152545,
-      "grad_norm": 1.4693831205368042,
-      "learning_rate": 9.054874371859296e-05,
-      "loss": 5.7467,
-      "step": 9906
-    },
-    {
-      "epoch": 5.166623207301173,
-      "grad_norm": 1.4842503070831299,
-      "learning_rate": 9.054773869346734e-05,
-      "loss": 5.8732,
-      "step": 9907
-    },
-    {
-      "epoch": 5.1671447196870925,
-      "grad_norm": 1.4684351682662964,
-      "learning_rate": 9.05467336683417e-05,
-      "loss": 5.9743,
-      "step": 9908
-    },
-    {
-      "epoch": 5.167666232073012,
-      "grad_norm": 1.3606348037719727,
-      "learning_rate": 9.054572864321608e-05,
-      "loss": 5.8519,
-      "step": 9909
-    },
-    {
-      "epoch": 5.168187744458931,
-      "grad_norm": 1.523033618927002,
-      "learning_rate": 9.054472361809045e-05,
-      "loss": 5.7302,
-      "step": 9910
-    },
-    {
-      "epoch": 5.16870925684485,
-      "grad_norm": 1.637416958808899,
-      "learning_rate": 9.054371859296483e-05,
-      "loss": 5.4082,
-      "step": 9911
-    },
-    {
-      "epoch": 5.1692307692307695,
-      "grad_norm": 1.3497196435928345,
-      "learning_rate": 9.05427135678392e-05,
-      "loss": 5.9323,
-      "step": 9912
-    },
-    {
-      "epoch": 5.169752281616688,
-      "grad_norm": 1.6495516300201416,
-      "learning_rate": 9.054170854271358e-05,
-      "loss": 5.5983,
-      "step": 9913
-    },
-    {
-      "epoch": 5.170273794002608,
-      "grad_norm": 1.4929183721542358,
-      "learning_rate": 9.054070351758795e-05,
-      "loss": 5.4216,
-      "step": 9914
-    },
-    {
-      "epoch": 5.170795306388527,
-      "grad_norm": 1.592025637626648,
-      "learning_rate": 9.053969849246231e-05,
-      "loss": 5.7795,
-      "step": 9915
-    },
-    {
-      "epoch": 5.171316818774446,
-      "grad_norm": 1.4331640005111694,
-      "learning_rate": 9.053869346733669e-05,
-      "loss": 5.8521,
-      "step": 9916
-    },
-    {
-      "epoch": 5.171838331160365,
-      "grad_norm": 1.3735859394073486,
-      "learning_rate": 9.053768844221105e-05,
-      "loss": 6.1483,
-      "step": 9917
-    },
-    {
-      "epoch": 5.1723598435462845,
-      "grad_norm": 1.6215449571609497,
-      "learning_rate": 9.053668341708543e-05,
-      "loss": 5.2722,
-      "step": 9918
-    },
-    {
-      "epoch": 5.172881355932203,
-      "grad_norm": 1.5381478071212769,
-      "learning_rate": 9.05356783919598e-05,
-      "loss": 5.4203,
-      "step": 9919
-    },
-    {
-      "epoch": 5.173402868318123,
-      "grad_norm": 1.3846725225448608,
-      "learning_rate": 9.053467336683417e-05,
-      "loss": 5.8898,
-      "step": 9920
-    },
-    {
-      "epoch": 5.173924380704042,
-      "grad_norm": 1.4400655031204224,
-      "learning_rate": 9.053366834170855e-05,
-      "loss": 5.7759,
-      "step": 9921
-    },
-    {
-      "epoch": 5.174445893089961,
-      "grad_norm": 1.5162898302078247,
-      "learning_rate": 9.053266331658293e-05,
-      "loss": 5.8293,
-      "step": 9922
-    },
-    {
-      "epoch": 5.17496740547588,
-      "grad_norm": 1.3530471324920654,
-      "learning_rate": 9.053165829145729e-05,
-      "loss": 6.0285,
-      "step": 9923
-    },
-    {
-      "epoch": 5.1754889178617995,
-      "grad_norm": 1.497591257095337,
-      "learning_rate": 9.053065326633167e-05,
-      "loss": 5.6772,
-      "step": 9924
-    },
-    {
-      "epoch": 5.176010430247718,
-      "grad_norm": 1.293992042541504,
-      "learning_rate": 9.052964824120603e-05,
-      "loss": 5.6945,
-      "step": 9925
-    },
-    {
-      "epoch": 5.176531942633638,
-      "grad_norm": 1.3455917835235596,
-      "learning_rate": 9.052864321608041e-05,
-      "loss": 5.9888,
-      "step": 9926
-    },
-    {
-      "epoch": 5.177053455019557,
-      "grad_norm": 1.4572997093200684,
-      "learning_rate": 9.052763819095478e-05,
-      "loss": 5.8011,
-      "step": 9927
-    },
-    {
-      "epoch": 5.177574967405476,
-      "grad_norm": 1.5736212730407715,
-      "learning_rate": 9.052663316582915e-05,
-      "loss": 5.276,
-      "step": 9928
-    },
-    {
-      "epoch": 5.178096479791395,
-      "grad_norm": 1.4594172239303589,
-      "learning_rate": 9.052562814070352e-05,
-      "loss": 5.5647,
-      "step": 9929
-    },
-    {
-      "epoch": 5.1786179921773146,
-      "grad_norm": 1.3668320178985596,
-      "learning_rate": 9.052462311557788e-05,
-      "loss": 5.0599,
-      "step": 9930
-    },
-    {
-      "epoch": 5.179139504563233,
-      "grad_norm": 1.4575880765914917,
-      "learning_rate": 9.052361809045226e-05,
-      "loss": 5.5572,
-      "step": 9931
-    },
-    {
-      "epoch": 5.179661016949153,
-      "grad_norm": 1.5186054706573486,
-      "learning_rate": 9.052261306532664e-05,
-      "loss": 4.8642,
-      "step": 9932
-    },
-    {
-      "epoch": 5.180182529335072,
-      "grad_norm": 1.4596121311187744,
-      "learning_rate": 9.052160804020102e-05,
-      "loss": 5.7415,
-      "step": 9933
-    },
-    {
-      "epoch": 5.180704041720991,
-      "grad_norm": 1.5136321783065796,
-      "learning_rate": 9.052060301507538e-05,
-      "loss": 5.6622,
-      "step": 9934
-    },
-    {
-      "epoch": 5.18122555410691,
-      "grad_norm": 1.470464825630188,
-      "learning_rate": 9.051959798994976e-05,
-      "loss": 6.0204,
-      "step": 9935
-    },
-    {
-      "epoch": 5.18174706649283,
-      "grad_norm": 1.466240406036377,
-      "learning_rate": 9.051859296482412e-05,
-      "loss": 5.9422,
-      "step": 9936
-    },
-    {
-      "epoch": 5.182268578878748,
-      "grad_norm": 1.3966820240020752,
-      "learning_rate": 9.05175879396985e-05,
-      "loss": 5.5859,
-      "step": 9937
-    },
-    {
-      "epoch": 5.182790091264668,
-      "grad_norm": 1.437476396560669,
-      "learning_rate": 9.051658291457286e-05,
-      "loss": 5.8368,
-      "step": 9938
-    },
-    {
-      "epoch": 5.183311603650587,
-      "grad_norm": 1.4285919666290283,
-      "learning_rate": 9.051557788944724e-05,
-      "loss": 5.9559,
-      "step": 9939
-    },
-    {
-      "epoch": 5.183833116036506,
-      "grad_norm": 1.519158124923706,
-      "learning_rate": 9.051457286432161e-05,
-      "loss": 5.531,
-      "step": 9940
-    },
-    {
-      "epoch": 5.184354628422425,
-      "grad_norm": 1.4807437658309937,
-      "learning_rate": 9.051356783919598e-05,
-      "loss": 5.4824,
-      "step": 9941
-    },
-    {
-      "epoch": 5.184876140808345,
-      "grad_norm": 1.4306085109710693,
-      "learning_rate": 9.051256281407036e-05,
-      "loss": 5.6454,
-      "step": 9942
-    },
-    {
-      "epoch": 5.185397653194263,
-      "grad_norm": 1.3726962804794312,
-      "learning_rate": 9.051155778894473e-05,
-      "loss": 5.8071,
-      "step": 9943
-    },
-    {
-      "epoch": 5.185919165580183,
-      "grad_norm": 1.9009023904800415,
-      "learning_rate": 9.05105527638191e-05,
-      "loss": 5.1974,
-      "step": 9944
-    },
-    {
-      "epoch": 5.186440677966102,
-      "grad_norm": 1.511934518814087,
-      "learning_rate": 9.050954773869347e-05,
-      "loss": 5.6521,
-      "step": 9945
-    },
-    {
-      "epoch": 5.186962190352021,
-      "grad_norm": 1.5876998901367188,
-      "learning_rate": 9.050854271356785e-05,
-      "loss": 5.5606,
-      "step": 9946
-    },
-    {
-      "epoch": 5.18748370273794,
-      "grad_norm": 1.417900800704956,
-      "learning_rate": 9.050753768844221e-05,
-      "loss": 5.8679,
-      "step": 9947
-    },
-    {
-      "epoch": 5.18800521512386,
-      "grad_norm": 1.4556668996810913,
-      "learning_rate": 9.050653266331659e-05,
-      "loss": 5.6389,
-      "step": 9948
-    },
-    {
-      "epoch": 5.188526727509778,
-      "grad_norm": 1.4087109565734863,
-      "learning_rate": 9.050552763819095e-05,
-      "loss": 5.5398,
-      "step": 9949
-    },
-    {
-      "epoch": 5.189048239895698,
-      "grad_norm": 1.4323667287826538,
-      "learning_rate": 9.050452261306533e-05,
-      "loss": 5.6941,
-      "step": 9950
-    },
-    {
-      "epoch": 5.189569752281617,
-      "grad_norm": 1.505454421043396,
-      "learning_rate": 9.05035175879397e-05,
-      "loss": 5.9104,
-      "step": 9951
-    },
-    {
-      "epoch": 5.190091264667536,
-      "grad_norm": 1.33050537109375,
-      "learning_rate": 9.050251256281407e-05,
-      "loss": 6.2153,
-      "step": 9952
-    },
-    {
-      "epoch": 5.190612777053455,
-      "grad_norm": 1.4077614545822144,
-      "learning_rate": 9.050150753768845e-05,
-      "loss": 5.3064,
-      "step": 9953
-    },
-    {
-      "epoch": 5.191134289439375,
-      "grad_norm": 1.4086315631866455,
-      "learning_rate": 9.050050251256283e-05,
-      "loss": 5.5545,
-      "step": 9954
-    },
-    {
-      "epoch": 5.191655801825293,
-      "grad_norm": 1.5410484075546265,
-      "learning_rate": 9.049949748743719e-05,
-      "loss": 5.6678,
-      "step": 9955
-    },
-    {
-      "epoch": 5.192177314211213,
-      "grad_norm": 1.4053854942321777,
-      "learning_rate": 9.049849246231156e-05,
-      "loss": 5.9258,
-      "step": 9956
-    },
-    {
-      "epoch": 5.192698826597132,
-      "grad_norm": 1.503434658050537,
-      "learning_rate": 9.049748743718594e-05,
-      "loss": 5.7043,
-      "step": 9957
-    },
-    {
-      "epoch": 5.193220338983051,
-      "grad_norm": 1.5614920854568481,
-      "learning_rate": 9.04964824120603e-05,
-      "loss": 5.3317,
-      "step": 9958
-    },
-    {
-      "epoch": 5.19374185136897,
-      "grad_norm": 1.7876731157302856,
-      "learning_rate": 9.049547738693468e-05,
-      "loss": 5.058,
-      "step": 9959
-    },
-    {
-      "epoch": 5.194263363754889,
-      "grad_norm": 1.5546025037765503,
-      "learning_rate": 9.049447236180904e-05,
-      "loss": 5.4056,
-      "step": 9960
-    },
-    {
-      "epoch": 5.194784876140808,
-      "grad_norm": 1.4160833358764648,
-      "learning_rate": 9.049346733668342e-05,
-      "loss": 5.8904,
-      "step": 9961
-    },
-    {
-      "epoch": 5.195306388526728,
-      "grad_norm": 1.4793853759765625,
-      "learning_rate": 9.04924623115578e-05,
-      "loss": 5.1416,
-      "step": 9962
-    },
-    {
-      "epoch": 5.195827900912647,
-      "grad_norm": 1.484444499015808,
-      "learning_rate": 9.049145728643218e-05,
-      "loss": 5.9224,
-      "step": 9963
-    },
-    {
-      "epoch": 5.196349413298566,
-      "grad_norm": 1.4560766220092773,
-      "learning_rate": 9.049045226130654e-05,
-      "loss": 5.6225,
-      "step": 9964
-    },
-    {
-      "epoch": 5.196870925684485,
-      "grad_norm": 1.7138055562973022,
-      "learning_rate": 9.048944723618092e-05,
-      "loss": 5.0761,
-      "step": 9965
-    },
-    {
-      "epoch": 5.197392438070404,
-      "grad_norm": 1.4021923542022705,
-      "learning_rate": 9.048844221105528e-05,
-      "loss": 6.0894,
-      "step": 9966
-    },
-    {
-      "epoch": 5.197913950456323,
-      "grad_norm": 1.4645551443099976,
-      "learning_rate": 9.048743718592966e-05,
-      "loss": 5.8846,
-      "step": 9967
-    },
-    {
-      "epoch": 5.198435462842243,
-      "grad_norm": 1.411424160003662,
-      "learning_rate": 9.048643216080402e-05,
-      "loss": 5.6751,
-      "step": 9968
-    },
-    {
-      "epoch": 5.198956975228161,
-      "grad_norm": 1.5094726085662842,
-      "learning_rate": 9.048542713567839e-05,
-      "loss": 5.7017,
-      "step": 9969
-    },
-    {
-      "epoch": 5.199478487614081,
-      "grad_norm": 1.5539816617965698,
-      "learning_rate": 9.048442211055277e-05,
-      "loss": 5.035,
-      "step": 9970
-    },
-    {
-      "epoch": 5.2,
-      "grad_norm": 1.4314384460449219,
-      "learning_rate": 9.048341708542713e-05,
-      "loss": 5.4059,
-      "step": 9971
-    },
-    {
-      "epoch": 5.200521512385919,
-      "grad_norm": 2.4178171157836914,
-      "learning_rate": 9.048241206030151e-05,
-      "loss": 5.4255,
-      "step": 9972
-    },
-    {
-      "epoch": 5.201043024771838,
-      "grad_norm": 1.4391679763793945,
-      "learning_rate": 9.048140703517589e-05,
-      "loss": 5.6224,
-      "step": 9973
-    },
-    {
-      "epoch": 5.201564537157758,
-      "grad_norm": 1.3902226686477661,
-      "learning_rate": 9.048040201005026e-05,
-      "loss": 5.8473,
-      "step": 9974
-    },
-    {
-      "epoch": 5.202086049543676,
-      "grad_norm": 1.543908715248108,
-      "learning_rate": 9.047939698492463e-05,
-      "loss": 5.5054,
-      "step": 9975
-    },
-    {
-      "epoch": 5.202607561929596,
-      "grad_norm": 1.5774601697921753,
-      "learning_rate": 9.0478391959799e-05,
-      "loss": 5.632,
-      "step": 9976
-    },
-    {
-      "epoch": 5.203129074315515,
-      "grad_norm": 1.4888261556625366,
-      "learning_rate": 9.047738693467337e-05,
-      "loss": 5.4613,
-      "step": 9977
-    },
-    {
-      "epoch": 5.203650586701434,
-      "grad_norm": 1.4399272203445435,
-      "learning_rate": 9.047638190954775e-05,
-      "loss": 5.4807,
-      "step": 9978
-    },
-    {
-      "epoch": 5.204172099087353,
-      "grad_norm": 1.4446275234222412,
-      "learning_rate": 9.047537688442211e-05,
-      "loss": 5.5286,
-      "step": 9979
-    },
-    {
-      "epoch": 5.204693611473273,
-      "grad_norm": 1.481410264968872,
-      "learning_rate": 9.047437185929649e-05,
-      "loss": 5.6388,
-      "step": 9980
-    },
-    {
-      "epoch": 5.205215123859191,
-      "grad_norm": 1.4537277221679688,
-      "learning_rate": 9.047336683417085e-05,
-      "loss": 5.7902,
-      "step": 9981
-    },
-    {
-      "epoch": 5.205736636245111,
-      "grad_norm": 1.4115307331085205,
-      "learning_rate": 9.047236180904523e-05,
-      "loss": 5.7082,
-      "step": 9982
-    },
-    {
-      "epoch": 5.20625814863103,
-      "grad_norm": 1.5536311864852905,
-      "learning_rate": 9.047135678391961e-05,
-      "loss": 4.8692,
-      "step": 9983
-    },
-    {
-      "epoch": 5.206779661016949,
-      "grad_norm": 1.576254963874817,
-      "learning_rate": 9.047035175879397e-05,
-      "loss": 5.6896,
-      "step": 9984
-    },
-    {
-      "epoch": 5.207301173402868,
-      "grad_norm": 1.5688122510910034,
-      "learning_rate": 9.046934673366835e-05,
-      "loss": 5.6857,
-      "step": 9985
-    },
-    {
-      "epoch": 5.207822685788788,
-      "grad_norm": 1.4543150663375854,
-      "learning_rate": 9.046834170854272e-05,
-      "loss": 5.583,
-      "step": 9986
-    },
-    {
-      "epoch": 5.208344198174706,
-      "grad_norm": 1.409266710281372,
-      "learning_rate": 9.04673366834171e-05,
-      "loss": 5.7489,
-      "step": 9987
-    },
-    {
-      "epoch": 5.208865710560626,
-      "grad_norm": 1.4648630619049072,
-      "learning_rate": 9.046633165829146e-05,
-      "loss": 5.6368,
-      "step": 9988
-    },
-    {
-      "epoch": 5.209387222946545,
-      "grad_norm": 1.4539110660552979,
-      "learning_rate": 9.046532663316584e-05,
-      "loss": 5.7356,
-      "step": 9989
-    },
-    {
-      "epoch": 5.209908735332464,
-      "grad_norm": 1.6872522830963135,
-      "learning_rate": 9.04643216080402e-05,
-      "loss": 5.0475,
-      "step": 9990
-    },
-    {
-      "epoch": 5.210430247718383,
-      "grad_norm": 1.5245758295059204,
-      "learning_rate": 9.046331658291458e-05,
-      "loss": 5.6195,
-      "step": 9991
-    },
-    {
-      "epoch": 5.210951760104303,
-      "grad_norm": 1.443320393562317,
-      "learning_rate": 9.046231155778894e-05,
-      "loss": 5.3965,
-      "step": 9992
-    },
-    {
-      "epoch": 5.211473272490221,
-      "grad_norm": 1.4111230373382568,
-      "learning_rate": 9.046130653266332e-05,
-      "loss": 5.5074,
-      "step": 9993
-    },
-    {
-      "epoch": 5.211994784876141,
-      "grad_norm": 1.4921717643737793,
-      "learning_rate": 9.04603015075377e-05,
-      "loss": 5.2305,
-      "step": 9994
-    },
-    {
-      "epoch": 5.21251629726206,
-      "grad_norm": 1.4672561883926392,
-      "learning_rate": 9.045929648241206e-05,
-      "loss": 5.6841,
-      "step": 9995
-    },
-    {
-      "epoch": 5.213037809647979,
-      "grad_norm": 1.5562853813171387,
-      "learning_rate": 9.045829145728644e-05,
-      "loss": 5.1881,
-      "step": 9996
-    },
-    {
-      "epoch": 5.213559322033898,
-      "grad_norm": 1.4447107315063477,
-      "learning_rate": 9.04572864321608e-05,
-      "loss": 5.8668,
-      "step": 9997
-    },
-    {
-      "epoch": 5.214080834419818,
-      "grad_norm": 1.523319125175476,
-      "learning_rate": 9.045628140703518e-05,
-      "loss": 5.4494,
-      "step": 9998
-    },
-    {
-      "epoch": 5.214602346805736,
-      "grad_norm": 1.4796721935272217,
-      "learning_rate": 9.045527638190955e-05,
-      "loss": 5.9487,
-      "step": 9999
-    },
-    {
-      "epoch": 5.215123859191656,
-      "grad_norm": 1.4420745372772217,
-      "learning_rate": 9.045427135678392e-05,
-      "loss": 5.3726,
-      "step": 10000
-    },
-    {
-      "epoch": 5.215123859191656,
-      "eval_loss": 5.693417549133301,
-      "eval_runtime": 42.6818,
-      "eval_samples_per_second": 28.724,
-      "eval_steps_per_second": 3.608,
-      "step": 10000
-    },
-    {
-      "epoch": 5.215645371577575,
-      "grad_norm": 1.4717859029769897,
-      "learning_rate": 9.045326633165829e-05,
-      "loss": 5.9161,
-      "step": 10001
-    },
-    {
-      "epoch": 5.216166883963494,
-      "grad_norm": 1.5446093082427979,
-      "learning_rate": 9.045226130653267e-05,
-      "loss": 5.8622,
-      "step": 10002
-    },
-    {
-      "epoch": 5.216688396349413,
-      "grad_norm": 1.3811769485473633,
-      "learning_rate": 9.045125628140703e-05,
-      "loss": 5.8583,
-      "step": 10003
-    },
-    {
-      "epoch": 5.217209908735333,
-      "grad_norm": 1.4074715375900269,
-      "learning_rate": 9.045025125628141e-05,
-      "loss": 5.6315,
-      "step": 10004
-    },
-    {
-      "epoch": 5.217731421121251,
-      "grad_norm": 1.37552011013031,
-      "learning_rate": 9.044924623115579e-05,
-      "loss": 5.5401,
-      "step": 10005
-    },
-    {
-      "epoch": 5.218252933507171,
-      "grad_norm": 1.5397658348083496,
-      "learning_rate": 9.044824120603016e-05,
-      "loss": 5.4292,
-      "step": 10006
-    },
-    {
-      "epoch": 5.21877444589309,
-      "grad_norm": 1.4655550718307495,
-      "learning_rate": 9.044723618090453e-05,
-      "loss": 5.3384,
-      "step": 10007
-    },
-    {
-      "epoch": 5.219295958279009,
-      "grad_norm": 1.4058303833007812,
-      "learning_rate": 9.044623115577889e-05,
-      "loss": 5.9096,
-      "step": 10008
-    },
-    {
-      "epoch": 5.219817470664928,
-      "grad_norm": 1.471301555633545,
-      "learning_rate": 9.044522613065327e-05,
-      "loss": 5.9044,
-      "step": 10009
-    },
-    {
-      "epoch": 5.220338983050848,
-      "grad_norm": 1.3151054382324219,
-      "learning_rate": 9.044422110552763e-05,
-      "loss": 5.7608,
-      "step": 10010
-    },
-    {
-      "epoch": 5.220860495436766,
-      "grad_norm": 1.332016944885254,
-      "learning_rate": 9.044321608040201e-05,
-      "loss": 5.9322,
-      "step": 10011
-    },
-    {
-      "epoch": 5.221382007822686,
-      "grad_norm": 1.55919349193573,
-      "learning_rate": 9.044221105527638e-05,
-      "loss": 5.4266,
-      "step": 10012
-    },
-    {
-      "epoch": 5.221903520208605,
-      "grad_norm": 1.512215256690979,
-      "learning_rate": 9.044120603015075e-05,
-      "loss": 5.553,
-      "step": 10013
-    },
-    {
-      "epoch": 5.222425032594524,
-      "grad_norm": 1.4980473518371582,
-      "learning_rate": 9.044020100502513e-05,
-      "loss": 5.3979,
-      "step": 10014
-    },
-    {
-      "epoch": 5.222946544980443,
-      "grad_norm": 1.3802341222763062,
-      "learning_rate": 9.043919597989951e-05,
-      "loss": 6.1245,
-      "step": 10015
-    },
-    {
-      "epoch": 5.223468057366363,
-      "grad_norm": 1.3954429626464844,
-      "learning_rate": 9.043819095477387e-05,
-      "loss": 5.479,
-      "step": 10016
-    },
-    {
-      "epoch": 5.223989569752281,
-      "grad_norm": 1.4645367860794067,
-      "learning_rate": 9.043718592964825e-05,
-      "loss": 5.3299,
-      "step": 10017
-    },
-    {
-      "epoch": 5.224511082138201,
-      "grad_norm": 1.4221478700637817,
-      "learning_rate": 9.043618090452262e-05,
-      "loss": 5.9462,
-      "step": 10018
-    },
-    {
-      "epoch": 5.22503259452412,
-      "grad_norm": 1.364375352859497,
-      "learning_rate": 9.0435175879397e-05,
-      "loss": 6.1325,
-      "step": 10019
-    },
-    {
-      "epoch": 5.225554106910039,
-      "grad_norm": 1.4787803888320923,
-      "learning_rate": 9.043417085427136e-05,
-      "loss": 5.7425,
-      "step": 10020
-    },
-    {
-      "epoch": 5.226075619295958,
-      "grad_norm": 1.407372236251831,
-      "learning_rate": 9.043316582914574e-05,
-      "loss": 5.6933,
-      "step": 10021
-    },
-    {
-      "epoch": 5.226597131681878,
-      "grad_norm": 1.4925684928894043,
-      "learning_rate": 9.04321608040201e-05,
-      "loss": 5.7532,
-      "step": 10022
-    },
-    {
-      "epoch": 5.227118644067796,
-      "grad_norm": 1.4356073141098022,
-      "learning_rate": 9.043115577889447e-05,
-      "loss": 5.5757,
-      "step": 10023
-    },
-    {
-      "epoch": 5.227640156453716,
-      "grad_norm": 1.456728458404541,
-      "learning_rate": 9.043015075376884e-05,
-      "loss": 5.7391,
-      "step": 10024
-    },
-    {
-      "epoch": 5.228161668839635,
-      "grad_norm": 1.4190521240234375,
-      "learning_rate": 9.042914572864322e-05,
-      "loss": 5.8676,
-      "step": 10025
-    },
-    {
-      "epoch": 5.228683181225554,
-      "grad_norm": 1.445654034614563,
-      "learning_rate": 9.04281407035176e-05,
-      "loss": 5.5109,
-      "step": 10026
-    },
-    {
-      "epoch": 5.229204693611473,
-      "grad_norm": 1.4841558933258057,
-      "learning_rate": 9.042713567839196e-05,
-      "loss": 5.6778,
-      "step": 10027
-    },
-    {
-      "epoch": 5.229726205997393,
-      "grad_norm": 1.6084147691726685,
-      "learning_rate": 9.042613065326634e-05,
-      "loss": 5.6225,
-      "step": 10028
-    },
-    {
-      "epoch": 5.230247718383311,
-      "grad_norm": 1.5230908393859863,
-      "learning_rate": 9.04251256281407e-05,
-      "loss": 5.7634,
-      "step": 10029
-    },
-    {
-      "epoch": 5.230769230769231,
-      "grad_norm": 1.4932098388671875,
-      "learning_rate": 9.042412060301508e-05,
-      "loss": 5.7,
-      "step": 10030
-    },
-    {
-      "epoch": 5.23129074315515,
-      "grad_norm": 1.4683929681777954,
-      "learning_rate": 9.042311557788945e-05,
-      "loss": 5.8991,
-      "step": 10031
-    },
-    {
-      "epoch": 5.231812255541069,
-      "grad_norm": 1.5104502439498901,
-      "learning_rate": 9.042211055276383e-05,
-      "loss": 5.7744,
-      "step": 10032
-    },
-    {
-      "epoch": 5.232333767926988,
-      "grad_norm": 1.6425775289535522,
-      "learning_rate": 9.042110552763819e-05,
-      "loss": 5.6253,
-      "step": 10033
-    },
-    {
-      "epoch": 5.232855280312908,
-      "grad_norm": 1.5350946187973022,
-      "learning_rate": 9.042010050251257e-05,
-      "loss": 5.2043,
-      "step": 10034
-    },
-    {
-      "epoch": 5.233376792698826,
-      "grad_norm": 1.3478316068649292,
-      "learning_rate": 9.041909547738695e-05,
-      "loss": 5.9213,
-      "step": 10035
-    },
-    {
-      "epoch": 5.233898305084746,
-      "grad_norm": 1.4705829620361328,
-      "learning_rate": 9.041809045226131e-05,
-      "loss": 5.7951,
-      "step": 10036
-    },
-    {
-      "epoch": 5.234419817470665,
-      "grad_norm": 1.5671786069869995,
-      "learning_rate": 9.041708542713569e-05,
-      "loss": 4.932,
-      "step": 10037
-    },
-    {
-      "epoch": 5.234941329856584,
-      "grad_norm": 1.5181962251663208,
-      "learning_rate": 9.041608040201005e-05,
-      "loss": 6.1354,
-      "step": 10038
-    },
-    {
-      "epoch": 5.235462842242503,
-      "grad_norm": 1.3497366905212402,
-      "learning_rate": 9.041507537688443e-05,
-      "loss": 5.4015,
-      "step": 10039
-    },
-    {
-      "epoch": 5.235984354628423,
-      "grad_norm": 1.4979101419448853,
-      "learning_rate": 9.04140703517588e-05,
-      "loss": 5.6389,
-      "step": 10040
-    },
-    {
-      "epoch": 5.236505867014341,
-      "grad_norm": 1.6562597751617432,
-      "learning_rate": 9.041306532663317e-05,
-      "loss": 5.3035,
-      "step": 10041
-    },
-    {
-      "epoch": 5.237027379400261,
-      "grad_norm": 1.4748013019561768,
-      "learning_rate": 9.041206030150754e-05,
-      "loss": 6.0393,
-      "step": 10042
-    },
-    {
-      "epoch": 5.23754889178618,
-      "grad_norm": 1.5297287702560425,
-      "learning_rate": 9.041105527638191e-05,
-      "loss": 5.8313,
-      "step": 10043
-    },
-    {
-      "epoch": 5.238070404172099,
-      "grad_norm": 1.5088549852371216,
-      "learning_rate": 9.041005025125628e-05,
-      "loss": 5.8193,
-      "step": 10044
-    },
-    {
-      "epoch": 5.238591916558018,
-      "grad_norm": 1.5079742670059204,
-      "learning_rate": 9.040904522613066e-05,
-      "loss": 5.5471,
-      "step": 10045
-    },
-    {
-      "epoch": 5.239113428943938,
-      "grad_norm": 1.3613719940185547,
-      "learning_rate": 9.040804020100503e-05,
-      "loss": 5.8865,
-      "step": 10046
-    },
-    {
-      "epoch": 5.239634941329856,
-      "grad_norm": 1.499681830406189,
-      "learning_rate": 9.040703517587941e-05,
-      "loss": 5.0419,
-      "step": 10047
-    },
-    {
-      "epoch": 5.240156453715776,
-      "grad_norm": 1.4805010557174683,
-      "learning_rate": 9.040603015075378e-05,
-      "loss": 5.449,
-      "step": 10048
-    },
-    {
-      "epoch": 5.240677966101695,
-      "grad_norm": 1.4263910055160522,
-      "learning_rate": 9.040502512562814e-05,
-      "loss": 5.6308,
-      "step": 10049
-    },
-    {
-      "epoch": 5.241199478487614,
-      "grad_norm": 1.4595465660095215,
-      "learning_rate": 9.040402010050252e-05,
-      "loss": 5.6999,
-      "step": 10050
-    },
-    {
-      "epoch": 5.241720990873533,
-      "grad_norm": 1.6155757904052734,
-      "learning_rate": 9.040301507537688e-05,
-      "loss": 4.6291,
-      "step": 10051
-    },
-    {
-      "epoch": 5.242242503259453,
-      "grad_norm": 1.448711633682251,
-      "learning_rate": 9.040201005025126e-05,
-      "loss": 5.5144,
-      "step": 10052
-    },
-    {
-      "epoch": 5.242764015645371,
-      "grad_norm": 1.4708187580108643,
-      "learning_rate": 9.040100502512562e-05,
-      "loss": 5.4994,
-      "step": 10053
-    },
-    {
-      "epoch": 5.243285528031291,
-      "grad_norm": 1.3723044395446777,
-      "learning_rate": 9.04e-05,
-      "loss": 5.5186,
-      "step": 10054
-    },
-    {
-      "epoch": 5.2438070404172095,
-      "grad_norm": 1.4838917255401611,
-      "learning_rate": 9.039899497487438e-05,
-      "loss": 5.3591,
-      "step": 10055
-    },
-    {
-      "epoch": 5.244328552803129,
-      "grad_norm": 1.454161524772644,
-      "learning_rate": 9.039798994974876e-05,
-      "loss": 5.7842,
-      "step": 10056
-    },
-    {
-      "epoch": 5.244850065189048,
-      "grad_norm": 1.4295337200164795,
-      "learning_rate": 9.039698492462312e-05,
-      "loss": 5.7393,
-      "step": 10057
-    },
-    {
-      "epoch": 5.245371577574968,
-      "grad_norm": 1.4544376134872437,
-      "learning_rate": 9.03959798994975e-05,
-      "loss": 5.292,
-      "step": 10058
-    },
-    {
-      "epoch": 5.245893089960886,
-      "grad_norm": 1.4494431018829346,
-      "learning_rate": 9.039497487437186e-05,
-      "loss": 5.5997,
-      "step": 10059
-    },
-    {
-      "epoch": 5.246414602346806,
-      "grad_norm": 1.491150140762329,
-      "learning_rate": 9.039396984924624e-05,
-      "loss": 5.858,
-      "step": 10060
-    },
-    {
-      "epoch": 5.2469361147327245,
-      "grad_norm": 1.5064276456832886,
-      "learning_rate": 9.03929648241206e-05,
-      "loss": 5.6776,
-      "step": 10061
-    },
-    {
-      "epoch": 5.247457627118644,
-      "grad_norm": 1.3450371026992798,
-      "learning_rate": 9.039195979899497e-05,
-      "loss": 5.7663,
-      "step": 10062
-    },
-    {
-      "epoch": 5.247979139504563,
-      "grad_norm": 1.3917235136032104,
-      "learning_rate": 9.039095477386935e-05,
-      "loss": 5.4919,
-      "step": 10063
-    },
-    {
-      "epoch": 5.248500651890482,
-      "grad_norm": 1.5457264184951782,
-      "learning_rate": 9.038994974874371e-05,
-      "loss": 5.8379,
-      "step": 10064
-    },
-    {
-      "epoch": 5.2490221642764014,
-      "grad_norm": 1.4693011045455933,
-      "learning_rate": 9.038894472361809e-05,
-      "loss": 5.6482,
-      "step": 10065
-    },
-    {
-      "epoch": 5.249543676662321,
-      "grad_norm": 1.490593671798706,
-      "learning_rate": 9.038793969849247e-05,
-      "loss": 5.3512,
-      "step": 10066
-    },
-    {
-      "epoch": 5.2500651890482395,
-      "grad_norm": 1.5154200792312622,
-      "learning_rate": 9.038693467336685e-05,
-      "loss": 5.6765,
-      "step": 10067
-    },
-    {
-      "epoch": 5.250586701434159,
-      "grad_norm": 1.402030348777771,
-      "learning_rate": 9.038592964824121e-05,
-      "loss": 5.8638,
-      "step": 10068
-    },
-    {
-      "epoch": 5.251108213820078,
-      "grad_norm": 1.4765713214874268,
-      "learning_rate": 9.038492462311559e-05,
-      "loss": 5.713,
-      "step": 10069
-    },
-    {
-      "epoch": 5.251629726205997,
-      "grad_norm": 1.348554015159607,
-      "learning_rate": 9.038391959798995e-05,
-      "loss": 6.0765,
-      "step": 10070
-    },
-    {
-      "epoch": 5.2521512385919165,
-      "grad_norm": 1.3900163173675537,
-      "learning_rate": 9.038291457286433e-05,
-      "loss": 5.7635,
-      "step": 10071
-    },
-    {
-      "epoch": 5.252672750977836,
-      "grad_norm": 1.3848965167999268,
-      "learning_rate": 9.03819095477387e-05,
-      "loss": 5.8575,
-      "step": 10072
-    },
-    {
-      "epoch": 5.2531942633637545,
-      "grad_norm": 1.469152569770813,
-      "learning_rate": 9.038090452261307e-05,
-      "loss": 5.4415,
-      "step": 10073
-    },
-    {
-      "epoch": 5.253715775749674,
-      "grad_norm": 1.501121997833252,
-      "learning_rate": 9.037989949748744e-05,
-      "loss": 5.668,
-      "step": 10074
-    },
-    {
-      "epoch": 5.254237288135593,
-      "grad_norm": 1.628551721572876,
-      "learning_rate": 9.037889447236181e-05,
-      "loss": 5.3139,
-      "step": 10075
-    },
-    {
-      "epoch": 5.254758800521512,
-      "grad_norm": 1.576874017715454,
-      "learning_rate": 9.037788944723619e-05,
-      "loss": 5.5314,
-      "step": 10076
-    },
-    {
-      "epoch": 5.2552803129074315,
-      "grad_norm": 1.5907871723175049,
-      "learning_rate": 9.037688442211056e-05,
-      "loss": 5.0692,
-      "step": 10077
-    },
-    {
-      "epoch": 5.255801825293351,
-      "grad_norm": 1.501671552658081,
-      "learning_rate": 9.037587939698493e-05,
-      "loss": 5.8724,
-      "step": 10078
-    },
-    {
-      "epoch": 5.2563233376792695,
-      "grad_norm": 1.4172048568725586,
-      "learning_rate": 9.03748743718593e-05,
-      "loss": 6.048,
-      "step": 10079
-    },
-    {
-      "epoch": 5.256844850065189,
-      "grad_norm": 1.3762807846069336,
-      "learning_rate": 9.037386934673368e-05,
-      "loss": 5.4745,
-      "step": 10080
-    },
-    {
-      "epoch": 5.257366362451108,
-      "grad_norm": 1.3691222667694092,
-      "learning_rate": 9.037286432160804e-05,
-      "loss": 5.6501,
-      "step": 10081
-    },
-    {
-      "epoch": 5.257887874837027,
-      "grad_norm": 1.4266730546951294,
-      "learning_rate": 9.037185929648242e-05,
-      "loss": 5.8197,
-      "step": 10082
-    },
-    {
-      "epoch": 5.2584093872229465,
-      "grad_norm": 1.4243927001953125,
-      "learning_rate": 9.037085427135678e-05,
-      "loss": 5.7378,
-      "step": 10083
-    },
-    {
-      "epoch": 5.258930899608866,
-      "grad_norm": 1.440061330795288,
-      "learning_rate": 9.036984924623116e-05,
-      "loss": 5.7501,
-      "step": 10084
-    },
-    {
-      "epoch": 5.2594524119947845,
-      "grad_norm": 1.4981987476348877,
-      "learning_rate": 9.036884422110552e-05,
-      "loss": 5.5786,
-      "step": 10085
-    },
-    {
-      "epoch": 5.259973924380704,
-      "grad_norm": 1.4598643779754639,
-      "learning_rate": 9.03678391959799e-05,
-      "loss": 4.6012,
-      "step": 10086
-    },
-    {
-      "epoch": 5.2604954367666235,
-      "grad_norm": 1.5161511898040771,
-      "learning_rate": 9.036683417085428e-05,
-      "loss": 5.312,
-      "step": 10087
-    },
-    {
-      "epoch": 5.261016949152542,
-      "grad_norm": 1.7268807888031006,
-      "learning_rate": 9.036582914572864e-05,
-      "loss": 5.2498,
-      "step": 10088
-    },
-    {
-      "epoch": 5.2615384615384615,
-      "grad_norm": 1.5388851165771484,
-      "learning_rate": 9.036482412060302e-05,
-      "loss": 5.3284,
-      "step": 10089
-    },
-    {
-      "epoch": 5.262059973924381,
-      "grad_norm": 1.4702500104904175,
-      "learning_rate": 9.036381909547739e-05,
-      "loss": 5.5491,
-      "step": 10090
-    },
-    {
-      "epoch": 5.2625814863102995,
-      "grad_norm": 1.4577462673187256,
-      "learning_rate": 9.036281407035176e-05,
-      "loss": 5.192,
-      "step": 10091
-    },
-    {
-      "epoch": 5.263102998696219,
-      "grad_norm": 1.6016701459884644,
-      "learning_rate": 9.036180904522613e-05,
-      "loss": 5.8432,
-      "step": 10092
-    },
-    {
-      "epoch": 5.2636245110821385,
-      "grad_norm": 1.4500590562820435,
-      "learning_rate": 9.03608040201005e-05,
-      "loss": 5.5604,
-      "step": 10093
-    },
-    {
-      "epoch": 5.264146023468057,
-      "grad_norm": 1.4224238395690918,
-      "learning_rate": 9.035979899497487e-05,
-      "loss": 5.474,
-      "step": 10094
-    },
-    {
-      "epoch": 5.2646675358539765,
-      "grad_norm": 1.4048842191696167,
-      "learning_rate": 9.035879396984925e-05,
-      "loss": 5.4519,
-      "step": 10095
-    },
-    {
-      "epoch": 5.265189048239896,
-      "grad_norm": 1.4362159967422485,
-      "learning_rate": 9.035778894472363e-05,
-      "loss": 5.798,
-      "step": 10096
-    },
-    {
-      "epoch": 5.2657105606258146,
-      "grad_norm": 1.4492383003234863,
-      "learning_rate": 9.0356783919598e-05,
-      "loss": 4.7177,
-      "step": 10097
-    },
-    {
-      "epoch": 5.266232073011734,
-      "grad_norm": 1.4999266862869263,
-      "learning_rate": 9.035577889447237e-05,
-      "loss": 5.8647,
-      "step": 10098
-    },
-    {
-      "epoch": 5.2667535853976535,
-      "grad_norm": 1.5726128816604614,
-      "learning_rate": 9.035477386934675e-05,
-      "loss": 5.254,
-      "step": 10099
-    },
-    {
-      "epoch": 5.267275097783572,
-      "grad_norm": 1.466484785079956,
-      "learning_rate": 9.035376884422111e-05,
-      "loss": 5.8386,
-      "step": 10100
-    },
-    {
-      "epoch": 5.2677966101694915,
-      "grad_norm": 1.5123391151428223,
-      "learning_rate": 9.035276381909549e-05,
-      "loss": 5.3832,
-      "step": 10101
-    },
-    {
-      "epoch": 5.268318122555411,
-      "grad_norm": 1.5916695594787598,
-      "learning_rate": 9.035175879396985e-05,
-      "loss": 5.2203,
-      "step": 10102
-    },
-    {
-      "epoch": 5.26883963494133,
-      "grad_norm": 1.3694965839385986,
-      "learning_rate": 9.035075376884422e-05,
-      "loss": 5.1611,
-      "step": 10103
-    },
-    {
-      "epoch": 5.269361147327249,
-      "grad_norm": 1.5151638984680176,
-      "learning_rate": 9.03497487437186e-05,
-      "loss": 5.9515,
-      "step": 10104
-    },
-    {
-      "epoch": 5.2698826597131685,
-      "grad_norm": 1.4841432571411133,
-      "learning_rate": 9.034874371859296e-05,
-      "loss": 5.7103,
-      "step": 10105
-    },
-    {
-      "epoch": 5.270404172099087,
-      "grad_norm": 1.4815245866775513,
-      "learning_rate": 9.034773869346734e-05,
-      "loss": 5.9809,
-      "step": 10106
-    },
-    {
-      "epoch": 5.2709256844850065,
-      "grad_norm": 1.4271259307861328,
-      "learning_rate": 9.034673366834172e-05,
-      "loss": 5.6026,
-      "step": 10107
-    },
-    {
-      "epoch": 5.271447196870926,
-      "grad_norm": 1.3901634216308594,
-      "learning_rate": 9.034572864321609e-05,
-      "loss": 5.9341,
-      "step": 10108
-    },
-    {
-      "epoch": 5.271968709256845,
-      "grad_norm": 1.4707971811294556,
-      "learning_rate": 9.034472361809046e-05,
-      "loss": 5.6568,
-      "step": 10109
-    },
-    {
-      "epoch": 5.272490221642764,
-      "grad_norm": 1.428215503692627,
-      "learning_rate": 9.034371859296483e-05,
-      "loss": 5.9458,
-      "step": 10110
-    },
-    {
-      "epoch": 5.2730117340286835,
-      "grad_norm": 1.3498142957687378,
-      "learning_rate": 9.03427135678392e-05,
-      "loss": 5.9059,
-      "step": 10111
-    },
-    {
-      "epoch": 5.273533246414602,
-      "grad_norm": 1.3889614343643188,
-      "learning_rate": 9.034170854271358e-05,
-      "loss": 5.501,
-      "step": 10112
-    },
-    {
-      "epoch": 5.2740547588005215,
-      "grad_norm": 1.4207524061203003,
-      "learning_rate": 9.034070351758794e-05,
-      "loss": 5.5392,
-      "step": 10113
-    },
-    {
-      "epoch": 5.274576271186441,
-      "grad_norm": 1.3346121311187744,
-      "learning_rate": 9.033969849246232e-05,
-      "loss": 5.434,
-      "step": 10114
-    },
-    {
-      "epoch": 5.27509778357236,
-      "grad_norm": 1.5477559566497803,
-      "learning_rate": 9.033869346733668e-05,
-      "loss": 5.4977,
-      "step": 10115
-    },
-    {
-      "epoch": 5.275619295958279,
-      "grad_norm": 1.4411602020263672,
-      "learning_rate": 9.033768844221106e-05,
-      "loss": 5.6205,
-      "step": 10116
-    },
-    {
-      "epoch": 5.2761408083441985,
-      "grad_norm": 1.404096245765686,
-      "learning_rate": 9.033668341708544e-05,
-      "loss": 5.7789,
-      "step": 10117
-    },
-    {
-      "epoch": 5.276662320730117,
-      "grad_norm": 1.3902307748794556,
-      "learning_rate": 9.03356783919598e-05,
-      "loss": 5.6599,
-      "step": 10118
-    },
-    {
-      "epoch": 5.277183833116037,
-      "grad_norm": 1.4467017650604248,
-      "learning_rate": 9.033467336683418e-05,
-      "loss": 5.8722,
-      "step": 10119
-    },
-    {
-      "epoch": 5.277705345501956,
-      "grad_norm": 1.5538296699523926,
-      "learning_rate": 9.033366834170855e-05,
-      "loss": 5.5758,
-      "step": 10120
-    },
-    {
-      "epoch": 5.278226857887875,
-      "grad_norm": 1.6112912893295288,
-      "learning_rate": 9.033266331658292e-05,
-      "loss": 5.4212,
-      "step": 10121
-    },
-    {
-      "epoch": 5.278748370273794,
-      "grad_norm": 1.451098084449768,
-      "learning_rate": 9.033165829145729e-05,
-      "loss": 5.9175,
-      "step": 10122
-    },
-    {
-      "epoch": 5.2792698826597135,
-      "grad_norm": 1.2418203353881836,
-      "learning_rate": 9.033065326633167e-05,
-      "loss": 5.2708,
-      "step": 10123
-    },
-    {
-      "epoch": 5.279791395045632,
-      "grad_norm": 1.3613461256027222,
-      "learning_rate": 9.032964824120603e-05,
-      "loss": 5.648,
-      "step": 10124
-    },
-    {
-      "epoch": 5.280312907431552,
-      "grad_norm": 1.3433095216751099,
-      "learning_rate": 9.032864321608041e-05,
-      "loss": 6.0698,
-      "step": 10125
-    },
-    {
-      "epoch": 5.280834419817471,
-      "grad_norm": 1.5903409719467163,
-      "learning_rate": 9.032763819095477e-05,
-      "loss": 5.3012,
-      "step": 10126
-    },
-    {
-      "epoch": 5.28135593220339,
-      "grad_norm": 1.3229511976242065,
-      "learning_rate": 9.032663316582915e-05,
-      "loss": 5.7769,
-      "step": 10127
-    },
-    {
-      "epoch": 5.281877444589309,
-      "grad_norm": 1.435637354850769,
-      "learning_rate": 9.032562814070353e-05,
-      "loss": 5.5244,
-      "step": 10128
-    },
-    {
-      "epoch": 5.2823989569752285,
-      "grad_norm": 1.488949179649353,
-      "learning_rate": 9.032462311557789e-05,
-      "loss": 5.8349,
-      "step": 10129
-    },
-    {
-      "epoch": 5.282920469361147,
-      "grad_norm": 1.4263296127319336,
-      "learning_rate": 9.032361809045227e-05,
-      "loss": 5.5644,
-      "step": 10130
-    },
-    {
-      "epoch": 5.283441981747067,
-      "grad_norm": 1.4931843280792236,
-      "learning_rate": 9.032261306532663e-05,
-      "loss": 5.1947,
-      "step": 10131
-    },
-    {
-      "epoch": 5.283963494132986,
-      "grad_norm": 1.652774691581726,
-      "learning_rate": 9.032160804020101e-05,
-      "loss": 5.7006,
-      "step": 10132
-    },
-    {
-      "epoch": 5.284485006518905,
-      "grad_norm": 1.3690481185913086,
-      "learning_rate": 9.032060301507538e-05,
-      "loss": 5.689,
-      "step": 10133
-    },
-    {
-      "epoch": 5.285006518904824,
-      "grad_norm": 1.6037957668304443,
-      "learning_rate": 9.031959798994975e-05,
-      "loss": 5.7437,
-      "step": 10134
-    },
-    {
-      "epoch": 5.2855280312907436,
-      "grad_norm": 1.5038952827453613,
-      "learning_rate": 9.031859296482412e-05,
-      "loss": 5.6974,
-      "step": 10135
-    },
-    {
-      "epoch": 5.286049543676662,
-      "grad_norm": 1.4818822145462036,
-      "learning_rate": 9.03175879396985e-05,
-      "loss": 5.8746,
-      "step": 10136
-    },
-    {
-      "epoch": 5.286571056062582,
-      "grad_norm": 1.4067444801330566,
-      "learning_rate": 9.031658291457287e-05,
-      "loss": 5.7685,
-      "step": 10137
-    },
-    {
-      "epoch": 5.287092568448501,
-      "grad_norm": 1.4086865186691284,
-      "learning_rate": 9.031557788944725e-05,
-      "loss": 5.7503,
-      "step": 10138
-    },
-    {
-      "epoch": 5.28761408083442,
-      "grad_norm": 1.4859592914581299,
-      "learning_rate": 9.031457286432162e-05,
-      "loss": 6.0192,
-      "step": 10139
-    },
-    {
-      "epoch": 5.288135593220339,
-      "grad_norm": 1.4527862071990967,
-      "learning_rate": 9.0313567839196e-05,
-      "loss": 5.6315,
-      "step": 10140
-    },
-    {
-      "epoch": 5.288657105606259,
-      "grad_norm": 1.4255590438842773,
-      "learning_rate": 9.031256281407036e-05,
-      "loss": 5.8929,
-      "step": 10141
-    },
-    {
-      "epoch": 5.289178617992177,
-      "grad_norm": 1.403523325920105,
-      "learning_rate": 9.031155778894472e-05,
-      "loss": 5.5291,
-      "step": 10142
-    },
-    {
-      "epoch": 5.289700130378097,
-      "grad_norm": 1.3358978033065796,
-      "learning_rate": 9.03105527638191e-05,
-      "loss": 5.4483,
-      "step": 10143
-    },
-    {
-      "epoch": 5.290221642764015,
-      "grad_norm": 1.4017674922943115,
-      "learning_rate": 9.030954773869346e-05,
-      "loss": 5.7032,
-      "step": 10144
-    },
-    {
-      "epoch": 5.290743155149935,
-      "grad_norm": 1.4424452781677246,
-      "learning_rate": 9.030854271356784e-05,
-      "loss": 5.6551,
-      "step": 10145
-    },
-    {
-      "epoch": 5.291264667535854,
-      "grad_norm": 1.4213212728500366,
-      "learning_rate": 9.03075376884422e-05,
-      "loss": 6.1358,
-      "step": 10146
-    },
-    {
-      "epoch": 5.291786179921774,
-      "grad_norm": 1.9626235961914062,
-      "learning_rate": 9.030653266331658e-05,
-      "loss": 4.9429,
-      "step": 10147
-    },
-    {
-      "epoch": 5.292307692307692,
-      "grad_norm": 1.463068962097168,
-      "learning_rate": 9.030552763819096e-05,
-      "loss": 5.5942,
-      "step": 10148
-    },
-    {
-      "epoch": 5.292829204693612,
-      "grad_norm": 1.4770915508270264,
-      "learning_rate": 9.030452261306534e-05,
-      "loss": 5.6926,
-      "step": 10149
-    },
-    {
-      "epoch": 5.29335071707953,
-      "grad_norm": 1.3663451671600342,
-      "learning_rate": 9.03035175879397e-05,
-      "loss": 5.9934,
-      "step": 10150
-    },
-    {
-      "epoch": 5.29387222946545,
-      "grad_norm": 1.4516716003417969,
-      "learning_rate": 9.030251256281408e-05,
-      "loss": 5.9346,
-      "step": 10151
-    },
-    {
-      "epoch": 5.294393741851369,
-      "grad_norm": 1.3995884656906128,
-      "learning_rate": 9.030150753768845e-05,
-      "loss": 5.9001,
-      "step": 10152
-    },
-    {
-      "epoch": 5.294915254237289,
-      "grad_norm": 1.6173863410949707,
-      "learning_rate": 9.030050251256282e-05,
-      "loss": 5.3547,
-      "step": 10153
-    },
-    {
-      "epoch": 5.295436766623207,
-      "grad_norm": 1.5221269130706787,
-      "learning_rate": 9.029949748743719e-05,
-      "loss": 5.6693,
-      "step": 10154
-    },
-    {
-      "epoch": 5.295958279009127,
-      "grad_norm": 1.488456130027771,
-      "learning_rate": 9.029849246231155e-05,
-      "loss": 5.5436,
-      "step": 10155
-    },
-    {
-      "epoch": 5.296479791395045,
-      "grad_norm": 1.3692303895950317,
-      "learning_rate": 9.029748743718593e-05,
-      "loss": 5.8905,
-      "step": 10156
-    },
-    {
-      "epoch": 5.297001303780965,
-      "grad_norm": 1.6565728187561035,
-      "learning_rate": 9.02964824120603e-05,
-      "loss": 5.4402,
-      "step": 10157
-    },
-    {
-      "epoch": 5.297522816166884,
-      "grad_norm": 1.3411543369293213,
-      "learning_rate": 9.029547738693467e-05,
-      "loss": 5.7118,
-      "step": 10158
-    },
-    {
-      "epoch": 5.298044328552803,
-      "grad_norm": 1.4342840909957886,
-      "learning_rate": 9.029447236180905e-05,
-      "loss": 6.162,
-      "step": 10159
-    },
-    {
-      "epoch": 5.298565840938722,
-      "grad_norm": 1.4498540163040161,
-      "learning_rate": 9.029346733668343e-05,
-      "loss": 5.5208,
-      "step": 10160
-    },
-    {
-      "epoch": 5.299087353324642,
-      "grad_norm": 1.4588541984558105,
-      "learning_rate": 9.029246231155779e-05,
-      "loss": 5.5778,
-      "step": 10161
-    },
-    {
-      "epoch": 5.29960886571056,
-      "grad_norm": 1.4278212785720825,
-      "learning_rate": 9.029145728643217e-05,
-      "loss": 5.8526,
-      "step": 10162
-    },
-    {
-      "epoch": 5.30013037809648,
-      "grad_norm": 1.4877732992172241,
-      "learning_rate": 9.029045226130653e-05,
-      "loss": 5.6603,
-      "step": 10163
-    },
-    {
-      "epoch": 5.300651890482399,
-      "grad_norm": 1.5012656450271606,
-      "learning_rate": 9.028944723618091e-05,
-      "loss": 5.5734,
-      "step": 10164
-    },
-    {
-      "epoch": 5.301173402868318,
-      "grad_norm": 1.3371511697769165,
-      "learning_rate": 9.028844221105528e-05,
-      "loss": 5.3745,
-      "step": 10165
-    },
-    {
-      "epoch": 5.301694915254237,
-      "grad_norm": 1.5119991302490234,
-      "learning_rate": 9.028743718592965e-05,
-      "loss": 5.4259,
-      "step": 10166
-    },
-    {
-      "epoch": 5.302216427640157,
-      "grad_norm": 1.4748679399490356,
-      "learning_rate": 9.028643216080402e-05,
-      "loss": 4.708,
-      "step": 10167
-    },
-    {
-      "epoch": 5.302737940026075,
-      "grad_norm": 1.6002538204193115,
-      "learning_rate": 9.02854271356784e-05,
-      "loss": 5.6048,
-      "step": 10168
-    },
-    {
-      "epoch": 5.303259452411995,
-      "grad_norm": 1.7044093608856201,
-      "learning_rate": 9.028442211055277e-05,
-      "loss": 4.8962,
-      "step": 10169
-    },
-    {
-      "epoch": 5.303780964797914,
-      "grad_norm": 1.4794741868972778,
-      "learning_rate": 9.028341708542714e-05,
-      "loss": 5.8414,
-      "step": 10170
-    },
-    {
-      "epoch": 5.304302477183833,
-      "grad_norm": 1.54389226436615,
-      "learning_rate": 9.028241206030152e-05,
-      "loss": 5.5981,
-      "step": 10171
-    },
-    {
-      "epoch": 5.304823989569752,
-      "grad_norm": 1.5560040473937988,
-      "learning_rate": 9.028140703517588e-05,
-      "loss": 5.6244,
-      "step": 10172
-    },
-    {
-      "epoch": 5.305345501955672,
-      "grad_norm": 1.4358760118484497,
-      "learning_rate": 9.028040201005026e-05,
-      "loss": 5.9609,
-      "step": 10173
-    },
-    {
-      "epoch": 5.30586701434159,
-      "grad_norm": 1.410981297492981,
-      "learning_rate": 9.027939698492462e-05,
-      "loss": 5.8108,
-      "step": 10174
-    },
-    {
-      "epoch": 5.30638852672751,
-      "grad_norm": 1.4781606197357178,
-      "learning_rate": 9.0278391959799e-05,
-      "loss": 5.7265,
-      "step": 10175
-    },
-    {
-      "epoch": 5.306910039113429,
-      "grad_norm": 1.4008976221084595,
-      "learning_rate": 9.027738693467337e-05,
-      "loss": 5.4647,
-      "step": 10176
-    },
-    {
-      "epoch": 5.307431551499348,
-      "grad_norm": 1.357210397720337,
-      "learning_rate": 9.027638190954774e-05,
-      "loss": 5.7917,
-      "step": 10177
-    },
-    {
-      "epoch": 5.307953063885267,
-      "grad_norm": 1.4074064493179321,
-      "learning_rate": 9.027537688442211e-05,
-      "loss": 5.697,
-      "step": 10178
-    },
-    {
-      "epoch": 5.308474576271187,
-      "grad_norm": 1.404657244682312,
-      "learning_rate": 9.027437185929648e-05,
-      "loss": 5.8148,
-      "step": 10179
-    },
-    {
-      "epoch": 5.308996088657105,
-      "grad_norm": 1.389084815979004,
-      "learning_rate": 9.027336683417086e-05,
-      "loss": 5.7696,
-      "step": 10180
-    },
-    {
-      "epoch": 5.309517601043025,
-      "grad_norm": 1.4845212697982788,
-      "learning_rate": 9.027236180904523e-05,
-      "loss": 5.6774,
-      "step": 10181
-    },
-    {
-      "epoch": 5.310039113428944,
-      "grad_norm": 1.489313006401062,
-      "learning_rate": 9.02713567839196e-05,
-      "loss": 5.7094,
-      "step": 10182
-    },
-    {
-      "epoch": 5.310560625814863,
-      "grad_norm": 1.4481770992279053,
-      "learning_rate": 9.027035175879397e-05,
-      "loss": 5.4335,
-      "step": 10183
-    },
-    {
-      "epoch": 5.311082138200782,
-      "grad_norm": 1.3650178909301758,
-      "learning_rate": 9.026934673366835e-05,
-      "loss": 5.6826,
-      "step": 10184
-    },
-    {
-      "epoch": 5.311603650586702,
-      "grad_norm": 1.5408092737197876,
-      "learning_rate": 9.026834170854271e-05,
-      "loss": 5.9372,
-      "step": 10185
-    },
-    {
-      "epoch": 5.31212516297262,
-      "grad_norm": 1.6812245845794678,
-      "learning_rate": 9.026733668341709e-05,
-      "loss": 5.8724,
-      "step": 10186
-    },
-    {
-      "epoch": 5.31264667535854,
-      "grad_norm": 1.5636911392211914,
-      "learning_rate": 9.026633165829145e-05,
-      "loss": 5.2088,
-      "step": 10187
-    },
-    {
-      "epoch": 5.313168187744459,
-      "grad_norm": 1.507303237915039,
-      "learning_rate": 9.026532663316583e-05,
-      "loss": 5.7872,
-      "step": 10188
-    },
-    {
-      "epoch": 5.313689700130378,
-      "grad_norm": 1.3101260662078857,
-      "learning_rate": 9.026432160804021e-05,
-      "loss": 5.7012,
-      "step": 10189
-    },
-    {
-      "epoch": 5.314211212516297,
-      "grad_norm": 1.4885419607162476,
-      "learning_rate": 9.026331658291459e-05,
-      "loss": 5.466,
-      "step": 10190
-    },
-    {
-      "epoch": 5.314732724902217,
-      "grad_norm": 1.385655164718628,
-      "learning_rate": 9.026231155778895e-05,
-      "loss": 5.8726,
-      "step": 10191
-    },
-    {
-      "epoch": 5.315254237288135,
-      "grad_norm": 1.4238570928573608,
-      "learning_rate": 9.026130653266333e-05,
-      "loss": 5.8543,
-      "step": 10192
-    },
-    {
-      "epoch": 5.315775749674055,
-      "grad_norm": 1.2176191806793213,
-      "learning_rate": 9.026030150753769e-05,
-      "loss": 5.9873,
-      "step": 10193
-    },
-    {
-      "epoch": 5.316297262059974,
-      "grad_norm": 1.4420068264007568,
-      "learning_rate": 9.025929648241207e-05,
-      "loss": 5.5586,
-      "step": 10194
-    },
-    {
-      "epoch": 5.316818774445893,
-      "grad_norm": 1.4391130208969116,
-      "learning_rate": 9.025829145728644e-05,
-      "loss": 5.8113,
-      "step": 10195
-    },
-    {
-      "epoch": 5.317340286831812,
-      "grad_norm": 1.4397854804992676,
-      "learning_rate": 9.02572864321608e-05,
-      "loss": 5.9375,
-      "step": 10196
-    },
-    {
-      "epoch": 5.317861799217732,
-      "grad_norm": 1.4512836933135986,
-      "learning_rate": 9.025628140703518e-05,
-      "loss": 5.548,
-      "step": 10197
-    },
-    {
-      "epoch": 5.31838331160365,
-      "grad_norm": 1.4951709508895874,
-      "learning_rate": 9.025527638190954e-05,
-      "loss": 5.8345,
-      "step": 10198
-    },
-    {
-      "epoch": 5.31890482398957,
-      "grad_norm": 1.5930612087249756,
-      "learning_rate": 9.025427135678392e-05,
-      "loss": 5.5683,
-      "step": 10199
-    },
-    {
-      "epoch": 5.319426336375489,
-      "grad_norm": 1.3639918565750122,
-      "learning_rate": 9.02532663316583e-05,
-      "loss": 5.9459,
-      "step": 10200
-    },
-    {
-      "epoch": 5.319947848761408,
-      "grad_norm": 1.4860025644302368,
-      "learning_rate": 9.025226130653268e-05,
-      "loss": 5.4401,
-      "step": 10201
-    },
-    {
-      "epoch": 5.320469361147327,
-      "grad_norm": 1.5039597749710083,
-      "learning_rate": 9.025125628140704e-05,
-      "loss": 5.9156,
-      "step": 10202
-    },
-    {
-      "epoch": 5.320990873533247,
-      "grad_norm": 1.4666608572006226,
-      "learning_rate": 9.025025125628142e-05,
-      "loss": 5.5488,
-      "step": 10203
-    },
-    {
-      "epoch": 5.321512385919165,
-      "grad_norm": 1.3759040832519531,
-      "learning_rate": 9.024924623115578e-05,
-      "loss": 4.2594,
-      "step": 10204
-    },
-    {
-      "epoch": 5.322033898305085,
-      "grad_norm": 1.4700740575790405,
-      "learning_rate": 9.024824120603016e-05,
-      "loss": 5.9183,
-      "step": 10205
-    },
-    {
-      "epoch": 5.322555410691004,
-      "grad_norm": 1.5964179039001465,
-      "learning_rate": 9.024723618090452e-05,
-      "loss": 5.6895,
-      "step": 10206
-    },
-    {
-      "epoch": 5.323076923076923,
-      "grad_norm": 1.4286209344863892,
-      "learning_rate": 9.02462311557789e-05,
-      "loss": 5.9161,
-      "step": 10207
-    },
-    {
-      "epoch": 5.323598435462842,
-      "grad_norm": 1.3978111743927002,
-      "learning_rate": 9.024522613065327e-05,
-      "loss": 6.1078,
-      "step": 10208
-    },
-    {
-      "epoch": 5.324119947848762,
-      "grad_norm": 1.5734169483184814,
-      "learning_rate": 9.024422110552764e-05,
-      "loss": 5.2924,
-      "step": 10209
-    },
-    {
-      "epoch": 5.32464146023468,
-      "grad_norm": 1.6161092519760132,
-      "learning_rate": 9.024321608040202e-05,
-      "loss": 5.6936,
-      "step": 10210
-    },
-    {
-      "epoch": 5.3251629726206,
-      "grad_norm": 1.5728449821472168,
-      "learning_rate": 9.024221105527639e-05,
-      "loss": 5.4333,
-      "step": 10211
-    },
-    {
-      "epoch": 5.325684485006519,
-      "grad_norm": 1.773559331893921,
-      "learning_rate": 9.024120603015076e-05,
-      "loss": 5.4845,
-      "step": 10212
-    },
-    {
-      "epoch": 5.326205997392438,
-      "grad_norm": 1.5202322006225586,
-      "learning_rate": 9.024020100502513e-05,
-      "loss": 5.5811,
-      "step": 10213
-    },
-    {
-      "epoch": 5.326727509778357,
-      "grad_norm": 1.4308488368988037,
-      "learning_rate": 9.02391959798995e-05,
-      "loss": 5.2266,
-      "step": 10214
-    },
-    {
-      "epoch": 5.327249022164277,
-      "grad_norm": 1.4669445753097534,
-      "learning_rate": 9.023819095477387e-05,
-      "loss": 5.6467,
-      "step": 10215
-    },
-    {
-      "epoch": 5.327770534550195,
-      "grad_norm": 1.5192683935165405,
-      "learning_rate": 9.023718592964825e-05,
-      "loss": 5.4324,
-      "step": 10216
-    },
-    {
-      "epoch": 5.328292046936115,
-      "grad_norm": 1.468769907951355,
-      "learning_rate": 9.023618090452261e-05,
-      "loss": 5.3246,
-      "step": 10217
-    },
-    {
-      "epoch": 5.328813559322034,
-      "grad_norm": 1.3620758056640625,
-      "learning_rate": 9.023517587939699e-05,
-      "loss": 5.8385,
-      "step": 10218
-    },
-    {
-      "epoch": 5.329335071707953,
-      "grad_norm": 1.3868898153305054,
-      "learning_rate": 9.023417085427135e-05,
-      "loss": 5.5634,
-      "step": 10219
-    },
-    {
-      "epoch": 5.329856584093872,
-      "grad_norm": 1.8883161544799805,
-      "learning_rate": 9.023316582914573e-05,
-      "loss": 5.0034,
-      "step": 10220
-    },
-    {
-      "epoch": 5.330378096479792,
-      "grad_norm": 1.5229982137680054,
-      "learning_rate": 9.023216080402011e-05,
-      "loss": 4.8879,
-      "step": 10221
-    },
-    {
-      "epoch": 5.33089960886571,
-      "grad_norm": 1.6382133960723877,
-      "learning_rate": 9.023115577889447e-05,
-      "loss": 5.4922,
-      "step": 10222
-    },
-    {
-      "epoch": 5.33142112125163,
-      "grad_norm": 1.6447757482528687,
-      "learning_rate": 9.023015075376885e-05,
-      "loss": 5.8495,
-      "step": 10223
-    },
-    {
-      "epoch": 5.331942633637549,
-      "grad_norm": 1.4234298467636108,
-      "learning_rate": 9.022914572864322e-05,
-      "loss": 5.5402,
-      "step": 10224
-    },
-    {
-      "epoch": 5.332464146023468,
-      "grad_norm": 1.7102653980255127,
-      "learning_rate": 9.02281407035176e-05,
-      "loss": 5.6542,
-      "step": 10225
-    },
-    {
-      "epoch": 5.332985658409387,
-      "grad_norm": 1.505831241607666,
-      "learning_rate": 9.022713567839196e-05,
-      "loss": 5.2729,
-      "step": 10226
-    },
-    {
-      "epoch": 5.333507170795307,
-      "grad_norm": 1.5086112022399902,
-      "learning_rate": 9.022613065326634e-05,
-      "loss": 5.6718,
-      "step": 10227
-    },
-    {
-      "epoch": 5.334028683181225,
-      "grad_norm": 1.5301624536514282,
-      "learning_rate": 9.02251256281407e-05,
-      "loss": 5.4514,
-      "step": 10228
-    },
-    {
-      "epoch": 5.334550195567145,
-      "grad_norm": 1.4157813787460327,
-      "learning_rate": 9.022412060301508e-05,
-      "loss": 5.673,
-      "step": 10229
-    },
-    {
-      "epoch": 5.335071707953064,
-      "grad_norm": 1.4857059717178345,
-      "learning_rate": 9.022311557788946e-05,
-      "loss": 5.7814,
-      "step": 10230
-    },
-    {
-      "epoch": 5.335593220338983,
-      "grad_norm": 1.4703820943832397,
-      "learning_rate": 9.022211055276383e-05,
-      "loss": 5.9424,
-      "step": 10231
-    },
-    {
-      "epoch": 5.336114732724902,
-      "grad_norm": 1.5231715440750122,
-      "learning_rate": 9.02211055276382e-05,
-      "loss": 5.835,
-      "step": 10232
-    },
-    {
-      "epoch": 5.336636245110822,
-      "grad_norm": 1.3984918594360352,
-      "learning_rate": 9.022010050251258e-05,
-      "loss": 5.8445,
-      "step": 10233
-    },
-    {
-      "epoch": 5.33715775749674,
-      "grad_norm": 1.4330426454544067,
-      "learning_rate": 9.021909547738694e-05,
-      "loss": 5.6257,
-      "step": 10234
-    },
-    {
-      "epoch": 5.33767926988266,
-      "grad_norm": 1.496041178703308,
-      "learning_rate": 9.02180904522613e-05,
-      "loss": 5.6275,
-      "step": 10235
-    },
-    {
-      "epoch": 5.338200782268579,
-      "grad_norm": 1.6019511222839355,
-      "learning_rate": 9.021708542713568e-05,
-      "loss": 4.8149,
-      "step": 10236
-    },
-    {
-      "epoch": 5.338722294654498,
-      "grad_norm": 1.4021120071411133,
-      "learning_rate": 9.021608040201005e-05,
-      "loss": 5.6017,
-      "step": 10237
-    },
-    {
-      "epoch": 5.339243807040417,
-      "grad_norm": 1.4280763864517212,
-      "learning_rate": 9.021507537688442e-05,
-      "loss": 5.6507,
-      "step": 10238
-    },
-    {
-      "epoch": 5.339765319426336,
-      "grad_norm": 1.4885696172714233,
-      "learning_rate": 9.021407035175879e-05,
-      "loss": 5.3205,
-      "step": 10239
-    },
-    {
-      "epoch": 5.340286831812255,
-      "grad_norm": 1.4745653867721558,
-      "learning_rate": 9.021306532663317e-05,
-      "loss": 5.5354,
-      "step": 10240
-    },
-    {
-      "epoch": 5.340808344198175,
-      "grad_norm": 1.3587496280670166,
-      "learning_rate": 9.021206030150754e-05,
-      "loss": 5.6468,
-      "step": 10241
-    },
-    {
-      "epoch": 5.341329856584094,
-      "grad_norm": 1.3264514207839966,
-      "learning_rate": 9.021105527638192e-05,
-      "loss": 5.8478,
-      "step": 10242
-    },
-    {
-      "epoch": 5.341851368970013,
-      "grad_norm": 1.437589168548584,
-      "learning_rate": 9.021005025125629e-05,
-      "loss": 5.8311,
-      "step": 10243
-    },
-    {
-      "epoch": 5.342372881355932,
-      "grad_norm": 1.343753457069397,
-      "learning_rate": 9.020904522613066e-05,
-      "loss": 5.2333,
-      "step": 10244
-    },
-    {
-      "epoch": 5.342894393741851,
-      "grad_norm": 1.4322502613067627,
-      "learning_rate": 9.020804020100503e-05,
-      "loss": 5.5893,
-      "step": 10245
-    },
-    {
-      "epoch": 5.34341590612777,
-      "grad_norm": 1.6897600889205933,
-      "learning_rate": 9.02070351758794e-05,
-      "loss": 5.3949,
-      "step": 10246
-    },
-    {
-      "epoch": 5.34393741851369,
-      "grad_norm": 1.4204440116882324,
-      "learning_rate": 9.020603015075377e-05,
-      "loss": 5.4185,
-      "step": 10247
-    },
-    {
-      "epoch": 5.344458930899609,
-      "grad_norm": 1.366647481918335,
-      "learning_rate": 9.020502512562813e-05,
-      "loss": 5.6635,
-      "step": 10248
-    },
-    {
-      "epoch": 5.344980443285528,
-      "grad_norm": 1.4737663269042969,
-      "learning_rate": 9.020402010050251e-05,
-      "loss": 5.6163,
-      "step": 10249
-    },
-    {
-      "epoch": 5.345501955671447,
-      "grad_norm": 1.4024121761322021,
-      "learning_rate": 9.020301507537689e-05,
-      "loss": 5.4959,
-      "step": 10250
-    },
-    {
-      "epoch": 5.346023468057366,
-      "grad_norm": 1.5999839305877686,
-      "learning_rate": 9.020201005025127e-05,
-      "loss": 5.4925,
-      "step": 10251
-    },
-    {
-      "epoch": 5.346544980443285,
-      "grad_norm": 1.316460132598877,
-      "learning_rate": 9.020100502512563e-05,
-      "loss": 5.7873,
-      "step": 10252
-    },
-    {
-      "epoch": 5.347066492829205,
-      "grad_norm": 1.504062533378601,
-      "learning_rate": 9.020000000000001e-05,
-      "loss": 5.1032,
-      "step": 10253
-    },
-    {
-      "epoch": 5.3475880052151235,
-      "grad_norm": 1.4356709718704224,
-      "learning_rate": 9.019899497487437e-05,
-      "loss": 5.7118,
-      "step": 10254
-    },
-    {
-      "epoch": 5.348109517601043,
-      "grad_norm": 1.5183720588684082,
-      "learning_rate": 9.019798994974875e-05,
-      "loss": 5.2328,
-      "step": 10255
-    },
-    {
-      "epoch": 5.348631029986962,
-      "grad_norm": 1.6235761642456055,
-      "learning_rate": 9.019698492462312e-05,
-      "loss": 5.1812,
-      "step": 10256
-    },
-    {
-      "epoch": 5.349152542372881,
-      "grad_norm": 1.476263165473938,
-      "learning_rate": 9.01959798994975e-05,
-      "loss": 5.8137,
-      "step": 10257
-    },
-    {
-      "epoch": 5.3496740547588,
-      "grad_norm": 1.47592031955719,
-      "learning_rate": 9.019497487437186e-05,
-      "loss": 4.9739,
-      "step": 10258
-    },
-    {
-      "epoch": 5.35019556714472,
-      "grad_norm": 1.382989525794983,
-      "learning_rate": 9.019396984924624e-05,
-      "loss": 5.8285,
-      "step": 10259
-    },
-    {
-      "epoch": 5.3507170795306385,
-      "grad_norm": 1.4239100217819214,
-      "learning_rate": 9.01929648241206e-05,
-      "loss": 5.3573,
-      "step": 10260
-    },
-    {
-      "epoch": 5.351238591916558,
-      "grad_norm": 1.7319763898849487,
-      "learning_rate": 9.019195979899498e-05,
-      "loss": 5.3675,
-      "step": 10261
-    },
-    {
-      "epoch": 5.351760104302477,
-      "grad_norm": 1.3405685424804688,
-      "learning_rate": 9.019095477386936e-05,
-      "loss": 5.9522,
-      "step": 10262
-    },
-    {
-      "epoch": 5.352281616688396,
-      "grad_norm": 1.3371466398239136,
-      "learning_rate": 9.018994974874372e-05,
-      "loss": 5.5198,
-      "step": 10263
-    },
-    {
-      "epoch": 5.352803129074315,
-      "grad_norm": 1.3950738906860352,
-      "learning_rate": 9.01889447236181e-05,
-      "loss": 6.0362,
-      "step": 10264
-    },
-    {
-      "epoch": 5.353324641460235,
-      "grad_norm": 1.435194730758667,
-      "learning_rate": 9.018793969849246e-05,
-      "loss": 5.6534,
-      "step": 10265
-    },
-    {
-      "epoch": 5.3538461538461535,
-      "grad_norm": 1.60768723487854,
-      "learning_rate": 9.018693467336684e-05,
-      "loss": 5.7581,
-      "step": 10266
-    },
-    {
-      "epoch": 5.354367666232073,
-      "grad_norm": 1.4624313116073608,
-      "learning_rate": 9.01859296482412e-05,
-      "loss": 5.6737,
-      "step": 10267
-    },
-    {
-      "epoch": 5.354889178617992,
-      "grad_norm": 1.5445939302444458,
-      "learning_rate": 9.018492462311558e-05,
-      "loss": 5.6502,
-      "step": 10268
-    },
-    {
-      "epoch": 5.355410691003911,
-      "grad_norm": 1.3557714223861694,
-      "learning_rate": 9.018391959798995e-05,
-      "loss": 5.9079,
-      "step": 10269
-    },
-    {
-      "epoch": 5.3559322033898304,
-      "grad_norm": 1.7879862785339355,
-      "learning_rate": 9.018291457286433e-05,
-      "loss": 5.1269,
-      "step": 10270
-    },
-    {
-      "epoch": 5.35645371577575,
-      "grad_norm": 1.4320110082626343,
-      "learning_rate": 9.01819095477387e-05,
-      "loss": 5.7046,
-      "step": 10271
-    },
-    {
-      "epoch": 5.3569752281616685,
-      "grad_norm": 1.4466145038604736,
-      "learning_rate": 9.018090452261308e-05,
-      "loss": 5.7731,
-      "step": 10272
-    },
-    {
-      "epoch": 5.357496740547588,
-      "grad_norm": 1.3664839267730713,
-      "learning_rate": 9.017989949748745e-05,
-      "loss": 5.5415,
-      "step": 10273
-    },
-    {
-      "epoch": 5.358018252933507,
-      "grad_norm": 1.432259202003479,
-      "learning_rate": 9.017889447236181e-05,
-      "loss": 6.1035,
-      "step": 10274
-    },
-    {
-      "epoch": 5.358539765319426,
-      "grad_norm": 1.4063067436218262,
-      "learning_rate": 9.017788944723619e-05,
-      "loss": 5.8523,
-      "step": 10275
-    },
-    {
-      "epoch": 5.3590612777053455,
-      "grad_norm": 1.6309561729431152,
-      "learning_rate": 9.017688442211055e-05,
-      "loss": 4.2884,
-      "step": 10276
-    },
-    {
-      "epoch": 5.359582790091265,
-      "grad_norm": 1.4899494647979736,
-      "learning_rate": 9.017587939698493e-05,
-      "loss": 5.7082,
-      "step": 10277
-    },
-    {
-      "epoch": 5.3601043024771835,
-      "grad_norm": 1.4235202074050903,
-      "learning_rate": 9.01748743718593e-05,
-      "loss": 5.8704,
-      "step": 10278
-    },
-    {
-      "epoch": 5.360625814863103,
-      "grad_norm": 1.5385639667510986,
-      "learning_rate": 9.017386934673367e-05,
-      "loss": 5.3477,
-      "step": 10279
-    },
-    {
-      "epoch": 5.361147327249022,
-      "grad_norm": 1.5799310207366943,
-      "learning_rate": 9.017286432160804e-05,
-      "loss": 5.9103,
-      "step": 10280
-    },
-    {
-      "epoch": 5.361668839634941,
-      "grad_norm": 1.677376389503479,
-      "learning_rate": 9.017185929648241e-05,
-      "loss": 5.6047,
-      "step": 10281
-    },
-    {
-      "epoch": 5.3621903520208605,
-      "grad_norm": 1.6116411685943604,
-      "learning_rate": 9.017085427135679e-05,
-      "loss": 4.9714,
-      "step": 10282
-    },
-    {
-      "epoch": 5.36271186440678,
-      "grad_norm": 1.559818148612976,
-      "learning_rate": 9.016984924623117e-05,
-      "loss": 5.093,
-      "step": 10283
-    },
-    {
-      "epoch": 5.3632333767926985,
-      "grad_norm": 1.4089144468307495,
-      "learning_rate": 9.016884422110553e-05,
-      "loss": 5.9716,
-      "step": 10284
-    },
-    {
-      "epoch": 5.363754889178618,
-      "grad_norm": 1.4709949493408203,
-      "learning_rate": 9.016783919597991e-05,
-      "loss": 5.6195,
-      "step": 10285
-    },
-    {
-      "epoch": 5.3642764015645374,
-      "grad_norm": 1.384634256362915,
-      "learning_rate": 9.016683417085428e-05,
-      "loss": 5.6891,
-      "step": 10286
-    },
-    {
-      "epoch": 5.364797913950456,
-      "grad_norm": 1.4020015001296997,
-      "learning_rate": 9.016582914572865e-05,
-      "loss": 4.4311,
-      "step": 10287
-    },
-    {
-      "epoch": 5.3653194263363755,
-      "grad_norm": 1.389190912246704,
-      "learning_rate": 9.016482412060302e-05,
-      "loss": 5.963,
-      "step": 10288
-    },
-    {
-      "epoch": 5.365840938722295,
-      "grad_norm": 1.4642819166183472,
-      "learning_rate": 9.016381909547738e-05,
-      "loss": 5.7096,
-      "step": 10289
-    },
-    {
-      "epoch": 5.3663624511082135,
-      "grad_norm": 1.5249472856521606,
-      "learning_rate": 9.016281407035176e-05,
-      "loss": 5.733,
-      "step": 10290
-    },
-    {
-      "epoch": 5.366883963494133,
-      "grad_norm": 1.4018474817276,
-      "learning_rate": 9.016180904522614e-05,
-      "loss": 5.8569,
-      "step": 10291
-    },
-    {
-      "epoch": 5.3674054758800525,
-      "grad_norm": 1.5382846593856812,
-      "learning_rate": 9.016080402010052e-05,
-      "loss": 5.5011,
-      "step": 10292
-    },
-    {
-      "epoch": 5.367926988265971,
-      "grad_norm": 1.5073137283325195,
-      "learning_rate": 9.015979899497488e-05,
-      "loss": 5.6861,
-      "step": 10293
-    },
-    {
-      "epoch": 5.3684485006518905,
-      "grad_norm": 1.3321858644485474,
-      "learning_rate": 9.015879396984926e-05,
-      "loss": 5.8215,
-      "step": 10294
-    },
-    {
-      "epoch": 5.36897001303781,
-      "grad_norm": 1.55703604221344,
-      "learning_rate": 9.015778894472362e-05,
-      "loss": 5.2027,
-      "step": 10295
-    },
-    {
-      "epoch": 5.3694915254237285,
-      "grad_norm": 1.644277572631836,
-      "learning_rate": 9.0156783919598e-05,
-      "loss": 5.6514,
-      "step": 10296
-    },
-    {
-      "epoch": 5.370013037809648,
-      "grad_norm": 1.5605595111846924,
-      "learning_rate": 9.015577889447236e-05,
-      "loss": 5.7261,
-      "step": 10297
-    },
-    {
-      "epoch": 5.3705345501955675,
-      "grad_norm": 1.5061653852462769,
-      "learning_rate": 9.015477386934674e-05,
-      "loss": 5.7205,
-      "step": 10298
-    },
-    {
-      "epoch": 5.371056062581486,
-      "grad_norm": 1.313819169998169,
-      "learning_rate": 9.01537688442211e-05,
-      "loss": 5.9347,
-      "step": 10299
-    },
-    {
-      "epoch": 5.3715775749674055,
-      "grad_norm": 1.448825716972351,
-      "learning_rate": 9.015276381909548e-05,
-      "loss": 5.5963,
-      "step": 10300
-    },
-    {
-      "epoch": 5.372099087353325,
-      "grad_norm": 1.6235548257827759,
-      "learning_rate": 9.015175879396985e-05,
-      "loss": 5.4979,
-      "step": 10301
-    },
-    {
-      "epoch": 5.3726205997392436,
-      "grad_norm": 1.4101418256759644,
-      "learning_rate": 9.015075376884423e-05,
-      "loss": 5.6974,
-      "step": 10302
-    },
-    {
-      "epoch": 5.373142112125163,
-      "grad_norm": 1.4671521186828613,
-      "learning_rate": 9.01497487437186e-05,
-      "loss": 5.1368,
-      "step": 10303
-    },
-    {
-      "epoch": 5.3736636245110825,
-      "grad_norm": 1.4604076147079468,
-      "learning_rate": 9.014874371859297e-05,
-      "loss": 5.3258,
-      "step": 10304
-    },
-    {
-      "epoch": 5.374185136897001,
-      "grad_norm": 1.5924981832504272,
-      "learning_rate": 9.014773869346735e-05,
-      "loss": 5.6054,
-      "step": 10305
-    },
-    {
-      "epoch": 5.3747066492829205,
-      "grad_norm": 1.3894344568252563,
-      "learning_rate": 9.014673366834171e-05,
-      "loss": 5.7446,
-      "step": 10306
-    },
-    {
-      "epoch": 5.37522816166884,
-      "grad_norm": 1.40617835521698,
-      "learning_rate": 9.014572864321609e-05,
-      "loss": 5.9261,
-      "step": 10307
-    },
-    {
-      "epoch": 5.375749674054759,
-      "grad_norm": 1.6797858476638794,
-      "learning_rate": 9.014472361809045e-05,
-      "loss": 5.0447,
-      "step": 10308
-    },
-    {
-      "epoch": 5.376271186440678,
-      "grad_norm": 1.363195538520813,
-      "learning_rate": 9.014371859296483e-05,
-      "loss": 5.5461,
-      "step": 10309
-    },
-    {
-      "epoch": 5.3767926988265975,
-      "grad_norm": 1.4109382629394531,
-      "learning_rate": 9.01427135678392e-05,
-      "loss": 5.6599,
-      "step": 10310
-    },
-    {
-      "epoch": 5.377314211212516,
-      "grad_norm": 1.3722935914993286,
-      "learning_rate": 9.014170854271357e-05,
-      "loss": 5.95,
-      "step": 10311
-    },
-    {
-      "epoch": 5.3778357235984355,
-      "grad_norm": 1.544797658920288,
-      "learning_rate": 9.014070351758795e-05,
-      "loss": 5.4314,
-      "step": 10312
-    },
-    {
-      "epoch": 5.378357235984355,
-      "grad_norm": 1.5354318618774414,
-      "learning_rate": 9.013969849246233e-05,
-      "loss": 5.6137,
-      "step": 10313
-    },
-    {
-      "epoch": 5.378878748370274,
-      "grad_norm": 1.3543375730514526,
-      "learning_rate": 9.013869346733669e-05,
-      "loss": 5.7397,
-      "step": 10314
-    },
-    {
-      "epoch": 5.379400260756193,
-      "grad_norm": 1.5135337114334106,
-      "learning_rate": 9.013768844221106e-05,
-      "loss": 5.6471,
-      "step": 10315
-    },
-    {
-      "epoch": 5.3799217731421125,
-      "grad_norm": 1.3590775728225708,
-      "learning_rate": 9.013668341708543e-05,
-      "loss": 5.8951,
-      "step": 10316
-    },
-    {
-      "epoch": 5.380443285528031,
-      "grad_norm": 1.4523017406463623,
-      "learning_rate": 9.01356783919598e-05,
-      "loss": 5.0874,
-      "step": 10317
-    },
-    {
-      "epoch": 5.3809647979139505,
-      "grad_norm": 1.4886469841003418,
-      "learning_rate": 9.013467336683418e-05,
-      "loss": 5.597,
-      "step": 10318
-    },
-    {
-      "epoch": 5.38148631029987,
-      "grad_norm": 1.4635052680969238,
-      "learning_rate": 9.013366834170854e-05,
-      "loss": 5.7498,
-      "step": 10319
-    },
-    {
-      "epoch": 5.382007822685789,
-      "grad_norm": 1.4065181016921997,
-      "learning_rate": 9.013266331658292e-05,
-      "loss": 5.8623,
-      "step": 10320
-    },
-    {
-      "epoch": 5.382529335071708,
-      "grad_norm": 1.755972981452942,
-      "learning_rate": 9.013165829145728e-05,
-      "loss": 5.6345,
-      "step": 10321
-    },
-    {
-      "epoch": 5.3830508474576275,
-      "grad_norm": 1.532627820968628,
-      "learning_rate": 9.013065326633166e-05,
-      "loss": 5.9674,
-      "step": 10322
-    },
-    {
-      "epoch": 5.383572359843546,
-      "grad_norm": 1.3523974418640137,
-      "learning_rate": 9.012964824120604e-05,
-      "loss": 5.5732,
-      "step": 10323
-    },
-    {
-      "epoch": 5.384093872229466,
-      "grad_norm": 1.3074290752410889,
-      "learning_rate": 9.012864321608042e-05,
-      "loss": 5.7384,
-      "step": 10324
-    },
-    {
-      "epoch": 5.384615384615385,
-      "grad_norm": 1.3894941806793213,
-      "learning_rate": 9.012763819095478e-05,
-      "loss": 6.211,
-      "step": 10325
-    },
-    {
-      "epoch": 5.385136897001304,
-      "grad_norm": 1.5131900310516357,
-      "learning_rate": 9.012663316582916e-05,
-      "loss": 5.2463,
-      "step": 10326
-    },
-    {
-      "epoch": 5.385658409387223,
-      "grad_norm": 1.3227185010910034,
-      "learning_rate": 9.012562814070352e-05,
-      "loss": 5.946,
-      "step": 10327
-    },
-    {
-      "epoch": 5.386179921773142,
-      "grad_norm": 1.3603779077529907,
-      "learning_rate": 9.012462311557789e-05,
-      "loss": 5.5044,
-      "step": 10328
-    },
-    {
-      "epoch": 5.386701434159061,
-      "grad_norm": 1.5274333953857422,
-      "learning_rate": 9.012361809045226e-05,
-      "loss": 5.7537,
-      "step": 10329
-    },
-    {
-      "epoch": 5.387222946544981,
-      "grad_norm": 1.4906944036483765,
-      "learning_rate": 9.012261306532663e-05,
-      "loss": 5.6782,
-      "step": 10330
-    },
-    {
-      "epoch": 5.3877444589309,
-      "grad_norm": 1.473856806755066,
-      "learning_rate": 9.0121608040201e-05,
-      "loss": 5.7622,
-      "step": 10331
-    },
-    {
-      "epoch": 5.388265971316819,
-      "grad_norm": 1.5477938652038574,
-      "learning_rate": 9.012060301507537e-05,
-      "loss": 5.281,
-      "step": 10332
-    },
-    {
-      "epoch": 5.388787483702738,
-      "grad_norm": 1.6846359968185425,
-      "learning_rate": 9.011959798994975e-05,
-      "loss": 4.9587,
-      "step": 10333
-    },
-    {
-      "epoch": 5.389308996088657,
-      "grad_norm": 1.5388920307159424,
-      "learning_rate": 9.011859296482413e-05,
-      "loss": 6.0003,
-      "step": 10334
-    },
-    {
-      "epoch": 5.389830508474576,
-      "grad_norm": 1.4506977796554565,
-      "learning_rate": 9.01175879396985e-05,
-      "loss": 5.8095,
-      "step": 10335
-    },
-    {
-      "epoch": 5.390352020860496,
-      "grad_norm": 1.391053557395935,
-      "learning_rate": 9.011658291457287e-05,
-      "loss": 6.0712,
-      "step": 10336
-    },
-    {
-      "epoch": 5.390873533246415,
-      "grad_norm": 1.425901174545288,
-      "learning_rate": 9.011557788944725e-05,
-      "loss": 5.5349,
-      "step": 10337
-    },
-    {
-      "epoch": 5.391395045632334,
-      "grad_norm": 1.3986331224441528,
-      "learning_rate": 9.011457286432161e-05,
-      "loss": 5.7131,
-      "step": 10338
-    },
-    {
-      "epoch": 5.391916558018253,
-      "grad_norm": 1.4293824434280396,
-      "learning_rate": 9.011356783919599e-05,
-      "loss": 5.9808,
-      "step": 10339
-    },
-    {
-      "epoch": 5.392438070404172,
-      "grad_norm": 1.7722313404083252,
-      "learning_rate": 9.011256281407035e-05,
-      "loss": 5.4427,
-      "step": 10340
-    },
-    {
-      "epoch": 5.392959582790091,
-      "grad_norm": 1.4018216133117676,
-      "learning_rate": 9.011155778894472e-05,
-      "loss": 5.7794,
-      "step": 10341
-    },
-    {
-      "epoch": 5.393481095176011,
-      "grad_norm": 1.395029067993164,
-      "learning_rate": 9.01105527638191e-05,
-      "loss": 5.6949,
-      "step": 10342
-    },
-    {
-      "epoch": 5.39400260756193,
-      "grad_norm": 1.4609378576278687,
-      "learning_rate": 9.010954773869347e-05,
-      "loss": 5.4976,
-      "step": 10343
-    },
-    {
-      "epoch": 5.394524119947849,
-      "grad_norm": 1.47409188747406,
-      "learning_rate": 9.010854271356785e-05,
-      "loss": 5.5711,
-      "step": 10344
-    },
-    {
-      "epoch": 5.395045632333768,
-      "grad_norm": 1.441180944442749,
-      "learning_rate": 9.010753768844222e-05,
-      "loss": 5.9536,
-      "step": 10345
-    },
-    {
-      "epoch": 5.395567144719687,
-      "grad_norm": 1.5181047916412354,
-      "learning_rate": 9.010653266331659e-05,
-      "loss": 5.8325,
-      "step": 10346
-    },
-    {
-      "epoch": 5.396088657105606,
-      "grad_norm": 1.6373478174209595,
-      "learning_rate": 9.010552763819096e-05,
-      "loss": 5.5028,
-      "step": 10347
-    },
-    {
-      "epoch": 5.396610169491526,
-      "grad_norm": 1.4294440746307373,
-      "learning_rate": 9.010452261306533e-05,
-      "loss": 5.6257,
-      "step": 10348
-    },
-    {
-      "epoch": 5.397131681877444,
-      "grad_norm": 1.5571463108062744,
-      "learning_rate": 9.01035175879397e-05,
-      "loss": 5.0681,
-      "step": 10349
-    },
-    {
-      "epoch": 5.397653194263364,
-      "grad_norm": 1.546956181526184,
-      "learning_rate": 9.010251256281408e-05,
-      "loss": 5.8283,
-      "step": 10350
-    },
-    {
-      "epoch": 5.398174706649283,
-      "grad_norm": 1.402212142944336,
-      "learning_rate": 9.010150753768844e-05,
-      "loss": 5.3116,
-      "step": 10351
-    },
-    {
-      "epoch": 5.398696219035202,
-      "grad_norm": 1.5935919284820557,
-      "learning_rate": 9.010050251256282e-05,
-      "loss": 5.2964,
-      "step": 10352
-    },
-    {
-      "epoch": 5.399217731421121,
-      "grad_norm": 1.4337387084960938,
-      "learning_rate": 9.009949748743718e-05,
-      "loss": 5.4802,
-      "step": 10353
-    },
-    {
-      "epoch": 5.399739243807041,
-      "grad_norm": 1.3245999813079834,
-      "learning_rate": 9.009849246231156e-05,
-      "loss": 5.709,
-      "step": 10354
-    },
-    {
-      "epoch": 5.400260756192959,
-      "grad_norm": 1.9014265537261963,
-      "learning_rate": 9.009748743718594e-05,
-      "loss": 5.4192,
-      "step": 10355
-    },
-    {
-      "epoch": 5.400782268578879,
-      "grad_norm": 1.7964805364608765,
-      "learning_rate": 9.00964824120603e-05,
-      "loss": 5.5363,
-      "step": 10356
-    },
-    {
-      "epoch": 5.401303780964798,
-      "grad_norm": 1.516506314277649,
-      "learning_rate": 9.009547738693468e-05,
-      "loss": 4.9884,
-      "step": 10357
-    },
-    {
-      "epoch": 5.401825293350717,
-      "grad_norm": 1.4831973314285278,
-      "learning_rate": 9.009447236180905e-05,
-      "loss": 5.979,
-      "step": 10358
-    },
-    {
-      "epoch": 5.402346805736636,
-      "grad_norm": 1.4365599155426025,
-      "learning_rate": 9.009346733668342e-05,
-      "loss": 5.5484,
-      "step": 10359
-    },
-    {
-      "epoch": 5.402868318122556,
-      "grad_norm": 1.5162920951843262,
-      "learning_rate": 9.009246231155779e-05,
-      "loss": 5.5688,
-      "step": 10360
-    },
-    {
-      "epoch": 5.403389830508474,
-      "grad_norm": 1.4060022830963135,
-      "learning_rate": 9.009145728643217e-05,
-      "loss": 5.4415,
-      "step": 10361
-    },
-    {
-      "epoch": 5.403911342894394,
-      "grad_norm": 1.5157737731933594,
-      "learning_rate": 9.009045226130653e-05,
-      "loss": 5.4192,
-      "step": 10362
-    },
-    {
-      "epoch": 5.404432855280313,
-      "grad_norm": 1.3115202188491821,
-      "learning_rate": 9.008944723618091e-05,
-      "loss": 6.0647,
-      "step": 10363
-    },
-    {
-      "epoch": 5.404954367666232,
-      "grad_norm": 1.4188824892044067,
-      "learning_rate": 9.008844221105529e-05,
-      "loss": 5.6286,
-      "step": 10364
-    },
-    {
-      "epoch": 5.405475880052151,
-      "grad_norm": 1.4039252996444702,
-      "learning_rate": 9.008743718592966e-05,
-      "loss": 5.7515,
-      "step": 10365
-    },
-    {
-      "epoch": 5.405997392438071,
-      "grad_norm": 1.4414304494857788,
-      "learning_rate": 9.008643216080403e-05,
-      "loss": 6.0306,
-      "step": 10366
-    },
-    {
-      "epoch": 5.406518904823989,
-      "grad_norm": 1.345015287399292,
-      "learning_rate": 9.008542713567839e-05,
-      "loss": 6.0037,
-      "step": 10367
-    },
-    {
-      "epoch": 5.407040417209909,
-      "grad_norm": 1.5031777620315552,
-      "learning_rate": 9.008442211055277e-05,
-      "loss": 5.6039,
-      "step": 10368
-    },
-    {
-      "epoch": 5.407561929595828,
-      "grad_norm": 1.443379521369934,
-      "learning_rate": 9.008341708542713e-05,
-      "loss": 5.6645,
-      "step": 10369
-    },
-    {
-      "epoch": 5.408083441981747,
-      "grad_norm": 1.408402442932129,
-      "learning_rate": 9.008241206030151e-05,
-      "loss": 5.8753,
-      "step": 10370
-    },
-    {
-      "epoch": 5.408604954367666,
-      "grad_norm": 1.420959711074829,
-      "learning_rate": 9.008140703517588e-05,
-      "loss": 5.7371,
-      "step": 10371
-    },
-    {
-      "epoch": 5.409126466753586,
-      "grad_norm": 1.4725791215896606,
-      "learning_rate": 9.008040201005025e-05,
-      "loss": 5.6141,
-      "step": 10372
-    },
-    {
-      "epoch": 5.409647979139504,
-      "grad_norm": 1.3845386505126953,
-      "learning_rate": 9.007939698492462e-05,
-      "loss": 5.6863,
-      "step": 10373
-    },
-    {
-      "epoch": 5.410169491525424,
-      "grad_norm": 1.3944177627563477,
-      "learning_rate": 9.0078391959799e-05,
-      "loss": 5.6025,
-      "step": 10374
-    },
-    {
-      "epoch": 5.410691003911343,
-      "grad_norm": 1.349969744682312,
-      "learning_rate": 9.007738693467337e-05,
-      "loss": 5.8363,
-      "step": 10375
-    },
-    {
-      "epoch": 5.411212516297262,
-      "grad_norm": 1.4599504470825195,
-      "learning_rate": 9.007638190954775e-05,
-      "loss": 5.7438,
-      "step": 10376
-    },
-    {
-      "epoch": 5.411734028683181,
-      "grad_norm": 1.8842430114746094,
-      "learning_rate": 9.007537688442212e-05,
-      "loss": 5.3469,
-      "step": 10377
-    },
-    {
-      "epoch": 5.412255541069101,
-      "grad_norm": 1.8325340747833252,
-      "learning_rate": 9.00743718592965e-05,
-      "loss": 5.3805,
-      "step": 10378
-    },
-    {
-      "epoch": 5.412777053455019,
-      "grad_norm": 1.5064162015914917,
-      "learning_rate": 9.007336683417086e-05,
-      "loss": 5.5533,
-      "step": 10379
-    },
-    {
-      "epoch": 5.413298565840939,
-      "grad_norm": 1.4795126914978027,
-      "learning_rate": 9.007236180904524e-05,
-      "loss": 5.8021,
-      "step": 10380
-    },
-    {
-      "epoch": 5.413820078226858,
-      "grad_norm": 1.48528254032135,
-      "learning_rate": 9.00713567839196e-05,
-      "loss": 5.5695,
-      "step": 10381
-    },
-    {
-      "epoch": 5.414341590612777,
-      "grad_norm": 1.524933099746704,
-      "learning_rate": 9.007035175879396e-05,
-      "loss": 5.5983,
-      "step": 10382
-    },
-    {
-      "epoch": 5.414863102998696,
-      "grad_norm": 1.5218279361724854,
-      "learning_rate": 9.006934673366834e-05,
-      "loss": 5.5254,
-      "step": 10383
-    },
-    {
-      "epoch": 5.415384615384616,
-      "grad_norm": 1.6243934631347656,
-      "learning_rate": 9.006834170854272e-05,
-      "loss": 5.4019,
-      "step": 10384
-    },
-    {
-      "epoch": 5.415906127770534,
-      "grad_norm": 1.4459525346755981,
-      "learning_rate": 9.00673366834171e-05,
-      "loss": 5.9141,
-      "step": 10385
-    },
-    {
-      "epoch": 5.416427640156454,
-      "grad_norm": 1.442171573638916,
-      "learning_rate": 9.006633165829146e-05,
-      "loss": 5.6479,
-      "step": 10386
-    },
-    {
-      "epoch": 5.416949152542373,
-      "grad_norm": 1.3921960592269897,
-      "learning_rate": 9.006532663316584e-05,
-      "loss": 6.0575,
-      "step": 10387
-    },
-    {
-      "epoch": 5.417470664928292,
-      "grad_norm": 1.4171091318130493,
-      "learning_rate": 9.00643216080402e-05,
-      "loss": 5.7419,
-      "step": 10388
-    },
-    {
-      "epoch": 5.417992177314211,
-      "grad_norm": 1.5791563987731934,
-      "learning_rate": 9.006331658291458e-05,
-      "loss": 4.8853,
-      "step": 10389
-    },
-    {
-      "epoch": 5.418513689700131,
-      "grad_norm": 1.4508466720581055,
-      "learning_rate": 9.006231155778895e-05,
-      "loss": 5.8386,
-      "step": 10390
-    },
-    {
-      "epoch": 5.419035202086049,
-      "grad_norm": 1.5315054655075073,
-      "learning_rate": 9.006130653266332e-05,
-      "loss": 5.0505,
-      "step": 10391
-    },
-    {
-      "epoch": 5.419556714471969,
-      "grad_norm": 1.4277361631393433,
-      "learning_rate": 9.006030150753769e-05,
-      "loss": 5.2858,
-      "step": 10392
-    },
-    {
-      "epoch": 5.420078226857888,
-      "grad_norm": 1.3090296983718872,
-      "learning_rate": 9.005929648241207e-05,
-      "loss": 5.8854,
-      "step": 10393
-    },
-    {
-      "epoch": 5.420599739243807,
-      "grad_norm": 1.3864718675613403,
-      "learning_rate": 9.005829145728643e-05,
-      "loss": 5.7432,
-      "step": 10394
-    },
-    {
-      "epoch": 5.421121251629726,
-      "grad_norm": 1.472598671913147,
-      "learning_rate": 9.005728643216081e-05,
-      "loss": 5.8862,
-      "step": 10395
-    },
-    {
-      "epoch": 5.421642764015646,
-      "grad_norm": 1.3983243703842163,
-      "learning_rate": 9.005628140703519e-05,
-      "loss": 5.0842,
-      "step": 10396
-    },
-    {
-      "epoch": 5.422164276401564,
-      "grad_norm": 1.3727399110794067,
-      "learning_rate": 9.005527638190955e-05,
-      "loss": 5.7735,
-      "step": 10397
-    },
-    {
-      "epoch": 5.422685788787484,
-      "grad_norm": 1.494642972946167,
-      "learning_rate": 9.005427135678393e-05,
-      "loss": 5.3708,
-      "step": 10398
-    },
-    {
-      "epoch": 5.423207301173403,
-      "grad_norm": 1.4355601072311401,
-      "learning_rate": 9.005326633165829e-05,
-      "loss": 5.5744,
-      "step": 10399
-    },
-    {
-      "epoch": 5.423728813559322,
-      "grad_norm": 1.460964560508728,
-      "learning_rate": 9.005226130653267e-05,
-      "loss": 5.8205,
-      "step": 10400
-    },
-    {
-      "epoch": 5.424250325945241,
-      "grad_norm": 1.5543617010116577,
-      "learning_rate": 9.005125628140703e-05,
-      "loss": 5.3298,
-      "step": 10401
-    },
-    {
-      "epoch": 5.424771838331161,
-      "grad_norm": 1.6537758111953735,
-      "learning_rate": 9.005025125628141e-05,
-      "loss": 5.3461,
-      "step": 10402
-    },
-    {
-      "epoch": 5.425293350717079,
-      "grad_norm": 1.6786447763442993,
-      "learning_rate": 9.004924623115578e-05,
-      "loss": 5.7416,
-      "step": 10403
-    },
-    {
-      "epoch": 5.425814863102999,
-      "grad_norm": 1.506950855255127,
-      "learning_rate": 9.004824120603015e-05,
-      "loss": 5.6922,
-      "step": 10404
-    },
-    {
-      "epoch": 5.426336375488918,
-      "grad_norm": 1.486511468887329,
-      "learning_rate": 9.004723618090453e-05,
-      "loss": 5.9835,
-      "step": 10405
-    },
-    {
-      "epoch": 5.426857887874837,
-      "grad_norm": 1.4177052974700928,
-      "learning_rate": 9.004623115577891e-05,
-      "loss": 5.6917,
-      "step": 10406
-    },
-    {
-      "epoch": 5.427379400260756,
-      "grad_norm": 1.8048750162124634,
-      "learning_rate": 9.004522613065327e-05,
-      "loss": 5.605,
-      "step": 10407
-    },
-    {
-      "epoch": 5.427900912646676,
-      "grad_norm": 1.6026074886322021,
-      "learning_rate": 9.004422110552764e-05,
-      "loss": 5.6529,
-      "step": 10408
-    },
-    {
-      "epoch": 5.428422425032594,
-      "grad_norm": 1.5728422403335571,
-      "learning_rate": 9.004321608040202e-05,
-      "loss": 5.3737,
-      "step": 10409
-    },
-    {
-      "epoch": 5.428943937418514,
-      "grad_norm": 1.5810431241989136,
-      "learning_rate": 9.004221105527638e-05,
-      "loss": 5.3264,
-      "step": 10410
-    },
-    {
-      "epoch": 5.429465449804433,
-      "grad_norm": 1.4311530590057373,
-      "learning_rate": 9.004120603015076e-05,
-      "loss": 5.6839,
-      "step": 10411
-    },
-    {
-      "epoch": 5.429986962190352,
-      "grad_norm": 1.4722007513046265,
-      "learning_rate": 9.004020100502512e-05,
-      "loss": 5.7458,
-      "step": 10412
-    },
-    {
-      "epoch": 5.430508474576271,
-      "grad_norm": 1.6334959268569946,
-      "learning_rate": 9.00391959798995e-05,
-      "loss": 5.4403,
-      "step": 10413
-    },
-    {
-      "epoch": 5.431029986962191,
-      "grad_norm": 1.2720927000045776,
-      "learning_rate": 9.003819095477387e-05,
-      "loss": 5.9236,
-      "step": 10414
-    },
-    {
-      "epoch": 5.431551499348109,
-      "grad_norm": 1.6285679340362549,
-      "learning_rate": 9.003718592964824e-05,
-      "loss": 5.1026,
-      "step": 10415
-    },
-    {
-      "epoch": 5.432073011734029,
-      "grad_norm": 1.603682041168213,
-      "learning_rate": 9.003618090452262e-05,
-      "loss": 5.4022,
-      "step": 10416
-    },
-    {
-      "epoch": 5.432594524119948,
-      "grad_norm": 1.5664619207382202,
-      "learning_rate": 9.0035175879397e-05,
-      "loss": 5.3534,
-      "step": 10417
-    },
-    {
-      "epoch": 5.433116036505867,
-      "grad_norm": 1.4744374752044678,
-      "learning_rate": 9.003417085427136e-05,
-      "loss": 5.4306,
-      "step": 10418
-    },
-    {
-      "epoch": 5.433637548891786,
-      "grad_norm": 1.4028116464614868,
-      "learning_rate": 9.003316582914574e-05,
-      "loss": 6.0245,
-      "step": 10419
-    },
-    {
-      "epoch": 5.434159061277706,
-      "grad_norm": 1.508144497871399,
-      "learning_rate": 9.00321608040201e-05,
-      "loss": 5.9922,
-      "step": 10420
-    },
-    {
-      "epoch": 5.434680573663624,
-      "grad_norm": 1.356785535812378,
-      "learning_rate": 9.003115577889447e-05,
-      "loss": 5.9281,
-      "step": 10421
-    },
-    {
-      "epoch": 5.435202086049544,
-      "grad_norm": 1.443791389465332,
-      "learning_rate": 9.003015075376885e-05,
-      "loss": 5.7952,
-      "step": 10422
-    },
-    {
-      "epoch": 5.435723598435462,
-      "grad_norm": 1.430184006690979,
-      "learning_rate": 9.002914572864321e-05,
-      "loss": 5.5113,
-      "step": 10423
-    },
-    {
-      "epoch": 5.436245110821382,
-      "grad_norm": 1.3964896202087402,
-      "learning_rate": 9.002814070351759e-05,
-      "loss": 5.8624,
-      "step": 10424
-    },
-    {
-      "epoch": 5.436766623207301,
-      "grad_norm": 1.4694124460220337,
-      "learning_rate": 9.002713567839197e-05,
-      "loss": 5.7595,
-      "step": 10425
-    },
-    {
-      "epoch": 5.437288135593221,
-      "grad_norm": 1.5289971828460693,
-      "learning_rate": 9.002613065326634e-05,
-      "loss": 5.6549,
-      "step": 10426
-    },
-    {
-      "epoch": 5.437809647979139,
-      "grad_norm": 1.3552371263504028,
-      "learning_rate": 9.002512562814071e-05,
-      "loss": 5.6389,
-      "step": 10427
-    },
-    {
-      "epoch": 5.438331160365059,
-      "grad_norm": 1.3424198627471924,
-      "learning_rate": 9.002412060301509e-05,
-      "loss": 5.9506,
-      "step": 10428
-    },
-    {
-      "epoch": 5.438852672750977,
-      "grad_norm": 1.3919686079025269,
-      "learning_rate": 9.002311557788945e-05,
-      "loss": 5.1115,
-      "step": 10429
-    },
-    {
-      "epoch": 5.439374185136897,
-      "grad_norm": 1.4892040491104126,
-      "learning_rate": 9.002211055276383e-05,
-      "loss": 5.6385,
-      "step": 10430
-    },
-    {
-      "epoch": 5.439895697522816,
-      "grad_norm": 1.4934167861938477,
-      "learning_rate": 9.00211055276382e-05,
-      "loss": 5.581,
-      "step": 10431
-    },
-    {
-      "epoch": 5.440417209908736,
-      "grad_norm": 1.3194286823272705,
-      "learning_rate": 9.002010050251257e-05,
-      "loss": 5.5712,
-      "step": 10432
-    },
-    {
-      "epoch": 5.440938722294654,
-      "grad_norm": 1.5284663438796997,
-      "learning_rate": 9.001909547738694e-05,
-      "loss": 4.9961,
-      "step": 10433
-    },
-    {
-      "epoch": 5.441460234680574,
-      "grad_norm": 1.4860661029815674,
-      "learning_rate": 9.00180904522613e-05,
-      "loss": 5.5736,
-      "step": 10434
-    },
-    {
-      "epoch": 5.441981747066492,
-      "grad_norm": 1.4548406600952148,
-      "learning_rate": 9.001708542713568e-05,
-      "loss": 5.8606,
-      "step": 10435
-    },
-    {
-      "epoch": 5.442503259452412,
-      "grad_norm": 1.420127272605896,
-      "learning_rate": 9.001608040201006e-05,
-      "loss": 6.01,
-      "step": 10436
-    },
-    {
-      "epoch": 5.443024771838331,
-      "grad_norm": 1.5450996160507202,
-      "learning_rate": 9.001507537688443e-05,
-      "loss": 5.6367,
-      "step": 10437
-    },
-    {
-      "epoch": 5.443546284224251,
-      "grad_norm": 1.3860366344451904,
-      "learning_rate": 9.00140703517588e-05,
-      "loss": 5.9346,
-      "step": 10438
-    },
-    {
-      "epoch": 5.444067796610169,
-      "grad_norm": 1.525604486465454,
-      "learning_rate": 9.001306532663318e-05,
-      "loss": 5.5312,
-      "step": 10439
-    },
-    {
-      "epoch": 5.444589308996089,
-      "grad_norm": 1.4250444173812866,
-      "learning_rate": 9.001206030150754e-05,
-      "loss": 5.3246,
-      "step": 10440
-    },
-    {
-      "epoch": 5.445110821382007,
-      "grad_norm": 1.4571789503097534,
-      "learning_rate": 9.001105527638192e-05,
-      "loss": 5.5194,
-      "step": 10441
-    },
-    {
-      "epoch": 5.445632333767927,
-      "grad_norm": 1.4137465953826904,
-      "learning_rate": 9.001005025125628e-05,
-      "loss": 5.726,
-      "step": 10442
-    },
-    {
-      "epoch": 5.446153846153846,
-      "grad_norm": 1.3715574741363525,
-      "learning_rate": 9.000904522613066e-05,
-      "loss": 5.3377,
-      "step": 10443
-    },
-    {
-      "epoch": 5.446675358539765,
-      "grad_norm": 1.4113187789916992,
-      "learning_rate": 9.000804020100502e-05,
-      "loss": 5.5306,
-      "step": 10444
-    },
-    {
-      "epoch": 5.447196870925684,
-      "grad_norm": 1.4104368686676025,
-      "learning_rate": 9.00070351758794e-05,
-      "loss": 5.8549,
-      "step": 10445
-    },
-    {
-      "epoch": 5.447718383311604,
-      "grad_norm": 1.5098892450332642,
-      "learning_rate": 9.000603015075378e-05,
-      "loss": 5.3358,
-      "step": 10446
-    },
-    {
-      "epoch": 5.448239895697522,
-      "grad_norm": 1.4788087606430054,
-      "learning_rate": 9.000502512562814e-05,
-      "loss": 5.8211,
-      "step": 10447
-    },
-    {
-      "epoch": 5.448761408083442,
-      "grad_norm": 1.470460057258606,
-      "learning_rate": 9.000402010050252e-05,
-      "loss": 5.5852,
-      "step": 10448
-    },
-    {
-      "epoch": 5.449282920469361,
-      "grad_norm": 1.3504749536514282,
-      "learning_rate": 9.000301507537689e-05,
-      "loss": 5.8415,
-      "step": 10449
-    },
-    {
-      "epoch": 5.44980443285528,
-      "grad_norm": 1.4444488286972046,
-      "learning_rate": 9.000201005025126e-05,
-      "loss": 5.629,
-      "step": 10450
-    },
-    {
-      "epoch": 5.450325945241199,
-      "grad_norm": 1.4733641147613525,
-      "learning_rate": 9.000100502512563e-05,
-      "loss": 5.8316,
-      "step": 10451
-    },
-    {
-      "epoch": 5.450847457627119,
-      "grad_norm": 1.4013129472732544,
-      "learning_rate": 9e-05,
-      "loss": 5.2798,
-      "step": 10452
-    },
-    {
-      "epoch": 5.451368970013037,
-      "grad_norm": 1.5350227355957031,
-      "learning_rate": 8.999899497487437e-05,
-      "loss": 5.8352,
-      "step": 10453
-    },
-    {
-      "epoch": 5.451890482398957,
-      "grad_norm": 1.4888927936553955,
-      "learning_rate": 8.999798994974875e-05,
-      "loss": 5.8298,
-      "step": 10454
-    },
-    {
-      "epoch": 5.452411994784876,
-      "grad_norm": 1.5187705755233765,
-      "learning_rate": 8.999698492462311e-05,
-      "loss": 5.5905,
-      "step": 10455
-    },
-    {
-      "epoch": 5.452933507170795,
-      "grad_norm": 1.486060380935669,
-      "learning_rate": 8.999597989949749e-05,
-      "loss": 5.7646,
-      "step": 10456
-    },
-    {
-      "epoch": 5.453455019556714,
-      "grad_norm": 1.5799533128738403,
-      "learning_rate": 8.999497487437187e-05,
-      "loss": 5.3956,
-      "step": 10457
-    },
-    {
-      "epoch": 5.453976531942634,
-      "grad_norm": 1.5062335729599,
-      "learning_rate": 8.999396984924625e-05,
-      "loss": 5.7285,
-      "step": 10458
-    },
-    {
-      "epoch": 5.4544980443285525,
-      "grad_norm": 1.3899563550949097,
-      "learning_rate": 8.999296482412061e-05,
-      "loss": 5.6554,
-      "step": 10459
-    },
-    {
-      "epoch": 5.455019556714472,
-      "grad_norm": 1.5313035249710083,
-      "learning_rate": 8.999195979899497e-05,
-      "loss": 5.0297,
-      "step": 10460
-    },
-    {
-      "epoch": 5.455541069100391,
-      "grad_norm": 1.498185157775879,
-      "learning_rate": 8.999095477386935e-05,
-      "loss": 5.9457,
-      "step": 10461
-    },
-    {
-      "epoch": 5.45606258148631,
-      "grad_norm": 1.371760606765747,
-      "learning_rate": 8.998994974874372e-05,
-      "loss": 5.8825,
-      "step": 10462
-    },
-    {
-      "epoch": 5.456584093872229,
-      "grad_norm": 1.4524531364440918,
-      "learning_rate": 8.99889447236181e-05,
-      "loss": 5.3893,
-      "step": 10463
-    },
-    {
-      "epoch": 5.457105606258149,
-      "grad_norm": 1.5481804609298706,
-      "learning_rate": 8.998793969849246e-05,
-      "loss": 5.2363,
-      "step": 10464
-    },
-    {
-      "epoch": 5.4576271186440675,
-      "grad_norm": 1.421221375465393,
-      "learning_rate": 8.998693467336684e-05,
-      "loss": 5.6129,
-      "step": 10465
-    },
-    {
-      "epoch": 5.458148631029987,
-      "grad_norm": 1.669452428817749,
-      "learning_rate": 8.998592964824121e-05,
-      "loss": 4.9568,
-      "step": 10466
-    },
-    {
-      "epoch": 5.458670143415906,
-      "grad_norm": 1.4031566381454468,
-      "learning_rate": 8.998492462311559e-05,
-      "loss": 5.6301,
-      "step": 10467
-    },
-    {
-      "epoch": 5.459191655801825,
-      "grad_norm": 1.5026051998138428,
-      "learning_rate": 8.998391959798996e-05,
-      "loss": 5.4623,
-      "step": 10468
-    },
-    {
-      "epoch": 5.459713168187744,
-      "grad_norm": 1.549522876739502,
-      "learning_rate": 8.998291457286433e-05,
-      "loss": 5.6162,
-      "step": 10469
-    },
-    {
-      "epoch": 5.460234680573664,
-      "grad_norm": 1.5481055974960327,
-      "learning_rate": 8.99819095477387e-05,
-      "loss": 5.5638,
-      "step": 10470
-    },
-    {
-      "epoch": 5.4607561929595825,
-      "grad_norm": 1.3635905981063843,
-      "learning_rate": 8.998090452261308e-05,
-      "loss": 5.6781,
-      "step": 10471
-    },
-    {
-      "epoch": 5.461277705345502,
-      "grad_norm": 1.4630109071731567,
-      "learning_rate": 8.997989949748744e-05,
-      "loss": 5.7109,
-      "step": 10472
-    },
-    {
-      "epoch": 5.461799217731421,
-      "grad_norm": 1.8126877546310425,
-      "learning_rate": 8.997889447236182e-05,
-      "loss": 5.443,
-      "step": 10473
-    },
-    {
-      "epoch": 5.46232073011734,
-      "grad_norm": 1.3666918277740479,
-      "learning_rate": 8.997788944723618e-05,
-      "loss": 5.8661,
-      "step": 10474
-    },
-    {
-      "epoch": 5.4628422425032594,
-      "grad_norm": 1.5255168676376343,
-      "learning_rate": 8.997688442211055e-05,
-      "loss": 5.4004,
-      "step": 10475
-    },
-    {
-      "epoch": 5.463363754889179,
-      "grad_norm": 1.488571286201477,
-      "learning_rate": 8.997587939698492e-05,
-      "loss": 5.7884,
-      "step": 10476
-    },
-    {
-      "epoch": 5.4638852672750975,
-      "grad_norm": 1.5981428623199463,
-      "learning_rate": 8.99748743718593e-05,
-      "loss": 5.505,
-      "step": 10477
-    },
-    {
-      "epoch": 5.464406779661017,
-      "grad_norm": 2.0508005619049072,
-      "learning_rate": 8.997386934673368e-05,
-      "loss": 5.2415,
-      "step": 10478
-    },
-    {
-      "epoch": 5.464928292046936,
-      "grad_norm": 1.5541373491287231,
-      "learning_rate": 8.997286432160804e-05,
-      "loss": 5.6101,
-      "step": 10479
-    },
-    {
-      "epoch": 5.465449804432855,
-      "grad_norm": 1.4701709747314453,
-      "learning_rate": 8.997185929648242e-05,
-      "loss": 6.086,
-      "step": 10480
-    },
-    {
-      "epoch": 5.4659713168187745,
-      "grad_norm": 1.4314124584197998,
-      "learning_rate": 8.997085427135679e-05,
-      "loss": 5.6981,
-      "step": 10481
-    },
-    {
-      "epoch": 5.466492829204694,
-      "grad_norm": 1.411684513092041,
-      "learning_rate": 8.996984924623116e-05,
-      "loss": 5.5636,
-      "step": 10482
-    },
-    {
-      "epoch": 5.4670143415906125,
-      "grad_norm": 1.3733701705932617,
-      "learning_rate": 8.996884422110553e-05,
-      "loss": 5.7607,
-      "step": 10483
-    },
-    {
-      "epoch": 5.467535853976532,
-      "grad_norm": 1.463255763053894,
-      "learning_rate": 8.99678391959799e-05,
-      "loss": 5.7746,
-      "step": 10484
-    },
-    {
-      "epoch": 5.468057366362451,
-      "grad_norm": 2.0161988735198975,
-      "learning_rate": 8.996683417085427e-05,
-      "loss": 5.6159,
-      "step": 10485
-    },
-    {
-      "epoch": 5.46857887874837,
-      "grad_norm": 1.43341863155365,
-      "learning_rate": 8.996582914572865e-05,
-      "loss": 5.5757,
-      "step": 10486
-    },
-    {
-      "epoch": 5.4691003911342895,
-      "grad_norm": 1.4315468072891235,
-      "learning_rate": 8.996482412060301e-05,
-      "loss": 5.6547,
-      "step": 10487
-    },
-    {
-      "epoch": 5.469621903520209,
-      "grad_norm": 1.3538097143173218,
-      "learning_rate": 8.996381909547739e-05,
-      "loss": 5.5146,
-      "step": 10488
-    },
-    {
-      "epoch": 5.4701434159061275,
-      "grad_norm": 1.4585342407226562,
-      "learning_rate": 8.996281407035177e-05,
-      "loss": 5.5846,
-      "step": 10489
-    },
-    {
-      "epoch": 5.470664928292047,
-      "grad_norm": 1.448188304901123,
-      "learning_rate": 8.996180904522613e-05,
-      "loss": 5.5696,
-      "step": 10490
-    },
-    {
-      "epoch": 5.4711864406779664,
-      "grad_norm": 1.3081797361373901,
-      "learning_rate": 8.996080402010051e-05,
-      "loss": 5.7939,
-      "step": 10491
-    },
-    {
-      "epoch": 5.471707953063885,
-      "grad_norm": 1.4029816389083862,
-      "learning_rate": 8.995979899497487e-05,
-      "loss": 5.978,
-      "step": 10492
-    },
-    {
-      "epoch": 5.4722294654498045,
-      "grad_norm": 1.3842742443084717,
-      "learning_rate": 8.995879396984925e-05,
-      "loss": 5.7605,
-      "step": 10493
-    },
-    {
-      "epoch": 5.472750977835724,
-      "grad_norm": 1.3958038091659546,
-      "learning_rate": 8.995778894472362e-05,
-      "loss": 5.9076,
-      "step": 10494
-    },
-    {
-      "epoch": 5.4732724902216425,
-      "grad_norm": 1.5873847007751465,
-      "learning_rate": 8.9956783919598e-05,
-      "loss": 5.4344,
-      "step": 10495
-    },
-    {
-      "epoch": 5.473794002607562,
-      "grad_norm": 1.4464925527572632,
-      "learning_rate": 8.995577889447236e-05,
-      "loss": 5.7357,
-      "step": 10496
-    },
-    {
-      "epoch": 5.4743155149934815,
-      "grad_norm": 1.6261836290359497,
-      "learning_rate": 8.995477386934674e-05,
-      "loss": 5.1764,
-      "step": 10497
-    },
-    {
-      "epoch": 5.4748370273794,
-      "grad_norm": 1.523697018623352,
-      "learning_rate": 8.995376884422111e-05,
-      "loss": 5.1895,
-      "step": 10498
-    },
-    {
-      "epoch": 5.4753585397653195,
-      "grad_norm": 1.4699403047561646,
-      "learning_rate": 8.995276381909549e-05,
-      "loss": 5.2165,
-      "step": 10499
-    },
-    {
-      "epoch": 5.475880052151239,
-      "grad_norm": 1.6217836141586304,
-      "learning_rate": 8.995175879396986e-05,
-      "loss": 5.5246,
-      "step": 10500
-    },
-    {
-      "epoch": 5.4764015645371575,
-      "grad_norm": 1.571385145187378,
-      "learning_rate": 8.995075376884422e-05,
-      "loss": 5.4143,
-      "step": 10501
-    },
-    {
-      "epoch": 5.476923076923077,
-      "grad_norm": 1.463789939880371,
-      "learning_rate": 8.99497487437186e-05,
-      "loss": 5.6644,
-      "step": 10502
-    },
-    {
-      "epoch": 5.4774445893089965,
-      "grad_norm": 1.4198336601257324,
-      "learning_rate": 8.994874371859296e-05,
-      "loss": 5.5575,
-      "step": 10503
-    },
-    {
-      "epoch": 5.477966101694915,
-      "grad_norm": 1.4517531394958496,
-      "learning_rate": 8.994773869346734e-05,
-      "loss": 5.8132,
-      "step": 10504
-    },
-    {
-      "epoch": 5.4784876140808345,
-      "grad_norm": 1.3597767353057861,
-      "learning_rate": 8.99467336683417e-05,
-      "loss": 6.0039,
-      "step": 10505
-    },
-    {
-      "epoch": 5.479009126466754,
-      "grad_norm": 1.5586787462234497,
-      "learning_rate": 8.994572864321608e-05,
-      "loss": 5.6402,
-      "step": 10506
-    },
-    {
-      "epoch": 5.4795306388526726,
-      "grad_norm": 1.3150954246520996,
-      "learning_rate": 8.994472361809045e-05,
-      "loss": 5.6866,
-      "step": 10507
-    },
-    {
-      "epoch": 5.480052151238592,
-      "grad_norm": 1.4818559885025024,
-      "learning_rate": 8.994371859296483e-05,
-      "loss": 5.6192,
-      "step": 10508
-    },
-    {
-      "epoch": 5.4805736636245115,
-      "grad_norm": 1.442344307899475,
-      "learning_rate": 8.99427135678392e-05,
-      "loss": 6.0299,
-      "step": 10509
-    },
-    {
-      "epoch": 5.48109517601043,
-      "grad_norm": 1.401706576347351,
-      "learning_rate": 8.994170854271358e-05,
-      "loss": 5.6443,
-      "step": 10510
-    },
-    {
-      "epoch": 5.4816166883963495,
-      "grad_norm": 1.3801261186599731,
-      "learning_rate": 8.994070351758795e-05,
-      "loss": 5.7512,
-      "step": 10511
-    },
-    {
-      "epoch": 5.482138200782269,
-      "grad_norm": 1.5764576196670532,
-      "learning_rate": 8.993969849246232e-05,
-      "loss": 5.3058,
-      "step": 10512
-    },
-    {
-      "epoch": 5.482659713168188,
-      "grad_norm": 1.63459312915802,
-      "learning_rate": 8.993869346733669e-05,
-      "loss": 5.0983,
-      "step": 10513
-    },
-    {
-      "epoch": 5.483181225554107,
-      "grad_norm": 1.4830843210220337,
-      "learning_rate": 8.993768844221105e-05,
-      "loss": 5.9483,
-      "step": 10514
-    },
-    {
-      "epoch": 5.4837027379400265,
-      "grad_norm": 1.5304527282714844,
-      "learning_rate": 8.993668341708543e-05,
-      "loss": 5.6194,
-      "step": 10515
-    },
-    {
-      "epoch": 5.484224250325945,
-      "grad_norm": 1.499582290649414,
-      "learning_rate": 8.99356783919598e-05,
-      "loss": 5.5716,
-      "step": 10516
-    },
-    {
-      "epoch": 5.4847457627118645,
-      "grad_norm": 1.3592082262039185,
-      "learning_rate": 8.993467336683417e-05,
-      "loss": 6.1083,
-      "step": 10517
-    },
-    {
-      "epoch": 5.485267275097783,
-      "grad_norm": 1.5519745349884033,
-      "learning_rate": 8.993366834170855e-05,
-      "loss": 5.4412,
-      "step": 10518
-    },
-    {
-      "epoch": 5.485788787483703,
-      "grad_norm": 1.4407771825790405,
-      "learning_rate": 8.993266331658293e-05,
-      "loss": 5.7656,
-      "step": 10519
-    },
-    {
-      "epoch": 5.486310299869622,
-      "grad_norm": 1.480281114578247,
-      "learning_rate": 8.993165829145729e-05,
-      "loss": 5.4931,
-      "step": 10520
-    },
-    {
-      "epoch": 5.4868318122555415,
-      "grad_norm": 1.3419010639190674,
-      "learning_rate": 8.993065326633167e-05,
-      "loss": 5.6337,
-      "step": 10521
-    },
-    {
-      "epoch": 5.48735332464146,
-      "grad_norm": 1.4074193239212036,
-      "learning_rate": 8.992964824120603e-05,
-      "loss": 5.8925,
-      "step": 10522
-    },
-    {
-      "epoch": 5.4878748370273795,
-      "grad_norm": 1.3500640392303467,
-      "learning_rate": 8.992864321608041e-05,
-      "loss": 5.5958,
-      "step": 10523
-    },
-    {
-      "epoch": 5.488396349413298,
-      "grad_norm": 1.3122191429138184,
-      "learning_rate": 8.992763819095478e-05,
-      "loss": 6.0418,
-      "step": 10524
-    },
-    {
-      "epoch": 5.488917861799218,
-      "grad_norm": 1.4566152095794678,
-      "learning_rate": 8.992663316582915e-05,
-      "loss": 5.188,
-      "step": 10525
-    },
-    {
-      "epoch": 5.489439374185137,
-      "grad_norm": Infinity,
-      "learning_rate": 8.992663316582915e-05,
-      "loss": 5.1295,
-      "step": 10526
-    },
-    {
-      "epoch": 5.4899608865710565,
-      "grad_norm": 1.4198960065841675,
-      "learning_rate": 8.992562814070352e-05,
-      "loss": 5.7096,
-      "step": 10527
-    },
-    {
-      "epoch": 5.490482398956975,
-      "grad_norm": 1.4923793077468872,
-      "learning_rate": 8.992462311557788e-05,
-      "loss": 5.4687,
-      "step": 10528
-    },
-    {
-      "epoch": 5.491003911342895,
-      "grad_norm": 1.4669264554977417,
-      "learning_rate": 8.992361809045226e-05,
-      "loss": 5.2823,
-      "step": 10529
-    },
-    {
-      "epoch": 5.491525423728813,
-      "grad_norm": 1.3114299774169922,
-      "learning_rate": 8.992261306532664e-05,
-      "loss": 5.67,
-      "step": 10530
-    },
-    {
-      "epoch": 5.492046936114733,
-      "grad_norm": 1.3704609870910645,
-      "learning_rate": 8.992160804020102e-05,
-      "loss": 5.568,
-      "step": 10531
-    },
-    {
-      "epoch": 5.492568448500652,
-      "grad_norm": 1.3184897899627686,
-      "learning_rate": 8.992060301507538e-05,
-      "loss": 5.702,
-      "step": 10532
-    },
-    {
-      "epoch": 5.493089960886571,
-      "grad_norm": 1.3617497682571411,
-      "learning_rate": 8.991959798994976e-05,
-      "loss": 5.3451,
-      "step": 10533
-    },
-    {
-      "epoch": 5.49361147327249,
-      "grad_norm": 1.4746029376983643,
-      "learning_rate": 8.991859296482412e-05,
-      "loss": 5.5987,
-      "step": 10534
-    },
-    {
-      "epoch": 5.49413298565841,
-      "grad_norm": 1.4093337059020996,
-      "learning_rate": 8.99175879396985e-05,
-      "loss": 5.7173,
-      "step": 10535
-    },
-    {
-      "epoch": 5.494654498044328,
-      "grad_norm": 1.4069545269012451,
-      "learning_rate": 8.991658291457286e-05,
-      "loss": 5.6955,
-      "step": 10536
-    },
-    {
-      "epoch": 5.495176010430248,
-      "grad_norm": 1.3906537294387817,
-      "learning_rate": 8.991557788944724e-05,
-      "loss": 5.9466,
-      "step": 10537
-    },
-    {
-      "epoch": 5.495697522816167,
-      "grad_norm": 1.407332181930542,
-      "learning_rate": 8.99145728643216e-05,
-      "loss": 5.4268,
-      "step": 10538
-    },
-    {
-      "epoch": 5.496219035202086,
-      "grad_norm": 1.3363192081451416,
-      "learning_rate": 8.991356783919598e-05,
-      "loss": 5.9909,
-      "step": 10539
-    },
-    {
-      "epoch": 5.496740547588005,
-      "grad_norm": 1.3480929136276245,
-      "learning_rate": 8.991256281407036e-05,
-      "loss": 5.886,
-      "step": 10540
-    },
-    {
-      "epoch": 5.497262059973925,
-      "grad_norm": 1.5294699668884277,
-      "learning_rate": 8.991155778894473e-05,
-      "loss": 5.1719,
-      "step": 10541
-    },
-    {
-      "epoch": 5.497783572359843,
-      "grad_norm": 1.4702763557434082,
-      "learning_rate": 8.99105527638191e-05,
-      "loss": 5.5723,
-      "step": 10542
-    },
-    {
-      "epoch": 5.498305084745763,
-      "grad_norm": 1.4049068689346313,
-      "learning_rate": 8.990954773869347e-05,
-      "loss": 5.7262,
-      "step": 10543
-    },
-    {
-      "epoch": 5.498826597131682,
-      "grad_norm": 1.4722094535827637,
-      "learning_rate": 8.990854271356785e-05,
-      "loss": 5.5266,
-      "step": 10544
-    },
-    {
-      "epoch": 5.499348109517601,
-      "grad_norm": 1.4187508821487427,
-      "learning_rate": 8.990753768844221e-05,
-      "loss": 5.7603,
-      "step": 10545
-    },
-    {
-      "epoch": 5.49986962190352,
-      "grad_norm": 1.4626790285110474,
-      "learning_rate": 8.990653266331659e-05,
-      "loss": 5.6251,
-      "step": 10546
-    },
-    {
-      "epoch": 5.50039113428944,
-      "grad_norm": 1.557761549949646,
-      "learning_rate": 8.990552763819095e-05,
-      "loss": 5.1311,
-      "step": 10547
-    },
-    {
-      "epoch": 5.500912646675358,
-      "grad_norm": 1.5824556350708008,
-      "learning_rate": 8.990452261306533e-05,
-      "loss": 5.7965,
-      "step": 10548
-    },
-    {
-      "epoch": 5.501434159061278,
-      "grad_norm": 1.4628400802612305,
-      "learning_rate": 8.99035175879397e-05,
-      "loss": 5.5349,
-      "step": 10549
-    },
-    {
-      "epoch": 5.501955671447197,
-      "grad_norm": 1.5432307720184326,
-      "learning_rate": 8.990251256281407e-05,
-      "loss": 5.6609,
-      "step": 10550
-    },
-    {
-      "epoch": 5.502477183833116,
-      "grad_norm": 1.4701995849609375,
-      "learning_rate": 8.990150753768845e-05,
-      "loss": 5.682,
-      "step": 10551
-    },
-    {
-      "epoch": 5.502998696219035,
-      "grad_norm": 1.5971382856369019,
-      "learning_rate": 8.990050251256283e-05,
-      "loss": 5.4788,
-      "step": 10552
-    },
-    {
-      "epoch": 5.503520208604955,
-      "grad_norm": 1.3728458881378174,
-      "learning_rate": 8.989949748743719e-05,
-      "loss": 5.7085,
-      "step": 10553
-    },
-    {
-      "epoch": 5.504041720990873,
-      "grad_norm": 1.4562474489212036,
-      "learning_rate": 8.989849246231157e-05,
-      "loss": 5.9446,
-      "step": 10554
-    },
-    {
-      "epoch": 5.504563233376793,
-      "grad_norm": 1.5724762678146362,
-      "learning_rate": 8.989748743718593e-05,
-      "loss": 5.259,
-      "step": 10555
-    },
-    {
-      "epoch": 5.505084745762712,
-      "grad_norm": 1.3382744789123535,
-      "learning_rate": 8.98964824120603e-05,
-      "loss": 5.6605,
-      "step": 10556
-    },
-    {
-      "epoch": 5.505606258148631,
-      "grad_norm": 1.3708651065826416,
-      "learning_rate": 8.989547738693468e-05,
-      "loss": 6.0056,
-      "step": 10557
-    },
-    {
-      "epoch": 5.50612777053455,
-      "grad_norm": 1.4889352321624756,
-      "learning_rate": 8.989447236180904e-05,
-      "loss": 5.1108,
-      "step": 10558
-    },
-    {
-      "epoch": 5.50664928292047,
-      "grad_norm": 1.4954710006713867,
-      "learning_rate": 8.989346733668342e-05,
-      "loss": 5.4475,
-      "step": 10559
-    },
-    {
-      "epoch": 5.507170795306388,
-      "grad_norm": 1.6496992111206055,
-      "learning_rate": 8.98924623115578e-05,
-      "loss": 4.9671,
-      "step": 10560
-    },
-    {
-      "epoch": 5.507692307692308,
-      "grad_norm": 1.398339867591858,
-      "learning_rate": 8.989145728643217e-05,
-      "loss": 5.4961,
-      "step": 10561
-    },
-    {
-      "epoch": 5.508213820078227,
-      "grad_norm": 1.358109474182129,
-      "learning_rate": 8.989045226130654e-05,
-      "loss": 5.5646,
-      "step": 10562
-    },
-    {
-      "epoch": 5.508735332464146,
-      "grad_norm": 1.439739465713501,
-      "learning_rate": 8.988944723618092e-05,
-      "loss": 5.8499,
-      "step": 10563
-    },
-    {
-      "epoch": 5.509256844850065,
-      "grad_norm": 1.5602704286575317,
-      "learning_rate": 8.988844221105528e-05,
-      "loss": 5.8156,
-      "step": 10564
-    },
-    {
-      "epoch": 5.509778357235985,
-      "grad_norm": 1.5778114795684814,
-      "learning_rate": 8.988743718592966e-05,
-      "loss": 5.7366,
-      "step": 10565
-    },
-    {
-      "epoch": 5.510299869621903,
-      "grad_norm": 1.4247666597366333,
-      "learning_rate": 8.988643216080402e-05,
-      "loss": 5.5688,
-      "step": 10566
-    },
-    {
-      "epoch": 5.510821382007823,
-      "grad_norm": 1.6040902137756348,
-      "learning_rate": 8.98854271356784e-05,
-      "loss": 5.7247,
-      "step": 10567
-    },
-    {
-      "epoch": 5.511342894393742,
-      "grad_norm": 1.5974342823028564,
-      "learning_rate": 8.988442211055276e-05,
-      "loss": 5.6316,
-      "step": 10568
-    },
-    {
-      "epoch": 5.511864406779661,
-      "grad_norm": 1.574756145477295,
-      "learning_rate": 8.988341708542713e-05,
-      "loss": 5.0064,
-      "step": 10569
-    },
-    {
-      "epoch": 5.51238591916558,
-      "grad_norm": 1.4884158372879028,
-      "learning_rate": 8.988241206030151e-05,
-      "loss": 5.8027,
-      "step": 10570
-    },
-    {
-      "epoch": 5.5129074315515,
-      "grad_norm": 1.619646430015564,
-      "learning_rate": 8.988140703517588e-05,
-      "loss": 5.8179,
-      "step": 10571
-    },
-    {
-      "epoch": 5.513428943937418,
-      "grad_norm": 1.6148234605789185,
-      "learning_rate": 8.988040201005026e-05,
-      "loss": 5.3252,
-      "step": 10572
-    },
-    {
-      "epoch": 5.513950456323338,
-      "grad_norm": 1.5819690227508545,
-      "learning_rate": 8.987939698492463e-05,
-      "loss": 5.5503,
-      "step": 10573
-    },
-    {
-      "epoch": 5.514471968709257,
-      "grad_norm": 1.5805248022079468,
-      "learning_rate": 8.9878391959799e-05,
-      "loss": 5.9462,
-      "step": 10574
-    },
-    {
-      "epoch": 5.514993481095176,
-      "grad_norm": 1.406213641166687,
-      "learning_rate": 8.987738693467337e-05,
-      "loss": 5.845,
-      "step": 10575
-    },
-    {
-      "epoch": 5.515514993481095,
-      "grad_norm": 1.6024516820907593,
-      "learning_rate": 8.987638190954775e-05,
-      "loss": 5.6243,
-      "step": 10576
-    },
-    {
-      "epoch": 5.516036505867015,
-      "grad_norm": 1.4450137615203857,
-      "learning_rate": 8.987537688442211e-05,
-      "loss": 5.6585,
-      "step": 10577
-    },
-    {
-      "epoch": 5.516558018252933,
-      "grad_norm": 1.4574956893920898,
-      "learning_rate": 8.987437185929649e-05,
-      "loss": 5.6695,
-      "step": 10578
-    },
-    {
-      "epoch": 5.517079530638853,
-      "grad_norm": 1.340336561203003,
-      "learning_rate": 8.987336683417085e-05,
-      "loss": 5.7985,
-      "step": 10579
-    },
-    {
-      "epoch": 5.517601043024772,
-      "grad_norm": 1.2795006036758423,
-      "learning_rate": 8.987236180904523e-05,
-      "loss": 5.8606,
-      "step": 10580
-    },
-    {
-      "epoch": 5.518122555410691,
-      "grad_norm": 1.5590077638626099,
-      "learning_rate": 8.987135678391961e-05,
-      "loss": 5.5806,
-      "step": 10581
-    },
-    {
-      "epoch": 5.51864406779661,
-      "grad_norm": 1.617225170135498,
-      "learning_rate": 8.987035175879397e-05,
-      "loss": 4.6873,
-      "step": 10582
-    },
-    {
-      "epoch": 5.51916558018253,
-      "grad_norm": 1.4698587656021118,
-      "learning_rate": 8.986934673366835e-05,
-      "loss": 6.0167,
-      "step": 10583
-    },
-    {
-      "epoch": 5.519687092568448,
-      "grad_norm": 1.4182524681091309,
-      "learning_rate": 8.986834170854272e-05,
-      "loss": 5.7454,
-      "step": 10584
-    },
-    {
-      "epoch": 5.520208604954368,
-      "grad_norm": 1.429352879524231,
-      "learning_rate": 8.986733668341709e-05,
-      "loss": 5.3285,
-      "step": 10585
-    },
-    {
-      "epoch": 5.520730117340287,
-      "grad_norm": 1.5638583898544312,
-      "learning_rate": 8.986633165829146e-05,
-      "loss": 5.0868,
-      "step": 10586
-    },
-    {
-      "epoch": 5.521251629726206,
-      "grad_norm": 1.6468461751937866,
-      "learning_rate": 8.986532663316584e-05,
-      "loss": 5.3661,
-      "step": 10587
-    },
-    {
-      "epoch": 5.521773142112125,
-      "grad_norm": 1.465616226196289,
-      "learning_rate": 8.98643216080402e-05,
-      "loss": 5.5858,
-      "step": 10588
-    },
-    {
-      "epoch": 5.522294654498045,
-      "grad_norm": 1.4571688175201416,
-      "learning_rate": 8.986331658291458e-05,
-      "loss": 5.5763,
-      "step": 10589
-    },
-    {
-      "epoch": 5.522816166883963,
-      "grad_norm": 1.4036407470703125,
-      "learning_rate": 8.986231155778894e-05,
-      "loss": 5.7032,
-      "step": 10590
-    },
-    {
-      "epoch": 5.523337679269883,
-      "grad_norm": 1.4435793161392212,
-      "learning_rate": 8.986130653266332e-05,
-      "loss": 5.3011,
-      "step": 10591
-    },
-    {
-      "epoch": 5.523859191655802,
-      "grad_norm": 1.4620122909545898,
-      "learning_rate": 8.98603015075377e-05,
-      "loss": 5.6311,
-      "step": 10592
-    },
-    {
-      "epoch": 5.524380704041721,
-      "grad_norm": 1.5172603130340576,
-      "learning_rate": 8.985929648241207e-05,
-      "loss": 5.853,
-      "step": 10593
-    },
-    {
-      "epoch": 5.52490221642764,
-      "grad_norm": 1.4170520305633545,
-      "learning_rate": 8.985829145728644e-05,
-      "loss": 5.4437,
-      "step": 10594
-    },
-    {
-      "epoch": 5.52542372881356,
-      "grad_norm": 1.4288333654403687,
-      "learning_rate": 8.98572864321608e-05,
-      "loss": 5.6101,
-      "step": 10595
-    },
-    {
-      "epoch": 5.525945241199478,
-      "grad_norm": 1.5294800996780396,
-      "learning_rate": 8.985628140703518e-05,
-      "loss": 5.4093,
-      "step": 10596
-    },
-    {
-      "epoch": 5.526466753585398,
-      "grad_norm": 1.3892121315002441,
-      "learning_rate": 8.985527638190955e-05,
-      "loss": 5.1796,
-      "step": 10597
-    },
-    {
-      "epoch": 5.526988265971317,
-      "grad_norm": 1.5848643779754639,
-      "learning_rate": 8.985427135678392e-05,
-      "loss": 5.3476,
-      "step": 10598
-    },
-    {
-      "epoch": 5.527509778357236,
-      "grad_norm": 1.5734790563583374,
-      "learning_rate": 8.985326633165829e-05,
-      "loss": 5.5475,
-      "step": 10599
-    },
-    {
-      "epoch": 5.528031290743155,
-      "grad_norm": 1.8406802415847778,
-      "learning_rate": 8.985226130653267e-05,
-      "loss": 5.2144,
-      "step": 10600
-    },
-    {
-      "epoch": 5.528552803129074,
-      "grad_norm": 1.4999756813049316,
-      "learning_rate": 8.985125628140704e-05,
-      "loss": 5.2671,
-      "step": 10601
-    },
-    {
-      "epoch": 5.529074315514993,
-      "grad_norm": 1.4691054821014404,
-      "learning_rate": 8.985025125628142e-05,
-      "loss": 5.7284,
-      "step": 10602
-    },
-    {
-      "epoch": 5.529595827900913,
-      "grad_norm": 1.4603995084762573,
-      "learning_rate": 8.984924623115579e-05,
-      "loss": 5.2496,
-      "step": 10603
-    },
-    {
-      "epoch": 5.530117340286832,
-      "grad_norm": 1.4993500709533691,
-      "learning_rate": 8.984824120603016e-05,
-      "loss": 5.475,
-      "step": 10604
-    },
-    {
-      "epoch": 5.530638852672751,
-      "grad_norm": 1.489986777305603,
-      "learning_rate": 8.984723618090453e-05,
-      "loss": 5.6777,
-      "step": 10605
-    },
-    {
-      "epoch": 5.53116036505867,
-      "grad_norm": 1.5664591789245605,
-      "learning_rate": 8.98462311557789e-05,
-      "loss": 5.1724,
-      "step": 10606
-    },
-    {
-      "epoch": 5.531681877444589,
-      "grad_norm": 1.8480204343795776,
-      "learning_rate": 8.984522613065327e-05,
-      "loss": 4.8966,
-      "step": 10607
-    },
-    {
-      "epoch": 5.532203389830508,
-      "grad_norm": 1.4696300029754639,
-      "learning_rate": 8.984422110552763e-05,
-      "loss": 5.4115,
-      "step": 10608
-    },
-    {
-      "epoch": 5.532724902216428,
-      "grad_norm": 1.3801639080047607,
-      "learning_rate": 8.984321608040201e-05,
-      "loss": 5.6578,
-      "step": 10609
-    },
-    {
-      "epoch": 5.533246414602347,
-      "grad_norm": 1.5254082679748535,
-      "learning_rate": 8.984221105527638e-05,
-      "loss": 5.3079,
-      "step": 10610
-    },
-    {
-      "epoch": 5.533767926988266,
-      "grad_norm": 1.5073308944702148,
-      "learning_rate": 8.984120603015075e-05,
-      "loss": 5.5272,
-      "step": 10611
-    },
-    {
-      "epoch": 5.534289439374185,
-      "grad_norm": 1.3704605102539062,
-      "learning_rate": 8.984020100502513e-05,
-      "loss": 5.6318,
-      "step": 10612
-    },
-    {
-      "epoch": 5.534810951760104,
-      "grad_norm": 1.4746230840682983,
-      "learning_rate": 8.983919597989951e-05,
-      "loss": 5.5075,
-      "step": 10613
-    },
-    {
-      "epoch": 5.535332464146023,
-      "grad_norm": 1.4431073665618896,
-      "learning_rate": 8.983819095477387e-05,
-      "loss": 5.389,
-      "step": 10614
-    },
-    {
-      "epoch": 5.535853976531943,
-      "grad_norm": 1.351820945739746,
-      "learning_rate": 8.983718592964825e-05,
-      "loss": 5.7438,
-      "step": 10615
-    },
-    {
-      "epoch": 5.536375488917862,
-      "grad_norm": 1.5364855527877808,
-      "learning_rate": 8.983618090452262e-05,
-      "loss": 5.4546,
-      "step": 10616
-    },
-    {
-      "epoch": 5.536897001303781,
-      "grad_norm": 1.5313864946365356,
-      "learning_rate": 8.9835175879397e-05,
-      "loss": 5.5136,
-      "step": 10617
-    },
-    {
-      "epoch": 5.5374185136897,
-      "grad_norm": 1.2924866676330566,
-      "learning_rate": 8.983417085427136e-05,
-      "loss": 6.0312,
-      "step": 10618
-    },
-    {
-      "epoch": 5.537940026075619,
-      "grad_norm": 1.4201401472091675,
-      "learning_rate": 8.983316582914574e-05,
-      "loss": 5.7921,
-      "step": 10619
-    },
-    {
-      "epoch": 5.538461538461538,
-      "grad_norm": 1.2767994403839111,
-      "learning_rate": 8.98321608040201e-05,
-      "loss": 6.0358,
-      "step": 10620
-    },
-    {
-      "epoch": 5.538983050847458,
-      "grad_norm": 1.6226348876953125,
-      "learning_rate": 8.983115577889448e-05,
-      "loss": 5.2615,
-      "step": 10621
-    },
-    {
-      "epoch": 5.539504563233377,
-      "grad_norm": 1.3227152824401855,
-      "learning_rate": 8.983015075376886e-05,
-      "loss": 5.7721,
-      "step": 10622
-    },
-    {
-      "epoch": 5.540026075619296,
-      "grad_norm": 1.4500418901443481,
-      "learning_rate": 8.982914572864322e-05,
-      "loss": 5.5227,
-      "step": 10623
-    },
-    {
-      "epoch": 5.540547588005215,
-      "grad_norm": 1.32511568069458,
-      "learning_rate": 8.98281407035176e-05,
-      "loss": 5.6303,
-      "step": 10624
-    },
-    {
-      "epoch": 5.541069100391134,
-      "grad_norm": 1.4139864444732666,
-      "learning_rate": 8.982713567839196e-05,
-      "loss": 5.779,
-      "step": 10625
-    },
-    {
-      "epoch": 5.541590612777053,
-      "grad_norm": 1.6047974824905396,
-      "learning_rate": 8.982613065326634e-05,
-      "loss": 5.8524,
-      "step": 10626
-    },
-    {
-      "epoch": 5.542112125162973,
-      "grad_norm": 1.4181280136108398,
-      "learning_rate": 8.98251256281407e-05,
-      "loss": 5.75,
-      "step": 10627
-    },
-    {
-      "epoch": 5.542633637548892,
-      "grad_norm": 1.4477341175079346,
-      "learning_rate": 8.982412060301508e-05,
-      "loss": 5.327,
-      "step": 10628
-    },
-    {
-      "epoch": 5.543155149934811,
-      "grad_norm": 1.451240062713623,
-      "learning_rate": 8.982311557788945e-05,
-      "loss": 5.4207,
-      "step": 10629
-    },
-    {
-      "epoch": 5.54367666232073,
-      "grad_norm": 1.4802944660186768,
-      "learning_rate": 8.982211055276382e-05,
-      "loss": 5.4428,
-      "step": 10630
-    },
-    {
-      "epoch": 5.544198174706649,
-      "grad_norm": 1.5272756814956665,
-      "learning_rate": 8.982110552763819e-05,
-      "loss": 5.185,
-      "step": 10631
-    },
-    {
-      "epoch": 5.544719687092568,
-      "grad_norm": 1.4353892803192139,
-      "learning_rate": 8.982010050251257e-05,
-      "loss": 5.6617,
-      "step": 10632
-    },
-    {
-      "epoch": 5.545241199478488,
-      "grad_norm": 1.3849009275436401,
-      "learning_rate": 8.981909547738694e-05,
-      "loss": 5.8204,
-      "step": 10633
-    },
-    {
-      "epoch": 5.545762711864406,
-      "grad_norm": 1.441770315170288,
-      "learning_rate": 8.981809045226131e-05,
-      "loss": 5.4206,
-      "step": 10634
-    },
-    {
-      "epoch": 5.546284224250326,
-      "grad_norm": 1.3511791229248047,
-      "learning_rate": 8.981708542713569e-05,
-      "loss": 6.1352,
-      "step": 10635
-    },
-    {
-      "epoch": 5.546805736636245,
-      "grad_norm": 1.5235979557037354,
-      "learning_rate": 8.981608040201005e-05,
-      "loss": 5.3545,
-      "step": 10636
-    },
-    {
-      "epoch": 5.547327249022164,
-      "grad_norm": 1.431562900543213,
-      "learning_rate": 8.981507537688443e-05,
-      "loss": 5.7072,
-      "step": 10637
-    },
-    {
-      "epoch": 5.547848761408083,
-      "grad_norm": 1.5068142414093018,
-      "learning_rate": 8.981407035175879e-05,
-      "loss": 5.2638,
-      "step": 10638
-    },
-    {
-      "epoch": 5.548370273794003,
-      "grad_norm": 1.5173410177230835,
-      "learning_rate": 8.981306532663317e-05,
-      "loss": 5.5781,
-      "step": 10639
-    },
-    {
-      "epoch": 5.548891786179921,
-      "grad_norm": 1.4697211980819702,
-      "learning_rate": 8.981206030150753e-05,
-      "loss": 5.7283,
-      "step": 10640
-    },
-    {
-      "epoch": 5.549413298565841,
-      "grad_norm": 1.4464901685714722,
-      "learning_rate": 8.981105527638191e-05,
-      "loss": 5.6286,
-      "step": 10641
-    },
-    {
-      "epoch": 5.54993481095176,
-      "grad_norm": 1.5917302370071411,
-      "learning_rate": 8.981005025125629e-05,
-      "loss": 5.4144,
-      "step": 10642
-    },
-    {
-      "epoch": 5.550456323337679,
-      "grad_norm": 1.7045180797576904,
-      "learning_rate": 8.980904522613067e-05,
-      "loss": 5.1914,
-      "step": 10643
-    },
-    {
-      "epoch": 5.550977835723598,
-      "grad_norm": 1.456911325454712,
-      "learning_rate": 8.980804020100503e-05,
-      "loss": 5.5286,
-      "step": 10644
-    },
-    {
-      "epoch": 5.551499348109518,
-      "grad_norm": 1.3162492513656616,
-      "learning_rate": 8.980703517587941e-05,
-      "loss": 5.6386,
-      "step": 10645
-    },
-    {
-      "epoch": 5.552020860495436,
-      "grad_norm": 1.572289228439331,
-      "learning_rate": 8.980603015075377e-05,
-      "loss": 5.259,
-      "step": 10646
-    },
-    {
-      "epoch": 5.552542372881356,
-      "grad_norm": 1.5026469230651855,
-      "learning_rate": 8.980502512562815e-05,
-      "loss": 5.6571,
-      "step": 10647
-    },
-    {
-      "epoch": 5.553063885267275,
-      "grad_norm": 1.4612823724746704,
-      "learning_rate": 8.980402010050252e-05,
-      "loss": 5.9021,
-      "step": 10648
-    },
-    {
-      "epoch": 5.553585397653194,
-      "grad_norm": 1.4366511106491089,
-      "learning_rate": 8.980301507537688e-05,
-      "loss": 5.5088,
-      "step": 10649
-    },
-    {
-      "epoch": 5.554106910039113,
-      "grad_norm": 1.3544087409973145,
-      "learning_rate": 8.980201005025126e-05,
-      "loss": 6.0035,
-      "step": 10650
-    },
-    {
-      "epoch": 5.554628422425033,
-      "grad_norm": 1.4471513032913208,
-      "learning_rate": 8.980100502512562e-05,
-      "loss": 5.6498,
-      "step": 10651
-    },
-    {
-      "epoch": 5.555149934810951,
-      "grad_norm": 1.4358118772506714,
-      "learning_rate": 8.98e-05,
-      "loss": 5.2415,
-      "step": 10652
-    },
-    {
-      "epoch": 5.555671447196871,
-      "grad_norm": 1.5437648296356201,
-      "learning_rate": 8.979899497487438e-05,
-      "loss": 5.0231,
-      "step": 10653
-    },
-    {
-      "epoch": 5.55619295958279,
-      "grad_norm": 1.5040374994277954,
-      "learning_rate": 8.979798994974876e-05,
-      "loss": 5.4041,
-      "step": 10654
-    },
-    {
-      "epoch": 5.556714471968709,
-      "grad_norm": 1.440911054611206,
-      "learning_rate": 8.979698492462312e-05,
-      "loss": 5.7352,
-      "step": 10655
-    },
-    {
-      "epoch": 5.557235984354628,
-      "grad_norm": 1.463666558265686,
-      "learning_rate": 8.97959798994975e-05,
-      "loss": 5.5876,
-      "step": 10656
-    },
-    {
-      "epoch": 5.557757496740548,
-      "grad_norm": 1.5259830951690674,
-      "learning_rate": 8.979497487437186e-05,
-      "loss": 5.2268,
-      "step": 10657
-    },
-    {
-      "epoch": 5.558279009126466,
-      "grad_norm": 1.4595916271209717,
-      "learning_rate": 8.979396984924624e-05,
-      "loss": 5.8421,
-      "step": 10658
-    },
-    {
-      "epoch": 5.558800521512386,
-      "grad_norm": 1.4905551671981812,
-      "learning_rate": 8.97929648241206e-05,
-      "loss": 5.5621,
-      "step": 10659
-    },
-    {
-      "epoch": 5.559322033898305,
-      "grad_norm": 1.4744524955749512,
-      "learning_rate": 8.979195979899498e-05,
-      "loss": 5.705,
-      "step": 10660
-    },
-    {
-      "epoch": 5.559843546284224,
-      "grad_norm": 1.5570566654205322,
-      "learning_rate": 8.979095477386935e-05,
-      "loss": 5.7084,
-      "step": 10661
-    },
-    {
-      "epoch": 5.560365058670143,
-      "grad_norm": 1.5039702653884888,
-      "learning_rate": 8.978994974874371e-05,
-      "loss": 5.3165,
-      "step": 10662
-    },
-    {
-      "epoch": 5.560886571056063,
-      "grad_norm": 1.373837947845459,
-      "learning_rate": 8.978894472361809e-05,
-      "loss": 5.6137,
-      "step": 10663
-    },
-    {
-      "epoch": 5.5614080834419815,
-      "grad_norm": 1.5403666496276855,
-      "learning_rate": 8.978793969849247e-05,
-      "loss": 5.9779,
-      "step": 10664
-    },
-    {
-      "epoch": 5.561929595827901,
-      "grad_norm": 1.3370065689086914,
-      "learning_rate": 8.978693467336684e-05,
-      "loss": 5.9706,
-      "step": 10665
-    },
-    {
-      "epoch": 5.56245110821382,
-      "grad_norm": 1.425170660018921,
-      "learning_rate": 8.978592964824121e-05,
-      "loss": 5.4408,
-      "step": 10666
-    },
-    {
-      "epoch": 5.562972620599739,
-      "grad_norm": 1.4637962579727173,
-      "learning_rate": 8.978492462311559e-05,
-      "loss": 5.1684,
-      "step": 10667
-    },
-    {
-      "epoch": 5.563494132985658,
-      "grad_norm": 1.6192643642425537,
-      "learning_rate": 8.978391959798995e-05,
-      "loss": 5.5964,
-      "step": 10668
-    },
-    {
-      "epoch": 5.564015645371578,
-      "grad_norm": 1.5467052459716797,
-      "learning_rate": 8.978291457286433e-05,
-      "loss": 4.8645,
-      "step": 10669
-    },
-    {
-      "epoch": 5.5645371577574965,
-      "grad_norm": 1.4735559225082397,
-      "learning_rate": 8.97819095477387e-05,
-      "loss": 5.4924,
-      "step": 10670
-    },
-    {
-      "epoch": 5.565058670143416,
-      "grad_norm": 1.4907965660095215,
-      "learning_rate": 8.978090452261307e-05,
-      "loss": 5.2758,
-      "step": 10671
-    },
-    {
-      "epoch": 5.565580182529335,
-      "grad_norm": 1.443861484527588,
-      "learning_rate": 8.977989949748744e-05,
-      "loss": 5.9669,
-      "step": 10672
-    },
-    {
-      "epoch": 5.566101694915254,
-      "grad_norm": 1.4585715532302856,
-      "learning_rate": 8.977889447236181e-05,
-      "loss": 5.257,
-      "step": 10673
-    },
-    {
-      "epoch": 5.566623207301173,
-      "grad_norm": 1.4580118656158447,
-      "learning_rate": 8.977788944723619e-05,
-      "loss": 5.4391,
-      "step": 10674
-    },
-    {
-      "epoch": 5.567144719687093,
-      "grad_norm": 1.3736151456832886,
-      "learning_rate": 8.977688442211056e-05,
-      "loss": 5.4986,
-      "step": 10675
-    },
-    {
-      "epoch": 5.5676662320730115,
-      "grad_norm": 1.4873274564743042,
-      "learning_rate": 8.977587939698493e-05,
-      "loss": 5.3497,
-      "step": 10676
-    },
-    {
-      "epoch": 5.568187744458931,
-      "grad_norm": 1.4092762470245361,
-      "learning_rate": 8.97748743718593e-05,
-      "loss": 5.8026,
-      "step": 10677
-    },
-    {
-      "epoch": 5.56870925684485,
-      "grad_norm": 1.4899189472198486,
-      "learning_rate": 8.977386934673368e-05,
-      "loss": 5.0795,
-      "step": 10678
-    },
-    {
-      "epoch": 5.569230769230769,
-      "grad_norm": 1.4775065183639526,
-      "learning_rate": 8.977286432160804e-05,
-      "loss": 5.588,
-      "step": 10679
-    },
-    {
-      "epoch": 5.5697522816166884,
-      "grad_norm": 1.4677801132202148,
-      "learning_rate": 8.977185929648242e-05,
-      "loss": 5.5959,
-      "step": 10680
-    },
-    {
-      "epoch": 5.570273794002608,
-      "grad_norm": 1.4771485328674316,
-      "learning_rate": 8.977085427135678e-05,
-      "loss": 5.6068,
-      "step": 10681
-    },
-    {
-      "epoch": 5.5707953063885265,
-      "grad_norm": 1.5151886940002441,
-      "learning_rate": 8.976984924623116e-05,
-      "loss": 5.7509,
-      "step": 10682
-    },
-    {
-      "epoch": 5.571316818774446,
-      "grad_norm": 1.349482774734497,
-      "learning_rate": 8.976884422110552e-05,
-      "loss": 5.5763,
-      "step": 10683
-    },
-    {
-      "epoch": 5.571838331160365,
-      "grad_norm": 1.3438308238983154,
-      "learning_rate": 8.97678391959799e-05,
-      "loss": 5.8156,
-      "step": 10684
-    },
-    {
-      "epoch": 5.572359843546284,
-      "grad_norm": 1.5698109865188599,
-      "learning_rate": 8.976683417085428e-05,
-      "loss": 4.9264,
-      "step": 10685
-    },
-    {
-      "epoch": 5.5728813559322035,
-      "grad_norm": 1.441450595855713,
-      "learning_rate": 8.976582914572866e-05,
-      "loss": 5.7915,
-      "step": 10686
-    },
-    {
-      "epoch": 5.573402868318123,
-      "grad_norm": 1.3790888786315918,
-      "learning_rate": 8.976482412060302e-05,
-      "loss": 5.5852,
-      "step": 10687
-    },
-    {
-      "epoch": 5.5739243807040415,
-      "grad_norm": 1.468972086906433,
-      "learning_rate": 8.976381909547739e-05,
-      "loss": 5.6414,
-      "step": 10688
-    },
-    {
-      "epoch": 5.574445893089961,
-      "grad_norm": 1.307621717453003,
-      "learning_rate": 8.976281407035176e-05,
-      "loss": 5.2198,
-      "step": 10689
-    },
-    {
-      "epoch": 5.57496740547588,
-      "grad_norm": 1.4904778003692627,
-      "learning_rate": 8.976180904522613e-05,
-      "loss": 5.4697,
-      "step": 10690
-    },
-    {
-      "epoch": 5.575488917861799,
-      "grad_norm": 1.4442285299301147,
-      "learning_rate": 8.97608040201005e-05,
-      "loss": 6.0406,
-      "step": 10691
-    },
-    {
-      "epoch": 5.5760104302477185,
-      "grad_norm": 1.4455108642578125,
-      "learning_rate": 8.975979899497487e-05,
-      "loss": 5.5153,
-      "step": 10692
-    },
-    {
-      "epoch": 5.576531942633638,
-      "grad_norm": 1.4157465696334839,
-      "learning_rate": 8.975879396984925e-05,
-      "loss": 5.4922,
-      "step": 10693
-    },
-    {
-      "epoch": 5.5770534550195565,
-      "grad_norm": 1.4414554834365845,
-      "learning_rate": 8.975778894472363e-05,
-      "loss": 5.6488,
-      "step": 10694
-    },
-    {
-      "epoch": 5.577574967405476,
-      "grad_norm": 1.3763607740402222,
-      "learning_rate": 8.9756783919598e-05,
-      "loss": 5.5192,
-      "step": 10695
-    },
-    {
-      "epoch": 5.578096479791395,
-      "grad_norm": 1.4902335405349731,
-      "learning_rate": 8.975577889447237e-05,
-      "loss": 5.6269,
-      "step": 10696
-    },
-    {
-      "epoch": 5.578617992177314,
-      "grad_norm": 1.3486636877059937,
-      "learning_rate": 8.975477386934675e-05,
-      "loss": 5.8181,
-      "step": 10697
-    },
-    {
-      "epoch": 5.5791395045632335,
-      "grad_norm": 1.4632179737091064,
-      "learning_rate": 8.975376884422111e-05,
-      "loss": 5.1638,
-      "step": 10698
-    },
-    {
-      "epoch": 5.579661016949153,
-      "grad_norm": 1.8298215866088867,
-      "learning_rate": 8.975276381909549e-05,
-      "loss": 5.6288,
-      "step": 10699
-    },
-    {
-      "epoch": 5.5801825293350715,
-      "grad_norm": 1.473177194595337,
-      "learning_rate": 8.975175879396985e-05,
-      "loss": 5.7395,
-      "step": 10700
-    },
-    {
-      "epoch": 5.580704041720991,
-      "grad_norm": 1.4965496063232422,
-      "learning_rate": 8.975075376884422e-05,
-      "loss": 5.8979,
-      "step": 10701
-    },
-    {
-      "epoch": 5.58122555410691,
-      "grad_norm": 1.496484637260437,
-      "learning_rate": 8.97497487437186e-05,
-      "loss": 5.6839,
-      "step": 10702
-    },
-    {
-      "epoch": 5.581747066492829,
-      "grad_norm": 1.6331276893615723,
-      "learning_rate": 8.974874371859296e-05,
-      "loss": 5.6053,
-      "step": 10703
-    },
-    {
-      "epoch": 5.5822685788787485,
-      "grad_norm": 1.4352926015853882,
-      "learning_rate": 8.974773869346734e-05,
-      "loss": 5.7001,
-      "step": 10704
-    },
-    {
-      "epoch": 5.582790091264668,
-      "grad_norm": 1.5482256412506104,
-      "learning_rate": 8.974673366834171e-05,
-      "loss": 5.3971,
-      "step": 10705
-    },
-    {
-      "epoch": 5.5833116036505865,
-      "grad_norm": 1.4308723211288452,
-      "learning_rate": 8.974572864321609e-05,
-      "loss": 5.8068,
-      "step": 10706
-    },
-    {
-      "epoch": 5.583833116036506,
-      "grad_norm": 1.468033790588379,
-      "learning_rate": 8.974472361809046e-05,
-      "loss": 5.4917,
-      "step": 10707
-    },
-    {
-      "epoch": 5.584354628422425,
-      "grad_norm": 1.4369900226593018,
-      "learning_rate": 8.974371859296483e-05,
-      "loss": 5.5902,
-      "step": 10708
-    },
-    {
-      "epoch": 5.584876140808344,
-      "grad_norm": 1.4233335256576538,
-      "learning_rate": 8.97427135678392e-05,
-      "loss": 5.9019,
-      "step": 10709
-    },
-    {
-      "epoch": 5.5853976531942635,
-      "grad_norm": 1.5117610692977905,
-      "learning_rate": 8.974170854271358e-05,
-      "loss": 5.5663,
-      "step": 10710
-    },
-    {
-      "epoch": 5.585919165580183,
-      "grad_norm": 1.6657543182373047,
-      "learning_rate": 8.974070351758794e-05,
-      "loss": 5.517,
-      "step": 10711
-    },
-    {
-      "epoch": 5.5864406779661016,
-      "grad_norm": 1.5158084630966187,
-      "learning_rate": 8.973969849246232e-05,
-      "loss": 5.2128,
-      "step": 10712
-    },
-    {
-      "epoch": 5.586962190352021,
-      "grad_norm": 1.4552608728408813,
-      "learning_rate": 8.973869346733668e-05,
-      "loss": 5.2406,
-      "step": 10713
-    },
-    {
-      "epoch": 5.58748370273794,
-      "grad_norm": 1.4881843328475952,
-      "learning_rate": 8.973768844221106e-05,
-      "loss": 5.299,
-      "step": 10714
-    },
-    {
-      "epoch": 5.588005215123859,
-      "grad_norm": 1.497010588645935,
-      "learning_rate": 8.973668341708544e-05,
-      "loss": 5.6019,
-      "step": 10715
-    },
-    {
-      "epoch": 5.5885267275097785,
-      "grad_norm": 1.6426019668579102,
-      "learning_rate": 8.97356783919598e-05,
-      "loss": 5.6731,
-      "step": 10716
-    },
-    {
-      "epoch": 5.589048239895698,
-      "grad_norm": 1.3704931735992432,
-      "learning_rate": 8.973467336683418e-05,
-      "loss": 5.5186,
-      "step": 10717
-    },
-    {
-      "epoch": 5.589569752281617,
-      "grad_norm": 1.3988368511199951,
-      "learning_rate": 8.973366834170854e-05,
-      "loss": 5.4417,
-      "step": 10718
-    },
-    {
-      "epoch": 5.590091264667536,
-      "grad_norm": 1.3900200128555298,
-      "learning_rate": 8.973266331658292e-05,
-      "loss": 5.7314,
-      "step": 10719
-    },
-    {
-      "epoch": 5.590612777053455,
-      "grad_norm": 1.4626874923706055,
-      "learning_rate": 8.973165829145729e-05,
-      "loss": 5.5044,
-      "step": 10720
-    },
-    {
-      "epoch": 5.591134289439374,
-      "grad_norm": 1.4698872566223145,
-      "learning_rate": 8.973065326633166e-05,
-      "loss": 5.5728,
-      "step": 10721
-    },
-    {
-      "epoch": 5.5916558018252935,
-      "grad_norm": 1.391894817352295,
-      "learning_rate": 8.972964824120603e-05,
-      "loss": 5.8829,
-      "step": 10722
-    },
-    {
-      "epoch": 5.592177314211213,
-      "grad_norm": 1.3896067142486572,
-      "learning_rate": 8.97286432160804e-05,
-      "loss": 5.8358,
-      "step": 10723
-    },
-    {
-      "epoch": 5.592698826597132,
-      "grad_norm": 1.4350017309188843,
-      "learning_rate": 8.972763819095477e-05,
-      "loss": 5.7083,
-      "step": 10724
-    },
-    {
-      "epoch": 5.593220338983051,
-      "grad_norm": 1.4307115077972412,
-      "learning_rate": 8.972663316582915e-05,
-      "loss": 5.4359,
-      "step": 10725
-    },
-    {
-      "epoch": 5.59374185136897,
-      "grad_norm": 2.6177988052368164,
-      "learning_rate": 8.972562814070353e-05,
-      "loss": 5.0889,
-      "step": 10726
-    },
-    {
-      "epoch": 5.594263363754889,
-      "grad_norm": 1.5480645895004272,
-      "learning_rate": 8.972462311557789e-05,
-      "loss": 5.5142,
-      "step": 10727
-    },
-    {
-      "epoch": 5.5947848761408085,
-      "grad_norm": 1.384631872177124,
-      "learning_rate": 8.972361809045227e-05,
-      "loss": 5.7605,
-      "step": 10728
-    },
-    {
-      "epoch": 5.595306388526727,
-      "grad_norm": 1.7982373237609863,
-      "learning_rate": 8.972261306532663e-05,
-      "loss": 4.7923,
-      "step": 10729
-    },
-    {
-      "epoch": 5.595827900912647,
-      "grad_norm": 1.5866572856903076,
-      "learning_rate": 8.972160804020101e-05,
-      "loss": 5.2396,
-      "step": 10730
-    },
-    {
-      "epoch": 5.596349413298566,
-      "grad_norm": 1.5016323328018188,
-      "learning_rate": 8.972060301507537e-05,
-      "loss": 5.9116,
-      "step": 10731
-    },
-    {
-      "epoch": 5.596870925684485,
-      "grad_norm": 1.4689751863479614,
-      "learning_rate": 8.971959798994975e-05,
-      "loss": 5.7532,
-      "step": 10732
-    },
-    {
-      "epoch": 5.597392438070404,
-      "grad_norm": 1.4844993352890015,
-      "learning_rate": 8.971859296482412e-05,
-      "loss": 5.8688,
-      "step": 10733
-    },
-    {
-      "epoch": 5.597913950456324,
-      "grad_norm": 1.5646308660507202,
-      "learning_rate": 8.97175879396985e-05,
-      "loss": 5.5851,
-      "step": 10734
-    },
-    {
-      "epoch": 5.598435462842242,
-      "grad_norm": 1.4569100141525269,
-      "learning_rate": 8.971658291457287e-05,
-      "loss": 5.8676,
-      "step": 10735
-    },
-    {
-      "epoch": 5.598956975228162,
-      "grad_norm": 1.5433564186096191,
-      "learning_rate": 8.971557788944725e-05,
-      "loss": 4.9837,
-      "step": 10736
-    },
-    {
-      "epoch": 5.599478487614081,
-      "grad_norm": 1.4119356870651245,
-      "learning_rate": 8.971457286432161e-05,
-      "loss": 5.6885,
-      "step": 10737
-    },
-    {
-      "epoch": 5.6,
-      "grad_norm": 1.3119248151779175,
-      "learning_rate": 8.971356783919599e-05,
-      "loss": 5.923,
-      "step": 10738
-    },
-    {
-      "epoch": 5.600521512385919,
-      "grad_norm": 1.4582245349884033,
-      "learning_rate": 8.971256281407036e-05,
-      "loss": 5.2739,
-      "step": 10739
-    },
-    {
-      "epoch": 5.601043024771839,
-      "grad_norm": 1.5176348686218262,
-      "learning_rate": 8.971155778894473e-05,
-      "loss": 5.6583,
-      "step": 10740
-    },
-    {
-      "epoch": 5.601564537157757,
-      "grad_norm": 1.5285310745239258,
-      "learning_rate": 8.97105527638191e-05,
-      "loss": 5.5315,
-      "step": 10741
-    },
-    {
-      "epoch": 5.602086049543677,
-      "grad_norm": 1.5655077695846558,
-      "learning_rate": 8.970954773869346e-05,
-      "loss": 5.3607,
-      "step": 10742
-    },
-    {
-      "epoch": 5.602607561929596,
-      "grad_norm": 1.5796623229980469,
-      "learning_rate": 8.970854271356784e-05,
-      "loss": 5.2698,
-      "step": 10743
-    },
-    {
-      "epoch": 5.603129074315515,
-      "grad_norm": 1.4663238525390625,
-      "learning_rate": 8.97075376884422e-05,
-      "loss": 5.6631,
-      "step": 10744
-    },
-    {
-      "epoch": 5.603650586701434,
-      "grad_norm": 1.4652966260910034,
-      "learning_rate": 8.970653266331658e-05,
-      "loss": 5.6178,
-      "step": 10745
-    },
-    {
-      "epoch": 5.604172099087354,
-      "grad_norm": 1.7010948657989502,
-      "learning_rate": 8.970552763819096e-05,
-      "loss": 4.4571,
-      "step": 10746
-    },
-    {
-      "epoch": 5.604693611473272,
-      "grad_norm": 1.422619104385376,
-      "learning_rate": 8.970452261306534e-05,
-      "loss": 5.6791,
-      "step": 10747
-    },
-    {
-      "epoch": 5.605215123859192,
-      "grad_norm": 1.5135239362716675,
-      "learning_rate": 8.97035175879397e-05,
-      "loss": 5.024,
-      "step": 10748
-    },
-    {
-      "epoch": 5.605736636245111,
-      "grad_norm": 1.374961495399475,
-      "learning_rate": 8.970251256281408e-05,
-      "loss": 5.9613,
-      "step": 10749
-    },
-    {
-      "epoch": 5.60625814863103,
-      "grad_norm": 1.4520517587661743,
-      "learning_rate": 8.970150753768845e-05,
-      "loss": 5.6929,
-      "step": 10750
-    },
-    {
-      "epoch": 5.606779661016949,
-      "grad_norm": 1.4950183629989624,
-      "learning_rate": 8.970050251256282e-05,
-      "loss": 5.6589,
-      "step": 10751
-    },
-    {
-      "epoch": 5.607301173402869,
-      "grad_norm": 1.620504379272461,
-      "learning_rate": 8.969949748743719e-05,
-      "loss": 5.3102,
-      "step": 10752
-    },
-    {
-      "epoch": 5.607822685788787,
-      "grad_norm": 1.5027135610580444,
-      "learning_rate": 8.969849246231157e-05,
-      "loss": 5.3994,
-      "step": 10753
-    },
-    {
-      "epoch": 5.608344198174707,
-      "grad_norm": 1.611693024635315,
-      "learning_rate": 8.969748743718593e-05,
-      "loss": 5.3491,
-      "step": 10754
-    },
-    {
-      "epoch": 5.608865710560626,
-      "grad_norm": 1.4383951425552368,
-      "learning_rate": 8.969648241206031e-05,
-      "loss": 5.6933,
-      "step": 10755
-    },
-    {
-      "epoch": 5.609387222946545,
-      "grad_norm": 1.414306402206421,
-      "learning_rate": 8.969547738693469e-05,
-      "loss": 5.8188,
-      "step": 10756
-    },
-    {
-      "epoch": 5.609908735332464,
-      "grad_norm": 1.493955135345459,
-      "learning_rate": 8.969447236180905e-05,
-      "loss": 5.1613,
-      "step": 10757
-    },
-    {
-      "epoch": 5.610430247718384,
-      "grad_norm": 1.6525912284851074,
-      "learning_rate": 8.969346733668343e-05,
-      "loss": 5.8144,
-      "step": 10758
-    },
-    {
-      "epoch": 5.610951760104302,
-      "grad_norm": 1.40420401096344,
-      "learning_rate": 8.969246231155779e-05,
-      "loss": 5.5869,
-      "step": 10759
-    },
-    {
-      "epoch": 5.611473272490222,
-      "grad_norm": 1.5402380228042603,
-      "learning_rate": 8.969145728643217e-05,
-      "loss": 5.3246,
-      "step": 10760
-    },
-    {
-      "epoch": 5.611994784876141,
-      "grad_norm": 1.4089709520339966,
-      "learning_rate": 8.969045226130653e-05,
-      "loss": 5.8276,
-      "step": 10761
-    },
-    {
-      "epoch": 5.61251629726206,
-      "grad_norm": 1.4307835102081299,
-      "learning_rate": 8.968944723618091e-05,
-      "loss": 5.7339,
-      "step": 10762
-    },
-    {
-      "epoch": 5.613037809647979,
-      "grad_norm": 1.4012186527252197,
-      "learning_rate": 8.968844221105528e-05,
-      "loss": 5.5585,
-      "step": 10763
-    },
-    {
-      "epoch": 5.613559322033899,
-      "grad_norm": 1.3981788158416748,
-      "learning_rate": 8.968743718592965e-05,
-      "loss": 6.0062,
-      "step": 10764
-    },
-    {
-      "epoch": 5.614080834419817,
-      "grad_norm": 1.535621166229248,
-      "learning_rate": 8.968643216080402e-05,
-      "loss": 5.3075,
-      "step": 10765
-    },
-    {
-      "epoch": 5.614602346805737,
-      "grad_norm": 1.3954375982284546,
-      "learning_rate": 8.96854271356784e-05,
-      "loss": 5.8018,
-      "step": 10766
-    },
-    {
-      "epoch": 5.615123859191656,
-      "grad_norm": 1.4822372198104858,
-      "learning_rate": 8.968442211055277e-05,
-      "loss": 5.6684,
-      "step": 10767
-    },
-    {
-      "epoch": 5.615645371577575,
-      "grad_norm": 1.35394287109375,
-      "learning_rate": 8.968341708542714e-05,
-      "loss": 5.7385,
-      "step": 10768
-    },
-    {
-      "epoch": 5.616166883963494,
-      "grad_norm": 1.413093090057373,
-      "learning_rate": 8.968241206030152e-05,
-      "loss": 5.8517,
-      "step": 10769
-    },
-    {
-      "epoch": 5.616688396349414,
-      "grad_norm": 1.531387209892273,
-      "learning_rate": 8.968140703517588e-05,
-      "loss": 5.2694,
-      "step": 10770
-    },
-    {
-      "epoch": 5.617209908735332,
-      "grad_norm": 1.4426822662353516,
-      "learning_rate": 8.968040201005026e-05,
-      "loss": 5.9047,
-      "step": 10771
-    },
-    {
-      "epoch": 5.617731421121252,
-      "grad_norm": 1.2957112789154053,
-      "learning_rate": 8.967939698492462e-05,
-      "loss": 5.0518,
-      "step": 10772
-    },
-    {
-      "epoch": 5.618252933507171,
-      "grad_norm": 1.6042314767837524,
-      "learning_rate": 8.9678391959799e-05,
-      "loss": 5.7134,
-      "step": 10773
-    },
-    {
-      "epoch": 5.61877444589309,
-      "grad_norm": 1.4272891283035278,
-      "learning_rate": 8.967738693467336e-05,
-      "loss": 6.0769,
-      "step": 10774
-    },
-    {
-      "epoch": 5.619295958279009,
-      "grad_norm": 1.5628610849380493,
-      "learning_rate": 8.967638190954774e-05,
-      "loss": 5.6745,
-      "step": 10775
-    },
-    {
-      "epoch": 5.619817470664929,
-      "grad_norm": 1.3754807710647583,
-      "learning_rate": 8.967537688442212e-05,
-      "loss": 5.8278,
-      "step": 10776
-    },
-    {
-      "epoch": 5.620338983050847,
-      "grad_norm": 1.4431676864624023,
-      "learning_rate": 8.96743718592965e-05,
-      "loss": 5.7195,
-      "step": 10777
-    },
-    {
-      "epoch": 5.620860495436767,
-      "grad_norm": 1.5327861309051514,
-      "learning_rate": 8.967336683417086e-05,
-      "loss": 5.4871,
-      "step": 10778
-    },
-    {
-      "epoch": 5.621382007822686,
-      "grad_norm": 1.5918577909469604,
-      "learning_rate": 8.967236180904524e-05,
-      "loss": 5.5217,
-      "step": 10779
-    },
-    {
-      "epoch": 5.621903520208605,
-      "grad_norm": 1.4155633449554443,
-      "learning_rate": 8.96713567839196e-05,
-      "loss": 5.6386,
-      "step": 10780
-    },
-    {
-      "epoch": 5.622425032594524,
-      "grad_norm": 1.5881943702697754,
-      "learning_rate": 8.967035175879397e-05,
-      "loss": 5.9649,
-      "step": 10781
-    },
-    {
-      "epoch": 5.622946544980444,
-      "grad_norm": 1.3059442043304443,
-      "learning_rate": 8.966934673366835e-05,
-      "loss": 5.9676,
-      "step": 10782
-    },
-    {
-      "epoch": 5.623468057366362,
-      "grad_norm": 1.6797270774841309,
-      "learning_rate": 8.966834170854271e-05,
-      "loss": 5.5308,
-      "step": 10783
-    },
-    {
-      "epoch": 5.623989569752282,
-      "grad_norm": 1.5364915132522583,
-      "learning_rate": 8.966733668341709e-05,
-      "loss": 5.4937,
-      "step": 10784
-    },
-    {
-      "epoch": 5.624511082138201,
-      "grad_norm": 1.6600849628448486,
-      "learning_rate": 8.966633165829145e-05,
-      "loss": 5.5003,
-      "step": 10785
-    },
-    {
-      "epoch": 5.62503259452412,
-      "grad_norm": 1.4681226015090942,
-      "learning_rate": 8.966532663316583e-05,
-      "loss": 5.872,
-      "step": 10786
-    },
-    {
-      "epoch": 5.625554106910039,
-      "grad_norm": 1.4834232330322266,
-      "learning_rate": 8.966432160804021e-05,
-      "loss": 5.0439,
-      "step": 10787
-    },
-    {
-      "epoch": 5.626075619295959,
-      "grad_norm": 1.5923528671264648,
-      "learning_rate": 8.966331658291459e-05,
-      "loss": 5.4792,
-      "step": 10788
-    },
-    {
-      "epoch": 5.626597131681877,
-      "grad_norm": 1.408738613128662,
-      "learning_rate": 8.966231155778895e-05,
-      "loss": 5.4845,
-      "step": 10789
-    },
-    {
-      "epoch": 5.627118644067797,
-      "grad_norm": 1.4079418182373047,
-      "learning_rate": 8.966130653266333e-05,
-      "loss": 5.9044,
-      "step": 10790
-    },
-    {
-      "epoch": 5.627640156453715,
-      "grad_norm": 1.492214322090149,
-      "learning_rate": 8.966030150753769e-05,
-      "loss": 5.3748,
-      "step": 10791
-    },
-    {
-      "epoch": 5.628161668839635,
-      "grad_norm": 1.4931862354278564,
-      "learning_rate": 8.965929648241207e-05,
-      "loss": 5.4204,
-      "step": 10792
-    },
-    {
-      "epoch": 5.628683181225554,
-      "grad_norm": 1.6052953004837036,
-      "learning_rate": 8.965829145728643e-05,
-      "loss": 4.7524,
-      "step": 10793
-    },
-    {
-      "epoch": 5.629204693611474,
-      "grad_norm": 1.421484112739563,
-      "learning_rate": 8.96572864321608e-05,
-      "loss": 5.9184,
-      "step": 10794
-    },
-    {
-      "epoch": 5.629726205997392,
-      "grad_norm": 1.387969970703125,
-      "learning_rate": 8.965628140703518e-05,
-      "loss": 5.8504,
-      "step": 10795
-    },
-    {
-      "epoch": 5.630247718383312,
-      "grad_norm": 1.38021981716156,
-      "learning_rate": 8.965527638190955e-05,
-      "loss": 5.9066,
-      "step": 10796
-    },
-    {
-      "epoch": 5.63076923076923,
-      "grad_norm": 1.6067055463790894,
-      "learning_rate": 8.965427135678393e-05,
-      "loss": 5.2257,
-      "step": 10797
-    },
-    {
-      "epoch": 5.63129074315515,
-      "grad_norm": 1.5525658130645752,
-      "learning_rate": 8.96532663316583e-05,
-      "loss": 5.0937,
-      "step": 10798
-    },
-    {
-      "epoch": 5.631812255541069,
-      "grad_norm": 1.4380319118499756,
-      "learning_rate": 8.965226130653267e-05,
-      "loss": 5.514,
-      "step": 10799
-    },
-    {
-      "epoch": 5.632333767926989,
-      "grad_norm": 1.5950044393539429,
-      "learning_rate": 8.965125628140704e-05,
-      "loss": 5.2153,
-      "step": 10800
-    },
-    {
-      "epoch": 5.632855280312907,
-      "grad_norm": 1.7496587038040161,
-      "learning_rate": 8.965025125628142e-05,
-      "loss": 5.4496,
-      "step": 10801
-    },
-    {
-      "epoch": 5.633376792698827,
-      "grad_norm": 1.4401425123214722,
-      "learning_rate": 8.964924623115578e-05,
-      "loss": 5.7522,
-      "step": 10802
-    },
-    {
-      "epoch": 5.633898305084745,
-      "grad_norm": 1.6351449489593506,
-      "learning_rate": 8.964824120603016e-05,
-      "loss": 5.2958,
-      "step": 10803
-    },
-    {
-      "epoch": 5.634419817470665,
-      "grad_norm": 1.4320244789123535,
-      "learning_rate": 8.964723618090452e-05,
-      "loss": 5.8712,
-      "step": 10804
-    },
-    {
-      "epoch": 5.634941329856584,
-      "grad_norm": 1.5549267530441284,
-      "learning_rate": 8.96462311557789e-05,
-      "loss": 5.6277,
-      "step": 10805
-    },
-    {
-      "epoch": 5.635462842242504,
-      "grad_norm": 1.5231528282165527,
-      "learning_rate": 8.964522613065326e-05,
-      "loss": 5.6973,
-      "step": 10806
-    },
-    {
-      "epoch": 5.635984354628422,
-      "grad_norm": 1.475305438041687,
-      "learning_rate": 8.964422110552764e-05,
-      "loss": 5.8017,
-      "step": 10807
-    },
-    {
-      "epoch": 5.636505867014342,
-      "grad_norm": 1.4046663045883179,
-      "learning_rate": 8.964321608040202e-05,
-      "loss": 5.5645,
-      "step": 10808
-    },
-    {
-      "epoch": 5.63702737940026,
-      "grad_norm": 1.4886283874511719,
-      "learning_rate": 8.964221105527638e-05,
-      "loss": 4.701,
-      "step": 10809
-    },
-    {
-      "epoch": 5.63754889178618,
-      "grad_norm": 1.4730623960494995,
-      "learning_rate": 8.964120603015076e-05,
-      "loss": 5.4826,
-      "step": 10810
-    },
-    {
-      "epoch": 5.638070404172099,
-      "grad_norm": 1.5099440813064575,
-      "learning_rate": 8.964020100502513e-05,
-      "loss": 5.2469,
-      "step": 10811
-    },
-    {
-      "epoch": 5.638591916558019,
-      "grad_norm": 1.4545295238494873,
-      "learning_rate": 8.96391959798995e-05,
-      "loss": 5.9159,
-      "step": 10812
-    },
-    {
-      "epoch": 5.639113428943937,
-      "grad_norm": 1.5190812349319458,
-      "learning_rate": 8.963819095477387e-05,
-      "loss": 5.6385,
-      "step": 10813
-    },
-    {
-      "epoch": 5.639634941329857,
-      "grad_norm": 1.3853734731674194,
-      "learning_rate": 8.963718592964825e-05,
-      "loss": 5.4425,
-      "step": 10814
-    },
-    {
-      "epoch": 5.640156453715775,
-      "grad_norm": 1.6198279857635498,
-      "learning_rate": 8.963618090452261e-05,
-      "loss": 5.4611,
-      "step": 10815
-    },
-    {
-      "epoch": 5.640677966101695,
-      "grad_norm": 1.6030395030975342,
-      "learning_rate": 8.963517587939699e-05,
-      "loss": 5.0929,
-      "step": 10816
-    },
-    {
-      "epoch": 5.641199478487614,
-      "grad_norm": 1.7084485292434692,
-      "learning_rate": 8.963417085427135e-05,
-      "loss": 4.9479,
-      "step": 10817
-    },
-    {
-      "epoch": 5.641720990873534,
-      "grad_norm": 1.5259308815002441,
-      "learning_rate": 8.963316582914573e-05,
-      "loss": 5.9477,
-      "step": 10818
-    },
-    {
-      "epoch": 5.642242503259452,
-      "grad_norm": 1.4764087200164795,
-      "learning_rate": 8.963216080402011e-05,
-      "loss": 5.5445,
-      "step": 10819
-    },
-    {
-      "epoch": 5.642764015645372,
-      "grad_norm": 1.4806472063064575,
-      "learning_rate": 8.963115577889447e-05,
-      "loss": 5.6853,
-      "step": 10820
-    },
-    {
-      "epoch": 5.64328552803129,
-      "grad_norm": 1.3320348262786865,
-      "learning_rate": 8.963015075376885e-05,
-      "loss": 5.7958,
-      "step": 10821
-    },
-    {
-      "epoch": 5.64380704041721,
-      "grad_norm": 1.4778175354003906,
-      "learning_rate": 8.962914572864322e-05,
-      "loss": 5.4333,
-      "step": 10822
-    },
-    {
-      "epoch": 5.644328552803129,
-      "grad_norm": 1.459382176399231,
-      "learning_rate": 8.962814070351759e-05,
-      "loss": 5.5178,
-      "step": 10823
-    },
-    {
-      "epoch": 5.644850065189048,
-      "grad_norm": 1.4744855165481567,
-      "learning_rate": 8.962713567839196e-05,
-      "loss": 5.7683,
-      "step": 10824
-    },
-    {
-      "epoch": 5.645371577574967,
-      "grad_norm": 1.4650914669036865,
-      "learning_rate": 8.962613065326634e-05,
-      "loss": 5.8183,
-      "step": 10825
-    },
-    {
-      "epoch": 5.645893089960887,
-      "grad_norm": 1.6759175062179565,
-      "learning_rate": 8.96251256281407e-05,
-      "loss": 5.5627,
-      "step": 10826
-    },
-    {
-      "epoch": 5.646414602346805,
-      "grad_norm": 1.444706916809082,
-      "learning_rate": 8.962412060301508e-05,
-      "loss": 5.8642,
-      "step": 10827
-    },
-    {
-      "epoch": 5.646936114732725,
-      "grad_norm": 1.4047213792800903,
-      "learning_rate": 8.962311557788945e-05,
-      "loss": 5.4089,
-      "step": 10828
-    },
-    {
-      "epoch": 5.647457627118644,
-      "grad_norm": 1.5012317895889282,
-      "learning_rate": 8.962211055276383e-05,
-      "loss": 5.6848,
-      "step": 10829
-    },
-    {
-      "epoch": 5.647979139504563,
-      "grad_norm": 1.4350608587265015,
-      "learning_rate": 8.96211055276382e-05,
-      "loss": 5.6241,
-      "step": 10830
-    },
-    {
-      "epoch": 5.648500651890482,
-      "grad_norm": 1.3215248584747314,
-      "learning_rate": 8.962010050251257e-05,
-      "loss": 5.7633,
-      "step": 10831
-    },
-    {
-      "epoch": 5.649022164276402,
-      "grad_norm": 1.6051334142684937,
-      "learning_rate": 8.961909547738694e-05,
-      "loss": 5.2729,
-      "step": 10832
-    },
-    {
-      "epoch": 5.64954367666232,
-      "grad_norm": 1.459229826927185,
-      "learning_rate": 8.961809045226132e-05,
-      "loss": 5.4403,
-      "step": 10833
-    },
-    {
-      "epoch": 5.65006518904824,
-      "grad_norm": 1.3123656511306763,
-      "learning_rate": 8.961708542713568e-05,
-      "loss": 5.1454,
-      "step": 10834
-    },
-    {
-      "epoch": 5.650586701434159,
-      "grad_norm": 1.4791381359100342,
-      "learning_rate": 8.961608040201005e-05,
-      "loss": 5.4484,
-      "step": 10835
-    },
-    {
-      "epoch": 5.651108213820078,
-      "grad_norm": 1.4722166061401367,
-      "learning_rate": 8.961507537688442e-05,
-      "loss": 5.4197,
-      "step": 10836
-    },
-    {
-      "epoch": 5.651629726205997,
-      "grad_norm": 1.3411507606506348,
-      "learning_rate": 8.961407035175879e-05,
-      "loss": 5.8679,
-      "step": 10837
-    },
-    {
-      "epoch": 5.652151238591917,
-      "grad_norm": 1.3844192028045654,
-      "learning_rate": 8.961306532663317e-05,
-      "loss": 5.5692,
-      "step": 10838
-    },
-    {
-      "epoch": 5.652672750977835,
-      "grad_norm": 1.4378823041915894,
-      "learning_rate": 8.961206030150754e-05,
-      "loss": 5.5874,
-      "step": 10839
-    },
-    {
-      "epoch": 5.653194263363755,
-      "grad_norm": 1.4036285877227783,
-      "learning_rate": 8.961105527638192e-05,
-      "loss": 5.6101,
-      "step": 10840
-    },
-    {
-      "epoch": 5.653715775749674,
-      "grad_norm": 1.4930002689361572,
-      "learning_rate": 8.961005025125629e-05,
-      "loss": 5.6957,
-      "step": 10841
-    },
-    {
-      "epoch": 5.654237288135593,
-      "grad_norm": 1.4854040145874023,
-      "learning_rate": 8.960904522613066e-05,
-      "loss": 5.7187,
-      "step": 10842
-    },
-    {
-      "epoch": 5.654758800521512,
-      "grad_norm": 1.4667001962661743,
-      "learning_rate": 8.960804020100503e-05,
-      "loss": 5.8265,
-      "step": 10843
-    },
-    {
-      "epoch": 5.655280312907432,
-      "grad_norm": 1.4470330476760864,
-      "learning_rate": 8.96070351758794e-05,
-      "loss": 5.7775,
-      "step": 10844
-    },
-    {
-      "epoch": 5.65580182529335,
-      "grad_norm": 1.4831807613372803,
-      "learning_rate": 8.960603015075377e-05,
-      "loss": 5.8434,
-      "step": 10845
-    },
-    {
-      "epoch": 5.65632333767927,
-      "grad_norm": 1.838244915008545,
-      "learning_rate": 8.960502512562815e-05,
-      "loss": 5.8163,
-      "step": 10846
-    },
-    {
-      "epoch": 5.656844850065189,
-      "grad_norm": 1.3642475605010986,
-      "learning_rate": 8.960402010050251e-05,
-      "loss": 5.8054,
-      "step": 10847
-    },
-    {
-      "epoch": 5.657366362451108,
-      "grad_norm": 1.3965799808502197,
-      "learning_rate": 8.960301507537689e-05,
-      "loss": 5.5258,
-      "step": 10848
-    },
-    {
-      "epoch": 5.657887874837027,
-      "grad_norm": 1.4310121536254883,
-      "learning_rate": 8.960201005025127e-05,
-      "loss": 5.7902,
-      "step": 10849
-    },
-    {
-      "epoch": 5.658409387222947,
-      "grad_norm": 1.4178639650344849,
-      "learning_rate": 8.960100502512563e-05,
-      "loss": 5.6708,
-      "step": 10850
-    },
-    {
-      "epoch": 5.658930899608865,
-      "grad_norm": 1.403517484664917,
-      "learning_rate": 8.960000000000001e-05,
-      "loss": 6.0024,
-      "step": 10851
-    },
-    {
-      "epoch": 5.659452411994785,
-      "grad_norm": 1.3901526927947998,
-      "learning_rate": 8.959899497487437e-05,
-      "loss": 5.7589,
-      "step": 10852
-    },
-    {
-      "epoch": 5.659973924380704,
-      "grad_norm": 1.5637240409851074,
-      "learning_rate": 8.959798994974875e-05,
-      "loss": 5.8144,
-      "step": 10853
-    },
-    {
-      "epoch": 5.660495436766623,
-      "grad_norm": 1.81907057762146,
-      "learning_rate": 8.959698492462312e-05,
-      "loss": 4.6165,
-      "step": 10854
-    },
-    {
-      "epoch": 5.661016949152542,
-      "grad_norm": 1.438905119895935,
-      "learning_rate": 8.95959798994975e-05,
-      "loss": 5.8098,
-      "step": 10855
-    },
-    {
-      "epoch": 5.661538461538462,
-      "grad_norm": 1.4679975509643555,
-      "learning_rate": 8.959497487437186e-05,
-      "loss": 5.6438,
-      "step": 10856
-    },
-    {
-      "epoch": 5.66205997392438,
-      "grad_norm": 1.4496254920959473,
-      "learning_rate": 8.959396984924624e-05,
-      "loss": 5.8134,
-      "step": 10857
-    },
-    {
-      "epoch": 5.6625814863103,
-      "grad_norm": 1.461247205734253,
-      "learning_rate": 8.95929648241206e-05,
-      "loss": 5.7992,
-      "step": 10858
-    },
-    {
-      "epoch": 5.663102998696219,
-      "grad_norm": 1.4455519914627075,
-      "learning_rate": 8.959195979899498e-05,
-      "loss": 5.777,
-      "step": 10859
-    },
-    {
-      "epoch": 5.663624511082138,
-      "grad_norm": 1.4913793802261353,
-      "learning_rate": 8.959095477386936e-05,
-      "loss": 5.6603,
-      "step": 10860
-    },
-    {
-      "epoch": 5.664146023468057,
-      "grad_norm": 1.4361423254013062,
-      "learning_rate": 8.958994974874372e-05,
-      "loss": 5.5089,
-      "step": 10861
-    },
-    {
-      "epoch": 5.664667535853977,
-      "grad_norm": 1.4435937404632568,
-      "learning_rate": 8.95889447236181e-05,
-      "loss": 5.3958,
-      "step": 10862
-    },
-    {
-      "epoch": 5.665189048239895,
-      "grad_norm": 1.5203258991241455,
-      "learning_rate": 8.958793969849246e-05,
-      "loss": 4.9937,
-      "step": 10863
-    },
-    {
-      "epoch": 5.665710560625815,
-      "grad_norm": 1.4278147220611572,
-      "learning_rate": 8.958693467336684e-05,
-      "loss": 5.3512,
-      "step": 10864
-    },
-    {
-      "epoch": 5.666232073011734,
-      "grad_norm": 1.4207921028137207,
-      "learning_rate": 8.95859296482412e-05,
-      "loss": 5.3578,
-      "step": 10865
-    },
-    {
-      "epoch": 5.666753585397653,
-      "grad_norm": 1.4400840997695923,
-      "learning_rate": 8.958492462311558e-05,
-      "loss": 5.4376,
-      "step": 10866
-    },
-    {
-      "epoch": 5.667275097783572,
-      "grad_norm": 1.3464784622192383,
-      "learning_rate": 8.958391959798995e-05,
-      "loss": 5.9813,
-      "step": 10867
-    },
-    {
-      "epoch": 5.667796610169492,
-      "grad_norm": 1.3826415538787842,
-      "learning_rate": 8.958291457286432e-05,
-      "loss": 5.6781,
-      "step": 10868
-    },
-    {
-      "epoch": 5.6683181225554105,
-      "grad_norm": 1.3927260637283325,
-      "learning_rate": 8.95819095477387e-05,
-      "loss": 5.8858,
-      "step": 10869
-    },
-    {
-      "epoch": 5.66883963494133,
-      "grad_norm": 1.358038306236267,
-      "learning_rate": 8.958090452261308e-05,
-      "loss": 5.9488,
-      "step": 10870
-    },
-    {
-      "epoch": 5.669361147327249,
-      "grad_norm": 1.544036865234375,
-      "learning_rate": 8.957989949748744e-05,
-      "loss": 5.5982,
-      "step": 10871
-    },
-    {
-      "epoch": 5.669882659713168,
-      "grad_norm": 1.4062224626541138,
-      "learning_rate": 8.957889447236182e-05,
-      "loss": 5.4303,
-      "step": 10872
-    },
-    {
-      "epoch": 5.670404172099087,
-      "grad_norm": 1.447485327720642,
-      "learning_rate": 8.957788944723619e-05,
-      "loss": 5.4872,
-      "step": 10873
-    },
-    {
-      "epoch": 5.670925684485007,
-      "grad_norm": 1.4075819253921509,
-      "learning_rate": 8.957688442211055e-05,
-      "loss": 5.7405,
-      "step": 10874
-    },
-    {
-      "epoch": 5.6714471968709255,
-      "grad_norm": 1.404503345489502,
-      "learning_rate": 8.957587939698493e-05,
-      "loss": 5.9383,
-      "step": 10875
-    },
-    {
-      "epoch": 5.671968709256845,
-      "grad_norm": 1.4012337923049927,
-      "learning_rate": 8.957487437185929e-05,
-      "loss": 5.4766,
-      "step": 10876
-    },
-    {
-      "epoch": 5.672490221642764,
-      "grad_norm": 1.7492698431015015,
-      "learning_rate": 8.957386934673367e-05,
-      "loss": 5.4612,
-      "step": 10877
-    },
-    {
-      "epoch": 5.673011734028683,
-      "grad_norm": 1.3542150259017944,
-      "learning_rate": 8.957286432160803e-05,
-      "loss": 5.8676,
-      "step": 10878
-    },
-    {
-      "epoch": 5.673533246414602,
-      "grad_norm": 1.6544008255004883,
-      "learning_rate": 8.957185929648241e-05,
-      "loss": 5.1669,
-      "step": 10879
-    },
-    {
-      "epoch": 5.674054758800521,
-      "grad_norm": 1.485646367073059,
-      "learning_rate": 8.957085427135679e-05,
-      "loss": 5.9936,
-      "step": 10880
-    },
-    {
-      "epoch": 5.6745762711864405,
-      "grad_norm": 1.6065763235092163,
-      "learning_rate": 8.956984924623117e-05,
-      "loss": 5.2191,
-      "step": 10881
-    },
-    {
-      "epoch": 5.67509778357236,
-      "grad_norm": 1.42881441116333,
-      "learning_rate": 8.956884422110553e-05,
-      "loss": 5.903,
-      "step": 10882
-    },
-    {
-      "epoch": 5.675619295958279,
-      "grad_norm": 1.604276418685913,
-      "learning_rate": 8.956783919597991e-05,
-      "loss": 5.742,
-      "step": 10883
-    },
-    {
-      "epoch": 5.676140808344198,
-      "grad_norm": 1.4211664199829102,
-      "learning_rate": 8.956683417085427e-05,
-      "loss": 5.7315,
-      "step": 10884
-    },
-    {
-      "epoch": 5.6766623207301175,
-      "grad_norm": 1.5545965433120728,
-      "learning_rate": 8.956582914572865e-05,
-      "loss": 5.7242,
-      "step": 10885
-    },
-    {
-      "epoch": 5.677183833116036,
-      "grad_norm": 1.5674493312835693,
-      "learning_rate": 8.956482412060302e-05,
-      "loss": 5.8122,
-      "step": 10886
-    },
-    {
-      "epoch": 5.6777053455019555,
-      "grad_norm": 1.4700905084609985,
-      "learning_rate": 8.956381909547738e-05,
-      "loss": 5.8799,
-      "step": 10887
-    },
-    {
-      "epoch": 5.678226857887875,
-      "grad_norm": 1.5269750356674194,
-      "learning_rate": 8.956281407035176e-05,
-      "loss": 5.5959,
-      "step": 10888
-    },
-    {
-      "epoch": 5.678748370273794,
-      "grad_norm": 1.5720492601394653,
-      "learning_rate": 8.956180904522614e-05,
-      "loss": 5.4541,
-      "step": 10889
-    },
-    {
-      "epoch": 5.679269882659713,
-      "grad_norm": 1.4046553373336792,
-      "learning_rate": 8.956080402010051e-05,
-      "loss": 5.6894,
-      "step": 10890
-    },
-    {
-      "epoch": 5.6797913950456325,
-      "grad_norm": 1.5193123817443848,
-      "learning_rate": 8.955979899497488e-05,
-      "loss": 5.3373,
-      "step": 10891
-    },
-    {
-      "epoch": 5.680312907431551,
-      "grad_norm": 1.3828458786010742,
-      "learning_rate": 8.955879396984926e-05,
-      "loss": 5.8327,
-      "step": 10892
-    },
-    {
-      "epoch": 5.6808344198174705,
-      "grad_norm": 1.5247118473052979,
-      "learning_rate": 8.955778894472362e-05,
-      "loss": 5.4157,
-      "step": 10893
-    },
-    {
-      "epoch": 5.68135593220339,
-      "grad_norm": 1.5297023057937622,
-      "learning_rate": 8.9556783919598e-05,
-      "loss": 5.6032,
-      "step": 10894
-    },
-    {
-      "epoch": 5.681877444589309,
-      "grad_norm": 1.4664528369903564,
-      "learning_rate": 8.955577889447236e-05,
-      "loss": 5.4147,
-      "step": 10895
-    },
-    {
-      "epoch": 5.682398956975228,
-      "grad_norm": 1.573445439338684,
-      "learning_rate": 8.955477386934674e-05,
-      "loss": 5.0414,
-      "step": 10896
-    },
-    {
-      "epoch": 5.6829204693611475,
-      "grad_norm": 1.6101837158203125,
-      "learning_rate": 8.95537688442211e-05,
-      "loss": 5.188,
-      "step": 10897
-    },
-    {
-      "epoch": 5.683441981747066,
-      "grad_norm": 1.4774879217147827,
-      "learning_rate": 8.955276381909548e-05,
-      "loss": 5.5439,
-      "step": 10898
-    },
-    {
-      "epoch": 5.6839634941329855,
-      "grad_norm": 1.341994285583496,
-      "learning_rate": 8.955175879396985e-05,
-      "loss": 6.0121,
-      "step": 10899
-    },
-    {
-      "epoch": 5.684485006518905,
-      "grad_norm": 1.5684735774993896,
-      "learning_rate": 8.955075376884422e-05,
-      "loss": 5.1122,
-      "step": 10900
-    },
-    {
-      "epoch": 5.6850065189048244,
-      "grad_norm": 1.5068870782852173,
-      "learning_rate": 8.95497487437186e-05,
-      "loss": 5.4564,
-      "step": 10901
-    },
-    {
-      "epoch": 5.685528031290743,
-      "grad_norm": 1.398180365562439,
-      "learning_rate": 8.954874371859297e-05,
-      "loss": 5.2319,
-      "step": 10902
-    },
-    {
-      "epoch": 5.6860495436766625,
-      "grad_norm": 1.425860047340393,
-      "learning_rate": 8.954773869346734e-05,
-      "loss": 5.4307,
-      "step": 10903
-    },
-    {
-      "epoch": 5.686571056062581,
-      "grad_norm": 1.578149676322937,
-      "learning_rate": 8.954673366834171e-05,
-      "loss": 5.4038,
-      "step": 10904
-    },
-    {
-      "epoch": 5.6870925684485005,
-      "grad_norm": 1.5110795497894287,
-      "learning_rate": 8.954572864321609e-05,
-      "loss": 5.1053,
-      "step": 10905
-    },
-    {
-      "epoch": 5.68761408083442,
-      "grad_norm": 1.4004648923873901,
-      "learning_rate": 8.954472361809045e-05,
-      "loss": 5.8621,
-      "step": 10906
-    },
-    {
-      "epoch": 5.6881355932203395,
-      "grad_norm": 1.4452509880065918,
-      "learning_rate": 8.954371859296483e-05,
-      "loss": 5.6713,
-      "step": 10907
-    },
-    {
-      "epoch": 5.688657105606258,
-      "grad_norm": 1.3804337978363037,
-      "learning_rate": 8.95427135678392e-05,
-      "loss": 5.801,
-      "step": 10908
-    },
-    {
-      "epoch": 5.6891786179921775,
-      "grad_norm": 1.4068528413772583,
-      "learning_rate": 8.954170854271357e-05,
-      "loss": 5.6451,
-      "step": 10909
-    },
-    {
-      "epoch": 5.689700130378096,
-      "grad_norm": 1.4152898788452148,
-      "learning_rate": 8.954070351758795e-05,
-      "loss": 5.5171,
-      "step": 10910
-    },
-    {
-      "epoch": 5.6902216427640155,
-      "grad_norm": 1.4434278011322021,
-      "learning_rate": 8.953969849246233e-05,
-      "loss": 5.7044,
-      "step": 10911
-    },
-    {
-      "epoch": 5.690743155149935,
-      "grad_norm": 1.3902643918991089,
-      "learning_rate": 8.953869346733669e-05,
-      "loss": 5.7079,
-      "step": 10912
-    },
-    {
-      "epoch": 5.6912646675358545,
-      "grad_norm": 1.4508655071258545,
-      "learning_rate": 8.953768844221107e-05,
-      "loss": 5.7864,
-      "step": 10913
-    },
-    {
-      "epoch": 5.691786179921773,
-      "grad_norm": 1.452014446258545,
-      "learning_rate": 8.953668341708543e-05,
-      "loss": 5.5031,
-      "step": 10914
-    },
-    {
-      "epoch": 5.6923076923076925,
-      "grad_norm": 1.376511812210083,
-      "learning_rate": 8.95356783919598e-05,
-      "loss": 5.7767,
-      "step": 10915
-    },
-    {
-      "epoch": 5.692829204693611,
-      "grad_norm": 1.4678125381469727,
-      "learning_rate": 8.953467336683418e-05,
-      "loss": 5.5009,
-      "step": 10916
-    },
-    {
-      "epoch": 5.6933507170795306,
-      "grad_norm": 1.4048317670822144,
-      "learning_rate": 8.953366834170854e-05,
-      "loss": 5.8665,
-      "step": 10917
-    },
-    {
-      "epoch": 5.69387222946545,
-      "grad_norm": 1.605446219444275,
-      "learning_rate": 8.953266331658292e-05,
-      "loss": 5.5289,
-      "step": 10918
-    },
-    {
-      "epoch": 5.694393741851369,
-      "grad_norm": 1.4213716983795166,
-      "learning_rate": 8.953165829145728e-05,
-      "loss": 5.7146,
-      "step": 10919
-    },
-    {
-      "epoch": 5.694915254237288,
-      "grad_norm": 1.523597240447998,
-      "learning_rate": 8.953065326633166e-05,
-      "loss": 5.691,
-      "step": 10920
-    },
-    {
-      "epoch": 5.6954367666232075,
-      "grad_norm": 1.4171901941299438,
-      "learning_rate": 8.952964824120604e-05,
-      "loss": 5.7422,
-      "step": 10921
-    },
-    {
-      "epoch": 5.695958279009126,
-      "grad_norm": 1.5542359352111816,
-      "learning_rate": 8.952864321608042e-05,
-      "loss": 5.2608,
-      "step": 10922
-    },
-    {
-      "epoch": 5.696479791395046,
-      "grad_norm": 1.3167171478271484,
-      "learning_rate": 8.952763819095478e-05,
-      "loss": 5.9786,
-      "step": 10923
-    },
-    {
-      "epoch": 5.697001303780965,
-      "grad_norm": 1.7536647319793701,
-      "learning_rate": 8.952663316582916e-05,
-      "loss": 5.6015,
-      "step": 10924
-    },
-    {
-      "epoch": 5.697522816166884,
-      "grad_norm": 1.4066106081008911,
-      "learning_rate": 8.952562814070352e-05,
-      "loss": 5.1558,
-      "step": 10925
-    },
-    {
-      "epoch": 5.698044328552803,
-      "grad_norm": 1.5758917331695557,
-      "learning_rate": 8.95246231155779e-05,
-      "loss": 5.6382,
-      "step": 10926
-    },
-    {
-      "epoch": 5.6985658409387225,
-      "grad_norm": 1.3920670747756958,
-      "learning_rate": 8.952361809045226e-05,
-      "loss": 5.924,
-      "step": 10927
-    },
-    {
-      "epoch": 5.699087353324641,
-      "grad_norm": 1.3939228057861328,
-      "learning_rate": 8.952261306532663e-05,
-      "loss": 5.4202,
-      "step": 10928
-    },
-    {
-      "epoch": 5.699608865710561,
-      "grad_norm": 1.463529348373413,
-      "learning_rate": 8.9521608040201e-05,
-      "loss": 5.4931,
-      "step": 10929
-    },
-    {
-      "epoch": 5.70013037809648,
-      "grad_norm": 1.5447313785552979,
-      "learning_rate": 8.952060301507538e-05,
-      "loss": 5.8251,
-      "step": 10930
-    },
-    {
-      "epoch": 5.700651890482399,
-      "grad_norm": 1.4554716348648071,
-      "learning_rate": 8.951959798994976e-05,
-      "loss": 5.4079,
-      "step": 10931
-    },
-    {
-      "epoch": 5.701173402868318,
-      "grad_norm": 1.4944186210632324,
-      "learning_rate": 8.951859296482413e-05,
-      "loss": 5.5237,
-      "step": 10932
-    },
-    {
-      "epoch": 5.7016949152542376,
-      "grad_norm": 1.6450525522232056,
-      "learning_rate": 8.95175879396985e-05,
-      "loss": 5.4721,
-      "step": 10933
-    },
-    {
-      "epoch": 5.702216427640156,
-      "grad_norm": 1.318076729774475,
-      "learning_rate": 8.951658291457287e-05,
-      "loss": 5.9095,
-      "step": 10934
-    },
-    {
-      "epoch": 5.702737940026076,
-      "grad_norm": 1.4043376445770264,
-      "learning_rate": 8.951557788944725e-05,
-      "loss": 5.7314,
-      "step": 10935
-    },
-    {
-      "epoch": 5.703259452411995,
-      "grad_norm": 1.3855525255203247,
-      "learning_rate": 8.951457286432161e-05,
-      "loss": 5.9091,
-      "step": 10936
-    },
-    {
-      "epoch": 5.703780964797914,
-      "grad_norm": 1.3910993337631226,
-      "learning_rate": 8.951356783919599e-05,
-      "loss": 5.6053,
-      "step": 10937
-    },
-    {
-      "epoch": 5.704302477183833,
-      "grad_norm": 1.4748504161834717,
-      "learning_rate": 8.951256281407035e-05,
-      "loss": 5.7609,
-      "step": 10938
-    },
-    {
-      "epoch": 5.704823989569753,
-      "grad_norm": 1.404198169708252,
-      "learning_rate": 8.951155778894473e-05,
-      "loss": 5.9651,
-      "step": 10939
-    },
-    {
-      "epoch": 5.705345501955671,
-      "grad_norm": 1.5238829851150513,
-      "learning_rate": 8.95105527638191e-05,
-      "loss": 5.2186,
-      "step": 10940
-    },
-    {
-      "epoch": 5.705867014341591,
-      "grad_norm": 1.435915470123291,
-      "learning_rate": 8.950954773869347e-05,
-      "loss": 5.5549,
-      "step": 10941
-    },
-    {
-      "epoch": 5.70638852672751,
-      "grad_norm": 1.5584204196929932,
-      "learning_rate": 8.950854271356785e-05,
-      "loss": 5.1931,
-      "step": 10942
-    },
-    {
-      "epoch": 5.706910039113429,
-      "grad_norm": 1.3807698488235474,
-      "learning_rate": 8.950753768844221e-05,
-      "loss": 5.2595,
-      "step": 10943
-    },
-    {
-      "epoch": 5.707431551499348,
-      "grad_norm": 1.4175143241882324,
-      "learning_rate": 8.950653266331659e-05,
-      "loss": 5.7175,
-      "step": 10944
-    },
-    {
-      "epoch": 5.707953063885268,
-      "grad_norm": 1.3361456394195557,
-      "learning_rate": 8.950552763819096e-05,
-      "loss": 5.734,
-      "step": 10945
-    },
-    {
-      "epoch": 5.708474576271186,
-      "grad_norm": 1.260603666305542,
-      "learning_rate": 8.950452261306533e-05,
-      "loss": 5.6654,
-      "step": 10946
-    },
-    {
-      "epoch": 5.708996088657106,
-      "grad_norm": 2.0194332599639893,
-      "learning_rate": 8.95035175879397e-05,
-      "loss": 5.0596,
-      "step": 10947
-    },
-    {
-      "epoch": 5.709517601043025,
-      "grad_norm": 1.5251694917678833,
-      "learning_rate": 8.950251256281408e-05,
-      "loss": 5.2217,
-      "step": 10948
-    },
-    {
-      "epoch": 5.710039113428944,
-      "grad_norm": 1.3567287921905518,
-      "learning_rate": 8.950150753768844e-05,
-      "loss": 5.9272,
-      "step": 10949
-    },
-    {
-      "epoch": 5.710560625814863,
-      "grad_norm": 1.4431544542312622,
-      "learning_rate": 8.950050251256282e-05,
-      "loss": 5.4764,
-      "step": 10950
-    },
-    {
-      "epoch": 5.711082138200783,
-      "grad_norm": 1.408400058746338,
-      "learning_rate": 8.94994974874372e-05,
-      "loss": 6.2005,
-      "step": 10951
-    },
-    {
-      "epoch": 5.711603650586701,
-      "grad_norm": 1.502579927444458,
-      "learning_rate": 8.949849246231157e-05,
-      "loss": 5.8148,
-      "step": 10952
-    },
-    {
-      "epoch": 5.712125162972621,
-      "grad_norm": 1.5935227870941162,
-      "learning_rate": 8.949748743718594e-05,
-      "loss": 5.3022,
-      "step": 10953
-    },
-    {
-      "epoch": 5.71264667535854,
-      "grad_norm": 1.394395112991333,
-      "learning_rate": 8.94964824120603e-05,
-      "loss": 5.3106,
-      "step": 10954
-    },
-    {
-      "epoch": 5.713168187744459,
-      "grad_norm": 1.4722621440887451,
-      "learning_rate": 8.949547738693468e-05,
-      "loss": 5.7117,
-      "step": 10955
-    },
-    {
-      "epoch": 5.713689700130378,
-      "grad_norm": 1.5680347681045532,
-      "learning_rate": 8.949447236180904e-05,
-      "loss": 5.2212,
-      "step": 10956
-    },
-    {
-      "epoch": 5.714211212516298,
-      "grad_norm": 1.603228211402893,
-      "learning_rate": 8.949346733668342e-05,
-      "loss": 5.3674,
-      "step": 10957
-    },
-    {
-      "epoch": 5.714732724902216,
-      "grad_norm": 1.4112343788146973,
-      "learning_rate": 8.949246231155779e-05,
-      "loss": 5.8224,
-      "step": 10958
-    },
-    {
-      "epoch": 5.715254237288136,
-      "grad_norm": 1.4585505723953247,
-      "learning_rate": 8.949145728643216e-05,
-      "loss": 5.7207,
-      "step": 10959
-    },
-    {
-      "epoch": 5.715775749674055,
-      "grad_norm": 1.5137776136398315,
-      "learning_rate": 8.949045226130653e-05,
-      "loss": 5.4966,
-      "step": 10960
-    },
-    {
-      "epoch": 5.716297262059974,
-      "grad_norm": 1.5856624841690063,
-      "learning_rate": 8.94894472361809e-05,
-      "loss": 5.5685,
-      "step": 10961
-    },
-    {
-      "epoch": 5.716818774445893,
-      "grad_norm": 1.4281328916549683,
-      "learning_rate": 8.948844221105528e-05,
-      "loss": 5.3968,
-      "step": 10962
-    },
-    {
-      "epoch": 5.717340286831813,
-      "grad_norm": 1.4963778257369995,
-      "learning_rate": 8.948743718592966e-05,
-      "loss": 5.6856,
-      "step": 10963
-    },
-    {
-      "epoch": 5.717861799217731,
-      "grad_norm": 1.5307120084762573,
-      "learning_rate": 8.948643216080403e-05,
-      "loss": 5.8347,
-      "step": 10964
-    },
-    {
-      "epoch": 5.718383311603651,
-      "grad_norm": 1.4833736419677734,
-      "learning_rate": 8.94854271356784e-05,
-      "loss": 5.399,
-      "step": 10965
-    },
-    {
-      "epoch": 5.71890482398957,
-      "grad_norm": 1.4166479110717773,
-      "learning_rate": 8.948442211055277e-05,
-      "loss": 5.6579,
-      "step": 10966
-    },
-    {
-      "epoch": 5.719426336375489,
-      "grad_norm": 1.4053109884262085,
-      "learning_rate": 8.948341708542713e-05,
-      "loss": 5.7199,
-      "step": 10967
-    },
-    {
-      "epoch": 5.719947848761408,
-      "grad_norm": 1.5429428815841675,
-      "learning_rate": 8.948241206030151e-05,
-      "loss": 5.4244,
-      "step": 10968
-    },
-    {
-      "epoch": 5.720469361147328,
-      "grad_norm": 1.529392957687378,
-      "learning_rate": 8.948140703517587e-05,
-      "loss": 5.0858,
-      "step": 10969
-    },
-    {
-      "epoch": 5.720990873533246,
-      "grad_norm": 1.5008496046066284,
-      "learning_rate": 8.948040201005025e-05,
-      "loss": 5.657,
-      "step": 10970
-    },
-    {
-      "epoch": 5.721512385919166,
-      "grad_norm": 1.4729105234146118,
-      "learning_rate": 8.947939698492463e-05,
-      "loss": 5.9612,
-      "step": 10971
-    },
-    {
-      "epoch": 5.722033898305085,
-      "grad_norm": 1.409668207168579,
-      "learning_rate": 8.947839195979901e-05,
-      "loss": 5.8277,
-      "step": 10972
-    },
-    {
-      "epoch": 5.722555410691004,
-      "grad_norm": 1.4998316764831543,
-      "learning_rate": 8.947738693467337e-05,
-      "loss": 5.8199,
-      "step": 10973
-    },
-    {
-      "epoch": 5.723076923076923,
-      "grad_norm": 1.370540976524353,
-      "learning_rate": 8.947638190954775e-05,
-      "loss": 5.5689,
-      "step": 10974
-    },
-    {
-      "epoch": 5.723598435462842,
-      "grad_norm": 1.4309996366500854,
-      "learning_rate": 8.947537688442211e-05,
-      "loss": 5.6935,
-      "step": 10975
-    },
-    {
-      "epoch": 5.724119947848761,
-      "grad_norm": 1.3925691843032837,
-      "learning_rate": 8.947437185929649e-05,
-      "loss": 5.8133,
-      "step": 10976
-    },
-    {
-      "epoch": 5.724641460234681,
-      "grad_norm": 1.4683897495269775,
-      "learning_rate": 8.947336683417086e-05,
-      "loss": 5.8919,
-      "step": 10977
-    },
-    {
-      "epoch": 5.7251629726206,
-      "grad_norm": 1.4365804195404053,
-      "learning_rate": 8.947236180904523e-05,
-      "loss": 5.5713,
-      "step": 10978
-    },
-    {
-      "epoch": 5.725684485006519,
-      "grad_norm": 1.537802815437317,
-      "learning_rate": 8.94713567839196e-05,
-      "loss": 5.5111,
-      "step": 10979
-    },
-    {
-      "epoch": 5.726205997392438,
-      "grad_norm": 1.5508480072021484,
-      "learning_rate": 8.947035175879396e-05,
-      "loss": 5.7635,
-      "step": 10980
-    },
-    {
-      "epoch": 5.726727509778357,
-      "grad_norm": 1.4963473081588745,
-      "learning_rate": 8.946934673366834e-05,
-      "loss": 5.3067,
-      "step": 10981
-    },
-    {
-      "epoch": 5.727249022164276,
-      "grad_norm": 1.3477096557617188,
-      "learning_rate": 8.946834170854272e-05,
-      "loss": 5.6427,
-      "step": 10982
-    },
-    {
-      "epoch": 5.727770534550196,
-      "grad_norm": 1.5321629047393799,
-      "learning_rate": 8.94673366834171e-05,
-      "loss": 5.6251,
-      "step": 10983
-    },
-    {
-      "epoch": 5.728292046936115,
-      "grad_norm": 1.311721920967102,
-      "learning_rate": 8.946633165829146e-05,
-      "loss": 5.7563,
-      "step": 10984
-    },
-    {
-      "epoch": 5.728813559322034,
-      "grad_norm": 1.5104773044586182,
-      "learning_rate": 8.946532663316584e-05,
-      "loss": 5.5172,
-      "step": 10985
-    },
-    {
-      "epoch": 5.729335071707953,
-      "grad_norm": 1.5570684671401978,
-      "learning_rate": 8.94643216080402e-05,
-      "loss": 5.4071,
-      "step": 10986
-    },
-    {
-      "epoch": 5.729856584093872,
-      "grad_norm": 1.3081616163253784,
-      "learning_rate": 8.946331658291458e-05,
-      "loss": 5.8312,
-      "step": 10987
-    },
-    {
-      "epoch": 5.730378096479791,
-      "grad_norm": 1.4622939825057983,
-      "learning_rate": 8.946231155778895e-05,
-      "loss": 5.4964,
-      "step": 10988
-    },
-    {
-      "epoch": 5.730899608865711,
-      "grad_norm": 1.5187374353408813,
-      "learning_rate": 8.946130653266332e-05,
-      "loss": 5.6395,
-      "step": 10989
-    },
-    {
-      "epoch": 5.73142112125163,
-      "grad_norm": 1.5808544158935547,
-      "learning_rate": 8.946030150753769e-05,
-      "loss": 5.5626,
-      "step": 10990
-    },
-    {
-      "epoch": 5.731942633637549,
-      "grad_norm": 1.4094609022140503,
-      "learning_rate": 8.945929648241207e-05,
-      "loss": 5.7477,
-      "step": 10991
-    },
-    {
-      "epoch": 5.732464146023468,
-      "grad_norm": 1.3323421478271484,
-      "learning_rate": 8.945829145728643e-05,
-      "loss": 5.7479,
-      "step": 10992
-    },
-    {
-      "epoch": 5.732985658409387,
-      "grad_norm": 1.4520392417907715,
-      "learning_rate": 8.945728643216081e-05,
-      "loss": 5.5093,
-      "step": 10993
-    },
-    {
-      "epoch": 5.733507170795306,
-      "grad_norm": 1.4037079811096191,
-      "learning_rate": 8.945628140703519e-05,
-      "loss": 5.6726,
-      "step": 10994
-    },
-    {
-      "epoch": 5.734028683181226,
-      "grad_norm": 1.3451672792434692,
-      "learning_rate": 8.945527638190955e-05,
-      "loss": 5.8028,
-      "step": 10995
-    },
-    {
-      "epoch": 5.734550195567145,
-      "grad_norm": 1.447629690170288,
-      "learning_rate": 8.945427135678393e-05,
-      "loss": 5.7794,
-      "step": 10996
-    },
-    {
-      "epoch": 5.735071707953064,
-      "grad_norm": 1.350425362586975,
-      "learning_rate": 8.945326633165829e-05,
-      "loss": 5.8462,
-      "step": 10997
-    },
-    {
-      "epoch": 5.735593220338983,
-      "grad_norm": 1.5589371919631958,
-      "learning_rate": 8.945226130653267e-05,
-      "loss": 5.1152,
-      "step": 10998
-    },
-    {
-      "epoch": 5.736114732724902,
-      "grad_norm": 1.5589927434921265,
-      "learning_rate": 8.945125628140703e-05,
-      "loss": 5.5062,
-      "step": 10999
-    },
-    {
-      "epoch": 5.736636245110821,
-      "grad_norm": 1.450381875038147,
-      "learning_rate": 8.945025125628141e-05,
-      "loss": 5.7672,
-      "step": 11000
-    },
-    {
-      "epoch": 5.737157757496741,
-      "grad_norm": 1.3628703355789185,
-      "learning_rate": 8.944924623115578e-05,
-      "loss": 5.6845,
-      "step": 11001
-    },
-    {
-      "epoch": 5.73767926988266,
-      "grad_norm": 1.3711543083190918,
-      "learning_rate": 8.944824120603015e-05,
-      "loss": 5.6646,
-      "step": 11002
-    },
-    {
-      "epoch": 5.738200782268579,
-      "grad_norm": 1.4496585130691528,
-      "learning_rate": 8.944723618090453e-05,
-      "loss": 5.2687,
-      "step": 11003
-    },
-    {
-      "epoch": 5.738722294654498,
-      "grad_norm": 1.3922224044799805,
-      "learning_rate": 8.944623115577891e-05,
-      "loss": 5.9829,
-      "step": 11004
-    },
-    {
-      "epoch": 5.739243807040417,
-      "grad_norm": 1.3956485986709595,
-      "learning_rate": 8.944522613065327e-05,
-      "loss": 6.0203,
-      "step": 11005
-    },
-    {
-      "epoch": 5.739765319426336,
-      "grad_norm": 1.4096879959106445,
-      "learning_rate": 8.944422110552765e-05,
-      "loss": 6.0832,
-      "step": 11006
-    },
-    {
-      "epoch": 5.740286831812256,
-      "grad_norm": 1.3465909957885742,
-      "learning_rate": 8.944321608040202e-05,
-      "loss": 5.6256,
-      "step": 11007
-    },
-    {
-      "epoch": 5.740808344198175,
-      "grad_norm": 1.356597900390625,
-      "learning_rate": 8.944221105527638e-05,
-      "loss": 5.4909,
-      "step": 11008
-    },
-    {
-      "epoch": 5.741329856584094,
-      "grad_norm": 1.4740556478500366,
-      "learning_rate": 8.944120603015076e-05,
-      "loss": 5.6433,
-      "step": 11009
-    },
-    {
-      "epoch": 5.741851368970013,
-      "grad_norm": 1.419491171836853,
-      "learning_rate": 8.944020100502512e-05,
-      "loss": 5.7459,
-      "step": 11010
-    },
-    {
-      "epoch": 5.742372881355932,
-      "grad_norm": 1.5163482427597046,
-      "learning_rate": 8.94391959798995e-05,
-      "loss": 5.7211,
-      "step": 11011
-    },
-    {
-      "epoch": 5.742894393741851,
-      "grad_norm": 1.496117353439331,
-      "learning_rate": 8.943819095477386e-05,
-      "loss": 5.5877,
-      "step": 11012
-    },
-    {
-      "epoch": 5.743415906127771,
-      "grad_norm": 1.4984523057937622,
-      "learning_rate": 8.943718592964824e-05,
-      "loss": 5.6667,
-      "step": 11013
-    },
-    {
-      "epoch": 5.743937418513689,
-      "grad_norm": 1.4717193841934204,
-      "learning_rate": 8.943618090452262e-05,
-      "loss": 5.4994,
-      "step": 11014
-    },
-    {
-      "epoch": 5.744458930899609,
-      "grad_norm": 1.6193602085113525,
-      "learning_rate": 8.9435175879397e-05,
-      "loss": 5.395,
-      "step": 11015
-    },
-    {
-      "epoch": 5.744980443285528,
-      "grad_norm": 1.4333332777023315,
-      "learning_rate": 8.943417085427136e-05,
-      "loss": 5.4213,
-      "step": 11016
-    },
-    {
-      "epoch": 5.745501955671447,
-      "grad_norm": 1.5097194910049438,
-      "learning_rate": 8.943316582914574e-05,
-      "loss": 5.9078,
-      "step": 11017
-    },
-    {
-      "epoch": 5.746023468057366,
-      "grad_norm": 1.432334542274475,
-      "learning_rate": 8.94321608040201e-05,
-      "loss": 5.1842,
-      "step": 11018
-    },
-    {
-      "epoch": 5.746544980443286,
-      "grad_norm": 1.4199100732803345,
-      "learning_rate": 8.943115577889448e-05,
-      "loss": 5.8599,
-      "step": 11019
-    },
-    {
-      "epoch": 5.747066492829204,
-      "grad_norm": 1.3694907426834106,
-      "learning_rate": 8.943015075376885e-05,
-      "loss": 5.927,
-      "step": 11020
-    },
-    {
-      "epoch": 5.747588005215124,
-      "grad_norm": 1.545912504196167,
-      "learning_rate": 8.942914572864321e-05,
-      "loss": 5.4135,
-      "step": 11021
-    },
-    {
-      "epoch": 5.748109517601043,
-      "grad_norm": 1.4554182291030884,
-      "learning_rate": 8.942814070351759e-05,
-      "loss": 5.826,
-      "step": 11022
-    },
-    {
-      "epoch": 5.748631029986962,
-      "grad_norm": 1.4468672275543213,
-      "learning_rate": 8.942713567839197e-05,
-      "loss": 5.5431,
-      "step": 11023
-    },
-    {
-      "epoch": 5.749152542372881,
-      "grad_norm": 1.3289719820022583,
-      "learning_rate": 8.942613065326634e-05,
-      "loss": 5.7963,
-      "step": 11024
-    },
-    {
-      "epoch": 5.749674054758801,
-      "grad_norm": 1.5375677347183228,
-      "learning_rate": 8.942512562814071e-05,
-      "loss": 5.4564,
-      "step": 11025
-    },
-    {
-      "epoch": 5.750195567144719,
-      "grad_norm": 1.456183910369873,
-      "learning_rate": 8.942412060301509e-05,
-      "loss": 5.8883,
-      "step": 11026
-    },
-    {
-      "epoch": 5.750717079530639,
-      "grad_norm": 1.474006175994873,
-      "learning_rate": 8.942311557788945e-05,
-      "loss": 5.1966,
-      "step": 11027
-    },
-    {
-      "epoch": 5.751238591916558,
-      "grad_norm": 1.4617528915405273,
-      "learning_rate": 8.942211055276383e-05,
-      "loss": 5.3073,
-      "step": 11028
-    },
-    {
-      "epoch": 5.751760104302477,
-      "grad_norm": 1.2781676054000854,
-      "learning_rate": 8.942110552763819e-05,
-      "loss": 5.6155,
-      "step": 11029
-    },
-    {
-      "epoch": 5.752281616688396,
-      "grad_norm": 1.4530922174453735,
-      "learning_rate": 8.942010050251257e-05,
-      "loss": 5.7863,
-      "step": 11030
-    },
-    {
-      "epoch": 5.752803129074316,
-      "grad_norm": 1.499869704246521,
-      "learning_rate": 8.941909547738693e-05,
-      "loss": 6.0168,
-      "step": 11031
-    },
-    {
-      "epoch": 5.753324641460234,
-      "grad_norm": 1.4238908290863037,
-      "learning_rate": 8.941809045226131e-05,
-      "loss": 5.6372,
-      "step": 11032
-    },
-    {
-      "epoch": 5.753846153846154,
-      "grad_norm": 1.2409641742706299,
-      "learning_rate": 8.941708542713568e-05,
-      "loss": 6.0574,
-      "step": 11033
-    },
-    {
-      "epoch": 5.754367666232073,
-      "grad_norm": 1.3865808248519897,
-      "learning_rate": 8.941608040201005e-05,
-      "loss": 5.89,
-      "step": 11034
-    },
-    {
-      "epoch": 5.754889178617992,
-      "grad_norm": 1.4461098909378052,
-      "learning_rate": 8.941507537688443e-05,
-      "loss": 5.7926,
-      "step": 11035
-    },
-    {
-      "epoch": 5.755410691003911,
-      "grad_norm": 1.344017744064331,
-      "learning_rate": 8.94140703517588e-05,
-      "loss": 5.6779,
-      "step": 11036
-    },
-    {
-      "epoch": 5.755932203389831,
-      "grad_norm": 1.4051610231399536,
-      "learning_rate": 8.941306532663317e-05,
-      "loss": 5.5074,
-      "step": 11037
-    },
-    {
-      "epoch": 5.756453715775749,
-      "grad_norm": 1.697458267211914,
-      "learning_rate": 8.941206030150754e-05,
-      "loss": 5.7422,
-      "step": 11038
-    },
-    {
-      "epoch": 5.756975228161669,
-      "grad_norm": 1.4263421297073364,
-      "learning_rate": 8.941105527638192e-05,
-      "loss": 5.6597,
-      "step": 11039
-    },
-    {
-      "epoch": 5.757496740547588,
-      "grad_norm": 1.430764079093933,
-      "learning_rate": 8.941005025125628e-05,
-      "loss": 5.9198,
-      "step": 11040
-    },
-    {
-      "epoch": 5.758018252933507,
-      "grad_norm": 1.4301328659057617,
-      "learning_rate": 8.940904522613066e-05,
-      "loss": 5.6439,
-      "step": 11041
-    },
-    {
-      "epoch": 5.758539765319426,
-      "grad_norm": 1.4273492097854614,
-      "learning_rate": 8.940804020100502e-05,
-      "loss": 5.2004,
-      "step": 11042
-    },
-    {
-      "epoch": 5.759061277705346,
-      "grad_norm": 1.3749626874923706,
-      "learning_rate": 8.94070351758794e-05,
-      "loss": 5.2141,
-      "step": 11043
-    },
-    {
-      "epoch": 5.759582790091264,
-      "grad_norm": 1.4186491966247559,
-      "learning_rate": 8.940603015075378e-05,
-      "loss": 5.7131,
-      "step": 11044
-    },
-    {
-      "epoch": 5.760104302477184,
-      "grad_norm": 1.457743763923645,
-      "learning_rate": 8.940502512562816e-05,
-      "loss": 5.35,
-      "step": 11045
-    },
-    {
-      "epoch": 5.760625814863103,
-      "grad_norm": 1.3756777048110962,
-      "learning_rate": 8.940402010050252e-05,
-      "loss": 5.6292,
-      "step": 11046
-    },
-    {
-      "epoch": 5.761147327249022,
-      "grad_norm": 1.5122383832931519,
-      "learning_rate": 8.940301507537688e-05,
-      "loss": 5.3864,
-      "step": 11047
-    },
-    {
-      "epoch": 5.761668839634941,
-      "grad_norm": 1.7569669485092163,
-      "learning_rate": 8.940201005025126e-05,
-      "loss": 5.3712,
-      "step": 11048
-    },
-    {
-      "epoch": 5.762190352020861,
-      "grad_norm": 1.3629153966903687,
-      "learning_rate": 8.940100502512563e-05,
-      "loss": 5.37,
-      "step": 11049
-    },
-    {
-      "epoch": 5.762711864406779,
-      "grad_norm": 1.326552152633667,
-      "learning_rate": 8.94e-05,
-      "loss": 5.9947,
-      "step": 11050
-    },
-    {
-      "epoch": 5.763233376792699,
-      "grad_norm": 1.3558276891708374,
-      "learning_rate": 8.939899497487437e-05,
-      "loss": 5.6667,
-      "step": 11051
-    },
-    {
-      "epoch": 5.763754889178618,
-      "grad_norm": 1.3653569221496582,
-      "learning_rate": 8.939798994974875e-05,
-      "loss": 5.8591,
-      "step": 11052
-    },
-    {
-      "epoch": 5.764276401564537,
-      "grad_norm": 1.493650197982788,
-      "learning_rate": 8.939698492462311e-05,
-      "loss": 5.5972,
-      "step": 11053
-    },
-    {
-      "epoch": 5.764797913950456,
-      "grad_norm": 1.481748342514038,
-      "learning_rate": 8.939597989949749e-05,
-      "loss": 5.1749,
-      "step": 11054
-    },
-    {
-      "epoch": 5.765319426336376,
-      "grad_norm": 1.3626928329467773,
-      "learning_rate": 8.939497487437187e-05,
-      "loss": 5.9639,
-      "step": 11055
-    },
-    {
-      "epoch": 5.765840938722294,
-      "grad_norm": 1.3350484371185303,
-      "learning_rate": 8.939396984924624e-05,
-      "loss": 5.6259,
-      "step": 11056
-    },
-    {
-      "epoch": 5.766362451108214,
-      "grad_norm": 1.4143821001052856,
-      "learning_rate": 8.939296482412061e-05,
-      "loss": 5.7037,
-      "step": 11057
-    },
-    {
-      "epoch": 5.766883963494133,
-      "grad_norm": 1.4553712606430054,
-      "learning_rate": 8.939195979899499e-05,
-      "loss": 6.1459,
-      "step": 11058
-    },
-    {
-      "epoch": 5.767405475880052,
-      "grad_norm": 1.3865184783935547,
-      "learning_rate": 8.939095477386935e-05,
-      "loss": 5.31,
-      "step": 11059
-    },
-    {
-      "epoch": 5.767926988265971,
-      "grad_norm": 1.4803879261016846,
-      "learning_rate": 8.938994974874372e-05,
-      "loss": 5.8505,
-      "step": 11060
-    },
-    {
-      "epoch": 5.768448500651891,
-      "grad_norm": 1.5032525062561035,
-      "learning_rate": 8.938894472361809e-05,
-      "loss": 5.665,
-      "step": 11061
-    },
-    {
-      "epoch": 5.768970013037809,
-      "grad_norm": 1.4906001091003418,
-      "learning_rate": 8.938793969849246e-05,
-      "loss": 5.6805,
-      "step": 11062
-    },
-    {
-      "epoch": 5.769491525423729,
-      "grad_norm": 1.5074368715286255,
-      "learning_rate": 8.938693467336684e-05,
-      "loss": 5.082,
-      "step": 11063
-    },
-    {
-      "epoch": 5.770013037809648,
-      "grad_norm": 1.3212162256240845,
-      "learning_rate": 8.938592964824121e-05,
-      "loss": 4.9176,
-      "step": 11064
-    },
-    {
-      "epoch": 5.770534550195567,
-      "grad_norm": 1.5805473327636719,
-      "learning_rate": 8.938492462311559e-05,
-      "loss": 5.4522,
-      "step": 11065
-    },
-    {
-      "epoch": 5.771056062581486,
-      "grad_norm": 1.3666032552719116,
-      "learning_rate": 8.938391959798996e-05,
-      "loss": 5.1997,
-      "step": 11066
-    },
-    {
-      "epoch": 5.771577574967406,
-      "grad_norm": 1.3874982595443726,
-      "learning_rate": 8.938291457286433e-05,
-      "loss": 5.6514,
-      "step": 11067
-    },
-    {
-      "epoch": 5.772099087353324,
-      "grad_norm": 1.5265312194824219,
-      "learning_rate": 8.93819095477387e-05,
-      "loss": 4.5437,
-      "step": 11068
-    },
-    {
-      "epoch": 5.772620599739244,
-      "grad_norm": 1.3165547847747803,
-      "learning_rate": 8.938090452261307e-05,
-      "loss": 5.8756,
-      "step": 11069
-    },
-    {
-      "epoch": 5.7731421121251625,
-      "grad_norm": 1.5182777643203735,
-      "learning_rate": 8.937989949748744e-05,
-      "loss": 5.2466,
-      "step": 11070
-    },
-    {
-      "epoch": 5.773663624511082,
-      "grad_norm": 1.555492639541626,
-      "learning_rate": 8.937889447236182e-05,
-      "loss": 5.0518,
-      "step": 11071
-    },
-    {
-      "epoch": 5.774185136897001,
-      "grad_norm": 1.4603099822998047,
-      "learning_rate": 8.937788944723618e-05,
-      "loss": 5.8957,
-      "step": 11072
-    },
-    {
-      "epoch": 5.774706649282921,
-      "grad_norm": 1.4643217325210571,
-      "learning_rate": 8.937688442211055e-05,
-      "loss": 5.5346,
-      "step": 11073
-    },
-    {
-      "epoch": 5.7752281616688395,
-      "grad_norm": 1.7149631977081299,
-      "learning_rate": 8.937587939698492e-05,
-      "loss": 5.227,
-      "step": 11074
-    },
-    {
-      "epoch": 5.775749674054759,
-      "grad_norm": 1.679935097694397,
-      "learning_rate": 8.93748743718593e-05,
-      "loss": 5.737,
-      "step": 11075
-    },
-    {
-      "epoch": 5.7762711864406775,
-      "grad_norm": 1.650301456451416,
-      "learning_rate": 8.937386934673368e-05,
-      "loss": 5.7239,
-      "step": 11076
-    },
-    {
-      "epoch": 5.776792698826597,
-      "grad_norm": 1.6086139678955078,
-      "learning_rate": 8.937286432160804e-05,
-      "loss": 5.4067,
-      "step": 11077
-    },
-    {
-      "epoch": 5.777314211212516,
-      "grad_norm": 1.38319730758667,
-      "learning_rate": 8.937185929648242e-05,
-      "loss": 5.8881,
-      "step": 11078
-    },
-    {
-      "epoch": 5.777835723598436,
-      "grad_norm": 1.432379961013794,
-      "learning_rate": 8.937085427135679e-05,
-      "loss": 5.9094,
-      "step": 11079
-    },
-    {
-      "epoch": 5.7783572359843545,
-      "grad_norm": 1.3226871490478516,
-      "learning_rate": 8.936984924623116e-05,
-      "loss": 5.8952,
-      "step": 11080
-    },
-    {
-      "epoch": 5.778878748370274,
-      "grad_norm": 1.5094563961029053,
-      "learning_rate": 8.936884422110553e-05,
-      "loss": 5.8823,
-      "step": 11081
-    },
-    {
-      "epoch": 5.7794002607561925,
-      "grad_norm": 1.4125616550445557,
-      "learning_rate": 8.93678391959799e-05,
-      "loss": 5.7093,
-      "step": 11082
-    },
-    {
-      "epoch": 5.779921773142112,
-      "grad_norm": 1.3918821811676025,
-      "learning_rate": 8.936683417085427e-05,
-      "loss": 5.8885,
-      "step": 11083
-    },
-    {
-      "epoch": 5.780443285528031,
-      "grad_norm": 1.442017912864685,
-      "learning_rate": 8.936582914572865e-05,
-      "loss": 5.5438,
-      "step": 11084
-    },
-    {
-      "epoch": 5.780964797913951,
-      "grad_norm": 1.5405405759811401,
-      "learning_rate": 8.936482412060303e-05,
-      "loss": 5.6053,
-      "step": 11085
-    },
-    {
-      "epoch": 5.7814863102998695,
-      "grad_norm": 1.3721171617507935,
-      "learning_rate": 8.936381909547739e-05,
-      "loss": 5.0246,
-      "step": 11086
-    },
-    {
-      "epoch": 5.782007822685789,
-      "grad_norm": 1.5957597494125366,
-      "learning_rate": 8.936281407035177e-05,
-      "loss": 5.8572,
-      "step": 11087
-    },
-    {
-      "epoch": 5.7825293350717075,
-      "grad_norm": 1.4496006965637207,
-      "learning_rate": 8.936180904522613e-05,
-      "loss": 5.6295,
-      "step": 11088
-    },
-    {
-      "epoch": 5.783050847457627,
-      "grad_norm": 1.6491602659225464,
-      "learning_rate": 8.936080402010051e-05,
-      "loss": 5.4069,
-      "step": 11089
-    },
-    {
-      "epoch": 5.7835723598435465,
-      "grad_norm": 1.4083802700042725,
-      "learning_rate": 8.935979899497487e-05,
-      "loss": 5.6554,
-      "step": 11090
-    },
-    {
-      "epoch": 5.784093872229466,
-      "grad_norm": 1.447011947631836,
-      "learning_rate": 8.935879396984925e-05,
-      "loss": 5.7874,
-      "step": 11091
-    },
-    {
-      "epoch": 5.7846153846153845,
-      "grad_norm": 1.5056390762329102,
-      "learning_rate": 8.935778894472362e-05,
-      "loss": 5.3976,
-      "step": 11092
-    },
-    {
-      "epoch": 5.785136897001304,
-      "grad_norm": 1.5651241540908813,
-      "learning_rate": 8.9356783919598e-05,
-      "loss": 4.9541,
-      "step": 11093
-    },
-    {
-      "epoch": 5.7856584093872225,
-      "grad_norm": 2.1426172256469727,
-      "learning_rate": 8.935577889447236e-05,
-      "loss": 5.3628,
-      "step": 11094
-    },
-    {
-      "epoch": 5.786179921773142,
-      "grad_norm": 1.5441268682479858,
-      "learning_rate": 8.935477386934674e-05,
-      "loss": 5.7623,
-      "step": 11095
-    },
-    {
-      "epoch": 5.7867014341590615,
-      "grad_norm": 1.530523419380188,
-      "learning_rate": 8.935376884422111e-05,
-      "loss": 6.1924,
-      "step": 11096
-    },
-    {
-      "epoch": 5.787222946544981,
-      "grad_norm": 1.4667168855667114,
-      "learning_rate": 8.935276381909549e-05,
-      "loss": 5.5349,
-      "step": 11097
-    },
-    {
-      "epoch": 5.7877444589308995,
-      "grad_norm": 1.4252454042434692,
-      "learning_rate": 8.935175879396986e-05,
-      "loss": 5.8541,
-      "step": 11098
-    },
-    {
-      "epoch": 5.788265971316819,
-      "grad_norm": 1.4585285186767578,
-      "learning_rate": 8.935075376884423e-05,
-      "loss": 5.3541,
-      "step": 11099
-    },
-    {
-      "epoch": 5.7887874837027375,
-      "grad_norm": 1.462559700012207,
-      "learning_rate": 8.93497487437186e-05,
-      "loss": 6.0104,
-      "step": 11100
-    },
-    {
-      "epoch": 5.789308996088657,
-      "grad_norm": 1.467221736907959,
-      "learning_rate": 8.934874371859296e-05,
-      "loss": 5.7553,
-      "step": 11101
-    },
-    {
-      "epoch": 5.7898305084745765,
-      "grad_norm": 1.3957103490829468,
-      "learning_rate": 8.934773869346734e-05,
-      "loss": 5.8107,
-      "step": 11102
-    },
-    {
-      "epoch": 5.790352020860495,
-      "grad_norm": 1.5709515810012817,
-      "learning_rate": 8.93467336683417e-05,
-      "loss": 5.9105,
-      "step": 11103
-    },
-    {
-      "epoch": 5.7908735332464145,
-      "grad_norm": 1.4631412029266357,
-      "learning_rate": 8.934572864321608e-05,
-      "loss": 5.7856,
-      "step": 11104
-    },
-    {
-      "epoch": 5.791395045632334,
-      "grad_norm": 1.3820090293884277,
-      "learning_rate": 8.934472361809046e-05,
-      "loss": 5.6592,
-      "step": 11105
-    },
-    {
-      "epoch": 5.791916558018253,
-      "grad_norm": 1.5266880989074707,
-      "learning_rate": 8.934371859296484e-05,
-      "loss": 5.8181,
-      "step": 11106
-    },
-    {
-      "epoch": 5.792438070404172,
-      "grad_norm": 1.4228094816207886,
-      "learning_rate": 8.93427135678392e-05,
-      "loss": 5.9663,
-      "step": 11107
-    },
-    {
-      "epoch": 5.7929595827900915,
-      "grad_norm": 1.4558594226837158,
-      "learning_rate": 8.934170854271358e-05,
-      "loss": 5.5827,
-      "step": 11108
-    },
-    {
-      "epoch": 5.79348109517601,
-      "grad_norm": 1.6535422801971436,
-      "learning_rate": 8.934070351758794e-05,
-      "loss": 5.1751,
-      "step": 11109
-    },
-    {
-      "epoch": 5.7940026075619295,
-      "grad_norm": 1.3984551429748535,
-      "learning_rate": 8.933969849246232e-05,
-      "loss": 6.0277,
-      "step": 11110
-    },
-    {
-      "epoch": 5.794524119947849,
-      "grad_norm": 1.4308351278305054,
-      "learning_rate": 8.933869346733669e-05,
-      "loss": 5.4417,
-      "step": 11111
-    },
-    {
-      "epoch": 5.795045632333768,
-      "grad_norm": 1.5795142650604248,
-      "learning_rate": 8.933768844221106e-05,
-      "loss": 5.1329,
-      "step": 11112
-    },
-    {
-      "epoch": 5.795567144719687,
-      "grad_norm": 1.3238779306411743,
-      "learning_rate": 8.933668341708543e-05,
-      "loss": 5.746,
-      "step": 11113
-    },
-    {
-      "epoch": 5.7960886571056065,
-      "grad_norm": 1.5570451021194458,
-      "learning_rate": 8.933567839195979e-05,
-      "loss": 5.6325,
-      "step": 11114
-    },
-    {
-      "epoch": 5.796610169491525,
-      "grad_norm": 1.3931790590286255,
-      "learning_rate": 8.933467336683417e-05,
-      "loss": 5.566,
-      "step": 11115
-    },
-    {
-      "epoch": 5.7971316818774445,
-      "grad_norm": 1.429088830947876,
-      "learning_rate": 8.933366834170855e-05,
-      "loss": 5.5243,
-      "step": 11116
-    },
-    {
-      "epoch": 5.797653194263364,
-      "grad_norm": 1.45830237865448,
-      "learning_rate": 8.933266331658293e-05,
-      "loss": 5.3583,
-      "step": 11117
-    },
-    {
-      "epoch": 5.798174706649283,
-      "grad_norm": 1.5274248123168945,
-      "learning_rate": 8.933165829145729e-05,
-      "loss": 5.3013,
-      "step": 11118
-    },
-    {
-      "epoch": 5.798696219035202,
-      "grad_norm": 1.4783509969711304,
-      "learning_rate": 8.933065326633167e-05,
-      "loss": 5.4999,
-      "step": 11119
-    },
-    {
-      "epoch": 5.7992177314211215,
-      "grad_norm": 1.3838443756103516,
-      "learning_rate": 8.932964824120603e-05,
-      "loss": 5.6652,
-      "step": 11120
-    },
-    {
-      "epoch": 5.79973924380704,
-      "grad_norm": 1.520408034324646,
-      "learning_rate": 8.932864321608041e-05,
-      "loss": 5.6449,
-      "step": 11121
-    },
-    {
-      "epoch": 5.8002607561929596,
-      "grad_norm": 1.590740442276001,
-      "learning_rate": 8.932763819095477e-05,
-      "loss": 5.2637,
-      "step": 11122
-    },
-    {
-      "epoch": 5.800782268578879,
-      "grad_norm": 1.6381679773330688,
-      "learning_rate": 8.932663316582915e-05,
-      "loss": 5.4653,
-      "step": 11123
-    },
-    {
-      "epoch": 5.801303780964798,
-      "grad_norm": 1.4744840860366821,
-      "learning_rate": 8.932562814070352e-05,
-      "loss": 5.5353,
-      "step": 11124
-    },
-    {
-      "epoch": 5.801825293350717,
-      "grad_norm": 1.555695652961731,
-      "learning_rate": 8.93246231155779e-05,
-      "loss": 5.2291,
-      "step": 11125
-    },
-    {
-      "epoch": 5.8023468057366365,
-      "grad_norm": 1.5720689296722412,
-      "learning_rate": 8.932361809045227e-05,
-      "loss": 5.4784,
-      "step": 11126
-    },
-    {
-      "epoch": 5.802868318122555,
-      "grad_norm": 1.4789228439331055,
-      "learning_rate": 8.932261306532664e-05,
-      "loss": 5.7007,
-      "step": 11127
-    },
-    {
-      "epoch": 5.803389830508475,
-      "grad_norm": 1.482513666152954,
-      "learning_rate": 8.932160804020101e-05,
-      "loss": 5.1978,
-      "step": 11128
-    },
-    {
-      "epoch": 5.803911342894394,
-      "grad_norm": 1.5388909578323364,
-      "learning_rate": 8.932060301507538e-05,
-      "loss": 5.8298,
-      "step": 11129
-    },
-    {
-      "epoch": 5.804432855280313,
-      "grad_norm": 1.469780445098877,
-      "learning_rate": 8.931959798994976e-05,
-      "loss": 5.3417,
-      "step": 11130
-    },
-    {
-      "epoch": 5.804954367666232,
-      "grad_norm": 1.4363586902618408,
-      "learning_rate": 8.931859296482412e-05,
-      "loss": 5.2953,
-      "step": 11131
-    },
-    {
-      "epoch": 5.8054758800521515,
-      "grad_norm": 1.5173383951187134,
-      "learning_rate": 8.93175879396985e-05,
-      "loss": 5.8719,
-      "step": 11132
-    },
-    {
-      "epoch": 5.80599739243807,
-      "grad_norm": 1.7427622079849243,
-      "learning_rate": 8.931658291457286e-05,
-      "loss": 5.9253,
-      "step": 11133
-    },
-    {
-      "epoch": 5.80651890482399,
-      "grad_norm": 1.2970123291015625,
-      "learning_rate": 8.931557788944724e-05,
-      "loss": 5.6237,
-      "step": 11134
-    },
-    {
-      "epoch": 5.807040417209909,
-      "grad_norm": 1.3498351573944092,
-      "learning_rate": 8.93145728643216e-05,
-      "loss": 5.8077,
-      "step": 11135
-    },
-    {
-      "epoch": 5.807561929595828,
-      "grad_norm": 1.4313232898712158,
-      "learning_rate": 8.931356783919598e-05,
-      "loss": 5.7617,
-      "step": 11136
-    },
-    {
-      "epoch": 5.808083441981747,
-      "grad_norm": 1.3648717403411865,
-      "learning_rate": 8.931256281407036e-05,
-      "loss": 5.7821,
-      "step": 11137
-    },
-    {
-      "epoch": 5.8086049543676666,
-      "grad_norm": 1.4350194931030273,
-      "learning_rate": 8.931155778894474e-05,
-      "loss": 5.5779,
-      "step": 11138
-    },
-    {
-      "epoch": 5.809126466753585,
-      "grad_norm": 1.3823164701461792,
-      "learning_rate": 8.93105527638191e-05,
-      "loss": 5.7715,
-      "step": 11139
-    },
-    {
-      "epoch": 5.809647979139505,
-      "grad_norm": 1.3920141458511353,
-      "learning_rate": 8.930954773869347e-05,
-      "loss": 5.8081,
-      "step": 11140
-    },
-    {
-      "epoch": 5.810169491525424,
-      "grad_norm": 1.4346535205841064,
-      "learning_rate": 8.930854271356784e-05,
-      "loss": 5.7096,
-      "step": 11141
-    },
-    {
-      "epoch": 5.810691003911343,
-      "grad_norm": 1.5402590036392212,
-      "learning_rate": 8.930753768844221e-05,
-      "loss": 5.6574,
-      "step": 11142
-    },
-    {
-      "epoch": 5.811212516297262,
-      "grad_norm": 1.358138084411621,
-      "learning_rate": 8.930653266331659e-05,
-      "loss": 5.6125,
-      "step": 11143
-    },
-    {
-      "epoch": 5.811734028683182,
-      "grad_norm": 1.6958190202713013,
-      "learning_rate": 8.930552763819095e-05,
-      "loss": 5.9064,
-      "step": 11144
-    },
-    {
-      "epoch": 5.8122555410691,
-      "grad_norm": 1.3471524715423584,
-      "learning_rate": 8.930452261306533e-05,
-      "loss": 5.8068,
-      "step": 11145
-    },
-    {
-      "epoch": 5.81277705345502,
-      "grad_norm": 1.4115896224975586,
-      "learning_rate": 8.93035175879397e-05,
-      "loss": 5.2993,
-      "step": 11146
-    },
-    {
-      "epoch": 5.813298565840939,
-      "grad_norm": 1.3721095323562622,
-      "learning_rate": 8.930251256281407e-05,
-      "loss": 5.7738,
-      "step": 11147
-    },
-    {
-      "epoch": 5.813820078226858,
-      "grad_norm": 1.3466501235961914,
-      "learning_rate": 8.930150753768845e-05,
-      "loss": 5.0943,
-      "step": 11148
-    },
-    {
-      "epoch": 5.814341590612777,
-      "grad_norm": 1.4687105417251587,
-      "learning_rate": 8.930050251256283e-05,
-      "loss": 5.6988,
-      "step": 11149
-    },
-    {
-      "epoch": 5.814863102998697,
-      "grad_norm": 1.5961259603500366,
-      "learning_rate": 8.929949748743719e-05,
-      "loss": 5.5994,
-      "step": 11150
-    },
-    {
-      "epoch": 5.815384615384615,
-      "grad_norm": 1.5093573331832886,
-      "learning_rate": 8.929849246231157e-05,
-      "loss": 5.7422,
-      "step": 11151
-    },
-    {
-      "epoch": 5.815906127770535,
-      "grad_norm": 1.3976436853408813,
-      "learning_rate": 8.929748743718593e-05,
-      "loss": 5.8555,
-      "step": 11152
-    },
-    {
-      "epoch": 5.816427640156454,
-      "grad_norm": 1.4196374416351318,
-      "learning_rate": 8.92964824120603e-05,
-      "loss": 5.7201,
-      "step": 11153
-    },
-    {
-      "epoch": 5.816949152542373,
-      "grad_norm": 1.4761029481887817,
-      "learning_rate": 8.929547738693468e-05,
-      "loss": 5.2788,
-      "step": 11154
-    },
-    {
-      "epoch": 5.817470664928292,
-      "grad_norm": 1.3595227003097534,
-      "learning_rate": 8.929447236180904e-05,
-      "loss": 5.6304,
-      "step": 11155
-    },
-    {
-      "epoch": 5.817992177314212,
-      "grad_norm": 1.4221817255020142,
-      "learning_rate": 8.929346733668342e-05,
-      "loss": 5.7275,
-      "step": 11156
-    },
-    {
-      "epoch": 5.81851368970013,
-      "grad_norm": 1.335982084274292,
-      "learning_rate": 8.92924623115578e-05,
-      "loss": 5.9086,
-      "step": 11157
-    },
-    {
-      "epoch": 5.81903520208605,
-      "grad_norm": 1.7649877071380615,
-      "learning_rate": 8.929145728643217e-05,
-      "loss": 5.435,
-      "step": 11158
-    },
-    {
-      "epoch": 5.819556714471969,
-      "grad_norm": 1.4133331775665283,
-      "learning_rate": 8.929045226130654e-05,
-      "loss": 5.5371,
-      "step": 11159
-    },
-    {
-      "epoch": 5.820078226857888,
-      "grad_norm": 1.4357035160064697,
-      "learning_rate": 8.928944723618092e-05,
-      "loss": 5.2902,
-      "step": 11160
-    },
-    {
-      "epoch": 5.820599739243807,
-      "grad_norm": 1.399936318397522,
-      "learning_rate": 8.928844221105528e-05,
-      "loss": 5.9164,
-      "step": 11161
-    },
-    {
-      "epoch": 5.821121251629727,
-      "grad_norm": 1.4258466958999634,
-      "learning_rate": 8.928743718592966e-05,
-      "loss": 5.5577,
-      "step": 11162
-    },
-    {
-      "epoch": 5.821642764015645,
-      "grad_norm": 1.5698837041854858,
-      "learning_rate": 8.928643216080402e-05,
-      "loss": 5.2503,
-      "step": 11163
-    },
-    {
-      "epoch": 5.822164276401565,
-      "grad_norm": 1.6133439540863037,
-      "learning_rate": 8.92854271356784e-05,
-      "loss": 5.5462,
-      "step": 11164
-    },
-    {
-      "epoch": 5.822685788787483,
-      "grad_norm": 1.7021983861923218,
-      "learning_rate": 8.928442211055276e-05,
-      "loss": 5.6466,
-      "step": 11165
-    },
-    {
-      "epoch": 5.823207301173403,
-      "grad_norm": 1.749825358390808,
-      "learning_rate": 8.928341708542713e-05,
-      "loss": 5.7529,
-      "step": 11166
-    },
-    {
-      "epoch": 5.823728813559322,
-      "grad_norm": 1.6092408895492554,
-      "learning_rate": 8.92824120603015e-05,
-      "loss": 5.4618,
-      "step": 11167
-    },
-    {
-      "epoch": 5.824250325945242,
-      "grad_norm": 1.5879006385803223,
-      "learning_rate": 8.928140703517588e-05,
-      "loss": 5.5362,
-      "step": 11168
-    },
-    {
-      "epoch": 5.82477183833116,
-      "grad_norm": 1.5245084762573242,
-      "learning_rate": 8.928040201005026e-05,
-      "loss": 5.7537,
-      "step": 11169
-    },
-    {
-      "epoch": 5.82529335071708,
-      "grad_norm": 1.5205795764923096,
-      "learning_rate": 8.927939698492463e-05,
-      "loss": 5.6498,
-      "step": 11170
-    },
-    {
-      "epoch": 5.825814863102998,
-      "grad_norm": 1.4795408248901367,
-      "learning_rate": 8.9278391959799e-05,
-      "loss": 5.6866,
-      "step": 11171
-    },
-    {
-      "epoch": 5.826336375488918,
-      "grad_norm": 1.6438924074172974,
-      "learning_rate": 8.927738693467337e-05,
-      "loss": 5.6655,
-      "step": 11172
-    },
-    {
-      "epoch": 5.826857887874837,
-      "grad_norm": 1.7088818550109863,
-      "learning_rate": 8.927638190954775e-05,
-      "loss": 5.7583,
-      "step": 11173
-    },
-    {
-      "epoch": 5.827379400260757,
-      "grad_norm": 1.390309453010559,
-      "learning_rate": 8.927537688442211e-05,
-      "loss": 5.8022,
-      "step": 11174
-    },
-    {
-      "epoch": 5.827900912646675,
-      "grad_norm": 1.4313936233520508,
-      "learning_rate": 8.927437185929649e-05,
-      "loss": 6.1077,
-      "step": 11175
-    },
-    {
-      "epoch": 5.828422425032595,
-      "grad_norm": 1.5214931964874268,
-      "learning_rate": 8.927336683417085e-05,
-      "loss": 5.5847,
-      "step": 11176
-    },
-    {
-      "epoch": 5.828943937418513,
-      "grad_norm": 1.4822449684143066,
-      "learning_rate": 8.927236180904523e-05,
-      "loss": 5.8708,
-      "step": 11177
-    },
-    {
-      "epoch": 5.829465449804433,
-      "grad_norm": 1.5202186107635498,
-      "learning_rate": 8.927135678391961e-05,
-      "loss": 5.4048,
-      "step": 11178
-    },
-    {
-      "epoch": 5.829986962190352,
-      "grad_norm": 1.358853816986084,
-      "learning_rate": 8.927035175879397e-05,
-      "loss": 5.9877,
-      "step": 11179
-    },
-    {
-      "epoch": 5.830508474576272,
-      "grad_norm": 1.3562949895858765,
-      "learning_rate": 8.926934673366835e-05,
-      "loss": 5.3845,
-      "step": 11180
-    },
-    {
-      "epoch": 5.83102998696219,
-      "grad_norm": 1.502917766571045,
-      "learning_rate": 8.926834170854271e-05,
-      "loss": 5.8421,
-      "step": 11181
-    },
-    {
-      "epoch": 5.83155149934811,
-      "grad_norm": 1.4197522401809692,
-      "learning_rate": 8.926733668341709e-05,
-      "loss": 5.5628,
-      "step": 11182
-    },
-    {
-      "epoch": 5.832073011734028,
-      "grad_norm": 1.4737796783447266,
-      "learning_rate": 8.926633165829146e-05,
-      "loss": 5.9039,
-      "step": 11183
-    },
-    {
-      "epoch": 5.832594524119948,
-      "grad_norm": 1.4002681970596313,
-      "learning_rate": 8.926532663316583e-05,
-      "loss": 5.8194,
-      "step": 11184
-    },
-    {
-      "epoch": 5.833116036505867,
-      "grad_norm": 1.3653066158294678,
-      "learning_rate": 8.92643216080402e-05,
-      "loss": 5.7555,
-      "step": 11185
-    },
-    {
-      "epoch": 5.833637548891787,
-      "grad_norm": 1.4676254987716675,
-      "learning_rate": 8.926331658291458e-05,
-      "loss": 5.9091,
-      "step": 11186
-    },
-    {
-      "epoch": 5.834159061277705,
-      "grad_norm": 1.4901639223098755,
-      "learning_rate": 8.926231155778894e-05,
-      "loss": 5.4499,
-      "step": 11187
-    },
-    {
-      "epoch": 5.834680573663625,
-      "grad_norm": 1.497543215751648,
-      "learning_rate": 8.926130653266332e-05,
-      "loss": 5.8084,
-      "step": 11188
-    },
-    {
-      "epoch": 5.835202086049543,
-      "grad_norm": 1.5010689496994019,
-      "learning_rate": 8.92603015075377e-05,
-      "loss": 5.7141,
-      "step": 11189
-    },
-    {
-      "epoch": 5.835723598435463,
-      "grad_norm": 1.4037083387374878,
-      "learning_rate": 8.925929648241207e-05,
-      "loss": 5.5224,
-      "step": 11190
-    },
-    {
-      "epoch": 5.836245110821382,
-      "grad_norm": 1.6191229820251465,
-      "learning_rate": 8.925829145728644e-05,
-      "loss": 5.7469,
-      "step": 11191
-    },
-    {
-      "epoch": 5.836766623207302,
-      "grad_norm": 1.4571068286895752,
-      "learning_rate": 8.925728643216082e-05,
-      "loss": 5.4179,
-      "step": 11192
-    },
-    {
-      "epoch": 5.83728813559322,
-      "grad_norm": 1.4861764907836914,
-      "learning_rate": 8.925628140703518e-05,
-      "loss": 5.7857,
-      "step": 11193
-    },
-    {
-      "epoch": 5.83780964797914,
-      "grad_norm": 1.3416666984558105,
-      "learning_rate": 8.925527638190954e-05,
-      "loss": 5.6452,
-      "step": 11194
-    },
-    {
-      "epoch": 5.838331160365058,
-      "grad_norm": 1.254943609237671,
-      "learning_rate": 8.925427135678392e-05,
-      "loss": 5.9883,
-      "step": 11195
-    },
-    {
-      "epoch": 5.838852672750978,
-      "grad_norm": 1.4150797128677368,
-      "learning_rate": 8.925326633165829e-05,
-      "loss": 5.7723,
-      "step": 11196
-    },
-    {
-      "epoch": 5.839374185136897,
-      "grad_norm": 1.4482008218765259,
-      "learning_rate": 8.925226130653266e-05,
-      "loss": 5.5461,
-      "step": 11197
-    },
-    {
-      "epoch": 5.839895697522816,
-      "grad_norm": 1.4165109395980835,
-      "learning_rate": 8.925125628140704e-05,
-      "loss": 5.7423,
-      "step": 11198
-    },
-    {
-      "epoch": 5.840417209908735,
-      "grad_norm": 1.3813997507095337,
-      "learning_rate": 8.925025125628142e-05,
-      "loss": 5.6766,
-      "step": 11199
-    },
-    {
-      "epoch": 5.840938722294655,
-      "grad_norm": 1.3078476190567017,
-      "learning_rate": 8.924924623115578e-05,
-      "loss": 5.8741,
-      "step": 11200
-    },
-    {
-      "epoch": 5.841460234680573,
-      "grad_norm": 1.6934423446655273,
-      "learning_rate": 8.924824120603016e-05,
-      "loss": 5.1995,
-      "step": 11201
-    },
-    {
-      "epoch": 5.841981747066493,
-      "grad_norm": 1.3250765800476074,
-      "learning_rate": 8.924723618090453e-05,
-      "loss": 6.0364,
-      "step": 11202
-    },
-    {
-      "epoch": 5.842503259452412,
-      "grad_norm": 1.3266997337341309,
-      "learning_rate": 8.92462311557789e-05,
-      "loss": 5.3317,
-      "step": 11203
-    },
-    {
-      "epoch": 5.843024771838331,
-      "grad_norm": 1.5322840213775635,
-      "learning_rate": 8.924522613065327e-05,
-      "loss": 5.8234,
-      "step": 11204
-    },
-    {
-      "epoch": 5.84354628422425,
-      "grad_norm": 1.4361432790756226,
-      "learning_rate": 8.924422110552765e-05,
-      "loss": 5.3023,
-      "step": 11205
-    },
-    {
-      "epoch": 5.84406779661017,
-      "grad_norm": 1.4494373798370361,
-      "learning_rate": 8.924321608040201e-05,
-      "loss": 5.2297,
-      "step": 11206
-    },
-    {
-      "epoch": 5.844589308996088,
-      "grad_norm": 1.3717869520187378,
-      "learning_rate": 8.924221105527637e-05,
-      "loss": 5.3078,
-      "step": 11207
-    },
-    {
-      "epoch": 5.845110821382008,
-      "grad_norm": 1.4638352394104004,
-      "learning_rate": 8.924120603015075e-05,
-      "loss": 5.7285,
-      "step": 11208
-    },
-    {
-      "epoch": 5.845632333767927,
-      "grad_norm": 1.4879920482635498,
-      "learning_rate": 8.924020100502513e-05,
-      "loss": 5.5045,
-      "step": 11209
-    },
-    {
-      "epoch": 5.846153846153846,
-      "grad_norm": 1.4486687183380127,
-      "learning_rate": 8.923919597989951e-05,
-      "loss": 5.7852,
-      "step": 11210
-    },
-    {
-      "epoch": 5.846675358539765,
-      "grad_norm": 1.638842225074768,
-      "learning_rate": 8.923819095477387e-05,
-      "loss": 5.5267,
-      "step": 11211
-    },
-    {
-      "epoch": 5.847196870925685,
-      "grad_norm": 1.6264042854309082,
-      "learning_rate": 8.923718592964825e-05,
-      "loss": 5.643,
-      "step": 11212
-    },
-    {
-      "epoch": 5.847718383311603,
-      "grad_norm": 1.5895634889602661,
-      "learning_rate": 8.923618090452261e-05,
-      "loss": 5.665,
-      "step": 11213
-    },
-    {
-      "epoch": 5.848239895697523,
-      "grad_norm": 1.5642768144607544,
-      "learning_rate": 8.923517587939699e-05,
-      "loss": 5.2479,
-      "step": 11214
-    },
-    {
-      "epoch": 5.848761408083442,
-      "grad_norm": 1.454853892326355,
-      "learning_rate": 8.923417085427136e-05,
-      "loss": 5.3818,
-      "step": 11215
-    },
-    {
-      "epoch": 5.849282920469361,
-      "grad_norm": 1.4737509489059448,
-      "learning_rate": 8.923316582914573e-05,
-      "loss": 5.5742,
-      "step": 11216
-    },
-    {
-      "epoch": 5.84980443285528,
-      "grad_norm": 1.7577749490737915,
-      "learning_rate": 8.92321608040201e-05,
-      "loss": 5.4552,
-      "step": 11217
-    },
-    {
-      "epoch": 5.8503259452412,
-      "grad_norm": 1.448093056678772,
-      "learning_rate": 8.923115577889448e-05,
-      "loss": 5.3674,
-      "step": 11218
-    },
-    {
-      "epoch": 5.850847457627118,
-      "grad_norm": 1.6210650205612183,
-      "learning_rate": 8.923015075376885e-05,
-      "loss": 5.4763,
-      "step": 11219
-    },
-    {
-      "epoch": 5.851368970013038,
-      "grad_norm": 1.369286298751831,
-      "learning_rate": 8.922914572864322e-05,
-      "loss": 5.8898,
-      "step": 11220
-    },
-    {
-      "epoch": 5.851890482398957,
-      "grad_norm": 1.4452582597732544,
-      "learning_rate": 8.92281407035176e-05,
-      "loss": 5.7388,
-      "step": 11221
-    },
-    {
-      "epoch": 5.852411994784876,
-      "grad_norm": 1.384647250175476,
-      "learning_rate": 8.922713567839196e-05,
-      "loss": 5.7958,
-      "step": 11222
-    },
-    {
-      "epoch": 5.852933507170795,
-      "grad_norm": 1.3613686561584473,
-      "learning_rate": 8.922613065326634e-05,
-      "loss": 5.9357,
-      "step": 11223
-    },
-    {
-      "epoch": 5.853455019556715,
-      "grad_norm": 1.3587231636047363,
-      "learning_rate": 8.92251256281407e-05,
-      "loss": 6.0704,
-      "step": 11224
-    },
-    {
-      "epoch": 5.853976531942633,
-      "grad_norm": 1.5404865741729736,
-      "learning_rate": 8.922412060301508e-05,
-      "loss": 5.6551,
-      "step": 11225
-    },
-    {
-      "epoch": 5.854498044328553,
-      "grad_norm": 1.4239871501922607,
-      "learning_rate": 8.922311557788945e-05,
-      "loss": 5.2827,
-      "step": 11226
-    },
-    {
-      "epoch": 5.855019556714472,
-      "grad_norm": 1.525867223739624,
-      "learning_rate": 8.922211055276382e-05,
-      "loss": 5.7399,
-      "step": 11227
-    },
-    {
-      "epoch": 5.855541069100391,
-      "grad_norm": 1.5702109336853027,
-      "learning_rate": 8.922110552763819e-05,
-      "loss": 5.0989,
-      "step": 11228
-    },
-    {
-      "epoch": 5.85606258148631,
-      "grad_norm": 1.4979186058044434,
-      "learning_rate": 8.922010050251257e-05,
-      "loss": 5.8947,
-      "step": 11229
-    },
-    {
-      "epoch": 5.85658409387223,
-      "grad_norm": 1.5616304874420166,
-      "learning_rate": 8.921909547738694e-05,
-      "loss": 5.3075,
-      "step": 11230
-    },
-    {
-      "epoch": 5.857105606258148,
-      "grad_norm": 1.3542653322219849,
-      "learning_rate": 8.921809045226132e-05,
-      "loss": 5.8263,
-      "step": 11231
-    },
-    {
-      "epoch": 5.857627118644068,
-      "grad_norm": 1.4963207244873047,
-      "learning_rate": 8.921708542713569e-05,
-      "loss": 5.6241,
-      "step": 11232
-    },
-    {
-      "epoch": 5.858148631029987,
-      "grad_norm": 1.391697883605957,
-      "learning_rate": 8.921608040201005e-05,
-      "loss": 5.8542,
-      "step": 11233
-    },
-    {
-      "epoch": 5.858670143415906,
-      "grad_norm": 1.4556691646575928,
-      "learning_rate": 8.921507537688443e-05,
-      "loss": 5.5906,
-      "step": 11234
-    },
-    {
-      "epoch": 5.859191655801825,
-      "grad_norm": 1.3006312847137451,
-      "learning_rate": 8.921407035175879e-05,
-      "loss": 5.8226,
-      "step": 11235
-    },
-    {
-      "epoch": 5.859713168187745,
-      "grad_norm": 1.2992162704467773,
-      "learning_rate": 8.921306532663317e-05,
-      "loss": 4.2548,
-      "step": 11236
-    },
-    {
-      "epoch": 5.860234680573663,
-      "grad_norm": 1.440169095993042,
-      "learning_rate": 8.921206030150753e-05,
-      "loss": 5.7552,
-      "step": 11237
-    },
-    {
-      "epoch": 5.860756192959583,
-      "grad_norm": 1.4144957065582275,
-      "learning_rate": 8.921105527638191e-05,
-      "loss": 5.8141,
-      "step": 11238
-    },
-    {
-      "epoch": 5.861277705345502,
-      "grad_norm": 1.377417802810669,
-      "learning_rate": 8.921005025125629e-05,
-      "loss": 5.5762,
-      "step": 11239
-    },
-    {
-      "epoch": 5.861799217731421,
-      "grad_norm": 1.4270271062850952,
-      "learning_rate": 8.920904522613067e-05,
-      "loss": 5.6306,
-      "step": 11240
-    },
-    {
-      "epoch": 5.86232073011734,
-      "grad_norm": 1.423186182975769,
-      "learning_rate": 8.920804020100503e-05,
-      "loss": 5.8154,
-      "step": 11241
-    },
-    {
-      "epoch": 5.86284224250326,
-      "grad_norm": 1.482991337776184,
-      "learning_rate": 8.920703517587941e-05,
-      "loss": 5.5338,
-      "step": 11242
-    },
-    {
-      "epoch": 5.863363754889178,
-      "grad_norm": 1.467436671257019,
-      "learning_rate": 8.920603015075377e-05,
-      "loss": 5.561,
-      "step": 11243
-    },
-    {
-      "epoch": 5.863885267275098,
-      "grad_norm": 1.4468964338302612,
-      "learning_rate": 8.920502512562815e-05,
-      "loss": 5.1277,
-      "step": 11244
-    },
-    {
-      "epoch": 5.864406779661017,
-      "grad_norm": 1.5056766271591187,
-      "learning_rate": 8.920402010050252e-05,
-      "loss": 5.5974,
-      "step": 11245
-    },
-    {
-      "epoch": 5.864928292046936,
-      "grad_norm": 1.5367873907089233,
-      "learning_rate": 8.920301507537688e-05,
-      "loss": 5.5052,
-      "step": 11246
-    },
-    {
-      "epoch": 5.865449804432855,
-      "grad_norm": 1.5522658824920654,
-      "learning_rate": 8.920201005025126e-05,
-      "loss": 5.4104,
-      "step": 11247
-    },
-    {
-      "epoch": 5.865971316818775,
-      "grad_norm": 1.349218487739563,
-      "learning_rate": 8.920100502512562e-05,
-      "loss": 5.9719,
-      "step": 11248
-    },
-    {
-      "epoch": 5.866492829204693,
-      "grad_norm": 1.5161595344543457,
-      "learning_rate": 8.92e-05,
-      "loss": 5.4518,
-      "step": 11249
-    },
-    {
-      "epoch": 5.867014341590613,
-      "grad_norm": 1.3313595056533813,
-      "learning_rate": 8.919899497487438e-05,
-      "loss": 5.4305,
-      "step": 11250
-    },
-    {
-      "epoch": 5.867535853976532,
-      "grad_norm": 1.5281251668930054,
-      "learning_rate": 8.919798994974876e-05,
-      "loss": 4.5744,
-      "step": 11251
-    },
-    {
-      "epoch": 5.868057366362451,
-      "grad_norm": 1.6386549472808838,
-      "learning_rate": 8.919698492462312e-05,
-      "loss": 5.1994,
-      "step": 11252
-    },
-    {
-      "epoch": 5.86857887874837,
-      "grad_norm": 1.4716168642044067,
-      "learning_rate": 8.91959798994975e-05,
-      "loss": 5.5595,
-      "step": 11253
-    },
-    {
-      "epoch": 5.869100391134289,
-      "grad_norm": 1.388664722442627,
-      "learning_rate": 8.919497487437186e-05,
-      "loss": 5.6521,
-      "step": 11254
-    },
-    {
-      "epoch": 5.869621903520208,
-      "grad_norm": 1.4519238471984863,
-      "learning_rate": 8.919396984924624e-05,
-      "loss": 5.3899,
-      "step": 11255
-    },
-    {
-      "epoch": 5.870143415906128,
-      "grad_norm": 1.4538769721984863,
-      "learning_rate": 8.91929648241206e-05,
-      "loss": 5.6714,
-      "step": 11256
-    },
-    {
-      "epoch": 5.870664928292047,
-      "grad_norm": 1.553675889968872,
-      "learning_rate": 8.919195979899498e-05,
-      "loss": 5.2295,
-      "step": 11257
-    },
-    {
-      "epoch": 5.871186440677966,
-      "grad_norm": 1.4229531288146973,
-      "learning_rate": 8.919095477386935e-05,
-      "loss": 5.37,
-      "step": 11258
-    },
-    {
-      "epoch": 5.871707953063885,
-      "grad_norm": 1.385155200958252,
-      "learning_rate": 8.918994974874372e-05,
-      "loss": 5.898,
-      "step": 11259
-    },
-    {
-      "epoch": 5.872229465449804,
-      "grad_norm": 1.3691452741622925,
-      "learning_rate": 8.91889447236181e-05,
-      "loss": 5.5142,
-      "step": 11260
-    },
-    {
-      "epoch": 5.872750977835723,
-      "grad_norm": 1.5023103952407837,
-      "learning_rate": 8.918793969849247e-05,
-      "loss": 5.5344,
-      "step": 11261
-    },
-    {
-      "epoch": 5.873272490221643,
-      "grad_norm": 1.3645367622375488,
-      "learning_rate": 8.918693467336684e-05,
-      "loss": 5.6034,
-      "step": 11262
-    },
-    {
-      "epoch": 5.873794002607562,
-      "grad_norm": 1.3820292949676514,
-      "learning_rate": 8.918592964824121e-05,
-      "loss": 5.9078,
-      "step": 11263
-    },
-    {
-      "epoch": 5.874315514993481,
-      "grad_norm": 1.4005110263824463,
-      "learning_rate": 8.918492462311559e-05,
-      "loss": 5.4086,
-      "step": 11264
-    },
-    {
-      "epoch": 5.8748370273794,
-      "grad_norm": 1.4512995481491089,
-      "learning_rate": 8.918391959798995e-05,
-      "loss": 5.5766,
-      "step": 11265
-    },
-    {
-      "epoch": 5.875358539765319,
-      "grad_norm": 1.5693503618240356,
-      "learning_rate": 8.918291457286433e-05,
-      "loss": 5.857,
-      "step": 11266
-    },
-    {
-      "epoch": 5.875880052151238,
-      "grad_norm": 1.3535808324813843,
-      "learning_rate": 8.918190954773869e-05,
-      "loss": 5.8553,
-      "step": 11267
-    },
-    {
-      "epoch": 5.876401564537158,
-      "grad_norm": 1.4316158294677734,
-      "learning_rate": 8.918090452261307e-05,
-      "loss": 5.763,
-      "step": 11268
-    },
-    {
-      "epoch": 5.876923076923077,
-      "grad_norm": 1.4741361141204834,
-      "learning_rate": 8.917989949748743e-05,
-      "loss": 5.3904,
-      "step": 11269
-    },
-    {
-      "epoch": 5.877444589308996,
-      "grad_norm": 1.6620171070098877,
-      "learning_rate": 8.917889447236181e-05,
-      "loss": 5.3642,
-      "step": 11270
-    },
-    {
-      "epoch": 5.877966101694915,
-      "grad_norm": 1.3702480792999268,
-      "learning_rate": 8.917788944723619e-05,
-      "loss": 5.7901,
-      "step": 11271
-    },
-    {
-      "epoch": 5.878487614080834,
-      "grad_norm": 1.4051774740219116,
-      "learning_rate": 8.917688442211055e-05,
-      "loss": 5.7447,
-      "step": 11272
-    },
-    {
-      "epoch": 5.879009126466753,
-      "grad_norm": 1.4597913026809692,
-      "learning_rate": 8.917587939698493e-05,
-      "loss": 5.5276,
-      "step": 11273
-    },
-    {
-      "epoch": 5.879530638852673,
-      "grad_norm": 1.403415322303772,
-      "learning_rate": 8.91748743718593e-05,
-      "loss": 5.4252,
-      "step": 11274
-    },
-    {
-      "epoch": 5.880052151238592,
-      "grad_norm": 1.4147672653198242,
-      "learning_rate": 8.917386934673367e-05,
-      "loss": 5.6639,
-      "step": 11275
-    },
-    {
-      "epoch": 5.880573663624511,
-      "grad_norm": 1.2771334648132324,
-      "learning_rate": 8.917286432160804e-05,
-      "loss": 6.0458,
-      "step": 11276
-    },
-    {
-      "epoch": 5.88109517601043,
-      "grad_norm": 1.4177157878875732,
-      "learning_rate": 8.917185929648242e-05,
-      "loss": 6.0046,
-      "step": 11277
-    },
-    {
-      "epoch": 5.881616688396349,
-      "grad_norm": 1.548216700553894,
-      "learning_rate": 8.917085427135678e-05,
-      "loss": 5.2175,
-      "step": 11278
-    },
-    {
-      "epoch": 5.8821382007822685,
-      "grad_norm": 1.3531478643417358,
-      "learning_rate": 8.916984924623116e-05,
-      "loss": 5.7483,
-      "step": 11279
-    },
-    {
-      "epoch": 5.882659713168188,
-      "grad_norm": 1.3845802545547485,
-      "learning_rate": 8.916884422110554e-05,
-      "loss": 5.6746,
-      "step": 11280
-    },
-    {
-      "epoch": 5.883181225554107,
-      "grad_norm": 1.51714026927948,
-      "learning_rate": 8.916783919597991e-05,
-      "loss": 5.4076,
-      "step": 11281
-    },
-    {
-      "epoch": 5.883702737940026,
-      "grad_norm": 1.4058209657669067,
-      "learning_rate": 8.916683417085428e-05,
-      "loss": 5.9292,
-      "step": 11282
-    },
-    {
-      "epoch": 5.884224250325945,
-      "grad_norm": 1.4360978603363037,
-      "learning_rate": 8.916582914572866e-05,
-      "loss": 5.7357,
-      "step": 11283
-    },
-    {
-      "epoch": 5.884745762711864,
-      "grad_norm": 1.4907541275024414,
-      "learning_rate": 8.916482412060302e-05,
-      "loss": 5.8058,
-      "step": 11284
-    },
-    {
-      "epoch": 5.8852672750977835,
-      "grad_norm": 1.4319883584976196,
-      "learning_rate": 8.91638190954774e-05,
-      "loss": 5.9572,
-      "step": 11285
-    },
-    {
-      "epoch": 5.885788787483703,
-      "grad_norm": 1.5984867811203003,
-      "learning_rate": 8.916281407035176e-05,
-      "loss": 5.3557,
-      "step": 11286
-    },
-    {
-      "epoch": 5.886310299869622,
-      "grad_norm": 1.3156123161315918,
-      "learning_rate": 8.916180904522613e-05,
-      "loss": 5.4029,
-      "step": 11287
-    },
-    {
-      "epoch": 5.886831812255541,
-      "grad_norm": 1.5979077816009521,
-      "learning_rate": 8.91608040201005e-05,
-      "loss": 5.324,
-      "step": 11288
-    },
-    {
-      "epoch": 5.88735332464146,
-      "grad_norm": 1.466156005859375,
-      "learning_rate": 8.915979899497487e-05,
-      "loss": 5.6915,
-      "step": 11289
-    },
-    {
-      "epoch": 5.887874837027379,
-      "grad_norm": 1.3098756074905396,
-      "learning_rate": 8.915879396984925e-05,
-      "loss": 4.8949,
-      "step": 11290
-    },
-    {
-      "epoch": 5.8883963494132985,
-      "grad_norm": 1.3836251497268677,
-      "learning_rate": 8.915778894472362e-05,
-      "loss": 5.6736,
-      "step": 11291
-    },
-    {
-      "epoch": 5.888917861799218,
-      "grad_norm": 1.4004181623458862,
-      "learning_rate": 8.9156783919598e-05,
-      "loss": 5.2314,
-      "step": 11292
-    },
-    {
-      "epoch": 5.8894393741851365,
-      "grad_norm": 1.5102849006652832,
-      "learning_rate": 8.915577889447237e-05,
-      "loss": 5.551,
-      "step": 11293
-    },
-    {
-      "epoch": 5.889960886571056,
-      "grad_norm": 1.3285257816314697,
-      "learning_rate": 8.915477386934674e-05,
-      "loss": 5.7873,
-      "step": 11294
-    },
-    {
-      "epoch": 5.8904823989569755,
-      "grad_norm": 1.5043864250183105,
-      "learning_rate": 8.915376884422111e-05,
-      "loss": 5.5391,
-      "step": 11295
-    },
-    {
-      "epoch": 5.891003911342894,
-      "grad_norm": 1.4117249250411987,
-      "learning_rate": 8.915276381909549e-05,
-      "loss": 5.5583,
-      "step": 11296
-    },
-    {
-      "epoch": 5.8915254237288135,
-      "grad_norm": 1.5790517330169678,
-      "learning_rate": 8.915175879396985e-05,
-      "loss": 5.3185,
-      "step": 11297
-    },
-    {
-      "epoch": 5.892046936114733,
-      "grad_norm": 1.515542984008789,
-      "learning_rate": 8.915075376884423e-05,
-      "loss": 5.4032,
-      "step": 11298
-    },
-    {
-      "epoch": 5.8925684485006515,
-      "grad_norm": 1.3816111087799072,
-      "learning_rate": 8.914974874371859e-05,
-      "loss": 5.5197,
-      "step": 11299
-    },
-    {
-      "epoch": 5.893089960886571,
-      "grad_norm": 1.368752360343933,
-      "learning_rate": 8.914874371859297e-05,
-      "loss": 5.797,
-      "step": 11300
-    },
-    {
-      "epoch": 5.8936114732724905,
-      "grad_norm": 1.4097065925598145,
-      "learning_rate": 8.914773869346735e-05,
-      "loss": 5.2533,
-      "step": 11301
-    },
-    {
-      "epoch": 5.894132985658409,
-      "grad_norm": 1.377744197845459,
-      "learning_rate": 8.914673366834171e-05,
-      "loss": 5.6341,
-      "step": 11302
-    },
-    {
-      "epoch": 5.8946544980443285,
-      "grad_norm": 1.3712315559387207,
-      "learning_rate": 8.914572864321609e-05,
-      "loss": 5.629,
-      "step": 11303
-    },
-    {
-      "epoch": 5.895176010430248,
-      "grad_norm": 1.3958592414855957,
-      "learning_rate": 8.914472361809046e-05,
-      "loss": 5.5864,
-      "step": 11304
-    },
-    {
-      "epoch": 5.8956975228161665,
-      "grad_norm": 1.4758470058441162,
-      "learning_rate": 8.914371859296483e-05,
-      "loss": 5.3796,
-      "step": 11305
-    },
-    {
-      "epoch": 5.896219035202086,
-      "grad_norm": 1.36129891872406,
-      "learning_rate": 8.91427135678392e-05,
-      "loss": 5.9166,
-      "step": 11306
-    },
-    {
-      "epoch": 5.8967405475880055,
-      "grad_norm": 1.490878939628601,
-      "learning_rate": 8.914170854271357e-05,
-      "loss": 5.418,
-      "step": 11307
-    },
-    {
-      "epoch": 5.897262059973924,
-      "grad_norm": 1.5105335712432861,
-      "learning_rate": 8.914070351758794e-05,
-      "loss": 5.5746,
-      "step": 11308
-    },
-    {
-      "epoch": 5.8977835723598435,
-      "grad_norm": 1.7786872386932373,
-      "learning_rate": 8.913969849246232e-05,
-      "loss": 5.7376,
-      "step": 11309
-    },
-    {
-      "epoch": 5.898305084745763,
-      "grad_norm": 1.5075725317001343,
-      "learning_rate": 8.913869346733668e-05,
-      "loss": 5.5121,
-      "step": 11310
-    },
-    {
-      "epoch": 5.898826597131682,
-      "grad_norm": 1.4755496978759766,
-      "learning_rate": 8.913768844221106e-05,
-      "loss": 5.7465,
-      "step": 11311
-    },
-    {
-      "epoch": 5.899348109517601,
-      "grad_norm": 1.4680331945419312,
-      "learning_rate": 8.913668341708544e-05,
-      "loss": 5.707,
-      "step": 11312
-    },
-    {
-      "epoch": 5.8998696219035205,
-      "grad_norm": 1.4212799072265625,
-      "learning_rate": 8.91356783919598e-05,
-      "loss": 5.7232,
-      "step": 11313
-    },
-    {
-      "epoch": 5.900391134289439,
-      "grad_norm": 1.486430048942566,
-      "learning_rate": 8.913467336683418e-05,
-      "loss": 5.2711,
-      "step": 11314
-    },
-    {
-      "epoch": 5.9009126466753585,
-      "grad_norm": 1.5364068746566772,
-      "learning_rate": 8.913366834170854e-05,
-      "loss": 5.7329,
-      "step": 11315
-    },
-    {
-      "epoch": 5.901434159061278,
-      "grad_norm": 1.506272554397583,
-      "learning_rate": 8.913266331658292e-05,
-      "loss": 5.5568,
-      "step": 11316
-    },
-    {
-      "epoch": 5.901955671447197,
-      "grad_norm": 1.6023550033569336,
-      "learning_rate": 8.913165829145729e-05,
-      "loss": 5.5899,
-      "step": 11317
-    },
-    {
-      "epoch": 5.902477183833116,
-      "grad_norm": 1.4245072603225708,
-      "learning_rate": 8.913065326633166e-05,
-      "loss": 5.377,
-      "step": 11318
-    },
-    {
-      "epoch": 5.9029986962190355,
-      "grad_norm": 1.4291298389434814,
-      "learning_rate": 8.912964824120603e-05,
-      "loss": 5.7934,
-      "step": 11319
-    },
-    {
-      "epoch": 5.903520208604954,
-      "grad_norm": 1.518874168395996,
-      "learning_rate": 8.91286432160804e-05,
-      "loss": 5.4947,
-      "step": 11320
-    },
-    {
-      "epoch": 5.9040417209908735,
-      "grad_norm": 1.404563546180725,
-      "learning_rate": 8.912763819095477e-05,
-      "loss": 5.5883,
-      "step": 11321
-    },
-    {
-      "epoch": 5.904563233376793,
-      "grad_norm": 1.5758540630340576,
-      "learning_rate": 8.912663316582915e-05,
-      "loss": 5.316,
-      "step": 11322
-    },
-    {
-      "epoch": 5.905084745762712,
-      "grad_norm": 1.3868333101272583,
-      "learning_rate": 8.912562814070353e-05,
-      "loss": 5.6828,
-      "step": 11323
-    },
-    {
-      "epoch": 5.905606258148631,
-      "grad_norm": 1.342061996459961,
-      "learning_rate": 8.91246231155779e-05,
-      "loss": 5.4567,
-      "step": 11324
-    },
-    {
-      "epoch": 5.9061277705345505,
-      "grad_norm": 1.5016229152679443,
-      "learning_rate": 8.912361809045227e-05,
-      "loss": 5.226,
-      "step": 11325
-    },
-    {
-      "epoch": 5.906649282920469,
-      "grad_norm": 1.6057603359222412,
-      "learning_rate": 8.912261306532663e-05,
-      "loss": 5.4608,
-      "step": 11326
-    },
-    {
-      "epoch": 5.9071707953063886,
-      "grad_norm": 1.3618230819702148,
-      "learning_rate": 8.912160804020101e-05,
-      "loss": 5.5455,
-      "step": 11327
-    },
-    {
-      "epoch": 5.907692307692308,
-      "grad_norm": 1.423213005065918,
-      "learning_rate": 8.912060301507537e-05,
-      "loss": 5.6373,
-      "step": 11328
-    },
-    {
-      "epoch": 5.908213820078227,
-      "grad_norm": 1.6753309965133667,
-      "learning_rate": 8.911959798994975e-05,
-      "loss": 5.2058,
-      "step": 11329
-    },
-    {
-      "epoch": 5.908735332464146,
-      "grad_norm": 1.3807997703552246,
-      "learning_rate": 8.911859296482412e-05,
-      "loss": 5.8074,
-      "step": 11330
-    },
-    {
-      "epoch": 5.9092568448500655,
-      "grad_norm": 1.4062408208847046,
-      "learning_rate": 8.91175879396985e-05,
-      "loss": 5.9126,
-      "step": 11331
-    },
-    {
-      "epoch": 5.909778357235984,
-      "grad_norm": 1.4261136054992676,
-      "learning_rate": 8.911658291457287e-05,
-      "loss": 5.7805,
-      "step": 11332
-    },
-    {
-      "epoch": 5.910299869621904,
-      "grad_norm": 1.5612930059432983,
-      "learning_rate": 8.911557788944725e-05,
-      "loss": 5.5183,
-      "step": 11333
-    },
-    {
-      "epoch": 5.910821382007823,
-      "grad_norm": 1.4853062629699707,
-      "learning_rate": 8.911457286432161e-05,
-      "loss": 5.7412,
-      "step": 11334
-    },
-    {
-      "epoch": 5.911342894393742,
-      "grad_norm": 1.520404577255249,
-      "learning_rate": 8.911356783919599e-05,
-      "loss": 5.9055,
-      "step": 11335
-    },
-    {
-      "epoch": 5.911864406779661,
-      "grad_norm": 1.4446961879730225,
-      "learning_rate": 8.911256281407036e-05,
-      "loss": 5.5019,
-      "step": 11336
-    },
-    {
-      "epoch": 5.9123859191655805,
-      "grad_norm": 1.4221844673156738,
-      "learning_rate": 8.911155778894473e-05,
-      "loss": 5.4092,
-      "step": 11337
-    },
-    {
-      "epoch": 5.912907431551499,
-      "grad_norm": 1.3140445947647095,
-      "learning_rate": 8.91105527638191e-05,
-      "loss": 5.9498,
-      "step": 11338
-    },
-    {
-      "epoch": 5.913428943937419,
-      "grad_norm": 1.4865093231201172,
-      "learning_rate": 8.910954773869346e-05,
-      "loss": 5.9765,
-      "step": 11339
-    },
-    {
-      "epoch": 5.913950456323338,
-      "grad_norm": 1.4483637809753418,
-      "learning_rate": 8.910854271356784e-05,
-      "loss": 5.9316,
-      "step": 11340
-    },
-    {
-      "epoch": 5.914471968709257,
-      "grad_norm": 1.4083068370819092,
-      "learning_rate": 8.91075376884422e-05,
-      "loss": 5.9515,
-      "step": 11341
-    },
-    {
-      "epoch": 5.914993481095176,
-      "grad_norm": 1.5300514698028564,
-      "learning_rate": 8.910653266331658e-05,
-      "loss": 5.5368,
-      "step": 11342
-    },
-    {
-      "epoch": 5.9155149934810956,
-      "grad_norm": 1.472047209739685,
-      "learning_rate": 8.910552763819096e-05,
-      "loss": 5.6447,
-      "step": 11343
-    },
-    {
-      "epoch": 5.916036505867014,
-      "grad_norm": 1.4264626502990723,
-      "learning_rate": 8.910452261306534e-05,
-      "loss": 5.7227,
-      "step": 11344
-    },
-    {
-      "epoch": 5.916558018252934,
-      "grad_norm": 1.372085690498352,
-      "learning_rate": 8.91035175879397e-05,
-      "loss": 5.4932,
-      "step": 11345
-    },
-    {
-      "epoch": 5.917079530638853,
-      "grad_norm": 1.514417052268982,
-      "learning_rate": 8.910251256281408e-05,
-      "loss": 5.5076,
-      "step": 11346
-    },
-    {
-      "epoch": 5.917601043024772,
-      "grad_norm": 1.5213464498519897,
-      "learning_rate": 8.910150753768844e-05,
-      "loss": 5.4914,
-      "step": 11347
-    },
-    {
-      "epoch": 5.918122555410691,
-      "grad_norm": 1.5110071897506714,
-      "learning_rate": 8.910050251256282e-05,
-      "loss": 5.8044,
-      "step": 11348
-    },
-    {
-      "epoch": 5.91864406779661,
-      "grad_norm": 1.4245574474334717,
-      "learning_rate": 8.909949748743719e-05,
-      "loss": 5.507,
-      "step": 11349
-    },
-    {
-      "epoch": 5.919165580182529,
-      "grad_norm": 1.375240445137024,
-      "learning_rate": 8.909849246231156e-05,
-      "loss": 5.8026,
-      "step": 11350
-    },
-    {
-      "epoch": 5.919687092568449,
-      "grad_norm": 1.4586305618286133,
-      "learning_rate": 8.909748743718593e-05,
-      "loss": 5.9376,
-      "step": 11351
-    },
-    {
-      "epoch": 5.920208604954368,
-      "grad_norm": 1.4838114976882935,
-      "learning_rate": 8.90964824120603e-05,
-      "loss": 5.8598,
-      "step": 11352
-    },
-    {
-      "epoch": 5.920730117340287,
-      "grad_norm": 1.3929152488708496,
-      "learning_rate": 8.909547738693468e-05,
-      "loss": 5.7718,
-      "step": 11353
-    },
-    {
-      "epoch": 5.921251629726206,
-      "grad_norm": 1.4700287580490112,
-      "learning_rate": 8.909447236180905e-05,
-      "loss": 5.7076,
-      "step": 11354
-    },
-    {
-      "epoch": 5.921773142112125,
-      "grad_norm": 1.4567495584487915,
-      "learning_rate": 8.909346733668343e-05,
-      "loss": 5.7887,
-      "step": 11355
-    },
-    {
-      "epoch": 5.922294654498044,
-      "grad_norm": 1.4027135372161865,
-      "learning_rate": 8.909246231155779e-05,
-      "loss": 5.7896,
-      "step": 11356
-    },
-    {
-      "epoch": 5.922816166883964,
-      "grad_norm": 1.3650051355361938,
-      "learning_rate": 8.909145728643217e-05,
-      "loss": 5.7979,
-      "step": 11357
-    },
-    {
-      "epoch": 5.923337679269883,
-      "grad_norm": 1.4105373620986938,
-      "learning_rate": 8.909045226130653e-05,
-      "loss": 5.5241,
-      "step": 11358
-    },
-    {
-      "epoch": 5.923859191655802,
-      "grad_norm": 1.3619812726974487,
-      "learning_rate": 8.908944723618091e-05,
-      "loss": 5.6616,
-      "step": 11359
-    },
-    {
-      "epoch": 5.924380704041721,
-      "grad_norm": 1.4172947406768799,
-      "learning_rate": 8.908844221105527e-05,
-      "loss": 5.9015,
-      "step": 11360
-    },
-    {
-      "epoch": 5.92490221642764,
-      "grad_norm": 1.6320081949234009,
-      "learning_rate": 8.908743718592965e-05,
-      "loss": 4.2789,
-      "step": 11361
-    },
-    {
-      "epoch": 5.925423728813559,
-      "grad_norm": 1.3815693855285645,
-      "learning_rate": 8.908643216080402e-05,
-      "loss": 5.9405,
-      "step": 11362
-    },
-    {
-      "epoch": 5.925945241199479,
-      "grad_norm": 1.435364007949829,
-      "learning_rate": 8.90854271356784e-05,
-      "loss": 5.6102,
-      "step": 11363
-    },
-    {
-      "epoch": 5.926466753585398,
-      "grad_norm": 1.3903636932373047,
-      "learning_rate": 8.908442211055277e-05,
-      "loss": 5.8981,
-      "step": 11364
-    },
-    {
-      "epoch": 5.926988265971317,
-      "grad_norm": 1.427901029586792,
-      "learning_rate": 8.908341708542715e-05,
-      "loss": 5.6857,
-      "step": 11365
-    },
-    {
-      "epoch": 5.927509778357236,
-      "grad_norm": 1.4351145029067993,
-      "learning_rate": 8.908241206030151e-05,
-      "loss": 5.8725,
-      "step": 11366
-    },
-    {
-      "epoch": 5.928031290743155,
-      "grad_norm": 1.4889357089996338,
-      "learning_rate": 8.908140703517588e-05,
-      "loss": 5.5546,
-      "step": 11367
-    },
-    {
-      "epoch": 5.928552803129074,
-      "grad_norm": 1.5373586416244507,
-      "learning_rate": 8.908040201005026e-05,
-      "loss": 5.6925,
-      "step": 11368
-    },
-    {
-      "epoch": 5.929074315514994,
-      "grad_norm": 1.474624752998352,
-      "learning_rate": 8.907939698492462e-05,
-      "loss": 5.5231,
-      "step": 11369
-    },
-    {
-      "epoch": 5.929595827900913,
-      "grad_norm": 1.4162204265594482,
-      "learning_rate": 8.9078391959799e-05,
-      "loss": 5.8624,
-      "step": 11370
-    },
-    {
-      "epoch": 5.930117340286832,
-      "grad_norm": 1.4377985000610352,
-      "learning_rate": 8.907738693467336e-05,
-      "loss": 5.496,
-      "step": 11371
-    },
-    {
-      "epoch": 5.930638852672751,
-      "grad_norm": 1.349130630493164,
-      "learning_rate": 8.907638190954774e-05,
-      "loss": 6.0411,
-      "step": 11372
-    },
-    {
-      "epoch": 5.93116036505867,
-      "grad_norm": 1.4302442073822021,
-      "learning_rate": 8.907537688442212e-05,
-      "loss": 5.5336,
-      "step": 11373
-    },
-    {
-      "epoch": 5.931681877444589,
-      "grad_norm": 1.4982818365097046,
-      "learning_rate": 8.90743718592965e-05,
-      "loss": 5.6895,
-      "step": 11374
-    },
-    {
-      "epoch": 5.932203389830509,
-      "grad_norm": 1.4246960878372192,
-      "learning_rate": 8.907336683417086e-05,
-      "loss": 5.567,
-      "step": 11375
-    },
-    {
-      "epoch": 5.932724902216428,
-      "grad_norm": 1.3444892168045044,
-      "learning_rate": 8.907236180904524e-05,
-      "loss": 5.4424,
-      "step": 11376
-    },
-    {
-      "epoch": 5.933246414602347,
-      "grad_norm": 1.3398813009262085,
-      "learning_rate": 8.90713567839196e-05,
-      "loss": 6.1753,
-      "step": 11377
-    },
-    {
-      "epoch": 5.933767926988266,
-      "grad_norm": 1.5791714191436768,
-      "learning_rate": 8.907035175879398e-05,
-      "loss": 5.5122,
-      "step": 11378
-    },
-    {
-      "epoch": 5.934289439374185,
-      "grad_norm": 1.46021568775177,
-      "learning_rate": 8.906934673366834e-05,
-      "loss": 5.779,
-      "step": 11379
-    },
-    {
-      "epoch": 5.934810951760104,
-      "grad_norm": 1.4690427780151367,
-      "learning_rate": 8.906834170854271e-05,
-      "loss": 5.5667,
-      "step": 11380
-    },
-    {
-      "epoch": 5.935332464146024,
-      "grad_norm": 1.5943454504013062,
-      "learning_rate": 8.906733668341709e-05,
-      "loss": 5.5241,
-      "step": 11381
-    },
-    {
-      "epoch": 5.935853976531943,
-      "grad_norm": 1.4608924388885498,
-      "learning_rate": 8.906633165829145e-05,
-      "loss": 6.0709,
-      "step": 11382
-    },
-    {
-      "epoch": 5.936375488917862,
-      "grad_norm": 1.455690860748291,
-      "learning_rate": 8.906532663316583e-05,
-      "loss": 5.777,
-      "step": 11383
-    },
-    {
-      "epoch": 5.936897001303781,
-      "grad_norm": 1.5414828062057495,
-      "learning_rate": 8.906432160804021e-05,
-      "loss": 5.2531,
-      "step": 11384
-    },
-    {
-      "epoch": 5.9374185136897,
-      "grad_norm": 1.4874972105026245,
-      "learning_rate": 8.906331658291458e-05,
-      "loss": 5.3322,
-      "step": 11385
-    },
-    {
-      "epoch": 5.937940026075619,
-      "grad_norm": 1.8665611743927002,
-      "learning_rate": 8.906231155778895e-05,
-      "loss": 5.5093,
-      "step": 11386
-    },
-    {
-      "epoch": 5.938461538461539,
-      "grad_norm": 1.4702199697494507,
-      "learning_rate": 8.906130653266333e-05,
-      "loss": 5.055,
-      "step": 11387
-    },
-    {
-      "epoch": 5.938983050847457,
-      "grad_norm": 1.4513535499572754,
-      "learning_rate": 8.906030150753769e-05,
-      "loss": 5.7584,
-      "step": 11388
-    },
-    {
-      "epoch": 5.939504563233377,
-      "grad_norm": 1.4360400438308716,
-      "learning_rate": 8.905929648241207e-05,
-      "loss": 5.6066,
-      "step": 11389
-    },
-    {
-      "epoch": 5.940026075619296,
-      "grad_norm": 1.4236586093902588,
-      "learning_rate": 8.905829145728643e-05,
-      "loss": 5.6627,
-      "step": 11390
-    },
-    {
-      "epoch": 5.940547588005215,
-      "grad_norm": 1.449127435684204,
-      "learning_rate": 8.905728643216081e-05,
-      "loss": 5.5126,
-      "step": 11391
-    },
-    {
-      "epoch": 5.941069100391134,
-      "grad_norm": 1.6454875469207764,
-      "learning_rate": 8.905628140703518e-05,
-      "loss": 5.342,
-      "step": 11392
-    },
-    {
-      "epoch": 5.941590612777054,
-      "grad_norm": 1.440747618675232,
-      "learning_rate": 8.905527638190955e-05,
-      "loss": 6.0258,
-      "step": 11393
-    },
-    {
-      "epoch": 5.942112125162972,
-      "grad_norm": 1.4467829465866089,
-      "learning_rate": 8.905427135678393e-05,
-      "loss": 5.8607,
-      "step": 11394
-    },
-    {
-      "epoch": 5.942633637548892,
-      "grad_norm": 1.3585546016693115,
-      "learning_rate": 8.90532663316583e-05,
-      "loss": 5.9992,
-      "step": 11395
-    },
-    {
-      "epoch": 5.943155149934811,
-      "grad_norm": 1.42280912399292,
-      "learning_rate": 8.905226130653267e-05,
-      "loss": 5.7084,
-      "step": 11396
-    },
-    {
-      "epoch": 5.94367666232073,
-      "grad_norm": 1.3828580379486084,
-      "learning_rate": 8.905125628140704e-05,
-      "loss": 5.6464,
-      "step": 11397
-    },
-    {
-      "epoch": 5.944198174706649,
-      "grad_norm": 1.433448076248169,
-      "learning_rate": 8.905025125628142e-05,
-      "loss": 5.6937,
-      "step": 11398
-    },
-    {
-      "epoch": 5.944719687092569,
-      "grad_norm": 1.3970472812652588,
-      "learning_rate": 8.904924623115578e-05,
-      "loss": 5.6277,
-      "step": 11399
-    },
-    {
-      "epoch": 5.945241199478487,
-      "grad_norm": 1.5787118673324585,
-      "learning_rate": 8.904824120603016e-05,
-      "loss": 5.1386,
-      "step": 11400
-    },
-    {
-      "epoch": 5.945762711864407,
-      "grad_norm": 1.4131290912628174,
-      "learning_rate": 8.904723618090452e-05,
-      "loss": 5.4191,
-      "step": 11401
-    },
-    {
-      "epoch": 5.946284224250326,
-      "grad_norm": 1.4641848802566528,
-      "learning_rate": 8.90462311557789e-05,
-      "loss": 5.3758,
-      "step": 11402
-    },
-    {
-      "epoch": 5.946805736636245,
-      "grad_norm": 1.3387724161148071,
-      "learning_rate": 8.904522613065326e-05,
-      "loss": 5.0605,
-      "step": 11403
-    },
-    {
-      "epoch": 5.947327249022164,
-      "grad_norm": 1.5031414031982422,
-      "learning_rate": 8.904422110552764e-05,
-      "loss": 5.5768,
-      "step": 11404
-    },
-    {
-      "epoch": 5.947848761408084,
-      "grad_norm": 1.4257253408432007,
-      "learning_rate": 8.904321608040202e-05,
-      "loss": 5.3292,
-      "step": 11405
-    },
-    {
-      "epoch": 5.948370273794002,
-      "grad_norm": 1.5816388130187988,
-      "learning_rate": 8.904221105527638e-05,
-      "loss": 5.7983,
-      "step": 11406
-    },
-    {
-      "epoch": 5.948891786179922,
-      "grad_norm": 1.448049783706665,
-      "learning_rate": 8.904120603015076e-05,
-      "loss": 5.1479,
-      "step": 11407
-    },
-    {
-      "epoch": 5.949413298565841,
-      "grad_norm": 1.49215567111969,
-      "learning_rate": 8.904020100502513e-05,
-      "loss": 5.5868,
-      "step": 11408
-    },
-    {
-      "epoch": 5.94993481095176,
-      "grad_norm": 1.53244948387146,
-      "learning_rate": 8.90391959798995e-05,
-      "loss": 6.0492,
-      "step": 11409
-    },
-    {
-      "epoch": 5.950456323337679,
-      "grad_norm": 1.6240018606185913,
-      "learning_rate": 8.903819095477387e-05,
-      "loss": 5.2907,
-      "step": 11410
-    },
-    {
-      "epoch": 5.950977835723599,
-      "grad_norm": 1.397289752960205,
-      "learning_rate": 8.903718592964825e-05,
-      "loss": 5.5432,
-      "step": 11411
-    },
-    {
-      "epoch": 5.951499348109517,
-      "grad_norm": 1.4843603372573853,
-      "learning_rate": 8.903618090452261e-05,
-      "loss": 5.1018,
-      "step": 11412
-    },
-    {
-      "epoch": 5.952020860495437,
-      "grad_norm": 1.651544213294983,
-      "learning_rate": 8.903517587939699e-05,
-      "loss": 5.0194,
-      "step": 11413
-    },
-    {
-      "epoch": 5.952542372881356,
-      "grad_norm": 1.4784749746322632,
-      "learning_rate": 8.903417085427137e-05,
-      "loss": 5.8767,
-      "step": 11414
-    },
-    {
-      "epoch": 5.953063885267275,
-      "grad_norm": 1.4332085847854614,
-      "learning_rate": 8.903316582914574e-05,
-      "loss": 5.7246,
-      "step": 11415
-    },
-    {
-      "epoch": 5.953585397653194,
-      "grad_norm": 1.6198838949203491,
-      "learning_rate": 8.903216080402011e-05,
-      "loss": 5.7252,
-      "step": 11416
-    },
-    {
-      "epoch": 5.954106910039114,
-      "grad_norm": 1.6024205684661865,
-      "learning_rate": 8.903115577889449e-05,
-      "loss": 5.3997,
-      "step": 11417
-    },
-    {
-      "epoch": 5.954628422425032,
-      "grad_norm": 1.45270574092865,
-      "learning_rate": 8.903015075376885e-05,
-      "loss": 5.579,
-      "step": 11418
-    },
-    {
-      "epoch": 5.955149934810952,
-      "grad_norm": 1.5584391355514526,
-      "learning_rate": 8.902914572864321e-05,
-      "loss": 4.9728,
-      "step": 11419
-    },
-    {
-      "epoch": 5.955671447196871,
-      "grad_norm": 1.6731576919555664,
-      "learning_rate": 8.902814070351759e-05,
-      "loss": 4.9593,
-      "step": 11420
-    },
-    {
-      "epoch": 5.95619295958279,
-      "grad_norm": 1.4699910879135132,
-      "learning_rate": 8.902713567839196e-05,
-      "loss": 5.5176,
-      "step": 11421
-    },
-    {
-      "epoch": 5.956714471968709,
-      "grad_norm": 1.4992434978485107,
-      "learning_rate": 8.902613065326633e-05,
-      "loss": 5.9015,
-      "step": 11422
-    },
-    {
-      "epoch": 5.957235984354629,
-      "grad_norm": 1.4685287475585938,
-      "learning_rate": 8.90251256281407e-05,
-      "loss": 5.6006,
-      "step": 11423
-    },
-    {
-      "epoch": 5.957757496740547,
-      "grad_norm": 1.4554989337921143,
-      "learning_rate": 8.902412060301508e-05,
-      "loss": 5.6344,
-      "step": 11424
-    },
-    {
-      "epoch": 5.958279009126467,
-      "grad_norm": 1.4735748767852783,
-      "learning_rate": 8.902311557788945e-05,
-      "loss": 5.0677,
-      "step": 11425
-    },
-    {
-      "epoch": 5.958800521512386,
-      "grad_norm": 1.4445641040802002,
-      "learning_rate": 8.902211055276383e-05,
-      "loss": 5.2927,
-      "step": 11426
-    },
-    {
-      "epoch": 5.959322033898305,
-      "grad_norm": 1.5215175151824951,
-      "learning_rate": 8.90211055276382e-05,
-      "loss": 5.6959,
-      "step": 11427
-    },
-    {
-      "epoch": 5.959843546284224,
-      "grad_norm": 1.54198157787323,
-      "learning_rate": 8.902010050251257e-05,
-      "loss": 5.5956,
-      "step": 11428
-    },
-    {
-      "epoch": 5.960365058670144,
-      "grad_norm": 1.460075855255127,
-      "learning_rate": 8.901909547738694e-05,
-      "loss": 5.5793,
-      "step": 11429
-    },
-    {
-      "epoch": 5.960886571056062,
-      "grad_norm": 1.3017932176589966,
-      "learning_rate": 8.901809045226132e-05,
-      "loss": 5.6999,
-      "step": 11430
-    },
-    {
-      "epoch": 5.961408083441982,
-      "grad_norm": 1.467511773109436,
-      "learning_rate": 8.901708542713568e-05,
-      "loss": 5.544,
-      "step": 11431
-    },
-    {
-      "epoch": 5.961929595827901,
-      "grad_norm": 1.4471137523651123,
-      "learning_rate": 8.901608040201004e-05,
-      "loss": 5.5961,
-      "step": 11432
-    },
-    {
-      "epoch": 5.96245110821382,
-      "grad_norm": 1.6525609493255615,
-      "learning_rate": 8.901507537688442e-05,
-      "loss": 5.1083,
-      "step": 11433
-    },
-    {
-      "epoch": 5.962972620599739,
-      "grad_norm": 1.4385781288146973,
-      "learning_rate": 8.90140703517588e-05,
-      "loss": 5.6165,
-      "step": 11434
-    },
-    {
-      "epoch": 5.963494132985659,
-      "grad_norm": 1.2901585102081299,
-      "learning_rate": 8.901306532663318e-05,
-      "loss": 6.1053,
-      "step": 11435
-    },
-    {
-      "epoch": 5.964015645371577,
-      "grad_norm": 1.438503623008728,
-      "learning_rate": 8.901206030150754e-05,
-      "loss": 5.821,
-      "step": 11436
-    },
-    {
-      "epoch": 5.964537157757497,
-      "grad_norm": 1.5674656629562378,
-      "learning_rate": 8.901105527638192e-05,
-      "loss": 5.1993,
-      "step": 11437
-    },
-    {
-      "epoch": 5.965058670143416,
-      "grad_norm": 1.355259656906128,
-      "learning_rate": 8.901005025125628e-05,
-      "loss": 5.4785,
-      "step": 11438
-    },
-    {
-      "epoch": 5.965580182529335,
-      "grad_norm": 1.4853551387786865,
-      "learning_rate": 8.900904522613066e-05,
-      "loss": 5.683,
-      "step": 11439
-    },
-    {
-      "epoch": 5.966101694915254,
-      "grad_norm": 1.393019437789917,
-      "learning_rate": 8.900804020100503e-05,
-      "loss": 5.7418,
-      "step": 11440
-    },
-    {
-      "epoch": 5.966623207301174,
-      "grad_norm": 1.4571832418441772,
-      "learning_rate": 8.90070351758794e-05,
-      "loss": 5.6825,
-      "step": 11441
-    },
-    {
-      "epoch": 5.967144719687092,
-      "grad_norm": 1.360360860824585,
-      "learning_rate": 8.900603015075377e-05,
-      "loss": 6.1209,
-      "step": 11442
-    },
-    {
-      "epoch": 5.967666232073012,
-      "grad_norm": 1.3024564981460571,
-      "learning_rate": 8.900502512562815e-05,
-      "loss": 5.8015,
-      "step": 11443
-    },
-    {
-      "epoch": 5.96818774445893,
-      "grad_norm": 1.4167948961257935,
-      "learning_rate": 8.900402010050251e-05,
-      "loss": 5.7435,
-      "step": 11444
-    },
-    {
-      "epoch": 5.96870925684485,
-      "grad_norm": 1.4393635988235474,
-      "learning_rate": 8.900301507537689e-05,
-      "loss": 5.5192,
-      "step": 11445
-    },
-    {
-      "epoch": 5.969230769230769,
-      "grad_norm": 1.4602512121200562,
-      "learning_rate": 8.900201005025127e-05,
-      "loss": 5.7962,
-      "step": 11446
-    },
-    {
-      "epoch": 5.969752281616689,
-      "grad_norm": 1.5254912376403809,
-      "learning_rate": 8.900100502512563e-05,
-      "loss": 5.4532,
-      "step": 11447
-    },
-    {
-      "epoch": 5.970273794002607,
-      "grad_norm": 1.371071696281433,
-      "learning_rate": 8.900000000000001e-05,
-      "loss": 5.9296,
-      "step": 11448
-    },
-    {
-      "epoch": 5.970795306388527,
-      "grad_norm": 1.363547444343567,
-      "learning_rate": 8.899899497487437e-05,
-      "loss": 5.6911,
-      "step": 11449
-    },
-    {
-      "epoch": 5.971316818774445,
-      "grad_norm": 1.5369521379470825,
-      "learning_rate": 8.899798994974875e-05,
-      "loss": 5.6967,
-      "step": 11450
-    },
-    {
-      "epoch": 5.971838331160365,
-      "grad_norm": 1.3703279495239258,
-      "learning_rate": 8.899698492462311e-05,
-      "loss": 5.8964,
-      "step": 11451
-    },
-    {
-      "epoch": 5.972359843546284,
-      "grad_norm": 1.4732915163040161,
-      "learning_rate": 8.899597989949749e-05,
-      "loss": 5.3285,
-      "step": 11452
-    },
-    {
-      "epoch": 5.972881355932204,
-      "grad_norm": 1.410583734512329,
-      "learning_rate": 8.899497487437186e-05,
-      "loss": 5.8994,
-      "step": 11453
-    },
-    {
-      "epoch": 5.973402868318122,
-      "grad_norm": 1.433791995048523,
-      "learning_rate": 8.899396984924623e-05,
-      "loss": 5.7509,
-      "step": 11454
-    },
-    {
-      "epoch": 5.973924380704042,
-      "grad_norm": 1.524071216583252,
-      "learning_rate": 8.899296482412061e-05,
-      "loss": 5.1739,
-      "step": 11455
-    },
-    {
-      "epoch": 5.97444589308996,
-      "grad_norm": 1.5249229669570923,
-      "learning_rate": 8.899195979899499e-05,
-      "loss": 5.2025,
-      "step": 11456
-    },
-    {
-      "epoch": 5.97496740547588,
-      "grad_norm": 1.3883343935012817,
-      "learning_rate": 8.899095477386935e-05,
-      "loss": 5.7042,
-      "step": 11457
-    },
-    {
-      "epoch": 5.975488917861799,
-      "grad_norm": 1.433491587638855,
-      "learning_rate": 8.898994974874373e-05,
-      "loss": 5.7486,
-      "step": 11458
-    },
-    {
-      "epoch": 5.976010430247719,
-      "grad_norm": 1.4000357389450073,
-      "learning_rate": 8.89889447236181e-05,
-      "loss": 6.129,
-      "step": 11459
-    },
-    {
-      "epoch": 5.976531942633637,
-      "grad_norm": 1.4899263381958008,
-      "learning_rate": 8.898793969849246e-05,
-      "loss": 5.64,
-      "step": 11460
-    },
-    {
-      "epoch": 5.977053455019557,
-      "grad_norm": 1.5438705682754517,
-      "learning_rate": 8.898693467336684e-05,
-      "loss": 5.4661,
-      "step": 11461
-    },
-    {
-      "epoch": 5.9775749674054754,
-      "grad_norm": 1.496922492980957,
-      "learning_rate": 8.89859296482412e-05,
-      "loss": 5.4025,
-      "step": 11462
-    },
-    {
-      "epoch": 5.978096479791395,
-      "grad_norm": 1.4808335304260254,
-      "learning_rate": 8.898492462311558e-05,
-      "loss": 5.5306,
-      "step": 11463
-    },
-    {
-      "epoch": 5.978617992177314,
-      "grad_norm": 1.411497950553894,
-      "learning_rate": 8.898391959798995e-05,
-      "loss": 5.7863,
-      "step": 11464
-    },
-    {
-      "epoch": 5.979139504563234,
-      "grad_norm": 1.6915209293365479,
-      "learning_rate": 8.898291457286432e-05,
-      "loss": 5.0461,
-      "step": 11465
-    },
-    {
-      "epoch": 5.979661016949152,
-      "grad_norm": 1.4937763214111328,
-      "learning_rate": 8.89819095477387e-05,
-      "loss": 5.9495,
-      "step": 11466
-    },
-    {
-      "epoch": 5.980182529335072,
-      "grad_norm": 1.571192741394043,
-      "learning_rate": 8.898090452261308e-05,
-      "loss": 5.2026,
-      "step": 11467
-    },
-    {
-      "epoch": 5.9807040417209905,
-      "grad_norm": 2.140212297439575,
-      "learning_rate": 8.897989949748744e-05,
-      "loss": 5.0994,
-      "step": 11468
-    },
-    {
-      "epoch": 5.98122555410691,
-      "grad_norm": 1.9276847839355469,
-      "learning_rate": 8.897889447236182e-05,
-      "loss": 5.8296,
-      "step": 11469
-    },
-    {
-      "epoch": 5.981747066492829,
-      "grad_norm": 1.4022644758224487,
-      "learning_rate": 8.897788944723619e-05,
-      "loss": 5.9593,
-      "step": 11470
-    },
-    {
-      "epoch": 5.982268578878749,
-      "grad_norm": 1.5062175989151,
-      "learning_rate": 8.897688442211056e-05,
-      "loss": 6.0316,
-      "step": 11471
-    },
-    {
-      "epoch": 5.982790091264667,
-      "grad_norm": 1.536367654800415,
-      "learning_rate": 8.897587939698493e-05,
-      "loss": 5.8859,
-      "step": 11472
-    },
-    {
-      "epoch": 5.983311603650587,
-      "grad_norm": 1.39445161819458,
-      "learning_rate": 8.897487437185929e-05,
-      "loss": 5.7954,
-      "step": 11473
-    },
-    {
-      "epoch": 5.9838331160365055,
-      "grad_norm": 1.6776553392410278,
-      "learning_rate": 8.897386934673367e-05,
-      "loss": 5.3543,
-      "step": 11474
-    },
-    {
-      "epoch": 5.984354628422425,
-      "grad_norm": 1.392533540725708,
-      "learning_rate": 8.897286432160805e-05,
-      "loss": 5.9663,
-      "step": 11475
-    },
-    {
-      "epoch": 5.984876140808344,
-      "grad_norm": 1.7007397413253784,
-      "learning_rate": 8.897185929648243e-05,
-      "loss": 5.1544,
-      "step": 11476
-    },
-    {
-      "epoch": 5.985397653194263,
-      "grad_norm": 1.4021518230438232,
-      "learning_rate": 8.897085427135679e-05,
-      "loss": 6.0448,
-      "step": 11477
-    },
-    {
-      "epoch": 5.985919165580182,
-      "grad_norm": 1.3803470134735107,
-      "learning_rate": 8.896984924623117e-05,
-      "loss": 5.9063,
-      "step": 11478
-    },
-    {
-      "epoch": 5.986440677966102,
-      "grad_norm": 1.388715147972107,
-      "learning_rate": 8.896884422110553e-05,
-      "loss": 5.6731,
-      "step": 11479
-    },
-    {
-      "epoch": 5.9869621903520205,
-      "grad_norm": 1.28661048412323,
-      "learning_rate": 8.896783919597991e-05,
-      "loss": 6.0246,
-      "step": 11480
-    },
-    {
-      "epoch": 5.98748370273794,
-      "grad_norm": 1.5021650791168213,
-      "learning_rate": 8.896683417085427e-05,
-      "loss": 5.6801,
-      "step": 11481
-    },
-    {
-      "epoch": 5.988005215123859,
-      "grad_norm": 1.6011439561843872,
-      "learning_rate": 8.896582914572865e-05,
-      "loss": 5.4163,
-      "step": 11482
-    },
-    {
-      "epoch": 5.988526727509778,
-      "grad_norm": 1.4522206783294678,
-      "learning_rate": 8.896482412060302e-05,
-      "loss": 6.0047,
-      "step": 11483
-    },
-    {
-      "epoch": 5.9890482398956975,
-      "grad_norm": 1.3615741729736328,
-      "learning_rate": 8.89638190954774e-05,
-      "loss": 5.4999,
-      "step": 11484
-    },
-    {
-      "epoch": 5.989569752281617,
-      "grad_norm": 1.5473908185958862,
-      "learning_rate": 8.896281407035176e-05,
-      "loss": 5.5526,
-      "step": 11485
-    },
-    {
-      "epoch": 5.9900912646675355,
-      "grad_norm": 1.435376763343811,
-      "learning_rate": 8.896180904522614e-05,
-      "loss": 5.7747,
-      "step": 11486
-    },
-    {
-      "epoch": 5.990612777053455,
-      "grad_norm": 1.4560998678207397,
-      "learning_rate": 8.896080402010051e-05,
-      "loss": 6.1515,
-      "step": 11487
-    },
-    {
-      "epoch": 5.991134289439374,
-      "grad_norm": 1.4988056421279907,
-      "learning_rate": 8.895979899497488e-05,
-      "loss": 5.1282,
-      "step": 11488
-    },
-    {
-      "epoch": 5.991655801825293,
-      "grad_norm": 1.594580054283142,
-      "learning_rate": 8.895879396984926e-05,
-      "loss": 5.9263,
-      "step": 11489
-    },
-    {
-      "epoch": 5.9921773142112125,
-      "grad_norm": 1.5053037405014038,
-      "learning_rate": 8.895778894472362e-05,
-      "loss": 5.8692,
-      "step": 11490
-    },
-    {
-      "epoch": 5.992698826597132,
-      "grad_norm": 1.5817185640335083,
-      "learning_rate": 8.8956783919598e-05,
-      "loss": 5.4713,
-      "step": 11491
-    },
-    {
-      "epoch": 5.9932203389830505,
-      "grad_norm": 1.4370204210281372,
-      "learning_rate": 8.895577889447236e-05,
-      "loss": 5.6285,
-      "step": 11492
-    },
-    {
-      "epoch": 5.99374185136897,
-      "grad_norm": 1.382635474205017,
-      "learning_rate": 8.895477386934674e-05,
-      "loss": 4.9485,
-      "step": 11493
-    },
-    {
-      "epoch": 5.994263363754889,
-      "grad_norm": 1.752809762954712,
-      "learning_rate": 8.89537688442211e-05,
-      "loss": 4.9758,
-      "step": 11494
-    },
-    {
-      "epoch": 5.994784876140808,
-      "grad_norm": 1.543208360671997,
-      "learning_rate": 8.895276381909548e-05,
-      "loss": 5.3748,
-      "step": 11495
-    },
-    {
-      "epoch": 5.9953063885267275,
-      "grad_norm": 1.6799103021621704,
-      "learning_rate": 8.895175879396985e-05,
-      "loss": 5.4698,
-      "step": 11496
-    },
-    {
-      "epoch": 5.995827900912647,
-      "grad_norm": 1.495679497718811,
-      "learning_rate": 8.895075376884422e-05,
-      "loss": 5.6099,
-      "step": 11497
-    },
-    {
-      "epoch": 5.9963494132985655,
-      "grad_norm": 1.4469408988952637,
-      "learning_rate": 8.89497487437186e-05,
-      "loss": 5.7036,
-      "step": 11498
-    },
-    {
-      "epoch": 5.996870925684485,
-      "grad_norm": 1.3990968465805054,
-      "learning_rate": 8.894874371859297e-05,
-      "loss": 5.4651,
-      "step": 11499
-    },
-    {
-      "epoch": 5.9973924380704045,
-      "grad_norm": 1.550546646118164,
-      "learning_rate": 8.894773869346734e-05,
-      "loss": 5.9033,
-      "step": 11500
-    },
-    {
-      "epoch": 5.997913950456323,
-      "grad_norm": 1.5084245204925537,
-      "learning_rate": 8.894673366834171e-05,
-      "loss": 5.7971,
-      "step": 11501
-    },
-    {
-      "epoch": 5.9984354628422425,
-      "grad_norm": 1.3780282735824585,
-      "learning_rate": 8.894572864321609e-05,
-      "loss": 5.4991,
-      "step": 11502
-    },
-    {
-      "epoch": 5.998956975228162,
-      "grad_norm": 1.5459003448486328,
-      "learning_rate": 8.894472361809045e-05,
-      "loss": 5.4689,
-      "step": 11503
-    },
-    {
-      "epoch": 5.9994784876140805,
-      "grad_norm": 1.7015087604522705,
-      "learning_rate": 8.894371859296483e-05,
-      "loss": 4.799,
-      "step": 11504
-    },
-    {
-      "epoch": 6.0,
-      "grad_norm": 1.493294358253479,
-      "learning_rate": 8.894271356783919e-05,
-      "loss": 6.2573,
-      "step": 11505
-    },
-    {
-      "epoch": 6.0005215123859195,
-      "grad_norm": 1.506612777709961,
-      "learning_rate": 8.894170854271357e-05,
-      "loss": 5.8965,
-      "step": 11506
-    },
-    {
-      "epoch": 6.001043024771838,
-      "grad_norm": 1.6368435621261597,
-      "learning_rate": 8.894070351758795e-05,
-      "loss": 5.431,
-      "step": 11507
-    },
-    {
-      "epoch": 6.0015645371577575,
-      "grad_norm": 1.4499064683914185,
-      "learning_rate": 8.893969849246233e-05,
-      "loss": 6.0299,
-      "step": 11508
-    },
-    {
-      "epoch": 6.002086049543677,
-      "grad_norm": 1.5409725904464722,
-      "learning_rate": 8.893869346733669e-05,
-      "loss": 5.775,
-      "step": 11509
-    },
-    {
-      "epoch": 6.0026075619295955,
-      "grad_norm": 1.525002360343933,
-      "learning_rate": 8.893768844221107e-05,
-      "loss": 5.3477,
-      "step": 11510
-    },
-    {
-      "epoch": 6.003129074315515,
-      "grad_norm": 1.3902626037597656,
-      "learning_rate": 8.893668341708543e-05,
-      "loss": 5.3153,
-      "step": 11511
-    },
-    {
-      "epoch": 6.0036505867014345,
-      "grad_norm": 1.407558798789978,
-      "learning_rate": 8.89356783919598e-05,
-      "loss": 5.848,
-      "step": 11512
-    },
-    {
-      "epoch": 6.004172099087353,
-      "grad_norm": 1.495614767074585,
-      "learning_rate": 8.893467336683417e-05,
-      "loss": 5.1514,
-      "step": 11513
-    },
-    {
-      "epoch": 6.0046936114732725,
-      "grad_norm": 1.3853685855865479,
-      "learning_rate": 8.893366834170854e-05,
-      "loss": 5.4911,
-      "step": 11514
-    },
-    {
-      "epoch": 6.005215123859192,
-      "grad_norm": 1.4366756677627563,
-      "learning_rate": 8.893266331658292e-05,
-      "loss": 5.7491,
-      "step": 11515
-    },
-    {
-      "epoch": 6.005736636245111,
-      "grad_norm": 1.4816884994506836,
-      "learning_rate": 8.893165829145728e-05,
-      "loss": 5.359,
-      "step": 11516
-    },
-    {
-      "epoch": 6.00625814863103,
-      "grad_norm": 1.650976300239563,
-      "learning_rate": 8.893065326633166e-05,
-      "loss": 5.2963,
-      "step": 11517
-    },
-    {
-      "epoch": 6.0067796610169495,
-      "grad_norm": 1.4794443845748901,
-      "learning_rate": 8.892964824120604e-05,
-      "loss": 5.352,
-      "step": 11518
-    },
-    {
-      "epoch": 6.007301173402868,
-      "grad_norm": 1.4016081094741821,
-      "learning_rate": 8.892864321608041e-05,
-      "loss": 5.5628,
-      "step": 11519
-    },
-    {
-      "epoch": 6.0078226857887875,
-      "grad_norm": 1.489331603050232,
-      "learning_rate": 8.892763819095478e-05,
-      "loss": 5.6822,
-      "step": 11520
-    },
-    {
-      "epoch": 6.008344198174707,
-      "grad_norm": 1.5036084651947021,
-      "learning_rate": 8.892663316582916e-05,
-      "loss": 5.6775,
-      "step": 11521
-    },
-    {
-      "epoch": 6.008865710560626,
-      "grad_norm": 1.5228267908096313,
-      "learning_rate": 8.892562814070352e-05,
-      "loss": 5.0648,
-      "step": 11522
-    },
-    {
-      "epoch": 6.009387222946545,
-      "grad_norm": 1.438232183456421,
-      "learning_rate": 8.89246231155779e-05,
-      "loss": 5.7455,
-      "step": 11523
-    },
-    {
-      "epoch": 6.0099087353324645,
-      "grad_norm": 1.4587231874465942,
-      "learning_rate": 8.892361809045226e-05,
-      "loss": 5.3596,
-      "step": 11524
-    },
-    {
-      "epoch": 6.010430247718383,
-      "grad_norm": 1.5024542808532715,
-      "learning_rate": 8.892261306532663e-05,
-      "loss": 5.8227,
-      "step": 11525
-    },
-    {
-      "epoch": 6.0109517601043025,
-      "grad_norm": 1.577108383178711,
-      "learning_rate": 8.8921608040201e-05,
-      "loss": 5.2668,
-      "step": 11526
-    },
-    {
-      "epoch": 6.011473272490222,
-      "grad_norm": 1.4503118991851807,
-      "learning_rate": 8.892060301507538e-05,
-      "loss": 5.5014,
-      "step": 11527
-    },
-    {
-      "epoch": 6.011994784876141,
-      "grad_norm": 1.5820109844207764,
-      "learning_rate": 8.891959798994976e-05,
-      "loss": 5.3072,
-      "step": 11528
-    },
-    {
-      "epoch": 6.01251629726206,
-      "grad_norm": 1.4806160926818848,
-      "learning_rate": 8.891859296482412e-05,
-      "loss": 5.783,
-      "step": 11529
-    },
-    {
-      "epoch": 6.0130378096479795,
-      "grad_norm": 1.5678306818008423,
-      "learning_rate": 8.89175879396985e-05,
-      "loss": 5.4599,
-      "step": 11530
-    },
-    {
-      "epoch": 6.013559322033898,
-      "grad_norm": 1.46477210521698,
-      "learning_rate": 8.891658291457287e-05,
-      "loss": 5.6101,
-      "step": 11531
-    },
-    {
-      "epoch": 6.014080834419818,
-      "grad_norm": 1.3827650547027588,
-      "learning_rate": 8.891557788944724e-05,
-      "loss": 5.6189,
-      "step": 11532
-    },
-    {
-      "epoch": 6.014602346805737,
-      "grad_norm": 1.3744256496429443,
-      "learning_rate": 8.891457286432161e-05,
-      "loss": 5.6499,
-      "step": 11533
-    },
-    {
-      "epoch": 6.015123859191656,
-      "grad_norm": 1.4532227516174316,
-      "learning_rate": 8.891356783919599e-05,
-      "loss": 5.3989,
-      "step": 11534
-    },
-    {
-      "epoch": 6.015645371577575,
-      "grad_norm": 1.350378394126892,
-      "learning_rate": 8.891256281407035e-05,
-      "loss": 5.5783,
-      "step": 11535
-    },
-    {
-      "epoch": 6.0161668839634945,
-      "grad_norm": 1.338955044746399,
-      "learning_rate": 8.891155778894473e-05,
-      "loss": 6.0902,
-      "step": 11536
-    },
-    {
-      "epoch": 6.016688396349413,
-      "grad_norm": 1.33506178855896,
-      "learning_rate": 8.891055276381909e-05,
-      "loss": 6.1081,
-      "step": 11537
-    },
-    {
-      "epoch": 6.017209908735333,
-      "grad_norm": 1.4431123733520508,
-      "learning_rate": 8.890954773869347e-05,
-      "loss": 5.3238,
-      "step": 11538
-    },
-    {
-      "epoch": 6.017731421121252,
-      "grad_norm": 1.391509771347046,
-      "learning_rate": 8.890854271356785e-05,
-      "loss": 5.4759,
-      "step": 11539
-    },
-    {
-      "epoch": 6.018252933507171,
-      "grad_norm": 1.3812015056610107,
-      "learning_rate": 8.890753768844221e-05,
-      "loss": 5.3988,
-      "step": 11540
-    },
-    {
-      "epoch": 6.01877444589309,
-      "grad_norm": 1.4787200689315796,
-      "learning_rate": 8.890653266331659e-05,
-      "loss": 5.7927,
-      "step": 11541
-    },
-    {
-      "epoch": 6.0192959582790095,
-      "grad_norm": 1.3573094606399536,
-      "learning_rate": 8.890552763819096e-05,
-      "loss": 5.8857,
-      "step": 11542
-    },
-    {
-      "epoch": 6.019817470664928,
-      "grad_norm": 1.5243029594421387,
-      "learning_rate": 8.890452261306533e-05,
-      "loss": 5.6116,
-      "step": 11543
-    },
-    {
-      "epoch": 6.020338983050848,
-      "grad_norm": 1.4293748140335083,
-      "learning_rate": 8.89035175879397e-05,
-      "loss": 5.7861,
-      "step": 11544
-    },
-    {
-      "epoch": 6.020860495436767,
-      "grad_norm": 1.40329110622406,
-      "learning_rate": 8.890251256281408e-05,
-      "loss": 5.5211,
-      "step": 11545
-    },
-    {
-      "epoch": 6.021382007822686,
-      "grad_norm": 1.320410132408142,
-      "learning_rate": 8.890150753768844e-05,
-      "loss": 5.8615,
-      "step": 11546
-    },
-    {
-      "epoch": 6.021903520208605,
-      "grad_norm": 1.4855666160583496,
-      "learning_rate": 8.890050251256282e-05,
-      "loss": 5.6527,
-      "step": 11547
-    },
-    {
-      "epoch": 6.0224250325945246,
-      "grad_norm": 1.480035424232483,
-      "learning_rate": 8.88994974874372e-05,
-      "loss": 5.154,
-      "step": 11548
-    },
-    {
-      "epoch": 6.022946544980443,
-      "grad_norm": 1.5009076595306396,
-      "learning_rate": 8.889849246231157e-05,
-      "loss": 5.6175,
-      "step": 11549
-    },
-    {
-      "epoch": 6.023468057366363,
-      "grad_norm": 1.49691903591156,
-      "learning_rate": 8.889748743718594e-05,
-      "loss": 5.1467,
-      "step": 11550
-    },
-    {
-      "epoch": 6.023989569752282,
-      "grad_norm": 1.355566143989563,
-      "learning_rate": 8.889648241206031e-05,
-      "loss": 5.5139,
-      "step": 11551
-    },
-    {
-      "epoch": 6.024511082138201,
-      "grad_norm": 1.4316003322601318,
-      "learning_rate": 8.889547738693468e-05,
-      "loss": 5.8601,
-      "step": 11552
-    },
-    {
-      "epoch": 6.02503259452412,
-      "grad_norm": 1.645242691040039,
-      "learning_rate": 8.889447236180904e-05,
-      "loss": 5.0952,
-      "step": 11553
-    },
-    {
-      "epoch": 6.025554106910039,
-      "grad_norm": 1.5290937423706055,
-      "learning_rate": 8.889346733668342e-05,
-      "loss": 5.2091,
-      "step": 11554
-    },
-    {
-      "epoch": 6.026075619295958,
-      "grad_norm": 1.5301264524459839,
-      "learning_rate": 8.889246231155779e-05,
-      "loss": 5.3883,
-      "step": 11555
-    },
-    {
-      "epoch": 6.026597131681878,
-      "grad_norm": 1.3776392936706543,
-      "learning_rate": 8.889145728643216e-05,
-      "loss": 5.8995,
-      "step": 11556
-    },
-    {
-      "epoch": 6.027118644067796,
-      "grad_norm": 1.407408356666565,
-      "learning_rate": 8.889045226130653e-05,
-      "loss": 5.5612,
-      "step": 11557
-    },
-    {
-      "epoch": 6.027640156453716,
-      "grad_norm": 1.337822675704956,
-      "learning_rate": 8.88894472361809e-05,
-      "loss": 5.5711,
-      "step": 11558
-    },
-    {
-      "epoch": 6.028161668839635,
-      "grad_norm": 1.384225606918335,
-      "learning_rate": 8.888844221105528e-05,
-      "loss": 5.6158,
-      "step": 11559
-    },
-    {
-      "epoch": 6.028683181225554,
-      "grad_norm": 1.5943576097488403,
-      "learning_rate": 8.888743718592966e-05,
-      "loss": 5.0925,
-      "step": 11560
-    },
-    {
-      "epoch": 6.029204693611473,
-      "grad_norm": 1.378939151763916,
-      "learning_rate": 8.888643216080403e-05,
-      "loss": 5.4412,
-      "step": 11561
-    },
-    {
-      "epoch": 6.029726205997393,
-      "grad_norm": 1.3964170217514038,
-      "learning_rate": 8.88854271356784e-05,
-      "loss": 5.84,
-      "step": 11562
-    },
-    {
-      "epoch": 6.030247718383311,
-      "grad_norm": 1.4324042797088623,
-      "learning_rate": 8.888442211055277e-05,
-      "loss": 5.5921,
-      "step": 11563
-    },
-    {
-      "epoch": 6.030769230769231,
-      "grad_norm": 1.347200870513916,
-      "learning_rate": 8.888341708542715e-05,
-      "loss": 5.8994,
-      "step": 11564
-    },
-    {
-      "epoch": 6.03129074315515,
-      "grad_norm": 1.4284348487854004,
-      "learning_rate": 8.888241206030151e-05,
-      "loss": 5.663,
-      "step": 11565
-    },
-    {
-      "epoch": 6.031812255541069,
-      "grad_norm": 1.361517071723938,
-      "learning_rate": 8.888140703517587e-05,
-      "loss": 5.8241,
-      "step": 11566
-    },
-    {
-      "epoch": 6.032333767926988,
-      "grad_norm": 1.410835862159729,
-      "learning_rate": 8.888040201005025e-05,
-      "loss": 5.5417,
-      "step": 11567
-    },
-    {
-      "epoch": 6.032855280312908,
-      "grad_norm": 1.5175306797027588,
-      "learning_rate": 8.887939698492463e-05,
-      "loss": 5.6678,
-      "step": 11568
-    },
-    {
-      "epoch": 6.033376792698826,
-      "grad_norm": 1.3770477771759033,
-      "learning_rate": 8.887839195979901e-05,
-      "loss": 5.9723,
-      "step": 11569
-    },
-    {
-      "epoch": 6.033898305084746,
-      "grad_norm": 1.539601445198059,
-      "learning_rate": 8.887738693467337e-05,
-      "loss": 5.4048,
-      "step": 11570
-    },
-    {
-      "epoch": 6.034419817470665,
-      "grad_norm": 1.4945060014724731,
-      "learning_rate": 8.887638190954775e-05,
-      "loss": 5.6874,
-      "step": 11571
-    },
-    {
-      "epoch": 6.034941329856584,
-      "grad_norm": 1.3982456922531128,
-      "learning_rate": 8.887537688442211e-05,
-      "loss": 5.5145,
-      "step": 11572
-    },
-    {
-      "epoch": 6.035462842242503,
-      "grad_norm": 1.5270003080368042,
-      "learning_rate": 8.887437185929649e-05,
-      "loss": 5.542,
-      "step": 11573
-    },
-    {
-      "epoch": 6.035984354628423,
-      "grad_norm": 1.3933851718902588,
-      "learning_rate": 8.887336683417086e-05,
-      "loss": 5.5848,
-      "step": 11574
-    },
-    {
-      "epoch": 6.036505867014341,
-      "grad_norm": 1.4296610355377197,
-      "learning_rate": 8.887236180904523e-05,
-      "loss": 5.7676,
-      "step": 11575
-    },
-    {
-      "epoch": 6.037027379400261,
-      "grad_norm": 1.3824865818023682,
-      "learning_rate": 8.88713567839196e-05,
-      "loss": 5.5698,
-      "step": 11576
-    },
-    {
-      "epoch": 6.03754889178618,
-      "grad_norm": 1.4425302743911743,
-      "learning_rate": 8.887035175879398e-05,
-      "loss": 5.4209,
-      "step": 11577
-    },
-    {
-      "epoch": 6.038070404172099,
-      "grad_norm": 1.3651498556137085,
-      "learning_rate": 8.886934673366834e-05,
-      "loss": 5.6307,
-      "step": 11578
-    },
-    {
-      "epoch": 6.038591916558018,
-      "grad_norm": 1.4273524284362793,
-      "learning_rate": 8.886834170854272e-05,
-      "loss": 5.3766,
-      "step": 11579
-    },
-    {
-      "epoch": 6.039113428943938,
-      "grad_norm": 1.528870701789856,
-      "learning_rate": 8.88673366834171e-05,
-      "loss": 5.6867,
-      "step": 11580
-    },
-    {
-      "epoch": 6.039634941329856,
-      "grad_norm": 1.3965225219726562,
-      "learning_rate": 8.886633165829146e-05,
-      "loss": 6.0731,
-      "step": 11581
-    },
-    {
-      "epoch": 6.040156453715776,
-      "grad_norm": 1.47834050655365,
-      "learning_rate": 8.886532663316584e-05,
-      "loss": 5.2698,
-      "step": 11582
-    },
-    {
-      "epoch": 6.040677966101695,
-      "grad_norm": 1.5256093740463257,
-      "learning_rate": 8.88643216080402e-05,
-      "loss": 5.1856,
-      "step": 11583
-    },
-    {
-      "epoch": 6.041199478487614,
-      "grad_norm": 1.538853645324707,
-      "learning_rate": 8.886331658291458e-05,
-      "loss": 5.4396,
-      "step": 11584
-    },
-    {
-      "epoch": 6.041720990873533,
-      "grad_norm": 1.454728126525879,
-      "learning_rate": 8.886231155778894e-05,
-      "loss": 5.1644,
-      "step": 11585
-    },
-    {
-      "epoch": 6.042242503259453,
-      "grad_norm": 1.583857536315918,
-      "learning_rate": 8.886130653266332e-05,
-      "loss": 5.5083,
-      "step": 11586
-    },
-    {
-      "epoch": 6.042764015645371,
-      "grad_norm": 1.436181902885437,
-      "learning_rate": 8.886030150753769e-05,
-      "loss": 5.4298,
-      "step": 11587
-    },
-    {
-      "epoch": 6.043285528031291,
-      "grad_norm": 1.5486282110214233,
-      "learning_rate": 8.885929648241206e-05,
-      "loss": 5.5792,
-      "step": 11588
-    },
-    {
-      "epoch": 6.04380704041721,
-      "grad_norm": 1.3282279968261719,
-      "learning_rate": 8.885829145728644e-05,
-      "loss": 5.9995,
-      "step": 11589
-    },
-    {
-      "epoch": 6.044328552803129,
-      "grad_norm": 1.665515661239624,
-      "learning_rate": 8.885728643216082e-05,
-      "loss": 4.9865,
-      "step": 11590
-    },
-    {
-      "epoch": 6.044850065189048,
-      "grad_norm": 1.7850240468978882,
-      "learning_rate": 8.885628140703518e-05,
-      "loss": 4.6307,
-      "step": 11591
-    },
-    {
-      "epoch": 6.045371577574968,
-      "grad_norm": 1.5753324031829834,
-      "learning_rate": 8.885527638190955e-05,
-      "loss": 5.8452,
-      "step": 11592
-    },
-    {
-      "epoch": 6.045893089960886,
-      "grad_norm": 1.6387650966644287,
-      "learning_rate": 8.885427135678393e-05,
-      "loss": 5.4738,
-      "step": 11593
-    },
-    {
-      "epoch": 6.046414602346806,
-      "grad_norm": 1.410549521446228,
-      "learning_rate": 8.885326633165829e-05,
-      "loss": 5.871,
-      "step": 11594
-    },
-    {
-      "epoch": 6.046936114732725,
-      "grad_norm": 1.5383152961730957,
-      "learning_rate": 8.885226130653267e-05,
-      "loss": 5.5865,
-      "step": 11595
-    },
-    {
-      "epoch": 6.047457627118644,
-      "grad_norm": 1.536430835723877,
-      "learning_rate": 8.885125628140703e-05,
-      "loss": 5.5154,
-      "step": 11596
-    },
-    {
-      "epoch": 6.047979139504563,
-      "grad_norm": 1.5007210969924927,
-      "learning_rate": 8.885025125628141e-05,
-      "loss": 5.6281,
-      "step": 11597
-    },
-    {
-      "epoch": 6.048500651890483,
-      "grad_norm": 1.5569000244140625,
-      "learning_rate": 8.884924623115577e-05,
-      "loss": 5.6222,
-      "step": 11598
-    },
-    {
-      "epoch": 6.049022164276401,
-      "grad_norm": 1.3333216905593872,
-      "learning_rate": 8.884824120603015e-05,
-      "loss": 5.4981,
-      "step": 11599
-    },
-    {
-      "epoch": 6.049543676662321,
-      "grad_norm": 1.4326790571212769,
-      "learning_rate": 8.884723618090453e-05,
-      "loss": 5.7576,
-      "step": 11600
-    },
-    {
-      "epoch": 6.05006518904824,
-      "grad_norm": 1.5845881700515747,
-      "learning_rate": 8.884623115577891e-05,
-      "loss": 4.4274,
-      "step": 11601
-    },
-    {
-      "epoch": 6.050586701434159,
-      "grad_norm": 1.3588321208953857,
-      "learning_rate": 8.884522613065327e-05,
-      "loss": 5.8819,
-      "step": 11602
-    },
-    {
-      "epoch": 6.051108213820078,
-      "grad_norm": 1.3394415378570557,
-      "learning_rate": 8.884422110552765e-05,
-      "loss": 5.8239,
-      "step": 11603
-    },
-    {
-      "epoch": 6.051629726205998,
-      "grad_norm": 1.5041611194610596,
-      "learning_rate": 8.884321608040201e-05,
-      "loss": 5.5807,
-      "step": 11604
-    },
-    {
-      "epoch": 6.052151238591916,
-      "grad_norm": 1.4032191038131714,
-      "learning_rate": 8.884221105527638e-05,
-      "loss": 5.518,
-      "step": 11605
-    },
-    {
-      "epoch": 6.052672750977836,
-      "grad_norm": 1.507315754890442,
-      "learning_rate": 8.884120603015076e-05,
-      "loss": 5.6709,
-      "step": 11606
-    },
-    {
-      "epoch": 6.053194263363755,
-      "grad_norm": 1.6724522113800049,
-      "learning_rate": 8.884020100502512e-05,
-      "loss": 5.7313,
-      "step": 11607
-    },
-    {
-      "epoch": 6.053715775749674,
-      "grad_norm": 1.3834474086761475,
-      "learning_rate": 8.88391959798995e-05,
-      "loss": 5.7011,
-      "step": 11608
-    },
-    {
-      "epoch": 6.054237288135593,
-      "grad_norm": 1.3845744132995605,
-      "learning_rate": 8.883819095477388e-05,
-      "loss": 5.8044,
-      "step": 11609
-    },
-    {
-      "epoch": 6.054758800521513,
-      "grad_norm": 1.4126548767089844,
-      "learning_rate": 8.883718592964825e-05,
-      "loss": 5.5829,
-      "step": 11610
-    },
-    {
-      "epoch": 6.055280312907431,
-      "grad_norm": 1.4210928678512573,
-      "learning_rate": 8.883618090452262e-05,
-      "loss": 6.0685,
-      "step": 11611
-    },
-    {
-      "epoch": 6.055801825293351,
-      "grad_norm": 1.4478729963302612,
-      "learning_rate": 8.8835175879397e-05,
-      "loss": 5.4942,
-      "step": 11612
-    },
-    {
-      "epoch": 6.05632333767927,
-      "grad_norm": 1.4040195941925049,
-      "learning_rate": 8.883417085427136e-05,
-      "loss": 5.154,
-      "step": 11613
-    },
-    {
-      "epoch": 6.056844850065189,
-      "grad_norm": 1.458060622215271,
-      "learning_rate": 8.883316582914574e-05,
-      "loss": 5.5009,
-      "step": 11614
-    },
-    {
-      "epoch": 6.057366362451108,
-      "grad_norm": 1.3442856073379517,
-      "learning_rate": 8.88321608040201e-05,
-      "loss": 5.4299,
-      "step": 11615
-    },
-    {
-      "epoch": 6.057887874837028,
-      "grad_norm": 1.3361666202545166,
-      "learning_rate": 8.883115577889448e-05,
-      "loss": 5.4157,
-      "step": 11616
-    },
-    {
-      "epoch": 6.058409387222946,
-      "grad_norm": 1.5110055208206177,
-      "learning_rate": 8.883015075376884e-05,
-      "loss": 5.8257,
-      "step": 11617
-    },
-    {
-      "epoch": 6.058930899608866,
-      "grad_norm": 1.4149893522262573,
-      "learning_rate": 8.882914572864321e-05,
-      "loss": 5.4033,
-      "step": 11618
-    },
-    {
-      "epoch": 6.059452411994785,
-      "grad_norm": 1.8572697639465332,
-      "learning_rate": 8.882814070351759e-05,
-      "loss": 4.8073,
-      "step": 11619
-    },
-    {
-      "epoch": 6.059973924380704,
-      "grad_norm": 1.3739112615585327,
-      "learning_rate": 8.882713567839196e-05,
-      "loss": 5.7173,
-      "step": 11620
-    },
-    {
-      "epoch": 6.060495436766623,
-      "grad_norm": 1.4313302040100098,
-      "learning_rate": 8.882613065326634e-05,
-      "loss": 5.768,
-      "step": 11621
-    },
-    {
-      "epoch": 6.061016949152543,
-      "grad_norm": 1.4166511297225952,
-      "learning_rate": 8.882512562814071e-05,
-      "loss": 5.6159,
-      "step": 11622
-    },
-    {
-      "epoch": 6.061538461538461,
-      "grad_norm": 1.4441481828689575,
-      "learning_rate": 8.882412060301508e-05,
-      "loss": 5.5485,
-      "step": 11623
-    },
-    {
-      "epoch": 6.062059973924381,
-      "grad_norm": 1.3893630504608154,
-      "learning_rate": 8.882311557788945e-05,
-      "loss": 5.7115,
-      "step": 11624
-    },
-    {
-      "epoch": 6.0625814863103,
-      "grad_norm": 1.4057269096374512,
-      "learning_rate": 8.882211055276383e-05,
-      "loss": 5.7058,
-      "step": 11625
-    },
-    {
-      "epoch": 6.063102998696219,
-      "grad_norm": 1.5195238590240479,
-      "learning_rate": 8.882110552763819e-05,
-      "loss": 5.7299,
-      "step": 11626
-    },
-    {
-      "epoch": 6.063624511082138,
-      "grad_norm": 1.432850956916809,
-      "learning_rate": 8.882010050251257e-05,
-      "loss": 5.8528,
-      "step": 11627
-    },
-    {
-      "epoch": 6.064146023468058,
-      "grad_norm": 1.605488657951355,
-      "learning_rate": 8.881909547738693e-05,
-      "loss": 5.7096,
-      "step": 11628
-    },
-    {
-      "epoch": 6.064667535853976,
-      "grad_norm": 1.761354923248291,
-      "learning_rate": 8.881809045226131e-05,
-      "loss": 4.8571,
-      "step": 11629
-    },
-    {
-      "epoch": 6.065189048239896,
-      "grad_norm": 1.3983947038650513,
-      "learning_rate": 8.881708542713569e-05,
-      "loss": 6.0137,
-      "step": 11630
-    },
-    {
-      "epoch": 6.065710560625815,
-      "grad_norm": 1.4955530166625977,
-      "learning_rate": 8.881608040201005e-05,
-      "loss": 5.5121,
-      "step": 11631
-    },
-    {
-      "epoch": 6.066232073011734,
-      "grad_norm": 1.3789114952087402,
-      "learning_rate": 8.881507537688443e-05,
-      "loss": 5.6556,
-      "step": 11632
-    },
-    {
-      "epoch": 6.066753585397653,
-      "grad_norm": 1.4820038080215454,
-      "learning_rate": 8.88140703517588e-05,
-      "loss": 5.6404,
-      "step": 11633
-    },
-    {
-      "epoch": 6.067275097783573,
-      "grad_norm": 1.62134850025177,
-      "learning_rate": 8.881306532663317e-05,
-      "loss": 5.1017,
-      "step": 11634
-    },
-    {
-      "epoch": 6.067796610169491,
-      "grad_norm": 1.5597553253173828,
-      "learning_rate": 8.881206030150754e-05,
-      "loss": 5.4375,
-      "step": 11635
-    },
-    {
-      "epoch": 6.068318122555411,
-      "grad_norm": 1.4159907102584839,
-      "learning_rate": 8.881105527638192e-05,
-      "loss": 5.8136,
-      "step": 11636
-    },
-    {
-      "epoch": 6.06883963494133,
-      "grad_norm": 1.380881905555725,
-      "learning_rate": 8.881005025125628e-05,
-      "loss": 5.7713,
-      "step": 11637
-    },
-    {
-      "epoch": 6.069361147327249,
-      "grad_norm": 1.4326351881027222,
-      "learning_rate": 8.880904522613066e-05,
-      "loss": 5.2108,
-      "step": 11638
-    },
-    {
-      "epoch": 6.069882659713168,
-      "grad_norm": 1.4535613059997559,
-      "learning_rate": 8.880804020100502e-05,
-      "loss": 5.5082,
-      "step": 11639
-    },
-    {
-      "epoch": 6.070404172099088,
-      "grad_norm": 1.4238014221191406,
-      "learning_rate": 8.88070351758794e-05,
-      "loss": 5.6508,
-      "step": 11640
-    },
-    {
-      "epoch": 6.070925684485006,
-      "grad_norm": 1.4178804159164429,
-      "learning_rate": 8.880603015075378e-05,
-      "loss": 5.8139,
-      "step": 11641
-    },
-    {
-      "epoch": 6.071447196870926,
-      "grad_norm": 1.4395684003829956,
-      "learning_rate": 8.880502512562816e-05,
-      "loss": 5.7993,
-      "step": 11642
-    },
-    {
-      "epoch": 6.071968709256845,
-      "grad_norm": 1.4325587749481201,
-      "learning_rate": 8.880402010050252e-05,
-      "loss": 5.7201,
-      "step": 11643
-    },
-    {
-      "epoch": 6.072490221642764,
-      "grad_norm": 1.4966754913330078,
-      "learning_rate": 8.88030150753769e-05,
-      "loss": 5.7462,
-      "step": 11644
-    },
-    {
-      "epoch": 6.073011734028683,
-      "grad_norm": 1.3985131978988647,
-      "learning_rate": 8.880201005025126e-05,
-      "loss": 5.8019,
-      "step": 11645
-    },
-    {
-      "epoch": 6.073533246414602,
-      "grad_norm": 1.5723531246185303,
-      "learning_rate": 8.880100502512563e-05,
-      "loss": 5.7066,
-      "step": 11646
-    },
-    {
-      "epoch": 6.074054758800521,
-      "grad_norm": 1.439120888710022,
-      "learning_rate": 8.88e-05,
-      "loss": 5.4052,
-      "step": 11647
-    },
-    {
-      "epoch": 6.074576271186441,
-      "grad_norm": 1.5430370569229126,
-      "learning_rate": 8.879899497487437e-05,
-      "loss": 5.6761,
-      "step": 11648
-    },
-    {
-      "epoch": 6.075097783572359,
-      "grad_norm": 1.5189270973205566,
-      "learning_rate": 8.879798994974875e-05,
-      "loss": 5.4173,
-      "step": 11649
-    },
-    {
-      "epoch": 6.075619295958279,
-      "grad_norm": 1.4400956630706787,
-      "learning_rate": 8.879698492462311e-05,
-      "loss": 5.532,
-      "step": 11650
-    },
-    {
-      "epoch": 6.076140808344198,
-      "grad_norm": 1.4955275058746338,
-      "learning_rate": 8.879597989949749e-05,
-      "loss": 5.3771,
-      "step": 11651
-    },
-    {
-      "epoch": 6.076662320730117,
-      "grad_norm": 1.4612884521484375,
-      "learning_rate": 8.879497487437187e-05,
-      "loss": 5.4253,
-      "step": 11652
-    },
-    {
-      "epoch": 6.077183833116036,
-      "grad_norm": 1.4572975635528564,
-      "learning_rate": 8.879396984924624e-05,
-      "loss": 5.7556,
-      "step": 11653
-    },
-    {
-      "epoch": 6.077705345501956,
-      "grad_norm": 1.5068236589431763,
-      "learning_rate": 8.879296482412061e-05,
-      "loss": 5.5518,
-      "step": 11654
-    },
-    {
-      "epoch": 6.078226857887874,
-      "grad_norm": 1.5407376289367676,
-      "learning_rate": 8.879195979899499e-05,
-      "loss": 5.5085,
-      "step": 11655
-    },
-    {
-      "epoch": 6.078748370273794,
-      "grad_norm": 1.4388582706451416,
-      "learning_rate": 8.879095477386935e-05,
-      "loss": 5.378,
-      "step": 11656
-    },
-    {
-      "epoch": 6.079269882659713,
-      "grad_norm": 1.4095635414123535,
-      "learning_rate": 8.878994974874373e-05,
-      "loss": 5.8043,
-      "step": 11657
-    },
-    {
-      "epoch": 6.079791395045632,
-      "grad_norm": 1.4487318992614746,
-      "learning_rate": 8.878894472361809e-05,
-      "loss": 5.4226,
-      "step": 11658
-    },
-    {
-      "epoch": 6.080312907431551,
-      "grad_norm": 1.4810349941253662,
-      "learning_rate": 8.878793969849246e-05,
-      "loss": 5.7132,
-      "step": 11659
-    },
-    {
-      "epoch": 6.080834419817471,
-      "grad_norm": 1.5533474683761597,
-      "learning_rate": 8.878693467336683e-05,
-      "loss": 5.4808,
-      "step": 11660
-    },
-    {
-      "epoch": 6.081355932203389,
-      "grad_norm": 1.6641393899917603,
-      "learning_rate": 8.878592964824121e-05,
-      "loss": 5.4347,
-      "step": 11661
-    },
-    {
-      "epoch": 6.081877444589309,
-      "grad_norm": 1.3789678812026978,
-      "learning_rate": 8.878492462311559e-05,
-      "loss": 5.5371,
-      "step": 11662
-    },
-    {
-      "epoch": 6.082398956975228,
-      "grad_norm": 1.6892882585525513,
-      "learning_rate": 8.878391959798995e-05,
-      "loss": 5.4211,
-      "step": 11663
-    },
-    {
-      "epoch": 6.082920469361147,
-      "grad_norm": 1.6255137920379639,
-      "learning_rate": 8.878291457286433e-05,
-      "loss": 5.4877,
-      "step": 11664
-    },
-    {
-      "epoch": 6.083441981747066,
-      "grad_norm": 1.39809250831604,
-      "learning_rate": 8.87819095477387e-05,
-      "loss": 5.9126,
-      "step": 11665
-    },
-    {
-      "epoch": 6.083963494132986,
-      "grad_norm": 1.4208098649978638,
-      "learning_rate": 8.878090452261307e-05,
-      "loss": 5.9038,
-      "step": 11666
-    },
-    {
-      "epoch": 6.0844850065189044,
-      "grad_norm": 1.321784257888794,
-      "learning_rate": 8.877989949748744e-05,
-      "loss": 5.7531,
-      "step": 11667
-    },
-    {
-      "epoch": 6.085006518904824,
-      "grad_norm": 1.4681508541107178,
-      "learning_rate": 8.877889447236182e-05,
-      "loss": 5.7306,
-      "step": 11668
-    },
-    {
-      "epoch": 6.085528031290743,
-      "grad_norm": 1.4576343297958374,
-      "learning_rate": 8.877788944723618e-05,
-      "loss": 5.2853,
-      "step": 11669
-    },
-    {
-      "epoch": 6.086049543676662,
-      "grad_norm": 1.4131999015808105,
-      "learning_rate": 8.877688442211056e-05,
-      "loss": 5.7182,
-      "step": 11670
-    },
-    {
-      "epoch": 6.086571056062581,
-      "grad_norm": 1.393664836883545,
-      "learning_rate": 8.877587939698492e-05,
-      "loss": 5.7543,
-      "step": 11671
-    },
-    {
-      "epoch": 6.087092568448501,
-      "grad_norm": 1.4018250703811646,
-      "learning_rate": 8.87748743718593e-05,
-      "loss": 6.014,
-      "step": 11672
-    },
-    {
-      "epoch": 6.0876140808344195,
-      "grad_norm": 1.5123642683029175,
-      "learning_rate": 8.877386934673368e-05,
-      "loss": 5.2173,
-      "step": 11673
-    },
-    {
-      "epoch": 6.088135593220339,
-      "grad_norm": 1.3852519989013672,
-      "learning_rate": 8.877286432160804e-05,
-      "loss": 5.5522,
-      "step": 11674
-    },
-    {
-      "epoch": 6.088657105606258,
-      "grad_norm": 1.513590931892395,
-      "learning_rate": 8.877185929648242e-05,
-      "loss": 5.5405,
-      "step": 11675
-    },
-    {
-      "epoch": 6.089178617992177,
-      "grad_norm": 1.454188585281372,
-      "learning_rate": 8.877085427135678e-05,
-      "loss": 5.6613,
-      "step": 11676
-    },
-    {
-      "epoch": 6.089700130378096,
-      "grad_norm": 1.3352017402648926,
-      "learning_rate": 8.876984924623116e-05,
-      "loss": 5.6963,
-      "step": 11677
-    },
-    {
-      "epoch": 6.090221642764016,
-      "grad_norm": 1.4689871072769165,
-      "learning_rate": 8.876884422110553e-05,
-      "loss": 5.7368,
-      "step": 11678
-    },
-    {
-      "epoch": 6.0907431551499345,
-      "grad_norm": 1.439414143562317,
-      "learning_rate": 8.87678391959799e-05,
-      "loss": 5.7839,
-      "step": 11679
-    },
-    {
-      "epoch": 6.091264667535854,
-      "grad_norm": 1.6185612678527832,
-      "learning_rate": 8.876683417085427e-05,
-      "loss": 4.8691,
-      "step": 11680
-    },
-    {
-      "epoch": 6.091786179921773,
-      "grad_norm": 1.4994142055511475,
-      "learning_rate": 8.876582914572865e-05,
-      "loss": 5.6558,
-      "step": 11681
-    },
-    {
-      "epoch": 6.092307692307692,
-      "grad_norm": 1.5859827995300293,
-      "learning_rate": 8.876482412060302e-05,
-      "loss": 5.3107,
-      "step": 11682
-    },
-    {
-      "epoch": 6.0928292046936114,
-      "grad_norm": 1.45134437084198,
-      "learning_rate": 8.87638190954774e-05,
-      "loss": 5.4972,
-      "step": 11683
-    },
-    {
-      "epoch": 6.093350717079531,
-      "grad_norm": 1.494283676147461,
-      "learning_rate": 8.876281407035177e-05,
-      "loss": 5.6476,
-      "step": 11684
-    },
-    {
-      "epoch": 6.0938722294654495,
-      "grad_norm": 1.471144676208496,
-      "learning_rate": 8.876180904522613e-05,
-      "loss": 5.474,
-      "step": 11685
-    },
-    {
-      "epoch": 6.094393741851369,
-      "grad_norm": 1.5183571577072144,
-      "learning_rate": 8.876080402010051e-05,
-      "loss": 5.7644,
-      "step": 11686
-    },
-    {
-      "epoch": 6.094915254237288,
-      "grad_norm": 1.5407963991165161,
-      "learning_rate": 8.875979899497487e-05,
-      "loss": 5.7056,
-      "step": 11687
-    },
-    {
-      "epoch": 6.095436766623207,
-      "grad_norm": 1.3944048881530762,
-      "learning_rate": 8.875879396984925e-05,
-      "loss": 5.7754,
-      "step": 11688
-    },
-    {
-      "epoch": 6.0959582790091265,
-      "grad_norm": 1.3787580728530884,
-      "learning_rate": 8.875778894472361e-05,
-      "loss": 5.5175,
-      "step": 11689
-    },
-    {
-      "epoch": 6.096479791395046,
-      "grad_norm": 1.3635141849517822,
-      "learning_rate": 8.875678391959799e-05,
-      "loss": 5.3298,
-      "step": 11690
-    },
-    {
-      "epoch": 6.0970013037809645,
-      "grad_norm": 1.5315465927124023,
-      "learning_rate": 8.875577889447236e-05,
-      "loss": 5.7511,
-      "step": 11691
-    },
-    {
-      "epoch": 6.097522816166884,
-      "grad_norm": 1.5471428632736206,
-      "learning_rate": 8.875477386934673e-05,
-      "loss": 5.6441,
-      "step": 11692
-    },
-    {
-      "epoch": 6.098044328552803,
-      "grad_norm": 1.3699969053268433,
-      "learning_rate": 8.875376884422111e-05,
-      "loss": 5.7577,
-      "step": 11693
-    },
-    {
-      "epoch": 6.098565840938722,
-      "grad_norm": 1.3851144313812256,
-      "learning_rate": 8.875276381909549e-05,
-      "loss": 5.9529,
-      "step": 11694
-    },
-    {
-      "epoch": 6.0990873533246415,
-      "grad_norm": 1.634214997291565,
-      "learning_rate": 8.875175879396985e-05,
-      "loss": 5.8345,
-      "step": 11695
-    },
-    {
-      "epoch": 6.099608865710561,
-      "grad_norm": 1.4744995832443237,
-      "learning_rate": 8.875075376884423e-05,
-      "loss": 5.5098,
-      "step": 11696
-    },
-    {
-      "epoch": 6.1001303780964795,
-      "grad_norm": 1.4078611135482788,
-      "learning_rate": 8.87497487437186e-05,
-      "loss": 5.6124,
-      "step": 11697
-    },
-    {
-      "epoch": 6.100651890482399,
-      "grad_norm": 1.3949488401412964,
-      "learning_rate": 8.874874371859296e-05,
-      "loss": 5.4422,
-      "step": 11698
-    },
-    {
-      "epoch": 6.101173402868318,
-      "grad_norm": 1.47588050365448,
-      "learning_rate": 8.874773869346734e-05,
-      "loss": 5.7768,
-      "step": 11699
-    },
-    {
-      "epoch": 6.101694915254237,
-      "grad_norm": 1.3734062910079956,
-      "learning_rate": 8.87467336683417e-05,
-      "loss": 5.3536,
-      "step": 11700
-    },
-    {
-      "epoch": 6.1022164276401565,
-      "grad_norm": 2.007230281829834,
-      "learning_rate": 8.874572864321608e-05,
-      "loss": 5.1544,
-      "step": 11701
-    },
-    {
-      "epoch": 6.102737940026076,
-      "grad_norm": 1.427955985069275,
-      "learning_rate": 8.874472361809046e-05,
-      "loss": 5.5794,
-      "step": 11702
-    },
-    {
-      "epoch": 6.1032594524119945,
-      "grad_norm": 1.4774253368377686,
-      "learning_rate": 8.874371859296484e-05,
-      "loss": 5.9474,
-      "step": 11703
-    },
-    {
-      "epoch": 6.103780964797914,
-      "grad_norm": 1.4604557752609253,
-      "learning_rate": 8.87427135678392e-05,
-      "loss": 5.4862,
-      "step": 11704
-    },
-    {
-      "epoch": 6.1043024771838335,
-      "grad_norm": 1.3595337867736816,
-      "learning_rate": 8.874170854271358e-05,
-      "loss": 5.8414,
-      "step": 11705
-    },
-    {
-      "epoch": 6.104823989569752,
-      "grad_norm": 1.4700387716293335,
-      "learning_rate": 8.874070351758794e-05,
-      "loss": 5.8489,
-      "step": 11706
-    },
-    {
-      "epoch": 6.1053455019556715,
-      "grad_norm": 1.3889039754867554,
-      "learning_rate": 8.873969849246232e-05,
-      "loss": 5.7875,
-      "step": 11707
-    },
-    {
-      "epoch": 6.105867014341591,
-      "grad_norm": 1.4499835968017578,
-      "learning_rate": 8.873869346733669e-05,
-      "loss": 4.8339,
-      "step": 11708
-    },
-    {
-      "epoch": 6.1063885267275095,
-      "grad_norm": 1.4780635833740234,
-      "learning_rate": 8.873768844221106e-05,
-      "loss": 5.6033,
-      "step": 11709
-    },
-    {
-      "epoch": 6.106910039113429,
-      "grad_norm": 1.5425723791122437,
-      "learning_rate": 8.873668341708543e-05,
-      "loss": 5.537,
-      "step": 11710
-    },
-    {
-      "epoch": 6.1074315514993485,
-      "grad_norm": 1.3761768341064453,
-      "learning_rate": 8.873567839195979e-05,
-      "loss": 5.8357,
-      "step": 11711
-    },
-    {
-      "epoch": 6.107953063885267,
-      "grad_norm": 1.3207741975784302,
-      "learning_rate": 8.873467336683417e-05,
-      "loss": 5.8168,
-      "step": 11712
-    },
-    {
-      "epoch": 6.1084745762711865,
-      "grad_norm": 1.4480620622634888,
-      "learning_rate": 8.873366834170855e-05,
-      "loss": 5.8813,
-      "step": 11713
-    },
-    {
-      "epoch": 6.108996088657106,
-      "grad_norm": 1.4407176971435547,
-      "learning_rate": 8.873266331658293e-05,
-      "loss": 4.728,
-      "step": 11714
-    },
-    {
-      "epoch": 6.1095176010430245,
-      "grad_norm": 1.5314844846725464,
-      "learning_rate": 8.873165829145729e-05,
-      "loss": 6.0628,
-      "step": 11715
-    },
-    {
-      "epoch": 6.110039113428944,
-      "grad_norm": 1.3178815841674805,
-      "learning_rate": 8.873065326633167e-05,
-      "loss": 6.0351,
-      "step": 11716
-    },
-    {
-      "epoch": 6.1105606258148635,
-      "grad_norm": 1.5768862962722778,
-      "learning_rate": 8.872964824120603e-05,
-      "loss": 5.7221,
-      "step": 11717
-    },
-    {
-      "epoch": 6.111082138200782,
-      "grad_norm": 1.5467371940612793,
-      "learning_rate": 8.872864321608041e-05,
-      "loss": 5.3905,
-      "step": 11718
-    },
-    {
-      "epoch": 6.1116036505867015,
-      "grad_norm": 1.3839572668075562,
-      "learning_rate": 8.872763819095477e-05,
-      "loss": 5.9048,
-      "step": 11719
-    },
-    {
-      "epoch": 6.112125162972621,
-      "grad_norm": 1.478502869606018,
-      "learning_rate": 8.872663316582915e-05,
-      "loss": 5.3227,
-      "step": 11720
-    },
-    {
-      "epoch": 6.11264667535854,
-      "grad_norm": 1.5848363637924194,
-      "learning_rate": 8.872562814070352e-05,
-      "loss": 5.2032,
-      "step": 11721
-    },
-    {
-      "epoch": 6.113168187744459,
-      "grad_norm": 1.4280774593353271,
-      "learning_rate": 8.87246231155779e-05,
-      "loss": 5.5671,
-      "step": 11722
-    },
-    {
-      "epoch": 6.1136897001303785,
-      "grad_norm": 1.6996437311172485,
-      "learning_rate": 8.872361809045227e-05,
-      "loss": 5.2829,
-      "step": 11723
-    },
-    {
-      "epoch": 6.114211212516297,
-      "grad_norm": 1.5391329526901245,
-      "learning_rate": 8.872261306532665e-05,
-      "loss": 5.6899,
-      "step": 11724
-    },
-    {
-      "epoch": 6.1147327249022165,
-      "grad_norm": 1.3922899961471558,
-      "learning_rate": 8.872160804020101e-05,
-      "loss": 5.7069,
-      "step": 11725
-    },
-    {
-      "epoch": 6.115254237288136,
-      "grad_norm": 1.5605159997940063,
-      "learning_rate": 8.872060301507538e-05,
-      "loss": 5.7086,
-      "step": 11726
-    },
-    {
-      "epoch": 6.115775749674055,
-      "grad_norm": 1.3838860988616943,
-      "learning_rate": 8.871959798994976e-05,
-      "loss": 5.4775,
-      "step": 11727
-    },
-    {
-      "epoch": 6.116297262059974,
-      "grad_norm": 1.8620003461837769,
-      "learning_rate": 8.871859296482412e-05,
-      "loss": 4.9248,
-      "step": 11728
-    },
-    {
-      "epoch": 6.1168187744458935,
-      "grad_norm": 1.4944069385528564,
-      "learning_rate": 8.87175879396985e-05,
-      "loss": 4.9139,
-      "step": 11729
-    },
-    {
-      "epoch": 6.117340286831812,
-      "grad_norm": 1.3432064056396484,
-      "learning_rate": 8.871658291457286e-05,
-      "loss": 5.7852,
-      "step": 11730
-    },
-    {
-      "epoch": 6.1178617992177315,
-      "grad_norm": 1.5580792427062988,
-      "learning_rate": 8.871557788944724e-05,
-      "loss": 5.5494,
-      "step": 11731
-    },
-    {
-      "epoch": 6.118383311603651,
-      "grad_norm": 1.5761548280715942,
-      "learning_rate": 8.87145728643216e-05,
-      "loss": 5.6777,
-      "step": 11732
-    },
-    {
-      "epoch": 6.11890482398957,
-      "grad_norm": 1.4622244834899902,
-      "learning_rate": 8.871356783919598e-05,
-      "loss": 5.6417,
-      "step": 11733
-    },
-    {
-      "epoch": 6.119426336375489,
-      "grad_norm": 1.561968445777893,
-      "learning_rate": 8.871256281407036e-05,
-      "loss": 5.5437,
-      "step": 11734
-    },
-    {
-      "epoch": 6.1199478487614085,
-      "grad_norm": 1.5147289037704468,
-      "learning_rate": 8.871155778894474e-05,
-      "loss": 5.5692,
-      "step": 11735
-    },
-    {
-      "epoch": 6.120469361147327,
-      "grad_norm": 1.3611663579940796,
-      "learning_rate": 8.87105527638191e-05,
-      "loss": 5.554,
-      "step": 11736
-    },
-    {
-      "epoch": 6.120990873533247,
-      "grad_norm": 1.492305874824524,
-      "learning_rate": 8.870954773869348e-05,
-      "loss": 5.5693,
-      "step": 11737
-    },
-    {
-      "epoch": 6.121512385919166,
-      "grad_norm": 1.4689035415649414,
-      "learning_rate": 8.870854271356784e-05,
-      "loss": 5.6215,
-      "step": 11738
-    },
-    {
-      "epoch": 6.122033898305085,
-      "grad_norm": 1.4971598386764526,
-      "learning_rate": 8.870753768844221e-05,
-      "loss": 5.7214,
-      "step": 11739
-    },
-    {
-      "epoch": 6.122555410691004,
-      "grad_norm": 1.374586582183838,
-      "learning_rate": 8.870653266331659e-05,
-      "loss": 5.991,
-      "step": 11740
-    },
-    {
-      "epoch": 6.123076923076923,
-      "grad_norm": 1.5213510990142822,
-      "learning_rate": 8.870552763819095e-05,
-      "loss": 5.5045,
-      "step": 11741
-    },
-    {
-      "epoch": 6.123598435462842,
-      "grad_norm": 1.480839729309082,
-      "learning_rate": 8.870452261306533e-05,
-      "loss": 5.6927,
-      "step": 11742
-    },
-    {
-      "epoch": 6.124119947848762,
-      "grad_norm": 1.4107108116149902,
-      "learning_rate": 8.87035175879397e-05,
-      "loss": 5.7727,
-      "step": 11743
-    },
-    {
-      "epoch": 6.12464146023468,
-      "grad_norm": 1.3872690200805664,
-      "learning_rate": 8.870251256281408e-05,
-      "loss": 5.9504,
-      "step": 11744
-    },
-    {
-      "epoch": 6.1251629726206,
-      "grad_norm": 1.3829232454299927,
-      "learning_rate": 8.870150753768845e-05,
-      "loss": 5.5219,
-      "step": 11745
-    },
-    {
-      "epoch": 6.125684485006519,
-      "grad_norm": 1.56228768825531,
-      "learning_rate": 8.870050251256283e-05,
-      "loss": 5.8073,
-      "step": 11746
-    },
-    {
-      "epoch": 6.126205997392438,
-      "grad_norm": 1.5938564538955688,
-      "learning_rate": 8.869949748743719e-05,
-      "loss": 5.4584,
-      "step": 11747
-    },
-    {
-      "epoch": 6.126727509778357,
-      "grad_norm": 1.4453153610229492,
-      "learning_rate": 8.869849246231157e-05,
-      "loss": 5.9468,
-      "step": 11748
-    },
-    {
-      "epoch": 6.127249022164277,
-      "grad_norm": 1.4430454969406128,
-      "learning_rate": 8.869748743718593e-05,
-      "loss": 5.6423,
-      "step": 11749
-    },
-    {
-      "epoch": 6.127770534550195,
-      "grad_norm": 1.6684865951538086,
-      "learning_rate": 8.869648241206031e-05,
-      "loss": 5.5481,
-      "step": 11750
-    },
-    {
-      "epoch": 6.128292046936115,
-      "grad_norm": 1.4122414588928223,
-      "learning_rate": 8.869547738693467e-05,
-      "loss": 5.6588,
-      "step": 11751
-    },
-    {
-      "epoch": 6.128813559322034,
-      "grad_norm": 1.363555669784546,
-      "learning_rate": 8.869447236180904e-05,
-      "loss": 6.0239,
-      "step": 11752
-    },
-    {
-      "epoch": 6.129335071707953,
-      "grad_norm": 1.432004690170288,
-      "learning_rate": 8.869346733668342e-05,
-      "loss": 5.7494,
-      "step": 11753
-    },
-    {
-      "epoch": 6.129856584093872,
-      "grad_norm": 1.430413007736206,
-      "learning_rate": 8.86924623115578e-05,
-      "loss": 5.6699,
-      "step": 11754
-    },
-    {
-      "epoch": 6.130378096479792,
-      "grad_norm": 1.4717974662780762,
-      "learning_rate": 8.869145728643217e-05,
-      "loss": 4.6029,
-      "step": 11755
-    },
-    {
-      "epoch": 6.13089960886571,
-      "grad_norm": 1.481825351715088,
-      "learning_rate": 8.869045226130654e-05,
-      "loss": 5.5618,
-      "step": 11756
-    },
-    {
-      "epoch": 6.13142112125163,
-      "grad_norm": 1.4975093603134155,
-      "learning_rate": 8.868944723618091e-05,
-      "loss": 5.3753,
-      "step": 11757
-    },
-    {
-      "epoch": 6.131942633637549,
-      "grad_norm": 1.5655969381332397,
-      "learning_rate": 8.868844221105528e-05,
-      "loss": 5.6315,
-      "step": 11758
-    },
-    {
-      "epoch": 6.132464146023468,
-      "grad_norm": 1.5698857307434082,
-      "learning_rate": 8.868743718592966e-05,
-      "loss": 5.2135,
-      "step": 11759
-    },
-    {
-      "epoch": 6.132985658409387,
-      "grad_norm": 1.5084257125854492,
-      "learning_rate": 8.868643216080402e-05,
-      "loss": 5.3691,
-      "step": 11760
-    },
-    {
-      "epoch": 6.133507170795307,
-      "grad_norm": 1.6932837963104248,
-      "learning_rate": 8.86854271356784e-05,
-      "loss": 5.7103,
-      "step": 11761
-    },
-    {
-      "epoch": 6.134028683181225,
-      "grad_norm": 1.4110647439956665,
-      "learning_rate": 8.868442211055276e-05,
-      "loss": 5.4994,
-      "step": 11762
-    },
-    {
-      "epoch": 6.134550195567145,
-      "grad_norm": 1.3253979682922363,
-      "learning_rate": 8.868341708542714e-05,
-      "loss": 5.7329,
-      "step": 11763
-    },
-    {
-      "epoch": 6.135071707953064,
-      "grad_norm": 1.3837213516235352,
-      "learning_rate": 8.868241206030152e-05,
-      "loss": 5.5544,
-      "step": 11764
-    },
-    {
-      "epoch": 6.135593220338983,
-      "grad_norm": 1.414543628692627,
-      "learning_rate": 8.868140703517588e-05,
-      "loss": 5.869,
-      "step": 11765
-    },
-    {
-      "epoch": 6.136114732724902,
-      "grad_norm": 1.4943103790283203,
-      "learning_rate": 8.868040201005026e-05,
-      "loss": 5.2877,
-      "step": 11766
-    },
-    {
-      "epoch": 6.136636245110822,
-      "grad_norm": 1.4421781301498413,
-      "learning_rate": 8.867939698492462e-05,
-      "loss": 5.5923,
-      "step": 11767
-    },
-    {
-      "epoch": 6.13715775749674,
-      "grad_norm": 1.3864333629608154,
-      "learning_rate": 8.8678391959799e-05,
-      "loss": 5.8791,
-      "step": 11768
-    },
-    {
-      "epoch": 6.13767926988266,
-      "grad_norm": 1.6303201913833618,
-      "learning_rate": 8.867738693467337e-05,
-      "loss": 5.3575,
-      "step": 11769
-    },
-    {
-      "epoch": 6.138200782268579,
-      "grad_norm": 1.5077857971191406,
-      "learning_rate": 8.867638190954774e-05,
-      "loss": 5.0241,
-      "step": 11770
-    },
-    {
-      "epoch": 6.138722294654498,
-      "grad_norm": 1.5140588283538818,
-      "learning_rate": 8.867537688442211e-05,
-      "loss": 5.9976,
-      "step": 11771
-    },
-    {
-      "epoch": 6.139243807040417,
-      "grad_norm": 1.4903478622436523,
-      "learning_rate": 8.867437185929649e-05,
-      "loss": 5.6388,
-      "step": 11772
-    },
-    {
-      "epoch": 6.139765319426337,
-      "grad_norm": 1.5027995109558105,
-      "learning_rate": 8.867336683417085e-05,
-      "loss": 5.7344,
-      "step": 11773
-    },
-    {
-      "epoch": 6.140286831812255,
-      "grad_norm": 1.402720332145691,
-      "learning_rate": 8.867236180904523e-05,
-      "loss": 5.7508,
-      "step": 11774
-    },
-    {
-      "epoch": 6.140808344198175,
-      "grad_norm": 1.4316145181655884,
-      "learning_rate": 8.86713567839196e-05,
-      "loss": 5.9082,
-      "step": 11775
-    },
-    {
-      "epoch": 6.141329856584094,
-      "grad_norm": 1.3722779750823975,
-      "learning_rate": 8.867035175879398e-05,
-      "loss": 5.8194,
-      "step": 11776
-    },
-    {
-      "epoch": 6.141851368970013,
-      "grad_norm": 1.4731295108795166,
-      "learning_rate": 8.866934673366835e-05,
-      "loss": 5.3685,
-      "step": 11777
-    },
-    {
-      "epoch": 6.142372881355932,
-      "grad_norm": 1.3905938863754272,
-      "learning_rate": 8.866834170854271e-05,
-      "loss": 5.8951,
-      "step": 11778
-    },
-    {
-      "epoch": 6.142894393741852,
-      "grad_norm": 1.3644099235534668,
-      "learning_rate": 8.866733668341709e-05,
-      "loss": 5.9503,
-      "step": 11779
-    },
-    {
-      "epoch": 6.14341590612777,
-      "grad_norm": 1.3189258575439453,
-      "learning_rate": 8.866633165829146e-05,
-      "loss": 5.8498,
-      "step": 11780
-    },
-    {
-      "epoch": 6.14393741851369,
-      "grad_norm": 1.4673799276351929,
-      "learning_rate": 8.866532663316583e-05,
-      "loss": 5.0007,
-      "step": 11781
-    },
-    {
-      "epoch": 6.144458930899609,
-      "grad_norm": 1.4968609809875488,
-      "learning_rate": 8.86643216080402e-05,
-      "loss": 5.3277,
-      "step": 11782
-    },
-    {
-      "epoch": 6.144980443285528,
-      "grad_norm": 1.353542685508728,
-      "learning_rate": 8.866331658291458e-05,
-      "loss": 5.6635,
-      "step": 11783
-    },
-    {
-      "epoch": 6.145501955671447,
-      "grad_norm": 1.3106651306152344,
-      "learning_rate": 8.866231155778895e-05,
-      "loss": 5.7497,
-      "step": 11784
-    },
-    {
-      "epoch": 6.146023468057367,
-      "grad_norm": 1.517261028289795,
-      "learning_rate": 8.866130653266333e-05,
-      "loss": 5.5916,
-      "step": 11785
-    },
-    {
-      "epoch": 6.146544980443285,
-      "grad_norm": 1.5019961595535278,
-      "learning_rate": 8.86603015075377e-05,
-      "loss": 5.528,
-      "step": 11786
-    },
-    {
-      "epoch": 6.147066492829205,
-      "grad_norm": 1.5186127424240112,
-      "learning_rate": 8.865929648241207e-05,
-      "loss": 5.1882,
-      "step": 11787
-    },
-    {
-      "epoch": 6.147588005215124,
-      "grad_norm": 1.4122878313064575,
-      "learning_rate": 8.865829145728644e-05,
-      "loss": 5.9559,
-      "step": 11788
-    },
-    {
-      "epoch": 6.148109517601043,
-      "grad_norm": 1.4793450832366943,
-      "learning_rate": 8.865728643216081e-05,
-      "loss": 5.4195,
-      "step": 11789
-    },
-    {
-      "epoch": 6.148631029986962,
-      "grad_norm": 1.4662984609603882,
-      "learning_rate": 8.865628140703518e-05,
-      "loss": 5.7764,
-      "step": 11790
-    },
-    {
-      "epoch": 6.149152542372882,
-      "grad_norm": 1.3539190292358398,
-      "learning_rate": 8.865527638190954e-05,
-      "loss": 5.8997,
-      "step": 11791
-    },
-    {
-      "epoch": 6.1496740547588,
-      "grad_norm": 1.5010665655136108,
-      "learning_rate": 8.865427135678392e-05,
-      "loss": 4.4386,
-      "step": 11792
-    },
-    {
-      "epoch": 6.15019556714472,
-      "grad_norm": 1.5357328653335571,
-      "learning_rate": 8.865326633165829e-05,
-      "loss": 5.6625,
-      "step": 11793
-    },
-    {
-      "epoch": 6.150717079530639,
-      "grad_norm": 1.4749938249588013,
-      "learning_rate": 8.865226130653266e-05,
-      "loss": 5.5029,
-      "step": 11794
-    },
-    {
-      "epoch": 6.151238591916558,
-      "grad_norm": 1.4935063123703003,
-      "learning_rate": 8.865125628140704e-05,
-      "loss": 5.788,
-      "step": 11795
-    },
-    {
-      "epoch": 6.151760104302477,
-      "grad_norm": 1.3627625703811646,
-      "learning_rate": 8.865025125628142e-05,
-      "loss": 5.7943,
-      "step": 11796
-    },
-    {
-      "epoch": 6.152281616688397,
-      "grad_norm": 1.435412049293518,
-      "learning_rate": 8.864924623115578e-05,
-      "loss": 5.6105,
-      "step": 11797
-    },
-    {
-      "epoch": 6.152803129074315,
-      "grad_norm": 1.6423413753509521,
-      "learning_rate": 8.864824120603016e-05,
-      "loss": 5.3461,
-      "step": 11798
-    },
-    {
-      "epoch": 6.153324641460235,
-      "grad_norm": 1.4718775749206543,
-      "learning_rate": 8.864723618090453e-05,
-      "loss": 5.5383,
-      "step": 11799
-    },
-    {
-      "epoch": 6.153846153846154,
-      "grad_norm": 1.345640778541565,
-      "learning_rate": 8.86462311557789e-05,
-      "loss": 6.0342,
-      "step": 11800
-    },
-    {
-      "epoch": 6.154367666232073,
-      "grad_norm": 1.5049800872802734,
-      "learning_rate": 8.864522613065327e-05,
-      "loss": 5.4265,
-      "step": 11801
-    },
-    {
-      "epoch": 6.154889178617992,
-      "grad_norm": 1.4488152265548706,
-      "learning_rate": 8.864422110552765e-05,
-      "loss": 5.9624,
-      "step": 11802
-    },
-    {
-      "epoch": 6.155410691003912,
-      "grad_norm": 1.4978324174880981,
-      "learning_rate": 8.864321608040201e-05,
-      "loss": 5.4616,
-      "step": 11803
-    },
-    {
-      "epoch": 6.15593220338983,
-      "grad_norm": 1.4560325145721436,
-      "learning_rate": 8.864221105527639e-05,
-      "loss": 5.7261,
-      "step": 11804
-    },
-    {
-      "epoch": 6.15645371577575,
-      "grad_norm": 1.474463939666748,
-      "learning_rate": 8.864120603015077e-05,
-      "loss": 4.8193,
-      "step": 11805
-    },
-    {
-      "epoch": 6.156975228161669,
-      "grad_norm": 1.3275582790374756,
-      "learning_rate": 8.864020100502513e-05,
-      "loss": 5.7143,
-      "step": 11806
-    },
-    {
-      "epoch": 6.157496740547588,
-      "grad_norm": 1.4722857475280762,
-      "learning_rate": 8.863919597989951e-05,
-      "loss": 5.299,
-      "step": 11807
-    },
-    {
-      "epoch": 6.158018252933507,
-      "grad_norm": 1.4630850553512573,
-      "learning_rate": 8.863819095477387e-05,
-      "loss": 5.797,
-      "step": 11808
-    },
-    {
-      "epoch": 6.158539765319427,
-      "grad_norm": 1.4254553318023682,
-      "learning_rate": 8.863718592964825e-05,
-      "loss": 5.3101,
-      "step": 11809
-    },
-    {
-      "epoch": 6.159061277705345,
-      "grad_norm": 1.5943248271942139,
-      "learning_rate": 8.863618090452261e-05,
-      "loss": 5.6027,
-      "step": 11810
-    },
-    {
-      "epoch": 6.159582790091265,
-      "grad_norm": 1.494563102722168,
-      "learning_rate": 8.863517587939699e-05,
-      "loss": 5.7263,
-      "step": 11811
-    },
-    {
-      "epoch": 6.160104302477184,
-      "grad_norm": 1.5194485187530518,
-      "learning_rate": 8.863417085427136e-05,
-      "loss": 5.4292,
-      "step": 11812
-    },
-    {
-      "epoch": 6.160625814863103,
-      "grad_norm": 1.4063482284545898,
-      "learning_rate": 8.863316582914573e-05,
-      "loss": 5.7952,
-      "step": 11813
-    },
-    {
-      "epoch": 6.161147327249022,
-      "grad_norm": 1.5100476741790771,
-      "learning_rate": 8.86321608040201e-05,
-      "loss": 5.689,
-      "step": 11814
-    },
-    {
-      "epoch": 6.161668839634942,
-      "grad_norm": 1.6573777198791504,
-      "learning_rate": 8.863115577889448e-05,
-      "loss": 5.7039,
-      "step": 11815
-    },
-    {
-      "epoch": 6.16219035202086,
-      "grad_norm": 1.5657254457473755,
-      "learning_rate": 8.863015075376885e-05,
-      "loss": 5.1507,
-      "step": 11816
-    },
-    {
-      "epoch": 6.16271186440678,
-      "grad_norm": 1.4246325492858887,
-      "learning_rate": 8.862914572864323e-05,
-      "loss": 5.6766,
-      "step": 11817
-    },
-    {
-      "epoch": 6.163233376792699,
-      "grad_norm": 1.5750113725662231,
-      "learning_rate": 8.86281407035176e-05,
-      "loss": 5.4888,
-      "step": 11818
-    },
-    {
-      "epoch": 6.163754889178618,
-      "grad_norm": 1.3645308017730713,
-      "learning_rate": 8.862713567839196e-05,
-      "loss": 5.6753,
-      "step": 11819
-    },
-    {
-      "epoch": 6.164276401564537,
-      "grad_norm": 1.4580389261245728,
-      "learning_rate": 8.862613065326634e-05,
-      "loss": 4.9128,
-      "step": 11820
-    },
-    {
-      "epoch": 6.164797913950457,
-      "grad_norm": 1.41416335105896,
-      "learning_rate": 8.86251256281407e-05,
-      "loss": 5.712,
-      "step": 11821
-    },
-    {
-      "epoch": 6.165319426336375,
-      "grad_norm": 1.420069694519043,
-      "learning_rate": 8.862412060301508e-05,
-      "loss": 5.6865,
-      "step": 11822
-    },
-    {
-      "epoch": 6.165840938722295,
-      "grad_norm": 1.4848796129226685,
-      "learning_rate": 8.862311557788944e-05,
-      "loss": 5.6346,
-      "step": 11823
-    },
-    {
-      "epoch": 6.166362451108214,
-      "grad_norm": 1.57956063747406,
-      "learning_rate": 8.862211055276382e-05,
-      "loss": 5.6165,
-      "step": 11824
-    },
-    {
-      "epoch": 6.166883963494133,
-      "grad_norm": 1.5811635255813599,
-      "learning_rate": 8.862110552763819e-05,
-      "loss": 5.146,
-      "step": 11825
-    },
-    {
-      "epoch": 6.167405475880052,
-      "grad_norm": 1.6357394456863403,
-      "learning_rate": 8.862010050251256e-05,
-      "loss": 4.7783,
-      "step": 11826
-    },
-    {
-      "epoch": 6.167926988265972,
-      "grad_norm": 1.397835612297058,
-      "learning_rate": 8.861909547738694e-05,
-      "loss": 5.3284,
-      "step": 11827
-    },
-    {
-      "epoch": 6.16844850065189,
-      "grad_norm": 1.4105814695358276,
-      "learning_rate": 8.861809045226132e-05,
-      "loss": 5.5837,
-      "step": 11828
-    },
-    {
-      "epoch": 6.16897001303781,
-      "grad_norm": 1.4551613330841064,
-      "learning_rate": 8.861708542713568e-05,
-      "loss": 5.5932,
-      "step": 11829
-    },
-    {
-      "epoch": 6.169491525423728,
-      "grad_norm": 1.4965585470199585,
-      "learning_rate": 8.861608040201006e-05,
-      "loss": 5.3643,
-      "step": 11830
-    },
-    {
-      "epoch": 6.170013037809648,
-      "grad_norm": 1.4951233863830566,
-      "learning_rate": 8.861507537688443e-05,
-      "loss": 5.3626,
-      "step": 11831
-    },
-    {
-      "epoch": 6.170534550195567,
-      "grad_norm": 1.5110570192337036,
-      "learning_rate": 8.861407035175879e-05,
-      "loss": 5.3436,
-      "step": 11832
-    },
-    {
-      "epoch": 6.171056062581487,
-      "grad_norm": 1.5097216367721558,
-      "learning_rate": 8.861306532663317e-05,
-      "loss": 5.0431,
-      "step": 11833
-    },
-    {
-      "epoch": 6.171577574967405,
-      "grad_norm": 1.5769481658935547,
-      "learning_rate": 8.861206030150753e-05,
-      "loss": 5.7407,
-      "step": 11834
-    },
-    {
-      "epoch": 6.172099087353325,
-      "grad_norm": 1.3957126140594482,
-      "learning_rate": 8.861105527638191e-05,
-      "loss": 5.7101,
-      "step": 11835
-    },
-    {
-      "epoch": 6.172620599739243,
-      "grad_norm": 1.4627699851989746,
-      "learning_rate": 8.861005025125629e-05,
-      "loss": 5.2413,
-      "step": 11836
-    },
-    {
-      "epoch": 6.173142112125163,
-      "grad_norm": 1.359870433807373,
-      "learning_rate": 8.860904522613067e-05,
-      "loss": 5.5997,
-      "step": 11837
-    },
-    {
-      "epoch": 6.173663624511082,
-      "grad_norm": 1.4907082319259644,
-      "learning_rate": 8.860804020100503e-05,
-      "loss": 4.8432,
-      "step": 11838
-    },
-    {
-      "epoch": 6.174185136897001,
-      "grad_norm": 1.6459912061691284,
-      "learning_rate": 8.860703517587941e-05,
-      "loss": 5.7904,
-      "step": 11839
-    },
-    {
-      "epoch": 6.17470664928292,
-      "grad_norm": 1.3764584064483643,
-      "learning_rate": 8.860603015075377e-05,
-      "loss": 5.7687,
-      "step": 11840
-    },
-    {
-      "epoch": 6.17522816166884,
-      "grad_norm": 1.4441922903060913,
-      "learning_rate": 8.860502512562815e-05,
-      "loss": 5.7576,
-      "step": 11841
-    },
-    {
-      "epoch": 6.175749674054758,
-      "grad_norm": 1.5991111993789673,
-      "learning_rate": 8.860402010050251e-05,
-      "loss": 4.8069,
-      "step": 11842
-    },
-    {
-      "epoch": 6.176271186440678,
-      "grad_norm": 1.6642111539840698,
-      "learning_rate": 8.860301507537689e-05,
-      "loss": 5.8093,
-      "step": 11843
-    },
-    {
-      "epoch": 6.176792698826597,
-      "grad_norm": 1.634757399559021,
-      "learning_rate": 8.860201005025126e-05,
-      "loss": 5.7146,
-      "step": 11844
-    },
-    {
-      "epoch": 6.177314211212516,
-      "grad_norm": 1.3511865139007568,
-      "learning_rate": 8.860100502512562e-05,
-      "loss": 5.4757,
-      "step": 11845
-    },
-    {
-      "epoch": 6.177835723598435,
-      "grad_norm": 1.4301635026931763,
-      "learning_rate": 8.86e-05,
-      "loss": 6.091,
-      "step": 11846
-    },
-    {
-      "epoch": 6.178357235984355,
-      "grad_norm": 1.4631774425506592,
-      "learning_rate": 8.859899497487438e-05,
-      "loss": 5.891,
-      "step": 11847
-    },
-    {
-      "epoch": 6.178878748370273,
-      "grad_norm": 1.7007461786270142,
-      "learning_rate": 8.859798994974875e-05,
-      "loss": 5.2152,
-      "step": 11848
-    },
-    {
-      "epoch": 6.179400260756193,
-      "grad_norm": 1.348065972328186,
-      "learning_rate": 8.859698492462312e-05,
-      "loss": 5.8798,
-      "step": 11849
-    },
-    {
-      "epoch": 6.179921773142112,
-      "grad_norm": 1.3159105777740479,
-      "learning_rate": 8.85959798994975e-05,
-      "loss": 5.523,
-      "step": 11850
-    },
-    {
-      "epoch": 6.180443285528031,
-      "grad_norm": 1.3755289316177368,
-      "learning_rate": 8.859497487437186e-05,
-      "loss": 6.0156,
-      "step": 11851
-    },
-    {
-      "epoch": 6.18096479791395,
-      "grad_norm": 1.5139750242233276,
-      "learning_rate": 8.859396984924624e-05,
-      "loss": 5.2929,
-      "step": 11852
-    },
-    {
-      "epoch": 6.18148631029987,
-      "grad_norm": 1.4493751525878906,
-      "learning_rate": 8.85929648241206e-05,
-      "loss": 5.5841,
-      "step": 11853
-    },
-    {
-      "epoch": 6.182007822685788,
-      "grad_norm": 1.423294186592102,
-      "learning_rate": 8.859195979899498e-05,
-      "loss": 5.7616,
-      "step": 11854
-    },
-    {
-      "epoch": 6.182529335071708,
-      "grad_norm": 1.3874956369400024,
-      "learning_rate": 8.859095477386934e-05,
-      "loss": 5.885,
-      "step": 11855
-    },
-    {
-      "epoch": 6.183050847457627,
-      "grad_norm": 1.2573902606964111,
-      "learning_rate": 8.858994974874372e-05,
-      "loss": 5.3385,
-      "step": 11856
-    },
-    {
-      "epoch": 6.183572359843546,
-      "grad_norm": 1.3708124160766602,
-      "learning_rate": 8.85889447236181e-05,
-      "loss": 5.8966,
-      "step": 11857
-    },
-    {
-      "epoch": 6.184093872229465,
-      "grad_norm": 1.6050593852996826,
-      "learning_rate": 8.858793969849246e-05,
-      "loss": 5.4642,
-      "step": 11858
-    },
-    {
-      "epoch": 6.184615384615385,
-      "grad_norm": 1.5259462594985962,
-      "learning_rate": 8.858693467336684e-05,
-      "loss": 5.7711,
-      "step": 11859
-    },
-    {
-      "epoch": 6.185136897001303,
-      "grad_norm": 1.584122896194458,
-      "learning_rate": 8.858592964824121e-05,
-      "loss": 4.8023,
-      "step": 11860
-    },
-    {
-      "epoch": 6.185658409387223,
-      "grad_norm": 1.5089665651321411,
-      "learning_rate": 8.858492462311558e-05,
-      "loss": 4.4374,
-      "step": 11861
-    },
-    {
-      "epoch": 6.186179921773142,
-      "grad_norm": 1.384785532951355,
-      "learning_rate": 8.858391959798995e-05,
-      "loss": 5.8398,
-      "step": 11862
-    },
-    {
-      "epoch": 6.186701434159061,
-      "grad_norm": 1.4850534200668335,
-      "learning_rate": 8.858291457286433e-05,
-      "loss": 5.6826,
-      "step": 11863
-    },
-    {
-      "epoch": 6.18722294654498,
-      "grad_norm": 1.4849025011062622,
-      "learning_rate": 8.858190954773869e-05,
-      "loss": 5.4371,
-      "step": 11864
-    },
-    {
-      "epoch": 6.1877444589309,
-      "grad_norm": 1.5544782876968384,
-      "learning_rate": 8.858090452261307e-05,
-      "loss": 5.7144,
-      "step": 11865
-    },
-    {
-      "epoch": 6.188265971316818,
-      "grad_norm": 1.4645835161209106,
-      "learning_rate": 8.857989949748743e-05,
-      "loss": 5.7876,
-      "step": 11866
-    },
-    {
-      "epoch": 6.188787483702738,
-      "grad_norm": 1.4199129343032837,
-      "learning_rate": 8.857889447236181e-05,
-      "loss": 5.3352,
-      "step": 11867
-    },
-    {
-      "epoch": 6.189308996088657,
-      "grad_norm": 1.4587538242340088,
-      "learning_rate": 8.857788944723619e-05,
-      "loss": 5.5971,
-      "step": 11868
-    },
-    {
-      "epoch": 6.189830508474576,
-      "grad_norm": 1.3427462577819824,
-      "learning_rate": 8.857688442211057e-05,
-      "loss": 5.7332,
-      "step": 11869
-    },
-    {
-      "epoch": 6.190352020860495,
-      "grad_norm": 1.4890226125717163,
-      "learning_rate": 8.857587939698493e-05,
-      "loss": 5.5312,
-      "step": 11870
-    },
-    {
-      "epoch": 6.190873533246415,
-      "grad_norm": 1.4173305034637451,
-      "learning_rate": 8.85748743718593e-05,
-      "loss": 5.7983,
-      "step": 11871
-    },
-    {
-      "epoch": 6.1913950456323334,
-      "grad_norm": 1.6279641389846802,
-      "learning_rate": 8.857386934673367e-05,
-      "loss": 5.5996,
-      "step": 11872
-    },
-    {
-      "epoch": 6.191916558018253,
-      "grad_norm": 1.6986795663833618,
-      "learning_rate": 8.857286432160804e-05,
-      "loss": 5.1919,
-      "step": 11873
-    },
-    {
-      "epoch": 6.192438070404172,
-      "grad_norm": 1.5237427949905396,
-      "learning_rate": 8.857185929648242e-05,
-      "loss": 5.4886,
-      "step": 11874
-    },
-    {
-      "epoch": 6.192959582790091,
-      "grad_norm": 1.4921013116836548,
-      "learning_rate": 8.857085427135678e-05,
-      "loss": 5.5943,
-      "step": 11875
-    },
-    {
-      "epoch": 6.19348109517601,
-      "grad_norm": 1.5208439826965332,
-      "learning_rate": 8.856984924623116e-05,
-      "loss": 5.2234,
-      "step": 11876
-    },
-    {
-      "epoch": 6.19400260756193,
-      "grad_norm": 1.498189091682434,
-      "learning_rate": 8.856884422110554e-05,
-      "loss": 5.6834,
-      "step": 11877
-    },
-    {
-      "epoch": 6.1945241199478485,
-      "grad_norm": 1.3653897047042847,
-      "learning_rate": 8.856783919597991e-05,
-      "loss": 5.8321,
-      "step": 11878
-    },
-    {
-      "epoch": 6.195045632333768,
-      "grad_norm": 1.4282323122024536,
-      "learning_rate": 8.856683417085428e-05,
-      "loss": 5.5926,
-      "step": 11879
-    },
-    {
-      "epoch": 6.195567144719687,
-      "grad_norm": 1.5079749822616577,
-      "learning_rate": 8.856582914572866e-05,
-      "loss": 5.3989,
-      "step": 11880
-    },
-    {
-      "epoch": 6.196088657105606,
-      "grad_norm": 1.409413456916809,
-      "learning_rate": 8.856482412060302e-05,
-      "loss": 5.4374,
-      "step": 11881
-    },
-    {
-      "epoch": 6.196610169491525,
-      "grad_norm": 1.499351978302002,
-      "learning_rate": 8.85638190954774e-05,
-      "loss": 5.7453,
-      "step": 11882
-    },
-    {
-      "epoch": 6.197131681877445,
-      "grad_norm": 1.4912595748901367,
-      "learning_rate": 8.856281407035176e-05,
-      "loss": 4.936,
-      "step": 11883
-    },
-    {
-      "epoch": 6.1976531942633635,
-      "grad_norm": 1.4058117866516113,
-      "learning_rate": 8.856180904522613e-05,
-      "loss": 5.8313,
-      "step": 11884
-    },
-    {
-      "epoch": 6.198174706649283,
-      "grad_norm": 1.445358395576477,
-      "learning_rate": 8.85608040201005e-05,
-      "loss": 5.4986,
-      "step": 11885
-    },
-    {
-      "epoch": 6.198696219035202,
-      "grad_norm": 1.545884132385254,
-      "learning_rate": 8.855979899497487e-05,
-      "loss": 5.1402,
-      "step": 11886
-    },
-    {
-      "epoch": 6.199217731421121,
-      "grad_norm": 1.3578044176101685,
-      "learning_rate": 8.855879396984925e-05,
-      "loss": 5.5937,
-      "step": 11887
-    },
-    {
-      "epoch": 6.1997392438070404,
-      "grad_norm": 1.389089822769165,
-      "learning_rate": 8.855778894472362e-05,
-      "loss": 5.8169,
-      "step": 11888
-    },
-    {
-      "epoch": 6.20026075619296,
-      "grad_norm": 1.643487811088562,
-      "learning_rate": 8.8556783919598e-05,
-      "loss": 5.136,
-      "step": 11889
-    },
-    {
-      "epoch": 6.2007822685788785,
-      "grad_norm": 1.4691789150238037,
-      "learning_rate": 8.855577889447237e-05,
-      "loss": 5.5834,
-      "step": 11890
-    },
-    {
-      "epoch": 6.201303780964798,
-      "grad_norm": 1.4292867183685303,
-      "learning_rate": 8.855477386934674e-05,
-      "loss": 5.6219,
-      "step": 11891
-    },
-    {
-      "epoch": 6.201825293350717,
-      "grad_norm": 1.4417500495910645,
-      "learning_rate": 8.855376884422111e-05,
-      "loss": 5.3366,
-      "step": 11892
-    },
-    {
-      "epoch": 6.202346805736636,
-      "grad_norm": 1.3738152980804443,
-      "learning_rate": 8.855276381909549e-05,
-      "loss": 5.7958,
-      "step": 11893
-    },
-    {
-      "epoch": 6.2028683181225555,
-      "grad_norm": 1.3699332475662231,
-      "learning_rate": 8.855175879396985e-05,
-      "loss": 5.6761,
-      "step": 11894
-    },
-    {
-      "epoch": 6.203389830508475,
-      "grad_norm": 1.5214205980300903,
-      "learning_rate": 8.855075376884423e-05,
-      "loss": 5.4801,
-      "step": 11895
-    },
-    {
-      "epoch": 6.2039113428943935,
-      "grad_norm": 1.4743198156356812,
-      "learning_rate": 8.854974874371859e-05,
-      "loss": 5.6001,
-      "step": 11896
-    },
-    {
-      "epoch": 6.204432855280313,
-      "grad_norm": 1.4038487672805786,
-      "learning_rate": 8.854874371859297e-05,
-      "loss": 5.467,
-      "step": 11897
-    },
-    {
-      "epoch": 6.204954367666232,
-      "grad_norm": 1.4264708757400513,
-      "learning_rate": 8.854773869346735e-05,
-      "loss": 5.2505,
-      "step": 11898
-    },
-    {
-      "epoch": 6.205475880052151,
-      "grad_norm": 1.4308420419692993,
-      "learning_rate": 8.854673366834171e-05,
-      "loss": 5.7043,
-      "step": 11899
-    },
-    {
-      "epoch": 6.2059973924380705,
-      "grad_norm": 1.6168981790542603,
-      "learning_rate": 8.854572864321609e-05,
-      "loss": 5.3636,
-      "step": 11900
-    },
-    {
-      "epoch": 6.20651890482399,
-      "grad_norm": 1.3622676134109497,
-      "learning_rate": 8.854472361809045e-05,
-      "loss": 5.8862,
-      "step": 11901
-    },
-    {
-      "epoch": 6.2070404172099085,
-      "grad_norm": 1.6386746168136597,
-      "learning_rate": 8.854371859296483e-05,
-      "loss": 4.7339,
-      "step": 11902
-    },
-    {
-      "epoch": 6.207561929595828,
-      "grad_norm": 1.375769853591919,
-      "learning_rate": 8.85427135678392e-05,
-      "loss": 5.9584,
-      "step": 11903
-    },
-    {
-      "epoch": 6.208083441981747,
-      "grad_norm": 1.7631757259368896,
-      "learning_rate": 8.854170854271357e-05,
-      "loss": 5.306,
-      "step": 11904
-    },
-    {
-      "epoch": 6.208604954367666,
-      "grad_norm": 1.7043920755386353,
-      "learning_rate": 8.854070351758794e-05,
-      "loss": 4.8314,
-      "step": 11905
-    },
-    {
-      "epoch": 6.2091264667535855,
-      "grad_norm": 1.4538995027542114,
-      "learning_rate": 8.853969849246232e-05,
-      "loss": 5.8703,
-      "step": 11906
-    },
-    {
-      "epoch": 6.209647979139505,
-      "grad_norm": 1.557161569595337,
-      "learning_rate": 8.853869346733668e-05,
-      "loss": 5.1377,
-      "step": 11907
-    },
-    {
-      "epoch": 6.2101694915254235,
-      "grad_norm": 1.4238965511322021,
-      "learning_rate": 8.853768844221106e-05,
-      "loss": 5.8377,
-      "step": 11908
-    },
-    {
-      "epoch": 6.210691003911343,
-      "grad_norm": 1.4127355813980103,
-      "learning_rate": 8.853668341708544e-05,
-      "loss": 5.6875,
-      "step": 11909
-    },
-    {
-      "epoch": 6.2112125162972625,
-      "grad_norm": 1.4903483390808105,
-      "learning_rate": 8.853567839195981e-05,
-      "loss": 5.7097,
-      "step": 11910
-    },
-    {
-      "epoch": 6.211734028683181,
-      "grad_norm": 1.4389986991882324,
-      "learning_rate": 8.853467336683418e-05,
-      "loss": 5.8747,
-      "step": 11911
-    },
-    {
-      "epoch": 6.2122555410691005,
-      "grad_norm": 1.4929208755493164,
-      "learning_rate": 8.853366834170854e-05,
-      "loss": 6.0806,
-      "step": 11912
-    },
-    {
-      "epoch": 6.21277705345502,
-      "grad_norm": 1.4764546155929565,
-      "learning_rate": 8.853266331658292e-05,
-      "loss": 5.7014,
-      "step": 11913
-    },
-    {
-      "epoch": 6.2132985658409385,
-      "grad_norm": 1.4485608339309692,
-      "learning_rate": 8.853165829145728e-05,
-      "loss": 5.7795,
-      "step": 11914
-    },
-    {
-      "epoch": 6.213820078226858,
-      "grad_norm": 1.5148439407348633,
-      "learning_rate": 8.853065326633166e-05,
-      "loss": 5.2809,
-      "step": 11915
-    },
-    {
-      "epoch": 6.2143415906127775,
-      "grad_norm": 1.5536158084869385,
-      "learning_rate": 8.852964824120603e-05,
-      "loss": 5.7626,
-      "step": 11916
-    },
-    {
-      "epoch": 6.214863102998696,
-      "grad_norm": 1.44167959690094,
-      "learning_rate": 8.85286432160804e-05,
-      "loss": 5.5751,
-      "step": 11917
-    },
-    {
-      "epoch": 6.2153846153846155,
-      "grad_norm": 1.5543429851531982,
-      "learning_rate": 8.852763819095478e-05,
-      "loss": 5.0301,
-      "step": 11918
-    },
-    {
-      "epoch": 6.215906127770535,
-      "grad_norm": 1.4895976781845093,
-      "learning_rate": 8.852663316582916e-05,
-      "loss": 5.496,
-      "step": 11919
-    },
-    {
-      "epoch": 6.2164276401564535,
-      "grad_norm": 1.4518386125564575,
-      "learning_rate": 8.852562814070352e-05,
-      "loss": 5.7203,
-      "step": 11920
-    },
-    {
-      "epoch": 6.216949152542373,
-      "grad_norm": 1.4725435972213745,
-      "learning_rate": 8.85246231155779e-05,
-      "loss": 5.6008,
-      "step": 11921
-    },
-    {
-      "epoch": 6.2174706649282925,
-      "grad_norm": 1.5274981260299683,
-      "learning_rate": 8.852361809045227e-05,
-      "loss": 5.6513,
-      "step": 11922
-    },
-    {
-      "epoch": 6.217992177314211,
-      "grad_norm": 1.5392793416976929,
-      "learning_rate": 8.852261306532664e-05,
-      "loss": 5.5344,
-      "step": 11923
-    },
-    {
-      "epoch": 6.2185136897001305,
-      "grad_norm": 1.430160403251648,
-      "learning_rate": 8.852160804020101e-05,
-      "loss": 5.6396,
-      "step": 11924
-    },
-    {
-      "epoch": 6.219035202086049,
-      "grad_norm": 1.3830451965332031,
-      "learning_rate": 8.852060301507537e-05,
-      "loss": 5.44,
-      "step": 11925
-    },
-    {
-      "epoch": 6.219556714471969,
-      "grad_norm": 1.6933056116104126,
-      "learning_rate": 8.851959798994975e-05,
-      "loss": 5.482,
-      "step": 11926
-    },
-    {
-      "epoch": 6.220078226857888,
-      "grad_norm": 1.5601814985275269,
-      "learning_rate": 8.851859296482411e-05,
-      "loss": 5.4514,
-      "step": 11927
-    },
-    {
-      "epoch": 6.2205997392438075,
-      "grad_norm": 1.3834760189056396,
-      "learning_rate": 8.851758793969849e-05,
-      "loss": 5.8367,
-      "step": 11928
-    },
-    {
-      "epoch": 6.221121251629726,
-      "grad_norm": 1.5393314361572266,
-      "learning_rate": 8.851658291457287e-05,
-      "loss": 5.793,
-      "step": 11929
-    },
-    {
-      "epoch": 6.2216427640156455,
-      "grad_norm": 1.4756754636764526,
-      "learning_rate": 8.851557788944725e-05,
-      "loss": 5.6314,
-      "step": 11930
-    },
-    {
-      "epoch": 6.222164276401564,
-      "grad_norm": 1.3858733177185059,
-      "learning_rate": 8.851457286432161e-05,
-      "loss": 5.2934,
-      "step": 11931
-    },
-    {
-      "epoch": 6.222685788787484,
-      "grad_norm": 1.4901131391525269,
-      "learning_rate": 8.851356783919599e-05,
-      "loss": 5.443,
-      "step": 11932
-    },
-    {
-      "epoch": 6.223207301173403,
-      "grad_norm": 1.4807761907577515,
-      "learning_rate": 8.851256281407035e-05,
-      "loss": 5.7936,
-      "step": 11933
-    },
-    {
-      "epoch": 6.223728813559322,
-      "grad_norm": 1.4142582416534424,
-      "learning_rate": 8.851155778894473e-05,
-      "loss": 5.6842,
-      "step": 11934
-    },
-    {
-      "epoch": 6.224250325945241,
-      "grad_norm": 1.4730491638183594,
-      "learning_rate": 8.85105527638191e-05,
-      "loss": 5.7457,
-      "step": 11935
-    },
-    {
-      "epoch": 6.2247718383311605,
-      "grad_norm": 1.6209534406661987,
-      "learning_rate": 8.850954773869347e-05,
-      "loss": 5.2467,
-      "step": 11936
-    },
-    {
-      "epoch": 6.225293350717079,
-      "grad_norm": 1.5777654647827148,
-      "learning_rate": 8.850854271356784e-05,
-      "loss": 5.9202,
-      "step": 11937
-    },
-    {
-      "epoch": 6.225814863102999,
-      "grad_norm": 1.443739414215088,
-      "learning_rate": 8.850753768844222e-05,
-      "loss": 5.5999,
-      "step": 11938
-    },
-    {
-      "epoch": 6.226336375488918,
-      "grad_norm": 1.3688842058181763,
-      "learning_rate": 8.85065326633166e-05,
-      "loss": 5.3899,
-      "step": 11939
-    },
-    {
-      "epoch": 6.226857887874837,
-      "grad_norm": 1.470777988433838,
-      "learning_rate": 8.850552763819096e-05,
-      "loss": 5.783,
-      "step": 11940
-    },
-    {
-      "epoch": 6.227379400260756,
-      "grad_norm": 1.5731507539749146,
-      "learning_rate": 8.850452261306534e-05,
-      "loss": 5.4962,
-      "step": 11941
-    },
-    {
-      "epoch": 6.227900912646676,
-      "grad_norm": 1.4565469026565552,
-      "learning_rate": 8.85035175879397e-05,
-      "loss": 5.7536,
-      "step": 11942
-    },
-    {
-      "epoch": 6.228422425032594,
-      "grad_norm": 1.4491099119186401,
-      "learning_rate": 8.850251256281408e-05,
-      "loss": 5.5301,
-      "step": 11943
-    },
-    {
-      "epoch": 6.228943937418514,
-      "grad_norm": 1.3789137601852417,
-      "learning_rate": 8.850150753768844e-05,
-      "loss": 5.6872,
-      "step": 11944
-    },
-    {
-      "epoch": 6.229465449804433,
-      "grad_norm": 1.3986762762069702,
-      "learning_rate": 8.850050251256282e-05,
-      "loss": 5.5919,
-      "step": 11945
-    },
-    {
-      "epoch": 6.229986962190352,
-      "grad_norm": 1.7459979057312012,
-      "learning_rate": 8.849949748743719e-05,
-      "loss": 4.7146,
-      "step": 11946
-    },
-    {
-      "epoch": 6.230508474576271,
-      "grad_norm": 1.395611047744751,
-      "learning_rate": 8.849849246231156e-05,
-      "loss": 5.5137,
-      "step": 11947
-    },
-    {
-      "epoch": 6.231029986962191,
-      "grad_norm": 1.446032166481018,
-      "learning_rate": 8.849748743718593e-05,
-      "loss": 5.4701,
-      "step": 11948
-    },
-    {
-      "epoch": 6.231551499348109,
-      "grad_norm": 1.48972487449646,
-      "learning_rate": 8.84964824120603e-05,
-      "loss": 5.8392,
-      "step": 11949
-    },
-    {
-      "epoch": 6.232073011734029,
-      "grad_norm": 1.4500093460083008,
-      "learning_rate": 8.849547738693468e-05,
-      "loss": 5.7625,
-      "step": 11950
-    },
-    {
-      "epoch": 6.232594524119948,
-      "grad_norm": 1.540531039237976,
-      "learning_rate": 8.849447236180905e-05,
-      "loss": 5.2494,
-      "step": 11951
-    },
-    {
-      "epoch": 6.233116036505867,
-      "grad_norm": 1.5164334774017334,
-      "learning_rate": 8.849346733668343e-05,
-      "loss": 5.6017,
-      "step": 11952
-    },
-    {
-      "epoch": 6.233637548891786,
-      "grad_norm": 1.4520556926727295,
-      "learning_rate": 8.849246231155779e-05,
-      "loss": 5.3142,
-      "step": 11953
-    },
-    {
-      "epoch": 6.234159061277706,
-      "grad_norm": 1.5704611539840698,
-      "learning_rate": 8.849145728643217e-05,
-      "loss": 4.9755,
-      "step": 11954
-    },
-    {
-      "epoch": 6.234680573663624,
-      "grad_norm": 1.717331051826477,
-      "learning_rate": 8.849045226130653e-05,
-      "loss": 5.8494,
-      "step": 11955
-    },
-    {
-      "epoch": 6.235202086049544,
-      "grad_norm": 1.488868236541748,
-      "learning_rate": 8.848944723618091e-05,
-      "loss": 5.3888,
-      "step": 11956
-    },
-    {
-      "epoch": 6.235723598435463,
-      "grad_norm": 1.455293893814087,
-      "learning_rate": 8.848844221105527e-05,
-      "loss": 5.1665,
-      "step": 11957
-    },
-    {
-      "epoch": 6.236245110821382,
-      "grad_norm": 1.4924079179763794,
-      "learning_rate": 8.848743718592965e-05,
-      "loss": 5.6594,
-      "step": 11958
-    },
-    {
-      "epoch": 6.236766623207301,
-      "grad_norm": 1.3830736875534058,
-      "learning_rate": 8.848643216080403e-05,
-      "loss": 5.6482,
-      "step": 11959
-    },
-    {
-      "epoch": 6.237288135593221,
-      "grad_norm": 1.3520116806030273,
-      "learning_rate": 8.848542713567841e-05,
-      "loss": 5.6981,
-      "step": 11960
-    },
-    {
-      "epoch": 6.237809647979139,
-      "grad_norm": 1.3648649454116821,
-      "learning_rate": 8.848442211055277e-05,
-      "loss": 5.7346,
-      "step": 11961
-    },
-    {
-      "epoch": 6.238331160365059,
-      "grad_norm": 1.3294916152954102,
-      "learning_rate": 8.848341708542715e-05,
-      "loss": 5.7448,
-      "step": 11962
-    },
-    {
-      "epoch": 6.238852672750978,
-      "grad_norm": 1.4094016551971436,
-      "learning_rate": 8.848241206030151e-05,
-      "loss": 5.7646,
-      "step": 11963
-    },
-    {
-      "epoch": 6.239374185136897,
-      "grad_norm": 1.5919846296310425,
-      "learning_rate": 8.848140703517588e-05,
-      "loss": 5.3905,
-      "step": 11964
-    },
-    {
-      "epoch": 6.239895697522816,
-      "grad_norm": 1.7241411209106445,
-      "learning_rate": 8.848040201005026e-05,
-      "loss": 5.4075,
-      "step": 11965
-    },
-    {
-      "epoch": 6.240417209908736,
-      "grad_norm": 1.454823613166809,
-      "learning_rate": 8.847939698492462e-05,
-      "loss": 5.2356,
-      "step": 11966
-    },
-    {
-      "epoch": 6.240938722294654,
-      "grad_norm": 1.4595212936401367,
-      "learning_rate": 8.8478391959799e-05,
-      "loss": 5.3216,
-      "step": 11967
-    },
-    {
-      "epoch": 6.241460234680574,
-      "grad_norm": 1.384568214416504,
-      "learning_rate": 8.847738693467336e-05,
-      "loss": 5.8206,
-      "step": 11968
-    },
-    {
-      "epoch": 6.241981747066493,
-      "grad_norm": 1.3129727840423584,
-      "learning_rate": 8.847638190954774e-05,
-      "loss": 5.7102,
-      "step": 11969
-    },
-    {
-      "epoch": 6.242503259452412,
-      "grad_norm": 1.3979953527450562,
-      "learning_rate": 8.847537688442212e-05,
-      "loss": 4.3481,
-      "step": 11970
-    },
-    {
-      "epoch": 6.243024771838331,
-      "grad_norm": 1.3557648658752441,
-      "learning_rate": 8.84743718592965e-05,
-      "loss": 5.6939,
-      "step": 11971
-    },
-    {
-      "epoch": 6.243546284224251,
-      "grad_norm": 1.3782685995101929,
-      "learning_rate": 8.847336683417086e-05,
-      "loss": 5.8809,
-      "step": 11972
-    },
-    {
-      "epoch": 6.244067796610169,
-      "grad_norm": 1.5902217626571655,
-      "learning_rate": 8.847236180904524e-05,
-      "loss": 5.787,
-      "step": 11973
-    },
-    {
-      "epoch": 6.244589308996089,
-      "grad_norm": 1.5943399667739868,
-      "learning_rate": 8.84713567839196e-05,
-      "loss": 5.1174,
-      "step": 11974
-    },
-    {
-      "epoch": 6.245110821382008,
-      "grad_norm": 1.4708154201507568,
-      "learning_rate": 8.847035175879398e-05,
-      "loss": 5.0109,
-      "step": 11975
-    },
-    {
-      "epoch": 6.245632333767927,
-      "grad_norm": 1.4456918239593506,
-      "learning_rate": 8.846934673366834e-05,
-      "loss": 5.7668,
-      "step": 11976
-    },
-    {
-      "epoch": 6.246153846153846,
-      "grad_norm": 1.4126795530319214,
-      "learning_rate": 8.846834170854271e-05,
-      "loss": 5.6281,
-      "step": 11977
-    },
-    {
-      "epoch": 6.246675358539766,
-      "grad_norm": 1.5016158819198608,
-      "learning_rate": 8.846733668341709e-05,
-      "loss": 5.8234,
-      "step": 11978
-    },
-    {
-      "epoch": 6.247196870925684,
-      "grad_norm": 1.4188992977142334,
-      "learning_rate": 8.846633165829145e-05,
-      "loss": 5.421,
-      "step": 11979
-    },
-    {
-      "epoch": 6.247718383311604,
-      "grad_norm": 1.401224136352539,
-      "learning_rate": 8.846532663316583e-05,
-      "loss": 5.8244,
-      "step": 11980
-    },
-    {
-      "epoch": 6.248239895697523,
-      "grad_norm": 1.3241792917251587,
-      "learning_rate": 8.84643216080402e-05,
-      "loss": 5.7332,
-      "step": 11981
-    },
-    {
-      "epoch": 6.248761408083442,
-      "grad_norm": 1.426016926765442,
-      "learning_rate": 8.846331658291458e-05,
-      "loss": 5.8855,
-      "step": 11982
-    },
-    {
-      "epoch": 6.249282920469361,
-      "grad_norm": 1.3285021781921387,
-      "learning_rate": 8.846231155778895e-05,
-      "loss": 5.4545,
-      "step": 11983
-    },
-    {
-      "epoch": 6.249804432855281,
-      "grad_norm": 1.4279423952102661,
-      "learning_rate": 8.846130653266333e-05,
-      "loss": 5.6828,
-      "step": 11984
-    },
-    {
-      "epoch": 6.250325945241199,
-      "grad_norm": 1.418649435043335,
-      "learning_rate": 8.846030150753769e-05,
-      "loss": 5.9416,
-      "step": 11985
-    },
-    {
-      "epoch": 6.250847457627119,
-      "grad_norm": 1.4831743240356445,
-      "learning_rate": 8.845929648241207e-05,
-      "loss": 5.5728,
-      "step": 11986
-    },
-    {
-      "epoch": 6.251368970013038,
-      "grad_norm": 1.3504599332809448,
-      "learning_rate": 8.845829145728643e-05,
-      "loss": 5.7973,
-      "step": 11987
-    },
-    {
-      "epoch": 6.251890482398957,
-      "grad_norm": 1.589269995689392,
-      "learning_rate": 8.845728643216081e-05,
-      "loss": 4.9596,
-      "step": 11988
-    },
-    {
-      "epoch": 6.252411994784876,
-      "grad_norm": 1.4082505702972412,
-      "learning_rate": 8.845628140703517e-05,
-      "loss": 5.6044,
-      "step": 11989
-    },
-    {
-      "epoch": 6.252933507170796,
-      "grad_norm": 1.5215415954589844,
-      "learning_rate": 8.845527638190955e-05,
-      "loss": 5.4683,
-      "step": 11990
-    },
-    {
-      "epoch": 6.253455019556714,
-      "grad_norm": 1.410811185836792,
-      "learning_rate": 8.845427135678393e-05,
-      "loss": 5.2327,
-      "step": 11991
-    },
-    {
-      "epoch": 6.253976531942634,
-      "grad_norm": 1.560035228729248,
-      "learning_rate": 8.84532663316583e-05,
-      "loss": 5.6108,
-      "step": 11992
-    },
-    {
-      "epoch": 6.254498044328553,
-      "grad_norm": 1.3670291900634766,
-      "learning_rate": 8.845226130653267e-05,
-      "loss": 5.8912,
-      "step": 11993
-    },
-    {
-      "epoch": 6.255019556714472,
-      "grad_norm": 1.5167073011398315,
-      "learning_rate": 8.845125628140704e-05,
-      "loss": 5.4399,
-      "step": 11994
-    },
-    {
-      "epoch": 6.255541069100391,
-      "grad_norm": 1.4532827138900757,
-      "learning_rate": 8.845025125628141e-05,
-      "loss": 5.3466,
-      "step": 11995
-    },
-    {
-      "epoch": 6.256062581486311,
-      "grad_norm": 1.3444067239761353,
-      "learning_rate": 8.844924623115578e-05,
-      "loss": 6.0429,
-      "step": 11996
-    },
-    {
-      "epoch": 6.256584093872229,
-      "grad_norm": 1.4738945960998535,
-      "learning_rate": 8.844824120603016e-05,
-      "loss": 5.4488,
-      "step": 11997
-    },
-    {
-      "epoch": 6.257105606258149,
-      "grad_norm": 1.4280755519866943,
-      "learning_rate": 8.844723618090452e-05,
-      "loss": 5.5526,
-      "step": 11998
-    },
-    {
-      "epoch": 6.257627118644068,
-      "grad_norm": 1.3474198579788208,
-      "learning_rate": 8.84462311557789e-05,
-      "loss": 5.5838,
-      "step": 11999
-    },
-    {
-      "epoch": 6.258148631029987,
-      "grad_norm": 1.6222712993621826,
-      "learning_rate": 8.844522613065326e-05,
-      "loss": 5.5771,
-      "step": 12000
-    },
-    {
-      "epoch": 6.258670143415906,
-      "grad_norm": 1.4202916622161865,
-      "learning_rate": 8.844422110552764e-05,
-      "loss": 5.3481,
-      "step": 12001
-    },
-    {
-      "epoch": 6.259191655801826,
-      "grad_norm": 1.4208847284317017,
-      "learning_rate": 8.844321608040202e-05,
-      "loss": 5.7071,
-      "step": 12002
-    },
-    {
-      "epoch": 6.259713168187744,
-      "grad_norm": 1.4509614706039429,
-      "learning_rate": 8.84422110552764e-05,
-      "loss": 5.801,
-      "step": 12003
-    },
-    {
-      "epoch": 6.260234680573664,
-      "grad_norm": 1.450060486793518,
-      "learning_rate": 8.844120603015076e-05,
-      "loss": 5.3519,
-      "step": 12004
-    },
-    {
-      "epoch": 6.260756192959583,
-      "grad_norm": 1.5493963956832886,
-      "learning_rate": 8.844020100502512e-05,
-      "loss": 5.0173,
-      "step": 12005
-    },
-    {
-      "epoch": 6.261277705345502,
-      "grad_norm": 1.6297320127487183,
-      "learning_rate": 8.84391959798995e-05,
-      "loss": 5.4839,
-      "step": 12006
-    },
-    {
-      "epoch": 6.261799217731421,
-      "grad_norm": 1.5515103340148926,
-      "learning_rate": 8.843819095477387e-05,
-      "loss": 5.6686,
-      "step": 12007
-    },
-    {
-      "epoch": 6.262320730117341,
-      "grad_norm": 1.5418565273284912,
-      "learning_rate": 8.843718592964824e-05,
-      "loss": 5.6546,
-      "step": 12008
-    },
-    {
-      "epoch": 6.262842242503259,
-      "grad_norm": 1.4638097286224365,
-      "learning_rate": 8.843618090452261e-05,
-      "loss": 5.5552,
-      "step": 12009
-    },
-    {
-      "epoch": 6.263363754889179,
-      "grad_norm": 1.5352773666381836,
-      "learning_rate": 8.843517587939699e-05,
-      "loss": 5.4533,
-      "step": 12010
-    },
-    {
-      "epoch": 6.263885267275098,
-      "grad_norm": 1.5890592336654663,
-      "learning_rate": 8.843417085427136e-05,
-      "loss": 5.6646,
-      "step": 12011
-    },
-    {
-      "epoch": 6.264406779661017,
-      "grad_norm": 1.627415657043457,
-      "learning_rate": 8.843316582914574e-05,
-      "loss": 5.4101,
-      "step": 12012
-    },
-    {
-      "epoch": 6.264928292046936,
-      "grad_norm": 1.501802682876587,
-      "learning_rate": 8.84321608040201e-05,
-      "loss": 5.7791,
-      "step": 12013
-    },
-    {
-      "epoch": 6.265449804432855,
-      "grad_norm": 1.5207936763763428,
-      "learning_rate": 8.843115577889448e-05,
-      "loss": 5.2369,
-      "step": 12014
-    },
-    {
-      "epoch": 6.265971316818774,
-      "grad_norm": 1.3493332862854004,
-      "learning_rate": 8.843015075376885e-05,
-      "loss": 5.9268,
-      "step": 12015
-    },
-    {
-      "epoch": 6.266492829204694,
-      "grad_norm": 1.3398523330688477,
-      "learning_rate": 8.842914572864323e-05,
-      "loss": 5.7881,
-      "step": 12016
-    },
-    {
-      "epoch": 6.267014341590613,
-      "grad_norm": 1.3919600248336792,
-      "learning_rate": 8.842814070351759e-05,
-      "loss": 5.8231,
-      "step": 12017
-    },
-    {
-      "epoch": 6.267535853976532,
-      "grad_norm": 1.4427608251571655,
-      "learning_rate": 8.842713567839196e-05,
-      "loss": 5.6764,
-      "step": 12018
-    },
-    {
-      "epoch": 6.268057366362451,
-      "grad_norm": 1.4395346641540527,
-      "learning_rate": 8.842613065326633e-05,
-      "loss": 5.8923,
-      "step": 12019
-    },
-    {
-      "epoch": 6.26857887874837,
-      "grad_norm": 1.5666147470474243,
-      "learning_rate": 8.84251256281407e-05,
-      "loss": 5.5648,
-      "step": 12020
-    },
-    {
-      "epoch": 6.269100391134289,
-      "grad_norm": 1.4631036520004272,
-      "learning_rate": 8.842412060301508e-05,
-      "loss": 5.3906,
-      "step": 12021
-    },
-    {
-      "epoch": 6.269621903520209,
-      "grad_norm": 1.498826503753662,
-      "learning_rate": 8.842311557788945e-05,
-      "loss": 6.0368,
-      "step": 12022
-    },
-    {
-      "epoch": 6.270143415906128,
-      "grad_norm": 1.4078971147537231,
-      "learning_rate": 8.842211055276383e-05,
-      "loss": 5.443,
-      "step": 12023
-    },
-    {
-      "epoch": 6.270664928292047,
-      "grad_norm": 1.3704824447631836,
-      "learning_rate": 8.84211055276382e-05,
-      "loss": 5.9619,
-      "step": 12024
-    },
-    {
-      "epoch": 6.271186440677966,
-      "grad_norm": 1.4154819250106812,
-      "learning_rate": 8.842010050251257e-05,
-      "loss": 5.7941,
-      "step": 12025
-    },
-    {
-      "epoch": 6.271707953063885,
-      "grad_norm": 1.5781630277633667,
-      "learning_rate": 8.841909547738694e-05,
-      "loss": 5.4478,
-      "step": 12026
-    },
-    {
-      "epoch": 6.272229465449804,
-      "grad_norm": 1.4118572473526,
-      "learning_rate": 8.841809045226131e-05,
-      "loss": 5.4527,
-      "step": 12027
-    },
-    {
-      "epoch": 6.272750977835724,
-      "grad_norm": 1.3601629734039307,
-      "learning_rate": 8.841708542713568e-05,
-      "loss": 5.2582,
-      "step": 12028
-    },
-    {
-      "epoch": 6.273272490221642,
-      "grad_norm": 1.5208673477172852,
-      "learning_rate": 8.841608040201006e-05,
-      "loss": 5.4067,
-      "step": 12029
-    },
-    {
-      "epoch": 6.273794002607562,
-      "grad_norm": 1.55955171585083,
-      "learning_rate": 8.841507537688442e-05,
-      "loss": 5.2712,
-      "step": 12030
-    },
-    {
-      "epoch": 6.274315514993481,
-      "grad_norm": 1.4853382110595703,
-      "learning_rate": 8.84140703517588e-05,
-      "loss": 5.2547,
-      "step": 12031
-    },
-    {
-      "epoch": 6.2748370273794,
-      "grad_norm": 1.4155550003051758,
-      "learning_rate": 8.841306532663318e-05,
-      "loss": 5.7208,
-      "step": 12032
-    },
-    {
-      "epoch": 6.275358539765319,
-      "grad_norm": 1.453655481338501,
-      "learning_rate": 8.841206030150754e-05,
-      "loss": 5.2229,
-      "step": 12033
-    },
-    {
-      "epoch": 6.275880052151239,
-      "grad_norm": 1.3679389953613281,
-      "learning_rate": 8.841105527638192e-05,
-      "loss": 5.8292,
-      "step": 12034
-    },
-    {
-      "epoch": 6.276401564537157,
-      "grad_norm": 1.4981225728988647,
-      "learning_rate": 8.841005025125628e-05,
-      "loss": 5.6897,
-      "step": 12035
-    },
-    {
-      "epoch": 6.276923076923077,
-      "grad_norm": 1.3583247661590576,
-      "learning_rate": 8.840904522613066e-05,
-      "loss": 5.469,
-      "step": 12036
-    },
-    {
-      "epoch": 6.277444589308996,
-      "grad_norm": 1.4579613208770752,
-      "learning_rate": 8.840804020100503e-05,
-      "loss": 5.0794,
-      "step": 12037
-    },
-    {
-      "epoch": 6.277966101694915,
-      "grad_norm": 1.503066062927246,
-      "learning_rate": 8.84070351758794e-05,
-      "loss": 5.804,
-      "step": 12038
-    },
-    {
-      "epoch": 6.278487614080834,
-      "grad_norm": 1.3480857610702515,
-      "learning_rate": 8.840603015075377e-05,
-      "loss": 5.6839,
-      "step": 12039
-    },
-    {
-      "epoch": 6.279009126466754,
-      "grad_norm": 1.3260538578033447,
-      "learning_rate": 8.840502512562815e-05,
-      "loss": 5.8841,
-      "step": 12040
-    },
-    {
-      "epoch": 6.279530638852672,
-      "grad_norm": 1.2928334474563599,
-      "learning_rate": 8.840402010050251e-05,
-      "loss": 5.7839,
-      "step": 12041
-    },
-    {
-      "epoch": 6.280052151238592,
-      "grad_norm": 1.4295276403427124,
-      "learning_rate": 8.840301507537689e-05,
-      "loss": 5.8267,
-      "step": 12042
-    },
-    {
-      "epoch": 6.280573663624511,
-      "grad_norm": 1.4535051584243774,
-      "learning_rate": 8.840201005025127e-05,
-      "loss": 5.0047,
-      "step": 12043
-    },
-    {
-      "epoch": 6.28109517601043,
-      "grad_norm": 1.391901969909668,
-      "learning_rate": 8.840100502512563e-05,
-      "loss": 5.6807,
-      "step": 12044
-    },
-    {
-      "epoch": 6.281616688396349,
-      "grad_norm": 1.4072856903076172,
-      "learning_rate": 8.840000000000001e-05,
-      "loss": 5.7186,
-      "step": 12045
-    },
-    {
-      "epoch": 6.282138200782269,
-      "grad_norm": 1.4760102033615112,
-      "learning_rate": 8.839899497487437e-05,
-      "loss": 5.6033,
-      "step": 12046
-    },
-    {
-      "epoch": 6.282659713168187,
-      "grad_norm": 1.4833502769470215,
-      "learning_rate": 8.839798994974875e-05,
-      "loss": 5.8992,
-      "step": 12047
-    },
-    {
-      "epoch": 6.283181225554107,
-      "grad_norm": 1.4219847917556763,
-      "learning_rate": 8.839698492462311e-05,
-      "loss": 5.8911,
-      "step": 12048
-    },
-    {
-      "epoch": 6.283702737940026,
-      "grad_norm": 1.5908845663070679,
-      "learning_rate": 8.839597989949749e-05,
-      "loss": 5.6855,
-      "step": 12049
-    },
-    {
-      "epoch": 6.284224250325945,
-      "grad_norm": 1.4310568571090698,
-      "learning_rate": 8.839497487437186e-05,
-      "loss": 5.4337,
-      "step": 12050
-    },
-    {
-      "epoch": 6.284745762711864,
-      "grad_norm": 1.5826183557510376,
-      "learning_rate": 8.839396984924623e-05,
-      "loss": 5.6392,
-      "step": 12051
-    },
-    {
-      "epoch": 6.285267275097784,
-      "grad_norm": 1.4675840139389038,
-      "learning_rate": 8.839296482412061e-05,
-      "loss": 5.4086,
-      "step": 12052
-    },
-    {
-      "epoch": 6.285788787483702,
-      "grad_norm": 1.5819295644760132,
-      "learning_rate": 8.839195979899499e-05,
-      "loss": 5.3551,
-      "step": 12053
-    },
-    {
-      "epoch": 6.286310299869622,
-      "grad_norm": 1.3784035444259644,
-      "learning_rate": 8.839095477386935e-05,
-      "loss": 5.4538,
-      "step": 12054
-    },
-    {
-      "epoch": 6.286831812255541,
-      "grad_norm": 1.453951120376587,
-      "learning_rate": 8.838994974874373e-05,
-      "loss": 5.1917,
-      "step": 12055
-    },
-    {
-      "epoch": 6.28735332464146,
-      "grad_norm": 1.4149829149246216,
-      "learning_rate": 8.83889447236181e-05,
-      "loss": 5.8249,
-      "step": 12056
-    },
-    {
-      "epoch": 6.287874837027379,
-      "grad_norm": 1.307054042816162,
-      "learning_rate": 8.838793969849246e-05,
-      "loss": 5.9435,
-      "step": 12057
-    },
-    {
-      "epoch": 6.288396349413299,
-      "grad_norm": 1.3327168226242065,
-      "learning_rate": 8.838693467336684e-05,
-      "loss": 5.3628,
-      "step": 12058
-    },
-    {
-      "epoch": 6.288917861799217,
-      "grad_norm": 1.4397270679473877,
-      "learning_rate": 8.83859296482412e-05,
-      "loss": 5.5649,
-      "step": 12059
-    },
-    {
-      "epoch": 6.289439374185137,
-      "grad_norm": 1.378709316253662,
-      "learning_rate": 8.838492462311558e-05,
-      "loss": 5.8172,
-      "step": 12060
-    },
-    {
-      "epoch": 6.289960886571056,
-      "grad_norm": 1.4211546182632446,
-      "learning_rate": 8.838391959798994e-05,
-      "loss": 5.4336,
-      "step": 12061
-    },
-    {
-      "epoch": 6.290482398956975,
-      "grad_norm": 1.3965972661972046,
-      "learning_rate": 8.838291457286432e-05,
-      "loss": 5.4897,
-      "step": 12062
-    },
-    {
-      "epoch": 6.291003911342894,
-      "grad_norm": 1.4729199409484863,
-      "learning_rate": 8.83819095477387e-05,
-      "loss": 5.2197,
-      "step": 12063
-    },
-    {
-      "epoch": 6.291525423728814,
-      "grad_norm": 1.5061179399490356,
-      "learning_rate": 8.838090452261308e-05,
-      "loss": 5.4361,
-      "step": 12064
-    },
-    {
-      "epoch": 6.292046936114732,
-      "grad_norm": 1.4339280128479004,
-      "learning_rate": 8.837989949748744e-05,
-      "loss": 5.6993,
-      "step": 12065
-    },
-    {
-      "epoch": 6.292568448500652,
-      "grad_norm": 1.4180926084518433,
-      "learning_rate": 8.837889447236182e-05,
-      "loss": 5.4446,
-      "step": 12066
-    },
-    {
-      "epoch": 6.293089960886571,
-      "grad_norm": 1.3643224239349365,
-      "learning_rate": 8.837788944723618e-05,
-      "loss": 4.8061,
-      "step": 12067
-    },
-    {
-      "epoch": 6.29361147327249,
-      "grad_norm": 1.5898480415344238,
-      "learning_rate": 8.837688442211056e-05,
-      "loss": 5.2628,
-      "step": 12068
-    },
-    {
-      "epoch": 6.294132985658409,
-      "grad_norm": 1.377640962600708,
-      "learning_rate": 8.837587939698493e-05,
-      "loss": 5.5901,
-      "step": 12069
-    },
-    {
-      "epoch": 6.294654498044329,
-      "grad_norm": 1.373678207397461,
-      "learning_rate": 8.837487437185929e-05,
-      "loss": 5.984,
-      "step": 12070
-    },
-    {
-      "epoch": 6.295176010430247,
-      "grad_norm": 1.59443199634552,
-      "learning_rate": 8.837386934673367e-05,
-      "loss": 4.9514,
-      "step": 12071
-    },
-    {
-      "epoch": 6.295697522816167,
-      "grad_norm": 1.4677097797393799,
-      "learning_rate": 8.837286432160805e-05,
-      "loss": 5.7602,
-      "step": 12072
-    },
-    {
-      "epoch": 6.296219035202086,
-      "grad_norm": 1.4832299947738647,
-      "learning_rate": 8.837185929648242e-05,
-      "loss": 4.8366,
-      "step": 12073
-    },
-    {
-      "epoch": 6.296740547588005,
-      "grad_norm": 1.3001480102539062,
-      "learning_rate": 8.837085427135679e-05,
-      "loss": 5.9922,
-      "step": 12074
-    },
-    {
-      "epoch": 6.297262059973924,
-      "grad_norm": 1.5553359985351562,
-      "learning_rate": 8.836984924623117e-05,
-      "loss": 5.2948,
-      "step": 12075
-    },
-    {
-      "epoch": 6.297783572359844,
-      "grad_norm": 1.4530389308929443,
-      "learning_rate": 8.836884422110553e-05,
-      "loss": 5.8257,
-      "step": 12076
-    },
-    {
-      "epoch": 6.2983050847457624,
-      "grad_norm": 1.4759005308151245,
-      "learning_rate": 8.836783919597991e-05,
-      "loss": 5.6376,
-      "step": 12077
-    },
-    {
-      "epoch": 6.298826597131682,
-      "grad_norm": 1.3651950359344482,
-      "learning_rate": 8.836683417085427e-05,
-      "loss": 5.5992,
-      "step": 12078
-    },
-    {
-      "epoch": 6.299348109517601,
-      "grad_norm": 1.5011557340621948,
-      "learning_rate": 8.836582914572865e-05,
-      "loss": 5.7003,
-      "step": 12079
-    },
-    {
-      "epoch": 6.29986962190352,
-      "grad_norm": 1.3884644508361816,
-      "learning_rate": 8.836482412060301e-05,
-      "loss": 5.9019,
-      "step": 12080
-    },
-    {
-      "epoch": 6.300391134289439,
-      "grad_norm": 1.4578614234924316,
-      "learning_rate": 8.836381909547739e-05,
-      "loss": 5.9988,
-      "step": 12081
-    },
-    {
-      "epoch": 6.300912646675359,
-      "grad_norm": 1.5090234279632568,
-      "learning_rate": 8.836281407035176e-05,
-      "loss": 5.5877,
-      "step": 12082
-    },
-    {
-      "epoch": 6.3014341590612775,
-      "grad_norm": 1.4326368570327759,
-      "learning_rate": 8.836180904522613e-05,
-      "loss": 5.8152,
-      "step": 12083
-    },
-    {
-      "epoch": 6.301955671447197,
-      "grad_norm": 1.7344613075256348,
-      "learning_rate": 8.836080402010051e-05,
-      "loss": 5.0787,
-      "step": 12084
-    },
-    {
-      "epoch": 6.302477183833116,
-      "grad_norm": 2.0626323223114014,
-      "learning_rate": 8.835979899497488e-05,
-      "loss": 4.6779,
-      "step": 12085
-    },
-    {
-      "epoch": 6.302998696219035,
-      "grad_norm": 1.5382282733917236,
-      "learning_rate": 8.835879396984925e-05,
-      "loss": 5.6686,
-      "step": 12086
-    },
-    {
-      "epoch": 6.303520208604954,
-      "grad_norm": 1.5743306875228882,
-      "learning_rate": 8.835778894472362e-05,
-      "loss": 5.3387,
-      "step": 12087
-    },
-    {
-      "epoch": 6.304041720990874,
-      "grad_norm": 1.3784205913543701,
-      "learning_rate": 8.8356783919598e-05,
-      "loss": 5.8669,
-      "step": 12088
-    },
-    {
-      "epoch": 6.3045632333767925,
-      "grad_norm": 1.3935288190841675,
-      "learning_rate": 8.835577889447236e-05,
-      "loss": 5.8338,
-      "step": 12089
-    },
-    {
-      "epoch": 6.305084745762712,
-      "grad_norm": 1.533415675163269,
-      "learning_rate": 8.835477386934674e-05,
-      "loss": 5.5201,
-      "step": 12090
-    },
-    {
-      "epoch": 6.305606258148631,
-      "grad_norm": 1.4763392210006714,
-      "learning_rate": 8.83537688442211e-05,
-      "loss": 5.4968,
-      "step": 12091
-    },
-    {
-      "epoch": 6.30612777053455,
-      "grad_norm": 1.3943382501602173,
-      "learning_rate": 8.835276381909548e-05,
-      "loss": 5.8321,
-      "step": 12092
-    },
-    {
-      "epoch": 6.3066492829204694,
-      "grad_norm": 1.5169154405593872,
-      "learning_rate": 8.835175879396986e-05,
-      "loss": 5.7063,
-      "step": 12093
-    },
-    {
-      "epoch": 6.307170795306389,
-      "grad_norm": 1.3771291971206665,
-      "learning_rate": 8.835075376884424e-05,
-      "loss": 5.9224,
-      "step": 12094
-    },
-    {
-      "epoch": 6.3076923076923075,
-      "grad_norm": 1.5625802278518677,
-      "learning_rate": 8.83497487437186e-05,
-      "loss": 5.6444,
-      "step": 12095
-    },
-    {
-      "epoch": 6.308213820078227,
-      "grad_norm": 1.4044710397720337,
-      "learning_rate": 8.834874371859298e-05,
-      "loss": 5.628,
-      "step": 12096
-    },
-    {
-      "epoch": 6.308735332464146,
-      "grad_norm": 1.3922559022903442,
-      "learning_rate": 8.834773869346734e-05,
-      "loss": 5.8079,
-      "step": 12097
-    },
-    {
-      "epoch": 6.309256844850065,
-      "grad_norm": 1.3689286708831787,
-      "learning_rate": 8.834673366834171e-05,
-      "loss": 5.9705,
-      "step": 12098
-    },
-    {
-      "epoch": 6.3097783572359845,
-      "grad_norm": 1.4325518608093262,
-      "learning_rate": 8.834572864321608e-05,
-      "loss": 5.7138,
-      "step": 12099
-    },
-    {
-      "epoch": 6.310299869621904,
-      "grad_norm": 1.475907325744629,
-      "learning_rate": 8.834472361809045e-05,
-      "loss": 5.4447,
-      "step": 12100
-    },
-    {
-      "epoch": 6.3108213820078225,
-      "grad_norm": 1.5030018091201782,
-      "learning_rate": 8.834371859296483e-05,
-      "loss": 5.1858,
-      "step": 12101
-    },
-    {
-      "epoch": 6.311342894393742,
-      "grad_norm": 1.2887983322143555,
-      "learning_rate": 8.834271356783919e-05,
-      "loss": 5.8435,
-      "step": 12102
-    },
-    {
-      "epoch": 6.311864406779661,
-      "grad_norm": 1.4396398067474365,
-      "learning_rate": 8.834170854271357e-05,
-      "loss": 5.5629,
-      "step": 12103
-    },
-    {
-      "epoch": 6.31238591916558,
-      "grad_norm": 1.5048037767410278,
-      "learning_rate": 8.834070351758795e-05,
-      "loss": 5.7272,
-      "step": 12104
-    },
-    {
-      "epoch": 6.3129074315514995,
-      "grad_norm": 1.51114821434021,
-      "learning_rate": 8.833969849246232e-05,
-      "loss": 5.825,
-      "step": 12105
-    },
-    {
-      "epoch": 6.313428943937419,
-      "grad_norm": 1.3946329355239868,
-      "learning_rate": 8.833869346733669e-05,
-      "loss": 5.8396,
-      "step": 12106
-    },
-    {
-      "epoch": 6.3139504563233375,
-      "grad_norm": 1.3331453800201416,
-      "learning_rate": 8.833768844221107e-05,
-      "loss": 5.8089,
-      "step": 12107
-    },
-    {
-      "epoch": 6.314471968709257,
-      "grad_norm": 1.3997777700424194,
-      "learning_rate": 8.833668341708543e-05,
-      "loss": 5.8215,
-      "step": 12108
-    },
-    {
-      "epoch": 6.3149934810951756,
-      "grad_norm": 1.4033890962600708,
-      "learning_rate": 8.833567839195981e-05,
-      "loss": 5.5975,
-      "step": 12109
-    },
-    {
-      "epoch": 6.315514993481095,
-      "grad_norm": 1.5121886730194092,
-      "learning_rate": 8.833467336683417e-05,
-      "loss": 5.6423,
-      "step": 12110
-    },
-    {
-      "epoch": 6.3160365058670145,
-      "grad_norm": 1.6517807245254517,
-      "learning_rate": 8.833366834170854e-05,
-      "loss": 5.2433,
-      "step": 12111
-    },
-    {
-      "epoch": 6.316558018252934,
-      "grad_norm": 1.4465112686157227,
-      "learning_rate": 8.833266331658292e-05,
-      "loss": 5.7615,
-      "step": 12112
-    },
-    {
-      "epoch": 6.3170795306388525,
-      "grad_norm": 1.4385017156600952,
-      "learning_rate": 8.833165829145729e-05,
-      "loss": 5.5185,
-      "step": 12113
-    },
-    {
-      "epoch": 6.317601043024772,
-      "grad_norm": 1.3362377882003784,
-      "learning_rate": 8.833065326633167e-05,
-      "loss": 5.5966,
-      "step": 12114
-    },
-    {
-      "epoch": 6.318122555410691,
-      "grad_norm": 1.467989444732666,
-      "learning_rate": 8.832964824120604e-05,
-      "loss": 5.769,
-      "step": 12115
-    },
-    {
-      "epoch": 6.31864406779661,
-      "grad_norm": 1.2873549461364746,
-      "learning_rate": 8.832864321608041e-05,
-      "loss": 5.8695,
-      "step": 12116
-    },
-    {
-      "epoch": 6.3191655801825295,
-      "grad_norm": 1.3832297325134277,
-      "learning_rate": 8.832763819095478e-05,
-      "loss": 5.4459,
-      "step": 12117
-    },
-    {
-      "epoch": 6.319687092568449,
-      "grad_norm": 1.3490833044052124,
-      "learning_rate": 8.832663316582916e-05,
-      "loss": 5.9918,
-      "step": 12118
-    },
-    {
-      "epoch": 6.3202086049543675,
-      "grad_norm": 1.6328246593475342,
-      "learning_rate": 8.832562814070352e-05,
-      "loss": 5.508,
-      "step": 12119
-    },
-    {
-      "epoch": 6.320730117340287,
-      "grad_norm": 1.3724193572998047,
-      "learning_rate": 8.83246231155779e-05,
-      "loss": 5.6168,
-      "step": 12120
-    },
-    {
-      "epoch": 6.321251629726206,
-      "grad_norm": 1.4536322355270386,
-      "learning_rate": 8.832361809045226e-05,
-      "loss": 5.013,
-      "step": 12121
-    },
-    {
-      "epoch": 6.321773142112125,
-      "grad_norm": 1.5350182056427002,
-      "learning_rate": 8.832261306532664e-05,
-      "loss": 5.1548,
-      "step": 12122
-    },
-    {
-      "epoch": 6.3222946544980445,
-      "grad_norm": 1.5707682371139526,
-      "learning_rate": 8.8321608040201e-05,
-      "loss": 5.1853,
-      "step": 12123
-    },
-    {
-      "epoch": 6.322816166883963,
-      "grad_norm": 1.5580750703811646,
-      "learning_rate": 8.832060301507538e-05,
-      "loss": 5.2579,
-      "step": 12124
-    },
-    {
-      "epoch": 6.3233376792698825,
-      "grad_norm": 1.3891470432281494,
-      "learning_rate": 8.831959798994976e-05,
-      "loss": 5.8914,
-      "step": 12125
-    },
-    {
-      "epoch": 6.323859191655802,
-      "grad_norm": 1.4920660257339478,
-      "learning_rate": 8.831859296482412e-05,
-      "loss": 5.503,
-      "step": 12126
-    },
-    {
-      "epoch": 6.324380704041721,
-      "grad_norm": 1.4493019580841064,
-      "learning_rate": 8.83175879396985e-05,
-      "loss": 5.5421,
-      "step": 12127
-    },
-    {
-      "epoch": 6.32490221642764,
-      "grad_norm": 1.3207662105560303,
-      "learning_rate": 8.831658291457287e-05,
-      "loss": 6.0284,
-      "step": 12128
-    },
-    {
-      "epoch": 6.3254237288135595,
-      "grad_norm": 1.3509985208511353,
-      "learning_rate": 8.831557788944724e-05,
-      "loss": 5.6847,
-      "step": 12129
-    },
-    {
-      "epoch": 6.325945241199478,
-      "grad_norm": 1.4327774047851562,
-      "learning_rate": 8.831457286432161e-05,
-      "loss": 5.5769,
-      "step": 12130
-    },
-    {
-      "epoch": 6.326466753585398,
-      "grad_norm": 1.3275338411331177,
-      "learning_rate": 8.831356783919599e-05,
-      "loss": 5.6197,
-      "step": 12131
-    },
-    {
-      "epoch": 6.326988265971317,
-      "grad_norm": 1.440687656402588,
-      "learning_rate": 8.831256281407035e-05,
-      "loss": 5.3316,
-      "step": 12132
-    },
-    {
-      "epoch": 6.327509778357236,
-      "grad_norm": 1.3027369976043701,
-      "learning_rate": 8.831155778894473e-05,
-      "loss": 6.0799,
-      "step": 12133
-    },
-    {
-      "epoch": 6.328031290743155,
-      "grad_norm": 1.628475546836853,
-      "learning_rate": 8.83105527638191e-05,
-      "loss": 5.1924,
-      "step": 12134
-    },
-    {
-      "epoch": 6.3285528031290745,
-      "grad_norm": 1.344660758972168,
-      "learning_rate": 8.830954773869348e-05,
-      "loss": 5.9238,
-      "step": 12135
-    },
-    {
-      "epoch": 6.329074315514993,
-      "grad_norm": 1.2707737684249878,
-      "learning_rate": 8.830854271356785e-05,
-      "loss": 5.9772,
-      "step": 12136
-    },
-    {
-      "epoch": 6.329595827900913,
-      "grad_norm": 1.6080198287963867,
-      "learning_rate": 8.830753768844221e-05,
-      "loss": 5.4598,
-      "step": 12137
-    },
-    {
-      "epoch": 6.330117340286832,
-      "grad_norm": 1.6154594421386719,
-      "learning_rate": 8.830653266331659e-05,
-      "loss": 5.2443,
-      "step": 12138
-    },
-    {
-      "epoch": 6.330638852672751,
-      "grad_norm": 1.4336411952972412,
-      "learning_rate": 8.830552763819095e-05,
-      "loss": 5.8761,
-      "step": 12139
-    },
-    {
-      "epoch": 6.33116036505867,
-      "grad_norm": 1.6947381496429443,
-      "learning_rate": 8.830452261306533e-05,
-      "loss": 5.4769,
-      "step": 12140
-    },
-    {
-      "epoch": 6.3316818774445895,
-      "grad_norm": 1.4000518321990967,
-      "learning_rate": 8.83035175879397e-05,
-      "loss": 5.8468,
-      "step": 12141
-    },
-    {
-      "epoch": 6.332203389830508,
-      "grad_norm": 1.4407309293746948,
-      "learning_rate": 8.830251256281407e-05,
-      "loss": 5.6561,
-      "step": 12142
-    },
-    {
-      "epoch": 6.332724902216428,
-      "grad_norm": 1.4863505363464355,
-      "learning_rate": 8.830150753768844e-05,
-      "loss": 5.4698,
-      "step": 12143
-    },
-    {
-      "epoch": 6.333246414602347,
-      "grad_norm": 1.5159229040145874,
-      "learning_rate": 8.830050251256282e-05,
-      "loss": 5.6421,
-      "step": 12144
-    },
-    {
-      "epoch": 6.333767926988266,
-      "grad_norm": 1.378055214881897,
-      "learning_rate": 8.82994974874372e-05,
-      "loss": 5.9447,
-      "step": 12145
-    },
-    {
-      "epoch": 6.334289439374185,
-      "grad_norm": 1.3307565450668335,
-      "learning_rate": 8.829849246231157e-05,
-      "loss": 5.6431,
-      "step": 12146
-    },
-    {
-      "epoch": 6.334810951760105,
-      "grad_norm": 1.526389718055725,
-      "learning_rate": 8.829748743718594e-05,
-      "loss": 5.5554,
-      "step": 12147
-    },
-    {
-      "epoch": 6.335332464146023,
-      "grad_norm": 1.4343065023422241,
-      "learning_rate": 8.829648241206031e-05,
-      "loss": 5.4554,
-      "step": 12148
-    },
-    {
-      "epoch": 6.335853976531943,
-      "grad_norm": 1.3296794891357422,
-      "learning_rate": 8.829547738693468e-05,
-      "loss": 5.9344,
-      "step": 12149
-    },
-    {
-      "epoch": 6.336375488917862,
-      "grad_norm": 1.435777187347412,
-      "learning_rate": 8.829447236180904e-05,
-      "loss": 5.6421,
-      "step": 12150
-    },
-    {
-      "epoch": 6.336897001303781,
-      "grad_norm": 1.3877465724945068,
-      "learning_rate": 8.829346733668342e-05,
-      "loss": 5.5561,
-      "step": 12151
-    },
-    {
-      "epoch": 6.3374185136897,
-      "grad_norm": 1.5047132968902588,
-      "learning_rate": 8.829246231155778e-05,
-      "loss": 5.3564,
-      "step": 12152
-    },
-    {
-      "epoch": 6.33794002607562,
-      "grad_norm": 1.3505843877792358,
-      "learning_rate": 8.829145728643216e-05,
-      "loss": 5.6948,
-      "step": 12153
-    },
-    {
-      "epoch": 6.338461538461538,
-      "grad_norm": 1.330729603767395,
-      "learning_rate": 8.829045226130653e-05,
-      "loss": 5.994,
-      "step": 12154
-    },
-    {
-      "epoch": 6.338983050847458,
-      "grad_norm": 1.5584123134613037,
-      "learning_rate": 8.82894472361809e-05,
-      "loss": 5.4732,
-      "step": 12155
-    },
-    {
-      "epoch": 6.339504563233377,
-      "grad_norm": 1.5374057292938232,
-      "learning_rate": 8.828844221105528e-05,
-      "loss": 5.5678,
-      "step": 12156
-    },
-    {
-      "epoch": 6.340026075619296,
-      "grad_norm": 1.4868539571762085,
-      "learning_rate": 8.828743718592966e-05,
-      "loss": 5.2561,
-      "step": 12157
-    },
-    {
-      "epoch": 6.340547588005215,
-      "grad_norm": 1.3896902799606323,
-      "learning_rate": 8.828643216080402e-05,
-      "loss": 5.3905,
-      "step": 12158
-    },
-    {
-      "epoch": 6.341069100391135,
-      "grad_norm": 1.3424299955368042,
-      "learning_rate": 8.82854271356784e-05,
-      "loss": 5.7307,
-      "step": 12159
-    },
-    {
-      "epoch": 6.341590612777053,
-      "grad_norm": 1.4135481119155884,
-      "learning_rate": 8.828442211055277e-05,
-      "loss": 5.763,
-      "step": 12160
-    },
-    {
-      "epoch": 6.342112125162973,
-      "grad_norm": 1.558407187461853,
-      "learning_rate": 8.828341708542714e-05,
-      "loss": 5.3588,
-      "step": 12161
-    },
-    {
-      "epoch": 6.342633637548892,
-      "grad_norm": 1.5697002410888672,
-      "learning_rate": 8.828241206030151e-05,
-      "loss": 5.3563,
-      "step": 12162
-    },
-    {
-      "epoch": 6.343155149934811,
-      "grad_norm": 1.4083223342895508,
-      "learning_rate": 8.828140703517587e-05,
-      "loss": 5.9154,
-      "step": 12163
-    },
-    {
-      "epoch": 6.34367666232073,
-      "grad_norm": 1.7371326684951782,
-      "learning_rate": 8.828040201005025e-05,
-      "loss": 5.8864,
-      "step": 12164
-    },
-    {
-      "epoch": 6.34419817470665,
-      "grad_norm": 1.6310065984725952,
-      "learning_rate": 8.827939698492463e-05,
-      "loss": 5.8073,
-      "step": 12165
-    },
-    {
-      "epoch": 6.344719687092568,
-      "grad_norm": 1.528121829032898,
-      "learning_rate": 8.8278391959799e-05,
-      "loss": 5.359,
-      "step": 12166
-    },
-    {
-      "epoch": 6.345241199478488,
-      "grad_norm": 1.533721923828125,
-      "learning_rate": 8.827738693467337e-05,
-      "loss": 5.5962,
-      "step": 12167
-    },
-    {
-      "epoch": 6.345762711864407,
-      "grad_norm": 1.4835429191589355,
-      "learning_rate": 8.827638190954775e-05,
-      "loss": 5.6752,
-      "step": 12168
-    },
-    {
-      "epoch": 6.346284224250326,
-      "grad_norm": 1.4723503589630127,
-      "learning_rate": 8.827537688442211e-05,
-      "loss": 5.6876,
-      "step": 12169
-    },
-    {
-      "epoch": 6.346805736636245,
-      "grad_norm": 1.506423830986023,
-      "learning_rate": 8.827437185929649e-05,
-      "loss": 5.0883,
-      "step": 12170
-    },
-    {
-      "epoch": 6.347327249022165,
-      "grad_norm": 1.251632809638977,
-      "learning_rate": 8.827336683417085e-05,
-      "loss": 5.9783,
-      "step": 12171
-    },
-    {
-      "epoch": 6.347848761408083,
-      "grad_norm": 1.4227330684661865,
-      "learning_rate": 8.827236180904523e-05,
-      "loss": 6.0599,
-      "step": 12172
-    },
-    {
-      "epoch": 6.348370273794003,
-      "grad_norm": 1.3502322435379028,
-      "learning_rate": 8.82713567839196e-05,
-      "loss": 5.4769,
-      "step": 12173
-    },
-    {
-      "epoch": 6.348891786179922,
-      "grad_norm": 1.4116591215133667,
-      "learning_rate": 8.827035175879397e-05,
-      "loss": 5.4481,
-      "step": 12174
-    },
-    {
-      "epoch": 6.349413298565841,
-      "grad_norm": 1.4646697044372559,
-      "learning_rate": 8.826934673366834e-05,
-      "loss": 5.6691,
-      "step": 12175
-    },
-    {
-      "epoch": 6.34993481095176,
-      "grad_norm": 1.4485068321228027,
-      "learning_rate": 8.826834170854272e-05,
-      "loss": 5.4802,
-      "step": 12176
-    },
-    {
-      "epoch": 6.35045632333768,
-      "grad_norm": 1.4672069549560547,
-      "learning_rate": 8.82673366834171e-05,
-      "loss": 5.5141,
-      "step": 12177
-    },
-    {
-      "epoch": 6.350977835723598,
-      "grad_norm": 1.296041488647461,
-      "learning_rate": 8.826633165829146e-05,
-      "loss": 5.8917,
-      "step": 12178
-    },
-    {
-      "epoch": 6.351499348109518,
-      "grad_norm": 1.480319857597351,
-      "learning_rate": 8.826532663316584e-05,
-      "loss": 5.8102,
-      "step": 12179
-    },
-    {
-      "epoch": 6.352020860495437,
-      "grad_norm": 1.5142804384231567,
-      "learning_rate": 8.82643216080402e-05,
-      "loss": 5.5116,
-      "step": 12180
-    },
-    {
-      "epoch": 6.352542372881356,
-      "grad_norm": 1.4915592670440674,
-      "learning_rate": 8.826331658291458e-05,
-      "loss": 5.2372,
-      "step": 12181
-    },
-    {
-      "epoch": 6.353063885267275,
-      "grad_norm": 1.5487685203552246,
-      "learning_rate": 8.826231155778894e-05,
-      "loss": 4.7799,
-      "step": 12182
-    },
-    {
-      "epoch": 6.353585397653195,
-      "grad_norm": 1.3946934938430786,
-      "learning_rate": 8.826130653266332e-05,
-      "loss": 5.6234,
-      "step": 12183
-    },
-    {
-      "epoch": 6.354106910039113,
-      "grad_norm": 1.5801653861999512,
-      "learning_rate": 8.826030150753769e-05,
-      "loss": 5.3972,
-      "step": 12184
-    },
-    {
-      "epoch": 6.354628422425033,
-      "grad_norm": 1.4909510612487793,
-      "learning_rate": 8.825929648241206e-05,
-      "loss": 5.3546,
-      "step": 12185
-    },
-    {
-      "epoch": 6.355149934810952,
-      "grad_norm": 1.4385992288589478,
-      "learning_rate": 8.825829145728644e-05,
-      "loss": 5.29,
-      "step": 12186
-    },
-    {
-      "epoch": 6.355671447196871,
-      "grad_norm": 1.4090741872787476,
-      "learning_rate": 8.825728643216082e-05,
-      "loss": 5.737,
-      "step": 12187
-    },
-    {
-      "epoch": 6.35619295958279,
-      "grad_norm": 1.4292742013931274,
-      "learning_rate": 8.825628140703518e-05,
-      "loss": 4.9245,
-      "step": 12188
-    },
-    {
-      "epoch": 6.35671447196871,
-      "grad_norm": 1.4844692945480347,
-      "learning_rate": 8.825527638190956e-05,
-      "loss": 5.7668,
-      "step": 12189
-    },
-    {
-      "epoch": 6.357235984354628,
-      "grad_norm": 1.4666094779968262,
-      "learning_rate": 8.825427135678393e-05,
-      "loss": 5.5606,
-      "step": 12190
-    },
-    {
-      "epoch": 6.357757496740548,
-      "grad_norm": 1.4430983066558838,
-      "learning_rate": 8.825326633165829e-05,
-      "loss": 5.5265,
-      "step": 12191
-    },
-    {
-      "epoch": 6.358279009126467,
-      "grad_norm": 1.4285188913345337,
-      "learning_rate": 8.825226130653267e-05,
-      "loss": 5.7811,
-      "step": 12192
-    },
-    {
-      "epoch": 6.358800521512386,
-      "grad_norm": 1.4459666013717651,
-      "learning_rate": 8.825125628140703e-05,
-      "loss": 5.7705,
-      "step": 12193
-    },
-    {
-      "epoch": 6.359322033898305,
-      "grad_norm": 1.3884646892547607,
-      "learning_rate": 8.825025125628141e-05,
-      "loss": 6.0039,
-      "step": 12194
-    },
-    {
-      "epoch": 6.359843546284225,
-      "grad_norm": 1.3178671598434448,
-      "learning_rate": 8.824924623115577e-05,
-      "loss": 5.9169,
-      "step": 12195
-    },
-    {
-      "epoch": 6.360365058670143,
-      "grad_norm": 1.510746717453003,
-      "learning_rate": 8.824824120603015e-05,
-      "loss": 5.4208,
-      "step": 12196
-    },
-    {
-      "epoch": 6.360886571056063,
-      "grad_norm": 1.4888139963150024,
-      "learning_rate": 8.824723618090453e-05,
-      "loss": 5.6368,
-      "step": 12197
-    },
-    {
-      "epoch": 6.361408083441981,
-      "grad_norm": 1.3968088626861572,
-      "learning_rate": 8.824623115577891e-05,
-      "loss": 5.4919,
-      "step": 12198
-    },
-    {
-      "epoch": 6.361929595827901,
-      "grad_norm": 1.405064344406128,
-      "learning_rate": 8.824522613065327e-05,
-      "loss": 5.4922,
-      "step": 12199
-    },
-    {
-      "epoch": 6.36245110821382,
-      "grad_norm": 2.084150791168213,
-      "learning_rate": 8.824422110552765e-05,
-      "loss": 4.2513,
-      "step": 12200
-    },
-    {
-      "epoch": 6.36297262059974,
-      "grad_norm": 1.5728200674057007,
-      "learning_rate": 8.824321608040201e-05,
-      "loss": 5.9125,
-      "step": 12201
-    },
-    {
-      "epoch": 6.363494132985658,
-      "grad_norm": 1.5503734350204468,
-      "learning_rate": 8.824221105527639e-05,
-      "loss": 5.9045,
-      "step": 12202
-    },
-    {
-      "epoch": 6.364015645371578,
-      "grad_norm": 1.5716289281845093,
-      "learning_rate": 8.824120603015076e-05,
-      "loss": 4.7721,
-      "step": 12203
-    },
-    {
-      "epoch": 6.364537157757496,
-      "grad_norm": 1.4194059371948242,
-      "learning_rate": 8.824020100502512e-05,
-      "loss": 5.519,
-      "step": 12204
-    },
-    {
-      "epoch": 6.365058670143416,
-      "grad_norm": 1.3935153484344482,
-      "learning_rate": 8.82391959798995e-05,
-      "loss": 5.4924,
-      "step": 12205
-    },
-    {
-      "epoch": 6.365580182529335,
-      "grad_norm": 1.4296725988388062,
-      "learning_rate": 8.823819095477388e-05,
-      "loss": 5.7943,
-      "step": 12206
-    },
-    {
-      "epoch": 6.366101694915255,
-      "grad_norm": 1.4515386819839478,
-      "learning_rate": 8.823718592964825e-05,
-      "loss": 5.4516,
-      "step": 12207
-    },
-    {
-      "epoch": 6.366623207301173,
-      "grad_norm": 1.3468106985092163,
-      "learning_rate": 8.823618090452262e-05,
-      "loss": 5.6462,
-      "step": 12208
-    },
-    {
-      "epoch": 6.367144719687093,
-      "grad_norm": 1.4124257564544678,
-      "learning_rate": 8.8235175879397e-05,
-      "loss": 5.7585,
-      "step": 12209
-    },
-    {
-      "epoch": 6.367666232073011,
-      "grad_norm": 1.381523609161377,
-      "learning_rate": 8.823417085427136e-05,
-      "loss": 5.6055,
-      "step": 12210
-    },
-    {
-      "epoch": 6.368187744458931,
-      "grad_norm": 1.4750747680664062,
-      "learning_rate": 8.823316582914574e-05,
-      "loss": 5.5155,
-      "step": 12211
-    },
-    {
-      "epoch": 6.36870925684485,
-      "grad_norm": 1.4453010559082031,
-      "learning_rate": 8.82321608040201e-05,
-      "loss": 5.5929,
-      "step": 12212
-    },
-    {
-      "epoch": 6.36923076923077,
-      "grad_norm": 1.5138916969299316,
-      "learning_rate": 8.823115577889448e-05,
-      "loss": 5.3795,
-      "step": 12213
-    },
-    {
-      "epoch": 6.369752281616688,
-      "grad_norm": 1.699020504951477,
-      "learning_rate": 8.823015075376884e-05,
-      "loss": 5.3188,
-      "step": 12214
-    },
-    {
-      "epoch": 6.370273794002608,
-      "grad_norm": 1.786414623260498,
-      "learning_rate": 8.822914572864322e-05,
-      "loss": 5.3821,
-      "step": 12215
-    },
-    {
-      "epoch": 6.370795306388526,
-      "grad_norm": 1.3971307277679443,
-      "learning_rate": 8.822814070351759e-05,
-      "loss": 5.52,
-      "step": 12216
-    },
-    {
-      "epoch": 6.371316818774446,
-      "grad_norm": 1.4634970426559448,
-      "learning_rate": 8.822713567839196e-05,
-      "loss": 5.8883,
-      "step": 12217
-    },
-    {
-      "epoch": 6.371838331160365,
-      "grad_norm": 1.4427783489227295,
-      "learning_rate": 8.822613065326634e-05,
-      "loss": 5.4954,
-      "step": 12218
-    },
-    {
-      "epoch": 6.372359843546284,
-      "grad_norm": 1.4870893955230713,
-      "learning_rate": 8.82251256281407e-05,
-      "loss": 5.7131,
-      "step": 12219
-    },
-    {
-      "epoch": 6.372881355932203,
-      "grad_norm": 1.4399052858352661,
-      "learning_rate": 8.822412060301508e-05,
-      "loss": 5.2521,
-      "step": 12220
-    },
-    {
-      "epoch": 6.373402868318123,
-      "grad_norm": 1.3952280282974243,
-      "learning_rate": 8.822311557788945e-05,
-      "loss": 5.3951,
-      "step": 12221
-    },
-    {
-      "epoch": 6.373924380704041,
-      "grad_norm": 1.397067666053772,
-      "learning_rate": 8.822211055276383e-05,
-      "loss": 5.618,
-      "step": 12222
-    },
-    {
-      "epoch": 6.374445893089961,
-      "grad_norm": 1.4171732664108276,
-      "learning_rate": 8.822110552763819e-05,
-      "loss": 5.548,
-      "step": 12223
-    },
-    {
-      "epoch": 6.37496740547588,
-      "grad_norm": 1.2858281135559082,
-      "learning_rate": 8.822010050251257e-05,
-      "loss": 6.0247,
-      "step": 12224
-    },
-    {
-      "epoch": 6.375488917861799,
-      "grad_norm": 1.3728171586990356,
-      "learning_rate": 8.821909547738693e-05,
-      "loss": 5.8506,
-      "step": 12225
-    },
-    {
-      "epoch": 6.376010430247718,
-      "grad_norm": 1.4187681674957275,
-      "learning_rate": 8.821809045226131e-05,
-      "loss": 5.4468,
-      "step": 12226
-    },
-    {
-      "epoch": 6.376531942633638,
-      "grad_norm": 1.3169276714324951,
-      "learning_rate": 8.821708542713569e-05,
-      "loss": 5.783,
-      "step": 12227
-    },
-    {
-      "epoch": 6.377053455019556,
-      "grad_norm": 1.49747633934021,
-      "learning_rate": 8.821608040201007e-05,
-      "loss": 5.4559,
-      "step": 12228
-    },
-    {
-      "epoch": 6.377574967405476,
-      "grad_norm": 1.409165859222412,
-      "learning_rate": 8.821507537688443e-05,
-      "loss": 5.7901,
-      "step": 12229
-    },
-    {
-      "epoch": 6.378096479791395,
-      "grad_norm": 1.4317535161972046,
-      "learning_rate": 8.82140703517588e-05,
-      "loss": 5.369,
-      "step": 12230
-    },
-    {
-      "epoch": 6.378617992177314,
-      "grad_norm": 1.6110883951187134,
-      "learning_rate": 8.821306532663317e-05,
-      "loss": 4.9418,
-      "step": 12231
-    },
-    {
-      "epoch": 6.379139504563233,
-      "grad_norm": 1.5186666250228882,
-      "learning_rate": 8.821206030150754e-05,
-      "loss": 5.358,
-      "step": 12232
-    },
-    {
-      "epoch": 6.379661016949153,
-      "grad_norm": 1.4872722625732422,
-      "learning_rate": 8.821105527638191e-05,
-      "loss": 5.7829,
-      "step": 12233
-    },
-    {
-      "epoch": 6.380182529335071,
-      "grad_norm": 1.4706742763519287,
-      "learning_rate": 8.821005025125628e-05,
-      "loss": 5.4025,
-      "step": 12234
-    },
-    {
-      "epoch": 6.380704041720991,
-      "grad_norm": 1.4497088193893433,
-      "learning_rate": 8.820904522613066e-05,
-      "loss": 5.3688,
-      "step": 12235
-    },
-    {
-      "epoch": 6.38122555410691,
-      "grad_norm": 1.4196417331695557,
-      "learning_rate": 8.820804020100502e-05,
-      "loss": 5.5335,
-      "step": 12236
-    },
-    {
-      "epoch": 6.381747066492829,
-      "grad_norm": 1.3944603204727173,
-      "learning_rate": 8.82070351758794e-05,
-      "loss": 5.7376,
-      "step": 12237
-    },
-    {
-      "epoch": 6.382268578878748,
-      "grad_norm": 1.5761350393295288,
-      "learning_rate": 8.820603015075378e-05,
-      "loss": 5.2931,
-      "step": 12238
-    },
-    {
-      "epoch": 6.382790091264668,
-      "grad_norm": 1.5221244096755981,
-      "learning_rate": 8.820502512562815e-05,
-      "loss": 5.3599,
-      "step": 12239
-    },
-    {
-      "epoch": 6.383311603650586,
-      "grad_norm": 1.382493019104004,
-      "learning_rate": 8.820402010050252e-05,
-      "loss": 5.9875,
-      "step": 12240
-    },
-    {
-      "epoch": 6.383833116036506,
-      "grad_norm": 1.335075855255127,
-      "learning_rate": 8.82030150753769e-05,
-      "loss": 5.873,
-      "step": 12241
-    },
-    {
-      "epoch": 6.384354628422425,
-      "grad_norm": 1.2675015926361084,
-      "learning_rate": 8.820201005025126e-05,
-      "loss": 5.8856,
-      "step": 12242
-    },
-    {
-      "epoch": 6.384876140808344,
-      "grad_norm": 1.3323925733566284,
-      "learning_rate": 8.820100502512562e-05,
-      "loss": 6.0389,
-      "step": 12243
-    },
-    {
-      "epoch": 6.385397653194263,
-      "grad_norm": 1.5182682275772095,
-      "learning_rate": 8.82e-05,
-      "loss": 5.3876,
-      "step": 12244
-    },
-    {
-      "epoch": 6.385919165580183,
-      "grad_norm": 1.4591470956802368,
-      "learning_rate": 8.819899497487437e-05,
-      "loss": 5.2304,
-      "step": 12245
-    },
-    {
-      "epoch": 6.386440677966101,
-      "grad_norm": 1.4802595376968384,
-      "learning_rate": 8.819798994974874e-05,
-      "loss": 5.5211,
-      "step": 12246
-    },
-    {
-      "epoch": 6.386962190352021,
-      "grad_norm": 1.4425345659255981,
-      "learning_rate": 8.819698492462312e-05,
-      "loss": 5.7371,
-      "step": 12247
-    },
-    {
-      "epoch": 6.38748370273794,
-      "grad_norm": 1.3161156177520752,
-      "learning_rate": 8.81959798994975e-05,
-      "loss": 5.8716,
-      "step": 12248
-    },
-    {
-      "epoch": 6.388005215123859,
-      "grad_norm": 1.360721230506897,
-      "learning_rate": 8.819497487437186e-05,
-      "loss": 5.9604,
-      "step": 12249
-    },
-    {
-      "epoch": 6.388526727509778,
-      "grad_norm": 1.4703439474105835,
-      "learning_rate": 8.819396984924624e-05,
-      "loss": 5.4409,
-      "step": 12250
-    },
-    {
-      "epoch": 6.389048239895698,
-      "grad_norm": 1.4632480144500732,
-      "learning_rate": 8.81929648241206e-05,
-      "loss": 5.305,
-      "step": 12251
-    },
-    {
-      "epoch": 6.389569752281616,
-      "grad_norm": 1.3937593698501587,
-      "learning_rate": 8.819195979899498e-05,
-      "loss": 5.7847,
-      "step": 12252
-    },
-    {
-      "epoch": 6.390091264667536,
-      "grad_norm": 1.3734287023544312,
-      "learning_rate": 8.819095477386935e-05,
-      "loss": 5.8971,
-      "step": 12253
-    },
-    {
-      "epoch": 6.390612777053455,
-      "grad_norm": 1.4256542921066284,
-      "learning_rate": 8.818994974874373e-05,
-      "loss": 5.3316,
-      "step": 12254
-    },
-    {
-      "epoch": 6.391134289439374,
-      "grad_norm": 1.2939655780792236,
-      "learning_rate": 8.818894472361809e-05,
-      "loss": 5.89,
-      "step": 12255
-    },
-    {
-      "epoch": 6.391655801825293,
-      "grad_norm": 1.4324458837509155,
-      "learning_rate": 8.818793969849246e-05,
-      "loss": 5.7079,
-      "step": 12256
-    },
-    {
-      "epoch": 6.392177314211213,
-      "grad_norm": 1.5167888402938843,
-      "learning_rate": 8.818693467336683e-05,
-      "loss": 5.7998,
-      "step": 12257
-    },
-    {
-      "epoch": 6.392698826597131,
-      "grad_norm": 1.5175211429595947,
-      "learning_rate": 8.818592964824121e-05,
-      "loss": 5.435,
-      "step": 12258
-    },
-    {
-      "epoch": 6.393220338983051,
-      "grad_norm": 1.528090238571167,
-      "learning_rate": 8.818492462311559e-05,
-      "loss": 5.5824,
-      "step": 12259
-    },
-    {
-      "epoch": 6.39374185136897,
-      "grad_norm": 1.4604538679122925,
-      "learning_rate": 8.818391959798995e-05,
-      "loss": 5.5871,
-      "step": 12260
-    },
-    {
-      "epoch": 6.394263363754889,
-      "grad_norm": 1.47371244430542,
-      "learning_rate": 8.818291457286433e-05,
-      "loss": 5.1875,
-      "step": 12261
-    },
-    {
-      "epoch": 6.394784876140808,
-      "grad_norm": 1.3886140584945679,
-      "learning_rate": 8.81819095477387e-05,
-      "loss": 5.491,
-      "step": 12262
-    },
-    {
-      "epoch": 6.395306388526728,
-      "grad_norm": 1.4838526248931885,
-      "learning_rate": 8.818090452261307e-05,
-      "loss": 5.4091,
-      "step": 12263
-    },
-    {
-      "epoch": 6.395827900912646,
-      "grad_norm": 1.3425787687301636,
-      "learning_rate": 8.817989949748744e-05,
-      "loss": 5.6645,
-      "step": 12264
-    },
-    {
-      "epoch": 6.396349413298566,
-      "grad_norm": 1.4481582641601562,
-      "learning_rate": 8.817889447236181e-05,
-      "loss": 5.8373,
-      "step": 12265
-    },
-    {
-      "epoch": 6.396870925684485,
-      "grad_norm": 1.518369436264038,
-      "learning_rate": 8.817788944723618e-05,
-      "loss": 5.3999,
-      "step": 12266
-    },
-    {
-      "epoch": 6.397392438070404,
-      "grad_norm": 1.3389809131622314,
-      "learning_rate": 8.817688442211056e-05,
-      "loss": 5.8671,
-      "step": 12267
-    },
-    {
-      "epoch": 6.397913950456323,
-      "grad_norm": 1.4639378786087036,
-      "learning_rate": 8.817587939698493e-05,
-      "loss": 5.7784,
-      "step": 12268
-    },
-    {
-      "epoch": 6.398435462842243,
-      "grad_norm": 1.5024561882019043,
-      "learning_rate": 8.817487437185931e-05,
-      "loss": 5.9637,
-      "step": 12269
-    },
-    {
-      "epoch": 6.398956975228161,
-      "grad_norm": 1.560990810394287,
-      "learning_rate": 8.817386934673368e-05,
-      "loss": 5.5154,
-      "step": 12270
-    },
-    {
-      "epoch": 6.399478487614081,
-      "grad_norm": 1.2266737222671509,
-      "learning_rate": 8.817286432160804e-05,
-      "loss": 4.8219,
-      "step": 12271
-    },
-    {
-      "epoch": 6.4,
-      "grad_norm": 1.4626342058181763,
-      "learning_rate": 8.817185929648242e-05,
-      "loss": 5.2933,
-      "step": 12272
-    },
-    {
-      "epoch": 6.400521512385919,
-      "grad_norm": 1.427392601966858,
-      "learning_rate": 8.817085427135678e-05,
-      "loss": 5.8878,
-      "step": 12273
-    },
-    {
-      "epoch": 6.401043024771838,
-      "grad_norm": 1.4779373407363892,
-      "learning_rate": 8.816984924623116e-05,
-      "loss": 5.6789,
-      "step": 12274
-    },
-    {
-      "epoch": 6.401564537157758,
-      "grad_norm": 1.3206887245178223,
-      "learning_rate": 8.816884422110553e-05,
-      "loss": 5.0854,
-      "step": 12275
-    },
-    {
-      "epoch": 6.402086049543676,
-      "grad_norm": 1.4355905055999756,
-      "learning_rate": 8.81678391959799e-05,
-      "loss": 5.2877,
-      "step": 12276
-    },
-    {
-      "epoch": 6.402607561929596,
-      "grad_norm": 1.578148603439331,
-      "learning_rate": 8.816683417085427e-05,
-      "loss": 5.3097,
-      "step": 12277
-    },
-    {
-      "epoch": 6.403129074315515,
-      "grad_norm": 1.5697033405303955,
-      "learning_rate": 8.816582914572865e-05,
-      "loss": 5.8357,
-      "step": 12278
-    },
-    {
-      "epoch": 6.403650586701434,
-      "grad_norm": 1.6093648672103882,
-      "learning_rate": 8.816482412060302e-05,
-      "loss": 5.2604,
-      "step": 12279
-    },
-    {
-      "epoch": 6.404172099087353,
-      "grad_norm": 1.4446955919265747,
-      "learning_rate": 8.81638190954774e-05,
-      "loss": 5.4395,
-      "step": 12280
-    },
-    {
-      "epoch": 6.404693611473273,
-      "grad_norm": 1.6522186994552612,
-      "learning_rate": 8.816281407035177e-05,
-      "loss": 5.6057,
-      "step": 12281
-    },
-    {
-      "epoch": 6.4052151238591915,
-      "grad_norm": 1.612200379371643,
-      "learning_rate": 8.816180904522614e-05,
-      "loss": 5.1585,
-      "step": 12282
-    },
-    {
-      "epoch": 6.405736636245111,
-      "grad_norm": 1.4451087713241577,
-      "learning_rate": 8.816080402010051e-05,
-      "loss": 5.6742,
-      "step": 12283
-    },
-    {
-      "epoch": 6.40625814863103,
-      "grad_norm": 1.4551714658737183,
-      "learning_rate": 8.815979899497487e-05,
-      "loss": 5.5968,
-      "step": 12284
-    },
-    {
-      "epoch": 6.406779661016949,
-      "grad_norm": 1.5619081258773804,
-      "learning_rate": 8.815879396984925e-05,
-      "loss": 5.5912,
-      "step": 12285
-    },
-    {
-      "epoch": 6.407301173402868,
-      "grad_norm": 1.412952184677124,
-      "learning_rate": 8.815778894472361e-05,
-      "loss": 5.7991,
-      "step": 12286
-    },
-    {
-      "epoch": 6.407822685788788,
-      "grad_norm": 1.4653518199920654,
-      "learning_rate": 8.815678391959799e-05,
-      "loss": 5.5906,
-      "step": 12287
-    },
-    {
-      "epoch": 6.4083441981747065,
-      "grad_norm": 1.572576880455017,
-      "learning_rate": 8.815577889447237e-05,
-      "loss": 5.319,
-      "step": 12288
-    },
-    {
-      "epoch": 6.408865710560626,
-      "grad_norm": 1.4404702186584473,
-      "learning_rate": 8.815477386934675e-05,
-      "loss": 5.3296,
-      "step": 12289
-    },
-    {
-      "epoch": 6.409387222946545,
-      "grad_norm": 1.482928991317749,
-      "learning_rate": 8.815376884422111e-05,
-      "loss": 5.2498,
-      "step": 12290
-    },
-    {
-      "epoch": 6.409908735332464,
-      "grad_norm": 1.4822885990142822,
-      "learning_rate": 8.815276381909549e-05,
-      "loss": 5.6699,
-      "step": 12291
-    },
-    {
-      "epoch": 6.410430247718383,
-      "grad_norm": 1.5829187631607056,
-      "learning_rate": 8.815175879396985e-05,
-      "loss": 5.616,
-      "step": 12292
-    },
-    {
-      "epoch": 6.410951760104302,
-      "grad_norm": 1.4730305671691895,
-      "learning_rate": 8.815075376884423e-05,
-      "loss": 5.4947,
-      "step": 12293
-    },
-    {
-      "epoch": 6.4114732724902215,
-      "grad_norm": 1.4638900756835938,
-      "learning_rate": 8.81497487437186e-05,
-      "loss": 5.2828,
-      "step": 12294
-    },
-    {
-      "epoch": 6.411994784876141,
-      "grad_norm": 1.6327484846115112,
-      "learning_rate": 8.814874371859297e-05,
-      "loss": 5.4319,
-      "step": 12295
-    },
-    {
-      "epoch": 6.41251629726206,
-      "grad_norm": 1.5855402946472168,
-      "learning_rate": 8.814773869346734e-05,
-      "loss": 5.4827,
-      "step": 12296
-    },
-    {
-      "epoch": 6.413037809647979,
-      "grad_norm": 1.5764530897140503,
-      "learning_rate": 8.81467336683417e-05,
-      "loss": 5.6907,
-      "step": 12297
-    },
-    {
-      "epoch": 6.4135593220338984,
-      "grad_norm": 1.3923953771591187,
-      "learning_rate": 8.814572864321608e-05,
-      "loss": 5.6651,
-      "step": 12298
-    },
-    {
-      "epoch": 6.414080834419817,
-      "grad_norm": 1.4534045457839966,
-      "learning_rate": 8.814472361809046e-05,
-      "loss": 5.7933,
-      "step": 12299
-    },
-    {
-      "epoch": 6.4146023468057365,
-      "grad_norm": 1.592774748802185,
-      "learning_rate": 8.814371859296484e-05,
-      "loss": 4.5249,
-      "step": 12300
-    },
-    {
-      "epoch": 6.415123859191656,
-      "grad_norm": 1.413705825805664,
-      "learning_rate": 8.81427135678392e-05,
-      "loss": 5.7643,
-      "step": 12301
-    },
-    {
-      "epoch": 6.415645371577575,
-      "grad_norm": 1.6202391386032104,
-      "learning_rate": 8.814170854271358e-05,
-      "loss": 5.2539,
-      "step": 12302
-    },
-    {
-      "epoch": 6.416166883963494,
-      "grad_norm": 1.7524176836013794,
-      "learning_rate": 8.814070351758794e-05,
-      "loss": 5.1757,
-      "step": 12303
-    },
-    {
-      "epoch": 6.4166883963494135,
-      "grad_norm": 1.5299781560897827,
-      "learning_rate": 8.813969849246232e-05,
-      "loss": 5.3008,
-      "step": 12304
-    },
-    {
-      "epoch": 6.417209908735332,
-      "grad_norm": 1.4312807321548462,
-      "learning_rate": 8.813869346733668e-05,
-      "loss": 5.6172,
-      "step": 12305
-    },
-    {
-      "epoch": 6.4177314211212515,
-      "grad_norm": 1.5321975946426392,
-      "learning_rate": 8.813768844221106e-05,
-      "loss": 5.5221,
-      "step": 12306
-    },
-    {
-      "epoch": 6.418252933507171,
-      "grad_norm": 1.4345386028289795,
-      "learning_rate": 8.813668341708543e-05,
-      "loss": 5.927,
-      "step": 12307
-    },
-    {
-      "epoch": 6.41877444589309,
-      "grad_norm": 1.618627905845642,
-      "learning_rate": 8.81356783919598e-05,
-      "loss": 4.8395,
-      "step": 12308
-    },
-    {
-      "epoch": 6.419295958279009,
-      "grad_norm": 1.5540077686309814,
-      "learning_rate": 8.813467336683417e-05,
-      "loss": 5.36,
-      "step": 12309
-    },
-    {
-      "epoch": 6.4198174706649285,
-      "grad_norm": 1.444062352180481,
-      "learning_rate": 8.813366834170855e-05,
-      "loss": 5.4994,
-      "step": 12310
-    },
-    {
-      "epoch": 6.420338983050847,
-      "grad_norm": 1.415596604347229,
-      "learning_rate": 8.813266331658292e-05,
-      "loss": 5.7162,
-      "step": 12311
-    },
-    {
-      "epoch": 6.4208604954367665,
-      "grad_norm": 1.401892066001892,
-      "learning_rate": 8.813165829145729e-05,
-      "loss": 5.5236,
-      "step": 12312
-    },
-    {
-      "epoch": 6.421382007822686,
-      "grad_norm": 1.5646836757659912,
-      "learning_rate": 8.813065326633167e-05,
-      "loss": 5.8446,
-      "step": 12313
-    },
-    {
-      "epoch": 6.4219035202086046,
-      "grad_norm": 1.491349458694458,
-      "learning_rate": 8.812964824120603e-05,
-      "loss": 5.74,
-      "step": 12314
-    },
-    {
-      "epoch": 6.422425032594524,
-      "grad_norm": 1.5022974014282227,
-      "learning_rate": 8.812864321608041e-05,
-      "loss": 5.5079,
-      "step": 12315
-    },
-    {
-      "epoch": 6.4229465449804435,
-      "grad_norm": 1.4012523889541626,
-      "learning_rate": 8.812763819095477e-05,
-      "loss": 5.8197,
-      "step": 12316
-    },
-    {
-      "epoch": 6.423468057366362,
-      "grad_norm": 1.3767344951629639,
-      "learning_rate": 8.812663316582915e-05,
-      "loss": 5.7991,
-      "step": 12317
-    },
-    {
-      "epoch": 6.4239895697522815,
-      "grad_norm": 1.9282090663909912,
-      "learning_rate": 8.812562814070351e-05,
-      "loss": 5.4469,
-      "step": 12318
-    },
-    {
-      "epoch": 6.424511082138201,
-      "grad_norm": 1.481793761253357,
-      "learning_rate": 8.812462311557789e-05,
-      "loss": 5.6434,
-      "step": 12319
-    },
-    {
-      "epoch": 6.42503259452412,
-      "grad_norm": 1.4800115823745728,
-      "learning_rate": 8.812361809045227e-05,
-      "loss": 5.6715,
-      "step": 12320
-    },
-    {
-      "epoch": 6.425554106910039,
-      "grad_norm": 1.646868348121643,
-      "learning_rate": 8.812261306532665e-05,
-      "loss": 5.3738,
-      "step": 12321
-    },
-    {
-      "epoch": 6.4260756192959585,
-      "grad_norm": 1.3726975917816162,
-      "learning_rate": 8.812160804020101e-05,
-      "loss": 5.983,
-      "step": 12322
-    },
-    {
-      "epoch": 6.426597131681877,
-      "grad_norm": 1.5390900373458862,
-      "learning_rate": 8.812060301507538e-05,
-      "loss": 5.3409,
-      "step": 12323
-    },
-    {
-      "epoch": 6.4271186440677965,
-      "grad_norm": 1.726402997970581,
-      "learning_rate": 8.811959798994975e-05,
-      "loss": 4.9655,
-      "step": 12324
-    },
-    {
-      "epoch": 6.427640156453716,
-      "grad_norm": 1.6351051330566406,
-      "learning_rate": 8.811859296482412e-05,
-      "loss": 5.1929,
-      "step": 12325
-    },
-    {
-      "epoch": 6.428161668839635,
-      "grad_norm": 1.4899537563323975,
-      "learning_rate": 8.81175879396985e-05,
-      "loss": 5.4563,
-      "step": 12326
-    },
-    {
-      "epoch": 6.428683181225554,
-      "grad_norm": 1.4254094362258911,
-      "learning_rate": 8.811658291457286e-05,
-      "loss": 5.7713,
-      "step": 12327
-    },
-    {
-      "epoch": 6.4292046936114735,
-      "grad_norm": 1.563011646270752,
-      "learning_rate": 8.811557788944724e-05,
-      "loss": 4.9107,
-      "step": 12328
-    },
-    {
-      "epoch": 6.429726205997392,
-      "grad_norm": 1.5598279237747192,
-      "learning_rate": 8.81145728643216e-05,
-      "loss": 5.8376,
-      "step": 12329
-    },
-    {
-      "epoch": 6.4302477183833116,
-      "grad_norm": 1.4684690237045288,
-      "learning_rate": 8.811356783919598e-05,
-      "loss": 5.6438,
-      "step": 12330
-    },
-    {
-      "epoch": 6.430769230769231,
-      "grad_norm": 1.5901228189468384,
-      "learning_rate": 8.811256281407036e-05,
-      "loss": 5.4757,
-      "step": 12331
-    },
-    {
-      "epoch": 6.43129074315515,
-      "grad_norm": 1.4514451026916504,
-      "learning_rate": 8.811155778894474e-05,
-      "loss": 5.7955,
-      "step": 12332
-    },
-    {
-      "epoch": 6.431812255541069,
-      "grad_norm": 1.461692214012146,
-      "learning_rate": 8.81105527638191e-05,
-      "loss": 5.4813,
-      "step": 12333
-    },
-    {
-      "epoch": 6.4323337679269885,
-      "grad_norm": 1.3234068155288696,
-      "learning_rate": 8.810954773869348e-05,
-      "loss": 5.5391,
-      "step": 12334
-    },
-    {
-      "epoch": 6.432855280312907,
-      "grad_norm": 1.4290522336959839,
-      "learning_rate": 8.810854271356784e-05,
-      "loss": 5.6446,
-      "step": 12335
-    },
-    {
-      "epoch": 6.433376792698827,
-      "grad_norm": 1.4482927322387695,
-      "learning_rate": 8.810753768844221e-05,
-      "loss": 5.3787,
-      "step": 12336
-    },
-    {
-      "epoch": 6.433898305084746,
-      "grad_norm": 1.3852169513702393,
-      "learning_rate": 8.810653266331658e-05,
-      "loss": 6.0821,
-      "step": 12337
-    },
-    {
-      "epoch": 6.434419817470665,
-      "grad_norm": 1.2715764045715332,
-      "learning_rate": 8.810552763819095e-05,
-      "loss": 6.0843,
-      "step": 12338
-    },
-    {
-      "epoch": 6.434941329856584,
-      "grad_norm": 2.1181516647338867,
-      "learning_rate": 8.810452261306533e-05,
-      "loss": 5.3792,
-      "step": 12339
-    },
-    {
-      "epoch": 6.4354628422425035,
-      "grad_norm": 1.448868989944458,
-      "learning_rate": 8.81035175879397e-05,
-      "loss": 5.6746,
-      "step": 12340
-    },
-    {
-      "epoch": 6.435984354628422,
-      "grad_norm": 1.4440921545028687,
-      "learning_rate": 8.810251256281408e-05,
-      "loss": 5.8219,
-      "step": 12341
-    },
-    {
-      "epoch": 6.436505867014342,
-      "grad_norm": 1.3736803531646729,
-      "learning_rate": 8.810150753768845e-05,
-      "loss": 5.7113,
-      "step": 12342
-    },
-    {
-      "epoch": 6.437027379400261,
-      "grad_norm": 1.3876579999923706,
-      "learning_rate": 8.810050251256282e-05,
-      "loss": 5.7087,
-      "step": 12343
-    },
-    {
-      "epoch": 6.43754889178618,
-      "grad_norm": 1.359338402748108,
-      "learning_rate": 8.809949748743719e-05,
-      "loss": 5.6221,
-      "step": 12344
-    },
-    {
-      "epoch": 6.438070404172099,
-      "grad_norm": 1.551284909248352,
-      "learning_rate": 8.809849246231157e-05,
-      "loss": 5.3263,
-      "step": 12345
-    },
-    {
-      "epoch": 6.4385919165580185,
-      "grad_norm": 1.3460392951965332,
-      "learning_rate": 8.809748743718593e-05,
-      "loss": 4.76,
-      "step": 12346
-    },
-    {
-      "epoch": 6.439113428943937,
-      "grad_norm": 1.465205192565918,
-      "learning_rate": 8.809648241206031e-05,
-      "loss": 5.9279,
-      "step": 12347
-    },
-    {
-      "epoch": 6.439634941329857,
-      "grad_norm": 1.500546932220459,
-      "learning_rate": 8.809547738693467e-05,
-      "loss": 5.6204,
-      "step": 12348
-    },
-    {
-      "epoch": 6.440156453715776,
-      "grad_norm": 1.4785327911376953,
-      "learning_rate": 8.809447236180904e-05,
-      "loss": 5.8211,
-      "step": 12349
-    },
-    {
-      "epoch": 6.440677966101695,
-      "grad_norm": 1.405735731124878,
-      "learning_rate": 8.809346733668342e-05,
-      "loss": 6.0353,
-      "step": 12350
-    },
-    {
-      "epoch": 6.441199478487614,
-      "grad_norm": 1.4556152820587158,
-      "learning_rate": 8.809246231155779e-05,
-      "loss": 5.7157,
-      "step": 12351
-    },
-    {
-      "epoch": 6.441720990873534,
-      "grad_norm": 1.434609055519104,
-      "learning_rate": 8.809145728643217e-05,
-      "loss": 4.8225,
-      "step": 12352
-    },
-    {
-      "epoch": 6.442242503259452,
-      "grad_norm": 1.7365261316299438,
-      "learning_rate": 8.809045226130654e-05,
-      "loss": 4.8069,
-      "step": 12353
-    },
-    {
-      "epoch": 6.442764015645372,
-      "grad_norm": 1.5130285024642944,
-      "learning_rate": 8.808944723618091e-05,
-      "loss": 5.7442,
-      "step": 12354
-    },
-    {
-      "epoch": 6.443285528031291,
-      "grad_norm": 1.674997329711914,
-      "learning_rate": 8.808844221105528e-05,
-      "loss": 4.9937,
-      "step": 12355
-    },
-    {
-      "epoch": 6.44380704041721,
-      "grad_norm": 1.4525433778762817,
-      "learning_rate": 8.808743718592966e-05,
-      "loss": 5.7793,
-      "step": 12356
-    },
-    {
-      "epoch": 6.444328552803129,
-      "grad_norm": 1.4050140380859375,
-      "learning_rate": 8.808643216080402e-05,
-      "loss": 5.5348,
-      "step": 12357
-    },
-    {
-      "epoch": 6.444850065189049,
-      "grad_norm": 1.4503260850906372,
-      "learning_rate": 8.80854271356784e-05,
-      "loss": 5.7861,
-      "step": 12358
-    },
-    {
-      "epoch": 6.445371577574967,
-      "grad_norm": 1.4740691184997559,
-      "learning_rate": 8.808442211055276e-05,
-      "loss": 5.8656,
-      "step": 12359
-    },
-    {
-      "epoch": 6.445893089960887,
-      "grad_norm": 1.3675254583358765,
-      "learning_rate": 8.808341708542714e-05,
-      "loss": 5.7501,
-      "step": 12360
-    },
-    {
-      "epoch": 6.446414602346806,
-      "grad_norm": 1.4367542266845703,
-      "learning_rate": 8.808241206030152e-05,
-      "loss": 5.7249,
-      "step": 12361
-    },
-    {
-      "epoch": 6.446936114732725,
-      "grad_norm": 1.5021365880966187,
-      "learning_rate": 8.80814070351759e-05,
-      "loss": 5.5976,
-      "step": 12362
-    },
-    {
-      "epoch": 6.447457627118644,
-      "grad_norm": 1.484460473060608,
-      "learning_rate": 8.808040201005026e-05,
-      "loss": 5.6577,
-      "step": 12363
-    },
-    {
-      "epoch": 6.447979139504564,
-      "grad_norm": 1.5684963464736938,
-      "learning_rate": 8.807939698492462e-05,
-      "loss": 5.2501,
-      "step": 12364
-    },
-    {
-      "epoch": 6.448500651890482,
-      "grad_norm": 1.54085373878479,
-      "learning_rate": 8.8078391959799e-05,
-      "loss": 5.1348,
-      "step": 12365
-    },
-    {
-      "epoch": 6.449022164276402,
-      "grad_norm": 1.7493807077407837,
-      "learning_rate": 8.807738693467337e-05,
-      "loss": 5.6407,
-      "step": 12366
-    },
-    {
-      "epoch": 6.449543676662321,
-      "grad_norm": 1.405956745147705,
-      "learning_rate": 8.807638190954774e-05,
-      "loss": 5.596,
-      "step": 12367
-    },
-    {
-      "epoch": 6.45006518904824,
-      "grad_norm": 1.3265783786773682,
-      "learning_rate": 8.807537688442211e-05,
-      "loss": 5.534,
-      "step": 12368
-    },
-    {
-      "epoch": 6.450586701434159,
-      "grad_norm": 1.5398539304733276,
-      "learning_rate": 8.807437185929649e-05,
-      "loss": 4.9131,
-      "step": 12369
-    },
-    {
-      "epoch": 6.451108213820079,
-      "grad_norm": 1.3642629384994507,
-      "learning_rate": 8.807336683417085e-05,
-      "loss": 5.8148,
-      "step": 12370
-    },
-    {
-      "epoch": 6.451629726205997,
-      "grad_norm": 1.3724558353424072,
-      "learning_rate": 8.807236180904523e-05,
-      "loss": 5.5896,
-      "step": 12371
-    },
-    {
-      "epoch": 6.452151238591917,
-      "grad_norm": 1.436241626739502,
-      "learning_rate": 8.80713567839196e-05,
-      "loss": 5.4805,
-      "step": 12372
-    },
-    {
-      "epoch": 6.452672750977836,
-      "grad_norm": 1.3802975416183472,
-      "learning_rate": 8.807035175879398e-05,
-      "loss": 5.6184,
-      "step": 12373
-    },
-    {
-      "epoch": 6.453194263363755,
-      "grad_norm": 1.4066349267959595,
-      "learning_rate": 8.806934673366835e-05,
-      "loss": 5.7961,
-      "step": 12374
-    },
-    {
-      "epoch": 6.453715775749674,
-      "grad_norm": 1.438141942024231,
-      "learning_rate": 8.806834170854273e-05,
-      "loss": 5.4654,
-      "step": 12375
-    },
-    {
-      "epoch": 6.454237288135594,
-      "grad_norm": 1.4204347133636475,
-      "learning_rate": 8.806733668341709e-05,
-      "loss": 5.5599,
-      "step": 12376
-    },
-    {
-      "epoch": 6.454758800521512,
-      "grad_norm": 1.5285028219223022,
-      "learning_rate": 8.806633165829145e-05,
-      "loss": 5.6744,
-      "step": 12377
-    },
-    {
-      "epoch": 6.455280312907432,
-      "grad_norm": 1.6296660900115967,
-      "learning_rate": 8.806532663316583e-05,
-      "loss": 5.083,
-      "step": 12378
-    },
-    {
-      "epoch": 6.455801825293351,
-      "grad_norm": 1.4418717622756958,
-      "learning_rate": 8.80643216080402e-05,
-      "loss": 5.4439,
-      "step": 12379
-    },
-    {
-      "epoch": 6.45632333767927,
-      "grad_norm": 1.4093581438064575,
-      "learning_rate": 8.806331658291457e-05,
-      "loss": 5.345,
-      "step": 12380
-    },
-    {
-      "epoch": 6.456844850065189,
-      "grad_norm": 1.611208438873291,
-      "learning_rate": 8.806231155778895e-05,
-      "loss": 5.7432,
-      "step": 12381
-    },
-    {
-      "epoch": 6.457366362451109,
-      "grad_norm": 1.590815782546997,
-      "learning_rate": 8.806130653266333e-05,
-      "loss": 5.3405,
-      "step": 12382
-    },
-    {
-      "epoch": 6.457887874837027,
-      "grad_norm": 1.3767541646957397,
-      "learning_rate": 8.80603015075377e-05,
-      "loss": 5.7199,
-      "step": 12383
-    },
-    {
-      "epoch": 6.458409387222947,
-      "grad_norm": 1.3843947649002075,
-      "learning_rate": 8.805929648241207e-05,
-      "loss": 5.7899,
-      "step": 12384
-    },
-    {
-      "epoch": 6.458930899608866,
-      "grad_norm": 1.4581185579299927,
-      "learning_rate": 8.805829145728644e-05,
-      "loss": 4.8054,
-      "step": 12385
-    },
-    {
-      "epoch": 6.459452411994785,
-      "grad_norm": 1.5644441843032837,
-      "learning_rate": 8.805728643216081e-05,
-      "loss": 5.0286,
-      "step": 12386
-    },
-    {
-      "epoch": 6.459973924380704,
-      "grad_norm": 1.5694894790649414,
-      "learning_rate": 8.805628140703518e-05,
-      "loss": 5.5962,
-      "step": 12387
-    },
-    {
-      "epoch": 6.460495436766623,
-      "grad_norm": 1.5887560844421387,
-      "learning_rate": 8.805527638190956e-05,
-      "loss": 5.2852,
-      "step": 12388
-    },
-    {
-      "epoch": 6.461016949152542,
-      "grad_norm": 1.4364598989486694,
-      "learning_rate": 8.805427135678392e-05,
-      "loss": 5.2558,
-      "step": 12389
-    },
-    {
-      "epoch": 6.461538461538462,
-      "grad_norm": 1.437892198562622,
-      "learning_rate": 8.805326633165828e-05,
-      "loss": 5.7539,
-      "step": 12390
-    },
-    {
-      "epoch": 6.462059973924381,
-      "grad_norm": 1.5709222555160522,
-      "learning_rate": 8.805226130653266e-05,
-      "loss": 5.1556,
-      "step": 12391
-    },
-    {
-      "epoch": 6.4625814863103,
-      "grad_norm": 1.368564486503601,
-      "learning_rate": 8.805125628140704e-05,
-      "loss": 5.7982,
-      "step": 12392
-    },
-    {
-      "epoch": 6.463102998696219,
-      "grad_norm": 1.7009506225585938,
-      "learning_rate": 8.805025125628142e-05,
-      "loss": 5.3254,
-      "step": 12393
-    },
-    {
-      "epoch": 6.463624511082138,
-      "grad_norm": 1.4114599227905273,
-      "learning_rate": 8.804924623115578e-05,
-      "loss": 5.3063,
-      "step": 12394
-    },
-    {
-      "epoch": 6.464146023468057,
-      "grad_norm": 1.4827611446380615,
-      "learning_rate": 8.804824120603016e-05,
-      "loss": 5.4065,
-      "step": 12395
-    },
-    {
-      "epoch": 6.464667535853977,
-      "grad_norm": 1.415856957435608,
-      "learning_rate": 8.804723618090452e-05,
-      "loss": 5.5719,
-      "step": 12396
-    },
-    {
-      "epoch": 6.465189048239896,
-      "grad_norm": 1.4004614353179932,
-      "learning_rate": 8.80462311557789e-05,
-      "loss": 5.4921,
-      "step": 12397
-    },
-    {
-      "epoch": 6.465710560625815,
-      "grad_norm": 1.4583027362823486,
-      "learning_rate": 8.804522613065327e-05,
-      "loss": 5.6516,
-      "step": 12398
-    },
-    {
-      "epoch": 6.466232073011734,
-      "grad_norm": 1.8655674457550049,
-      "learning_rate": 8.804422110552764e-05,
-      "loss": 5.4548,
-      "step": 12399
-    },
-    {
-      "epoch": 6.466753585397653,
-      "grad_norm": 1.4628874063491821,
-      "learning_rate": 8.804321608040201e-05,
-      "loss": 6.0187,
-      "step": 12400
-    },
-    {
-      "epoch": 6.467275097783572,
-      "grad_norm": 1.3126760721206665,
-      "learning_rate": 8.804221105527639e-05,
-      "loss": 5.7048,
-      "step": 12401
-    },
-    {
-      "epoch": 6.467796610169492,
-      "grad_norm": 1.3370603322982788,
-      "learning_rate": 8.804120603015076e-05,
-      "loss": 5.616,
-      "step": 12402
-    },
-    {
-      "epoch": 6.46831812255541,
-      "grad_norm": 1.4881099462509155,
-      "learning_rate": 8.804020100502513e-05,
-      "loss": 5.7212,
-      "step": 12403
-    },
-    {
-      "epoch": 6.46883963494133,
-      "grad_norm": 1.3161354064941406,
-      "learning_rate": 8.80391959798995e-05,
-      "loss": 5.9275,
-      "step": 12404
-    },
-    {
-      "epoch": 6.469361147327249,
-      "grad_norm": 1.446272611618042,
-      "learning_rate": 8.803819095477387e-05,
-      "loss": 4.7986,
-      "step": 12405
-    },
-    {
-      "epoch": 6.469882659713168,
-      "grad_norm": 1.4071003198623657,
-      "learning_rate": 8.803718592964825e-05,
-      "loss": 5.9418,
-      "step": 12406
-    },
-    {
-      "epoch": 6.470404172099087,
-      "grad_norm": 1.5117669105529785,
-      "learning_rate": 8.803618090452261e-05,
-      "loss": 5.6083,
-      "step": 12407
-    },
-    {
-      "epoch": 6.470925684485007,
-      "grad_norm": 1.4707705974578857,
-      "learning_rate": 8.803517587939699e-05,
-      "loss": 5.3514,
-      "step": 12408
-    },
-    {
-      "epoch": 6.471447196870925,
-      "grad_norm": 1.7434885501861572,
-      "learning_rate": 8.803417085427135e-05,
-      "loss": 5.252,
-      "step": 12409
-    },
-    {
-      "epoch": 6.471968709256845,
-      "grad_norm": 1.5283344984054565,
-      "learning_rate": 8.803316582914573e-05,
-      "loss": 5.7439,
-      "step": 12410
-    },
-    {
-      "epoch": 6.472490221642764,
-      "grad_norm": 1.5107430219650269,
-      "learning_rate": 8.80321608040201e-05,
-      "loss": 5.8647,
-      "step": 12411
-    },
-    {
-      "epoch": 6.473011734028683,
-      "grad_norm": 1.4917230606079102,
-      "learning_rate": 8.803115577889447e-05,
-      "loss": 5.3346,
-      "step": 12412
-    },
-    {
-      "epoch": 6.473533246414602,
-      "grad_norm": 1.560365080833435,
-      "learning_rate": 8.803015075376885e-05,
-      "loss": 5.6884,
-      "step": 12413
-    },
-    {
-      "epoch": 6.474054758800522,
-      "grad_norm": 1.4106907844543457,
-      "learning_rate": 8.802914572864323e-05,
-      "loss": 5.7369,
-      "step": 12414
-    },
-    {
-      "epoch": 6.47457627118644,
-      "grad_norm": 1.7385326623916626,
-      "learning_rate": 8.80281407035176e-05,
-      "loss": 5.4126,
-      "step": 12415
-    },
-    {
-      "epoch": 6.47509778357236,
-      "grad_norm": 1.4627634286880493,
-      "learning_rate": 8.802713567839196e-05,
-      "loss": 5.4544,
-      "step": 12416
-    },
-    {
-      "epoch": 6.475619295958279,
-      "grad_norm": 1.4005321264266968,
-      "learning_rate": 8.802613065326634e-05,
-      "loss": 5.687,
-      "step": 12417
-    },
-    {
-      "epoch": 6.476140808344198,
-      "grad_norm": 1.6482642889022827,
-      "learning_rate": 8.80251256281407e-05,
-      "loss": 5.6323,
-      "step": 12418
-    },
-    {
-      "epoch": 6.476662320730117,
-      "grad_norm": 1.5998446941375732,
-      "learning_rate": 8.802412060301508e-05,
-      "loss": 5.0683,
-      "step": 12419
-    },
-    {
-      "epoch": 6.477183833116037,
-      "grad_norm": 1.469269037246704,
-      "learning_rate": 8.802311557788944e-05,
-      "loss": 5.9523,
-      "step": 12420
-    },
-    {
-      "epoch": 6.477705345501955,
-      "grad_norm": 1.5380696058273315,
-      "learning_rate": 8.802211055276382e-05,
-      "loss": 5.6246,
-      "step": 12421
-    },
-    {
-      "epoch": 6.478226857887875,
-      "grad_norm": 1.4603004455566406,
-      "learning_rate": 8.80211055276382e-05,
-      "loss": 5.7065,
-      "step": 12422
-    },
-    {
-      "epoch": 6.478748370273794,
-      "grad_norm": 1.5581802129745483,
-      "learning_rate": 8.802010050251258e-05,
-      "loss": 5.0572,
-      "step": 12423
-    },
-    {
-      "epoch": 6.479269882659713,
-      "grad_norm": 1.5039197206497192,
-      "learning_rate": 8.801909547738694e-05,
-      "loss": 5.6126,
-      "step": 12424
-    },
-    {
-      "epoch": 6.479791395045632,
-      "grad_norm": 1.5614122152328491,
-      "learning_rate": 8.801809045226132e-05,
-      "loss": 5.5312,
-      "step": 12425
-    },
-    {
-      "epoch": 6.480312907431552,
-      "grad_norm": 1.3942195177078247,
-      "learning_rate": 8.801708542713568e-05,
-      "loss": 5.7958,
-      "step": 12426
-    },
-    {
-      "epoch": 6.48083441981747,
-      "grad_norm": 1.468167781829834,
-      "learning_rate": 8.801608040201006e-05,
-      "loss": 5.6224,
-      "step": 12427
-    },
-    {
-      "epoch": 6.48135593220339,
-      "grad_norm": 1.4568753242492676,
-      "learning_rate": 8.801507537688443e-05,
-      "loss": 5.6129,
-      "step": 12428
-    },
-    {
-      "epoch": 6.481877444589309,
-      "grad_norm": 1.447894811630249,
-      "learning_rate": 8.801407035175879e-05,
-      "loss": 5.1642,
-      "step": 12429
-    },
-    {
-      "epoch": 6.482398956975228,
-      "grad_norm": 1.4283291101455688,
-      "learning_rate": 8.801306532663317e-05,
-      "loss": 5.7924,
-      "step": 12430
-    },
-    {
-      "epoch": 6.482920469361147,
-      "grad_norm": 1.4039815664291382,
-      "learning_rate": 8.801206030150753e-05,
-      "loss": 5.8967,
-      "step": 12431
-    },
-    {
-      "epoch": 6.483441981747067,
-      "grad_norm": 1.5332776308059692,
-      "learning_rate": 8.801105527638191e-05,
-      "loss": 5.5242,
-      "step": 12432
-    },
-    {
-      "epoch": 6.483963494132985,
-      "grad_norm": 1.5946636199951172,
-      "learning_rate": 8.801005025125629e-05,
-      "loss": 5.2754,
-      "step": 12433
-    },
-    {
-      "epoch": 6.484485006518905,
-      "grad_norm": 1.5294524431228638,
-      "learning_rate": 8.800904522613067e-05,
-      "loss": 5.3223,
-      "step": 12434
-    },
-    {
-      "epoch": 6.485006518904824,
-      "grad_norm": 1.486592173576355,
-      "learning_rate": 8.800804020100503e-05,
-      "loss": 5.2983,
-      "step": 12435
-    },
-    {
-      "epoch": 6.485528031290743,
-      "grad_norm": 1.672872543334961,
-      "learning_rate": 8.800703517587941e-05,
-      "loss": 5.2014,
-      "step": 12436
-    },
-    {
-      "epoch": 6.486049543676662,
-      "grad_norm": 1.4293420314788818,
-      "learning_rate": 8.800603015075377e-05,
-      "loss": 5.8217,
-      "step": 12437
-    },
-    {
-      "epoch": 6.486571056062582,
-      "grad_norm": 1.465050220489502,
-      "learning_rate": 8.800502512562815e-05,
-      "loss": 5.6198,
-      "step": 12438
-    },
-    {
-      "epoch": 6.4870925684485,
-      "grad_norm": 1.4624377489089966,
-      "learning_rate": 8.800402010050251e-05,
-      "loss": 5.525,
-      "step": 12439
-    },
-    {
-      "epoch": 6.48761408083442,
-      "grad_norm": 1.5186265707015991,
-      "learning_rate": 8.800301507537689e-05,
-      "loss": 5.5175,
-      "step": 12440
-    },
-    {
-      "epoch": 6.488135593220339,
-      "grad_norm": 1.3876105546951294,
-      "learning_rate": 8.800201005025126e-05,
-      "loss": 5.6541,
-      "step": 12441
-    },
-    {
-      "epoch": 6.488657105606258,
-      "grad_norm": 1.6032865047454834,
-      "learning_rate": 8.800100502512563e-05,
-      "loss": 5.2484,
-      "step": 12442
-    },
-    {
-      "epoch": 6.489178617992177,
-      "grad_norm": 1.4521745443344116,
-      "learning_rate": 8.800000000000001e-05,
-      "loss": 5.792,
-      "step": 12443
-    },
-    {
-      "epoch": 6.489700130378097,
-      "grad_norm": 1.413445234298706,
-      "learning_rate": 8.799899497487438e-05,
-      "loss": 5.7826,
-      "step": 12444
-    },
-    {
-      "epoch": 6.490221642764015,
-      "grad_norm": 1.2929264307022095,
-      "learning_rate": 8.799798994974875e-05,
-      "loss": 5.7395,
-      "step": 12445
-    },
-    {
-      "epoch": 6.490743155149935,
-      "grad_norm": 1.49071204662323,
-      "learning_rate": 8.799698492462312e-05,
-      "loss": 5.5639,
-      "step": 12446
-    },
-    {
-      "epoch": 6.491264667535854,
-      "grad_norm": 1.4352720975875854,
-      "learning_rate": 8.79959798994975e-05,
-      "loss": 5.5656,
-      "step": 12447
-    },
-    {
-      "epoch": 6.491786179921773,
-      "grad_norm": 1.4231700897216797,
-      "learning_rate": 8.799497487437186e-05,
-      "loss": 5.4795,
-      "step": 12448
-    },
-    {
-      "epoch": 6.492307692307692,
-      "grad_norm": 1.3763982057571411,
-      "learning_rate": 8.799396984924624e-05,
-      "loss": 6.0767,
-      "step": 12449
-    },
-    {
-      "epoch": 6.492829204693612,
-      "grad_norm": 1.5284106731414795,
-      "learning_rate": 8.79929648241206e-05,
-      "loss": 5.6098,
-      "step": 12450
-    },
-    {
-      "epoch": 6.49335071707953,
-      "grad_norm": 1.3937326669692993,
-      "learning_rate": 8.799195979899498e-05,
-      "loss": 4.7173,
-      "step": 12451
-    },
-    {
-      "epoch": 6.49387222946545,
-      "grad_norm": 1.4928662776947021,
-      "learning_rate": 8.799095477386934e-05,
-      "loss": 5.4585,
-      "step": 12452
-    },
-    {
-      "epoch": 6.494393741851369,
-      "grad_norm": 1.3493150472640991,
-      "learning_rate": 8.798994974874372e-05,
-      "loss": 5.815,
-      "step": 12453
-    },
-    {
-      "epoch": 6.494915254237288,
-      "grad_norm": 1.5055917501449585,
-      "learning_rate": 8.79889447236181e-05,
-      "loss": 4.9744,
-      "step": 12454
-    },
-    {
-      "epoch": 6.495436766623207,
-      "grad_norm": 1.4732328653335571,
-      "learning_rate": 8.798793969849248e-05,
-      "loss": 5.3897,
-      "step": 12455
-    },
-    {
-      "epoch": 6.495958279009127,
-      "grad_norm": 1.3425999879837036,
-      "learning_rate": 8.798693467336684e-05,
-      "loss": 4.9154,
-      "step": 12456
-    },
-    {
-      "epoch": 6.496479791395045,
-      "grad_norm": 1.4671053886413574,
-      "learning_rate": 8.79859296482412e-05,
-      "loss": 5.6627,
-      "step": 12457
-    },
-    {
-      "epoch": 6.497001303780965,
-      "grad_norm": 1.441741943359375,
-      "learning_rate": 8.798492462311558e-05,
-      "loss": 5.1425,
-      "step": 12458
-    },
-    {
-      "epoch": 6.497522816166884,
-      "grad_norm": 1.4229652881622314,
-      "learning_rate": 8.798391959798995e-05,
-      "loss": 5.7178,
-      "step": 12459
-    },
-    {
-      "epoch": 6.498044328552803,
-      "grad_norm": 1.3382782936096191,
-      "learning_rate": 8.798291457286433e-05,
-      "loss": 5.4547,
-      "step": 12460
-    },
-    {
-      "epoch": 6.498565840938722,
-      "grad_norm": 1.5382015705108643,
-      "learning_rate": 8.798190954773869e-05,
-      "loss": 5.7715,
-      "step": 12461
-    },
-    {
-      "epoch": 6.499087353324642,
-      "grad_norm": 1.4722826480865479,
-      "learning_rate": 8.798090452261307e-05,
-      "loss": 5.7657,
-      "step": 12462
-    },
-    {
-      "epoch": 6.49960886571056,
-      "grad_norm": 1.4082194566726685,
-      "learning_rate": 8.797989949748745e-05,
-      "loss": 5.3909,
-      "step": 12463
-    },
-    {
-      "epoch": 6.50013037809648,
-      "grad_norm": 1.464550256729126,
-      "learning_rate": 8.797889447236182e-05,
-      "loss": 5.6014,
-      "step": 12464
-    },
-    {
-      "epoch": 6.500651890482399,
-      "grad_norm": 1.4342536926269531,
-      "learning_rate": 8.797788944723619e-05,
-      "loss": 5.5927,
-      "step": 12465
-    },
-    {
-      "epoch": 6.501173402868318,
-      "grad_norm": 1.4378613233566284,
-      "learning_rate": 8.797688442211057e-05,
-      "loss": 5.7741,
-      "step": 12466
-    },
-    {
-      "epoch": 6.501694915254237,
-      "grad_norm": 1.3294721841812134,
-      "learning_rate": 8.797587939698493e-05,
-      "loss": 5.7286,
-      "step": 12467
-    },
-    {
-      "epoch": 6.502216427640157,
-      "grad_norm": 1.5109871625900269,
-      "learning_rate": 8.797487437185931e-05,
-      "loss": 5.64,
-      "step": 12468
-    },
-    {
-      "epoch": 6.502737940026075,
-      "grad_norm": 1.5205719470977783,
-      "learning_rate": 8.797386934673367e-05,
-      "loss": 5.915,
-      "step": 12469
-    },
-    {
-      "epoch": 6.503259452411995,
-      "grad_norm": 1.4309018850326538,
-      "learning_rate": 8.797286432160804e-05,
-      "loss": 5.5132,
-      "step": 12470
-    },
-    {
-      "epoch": 6.5037809647979135,
-      "grad_norm": 1.332653522491455,
-      "learning_rate": 8.797185929648241e-05,
-      "loss": 5.7183,
-      "step": 12471
-    },
-    {
-      "epoch": 6.504302477183833,
-      "grad_norm": 1.5436627864837646,
-      "learning_rate": 8.797085427135678e-05,
-      "loss": 5.4086,
-      "step": 12472
-    },
-    {
-      "epoch": 6.504823989569752,
-      "grad_norm": 1.407082200050354,
-      "learning_rate": 8.796984924623116e-05,
-      "loss": 5.8656,
-      "step": 12473
-    },
-    {
-      "epoch": 6.505345501955672,
-      "grad_norm": 1.5265003442764282,
-      "learning_rate": 8.796884422110553e-05,
-      "loss": 5.5808,
-      "step": 12474
-    },
-    {
-      "epoch": 6.50586701434159,
-      "grad_norm": 1.3611451387405396,
-      "learning_rate": 8.796783919597991e-05,
-      "loss": 5.4537,
-      "step": 12475
-    },
-    {
-      "epoch": 6.50638852672751,
-      "grad_norm": 1.3606764078140259,
-      "learning_rate": 8.796683417085428e-05,
-      "loss": 5.798,
-      "step": 12476
-    },
-    {
-      "epoch": 6.5069100391134285,
-      "grad_norm": 1.628183364868164,
-      "learning_rate": 8.796582914572865e-05,
-      "loss": 5.0206,
-      "step": 12477
-    },
-    {
-      "epoch": 6.507431551499348,
-      "grad_norm": 1.4681569337844849,
-      "learning_rate": 8.796482412060302e-05,
-      "loss": 5.4871,
-      "step": 12478
-    },
-    {
-      "epoch": 6.507953063885267,
-      "grad_norm": 1.3806575536727905,
-      "learning_rate": 8.79638190954774e-05,
-      "loss": 5.4177,
-      "step": 12479
-    },
-    {
-      "epoch": 6.508474576271187,
-      "grad_norm": 1.4010251760482788,
-      "learning_rate": 8.796281407035176e-05,
-      "loss": 5.9606,
-      "step": 12480
-    },
-    {
-      "epoch": 6.508996088657105,
-      "grad_norm": 1.4544035196304321,
-      "learning_rate": 8.796180904522614e-05,
-      "loss": 5.6542,
-      "step": 12481
-    },
-    {
-      "epoch": 6.509517601043025,
-      "grad_norm": 1.456703543663025,
-      "learning_rate": 8.79608040201005e-05,
-      "loss": 5.5143,
-      "step": 12482
-    },
-    {
-      "epoch": 6.5100391134289435,
-      "grad_norm": 1.3905287981033325,
-      "learning_rate": 8.795979899497487e-05,
-      "loss": 5.9592,
-      "step": 12483
-    },
-    {
-      "epoch": 6.510560625814863,
-      "grad_norm": 1.462240219116211,
-      "learning_rate": 8.795879396984924e-05,
-      "loss": 5.8446,
-      "step": 12484
-    },
-    {
-      "epoch": 6.511082138200782,
-      "grad_norm": 1.3944820165634155,
-      "learning_rate": 8.795778894472362e-05,
-      "loss": 5.6792,
-      "step": 12485
-    },
-    {
-      "epoch": 6.511603650586702,
-      "grad_norm": 1.3809010982513428,
-      "learning_rate": 8.7956783919598e-05,
-      "loss": 5.407,
-      "step": 12486
-    },
-    {
-      "epoch": 6.5121251629726205,
-      "grad_norm": 1.3088277578353882,
-      "learning_rate": 8.795577889447236e-05,
-      "loss": 5.9453,
-      "step": 12487
-    },
-    {
-      "epoch": 6.51264667535854,
-      "grad_norm": 1.4226888418197632,
-      "learning_rate": 8.795477386934674e-05,
-      "loss": 5.186,
-      "step": 12488
-    },
-    {
-      "epoch": 6.5131681877444585,
-      "grad_norm": 1.3380358219146729,
-      "learning_rate": 8.79537688442211e-05,
-      "loss": 6.0949,
-      "step": 12489
-    },
-    {
-      "epoch": 6.513689700130378,
-      "grad_norm": 1.5355838537216187,
-      "learning_rate": 8.795276381909548e-05,
-      "loss": 5.2893,
-      "step": 12490
-    },
-    {
-      "epoch": 6.514211212516297,
-      "grad_norm": 1.3450136184692383,
-      "learning_rate": 8.795175879396985e-05,
-      "loss": 5.9804,
-      "step": 12491
-    },
-    {
-      "epoch": 6.514732724902217,
-      "grad_norm": 1.4839774370193481,
-      "learning_rate": 8.795075376884423e-05,
-      "loss": 5.6333,
-      "step": 12492
-    },
-    {
-      "epoch": 6.5152542372881355,
-      "grad_norm": 1.47777259349823,
-      "learning_rate": 8.794974874371859e-05,
-      "loss": 5.4413,
-      "step": 12493
-    },
-    {
-      "epoch": 6.515775749674055,
-      "grad_norm": 1.423406958580017,
-      "learning_rate": 8.794874371859297e-05,
-      "loss": 5.4427,
-      "step": 12494
-    },
-    {
-      "epoch": 6.5162972620599735,
-      "grad_norm": 1.4609050750732422,
-      "learning_rate": 8.794773869346735e-05,
-      "loss": 5.3153,
-      "step": 12495
-    },
-    {
-      "epoch": 6.516818774445893,
-      "grad_norm": 1.584741473197937,
-      "learning_rate": 8.794673366834171e-05,
-      "loss": 5.1374,
-      "step": 12496
-    },
-    {
-      "epoch": 6.517340286831812,
-      "grad_norm": 1.5071356296539307,
-      "learning_rate": 8.794572864321609e-05,
-      "loss": 5.4185,
-      "step": 12497
-    },
-    {
-      "epoch": 6.517861799217732,
-      "grad_norm": 1.4606349468231201,
-      "learning_rate": 8.794472361809045e-05,
-      "loss": 5.3414,
-      "step": 12498
-    },
-    {
-      "epoch": 6.5183833116036505,
-      "grad_norm": 1.4920042753219604,
-      "learning_rate": 8.794371859296483e-05,
-      "loss": 5.5956,
-      "step": 12499
-    },
-    {
-      "epoch": 6.51890482398957,
-      "grad_norm": 1.4514347314834595,
-      "learning_rate": 8.79427135678392e-05,
-      "loss": 5.4224,
-      "step": 12500
-    },
-    {
-      "epoch": 6.5194263363754885,
-      "grad_norm": 1.418641448020935,
-      "learning_rate": 8.794170854271357e-05,
-      "loss": 5.7454,
-      "step": 12501
-    },
-    {
-      "epoch": 6.519947848761408,
-      "grad_norm": 1.430914282798767,
-      "learning_rate": 8.794070351758794e-05,
-      "loss": 5.4475,
-      "step": 12502
-    },
-    {
-      "epoch": 6.5204693611473274,
-      "grad_norm": 1.6658300161361694,
-      "learning_rate": 8.793969849246232e-05,
-      "loss": 5.6307,
-      "step": 12503
-    },
-    {
-      "epoch": 6.520990873533247,
-      "grad_norm": 1.5116956233978271,
-      "learning_rate": 8.793869346733668e-05,
-      "loss": 4.508,
-      "step": 12504
-    },
-    {
-      "epoch": 6.5215123859191655,
-      "grad_norm": 1.659289002418518,
-      "learning_rate": 8.793768844221106e-05,
-      "loss": 4.9623,
-      "step": 12505
-    },
-    {
-      "epoch": 6.522033898305085,
-      "grad_norm": 1.4499802589416504,
-      "learning_rate": 8.793668341708543e-05,
-      "loss": 5.3696,
-      "step": 12506
-    },
-    {
-      "epoch": 6.5225554106910035,
-      "grad_norm": 1.4461456537246704,
-      "learning_rate": 8.793567839195981e-05,
-      "loss": 5.6694,
-      "step": 12507
-    },
-    {
-      "epoch": 6.523076923076923,
-      "grad_norm": 1.4375935792922974,
-      "learning_rate": 8.793467336683418e-05,
-      "loss": 5.476,
-      "step": 12508
-    },
-    {
-      "epoch": 6.5235984354628425,
-      "grad_norm": 1.4778969287872314,
-      "learning_rate": 8.793366834170854e-05,
-      "loss": 5.8047,
-      "step": 12509
-    },
-    {
-      "epoch": 6.524119947848761,
-      "grad_norm": 1.355550765991211,
-      "learning_rate": 8.793266331658292e-05,
-      "loss": 6.1013,
-      "step": 12510
-    },
-    {
-      "epoch": 6.5246414602346805,
-      "grad_norm": 1.4744062423706055,
-      "learning_rate": 8.793165829145728e-05,
-      "loss": 5.1813,
-      "step": 12511
-    },
-    {
-      "epoch": 6.5251629726206,
-      "grad_norm": 1.3988584280014038,
-      "learning_rate": 8.793065326633166e-05,
-      "loss": 5.7425,
-      "step": 12512
-    },
-    {
-      "epoch": 6.5256844850065185,
-      "grad_norm": 1.4191807508468628,
-      "learning_rate": 8.792964824120603e-05,
-      "loss": 5.6828,
-      "step": 12513
-    },
-    {
-      "epoch": 6.526205997392438,
-      "grad_norm": 1.398184061050415,
-      "learning_rate": 8.79286432160804e-05,
-      "loss": 5.3636,
-      "step": 12514
-    },
-    {
-      "epoch": 6.5267275097783575,
-      "grad_norm": 1.3990029096603394,
-      "learning_rate": 8.792763819095478e-05,
-      "loss": 5.2309,
-      "step": 12515
-    },
-    {
-      "epoch": 6.527249022164276,
-      "grad_norm": 1.3994126319885254,
-      "learning_rate": 8.792663316582916e-05,
-      "loss": 6.0042,
-      "step": 12516
-    },
-    {
-      "epoch": 6.5277705345501955,
-      "grad_norm": 1.4808038473129272,
-      "learning_rate": 8.792562814070352e-05,
-      "loss": 5.5677,
-      "step": 12517
-    },
-    {
-      "epoch": 6.528292046936115,
-      "grad_norm": 1.4258002042770386,
-      "learning_rate": 8.79246231155779e-05,
-      "loss": 5.7161,
-      "step": 12518
-    },
-    {
-      "epoch": 6.5288135593220336,
-      "grad_norm": 1.622502088546753,
-      "learning_rate": 8.792361809045227e-05,
-      "loss": 5.5452,
-      "step": 12519
-    },
-    {
-      "epoch": 6.529335071707953,
-      "grad_norm": 1.3774826526641846,
-      "learning_rate": 8.792261306532664e-05,
-      "loss": 5.5119,
-      "step": 12520
-    },
-    {
-      "epoch": 6.5298565840938725,
-      "grad_norm": 1.5008872747421265,
-      "learning_rate": 8.792160804020101e-05,
-      "loss": 5.44,
-      "step": 12521
-    },
-    {
-      "epoch": 6.530378096479791,
-      "grad_norm": 1.4999505281448364,
-      "learning_rate": 8.792060301507537e-05,
-      "loss": 5.6408,
-      "step": 12522
-    },
-    {
-      "epoch": 6.5308996088657105,
-      "grad_norm": 1.329841136932373,
-      "learning_rate": 8.791959798994975e-05,
-      "loss": 5.9098,
-      "step": 12523
-    },
-    {
-      "epoch": 6.53142112125163,
-      "grad_norm": 1.4376024007797241,
-      "learning_rate": 8.791859296482411e-05,
-      "loss": 5.5938,
-      "step": 12524
-    },
-    {
-      "epoch": 6.531942633637549,
-      "grad_norm": 1.4345446825027466,
-      "learning_rate": 8.791758793969849e-05,
-      "loss": 5.197,
-      "step": 12525
-    },
-    {
-      "epoch": 6.532464146023468,
-      "grad_norm": 1.3469228744506836,
-      "learning_rate": 8.791658291457287e-05,
-      "loss": 5.9648,
-      "step": 12526
-    },
-    {
-      "epoch": 6.5329856584093875,
-      "grad_norm": 1.4169988632202148,
-      "learning_rate": 8.791557788944725e-05,
-      "loss": 5.8552,
-      "step": 12527
-    },
-    {
-      "epoch": 6.533507170795306,
-      "grad_norm": 1.3211077451705933,
-      "learning_rate": 8.791457286432161e-05,
-      "loss": 5.7412,
-      "step": 12528
-    },
-    {
-      "epoch": 6.5340286831812255,
-      "grad_norm": 1.3898082971572876,
-      "learning_rate": 8.791356783919599e-05,
-      "loss": 5.2091,
-      "step": 12529
-    },
-    {
-      "epoch": 6.534550195567145,
-      "grad_norm": 1.45436429977417,
-      "learning_rate": 8.791256281407035e-05,
-      "loss": 5.5723,
-      "step": 12530
-    },
-    {
-      "epoch": 6.535071707953064,
-      "grad_norm": 1.3811384439468384,
-      "learning_rate": 8.791155778894473e-05,
-      "loss": 5.8656,
-      "step": 12531
-    },
-    {
-      "epoch": 6.535593220338983,
-      "grad_norm": 1.4799333810806274,
-      "learning_rate": 8.79105527638191e-05,
-      "loss": 5.5559,
-      "step": 12532
-    },
-    {
-      "epoch": 6.5361147327249025,
-      "grad_norm": 1.3495765924453735,
-      "learning_rate": 8.790954773869347e-05,
-      "loss": 5.614,
-      "step": 12533
-    },
-    {
-      "epoch": 6.536636245110821,
-      "grad_norm": 1.653756856918335,
-      "learning_rate": 8.790854271356784e-05,
-      "loss": 5.2663,
-      "step": 12534
-    },
-    {
-      "epoch": 6.5371577574967406,
-      "grad_norm": 1.469201683998108,
-      "learning_rate": 8.790753768844222e-05,
-      "loss": 5.5156,
-      "step": 12535
-    },
-    {
-      "epoch": 6.53767926988266,
-      "grad_norm": 1.4156748056411743,
-      "learning_rate": 8.79065326633166e-05,
-      "loss": 5.8851,
-      "step": 12536
-    },
-    {
-      "epoch": 6.538200782268579,
-      "grad_norm": 1.4300384521484375,
-      "learning_rate": 8.790552763819096e-05,
-      "loss": 5.5242,
-      "step": 12537
-    },
-    {
-      "epoch": 6.538722294654498,
-      "grad_norm": 1.4932236671447754,
-      "learning_rate": 8.790452261306534e-05,
-      "loss": 4.9289,
-      "step": 12538
-    },
-    {
-      "epoch": 6.5392438070404175,
-      "grad_norm": 1.4619979858398438,
-      "learning_rate": 8.79035175879397e-05,
-      "loss": 5.6292,
-      "step": 12539
-    },
-    {
-      "epoch": 6.539765319426336,
-      "grad_norm": 1.422136902809143,
-      "learning_rate": 8.790251256281408e-05,
-      "loss": 5.8012,
-      "step": 12540
-    },
-    {
-      "epoch": 6.540286831812256,
-      "grad_norm": 1.4003074169158936,
-      "learning_rate": 8.790150753768844e-05,
-      "loss": 5.213,
-      "step": 12541
-    },
-    {
-      "epoch": 6.540808344198175,
-      "grad_norm": 1.4106879234313965,
-      "learning_rate": 8.790050251256282e-05,
-      "loss": 5.5448,
-      "step": 12542
-    },
-    {
-      "epoch": 6.541329856584094,
-      "grad_norm": 1.482836127281189,
-      "learning_rate": 8.789949748743718e-05,
-      "loss": 5.465,
-      "step": 12543
-    },
-    {
-      "epoch": 6.541851368970013,
-      "grad_norm": 1.408078908920288,
-      "learning_rate": 8.789849246231156e-05,
-      "loss": 5.7231,
-      "step": 12544
-    },
-    {
-      "epoch": 6.5423728813559325,
-      "grad_norm": 1.3140257596969604,
-      "learning_rate": 8.789748743718593e-05,
-      "loss": 5.6402,
-      "step": 12545
-    },
-    {
-      "epoch": 6.542894393741851,
-      "grad_norm": 1.4012560844421387,
-      "learning_rate": 8.78964824120603e-05,
-      "loss": 5.7401,
-      "step": 12546
-    },
-    {
-      "epoch": 6.543415906127771,
-      "grad_norm": 1.5670433044433594,
-      "learning_rate": 8.789547738693468e-05,
-      "loss": 5.1313,
-      "step": 12547
-    },
-    {
-      "epoch": 6.54393741851369,
-      "grad_norm": 1.430067539215088,
-      "learning_rate": 8.789447236180906e-05,
-      "loss": 5.6382,
-      "step": 12548
-    },
-    {
-      "epoch": 6.544458930899609,
-      "grad_norm": 1.3594905138015747,
-      "learning_rate": 8.789346733668342e-05,
-      "loss": 5.7503,
-      "step": 12549
-    },
-    {
-      "epoch": 6.544980443285528,
-      "grad_norm": 1.4518684148788452,
-      "learning_rate": 8.789246231155779e-05,
-      "loss": 5.649,
-      "step": 12550
-    },
-    {
-      "epoch": 6.5455019556714475,
-      "grad_norm": 1.4472609758377075,
-      "learning_rate": 8.789145728643217e-05,
-      "loss": 5.1447,
-      "step": 12551
-    },
-    {
-      "epoch": 6.546023468057366,
-      "grad_norm": 1.4639800786972046,
-      "learning_rate": 8.789045226130653e-05,
-      "loss": 5.756,
-      "step": 12552
-    },
-    {
-      "epoch": 6.546544980443286,
-      "grad_norm": 1.3664597272872925,
-      "learning_rate": 8.788944723618091e-05,
-      "loss": 5.9137,
-      "step": 12553
-    },
-    {
-      "epoch": 6.547066492829205,
-      "grad_norm": 1.4685629606246948,
-      "learning_rate": 8.788844221105527e-05,
-      "loss": 5.2428,
-      "step": 12554
-    },
-    {
-      "epoch": 6.547588005215124,
-      "grad_norm": 1.3590062856674194,
-      "learning_rate": 8.788743718592965e-05,
-      "loss": 5.8129,
-      "step": 12555
-    },
-    {
-      "epoch": 6.548109517601043,
-      "grad_norm": 1.6284739971160889,
-      "learning_rate": 8.788643216080403e-05,
-      "loss": 4.9765,
-      "step": 12556
-    },
-    {
-      "epoch": 6.548631029986963,
-      "grad_norm": 1.492897391319275,
-      "learning_rate": 8.78854271356784e-05,
-      "loss": 5.2583,
-      "step": 12557
-    },
-    {
-      "epoch": 6.549152542372881,
-      "grad_norm": 1.4101232290267944,
-      "learning_rate": 8.788442211055277e-05,
-      "loss": 5.6307,
-      "step": 12558
-    },
-    {
-      "epoch": 6.549674054758801,
-      "grad_norm": 1.3859524726867676,
-      "learning_rate": 8.788341708542715e-05,
-      "loss": 6.0056,
-      "step": 12559
-    },
-    {
-      "epoch": 6.55019556714472,
-      "grad_norm": 1.4319469928741455,
-      "learning_rate": 8.788241206030151e-05,
-      "loss": 5.6361,
-      "step": 12560
-    },
-    {
-      "epoch": 6.550717079530639,
-      "grad_norm": 1.3612784147262573,
-      "learning_rate": 8.788140703517589e-05,
-      "loss": 5.847,
-      "step": 12561
-    },
-    {
-      "epoch": 6.551238591916558,
-      "grad_norm": 1.5109992027282715,
-      "learning_rate": 8.788040201005025e-05,
-      "loss": 5.6937,
-      "step": 12562
-    },
-    {
-      "epoch": 6.551760104302478,
-      "grad_norm": 1.5105756521224976,
-      "learning_rate": 8.787939698492462e-05,
-      "loss": 5.1252,
-      "step": 12563
-    },
-    {
-      "epoch": 6.552281616688396,
-      "grad_norm": 1.5078566074371338,
-      "learning_rate": 8.7878391959799e-05,
-      "loss": 5.7021,
-      "step": 12564
-    },
-    {
-      "epoch": 6.552803129074316,
-      "grad_norm": 1.373811960220337,
-      "learning_rate": 8.787738693467336e-05,
-      "loss": 5.5237,
-      "step": 12565
-    },
-    {
-      "epoch": 6.553324641460234,
-      "grad_norm": 1.5315057039260864,
-      "learning_rate": 8.787638190954774e-05,
-      "loss": 4.8549,
-      "step": 12566
-    },
-    {
-      "epoch": 6.553846153846154,
-      "grad_norm": 1.3391033411026,
-      "learning_rate": 8.787537688442212e-05,
-      "loss": 5.328,
-      "step": 12567
-    },
-    {
-      "epoch": 6.554367666232073,
-      "grad_norm": 1.4832059144973755,
-      "learning_rate": 8.78743718592965e-05,
-      "loss": 5.5747,
-      "step": 12568
-    },
-    {
-      "epoch": 6.554889178617993,
-      "grad_norm": 1.3858699798583984,
-      "learning_rate": 8.787336683417086e-05,
-      "loss": 5.9456,
-      "step": 12569
-    },
-    {
-      "epoch": 6.555410691003911,
-      "grad_norm": 1.2609801292419434,
-      "learning_rate": 8.787236180904524e-05,
-      "loss": 5.4284,
-      "step": 12570
-    },
-    {
-      "epoch": 6.555932203389831,
-      "grad_norm": 1.4727749824523926,
-      "learning_rate": 8.78713567839196e-05,
-      "loss": 5.7881,
-      "step": 12571
-    },
-    {
-      "epoch": 6.556453715775749,
-      "grad_norm": 1.4419755935668945,
-      "learning_rate": 8.787035175879398e-05,
-      "loss": 5.5763,
-      "step": 12572
-    },
-    {
-      "epoch": 6.556975228161669,
-      "grad_norm": 1.359143614768982,
-      "learning_rate": 8.786934673366834e-05,
-      "loss": 5.753,
-      "step": 12573
-    },
-    {
-      "epoch": 6.557496740547588,
-      "grad_norm": 1.4163727760314941,
-      "learning_rate": 8.786834170854272e-05,
-      "loss": 6.0008,
-      "step": 12574
-    },
-    {
-      "epoch": 6.558018252933508,
-      "grad_norm": 1.5440621376037598,
-      "learning_rate": 8.786733668341708e-05,
-      "loss": 5.464,
-      "step": 12575
-    },
-    {
-      "epoch": 6.558539765319426,
-      "grad_norm": 1.50942862033844,
-      "learning_rate": 8.786633165829146e-05,
-      "loss": 5.3041,
-      "step": 12576
-    },
-    {
-      "epoch": 6.559061277705346,
-      "grad_norm": 1.4390968084335327,
-      "learning_rate": 8.786532663316584e-05,
-      "loss": 5.7568,
-      "step": 12577
-    },
-    {
-      "epoch": 6.559582790091264,
-      "grad_norm": 1.6266926527023315,
-      "learning_rate": 8.78643216080402e-05,
-      "loss": 5.3408,
-      "step": 12578
-    },
-    {
-      "epoch": 6.560104302477184,
-      "grad_norm": 1.3825894594192505,
-      "learning_rate": 8.786331658291458e-05,
-      "loss": 5.6441,
-      "step": 12579
-    },
-    {
-      "epoch": 6.560625814863103,
-      "grad_norm": 1.4531939029693604,
-      "learning_rate": 8.786231155778895e-05,
-      "loss": 5.1761,
-      "step": 12580
-    },
-    {
-      "epoch": 6.561147327249023,
-      "grad_norm": 1.4138857126235962,
-      "learning_rate": 8.786130653266332e-05,
-      "loss": 5.2873,
-      "step": 12581
-    },
-    {
-      "epoch": 6.561668839634941,
-      "grad_norm": 1.2894039154052734,
-      "learning_rate": 8.786030150753769e-05,
-      "loss": 5.8891,
-      "step": 12582
-    },
-    {
-      "epoch": 6.562190352020861,
-      "grad_norm": 1.5632343292236328,
-      "learning_rate": 8.785929648241207e-05,
-      "loss": 5.3088,
-      "step": 12583
-    },
-    {
-      "epoch": 6.562711864406779,
-      "grad_norm": 1.4690945148468018,
-      "learning_rate": 8.785829145728643e-05,
-      "loss": 5.4137,
-      "step": 12584
-    },
-    {
-      "epoch": 6.563233376792699,
-      "grad_norm": 1.4702199697494507,
-      "learning_rate": 8.785728643216081e-05,
-      "loss": 5.2498,
-      "step": 12585
-    },
-    {
-      "epoch": 6.563754889178618,
-      "grad_norm": 1.4037480354309082,
-      "learning_rate": 8.785628140703517e-05,
-      "loss": 5.5399,
-      "step": 12586
-    },
-    {
-      "epoch": 6.564276401564538,
-      "grad_norm": 1.5832880735397339,
-      "learning_rate": 8.785527638190955e-05,
-      "loss": 5.3659,
-      "step": 12587
-    },
-    {
-      "epoch": 6.564797913950456,
-      "grad_norm": 1.5006401538848877,
-      "learning_rate": 8.785427135678393e-05,
-      "loss": 5.6755,
-      "step": 12588
-    },
-    {
-      "epoch": 6.565319426336376,
-      "grad_norm": 1.4322131872177124,
-      "learning_rate": 8.78532663316583e-05,
-      "loss": 5.7679,
-      "step": 12589
-    },
-    {
-      "epoch": 6.565840938722294,
-      "grad_norm": 1.420424222946167,
-      "learning_rate": 8.785226130653267e-05,
-      "loss": 5.5745,
-      "step": 12590
-    },
-    {
-      "epoch": 6.566362451108214,
-      "grad_norm": 1.459281086921692,
-      "learning_rate": 8.785125628140704e-05,
-      "loss": 5.7776,
-      "step": 12591
-    },
-    {
-      "epoch": 6.566883963494133,
-      "grad_norm": 1.4524139165878296,
-      "learning_rate": 8.785025125628141e-05,
-      "loss": 5.732,
-      "step": 12592
-    },
-    {
-      "epoch": 6.567405475880053,
-      "grad_norm": 1.4340914487838745,
-      "learning_rate": 8.784924623115578e-05,
-      "loss": 5.5924,
-      "step": 12593
-    },
-    {
-      "epoch": 6.567926988265971,
-      "grad_norm": 1.532705545425415,
-      "learning_rate": 8.784824120603016e-05,
-      "loss": 5.1333,
-      "step": 12594
-    },
-    {
-      "epoch": 6.568448500651891,
-      "grad_norm": 1.683125615119934,
-      "learning_rate": 8.784723618090452e-05,
-      "loss": 5.4298,
-      "step": 12595
-    },
-    {
-      "epoch": 6.568970013037809,
-      "grad_norm": 1.6345151662826538,
-      "learning_rate": 8.78462311557789e-05,
-      "loss": 5.5559,
-      "step": 12596
-    },
-    {
-      "epoch": 6.569491525423729,
-      "grad_norm": 1.471537709236145,
-      "learning_rate": 8.784522613065328e-05,
-      "loss": 5.5485,
-      "step": 12597
-    },
-    {
-      "epoch": 6.570013037809648,
-      "grad_norm": 1.5202357769012451,
-      "learning_rate": 8.784422110552765e-05,
-      "loss": 5.4631,
-      "step": 12598
-    },
-    {
-      "epoch": 6.570534550195567,
-      "grad_norm": 1.4303057193756104,
-      "learning_rate": 8.784321608040202e-05,
-      "loss": 5.65,
-      "step": 12599
-    },
-    {
-      "epoch": 6.571056062581486,
-      "grad_norm": 1.4635146856307983,
-      "learning_rate": 8.78422110552764e-05,
-      "loss": 5.6473,
-      "step": 12600
-    },
-    {
-      "epoch": 6.571577574967406,
-      "grad_norm": 1.347988486289978,
-      "learning_rate": 8.784120603015076e-05,
-      "loss": 5.7522,
-      "step": 12601
-    },
-    {
-      "epoch": 6.572099087353324,
-      "grad_norm": 1.3689754009246826,
-      "learning_rate": 8.784020100502512e-05,
-      "loss": 5.4076,
-      "step": 12602
-    },
-    {
-      "epoch": 6.572620599739244,
-      "grad_norm": 1.4595489501953125,
-      "learning_rate": 8.78391959798995e-05,
-      "loss": 4.8959,
-      "step": 12603
-    },
-    {
-      "epoch": 6.573142112125163,
-      "grad_norm": 1.5766183137893677,
-      "learning_rate": 8.783819095477387e-05,
-      "loss": 5.0713,
-      "step": 12604
-    },
-    {
-      "epoch": 6.573663624511082,
-      "grad_norm": 1.424189805984497,
-      "learning_rate": 8.783718592964824e-05,
-      "loss": 5.6537,
-      "step": 12605
-    },
-    {
-      "epoch": 6.574185136897001,
-      "grad_norm": 1.4660588502883911,
-      "learning_rate": 8.783618090452261e-05,
-      "loss": 5.6814,
-      "step": 12606
-    },
-    {
-      "epoch": 6.574706649282921,
-      "grad_norm": 1.4175626039505005,
-      "learning_rate": 8.783517587939699e-05,
-      "loss": 5.4575,
-      "step": 12607
-    },
-    {
-      "epoch": 6.575228161668839,
-      "grad_norm": 1.4180727005004883,
-      "learning_rate": 8.783417085427136e-05,
-      "loss": 5.694,
-      "step": 12608
-    },
-    {
-      "epoch": 6.575749674054759,
-      "grad_norm": 1.3509196043014526,
-      "learning_rate": 8.783316582914574e-05,
-      "loss": 5.9463,
-      "step": 12609
-    },
-    {
-      "epoch": 6.576271186440678,
-      "grad_norm": 1.5047180652618408,
-      "learning_rate": 8.78321608040201e-05,
-      "loss": 5.7021,
-      "step": 12610
-    },
-    {
-      "epoch": 6.576792698826597,
-      "grad_norm": 1.4767088890075684,
-      "learning_rate": 8.783115577889448e-05,
-      "loss": 5.4066,
-      "step": 12611
-    },
-    {
-      "epoch": 6.577314211212516,
-      "grad_norm": 1.562290906906128,
-      "learning_rate": 8.783015075376885e-05,
-      "loss": 5.211,
-      "step": 12612
-    },
-    {
-      "epoch": 6.577835723598436,
-      "grad_norm": 1.3040703535079956,
-      "learning_rate": 8.782914572864323e-05,
-      "loss": 5.796,
-      "step": 12613
-    },
-    {
-      "epoch": 6.578357235984354,
-      "grad_norm": 1.5241940021514893,
-      "learning_rate": 8.782814070351759e-05,
-      "loss": 5.2323,
-      "step": 12614
-    },
-    {
-      "epoch": 6.578878748370274,
-      "grad_norm": 1.440870761871338,
-      "learning_rate": 8.782713567839195e-05,
-      "loss": 5.4726,
-      "step": 12615
-    },
-    {
-      "epoch": 6.579400260756193,
-      "grad_norm": 1.5338983535766602,
-      "learning_rate": 8.782613065326633e-05,
-      "loss": 5.529,
-      "step": 12616
-    },
-    {
-      "epoch": 6.579921773142112,
-      "grad_norm": 1.4736131429672241,
-      "learning_rate": 8.782512562814071e-05,
-      "loss": 4.5135,
-      "step": 12617
-    },
-    {
-      "epoch": 6.580443285528031,
-      "grad_norm": 1.5682207345962524,
-      "learning_rate": 8.782412060301509e-05,
-      "loss": 5.0513,
-      "step": 12618
-    },
-    {
-      "epoch": 6.580964797913951,
-      "grad_norm": 1.7090058326721191,
-      "learning_rate": 8.782311557788945e-05,
-      "loss": 5.0464,
-      "step": 12619
-    },
-    {
-      "epoch": 6.581486310299869,
-      "grad_norm": 1.4685333967208862,
-      "learning_rate": 8.782211055276383e-05,
-      "loss": 5.3914,
-      "step": 12620
-    },
-    {
-      "epoch": 6.582007822685789,
-      "grad_norm": 1.3985859155654907,
-      "learning_rate": 8.78211055276382e-05,
-      "loss": 5.8129,
-      "step": 12621
-    },
-    {
-      "epoch": 6.582529335071708,
-      "grad_norm": 1.6404573917388916,
-      "learning_rate": 8.782010050251257e-05,
-      "loss": 5.424,
-      "step": 12622
-    },
-    {
-      "epoch": 6.583050847457627,
-      "grad_norm": 1.3094804286956787,
-      "learning_rate": 8.781909547738694e-05,
-      "loss": 5.7867,
-      "step": 12623
-    },
-    {
-      "epoch": 6.583572359843546,
-      "grad_norm": 1.485745906829834,
-      "learning_rate": 8.781809045226131e-05,
-      "loss": 5.875,
-      "step": 12624
-    },
-    {
-      "epoch": 6.584093872229466,
-      "grad_norm": 1.517445683479309,
-      "learning_rate": 8.781708542713568e-05,
-      "loss": 5.7562,
-      "step": 12625
-    },
-    {
-      "epoch": 6.584615384615384,
-      "grad_norm": 1.4216309785842896,
-      "learning_rate": 8.781608040201006e-05,
-      "loss": 5.4867,
-      "step": 12626
-    },
-    {
-      "epoch": 6.585136897001304,
-      "grad_norm": 1.3893240690231323,
-      "learning_rate": 8.781507537688442e-05,
-      "loss": 5.606,
-      "step": 12627
-    },
-    {
-      "epoch": 6.585658409387223,
-      "grad_norm": 1.4484920501708984,
-      "learning_rate": 8.78140703517588e-05,
-      "loss": 5.6204,
-      "step": 12628
-    },
-    {
-      "epoch": 6.586179921773142,
-      "grad_norm": 1.5166114568710327,
-      "learning_rate": 8.781306532663318e-05,
-      "loss": 5.7967,
-      "step": 12629
-    },
-    {
-      "epoch": 6.586701434159061,
-      "grad_norm": 1.3741916418075562,
-      "learning_rate": 8.781206030150754e-05,
-      "loss": 5.8036,
-      "step": 12630
-    },
-    {
-      "epoch": 6.587222946544981,
-      "grad_norm": 1.2892519235610962,
-      "learning_rate": 8.781105527638192e-05,
-      "loss": 6.0055,
-      "step": 12631
-    },
-    {
-      "epoch": 6.587744458930899,
-      "grad_norm": 1.5957785844802856,
-      "learning_rate": 8.781005025125628e-05,
-      "loss": 5.3736,
-      "step": 12632
-    },
-    {
-      "epoch": 6.588265971316819,
-      "grad_norm": 1.5308053493499756,
-      "learning_rate": 8.780904522613066e-05,
-      "loss": 5.2508,
-      "step": 12633
-    },
-    {
-      "epoch": 6.588787483702738,
-      "grad_norm": 1.4613311290740967,
-      "learning_rate": 8.780804020100502e-05,
-      "loss": 5.8208,
-      "step": 12634
-    },
-    {
-      "epoch": 6.589308996088657,
-      "grad_norm": 1.397038221359253,
-      "learning_rate": 8.78070351758794e-05,
-      "loss": 5.4577,
-      "step": 12635
-    },
-    {
-      "epoch": 6.589830508474576,
-      "grad_norm": 1.3209458589553833,
-      "learning_rate": 8.780603015075377e-05,
-      "loss": 5.9181,
-      "step": 12636
-    },
-    {
-      "epoch": 6.590352020860496,
-      "grad_norm": 1.50836181640625,
-      "learning_rate": 8.780502512562814e-05,
-      "loss": 5.425,
-      "step": 12637
-    },
-    {
-      "epoch": 6.590873533246414,
-      "grad_norm": 1.4640510082244873,
-      "learning_rate": 8.780402010050251e-05,
-      "loss": 5.7029,
-      "step": 12638
-    },
-    {
-      "epoch": 6.591395045632334,
-      "grad_norm": 1.4712388515472412,
-      "learning_rate": 8.780301507537689e-05,
-      "loss": 5.7552,
-      "step": 12639
-    },
-    {
-      "epoch": 6.591916558018253,
-      "grad_norm": 1.5009455680847168,
-      "learning_rate": 8.780201005025126e-05,
-      "loss": 5.4579,
-      "step": 12640
-    },
-    {
-      "epoch": 6.592438070404172,
-      "grad_norm": 1.5606987476348877,
-      "learning_rate": 8.780100502512564e-05,
-      "loss": 5.7232,
-      "step": 12641
-    },
-    {
-      "epoch": 6.592959582790091,
-      "grad_norm": 1.4215489625930786,
-      "learning_rate": 8.78e-05,
-      "loss": 5.6592,
-      "step": 12642
-    },
-    {
-      "epoch": 6.593481095176011,
-      "grad_norm": 1.6158361434936523,
-      "learning_rate": 8.779899497487437e-05,
-      "loss": 5.1185,
-      "step": 12643
-    },
-    {
-      "epoch": 6.594002607561929,
-      "grad_norm": 1.4108761548995972,
-      "learning_rate": 8.779798994974875e-05,
-      "loss": 5.3929,
-      "step": 12644
-    },
-    {
-      "epoch": 6.594524119947849,
-      "grad_norm": 1.5046839714050293,
-      "learning_rate": 8.779698492462311e-05,
-      "loss": 5.5334,
-      "step": 12645
-    },
-    {
-      "epoch": 6.595045632333768,
-      "grad_norm": 1.4471229314804077,
-      "learning_rate": 8.779597989949749e-05,
-      "loss": 5.7661,
-      "step": 12646
-    },
-    {
-      "epoch": 6.595567144719687,
-      "grad_norm": 1.4066576957702637,
-      "learning_rate": 8.779497487437185e-05,
-      "loss": 5.5804,
-      "step": 12647
-    },
-    {
-      "epoch": 6.596088657105606,
-      "grad_norm": 1.4027494192123413,
-      "learning_rate": 8.779396984924623e-05,
-      "loss": 4.9957,
-      "step": 12648
-    },
-    {
-      "epoch": 6.596610169491526,
-      "grad_norm": 1.4992258548736572,
-      "learning_rate": 8.779296482412061e-05,
-      "loss": 5.2928,
-      "step": 12649
-    },
-    {
-      "epoch": 6.597131681877444,
-      "grad_norm": 1.5492808818817139,
-      "learning_rate": 8.779195979899499e-05,
-      "loss": 5.2913,
-      "step": 12650
-    },
-    {
-      "epoch": 6.597653194263364,
-      "grad_norm": 1.4365154504776,
-      "learning_rate": 8.779095477386935e-05,
-      "loss": 5.6598,
-      "step": 12651
-    },
-    {
-      "epoch": 6.598174706649283,
-      "grad_norm": 1.3569459915161133,
-      "learning_rate": 8.778994974874373e-05,
-      "loss": 5.8414,
-      "step": 12652
-    },
-    {
-      "epoch": 6.598696219035202,
-      "grad_norm": 1.3475803136825562,
-      "learning_rate": 8.77889447236181e-05,
-      "loss": 5.5124,
-      "step": 12653
-    },
-    {
-      "epoch": 6.599217731421121,
-      "grad_norm": 1.4482766389846802,
-      "learning_rate": 8.778793969849247e-05,
-      "loss": 5.0951,
-      "step": 12654
-    },
-    {
-      "epoch": 6.599739243807041,
-      "grad_norm": 1.4267802238464355,
-      "learning_rate": 8.778693467336684e-05,
-      "loss": 5.8302,
-      "step": 12655
-    },
-    {
-      "epoch": 6.600260756192959,
-      "grad_norm": 1.3828619718551636,
-      "learning_rate": 8.77859296482412e-05,
-      "loss": 5.7241,
-      "step": 12656
-    },
-    {
-      "epoch": 6.600782268578879,
-      "grad_norm": 1.3185272216796875,
-      "learning_rate": 8.778492462311558e-05,
-      "loss": 5.7261,
-      "step": 12657
-    },
-    {
-      "epoch": 6.601303780964798,
-      "grad_norm": 1.3820860385894775,
-      "learning_rate": 8.778391959798994e-05,
-      "loss": 5.8836,
-      "step": 12658
-    },
-    {
-      "epoch": 6.601825293350717,
-      "grad_norm": 1.5078847408294678,
-      "learning_rate": 8.778291457286432e-05,
-      "loss": 5.8358,
-      "step": 12659
-    },
-    {
-      "epoch": 6.602346805736636,
-      "grad_norm": 1.4330273866653442,
-      "learning_rate": 8.77819095477387e-05,
-      "loss": 5.6504,
-      "step": 12660
-    },
-    {
-      "epoch": 6.602868318122555,
-      "grad_norm": 1.4467077255249023,
-      "learning_rate": 8.778090452261308e-05,
-      "loss": 5.7501,
-      "step": 12661
-    },
-    {
-      "epoch": 6.603389830508474,
-      "grad_norm": 1.5018937587738037,
-      "learning_rate": 8.777989949748744e-05,
-      "loss": 5.6609,
-      "step": 12662
-    },
-    {
-      "epoch": 6.603911342894394,
-      "grad_norm": 2.646347999572754,
-      "learning_rate": 8.777889447236182e-05,
-      "loss": 5.5618,
-      "step": 12663
-    },
-    {
-      "epoch": 6.604432855280313,
-      "grad_norm": 1.4570772647857666,
-      "learning_rate": 8.777788944723618e-05,
-      "loss": 5.7391,
-      "step": 12664
-    },
-    {
-      "epoch": 6.604954367666232,
-      "grad_norm": 1.4348684549331665,
-      "learning_rate": 8.777688442211056e-05,
-      "loss": 5.9916,
-      "step": 12665
-    },
-    {
-      "epoch": 6.605475880052151,
-      "grad_norm": 2.033144235610962,
-      "learning_rate": 8.777587939698493e-05,
-      "loss": 4.8194,
-      "step": 12666
-    },
-    {
-      "epoch": 6.60599739243807,
-      "grad_norm": 1.4769818782806396,
-      "learning_rate": 8.77748743718593e-05,
-      "loss": 5.572,
-      "step": 12667
-    },
-    {
-      "epoch": 6.606518904823989,
-      "grad_norm": 1.595062017440796,
-      "learning_rate": 8.777386934673367e-05,
-      "loss": 5.3813,
-      "step": 12668
-    },
-    {
-      "epoch": 6.607040417209909,
-      "grad_norm": 1.3825657367706299,
-      "learning_rate": 8.777286432160805e-05,
-      "loss": 5.9431,
-      "step": 12669
-    },
-    {
-      "epoch": 6.607561929595828,
-      "grad_norm": 1.4021836519241333,
-      "learning_rate": 8.777185929648242e-05,
-      "loss": 5.6312,
-      "step": 12670
-    },
-    {
-      "epoch": 6.608083441981747,
-      "grad_norm": 1.646450161933899,
-      "learning_rate": 8.777085427135679e-05,
-      "loss": 5.071,
-      "step": 12671
-    },
-    {
-      "epoch": 6.608604954367666,
-      "grad_norm": 1.6757258176803589,
-      "learning_rate": 8.776984924623117e-05,
-      "loss": 5.3128,
-      "step": 12672
-    },
-    {
-      "epoch": 6.609126466753585,
-      "grad_norm": 1.4847173690795898,
-      "learning_rate": 8.776884422110553e-05,
-      "loss": 5.8981,
-      "step": 12673
-    },
-    {
-      "epoch": 6.609647979139504,
-      "grad_norm": 1.5196614265441895,
-      "learning_rate": 8.776783919597991e-05,
-      "loss": 5.683,
-      "step": 12674
-    },
-    {
-      "epoch": 6.610169491525424,
-      "grad_norm": 1.5948742628097534,
-      "learning_rate": 8.776683417085427e-05,
-      "loss": 5.4589,
-      "step": 12675
-    },
-    {
-      "epoch": 6.610691003911343,
-      "grad_norm": 1.5615875720977783,
-      "learning_rate": 8.776582914572865e-05,
-      "loss": 5.8068,
-      "step": 12676
-    },
-    {
-      "epoch": 6.611212516297262,
-      "grad_norm": 1.4867985248565674,
-      "learning_rate": 8.776482412060301e-05,
-      "loss": 5.5718,
-      "step": 12677
-    },
-    {
-      "epoch": 6.611734028683181,
-      "grad_norm": 1.6656172275543213,
-      "learning_rate": 8.776381909547739e-05,
-      "loss": 5.5089,
-      "step": 12678
-    },
-    {
-      "epoch": 6.6122555410691,
-      "grad_norm": 1.5028811693191528,
-      "learning_rate": 8.776281407035176e-05,
-      "loss": 5.55,
-      "step": 12679
-    },
-    {
-      "epoch": 6.612777053455019,
-      "grad_norm": 1.4071787595748901,
-      "learning_rate": 8.776180904522613e-05,
-      "loss": 5.9175,
-      "step": 12680
-    },
-    {
-      "epoch": 6.613298565840939,
-      "grad_norm": 1.4265327453613281,
-      "learning_rate": 8.776080402010051e-05,
-      "loss": 5.2983,
-      "step": 12681
-    },
-    {
-      "epoch": 6.613820078226858,
-      "grad_norm": 1.4804184436798096,
-      "learning_rate": 8.775979899497488e-05,
-      "loss": 5.441,
-      "step": 12682
-    },
-    {
-      "epoch": 6.614341590612777,
-      "grad_norm": 1.4344085454940796,
-      "learning_rate": 8.775879396984925e-05,
-      "loss": 5.4587,
-      "step": 12683
-    },
-    {
-      "epoch": 6.614863102998696,
-      "grad_norm": 1.426405906677246,
-      "learning_rate": 8.775778894472362e-05,
-      "loss": 5.029,
-      "step": 12684
-    },
-    {
-      "epoch": 6.615384615384615,
-      "grad_norm": 1.4278090000152588,
-      "learning_rate": 8.7756783919598e-05,
-      "loss": 5.7661,
-      "step": 12685
-    },
-    {
-      "epoch": 6.615906127770534,
-      "grad_norm": 1.3954023122787476,
-      "learning_rate": 8.775577889447236e-05,
-      "loss": 5.6818,
-      "step": 12686
-    },
-    {
-      "epoch": 6.616427640156454,
-      "grad_norm": 1.458397388458252,
-      "learning_rate": 8.775477386934674e-05,
-      "loss": 5.6246,
-      "step": 12687
-    },
-    {
-      "epoch": 6.616949152542373,
-      "grad_norm": 1.4321602582931519,
-      "learning_rate": 8.77537688442211e-05,
-      "loss": 5.6714,
-      "step": 12688
-    },
-    {
-      "epoch": 6.617470664928292,
-      "grad_norm": 1.4184343814849854,
-      "learning_rate": 8.775276381909548e-05,
-      "loss": 5.7405,
-      "step": 12689
-    },
-    {
-      "epoch": 6.617992177314211,
-      "grad_norm": 1.5464221239089966,
-      "learning_rate": 8.775175879396986e-05,
-      "loss": 4.666,
-      "step": 12690
-    },
-    {
-      "epoch": 6.61851368970013,
-      "grad_norm": 1.6467781066894531,
-      "learning_rate": 8.775075376884424e-05,
-      "loss": 5.4469,
-      "step": 12691
-    },
-    {
-      "epoch": 6.6190352020860495,
-      "grad_norm": 1.391797423362732,
-      "learning_rate": 8.77497487437186e-05,
-      "loss": 5.8458,
-      "step": 12692
-    },
-    {
-      "epoch": 6.619556714471969,
-      "grad_norm": 1.3713209629058838,
-      "learning_rate": 8.774874371859298e-05,
-      "loss": 5.9564,
-      "step": 12693
-    },
-    {
-      "epoch": 6.6200782268578875,
-      "grad_norm": 1.374988079071045,
-      "learning_rate": 8.774773869346734e-05,
-      "loss": 5.0061,
-      "step": 12694
-    },
-    {
-      "epoch": 6.620599739243807,
-      "grad_norm": 1.3422836065292358,
-      "learning_rate": 8.77467336683417e-05,
-      "loss": 5.7642,
-      "step": 12695
-    },
-    {
-      "epoch": 6.621121251629726,
-      "grad_norm": 1.5050263404846191,
-      "learning_rate": 8.774572864321608e-05,
-      "loss": 5.3079,
-      "step": 12696
-    },
-    {
-      "epoch": 6.621642764015645,
-      "grad_norm": 1.3768254518508911,
-      "learning_rate": 8.774472361809045e-05,
-      "loss": 5.3917,
-      "step": 12697
-    },
-    {
-      "epoch": 6.6221642764015645,
-      "grad_norm": 1.417422890663147,
-      "learning_rate": 8.774371859296483e-05,
-      "loss": 5.8489,
-      "step": 12698
-    },
-    {
-      "epoch": 6.622685788787484,
-      "grad_norm": 1.5776814222335815,
-      "learning_rate": 8.774271356783919e-05,
-      "loss": 5.159,
-      "step": 12699
-    },
-    {
-      "epoch": 6.6232073011734025,
-      "grad_norm": 1.6243717670440674,
-      "learning_rate": 8.774170854271357e-05,
-      "loss": 5.5926,
-      "step": 12700
-    },
-    {
-      "epoch": 6.623728813559322,
-      "grad_norm": 1.4073995351791382,
-      "learning_rate": 8.774070351758795e-05,
-      "loss": 5.734,
-      "step": 12701
-    },
-    {
-      "epoch": 6.624250325945241,
-      "grad_norm": 1.3392020463943481,
-      "learning_rate": 8.773969849246232e-05,
-      "loss": 5.5972,
-      "step": 12702
-    },
-    {
-      "epoch": 6.62477183833116,
-      "grad_norm": 1.5713303089141846,
-      "learning_rate": 8.773869346733669e-05,
-      "loss": 5.7704,
-      "step": 12703
-    },
-    {
-      "epoch": 6.6252933507170795,
-      "grad_norm": 1.379541039466858,
-      "learning_rate": 8.773768844221107e-05,
-      "loss": 5.631,
-      "step": 12704
-    },
-    {
-      "epoch": 6.625814863102999,
-      "grad_norm": 1.4228026866912842,
-      "learning_rate": 8.773668341708543e-05,
-      "loss": 5.6018,
-      "step": 12705
-    },
-    {
-      "epoch": 6.6263363754889175,
-      "grad_norm": 1.4532055854797363,
-      "learning_rate": 8.773567839195981e-05,
-      "loss": 5.7198,
-      "step": 12706
-    },
-    {
-      "epoch": 6.626857887874837,
-      "grad_norm": 1.4348878860473633,
-      "learning_rate": 8.773467336683417e-05,
-      "loss": 5.4645,
-      "step": 12707
-    },
-    {
-      "epoch": 6.6273794002607564,
-      "grad_norm": 1.5477967262268066,
-      "learning_rate": 8.773366834170854e-05,
-      "loss": 5.3382,
-      "step": 12708
-    },
-    {
-      "epoch": 6.627900912646675,
-      "grad_norm": 1.6413317918777466,
-      "learning_rate": 8.773266331658291e-05,
-      "loss": 5.231,
-      "step": 12709
-    },
-    {
-      "epoch": 6.6284224250325945,
-      "grad_norm": 1.4687913656234741,
-      "learning_rate": 8.773165829145729e-05,
-      "loss": 5.2688,
-      "step": 12710
-    },
-    {
-      "epoch": 6.628943937418514,
-      "grad_norm": 1.382320761680603,
-      "learning_rate": 8.773065326633167e-05,
-      "loss": 5.2741,
-      "step": 12711
-    },
-    {
-      "epoch": 6.6294654498044325,
-      "grad_norm": 1.3559846878051758,
-      "learning_rate": 8.772964824120603e-05,
-      "loss": 5.904,
-      "step": 12712
-    },
-    {
-      "epoch": 6.629986962190352,
-      "grad_norm": 1.300837516784668,
-      "learning_rate": 8.772864321608041e-05,
-      "loss": 6.0428,
-      "step": 12713
-    },
-    {
-      "epoch": 6.6305084745762715,
-      "grad_norm": 1.3721174001693726,
-      "learning_rate": 8.772763819095478e-05,
-      "loss": 5.5082,
-      "step": 12714
-    },
-    {
-      "epoch": 6.63102998696219,
-      "grad_norm": 1.4383790493011475,
-      "learning_rate": 8.772663316582915e-05,
-      "loss": 5.3882,
-      "step": 12715
-    },
-    {
-      "epoch": 6.6315514993481095,
-      "grad_norm": 1.5193337202072144,
-      "learning_rate": 8.772562814070352e-05,
-      "loss": 5.4044,
-      "step": 12716
-    },
-    {
-      "epoch": 6.632073011734029,
-      "grad_norm": 1.5063661336898804,
-      "learning_rate": 8.77246231155779e-05,
-      "loss": 5.5102,
-      "step": 12717
-    },
-    {
-      "epoch": 6.6325945241199475,
-      "grad_norm": 1.6173903942108154,
-      "learning_rate": 8.772361809045226e-05,
-      "loss": 5.1089,
-      "step": 12718
-    },
-    {
-      "epoch": 6.633116036505867,
-      "grad_norm": 1.4732931852340698,
-      "learning_rate": 8.772261306532664e-05,
-      "loss": 5.6493,
-      "step": 12719
-    },
-    {
-      "epoch": 6.6336375488917865,
-      "grad_norm": 1.3904318809509277,
-      "learning_rate": 8.7721608040201e-05,
-      "loss": 5.7361,
-      "step": 12720
-    },
-    {
-      "epoch": 6.634159061277705,
-      "grad_norm": 1.4752448797225952,
-      "learning_rate": 8.772060301507538e-05,
-      "loss": 5.3505,
-      "step": 12721
-    },
-    {
-      "epoch": 6.6346805736636245,
-      "grad_norm": 1.310181736946106,
-      "learning_rate": 8.771959798994976e-05,
-      "loss": 5.1371,
-      "step": 12722
-    },
-    {
-      "epoch": 6.635202086049544,
-      "grad_norm": 1.5187984704971313,
-      "learning_rate": 8.771859296482412e-05,
-      "loss": 5.3394,
-      "step": 12723
-    },
-    {
-      "epoch": 6.6357235984354626,
-      "grad_norm": 1.3388502597808838,
-      "learning_rate": 8.77175879396985e-05,
-      "loss": 5.5252,
-      "step": 12724
-    },
-    {
-      "epoch": 6.636245110821382,
-      "grad_norm": 1.4206173419952393,
-      "learning_rate": 8.771658291457286e-05,
-      "loss": 5.7729,
-      "step": 12725
-    },
-    {
-      "epoch": 6.6367666232073015,
-      "grad_norm": 1.3759329319000244,
-      "learning_rate": 8.771557788944724e-05,
-      "loss": 5.842,
-      "step": 12726
-    },
-    {
-      "epoch": 6.63728813559322,
-      "grad_norm": 1.243151068687439,
-      "learning_rate": 8.771457286432161e-05,
-      "loss": 6.1274,
-      "step": 12727
-    },
-    {
-      "epoch": 6.6378096479791395,
-      "grad_norm": 1.3851327896118164,
-      "learning_rate": 8.771356783919598e-05,
-      "loss": 5.8062,
-      "step": 12728
-    },
-    {
-      "epoch": 6.638331160365059,
-      "grad_norm": 1.492940068244934,
-      "learning_rate": 8.771256281407035e-05,
-      "loss": 5.3939,
-      "step": 12729
-    },
-    {
-      "epoch": 6.638852672750978,
-      "grad_norm": 1.4239271879196167,
-      "learning_rate": 8.771155778894473e-05,
-      "loss": 5.9161,
-      "step": 12730
-    },
-    {
-      "epoch": 6.639374185136897,
-      "grad_norm": 1.495632529258728,
-      "learning_rate": 8.77105527638191e-05,
-      "loss": 5.8951,
-      "step": 12731
-    },
-    {
-      "epoch": 6.6398956975228165,
-      "grad_norm": 1.5401967763900757,
-      "learning_rate": 8.770954773869348e-05,
-      "loss": 5.4034,
-      "step": 12732
-    },
-    {
-      "epoch": 6.640417209908735,
-      "grad_norm": 1.3943099975585938,
-      "learning_rate": 8.770854271356785e-05,
-      "loss": 5.4726,
-      "step": 12733
-    },
-    {
-      "epoch": 6.6409387222946545,
-      "grad_norm": 1.4532575607299805,
-      "learning_rate": 8.770753768844222e-05,
-      "loss": 5.7614,
-      "step": 12734
-    },
-    {
-      "epoch": 6.641460234680574,
-      "grad_norm": 1.4067597389221191,
-      "learning_rate": 8.770653266331659e-05,
-      "loss": 5.3414,
-      "step": 12735
-    },
-    {
-      "epoch": 6.641981747066493,
-      "grad_norm": 1.455438256263733,
-      "learning_rate": 8.770552763819095e-05,
-      "loss": 5.7926,
-      "step": 12736
-    },
-    {
-      "epoch": 6.642503259452412,
-      "grad_norm": 1.365914225578308,
-      "learning_rate": 8.770452261306533e-05,
-      "loss": 5.5659,
-      "step": 12737
-    },
-    {
-      "epoch": 6.6430247718383315,
-      "grad_norm": 1.4311046600341797,
-      "learning_rate": 8.77035175879397e-05,
-      "loss": 5.617,
-      "step": 12738
-    },
-    {
-      "epoch": 6.64354628422425,
-      "grad_norm": 1.3220605850219727,
-      "learning_rate": 8.770251256281407e-05,
-      "loss": 5.7599,
-      "step": 12739
-    },
-    {
-      "epoch": 6.6440677966101696,
-      "grad_norm": 1.304575800895691,
-      "learning_rate": 8.770150753768844e-05,
-      "loss": 5.8248,
-      "step": 12740
-    },
-    {
-      "epoch": 6.644589308996089,
-      "grad_norm": 1.4549000263214111,
-      "learning_rate": 8.770050251256282e-05,
-      "loss": 5.7301,
-      "step": 12741
-    },
-    {
-      "epoch": 6.645110821382008,
-      "grad_norm": 1.3772786855697632,
-      "learning_rate": 8.769949748743719e-05,
-      "loss": 5.3651,
-      "step": 12742
-    },
-    {
-      "epoch": 6.645632333767927,
-      "grad_norm": 1.3597321510314941,
-      "learning_rate": 8.769849246231157e-05,
-      "loss": 5.7362,
-      "step": 12743
-    },
-    {
-      "epoch": 6.6461538461538465,
-      "grad_norm": 1.366639256477356,
-      "learning_rate": 8.769748743718593e-05,
-      "loss": 5.8753,
-      "step": 12744
-    },
-    {
-      "epoch": 6.646675358539765,
-      "grad_norm": 1.414522409439087,
-      "learning_rate": 8.769648241206031e-05,
-      "loss": 4.8106,
-      "step": 12745
-    },
-    {
-      "epoch": 6.647196870925685,
-      "grad_norm": 1.3914718627929688,
-      "learning_rate": 8.769547738693468e-05,
-      "loss": 5.7686,
-      "step": 12746
-    },
-    {
-      "epoch": 6.647718383311604,
-      "grad_norm": 1.4633933305740356,
-      "learning_rate": 8.769447236180905e-05,
-      "loss": 5.9685,
-      "step": 12747
-    },
-    {
-      "epoch": 6.648239895697523,
-      "grad_norm": 1.4253698587417603,
-      "learning_rate": 8.769346733668342e-05,
-      "loss": 5.2858,
-      "step": 12748
-    },
-    {
-      "epoch": 6.648761408083442,
-      "grad_norm": 1.4143096208572388,
-      "learning_rate": 8.769246231155778e-05,
-      "loss": 5.5635,
-      "step": 12749
-    },
-    {
-      "epoch": 6.6492829204693615,
-      "grad_norm": 1.2959723472595215,
-      "learning_rate": 8.769145728643216e-05,
-      "loss": 6.0348,
-      "step": 12750
-    },
-    {
-      "epoch": 6.64980443285528,
-      "grad_norm": 1.3806322813034058,
-      "learning_rate": 8.769045226130654e-05,
-      "loss": 5.7625,
-      "step": 12751
-    },
-    {
-      "epoch": 6.6503259452412,
-      "grad_norm": 1.4848254919052124,
-      "learning_rate": 8.768944723618092e-05,
-      "loss": 5.3936,
-      "step": 12752
-    },
-    {
-      "epoch": 6.650847457627119,
-      "grad_norm": 1.328060507774353,
-      "learning_rate": 8.768844221105528e-05,
-      "loss": 5.7045,
-      "step": 12753
-    },
-    {
-      "epoch": 6.651368970013038,
-      "grad_norm": 1.4735866785049438,
-      "learning_rate": 8.768743718592966e-05,
-      "loss": 5.3884,
-      "step": 12754
-    },
-    {
-      "epoch": 6.651890482398957,
-      "grad_norm": 1.5322743654251099,
-      "learning_rate": 8.768643216080402e-05,
-      "loss": 5.3671,
-      "step": 12755
-    },
-    {
-      "epoch": 6.652411994784876,
-      "grad_norm": 1.2741072177886963,
-      "learning_rate": 8.76854271356784e-05,
-      "loss": 5.7149,
-      "step": 12756
-    },
-    {
-      "epoch": 6.652933507170795,
-      "grad_norm": 1.458165168762207,
-      "learning_rate": 8.768442211055277e-05,
-      "loss": 5.5331,
-      "step": 12757
-    },
-    {
-      "epoch": 6.653455019556715,
-      "grad_norm": 1.3718332052230835,
-      "learning_rate": 8.768341708542714e-05,
-      "loss": 5.9309,
-      "step": 12758
-    },
-    {
-      "epoch": 6.653976531942634,
-      "grad_norm": 1.5553209781646729,
-      "learning_rate": 8.768241206030151e-05,
-      "loss": 4.8039,
-      "step": 12759
-    },
-    {
-      "epoch": 6.654498044328553,
-      "grad_norm": 1.4828732013702393,
-      "learning_rate": 8.768140703517589e-05,
-      "loss": 5.8454,
-      "step": 12760
-    },
-    {
-      "epoch": 6.655019556714472,
-      "grad_norm": 1.466138243675232,
-      "learning_rate": 8.768040201005025e-05,
-      "loss": 5.7301,
-      "step": 12761
-    },
-    {
-      "epoch": 6.655541069100391,
-      "grad_norm": 1.4061579704284668,
-      "learning_rate": 8.767939698492463e-05,
-      "loss": 5.5757,
-      "step": 12762
-    },
-    {
-      "epoch": 6.65606258148631,
-      "grad_norm": 1.4469388723373413,
-      "learning_rate": 8.7678391959799e-05,
-      "loss": 5.4785,
-      "step": 12763
-    },
-    {
-      "epoch": 6.65658409387223,
-      "grad_norm": 1.3981482982635498,
-      "learning_rate": 8.767738693467337e-05,
-      "loss": 5.6365,
-      "step": 12764
-    },
-    {
-      "epoch": 6.657105606258149,
-      "grad_norm": 1.6990699768066406,
-      "learning_rate": 8.767638190954775e-05,
-      "loss": 5.5737,
-      "step": 12765
-    },
-    {
-      "epoch": 6.657627118644068,
-      "grad_norm": 1.548140287399292,
-      "learning_rate": 8.767537688442211e-05,
-      "loss": 5.7171,
-      "step": 12766
-    },
-    {
-      "epoch": 6.658148631029987,
-      "grad_norm": 1.5899220705032349,
-      "learning_rate": 8.767437185929649e-05,
-      "loss": 5.3846,
-      "step": 12767
-    },
-    {
-      "epoch": 6.658670143415906,
-      "grad_norm": 1.3387292623519897,
-      "learning_rate": 8.767336683417085e-05,
-      "loss": 6.0568,
-      "step": 12768
-    },
-    {
-      "epoch": 6.659191655801825,
-      "grad_norm": 1.3478761911392212,
-      "learning_rate": 8.767236180904523e-05,
-      "loss": 5.7805,
-      "step": 12769
-    },
-    {
-      "epoch": 6.659713168187745,
-      "grad_norm": 1.42526113986969,
-      "learning_rate": 8.76713567839196e-05,
-      "loss": 5.8357,
-      "step": 12770
-    },
-    {
-      "epoch": 6.660234680573664,
-      "grad_norm": 1.4818952083587646,
-      "learning_rate": 8.767035175879397e-05,
-      "loss": 5.8157,
-      "step": 12771
-    },
-    {
-      "epoch": 6.660756192959583,
-      "grad_norm": 1.4578192234039307,
-      "learning_rate": 8.766934673366835e-05,
-      "loss": 5.2855,
-      "step": 12772
-    },
-    {
-      "epoch": 6.661277705345502,
-      "grad_norm": 1.4734182357788086,
-      "learning_rate": 8.766834170854273e-05,
-      "loss": 5.6158,
-      "step": 12773
-    },
-    {
-      "epoch": 6.661799217731421,
-      "grad_norm": 1.6257375478744507,
-      "learning_rate": 8.76673366834171e-05,
-      "loss": 5.6439,
-      "step": 12774
-    },
-    {
-      "epoch": 6.66232073011734,
-      "grad_norm": 1.4205780029296875,
-      "learning_rate": 8.766633165829146e-05,
-      "loss": 5.2789,
-      "step": 12775
-    },
-    {
-      "epoch": 6.66284224250326,
-      "grad_norm": 1.5704612731933594,
-      "learning_rate": 8.766532663316584e-05,
-      "loss": 5.4904,
-      "step": 12776
-    },
-    {
-      "epoch": 6.663363754889179,
-      "grad_norm": 1.4090694189071655,
-      "learning_rate": 8.76643216080402e-05,
-      "loss": 5.3308,
-      "step": 12777
-    },
-    {
-      "epoch": 6.663885267275098,
-      "grad_norm": 1.536839485168457,
-      "learning_rate": 8.766331658291458e-05,
-      "loss": 5.5016,
-      "step": 12778
-    },
-    {
-      "epoch": 6.664406779661017,
-      "grad_norm": 1.3951611518859863,
-      "learning_rate": 8.766231155778894e-05,
-      "loss": 5.5258,
-      "step": 12779
-    },
-    {
-      "epoch": 6.664928292046936,
-      "grad_norm": 1.523594617843628,
-      "learning_rate": 8.766130653266332e-05,
-      "loss": 5.5442,
-      "step": 12780
-    },
-    {
-      "epoch": 6.665449804432855,
-      "grad_norm": 1.4146018028259277,
-      "learning_rate": 8.766030150753768e-05,
-      "loss": 5.62,
-      "step": 12781
-    },
-    {
-      "epoch": 6.665971316818775,
-      "grad_norm": 1.394034743309021,
-      "learning_rate": 8.765929648241206e-05,
-      "loss": 5.6149,
-      "step": 12782
-    },
-    {
-      "epoch": 6.666492829204694,
-      "grad_norm": 1.3811874389648438,
-      "learning_rate": 8.765829145728644e-05,
-      "loss": 5.7538,
-      "step": 12783
-    },
-    {
-      "epoch": 6.667014341590613,
-      "grad_norm": 1.4417401552200317,
-      "learning_rate": 8.765728643216082e-05,
-      "loss": 5.6746,
-      "step": 12784
-    },
-    {
-      "epoch": 6.667535853976532,
-      "grad_norm": 1.4531605243682861,
-      "learning_rate": 8.765628140703518e-05,
-      "loss": 5.2057,
-      "step": 12785
-    },
-    {
-      "epoch": 6.668057366362451,
-      "grad_norm": 1.3937252759933472,
-      "learning_rate": 8.765527638190956e-05,
-      "loss": 5.8601,
-      "step": 12786
-    },
-    {
-      "epoch": 6.66857887874837,
-      "grad_norm": 1.3796398639678955,
-      "learning_rate": 8.765427135678392e-05,
-      "loss": 5.9233,
-      "step": 12787
-    },
-    {
-      "epoch": 6.66910039113429,
-      "grad_norm": 1.3045505285263062,
-      "learning_rate": 8.765326633165829e-05,
-      "loss": 5.0307,
-      "step": 12788
-    },
-    {
-      "epoch": 6.669621903520208,
-      "grad_norm": 1.3593705892562866,
-      "learning_rate": 8.765226130653267e-05,
-      "loss": 5.9502,
-      "step": 12789
-    },
-    {
-      "epoch": 6.670143415906128,
-      "grad_norm": 1.4448844194412231,
-      "learning_rate": 8.765125628140703e-05,
-      "loss": 5.7594,
-      "step": 12790
-    },
-    {
-      "epoch": 6.670664928292047,
-      "grad_norm": 1.425901174545288,
-      "learning_rate": 8.765025125628141e-05,
-      "loss": 6.0024,
-      "step": 12791
-    },
-    {
-      "epoch": 6.671186440677966,
-      "grad_norm": 1.4235531091690063,
-      "learning_rate": 8.764924623115579e-05,
-      "loss": 5.7386,
-      "step": 12792
-    },
-    {
-      "epoch": 6.671707953063885,
-      "grad_norm": 1.4574190378189087,
-      "learning_rate": 8.764824120603016e-05,
-      "loss": 4.8158,
-      "step": 12793
-    },
-    {
-      "epoch": 6.672229465449805,
-      "grad_norm": 1.4605151414871216,
-      "learning_rate": 8.764723618090453e-05,
-      "loss": 5.8209,
-      "step": 12794
-    },
-    {
-      "epoch": 6.672750977835723,
-      "grad_norm": 1.4213238954544067,
-      "learning_rate": 8.76462311557789e-05,
-      "loss": 5.5796,
-      "step": 12795
-    },
-    {
-      "epoch": 6.673272490221643,
-      "grad_norm": 1.410429835319519,
-      "learning_rate": 8.764522613065327e-05,
-      "loss": 5.4916,
-      "step": 12796
-    },
-    {
-      "epoch": 6.673794002607562,
-      "grad_norm": 1.3719998598098755,
-      "learning_rate": 8.764422110552765e-05,
-      "loss": 5.4208,
-      "step": 12797
-    },
-    {
-      "epoch": 6.674315514993481,
-      "grad_norm": 1.3995013236999512,
-      "learning_rate": 8.764321608040201e-05,
-      "loss": 5.2556,
-      "step": 12798
-    },
-    {
-      "epoch": 6.6748370273794,
-      "grad_norm": 1.488429307937622,
-      "learning_rate": 8.764221105527639e-05,
-      "loss": 5.428,
-      "step": 12799
-    },
-    {
-      "epoch": 6.67535853976532,
-      "grad_norm": 1.4296579360961914,
-      "learning_rate": 8.764120603015075e-05,
-      "loss": 6.0151,
-      "step": 12800
-    },
-    {
-      "epoch": 6.675880052151238,
-      "grad_norm": 1.5000637769699097,
-      "learning_rate": 8.764020100502512e-05,
-      "loss": 5.6728,
-      "step": 12801
-    },
-    {
-      "epoch": 6.676401564537158,
-      "grad_norm": 1.3603116273880005,
-      "learning_rate": 8.76391959798995e-05,
-      "loss": 5.7496,
-      "step": 12802
-    },
-    {
-      "epoch": 6.676923076923077,
-      "grad_norm": 1.472214698791504,
-      "learning_rate": 8.763819095477387e-05,
-      "loss": 5.4024,
-      "step": 12803
-    },
-    {
-      "epoch": 6.677444589308996,
-      "grad_norm": 1.5616724491119385,
-      "learning_rate": 8.763718592964825e-05,
-      "loss": 5.3116,
-      "step": 12804
-    },
-    {
-      "epoch": 6.677966101694915,
-      "grad_norm": 1.351335048675537,
-      "learning_rate": 8.763618090452262e-05,
-      "loss": 5.6412,
-      "step": 12805
-    },
-    {
-      "epoch": 6.678487614080835,
-      "grad_norm": 1.3945475816726685,
-      "learning_rate": 8.7635175879397e-05,
-      "loss": 5.1248,
-      "step": 12806
-    },
-    {
-      "epoch": 6.679009126466753,
-      "grad_norm": 1.5649261474609375,
-      "learning_rate": 8.763417085427136e-05,
-      "loss": 4.9444,
-      "step": 12807
-    },
-    {
-      "epoch": 6.679530638852673,
-      "grad_norm": 1.782228946685791,
-      "learning_rate": 8.763316582914574e-05,
-      "loss": 4.9237,
-      "step": 12808
-    },
-    {
-      "epoch": 6.680052151238592,
-      "grad_norm": 1.4529575109481812,
-      "learning_rate": 8.76321608040201e-05,
-      "loss": 5.2834,
-      "step": 12809
-    },
-    {
-      "epoch": 6.680573663624511,
-      "grad_norm": 1.6766923666000366,
-      "learning_rate": 8.763115577889448e-05,
-      "loss": 5.5437,
-      "step": 12810
-    },
-    {
-      "epoch": 6.68109517601043,
-      "grad_norm": 1.367929458618164,
-      "learning_rate": 8.763015075376884e-05,
-      "loss": 5.4899,
-      "step": 12811
-    },
-    {
-      "epoch": 6.68161668839635,
-      "grad_norm": 1.438033103942871,
-      "learning_rate": 8.762914572864322e-05,
-      "loss": 5.3234,
-      "step": 12812
-    },
-    {
-      "epoch": 6.682138200782268,
-      "grad_norm": 1.445300579071045,
-      "learning_rate": 8.762814070351758e-05,
-      "loss": 5.319,
-      "step": 12813
-    },
-    {
-      "epoch": 6.682659713168188,
-      "grad_norm": 1.4838016033172607,
-      "learning_rate": 8.762713567839196e-05,
-      "loss": 5.4681,
-      "step": 12814
-    },
-    {
-      "epoch": 6.683181225554107,
-      "grad_norm": 1.3934568166732788,
-      "learning_rate": 8.762613065326634e-05,
-      "loss": 6.2128,
-      "step": 12815
-    },
-    {
-      "epoch": 6.683702737940026,
-      "grad_norm": 1.3830921649932861,
-      "learning_rate": 8.76251256281407e-05,
-      "loss": 5.6773,
-      "step": 12816
-    },
-    {
-      "epoch": 6.684224250325945,
-      "grad_norm": 1.567723035812378,
-      "learning_rate": 8.762412060301508e-05,
-      "loss": 5.415,
-      "step": 12817
-    },
-    {
-      "epoch": 6.684745762711865,
-      "grad_norm": 1.479431390762329,
-      "learning_rate": 8.762311557788945e-05,
-      "loss": 5.8143,
-      "step": 12818
-    },
-    {
-      "epoch": 6.685267275097783,
-      "grad_norm": 1.4677802324295044,
-      "learning_rate": 8.762211055276382e-05,
-      "loss": 5.733,
-      "step": 12819
-    },
-    {
-      "epoch": 6.685788787483703,
-      "grad_norm": 1.3229190111160278,
-      "learning_rate": 8.762110552763819e-05,
-      "loss": 5.4112,
-      "step": 12820
-    },
-    {
-      "epoch": 6.686310299869622,
-      "grad_norm": 1.5249003171920776,
-      "learning_rate": 8.762010050251257e-05,
-      "loss": 5.551,
-      "step": 12821
-    },
-    {
-      "epoch": 6.686831812255541,
-      "grad_norm": 1.4870644807815552,
-      "learning_rate": 8.761909547738693e-05,
-      "loss": 5.5876,
-      "step": 12822
-    },
-    {
-      "epoch": 6.68735332464146,
-      "grad_norm": 1.4516531229019165,
-      "learning_rate": 8.761809045226131e-05,
-      "loss": 5.3875,
-      "step": 12823
-    },
-    {
-      "epoch": 6.68787483702738,
-      "grad_norm": 1.5230038166046143,
-      "learning_rate": 8.761708542713569e-05,
-      "loss": 5.5145,
-      "step": 12824
-    },
-    {
-      "epoch": 6.688396349413298,
-      "grad_norm": 1.3259910345077515,
-      "learning_rate": 8.761608040201006e-05,
-      "loss": 6.002,
-      "step": 12825
-    },
-    {
-      "epoch": 6.688917861799218,
-      "grad_norm": 1.4712027311325073,
-      "learning_rate": 8.761507537688443e-05,
-      "loss": 5.5068,
-      "step": 12826
-    },
-    {
-      "epoch": 6.689439374185137,
-      "grad_norm": 1.4919196367263794,
-      "learning_rate": 8.761407035175881e-05,
-      "loss": 5.0339,
-      "step": 12827
-    },
-    {
-      "epoch": 6.689960886571056,
-      "grad_norm": 1.5420600175857544,
-      "learning_rate": 8.761306532663317e-05,
-      "loss": 5.6854,
-      "step": 12828
-    },
-    {
-      "epoch": 6.690482398956975,
-      "grad_norm": 1.422290325164795,
-      "learning_rate": 8.761206030150754e-05,
-      "loss": 5.6722,
-      "step": 12829
-    },
-    {
-      "epoch": 6.691003911342895,
-      "grad_norm": 1.423728346824646,
-      "learning_rate": 8.761105527638191e-05,
-      "loss": 5.6957,
-      "step": 12830
-    },
-    {
-      "epoch": 6.691525423728813,
-      "grad_norm": 1.457627296447754,
-      "learning_rate": 8.761005025125628e-05,
-      "loss": 5.9858,
-      "step": 12831
-    },
-    {
-      "epoch": 6.692046936114733,
-      "grad_norm": 1.4462013244628906,
-      "learning_rate": 8.760904522613066e-05,
-      "loss": 5.7718,
-      "step": 12832
-    },
-    {
-      "epoch": 6.692568448500652,
-      "grad_norm": 1.5031582117080688,
-      "learning_rate": 8.760804020100502e-05,
-      "loss": 5.622,
-      "step": 12833
-    },
-    {
-      "epoch": 6.693089960886571,
-      "grad_norm": 1.4214770793914795,
-      "learning_rate": 8.76070351758794e-05,
-      "loss": 5.9267,
-      "step": 12834
-    },
-    {
-      "epoch": 6.69361147327249,
-      "grad_norm": 1.6774251461029053,
-      "learning_rate": 8.760603015075378e-05,
-      "loss": 5.2405,
-      "step": 12835
-    },
-    {
-      "epoch": 6.69413298565841,
-      "grad_norm": 1.5648654699325562,
-      "learning_rate": 8.760502512562815e-05,
-      "loss": 4.3913,
-      "step": 12836
-    },
-    {
-      "epoch": 6.694654498044328,
-      "grad_norm": 1.455964207649231,
-      "learning_rate": 8.760402010050252e-05,
-      "loss": 5.5786,
-      "step": 12837
-    },
-    {
-      "epoch": 6.695176010430248,
-      "grad_norm": 1.5922733545303345,
-      "learning_rate": 8.76030150753769e-05,
-      "loss": 5.6251,
-      "step": 12838
-    },
-    {
-      "epoch": 6.695697522816167,
-      "grad_norm": 1.4571183919906616,
-      "learning_rate": 8.760201005025126e-05,
-      "loss": 5.5927,
-      "step": 12839
-    },
-    {
-      "epoch": 6.696219035202086,
-      "grad_norm": 1.420799970626831,
-      "learning_rate": 8.760100502512564e-05,
-      "loss": 5.5727,
-      "step": 12840
-    },
-    {
-      "epoch": 6.696740547588005,
-      "grad_norm": 1.4972995519638062,
-      "learning_rate": 8.76e-05,
-      "loss": 5.7641,
-      "step": 12841
-    },
-    {
-      "epoch": 6.697262059973925,
-      "grad_norm": 1.6494814157485962,
-      "learning_rate": 8.759899497487437e-05,
-      "loss": 5.2854,
-      "step": 12842
-    },
-    {
-      "epoch": 6.697783572359843,
-      "grad_norm": 1.5611070394515991,
-      "learning_rate": 8.759798994974874e-05,
-      "loss": 5.6837,
-      "step": 12843
-    },
-    {
-      "epoch": 6.698305084745763,
-      "grad_norm": 1.448598861694336,
-      "learning_rate": 8.759698492462312e-05,
-      "loss": 5.5205,
-      "step": 12844
-    },
-    {
-      "epoch": 6.698826597131681,
-      "grad_norm": 1.506805658340454,
-      "learning_rate": 8.75959798994975e-05,
-      "loss": 5.5927,
-      "step": 12845
-    },
-    {
-      "epoch": 6.699348109517601,
-      "grad_norm": 1.4841135740280151,
-      "learning_rate": 8.759497487437186e-05,
-      "loss": 5.7315,
-      "step": 12846
-    },
-    {
-      "epoch": 6.69986962190352,
-      "grad_norm": 1.4664496183395386,
-      "learning_rate": 8.759396984924624e-05,
-      "loss": 5.8964,
-      "step": 12847
-    },
-    {
-      "epoch": 6.70039113428944,
-      "grad_norm": 1.5546787977218628,
-      "learning_rate": 8.75929648241206e-05,
-      "loss": 5.2254,
-      "step": 12848
-    },
-    {
-      "epoch": 6.700912646675358,
-      "grad_norm": 1.4552676677703857,
-      "learning_rate": 8.759195979899498e-05,
-      "loss": 5.1671,
-      "step": 12849
-    },
-    {
-      "epoch": 6.701434159061278,
-      "grad_norm": 1.4057780504226685,
-      "learning_rate": 8.759095477386935e-05,
-      "loss": 5.464,
-      "step": 12850
-    },
-    {
-      "epoch": 6.701955671447196,
-      "grad_norm": 1.4960026741027832,
-      "learning_rate": 8.758994974874373e-05,
-      "loss": 5.165,
-      "step": 12851
-    },
-    {
-      "epoch": 6.702477183833116,
-      "grad_norm": 1.4670170545578003,
-      "learning_rate": 8.758894472361809e-05,
-      "loss": 5.5209,
-      "step": 12852
-    },
-    {
-      "epoch": 6.702998696219035,
-      "grad_norm": 1.3772598505020142,
-      "learning_rate": 8.758793969849247e-05,
-      "loss": 5.8247,
-      "step": 12853
-    },
-    {
-      "epoch": 6.703520208604955,
-      "grad_norm": 1.6594270467758179,
-      "learning_rate": 8.758693467336683e-05,
-      "loss": 5.1074,
-      "step": 12854
-    },
-    {
-      "epoch": 6.704041720990873,
-      "grad_norm": 1.5321290493011475,
-      "learning_rate": 8.758592964824121e-05,
-      "loss": 5.3988,
-      "step": 12855
-    },
-    {
-      "epoch": 6.704563233376793,
-      "grad_norm": 1.4614876508712769,
-      "learning_rate": 8.758492462311559e-05,
-      "loss": 5.3676,
-      "step": 12856
-    },
-    {
-      "epoch": 6.705084745762711,
-      "grad_norm": 1.4088186025619507,
-      "learning_rate": 8.758391959798995e-05,
-      "loss": 5.5491,
-      "step": 12857
-    },
-    {
-      "epoch": 6.705606258148631,
-      "grad_norm": 1.354395866394043,
-      "learning_rate": 8.758291457286433e-05,
-      "loss": 5.9257,
-      "step": 12858
-    },
-    {
-      "epoch": 6.70612777053455,
-      "grad_norm": 1.4872767925262451,
-      "learning_rate": 8.75819095477387e-05,
-      "loss": 5.7384,
-      "step": 12859
-    },
-    {
-      "epoch": 6.70664928292047,
-      "grad_norm": 1.5058050155639648,
-      "learning_rate": 8.758090452261307e-05,
-      "loss": 5.701,
-      "step": 12860
-    },
-    {
-      "epoch": 6.707170795306388,
-      "grad_norm": 1.4931447505950928,
-      "learning_rate": 8.757989949748744e-05,
-      "loss": 5.314,
-      "step": 12861
-    },
-    {
-      "epoch": 6.707692307692308,
-      "grad_norm": 1.3797168731689453,
-      "learning_rate": 8.757889447236181e-05,
-      "loss": 5.4726,
-      "step": 12862
-    },
-    {
-      "epoch": 6.708213820078226,
-      "grad_norm": 1.3089187145233154,
-      "learning_rate": 8.757788944723618e-05,
-      "loss": 5.0096,
-      "step": 12863
-    },
-    {
-      "epoch": 6.708735332464146,
-      "grad_norm": 1.4904160499572754,
-      "learning_rate": 8.757688442211056e-05,
-      "loss": 5.8512,
-      "step": 12864
-    },
-    {
-      "epoch": 6.709256844850065,
-      "grad_norm": 1.6877117156982422,
-      "learning_rate": 8.757587939698493e-05,
-      "loss": 5.2088,
-      "step": 12865
-    },
-    {
-      "epoch": 6.709778357235985,
-      "grad_norm": 1.397830843925476,
-      "learning_rate": 8.757487437185931e-05,
-      "loss": 5.4284,
-      "step": 12866
-    },
-    {
-      "epoch": 6.710299869621903,
-      "grad_norm": 1.425405502319336,
-      "learning_rate": 8.757386934673368e-05,
-      "loss": 5.8197,
-      "step": 12867
-    },
-    {
-      "epoch": 6.710821382007823,
-      "grad_norm": 1.35599684715271,
-      "learning_rate": 8.757286432160804e-05,
-      "loss": 5.4481,
-      "step": 12868
-    },
-    {
-      "epoch": 6.711342894393741,
-      "grad_norm": 1.4064160585403442,
-      "learning_rate": 8.757185929648242e-05,
-      "loss": 5.3225,
-      "step": 12869
-    },
-    {
-      "epoch": 6.711864406779661,
-      "grad_norm": 1.5886445045471191,
-      "learning_rate": 8.757085427135678e-05,
-      "loss": 5.5014,
-      "step": 12870
-    },
-    {
-      "epoch": 6.71238591916558,
-      "grad_norm": 1.5648869276046753,
-      "learning_rate": 8.756984924623116e-05,
-      "loss": 5.312,
-      "step": 12871
-    },
-    {
-      "epoch": 6.7129074315515,
-      "grad_norm": 1.6132185459136963,
-      "learning_rate": 8.756884422110552e-05,
-      "loss": 5.8742,
-      "step": 12872
-    },
-    {
-      "epoch": 6.713428943937418,
-      "grad_norm": 1.5035642385482788,
-      "learning_rate": 8.75678391959799e-05,
-      "loss": 5.4092,
-      "step": 12873
-    },
-    {
-      "epoch": 6.713950456323338,
-      "grad_norm": 1.4022771120071411,
-      "learning_rate": 8.756683417085427e-05,
-      "loss": 5.8416,
-      "step": 12874
-    },
-    {
-      "epoch": 6.7144719687092564,
-      "grad_norm": 1.4623056650161743,
-      "learning_rate": 8.756582914572864e-05,
-      "loss": 5.7253,
-      "step": 12875
-    },
-    {
-      "epoch": 6.714993481095176,
-      "grad_norm": 1.438920021057129,
-      "learning_rate": 8.756482412060302e-05,
-      "loss": 5.4507,
-      "step": 12876
-    },
-    {
-      "epoch": 6.715514993481095,
-      "grad_norm": 1.503581166267395,
-      "learning_rate": 8.75638190954774e-05,
-      "loss": 5.2992,
-      "step": 12877
-    },
-    {
-      "epoch": 6.716036505867015,
-      "grad_norm": 1.7550193071365356,
-      "learning_rate": 8.756281407035176e-05,
-      "loss": 4.1992,
-      "step": 12878
-    },
-    {
-      "epoch": 6.716558018252933,
-      "grad_norm": 1.4273511171340942,
-      "learning_rate": 8.756180904522614e-05,
-      "loss": 5.6976,
-      "step": 12879
-    },
-    {
-      "epoch": 6.717079530638853,
-      "grad_norm": 1.417890191078186,
-      "learning_rate": 8.75608040201005e-05,
-      "loss": 6.0029,
-      "step": 12880
-    },
-    {
-      "epoch": 6.7176010430247715,
-      "grad_norm": 1.38090980052948,
-      "learning_rate": 8.755979899497487e-05,
-      "loss": 5.5322,
-      "step": 12881
-    },
-    {
-      "epoch": 6.718122555410691,
-      "grad_norm": 1.5042381286621094,
-      "learning_rate": 8.755879396984925e-05,
-      "loss": 5.5812,
-      "step": 12882
-    },
-    {
-      "epoch": 6.71864406779661,
-      "grad_norm": 1.400622010231018,
-      "learning_rate": 8.755778894472361e-05,
-      "loss": 5.8757,
-      "step": 12883
-    },
-    {
-      "epoch": 6.719165580182529,
-      "grad_norm": 1.306994080543518,
-      "learning_rate": 8.755678391959799e-05,
-      "loss": 5.9408,
-      "step": 12884
-    },
-    {
-      "epoch": 6.719687092568448,
-      "grad_norm": 1.3501349687576294,
-      "learning_rate": 8.755577889447237e-05,
-      "loss": 5.9628,
-      "step": 12885
-    },
-    {
-      "epoch": 6.720208604954368,
-      "grad_norm": 1.4997843503952026,
-      "learning_rate": 8.755477386934675e-05,
-      "loss": 5.3667,
-      "step": 12886
-    },
-    {
-      "epoch": 6.7207301173402865,
-      "grad_norm": 1.3222925662994385,
-      "learning_rate": 8.755376884422111e-05,
-      "loss": 5.8042,
-      "step": 12887
-    },
-    {
-      "epoch": 6.721251629726206,
-      "grad_norm": 1.421418309211731,
-      "learning_rate": 8.755276381909549e-05,
-      "loss": 5.6018,
-      "step": 12888
-    },
-    {
-      "epoch": 6.721773142112125,
-      "grad_norm": 1.5302369594573975,
-      "learning_rate": 8.755175879396985e-05,
-      "loss": 5.455,
-      "step": 12889
-    },
-    {
-      "epoch": 6.722294654498044,
-      "grad_norm": 1.4184293746948242,
-      "learning_rate": 8.755075376884423e-05,
-      "loss": 5.8157,
-      "step": 12890
-    },
-    {
-      "epoch": 6.722816166883963,
-      "grad_norm": 1.4871785640716553,
-      "learning_rate": 8.75497487437186e-05,
-      "loss": 5.3941,
-      "step": 12891
-    },
-    {
-      "epoch": 6.723337679269883,
-      "grad_norm": 1.353327751159668,
-      "learning_rate": 8.754874371859297e-05,
-      "loss": 5.8094,
-      "step": 12892
-    },
-    {
-      "epoch": 6.7238591916558015,
-      "grad_norm": 1.4081494808197021,
-      "learning_rate": 8.754773869346734e-05,
-      "loss": 5.6403,
-      "step": 12893
-    },
-    {
-      "epoch": 6.724380704041721,
-      "grad_norm": 1.4502794742584229,
-      "learning_rate": 8.75467336683417e-05,
-      "loss": 5.5797,
-      "step": 12894
-    },
-    {
-      "epoch": 6.72490221642764,
-      "grad_norm": 1.3688503503799438,
-      "learning_rate": 8.754572864321608e-05,
-      "loss": 5.8725,
-      "step": 12895
-    },
-    {
-      "epoch": 6.725423728813559,
-      "grad_norm": 1.5035464763641357,
-      "learning_rate": 8.754472361809046e-05,
-      "loss": 5.3924,
-      "step": 12896
-    },
-    {
-      "epoch": 6.7259452411994785,
-      "grad_norm": 1.286468267440796,
-      "learning_rate": 8.754371859296483e-05,
-      "loss": 6.0677,
-      "step": 12897
-    },
-    {
-      "epoch": 6.726466753585398,
-      "grad_norm": 1.5716204643249512,
-      "learning_rate": 8.75427135678392e-05,
-      "loss": 4.645,
-      "step": 12898
-    },
-    {
-      "epoch": 6.7269882659713165,
-      "grad_norm": 1.5926568508148193,
-      "learning_rate": 8.754170854271358e-05,
-      "loss": 5.5912,
-      "step": 12899
-    },
-    {
-      "epoch": 6.727509778357236,
-      "grad_norm": 1.3775511980056763,
-      "learning_rate": 8.754070351758794e-05,
-      "loss": 5.3362,
-      "step": 12900
-    },
-    {
-      "epoch": 6.728031290743155,
-      "grad_norm": 1.4979419708251953,
-      "learning_rate": 8.753969849246232e-05,
-      "loss": 5.7803,
-      "step": 12901
-    },
-    {
-      "epoch": 6.728552803129074,
-      "grad_norm": 1.529505968093872,
-      "learning_rate": 8.753869346733668e-05,
-      "loss": 5.2302,
-      "step": 12902
-    },
-    {
-      "epoch": 6.7290743155149935,
-      "grad_norm": 1.518815279006958,
-      "learning_rate": 8.753768844221106e-05,
-      "loss": 5.6444,
-      "step": 12903
-    },
-    {
-      "epoch": 6.729595827900913,
-      "grad_norm": 1.6006195545196533,
-      "learning_rate": 8.753668341708543e-05,
-      "loss": 5.3328,
-      "step": 12904
-    },
-    {
-      "epoch": 6.7301173402868315,
-      "grad_norm": 1.432692050933838,
-      "learning_rate": 8.75356783919598e-05,
-      "loss": 5.5458,
-      "step": 12905
-    },
-    {
-      "epoch": 6.730638852672751,
-      "grad_norm": 1.5494927167892456,
-      "learning_rate": 8.753467336683418e-05,
-      "loss": 5.3325,
-      "step": 12906
-    },
-    {
-      "epoch": 6.73116036505867,
-      "grad_norm": 1.7235878705978394,
-      "learning_rate": 8.753366834170856e-05,
-      "loss": 4.7747,
-      "step": 12907
-    },
-    {
-      "epoch": 6.731681877444589,
-      "grad_norm": 1.5326013565063477,
-      "learning_rate": 8.753266331658292e-05,
-      "loss": 5.5954,
-      "step": 12908
-    },
-    {
-      "epoch": 6.7322033898305085,
-      "grad_norm": 1.4726730585098267,
-      "learning_rate": 8.753165829145729e-05,
-      "loss": 5.4155,
-      "step": 12909
-    },
-    {
-      "epoch": 6.732724902216428,
-      "grad_norm": 1.5503407716751099,
-      "learning_rate": 8.753065326633167e-05,
-      "loss": 5.3932,
-      "step": 12910
-    },
-    {
-      "epoch": 6.7332464146023465,
-      "grad_norm": 1.3695964813232422,
-      "learning_rate": 8.752964824120603e-05,
-      "loss": 5.6673,
-      "step": 12911
-    },
-    {
-      "epoch": 6.733767926988266,
-      "grad_norm": 1.41269052028656,
-      "learning_rate": 8.752864321608041e-05,
-      "loss": 6.0206,
-      "step": 12912
-    },
-    {
-      "epoch": 6.7342894393741854,
-      "grad_norm": 1.4139107465744019,
-      "learning_rate": 8.752763819095477e-05,
-      "loss": 5.3379,
-      "step": 12913
-    },
-    {
-      "epoch": 6.734810951760104,
-      "grad_norm": 1.3406486511230469,
-      "learning_rate": 8.752663316582915e-05,
-      "loss": 5.4926,
-      "step": 12914
-    },
-    {
-      "epoch": 6.7353324641460235,
-      "grad_norm": 1.3313837051391602,
-      "learning_rate": 8.752562814070351e-05,
-      "loss": 5.3679,
-      "step": 12915
-    },
-    {
-      "epoch": 6.735853976531943,
-      "grad_norm": 1.3869768381118774,
-      "learning_rate": 8.752462311557789e-05,
-      "loss": 5.4576,
-      "step": 12916
-    },
-    {
-      "epoch": 6.7363754889178615,
-      "grad_norm": 1.4364079236984253,
-      "learning_rate": 8.752361809045227e-05,
-      "loss": 5.5214,
-      "step": 12917
-    },
-    {
-      "epoch": 6.736897001303781,
-      "grad_norm": 1.432847023010254,
-      "learning_rate": 8.752261306532665e-05,
-      "loss": 5.6061,
-      "step": 12918
-    },
-    {
-      "epoch": 6.7374185136897005,
-      "grad_norm": 1.4812605381011963,
-      "learning_rate": 8.752160804020101e-05,
-      "loss": 5.1621,
-      "step": 12919
-    },
-    {
-      "epoch": 6.737940026075619,
-      "grad_norm": 1.572106122970581,
-      "learning_rate": 8.752060301507539e-05,
-      "loss": 5.7396,
-      "step": 12920
-    },
-    {
-      "epoch": 6.7384615384615385,
-      "grad_norm": 1.3806670904159546,
-      "learning_rate": 8.751959798994975e-05,
-      "loss": 5.6612,
-      "step": 12921
-    },
-    {
-      "epoch": 6.738983050847458,
-      "grad_norm": 1.4857136011123657,
-      "learning_rate": 8.751859296482412e-05,
-      "loss": 5.2353,
-      "step": 12922
-    },
-    {
-      "epoch": 6.7395045632333765,
-      "grad_norm": 1.5276367664337158,
-      "learning_rate": 8.75175879396985e-05,
-      "loss": 5.2185,
-      "step": 12923
-    },
-    {
-      "epoch": 6.740026075619296,
-      "grad_norm": 1.446948766708374,
-      "learning_rate": 8.751658291457286e-05,
-      "loss": 5.4536,
-      "step": 12924
-    },
-    {
-      "epoch": 6.7405475880052155,
-      "grad_norm": 1.9266706705093384,
-      "learning_rate": 8.751557788944724e-05,
-      "loss": 5.5783,
-      "step": 12925
-    },
-    {
-      "epoch": 6.741069100391134,
-      "grad_norm": 1.3439608812332153,
-      "learning_rate": 8.751457286432162e-05,
-      "loss": 5.6052,
-      "step": 12926
-    },
-    {
-      "epoch": 6.7415906127770535,
-      "grad_norm": 1.3757380247116089,
-      "learning_rate": 8.7513567839196e-05,
-      "loss": 5.6875,
-      "step": 12927
-    },
-    {
-      "epoch": 6.742112125162973,
-      "grad_norm": 1.377844214439392,
-      "learning_rate": 8.751256281407036e-05,
-      "loss": 5.9908,
-      "step": 12928
-    },
-    {
-      "epoch": 6.742633637548892,
-      "grad_norm": 1.3863346576690674,
-      "learning_rate": 8.751155778894474e-05,
-      "loss": 5.7368,
-      "step": 12929
-    },
-    {
-      "epoch": 6.743155149934811,
-      "grad_norm": 1.2930681705474854,
-      "learning_rate": 8.75105527638191e-05,
-      "loss": 5.8552,
-      "step": 12930
-    },
-    {
-      "epoch": 6.7436766623207305,
-      "grad_norm": 1.4953399896621704,
-      "learning_rate": 8.750954773869348e-05,
-      "loss": 5.6349,
-      "step": 12931
-    },
-    {
-      "epoch": 6.744198174706649,
-      "grad_norm": 1.5454089641571045,
-      "learning_rate": 8.750854271356784e-05,
-      "loss": 5.5474,
-      "step": 12932
-    },
-    {
-      "epoch": 6.7447196870925685,
-      "grad_norm": 1.4068899154663086,
-      "learning_rate": 8.750753768844222e-05,
-      "loss": 5.7806,
-      "step": 12933
-    },
-    {
-      "epoch": 6.745241199478488,
-      "grad_norm": 1.4993643760681152,
-      "learning_rate": 8.750653266331658e-05,
-      "loss": 4.7752,
-      "step": 12934
-    },
-    {
-      "epoch": 6.745762711864407,
-      "grad_norm": 1.5210257768630981,
-      "learning_rate": 8.750552763819095e-05,
-      "loss": 5.4497,
-      "step": 12935
-    },
-    {
-      "epoch": 6.746284224250326,
-      "grad_norm": 1.435760736465454,
-      "learning_rate": 8.750452261306533e-05,
-      "loss": 5.6749,
-      "step": 12936
-    },
-    {
-      "epoch": 6.7468057366362455,
-      "grad_norm": 1.5299052000045776,
-      "learning_rate": 8.75035175879397e-05,
-      "loss": 5.8583,
-      "step": 12937
-    },
-    {
-      "epoch": 6.747327249022164,
-      "grad_norm": 1.3997024297714233,
-      "learning_rate": 8.750251256281408e-05,
-      "loss": 5.4734,
-      "step": 12938
-    },
-    {
-      "epoch": 6.7478487614080835,
-      "grad_norm": 1.3807650804519653,
-      "learning_rate": 8.750150753768845e-05,
-      "loss": 5.8145,
-      "step": 12939
-    },
-    {
-      "epoch": 6.748370273794002,
-      "grad_norm": 1.3790351152420044,
-      "learning_rate": 8.750050251256282e-05,
-      "loss": 5.7283,
-      "step": 12940
-    },
-    {
-      "epoch": 6.748891786179922,
-      "grad_norm": 1.414139986038208,
-      "learning_rate": 8.749949748743719e-05,
-      "loss": 5.8767,
-      "step": 12941
-    },
-    {
-      "epoch": 6.749413298565841,
-      "grad_norm": 1.413427710533142,
-      "learning_rate": 8.749849246231157e-05,
-      "loss": 5.6973,
-      "step": 12942
-    },
-    {
-      "epoch": 6.7499348109517605,
-      "grad_norm": 1.5015395879745483,
-      "learning_rate": 8.749748743718593e-05,
-      "loss": 5.5305,
-      "step": 12943
-    },
-    {
-      "epoch": 6.750456323337679,
-      "grad_norm": 1.5135955810546875,
-      "learning_rate": 8.749648241206031e-05,
-      "loss": 5.6143,
-      "step": 12944
-    },
-    {
-      "epoch": 6.7509778357235986,
-      "grad_norm": 1.4148367643356323,
-      "learning_rate": 8.749547738693467e-05,
-      "loss": 5.4748,
-      "step": 12945
-    },
-    {
-      "epoch": 6.751499348109517,
-      "grad_norm": 1.4204000234603882,
-      "learning_rate": 8.749447236180905e-05,
-      "loss": 5.8322,
-      "step": 12946
-    },
-    {
-      "epoch": 6.752020860495437,
-      "grad_norm": 1.5825470685958862,
-      "learning_rate": 8.749346733668343e-05,
-      "loss": 5.3464,
-      "step": 12947
-    },
-    {
-      "epoch": 6.752542372881356,
-      "grad_norm": 1.4701452255249023,
-      "learning_rate": 8.749246231155779e-05,
-      "loss": 5.5515,
-      "step": 12948
-    },
-    {
-      "epoch": 6.7530638852672755,
-      "grad_norm": 1.4197903871536255,
-      "learning_rate": 8.749145728643217e-05,
-      "loss": 5.6421,
-      "step": 12949
-    },
-    {
-      "epoch": 6.753585397653194,
-      "grad_norm": 1.5351446866989136,
-      "learning_rate": 8.749045226130653e-05,
-      "loss": 5.246,
-      "step": 12950
-    },
-    {
-      "epoch": 6.754106910039114,
-      "grad_norm": 1.487221598625183,
-      "learning_rate": 8.748944723618091e-05,
-      "loss": 5.5169,
-      "step": 12951
-    },
-    {
-      "epoch": 6.754628422425032,
-      "grad_norm": 1.5500482320785522,
-      "learning_rate": 8.748844221105528e-05,
-      "loss": 4.8799,
-      "step": 12952
-    },
-    {
-      "epoch": 6.755149934810952,
-      "grad_norm": 1.5753501653671265,
-      "learning_rate": 8.748743718592965e-05,
-      "loss": 5.6033,
-      "step": 12953
-    },
-    {
-      "epoch": 6.755671447196871,
-      "grad_norm": 1.5642931461334229,
-      "learning_rate": 8.748643216080402e-05,
-      "loss": 5.4088,
-      "step": 12954
-    },
-    {
-      "epoch": 6.7561929595827905,
-      "grad_norm": 1.4574861526489258,
-      "learning_rate": 8.74854271356784e-05,
-      "loss": 5.2381,
-      "step": 12955
-    },
-    {
-      "epoch": 6.756714471968709,
-      "grad_norm": 1.3989100456237793,
-      "learning_rate": 8.748442211055276e-05,
-      "loss": 5.6792,
-      "step": 12956
-    },
-    {
-      "epoch": 6.757235984354629,
-      "grad_norm": 1.4543070793151855,
-      "learning_rate": 8.748341708542714e-05,
-      "loss": 5.8931,
-      "step": 12957
-    },
-    {
-      "epoch": 6.757757496740547,
-      "grad_norm": 1.556498646736145,
-      "learning_rate": 8.748241206030152e-05,
-      "loss": 5.2696,
-      "step": 12958
-    },
-    {
-      "epoch": 6.758279009126467,
-      "grad_norm": 1.520209550857544,
-      "learning_rate": 8.74814070351759e-05,
-      "loss": 5.7003,
-      "step": 12959
-    },
-    {
-      "epoch": 6.758800521512386,
-      "grad_norm": 1.4835176467895508,
-      "learning_rate": 8.748040201005026e-05,
-      "loss": 5.4375,
-      "step": 12960
-    },
-    {
-      "epoch": 6.7593220338983055,
-      "grad_norm": 1.5834152698516846,
-      "learning_rate": 8.747939698492462e-05,
-      "loss": 5.167,
-      "step": 12961
-    },
-    {
-      "epoch": 6.759843546284224,
-      "grad_norm": 1.3461530208587646,
-      "learning_rate": 8.7478391959799e-05,
-      "loss": 5.6189,
-      "step": 12962
-    },
-    {
-      "epoch": 6.760365058670144,
-      "grad_norm": 1.4783391952514648,
-      "learning_rate": 8.747738693467336e-05,
-      "loss": 5.4832,
-      "step": 12963
-    },
-    {
-      "epoch": 6.760886571056062,
-      "grad_norm": 1.547647476196289,
-      "learning_rate": 8.747638190954774e-05,
-      "loss": 5.3697,
-      "step": 12964
-    },
-    {
-      "epoch": 6.761408083441982,
-      "grad_norm": 1.3840641975402832,
-      "learning_rate": 8.747537688442211e-05,
-      "loss": 5.8469,
-      "step": 12965
-    },
-    {
-      "epoch": 6.761929595827901,
-      "grad_norm": 1.3153142929077148,
-      "learning_rate": 8.747437185929648e-05,
-      "loss": 6.0385,
-      "step": 12966
-    },
-    {
-      "epoch": 6.762451108213821,
-      "grad_norm": 1.418047308921814,
-      "learning_rate": 8.747336683417085e-05,
-      "loss": 5.4114,
-      "step": 12967
-    },
-    {
-      "epoch": 6.762972620599739,
-      "grad_norm": 1.3821126222610474,
-      "learning_rate": 8.747236180904523e-05,
-      "loss": 5.7989,
-      "step": 12968
-    },
-    {
-      "epoch": 6.763494132985659,
-      "grad_norm": 1.5547492504119873,
-      "learning_rate": 8.74713567839196e-05,
-      "loss": 5.2473,
-      "step": 12969
-    },
-    {
-      "epoch": 6.764015645371577,
-      "grad_norm": 1.3935418128967285,
-      "learning_rate": 8.747035175879398e-05,
-      "loss": 5.7549,
-      "step": 12970
-    },
-    {
-      "epoch": 6.764537157757497,
-      "grad_norm": 1.4017019271850586,
-      "learning_rate": 8.746934673366835e-05,
-      "loss": 5.5046,
-      "step": 12971
-    },
-    {
-      "epoch": 6.765058670143416,
-      "grad_norm": 1.466787576675415,
-      "learning_rate": 8.746834170854272e-05,
-      "loss": 5.6034,
-      "step": 12972
-    },
-    {
-      "epoch": 6.765580182529335,
-      "grad_norm": 1.405820608139038,
-      "learning_rate": 8.746733668341709e-05,
-      "loss": 5.7176,
-      "step": 12973
-    },
-    {
-      "epoch": 6.766101694915254,
-      "grad_norm": 1.4241139888763428,
-      "learning_rate": 8.746633165829145e-05,
-      "loss": 5.7412,
-      "step": 12974
-    },
-    {
-      "epoch": 6.766623207301174,
-      "grad_norm": 1.5310122966766357,
-      "learning_rate": 8.746532663316583e-05,
-      "loss": 5.049,
-      "step": 12975
-    },
-    {
-      "epoch": 6.767144719687092,
-      "grad_norm": 1.5013501644134521,
-      "learning_rate": 8.74643216080402e-05,
-      "loss": 5.6641,
-      "step": 12976
-    },
-    {
-      "epoch": 6.767666232073012,
-      "grad_norm": 1.8976328372955322,
-      "learning_rate": 8.746331658291457e-05,
-      "loss": 5.3523,
-      "step": 12977
-    },
-    {
-      "epoch": 6.768187744458931,
-      "grad_norm": 1.2967411279678345,
-      "learning_rate": 8.746231155778895e-05,
-      "loss": 5.9684,
-      "step": 12978
-    },
-    {
-      "epoch": 6.76870925684485,
-      "grad_norm": 1.5587066411972046,
-      "learning_rate": 8.746130653266333e-05,
-      "loss": 5.2376,
-      "step": 12979
-    },
-    {
-      "epoch": 6.769230769230769,
-      "grad_norm": 1.3736681938171387,
-      "learning_rate": 8.746030150753769e-05,
-      "loss": 5.9197,
-      "step": 12980
-    },
-    {
-      "epoch": 6.769752281616689,
-      "grad_norm": 1.3633453845977783,
-      "learning_rate": 8.745929648241207e-05,
-      "loss": 5.603,
-      "step": 12981
-    },
-    {
-      "epoch": 6.770273794002607,
-      "grad_norm": 1.431043267250061,
-      "learning_rate": 8.745829145728644e-05,
-      "loss": 5.597,
-      "step": 12982
-    },
-    {
-      "epoch": 6.770795306388527,
-      "grad_norm": 1.4546245336532593,
-      "learning_rate": 8.745728643216081e-05,
-      "loss": 5.6864,
-      "step": 12983
-    },
-    {
-      "epoch": 6.771316818774446,
-      "grad_norm": 1.4225929975509644,
-      "learning_rate": 8.745628140703518e-05,
-      "loss": 5.556,
-      "step": 12984
-    },
-    {
-      "epoch": 6.771838331160365,
-      "grad_norm": 1.413989782333374,
-      "learning_rate": 8.745527638190955e-05,
-      "loss": 5.9142,
-      "step": 12985
-    },
-    {
-      "epoch": 6.772359843546284,
-      "grad_norm": 1.3322391510009766,
-      "learning_rate": 8.745427135678392e-05,
-      "loss": 5.9289,
-      "step": 12986
-    },
-    {
-      "epoch": 6.772881355932204,
-      "grad_norm": 1.3640295267105103,
-      "learning_rate": 8.74532663316583e-05,
-      "loss": 5.7926,
-      "step": 12987
-    },
-    {
-      "epoch": 6.773402868318122,
-      "grad_norm": 1.5264508724212646,
-      "learning_rate": 8.745226130653266e-05,
-      "loss": 5.3824,
-      "step": 12988
-    },
-    {
-      "epoch": 6.773924380704042,
-      "grad_norm": 1.8715994358062744,
-      "learning_rate": 8.745125628140704e-05,
-      "loss": 5.2562,
-      "step": 12989
-    },
-    {
-      "epoch": 6.774445893089961,
-      "grad_norm": 1.4258023500442505,
-      "learning_rate": 8.745025125628142e-05,
-      "loss": 5.6618,
-      "step": 12990
-    },
-    {
-      "epoch": 6.77496740547588,
-      "grad_norm": 1.7969495058059692,
-      "learning_rate": 8.744924623115578e-05,
-      "loss": 5.6377,
-      "step": 12991
-    },
-    {
-      "epoch": 6.775488917861799,
-      "grad_norm": 1.3364239931106567,
-      "learning_rate": 8.744824120603016e-05,
-      "loss": 5.5466,
-      "step": 12992
-    },
-    {
-      "epoch": 6.776010430247719,
-      "grad_norm": 1.4687128067016602,
-      "learning_rate": 8.744723618090452e-05,
-      "loss": 5.7711,
-      "step": 12993
-    },
-    {
-      "epoch": 6.776531942633637,
-      "grad_norm": 1.484465479850769,
-      "learning_rate": 8.74462311557789e-05,
-      "loss": 5.7482,
-      "step": 12994
-    },
-    {
-      "epoch": 6.777053455019557,
-      "grad_norm": 1.5390015840530396,
-      "learning_rate": 8.744522613065327e-05,
-      "loss": 5.0742,
-      "step": 12995
-    },
-    {
-      "epoch": 6.777574967405476,
-      "grad_norm": 1.407578945159912,
-      "learning_rate": 8.744422110552764e-05,
-      "loss": 5.764,
-      "step": 12996
-    },
-    {
-      "epoch": 6.778096479791395,
-      "grad_norm": 1.3707042932510376,
-      "learning_rate": 8.744321608040201e-05,
-      "loss": 5.9204,
-      "step": 12997
-    },
-    {
-      "epoch": 6.778617992177314,
-      "grad_norm": 1.3791706562042236,
-      "learning_rate": 8.744221105527639e-05,
-      "loss": 5.2279,
-      "step": 12998
-    },
-    {
-      "epoch": 6.779139504563234,
-      "grad_norm": 1.417040467262268,
-      "learning_rate": 8.744120603015076e-05,
-      "loss": 5.4195,
-      "step": 12999
-    },
-    {
-      "epoch": 6.779661016949152,
-      "grad_norm": 1.5586284399032593,
-      "learning_rate": 8.744020100502514e-05,
-      "loss": 5.4982,
-      "step": 13000
-    },
-    {
-      "epoch": 6.780182529335072,
-      "grad_norm": 1.5421696901321411,
-      "learning_rate": 8.74391959798995e-05,
-      "loss": 5.245,
-      "step": 13001
-    },
-    {
-      "epoch": 6.780704041720991,
-      "grad_norm": 1.4511457681655884,
-      "learning_rate": 8.743819095477387e-05,
-      "loss": 5.4794,
-      "step": 13002
-    },
-    {
-      "epoch": 6.78122555410691,
-      "grad_norm": 1.533961296081543,
-      "learning_rate": 8.743718592964825e-05,
-      "loss": 5.7046,
-      "step": 13003
-    },
-    {
-      "epoch": 6.781747066492829,
-      "grad_norm": 2.3832619190216064,
-      "learning_rate": 8.743618090452261e-05,
-      "loss": 5.0649,
-      "step": 13004
-    },
-    {
-      "epoch": 6.782268578878749,
-      "grad_norm": 1.5229096412658691,
-      "learning_rate": 8.743517587939699e-05,
-      "loss": 5.4284,
-      "step": 13005
-    },
-    {
-      "epoch": 6.782790091264667,
-      "grad_norm": 1.560542106628418,
-      "learning_rate": 8.743417085427135e-05,
-      "loss": 5.2422,
-      "step": 13006
-    },
-    {
-      "epoch": 6.783311603650587,
-      "grad_norm": 1.4768189191818237,
-      "learning_rate": 8.743316582914573e-05,
-      "loss": 5.6699,
-      "step": 13007
-    },
-    {
-      "epoch": 6.783833116036506,
-      "grad_norm": 1.5079182386398315,
-      "learning_rate": 8.74321608040201e-05,
-      "loss": 5.6358,
-      "step": 13008
-    },
-    {
-      "epoch": 6.784354628422425,
-      "grad_norm": 1.4047681093215942,
-      "learning_rate": 8.743115577889447e-05,
-      "loss": 5.3879,
-      "step": 13009
-    },
-    {
-      "epoch": 6.784876140808344,
-      "grad_norm": 1.4878028631210327,
-      "learning_rate": 8.743015075376885e-05,
-      "loss": 5.4708,
-      "step": 13010
-    },
-    {
-      "epoch": 6.785397653194264,
-      "grad_norm": 1.3969591856002808,
-      "learning_rate": 8.742914572864323e-05,
-      "loss": 5.5117,
-      "step": 13011
-    },
-    {
-      "epoch": 6.785919165580182,
-      "grad_norm": 1.5361905097961426,
-      "learning_rate": 8.74281407035176e-05,
-      "loss": 5.6062,
-      "step": 13012
-    },
-    {
-      "epoch": 6.786440677966102,
-      "grad_norm": 1.4069421291351318,
-      "learning_rate": 8.742713567839197e-05,
-      "loss": 6.0374,
-      "step": 13013
-    },
-    {
-      "epoch": 6.786962190352021,
-      "grad_norm": 1.290174961090088,
-      "learning_rate": 8.742613065326634e-05,
-      "loss": 5.8874,
-      "step": 13014
-    },
-    {
-      "epoch": 6.78748370273794,
-      "grad_norm": 1.3905510902404785,
-      "learning_rate": 8.74251256281407e-05,
-      "loss": 5.6522,
-      "step": 13015
-    },
-    {
-      "epoch": 6.788005215123859,
-      "grad_norm": 1.504899501800537,
-      "learning_rate": 8.742412060301508e-05,
-      "loss": 5.6411,
-      "step": 13016
-    },
-    {
-      "epoch": 6.788526727509779,
-      "grad_norm": 1.5551254749298096,
-      "learning_rate": 8.742311557788944e-05,
-      "loss": 5.0454,
-      "step": 13017
-    },
-    {
-      "epoch": 6.789048239895697,
-      "grad_norm": 1.402003526687622,
-      "learning_rate": 8.742211055276382e-05,
-      "loss": 5.6868,
-      "step": 13018
-    },
-    {
-      "epoch": 6.789569752281617,
-      "grad_norm": 1.5429739952087402,
-      "learning_rate": 8.74211055276382e-05,
-      "loss": 5.7435,
-      "step": 13019
-    },
-    {
-      "epoch": 6.790091264667536,
-      "grad_norm": 1.4808398485183716,
-      "learning_rate": 8.742010050251258e-05,
-      "loss": 5.3817,
-      "step": 13020
-    },
-    {
-      "epoch": 6.790612777053455,
-      "grad_norm": 1.4027894735336304,
-      "learning_rate": 8.741909547738694e-05,
-      "loss": 5.768,
-      "step": 13021
-    },
-    {
-      "epoch": 6.791134289439374,
-      "grad_norm": 1.3579949140548706,
-      "learning_rate": 8.741809045226132e-05,
-      "loss": 5.853,
-      "step": 13022
-    },
-    {
-      "epoch": 6.791655801825294,
-      "grad_norm": 1.3913379907608032,
-      "learning_rate": 8.741708542713568e-05,
-      "loss": 5.5749,
-      "step": 13023
-    },
-    {
-      "epoch": 6.792177314211212,
-      "grad_norm": 1.5369387865066528,
-      "learning_rate": 8.741608040201006e-05,
-      "loss": 5.3369,
-      "step": 13024
-    },
-    {
-      "epoch": 6.792698826597132,
-      "grad_norm": 1.3365228176116943,
-      "learning_rate": 8.741507537688442e-05,
-      "loss": 4.9083,
-      "step": 13025
-    },
-    {
-      "epoch": 6.793220338983051,
-      "grad_norm": 1.3750261068344116,
-      "learning_rate": 8.74140703517588e-05,
-      "loss": 5.8712,
-      "step": 13026
-    },
-    {
-      "epoch": 6.79374185136897,
-      "grad_norm": 1.5418564081192017,
-      "learning_rate": 8.741306532663317e-05,
-      "loss": 5.6122,
-      "step": 13027
-    },
-    {
-      "epoch": 6.794263363754889,
-      "grad_norm": 1.4683575630187988,
-      "learning_rate": 8.741206030150753e-05,
-      "loss": 5.1398,
-      "step": 13028
-    },
-    {
-      "epoch": 6.794784876140809,
-      "grad_norm": 1.4433772563934326,
-      "learning_rate": 8.741105527638191e-05,
-      "loss": 5.2609,
-      "step": 13029
-    },
-    {
-      "epoch": 6.795306388526727,
-      "grad_norm": 1.3538075685501099,
-      "learning_rate": 8.741005025125629e-05,
-      "loss": 5.8461,
-      "step": 13030
-    },
-    {
-      "epoch": 6.795827900912647,
-      "grad_norm": 1.7590866088867188,
-      "learning_rate": 8.740904522613066e-05,
-      "loss": 4.9492,
-      "step": 13031
-    },
-    {
-      "epoch": 6.796349413298566,
-      "grad_norm": 1.3921585083007812,
-      "learning_rate": 8.740804020100503e-05,
-      "loss": 5.4751,
-      "step": 13032
-    },
-    {
-      "epoch": 6.796870925684485,
-      "grad_norm": 1.394156575202942,
-      "learning_rate": 8.74070351758794e-05,
-      "loss": 6.0373,
-      "step": 13033
-    },
-    {
-      "epoch": 6.797392438070404,
-      "grad_norm": 1.454831838607788,
-      "learning_rate": 8.740603015075377e-05,
-      "loss": 5.6624,
-      "step": 13034
-    },
-    {
-      "epoch": 6.797913950456323,
-      "grad_norm": 1.3722530603408813,
-      "learning_rate": 8.740502512562815e-05,
-      "loss": 5.1511,
-      "step": 13035
-    },
-    {
-      "epoch": 6.798435462842242,
-      "grad_norm": 1.4915679693222046,
-      "learning_rate": 8.740402010050251e-05,
-      "loss": 5.3913,
-      "step": 13036
-    },
-    {
-      "epoch": 6.798956975228162,
-      "grad_norm": 1.481310248374939,
-      "learning_rate": 8.740301507537689e-05,
-      "loss": 5.7484,
-      "step": 13037
-    },
-    {
-      "epoch": 6.799478487614081,
-      "grad_norm": 1.4843041896820068,
-      "learning_rate": 8.740201005025125e-05,
-      "loss": 5.5584,
-      "step": 13038
-    },
-    {
-      "epoch": 6.8,
-      "grad_norm": 1.4647318124771118,
-      "learning_rate": 8.740100502512563e-05,
-      "loss": 5.5884,
-      "step": 13039
-    },
-    {
-      "epoch": 6.800521512385919,
-      "grad_norm": 1.444783329963684,
-      "learning_rate": 8.740000000000001e-05,
-      "loss": 5.9094,
-      "step": 13040
-    },
-    {
-      "epoch": 6.801043024771838,
-      "grad_norm": 1.4303377866744995,
-      "learning_rate": 8.739899497487437e-05,
-      "loss": 5.5005,
-      "step": 13041
-    },
-    {
-      "epoch": 6.801564537157757,
-      "grad_norm": 1.4127198457717896,
-      "learning_rate": 8.739798994974875e-05,
-      "loss": 5.465,
-      "step": 13042
-    },
-    {
-      "epoch": 6.802086049543677,
-      "grad_norm": 1.520289421081543,
-      "learning_rate": 8.739698492462312e-05,
-      "loss": 5.6286,
-      "step": 13043
-    },
-    {
-      "epoch": 6.802607561929596,
-      "grad_norm": 1.4306286573410034,
-      "learning_rate": 8.73959798994975e-05,
-      "loss": 5.452,
-      "step": 13044
-    },
-    {
-      "epoch": 6.803129074315515,
-      "grad_norm": 1.5512374639511108,
-      "learning_rate": 8.739497487437186e-05,
-      "loss": 5.7288,
-      "step": 13045
-    },
-    {
-      "epoch": 6.803650586701434,
-      "grad_norm": 1.3893169164657593,
-      "learning_rate": 8.739396984924624e-05,
-      "loss": 5.7595,
-      "step": 13046
-    },
-    {
-      "epoch": 6.804172099087353,
-      "grad_norm": 1.4788779020309448,
-      "learning_rate": 8.73929648241206e-05,
-      "loss": 5.7912,
-      "step": 13047
-    },
-    {
-      "epoch": 6.804693611473272,
-      "grad_norm": 1.4758238792419434,
-      "learning_rate": 8.739195979899498e-05,
-      "loss": 5.6745,
-      "step": 13048
-    },
-    {
-      "epoch": 6.805215123859192,
-      "grad_norm": 1.554914951324463,
-      "learning_rate": 8.739095477386934e-05,
-      "loss": 5.2051,
-      "step": 13049
-    },
-    {
-      "epoch": 6.805736636245111,
-      "grad_norm": 1.3898274898529053,
-      "learning_rate": 8.738994974874372e-05,
-      "loss": 5.7279,
-      "step": 13050
-    },
-    {
-      "epoch": 6.80625814863103,
-      "grad_norm": 1.4800206422805786,
-      "learning_rate": 8.73889447236181e-05,
-      "loss": 5.6009,
-      "step": 13051
-    },
-    {
-      "epoch": 6.806779661016949,
-      "grad_norm": 1.4042654037475586,
-      "learning_rate": 8.738793969849248e-05,
-      "loss": 5.8474,
-      "step": 13052
-    },
-    {
-      "epoch": 6.807301173402868,
-      "grad_norm": 1.4371329545974731,
-      "learning_rate": 8.738693467336684e-05,
-      "loss": 5.4417,
-      "step": 13053
-    },
-    {
-      "epoch": 6.807822685788787,
-      "grad_norm": 1.5045546293258667,
-      "learning_rate": 8.73859296482412e-05,
-      "loss": 5.4124,
-      "step": 13054
-    },
-    {
-      "epoch": 6.808344198174707,
-      "grad_norm": 1.6229749917984009,
-      "learning_rate": 8.738492462311558e-05,
-      "loss": 4.8445,
-      "step": 13055
-    },
-    {
-      "epoch": 6.808865710560626,
-      "grad_norm": 1.2861509323120117,
-      "learning_rate": 8.738391959798995e-05,
-      "loss": 5.0374,
-      "step": 13056
-    },
-    {
-      "epoch": 6.809387222946545,
-      "grad_norm": 1.3762145042419434,
-      "learning_rate": 8.738291457286432e-05,
-      "loss": 5.8783,
-      "step": 13057
-    },
-    {
-      "epoch": 6.809908735332464,
-      "grad_norm": 1.5025694370269775,
-      "learning_rate": 8.738190954773869e-05,
-      "loss": 5.5861,
-      "step": 13058
-    },
-    {
-      "epoch": 6.810430247718383,
-      "grad_norm": 1.3696616888046265,
-      "learning_rate": 8.738090452261307e-05,
-      "loss": 5.7225,
-      "step": 13059
-    },
-    {
-      "epoch": 6.810951760104302,
-      "grad_norm": 1.5139281749725342,
-      "learning_rate": 8.737989949748744e-05,
-      "loss": 5.4498,
-      "step": 13060
-    },
-    {
-      "epoch": 6.811473272490222,
-      "grad_norm": 1.5652334690093994,
-      "learning_rate": 8.737889447236182e-05,
-      "loss": 5.3399,
-      "step": 13061
-    },
-    {
-      "epoch": 6.811994784876141,
-      "grad_norm": 1.4104660749435425,
-      "learning_rate": 8.737788944723619e-05,
-      "loss": 5.6597,
-      "step": 13062
-    },
-    {
-      "epoch": 6.81251629726206,
-      "grad_norm": 1.5945795774459839,
-      "learning_rate": 8.737688442211056e-05,
-      "loss": 5.4533,
-      "step": 13063
-    },
-    {
-      "epoch": 6.813037809647979,
-      "grad_norm": 1.4883118867874146,
-      "learning_rate": 8.737587939698493e-05,
-      "loss": 5.7554,
-      "step": 13064
-    },
-    {
-      "epoch": 6.813559322033898,
-      "grad_norm": 1.5934516191482544,
-      "learning_rate": 8.737487437185931e-05,
-      "loss": 5.2979,
-      "step": 13065
-    },
-    {
-      "epoch": 6.814080834419817,
-      "grad_norm": 1.4103431701660156,
-      "learning_rate": 8.737386934673367e-05,
-      "loss": 5.7426,
-      "step": 13066
-    },
-    {
-      "epoch": 6.814602346805737,
-      "grad_norm": 1.4790726900100708,
-      "learning_rate": 8.737286432160804e-05,
-      "loss": 5.495,
-      "step": 13067
-    },
-    {
-      "epoch": 6.815123859191655,
-      "grad_norm": 1.421108365058899,
-      "learning_rate": 8.737185929648241e-05,
-      "loss": 5.6065,
-      "step": 13068
-    },
-    {
-      "epoch": 6.815645371577575,
-      "grad_norm": 1.347221851348877,
-      "learning_rate": 8.737085427135678e-05,
-      "loss": 5.8322,
-      "step": 13069
-    },
-    {
-      "epoch": 6.816166883963494,
-      "grad_norm": 1.4749139547348022,
-      "learning_rate": 8.736984924623116e-05,
-      "loss": 5.6167,
-      "step": 13070
-    },
-    {
-      "epoch": 6.816688396349413,
-      "grad_norm": 1.4353291988372803,
-      "learning_rate": 8.736884422110553e-05,
-      "loss": 5.4377,
-      "step": 13071
-    },
-    {
-      "epoch": 6.817209908735332,
-      "grad_norm": 1.2957545518875122,
-      "learning_rate": 8.736783919597991e-05,
-      "loss": 5.8858,
-      "step": 13072
-    },
-    {
-      "epoch": 6.817731421121252,
-      "grad_norm": 1.396743655204773,
-      "learning_rate": 8.736683417085428e-05,
-      "loss": 5.3925,
-      "step": 13073
-    },
-    {
-      "epoch": 6.81825293350717,
-      "grad_norm": 1.5138754844665527,
-      "learning_rate": 8.736582914572865e-05,
-      "loss": 5.572,
-      "step": 13074
-    },
-    {
-      "epoch": 6.81877444589309,
-      "grad_norm": 1.4415092468261719,
-      "learning_rate": 8.736482412060302e-05,
-      "loss": 5.6987,
-      "step": 13075
-    },
-    {
-      "epoch": 6.819295958279009,
-      "grad_norm": 1.4668790102005005,
-      "learning_rate": 8.73638190954774e-05,
-      "loss": 5.3156,
-      "step": 13076
-    },
-    {
-      "epoch": 6.819817470664928,
-      "grad_norm": 1.4260199069976807,
-      "learning_rate": 8.736281407035176e-05,
-      "loss": 5.6959,
-      "step": 13077
-    },
-    {
-      "epoch": 6.820338983050847,
-      "grad_norm": 1.6572165489196777,
-      "learning_rate": 8.736180904522614e-05,
-      "loss": 5.6058,
-      "step": 13078
-    },
-    {
-      "epoch": 6.820860495436767,
-      "grad_norm": 1.5033843517303467,
-      "learning_rate": 8.73608040201005e-05,
-      "loss": 5.5271,
-      "step": 13079
-    },
-    {
-      "epoch": 6.8213820078226854,
-      "grad_norm": 1.427453637123108,
-      "learning_rate": 8.735979899497488e-05,
-      "loss": 5.3157,
-      "step": 13080
-    },
-    {
-      "epoch": 6.821903520208605,
-      "grad_norm": 1.3666183948516846,
-      "learning_rate": 8.735879396984926e-05,
-      "loss": 5.8858,
-      "step": 13081
-    },
-    {
-      "epoch": 6.822425032594524,
-      "grad_norm": 1.4239518642425537,
-      "learning_rate": 8.735778894472362e-05,
-      "loss": 5.3818,
-      "step": 13082
-    },
-    {
-      "epoch": 6.822946544980443,
-      "grad_norm": 1.3181743621826172,
-      "learning_rate": 8.7356783919598e-05,
-      "loss": 5.7137,
-      "step": 13083
-    },
-    {
-      "epoch": 6.823468057366362,
-      "grad_norm": 1.5173778533935547,
-      "learning_rate": 8.735577889447236e-05,
-      "loss": 5.3016,
-      "step": 13084
-    },
-    {
-      "epoch": 6.823989569752282,
-      "grad_norm": 1.4261589050292969,
-      "learning_rate": 8.735477386934674e-05,
-      "loss": 5.5066,
-      "step": 13085
-    },
-    {
-      "epoch": 6.8245110821382005,
-      "grad_norm": 1.3882603645324707,
-      "learning_rate": 8.73537688442211e-05,
-      "loss": 5.5096,
-      "step": 13086
-    },
-    {
-      "epoch": 6.82503259452412,
-      "grad_norm": 1.5649189949035645,
-      "learning_rate": 8.735276381909548e-05,
-      "loss": 5.1197,
-      "step": 13087
-    },
-    {
-      "epoch": 6.825554106910039,
-      "grad_norm": 1.3453960418701172,
-      "learning_rate": 8.735175879396985e-05,
-      "loss": 5.7102,
-      "step": 13088
-    },
-    {
-      "epoch": 6.826075619295958,
-      "grad_norm": 1.4414796829223633,
-      "learning_rate": 8.735075376884423e-05,
-      "loss": 5.4909,
-      "step": 13089
-    },
-    {
-      "epoch": 6.826597131681877,
-      "grad_norm": 1.4232909679412842,
-      "learning_rate": 8.734974874371859e-05,
-      "loss": 5.5597,
-      "step": 13090
-    },
-    {
-      "epoch": 6.827118644067797,
-      "grad_norm": 1.464868426322937,
-      "learning_rate": 8.734874371859297e-05,
-      "loss": 5.0611,
-      "step": 13091
-    },
-    {
-      "epoch": 6.8276401564537155,
-      "grad_norm": 1.4110345840454102,
-      "learning_rate": 8.734773869346735e-05,
-      "loss": 5.8023,
-      "step": 13092
-    },
-    {
-      "epoch": 6.828161668839635,
-      "grad_norm": 1.4148502349853516,
-      "learning_rate": 8.734673366834172e-05,
-      "loss": 5.6084,
-      "step": 13093
-    },
-    {
-      "epoch": 6.828683181225554,
-      "grad_norm": 1.4783415794372559,
-      "learning_rate": 8.734572864321609e-05,
-      "loss": 5.6349,
-      "step": 13094
-    },
-    {
-      "epoch": 6.829204693611473,
-      "grad_norm": 1.4886547327041626,
-      "learning_rate": 8.734472361809045e-05,
-      "loss": 5.4447,
-      "step": 13095
-    },
-    {
-      "epoch": 6.829726205997392,
-      "grad_norm": 1.335707187652588,
-      "learning_rate": 8.734371859296483e-05,
-      "loss": 5.9977,
-      "step": 13096
-    },
-    {
-      "epoch": 6.830247718383312,
-      "grad_norm": 1.4298427104949951,
-      "learning_rate": 8.73427135678392e-05,
-      "loss": 5.8318,
-      "step": 13097
-    },
-    {
-      "epoch": 6.8307692307692305,
-      "grad_norm": 1.414836049079895,
-      "learning_rate": 8.734170854271357e-05,
-      "loss": 5.7182,
-      "step": 13098
-    },
-    {
-      "epoch": 6.83129074315515,
-      "grad_norm": 1.4735349416732788,
-      "learning_rate": 8.734070351758794e-05,
-      "loss": 5.626,
-      "step": 13099
-    },
-    {
-      "epoch": 6.831812255541069,
-      "grad_norm": 1.4156290292739868,
-      "learning_rate": 8.733969849246231e-05,
-      "loss": 5.6918,
-      "step": 13100
-    },
-    {
-      "epoch": 6.832333767926988,
-      "grad_norm": 1.4047309160232544,
-      "learning_rate": 8.733869346733669e-05,
-      "loss": 5.5215,
-      "step": 13101
-    },
-    {
-      "epoch": 6.8328552803129075,
-      "grad_norm": 1.494475245475769,
-      "learning_rate": 8.733768844221107e-05,
-      "loss": 5.6677,
-      "step": 13102
-    },
-    {
-      "epoch": 6.833376792698827,
-      "grad_norm": 1.7289236783981323,
-      "learning_rate": 8.733668341708543e-05,
-      "loss": 4.7159,
-      "step": 13103
-    },
-    {
-      "epoch": 6.8338983050847455,
-      "grad_norm": 1.4428136348724365,
-      "learning_rate": 8.733567839195981e-05,
-      "loss": 5.8102,
-      "step": 13104
-    },
-    {
-      "epoch": 6.834419817470665,
-      "grad_norm": 1.3666611909866333,
-      "learning_rate": 8.733467336683418e-05,
-      "loss": 5.7284,
-      "step": 13105
-    },
-    {
-      "epoch": 6.834941329856584,
-      "grad_norm": 1.3787580728530884,
-      "learning_rate": 8.733366834170855e-05,
-      "loss": 5.5316,
-      "step": 13106
-    },
-    {
-      "epoch": 6.835462842242503,
-      "grad_norm": 1.4141874313354492,
-      "learning_rate": 8.733266331658292e-05,
-      "loss": 5.7226,
-      "step": 13107
-    },
-    {
-      "epoch": 6.8359843546284225,
-      "grad_norm": 1.5294097661972046,
-      "learning_rate": 8.733165829145728e-05,
-      "loss": 4.9287,
-      "step": 13108
-    },
-    {
-      "epoch": 6.836505867014342,
-      "grad_norm": 1.4488030672073364,
-      "learning_rate": 8.733065326633166e-05,
-      "loss": 5.3022,
-      "step": 13109
-    },
-    {
-      "epoch": 6.8370273794002605,
-      "grad_norm": 1.5019017457962036,
-      "learning_rate": 8.732964824120602e-05,
-      "loss": 5.5886,
-      "step": 13110
-    },
-    {
-      "epoch": 6.83754889178618,
-      "grad_norm": 1.4181054830551147,
-      "learning_rate": 8.73286432160804e-05,
-      "loss": 5.9906,
-      "step": 13111
-    },
-    {
-      "epoch": 6.838070404172099,
-      "grad_norm": 1.4252159595489502,
-      "learning_rate": 8.732763819095478e-05,
-      "loss": 5.7324,
-      "step": 13112
-    },
-    {
-      "epoch": 6.838591916558018,
-      "grad_norm": 1.4503065347671509,
-      "learning_rate": 8.732663316582916e-05,
-      "loss": 5.6754,
-      "step": 13113
-    },
-    {
-      "epoch": 6.8391134289439375,
-      "grad_norm": 1.3199219703674316,
-      "learning_rate": 8.732562814070352e-05,
-      "loss": 5.9995,
-      "step": 13114
-    },
-    {
-      "epoch": 6.839634941329857,
-      "grad_norm": 1.496468186378479,
-      "learning_rate": 8.73246231155779e-05,
-      "loss": 5.1386,
-      "step": 13115
-    },
-    {
-      "epoch": 6.8401564537157755,
-      "grad_norm": 1.4446656703948975,
-      "learning_rate": 8.732361809045226e-05,
-      "loss": 5.0613,
-      "step": 13116
-    },
-    {
-      "epoch": 6.840677966101695,
-      "grad_norm": 1.3997827768325806,
-      "learning_rate": 8.732261306532664e-05,
-      "loss": 5.2641,
-      "step": 13117
-    },
-    {
-      "epoch": 6.8411994784876144,
-      "grad_norm": Infinity,
-      "learning_rate": 8.732261306532664e-05,
-      "loss": 5.361,
-      "step": 13118
-    },
-    {
-      "epoch": 6.841720990873533,
-      "grad_norm": 1.3405433893203735,
-      "learning_rate": 8.7321608040201e-05,
-      "loss": 5.6263,
-      "step": 13119
-    },
-    {
-      "epoch": 6.8422425032594525,
-      "grad_norm": 1.3896105289459229,
-      "learning_rate": 8.732060301507538e-05,
-      "loss": 5.8165,
-      "step": 13120
-    },
-    {
-      "epoch": 6.842764015645372,
-      "grad_norm": 1.392599105834961,
-      "learning_rate": 8.731959798994975e-05,
-      "loss": 5.7815,
-      "step": 13121
-    },
-    {
-      "epoch": 6.8432855280312905,
-      "grad_norm": 1.6999871730804443,
-      "learning_rate": 8.731859296482413e-05,
-      "loss": 5.5647,
-      "step": 13122
-    },
-    {
-      "epoch": 6.84380704041721,
-      "grad_norm": 1.4028860330581665,
-      "learning_rate": 8.73175879396985e-05,
-      "loss": 5.2894,
-      "step": 13123
-    },
-    {
-      "epoch": 6.8443285528031295,
-      "grad_norm": 1.4749451875686646,
-      "learning_rate": 8.731658291457287e-05,
-      "loss": 6.0361,
-      "step": 13124
-    },
-    {
-      "epoch": 6.844850065189048,
-      "grad_norm": 1.3465672731399536,
-      "learning_rate": 8.731557788944725e-05,
-      "loss": 6.0044,
-      "step": 13125
-    },
-    {
-      "epoch": 6.8453715775749675,
-      "grad_norm": 1.495519757270813,
-      "learning_rate": 8.731457286432161e-05,
-      "loss": 5.6031,
-      "step": 13126
-    },
-    {
-      "epoch": 6.845893089960887,
-      "grad_norm": 1.4587502479553223,
-      "learning_rate": 8.731356783919599e-05,
-      "loss": 5.5831,
-      "step": 13127
-    },
-    {
-      "epoch": 6.8464146023468055,
-      "grad_norm": 1.5295110940933228,
-      "learning_rate": 8.731256281407035e-05,
-      "loss": 5.8104,
-      "step": 13128
-    },
-    {
-      "epoch": 6.846936114732725,
-      "grad_norm": 1.461858868598938,
-      "learning_rate": 8.731155778894473e-05,
-      "loss": 4.9841,
-      "step": 13129
-    },
-    {
-      "epoch": 6.847457627118644,
-      "grad_norm": 1.4394958019256592,
-      "learning_rate": 8.73105527638191e-05,
-      "loss": 5.5966,
-      "step": 13130
-    },
-    {
-      "epoch": 6.847979139504563,
-      "grad_norm": 1.6307897567749023,
-      "learning_rate": 8.730954773869347e-05,
-      "loss": 5.3414,
-      "step": 13131
-    },
-    {
-      "epoch": 6.8485006518904825,
-      "grad_norm": 1.3828166723251343,
-      "learning_rate": 8.730854271356784e-05,
-      "loss": 5.4863,
-      "step": 13132
-    },
-    {
-      "epoch": 6.849022164276402,
-      "grad_norm": 1.5379228591918945,
-      "learning_rate": 8.730753768844221e-05,
-      "loss": 5.478,
-      "step": 13133
-    },
-    {
-      "epoch": 6.849543676662321,
-      "grad_norm": 1.4205949306488037,
-      "learning_rate": 8.730653266331659e-05,
-      "loss": 5.4469,
-      "step": 13134
-    },
-    {
-      "epoch": 6.85006518904824,
-      "grad_norm": 1.390470266342163,
-      "learning_rate": 8.730552763819096e-05,
-      "loss": 5.7008,
-      "step": 13135
-    },
-    {
-      "epoch": 6.850586701434159,
-      "grad_norm": 1.335784673690796,
-      "learning_rate": 8.730452261306533e-05,
-      "loss": 5.9986,
-      "step": 13136
-    },
-    {
-      "epoch": 6.851108213820078,
-      "grad_norm": 1.372817039489746,
-      "learning_rate": 8.73035175879397e-05,
-      "loss": 5.7393,
-      "step": 13137
-    },
-    {
-      "epoch": 6.8516297262059975,
-      "grad_norm": 1.3501542806625366,
-      "learning_rate": 8.730251256281408e-05,
-      "loss": 5.6741,
-      "step": 13138
-    },
-    {
-      "epoch": 6.852151238591917,
-      "grad_norm": 1.5561789274215698,
-      "learning_rate": 8.730150753768844e-05,
-      "loss": 5.5,
-      "step": 13139
-    },
-    {
-      "epoch": 6.852672750977836,
-      "grad_norm": 1.3708488941192627,
-      "learning_rate": 8.730050251256282e-05,
-      "loss": 5.6536,
-      "step": 13140
-    },
-    {
-      "epoch": 6.853194263363755,
-      "grad_norm": 1.3966197967529297,
-      "learning_rate": 8.729949748743718e-05,
-      "loss": 5.3566,
-      "step": 13141
-    },
-    {
-      "epoch": 6.853715775749674,
-      "grad_norm": 1.4969977140426636,
-      "learning_rate": 8.729849246231156e-05,
-      "loss": 5.0373,
-      "step": 13142
-    },
-    {
-      "epoch": 6.854237288135593,
-      "grad_norm": 1.532609462738037,
-      "learning_rate": 8.729748743718593e-05,
-      "loss": 5.8968,
-      "step": 13143
-    },
-    {
-      "epoch": 6.8547588005215125,
-      "grad_norm": 1.4303885698318481,
-      "learning_rate": 8.72964824120603e-05,
-      "loss": 4.8713,
-      "step": 13144
-    },
-    {
-      "epoch": 6.855280312907432,
-      "grad_norm": 1.4966503381729126,
-      "learning_rate": 8.729547738693468e-05,
-      "loss": 5.9034,
-      "step": 13145
-    },
-    {
-      "epoch": 6.855801825293351,
-      "grad_norm": 1.4323278665542603,
-      "learning_rate": 8.729447236180906e-05,
-      "loss": 5.3319,
-      "step": 13146
-    },
-    {
-      "epoch": 6.85632333767927,
-      "grad_norm": 1.4096976518630981,
-      "learning_rate": 8.729346733668342e-05,
-      "loss": 5.8221,
-      "step": 13147
-    },
-    {
-      "epoch": 6.856844850065189,
-      "grad_norm": 1.512694001197815,
-      "learning_rate": 8.729246231155779e-05,
-      "loss": 5.2648,
-      "step": 13148
-    },
-    {
-      "epoch": 6.857366362451108,
-      "grad_norm": 1.4097522497177124,
-      "learning_rate": 8.729145728643217e-05,
-      "loss": 5.6509,
-      "step": 13149
-    },
-    {
-      "epoch": 6.8578878748370276,
-      "grad_norm": 1.4021409749984741,
-      "learning_rate": 8.729045226130653e-05,
-      "loss": 5.8579,
-      "step": 13150
-    },
-    {
-      "epoch": 6.858409387222947,
-      "grad_norm": 1.6009441614151,
-      "learning_rate": 8.728944723618091e-05,
-      "loss": 5.2366,
-      "step": 13151
-    },
-    {
-      "epoch": 6.858930899608866,
-      "grad_norm": 1.5279228687286377,
-      "learning_rate": 8.728844221105527e-05,
-      "loss": 4.8398,
-      "step": 13152
-    },
-    {
-      "epoch": 6.859452411994785,
-      "grad_norm": 1.3921234607696533,
-      "learning_rate": 8.728743718592965e-05,
-      "loss": 5.796,
-      "step": 13153
-    },
-    {
-      "epoch": 6.859973924380704,
-      "grad_norm": 1.5485622882843018,
-      "learning_rate": 8.728643216080403e-05,
-      "loss": 5.3697,
-      "step": 13154
-    },
-    {
-      "epoch": 6.860495436766623,
-      "grad_norm": 1.4151583909988403,
-      "learning_rate": 8.72854271356784e-05,
-      "loss": 5.8837,
-      "step": 13155
-    },
-    {
-      "epoch": 6.861016949152543,
-      "grad_norm": 1.3956092596054077,
-      "learning_rate": 8.728442211055277e-05,
-      "loss": 5.8138,
-      "step": 13156
-    },
-    {
-      "epoch": 6.861538461538462,
-      "grad_norm": 1.4662173986434937,
-      "learning_rate": 8.728341708542715e-05,
-      "loss": 5.3054,
-      "step": 13157
-    },
-    {
-      "epoch": 6.862059973924381,
-      "grad_norm": 1.5415301322937012,
-      "learning_rate": 8.728241206030151e-05,
-      "loss": 4.9936,
-      "step": 13158
-    },
-    {
-      "epoch": 6.8625814863103,
-      "grad_norm": 1.335253119468689,
-      "learning_rate": 8.728140703517589e-05,
-      "loss": 5.7895,
-      "step": 13159
-    },
-    {
-      "epoch": 6.863102998696219,
-      "grad_norm": 1.4056620597839355,
-      "learning_rate": 8.728040201005025e-05,
-      "loss": 5.115,
-      "step": 13160
-    },
-    {
-      "epoch": 6.863624511082138,
-      "grad_norm": 1.553186058998108,
-      "learning_rate": 8.727939698492462e-05,
-      "loss": 4.7464,
-      "step": 13161
-    },
-    {
-      "epoch": 6.864146023468058,
-      "grad_norm": 1.364774465560913,
-      "learning_rate": 8.7278391959799e-05,
-      "loss": 5.6091,
-      "step": 13162
-    },
-    {
-      "epoch": 6.864667535853976,
-      "grad_norm": 1.4432374238967896,
-      "learning_rate": 8.727738693467336e-05,
-      "loss": 5.1616,
-      "step": 13163
-    },
-    {
-      "epoch": 6.865189048239896,
-      "grad_norm": 1.4214255809783936,
-      "learning_rate": 8.727638190954774e-05,
-      "loss": 5.3747,
-      "step": 13164
-    },
-    {
-      "epoch": 6.865710560625815,
-      "grad_norm": 1.4623116254806519,
-      "learning_rate": 8.727537688442212e-05,
-      "loss": 5.699,
-      "step": 13165
-    },
-    {
-      "epoch": 6.866232073011734,
-      "grad_norm": 1.4343132972717285,
-      "learning_rate": 8.72743718592965e-05,
-      "loss": 5.6034,
-      "step": 13166
-    },
-    {
-      "epoch": 6.866753585397653,
-      "grad_norm": 1.4713164567947388,
-      "learning_rate": 8.727336683417086e-05,
-      "loss": 5.3098,
-      "step": 13167
-    },
-    {
-      "epoch": 6.867275097783573,
-      "grad_norm": 1.3502472639083862,
-      "learning_rate": 8.727236180904524e-05,
-      "loss": 5.0736,
-      "step": 13168
-    },
-    {
-      "epoch": 6.867796610169491,
-      "grad_norm": 1.4484864473342896,
-      "learning_rate": 8.72713567839196e-05,
-      "loss": 5.2323,
-      "step": 13169
-    },
-    {
-      "epoch": 6.868318122555411,
-      "grad_norm": 1.3253417015075684,
-      "learning_rate": 8.727035175879398e-05,
-      "loss": 5.9456,
-      "step": 13170
-    },
-    {
-      "epoch": 6.86883963494133,
-      "grad_norm": 1.4791173934936523,
-      "learning_rate": 8.726934673366834e-05,
-      "loss": 5.4112,
-      "step": 13171
-    },
-    {
-      "epoch": 6.869361147327249,
-      "grad_norm": 1.4555723667144775,
-      "learning_rate": 8.726834170854272e-05,
-      "loss": 5.6912,
-      "step": 13172
-    },
-    {
-      "epoch": 6.869882659713168,
-      "grad_norm": 1.3762667179107666,
-      "learning_rate": 8.726733668341708e-05,
-      "loss": 5.7866,
-      "step": 13173
-    },
-    {
-      "epoch": 6.870404172099088,
-      "grad_norm": 1.5583804845809937,
-      "learning_rate": 8.726633165829146e-05,
-      "loss": 5.5142,
-      "step": 13174
-    },
-    {
-      "epoch": 6.870925684485006,
-      "grad_norm": 1.2752480506896973,
-      "learning_rate": 8.726532663316584e-05,
-      "loss": 5.5938,
-      "step": 13175
-    },
-    {
-      "epoch": 6.871447196870926,
-      "grad_norm": 1.5201846361160278,
-      "learning_rate": 8.72643216080402e-05,
-      "loss": 5.1749,
-      "step": 13176
-    },
-    {
-      "epoch": 6.871968709256845,
-      "grad_norm": 1.4283307790756226,
-      "learning_rate": 8.726331658291458e-05,
-      "loss": 5.5305,
-      "step": 13177
-    },
-    {
-      "epoch": 6.872490221642764,
-      "grad_norm": 1.3894824981689453,
-      "learning_rate": 8.726231155778895e-05,
-      "loss": 5.2629,
-      "step": 13178
-    },
-    {
-      "epoch": 6.873011734028683,
-      "grad_norm": 1.442294716835022,
-      "learning_rate": 8.726130653266332e-05,
-      "loss": 5.4619,
-      "step": 13179
-    },
-    {
-      "epoch": 6.873533246414603,
-      "grad_norm": 1.4326878786087036,
-      "learning_rate": 8.726030150753769e-05,
-      "loss": 5.865,
-      "step": 13180
-    },
-    {
-      "epoch": 6.874054758800521,
-      "grad_norm": 1.3078805208206177,
-      "learning_rate": 8.725929648241207e-05,
-      "loss": 6.0233,
-      "step": 13181
-    },
-    {
-      "epoch": 6.874576271186441,
-      "grad_norm": 1.4145395755767822,
-      "learning_rate": 8.725829145728643e-05,
-      "loss": 5.3984,
-      "step": 13182
-    },
-    {
-      "epoch": 6.87509778357236,
-      "grad_norm": 1.3141976594924927,
-      "learning_rate": 8.725728643216081e-05,
-      "loss": 5.8508,
-      "step": 13183
-    },
-    {
-      "epoch": 6.875619295958279,
-      "grad_norm": 1.3464046716690063,
-      "learning_rate": 8.725628140703517e-05,
-      "loss": 5.7132,
-      "step": 13184
-    },
-    {
-      "epoch": 6.876140808344198,
-      "grad_norm": 1.4892749786376953,
-      "learning_rate": 8.725527638190955e-05,
-      "loss": 5.4607,
-      "step": 13185
-    },
-    {
-      "epoch": 6.876662320730118,
-      "grad_norm": 1.3456188440322876,
-      "learning_rate": 8.725427135678393e-05,
-      "loss": 5.9198,
-      "step": 13186
-    },
-    {
-      "epoch": 6.877183833116036,
-      "grad_norm": 1.6661677360534668,
-      "learning_rate": 8.72532663316583e-05,
-      "loss": 5.585,
-      "step": 13187
-    },
-    {
-      "epoch": 6.877705345501956,
-      "grad_norm": 1.4901773929595947,
-      "learning_rate": 8.725226130653267e-05,
-      "loss": 5.5624,
-      "step": 13188
-    },
-    {
-      "epoch": 6.878226857887875,
-      "grad_norm": 1.4497349262237549,
-      "learning_rate": 8.725125628140703e-05,
-      "loss": 5.5006,
-      "step": 13189
-    },
-    {
-      "epoch": 6.878748370273794,
-      "grad_norm": 1.4228477478027344,
-      "learning_rate": 8.725025125628141e-05,
-      "loss": 5.9949,
-      "step": 13190
-    },
-    {
-      "epoch": 6.879269882659713,
-      "grad_norm": 1.5094822645187378,
-      "learning_rate": 8.724924623115578e-05,
-      "loss": 5.7937,
-      "step": 13191
-    },
-    {
-      "epoch": 6.879791395045633,
-      "grad_norm": 1.6918152570724487,
-      "learning_rate": 8.724824120603015e-05,
-      "loss": 5.4494,
-      "step": 13192
-    },
-    {
-      "epoch": 6.880312907431551,
-      "grad_norm": 1.3921840190887451,
-      "learning_rate": 8.724723618090452e-05,
-      "loss": 5.6562,
-      "step": 13193
-    },
-    {
-      "epoch": 6.880834419817471,
-      "grad_norm": 1.324840784072876,
-      "learning_rate": 8.72462311557789e-05,
-      "loss": 5.8676,
-      "step": 13194
-    },
-    {
-      "epoch": 6.88135593220339,
-      "grad_norm": 1.4437055587768555,
-      "learning_rate": 8.724522613065327e-05,
-      "loss": 5.2625,
-      "step": 13195
-    },
-    {
-      "epoch": 6.881877444589309,
-      "grad_norm": 1.4696484804153442,
-      "learning_rate": 8.724422110552765e-05,
-      "loss": 5.8266,
-      "step": 13196
-    },
-    {
-      "epoch": 6.882398956975228,
-      "grad_norm": 1.4741723537445068,
-      "learning_rate": 8.724321608040202e-05,
-      "loss": 5.7074,
-      "step": 13197
-    },
-    {
-      "epoch": 6.882920469361148,
-      "grad_norm": 1.3416105508804321,
-      "learning_rate": 8.72422110552764e-05,
-      "loss": 5.9414,
-      "step": 13198
-    },
-    {
-      "epoch": 6.883441981747066,
-      "grad_norm": 1.3794000148773193,
-      "learning_rate": 8.724120603015076e-05,
-      "loss": 5.491,
-      "step": 13199
-    },
-    {
-      "epoch": 6.883963494132986,
-      "grad_norm": 1.409883737564087,
-      "learning_rate": 8.724020100502514e-05,
-      "loss": 5.8139,
-      "step": 13200
-    },
-    {
-      "epoch": 6.884485006518905,
-      "grad_norm": 1.5332021713256836,
-      "learning_rate": 8.72391959798995e-05,
-      "loss": 5.5962,
-      "step": 13201
-    },
-    {
-      "epoch": 6.885006518904824,
-      "grad_norm": 1.505989670753479,
-      "learning_rate": 8.723819095477386e-05,
-      "loss": 5.6001,
-      "step": 13202
-    },
-    {
-      "epoch": 6.885528031290743,
-      "grad_norm": 1.3647571802139282,
-      "learning_rate": 8.723718592964824e-05,
-      "loss": 5.6893,
-      "step": 13203
-    },
-    {
-      "epoch": 6.886049543676663,
-      "grad_norm": 1.4559099674224854,
-      "learning_rate": 8.723618090452261e-05,
-      "loss": 5.4508,
-      "step": 13204
-    },
-    {
-      "epoch": 6.886571056062581,
-      "grad_norm": 1.4732754230499268,
-      "learning_rate": 8.723517587939698e-05,
-      "loss": 5.5885,
-      "step": 13205
-    },
-    {
-      "epoch": 6.887092568448501,
-      "grad_norm": 1.3421133756637573,
-      "learning_rate": 8.723417085427136e-05,
-      "loss": 5.44,
-      "step": 13206
-    },
-    {
-      "epoch": 6.88761408083442,
-      "grad_norm": 1.4015789031982422,
-      "learning_rate": 8.723316582914574e-05,
-      "loss": 5.8009,
-      "step": 13207
-    },
-    {
-      "epoch": 6.888135593220339,
-      "grad_norm": 1.5020679235458374,
-      "learning_rate": 8.72321608040201e-05,
-      "loss": 5.4555,
-      "step": 13208
-    },
-    {
-      "epoch": 6.888657105606258,
-      "grad_norm": 1.4048621654510498,
-      "learning_rate": 8.723115577889448e-05,
-      "loss": 6.0182,
-      "step": 13209
-    },
-    {
-      "epoch": 6.889178617992178,
-      "grad_norm": 1.480368733406067,
-      "learning_rate": 8.723015075376885e-05,
-      "loss": 5.6949,
-      "step": 13210
-    },
-    {
-      "epoch": 6.889700130378096,
-      "grad_norm": 1.453173279762268,
-      "learning_rate": 8.722914572864322e-05,
-      "loss": 5.549,
-      "step": 13211
-    },
-    {
-      "epoch": 6.890221642764016,
-      "grad_norm": 1.3605462312698364,
-      "learning_rate": 8.722814070351759e-05,
-      "loss": 5.7951,
-      "step": 13212
-    },
-    {
-      "epoch": 6.890743155149935,
-      "grad_norm": 1.3382529020309448,
-      "learning_rate": 8.722713567839197e-05,
-      "loss": 6.1252,
-      "step": 13213
-    },
-    {
-      "epoch": 6.891264667535854,
-      "grad_norm": 1.4121787548065186,
-      "learning_rate": 8.722613065326633e-05,
-      "loss": 5.169,
-      "step": 13214
-    },
-    {
-      "epoch": 6.891786179921773,
-      "grad_norm": 1.4558900594711304,
-      "learning_rate": 8.722512562814071e-05,
-      "loss": 5.5668,
-      "step": 13215
-    },
-    {
-      "epoch": 6.892307692307693,
-      "grad_norm": 1.3969424962997437,
-      "learning_rate": 8.722412060301509e-05,
-      "loss": 5.9214,
-      "step": 13216
-    },
-    {
-      "epoch": 6.892829204693611,
-      "grad_norm": 1.4808926582336426,
-      "learning_rate": 8.722311557788945e-05,
-      "loss": 5.6017,
-      "step": 13217
-    },
-    {
-      "epoch": 6.893350717079531,
-      "grad_norm": 1.392142415046692,
-      "learning_rate": 8.722211055276383e-05,
-      "loss": 5.5681,
-      "step": 13218
-    },
-    {
-      "epoch": 6.893872229465449,
-      "grad_norm": 1.3810856342315674,
-      "learning_rate": 8.722110552763819e-05,
-      "loss": 5.2584,
-      "step": 13219
-    },
-    {
-      "epoch": 6.894393741851369,
-      "grad_norm": 1.4694420099258423,
-      "learning_rate": 8.722010050251257e-05,
-      "loss": 5.6668,
-      "step": 13220
-    },
-    {
-      "epoch": 6.894915254237288,
-      "grad_norm": 1.3267583847045898,
-      "learning_rate": 8.721909547738694e-05,
-      "loss": 5.6424,
-      "step": 13221
-    },
-    {
-      "epoch": 6.895436766623208,
-      "grad_norm": 1.3700863122940063,
-      "learning_rate": 8.721809045226131e-05,
-      "loss": 6.0408,
-      "step": 13222
-    },
-    {
-      "epoch": 6.895958279009126,
-      "grad_norm": 1.3975127935409546,
-      "learning_rate": 8.721708542713568e-05,
-      "loss": 5.7675,
-      "step": 13223
-    },
-    {
-      "epoch": 6.896479791395046,
-      "grad_norm": 1.8679968118667603,
-      "learning_rate": 8.721608040201005e-05,
-      "loss": 4.4971,
-      "step": 13224
-    },
-    {
-      "epoch": 6.897001303780964,
-      "grad_norm": 1.4413628578186035,
-      "learning_rate": 8.721507537688442e-05,
-      "loss": 5.3969,
-      "step": 13225
-    },
-    {
-      "epoch": 6.897522816166884,
-      "grad_norm": 1.5381438732147217,
-      "learning_rate": 8.72140703517588e-05,
-      "loss": 5.8236,
-      "step": 13226
-    },
-    {
-      "epoch": 6.898044328552803,
-      "grad_norm": 1.4670847654342651,
-      "learning_rate": 8.721306532663317e-05,
-      "loss": 5.6581,
-      "step": 13227
-    },
-    {
-      "epoch": 6.898565840938723,
-      "grad_norm": 1.5865319967269897,
-      "learning_rate": 8.721206030150754e-05,
-      "loss": 5.2943,
-      "step": 13228
-    },
-    {
-      "epoch": 6.899087353324641,
-      "grad_norm": 1.394509196281433,
-      "learning_rate": 8.721105527638192e-05,
-      "loss": 5.9158,
-      "step": 13229
-    },
-    {
-      "epoch": 6.899608865710561,
-      "grad_norm": 1.3617596626281738,
-      "learning_rate": 8.721005025125628e-05,
-      "loss": 5.9961,
-      "step": 13230
-    },
-    {
-      "epoch": 6.900130378096479,
-      "grad_norm": 1.4174559116363525,
-      "learning_rate": 8.720904522613066e-05,
-      "loss": 5.5684,
-      "step": 13231
-    },
-    {
-      "epoch": 6.900651890482399,
-      "grad_norm": 1.3633148670196533,
-      "learning_rate": 8.720804020100502e-05,
-      "loss": 5.644,
-      "step": 13232
-    },
-    {
-      "epoch": 6.901173402868318,
-      "grad_norm": 1.4199458360671997,
-      "learning_rate": 8.72070351758794e-05,
-      "loss": 5.7105,
-      "step": 13233
-    },
-    {
-      "epoch": 6.901694915254238,
-      "grad_norm": 1.4530327320098877,
-      "learning_rate": 8.720603015075377e-05,
-      "loss": 5.4772,
-      "step": 13234
-    },
-    {
-      "epoch": 6.902216427640156,
-      "grad_norm": 1.3954952955245972,
-      "learning_rate": 8.720502512562814e-05,
-      "loss": 5.5851,
-      "step": 13235
-    },
-    {
-      "epoch": 6.902737940026076,
-      "grad_norm": 1.3727400302886963,
-      "learning_rate": 8.720402010050252e-05,
-      "loss": 5.9219,
-      "step": 13236
-    },
-    {
-      "epoch": 6.903259452411994,
-      "grad_norm": 1.3902932405471802,
-      "learning_rate": 8.72030150753769e-05,
-      "loss": 5.7362,
-      "step": 13237
-    },
-    {
-      "epoch": 6.903780964797914,
-      "grad_norm": 1.447379231452942,
-      "learning_rate": 8.720201005025126e-05,
-      "loss": 5.4472,
-      "step": 13238
-    },
-    {
-      "epoch": 6.904302477183833,
-      "grad_norm": 1.3186500072479248,
-      "learning_rate": 8.720100502512564e-05,
-      "loss": 5.7276,
-      "step": 13239
-    },
-    {
-      "epoch": 6.904823989569753,
-      "grad_norm": 1.385095477104187,
-      "learning_rate": 8.72e-05,
-      "loss": 5.3252,
-      "step": 13240
-    },
-    {
-      "epoch": 6.905345501955671,
-      "grad_norm": 1.3445531129837036,
-      "learning_rate": 8.719899497487437e-05,
-      "loss": 5.443,
-      "step": 13241
-    },
-    {
-      "epoch": 6.905867014341591,
-      "grad_norm": 1.3237533569335938,
-      "learning_rate": 8.719798994974875e-05,
-      "loss": 5.8137,
-      "step": 13242
-    },
-    {
-      "epoch": 6.906388526727509,
-      "grad_norm": 1.2612031698226929,
-      "learning_rate": 8.719698492462311e-05,
-      "loss": 5.8713,
-      "step": 13243
-    },
-    {
-      "epoch": 6.906910039113429,
-      "grad_norm": 1.385340690612793,
-      "learning_rate": 8.719597989949749e-05,
-      "loss": 5.498,
-      "step": 13244
-    },
-    {
-      "epoch": 6.907431551499348,
-      "grad_norm": 1.701211929321289,
-      "learning_rate": 8.719497487437185e-05,
-      "loss": 5.5625,
-      "step": 13245
-    },
-    {
-      "epoch": 6.907953063885268,
-      "grad_norm": 1.4198564291000366,
-      "learning_rate": 8.719396984924623e-05,
-      "loss": 5.8423,
-      "step": 13246
-    },
-    {
-      "epoch": 6.908474576271186,
-      "grad_norm": 1.3155579566955566,
-      "learning_rate": 8.719296482412061e-05,
-      "loss": 5.8663,
-      "step": 13247
-    },
-    {
-      "epoch": 6.908996088657106,
-      "grad_norm": 1.3304059505462646,
-      "learning_rate": 8.719195979899499e-05,
-      "loss": 5.7987,
-      "step": 13248
-    },
-    {
-      "epoch": 6.909517601043024,
-      "grad_norm": 1.4307845830917358,
-      "learning_rate": 8.719095477386935e-05,
-      "loss": 5.7693,
-      "step": 13249
-    },
-    {
-      "epoch": 6.910039113428944,
-      "grad_norm": 1.343855619430542,
-      "learning_rate": 8.718994974874373e-05,
-      "loss": 5.8502,
-      "step": 13250
-    },
-    {
-      "epoch": 6.910560625814863,
-      "grad_norm": 1.706536889076233,
-      "learning_rate": 8.71889447236181e-05,
-      "loss": 5.3318,
-      "step": 13251
-    },
-    {
-      "epoch": 6.911082138200783,
-      "grad_norm": 1.364210844039917,
-      "learning_rate": 8.718793969849247e-05,
-      "loss": 5.7844,
-      "step": 13252
-    },
-    {
-      "epoch": 6.911603650586701,
-      "grad_norm": 1.4805934429168701,
-      "learning_rate": 8.718693467336684e-05,
-      "loss": 5.7988,
-      "step": 13253
-    },
-    {
-      "epoch": 6.912125162972621,
-      "grad_norm": 1.3838690519332886,
-      "learning_rate": 8.71859296482412e-05,
-      "loss": 5.6777,
-      "step": 13254
-    },
-    {
-      "epoch": 6.912646675358539,
-      "grad_norm": 1.459098219871521,
-      "learning_rate": 8.718492462311558e-05,
-      "loss": 5.3691,
-      "step": 13255
-    },
-    {
-      "epoch": 6.913168187744459,
-      "grad_norm": 1.3118094205856323,
-      "learning_rate": 8.718391959798996e-05,
-      "loss": 5.8146,
-      "step": 13256
-    },
-    {
-      "epoch": 6.913689700130378,
-      "grad_norm": 1.4691168069839478,
-      "learning_rate": 8.718291457286433e-05,
-      "loss": 5.5871,
-      "step": 13257
-    },
-    {
-      "epoch": 6.914211212516297,
-      "grad_norm": 1.2898691892623901,
-      "learning_rate": 8.71819095477387e-05,
-      "loss": 5.6116,
-      "step": 13258
-    },
-    {
-      "epoch": 6.914732724902216,
-      "grad_norm": 1.5554901361465454,
-      "learning_rate": 8.718090452261308e-05,
-      "loss": 5.5725,
-      "step": 13259
-    },
-    {
-      "epoch": 6.915254237288136,
-      "grad_norm": 1.4737842082977295,
-      "learning_rate": 8.717989949748744e-05,
-      "loss": 5.7095,
-      "step": 13260
-    },
-    {
-      "epoch": 6.915775749674054,
-      "grad_norm": 1.4652020931243896,
-      "learning_rate": 8.717889447236182e-05,
-      "loss": 5.3408,
-      "step": 13261
-    },
-    {
-      "epoch": 6.916297262059974,
-      "grad_norm": 1.3736578226089478,
-      "learning_rate": 8.717788944723618e-05,
-      "loss": 5.7088,
-      "step": 13262
-    },
-    {
-      "epoch": 6.916818774445893,
-      "grad_norm": 1.4180052280426025,
-      "learning_rate": 8.717688442211056e-05,
-      "loss": 5.5649,
-      "step": 13263
-    },
-    {
-      "epoch": 6.917340286831812,
-      "grad_norm": 1.4963427782058716,
-      "learning_rate": 8.717587939698492e-05,
-      "loss": 5.4054,
-      "step": 13264
-    },
-    {
-      "epoch": 6.917861799217731,
-      "grad_norm": 1.4722579717636108,
-      "learning_rate": 8.71748743718593e-05,
-      "loss": 5.5226,
-      "step": 13265
-    },
-    {
-      "epoch": 6.918383311603651,
-      "grad_norm": 1.46590256690979,
-      "learning_rate": 8.717386934673367e-05,
-      "loss": 5.2344,
-      "step": 13266
-    },
-    {
-      "epoch": 6.918904823989569,
-      "grad_norm": 1.4531309604644775,
-      "learning_rate": 8.717286432160804e-05,
-      "loss": 5.4399,
-      "step": 13267
-    },
-    {
-      "epoch": 6.919426336375489,
-      "grad_norm": 1.354246735572815,
-      "learning_rate": 8.717185929648242e-05,
-      "loss": 5.66,
-      "step": 13268
-    },
-    {
-      "epoch": 6.919947848761408,
-      "grad_norm": 1.443540334701538,
-      "learning_rate": 8.717085427135679e-05,
-      "loss": 5.4641,
-      "step": 13269
-    },
-    {
-      "epoch": 6.920469361147327,
-      "grad_norm": 1.4912827014923096,
-      "learning_rate": 8.716984924623116e-05,
-      "loss": 5.8973,
-      "step": 13270
-    },
-    {
-      "epoch": 6.920990873533246,
-      "grad_norm": 1.3750600814819336,
-      "learning_rate": 8.716884422110553e-05,
-      "loss": 5.5577,
-      "step": 13271
-    },
-    {
-      "epoch": 6.921512385919166,
-      "grad_norm": 1.5366754531860352,
-      "learning_rate": 8.71678391959799e-05,
-      "loss": 5.0216,
-      "step": 13272
-    },
-    {
-      "epoch": 6.922033898305084,
-      "grad_norm": 1.6127934455871582,
-      "learning_rate": 8.716683417085427e-05,
-      "loss": 5.4429,
-      "step": 13273
-    },
-    {
-      "epoch": 6.922555410691004,
-      "grad_norm": 1.4167646169662476,
-      "learning_rate": 8.716582914572865e-05,
-      "loss": 5.8597,
-      "step": 13274
-    },
-    {
-      "epoch": 6.923076923076923,
-      "grad_norm": 1.4411351680755615,
-      "learning_rate": 8.716482412060301e-05,
-      "loss": 5.5687,
-      "step": 13275
-    },
-    {
-      "epoch": 6.923598435462842,
-      "grad_norm": 1.4399065971374512,
-      "learning_rate": 8.716381909547739e-05,
-      "loss": 5.8471,
-      "step": 13276
-    },
-    {
-      "epoch": 6.924119947848761,
-      "grad_norm": 1.3144664764404297,
-      "learning_rate": 8.716281407035177e-05,
-      "loss": 5.6316,
-      "step": 13277
-    },
-    {
-      "epoch": 6.924641460234681,
-      "grad_norm": 1.5150513648986816,
-      "learning_rate": 8.716180904522615e-05,
-      "loss": 5.3082,
-      "step": 13278
-    },
-    {
-      "epoch": 6.925162972620599,
-      "grad_norm": 1.4315844774246216,
-      "learning_rate": 8.716080402010051e-05,
-      "loss": 5.8042,
-      "step": 13279
-    },
-    {
-      "epoch": 6.925684485006519,
-      "grad_norm": 1.4487781524658203,
-      "learning_rate": 8.715979899497489e-05,
-      "loss": 5.6111,
-      "step": 13280
-    },
-    {
-      "epoch": 6.926205997392438,
-      "grad_norm": 1.4278978109359741,
-      "learning_rate": 8.715879396984925e-05,
-      "loss": 5.785,
-      "step": 13281
-    },
-    {
-      "epoch": 6.926727509778357,
-      "grad_norm": 1.4267510175704956,
-      "learning_rate": 8.715778894472362e-05,
-      "loss": 5.5838,
-      "step": 13282
-    },
-    {
-      "epoch": 6.927249022164276,
-      "grad_norm": 1.3004896640777588,
-      "learning_rate": 8.7156783919598e-05,
-      "loss": 5.8541,
-      "step": 13283
-    },
-    {
-      "epoch": 6.927770534550196,
-      "grad_norm": 1.5250394344329834,
-      "learning_rate": 8.715577889447236e-05,
-      "loss": 5.7226,
-      "step": 13284
-    },
-    {
-      "epoch": 6.9282920469361144,
-      "grad_norm": 1.5303797721862793,
-      "learning_rate": 8.715477386934674e-05,
-      "loss": 5.9732,
-      "step": 13285
-    },
-    {
-      "epoch": 6.928813559322034,
-      "grad_norm": 1.4950717687606812,
-      "learning_rate": 8.71537688442211e-05,
-      "loss": 4.8119,
-      "step": 13286
-    },
-    {
-      "epoch": 6.929335071707953,
-      "grad_norm": 1.4639976024627686,
-      "learning_rate": 8.715276381909548e-05,
-      "loss": 5.6793,
-      "step": 13287
-    },
-    {
-      "epoch": 6.929856584093872,
-      "grad_norm": 1.38613760471344,
-      "learning_rate": 8.715175879396986e-05,
-      "loss": 5.5166,
-      "step": 13288
-    },
-    {
-      "epoch": 6.930378096479791,
-      "grad_norm": 1.4668622016906738,
-      "learning_rate": 8.715075376884423e-05,
-      "loss": 5.8096,
-      "step": 13289
-    },
-    {
-      "epoch": 6.930899608865711,
-      "grad_norm": 1.3955070972442627,
-      "learning_rate": 8.71497487437186e-05,
-      "loss": 5.7188,
-      "step": 13290
-    },
-    {
-      "epoch": 6.9314211212516295,
-      "grad_norm": 1.3983622789382935,
-      "learning_rate": 8.714874371859298e-05,
-      "loss": 5.4946,
-      "step": 13291
-    },
-    {
-      "epoch": 6.931942633637549,
-      "grad_norm": 1.6636418104171753,
-      "learning_rate": 8.714773869346734e-05,
-      "loss": 5.3901,
-      "step": 13292
-    },
-    {
-      "epoch": 6.932464146023468,
-      "grad_norm": 1.5902067422866821,
-      "learning_rate": 8.714673366834172e-05,
-      "loss": 5.5319,
-      "step": 13293
-    },
-    {
-      "epoch": 6.932985658409387,
-      "grad_norm": 1.3940727710723877,
-      "learning_rate": 8.714572864321608e-05,
-      "loss": 5.6151,
-      "step": 13294
-    },
-    {
-      "epoch": 6.933507170795306,
-      "grad_norm": 1.5068354606628418,
-      "learning_rate": 8.714472361809045e-05,
-      "loss": 5.3728,
-      "step": 13295
-    },
-    {
-      "epoch": 6.934028683181226,
-      "grad_norm": 1.6044501066207886,
-      "learning_rate": 8.714371859296482e-05,
-      "loss": 5.5412,
-      "step": 13296
-    },
-    {
-      "epoch": 6.9345501955671445,
-      "grad_norm": 1.5169472694396973,
-      "learning_rate": 8.714271356783919e-05,
-      "loss": 5.1521,
-      "step": 13297
-    },
-    {
-      "epoch": 6.935071707953064,
-      "grad_norm": 1.4585224390029907,
-      "learning_rate": 8.714170854271357e-05,
-      "loss": 5.432,
-      "step": 13298
-    },
-    {
-      "epoch": 6.935593220338983,
-      "grad_norm": 1.4531594514846802,
-      "learning_rate": 8.714070351758794e-05,
-      "loss": 5.5702,
-      "step": 13299
-    },
-    {
-      "epoch": 6.936114732724902,
-      "grad_norm": 1.396121621131897,
-      "learning_rate": 8.713969849246232e-05,
-      "loss": 5.7547,
-      "step": 13300
-    },
-    {
-      "epoch": 6.936636245110821,
-      "grad_norm": 1.4604768753051758,
-      "learning_rate": 8.713869346733669e-05,
-      "loss": 5.4952,
-      "step": 13301
-    },
-    {
-      "epoch": 6.937157757496741,
-      "grad_norm": 1.4175890684127808,
-      "learning_rate": 8.713768844221106e-05,
-      "loss": 5.6259,
-      "step": 13302
-    },
-    {
-      "epoch": 6.9376792698826595,
-      "grad_norm": 1.452567219734192,
-      "learning_rate": 8.713668341708543e-05,
-      "loss": 5.5131,
-      "step": 13303
-    },
-    {
-      "epoch": 6.938200782268579,
-      "grad_norm": 1.3835870027542114,
-      "learning_rate": 8.713567839195981e-05,
-      "loss": 5.7012,
-      "step": 13304
-    },
-    {
-      "epoch": 6.938722294654498,
-      "grad_norm": 1.4527009725570679,
-      "learning_rate": 8.713467336683417e-05,
-      "loss": 5.474,
-      "step": 13305
-    },
-    {
-      "epoch": 6.939243807040417,
-      "grad_norm": 1.482246994972229,
-      "learning_rate": 8.713366834170855e-05,
-      "loss": 5.391,
-      "step": 13306
-    },
-    {
-      "epoch": 6.9397653194263365,
-      "grad_norm": 1.6015952825546265,
-      "learning_rate": 8.713266331658291e-05,
-      "loss": 4.9857,
-      "step": 13307
-    },
-    {
-      "epoch": 6.940286831812256,
-      "grad_norm": 1.4695680141448975,
-      "learning_rate": 8.713165829145729e-05,
-      "loss": 5.6018,
-      "step": 13308
-    },
-    {
-      "epoch": 6.9408083441981745,
-      "grad_norm": 1.4726529121398926,
-      "learning_rate": 8.713065326633167e-05,
-      "loss": 5.0986,
-      "step": 13309
-    },
-    {
-      "epoch": 6.941329856584094,
-      "grad_norm": 1.4324073791503906,
-      "learning_rate": 8.712964824120603e-05,
-      "loss": 5.3927,
-      "step": 13310
-    },
-    {
-      "epoch": 6.941851368970013,
-      "grad_norm": 1.4207093715667725,
-      "learning_rate": 8.712864321608041e-05,
-      "loss": 5.8956,
-      "step": 13311
-    },
-    {
-      "epoch": 6.942372881355932,
-      "grad_norm": 1.4475767612457275,
-      "learning_rate": 8.712763819095478e-05,
-      "loss": 5.5813,
-      "step": 13312
-    },
-    {
-      "epoch": 6.9428943937418515,
-      "grad_norm": 1.5410809516906738,
-      "learning_rate": 8.712663316582915e-05,
-      "loss": 5.635,
-      "step": 13313
-    },
-    {
-      "epoch": 6.94341590612777,
-      "grad_norm": 1.4451422691345215,
-      "learning_rate": 8.712562814070352e-05,
-      "loss": 5.8924,
-      "step": 13314
-    },
-    {
-      "epoch": 6.9439374185136895,
-      "grad_norm": 1.508739948272705,
-      "learning_rate": 8.71246231155779e-05,
-      "loss": 5.7591,
-      "step": 13315
-    },
-    {
-      "epoch": 6.944458930899609,
-      "grad_norm": 1.3732361793518066,
-      "learning_rate": 8.712361809045226e-05,
-      "loss": 5.6796,
-      "step": 13316
-    },
-    {
-      "epoch": 6.944980443285528,
-      "grad_norm": 1.5606842041015625,
-      "learning_rate": 8.712261306532664e-05,
-      "loss": 5.3723,
-      "step": 13317
-    },
-    {
-      "epoch": 6.945501955671447,
-      "grad_norm": 1.4442291259765625,
-      "learning_rate": 8.7121608040201e-05,
-      "loss": 5.4069,
-      "step": 13318
-    },
-    {
-      "epoch": 6.9460234680573665,
-      "grad_norm": 1.581460952758789,
-      "learning_rate": 8.712060301507538e-05,
-      "loss": 5.2218,
-      "step": 13319
-    },
-    {
-      "epoch": 6.946544980443285,
-      "grad_norm": 1.3619707822799683,
-      "learning_rate": 8.711959798994976e-05,
-      "loss": 5.7508,
-      "step": 13320
-    },
-    {
-      "epoch": 6.9470664928292045,
-      "grad_norm": 1.5278903245925903,
-      "learning_rate": 8.711859296482412e-05,
-      "loss": 5.8552,
-      "step": 13321
-    },
-    {
-      "epoch": 6.947588005215124,
-      "grad_norm": 1.3526830673217773,
-      "learning_rate": 8.71175879396985e-05,
-      "loss": 5.7088,
-      "step": 13322
-    },
-    {
-      "epoch": 6.9481095176010434,
-      "grad_norm": 1.4586069583892822,
-      "learning_rate": 8.711658291457286e-05,
-      "loss": 5.0016,
-      "step": 13323
-    },
-    {
-      "epoch": 6.948631029986962,
-      "grad_norm": 1.407771110534668,
-      "learning_rate": 8.711557788944724e-05,
-      "loss": 5.5982,
-      "step": 13324
-    },
-    {
-      "epoch": 6.9491525423728815,
-      "grad_norm": 1.3836181163787842,
-      "learning_rate": 8.71145728643216e-05,
-      "loss": 5.618,
-      "step": 13325
-    },
-    {
-      "epoch": 6.9496740547588,
-      "grad_norm": 1.3673350811004639,
-      "learning_rate": 8.711356783919598e-05,
-      "loss": 5.7113,
-      "step": 13326
-    },
-    {
-      "epoch": 6.9501955671447195,
-      "grad_norm": 1.4720780849456787,
-      "learning_rate": 8.711256281407035e-05,
-      "loss": 5.5075,
-      "step": 13327
-    },
-    {
-      "epoch": 6.950717079530639,
-      "grad_norm": 1.395456075668335,
-      "learning_rate": 8.711155778894473e-05,
-      "loss": 5.7791,
-      "step": 13328
-    },
-    {
-      "epoch": 6.9512385919165585,
-      "grad_norm": 1.4478695392608643,
-      "learning_rate": 8.71105527638191e-05,
-      "loss": 5.8729,
-      "step": 13329
-    },
-    {
-      "epoch": 6.951760104302477,
-      "grad_norm": 1.3494130373001099,
-      "learning_rate": 8.710954773869348e-05,
-      "loss": 5.9408,
-      "step": 13330
-    },
-    {
-      "epoch": 6.9522816166883965,
-      "grad_norm": 1.450488567352295,
-      "learning_rate": 8.710854271356785e-05,
-      "loss": 5.2868,
-      "step": 13331
-    },
-    {
-      "epoch": 6.952803129074315,
-      "grad_norm": 1.5069586038589478,
-      "learning_rate": 8.710753768844222e-05,
-      "loss": 5.4187,
-      "step": 13332
-    },
-    {
-      "epoch": 6.9533246414602345,
-      "grad_norm": 1.3240537643432617,
-      "learning_rate": 8.710653266331659e-05,
-      "loss": 5.4792,
-      "step": 13333
-    },
-    {
-      "epoch": 6.953846153846154,
-      "grad_norm": 1.6023038625717163,
-      "learning_rate": 8.710552763819095e-05,
-      "loss": 5.6603,
-      "step": 13334
-    },
-    {
-      "epoch": 6.9543676662320735,
-      "grad_norm": 1.555156946182251,
-      "learning_rate": 8.710452261306533e-05,
-      "loss": 5.9118,
-      "step": 13335
-    },
-    {
-      "epoch": 6.954889178617992,
-      "grad_norm": 1.3293298482894897,
-      "learning_rate": 8.71035175879397e-05,
-      "loss": 5.9274,
-      "step": 13336
-    },
-    {
-      "epoch": 6.9554106910039115,
-      "grad_norm": 1.4671381711959839,
-      "learning_rate": 8.710251256281407e-05,
-      "loss": 5.5823,
-      "step": 13337
-    },
-    {
-      "epoch": 6.95593220338983,
-      "grad_norm": 1.4281806945800781,
-      "learning_rate": 8.710150753768844e-05,
-      "loss": 5.6768,
-      "step": 13338
-    },
-    {
-      "epoch": 6.95645371577575,
-      "grad_norm": 1.4684056043624878,
-      "learning_rate": 8.710050251256281e-05,
-      "loss": 5.787,
-      "step": 13339
-    },
-    {
-      "epoch": 6.956975228161669,
-      "grad_norm": 1.3198171854019165,
-      "learning_rate": 8.709949748743719e-05,
-      "loss": 5.4559,
-      "step": 13340
-    },
-    {
-      "epoch": 6.9574967405475885,
-      "grad_norm": 1.5199273824691772,
-      "learning_rate": 8.709849246231157e-05,
-      "loss": 5.5438,
-      "step": 13341
-    },
-    {
-      "epoch": 6.958018252933507,
-      "grad_norm": 1.3330984115600586,
-      "learning_rate": 8.709748743718593e-05,
-      "loss": 6.1115,
-      "step": 13342
-    },
-    {
-      "epoch": 6.9585397653194265,
-      "grad_norm": 1.3391082286834717,
-      "learning_rate": 8.709648241206031e-05,
-      "loss": 5.5991,
-      "step": 13343
-    },
-    {
-      "epoch": 6.959061277705345,
-      "grad_norm": 1.4201982021331787,
-      "learning_rate": 8.709547738693468e-05,
-      "loss": 5.8293,
-      "step": 13344
-    },
-    {
-      "epoch": 6.959582790091265,
-      "grad_norm": 1.4060670137405396,
-      "learning_rate": 8.709447236180905e-05,
-      "loss": 5.2834,
-      "step": 13345
-    },
-    {
-      "epoch": 6.960104302477184,
-      "grad_norm": 1.4142917394638062,
-      "learning_rate": 8.709346733668342e-05,
-      "loss": 5.8882,
-      "step": 13346
-    },
-    {
-      "epoch": 6.960625814863103,
-      "grad_norm": 1.42680025100708,
-      "learning_rate": 8.709246231155778e-05,
-      "loss": 5.602,
-      "step": 13347
-    },
-    {
-      "epoch": 6.961147327249022,
-      "grad_norm": 1.4544587135314941,
-      "learning_rate": 8.709145728643216e-05,
-      "loss": 5.63,
-      "step": 13348
-    },
-    {
-      "epoch": 6.9616688396349415,
-      "grad_norm": 1.4432710409164429,
-      "learning_rate": 8.709045226130654e-05,
-      "loss": 5.746,
-      "step": 13349
-    },
-    {
-      "epoch": 6.96219035202086,
-      "grad_norm": 1.4538097381591797,
-      "learning_rate": 8.708944723618092e-05,
-      "loss": 5.5986,
-      "step": 13350
-    },
-    {
-      "epoch": 6.96271186440678,
-      "grad_norm": 1.4957667589187622,
-      "learning_rate": 8.708844221105528e-05,
-      "loss": 5.4376,
-      "step": 13351
-    },
-    {
-      "epoch": 6.963233376792699,
-      "grad_norm": 1.4634274244308472,
-      "learning_rate": 8.708743718592966e-05,
-      "loss": 5.6969,
-      "step": 13352
-    },
-    {
-      "epoch": 6.963754889178618,
-      "grad_norm": 1.4709599018096924,
-      "learning_rate": 8.708643216080402e-05,
-      "loss": 5.7251,
-      "step": 13353
-    },
-    {
-      "epoch": 6.964276401564537,
-      "grad_norm": 1.4052854776382446,
-      "learning_rate": 8.70854271356784e-05,
-      "loss": 5.4901,
-      "step": 13354
-    },
-    {
-      "epoch": 6.9647979139504566,
-      "grad_norm": 1.3707952499389648,
-      "learning_rate": 8.708442211055276e-05,
-      "loss": 5.7167,
-      "step": 13355
-    },
-    {
-      "epoch": 6.965319426336375,
-      "grad_norm": 1.4914451837539673,
-      "learning_rate": 8.708341708542714e-05,
-      "loss": 5.1317,
-      "step": 13356
-    },
-    {
-      "epoch": 6.965840938722295,
-      "grad_norm": 1.4216047525405884,
-      "learning_rate": 8.70824120603015e-05,
-      "loss": 5.161,
-      "step": 13357
-    },
-    {
-      "epoch": 6.966362451108214,
-      "grad_norm": 1.4134098291397095,
-      "learning_rate": 8.708140703517588e-05,
-      "loss": 5.1967,
-      "step": 13358
-    },
-    {
-      "epoch": 6.966883963494133,
-      "grad_norm": 1.4451937675476074,
-      "learning_rate": 8.708040201005025e-05,
-      "loss": 5.5869,
-      "step": 13359
-    },
-    {
-      "epoch": 6.967405475880052,
-      "grad_norm": 1.3835577964782715,
-      "learning_rate": 8.707939698492463e-05,
-      "loss": 5.7478,
-      "step": 13360
-    },
-    {
-      "epoch": 6.967926988265972,
-      "grad_norm": 1.659940481185913,
-      "learning_rate": 8.7078391959799e-05,
-      "loss": 5.3998,
-      "step": 13361
-    },
-    {
-      "epoch": 6.96844850065189,
-      "grad_norm": 1.451550006866455,
-      "learning_rate": 8.707738693467337e-05,
-      "loss": 5.1264,
-      "step": 13362
-    },
-    {
-      "epoch": 6.96897001303781,
-      "grad_norm": 1.3033826351165771,
-      "learning_rate": 8.707638190954775e-05,
-      "loss": 5.9982,
-      "step": 13363
-    },
-    {
-      "epoch": 6.969491525423729,
-      "grad_norm": 1.4470417499542236,
-      "learning_rate": 8.707537688442211e-05,
-      "loss": 5.9449,
-      "step": 13364
-    },
-    {
-      "epoch": 6.970013037809648,
-      "grad_norm": 1.505178689956665,
-      "learning_rate": 8.707437185929649e-05,
-      "loss": 5.3828,
-      "step": 13365
-    },
-    {
-      "epoch": 6.970534550195567,
-      "grad_norm": 1.4007105827331543,
-      "learning_rate": 8.707336683417085e-05,
-      "loss": 5.0486,
-      "step": 13366
-    },
-    {
-      "epoch": 6.971056062581487,
-      "grad_norm": 1.3721015453338623,
-      "learning_rate": 8.707236180904523e-05,
-      "loss": 5.843,
-      "step": 13367
-    },
-    {
-      "epoch": 6.971577574967405,
-      "grad_norm": 1.439216136932373,
-      "learning_rate": 8.70713567839196e-05,
-      "loss": 5.4816,
-      "step": 13368
-    },
-    {
-      "epoch": 6.972099087353325,
-      "grad_norm": 1.418509840965271,
-      "learning_rate": 8.707035175879397e-05,
-      "loss": 5.6509,
-      "step": 13369
-    },
-    {
-      "epoch": 6.972620599739244,
-      "grad_norm": 1.4475274085998535,
-      "learning_rate": 8.706934673366835e-05,
-      "loss": 5.8797,
-      "step": 13370
-    },
-    {
-      "epoch": 6.973142112125163,
-      "grad_norm": 1.4043692350387573,
-      "learning_rate": 8.706834170854273e-05,
-      "loss": 5.8034,
-      "step": 13371
-    },
-    {
-      "epoch": 6.973663624511082,
-      "grad_norm": 1.3142307996749878,
-      "learning_rate": 8.706733668341709e-05,
-      "loss": 5.5609,
-      "step": 13372
-    },
-    {
-      "epoch": 6.974185136897002,
-      "grad_norm": 1.4457515478134155,
-      "learning_rate": 8.706633165829147e-05,
-      "loss": 5.459,
-      "step": 13373
-    },
-    {
-      "epoch": 6.97470664928292,
-      "grad_norm": 1.3880631923675537,
-      "learning_rate": 8.706532663316583e-05,
-      "loss": 5.5104,
-      "step": 13374
-    },
-    {
-      "epoch": 6.97522816166884,
-      "grad_norm": 1.508703589439392,
-      "learning_rate": 8.70643216080402e-05,
-      "loss": 5.5321,
-      "step": 13375
-    },
-    {
-      "epoch": 6.975749674054759,
-      "grad_norm": 1.346994161605835,
-      "learning_rate": 8.706331658291458e-05,
-      "loss": 5.8971,
-      "step": 13376
-    },
-    {
-      "epoch": 6.976271186440678,
-      "grad_norm": 1.3925755023956299,
-      "learning_rate": 8.706231155778894e-05,
-      "loss": 5.4844,
-      "step": 13377
-    },
-    {
-      "epoch": 6.976792698826597,
-      "grad_norm": 1.409274697303772,
-      "learning_rate": 8.706130653266332e-05,
-      "loss": 5.3255,
-      "step": 13378
-    },
-    {
-      "epoch": 6.977314211212517,
-      "grad_norm": 1.4572656154632568,
-      "learning_rate": 8.706030150753768e-05,
-      "loss": 5.5552,
-      "step": 13379
-    },
-    {
-      "epoch": 6.977835723598435,
-      "grad_norm": 1.3959450721740723,
-      "learning_rate": 8.705929648241206e-05,
-      "loss": 5.5771,
-      "step": 13380
-    },
-    {
-      "epoch": 6.978357235984355,
-      "grad_norm": 1.3494313955307007,
-      "learning_rate": 8.705829145728644e-05,
-      "loss": 5.6901,
-      "step": 13381
-    },
-    {
-      "epoch": 6.978878748370274,
-      "grad_norm": 1.3455893993377686,
-      "learning_rate": 8.705728643216082e-05,
-      "loss": 5.424,
-      "step": 13382
-    },
-    {
-      "epoch": 6.979400260756193,
-      "grad_norm": 1.5395665168762207,
-      "learning_rate": 8.705628140703518e-05,
-      "loss": 5.8227,
-      "step": 13383
-    },
-    {
-      "epoch": 6.979921773142112,
-      "grad_norm": 1.3639477491378784,
-      "learning_rate": 8.705527638190956e-05,
-      "loss": 5.9073,
-      "step": 13384
-    },
-    {
-      "epoch": 6.980443285528032,
-      "grad_norm": 1.3254749774932861,
-      "learning_rate": 8.705427135678392e-05,
-      "loss": 5.5652,
-      "step": 13385
-    },
-    {
-      "epoch": 6.98096479791395,
-      "grad_norm": 1.4098268747329712,
-      "learning_rate": 8.70532663316583e-05,
-      "loss": 5.5309,
-      "step": 13386
-    },
-    {
-      "epoch": 6.98148631029987,
-      "grad_norm": 1.4973217248916626,
-      "learning_rate": 8.705226130653267e-05,
-      "loss": 5.689,
-      "step": 13387
-    },
-    {
-      "epoch": 6.982007822685789,
-      "grad_norm": 1.577728271484375,
-      "learning_rate": 8.705125628140703e-05,
-      "loss": 5.507,
-      "step": 13388
-    },
-    {
-      "epoch": 6.982529335071708,
-      "grad_norm": 1.4927242994308472,
-      "learning_rate": 8.705025125628141e-05,
-      "loss": 5.5177,
-      "step": 13389
-    },
-    {
-      "epoch": 6.983050847457627,
-      "grad_norm": 1.638493537902832,
-      "learning_rate": 8.704924623115579e-05,
-      "loss": 5.5891,
-      "step": 13390
-    },
-    {
-      "epoch": 6.983572359843547,
-      "grad_norm": 1.4359697103500366,
-      "learning_rate": 8.704824120603016e-05,
-      "loss": 5.4611,
-      "step": 13391
-    },
-    {
-      "epoch": 6.984093872229465,
-      "grad_norm": 1.3334813117980957,
-      "learning_rate": 8.704723618090453e-05,
-      "loss": 5.8773,
-      "step": 13392
-    },
-    {
-      "epoch": 6.984615384615385,
-      "grad_norm": 1.4892915487289429,
-      "learning_rate": 8.70462311557789e-05,
-      "loss": 5.6921,
-      "step": 13393
-    },
-    {
-      "epoch": 6.985136897001304,
-      "grad_norm": 1.5081355571746826,
-      "learning_rate": 8.704522613065327e-05,
-      "loss": 5.2545,
-      "step": 13394
-    },
-    {
-      "epoch": 6.985658409387223,
-      "grad_norm": 1.355001449584961,
-      "learning_rate": 8.704422110552765e-05,
-      "loss": 5.555,
-      "step": 13395
-    },
-    {
-      "epoch": 6.986179921773142,
-      "grad_norm": 1.3713586330413818,
-      "learning_rate": 8.704321608040201e-05,
-      "loss": 5.6285,
-      "step": 13396
-    },
-    {
-      "epoch": 6.986701434159062,
-      "grad_norm": 1.4931268692016602,
-      "learning_rate": 8.704221105527639e-05,
-      "loss": 5.6208,
-      "step": 13397
-    },
-    {
-      "epoch": 6.98722294654498,
-      "grad_norm": 1.436623215675354,
-      "learning_rate": 8.704120603015075e-05,
-      "loss": 5.6248,
-      "step": 13398
-    },
-    {
-      "epoch": 6.9877444589309,
-      "grad_norm": 1.3965542316436768,
-      "learning_rate": 8.704020100502513e-05,
-      "loss": 5.8099,
-      "step": 13399
-    },
-    {
-      "epoch": 6.988265971316819,
-      "grad_norm": 1.3200595378875732,
-      "learning_rate": 8.70391959798995e-05,
-      "loss": 5.9473,
-      "step": 13400
-    },
-    {
-      "epoch": 6.988787483702738,
-      "grad_norm": 1.465320110321045,
-      "learning_rate": 8.703819095477387e-05,
-      "loss": 5.6862,
-      "step": 13401
-    },
-    {
-      "epoch": 6.989308996088657,
-      "grad_norm": 1.3251965045928955,
-      "learning_rate": 8.703718592964825e-05,
-      "loss": 6.0016,
-      "step": 13402
-    },
-    {
-      "epoch": 6.989830508474577,
-      "grad_norm": 1.3845266103744507,
-      "learning_rate": 8.703618090452262e-05,
-      "loss": 5.4742,
-      "step": 13403
-    },
-    {
-      "epoch": 6.990352020860495,
-      "grad_norm": 1.3564001321792603,
-      "learning_rate": 8.7035175879397e-05,
-      "loss": 5.9768,
-      "step": 13404
-    },
-    {
-      "epoch": 6.990873533246415,
-      "grad_norm": 1.4856059551239014,
-      "learning_rate": 8.703417085427136e-05,
-      "loss": 5.3674,
-      "step": 13405
-    },
-    {
-      "epoch": 6.991395045632334,
-      "grad_norm": 1.3881319761276245,
-      "learning_rate": 8.703316582914574e-05,
-      "loss": 5.6545,
-      "step": 13406
-    },
-    {
-      "epoch": 6.991916558018253,
-      "grad_norm": 1.5049464702606201,
-      "learning_rate": 8.70321608040201e-05,
-      "loss": 5.6046,
-      "step": 13407
-    },
-    {
-      "epoch": 6.992438070404172,
-      "grad_norm": 1.3670753240585327,
-      "learning_rate": 8.703115577889448e-05,
-      "loss": 5.6352,
-      "step": 13408
-    },
-    {
-      "epoch": 6.992959582790091,
-      "grad_norm": 1.3510066270828247,
-      "learning_rate": 8.703015075376884e-05,
-      "loss": 5.8483,
-      "step": 13409
-    },
-    {
-      "epoch": 6.99348109517601,
-      "grad_norm": 1.4152847528457642,
-      "learning_rate": 8.702914572864322e-05,
-      "loss": 5.5295,
-      "step": 13410
-    },
-    {
-      "epoch": 6.99400260756193,
-      "grad_norm": 1.3526033163070679,
-      "learning_rate": 8.70281407035176e-05,
-      "loss": 5.8497,
-      "step": 13411
-    },
-    {
-      "epoch": 6.994524119947849,
-      "grad_norm": 1.4955945014953613,
-      "learning_rate": 8.702713567839198e-05,
-      "loss": 5.5484,
-      "step": 13412
-    },
-    {
-      "epoch": 6.995045632333768,
-      "grad_norm": 1.4018276929855347,
-      "learning_rate": 8.702613065326634e-05,
-      "loss": 5.9126,
-      "step": 13413
-    },
-    {
-      "epoch": 6.995567144719687,
-      "grad_norm": 1.5255998373031616,
-      "learning_rate": 8.70251256281407e-05,
-      "loss": 5.7088,
-      "step": 13414
-    },
-    {
-      "epoch": 6.996088657105606,
-      "grad_norm": 1.4157768487930298,
-      "learning_rate": 8.702412060301508e-05,
-      "loss": 5.5033,
-      "step": 13415
-    },
-    {
-      "epoch": 6.996610169491525,
-      "grad_norm": 1.5596543550491333,
-      "learning_rate": 8.702311557788945e-05,
-      "loss": 5.1652,
-      "step": 13416
-    },
-    {
-      "epoch": 6.997131681877445,
-      "grad_norm": 1.5296235084533691,
-      "learning_rate": 8.702211055276382e-05,
-      "loss": 4.7977,
-      "step": 13417
-    },
-    {
-      "epoch": 6.997653194263364,
-      "grad_norm": 1.4895062446594238,
-      "learning_rate": 8.702110552763819e-05,
-      "loss": 5.6861,
-      "step": 13418
-    },
-    {
-      "epoch": 6.998174706649283,
-      "grad_norm": 1.4128345251083374,
-      "learning_rate": 8.702010050251257e-05,
-      "loss": 5.6152,
-      "step": 13419
-    },
-    {
-      "epoch": 6.998696219035202,
-      "grad_norm": 1.6972578763961792,
-      "learning_rate": 8.701909547738693e-05,
-      "loss": 5.0183,
-      "step": 13420
-    },
-    {
-      "epoch": 6.999217731421121,
-      "grad_norm": 1.426389217376709,
-      "learning_rate": 8.701809045226131e-05,
-      "loss": 5.5888,
-      "step": 13421
-    },
-    {
-      "epoch": 6.99973924380704,
-      "grad_norm": 1.4051347970962524,
-      "learning_rate": 8.701708542713569e-05,
-      "loss": 5.4924,
-      "step": 13422
-    },
-    {
-      "epoch": 7.00026075619296,
-      "grad_norm": 1.8485748767852783,
-      "learning_rate": 8.701608040201006e-05,
-      "loss": 5.6172,
-      "step": 13423
-    },
-    {
-      "epoch": 7.000782268578878,
-      "grad_norm": 1.4821194410324097,
-      "learning_rate": 8.701507537688443e-05,
-      "loss": 5.2798,
-      "step": 13424
-    },
-    {
-      "epoch": 7.001303780964798,
-      "grad_norm": 1.610002040863037,
-      "learning_rate": 8.70140703517588e-05,
-      "loss": 5.4295,
-      "step": 13425
-    },
-    {
-      "epoch": 7.001825293350717,
-      "grad_norm": 1.4609360694885254,
-      "learning_rate": 8.701306532663317e-05,
-      "loss": 5.311,
-      "step": 13426
-    },
-    {
-      "epoch": 7.002346805736636,
-      "grad_norm": 1.4490759372711182,
-      "learning_rate": 8.701206030150753e-05,
-      "loss": 5.0553,
-      "step": 13427
-    },
-    {
-      "epoch": 7.002868318122555,
-      "grad_norm": 1.7876455783843994,
-      "learning_rate": 8.701105527638191e-05,
-      "loss": 5.6138,
-      "step": 13428
-    },
-    {
-      "epoch": 7.003389830508475,
-      "grad_norm": 1.4610207080841064,
-      "learning_rate": 8.701005025125628e-05,
-      "loss": 5.5573,
-      "step": 13429
-    },
-    {
-      "epoch": 7.003911342894393,
-      "grad_norm": 1.4007673263549805,
-      "learning_rate": 8.700904522613065e-05,
-      "loss": 5.4963,
-      "step": 13430
-    },
-    {
-      "epoch": 7.004432855280313,
-      "grad_norm": 1.5872492790222168,
-      "learning_rate": 8.700804020100503e-05,
-      "loss": 5.518,
-      "step": 13431
-    },
-    {
-      "epoch": 7.004954367666232,
-      "grad_norm": 1.4385663270950317,
-      "learning_rate": 8.700703517587941e-05,
-      "loss": 5.2833,
-      "step": 13432
-    },
-    {
-      "epoch": 7.005475880052151,
-      "grad_norm": 1.425424575805664,
-      "learning_rate": 8.700603015075377e-05,
-      "loss": 5.4996,
-      "step": 13433
-    },
-    {
-      "epoch": 7.00599739243807,
-      "grad_norm": 1.5454621315002441,
-      "learning_rate": 8.700502512562815e-05,
-      "loss": 5.3359,
-      "step": 13434
-    },
-    {
-      "epoch": 7.00651890482399,
-      "grad_norm": 1.3315880298614502,
-      "learning_rate": 8.700402010050252e-05,
-      "loss": 5.838,
-      "step": 13435
-    },
-    {
-      "epoch": 7.007040417209908,
-      "grad_norm": 1.322046160697937,
-      "learning_rate": 8.70030150753769e-05,
-      "loss": 5.8882,
-      "step": 13436
-    },
-    {
-      "epoch": 7.007561929595828,
-      "grad_norm": 1.3955097198486328,
-      "learning_rate": 8.700201005025126e-05,
-      "loss": 5.7727,
-      "step": 13437
-    },
-    {
-      "epoch": 7.008083441981747,
-      "grad_norm": 1.434022068977356,
-      "learning_rate": 8.700100502512564e-05,
-      "loss": 5.4132,
-      "step": 13438
-    },
-    {
-      "epoch": 7.008604954367666,
-      "grad_norm": 1.4591773748397827,
-      "learning_rate": 8.7e-05,
-      "loss": 5.4758,
-      "step": 13439
-    },
-    {
-      "epoch": 7.009126466753585,
-      "grad_norm": 1.4431689977645874,
-      "learning_rate": 8.699899497487438e-05,
-      "loss": 5.9166,
-      "step": 13440
-    },
-    {
-      "epoch": 7.009647979139505,
-      "grad_norm": 1.3714346885681152,
-      "learning_rate": 8.699798994974874e-05,
-      "loss": 5.9118,
-      "step": 13441
-    },
-    {
-      "epoch": 7.010169491525423,
-      "grad_norm": 1.4963582754135132,
-      "learning_rate": 8.699698492462312e-05,
-      "loss": 5.6545,
-      "step": 13442
-    },
-    {
-      "epoch": 7.010691003911343,
-      "grad_norm": 1.4886062145233154,
-      "learning_rate": 8.69959798994975e-05,
-      "loss": 5.4472,
-      "step": 13443
-    },
-    {
-      "epoch": 7.011212516297262,
-      "grad_norm": 1.5133861303329468,
-      "learning_rate": 8.699497487437186e-05,
-      "loss": 5.708,
-      "step": 13444
-    },
-    {
-      "epoch": 7.011734028683181,
-      "grad_norm": 1.6591498851776123,
-      "learning_rate": 8.699396984924624e-05,
-      "loss": 5.3799,
-      "step": 13445
-    },
-    {
-      "epoch": 7.0122555410691,
-      "grad_norm": 1.5151041746139526,
-      "learning_rate": 8.69929648241206e-05,
-      "loss": 5.5876,
-      "step": 13446
-    },
-    {
-      "epoch": 7.01277705345502,
-      "grad_norm": 1.502970576286316,
-      "learning_rate": 8.699195979899498e-05,
-      "loss": 5.3159,
-      "step": 13447
-    },
-    {
-      "epoch": 7.013298565840938,
-      "grad_norm": 1.4799509048461914,
-      "learning_rate": 8.699095477386935e-05,
-      "loss": 5.8265,
-      "step": 13448
-    },
-    {
-      "epoch": 7.013820078226858,
-      "grad_norm": 1.3858861923217773,
-      "learning_rate": 8.698994974874372e-05,
-      "loss": 5.8121,
-      "step": 13449
-    },
-    {
-      "epoch": 7.014341590612777,
-      "grad_norm": 1.400848388671875,
-      "learning_rate": 8.698894472361809e-05,
-      "loss": 5.3686,
-      "step": 13450
-    },
-    {
-      "epoch": 7.014863102998696,
-      "grad_norm": 1.4534039497375488,
-      "learning_rate": 8.698793969849247e-05,
-      "loss": 5.6712,
-      "step": 13451
-    },
-    {
-      "epoch": 7.015384615384615,
-      "grad_norm": 1.4710307121276855,
-      "learning_rate": 8.698693467336684e-05,
-      "loss": 5.8047,
-      "step": 13452
-    },
-    {
-      "epoch": 7.015906127770535,
-      "grad_norm": 1.4461052417755127,
-      "learning_rate": 8.698592964824122e-05,
-      "loss": 5.6924,
-      "step": 13453
-    },
-    {
-      "epoch": 7.016427640156453,
-      "grad_norm": 1.4467244148254395,
-      "learning_rate": 8.698492462311559e-05,
-      "loss": 5.7575,
-      "step": 13454
-    },
-    {
-      "epoch": 7.016949152542373,
-      "grad_norm": 1.2938076257705688,
-      "learning_rate": 8.698391959798995e-05,
-      "loss": 5.3051,
-      "step": 13455
-    },
-    {
-      "epoch": 7.017470664928292,
-      "grad_norm": 1.4672337770462036,
-      "learning_rate": 8.698291457286433e-05,
-      "loss": 5.6897,
-      "step": 13456
-    },
-    {
-      "epoch": 7.017992177314211,
-      "grad_norm": 1.4397361278533936,
-      "learning_rate": 8.698190954773869e-05,
-      "loss": 5.7627,
-      "step": 13457
-    },
-    {
-      "epoch": 7.01851368970013,
-      "grad_norm": 1.5544052124023438,
-      "learning_rate": 8.698090452261307e-05,
-      "loss": 5.0664,
-      "step": 13458
-    },
-    {
-      "epoch": 7.01903520208605,
-      "grad_norm": 1.4260209798812866,
-      "learning_rate": 8.697989949748744e-05,
-      "loss": 5.7323,
-      "step": 13459
-    },
-    {
-      "epoch": 7.019556714471968,
-      "grad_norm": 1.4896206855773926,
-      "learning_rate": 8.697889447236181e-05,
-      "loss": 5.6959,
-      "step": 13460
-    },
-    {
-      "epoch": 7.020078226857888,
-      "grad_norm": 1.3778749704360962,
-      "learning_rate": 8.697788944723618e-05,
-      "loss": 5.0909,
-      "step": 13461
-    },
-    {
-      "epoch": 7.020599739243807,
-      "grad_norm": 1.3835723400115967,
-      "learning_rate": 8.697688442211056e-05,
-      "loss": 5.4147,
-      "step": 13462
-    },
-    {
-      "epoch": 7.021121251629726,
-      "grad_norm": 1.4496538639068604,
-      "learning_rate": 8.697587939698493e-05,
-      "loss": 5.148,
-      "step": 13463
-    },
-    {
-      "epoch": 7.021642764015645,
-      "grad_norm": 1.460734486579895,
-      "learning_rate": 8.697487437185931e-05,
-      "loss": 5.3684,
-      "step": 13464
-    },
-    {
-      "epoch": 7.022164276401565,
-      "grad_norm": 1.3311572074890137,
-      "learning_rate": 8.697386934673367e-05,
-      "loss": 5.7251,
-      "step": 13465
-    },
-    {
-      "epoch": 7.022685788787483,
-      "grad_norm": 1.3198803663253784,
-      "learning_rate": 8.697286432160805e-05,
-      "loss": 5.4769,
-      "step": 13466
-    },
-    {
-      "epoch": 7.023207301173403,
-      "grad_norm": 1.4641104936599731,
-      "learning_rate": 8.697185929648242e-05,
-      "loss": 5.5742,
-      "step": 13467
-    },
-    {
-      "epoch": 7.023728813559322,
-      "grad_norm": 1.3770196437835693,
-      "learning_rate": 8.697085427135678e-05,
-      "loss": 5.8993,
-      "step": 13468
-    },
-    {
-      "epoch": 7.024250325945241,
-      "grad_norm": 1.4775896072387695,
-      "learning_rate": 8.696984924623116e-05,
-      "loss": 5.5317,
-      "step": 13469
-    },
-    {
-      "epoch": 7.02477183833116,
-      "grad_norm": 1.4371564388275146,
-      "learning_rate": 8.696884422110552e-05,
-      "loss": 5.4453,
-      "step": 13470
-    },
-    {
-      "epoch": 7.02529335071708,
-      "grad_norm": 1.4484115839004517,
-      "learning_rate": 8.69678391959799e-05,
-      "loss": 5.3682,
-      "step": 13471
-    },
-    {
-      "epoch": 7.025814863102998,
-      "grad_norm": 1.4084595441818237,
-      "learning_rate": 8.696683417085427e-05,
-      "loss": 5.4254,
-      "step": 13472
-    },
-    {
-      "epoch": 7.026336375488918,
-      "grad_norm": 1.4904628992080688,
-      "learning_rate": 8.696582914572864e-05,
-      "loss": 5.5454,
-      "step": 13473
-    },
-    {
-      "epoch": 7.026857887874837,
-      "grad_norm": 1.6426067352294922,
-      "learning_rate": 8.696482412060302e-05,
-      "loss": 5.2355,
-      "step": 13474
-    },
-    {
-      "epoch": 7.027379400260756,
-      "grad_norm": 1.4853923320770264,
-      "learning_rate": 8.69638190954774e-05,
-      "loss": 5.5352,
-      "step": 13475
-    },
-    {
-      "epoch": 7.027900912646675,
-      "grad_norm": 1.371253252029419,
-      "learning_rate": 8.696281407035176e-05,
-      "loss": 5.7212,
-      "step": 13476
-    },
-    {
-      "epoch": 7.028422425032595,
-      "grad_norm": 1.5297445058822632,
-      "learning_rate": 8.696180904522614e-05,
-      "loss": 5.3758,
-      "step": 13477
-    },
-    {
-      "epoch": 7.028943937418513,
-      "grad_norm": 1.5484615564346313,
-      "learning_rate": 8.69608040201005e-05,
-      "loss": 5.1708,
-      "step": 13478
-    },
-    {
-      "epoch": 7.029465449804433,
-      "grad_norm": 1.3931173086166382,
-      "learning_rate": 8.695979899497488e-05,
-      "loss": 5.7852,
-      "step": 13479
-    },
-    {
-      "epoch": 7.029986962190352,
-      "grad_norm": 1.4016035795211792,
-      "learning_rate": 8.695879396984925e-05,
-      "loss": 5.7639,
-      "step": 13480
-    },
-    {
-      "epoch": 7.030508474576271,
-      "grad_norm": 1.443292260169983,
-      "learning_rate": 8.695778894472361e-05,
-      "loss": 5.4668,
-      "step": 13481
-    },
-    {
-      "epoch": 7.03102998696219,
-      "grad_norm": 1.3710988759994507,
-      "learning_rate": 8.695678391959799e-05,
-      "loss": 5.8426,
-      "step": 13482
-    },
-    {
-      "epoch": 7.03155149934811,
-      "grad_norm": 1.5292260646820068,
-      "learning_rate": 8.695577889447237e-05,
-      "loss": 5.4736,
-      "step": 13483
-    },
-    {
-      "epoch": 7.032073011734028,
-      "grad_norm": 1.471585750579834,
-      "learning_rate": 8.695477386934675e-05,
-      "loss": 5.0398,
-      "step": 13484
-    },
-    {
-      "epoch": 7.032594524119948,
-      "grad_norm": 1.4033136367797852,
-      "learning_rate": 8.695376884422111e-05,
-      "loss": 5.7082,
-      "step": 13485
-    },
-    {
-      "epoch": 7.033116036505867,
-      "grad_norm": 1.3297590017318726,
-      "learning_rate": 8.695276381909549e-05,
-      "loss": 5.1385,
-      "step": 13486
-    },
-    {
-      "epoch": 7.033637548891786,
-      "grad_norm": 1.3944824934005737,
-      "learning_rate": 8.695175879396985e-05,
-      "loss": 5.7619,
-      "step": 13487
-    },
-    {
-      "epoch": 7.034159061277705,
-      "grad_norm": 1.3541886806488037,
-      "learning_rate": 8.695075376884423e-05,
-      "loss": 5.3419,
-      "step": 13488
-    },
-    {
-      "epoch": 7.034680573663625,
-      "grad_norm": 1.4420626163482666,
-      "learning_rate": 8.69497487437186e-05,
-      "loss": 5.7486,
-      "step": 13489
-    },
-    {
-      "epoch": 7.0352020860495434,
-      "grad_norm": 1.4421377182006836,
-      "learning_rate": 8.694874371859297e-05,
-      "loss": 5.756,
-      "step": 13490
-    },
-    {
-      "epoch": 7.035723598435463,
-      "grad_norm": 1.3632266521453857,
-      "learning_rate": 8.694773869346734e-05,
-      "loss": 5.7661,
-      "step": 13491
-    },
-    {
-      "epoch": 7.036245110821382,
-      "grad_norm": 1.4393789768218994,
-      "learning_rate": 8.694673366834171e-05,
-      "loss": 5.3793,
-      "step": 13492
-    },
-    {
-      "epoch": 7.036766623207301,
-      "grad_norm": 1.511124849319458,
-      "learning_rate": 8.694572864321608e-05,
-      "loss": 5.7117,
-      "step": 13493
-    },
-    {
-      "epoch": 7.03728813559322,
-      "grad_norm": 1.382969617843628,
-      "learning_rate": 8.694472361809046e-05,
-      "loss": 5.7368,
-      "step": 13494
-    },
-    {
-      "epoch": 7.03780964797914,
-      "grad_norm": 1.4124302864074707,
-      "learning_rate": 8.694371859296483e-05,
-      "loss": 5.9784,
-      "step": 13495
-    },
-    {
-      "epoch": 7.0383311603650585,
-      "grad_norm": 1.3480901718139648,
-      "learning_rate": 8.69427135678392e-05,
-      "loss": 5.4335,
-      "step": 13496
-    },
-    {
-      "epoch": 7.038852672750978,
-      "grad_norm": 1.4954389333724976,
-      "learning_rate": 8.694170854271358e-05,
-      "loss": 5.4678,
-      "step": 13497
-    },
-    {
-      "epoch": 7.039374185136897,
-      "grad_norm": 1.579001784324646,
-      "learning_rate": 8.694070351758794e-05,
-      "loss": 5.4604,
-      "step": 13498
-    },
-    {
-      "epoch": 7.039895697522816,
-      "grad_norm": 1.4330686330795288,
-      "learning_rate": 8.693969849246232e-05,
-      "loss": 5.6495,
-      "step": 13499
-    },
-    {
-      "epoch": 7.040417209908735,
-      "grad_norm": 1.358410120010376,
-      "learning_rate": 8.693869346733668e-05,
-      "loss": 5.8587,
-      "step": 13500
-    },
-    {
-      "epoch": 7.040938722294655,
-      "grad_norm": 1.4798851013183594,
-      "learning_rate": 8.693768844221106e-05,
-      "loss": 5.0451,
-      "step": 13501
-    },
-    {
-      "epoch": 7.0414602346805735,
-      "grad_norm": 1.5255372524261475,
-      "learning_rate": 8.693668341708542e-05,
-      "loss": 5.1084,
-      "step": 13502
-    },
-    {
-      "epoch": 7.041981747066493,
-      "grad_norm": 1.4601895809173584,
-      "learning_rate": 8.69356783919598e-05,
-      "loss": 5.433,
-      "step": 13503
-    },
-    {
-      "epoch": 7.042503259452412,
-      "grad_norm": 1.4498512744903564,
-      "learning_rate": 8.693467336683418e-05,
-      "loss": 5.7736,
-      "step": 13504
-    },
-    {
-      "epoch": 7.043024771838331,
-      "grad_norm": 1.4497058391571045,
-      "learning_rate": 8.693366834170856e-05,
-      "loss": 5.009,
-      "step": 13505
-    },
-    {
-      "epoch": 7.04354628422425,
-      "grad_norm": 1.3038369417190552,
-      "learning_rate": 8.693266331658292e-05,
-      "loss": 5.8339,
-      "step": 13506
-    },
-    {
-      "epoch": 7.04406779661017,
-      "grad_norm": 1.395769715309143,
-      "learning_rate": 8.693165829145729e-05,
-      "loss": 5.7608,
-      "step": 13507
-    },
-    {
-      "epoch": 7.0445893089960885,
-      "grad_norm": 1.4769400358200073,
-      "learning_rate": 8.693065326633166e-05,
-      "loss": 5.5145,
-      "step": 13508
-    },
-    {
-      "epoch": 7.045110821382008,
-      "grad_norm": 1.4044547080993652,
-      "learning_rate": 8.692964824120603e-05,
-      "loss": 5.7547,
-      "step": 13509
-    },
-    {
-      "epoch": 7.045632333767927,
-      "grad_norm": 1.4802017211914062,
-      "learning_rate": 8.69286432160804e-05,
-      "loss": 5.116,
-      "step": 13510
-    },
-    {
-      "epoch": 7.046153846153846,
-      "grad_norm": 1.4355244636535645,
-      "learning_rate": 8.692763819095477e-05,
-      "loss": 5.7684,
-      "step": 13511
-    },
-    {
-      "epoch": 7.0466753585397655,
-      "grad_norm": 1.5294073820114136,
-      "learning_rate": 8.692663316582915e-05,
-      "loss": 5.6104,
-      "step": 13512
-    },
-    {
-      "epoch": 7.047196870925685,
-      "grad_norm": 1.4788920879364014,
-      "learning_rate": 8.692562814070351e-05,
-      "loss": 5.6739,
-      "step": 13513
-    },
-    {
-      "epoch": 7.0477183833116035,
-      "grad_norm": 1.4485869407653809,
-      "learning_rate": 8.692462311557789e-05,
-      "loss": 5.8912,
-      "step": 13514
-    },
-    {
-      "epoch": 7.048239895697523,
-      "grad_norm": 1.4866095781326294,
-      "learning_rate": 8.692361809045227e-05,
-      "loss": 5.3231,
-      "step": 13515
-    },
-    {
-      "epoch": 7.0487614080834415,
-      "grad_norm": 1.4245738983154297,
-      "learning_rate": 8.692261306532665e-05,
-      "loss": 5.608,
-      "step": 13516
-    },
-    {
-      "epoch": 7.049282920469361,
-      "grad_norm": 1.5063649415969849,
-      "learning_rate": 8.692160804020101e-05,
-      "loss": 5.2061,
-      "step": 13517
-    },
-    {
-      "epoch": 7.0498044328552805,
-      "grad_norm": 1.6474918127059937,
-      "learning_rate": 8.692060301507539e-05,
-      "loss": 5.2798,
-      "step": 13518
-    },
-    {
-      "epoch": 7.050325945241199,
-      "grad_norm": 1.2920011281967163,
-      "learning_rate": 8.691959798994975e-05,
-      "loss": 5.1269,
-      "step": 13519
-    },
-    {
-      "epoch": 7.0508474576271185,
-      "grad_norm": 1.4522672891616821,
-      "learning_rate": 8.691859296482412e-05,
-      "loss": 5.5168,
-      "step": 13520
-    },
-    {
-      "epoch": 7.051368970013038,
-      "grad_norm": 1.417240858078003,
-      "learning_rate": 8.69175879396985e-05,
-      "loss": 5.5638,
-      "step": 13521
-    },
-    {
-      "epoch": 7.0518904823989566,
-      "grad_norm": 1.3392002582550049,
-      "learning_rate": 8.691658291457286e-05,
-      "loss": 5.0803,
-      "step": 13522
-    },
-    {
-      "epoch": 7.052411994784876,
-      "grad_norm": 1.4109644889831543,
-      "learning_rate": 8.691557788944724e-05,
-      "loss": 5.7814,
-      "step": 13523
-    },
-    {
-      "epoch": 7.0529335071707955,
-      "grad_norm": 1.3801614046096802,
-      "learning_rate": 8.691457286432161e-05,
-      "loss": 5.2116,
-      "step": 13524
-    },
-    {
-      "epoch": 7.053455019556714,
-      "grad_norm": 1.4768915176391602,
-      "learning_rate": 8.691356783919599e-05,
-      "loss": 5.0764,
-      "step": 13525
-    },
-    {
-      "epoch": 7.0539765319426335,
-      "grad_norm": 1.4160035848617554,
-      "learning_rate": 8.691256281407036e-05,
-      "loss": 5.5037,
-      "step": 13526
-    },
-    {
-      "epoch": 7.054498044328553,
-      "grad_norm": 1.4442335367202759,
-      "learning_rate": 8.691155778894473e-05,
-      "loss": 5.4608,
-      "step": 13527
-    },
-    {
-      "epoch": 7.055019556714472,
-      "grad_norm": 1.4763801097869873,
-      "learning_rate": 8.69105527638191e-05,
-      "loss": 5.323,
-      "step": 13528
-    },
-    {
-      "epoch": 7.055541069100391,
-      "grad_norm": 1.3409476280212402,
-      "learning_rate": 8.690954773869348e-05,
-      "loss": 5.5344,
-      "step": 13529
-    },
-    {
-      "epoch": 7.0560625814863105,
-      "grad_norm": 1.46872878074646,
-      "learning_rate": 8.690854271356784e-05,
-      "loss": 5.0455,
-      "step": 13530
-    },
-    {
-      "epoch": 7.056584093872229,
-      "grad_norm": 1.4129635095596313,
-      "learning_rate": 8.690753768844222e-05,
-      "loss": 5.5784,
-      "step": 13531
-    },
-    {
-      "epoch": 7.0571056062581485,
-      "grad_norm": 1.2868163585662842,
-      "learning_rate": 8.690653266331658e-05,
-      "loss": 4.9813,
-      "step": 13532
-    },
-    {
-      "epoch": 7.057627118644068,
-      "grad_norm": 1.399598240852356,
-      "learning_rate": 8.690552763819096e-05,
-      "loss": 5.0879,
-      "step": 13533
-    },
-    {
-      "epoch": 7.058148631029987,
-      "grad_norm": 1.3907204866409302,
-      "learning_rate": 8.690452261306532e-05,
-      "loss": 5.6694,
-      "step": 13534
-    },
-    {
-      "epoch": 7.058670143415906,
-      "grad_norm": 1.6101486682891846,
-      "learning_rate": 8.69035175879397e-05,
-      "loss": 5.1895,
-      "step": 13535
-    },
-    {
-      "epoch": 7.0591916558018255,
-      "grad_norm": 1.5695613622665405,
-      "learning_rate": 8.690251256281408e-05,
-      "loss": 5.6528,
-      "step": 13536
-    },
-    {
-      "epoch": 7.059713168187744,
-      "grad_norm": 1.4829063415527344,
-      "learning_rate": 8.690150753768844e-05,
-      "loss": 5.6244,
-      "step": 13537
-    },
-    {
-      "epoch": 7.0602346805736635,
-      "grad_norm": 1.4469066858291626,
-      "learning_rate": 8.690050251256282e-05,
-      "loss": 5.6093,
-      "step": 13538
-    },
-    {
-      "epoch": 7.060756192959583,
-      "grad_norm": 1.427062749862671,
-      "learning_rate": 8.689949748743719e-05,
-      "loss": 5.6679,
-      "step": 13539
-    },
-    {
-      "epoch": 7.061277705345502,
-      "grad_norm": 1.4258912801742554,
-      "learning_rate": 8.689849246231156e-05,
-      "loss": 5.6927,
-      "step": 13540
-    },
-    {
-      "epoch": 7.061799217731421,
-      "grad_norm": 1.3217735290527344,
-      "learning_rate": 8.689748743718593e-05,
-      "loss": 5.9511,
-      "step": 13541
-    },
-    {
-      "epoch": 7.0623207301173405,
-      "grad_norm": 1.3539990186691284,
-      "learning_rate": 8.689648241206031e-05,
-      "loss": 5.8172,
-      "step": 13542
-    },
-    {
-      "epoch": 7.062842242503259,
-      "grad_norm": 1.369462251663208,
-      "learning_rate": 8.689547738693467e-05,
-      "loss": 5.5092,
-      "step": 13543
-    },
-    {
-      "epoch": 7.063363754889179,
-      "grad_norm": 1.4070191383361816,
-      "learning_rate": 8.689447236180905e-05,
-      "loss": 5.423,
-      "step": 13544
-    },
-    {
-      "epoch": 7.063885267275098,
-      "grad_norm": 1.3957527875900269,
-      "learning_rate": 8.689346733668343e-05,
-      "loss": 5.6322,
-      "step": 13545
-    },
-    {
-      "epoch": 7.064406779661017,
-      "grad_norm": 1.4492450952529907,
-      "learning_rate": 8.68924623115578e-05,
-      "loss": 5.5996,
-      "step": 13546
-    },
-    {
-      "epoch": 7.064928292046936,
-      "grad_norm": 1.5721734762191772,
-      "learning_rate": 8.689145728643217e-05,
-      "loss": 5.325,
-      "step": 13547
-    },
-    {
-      "epoch": 7.0654498044328555,
-      "grad_norm": 1.4552375078201294,
-      "learning_rate": 8.689045226130653e-05,
-      "loss": 5.5795,
-      "step": 13548
-    },
-    {
-      "epoch": 7.065971316818774,
-      "grad_norm": 1.4627131223678589,
-      "learning_rate": 8.688944723618091e-05,
-      "loss": 5.6862,
-      "step": 13549
-    },
-    {
-      "epoch": 7.066492829204694,
-      "grad_norm": 1.556059718132019,
-      "learning_rate": 8.688844221105528e-05,
-      "loss": 5.3182,
-      "step": 13550
-    },
-    {
-      "epoch": 7.067014341590613,
-      "grad_norm": 1.5156433582305908,
-      "learning_rate": 8.688743718592965e-05,
-      "loss": 5.4054,
-      "step": 13551
-    },
-    {
-      "epoch": 7.067535853976532,
-      "grad_norm": 1.3692740201950073,
-      "learning_rate": 8.688643216080402e-05,
-      "loss": 5.9718,
-      "step": 13552
-    },
-    {
-      "epoch": 7.068057366362451,
-      "grad_norm": 1.3856106996536255,
-      "learning_rate": 8.68854271356784e-05,
-      "loss": 5.4612,
-      "step": 13553
-    },
-    {
-      "epoch": 7.0685788787483705,
-      "grad_norm": 1.4792388677597046,
-      "learning_rate": 8.688442211055276e-05,
-      "loss": 5.2844,
-      "step": 13554
-    },
-    {
-      "epoch": 7.069100391134289,
-      "grad_norm": 1.3618789911270142,
-      "learning_rate": 8.688341708542714e-05,
-      "loss": 5.5473,
-      "step": 13555
-    },
-    {
-      "epoch": 7.069621903520209,
-      "grad_norm": 1.6274526119232178,
-      "learning_rate": 8.688241206030152e-05,
-      "loss": 4.9174,
-      "step": 13556
-    },
-    {
-      "epoch": 7.070143415906128,
-      "grad_norm": 1.492628574371338,
-      "learning_rate": 8.688140703517589e-05,
-      "loss": 5.5222,
-      "step": 13557
-    },
-    {
-      "epoch": 7.070664928292047,
-      "grad_norm": 1.421893835067749,
-      "learning_rate": 8.688040201005026e-05,
-      "loss": 5.7627,
-      "step": 13558
-    },
-    {
-      "epoch": 7.071186440677966,
-      "grad_norm": 1.4931410551071167,
-      "learning_rate": 8.687939698492464e-05,
-      "loss": 5.8527,
-      "step": 13559
-    },
-    {
-      "epoch": 7.0717079530638856,
-      "grad_norm": 1.4539486169815063,
-      "learning_rate": 8.6878391959799e-05,
-      "loss": 5.8022,
-      "step": 13560
-    },
-    {
-      "epoch": 7.072229465449804,
-      "grad_norm": 1.4285629987716675,
-      "learning_rate": 8.687738693467336e-05,
-      "loss": 5.3471,
-      "step": 13561
-    },
-    {
-      "epoch": 7.072750977835724,
-      "grad_norm": 1.4815881252288818,
-      "learning_rate": 8.687638190954774e-05,
-      "loss": 5.7462,
-      "step": 13562
-    },
-    {
-      "epoch": 7.073272490221643,
-      "grad_norm": 1.5508919954299927,
-      "learning_rate": 8.68753768844221e-05,
-      "loss": 5.1894,
-      "step": 13563
-    },
-    {
-      "epoch": 7.073794002607562,
-      "grad_norm": 1.4936853647232056,
-      "learning_rate": 8.687437185929648e-05,
-      "loss": 5.6241,
-      "step": 13564
-    },
-    {
-      "epoch": 7.074315514993481,
-      "grad_norm": 1.5734385251998901,
-      "learning_rate": 8.687336683417086e-05,
-      "loss": 5.684,
-      "step": 13565
-    },
-    {
-      "epoch": 7.074837027379401,
-      "grad_norm": 1.3043161630630493,
-      "learning_rate": 8.687236180904524e-05,
-      "loss": 5.8608,
-      "step": 13566
-    },
-    {
-      "epoch": 7.075358539765319,
-      "grad_norm": 1.4227317571640015,
-      "learning_rate": 8.68713567839196e-05,
-      "loss": 5.0418,
-      "step": 13567
-    },
-    {
-      "epoch": 7.075880052151239,
-      "grad_norm": 1.2740521430969238,
-      "learning_rate": 8.687035175879398e-05,
-      "loss": 5.6072,
-      "step": 13568
-    },
-    {
-      "epoch": 7.076401564537158,
-      "grad_norm": 1.4511280059814453,
-      "learning_rate": 8.686934673366835e-05,
-      "loss": 5.6268,
-      "step": 13569
-    },
-    {
-      "epoch": 7.076923076923077,
-      "grad_norm": 1.455051302909851,
-      "learning_rate": 8.686834170854272e-05,
-      "loss": 5.6008,
-      "step": 13570
-    },
-    {
-      "epoch": 7.077444589308996,
-      "grad_norm": 1.5562998056411743,
-      "learning_rate": 8.686733668341709e-05,
-      "loss": 5.4041,
-      "step": 13571
-    },
-    {
-      "epoch": 7.077966101694916,
-      "grad_norm": 1.5050525665283203,
-      "learning_rate": 8.686633165829147e-05,
-      "loss": 5.6107,
-      "step": 13572
-    },
-    {
-      "epoch": 7.078487614080834,
-      "grad_norm": 1.4060953855514526,
-      "learning_rate": 8.686532663316583e-05,
-      "loss": 5.7125,
-      "step": 13573
-    },
-    {
-      "epoch": 7.079009126466754,
-      "grad_norm": 1.5273085832595825,
-      "learning_rate": 8.68643216080402e-05,
-      "loss": 4.9446,
-      "step": 13574
-    },
-    {
-      "epoch": 7.079530638852673,
-      "grad_norm": 1.5765935182571411,
-      "learning_rate": 8.686331658291457e-05,
-      "loss": 5.2055,
-      "step": 13575
-    },
-    {
-      "epoch": 7.080052151238592,
-      "grad_norm": 1.5833678245544434,
-      "learning_rate": 8.686231155778895e-05,
-      "loss": 5.3419,
-      "step": 13576
-    },
-    {
-      "epoch": 7.080573663624511,
-      "grad_norm": 1.6957190036773682,
-      "learning_rate": 8.686130653266333e-05,
-      "loss": 4.5964,
-      "step": 13577
-    },
-    {
-      "epoch": 7.081095176010431,
-      "grad_norm": 1.4957491159439087,
-      "learning_rate": 8.686030150753769e-05,
-      "loss": 5.6978,
-      "step": 13578
-    },
-    {
-      "epoch": 7.081616688396349,
-      "grad_norm": 1.4473893642425537,
-      "learning_rate": 8.685929648241207e-05,
-      "loss": 5.9766,
-      "step": 13579
-    },
-    {
-      "epoch": 7.082138200782269,
-      "grad_norm": 1.5708186626434326,
-      "learning_rate": 8.685829145728643e-05,
-      "loss": 5.2648,
-      "step": 13580
-    },
-    {
-      "epoch": 7.082659713168188,
-      "grad_norm": 1.546856164932251,
-      "learning_rate": 8.685728643216081e-05,
-      "loss": 5.5832,
-      "step": 13581
-    },
-    {
-      "epoch": 7.083181225554107,
-      "grad_norm": 1.6524760723114014,
-      "learning_rate": 8.685628140703518e-05,
-      "loss": 5.1632,
-      "step": 13582
-    },
-    {
-      "epoch": 7.083702737940026,
-      "grad_norm": 1.5229398012161255,
-      "learning_rate": 8.685527638190955e-05,
-      "loss": 5.5758,
-      "step": 13583
-    },
-    {
-      "epoch": 7.084224250325946,
-      "grad_norm": 1.4050939083099365,
-      "learning_rate": 8.685427135678392e-05,
-      "loss": 5.4897,
-      "step": 13584
-    },
-    {
-      "epoch": 7.084745762711864,
-      "grad_norm": 1.4276978969573975,
-      "learning_rate": 8.68532663316583e-05,
-      "loss": 5.7754,
-      "step": 13585
-    },
-    {
-      "epoch": 7.085267275097784,
-      "grad_norm": 1.5231478214263916,
-      "learning_rate": 8.685226130653267e-05,
-      "loss": 5.4932,
-      "step": 13586
-    },
-    {
-      "epoch": 7.085788787483703,
-      "grad_norm": 1.5633713006973267,
-      "learning_rate": 8.685125628140704e-05,
-      "loss": 5.1065,
-      "step": 13587
-    },
-    {
-      "epoch": 7.086310299869622,
-      "grad_norm": 1.30764639377594,
-      "learning_rate": 8.685025125628142e-05,
-      "loss": 5.5082,
-      "step": 13588
-    },
-    {
-      "epoch": 7.086831812255541,
-      "grad_norm": 1.5375173091888428,
-      "learning_rate": 8.684924623115578e-05,
-      "loss": 5.4468,
-      "step": 13589
-    },
-    {
-      "epoch": 7.087353324641461,
-      "grad_norm": 1.376344919204712,
-      "learning_rate": 8.684824120603016e-05,
-      "loss": 5.6734,
-      "step": 13590
-    },
-    {
-      "epoch": 7.087874837027379,
-      "grad_norm": 1.3691282272338867,
-      "learning_rate": 8.684723618090452e-05,
-      "loss": 5.6659,
-      "step": 13591
-    },
-    {
-      "epoch": 7.088396349413299,
-      "grad_norm": 1.4925811290740967,
-      "learning_rate": 8.68462311557789e-05,
-      "loss": 5.6942,
-      "step": 13592
-    },
-    {
-      "epoch": 7.088917861799218,
-      "grad_norm": 1.5294550657272339,
-      "learning_rate": 8.684522613065326e-05,
-      "loss": 5.2502,
-      "step": 13593
-    },
-    {
-      "epoch": 7.089439374185137,
-      "grad_norm": 1.4407625198364258,
-      "learning_rate": 8.684422110552764e-05,
-      "loss": 5.5743,
-      "step": 13594
-    },
-    {
-      "epoch": 7.089960886571056,
-      "grad_norm": 1.5639410018920898,
-      "learning_rate": 8.6843216080402e-05,
-      "loss": 5.3387,
-      "step": 13595
-    },
-    {
-      "epoch": 7.090482398956976,
-      "grad_norm": 1.6965670585632324,
-      "learning_rate": 8.684221105527638e-05,
-      "loss": 4.8863,
-      "step": 13596
-    },
-    {
-      "epoch": 7.091003911342894,
-      "grad_norm": 1.4019525051116943,
-      "learning_rate": 8.684120603015076e-05,
-      "loss": 5.8264,
-      "step": 13597
-    },
-    {
-      "epoch": 7.091525423728814,
-      "grad_norm": 1.390803337097168,
-      "learning_rate": 8.684020100502514e-05,
-      "loss": 5.7193,
-      "step": 13598
-    },
-    {
-      "epoch": 7.092046936114733,
-      "grad_norm": 1.3046754598617554,
-      "learning_rate": 8.68391959798995e-05,
-      "loss": 4.8235,
-      "step": 13599
-    },
-    {
-      "epoch": 7.092568448500652,
-      "grad_norm": 1.4319632053375244,
-      "learning_rate": 8.683819095477387e-05,
-      "loss": 5.7886,
-      "step": 13600
-    },
-    {
-      "epoch": 7.093089960886571,
-      "grad_norm": 1.4180150032043457,
-      "learning_rate": 8.683718592964825e-05,
-      "loss": 5.58,
-      "step": 13601
-    },
-    {
-      "epoch": 7.093611473272491,
-      "grad_norm": 1.4659533500671387,
-      "learning_rate": 8.683618090452261e-05,
-      "loss": 5.5936,
-      "step": 13602
-    },
-    {
-      "epoch": 7.094132985658409,
-      "grad_norm": 1.4651235342025757,
-      "learning_rate": 8.683517587939699e-05,
-      "loss": 5.6928,
-      "step": 13603
-    },
-    {
-      "epoch": 7.094654498044329,
-      "grad_norm": 1.5278723239898682,
-      "learning_rate": 8.683417085427135e-05,
-      "loss": 5.3064,
-      "step": 13604
-    },
-    {
-      "epoch": 7.095176010430248,
-      "grad_norm": 1.3299847841262817,
-      "learning_rate": 8.683316582914573e-05,
-      "loss": 5.0405,
-      "step": 13605
-    },
-    {
-      "epoch": 7.095697522816167,
-      "grad_norm": 1.595296025276184,
-      "learning_rate": 8.683216080402011e-05,
-      "loss": 5.4664,
-      "step": 13606
-    },
-    {
-      "epoch": 7.096219035202086,
-      "grad_norm": 1.4436908960342407,
-      "learning_rate": 8.683115577889449e-05,
-      "loss": 5.3747,
-      "step": 13607
-    },
-    {
-      "epoch": 7.096740547588006,
-      "grad_norm": 1.4779636859893799,
-      "learning_rate": 8.683015075376885e-05,
-      "loss": 5.4351,
-      "step": 13608
-    },
-    {
-      "epoch": 7.097262059973924,
-      "grad_norm": 1.3523204326629639,
-      "learning_rate": 8.682914572864323e-05,
-      "loss": 5.87,
-      "step": 13609
-    },
-    {
-      "epoch": 7.097783572359844,
-      "grad_norm": 1.7195734977722168,
-      "learning_rate": 8.682814070351759e-05,
-      "loss": 5.1955,
-      "step": 13610
-    },
-    {
-      "epoch": 7.098305084745762,
-      "grad_norm": 1.4393012523651123,
-      "learning_rate": 8.682713567839197e-05,
-      "loss": 5.7195,
-      "step": 13611
-    },
-    {
-      "epoch": 7.098826597131682,
-      "grad_norm": 1.3339797258377075,
-      "learning_rate": 8.682613065326633e-05,
-      "loss": 5.1458,
-      "step": 13612
-    },
-    {
-      "epoch": 7.099348109517601,
-      "grad_norm": 1.31460440158844,
-      "learning_rate": 8.68251256281407e-05,
-      "loss": 5.0231,
-      "step": 13613
-    },
-    {
-      "epoch": 7.09986962190352,
-      "grad_norm": 1.3422274589538574,
-      "learning_rate": 8.682412060301508e-05,
-      "loss": 5.889,
-      "step": 13614
-    },
-    {
-      "epoch": 7.100391134289439,
-      "grad_norm": 1.5090866088867188,
-      "learning_rate": 8.682311557788944e-05,
-      "loss": 5.6441,
-      "step": 13615
-    },
-    {
-      "epoch": 7.100912646675359,
-      "grad_norm": 1.4507813453674316,
-      "learning_rate": 8.682211055276382e-05,
-      "loss": 5.3904,
-      "step": 13616
-    },
-    {
-      "epoch": 7.101434159061277,
-      "grad_norm": 1.4274462461471558,
-      "learning_rate": 8.68211055276382e-05,
-      "loss": 5.4279,
-      "step": 13617
-    },
-    {
-      "epoch": 7.101955671447197,
-      "grad_norm": 1.4477678537368774,
-      "learning_rate": 8.682010050251257e-05,
-      "loss": 5.7736,
-      "step": 13618
-    },
-    {
-      "epoch": 7.102477183833116,
-      "grad_norm": 1.4626460075378418,
-      "learning_rate": 8.681909547738694e-05,
-      "loss": 5.6208,
-      "step": 13619
-    },
-    {
-      "epoch": 7.102998696219035,
-      "grad_norm": 1.4424768686294556,
-      "learning_rate": 8.681809045226132e-05,
-      "loss": 5.5077,
-      "step": 13620
-    },
-    {
-      "epoch": 7.103520208604954,
-      "grad_norm": 1.4640727043151855,
-      "learning_rate": 8.681708542713568e-05,
-      "loss": 5.5983,
-      "step": 13621
-    },
-    {
-      "epoch": 7.104041720990874,
-      "grad_norm": 1.9582855701446533,
-      "learning_rate": 8.681608040201006e-05,
-      "loss": 4.5508,
-      "step": 13622
-    },
-    {
-      "epoch": 7.104563233376792,
-      "grad_norm": 1.5002244710922241,
-      "learning_rate": 8.681507537688442e-05,
-      "loss": 5.7418,
-      "step": 13623
-    },
-    {
-      "epoch": 7.105084745762712,
-      "grad_norm": 1.4540371894836426,
-      "learning_rate": 8.68140703517588e-05,
-      "loss": 5.8811,
-      "step": 13624
-    },
-    {
-      "epoch": 7.105606258148631,
-      "grad_norm": 1.4526832103729248,
-      "learning_rate": 8.681306532663317e-05,
-      "loss": 5.5539,
-      "step": 13625
-    },
-    {
-      "epoch": 7.10612777053455,
-      "grad_norm": 1.5479185581207275,
-      "learning_rate": 8.681206030150754e-05,
-      "loss": 5.2525,
-      "step": 13626
-    },
-    {
-      "epoch": 7.106649282920469,
-      "grad_norm": 1.3460255861282349,
-      "learning_rate": 8.681105527638191e-05,
-      "loss": 5.6054,
-      "step": 13627
-    },
-    {
-      "epoch": 7.107170795306389,
-      "grad_norm": 1.4631646871566772,
-      "learning_rate": 8.681005025125629e-05,
-      "loss": 5.7908,
-      "step": 13628
-    },
-    {
-      "epoch": 7.107692307692307,
-      "grad_norm": 1.329628586769104,
-      "learning_rate": 8.680904522613066e-05,
-      "loss": 5.5694,
-      "step": 13629
-    },
-    {
-      "epoch": 7.108213820078227,
-      "grad_norm": 1.5061373710632324,
-      "learning_rate": 8.680804020100503e-05,
-      "loss": 5.6819,
-      "step": 13630
-    },
-    {
-      "epoch": 7.108735332464146,
-      "grad_norm": 1.5007331371307373,
-      "learning_rate": 8.68070351758794e-05,
-      "loss": 5.2998,
-      "step": 13631
-    },
-    {
-      "epoch": 7.109256844850065,
-      "grad_norm": 1.515008568763733,
-      "learning_rate": 8.680603015075377e-05,
-      "loss": 5.6694,
-      "step": 13632
-    },
-    {
-      "epoch": 7.109778357235984,
-      "grad_norm": 1.3977910280227661,
-      "learning_rate": 8.680502512562815e-05,
-      "loss": 5.5662,
-      "step": 13633
-    },
-    {
-      "epoch": 7.110299869621904,
-      "grad_norm": 1.730996012687683,
-      "learning_rate": 8.680402010050251e-05,
-      "loss": 4.9271,
-      "step": 13634
-    },
-    {
-      "epoch": 7.110821382007822,
-      "grad_norm": 1.5520387887954712,
-      "learning_rate": 8.680301507537689e-05,
-      "loss": 5.24,
-      "step": 13635
-    },
-    {
-      "epoch": 7.111342894393742,
-      "grad_norm": 1.572482943534851,
-      "learning_rate": 8.680201005025125e-05,
-      "loss": 5.6464,
-      "step": 13636
-    },
-    {
-      "epoch": 7.111864406779661,
-      "grad_norm": 1.4917972087860107,
-      "learning_rate": 8.680100502512563e-05,
-      "loss": 5.1299,
-      "step": 13637
-    },
-    {
-      "epoch": 7.11238591916558,
-      "grad_norm": 1.5377639532089233,
-      "learning_rate": 8.680000000000001e-05,
-      "loss": 5.7582,
-      "step": 13638
-    },
-    {
-      "epoch": 7.112907431551499,
-      "grad_norm": 1.372374415397644,
-      "learning_rate": 8.679899497487439e-05,
-      "loss": 5.8432,
-      "step": 13639
-    },
-    {
-      "epoch": 7.113428943937419,
-      "grad_norm": 1.4222642183303833,
-      "learning_rate": 8.679798994974875e-05,
-      "loss": 5.5761,
-      "step": 13640
-    },
-    {
-      "epoch": 7.113950456323337,
-      "grad_norm": 1.6515171527862549,
-      "learning_rate": 8.679698492462312e-05,
-      "loss": 5.0948,
-      "step": 13641
-    },
-    {
-      "epoch": 7.114471968709257,
-      "grad_norm": 1.5630109310150146,
-      "learning_rate": 8.67959798994975e-05,
-      "loss": 5.3326,
-      "step": 13642
-    },
-    {
-      "epoch": 7.114993481095176,
-      "grad_norm": 1.3394486904144287,
-      "learning_rate": 8.679497487437186e-05,
-      "loss": 5.3268,
-      "step": 13643
-    },
-    {
-      "epoch": 7.115514993481095,
-      "grad_norm": 1.7022830247879028,
-      "learning_rate": 8.679396984924624e-05,
-      "loss": 4.6412,
-      "step": 13644
-    },
-    {
-      "epoch": 7.116036505867014,
-      "grad_norm": 1.7195699214935303,
-      "learning_rate": 8.67929648241206e-05,
-      "loss": 5.3261,
-      "step": 13645
-    },
-    {
-      "epoch": 7.116558018252934,
-      "grad_norm": 1.6442387104034424,
-      "learning_rate": 8.679195979899498e-05,
-      "loss": 5.5225,
-      "step": 13646
-    },
-    {
-      "epoch": 7.117079530638852,
-      "grad_norm": 1.6683534383773804,
-      "learning_rate": 8.679095477386934e-05,
-      "loss": 5.1136,
-      "step": 13647
-    },
-    {
-      "epoch": 7.117601043024772,
-      "grad_norm": 1.6192593574523926,
-      "learning_rate": 8.678994974874372e-05,
-      "loss": 5.1716,
-      "step": 13648
-    },
-    {
-      "epoch": 7.118122555410691,
-      "grad_norm": 1.4473369121551514,
-      "learning_rate": 8.67889447236181e-05,
-      "loss": 5.4246,
-      "step": 13649
-    },
-    {
-      "epoch": 7.11864406779661,
-      "grad_norm": 1.6402381658554077,
-      "learning_rate": 8.678793969849248e-05,
-      "loss": 5.5722,
-      "step": 13650
-    },
-    {
-      "epoch": 7.119165580182529,
-      "grad_norm": 1.5254406929016113,
-      "learning_rate": 8.678693467336684e-05,
-      "loss": 5.4803,
-      "step": 13651
-    },
-    {
-      "epoch": 7.119687092568449,
-      "grad_norm": 1.4192821979522705,
-      "learning_rate": 8.678592964824122e-05,
-      "loss": 5.5274,
-      "step": 13652
-    },
-    {
-      "epoch": 7.120208604954367,
-      "grad_norm": 1.4525632858276367,
-      "learning_rate": 8.678492462311558e-05,
-      "loss": 5.7034,
-      "step": 13653
-    },
-    {
-      "epoch": 7.120730117340287,
-      "grad_norm": 1.5107717514038086,
-      "learning_rate": 8.678391959798995e-05,
-      "loss": 4.9533,
-      "step": 13654
-    },
-    {
-      "epoch": 7.121251629726206,
-      "grad_norm": 1.3504157066345215,
-      "learning_rate": 8.678291457286432e-05,
-      "loss": 5.9177,
-      "step": 13655
-    },
-    {
-      "epoch": 7.121773142112125,
-      "grad_norm": 1.297222375869751,
-      "learning_rate": 8.678190954773869e-05,
-      "loss": 5.702,
-      "step": 13656
-    },
-    {
-      "epoch": 7.122294654498044,
-      "grad_norm": 1.4336378574371338,
-      "learning_rate": 8.678090452261307e-05,
-      "loss": 5.8329,
-      "step": 13657
-    },
-    {
-      "epoch": 7.122816166883964,
-      "grad_norm": 1.4444360733032227,
-      "learning_rate": 8.677989949748744e-05,
-      "loss": 5.476,
-      "step": 13658
-    },
-    {
-      "epoch": 7.123337679269882,
-      "grad_norm": 1.497109055519104,
-      "learning_rate": 8.677889447236182e-05,
-      "loss": 5.5574,
-      "step": 13659
-    },
-    {
-      "epoch": 7.123859191655802,
-      "grad_norm": 1.5590578317642212,
-      "learning_rate": 8.677788944723619e-05,
-      "loss": 4.7393,
-      "step": 13660
-    },
-    {
-      "epoch": 7.124380704041721,
-      "grad_norm": 1.4764740467071533,
-      "learning_rate": 8.677688442211056e-05,
-      "loss": 5.3534,
-      "step": 13661
-    },
-    {
-      "epoch": 7.12490221642764,
-      "grad_norm": 1.4598698616027832,
-      "learning_rate": 8.677587939698493e-05,
-      "loss": 5.6324,
-      "step": 13662
-    },
-    {
-      "epoch": 7.125423728813559,
-      "grad_norm": 1.5051214694976807,
-      "learning_rate": 8.67748743718593e-05,
-      "loss": 5.8152,
-      "step": 13663
-    },
-    {
-      "epoch": 7.125945241199479,
-      "grad_norm": 1.4194072484970093,
-      "learning_rate": 8.677386934673367e-05,
-      "loss": 5.062,
-      "step": 13664
-    },
-    {
-      "epoch": 7.126466753585397,
-      "grad_norm": 1.5226991176605225,
-      "learning_rate": 8.677286432160805e-05,
-      "loss": 5.6502,
-      "step": 13665
-    },
-    {
-      "epoch": 7.126988265971317,
-      "grad_norm": 1.4377437829971313,
-      "learning_rate": 8.677185929648241e-05,
-      "loss": 5.6032,
-      "step": 13666
-    },
-    {
-      "epoch": 7.127509778357236,
-      "grad_norm": 1.485182285308838,
-      "learning_rate": 8.677085427135678e-05,
-      "loss": 5.4856,
-      "step": 13667
-    },
-    {
-      "epoch": 7.128031290743155,
-      "grad_norm": 1.5292298793792725,
-      "learning_rate": 8.676984924623115e-05,
-      "loss": 5.2058,
-      "step": 13668
-    },
-    {
-      "epoch": 7.128552803129074,
-      "grad_norm": 1.403167486190796,
-      "learning_rate": 8.676884422110553e-05,
-      "loss": 5.3197,
-      "step": 13669
-    },
-    {
-      "epoch": 7.129074315514994,
-      "grad_norm": 1.5798016786575317,
-      "learning_rate": 8.676783919597991e-05,
-      "loss": 5.3361,
-      "step": 13670
-    },
-    {
-      "epoch": 7.129595827900912,
-      "grad_norm": 1.5143396854400635,
-      "learning_rate": 8.676683417085427e-05,
-      "loss": 5.3996,
-      "step": 13671
-    },
-    {
-      "epoch": 7.130117340286832,
-      "grad_norm": 1.3972196578979492,
-      "learning_rate": 8.676582914572865e-05,
-      "loss": 5.7792,
-      "step": 13672
-    },
-    {
-      "epoch": 7.130638852672751,
-      "grad_norm": 1.501516342163086,
-      "learning_rate": 8.676482412060302e-05,
-      "loss": 5.2751,
-      "step": 13673
-    },
-    {
-      "epoch": 7.13116036505867,
-      "grad_norm": 1.4885339736938477,
-      "learning_rate": 8.67638190954774e-05,
-      "loss": 5.8677,
-      "step": 13674
-    },
-    {
-      "epoch": 7.131681877444589,
-      "grad_norm": 1.5981541872024536,
-      "learning_rate": 8.676281407035176e-05,
-      "loss": 5.1525,
-      "step": 13675
-    },
-    {
-      "epoch": 7.132203389830509,
-      "grad_norm": 1.4468238353729248,
-      "learning_rate": 8.676180904522614e-05,
-      "loss": 5.119,
-      "step": 13676
-    },
-    {
-      "epoch": 7.132724902216427,
-      "grad_norm": 1.4865953922271729,
-      "learning_rate": 8.67608040201005e-05,
-      "loss": 5.6191,
-      "step": 13677
-    },
-    {
-      "epoch": 7.133246414602347,
-      "grad_norm": 1.5246223211288452,
-      "learning_rate": 8.675979899497488e-05,
-      "loss": 5.2911,
-      "step": 13678
-    },
-    {
-      "epoch": 7.133767926988266,
-      "grad_norm": 1.4922221899032593,
-      "learning_rate": 8.675879396984926e-05,
-      "loss": 5.5074,
-      "step": 13679
-    },
-    {
-      "epoch": 7.134289439374185,
-      "grad_norm": 1.3871506452560425,
-      "learning_rate": 8.675778894472362e-05,
-      "loss": 5.4585,
-      "step": 13680
-    },
-    {
-      "epoch": 7.134810951760104,
-      "grad_norm": 1.4123525619506836,
-      "learning_rate": 8.6756783919598e-05,
-      "loss": 5.6936,
-      "step": 13681
-    },
-    {
-      "epoch": 7.135332464146024,
-      "grad_norm": 1.4054672718048096,
-      "learning_rate": 8.675577889447236e-05,
-      "loss": 5.7809,
-      "step": 13682
-    },
-    {
-      "epoch": 7.135853976531942,
-      "grad_norm": 1.6185574531555176,
-      "learning_rate": 8.675477386934674e-05,
-      "loss": 5.424,
-      "step": 13683
-    },
-    {
-      "epoch": 7.136375488917862,
-      "grad_norm": 1.5688014030456543,
-      "learning_rate": 8.67537688442211e-05,
-      "loss": 5.4405,
-      "step": 13684
-    },
-    {
-      "epoch": 7.136897001303781,
-      "grad_norm": 1.365963101387024,
-      "learning_rate": 8.675276381909548e-05,
-      "loss": 5.9682,
-      "step": 13685
-    },
-    {
-      "epoch": 7.1374185136897,
-      "grad_norm": 1.4275968074798584,
-      "learning_rate": 8.675175879396985e-05,
-      "loss": 5.4065,
-      "step": 13686
-    },
-    {
-      "epoch": 7.137940026075619,
-      "grad_norm": 1.3992513418197632,
-      "learning_rate": 8.675075376884422e-05,
-      "loss": 5.6803,
-      "step": 13687
-    },
-    {
-      "epoch": 7.138461538461539,
-      "grad_norm": 1.3723068237304688,
-      "learning_rate": 8.674974874371859e-05,
-      "loss": 5.7803,
-      "step": 13688
-    },
-    {
-      "epoch": 7.138983050847457,
-      "grad_norm": 1.4092477560043335,
-      "learning_rate": 8.674874371859297e-05,
-      "loss": 5.5841,
-      "step": 13689
-    },
-    {
-      "epoch": 7.139504563233377,
-      "grad_norm": 1.4170984029769897,
-      "learning_rate": 8.674773869346734e-05,
-      "loss": 5.5125,
-      "step": 13690
-    },
-    {
-      "epoch": 7.140026075619296,
-      "grad_norm": 1.399552345275879,
-      "learning_rate": 8.674673366834172e-05,
-      "loss": 5.8054,
-      "step": 13691
-    },
-    {
-      "epoch": 7.140547588005215,
-      "grad_norm": 1.4512799978256226,
-      "learning_rate": 8.674572864321609e-05,
-      "loss": 5.1677,
-      "step": 13692
-    },
-    {
-      "epoch": 7.141069100391134,
-      "grad_norm": 1.4772684574127197,
-      "learning_rate": 8.674472361809045e-05,
-      "loss": 5.7397,
-      "step": 13693
-    },
-    {
-      "epoch": 7.141590612777054,
-      "grad_norm": 1.4428117275238037,
-      "learning_rate": 8.674371859296483e-05,
-      "loss": 5.5534,
-      "step": 13694
-    },
-    {
-      "epoch": 7.1421121251629724,
-      "grad_norm": 1.3874369859695435,
-      "learning_rate": 8.674271356783919e-05,
-      "loss": 5.7227,
-      "step": 13695
-    },
-    {
-      "epoch": 7.142633637548892,
-      "grad_norm": 1.4359149932861328,
-      "learning_rate": 8.674170854271357e-05,
-      "loss": 5.8466,
-      "step": 13696
-    },
-    {
-      "epoch": 7.143155149934811,
-      "grad_norm": 1.3171013593673706,
-      "learning_rate": 8.674070351758794e-05,
-      "loss": 5.7037,
-      "step": 13697
-    },
-    {
-      "epoch": 7.14367666232073,
-      "grad_norm": 1.3525609970092773,
-      "learning_rate": 8.673969849246231e-05,
-      "loss": 5.5682,
-      "step": 13698
-    },
-    {
-      "epoch": 7.144198174706649,
-      "grad_norm": 1.4392998218536377,
-      "learning_rate": 8.673869346733669e-05,
-      "loss": 5.6141,
-      "step": 13699
-    },
-    {
-      "epoch": 7.144719687092568,
-      "grad_norm": 1.593989372253418,
-      "learning_rate": 8.673768844221107e-05,
-      "loss": 5.436,
-      "step": 13700
-    },
-    {
-      "epoch": 7.1452411994784875,
-      "grad_norm": 1.3460668325424194,
-      "learning_rate": 8.673668341708543e-05,
-      "loss": 5.6313,
-      "step": 13701
-    },
-    {
-      "epoch": 7.145762711864407,
-      "grad_norm": 1.3352442979812622,
-      "learning_rate": 8.673567839195981e-05,
-      "loss": 5.1501,
-      "step": 13702
-    },
-    {
-      "epoch": 7.146284224250326,
-      "grad_norm": 1.4672331809997559,
-      "learning_rate": 8.673467336683417e-05,
-      "loss": 5.7272,
-      "step": 13703
-    },
-    {
-      "epoch": 7.146805736636245,
-      "grad_norm": 1.6010291576385498,
-      "learning_rate": 8.673366834170855e-05,
-      "loss": 5.091,
-      "step": 13704
-    },
-    {
-      "epoch": 7.147327249022164,
-      "grad_norm": 1.3806794881820679,
-      "learning_rate": 8.673266331658292e-05,
-      "loss": 5.709,
-      "step": 13705
-    },
-    {
-      "epoch": 7.147848761408083,
-      "grad_norm": 1.4654204845428467,
-      "learning_rate": 8.673165829145728e-05,
-      "loss": 5.0132,
-      "step": 13706
-    },
-    {
-      "epoch": 7.1483702737940025,
-      "grad_norm": 1.416214942932129,
-      "learning_rate": 8.673065326633166e-05,
-      "loss": 5.799,
-      "step": 13707
-    },
-    {
-      "epoch": 7.148891786179922,
-      "grad_norm": 1.348501205444336,
-      "learning_rate": 8.672964824120602e-05,
-      "loss": 5.6978,
-      "step": 13708
-    },
-    {
-      "epoch": 7.1494132985658405,
-      "grad_norm": 1.3492417335510254,
-      "learning_rate": 8.67286432160804e-05,
-      "loss": 5.6261,
-      "step": 13709
-    },
-    {
-      "epoch": 7.14993481095176,
-      "grad_norm": 1.4385771751403809,
-      "learning_rate": 8.672763819095478e-05,
-      "loss": 4.9238,
-      "step": 13710
-    },
-    {
-      "epoch": 7.150456323337679,
-      "grad_norm": 1.4635928869247437,
-      "learning_rate": 8.672663316582916e-05,
-      "loss": 5.9071,
-      "step": 13711
-    },
-    {
-      "epoch": 7.150977835723598,
-      "grad_norm": 1.5019280910491943,
-      "learning_rate": 8.672562814070352e-05,
-      "loss": 5.6247,
-      "step": 13712
-    },
-    {
-      "epoch": 7.1514993481095175,
-      "grad_norm": 1.4594782590866089,
-      "learning_rate": 8.67246231155779e-05,
-      "loss": 5.3033,
-      "step": 13713
-    },
-    {
-      "epoch": 7.152020860495437,
-      "grad_norm": 1.454990267753601,
-      "learning_rate": 8.672361809045226e-05,
-      "loss": 5.7657,
-      "step": 13714
-    },
-    {
-      "epoch": 7.1525423728813555,
-      "grad_norm": 1.6095958948135376,
-      "learning_rate": 8.672261306532664e-05,
-      "loss": 5.5021,
-      "step": 13715
-    },
-    {
-      "epoch": 7.153063885267275,
-      "grad_norm": 1.3762080669403076,
-      "learning_rate": 8.6721608040201e-05,
-      "loss": 5.8533,
-      "step": 13716
-    },
-    {
-      "epoch": 7.1535853976531945,
-      "grad_norm": 2.1343610286712646,
-      "learning_rate": 8.672060301507538e-05,
-      "loss": 5.0848,
-      "step": 13717
-    },
-    {
-      "epoch": 7.154106910039113,
-      "grad_norm": 1.5194721221923828,
-      "learning_rate": 8.671959798994975e-05,
-      "loss": 5.6722,
-      "step": 13718
-    },
-    {
-      "epoch": 7.1546284224250325,
-      "grad_norm": 1.5119245052337646,
-      "learning_rate": 8.671859296482413e-05,
-      "loss": 5.0735,
-      "step": 13719
-    },
-    {
-      "epoch": 7.155149934810952,
-      "grad_norm": 1.4835466146469116,
-      "learning_rate": 8.67175879396985e-05,
-      "loss": 5.9239,
-      "step": 13720
-    },
-    {
-      "epoch": 7.1556714471968705,
-      "grad_norm": 1.8728574514389038,
-      "learning_rate": 8.671658291457287e-05,
-      "loss": 5.1611,
-      "step": 13721
-    },
-    {
-      "epoch": 7.15619295958279,
-      "grad_norm": 1.4794845581054688,
-      "learning_rate": 8.671557788944725e-05,
-      "loss": 5.9314,
-      "step": 13722
-    },
-    {
-      "epoch": 7.1567144719687095,
-      "grad_norm": 1.4493411779403687,
-      "learning_rate": 8.671457286432161e-05,
-      "loss": 5.8483,
-      "step": 13723
-    },
-    {
-      "epoch": 7.157235984354628,
-      "grad_norm": 1.4273525476455688,
-      "learning_rate": 8.671356783919599e-05,
-      "loss": 5.0698,
-      "step": 13724
-    },
-    {
-      "epoch": 7.1577574967405475,
-      "grad_norm": 1.4652984142303467,
-      "learning_rate": 8.671256281407035e-05,
-      "loss": 5.4427,
-      "step": 13725
-    },
-    {
-      "epoch": 7.158279009126467,
-      "grad_norm": 1.5137547254562378,
-      "learning_rate": 8.671155778894473e-05,
-      "loss": 5.6676,
-      "step": 13726
-    },
-    {
-      "epoch": 7.1588005215123856,
-      "grad_norm": 1.380881428718567,
-      "learning_rate": 8.67105527638191e-05,
-      "loss": 5.6083,
-      "step": 13727
-    },
-    {
-      "epoch": 7.159322033898305,
-      "grad_norm": 1.395025372505188,
-      "learning_rate": 8.670954773869347e-05,
-      "loss": 5.2951,
-      "step": 13728
-    },
-    {
-      "epoch": 7.1598435462842245,
-      "grad_norm": 1.5917667150497437,
-      "learning_rate": 8.670854271356784e-05,
-      "loss": 5.392,
-      "step": 13729
-    },
-    {
-      "epoch": 7.160365058670143,
-      "grad_norm": 1.5410608053207397,
-      "learning_rate": 8.670753768844221e-05,
-      "loss": 5.3055,
-      "step": 13730
-    },
-    {
-      "epoch": 7.1608865710560625,
-      "grad_norm": 1.3579561710357666,
-      "learning_rate": 8.670653266331659e-05,
-      "loss": 6.0048,
-      "step": 13731
-    },
-    {
-      "epoch": 7.161408083441982,
-      "grad_norm": 1.3876606225967407,
-      "learning_rate": 8.670552763819097e-05,
-      "loss": 5.8546,
-      "step": 13732
-    },
-    {
-      "epoch": 7.161929595827901,
-      "grad_norm": 1.3969610929489136,
-      "learning_rate": 8.670452261306533e-05,
-      "loss": 5.6641,
-      "step": 13733
-    },
-    {
-      "epoch": 7.16245110821382,
-      "grad_norm": 1.5410056114196777,
-      "learning_rate": 8.67035175879397e-05,
-      "loss": 5.3438,
-      "step": 13734
-    },
-    {
-      "epoch": 7.1629726205997395,
-      "grad_norm": 1.4928194284439087,
-      "learning_rate": 8.670251256281408e-05,
-      "loss": 5.5356,
-      "step": 13735
-    },
-    {
-      "epoch": 7.163494132985658,
-      "grad_norm": 1.6006301641464233,
-      "learning_rate": 8.670150753768844e-05,
-      "loss": 5.6162,
-      "step": 13736
-    },
-    {
-      "epoch": 7.1640156453715775,
-      "grad_norm": 1.477206826210022,
-      "learning_rate": 8.670050251256282e-05,
-      "loss": 5.0868,
-      "step": 13737
-    },
-    {
-      "epoch": 7.164537157757497,
-      "grad_norm": 1.3133422136306763,
-      "learning_rate": 8.669949748743718e-05,
-      "loss": 5.854,
-      "step": 13738
-    },
-    {
-      "epoch": 7.165058670143416,
-      "grad_norm": 1.5052441358566284,
-      "learning_rate": 8.669849246231156e-05,
-      "loss": 5.6181,
-      "step": 13739
-    },
-    {
-      "epoch": 7.165580182529335,
-      "grad_norm": 1.4646841287612915,
-      "learning_rate": 8.669748743718594e-05,
-      "loss": 5.8571,
-      "step": 13740
-    },
-    {
-      "epoch": 7.1661016949152545,
-      "grad_norm": 1.4477499723434448,
-      "learning_rate": 8.669648241206032e-05,
-      "loss": 5.6897,
-      "step": 13741
-    },
-    {
-      "epoch": 7.166623207301173,
-      "grad_norm": 1.5311987400054932,
-      "learning_rate": 8.669547738693468e-05,
-      "loss": 5.8968,
-      "step": 13742
-    },
-    {
-      "epoch": 7.1671447196870925,
-      "grad_norm": 1.4337255954742432,
-      "learning_rate": 8.669447236180906e-05,
-      "loss": 5.6948,
-      "step": 13743
-    },
-    {
-      "epoch": 7.167666232073012,
-      "grad_norm": 1.4166909456253052,
-      "learning_rate": 8.669346733668342e-05,
-      "loss": 5.6831,
-      "step": 13744
-    },
-    {
-      "epoch": 7.168187744458931,
-      "grad_norm": 1.4642064571380615,
-      "learning_rate": 8.66924623115578e-05,
-      "loss": 5.7748,
-      "step": 13745
-    },
-    {
-      "epoch": 7.16870925684485,
-      "grad_norm": 1.3687443733215332,
-      "learning_rate": 8.669145728643216e-05,
-      "loss": 5.8872,
-      "step": 13746
-    },
-    {
-      "epoch": 7.1692307692307695,
-      "grad_norm": 1.5333423614501953,
-      "learning_rate": 8.669045226130653e-05,
-      "loss": 5.0032,
-      "step": 13747
-    },
-    {
-      "epoch": 7.169752281616688,
-      "grad_norm": 1.3645167350769043,
-      "learning_rate": 8.66894472361809e-05,
-      "loss": 5.7995,
-      "step": 13748
-    },
-    {
-      "epoch": 7.170273794002608,
-      "grad_norm": 1.5342808961868286,
-      "learning_rate": 8.668844221105527e-05,
-      "loss": 5.8671,
-      "step": 13749
-    },
-    {
-      "epoch": 7.170795306388527,
-      "grad_norm": 1.4890203475952148,
-      "learning_rate": 8.668743718592965e-05,
-      "loss": 5.395,
-      "step": 13750
-    },
-    {
-      "epoch": 7.171316818774446,
-      "grad_norm": 1.5375837087631226,
-      "learning_rate": 8.668643216080403e-05,
-      "loss": 5.7229,
-      "step": 13751
-    },
-    {
-      "epoch": 7.171838331160365,
-      "grad_norm": 1.4149529933929443,
-      "learning_rate": 8.66854271356784e-05,
-      "loss": 5.729,
-      "step": 13752
-    },
-    {
-      "epoch": 7.1723598435462845,
-      "grad_norm": 1.4262226819992065,
-      "learning_rate": 8.668442211055277e-05,
-      "loss": 5.4925,
-      "step": 13753
-    },
-    {
-      "epoch": 7.172881355932203,
-      "grad_norm": 1.3137096166610718,
-      "learning_rate": 8.668341708542715e-05,
-      "loss": 5.9522,
-      "step": 13754
-    },
-    {
-      "epoch": 7.173402868318123,
-      "grad_norm": 1.6734628677368164,
-      "learning_rate": 8.668241206030151e-05,
-      "loss": 5.5751,
-      "step": 13755
-    },
-    {
-      "epoch": 7.173924380704042,
-      "grad_norm": 1.39096200466156,
-      "learning_rate": 8.668140703517589e-05,
-      "loss": 6.1222,
-      "step": 13756
-    },
-    {
-      "epoch": 7.174445893089961,
-      "grad_norm": 1.5020592212677002,
-      "learning_rate": 8.668040201005025e-05,
-      "loss": 5.4956,
-      "step": 13757
-    },
-    {
-      "epoch": 7.17496740547588,
-      "grad_norm": 1.5459926128387451,
-      "learning_rate": 8.667939698492463e-05,
-      "loss": 5.4374,
-      "step": 13758
-    },
-    {
-      "epoch": 7.1754889178617995,
-      "grad_norm": 1.4007116556167603,
-      "learning_rate": 8.6678391959799e-05,
-      "loss": 5.7001,
-      "step": 13759
-    },
-    {
-      "epoch": 7.176010430247718,
-      "grad_norm": 1.4595884084701538,
-      "learning_rate": 8.667738693467337e-05,
-      "loss": 5.6649,
-      "step": 13760
-    },
-    {
-      "epoch": 7.176531942633638,
-      "grad_norm": 1.3825914859771729,
-      "learning_rate": 8.667638190954775e-05,
-      "loss": 5.7271,
-      "step": 13761
-    },
-    {
-      "epoch": 7.177053455019557,
-      "grad_norm": 1.50836980342865,
-      "learning_rate": 8.667537688442211e-05,
-      "loss": 5.2057,
-      "step": 13762
-    },
-    {
-      "epoch": 7.177574967405476,
-      "grad_norm": 1.475838303565979,
-      "learning_rate": 8.667437185929649e-05,
-      "loss": 5.4637,
-      "step": 13763
-    },
-    {
-      "epoch": 7.178096479791395,
-      "grad_norm": 1.5173999071121216,
-      "learning_rate": 8.667336683417086e-05,
-      "loss": 5.1034,
-      "step": 13764
-    },
-    {
-      "epoch": 7.1786179921773146,
-      "grad_norm": 1.4801719188690186,
-      "learning_rate": 8.667236180904523e-05,
-      "loss": 5.474,
-      "step": 13765
-    },
-    {
-      "epoch": 7.179139504563233,
-      "grad_norm": 1.4594944715499878,
-      "learning_rate": 8.66713567839196e-05,
-      "loss": 5.5546,
-      "step": 13766
-    },
-    {
-      "epoch": 7.179661016949153,
-      "grad_norm": 1.3547470569610596,
-      "learning_rate": 8.667035175879398e-05,
-      "loss": 5.476,
-      "step": 13767
-    },
-    {
-      "epoch": 7.180182529335072,
-      "grad_norm": 1.4691829681396484,
-      "learning_rate": 8.666934673366834e-05,
-      "loss": 5.2176,
-      "step": 13768
-    },
-    {
-      "epoch": 7.180704041720991,
-      "grad_norm": 1.3840066194534302,
-      "learning_rate": 8.666834170854272e-05,
-      "loss": 5.6359,
-      "step": 13769
-    },
-    {
-      "epoch": 7.18122555410691,
-      "grad_norm": 1.4959218502044678,
-      "learning_rate": 8.666733668341708e-05,
-      "loss": 5.5634,
-      "step": 13770
-    },
-    {
-      "epoch": 7.18174706649283,
-      "grad_norm": 1.4436759948730469,
-      "learning_rate": 8.666633165829146e-05,
-      "loss": 5.666,
-      "step": 13771
-    },
-    {
-      "epoch": 7.182268578878748,
-      "grad_norm": 1.4756942987442017,
-      "learning_rate": 8.666532663316584e-05,
-      "loss": 5.5328,
-      "step": 13772
-    },
-    {
-      "epoch": 7.182790091264668,
-      "grad_norm": 1.3193011283874512,
-      "learning_rate": 8.66643216080402e-05,
-      "loss": 5.6252,
-      "step": 13773
-    },
-    {
-      "epoch": 7.183311603650587,
-      "grad_norm": 1.611114263534546,
-      "learning_rate": 8.666331658291458e-05,
-      "loss": 5.2047,
-      "step": 13774
-    },
-    {
-      "epoch": 7.183833116036506,
-      "grad_norm": 1.4112190008163452,
-      "learning_rate": 8.666231155778894e-05,
-      "loss": 5.3845,
-      "step": 13775
-    },
-    {
-      "epoch": 7.184354628422425,
-      "grad_norm": 1.5705552101135254,
-      "learning_rate": 8.666130653266332e-05,
-      "loss": 5.151,
-      "step": 13776
-    },
-    {
-      "epoch": 7.184876140808345,
-      "grad_norm": 1.465720295906067,
-      "learning_rate": 8.666030150753769e-05,
-      "loss": 5.5399,
-      "step": 13777
-    },
-    {
-      "epoch": 7.185397653194263,
-      "grad_norm": 1.4347004890441895,
-      "learning_rate": 8.665929648241206e-05,
-      "loss": 5.187,
-      "step": 13778
-    },
-    {
-      "epoch": 7.185919165580183,
-      "grad_norm": 1.3448377847671509,
-      "learning_rate": 8.665829145728643e-05,
-      "loss": 5.8386,
-      "step": 13779
-    },
-    {
-      "epoch": 7.186440677966102,
-      "grad_norm": 1.4741203784942627,
-      "learning_rate": 8.665728643216081e-05,
-      "loss": 5.4708,
-      "step": 13780
-    },
-    {
-      "epoch": 7.186962190352021,
-      "grad_norm": 1.4139752388000488,
-      "learning_rate": 8.665628140703518e-05,
-      "loss": 5.7875,
-      "step": 13781
-    },
-    {
-      "epoch": 7.18748370273794,
-      "grad_norm": 1.4072082042694092,
-      "learning_rate": 8.665527638190956e-05,
-      "loss": 5.7757,
-      "step": 13782
-    },
-    {
-      "epoch": 7.18800521512386,
-      "grad_norm": 1.4167219400405884,
-      "learning_rate": 8.665427135678393e-05,
-      "loss": 5.3975,
-      "step": 13783
-    },
-    {
-      "epoch": 7.188526727509778,
-      "grad_norm": 1.4888793230056763,
-      "learning_rate": 8.66532663316583e-05,
-      "loss": 5.0454,
-      "step": 13784
-    },
-    {
-      "epoch": 7.189048239895698,
-      "grad_norm": 1.6166670322418213,
-      "learning_rate": 8.665226130653267e-05,
-      "loss": 5.3865,
-      "step": 13785
-    },
-    {
-      "epoch": 7.189569752281617,
-      "grad_norm": 1.3914917707443237,
-      "learning_rate": 8.665125628140703e-05,
-      "loss": 5.6687,
-      "step": 13786
-    },
-    {
-      "epoch": 7.190091264667536,
-      "grad_norm": 1.485008716583252,
-      "learning_rate": 8.665025125628141e-05,
-      "loss": 5.0224,
-      "step": 13787
-    },
-    {
-      "epoch": 7.190612777053455,
-      "grad_norm": 1.5275533199310303,
-      "learning_rate": 8.664924623115578e-05,
-      "loss": 5.325,
-      "step": 13788
-    },
-    {
-      "epoch": 7.191134289439375,
-      "grad_norm": 1.5702720880508423,
-      "learning_rate": 8.664824120603015e-05,
-      "loss": 5.3958,
-      "step": 13789
-    },
-    {
-      "epoch": 7.191655801825293,
-      "grad_norm": 1.8940235376358032,
-      "learning_rate": 8.664723618090452e-05,
-      "loss": 5.1567,
-      "step": 13790
-    },
-    {
-      "epoch": 7.192177314211213,
-      "grad_norm": 1.4270268678665161,
-      "learning_rate": 8.66462311557789e-05,
-      "loss": 5.489,
-      "step": 13791
-    },
-    {
-      "epoch": 7.192698826597132,
-      "grad_norm": 1.486464023590088,
-      "learning_rate": 8.664522613065327e-05,
-      "loss": 5.6908,
-      "step": 13792
-    },
-    {
-      "epoch": 7.193220338983051,
-      "grad_norm": 1.3854243755340576,
-      "learning_rate": 8.664422110552765e-05,
-      "loss": 5.4293,
-      "step": 13793
-    },
-    {
-      "epoch": 7.19374185136897,
-      "grad_norm": 1.4980734586715698,
-      "learning_rate": 8.664321608040202e-05,
-      "loss": 5.3788,
-      "step": 13794
-    },
-    {
-      "epoch": 7.194263363754889,
-      "grad_norm": 1.5902016162872314,
-      "learning_rate": 8.664221105527639e-05,
-      "loss": 5.3741,
-      "step": 13795
-    },
-    {
-      "epoch": 7.194784876140808,
-      "grad_norm": 1.5114284753799438,
-      "learning_rate": 8.664120603015076e-05,
-      "loss": 5.7927,
-      "step": 13796
-    },
-    {
-      "epoch": 7.195306388526728,
-      "grad_norm": 1.5146446228027344,
-      "learning_rate": 8.664020100502514e-05,
-      "loss": 5.5852,
-      "step": 13797
-    },
-    {
-      "epoch": 7.195827900912647,
-      "grad_norm": 1.4084800481796265,
-      "learning_rate": 8.66391959798995e-05,
-      "loss": 5.8451,
-      "step": 13798
-    },
-    {
-      "epoch": 7.196349413298566,
-      "grad_norm": 1.4775413274765015,
-      "learning_rate": 8.663819095477388e-05,
-      "loss": 5.6098,
-      "step": 13799
-    },
-    {
-      "epoch": 7.196870925684485,
-      "grad_norm": 1.4530738592147827,
-      "learning_rate": 8.663718592964824e-05,
-      "loss": 5.9596,
-      "step": 13800
-    },
-    {
-      "epoch": 7.197392438070404,
-      "grad_norm": 1.28622305393219,
-      "learning_rate": 8.66361809045226e-05,
-      "loss": 5.9268,
-      "step": 13801
-    },
-    {
-      "epoch": 7.197913950456323,
-      "grad_norm": 1.4587386846542358,
-      "learning_rate": 8.663517587939698e-05,
-      "loss": 5.8263,
-      "step": 13802
-    },
-    {
-      "epoch": 7.198435462842243,
-      "grad_norm": 1.3289525508880615,
-      "learning_rate": 8.663417085427136e-05,
-      "loss": 6.0443,
-      "step": 13803
-    },
-    {
-      "epoch": 7.198956975228161,
-      "grad_norm": 1.5427478551864624,
-      "learning_rate": 8.663316582914574e-05,
-      "loss": 5.3255,
-      "step": 13804
-    },
-    {
-      "epoch": 7.199478487614081,
-      "grad_norm": 1.4756428003311157,
-      "learning_rate": 8.66321608040201e-05,
-      "loss": 5.877,
-      "step": 13805
-    },
-    {
-      "epoch": 7.2,
-      "grad_norm": 1.5827785730361938,
-      "learning_rate": 8.663115577889448e-05,
-      "loss": 5.9185,
-      "step": 13806
-    },
-    {
-      "epoch": 7.200521512385919,
-      "grad_norm": 1.5714043378829956,
-      "learning_rate": 8.663015075376885e-05,
-      "loss": 5.3746,
-      "step": 13807
-    },
-    {
-      "epoch": 7.201043024771838,
-      "grad_norm": 1.602576732635498,
-      "learning_rate": 8.662914572864322e-05,
-      "loss": 5.4826,
-      "step": 13808
-    },
-    {
-      "epoch": 7.201564537157758,
-      "grad_norm": 1.5858824253082275,
-      "learning_rate": 8.662814070351759e-05,
-      "loss": 5.4808,
-      "step": 13809
-    },
-    {
-      "epoch": 7.202086049543676,
-      "grad_norm": 1.4796278476715088,
-      "learning_rate": 8.662713567839197e-05,
-      "loss": 5.7076,
-      "step": 13810
-    },
-    {
-      "epoch": 7.202607561929596,
-      "grad_norm": 1.4641693830490112,
-      "learning_rate": 8.662613065326633e-05,
-      "loss": 5.4828,
-      "step": 13811
-    },
-    {
-      "epoch": 7.203129074315515,
-      "grad_norm": 1.3879826068878174,
-      "learning_rate": 8.662512562814071e-05,
-      "loss": 5.4879,
-      "step": 13812
-    },
-    {
-      "epoch": 7.203650586701434,
-      "grad_norm": 1.446555733680725,
-      "learning_rate": 8.662412060301509e-05,
-      "loss": 5.9207,
-      "step": 13813
-    },
-    {
-      "epoch": 7.204172099087353,
-      "grad_norm": 1.3338100910186768,
-      "learning_rate": 8.662311557788945e-05,
-      "loss": 5.8018,
-      "step": 13814
-    },
-    {
-      "epoch": 7.204693611473273,
-      "grad_norm": 1.3744752407073975,
-      "learning_rate": 8.662211055276383e-05,
-      "loss": 5.5258,
-      "step": 13815
-    },
-    {
-      "epoch": 7.205215123859191,
-      "grad_norm": 1.5686001777648926,
-      "learning_rate": 8.662110552763819e-05,
-      "loss": 5.4953,
-      "step": 13816
-    },
-    {
-      "epoch": 7.205736636245111,
-      "grad_norm": 1.3690036535263062,
-      "learning_rate": 8.662010050251257e-05,
-      "loss": 5.6988,
-      "step": 13817
-    },
-    {
-      "epoch": 7.20625814863103,
-      "grad_norm": 1.290788173675537,
-      "learning_rate": 8.661909547738693e-05,
-      "loss": 6.1169,
-      "step": 13818
-    },
-    {
-      "epoch": 7.206779661016949,
-      "grad_norm": 1.6079181432724,
-      "learning_rate": 8.661809045226131e-05,
-      "loss": 5.3602,
-      "step": 13819
-    },
-    {
-      "epoch": 7.207301173402868,
-      "grad_norm": 1.5802100896835327,
-      "learning_rate": 8.661708542713568e-05,
-      "loss": 5.8747,
-      "step": 13820
-    },
-    {
-      "epoch": 7.207822685788788,
-      "grad_norm": 1.5026919841766357,
-      "learning_rate": 8.661608040201005e-05,
-      "loss": 5.598,
-      "step": 13821
-    },
-    {
-      "epoch": 7.208344198174706,
-      "grad_norm": 1.445643424987793,
-      "learning_rate": 8.661507537688442e-05,
-      "loss": 5.8844,
-      "step": 13822
-    },
-    {
-      "epoch": 7.208865710560626,
-      "grad_norm": 1.4263205528259277,
-      "learning_rate": 8.66140703517588e-05,
-      "loss": 5.6695,
-      "step": 13823
-    },
-    {
-      "epoch": 7.209387222946545,
-      "grad_norm": 1.4534839391708374,
-      "learning_rate": 8.661306532663317e-05,
-      "loss": 5.8565,
-      "step": 13824
-    },
-    {
-      "epoch": 7.209908735332464,
-      "grad_norm": 1.4577950239181519,
-      "learning_rate": 8.661206030150755e-05,
-      "loss": 5.4555,
-      "step": 13825
-    },
-    {
-      "epoch": 7.210430247718383,
-      "grad_norm": 1.5838875770568848,
-      "learning_rate": 8.661105527638192e-05,
-      "loss": 5.6712,
-      "step": 13826
-    },
-    {
-      "epoch": 7.210951760104303,
-      "grad_norm": 1.571539044380188,
-      "learning_rate": 8.661005025125628e-05,
-      "loss": 5.5337,
-      "step": 13827
-    },
-    {
-      "epoch": 7.211473272490221,
-      "grad_norm": 1.3274136781692505,
-      "learning_rate": 8.660904522613066e-05,
-      "loss": 5.7376,
-      "step": 13828
-    },
-    {
-      "epoch": 7.211994784876141,
-      "grad_norm": 1.4326677322387695,
-      "learning_rate": 8.660804020100502e-05,
-      "loss": 5.3089,
-      "step": 13829
-    },
-    {
-      "epoch": 7.21251629726206,
-      "grad_norm": 1.5162547826766968,
-      "learning_rate": 8.66070351758794e-05,
-      "loss": 5.0942,
-      "step": 13830
-    },
-    {
-      "epoch": 7.213037809647979,
-      "grad_norm": 1.4269556999206543,
-      "learning_rate": 8.660603015075376e-05,
-      "loss": 5.9023,
-      "step": 13831
-    },
-    {
-      "epoch": 7.213559322033898,
-      "grad_norm": 1.380919337272644,
-      "learning_rate": 8.660502512562814e-05,
-      "loss": 6.0221,
-      "step": 13832
-    },
-    {
-      "epoch": 7.214080834419818,
-      "grad_norm": 1.4417619705200195,
-      "learning_rate": 8.660402010050252e-05,
-      "loss": 5.618,
-      "step": 13833
-    },
-    {
-      "epoch": 7.214602346805736,
-      "grad_norm": 1.4398239850997925,
-      "learning_rate": 8.66030150753769e-05,
-      "loss": 5.7128,
-      "step": 13834
-    },
-    {
-      "epoch": 7.215123859191656,
-      "grad_norm": 1.526016354560852,
-      "learning_rate": 8.660201005025126e-05,
-      "loss": 5.7215,
-      "step": 13835
-    },
-    {
-      "epoch": 7.215645371577575,
-      "grad_norm": 1.7095297574996948,
-      "learning_rate": 8.660100502512564e-05,
-      "loss": 5.1353,
-      "step": 13836
-    },
-    {
-      "epoch": 7.216166883963494,
-      "grad_norm": 1.6299515962600708,
-      "learning_rate": 8.66e-05,
-      "loss": 5.4414,
-      "step": 13837
-    },
-    {
-      "epoch": 7.216688396349413,
-      "grad_norm": 1.4907594919204712,
-      "learning_rate": 8.659899497487438e-05,
-      "loss": 5.8544,
-      "step": 13838
-    },
-    {
-      "epoch": 7.217209908735333,
-      "grad_norm": 1.2853449583053589,
-      "learning_rate": 8.659798994974875e-05,
-      "loss": 5.1076,
-      "step": 13839
-    },
-    {
-      "epoch": 7.217731421121251,
-      "grad_norm": 1.388832449913025,
-      "learning_rate": 8.659698492462311e-05,
-      "loss": 5.5784,
-      "step": 13840
-    },
-    {
-      "epoch": 7.218252933507171,
-      "grad_norm": 1.4117907285690308,
-      "learning_rate": 8.659597989949749e-05,
-      "loss": 5.0705,
-      "step": 13841
-    },
-    {
-      "epoch": 7.21877444589309,
-      "grad_norm": 1.7573840618133545,
-      "learning_rate": 8.659497487437185e-05,
-      "loss": 5.339,
-      "step": 13842
-    },
-    {
-      "epoch": 7.219295958279009,
-      "grad_norm": 1.5462278127670288,
-      "learning_rate": 8.659396984924623e-05,
-      "loss": 5.2077,
-      "step": 13843
-    },
-    {
-      "epoch": 7.219817470664928,
-      "grad_norm": 1.4318897724151611,
-      "learning_rate": 8.659296482412061e-05,
-      "loss": 5.7215,
-      "step": 13844
-    },
-    {
-      "epoch": 7.220338983050848,
-      "grad_norm": 1.4130226373672485,
-      "learning_rate": 8.659195979899499e-05,
-      "loss": 5.7834,
-      "step": 13845
-    },
-    {
-      "epoch": 7.220860495436766,
-      "grad_norm": 1.4549789428710938,
-      "learning_rate": 8.659095477386935e-05,
-      "loss": 5.5173,
-      "step": 13846
-    },
-    {
-      "epoch": 7.221382007822686,
-      "grad_norm": 1.41904616355896,
-      "learning_rate": 8.658994974874373e-05,
-      "loss": 5.7609,
-      "step": 13847
-    },
-    {
-      "epoch": 7.221903520208605,
-      "grad_norm": 1.4960914850234985,
-      "learning_rate": 8.658894472361809e-05,
-      "loss": 5.6916,
-      "step": 13848
-    },
-    {
-      "epoch": 7.222425032594524,
-      "grad_norm": 1.4038286209106445,
-      "learning_rate": 8.658793969849247e-05,
-      "loss": 5.5725,
-      "step": 13849
-    },
-    {
-      "epoch": 7.222946544980443,
-      "grad_norm": 1.471781611442566,
-      "learning_rate": 8.658693467336683e-05,
-      "loss": 5.7024,
-      "step": 13850
-    },
-    {
-      "epoch": 7.223468057366363,
-      "grad_norm": 1.4794539213180542,
-      "learning_rate": 8.658592964824121e-05,
-      "loss": 4.8575,
-      "step": 13851
-    },
-    {
-      "epoch": 7.223989569752281,
-      "grad_norm": 1.3971178531646729,
-      "learning_rate": 8.658492462311558e-05,
-      "loss": 5.8211,
-      "step": 13852
-    },
-    {
-      "epoch": 7.224511082138201,
-      "grad_norm": 1.606467843055725,
-      "learning_rate": 8.658391959798995e-05,
-      "loss": 5.3796,
-      "step": 13853
-    },
-    {
-      "epoch": 7.22503259452412,
-      "grad_norm": 1.379504919052124,
-      "learning_rate": 8.658291457286433e-05,
-      "loss": 5.7754,
-      "step": 13854
-    },
-    {
-      "epoch": 7.225554106910039,
-      "grad_norm": 1.4252222776412964,
-      "learning_rate": 8.65819095477387e-05,
-      "loss": 5.7975,
-      "step": 13855
-    },
-    {
-      "epoch": 7.226075619295958,
-      "grad_norm": 1.3371676206588745,
-      "learning_rate": 8.658090452261307e-05,
-      "loss": 5.5498,
-      "step": 13856
-    },
-    {
-      "epoch": 7.226597131681878,
-      "grad_norm": 1.43900465965271,
-      "learning_rate": 8.657989949748744e-05,
-      "loss": 5.3648,
-      "step": 13857
-    },
-    {
-      "epoch": 7.227118644067796,
-      "grad_norm": 1.4741994142532349,
-      "learning_rate": 8.657889447236182e-05,
-      "loss": 5.4846,
-      "step": 13858
-    },
-    {
-      "epoch": 7.227640156453716,
-      "grad_norm": 1.5168529748916626,
-      "learning_rate": 8.657788944723618e-05,
-      "loss": 5.5055,
-      "step": 13859
-    },
-    {
-      "epoch": 7.228161668839635,
-      "grad_norm": 1.3755425214767456,
-      "learning_rate": 8.657688442211056e-05,
-      "loss": 5.5566,
-      "step": 13860
-    },
-    {
-      "epoch": 7.228683181225554,
-      "grad_norm": 1.393425464630127,
-      "learning_rate": 8.657587939698492e-05,
-      "loss": 5.5367,
-      "step": 13861
-    },
-    {
-      "epoch": 7.229204693611473,
-      "grad_norm": 1.5195808410644531,
-      "learning_rate": 8.65748743718593e-05,
-      "loss": 5.0902,
-      "step": 13862
-    },
-    {
-      "epoch": 7.229726205997393,
-      "grad_norm": 1.3544471263885498,
-      "learning_rate": 8.657386934673367e-05,
-      "loss": 5.1259,
-      "step": 13863
-    },
-    {
-      "epoch": 7.230247718383311,
-      "grad_norm": 1.2949868440628052,
-      "learning_rate": 8.657286432160804e-05,
-      "loss": 5.8949,
-      "step": 13864
-    },
-    {
-      "epoch": 7.230769230769231,
-      "grad_norm": 1.3344718217849731,
-      "learning_rate": 8.657185929648242e-05,
-      "loss": 5.939,
-      "step": 13865
-    },
-    {
-      "epoch": 7.23129074315515,
-      "grad_norm": 1.58356511592865,
-      "learning_rate": 8.657085427135679e-05,
-      "loss": 5.7261,
-      "step": 13866
-    },
-    {
-      "epoch": 7.231812255541069,
-      "grad_norm": 1.352306842803955,
-      "learning_rate": 8.656984924623116e-05,
-      "loss": 5.7658,
-      "step": 13867
-    },
-    {
-      "epoch": 7.232333767926988,
-      "grad_norm": 1.534409999847412,
-      "learning_rate": 8.656884422110553e-05,
-      "loss": 5.5348,
-      "step": 13868
-    },
-    {
-      "epoch": 7.232855280312908,
-      "grad_norm": 1.5511760711669922,
-      "learning_rate": 8.65678391959799e-05,
-      "loss": 5.496,
-      "step": 13869
-    },
-    {
-      "epoch": 7.233376792698826,
-      "grad_norm": 1.5484240055084229,
-      "learning_rate": 8.656683417085427e-05,
-      "loss": 5.3226,
-      "step": 13870
-    },
-    {
-      "epoch": 7.233898305084746,
-      "grad_norm": 1.3928508758544922,
-      "learning_rate": 8.656582914572865e-05,
-      "loss": 5.5875,
-      "step": 13871
-    },
-    {
-      "epoch": 7.234419817470665,
-      "grad_norm": 1.4767929315567017,
-      "learning_rate": 8.656482412060301e-05,
-      "loss": 5.4868,
-      "step": 13872
-    },
-    {
-      "epoch": 7.234941329856584,
-      "grad_norm": 1.4403691291809082,
-      "learning_rate": 8.656381909547739e-05,
-      "loss": 5.3799,
-      "step": 13873
-    },
-    {
-      "epoch": 7.235462842242503,
-      "grad_norm": 1.6841845512390137,
-      "learning_rate": 8.656281407035177e-05,
-      "loss": 5.1363,
-      "step": 13874
-    },
-    {
-      "epoch": 7.235984354628423,
-      "grad_norm": 1.3859190940856934,
-      "learning_rate": 8.656180904522614e-05,
-      "loss": 5.8355,
-      "step": 13875
-    },
-    {
-      "epoch": 7.236505867014341,
-      "grad_norm": 1.4919509887695312,
-      "learning_rate": 8.656080402010051e-05,
-      "loss": 5.4711,
-      "step": 13876
-    },
-    {
-      "epoch": 7.237027379400261,
-      "grad_norm": 1.3247184753417969,
-      "learning_rate": 8.655979899497489e-05,
-      "loss": 5.4693,
-      "step": 13877
-    },
-    {
-      "epoch": 7.23754889178618,
-      "grad_norm": 1.345331072807312,
-      "learning_rate": 8.655879396984925e-05,
-      "loss": 5.7911,
-      "step": 13878
-    },
-    {
-      "epoch": 7.238070404172099,
-      "grad_norm": 1.5611610412597656,
-      "learning_rate": 8.655778894472362e-05,
-      "loss": 5.0772,
-      "step": 13879
-    },
-    {
-      "epoch": 7.238591916558018,
-      "grad_norm": 1.384142279624939,
-      "learning_rate": 8.6556783919598e-05,
-      "loss": 5.3248,
-      "step": 13880
-    },
-    {
-      "epoch": 7.239113428943938,
-      "grad_norm": 1.534067988395691,
-      "learning_rate": 8.655577889447236e-05,
-      "loss": 5.6008,
-      "step": 13881
-    },
-    {
-      "epoch": 7.239634941329856,
-      "grad_norm": 1.6124788522720337,
-      "learning_rate": 8.655477386934674e-05,
-      "loss": 5.29,
-      "step": 13882
-    },
-    {
-      "epoch": 7.240156453715776,
-      "grad_norm": 1.5609769821166992,
-      "learning_rate": 8.65537688442211e-05,
-      "loss": 5.7622,
-      "step": 13883
-    },
-    {
-      "epoch": 7.240677966101695,
-      "grad_norm": 1.373262882232666,
-      "learning_rate": 8.655276381909548e-05,
-      "loss": 5.6946,
-      "step": 13884
-    },
-    {
-      "epoch": 7.241199478487614,
-      "grad_norm": 1.4647949934005737,
-      "learning_rate": 8.655175879396986e-05,
-      "loss": 5.1776,
-      "step": 13885
-    },
-    {
-      "epoch": 7.241720990873533,
-      "grad_norm": 1.4784338474273682,
-      "learning_rate": 8.655075376884423e-05,
-      "loss": 5.6479,
-      "step": 13886
-    },
-    {
-      "epoch": 7.242242503259453,
-      "grad_norm": 1.5353575944900513,
-      "learning_rate": 8.65497487437186e-05,
-      "loss": 5.4822,
-      "step": 13887
-    },
-    {
-      "epoch": 7.242764015645371,
-      "grad_norm": 1.4073857069015503,
-      "learning_rate": 8.654874371859298e-05,
-      "loss": 5.7631,
-      "step": 13888
-    },
-    {
-      "epoch": 7.243285528031291,
-      "grad_norm": 1.4472273588180542,
-      "learning_rate": 8.654773869346734e-05,
-      "loss": 5.5994,
-      "step": 13889
-    },
-    {
-      "epoch": 7.2438070404172095,
-      "grad_norm": 1.391561508178711,
-      "learning_rate": 8.654673366834172e-05,
-      "loss": 5.8855,
-      "step": 13890
-    },
-    {
-      "epoch": 7.244328552803129,
-      "grad_norm": 1.4057506322860718,
-      "learning_rate": 8.654572864321608e-05,
-      "loss": 5.8038,
-      "step": 13891
-    },
-    {
-      "epoch": 7.244850065189048,
-      "grad_norm": 1.361465573310852,
-      "learning_rate": 8.654472361809046e-05,
-      "loss": 5.8476,
-      "step": 13892
-    },
-    {
-      "epoch": 7.245371577574968,
-      "grad_norm": 1.4027957916259766,
-      "learning_rate": 8.654371859296482e-05,
-      "loss": 5.9451,
-      "step": 13893
-    },
-    {
-      "epoch": 7.245893089960886,
-      "grad_norm": 1.3662689924240112,
-      "learning_rate": 8.65427135678392e-05,
-      "loss": 5.9848,
-      "step": 13894
-    },
-    {
-      "epoch": 7.246414602346806,
-      "grad_norm": 1.7352598905563354,
-      "learning_rate": 8.654170854271358e-05,
-      "loss": 4.7504,
-      "step": 13895
-    },
-    {
-      "epoch": 7.2469361147327245,
-      "grad_norm": 1.3858022689819336,
-      "learning_rate": 8.654070351758794e-05,
-      "loss": 5.8145,
-      "step": 13896
-    },
-    {
-      "epoch": 7.247457627118644,
-      "grad_norm": 1.515507459640503,
-      "learning_rate": 8.653969849246232e-05,
-      "loss": 5.3422,
-      "step": 13897
-    },
-    {
-      "epoch": 7.247979139504563,
-      "grad_norm": 1.4698268175125122,
-      "learning_rate": 8.653869346733669e-05,
-      "loss": 5.3281,
-      "step": 13898
-    },
-    {
-      "epoch": 7.248500651890482,
-      "grad_norm": 1.384896993637085,
-      "learning_rate": 8.653768844221106e-05,
-      "loss": 5.9505,
-      "step": 13899
-    },
-    {
-      "epoch": 7.2490221642764014,
-      "grad_norm": 1.4043617248535156,
-      "learning_rate": 8.653668341708543e-05,
-      "loss": 5.6624,
-      "step": 13900
-    },
-    {
-      "epoch": 7.249543676662321,
-      "grad_norm": 1.4650789499282837,
-      "learning_rate": 8.65356783919598e-05,
-      "loss": 5.2496,
-      "step": 13901
-    },
-    {
-      "epoch": 7.2500651890482395,
-      "grad_norm": 1.5133768320083618,
-      "learning_rate": 8.653467336683417e-05,
-      "loss": 5.5005,
-      "step": 13902
-    },
-    {
-      "epoch": 7.250586701434159,
-      "grad_norm": 1.4040523767471313,
-      "learning_rate": 8.653366834170855e-05,
-      "loss": 5.7626,
-      "step": 13903
-    },
-    {
-      "epoch": 7.251108213820078,
-      "grad_norm": 1.4309961795806885,
-      "learning_rate": 8.653266331658291e-05,
-      "loss": 5.6769,
-      "step": 13904
-    },
-    {
-      "epoch": 7.251629726205997,
-      "grad_norm": 1.3854128122329712,
-      "learning_rate": 8.653165829145729e-05,
-      "loss": 5.9449,
-      "step": 13905
-    },
-    {
-      "epoch": 7.2521512385919165,
-      "grad_norm": 1.3234103918075562,
-      "learning_rate": 8.653065326633167e-05,
-      "loss": 5.6511,
-      "step": 13906
-    },
-    {
-      "epoch": 7.252672750977836,
-      "grad_norm": 1.3531053066253662,
-      "learning_rate": 8.652964824120603e-05,
-      "loss": 5.9423,
-      "step": 13907
-    },
-    {
-      "epoch": 7.2531942633637545,
-      "grad_norm": 1.396986961364746,
-      "learning_rate": 8.652864321608041e-05,
-      "loss": 5.2624,
-      "step": 13908
-    },
-    {
-      "epoch": 7.253715775749674,
-      "grad_norm": 1.416448950767517,
-      "learning_rate": 8.652763819095477e-05,
-      "loss": 5.5106,
-      "step": 13909
-    },
-    {
-      "epoch": 7.254237288135593,
-      "grad_norm": 1.3871735334396362,
-      "learning_rate": 8.652663316582915e-05,
-      "loss": 5.6092,
-      "step": 13910
-    },
-    {
-      "epoch": 7.254758800521512,
-      "grad_norm": 1.4899522066116333,
-      "learning_rate": 8.652562814070352e-05,
-      "loss": 5.6682,
-      "step": 13911
-    },
-    {
-      "epoch": 7.2552803129074315,
-      "grad_norm": 1.3787933588027954,
-      "learning_rate": 8.65246231155779e-05,
-      "loss": 5.2395,
-      "step": 13912
-    },
-    {
-      "epoch": 7.255801825293351,
-      "grad_norm": 1.376882791519165,
-      "learning_rate": 8.652361809045226e-05,
-      "loss": 5.8448,
-      "step": 13913
-    },
-    {
-      "epoch": 7.2563233376792695,
-      "grad_norm": 1.4317625761032104,
-      "learning_rate": 8.652261306532664e-05,
-      "loss": 5.9074,
-      "step": 13914
-    },
-    {
-      "epoch": 7.256844850065189,
-      "grad_norm": 1.4317023754119873,
-      "learning_rate": 8.652160804020101e-05,
-      "loss": 5.4882,
-      "step": 13915
-    },
-    {
-      "epoch": 7.257366362451108,
-      "grad_norm": 1.519873023033142,
-      "learning_rate": 8.652060301507539e-05,
-      "loss": 5.3983,
-      "step": 13916
-    },
-    {
-      "epoch": 7.257887874837027,
-      "grad_norm": 1.4214595556259155,
-      "learning_rate": 8.651959798994976e-05,
-      "loss": 5.4822,
-      "step": 13917
-    },
-    {
-      "epoch": 7.2584093872229465,
-      "grad_norm": 1.426478385925293,
-      "learning_rate": 8.651859296482413e-05,
-      "loss": 5.2393,
-      "step": 13918
-    },
-    {
-      "epoch": 7.258930899608866,
-      "grad_norm": 1.3471466302871704,
-      "learning_rate": 8.65175879396985e-05,
-      "loss": 5.5461,
-      "step": 13919
-    },
-    {
-      "epoch": 7.2594524119947845,
-      "grad_norm": 1.3822931051254272,
-      "learning_rate": 8.651658291457286e-05,
-      "loss": 5.8928,
-      "step": 13920
-    },
-    {
-      "epoch": 7.259973924380704,
-      "grad_norm": 1.4765878915786743,
-      "learning_rate": 8.651557788944724e-05,
-      "loss": 5.5311,
-      "step": 13921
-    },
-    {
-      "epoch": 7.2604954367666235,
-      "grad_norm": 1.3334802389144897,
-      "learning_rate": 8.65145728643216e-05,
-      "loss": 5.2608,
-      "step": 13922
-    },
-    {
-      "epoch": 7.261016949152542,
-      "grad_norm": 1.4536545276641846,
-      "learning_rate": 8.651356783919598e-05,
-      "loss": 5.5716,
-      "step": 13923
-    },
-    {
-      "epoch": 7.2615384615384615,
-      "grad_norm": 1.436621069908142,
-      "learning_rate": 8.651256281407035e-05,
-      "loss": 5.4864,
-      "step": 13924
-    },
-    {
-      "epoch": 7.262059973924381,
-      "grad_norm": 1.6455384492874146,
-      "learning_rate": 8.651155778894472e-05,
-      "loss": 5.2419,
-      "step": 13925
-    },
-    {
-      "epoch": 7.2625814863102995,
-      "grad_norm": 1.4164351224899292,
-      "learning_rate": 8.65105527638191e-05,
-      "loss": 5.7885,
-      "step": 13926
-    },
-    {
-      "epoch": 7.263102998696219,
-      "grad_norm": 1.4594738483428955,
-      "learning_rate": 8.650954773869348e-05,
-      "loss": 5.7336,
-      "step": 13927
-    },
-    {
-      "epoch": 7.2636245110821385,
-      "grad_norm": 1.5486878156661987,
-      "learning_rate": 8.650854271356784e-05,
-      "loss": 5.3165,
-      "step": 13928
-    },
-    {
-      "epoch": 7.264146023468057,
-      "grad_norm": 1.3693488836288452,
-      "learning_rate": 8.650753768844222e-05,
-      "loss": 5.9375,
-      "step": 13929
-    },
-    {
-      "epoch": 7.2646675358539765,
-      "grad_norm": 1.5233376026153564,
-      "learning_rate": 8.650653266331659e-05,
-      "loss": 5.4604,
-      "step": 13930
-    },
-    {
-      "epoch": 7.265189048239896,
-      "grad_norm": 1.3979462385177612,
-      "learning_rate": 8.650552763819096e-05,
-      "loss": 5.8789,
-      "step": 13931
-    },
-    {
-      "epoch": 7.2657105606258146,
-      "grad_norm": 1.4136751890182495,
-      "learning_rate": 8.650452261306533e-05,
-      "loss": 5.5115,
-      "step": 13932
-    },
-    {
-      "epoch": 7.266232073011734,
-      "grad_norm": 1.3632471561431885,
-      "learning_rate": 8.650351758793969e-05,
-      "loss": 5.7081,
-      "step": 13933
-    },
-    {
-      "epoch": 7.2667535853976535,
-      "grad_norm": 1.3664360046386719,
-      "learning_rate": 8.650251256281407e-05,
-      "loss": 5.8476,
-      "step": 13934
-    },
-    {
-      "epoch": 7.267275097783572,
-      "grad_norm": 1.2854349613189697,
-      "learning_rate": 8.650150753768845e-05,
-      "loss": 5.9687,
-      "step": 13935
-    },
-    {
-      "epoch": 7.2677966101694915,
-      "grad_norm": 1.4060533046722412,
-      "learning_rate": 8.650050251256283e-05,
-      "loss": 5.5127,
-      "step": 13936
-    },
-    {
-      "epoch": 7.268318122555411,
-      "grad_norm": 1.4356714487075806,
-      "learning_rate": 8.649949748743719e-05,
-      "loss": 5.4107,
-      "step": 13937
-    },
-    {
-      "epoch": 7.26883963494133,
-      "grad_norm": 1.4627318382263184,
-      "learning_rate": 8.649849246231157e-05,
-      "loss": 5.6968,
-      "step": 13938
-    },
-    {
-      "epoch": 7.269361147327249,
-      "grad_norm": 1.5060369968414307,
-      "learning_rate": 8.649748743718593e-05,
-      "loss": 5.4215,
-      "step": 13939
-    },
-    {
-      "epoch": 7.2698826597131685,
-      "grad_norm": 1.5632301568984985,
-      "learning_rate": 8.649648241206031e-05,
-      "loss": 5.4403,
-      "step": 13940
-    },
-    {
-      "epoch": 7.270404172099087,
-      "grad_norm": 1.3948538303375244,
-      "learning_rate": 8.649547738693468e-05,
-      "loss": 5.4469,
-      "step": 13941
-    },
-    {
-      "epoch": 7.2709256844850065,
-      "grad_norm": 1.414812445640564,
-      "learning_rate": 8.649447236180905e-05,
-      "loss": 5.3449,
-      "step": 13942
-    },
-    {
-      "epoch": 7.271447196870926,
-      "grad_norm": 1.3936457633972168,
-      "learning_rate": 8.649346733668342e-05,
-      "loss": 5.6481,
-      "step": 13943
-    },
-    {
-      "epoch": 7.271968709256845,
-      "grad_norm": 1.3703370094299316,
-      "learning_rate": 8.64924623115578e-05,
-      "loss": 5.6422,
-      "step": 13944
-    },
-    {
-      "epoch": 7.272490221642764,
-      "grad_norm": 1.4949051141738892,
-      "learning_rate": 8.649145728643216e-05,
-      "loss": 5.4538,
-      "step": 13945
-    },
-    {
-      "epoch": 7.2730117340286835,
-      "grad_norm": 1.5425009727478027,
-      "learning_rate": 8.649045226130654e-05,
-      "loss": 5.3842,
-      "step": 13946
-    },
-    {
-      "epoch": 7.273533246414602,
-      "grad_norm": 1.4803504943847656,
-      "learning_rate": 8.648944723618091e-05,
-      "loss": 5.7159,
-      "step": 13947
-    },
-    {
-      "epoch": 7.2740547588005215,
-      "grad_norm": 1.4635493755340576,
-      "learning_rate": 8.648844221105528e-05,
-      "loss": 5.4063,
-      "step": 13948
-    },
-    {
-      "epoch": 7.274576271186441,
-      "grad_norm": 1.4413903951644897,
-      "learning_rate": 8.648743718592966e-05,
-      "loss": 5.4398,
-      "step": 13949
-    },
-    {
-      "epoch": 7.27509778357236,
-      "grad_norm": 1.7470332384109497,
-      "learning_rate": 8.648643216080402e-05,
-      "loss": 5.207,
-      "step": 13950
-    },
-    {
-      "epoch": 7.275619295958279,
-      "grad_norm": 1.4892231225967407,
-      "learning_rate": 8.64854271356784e-05,
-      "loss": 5.7095,
-      "step": 13951
-    },
-    {
-      "epoch": 7.2761408083441985,
-      "grad_norm": 1.4328135251998901,
-      "learning_rate": 8.648442211055276e-05,
-      "loss": 5.8331,
-      "step": 13952
-    },
-    {
-      "epoch": 7.276662320730117,
-      "grad_norm": 1.4619526863098145,
-      "learning_rate": 8.648341708542714e-05,
-      "loss": 5.6052,
-      "step": 13953
-    },
-    {
-      "epoch": 7.277183833116037,
-      "grad_norm": 1.5187896490097046,
-      "learning_rate": 8.64824120603015e-05,
-      "loss": 5.1837,
-      "step": 13954
-    },
-    {
-      "epoch": 7.277705345501956,
-      "grad_norm": 1.334855318069458,
-      "learning_rate": 8.648140703517588e-05,
-      "loss": 5.5522,
-      "step": 13955
-    },
-    {
-      "epoch": 7.278226857887875,
-      "grad_norm": 1.3558253049850464,
-      "learning_rate": 8.648040201005025e-05,
-      "loss": 5.5673,
-      "step": 13956
-    },
-    {
-      "epoch": 7.278748370273794,
-      "grad_norm": 1.40957510471344,
-      "learning_rate": 8.647939698492463e-05,
-      "loss": 5.6261,
-      "step": 13957
-    },
-    {
-      "epoch": 7.2792698826597135,
-      "grad_norm": 1.423427700996399,
-      "learning_rate": 8.6478391959799e-05,
-      "loss": 4.9521,
-      "step": 13958
-    },
-    {
-      "epoch": 7.279791395045632,
-      "grad_norm": 1.3916763067245483,
-      "learning_rate": 8.647738693467337e-05,
-      "loss": 5.9292,
-      "step": 13959
-    },
-    {
-      "epoch": 7.280312907431552,
-      "grad_norm": 1.3993924856185913,
-      "learning_rate": 8.647638190954775e-05,
-      "loss": 5.631,
-      "step": 13960
-    },
-    {
-      "epoch": 7.280834419817471,
-      "grad_norm": 1.441568374633789,
-      "learning_rate": 8.647537688442211e-05,
-      "loss": 4.9591,
-      "step": 13961
-    },
-    {
-      "epoch": 7.28135593220339,
-      "grad_norm": 1.2964200973510742,
-      "learning_rate": 8.647437185929649e-05,
-      "loss": 5.8975,
-      "step": 13962
-    },
-    {
-      "epoch": 7.281877444589309,
-      "grad_norm": 1.4882776737213135,
-      "learning_rate": 8.647336683417085e-05,
-      "loss": 5.5268,
-      "step": 13963
-    },
-    {
-      "epoch": 7.2823989569752285,
-      "grad_norm": 1.4841235876083374,
-      "learning_rate": 8.647236180904523e-05,
-      "loss": 5.8094,
-      "step": 13964
-    },
-    {
-      "epoch": 7.282920469361147,
-      "grad_norm": 1.41879141330719,
-      "learning_rate": 8.64713567839196e-05,
-      "loss": 5.435,
-      "step": 13965
-    },
-    {
-      "epoch": 7.283441981747067,
-      "grad_norm": 1.3464205265045166,
-      "learning_rate": 8.647035175879397e-05,
-      "loss": 5.8486,
-      "step": 13966
-    },
-    {
-      "epoch": 7.283963494132986,
-      "grad_norm": 1.497770071029663,
-      "learning_rate": 8.646934673366835e-05,
-      "loss": 5.617,
-      "step": 13967
-    },
-    {
-      "epoch": 7.284485006518905,
-      "grad_norm": 1.4082905054092407,
-      "learning_rate": 8.646834170854273e-05,
-      "loss": 5.5819,
-      "step": 13968
-    },
-    {
-      "epoch": 7.285006518904824,
-      "grad_norm": 1.4159069061279297,
-      "learning_rate": 8.646733668341709e-05,
-      "loss": 5.5347,
-      "step": 13969
-    },
-    {
-      "epoch": 7.2855280312907436,
-      "grad_norm": 1.3514378070831299,
-      "learning_rate": 8.646633165829147e-05,
-      "loss": 5.0306,
-      "step": 13970
-    },
-    {
-      "epoch": 7.286049543676662,
-      "grad_norm": 1.4812904596328735,
-      "learning_rate": 8.646532663316583e-05,
-      "loss": 5.6348,
-      "step": 13971
-    },
-    {
-      "epoch": 7.286571056062582,
-      "grad_norm": 1.380391001701355,
-      "learning_rate": 8.64643216080402e-05,
-      "loss": 6.0167,
-      "step": 13972
-    },
-    {
-      "epoch": 7.287092568448501,
-      "grad_norm": 1.4003443717956543,
-      "learning_rate": 8.646331658291458e-05,
-      "loss": 5.5624,
-      "step": 13973
-    },
-    {
-      "epoch": 7.28761408083442,
-      "grad_norm": 1.4305994510650635,
-      "learning_rate": 8.646231155778894e-05,
-      "loss": 5.9424,
-      "step": 13974
-    },
-    {
-      "epoch": 7.288135593220339,
-      "grad_norm": 1.419061541557312,
-      "learning_rate": 8.646130653266332e-05,
-      "loss": 5.9904,
-      "step": 13975
-    },
-    {
-      "epoch": 7.288657105606259,
-      "grad_norm": 1.5180294513702393,
-      "learning_rate": 8.646030150753768e-05,
-      "loss": 5.2705,
-      "step": 13976
-    },
-    {
-      "epoch": 7.289178617992177,
-      "grad_norm": 1.5428037643432617,
-      "learning_rate": 8.645929648241206e-05,
-      "loss": 5.787,
-      "step": 13977
-    },
-    {
-      "epoch": 7.289700130378097,
-      "grad_norm": 1.4225468635559082,
-      "learning_rate": 8.645829145728644e-05,
-      "loss": 5.4238,
-      "step": 13978
-    },
-    {
-      "epoch": 7.290221642764015,
-      "grad_norm": 1.3030236959457397,
-      "learning_rate": 8.645728643216082e-05,
-      "loss": 5.4661,
-      "step": 13979
-    },
-    {
-      "epoch": 7.290743155149935,
-      "grad_norm": 1.369858980178833,
-      "learning_rate": 8.645628140703518e-05,
-      "loss": 5.5244,
-      "step": 13980
-    },
-    {
-      "epoch": 7.291264667535854,
-      "grad_norm": 1.536669135093689,
-      "learning_rate": 8.645527638190956e-05,
-      "loss": 5.4945,
-      "step": 13981
-    },
-    {
-      "epoch": 7.291786179921774,
-      "grad_norm": 1.4402844905853271,
-      "learning_rate": 8.645427135678392e-05,
-      "loss": 5.644,
-      "step": 13982
-    },
-    {
-      "epoch": 7.292307692307692,
-      "grad_norm": 1.3958327770233154,
-      "learning_rate": 8.64532663316583e-05,
-      "loss": 5.6003,
-      "step": 13983
-    },
-    {
-      "epoch": 7.292829204693612,
-      "grad_norm": 1.4838604927062988,
-      "learning_rate": 8.645226130653266e-05,
-      "loss": 5.3198,
-      "step": 13984
-    },
-    {
-      "epoch": 7.29335071707953,
-      "grad_norm": 1.375259518623352,
-      "learning_rate": 8.645125628140704e-05,
-      "loss": 5.855,
-      "step": 13985
-    },
-    {
-      "epoch": 7.29387222946545,
-      "grad_norm": 1.4475239515304565,
-      "learning_rate": 8.64502512562814e-05,
-      "loss": 5.4398,
-      "step": 13986
-    },
-    {
-      "epoch": 7.294393741851369,
-      "grad_norm": 1.4006599187850952,
-      "learning_rate": 8.644924623115578e-05,
-      "loss": 5.9076,
-      "step": 13987
-    },
-    {
-      "epoch": 7.294915254237289,
-      "grad_norm": 1.486405849456787,
-      "learning_rate": 8.644824120603016e-05,
-      "loss": 5.5671,
-      "step": 13988
-    },
-    {
-      "epoch": 7.295436766623207,
-      "grad_norm": 1.4042093753814697,
-      "learning_rate": 8.644723618090453e-05,
-      "loss": 5.7671,
-      "step": 13989
-    },
-    {
-      "epoch": 7.295958279009127,
-      "grad_norm": 1.550836443901062,
-      "learning_rate": 8.64462311557789e-05,
-      "loss": 5.1654,
-      "step": 13990
-    },
-    {
-      "epoch": 7.296479791395045,
-      "grad_norm": 1.4479562044143677,
-      "learning_rate": 8.644522613065327e-05,
-      "loss": 5.5512,
-      "step": 13991
-    },
-    {
-      "epoch": 7.297001303780965,
-      "grad_norm": 1.47080659866333,
-      "learning_rate": 8.644422110552765e-05,
-      "loss": 5.7425,
-      "step": 13992
-    },
-    {
-      "epoch": 7.297522816166884,
-      "grad_norm": 1.6019738912582397,
-      "learning_rate": 8.644321608040201e-05,
-      "loss": 5.4217,
-      "step": 13993
-    },
-    {
-      "epoch": 7.298044328552803,
-      "grad_norm": 1.4677263498306274,
-      "learning_rate": 8.644221105527639e-05,
-      "loss": 5.3774,
-      "step": 13994
-    },
-    {
-      "epoch": 7.298565840938722,
-      "grad_norm": 1.6464918851852417,
-      "learning_rate": 8.644120603015075e-05,
-      "loss": 5.3443,
-      "step": 13995
-    },
-    {
-      "epoch": 7.299087353324642,
-      "grad_norm": 1.5800527334213257,
-      "learning_rate": 8.644020100502513e-05,
-      "loss": 5.2024,
-      "step": 13996
-    },
-    {
-      "epoch": 7.29960886571056,
-      "grad_norm": 1.4476515054702759,
-      "learning_rate": 8.64391959798995e-05,
-      "loss": 5.7206,
-      "step": 13997
-    },
-    {
-      "epoch": 7.30013037809648,
-      "grad_norm": 1.4453465938568115,
-      "learning_rate": 8.643819095477387e-05,
-      "loss": 5.7064,
-      "step": 13998
-    },
-    {
-      "epoch": 7.300651890482399,
-      "grad_norm": 1.5834791660308838,
-      "learning_rate": 8.643718592964825e-05,
-      "loss": 5.1279,
-      "step": 13999
-    },
-    {
-      "epoch": 7.301173402868318,
-      "grad_norm": 1.8103598356246948,
-      "learning_rate": 8.643618090452261e-05,
-      "loss": 5.2943,
-      "step": 14000
-    },
-    {
-      "epoch": 7.301694915254237,
-      "grad_norm": 1.6738951206207275,
-      "learning_rate": 8.643517587939699e-05,
-      "loss": 5.5403,
-      "step": 14001
-    },
-    {
-      "epoch": 7.302216427640157,
-      "grad_norm": 1.5249449014663696,
-      "learning_rate": 8.643417085427136e-05,
-      "loss": 5.3345,
-      "step": 14002
-    },
-    {
-      "epoch": 7.302737940026075,
-      "grad_norm": 1.4163775444030762,
-      "learning_rate": 8.643316582914573e-05,
-      "loss": 5.8385,
-      "step": 14003
-    },
-    {
-      "epoch": 7.303259452411995,
-      "grad_norm": 1.4143809080123901,
-      "learning_rate": 8.64321608040201e-05,
-      "loss": 5.4805,
-      "step": 14004
-    },
-    {
-      "epoch": 7.303780964797914,
-      "grad_norm": 1.3905314207077026,
-      "learning_rate": 8.643115577889448e-05,
-      "loss": 5.5365,
-      "step": 14005
-    },
-    {
-      "epoch": 7.304302477183833,
-      "grad_norm": 1.403387427330017,
-      "learning_rate": 8.643015075376884e-05,
-      "loss": 5.9316,
-      "step": 14006
-    },
-    {
-      "epoch": 7.304823989569752,
-      "grad_norm": 1.5198328495025635,
-      "learning_rate": 8.642914572864322e-05,
-      "loss": 5.6235,
-      "step": 14007
-    },
-    {
-      "epoch": 7.305345501955672,
-      "grad_norm": 1.6171300411224365,
-      "learning_rate": 8.64281407035176e-05,
-      "loss": 5.3242,
-      "step": 14008
-    },
-    {
-      "epoch": 7.30586701434159,
-      "grad_norm": 1.378237009048462,
-      "learning_rate": 8.642713567839197e-05,
-      "loss": 5.6539,
-      "step": 14009
-    },
-    {
-      "epoch": 7.30638852672751,
-      "grad_norm": 1.3996731042861938,
-      "learning_rate": 8.642613065326634e-05,
-      "loss": 5.6495,
-      "step": 14010
-    },
-    {
-      "epoch": 7.306910039113429,
-      "grad_norm": 1.5628018379211426,
-      "learning_rate": 8.642512562814072e-05,
-      "loss": 5.5354,
-      "step": 14011
-    },
-    {
-      "epoch": 7.307431551499348,
-      "grad_norm": 1.536037564277649,
-      "learning_rate": 8.642412060301508e-05,
-      "loss": 5.3698,
-      "step": 14012
-    },
-    {
-      "epoch": 7.307953063885267,
-      "grad_norm": 1.4203882217407227,
-      "learning_rate": 8.642311557788944e-05,
-      "loss": 5.5838,
-      "step": 14013
-    },
-    {
-      "epoch": 7.308474576271187,
-      "grad_norm": 1.611579179763794,
-      "learning_rate": 8.642211055276382e-05,
-      "loss": 5.2257,
-      "step": 14014
-    },
-    {
-      "epoch": 7.308996088657105,
-      "grad_norm": 1.7825345993041992,
-      "learning_rate": 8.642110552763819e-05,
-      "loss": 5.3274,
-      "step": 14015
-    },
-    {
-      "epoch": 7.309517601043025,
-      "grad_norm": 1.5274690389633179,
-      "learning_rate": 8.642010050251256e-05,
-      "loss": 5.5968,
-      "step": 14016
-    },
-    {
-      "epoch": 7.310039113428944,
-      "grad_norm": 1.407712697982788,
-      "learning_rate": 8.641909547738693e-05,
-      "loss": 5.7945,
-      "step": 14017
-    },
-    {
-      "epoch": 7.310560625814863,
-      "grad_norm": 1.454009771347046,
-      "learning_rate": 8.641809045226131e-05,
-      "loss": 5.3691,
-      "step": 14018
-    },
-    {
-      "epoch": 7.311082138200782,
-      "grad_norm": 1.4856528043746948,
-      "learning_rate": 8.641708542713568e-05,
-      "loss": 5.0777,
-      "step": 14019
-    },
-    {
-      "epoch": 7.311603650586702,
-      "grad_norm": 1.525679111480713,
-      "learning_rate": 8.641608040201006e-05,
-      "loss": 4.8728,
-      "step": 14020
-    },
-    {
-      "epoch": 7.31212516297262,
-      "grad_norm": 1.3854422569274902,
-      "learning_rate": 8.641507537688443e-05,
-      "loss": 5.5315,
-      "step": 14021
-    },
-    {
-      "epoch": 7.31264667535854,
-      "grad_norm": 1.365897536277771,
-      "learning_rate": 8.64140703517588e-05,
-      "loss": 5.3863,
-      "step": 14022
-    },
-    {
-      "epoch": 7.313168187744459,
-      "grad_norm": 1.3100353479385376,
-      "learning_rate": 8.641306532663317e-05,
-      "loss": 5.5204,
-      "step": 14023
-    },
-    {
-      "epoch": 7.313689700130378,
-      "grad_norm": 1.429561972618103,
-      "learning_rate": 8.641206030150755e-05,
-      "loss": 5.7937,
-      "step": 14024
-    },
-    {
-      "epoch": 7.314211212516297,
-      "grad_norm": 1.5019487142562866,
-      "learning_rate": 8.641105527638191e-05,
-      "loss": 5.6593,
-      "step": 14025
-    },
-    {
-      "epoch": 7.314732724902217,
-      "grad_norm": 1.2921549081802368,
-      "learning_rate": 8.641005025125628e-05,
-      "loss": 5.7422,
-      "step": 14026
-    },
-    {
-      "epoch": 7.315254237288135,
-      "grad_norm": 1.4780654907226562,
-      "learning_rate": 8.640904522613065e-05,
-      "loss": 5.6734,
-      "step": 14027
-    },
-    {
-      "epoch": 7.315775749674055,
-      "grad_norm": 1.5460532903671265,
-      "learning_rate": 8.640804020100503e-05,
-      "loss": 5.18,
-      "step": 14028
-    },
-    {
-      "epoch": 7.316297262059974,
-      "grad_norm": 1.673261284828186,
-      "learning_rate": 8.640703517587941e-05,
-      "loss": 5.6863,
-      "step": 14029
-    },
-    {
-      "epoch": 7.316818774445893,
-      "grad_norm": 1.5857906341552734,
-      "learning_rate": 8.640603015075377e-05,
-      "loss": 5.2165,
-      "step": 14030
-    },
-    {
-      "epoch": 7.317340286831812,
-      "grad_norm": 1.2880761623382568,
-      "learning_rate": 8.640502512562815e-05,
-      "loss": 6.1114,
-      "step": 14031
-    },
-    {
-      "epoch": 7.317861799217732,
-      "grad_norm": 1.600763201713562,
-      "learning_rate": 8.640402010050252e-05,
-      "loss": 4.9352,
-      "step": 14032
-    },
-    {
-      "epoch": 7.31838331160365,
-      "grad_norm": 1.4753471612930298,
-      "learning_rate": 8.640301507537689e-05,
-      "loss": 5.224,
-      "step": 14033
-    },
-    {
-      "epoch": 7.31890482398957,
-      "grad_norm": 1.3306764364242554,
-      "learning_rate": 8.640201005025126e-05,
-      "loss": 5.7428,
-      "step": 14034
-    },
-    {
-      "epoch": 7.319426336375489,
-      "grad_norm": 1.3086230754852295,
-      "learning_rate": 8.640100502512564e-05,
-      "loss": 5.7517,
-      "step": 14035
-    },
-    {
-      "epoch": 7.319947848761408,
-      "grad_norm": 1.421722650527954,
-      "learning_rate": 8.64e-05,
-      "loss": 5.7174,
-      "step": 14036
-    },
-    {
-      "epoch": 7.320469361147327,
-      "grad_norm": 1.365938663482666,
-      "learning_rate": 8.639899497487438e-05,
-      "loss": 5.7607,
-      "step": 14037
-    },
-    {
-      "epoch": 7.320990873533247,
-      "grad_norm": 1.5328855514526367,
-      "learning_rate": 8.639798994974874e-05,
-      "loss": 5.2778,
-      "step": 14038
-    },
-    {
-      "epoch": 7.321512385919165,
-      "grad_norm": 1.4467675685882568,
-      "learning_rate": 8.639698492462312e-05,
-      "loss": 5.6991,
-      "step": 14039
-    },
-    {
-      "epoch": 7.322033898305085,
-      "grad_norm": 1.4114993810653687,
-      "learning_rate": 8.63959798994975e-05,
-      "loss": 5.5077,
-      "step": 14040
-    },
-    {
-      "epoch": 7.322555410691004,
-      "grad_norm": 1.5762125253677368,
-      "learning_rate": 8.639497487437186e-05,
-      "loss": 5.3227,
-      "step": 14041
-    },
-    {
-      "epoch": 7.323076923076923,
-      "grad_norm": 1.4283225536346436,
-      "learning_rate": 8.639396984924624e-05,
-      "loss": 5.7728,
-      "step": 14042
-    },
-    {
-      "epoch": 7.323598435462842,
-      "grad_norm": 1.5265038013458252,
-      "learning_rate": 8.63929648241206e-05,
-      "loss": 5.4598,
-      "step": 14043
-    },
-    {
-      "epoch": 7.324119947848762,
-      "grad_norm": 1.6257599592208862,
-      "learning_rate": 8.639195979899498e-05,
-      "loss": 5.2035,
-      "step": 14044
-    },
-    {
-      "epoch": 7.32464146023468,
-      "grad_norm": 1.5463647842407227,
-      "learning_rate": 8.639095477386935e-05,
-      "loss": 5.4449,
-      "step": 14045
-    },
-    {
-      "epoch": 7.3251629726206,
-      "grad_norm": 1.5769624710083008,
-      "learning_rate": 8.638994974874372e-05,
-      "loss": 4.8303,
-      "step": 14046
-    },
-    {
-      "epoch": 7.325684485006519,
-      "grad_norm": 1.517297387123108,
-      "learning_rate": 8.638894472361809e-05,
-      "loss": 5.5122,
-      "step": 14047
-    },
-    {
-      "epoch": 7.326205997392438,
-      "grad_norm": 1.4957104921340942,
-      "learning_rate": 8.638793969849247e-05,
-      "loss": 5.418,
-      "step": 14048
-    },
-    {
-      "epoch": 7.326727509778357,
-      "grad_norm": 1.4642125368118286,
-      "learning_rate": 8.638693467336684e-05,
-      "loss": 5.5927,
-      "step": 14049
-    },
-    {
-      "epoch": 7.327249022164277,
-      "grad_norm": 1.4118324518203735,
-      "learning_rate": 8.638592964824122e-05,
-      "loss": 5.8966,
-      "step": 14050
-    },
-    {
-      "epoch": 7.327770534550195,
-      "grad_norm": 1.5057858228683472,
-      "learning_rate": 8.638492462311559e-05,
-      "loss": 4.9099,
-      "step": 14051
-    },
-    {
-      "epoch": 7.328292046936115,
-      "grad_norm": 1.4634974002838135,
-      "learning_rate": 8.638391959798995e-05,
-      "loss": 5.7176,
-      "step": 14052
-    },
-    {
-      "epoch": 7.328813559322034,
-      "grad_norm": 1.3176823854446411,
-      "learning_rate": 8.638291457286433e-05,
-      "loss": 5.8532,
-      "step": 14053
-    },
-    {
-      "epoch": 7.329335071707953,
-      "grad_norm": 1.4588236808776855,
-      "learning_rate": 8.638190954773869e-05,
-      "loss": 4.7157,
-      "step": 14054
-    },
-    {
-      "epoch": 7.329856584093872,
-      "grad_norm": 1.3887196779251099,
-      "learning_rate": 8.638090452261307e-05,
-      "loss": 5.862,
-      "step": 14055
-    },
-    {
-      "epoch": 7.330378096479792,
-      "grad_norm": 1.4252700805664062,
-      "learning_rate": 8.637989949748743e-05,
-      "loss": 5.6735,
-      "step": 14056
-    },
-    {
-      "epoch": 7.33089960886571,
-      "grad_norm": 1.5182180404663086,
-      "learning_rate": 8.637889447236181e-05,
-      "loss": 5.7014,
-      "step": 14057
-    },
-    {
-      "epoch": 7.33142112125163,
-      "grad_norm": 1.3444551229476929,
-      "learning_rate": 8.637788944723618e-05,
-      "loss": 5.4334,
-      "step": 14058
-    },
-    {
-      "epoch": 7.331942633637549,
-      "grad_norm": 1.405275583267212,
-      "learning_rate": 8.637688442211055e-05,
-      "loss": 5.6922,
-      "step": 14059
-    },
-    {
-      "epoch": 7.332464146023468,
-      "grad_norm": 1.2852128744125366,
-      "learning_rate": 8.637587939698493e-05,
-      "loss": 6.1625,
-      "step": 14060
-    },
-    {
-      "epoch": 7.332985658409387,
-      "grad_norm": 1.7548332214355469,
-      "learning_rate": 8.637487437185931e-05,
-      "loss": 4.9837,
-      "step": 14061
-    },
-    {
-      "epoch": 7.333507170795307,
-      "grad_norm": 1.4273828268051147,
-      "learning_rate": 8.637386934673367e-05,
-      "loss": 5.5702,
-      "step": 14062
-    },
-    {
-      "epoch": 7.334028683181225,
-      "grad_norm": 1.4239511489868164,
-      "learning_rate": 8.637286432160805e-05,
-      "loss": 5.7775,
-      "step": 14063
-    },
-    {
-      "epoch": 7.334550195567145,
-      "grad_norm": 1.5540846586227417,
-      "learning_rate": 8.637185929648242e-05,
-      "loss": 5.0651,
-      "step": 14064
-    },
-    {
-      "epoch": 7.335071707953064,
-      "grad_norm": 1.4383010864257812,
-      "learning_rate": 8.637085427135678e-05,
-      "loss": 4.9367,
-      "step": 14065
-    },
-    {
-      "epoch": 7.335593220338983,
-      "grad_norm": 1.4802595376968384,
-      "learning_rate": 8.636984924623116e-05,
-      "loss": 5.4442,
-      "step": 14066
-    },
-    {
-      "epoch": 7.336114732724902,
-      "grad_norm": 1.548940658569336,
-      "learning_rate": 8.636884422110552e-05,
-      "loss": 5.2126,
-      "step": 14067
-    },
-    {
-      "epoch": 7.336636245110822,
-      "grad_norm": 1.4102771282196045,
-      "learning_rate": 8.63678391959799e-05,
-      "loss": 5.2937,
-      "step": 14068
-    },
-    {
-      "epoch": 7.33715775749674,
-      "grad_norm": 1.4836252927780151,
-      "learning_rate": 8.636683417085428e-05,
-      "loss": 5.4667,
-      "step": 14069
-    },
-    {
-      "epoch": 7.33767926988266,
-      "grad_norm": 1.7689772844314575,
-      "learning_rate": 8.636582914572866e-05,
-      "loss": 4.8048,
-      "step": 14070
-    },
-    {
-      "epoch": 7.338200782268579,
-      "grad_norm": 1.3359594345092773,
-      "learning_rate": 8.636482412060302e-05,
-      "loss": 5.8243,
-      "step": 14071
-    },
-    {
-      "epoch": 7.338722294654498,
-      "grad_norm": 1.3384689092636108,
-      "learning_rate": 8.63638190954774e-05,
-      "loss": 5.7357,
-      "step": 14072
-    },
-    {
-      "epoch": 7.339243807040417,
-      "grad_norm": 1.5541619062423706,
-      "learning_rate": 8.636281407035176e-05,
-      "loss": 5.0173,
-      "step": 14073
-    },
-    {
-      "epoch": 7.339765319426336,
-      "grad_norm": 1.5428541898727417,
-      "learning_rate": 8.636180904522614e-05,
-      "loss": 5.0967,
-      "step": 14074
-    },
-    {
-      "epoch": 7.340286831812255,
-      "grad_norm": 1.3656858205795288,
-      "learning_rate": 8.63608040201005e-05,
-      "loss": 5.8534,
-      "step": 14075
-    },
-    {
-      "epoch": 7.340808344198175,
-      "grad_norm": 1.4813761711120605,
-      "learning_rate": 8.635979899497488e-05,
-      "loss": 5.2809,
-      "step": 14076
-    },
-    {
-      "epoch": 7.341329856584094,
-      "grad_norm": 1.468874454498291,
-      "learning_rate": 8.635879396984925e-05,
-      "loss": 5.1922,
-      "step": 14077
-    },
-    {
-      "epoch": 7.341851368970013,
-      "grad_norm": 1.5164567232131958,
-      "learning_rate": 8.635778894472362e-05,
-      "loss": 5.3245,
-      "step": 14078
-    },
-    {
-      "epoch": 7.342372881355932,
-      "grad_norm": 1.3253073692321777,
-      "learning_rate": 8.635678391959799e-05,
-      "loss": 5.5998,
-      "step": 14079
-    },
-    {
-      "epoch": 7.342894393741851,
-      "grad_norm": 1.5097453594207764,
-      "learning_rate": 8.635577889447237e-05,
-      "loss": 4.9828,
-      "step": 14080
-    },
-    {
-      "epoch": 7.34341590612777,
-      "grad_norm": 1.4521331787109375,
-      "learning_rate": 8.635477386934674e-05,
-      "loss": 5.0385,
-      "step": 14081
-    },
-    {
-      "epoch": 7.34393741851369,
-      "grad_norm": 1.426523208618164,
-      "learning_rate": 8.635376884422111e-05,
-      "loss": 5.4786,
-      "step": 14082
-    },
-    {
-      "epoch": 7.344458930899609,
-      "grad_norm": 1.3585010766983032,
-      "learning_rate": 8.635276381909549e-05,
-      "loss": 5.3661,
-      "step": 14083
-    },
-    {
-      "epoch": 7.344980443285528,
-      "grad_norm": 1.3712060451507568,
-      "learning_rate": 8.635175879396985e-05,
-      "loss": 5.5181,
-      "step": 14084
-    },
-    {
-      "epoch": 7.345501955671447,
-      "grad_norm": 1.387413501739502,
-      "learning_rate": 8.635075376884423e-05,
-      "loss": 5.4512,
-      "step": 14085
-    },
-    {
-      "epoch": 7.346023468057366,
-      "grad_norm": 1.4190168380737305,
-      "learning_rate": 8.634974874371859e-05,
-      "loss": 5.65,
-      "step": 14086
-    },
-    {
-      "epoch": 7.346544980443285,
-      "grad_norm": 1.4257169961929321,
-      "learning_rate": 8.634874371859297e-05,
-      "loss": 5.2987,
-      "step": 14087
-    },
-    {
-      "epoch": 7.347066492829205,
-      "grad_norm": 1.4336555004119873,
-      "learning_rate": 8.634773869346733e-05,
-      "loss": 5.6053,
-      "step": 14088
-    },
-    {
-      "epoch": 7.3475880052151235,
-      "grad_norm": 1.525833010673523,
-      "learning_rate": 8.634673366834171e-05,
-      "loss": 5.0805,
-      "step": 14089
-    },
-    {
-      "epoch": 7.348109517601043,
-      "grad_norm": 1.6121132373809814,
-      "learning_rate": 8.634572864321609e-05,
-      "loss": 5.6931,
-      "step": 14090
-    },
-    {
-      "epoch": 7.348631029986962,
-      "grad_norm": 1.5063132047653198,
-      "learning_rate": 8.634472361809047e-05,
-      "loss": 5.856,
-      "step": 14091
-    },
-    {
-      "epoch": 7.349152542372881,
-      "grad_norm": 1.3940805196762085,
-      "learning_rate": 8.634371859296483e-05,
-      "loss": 5.6085,
-      "step": 14092
-    },
-    {
-      "epoch": 7.3496740547588,
-      "grad_norm": 1.4855287075042725,
-      "learning_rate": 8.63427135678392e-05,
-      "loss": 5.5625,
-      "step": 14093
-    },
-    {
-      "epoch": 7.35019556714472,
-      "grad_norm": 1.4418648481369019,
-      "learning_rate": 8.634170854271357e-05,
-      "loss": 6.0882,
-      "step": 14094
-    },
-    {
-      "epoch": 7.3507170795306385,
-      "grad_norm": 1.4847739934921265,
-      "learning_rate": 8.634070351758794e-05,
-      "loss": 5.6298,
-      "step": 14095
-    },
-    {
-      "epoch": 7.351238591916558,
-      "grad_norm": 1.4054909944534302,
-      "learning_rate": 8.633969849246232e-05,
-      "loss": 5.494,
-      "step": 14096
-    },
-    {
-      "epoch": 7.351760104302477,
-      "grad_norm": 1.4102665185928345,
-      "learning_rate": 8.633869346733668e-05,
-      "loss": 5.842,
-      "step": 14097
-    },
-    {
-      "epoch": 7.352281616688396,
-      "grad_norm": 1.5243723392486572,
-      "learning_rate": 8.633768844221106e-05,
-      "loss": 5.2948,
-      "step": 14098
-    },
-    {
-      "epoch": 7.352803129074315,
-      "grad_norm": 1.3316948413848877,
-      "learning_rate": 8.633668341708542e-05,
-      "loss": 5.5262,
-      "step": 14099
-    },
-    {
-      "epoch": 7.353324641460235,
-      "grad_norm": 1.4595082998275757,
-      "learning_rate": 8.63356783919598e-05,
-      "loss": 5.8955,
-      "step": 14100
-    },
-    {
-      "epoch": 7.3538461538461535,
-      "grad_norm": 1.4924750328063965,
-      "learning_rate": 8.633467336683418e-05,
-      "loss": 5.5194,
-      "step": 14101
-    },
-    {
-      "epoch": 7.354367666232073,
-      "grad_norm": 1.5112718343734741,
-      "learning_rate": 8.633366834170856e-05,
-      "loss": 5.3683,
-      "step": 14102
-    },
-    {
-      "epoch": 7.354889178617992,
-      "grad_norm": 1.3881654739379883,
-      "learning_rate": 8.633266331658292e-05,
-      "loss": 5.3937,
-      "step": 14103
-    },
-    {
-      "epoch": 7.355410691003911,
-      "grad_norm": 1.3978618383407593,
-      "learning_rate": 8.63316582914573e-05,
-      "loss": 5.617,
-      "step": 14104
-    },
-    {
-      "epoch": 7.3559322033898304,
-      "grad_norm": 1.425763726234436,
-      "learning_rate": 8.633065326633166e-05,
-      "loss": 5.7066,
-      "step": 14105
-    },
-    {
-      "epoch": 7.35645371577575,
-      "grad_norm": 1.5361095666885376,
-      "learning_rate": 8.632964824120603e-05,
-      "loss": 5.6143,
-      "step": 14106
-    },
-    {
-      "epoch": 7.3569752281616685,
-      "grad_norm": 1.532286286354065,
-      "learning_rate": 8.63286432160804e-05,
-      "loss": 5.5893,
-      "step": 14107
-    },
-    {
-      "epoch": 7.357496740547588,
-      "grad_norm": 1.708723545074463,
-      "learning_rate": 8.632763819095477e-05,
-      "loss": 4.7609,
-      "step": 14108
-    },
-    {
-      "epoch": 7.358018252933507,
-      "grad_norm": 1.4268895387649536,
-      "learning_rate": 8.632663316582915e-05,
-      "loss": 5.6622,
-      "step": 14109
-    },
-    {
-      "epoch": 7.358539765319426,
-      "grad_norm": 1.3774293661117554,
-      "learning_rate": 8.632562814070353e-05,
-      "loss": 5.8767,
-      "step": 14110
-    },
-    {
-      "epoch": 7.3590612777053455,
-      "grad_norm": 1.5874992609024048,
-      "learning_rate": 8.63246231155779e-05,
-      "loss": 5.2593,
-      "step": 14111
-    },
-    {
-      "epoch": 7.359582790091265,
-      "grad_norm": 1.436953067779541,
-      "learning_rate": 8.632361809045227e-05,
-      "loss": 5.6697,
-      "step": 14112
-    },
-    {
-      "epoch": 7.3601043024771835,
-      "grad_norm": 1.5072623491287231,
-      "learning_rate": 8.632261306532664e-05,
-      "loss": 5.527,
-      "step": 14113
-    },
-    {
-      "epoch": 7.360625814863103,
-      "grad_norm": 1.3641146421432495,
-      "learning_rate": 8.632160804020101e-05,
-      "loss": 5.9778,
-      "step": 14114
-    },
-    {
-      "epoch": 7.361147327249022,
-      "grad_norm": 1.4975948333740234,
-      "learning_rate": 8.632060301507539e-05,
-      "loss": 5.4439,
-      "step": 14115
-    },
-    {
-      "epoch": 7.361668839634941,
-      "grad_norm": 1.3952375650405884,
-      "learning_rate": 8.631959798994975e-05,
-      "loss": 5.9967,
-      "step": 14116
-    },
-    {
-      "epoch": 7.3621903520208605,
-      "grad_norm": 1.4663046598434448,
-      "learning_rate": 8.631859296482413e-05,
-      "loss": 5.404,
-      "step": 14117
-    },
-    {
-      "epoch": 7.36271186440678,
-      "grad_norm": 1.351913332939148,
-      "learning_rate": 8.63175879396985e-05,
-      "loss": 6.0953,
-      "step": 14118
-    },
-    {
-      "epoch": 7.3632333767926985,
-      "grad_norm": 1.3791661262512207,
-      "learning_rate": 8.631658291457286e-05,
-      "loss": 5.8144,
-      "step": 14119
-    },
-    {
-      "epoch": 7.363754889178618,
-      "grad_norm": 1.4670295715332031,
-      "learning_rate": 8.631557788944724e-05,
-      "loss": 5.8817,
-      "step": 14120
-    },
-    {
-      "epoch": 7.3642764015645374,
-      "grad_norm": 1.4128338098526,
-      "learning_rate": 8.631457286432161e-05,
-      "loss": 5.7477,
-      "step": 14121
-    },
-    {
-      "epoch": 7.364797913950456,
-      "grad_norm": 1.478087306022644,
-      "learning_rate": 8.631356783919599e-05,
-      "loss": 5.4937,
-      "step": 14122
-    },
-    {
-      "epoch": 7.3653194263363755,
-      "grad_norm": 1.5136597156524658,
-      "learning_rate": 8.631256281407036e-05,
-      "loss": 5.5861,
-      "step": 14123
-    },
-    {
-      "epoch": 7.365840938722295,
-      "grad_norm": 1.5503915548324585,
-      "learning_rate": 8.631155778894473e-05,
-      "loss": 5.8446,
-      "step": 14124
-    },
-    {
-      "epoch": 7.3663624511082135,
-      "grad_norm": 1.4703823328018188,
-      "learning_rate": 8.63105527638191e-05,
-      "loss": 5.9279,
-      "step": 14125
-    },
-    {
-      "epoch": 7.366883963494133,
-      "grad_norm": 1.454862117767334,
-      "learning_rate": 8.630954773869348e-05,
-      "loss": 5.1826,
-      "step": 14126
-    },
-    {
-      "epoch": 7.3674054758800525,
-      "grad_norm": 1.3330039978027344,
-      "learning_rate": 8.630854271356784e-05,
-      "loss": 5.2879,
-      "step": 14127
-    },
-    {
-      "epoch": 7.367926988265971,
-      "grad_norm": 1.427688479423523,
-      "learning_rate": 8.630753768844222e-05,
-      "loss": 5.3737,
-      "step": 14128
-    },
-    {
-      "epoch": 7.3684485006518905,
-      "grad_norm": 1.67949640750885,
-      "learning_rate": 8.630653266331658e-05,
-      "loss": 4.4529,
-      "step": 14129
-    },
-    {
-      "epoch": 7.36897001303781,
-      "grad_norm": 1.4954167604446411,
-      "learning_rate": 8.630552763819096e-05,
-      "loss": 5.6532,
-      "step": 14130
-    },
-    {
-      "epoch": 7.3694915254237285,
-      "grad_norm": 1.5530565977096558,
-      "learning_rate": 8.630452261306532e-05,
-      "loss": 5.8876,
-      "step": 14131
-    },
-    {
-      "epoch": 7.370013037809648,
-      "grad_norm": 1.4944267272949219,
-      "learning_rate": 8.63035175879397e-05,
-      "loss": 5.806,
-      "step": 14132
-    },
-    {
-      "epoch": 7.3705345501955675,
-      "grad_norm": 1.4697961807250977,
-      "learning_rate": 8.630251256281408e-05,
-      "loss": 5.4654,
-      "step": 14133
-    },
-    {
-      "epoch": 7.371056062581486,
-      "grad_norm": 1.4897717237472534,
-      "learning_rate": 8.630150753768844e-05,
-      "loss": 5.5045,
-      "step": 14134
-    },
-    {
-      "epoch": 7.3715775749674055,
-      "grad_norm": 1.4835759401321411,
-      "learning_rate": 8.630050251256282e-05,
-      "loss": 5.4403,
-      "step": 14135
-    },
-    {
-      "epoch": 7.372099087353325,
-      "grad_norm": 1.3518813848495483,
-      "learning_rate": 8.629949748743719e-05,
-      "loss": 5.8997,
-      "step": 14136
-    },
-    {
-      "epoch": 7.3726205997392436,
-      "grad_norm": 2.0226664543151855,
-      "learning_rate": 8.629849246231156e-05,
-      "loss": 4.6146,
-      "step": 14137
-    },
-    {
-      "epoch": 7.373142112125163,
-      "grad_norm": 1.4120374917984009,
-      "learning_rate": 8.629748743718593e-05,
-      "loss": 5.9279,
-      "step": 14138
-    },
-    {
-      "epoch": 7.3736636245110825,
-      "grad_norm": 1.7325031757354736,
-      "learning_rate": 8.62964824120603e-05,
-      "loss": 5.1797,
-      "step": 14139
-    },
-    {
-      "epoch": 7.374185136897001,
-      "grad_norm": 1.496110200881958,
-      "learning_rate": 8.629547738693467e-05,
-      "loss": 5.7291,
-      "step": 14140
-    },
-    {
-      "epoch": 7.3747066492829205,
-      "grad_norm": 1.2764310836791992,
-      "learning_rate": 8.629447236180905e-05,
-      "loss": 6.0163,
-      "step": 14141
-    },
-    {
-      "epoch": 7.37522816166884,
-      "grad_norm": 1.3636783361434937,
-      "learning_rate": 8.629346733668343e-05,
-      "loss": 5.7685,
-      "step": 14142
-    },
-    {
-      "epoch": 7.375749674054759,
-      "grad_norm": 1.4744255542755127,
-      "learning_rate": 8.62924623115578e-05,
-      "loss": 5.4804,
-      "step": 14143
-    },
-    {
-      "epoch": 7.376271186440678,
-      "grad_norm": 1.4802439212799072,
-      "learning_rate": 8.629145728643217e-05,
-      "loss": 5.6106,
-      "step": 14144
-    },
-    {
-      "epoch": 7.3767926988265975,
-      "grad_norm": 1.6065635681152344,
-      "learning_rate": 8.629045226130653e-05,
-      "loss": 5.3015,
-      "step": 14145
-    },
-    {
-      "epoch": 7.377314211212516,
-      "grad_norm": 1.4100929498672485,
-      "learning_rate": 8.628944723618091e-05,
-      "loss": 5.8161,
-      "step": 14146
-    },
-    {
-      "epoch": 7.3778357235984355,
-      "grad_norm": 1.4408624172210693,
-      "learning_rate": 8.628844221105527e-05,
-      "loss": 5.5822,
-      "step": 14147
-    },
-    {
-      "epoch": 7.378357235984355,
-      "grad_norm": 1.3833177089691162,
-      "learning_rate": 8.628743718592965e-05,
-      "loss": 5.9259,
-      "step": 14148
-    },
-    {
-      "epoch": 7.378878748370274,
-      "grad_norm": 1.8256405591964722,
-      "learning_rate": 8.628643216080402e-05,
-      "loss": 4.9419,
-      "step": 14149
-    },
-    {
-      "epoch": 7.379400260756193,
-      "grad_norm": 1.5105611085891724,
-      "learning_rate": 8.62854271356784e-05,
-      "loss": 5.689,
-      "step": 14150
-    },
-    {
-      "epoch": 7.3799217731421125,
-      "grad_norm": 1.714981198310852,
-      "learning_rate": 8.628442211055276e-05,
-      "loss": 4.9384,
-      "step": 14151
-    },
-    {
-      "epoch": 7.380443285528031,
-      "grad_norm": 1.4923653602600098,
-      "learning_rate": 8.628341708542714e-05,
-      "loss": 5.4446,
-      "step": 14152
-    },
-    {
-      "epoch": 7.3809647979139505,
-      "grad_norm": 1.4777165651321411,
-      "learning_rate": 8.628241206030151e-05,
-      "loss": 5.914,
-      "step": 14153
-    },
-    {
-      "epoch": 7.38148631029987,
-      "grad_norm": 1.5271995067596436,
-      "learning_rate": 8.628140703517589e-05,
-      "loss": 5.3885,
-      "step": 14154
-    },
-    {
-      "epoch": 7.382007822685789,
-      "grad_norm": 1.8909293413162231,
-      "learning_rate": 8.628040201005026e-05,
-      "loss": 4.7539,
-      "step": 14155
-    },
-    {
-      "epoch": 7.382529335071708,
-      "grad_norm": 1.5430923700332642,
-      "learning_rate": 8.627939698492463e-05,
-      "loss": 4.9579,
-      "step": 14156
-    },
-    {
-      "epoch": 7.3830508474576275,
-      "grad_norm": 1.4381492137908936,
-      "learning_rate": 8.6278391959799e-05,
-      "loss": 5.648,
-      "step": 14157
-    },
-    {
-      "epoch": 7.383572359843546,
-      "grad_norm": 1.5892207622528076,
-      "learning_rate": 8.627738693467336e-05,
-      "loss": 5.3703,
-      "step": 14158
-    },
-    {
-      "epoch": 7.384093872229466,
-      "grad_norm": 1.285823106765747,
-      "learning_rate": 8.627638190954774e-05,
-      "loss": 5.3019,
-      "step": 14159
-    },
-    {
-      "epoch": 7.384615384615385,
-      "grad_norm": 1.4033992290496826,
-      "learning_rate": 8.62753768844221e-05,
-      "loss": 5.8053,
-      "step": 14160
-    },
-    {
-      "epoch": 7.385136897001304,
-      "grad_norm": 1.4279005527496338,
-      "learning_rate": 8.627437185929648e-05,
-      "loss": 5.8314,
-      "step": 14161
-    },
-    {
-      "epoch": 7.385658409387223,
-      "grad_norm": 1.3596521615982056,
-      "learning_rate": 8.627336683417086e-05,
-      "loss": 5.8954,
-      "step": 14162
-    },
-    {
-      "epoch": 7.386179921773142,
-      "grad_norm": 1.4434218406677246,
-      "learning_rate": 8.627236180904524e-05,
-      "loss": 5.2857,
-      "step": 14163
-    },
-    {
-      "epoch": 7.386701434159061,
-      "grad_norm": 1.4004137516021729,
-      "learning_rate": 8.62713567839196e-05,
-      "loss": 5.6606,
-      "step": 14164
-    },
-    {
-      "epoch": 7.387222946544981,
-      "grad_norm": 1.4420850276947021,
-      "learning_rate": 8.627035175879398e-05,
-      "loss": 5.6736,
-      "step": 14165
-    },
-    {
-      "epoch": 7.3877444589309,
-      "grad_norm": 1.382262110710144,
-      "learning_rate": 8.626934673366834e-05,
-      "loss": 5.5453,
-      "step": 14166
-    },
-    {
-      "epoch": 7.388265971316819,
-      "grad_norm": 1.3904736042022705,
-      "learning_rate": 8.626834170854272e-05,
-      "loss": 5.9901,
-      "step": 14167
-    },
-    {
-      "epoch": 7.388787483702738,
-      "grad_norm": 1.4048441648483276,
-      "learning_rate": 8.626733668341709e-05,
-      "loss": 5.4459,
-      "step": 14168
-    },
-    {
-      "epoch": 7.389308996088657,
-      "grad_norm": 1.4792418479919434,
-      "learning_rate": 8.626633165829146e-05,
-      "loss": 5.8025,
-      "step": 14169
-    },
-    {
-      "epoch": 7.389830508474576,
-      "grad_norm": 1.386809229850769,
-      "learning_rate": 8.626532663316583e-05,
-      "loss": 5.4546,
-      "step": 14170
-    },
-    {
-      "epoch": 7.390352020860496,
-      "grad_norm": 1.4145687818527222,
-      "learning_rate": 8.62643216080402e-05,
-      "loss": 5.4652,
-      "step": 14171
-    },
-    {
-      "epoch": 7.390873533246415,
-      "grad_norm": 1.4539874792099,
-      "learning_rate": 8.626331658291457e-05,
-      "loss": 5.6454,
-      "step": 14172
-    },
-    {
-      "epoch": 7.391395045632334,
-      "grad_norm": 1.400189995765686,
-      "learning_rate": 8.626231155778895e-05,
-      "loss": 5.6029,
-      "step": 14173
-    },
-    {
-      "epoch": 7.391916558018253,
-      "grad_norm": 1.529093861579895,
-      "learning_rate": 8.626130653266333e-05,
-      "loss": 5.8987,
-      "step": 14174
-    },
-    {
-      "epoch": 7.392438070404172,
-      "grad_norm": 1.6764836311340332,
-      "learning_rate": 8.626030150753769e-05,
-      "loss": 5.0425,
-      "step": 14175
-    },
-    {
-      "epoch": 7.392959582790091,
-      "grad_norm": 1.3173460960388184,
-      "learning_rate": 8.625929648241207e-05,
-      "loss": 5.6325,
-      "step": 14176
-    },
-    {
-      "epoch": 7.393481095176011,
-      "grad_norm": 1.4846165180206299,
-      "learning_rate": 8.625829145728643e-05,
-      "loss": 5.4129,
-      "step": 14177
-    },
-    {
-      "epoch": 7.39400260756193,
-      "grad_norm": 1.4071072340011597,
-      "learning_rate": 8.625728643216081e-05,
-      "loss": 5.5546,
-      "step": 14178
-    },
-    {
-      "epoch": 7.394524119947849,
-      "grad_norm": 1.435632348060608,
-      "learning_rate": 8.625628140703518e-05,
-      "loss": 5.7503,
-      "step": 14179
-    },
-    {
-      "epoch": 7.395045632333768,
-      "grad_norm": 1.408699631690979,
-      "learning_rate": 8.625527638190955e-05,
-      "loss": 5.6445,
-      "step": 14180
-    },
-    {
-      "epoch": 7.395567144719687,
-      "grad_norm": 1.5238648653030396,
-      "learning_rate": 8.625427135678392e-05,
-      "loss": 4.9994,
-      "step": 14181
-    },
-    {
-      "epoch": 7.396088657105606,
-      "grad_norm": 1.5956357717514038,
-      "learning_rate": 8.62532663316583e-05,
-      "loss": 5.4867,
-      "step": 14182
-    },
-    {
-      "epoch": 7.396610169491526,
-      "grad_norm": 1.3702948093414307,
-      "learning_rate": 8.625226130653267e-05,
-      "loss": 5.6506,
-      "step": 14183
-    },
-    {
-      "epoch": 7.397131681877444,
-      "grad_norm": 1.5504330396652222,
-      "learning_rate": 8.625125628140705e-05,
-      "loss": 5.5675,
-      "step": 14184
-    },
-    {
-      "epoch": 7.397653194263364,
-      "grad_norm": 1.352691888809204,
-      "learning_rate": 8.625025125628141e-05,
-      "loss": 5.9027,
-      "step": 14185
-    },
-    {
-      "epoch": 7.398174706649283,
-      "grad_norm": 1.3896793127059937,
-      "learning_rate": 8.624924623115578e-05,
-      "loss": 5.4436,
-      "step": 14186
-    },
-    {
-      "epoch": 7.398696219035202,
-      "grad_norm": 1.3449150323867798,
-      "learning_rate": 8.624824120603016e-05,
-      "loss": 5.4449,
-      "step": 14187
-    },
-    {
-      "epoch": 7.399217731421121,
-      "grad_norm": 1.524226188659668,
-      "learning_rate": 8.624723618090452e-05,
-      "loss": 4.974,
-      "step": 14188
-    },
-    {
-      "epoch": 7.399739243807041,
-      "grad_norm": 1.511344313621521,
-      "learning_rate": 8.62462311557789e-05,
-      "loss": 5.4156,
-      "step": 14189
-    },
-    {
-      "epoch": 7.400260756192959,
-      "grad_norm": 1.4743949174880981,
-      "learning_rate": 8.624522613065326e-05,
-      "loss": 5.1156,
-      "step": 14190
-    },
-    {
-      "epoch": 7.400782268578879,
-      "grad_norm": 1.6428529024124146,
-      "learning_rate": 8.624422110552764e-05,
-      "loss": 5.4457,
-      "step": 14191
-    },
-    {
-      "epoch": 7.401303780964798,
-      "grad_norm": 1.577919840812683,
-      "learning_rate": 8.6243216080402e-05,
-      "loss": 5.1141,
-      "step": 14192
-    },
-    {
-      "epoch": 7.401825293350717,
-      "grad_norm": 1.4119775295257568,
-      "learning_rate": 8.624221105527638e-05,
-      "loss": 5.5281,
-      "step": 14193
-    },
-    {
-      "epoch": 7.402346805736636,
-      "grad_norm": 1.4805152416229248,
-      "learning_rate": 8.624120603015076e-05,
-      "loss": 5.753,
-      "step": 14194
-    },
-    {
-      "epoch": 7.402868318122556,
-      "grad_norm": 1.2799268960952759,
-      "learning_rate": 8.624020100502514e-05,
-      "loss": 5.2896,
-      "step": 14195
-    },
-    {
-      "epoch": 7.403389830508474,
-      "grad_norm": 1.3589462041854858,
-      "learning_rate": 8.62391959798995e-05,
-      "loss": 5.8541,
-      "step": 14196
-    },
-    {
-      "epoch": 7.403911342894394,
-      "grad_norm": 1.2868988513946533,
-      "learning_rate": 8.623819095477388e-05,
-      "loss": 5.6222,
-      "step": 14197
-    },
-    {
-      "epoch": 7.404432855280313,
-      "grad_norm": 1.3450533151626587,
-      "learning_rate": 8.623718592964825e-05,
-      "loss": 5.5451,
-      "step": 14198
-    },
-    {
-      "epoch": 7.404954367666232,
-      "grad_norm": 1.5555152893066406,
-      "learning_rate": 8.623618090452261e-05,
-      "loss": 5.0558,
-      "step": 14199
-    },
-    {
-      "epoch": 7.405475880052151,
-      "grad_norm": 1.4277386665344238,
-      "learning_rate": 8.623517587939699e-05,
-      "loss": 5.4719,
-      "step": 14200
-    },
-    {
-      "epoch": 7.405997392438071,
-      "grad_norm": 1.464926838874817,
-      "learning_rate": 8.623417085427135e-05,
-      "loss": 5.4429,
-      "step": 14201
-    },
-    {
-      "epoch": 7.406518904823989,
-      "grad_norm": 1.5777472257614136,
-      "learning_rate": 8.623316582914573e-05,
-      "loss": 5.0734,
-      "step": 14202
-    },
-    {
-      "epoch": 7.407040417209909,
-      "grad_norm": 1.4172298908233643,
-      "learning_rate": 8.623216080402011e-05,
-      "loss": 5.7,
-      "step": 14203
-    },
-    {
-      "epoch": 7.407561929595828,
-      "grad_norm": 1.48367178440094,
-      "learning_rate": 8.623115577889449e-05,
-      "loss": 5.3557,
-      "step": 14204
-    },
-    {
-      "epoch": 7.408083441981747,
-      "grad_norm": 1.4576481580734253,
-      "learning_rate": 8.623015075376885e-05,
-      "loss": 5.5282,
-      "step": 14205
-    },
-    {
-      "epoch": 7.408604954367666,
-      "grad_norm": 1.548510193824768,
-      "learning_rate": 8.622914572864323e-05,
-      "loss": 5.2592,
-      "step": 14206
-    },
-    {
-      "epoch": 7.409126466753586,
-      "grad_norm": 1.6012077331542969,
-      "learning_rate": 8.622814070351759e-05,
-      "loss": 5.7055,
-      "step": 14207
-    },
-    {
-      "epoch": 7.409647979139504,
-      "grad_norm": 1.425754189491272,
-      "learning_rate": 8.622713567839197e-05,
-      "loss": 6.0353,
-      "step": 14208
-    },
-    {
-      "epoch": 7.410169491525424,
-      "grad_norm": 1.463584303855896,
-      "learning_rate": 8.622613065326633e-05,
-      "loss": 4.9069,
-      "step": 14209
-    },
-    {
-      "epoch": 7.410691003911343,
-      "grad_norm": 1.405301570892334,
-      "learning_rate": 8.622512562814071e-05,
-      "loss": 5.4763,
-      "step": 14210
-    },
-    {
-      "epoch": 7.411212516297262,
-      "grad_norm": 1.4896106719970703,
-      "learning_rate": 8.622412060301508e-05,
-      "loss": 5.2585,
-      "step": 14211
-    },
-    {
-      "epoch": 7.411734028683181,
-      "grad_norm": 1.3785629272460938,
-      "learning_rate": 8.622311557788944e-05,
-      "loss": 5.569,
-      "step": 14212
-    },
-    {
-      "epoch": 7.412255541069101,
-      "grad_norm": 1.3498270511627197,
-      "learning_rate": 8.622211055276382e-05,
-      "loss": 5.922,
-      "step": 14213
-    },
-    {
-      "epoch": 7.412777053455019,
-      "grad_norm": 1.3219002485275269,
-      "learning_rate": 8.62211055276382e-05,
-      "loss": 5.6531,
-      "step": 14214
-    },
-    {
-      "epoch": 7.413298565840939,
-      "grad_norm": 1.4623290300369263,
-      "learning_rate": 8.622010050251257e-05,
-      "loss": 5.4577,
-      "step": 14215
-    },
-    {
-      "epoch": 7.413820078226858,
-      "grad_norm": 1.515224814414978,
-      "learning_rate": 8.621909547738694e-05,
-      "loss": 5.167,
-      "step": 14216
-    },
-    {
-      "epoch": 7.414341590612777,
-      "grad_norm": 1.4089800119400024,
-      "learning_rate": 8.621809045226132e-05,
-      "loss": 5.3811,
-      "step": 14217
-    },
-    {
-      "epoch": 7.414863102998696,
-      "grad_norm": 1.4516761302947998,
-      "learning_rate": 8.621708542713568e-05,
-      "loss": 5.5102,
-      "step": 14218
-    },
-    {
-      "epoch": 7.415384615384616,
-      "grad_norm": 1.4150274991989136,
-      "learning_rate": 8.621608040201006e-05,
-      "loss": 5.2768,
-      "step": 14219
-    },
-    {
-      "epoch": 7.415906127770534,
-      "grad_norm": 1.3778207302093506,
-      "learning_rate": 8.621507537688442e-05,
-      "loss": 5.6867,
-      "step": 14220
-    },
-    {
-      "epoch": 7.416427640156454,
-      "grad_norm": 1.5403506755828857,
-      "learning_rate": 8.62140703517588e-05,
-      "loss": 5.6693,
-      "step": 14221
-    },
-    {
-      "epoch": 7.416949152542373,
-      "grad_norm": 1.5226300954818726,
-      "learning_rate": 8.621306532663316e-05,
-      "loss": 5.4529,
-      "step": 14222
-    },
-    {
-      "epoch": 7.417470664928292,
-      "grad_norm": 1.405263900756836,
-      "learning_rate": 8.621206030150754e-05,
-      "loss": 5.4865,
-      "step": 14223
-    },
-    {
-      "epoch": 7.417992177314211,
-      "grad_norm": 1.376161813735962,
-      "learning_rate": 8.621105527638192e-05,
-      "loss": 5.8112,
-      "step": 14224
-    },
-    {
-      "epoch": 7.418513689700131,
-      "grad_norm": 1.4538315534591675,
-      "learning_rate": 8.621005025125628e-05,
-      "loss": 5.9042,
-      "step": 14225
-    },
-    {
-      "epoch": 7.419035202086049,
-      "grad_norm": 1.4149014949798584,
-      "learning_rate": 8.620904522613066e-05,
-      "loss": 5.8692,
-      "step": 14226
-    },
-    {
-      "epoch": 7.419556714471969,
-      "grad_norm": 1.4087064266204834,
-      "learning_rate": 8.620804020100503e-05,
-      "loss": 5.8994,
-      "step": 14227
-    },
-    {
-      "epoch": 7.420078226857888,
-      "grad_norm": 1.4016448259353638,
-      "learning_rate": 8.62070351758794e-05,
-      "loss": 5.622,
-      "step": 14228
-    },
-    {
-      "epoch": 7.420599739243807,
-      "grad_norm": 1.2972010374069214,
-      "learning_rate": 8.620603015075377e-05,
-      "loss": 5.7947,
-      "step": 14229
-    },
-    {
-      "epoch": 7.421121251629726,
-      "grad_norm": 1.474238634109497,
-      "learning_rate": 8.620502512562815e-05,
-      "loss": 5.0965,
-      "step": 14230
-    },
-    {
-      "epoch": 7.421642764015646,
-      "grad_norm": 1.372192144393921,
-      "learning_rate": 8.620402010050251e-05,
-      "loss": 5.6062,
-      "step": 14231
-    },
-    {
-      "epoch": 7.422164276401564,
-      "grad_norm": 1.4769084453582764,
-      "learning_rate": 8.620301507537689e-05,
-      "loss": 5.4419,
-      "step": 14232
-    },
-    {
-      "epoch": 7.422685788787484,
-      "grad_norm": 1.513552188873291,
-      "learning_rate": 8.620201005025125e-05,
-      "loss": 5.0112,
-      "step": 14233
-    },
-    {
-      "epoch": 7.423207301173403,
-      "grad_norm": 1.4308239221572876,
-      "learning_rate": 8.620100502512563e-05,
-      "loss": 5.4621,
-      "step": 14234
-    },
-    {
-      "epoch": 7.423728813559322,
-      "grad_norm": 1.5503261089324951,
-      "learning_rate": 8.620000000000001e-05,
-      "loss": 5.3774,
-      "step": 14235
-    },
-    {
-      "epoch": 7.424250325945241,
-      "grad_norm": 1.4259485006332397,
-      "learning_rate": 8.619899497487439e-05,
-      "loss": 5.779,
-      "step": 14236
-    },
-    {
-      "epoch": 7.424771838331161,
-      "grad_norm": 1.3928489685058594,
-      "learning_rate": 8.619798994974875e-05,
-      "loss": 5.4128,
-      "step": 14237
-    },
-    {
-      "epoch": 7.425293350717079,
-      "grad_norm": 1.4417411088943481,
-      "learning_rate": 8.619698492462311e-05,
-      "loss": 5.6561,
-      "step": 14238
-    },
-    {
-      "epoch": 7.425814863102999,
-      "grad_norm": 1.4377459287643433,
-      "learning_rate": 8.619597989949749e-05,
-      "loss": 5.7665,
-      "step": 14239
-    },
-    {
-      "epoch": 7.426336375488918,
-      "grad_norm": 1.379299521446228,
-      "learning_rate": 8.619497487437186e-05,
-      "loss": 5.7633,
-      "step": 14240
-    },
-    {
-      "epoch": 7.426857887874837,
-      "grad_norm": 1.4549612998962402,
-      "learning_rate": 8.619396984924623e-05,
-      "loss": 5.7221,
-      "step": 14241
-    },
-    {
-      "epoch": 7.427379400260756,
-      "grad_norm": 1.4235965013504028,
-      "learning_rate": 8.61929648241206e-05,
-      "loss": 5.6509,
-      "step": 14242
-    },
-    {
-      "epoch": 7.427900912646676,
-      "grad_norm": 1.4879059791564941,
-      "learning_rate": 8.619195979899498e-05,
-      "loss": 5.3719,
-      "step": 14243
-    },
-    {
-      "epoch": 7.428422425032594,
-      "grad_norm": 1.3479435443878174,
-      "learning_rate": 8.619095477386935e-05,
-      "loss": 5.8037,
-      "step": 14244
-    },
-    {
-      "epoch": 7.428943937418514,
-      "grad_norm": 1.3629789352416992,
-      "learning_rate": 8.618994974874373e-05,
-      "loss": 5.2061,
-      "step": 14245
-    },
-    {
-      "epoch": 7.429465449804433,
-      "grad_norm": 1.4932838678359985,
-      "learning_rate": 8.61889447236181e-05,
-      "loss": 5.6273,
-      "step": 14246
-    },
-    {
-      "epoch": 7.429986962190352,
-      "grad_norm": 1.4447717666625977,
-      "learning_rate": 8.618793969849247e-05,
-      "loss": 5.1216,
-      "step": 14247
-    },
-    {
-      "epoch": 7.430508474576271,
-      "grad_norm": 1.4133471250534058,
-      "learning_rate": 8.618693467336684e-05,
-      "loss": 5.4463,
-      "step": 14248
-    },
-    {
-      "epoch": 7.431029986962191,
-      "grad_norm": 1.4122138023376465,
-      "learning_rate": 8.618592964824122e-05,
-      "loss": 5.7575,
-      "step": 14249
-    },
-    {
-      "epoch": 7.431551499348109,
-      "grad_norm": 1.5188621282577515,
-      "learning_rate": 8.618492462311558e-05,
-      "loss": 5.3654,
-      "step": 14250
-    },
-    {
-      "epoch": 7.432073011734029,
-      "grad_norm": 1.442542314529419,
-      "learning_rate": 8.618391959798996e-05,
-      "loss": 5.8125,
-      "step": 14251
-    },
-    {
-      "epoch": 7.432594524119948,
-      "grad_norm": 1.6895517110824585,
-      "learning_rate": 8.618291457286432e-05,
-      "loss": 4.7752,
-      "step": 14252
-    },
-    {
-      "epoch": 7.433116036505867,
-      "grad_norm": 1.4602105617523193,
-      "learning_rate": 8.618190954773869e-05,
-      "loss": 5.6572,
-      "step": 14253
-    },
-    {
-      "epoch": 7.433637548891786,
-      "grad_norm": 1.510026216506958,
-      "learning_rate": 8.618090452261306e-05,
-      "loss": 5.8225,
-      "step": 14254
-    },
-    {
-      "epoch": 7.434159061277706,
-      "grad_norm": 1.5581762790679932,
-      "learning_rate": 8.617989949748744e-05,
-      "loss": 4.9862,
-      "step": 14255
-    },
-    {
-      "epoch": 7.434680573663624,
-      "grad_norm": 1.412340760231018,
-      "learning_rate": 8.617889447236182e-05,
-      "loss": 5.5487,
-      "step": 14256
-    },
-    {
-      "epoch": 7.435202086049544,
-      "grad_norm": 1.6891721487045288,
-      "learning_rate": 8.617788944723618e-05,
-      "loss": 5.444,
-      "step": 14257
-    },
-    {
-      "epoch": 7.435723598435462,
-      "grad_norm": 1.4973411560058594,
-      "learning_rate": 8.617688442211056e-05,
-      "loss": 5.5478,
-      "step": 14258
-    },
-    {
-      "epoch": 7.436245110821382,
-      "grad_norm": 1.429060697555542,
-      "learning_rate": 8.617587939698493e-05,
-      "loss": 5.7634,
-      "step": 14259
-    },
-    {
-      "epoch": 7.436766623207301,
-      "grad_norm": 1.5158412456512451,
-      "learning_rate": 8.61748743718593e-05,
-      "loss": 5.6012,
-      "step": 14260
-    },
-    {
-      "epoch": 7.437288135593221,
-      "grad_norm": 1.5324784517288208,
-      "learning_rate": 8.617386934673367e-05,
-      "loss": 5.4205,
-      "step": 14261
-    },
-    {
-      "epoch": 7.437809647979139,
-      "grad_norm": 1.5659664869308472,
-      "learning_rate": 8.617286432160805e-05,
-      "loss": 5.3359,
-      "step": 14262
-    },
-    {
-      "epoch": 7.438331160365059,
-      "grad_norm": 1.3550336360931396,
-      "learning_rate": 8.617185929648241e-05,
-      "loss": 5.5797,
-      "step": 14263
-    },
-    {
-      "epoch": 7.438852672750977,
-      "grad_norm": 1.6186634302139282,
-      "learning_rate": 8.617085427135679e-05,
-      "loss": 5.8263,
-      "step": 14264
-    },
-    {
-      "epoch": 7.439374185136897,
-      "grad_norm": 1.4324424266815186,
-      "learning_rate": 8.616984924623117e-05,
-      "loss": 5.6028,
-      "step": 14265
-    },
-    {
-      "epoch": 7.439895697522816,
-      "grad_norm": 1.4691213369369507,
-      "learning_rate": 8.616884422110553e-05,
-      "loss": 5.6158,
-      "step": 14266
-    },
-    {
-      "epoch": 7.440417209908736,
-      "grad_norm": 1.4039791822433472,
-      "learning_rate": 8.616783919597991e-05,
-      "loss": 5.4533,
-      "step": 14267
-    },
-    {
-      "epoch": 7.440938722294654,
-      "grad_norm": 1.518585443496704,
-      "learning_rate": 8.616683417085427e-05,
-      "loss": 5.4842,
-      "step": 14268
-    },
-    {
-      "epoch": 7.441460234680574,
-      "grad_norm": 1.4168916940689087,
-      "learning_rate": 8.616582914572865e-05,
-      "loss": 5.9592,
-      "step": 14269
-    },
-    {
-      "epoch": 7.441981747066492,
-      "grad_norm": 1.4451208114624023,
-      "learning_rate": 8.616482412060302e-05,
-      "loss": 5.6947,
-      "step": 14270
-    },
-    {
-      "epoch": 7.442503259452412,
-      "grad_norm": 1.381415843963623,
-      "learning_rate": 8.616381909547739e-05,
-      "loss": 5.4075,
-      "step": 14271
-    },
-    {
-      "epoch": 7.443024771838331,
-      "grad_norm": 1.3678151369094849,
-      "learning_rate": 8.616281407035176e-05,
-      "loss": 5.5872,
-      "step": 14272
-    },
-    {
-      "epoch": 7.443546284224251,
-      "grad_norm": 1.3448539972305298,
-      "learning_rate": 8.616180904522614e-05,
-      "loss": 5.5019,
-      "step": 14273
-    },
-    {
-      "epoch": 7.444067796610169,
-      "grad_norm": 1.3953989744186401,
-      "learning_rate": 8.61608040201005e-05,
-      "loss": 5.6416,
-      "step": 14274
-    },
-    {
-      "epoch": 7.444589308996089,
-      "grad_norm": 1.479372262954712,
-      "learning_rate": 8.615979899497488e-05,
-      "loss": 5.5388,
-      "step": 14275
-    },
-    {
-      "epoch": 7.445110821382007,
-      "grad_norm": 1.4568132162094116,
-      "learning_rate": 8.615879396984926e-05,
-      "loss": 4.832,
-      "step": 14276
-    },
-    {
-      "epoch": 7.445632333767927,
-      "grad_norm": 1.3524776697158813,
-      "learning_rate": 8.615778894472363e-05,
-      "loss": 5.7165,
-      "step": 14277
-    },
-    {
-      "epoch": 7.446153846153846,
-      "grad_norm": 1.3663841485977173,
-      "learning_rate": 8.6156783919598e-05,
-      "loss": 5.6367,
-      "step": 14278
-    },
-    {
-      "epoch": 7.446675358539765,
-      "grad_norm": 1.4485597610473633,
-      "learning_rate": 8.615577889447236e-05,
-      "loss": 5.753,
-      "step": 14279
-    },
-    {
-      "epoch": 7.447196870925684,
-      "grad_norm": 1.4402482509613037,
-      "learning_rate": 8.615477386934674e-05,
-      "loss": 5.7825,
-      "step": 14280
-    },
-    {
-      "epoch": 7.447718383311604,
-      "grad_norm": 1.3657389879226685,
-      "learning_rate": 8.61537688442211e-05,
-      "loss": 5.356,
-      "step": 14281
-    },
-    {
-      "epoch": 7.448239895697522,
-      "grad_norm": 1.363399863243103,
-      "learning_rate": 8.615276381909548e-05,
-      "loss": 5.6831,
-      "step": 14282
-    },
-    {
-      "epoch": 7.448761408083442,
-      "grad_norm": 1.4248377084732056,
-      "learning_rate": 8.615175879396985e-05,
-      "loss": 5.8573,
-      "step": 14283
-    },
-    {
-      "epoch": 7.449282920469361,
-      "grad_norm": 1.4792426824569702,
-      "learning_rate": 8.615075376884422e-05,
-      "loss": 5.5372,
-      "step": 14284
-    },
-    {
-      "epoch": 7.44980443285528,
-      "grad_norm": 1.4645453691482544,
-      "learning_rate": 8.614974874371859e-05,
-      "loss": 5.0142,
-      "step": 14285
-    },
-    {
-      "epoch": 7.450325945241199,
-      "grad_norm": 1.5381807088851929,
-      "learning_rate": 8.614874371859297e-05,
-      "loss": 4.9206,
-      "step": 14286
-    },
-    {
-      "epoch": 7.450847457627119,
-      "grad_norm": 1.4675301313400269,
-      "learning_rate": 8.614773869346734e-05,
-      "loss": 4.806,
-      "step": 14287
-    },
-    {
-      "epoch": 7.451368970013037,
-      "grad_norm": 1.3951455354690552,
-      "learning_rate": 8.614673366834172e-05,
-      "loss": 5.1818,
-      "step": 14288
-    },
-    {
-      "epoch": 7.451890482398957,
-      "grad_norm": 1.4525424242019653,
-      "learning_rate": 8.614572864321609e-05,
-      "loss": 5.5625,
-      "step": 14289
-    },
-    {
-      "epoch": 7.452411994784876,
-      "grad_norm": 1.3821938037872314,
-      "learning_rate": 8.614472361809046e-05,
-      "loss": 5.8409,
-      "step": 14290
-    },
-    {
-      "epoch": 7.452933507170795,
-      "grad_norm": 1.5408557653427124,
-      "learning_rate": 8.614371859296483e-05,
-      "loss": 5.2333,
-      "step": 14291
-    },
-    {
-      "epoch": 7.453455019556714,
-      "grad_norm": 1.34048330783844,
-      "learning_rate": 8.614271356783919e-05,
-      "loss": 5.6896,
-      "step": 14292
-    },
-    {
-      "epoch": 7.453976531942634,
-      "grad_norm": 1.7631473541259766,
-      "learning_rate": 8.614170854271357e-05,
-      "loss": 5.2045,
-      "step": 14293
-    },
-    {
-      "epoch": 7.4544980443285525,
-      "grad_norm": 1.4007327556610107,
-      "learning_rate": 8.614070351758793e-05,
-      "loss": 5.7203,
-      "step": 14294
-    },
-    {
-      "epoch": 7.455019556714472,
-      "grad_norm": 1.5579476356506348,
-      "learning_rate": 8.613969849246231e-05,
-      "loss": 5.6567,
-      "step": 14295
-    },
-    {
-      "epoch": 7.455541069100391,
-      "grad_norm": 1.3906112909317017,
-      "learning_rate": 8.613869346733669e-05,
-      "loss": 5.431,
-      "step": 14296
-    },
-    {
-      "epoch": 7.45606258148631,
-      "grad_norm": 1.4342093467712402,
-      "learning_rate": 8.613768844221107e-05,
-      "loss": 5.6358,
-      "step": 14297
-    },
-    {
-      "epoch": 7.456584093872229,
-      "grad_norm": 1.5345861911773682,
-      "learning_rate": 8.613668341708543e-05,
-      "loss": 5.4142,
-      "step": 14298
-    },
-    {
-      "epoch": 7.457105606258149,
-      "grad_norm": 1.5842835903167725,
-      "learning_rate": 8.613567839195981e-05,
-      "loss": 5.3757,
-      "step": 14299
-    },
-    {
-      "epoch": 7.4576271186440675,
-      "grad_norm": 1.519406795501709,
-      "learning_rate": 8.613467336683417e-05,
-      "loss": 5.73,
-      "step": 14300
-    },
-    {
-      "epoch": 7.458148631029987,
-      "grad_norm": 1.5290672779083252,
-      "learning_rate": 8.613366834170855e-05,
-      "loss": 5.6529,
-      "step": 14301
-    },
-    {
-      "epoch": 7.458670143415906,
-      "grad_norm": 1.4627504348754883,
-      "learning_rate": 8.613266331658292e-05,
-      "loss": 5.4367,
-      "step": 14302
-    },
-    {
-      "epoch": 7.459191655801825,
-      "grad_norm": 1.431751012802124,
-      "learning_rate": 8.61316582914573e-05,
-      "loss": 5.3066,
-      "step": 14303
-    },
-    {
-      "epoch": 7.459713168187744,
-      "grad_norm": 1.4224778413772583,
-      "learning_rate": 8.613065326633166e-05,
-      "loss": 5.5867,
-      "step": 14304
-    },
-    {
-      "epoch": 7.460234680573664,
-      "grad_norm": 1.473611831665039,
-      "learning_rate": 8.612964824120602e-05,
-      "loss": 5.2475,
-      "step": 14305
-    },
-    {
-      "epoch": 7.4607561929595825,
-      "grad_norm": 1.4644843339920044,
-      "learning_rate": 8.61286432160804e-05,
-      "loss": 5.4721,
-      "step": 14306
-    },
-    {
-      "epoch": 7.461277705345502,
-      "grad_norm": 1.4035083055496216,
-      "learning_rate": 8.612763819095478e-05,
-      "loss": 5.5835,
-      "step": 14307
-    },
-    {
-      "epoch": 7.461799217731421,
-      "grad_norm": 1.5075477361679077,
-      "learning_rate": 8.612663316582916e-05,
-      "loss": 5.3304,
-      "step": 14308
-    },
-    {
-      "epoch": 7.46232073011734,
-      "grad_norm": 1.4983278512954712,
-      "learning_rate": 8.612562814070352e-05,
-      "loss": 5.432,
-      "step": 14309
-    },
-    {
-      "epoch": 7.4628422425032594,
-      "grad_norm": 1.5182899236679077,
-      "learning_rate": 8.61246231155779e-05,
-      "loss": 5.524,
-      "step": 14310
-    },
-    {
-      "epoch": 7.463363754889179,
-      "grad_norm": 1.2855689525604248,
-      "learning_rate": 8.612361809045226e-05,
-      "loss": 5.911,
-      "step": 14311
-    },
-    {
-      "epoch": 7.4638852672750975,
-      "grad_norm": 1.466791033744812,
-      "learning_rate": 8.612261306532664e-05,
-      "loss": 5.7725,
-      "step": 14312
-    },
-    {
-      "epoch": 7.464406779661017,
-      "grad_norm": 1.6274499893188477,
-      "learning_rate": 8.6121608040201e-05,
-      "loss": 5.0167,
-      "step": 14313
-    },
-    {
-      "epoch": 7.464928292046936,
-      "grad_norm": 1.4677599668502808,
-      "learning_rate": 8.612060301507538e-05,
-      "loss": 5.6113,
-      "step": 14314
-    },
-    {
-      "epoch": 7.465449804432855,
-      "grad_norm": 1.3876099586486816,
-      "learning_rate": 8.611959798994975e-05,
-      "loss": 5.6266,
-      "step": 14315
-    },
-    {
-      "epoch": 7.4659713168187745,
-      "grad_norm": 1.4407918453216553,
-      "learning_rate": 8.611859296482412e-05,
-      "loss": 5.6032,
-      "step": 14316
-    },
-    {
-      "epoch": 7.466492829204694,
-      "grad_norm": 1.4900299310684204,
-      "learning_rate": 8.61175879396985e-05,
-      "loss": 5.7813,
-      "step": 14317
-    },
-    {
-      "epoch": 7.4670143415906125,
-      "grad_norm": 1.4530835151672363,
-      "learning_rate": 8.611658291457287e-05,
-      "loss": 5.5158,
-      "step": 14318
-    },
-    {
-      "epoch": 7.467535853976532,
-      "grad_norm": 1.6213852167129517,
-      "learning_rate": 8.611557788944724e-05,
-      "loss": 5.4528,
-      "step": 14319
-    },
-    {
-      "epoch": 7.468057366362451,
-      "grad_norm": 1.4295357465744019,
-      "learning_rate": 8.611457286432161e-05,
-      "loss": 5.5588,
-      "step": 14320
-    },
-    {
-      "epoch": 7.46857887874837,
-      "grad_norm": 1.5892099142074585,
-      "learning_rate": 8.611356783919599e-05,
-      "loss": 5.2406,
-      "step": 14321
-    },
-    {
-      "epoch": 7.4691003911342895,
-      "grad_norm": 1.3827917575836182,
-      "learning_rate": 8.611256281407035e-05,
-      "loss": 5.6161,
-      "step": 14322
-    },
-    {
-      "epoch": 7.469621903520209,
-      "grad_norm": 1.5175367593765259,
-      "learning_rate": 8.611155778894473e-05,
-      "loss": 5.4232,
-      "step": 14323
-    },
-    {
-      "epoch": 7.4701434159061275,
-      "grad_norm": 1.5817011594772339,
-      "learning_rate": 8.611055276381909e-05,
-      "loss": 5.4281,
-      "step": 14324
-    },
-    {
-      "epoch": 7.470664928292047,
-      "grad_norm": 1.4407696723937988,
-      "learning_rate": 8.610954773869347e-05,
-      "loss": 5.9789,
-      "step": 14325
-    },
-    {
-      "epoch": 7.4711864406779664,
-      "grad_norm": 1.514825701713562,
-      "learning_rate": 8.610854271356783e-05,
-      "loss": 5.6888,
-      "step": 14326
-    },
-    {
-      "epoch": 7.471707953063885,
-      "grad_norm": 1.4803707599639893,
-      "learning_rate": 8.610753768844221e-05,
-      "loss": 5.1423,
-      "step": 14327
-    },
-    {
-      "epoch": 7.4722294654498045,
-      "grad_norm": 1.4762588739395142,
-      "learning_rate": 8.610653266331659e-05,
-      "loss": 5.6034,
-      "step": 14328
-    },
-    {
-      "epoch": 7.472750977835724,
-      "grad_norm": 1.3832391500473022,
-      "learning_rate": 8.610552763819097e-05,
-      "loss": 5.4574,
-      "step": 14329
-    },
-    {
-      "epoch": 7.4732724902216425,
-      "grad_norm": 1.4260191917419434,
-      "learning_rate": 8.610452261306533e-05,
-      "loss": 5.464,
-      "step": 14330
-    },
-    {
-      "epoch": 7.473794002607562,
-      "grad_norm": 1.5728232860565186,
-      "learning_rate": 8.61035175879397e-05,
-      "loss": 5.539,
-      "step": 14331
-    },
-    {
-      "epoch": 7.4743155149934815,
-      "grad_norm": 1.6049944162368774,
-      "learning_rate": 8.610251256281407e-05,
-      "loss": 5.23,
-      "step": 14332
-    },
-    {
-      "epoch": 7.4748370273794,
-      "grad_norm": 1.5035383701324463,
-      "learning_rate": 8.610150753768844e-05,
-      "loss": 5.4694,
-      "step": 14333
-    },
-    {
-      "epoch": 7.4753585397653195,
-      "grad_norm": 1.4754914045333862,
-      "learning_rate": 8.610050251256282e-05,
-      "loss": 5.7656,
-      "step": 14334
-    },
-    {
-      "epoch": 7.475880052151239,
-      "grad_norm": 1.4320921897888184,
-      "learning_rate": 8.609949748743718e-05,
-      "loss": 5.5553,
-      "step": 14335
-    },
-    {
-      "epoch": 7.4764015645371575,
-      "grad_norm": 1.3926585912704468,
-      "learning_rate": 8.609849246231156e-05,
-      "loss": 5.4793,
-      "step": 14336
-    },
-    {
-      "epoch": 7.476923076923077,
-      "grad_norm": 1.3676341772079468,
-      "learning_rate": 8.609748743718594e-05,
-      "loss": 5.7603,
-      "step": 14337
-    },
-    {
-      "epoch": 7.4774445893089965,
-      "grad_norm": 1.3652431964874268,
-      "learning_rate": 8.609648241206031e-05,
-      "loss": 5.918,
-      "step": 14338
-    },
-    {
-      "epoch": 7.477966101694915,
-      "grad_norm": 1.4791314601898193,
-      "learning_rate": 8.609547738693468e-05,
-      "loss": 5.3456,
-      "step": 14339
-    },
-    {
-      "epoch": 7.4784876140808345,
-      "grad_norm": 1.4573028087615967,
-      "learning_rate": 8.609447236180906e-05,
-      "loss": 5.94,
-      "step": 14340
-    },
-    {
-      "epoch": 7.479009126466754,
-      "grad_norm": 1.5230895280838013,
-      "learning_rate": 8.609346733668342e-05,
-      "loss": 5.3826,
-      "step": 14341
-    },
-    {
-      "epoch": 7.4795306388526726,
-      "grad_norm": 1.504063606262207,
-      "learning_rate": 8.60924623115578e-05,
-      "loss": 5.5283,
-      "step": 14342
-    },
-    {
-      "epoch": 7.480052151238592,
-      "grad_norm": 2.577378273010254,
-      "learning_rate": 8.609145728643216e-05,
-      "loss": 4.5913,
-      "step": 14343
-    },
-    {
-      "epoch": 7.4805736636245115,
-      "grad_norm": 1.601867914199829,
-      "learning_rate": 8.609045226130654e-05,
-      "loss": 5.655,
-      "step": 14344
-    },
-    {
-      "epoch": 7.48109517601043,
-      "grad_norm": 1.509210228919983,
-      "learning_rate": 8.60894472361809e-05,
-      "loss": 5.6375,
-      "step": 14345
-    },
-    {
-      "epoch": 7.4816166883963495,
-      "grad_norm": 1.4378777742385864,
-      "learning_rate": 8.608844221105527e-05,
-      "loss": 5.5022,
-      "step": 14346
-    },
-    {
-      "epoch": 7.482138200782269,
-      "grad_norm": 1.5622271299362183,
-      "learning_rate": 8.608743718592965e-05,
-      "loss": 5.4502,
-      "step": 14347
-    },
-    {
-      "epoch": 7.482659713168188,
-      "grad_norm": 1.3820501565933228,
-      "learning_rate": 8.608643216080403e-05,
-      "loss": 5.7039,
-      "step": 14348
-    },
-    {
-      "epoch": 7.483181225554107,
-      "grad_norm": 1.594468355178833,
-      "learning_rate": 8.60854271356784e-05,
-      "loss": 4.5349,
-      "step": 14349
-    },
-    {
-      "epoch": 7.4837027379400265,
-      "grad_norm": 1.524670124053955,
-      "learning_rate": 8.608442211055277e-05,
-      "loss": 4.9033,
-      "step": 14350
-    },
-    {
-      "epoch": 7.484224250325945,
-      "grad_norm": 1.510387897491455,
-      "learning_rate": 8.608341708542715e-05,
-      "loss": 5.3089,
-      "step": 14351
-    },
-    {
-      "epoch": 7.4847457627118645,
-      "grad_norm": 1.419113278388977,
-      "learning_rate": 8.608241206030151e-05,
-      "loss": 5.8636,
-      "step": 14352
-    },
-    {
-      "epoch": 7.485267275097783,
-      "grad_norm": 1.3459213972091675,
-      "learning_rate": 8.608140703517589e-05,
-      "loss": 5.9803,
-      "step": 14353
-    },
-    {
-      "epoch": 7.485788787483703,
-      "grad_norm": 1.4623483419418335,
-      "learning_rate": 8.608040201005025e-05,
-      "loss": 5.2705,
-      "step": 14354
-    },
-    {
-      "epoch": 7.486310299869622,
-      "grad_norm": 1.3900104761123657,
-      "learning_rate": 8.607939698492463e-05,
-      "loss": 5.5013,
-      "step": 14355
-    },
-    {
-      "epoch": 7.4868318122555415,
-      "grad_norm": 1.4796334505081177,
-      "learning_rate": 8.6078391959799e-05,
-      "loss": 5.4071,
-      "step": 14356
-    },
-    {
-      "epoch": 7.48735332464146,
-      "grad_norm": 1.5062527656555176,
-      "learning_rate": 8.607738693467337e-05,
-      "loss": 5.412,
-      "step": 14357
-    },
-    {
-      "epoch": 7.4878748370273795,
-      "grad_norm": 1.3945561647415161,
-      "learning_rate": 8.607638190954775e-05,
-      "loss": 5.5428,
-      "step": 14358
-    },
-    {
-      "epoch": 7.488396349413298,
-      "grad_norm": 1.3692220449447632,
-      "learning_rate": 8.607537688442211e-05,
-      "loss": 5.7688,
-      "step": 14359
-    },
-    {
-      "epoch": 7.488917861799218,
-      "grad_norm": 1.6224329471588135,
-      "learning_rate": 8.607437185929649e-05,
-      "loss": 5.6577,
-      "step": 14360
-    },
-    {
-      "epoch": 7.489439374185137,
-      "grad_norm": 1.385735034942627,
-      "learning_rate": 8.607336683417086e-05,
-      "loss": 5.6236,
-      "step": 14361
-    },
-    {
-      "epoch": 7.4899608865710565,
-      "grad_norm": 1.4128612279891968,
-      "learning_rate": 8.607236180904523e-05,
-      "loss": 5.6155,
-      "step": 14362
-    },
-    {
-      "epoch": 7.490482398956975,
-      "grad_norm": 1.532691478729248,
-      "learning_rate": 8.60713567839196e-05,
-      "loss": 5.1946,
-      "step": 14363
-    },
-    {
-      "epoch": 7.491003911342895,
-      "grad_norm": 1.4949405193328857,
-      "learning_rate": 8.607035175879398e-05,
-      "loss": 5.4635,
-      "step": 14364
-    },
-    {
-      "epoch": 7.491525423728813,
-      "grad_norm": 1.409379243850708,
-      "learning_rate": 8.606934673366834e-05,
-      "loss": 5.5668,
-      "step": 14365
-    },
-    {
-      "epoch": 7.492046936114733,
-      "grad_norm": 1.4646391868591309,
-      "learning_rate": 8.606834170854272e-05,
-      "loss": 5.8451,
-      "step": 14366
-    },
-    {
-      "epoch": 7.492568448500652,
-      "grad_norm": 1.3866653442382812,
-      "learning_rate": 8.606733668341708e-05,
-      "loss": 5.6677,
-      "step": 14367
-    },
-    {
-      "epoch": 7.493089960886571,
-      "grad_norm": 1.3936150074005127,
-      "learning_rate": 8.606633165829146e-05,
-      "loss": 5.5154,
-      "step": 14368
-    },
-    {
-      "epoch": 7.49361147327249,
-      "grad_norm": 1.4900511503219604,
-      "learning_rate": 8.606532663316584e-05,
-      "loss": 5.5131,
-      "step": 14369
-    },
-    {
-      "epoch": 7.49413298565841,
-      "grad_norm": 1.5182266235351562,
-      "learning_rate": 8.606432160804022e-05,
-      "loss": 5.3355,
-      "step": 14370
-    },
-    {
-      "epoch": 7.494654498044328,
-      "grad_norm": 1.418129324913025,
-      "learning_rate": 8.606331658291458e-05,
-      "loss": 5.5601,
-      "step": 14371
-    },
-    {
-      "epoch": 7.495176010430248,
-      "grad_norm": 1.3847951889038086,
-      "learning_rate": 8.606231155778894e-05,
-      "loss": 5.6372,
-      "step": 14372
-    },
-    {
-      "epoch": 7.495697522816167,
-      "grad_norm": 1.40888512134552,
-      "learning_rate": 8.606130653266332e-05,
-      "loss": 5.418,
-      "step": 14373
-    },
-    {
-      "epoch": 7.496219035202086,
-      "grad_norm": 1.4367265701293945,
-      "learning_rate": 8.606030150753769e-05,
-      "loss": 5.5986,
-      "step": 14374
-    },
-    {
-      "epoch": 7.496740547588005,
-      "grad_norm": 1.5357054471969604,
-      "learning_rate": 8.605929648241206e-05,
-      "loss": 5.6421,
-      "step": 14375
-    },
-    {
-      "epoch": 7.497262059973925,
-      "grad_norm": 1.4539191722869873,
-      "learning_rate": 8.605829145728643e-05,
-      "loss": 5.4075,
-      "step": 14376
-    },
-    {
-      "epoch": 7.497783572359843,
-      "grad_norm": 1.3913840055465698,
-      "learning_rate": 8.60572864321608e-05,
-      "loss": 5.6927,
-      "step": 14377
-    },
-    {
-      "epoch": 7.498305084745763,
-      "grad_norm": 1.5202126502990723,
-      "learning_rate": 8.605628140703518e-05,
-      "loss": 4.9966,
-      "step": 14378
-    },
-    {
-      "epoch": 7.498826597131682,
-      "grad_norm": 1.3729082345962524,
-      "learning_rate": 8.605527638190956e-05,
-      "loss": 5.2882,
-      "step": 14379
-    },
-    {
-      "epoch": 7.499348109517601,
-      "grad_norm": 1.469238519668579,
-      "learning_rate": 8.605427135678393e-05,
-      "loss": 5.3945,
-      "step": 14380
-    },
-    {
-      "epoch": 7.49986962190352,
-      "grad_norm": 1.5600395202636719,
-      "learning_rate": 8.60532663316583e-05,
-      "loss": 5.7606,
-      "step": 14381
-    },
-    {
-      "epoch": 7.50039113428944,
-      "grad_norm": 1.3746979236602783,
-      "learning_rate": 8.605226130653267e-05,
-      "loss": 5.7023,
-      "step": 14382
-    },
-    {
-      "epoch": 7.500912646675358,
-      "grad_norm": 1.4826563596725464,
-      "learning_rate": 8.605125628140705e-05,
-      "loss": 5.389,
-      "step": 14383
-    },
-    {
-      "epoch": 7.501434159061278,
-      "grad_norm": 1.403035044670105,
-      "learning_rate": 8.605025125628141e-05,
-      "loss": 5.3065,
-      "step": 14384
-    },
-    {
-      "epoch": 7.501955671447197,
-      "grad_norm": 1.4151207208633423,
-      "learning_rate": 8.604924623115577e-05,
-      "loss": 5.562,
-      "step": 14385
-    },
-    {
-      "epoch": 7.502477183833116,
-      "grad_norm": 1.4127271175384521,
-      "learning_rate": 8.604824120603015e-05,
-      "loss": 5.6924,
-      "step": 14386
-    },
-    {
-      "epoch": 7.502998696219035,
-      "grad_norm": 1.569226861000061,
-      "learning_rate": 8.604723618090452e-05,
-      "loss": 5.9015,
-      "step": 14387
-    },
-    {
-      "epoch": 7.503520208604955,
-      "grad_norm": 1.4697234630584717,
-      "learning_rate": 8.60462311557789e-05,
-      "loss": 5.6387,
-      "step": 14388
-    },
-    {
-      "epoch": 7.504041720990873,
-      "grad_norm": 1.353944182395935,
-      "learning_rate": 8.604522613065327e-05,
-      "loss": 5.8782,
-      "step": 14389
-    },
-    {
-      "epoch": 7.504563233376793,
-      "grad_norm": 1.2924259901046753,
-      "learning_rate": 8.604422110552765e-05,
-      "loss": 5.8684,
-      "step": 14390
-    },
-    {
-      "epoch": 7.505084745762712,
-      "grad_norm": 1.4000943899154663,
-      "learning_rate": 8.604321608040201e-05,
-      "loss": 4.7434,
-      "step": 14391
-    },
-    {
-      "epoch": 7.505606258148631,
-      "grad_norm": 1.4217276573181152,
-      "learning_rate": 8.604221105527639e-05,
-      "loss": 5.3836,
-      "step": 14392
-    },
-    {
-      "epoch": 7.50612777053455,
-      "grad_norm": 1.4223589897155762,
-      "learning_rate": 8.604120603015076e-05,
-      "loss": 5.45,
-      "step": 14393
-    },
-    {
-      "epoch": 7.50664928292047,
-      "grad_norm": 1.4876317977905273,
-      "learning_rate": 8.604020100502513e-05,
-      "loss": 5.0462,
-      "step": 14394
-    },
-    {
-      "epoch": 7.507170795306388,
-      "grad_norm": 1.4112242460250854,
-      "learning_rate": 8.60391959798995e-05,
-      "loss": 5.4269,
-      "step": 14395
-    },
-    {
-      "epoch": 7.507692307692308,
-      "grad_norm": 1.35599946975708,
-      "learning_rate": 8.603819095477388e-05,
-      "loss": 5.1399,
-      "step": 14396
-    },
-    {
-      "epoch": 7.508213820078227,
-      "grad_norm": 1.4037548303604126,
-      "learning_rate": 8.603718592964824e-05,
-      "loss": 5.7485,
-      "step": 14397
-    },
-    {
-      "epoch": 7.508735332464146,
-      "grad_norm": 1.39333975315094,
-      "learning_rate": 8.603618090452262e-05,
-      "loss": 5.6734,
-      "step": 14398
-    },
-    {
-      "epoch": 7.509256844850065,
-      "grad_norm": 1.3511556386947632,
-      "learning_rate": 8.6035175879397e-05,
-      "loss": 5.9106,
-      "step": 14399
-    },
-    {
-      "epoch": 7.509778357235985,
-      "grad_norm": 1.4763686656951904,
-      "learning_rate": 8.603417085427136e-05,
-      "loss": 5.1948,
-      "step": 14400
-    },
-    {
-      "epoch": 7.510299869621903,
-      "grad_norm": 1.3850815296173096,
-      "learning_rate": 8.603316582914574e-05,
-      "loss": 5.9674,
-      "step": 14401
-    },
-    {
-      "epoch": 7.510821382007823,
-      "grad_norm": 1.3484561443328857,
-      "learning_rate": 8.60321608040201e-05,
-      "loss": 5.6702,
-      "step": 14402
-    },
-    {
-      "epoch": 7.511342894393742,
-      "grad_norm": 1.4632606506347656,
-      "learning_rate": 8.603115577889448e-05,
-      "loss": 5.7945,
-      "step": 14403
-    },
-    {
-      "epoch": 7.511864406779661,
-      "grad_norm": 1.4342886209487915,
-      "learning_rate": 8.603015075376884e-05,
-      "loss": 5.8189,
-      "step": 14404
-    },
-    {
-      "epoch": 7.51238591916558,
-      "grad_norm": 1.3690242767333984,
-      "learning_rate": 8.602914572864322e-05,
-      "loss": 5.6657,
-      "step": 14405
-    },
-    {
-      "epoch": 7.5129074315515,
-      "grad_norm": 1.4263441562652588,
-      "learning_rate": 8.602814070351759e-05,
-      "loss": 5.3018,
-      "step": 14406
-    },
-    {
-      "epoch": 7.513428943937418,
-      "grad_norm": 1.4232409000396729,
-      "learning_rate": 8.602713567839196e-05,
-      "loss": 5.5999,
-      "step": 14407
-    },
-    {
-      "epoch": 7.513950456323338,
-      "grad_norm": 1.4232099056243896,
-      "learning_rate": 8.602613065326633e-05,
-      "loss": 5.4424,
-      "step": 14408
-    },
-    {
-      "epoch": 7.514471968709257,
-      "grad_norm": 1.4716486930847168,
-      "learning_rate": 8.60251256281407e-05,
-      "loss": 5.3518,
-      "step": 14409
-    },
-    {
-      "epoch": 7.514993481095176,
-      "grad_norm": 1.5529555082321167,
-      "learning_rate": 8.602412060301508e-05,
-      "loss": 4.8271,
-      "step": 14410
-    },
-    {
-      "epoch": 7.515514993481095,
-      "grad_norm": 1.3759700059890747,
-      "learning_rate": 8.602311557788945e-05,
-      "loss": 5.6752,
-      "step": 14411
-    },
-    {
-      "epoch": 7.516036505867015,
-      "grad_norm": 1.3353825807571411,
-      "learning_rate": 8.602211055276383e-05,
-      "loss": 5.7744,
-      "step": 14412
-    },
-    {
-      "epoch": 7.516558018252933,
-      "grad_norm": 1.339860439300537,
-      "learning_rate": 8.602110552763819e-05,
-      "loss": 5.5908,
-      "step": 14413
-    },
-    {
-      "epoch": 7.517079530638853,
-      "grad_norm": 1.3372331857681274,
-      "learning_rate": 8.602010050251257e-05,
-      "loss": 5.8045,
-      "step": 14414
-    },
-    {
-      "epoch": 7.517601043024772,
-      "grad_norm": 1.5018301010131836,
-      "learning_rate": 8.601909547738693e-05,
-      "loss": 5.5836,
-      "step": 14415
-    },
-    {
-      "epoch": 7.518122555410691,
-      "grad_norm": 1.3613617420196533,
-      "learning_rate": 8.601809045226131e-05,
-      "loss": 5.7909,
-      "step": 14416
-    },
-    {
-      "epoch": 7.51864406779661,
-      "grad_norm": 1.4318809509277344,
-      "learning_rate": 8.601708542713568e-05,
-      "loss": 5.427,
-      "step": 14417
-    },
-    {
-      "epoch": 7.51916558018253,
-      "grad_norm": 1.3824576139450073,
-      "learning_rate": 8.601608040201005e-05,
-      "loss": 5.3923,
-      "step": 14418
-    },
-    {
-      "epoch": 7.519687092568448,
-      "grad_norm": 1.4758055210113525,
-      "learning_rate": 8.601507537688443e-05,
-      "loss": 4.8623,
-      "step": 14419
-    },
-    {
-      "epoch": 7.520208604954368,
-      "grad_norm": 1.4531676769256592,
-      "learning_rate": 8.601407035175881e-05,
-      "loss": 5.6739,
-      "step": 14420
-    },
-    {
-      "epoch": 7.520730117340287,
-      "grad_norm": 1.496105670928955,
-      "learning_rate": 8.601306532663317e-05,
-      "loss": 5.1408,
-      "step": 14421
-    },
-    {
-      "epoch": 7.521251629726206,
-      "grad_norm": 1.534218192100525,
-      "learning_rate": 8.601206030150755e-05,
-      "loss": 5.451,
-      "step": 14422
-    },
-    {
-      "epoch": 7.521773142112125,
-      "grad_norm": 1.4804588556289673,
-      "learning_rate": 8.601105527638191e-05,
-      "loss": 5.54,
-      "step": 14423
-    },
-    {
-      "epoch": 7.522294654498045,
-      "grad_norm": 1.330087423324585,
-      "learning_rate": 8.601005025125628e-05,
-      "loss": 5.4902,
-      "step": 14424
-    },
-    {
-      "epoch": 7.522816166883963,
-      "grad_norm": 1.3668454885482788,
-      "learning_rate": 8.600904522613066e-05,
-      "loss": 5.2132,
-      "step": 14425
-    },
-    {
-      "epoch": 7.523337679269883,
-      "grad_norm": 1.491457462310791,
-      "learning_rate": 8.600804020100502e-05,
-      "loss": 5.6804,
-      "step": 14426
-    },
-    {
-      "epoch": 7.523859191655802,
-      "grad_norm": 1.4169975519180298,
-      "learning_rate": 8.60070351758794e-05,
-      "loss": 5.6025,
-      "step": 14427
-    },
-    {
-      "epoch": 7.524380704041721,
-      "grad_norm": 1.469190239906311,
-      "learning_rate": 8.600603015075376e-05,
-      "loss": 4.8548,
-      "step": 14428
-    },
-    {
-      "epoch": 7.52490221642764,
-      "grad_norm": 1.3302276134490967,
-      "learning_rate": 8.600502512562814e-05,
-      "loss": 5.7888,
-      "step": 14429
-    },
-    {
-      "epoch": 7.52542372881356,
-      "grad_norm": 1.4936333894729614,
-      "learning_rate": 8.600402010050252e-05,
-      "loss": 5.2033,
-      "step": 14430
-    },
-    {
-      "epoch": 7.525945241199478,
-      "grad_norm": 1.512298822402954,
-      "learning_rate": 8.60030150753769e-05,
-      "loss": 5.5871,
-      "step": 14431
-    },
-    {
-      "epoch": 7.526466753585398,
-      "grad_norm": 1.3258998394012451,
-      "learning_rate": 8.600201005025126e-05,
-      "loss": 5.8521,
-      "step": 14432
-    },
-    {
-      "epoch": 7.526988265971317,
-      "grad_norm": 1.3351576328277588,
-      "learning_rate": 8.600100502512564e-05,
-      "loss": 5.7803,
-      "step": 14433
-    },
-    {
-      "epoch": 7.527509778357236,
-      "grad_norm": 1.3888005018234253,
-      "learning_rate": 8.6e-05,
-      "loss": 5.5329,
-      "step": 14434
-    },
-    {
-      "epoch": 7.528031290743155,
-      "grad_norm": 1.3629043102264404,
-      "learning_rate": 8.599899497487438e-05,
-      "loss": 5.6424,
-      "step": 14435
-    },
-    {
-      "epoch": 7.528552803129074,
-      "grad_norm": 1.4096856117248535,
-      "learning_rate": 8.599798994974875e-05,
-      "loss": 5.2571,
-      "step": 14436
-    },
-    {
-      "epoch": 7.529074315514993,
-      "grad_norm": 1.2979843616485596,
-      "learning_rate": 8.599698492462312e-05,
-      "loss": 5.6272,
-      "step": 14437
-    },
-    {
-      "epoch": 7.529595827900913,
-      "grad_norm": 1.3374485969543457,
-      "learning_rate": 8.599597989949749e-05,
-      "loss": 5.9124,
-      "step": 14438
-    },
-    {
-      "epoch": 7.530117340286832,
-      "grad_norm": 1.3806493282318115,
-      "learning_rate": 8.599497487437187e-05,
-      "loss": 5.3851,
-      "step": 14439
-    },
-    {
-      "epoch": 7.530638852672751,
-      "grad_norm": 1.3650271892547607,
-      "learning_rate": 8.599396984924624e-05,
-      "loss": 5.5857,
-      "step": 14440
-    },
-    {
-      "epoch": 7.53116036505867,
-      "grad_norm": 1.3986033201217651,
-      "learning_rate": 8.599296482412061e-05,
-      "loss": 6.0091,
-      "step": 14441
-    },
-    {
-      "epoch": 7.531681877444589,
-      "grad_norm": 1.366439938545227,
-      "learning_rate": 8.599195979899499e-05,
-      "loss": 5.5646,
-      "step": 14442
-    },
-    {
-      "epoch": 7.532203389830508,
-      "grad_norm": 1.4063334465026855,
-      "learning_rate": 8.599095477386935e-05,
-      "loss": 5.4478,
-      "step": 14443
-    },
-    {
-      "epoch": 7.532724902216428,
-      "grad_norm": 1.5154016017913818,
-      "learning_rate": 8.598994974874373e-05,
-      "loss": 5.284,
-      "step": 14444
-    },
-    {
-      "epoch": 7.533246414602347,
-      "grad_norm": 1.3358287811279297,
-      "learning_rate": 8.598894472361809e-05,
-      "loss": 5.3807,
-      "step": 14445
-    },
-    {
-      "epoch": 7.533767926988266,
-      "grad_norm": 1.4691824913024902,
-      "learning_rate": 8.598793969849247e-05,
-      "loss": 5.7575,
-      "step": 14446
-    },
-    {
-      "epoch": 7.534289439374185,
-      "grad_norm": 1.4193345308303833,
-      "learning_rate": 8.598693467336683e-05,
-      "loss": 5.4015,
-      "step": 14447
-    },
-    {
-      "epoch": 7.534810951760104,
-      "grad_norm": 1.415890097618103,
-      "learning_rate": 8.598592964824121e-05,
-      "loss": 5.8251,
-      "step": 14448
-    },
-    {
-      "epoch": 7.535332464146023,
-      "grad_norm": 1.4307185411453247,
-      "learning_rate": 8.598492462311558e-05,
-      "loss": 4.9567,
-      "step": 14449
-    },
-    {
-      "epoch": 7.535853976531943,
-      "grad_norm": 1.449263334274292,
-      "learning_rate": 8.598391959798995e-05,
-      "loss": 5.6431,
-      "step": 14450
-    },
-    {
-      "epoch": 7.536375488917862,
-      "grad_norm": 1.53867506980896,
-      "learning_rate": 8.598291457286433e-05,
-      "loss": 5.398,
-      "step": 14451
-    },
-    {
-      "epoch": 7.536897001303781,
-      "grad_norm": 1.434108018875122,
-      "learning_rate": 8.59819095477387e-05,
-      "loss": 5.5036,
-      "step": 14452
-    },
-    {
-      "epoch": 7.5374185136897,
-      "grad_norm": 1.3901947736740112,
-      "learning_rate": 8.598090452261307e-05,
-      "loss": 5.622,
-      "step": 14453
-    },
-    {
-      "epoch": 7.537940026075619,
-      "grad_norm": 1.5232458114624023,
-      "learning_rate": 8.597989949748744e-05,
-      "loss": 5.5522,
-      "step": 14454
-    },
-    {
-      "epoch": 7.538461538461538,
-      "grad_norm": 1.437117338180542,
-      "learning_rate": 8.597889447236182e-05,
-      "loss": 5.8877,
-      "step": 14455
-    },
-    {
-      "epoch": 7.538983050847458,
-      "grad_norm": 1.5587292909622192,
-      "learning_rate": 8.597788944723618e-05,
-      "loss": 5.2392,
-      "step": 14456
-    },
-    {
-      "epoch": 7.539504563233377,
-      "grad_norm": 1.5550967454910278,
-      "learning_rate": 8.597688442211056e-05,
-      "loss": 5.7528,
-      "step": 14457
-    },
-    {
-      "epoch": 7.540026075619296,
-      "grad_norm": 1.2937589883804321,
-      "learning_rate": 8.597587939698492e-05,
-      "loss": 6.0642,
-      "step": 14458
-    },
-    {
-      "epoch": 7.540547588005215,
-      "grad_norm": 1.3744772672653198,
-      "learning_rate": 8.59748743718593e-05,
-      "loss": 5.6839,
-      "step": 14459
-    },
-    {
-      "epoch": 7.541069100391134,
-      "grad_norm": 1.498262643814087,
-      "learning_rate": 8.597386934673366e-05,
-      "loss": 5.6816,
-      "step": 14460
-    },
-    {
-      "epoch": 7.541590612777053,
-      "grad_norm": 1.398675799369812,
-      "learning_rate": 8.597286432160804e-05,
-      "loss": 5.2682,
-      "step": 14461
-    },
-    {
-      "epoch": 7.542112125162973,
-      "grad_norm": 1.4919991493225098,
-      "learning_rate": 8.597185929648242e-05,
-      "loss": 5.569,
-      "step": 14462
-    },
-    {
-      "epoch": 7.542633637548892,
-      "grad_norm": 1.3810702562332153,
-      "learning_rate": 8.59708542713568e-05,
-      "loss": 5.3449,
-      "step": 14463
-    },
-    {
-      "epoch": 7.543155149934811,
-      "grad_norm": 1.480148434638977,
-      "learning_rate": 8.596984924623116e-05,
-      "loss": 5.5025,
-      "step": 14464
-    },
-    {
-      "epoch": 7.54367666232073,
-      "grad_norm": 1.4620416164398193,
-      "learning_rate": 8.596884422110553e-05,
-      "loss": 5.5,
-      "step": 14465
-    },
-    {
-      "epoch": 7.544198174706649,
-      "grad_norm": 1.3785277605056763,
-      "learning_rate": 8.59678391959799e-05,
-      "loss": 5.8036,
-      "step": 14466
-    },
-    {
-      "epoch": 7.544719687092568,
-      "grad_norm": 1.348463535308838,
-      "learning_rate": 8.596683417085427e-05,
-      "loss": 5.9949,
-      "step": 14467
-    },
-    {
-      "epoch": 7.545241199478488,
-      "grad_norm": 1.608410120010376,
-      "learning_rate": 8.596582914572865e-05,
-      "loss": 4.9123,
-      "step": 14468
-    },
-    {
-      "epoch": 7.545762711864406,
-      "grad_norm": 1.5562827587127686,
-      "learning_rate": 8.596482412060301e-05,
-      "loss": 5.4268,
-      "step": 14469
-    },
-    {
-      "epoch": 7.546284224250326,
-      "grad_norm": 1.3261356353759766,
-      "learning_rate": 8.596381909547739e-05,
-      "loss": 5.9104,
-      "step": 14470
-    },
-    {
-      "epoch": 7.546805736636245,
-      "grad_norm": 1.419384479522705,
-      "learning_rate": 8.596281407035177e-05,
-      "loss": 5.4102,
-      "step": 14471
-    },
-    {
-      "epoch": 7.547327249022164,
-      "grad_norm": 1.487569808959961,
-      "learning_rate": 8.596180904522614e-05,
-      "loss": 5.7281,
-      "step": 14472
-    },
-    {
-      "epoch": 7.547848761408083,
-      "grad_norm": 1.5440367460250854,
-      "learning_rate": 8.596080402010051e-05,
-      "loss": 5.6354,
-      "step": 14473
-    },
-    {
-      "epoch": 7.548370273794003,
-      "grad_norm": 1.8302373886108398,
-      "learning_rate": 8.595979899497489e-05,
-      "loss": 5.5104,
-      "step": 14474
-    },
-    {
-      "epoch": 7.548891786179921,
-      "grad_norm": 1.380648136138916,
-      "learning_rate": 8.595879396984925e-05,
-      "loss": 5.6028,
-      "step": 14475
-    },
-    {
-      "epoch": 7.549413298565841,
-      "grad_norm": 1.5158854722976685,
-      "learning_rate": 8.595778894472363e-05,
-      "loss": 5.3443,
-      "step": 14476
-    },
-    {
-      "epoch": 7.54993481095176,
-      "grad_norm": 1.4212790727615356,
-      "learning_rate": 8.595678391959799e-05,
-      "loss": 5.7541,
-      "step": 14477
-    },
-    {
-      "epoch": 7.550456323337679,
-      "grad_norm": 1.4399834871292114,
-      "learning_rate": 8.595577889447236e-05,
-      "loss": 5.4171,
-      "step": 14478
-    },
-    {
-      "epoch": 7.550977835723598,
-      "grad_norm": 1.3858368396759033,
-      "learning_rate": 8.595477386934673e-05,
-      "loss": 5.6575,
-      "step": 14479
-    },
-    {
-      "epoch": 7.551499348109518,
-      "grad_norm": 1.3507689237594604,
-      "learning_rate": 8.59537688442211e-05,
-      "loss": 5.5356,
-      "step": 14480
-    },
-    {
-      "epoch": 7.552020860495436,
-      "grad_norm": 1.3642578125,
-      "learning_rate": 8.595276381909548e-05,
-      "loss": 5.8055,
-      "step": 14481
-    },
-    {
-      "epoch": 7.552542372881356,
-      "grad_norm": 1.4757659435272217,
-      "learning_rate": 8.595175879396985e-05,
-      "loss": 5.4589,
-      "step": 14482
-    },
-    {
-      "epoch": 7.553063885267275,
-      "grad_norm": 1.4366340637207031,
-      "learning_rate": 8.595075376884423e-05,
-      "loss": 5.456,
-      "step": 14483
-    },
-    {
-      "epoch": 7.553585397653194,
-      "grad_norm": 1.3678559064865112,
-      "learning_rate": 8.59497487437186e-05,
-      "loss": 5.866,
-      "step": 14484
-    },
-    {
-      "epoch": 7.554106910039113,
-      "grad_norm": 1.377436876296997,
-      "learning_rate": 8.594874371859297e-05,
-      "loss": 5.7199,
-      "step": 14485
-    },
-    {
-      "epoch": 7.554628422425033,
-      "grad_norm": 1.4054288864135742,
-      "learning_rate": 8.594773869346734e-05,
-      "loss": 5.829,
-      "step": 14486
-    },
-    {
-      "epoch": 7.555149934810951,
-      "grad_norm": 1.4703978300094604,
-      "learning_rate": 8.594673366834172e-05,
-      "loss": 5.4015,
-      "step": 14487
-    },
-    {
-      "epoch": 7.555671447196871,
-      "grad_norm": 1.502322793006897,
-      "learning_rate": 8.594572864321608e-05,
-      "loss": 5.4738,
-      "step": 14488
-    },
-    {
-      "epoch": 7.55619295958279,
-      "grad_norm": 1.4493664503097534,
-      "learning_rate": 8.594472361809046e-05,
-      "loss": 5.5929,
-      "step": 14489
-    },
-    {
-      "epoch": 7.556714471968709,
-      "grad_norm": 1.4948803186416626,
-      "learning_rate": 8.594371859296482e-05,
-      "loss": 5.6785,
-      "step": 14490
-    },
-    {
-      "epoch": 7.557235984354628,
-      "grad_norm": 1.4655007123947144,
-      "learning_rate": 8.59427135678392e-05,
-      "loss": 6.0633,
-      "step": 14491
-    },
-    {
-      "epoch": 7.557757496740548,
-      "grad_norm": 1.351957082748413,
-      "learning_rate": 8.594170854271358e-05,
-      "loss": 5.2559,
-      "step": 14492
-    },
-    {
-      "epoch": 7.558279009126466,
-      "grad_norm": 1.675897479057312,
-      "learning_rate": 8.594070351758794e-05,
-      "loss": 5.0084,
-      "step": 14493
-    },
-    {
-      "epoch": 7.558800521512386,
-      "grad_norm": 1.486124038696289,
-      "learning_rate": 8.593969849246232e-05,
-      "loss": 5.3124,
-      "step": 14494
-    },
-    {
-      "epoch": 7.559322033898305,
-      "grad_norm": 1.5843571424484253,
-      "learning_rate": 8.593869346733668e-05,
-      "loss": 4.7213,
-      "step": 14495
-    },
-    {
-      "epoch": 7.559843546284224,
-      "grad_norm": 1.4767789840698242,
-      "learning_rate": 8.593768844221106e-05,
-      "loss": 5.6542,
-      "step": 14496
-    },
-    {
-      "epoch": 7.560365058670143,
-      "grad_norm": 1.5027638673782349,
-      "learning_rate": 8.593668341708543e-05,
-      "loss": 5.6343,
-      "step": 14497
-    },
-    {
-      "epoch": 7.560886571056063,
-      "grad_norm": 1.38805091381073,
-      "learning_rate": 8.59356783919598e-05,
-      "loss": 5.9068,
-      "step": 14498
-    },
-    {
-      "epoch": 7.5614080834419815,
-      "grad_norm": 1.3647178411483765,
-      "learning_rate": 8.593467336683417e-05,
-      "loss": 5.8438,
-      "step": 14499
-    },
-    {
-      "epoch": 7.561929595827901,
-      "grad_norm": 1.3861160278320312,
-      "learning_rate": 8.593366834170855e-05,
-      "loss": 5.5691,
-      "step": 14500
-    },
-    {
-      "epoch": 7.56245110821382,
-      "grad_norm": 1.3433409929275513,
-      "learning_rate": 8.593266331658291e-05,
-      "loss": 5.8111,
-      "step": 14501
-    },
-    {
-      "epoch": 7.562972620599739,
-      "grad_norm": 1.4534703493118286,
-      "learning_rate": 8.593165829145729e-05,
-      "loss": 5.737,
-      "step": 14502
-    },
-    {
-      "epoch": 7.563494132985658,
-      "grad_norm": 1.4970481395721436,
-      "learning_rate": 8.593065326633167e-05,
-      "loss": 5.6365,
-      "step": 14503
-    },
-    {
-      "epoch": 7.564015645371578,
-      "grad_norm": 1.42659592628479,
-      "learning_rate": 8.592964824120603e-05,
-      "loss": 5.668,
-      "step": 14504
-    },
-    {
-      "epoch": 7.5645371577574965,
-      "grad_norm": 1.5125224590301514,
-      "learning_rate": 8.592864321608041e-05,
-      "loss": 5.5271,
-      "step": 14505
-    },
-    {
-      "epoch": 7.565058670143416,
-      "grad_norm": 1.475508689880371,
-      "learning_rate": 8.592763819095477e-05,
-      "loss": 5.8637,
-      "step": 14506
-    },
-    {
-      "epoch": 7.565580182529335,
-      "grad_norm": 1.5756394863128662,
-      "learning_rate": 8.592663316582915e-05,
-      "loss": 5.1239,
-      "step": 14507
-    },
-    {
-      "epoch": 7.566101694915254,
-      "grad_norm": 1.4717713594436646,
-      "learning_rate": 8.592562814070352e-05,
-      "loss": 5.0009,
-      "step": 14508
-    },
-    {
-      "epoch": 7.566623207301173,
-      "grad_norm": 1.3063156604766846,
-      "learning_rate": 8.592462311557789e-05,
-      "loss": 6.163,
-      "step": 14509
-    },
-    {
-      "epoch": 7.567144719687093,
-      "grad_norm": 1.4517159461975098,
-      "learning_rate": 8.592361809045226e-05,
-      "loss": 5.211,
-      "step": 14510
-    },
-    {
-      "epoch": 7.5676662320730115,
-      "grad_norm": 1.3912851810455322,
-      "learning_rate": 8.592261306532664e-05,
-      "loss": 5.778,
-      "step": 14511
-    },
-    {
-      "epoch": 7.568187744458931,
-      "grad_norm": 1.343937873840332,
-      "learning_rate": 8.592160804020101e-05,
-      "loss": 5.5559,
-      "step": 14512
-    },
-    {
-      "epoch": 7.56870925684485,
-      "grad_norm": 1.5128029584884644,
-      "learning_rate": 8.592060301507539e-05,
-      "loss": 5.522,
-      "step": 14513
-    },
-    {
-      "epoch": 7.569230769230769,
-      "grad_norm": 1.5881611108779907,
-      "learning_rate": 8.591959798994976e-05,
-      "loss": 5.4716,
-      "step": 14514
-    },
-    {
-      "epoch": 7.5697522816166884,
-      "grad_norm": 1.3552659749984741,
-      "learning_rate": 8.591859296482413e-05,
-      "loss": 5.87,
-      "step": 14515
-    },
-    {
-      "epoch": 7.570273794002608,
-      "grad_norm": 1.4371100664138794,
-      "learning_rate": 8.59175879396985e-05,
-      "loss": 5.4748,
-      "step": 14516
-    },
-    {
-      "epoch": 7.5707953063885265,
-      "grad_norm": 1.406935453414917,
-      "learning_rate": 8.591658291457286e-05,
-      "loss": 5.4257,
-      "step": 14517
-    },
-    {
-      "epoch": 7.571316818774446,
-      "grad_norm": 1.3984942436218262,
-      "learning_rate": 8.591557788944724e-05,
-      "loss": 5.3912,
-      "step": 14518
-    },
-    {
-      "epoch": 7.571838331160365,
-      "grad_norm": 1.581470251083374,
-      "learning_rate": 8.59145728643216e-05,
-      "loss": 5.1202,
-      "step": 14519
-    },
-    {
-      "epoch": 7.572359843546284,
-      "grad_norm": 1.4650976657867432,
-      "learning_rate": 8.591356783919598e-05,
-      "loss": 5.7334,
-      "step": 14520
-    },
-    {
-      "epoch": 7.5728813559322035,
-      "grad_norm": 1.5230457782745361,
-      "learning_rate": 8.591256281407035e-05,
-      "loss": 5.4284,
-      "step": 14521
-    },
-    {
-      "epoch": 7.573402868318123,
-      "grad_norm": 1.5607821941375732,
-      "learning_rate": 8.591155778894472e-05,
-      "loss": 5.5479,
-      "step": 14522
-    },
-    {
-      "epoch": 7.5739243807040415,
-      "grad_norm": 1.4428925514221191,
-      "learning_rate": 8.59105527638191e-05,
-      "loss": 5.8909,
-      "step": 14523
-    },
-    {
-      "epoch": 7.574445893089961,
-      "grad_norm": 1.5931625366210938,
-      "learning_rate": 8.590954773869348e-05,
-      "loss": 5.0729,
-      "step": 14524
-    },
-    {
-      "epoch": 7.57496740547588,
-      "grad_norm": 1.4365637302398682,
-      "learning_rate": 8.590854271356784e-05,
-      "loss": 5.4241,
-      "step": 14525
-    },
-    {
-      "epoch": 7.575488917861799,
-      "grad_norm": 1.6107031106948853,
-      "learning_rate": 8.590753768844222e-05,
-      "loss": 4.9528,
-      "step": 14526
-    },
-    {
-      "epoch": 7.5760104302477185,
-      "grad_norm": 1.5184694528579712,
-      "learning_rate": 8.590653266331659e-05,
-      "loss": 5.1976,
-      "step": 14527
-    },
-    {
-      "epoch": 7.576531942633638,
-      "grad_norm": 1.4071118831634521,
-      "learning_rate": 8.590552763819096e-05,
-      "loss": 5.8603,
-      "step": 14528
-    },
-    {
-      "epoch": 7.5770534550195565,
-      "grad_norm": 1.459815263748169,
-      "learning_rate": 8.590452261306533e-05,
-      "loss": 5.7651,
-      "step": 14529
-    },
-    {
-      "epoch": 7.577574967405476,
-      "grad_norm": 1.5541770458221436,
-      "learning_rate": 8.59035175879397e-05,
-      "loss": 5.2337,
-      "step": 14530
-    },
-    {
-      "epoch": 7.578096479791395,
-      "grad_norm": 1.3894541263580322,
-      "learning_rate": 8.590251256281407e-05,
-      "loss": 5.6794,
-      "step": 14531
-    },
-    {
-      "epoch": 7.578617992177314,
-      "grad_norm": 1.4117916822433472,
-      "learning_rate": 8.590150753768845e-05,
-      "loss": 5.4685,
-      "step": 14532
-    },
-    {
-      "epoch": 7.5791395045632335,
-      "grad_norm": 1.4792473316192627,
-      "learning_rate": 8.590050251256283e-05,
-      "loss": 5.9351,
-      "step": 14533
-    },
-    {
-      "epoch": 7.579661016949153,
-      "grad_norm": 1.4343838691711426,
-      "learning_rate": 8.589949748743719e-05,
-      "loss": 5.4192,
-      "step": 14534
-    },
-    {
-      "epoch": 7.5801825293350715,
-      "grad_norm": 1.3310928344726562,
-      "learning_rate": 8.589849246231157e-05,
-      "loss": 5.6716,
-      "step": 14535
-    },
-    {
-      "epoch": 7.580704041720991,
-      "grad_norm": 1.4210768938064575,
-      "learning_rate": 8.589748743718593e-05,
-      "loss": 5.8952,
-      "step": 14536
-    },
-    {
-      "epoch": 7.58122555410691,
-      "grad_norm": 1.4781638383865356,
-      "learning_rate": 8.589648241206031e-05,
-      "loss": 5.6032,
-      "step": 14537
-    },
-    {
-      "epoch": 7.581747066492829,
-      "grad_norm": 1.4891130924224854,
-      "learning_rate": 8.589547738693467e-05,
-      "loss": 5.3457,
-      "step": 14538
-    },
-    {
-      "epoch": 7.5822685788787485,
-      "grad_norm": 1.453339695930481,
-      "learning_rate": 8.589447236180905e-05,
-      "loss": 5.626,
-      "step": 14539
-    },
-    {
-      "epoch": 7.582790091264668,
-      "grad_norm": 1.4110653400421143,
-      "learning_rate": 8.589346733668342e-05,
-      "loss": 5.4056,
-      "step": 14540
-    },
-    {
-      "epoch": 7.5833116036505865,
-      "grad_norm": 1.3479136228561401,
-      "learning_rate": 8.58924623115578e-05,
-      "loss": 5.2099,
-      "step": 14541
-    },
-    {
-      "epoch": 7.583833116036506,
-      "grad_norm": 1.3129987716674805,
-      "learning_rate": 8.589145728643216e-05,
-      "loss": 5.8359,
-      "step": 14542
-    },
-    {
-      "epoch": 7.584354628422425,
-      "grad_norm": 1.4158906936645508,
-      "learning_rate": 8.589045226130654e-05,
-      "loss": 4.9837,
-      "step": 14543
-    },
-    {
-      "epoch": 7.584876140808344,
-      "grad_norm": 1.5240343809127808,
-      "learning_rate": 8.588944723618091e-05,
-      "loss": 5.398,
-      "step": 14544
-    },
-    {
-      "epoch": 7.5853976531942635,
-      "grad_norm": 1.4125995635986328,
-      "learning_rate": 8.588844221105528e-05,
-      "loss": 5.6063,
-      "step": 14545
-    },
-    {
-      "epoch": 7.585919165580183,
-      "grad_norm": 1.7432910203933716,
-      "learning_rate": 8.588743718592966e-05,
-      "loss": 5.3051,
-      "step": 14546
-    },
-    {
-      "epoch": 7.5864406779661016,
-      "grad_norm": 1.4536397457122803,
-      "learning_rate": 8.588643216080402e-05,
-      "loss": 5.6085,
-      "step": 14547
-    },
-    {
-      "epoch": 7.586962190352021,
-      "grad_norm": 1.5073397159576416,
-      "learning_rate": 8.58854271356784e-05,
-      "loss": 5.3252,
-      "step": 14548
-    },
-    {
-      "epoch": 7.58748370273794,
-      "grad_norm": 1.4076570272445679,
-      "learning_rate": 8.588442211055276e-05,
-      "loss": 6.083,
-      "step": 14549
-    },
-    {
-      "epoch": 7.588005215123859,
-      "grad_norm": 1.5253801345825195,
-      "learning_rate": 8.588341708542714e-05,
-      "loss": 5.5869,
-      "step": 14550
-    },
-    {
-      "epoch": 7.5885267275097785,
-      "grad_norm": 1.4401565790176392,
-      "learning_rate": 8.58824120603015e-05,
-      "loss": 5.3567,
-      "step": 14551
-    },
-    {
-      "epoch": 7.589048239895698,
-      "grad_norm": 1.4530861377716064,
-      "learning_rate": 8.588140703517588e-05,
-      "loss": 5.7219,
-      "step": 14552
-    },
-    {
-      "epoch": 7.589569752281617,
-      "grad_norm": 1.6317775249481201,
-      "learning_rate": 8.588040201005026e-05,
-      "loss": 5.271,
-      "step": 14553
-    },
-    {
-      "epoch": 7.590091264667536,
-      "grad_norm": 1.4088332653045654,
-      "learning_rate": 8.587939698492464e-05,
-      "loss": 5.887,
-      "step": 14554
-    },
-    {
-      "epoch": 7.590612777053455,
-      "grad_norm": 1.4815906286239624,
-      "learning_rate": 8.5878391959799e-05,
-      "loss": 5.3528,
-      "step": 14555
-    },
-    {
-      "epoch": 7.591134289439374,
-      "grad_norm": 1.4869357347488403,
-      "learning_rate": 8.587738693467338e-05,
-      "loss": 5.5335,
-      "step": 14556
-    },
-    {
-      "epoch": 7.5916558018252935,
-      "grad_norm": 1.6332803964614868,
-      "learning_rate": 8.587638190954774e-05,
-      "loss": 4.9895,
-      "step": 14557
-    },
-    {
-      "epoch": 7.592177314211213,
-      "grad_norm": 1.4781339168548584,
-      "learning_rate": 8.587537688442211e-05,
-      "loss": 5.1508,
-      "step": 14558
-    },
-    {
-      "epoch": 7.592698826597132,
-      "grad_norm": 1.6180503368377686,
-      "learning_rate": 8.587437185929649e-05,
-      "loss": 5.1106,
-      "step": 14559
-    },
-    {
-      "epoch": 7.593220338983051,
-      "grad_norm": 1.3682177066802979,
-      "learning_rate": 8.587336683417085e-05,
-      "loss": 5.3729,
-      "step": 14560
-    },
-    {
-      "epoch": 7.59374185136897,
-      "grad_norm": 1.426377296447754,
-      "learning_rate": 8.587236180904523e-05,
-      "loss": 5.187,
-      "step": 14561
-    },
-    {
-      "epoch": 7.594263363754889,
-      "grad_norm": 1.3945800065994263,
-      "learning_rate": 8.587135678391959e-05,
-      "loss": 5.6364,
-      "step": 14562
-    },
-    {
-      "epoch": 7.5947848761408085,
-      "grad_norm": 1.5481476783752441,
-      "learning_rate": 8.587035175879397e-05,
-      "loss": 5.8999,
-      "step": 14563
-    },
-    {
-      "epoch": 7.595306388526727,
-      "grad_norm": 1.465919852256775,
-      "learning_rate": 8.586934673366835e-05,
-      "loss": 5.9348,
-      "step": 14564
-    },
-    {
-      "epoch": 7.595827900912647,
-      "grad_norm": 1.4737963676452637,
-      "learning_rate": 8.586834170854273e-05,
-      "loss": 5.6371,
-      "step": 14565
-    },
-    {
-      "epoch": 7.596349413298566,
-      "grad_norm": 1.4509172439575195,
-      "learning_rate": 8.586733668341709e-05,
-      "loss": 5.5305,
-      "step": 14566
-    },
-    {
-      "epoch": 7.596870925684485,
-      "grad_norm": 1.4933651685714722,
-      "learning_rate": 8.586633165829147e-05,
-      "loss": 5.4979,
-      "step": 14567
-    },
-    {
-      "epoch": 7.597392438070404,
-      "grad_norm": 1.4171520471572876,
-      "learning_rate": 8.586532663316583e-05,
-      "loss": 5.9455,
-      "step": 14568
-    },
-    {
-      "epoch": 7.597913950456324,
-      "grad_norm": 1.4872018098831177,
-      "learning_rate": 8.586432160804021e-05,
-      "loss": 5.6514,
-      "step": 14569
-    },
-    {
-      "epoch": 7.598435462842242,
-      "grad_norm": 1.4347187280654907,
-      "learning_rate": 8.586331658291457e-05,
-      "loss": 5.5563,
-      "step": 14570
-    },
-    {
-      "epoch": 7.598956975228162,
-      "grad_norm": 1.3686541318893433,
-      "learning_rate": 8.586231155778894e-05,
-      "loss": 5.4448,
-      "step": 14571
-    },
-    {
-      "epoch": 7.599478487614081,
-      "grad_norm": 1.5874465703964233,
-      "learning_rate": 8.586130653266332e-05,
-      "loss": 5.4713,
-      "step": 14572
-    },
-    {
-      "epoch": 7.6,
-      "grad_norm": 1.3978043794631958,
-      "learning_rate": 8.58603015075377e-05,
-      "loss": 5.678,
-      "step": 14573
-    },
-    {
-      "epoch": 7.600521512385919,
-      "grad_norm": 1.8229976892471313,
-      "learning_rate": 8.585929648241207e-05,
-      "loss": 5.5195,
-      "step": 14574
-    },
-    {
-      "epoch": 7.601043024771839,
-      "grad_norm": 1.413501262664795,
-      "learning_rate": 8.585829145728644e-05,
-      "loss": 5.7271,
-      "step": 14575
-    },
-    {
-      "epoch": 7.601564537157757,
-      "grad_norm": 1.420708179473877,
-      "learning_rate": 8.585728643216081e-05,
-      "loss": 5.3309,
-      "step": 14576
-    },
-    {
-      "epoch": 7.602086049543677,
-      "grad_norm": 1.5734912157058716,
-      "learning_rate": 8.585628140703518e-05,
-      "loss": 5.5062,
-      "step": 14577
-    },
-    {
-      "epoch": 7.602607561929596,
-      "grad_norm": 1.4779011011123657,
-      "learning_rate": 8.585527638190956e-05,
-      "loss": 5.5413,
-      "step": 14578
-    },
-    {
-      "epoch": 7.603129074315515,
-      "grad_norm": 1.727867603302002,
-      "learning_rate": 8.585427135678392e-05,
-      "loss": 5.3585,
-      "step": 14579
-    },
-    {
-      "epoch": 7.603650586701434,
-      "grad_norm": 1.494776964187622,
-      "learning_rate": 8.58532663316583e-05,
-      "loss": 5.5079,
-      "step": 14580
-    },
-    {
-      "epoch": 7.604172099087354,
-      "grad_norm": 1.4733668565750122,
-      "learning_rate": 8.585226130653266e-05,
-      "loss": 5.7606,
-      "step": 14581
-    },
-    {
-      "epoch": 7.604693611473272,
-      "grad_norm": 1.4775092601776123,
-      "learning_rate": 8.585125628140704e-05,
-      "loss": 5.0424,
-      "step": 14582
-    },
-    {
-      "epoch": 7.605215123859192,
-      "grad_norm": 1.3640263080596924,
-      "learning_rate": 8.58502512562814e-05,
-      "loss": 5.8514,
-      "step": 14583
-    },
-    {
-      "epoch": 7.605736636245111,
-      "grad_norm": 1.4299324750900269,
-      "learning_rate": 8.584924623115578e-05,
-      "loss": 5.5092,
-      "step": 14584
-    },
-    {
-      "epoch": 7.60625814863103,
-      "grad_norm": 1.3825961351394653,
-      "learning_rate": 8.584824120603016e-05,
-      "loss": 5.7907,
-      "step": 14585
-    },
-    {
-      "epoch": 7.606779661016949,
-      "grad_norm": 1.4047893285751343,
-      "learning_rate": 8.584723618090453e-05,
-      "loss": 5.266,
-      "step": 14586
-    },
-    {
-      "epoch": 7.607301173402869,
-      "grad_norm": 1.6981569528579712,
-      "learning_rate": 8.58462311557789e-05,
-      "loss": 5.5537,
-      "step": 14587
-    },
-    {
-      "epoch": 7.607822685788787,
-      "grad_norm": 1.4846986532211304,
-      "learning_rate": 8.584522613065327e-05,
-      "loss": 5.2103,
-      "step": 14588
-    },
-    {
-      "epoch": 7.608344198174707,
-      "grad_norm": 1.4243905544281006,
-      "learning_rate": 8.584422110552765e-05,
-      "loss": 5.3094,
-      "step": 14589
-    },
-    {
-      "epoch": 7.608865710560626,
-      "grad_norm": 1.459595799446106,
-      "learning_rate": 8.584321608040201e-05,
-      "loss": 5.7184,
-      "step": 14590
-    },
-    {
-      "epoch": 7.609387222946545,
-      "grad_norm": 1.4177128076553345,
-      "learning_rate": 8.584221105527639e-05,
-      "loss": 5.876,
-      "step": 14591
-    },
-    {
-      "epoch": 7.609908735332464,
-      "grad_norm": 1.4679104089736938,
-      "learning_rate": 8.584120603015075e-05,
-      "loss": 5.6221,
-      "step": 14592
-    },
-    {
-      "epoch": 7.610430247718384,
-      "grad_norm": 1.353390097618103,
-      "learning_rate": 8.584020100502513e-05,
-      "loss": 5.8467,
-      "step": 14593
-    },
-    {
-      "epoch": 7.610951760104302,
-      "grad_norm": 1.4778640270233154,
-      "learning_rate": 8.583919597989951e-05,
-      "loss": 5.1659,
-      "step": 14594
-    },
-    {
-      "epoch": 7.611473272490222,
-      "grad_norm": 1.3148934841156006,
-      "learning_rate": 8.583819095477388e-05,
-      "loss": 5.8654,
-      "step": 14595
-    },
-    {
-      "epoch": 7.611994784876141,
-      "grad_norm": 1.4188932180404663,
-      "learning_rate": 8.583718592964825e-05,
-      "loss": 5.2034,
-      "step": 14596
-    },
-    {
-      "epoch": 7.61251629726206,
-      "grad_norm": 1.5931352376937866,
-      "learning_rate": 8.583618090452261e-05,
-      "loss": 4.7301,
-      "step": 14597
-    },
-    {
-      "epoch": 7.613037809647979,
-      "grad_norm": 1.4532877206802368,
-      "learning_rate": 8.583517587939699e-05,
-      "loss": 5.9231,
-      "step": 14598
-    },
-    {
-      "epoch": 7.613559322033899,
-      "grad_norm": 1.4350519180297852,
-      "learning_rate": 8.583417085427136e-05,
-      "loss": 5.5897,
-      "step": 14599
-    },
-    {
-      "epoch": 7.614080834419817,
-      "grad_norm": 1.4797754287719727,
-      "learning_rate": 8.583316582914573e-05,
-      "loss": 5.016,
-      "step": 14600
-    },
-    {
-      "epoch": 7.614602346805737,
-      "grad_norm": 1.3961119651794434,
-      "learning_rate": 8.58321608040201e-05,
-      "loss": 5.5914,
-      "step": 14601
-    },
-    {
-      "epoch": 7.615123859191656,
-      "grad_norm": 1.3779646158218384,
-      "learning_rate": 8.583115577889448e-05,
-      "loss": 5.4804,
-      "step": 14602
-    },
-    {
-      "epoch": 7.615645371577575,
-      "grad_norm": 1.3455175161361694,
-      "learning_rate": 8.583015075376884e-05,
-      "loss": 5.4765,
-      "step": 14603
-    },
-    {
-      "epoch": 7.616166883963494,
-      "grad_norm": 1.3976656198501587,
-      "learning_rate": 8.582914572864322e-05,
-      "loss": 5.1494,
-      "step": 14604
-    },
-    {
-      "epoch": 7.616688396349414,
-      "grad_norm": 1.3625767230987549,
-      "learning_rate": 8.58281407035176e-05,
-      "loss": 5.6284,
-      "step": 14605
-    },
-    {
-      "epoch": 7.617209908735332,
-      "grad_norm": 1.4791330099105835,
-      "learning_rate": 8.582713567839197e-05,
-      "loss": 5.4715,
-      "step": 14606
-    },
-    {
-      "epoch": 7.617731421121252,
-      "grad_norm": 1.4508873224258423,
-      "learning_rate": 8.582613065326634e-05,
-      "loss": 5.4205,
-      "step": 14607
-    },
-    {
-      "epoch": 7.618252933507171,
-      "grad_norm": 1.375443935394287,
-      "learning_rate": 8.582512562814072e-05,
-      "loss": 5.6043,
-      "step": 14608
-    },
-    {
-      "epoch": 7.61877444589309,
-      "grad_norm": 1.3604334592819214,
-      "learning_rate": 8.582412060301508e-05,
-      "loss": 5.3671,
-      "step": 14609
-    },
-    {
-      "epoch": 7.619295958279009,
-      "grad_norm": 1.6541023254394531,
-      "learning_rate": 8.582311557788946e-05,
-      "loss": 5.0596,
-      "step": 14610
-    },
-    {
-      "epoch": 7.619817470664929,
-      "grad_norm": 1.3720474243164062,
-      "learning_rate": 8.582211055276382e-05,
-      "loss": 5.2987,
-      "step": 14611
-    },
-    {
-      "epoch": 7.620338983050847,
-      "grad_norm": 1.4350706338882446,
-      "learning_rate": 8.582110552763819e-05,
-      "loss": 5.6824,
-      "step": 14612
-    },
-    {
-      "epoch": 7.620860495436767,
-      "grad_norm": 1.3727245330810547,
-      "learning_rate": 8.582010050251256e-05,
-      "loss": 5.4298,
-      "step": 14613
-    },
-    {
-      "epoch": 7.621382007822686,
-      "grad_norm": 1.4559354782104492,
-      "learning_rate": 8.581909547738693e-05,
-      "loss": 5.5784,
-      "step": 14614
-    },
-    {
-      "epoch": 7.621903520208605,
-      "grad_norm": 1.4342036247253418,
-      "learning_rate": 8.58180904522613e-05,
-      "loss": 5.5907,
-      "step": 14615
-    },
-    {
-      "epoch": 7.622425032594524,
-      "grad_norm": 1.3768900632858276,
-      "learning_rate": 8.581708542713568e-05,
-      "loss": 5.6485,
-      "step": 14616
-    },
-    {
-      "epoch": 7.622946544980444,
-      "grad_norm": 1.4285547733306885,
-      "learning_rate": 8.581608040201006e-05,
-      "loss": 5.7578,
-      "step": 14617
-    },
-    {
-      "epoch": 7.623468057366362,
-      "grad_norm": 1.675321340560913,
-      "learning_rate": 8.581507537688443e-05,
-      "loss": 5.3942,
-      "step": 14618
-    },
-    {
-      "epoch": 7.623989569752282,
-      "grad_norm": 1.5953563451766968,
-      "learning_rate": 8.58140703517588e-05,
-      "loss": 4.6846,
-      "step": 14619
-    },
-    {
-      "epoch": 7.624511082138201,
-      "grad_norm": 1.4773523807525635,
-      "learning_rate": 8.581306532663317e-05,
-      "loss": 5.4585,
-      "step": 14620
-    },
-    {
-      "epoch": 7.62503259452412,
-      "grad_norm": 1.5057294368743896,
-      "learning_rate": 8.581206030150755e-05,
-      "loss": 5.4498,
-      "step": 14621
-    },
-    {
-      "epoch": 7.625554106910039,
-      "grad_norm": 1.4820457696914673,
-      "learning_rate": 8.581105527638191e-05,
-      "loss": 5.6534,
-      "step": 14622
-    },
-    {
-      "epoch": 7.626075619295959,
-      "grad_norm": 1.4526172876358032,
-      "learning_rate": 8.581005025125629e-05,
-      "loss": 5.552,
-      "step": 14623
-    },
-    {
-      "epoch": 7.626597131681877,
-      "grad_norm": 1.4059566259384155,
-      "learning_rate": 8.580904522613065e-05,
-      "loss": 5.421,
-      "step": 14624
-    },
-    {
-      "epoch": 7.627118644067797,
-      "grad_norm": 1.3826589584350586,
-      "learning_rate": 8.580804020100503e-05,
-      "loss": 5.772,
-      "step": 14625
-    },
-    {
-      "epoch": 7.627640156453715,
-      "grad_norm": 1.397700548171997,
-      "learning_rate": 8.580703517587941e-05,
-      "loss": 5.7374,
-      "step": 14626
-    },
-    {
-      "epoch": 7.628161668839635,
-      "grad_norm": 1.5908472537994385,
-      "learning_rate": 8.580603015075377e-05,
-      "loss": 5.1126,
-      "step": 14627
-    },
-    {
-      "epoch": 7.628683181225554,
-      "grad_norm": 1.555870771408081,
-      "learning_rate": 8.580502512562815e-05,
-      "loss": 5.3093,
-      "step": 14628
-    },
-    {
-      "epoch": 7.629204693611474,
-      "grad_norm": 1.5971022844314575,
-      "learning_rate": 8.580402010050251e-05,
-      "loss": 5.5678,
-      "step": 14629
-    },
-    {
-      "epoch": 7.629726205997392,
-      "grad_norm": 1.4287526607513428,
-      "learning_rate": 8.580301507537689e-05,
-      "loss": 5.5924,
-      "step": 14630
-    },
-    {
-      "epoch": 7.630247718383312,
-      "grad_norm": 1.3843683004379272,
-      "learning_rate": 8.580201005025126e-05,
-      "loss": 5.6856,
-      "step": 14631
-    },
-    {
-      "epoch": 7.63076923076923,
-      "grad_norm": 1.3418704271316528,
-      "learning_rate": 8.580100502512563e-05,
-      "loss": 5.2955,
-      "step": 14632
-    },
-    {
-      "epoch": 7.63129074315515,
-      "grad_norm": 1.382317066192627,
-      "learning_rate": 8.58e-05,
-      "loss": 5.4985,
-      "step": 14633
-    },
-    {
-      "epoch": 7.631812255541069,
-      "grad_norm": 1.432120680809021,
-      "learning_rate": 8.579899497487438e-05,
-      "loss": 5.3769,
-      "step": 14634
-    },
-    {
-      "epoch": 7.632333767926989,
-      "grad_norm": 1.3869717121124268,
-      "learning_rate": 8.579798994974874e-05,
-      "loss": 5.8852,
-      "step": 14635
-    },
-    {
-      "epoch": 7.632855280312907,
-      "grad_norm": 1.2733527421951294,
-      "learning_rate": 8.579698492462312e-05,
-      "loss": 5.9152,
-      "step": 14636
-    },
-    {
-      "epoch": 7.633376792698827,
-      "grad_norm": 1.4165894985198975,
-      "learning_rate": 8.57959798994975e-05,
-      "loss": 5.8174,
-      "step": 14637
-    },
-    {
-      "epoch": 7.633898305084745,
-      "grad_norm": 1.4351458549499512,
-      "learning_rate": 8.579497487437186e-05,
-      "loss": 5.8018,
-      "step": 14638
-    },
-    {
-      "epoch": 7.634419817470665,
-      "grad_norm": 1.3981125354766846,
-      "learning_rate": 8.579396984924624e-05,
-      "loss": 5.7204,
-      "step": 14639
-    },
-    {
-      "epoch": 7.634941329856584,
-      "grad_norm": 1.5012327432632446,
-      "learning_rate": 8.57929648241206e-05,
-      "loss": 5.7069,
-      "step": 14640
-    },
-    {
-      "epoch": 7.635462842242504,
-      "grad_norm": 1.5171332359313965,
-      "learning_rate": 8.579195979899498e-05,
-      "loss": 5.3609,
-      "step": 14641
-    },
-    {
-      "epoch": 7.635984354628422,
-      "grad_norm": 1.4870562553405762,
-      "learning_rate": 8.579095477386934e-05,
-      "loss": 5.685,
-      "step": 14642
-    },
-    {
-      "epoch": 7.636505867014342,
-      "grad_norm": 1.4352293014526367,
-      "learning_rate": 8.578994974874372e-05,
-      "loss": 5.6263,
-      "step": 14643
-    },
-    {
-      "epoch": 7.63702737940026,
-      "grad_norm": 1.4950265884399414,
-      "learning_rate": 8.578894472361809e-05,
-      "loss": 5.5199,
-      "step": 14644
-    },
-    {
-      "epoch": 7.63754889178618,
-      "grad_norm": 1.4033253192901611,
-      "learning_rate": 8.578793969849246e-05,
-      "loss": 5.3981,
-      "step": 14645
-    },
-    {
-      "epoch": 7.638070404172099,
-      "grad_norm": 1.4909754991531372,
-      "learning_rate": 8.578693467336684e-05,
-      "loss": 5.5713,
-      "step": 14646
-    },
-    {
-      "epoch": 7.638591916558019,
-      "grad_norm": 1.42000150680542,
-      "learning_rate": 8.578592964824122e-05,
-      "loss": 5.6759,
-      "step": 14647
-    },
-    {
-      "epoch": 7.639113428943937,
-      "grad_norm": 1.503796100616455,
-      "learning_rate": 8.578492462311558e-05,
-      "loss": 5.5221,
-      "step": 14648
-    },
-    {
-      "epoch": 7.639634941329857,
-      "grad_norm": 1.4921307563781738,
-      "learning_rate": 8.578391959798996e-05,
-      "loss": 5.2454,
-      "step": 14649
-    },
-    {
-      "epoch": 7.640156453715775,
-      "grad_norm": 1.5497536659240723,
-      "learning_rate": 8.578291457286433e-05,
-      "loss": 5.7581,
-      "step": 14650
-    },
-    {
-      "epoch": 7.640677966101695,
-      "grad_norm": 1.3489325046539307,
-      "learning_rate": 8.578190954773869e-05,
-      "loss": 5.5266,
-      "step": 14651
-    },
-    {
-      "epoch": 7.641199478487614,
-      "grad_norm": 1.3150960206985474,
-      "learning_rate": 8.578090452261307e-05,
-      "loss": 5.7612,
-      "step": 14652
-    },
-    {
-      "epoch": 7.641720990873534,
-      "grad_norm": 1.3955278396606445,
-      "learning_rate": 8.577989949748743e-05,
-      "loss": 5.5386,
-      "step": 14653
-    },
-    {
-      "epoch": 7.642242503259452,
-      "grad_norm": 1.4535070657730103,
-      "learning_rate": 8.577889447236181e-05,
-      "loss": 5.7066,
-      "step": 14654
-    },
-    {
-      "epoch": 7.642764015645372,
-      "grad_norm": 1.4087462425231934,
-      "learning_rate": 8.577788944723618e-05,
-      "loss": 5.8026,
-      "step": 14655
-    },
-    {
-      "epoch": 7.64328552803129,
-      "grad_norm": 1.3564918041229248,
-      "learning_rate": 8.577688442211055e-05,
-      "loss": 5.2437,
-      "step": 14656
-    },
-    {
-      "epoch": 7.64380704041721,
-      "grad_norm": 1.5357944965362549,
-      "learning_rate": 8.577587939698493e-05,
-      "loss": 5.0253,
-      "step": 14657
-    },
-    {
-      "epoch": 7.644328552803129,
-      "grad_norm": 1.5106899738311768,
-      "learning_rate": 8.577487437185931e-05,
-      "loss": 5.565,
-      "step": 14658
-    },
-    {
-      "epoch": 7.644850065189048,
-      "grad_norm": 1.4094544649124146,
-      "learning_rate": 8.577386934673367e-05,
-      "loss": 5.2372,
-      "step": 14659
-    },
-    {
-      "epoch": 7.645371577574967,
-      "grad_norm": 1.3909111022949219,
-      "learning_rate": 8.577286432160805e-05,
-      "loss": 5.9288,
-      "step": 14660
-    },
-    {
-      "epoch": 7.645893089960887,
-      "grad_norm": 1.3765878677368164,
-      "learning_rate": 8.577185929648241e-05,
-      "loss": 5.8483,
-      "step": 14661
-    },
-    {
-      "epoch": 7.646414602346805,
-      "grad_norm": 1.544650912284851,
-      "learning_rate": 8.577085427135679e-05,
-      "loss": 5.2543,
-      "step": 14662
-    },
-    {
-      "epoch": 7.646936114732725,
-      "grad_norm": 1.6233195066452026,
-      "learning_rate": 8.576984924623116e-05,
-      "loss": 5.7661,
-      "step": 14663
-    },
-    {
-      "epoch": 7.647457627118644,
-      "grad_norm": 1.4827147722244263,
-      "learning_rate": 8.576884422110552e-05,
-      "loss": 5.5571,
-      "step": 14664
-    },
-    {
-      "epoch": 7.647979139504563,
-      "grad_norm": 1.4367491006851196,
-      "learning_rate": 8.57678391959799e-05,
-      "loss": 5.6647,
-      "step": 14665
-    },
-    {
-      "epoch": 7.648500651890482,
-      "grad_norm": 1.475691318511963,
-      "learning_rate": 8.576683417085428e-05,
-      "loss": 5.2522,
-      "step": 14666
-    },
-    {
-      "epoch": 7.649022164276402,
-      "grad_norm": 1.486930012702942,
-      "learning_rate": 8.576582914572865e-05,
-      "loss": 5.5879,
-      "step": 14667
-    },
-    {
-      "epoch": 7.64954367666232,
-      "grad_norm": 1.3296513557434082,
-      "learning_rate": 8.576482412060302e-05,
-      "loss": 5.947,
-      "step": 14668
-    },
-    {
-      "epoch": 7.65006518904824,
-      "grad_norm": 1.3885064125061035,
-      "learning_rate": 8.57638190954774e-05,
-      "loss": 5.8143,
-      "step": 14669
-    },
-    {
-      "epoch": 7.650586701434159,
-      "grad_norm": 1.4947150945663452,
-      "learning_rate": 8.576281407035176e-05,
-      "loss": 5.7642,
-      "step": 14670
-    },
-    {
-      "epoch": 7.651108213820078,
-      "grad_norm": 1.5866930484771729,
-      "learning_rate": 8.576180904522614e-05,
-      "loss": 5.2412,
-      "step": 14671
-    },
-    {
-      "epoch": 7.651629726205997,
-      "grad_norm": 1.5581918954849243,
-      "learning_rate": 8.57608040201005e-05,
-      "loss": 4.9765,
-      "step": 14672
-    },
-    {
-      "epoch": 7.652151238591917,
-      "grad_norm": 1.4363638162612915,
-      "learning_rate": 8.575979899497488e-05,
-      "loss": 5.7409,
-      "step": 14673
-    },
-    {
-      "epoch": 7.652672750977835,
-      "grad_norm": 1.4252647161483765,
-      "learning_rate": 8.575879396984925e-05,
-      "loss": 5.6977,
-      "step": 14674
-    },
-    {
-      "epoch": 7.653194263363755,
-      "grad_norm": 1.4991096258163452,
-      "learning_rate": 8.575778894472362e-05,
-      "loss": 6.0273,
-      "step": 14675
-    },
-    {
-      "epoch": 7.653715775749674,
-      "grad_norm": 1.421834111213684,
-      "learning_rate": 8.575678391959799e-05,
-      "loss": 5.0179,
-      "step": 14676
-    },
-    {
-      "epoch": 7.654237288135593,
-      "grad_norm": 1.4294553995132446,
-      "learning_rate": 8.575577889447237e-05,
-      "loss": 5.4072,
-      "step": 14677
-    },
-    {
-      "epoch": 7.654758800521512,
-      "grad_norm": 1.4207764863967896,
-      "learning_rate": 8.575477386934674e-05,
-      "loss": 5.2973,
-      "step": 14678
-    },
-    {
-      "epoch": 7.655280312907432,
-      "grad_norm": 1.3687636852264404,
-      "learning_rate": 8.575376884422111e-05,
-      "loss": 5.4252,
-      "step": 14679
-    },
-    {
-      "epoch": 7.65580182529335,
-      "grad_norm": 1.4813637733459473,
-      "learning_rate": 8.575276381909549e-05,
-      "loss": 5.8017,
-      "step": 14680
-    },
-    {
-      "epoch": 7.65632333767927,
-      "grad_norm": 1.5380475521087646,
-      "learning_rate": 8.575175879396985e-05,
-      "loss": 5.5064,
-      "step": 14681
-    },
-    {
-      "epoch": 7.656844850065189,
-      "grad_norm": 1.491517424583435,
-      "learning_rate": 8.575075376884423e-05,
-      "loss": 5.3233,
-      "step": 14682
-    },
-    {
-      "epoch": 7.657366362451108,
-      "grad_norm": 1.4311333894729614,
-      "learning_rate": 8.574974874371859e-05,
-      "loss": 5.3413,
-      "step": 14683
-    },
-    {
-      "epoch": 7.657887874837027,
-      "grad_norm": 1.3998380899429321,
-      "learning_rate": 8.574874371859297e-05,
-      "loss": 5.5189,
-      "step": 14684
-    },
-    {
-      "epoch": 7.658409387222947,
-      "grad_norm": 1.292834997177124,
-      "learning_rate": 8.574773869346733e-05,
-      "loss": 5.8297,
-      "step": 14685
-    },
-    {
-      "epoch": 7.658930899608865,
-      "grad_norm": 1.4793167114257812,
-      "learning_rate": 8.574673366834171e-05,
-      "loss": 6.0114,
-      "step": 14686
-    },
-    {
-      "epoch": 7.659452411994785,
-      "grad_norm": 1.5299097299575806,
-      "learning_rate": 8.574572864321609e-05,
-      "loss": 5.881,
-      "step": 14687
-    },
-    {
-      "epoch": 7.659973924380704,
-      "grad_norm": 1.321669340133667,
-      "learning_rate": 8.574472361809047e-05,
-      "loss": 5.3794,
-      "step": 14688
-    },
-    {
-      "epoch": 7.660495436766623,
-      "grad_norm": 1.3716124296188354,
-      "learning_rate": 8.574371859296483e-05,
-      "loss": 5.6012,
-      "step": 14689
-    },
-    {
-      "epoch": 7.661016949152542,
-      "grad_norm": 1.4647706747055054,
-      "learning_rate": 8.57427135678392e-05,
-      "loss": 5.5811,
-      "step": 14690
-    },
-    {
-      "epoch": 7.661538461538462,
-      "grad_norm": 1.3908463716506958,
-      "learning_rate": 8.574170854271357e-05,
-      "loss": 5.8113,
-      "step": 14691
-    },
-    {
-      "epoch": 7.66205997392438,
-      "grad_norm": 1.5413529872894287,
-      "learning_rate": 8.574070351758794e-05,
-      "loss": 5.652,
-      "step": 14692
-    },
-    {
-      "epoch": 7.6625814863103,
-      "grad_norm": 1.4689393043518066,
-      "learning_rate": 8.573969849246232e-05,
-      "loss": 5.8838,
-      "step": 14693
-    },
-    {
-      "epoch": 7.663102998696219,
-      "grad_norm": 1.5362683534622192,
-      "learning_rate": 8.573869346733668e-05,
-      "loss": 4.8633,
-      "step": 14694
-    },
-    {
-      "epoch": 7.663624511082138,
-      "grad_norm": 1.3247417211532593,
-      "learning_rate": 8.573768844221106e-05,
-      "loss": 5.7853,
-      "step": 14695
-    },
-    {
-      "epoch": 7.664146023468057,
-      "grad_norm": 1.548058271408081,
-      "learning_rate": 8.573668341708542e-05,
-      "loss": 4.7817,
-      "step": 14696
-    },
-    {
-      "epoch": 7.664667535853977,
-      "grad_norm": 1.520589828491211,
-      "learning_rate": 8.57356783919598e-05,
-      "loss": 5.7826,
-      "step": 14697
-    },
-    {
-      "epoch": 7.665189048239895,
-      "grad_norm": 1.7015143632888794,
-      "learning_rate": 8.573467336683418e-05,
-      "loss": 5.5699,
-      "step": 14698
-    },
-    {
-      "epoch": 7.665710560625815,
-      "grad_norm": 1.4458372592926025,
-      "learning_rate": 8.573366834170856e-05,
-      "loss": 5.3857,
-      "step": 14699
-    },
-    {
-      "epoch": 7.666232073011734,
-      "grad_norm": 1.6074435710906982,
-      "learning_rate": 8.573266331658292e-05,
-      "loss": 5.8484,
-      "step": 14700
-    },
-    {
-      "epoch": 7.666753585397653,
-      "grad_norm": 1.624144434928894,
-      "learning_rate": 8.57316582914573e-05,
-      "loss": 5.8928,
-      "step": 14701
-    },
-    {
-      "epoch": 7.667275097783572,
-      "grad_norm": 1.5539202690124512,
-      "learning_rate": 8.573065326633166e-05,
-      "loss": 5.3345,
-      "step": 14702
-    },
-    {
-      "epoch": 7.667796610169492,
-      "grad_norm": 1.6379830837249756,
-      "learning_rate": 8.572964824120604e-05,
-      "loss": 5.0391,
-      "step": 14703
-    },
-    {
-      "epoch": 7.6683181225554105,
-      "grad_norm": 1.4335107803344727,
-      "learning_rate": 8.57286432160804e-05,
-      "loss": 5.6575,
-      "step": 14704
-    },
-    {
-      "epoch": 7.66883963494133,
-      "grad_norm": 1.4010604619979858,
-      "learning_rate": 8.572763819095477e-05,
-      "loss": 5.3618,
-      "step": 14705
-    },
-    {
-      "epoch": 7.669361147327249,
-      "grad_norm": 1.437680721282959,
-      "learning_rate": 8.572663316582915e-05,
-      "loss": 5.0952,
-      "step": 14706
-    },
-    {
-      "epoch": 7.669882659713168,
-      "grad_norm": 1.45354163646698,
-      "learning_rate": 8.572562814070352e-05,
-      "loss": 5.221,
-      "step": 14707
-    },
-    {
-      "epoch": 7.670404172099087,
-      "grad_norm": 1.4506324529647827,
-      "learning_rate": 8.57246231155779e-05,
-      "loss": 5.2993,
-      "step": 14708
-    },
-    {
-      "epoch": 7.670925684485007,
-      "grad_norm": 1.4666541814804077,
-      "learning_rate": 8.572361809045227e-05,
-      "loss": 5.5502,
-      "step": 14709
-    },
-    {
-      "epoch": 7.6714471968709255,
-      "grad_norm": 1.4368038177490234,
-      "learning_rate": 8.572261306532664e-05,
-      "loss": 5.347,
-      "step": 14710
-    },
-    {
-      "epoch": 7.671968709256845,
-      "grad_norm": 1.4859592914581299,
-      "learning_rate": 8.572160804020101e-05,
-      "loss": 5.4601,
-      "step": 14711
-    },
-    {
-      "epoch": 7.672490221642764,
-      "grad_norm": 1.5199806690216064,
-      "learning_rate": 8.572060301507539e-05,
-      "loss": 5.7809,
-      "step": 14712
-    },
-    {
-      "epoch": 7.673011734028683,
-      "grad_norm": 1.3522474765777588,
-      "learning_rate": 8.571959798994975e-05,
-      "loss": 5.8287,
-      "step": 14713
-    },
-    {
-      "epoch": 7.673533246414602,
-      "grad_norm": 1.5956192016601562,
-      "learning_rate": 8.571859296482413e-05,
-      "loss": 5.4111,
-      "step": 14714
-    },
-    {
-      "epoch": 7.674054758800521,
-      "grad_norm": 1.495283603668213,
-      "learning_rate": 8.571758793969849e-05,
-      "loss": 5.7645,
-      "step": 14715
-    },
-    {
-      "epoch": 7.6745762711864405,
-      "grad_norm": 1.3905409574508667,
-      "learning_rate": 8.571658291457287e-05,
-      "loss": 5.7638,
-      "step": 14716
-    },
-    {
-      "epoch": 7.67509778357236,
-      "grad_norm": 1.4557733535766602,
-      "learning_rate": 8.571557788944723e-05,
-      "loss": 5.4745,
-      "step": 14717
-    },
-    {
-      "epoch": 7.675619295958279,
-      "grad_norm": 1.392626404762268,
-      "learning_rate": 8.571457286432161e-05,
-      "loss": 5.4911,
-      "step": 14718
-    },
-    {
-      "epoch": 7.676140808344198,
-      "grad_norm": 1.3730162382125854,
-      "learning_rate": 8.571356783919599e-05,
-      "loss": 5.8842,
-      "step": 14719
-    },
-    {
-      "epoch": 7.6766623207301175,
-      "grad_norm": 1.467215895652771,
-      "learning_rate": 8.571256281407035e-05,
-      "loss": 5.5235,
-      "step": 14720
-    },
-    {
-      "epoch": 7.677183833116036,
-      "grad_norm": 1.4114973545074463,
-      "learning_rate": 8.571155778894473e-05,
-      "loss": 5.7143,
-      "step": 14721
-    },
-    {
-      "epoch": 7.6777053455019555,
-      "grad_norm": 1.3557264804840088,
-      "learning_rate": 8.57105527638191e-05,
-      "loss": 5.5797,
-      "step": 14722
-    },
-    {
-      "epoch": 7.678226857887875,
-      "grad_norm": 1.39503812789917,
-      "learning_rate": 8.570954773869347e-05,
-      "loss": 5.5436,
-      "step": 14723
-    },
-    {
-      "epoch": 7.678748370273794,
-      "grad_norm": 1.5187641382217407,
-      "learning_rate": 8.570854271356784e-05,
-      "loss": 5.6563,
-      "step": 14724
-    },
-    {
-      "epoch": 7.679269882659713,
-      "grad_norm": 1.4089620113372803,
-      "learning_rate": 8.570753768844222e-05,
-      "loss": 5.5209,
-      "step": 14725
-    },
-    {
-      "epoch": 7.6797913950456325,
-      "grad_norm": 1.508353352546692,
-      "learning_rate": 8.570653266331658e-05,
-      "loss": 5.2482,
-      "step": 14726
-    },
-    {
-      "epoch": 7.680312907431551,
-      "grad_norm": 1.4408546686172485,
-      "learning_rate": 8.570552763819096e-05,
-      "loss": 6.01,
-      "step": 14727
-    },
-    {
-      "epoch": 7.6808344198174705,
-      "grad_norm": 1.4495387077331543,
-      "learning_rate": 8.570452261306534e-05,
-      "loss": 5.5215,
-      "step": 14728
-    },
-    {
-      "epoch": 7.68135593220339,
-      "grad_norm": 1.3931432962417603,
-      "learning_rate": 8.570351758793971e-05,
-      "loss": 5.3699,
-      "step": 14729
-    },
-    {
-      "epoch": 7.681877444589309,
-      "grad_norm": 1.4001067876815796,
-      "learning_rate": 8.570251256281408e-05,
-      "loss": 5.0847,
-      "step": 14730
-    },
-    {
-      "epoch": 7.682398956975228,
-      "grad_norm": 1.3747936487197876,
-      "learning_rate": 8.570150753768844e-05,
-      "loss": 5.6811,
-      "step": 14731
-    },
-    {
-      "epoch": 7.6829204693611475,
-      "grad_norm": 1.3512450456619263,
-      "learning_rate": 8.570050251256282e-05,
-      "loss": 5.6691,
-      "step": 14732
-    },
-    {
-      "epoch": 7.683441981747066,
-      "grad_norm": 1.45374596118927,
-      "learning_rate": 8.569949748743718e-05,
-      "loss": 5.0967,
-      "step": 14733
-    },
-    {
-      "epoch": 7.6839634941329855,
-      "grad_norm": 1.5682382583618164,
-      "learning_rate": 8.569849246231156e-05,
-      "loss": 5.071,
-      "step": 14734
-    },
-    {
-      "epoch": 7.684485006518905,
-      "grad_norm": 1.4758872985839844,
-      "learning_rate": 8.569748743718593e-05,
-      "loss": 5.8897,
-      "step": 14735
-    },
-    {
-      "epoch": 7.6850065189048244,
-      "grad_norm": 1.4261268377304077,
-      "learning_rate": 8.56964824120603e-05,
-      "loss": 5.4662,
-      "step": 14736
-    },
-    {
-      "epoch": 7.685528031290743,
-      "grad_norm": 1.362537145614624,
-      "learning_rate": 8.569547738693467e-05,
-      "loss": 5.7164,
-      "step": 14737
-    },
-    {
-      "epoch": 7.6860495436766625,
-      "grad_norm": 1.335270643234253,
-      "learning_rate": 8.569447236180905e-05,
-      "loss": 5.755,
-      "step": 14738
-    },
-    {
-      "epoch": 7.686571056062581,
-      "grad_norm": 1.4678642749786377,
-      "learning_rate": 8.569346733668342e-05,
-      "loss": 5.8006,
-      "step": 14739
-    },
-    {
-      "epoch": 7.6870925684485005,
-      "grad_norm": 1.2940086126327515,
-      "learning_rate": 8.56924623115578e-05,
-      "loss": 6.0066,
-      "step": 14740
-    },
-    {
-      "epoch": 7.68761408083442,
-      "grad_norm": 1.4905263185501099,
-      "learning_rate": 8.569145728643217e-05,
-      "loss": 5.6827,
-      "step": 14741
-    },
-    {
-      "epoch": 7.6881355932203395,
-      "grad_norm": 1.500954031944275,
-      "learning_rate": 8.569045226130654e-05,
-      "loss": 5.2072,
-      "step": 14742
-    },
-    {
-      "epoch": 7.688657105606258,
-      "grad_norm": 1.471570372581482,
-      "learning_rate": 8.568944723618091e-05,
-      "loss": 5.8793,
-      "step": 14743
-    },
-    {
-      "epoch": 7.6891786179921775,
-      "grad_norm": 1.6480016708374023,
-      "learning_rate": 8.568844221105527e-05,
-      "loss": 5.3504,
-      "step": 14744
-    },
-    {
-      "epoch": 7.689700130378096,
-      "grad_norm": 1.4533756971359253,
-      "learning_rate": 8.568743718592965e-05,
-      "loss": 5.3577,
-      "step": 14745
-    },
-    {
-      "epoch": 7.6902216427640155,
-      "grad_norm": 1.4899380207061768,
-      "learning_rate": 8.568643216080402e-05,
-      "loss": 5.2142,
-      "step": 14746
-    },
-    {
-      "epoch": 7.690743155149935,
-      "grad_norm": 1.3690046072006226,
-      "learning_rate": 8.568542713567839e-05,
-      "loss": 5.5417,
-      "step": 14747
-    },
-    {
-      "epoch": 7.6912646675358545,
-      "grad_norm": 1.3653653860092163,
-      "learning_rate": 8.568442211055277e-05,
-      "loss": 5.3554,
-      "step": 14748
-    },
-    {
-      "epoch": 7.691786179921773,
-      "grad_norm": 1.3265948295593262,
-      "learning_rate": 8.568341708542715e-05,
-      "loss": 5.7507,
-      "step": 14749
-    },
-    {
-      "epoch": 7.6923076923076925,
-      "grad_norm": 1.4186019897460938,
-      "learning_rate": 8.568241206030151e-05,
-      "loss": 5.7338,
-      "step": 14750
-    },
-    {
-      "epoch": 7.692829204693611,
-      "grad_norm": 1.8793933391571045,
-      "learning_rate": 8.568140703517589e-05,
-      "loss": 4.9948,
-      "step": 14751
-    },
-    {
-      "epoch": 7.6933507170795306,
-      "grad_norm": 1.34813392162323,
-      "learning_rate": 8.568040201005026e-05,
-      "loss": 5.4168,
-      "step": 14752
-    },
-    {
-      "epoch": 7.69387222946545,
-      "grad_norm": 1.4172838926315308,
-      "learning_rate": 8.567939698492463e-05,
-      "loss": 5.6237,
-      "step": 14753
-    },
-    {
-      "epoch": 7.694393741851369,
-      "grad_norm": 1.3378936052322388,
-      "learning_rate": 8.5678391959799e-05,
-      "loss": 5.5808,
-      "step": 14754
-    },
-    {
-      "epoch": 7.694915254237288,
-      "grad_norm": 1.472091794013977,
-      "learning_rate": 8.567738693467338e-05,
-      "loss": 5.2642,
-      "step": 14755
-    },
-    {
-      "epoch": 7.6954367666232075,
-      "grad_norm": 1.5203274488449097,
-      "learning_rate": 8.567638190954774e-05,
-      "loss": 5.643,
-      "step": 14756
-    },
-    {
-      "epoch": 7.695958279009126,
-      "grad_norm": 1.6018379926681519,
-      "learning_rate": 8.56753768844221e-05,
-      "loss": 4.9829,
-      "step": 14757
-    },
-    {
-      "epoch": 7.696479791395046,
-      "grad_norm": 1.3405405282974243,
-      "learning_rate": 8.567437185929648e-05,
-      "loss": 5.711,
-      "step": 14758
-    },
-    {
-      "epoch": 7.697001303780965,
-      "grad_norm": 1.4114539623260498,
-      "learning_rate": 8.567336683417086e-05,
-      "loss": 5.7309,
-      "step": 14759
-    },
-    {
-      "epoch": 7.697522816166884,
-      "grad_norm": 1.4446299076080322,
-      "learning_rate": 8.567236180904524e-05,
-      "loss": 5.6742,
-      "step": 14760
-    },
-    {
-      "epoch": 7.698044328552803,
-      "grad_norm": 1.5071051120758057,
-      "learning_rate": 8.56713567839196e-05,
-      "loss": 5.8511,
-      "step": 14761
-    },
-    {
-      "epoch": 7.6985658409387225,
-      "grad_norm": 1.4111144542694092,
-      "learning_rate": 8.567035175879398e-05,
-      "loss": 5.9173,
-      "step": 14762
-    },
-    {
-      "epoch": 7.699087353324641,
-      "grad_norm": 1.6788824796676636,
-      "learning_rate": 8.566934673366834e-05,
-      "loss": 5.2343,
-      "step": 14763
-    },
-    {
-      "epoch": 7.699608865710561,
-      "grad_norm": 1.4079809188842773,
-      "learning_rate": 8.566834170854272e-05,
-      "loss": 5.7582,
-      "step": 14764
-    },
-    {
-      "epoch": 7.70013037809648,
-      "grad_norm": 1.4450013637542725,
-      "learning_rate": 8.566733668341709e-05,
-      "loss": 5.3325,
-      "step": 14765
-    },
-    {
-      "epoch": 7.700651890482399,
-      "grad_norm": 1.4873801469802856,
-      "learning_rate": 8.566633165829146e-05,
-      "loss": 5.4086,
-      "step": 14766
-    },
-    {
-      "epoch": 7.701173402868318,
-      "grad_norm": 1.3587675094604492,
-      "learning_rate": 8.566532663316583e-05,
-      "loss": 5.6883,
-      "step": 14767
-    },
-    {
-      "epoch": 7.7016949152542376,
-      "grad_norm": 1.3025877475738525,
-      "learning_rate": 8.56643216080402e-05,
-      "loss": 5.8786,
-      "step": 14768
-    },
-    {
-      "epoch": 7.702216427640156,
-      "grad_norm": 1.4709336757659912,
-      "learning_rate": 8.566331658291458e-05,
-      "loss": 5.6206,
-      "step": 14769
-    },
-    {
-      "epoch": 7.702737940026076,
-      "grad_norm": 1.413230299949646,
-      "learning_rate": 8.566231155778895e-05,
-      "loss": 5.6284,
-      "step": 14770
-    },
-    {
-      "epoch": 7.703259452411995,
-      "grad_norm": 1.6142833232879639,
-      "learning_rate": 8.566130653266333e-05,
-      "loss": 5.1408,
-      "step": 14771
-    },
-    {
-      "epoch": 7.703780964797914,
-      "grad_norm": 1.6423684358596802,
-      "learning_rate": 8.566030150753769e-05,
-      "loss": 5.6306,
-      "step": 14772
-    },
-    {
-      "epoch": 7.704302477183833,
-      "grad_norm": 1.535393476486206,
-      "learning_rate": 8.565929648241207e-05,
-      "loss": 5.4607,
-      "step": 14773
-    },
-    {
-      "epoch": 7.704823989569753,
-      "grad_norm": 1.5291744470596313,
-      "learning_rate": 8.565829145728643e-05,
-      "loss": 5.6793,
-      "step": 14774
-    },
-    {
-      "epoch": 7.705345501955671,
-      "grad_norm": 1.39779794216156,
-      "learning_rate": 8.565728643216081e-05,
-      "loss": 5.811,
-      "step": 14775
-    },
-    {
-      "epoch": 7.705867014341591,
-      "grad_norm": 1.5096255540847778,
-      "learning_rate": 8.565628140703517e-05,
-      "loss": 5.6343,
-      "step": 14776
-    },
-    {
-      "epoch": 7.70638852672751,
-      "grad_norm": 1.4525707960128784,
-      "learning_rate": 8.565527638190955e-05,
-      "loss": 5.6238,
-      "step": 14777
-    },
-    {
-      "epoch": 7.706910039113429,
-      "grad_norm": 1.4152907133102417,
-      "learning_rate": 8.565427135678392e-05,
-      "loss": 5.1513,
-      "step": 14778
-    },
-    {
-      "epoch": 7.707431551499348,
-      "grad_norm": 1.3965309858322144,
-      "learning_rate": 8.56532663316583e-05,
-      "loss": 5.8588,
-      "step": 14779
-    },
-    {
-      "epoch": 7.707953063885268,
-      "grad_norm": 1.5874367952346802,
-      "learning_rate": 8.565226130653267e-05,
-      "loss": 5.3228,
-      "step": 14780
-    },
-    {
-      "epoch": 7.708474576271186,
-      "grad_norm": 1.5047407150268555,
-      "learning_rate": 8.565125628140705e-05,
-      "loss": 4.8841,
-      "step": 14781
-    },
-    {
-      "epoch": 7.708996088657106,
-      "grad_norm": 1.4959686994552612,
-      "learning_rate": 8.565025125628141e-05,
-      "loss": 5.604,
-      "step": 14782
-    },
-    {
-      "epoch": 7.709517601043025,
-      "grad_norm": 1.529354453086853,
-      "learning_rate": 8.564924623115578e-05,
-      "loss": 5.3697,
-      "step": 14783
-    },
-    {
-      "epoch": 7.710039113428944,
-      "grad_norm": 1.5478074550628662,
-      "learning_rate": 8.564824120603016e-05,
-      "loss": 5.3552,
-      "step": 14784
-    },
-    {
-      "epoch": 7.710560625814863,
-      "grad_norm": 1.5828460454940796,
-      "learning_rate": 8.564723618090452e-05,
-      "loss": 5.9121,
-      "step": 14785
-    },
-    {
-      "epoch": 7.711082138200783,
-      "grad_norm": 1.5950356721878052,
-      "learning_rate": 8.56462311557789e-05,
-      "loss": 5.6399,
-      "step": 14786
-    },
-    {
-      "epoch": 7.711603650586701,
-      "grad_norm": 1.4942759275436401,
-      "learning_rate": 8.564522613065326e-05,
-      "loss": 5.7578,
-      "step": 14787
-    },
-    {
-      "epoch": 7.712125162972621,
-      "grad_norm": 1.3950265645980835,
-      "learning_rate": 8.564422110552764e-05,
-      "loss": 5.5435,
-      "step": 14788
-    },
-    {
-      "epoch": 7.71264667535854,
-      "grad_norm": 1.3896054029464722,
-      "learning_rate": 8.5643216080402e-05,
-      "loss": 5.8613,
-      "step": 14789
-    },
-    {
-      "epoch": 7.713168187744459,
-      "grad_norm": 1.562682032585144,
-      "learning_rate": 8.564221105527638e-05,
-      "loss": 5.1252,
-      "step": 14790
-    },
-    {
-      "epoch": 7.713689700130378,
-      "grad_norm": 1.450722575187683,
-      "learning_rate": 8.564120603015076e-05,
-      "loss": 5.37,
-      "step": 14791
-    },
-    {
-      "epoch": 7.714211212516298,
-      "grad_norm": 1.90537691116333,
-      "learning_rate": 8.564020100502514e-05,
-      "loss": 5.0483,
-      "step": 14792
-    },
-    {
-      "epoch": 7.714732724902216,
-      "grad_norm": 1.4214389324188232,
-      "learning_rate": 8.56391959798995e-05,
-      "loss": 4.7918,
-      "step": 14793
-    },
-    {
-      "epoch": 7.715254237288136,
-      "grad_norm": 1.469675898551941,
-      "learning_rate": 8.563819095477388e-05,
-      "loss": 5.4333,
-      "step": 14794
-    },
-    {
-      "epoch": 7.715775749674055,
-      "grad_norm": 1.411699891090393,
-      "learning_rate": 8.563718592964824e-05,
-      "loss": 5.4798,
-      "step": 14795
-    },
-    {
-      "epoch": 7.716297262059974,
-      "grad_norm": 1.6021950244903564,
-      "learning_rate": 8.563618090452262e-05,
-      "loss": 5.5775,
-      "step": 14796
-    },
-    {
-      "epoch": 7.716818774445893,
-      "grad_norm": 1.5753060579299927,
-      "learning_rate": 8.563517587939699e-05,
-      "loss": 4.8378,
-      "step": 14797
-    },
-    {
-      "epoch": 7.717340286831813,
-      "grad_norm": 1.5072102546691895,
-      "learning_rate": 8.563417085427135e-05,
-      "loss": 5.5738,
-      "step": 14798
-    },
-    {
-      "epoch": 7.717861799217731,
-      "grad_norm": 1.3707101345062256,
-      "learning_rate": 8.563316582914573e-05,
-      "loss": 5.7431,
-      "step": 14799
-    },
-    {
-      "epoch": 7.718383311603651,
-      "grad_norm": 1.4262690544128418,
-      "learning_rate": 8.56321608040201e-05,
-      "loss": 5.663,
-      "step": 14800
-    },
-    {
-      "epoch": 7.71890482398957,
-      "grad_norm": 1.348139762878418,
-      "learning_rate": 8.563115577889448e-05,
-      "loss": 5.1651,
-      "step": 14801
-    },
-    {
-      "epoch": 7.719426336375489,
-      "grad_norm": 1.3446165323257446,
-      "learning_rate": 8.563015075376885e-05,
-      "loss": 5.5771,
-      "step": 14802
-    },
-    {
-      "epoch": 7.719947848761408,
-      "grad_norm": 1.4711182117462158,
-      "learning_rate": 8.562914572864323e-05,
-      "loss": 5.6746,
-      "step": 14803
-    },
-    {
-      "epoch": 7.720469361147328,
-      "grad_norm": 1.3799740076065063,
-      "learning_rate": 8.562814070351759e-05,
-      "loss": 5.8473,
-      "step": 14804
-    },
-    {
-      "epoch": 7.720990873533246,
-      "grad_norm": 1.3516179323196411,
-      "learning_rate": 8.562713567839197e-05,
-      "loss": 5.6321,
-      "step": 14805
-    },
-    {
-      "epoch": 7.721512385919166,
-      "grad_norm": 1.6261240243911743,
-      "learning_rate": 8.562613065326633e-05,
-      "loss": 5.7353,
-      "step": 14806
-    },
-    {
-      "epoch": 7.722033898305085,
-      "grad_norm": 1.3962204456329346,
-      "learning_rate": 8.562512562814071e-05,
-      "loss": 5.5102,
-      "step": 14807
-    },
-    {
-      "epoch": 7.722555410691004,
-      "grad_norm": 1.4644442796707153,
-      "learning_rate": 8.562412060301507e-05,
-      "loss": 5.2088,
-      "step": 14808
-    },
-    {
-      "epoch": 7.723076923076923,
-      "grad_norm": 1.487269401550293,
-      "learning_rate": 8.562311557788945e-05,
-      "loss": 5.4267,
-      "step": 14809
-    },
-    {
-      "epoch": 7.723598435462842,
-      "grad_norm": 1.3908530473709106,
-      "learning_rate": 8.562211055276382e-05,
-      "loss": 5.3987,
-      "step": 14810
-    },
-    {
-      "epoch": 7.724119947848761,
-      "grad_norm": 1.5016013383865356,
-      "learning_rate": 8.56211055276382e-05,
-      "loss": 5.7055,
-      "step": 14811
-    },
-    {
-      "epoch": 7.724641460234681,
-      "grad_norm": 1.3737385272979736,
-      "learning_rate": 8.562010050251257e-05,
-      "loss": 5.939,
-      "step": 14812
-    },
-    {
-      "epoch": 7.7251629726206,
-      "grad_norm": 1.5205028057098389,
-      "learning_rate": 8.561909547738694e-05,
-      "loss": 5.5123,
-      "step": 14813
-    },
-    {
-      "epoch": 7.725684485006519,
-      "grad_norm": 1.3895668983459473,
-      "learning_rate": 8.561809045226131e-05,
-      "loss": 5.6332,
-      "step": 14814
-    },
-    {
-      "epoch": 7.726205997392438,
-      "grad_norm": 1.416590690612793,
-      "learning_rate": 8.561708542713568e-05,
-      "loss": 5.4861,
-      "step": 14815
-    },
-    {
-      "epoch": 7.726727509778357,
-      "grad_norm": 1.5170279741287231,
-      "learning_rate": 8.561608040201006e-05,
-      "loss": 4.6731,
-      "step": 14816
-    },
-    {
-      "epoch": 7.727249022164276,
-      "grad_norm": 1.4652515649795532,
-      "learning_rate": 8.561507537688442e-05,
-      "loss": 5.6669,
-      "step": 14817
-    },
-    {
-      "epoch": 7.727770534550196,
-      "grad_norm": 1.4400982856750488,
-      "learning_rate": 8.56140703517588e-05,
-      "loss": 5.8589,
-      "step": 14818
-    },
-    {
-      "epoch": 7.728292046936115,
-      "grad_norm": 1.37783682346344,
-      "learning_rate": 8.561306532663316e-05,
-      "loss": 5.492,
-      "step": 14819
-    },
-    {
-      "epoch": 7.728813559322034,
-      "grad_norm": 1.6035935878753662,
-      "learning_rate": 8.561206030150754e-05,
-      "loss": 4.804,
-      "step": 14820
-    },
-    {
-      "epoch": 7.729335071707953,
-      "grad_norm": 1.47847580909729,
-      "learning_rate": 8.561105527638192e-05,
-      "loss": 5.328,
-      "step": 14821
-    },
-    {
-      "epoch": 7.729856584093872,
-      "grad_norm": 1.4305375814437866,
-      "learning_rate": 8.56100502512563e-05,
-      "loss": 5.5318,
-      "step": 14822
-    },
-    {
-      "epoch": 7.730378096479791,
-      "grad_norm": 1.6848556995391846,
-      "learning_rate": 8.560904522613066e-05,
-      "loss": 5.5313,
-      "step": 14823
-    },
-    {
-      "epoch": 7.730899608865711,
-      "grad_norm": 1.4849793910980225,
-      "learning_rate": 8.560804020100503e-05,
-      "loss": 5.3012,
-      "step": 14824
-    },
-    {
-      "epoch": 7.73142112125163,
-      "grad_norm": 1.3191434144973755,
-      "learning_rate": 8.56070351758794e-05,
-      "loss": 5.7006,
-      "step": 14825
-    },
-    {
-      "epoch": 7.731942633637549,
-      "grad_norm": 1.4468410015106201,
-      "learning_rate": 8.560603015075377e-05,
-      "loss": 5.6186,
-      "step": 14826
-    },
-    {
-      "epoch": 7.732464146023468,
-      "grad_norm": 1.4431235790252686,
-      "learning_rate": 8.560502512562815e-05,
-      "loss": 5.4975,
-      "step": 14827
-    },
-    {
-      "epoch": 7.732985658409387,
-      "grad_norm": 1.5484137535095215,
-      "learning_rate": 8.560402010050251e-05,
-      "loss": 5.7887,
-      "step": 14828
-    },
-    {
-      "epoch": 7.733507170795306,
-      "grad_norm": 1.529132604598999,
-      "learning_rate": 8.560301507537689e-05,
-      "loss": 5.6447,
-      "step": 14829
-    },
-    {
-      "epoch": 7.734028683181226,
-      "grad_norm": 1.2961342334747314,
-      "learning_rate": 8.560201005025125e-05,
-      "loss": 5.9338,
-      "step": 14830
-    },
-    {
-      "epoch": 7.734550195567145,
-      "grad_norm": 1.599540114402771,
-      "learning_rate": 8.560100502512563e-05,
-      "loss": 4.8755,
-      "step": 14831
-    },
-    {
-      "epoch": 7.735071707953064,
-      "grad_norm": 1.6193578243255615,
-      "learning_rate": 8.560000000000001e-05,
-      "loss": 5.1817,
-      "step": 14832
-    },
-    {
-      "epoch": 7.735593220338983,
-      "grad_norm": 1.616989016532898,
-      "learning_rate": 8.559899497487438e-05,
-      "loss": 4.9347,
-      "step": 14833
-    },
-    {
-      "epoch": 7.736114732724902,
-      "grad_norm": 1.3978490829467773,
-      "learning_rate": 8.559798994974875e-05,
-      "loss": 5.5828,
-      "step": 14834
-    },
-    {
-      "epoch": 7.736636245110821,
-      "grad_norm": 1.7077770233154297,
-      "learning_rate": 8.559698492462313e-05,
-      "loss": 5.2395,
-      "step": 14835
-    },
-    {
-      "epoch": 7.737157757496741,
-      "grad_norm": 1.3986114263534546,
-      "learning_rate": 8.559597989949749e-05,
-      "loss": 5.6592,
-      "step": 14836
-    },
-    {
-      "epoch": 7.73767926988266,
-      "grad_norm": 1.5664916038513184,
-      "learning_rate": 8.559497487437186e-05,
-      "loss": 4.9861,
-      "step": 14837
-    },
-    {
-      "epoch": 7.738200782268579,
-      "grad_norm": 1.341298222541809,
-      "learning_rate": 8.559396984924623e-05,
-      "loss": 5.8691,
-      "step": 14838
-    },
-    {
-      "epoch": 7.738722294654498,
-      "grad_norm": 1.3619561195373535,
-      "learning_rate": 8.55929648241206e-05,
-      "loss": 5.5233,
-      "step": 14839
-    },
-    {
-      "epoch": 7.739243807040417,
-      "grad_norm": 1.5581225156784058,
-      "learning_rate": 8.559195979899498e-05,
-      "loss": 5.4068,
-      "step": 14840
-    },
-    {
-      "epoch": 7.739765319426336,
-      "grad_norm": 1.3480321168899536,
-      "learning_rate": 8.559095477386935e-05,
-      "loss": 5.5538,
-      "step": 14841
-    },
-    {
-      "epoch": 7.740286831812256,
-      "grad_norm": 1.4259237051010132,
-      "learning_rate": 8.558994974874373e-05,
-      "loss": 5.7791,
-      "step": 14842
-    },
-    {
-      "epoch": 7.740808344198175,
-      "grad_norm": 1.422835111618042,
-      "learning_rate": 8.55889447236181e-05,
-      "loss": 5.1687,
-      "step": 14843
-    },
-    {
-      "epoch": 7.741329856584094,
-      "grad_norm": 1.4099276065826416,
-      "learning_rate": 8.558793969849247e-05,
-      "loss": 5.367,
-      "step": 14844
-    },
-    {
-      "epoch": 7.741851368970013,
-      "grad_norm": 1.6038451194763184,
-      "learning_rate": 8.558693467336684e-05,
-      "loss": 5.4124,
-      "step": 14845
-    },
-    {
-      "epoch": 7.742372881355932,
-      "grad_norm": 1.406217336654663,
-      "learning_rate": 8.558592964824122e-05,
-      "loss": 5.7604,
-      "step": 14846
-    },
-    {
-      "epoch": 7.742894393741851,
-      "grad_norm": 1.3630237579345703,
-      "learning_rate": 8.558492462311558e-05,
-      "loss": 5.667,
-      "step": 14847
-    },
-    {
-      "epoch": 7.743415906127771,
-      "grad_norm": 1.2810211181640625,
-      "learning_rate": 8.558391959798996e-05,
-      "loss": 5.9569,
-      "step": 14848
-    },
-    {
-      "epoch": 7.743937418513689,
-      "grad_norm": 1.3420000076293945,
-      "learning_rate": 8.558291457286432e-05,
-      "loss": 5.8974,
-      "step": 14849
-    },
-    {
-      "epoch": 7.744458930899609,
-      "grad_norm": 1.4180821180343628,
-      "learning_rate": 8.558190954773869e-05,
-      "loss": 5.804,
-      "step": 14850
-    },
-    {
-      "epoch": 7.744980443285528,
-      "grad_norm": 1.4010448455810547,
-      "learning_rate": 8.558090452261306e-05,
-      "loss": 5.5227,
-      "step": 14851
-    },
-    {
-      "epoch": 7.745501955671447,
-      "grad_norm": 1.4597938060760498,
-      "learning_rate": 8.557989949748744e-05,
-      "loss": 5.2178,
-      "step": 14852
-    },
-    {
-      "epoch": 7.746023468057366,
-      "grad_norm": 1.3471705913543701,
-      "learning_rate": 8.557889447236182e-05,
-      "loss": 5.6795,
-      "step": 14853
-    },
-    {
-      "epoch": 7.746544980443286,
-      "grad_norm": 1.351319670677185,
-      "learning_rate": 8.557788944723618e-05,
-      "loss": 5.7393,
-      "step": 14854
-    },
-    {
-      "epoch": 7.747066492829204,
-      "grad_norm": 1.4462369680404663,
-      "learning_rate": 8.557688442211056e-05,
-      "loss": 5.6822,
-      "step": 14855
-    },
-    {
-      "epoch": 7.747588005215124,
-      "grad_norm": 1.5280177593231201,
-      "learning_rate": 8.557587939698493e-05,
-      "loss": 5.321,
-      "step": 14856
-    },
-    {
-      "epoch": 7.748109517601043,
-      "grad_norm": 1.3730849027633667,
-      "learning_rate": 8.55748743718593e-05,
-      "loss": 5.8742,
-      "step": 14857
-    },
-    {
-      "epoch": 7.748631029986962,
-      "grad_norm": 1.318661093711853,
-      "learning_rate": 8.557386934673367e-05,
-      "loss": 5.4389,
-      "step": 14858
-    },
-    {
-      "epoch": 7.749152542372881,
-      "grad_norm": 1.4118945598602295,
-      "learning_rate": 8.557286432160805e-05,
-      "loss": 5.2063,
-      "step": 14859
-    },
-    {
-      "epoch": 7.749674054758801,
-      "grad_norm": 1.4344338178634644,
-      "learning_rate": 8.557185929648241e-05,
-      "loss": 5.5654,
-      "step": 14860
-    },
-    {
-      "epoch": 7.750195567144719,
-      "grad_norm": 1.392613172531128,
-      "learning_rate": 8.557085427135679e-05,
-      "loss": 5.5008,
-      "step": 14861
-    },
-    {
-      "epoch": 7.750717079530639,
-      "grad_norm": 1.3882275819778442,
-      "learning_rate": 8.556984924623117e-05,
-      "loss": 5.6829,
-      "step": 14862
-    },
-    {
-      "epoch": 7.751238591916558,
-      "grad_norm": 1.5139144659042358,
-      "learning_rate": 8.556884422110553e-05,
-      "loss": 5.4817,
-      "step": 14863
-    },
-    {
-      "epoch": 7.751760104302477,
-      "grad_norm": 1.4410115480422974,
-      "learning_rate": 8.556783919597991e-05,
-      "loss": 5.467,
-      "step": 14864
-    },
-    {
-      "epoch": 7.752281616688396,
-      "grad_norm": 1.3939090967178345,
-      "learning_rate": 8.556683417085427e-05,
-      "loss": 5.6303,
-      "step": 14865
-    },
-    {
-      "epoch": 7.752803129074316,
-      "grad_norm": 1.6484264135360718,
-      "learning_rate": 8.556582914572865e-05,
-      "loss": 4.5525,
-      "step": 14866
-    },
-    {
-      "epoch": 7.753324641460234,
-      "grad_norm": 1.347132682800293,
-      "learning_rate": 8.556482412060301e-05,
-      "loss": 5.404,
-      "step": 14867
-    },
-    {
-      "epoch": 7.753846153846154,
-      "grad_norm": 1.3919563293457031,
-      "learning_rate": 8.556381909547739e-05,
-      "loss": 5.5045,
-      "step": 14868
-    },
-    {
-      "epoch": 7.754367666232073,
-      "grad_norm": 1.3859913349151611,
-      "learning_rate": 8.556281407035176e-05,
-      "loss": 5.5335,
-      "step": 14869
-    },
-    {
-      "epoch": 7.754889178617992,
-      "grad_norm": 1.341484546661377,
-      "learning_rate": 8.556180904522613e-05,
-      "loss": 5.4631,
-      "step": 14870
-    },
-    {
-      "epoch": 7.755410691003911,
-      "grad_norm": 1.4081883430480957,
-      "learning_rate": 8.55608040201005e-05,
-      "loss": 5.6345,
-      "step": 14871
-    },
-    {
-      "epoch": 7.755932203389831,
-      "grad_norm": 1.4294639825820923,
-      "learning_rate": 8.555979899497488e-05,
-      "loss": 5.4696,
-      "step": 14872
-    },
-    {
-      "epoch": 7.756453715775749,
-      "grad_norm": 1.3975739479064941,
-      "learning_rate": 8.555879396984925e-05,
-      "loss": 5.8295,
-      "step": 14873
-    },
-    {
-      "epoch": 7.756975228161669,
-      "grad_norm": 1.3718607425689697,
-      "learning_rate": 8.555778894472363e-05,
-      "loss": 5.7622,
-      "step": 14874
-    },
-    {
-      "epoch": 7.757496740547588,
-      "grad_norm": 1.4000059366226196,
-      "learning_rate": 8.5556783919598e-05,
-      "loss": 5.8113,
-      "step": 14875
-    },
-    {
-      "epoch": 7.758018252933507,
-      "grad_norm": 1.6417006254196167,
-      "learning_rate": 8.555577889447236e-05,
-      "loss": 4.9166,
-      "step": 14876
-    },
-    {
-      "epoch": 7.758539765319426,
-      "grad_norm": 1.366329550743103,
-      "learning_rate": 8.555477386934674e-05,
-      "loss": 5.8791,
-      "step": 14877
-    },
-    {
-      "epoch": 7.759061277705346,
-      "grad_norm": 1.49155855178833,
-      "learning_rate": 8.55537688442211e-05,
-      "loss": 5.6821,
-      "step": 14878
-    },
-    {
-      "epoch": 7.759582790091264,
-      "grad_norm": 1.4981639385223389,
-      "learning_rate": 8.555276381909548e-05,
-      "loss": 5.2831,
-      "step": 14879
-    },
-    {
-      "epoch": 7.760104302477184,
-      "grad_norm": 1.4202502965927124,
-      "learning_rate": 8.555175879396984e-05,
-      "loss": 5.4306,
-      "step": 14880
-    },
-    {
-      "epoch": 7.760625814863103,
-      "grad_norm": 1.4384582042694092,
-      "learning_rate": 8.555075376884422e-05,
-      "loss": 5.6187,
-      "step": 14881
-    },
-    {
-      "epoch": 7.761147327249022,
-      "grad_norm": 1.5004130601882935,
-      "learning_rate": 8.55497487437186e-05,
-      "loss": 5.5152,
-      "step": 14882
-    },
-    {
-      "epoch": 7.761668839634941,
-      "grad_norm": 1.4274336099624634,
-      "learning_rate": 8.554874371859298e-05,
-      "loss": 5.6489,
-      "step": 14883
-    },
-    {
-      "epoch": 7.762190352020861,
-      "grad_norm": 1.4857982397079468,
-      "learning_rate": 8.554773869346734e-05,
-      "loss": 5.6123,
-      "step": 14884
-    },
-    {
-      "epoch": 7.762711864406779,
-      "grad_norm": 1.3764257431030273,
-      "learning_rate": 8.554673366834172e-05,
-      "loss": 5.6241,
-      "step": 14885
-    },
-    {
-      "epoch": 7.763233376792699,
-      "grad_norm": 1.520395278930664,
-      "learning_rate": 8.554572864321608e-05,
-      "loss": 5.1258,
-      "step": 14886
-    },
-    {
-      "epoch": 7.763754889178618,
-      "grad_norm": 1.4276330471038818,
-      "learning_rate": 8.554472361809046e-05,
-      "loss": 5.3689,
-      "step": 14887
-    },
-    {
-      "epoch": 7.764276401564537,
-      "grad_norm": 1.5538287162780762,
-      "learning_rate": 8.554371859296483e-05,
-      "loss": 5.4306,
-      "step": 14888
-    },
-    {
-      "epoch": 7.764797913950456,
-      "grad_norm": 1.5398759841918945,
-      "learning_rate": 8.55427135678392e-05,
-      "loss": 5.3133,
-      "step": 14889
-    },
-    {
-      "epoch": 7.765319426336376,
-      "grad_norm": 1.3996349573135376,
-      "learning_rate": 8.554170854271357e-05,
-      "loss": 5.58,
-      "step": 14890
-    },
-    {
-      "epoch": 7.765840938722294,
-      "grad_norm": 1.4135024547576904,
-      "learning_rate": 8.554070351758793e-05,
-      "loss": 5.8082,
-      "step": 14891
-    },
-    {
-      "epoch": 7.766362451108214,
-      "grad_norm": 1.3489164113998413,
-      "learning_rate": 8.553969849246231e-05,
-      "loss": 5.8045,
-      "step": 14892
-    },
-    {
-      "epoch": 7.766883963494133,
-      "grad_norm": 1.4408648014068604,
-      "learning_rate": 8.553869346733669e-05,
-      "loss": 5.5447,
-      "step": 14893
-    },
-    {
-      "epoch": 7.767405475880052,
-      "grad_norm": 1.426633358001709,
-      "learning_rate": 8.553768844221107e-05,
-      "loss": 5.7946,
-      "step": 14894
-    },
-    {
-      "epoch": 7.767926988265971,
-      "grad_norm": 1.4904190301895142,
-      "learning_rate": 8.553668341708543e-05,
-      "loss": 5.7233,
-      "step": 14895
-    },
-    {
-      "epoch": 7.768448500651891,
-      "grad_norm": 1.3592597246170044,
-      "learning_rate": 8.553567839195981e-05,
-      "loss": 5.8029,
-      "step": 14896
-    },
-    {
-      "epoch": 7.768970013037809,
-      "grad_norm": 1.4270555973052979,
-      "learning_rate": 8.553467336683417e-05,
-      "loss": 5.3394,
-      "step": 14897
-    },
-    {
-      "epoch": 7.769491525423729,
-      "grad_norm": 1.3691420555114746,
-      "learning_rate": 8.553366834170855e-05,
-      "loss": 5.7616,
-      "step": 14898
-    },
-    {
-      "epoch": 7.770013037809648,
-      "grad_norm": 1.4500921964645386,
-      "learning_rate": 8.553266331658292e-05,
-      "loss": 5.4399,
-      "step": 14899
-    },
-    {
-      "epoch": 7.770534550195567,
-      "grad_norm": 1.4810245037078857,
-      "learning_rate": 8.553165829145729e-05,
-      "loss": 5.5693,
-      "step": 14900
-    },
-    {
-      "epoch": 7.771056062581486,
-      "grad_norm": 1.4829703569412231,
-      "learning_rate": 8.553065326633166e-05,
-      "loss": 5.2089,
-      "step": 14901
-    },
-    {
-      "epoch": 7.771577574967406,
-      "grad_norm": 1.5846468210220337,
-      "learning_rate": 8.552964824120603e-05,
-      "loss": 5.3582,
-      "step": 14902
-    },
-    {
-      "epoch": 7.772099087353324,
-      "grad_norm": 1.5076220035552979,
-      "learning_rate": 8.552864321608041e-05,
-      "loss": 4.9961,
-      "step": 14903
-    },
-    {
-      "epoch": 7.772620599739244,
-      "grad_norm": 1.4856815338134766,
-      "learning_rate": 8.552763819095478e-05,
-      "loss": 5.5356,
-      "step": 14904
-    },
-    {
-      "epoch": 7.7731421121251625,
-      "grad_norm": 1.3288263082504272,
-      "learning_rate": 8.552663316582915e-05,
-      "loss": 5.6145,
-      "step": 14905
-    },
-    {
-      "epoch": 7.773663624511082,
-      "grad_norm": 1.5263499021530151,
-      "learning_rate": 8.552562814070352e-05,
-      "loss": 5.3872,
-      "step": 14906
-    },
-    {
-      "epoch": 7.774185136897001,
-      "grad_norm": 1.4119091033935547,
-      "learning_rate": 8.55246231155779e-05,
-      "loss": 5.485,
-      "step": 14907
-    },
-    {
-      "epoch": 7.774706649282921,
-      "grad_norm": 1.3823961019515991,
-      "learning_rate": 8.552361809045226e-05,
-      "loss": 5.7401,
-      "step": 14908
-    },
-    {
-      "epoch": 7.7752281616688395,
-      "grad_norm": 1.4780751466751099,
-      "learning_rate": 8.552261306532664e-05,
-      "loss": 5.5376,
-      "step": 14909
-    },
-    {
-      "epoch": 7.775749674054759,
-      "grad_norm": 1.429235577583313,
-      "learning_rate": 8.5521608040201e-05,
-      "loss": 5.4443,
-      "step": 14910
-    },
-    {
-      "epoch": 7.7762711864406775,
-      "grad_norm": 1.5963925123214722,
-      "learning_rate": 8.552060301507538e-05,
-      "loss": 5.2469,
-      "step": 14911
-    },
-    {
-      "epoch": 7.776792698826597,
-      "grad_norm": 1.5040754079818726,
-      "learning_rate": 8.551959798994975e-05,
-      "loss": 5.5508,
-      "step": 14912
-    },
-    {
-      "epoch": 7.777314211212516,
-      "grad_norm": 1.386915683746338,
-      "learning_rate": 8.551859296482412e-05,
-      "loss": 5.8192,
-      "step": 14913
-    },
-    {
-      "epoch": 7.777835723598436,
-      "grad_norm": 1.442958950996399,
-      "learning_rate": 8.55175879396985e-05,
-      "loss": 5.5002,
-      "step": 14914
-    },
-    {
-      "epoch": 7.7783572359843545,
-      "grad_norm": 1.4908467531204224,
-      "learning_rate": 8.551658291457288e-05,
-      "loss": 4.9773,
-      "step": 14915
-    },
-    {
-      "epoch": 7.778878748370274,
-      "grad_norm": 1.4285255670547485,
-      "learning_rate": 8.551557788944724e-05,
-      "loss": 5.6449,
-      "step": 14916
-    },
-    {
-      "epoch": 7.7794002607561925,
-      "grad_norm": 1.462139368057251,
-      "learning_rate": 8.551457286432161e-05,
-      "loss": 5.1195,
-      "step": 14917
-    },
-    {
-      "epoch": 7.779921773142112,
-      "grad_norm": 1.5461052656173706,
-      "learning_rate": 8.551356783919599e-05,
-      "loss": 5.4099,
-      "step": 14918
-    },
-    {
-      "epoch": 7.780443285528031,
-      "grad_norm": 1.354026436805725,
-      "learning_rate": 8.551256281407035e-05,
-      "loss": 5.6542,
-      "step": 14919
-    },
-    {
-      "epoch": 7.780964797913951,
-      "grad_norm": 1.4298937320709229,
-      "learning_rate": 8.551155778894473e-05,
-      "loss": 5.5751,
-      "step": 14920
-    },
-    {
-      "epoch": 7.7814863102998695,
-      "grad_norm": 1.3176008462905884,
-      "learning_rate": 8.551055276381909e-05,
-      "loss": 5.684,
-      "step": 14921
-    },
-    {
-      "epoch": 7.782007822685789,
-      "grad_norm": 1.3620240688323975,
-      "learning_rate": 8.550954773869347e-05,
-      "loss": 5.8003,
-      "step": 14922
-    },
-    {
-      "epoch": 7.7825293350717075,
-      "grad_norm": 1.5781197547912598,
-      "learning_rate": 8.550854271356785e-05,
-      "loss": 5.6626,
-      "step": 14923
-    },
-    {
-      "epoch": 7.783050847457627,
-      "grad_norm": 1.5242420434951782,
-      "learning_rate": 8.550753768844223e-05,
-      "loss": 5.2804,
-      "step": 14924
-    },
-    {
-      "epoch": 7.7835723598435465,
-      "grad_norm": 1.3122990131378174,
-      "learning_rate": 8.550653266331659e-05,
-      "loss": 5.811,
-      "step": 14925
-    },
-    {
-      "epoch": 7.784093872229466,
-      "grad_norm": 1.4122151136398315,
-      "learning_rate": 8.550552763819097e-05,
-      "loss": 5.5895,
-      "step": 14926
-    },
-    {
-      "epoch": 7.7846153846153845,
-      "grad_norm": 1.452530860900879,
-      "learning_rate": 8.550452261306533e-05,
-      "loss": 5.5226,
-      "step": 14927
-    },
-    {
-      "epoch": 7.785136897001304,
-      "grad_norm": 1.4963266849517822,
-      "learning_rate": 8.550351758793971e-05,
-      "loss": 5.7235,
-      "step": 14928
-    },
-    {
-      "epoch": 7.7856584093872225,
-      "grad_norm": 1.647230863571167,
-      "learning_rate": 8.550251256281407e-05,
-      "loss": 5.2832,
-      "step": 14929
-    },
-    {
-      "epoch": 7.786179921773142,
-      "grad_norm": 1.3008484840393066,
-      "learning_rate": 8.550150753768844e-05,
-      "loss": 5.8745,
-      "step": 14930
-    },
-    {
-      "epoch": 7.7867014341590615,
-      "grad_norm": 1.3708586692810059,
-      "learning_rate": 8.550050251256282e-05,
-      "loss": 5.6008,
-      "step": 14931
-    },
-    {
-      "epoch": 7.787222946544981,
-      "grad_norm": 1.56390380859375,
-      "learning_rate": 8.549949748743718e-05,
-      "loss": 5.6742,
-      "step": 14932
-    },
-    {
-      "epoch": 7.7877444589308995,
-      "grad_norm": 1.610580563545227,
-      "learning_rate": 8.549849246231156e-05,
-      "loss": 5.3062,
-      "step": 14933
-    },
-    {
-      "epoch": 7.788265971316819,
-      "grad_norm": 1.531510353088379,
-      "learning_rate": 8.549748743718594e-05,
-      "loss": 5.1024,
-      "step": 14934
-    },
-    {
-      "epoch": 7.7887874837027375,
-      "grad_norm": 1.4622738361358643,
-      "learning_rate": 8.549648241206031e-05,
-      "loss": 5.4442,
-      "step": 14935
-    },
-    {
-      "epoch": 7.789308996088657,
-      "grad_norm": 1.540500283241272,
-      "learning_rate": 8.549547738693468e-05,
-      "loss": 5.9065,
-      "step": 14936
-    },
-    {
-      "epoch": 7.7898305084745765,
-      "grad_norm": 1.4684021472930908,
-      "learning_rate": 8.549447236180906e-05,
-      "loss": 5.6382,
-      "step": 14937
-    },
-    {
-      "epoch": 7.790352020860495,
-      "grad_norm": 1.5925712585449219,
-      "learning_rate": 8.549346733668342e-05,
-      "loss": 5.2667,
-      "step": 14938
-    },
-    {
-      "epoch": 7.7908735332464145,
-      "grad_norm": 1.4803155660629272,
-      "learning_rate": 8.54924623115578e-05,
-      "loss": 5.7217,
-      "step": 14939
-    },
-    {
-      "epoch": 7.791395045632334,
-      "grad_norm": 1.3818414211273193,
-      "learning_rate": 8.549145728643216e-05,
-      "loss": 5.4047,
-      "step": 14940
-    },
-    {
-      "epoch": 7.791916558018253,
-      "grad_norm": 1.4123834371566772,
-      "learning_rate": 8.549045226130654e-05,
-      "loss": 5.1711,
-      "step": 14941
-    },
-    {
-      "epoch": 7.792438070404172,
-      "grad_norm": 1.3774057626724243,
-      "learning_rate": 8.54894472361809e-05,
-      "loss": 5.7568,
-      "step": 14942
-    },
-    {
-      "epoch": 7.7929595827900915,
-      "grad_norm": 1.3357653617858887,
-      "learning_rate": 8.548844221105527e-05,
-      "loss": 5.8832,
-      "step": 14943
-    },
-    {
-      "epoch": 7.79348109517601,
-      "grad_norm": 1.3212697505950928,
-      "learning_rate": 8.548743718592965e-05,
-      "loss": 5.5552,
-      "step": 14944
-    },
-    {
-      "epoch": 7.7940026075619295,
-      "grad_norm": 1.335204839706421,
-      "learning_rate": 8.548643216080402e-05,
-      "loss": 6.042,
-      "step": 14945
-    },
-    {
-      "epoch": 7.794524119947849,
-      "grad_norm": 1.3908599615097046,
-      "learning_rate": 8.54854271356784e-05,
-      "loss": 5.3363,
-      "step": 14946
-    },
-    {
-      "epoch": 7.795045632333768,
-      "grad_norm": 1.30963134765625,
-      "learning_rate": 8.548442211055277e-05,
-      "loss": 5.7074,
-      "step": 14947
-    },
-    {
-      "epoch": 7.795567144719687,
-      "grad_norm": 1.3709102869033813,
-      "learning_rate": 8.548341708542714e-05,
-      "loss": 5.5427,
-      "step": 14948
-    },
-    {
-      "epoch": 7.7960886571056065,
-      "grad_norm": 1.6157333850860596,
-      "learning_rate": 8.548241206030151e-05,
-      "loss": 5.2273,
-      "step": 14949
-    },
-    {
-      "epoch": 7.796610169491525,
-      "grad_norm": 1.4239939451217651,
-      "learning_rate": 8.548140703517589e-05,
-      "loss": 5.1344,
-      "step": 14950
-    },
-    {
-      "epoch": 7.7971316818774445,
-      "grad_norm": 1.435671091079712,
-      "learning_rate": 8.548040201005025e-05,
-      "loss": 5.9393,
-      "step": 14951
-    },
-    {
-      "epoch": 7.797653194263364,
-      "grad_norm": 1.5465866327285767,
-      "learning_rate": 8.547939698492463e-05,
-      "loss": 5.3754,
-      "step": 14952
-    },
-    {
-      "epoch": 7.798174706649283,
-      "grad_norm": 1.487158179283142,
-      "learning_rate": 8.547839195979899e-05,
-      "loss": 5.3784,
-      "step": 14953
-    },
-    {
-      "epoch": 7.798696219035202,
-      "grad_norm": 1.4286308288574219,
-      "learning_rate": 8.547738693467337e-05,
-      "loss": 5.4272,
-      "step": 14954
-    },
-    {
-      "epoch": 7.7992177314211215,
-      "grad_norm": 1.3836826086044312,
-      "learning_rate": 8.547638190954775e-05,
-      "loss": 5.7135,
-      "step": 14955
-    },
-    {
-      "epoch": 7.79973924380704,
-      "grad_norm": 1.4383779764175415,
-      "learning_rate": 8.547537688442211e-05,
-      "loss": 5.5651,
-      "step": 14956
-    },
-    {
-      "epoch": 7.8002607561929596,
-      "grad_norm": 1.3960251808166504,
-      "learning_rate": 8.547437185929649e-05,
-      "loss": 5.6825,
-      "step": 14957
-    },
-    {
-      "epoch": 7.800782268578879,
-      "grad_norm": 1.378402590751648,
-      "learning_rate": 8.547336683417085e-05,
-      "loss": 5.7609,
-      "step": 14958
-    },
-    {
-      "epoch": 7.801303780964798,
-      "grad_norm": 1.3570935726165771,
-      "learning_rate": 8.547236180904523e-05,
-      "loss": 5.8092,
-      "step": 14959
-    },
-    {
-      "epoch": 7.801825293350717,
-      "grad_norm": 1.3703584671020508,
-      "learning_rate": 8.54713567839196e-05,
-      "loss": 5.7603,
-      "step": 14960
-    },
-    {
-      "epoch": 7.8023468057366365,
-      "grad_norm": 1.4241948127746582,
-      "learning_rate": 8.547035175879397e-05,
-      "loss": 5.7046,
-      "step": 14961
-    },
-    {
-      "epoch": 7.802868318122555,
-      "grad_norm": 1.4227250814437866,
-      "learning_rate": 8.546934673366834e-05,
-      "loss": 5.5059,
-      "step": 14962
-    },
-    {
-      "epoch": 7.803389830508475,
-      "grad_norm": 1.313709020614624,
-      "learning_rate": 8.546834170854272e-05,
-      "loss": 5.6134,
-      "step": 14963
-    },
-    {
-      "epoch": 7.803911342894394,
-      "grad_norm": 1.4118183851242065,
-      "learning_rate": 8.546733668341708e-05,
-      "loss": 5.5656,
-      "step": 14964
-    },
-    {
-      "epoch": 7.804432855280313,
-      "grad_norm": 1.5375540256500244,
-      "learning_rate": 8.546633165829146e-05,
-      "loss": 5.5011,
-      "step": 14965
-    },
-    {
-      "epoch": 7.804954367666232,
-      "grad_norm": 1.3843860626220703,
-      "learning_rate": 8.546532663316584e-05,
-      "loss": 5.5353,
-      "step": 14966
-    },
-    {
-      "epoch": 7.8054758800521515,
-      "grad_norm": 1.4926122426986694,
-      "learning_rate": 8.546432160804021e-05,
-      "loss": 5.6153,
-      "step": 14967
-    },
-    {
-      "epoch": 7.80599739243807,
-      "grad_norm": 1.3765252828598022,
-      "learning_rate": 8.546331658291458e-05,
-      "loss": 5.6661,
-      "step": 14968
-    },
-    {
-      "epoch": 7.80651890482399,
-      "grad_norm": 1.4020739793777466,
-      "learning_rate": 8.546231155778894e-05,
-      "loss": 5.671,
-      "step": 14969
-    },
-    {
-      "epoch": 7.807040417209909,
-      "grad_norm": 1.4202426671981812,
-      "learning_rate": 8.546130653266332e-05,
-      "loss": 6.0366,
-      "step": 14970
-    },
-    {
-      "epoch": 7.807561929595828,
-      "grad_norm": 1.4802809953689575,
-      "learning_rate": 8.546030150753768e-05,
-      "loss": 5.2948,
-      "step": 14971
-    },
-    {
-      "epoch": 7.808083441981747,
-      "grad_norm": 1.3885397911071777,
-      "learning_rate": 8.545929648241206e-05,
-      "loss": 5.7996,
-      "step": 14972
-    },
-    {
-      "epoch": 7.8086049543676666,
-      "grad_norm": 1.4254744052886963,
-      "learning_rate": 8.545829145728643e-05,
-      "loss": 5.5984,
-      "step": 14973
-    },
-    {
-      "epoch": 7.809126466753585,
-      "grad_norm": 1.3783246278762817,
-      "learning_rate": 8.54572864321608e-05,
-      "loss": 5.9201,
-      "step": 14974
-    },
-    {
-      "epoch": 7.809647979139505,
-      "grad_norm": 1.5453559160232544,
-      "learning_rate": 8.545628140703518e-05,
-      "loss": 4.978,
-      "step": 14975
-    },
-    {
-      "epoch": 7.810169491525424,
-      "grad_norm": 1.3881481885910034,
-      "learning_rate": 8.545527638190956e-05,
-      "loss": 5.7656,
-      "step": 14976
-    },
-    {
-      "epoch": 7.810691003911343,
-      "grad_norm": 1.4385395050048828,
-      "learning_rate": 8.545427135678392e-05,
-      "loss": 5.6781,
-      "step": 14977
-    },
-    {
-      "epoch": 7.811212516297262,
-      "grad_norm": 1.3628040552139282,
-      "learning_rate": 8.54532663316583e-05,
-      "loss": 5.8769,
-      "step": 14978
-    },
-    {
-      "epoch": 7.811734028683182,
-      "grad_norm": 1.5564589500427246,
-      "learning_rate": 8.545226130653267e-05,
-      "loss": 4.9415,
-      "step": 14979
-    },
-    {
-      "epoch": 7.8122555410691,
-      "grad_norm": 1.4005335569381714,
-      "learning_rate": 8.545125628140704e-05,
-      "loss": 5.4438,
-      "step": 14980
-    },
-    {
-      "epoch": 7.81277705345502,
-      "grad_norm": 1.3782708644866943,
-      "learning_rate": 8.545025125628141e-05,
-      "loss": 5.674,
-      "step": 14981
-    },
-    {
-      "epoch": 7.813298565840939,
-      "grad_norm": 1.6356866359710693,
-      "learning_rate": 8.544924623115579e-05,
-      "loss": 5.0197,
-      "step": 14982
-    },
-    {
-      "epoch": 7.813820078226858,
-      "grad_norm": 1.4557808637619019,
-      "learning_rate": 8.544824120603015e-05,
-      "loss": 5.6339,
-      "step": 14983
-    },
-    {
-      "epoch": 7.814341590612777,
-      "grad_norm": 1.5113551616668701,
-      "learning_rate": 8.544723618090452e-05,
-      "loss": 4.9117,
-      "step": 14984
-    },
-    {
-      "epoch": 7.814863102998697,
-      "grad_norm": 1.6178703308105469,
-      "learning_rate": 8.54462311557789e-05,
-      "loss": 5.3937,
-      "step": 14985
-    },
-    {
-      "epoch": 7.815384615384615,
-      "grad_norm": 1.593268632888794,
-      "learning_rate": 8.544522613065327e-05,
-      "loss": 5.494,
-      "step": 14986
-    },
-    {
-      "epoch": 7.815906127770535,
-      "grad_norm": 1.622837781906128,
-      "learning_rate": 8.544422110552765e-05,
-      "loss": 5.3244,
-      "step": 14987
-    },
-    {
-      "epoch": 7.816427640156454,
-      "grad_norm": 1.4447187185287476,
-      "learning_rate": 8.544321608040201e-05,
-      "loss": 5.758,
-      "step": 14988
-    },
-    {
-      "epoch": 7.816949152542373,
-      "grad_norm": 1.3926409482955933,
-      "learning_rate": 8.544221105527639e-05,
-      "loss": 5.6213,
-      "step": 14989
-    },
-    {
-      "epoch": 7.817470664928292,
-      "grad_norm": 1.4714878797531128,
-      "learning_rate": 8.544120603015076e-05,
-      "loss": 5.3579,
-      "step": 14990
-    },
-    {
-      "epoch": 7.817992177314212,
-      "grad_norm": 1.4858633279800415,
-      "learning_rate": 8.544020100502513e-05,
-      "loss": 5.6764,
-      "step": 14991
-    },
-    {
-      "epoch": 7.81851368970013,
-      "grad_norm": 1.4992055892944336,
-      "learning_rate": 8.54391959798995e-05,
-      "loss": 5.7646,
-      "step": 14992
-    },
-    {
-      "epoch": 7.81903520208605,
-      "grad_norm": 1.3898687362670898,
-      "learning_rate": 8.543819095477388e-05,
-      "loss": 5.3982,
-      "step": 14993
-    },
-    {
-      "epoch": 7.819556714471969,
-      "grad_norm": 1.3627413511276245,
-      "learning_rate": 8.543718592964824e-05,
-      "loss": 5.9543,
-      "step": 14994
-    },
-    {
-      "epoch": 7.820078226857888,
-      "grad_norm": 1.4318257570266724,
-      "learning_rate": 8.543618090452262e-05,
-      "loss": 5.6955,
-      "step": 14995
-    },
-    {
-      "epoch": 7.820599739243807,
-      "grad_norm": 1.4412955045700073,
-      "learning_rate": 8.5435175879397e-05,
-      "loss": 5.6002,
-      "step": 14996
-    },
-    {
-      "epoch": 7.821121251629727,
-      "grad_norm": 1.37083899974823,
-      "learning_rate": 8.543417085427136e-05,
-      "loss": 5.9952,
-      "step": 14997
-    },
-    {
-      "epoch": 7.821642764015645,
-      "grad_norm": 1.443403959274292,
-      "learning_rate": 8.543316582914574e-05,
-      "loss": 5.3132,
-      "step": 14998
-    },
-    {
-      "epoch": 7.822164276401565,
-      "grad_norm": 1.4042354822158813,
-      "learning_rate": 8.54321608040201e-05,
-      "loss": 5.7752,
-      "step": 14999
-    },
-    {
-      "epoch": 7.822685788787483,
-      "grad_norm": 1.4015518426895142,
-      "learning_rate": 8.543115577889448e-05,
-      "loss": 5.4686,
-      "step": 15000
-    },
-    {
-      "epoch": 7.822685788787483,
-      "eval_loss": 5.586970806121826,
-      "eval_runtime": 42.6841,
-      "eval_samples_per_second": 28.723,
-      "eval_steps_per_second": 3.608,
-      "step": 15000
-    },
-    {
-      "epoch": 7.823207301173403,
-      "grad_norm": 1.6287904977798462,
-      "learning_rate": 8.543015075376884e-05,
-      "loss": 4.6893,
-      "step": 15001
-    },
-    {
-      "epoch": 7.823728813559322,
-      "grad_norm": 1.4765905141830444,
-      "learning_rate": 8.542914572864322e-05,
-      "loss": 5.3284,
-      "step": 15002
-    },
-    {
-      "epoch": 7.824250325945242,
-      "grad_norm": 1.3667665719985962,
-      "learning_rate": 8.542814070351759e-05,
-      "loss": 5.5729,
-      "step": 15003
-    },
-    {
-      "epoch": 7.82477183833116,
-      "grad_norm": 1.5398184061050415,
-      "learning_rate": 8.542713567839196e-05,
-      "loss": 5.0842,
-      "step": 15004
-    },
-    {
-      "epoch": 7.82529335071708,
-      "grad_norm": 1.7367740869522095,
-      "learning_rate": 8.542613065326633e-05,
-      "loss": 5.0605,
-      "step": 15005
-    },
-    {
-      "epoch": 7.825814863102998,
-      "grad_norm": 1.4108339548110962,
-      "learning_rate": 8.54251256281407e-05,
-      "loss": 5.1719,
-      "step": 15006
-    },
-    {
-      "epoch": 7.826336375488918,
-      "grad_norm": 1.4791722297668457,
-      "learning_rate": 8.542412060301508e-05,
-      "loss": 5.2301,
-      "step": 15007
-    },
-    {
-      "epoch": 7.826857887874837,
-      "grad_norm": 1.4308544397354126,
-      "learning_rate": 8.542311557788946e-05,
-      "loss": 5.3753,
-      "step": 15008
-    },
-    {
-      "epoch": 7.827379400260757,
-      "grad_norm": 1.4527614116668701,
-      "learning_rate": 8.542211055276383e-05,
-      "loss": 5.9292,
-      "step": 15009
-    },
-    {
-      "epoch": 7.827900912646675,
-      "grad_norm": 1.535609245300293,
-      "learning_rate": 8.542110552763819e-05,
-      "loss": 5.3159,
-      "step": 15010
-    },
-    {
-      "epoch": 7.828422425032595,
-      "grad_norm": 1.4543719291687012,
-      "learning_rate": 8.542010050251257e-05,
-      "loss": 5.8706,
-      "step": 15011
-    },
-    {
-      "epoch": 7.828943937418513,
-      "grad_norm": 1.4280061721801758,
-      "learning_rate": 8.541909547738693e-05,
-      "loss": 5.5142,
-      "step": 15012
-    },
-    {
-      "epoch": 7.829465449804433,
-      "grad_norm": 1.5681068897247314,
-      "learning_rate": 8.541809045226131e-05,
-      "loss": 4.9883,
-      "step": 15013
-    },
-    {
-      "epoch": 7.829986962190352,
-      "grad_norm": 1.482130765914917,
-      "learning_rate": 8.541708542713567e-05,
-      "loss": 5.3713,
-      "step": 15014
-    },
-    {
-      "epoch": 7.830508474576272,
-      "grad_norm": 1.4447333812713623,
-      "learning_rate": 8.541608040201005e-05,
-      "loss": 5.7443,
-      "step": 15015
-    },
-    {
-      "epoch": 7.83102998696219,
-      "grad_norm": 1.3203773498535156,
-      "learning_rate": 8.541507537688443e-05,
-      "loss": 5.7947,
-      "step": 15016
-    },
-    {
-      "epoch": 7.83155149934811,
-      "grad_norm": 1.375363826751709,
-      "learning_rate": 8.541407035175881e-05,
-      "loss": 5.5196,
-      "step": 15017
-    },
-    {
-      "epoch": 7.832073011734028,
-      "grad_norm": 1.3012640476226807,
-      "learning_rate": 8.541306532663317e-05,
-      "loss": 6.0431,
-      "step": 15018
-    },
-    {
-      "epoch": 7.832594524119948,
-      "grad_norm": 1.3550634384155273,
-      "learning_rate": 8.541206030150755e-05,
-      "loss": 5.7213,
-      "step": 15019
-    },
-    {
-      "epoch": 7.833116036505867,
-      "grad_norm": 1.5413992404937744,
-      "learning_rate": 8.541105527638191e-05,
-      "loss": 5.1833,
-      "step": 15020
-    },
-    {
-      "epoch": 7.833637548891787,
-      "grad_norm": 1.3638540506362915,
-      "learning_rate": 8.541005025125629e-05,
-      "loss": 5.5447,
-      "step": 15021
-    },
-    {
-      "epoch": 7.834159061277705,
-      "grad_norm": 1.4122185707092285,
-      "learning_rate": 8.540904522613066e-05,
-      "loss": 5.8244,
-      "step": 15022
-    },
-    {
-      "epoch": 7.834680573663625,
-      "grad_norm": 1.6213678121566772,
-      "learning_rate": 8.540804020100502e-05,
-      "loss": 5.4667,
-      "step": 15023
-    },
-    {
-      "epoch": 7.835202086049543,
-      "grad_norm": 1.5544401407241821,
-      "learning_rate": 8.54070351758794e-05,
-      "loss": 5.622,
-      "step": 15024
-    },
-    {
-      "epoch": 7.835723598435463,
-      "grad_norm": 1.3995805978775024,
-      "learning_rate": 8.540603015075376e-05,
-      "loss": 5.679,
-      "step": 15025
-    },
-    {
-      "epoch": 7.836245110821382,
-      "grad_norm": 1.2829991579055786,
-      "learning_rate": 8.540502512562814e-05,
-      "loss": 5.8957,
-      "step": 15026
-    },
-    {
-      "epoch": 7.836766623207302,
-      "grad_norm": 1.494748592376709,
-      "learning_rate": 8.540402010050252e-05,
-      "loss": 5.2327,
-      "step": 15027
-    },
-    {
-      "epoch": 7.83728813559322,
-      "grad_norm": 1.6106172800064087,
-      "learning_rate": 8.54030150753769e-05,
-      "loss": 5.2702,
-      "step": 15028
-    },
-    {
-      "epoch": 7.83780964797914,
-      "grad_norm": 1.4900389909744263,
-      "learning_rate": 8.540201005025126e-05,
-      "loss": 5.4684,
-      "step": 15029
-    },
-    {
-      "epoch": 7.838331160365058,
-      "grad_norm": 1.3579721450805664,
-      "learning_rate": 8.540100502512564e-05,
-      "loss": 5.5901,
-      "step": 15030
-    },
-    {
-      "epoch": 7.838852672750978,
-      "grad_norm": 1.5330694913864136,
-      "learning_rate": 8.54e-05,
-      "loss": 5.103,
-      "step": 15031
-    },
-    {
-      "epoch": 7.839374185136897,
-      "grad_norm": 1.4363458156585693,
-      "learning_rate": 8.539899497487438e-05,
-      "loss": 5.6268,
-      "step": 15032
-    },
-    {
-      "epoch": 7.839895697522816,
-      "grad_norm": 1.6081088781356812,
-      "learning_rate": 8.539798994974874e-05,
-      "loss": 5.439,
-      "step": 15033
-    },
-    {
-      "epoch": 7.840417209908735,
-      "grad_norm": 1.3614498376846313,
-      "learning_rate": 8.539698492462312e-05,
-      "loss": 5.1323,
-      "step": 15034
-    },
-    {
-      "epoch": 7.840938722294655,
-      "grad_norm": 1.5176000595092773,
-      "learning_rate": 8.539597989949749e-05,
-      "loss": 5.6593,
-      "step": 15035
-    },
-    {
-      "epoch": 7.841460234680573,
-      "grad_norm": 1.502511978149414,
-      "learning_rate": 8.539497487437186e-05,
-      "loss": 5.8224,
-      "step": 15036
-    },
-    {
-      "epoch": 7.841981747066493,
-      "grad_norm": 1.3678221702575684,
-      "learning_rate": 8.539396984924624e-05,
-      "loss": 5.6818,
-      "step": 15037
-    },
-    {
-      "epoch": 7.842503259452412,
-      "grad_norm": 1.548999547958374,
-      "learning_rate": 8.53929648241206e-05,
-      "loss": 5.5301,
-      "step": 15038
-    },
-    {
-      "epoch": 7.843024771838331,
-      "grad_norm": 1.56722891330719,
-      "learning_rate": 8.539195979899498e-05,
-      "loss": 5.5254,
-      "step": 15039
-    },
-    {
-      "epoch": 7.84354628422425,
-      "grad_norm": 1.4623483419418335,
-      "learning_rate": 8.539095477386935e-05,
-      "loss": 5.6195,
-      "step": 15040
-    },
-    {
-      "epoch": 7.84406779661017,
-      "grad_norm": 1.4051355123519897,
-      "learning_rate": 8.538994974874373e-05,
-      "loss": 5.6393,
-      "step": 15041
-    },
-    {
-      "epoch": 7.844589308996088,
-      "grad_norm": 1.5261871814727783,
-      "learning_rate": 8.538894472361809e-05,
-      "loss": 4.897,
-      "step": 15042
-    },
-    {
-      "epoch": 7.845110821382008,
-      "grad_norm": 1.5428394079208374,
-      "learning_rate": 8.538793969849247e-05,
-      "loss": 5.4516,
-      "step": 15043
-    },
-    {
-      "epoch": 7.845632333767927,
-      "grad_norm": 1.5369852781295776,
-      "learning_rate": 8.538693467336683e-05,
-      "loss": 5.4793,
-      "step": 15044
-    },
-    {
-      "epoch": 7.846153846153846,
-      "grad_norm": 1.4259412288665771,
-      "learning_rate": 8.538592964824121e-05,
-      "loss": 5.7292,
-      "step": 15045
-    },
-    {
-      "epoch": 7.846675358539765,
-      "grad_norm": 1.391097903251648,
-      "learning_rate": 8.538492462311557e-05,
-      "loss": 6.0522,
-      "step": 15046
-    },
-    {
-      "epoch": 7.847196870925685,
-      "grad_norm": 1.5544253587722778,
-      "learning_rate": 8.538391959798995e-05,
-      "loss": 5.4924,
-      "step": 15047
-    },
-    {
-      "epoch": 7.847718383311603,
-      "grad_norm": 1.4836848974227905,
-      "learning_rate": 8.538291457286433e-05,
-      "loss": 5.4672,
-      "step": 15048
-    },
-    {
-      "epoch": 7.848239895697523,
-      "grad_norm": 1.4115604162216187,
-      "learning_rate": 8.53819095477387e-05,
-      "loss": 5.5548,
-      "step": 15049
-    },
-    {
-      "epoch": 7.848761408083442,
-      "grad_norm": 1.4422063827514648,
-      "learning_rate": 8.538090452261307e-05,
-      "loss": 5.5241,
-      "step": 15050
-    },
-    {
-      "epoch": 7.849282920469361,
-      "grad_norm": 1.9059169292449951,
-      "learning_rate": 8.537989949748744e-05,
-      "loss": 5.2593,
-      "step": 15051
-    },
-    {
-      "epoch": 7.84980443285528,
-      "grad_norm": 1.6694374084472656,
-      "learning_rate": 8.537889447236181e-05,
-      "loss": 5.5679,
-      "step": 15052
-    },
-    {
-      "epoch": 7.8503259452412,
-      "grad_norm": 1.4957221746444702,
-      "learning_rate": 8.537788944723618e-05,
-      "loss": 5.1374,
-      "step": 15053
-    },
-    {
-      "epoch": 7.850847457627118,
-      "grad_norm": 1.4089535474777222,
-      "learning_rate": 8.537688442211056e-05,
-      "loss": 5.709,
-      "step": 15054
-    },
-    {
-      "epoch": 7.851368970013038,
-      "grad_norm": 1.2785613536834717,
-      "learning_rate": 8.537587939698492e-05,
-      "loss": 5.8871,
-      "step": 15055
-    },
-    {
-      "epoch": 7.851890482398957,
-      "grad_norm": 1.4616209268569946,
-      "learning_rate": 8.53748743718593e-05,
-      "loss": 5.6016,
-      "step": 15056
-    },
-    {
-      "epoch": 7.852411994784876,
-      "grad_norm": 1.4319452047348022,
-      "learning_rate": 8.537386934673368e-05,
-      "loss": 5.6519,
-      "step": 15057
-    },
-    {
-      "epoch": 7.852933507170795,
-      "grad_norm": 1.5094350576400757,
-      "learning_rate": 8.537286432160805e-05,
-      "loss": 5.0269,
-      "step": 15058
-    },
-    {
-      "epoch": 7.853455019556715,
-      "grad_norm": 1.488564372062683,
-      "learning_rate": 8.537185929648242e-05,
-      "loss": 5.4592,
-      "step": 15059
-    },
-    {
-      "epoch": 7.853976531942633,
-      "grad_norm": 1.3258891105651855,
-      "learning_rate": 8.53708542713568e-05,
-      "loss": 5.2956,
-      "step": 15060
-    },
-    {
-      "epoch": 7.854498044328553,
-      "grad_norm": 1.536194086074829,
-      "learning_rate": 8.536984924623116e-05,
-      "loss": 5.4889,
-      "step": 15061
-    },
-    {
-      "epoch": 7.855019556714472,
-      "grad_norm": 1.4162681102752686,
-      "learning_rate": 8.536884422110554e-05,
-      "loss": 5.9623,
-      "step": 15062
-    },
-    {
-      "epoch": 7.855541069100391,
-      "grad_norm": 1.4789527654647827,
-      "learning_rate": 8.53678391959799e-05,
-      "loss": 5.5669,
-      "step": 15063
-    },
-    {
-      "epoch": 7.85606258148631,
-      "grad_norm": 1.4228535890579224,
-      "learning_rate": 8.536683417085427e-05,
-      "loss": 5.0022,
-      "step": 15064
-    },
-    {
-      "epoch": 7.85658409387223,
-      "grad_norm": 1.528040885925293,
-      "learning_rate": 8.536582914572865e-05,
-      "loss": 5.5102,
-      "step": 15065
-    },
-    {
-      "epoch": 7.857105606258148,
-      "grad_norm": 1.3114972114562988,
-      "learning_rate": 8.536482412060301e-05,
-      "loss": 6.0327,
-      "step": 15066
-    },
-    {
-      "epoch": 7.857627118644068,
-      "grad_norm": 1.3635048866271973,
-      "learning_rate": 8.536381909547739e-05,
-      "loss": 5.7595,
-      "step": 15067
-    },
-    {
-      "epoch": 7.858148631029987,
-      "grad_norm": 1.3831486701965332,
-      "learning_rate": 8.536281407035177e-05,
-      "loss": 5.3657,
-      "step": 15068
-    },
-    {
-      "epoch": 7.858670143415906,
-      "grad_norm": 1.4145101308822632,
-      "learning_rate": 8.536180904522614e-05,
-      "loss": 5.8606,
-      "step": 15069
-    },
-    {
-      "epoch": 7.859191655801825,
-      "grad_norm": 1.4955402612686157,
-      "learning_rate": 8.536080402010051e-05,
-      "loss": 5.5973,
-      "step": 15070
-    },
-    {
-      "epoch": 7.859713168187745,
-      "grad_norm": 1.379913091659546,
-      "learning_rate": 8.535979899497489e-05,
-      "loss": 5.827,
-      "step": 15071
-    },
-    {
-      "epoch": 7.860234680573663,
-      "grad_norm": 1.4982719421386719,
-      "learning_rate": 8.535879396984925e-05,
-      "loss": 5.3341,
-      "step": 15072
-    },
-    {
-      "epoch": 7.860756192959583,
-      "grad_norm": 1.4902962446212769,
-      "learning_rate": 8.535778894472363e-05,
-      "loss": 5.4331,
-      "step": 15073
-    },
-    {
-      "epoch": 7.861277705345502,
-      "grad_norm": 1.463169813156128,
-      "learning_rate": 8.535678391959799e-05,
-      "loss": 5.4924,
-      "step": 15074
-    },
-    {
-      "epoch": 7.861799217731421,
-      "grad_norm": 1.4087530374526978,
-      "learning_rate": 8.535577889447237e-05,
-      "loss": 5.5269,
-      "step": 15075
-    },
-    {
-      "epoch": 7.86232073011734,
-      "grad_norm": 1.3833088874816895,
-      "learning_rate": 8.535477386934673e-05,
-      "loss": 5.557,
-      "step": 15076
-    },
-    {
-      "epoch": 7.86284224250326,
-      "grad_norm": 1.4652096033096313,
-      "learning_rate": 8.535376884422111e-05,
-      "loss": 5.6565,
-      "step": 15077
-    },
-    {
-      "epoch": 7.863363754889178,
-      "grad_norm": 1.461596965789795,
-      "learning_rate": 8.535276381909549e-05,
-      "loss": 5.86,
-      "step": 15078
-    },
-    {
-      "epoch": 7.863885267275098,
-      "grad_norm": 1.4087460041046143,
-      "learning_rate": 8.535175879396985e-05,
-      "loss": 5.4279,
-      "step": 15079
-    },
-    {
-      "epoch": 7.864406779661017,
-      "grad_norm": 1.4727836847305298,
-      "learning_rate": 8.535075376884423e-05,
-      "loss": 5.4585,
-      "step": 15080
-    },
-    {
-      "epoch": 7.864928292046936,
-      "grad_norm": 1.4426206350326538,
-      "learning_rate": 8.53497487437186e-05,
-      "loss": 5.5214,
-      "step": 15081
-    },
-    {
-      "epoch": 7.865449804432855,
-      "grad_norm": 1.5194305181503296,
-      "learning_rate": 8.534874371859297e-05,
-      "loss": 5.4417,
-      "step": 15082
-    },
-    {
-      "epoch": 7.865971316818775,
-      "grad_norm": 1.4611155986785889,
-      "learning_rate": 8.534773869346734e-05,
-      "loss": 5.8036,
-      "step": 15083
-    },
-    {
-      "epoch": 7.866492829204693,
-      "grad_norm": 1.2946679592132568,
-      "learning_rate": 8.534673366834172e-05,
-      "loss": 5.8575,
-      "step": 15084
-    },
-    {
-      "epoch": 7.867014341590613,
-      "grad_norm": 1.3799494504928589,
-      "learning_rate": 8.534572864321608e-05,
-      "loss": 5.5527,
-      "step": 15085
-    },
-    {
-      "epoch": 7.867535853976532,
-      "grad_norm": 1.3568342924118042,
-      "learning_rate": 8.534472361809046e-05,
-      "loss": 5.3962,
-      "step": 15086
-    },
-    {
-      "epoch": 7.868057366362451,
-      "grad_norm": 1.3647531270980835,
-      "learning_rate": 8.534371859296482e-05,
-      "loss": 5.8666,
-      "step": 15087
-    },
-    {
-      "epoch": 7.86857887874837,
-      "grad_norm": 1.4006540775299072,
-      "learning_rate": 8.53427135678392e-05,
-      "loss": 4.536,
-      "step": 15088
-    },
-    {
-      "epoch": 7.869100391134289,
-      "grad_norm": 1.3616218566894531,
-      "learning_rate": 8.534170854271358e-05,
-      "loss": 4.4246,
-      "step": 15089
-    },
-    {
-      "epoch": 7.869621903520208,
-      "grad_norm": 1.4096369743347168,
-      "learning_rate": 8.534070351758794e-05,
-      "loss": 5.2451,
-      "step": 15090
-    },
-    {
-      "epoch": 7.870143415906128,
-      "grad_norm": 1.5202655792236328,
-      "learning_rate": 8.533969849246232e-05,
-      "loss": 5.8854,
-      "step": 15091
-    },
-    {
-      "epoch": 7.870664928292047,
-      "grad_norm": 1.396000862121582,
-      "learning_rate": 8.533869346733668e-05,
-      "loss": 5.5469,
-      "step": 15092
-    },
-    {
-      "epoch": 7.871186440677966,
-      "grad_norm": 1.4148433208465576,
-      "learning_rate": 8.533768844221106e-05,
-      "loss": 5.707,
-      "step": 15093
-    },
-    {
-      "epoch": 7.871707953063885,
-      "grad_norm": 1.358188509941101,
-      "learning_rate": 8.533668341708543e-05,
-      "loss": 5.6775,
-      "step": 15094
-    },
-    {
-      "epoch": 7.872229465449804,
-      "grad_norm": 1.3573997020721436,
-      "learning_rate": 8.53356783919598e-05,
-      "loss": 5.9248,
-      "step": 15095
-    },
-    {
-      "epoch": 7.872750977835723,
-      "grad_norm": 1.4630261659622192,
-      "learning_rate": 8.533467336683417e-05,
-      "loss": 5.3223,
-      "step": 15096
-    },
-    {
-      "epoch": 7.873272490221643,
-      "grad_norm": 1.6047700643539429,
-      "learning_rate": 8.533366834170855e-05,
-      "loss": 5.0501,
-      "step": 15097
-    },
-    {
-      "epoch": 7.873794002607562,
-      "grad_norm": 1.3787583112716675,
-      "learning_rate": 8.533266331658292e-05,
-      "loss": 5.7106,
-      "step": 15098
-    },
-    {
-      "epoch": 7.874315514993481,
-      "grad_norm": 1.4319475889205933,
-      "learning_rate": 8.53316582914573e-05,
-      "loss": 5.7008,
-      "step": 15099
-    },
-    {
-      "epoch": 7.8748370273794,
-      "grad_norm": 1.501732587814331,
-      "learning_rate": 8.533065326633167e-05,
-      "loss": 5.1752,
-      "step": 15100
-    },
-    {
-      "epoch": 7.875358539765319,
-      "grad_norm": 1.4385813474655151,
-      "learning_rate": 8.532964824120604e-05,
-      "loss": 5.9613,
-      "step": 15101
-    },
-    {
-      "epoch": 7.875880052151238,
-      "grad_norm": 1.4393730163574219,
-      "learning_rate": 8.532864321608041e-05,
-      "loss": 5.453,
-      "step": 15102
-    },
-    {
-      "epoch": 7.876401564537158,
-      "grad_norm": 1.3807787895202637,
-      "learning_rate": 8.532763819095477e-05,
-      "loss": 5.4495,
-      "step": 15103
-    },
-    {
-      "epoch": 7.876923076923077,
-      "grad_norm": 1.4870423078536987,
-      "learning_rate": 8.532663316582915e-05,
-      "loss": 5.1602,
-      "step": 15104
-    },
-    {
-      "epoch": 7.877444589308996,
-      "grad_norm": 1.4968969821929932,
-      "learning_rate": 8.532562814070351e-05,
-      "loss": 5.196,
-      "step": 15105
-    },
-    {
-      "epoch": 7.877966101694915,
-      "grad_norm": 1.4364585876464844,
-      "learning_rate": 8.532462311557789e-05,
-      "loss": 5.4205,
-      "step": 15106
-    },
-    {
-      "epoch": 7.878487614080834,
-      "grad_norm": 1.5904603004455566,
-      "learning_rate": 8.532361809045226e-05,
-      "loss": 5.3567,
-      "step": 15107
-    },
-    {
-      "epoch": 7.879009126466753,
-      "grad_norm": 1.4066187143325806,
-      "learning_rate": 8.532261306532663e-05,
-      "loss": 5.4079,
-      "step": 15108
-    },
-    {
-      "epoch": 7.879530638852673,
-      "grad_norm": 1.4495388269424438,
-      "learning_rate": 8.532160804020101e-05,
-      "loss": 5.956,
-      "step": 15109
-    },
-    {
-      "epoch": 7.880052151238592,
-      "grad_norm": 1.6409300565719604,
-      "learning_rate": 8.532060301507539e-05,
-      "loss": 5.1347,
-      "step": 15110
-    },
-    {
-      "epoch": 7.880573663624511,
-      "grad_norm": 1.520734190940857,
-      "learning_rate": 8.531959798994975e-05,
-      "loss": 5.0023,
-      "step": 15111
-    },
-    {
-      "epoch": 7.88109517601043,
-      "grad_norm": 1.4600203037261963,
-      "learning_rate": 8.531859296482413e-05,
-      "loss": 5.5362,
-      "step": 15112
-    },
-    {
-      "epoch": 7.881616688396349,
-      "grad_norm": 1.3994910717010498,
-      "learning_rate": 8.53175879396985e-05,
-      "loss": 5.5924,
-      "step": 15113
-    },
-    {
-      "epoch": 7.8821382007822685,
-      "grad_norm": 1.2789793014526367,
-      "learning_rate": 8.531658291457287e-05,
-      "loss": 5.384,
-      "step": 15114
-    },
-    {
-      "epoch": 7.882659713168188,
-      "grad_norm": 1.4117274284362793,
-      "learning_rate": 8.531557788944724e-05,
-      "loss": 5.6372,
-      "step": 15115
-    },
-    {
-      "epoch": 7.883181225554107,
-      "grad_norm": 1.39469575881958,
-      "learning_rate": 8.53145728643216e-05,
-      "loss": 5.8044,
-      "step": 15116
-    },
-    {
-      "epoch": 7.883702737940026,
-      "grad_norm": 1.4595617055892944,
-      "learning_rate": 8.531356783919598e-05,
-      "loss": 5.55,
-      "step": 15117
-    },
-    {
-      "epoch": 7.884224250325945,
-      "grad_norm": 1.6286848783493042,
-      "learning_rate": 8.531256281407034e-05,
-      "loss": 4.5939,
-      "step": 15118
-    },
-    {
-      "epoch": 7.884745762711864,
-      "grad_norm": 1.3540353775024414,
-      "learning_rate": 8.531155778894472e-05,
-      "loss": 5.837,
-      "step": 15119
-    },
-    {
-      "epoch": 7.8852672750977835,
-      "grad_norm": 1.4304085969924927,
-      "learning_rate": 8.53105527638191e-05,
-      "loss": 5.3405,
-      "step": 15120
-    },
-    {
-      "epoch": 7.885788787483703,
-      "grad_norm": 1.350635051727295,
-      "learning_rate": 8.530954773869348e-05,
-      "loss": 5.5468,
-      "step": 15121
-    },
-    {
-      "epoch": 7.886310299869622,
-      "grad_norm": 1.5186350345611572,
-      "learning_rate": 8.530854271356784e-05,
-      "loss": 5.1778,
-      "step": 15122
-    },
-    {
-      "epoch": 7.886831812255541,
-      "grad_norm": 1.4620275497436523,
-      "learning_rate": 8.530753768844222e-05,
-      "loss": 5.5753,
-      "step": 15123
-    },
-    {
-      "epoch": 7.88735332464146,
-      "grad_norm": 1.4612905979156494,
-      "learning_rate": 8.530653266331658e-05,
-      "loss": 5.1761,
-      "step": 15124
-    },
-    {
-      "epoch": 7.887874837027379,
-      "grad_norm": 1.4512892961502075,
-      "learning_rate": 8.530552763819096e-05,
-      "loss": 5.5808,
-      "step": 15125
-    },
-    {
-      "epoch": 7.8883963494132985,
-      "grad_norm": 1.3227237462997437,
-      "learning_rate": 8.530452261306533e-05,
-      "loss": 4.9647,
-      "step": 15126
-    },
-    {
-      "epoch": 7.888917861799218,
-      "grad_norm": 1.318312406539917,
-      "learning_rate": 8.53035175879397e-05,
-      "loss": 5.7042,
-      "step": 15127
-    },
-    {
-      "epoch": 7.8894393741851365,
-      "grad_norm": 1.3888776302337646,
-      "learning_rate": 8.530251256281407e-05,
-      "loss": 5.8986,
-      "step": 15128
-    },
-    {
-      "epoch": 7.889960886571056,
-      "grad_norm": 1.4443271160125732,
-      "learning_rate": 8.530150753768845e-05,
-      "loss": 5.4043,
-      "step": 15129
-    },
-    {
-      "epoch": 7.8904823989569755,
-      "grad_norm": 1.4924161434173584,
-      "learning_rate": 8.530050251256282e-05,
-      "loss": 5.5869,
-      "step": 15130
-    },
-    {
-      "epoch": 7.891003911342894,
-      "grad_norm": 1.273324728012085,
-      "learning_rate": 8.529949748743719e-05,
-      "loss": 5.7011,
-      "step": 15131
-    },
-    {
-      "epoch": 7.8915254237288135,
-      "grad_norm": 1.4935715198516846,
-      "learning_rate": 8.529849246231157e-05,
-      "loss": 5.3492,
-      "step": 15132
-    },
-    {
-      "epoch": 7.892046936114733,
-      "grad_norm": 1.3872042894363403,
-      "learning_rate": 8.529748743718593e-05,
-      "loss": 5.7037,
-      "step": 15133
-    },
-    {
-      "epoch": 7.8925684485006515,
-      "grad_norm": 1.4551434516906738,
-      "learning_rate": 8.529648241206031e-05,
-      "loss": 5.3256,
-      "step": 15134
-    },
-    {
-      "epoch": 7.893089960886571,
-      "grad_norm": 1.3485249280929565,
-      "learning_rate": 8.529547738693467e-05,
-      "loss": 5.6967,
-      "step": 15135
-    },
-    {
-      "epoch": 7.8936114732724905,
-      "grad_norm": 1.5651288032531738,
-      "learning_rate": 8.529447236180905e-05,
-      "loss": 5.3123,
-      "step": 15136
-    },
-    {
-      "epoch": 7.894132985658409,
-      "grad_norm": 1.4877749681472778,
-      "learning_rate": 8.529346733668342e-05,
-      "loss": 5.5541,
-      "step": 15137
-    },
-    {
-      "epoch": 7.8946544980443285,
-      "grad_norm": 1.3793089389801025,
-      "learning_rate": 8.529246231155779e-05,
-      "loss": 5.8982,
-      "step": 15138
-    },
-    {
-      "epoch": 7.895176010430248,
-      "grad_norm": 1.5156059265136719,
-      "learning_rate": 8.529145728643216e-05,
-      "loss": 5.4214,
-      "step": 15139
-    },
-    {
-      "epoch": 7.8956975228161665,
-      "grad_norm": 1.454900860786438,
-      "learning_rate": 8.529045226130654e-05,
-      "loss": 5.7453,
-      "step": 15140
-    },
-    {
-      "epoch": 7.896219035202086,
-      "grad_norm": 1.4198014736175537,
-      "learning_rate": 8.528944723618091e-05,
-      "loss": 5.6704,
-      "step": 15141
-    },
-    {
-      "epoch": 7.8967405475880055,
-      "grad_norm": 1.3729057312011719,
-      "learning_rate": 8.528844221105528e-05,
-      "loss": 5.6246,
-      "step": 15142
-    },
-    {
-      "epoch": 7.897262059973924,
-      "grad_norm": 1.478641152381897,
-      "learning_rate": 8.528743718592965e-05,
-      "loss": 5.467,
-      "step": 15143
-    },
-    {
-      "epoch": 7.8977835723598435,
-      "grad_norm": 1.4752105474472046,
-      "learning_rate": 8.528643216080402e-05,
-      "loss": 5.2984,
-      "step": 15144
-    },
-    {
-      "epoch": 7.898305084745763,
-      "grad_norm": 1.5561600923538208,
-      "learning_rate": 8.52854271356784e-05,
-      "loss": 5.2162,
-      "step": 15145
-    },
-    {
-      "epoch": 7.898826597131682,
-      "grad_norm": 1.4026646614074707,
-      "learning_rate": 8.528442211055276e-05,
-      "loss": 5.1414,
-      "step": 15146
-    },
-    {
-      "epoch": 7.899348109517601,
-      "grad_norm": 1.4712929725646973,
-      "learning_rate": 8.528341708542714e-05,
-      "loss": 5.6309,
-      "step": 15147
-    },
-    {
-      "epoch": 7.8998696219035205,
-      "grad_norm": 1.3349515199661255,
-      "learning_rate": 8.52824120603015e-05,
-      "loss": 5.6453,
-      "step": 15148
-    },
-    {
-      "epoch": 7.900391134289439,
-      "grad_norm": 1.5216326713562012,
-      "learning_rate": 8.528140703517588e-05,
-      "loss": 5.1009,
-      "step": 15149
-    },
-    {
-      "epoch": 7.9009126466753585,
-      "grad_norm": 1.511669635772705,
-      "learning_rate": 8.528040201005026e-05,
-      "loss": 5.0473,
-      "step": 15150
-    },
-    {
-      "epoch": 7.901434159061278,
-      "grad_norm": 1.4194422960281372,
-      "learning_rate": 8.527939698492464e-05,
-      "loss": 5.5475,
-      "step": 15151
-    },
-    {
-      "epoch": 7.901955671447197,
-      "grad_norm": 1.4626944065093994,
-      "learning_rate": 8.5278391959799e-05,
-      "loss": 5.9754,
-      "step": 15152
-    },
-    {
-      "epoch": 7.902477183833116,
-      "grad_norm": 1.4818096160888672,
-      "learning_rate": 8.527738693467338e-05,
-      "loss": 5.3183,
-      "step": 15153
-    },
-    {
-      "epoch": 7.9029986962190355,
-      "grad_norm": 1.3330494165420532,
-      "learning_rate": 8.527638190954774e-05,
-      "loss": 5.7167,
-      "step": 15154
-    },
-    {
-      "epoch": 7.903520208604954,
-      "grad_norm": 1.3322207927703857,
-      "learning_rate": 8.527537688442212e-05,
-      "loss": 5.8472,
-      "step": 15155
-    },
-    {
-      "epoch": 7.9040417209908735,
-      "grad_norm": 1.426382303237915,
-      "learning_rate": 8.527437185929649e-05,
-      "loss": 5.7864,
-      "step": 15156
-    },
-    {
-      "epoch": 7.904563233376793,
-      "grad_norm": 1.382879376411438,
-      "learning_rate": 8.527336683417085e-05,
-      "loss": 5.6453,
-      "step": 15157
-    },
-    {
-      "epoch": 7.905084745762712,
-      "grad_norm": 1.369033932685852,
-      "learning_rate": 8.527236180904523e-05,
-      "loss": 5.5535,
-      "step": 15158
-    },
-    {
-      "epoch": 7.905606258148631,
-      "grad_norm": 1.4164831638336182,
-      "learning_rate": 8.527135678391959e-05,
-      "loss": 5.7967,
-      "step": 15159
-    },
-    {
-      "epoch": 7.9061277705345505,
-      "grad_norm": 1.4145272970199585,
-      "learning_rate": 8.527035175879397e-05,
-      "loss": 5.5071,
-      "step": 15160
-    },
-    {
-      "epoch": 7.906649282920469,
-      "grad_norm": 1.4025137424468994,
-      "learning_rate": 8.526934673366835e-05,
-      "loss": 5.9705,
-      "step": 15161
-    },
-    {
-      "epoch": 7.9071707953063886,
-      "grad_norm": 1.6208710670471191,
-      "learning_rate": 8.526834170854273e-05,
-      "loss": 4.9091,
-      "step": 15162
-    },
-    {
-      "epoch": 7.907692307692308,
-      "grad_norm": 1.4814720153808594,
-      "learning_rate": 8.526733668341709e-05,
-      "loss": 5.6513,
-      "step": 15163
-    },
-    {
-      "epoch": 7.908213820078227,
-      "grad_norm": 1.49767005443573,
-      "learning_rate": 8.526633165829147e-05,
-      "loss": 5.6786,
-      "step": 15164
-    },
-    {
-      "epoch": 7.908735332464146,
-      "grad_norm": 1.4896337985992432,
-      "learning_rate": 8.526532663316583e-05,
-      "loss": 5.8827,
-      "step": 15165
-    },
-    {
-      "epoch": 7.9092568448500655,
-      "grad_norm": 1.4572288990020752,
-      "learning_rate": 8.526432160804021e-05,
-      "loss": 5.241,
-      "step": 15166
-    },
-    {
-      "epoch": 7.909778357235984,
-      "grad_norm": 1.257376790046692,
-      "learning_rate": 8.526331658291457e-05,
-      "loss": 5.3694,
-      "step": 15167
-    },
-    {
-      "epoch": 7.910299869621904,
-      "grad_norm": 1.6026183366775513,
-      "learning_rate": 8.526231155778895e-05,
-      "loss": 5.2162,
-      "step": 15168
-    },
-    {
-      "epoch": 7.910821382007823,
-      "grad_norm": 1.5042181015014648,
-      "learning_rate": 8.526130653266332e-05,
-      "loss": 5.3845,
-      "step": 15169
-    },
-    {
-      "epoch": 7.911342894393742,
-      "grad_norm": 1.6723257303237915,
-      "learning_rate": 8.52603015075377e-05,
-      "loss": 5.4323,
-      "step": 15170
-    },
-    {
-      "epoch": 7.911864406779661,
-      "grad_norm": 1.412706732749939,
-      "learning_rate": 8.525929648241207e-05,
-      "loss": 5.3419,
-      "step": 15171
-    },
-    {
-      "epoch": 7.9123859191655805,
-      "grad_norm": 1.4601646661758423,
-      "learning_rate": 8.525829145728644e-05,
-      "loss": 5.8551,
-      "step": 15172
-    },
-    {
-      "epoch": 7.912907431551499,
-      "grad_norm": 1.5339144468307495,
-      "learning_rate": 8.525728643216081e-05,
-      "loss": 4.904,
-      "step": 15173
-    },
-    {
-      "epoch": 7.913428943937419,
-      "grad_norm": 1.5483357906341553,
-      "learning_rate": 8.525628140703518e-05,
-      "loss": 5.4289,
-      "step": 15174
-    },
-    {
-      "epoch": 7.913950456323338,
-      "grad_norm": 1.4401049613952637,
-      "learning_rate": 8.525527638190956e-05,
-      "loss": 5.8031,
-      "step": 15175
-    },
-    {
-      "epoch": 7.914471968709257,
-      "grad_norm": 1.4061312675476074,
-      "learning_rate": 8.525427135678392e-05,
-      "loss": 5.1282,
-      "step": 15176
-    },
-    {
-      "epoch": 7.914993481095176,
-      "grad_norm": 1.3567041158676147,
-      "learning_rate": 8.52532663316583e-05,
-      "loss": 5.6051,
-      "step": 15177
-    },
-    {
-      "epoch": 7.9155149934810956,
-      "grad_norm": 1.4950910806655884,
-      "learning_rate": 8.525226130653266e-05,
-      "loss": 5.5834,
-      "step": 15178
-    },
-    {
-      "epoch": 7.916036505867014,
-      "grad_norm": 1.4400360584259033,
-      "learning_rate": 8.525125628140704e-05,
-      "loss": 5.6582,
-      "step": 15179
-    },
-    {
-      "epoch": 7.916558018252934,
-      "grad_norm": 1.5387053489685059,
-      "learning_rate": 8.52502512562814e-05,
-      "loss": 5.2724,
-      "step": 15180
-    },
-    {
-      "epoch": 7.917079530638853,
-      "grad_norm": 1.4269241094589233,
-      "learning_rate": 8.524924623115578e-05,
-      "loss": 5.3669,
-      "step": 15181
-    },
-    {
-      "epoch": 7.917601043024772,
-      "grad_norm": 1.4306923151016235,
-      "learning_rate": 8.524824120603016e-05,
-      "loss": 5.446,
-      "step": 15182
-    },
-    {
-      "epoch": 7.918122555410691,
-      "grad_norm": 1.3901735544204712,
-      "learning_rate": 8.524723618090452e-05,
-      "loss": 5.4479,
-      "step": 15183
-    },
-    {
-      "epoch": 7.91864406779661,
-      "grad_norm": 1.539788007736206,
-      "learning_rate": 8.52462311557789e-05,
-      "loss": 5.4663,
-      "step": 15184
-    },
-    {
-      "epoch": 7.919165580182529,
-      "grad_norm": 1.4476512670516968,
-      "learning_rate": 8.524522613065327e-05,
-      "loss": 5.6078,
-      "step": 15185
-    },
-    {
-      "epoch": 7.919687092568449,
-      "grad_norm": 1.5279911756515503,
-      "learning_rate": 8.524422110552764e-05,
-      "loss": 5.214,
-      "step": 15186
-    },
-    {
-      "epoch": 7.920208604954368,
-      "grad_norm": 1.4672143459320068,
-      "learning_rate": 8.524321608040201e-05,
-      "loss": 5.6883,
-      "step": 15187
-    },
-    {
-      "epoch": 7.920730117340287,
-      "grad_norm": 1.6178632974624634,
-      "learning_rate": 8.524221105527639e-05,
-      "loss": 5.2662,
-      "step": 15188
-    },
-    {
-      "epoch": 7.921251629726206,
-      "grad_norm": 1.4577678442001343,
-      "learning_rate": 8.524120603015075e-05,
-      "loss": 5.2795,
-      "step": 15189
-    },
-    {
-      "epoch": 7.921773142112125,
-      "grad_norm": 1.5147814750671387,
-      "learning_rate": 8.524020100502513e-05,
-      "loss": 5.3702,
-      "step": 15190
-    },
-    {
-      "epoch": 7.922294654498044,
-      "grad_norm": 1.6658542156219482,
-      "learning_rate": 8.52391959798995e-05,
-      "loss": 4.8294,
-      "step": 15191
-    },
-    {
-      "epoch": 7.922816166883964,
-      "grad_norm": 1.4964237213134766,
-      "learning_rate": 8.523819095477388e-05,
-      "loss": 5.3909,
-      "step": 15192
-    },
-    {
-      "epoch": 7.923337679269883,
-      "grad_norm": 1.4004024267196655,
-      "learning_rate": 8.523718592964825e-05,
-      "loss": 5.0614,
-      "step": 15193
-    },
-    {
-      "epoch": 7.923859191655802,
-      "grad_norm": 1.4363291263580322,
-      "learning_rate": 8.523618090452263e-05,
-      "loss": 5.6777,
-      "step": 15194
-    },
-    {
-      "epoch": 7.924380704041721,
-      "grad_norm": 1.3250662088394165,
-      "learning_rate": 8.523517587939699e-05,
-      "loss": 5.894,
-      "step": 15195
-    },
-    {
-      "epoch": 7.92490221642764,
-      "grad_norm": 1.3632279634475708,
-      "learning_rate": 8.523417085427135e-05,
-      "loss": 5.8331,
-      "step": 15196
-    },
-    {
-      "epoch": 7.925423728813559,
-      "grad_norm": 1.5371427536010742,
-      "learning_rate": 8.523316582914573e-05,
-      "loss": 5.1806,
-      "step": 15197
-    },
-    {
-      "epoch": 7.925945241199479,
-      "grad_norm": 1.4039794206619263,
-      "learning_rate": 8.52321608040201e-05,
-      "loss": 5.5982,
-      "step": 15198
-    },
-    {
-      "epoch": 7.926466753585398,
-      "grad_norm": 1.425973892211914,
-      "learning_rate": 8.523115577889447e-05,
-      "loss": 5.7676,
-      "step": 15199
-    },
-    {
-      "epoch": 7.926988265971317,
-      "grad_norm": 1.4697774648666382,
-      "learning_rate": 8.523015075376884e-05,
-      "loss": 5.4149,
-      "step": 15200
-    },
-    {
-      "epoch": 7.927509778357236,
-      "grad_norm": 1.3331897258758545,
-      "learning_rate": 8.522914572864322e-05,
-      "loss": 5.9802,
-      "step": 15201
-    },
-    {
-      "epoch": 7.928031290743155,
-      "grad_norm": 1.3886746168136597,
-      "learning_rate": 8.52281407035176e-05,
-      "loss": 5.3149,
-      "step": 15202
-    },
-    {
-      "epoch": 7.928552803129074,
-      "grad_norm": 1.3997042179107666,
-      "learning_rate": 8.522713567839197e-05,
-      "loss": 5.9363,
-      "step": 15203
-    },
-    {
-      "epoch": 7.929074315514994,
-      "grad_norm": 1.5172057151794434,
-      "learning_rate": 8.522613065326634e-05,
-      "loss": 5.2101,
-      "step": 15204
-    },
-    {
-      "epoch": 7.929595827900913,
-      "grad_norm": 1.4293462038040161,
-      "learning_rate": 8.522512562814071e-05,
-      "loss": 5.5653,
-      "step": 15205
-    },
-    {
-      "epoch": 7.930117340286832,
-      "grad_norm": 1.517549753189087,
-      "learning_rate": 8.522412060301508e-05,
-      "loss": 5.2287,
-      "step": 15206
-    },
-    {
-      "epoch": 7.930638852672751,
-      "grad_norm": 1.5149362087249756,
-      "learning_rate": 8.522311557788946e-05,
-      "loss": 4.5762,
-      "step": 15207
-    },
-    {
-      "epoch": 7.93116036505867,
-      "grad_norm": 1.5136775970458984,
-      "learning_rate": 8.522211055276382e-05,
-      "loss": 5.4389,
-      "step": 15208
-    },
-    {
-      "epoch": 7.931681877444589,
-      "grad_norm": 1.54416823387146,
-      "learning_rate": 8.522110552763819e-05,
-      "loss": 5.1876,
-      "step": 15209
-    },
-    {
-      "epoch": 7.932203389830509,
-      "grad_norm": 1.5124694108963013,
-      "learning_rate": 8.522010050251256e-05,
-      "loss": 4.9563,
-      "step": 15210
-    },
-    {
-      "epoch": 7.932724902216428,
-      "grad_norm": 1.5802602767944336,
-      "learning_rate": 8.521909547738694e-05,
-      "loss": 4.8955,
-      "step": 15211
-    },
-    {
-      "epoch": 7.933246414602347,
-      "grad_norm": 1.409733533859253,
-      "learning_rate": 8.521809045226132e-05,
-      "loss": 5.9149,
-      "step": 15212
-    },
-    {
-      "epoch": 7.933767926988266,
-      "grad_norm": 1.3385579586029053,
-      "learning_rate": 8.521708542713568e-05,
-      "loss": 5.2984,
-      "step": 15213
-    },
-    {
-      "epoch": 7.934289439374185,
-      "grad_norm": 1.5719866752624512,
-      "learning_rate": 8.521608040201006e-05,
-      "loss": 4.4866,
-      "step": 15214
-    },
-    {
-      "epoch": 7.934810951760104,
-      "grad_norm": 1.3830783367156982,
-      "learning_rate": 8.521507537688442e-05,
-      "loss": 5.5335,
-      "step": 15215
-    },
-    {
-      "epoch": 7.935332464146024,
-      "grad_norm": 1.3493869304656982,
-      "learning_rate": 8.52140703517588e-05,
-      "loss": 5.1096,
-      "step": 15216
-    },
-    {
-      "epoch": 7.935853976531943,
-      "grad_norm": 1.394181728363037,
-      "learning_rate": 8.521306532663317e-05,
-      "loss": 5.7092,
-      "step": 15217
-    },
-    {
-      "epoch": 7.936375488917862,
-      "grad_norm": 1.3165771961212158,
-      "learning_rate": 8.521206030150754e-05,
-      "loss": 4.3173,
-      "step": 15218
-    },
-    {
-      "epoch": 7.936897001303781,
-      "grad_norm": 1.4170275926589966,
-      "learning_rate": 8.521105527638191e-05,
-      "loss": 5.4556,
-      "step": 15219
-    },
-    {
-      "epoch": 7.9374185136897,
-      "grad_norm": 1.3206665515899658,
-      "learning_rate": 8.521005025125629e-05,
-      "loss": 5.699,
-      "step": 15220
-    },
-    {
-      "epoch": 7.937940026075619,
-      "grad_norm": 1.4654545783996582,
-      "learning_rate": 8.520904522613065e-05,
-      "loss": 5.3956,
-      "step": 15221
-    },
-    {
-      "epoch": 7.938461538461539,
-      "grad_norm": 1.3279049396514893,
-      "learning_rate": 8.520804020100503e-05,
-      "loss": 5.9053,
-      "step": 15222
-    },
-    {
-      "epoch": 7.938983050847457,
-      "grad_norm": 1.6796492338180542,
-      "learning_rate": 8.520703517587941e-05,
-      "loss": 5.3831,
-      "step": 15223
-    },
-    {
-      "epoch": 7.939504563233377,
-      "grad_norm": 1.3637524843215942,
-      "learning_rate": 8.520603015075377e-05,
-      "loss": 5.6448,
-      "step": 15224
-    },
-    {
-      "epoch": 7.940026075619296,
-      "grad_norm": 1.4378116130828857,
-      "learning_rate": 8.520502512562815e-05,
-      "loss": 5.6229,
-      "step": 15225
-    },
-    {
-      "epoch": 7.940547588005215,
-      "grad_norm": 1.4071587324142456,
-      "learning_rate": 8.520402010050251e-05,
-      "loss": 5.7901,
-      "step": 15226
-    },
-    {
-      "epoch": 7.941069100391134,
-      "grad_norm": 1.399428367614746,
-      "learning_rate": 8.520301507537689e-05,
-      "loss": 5.3863,
-      "step": 15227
-    },
-    {
-      "epoch": 7.941590612777054,
-      "grad_norm": 1.6730403900146484,
-      "learning_rate": 8.520201005025126e-05,
-      "loss": 5.0204,
-      "step": 15228
-    },
-    {
-      "epoch": 7.942112125162972,
-      "grad_norm": 1.6284486055374146,
-      "learning_rate": 8.520100502512563e-05,
-      "loss": 4.9125,
-      "step": 15229
-    },
-    {
-      "epoch": 7.942633637548892,
-      "grad_norm": 1.3896194696426392,
-      "learning_rate": 8.52e-05,
-      "loss": 5.8081,
-      "step": 15230
-    },
-    {
-      "epoch": 7.943155149934811,
-      "grad_norm": 1.5234479904174805,
-      "learning_rate": 8.519899497487438e-05,
-      "loss": 5.317,
-      "step": 15231
-    },
-    {
-      "epoch": 7.94367666232073,
-      "grad_norm": 1.3589720726013184,
-      "learning_rate": 8.519798994974875e-05,
-      "loss": 5.8638,
-      "step": 15232
-    },
-    {
-      "epoch": 7.944198174706649,
-      "grad_norm": 1.5559979677200317,
-      "learning_rate": 8.519698492462313e-05,
-      "loss": 5.0755,
-      "step": 15233
-    },
-    {
-      "epoch": 7.944719687092569,
-      "grad_norm": 1.623522400856018,
-      "learning_rate": 8.51959798994975e-05,
-      "loss": 5.3817,
-      "step": 15234
-    },
-    {
-      "epoch": 7.945241199478487,
-      "grad_norm": 1.4860962629318237,
-      "learning_rate": 8.519497487437186e-05,
-      "loss": 5.4415,
-      "step": 15235
-    },
-    {
-      "epoch": 7.945762711864407,
-      "grad_norm": 1.375150442123413,
-      "learning_rate": 8.519396984924624e-05,
-      "loss": 5.8364,
-      "step": 15236
-    },
-    {
-      "epoch": 7.946284224250326,
-      "grad_norm": 1.5239845514297485,
-      "learning_rate": 8.51929648241206e-05,
-      "loss": 5.1835,
-      "step": 15237
-    },
-    {
-      "epoch": 7.946805736636245,
-      "grad_norm": 1.393642544746399,
-      "learning_rate": 8.519195979899498e-05,
-      "loss": 6.0036,
-      "step": 15238
-    },
-    {
-      "epoch": 7.947327249022164,
-      "grad_norm": 1.6617746353149414,
-      "learning_rate": 8.519095477386934e-05,
-      "loss": 5.0826,
-      "step": 15239
-    },
-    {
-      "epoch": 7.947848761408084,
-      "grad_norm": 1.4281734228134155,
-      "learning_rate": 8.518994974874372e-05,
-      "loss": 5.6755,
-      "step": 15240
-    },
-    {
-      "epoch": 7.948370273794002,
-      "grad_norm": 1.451132893562317,
-      "learning_rate": 8.518894472361809e-05,
-      "loss": 5.5549,
-      "step": 15241
-    },
-    {
-      "epoch": 7.948891786179922,
-      "grad_norm": 1.3334482908248901,
-      "learning_rate": 8.518793969849246e-05,
-      "loss": 5.3643,
-      "step": 15242
-    },
-    {
-      "epoch": 7.949413298565841,
-      "grad_norm": 1.4052555561065674,
-      "learning_rate": 8.518693467336684e-05,
-      "loss": 5.5226,
-      "step": 15243
-    },
-    {
-      "epoch": 7.94993481095176,
-      "grad_norm": 1.3117587566375732,
-      "learning_rate": 8.518592964824122e-05,
-      "loss": 5.2404,
-      "step": 15244
-    },
-    {
-      "epoch": 7.950456323337679,
-      "grad_norm": 1.4296751022338867,
-      "learning_rate": 8.518492462311558e-05,
-      "loss": 5.3118,
-      "step": 15245
-    },
-    {
-      "epoch": 7.950977835723599,
-      "grad_norm": 1.4126226902008057,
-      "learning_rate": 8.518391959798996e-05,
-      "loss": 5.6899,
-      "step": 15246
-    },
-    {
-      "epoch": 7.951499348109517,
-      "grad_norm": 1.4016408920288086,
-      "learning_rate": 8.518291457286433e-05,
-      "loss": 5.4898,
-      "step": 15247
-    },
-    {
-      "epoch": 7.952020860495437,
-      "grad_norm": 1.5280274152755737,
-      "learning_rate": 8.51819095477387e-05,
-      "loss": 5.7939,
-      "step": 15248
-    },
-    {
-      "epoch": 7.952542372881356,
-      "grad_norm": 1.5310653448104858,
-      "learning_rate": 8.518090452261307e-05,
-      "loss": 5.4623,
-      "step": 15249
-    },
-    {
-      "epoch": 7.953063885267275,
-      "grad_norm": 1.4558191299438477,
-      "learning_rate": 8.517989949748743e-05,
-      "loss": 5.7817,
-      "step": 15250
-    },
-    {
-      "epoch": 7.953585397653194,
-      "grad_norm": 1.4668272733688354,
-      "learning_rate": 8.517889447236181e-05,
-      "loss": 5.7941,
-      "step": 15251
-    },
-    {
-      "epoch": 7.954106910039114,
-      "grad_norm": 1.4545966386795044,
-      "learning_rate": 8.517788944723619e-05,
-      "loss": 5.5657,
-      "step": 15252
-    },
-    {
-      "epoch": 7.954628422425032,
-      "grad_norm": 1.4763559103012085,
-      "learning_rate": 8.517688442211057e-05,
-      "loss": 5.7165,
-      "step": 15253
-    },
-    {
-      "epoch": 7.955149934810952,
-      "grad_norm": 1.331691026687622,
-      "learning_rate": 8.517587939698493e-05,
-      "loss": 5.7758,
-      "step": 15254
-    },
-    {
-      "epoch": 7.955671447196871,
-      "grad_norm": 1.5719778537750244,
-      "learning_rate": 8.517487437185931e-05,
-      "loss": 5.1272,
-      "step": 15255
-    },
-    {
-      "epoch": 7.95619295958279,
-      "grad_norm": 1.61163330078125,
-      "learning_rate": 8.517386934673367e-05,
-      "loss": 5.4316,
-      "step": 15256
-    },
-    {
-      "epoch": 7.956714471968709,
-      "grad_norm": 1.4955809116363525,
-      "learning_rate": 8.517286432160805e-05,
-      "loss": 5.5667,
-      "step": 15257
-    },
-    {
-      "epoch": 7.957235984354629,
-      "grad_norm": 1.492393970489502,
-      "learning_rate": 8.517185929648241e-05,
-      "loss": 5.0672,
-      "step": 15258
-    },
-    {
-      "epoch": 7.957757496740547,
-      "grad_norm": 1.394473671913147,
-      "learning_rate": 8.517085427135679e-05,
-      "loss": 5.2745,
-      "step": 15259
-    },
-    {
-      "epoch": 7.958279009126467,
-      "grad_norm": 1.5231703519821167,
-      "learning_rate": 8.516984924623116e-05,
-      "loss": 5.2811,
-      "step": 15260
-    },
-    {
-      "epoch": 7.958800521512386,
-      "grad_norm": 1.4779448509216309,
-      "learning_rate": 8.516884422110553e-05,
-      "loss": 5.0882,
-      "step": 15261
-    },
-    {
-      "epoch": 7.959322033898305,
-      "grad_norm": 1.3228049278259277,
-      "learning_rate": 8.51678391959799e-05,
-      "loss": 5.9125,
-      "step": 15262
-    },
-    {
-      "epoch": 7.959843546284224,
-      "grad_norm": 1.433498740196228,
-      "learning_rate": 8.516683417085428e-05,
-      "loss": 5.6853,
-      "step": 15263
-    },
-    {
-      "epoch": 7.960365058670144,
-      "grad_norm": 1.2500802278518677,
-      "learning_rate": 8.516582914572865e-05,
-      "loss": 5.8688,
-      "step": 15264
-    },
-    {
-      "epoch": 7.960886571056062,
-      "grad_norm": 1.4490346908569336,
-      "learning_rate": 8.516482412060302e-05,
-      "loss": 5.6436,
-      "step": 15265
-    },
-    {
-      "epoch": 7.961408083441982,
-      "grad_norm": 1.535935878753662,
-      "learning_rate": 8.51638190954774e-05,
-      "loss": 5.4192,
-      "step": 15266
-    },
-    {
-      "epoch": 7.961929595827901,
-      "grad_norm": 1.5511960983276367,
-      "learning_rate": 8.516281407035176e-05,
-      "loss": 5.2033,
-      "step": 15267
-    },
-    {
-      "epoch": 7.96245110821382,
-      "grad_norm": 1.6857048273086548,
-      "learning_rate": 8.516180904522614e-05,
-      "loss": 4.7125,
-      "step": 15268
-    },
-    {
-      "epoch": 7.962972620599739,
-      "grad_norm": 1.3394516706466675,
-      "learning_rate": 8.51608040201005e-05,
-      "loss": 5.7048,
-      "step": 15269
-    },
-    {
-      "epoch": 7.963494132985659,
-      "grad_norm": 1.423370122909546,
-      "learning_rate": 8.515979899497488e-05,
-      "loss": 5.6357,
-      "step": 15270
-    },
-    {
-      "epoch": 7.964015645371577,
-      "grad_norm": 1.4120911359786987,
-      "learning_rate": 8.515879396984924e-05,
-      "loss": 5.558,
-      "step": 15271
-    },
-    {
-      "epoch": 7.964537157757497,
-      "grad_norm": 1.4099324941635132,
-      "learning_rate": 8.515778894472362e-05,
-      "loss": 5.4853,
-      "step": 15272
-    },
-    {
-      "epoch": 7.965058670143416,
-      "grad_norm": 1.3890247344970703,
-      "learning_rate": 8.515678391959799e-05,
-      "loss": 5.8892,
-      "step": 15273
-    },
-    {
-      "epoch": 7.965580182529335,
-      "grad_norm": 1.496289610862732,
-      "learning_rate": 8.515577889447236e-05,
-      "loss": 5.577,
-      "step": 15274
-    },
-    {
-      "epoch": 7.966101694915254,
-      "grad_norm": 1.3353215456008911,
-      "learning_rate": 8.515477386934674e-05,
-      "loss": 5.7682,
-      "step": 15275
-    },
-    {
-      "epoch": 7.966623207301174,
-      "grad_norm": 1.3960866928100586,
-      "learning_rate": 8.51537688442211e-05,
-      "loss": 5.8876,
-      "step": 15276
-    },
-    {
-      "epoch": 7.967144719687092,
-      "grad_norm": 1.4779362678527832,
-      "learning_rate": 8.515276381909548e-05,
-      "loss": 5.8568,
-      "step": 15277
-    },
-    {
-      "epoch": 7.967666232073012,
-      "grad_norm": 1.4771121740341187,
-      "learning_rate": 8.515175879396985e-05,
-      "loss": 5.7406,
-      "step": 15278
-    },
-    {
-      "epoch": 7.96818774445893,
-      "grad_norm": 1.441556453704834,
-      "learning_rate": 8.515075376884423e-05,
-      "loss": 5.584,
-      "step": 15279
-    },
-    {
-      "epoch": 7.96870925684485,
-      "grad_norm": 1.5398660898208618,
-      "learning_rate": 8.514974874371859e-05,
-      "loss": 5.6864,
-      "step": 15280
-    },
-    {
-      "epoch": 7.969230769230769,
-      "grad_norm": 1.3906174898147583,
-      "learning_rate": 8.514874371859297e-05,
-      "loss": 4.9637,
-      "step": 15281
-    },
-    {
-      "epoch": 7.969752281616689,
-      "grad_norm": 1.4767365455627441,
-      "learning_rate": 8.514773869346733e-05,
-      "loss": 5.6841,
-      "step": 15282
-    },
-    {
-      "epoch": 7.970273794002607,
-      "grad_norm": 1.4871493577957153,
-      "learning_rate": 8.514673366834171e-05,
-      "loss": 5.6768,
-      "step": 15283
-    },
-    {
-      "epoch": 7.970795306388527,
-      "grad_norm": 1.4496350288391113,
-      "learning_rate": 8.514572864321609e-05,
-      "loss": 5.7647,
-      "step": 15284
-    },
-    {
-      "epoch": 7.971316818774445,
-      "grad_norm": 1.571225643157959,
-      "learning_rate": 8.514472361809047e-05,
-      "loss": 5.3357,
-      "step": 15285
-    },
-    {
-      "epoch": 7.971838331160365,
-      "grad_norm": 1.4347412586212158,
-      "learning_rate": 8.514371859296483e-05,
-      "loss": 5.4819,
-      "step": 15286
-    },
-    {
-      "epoch": 7.972359843546284,
-      "grad_norm": 1.4508397579193115,
-      "learning_rate": 8.514271356783921e-05,
-      "loss": 5.5535,
-      "step": 15287
-    },
-    {
-      "epoch": 7.972881355932204,
-      "grad_norm": 1.8567384481430054,
-      "learning_rate": 8.514170854271357e-05,
-      "loss": 5.2025,
-      "step": 15288
-    },
-    {
-      "epoch": 7.973402868318122,
-      "grad_norm": 1.5012658834457397,
-      "learning_rate": 8.514070351758794e-05,
-      "loss": 5.6018,
-      "step": 15289
-    },
-    {
-      "epoch": 7.973924380704042,
-      "grad_norm": 1.3899730443954468,
-      "learning_rate": 8.513969849246231e-05,
-      "loss": 5.4807,
-      "step": 15290
-    },
-    {
-      "epoch": 7.97444589308996,
-      "grad_norm": 1.4612985849380493,
-      "learning_rate": 8.513869346733668e-05,
-      "loss": 5.4457,
-      "step": 15291
-    },
-    {
-      "epoch": 7.97496740547588,
-      "grad_norm": 1.4013341665267944,
-      "learning_rate": 8.513768844221106e-05,
-      "loss": 5.5916,
-      "step": 15292
-    },
-    {
-      "epoch": 7.975488917861799,
-      "grad_norm": 1.5835975408554077,
-      "learning_rate": 8.513668341708542e-05,
-      "loss": 5.4713,
-      "step": 15293
-    },
-    {
-      "epoch": 7.976010430247719,
-      "grad_norm": 1.3606103658676147,
-      "learning_rate": 8.51356783919598e-05,
-      "loss": 5.7688,
-      "step": 15294
-    },
-    {
-      "epoch": 7.976531942633637,
-      "grad_norm": 1.4576640129089355,
-      "learning_rate": 8.513467336683418e-05,
-      "loss": 5.743,
-      "step": 15295
-    },
-    {
-      "epoch": 7.977053455019557,
-      "grad_norm": 1.417160987854004,
-      "learning_rate": 8.513366834170855e-05,
-      "loss": 5.8171,
-      "step": 15296
-    },
-    {
-      "epoch": 7.9775749674054754,
-      "grad_norm": 1.2952017784118652,
-      "learning_rate": 8.513266331658292e-05,
-      "loss": 5.8054,
-      "step": 15297
-    },
-    {
-      "epoch": 7.978096479791395,
-      "grad_norm": 1.3735994100570679,
-      "learning_rate": 8.51316582914573e-05,
-      "loss": 5.3766,
-      "step": 15298
-    },
-    {
-      "epoch": 7.978617992177314,
-      "grad_norm": 1.5863746404647827,
-      "learning_rate": 8.513065326633166e-05,
-      "loss": 5.2855,
-      "step": 15299
-    },
-    {
-      "epoch": 7.979139504563234,
-      "grad_norm": 1.5093586444854736,
-      "learning_rate": 8.512964824120604e-05,
-      "loss": 5.7473,
-      "step": 15300
-    },
-    {
-      "epoch": 7.979661016949152,
-      "grad_norm": 1.3143222332000732,
-      "learning_rate": 8.51286432160804e-05,
-      "loss": 5.46,
-      "step": 15301
-    },
-    {
-      "epoch": 7.980182529335072,
-      "grad_norm": 1.551090121269226,
-      "learning_rate": 8.512763819095477e-05,
-      "loss": 5.1317,
-      "step": 15302
-    },
-    {
-      "epoch": 7.9807040417209905,
-      "grad_norm": 1.493478536605835,
-      "learning_rate": 8.512663316582915e-05,
-      "loss": 5.3051,
-      "step": 15303
-    },
-    {
-      "epoch": 7.98122555410691,
-      "grad_norm": 1.488033413887024,
-      "learning_rate": 8.512562814070352e-05,
-      "loss": 5.6546,
-      "step": 15304
-    },
-    {
-      "epoch": 7.981747066492829,
-      "grad_norm": 1.5889873504638672,
-      "learning_rate": 8.51246231155779e-05,
-      "loss": 4.5752,
-      "step": 15305
-    },
-    {
-      "epoch": 7.982268578878749,
-      "grad_norm": 1.4546301364898682,
-      "learning_rate": 8.512361809045227e-05,
-      "loss": 5.2777,
-      "step": 15306
-    },
-    {
-      "epoch": 7.982790091264667,
-      "grad_norm": 1.295559048652649,
-      "learning_rate": 8.512261306532664e-05,
-      "loss": 5.8592,
-      "step": 15307
-    },
-    {
-      "epoch": 7.983311603650587,
-      "grad_norm": 1.630520224571228,
-      "learning_rate": 8.512160804020101e-05,
-      "loss": 5.3112,
-      "step": 15308
-    },
-    {
-      "epoch": 7.9838331160365055,
-      "grad_norm": 1.4489284753799438,
-      "learning_rate": 8.512060301507539e-05,
-      "loss": 5.6933,
-      "step": 15309
-    },
-    {
-      "epoch": 7.984354628422425,
-      "grad_norm": 1.3093754053115845,
-      "learning_rate": 8.511959798994975e-05,
-      "loss": 5.8363,
-      "step": 15310
-    },
-    {
-      "epoch": 7.984876140808344,
-      "grad_norm": 1.3743376731872559,
-      "learning_rate": 8.511859296482413e-05,
-      "loss": 5.2697,
-      "step": 15311
-    },
-    {
-      "epoch": 7.985397653194263,
-      "grad_norm": 1.4283168315887451,
-      "learning_rate": 8.511758793969849e-05,
-      "loss": 5.7218,
-      "step": 15312
-    },
-    {
-      "epoch": 7.985919165580182,
-      "grad_norm": 1.4341809749603271,
-      "learning_rate": 8.511658291457287e-05,
-      "loss": 5.497,
-      "step": 15313
-    },
-    {
-      "epoch": 7.986440677966102,
-      "grad_norm": 1.3436769247055054,
-      "learning_rate": 8.511557788944723e-05,
-      "loss": 5.5615,
-      "step": 15314
-    },
-    {
-      "epoch": 7.9869621903520205,
-      "grad_norm": 1.420452356338501,
-      "learning_rate": 8.511457286432161e-05,
-      "loss": 5.3189,
-      "step": 15315
-    },
-    {
-      "epoch": 7.98748370273794,
-      "grad_norm": 1.559373140335083,
-      "learning_rate": 8.511356783919599e-05,
-      "loss": 5.6767,
-      "step": 15316
-    },
-    {
-      "epoch": 7.988005215123859,
-      "grad_norm": 1.3048617839813232,
-      "learning_rate": 8.511256281407035e-05,
-      "loss": 6.0294,
-      "step": 15317
-    },
-    {
-      "epoch": 7.988526727509778,
-      "grad_norm": 1.6018431186676025,
-      "learning_rate": 8.511155778894473e-05,
-      "loss": 5.421,
-      "step": 15318
-    },
-    {
-      "epoch": 7.9890482398956975,
-      "grad_norm": 1.4375603199005127,
-      "learning_rate": 8.51105527638191e-05,
-      "loss": 5.8983,
-      "step": 15319
-    },
-    {
-      "epoch": 7.989569752281617,
-      "grad_norm": 1.5743921995162964,
-      "learning_rate": 8.510954773869347e-05,
-      "loss": 5.5944,
-      "step": 15320
-    },
-    {
-      "epoch": 7.9900912646675355,
-      "grad_norm": 1.795737385749817,
-      "learning_rate": 8.510854271356784e-05,
-      "loss": 5.2687,
-      "step": 15321
-    },
-    {
-      "epoch": 7.990612777053455,
-      "grad_norm": 1.385797381401062,
-      "learning_rate": 8.510753768844222e-05,
-      "loss": 5.6302,
-      "step": 15322
-    },
-    {
-      "epoch": 7.991134289439374,
-      "grad_norm": 1.385337233543396,
-      "learning_rate": 8.510653266331658e-05,
-      "loss": 5.7713,
-      "step": 15323
-    },
-    {
-      "epoch": 7.991655801825293,
-      "grad_norm": 1.3908530473709106,
-      "learning_rate": 8.510552763819096e-05,
-      "loss": 5.4136,
-      "step": 15324
-    },
-    {
-      "epoch": 7.9921773142112125,
-      "grad_norm": 1.4376829862594604,
-      "learning_rate": 8.510452261306534e-05,
-      "loss": 5.673,
-      "step": 15325
-    },
-    {
-      "epoch": 7.992698826597132,
-      "grad_norm": 1.315650463104248,
-      "learning_rate": 8.510351758793971e-05,
-      "loss": 6.0592,
-      "step": 15326
-    },
-    {
-      "epoch": 7.9932203389830505,
-      "grad_norm": 1.445202112197876,
-      "learning_rate": 8.510251256281408e-05,
-      "loss": 5.3184,
-      "step": 15327
-    },
-    {
-      "epoch": 7.99374185136897,
-      "grad_norm": 1.4960308074951172,
-      "learning_rate": 8.510150753768844e-05,
-      "loss": 5.4449,
-      "step": 15328
-    },
-    {
-      "epoch": 7.994263363754889,
-      "grad_norm": 1.4911420345306396,
-      "learning_rate": 8.510050251256282e-05,
-      "loss": 5.1792,
-      "step": 15329
-    },
-    {
-      "epoch": 7.994784876140808,
-      "grad_norm": 1.3966928720474243,
-      "learning_rate": 8.509949748743718e-05,
-      "loss": 5.6183,
-      "step": 15330
-    },
-    {
-      "epoch": 7.9953063885267275,
-      "grad_norm": 1.4473296403884888,
-      "learning_rate": 8.509849246231156e-05,
-      "loss": 5.5198,
-      "step": 15331
-    },
-    {
-      "epoch": 7.995827900912647,
-      "grad_norm": 1.639607310295105,
-      "learning_rate": 8.509748743718593e-05,
-      "loss": 5.4604,
-      "step": 15332
-    },
-    {
-      "epoch": 7.9963494132985655,
-      "grad_norm": 1.513789415359497,
-      "learning_rate": 8.50964824120603e-05,
-      "loss": 5.764,
-      "step": 15333
-    },
-    {
-      "epoch": 7.996870925684485,
-      "grad_norm": 1.4958255290985107,
-      "learning_rate": 8.509547738693467e-05,
-      "loss": 5.0624,
-      "step": 15334
-    },
-    {
-      "epoch": 7.9973924380704045,
-      "grad_norm": 1.4431947469711304,
-      "learning_rate": 8.509447236180905e-05,
-      "loss": 5.7611,
-      "step": 15335
-    },
-    {
-      "epoch": 7.997913950456323,
-      "grad_norm": 1.5927302837371826,
-      "learning_rate": 8.509346733668342e-05,
-      "loss": 5.2051,
-      "step": 15336
-    },
-    {
-      "epoch": 7.9984354628422425,
-      "grad_norm": 1.6910638809204102,
-      "learning_rate": 8.50924623115578e-05,
-      "loss": 5.5017,
-      "step": 15337
-    },
-    {
-      "epoch": 7.998956975228162,
-      "grad_norm": 1.3044342994689941,
-      "learning_rate": 8.509145728643217e-05,
-      "loss": 5.9232,
-      "step": 15338
-    },
-    {
-      "epoch": 7.9994784876140805,
-      "grad_norm": 1.3677867650985718,
-      "learning_rate": 8.509045226130654e-05,
-      "loss": 5.5916,
-      "step": 15339
-    },
-    {
-      "epoch": 8.0,
-      "grad_norm": 1.5175011157989502,
-      "learning_rate": 8.508944723618091e-05,
-      "loss": 5.7852,
-      "step": 15340
-    },
-    {
-      "epoch": 8.00052151238592,
-      "grad_norm": 1.390304684638977,
-      "learning_rate": 8.508844221105529e-05,
-      "loss": 5.5628,
-      "step": 15341
-    },
-    {
-      "epoch": 8.001043024771839,
-      "grad_norm": 1.446747064590454,
-      "learning_rate": 8.508743718592965e-05,
-      "loss": 5.1767,
-      "step": 15342
-    },
-    {
-      "epoch": 8.001564537157757,
-      "grad_norm": 1.4341310262680054,
-      "learning_rate": 8.508643216080401e-05,
-      "loss": 5.7307,
-      "step": 15343
-    },
-    {
-      "epoch": 8.002086049543676,
-      "grad_norm": 1.3088773488998413,
-      "learning_rate": 8.508542713567839e-05,
-      "loss": 5.832,
-      "step": 15344
-    },
-    {
-      "epoch": 8.002607561929596,
-      "grad_norm": 1.376611351966858,
-      "learning_rate": 8.508442211055277e-05,
-      "loss": 5.71,
-      "step": 15345
-    },
-    {
-      "epoch": 8.003129074315515,
-      "grad_norm": 1.4603737592697144,
-      "learning_rate": 8.508341708542715e-05,
-      "loss": 5.3734,
-      "step": 15346
-    },
-    {
-      "epoch": 8.003650586701434,
-      "grad_norm": 1.4773428440093994,
-      "learning_rate": 8.508241206030151e-05,
-      "loss": 5.7034,
-      "step": 15347
-    },
-    {
-      "epoch": 8.004172099087354,
-      "grad_norm": 1.3950238227844238,
-      "learning_rate": 8.508140703517589e-05,
-      "loss": 5.489,
-      "step": 15348
-    },
-    {
-      "epoch": 8.004693611473272,
-      "grad_norm": 1.3618615865707397,
-      "learning_rate": 8.508040201005025e-05,
-      "loss": 5.4327,
-      "step": 15349
-    },
-    {
-      "epoch": 8.005215123859191,
-      "grad_norm": 1.5947985649108887,
-      "learning_rate": 8.507939698492463e-05,
-      "loss": 5.4367,
-      "step": 15350
-    },
-    {
-      "epoch": 8.00573663624511,
-      "grad_norm": 1.4083995819091797,
-      "learning_rate": 8.5078391959799e-05,
-      "loss": 5.853,
-      "step": 15351
-    },
-    {
-      "epoch": 8.00625814863103,
-      "grad_norm": 1.5465725660324097,
-      "learning_rate": 8.507738693467337e-05,
-      "loss": 5.5842,
-      "step": 15352
-    },
-    {
-      "epoch": 8.00677966101695,
-      "grad_norm": 1.5305765867233276,
-      "learning_rate": 8.507638190954774e-05,
-      "loss": 5.7323,
-      "step": 15353
-    },
-    {
-      "epoch": 8.007301173402869,
-      "grad_norm": 1.5905872583389282,
-      "learning_rate": 8.507537688442212e-05,
-      "loss": 5.4104,
-      "step": 15354
-    },
-    {
-      "epoch": 8.007822685788787,
-      "grad_norm": 1.66057550907135,
-      "learning_rate": 8.507437185929648e-05,
-      "loss": 5.3539,
-      "step": 15355
-    },
-    {
-      "epoch": 8.008344198174706,
-      "grad_norm": 1.5634052753448486,
-      "learning_rate": 8.507336683417086e-05,
-      "loss": 5.3184,
-      "step": 15356
-    },
-    {
-      "epoch": 8.008865710560626,
-      "grad_norm": 1.5202713012695312,
-      "learning_rate": 8.507236180904524e-05,
-      "loss": 5.5682,
-      "step": 15357
-    },
-    {
-      "epoch": 8.009387222946545,
-      "grad_norm": 1.3338435888290405,
-      "learning_rate": 8.50713567839196e-05,
-      "loss": 5.784,
-      "step": 15358
-    },
-    {
-      "epoch": 8.009908735332465,
-      "grad_norm": 1.3778048753738403,
-      "learning_rate": 8.507035175879398e-05,
-      "loss": 5.3451,
-      "step": 15359
-    },
-    {
-      "epoch": 8.010430247718384,
-      "grad_norm": 1.5769524574279785,
-      "learning_rate": 8.506934673366834e-05,
-      "loss": 5.4301,
-      "step": 15360
-    },
-    {
-      "epoch": 8.010951760104302,
-      "grad_norm": 1.6457512378692627,
-      "learning_rate": 8.506834170854272e-05,
-      "loss": 5.336,
-      "step": 15361
-    },
-    {
-      "epoch": 8.011473272490221,
-      "grad_norm": 1.4528099298477173,
-      "learning_rate": 8.506733668341708e-05,
-      "loss": 5.4424,
-      "step": 15362
-    },
-    {
-      "epoch": 8.01199478487614,
-      "grad_norm": 1.3624287843704224,
-      "learning_rate": 8.506633165829146e-05,
-      "loss": 5.1465,
-      "step": 15363
-    },
-    {
-      "epoch": 8.01251629726206,
-      "grad_norm": 1.390187382698059,
-      "learning_rate": 8.506532663316583e-05,
-      "loss": 5.8776,
-      "step": 15364
-    },
-    {
-      "epoch": 8.01303780964798,
-      "grad_norm": 1.5628948211669922,
-      "learning_rate": 8.50643216080402e-05,
-      "loss": 5.3956,
-      "step": 15365
-    },
-    {
-      "epoch": 8.013559322033899,
-      "grad_norm": 1.4415775537490845,
-      "learning_rate": 8.506331658291458e-05,
-      "loss": 5.8267,
-      "step": 15366
-    },
-    {
-      "epoch": 8.014080834419817,
-      "grad_norm": 1.4305692911148071,
-      "learning_rate": 8.506231155778896e-05,
-      "loss": 5.6353,
-      "step": 15367
-    },
-    {
-      "epoch": 8.014602346805736,
-      "grad_norm": 1.8618881702423096,
-      "learning_rate": 8.506130653266332e-05,
-      "loss": 4.9697,
-      "step": 15368
-    },
-    {
-      "epoch": 8.015123859191656,
-      "grad_norm": 1.418805480003357,
-      "learning_rate": 8.506030150753769e-05,
-      "loss": 5.7107,
-      "step": 15369
-    },
-    {
-      "epoch": 8.015645371577575,
-      "grad_norm": 1.3926386833190918,
-      "learning_rate": 8.505929648241207e-05,
-      "loss": 5.7812,
-      "step": 15370
-    },
-    {
-      "epoch": 8.016166883963495,
-      "grad_norm": 1.564953327178955,
-      "learning_rate": 8.505829145728643e-05,
-      "loss": 5.0464,
-      "step": 15371
-    },
-    {
-      "epoch": 8.016688396349414,
-      "grad_norm": 1.4493554830551147,
-      "learning_rate": 8.505728643216081e-05,
-      "loss": 5.635,
-      "step": 15372
-    },
-    {
-      "epoch": 8.017209908735332,
-      "grad_norm": 1.5577951669692993,
-      "learning_rate": 8.505628140703517e-05,
-      "loss": 4.5504,
-      "step": 15373
-    },
-    {
-      "epoch": 8.017731421121251,
-      "grad_norm": 1.4878005981445312,
-      "learning_rate": 8.505527638190955e-05,
-      "loss": 5.5053,
-      "step": 15374
-    },
-    {
-      "epoch": 8.01825293350717,
-      "grad_norm": 1.4111437797546387,
-      "learning_rate": 8.505427135678392e-05,
-      "loss": 5.8786,
-      "step": 15375
-    },
-    {
-      "epoch": 8.01877444589309,
-      "grad_norm": 1.3699852228164673,
-      "learning_rate": 8.505326633165829e-05,
-      "loss": 5.6801,
-      "step": 15376
-    },
-    {
-      "epoch": 8.01929595827901,
-      "grad_norm": 1.3855986595153809,
-      "learning_rate": 8.505226130653267e-05,
-      "loss": 5.524,
-      "step": 15377
-    },
-    {
-      "epoch": 8.019817470664929,
-      "grad_norm": 1.4664092063903809,
-      "learning_rate": 8.505125628140705e-05,
-      "loss": 5.5515,
-      "step": 15378
-    },
-    {
-      "epoch": 8.020338983050847,
-      "grad_norm": 1.5992985963821411,
-      "learning_rate": 8.505025125628141e-05,
-      "loss": 5.0621,
-      "step": 15379
-    },
-    {
-      "epoch": 8.020860495436766,
-      "grad_norm": 1.3935317993164062,
-      "learning_rate": 8.504924623115579e-05,
-      "loss": 5.9358,
-      "step": 15380
-    },
-    {
-      "epoch": 8.021382007822686,
-      "grad_norm": 1.2516634464263916,
-      "learning_rate": 8.504824120603015e-05,
-      "loss": 5.0581,
-      "step": 15381
-    },
-    {
-      "epoch": 8.021903520208605,
-      "grad_norm": 1.3793138265609741,
-      "learning_rate": 8.504723618090452e-05,
-      "loss": 5.7427,
-      "step": 15382
-    },
-    {
-      "epoch": 8.022425032594525,
-      "grad_norm": 1.437124490737915,
-      "learning_rate": 8.50462311557789e-05,
-      "loss": 5.6964,
-      "step": 15383
-    },
-    {
-      "epoch": 8.022946544980444,
-      "grad_norm": 1.5285338163375854,
-      "learning_rate": 8.504522613065326e-05,
-      "loss": 5.3843,
-      "step": 15384
-    },
-    {
-      "epoch": 8.023468057366362,
-      "grad_norm": 1.4900380373001099,
-      "learning_rate": 8.504422110552764e-05,
-      "loss": 5.5768,
-      "step": 15385
-    },
-    {
-      "epoch": 8.023989569752281,
-      "grad_norm": 1.3816604614257812,
-      "learning_rate": 8.504321608040202e-05,
-      "loss": 5.6575,
-      "step": 15386
-    },
-    {
-      "epoch": 8.0245110821382,
-      "grad_norm": 1.5043259859085083,
-      "learning_rate": 8.50422110552764e-05,
-      "loss": 4.5592,
-      "step": 15387
-    },
-    {
-      "epoch": 8.02503259452412,
-      "grad_norm": 1.421183705329895,
-      "learning_rate": 8.504120603015076e-05,
-      "loss": 5.6438,
-      "step": 15388
-    },
-    {
-      "epoch": 8.02555410691004,
-      "grad_norm": 1.2909908294677734,
-      "learning_rate": 8.504020100502514e-05,
-      "loss": 5.9081,
-      "step": 15389
-    },
-    {
-      "epoch": 8.026075619295959,
-      "grad_norm": 1.41033935546875,
-      "learning_rate": 8.50391959798995e-05,
-      "loss": 5.3814,
-      "step": 15390
-    },
-    {
-      "epoch": 8.026597131681877,
-      "grad_norm": 1.4824128150939941,
-      "learning_rate": 8.503819095477388e-05,
-      "loss": 5.3274,
-      "step": 15391
-    },
-    {
-      "epoch": 8.027118644067796,
-      "grad_norm": 1.3800396919250488,
-      "learning_rate": 8.503718592964824e-05,
-      "loss": 5.5972,
-      "step": 15392
-    },
-    {
-      "epoch": 8.027640156453716,
-      "grad_norm": 1.3465046882629395,
-      "learning_rate": 8.503618090452262e-05,
-      "loss": 5.9861,
-      "step": 15393
-    },
-    {
-      "epoch": 8.028161668839635,
-      "grad_norm": 1.3072959184646606,
-      "learning_rate": 8.503517587939699e-05,
-      "loss": 5.6239,
-      "step": 15394
-    },
-    {
-      "epoch": 8.028683181225555,
-      "grad_norm": 1.6013771295547485,
-      "learning_rate": 8.503417085427135e-05,
-      "loss": 4.7791,
-      "step": 15395
-    },
-    {
-      "epoch": 8.029204693611474,
-      "grad_norm": 1.400961995124817,
-      "learning_rate": 8.503316582914573e-05,
-      "loss": 5.5055,
-      "step": 15396
-    },
-    {
-      "epoch": 8.029726205997392,
-      "grad_norm": 1.4158071279525757,
-      "learning_rate": 8.50321608040201e-05,
-      "loss": 5.3556,
-      "step": 15397
-    },
-    {
-      "epoch": 8.030247718383311,
-      "grad_norm": 1.4595032930374146,
-      "learning_rate": 8.503115577889448e-05,
-      "loss": 5.0704,
-      "step": 15398
-    },
-    {
-      "epoch": 8.03076923076923,
-      "grad_norm": 1.455165982246399,
-      "learning_rate": 8.503015075376885e-05,
-      "loss": 5.6415,
-      "step": 15399
-    },
-    {
-      "epoch": 8.03129074315515,
-      "grad_norm": 1.4311209917068481,
-      "learning_rate": 8.502914572864323e-05,
-      "loss": 5.5771,
-      "step": 15400
-    },
-    {
-      "epoch": 8.03181225554107,
-      "grad_norm": 1.413723349571228,
-      "learning_rate": 8.502814070351759e-05,
-      "loss": 5.4104,
-      "step": 15401
-    },
-    {
-      "epoch": 8.032333767926989,
-      "grad_norm": 1.4406311511993408,
-      "learning_rate": 8.502713567839197e-05,
-      "loss": 5.5958,
-      "step": 15402
-    },
-    {
-      "epoch": 8.032855280312907,
-      "grad_norm": 1.378806471824646,
-      "learning_rate": 8.502613065326633e-05,
-      "loss": 5.4055,
-      "step": 15403
-    },
-    {
-      "epoch": 8.033376792698826,
-      "grad_norm": 1.4509977102279663,
-      "learning_rate": 8.502512562814071e-05,
-      "loss": 5.4141,
-      "step": 15404
-    },
-    {
-      "epoch": 8.033898305084746,
-      "grad_norm": 1.394758701324463,
-      "learning_rate": 8.502412060301507e-05,
-      "loss": 5.4701,
-      "step": 15405
-    },
-    {
-      "epoch": 8.034419817470665,
-      "grad_norm": 1.471449613571167,
-      "learning_rate": 8.502311557788945e-05,
-      "loss": 5.3859,
-      "step": 15406
-    },
-    {
-      "epoch": 8.034941329856585,
-      "grad_norm": 1.5800950527191162,
-      "learning_rate": 8.502211055276383e-05,
-      "loss": 5.5598,
-      "step": 15407
-    },
-    {
-      "epoch": 8.035462842242504,
-      "grad_norm": 1.5112214088439941,
-      "learning_rate": 8.50211055276382e-05,
-      "loss": 5.134,
-      "step": 15408
-    },
-    {
-      "epoch": 8.035984354628422,
-      "grad_norm": 1.603190541267395,
-      "learning_rate": 8.502010050251257e-05,
-      "loss": 4.6321,
-      "step": 15409
-    },
-    {
-      "epoch": 8.036505867014341,
-      "grad_norm": 1.8280936479568481,
-      "learning_rate": 8.501909547738694e-05,
-      "loss": 5.029,
-      "step": 15410
-    },
-    {
-      "epoch": 8.03702737940026,
-      "grad_norm": 1.4177597761154175,
-      "learning_rate": 8.501809045226131e-05,
-      "loss": 5.2904,
-      "step": 15411
-    },
-    {
-      "epoch": 8.03754889178618,
-      "grad_norm": 1.3215174674987793,
-      "learning_rate": 8.501708542713568e-05,
-      "loss": 5.94,
-      "step": 15412
-    },
-    {
-      "epoch": 8.0380704041721,
-      "grad_norm": 1.4715266227722168,
-      "learning_rate": 8.501608040201006e-05,
-      "loss": 5.3254,
-      "step": 15413
-    },
-    {
-      "epoch": 8.038591916558019,
-      "grad_norm": 1.523896336555481,
-      "learning_rate": 8.501507537688442e-05,
-      "loss": 4.7039,
-      "step": 15414
-    },
-    {
-      "epoch": 8.039113428943937,
-      "grad_norm": 1.4576537609100342,
-      "learning_rate": 8.50140703517588e-05,
-      "loss": 5.1779,
-      "step": 15415
-    },
-    {
-      "epoch": 8.039634941329856,
-      "grad_norm": 1.5641121864318848,
-      "learning_rate": 8.501306532663316e-05,
-      "loss": 5.3889,
-      "step": 15416
-    },
-    {
-      "epoch": 8.040156453715776,
-      "grad_norm": 1.4114723205566406,
-      "learning_rate": 8.501206030150754e-05,
-      "loss": 5.658,
-      "step": 15417
-    },
-    {
-      "epoch": 8.040677966101695,
-      "grad_norm": 1.432917833328247,
-      "learning_rate": 8.501105527638192e-05,
-      "loss": 5.6712,
-      "step": 15418
-    },
-    {
-      "epoch": 8.041199478487615,
-      "grad_norm": 1.4624650478363037,
-      "learning_rate": 8.50100502512563e-05,
-      "loss": 5.5913,
-      "step": 15419
-    },
-    {
-      "epoch": 8.041720990873534,
-      "grad_norm": 1.6010112762451172,
-      "learning_rate": 8.500904522613066e-05,
-      "loss": 5.4416,
-      "step": 15420
-    },
-    {
-      "epoch": 8.042242503259452,
-      "grad_norm": 1.4051388502120972,
-      "learning_rate": 8.500804020100504e-05,
-      "loss": 5.8383,
-      "step": 15421
-    },
-    {
-      "epoch": 8.042764015645371,
-      "grad_norm": 1.410883903503418,
-      "learning_rate": 8.50070351758794e-05,
-      "loss": 5.663,
-      "step": 15422
-    },
-    {
-      "epoch": 8.04328552803129,
-      "grad_norm": 1.731884241104126,
-      "learning_rate": 8.500603015075377e-05,
-      "loss": 4.7957,
-      "step": 15423
-    },
-    {
-      "epoch": 8.04380704041721,
-      "grad_norm": 2.5581672191619873,
-      "learning_rate": 8.500502512562814e-05,
-      "loss": 5.4425,
-      "step": 15424
-    },
-    {
-      "epoch": 8.04432855280313,
-      "grad_norm": 1.8473352193832397,
-      "learning_rate": 8.500402010050251e-05,
-      "loss": 5.0052,
-      "step": 15425
-    },
-    {
-      "epoch": 8.044850065189049,
-      "grad_norm": 1.5868275165557861,
-      "learning_rate": 8.500301507537689e-05,
-      "loss": 5.5932,
-      "step": 15426
-    },
-    {
-      "epoch": 8.045371577574967,
-      "grad_norm": 1.4267325401306152,
-      "learning_rate": 8.500201005025126e-05,
-      "loss": 5.4899,
-      "step": 15427
-    },
-    {
-      "epoch": 8.045893089960886,
-      "grad_norm": 1.4029301404953003,
-      "learning_rate": 8.500100502512564e-05,
-      "loss": 5.5891,
-      "step": 15428
-    },
-    {
-      "epoch": 8.046414602346806,
-      "grad_norm": 1.5177831649780273,
-      "learning_rate": 8.5e-05,
-      "loss": 5.2651,
-      "step": 15429
-    },
-    {
-      "epoch": 8.046936114732725,
-      "grad_norm": 1.4123040437698364,
-      "learning_rate": 8.499899497487438e-05,
-      "loss": 5.718,
-      "step": 15430
-    },
-    {
-      "epoch": 8.047457627118645,
-      "grad_norm": 1.53366219997406,
-      "learning_rate": 8.499798994974875e-05,
-      "loss": 5.3367,
-      "step": 15431
-    },
-    {
-      "epoch": 8.047979139504564,
-      "grad_norm": 1.4641448259353638,
-      "learning_rate": 8.499698492462313e-05,
-      "loss": 5.4563,
-      "step": 15432
-    },
-    {
-      "epoch": 8.048500651890482,
-      "grad_norm": 1.5163789987564087,
-      "learning_rate": 8.499597989949749e-05,
-      "loss": 5.3743,
-      "step": 15433
-    },
-    {
-      "epoch": 8.049022164276401,
-      "grad_norm": 1.433473825454712,
-      "learning_rate": 8.499497487437187e-05,
-      "loss": 5.7143,
-      "step": 15434
-    },
-    {
-      "epoch": 8.04954367666232,
-      "grad_norm": 1.4664863348007202,
-      "learning_rate": 8.499396984924623e-05,
-      "loss": 5.6089,
-      "step": 15435
-    },
-    {
-      "epoch": 8.05006518904824,
-      "grad_norm": 1.5481555461883545,
-      "learning_rate": 8.49929648241206e-05,
-      "loss": 5.7872,
-      "step": 15436
-    },
-    {
-      "epoch": 8.05058670143416,
-      "grad_norm": 1.5192092657089233,
-      "learning_rate": 8.499195979899497e-05,
-      "loss": 5.8185,
-      "step": 15437
-    },
-    {
-      "epoch": 8.051108213820077,
-      "grad_norm": 1.4179058074951172,
-      "learning_rate": 8.499095477386935e-05,
-      "loss": 5.7085,
-      "step": 15438
-    },
-    {
-      "epoch": 8.051629726205997,
-      "grad_norm": 1.5664829015731812,
-      "learning_rate": 8.498994974874373e-05,
-      "loss": 5.1257,
-      "step": 15439
-    },
-    {
-      "epoch": 8.052151238591916,
-      "grad_norm": 1.6102935075759888,
-      "learning_rate": 8.49889447236181e-05,
-      "loss": 5.3093,
-      "step": 15440
-    },
-    {
-      "epoch": 8.052672750977836,
-      "grad_norm": 1.4769339561462402,
-      "learning_rate": 8.498793969849247e-05,
-      "loss": 5.2589,
-      "step": 15441
-    },
-    {
-      "epoch": 8.053194263363755,
-      "grad_norm": 1.4127150774002075,
-      "learning_rate": 8.498693467336684e-05,
-      "loss": 5.6652,
-      "step": 15442
-    },
-    {
-      "epoch": 8.053715775749675,
-      "grad_norm": 1.338387131690979,
-      "learning_rate": 8.498592964824121e-05,
-      "loss": 5.3556,
-      "step": 15443
-    },
-    {
-      "epoch": 8.054237288135592,
-      "grad_norm": 1.3104321956634521,
-      "learning_rate": 8.498492462311558e-05,
-      "loss": 5.7493,
-      "step": 15444
-    },
-    {
-      "epoch": 8.054758800521512,
-      "grad_norm": 1.9962481260299683,
-      "learning_rate": 8.498391959798996e-05,
-      "loss": 5.2251,
-      "step": 15445
-    },
-    {
-      "epoch": 8.055280312907431,
-      "grad_norm": 1.5775083303451538,
-      "learning_rate": 8.498291457286432e-05,
-      "loss": 5.1545,
-      "step": 15446
-    },
-    {
-      "epoch": 8.05580182529335,
-      "grad_norm": 1.5940611362457275,
-      "learning_rate": 8.49819095477387e-05,
-      "loss": 5.5182,
-      "step": 15447
-    },
-    {
-      "epoch": 8.05632333767927,
-      "grad_norm": 1.4766106605529785,
-      "learning_rate": 8.498090452261306e-05,
-      "loss": 5.6882,
-      "step": 15448
-    },
-    {
-      "epoch": 8.05684485006519,
-      "grad_norm": 1.7149903774261475,
-      "learning_rate": 8.497989949748744e-05,
-      "loss": 4.9646,
-      "step": 15449
-    },
-    {
-      "epoch": 8.057366362451107,
-      "grad_norm": 1.4124594926834106,
-      "learning_rate": 8.497889447236182e-05,
-      "loss": 5.1177,
-      "step": 15450
-    },
-    {
-      "epoch": 8.057887874837027,
-      "grad_norm": 1.469590663909912,
-      "learning_rate": 8.497788944723618e-05,
-      "loss": 4.7279,
-      "step": 15451
-    },
-    {
-      "epoch": 8.058409387222946,
-      "grad_norm": 1.5134634971618652,
-      "learning_rate": 8.497688442211056e-05,
-      "loss": 5.6592,
-      "step": 15452
-    },
-    {
-      "epoch": 8.058930899608866,
-      "grad_norm": 1.416188359260559,
-      "learning_rate": 8.497587939698492e-05,
-      "loss": 5.6155,
-      "step": 15453
-    },
-    {
-      "epoch": 8.059452411994785,
-      "grad_norm": 1.485105037689209,
-      "learning_rate": 8.49748743718593e-05,
-      "loss": 5.8069,
-      "step": 15454
-    },
-    {
-      "epoch": 8.059973924380705,
-      "grad_norm": 1.4645200967788696,
-      "learning_rate": 8.497386934673367e-05,
-      "loss": 5.0796,
-      "step": 15455
-    },
-    {
-      "epoch": 8.060495436766622,
-      "grad_norm": 1.4247392416000366,
-      "learning_rate": 8.497286432160804e-05,
-      "loss": 5.8166,
-      "step": 15456
-    },
-    {
-      "epoch": 8.061016949152542,
-      "grad_norm": 1.487082839012146,
-      "learning_rate": 8.497185929648241e-05,
-      "loss": 5.5928,
-      "step": 15457
-    },
-    {
-      "epoch": 8.061538461538461,
-      "grad_norm": 1.3990180492401123,
-      "learning_rate": 8.497085427135679e-05,
-      "loss": 5.6357,
-      "step": 15458
-    },
-    {
-      "epoch": 8.06205997392438,
-      "grad_norm": 1.3720003366470337,
-      "learning_rate": 8.496984924623116e-05,
-      "loss": 5.6588,
-      "step": 15459
-    },
-    {
-      "epoch": 8.0625814863103,
-      "grad_norm": 1.586195707321167,
-      "learning_rate": 8.496884422110554e-05,
-      "loss": 5.1869,
-      "step": 15460
-    },
-    {
-      "epoch": 8.06310299869622,
-      "grad_norm": 1.4899824857711792,
-      "learning_rate": 8.496783919597991e-05,
-      "loss": 5.6204,
-      "step": 15461
-    },
-    {
-      "epoch": 8.063624511082137,
-      "grad_norm": 1.422252893447876,
-      "learning_rate": 8.496683417085427e-05,
-      "loss": 5.5023,
-      "step": 15462
-    },
-    {
-      "epoch": 8.064146023468057,
-      "grad_norm": 1.4452879428863525,
-      "learning_rate": 8.496582914572865e-05,
-      "loss": 5.6389,
-      "step": 15463
-    },
-    {
-      "epoch": 8.064667535853976,
-      "grad_norm": 1.4618357419967651,
-      "learning_rate": 8.496482412060301e-05,
-      "loss": 5.5253,
-      "step": 15464
-    },
-    {
-      "epoch": 8.065189048239896,
-      "grad_norm": 1.4778043031692505,
-      "learning_rate": 8.496381909547739e-05,
-      "loss": 5.1973,
-      "step": 15465
-    },
-    {
-      "epoch": 8.065710560625815,
-      "grad_norm": 1.5097606182098389,
-      "learning_rate": 8.496281407035176e-05,
-      "loss": 5.6891,
-      "step": 15466
-    },
-    {
-      "epoch": 8.066232073011735,
-      "grad_norm": 1.5006811618804932,
-      "learning_rate": 8.496180904522613e-05,
-      "loss": 5.3355,
-      "step": 15467
-    },
-    {
-      "epoch": 8.066753585397652,
-      "grad_norm": 1.4825302362442017,
-      "learning_rate": 8.49608040201005e-05,
-      "loss": 5.3646,
-      "step": 15468
-    },
-    {
-      "epoch": 8.067275097783572,
-      "grad_norm": 1.3978360891342163,
-      "learning_rate": 8.495979899497488e-05,
-      "loss": 5.7727,
-      "step": 15469
-    },
-    {
-      "epoch": 8.067796610169491,
-      "grad_norm": 1.712721347808838,
-      "learning_rate": 8.495879396984925e-05,
-      "loss": 4.803,
-      "step": 15470
-    },
-    {
-      "epoch": 8.06831812255541,
-      "grad_norm": 1.4304463863372803,
-      "learning_rate": 8.495778894472363e-05,
-      "loss": 5.1077,
-      "step": 15471
-    },
-    {
-      "epoch": 8.06883963494133,
-      "grad_norm": 1.5150009393692017,
-      "learning_rate": 8.4956783919598e-05,
-      "loss": 5.1705,
-      "step": 15472
-    },
-    {
-      "epoch": 8.06936114732725,
-      "grad_norm": 1.4580217599868774,
-      "learning_rate": 8.495577889447237e-05,
-      "loss": 5.9172,
-      "step": 15473
-    },
-    {
-      "epoch": 8.069882659713167,
-      "grad_norm": 1.4772493839263916,
-      "learning_rate": 8.495477386934674e-05,
-      "loss": 5.6573,
-      "step": 15474
-    },
-    {
-      "epoch": 8.070404172099087,
-      "grad_norm": 1.4542133808135986,
-      "learning_rate": 8.49537688442211e-05,
-      "loss": 5.5825,
-      "step": 15475
-    },
-    {
-      "epoch": 8.070925684485006,
-      "grad_norm": 1.392167091369629,
-      "learning_rate": 8.495276381909548e-05,
-      "loss": 5.7194,
-      "step": 15476
-    },
-    {
-      "epoch": 8.071447196870926,
-      "grad_norm": 1.408933401107788,
-      "learning_rate": 8.495175879396984e-05,
-      "loss": 5.626,
-      "step": 15477
-    },
-    {
-      "epoch": 8.071968709256845,
-      "grad_norm": Infinity,
-      "learning_rate": 8.495175879396984e-05,
-      "loss": 5.1782,
-      "step": 15478
-    },
-    {
-      "epoch": 8.072490221642765,
-      "grad_norm": 1.4997482299804688,
-      "learning_rate": 8.495075376884422e-05,
-      "loss": 5.391,
-      "step": 15479
-    },
-    {
-      "epoch": 8.073011734028682,
-      "grad_norm": 1.5334839820861816,
-      "learning_rate": 8.49497487437186e-05,
-      "loss": 5.957,
-      "step": 15480
-    },
-    {
-      "epoch": 8.073533246414602,
-      "grad_norm": 1.5397247076034546,
-      "learning_rate": 8.494874371859298e-05,
-      "loss": 5.3534,
-      "step": 15481
-    },
-    {
-      "epoch": 8.074054758800521,
-      "grad_norm": 1.359954833984375,
-      "learning_rate": 8.494773869346734e-05,
-      "loss": 5.9303,
-      "step": 15482
-    },
-    {
-      "epoch": 8.07457627118644,
-      "grad_norm": 1.5373835563659668,
-      "learning_rate": 8.494673366834172e-05,
-      "loss": 5.5577,
-      "step": 15483
-    },
-    {
-      "epoch": 8.07509778357236,
-      "grad_norm": 1.4575697183609009,
-      "learning_rate": 8.494572864321608e-05,
-      "loss": 5.537,
-      "step": 15484
-    },
-    {
-      "epoch": 8.07561929595828,
-      "grad_norm": 1.4422193765640259,
-      "learning_rate": 8.494472361809046e-05,
-      "loss": 5.9706,
-      "step": 15485
-    },
-    {
-      "epoch": 8.076140808344197,
-      "grad_norm": 1.4497785568237305,
-      "learning_rate": 8.494371859296483e-05,
-      "loss": 5.5231,
-      "step": 15486
-    },
-    {
-      "epoch": 8.076662320730117,
-      "grad_norm": 1.369357705116272,
-      "learning_rate": 8.49427135678392e-05,
-      "loss": 5.6282,
-      "step": 15487
-    },
-    {
-      "epoch": 8.077183833116036,
-      "grad_norm": 1.682360053062439,
-      "learning_rate": 8.494170854271357e-05,
-      "loss": 5.0933,
-      "step": 15488
-    },
-    {
-      "epoch": 8.077705345501956,
-      "grad_norm": 1.4189612865447998,
-      "learning_rate": 8.494070351758793e-05,
-      "loss": 5.6542,
-      "step": 15489
-    },
-    {
-      "epoch": 8.078226857887875,
-      "grad_norm": 1.475553274154663,
-      "learning_rate": 8.493969849246231e-05,
-      "loss": 5.1978,
-      "step": 15490
-    },
-    {
-      "epoch": 8.078748370273795,
-      "grad_norm": 1.4298346042633057,
-      "learning_rate": 8.493869346733669e-05,
-      "loss": 5.5747,
-      "step": 15491
-    },
-    {
-      "epoch": 8.079269882659712,
-      "grad_norm": 1.511306881904602,
-      "learning_rate": 8.493768844221107e-05,
-      "loss": 5.3114,
-      "step": 15492
-    },
-    {
-      "epoch": 8.079791395045632,
-      "grad_norm": 1.3894777297973633,
-      "learning_rate": 8.493668341708543e-05,
-      "loss": 5.6377,
-      "step": 15493
-    },
-    {
-      "epoch": 8.080312907431551,
-      "grad_norm": 1.4410673379898071,
-      "learning_rate": 8.493567839195981e-05,
-      "loss": 5.3914,
-      "step": 15494
-    },
-    {
-      "epoch": 8.08083441981747,
-      "grad_norm": 1.3325062990188599,
-      "learning_rate": 8.493467336683417e-05,
-      "loss": 5.6608,
-      "step": 15495
-    },
-    {
-      "epoch": 8.08135593220339,
-      "grad_norm": 1.4935847520828247,
-      "learning_rate": 8.493366834170855e-05,
-      "loss": 5.7059,
-      "step": 15496
-    },
-    {
-      "epoch": 8.08187744458931,
-      "grad_norm": 1.532165765762329,
-      "learning_rate": 8.493266331658291e-05,
-      "loss": 5.569,
-      "step": 15497
-    },
-    {
-      "epoch": 8.082398956975227,
-      "grad_norm": 1.5525789260864258,
-      "learning_rate": 8.493165829145729e-05,
-      "loss": 4.8551,
-      "step": 15498
-    },
-    {
-      "epoch": 8.082920469361147,
-      "grad_norm": 1.4461520910263062,
-      "learning_rate": 8.493065326633166e-05,
-      "loss": 5.5708,
-      "step": 15499
-    },
-    {
-      "epoch": 8.083441981747066,
-      "grad_norm": 1.6441563367843628,
-      "learning_rate": 8.492964824120603e-05,
-      "loss": 5.2709,
-      "step": 15500
-    },
-    {
-      "epoch": 8.083963494132986,
-      "grad_norm": 1.3830727338790894,
-      "learning_rate": 8.492864321608041e-05,
-      "loss": 5.6674,
-      "step": 15501
-    },
-    {
-      "epoch": 8.084485006518905,
-      "grad_norm": 1.392408847808838,
-      "learning_rate": 8.492763819095478e-05,
-      "loss": 5.8682,
-      "step": 15502
-    },
-    {
-      "epoch": 8.085006518904825,
-      "grad_norm": 1.5038927793502808,
-      "learning_rate": 8.492663316582915e-05,
-      "loss": 5.5057,
-      "step": 15503
-    },
-    {
-      "epoch": 8.085528031290742,
-      "grad_norm": 1.3829306364059448,
-      "learning_rate": 8.492562814070352e-05,
-      "loss": 5.7114,
-      "step": 15504
-    },
-    {
-      "epoch": 8.086049543676662,
-      "grad_norm": 1.4830255508422852,
-      "learning_rate": 8.49246231155779e-05,
-      "loss": 5.0291,
-      "step": 15505
-    },
-    {
-      "epoch": 8.086571056062581,
-      "grad_norm": 1.4965134859085083,
-      "learning_rate": 8.492361809045226e-05,
-      "loss": 5.616,
-      "step": 15506
-    },
-    {
-      "epoch": 8.0870925684485,
-      "grad_norm": 1.433705449104309,
-      "learning_rate": 8.492261306532664e-05,
-      "loss": 5.5876,
-      "step": 15507
-    },
-    {
-      "epoch": 8.08761408083442,
-      "grad_norm": 1.637774109840393,
-      "learning_rate": 8.4921608040201e-05,
-      "loss": 5.3905,
-      "step": 15508
-    },
-    {
-      "epoch": 8.08813559322034,
-      "grad_norm": 1.4559588432312012,
-      "learning_rate": 8.492060301507538e-05,
-      "loss": 5.2745,
-      "step": 15509
-    },
-    {
-      "epoch": 8.088657105606258,
-      "grad_norm": 1.4403352737426758,
-      "learning_rate": 8.491959798994974e-05,
-      "loss": 5.7179,
-      "step": 15510
-    },
-    {
-      "epoch": 8.089178617992177,
-      "grad_norm": 1.5928643941879272,
-      "learning_rate": 8.491859296482412e-05,
-      "loss": 4.9602,
-      "step": 15511
-    },
-    {
-      "epoch": 8.089700130378096,
-      "grad_norm": 1.4875457286834717,
-      "learning_rate": 8.49175879396985e-05,
-      "loss": 5.7121,
-      "step": 15512
-    },
-    {
-      "epoch": 8.090221642764016,
-      "grad_norm": 1.3799540996551514,
-      "learning_rate": 8.491658291457288e-05,
-      "loss": 5.7536,
-      "step": 15513
-    },
-    {
-      "epoch": 8.090743155149935,
-      "grad_norm": 1.3551236391067505,
-      "learning_rate": 8.491557788944724e-05,
-      "loss": 5.799,
-      "step": 15514
-    },
-    {
-      "epoch": 8.091264667535855,
-      "grad_norm": 1.4892292022705078,
-      "learning_rate": 8.491457286432162e-05,
-      "loss": 5.7994,
-      "step": 15515
-    },
-    {
-      "epoch": 8.091786179921773,
-      "grad_norm": 1.520025372505188,
-      "learning_rate": 8.491356783919598e-05,
-      "loss": 5.7286,
-      "step": 15516
-    },
-    {
-      "epoch": 8.092307692307692,
-      "grad_norm": 1.4459669589996338,
-      "learning_rate": 8.491256281407035e-05,
-      "loss": 5.3076,
-      "step": 15517
-    },
-    {
-      "epoch": 8.092829204693611,
-      "grad_norm": 1.4724669456481934,
-      "learning_rate": 8.491155778894473e-05,
-      "loss": 5.4054,
-      "step": 15518
-    },
-    {
-      "epoch": 8.093350717079531,
-      "grad_norm": 1.446352481842041,
-      "learning_rate": 8.491055276381909e-05,
-      "loss": 5.3088,
-      "step": 15519
-    },
-    {
-      "epoch": 8.09387222946545,
-      "grad_norm": 1.6142348051071167,
-      "learning_rate": 8.490954773869347e-05,
-      "loss": 5.4955,
-      "step": 15520
-    },
-    {
-      "epoch": 8.09439374185137,
-      "grad_norm": 1.6913491487503052,
-      "learning_rate": 8.490854271356785e-05,
-      "loss": 5.611,
-      "step": 15521
-    },
-    {
-      "epoch": 8.094915254237288,
-      "grad_norm": 1.3881362676620483,
-      "learning_rate": 8.490753768844222e-05,
-      "loss": 5.3703,
-      "step": 15522
-    },
-    {
-      "epoch": 8.095436766623207,
-      "grad_norm": 1.3482580184936523,
-      "learning_rate": 8.490653266331659e-05,
-      "loss": 5.8333,
-      "step": 15523
-    },
-    {
-      "epoch": 8.095958279009126,
-      "grad_norm": 1.392115592956543,
-      "learning_rate": 8.490552763819097e-05,
-      "loss": 5.0755,
-      "step": 15524
-    },
-    {
-      "epoch": 8.096479791395046,
-      "grad_norm": 1.3963944911956787,
-      "learning_rate": 8.490452261306533e-05,
-      "loss": 5.6712,
-      "step": 15525
-    },
-    {
-      "epoch": 8.097001303780965,
-      "grad_norm": 1.5378104448318481,
-      "learning_rate": 8.490351758793971e-05,
-      "loss": 5.0091,
-      "step": 15526
-    },
-    {
-      "epoch": 8.097522816166883,
-      "grad_norm": 1.3617117404937744,
-      "learning_rate": 8.490251256281407e-05,
-      "loss": 5.5242,
-      "step": 15527
-    },
-    {
-      "epoch": 8.098044328552803,
-      "grad_norm": 1.394676685333252,
-      "learning_rate": 8.490150753768845e-05,
-      "loss": 5.5016,
-      "step": 15528
-    },
-    {
-      "epoch": 8.098565840938722,
-      "grad_norm": 1.4392294883728027,
-      "learning_rate": 8.490050251256281e-05,
-      "loss": 5.6538,
-      "step": 15529
-    },
-    {
-      "epoch": 8.099087353324641,
-      "grad_norm": 1.353211522102356,
-      "learning_rate": 8.489949748743718e-05,
-      "loss": 5.6043,
-      "step": 15530
-    },
-    {
-      "epoch": 8.099608865710561,
-      "grad_norm": 1.3959040641784668,
-      "learning_rate": 8.489849246231156e-05,
-      "loss": 5.8259,
-      "step": 15531
-    },
-    {
-      "epoch": 8.10013037809648,
-      "grad_norm": 1.4671725034713745,
-      "learning_rate": 8.489748743718593e-05,
-      "loss": 4.787,
-      "step": 15532
-    },
-    {
-      "epoch": 8.100651890482398,
-      "grad_norm": 1.4205200672149658,
-      "learning_rate": 8.489648241206031e-05,
-      "loss": 5.4802,
-      "step": 15533
-    },
-    {
-      "epoch": 8.101173402868318,
-      "grad_norm": 1.4414615631103516,
-      "learning_rate": 8.489547738693468e-05,
-      "loss": 5.4476,
-      "step": 15534
-    },
-    {
-      "epoch": 8.101694915254237,
-      "grad_norm": 1.4431724548339844,
-      "learning_rate": 8.489447236180905e-05,
-      "loss": 5.6447,
-      "step": 15535
-    },
-    {
-      "epoch": 8.102216427640156,
-      "grad_norm": 1.4930686950683594,
-      "learning_rate": 8.489346733668342e-05,
-      "loss": 5.6196,
-      "step": 15536
-    },
-    {
-      "epoch": 8.102737940026076,
-      "grad_norm": 1.3843939304351807,
-      "learning_rate": 8.48924623115578e-05,
-      "loss": 5.3746,
-      "step": 15537
-    },
-    {
-      "epoch": 8.103259452411995,
-      "grad_norm": 1.3679341077804565,
-      "learning_rate": 8.489145728643216e-05,
-      "loss": 5.6248,
-      "step": 15538
-    },
-    {
-      "epoch": 8.103780964797913,
-      "grad_norm": 1.3605953454971313,
-      "learning_rate": 8.489045226130654e-05,
-      "loss": 5.6407,
-      "step": 15539
-    },
-    {
-      "epoch": 8.104302477183833,
-      "grad_norm": 1.3607312440872192,
-      "learning_rate": 8.48894472361809e-05,
-      "loss": 6.0562,
-      "step": 15540
-    },
-    {
-      "epoch": 8.104823989569752,
-      "grad_norm": 1.4508575201034546,
-      "learning_rate": 8.488844221105528e-05,
-      "loss": 5.4414,
-      "step": 15541
-    },
-    {
-      "epoch": 8.105345501955671,
-      "grad_norm": 1.4736027717590332,
-      "learning_rate": 8.488743718592966e-05,
-      "loss": 5.7324,
-      "step": 15542
-    },
-    {
-      "epoch": 8.105867014341591,
-      "grad_norm": 1.7025319337844849,
-      "learning_rate": 8.488643216080402e-05,
-      "loss": 5.1381,
-      "step": 15543
-    },
-    {
-      "epoch": 8.10638852672751,
-      "grad_norm": 1.571522831916809,
-      "learning_rate": 8.48854271356784e-05,
-      "loss": 4.866,
-      "step": 15544
-    },
-    {
-      "epoch": 8.106910039113428,
-      "grad_norm": 1.3378335237503052,
-      "learning_rate": 8.488442211055277e-05,
-      "loss": 5.4035,
-      "step": 15545
-    },
-    {
-      "epoch": 8.107431551499348,
-      "grad_norm": 1.3754727840423584,
-      "learning_rate": 8.488341708542714e-05,
-      "loss": 5.224,
-      "step": 15546
-    },
-    {
-      "epoch": 8.107953063885267,
-      "grad_norm": 1.5679142475128174,
-      "learning_rate": 8.488241206030151e-05,
-      "loss": 4.6494,
-      "step": 15547
-    },
-    {
-      "epoch": 8.108474576271187,
-      "grad_norm": 1.496989130973816,
-      "learning_rate": 8.488140703517589e-05,
-      "loss": 5.7979,
-      "step": 15548
-    },
-    {
-      "epoch": 8.108996088657106,
-      "grad_norm": 1.4647834300994873,
-      "learning_rate": 8.488040201005025e-05,
-      "loss": 5.2785,
-      "step": 15549
-    },
-    {
-      "epoch": 8.109517601043025,
-      "grad_norm": 1.3811612129211426,
-      "learning_rate": 8.487939698492463e-05,
-      "loss": 5.7119,
-      "step": 15550
-    },
-    {
-      "epoch": 8.110039113428943,
-      "grad_norm": 1.5355618000030518,
-      "learning_rate": 8.487839195979899e-05,
-      "loss": 5.2248,
-      "step": 15551
-    },
-    {
-      "epoch": 8.110560625814863,
-      "grad_norm": 1.4873042106628418,
-      "learning_rate": 8.487738693467337e-05,
-      "loss": 5.4367,
-      "step": 15552
-    },
-    {
-      "epoch": 8.111082138200782,
-      "grad_norm": 1.3422654867172241,
-      "learning_rate": 8.487638190954775e-05,
-      "loss": 5.9475,
-      "step": 15553
-    },
-    {
-      "epoch": 8.111603650586702,
-      "grad_norm": 1.499496340751648,
-      "learning_rate": 8.487537688442212e-05,
-      "loss": 5.5472,
-      "step": 15554
-    },
-    {
-      "epoch": 8.112125162972621,
-      "grad_norm": 1.4672340154647827,
-      "learning_rate": 8.487437185929649e-05,
-      "loss": 5.5653,
-      "step": 15555
-    },
-    {
-      "epoch": 8.11264667535854,
-      "grad_norm": 1.346651315689087,
-      "learning_rate": 8.487336683417085e-05,
-      "loss": 5.1771,
-      "step": 15556
-    },
-    {
-      "epoch": 8.113168187744458,
-      "grad_norm": 1.8967205286026,
-      "learning_rate": 8.487236180904523e-05,
-      "loss": 4.9222,
-      "step": 15557
-    },
-    {
-      "epoch": 8.113689700130378,
-      "grad_norm": 1.4375241994857788,
-      "learning_rate": 8.48713567839196e-05,
-      "loss": 5.379,
-      "step": 15558
-    },
-    {
-      "epoch": 8.114211212516297,
-      "grad_norm": 1.6120744943618774,
-      "learning_rate": 8.487035175879397e-05,
-      "loss": 5.1128,
-      "step": 15559
-    },
-    {
-      "epoch": 8.114732724902217,
-      "grad_norm": 1.300262451171875,
-      "learning_rate": 8.486934673366834e-05,
-      "loss": 6.1117,
-      "step": 15560
-    },
-    {
-      "epoch": 8.115254237288136,
-      "grad_norm": 1.4084118604660034,
-      "learning_rate": 8.486834170854272e-05,
-      "loss": 5.8308,
-      "step": 15561
-    },
-    {
-      "epoch": 8.115775749674055,
-      "grad_norm": 1.3975722789764404,
-      "learning_rate": 8.48673366834171e-05,
-      "loss": 5.889,
-      "step": 15562
-    },
-    {
-      "epoch": 8.116297262059973,
-      "grad_norm": 1.4709371328353882,
-      "learning_rate": 8.486633165829147e-05,
-      "loss": 5.5223,
-      "step": 15563
-    },
-    {
-      "epoch": 8.116818774445893,
-      "grad_norm": 1.381808876991272,
-      "learning_rate": 8.486532663316584e-05,
-      "loss": 5.9209,
-      "step": 15564
-    },
-    {
-      "epoch": 8.117340286831812,
-      "grad_norm": 1.5552624464035034,
-      "learning_rate": 8.486432160804021e-05,
-      "loss": 5.3573,
-      "step": 15565
-    },
-    {
-      "epoch": 8.117861799217732,
-      "grad_norm": 1.4626195430755615,
-      "learning_rate": 8.486331658291458e-05,
-      "loss": 4.7643,
-      "step": 15566
-    },
-    {
-      "epoch": 8.118383311603651,
-      "grad_norm": 1.4831053018569946,
-      "learning_rate": 8.486231155778896e-05,
-      "loss": 5.2142,
-      "step": 15567
-    },
-    {
-      "epoch": 8.11890482398957,
-      "grad_norm": 1.4958866834640503,
-      "learning_rate": 8.486130653266332e-05,
-      "loss": 5.1064,
-      "step": 15568
-    },
-    {
-      "epoch": 8.119426336375488,
-      "grad_norm": 1.3202338218688965,
-      "learning_rate": 8.486030150753768e-05,
-      "loss": 5.9428,
-      "step": 15569
-    },
-    {
-      "epoch": 8.119947848761408,
-      "grad_norm": 1.3805500268936157,
-      "learning_rate": 8.485929648241206e-05,
-      "loss": 5.6702,
-      "step": 15570
-    },
-    {
-      "epoch": 8.120469361147327,
-      "grad_norm": 1.4750313758850098,
-      "learning_rate": 8.485829145728643e-05,
-      "loss": 5.3515,
-      "step": 15571
-    },
-    {
-      "epoch": 8.120990873533247,
-      "grad_norm": 1.4101132154464722,
-      "learning_rate": 8.48572864321608e-05,
-      "loss": 5.0853,
-      "step": 15572
-    },
-    {
-      "epoch": 8.121512385919166,
-      "grad_norm": 1.5301276445388794,
-      "learning_rate": 8.485628140703518e-05,
-      "loss": 5.3527,
-      "step": 15573
-    },
-    {
-      "epoch": 8.122033898305085,
-      "grad_norm": 1.4019984006881714,
-      "learning_rate": 8.485527638190956e-05,
-      "loss": 5.2153,
-      "step": 15574
-    },
-    {
-      "epoch": 8.122555410691003,
-      "grad_norm": 1.3314241170883179,
-      "learning_rate": 8.485427135678392e-05,
-      "loss": 5.6257,
-      "step": 15575
-    },
-    {
-      "epoch": 8.123076923076923,
-      "grad_norm": 1.4106816053390503,
-      "learning_rate": 8.48532663316583e-05,
-      "loss": 5.5899,
-      "step": 15576
-    },
-    {
-      "epoch": 8.123598435462842,
-      "grad_norm": 1.7705309391021729,
-      "learning_rate": 8.485226130653267e-05,
-      "loss": 5.594,
-      "step": 15577
-    },
-    {
-      "epoch": 8.124119947848762,
-      "grad_norm": 1.337105631828308,
-      "learning_rate": 8.485125628140704e-05,
-      "loss": 5.5343,
-      "step": 15578
-    },
-    {
-      "epoch": 8.124641460234681,
-      "grad_norm": 1.3498470783233643,
-      "learning_rate": 8.485025125628141e-05,
-      "loss": 5.9471,
-      "step": 15579
-    },
-    {
-      "epoch": 8.1251629726206,
-      "grad_norm": 1.4374278783798218,
-      "learning_rate": 8.484924623115579e-05,
-      "loss": 5.5415,
-      "step": 15580
-    },
-    {
-      "epoch": 8.125684485006518,
-      "grad_norm": 1.3575382232666016,
-      "learning_rate": 8.484824120603015e-05,
-      "loss": 5.5804,
-      "step": 15581
-    },
-    {
-      "epoch": 8.126205997392438,
-      "grad_norm": 1.4507306814193726,
-      "learning_rate": 8.484723618090453e-05,
-      "loss": 5.4324,
-      "step": 15582
-    },
-    {
-      "epoch": 8.126727509778357,
-      "grad_norm": 1.3927377462387085,
-      "learning_rate": 8.48462311557789e-05,
-      "loss": 5.7295,
-      "step": 15583
-    },
-    {
-      "epoch": 8.127249022164277,
-      "grad_norm": 1.5501129627227783,
-      "learning_rate": 8.484522613065327e-05,
-      "loss": 5.1406,
-      "step": 15584
-    },
-    {
-      "epoch": 8.127770534550196,
-      "grad_norm": 1.673203468322754,
-      "learning_rate": 8.484422110552765e-05,
-      "loss": 5.1729,
-      "step": 15585
-    },
-    {
-      "epoch": 8.128292046936116,
-      "grad_norm": 1.396837592124939,
-      "learning_rate": 8.484321608040201e-05,
-      "loss": 5.6673,
-      "step": 15586
-    },
-    {
-      "epoch": 8.128813559322033,
-      "grad_norm": 1.453805685043335,
-      "learning_rate": 8.484221105527639e-05,
-      "loss": 5.8017,
-      "step": 15587
-    },
-    {
-      "epoch": 8.129335071707953,
-      "grad_norm": 1.4538160562515259,
-      "learning_rate": 8.484120603015075e-05,
-      "loss": 5.2467,
-      "step": 15588
-    },
-    {
-      "epoch": 8.129856584093872,
-      "grad_norm": 1.3509242534637451,
-      "learning_rate": 8.484020100502513e-05,
-      "loss": 5.9653,
-      "step": 15589
-    },
-    {
-      "epoch": 8.130378096479792,
-      "grad_norm": 1.3285996913909912,
-      "learning_rate": 8.48391959798995e-05,
-      "loss": 5.6723,
-      "step": 15590
-    },
-    {
-      "epoch": 8.130899608865711,
-      "grad_norm": 1.4589732885360718,
-      "learning_rate": 8.483819095477387e-05,
-      "loss": 5.2919,
-      "step": 15591
-    },
-    {
-      "epoch": 8.13142112125163,
-      "grad_norm": 1.4421581029891968,
-      "learning_rate": 8.483718592964824e-05,
-      "loss": 5.4921,
-      "step": 15592
-    },
-    {
-      "epoch": 8.131942633637548,
-      "grad_norm": 1.5017290115356445,
-      "learning_rate": 8.483618090452262e-05,
-      "loss": 5.576,
-      "step": 15593
-    },
-    {
-      "epoch": 8.132464146023468,
-      "grad_norm": 1.409816861152649,
-      "learning_rate": 8.4835175879397e-05,
-      "loss": 5.3614,
-      "step": 15594
-    },
-    {
-      "epoch": 8.132985658409387,
-      "grad_norm": 1.415540099143982,
-      "learning_rate": 8.483417085427136e-05,
-      "loss": 5.4161,
-      "step": 15595
-    },
-    {
-      "epoch": 8.133507170795307,
-      "grad_norm": 1.5131416320800781,
-      "learning_rate": 8.483316582914574e-05,
-      "loss": 5.5096,
-      "step": 15596
-    },
-    {
-      "epoch": 8.134028683181226,
-      "grad_norm": 1.2805544137954712,
-      "learning_rate": 8.48321608040201e-05,
-      "loss": 5.7007,
-      "step": 15597
-    },
-    {
-      "epoch": 8.134550195567146,
-      "grad_norm": 1.4210224151611328,
-      "learning_rate": 8.483115577889448e-05,
-      "loss": 5.5568,
-      "step": 15598
-    },
-    {
-      "epoch": 8.135071707953063,
-      "grad_norm": 1.4034459590911865,
-      "learning_rate": 8.483015075376884e-05,
-      "loss": 5.6493,
-      "step": 15599
-    },
-    {
-      "epoch": 8.135593220338983,
-      "grad_norm": 1.5517778396606445,
-      "learning_rate": 8.482914572864322e-05,
-      "loss": 5.6293,
-      "step": 15600
-    },
-    {
-      "epoch": 8.136114732724902,
-      "grad_norm": 1.581545352935791,
-      "learning_rate": 8.482814070351758e-05,
-      "loss": 5.3953,
-      "step": 15601
-    },
-    {
-      "epoch": 8.136636245110822,
-      "grad_norm": 1.5390088558197021,
-      "learning_rate": 8.482713567839196e-05,
-      "loss": 5.1723,
-      "step": 15602
-    },
-    {
-      "epoch": 8.137157757496741,
-      "grad_norm": 1.5111488103866577,
-      "learning_rate": 8.482613065326633e-05,
-      "loss": 5.2074,
-      "step": 15603
-    },
-    {
-      "epoch": 8.13767926988266,
-      "grad_norm": 1.5426474809646606,
-      "learning_rate": 8.48251256281407e-05,
-      "loss": 5.6368,
-      "step": 15604
-    },
-    {
-      "epoch": 8.138200782268578,
-      "grad_norm": 1.4915671348571777,
-      "learning_rate": 8.482412060301508e-05,
-      "loss": 5.311,
-      "step": 15605
-    },
-    {
-      "epoch": 8.138722294654498,
-      "grad_norm": 1.4542789459228516,
-      "learning_rate": 8.482311557788946e-05,
-      "loss": 5.599,
-      "step": 15606
-    },
-    {
-      "epoch": 8.139243807040417,
-      "grad_norm": 1.4096118211746216,
-      "learning_rate": 8.482211055276382e-05,
-      "loss": 5.9645,
-      "step": 15607
-    },
-    {
-      "epoch": 8.139765319426337,
-      "grad_norm": 1.5096935033798218,
-      "learning_rate": 8.48211055276382e-05,
-      "loss": 5.6701,
-      "step": 15608
-    },
-    {
-      "epoch": 8.140286831812256,
-      "grad_norm": 1.592988133430481,
-      "learning_rate": 8.482010050251257e-05,
-      "loss": 5.3796,
-      "step": 15609
-    },
-    {
-      "epoch": 8.140808344198176,
-      "grad_norm": 1.453254222869873,
-      "learning_rate": 8.481909547738693e-05,
-      "loss": 5.5625,
-      "step": 15610
-    },
-    {
-      "epoch": 8.141329856584093,
-      "grad_norm": 1.572135090827942,
-      "learning_rate": 8.481809045226131e-05,
-      "loss": 5.5304,
-      "step": 15611
-    },
-    {
-      "epoch": 8.141851368970013,
-      "grad_norm": 1.3350560665130615,
-      "learning_rate": 8.481708542713567e-05,
-      "loss": 5.5486,
-      "step": 15612
-    },
-    {
-      "epoch": 8.142372881355932,
-      "grad_norm": 1.4541770219802856,
-      "learning_rate": 8.481608040201005e-05,
-      "loss": 5.7234,
-      "step": 15613
-    },
-    {
-      "epoch": 8.142894393741852,
-      "grad_norm": 1.4264473915100098,
-      "learning_rate": 8.481507537688443e-05,
-      "loss": 5.6934,
-      "step": 15614
-    },
-    {
-      "epoch": 8.143415906127771,
-      "grad_norm": 1.3341819047927856,
-      "learning_rate": 8.48140703517588e-05,
-      "loss": 5.4834,
-      "step": 15615
-    },
-    {
-      "epoch": 8.14393741851369,
-      "grad_norm": 1.3868813514709473,
-      "learning_rate": 8.481306532663317e-05,
-      "loss": 5.2922,
-      "step": 15616
-    },
-    {
-      "epoch": 8.144458930899608,
-      "grad_norm": 1.4651952981948853,
-      "learning_rate": 8.481206030150755e-05,
-      "loss": 5.6416,
-      "step": 15617
-    },
-    {
-      "epoch": 8.144980443285528,
-      "grad_norm": 1.4945381879806519,
-      "learning_rate": 8.481105527638191e-05,
-      "loss": 5.453,
-      "step": 15618
-    },
-    {
-      "epoch": 8.145501955671447,
-      "grad_norm": 1.509943962097168,
-      "learning_rate": 8.481005025125629e-05,
-      "loss": 5.3416,
-      "step": 15619
-    },
-    {
-      "epoch": 8.146023468057367,
-      "grad_norm": 1.369494915008545,
-      "learning_rate": 8.480904522613066e-05,
-      "loss": 5.5599,
-      "step": 15620
-    },
-    {
-      "epoch": 8.146544980443286,
-      "grad_norm": 1.5115758180618286,
-      "learning_rate": 8.480804020100503e-05,
-      "loss": 5.541,
-      "step": 15621
-    },
-    {
-      "epoch": 8.147066492829204,
-      "grad_norm": 1.4213604927062988,
-      "learning_rate": 8.48070351758794e-05,
-      "loss": 5.4217,
-      "step": 15622
-    },
-    {
-      "epoch": 8.147588005215123,
-      "grad_norm": 1.5920116901397705,
-      "learning_rate": 8.480603015075376e-05,
-      "loss": 5.4849,
-      "step": 15623
-    },
-    {
-      "epoch": 8.148109517601043,
-      "grad_norm": 1.3764736652374268,
-      "learning_rate": 8.480502512562814e-05,
-      "loss": 5.7304,
-      "step": 15624
-    },
-    {
-      "epoch": 8.148631029986962,
-      "grad_norm": 1.476110816001892,
-      "learning_rate": 8.480402010050252e-05,
-      "loss": 5.7648,
-      "step": 15625
-    },
-    {
-      "epoch": 8.149152542372882,
-      "grad_norm": 1.3749607801437378,
-      "learning_rate": 8.48030150753769e-05,
-      "loss": 5.7916,
-      "step": 15626
-    },
-    {
-      "epoch": 8.149674054758801,
-      "grad_norm": 1.4094314575195312,
-      "learning_rate": 8.480201005025126e-05,
-      "loss": 5.5941,
-      "step": 15627
-    },
-    {
-      "epoch": 8.150195567144719,
-      "grad_norm": 1.385056495666504,
-      "learning_rate": 8.480100502512564e-05,
-      "loss": 5.2933,
-      "step": 15628
-    },
-    {
-      "epoch": 8.150717079530638,
-      "grad_norm": 1.6045433282852173,
-      "learning_rate": 8.48e-05,
-      "loss": 5.4459,
-      "step": 15629
-    },
-    {
-      "epoch": 8.151238591916558,
-      "grad_norm": 1.5553525686264038,
-      "learning_rate": 8.479899497487438e-05,
-      "loss": 5.4154,
-      "step": 15630
-    },
-    {
-      "epoch": 8.151760104302477,
-      "grad_norm": 1.4157049655914307,
-      "learning_rate": 8.479798994974874e-05,
-      "loss": 5.9612,
-      "step": 15631
-    },
-    {
-      "epoch": 8.152281616688397,
-      "grad_norm": 1.3862642049789429,
-      "learning_rate": 8.479698492462312e-05,
-      "loss": 5.811,
-      "step": 15632
-    },
-    {
-      "epoch": 8.152803129074316,
-      "grad_norm": 1.3505843877792358,
-      "learning_rate": 8.479597989949749e-05,
-      "loss": 6.0447,
-      "step": 15633
-    },
-    {
-      "epoch": 8.153324641460234,
-      "grad_norm": 1.4560343027114868,
-      "learning_rate": 8.479497487437186e-05,
-      "loss": 5.2017,
-      "step": 15634
-    },
-    {
-      "epoch": 8.153846153846153,
-      "grad_norm": 1.427320122718811,
-      "learning_rate": 8.479396984924624e-05,
-      "loss": 5.519,
-      "step": 15635
-    },
-    {
-      "epoch": 8.154367666232073,
-      "grad_norm": 1.3723108768463135,
-      "learning_rate": 8.47929648241206e-05,
-      "loss": 5.3713,
-      "step": 15636
-    },
-    {
-      "epoch": 8.154889178617992,
-      "grad_norm": 1.4132267236709595,
-      "learning_rate": 8.479195979899498e-05,
-      "loss": 5.1862,
-      "step": 15637
-    },
-    {
-      "epoch": 8.155410691003912,
-      "grad_norm": 1.5348517894744873,
-      "learning_rate": 8.479095477386935e-05,
-      "loss": 5.3815,
-      "step": 15638
-    },
-    {
-      "epoch": 8.155932203389831,
-      "grad_norm": 1.4944921731948853,
-      "learning_rate": 8.478994974874373e-05,
-      "loss": 5.2171,
-      "step": 15639
-    },
-    {
-      "epoch": 8.156453715775749,
-      "grad_norm": 1.5292121171951294,
-      "learning_rate": 8.478894472361809e-05,
-      "loss": 5.6398,
-      "step": 15640
-    },
-    {
-      "epoch": 8.156975228161668,
-      "grad_norm": 1.3823909759521484,
-      "learning_rate": 8.478793969849247e-05,
-      "loss": 5.3769,
-      "step": 15641
-    },
-    {
-      "epoch": 8.157496740547588,
-      "grad_norm": 1.4275274276733398,
-      "learning_rate": 8.478693467336683e-05,
-      "loss": 5.5798,
-      "step": 15642
-    },
-    {
-      "epoch": 8.158018252933507,
-      "grad_norm": 1.7292616367340088,
-      "learning_rate": 8.478592964824121e-05,
-      "loss": 5.3565,
-      "step": 15643
-    },
-    {
-      "epoch": 8.158539765319427,
-      "grad_norm": 1.463364601135254,
-      "learning_rate": 8.478492462311557e-05,
-      "loss": 5.7937,
-      "step": 15644
-    },
-    {
-      "epoch": 8.159061277705346,
-      "grad_norm": 1.5435712337493896,
-      "learning_rate": 8.478391959798995e-05,
-      "loss": 5.3912,
-      "step": 15645
-    },
-    {
-      "epoch": 8.159582790091264,
-      "grad_norm": 1.4186781644821167,
-      "learning_rate": 8.478291457286433e-05,
-      "loss": 5.1981,
-      "step": 15646
-    },
-    {
-      "epoch": 8.160104302477183,
-      "grad_norm": 1.4178798198699951,
-      "learning_rate": 8.478190954773871e-05,
-      "loss": 5.4085,
-      "step": 15647
-    },
-    {
-      "epoch": 8.160625814863103,
-      "grad_norm": 1.4930795431137085,
-      "learning_rate": 8.478090452261307e-05,
-      "loss": 5.1776,
-      "step": 15648
-    },
-    {
-      "epoch": 8.161147327249022,
-      "grad_norm": 1.518615484237671,
-      "learning_rate": 8.477989949748744e-05,
-      "loss": 5.5114,
-      "step": 15649
-    },
-    {
-      "epoch": 8.161668839634942,
-      "grad_norm": 1.4567126035690308,
-      "learning_rate": 8.477889447236181e-05,
-      "loss": 5.4336,
-      "step": 15650
-    },
-    {
-      "epoch": 8.162190352020861,
-      "grad_norm": 1.5706982612609863,
-      "learning_rate": 8.477788944723618e-05,
-      "loss": 5.475,
-      "step": 15651
-    },
-    {
-      "epoch": 8.162711864406779,
-      "grad_norm": 1.4382402896881104,
-      "learning_rate": 8.477688442211056e-05,
-      "loss": 5.3548,
-      "step": 15652
-    },
-    {
-      "epoch": 8.163233376792698,
-      "grad_norm": 1.4422733783721924,
-      "learning_rate": 8.477587939698492e-05,
-      "loss": 5.5328,
-      "step": 15653
-    },
-    {
-      "epoch": 8.163754889178618,
-      "grad_norm": 1.310920238494873,
-      "learning_rate": 8.47748743718593e-05,
-      "loss": 5.9082,
-      "step": 15654
-    },
-    {
-      "epoch": 8.164276401564537,
-      "grad_norm": 1.4188412427902222,
-      "learning_rate": 8.477386934673368e-05,
-      "loss": 5.6939,
-      "step": 15655
-    },
-    {
-      "epoch": 8.164797913950457,
-      "grad_norm": 1.429704189300537,
-      "learning_rate": 8.477286432160805e-05,
-      "loss": 5.8276,
-      "step": 15656
-    },
-    {
-      "epoch": 8.165319426336376,
-      "grad_norm": 1.486853837966919,
-      "learning_rate": 8.477185929648242e-05,
-      "loss": 5.1127,
-      "step": 15657
-    },
-    {
-      "epoch": 8.165840938722294,
-      "grad_norm": 1.6279070377349854,
-      "learning_rate": 8.47708542713568e-05,
-      "loss": 4.5894,
-      "step": 15658
-    },
-    {
-      "epoch": 8.166362451108213,
-      "grad_norm": 1.485946536064148,
-      "learning_rate": 8.476984924623116e-05,
-      "loss": 5.0548,
-      "step": 15659
-    },
-    {
-      "epoch": 8.166883963494133,
-      "grad_norm": 1.537684679031372,
-      "learning_rate": 8.476884422110554e-05,
-      "loss": 5.1216,
-      "step": 15660
-    },
-    {
-      "epoch": 8.167405475880052,
-      "grad_norm": 1.4714460372924805,
-      "learning_rate": 8.47678391959799e-05,
-      "loss": 5.6528,
-      "step": 15661
-    },
-    {
-      "epoch": 8.167926988265972,
-      "grad_norm": 1.4220597743988037,
-      "learning_rate": 8.476683417085427e-05,
-      "loss": 5.2816,
-      "step": 15662
-    },
-    {
-      "epoch": 8.168448500651891,
-      "grad_norm": 1.3877047300338745,
-      "learning_rate": 8.476582914572864e-05,
-      "loss": 5.6596,
-      "step": 15663
-    },
-    {
-      "epoch": 8.168970013037809,
-      "grad_norm": 1.4273946285247803,
-      "learning_rate": 8.476482412060301e-05,
-      "loss": 5.6472,
-      "step": 15664
-    },
-    {
-      "epoch": 8.169491525423728,
-      "grad_norm": 1.5067089796066284,
-      "learning_rate": 8.476381909547739e-05,
-      "loss": 5.1788,
-      "step": 15665
-    },
-    {
-      "epoch": 8.170013037809648,
-      "grad_norm": 1.725943684577942,
-      "learning_rate": 8.476281407035176e-05,
-      "loss": 4.6123,
-      "step": 15666
-    },
-    {
-      "epoch": 8.170534550195567,
-      "grad_norm": 1.4579700231552124,
-      "learning_rate": 8.476180904522614e-05,
-      "loss": 5.5621,
-      "step": 15667
-    },
-    {
-      "epoch": 8.171056062581487,
-      "grad_norm": 1.3820748329162598,
-      "learning_rate": 8.47608040201005e-05,
-      "loss": 5.663,
-      "step": 15668
-    },
-    {
-      "epoch": 8.171577574967406,
-      "grad_norm": 1.5525779724121094,
-      "learning_rate": 8.475979899497488e-05,
-      "loss": 5.1679,
-      "step": 15669
-    },
-    {
-      "epoch": 8.172099087353324,
-      "grad_norm": 1.4974778890609741,
-      "learning_rate": 8.475879396984925e-05,
-      "loss": 5.2157,
-      "step": 15670
-    },
-    {
-      "epoch": 8.172620599739243,
-      "grad_norm": 1.4455845355987549,
-      "learning_rate": 8.475778894472363e-05,
-      "loss": 5.5693,
-      "step": 15671
-    },
-    {
-      "epoch": 8.173142112125163,
-      "grad_norm": 1.3637913465499878,
-      "learning_rate": 8.475678391959799e-05,
-      "loss": 5.9652,
-      "step": 15672
-    },
-    {
-      "epoch": 8.173663624511082,
-      "grad_norm": 1.5420129299163818,
-      "learning_rate": 8.475577889447237e-05,
-      "loss": 5.4046,
-      "step": 15673
-    },
-    {
-      "epoch": 8.174185136897002,
-      "grad_norm": 1.6305127143859863,
-      "learning_rate": 8.475477386934673e-05,
-      "loss": 5.4738,
-      "step": 15674
-    },
-    {
-      "epoch": 8.174706649282921,
-      "grad_norm": 1.4297443628311157,
-      "learning_rate": 8.475376884422111e-05,
-      "loss": 4.7748,
-      "step": 15675
-    },
-    {
-      "epoch": 8.175228161668839,
-      "grad_norm": 1.500221610069275,
-      "learning_rate": 8.475276381909549e-05,
-      "loss": 4.8599,
-      "step": 15676
-    },
-    {
-      "epoch": 8.175749674054758,
-      "grad_norm": 1.3496217727661133,
-      "learning_rate": 8.475175879396985e-05,
-      "loss": 5.5598,
-      "step": 15677
-    },
-    {
-      "epoch": 8.176271186440678,
-      "grad_norm": 1.4575495719909668,
-      "learning_rate": 8.475075376884423e-05,
-      "loss": 5.3015,
-      "step": 15678
-    },
-    {
-      "epoch": 8.176792698826597,
-      "grad_norm": 1.43996262550354,
-      "learning_rate": 8.47497487437186e-05,
-      "loss": 5.7395,
-      "step": 15679
-    },
-    {
-      "epoch": 8.177314211212517,
-      "grad_norm": 1.385510802268982,
-      "learning_rate": 8.474874371859297e-05,
-      "loss": 5.6473,
-      "step": 15680
-    },
-    {
-      "epoch": 8.177835723598436,
-      "grad_norm": 1.2863478660583496,
-      "learning_rate": 8.474773869346734e-05,
-      "loss": 5.811,
-      "step": 15681
-    },
-    {
-      "epoch": 8.178357235984354,
-      "grad_norm": 1.4524357318878174,
-      "learning_rate": 8.474673366834171e-05,
-      "loss": 5.8097,
-      "step": 15682
-    },
-    {
-      "epoch": 8.178878748370273,
-      "grad_norm": 1.5861839056015015,
-      "learning_rate": 8.474572864321608e-05,
-      "loss": 5.3148,
-      "step": 15683
-    },
-    {
-      "epoch": 8.179400260756193,
-      "grad_norm": 1.3613831996917725,
-      "learning_rate": 8.474472361809046e-05,
-      "loss": 5.7718,
-      "step": 15684
-    },
-    {
-      "epoch": 8.179921773142112,
-      "grad_norm": 1.4768444299697876,
-      "learning_rate": 8.474371859296482e-05,
-      "loss": 5.4666,
-      "step": 15685
-    },
-    {
-      "epoch": 8.180443285528032,
-      "grad_norm": 1.5379939079284668,
-      "learning_rate": 8.47427135678392e-05,
-      "loss": 5.0985,
-      "step": 15686
-    },
-    {
-      "epoch": 8.180964797913951,
-      "grad_norm": 1.4930436611175537,
-      "learning_rate": 8.474170854271358e-05,
-      "loss": 5.9038,
-      "step": 15687
-    },
-    {
-      "epoch": 8.181486310299869,
-      "grad_norm": 1.481609582901001,
-      "learning_rate": 8.474070351758794e-05,
-      "loss": 5.2875,
-      "step": 15688
-    },
-    {
-      "epoch": 8.182007822685788,
-      "grad_norm": 1.5259296894073486,
-      "learning_rate": 8.473969849246232e-05,
-      "loss": 5.2092,
-      "step": 15689
-    },
-    {
-      "epoch": 8.182529335071708,
-      "grad_norm": 1.4880670309066772,
-      "learning_rate": 8.473869346733668e-05,
-      "loss": 5.6307,
-      "step": 15690
-    },
-    {
-      "epoch": 8.183050847457627,
-      "grad_norm": 1.515535593032837,
-      "learning_rate": 8.473768844221106e-05,
-      "loss": 4.9057,
-      "step": 15691
-    },
-    {
-      "epoch": 8.183572359843547,
-      "grad_norm": 1.462506890296936,
-      "learning_rate": 8.473668341708542e-05,
-      "loss": 5.9277,
-      "step": 15692
-    },
-    {
-      "epoch": 8.184093872229466,
-      "grad_norm": 1.5277743339538574,
-      "learning_rate": 8.47356783919598e-05,
-      "loss": 5.8953,
-      "step": 15693
-    },
-    {
-      "epoch": 8.184615384615384,
-      "grad_norm": 1.4780571460723877,
-      "learning_rate": 8.473467336683417e-05,
-      "loss": 5.6029,
-      "step": 15694
-    },
-    {
-      "epoch": 8.185136897001303,
-      "grad_norm": 1.3528010845184326,
-      "learning_rate": 8.473366834170854e-05,
-      "loss": 5.7132,
-      "step": 15695
-    },
-    {
-      "epoch": 8.185658409387223,
-      "grad_norm": 1.5806350708007812,
-      "learning_rate": 8.473266331658292e-05,
-      "loss": 5.4253,
-      "step": 15696
-    },
-    {
-      "epoch": 8.186179921773142,
-      "grad_norm": 1.4897887706756592,
-      "learning_rate": 8.47316582914573e-05,
-      "loss": 5.1961,
-      "step": 15697
-    },
-    {
-      "epoch": 8.186701434159062,
-      "grad_norm": 1.4663089513778687,
-      "learning_rate": 8.473065326633166e-05,
-      "loss": 5.5321,
-      "step": 15698
-    },
-    {
-      "epoch": 8.187222946544981,
-      "grad_norm": 1.5161285400390625,
-      "learning_rate": 8.472964824120604e-05,
-      "loss": 5.3861,
-      "step": 15699
-    },
-    {
-      "epoch": 8.187744458930899,
-      "grad_norm": 1.588722825050354,
-      "learning_rate": 8.472864321608041e-05,
-      "loss": 5.1899,
-      "step": 15700
-    },
-    {
-      "epoch": 8.188265971316818,
-      "grad_norm": 1.4283664226531982,
-      "learning_rate": 8.472763819095478e-05,
-      "loss": 5.7274,
-      "step": 15701
-    },
-    {
-      "epoch": 8.188787483702738,
-      "grad_norm": 1.3967450857162476,
-      "learning_rate": 8.472663316582915e-05,
-      "loss": 5.8239,
-      "step": 15702
-    },
-    {
-      "epoch": 8.189308996088657,
-      "grad_norm": 1.4912315607070923,
-      "learning_rate": 8.472562814070351e-05,
-      "loss": 5.3167,
-      "step": 15703
-    },
-    {
-      "epoch": 8.189830508474577,
-      "grad_norm": 1.4418059587478638,
-      "learning_rate": 8.472462311557789e-05,
-      "loss": 5.2679,
-      "step": 15704
-    },
-    {
-      "epoch": 8.190352020860496,
-      "grad_norm": 1.4699418544769287,
-      "learning_rate": 8.472361809045226e-05,
-      "loss": 5.6366,
-      "step": 15705
-    },
-    {
-      "epoch": 8.190873533246414,
-      "grad_norm": 1.4006768465042114,
-      "learning_rate": 8.472261306532663e-05,
-      "loss": 6.0613,
-      "step": 15706
-    },
-    {
-      "epoch": 8.191395045632333,
-      "grad_norm": 1.5386254787445068,
-      "learning_rate": 8.472160804020101e-05,
-      "loss": 5.5841,
-      "step": 15707
-    },
-    {
-      "epoch": 8.191916558018253,
-      "grad_norm": 1.5806063413619995,
-      "learning_rate": 8.472060301507539e-05,
-      "loss": 5.6865,
-      "step": 15708
-    },
-    {
-      "epoch": 8.192438070404172,
-      "grad_norm": 1.5957324504852295,
-      "learning_rate": 8.471959798994975e-05,
-      "loss": 5.4481,
-      "step": 15709
-    },
-    {
-      "epoch": 8.192959582790092,
-      "grad_norm": 1.435654878616333,
-      "learning_rate": 8.471859296482413e-05,
-      "loss": 5.8497,
-      "step": 15710
-    },
-    {
-      "epoch": 8.193481095176011,
-      "grad_norm": 1.4764268398284912,
-      "learning_rate": 8.47175879396985e-05,
-      "loss": 5.2377,
-      "step": 15711
-    },
-    {
-      "epoch": 8.194002607561929,
-      "grad_norm": 1.3605573177337646,
-      "learning_rate": 8.471658291457287e-05,
-      "loss": 5.8971,
-      "step": 15712
-    },
-    {
-      "epoch": 8.194524119947848,
-      "grad_norm": 1.3622684478759766,
-      "learning_rate": 8.471557788944724e-05,
-      "loss": 5.7264,
-      "step": 15713
-    },
-    {
-      "epoch": 8.195045632333768,
-      "grad_norm": 1.4610645771026611,
-      "learning_rate": 8.471457286432162e-05,
-      "loss": 5.8218,
-      "step": 15714
-    },
-    {
-      "epoch": 8.195567144719687,
-      "grad_norm": 1.4439469575881958,
-      "learning_rate": 8.471356783919598e-05,
-      "loss": 5.4333,
-      "step": 15715
-    },
-    {
-      "epoch": 8.196088657105607,
-      "grad_norm": 1.4013832807540894,
-      "learning_rate": 8.471256281407036e-05,
-      "loss": 5.3492,
-      "step": 15716
-    },
-    {
-      "epoch": 8.196610169491525,
-      "grad_norm": 1.4151861667633057,
-      "learning_rate": 8.471155778894474e-05,
-      "loss": 5.3737,
-      "step": 15717
-    },
-    {
-      "epoch": 8.197131681877444,
-      "grad_norm": 1.4271551370620728,
-      "learning_rate": 8.47105527638191e-05,
-      "loss": 5.9077,
-      "step": 15718
-    },
-    {
-      "epoch": 8.197653194263363,
-      "grad_norm": 1.4510663747787476,
-      "learning_rate": 8.470954773869348e-05,
-      "loss": 5.6372,
-      "step": 15719
-    },
-    {
-      "epoch": 8.198174706649283,
-      "grad_norm": 1.3848557472229004,
-      "learning_rate": 8.470854271356784e-05,
-      "loss": 5.5285,
-      "step": 15720
-    },
-    {
-      "epoch": 8.198696219035202,
-      "grad_norm": 1.4996144771575928,
-      "learning_rate": 8.470753768844222e-05,
-      "loss": 5.1585,
-      "step": 15721
-    },
-    {
-      "epoch": 8.199217731421122,
-      "grad_norm": 1.5980603694915771,
-      "learning_rate": 8.470653266331658e-05,
-      "loss": 5.3544,
-      "step": 15722
-    },
-    {
-      "epoch": 8.19973924380704,
-      "grad_norm": 1.5464061498641968,
-      "learning_rate": 8.470552763819096e-05,
-      "loss": 5.0331,
-      "step": 15723
-    },
-    {
-      "epoch": 8.200260756192959,
-      "grad_norm": 1.4435473680496216,
-      "learning_rate": 8.470452261306533e-05,
-      "loss": 5.0285,
-      "step": 15724
-    },
-    {
-      "epoch": 8.200782268578878,
-      "grad_norm": 1.419960856437683,
-      "learning_rate": 8.47035175879397e-05,
-      "loss": 5.7502,
-      "step": 15725
-    },
-    {
-      "epoch": 8.201303780964798,
-      "grad_norm": 1.5281155109405518,
-      "learning_rate": 8.470251256281407e-05,
-      "loss": 5.628,
-      "step": 15726
-    },
-    {
-      "epoch": 8.201825293350717,
-      "grad_norm": 1.7027063369750977,
-      "learning_rate": 8.470150753768845e-05,
-      "loss": 5.2764,
-      "step": 15727
-    },
-    {
-      "epoch": 8.202346805736637,
-      "grad_norm": 1.434435248374939,
-      "learning_rate": 8.470050251256282e-05,
-      "loss": 5.0294,
-      "step": 15728
-    },
-    {
-      "epoch": 8.202868318122555,
-      "grad_norm": 1.4521324634552002,
-      "learning_rate": 8.469949748743719e-05,
-      "loss": 5.4083,
-      "step": 15729
-    },
-    {
-      "epoch": 8.203389830508474,
-      "grad_norm": 1.4004123210906982,
-      "learning_rate": 8.469849246231157e-05,
-      "loss": 5.6841,
-      "step": 15730
-    },
-    {
-      "epoch": 8.203911342894393,
-      "grad_norm": 1.5111312866210938,
-      "learning_rate": 8.469748743718593e-05,
-      "loss": 5.6646,
-      "step": 15731
-    },
-    {
-      "epoch": 8.204432855280313,
-      "grad_norm": 1.3249565362930298,
-      "learning_rate": 8.469648241206031e-05,
-      "loss": 5.8177,
-      "step": 15732
-    },
-    {
-      "epoch": 8.204954367666232,
-      "grad_norm": 1.3625563383102417,
-      "learning_rate": 8.469547738693467e-05,
-      "loss": 5.6257,
-      "step": 15733
-    },
-    {
-      "epoch": 8.205475880052152,
-      "grad_norm": 1.4290738105773926,
-      "learning_rate": 8.469447236180905e-05,
-      "loss": 5.7234,
-      "step": 15734
-    },
-    {
-      "epoch": 8.20599739243807,
-      "grad_norm": 1.4249438047409058,
-      "learning_rate": 8.469346733668341e-05,
-      "loss": 5.6813,
-      "step": 15735
-    },
-    {
-      "epoch": 8.206518904823989,
-      "grad_norm": 1.4201111793518066,
-      "learning_rate": 8.469246231155779e-05,
-      "loss": 5.4823,
-      "step": 15736
-    },
-    {
-      "epoch": 8.207040417209909,
-      "grad_norm": 1.4363564252853394,
-      "learning_rate": 8.469145728643217e-05,
-      "loss": 5.5656,
-      "step": 15737
-    },
-    {
-      "epoch": 8.207561929595828,
-      "grad_norm": 1.4932661056518555,
-      "learning_rate": 8.469045226130655e-05,
-      "loss": 5.5949,
-      "step": 15738
-    },
-    {
-      "epoch": 8.208083441981747,
-      "grad_norm": 1.4237985610961914,
-      "learning_rate": 8.468944723618091e-05,
-      "loss": 5.6583,
-      "step": 15739
-    },
-    {
-      "epoch": 8.208604954367667,
-      "grad_norm": 1.387248158454895,
-      "learning_rate": 8.468844221105529e-05,
-      "loss": 5.6805,
-      "step": 15740
-    },
-    {
-      "epoch": 8.209126466753585,
-      "grad_norm": 1.3265591859817505,
-      "learning_rate": 8.468743718592965e-05,
-      "loss": 5.662,
-      "step": 15741
-    },
-    {
-      "epoch": 8.209647979139504,
-      "grad_norm": 1.3854267597198486,
-      "learning_rate": 8.468643216080402e-05,
-      "loss": 5.8296,
-      "step": 15742
-    },
-    {
-      "epoch": 8.210169491525424,
-      "grad_norm": 1.3817492723464966,
-      "learning_rate": 8.46854271356784e-05,
-      "loss": 5.4144,
-      "step": 15743
-    },
-    {
-      "epoch": 8.210691003911343,
-      "grad_norm": 1.4484914541244507,
-      "learning_rate": 8.468442211055276e-05,
-      "loss": 5.2767,
-      "step": 15744
-    },
-    {
-      "epoch": 8.211212516297262,
-      "grad_norm": 1.366478443145752,
-      "learning_rate": 8.468341708542714e-05,
-      "loss": 5.7542,
-      "step": 15745
-    },
-    {
-      "epoch": 8.211734028683182,
-      "grad_norm": 1.3785945177078247,
-      "learning_rate": 8.46824120603015e-05,
-      "loss": 5.7536,
-      "step": 15746
-    },
-    {
-      "epoch": 8.2122555410691,
-      "grad_norm": 1.382473349571228,
-      "learning_rate": 8.468140703517588e-05,
-      "loss": 5.6491,
-      "step": 15747
-    },
-    {
-      "epoch": 8.212777053455019,
-      "grad_norm": 1.3798565864562988,
-      "learning_rate": 8.468040201005026e-05,
-      "loss": 5.6605,
-      "step": 15748
-    },
-    {
-      "epoch": 8.213298565840939,
-      "grad_norm": 1.4109631776809692,
-      "learning_rate": 8.467939698492464e-05,
-      "loss": 5.428,
-      "step": 15749
-    },
-    {
-      "epoch": 8.213820078226858,
-      "grad_norm": 1.3077722787857056,
-      "learning_rate": 8.4678391959799e-05,
-      "loss": 5.6788,
-      "step": 15750
-    },
-    {
-      "epoch": 8.214341590612777,
-      "grad_norm": 1.3707021474838257,
-      "learning_rate": 8.467738693467338e-05,
-      "loss": 5.7515,
-      "step": 15751
-    },
-    {
-      "epoch": 8.214863102998697,
-      "grad_norm": 1.4192339181900024,
-      "learning_rate": 8.467638190954774e-05,
-      "loss": 5.3697,
-      "step": 15752
-    },
-    {
-      "epoch": 8.215384615384615,
-      "grad_norm": 1.508329153060913,
-      "learning_rate": 8.467537688442212e-05,
-      "loss": 5.3475,
-      "step": 15753
-    },
-    {
-      "epoch": 8.215906127770534,
-      "grad_norm": 1.535009503364563,
-      "learning_rate": 8.467437185929648e-05,
-      "loss": 5.8883,
-      "step": 15754
-    },
-    {
-      "epoch": 8.216427640156454,
-      "grad_norm": 1.476557731628418,
-      "learning_rate": 8.467336683417085e-05,
-      "loss": 5.2315,
-      "step": 15755
-    },
-    {
-      "epoch": 8.216949152542373,
-      "grad_norm": 1.41514253616333,
-      "learning_rate": 8.467236180904523e-05,
-      "loss": 5.3246,
-      "step": 15756
-    },
-    {
-      "epoch": 8.217470664928292,
-      "grad_norm": 1.5276771783828735,
-      "learning_rate": 8.46713567839196e-05,
-      "loss": 5.1752,
-      "step": 15757
-    },
-    {
-      "epoch": 8.217992177314212,
-      "grad_norm": 1.4018306732177734,
-      "learning_rate": 8.467035175879398e-05,
-      "loss": 5.8253,
-      "step": 15758
-    },
-    {
-      "epoch": 8.21851368970013,
-      "grad_norm": 1.5412282943725586,
-      "learning_rate": 8.466934673366835e-05,
-      "loss": 5.1982,
-      "step": 15759
-    },
-    {
-      "epoch": 8.219035202086049,
-      "grad_norm": 1.49917471408844,
-      "learning_rate": 8.466834170854272e-05,
-      "loss": 5.4657,
-      "step": 15760
-    },
-    {
-      "epoch": 8.219556714471969,
-      "grad_norm": 1.4708592891693115,
-      "learning_rate": 8.466733668341709e-05,
-      "loss": 4.9168,
-      "step": 15761
-    },
-    {
-      "epoch": 8.220078226857888,
-      "grad_norm": 1.4547802209854126,
-      "learning_rate": 8.466633165829147e-05,
-      "loss": 5.5119,
-      "step": 15762
-    },
-    {
-      "epoch": 8.220599739243807,
-      "grad_norm": 1.4105051755905151,
-      "learning_rate": 8.466532663316583e-05,
-      "loss": 5.4833,
-      "step": 15763
-    },
-    {
-      "epoch": 8.221121251629727,
-      "grad_norm": 1.4455944299697876,
-      "learning_rate": 8.466432160804021e-05,
-      "loss": 5.6255,
-      "step": 15764
-    },
-    {
-      "epoch": 8.221642764015645,
-      "grad_norm": 1.3468468189239502,
-      "learning_rate": 8.466331658291457e-05,
-      "loss": 5.506,
-      "step": 15765
-    },
-    {
-      "epoch": 8.222164276401564,
-      "grad_norm": 1.447231650352478,
-      "learning_rate": 8.466231155778895e-05,
-      "loss": 5.7529,
-      "step": 15766
-    },
-    {
-      "epoch": 8.222685788787484,
-      "grad_norm": 1.423258662223816,
-      "learning_rate": 8.466130653266331e-05,
-      "loss": 5.5575,
-      "step": 15767
-    },
-    {
-      "epoch": 8.223207301173403,
-      "grad_norm": 1.4167636632919312,
-      "learning_rate": 8.466030150753769e-05,
-      "loss": 5.7239,
-      "step": 15768
-    },
-    {
-      "epoch": 8.223728813559323,
-      "grad_norm": 1.4478312730789185,
-      "learning_rate": 8.465929648241207e-05,
-      "loss": 5.295,
-      "step": 15769
-    },
-    {
-      "epoch": 8.224250325945242,
-      "grad_norm": 1.256091833114624,
-      "learning_rate": 8.465829145728643e-05,
-      "loss": 5.8064,
-      "step": 15770
-    },
-    {
-      "epoch": 8.22477183833116,
-      "grad_norm": 1.4325037002563477,
-      "learning_rate": 8.465728643216081e-05,
-      "loss": 5.8573,
-      "step": 15771
-    },
-    {
-      "epoch": 8.22529335071708,
-      "grad_norm": 1.4392352104187012,
-      "learning_rate": 8.465628140703518e-05,
-      "loss": 5.7699,
-      "step": 15772
-    },
-    {
-      "epoch": 8.225814863102999,
-      "grad_norm": 1.2974876165390015,
-      "learning_rate": 8.465527638190955e-05,
-      "loss": 5.901,
-      "step": 15773
-    },
-    {
-      "epoch": 8.226336375488918,
-      "grad_norm": 1.5390645265579224,
-      "learning_rate": 8.465427135678392e-05,
-      "loss": 5.0797,
-      "step": 15774
-    },
-    {
-      "epoch": 8.226857887874838,
-      "grad_norm": 1.4045251607894897,
-      "learning_rate": 8.46532663316583e-05,
-      "loss": 5.3441,
-      "step": 15775
-    },
-    {
-      "epoch": 8.227379400260757,
-      "grad_norm": 1.418887734413147,
-      "learning_rate": 8.465226130653266e-05,
-      "loss": 5.4019,
-      "step": 15776
-    },
-    {
-      "epoch": 8.227900912646675,
-      "grad_norm": 1.4092835187911987,
-      "learning_rate": 8.465125628140704e-05,
-      "loss": 5.5853,
-      "step": 15777
-    },
-    {
-      "epoch": 8.228422425032594,
-      "grad_norm": 1.4801493883132935,
-      "learning_rate": 8.46502512562814e-05,
-      "loss": 5.7136,
-      "step": 15778
-    },
-    {
-      "epoch": 8.228943937418514,
-      "grad_norm": 1.5004867315292358,
-      "learning_rate": 8.464924623115578e-05,
-      "loss": 5.0606,
-      "step": 15779
-    },
-    {
-      "epoch": 8.229465449804433,
-      "grad_norm": 1.4344185590744019,
-      "learning_rate": 8.464824120603016e-05,
-      "loss": 5.6834,
-      "step": 15780
-    },
-    {
-      "epoch": 8.229986962190353,
-      "grad_norm": 1.3830920457839966,
-      "learning_rate": 8.464723618090452e-05,
-      "loss": 5.6339,
-      "step": 15781
-    },
-    {
-      "epoch": 8.230508474576272,
-      "grad_norm": 1.3700313568115234,
-      "learning_rate": 8.46462311557789e-05,
-      "loss": 5.7486,
-      "step": 15782
-    },
-    {
-      "epoch": 8.23102998696219,
-      "grad_norm": 1.3572949171066284,
-      "learning_rate": 8.464522613065327e-05,
-      "loss": 5.2489,
-      "step": 15783
-    },
-    {
-      "epoch": 8.23155149934811,
-      "grad_norm": 1.505470633506775,
-      "learning_rate": 8.464422110552764e-05,
-      "loss": 5.3648,
-      "step": 15784
-    },
-    {
-      "epoch": 8.232073011734029,
-      "grad_norm": 1.5689599514007568,
-      "learning_rate": 8.464321608040201e-05,
-      "loss": 5.3144,
-      "step": 15785
-    },
-    {
-      "epoch": 8.232594524119948,
-      "grad_norm": 1.3889964818954468,
-      "learning_rate": 8.464221105527639e-05,
-      "loss": 5.7106,
-      "step": 15786
-    },
-    {
-      "epoch": 8.233116036505868,
-      "grad_norm": 1.4004110097885132,
-      "learning_rate": 8.464120603015075e-05,
-      "loss": 5.8092,
-      "step": 15787
-    },
-    {
-      "epoch": 8.233637548891787,
-      "grad_norm": 1.3814398050308228,
-      "learning_rate": 8.464020100502513e-05,
-      "loss": 5.2431,
-      "step": 15788
-    },
-    {
-      "epoch": 8.234159061277705,
-      "grad_norm": 1.454531192779541,
-      "learning_rate": 8.46391959798995e-05,
-      "loss": 5.4928,
-      "step": 15789
-    },
-    {
-      "epoch": 8.234680573663624,
-      "grad_norm": 1.3786059617996216,
-      "learning_rate": 8.463819095477388e-05,
-      "loss": 5.8994,
-      "step": 15790
-    },
-    {
-      "epoch": 8.235202086049544,
-      "grad_norm": 1.338403344154358,
-      "learning_rate": 8.463718592964825e-05,
-      "loss": 5.917,
-      "step": 15791
-    },
-    {
-      "epoch": 8.235723598435463,
-      "grad_norm": 1.7408145666122437,
-      "learning_rate": 8.463618090452262e-05,
-      "loss": 5.3605,
-      "step": 15792
-    },
-    {
-      "epoch": 8.236245110821383,
-      "grad_norm": 1.424072027206421,
-      "learning_rate": 8.463517587939699e-05,
-      "loss": 5.9482,
-      "step": 15793
-    },
-    {
-      "epoch": 8.236766623207302,
-      "grad_norm": 1.5715824365615845,
-      "learning_rate": 8.463417085427137e-05,
-      "loss": 4.7835,
-      "step": 15794
-    },
-    {
-      "epoch": 8.23728813559322,
-      "grad_norm": 1.445021152496338,
-      "learning_rate": 8.463316582914573e-05,
-      "loss": 5.8397,
-      "step": 15795
-    },
-    {
-      "epoch": 8.23780964797914,
-      "grad_norm": 1.4627408981323242,
-      "learning_rate": 8.46321608040201e-05,
-      "loss": 5.5774,
-      "step": 15796
-    },
-    {
-      "epoch": 8.238331160365059,
-      "grad_norm": 1.5527275800704956,
-      "learning_rate": 8.463115577889447e-05,
-      "loss": 5.2931,
-      "step": 15797
-    },
-    {
-      "epoch": 8.238852672750978,
-      "grad_norm": 1.5406156778335571,
-      "learning_rate": 8.463015075376884e-05,
-      "loss": 5.6067,
-      "step": 15798
-    },
-    {
-      "epoch": 8.239374185136898,
-      "grad_norm": 1.417326807975769,
-      "learning_rate": 8.462914572864322e-05,
-      "loss": 5.9641,
-      "step": 15799
-    },
-    {
-      "epoch": 8.239895697522817,
-      "grad_norm": 1.5441179275512695,
-      "learning_rate": 8.46281407035176e-05,
-      "loss": 5.4097,
-      "step": 15800
-    },
-    {
-      "epoch": 8.240417209908735,
-      "grad_norm": 1.6008589267730713,
-      "learning_rate": 8.462713567839197e-05,
-      "loss": 5.6699,
-      "step": 15801
-    },
-    {
-      "epoch": 8.240938722294654,
-      "grad_norm": 1.3133982419967651,
-      "learning_rate": 8.462613065326634e-05,
-      "loss": 5.8193,
-      "step": 15802
-    },
-    {
-      "epoch": 8.241460234680574,
-      "grad_norm": 1.5236480236053467,
-      "learning_rate": 8.462512562814071e-05,
-      "loss": 5.548,
-      "step": 15803
-    },
-    {
-      "epoch": 8.241981747066493,
-      "grad_norm": 1.5558503866195679,
-      "learning_rate": 8.462412060301508e-05,
-      "loss": 5.6978,
-      "step": 15804
-    },
-    {
-      "epoch": 8.242503259452413,
-      "grad_norm": 1.5416392087936401,
-      "learning_rate": 8.462311557788946e-05,
-      "loss": 5.7067,
-      "step": 15805
-    },
-    {
-      "epoch": 8.243024771838332,
-      "grad_norm": 1.562567114830017,
-      "learning_rate": 8.462211055276382e-05,
-      "loss": 5.1771,
-      "step": 15806
-    },
-    {
-      "epoch": 8.24354628422425,
-      "grad_norm": 1.5003142356872559,
-      "learning_rate": 8.46211055276382e-05,
-      "loss": 5.2657,
-      "step": 15807
-    },
-    {
-      "epoch": 8.24406779661017,
-      "grad_norm": 1.6471900939941406,
-      "learning_rate": 8.462010050251256e-05,
-      "loss": 5.5614,
-      "step": 15808
-    },
-    {
-      "epoch": 8.244589308996089,
-      "grad_norm": 1.534129023551941,
-      "learning_rate": 8.461909547738694e-05,
-      "loss": 5.5332,
-      "step": 15809
-    },
-    {
-      "epoch": 8.245110821382008,
-      "grad_norm": 1.5146350860595703,
-      "learning_rate": 8.461809045226132e-05,
-      "loss": 5.2022,
-      "step": 15810
-    },
-    {
-      "epoch": 8.245632333767928,
-      "grad_norm": 1.4150269031524658,
-      "learning_rate": 8.461708542713568e-05,
-      "loss": 5.7878,
-      "step": 15811
-    },
-    {
-      "epoch": 8.246153846153845,
-      "grad_norm": 1.5084728002548218,
-      "learning_rate": 8.461608040201006e-05,
-      "loss": 5.2717,
-      "step": 15812
-    },
-    {
-      "epoch": 8.246675358539765,
-      "grad_norm": 1.3794447183609009,
-      "learning_rate": 8.461507537688442e-05,
-      "loss": 5.801,
-      "step": 15813
-    },
-    {
-      "epoch": 8.247196870925684,
-      "grad_norm": 1.4749020338058472,
-      "learning_rate": 8.46140703517588e-05,
-      "loss": 5.8158,
-      "step": 15814
-    },
-    {
-      "epoch": 8.247718383311604,
-      "grad_norm": 1.3189536333084106,
-      "learning_rate": 8.461306532663317e-05,
-      "loss": 5.7886,
-      "step": 15815
-    },
-    {
-      "epoch": 8.248239895697523,
-      "grad_norm": 1.4136351346969604,
-      "learning_rate": 8.461206030150754e-05,
-      "loss": 5.5895,
-      "step": 15816
-    },
-    {
-      "epoch": 8.248761408083443,
-      "grad_norm": 1.570310354232788,
-      "learning_rate": 8.461105527638191e-05,
-      "loss": 5.8626,
-      "step": 15817
-    },
-    {
-      "epoch": 8.24928292046936,
-      "grad_norm": 1.4577895402908325,
-      "learning_rate": 8.461005025125629e-05,
-      "loss": 5.5529,
-      "step": 15818
-    },
-    {
-      "epoch": 8.24980443285528,
-      "grad_norm": 1.8627870082855225,
-      "learning_rate": 8.460904522613065e-05,
-      "loss": 4.4952,
-      "step": 15819
-    },
-    {
-      "epoch": 8.2503259452412,
-      "grad_norm": 1.4534896612167358,
-      "learning_rate": 8.460804020100503e-05,
-      "loss": 5.6476,
-      "step": 15820
-    },
-    {
-      "epoch": 8.250847457627119,
-      "grad_norm": 1.475164532661438,
-      "learning_rate": 8.46070351758794e-05,
-      "loss": 5.5647,
-      "step": 15821
-    },
-    {
-      "epoch": 8.251368970013038,
-      "grad_norm": 1.484103798866272,
-      "learning_rate": 8.460603015075377e-05,
-      "loss": 5.5263,
-      "step": 15822
-    },
-    {
-      "epoch": 8.251890482398958,
-      "grad_norm": 1.5220357179641724,
-      "learning_rate": 8.460502512562815e-05,
-      "loss": 5.4031,
-      "step": 15823
-    },
-    {
-      "epoch": 8.252411994784875,
-      "grad_norm": 1.455743432044983,
-      "learning_rate": 8.460402010050251e-05,
-      "loss": 5.2914,
-      "step": 15824
-    },
-    {
-      "epoch": 8.252933507170795,
-      "grad_norm": 1.4494571685791016,
-      "learning_rate": 8.460301507537689e-05,
-      "loss": 5.6511,
-      "step": 15825
-    },
-    {
-      "epoch": 8.253455019556714,
-      "grad_norm": 1.6204783916473389,
-      "learning_rate": 8.460201005025125e-05,
-      "loss": 5.8298,
-      "step": 15826
-    },
-    {
-      "epoch": 8.253976531942634,
-      "grad_norm": 1.4394683837890625,
-      "learning_rate": 8.460100502512563e-05,
-      "loss": 5.2912,
-      "step": 15827
-    },
-    {
-      "epoch": 8.254498044328553,
-      "grad_norm": 1.5136851072311401,
-      "learning_rate": 8.46e-05,
-      "loss": 5.2703,
-      "step": 15828
-    },
-    {
-      "epoch": 8.255019556714473,
-      "grad_norm": 1.4477945566177368,
-      "learning_rate": 8.459899497487437e-05,
-      "loss": 5.4448,
-      "step": 15829
-    },
-    {
-      "epoch": 8.25554106910039,
-      "grad_norm": 1.4141771793365479,
-      "learning_rate": 8.459798994974875e-05,
-      "loss": 5.7948,
-      "step": 15830
-    },
-    {
-      "epoch": 8.25606258148631,
-      "grad_norm": 1.4768807888031006,
-      "learning_rate": 8.459698492462313e-05,
-      "loss": 5.8481,
-      "step": 15831
-    },
-    {
-      "epoch": 8.25658409387223,
-      "grad_norm": 1.4375606775283813,
-      "learning_rate": 8.45959798994975e-05,
-      "loss": 6.0027,
-      "step": 15832
-    },
-    {
-      "epoch": 8.257105606258149,
-      "grad_norm": 1.406207799911499,
-      "learning_rate": 8.459497487437187e-05,
-      "loss": 5.3202,
-      "step": 15833
-    },
-    {
-      "epoch": 8.257627118644068,
-      "grad_norm": 1.482801914215088,
-      "learning_rate": 8.459396984924624e-05,
-      "loss": 5.4423,
-      "step": 15834
-    },
-    {
-      "epoch": 8.258148631029988,
-      "grad_norm": 1.5668821334838867,
-      "learning_rate": 8.45929648241206e-05,
-      "loss": 5.4518,
-      "step": 15835
-    },
-    {
-      "epoch": 8.258670143415905,
-      "grad_norm": 1.4035617113113403,
-      "learning_rate": 8.459195979899498e-05,
-      "loss": 5.6356,
-      "step": 15836
-    },
-    {
-      "epoch": 8.259191655801825,
-      "grad_norm": 1.5411652326583862,
-      "learning_rate": 8.459095477386934e-05,
-      "loss": 5.2687,
-      "step": 15837
-    },
-    {
-      "epoch": 8.259713168187744,
-      "grad_norm": 1.4445432424545288,
-      "learning_rate": 8.458994974874372e-05,
-      "loss": 5.6304,
-      "step": 15838
-    },
-    {
-      "epoch": 8.260234680573664,
-      "grad_norm": 1.5861488580703735,
-      "learning_rate": 8.458894472361808e-05,
-      "loss": 5.1478,
-      "step": 15839
-    },
-    {
-      "epoch": 8.260756192959583,
-      "grad_norm": 1.5702685117721558,
-      "learning_rate": 8.458793969849246e-05,
-      "loss": 5.166,
-      "step": 15840
-    },
-    {
-      "epoch": 8.261277705345503,
-      "grad_norm": 1.558767318725586,
-      "learning_rate": 8.458693467336684e-05,
-      "loss": 5.2816,
-      "step": 15841
-    },
-    {
-      "epoch": 8.26179921773142,
-      "grad_norm": 1.56748366355896,
-      "learning_rate": 8.458592964824122e-05,
-      "loss": 5.3942,
-      "step": 15842
-    },
-    {
-      "epoch": 8.26232073011734,
-      "grad_norm": 1.5700925588607788,
-      "learning_rate": 8.458492462311558e-05,
-      "loss": 4.9963,
-      "step": 15843
-    },
-    {
-      "epoch": 8.26284224250326,
-      "grad_norm": 1.4957177639007568,
-      "learning_rate": 8.458391959798996e-05,
-      "loss": 5.4783,
-      "step": 15844
-    },
-    {
-      "epoch": 8.263363754889179,
-      "grad_norm": 1.546433925628662,
-      "learning_rate": 8.458291457286432e-05,
-      "loss": 5.7368,
-      "step": 15845
-    },
-    {
-      "epoch": 8.263885267275098,
-      "grad_norm": 1.5821900367736816,
-      "learning_rate": 8.45819095477387e-05,
-      "loss": 5.5289,
-      "step": 15846
-    },
-    {
-      "epoch": 8.264406779661018,
-      "grad_norm": 1.3721505403518677,
-      "learning_rate": 8.458090452261307e-05,
-      "loss": 5.6886,
-      "step": 15847
-    },
-    {
-      "epoch": 8.264928292046935,
-      "grad_norm": 1.3780944347381592,
-      "learning_rate": 8.457989949748743e-05,
-      "loss": 5.6399,
-      "step": 15848
-    },
-    {
-      "epoch": 8.265449804432855,
-      "grad_norm": 1.4012144804000854,
-      "learning_rate": 8.457889447236181e-05,
-      "loss": 5.4458,
-      "step": 15849
-    },
-    {
-      "epoch": 8.265971316818774,
-      "grad_norm": 1.578778624534607,
-      "learning_rate": 8.457788944723619e-05,
-      "loss": 5.3058,
-      "step": 15850
-    },
-    {
-      "epoch": 8.266492829204694,
-      "grad_norm": 1.4275809526443481,
-      "learning_rate": 8.457688442211056e-05,
-      "loss": 5.5585,
-      "step": 15851
-    },
-    {
-      "epoch": 8.267014341590613,
-      "grad_norm": 1.37882399559021,
-      "learning_rate": 8.457587939698493e-05,
-      "loss": 5.5629,
-      "step": 15852
-    },
-    {
-      "epoch": 8.267535853976533,
-      "grad_norm": 1.4180371761322021,
-      "learning_rate": 8.45748743718593e-05,
-      "loss": 5.6888,
-      "step": 15853
-    },
-    {
-      "epoch": 8.26805736636245,
-      "grad_norm": 1.3796091079711914,
-      "learning_rate": 8.457386934673367e-05,
-      "loss": 5.9032,
-      "step": 15854
-    },
-    {
-      "epoch": 8.26857887874837,
-      "grad_norm": 1.4430314302444458,
-      "learning_rate": 8.457286432160805e-05,
-      "loss": 5.0611,
-      "step": 15855
-    },
-    {
-      "epoch": 8.26910039113429,
-      "grad_norm": 1.3288707733154297,
-      "learning_rate": 8.457185929648241e-05,
-      "loss": 5.9358,
-      "step": 15856
-    },
-    {
-      "epoch": 8.269621903520209,
-      "grad_norm": 1.5671818256378174,
-      "learning_rate": 8.457085427135679e-05,
-      "loss": 5.3142,
-      "step": 15857
-    },
-    {
-      "epoch": 8.270143415906128,
-      "grad_norm": 1.3701246976852417,
-      "learning_rate": 8.456984924623116e-05,
-      "loss": 5.8527,
-      "step": 15858
-    },
-    {
-      "epoch": 8.270664928292048,
-      "grad_norm": 1.4828314781188965,
-      "learning_rate": 8.456884422110553e-05,
-      "loss": 5.4784,
-      "step": 15859
-    },
-    {
-      "epoch": 8.271186440677965,
-      "grad_norm": 1.5456962585449219,
-      "learning_rate": 8.45678391959799e-05,
-      "loss": 5.5513,
-      "step": 15860
-    },
-    {
-      "epoch": 8.271707953063885,
-      "grad_norm": 1.474442720413208,
-      "learning_rate": 8.456683417085427e-05,
-      "loss": 5.451,
-      "step": 15861
-    },
-    {
-      "epoch": 8.272229465449804,
-      "grad_norm": 1.7194995880126953,
-      "learning_rate": 8.456582914572865e-05,
-      "loss": 5.4999,
-      "step": 15862
-    },
-    {
-      "epoch": 8.272750977835724,
-      "grad_norm": 1.4589446783065796,
-      "learning_rate": 8.456482412060302e-05,
-      "loss": 5.6123,
-      "step": 15863
-    },
-    {
-      "epoch": 8.273272490221643,
-      "grad_norm": 1.4888437986373901,
-      "learning_rate": 8.45638190954774e-05,
-      "loss": 5.4977,
-      "step": 15864
-    },
-    {
-      "epoch": 8.273794002607563,
-      "grad_norm": 1.4184558391571045,
-      "learning_rate": 8.456281407035176e-05,
-      "loss": 5.2206,
-      "step": 15865
-    },
-    {
-      "epoch": 8.27431551499348,
-      "grad_norm": 1.6213154792785645,
-      "learning_rate": 8.456180904522614e-05,
-      "loss": 5.3435,
-      "step": 15866
-    },
-    {
-      "epoch": 8.2748370273794,
-      "grad_norm": 1.397697925567627,
-      "learning_rate": 8.45608040201005e-05,
-      "loss": 5.872,
-      "step": 15867
-    },
-    {
-      "epoch": 8.27535853976532,
-      "grad_norm": 1.3756823539733887,
-      "learning_rate": 8.455979899497488e-05,
-      "loss": 4.6247,
-      "step": 15868
-    },
-    {
-      "epoch": 8.275880052151239,
-      "grad_norm": 1.330529808998108,
-      "learning_rate": 8.455879396984924e-05,
-      "loss": 5.8374,
-      "step": 15869
-    },
-    {
-      "epoch": 8.276401564537158,
-      "grad_norm": 1.3128541707992554,
-      "learning_rate": 8.455778894472362e-05,
-      "loss": 5.8602,
-      "step": 15870
-    },
-    {
-      "epoch": 8.276923076923078,
-      "grad_norm": 1.377893090248108,
-      "learning_rate": 8.4556783919598e-05,
-      "loss": 5.7595,
-      "step": 15871
-    },
-    {
-      "epoch": 8.277444589308995,
-      "grad_norm": 1.273043155670166,
-      "learning_rate": 8.455577889447238e-05,
-      "loss": 5.8662,
-      "step": 15872
-    },
-    {
-      "epoch": 8.277966101694915,
-      "grad_norm": 1.4187883138656616,
-      "learning_rate": 8.455477386934674e-05,
-      "loss": 5.7781,
-      "step": 15873
-    },
-    {
-      "epoch": 8.278487614080834,
-      "grad_norm": 1.4822087287902832,
-      "learning_rate": 8.455376884422112e-05,
-      "loss": 4.9331,
-      "step": 15874
-    },
-    {
-      "epoch": 8.279009126466754,
-      "grad_norm": 1.4608142375946045,
-      "learning_rate": 8.455276381909548e-05,
-      "loss": 5.7062,
-      "step": 15875
-    },
-    {
-      "epoch": 8.279530638852673,
-      "grad_norm": 1.3936388492584229,
-      "learning_rate": 8.455175879396985e-05,
-      "loss": 5.7084,
-      "step": 15876
-    },
-    {
-      "epoch": 8.280052151238593,
-      "grad_norm": 1.477307915687561,
-      "learning_rate": 8.455075376884423e-05,
-      "loss": 5.5546,
-      "step": 15877
-    },
-    {
-      "epoch": 8.28057366362451,
-      "grad_norm": 1.495760202407837,
-      "learning_rate": 8.454974874371859e-05,
-      "loss": 5.5461,
-      "step": 15878
-    },
-    {
-      "epoch": 8.28109517601043,
-      "grad_norm": 1.4030472040176392,
-      "learning_rate": 8.454874371859297e-05,
-      "loss": 5.6119,
-      "step": 15879
-    },
-    {
-      "epoch": 8.28161668839635,
-      "grad_norm": 1.4525583982467651,
-      "learning_rate": 8.454773869346733e-05,
-      "loss": 5.8582,
-      "step": 15880
-    },
-    {
-      "epoch": 8.282138200782269,
-      "grad_norm": 1.4331893920898438,
-      "learning_rate": 8.454673366834171e-05,
-      "loss": 5.0769,
-      "step": 15881
-    },
-    {
-      "epoch": 8.282659713168188,
-      "grad_norm": 1.4314820766448975,
-      "learning_rate": 8.454572864321609e-05,
-      "loss": 5.405,
-      "step": 15882
-    },
-    {
-      "epoch": 8.283181225554108,
-      "grad_norm": 1.67923104763031,
-      "learning_rate": 8.454472361809047e-05,
-      "loss": 4.7148,
-      "step": 15883
-    },
-    {
-      "epoch": 8.283702737940025,
-      "grad_norm": 1.4646562337875366,
-      "learning_rate": 8.454371859296483e-05,
-      "loss": 5.2748,
-      "step": 15884
-    },
-    {
-      "epoch": 8.284224250325945,
-      "grad_norm": 1.3863946199417114,
-      "learning_rate": 8.454271356783921e-05,
-      "loss": 5.5994,
-      "step": 15885
-    },
-    {
-      "epoch": 8.284745762711864,
-      "grad_norm": 1.3674498796463013,
-      "learning_rate": 8.454170854271357e-05,
-      "loss": 5.9057,
-      "step": 15886
-    },
-    {
-      "epoch": 8.285267275097784,
-      "grad_norm": 1.4301966428756714,
-      "learning_rate": 8.454070351758795e-05,
-      "loss": 5.3135,
-      "step": 15887
-    },
-    {
-      "epoch": 8.285788787483703,
-      "grad_norm": 1.4029064178466797,
-      "learning_rate": 8.453969849246231e-05,
-      "loss": 5.8095,
-      "step": 15888
-    },
-    {
-      "epoch": 8.286310299869623,
-      "grad_norm": 1.363068699836731,
-      "learning_rate": 8.453869346733668e-05,
-      "loss": 5.5714,
-      "step": 15889
-    },
-    {
-      "epoch": 8.28683181225554,
-      "grad_norm": 1.5678194761276245,
-      "learning_rate": 8.453768844221106e-05,
-      "loss": 5.3005,
-      "step": 15890
-    },
-    {
-      "epoch": 8.28735332464146,
-      "grad_norm": 1.3899160623550415,
-      "learning_rate": 8.453668341708543e-05,
-      "loss": 5.6153,
-      "step": 15891
-    },
-    {
-      "epoch": 8.28787483702738,
-      "grad_norm": 1.6290831565856934,
-      "learning_rate": 8.453567839195981e-05,
-      "loss": 5.3544,
-      "step": 15892
-    },
-    {
-      "epoch": 8.288396349413299,
-      "grad_norm": 1.3395203351974487,
-      "learning_rate": 8.453467336683418e-05,
-      "loss": 5.8449,
-      "step": 15893
-    },
-    {
-      "epoch": 8.288917861799218,
-      "grad_norm": 1.426565170288086,
-      "learning_rate": 8.453366834170855e-05,
-      "loss": 5.7371,
-      "step": 15894
-    },
-    {
-      "epoch": 8.289439374185136,
-      "grad_norm": 1.3848758935928345,
-      "learning_rate": 8.453266331658292e-05,
-      "loss": 5.7198,
-      "step": 15895
-    },
-    {
-      "epoch": 8.289960886571055,
-      "grad_norm": 1.4147223234176636,
-      "learning_rate": 8.45316582914573e-05,
-      "loss": 5.6899,
-      "step": 15896
-    },
-    {
-      "epoch": 8.290482398956975,
-      "grad_norm": 1.3713607788085938,
-      "learning_rate": 8.453065326633166e-05,
-      "loss": 5.3943,
-      "step": 15897
-    },
-    {
-      "epoch": 8.291003911342894,
-      "grad_norm": 1.4115911722183228,
-      "learning_rate": 8.452964824120604e-05,
-      "loss": 5.4378,
-      "step": 15898
-    },
-    {
-      "epoch": 8.291525423728814,
-      "grad_norm": 1.4592313766479492,
-      "learning_rate": 8.45286432160804e-05,
-      "loss": 5.9101,
-      "step": 15899
-    },
-    {
-      "epoch": 8.292046936114733,
-      "grad_norm": 1.4952404499053955,
-      "learning_rate": 8.452763819095478e-05,
-      "loss": 5.4142,
-      "step": 15900
-    },
-    {
-      "epoch": 8.292568448500653,
-      "grad_norm": 1.5101289749145508,
-      "learning_rate": 8.452663316582914e-05,
-      "loss": 5.2558,
-      "step": 15901
-    },
-    {
-      "epoch": 8.29308996088657,
-      "grad_norm": 1.3972727060317993,
-      "learning_rate": 8.452562814070352e-05,
-      "loss": 5.6953,
-      "step": 15902
-    },
-    {
-      "epoch": 8.29361147327249,
-      "grad_norm": 1.4211363792419434,
-      "learning_rate": 8.45246231155779e-05,
-      "loss": 4.922,
-      "step": 15903
-    },
-    {
-      "epoch": 8.29413298565841,
-      "grad_norm": 1.3111618757247925,
-      "learning_rate": 8.452361809045226e-05,
-      "loss": 5.9986,
-      "step": 15904
-    },
-    {
-      "epoch": 8.294654498044329,
-      "grad_norm": 1.5290096998214722,
-      "learning_rate": 8.452261306532664e-05,
-      "loss": 5.6036,
-      "step": 15905
-    },
-    {
-      "epoch": 8.295176010430248,
-      "grad_norm": 1.477309226989746,
-      "learning_rate": 8.4521608040201e-05,
-      "loss": 5.8305,
-      "step": 15906
-    },
-    {
-      "epoch": 8.295697522816166,
-      "grad_norm": 1.2797796726226807,
-      "learning_rate": 8.452060301507538e-05,
-      "loss": 5.626,
-      "step": 15907
-    },
-    {
-      "epoch": 8.296219035202085,
-      "grad_norm": 1.4107009172439575,
-      "learning_rate": 8.451959798994975e-05,
-      "loss": 5.4285,
-      "step": 15908
-    },
-    {
-      "epoch": 8.296740547588005,
-      "grad_norm": 1.4235445261001587,
-      "learning_rate": 8.451859296482413e-05,
-      "loss": 5.6693,
-      "step": 15909
-    },
-    {
-      "epoch": 8.297262059973924,
-      "grad_norm": 1.4250483512878418,
-      "learning_rate": 8.451758793969849e-05,
-      "loss": 5.7822,
-      "step": 15910
-    },
-    {
-      "epoch": 8.297783572359844,
-      "grad_norm": 1.4717257022857666,
-      "learning_rate": 8.451658291457287e-05,
-      "loss": 5.4109,
-      "step": 15911
-    },
-    {
-      "epoch": 8.298305084745763,
-      "grad_norm": 1.503605604171753,
-      "learning_rate": 8.451557788944725e-05,
-      "loss": 5.27,
-      "step": 15912
-    },
-    {
-      "epoch": 8.298826597131681,
-      "grad_norm": 1.4258735179901123,
-      "learning_rate": 8.451457286432162e-05,
-      "loss": 5.3671,
-      "step": 15913
-    },
-    {
-      "epoch": 8.2993481095176,
-      "grad_norm": 1.4839059114456177,
-      "learning_rate": 8.451356783919599e-05,
-      "loss": 5.6198,
-      "step": 15914
-    },
-    {
-      "epoch": 8.29986962190352,
-      "grad_norm": 1.4041774272918701,
-      "learning_rate": 8.451256281407035e-05,
-      "loss": 5.685,
-      "step": 15915
-    },
-    {
-      "epoch": 8.30039113428944,
-      "grad_norm": 1.649997591972351,
-      "learning_rate": 8.451155778894473e-05,
-      "loss": 5.733,
-      "step": 15916
-    },
-    {
-      "epoch": 8.300912646675359,
-      "grad_norm": 1.3233736753463745,
-      "learning_rate": 8.45105527638191e-05,
-      "loss": 5.9413,
-      "step": 15917
-    },
-    {
-      "epoch": 8.301434159061278,
-      "grad_norm": 1.3886852264404297,
-      "learning_rate": 8.450954773869347e-05,
-      "loss": 5.6069,
-      "step": 15918
-    },
-    {
-      "epoch": 8.301955671447196,
-      "grad_norm": 1.3759963512420654,
-      "learning_rate": 8.450854271356784e-05,
-      "loss": 5.446,
-      "step": 15919
-    },
-    {
-      "epoch": 8.302477183833116,
-      "grad_norm": 1.6840280294418335,
-      "learning_rate": 8.450753768844221e-05,
-      "loss": 4.9565,
-      "step": 15920
-    },
-    {
-      "epoch": 8.302998696219035,
-      "grad_norm": 1.5398145914077759,
-      "learning_rate": 8.450653266331658e-05,
-      "loss": 5.3624,
-      "step": 15921
-    },
-    {
-      "epoch": 8.303520208604954,
-      "grad_norm": 1.6161919832229614,
-      "learning_rate": 8.450552763819096e-05,
-      "loss": 5.122,
-      "step": 15922
-    },
-    {
-      "epoch": 8.304041720990874,
-      "grad_norm": 1.5116100311279297,
-      "learning_rate": 8.450452261306533e-05,
-      "loss": 5.1173,
-      "step": 15923
-    },
-    {
-      "epoch": 8.304563233376793,
-      "grad_norm": 1.438167929649353,
-      "learning_rate": 8.450351758793971e-05,
-      "loss": 5.6139,
-      "step": 15924
-    },
-    {
-      "epoch": 8.305084745762711,
-      "grad_norm": 1.711519718170166,
-      "learning_rate": 8.450251256281408e-05,
-      "loss": 4.8603,
-      "step": 15925
-    },
-    {
-      "epoch": 8.30560625814863,
-      "grad_norm": 1.4468477964401245,
-      "learning_rate": 8.450150753768845e-05,
-      "loss": 5.3458,
-      "step": 15926
-    },
-    {
-      "epoch": 8.30612777053455,
-      "grad_norm": 1.5403406620025635,
-      "learning_rate": 8.450050251256282e-05,
-      "loss": 5.753,
-      "step": 15927
-    },
-    {
-      "epoch": 8.30664928292047,
-      "grad_norm": 1.4824362993240356,
-      "learning_rate": 8.449949748743718e-05,
-      "loss": 5.8273,
-      "step": 15928
-    },
-    {
-      "epoch": 8.307170795306389,
-      "grad_norm": 1.5934383869171143,
-      "learning_rate": 8.449849246231156e-05,
-      "loss": 5.1556,
-      "step": 15929
-    },
-    {
-      "epoch": 8.307692307692308,
-      "grad_norm": 1.4519424438476562,
-      "learning_rate": 8.449748743718592e-05,
-      "loss": 5.8314,
-      "step": 15930
-    },
-    {
-      "epoch": 8.308213820078226,
-      "grad_norm": 1.5523474216461182,
-      "learning_rate": 8.44964824120603e-05,
-      "loss": 5.358,
-      "step": 15931
-    },
-    {
-      "epoch": 8.308735332464146,
-      "grad_norm": 1.3591452836990356,
-      "learning_rate": 8.449547738693468e-05,
-      "loss": 5.9337,
-      "step": 15932
-    },
-    {
-      "epoch": 8.309256844850065,
-      "grad_norm": 1.624999761581421,
-      "learning_rate": 8.449447236180906e-05,
-      "loss": 4.902,
-      "step": 15933
-    },
-    {
-      "epoch": 8.309778357235984,
-      "grad_norm": 1.460902452468872,
-      "learning_rate": 8.449346733668342e-05,
-      "loss": 5.3314,
-      "step": 15934
-    },
-    {
-      "epoch": 8.310299869621904,
-      "grad_norm": 1.4620457887649536,
-      "learning_rate": 8.44924623115578e-05,
-      "loss": 5.4444,
-      "step": 15935
-    },
-    {
-      "epoch": 8.310821382007823,
-      "grad_norm": 1.4059715270996094,
-      "learning_rate": 8.449145728643216e-05,
-      "loss": 5.6099,
-      "step": 15936
-    },
-    {
-      "epoch": 8.311342894393741,
-      "grad_norm": 1.5995643138885498,
-      "learning_rate": 8.449045226130654e-05,
-      "loss": 5.1441,
-      "step": 15937
-    },
-    {
-      "epoch": 8.31186440677966,
-      "grad_norm": 1.4582644701004028,
-      "learning_rate": 8.448944723618091e-05,
-      "loss": 5.4897,
-      "step": 15938
-    },
-    {
-      "epoch": 8.31238591916558,
-      "grad_norm": 1.4212092161178589,
-      "learning_rate": 8.448844221105528e-05,
-      "loss": 5.642,
-      "step": 15939
-    },
-    {
-      "epoch": 8.3129074315515,
-      "grad_norm": 1.4048030376434326,
-      "learning_rate": 8.448743718592965e-05,
-      "loss": 5.5626,
-      "step": 15940
-    },
-    {
-      "epoch": 8.313428943937419,
-      "grad_norm": 1.4095392227172852,
-      "learning_rate": 8.448643216080401e-05,
-      "loss": 5.595,
-      "step": 15941
-    },
-    {
-      "epoch": 8.313950456323338,
-      "grad_norm": 1.3824983835220337,
-      "learning_rate": 8.448542713567839e-05,
-      "loss": 5.5757,
-      "step": 15942
-    },
-    {
-      "epoch": 8.314471968709256,
-      "grad_norm": 1.466179609298706,
-      "learning_rate": 8.448442211055277e-05,
-      "loss": 5.5241,
-      "step": 15943
-    },
-    {
-      "epoch": 8.314993481095176,
-      "grad_norm": 1.5262266397476196,
-      "learning_rate": 8.448341708542715e-05,
-      "loss": 5.3025,
-      "step": 15944
-    },
-    {
-      "epoch": 8.315514993481095,
-      "grad_norm": 1.3975412845611572,
-      "learning_rate": 8.448241206030151e-05,
-      "loss": 5.5968,
-      "step": 15945
-    },
-    {
-      "epoch": 8.316036505867014,
-      "grad_norm": 1.4115992784500122,
-      "learning_rate": 8.448140703517589e-05,
-      "loss": 5.5584,
-      "step": 15946
-    },
-    {
-      "epoch": 8.316558018252934,
-      "grad_norm": 1.307379961013794,
-      "learning_rate": 8.448040201005025e-05,
-      "loss": 5.7984,
-      "step": 15947
-    },
-    {
-      "epoch": 8.317079530638853,
-      "grad_norm": 1.4425749778747559,
-      "learning_rate": 8.447939698492463e-05,
-      "loss": 5.6679,
-      "step": 15948
-    },
-    {
-      "epoch": 8.317601043024771,
-      "grad_norm": 1.3776566982269287,
-      "learning_rate": 8.4478391959799e-05,
-      "loss": 5.1993,
-      "step": 15949
-    },
-    {
-      "epoch": 8.31812255541069,
-      "grad_norm": 1.3335384130477905,
-      "learning_rate": 8.447738693467337e-05,
-      "loss": 5.7974,
-      "step": 15950
-    },
-    {
-      "epoch": 8.31864406779661,
-      "grad_norm": 1.291144609451294,
-      "learning_rate": 8.447638190954774e-05,
-      "loss": 5.7945,
-      "step": 15951
-    },
-    {
-      "epoch": 8.31916558018253,
-      "grad_norm": 1.39948570728302,
-      "learning_rate": 8.447537688442212e-05,
-      "loss": 5.1874,
-      "step": 15952
-    },
-    {
-      "epoch": 8.319687092568449,
-      "grad_norm": 1.533009648323059,
-      "learning_rate": 8.447437185929648e-05,
-      "loss": 5.2387,
-      "step": 15953
-    },
-    {
-      "epoch": 8.320208604954368,
-      "grad_norm": 1.42972993850708,
-      "learning_rate": 8.447336683417086e-05,
-      "loss": 5.8858,
-      "step": 15954
-    },
-    {
-      "epoch": 8.320730117340286,
-      "grad_norm": 1.391144871711731,
-      "learning_rate": 8.447236180904524e-05,
-      "loss": 5.6155,
-      "step": 15955
-    },
-    {
-      "epoch": 8.321251629726206,
-      "grad_norm": 1.4496901035308838,
-      "learning_rate": 8.44713567839196e-05,
-      "loss": 5.8481,
-      "step": 15956
-    },
-    {
-      "epoch": 8.321773142112125,
-      "grad_norm": 1.4705294370651245,
-      "learning_rate": 8.447035175879398e-05,
-      "loss": 4.9626,
-      "step": 15957
-    },
-    {
-      "epoch": 8.322294654498045,
-      "grad_norm": 1.395022988319397,
-      "learning_rate": 8.446934673366834e-05,
-      "loss": 5.3669,
-      "step": 15958
-    },
-    {
-      "epoch": 8.322816166883964,
-      "grad_norm": 1.4462367296218872,
-      "learning_rate": 8.446834170854272e-05,
-      "loss": 5.6842,
-      "step": 15959
-    },
-    {
-      "epoch": 8.323337679269883,
-      "grad_norm": 1.3980281352996826,
-      "learning_rate": 8.446733668341708e-05,
-      "loss": 5.3362,
-      "step": 15960
-    },
-    {
-      "epoch": 8.323859191655801,
-      "grad_norm": 1.4465581178665161,
-      "learning_rate": 8.446633165829146e-05,
-      "loss": 5.3237,
-      "step": 15961
-    },
-    {
-      "epoch": 8.32438070404172,
-      "grad_norm": 1.3908741474151611,
-      "learning_rate": 8.446532663316583e-05,
-      "loss": 5.3435,
-      "step": 15962
-    },
-    {
-      "epoch": 8.32490221642764,
-      "grad_norm": 1.3185327053070068,
-      "learning_rate": 8.44643216080402e-05,
-      "loss": 5.6351,
-      "step": 15963
-    },
-    {
-      "epoch": 8.32542372881356,
-      "grad_norm": 1.4448957443237305,
-      "learning_rate": 8.446331658291458e-05,
-      "loss": 5.7112,
-      "step": 15964
-    },
-    {
-      "epoch": 8.325945241199479,
-      "grad_norm": 1.4519954919815063,
-      "learning_rate": 8.446231155778896e-05,
-      "loss": 5.7076,
-      "step": 15965
-    },
-    {
-      "epoch": 8.326466753585398,
-      "grad_norm": 1.2674859762191772,
-      "learning_rate": 8.446130653266332e-05,
-      "loss": 5.7608,
-      "step": 15966
-    },
-    {
-      "epoch": 8.326988265971316,
-      "grad_norm": 1.4240796566009521,
-      "learning_rate": 8.44603015075377e-05,
-      "loss": 5.8488,
-      "step": 15967
-    },
-    {
-      "epoch": 8.327509778357236,
-      "grad_norm": 1.5564864873886108,
-      "learning_rate": 8.445929648241207e-05,
-      "loss": 5.568,
-      "step": 15968
-    },
-    {
-      "epoch": 8.328031290743155,
-      "grad_norm": 1.602820634841919,
-      "learning_rate": 8.445829145728643e-05,
-      "loss": 5.7399,
-      "step": 15969
-    },
-    {
-      "epoch": 8.328552803129075,
-      "grad_norm": 1.5959762334823608,
-      "learning_rate": 8.445728643216081e-05,
-      "loss": 5.5063,
-      "step": 15970
-    },
-    {
-      "epoch": 8.329074315514994,
-      "grad_norm": 1.5072026252746582,
-      "learning_rate": 8.445628140703517e-05,
-      "loss": 5.5989,
-      "step": 15971
-    },
-    {
-      "epoch": 8.329595827900913,
-      "grad_norm": 1.5525572299957275,
-      "learning_rate": 8.445527638190955e-05,
-      "loss": 5.297,
-      "step": 15972
-    },
-    {
-      "epoch": 8.330117340286831,
-      "grad_norm": 1.4932000637054443,
-      "learning_rate": 8.445427135678391e-05,
-      "loss": 5.6033,
-      "step": 15973
-    },
-    {
-      "epoch": 8.33063885267275,
-      "grad_norm": 1.399903655052185,
-      "learning_rate": 8.445326633165829e-05,
-      "loss": 5.6169,
-      "step": 15974
-    },
-    {
-      "epoch": 8.33116036505867,
-      "grad_norm": 1.3966963291168213,
-      "learning_rate": 8.445226130653267e-05,
-      "loss": 5.1716,
-      "step": 15975
-    },
-    {
-      "epoch": 8.33168187744459,
-      "grad_norm": 1.529606819152832,
-      "learning_rate": 8.445125628140705e-05,
-      "loss": 5.4551,
-      "step": 15976
-    },
-    {
-      "epoch": 8.332203389830509,
-      "grad_norm": 1.478464126586914,
-      "learning_rate": 8.445025125628141e-05,
-      "loss": 5.5018,
-      "step": 15977
-    },
-    {
-      "epoch": 8.332724902216428,
-      "grad_norm": 1.4331082105636597,
-      "learning_rate": 8.444924623115579e-05,
-      "loss": 5.383,
-      "step": 15978
-    },
-    {
-      "epoch": 8.333246414602346,
-      "grad_norm": 1.5272414684295654,
-      "learning_rate": 8.444824120603015e-05,
-      "loss": 5.4435,
-      "step": 15979
-    },
-    {
-      "epoch": 8.333767926988266,
-      "grad_norm": 1.4849690198898315,
-      "learning_rate": 8.444723618090453e-05,
-      "loss": 5.4343,
-      "step": 15980
-    },
-    {
-      "epoch": 8.334289439374185,
-      "grad_norm": 1.4924222230911255,
-      "learning_rate": 8.44462311557789e-05,
-      "loss": 5.3274,
-      "step": 15981
-    },
-    {
-      "epoch": 8.334810951760105,
-      "grad_norm": 1.4785796403884888,
-      "learning_rate": 8.444522613065326e-05,
-      "loss": 5.2807,
-      "step": 15982
-    },
-    {
-      "epoch": 8.335332464146024,
-      "grad_norm": 1.3742555379867554,
-      "learning_rate": 8.444422110552764e-05,
-      "loss": 5.4324,
-      "step": 15983
-    },
-    {
-      "epoch": 8.335853976531943,
-      "grad_norm": 1.3757057189941406,
-      "learning_rate": 8.444321608040202e-05,
-      "loss": 5.782,
-      "step": 15984
-    },
-    {
-      "epoch": 8.336375488917861,
-      "grad_norm": 1.4534735679626465,
-      "learning_rate": 8.44422110552764e-05,
-      "loss": 5.1405,
-      "step": 15985
-    },
-    {
-      "epoch": 8.33689700130378,
-      "grad_norm": 1.5605055093765259,
-      "learning_rate": 8.444120603015076e-05,
-      "loss": 5.4312,
-      "step": 15986
-    },
-    {
-      "epoch": 8.3374185136897,
-      "grad_norm": 1.5666334629058838,
-      "learning_rate": 8.444020100502514e-05,
-      "loss": 5.3037,
-      "step": 15987
-    },
-    {
-      "epoch": 8.33794002607562,
-      "grad_norm": 1.497889518737793,
-      "learning_rate": 8.44391959798995e-05,
-      "loss": 5.276,
-      "step": 15988
-    },
-    {
-      "epoch": 8.338461538461539,
-      "grad_norm": 1.3910679817199707,
-      "learning_rate": 8.443819095477388e-05,
-      "loss": 5.5844,
-      "step": 15989
-    },
-    {
-      "epoch": 8.338983050847457,
-      "grad_norm": 1.362622857093811,
-      "learning_rate": 8.443718592964824e-05,
-      "loss": 5.6741,
-      "step": 15990
-    },
-    {
-      "epoch": 8.339504563233376,
-      "grad_norm": 1.4490845203399658,
-      "learning_rate": 8.443618090452262e-05,
-      "loss": 5.5092,
-      "step": 15991
-    },
-    {
-      "epoch": 8.340026075619296,
-      "grad_norm": 1.3583098649978638,
-      "learning_rate": 8.443517587939698e-05,
-      "loss": 5.4207,
-      "step": 15992
-    },
-    {
-      "epoch": 8.340547588005215,
-      "grad_norm": 1.3696868419647217,
-      "learning_rate": 8.443417085427136e-05,
-      "loss": 5.7722,
-      "step": 15993
-    },
-    {
-      "epoch": 8.341069100391135,
-      "grad_norm": 1.4488942623138428,
-      "learning_rate": 8.443316582914573e-05,
-      "loss": 5.8139,
-      "step": 15994
-    },
-    {
-      "epoch": 8.341590612777054,
-      "grad_norm": 1.3814903497695923,
-      "learning_rate": 8.44321608040201e-05,
-      "loss": 5.6847,
-      "step": 15995
-    },
-    {
-      "epoch": 8.342112125162974,
-      "grad_norm": 1.458188533782959,
-      "learning_rate": 8.443115577889448e-05,
-      "loss": 5.3698,
-      "step": 15996
-    },
-    {
-      "epoch": 8.342633637548891,
-      "grad_norm": 1.3734210729599,
-      "learning_rate": 8.443015075376885e-05,
-      "loss": 5.4613,
-      "step": 15997
-    },
-    {
-      "epoch": 8.34315514993481,
-      "grad_norm": 1.4795769453048706,
-      "learning_rate": 8.442914572864322e-05,
-      "loss": 5.232,
-      "step": 15998
-    },
-    {
-      "epoch": 8.34367666232073,
-      "grad_norm": 1.5418455600738525,
-      "learning_rate": 8.442814070351759e-05,
-      "loss": 5.4994,
-      "step": 15999
-    },
-    {
-      "epoch": 8.34419817470665,
-      "grad_norm": 1.504374623298645,
-      "learning_rate": 8.442713567839197e-05,
-      "loss": 5.55,
-      "step": 16000
-    },
-    {
-      "epoch": 8.344719687092569,
-      "grad_norm": 1.6758397817611694,
-      "learning_rate": 8.442613065326633e-05,
-      "loss": 4.9375,
-      "step": 16001
-    },
-    {
-      "epoch": 8.345241199478487,
-      "grad_norm": 1.4920449256896973,
-      "learning_rate": 8.442512562814071e-05,
-      "loss": 5.2795,
-      "step": 16002
-    },
-    {
-      "epoch": 8.345762711864406,
-      "grad_norm": 1.4816075563430786,
-      "learning_rate": 8.442412060301507e-05,
-      "loss": 5.3093,
-      "step": 16003
-    },
-    {
-      "epoch": 8.346284224250326,
-      "grad_norm": 1.4014040231704712,
-      "learning_rate": 8.442311557788945e-05,
-      "loss": 5.0151,
-      "step": 16004
-    },
-    {
-      "epoch": 8.346805736636245,
-      "grad_norm": 1.4864224195480347,
-      "learning_rate": 8.442211055276383e-05,
-      "loss": 5.097,
-      "step": 16005
-    },
-    {
-      "epoch": 8.347327249022165,
-      "grad_norm": 1.414717435836792,
-      "learning_rate": 8.44211055276382e-05,
-      "loss": 5.7251,
-      "step": 16006
-    },
-    {
-      "epoch": 8.347848761408084,
-      "grad_norm": 1.385803461074829,
-      "learning_rate": 8.442010050251257e-05,
-      "loss": 5.6724,
-      "step": 16007
-    },
-    {
-      "epoch": 8.348370273794002,
-      "grad_norm": 1.4936527013778687,
-      "learning_rate": 8.441909547738693e-05,
-      "loss": 5.3157,
-      "step": 16008
-    },
-    {
-      "epoch": 8.348891786179921,
-      "grad_norm": 1.3844199180603027,
-      "learning_rate": 8.441809045226131e-05,
-      "loss": 5.7103,
-      "step": 16009
-    },
-    {
-      "epoch": 8.34941329856584,
-      "grad_norm": 1.628217339515686,
-      "learning_rate": 8.441708542713568e-05,
-      "loss": 5.5027,
-      "step": 16010
-    },
-    {
-      "epoch": 8.34993481095176,
-      "grad_norm": 1.4541045427322388,
-      "learning_rate": 8.441608040201005e-05,
-      "loss": 5.554,
-      "step": 16011
-    },
-    {
-      "epoch": 8.35045632333768,
-      "grad_norm": 1.4511072635650635,
-      "learning_rate": 8.441507537688442e-05,
-      "loss": 5.6127,
-      "step": 16012
-    },
-    {
-      "epoch": 8.350977835723599,
-      "grad_norm": 1.5201529264450073,
-      "learning_rate": 8.44140703517588e-05,
-      "loss": 5.8629,
-      "step": 16013
-    },
-    {
-      "epoch": 8.351499348109517,
-      "grad_norm": 1.3730664253234863,
-      "learning_rate": 8.441306532663316e-05,
-      "loss": 5.7893,
-      "step": 16014
-    },
-    {
-      "epoch": 8.352020860495436,
-      "grad_norm": 1.4325379133224487,
-      "learning_rate": 8.441206030150754e-05,
-      "loss": 5.8405,
-      "step": 16015
-    },
-    {
-      "epoch": 8.352542372881356,
-      "grad_norm": 1.4827306270599365,
-      "learning_rate": 8.441105527638192e-05,
-      "loss": 5.6013,
-      "step": 16016
-    },
-    {
-      "epoch": 8.353063885267275,
-      "grad_norm": 1.427308201789856,
-      "learning_rate": 8.44100502512563e-05,
-      "loss": 5.5002,
-      "step": 16017
-    },
-    {
-      "epoch": 8.353585397653195,
-      "grad_norm": 1.4410616159439087,
-      "learning_rate": 8.440904522613066e-05,
-      "loss": 5.5908,
-      "step": 16018
-    },
-    {
-      "epoch": 8.354106910039114,
-      "grad_norm": 1.468735933303833,
-      "learning_rate": 8.440804020100504e-05,
-      "loss": 5.7491,
-      "step": 16019
-    },
-    {
-      "epoch": 8.354628422425032,
-      "grad_norm": 1.4429700374603271,
-      "learning_rate": 8.44070351758794e-05,
-      "loss": 5.6745,
-      "step": 16020
-    },
-    {
-      "epoch": 8.355149934810951,
-      "grad_norm": 1.3804229497909546,
-      "learning_rate": 8.440603015075377e-05,
-      "loss": 5.6305,
-      "step": 16021
-    },
-    {
-      "epoch": 8.35567144719687,
-      "grad_norm": 1.4659391641616821,
-      "learning_rate": 8.440502512562814e-05,
-      "loss": 5.838,
-      "step": 16022
-    },
-    {
-      "epoch": 8.35619295958279,
-      "grad_norm": 1.3757655620574951,
-      "learning_rate": 8.440402010050251e-05,
-      "loss": 5.6665,
-      "step": 16023
-    },
-    {
-      "epoch": 8.35671447196871,
-      "grad_norm": 1.3690297603607178,
-      "learning_rate": 8.440301507537689e-05,
-      "loss": 5.6853,
-      "step": 16024
-    },
-    {
-      "epoch": 8.357235984354629,
-      "grad_norm": 1.6084791421890259,
-      "learning_rate": 8.440201005025126e-05,
-      "loss": 5.5372,
-      "step": 16025
-    },
-    {
-      "epoch": 8.357757496740547,
-      "grad_norm": 1.384351372718811,
-      "learning_rate": 8.440100502512564e-05,
-      "loss": 5.8988,
-      "step": 16026
-    },
-    {
-      "epoch": 8.358279009126466,
-      "grad_norm": 1.387410044670105,
-      "learning_rate": 8.44e-05,
-      "loss": 5.8066,
-      "step": 16027
-    },
-    {
-      "epoch": 8.358800521512386,
-      "grad_norm": 1.3674942255020142,
-      "learning_rate": 8.439899497487438e-05,
-      "loss": 5.6325,
-      "step": 16028
-    },
-    {
-      "epoch": 8.359322033898305,
-      "grad_norm": 1.5085363388061523,
-      "learning_rate": 8.439798994974875e-05,
-      "loss": 5.2262,
-      "step": 16029
-    },
-    {
-      "epoch": 8.359843546284225,
-      "grad_norm": 1.5006816387176514,
-      "learning_rate": 8.439698492462313e-05,
-      "loss": 5.5047,
-      "step": 16030
-    },
-    {
-      "epoch": 8.360365058670144,
-      "grad_norm": 1.5318262577056885,
-      "learning_rate": 8.439597989949749e-05,
-      "loss": 5.1853,
-      "step": 16031
-    },
-    {
-      "epoch": 8.360886571056062,
-      "grad_norm": 1.7897796630859375,
-      "learning_rate": 8.439497487437187e-05,
-      "loss": 5.2258,
-      "step": 16032
-    },
-    {
-      "epoch": 8.361408083441981,
-      "grad_norm": 1.5215789079666138,
-      "learning_rate": 8.439396984924623e-05,
-      "loss": 5.7157,
-      "step": 16033
-    },
-    {
-      "epoch": 8.3619295958279,
-      "grad_norm": 1.3863027095794678,
-      "learning_rate": 8.43929648241206e-05,
-      "loss": 5.4673,
-      "step": 16034
-    },
-    {
-      "epoch": 8.36245110821382,
-      "grad_norm": 1.4120126962661743,
-      "learning_rate": 8.439195979899497e-05,
-      "loss": 5.5129,
-      "step": 16035
-    },
-    {
-      "epoch": 8.36297262059974,
-      "grad_norm": 1.4492841958999634,
-      "learning_rate": 8.439095477386935e-05,
-      "loss": 5.3652,
-      "step": 16036
-    },
-    {
-      "epoch": 8.36349413298566,
-      "grad_norm": 1.4481147527694702,
-      "learning_rate": 8.438994974874373e-05,
-      "loss": 4.968,
-      "step": 16037
-    },
-    {
-      "epoch": 8.364015645371577,
-      "grad_norm": 1.4658496379852295,
-      "learning_rate": 8.43889447236181e-05,
-      "loss": 5.1149,
-      "step": 16038
-    },
-    {
-      "epoch": 8.364537157757496,
-      "grad_norm": 1.4367713928222656,
-      "learning_rate": 8.438793969849247e-05,
-      "loss": 5.533,
-      "step": 16039
-    },
-    {
-      "epoch": 8.365058670143416,
-      "grad_norm": 1.3960587978363037,
-      "learning_rate": 8.438693467336684e-05,
-      "loss": 5.5039,
-      "step": 16040
-    },
-    {
-      "epoch": 8.365580182529335,
-      "grad_norm": 1.346131682395935,
-      "learning_rate": 8.438592964824121e-05,
-      "loss": 5.1948,
-      "step": 16041
-    },
-    {
-      "epoch": 8.366101694915255,
-      "grad_norm": 1.4895384311676025,
-      "learning_rate": 8.438492462311558e-05,
-      "loss": 4.7733,
-      "step": 16042
-    },
-    {
-      "epoch": 8.366623207301174,
-      "grad_norm": 1.53669011592865,
-      "learning_rate": 8.438391959798996e-05,
-      "loss": 5.6217,
-      "step": 16043
-    },
-    {
-      "epoch": 8.367144719687092,
-      "grad_norm": 1.3956218957901,
-      "learning_rate": 8.438291457286432e-05,
-      "loss": 5.825,
-      "step": 16044
-    },
-    {
-      "epoch": 8.367666232073011,
-      "grad_norm": 1.4116652011871338,
-      "learning_rate": 8.43819095477387e-05,
-      "loss": 5.4489,
-      "step": 16045
-    },
-    {
-      "epoch": 8.36818774445893,
-      "grad_norm": 1.431934118270874,
-      "learning_rate": 8.438090452261308e-05,
-      "loss": 5.8106,
-      "step": 16046
-    },
-    {
-      "epoch": 8.36870925684485,
-      "grad_norm": 1.4541420936584473,
-      "learning_rate": 8.437989949748744e-05,
-      "loss": 5.8838,
-      "step": 16047
-    },
-    {
-      "epoch": 8.36923076923077,
-      "grad_norm": 1.4531906843185425,
-      "learning_rate": 8.437889447236182e-05,
-      "loss": 5.3466,
-      "step": 16048
-    },
-    {
-      "epoch": 8.36975228161669,
-      "grad_norm": 1.338457465171814,
-      "learning_rate": 8.437788944723618e-05,
-      "loss": 5.8195,
-      "step": 16049
-    },
-    {
-      "epoch": 8.370273794002607,
-      "grad_norm": 1.4263468980789185,
-      "learning_rate": 8.437688442211056e-05,
-      "loss": 5.5701,
-      "step": 16050
-    },
-    {
-      "epoch": 8.370795306388526,
-      "grad_norm": 1.570862889289856,
-      "learning_rate": 8.437587939698492e-05,
-      "loss": 5.4856,
-      "step": 16051
-    },
-    {
-      "epoch": 8.371316818774446,
-      "grad_norm": 1.418013095855713,
-      "learning_rate": 8.43748743718593e-05,
-      "loss": 5.1883,
-      "step": 16052
-    },
-    {
-      "epoch": 8.371838331160365,
-      "grad_norm": 1.3834482431411743,
-      "learning_rate": 8.437386934673367e-05,
-      "loss": 5.2519,
-      "step": 16053
-    },
-    {
-      "epoch": 8.372359843546285,
-      "grad_norm": 1.5436902046203613,
-      "learning_rate": 8.437286432160804e-05,
-      "loss": 5.3318,
-      "step": 16054
-    },
-    {
-      "epoch": 8.372881355932204,
-      "grad_norm": 1.510059118270874,
-      "learning_rate": 8.437185929648241e-05,
-      "loss": 5.1861,
-      "step": 16055
-    },
-    {
-      "epoch": 8.373402868318122,
-      "grad_norm": 1.556776762008667,
-      "learning_rate": 8.437085427135679e-05,
-      "loss": 4.9405,
-      "step": 16056
-    },
-    {
-      "epoch": 8.373924380704041,
-      "grad_norm": 1.4405328035354614,
-      "learning_rate": 8.436984924623116e-05,
-      "loss": 5.5123,
-      "step": 16057
-    },
-    {
-      "epoch": 8.37444589308996,
-      "grad_norm": 1.4164496660232544,
-      "learning_rate": 8.436884422110554e-05,
-      "loss": 5.2874,
-      "step": 16058
-    },
-    {
-      "epoch": 8.37496740547588,
-      "grad_norm": 1.3285411596298218,
-      "learning_rate": 8.43678391959799e-05,
-      "loss": 5.517,
-      "step": 16059
-    },
-    {
-      "epoch": 8.3754889178618,
-      "grad_norm": 1.5317302942276,
-      "learning_rate": 8.436683417085428e-05,
-      "loss": 4.6043,
-      "step": 16060
-    },
-    {
-      "epoch": 8.37601043024772,
-      "grad_norm": 1.3769731521606445,
-      "learning_rate": 8.436582914572865e-05,
-      "loss": 5.9274,
-      "step": 16061
-    },
-    {
-      "epoch": 8.376531942633637,
-      "grad_norm": 1.3655836582183838,
-      "learning_rate": 8.436482412060301e-05,
-      "loss": 5.7668,
-      "step": 16062
-    },
-    {
-      "epoch": 8.377053455019556,
-      "grad_norm": 1.5482333898544312,
-      "learning_rate": 8.436381909547739e-05,
-      "loss": 5.5191,
-      "step": 16063
-    },
-    {
-      "epoch": 8.377574967405476,
-      "grad_norm": 1.497146487236023,
-      "learning_rate": 8.436281407035175e-05,
-      "loss": 5.0571,
-      "step": 16064
-    },
-    {
-      "epoch": 8.378096479791395,
-      "grad_norm": 1.4152060747146606,
-      "learning_rate": 8.436180904522613e-05,
-      "loss": 5.487,
-      "step": 16065
-    },
-    {
-      "epoch": 8.378617992177315,
-      "grad_norm": 1.423811912536621,
-      "learning_rate": 8.436080402010051e-05,
-      "loss": 5.7586,
-      "step": 16066
-    },
-    {
-      "epoch": 8.379139504563234,
-      "grad_norm": 1.376279592514038,
-      "learning_rate": 8.435979899497489e-05,
-      "loss": 5.256,
-      "step": 16067
-    },
-    {
-      "epoch": 8.379661016949152,
-      "grad_norm": 1.3610820770263672,
-      "learning_rate": 8.435879396984925e-05,
-      "loss": 5.8684,
-      "step": 16068
-    },
-    {
-      "epoch": 8.380182529335071,
-      "grad_norm": 1.4266982078552246,
-      "learning_rate": 8.435778894472363e-05,
-      "loss": 5.6716,
-      "step": 16069
-    },
-    {
-      "epoch": 8.38070404172099,
-      "grad_norm": 1.4553724527359009,
-      "learning_rate": 8.4356783919598e-05,
-      "loss": 5.653,
-      "step": 16070
-    },
-    {
-      "epoch": 8.38122555410691,
-      "grad_norm": 1.3894379138946533,
-      "learning_rate": 8.435577889447237e-05,
-      "loss": 5.6149,
-      "step": 16071
-    },
-    {
-      "epoch": 8.38174706649283,
-      "grad_norm": 1.5245280265808105,
-      "learning_rate": 8.435477386934674e-05,
-      "loss": 5.2165,
-      "step": 16072
-    },
-    {
-      "epoch": 8.38226857887875,
-      "grad_norm": 1.4296437501907349,
-      "learning_rate": 8.435376884422111e-05,
-      "loss": 5.8494,
-      "step": 16073
-    },
-    {
-      "epoch": 8.382790091264667,
-      "grad_norm": 1.4475390911102295,
-      "learning_rate": 8.435276381909548e-05,
-      "loss": 5.3021,
-      "step": 16074
-    },
-    {
-      "epoch": 8.383311603650586,
-      "grad_norm": 1.459594964981079,
-      "learning_rate": 8.435175879396984e-05,
-      "loss": 5.342,
-      "step": 16075
-    },
-    {
-      "epoch": 8.383833116036506,
-      "grad_norm": 1.513715147972107,
-      "learning_rate": 8.435075376884422e-05,
-      "loss": 5.1607,
-      "step": 16076
-    },
-    {
-      "epoch": 8.384354628422425,
-      "grad_norm": 1.404284954071045,
-      "learning_rate": 8.43497487437186e-05,
-      "loss": 5.635,
-      "step": 16077
-    },
-    {
-      "epoch": 8.384876140808345,
-      "grad_norm": 1.4496219158172607,
-      "learning_rate": 8.434874371859298e-05,
-      "loss": 5.5622,
-      "step": 16078
-    },
-    {
-      "epoch": 8.385397653194264,
-      "grad_norm": 1.9338045120239258,
-      "learning_rate": 8.434773869346734e-05,
-      "loss": 4.9661,
-      "step": 16079
-    },
-    {
-      "epoch": 8.385919165580182,
-      "grad_norm": 1.490051031112671,
-      "learning_rate": 8.434673366834172e-05,
-      "loss": 5.299,
-      "step": 16080
-    },
-    {
-      "epoch": 8.386440677966101,
-      "grad_norm": 1.4671480655670166,
-      "learning_rate": 8.434572864321608e-05,
-      "loss": 5.63,
-      "step": 16081
-    },
-    {
-      "epoch": 8.38696219035202,
-      "grad_norm": 1.4292240142822266,
-      "learning_rate": 8.434472361809046e-05,
-      "loss": 5.6074,
-      "step": 16082
-    },
-    {
-      "epoch": 8.38748370273794,
-      "grad_norm": 1.665950059890747,
-      "learning_rate": 8.434371859296482e-05,
-      "loss": 5.3331,
-      "step": 16083
-    },
-    {
-      "epoch": 8.38800521512386,
-      "grad_norm": 1.5039023160934448,
-      "learning_rate": 8.43427135678392e-05,
-      "loss": 5.8048,
-      "step": 16084
-    },
-    {
-      "epoch": 8.388526727509777,
-      "grad_norm": 1.3658021688461304,
-      "learning_rate": 8.434170854271357e-05,
-      "loss": 5.3961,
-      "step": 16085
-    },
-    {
-      "epoch": 8.389048239895697,
-      "grad_norm": 1.390698790550232,
-      "learning_rate": 8.434070351758794e-05,
-      "loss": 5.4883,
-      "step": 16086
-    },
-    {
-      "epoch": 8.389569752281616,
-      "grad_norm": 1.3121788501739502,
-      "learning_rate": 8.433969849246232e-05,
-      "loss": 5.8658,
-      "step": 16087
-    },
-    {
-      "epoch": 8.390091264667536,
-      "grad_norm": 1.4939700365066528,
-      "learning_rate": 8.433869346733669e-05,
-      "loss": 5.405,
-      "step": 16088
-    },
-    {
-      "epoch": 8.390612777053455,
-      "grad_norm": 1.4339754581451416,
-      "learning_rate": 8.433768844221106e-05,
-      "loss": 5.6689,
-      "step": 16089
-    },
-    {
-      "epoch": 8.391134289439375,
-      "grad_norm": 1.4499372243881226,
-      "learning_rate": 8.433668341708543e-05,
-      "loss": 5.6579,
-      "step": 16090
-    },
-    {
-      "epoch": 8.391655801825294,
-      "grad_norm": 1.4279826879501343,
-      "learning_rate": 8.43356783919598e-05,
-      "loss": 5.5256,
-      "step": 16091
-    },
-    {
-      "epoch": 8.392177314211212,
-      "grad_norm": 1.3224105834960938,
-      "learning_rate": 8.433467336683417e-05,
-      "loss": 5.6988,
-      "step": 16092
-    },
-    {
-      "epoch": 8.392698826597131,
-      "grad_norm": 1.379210114479065,
-      "learning_rate": 8.433366834170855e-05,
-      "loss": 5.9081,
-      "step": 16093
-    },
-    {
-      "epoch": 8.39322033898305,
-      "grad_norm": 1.3657768964767456,
-      "learning_rate": 8.433266331658291e-05,
-      "loss": 5.2152,
-      "step": 16094
-    },
-    {
-      "epoch": 8.39374185136897,
-      "grad_norm": 1.5339962244033813,
-      "learning_rate": 8.433165829145729e-05,
-      "loss": 5.5086,
-      "step": 16095
-    },
-    {
-      "epoch": 8.39426336375489,
-      "grad_norm": 1.5190377235412598,
-      "learning_rate": 8.433065326633166e-05,
-      "loss": 5.6024,
-      "step": 16096
-    },
-    {
-      "epoch": 8.394784876140807,
-      "grad_norm": 1.3834750652313232,
-      "learning_rate": 8.432964824120603e-05,
-      "loss": 5.8988,
-      "step": 16097
-    },
-    {
-      "epoch": 8.395306388526727,
-      "grad_norm": 1.5147089958190918,
-      "learning_rate": 8.432864321608041e-05,
-      "loss": 5.4647,
-      "step": 16098
-    },
-    {
-      "epoch": 8.395827900912646,
-      "grad_norm": 1.4412109851837158,
-      "learning_rate": 8.432763819095479e-05,
-      "loss": 5.4041,
-      "step": 16099
-    },
-    {
-      "epoch": 8.396349413298566,
-      "grad_norm": 1.5783789157867432,
-      "learning_rate": 8.432663316582915e-05,
-      "loss": 4.9433,
-      "step": 16100
-    },
-    {
-      "epoch": 8.396870925684485,
-      "grad_norm": 1.4107398986816406,
-      "learning_rate": 8.432562814070352e-05,
-      "loss": 5.6723,
-      "step": 16101
-    },
-    {
-      "epoch": 8.397392438070405,
-      "grad_norm": 1.4339709281921387,
-      "learning_rate": 8.43246231155779e-05,
-      "loss": 5.1507,
-      "step": 16102
-    },
-    {
-      "epoch": 8.397913950456322,
-      "grad_norm": 1.4619797468185425,
-      "learning_rate": 8.432361809045226e-05,
-      "loss": 5.2728,
-      "step": 16103
-    },
-    {
-      "epoch": 8.398435462842242,
-      "grad_norm": 1.5045579671859741,
-      "learning_rate": 8.432261306532664e-05,
-      "loss": 5.6034,
-      "step": 16104
-    },
-    {
-      "epoch": 8.398956975228161,
-      "grad_norm": 1.406790852546692,
-      "learning_rate": 8.4321608040201e-05,
-      "loss": 5.1439,
-      "step": 16105
-    },
-    {
-      "epoch": 8.399478487614081,
-      "grad_norm": 1.4507883787155151,
-      "learning_rate": 8.432060301507538e-05,
-      "loss": 5.7224,
-      "step": 16106
-    },
-    {
-      "epoch": 8.4,
-      "grad_norm": 1.595935344696045,
-      "learning_rate": 8.431959798994974e-05,
-      "loss": 5.1339,
-      "step": 16107
-    },
-    {
-      "epoch": 8.40052151238592,
-      "grad_norm": 1.4726231098175049,
-      "learning_rate": 8.431859296482412e-05,
-      "loss": 5.3194,
-      "step": 16108
-    },
-    {
-      "epoch": 8.401043024771838,
-      "grad_norm": 1.4438586235046387,
-      "learning_rate": 8.43175879396985e-05,
-      "loss": 5.6857,
-      "step": 16109
-    },
-    {
-      "epoch": 8.401564537157757,
-      "grad_norm": 1.3812757730484009,
-      "learning_rate": 8.431658291457288e-05,
-      "loss": 5.5029,
-      "step": 16110
-    },
-    {
-      "epoch": 8.402086049543676,
-      "grad_norm": 1.4957808256149292,
-      "learning_rate": 8.431557788944724e-05,
-      "loss": 4.9263,
-      "step": 16111
-    },
-    {
-      "epoch": 8.402607561929596,
-      "grad_norm": 1.3335469961166382,
-      "learning_rate": 8.431457286432162e-05,
-      "loss": 5.6713,
-      "step": 16112
-    },
-    {
-      "epoch": 8.403129074315515,
-      "grad_norm": 1.3659363985061646,
-      "learning_rate": 8.431356783919598e-05,
-      "loss": 5.4633,
-      "step": 16113
-    },
-    {
-      "epoch": 8.403650586701435,
-      "grad_norm": 1.3799513578414917,
-      "learning_rate": 8.431256281407035e-05,
-      "loss": 4.9012,
-      "step": 16114
-    },
-    {
-      "epoch": 8.404172099087353,
-      "grad_norm": 1.5219062566757202,
-      "learning_rate": 8.431155778894473e-05,
-      "loss": 5.3564,
-      "step": 16115
-    },
-    {
-      "epoch": 8.404693611473272,
-      "grad_norm": 1.5627562999725342,
-      "learning_rate": 8.431055276381909e-05,
-      "loss": 5.3155,
-      "step": 16116
-    },
-    {
-      "epoch": 8.405215123859191,
-      "grad_norm": 1.5942436456680298,
-      "learning_rate": 8.430954773869347e-05,
-      "loss": 5.1577,
-      "step": 16117
-    },
-    {
-      "epoch": 8.405736636245111,
-      "grad_norm": 1.4543038606643677,
-      "learning_rate": 8.430854271356785e-05,
-      "loss": 5.566,
-      "step": 16118
-    },
-    {
-      "epoch": 8.40625814863103,
-      "grad_norm": 1.3444411754608154,
-      "learning_rate": 8.430753768844222e-05,
-      "loss": 5.7669,
-      "step": 16119
-    },
-    {
-      "epoch": 8.40677966101695,
-      "grad_norm": 1.3484081029891968,
-      "learning_rate": 8.430653266331659e-05,
-      "loss": 5.895,
-      "step": 16120
-    },
-    {
-      "epoch": 8.407301173402868,
-      "grad_norm": 1.4837757349014282,
-      "learning_rate": 8.430552763819097e-05,
-      "loss": 5.7888,
-      "step": 16121
-    },
-    {
-      "epoch": 8.407822685788787,
-      "grad_norm": 1.4480828046798706,
-      "learning_rate": 8.430452261306533e-05,
-      "loss": 5.5082,
-      "step": 16122
-    },
-    {
-      "epoch": 8.408344198174706,
-      "grad_norm": 1.4532753229141235,
-      "learning_rate": 8.430351758793971e-05,
-      "loss": 5.7895,
-      "step": 16123
-    },
-    {
-      "epoch": 8.408865710560626,
-      "grad_norm": 1.4868870973587036,
-      "learning_rate": 8.430251256281407e-05,
-      "loss": 5.4205,
-      "step": 16124
-    },
-    {
-      "epoch": 8.409387222946545,
-      "grad_norm": 1.5160789489746094,
-      "learning_rate": 8.430150753768845e-05,
-      "loss": 5.5973,
-      "step": 16125
-    },
-    {
-      "epoch": 8.409908735332465,
-      "grad_norm": 1.4569697380065918,
-      "learning_rate": 8.430050251256281e-05,
-      "loss": 5.3046,
-      "step": 16126
-    },
-    {
-      "epoch": 8.410430247718383,
-      "grad_norm": 1.642928957939148,
-      "learning_rate": 8.429949748743718e-05,
-      "loss": 5.179,
-      "step": 16127
-    },
-    {
-      "epoch": 8.410951760104302,
-      "grad_norm": 1.5835330486297607,
-      "learning_rate": 8.429849246231156e-05,
-      "loss": 5.3991,
-      "step": 16128
-    },
-    {
-      "epoch": 8.411473272490221,
-      "grad_norm": 1.4960864782333374,
-      "learning_rate": 8.429748743718593e-05,
-      "loss": 5.5054,
-      "step": 16129
-    },
-    {
-      "epoch": 8.411994784876141,
-      "grad_norm": 1.587125539779663,
-      "learning_rate": 8.429648241206031e-05,
-      "loss": 5.3242,
-      "step": 16130
-    },
-    {
-      "epoch": 8.41251629726206,
-      "grad_norm": 1.3519264459609985,
-      "learning_rate": 8.429547738693468e-05,
-      "loss": 5.7859,
-      "step": 16131
-    },
-    {
-      "epoch": 8.41303780964798,
-      "grad_norm": 1.5771528482437134,
-      "learning_rate": 8.429447236180905e-05,
-      "loss": 4.9605,
-      "step": 16132
-    },
-    {
-      "epoch": 8.413559322033898,
-      "grad_norm": 1.4783871173858643,
-      "learning_rate": 8.429346733668342e-05,
-      "loss": 5.6059,
-      "step": 16133
-    },
-    {
-      "epoch": 8.414080834419817,
-      "grad_norm": 1.2771652936935425,
-      "learning_rate": 8.42924623115578e-05,
-      "loss": 5.7807,
-      "step": 16134
-    },
-    {
-      "epoch": 8.414602346805736,
-      "grad_norm": 1.3679823875427246,
-      "learning_rate": 8.429145728643216e-05,
-      "loss": 5.7281,
-      "step": 16135
-    },
-    {
-      "epoch": 8.415123859191656,
-      "grad_norm": 1.4651376008987427,
-      "learning_rate": 8.429045226130654e-05,
-      "loss": 5.8054,
-      "step": 16136
-    },
-    {
-      "epoch": 8.415645371577575,
-      "grad_norm": 1.471177339553833,
-      "learning_rate": 8.42894472361809e-05,
-      "loss": 5.4151,
-      "step": 16137
-    },
-    {
-      "epoch": 8.416166883963495,
-      "grad_norm": 1.5430511236190796,
-      "learning_rate": 8.428844221105528e-05,
-      "loss": 4.9125,
-      "step": 16138
-    },
-    {
-      "epoch": 8.416688396349413,
-      "grad_norm": 1.4281829595565796,
-      "learning_rate": 8.428743718592966e-05,
-      "loss": 5.6038,
-      "step": 16139
-    },
-    {
-      "epoch": 8.417209908735332,
-      "grad_norm": 1.3441716432571411,
-      "learning_rate": 8.428643216080402e-05,
-      "loss": 5.7535,
-      "step": 16140
-    },
-    {
-      "epoch": 8.417731421121252,
-      "grad_norm": 1.3890423774719238,
-      "learning_rate": 8.42854271356784e-05,
-      "loss": 5.8084,
-      "step": 16141
-    },
-    {
-      "epoch": 8.418252933507171,
-      "grad_norm": 1.4765045642852783,
-      "learning_rate": 8.428442211055276e-05,
-      "loss": 5.2657,
-      "step": 16142
-    },
-    {
-      "epoch": 8.41877444589309,
-      "grad_norm": 1.5323928594589233,
-      "learning_rate": 8.428341708542714e-05,
-      "loss": 5.7274,
-      "step": 16143
-    },
-    {
-      "epoch": 8.41929595827901,
-      "grad_norm": 1.3907129764556885,
-      "learning_rate": 8.42824120603015e-05,
-      "loss": 5.4728,
-      "step": 16144
-    },
-    {
-      "epoch": 8.419817470664928,
-      "grad_norm": 1.3522963523864746,
-      "learning_rate": 8.428140703517588e-05,
-      "loss": 5.2692,
-      "step": 16145
-    },
-    {
-      "epoch": 8.420338983050847,
-      "grad_norm": 1.3624845743179321,
-      "learning_rate": 8.428040201005025e-05,
-      "loss": 5.5769,
-      "step": 16146
-    },
-    {
-      "epoch": 8.420860495436767,
-      "grad_norm": 1.4033408164978027,
-      "learning_rate": 8.427939698492463e-05,
-      "loss": 5.7202,
-      "step": 16147
-    },
-    {
-      "epoch": 8.421382007822686,
-      "grad_norm": 1.36985182762146,
-      "learning_rate": 8.427839195979899e-05,
-      "loss": 5.7425,
-      "step": 16148
-    },
-    {
-      "epoch": 8.421903520208605,
-      "grad_norm": 1.5910874605178833,
-      "learning_rate": 8.427738693467337e-05,
-      "loss": 5.5441,
-      "step": 16149
-    },
-    {
-      "epoch": 8.422425032594525,
-      "grad_norm": 1.4702595472335815,
-      "learning_rate": 8.427638190954775e-05,
-      "loss": 5.6634,
-      "step": 16150
-    },
-    {
-      "epoch": 8.422946544980443,
-      "grad_norm": 1.5648807287216187,
-      "learning_rate": 8.427537688442212e-05,
-      "loss": 5.3563,
-      "step": 16151
-    },
-    {
-      "epoch": 8.423468057366362,
-      "grad_norm": 1.4909021854400635,
-      "learning_rate": 8.427437185929649e-05,
-      "loss": 5.4283,
-      "step": 16152
-    },
-    {
-      "epoch": 8.423989569752282,
-      "grad_norm": 1.4872814416885376,
-      "learning_rate": 8.427336683417087e-05,
-      "loss": 5.2605,
-      "step": 16153
-    },
-    {
-      "epoch": 8.424511082138201,
-      "grad_norm": 1.478987455368042,
-      "learning_rate": 8.427236180904523e-05,
-      "loss": 5.3484,
-      "step": 16154
-    },
-    {
-      "epoch": 8.42503259452412,
-      "grad_norm": 1.35179603099823,
-      "learning_rate": 8.42713567839196e-05,
-      "loss": 5.4565,
-      "step": 16155
-    },
-    {
-      "epoch": 8.42555410691004,
-      "grad_norm": 1.5015243291854858,
-      "learning_rate": 8.427035175879397e-05,
-      "loss": 5.4256,
-      "step": 16156
-    },
-    {
-      "epoch": 8.426075619295958,
-      "grad_norm": 1.5357385873794556,
-      "learning_rate": 8.426934673366834e-05,
-      "loss": 4.8303,
-      "step": 16157
-    },
-    {
-      "epoch": 8.426597131681877,
-      "grad_norm": 1.368300199508667,
-      "learning_rate": 8.426834170854271e-05,
-      "loss": 5.5543,
-      "step": 16158
-    },
-    {
-      "epoch": 8.427118644067797,
-      "grad_norm": 1.3007712364196777,
-      "learning_rate": 8.426733668341709e-05,
-      "loss": 5.9314,
-      "step": 16159
-    },
-    {
-      "epoch": 8.427640156453716,
-      "grad_norm": 1.4460233449935913,
-      "learning_rate": 8.426633165829147e-05,
-      "loss": 5.7066,
-      "step": 16160
-    },
-    {
-      "epoch": 8.428161668839635,
-      "grad_norm": 1.3938862085342407,
-      "learning_rate": 8.426532663316583e-05,
-      "loss": 5.7754,
-      "step": 16161
-    },
-    {
-      "epoch": 8.428683181225555,
-      "grad_norm": 1.7233937978744507,
-      "learning_rate": 8.426432160804021e-05,
-      "loss": 5.1365,
-      "step": 16162
-    },
-    {
-      "epoch": 8.429204693611473,
-      "grad_norm": 1.2903954982757568,
-      "learning_rate": 8.426331658291458e-05,
-      "loss": 6.0814,
-      "step": 16163
-    },
-    {
-      "epoch": 8.429726205997392,
-      "grad_norm": 1.4944130182266235,
-      "learning_rate": 8.426231155778895e-05,
-      "loss": 5.193,
-      "step": 16164
-    },
-    {
-      "epoch": 8.430247718383312,
-      "grad_norm": 1.4338456392288208,
-      "learning_rate": 8.426130653266332e-05,
-      "loss": 5.0643,
-      "step": 16165
-    },
-    {
-      "epoch": 8.430769230769231,
-      "grad_norm": 1.4309262037277222,
-      "learning_rate": 8.42603015075377e-05,
-      "loss": 4.9456,
-      "step": 16166
-    },
-    {
-      "epoch": 8.43129074315515,
-      "grad_norm": 1.6960926055908203,
-      "learning_rate": 8.425929648241206e-05,
-      "loss": 5.3937,
-      "step": 16167
-    },
-    {
-      "epoch": 8.43181225554107,
-      "grad_norm": 1.5381755828857422,
-      "learning_rate": 8.425829145728643e-05,
-      "loss": 5.6298,
-      "step": 16168
-    },
-    {
-      "epoch": 8.432333767926988,
-      "grad_norm": 1.3458833694458008,
-      "learning_rate": 8.42572864321608e-05,
-      "loss": 5.3015,
-      "step": 16169
-    },
-    {
-      "epoch": 8.432855280312907,
-      "grad_norm": 1.5250509977340698,
-      "learning_rate": 8.425628140703518e-05,
-      "loss": 5.1764,
-      "step": 16170
-    },
-    {
-      "epoch": 8.433376792698827,
-      "grad_norm": 1.4611537456512451,
-      "learning_rate": 8.425527638190956e-05,
-      "loss": 5.1801,
-      "step": 16171
-    },
-    {
-      "epoch": 8.433898305084746,
-      "grad_norm": 1.4016579389572144,
-      "learning_rate": 8.425427135678392e-05,
-      "loss": 5.7022,
-      "step": 16172
-    },
-    {
-      "epoch": 8.434419817470665,
-      "grad_norm": 1.4487282037734985,
-      "learning_rate": 8.42532663316583e-05,
-      "loss": 5.3909,
-      "step": 16173
-    },
-    {
-      "epoch": 8.434941329856585,
-      "grad_norm": 1.4919952154159546,
-      "learning_rate": 8.425226130653266e-05,
-      "loss": 5.0875,
-      "step": 16174
-    },
-    {
-      "epoch": 8.435462842242503,
-      "grad_norm": 1.4578818082809448,
-      "learning_rate": 8.425125628140704e-05,
-      "loss": 5.5867,
-      "step": 16175
-    },
-    {
-      "epoch": 8.435984354628422,
-      "grad_norm": 1.4479092359542847,
-      "learning_rate": 8.425025125628141e-05,
-      "loss": 5.0202,
-      "step": 16176
-    },
-    {
-      "epoch": 8.436505867014342,
-      "grad_norm": 1.6090083122253418,
-      "learning_rate": 8.424924623115578e-05,
-      "loss": 5.3188,
-      "step": 16177
-    },
-    {
-      "epoch": 8.437027379400261,
-      "grad_norm": 1.4183322191238403,
-      "learning_rate": 8.424824120603015e-05,
-      "loss": 5.5488,
-      "step": 16178
-    },
-    {
-      "epoch": 8.43754889178618,
-      "grad_norm": 1.3837671279907227,
-      "learning_rate": 8.424723618090453e-05,
-      "loss": 5.3759,
-      "step": 16179
-    },
-    {
-      "epoch": 8.438070404172098,
-      "grad_norm": 1.4647283554077148,
-      "learning_rate": 8.42462311557789e-05,
-      "loss": 5.5902,
-      "step": 16180
-    },
-    {
-      "epoch": 8.438591916558018,
-      "grad_norm": 1.438813328742981,
-      "learning_rate": 8.424522613065327e-05,
-      "loss": 5.696,
-      "step": 16181
-    },
-    {
-      "epoch": 8.439113428943937,
-      "grad_norm": 1.6649941205978394,
-      "learning_rate": 8.424422110552765e-05,
-      "loss": 5.5085,
-      "step": 16182
-    },
-    {
-      "epoch": 8.439634941329857,
-      "grad_norm": 1.462525486946106,
-      "learning_rate": 8.424321608040201e-05,
-      "loss": 5.1229,
-      "step": 16183
-    },
-    {
-      "epoch": 8.440156453715776,
-      "grad_norm": 1.3341935873031616,
-      "learning_rate": 8.424221105527639e-05,
-      "loss": 5.9746,
-      "step": 16184
-    },
-    {
-      "epoch": 8.440677966101696,
-      "grad_norm": 1.5025049448013306,
-      "learning_rate": 8.424120603015075e-05,
-      "loss": 5.0136,
-      "step": 16185
-    },
-    {
-      "epoch": 8.441199478487615,
-      "grad_norm": 1.4485257863998413,
-      "learning_rate": 8.424020100502513e-05,
-      "loss": 5.3904,
-      "step": 16186
-    },
-    {
-      "epoch": 8.441720990873533,
-      "grad_norm": 1.3277555704116821,
-      "learning_rate": 8.42391959798995e-05,
-      "loss": 5.6961,
-      "step": 16187
-    },
-    {
-      "epoch": 8.442242503259452,
-      "grad_norm": 1.4968492984771729,
-      "learning_rate": 8.423819095477387e-05,
-      "loss": 5.1768,
-      "step": 16188
-    },
-    {
-      "epoch": 8.442764015645372,
-      "grad_norm": 1.546457290649414,
-      "learning_rate": 8.423718592964824e-05,
-      "loss": 5.2617,
-      "step": 16189
-    },
-    {
-      "epoch": 8.443285528031291,
-      "grad_norm": 1.3422008752822876,
-      "learning_rate": 8.423618090452262e-05,
-      "loss": 5.0841,
-      "step": 16190
-    },
-    {
-      "epoch": 8.44380704041721,
-      "grad_norm": 1.4390556812286377,
-      "learning_rate": 8.423517587939699e-05,
-      "loss": 5.5186,
-      "step": 16191
-    },
-    {
-      "epoch": 8.444328552803128,
-      "grad_norm": 1.4618686437606812,
-      "learning_rate": 8.423417085427137e-05,
-      "loss": 5.8814,
-      "step": 16192
-    },
-    {
-      "epoch": 8.444850065189048,
-      "grad_norm": 1.3392894268035889,
-      "learning_rate": 8.423316582914574e-05,
-      "loss": 5.7298,
-      "step": 16193
-    },
-    {
-      "epoch": 8.445371577574967,
-      "grad_norm": 1.4093984365463257,
-      "learning_rate": 8.42321608040201e-05,
-      "loss": 5.082,
-      "step": 16194
-    },
-    {
-      "epoch": 8.445893089960887,
-      "grad_norm": 1.8295124769210815,
-      "learning_rate": 8.423115577889448e-05,
-      "loss": 5.1186,
-      "step": 16195
-    },
-    {
-      "epoch": 8.446414602346806,
-      "grad_norm": 1.4293227195739746,
-      "learning_rate": 8.423015075376884e-05,
-      "loss": 5.4479,
-      "step": 16196
-    },
-    {
-      "epoch": 8.446936114732726,
-      "grad_norm": 1.389785885810852,
-      "learning_rate": 8.422914572864322e-05,
-      "loss": 4.9152,
-      "step": 16197
-    },
-    {
-      "epoch": 8.447457627118643,
-      "grad_norm": 1.395507574081421,
-      "learning_rate": 8.422814070351758e-05,
-      "loss": 5.639,
-      "step": 16198
-    },
-    {
-      "epoch": 8.447979139504563,
-      "grad_norm": 1.279334306716919,
-      "learning_rate": 8.422713567839196e-05,
-      "loss": 5.9153,
-      "step": 16199
-    },
-    {
-      "epoch": 8.448500651890482,
-      "grad_norm": 1.463991641998291,
-      "learning_rate": 8.422613065326634e-05,
-      "loss": 5.404,
-      "step": 16200
-    },
-    {
-      "epoch": 8.449022164276402,
-      "grad_norm": 1.4850136041641235,
-      "learning_rate": 8.422512562814072e-05,
-      "loss": 5.6013,
-      "step": 16201
-    },
-    {
-      "epoch": 8.449543676662321,
-      "grad_norm": 1.3926880359649658,
-      "learning_rate": 8.422412060301508e-05,
-      "loss": 5.2944,
-      "step": 16202
-    },
-    {
-      "epoch": 8.45006518904824,
-      "grad_norm": 1.3785253763198853,
-      "learning_rate": 8.422311557788946e-05,
-      "loss": 5.6757,
-      "step": 16203
-    },
-    {
-      "epoch": 8.450586701434158,
-      "grad_norm": 1.3904322385787964,
-      "learning_rate": 8.422211055276382e-05,
-      "loss": 5.4191,
-      "step": 16204
-    },
-    {
-      "epoch": 8.451108213820078,
-      "grad_norm": 1.4383984804153442,
-      "learning_rate": 8.42211055276382e-05,
-      "loss": 5.5075,
-      "step": 16205
-    },
-    {
-      "epoch": 8.451629726205997,
-      "grad_norm": 1.3795032501220703,
-      "learning_rate": 8.422010050251257e-05,
-      "loss": 5.379,
-      "step": 16206
-    },
-    {
-      "epoch": 8.452151238591917,
-      "grad_norm": 1.3220949172973633,
-      "learning_rate": 8.421909547738693e-05,
-      "loss": 5.7296,
-      "step": 16207
-    },
-    {
-      "epoch": 8.452672750977836,
-      "grad_norm": 1.5404318571090698,
-      "learning_rate": 8.421809045226131e-05,
-      "loss": 5.5361,
-      "step": 16208
-    },
-    {
-      "epoch": 8.453194263363756,
-      "grad_norm": 1.310841679573059,
-      "learning_rate": 8.421708542713567e-05,
-      "loss": 5.8339,
-      "step": 16209
-    },
-    {
-      "epoch": 8.453715775749673,
-      "grad_norm": 1.4948856830596924,
-      "learning_rate": 8.421608040201005e-05,
-      "loss": 5.2088,
-      "step": 16210
-    },
-    {
-      "epoch": 8.454237288135593,
-      "grad_norm": 1.5076905488967896,
-      "learning_rate": 8.421507537688443e-05,
-      "loss": 5.2338,
-      "step": 16211
-    },
-    {
-      "epoch": 8.454758800521512,
-      "grad_norm": 1.3501328229904175,
-      "learning_rate": 8.42140703517588e-05,
-      "loss": 5.9439,
-      "step": 16212
-    },
-    {
-      "epoch": 8.455280312907432,
-      "grad_norm": 1.415837049484253,
-      "learning_rate": 8.421306532663317e-05,
-      "loss": 5.5858,
-      "step": 16213
-    },
-    {
-      "epoch": 8.455801825293351,
-      "grad_norm": 1.4384658336639404,
-      "learning_rate": 8.421206030150755e-05,
-      "loss": 5.6015,
-      "step": 16214
-    },
-    {
-      "epoch": 8.45632333767927,
-      "grad_norm": 1.4486894607543945,
-      "learning_rate": 8.421105527638191e-05,
-      "loss": 5.6896,
-      "step": 16215
-    },
-    {
-      "epoch": 8.456844850065188,
-      "grad_norm": 1.3354312181472778,
-      "learning_rate": 8.421005025125629e-05,
-      "loss": 5.5815,
-      "step": 16216
-    },
-    {
-      "epoch": 8.457366362451108,
-      "grad_norm": 1.366239309310913,
-      "learning_rate": 8.420904522613065e-05,
-      "loss": 5.9688,
-      "step": 16217
-    },
-    {
-      "epoch": 8.457887874837027,
-      "grad_norm": 1.5704858303070068,
-      "learning_rate": 8.420804020100503e-05,
-      "loss": 4.7469,
-      "step": 16218
-    },
-    {
-      "epoch": 8.458409387222947,
-      "grad_norm": 1.3572789430618286,
-      "learning_rate": 8.42070351758794e-05,
-      "loss": 5.712,
-      "step": 16219
-    },
-    {
-      "epoch": 8.458930899608866,
-      "grad_norm": 1.339353322982788,
-      "learning_rate": 8.420603015075377e-05,
-      "loss": 5.3306,
-      "step": 16220
-    },
-    {
-      "epoch": 8.459452411994786,
-      "grad_norm": 1.3810932636260986,
-      "learning_rate": 8.420502512562815e-05,
-      "loss": 5.2048,
-      "step": 16221
-    },
-    {
-      "epoch": 8.459973924380703,
-      "grad_norm": 1.4041317701339722,
-      "learning_rate": 8.420402010050252e-05,
-      "loss": 5.6239,
-      "step": 16222
-    },
-    {
-      "epoch": 8.460495436766623,
-      "grad_norm": 1.4304488897323608,
-      "learning_rate": 8.42030150753769e-05,
-      "loss": 5.6196,
-      "step": 16223
-    },
-    {
-      "epoch": 8.461016949152542,
-      "grad_norm": 1.405139446258545,
-      "learning_rate": 8.420201005025126e-05,
-      "loss": 5.538,
-      "step": 16224
-    },
-    {
-      "epoch": 8.461538461538462,
-      "grad_norm": 1.3897625207901,
-      "learning_rate": 8.420100502512564e-05,
-      "loss": 5.7567,
-      "step": 16225
-    },
-    {
-      "epoch": 8.462059973924381,
-      "grad_norm": 1.551687479019165,
-      "learning_rate": 8.42e-05,
-      "loss": 5.621,
-      "step": 16226
-    },
-    {
-      "epoch": 8.4625814863103,
-      "grad_norm": 1.7834742069244385,
-      "learning_rate": 8.419899497487438e-05,
-      "loss": 4.929,
-      "step": 16227
-    },
-    {
-      "epoch": 8.463102998696218,
-      "grad_norm": 1.3782892227172852,
-      "learning_rate": 8.419798994974874e-05,
-      "loss": 5.2548,
-      "step": 16228
-    },
-    {
-      "epoch": 8.463624511082138,
-      "grad_norm": 1.401948094367981,
-      "learning_rate": 8.419698492462312e-05,
-      "loss": 4.6291,
-      "step": 16229
-    },
-    {
-      "epoch": 8.464146023468057,
-      "grad_norm": 1.503957986831665,
-      "learning_rate": 8.419597989949748e-05,
-      "loss": 5.6403,
-      "step": 16230
-    },
-    {
-      "epoch": 8.464667535853977,
-      "grad_norm": 1.5762001276016235,
-      "learning_rate": 8.419497487437186e-05,
-      "loss": 5.4079,
-      "step": 16231
-    },
-    {
-      "epoch": 8.465189048239896,
-      "grad_norm": 1.459452509880066,
-      "learning_rate": 8.419396984924624e-05,
-      "loss": 5.5275,
-      "step": 16232
-    },
-    {
-      "epoch": 8.465710560625816,
-      "grad_norm": 1.5235280990600586,
-      "learning_rate": 8.419296482412062e-05,
-      "loss": 5.3013,
-      "step": 16233
-    },
-    {
-      "epoch": 8.466232073011733,
-      "grad_norm": 1.5874536037445068,
-      "learning_rate": 8.419195979899498e-05,
-      "loss": 4.9698,
-      "step": 16234
-    },
-    {
-      "epoch": 8.466753585397653,
-      "grad_norm": 1.3900612592697144,
-      "learning_rate": 8.419095477386935e-05,
-      "loss": 5.3742,
-      "step": 16235
-    },
-    {
-      "epoch": 8.467275097783572,
-      "grad_norm": 1.3569087982177734,
-      "learning_rate": 8.418994974874372e-05,
-      "loss": 5.1134,
-      "step": 16236
-    },
-    {
-      "epoch": 8.467796610169492,
-      "grad_norm": 1.4767959117889404,
-      "learning_rate": 8.418894472361809e-05,
-      "loss": 5.8616,
-      "step": 16237
-    },
-    {
-      "epoch": 8.468318122555411,
-      "grad_norm": 1.5518466234207153,
-      "learning_rate": 8.418793969849247e-05,
-      "loss": 5.3588,
-      "step": 16238
-    },
-    {
-      "epoch": 8.46883963494133,
-      "grad_norm": 1.4954818487167358,
-      "learning_rate": 8.418693467336683e-05,
-      "loss": 5.1225,
-      "step": 16239
-    },
-    {
-      "epoch": 8.469361147327248,
-      "grad_norm": 1.4030823707580566,
-      "learning_rate": 8.418592964824121e-05,
-      "loss": 5.2526,
-      "step": 16240
-    },
-    {
-      "epoch": 8.469882659713168,
-      "grad_norm": 1.41080904006958,
-      "learning_rate": 8.418492462311559e-05,
-      "loss": 5.6533,
-      "step": 16241
-    },
-    {
-      "epoch": 8.470404172099087,
-      "grad_norm": 1.4401237964630127,
-      "learning_rate": 8.418391959798996e-05,
-      "loss": 5.6491,
-      "step": 16242
-    },
-    {
-      "epoch": 8.470925684485007,
-      "grad_norm": 1.4135055541992188,
-      "learning_rate": 8.418291457286433e-05,
-      "loss": 5.2678,
-      "step": 16243
-    },
-    {
-      "epoch": 8.471447196870926,
-      "grad_norm": 1.4171780347824097,
-      "learning_rate": 8.41819095477387e-05,
-      "loss": 5.8043,
-      "step": 16244
-    },
-    {
-      "epoch": 8.471968709256846,
-      "grad_norm": 1.4009233713150024,
-      "learning_rate": 8.418090452261307e-05,
-      "loss": 5.7416,
-      "step": 16245
-    },
-    {
-      "epoch": 8.472490221642763,
-      "grad_norm": 1.4349644184112549,
-      "learning_rate": 8.417989949748745e-05,
-      "loss": 5.5646,
-      "step": 16246
-    },
-    {
-      "epoch": 8.473011734028683,
-      "grad_norm": 1.3982341289520264,
-      "learning_rate": 8.417889447236181e-05,
-      "loss": 5.4435,
-      "step": 16247
-    },
-    {
-      "epoch": 8.473533246414602,
-      "grad_norm": 1.3714706897735596,
-      "learning_rate": 8.417788944723618e-05,
-      "loss": 5.7795,
-      "step": 16248
-    },
-    {
-      "epoch": 8.474054758800522,
-      "grad_norm": 1.396523118019104,
-      "learning_rate": 8.417688442211055e-05,
-      "loss": 5.2386,
-      "step": 16249
-    },
-    {
-      "epoch": 8.474576271186441,
-      "grad_norm": 1.6258031129837036,
-      "learning_rate": 8.417587939698492e-05,
-      "loss": 5.3586,
-      "step": 16250
-    },
-    {
-      "epoch": 8.47509778357236,
-      "grad_norm": 1.5053828954696655,
-      "learning_rate": 8.41748743718593e-05,
-      "loss": 5.6597,
-      "step": 16251
-    },
-    {
-      "epoch": 8.475619295958278,
-      "grad_norm": 1.5445096492767334,
-      "learning_rate": 8.417386934673367e-05,
-      "loss": 5.8149,
-      "step": 16252
-    },
-    {
-      "epoch": 8.476140808344198,
-      "grad_norm": 1.4885343313217163,
-      "learning_rate": 8.417286432160805e-05,
-      "loss": 5.7406,
-      "step": 16253
-    },
-    {
-      "epoch": 8.476662320730117,
-      "grad_norm": 1.4981075525283813,
-      "learning_rate": 8.417185929648242e-05,
-      "loss": 5.4396,
-      "step": 16254
-    },
-    {
-      "epoch": 8.477183833116037,
-      "grad_norm": 1.3831754922866821,
-      "learning_rate": 8.41708542713568e-05,
-      "loss": 5.9081,
-      "step": 16255
-    },
-    {
-      "epoch": 8.477705345501956,
-      "grad_norm": 1.5725641250610352,
-      "learning_rate": 8.416984924623116e-05,
-      "loss": 5.1548,
-      "step": 16256
-    },
-    {
-      "epoch": 8.478226857887876,
-      "grad_norm": 1.40445077419281,
-      "learning_rate": 8.416884422110554e-05,
-      "loss": 5.2288,
-      "step": 16257
-    },
-    {
-      "epoch": 8.478748370273793,
-      "grad_norm": 1.4143438339233398,
-      "learning_rate": 8.41678391959799e-05,
-      "loss": 5.4474,
-      "step": 16258
-    },
-    {
-      "epoch": 8.479269882659713,
-      "grad_norm": 1.4662890434265137,
-      "learning_rate": 8.416683417085428e-05,
-      "loss": 5.7607,
-      "step": 16259
-    },
-    {
-      "epoch": 8.479791395045632,
-      "grad_norm": 1.3890568017959595,
-      "learning_rate": 8.416582914572864e-05,
-      "loss": 5.5942,
-      "step": 16260
-    },
-    {
-      "epoch": 8.480312907431552,
-      "grad_norm": 1.4208444356918335,
-      "learning_rate": 8.416482412060302e-05,
-      "loss": 5.4662,
-      "step": 16261
-    },
-    {
-      "epoch": 8.480834419817471,
-      "grad_norm": 1.682942509651184,
-      "learning_rate": 8.41638190954774e-05,
-      "loss": 5.2358,
-      "step": 16262
-    },
-    {
-      "epoch": 8.48135593220339,
-      "grad_norm": 1.386042594909668,
-      "learning_rate": 8.416281407035176e-05,
-      "loss": 5.2574,
-      "step": 16263
-    },
-    {
-      "epoch": 8.481877444589308,
-      "grad_norm": 1.3789124488830566,
-      "learning_rate": 8.416180904522614e-05,
-      "loss": 5.5634,
-      "step": 16264
-    },
-    {
-      "epoch": 8.482398956975228,
-      "grad_norm": 1.3166377544403076,
-      "learning_rate": 8.41608040201005e-05,
-      "loss": 5.8143,
-      "step": 16265
-    },
-    {
-      "epoch": 8.482920469361147,
-      "grad_norm": 1.4488633871078491,
-      "learning_rate": 8.415979899497488e-05,
-      "loss": 5.2373,
-      "step": 16266
-    },
-    {
-      "epoch": 8.483441981747067,
-      "grad_norm": 1.5035679340362549,
-      "learning_rate": 8.415879396984925e-05,
-      "loss": 5.2976,
-      "step": 16267
-    },
-    {
-      "epoch": 8.483963494132986,
-      "grad_norm": 1.4377559423446655,
-      "learning_rate": 8.415778894472363e-05,
-      "loss": 5.0286,
-      "step": 16268
-    },
-    {
-      "epoch": 8.484485006518906,
-      "grad_norm": 1.4902321100234985,
-      "learning_rate": 8.415678391959799e-05,
-      "loss": 5.8942,
-      "step": 16269
-    },
-    {
-      "epoch": 8.485006518904823,
-      "grad_norm": 1.3802274465560913,
-      "learning_rate": 8.415577889447237e-05,
-      "loss": 5.7178,
-      "step": 16270
-    },
-    {
-      "epoch": 8.485528031290743,
-      "grad_norm": 1.4650157690048218,
-      "learning_rate": 8.415477386934673e-05,
-      "loss": 5.6931,
-      "step": 16271
-    },
-    {
-      "epoch": 8.486049543676662,
-      "grad_norm": 1.3341090679168701,
-      "learning_rate": 8.415376884422111e-05,
-      "loss": 5.8694,
-      "step": 16272
-    },
-    {
-      "epoch": 8.486571056062582,
-      "grad_norm": 1.4870949983596802,
-      "learning_rate": 8.415276381909549e-05,
-      "loss": 5.702,
-      "step": 16273
-    },
-    {
-      "epoch": 8.487092568448501,
-      "grad_norm": 1.424792766571045,
-      "learning_rate": 8.415175879396985e-05,
-      "loss": 5.6508,
-      "step": 16274
-    },
-    {
-      "epoch": 8.487614080834419,
-      "grad_norm": 1.6119937896728516,
-      "learning_rate": 8.415075376884423e-05,
-      "loss": 5.2143,
-      "step": 16275
-    },
-    {
-      "epoch": 8.488135593220338,
-      "grad_norm": 1.3839800357818604,
-      "learning_rate": 8.41497487437186e-05,
-      "loss": 5.6363,
-      "step": 16276
-    },
-    {
-      "epoch": 8.488657105606258,
-      "grad_norm": 1.3895314931869507,
-      "learning_rate": 8.414874371859297e-05,
-      "loss": 5.5482,
-      "step": 16277
-    },
-    {
-      "epoch": 8.489178617992177,
-      "grad_norm": 1.4103795289993286,
-      "learning_rate": 8.414773869346734e-05,
-      "loss": 5.7319,
-      "step": 16278
-    },
-    {
-      "epoch": 8.489700130378097,
-      "grad_norm": 1.6477429866790771,
-      "learning_rate": 8.414673366834171e-05,
-      "loss": 5.0811,
-      "step": 16279
-    },
-    {
-      "epoch": 8.490221642764016,
-      "grad_norm": 1.4250315427780151,
-      "learning_rate": 8.414572864321608e-05,
-      "loss": 5.6652,
-      "step": 16280
-    },
-    {
-      "epoch": 8.490743155149936,
-      "grad_norm": 1.3829952478408813,
-      "learning_rate": 8.414472361809046e-05,
-      "loss": 5.5521,
-      "step": 16281
-    },
-    {
-      "epoch": 8.491264667535853,
-      "grad_norm": 1.5686968564987183,
-      "learning_rate": 8.414371859296482e-05,
-      "loss": 5.7742,
-      "step": 16282
-    },
-    {
-      "epoch": 8.491786179921773,
-      "grad_norm": 1.3568753004074097,
-      "learning_rate": 8.41427135678392e-05,
-      "loss": 5.5451,
-      "step": 16283
-    },
-    {
-      "epoch": 8.492307692307692,
-      "grad_norm": 1.4286757707595825,
-      "learning_rate": 8.414170854271358e-05,
-      "loss": 5.4459,
-      "step": 16284
-    },
-    {
-      "epoch": 8.492829204693612,
-      "grad_norm": 1.4185134172439575,
-      "learning_rate": 8.414070351758795e-05,
-      "loss": 5.771,
-      "step": 16285
-    },
-    {
-      "epoch": 8.493350717079531,
-      "grad_norm": 1.4366488456726074,
-      "learning_rate": 8.413969849246232e-05,
-      "loss": 5.5368,
-      "step": 16286
-    },
-    {
-      "epoch": 8.493872229465449,
-      "grad_norm": 1.4074499607086182,
-      "learning_rate": 8.413869346733668e-05,
-      "loss": 5.6121,
-      "step": 16287
-    },
-    {
-      "epoch": 8.494393741851368,
-      "grad_norm": 1.5088961124420166,
-      "learning_rate": 8.413768844221106e-05,
-      "loss": 5.3589,
-      "step": 16288
-    },
-    {
-      "epoch": 8.494915254237288,
-      "grad_norm": 1.461081624031067,
-      "learning_rate": 8.413668341708542e-05,
-      "loss": 4.7038,
-      "step": 16289
-    },
-    {
-      "epoch": 8.495436766623207,
-      "grad_norm": 1.372451901435852,
-      "learning_rate": 8.41356783919598e-05,
-      "loss": 5.9244,
-      "step": 16290
-    },
-    {
-      "epoch": 8.495958279009127,
-      "grad_norm": 1.7016204595565796,
-      "learning_rate": 8.413467336683417e-05,
-      "loss": 5.0494,
-      "step": 16291
-    },
-    {
-      "epoch": 8.496479791395046,
-      "grad_norm": 1.4344096183776855,
-      "learning_rate": 8.413366834170854e-05,
-      "loss": 5.7458,
-      "step": 16292
-    },
-    {
-      "epoch": 8.497001303780964,
-      "grad_norm": 1.51285982131958,
-      "learning_rate": 8.413266331658292e-05,
-      "loss": 5.2118,
-      "step": 16293
-    },
-    {
-      "epoch": 8.497522816166883,
-      "grad_norm": 1.4368088245391846,
-      "learning_rate": 8.41316582914573e-05,
-      "loss": 5.1611,
-      "step": 16294
-    },
-    {
-      "epoch": 8.498044328552803,
-      "grad_norm": 1.4745235443115234,
-      "learning_rate": 8.413065326633166e-05,
-      "loss": 5.7714,
-      "step": 16295
-    },
-    {
-      "epoch": 8.498565840938722,
-      "grad_norm": 1.392412543296814,
-      "learning_rate": 8.412964824120604e-05,
-      "loss": 5.7046,
-      "step": 16296
-    },
-    {
-      "epoch": 8.499087353324642,
-      "grad_norm": 1.2590153217315674,
-      "learning_rate": 8.41286432160804e-05,
-      "loss": 5.0089,
-      "step": 16297
-    },
-    {
-      "epoch": 8.499608865710561,
-      "grad_norm": 1.3884339332580566,
-      "learning_rate": 8.412763819095478e-05,
-      "loss": 5.0125,
-      "step": 16298
-    },
-    {
-      "epoch": 8.500130378096479,
-      "grad_norm": 1.4788516759872437,
-      "learning_rate": 8.412663316582915e-05,
-      "loss": 5.3835,
-      "step": 16299
-    },
-    {
-      "epoch": 8.500651890482398,
-      "grad_norm": 1.5755811929702759,
-      "learning_rate": 8.412562814070351e-05,
-      "loss": 5.1858,
-      "step": 16300
-    },
-    {
-      "epoch": 8.501173402868318,
-      "grad_norm": 1.4121792316436768,
-      "learning_rate": 8.412462311557789e-05,
-      "loss": 5.7234,
-      "step": 16301
-    },
-    {
-      "epoch": 8.501694915254237,
-      "grad_norm": 1.5003864765167236,
-      "learning_rate": 8.412361809045225e-05,
-      "loss": 5.6416,
-      "step": 16302
-    },
-    {
-      "epoch": 8.502216427640157,
-      "grad_norm": 1.3301002979278564,
-      "learning_rate": 8.412261306532663e-05,
-      "loss": 6.0105,
-      "step": 16303
-    },
-    {
-      "epoch": 8.502737940026076,
-      "grad_norm": 1.5484566688537598,
-      "learning_rate": 8.412160804020101e-05,
-      "loss": 5.2331,
-      "step": 16304
-    },
-    {
-      "epoch": 8.503259452411994,
-      "grad_norm": 1.3522982597351074,
-      "learning_rate": 8.412060301507539e-05,
-      "loss": 5.9683,
-      "step": 16305
-    },
-    {
-      "epoch": 8.503780964797913,
-      "grad_norm": 1.4320651292800903,
-      "learning_rate": 8.411959798994975e-05,
-      "loss": 5.6104,
-      "step": 16306
-    },
-    {
-      "epoch": 8.504302477183833,
-      "grad_norm": 1.5427515506744385,
-      "learning_rate": 8.411859296482413e-05,
-      "loss": 5.0654,
-      "step": 16307
-    },
-    {
-      "epoch": 8.504823989569752,
-      "grad_norm": 1.4770077466964722,
-      "learning_rate": 8.41175879396985e-05,
-      "loss": 5.6556,
-      "step": 16308
-    },
-    {
-      "epoch": 8.505345501955672,
-      "grad_norm": 1.4306730031967163,
-      "learning_rate": 8.411658291457287e-05,
-      "loss": 5.423,
-      "step": 16309
-    },
-    {
-      "epoch": 8.505867014341591,
-      "grad_norm": 1.4186207056045532,
-      "learning_rate": 8.411557788944724e-05,
-      "loss": 5.6707,
-      "step": 16310
-    },
-    {
-      "epoch": 8.506388526727509,
-      "grad_norm": 1.4631394147872925,
-      "learning_rate": 8.411457286432161e-05,
-      "loss": 5.4871,
-      "step": 16311
-    },
-    {
-      "epoch": 8.506910039113428,
-      "grad_norm": 1.4884411096572876,
-      "learning_rate": 8.411356783919598e-05,
-      "loss": 5.4567,
-      "step": 16312
-    },
-    {
-      "epoch": 8.507431551499348,
-      "grad_norm": 1.4493452310562134,
-      "learning_rate": 8.411256281407036e-05,
-      "loss": 5.5605,
-      "step": 16313
-    },
-    {
-      "epoch": 8.507953063885267,
-      "grad_norm": 1.3339916467666626,
-      "learning_rate": 8.411155778894473e-05,
-      "loss": 5.9729,
-      "step": 16314
-    },
-    {
-      "epoch": 8.508474576271187,
-      "grad_norm": 1.523606300354004,
-      "learning_rate": 8.41105527638191e-05,
-      "loss": 5.0904,
-      "step": 16315
-    },
-    {
-      "epoch": 8.508996088657106,
-      "grad_norm": 1.533424973487854,
-      "learning_rate": 8.410954773869348e-05,
-      "loss": 5.1596,
-      "step": 16316
-    },
-    {
-      "epoch": 8.509517601043024,
-      "grad_norm": 1.4198166131973267,
-      "learning_rate": 8.410854271356784e-05,
-      "loss": 5.532,
-      "step": 16317
-    },
-    {
-      "epoch": 8.510039113428943,
-      "grad_norm": 1.3499614000320435,
-      "learning_rate": 8.410753768844222e-05,
-      "loss": 5.5899,
-      "step": 16318
-    },
-    {
-      "epoch": 8.510560625814863,
-      "grad_norm": 1.3293159008026123,
-      "learning_rate": 8.410653266331658e-05,
-      "loss": 5.8062,
-      "step": 16319
-    },
-    {
-      "epoch": 8.511082138200782,
-      "grad_norm": 1.3016242980957031,
-      "learning_rate": 8.410552763819096e-05,
-      "loss": 5.8828,
-      "step": 16320
-    },
-    {
-      "epoch": 8.511603650586702,
-      "grad_norm": 1.4299463033676147,
-      "learning_rate": 8.410452261306532e-05,
-      "loss": 4.7974,
-      "step": 16321
-    },
-    {
-      "epoch": 8.512125162972621,
-      "grad_norm": 1.4880151748657227,
-      "learning_rate": 8.41035175879397e-05,
-      "loss": 5.4461,
-      "step": 16322
-    },
-    {
-      "epoch": 8.512646675358539,
-      "grad_norm": 1.4608279466629028,
-      "learning_rate": 8.410251256281407e-05,
-      "loss": 5.7073,
-      "step": 16323
-    },
-    {
-      "epoch": 8.513168187744458,
-      "grad_norm": 1.4951132535934448,
-      "learning_rate": 8.410150753768844e-05,
-      "loss": 5.318,
-      "step": 16324
-    },
-    {
-      "epoch": 8.513689700130378,
-      "grad_norm": 1.4994184970855713,
-      "learning_rate": 8.410050251256282e-05,
-      "loss": 5.2159,
-      "step": 16325
-    },
-    {
-      "epoch": 8.514211212516297,
-      "grad_norm": 1.4930249452590942,
-      "learning_rate": 8.40994974874372e-05,
-      "loss": 5.4586,
-      "step": 16326
-    },
-    {
-      "epoch": 8.514732724902217,
-      "grad_norm": 1.363211989402771,
-      "learning_rate": 8.409849246231156e-05,
-      "loss": 5.5099,
-      "step": 16327
-    },
-    {
-      "epoch": 8.515254237288136,
-      "grad_norm": 1.3574644327163696,
-      "learning_rate": 8.409748743718593e-05,
-      "loss": 5.8946,
-      "step": 16328
-    },
-    {
-      "epoch": 8.515775749674054,
-      "grad_norm": 1.417824625968933,
-      "learning_rate": 8.40964824120603e-05,
-      "loss": 5.7409,
-      "step": 16329
-    },
-    {
-      "epoch": 8.516297262059974,
-      "grad_norm": 1.4192311763763428,
-      "learning_rate": 8.409547738693467e-05,
-      "loss": 5.4264,
-      "step": 16330
-    },
-    {
-      "epoch": 8.516818774445893,
-      "grad_norm": 1.434242844581604,
-      "learning_rate": 8.409447236180905e-05,
-      "loss": 5.1939,
-      "step": 16331
-    },
-    {
-      "epoch": 8.517340286831812,
-      "grad_norm": 1.310712218284607,
-      "learning_rate": 8.409346733668341e-05,
-      "loss": 5.9505,
-      "step": 16332
-    },
-    {
-      "epoch": 8.517861799217732,
-      "grad_norm": 1.4450511932373047,
-      "learning_rate": 8.409246231155779e-05,
-      "loss": 5.2603,
-      "step": 16333
-    },
-    {
-      "epoch": 8.518383311603651,
-      "grad_norm": 1.393729329109192,
-      "learning_rate": 8.409145728643217e-05,
-      "loss": 5.5123,
-      "step": 16334
-    },
-    {
-      "epoch": 8.518904823989569,
-      "grad_norm": 1.3721262216567993,
-      "learning_rate": 8.409045226130655e-05,
-      "loss": 5.5054,
-      "step": 16335
-    },
-    {
-      "epoch": 8.519426336375489,
-      "grad_norm": 1.4702799320220947,
-      "learning_rate": 8.408944723618091e-05,
-      "loss": 5.5488,
-      "step": 16336
-    },
-    {
-      "epoch": 8.519947848761408,
-      "grad_norm": 1.3862028121948242,
-      "learning_rate": 8.408844221105529e-05,
-      "loss": 5.4412,
-      "step": 16337
-    },
-    {
-      "epoch": 8.520469361147327,
-      "grad_norm": 1.440938115119934,
-      "learning_rate": 8.408743718592965e-05,
-      "loss": 5.7279,
-      "step": 16338
-    },
-    {
-      "epoch": 8.520990873533247,
-      "grad_norm": 1.5683718919754028,
-      "learning_rate": 8.408643216080403e-05,
-      "loss": 5.4503,
-      "step": 16339
-    },
-    {
-      "epoch": 8.521512385919166,
-      "grad_norm": 1.348432183265686,
-      "learning_rate": 8.40854271356784e-05,
-      "loss": 5.7932,
-      "step": 16340
-    },
-    {
-      "epoch": 8.522033898305084,
-      "grad_norm": 1.519518494606018,
-      "learning_rate": 8.408442211055276e-05,
-      "loss": 4.9353,
-      "step": 16341
-    },
-    {
-      "epoch": 8.522555410691004,
-      "grad_norm": 1.4163525104522705,
-      "learning_rate": 8.408341708542714e-05,
-      "loss": 5.6686,
-      "step": 16342
-    },
-    {
-      "epoch": 8.523076923076923,
-      "grad_norm": 1.3802011013031006,
-      "learning_rate": 8.40824120603015e-05,
-      "loss": 5.9269,
-      "step": 16343
-    },
-    {
-      "epoch": 8.523598435462842,
-      "grad_norm": 1.5114336013793945,
-      "learning_rate": 8.408140703517588e-05,
-      "loss": 4.9831,
-      "step": 16344
-    },
-    {
-      "epoch": 8.524119947848762,
-      "grad_norm": 1.3513715267181396,
-      "learning_rate": 8.408040201005026e-05,
-      "loss": 5.5953,
-      "step": 16345
-    },
-    {
-      "epoch": 8.524641460234681,
-      "grad_norm": 1.4421672821044922,
-      "learning_rate": 8.407939698492463e-05,
-      "loss": 5.3267,
-      "step": 16346
-    },
-    {
-      "epoch": 8.525162972620599,
-      "grad_norm": 1.476525902748108,
-      "learning_rate": 8.4078391959799e-05,
-      "loss": 5.7631,
-      "step": 16347
-    },
-    {
-      "epoch": 8.525684485006519,
-      "grad_norm": 1.523127555847168,
-      "learning_rate": 8.407738693467338e-05,
-      "loss": 5.4638,
-      "step": 16348
-    },
-    {
-      "epoch": 8.526205997392438,
-      "grad_norm": 1.402724027633667,
-      "learning_rate": 8.407638190954774e-05,
-      "loss": 5.699,
-      "step": 16349
-    },
-    {
-      "epoch": 8.526727509778357,
-      "grad_norm": 1.3704191446304321,
-      "learning_rate": 8.407537688442212e-05,
-      "loss": 5.7752,
-      "step": 16350
-    },
-    {
-      "epoch": 8.527249022164277,
-      "grad_norm": 1.3813397884368896,
-      "learning_rate": 8.407437185929648e-05,
-      "loss": 5.7721,
-      "step": 16351
-    },
-    {
-      "epoch": 8.527770534550196,
-      "grad_norm": 1.4107345342636108,
-      "learning_rate": 8.407336683417086e-05,
-      "loss": 5.2107,
-      "step": 16352
-    },
-    {
-      "epoch": 8.528292046936114,
-      "grad_norm": 1.4279944896697998,
-      "learning_rate": 8.407236180904523e-05,
-      "loss": 5.4633,
-      "step": 16353
-    },
-    {
-      "epoch": 8.528813559322034,
-      "grad_norm": 1.469723105430603,
-      "learning_rate": 8.40713567839196e-05,
-      "loss": 5.5019,
-      "step": 16354
-    },
-    {
-      "epoch": 8.529335071707953,
-      "grad_norm": 1.328547477722168,
-      "learning_rate": 8.407035175879398e-05,
-      "loss": 5.6764,
-      "step": 16355
-    },
-    {
-      "epoch": 8.529856584093872,
-      "grad_norm": 1.3986341953277588,
-      "learning_rate": 8.406934673366835e-05,
-      "loss": 5.6489,
-      "step": 16356
-    },
-    {
-      "epoch": 8.530378096479792,
-      "grad_norm": 1.4318324327468872,
-      "learning_rate": 8.406834170854272e-05,
-      "loss": 4.8899,
-      "step": 16357
-    },
-    {
-      "epoch": 8.53089960886571,
-      "grad_norm": 1.440152883529663,
-      "learning_rate": 8.406733668341709e-05,
-      "loss": 5.478,
-      "step": 16358
-    },
-    {
-      "epoch": 8.531421121251629,
-      "grad_norm": 1.4767831563949585,
-      "learning_rate": 8.406633165829147e-05,
-      "loss": 5.5326,
-      "step": 16359
-    },
-    {
-      "epoch": 8.531942633637549,
-      "grad_norm": 1.2952995300292969,
-      "learning_rate": 8.406532663316583e-05,
-      "loss": 5.9321,
-      "step": 16360
-    },
-    {
-      "epoch": 8.532464146023468,
-      "grad_norm": 1.4433907270431519,
-      "learning_rate": 8.406432160804021e-05,
-      "loss": 5.2721,
-      "step": 16361
-    },
-    {
-      "epoch": 8.532985658409387,
-      "grad_norm": 1.444433569908142,
-      "learning_rate": 8.406331658291457e-05,
-      "loss": 5.6761,
-      "step": 16362
-    },
-    {
-      "epoch": 8.533507170795307,
-      "grad_norm": 1.4009400606155396,
-      "learning_rate": 8.406231155778895e-05,
-      "loss": 5.1576,
-      "step": 16363
-    },
-    {
-      "epoch": 8.534028683181226,
-      "grad_norm": 1.5646318197250366,
-      "learning_rate": 8.406130653266331e-05,
-      "loss": 5.7697,
-      "step": 16364
-    },
-    {
-      "epoch": 8.534550195567144,
-      "grad_norm": 1.6418472528457642,
-      "learning_rate": 8.406030150753769e-05,
-      "loss": 5.3607,
-      "step": 16365
-    },
-    {
-      "epoch": 8.535071707953064,
-      "grad_norm": 1.4822769165039062,
-      "learning_rate": 8.405929648241207e-05,
-      "loss": 5.6217,
-      "step": 16366
-    },
-    {
-      "epoch": 8.535593220338983,
-      "grad_norm": 1.5673905611038208,
-      "learning_rate": 8.405829145728643e-05,
-      "loss": 4.9887,
-      "step": 16367
-    },
-    {
-      "epoch": 8.536114732724903,
-      "grad_norm": 1.3535528182983398,
-      "learning_rate": 8.405728643216081e-05,
-      "loss": 5.4071,
-      "step": 16368
-    },
-    {
-      "epoch": 8.536636245110822,
-      "grad_norm": 1.4707698822021484,
-      "learning_rate": 8.405628140703518e-05,
-      "loss": 5.6665,
-      "step": 16369
-    },
-    {
-      "epoch": 8.53715775749674,
-      "grad_norm": 1.6002399921417236,
-      "learning_rate": 8.405527638190955e-05,
-      "loss": 5.7226,
-      "step": 16370
-    },
-    {
-      "epoch": 8.53767926988266,
-      "grad_norm": 1.4739612340927124,
-      "learning_rate": 8.405427135678392e-05,
-      "loss": 5.3693,
-      "step": 16371
-    },
-    {
-      "epoch": 8.538200782268579,
-      "grad_norm": 1.4879001379013062,
-      "learning_rate": 8.40532663316583e-05,
-      "loss": 5.4926,
-      "step": 16372
-    },
-    {
-      "epoch": 8.538722294654498,
-      "grad_norm": 1.5028101205825806,
-      "learning_rate": 8.405226130653266e-05,
-      "loss": 5.3624,
-      "step": 16373
-    },
-    {
-      "epoch": 8.539243807040418,
-      "grad_norm": 1.3963958024978638,
-      "learning_rate": 8.405125628140704e-05,
-      "loss": 5.7082,
-      "step": 16374
-    },
-    {
-      "epoch": 8.539765319426337,
-      "grad_norm": 1.4851833581924438,
-      "learning_rate": 8.405025125628142e-05,
-      "loss": 5.1404,
-      "step": 16375
-    },
-    {
-      "epoch": 8.540286831812256,
-      "grad_norm": 1.4073140621185303,
-      "learning_rate": 8.40492462311558e-05,
-      "loss": 5.7296,
-      "step": 16376
-    },
-    {
-      "epoch": 8.540808344198174,
-      "grad_norm": 1.3304623365402222,
-      "learning_rate": 8.404824120603016e-05,
-      "loss": 6.0101,
-      "step": 16377
-    },
-    {
-      "epoch": 8.541329856584094,
-      "grad_norm": 1.4013386964797974,
-      "learning_rate": 8.404723618090454e-05,
-      "loss": 5.6639,
-      "step": 16378
-    },
-    {
-      "epoch": 8.541851368970013,
-      "grad_norm": 1.3646821975708008,
-      "learning_rate": 8.40462311557789e-05,
-      "loss": 5.5512,
-      "step": 16379
-    },
-    {
-      "epoch": 8.542372881355933,
-      "grad_norm": 1.4587630033493042,
-      "learning_rate": 8.404522613065326e-05,
-      "loss": 5.4475,
-      "step": 16380
-    },
-    {
-      "epoch": 8.542894393741852,
-      "grad_norm": 1.4335111379623413,
-      "learning_rate": 8.404422110552764e-05,
-      "loss": 5.875,
-      "step": 16381
-    },
-    {
-      "epoch": 8.54341590612777,
-      "grad_norm": 1.3422613143920898,
-      "learning_rate": 8.4043216080402e-05,
-      "loss": 5.8181,
-      "step": 16382
-    },
-    {
-      "epoch": 8.54393741851369,
-      "grad_norm": 1.4235994815826416,
-      "learning_rate": 8.404221105527638e-05,
-      "loss": 5.6201,
-      "step": 16383
-    },
-    {
-      "epoch": 8.544458930899609,
-      "grad_norm": 1.339044213294983,
-      "learning_rate": 8.404120603015075e-05,
-      "loss": 5.6226,
-      "step": 16384
-    },
-    {
-      "epoch": 8.544980443285528,
-      "grad_norm": 1.3962174654006958,
-      "learning_rate": 8.404020100502513e-05,
-      "loss": 5.8274,
-      "step": 16385
-    },
-    {
-      "epoch": 8.545501955671448,
-      "grad_norm": 1.3991062641143799,
-      "learning_rate": 8.40391959798995e-05,
-      "loss": 5.8344,
-      "step": 16386
-    },
-    {
-      "epoch": 8.546023468057367,
-      "grad_norm": 1.4849735498428345,
-      "learning_rate": 8.403819095477388e-05,
-      "loss": 5.4119,
-      "step": 16387
-    },
-    {
-      "epoch": 8.546544980443285,
-      "grad_norm": 1.3055260181427002,
-      "learning_rate": 8.403718592964825e-05,
-      "loss": 5.7153,
-      "step": 16388
-    },
-    {
-      "epoch": 8.547066492829204,
-      "grad_norm": 1.406346082687378,
-      "learning_rate": 8.403618090452262e-05,
-      "loss": 5.215,
-      "step": 16389
-    },
-    {
-      "epoch": 8.547588005215124,
-      "grad_norm": 1.384652853012085,
-      "learning_rate": 8.403517587939699e-05,
-      "loss": 5.7013,
-      "step": 16390
-    },
-    {
-      "epoch": 8.548109517601043,
-      "grad_norm": 1.4237455129623413,
-      "learning_rate": 8.403417085427137e-05,
-      "loss": 5.6436,
-      "step": 16391
-    },
-    {
-      "epoch": 8.548631029986963,
-      "grad_norm": 1.447649359703064,
-      "learning_rate": 8.403316582914573e-05,
-      "loss": 4.9642,
-      "step": 16392
-    },
-    {
-      "epoch": 8.549152542372882,
-      "grad_norm": 1.4991704225540161,
-      "learning_rate": 8.40321608040201e-05,
-      "loss": 5.1046,
-      "step": 16393
-    },
-    {
-      "epoch": 8.5496740547588,
-      "grad_norm": 1.4957998991012573,
-      "learning_rate": 8.403115577889447e-05,
-      "loss": 5.1178,
-      "step": 16394
-    },
-    {
-      "epoch": 8.55019556714472,
-      "grad_norm": 1.3411298990249634,
-      "learning_rate": 8.403015075376885e-05,
-      "loss": 5.8324,
-      "step": 16395
-    },
-    {
-      "epoch": 8.550717079530639,
-      "grad_norm": 1.338019609451294,
-      "learning_rate": 8.402914572864323e-05,
-      "loss": 5.5814,
-      "step": 16396
-    },
-    {
-      "epoch": 8.551238591916558,
-      "grad_norm": 1.4990488290786743,
-      "learning_rate": 8.402814070351759e-05,
-      "loss": 5.1198,
-      "step": 16397
-    },
-    {
-      "epoch": 8.551760104302478,
-      "grad_norm": 1.44180428981781,
-      "learning_rate": 8.402713567839197e-05,
-      "loss": 5.8572,
-      "step": 16398
-    },
-    {
-      "epoch": 8.552281616688397,
-      "grad_norm": 1.4442390203475952,
-      "learning_rate": 8.402613065326633e-05,
-      "loss": 5.3527,
-      "step": 16399
-    },
-    {
-      "epoch": 8.552803129074315,
-      "grad_norm": 1.495782494544983,
-      "learning_rate": 8.402512562814071e-05,
-      "loss": 5.7837,
-      "step": 16400
-    },
-    {
-      "epoch": 8.553324641460234,
-      "grad_norm": 1.3417143821716309,
-      "learning_rate": 8.402412060301508e-05,
-      "loss": 5.8992,
-      "step": 16401
-    },
-    {
-      "epoch": 8.553846153846154,
-      "grad_norm": 1.4816838502883911,
-      "learning_rate": 8.402311557788945e-05,
-      "loss": 5.7034,
-      "step": 16402
-    },
-    {
-      "epoch": 8.554367666232073,
-      "grad_norm": 1.5080004930496216,
-      "learning_rate": 8.402211055276382e-05,
-      "loss": 5.4601,
-      "step": 16403
-    },
-    {
-      "epoch": 8.554889178617993,
-      "grad_norm": 1.4068087339401245,
-      "learning_rate": 8.40211055276382e-05,
-      "loss": 5.8374,
-      "step": 16404
-    },
-    {
-      "epoch": 8.555410691003912,
-      "grad_norm": 1.4894593954086304,
-      "learning_rate": 8.402010050251256e-05,
-      "loss": 5.6616,
-      "step": 16405
-    },
-    {
-      "epoch": 8.55593220338983,
-      "grad_norm": 1.3798271417617798,
-      "learning_rate": 8.401909547738694e-05,
-      "loss": 5.8787,
-      "step": 16406
-    },
-    {
-      "epoch": 8.55645371577575,
-      "grad_norm": 1.3827537298202515,
-      "learning_rate": 8.401809045226132e-05,
-      "loss": 5.6701,
-      "step": 16407
-    },
-    {
-      "epoch": 8.556975228161669,
-      "grad_norm": 1.3991239070892334,
-      "learning_rate": 8.401708542713568e-05,
-      "loss": 5.4336,
-      "step": 16408
-    },
-    {
-      "epoch": 8.557496740547588,
-      "grad_norm": 1.4078681468963623,
-      "learning_rate": 8.401608040201006e-05,
-      "loss": 5.8665,
-      "step": 16409
-    },
-    {
-      "epoch": 8.558018252933508,
-      "grad_norm": 1.5952829122543335,
-      "learning_rate": 8.401507537688442e-05,
-      "loss": 5.5405,
-      "step": 16410
-    },
-    {
-      "epoch": 8.558539765319427,
-      "grad_norm": 1.4282077550888062,
-      "learning_rate": 8.40140703517588e-05,
-      "loss": 5.9755,
-      "step": 16411
-    },
-    {
-      "epoch": 8.559061277705345,
-      "grad_norm": 1.3922971487045288,
-      "learning_rate": 8.401306532663316e-05,
-      "loss": 5.4017,
-      "step": 16412
-    },
-    {
-      "epoch": 8.559582790091264,
-      "grad_norm": 1.5971148014068604,
-      "learning_rate": 8.401206030150754e-05,
-      "loss": 4.9622,
-      "step": 16413
-    },
-    {
-      "epoch": 8.560104302477184,
-      "grad_norm": 1.4166117906570435,
-      "learning_rate": 8.401105527638191e-05,
-      "loss": 5.2486,
-      "step": 16414
-    },
-    {
-      "epoch": 8.560625814863103,
-      "grad_norm": 1.4766515493392944,
-      "learning_rate": 8.401005025125628e-05,
-      "loss": 5.443,
-      "step": 16415
-    },
-    {
-      "epoch": 8.561147327249023,
-      "grad_norm": 1.3745697736740112,
-      "learning_rate": 8.400904522613066e-05,
-      "loss": 4.9973,
-      "step": 16416
-    },
-    {
-      "epoch": 8.561668839634942,
-      "grad_norm": 1.4124486446380615,
-      "learning_rate": 8.400804020100504e-05,
-      "loss": 5.4626,
-      "step": 16417
-    },
-    {
-      "epoch": 8.56219035202086,
-      "grad_norm": 1.377261996269226,
-      "learning_rate": 8.40070351758794e-05,
-      "loss": 5.6449,
-      "step": 16418
-    },
-    {
-      "epoch": 8.56271186440678,
-      "grad_norm": 1.4113552570343018,
-      "learning_rate": 8.400603015075378e-05,
-      "loss": 5.5016,
-      "step": 16419
-    },
-    {
-      "epoch": 8.563233376792699,
-      "grad_norm": 1.4962824583053589,
-      "learning_rate": 8.400502512562815e-05,
-      "loss": 5.4951,
-      "step": 16420
-    },
-    {
-      "epoch": 8.563754889178618,
-      "grad_norm": 1.4838231801986694,
-      "learning_rate": 8.400402010050251e-05,
-      "loss": 5.5482,
-      "step": 16421
-    },
-    {
-      "epoch": 8.564276401564538,
-      "grad_norm": 1.3402056694030762,
-      "learning_rate": 8.400301507537689e-05,
-      "loss": 5.8117,
-      "step": 16422
-    },
-    {
-      "epoch": 8.564797913950457,
-      "grad_norm": 1.440248727798462,
-      "learning_rate": 8.400201005025125e-05,
-      "loss": 5.5855,
-      "step": 16423
-    },
-    {
-      "epoch": 8.565319426336375,
-      "grad_norm": 1.4750702381134033,
-      "learning_rate": 8.400100502512563e-05,
-      "loss": 5.4166,
-      "step": 16424
-    },
-    {
-      "epoch": 8.565840938722294,
-      "grad_norm": 1.5666407346725464,
-      "learning_rate": 8.4e-05,
-      "loss": 5.2097,
-      "step": 16425
-    },
-    {
-      "epoch": 8.566362451108214,
-      "grad_norm": 1.4203320741653442,
-      "learning_rate": 8.399899497487437e-05,
-      "loss": 5.6418,
-      "step": 16426
-    },
-    {
-      "epoch": 8.566883963494133,
-      "grad_norm": 1.3683960437774658,
-      "learning_rate": 8.399798994974875e-05,
-      "loss": 5.5084,
-      "step": 16427
-    },
-    {
-      "epoch": 8.567405475880053,
-      "grad_norm": 1.5285090208053589,
-      "learning_rate": 8.399698492462313e-05,
-      "loss": 5.5311,
-      "step": 16428
-    },
-    {
-      "epoch": 8.567926988265972,
-      "grad_norm": 1.4183475971221924,
-      "learning_rate": 8.399597989949749e-05,
-      "loss": 5.4719,
-      "step": 16429
-    },
-    {
-      "epoch": 8.56844850065189,
-      "grad_norm": 1.4531680345535278,
-      "learning_rate": 8.399497487437187e-05,
-      "loss": 5.5417,
-      "step": 16430
-    },
-    {
-      "epoch": 8.56897001303781,
-      "grad_norm": 1.42386794090271,
-      "learning_rate": 8.399396984924624e-05,
-      "loss": 5.4014,
-      "step": 16431
-    },
-    {
-      "epoch": 8.569491525423729,
-      "grad_norm": 1.4993036985397339,
-      "learning_rate": 8.399296482412061e-05,
-      "loss": 5.2145,
-      "step": 16432
-    },
-    {
-      "epoch": 8.570013037809648,
-      "grad_norm": 1.4096256494522095,
-      "learning_rate": 8.399195979899498e-05,
-      "loss": 5.9852,
-      "step": 16433
-    },
-    {
-      "epoch": 8.570534550195568,
-      "grad_norm": 1.475264549255371,
-      "learning_rate": 8.399095477386934e-05,
-      "loss": 5.5759,
-      "step": 16434
-    },
-    {
-      "epoch": 8.571056062581487,
-      "grad_norm": 1.562327265739441,
-      "learning_rate": 8.398994974874372e-05,
-      "loss": 5.5271,
-      "step": 16435
-    },
-    {
-      "epoch": 8.571577574967405,
-      "grad_norm": 1.4583144187927246,
-      "learning_rate": 8.398894472361808e-05,
-      "loss": 5.5762,
-      "step": 16436
-    },
-    {
-      "epoch": 8.572099087353324,
-      "grad_norm": 1.3762739896774292,
-      "learning_rate": 8.398793969849246e-05,
-      "loss": 5.1247,
-      "step": 16437
-    },
-    {
-      "epoch": 8.572620599739244,
-      "grad_norm": 1.5386998653411865,
-      "learning_rate": 8.398693467336684e-05,
-      "loss": 5.5611,
-      "step": 16438
-    },
-    {
-      "epoch": 8.573142112125163,
-      "grad_norm": 1.4216035604476929,
-      "learning_rate": 8.398592964824122e-05,
-      "loss": 6.0088,
-      "step": 16439
-    },
-    {
-      "epoch": 8.573663624511083,
-      "grad_norm": 1.327891230583191,
-      "learning_rate": 8.398492462311558e-05,
-      "loss": 5.7278,
-      "step": 16440
-    },
-    {
-      "epoch": 8.574185136897002,
-      "grad_norm": 1.3435205221176147,
-      "learning_rate": 8.398391959798996e-05,
-      "loss": 5.5298,
-      "step": 16441
-    },
-    {
-      "epoch": 8.57470664928292,
-      "grad_norm": 1.4175007343292236,
-      "learning_rate": 8.398291457286432e-05,
-      "loss": 5.8152,
-      "step": 16442
-    },
-    {
-      "epoch": 8.57522816166884,
-      "grad_norm": 1.4266506433486938,
-      "learning_rate": 8.39819095477387e-05,
-      "loss": 5.549,
-      "step": 16443
-    },
-    {
-      "epoch": 8.575749674054759,
-      "grad_norm": 1.3712613582611084,
-      "learning_rate": 8.398090452261307e-05,
-      "loss": 5.8054,
-      "step": 16444
-    },
-    {
-      "epoch": 8.576271186440678,
-      "grad_norm": 1.294525384902954,
-      "learning_rate": 8.397989949748744e-05,
-      "loss": 5.6886,
-      "step": 16445
-    },
-    {
-      "epoch": 8.576792698826598,
-      "grad_norm": 1.3315651416778564,
-      "learning_rate": 8.397889447236181e-05,
-      "loss": 5.6663,
-      "step": 16446
-    },
-    {
-      "epoch": 8.577314211212517,
-      "grad_norm": 1.4869273900985718,
-      "learning_rate": 8.397788944723619e-05,
-      "loss": 4.5788,
-      "step": 16447
-    },
-    {
-      "epoch": 8.577835723598435,
-      "grad_norm": 1.3594143390655518,
-      "learning_rate": 8.397688442211056e-05,
-      "loss": 5.6146,
-      "step": 16448
-    },
-    {
-      "epoch": 8.578357235984354,
-      "grad_norm": 1.3804850578308105,
-      "learning_rate": 8.397587939698493e-05,
-      "loss": 4.865,
-      "step": 16449
-    },
-    {
-      "epoch": 8.578878748370274,
-      "grad_norm": 1.5370855331420898,
-      "learning_rate": 8.39748743718593e-05,
-      "loss": 5.6898,
-      "step": 16450
-    },
-    {
-      "epoch": 8.579400260756193,
-      "grad_norm": 1.639402151107788,
-      "learning_rate": 8.397386934673367e-05,
-      "loss": 5.1599,
-      "step": 16451
-    },
-    {
-      "epoch": 8.579921773142113,
-      "grad_norm": 1.4942504167556763,
-      "learning_rate": 8.397286432160805e-05,
-      "loss": 5.1904,
-      "step": 16452
-    },
-    {
-      "epoch": 8.58044328552803,
-      "grad_norm": 1.413206934928894,
-      "learning_rate": 8.397185929648241e-05,
-      "loss": 5.4894,
-      "step": 16453
-    },
-    {
-      "epoch": 8.58096479791395,
-      "grad_norm": 1.6142665147781372,
-      "learning_rate": 8.397085427135679e-05,
-      "loss": 5.7246,
-      "step": 16454
-    },
-    {
-      "epoch": 8.58148631029987,
-      "grad_norm": 1.4179819822311401,
-      "learning_rate": 8.396984924623115e-05,
-      "loss": 5.7597,
-      "step": 16455
-    },
-    {
-      "epoch": 8.582007822685789,
-      "grad_norm": 1.4921118021011353,
-      "learning_rate": 8.396884422110553e-05,
-      "loss": 5.2921,
-      "step": 16456
-    },
-    {
-      "epoch": 8.582529335071708,
-      "grad_norm": 1.3822928667068481,
-      "learning_rate": 8.39678391959799e-05,
-      "loss": 5.851,
-      "step": 16457
-    },
-    {
-      "epoch": 8.583050847457628,
-      "grad_norm": 1.4476432800292969,
-      "learning_rate": 8.396683417085427e-05,
-      "loss": 5.8784,
-      "step": 16458
-    },
-    {
-      "epoch": 8.583572359843547,
-      "grad_norm": 1.359006643295288,
-      "learning_rate": 8.396582914572865e-05,
-      "loss": 5.6763,
-      "step": 16459
-    },
-    {
-      "epoch": 8.584093872229465,
-      "grad_norm": 1.4410808086395264,
-      "learning_rate": 8.396482412060302e-05,
-      "loss": 5.5889,
-      "step": 16460
-    },
-    {
-      "epoch": 8.584615384615384,
-      "grad_norm": 1.4647161960601807,
-      "learning_rate": 8.39638190954774e-05,
-      "loss": 5.2294,
-      "step": 16461
-    },
-    {
-      "epoch": 8.585136897001304,
-      "grad_norm": 1.5320937633514404,
-      "learning_rate": 8.396281407035176e-05,
-      "loss": 5.4201,
-      "step": 16462
-    },
-    {
-      "epoch": 8.585658409387223,
-      "grad_norm": 1.5162944793701172,
-      "learning_rate": 8.396180904522614e-05,
-      "loss": 5.6362,
-      "step": 16463
-    },
-    {
-      "epoch": 8.586179921773143,
-      "grad_norm": 1.4587829113006592,
-      "learning_rate": 8.39608040201005e-05,
-      "loss": 5.4805,
-      "step": 16464
-    },
-    {
-      "epoch": 8.58670143415906,
-      "grad_norm": 1.5023447275161743,
-      "learning_rate": 8.395979899497488e-05,
-      "loss": 5.7375,
-      "step": 16465
-    },
-    {
-      "epoch": 8.58722294654498,
-      "grad_norm": 1.3555914163589478,
-      "learning_rate": 8.395879396984924e-05,
-      "loss": 5.5202,
-      "step": 16466
-    },
-    {
-      "epoch": 8.5877444589309,
-      "grad_norm": 1.4481830596923828,
-      "learning_rate": 8.395778894472362e-05,
-      "loss": 5.7434,
-      "step": 16467
-    },
-    {
-      "epoch": 8.588265971316819,
-      "grad_norm": 1.4838504791259766,
-      "learning_rate": 8.3956783919598e-05,
-      "loss": 5.3673,
-      "step": 16468
-    },
-    {
-      "epoch": 8.588787483702738,
-      "grad_norm": 1.47458815574646,
-      "learning_rate": 8.395577889447238e-05,
-      "loss": 5.1476,
-      "step": 16469
-    },
-    {
-      "epoch": 8.589308996088658,
-      "grad_norm": 1.4146173000335693,
-      "learning_rate": 8.395477386934674e-05,
-      "loss": 5.7709,
-      "step": 16470
-    },
-    {
-      "epoch": 8.589830508474577,
-      "grad_norm": 1.5466421842575073,
-      "learning_rate": 8.395376884422112e-05,
-      "loss": 5.3961,
-      "step": 16471
-    },
-    {
-      "epoch": 8.590352020860495,
-      "grad_norm": 1.449985146522522,
-      "learning_rate": 8.395276381909548e-05,
-      "loss": 5.76,
-      "step": 16472
-    },
-    {
-      "epoch": 8.590873533246414,
-      "grad_norm": 1.4856599569320679,
-      "learning_rate": 8.395175879396985e-05,
-      "loss": 5.5067,
-      "step": 16473
-    },
-    {
-      "epoch": 8.591395045632334,
-      "grad_norm": 1.3422532081604004,
-      "learning_rate": 8.395075376884422e-05,
-      "loss": 5.8052,
-      "step": 16474
-    },
-    {
-      "epoch": 8.591916558018253,
-      "grad_norm": 1.4634809494018555,
-      "learning_rate": 8.394974874371859e-05,
-      "loss": 5.4581,
-      "step": 16475
-    },
-    {
-      "epoch": 8.592438070404173,
-      "grad_norm": 1.4853835105895996,
-      "learning_rate": 8.394874371859297e-05,
-      "loss": 5.3119,
-      "step": 16476
-    },
-    {
-      "epoch": 8.59295958279009,
-      "grad_norm": 1.4685561656951904,
-      "learning_rate": 8.394773869346733e-05,
-      "loss": 5.7229,
-      "step": 16477
-    },
-    {
-      "epoch": 8.59348109517601,
-      "grad_norm": 1.4492119550704956,
-      "learning_rate": 8.394673366834171e-05,
-      "loss": 5.7958,
-      "step": 16478
-    },
-    {
-      "epoch": 8.59400260756193,
-      "grad_norm": 1.34714937210083,
-      "learning_rate": 8.394572864321609e-05,
-      "loss": 6.0179,
-      "step": 16479
-    },
-    {
-      "epoch": 8.594524119947849,
-      "grad_norm": 1.346030592918396,
-      "learning_rate": 8.394472361809046e-05,
-      "loss": 5.5636,
-      "step": 16480
-    },
-    {
-      "epoch": 8.595045632333768,
-      "grad_norm": 1.476633071899414,
-      "learning_rate": 8.394371859296483e-05,
-      "loss": 4.9938,
-      "step": 16481
-    },
-    {
-      "epoch": 8.595567144719688,
-      "grad_norm": 1.4267587661743164,
-      "learning_rate": 8.39427135678392e-05,
-      "loss": 5.5708,
-      "step": 16482
-    },
-    {
-      "epoch": 8.596088657105605,
-      "grad_norm": 1.5029383897781372,
-      "learning_rate": 8.394170854271357e-05,
-      "loss": 5.4351,
-      "step": 16483
-    },
-    {
-      "epoch": 8.596610169491525,
-      "grad_norm": 1.6294695138931274,
-      "learning_rate": 8.394070351758795e-05,
-      "loss": 5.5123,
-      "step": 16484
-    },
-    {
-      "epoch": 8.597131681877444,
-      "grad_norm": 1.4931389093399048,
-      "learning_rate": 8.393969849246231e-05,
-      "loss": 5.8067,
-      "step": 16485
-    },
-    {
-      "epoch": 8.597653194263364,
-      "grad_norm": 1.5537701845169067,
-      "learning_rate": 8.393869346733668e-05,
-      "loss": 5.6398,
-      "step": 16486
-    },
-    {
-      "epoch": 8.598174706649283,
-      "grad_norm": 1.4526457786560059,
-      "learning_rate": 8.393768844221105e-05,
-      "loss": 5.6443,
-      "step": 16487
-    },
-    {
-      "epoch": 8.598696219035203,
-      "grad_norm": 1.6119247674942017,
-      "learning_rate": 8.393668341708543e-05,
-      "loss": 5.1907,
-      "step": 16488
-    },
-    {
-      "epoch": 8.59921773142112,
-      "grad_norm": 1.4856294393539429,
-      "learning_rate": 8.393567839195981e-05,
-      "loss": 5.8804,
-      "step": 16489
-    },
-    {
-      "epoch": 8.59973924380704,
-      "grad_norm": 1.636724829673767,
-      "learning_rate": 8.393467336683417e-05,
-      "loss": 4.9076,
-      "step": 16490
-    },
-    {
-      "epoch": 8.60026075619296,
-      "grad_norm": 1.4764469861984253,
-      "learning_rate": 8.393366834170855e-05,
-      "loss": 5.7675,
-      "step": 16491
-    },
-    {
-      "epoch": 8.600782268578879,
-      "grad_norm": 1.4495655298233032,
-      "learning_rate": 8.393266331658292e-05,
-      "loss": 5.1568,
-      "step": 16492
-    },
-    {
-      "epoch": 8.601303780964798,
-      "grad_norm": 1.519147515296936,
-      "learning_rate": 8.39316582914573e-05,
-      "loss": 5.6042,
-      "step": 16493
-    },
-    {
-      "epoch": 8.601825293350718,
-      "grad_norm": 1.677255392074585,
-      "learning_rate": 8.393065326633166e-05,
-      "loss": 5.8338,
-      "step": 16494
-    },
-    {
-      "epoch": 8.602346805736635,
-      "grad_norm": 1.4560002088546753,
-      "learning_rate": 8.392964824120604e-05,
-      "loss": 5.6589,
-      "step": 16495
-    },
-    {
-      "epoch": 8.602868318122555,
-      "grad_norm": 1.3913286924362183,
-      "learning_rate": 8.39286432160804e-05,
-      "loss": 5.6062,
-      "step": 16496
-    },
-    {
-      "epoch": 8.603389830508474,
-      "grad_norm": 1.362370491027832,
-      "learning_rate": 8.392763819095478e-05,
-      "loss": 5.8623,
-      "step": 16497
-    },
-    {
-      "epoch": 8.603911342894394,
-      "grad_norm": 1.344109058380127,
-      "learning_rate": 8.392663316582914e-05,
-      "loss": 5.4936,
-      "step": 16498
-    },
-    {
-      "epoch": 8.604432855280313,
-      "grad_norm": 1.5962419509887695,
-      "learning_rate": 8.392562814070352e-05,
-      "loss": 5.4933,
-      "step": 16499
-    },
-    {
-      "epoch": 8.604954367666233,
-      "grad_norm": 1.5037920475006104,
-      "learning_rate": 8.39246231155779e-05,
-      "loss": 5.3841,
-      "step": 16500
-    },
-    {
-      "epoch": 8.60547588005215,
-      "grad_norm": 1.4102195501327515,
-      "learning_rate": 8.392361809045226e-05,
-      "loss": 5.809,
-      "step": 16501
-    },
-    {
-      "epoch": 8.60599739243807,
-      "grad_norm": 1.4363417625427246,
-      "learning_rate": 8.392261306532664e-05,
-      "loss": 5.4381,
-      "step": 16502
-    },
-    {
-      "epoch": 8.60651890482399,
-      "grad_norm": 1.4897860288619995,
-      "learning_rate": 8.3921608040201e-05,
-      "loss": 5.2644,
-      "step": 16503
-    },
-    {
-      "epoch": 8.607040417209909,
-      "grad_norm": 1.5019948482513428,
-      "learning_rate": 8.392060301507538e-05,
-      "loss": 5.1836,
-      "step": 16504
-    },
-    {
-      "epoch": 8.607561929595828,
-      "grad_norm": 1.5264772176742554,
-      "learning_rate": 8.391959798994975e-05,
-      "loss": 5.235,
-      "step": 16505
-    },
-    {
-      "epoch": 8.608083441981748,
-      "grad_norm": 1.3974666595458984,
-      "learning_rate": 8.391859296482413e-05,
-      "loss": 5.6258,
-      "step": 16506
-    },
-    {
-      "epoch": 8.608604954367665,
-      "grad_norm": 1.443668007850647,
-      "learning_rate": 8.391758793969849e-05,
-      "loss": 5.6057,
-      "step": 16507
-    },
-    {
-      "epoch": 8.609126466753585,
-      "grad_norm": 1.3504314422607422,
-      "learning_rate": 8.391658291457287e-05,
-      "loss": 5.7339,
-      "step": 16508
-    },
-    {
-      "epoch": 8.609647979139504,
-      "grad_norm": 1.3409943580627441,
-      "learning_rate": 8.391557788944725e-05,
-      "loss": 5.7428,
-      "step": 16509
-    },
-    {
-      "epoch": 8.610169491525424,
-      "grad_norm": 1.45543372631073,
-      "learning_rate": 8.391457286432162e-05,
-      "loss": 5.6455,
-      "step": 16510
-    },
-    {
-      "epoch": 8.610691003911343,
-      "grad_norm": 1.395363211631775,
-      "learning_rate": 8.391356783919599e-05,
-      "loss": 5.8471,
-      "step": 16511
-    },
-    {
-      "epoch": 8.611212516297263,
-      "grad_norm": 1.2858837842941284,
-      "learning_rate": 8.391256281407036e-05,
-      "loss": 5.4495,
-      "step": 16512
-    },
-    {
-      "epoch": 8.61173402868318,
-      "grad_norm": 1.366391658782959,
-      "learning_rate": 8.391155778894473e-05,
-      "loss": 5.7874,
-      "step": 16513
-    },
-    {
-      "epoch": 8.6122555410691,
-      "grad_norm": 1.6932330131530762,
-      "learning_rate": 8.39105527638191e-05,
-      "loss": 4.9325,
-      "step": 16514
-    },
-    {
-      "epoch": 8.61277705345502,
-      "grad_norm": 1.3596315383911133,
-      "learning_rate": 8.390954773869347e-05,
-      "loss": 5.225,
-      "step": 16515
-    },
-    {
-      "epoch": 8.613298565840939,
-      "grad_norm": 1.458364486694336,
-      "learning_rate": 8.390854271356784e-05,
-      "loss": 5.3605,
-      "step": 16516
-    },
-    {
-      "epoch": 8.613820078226858,
-      "grad_norm": 1.4696861505508423,
-      "learning_rate": 8.390753768844221e-05,
-      "loss": 5.5098,
-      "step": 16517
-    },
-    {
-      "epoch": 8.614341590612778,
-      "grad_norm": 1.6012495756149292,
-      "learning_rate": 8.390653266331658e-05,
-      "loss": 5.4593,
-      "step": 16518
-    },
-    {
-      "epoch": 8.614863102998696,
-      "grad_norm": 1.3780916929244995,
-      "learning_rate": 8.390552763819096e-05,
-      "loss": 5.1492,
-      "step": 16519
-    },
-    {
-      "epoch": 8.615384615384615,
-      "grad_norm": 1.4615017175674438,
-      "learning_rate": 8.390452261306533e-05,
-      "loss": 5.5484,
-      "step": 16520
-    },
-    {
-      "epoch": 8.615906127770534,
-      "grad_norm": 1.3668862581253052,
-      "learning_rate": 8.390351758793971e-05,
-      "loss": 5.5416,
-      "step": 16521
-    },
-    {
-      "epoch": 8.616427640156454,
-      "grad_norm": 1.4412810802459717,
-      "learning_rate": 8.390251256281408e-05,
-      "loss": 5.7227,
-      "step": 16522
-    },
-    {
-      "epoch": 8.616949152542373,
-      "grad_norm": 1.4808135032653809,
-      "learning_rate": 8.390150753768845e-05,
-      "loss": 5.1615,
-      "step": 16523
-    },
-    {
-      "epoch": 8.617470664928293,
-      "grad_norm": 1.5771206617355347,
-      "learning_rate": 8.390050251256282e-05,
-      "loss": 5.713,
-      "step": 16524
-    },
-    {
-      "epoch": 8.61799217731421,
-      "grad_norm": 1.3992239236831665,
-      "learning_rate": 8.38994974874372e-05,
-      "loss": 5.503,
-      "step": 16525
-    },
-    {
-      "epoch": 8.61851368970013,
-      "grad_norm": 1.4297034740447998,
-      "learning_rate": 8.389849246231156e-05,
-      "loss": 5.7046,
-      "step": 16526
-    },
-    {
-      "epoch": 8.61903520208605,
-      "grad_norm": 1.4618194103240967,
-      "learning_rate": 8.389748743718592e-05,
-      "loss": 5.2111,
-      "step": 16527
-    },
-    {
-      "epoch": 8.619556714471969,
-      "grad_norm": 1.409436821937561,
-      "learning_rate": 8.38964824120603e-05,
-      "loss": 5.26,
-      "step": 16528
-    },
-    {
-      "epoch": 8.620078226857888,
-      "grad_norm": 1.481938123703003,
-      "learning_rate": 8.389547738693468e-05,
-      "loss": 5.5337,
-      "step": 16529
-    },
-    {
-      "epoch": 8.620599739243808,
-      "grad_norm": 1.4171274900436401,
-      "learning_rate": 8.389447236180906e-05,
-      "loss": 5.8327,
-      "step": 16530
-    },
-    {
-      "epoch": 8.621121251629726,
-      "grad_norm": 1.403080940246582,
-      "learning_rate": 8.389346733668342e-05,
-      "loss": 5.6899,
-      "step": 16531
-    },
-    {
-      "epoch": 8.621642764015645,
-      "grad_norm": 1.2920950651168823,
-      "learning_rate": 8.38924623115578e-05,
-      "loss": 5.9852,
-      "step": 16532
-    },
-    {
-      "epoch": 8.622164276401564,
-      "grad_norm": 1.4430574178695679,
-      "learning_rate": 8.389145728643216e-05,
-      "loss": 5.5321,
-      "step": 16533
-    },
-    {
-      "epoch": 8.622685788787484,
-      "grad_norm": 1.2973804473876953,
-      "learning_rate": 8.389045226130654e-05,
-      "loss": 5.921,
-      "step": 16534
-    },
-    {
-      "epoch": 8.623207301173403,
-      "grad_norm": 1.6506587266921997,
-      "learning_rate": 8.38894472361809e-05,
-      "loss": 5.7581,
-      "step": 16535
-    },
-    {
-      "epoch": 8.623728813559323,
-      "grad_norm": 1.521450161933899,
-      "learning_rate": 8.388844221105528e-05,
-      "loss": 5.4354,
-      "step": 16536
-    },
-    {
-      "epoch": 8.62425032594524,
-      "grad_norm": 1.596073865890503,
-      "learning_rate": 8.388743718592965e-05,
-      "loss": 5.5277,
-      "step": 16537
-    },
-    {
-      "epoch": 8.62477183833116,
-      "grad_norm": 1.5043069124221802,
-      "learning_rate": 8.388643216080403e-05,
-      "loss": 5.2578,
-      "step": 16538
-    },
-    {
-      "epoch": 8.62529335071708,
-      "grad_norm": 1.3466498851776123,
-      "learning_rate": 8.388542713567839e-05,
-      "loss": 5.6852,
-      "step": 16539
-    },
-    {
-      "epoch": 8.625814863102999,
-      "grad_norm": 1.5276778936386108,
-      "learning_rate": 8.388442211055277e-05,
-      "loss": 5.5039,
-      "step": 16540
-    },
-    {
-      "epoch": 8.626336375488918,
-      "grad_norm": 1.50129234790802,
-      "learning_rate": 8.388341708542715e-05,
-      "loss": 5.4158,
-      "step": 16541
-    },
-    {
-      "epoch": 8.626857887874838,
-      "grad_norm": 1.472386121749878,
-      "learning_rate": 8.388241206030151e-05,
-      "loss": 5.3504,
-      "step": 16542
-    },
-    {
-      "epoch": 8.627379400260756,
-      "grad_norm": 1.4357982873916626,
-      "learning_rate": 8.388140703517589e-05,
-      "loss": 5.705,
-      "step": 16543
-    },
-    {
-      "epoch": 8.627900912646675,
-      "grad_norm": 1.3610848188400269,
-      "learning_rate": 8.388040201005025e-05,
-      "loss": 5.5192,
-      "step": 16544
-    },
-    {
-      "epoch": 8.628422425032594,
-      "grad_norm": 1.3610913753509521,
-      "learning_rate": 8.387939698492463e-05,
-      "loss": 5.8306,
-      "step": 16545
-    },
-    {
-      "epoch": 8.628943937418514,
-      "grad_norm": 1.36136794090271,
-      "learning_rate": 8.3878391959799e-05,
-      "loss": 5.6766,
-      "step": 16546
-    },
-    {
-      "epoch": 8.629465449804433,
-      "grad_norm": 1.440032958984375,
-      "learning_rate": 8.387738693467337e-05,
-      "loss": 5.4307,
-      "step": 16547
-    },
-    {
-      "epoch": 8.629986962190351,
-      "grad_norm": 1.3191920518875122,
-      "learning_rate": 8.387638190954774e-05,
-      "loss": 5.6834,
-      "step": 16548
-    },
-    {
-      "epoch": 8.63050847457627,
-      "grad_norm": 1.4927971363067627,
-      "learning_rate": 8.387537688442211e-05,
-      "loss": 5.5975,
-      "step": 16549
-    },
-    {
-      "epoch": 8.63102998696219,
-      "grad_norm": 1.393389105796814,
-      "learning_rate": 8.387437185929649e-05,
-      "loss": 5.3358,
-      "step": 16550
-    },
-    {
-      "epoch": 8.63155149934811,
-      "grad_norm": 1.4889214038848877,
-      "learning_rate": 8.387336683417087e-05,
-      "loss": 5.0777,
-      "step": 16551
-    },
-    {
-      "epoch": 8.632073011734029,
-      "grad_norm": 1.4875729084014893,
-      "learning_rate": 8.387236180904523e-05,
-      "loss": 5.3896,
-      "step": 16552
-    },
-    {
-      "epoch": 8.632594524119948,
-      "grad_norm": 1.5702426433563232,
-      "learning_rate": 8.38713567839196e-05,
-      "loss": 5.4024,
-      "step": 16553
-    },
-    {
-      "epoch": 8.633116036505868,
-      "grad_norm": 1.4722286462783813,
-      "learning_rate": 8.387035175879398e-05,
-      "loss": 5.4165,
-      "step": 16554
-    },
-    {
-      "epoch": 8.633637548891786,
-      "grad_norm": 1.5269958972930908,
-      "learning_rate": 8.386934673366834e-05,
-      "loss": 5.2223,
-      "step": 16555
-    },
-    {
-      "epoch": 8.634159061277705,
-      "grad_norm": 1.428118109703064,
-      "learning_rate": 8.386834170854272e-05,
-      "loss": 5.4126,
-      "step": 16556
-    },
-    {
-      "epoch": 8.634680573663625,
-      "grad_norm": 1.4544731378555298,
-      "learning_rate": 8.386733668341708e-05,
-      "loss": 5.2815,
-      "step": 16557
-    },
-    {
-      "epoch": 8.635202086049544,
-      "grad_norm": 1.3946664333343506,
-      "learning_rate": 8.386633165829146e-05,
-      "loss": 5.5954,
-      "step": 16558
-    },
-    {
-      "epoch": 8.635723598435463,
-      "grad_norm": 1.3931223154067993,
-      "learning_rate": 8.386532663316582e-05,
-      "loss": 5.5907,
-      "step": 16559
-    },
-    {
-      "epoch": 8.636245110821381,
-      "grad_norm": 1.459550142288208,
-      "learning_rate": 8.38643216080402e-05,
-      "loss": 5.5241,
-      "step": 16560
-    },
-    {
-      "epoch": 8.6367666232073,
-      "grad_norm": 1.4300659894943237,
-      "learning_rate": 8.386331658291458e-05,
-      "loss": 5.8512,
-      "step": 16561
-    },
-    {
-      "epoch": 8.63728813559322,
-      "grad_norm": 1.5030372142791748,
-      "learning_rate": 8.386231155778896e-05,
-      "loss": 5.6323,
-      "step": 16562
-    },
-    {
-      "epoch": 8.63780964797914,
-      "grad_norm": 1.405116319656372,
-      "learning_rate": 8.386130653266332e-05,
-      "loss": 5.3041,
-      "step": 16563
-    },
-    {
-      "epoch": 8.638331160365059,
-      "grad_norm": 1.4750925302505493,
-      "learning_rate": 8.38603015075377e-05,
-      "loss": 4.863,
-      "step": 16564
-    },
-    {
-      "epoch": 8.638852672750978,
-      "grad_norm": 1.569077968597412,
-      "learning_rate": 8.385929648241206e-05,
-      "loss": 4.8725,
-      "step": 16565
-    },
-    {
-      "epoch": 8.639374185136898,
-      "grad_norm": 1.5400176048278809,
-      "learning_rate": 8.385829145728643e-05,
-      "loss": 4.9517,
-      "step": 16566
-    },
-    {
-      "epoch": 8.639895697522816,
-      "grad_norm": 1.4053049087524414,
-      "learning_rate": 8.38572864321608e-05,
-      "loss": 5.6578,
-      "step": 16567
-    },
-    {
-      "epoch": 8.640417209908735,
-      "grad_norm": 1.3419307470321655,
-      "learning_rate": 8.385628140703517e-05,
-      "loss": 5.7574,
-      "step": 16568
-    },
-    {
-      "epoch": 8.640938722294655,
-      "grad_norm": 1.4492665529251099,
-      "learning_rate": 8.385527638190955e-05,
-      "loss": 5.1601,
-      "step": 16569
-    },
-    {
-      "epoch": 8.641460234680574,
-      "grad_norm": 1.487099051475525,
-      "learning_rate": 8.385427135678393e-05,
-      "loss": 5.2153,
-      "step": 16570
-    },
-    {
-      "epoch": 8.641981747066493,
-      "grad_norm": 1.4220266342163086,
-      "learning_rate": 8.38532663316583e-05,
-      "loss": 5.7231,
-      "step": 16571
-    },
-    {
-      "epoch": 8.642503259452411,
-      "grad_norm": 1.453260898590088,
-      "learning_rate": 8.385226130653267e-05,
-      "loss": 5.7866,
-      "step": 16572
-    },
-    {
-      "epoch": 8.64302477183833,
-      "grad_norm": 1.3707654476165771,
-      "learning_rate": 8.385125628140705e-05,
-      "loss": 5.7274,
-      "step": 16573
-    },
-    {
-      "epoch": 8.64354628422425,
-      "grad_norm": 1.487196445465088,
-      "learning_rate": 8.385025125628141e-05,
-      "loss": 5.173,
-      "step": 16574
-    },
-    {
-      "epoch": 8.64406779661017,
-      "grad_norm": 1.6495578289031982,
-      "learning_rate": 8.384924623115579e-05,
-      "loss": 5.1331,
-      "step": 16575
-    },
-    {
-      "epoch": 8.644589308996089,
-      "grad_norm": 1.3457738161087036,
-      "learning_rate": 8.384824120603015e-05,
-      "loss": 5.737,
-      "step": 16576
-    },
-    {
-      "epoch": 8.645110821382008,
-      "grad_norm": 1.4844590425491333,
-      "learning_rate": 8.384723618090453e-05,
-      "loss": 5.5775,
-      "step": 16577
-    },
-    {
-      "epoch": 8.645632333767926,
-      "grad_norm": 1.3897250890731812,
-      "learning_rate": 8.38462311557789e-05,
-      "loss": 5.5134,
-      "step": 16578
-    },
-    {
-      "epoch": 8.646153846153846,
-      "grad_norm": 1.4173873662948608,
-      "learning_rate": 8.384522613065326e-05,
-      "loss": 5.6111,
-      "step": 16579
-    },
-    {
-      "epoch": 8.646675358539765,
-      "grad_norm": 1.5056480169296265,
-      "learning_rate": 8.384422110552764e-05,
-      "loss": 5.6111,
-      "step": 16580
-    },
-    {
-      "epoch": 8.647196870925685,
-      "grad_norm": 1.397499442100525,
-      "learning_rate": 8.384321608040201e-05,
-      "loss": 5.6046,
-      "step": 16581
-    },
-    {
-      "epoch": 8.647718383311604,
-      "grad_norm": 1.5398305654525757,
-      "learning_rate": 8.384221105527639e-05,
-      "loss": 5.4465,
-      "step": 16582
-    },
-    {
-      "epoch": 8.648239895697523,
-      "grad_norm": 1.4561458826065063,
-      "learning_rate": 8.384120603015076e-05,
-      "loss": 5.5711,
-      "step": 16583
-    },
-    {
-      "epoch": 8.648761408083441,
-      "grad_norm": 1.5548501014709473,
-      "learning_rate": 8.384020100502513e-05,
-      "loss": 5.0746,
-      "step": 16584
-    },
-    {
-      "epoch": 8.64928292046936,
-      "grad_norm": 1.4613229036331177,
-      "learning_rate": 8.38391959798995e-05,
-      "loss": 5.7759,
-      "step": 16585
-    },
-    {
-      "epoch": 8.64980443285528,
-      "grad_norm": 1.5699530839920044,
-      "learning_rate": 8.383819095477388e-05,
-      "loss": 5.2883,
-      "step": 16586
-    },
-    {
-      "epoch": 8.6503259452412,
-      "grad_norm": 1.3970142602920532,
-      "learning_rate": 8.383718592964824e-05,
-      "loss": 5.6803,
-      "step": 16587
-    },
-    {
-      "epoch": 8.650847457627119,
-      "grad_norm": 1.49769127368927,
-      "learning_rate": 8.383618090452262e-05,
-      "loss": 5.1077,
-      "step": 16588
-    },
-    {
-      "epoch": 8.651368970013039,
-      "grad_norm": 1.5174593925476074,
-      "learning_rate": 8.383517587939698e-05,
-      "loss": 5.6081,
-      "step": 16589
-    },
-    {
-      "epoch": 8.651890482398956,
-      "grad_norm": 1.519070029258728,
-      "learning_rate": 8.383417085427136e-05,
-      "loss": 5.4573,
-      "step": 16590
-    },
-    {
-      "epoch": 8.652411994784876,
-      "grad_norm": 1.391111969947815,
-      "learning_rate": 8.383316582914574e-05,
-      "loss": 5.6005,
-      "step": 16591
-    },
-    {
-      "epoch": 8.652933507170795,
-      "grad_norm": 1.52463960647583,
-      "learning_rate": 8.38321608040201e-05,
-      "loss": 5.5859,
-      "step": 16592
-    },
-    {
-      "epoch": 8.653455019556715,
-      "grad_norm": 1.5849878787994385,
-      "learning_rate": 8.383115577889448e-05,
-      "loss": 5.3983,
-      "step": 16593
-    },
-    {
-      "epoch": 8.653976531942634,
-      "grad_norm": 1.430938959121704,
-      "learning_rate": 8.383015075376885e-05,
-      "loss": 5.4646,
-      "step": 16594
-    },
-    {
-      "epoch": 8.654498044328554,
-      "grad_norm": 1.525486946105957,
-      "learning_rate": 8.382914572864322e-05,
-      "loss": 5.3287,
-      "step": 16595
-    },
-    {
-      "epoch": 8.655019556714471,
-      "grad_norm": 1.5679303407669067,
-      "learning_rate": 8.382814070351759e-05,
-      "loss": 5.2893,
-      "step": 16596
-    },
-    {
-      "epoch": 8.65554106910039,
-      "grad_norm": 1.312322735786438,
-      "learning_rate": 8.382713567839197e-05,
-      "loss": 5.9517,
-      "step": 16597
-    },
-    {
-      "epoch": 8.65606258148631,
-      "grad_norm": 1.4227416515350342,
-      "learning_rate": 8.382613065326633e-05,
-      "loss": 5.1929,
-      "step": 16598
-    },
-    {
-      "epoch": 8.65658409387223,
-      "grad_norm": 1.429144263267517,
-      "learning_rate": 8.382512562814071e-05,
-      "loss": 5.5503,
-      "step": 16599
-    },
-    {
-      "epoch": 8.657105606258149,
-      "grad_norm": 1.4796130657196045,
-      "learning_rate": 8.382412060301507e-05,
-      "loss": 5.2019,
-      "step": 16600
-    },
-    {
-      "epoch": 8.657627118644069,
-      "grad_norm": 1.555585503578186,
-      "learning_rate": 8.382311557788945e-05,
-      "loss": 5.1949,
-      "step": 16601
-    },
-    {
-      "epoch": 8.658148631029986,
-      "grad_norm": 1.5434623956680298,
-      "learning_rate": 8.382211055276383e-05,
-      "loss": 5.7691,
-      "step": 16602
-    },
-    {
-      "epoch": 8.658670143415906,
-      "grad_norm": 1.4552699327468872,
-      "learning_rate": 8.38211055276382e-05,
-      "loss": 5.5276,
-      "step": 16603
-    },
-    {
-      "epoch": 8.659191655801825,
-      "grad_norm": 1.494680643081665,
-      "learning_rate": 8.382010050251257e-05,
-      "loss": 5.3751,
-      "step": 16604
-    },
-    {
-      "epoch": 8.659713168187745,
-      "grad_norm": 1.4369823932647705,
-      "learning_rate": 8.381909547738695e-05,
-      "loss": 5.6482,
-      "step": 16605
-    },
-    {
-      "epoch": 8.660234680573664,
-      "grad_norm": 1.360063076019287,
-      "learning_rate": 8.381809045226131e-05,
-      "loss": 5.3941,
-      "step": 16606
-    },
-    {
-      "epoch": 8.660756192959584,
-      "grad_norm": 1.672528862953186,
-      "learning_rate": 8.381708542713568e-05,
-      "loss": 4.9607,
-      "step": 16607
-    },
-    {
-      "epoch": 8.661277705345501,
-      "grad_norm": 1.5254144668579102,
-      "learning_rate": 8.381608040201005e-05,
-      "loss": 5.7766,
-      "step": 16608
-    },
-    {
-      "epoch": 8.66179921773142,
-      "grad_norm": 1.5074207782745361,
-      "learning_rate": 8.381507537688442e-05,
-      "loss": 5.329,
-      "step": 16609
-    },
-    {
-      "epoch": 8.66232073011734,
-      "grad_norm": 1.401104211807251,
-      "learning_rate": 8.38140703517588e-05,
-      "loss": 5.6074,
-      "step": 16610
-    },
-    {
-      "epoch": 8.66284224250326,
-      "grad_norm": 1.4910780191421509,
-      "learning_rate": 8.381306532663316e-05,
-      "loss": 5.5213,
-      "step": 16611
-    },
-    {
-      "epoch": 8.663363754889179,
-      "grad_norm": 1.4465361833572388,
-      "learning_rate": 8.381206030150754e-05,
-      "loss": 5.2407,
-      "step": 16612
-    },
-    {
-      "epoch": 8.663885267275099,
-      "grad_norm": 1.5335185527801514,
-      "learning_rate": 8.381105527638192e-05,
-      "loss": 5.018,
-      "step": 16613
-    },
-    {
-      "epoch": 8.664406779661016,
-      "grad_norm": 1.4419161081314087,
-      "learning_rate": 8.38100502512563e-05,
-      "loss": 5.4165,
-      "step": 16614
-    },
-    {
-      "epoch": 8.664928292046936,
-      "grad_norm": 1.4816707372665405,
-      "learning_rate": 8.380904522613066e-05,
-      "loss": 5.7111,
-      "step": 16615
-    },
-    {
-      "epoch": 8.665449804432855,
-      "grad_norm": 1.4090548753738403,
-      "learning_rate": 8.380804020100504e-05,
-      "loss": 5.0344,
-      "step": 16616
-    },
-    {
-      "epoch": 8.665971316818775,
-      "grad_norm": 1.5145695209503174,
-      "learning_rate": 8.38070351758794e-05,
-      "loss": 5.4544,
-      "step": 16617
-    },
-    {
-      "epoch": 8.666492829204694,
-      "grad_norm": 1.4700844287872314,
-      "learning_rate": 8.380603015075378e-05,
-      "loss": 5.3024,
-      "step": 16618
-    },
-    {
-      "epoch": 8.667014341590614,
-      "grad_norm": 1.4595328569412231,
-      "learning_rate": 8.380502512562814e-05,
-      "loss": 5.4788,
-      "step": 16619
-    },
-    {
-      "epoch": 8.667535853976531,
-      "grad_norm": 1.3690799474716187,
-      "learning_rate": 8.38040201005025e-05,
-      "loss": 5.5844,
-      "step": 16620
-    },
-    {
-      "epoch": 8.66805736636245,
-      "grad_norm": 1.3423594236373901,
-      "learning_rate": 8.380301507537688e-05,
-      "loss": 5.4903,
-      "step": 16621
-    },
-    {
-      "epoch": 8.66857887874837,
-      "grad_norm": 1.449497103691101,
-      "learning_rate": 8.380201005025126e-05,
-      "loss": 5.3351,
-      "step": 16622
-    },
-    {
-      "epoch": 8.66910039113429,
-      "grad_norm": 1.5126217603683472,
-      "learning_rate": 8.380100502512564e-05,
-      "loss": 5.219,
-      "step": 16623
-    },
-    {
-      "epoch": 8.66962190352021,
-      "grad_norm": 1.4892776012420654,
-      "learning_rate": 8.38e-05,
-      "loss": 5.1101,
-      "step": 16624
-    },
-    {
-      "epoch": 8.670143415906129,
-      "grad_norm": 1.4260246753692627,
-      "learning_rate": 8.379899497487438e-05,
-      "loss": 5.3634,
-      "step": 16625
-    },
-    {
-      "epoch": 8.670664928292046,
-      "grad_norm": 1.4476381540298462,
-      "learning_rate": 8.379798994974875e-05,
-      "loss": 5.5392,
-      "step": 16626
-    },
-    {
-      "epoch": 8.671186440677966,
-      "grad_norm": 1.4339954853057861,
-      "learning_rate": 8.379698492462312e-05,
-      "loss": 5.4832,
-      "step": 16627
-    },
-    {
-      "epoch": 8.671707953063885,
-      "grad_norm": 1.371097207069397,
-      "learning_rate": 8.379597989949749e-05,
-      "loss": 5.9241,
-      "step": 16628
-    },
-    {
-      "epoch": 8.672229465449805,
-      "grad_norm": 1.6015836000442505,
-      "learning_rate": 8.379497487437187e-05,
-      "loss": 4.8434,
-      "step": 16629
-    },
-    {
-      "epoch": 8.672750977835724,
-      "grad_norm": 1.5464744567871094,
-      "learning_rate": 8.379396984924623e-05,
-      "loss": 5.4228,
-      "step": 16630
-    },
-    {
-      "epoch": 8.673272490221644,
-      "grad_norm": 1.3944470882415771,
-      "learning_rate": 8.379296482412061e-05,
-      "loss": 5.4343,
-      "step": 16631
-    },
-    {
-      "epoch": 8.673794002607561,
-      "grad_norm": 1.5173838138580322,
-      "learning_rate": 8.379195979899497e-05,
-      "loss": 5.1319,
-      "step": 16632
-    },
-    {
-      "epoch": 8.67431551499348,
-      "grad_norm": 1.4492597579956055,
-      "learning_rate": 8.379095477386935e-05,
-      "loss": 5.2887,
-      "step": 16633
-    },
-    {
-      "epoch": 8.6748370273794,
-      "grad_norm": 1.365016222000122,
-      "learning_rate": 8.378994974874373e-05,
-      "loss": 5.6963,
-      "step": 16634
-    },
-    {
-      "epoch": 8.67535853976532,
-      "grad_norm": 1.3998199701309204,
-      "learning_rate": 8.378894472361809e-05,
-      "loss": 5.6116,
-      "step": 16635
-    },
-    {
-      "epoch": 8.67588005215124,
-      "grad_norm": 1.4436501264572144,
-      "learning_rate": 8.378793969849247e-05,
-      "loss": 5.2586,
-      "step": 16636
-    },
-    {
-      "epoch": 8.676401564537159,
-      "grad_norm": 1.3133180141448975,
-      "learning_rate": 8.378693467336683e-05,
-      "loss": 5.705,
-      "step": 16637
-    },
-    {
-      "epoch": 8.676923076923076,
-      "grad_norm": 1.3691469430923462,
-      "learning_rate": 8.378592964824121e-05,
-      "loss": 5.7805,
-      "step": 16638
-    },
-    {
-      "epoch": 8.677444589308996,
-      "grad_norm": 1.362877368927002,
-      "learning_rate": 8.378492462311558e-05,
-      "loss": 5.6339,
-      "step": 16639
-    },
-    {
-      "epoch": 8.677966101694915,
-      "grad_norm": 1.380416989326477,
-      "learning_rate": 8.378391959798995e-05,
-      "loss": 5.6602,
-      "step": 16640
-    },
-    {
-      "epoch": 8.678487614080835,
-      "grad_norm": 1.470716953277588,
-      "learning_rate": 8.378291457286432e-05,
-      "loss": 5.5072,
-      "step": 16641
-    },
-    {
-      "epoch": 8.679009126466754,
-      "grad_norm": 1.3453530073165894,
-      "learning_rate": 8.37819095477387e-05,
-      "loss": 5.8824,
-      "step": 16642
-    },
-    {
-      "epoch": 8.679530638852672,
-      "grad_norm": 1.406777024269104,
-      "learning_rate": 8.378090452261307e-05,
-      "loss": 5.8271,
-      "step": 16643
-    },
-    {
-      "epoch": 8.680052151238591,
-      "grad_norm": 1.4406081438064575,
-      "learning_rate": 8.377989949748745e-05,
-      "loss": 5.559,
-      "step": 16644
-    },
-    {
-      "epoch": 8.68057366362451,
-      "grad_norm": 1.3449913263320923,
-      "learning_rate": 8.377889447236182e-05,
-      "loss": 5.802,
-      "step": 16645
-    },
-    {
-      "epoch": 8.68109517601043,
-      "grad_norm": 1.348537802696228,
-      "learning_rate": 8.377788944723618e-05,
-      "loss": 5.8322,
-      "step": 16646
-    },
-    {
-      "epoch": 8.68161668839635,
-      "grad_norm": 1.546535611152649,
-      "learning_rate": 8.377688442211056e-05,
-      "loss": 5.4605,
-      "step": 16647
-    },
-    {
-      "epoch": 8.68213820078227,
-      "grad_norm": 1.3823496103286743,
-      "learning_rate": 8.377587939698492e-05,
-      "loss": 5.834,
-      "step": 16648
-    },
-    {
-      "epoch": 8.682659713168189,
-      "grad_norm": 1.4969735145568848,
-      "learning_rate": 8.37748743718593e-05,
-      "loss": 5.7348,
-      "step": 16649
-    },
-    {
-      "epoch": 8.683181225554106,
-      "grad_norm": 1.5523357391357422,
-      "learning_rate": 8.377386934673366e-05,
-      "loss": 5.3828,
-      "step": 16650
-    },
-    {
-      "epoch": 8.683702737940026,
-      "grad_norm": 1.404862403869629,
-      "learning_rate": 8.377286432160804e-05,
-      "loss": 5.4521,
-      "step": 16651
-    },
-    {
-      "epoch": 8.684224250325945,
-      "grad_norm": 1.3687931299209595,
-      "learning_rate": 8.377185929648241e-05,
-      "loss": 5.8628,
-      "step": 16652
-    },
-    {
-      "epoch": 8.684745762711865,
-      "grad_norm": 1.393282413482666,
-      "learning_rate": 8.377085427135678e-05,
-      "loss": 4.9944,
-      "step": 16653
-    },
-    {
-      "epoch": 8.685267275097784,
-      "grad_norm": 1.4880640506744385,
-      "learning_rate": 8.376984924623116e-05,
-      "loss": 5.2984,
-      "step": 16654
-    },
-    {
-      "epoch": 8.685788787483702,
-      "grad_norm": 1.565285563468933,
-      "learning_rate": 8.376884422110554e-05,
-      "loss": 4.7236,
-      "step": 16655
-    },
-    {
-      "epoch": 8.686310299869621,
-      "grad_norm": 1.3942184448242188,
-      "learning_rate": 8.37678391959799e-05,
-      "loss": 5.6128,
-      "step": 16656
-    },
-    {
-      "epoch": 8.68683181225554,
-      "grad_norm": 1.3721123933792114,
-      "learning_rate": 8.376683417085428e-05,
-      "loss": 5.9228,
-      "step": 16657
-    },
-    {
-      "epoch": 8.68735332464146,
-      "grad_norm": 1.5190467834472656,
-      "learning_rate": 8.376582914572865e-05,
-      "loss": 5.3673,
-      "step": 16658
-    },
-    {
-      "epoch": 8.68787483702738,
-      "grad_norm": 1.3845018148422241,
-      "learning_rate": 8.376482412060301e-05,
-      "loss": 5.6038,
-      "step": 16659
-    },
-    {
-      "epoch": 8.6883963494133,
-      "grad_norm": 1.3710434436798096,
-      "learning_rate": 8.376381909547739e-05,
-      "loss": 4.7887,
-      "step": 16660
-    },
-    {
-      "epoch": 8.688917861799219,
-      "grad_norm": 1.3422789573669434,
-      "learning_rate": 8.376281407035175e-05,
-      "loss": 5.8356,
-      "step": 16661
-    },
-    {
-      "epoch": 8.689439374185136,
-      "grad_norm": 1.3422796726226807,
-      "learning_rate": 8.376180904522613e-05,
-      "loss": 5.8807,
-      "step": 16662
-    },
-    {
-      "epoch": 8.689960886571056,
-      "grad_norm": 1.3242710828781128,
-      "learning_rate": 8.376080402010051e-05,
-      "loss": 5.7034,
-      "step": 16663
-    },
-    {
-      "epoch": 8.690482398956975,
-      "grad_norm": 1.6194626092910767,
-      "learning_rate": 8.375979899497489e-05,
-      "loss": 5.4305,
-      "step": 16664
-    },
-    {
-      "epoch": 8.691003911342895,
-      "grad_norm": 1.565456509590149,
-      "learning_rate": 8.375879396984925e-05,
-      "loss": 5.7412,
-      "step": 16665
-    },
-    {
-      "epoch": 8.691525423728814,
-      "grad_norm": 1.3932210206985474,
-      "learning_rate": 8.375778894472363e-05,
-      "loss": 5.587,
-      "step": 16666
-    },
-    {
-      "epoch": 8.692046936114732,
-      "grad_norm": 1.6675975322723389,
-      "learning_rate": 8.375678391959799e-05,
-      "loss": 5.4927,
-      "step": 16667
-    },
-    {
-      "epoch": 8.692568448500651,
-      "grad_norm": 1.529707431793213,
-      "learning_rate": 8.375577889447237e-05,
-      "loss": 4.7833,
-      "step": 16668
-    },
-    {
-      "epoch": 8.69308996088657,
-      "grad_norm": 1.6564233303070068,
-      "learning_rate": 8.375477386934674e-05,
-      "loss": 5.2622,
-      "step": 16669
-    },
-    {
-      "epoch": 8.69361147327249,
-      "grad_norm": 1.4605923891067505,
-      "learning_rate": 8.375376884422111e-05,
-      "loss": 5.4845,
-      "step": 16670
-    },
-    {
-      "epoch": 8.69413298565841,
-      "grad_norm": 1.3680305480957031,
-      "learning_rate": 8.375276381909548e-05,
-      "loss": 5.8573,
-      "step": 16671
-    },
-    {
-      "epoch": 8.69465449804433,
-      "grad_norm": 1.4333151578903198,
-      "learning_rate": 8.375175879396984e-05,
-      "loss": 6.0871,
-      "step": 16672
-    },
-    {
-      "epoch": 8.695176010430247,
-      "grad_norm": 1.4516117572784424,
-      "learning_rate": 8.375075376884422e-05,
-      "loss": 5.1747,
-      "step": 16673
-    },
-    {
-      "epoch": 8.695697522816166,
-      "grad_norm": 1.4465407133102417,
-      "learning_rate": 8.37497487437186e-05,
-      "loss": 5.7295,
-      "step": 16674
-    },
-    {
-      "epoch": 8.696219035202086,
-      "grad_norm": 1.3650143146514893,
-      "learning_rate": 8.374874371859298e-05,
-      "loss": 5.5376,
-      "step": 16675
-    },
-    {
-      "epoch": 8.696740547588005,
-      "grad_norm": 1.3408390283584595,
-      "learning_rate": 8.374773869346734e-05,
-      "loss": 5.858,
-      "step": 16676
-    },
-    {
-      "epoch": 8.697262059973925,
-      "grad_norm": 1.3400743007659912,
-      "learning_rate": 8.374673366834172e-05,
-      "loss": 5.7518,
-      "step": 16677
-    },
-    {
-      "epoch": 8.697783572359844,
-      "grad_norm": 1.3094632625579834,
-      "learning_rate": 8.374572864321608e-05,
-      "loss": 5.697,
-      "step": 16678
-    },
-    {
-      "epoch": 8.698305084745762,
-      "grad_norm": 1.4265309572219849,
-      "learning_rate": 8.374472361809046e-05,
-      "loss": 5.5808,
-      "step": 16679
-    },
-    {
-      "epoch": 8.698826597131681,
-      "grad_norm": 1.5020571947097778,
-      "learning_rate": 8.374371859296482e-05,
-      "loss": 5.305,
-      "step": 16680
-    },
-    {
-      "epoch": 8.6993481095176,
-      "grad_norm": 1.4864927530288696,
-      "learning_rate": 8.37427135678392e-05,
-      "loss": 5.3129,
-      "step": 16681
-    },
-    {
-      "epoch": 8.69986962190352,
-      "grad_norm": 1.3416062593460083,
-      "learning_rate": 8.374170854271357e-05,
-      "loss": 5.7326,
-      "step": 16682
-    },
-    {
-      "epoch": 8.70039113428944,
-      "grad_norm": 1.530544400215149,
-      "learning_rate": 8.374070351758794e-05,
-      "loss": 5.1403,
-      "step": 16683
-    },
-    {
-      "epoch": 8.70091264667536,
-      "grad_norm": 1.2932556867599487,
-      "learning_rate": 8.373969849246232e-05,
-      "loss": 5.6282,
-      "step": 16684
-    },
-    {
-      "epoch": 8.701434159061277,
-      "grad_norm": 1.4915558099746704,
-      "learning_rate": 8.37386934673367e-05,
-      "loss": 5.6674,
-      "step": 16685
-    },
-    {
-      "epoch": 8.701955671447196,
-      "grad_norm": 1.4061027765274048,
-      "learning_rate": 8.373768844221106e-05,
-      "loss": 5.6555,
-      "step": 16686
-    },
-    {
-      "epoch": 8.702477183833116,
-      "grad_norm": 1.2282124757766724,
-      "learning_rate": 8.373668341708543e-05,
-      "loss": 5.1559,
-      "step": 16687
-    },
-    {
-      "epoch": 8.702998696219035,
-      "grad_norm": 1.5652388334274292,
-      "learning_rate": 8.37356783919598e-05,
-      "loss": 5.4897,
-      "step": 16688
-    },
-    {
-      "epoch": 8.703520208604955,
-      "grad_norm": 1.5356403589248657,
-      "learning_rate": 8.373467336683417e-05,
-      "loss": 5.7543,
-      "step": 16689
-    },
-    {
-      "epoch": 8.704041720990874,
-      "grad_norm": 1.4982415437698364,
-      "learning_rate": 8.373366834170855e-05,
-      "loss": 5.8602,
-      "step": 16690
-    },
-    {
-      "epoch": 8.704563233376792,
-      "grad_norm": 1.5782814025878906,
-      "learning_rate": 8.373266331658291e-05,
-      "loss": 5.4121,
-      "step": 16691
-    },
-    {
-      "epoch": 8.705084745762711,
-      "grad_norm": 1.6122368574142456,
-      "learning_rate": 8.373165829145729e-05,
-      "loss": 5.1218,
-      "step": 16692
-    },
-    {
-      "epoch": 8.70560625814863,
-      "grad_norm": 1.4120765924453735,
-      "learning_rate": 8.373065326633165e-05,
-      "loss": 5.5466,
-      "step": 16693
-    },
-    {
-      "epoch": 8.70612777053455,
-      "grad_norm": 1.5860767364501953,
-      "learning_rate": 8.372964824120603e-05,
-      "loss": 5.7217,
-      "step": 16694
-    },
-    {
-      "epoch": 8.70664928292047,
-      "grad_norm": 1.471635341644287,
-      "learning_rate": 8.372864321608041e-05,
-      "loss": 5.7475,
-      "step": 16695
-    },
-    {
-      "epoch": 8.70717079530639,
-      "grad_norm": 1.5463684797286987,
-      "learning_rate": 8.372763819095479e-05,
-      "loss": 5.4705,
-      "step": 16696
-    },
-    {
-      "epoch": 8.707692307692307,
-      "grad_norm": 1.5364494323730469,
-      "learning_rate": 8.372663316582915e-05,
-      "loss": 5.5806,
-      "step": 16697
-    },
-    {
-      "epoch": 8.708213820078226,
-      "grad_norm": 1.5216912031173706,
-      "learning_rate": 8.372562814070353e-05,
-      "loss": 5.293,
-      "step": 16698
-    },
-    {
-      "epoch": 8.708735332464146,
-      "grad_norm": 1.46867835521698,
-      "learning_rate": 8.37246231155779e-05,
-      "loss": 5.7486,
-      "step": 16699
-    },
-    {
-      "epoch": 8.709256844850065,
-      "grad_norm": 1.4986095428466797,
-      "learning_rate": 8.372361809045226e-05,
-      "loss": 5.6637,
-      "step": 16700
-    },
-    {
-      "epoch": 8.709778357235985,
-      "grad_norm": 1.4553935527801514,
-      "learning_rate": 8.372261306532664e-05,
-      "loss": 5.3883,
-      "step": 16701
-    },
-    {
-      "epoch": 8.710299869621904,
-      "grad_norm": 1.3768388032913208,
-      "learning_rate": 8.3721608040201e-05,
-      "loss": 5.6687,
-      "step": 16702
-    },
-    {
-      "epoch": 8.710821382007822,
-      "grad_norm": 1.4736624956130981,
-      "learning_rate": 8.372060301507538e-05,
-      "loss": 5.7795,
-      "step": 16703
-    },
-    {
-      "epoch": 8.711342894393741,
-      "grad_norm": 1.6664537191390991,
-      "learning_rate": 8.371959798994976e-05,
-      "loss": 5.4556,
-      "step": 16704
-    },
-    {
-      "epoch": 8.711864406779661,
-      "grad_norm": 1.3951318264007568,
-      "learning_rate": 8.371859296482413e-05,
-      "loss": 5.6392,
-      "step": 16705
-    },
-    {
-      "epoch": 8.71238591916558,
-      "grad_norm": 1.4170111417770386,
-      "learning_rate": 8.37175879396985e-05,
-      "loss": 5.444,
-      "step": 16706
-    },
-    {
-      "epoch": 8.7129074315515,
-      "grad_norm": 1.4265649318695068,
-      "learning_rate": 8.371658291457288e-05,
-      "loss": 5.3132,
-      "step": 16707
-    },
-    {
-      "epoch": 8.71342894393742,
-      "grad_norm": 1.3890067338943481,
-      "learning_rate": 8.371557788944724e-05,
-      "loss": 5.1139,
-      "step": 16708
-    },
-    {
-      "epoch": 8.713950456323337,
-      "grad_norm": 1.4727482795715332,
-      "learning_rate": 8.371457286432162e-05,
-      "loss": 5.129,
-      "step": 16709
-    },
-    {
-      "epoch": 8.714471968709256,
-      "grad_norm": 1.387488603591919,
-      "learning_rate": 8.371356783919598e-05,
-      "loss": 5.2465,
-      "step": 16710
-    },
-    {
-      "epoch": 8.714993481095176,
-      "grad_norm": 1.515058994293213,
-      "learning_rate": 8.371256281407036e-05,
-      "loss": 5.4956,
-      "step": 16711
-    },
-    {
-      "epoch": 8.715514993481095,
-      "grad_norm": 1.3480738401412964,
-      "learning_rate": 8.371155778894472e-05,
-      "loss": 5.707,
-      "step": 16712
-    },
-    {
-      "epoch": 8.716036505867015,
-      "grad_norm": 1.4705146551132202,
-      "learning_rate": 8.371055276381909e-05,
-      "loss": 5.0665,
-      "step": 16713
-    },
-    {
-      "epoch": 8.716558018252934,
-      "grad_norm": 1.3832088708877563,
-      "learning_rate": 8.370954773869347e-05,
-      "loss": 5.4319,
-      "step": 16714
-    },
-    {
-      "epoch": 8.717079530638852,
-      "grad_norm": 1.373336911201477,
-      "learning_rate": 8.370854271356784e-05,
-      "loss": 5.9443,
-      "step": 16715
-    },
-    {
-      "epoch": 8.717601043024771,
-      "grad_norm": 1.3842707872390747,
-      "learning_rate": 8.370753768844222e-05,
-      "loss": 5.8435,
-      "step": 16716
-    },
-    {
-      "epoch": 8.718122555410691,
-      "grad_norm": 1.469234824180603,
-      "learning_rate": 8.370653266331659e-05,
-      "loss": 5.3407,
-      "step": 16717
-    },
-    {
-      "epoch": 8.71864406779661,
-      "grad_norm": 1.4248286485671997,
-      "learning_rate": 8.370552763819096e-05,
-      "loss": 5.547,
-      "step": 16718
-    },
-    {
-      "epoch": 8.71916558018253,
-      "grad_norm": 1.7512394189834595,
-      "learning_rate": 8.370452261306533e-05,
-      "loss": 5.2079,
-      "step": 16719
-    },
-    {
-      "epoch": 8.71968709256845,
-      "grad_norm": 1.4802579879760742,
-      "learning_rate": 8.37035175879397e-05,
-      "loss": 5.4379,
-      "step": 16720
-    },
-    {
-      "epoch": 8.720208604954367,
-      "grad_norm": 1.4202677011489868,
-      "learning_rate": 8.370251256281407e-05,
-      "loss": 5.8019,
-      "step": 16721
-    },
-    {
-      "epoch": 8.720730117340286,
-      "grad_norm": 1.4714112281799316,
-      "learning_rate": 8.370150753768845e-05,
-      "loss": 5.6391,
-      "step": 16722
-    },
-    {
-      "epoch": 8.721251629726206,
-      "grad_norm": 1.4480019807815552,
-      "learning_rate": 8.370050251256281e-05,
-      "loss": 5.2298,
-      "step": 16723
-    },
-    {
-      "epoch": 8.721773142112125,
-      "grad_norm": 1.3937593698501587,
-      "learning_rate": 8.369949748743719e-05,
-      "loss": 5.4182,
-      "step": 16724
-    },
-    {
-      "epoch": 8.722294654498045,
-      "grad_norm": 1.4685941934585571,
-      "learning_rate": 8.369849246231157e-05,
-      "loss": 5.4653,
-      "step": 16725
-    },
-    {
-      "epoch": 8.722816166883963,
-      "grad_norm": 1.3177833557128906,
-      "learning_rate": 8.369748743718593e-05,
-      "loss": 5.7544,
-      "step": 16726
-    },
-    {
-      "epoch": 8.723337679269882,
-      "grad_norm": 1.4430752992630005,
-      "learning_rate": 8.369648241206031e-05,
-      "loss": 5.6831,
-      "step": 16727
-    },
-    {
-      "epoch": 8.723859191655801,
-      "grad_norm": 1.4679450988769531,
-      "learning_rate": 8.369547738693467e-05,
-      "loss": 5.1022,
-      "step": 16728
-    },
-    {
-      "epoch": 8.724380704041721,
-      "grad_norm": 1.4259769916534424,
-      "learning_rate": 8.369447236180905e-05,
-      "loss": 5.7719,
-      "step": 16729
-    },
-    {
-      "epoch": 8.72490221642764,
-      "grad_norm": 1.3920800685882568,
-      "learning_rate": 8.369346733668342e-05,
-      "loss": 5.8762,
-      "step": 16730
-    },
-    {
-      "epoch": 8.72542372881356,
-      "grad_norm": 1.4668123722076416,
-      "learning_rate": 8.36924623115578e-05,
-      "loss": 5.2653,
-      "step": 16731
-    },
-    {
-      "epoch": 8.72594524119948,
-      "grad_norm": 1.3466932773590088,
-      "learning_rate": 8.369145728643216e-05,
-      "loss": 5.6474,
-      "step": 16732
-    },
-    {
-      "epoch": 8.726466753585397,
-      "grad_norm": 1.4934674501419067,
-      "learning_rate": 8.369045226130654e-05,
-      "loss": 5.0215,
-      "step": 16733
-    },
-    {
-      "epoch": 8.726988265971316,
-      "grad_norm": 1.4377673864364624,
-      "learning_rate": 8.36894472361809e-05,
-      "loss": 5.0752,
-      "step": 16734
-    },
-    {
-      "epoch": 8.727509778357236,
-      "grad_norm": 1.3366882801055908,
-      "learning_rate": 8.368844221105528e-05,
-      "loss": 5.8394,
-      "step": 16735
-    },
-    {
-      "epoch": 8.728031290743155,
-      "grad_norm": 1.4576045274734497,
-      "learning_rate": 8.368743718592966e-05,
-      "loss": 5.136,
-      "step": 16736
-    },
-    {
-      "epoch": 8.728552803129075,
-      "grad_norm": 1.4368394613265991,
-      "learning_rate": 8.368643216080403e-05,
-      "loss": 5.8123,
-      "step": 16737
-    },
-    {
-      "epoch": 8.729074315514993,
-      "grad_norm": 1.4711084365844727,
-      "learning_rate": 8.36854271356784e-05,
-      "loss": 5.5184,
-      "step": 16738
-    },
-    {
-      "epoch": 8.729595827900912,
-      "grad_norm": 1.5634417533874512,
-      "learning_rate": 8.368442211055276e-05,
-      "loss": 5.5109,
-      "step": 16739
-    },
-    {
-      "epoch": 8.730117340286832,
-      "grad_norm": 1.3339828252792358,
-      "learning_rate": 8.368341708542714e-05,
-      "loss": 5.5781,
-      "step": 16740
-    },
-    {
-      "epoch": 8.730638852672751,
-      "grad_norm": 1.352977991104126,
-      "learning_rate": 8.36824120603015e-05,
-      "loss": 5.7356,
-      "step": 16741
-    },
-    {
-      "epoch": 8.73116036505867,
-      "grad_norm": 1.5020301342010498,
-      "learning_rate": 8.368140703517588e-05,
-      "loss": 5.5498,
-      "step": 16742
-    },
-    {
-      "epoch": 8.73168187744459,
-      "grad_norm": 1.383081078529358,
-      "learning_rate": 8.368040201005025e-05,
-      "loss": 5.9311,
-      "step": 16743
-    },
-    {
-      "epoch": 8.73220338983051,
-      "grad_norm": 1.4595805406570435,
-      "learning_rate": 8.367939698492463e-05,
-      "loss": 5.4031,
-      "step": 16744
-    },
-    {
-      "epoch": 8.732724902216427,
-      "grad_norm": 1.3481124639511108,
-      "learning_rate": 8.3678391959799e-05,
-      "loss": 5.8381,
-      "step": 16745
-    },
-    {
-      "epoch": 8.733246414602347,
-      "grad_norm": 1.5753884315490723,
-      "learning_rate": 8.367738693467338e-05,
-      "loss": 5.5978,
-      "step": 16746
-    },
-    {
-      "epoch": 8.733767926988266,
-      "grad_norm": 1.4132298231124878,
-      "learning_rate": 8.367638190954775e-05,
-      "loss": 5.4811,
-      "step": 16747
-    },
-    {
-      "epoch": 8.734289439374185,
-      "grad_norm": 1.4920029640197754,
-      "learning_rate": 8.367537688442212e-05,
-      "loss": 5.7596,
-      "step": 16748
-    },
-    {
-      "epoch": 8.734810951760105,
-      "grad_norm": 1.5488255023956299,
-      "learning_rate": 8.367437185929649e-05,
-      "loss": 5.353,
-      "step": 16749
-    },
-    {
-      "epoch": 8.735332464146023,
-      "grad_norm": 1.4874367713928223,
-      "learning_rate": 8.367336683417086e-05,
-      "loss": 5.3531,
-      "step": 16750
-    },
-    {
-      "epoch": 8.735853976531942,
-      "grad_norm": 1.3697317838668823,
-      "learning_rate": 8.367236180904523e-05,
-      "loss": 5.5202,
-      "step": 16751
-    },
-    {
-      "epoch": 8.736375488917862,
-      "grad_norm": 1.5995389223098755,
-      "learning_rate": 8.36713567839196e-05,
-      "loss": 5.307,
-      "step": 16752
-    },
-    {
-      "epoch": 8.736897001303781,
-      "grad_norm": 1.3747880458831787,
-      "learning_rate": 8.367035175879397e-05,
-      "loss": 5.2946,
-      "step": 16753
-    },
-    {
-      "epoch": 8.7374185136897,
-      "grad_norm": 1.4809263944625854,
-      "learning_rate": 8.366934673366834e-05,
-      "loss": 5.6079,
-      "step": 16754
-    },
-    {
-      "epoch": 8.73794002607562,
-      "grad_norm": 1.4195374250411987,
-      "learning_rate": 8.366834170854271e-05,
-      "loss": 5.3515,
-      "step": 16755
-    },
-    {
-      "epoch": 8.73846153846154,
-      "grad_norm": 1.4479291439056396,
-      "learning_rate": 8.366733668341709e-05,
-      "loss": 5.2242,
-      "step": 16756
-    },
-    {
-      "epoch": 8.738983050847457,
-      "grad_norm": 1.4068193435668945,
-      "learning_rate": 8.366633165829147e-05,
-      "loss": 5.4452,
-      "step": 16757
-    },
-    {
-      "epoch": 8.739504563233377,
-      "grad_norm": 1.4583686590194702,
-      "learning_rate": 8.366532663316583e-05,
-      "loss": 5.5487,
-      "step": 16758
-    },
-    {
-      "epoch": 8.740026075619296,
-      "grad_norm": 1.3693798780441284,
-      "learning_rate": 8.366432160804021e-05,
-      "loss": 5.1096,
-      "step": 16759
-    },
-    {
-      "epoch": 8.740547588005215,
-      "grad_norm": 1.4219781160354614,
-      "learning_rate": 8.366331658291458e-05,
-      "loss": 5.6573,
-      "step": 16760
-    },
-    {
-      "epoch": 8.741069100391135,
-      "grad_norm": 1.3781505823135376,
-      "learning_rate": 8.366231155778895e-05,
-      "loss": 5.5478,
-      "step": 16761
-    },
-    {
-      "epoch": 8.741590612777053,
-      "grad_norm": 1.506433367729187,
-      "learning_rate": 8.366130653266332e-05,
-      "loss": 5.4081,
-      "step": 16762
-    },
-    {
-      "epoch": 8.742112125162972,
-      "grad_norm": 1.355911374092102,
-      "learning_rate": 8.36603015075377e-05,
-      "loss": 5.1731,
-      "step": 16763
-    },
-    {
-      "epoch": 8.742633637548892,
-      "grad_norm": 1.3465626239776611,
-      "learning_rate": 8.365929648241206e-05,
-      "loss": 5.8523,
-      "step": 16764
-    },
-    {
-      "epoch": 8.743155149934811,
-      "grad_norm": 1.2796212434768677,
-      "learning_rate": 8.365829145728642e-05,
-      "loss": 5.3959,
-      "step": 16765
-    },
-    {
-      "epoch": 8.74367666232073,
-      "grad_norm": 1.3779938220977783,
-      "learning_rate": 8.36572864321608e-05,
-      "loss": 5.4638,
-      "step": 16766
-    },
-    {
-      "epoch": 8.74419817470665,
-      "grad_norm": 1.3357949256896973,
-      "learning_rate": 8.365628140703518e-05,
-      "loss": 5.7988,
-      "step": 16767
-    },
-    {
-      "epoch": 8.744719687092568,
-      "grad_norm": 1.4129763841629028,
-      "learning_rate": 8.365527638190956e-05,
-      "loss": 5.6914,
-      "step": 16768
-    },
-    {
-      "epoch": 8.745241199478487,
-      "grad_norm": 1.4481457471847534,
-      "learning_rate": 8.365427135678392e-05,
-      "loss": 4.9778,
-      "step": 16769
-    },
-    {
-      "epoch": 8.745762711864407,
-      "grad_norm": 1.4216219186782837,
-      "learning_rate": 8.36532663316583e-05,
-      "loss": 5.6847,
-      "step": 16770
-    },
-    {
-      "epoch": 8.746284224250326,
-      "grad_norm": 1.3897520303726196,
-      "learning_rate": 8.365226130653266e-05,
-      "loss": 5.8577,
-      "step": 16771
-    },
-    {
-      "epoch": 8.746805736636245,
-      "grad_norm": 1.412482738494873,
-      "learning_rate": 8.365125628140704e-05,
-      "loss": 5.5371,
-      "step": 16772
-    },
-    {
-      "epoch": 8.747327249022165,
-      "grad_norm": 1.2854068279266357,
-      "learning_rate": 8.36502512562814e-05,
-      "loss": 5.7943,
-      "step": 16773
-    },
-    {
-      "epoch": 8.747848761408083,
-      "grad_norm": 1.350219488143921,
-      "learning_rate": 8.364924623115578e-05,
-      "loss": 5.9674,
-      "step": 16774
-    },
-    {
-      "epoch": 8.748370273794002,
-      "grad_norm": 1.3814570903778076,
-      "learning_rate": 8.364824120603015e-05,
-      "loss": 5.7001,
-      "step": 16775
-    },
-    {
-      "epoch": 8.748891786179922,
-      "grad_norm": 1.4835392236709595,
-      "learning_rate": 8.364723618090453e-05,
-      "loss": 5.1368,
-      "step": 16776
-    },
-    {
-      "epoch": 8.749413298565841,
-      "grad_norm": 1.3639581203460693,
-      "learning_rate": 8.36462311557789e-05,
-      "loss": 5.3531,
-      "step": 16777
-    },
-    {
-      "epoch": 8.74993481095176,
-      "grad_norm": 1.427730679512024,
-      "learning_rate": 8.364522613065328e-05,
-      "loss": 5.6159,
-      "step": 16778
-    },
-    {
-      "epoch": 8.75045632333768,
-      "grad_norm": 1.382274866104126,
-      "learning_rate": 8.364422110552765e-05,
-      "loss": 5.496,
-      "step": 16779
-    },
-    {
-      "epoch": 8.750977835723598,
-      "grad_norm": 1.4300868511199951,
-      "learning_rate": 8.364321608040201e-05,
-      "loss": 5.7387,
-      "step": 16780
-    },
-    {
-      "epoch": 8.751499348109517,
-      "grad_norm": 1.4523026943206787,
-      "learning_rate": 8.364221105527639e-05,
-      "loss": 5.6092,
-      "step": 16781
-    },
-    {
-      "epoch": 8.752020860495437,
-      "grad_norm": 1.456408143043518,
-      "learning_rate": 8.364120603015075e-05,
-      "loss": 5.318,
-      "step": 16782
-    },
-    {
-      "epoch": 8.752542372881356,
-      "grad_norm": 1.4793192148208618,
-      "learning_rate": 8.364020100502513e-05,
-      "loss": 5.8044,
-      "step": 16783
-    },
-    {
-      "epoch": 8.753063885267276,
-      "grad_norm": 1.406471848487854,
-      "learning_rate": 8.36391959798995e-05,
-      "loss": 5.6701,
-      "step": 16784
-    },
-    {
-      "epoch": 8.753585397653195,
-      "grad_norm": 1.3408664464950562,
-      "learning_rate": 8.363819095477387e-05,
-      "loss": 6.0207,
-      "step": 16785
-    },
-    {
-      "epoch": 8.754106910039113,
-      "grad_norm": 1.3876365423202515,
-      "learning_rate": 8.363718592964824e-05,
-      "loss": 5.67,
-      "step": 16786
-    },
-    {
-      "epoch": 8.754628422425032,
-      "grad_norm": 1.465591549873352,
-      "learning_rate": 8.363618090452261e-05,
-      "loss": 5.6,
-      "step": 16787
-    },
-    {
-      "epoch": 8.755149934810952,
-      "grad_norm": 1.5773639678955078,
-      "learning_rate": 8.363517587939699e-05,
-      "loss": 5.6216,
-      "step": 16788
-    },
-    {
-      "epoch": 8.755671447196871,
-      "grad_norm": 1.3617335557937622,
-      "learning_rate": 8.363417085427137e-05,
-      "loss": 5.6771,
-      "step": 16789
-    },
-    {
-      "epoch": 8.75619295958279,
-      "grad_norm": 1.5147019624710083,
-      "learning_rate": 8.363316582914573e-05,
-      "loss": 5.5599,
-      "step": 16790
-    },
-    {
-      "epoch": 8.75671447196871,
-      "grad_norm": 1.590320110321045,
-      "learning_rate": 8.363216080402011e-05,
-      "loss": 5.4066,
-      "step": 16791
-    },
-    {
-      "epoch": 8.757235984354628,
-      "grad_norm": 1.530150294303894,
-      "learning_rate": 8.363115577889448e-05,
-      "loss": 4.8645,
-      "step": 16792
-    },
-    {
-      "epoch": 8.757757496740547,
-      "grad_norm": 1.52835214138031,
-      "learning_rate": 8.363015075376884e-05,
-      "loss": 5.376,
-      "step": 16793
-    },
-    {
-      "epoch": 8.758279009126467,
-      "grad_norm": 1.4662717580795288,
-      "learning_rate": 8.362914572864322e-05,
-      "loss": 5.2965,
-      "step": 16794
-    },
-    {
-      "epoch": 8.758800521512386,
-      "grad_norm": 1.5479800701141357,
-      "learning_rate": 8.362814070351758e-05,
-      "loss": 5.2467,
-      "step": 16795
-    },
-    {
-      "epoch": 8.759322033898306,
-      "grad_norm": 1.5133029222488403,
-      "learning_rate": 8.362713567839196e-05,
-      "loss": 5.3545,
-      "step": 16796
-    },
-    {
-      "epoch": 8.759843546284225,
-      "grad_norm": 1.3063569068908691,
-      "learning_rate": 8.362613065326634e-05,
-      "loss": 5.7658,
-      "step": 16797
-    },
-    {
-      "epoch": 8.760365058670143,
-      "grad_norm": 1.4074227809906006,
-      "learning_rate": 8.362512562814072e-05,
-      "loss": 5.8603,
-      "step": 16798
-    },
-    {
-      "epoch": 8.760886571056062,
-      "grad_norm": 1.5554444789886475,
-      "learning_rate": 8.362412060301508e-05,
-      "loss": 5.6723,
-      "step": 16799
-    },
-    {
-      "epoch": 8.761408083441982,
-      "grad_norm": 1.5207096338272095,
-      "learning_rate": 8.362311557788946e-05,
-      "loss": 5.6313,
-      "step": 16800
-    },
-    {
-      "epoch": 8.761929595827901,
-      "grad_norm": 1.3871796131134033,
-      "learning_rate": 8.362211055276382e-05,
-      "loss": 5.7474,
-      "step": 16801
-    },
-    {
-      "epoch": 8.76245110821382,
-      "grad_norm": 1.2831425666809082,
-      "learning_rate": 8.36211055276382e-05,
-      "loss": 6.0772,
-      "step": 16802
-    },
-    {
-      "epoch": 8.76297262059974,
-      "grad_norm": 1.3616855144500732,
-      "learning_rate": 8.362010050251256e-05,
-      "loss": 5.5427,
-      "step": 16803
-    },
-    {
-      "epoch": 8.763494132985658,
-      "grad_norm": 1.3339133262634277,
-      "learning_rate": 8.361909547738694e-05,
-      "loss": 6.0597,
-      "step": 16804
-    },
-    {
-      "epoch": 8.764015645371577,
-      "grad_norm": 1.4548790454864502,
-      "learning_rate": 8.36180904522613e-05,
-      "loss": 5.634,
-      "step": 16805
-    },
-    {
-      "epoch": 8.764537157757497,
-      "grad_norm": 1.4748520851135254,
-      "learning_rate": 8.361708542713567e-05,
-      "loss": 5.3476,
-      "step": 16806
-    },
-    {
-      "epoch": 8.765058670143416,
-      "grad_norm": 1.413817286491394,
-      "learning_rate": 8.361608040201005e-05,
-      "loss": 5.3578,
-      "step": 16807
-    },
-    {
-      "epoch": 8.765580182529336,
-      "grad_norm": 1.5144597291946411,
-      "learning_rate": 8.361507537688443e-05,
-      "loss": 5.2415,
-      "step": 16808
-    },
-    {
-      "epoch": 8.766101694915255,
-      "grad_norm": 1.406569242477417,
-      "learning_rate": 8.36140703517588e-05,
-      "loss": 5.2083,
-      "step": 16809
-    },
-    {
-      "epoch": 8.766623207301173,
-      "grad_norm": 1.4216920137405396,
-      "learning_rate": 8.361306532663317e-05,
-      "loss": 5.2735,
-      "step": 16810
-    },
-    {
-      "epoch": 8.767144719687092,
-      "grad_norm": 1.4975683689117432,
-      "learning_rate": 8.361206030150755e-05,
-      "loss": 5.4491,
-      "step": 16811
-    },
-    {
-      "epoch": 8.767666232073012,
-      "grad_norm": 1.5107976198196411,
-      "learning_rate": 8.361105527638191e-05,
-      "loss": 5.0824,
-      "step": 16812
-    },
-    {
-      "epoch": 8.768187744458931,
-      "grad_norm": 1.3963868618011475,
-      "learning_rate": 8.361005025125629e-05,
-      "loss": 5.7164,
-      "step": 16813
-    },
-    {
-      "epoch": 8.76870925684485,
-      "grad_norm": 1.3925755023956299,
-      "learning_rate": 8.360904522613065e-05,
-      "loss": 5.8144,
-      "step": 16814
-    },
-    {
-      "epoch": 8.76923076923077,
-      "grad_norm": 1.4478256702423096,
-      "learning_rate": 8.360804020100503e-05,
-      "loss": 5.3605,
-      "step": 16815
-    },
-    {
-      "epoch": 8.769752281616688,
-      "grad_norm": 1.50880765914917,
-      "learning_rate": 8.36070351758794e-05,
-      "loss": 5.544,
-      "step": 16816
-    },
-    {
-      "epoch": 8.770273794002607,
-      "grad_norm": 1.4451451301574707,
-      "learning_rate": 8.360603015075377e-05,
-      "loss": 5.3065,
-      "step": 16817
-    },
-    {
-      "epoch": 8.770795306388527,
-      "grad_norm": 1.4471367597579956,
-      "learning_rate": 8.360502512562815e-05,
-      "loss": 5.7304,
-      "step": 16818
-    },
-    {
-      "epoch": 8.771316818774446,
-      "grad_norm": 1.5849237442016602,
-      "learning_rate": 8.360402010050251e-05,
-      "loss": 5.2789,
-      "step": 16819
-    },
-    {
-      "epoch": 8.771838331160366,
-      "grad_norm": 1.4914264678955078,
-      "learning_rate": 8.360301507537689e-05,
-      "loss": 5.2538,
-      "step": 16820
-    },
-    {
-      "epoch": 8.772359843546283,
-      "grad_norm": 1.4861900806427002,
-      "learning_rate": 8.360201005025126e-05,
-      "loss": 5.4958,
-      "step": 16821
-    },
-    {
-      "epoch": 8.772881355932203,
-      "grad_norm": 1.508380651473999,
-      "learning_rate": 8.360100502512563e-05,
-      "loss": 5.2876,
-      "step": 16822
-    },
-    {
-      "epoch": 8.773402868318122,
-      "grad_norm": 1.3423089981079102,
-      "learning_rate": 8.36e-05,
-      "loss": 5.6757,
-      "step": 16823
-    },
-    {
-      "epoch": 8.773924380704042,
-      "grad_norm": 1.510724663734436,
-      "learning_rate": 8.359899497487438e-05,
-      "loss": 4.9152,
-      "step": 16824
-    },
-    {
-      "epoch": 8.774445893089961,
-      "grad_norm": 1.474795937538147,
-      "learning_rate": 8.359798994974874e-05,
-      "loss": 5.0595,
-      "step": 16825
-    },
-    {
-      "epoch": 8.77496740547588,
-      "grad_norm": 1.3546725511550903,
-      "learning_rate": 8.359698492462312e-05,
-      "loss": 5.6638,
-      "step": 16826
-    },
-    {
-      "epoch": 8.7754889178618,
-      "grad_norm": 1.5240637063980103,
-      "learning_rate": 8.359597989949748e-05,
-      "loss": 5.3462,
-      "step": 16827
-    },
-    {
-      "epoch": 8.776010430247718,
-      "grad_norm": 1.5037168264389038,
-      "learning_rate": 8.359497487437186e-05,
-      "loss": 4.6578,
-      "step": 16828
-    },
-    {
-      "epoch": 8.776531942633637,
-      "grad_norm": 1.3781753778457642,
-      "learning_rate": 8.359396984924624e-05,
-      "loss": 5.6199,
-      "step": 16829
-    },
-    {
-      "epoch": 8.777053455019557,
-      "grad_norm": 1.2693275213241577,
-      "learning_rate": 8.359296482412062e-05,
-      "loss": 5.761,
-      "step": 16830
-    },
-    {
-      "epoch": 8.777574967405476,
-      "grad_norm": 1.3607226610183716,
-      "learning_rate": 8.359195979899498e-05,
-      "loss": 5.0678,
-      "step": 16831
-    },
-    {
-      "epoch": 8.778096479791396,
-      "grad_norm": 1.5265661478042603,
-      "learning_rate": 8.359095477386935e-05,
-      "loss": 5.0558,
-      "step": 16832
-    },
-    {
-      "epoch": 8.778617992177313,
-      "grad_norm": 1.358443021774292,
-      "learning_rate": 8.358994974874372e-05,
-      "loss": 5.824,
-      "step": 16833
-    },
-    {
-      "epoch": 8.779139504563233,
-      "grad_norm": 1.433332920074463,
-      "learning_rate": 8.358894472361809e-05,
-      "loss": 5.1886,
-      "step": 16834
-    },
-    {
-      "epoch": 8.779661016949152,
-      "grad_norm": 1.7115193605422974,
-      "learning_rate": 8.358793969849247e-05,
-      "loss": 5.2058,
-      "step": 16835
-    },
-    {
-      "epoch": 8.780182529335072,
-      "grad_norm": 1.6949700117111206,
-      "learning_rate": 8.358693467336683e-05,
-      "loss": 5.3628,
-      "step": 16836
-    },
-    {
-      "epoch": 8.780704041720991,
-      "grad_norm": 1.4137567281723022,
-      "learning_rate": 8.358592964824121e-05,
-      "loss": 5.8545,
-      "step": 16837
-    },
-    {
-      "epoch": 8.78122555410691,
-      "grad_norm": 1.4295179843902588,
-      "learning_rate": 8.358492462311559e-05,
-      "loss": 5.478,
-      "step": 16838
-    },
-    {
-      "epoch": 8.78174706649283,
-      "grad_norm": 1.4594531059265137,
-      "learning_rate": 8.358391959798996e-05,
-      "loss": 5.3657,
-      "step": 16839
-    },
-    {
-      "epoch": 8.782268578878748,
-      "grad_norm": 1.3532254695892334,
-      "learning_rate": 8.358291457286433e-05,
-      "loss": 6.0747,
-      "step": 16840
-    },
-    {
-      "epoch": 8.782790091264667,
-      "grad_norm": 1.3369441032409668,
-      "learning_rate": 8.35819095477387e-05,
-      "loss": 5.6178,
-      "step": 16841
-    },
-    {
-      "epoch": 8.783311603650587,
-      "grad_norm": 1.5422906875610352,
-      "learning_rate": 8.358090452261307e-05,
-      "loss": 5.7111,
-      "step": 16842
-    },
-    {
-      "epoch": 8.783833116036506,
-      "grad_norm": 1.5672085285186768,
-      "learning_rate": 8.357989949748745e-05,
-      "loss": 5.0497,
-      "step": 16843
-    },
-    {
-      "epoch": 8.784354628422426,
-      "grad_norm": 1.7414251565933228,
-      "learning_rate": 8.357889447236181e-05,
-      "loss": 5.3783,
-      "step": 16844
-    },
-    {
-      "epoch": 8.784876140808343,
-      "grad_norm": 1.4242902994155884,
-      "learning_rate": 8.357788944723618e-05,
-      "loss": 5.2149,
-      "step": 16845
-    },
-    {
-      "epoch": 8.785397653194263,
-      "grad_norm": 1.6199777126312256,
-      "learning_rate": 8.357688442211055e-05,
-      "loss": 5.0149,
-      "step": 16846
-    },
-    {
-      "epoch": 8.785919165580182,
-      "grad_norm": 1.433197021484375,
-      "learning_rate": 8.357587939698492e-05,
-      "loss": 5.1928,
-      "step": 16847
-    },
-    {
-      "epoch": 8.786440677966102,
-      "grad_norm": 1.3965153694152832,
-      "learning_rate": 8.35748743718593e-05,
-      "loss": 5.7487,
-      "step": 16848
-    },
-    {
-      "epoch": 8.786962190352021,
-      "grad_norm": 1.5306296348571777,
-      "learning_rate": 8.357386934673367e-05,
-      "loss": 5.5879,
-      "step": 16849
-    },
-    {
-      "epoch": 8.78748370273794,
-      "grad_norm": 1.4932234287261963,
-      "learning_rate": 8.357286432160805e-05,
-      "loss": 5.7642,
-      "step": 16850
-    },
-    {
-      "epoch": 8.78800521512386,
-      "grad_norm": 1.5806595087051392,
-      "learning_rate": 8.357185929648242e-05,
-      "loss": 5.5309,
-      "step": 16851
-    },
-    {
-      "epoch": 8.788526727509778,
-      "grad_norm": 1.6302812099456787,
-      "learning_rate": 8.35708542713568e-05,
-      "loss": 5.1454,
-      "step": 16852
-    },
-    {
-      "epoch": 8.789048239895697,
-      "grad_norm": 1.476861834526062,
-      "learning_rate": 8.356984924623116e-05,
-      "loss": 5.4957,
-      "step": 16853
-    },
-    {
-      "epoch": 8.789569752281617,
-      "grad_norm": 1.4508271217346191,
-      "learning_rate": 8.356884422110554e-05,
-      "loss": 5.7729,
-      "step": 16854
-    },
-    {
-      "epoch": 8.790091264667536,
-      "grad_norm": 1.4186556339263916,
-      "learning_rate": 8.35678391959799e-05,
-      "loss": 5.7472,
-      "step": 16855
-    },
-    {
-      "epoch": 8.790612777053456,
-      "grad_norm": 1.5407264232635498,
-      "learning_rate": 8.356683417085428e-05,
-      "loss": 5.3061,
-      "step": 16856
-    },
-    {
-      "epoch": 8.791134289439373,
-      "grad_norm": 1.4556219577789307,
-      "learning_rate": 8.356582914572864e-05,
-      "loss": 5.6796,
-      "step": 16857
-    },
-    {
-      "epoch": 8.791655801825293,
-      "grad_norm": 1.5846318006515503,
-      "learning_rate": 8.356482412060302e-05,
-      "loss": 5.3805,
-      "step": 16858
-    },
-    {
-      "epoch": 8.792177314211212,
-      "grad_norm": 1.413019061088562,
-      "learning_rate": 8.35638190954774e-05,
-      "loss": 5.4452,
-      "step": 16859
-    },
-    {
-      "epoch": 8.792698826597132,
-      "grad_norm": 1.3643913269042969,
-      "learning_rate": 8.356281407035176e-05,
-      "loss": 5.4753,
-      "step": 16860
-    },
-    {
-      "epoch": 8.793220338983051,
-      "grad_norm": 1.8906751871109009,
-      "learning_rate": 8.356180904522614e-05,
-      "loss": 5.295,
-      "step": 16861
-    },
-    {
-      "epoch": 8.79374185136897,
-      "grad_norm": 1.4246219396591187,
-      "learning_rate": 8.35608040201005e-05,
-      "loss": 5.3317,
-      "step": 16862
-    },
-    {
-      "epoch": 8.794263363754888,
-      "grad_norm": 1.7063485383987427,
-      "learning_rate": 8.355979899497488e-05,
-      "loss": 5.1487,
-      "step": 16863
-    },
-    {
-      "epoch": 8.794784876140808,
-      "grad_norm": 1.4483554363250732,
-      "learning_rate": 8.355879396984925e-05,
-      "loss": 5.5261,
-      "step": 16864
-    },
-    {
-      "epoch": 8.795306388526727,
-      "grad_norm": 1.5707721710205078,
-      "learning_rate": 8.355778894472362e-05,
-      "loss": 5.55,
-      "step": 16865
-    },
-    {
-      "epoch": 8.795827900912647,
-      "grad_norm": 1.3515585660934448,
-      "learning_rate": 8.355678391959799e-05,
-      "loss": 5.9651,
-      "step": 16866
-    },
-    {
-      "epoch": 8.796349413298566,
-      "grad_norm": 1.337493658065796,
-      "learning_rate": 8.355577889447237e-05,
-      "loss": 5.5937,
-      "step": 16867
-    },
-    {
-      "epoch": 8.796870925684486,
-      "grad_norm": 1.433868408203125,
-      "learning_rate": 8.355477386934673e-05,
-      "loss": 5.1979,
-      "step": 16868
-    },
-    {
-      "epoch": 8.797392438070403,
-      "grad_norm": 1.3700644969940186,
-      "learning_rate": 8.355376884422111e-05,
-      "loss": 5.7416,
-      "step": 16869
-    },
-    {
-      "epoch": 8.797913950456323,
-      "grad_norm": 1.6479696035385132,
-      "learning_rate": 8.355276381909549e-05,
-      "loss": 5.0886,
-      "step": 16870
-    },
-    {
-      "epoch": 8.798435462842242,
-      "grad_norm": 1.396196961402893,
-      "learning_rate": 8.355175879396986e-05,
-      "loss": 5.4423,
-      "step": 16871
-    },
-    {
-      "epoch": 8.798956975228162,
-      "grad_norm": 1.4713166952133179,
-      "learning_rate": 8.355075376884423e-05,
-      "loss": 5.6392,
-      "step": 16872
-    },
-    {
-      "epoch": 8.799478487614081,
-      "grad_norm": 1.455247402191162,
-      "learning_rate": 8.354974874371859e-05,
-      "loss": 5.1469,
-      "step": 16873
-    },
-    {
-      "epoch": 8.8,
-      "grad_norm": 1.3316524028778076,
-      "learning_rate": 8.354874371859297e-05,
-      "loss": 5.6778,
-      "step": 16874
-    },
-    {
-      "epoch": 8.800521512385918,
-      "grad_norm": 1.4325711727142334,
-      "learning_rate": 8.354773869346733e-05,
-      "loss": 5.1406,
-      "step": 16875
-    },
-    {
-      "epoch": 8.801043024771838,
-      "grad_norm": 1.462689757347107,
-      "learning_rate": 8.354673366834171e-05,
-      "loss": 5.2024,
-      "step": 16876
-    },
-    {
-      "epoch": 8.801564537157757,
-      "grad_norm": 1.358154296875,
-      "learning_rate": 8.354572864321608e-05,
-      "loss": 5.2823,
-      "step": 16877
-    },
-    {
-      "epoch": 8.802086049543677,
-      "grad_norm": 1.4956189393997192,
-      "learning_rate": 8.354472361809045e-05,
-      "loss": 5.3321,
-      "step": 16878
-    },
-    {
-      "epoch": 8.802607561929596,
-      "grad_norm": 1.3645659685134888,
-      "learning_rate": 8.354371859296483e-05,
-      "loss": 5.6986,
-      "step": 16879
-    },
-    {
-      "epoch": 8.803129074315516,
-      "grad_norm": 1.381750226020813,
-      "learning_rate": 8.354271356783921e-05,
-      "loss": 5.5485,
-      "step": 16880
-    },
-    {
-      "epoch": 8.803650586701433,
-      "grad_norm": 1.297747254371643,
-      "learning_rate": 8.354170854271357e-05,
-      "loss": 5.5325,
-      "step": 16881
-    },
-    {
-      "epoch": 8.804172099087353,
-      "grad_norm": 1.3522764444351196,
-      "learning_rate": 8.354070351758795e-05,
-      "loss": 5.539,
-      "step": 16882
-    },
-    {
-      "epoch": 8.804693611473272,
-      "grad_norm": 1.368298053741455,
-      "learning_rate": 8.353969849246232e-05,
-      "loss": 5.9373,
-      "step": 16883
-    },
-    {
-      "epoch": 8.805215123859192,
-      "grad_norm": 1.4929381608963013,
-      "learning_rate": 8.35386934673367e-05,
-      "loss": 5.2503,
-      "step": 16884
-    },
-    {
-      "epoch": 8.805736636245111,
-      "grad_norm": 1.6607733964920044,
-      "learning_rate": 8.353768844221106e-05,
-      "loss": 5.2451,
-      "step": 16885
-    },
-    {
-      "epoch": 8.80625814863103,
-      "grad_norm": 1.3980218172073364,
-      "learning_rate": 8.353668341708542e-05,
-      "loss": 5.6394,
-      "step": 16886
-    },
-    {
-      "epoch": 8.806779661016948,
-      "grad_norm": 2.247128963470459,
-      "learning_rate": 8.35356783919598e-05,
-      "loss": 4.8425,
-      "step": 16887
-    },
-    {
-      "epoch": 8.807301173402868,
-      "grad_norm": 1.5632272958755493,
-      "learning_rate": 8.353467336683416e-05,
-      "loss": 5.2634,
-      "step": 16888
-    },
-    {
-      "epoch": 8.807822685788787,
-      "grad_norm": 1.433674931526184,
-      "learning_rate": 8.353366834170854e-05,
-      "loss": 5.3955,
-      "step": 16889
-    },
-    {
-      "epoch": 8.808344198174707,
-      "grad_norm": 1.3963311910629272,
-      "learning_rate": 8.353266331658292e-05,
-      "loss": 5.6262,
-      "step": 16890
-    },
-    {
-      "epoch": 8.808865710560626,
-      "grad_norm": 1.4912768602371216,
-      "learning_rate": 8.35316582914573e-05,
-      "loss": 5.2931,
-      "step": 16891
-    },
-    {
-      "epoch": 8.809387222946546,
-      "grad_norm": 1.5207792520523071,
-      "learning_rate": 8.353065326633166e-05,
-      "loss": 5.2045,
-      "step": 16892
-    },
-    {
-      "epoch": 8.809908735332463,
-      "grad_norm": 1.3671205043792725,
-      "learning_rate": 8.352964824120604e-05,
-      "loss": 5.5073,
-      "step": 16893
-    },
-    {
-      "epoch": 8.810430247718383,
-      "grad_norm": 1.3994511365890503,
-      "learning_rate": 8.35286432160804e-05,
-      "loss": 5.8148,
-      "step": 16894
-    },
-    {
-      "epoch": 8.810951760104302,
-      "grad_norm": 1.49443519115448,
-      "learning_rate": 8.352763819095478e-05,
-      "loss": 5.4196,
-      "step": 16895
-    },
-    {
-      "epoch": 8.811473272490222,
-      "grad_norm": 1.3969900608062744,
-      "learning_rate": 8.352663316582915e-05,
-      "loss": 5.453,
-      "step": 16896
-    },
-    {
-      "epoch": 8.811994784876141,
-      "grad_norm": 1.392042636871338,
-      "learning_rate": 8.352562814070352e-05,
-      "loss": 5.8502,
-      "step": 16897
-    },
-    {
-      "epoch": 8.81251629726206,
-      "grad_norm": 1.4709107875823975,
-      "learning_rate": 8.352462311557789e-05,
-      "loss": 5.8584,
-      "step": 16898
-    },
-    {
-      "epoch": 8.813037809647978,
-      "grad_norm": 1.3911820650100708,
-      "learning_rate": 8.352361809045227e-05,
-      "loss": 5.337,
-      "step": 16899
-    },
-    {
-      "epoch": 8.813559322033898,
-      "grad_norm": 1.4821738004684448,
-      "learning_rate": 8.352261306532664e-05,
-      "loss": 5.603,
-      "step": 16900
-    },
-    {
-      "epoch": 8.814080834419817,
-      "grad_norm": 1.4141621589660645,
-      "learning_rate": 8.352160804020101e-05,
-      "loss": 5.4516,
-      "step": 16901
-    },
-    {
-      "epoch": 8.814602346805737,
-      "grad_norm": 1.3985174894332886,
-      "learning_rate": 8.352060301507539e-05,
-      "loss": 5.5283,
-      "step": 16902
-    },
-    {
-      "epoch": 8.815123859191656,
-      "grad_norm": 1.4878042936325073,
-      "learning_rate": 8.351959798994975e-05,
-      "loss": 5.084,
-      "step": 16903
-    },
-    {
-      "epoch": 8.815645371577576,
-      "grad_norm": 1.3323835134506226,
-      "learning_rate": 8.351859296482413e-05,
-      "loss": 5.6111,
-      "step": 16904
-    },
-    {
-      "epoch": 8.816166883963493,
-      "grad_norm": 1.53725266456604,
-      "learning_rate": 8.351758793969849e-05,
-      "loss": 5.4393,
-      "step": 16905
-    },
-    {
-      "epoch": 8.816688396349413,
-      "grad_norm": 1.4048932790756226,
-      "learning_rate": 8.351658291457287e-05,
-      "loss": 4.9404,
-      "step": 16906
-    },
-    {
-      "epoch": 8.817209908735332,
-      "grad_norm": 1.3381874561309814,
-      "learning_rate": 8.351557788944724e-05,
-      "loss": 5.4357,
-      "step": 16907
-    },
-    {
-      "epoch": 8.817731421121252,
-      "grad_norm": 1.2154691219329834,
-      "learning_rate": 8.351457286432161e-05,
-      "loss": 4.7563,
-      "step": 16908
-    },
-    {
-      "epoch": 8.818252933507171,
-      "grad_norm": 1.4678230285644531,
-      "learning_rate": 8.351356783919598e-05,
-      "loss": 5.6308,
-      "step": 16909
-    },
-    {
-      "epoch": 8.81877444589309,
-      "grad_norm": 1.4678895473480225,
-      "learning_rate": 8.351256281407036e-05,
-      "loss": 5.117,
-      "step": 16910
-    },
-    {
-      "epoch": 8.819295958279008,
-      "grad_norm": 1.517960786819458,
-      "learning_rate": 8.351155778894473e-05,
-      "loss": 4.9413,
-      "step": 16911
-    },
-    {
-      "epoch": 8.819817470664928,
-      "grad_norm": 1.364450454711914,
-      "learning_rate": 8.35105527638191e-05,
-      "loss": 5.7033,
-      "step": 16912
-    },
-    {
-      "epoch": 8.820338983050847,
-      "grad_norm": 1.479051113128662,
-      "learning_rate": 8.350954773869348e-05,
-      "loss": 5.8627,
-      "step": 16913
-    },
-    {
-      "epoch": 8.820860495436767,
-      "grad_norm": 1.5600918531417847,
-      "learning_rate": 8.350854271356784e-05,
-      "loss": 5.1991,
-      "step": 16914
-    },
-    {
-      "epoch": 8.821382007822686,
-      "grad_norm": 1.2732475996017456,
-      "learning_rate": 8.350753768844222e-05,
-      "loss": 5.9649,
-      "step": 16915
-    },
-    {
-      "epoch": 8.821903520208604,
-      "grad_norm": 1.341502070426941,
-      "learning_rate": 8.350653266331658e-05,
-      "loss": 5.5747,
-      "step": 16916
-    },
-    {
-      "epoch": 8.822425032594523,
-      "grad_norm": 1.460534691810608,
-      "learning_rate": 8.350552763819096e-05,
-      "loss": 5.4994,
-      "step": 16917
-    },
-    {
-      "epoch": 8.822946544980443,
-      "grad_norm": 1.393713116645813,
-      "learning_rate": 8.350452261306532e-05,
-      "loss": 5.6668,
-      "step": 16918
-    },
-    {
-      "epoch": 8.823468057366362,
-      "grad_norm": 1.3329353332519531,
-      "learning_rate": 8.35035175879397e-05,
-      "loss": 5.8865,
-      "step": 16919
-    },
-    {
-      "epoch": 8.823989569752282,
-      "grad_norm": 1.5972837209701538,
-      "learning_rate": 8.350251256281408e-05,
-      "loss": 4.7382,
-      "step": 16920
-    },
-    {
-      "epoch": 8.824511082138201,
-      "grad_norm": 1.5040769577026367,
-      "learning_rate": 8.350150753768846e-05,
-      "loss": 5.2618,
-      "step": 16921
-    },
-    {
-      "epoch": 8.82503259452412,
-      "grad_norm": 1.3760379552841187,
-      "learning_rate": 8.350050251256282e-05,
-      "loss": 5.7294,
-      "step": 16922
-    },
-    {
-      "epoch": 8.825554106910038,
-      "grad_norm": 1.50537109375,
-      "learning_rate": 8.34994974874372e-05,
-      "loss": 5.5036,
-      "step": 16923
-    },
-    {
-      "epoch": 8.826075619295958,
-      "grad_norm": 1.4735703468322754,
-      "learning_rate": 8.349849246231156e-05,
-      "loss": 5.5226,
-      "step": 16924
-    },
-    {
-      "epoch": 8.826597131681877,
-      "grad_norm": 1.377119779586792,
-      "learning_rate": 8.349748743718593e-05,
-      "loss": 5.6651,
-      "step": 16925
-    },
-    {
-      "epoch": 8.827118644067797,
-      "grad_norm": 1.453910231590271,
-      "learning_rate": 8.34964824120603e-05,
-      "loss": 5.3666,
-      "step": 16926
-    },
-    {
-      "epoch": 8.827640156453716,
-      "grad_norm": 1.5694024562835693,
-      "learning_rate": 8.349547738693467e-05,
-      "loss": 5.5927,
-      "step": 16927
-    },
-    {
-      "epoch": 8.828161668839634,
-      "grad_norm": 1.4811666011810303,
-      "learning_rate": 8.349447236180905e-05,
-      "loss": 5.4048,
-      "step": 16928
-    },
-    {
-      "epoch": 8.828683181225554,
-      "grad_norm": 1.4881889820098877,
-      "learning_rate": 8.349346733668341e-05,
-      "loss": 5.1142,
-      "step": 16929
-    },
-    {
-      "epoch": 8.829204693611473,
-      "grad_norm": 1.6575106382369995,
-      "learning_rate": 8.349246231155779e-05,
-      "loss": 4.9174,
-      "step": 16930
-    },
-    {
-      "epoch": 8.829726205997392,
-      "grad_norm": 1.5757826566696167,
-      "learning_rate": 8.349145728643217e-05,
-      "loss": 4.8897,
-      "step": 16931
-    },
-    {
-      "epoch": 8.830247718383312,
-      "grad_norm": 1.513580083847046,
-      "learning_rate": 8.349045226130655e-05,
-      "loss": 5.2723,
-      "step": 16932
-    },
-    {
-      "epoch": 8.830769230769231,
-      "grad_norm": 1.440398931503296,
-      "learning_rate": 8.348944723618091e-05,
-      "loss": 4.9353,
-      "step": 16933
-    },
-    {
-      "epoch": 8.83129074315515,
-      "grad_norm": 1.471454381942749,
-      "learning_rate": 8.348844221105529e-05,
-      "loss": 4.4621,
-      "step": 16934
-    },
-    {
-      "epoch": 8.831812255541069,
-      "grad_norm": 1.472762107849121,
-      "learning_rate": 8.348743718592965e-05,
-      "loss": 5.4558,
-      "step": 16935
-    },
-    {
-      "epoch": 8.832333767926988,
-      "grad_norm": 1.4964758157730103,
-      "learning_rate": 8.348643216080403e-05,
-      "loss": 5.6354,
-      "step": 16936
-    },
-    {
-      "epoch": 8.832855280312907,
-      "grad_norm": 1.433240294456482,
-      "learning_rate": 8.34854271356784e-05,
-      "loss": 5.107,
-      "step": 16937
-    },
-    {
-      "epoch": 8.833376792698827,
-      "grad_norm": 1.4073988199234009,
-      "learning_rate": 8.348442211055276e-05,
-      "loss": 5.387,
-      "step": 16938
-    },
-    {
-      "epoch": 8.833898305084746,
-      "grad_norm": 1.5243284702301025,
-      "learning_rate": 8.348341708542714e-05,
-      "loss": 5.4604,
-      "step": 16939
-    },
-    {
-      "epoch": 8.834419817470664,
-      "grad_norm": 1.4080723524093628,
-      "learning_rate": 8.34824120603015e-05,
-      "loss": 5.4903,
-      "step": 16940
-    },
-    {
-      "epoch": 8.834941329856584,
-      "grad_norm": 1.5761204957962036,
-      "learning_rate": 8.348140703517588e-05,
-      "loss": 5.2512,
-      "step": 16941
-    },
-    {
-      "epoch": 8.835462842242503,
-      "grad_norm": 1.3451896905899048,
-      "learning_rate": 8.348040201005026e-05,
-      "loss": 5.8983,
-      "step": 16942
-    },
-    {
-      "epoch": 8.835984354628422,
-      "grad_norm": 1.377536654472351,
-      "learning_rate": 8.347939698492463e-05,
-      "loss": 5.704,
-      "step": 16943
-    },
-    {
-      "epoch": 8.836505867014342,
-      "grad_norm": 1.3376017808914185,
-      "learning_rate": 8.3478391959799e-05,
-      "loss": 5.5209,
-      "step": 16944
-    },
-    {
-      "epoch": 8.837027379400261,
-      "grad_norm": 1.3826223611831665,
-      "learning_rate": 8.347738693467338e-05,
-      "loss": 5.4096,
-      "step": 16945
-    },
-    {
-      "epoch": 8.83754889178618,
-      "grad_norm": 1.4316983222961426,
-      "learning_rate": 8.347638190954774e-05,
-      "loss": 5.5008,
-      "step": 16946
-    },
-    {
-      "epoch": 8.838070404172099,
-      "grad_norm": 1.2980077266693115,
-      "learning_rate": 8.347537688442212e-05,
-      "loss": 5.8769,
-      "step": 16947
-    },
-    {
-      "epoch": 8.838591916558018,
-      "grad_norm": 1.366762638092041,
-      "learning_rate": 8.347437185929648e-05,
-      "loss": 5.544,
-      "step": 16948
-    },
-    {
-      "epoch": 8.839113428943937,
-      "grad_norm": 1.3625450134277344,
-      "learning_rate": 8.347336683417086e-05,
-      "loss": 5.6355,
-      "step": 16949
-    },
-    {
-      "epoch": 8.839634941329857,
-      "grad_norm": 1.3293616771697998,
-      "learning_rate": 8.347236180904522e-05,
-      "loss": 5.6614,
-      "step": 16950
-    },
-    {
-      "epoch": 8.840156453715776,
-      "grad_norm": 1.3836934566497803,
-      "learning_rate": 8.34713567839196e-05,
-      "loss": 5.6594,
-      "step": 16951
-    },
-    {
-      "epoch": 8.840677966101694,
-      "grad_norm": 1.4554105997085571,
-      "learning_rate": 8.347035175879398e-05,
-      "loss": 5.6642,
-      "step": 16952
-    },
-    {
-      "epoch": 8.841199478487614,
-      "grad_norm": 1.4288793802261353,
-      "learning_rate": 8.346934673366834e-05,
-      "loss": 5.9669,
-      "step": 16953
-    },
-    {
-      "epoch": 8.841720990873533,
-      "grad_norm": 1.3535306453704834,
-      "learning_rate": 8.346834170854272e-05,
-      "loss": 5.6109,
-      "step": 16954
-    },
-    {
-      "epoch": 8.842242503259452,
-      "grad_norm": 1.4145543575286865,
-      "learning_rate": 8.346733668341709e-05,
-      "loss": 5.5129,
-      "step": 16955
-    },
-    {
-      "epoch": 8.842764015645372,
-      "grad_norm": 2.244659900665283,
-      "learning_rate": 8.346633165829146e-05,
-      "loss": 5.3592,
-      "step": 16956
-    },
-    {
-      "epoch": 8.843285528031291,
-      "grad_norm": 1.3620402812957764,
-      "learning_rate": 8.346532663316583e-05,
-      "loss": 5.1688,
-      "step": 16957
-    },
-    {
-      "epoch": 8.843807040417209,
-      "grad_norm": 1.4569514989852905,
-      "learning_rate": 8.34643216080402e-05,
-      "loss": 5.6433,
-      "step": 16958
-    },
-    {
-      "epoch": 8.844328552803129,
-      "grad_norm": 1.5284662246704102,
-      "learning_rate": 8.346331658291457e-05,
-      "loss": 5.8126,
-      "step": 16959
-    },
-    {
-      "epoch": 8.844850065189048,
-      "grad_norm": 1.5196880102157593,
-      "learning_rate": 8.346231155778895e-05,
-      "loss": 5.1741,
-      "step": 16960
-    },
-    {
-      "epoch": 8.845371577574968,
-      "grad_norm": 1.3637324571609497,
-      "learning_rate": 8.346130653266331e-05,
-      "loss": 5.7165,
-      "step": 16961
-    },
-    {
-      "epoch": 8.845893089960887,
-      "grad_norm": 1.616620659828186,
-      "learning_rate": 8.346030150753769e-05,
-      "loss": 4.8574,
-      "step": 16962
-    },
-    {
-      "epoch": 8.846414602346806,
-      "grad_norm": 1.5080894231796265,
-      "learning_rate": 8.345929648241207e-05,
-      "loss": 5.1315,
-      "step": 16963
-    },
-    {
-      "epoch": 8.846936114732724,
-      "grad_norm": 1.3144590854644775,
-      "learning_rate": 8.345829145728645e-05,
-      "loss": 5.703,
-      "step": 16964
-    },
-    {
-      "epoch": 8.847457627118644,
-      "grad_norm": 1.377013921737671,
-      "learning_rate": 8.345728643216081e-05,
-      "loss": 5.9419,
-      "step": 16965
-    },
-    {
-      "epoch": 8.847979139504563,
-      "grad_norm": 1.4240809679031372,
-      "learning_rate": 8.345628140703517e-05,
-      "loss": 5.2925,
-      "step": 16966
-    },
-    {
-      "epoch": 8.848500651890483,
-      "grad_norm": 1.557939887046814,
-      "learning_rate": 8.345527638190955e-05,
-      "loss": 5.0437,
-      "step": 16967
-    },
-    {
-      "epoch": 8.849022164276402,
-      "grad_norm": 1.432033896446228,
-      "learning_rate": 8.345427135678392e-05,
-      "loss": 5.1258,
-      "step": 16968
-    },
-    {
-      "epoch": 8.849543676662321,
-      "grad_norm": 1.6672288179397583,
-      "learning_rate": 8.34532663316583e-05,
-      "loss": 5.9851,
-      "step": 16969
-    },
-    {
-      "epoch": 8.85006518904824,
-      "grad_norm": 1.5628514289855957,
-      "learning_rate": 8.345226130653266e-05,
-      "loss": 5.653,
-      "step": 16970
-    },
-    {
-      "epoch": 8.850586701434159,
-      "grad_norm": 1.4421374797821045,
-      "learning_rate": 8.345125628140704e-05,
-      "loss": 5.4535,
-      "step": 16971
-    },
-    {
-      "epoch": 8.851108213820078,
-      "grad_norm": 1.4180487394332886,
-      "learning_rate": 8.345025125628141e-05,
-      "loss": 5.7102,
-      "step": 16972
-    },
-    {
-      "epoch": 8.851629726205998,
-      "grad_norm": 1.468701720237732,
-      "learning_rate": 8.344924623115579e-05,
-      "loss": 5.6521,
-      "step": 16973
-    },
-    {
-      "epoch": 8.852151238591917,
-      "grad_norm": 1.4608471393585205,
-      "learning_rate": 8.344824120603016e-05,
-      "loss": 5.3677,
-      "step": 16974
-    },
-    {
-      "epoch": 8.852672750977836,
-      "grad_norm": 1.4559251070022583,
-      "learning_rate": 8.344723618090453e-05,
-      "loss": 5.1728,
-      "step": 16975
-    },
-    {
-      "epoch": 8.853194263363754,
-      "grad_norm": 1.4544708728790283,
-      "learning_rate": 8.34462311557789e-05,
-      "loss": 5.401,
-      "step": 16976
-    },
-    {
-      "epoch": 8.853715775749674,
-      "grad_norm": 1.5114163160324097,
-      "learning_rate": 8.344522613065328e-05,
-      "loss": 5.1762,
-      "step": 16977
-    },
-    {
-      "epoch": 8.854237288135593,
-      "grad_norm": 1.4597363471984863,
-      "learning_rate": 8.344422110552764e-05,
-      "loss": 5.3442,
-      "step": 16978
-    },
-    {
-      "epoch": 8.854758800521513,
-      "grad_norm": 1.66215181350708,
-      "learning_rate": 8.3443216080402e-05,
-      "loss": 5.8399,
-      "step": 16979
-    },
-    {
-      "epoch": 8.855280312907432,
-      "grad_norm": 1.31671142578125,
-      "learning_rate": 8.344221105527638e-05,
-      "loss": 5.3776,
-      "step": 16980
-    },
-    {
-      "epoch": 8.855801825293351,
-      "grad_norm": 1.4573251008987427,
-      "learning_rate": 8.344120603015075e-05,
-      "loss": 5.4253,
-      "step": 16981
-    },
-    {
-      "epoch": 8.85632333767927,
-      "grad_norm": 1.4840813875198364,
-      "learning_rate": 8.344020100502513e-05,
-      "loss": 5.8279,
-      "step": 16982
-    },
-    {
-      "epoch": 8.856844850065189,
-      "grad_norm": 1.892006754875183,
-      "learning_rate": 8.34391959798995e-05,
-      "loss": 4.9287,
-      "step": 16983
-    },
-    {
-      "epoch": 8.857366362451108,
-      "grad_norm": 1.381309151649475,
-      "learning_rate": 8.343819095477388e-05,
-      "loss": 5.4582,
-      "step": 16984
-    },
-    {
-      "epoch": 8.857887874837028,
-      "grad_norm": 1.5987880229949951,
-      "learning_rate": 8.343718592964825e-05,
-      "loss": 5.0217,
-      "step": 16985
-    },
-    {
-      "epoch": 8.858409387222947,
-      "grad_norm": 1.5874238014221191,
-      "learning_rate": 8.343618090452262e-05,
-      "loss": 5.425,
-      "step": 16986
-    },
-    {
-      "epoch": 8.858930899608866,
-      "grad_norm": 1.377091407775879,
-      "learning_rate": 8.343517587939699e-05,
-      "loss": 5.3538,
-      "step": 16987
-    },
-    {
-      "epoch": 8.859452411994784,
-      "grad_norm": 1.6570526361465454,
-      "learning_rate": 8.343417085427137e-05,
-      "loss": 5.2637,
-      "step": 16988
-    },
-    {
-      "epoch": 8.859973924380704,
-      "grad_norm": 1.389289379119873,
-      "learning_rate": 8.343316582914573e-05,
-      "loss": 5.5364,
-      "step": 16989
-    },
-    {
-      "epoch": 8.860495436766623,
-      "grad_norm": 1.4337373971939087,
-      "learning_rate": 8.343216080402011e-05,
-      "loss": 5.5896,
-      "step": 16990
-    },
-    {
-      "epoch": 8.861016949152543,
-      "grad_norm": 1.3667356967926025,
-      "learning_rate": 8.343115577889447e-05,
-      "loss": 5.5545,
-      "step": 16991
-    },
-    {
-      "epoch": 8.861538461538462,
-      "grad_norm": 1.3697766065597534,
-      "learning_rate": 8.343015075376885e-05,
-      "loss": 5.8303,
-      "step": 16992
-    },
-    {
-      "epoch": 8.862059973924381,
-      "grad_norm": 1.3990833759307861,
-      "learning_rate": 8.342914572864323e-05,
-      "loss": 5.7169,
-      "step": 16993
-    },
-    {
-      "epoch": 8.8625814863103,
-      "grad_norm": 1.2958526611328125,
-      "learning_rate": 8.342814070351759e-05,
-      "loss": 5.8773,
-      "step": 16994
-    },
-    {
-      "epoch": 8.863102998696219,
-      "grad_norm": 1.4506174325942993,
-      "learning_rate": 8.342713567839197e-05,
-      "loss": 5.3704,
-      "step": 16995
-    },
-    {
-      "epoch": 8.863624511082138,
-      "grad_norm": 1.6168773174285889,
-      "learning_rate": 8.342613065326633e-05,
-      "loss": 4.7147,
-      "step": 16996
-    },
-    {
-      "epoch": 8.864146023468058,
-      "grad_norm": 1.447791576385498,
-      "learning_rate": 8.342512562814071e-05,
-      "loss": 5.6523,
-      "step": 16997
-    },
-    {
-      "epoch": 8.864667535853977,
-      "grad_norm": 1.4244554042816162,
-      "learning_rate": 8.342412060301508e-05,
-      "loss": 5.4265,
-      "step": 16998
-    },
-    {
-      "epoch": 8.865189048239897,
-      "grad_norm": 1.4577950239181519,
-      "learning_rate": 8.342311557788945e-05,
-      "loss": 5.5983,
-      "step": 16999
-    },
-    {
-      "epoch": 8.865710560625814,
-      "grad_norm": 1.4404419660568237,
-      "learning_rate": 8.342211055276382e-05,
-      "loss": 5.7697,
-      "step": 17000
-    },
-    {
-      "epoch": 8.866232073011734,
-      "grad_norm": 1.5066076517105103,
-      "learning_rate": 8.34211055276382e-05,
-      "loss": 5.7429,
-      "step": 17001
-    },
-    {
-      "epoch": 8.866753585397653,
-      "grad_norm": 1.450811743736267,
-      "learning_rate": 8.342010050251256e-05,
-      "loss": 5.3872,
-      "step": 17002
-    },
-    {
-      "epoch": 8.867275097783573,
-      "grad_norm": 1.483816146850586,
-      "learning_rate": 8.341909547738694e-05,
-      "loss": 5.7021,
-      "step": 17003
-    },
-    {
-      "epoch": 8.867796610169492,
-      "grad_norm": 1.6130540370941162,
-      "learning_rate": 8.341809045226132e-05,
-      "loss": 5.2377,
-      "step": 17004
-    },
-    {
-      "epoch": 8.868318122555412,
-      "grad_norm": 1.522417664527893,
-      "learning_rate": 8.341708542713568e-05,
-      "loss": 5.0206,
-      "step": 17005
-    },
-    {
-      "epoch": 8.86883963494133,
-      "grad_norm": 1.358878254890442,
-      "learning_rate": 8.341608040201006e-05,
-      "loss": 5.3277,
-      "step": 17006
-    },
-    {
-      "epoch": 8.869361147327249,
-      "grad_norm": 1.41798734664917,
-      "learning_rate": 8.341507537688442e-05,
-      "loss": 5.6458,
-      "step": 17007
-    },
-    {
-      "epoch": 8.869882659713168,
-      "grad_norm": 1.4129235744476318,
-      "learning_rate": 8.34140703517588e-05,
-      "loss": 5.5138,
-      "step": 17008
-    },
-    {
-      "epoch": 8.870404172099088,
-      "grad_norm": 1.3933632373809814,
-      "learning_rate": 8.341306532663316e-05,
-      "loss": 5.6212,
-      "step": 17009
-    },
-    {
-      "epoch": 8.870925684485007,
-      "grad_norm": 1.2871266603469849,
-      "learning_rate": 8.341206030150754e-05,
-      "loss": 5.8201,
-      "step": 17010
-    },
-    {
-      "epoch": 8.871447196870925,
-      "grad_norm": 1.3837333917617798,
-      "learning_rate": 8.34110552763819e-05,
-      "loss": 5.7127,
-      "step": 17011
-    },
-    {
-      "epoch": 8.871968709256844,
-      "grad_norm": 1.3612205982208252,
-      "learning_rate": 8.341005025125628e-05,
-      "loss": 5.4922,
-      "step": 17012
-    },
-    {
-      "epoch": 8.872490221642764,
-      "grad_norm": 1.4796544313430786,
-      "learning_rate": 8.340904522613066e-05,
-      "loss": 5.4851,
-      "step": 17013
-    },
-    {
-      "epoch": 8.873011734028683,
-      "grad_norm": 1.4177509546279907,
-      "learning_rate": 8.340804020100504e-05,
-      "loss": 5.2942,
-      "step": 17014
-    },
-    {
-      "epoch": 8.873533246414603,
-      "grad_norm": 1.3008191585540771,
-      "learning_rate": 8.34070351758794e-05,
-      "loss": 5.7318,
-      "step": 17015
-    },
-    {
-      "epoch": 8.874054758800522,
-      "grad_norm": 1.3889836072921753,
-      "learning_rate": 8.340603015075378e-05,
-      "loss": 5.2947,
-      "step": 17016
-    },
-    {
-      "epoch": 8.874576271186442,
-      "grad_norm": 1.5064547061920166,
-      "learning_rate": 8.340502512562815e-05,
-      "loss": 5.4118,
-      "step": 17017
-    },
-    {
-      "epoch": 8.87509778357236,
-      "grad_norm": 1.4889479875564575,
-      "learning_rate": 8.340402010050251e-05,
-      "loss": 5.5919,
-      "step": 17018
-    },
-    {
-      "epoch": 8.875619295958279,
-      "grad_norm": 1.3423919677734375,
-      "learning_rate": 8.340301507537689e-05,
-      "loss": 5.7147,
-      "step": 17019
-    },
-    {
-      "epoch": 8.876140808344198,
-      "grad_norm": 1.381288766860962,
-      "learning_rate": 8.340201005025125e-05,
-      "loss": 5.3398,
-      "step": 17020
-    },
-    {
-      "epoch": 8.876662320730118,
-      "grad_norm": 1.491529941558838,
-      "learning_rate": 8.340100502512563e-05,
-      "loss": 5.7879,
-      "step": 17021
-    },
-    {
-      "epoch": 8.877183833116037,
-      "grad_norm": 1.4192219972610474,
-      "learning_rate": 8.34e-05,
-      "loss": 5.5968,
-      "step": 17022
-    },
-    {
-      "epoch": 8.877705345501955,
-      "grad_norm": 1.439959168434143,
-      "learning_rate": 8.339899497487437e-05,
-      "loss": 5.4637,
-      "step": 17023
-    },
-    {
-      "epoch": 8.878226857887874,
-      "grad_norm": 1.5505597591400146,
-      "learning_rate": 8.339798994974875e-05,
-      "loss": 5.1219,
-      "step": 17024
-    },
-    {
-      "epoch": 8.878748370273794,
-      "grad_norm": 1.4464688301086426,
-      "learning_rate": 8.339698492462313e-05,
-      "loss": 5.3009,
-      "step": 17025
-    },
-    {
-      "epoch": 8.879269882659713,
-      "grad_norm": 1.4401503801345825,
-      "learning_rate": 8.339597989949749e-05,
-      "loss": 5.6133,
-      "step": 17026
-    },
-    {
-      "epoch": 8.879791395045633,
-      "grad_norm": 1.3498351573944092,
-      "learning_rate": 8.339497487437187e-05,
-      "loss": 5.6691,
-      "step": 17027
-    },
-    {
-      "epoch": 8.880312907431552,
-      "grad_norm": 1.4354692697525024,
-      "learning_rate": 8.339396984924623e-05,
-      "loss": 5.7865,
-      "step": 17028
-    },
-    {
-      "epoch": 8.880834419817472,
-      "grad_norm": 1.3337249755859375,
-      "learning_rate": 8.339296482412061e-05,
-      "loss": 5.9128,
-      "step": 17029
-    },
-    {
-      "epoch": 8.88135593220339,
-      "grad_norm": 1.3404130935668945,
-      "learning_rate": 8.339195979899498e-05,
-      "loss": 5.3865,
-      "step": 17030
-    },
-    {
-      "epoch": 8.881877444589309,
-      "grad_norm": 1.3697166442871094,
-      "learning_rate": 8.339095477386934e-05,
-      "loss": 5.7218,
-      "step": 17031
-    },
-    {
-      "epoch": 8.882398956975228,
-      "grad_norm": 1.3653641939163208,
-      "learning_rate": 8.338994974874372e-05,
-      "loss": 5.6249,
-      "step": 17032
-    },
-    {
-      "epoch": 8.882920469361148,
-      "grad_norm": 1.5404266119003296,
-      "learning_rate": 8.33889447236181e-05,
-      "loss": 5.7161,
-      "step": 17033
-    },
-    {
-      "epoch": 8.883441981747067,
-      "grad_norm": 1.2855253219604492,
-      "learning_rate": 8.338793969849247e-05,
-      "loss": 4.8932,
-      "step": 17034
-    },
-    {
-      "epoch": 8.883963494132985,
-      "grad_norm": 1.4245060682296753,
-      "learning_rate": 8.338693467336684e-05,
-      "loss": 5.2884,
-      "step": 17035
-    },
-    {
-      "epoch": 8.884485006518904,
-      "grad_norm": 1.217742681503296,
-      "learning_rate": 8.338592964824122e-05,
-      "loss": 5.6901,
-      "step": 17036
-    },
-    {
-      "epoch": 8.885006518904824,
-      "grad_norm": 1.3787429332733154,
-      "learning_rate": 8.338492462311558e-05,
-      "loss": 5.0566,
-      "step": 17037
-    },
-    {
-      "epoch": 8.885528031290743,
-      "grad_norm": 1.331830620765686,
-      "learning_rate": 8.338391959798996e-05,
-      "loss": 5.7823,
-      "step": 17038
-    },
-    {
-      "epoch": 8.886049543676663,
-      "grad_norm": 1.393920660018921,
-      "learning_rate": 8.338291457286432e-05,
-      "loss": 5.4975,
-      "step": 17039
-    },
-    {
-      "epoch": 8.886571056062582,
-      "grad_norm": 1.3951112031936646,
-      "learning_rate": 8.33819095477387e-05,
-      "loss": 5.7929,
-      "step": 17040
-    },
-    {
-      "epoch": 8.887092568448502,
-      "grad_norm": 1.4013617038726807,
-      "learning_rate": 8.338090452261306e-05,
-      "loss": 5.6447,
-      "step": 17041
-    },
-    {
-      "epoch": 8.88761408083442,
-      "grad_norm": 1.4405478239059448,
-      "learning_rate": 8.337989949748744e-05,
-      "loss": 4.9556,
-      "step": 17042
-    },
-    {
-      "epoch": 8.888135593220339,
-      "grad_norm": 1.5156874656677246,
-      "learning_rate": 8.33788944723618e-05,
-      "loss": 5.0333,
-      "step": 17043
-    },
-    {
-      "epoch": 8.888657105606258,
-      "grad_norm": 1.5013337135314941,
-      "learning_rate": 8.337788944723618e-05,
-      "loss": 5.6318,
-      "step": 17044
-    },
-    {
-      "epoch": 8.889178617992178,
-      "grad_norm": 1.3334108591079712,
-      "learning_rate": 8.337688442211056e-05,
-      "loss": 5.8097,
-      "step": 17045
-    },
-    {
-      "epoch": 8.889700130378097,
-      "grad_norm": 1.4015549421310425,
-      "learning_rate": 8.337587939698493e-05,
-      "loss": 5.6382,
-      "step": 17046
-    },
-    {
-      "epoch": 8.890221642764015,
-      "grad_norm": 1.3645899295806885,
-      "learning_rate": 8.33748743718593e-05,
-      "loss": 4.9561,
-      "step": 17047
-    },
-    {
-      "epoch": 8.890743155149934,
-      "grad_norm": 1.3264247179031372,
-      "learning_rate": 8.337386934673367e-05,
-      "loss": 5.4444,
-      "step": 17048
-    },
-    {
-      "epoch": 8.891264667535854,
-      "grad_norm": 1.3543262481689453,
-      "learning_rate": 8.337286432160805e-05,
-      "loss": 5.6786,
-      "step": 17049
-    },
-    {
-      "epoch": 8.891786179921773,
-      "grad_norm": 1.3420767784118652,
-      "learning_rate": 8.337185929648241e-05,
-      "loss": 5.7155,
-      "step": 17050
-    },
-    {
-      "epoch": 8.892307692307693,
-      "grad_norm": 1.298834204673767,
-      "learning_rate": 8.337085427135679e-05,
-      "loss": 6.0569,
-      "step": 17051
-    },
-    {
-      "epoch": 8.892829204693612,
-      "grad_norm": 1.3356413841247559,
-      "learning_rate": 8.336984924623115e-05,
-      "loss": 5.9322,
-      "step": 17052
-    },
-    {
-      "epoch": 8.89335071707953,
-      "grad_norm": 1.4659143686294556,
-      "learning_rate": 8.336884422110553e-05,
-      "loss": 5.6301,
-      "step": 17053
-    },
-    {
-      "epoch": 8.89387222946545,
-      "grad_norm": 1.458458423614502,
-      "learning_rate": 8.336783919597991e-05,
-      "loss": 5.5794,
-      "step": 17054
-    },
-    {
-      "epoch": 8.894393741851369,
-      "grad_norm": 1.4793978929519653,
-      "learning_rate": 8.336683417085429e-05,
-      "loss": 5.2095,
-      "step": 17055
-    },
-    {
-      "epoch": 8.894915254237288,
-      "grad_norm": 1.438133955001831,
-      "learning_rate": 8.336582914572865e-05,
-      "loss": 5.4103,
-      "step": 17056
-    },
-    {
-      "epoch": 8.895436766623208,
-      "grad_norm": 1.3856194019317627,
-      "learning_rate": 8.336482412060303e-05,
-      "loss": 5.7321,
-      "step": 17057
-    },
-    {
-      "epoch": 8.895958279009127,
-      "grad_norm": 1.6212068796157837,
-      "learning_rate": 8.336381909547739e-05,
-      "loss": 5.4017,
-      "step": 17058
-    },
-    {
-      "epoch": 8.896479791395045,
-      "grad_norm": 1.5043728351593018,
-      "learning_rate": 8.336281407035176e-05,
-      "loss": 5.3008,
-      "step": 17059
-    },
-    {
-      "epoch": 8.897001303780964,
-      "grad_norm": 1.4352198839187622,
-      "learning_rate": 8.336180904522613e-05,
-      "loss": 5.4944,
-      "step": 17060
-    },
-    {
-      "epoch": 8.897522816166884,
-      "grad_norm": 1.417343258857727,
-      "learning_rate": 8.33608040201005e-05,
-      "loss": 5.656,
-      "step": 17061
-    },
-    {
-      "epoch": 8.898044328552803,
-      "grad_norm": 1.6876219511032104,
-      "learning_rate": 8.335979899497488e-05,
-      "loss": 5.3495,
-      "step": 17062
-    },
-    {
-      "epoch": 8.898565840938723,
-      "grad_norm": 1.3979650735855103,
-      "learning_rate": 8.335879396984924e-05,
-      "loss": 5.4753,
-      "step": 17063
-    },
-    {
-      "epoch": 8.899087353324642,
-      "grad_norm": 1.3552018404006958,
-      "learning_rate": 8.335778894472362e-05,
-      "loss": 5.235,
-      "step": 17064
-    },
-    {
-      "epoch": 8.89960886571056,
-      "grad_norm": 1.5721842050552368,
-      "learning_rate": 8.3356783919598e-05,
-      "loss": 5.2712,
-      "step": 17065
-    },
-    {
-      "epoch": 8.90013037809648,
-      "grad_norm": 1.8373327255249023,
-      "learning_rate": 8.335577889447237e-05,
-      "loss": 5.2961,
-      "step": 17066
-    },
-    {
-      "epoch": 8.900651890482399,
-      "grad_norm": 1.507252812385559,
-      "learning_rate": 8.335477386934674e-05,
-      "loss": 5.9541,
-      "step": 17067
-    },
-    {
-      "epoch": 8.901173402868318,
-      "grad_norm": 1.4344172477722168,
-      "learning_rate": 8.335376884422112e-05,
-      "loss": 5.3591,
-      "step": 17068
-    },
-    {
-      "epoch": 8.901694915254238,
-      "grad_norm": 1.2831131219863892,
-      "learning_rate": 8.335276381909548e-05,
-      "loss": 5.7469,
-      "step": 17069
-    },
-    {
-      "epoch": 8.902216427640157,
-      "grad_norm": 1.346082329750061,
-      "learning_rate": 8.335175879396986e-05,
-      "loss": 5.4401,
-      "step": 17070
-    },
-    {
-      "epoch": 8.902737940026075,
-      "grad_norm": 1.490307092666626,
-      "learning_rate": 8.335075376884422e-05,
-      "loss": 5.2968,
-      "step": 17071
-    },
-    {
-      "epoch": 8.903259452411994,
-      "grad_norm": 1.3915188312530518,
-      "learning_rate": 8.334974874371859e-05,
-      "loss": 5.7367,
-      "step": 17072
-    },
-    {
-      "epoch": 8.903780964797914,
-      "grad_norm": 1.520064353942871,
-      "learning_rate": 8.334874371859297e-05,
-      "loss": 5.4297,
-      "step": 17073
-    },
-    {
-      "epoch": 8.904302477183833,
-      "grad_norm": 1.380345344543457,
-      "learning_rate": 8.334773869346734e-05,
-      "loss": 5.5236,
-      "step": 17074
-    },
-    {
-      "epoch": 8.904823989569753,
-      "grad_norm": 1.365453839302063,
-      "learning_rate": 8.334673366834172e-05,
-      "loss": 5.3083,
-      "step": 17075
-    },
-    {
-      "epoch": 8.905345501955672,
-      "grad_norm": 1.5276308059692383,
-      "learning_rate": 8.334572864321609e-05,
-      "loss": 4.6554,
-      "step": 17076
-    },
-    {
-      "epoch": 8.90586701434159,
-      "grad_norm": 1.5000911951065063,
-      "learning_rate": 8.334472361809046e-05,
-      "loss": 5.7593,
-      "step": 17077
-    },
-    {
-      "epoch": 8.90638852672751,
-      "grad_norm": 1.5297530889511108,
-      "learning_rate": 8.334371859296483e-05,
-      "loss": 5.373,
-      "step": 17078
-    },
-    {
-      "epoch": 8.906910039113429,
-      "grad_norm": 1.4709762334823608,
-      "learning_rate": 8.33427135678392e-05,
-      "loss": 5.7279,
-      "step": 17079
-    },
-    {
-      "epoch": 8.907431551499348,
-      "grad_norm": 1.4669662714004517,
-      "learning_rate": 8.334170854271357e-05,
-      "loss": 5.0233,
-      "step": 17080
-    },
-    {
-      "epoch": 8.907953063885268,
-      "grad_norm": 1.3567931652069092,
-      "learning_rate": 8.334070351758795e-05,
-      "loss": 5.818,
-      "step": 17081
-    },
-    {
-      "epoch": 8.908474576271187,
-      "grad_norm": 1.4955748319625854,
-      "learning_rate": 8.333969849246231e-05,
-      "loss": 5.5165,
-      "step": 17082
-    },
-    {
-      "epoch": 8.908996088657105,
-      "grad_norm": 1.8687615394592285,
-      "learning_rate": 8.333869346733669e-05,
-      "loss": 5.3448,
-      "step": 17083
-    },
-    {
-      "epoch": 8.909517601043024,
-      "grad_norm": 1.4678385257720947,
-      "learning_rate": 8.333768844221105e-05,
-      "loss": 4.9365,
-      "step": 17084
-    },
-    {
-      "epoch": 8.910039113428944,
-      "grad_norm": 1.36420476436615,
-      "learning_rate": 8.333668341708543e-05,
-      "loss": 5.7715,
-      "step": 17085
-    },
-    {
-      "epoch": 8.910560625814863,
-      "grad_norm": 1.4271156787872314,
-      "learning_rate": 8.333567839195981e-05,
-      "loss": 5.3347,
-      "step": 17086
-    },
-    {
-      "epoch": 8.911082138200783,
-      "grad_norm": 1.390344262123108,
-      "learning_rate": 8.333467336683417e-05,
-      "loss": 5.6018,
-      "step": 17087
-    },
-    {
-      "epoch": 8.911603650586702,
-      "grad_norm": 1.4372127056121826,
-      "learning_rate": 8.333366834170855e-05,
-      "loss": 5.3205,
-      "step": 17088
-    },
-    {
-      "epoch": 8.91212516297262,
-      "grad_norm": 1.480141043663025,
-      "learning_rate": 8.333266331658292e-05,
-      "loss": 5.5567,
-      "step": 17089
-    },
-    {
-      "epoch": 8.91264667535854,
-      "grad_norm": 1.500901222229004,
-      "learning_rate": 8.33316582914573e-05,
-      "loss": 5.497,
-      "step": 17090
-    },
-    {
-      "epoch": 8.913168187744459,
-      "grad_norm": 1.415892243385315,
-      "learning_rate": 8.333065326633166e-05,
-      "loss": 5.5679,
-      "step": 17091
-    },
-    {
-      "epoch": 8.913689700130378,
-      "grad_norm": 1.5265836715698242,
-      "learning_rate": 8.332964824120604e-05,
-      "loss": 5.1616,
-      "step": 17092
-    },
-    {
-      "epoch": 8.914211212516298,
-      "grad_norm": 1.3410894870758057,
-      "learning_rate": 8.33286432160804e-05,
-      "loss": 5.7341,
-      "step": 17093
-    },
-    {
-      "epoch": 8.914732724902217,
-      "grad_norm": 1.4134284257888794,
-      "learning_rate": 8.332763819095478e-05,
-      "loss": 5.2393,
-      "step": 17094
-    },
-    {
-      "epoch": 8.915254237288135,
-      "grad_norm": 1.4884827136993408,
-      "learning_rate": 8.332663316582914e-05,
-      "loss": 5.4105,
-      "step": 17095
-    },
-    {
-      "epoch": 8.915775749674054,
-      "grad_norm": 1.4100183248519897,
-      "learning_rate": 8.332562814070352e-05,
-      "loss": 5.5505,
-      "step": 17096
-    },
-    {
-      "epoch": 8.916297262059974,
-      "grad_norm": 1.3131128549575806,
-      "learning_rate": 8.33246231155779e-05,
-      "loss": 5.3032,
-      "step": 17097
-    },
-    {
-      "epoch": 8.916818774445893,
-      "grad_norm": 1.5243116617202759,
-      "learning_rate": 8.332361809045226e-05,
-      "loss": 5.8592,
-      "step": 17098
-    },
-    {
-      "epoch": 8.917340286831813,
-      "grad_norm": 1.4357651472091675,
-      "learning_rate": 8.332261306532664e-05,
-      "loss": 5.3969,
-      "step": 17099
-    },
-    {
-      "epoch": 8.917861799217732,
-      "grad_norm": 1.4844194650650024,
-      "learning_rate": 8.3321608040201e-05,
-      "loss": 5.5607,
-      "step": 17100
-    },
-    {
-      "epoch": 8.91838331160365,
-      "grad_norm": 1.4648679494857788,
-      "learning_rate": 8.332060301507538e-05,
-      "loss": 5.7571,
-      "step": 17101
-    },
-    {
-      "epoch": 8.91890482398957,
-      "grad_norm": 1.299300193786621,
-      "learning_rate": 8.331959798994975e-05,
-      "loss": 5.4771,
-      "step": 17102
-    },
-    {
-      "epoch": 8.919426336375489,
-      "grad_norm": 1.5151951313018799,
-      "learning_rate": 8.331859296482412e-05,
-      "loss": 5.4314,
-      "step": 17103
-    },
-    {
-      "epoch": 8.919947848761408,
-      "grad_norm": 1.5636613368988037,
-      "learning_rate": 8.331758793969849e-05,
-      "loss": 4.9766,
-      "step": 17104
-    },
-    {
-      "epoch": 8.920469361147328,
-      "grad_norm": 1.5324138402938843,
-      "learning_rate": 8.331658291457287e-05,
-      "loss": 5.1772,
-      "step": 17105
-    },
-    {
-      "epoch": 8.920990873533245,
-      "grad_norm": 1.3634483814239502,
-      "learning_rate": 8.331557788944724e-05,
-      "loss": 5.55,
-      "step": 17106
-    },
-    {
-      "epoch": 8.921512385919165,
-      "grad_norm": 1.3414641618728638,
-      "learning_rate": 8.331457286432162e-05,
-      "loss": 5.5048,
-      "step": 17107
-    },
-    {
-      "epoch": 8.922033898305084,
-      "grad_norm": 1.5255392789840698,
-      "learning_rate": 8.331356783919599e-05,
-      "loss": 5.3767,
-      "step": 17108
-    },
-    {
-      "epoch": 8.922555410691004,
-      "grad_norm": 1.3961032629013062,
-      "learning_rate": 8.331256281407036e-05,
-      "loss": 5.1272,
-      "step": 17109
-    },
-    {
-      "epoch": 8.923076923076923,
-      "grad_norm": 1.5240453481674194,
-      "learning_rate": 8.331155778894473e-05,
-      "loss": 5.6604,
-      "step": 17110
-    },
-    {
-      "epoch": 8.923598435462843,
-      "grad_norm": 1.4089477062225342,
-      "learning_rate": 8.331055276381909e-05,
-      "loss": 5.3614,
-      "step": 17111
-    },
-    {
-      "epoch": 8.924119947848762,
-      "grad_norm": 1.4352165460586548,
-      "learning_rate": 8.330954773869347e-05,
-      "loss": 5.1746,
-      "step": 17112
-    },
-    {
-      "epoch": 8.92464146023468,
-      "grad_norm": 1.444793939590454,
-      "learning_rate": 8.330854271356783e-05,
-      "loss": 5.4358,
-      "step": 17113
-    },
-    {
-      "epoch": 8.9251629726206,
-      "grad_norm": 1.3942513465881348,
-      "learning_rate": 8.330753768844221e-05,
-      "loss": 5.9376,
-      "step": 17114
-    },
-    {
-      "epoch": 8.925684485006519,
-      "grad_norm": 1.451870322227478,
-      "learning_rate": 8.330653266331658e-05,
-      "loss": 5.8044,
-      "step": 17115
-    },
-    {
-      "epoch": 8.926205997392438,
-      "grad_norm": 1.3676447868347168,
-      "learning_rate": 8.330552763819095e-05,
-      "loss": 5.8257,
-      "step": 17116
-    },
-    {
-      "epoch": 8.926727509778358,
-      "grad_norm": 1.4487427473068237,
-      "learning_rate": 8.330452261306533e-05,
-      "loss": 5.64,
-      "step": 17117
-    },
-    {
-      "epoch": 8.927249022164276,
-      "grad_norm": 1.390188455581665,
-      "learning_rate": 8.330351758793971e-05,
-      "loss": 5.5017,
-      "step": 17118
-    },
-    {
-      "epoch": 8.927770534550195,
-      "grad_norm": 1.4026213884353638,
-      "learning_rate": 8.330251256281407e-05,
-      "loss": 5.6648,
-      "step": 17119
-    },
-    {
-      "epoch": 8.928292046936114,
-      "grad_norm": 1.46074640750885,
-      "learning_rate": 8.330150753768845e-05,
-      "loss": 4.9842,
-      "step": 17120
-    },
-    {
-      "epoch": 8.928813559322034,
-      "grad_norm": 1.4290302991867065,
-      "learning_rate": 8.330050251256282e-05,
-      "loss": 5.3563,
-      "step": 17121
-    },
-    {
-      "epoch": 8.929335071707953,
-      "grad_norm": 1.4640207290649414,
-      "learning_rate": 8.32994974874372e-05,
-      "loss": 5.4237,
-      "step": 17122
-    },
-    {
-      "epoch": 8.929856584093873,
-      "grad_norm": 1.5441086292266846,
-      "learning_rate": 8.329849246231156e-05,
-      "loss": 5.4458,
-      "step": 17123
-    },
-    {
-      "epoch": 8.930378096479792,
-      "grad_norm": 1.6134448051452637,
-      "learning_rate": 8.329748743718592e-05,
-      "loss": 5.5827,
-      "step": 17124
-    },
-    {
-      "epoch": 8.93089960886571,
-      "grad_norm": 1.3906546831130981,
-      "learning_rate": 8.32964824120603e-05,
-      "loss": 5.8959,
-      "step": 17125
-    },
-    {
-      "epoch": 8.93142112125163,
-      "grad_norm": 1.3615833520889282,
-      "learning_rate": 8.329547738693468e-05,
-      "loss": 5.5994,
-      "step": 17126
-    },
-    {
-      "epoch": 8.931942633637549,
-      "grad_norm": 1.5156500339508057,
-      "learning_rate": 8.329447236180906e-05,
-      "loss": 4.4772,
-      "step": 17127
-    },
-    {
-      "epoch": 8.932464146023468,
-      "grad_norm": 1.4692529439926147,
-      "learning_rate": 8.329346733668342e-05,
-      "loss": 5.3978,
-      "step": 17128
-    },
-    {
-      "epoch": 8.932985658409388,
-      "grad_norm": 1.4153026342391968,
-      "learning_rate": 8.32924623115578e-05,
-      "loss": 5.4921,
-      "step": 17129
-    },
-    {
-      "epoch": 8.933507170795306,
-      "grad_norm": 1.3382090330123901,
-      "learning_rate": 8.329145728643216e-05,
-      "loss": 5.7855,
-      "step": 17130
-    },
-    {
-      "epoch": 8.934028683181225,
-      "grad_norm": 1.4358071088790894,
-      "learning_rate": 8.329045226130654e-05,
-      "loss": 5.7591,
-      "step": 17131
-    },
-    {
-      "epoch": 8.934550195567144,
-      "grad_norm": 1.3905256986618042,
-      "learning_rate": 8.32894472361809e-05,
-      "loss": 5.4849,
-      "step": 17132
-    },
-    {
-      "epoch": 8.935071707953064,
-      "grad_norm": 1.313779354095459,
-      "learning_rate": 8.328844221105528e-05,
-      "loss": 5.8745,
-      "step": 17133
-    },
-    {
-      "epoch": 8.935593220338983,
-      "grad_norm": 1.3895442485809326,
-      "learning_rate": 8.328743718592965e-05,
-      "loss": 5.4586,
-      "step": 17134
-    },
-    {
-      "epoch": 8.936114732724903,
-      "grad_norm": 1.3851535320281982,
-      "learning_rate": 8.328643216080402e-05,
-      "loss": 5.3865,
-      "step": 17135
-    },
-    {
-      "epoch": 8.93663624511082,
-      "grad_norm": 1.594976544380188,
-      "learning_rate": 8.328542713567839e-05,
-      "loss": 4.9754,
-      "step": 17136
-    },
-    {
-      "epoch": 8.93715775749674,
-      "grad_norm": 1.4534224271774292,
-      "learning_rate": 8.328442211055277e-05,
-      "loss": 5.239,
-      "step": 17137
-    },
-    {
-      "epoch": 8.93767926988266,
-      "grad_norm": 1.4070262908935547,
-      "learning_rate": 8.328341708542714e-05,
-      "loss": 4.8981,
-      "step": 17138
-    },
-    {
-      "epoch": 8.938200782268579,
-      "grad_norm": 1.4809869527816772,
-      "learning_rate": 8.328241206030151e-05,
-      "loss": 5.8516,
-      "step": 17139
-    },
-    {
-      "epoch": 8.938722294654498,
-      "grad_norm": 1.370180368423462,
-      "learning_rate": 8.328140703517589e-05,
-      "loss": 5.5463,
-      "step": 17140
-    },
-    {
-      "epoch": 8.939243807040418,
-      "grad_norm": 1.2913397550582886,
-      "learning_rate": 8.328040201005025e-05,
-      "loss": 5.7666,
-      "step": 17141
-    },
-    {
-      "epoch": 8.939765319426336,
-      "grad_norm": 1.4915170669555664,
-      "learning_rate": 8.327939698492463e-05,
-      "loss": 5.5432,
-      "step": 17142
-    },
-    {
-      "epoch": 8.940286831812255,
-      "grad_norm": 1.4622982740402222,
-      "learning_rate": 8.327839195979899e-05,
-      "loss": 5.4715,
-      "step": 17143
-    },
-    {
-      "epoch": 8.940808344198174,
-      "grad_norm": 1.4105147123336792,
-      "learning_rate": 8.327738693467337e-05,
-      "loss": 5.797,
-      "step": 17144
-    },
-    {
-      "epoch": 8.941329856584094,
-      "grad_norm": 1.4037774801254272,
-      "learning_rate": 8.327638190954774e-05,
-      "loss": 5.0832,
-      "step": 17145
-    },
-    {
-      "epoch": 8.941851368970013,
-      "grad_norm": 1.387237310409546,
-      "learning_rate": 8.327537688442211e-05,
-      "loss": 5.2959,
-      "step": 17146
-    },
-    {
-      "epoch": 8.942372881355933,
-      "grad_norm": 1.4382861852645874,
-      "learning_rate": 8.327437185929649e-05,
-      "loss": 5.4297,
-      "step": 17147
-    },
-    {
-      "epoch": 8.94289439374185,
-      "grad_norm": 1.4246585369110107,
-      "learning_rate": 8.327336683417087e-05,
-      "loss": 5.8101,
-      "step": 17148
-    },
-    {
-      "epoch": 8.94341590612777,
-      "grad_norm": 1.3523129224777222,
-      "learning_rate": 8.327236180904523e-05,
-      "loss": 5.7402,
-      "step": 17149
-    },
-    {
-      "epoch": 8.94393741851369,
-      "grad_norm": 1.4960265159606934,
-      "learning_rate": 8.327135678391961e-05,
-      "loss": 5.2109,
-      "step": 17150
-    },
-    {
-      "epoch": 8.944458930899609,
-      "grad_norm": 1.402561068534851,
-      "learning_rate": 8.327035175879398e-05,
-      "loss": 5.6836,
-      "step": 17151
-    },
-    {
-      "epoch": 8.944980443285528,
-      "grad_norm": 1.3994759321212769,
-      "learning_rate": 8.326934673366834e-05,
-      "loss": 5.6714,
-      "step": 17152
-    },
-    {
-      "epoch": 8.945501955671448,
-      "grad_norm": 1.4177947044372559,
-      "learning_rate": 8.326834170854272e-05,
-      "loss": 5.4891,
-      "step": 17153
-    },
-    {
-      "epoch": 8.946023468057366,
-      "grad_norm": 1.376839518547058,
-      "learning_rate": 8.326733668341708e-05,
-      "loss": 5.7043,
-      "step": 17154
-    },
-    {
-      "epoch": 8.946544980443285,
-      "grad_norm": 1.4017196893692017,
-      "learning_rate": 8.326633165829146e-05,
-      "loss": 5.3827,
-      "step": 17155
-    },
-    {
-      "epoch": 8.947066492829205,
-      "grad_norm": 1.3988215923309326,
-      "learning_rate": 8.326532663316582e-05,
-      "loss": 5.4928,
-      "step": 17156
-    },
-    {
-      "epoch": 8.947588005215124,
-      "grad_norm": 1.4062068462371826,
-      "learning_rate": 8.32643216080402e-05,
-      "loss": 5.5322,
-      "step": 17157
-    },
-    {
-      "epoch": 8.948109517601043,
-      "grad_norm": 1.5087904930114746,
-      "learning_rate": 8.326331658291458e-05,
-      "loss": 5.2961,
-      "step": 17158
-    },
-    {
-      "epoch": 8.948631029986963,
-      "grad_norm": 1.4344910383224487,
-      "learning_rate": 8.326231155778896e-05,
-      "loss": 5.5631,
-      "step": 17159
-    },
-    {
-      "epoch": 8.94915254237288,
-      "grad_norm": 1.397058367729187,
-      "learning_rate": 8.326130653266332e-05,
-      "loss": 5.1979,
-      "step": 17160
-    },
-    {
-      "epoch": 8.9496740547588,
-      "grad_norm": 1.4380309581756592,
-      "learning_rate": 8.32603015075377e-05,
-      "loss": 5.1578,
-      "step": 17161
-    },
-    {
-      "epoch": 8.95019556714472,
-      "grad_norm": 1.4572501182556152,
-      "learning_rate": 8.325929648241206e-05,
-      "loss": 5.9931,
-      "step": 17162
-    },
-    {
-      "epoch": 8.950717079530639,
-      "grad_norm": 1.3975975513458252,
-      "learning_rate": 8.325829145728644e-05,
-      "loss": 5.3183,
-      "step": 17163
-    },
-    {
-      "epoch": 8.951238591916558,
-      "grad_norm": 1.431423306465149,
-      "learning_rate": 8.32572864321608e-05,
-      "loss": 5.472,
-      "step": 17164
-    },
-    {
-      "epoch": 8.951760104302478,
-      "grad_norm": 1.4654908180236816,
-      "learning_rate": 8.325628140703517e-05,
-      "loss": 5.6797,
-      "step": 17165
-    },
-    {
-      "epoch": 8.952281616688396,
-      "grad_norm": 1.4292670488357544,
-      "learning_rate": 8.325527638190955e-05,
-      "loss": 5.6124,
-      "step": 17166
-    },
-    {
-      "epoch": 8.952803129074315,
-      "grad_norm": 1.3400132656097412,
-      "learning_rate": 8.325427135678393e-05,
-      "loss": 5.7285,
-      "step": 17167
-    },
-    {
-      "epoch": 8.953324641460235,
-      "grad_norm": 1.3689686059951782,
-      "learning_rate": 8.32532663316583e-05,
-      "loss": 5.5468,
-      "step": 17168
-    },
-    {
-      "epoch": 8.953846153846154,
-      "grad_norm": 1.452040195465088,
-      "learning_rate": 8.325226130653267e-05,
-      "loss": 5.4681,
-      "step": 17169
-    },
-    {
-      "epoch": 8.954367666232073,
-      "grad_norm": 1.5705493688583374,
-      "learning_rate": 8.325125628140705e-05,
-      "loss": 5.0562,
-      "step": 17170
-    },
-    {
-      "epoch": 8.954889178617993,
-      "grad_norm": 1.4372966289520264,
-      "learning_rate": 8.325025125628141e-05,
-      "loss": 5.3552,
-      "step": 17171
-    },
-    {
-      "epoch": 8.95541069100391,
-      "grad_norm": 1.3601624965667725,
-      "learning_rate": 8.324924623115579e-05,
-      "loss": 5.5806,
-      "step": 17172
-    },
-    {
-      "epoch": 8.95593220338983,
-      "grad_norm": 1.4037408828735352,
-      "learning_rate": 8.324824120603015e-05,
-      "loss": 5.25,
-      "step": 17173
-    },
-    {
-      "epoch": 8.95645371577575,
-      "grad_norm": 1.609368085861206,
-      "learning_rate": 8.324723618090453e-05,
-      "loss": 5.4643,
-      "step": 17174
-    },
-    {
-      "epoch": 8.956975228161669,
-      "grad_norm": 1.3728997707366943,
-      "learning_rate": 8.32462311557789e-05,
-      "loss": 5.878,
-      "step": 17175
-    },
-    {
-      "epoch": 8.957496740547588,
-      "grad_norm": 1.404362678527832,
-      "learning_rate": 8.324522613065327e-05,
-      "loss": 5.4491,
-      "step": 17176
-    },
-    {
-      "epoch": 8.958018252933508,
-      "grad_norm": 1.4871209859848022,
-      "learning_rate": 8.324422110552764e-05,
-      "loss": 5.4258,
-      "step": 17177
-    },
-    {
-      "epoch": 8.958539765319426,
-      "grad_norm": 1.4671587944030762,
-      "learning_rate": 8.324321608040201e-05,
-      "loss": 5.1731,
-      "step": 17178
-    },
-    {
-      "epoch": 8.959061277705345,
-      "grad_norm": 1.4353214502334595,
-      "learning_rate": 8.324221105527639e-05,
-      "loss": 5.6285,
-      "step": 17179
-    },
-    {
-      "epoch": 8.959582790091265,
-      "grad_norm": 1.5662236213684082,
-      "learning_rate": 8.324120603015076e-05,
-      "loss": 5.4203,
-      "step": 17180
-    },
-    {
-      "epoch": 8.960104302477184,
-      "grad_norm": 1.4547046422958374,
-      "learning_rate": 8.324020100502513e-05,
-      "loss": 5.2743,
-      "step": 17181
-    },
-    {
-      "epoch": 8.960625814863103,
-      "grad_norm": 1.475999355316162,
-      "learning_rate": 8.32391959798995e-05,
-      "loss": 5.459,
-      "step": 17182
-    },
-    {
-      "epoch": 8.961147327249023,
-      "grad_norm": 1.5241751670837402,
-      "learning_rate": 8.323819095477388e-05,
-      "loss": 5.6888,
-      "step": 17183
-    },
-    {
-      "epoch": 8.96166883963494,
-      "grad_norm": 1.439688801765442,
-      "learning_rate": 8.323718592964824e-05,
-      "loss": 5.8426,
-      "step": 17184
-    },
-    {
-      "epoch": 8.96219035202086,
-      "grad_norm": 1.4469650983810425,
-      "learning_rate": 8.323618090452262e-05,
-      "loss": 5.7653,
-      "step": 17185
-    },
-    {
-      "epoch": 8.96271186440678,
-      "grad_norm": 1.5088584423065186,
-      "learning_rate": 8.323517587939698e-05,
-      "loss": 5.8094,
-      "step": 17186
-    },
-    {
-      "epoch": 8.963233376792699,
-      "grad_norm": 1.3987655639648438,
-      "learning_rate": 8.323417085427136e-05,
-      "loss": 5.6686,
-      "step": 17187
-    },
-    {
-      "epoch": 8.963754889178619,
-      "grad_norm": 1.471657633781433,
-      "learning_rate": 8.323316582914574e-05,
-      "loss": 5.4917,
-      "step": 17188
-    },
-    {
-      "epoch": 8.964276401564538,
-      "grad_norm": 1.5371453762054443,
-      "learning_rate": 8.323216080402012e-05,
-      "loss": 5.5069,
-      "step": 17189
-    },
-    {
-      "epoch": 8.964797913950456,
-      "grad_norm": 1.4816153049468994,
-      "learning_rate": 8.323115577889448e-05,
-      "loss": 5.6106,
-      "step": 17190
-    },
-    {
-      "epoch": 8.965319426336375,
-      "grad_norm": 1.3056435585021973,
-      "learning_rate": 8.323015075376884e-05,
-      "loss": 5.6304,
-      "step": 17191
-    },
-    {
-      "epoch": 8.965840938722295,
-      "grad_norm": 1.4310269355773926,
-      "learning_rate": 8.322914572864322e-05,
-      "loss": 5.7725,
-      "step": 17192
-    },
-    {
-      "epoch": 8.966362451108214,
-      "grad_norm": 1.3863972425460815,
-      "learning_rate": 8.322814070351759e-05,
-      "loss": 5.4714,
-      "step": 17193
-    },
-    {
-      "epoch": 8.966883963494134,
-      "grad_norm": 1.6251235008239746,
-      "learning_rate": 8.322713567839196e-05,
-      "loss": 5.3596,
-      "step": 17194
-    },
-    {
-      "epoch": 8.967405475880053,
-      "grad_norm": 1.4058809280395508,
-      "learning_rate": 8.322613065326633e-05,
-      "loss": 5.3678,
-      "step": 17195
-    },
-    {
-      "epoch": 8.96792698826597,
-      "grad_norm": 1.4945447444915771,
-      "learning_rate": 8.32251256281407e-05,
-      "loss": 5.4943,
-      "step": 17196
-    },
-    {
-      "epoch": 8.96844850065189,
-      "grad_norm": 1.4155802726745605,
-      "learning_rate": 8.322412060301507e-05,
-      "loss": 5.3265,
-      "step": 17197
-    },
-    {
-      "epoch": 8.96897001303781,
-      "grad_norm": 1.5083956718444824,
-      "learning_rate": 8.322311557788945e-05,
-      "loss": 5.5104,
-      "step": 17198
-    },
-    {
-      "epoch": 8.969491525423729,
-      "grad_norm": 1.5089318752288818,
-      "learning_rate": 8.322211055276383e-05,
-      "loss": 5.0366,
-      "step": 17199
-    },
-    {
-      "epoch": 8.970013037809649,
-      "grad_norm": 1.4194135665893555,
-      "learning_rate": 8.32211055276382e-05,
-      "loss": 5.6467,
-      "step": 17200
-    },
-    {
-      "epoch": 8.970534550195566,
-      "grad_norm": 1.4024479389190674,
-      "learning_rate": 8.322010050251257e-05,
-      "loss": 5.5535,
-      "step": 17201
-    },
-    {
-      "epoch": 8.971056062581486,
-      "grad_norm": 1.4617587327957153,
-      "learning_rate": 8.321909547738695e-05,
-      "loss": 5.131,
-      "step": 17202
-    },
-    {
-      "epoch": 8.971577574967405,
-      "grad_norm": 1.3386296033859253,
-      "learning_rate": 8.321809045226131e-05,
-      "loss": 5.2928,
-      "step": 17203
-    },
-    {
-      "epoch": 8.972099087353325,
-      "grad_norm": 1.3522703647613525,
-      "learning_rate": 8.321708542713567e-05,
-      "loss": 4.5587,
-      "step": 17204
-    },
-    {
-      "epoch": 8.972620599739244,
-      "grad_norm": 1.4241321086883545,
-      "learning_rate": 8.321608040201005e-05,
-      "loss": 5.3886,
-      "step": 17205
-    },
-    {
-      "epoch": 8.973142112125164,
-      "grad_norm": 1.6165413856506348,
-      "learning_rate": 8.321507537688442e-05,
-      "loss": 5.2495,
-      "step": 17206
-    },
-    {
-      "epoch": 8.973663624511083,
-      "grad_norm": 1.4865392446517944,
-      "learning_rate": 8.32140703517588e-05,
-      "loss": 5.6458,
-      "step": 17207
-    },
-    {
-      "epoch": 8.974185136897,
-      "grad_norm": 1.3910518884658813,
-      "learning_rate": 8.321306532663317e-05,
-      "loss": 5.6272,
-      "step": 17208
-    },
-    {
-      "epoch": 8.97470664928292,
-      "grad_norm": 1.538120150566101,
-      "learning_rate": 8.321206030150755e-05,
-      "loss": 4.7425,
-      "step": 17209
-    },
-    {
-      "epoch": 8.97522816166884,
-      "grad_norm": 1.379615306854248,
-      "learning_rate": 8.321105527638191e-05,
-      "loss": 5.5212,
-      "step": 17210
-    },
-    {
-      "epoch": 8.975749674054759,
-      "grad_norm": 1.3391060829162598,
-      "learning_rate": 8.321005025125629e-05,
-      "loss": 5.874,
-      "step": 17211
-    },
-    {
-      "epoch": 8.976271186440679,
-      "grad_norm": 1.5537376403808594,
-      "learning_rate": 8.320904522613066e-05,
-      "loss": 5.3268,
-      "step": 17212
-    },
-    {
-      "epoch": 8.976792698826596,
-      "grad_norm": 1.719254732131958,
-      "learning_rate": 8.320804020100503e-05,
-      "loss": 4.8647,
-      "step": 17213
-    },
-    {
-      "epoch": 8.977314211212516,
-      "grad_norm": 1.3998106718063354,
-      "learning_rate": 8.32070351758794e-05,
-      "loss": 5.1449,
-      "step": 17214
-    },
-    {
-      "epoch": 8.977835723598435,
-      "grad_norm": 1.490836501121521,
-      "learning_rate": 8.320603015075378e-05,
-      "loss": 5.778,
-      "step": 17215
-    },
-    {
-      "epoch": 8.978357235984355,
-      "grad_norm": 1.3912994861602783,
-      "learning_rate": 8.320502512562814e-05,
-      "loss": 5.5397,
-      "step": 17216
-    },
-    {
-      "epoch": 8.978878748370274,
-      "grad_norm": 1.4528592824935913,
-      "learning_rate": 8.32040201005025e-05,
-      "loss": 5.7737,
-      "step": 17217
-    },
-    {
-      "epoch": 8.979400260756194,
-      "grad_norm": 1.454712986946106,
-      "learning_rate": 8.320301507537688e-05,
-      "loss": 5.1977,
-      "step": 17218
-    },
-    {
-      "epoch": 8.979921773142113,
-      "grad_norm": 1.427297830581665,
-      "learning_rate": 8.320201005025126e-05,
-      "loss": 5.6389,
-      "step": 17219
-    },
-    {
-      "epoch": 8.98044328552803,
-      "grad_norm": 1.6764613389968872,
-      "learning_rate": 8.320100502512564e-05,
-      "loss": 5.2489,
-      "step": 17220
-    },
-    {
-      "epoch": 8.98096479791395,
-      "grad_norm": 1.4285928010940552,
-      "learning_rate": 8.32e-05,
-      "loss": 5.6645,
-      "step": 17221
-    },
-    {
-      "epoch": 8.98148631029987,
-      "grad_norm": 1.5021356344223022,
-      "learning_rate": 8.319899497487438e-05,
-      "loss": 5.5771,
-      "step": 17222
-    },
-    {
-      "epoch": 8.98200782268579,
-      "grad_norm": 1.5852656364440918,
-      "learning_rate": 8.319798994974875e-05,
-      "loss": 5.3452,
-      "step": 17223
-    },
-    {
-      "epoch": 8.982529335071709,
-      "grad_norm": 1.424446940422058,
-      "learning_rate": 8.319698492462312e-05,
-      "loss": 5.664,
-      "step": 17224
-    },
-    {
-      "epoch": 8.983050847457626,
-      "grad_norm": 1.4138952493667603,
-      "learning_rate": 8.319597989949749e-05,
-      "loss": 5.0664,
-      "step": 17225
-    },
-    {
-      "epoch": 8.983572359843546,
-      "grad_norm": 1.50666344165802,
-      "learning_rate": 8.319497487437187e-05,
-      "loss": 5.565,
-      "step": 17226
-    },
-    {
-      "epoch": 8.984093872229465,
-      "grad_norm": 1.3904057741165161,
-      "learning_rate": 8.319396984924623e-05,
-      "loss": 5.6808,
-      "step": 17227
-    },
-    {
-      "epoch": 8.984615384615385,
-      "grad_norm": 1.38749098777771,
-      "learning_rate": 8.319296482412061e-05,
-      "loss": 5.6827,
-      "step": 17228
-    },
-    {
-      "epoch": 8.985136897001304,
-      "grad_norm": 1.340425729751587,
-      "learning_rate": 8.319195979899498e-05,
-      "loss": 5.8358,
-      "step": 17229
-    },
-    {
-      "epoch": 8.985658409387224,
-      "grad_norm": 1.5228389501571655,
-      "learning_rate": 8.319095477386936e-05,
-      "loss": 5.3213,
-      "step": 17230
-    },
-    {
-      "epoch": 8.986179921773141,
-      "grad_norm": 1.5093159675598145,
-      "learning_rate": 8.318994974874373e-05,
-      "loss": 5.0826,
-      "step": 17231
-    },
-    {
-      "epoch": 8.98670143415906,
-      "grad_norm": 1.4177839756011963,
-      "learning_rate": 8.318894472361809e-05,
-      "loss": 4.9367,
-      "step": 17232
-    },
-    {
-      "epoch": 8.98722294654498,
-      "grad_norm": 1.586360216140747,
-      "learning_rate": 8.318793969849247e-05,
-      "loss": 5.6794,
-      "step": 17233
-    },
-    {
-      "epoch": 8.9877444589309,
-      "grad_norm": 1.5562132596969604,
-      "learning_rate": 8.318693467336683e-05,
-      "loss": 5.641,
-      "step": 17234
-    },
-    {
-      "epoch": 8.98826597131682,
-      "grad_norm": 1.4377762079238892,
-      "learning_rate": 8.318592964824121e-05,
-      "loss": 5.395,
-      "step": 17235
-    },
-    {
-      "epoch": 8.988787483702739,
-      "grad_norm": 1.4195791482925415,
-      "learning_rate": 8.318492462311558e-05,
-      "loss": 5.5753,
-      "step": 17236
-    },
-    {
-      "epoch": 8.989308996088656,
-      "grad_norm": 1.4816346168518066,
-      "learning_rate": 8.318391959798995e-05,
-      "loss": 5.6385,
-      "step": 17237
-    },
-    {
-      "epoch": 8.989830508474576,
-      "grad_norm": 1.4216594696044922,
-      "learning_rate": 8.318291457286432e-05,
-      "loss": 5.3342,
-      "step": 17238
-    },
-    {
-      "epoch": 8.990352020860495,
-      "grad_norm": 1.2920938730239868,
-      "learning_rate": 8.31819095477387e-05,
-      "loss": 4.6397,
-      "step": 17239
-    },
-    {
-      "epoch": 8.990873533246415,
-      "grad_norm": 1.636558175086975,
-      "learning_rate": 8.318090452261307e-05,
-      "loss": 5.5482,
-      "step": 17240
-    },
-    {
-      "epoch": 8.991395045632334,
-      "grad_norm": 1.4978289604187012,
-      "learning_rate": 8.317989949748745e-05,
-      "loss": 5.1398,
-      "step": 17241
-    },
-    {
-      "epoch": 8.991916558018254,
-      "grad_norm": 1.4544918537139893,
-      "learning_rate": 8.317889447236182e-05,
-      "loss": 5.8264,
-      "step": 17242
-    },
-    {
-      "epoch": 8.992438070404171,
-      "grad_norm": 1.4580154418945312,
-      "learning_rate": 8.31778894472362e-05,
-      "loss": 5.5585,
-      "step": 17243
-    },
-    {
-      "epoch": 8.99295958279009,
-      "grad_norm": 1.3980594873428345,
-      "learning_rate": 8.317688442211056e-05,
-      "loss": 5.4949,
-      "step": 17244
-    },
-    {
-      "epoch": 8.99348109517601,
-      "grad_norm": 1.5808533430099487,
-      "learning_rate": 8.317587939698492e-05,
-      "loss": 5.4413,
-      "step": 17245
-    },
-    {
-      "epoch": 8.99400260756193,
-      "grad_norm": 1.5741877555847168,
-      "learning_rate": 8.31748743718593e-05,
-      "loss": 5.6705,
-      "step": 17246
-    },
-    {
-      "epoch": 8.99452411994785,
-      "grad_norm": 1.430740237236023,
-      "learning_rate": 8.317386934673366e-05,
-      "loss": 6.04,
-      "step": 17247
-    },
-    {
-      "epoch": 8.995045632333769,
-      "grad_norm": 1.486871361732483,
-      "learning_rate": 8.317286432160804e-05,
-      "loss": 5.6867,
-      "step": 17248
-    },
-    {
-      "epoch": 8.995567144719686,
-      "grad_norm": 1.4932712316513062,
-      "learning_rate": 8.317185929648242e-05,
-      "loss": 5.8603,
-      "step": 17249
-    },
-    {
-      "epoch": 8.996088657105606,
-      "grad_norm": 1.428758144378662,
-      "learning_rate": 8.31708542713568e-05,
-      "loss": 5.8367,
-      "step": 17250
-    },
-    {
-      "epoch": 8.996610169491525,
-      "grad_norm": 1.4661887884140015,
-      "learning_rate": 8.316984924623116e-05,
-      "loss": 5.2528,
-      "step": 17251
-    },
-    {
-      "epoch": 8.997131681877445,
-      "grad_norm": 1.3328206539154053,
-      "learning_rate": 8.316884422110554e-05,
-      "loss": 5.4592,
-      "step": 17252
-    },
-    {
-      "epoch": 8.997653194263364,
-      "grad_norm": 1.4775805473327637,
-      "learning_rate": 8.31678391959799e-05,
-      "loss": 5.8215,
-      "step": 17253
-    },
-    {
-      "epoch": 8.998174706649284,
-      "grad_norm": 1.4234619140625,
-      "learning_rate": 8.316683417085428e-05,
-      "loss": 5.8893,
-      "step": 17254
-    },
-    {
-      "epoch": 8.998696219035201,
-      "grad_norm": 1.4394845962524414,
-      "learning_rate": 8.316582914572865e-05,
-      "loss": 5.51,
-      "step": 17255
-    },
-    {
-      "epoch": 8.99921773142112,
-      "grad_norm": 1.3854269981384277,
-      "learning_rate": 8.316482412060302e-05,
-      "loss": 5.4478,
-      "step": 17256
-    },
-    {
-      "epoch": 8.99973924380704,
-      "grad_norm": 1.4971160888671875,
-      "learning_rate": 8.316381909547739e-05,
-      "loss": 5.1449,
-      "step": 17257
-    },
-    {
-      "epoch": 9.00026075619296,
-      "grad_norm": 1.5786570310592651,
-      "learning_rate": 8.316281407035175e-05,
-      "loss": 5.5181,
-      "step": 17258
-    },
-    {
-      "epoch": 9.00078226857888,
-      "grad_norm": 1.384230136871338,
-      "learning_rate": 8.316180904522613e-05,
-      "loss": 5.6279,
-      "step": 17259
-    },
-    {
-      "epoch": 9.001303780964799,
-      "grad_norm": 1.6187585592269897,
-      "learning_rate": 8.316080402010051e-05,
-      "loss": 5.567,
-      "step": 17260
-    },
-    {
-      "epoch": 9.001825293350716,
-      "grad_norm": 1.3567012548446655,
-      "learning_rate": 8.315979899497489e-05,
-      "loss": 5.4745,
-      "step": 17261
-    },
-    {
-      "epoch": 9.002346805736636,
-      "grad_norm": 1.606665849685669,
-      "learning_rate": 8.315879396984925e-05,
-      "loss": 5.2077,
-      "step": 17262
-    },
-    {
-      "epoch": 9.002868318122555,
-      "grad_norm": 1.4134471416473389,
-      "learning_rate": 8.315778894472363e-05,
-      "loss": 5.0948,
-      "step": 17263
-    },
-    {
-      "epoch": 9.003389830508475,
-      "grad_norm": 1.4917067289352417,
-      "learning_rate": 8.315678391959799e-05,
-      "loss": 5.6457,
-      "step": 17264
-    },
-    {
-      "epoch": 9.003911342894394,
-      "grad_norm": 1.3551150560379028,
-      "learning_rate": 8.315577889447237e-05,
-      "loss": 5.7812,
-      "step": 17265
-    },
-    {
-      "epoch": 9.004432855280314,
-      "grad_norm": 1.3671259880065918,
-      "learning_rate": 8.315477386934673e-05,
-      "loss": 5.7403,
-      "step": 17266
-    },
-    {
-      "epoch": 9.004954367666231,
-      "grad_norm": 1.444583773612976,
-      "learning_rate": 8.315376884422111e-05,
-      "loss": 5.6212,
-      "step": 17267
-    },
-    {
-      "epoch": 9.00547588005215,
-      "grad_norm": 1.4186958074569702,
-      "learning_rate": 8.315276381909548e-05,
-      "loss": 5.7717,
-      "step": 17268
-    },
-    {
-      "epoch": 9.00599739243807,
-      "grad_norm": 1.2992233037948608,
-      "learning_rate": 8.315175879396985e-05,
-      "loss": 5.7591,
-      "step": 17269
-    },
-    {
-      "epoch": 9.00651890482399,
-      "grad_norm": 1.2047125101089478,
-      "learning_rate": 8.315075376884422e-05,
-      "loss": 4.8728,
-      "step": 17270
-    },
-    {
-      "epoch": 9.00704041720991,
-      "grad_norm": 1.3439866304397583,
-      "learning_rate": 8.31497487437186e-05,
-      "loss": 4.9014,
-      "step": 17271
-    },
-    {
-      "epoch": 9.007561929595829,
-      "grad_norm": 1.5586837530136108,
-      "learning_rate": 8.314874371859297e-05,
-      "loss": 5.1273,
-      "step": 17272
-    },
-    {
-      "epoch": 9.008083441981746,
-      "grad_norm": 1.4726762771606445,
-      "learning_rate": 8.314773869346734e-05,
-      "loss": 5.3694,
-      "step": 17273
-    },
-    {
-      "epoch": 9.008604954367666,
-      "grad_norm": 1.4158573150634766,
-      "learning_rate": 8.314673366834172e-05,
-      "loss": 5.9375,
-      "step": 17274
-    },
-    {
-      "epoch": 9.009126466753585,
-      "grad_norm": 1.310899019241333,
-      "learning_rate": 8.314572864321608e-05,
-      "loss": 5.6242,
-      "step": 17275
-    },
-    {
-      "epoch": 9.009647979139505,
-      "grad_norm": 1.2785108089447021,
-      "learning_rate": 8.314472361809046e-05,
-      "loss": 5.4763,
-      "step": 17276
-    },
-    {
-      "epoch": 9.010169491525424,
-      "grad_norm": 1.3256282806396484,
-      "learning_rate": 8.314371859296482e-05,
-      "loss": 5.9303,
-      "step": 17277
-    },
-    {
-      "epoch": 9.010691003911344,
-      "grad_norm": 1.2801897525787354,
-      "learning_rate": 8.31427135678392e-05,
-      "loss": 5.8318,
-      "step": 17278
-    },
-    {
-      "epoch": 9.011212516297261,
-      "grad_norm": 1.3779349327087402,
-      "learning_rate": 8.314170854271356e-05,
-      "loss": 5.8494,
-      "step": 17279
-    },
-    {
-      "epoch": 9.01173402868318,
-      "grad_norm": 1.4310824871063232,
-      "learning_rate": 8.314070351758794e-05,
-      "loss": 5.7356,
-      "step": 17280
-    },
-    {
-      "epoch": 9.0122555410691,
-      "grad_norm": 1.418982744216919,
-      "learning_rate": 8.313969849246232e-05,
-      "loss": 5.8391,
-      "step": 17281
-    },
-    {
-      "epoch": 9.01277705345502,
-      "grad_norm": 1.5339874029159546,
-      "learning_rate": 8.31386934673367e-05,
-      "loss": 5.0038,
-      "step": 17282
-    },
-    {
-      "epoch": 9.01329856584094,
-      "grad_norm": 1.429993987083435,
-      "learning_rate": 8.313768844221106e-05,
-      "loss": 5.7216,
-      "step": 17283
-    },
-    {
-      "epoch": 9.013820078226859,
-      "grad_norm": 1.5197460651397705,
-      "learning_rate": 8.313668341708543e-05,
-      "loss": 4.9276,
-      "step": 17284
-    },
-    {
-      "epoch": 9.014341590612776,
-      "grad_norm": 1.3862816095352173,
-      "learning_rate": 8.31356783919598e-05,
-      "loss": 5.5973,
-      "step": 17285
-    },
-    {
-      "epoch": 9.014863102998696,
-      "grad_norm": 1.3136045932769775,
-      "learning_rate": 8.313467336683417e-05,
-      "loss": 6.0376,
-      "step": 17286
-    },
-    {
-      "epoch": 9.015384615384615,
-      "grad_norm": 1.3364133834838867,
-      "learning_rate": 8.313366834170855e-05,
-      "loss": 5.9168,
-      "step": 17287
-    },
-    {
-      "epoch": 9.015906127770535,
-      "grad_norm": 1.4384169578552246,
-      "learning_rate": 8.313266331658291e-05,
-      "loss": 5.5916,
-      "step": 17288
-    },
-    {
-      "epoch": 9.016427640156454,
-      "grad_norm": 1.4226306676864624,
-      "learning_rate": 8.313165829145729e-05,
-      "loss": 5.1922,
-      "step": 17289
-    },
-    {
-      "epoch": 9.016949152542374,
-      "grad_norm": 1.2863487005233765,
-      "learning_rate": 8.313065326633165e-05,
-      "loss": 5.9147,
-      "step": 17290
-    },
-    {
-      "epoch": 9.017470664928291,
-      "grad_norm": 1.5965043306350708,
-      "learning_rate": 8.312964824120603e-05,
-      "loss": 5.3932,
-      "step": 17291
-    },
-    {
-      "epoch": 9.01799217731421,
-      "grad_norm": 1.622576117515564,
-      "learning_rate": 8.312864321608041e-05,
-      "loss": 4.6088,
-      "step": 17292
-    },
-    {
-      "epoch": 9.01851368970013,
-      "grad_norm": 1.4195871353149414,
-      "learning_rate": 8.312763819095479e-05,
-      "loss": 5.5237,
-      "step": 17293
-    },
-    {
-      "epoch": 9.01903520208605,
-      "grad_norm": 1.4993692636489868,
-      "learning_rate": 8.312663316582915e-05,
-      "loss": 5.5925,
-      "step": 17294
-    },
-    {
-      "epoch": 9.01955671447197,
-      "grad_norm": 1.4799340963363647,
-      "learning_rate": 8.312562814070353e-05,
-      "loss": 5.3647,
-      "step": 17295
-    },
-    {
-      "epoch": 9.020078226857889,
-      "grad_norm": 1.5004442930221558,
-      "learning_rate": 8.312462311557789e-05,
-      "loss": 5.459,
-      "step": 17296
-    },
-    {
-      "epoch": 9.020599739243806,
-      "grad_norm": 1.4664863348007202,
-      "learning_rate": 8.312361809045226e-05,
-      "loss": 5.3069,
-      "step": 17297
-    },
-    {
-      "epoch": 9.021121251629726,
-      "grad_norm": 1.2745188474655151,
-      "learning_rate": 8.312261306532663e-05,
-      "loss": 5.5265,
-      "step": 17298
-    },
-    {
-      "epoch": 9.021642764015645,
-      "grad_norm": 1.3095803260803223,
-      "learning_rate": 8.3121608040201e-05,
-      "loss": 5.3468,
-      "step": 17299
-    },
-    {
-      "epoch": 9.022164276401565,
-      "grad_norm": 1.4460837841033936,
-      "learning_rate": 8.312060301507538e-05,
-      "loss": 5.4027,
-      "step": 17300
-    },
-    {
-      "epoch": 9.022685788787484,
-      "grad_norm": 1.3879367113113403,
-      "learning_rate": 8.311959798994975e-05,
-      "loss": 5.5874,
-      "step": 17301
-    },
-    {
-      "epoch": 9.023207301173404,
-      "grad_norm": 1.3812978267669678,
-      "learning_rate": 8.311859296482413e-05,
-      "loss": 5.5985,
-      "step": 17302
-    },
-    {
-      "epoch": 9.023728813559321,
-      "grad_norm": 1.3877918720245361,
-      "learning_rate": 8.31175879396985e-05,
-      "loss": 5.5825,
-      "step": 17303
-    },
-    {
-      "epoch": 9.024250325945241,
-      "grad_norm": 1.4428832530975342,
-      "learning_rate": 8.311658291457287e-05,
-      "loss": 5.5169,
-      "step": 17304
-    },
-    {
-      "epoch": 9.02477183833116,
-      "grad_norm": 1.5336729288101196,
-      "learning_rate": 8.311557788944724e-05,
-      "loss": 5.0039,
-      "step": 17305
-    },
-    {
-      "epoch": 9.02529335071708,
-      "grad_norm": 1.446858525276184,
-      "learning_rate": 8.311457286432162e-05,
-      "loss": 5.1676,
-      "step": 17306
-    },
-    {
-      "epoch": 9.025814863103,
-      "grad_norm": 1.446319818496704,
-      "learning_rate": 8.311356783919598e-05,
-      "loss": 5.5209,
-      "step": 17307
-    },
-    {
-      "epoch": 9.026336375488917,
-      "grad_norm": 1.50642991065979,
-      "learning_rate": 8.311256281407036e-05,
-      "loss": 5.2162,
-      "step": 17308
-    },
-    {
-      "epoch": 9.026857887874836,
-      "grad_norm": 1.4276093244552612,
-      "learning_rate": 8.311155778894472e-05,
-      "loss": 4.945,
-      "step": 17309
-    },
-    {
-      "epoch": 9.027379400260756,
-      "grad_norm": 1.4663605690002441,
-      "learning_rate": 8.311055276381909e-05,
-      "loss": 5.4768,
-      "step": 17310
-    },
-    {
-      "epoch": 9.027900912646675,
-      "grad_norm": 1.2230736017227173,
-      "learning_rate": 8.310954773869347e-05,
-      "loss": 5.9437,
-      "step": 17311
-    },
-    {
-      "epoch": 9.028422425032595,
-      "grad_norm": 1.4952106475830078,
-      "learning_rate": 8.310854271356784e-05,
-      "loss": 5.3185,
-      "step": 17312
-    },
-    {
-      "epoch": 9.028943937418514,
-      "grad_norm": 1.389928936958313,
-      "learning_rate": 8.310753768844222e-05,
-      "loss": 5.5904,
-      "step": 17313
-    },
-    {
-      "epoch": 9.029465449804432,
-      "grad_norm": 1.3805091381072998,
-      "learning_rate": 8.310653266331659e-05,
-      "loss": 5.4408,
-      "step": 17314
-    },
-    {
-      "epoch": 9.029986962190351,
-      "grad_norm": 1.4228839874267578,
-      "learning_rate": 8.310552763819096e-05,
-      "loss": 5.9412,
-      "step": 17315
-    },
-    {
-      "epoch": 9.030508474576271,
-      "grad_norm": 1.4413777589797974,
-      "learning_rate": 8.310452261306533e-05,
-      "loss": 5.201,
-      "step": 17316
-    },
-    {
-      "epoch": 9.03102998696219,
-      "grad_norm": 1.4316178560256958,
-      "learning_rate": 8.31035175879397e-05,
-      "loss": 5.7298,
-      "step": 17317
-    },
-    {
-      "epoch": 9.03155149934811,
-      "grad_norm": 1.4059079885482788,
-      "learning_rate": 8.310251256281407e-05,
-      "loss": 5.9132,
-      "step": 17318
-    },
-    {
-      "epoch": 9.03207301173403,
-      "grad_norm": 1.5063568353652954,
-      "learning_rate": 8.310150753768845e-05,
-      "loss": 5.6178,
-      "step": 17319
-    },
-    {
-      "epoch": 9.032594524119947,
-      "grad_norm": 1.3380324840545654,
-      "learning_rate": 8.310050251256281e-05,
-      "loss": 5.7459,
-      "step": 17320
-    },
-    {
-      "epoch": 9.033116036505866,
-      "grad_norm": 1.4272377490997314,
-      "learning_rate": 8.309949748743719e-05,
-      "loss": 5.2959,
-      "step": 17321
-    },
-    {
-      "epoch": 9.033637548891786,
-      "grad_norm": 1.2505565881729126,
-      "learning_rate": 8.309849246231157e-05,
-      "loss": 5.6842,
-      "step": 17322
-    },
-    {
-      "epoch": 9.034159061277705,
-      "grad_norm": 1.3441941738128662,
-      "learning_rate": 8.309748743718595e-05,
-      "loss": 5.7448,
-      "step": 17323
-    },
-    {
-      "epoch": 9.034680573663625,
-      "grad_norm": 1.358316421508789,
-      "learning_rate": 8.309648241206031e-05,
-      "loss": 5.6825,
-      "step": 17324
-    },
-    {
-      "epoch": 9.035202086049544,
-      "grad_norm": 1.3717707395553589,
-      "learning_rate": 8.309547738693467e-05,
-      "loss": 5.4427,
-      "step": 17325
-    },
-    {
-      "epoch": 9.035723598435462,
-      "grad_norm": 1.417128324508667,
-      "learning_rate": 8.309447236180905e-05,
-      "loss": 5.6936,
-      "step": 17326
-    },
-    {
-      "epoch": 9.036245110821381,
-      "grad_norm": 1.3807530403137207,
-      "learning_rate": 8.309346733668342e-05,
-      "loss": 5.5494,
-      "step": 17327
-    },
-    {
-      "epoch": 9.036766623207301,
-      "grad_norm": 1.49567449092865,
-      "learning_rate": 8.30924623115578e-05,
-      "loss": 5.1268,
-      "step": 17328
-    },
-    {
-      "epoch": 9.03728813559322,
-      "grad_norm": 1.4016844034194946,
-      "learning_rate": 8.309145728643216e-05,
-      "loss": 5.6063,
-      "step": 17329
-    },
-    {
-      "epoch": 9.03780964797914,
-      "grad_norm": 1.5244462490081787,
-      "learning_rate": 8.309045226130654e-05,
-      "loss": 5.1669,
-      "step": 17330
-    },
-    {
-      "epoch": 9.03833116036506,
-      "grad_norm": 1.5828564167022705,
-      "learning_rate": 8.30894472361809e-05,
-      "loss": 5.0767,
-      "step": 17331
-    },
-    {
-      "epoch": 9.038852672750977,
-      "grad_norm": 1.3195710182189941,
-      "learning_rate": 8.308844221105528e-05,
-      "loss": 5.6305,
-      "step": 17332
-    },
-    {
-      "epoch": 9.039374185136897,
-      "grad_norm": 1.3506046533584595,
-      "learning_rate": 8.308743718592966e-05,
-      "loss": 5.714,
-      "step": 17333
-    },
-    {
-      "epoch": 9.039895697522816,
-      "grad_norm": 1.5116578340530396,
-      "learning_rate": 8.308643216080403e-05,
-      "loss": 5.0102,
-      "step": 17334
-    },
-    {
-      "epoch": 9.040417209908735,
-      "grad_norm": 1.4128819704055786,
-      "learning_rate": 8.30854271356784e-05,
-      "loss": 5.0252,
-      "step": 17335
-    },
-    {
-      "epoch": 9.040938722294655,
-      "grad_norm": 1.451920509338379,
-      "learning_rate": 8.308442211055278e-05,
-      "loss": 4.8412,
-      "step": 17336
-    },
-    {
-      "epoch": 9.041460234680574,
-      "grad_norm": 1.5551118850708008,
-      "learning_rate": 8.308341708542714e-05,
-      "loss": 5.1303,
-      "step": 17337
-    },
-    {
-      "epoch": 9.041981747066492,
-      "grad_norm": 1.519775152206421,
-      "learning_rate": 8.30824120603015e-05,
-      "loss": 5.2272,
-      "step": 17338
-    },
-    {
-      "epoch": 9.042503259452412,
-      "grad_norm": 1.5881339311599731,
-      "learning_rate": 8.308140703517588e-05,
-      "loss": 5.0547,
-      "step": 17339
-    },
-    {
-      "epoch": 9.043024771838331,
-      "grad_norm": 1.3865727186203003,
-      "learning_rate": 8.308040201005025e-05,
-      "loss": 5.3989,
-      "step": 17340
-    },
-    {
-      "epoch": 9.04354628422425,
-      "grad_norm": 1.478986144065857,
-      "learning_rate": 8.307939698492462e-05,
-      "loss": 5.2753,
-      "step": 17341
-    },
-    {
-      "epoch": 9.04406779661017,
-      "grad_norm": 1.4215189218521118,
-      "learning_rate": 8.3078391959799e-05,
-      "loss": 5.7989,
-      "step": 17342
-    },
-    {
-      "epoch": 9.04458930899609,
-      "grad_norm": 1.4156526327133179,
-      "learning_rate": 8.307738693467338e-05,
-      "loss": 5.6127,
-      "step": 17343
-    },
-    {
-      "epoch": 9.045110821382007,
-      "grad_norm": 1.3703389167785645,
-      "learning_rate": 8.307638190954774e-05,
-      "loss": 5.5435,
-      "step": 17344
-    },
-    {
-      "epoch": 9.045632333767927,
-      "grad_norm": 1.310840129852295,
-      "learning_rate": 8.307537688442212e-05,
-      "loss": 5.9149,
-      "step": 17345
-    },
-    {
-      "epoch": 9.046153846153846,
-      "grad_norm": 1.5163453817367554,
-      "learning_rate": 8.307437185929649e-05,
-      "loss": 5.2806,
-      "step": 17346
-    },
-    {
-      "epoch": 9.046675358539765,
-      "grad_norm": 1.6051056385040283,
-      "learning_rate": 8.307336683417086e-05,
-      "loss": 5.1681,
-      "step": 17347
-    },
-    {
-      "epoch": 9.047196870925685,
-      "grad_norm": 1.460837721824646,
-      "learning_rate": 8.307236180904523e-05,
-      "loss": 5.86,
-      "step": 17348
-    },
-    {
-      "epoch": 9.047718383311604,
-      "grad_norm": 1.6169523000717163,
-      "learning_rate": 8.30713567839196e-05,
-      "loss": 5.2894,
-      "step": 17349
-    },
-    {
-      "epoch": 9.048239895697522,
-      "grad_norm": 1.3713034391403198,
-      "learning_rate": 8.307035175879397e-05,
-      "loss": 5.6633,
-      "step": 17350
-    },
-    {
-      "epoch": 9.048761408083442,
-      "grad_norm": 1.4699265956878662,
-      "learning_rate": 8.306934673366833e-05,
-      "loss": 5.1366,
-      "step": 17351
-    },
-    {
-      "epoch": 9.049282920469361,
-      "grad_norm": 1.3074005842208862,
-      "learning_rate": 8.306834170854271e-05,
-      "loss": 5.9986,
-      "step": 17352
-    },
-    {
-      "epoch": 9.04980443285528,
-      "grad_norm": 1.4643789529800415,
-      "learning_rate": 8.306733668341709e-05,
-      "loss": 5.5953,
-      "step": 17353
-    },
-    {
-      "epoch": 9.0503259452412,
-      "grad_norm": 1.423485517501831,
-      "learning_rate": 8.306633165829147e-05,
-      "loss": 5.7128,
-      "step": 17354
-    },
-    {
-      "epoch": 9.05084745762712,
-      "grad_norm": 1.4116674661636353,
-      "learning_rate": 8.306532663316583e-05,
-      "loss": 5.7514,
-      "step": 17355
-    },
-    {
-      "epoch": 9.051368970013037,
-      "grad_norm": 1.5203075408935547,
-      "learning_rate": 8.306432160804021e-05,
-      "loss": 4.9239,
-      "step": 17356
-    },
-    {
-      "epoch": 9.051890482398957,
-      "grad_norm": 1.6594421863555908,
-      "learning_rate": 8.306331658291457e-05,
-      "loss": 5.5951,
-      "step": 17357
-    },
-    {
-      "epoch": 9.052411994784876,
-      "grad_norm": 1.4546432495117188,
-      "learning_rate": 8.306231155778895e-05,
-      "loss": 5.6273,
-      "step": 17358
-    },
-    {
-      "epoch": 9.052933507170795,
-      "grad_norm": 1.5356796979904175,
-      "learning_rate": 8.306130653266332e-05,
-      "loss": 5.7414,
-      "step": 17359
-    },
-    {
-      "epoch": 9.053455019556715,
-      "grad_norm": 1.3447059392929077,
-      "learning_rate": 8.30603015075377e-05,
-      "loss": 5.8532,
-      "step": 17360
-    },
-    {
-      "epoch": 9.053976531942634,
-      "grad_norm": 1.4360140562057495,
-      "learning_rate": 8.305929648241206e-05,
-      "loss": 5.8874,
-      "step": 17361
-    },
-    {
-      "epoch": 9.054498044328552,
-      "grad_norm": 1.5309480428695679,
-      "learning_rate": 8.305829145728644e-05,
-      "loss": 5.3664,
-      "step": 17362
-    },
-    {
-      "epoch": 9.055019556714472,
-      "grad_norm": 1.4308847188949585,
-      "learning_rate": 8.305728643216081e-05,
-      "loss": 5.4792,
-      "step": 17363
-    },
-    {
-      "epoch": 9.055541069100391,
-      "grad_norm": 1.4932876825332642,
-      "learning_rate": 8.305628140703518e-05,
-      "loss": 5.3515,
-      "step": 17364
-    },
-    {
-      "epoch": 9.05606258148631,
-      "grad_norm": 1.457777500152588,
-      "learning_rate": 8.305527638190956e-05,
-      "loss": 5.5464,
-      "step": 17365
-    },
-    {
-      "epoch": 9.05658409387223,
-      "grad_norm": 1.412912368774414,
-      "learning_rate": 8.305427135678392e-05,
-      "loss": 5.4111,
-      "step": 17366
-    },
-    {
-      "epoch": 9.05710560625815,
-      "grad_norm": 1.4156126976013184,
-      "learning_rate": 8.30532663316583e-05,
-      "loss": 5.7497,
-      "step": 17367
-    },
-    {
-      "epoch": 9.057627118644067,
-      "grad_norm": 1.5218912363052368,
-      "learning_rate": 8.305226130653266e-05,
-      "loss": 5.3347,
-      "step": 17368
-    },
-    {
-      "epoch": 9.058148631029987,
-      "grad_norm": 1.3184617757797241,
-      "learning_rate": 8.305125628140704e-05,
-      "loss": 5.5071,
-      "step": 17369
-    },
-    {
-      "epoch": 9.058670143415906,
-      "grad_norm": 1.438066005706787,
-      "learning_rate": 8.30502512562814e-05,
-      "loss": 5.4316,
-      "step": 17370
-    },
-    {
-      "epoch": 9.059191655801826,
-      "grad_norm": 1.3778554201126099,
-      "learning_rate": 8.304924623115578e-05,
-      "loss": 5.7027,
-      "step": 17371
-    },
-    {
-      "epoch": 9.059713168187745,
-      "grad_norm": 1.3240668773651123,
-      "learning_rate": 8.304824120603015e-05,
-      "loss": 5.2936,
-      "step": 17372
-    },
-    {
-      "epoch": 9.060234680573664,
-      "grad_norm": 1.397803783416748,
-      "learning_rate": 8.304723618090452e-05,
-      "loss": 5.7948,
-      "step": 17373
-    },
-    {
-      "epoch": 9.060756192959582,
-      "grad_norm": 1.392176628112793,
-      "learning_rate": 8.30462311557789e-05,
-      "loss": 5.3707,
-      "step": 17374
-    },
-    {
-      "epoch": 9.061277705345502,
-      "grad_norm": 1.3226953744888306,
-      "learning_rate": 8.304522613065328e-05,
-      "loss": 5.8663,
-      "step": 17375
-    },
-    {
-      "epoch": 9.061799217731421,
-      "grad_norm": 1.4812403917312622,
-      "learning_rate": 8.304422110552764e-05,
-      "loss": 5.3976,
-      "step": 17376
-    },
-    {
-      "epoch": 9.06232073011734,
-      "grad_norm": 1.3814200162887573,
-      "learning_rate": 8.304321608040201e-05,
-      "loss": 5.5654,
-      "step": 17377
-    },
-    {
-      "epoch": 9.06284224250326,
-      "grad_norm": 1.6280336380004883,
-      "learning_rate": 8.304221105527639e-05,
-      "loss": 5.4208,
-      "step": 17378
-    },
-    {
-      "epoch": 9.06336375488918,
-      "grad_norm": 1.4174515008926392,
-      "learning_rate": 8.304120603015075e-05,
-      "loss": 5.8252,
-      "step": 17379
-    },
-    {
-      "epoch": 9.063885267275097,
-      "grad_norm": 1.4717650413513184,
-      "learning_rate": 8.304020100502513e-05,
-      "loss": 5.554,
-      "step": 17380
-    },
-    {
-      "epoch": 9.064406779661017,
-      "grad_norm": 1.3201528787612915,
-      "learning_rate": 8.30391959798995e-05,
-      "loss": 5.8354,
-      "step": 17381
-    },
-    {
-      "epoch": 9.064928292046936,
-      "grad_norm": 1.5411276817321777,
-      "learning_rate": 8.303819095477387e-05,
-      "loss": 5.8163,
-      "step": 17382
-    },
-    {
-      "epoch": 9.065449804432856,
-      "grad_norm": 1.4123568534851074,
-      "learning_rate": 8.303718592964825e-05,
-      "loss": 5.7554,
-      "step": 17383
-    },
-    {
-      "epoch": 9.065971316818775,
-      "grad_norm": 1.412353754043579,
-      "learning_rate": 8.303618090452263e-05,
-      "loss": 5.7539,
-      "step": 17384
-    },
-    {
-      "epoch": 9.066492829204694,
-      "grad_norm": 1.4115382432937622,
-      "learning_rate": 8.303517587939699e-05,
-      "loss": 5.3396,
-      "step": 17385
-    },
-    {
-      "epoch": 9.067014341590612,
-      "grad_norm": 1.37235426902771,
-      "learning_rate": 8.303417085427137e-05,
-      "loss": 5.5629,
-      "step": 17386
-    },
-    {
-      "epoch": 9.067535853976532,
-      "grad_norm": 1.6963540315628052,
-      "learning_rate": 8.303316582914573e-05,
-      "loss": 5.3108,
-      "step": 17387
-    },
-    {
-      "epoch": 9.068057366362451,
-      "grad_norm": 1.3113943338394165,
-      "learning_rate": 8.303216080402011e-05,
-      "loss": 5.3469,
-      "step": 17388
-    },
-    {
-      "epoch": 9.06857887874837,
-      "grad_norm": 1.4021053314208984,
-      "learning_rate": 8.303115577889448e-05,
-      "loss": 5.7066,
-      "step": 17389
-    },
-    {
-      "epoch": 9.06910039113429,
-      "grad_norm": 2.1672706604003906,
-      "learning_rate": 8.303015075376884e-05,
-      "loss": 5.0522,
-      "step": 17390
-    },
-    {
-      "epoch": 9.06962190352021,
-      "grad_norm": 1.5698199272155762,
-      "learning_rate": 8.302914572864322e-05,
-      "loss": 5.3889,
-      "step": 17391
-    },
-    {
-      "epoch": 9.070143415906127,
-      "grad_norm": 1.4200937747955322,
-      "learning_rate": 8.302814070351758e-05,
-      "loss": 5.7938,
-      "step": 17392
-    },
-    {
-      "epoch": 9.070664928292047,
-      "grad_norm": 1.5251377820968628,
-      "learning_rate": 8.302713567839196e-05,
-      "loss": 5.3933,
-      "step": 17393
-    },
-    {
-      "epoch": 9.071186440677966,
-      "grad_norm": 1.4254982471466064,
-      "learning_rate": 8.302613065326634e-05,
-      "loss": 5.2304,
-      "step": 17394
-    },
-    {
-      "epoch": 9.071707953063886,
-      "grad_norm": 1.4910202026367188,
-      "learning_rate": 8.302512562814072e-05,
-      "loss": 5.8362,
-      "step": 17395
-    },
-    {
-      "epoch": 9.072229465449805,
-      "grad_norm": 1.446841835975647,
-      "learning_rate": 8.302412060301508e-05,
-      "loss": 5.736,
-      "step": 17396
-    },
-    {
-      "epoch": 9.072750977835724,
-      "grad_norm": 1.2596241235733032,
-      "learning_rate": 8.302311557788946e-05,
-      "loss": 4.8413,
-      "step": 17397
-    },
-    {
-      "epoch": 9.073272490221642,
-      "grad_norm": 1.5027480125427246,
-      "learning_rate": 8.302211055276382e-05,
-      "loss": 5.2697,
-      "step": 17398
-    },
-    {
-      "epoch": 9.073794002607562,
-      "grad_norm": 1.414673089981079,
-      "learning_rate": 8.30211055276382e-05,
-      "loss": 5.6959,
-      "step": 17399
-    },
-    {
-      "epoch": 9.074315514993481,
-      "grad_norm": 1.580088496208191,
-      "learning_rate": 8.302010050251256e-05,
-      "loss": 5.6353,
-      "step": 17400
-    },
-    {
-      "epoch": 9.0748370273794,
-      "grad_norm": 1.3119784593582153,
-      "learning_rate": 8.301909547738694e-05,
-      "loss": 5.9653,
-      "step": 17401
-    },
-    {
-      "epoch": 9.07535853976532,
-      "grad_norm": 1.441422939300537,
-      "learning_rate": 8.30180904522613e-05,
-      "loss": 5.4862,
-      "step": 17402
-    },
-    {
-      "epoch": 9.075880052151238,
-      "grad_norm": 1.4223519563674927,
-      "learning_rate": 8.301708542713568e-05,
-      "loss": 5.8194,
-      "step": 17403
-    },
-    {
-      "epoch": 9.076401564537157,
-      "grad_norm": 1.5598829984664917,
-      "learning_rate": 8.301608040201006e-05,
-      "loss": 5.5011,
-      "step": 17404
-    },
-    {
-      "epoch": 9.076923076923077,
-      "grad_norm": 1.478488564491272,
-      "learning_rate": 8.301507537688443e-05,
-      "loss": 5.3334,
-      "step": 17405
-    },
-    {
-      "epoch": 9.077444589308996,
-      "grad_norm": 1.3402127027511597,
-      "learning_rate": 8.30140703517588e-05,
-      "loss": 5.8466,
-      "step": 17406
-    },
-    {
-      "epoch": 9.077966101694916,
-      "grad_norm": 1.4737426042556763,
-      "learning_rate": 8.301306532663317e-05,
-      "loss": 5.646,
-      "step": 17407
-    },
-    {
-      "epoch": 9.078487614080835,
-      "grad_norm": 1.4537936449050903,
-      "learning_rate": 8.301206030150755e-05,
-      "loss": 5.9012,
-      "step": 17408
-    },
-    {
-      "epoch": 9.079009126466753,
-      "grad_norm": 1.4110138416290283,
-      "learning_rate": 8.301105527638191e-05,
-      "loss": 5.7321,
-      "step": 17409
-    },
-    {
-      "epoch": 9.079530638852672,
-      "grad_norm": 1.7334249019622803,
-      "learning_rate": 8.301005025125629e-05,
-      "loss": 5.2519,
-      "step": 17410
-    },
-    {
-      "epoch": 9.080052151238592,
-      "grad_norm": 1.6736717224121094,
-      "learning_rate": 8.300904522613065e-05,
-      "loss": 4.9881,
-      "step": 17411
-    },
-    {
-      "epoch": 9.080573663624511,
-      "grad_norm": 1.726289987564087,
-      "learning_rate": 8.300804020100503e-05,
-      "loss": 5.2542,
-      "step": 17412
-    },
-    {
-      "epoch": 9.08109517601043,
-      "grad_norm": 1.5112303495407104,
-      "learning_rate": 8.30070351758794e-05,
-      "loss": 5.8662,
-      "step": 17413
-    },
-    {
-      "epoch": 9.08161668839635,
-      "grad_norm": 1.8039460182189941,
-      "learning_rate": 8.300603015075377e-05,
-      "loss": 5.3206,
-      "step": 17414
-    },
-    {
-      "epoch": 9.082138200782268,
-      "grad_norm": 1.5248981714248657,
-      "learning_rate": 8.300502512562815e-05,
-      "loss": 5.3502,
-      "step": 17415
-    },
-    {
-      "epoch": 9.082659713168187,
-      "grad_norm": 1.3873485326766968,
-      "learning_rate": 8.300402010050253e-05,
-      "loss": 5.241,
-      "step": 17416
-    },
-    {
-      "epoch": 9.083181225554107,
-      "grad_norm": 1.4890539646148682,
-      "learning_rate": 8.300301507537689e-05,
-      "loss": 5.126,
-      "step": 17417
-    },
-    {
-      "epoch": 9.083702737940026,
-      "grad_norm": 1.336972951889038,
-      "learning_rate": 8.300201005025126e-05,
-      "loss": 5.9818,
-      "step": 17418
-    },
-    {
-      "epoch": 9.084224250325946,
-      "grad_norm": 1.4566400051116943,
-      "learning_rate": 8.300100502512563e-05,
-      "loss": 5.3876,
-      "step": 17419
-    },
-    {
-      "epoch": 9.084745762711865,
-      "grad_norm": 1.376192569732666,
-      "learning_rate": 8.3e-05,
-      "loss": 5.4696,
-      "step": 17420
-    },
-    {
-      "epoch": 9.085267275097783,
-      "grad_norm": 1.5269981622695923,
-      "learning_rate": 8.299899497487438e-05,
-      "loss": 5.351,
-      "step": 17421
-    },
-    {
-      "epoch": 9.085788787483702,
-      "grad_norm": 1.4576040506362915,
-      "learning_rate": 8.299798994974874e-05,
-      "loss": 5.1772,
-      "step": 17422
-    },
-    {
-      "epoch": 9.086310299869622,
-      "grad_norm": 1.4987351894378662,
-      "learning_rate": 8.299698492462312e-05,
-      "loss": 5.3615,
-      "step": 17423
-    },
-    {
-      "epoch": 9.086831812255541,
-      "grad_norm": 1.460443377494812,
-      "learning_rate": 8.299597989949748e-05,
-      "loss": 5.9458,
-      "step": 17424
-    },
-    {
-      "epoch": 9.08735332464146,
-      "grad_norm": 1.3801147937774658,
-      "learning_rate": 8.299497487437186e-05,
-      "loss": 5.702,
-      "step": 17425
-    },
-    {
-      "epoch": 9.08787483702738,
-      "grad_norm": 1.62882399559021,
-      "learning_rate": 8.299396984924624e-05,
-      "loss": 5.7623,
-      "step": 17426
-    },
-    {
-      "epoch": 9.088396349413298,
-      "grad_norm": 1.3255188465118408,
-      "learning_rate": 8.299296482412062e-05,
-      "loss": 5.7723,
-      "step": 17427
-    },
-    {
-      "epoch": 9.088917861799217,
-      "grad_norm": 1.4351930618286133,
-      "learning_rate": 8.299195979899498e-05,
-      "loss": 5.6463,
-      "step": 17428
-    },
-    {
-      "epoch": 9.089439374185137,
-      "grad_norm": 1.2872576713562012,
-      "learning_rate": 8.299095477386936e-05,
-      "loss": 5.8058,
-      "step": 17429
-    },
-    {
-      "epoch": 9.089960886571056,
-      "grad_norm": 1.4364506006240845,
-      "learning_rate": 8.298994974874372e-05,
-      "loss": 5.4134,
-      "step": 17430
-    },
-    {
-      "epoch": 9.090482398956976,
-      "grad_norm": 1.5186632871627808,
-      "learning_rate": 8.298894472361809e-05,
-      "loss": 5.4517,
-      "step": 17431
-    },
-    {
-      "epoch": 9.091003911342895,
-      "grad_norm": 1.4624884128570557,
-      "learning_rate": 8.298793969849246e-05,
-      "loss": 5.4913,
-      "step": 17432
-    },
-    {
-      "epoch": 9.091525423728813,
-      "grad_norm": 1.5325835943222046,
-      "learning_rate": 8.298693467336683e-05,
-      "loss": 5.2369,
-      "step": 17433
-    },
-    {
-      "epoch": 9.092046936114732,
-      "grad_norm": 1.5087807178497314,
-      "learning_rate": 8.29859296482412e-05,
-      "loss": 5.385,
-      "step": 17434
-    },
-    {
-      "epoch": 9.092568448500652,
-      "grad_norm": 1.4986335039138794,
-      "learning_rate": 8.298492462311558e-05,
-      "loss": 5.421,
-      "step": 17435
-    },
-    {
-      "epoch": 9.093089960886571,
-      "grad_norm": 1.4067739248275757,
-      "learning_rate": 8.298391959798996e-05,
-      "loss": 5.5257,
-      "step": 17436
-    },
-    {
-      "epoch": 9.09361147327249,
-      "grad_norm": 1.4479731321334839,
-      "learning_rate": 8.298291457286433e-05,
-      "loss": 5.2703,
-      "step": 17437
-    },
-    {
-      "epoch": 9.09413298565841,
-      "grad_norm": 1.4827561378479004,
-      "learning_rate": 8.29819095477387e-05,
-      "loss": 5.3402,
-      "step": 17438
-    },
-    {
-      "epoch": 9.094654498044328,
-      "grad_norm": 1.557031273841858,
-      "learning_rate": 8.298090452261307e-05,
-      "loss": 4.7915,
-      "step": 17439
-    },
-    {
-      "epoch": 9.095176010430247,
-      "grad_norm": 1.4283612966537476,
-      "learning_rate": 8.297989949748745e-05,
-      "loss": 5.7033,
-      "step": 17440
-    },
-    {
-      "epoch": 9.095697522816167,
-      "grad_norm": 1.421241283416748,
-      "learning_rate": 8.297889447236181e-05,
-      "loss": 5.7455,
-      "step": 17441
-    },
-    {
-      "epoch": 9.096219035202086,
-      "grad_norm": 1.439592957496643,
-      "learning_rate": 8.297788944723619e-05,
-      "loss": 5.7228,
-      "step": 17442
-    },
-    {
-      "epoch": 9.096740547588006,
-      "grad_norm": 1.3568246364593506,
-      "learning_rate": 8.297688442211055e-05,
-      "loss": 5.612,
-      "step": 17443
-    },
-    {
-      "epoch": 9.097262059973925,
-      "grad_norm": 1.4010051488876343,
-      "learning_rate": 8.297587939698492e-05,
-      "loss": 5.2714,
-      "step": 17444
-    },
-    {
-      "epoch": 9.097783572359843,
-      "grad_norm": 1.5350604057312012,
-      "learning_rate": 8.29748743718593e-05,
-      "loss": 4.9473,
-      "step": 17445
-    },
-    {
-      "epoch": 9.098305084745762,
-      "grad_norm": 1.5072736740112305,
-      "learning_rate": 8.297386934673367e-05,
-      "loss": 5.3916,
-      "step": 17446
-    },
-    {
-      "epoch": 9.098826597131682,
-      "grad_norm": 1.3866479396820068,
-      "learning_rate": 8.297286432160805e-05,
-      "loss": 5.5689,
-      "step": 17447
-    },
-    {
-      "epoch": 9.099348109517601,
-      "grad_norm": 1.3451699018478394,
-      "learning_rate": 8.297185929648241e-05,
-      "loss": 4.9109,
-      "step": 17448
-    },
-    {
-      "epoch": 9.09986962190352,
-      "grad_norm": 1.4360448122024536,
-      "learning_rate": 8.297085427135679e-05,
-      "loss": 5.2072,
-      "step": 17449
-    },
-    {
-      "epoch": 9.10039113428944,
-      "grad_norm": 1.5435600280761719,
-      "learning_rate": 8.296984924623116e-05,
-      "loss": 5.2725,
-      "step": 17450
-    },
-    {
-      "epoch": 9.100912646675358,
-      "grad_norm": 1.5193167924880981,
-      "learning_rate": 8.296884422110553e-05,
-      "loss": 5.9036,
-      "step": 17451
-    },
-    {
-      "epoch": 9.101434159061277,
-      "grad_norm": 1.440637230873108,
-      "learning_rate": 8.29678391959799e-05,
-      "loss": 5.225,
-      "step": 17452
-    },
-    {
-      "epoch": 9.101955671447197,
-      "grad_norm": 1.3297873735427856,
-      "learning_rate": 8.296683417085428e-05,
-      "loss": 5.7108,
-      "step": 17453
-    },
-    {
-      "epoch": 9.102477183833116,
-      "grad_norm": 1.4542567729949951,
-      "learning_rate": 8.296582914572864e-05,
-      "loss": 5.5267,
-      "step": 17454
-    },
-    {
-      "epoch": 9.102998696219036,
-      "grad_norm": 1.4624172449111938,
-      "learning_rate": 8.296482412060302e-05,
-      "loss": 5.6338,
-      "step": 17455
-    },
-    {
-      "epoch": 9.103520208604955,
-      "grad_norm": 1.3213274478912354,
-      "learning_rate": 8.29638190954774e-05,
-      "loss": 6.0327,
-      "step": 17456
-    },
-    {
-      "epoch": 9.104041720990873,
-      "grad_norm": 1.4577445983886719,
-      "learning_rate": 8.296281407035176e-05,
-      "loss": 5.2787,
-      "step": 17457
-    },
-    {
-      "epoch": 9.104563233376792,
-      "grad_norm": 1.3276745080947876,
-      "learning_rate": 8.296180904522614e-05,
-      "loss": 5.4953,
-      "step": 17458
-    },
-    {
-      "epoch": 9.105084745762712,
-      "grad_norm": 1.381100058555603,
-      "learning_rate": 8.29608040201005e-05,
-      "loss": 5.7648,
-      "step": 17459
-    },
-    {
-      "epoch": 9.105606258148631,
-      "grad_norm": 1.4709526300430298,
-      "learning_rate": 8.295979899497488e-05,
-      "loss": 5.0387,
-      "step": 17460
-    },
-    {
-      "epoch": 9.10612777053455,
-      "grad_norm": 1.349472999572754,
-      "learning_rate": 8.295879396984925e-05,
-      "loss": 5.4266,
-      "step": 17461
-    },
-    {
-      "epoch": 9.10664928292047,
-      "grad_norm": 1.4498356580734253,
-      "learning_rate": 8.295778894472362e-05,
-      "loss": 5.3874,
-      "step": 17462
-    },
-    {
-      "epoch": 9.107170795306388,
-      "grad_norm": 1.4817912578582764,
-      "learning_rate": 8.295678391959799e-05,
-      "loss": 5.569,
-      "step": 17463
-    },
-    {
-      "epoch": 9.107692307692307,
-      "grad_norm": 1.416117787361145,
-      "learning_rate": 8.295577889447237e-05,
-      "loss": 5.3167,
-      "step": 17464
-    },
-    {
-      "epoch": 9.108213820078227,
-      "grad_norm": 1.3756608963012695,
-      "learning_rate": 8.295477386934673e-05,
-      "loss": 5.6054,
-      "step": 17465
-    },
-    {
-      "epoch": 9.108735332464146,
-      "grad_norm": 1.3864023685455322,
-      "learning_rate": 8.295376884422111e-05,
-      "loss": 5.2917,
-      "step": 17466
-    },
-    {
-      "epoch": 9.109256844850066,
-      "grad_norm": 1.3921328783035278,
-      "learning_rate": 8.295276381909549e-05,
-      "loss": 5.4259,
-      "step": 17467
-    },
-    {
-      "epoch": 9.109778357235985,
-      "grad_norm": 1.4250743389129639,
-      "learning_rate": 8.295175879396986e-05,
-      "loss": 5.6739,
-      "step": 17468
-    },
-    {
-      "epoch": 9.110299869621903,
-      "grad_norm": 1.424360990524292,
-      "learning_rate": 8.295075376884423e-05,
-      "loss": 5.418,
-      "step": 17469
-    },
-    {
-      "epoch": 9.110821382007822,
-      "grad_norm": 1.4408375024795532,
-      "learning_rate": 8.294974874371859e-05,
-      "loss": 5.5368,
-      "step": 17470
-    },
-    {
-      "epoch": 9.111342894393742,
-      "grad_norm": 1.364951491355896,
-      "learning_rate": 8.294874371859297e-05,
-      "loss": 5.2961,
-      "step": 17471
-    },
-    {
-      "epoch": 9.111864406779661,
-      "grad_norm": 1.513312816619873,
-      "learning_rate": 8.294773869346733e-05,
-      "loss": 5.7127,
-      "step": 17472
-    },
-    {
-      "epoch": 9.11238591916558,
-      "grad_norm": 1.5675184726715088,
-      "learning_rate": 8.294673366834171e-05,
-      "loss": 5.3082,
-      "step": 17473
-    },
-    {
-      "epoch": 9.1129074315515,
-      "grad_norm": 1.508382797241211,
-      "learning_rate": 8.294572864321608e-05,
-      "loss": 5.6338,
-      "step": 17474
-    },
-    {
-      "epoch": 9.113428943937418,
-      "grad_norm": 1.4550862312316895,
-      "learning_rate": 8.294472361809045e-05,
-      "loss": 5.321,
-      "step": 17475
-    },
-    {
-      "epoch": 9.113950456323337,
-      "grad_norm": 1.6515109539031982,
-      "learning_rate": 8.294371859296483e-05,
-      "loss": 4.763,
-      "step": 17476
-    },
-    {
-      "epoch": 9.114471968709257,
-      "grad_norm": 1.5508466958999634,
-      "learning_rate": 8.294271356783921e-05,
-      "loss": 5.1989,
-      "step": 17477
-    },
-    {
-      "epoch": 9.114993481095176,
-      "grad_norm": 1.4244736433029175,
-      "learning_rate": 8.294170854271357e-05,
-      "loss": 5.5589,
-      "step": 17478
-    },
-    {
-      "epoch": 9.115514993481096,
-      "grad_norm": 1.4555209875106812,
-      "learning_rate": 8.294070351758795e-05,
-      "loss": 5.7463,
-      "step": 17479
-    },
-    {
-      "epoch": 9.116036505867015,
-      "grad_norm": 1.474456548690796,
-      "learning_rate": 8.293969849246232e-05,
-      "loss": 5.581,
-      "step": 17480
-    },
-    {
-      "epoch": 9.116558018252933,
-      "grad_norm": 1.2895766496658325,
-      "learning_rate": 8.29386934673367e-05,
-      "loss": 6.0398,
-      "step": 17481
-    },
-    {
-      "epoch": 9.117079530638852,
-      "grad_norm": 1.5545542240142822,
-      "learning_rate": 8.293768844221106e-05,
-      "loss": 5.29,
-      "step": 17482
-    },
-    {
-      "epoch": 9.117601043024772,
-      "grad_norm": 1.3431050777435303,
-      "learning_rate": 8.293668341708542e-05,
-      "loss": 5.2334,
-      "step": 17483
-    },
-    {
-      "epoch": 9.118122555410691,
-      "grad_norm": 1.355223298072815,
-      "learning_rate": 8.29356783919598e-05,
-      "loss": 5.2276,
-      "step": 17484
-    },
-    {
-      "epoch": 9.11864406779661,
-      "grad_norm": 1.3577569723129272,
-      "learning_rate": 8.293467336683416e-05,
-      "loss": 5.655,
-      "step": 17485
-    },
-    {
-      "epoch": 9.11916558018253,
-      "grad_norm": 1.4809763431549072,
-      "learning_rate": 8.293366834170854e-05,
-      "loss": 5.503,
-      "step": 17486
-    },
-    {
-      "epoch": 9.119687092568448,
-      "grad_norm": 1.509393334388733,
-      "learning_rate": 8.293266331658292e-05,
-      "loss": 5.3449,
-      "step": 17487
-    },
-    {
-      "epoch": 9.120208604954367,
-      "grad_norm": 1.4681872129440308,
-      "learning_rate": 8.29316582914573e-05,
-      "loss": 5.5672,
-      "step": 17488
-    },
-    {
-      "epoch": 9.120730117340287,
-      "grad_norm": 1.5185085535049438,
-      "learning_rate": 8.293065326633166e-05,
-      "loss": 5.6478,
-      "step": 17489
-    },
-    {
-      "epoch": 9.121251629726206,
-      "grad_norm": 1.5685802698135376,
-      "learning_rate": 8.292964824120604e-05,
-      "loss": 4.8326,
-      "step": 17490
-    },
-    {
-      "epoch": 9.121773142112126,
-      "grad_norm": 1.4697155952453613,
-      "learning_rate": 8.29286432160804e-05,
-      "loss": 5.7857,
-      "step": 17491
-    },
-    {
-      "epoch": 9.122294654498045,
-      "grad_norm": 1.4286540746688843,
-      "learning_rate": 8.292763819095478e-05,
-      "loss": 5.3961,
-      "step": 17492
-    },
-    {
-      "epoch": 9.122816166883963,
-      "grad_norm": 1.531938910484314,
-      "learning_rate": 8.292663316582915e-05,
-      "loss": 5.2973,
-      "step": 17493
-    },
-    {
-      "epoch": 9.123337679269882,
-      "grad_norm": 1.317456841468811,
-      "learning_rate": 8.292562814070352e-05,
-      "loss": 5.5813,
-      "step": 17494
-    },
-    {
-      "epoch": 9.123859191655802,
-      "grad_norm": 1.4007821083068848,
-      "learning_rate": 8.292462311557789e-05,
-      "loss": 5.5097,
-      "step": 17495
-    },
-    {
-      "epoch": 9.124380704041721,
-      "grad_norm": 1.4721264839172363,
-      "learning_rate": 8.292361809045227e-05,
-      "loss": 4.7658,
-      "step": 17496
-    },
-    {
-      "epoch": 9.12490221642764,
-      "grad_norm": 1.4705638885498047,
-      "learning_rate": 8.292261306532664e-05,
-      "loss": 5.6728,
-      "step": 17497
-    },
-    {
-      "epoch": 9.125423728813558,
-      "grad_norm": 1.3995777368545532,
-      "learning_rate": 8.292160804020101e-05,
-      "loss": 5.6023,
-      "step": 17498
-    },
-    {
-      "epoch": 9.125945241199478,
-      "grad_norm": 1.3988995552062988,
-      "learning_rate": 8.292060301507539e-05,
-      "loss": 5.5977,
-      "step": 17499
-    },
-    {
-      "epoch": 9.126466753585397,
-      "grad_norm": 1.3367202281951904,
-      "learning_rate": 8.291959798994975e-05,
-      "loss": 5.9081,
-      "step": 17500
-    },
-    {
-      "epoch": 9.126988265971317,
-      "grad_norm": 1.4216432571411133,
-      "learning_rate": 8.291859296482413e-05,
-      "loss": 5.3788,
-      "step": 17501
-    },
-    {
-      "epoch": 9.127509778357236,
-      "grad_norm": 1.3712421655654907,
-      "learning_rate": 8.291758793969849e-05,
-      "loss": 5.4006,
-      "step": 17502
-    },
-    {
-      "epoch": 9.128031290743156,
-      "grad_norm": 1.5073949098587036,
-      "learning_rate": 8.291658291457287e-05,
-      "loss": 5.3384,
-      "step": 17503
-    },
-    {
-      "epoch": 9.128552803129073,
-      "grad_norm": 1.4523956775665283,
-      "learning_rate": 8.291557788944723e-05,
-      "loss": 5.2102,
-      "step": 17504
-    },
-    {
-      "epoch": 9.129074315514993,
-      "grad_norm": 1.5132452249526978,
-      "learning_rate": 8.291457286432161e-05,
-      "loss": 5.1829,
-      "step": 17505
-    },
-    {
-      "epoch": 9.129595827900912,
-      "grad_norm": 1.4598431587219238,
-      "learning_rate": 8.291356783919598e-05,
-      "loss": 5.5705,
-      "step": 17506
-    },
-    {
-      "epoch": 9.130117340286832,
-      "grad_norm": 1.4349600076675415,
-      "learning_rate": 8.291256281407035e-05,
-      "loss": 5.5997,
-      "step": 17507
-    },
-    {
-      "epoch": 9.130638852672751,
-      "grad_norm": 1.4027385711669922,
-      "learning_rate": 8.291155778894473e-05,
-      "loss": 5.9634,
-      "step": 17508
-    },
-    {
-      "epoch": 9.13116036505867,
-      "grad_norm": 1.5839043855667114,
-      "learning_rate": 8.291055276381911e-05,
-      "loss": 5.7256,
-      "step": 17509
-    },
-    {
-      "epoch": 9.131681877444588,
-      "grad_norm": 1.6967203617095947,
-      "learning_rate": 8.290954773869347e-05,
-      "loss": 5.18,
-      "step": 17510
-    },
-    {
-      "epoch": 9.132203389830508,
-      "grad_norm": 1.4456632137298584,
-      "learning_rate": 8.290854271356784e-05,
-      "loss": 5.4095,
-      "step": 17511
-    },
-    {
-      "epoch": 9.132724902216427,
-      "grad_norm": 1.3805888891220093,
-      "learning_rate": 8.290753768844222e-05,
-      "loss": 5.6674,
-      "step": 17512
-    },
-    {
-      "epoch": 9.133246414602347,
-      "grad_norm": 1.3847852945327759,
-      "learning_rate": 8.290653266331658e-05,
-      "loss": 5.7867,
-      "step": 17513
-    },
-    {
-      "epoch": 9.133767926988266,
-      "grad_norm": 1.4477274417877197,
-      "learning_rate": 8.290552763819096e-05,
-      "loss": 5.0835,
-      "step": 17514
-    },
-    {
-      "epoch": 9.134289439374186,
-      "grad_norm": 1.3245892524719238,
-      "learning_rate": 8.290452261306532e-05,
-      "loss": 5.2861,
-      "step": 17515
-    },
-    {
-      "epoch": 9.134810951760103,
-      "grad_norm": 1.3178510665893555,
-      "learning_rate": 8.29035175879397e-05,
-      "loss": 5.1574,
-      "step": 17516
-    },
-    {
-      "epoch": 9.135332464146023,
-      "grad_norm": 1.418753743171692,
-      "learning_rate": 8.290251256281408e-05,
-      "loss": 5.545,
-      "step": 17517
-    },
-    {
-      "epoch": 9.135853976531942,
-      "grad_norm": 1.3742074966430664,
-      "learning_rate": 8.290150753768846e-05,
-      "loss": 5.7524,
-      "step": 17518
-    },
-    {
-      "epoch": 9.136375488917862,
-      "grad_norm": 1.5047922134399414,
-      "learning_rate": 8.290050251256282e-05,
-      "loss": 5.403,
-      "step": 17519
-    },
-    {
-      "epoch": 9.136897001303781,
-      "grad_norm": 1.4077062606811523,
-      "learning_rate": 8.28994974874372e-05,
-      "loss": 5.6404,
-      "step": 17520
-    },
-    {
-      "epoch": 9.1374185136897,
-      "grad_norm": 1.5135817527770996,
-      "learning_rate": 8.289849246231156e-05,
-      "loss": 5.0699,
-      "step": 17521
-    },
-    {
-      "epoch": 9.137940026075619,
-      "grad_norm": 1.4100223779678345,
-      "learning_rate": 8.289748743718594e-05,
-      "loss": 5.4393,
-      "step": 17522
-    },
-    {
-      "epoch": 9.138461538461538,
-      "grad_norm": 1.3604862689971924,
-      "learning_rate": 8.28964824120603e-05,
-      "loss": 5.8098,
-      "step": 17523
-    },
-    {
-      "epoch": 9.138983050847457,
-      "grad_norm": 1.3384654521942139,
-      "learning_rate": 8.289547738693467e-05,
-      "loss": 4.9263,
-      "step": 17524
-    },
-    {
-      "epoch": 9.139504563233377,
-      "grad_norm": 1.4652363061904907,
-      "learning_rate": 8.289447236180905e-05,
-      "loss": 5.4594,
-      "step": 17525
-    },
-    {
-      "epoch": 9.140026075619296,
-      "grad_norm": 1.56032133102417,
-      "learning_rate": 8.289346733668341e-05,
-      "loss": 5.3084,
-      "step": 17526
-    },
-    {
-      "epoch": 9.140547588005216,
-      "grad_norm": 1.4480454921722412,
-      "learning_rate": 8.289246231155779e-05,
-      "loss": 5.663,
-      "step": 17527
-    },
-    {
-      "epoch": 9.141069100391134,
-      "grad_norm": 1.4233883619308472,
-      "learning_rate": 8.289145728643217e-05,
-      "loss": 5.6823,
-      "step": 17528
-    },
-    {
-      "epoch": 9.141590612777053,
-      "grad_norm": 1.4505610466003418,
-      "learning_rate": 8.289045226130654e-05,
-      "loss": 5.2032,
-      "step": 17529
-    },
-    {
-      "epoch": 9.142112125162972,
-      "grad_norm": 1.4688643217086792,
-      "learning_rate": 8.288944723618091e-05,
-      "loss": 5.4064,
-      "step": 17530
-    },
-    {
-      "epoch": 9.142633637548892,
-      "grad_norm": 1.4583970308303833,
-      "learning_rate": 8.288844221105529e-05,
-      "loss": 5.5059,
-      "step": 17531
-    },
-    {
-      "epoch": 9.143155149934811,
-      "grad_norm": 1.4218381643295288,
-      "learning_rate": 8.288743718592965e-05,
-      "loss": 5.2915,
-      "step": 17532
-    },
-    {
-      "epoch": 9.14367666232073,
-      "grad_norm": 1.5849950313568115,
-      "learning_rate": 8.288643216080403e-05,
-      "loss": 5.0037,
-      "step": 17533
-    },
-    {
-      "epoch": 9.144198174706649,
-      "grad_norm": 1.4498271942138672,
-      "learning_rate": 8.288542713567839e-05,
-      "loss": 5.5432,
-      "step": 17534
-    },
-    {
-      "epoch": 9.144719687092568,
-      "grad_norm": 1.4505702257156372,
-      "learning_rate": 8.288442211055277e-05,
-      "loss": 5.26,
-      "step": 17535
-    },
-    {
-      "epoch": 9.145241199478487,
-      "grad_norm": 1.4658080339431763,
-      "learning_rate": 8.288341708542714e-05,
-      "loss": 5.5097,
-      "step": 17536
-    },
-    {
-      "epoch": 9.145762711864407,
-      "grad_norm": 1.3829625844955444,
-      "learning_rate": 8.288241206030151e-05,
-      "loss": 5.7773,
-      "step": 17537
-    },
-    {
-      "epoch": 9.146284224250326,
-      "grad_norm": 1.4326821565628052,
-      "learning_rate": 8.288140703517589e-05,
-      "loss": 5.5707,
-      "step": 17538
-    },
-    {
-      "epoch": 9.146805736636246,
-      "grad_norm": 1.4130661487579346,
-      "learning_rate": 8.288040201005025e-05,
-      "loss": 5.5991,
-      "step": 17539
-    },
-    {
-      "epoch": 9.147327249022164,
-      "grad_norm": 1.3829634189605713,
-      "learning_rate": 8.287939698492463e-05,
-      "loss": 5.5918,
-      "step": 17540
-    },
-    {
-      "epoch": 9.147848761408083,
-      "grad_norm": 1.3999228477478027,
-      "learning_rate": 8.2878391959799e-05,
-      "loss": 5.5576,
-      "step": 17541
-    },
-    {
-      "epoch": 9.148370273794002,
-      "grad_norm": 1.4937453269958496,
-      "learning_rate": 8.287738693467337e-05,
-      "loss": 5.6063,
-      "step": 17542
-    },
-    {
-      "epoch": 9.148891786179922,
-      "grad_norm": 1.5668882131576538,
-      "learning_rate": 8.287638190954774e-05,
-      "loss": 4.865,
-      "step": 17543
-    },
-    {
-      "epoch": 9.149413298565841,
-      "grad_norm": 1.6915048360824585,
-      "learning_rate": 8.287537688442212e-05,
-      "loss": 5.0527,
-      "step": 17544
-    },
-    {
-      "epoch": 9.14993481095176,
-      "grad_norm": 1.525158166885376,
-      "learning_rate": 8.287437185929648e-05,
-      "loss": 5.3446,
-      "step": 17545
-    },
-    {
-      "epoch": 9.150456323337679,
-      "grad_norm": 1.420320749282837,
-      "learning_rate": 8.287336683417086e-05,
-      "loss": 5.8751,
-      "step": 17546
-    },
-    {
-      "epoch": 9.150977835723598,
-      "grad_norm": 1.4821721315383911,
-      "learning_rate": 8.287236180904522e-05,
-      "loss": 5.5025,
-      "step": 17547
-    },
-    {
-      "epoch": 9.151499348109517,
-      "grad_norm": 1.4563132524490356,
-      "learning_rate": 8.28713567839196e-05,
-      "loss": 5.3199,
-      "step": 17548
-    },
-    {
-      "epoch": 9.152020860495437,
-      "grad_norm": 1.3764809370040894,
-      "learning_rate": 8.287035175879398e-05,
-      "loss": 5.6727,
-      "step": 17549
-    },
-    {
-      "epoch": 9.152542372881356,
-      "grad_norm": 1.3452060222625732,
-      "learning_rate": 8.286934673366834e-05,
-      "loss": 5.5378,
-      "step": 17550
-    },
-    {
-      "epoch": 9.153063885267276,
-      "grad_norm": 1.4641693830490112,
-      "learning_rate": 8.286834170854272e-05,
-      "loss": 5.5167,
-      "step": 17551
-    },
-    {
-      "epoch": 9.153585397653194,
-      "grad_norm": 1.349853515625,
-      "learning_rate": 8.286733668341709e-05,
-      "loss": 5.7151,
-      "step": 17552
-    },
-    {
-      "epoch": 9.154106910039113,
-      "grad_norm": 1.6118121147155762,
-      "learning_rate": 8.286633165829146e-05,
-      "loss": 5.4501,
-      "step": 17553
-    },
-    {
-      "epoch": 9.154628422425032,
-      "grad_norm": 1.591229796409607,
-      "learning_rate": 8.286532663316583e-05,
-      "loss": 5.3909,
-      "step": 17554
-    },
-    {
-      "epoch": 9.155149934810952,
-      "grad_norm": 1.4568344354629517,
-      "learning_rate": 8.28643216080402e-05,
-      "loss": 5.4359,
-      "step": 17555
-    },
-    {
-      "epoch": 9.155671447196871,
-      "grad_norm": 1.4254376888275146,
-      "learning_rate": 8.286331658291457e-05,
-      "loss": 5.698,
-      "step": 17556
-    },
-    {
-      "epoch": 9.156192959582791,
-      "grad_norm": 1.3674976825714111,
-      "learning_rate": 8.286231155778895e-05,
-      "loss": 5.515,
-      "step": 17557
-    },
-    {
-      "epoch": 9.156714471968709,
-      "grad_norm": 1.417150616645813,
-      "learning_rate": 8.286130653266333e-05,
-      "loss": 5.5988,
-      "step": 17558
-    },
-    {
-      "epoch": 9.157235984354628,
-      "grad_norm": 1.5065217018127441,
-      "learning_rate": 8.28603015075377e-05,
-      "loss": 4.7749,
-      "step": 17559
-    },
-    {
-      "epoch": 9.157757496740548,
-      "grad_norm": 1.4005374908447266,
-      "learning_rate": 8.285929648241207e-05,
-      "loss": 5.6332,
-      "step": 17560
-    },
-    {
-      "epoch": 9.158279009126467,
-      "grad_norm": 1.4230185747146606,
-      "learning_rate": 8.285829145728645e-05,
-      "loss": 5.9786,
-      "step": 17561
-    },
-    {
-      "epoch": 9.158800521512386,
-      "grad_norm": 1.3802645206451416,
-      "learning_rate": 8.285728643216081e-05,
-      "loss": 5.6485,
-      "step": 17562
-    },
-    {
-      "epoch": 9.159322033898306,
-      "grad_norm": 1.4748741388320923,
-      "learning_rate": 8.285628140703517e-05,
-      "loss": 5.4553,
-      "step": 17563
-    },
-    {
-      "epoch": 9.159843546284224,
-      "grad_norm": 1.4272775650024414,
-      "learning_rate": 8.285527638190955e-05,
-      "loss": 5.9084,
-      "step": 17564
-    },
-    {
-      "epoch": 9.160365058670143,
-      "grad_norm": 1.4472540616989136,
-      "learning_rate": 8.285427135678392e-05,
-      "loss": 5.4636,
-      "step": 17565
-    },
-    {
-      "epoch": 9.160886571056063,
-      "grad_norm": 1.4482941627502441,
-      "learning_rate": 8.28532663316583e-05,
-      "loss": 5.3368,
-      "step": 17566
-    },
-    {
-      "epoch": 9.161408083441982,
-      "grad_norm": 1.3999342918395996,
-      "learning_rate": 8.285226130653266e-05,
-      "loss": 5.6753,
-      "step": 17567
-    },
-    {
-      "epoch": 9.161929595827901,
-      "grad_norm": 1.4229156970977783,
-      "learning_rate": 8.285125628140704e-05,
-      "loss": 5.4792,
-      "step": 17568
-    },
-    {
-      "epoch": 9.162451108213821,
-      "grad_norm": 1.405462622642517,
-      "learning_rate": 8.285025125628141e-05,
-      "loss": 5.5454,
-      "step": 17569
-    },
-    {
-      "epoch": 9.162972620599739,
-      "grad_norm": 1.282599925994873,
-      "learning_rate": 8.284924623115579e-05,
-      "loss": 5.6419,
-      "step": 17570
-    },
-    {
-      "epoch": 9.163494132985658,
-      "grad_norm": 1.30425226688385,
-      "learning_rate": 8.284824120603016e-05,
-      "loss": 5.8811,
-      "step": 17571
-    },
-    {
-      "epoch": 9.164015645371578,
-      "grad_norm": 1.3920621871948242,
-      "learning_rate": 8.284723618090453e-05,
-      "loss": 5.601,
-      "step": 17572
-    },
-    {
-      "epoch": 9.164537157757497,
-      "grad_norm": 1.48684823513031,
-      "learning_rate": 8.28462311557789e-05,
-      "loss": 5.3812,
-      "step": 17573
-    },
-    {
-      "epoch": 9.165058670143416,
-      "grad_norm": 1.463987112045288,
-      "learning_rate": 8.284522613065328e-05,
-      "loss": 5.4253,
-      "step": 17574
-    },
-    {
-      "epoch": 9.165580182529336,
-      "grad_norm": 1.3474451303482056,
-      "learning_rate": 8.284422110552764e-05,
-      "loss": 5.3419,
-      "step": 17575
-    },
-    {
-      "epoch": 9.166101694915254,
-      "grad_norm": 1.5164052248001099,
-      "learning_rate": 8.2843216080402e-05,
-      "loss": 5.5013,
-      "step": 17576
-    },
-    {
-      "epoch": 9.166623207301173,
-      "grad_norm": 1.4014441967010498,
-      "learning_rate": 8.284221105527638e-05,
-      "loss": 5.656,
-      "step": 17577
-    },
-    {
-      "epoch": 9.167144719687093,
-      "grad_norm": 1.3937108516693115,
-      "learning_rate": 8.284120603015076e-05,
-      "loss": 5.4745,
-      "step": 17578
-    },
-    {
-      "epoch": 9.167666232073012,
-      "grad_norm": 1.3265430927276611,
-      "learning_rate": 8.284020100502514e-05,
-      "loss": 5.7653,
-      "step": 17579
-    },
-    {
-      "epoch": 9.168187744458931,
-      "grad_norm": 1.3760735988616943,
-      "learning_rate": 8.28391959798995e-05,
-      "loss": 5.7452,
-      "step": 17580
-    },
-    {
-      "epoch": 9.16870925684485,
-      "grad_norm": 1.3954495191574097,
-      "learning_rate": 8.283819095477388e-05,
-      "loss": 5.6599,
-      "step": 17581
-    },
-    {
-      "epoch": 9.169230769230769,
-      "grad_norm": 1.5597609281539917,
-      "learning_rate": 8.283718592964824e-05,
-      "loss": 5.8035,
-      "step": 17582
-    },
-    {
-      "epoch": 9.169752281616688,
-      "grad_norm": 1.377284049987793,
-      "learning_rate": 8.283618090452262e-05,
-      "loss": 5.7174,
-      "step": 17583
-    },
-    {
-      "epoch": 9.170273794002608,
-      "grad_norm": 1.4259051084518433,
-      "learning_rate": 8.283517587939699e-05,
-      "loss": 5.7643,
-      "step": 17584
-    },
-    {
-      "epoch": 9.170795306388527,
-      "grad_norm": 1.56025230884552,
-      "learning_rate": 8.283417085427136e-05,
-      "loss": 4.7893,
-      "step": 17585
-    },
-    {
-      "epoch": 9.171316818774446,
-      "grad_norm": 1.4201688766479492,
-      "learning_rate": 8.283316582914573e-05,
-      "loss": 5.2393,
-      "step": 17586
-    },
-    {
-      "epoch": 9.171838331160366,
-      "grad_norm": 1.4882559776306152,
-      "learning_rate": 8.28321608040201e-05,
-      "loss": 5.1553,
-      "step": 17587
-    },
-    {
-      "epoch": 9.172359843546284,
-      "grad_norm": 1.48074471950531,
-      "learning_rate": 8.283115577889447e-05,
-      "loss": 5.1222,
-      "step": 17588
-    },
-    {
-      "epoch": 9.172881355932203,
-      "grad_norm": 1.3785502910614014,
-      "learning_rate": 8.283015075376885e-05,
-      "loss": 5.5778,
-      "step": 17589
-    },
-    {
-      "epoch": 9.173402868318123,
-      "grad_norm": 1.4323077201843262,
-      "learning_rate": 8.282914572864323e-05,
-      "loss": 5.7063,
-      "step": 17590
-    },
-    {
-      "epoch": 9.173924380704042,
-      "grad_norm": 1.4603569507598877,
-      "learning_rate": 8.282814070351759e-05,
-      "loss": 5.5582,
-      "step": 17591
-    },
-    {
-      "epoch": 9.174445893089962,
-      "grad_norm": 1.5300102233886719,
-      "learning_rate": 8.282713567839197e-05,
-      "loss": 5.5465,
-      "step": 17592
-    },
-    {
-      "epoch": 9.17496740547588,
-      "grad_norm": 1.458500623703003,
-      "learning_rate": 8.282613065326633e-05,
-      "loss": 5.1887,
-      "step": 17593
-    },
-    {
-      "epoch": 9.175488917861799,
-      "grad_norm": 1.482334017753601,
-      "learning_rate": 8.282512562814071e-05,
-      "loss": 5.401,
-      "step": 17594
-    },
-    {
-      "epoch": 9.176010430247718,
-      "grad_norm": 1.440600037574768,
-      "learning_rate": 8.282412060301507e-05,
-      "loss": 5.6056,
-      "step": 17595
-    },
-    {
-      "epoch": 9.176531942633638,
-      "grad_norm": 1.4747061729431152,
-      "learning_rate": 8.282311557788945e-05,
-      "loss": 5.5679,
-      "step": 17596
-    },
-    {
-      "epoch": 9.177053455019557,
-      "grad_norm": 1.4951287508010864,
-      "learning_rate": 8.282211055276382e-05,
-      "loss": 5.5671,
-      "step": 17597
-    },
-    {
-      "epoch": 9.177574967405477,
-      "grad_norm": 1.364728569984436,
-      "learning_rate": 8.28211055276382e-05,
-      "loss": 5.63,
-      "step": 17598
-    },
-    {
-      "epoch": 9.178096479791394,
-      "grad_norm": 1.4617865085601807,
-      "learning_rate": 8.282010050251256e-05,
-      "loss": 5.2295,
-      "step": 17599
-    },
-    {
-      "epoch": 9.178617992177314,
-      "grad_norm": 1.9245483875274658,
-      "learning_rate": 8.281909547738694e-05,
-      "loss": 5.4074,
-      "step": 17600
-    },
-    {
-      "epoch": 9.179139504563233,
-      "grad_norm": 1.4554659128189087,
-      "learning_rate": 8.281809045226131e-05,
-      "loss": 5.7127,
-      "step": 17601
-    },
-    {
-      "epoch": 9.179661016949153,
-      "grad_norm": 1.355150818824768,
-      "learning_rate": 8.281708542713569e-05,
-      "loss": 5.7414,
-      "step": 17602
-    },
-    {
-      "epoch": 9.180182529335072,
-      "grad_norm": 1.4895449876785278,
-      "learning_rate": 8.281608040201006e-05,
-      "loss": 5.5627,
-      "step": 17603
-    },
-    {
-      "epoch": 9.180704041720992,
-      "grad_norm": 1.3890935182571411,
-      "learning_rate": 8.281507537688442e-05,
-      "loss": 5.4939,
-      "step": 17604
-    },
-    {
-      "epoch": 9.18122555410691,
-      "grad_norm": 1.5726248025894165,
-      "learning_rate": 8.28140703517588e-05,
-      "loss": 5.2488,
-      "step": 17605
-    },
-    {
-      "epoch": 9.181747066492829,
-      "grad_norm": 1.4720786809921265,
-      "learning_rate": 8.281306532663316e-05,
-      "loss": 5.9398,
-      "step": 17606
-    },
-    {
-      "epoch": 9.182268578878748,
-      "grad_norm": 1.4492038488388062,
-      "learning_rate": 8.281206030150754e-05,
-      "loss": 5.5169,
-      "step": 17607
-    },
-    {
-      "epoch": 9.182790091264668,
-      "grad_norm": 1.5100561380386353,
-      "learning_rate": 8.28110552763819e-05,
-      "loss": 5.426,
-      "step": 17608
-    },
-    {
-      "epoch": 9.183311603650587,
-      "grad_norm": 1.3984851837158203,
-      "learning_rate": 8.281005025125628e-05,
-      "loss": 5.514,
-      "step": 17609
-    },
-    {
-      "epoch": 9.183833116036507,
-      "grad_norm": 1.3936129808425903,
-      "learning_rate": 8.280904522613066e-05,
-      "loss": 5.6657,
-      "step": 17610
-    },
-    {
-      "epoch": 9.184354628422424,
-      "grad_norm": 1.3618017435073853,
-      "learning_rate": 8.280804020100504e-05,
-      "loss": 5.5397,
-      "step": 17611
-    },
-    {
-      "epoch": 9.184876140808344,
-      "grad_norm": 1.537179946899414,
-      "learning_rate": 8.28070351758794e-05,
-      "loss": 5.4335,
-      "step": 17612
-    },
-    {
-      "epoch": 9.185397653194263,
-      "grad_norm": 1.502042293548584,
-      "learning_rate": 8.280603015075378e-05,
-      "loss": 5.4586,
-      "step": 17613
-    },
-    {
-      "epoch": 9.185919165580183,
-      "grad_norm": 1.4345784187316895,
-      "learning_rate": 8.280502512562814e-05,
-      "loss": 5.5487,
-      "step": 17614
-    },
-    {
-      "epoch": 9.186440677966102,
-      "grad_norm": 1.3357983827590942,
-      "learning_rate": 8.280402010050252e-05,
-      "loss": 4.9328,
-      "step": 17615
-    },
-    {
-      "epoch": 9.186962190352022,
-      "grad_norm": 1.6318801641464233,
-      "learning_rate": 8.280301507537689e-05,
-      "loss": 5.2136,
-      "step": 17616
-    },
-    {
-      "epoch": 9.18748370273794,
-      "grad_norm": 1.5728073120117188,
-      "learning_rate": 8.280201005025125e-05,
-      "loss": 5.1954,
-      "step": 17617
-    },
-    {
-      "epoch": 9.188005215123859,
-      "grad_norm": 1.7080765962600708,
-      "learning_rate": 8.280100502512563e-05,
-      "loss": 5.4728,
-      "step": 17618
-    },
-    {
-      "epoch": 9.188526727509778,
-      "grad_norm": 1.4476102590560913,
-      "learning_rate": 8.28e-05,
-      "loss": 4.7064,
-      "step": 17619
-    },
-    {
-      "epoch": 9.189048239895698,
-      "grad_norm": 1.7387069463729858,
-      "learning_rate": 8.279899497487437e-05,
-      "loss": 5.3665,
-      "step": 17620
-    },
-    {
-      "epoch": 9.189569752281617,
-      "grad_norm": 1.3859039545059204,
-      "learning_rate": 8.279798994974875e-05,
-      "loss": 5.635,
-      "step": 17621
-    },
-    {
-      "epoch": 9.190091264667537,
-      "grad_norm": 1.3331228494644165,
-      "learning_rate": 8.279698492462313e-05,
-      "loss": 5.6012,
-      "step": 17622
-    },
-    {
-      "epoch": 9.190612777053454,
-      "grad_norm": 1.3810651302337646,
-      "learning_rate": 8.279597989949749e-05,
-      "loss": 5.4908,
-      "step": 17623
-    },
-    {
-      "epoch": 9.191134289439374,
-      "grad_norm": 1.423636555671692,
-      "learning_rate": 8.279497487437187e-05,
-      "loss": 5.4396,
-      "step": 17624
-    },
-    {
-      "epoch": 9.191655801825293,
-      "grad_norm": 1.4310568571090698,
-      "learning_rate": 8.279396984924623e-05,
-      "loss": 5.5062,
-      "step": 17625
-    },
-    {
-      "epoch": 9.192177314211213,
-      "grad_norm": 1.2940860986709595,
-      "learning_rate": 8.279296482412061e-05,
-      "loss": 5.7708,
-      "step": 17626
-    },
-    {
-      "epoch": 9.192698826597132,
-      "grad_norm": 1.371362328529358,
-      "learning_rate": 8.279195979899498e-05,
-      "loss": 5.4643,
-      "step": 17627
-    },
-    {
-      "epoch": 9.193220338983052,
-      "grad_norm": 1.3716537952423096,
-      "learning_rate": 8.279095477386935e-05,
-      "loss": 5.6986,
-      "step": 17628
-    },
-    {
-      "epoch": 9.19374185136897,
-      "grad_norm": 1.4497904777526855,
-      "learning_rate": 8.278994974874372e-05,
-      "loss": 5.6092,
-      "step": 17629
-    },
-    {
-      "epoch": 9.194263363754889,
-      "grad_norm": 1.3885223865509033,
-      "learning_rate": 8.27889447236181e-05,
-      "loss": 5.7769,
-      "step": 17630
-    },
-    {
-      "epoch": 9.194784876140808,
-      "grad_norm": 1.4163585901260376,
-      "learning_rate": 8.278793969849247e-05,
-      "loss": 5.6964,
-      "step": 17631
-    },
-    {
-      "epoch": 9.195306388526728,
-      "grad_norm": 1.423647165298462,
-      "learning_rate": 8.278693467336684e-05,
-      "loss": 5.7558,
-      "step": 17632
-    },
-    {
-      "epoch": 9.195827900912647,
-      "grad_norm": 1.3835922479629517,
-      "learning_rate": 8.278592964824122e-05,
-      "loss": 5.5271,
-      "step": 17633
-    },
-    {
-      "epoch": 9.196349413298567,
-      "grad_norm": 1.454591155052185,
-      "learning_rate": 8.278492462311558e-05,
-      "loss": 5.5688,
-      "step": 17634
-    },
-    {
-      "epoch": 9.196870925684484,
-      "grad_norm": 1.486068844795227,
-      "learning_rate": 8.278391959798996e-05,
-      "loss": 5.4146,
-      "step": 17635
-    },
-    {
-      "epoch": 9.197392438070404,
-      "grad_norm": 1.4386987686157227,
-      "learning_rate": 8.278291457286432e-05,
-      "loss": 5.4193,
-      "step": 17636
-    },
-    {
-      "epoch": 9.197913950456323,
-      "grad_norm": 1.450072169303894,
-      "learning_rate": 8.27819095477387e-05,
-      "loss": 5.8757,
-      "step": 17637
-    },
-    {
-      "epoch": 9.198435462842243,
-      "grad_norm": 1.4008394479751587,
-      "learning_rate": 8.278090452261306e-05,
-      "loss": 5.4398,
-      "step": 17638
-    },
-    {
-      "epoch": 9.198956975228162,
-      "grad_norm": 1.447314739227295,
-      "learning_rate": 8.277989949748744e-05,
-      "loss": 5.1488,
-      "step": 17639
-    },
-    {
-      "epoch": 9.199478487614082,
-      "grad_norm": 1.4682559967041016,
-      "learning_rate": 8.27788944723618e-05,
-      "loss": 4.8958,
-      "step": 17640
-    },
-    {
-      "epoch": 9.2,
-      "grad_norm": 1.4638503789901733,
-      "learning_rate": 8.277788944723618e-05,
-      "loss": 5.4602,
-      "step": 17641
-    },
-    {
-      "epoch": 9.200521512385919,
-      "grad_norm": 1.723026156425476,
-      "learning_rate": 8.277688442211056e-05,
-      "loss": 4.9565,
-      "step": 17642
-    },
-    {
-      "epoch": 9.201043024771838,
-      "grad_norm": 1.330110788345337,
-      "learning_rate": 8.277587939698493e-05,
-      "loss": 5.6894,
-      "step": 17643
-    },
-    {
-      "epoch": 9.201564537157758,
-      "grad_norm": 1.5313704013824463,
-      "learning_rate": 8.27748743718593e-05,
-      "loss": 5.2646,
-      "step": 17644
-    },
-    {
-      "epoch": 9.202086049543677,
-      "grad_norm": 1.430442214012146,
-      "learning_rate": 8.277386934673367e-05,
-      "loss": 5.5454,
-      "step": 17645
-    },
-    {
-      "epoch": 9.202607561929597,
-      "grad_norm": 1.3994237184524536,
-      "learning_rate": 8.277286432160805e-05,
-      "loss": 5.5042,
-      "step": 17646
-    },
-    {
-      "epoch": 9.203129074315514,
-      "grad_norm": 1.3784518241882324,
-      "learning_rate": 8.277185929648241e-05,
-      "loss": 5.2501,
-      "step": 17647
-    },
-    {
-      "epoch": 9.203650586701434,
-      "grad_norm": 1.4315900802612305,
-      "learning_rate": 8.277085427135679e-05,
-      "loss": 5.4179,
-      "step": 17648
-    },
-    {
-      "epoch": 9.204172099087353,
-      "grad_norm": 1.4309481382369995,
-      "learning_rate": 8.276984924623115e-05,
-      "loss": 5.5097,
-      "step": 17649
-    },
-    {
-      "epoch": 9.204693611473273,
-      "grad_norm": 1.5385570526123047,
-      "learning_rate": 8.276884422110553e-05,
-      "loss": 5.3369,
-      "step": 17650
-    },
-    {
-      "epoch": 9.205215123859192,
-      "grad_norm": 1.3369423151016235,
-      "learning_rate": 8.276783919597991e-05,
-      "loss": 5.7796,
-      "step": 17651
-    },
-    {
-      "epoch": 9.205736636245112,
-      "grad_norm": 3.2356762886047363,
-      "learning_rate": 8.276683417085429e-05,
-      "loss": 4.7624,
-      "step": 17652
-    },
-    {
-      "epoch": 9.20625814863103,
-      "grad_norm": 1.574173927307129,
-      "learning_rate": 8.276582914572865e-05,
-      "loss": 5.7549,
-      "step": 17653
-    },
-    {
-      "epoch": 9.206779661016949,
-      "grad_norm": 1.439113736152649,
-      "learning_rate": 8.276482412060303e-05,
-      "loss": 5.7702,
-      "step": 17654
-    },
-    {
-      "epoch": 9.207301173402868,
-      "grad_norm": 1.8748517036437988,
-      "learning_rate": 8.276381909547739e-05,
-      "loss": 5.1836,
-      "step": 17655
-    },
-    {
-      "epoch": 9.207822685788788,
-      "grad_norm": 1.4921178817749023,
-      "learning_rate": 8.276281407035176e-05,
-      "loss": 5.6261,
-      "step": 17656
-    },
-    {
-      "epoch": 9.208344198174707,
-      "grad_norm": 1.6219825744628906,
-      "learning_rate": 8.276180904522613e-05,
-      "loss": 5.3317,
-      "step": 17657
-    },
-    {
-      "epoch": 9.208865710560627,
-      "grad_norm": 1.4128663539886475,
-      "learning_rate": 8.27608040201005e-05,
-      "loss": 5.7334,
-      "step": 17658
-    },
-    {
-      "epoch": 9.209387222946544,
-      "grad_norm": 1.4538252353668213,
-      "learning_rate": 8.275979899497488e-05,
-      "loss": 5.2933,
-      "step": 17659
-    },
-    {
-      "epoch": 9.209908735332464,
-      "grad_norm": 1.4339607954025269,
-      "learning_rate": 8.275879396984924e-05,
-      "loss": 4.9229,
-      "step": 17660
-    },
-    {
-      "epoch": 9.210430247718383,
-      "grad_norm": 1.4889628887176514,
-      "learning_rate": 8.275778894472362e-05,
-      "loss": 5.8008,
-      "step": 17661
-    },
-    {
-      "epoch": 9.210951760104303,
-      "grad_norm": 1.4561161994934082,
-      "learning_rate": 8.2756783919598e-05,
-      "loss": 5.6648,
-      "step": 17662
-    },
-    {
-      "epoch": 9.211473272490222,
-      "grad_norm": 1.4771101474761963,
-      "learning_rate": 8.275577889447237e-05,
-      "loss": 5.5873,
-      "step": 17663
-    },
-    {
-      "epoch": 9.211994784876142,
-      "grad_norm": 1.453764796257019,
-      "learning_rate": 8.275477386934674e-05,
-      "loss": 5.1223,
-      "step": 17664
-    },
-    {
-      "epoch": 9.21251629726206,
-      "grad_norm": 1.5894732475280762,
-      "learning_rate": 8.275376884422112e-05,
-      "loss": 5.1655,
-      "step": 17665
-    },
-    {
-      "epoch": 9.213037809647979,
-      "grad_norm": 1.532426118850708,
-      "learning_rate": 8.275276381909548e-05,
-      "loss": 5.0926,
-      "step": 17666
-    },
-    {
-      "epoch": 9.213559322033898,
-      "grad_norm": 1.4770989418029785,
-      "learning_rate": 8.275175879396986e-05,
-      "loss": 5.681,
-      "step": 17667
-    },
-    {
-      "epoch": 9.214080834419818,
-      "grad_norm": 1.4984917640686035,
-      "learning_rate": 8.275075376884422e-05,
-      "loss": 5.4786,
-      "step": 17668
-    },
-    {
-      "epoch": 9.214602346805737,
-      "grad_norm": 1.602663278579712,
-      "learning_rate": 8.274974874371859e-05,
-      "loss": 5.2213,
-      "step": 17669
-    },
-    {
-      "epoch": 9.215123859191657,
-      "grad_norm": 1.6268552541732788,
-      "learning_rate": 8.274874371859296e-05,
-      "loss": 5.5139,
-      "step": 17670
-    },
-    {
-      "epoch": 9.215645371577574,
-      "grad_norm": 1.418068528175354,
-      "learning_rate": 8.274773869346734e-05,
-      "loss": 5.5173,
-      "step": 17671
-    },
-    {
-      "epoch": 9.216166883963494,
-      "grad_norm": 1.482157826423645,
-      "learning_rate": 8.274673366834172e-05,
-      "loss": 5.327,
-      "step": 17672
-    },
-    {
-      "epoch": 9.216688396349413,
-      "grad_norm": 1.4569991827011108,
-      "learning_rate": 8.274572864321608e-05,
-      "loss": 5.6689,
-      "step": 17673
-    },
-    {
-      "epoch": 9.217209908735333,
-      "grad_norm": 1.5634193420410156,
-      "learning_rate": 8.274472361809046e-05,
-      "loss": 5.5763,
-      "step": 17674
-    },
-    {
-      "epoch": 9.217731421121252,
-      "grad_norm": 1.4919040203094482,
-      "learning_rate": 8.274371859296483e-05,
-      "loss": 5.6473,
-      "step": 17675
-    },
-    {
-      "epoch": 9.21825293350717,
-      "grad_norm": 1.368727684020996,
-      "learning_rate": 8.27427135678392e-05,
-      "loss": 5.3042,
-      "step": 17676
-    },
-    {
-      "epoch": 9.21877444589309,
-      "grad_norm": 1.451511025428772,
-      "learning_rate": 8.274170854271357e-05,
-      "loss": 5.0175,
-      "step": 17677
-    },
-    {
-      "epoch": 9.219295958279009,
-      "grad_norm": 1.4704653024673462,
-      "learning_rate": 8.274070351758795e-05,
-      "loss": 5.8526,
-      "step": 17678
-    },
-    {
-      "epoch": 9.219817470664928,
-      "grad_norm": 1.3986951112747192,
-      "learning_rate": 8.273969849246231e-05,
-      "loss": 5.101,
-      "step": 17679
-    },
-    {
-      "epoch": 9.220338983050848,
-      "grad_norm": 1.4220006465911865,
-      "learning_rate": 8.273869346733669e-05,
-      "loss": 6.0345,
-      "step": 17680
-    },
-    {
-      "epoch": 9.220860495436767,
-      "grad_norm": 1.3828160762786865,
-      "learning_rate": 8.273768844221105e-05,
-      "loss": 5.365,
-      "step": 17681
-    },
-    {
-      "epoch": 9.221382007822687,
-      "grad_norm": 1.3801802396774292,
-      "learning_rate": 8.273668341708543e-05,
-      "loss": 5.4237,
-      "step": 17682
-    },
-    {
-      "epoch": 9.221903520208604,
-      "grad_norm": 1.5068162679672241,
-      "learning_rate": 8.273567839195981e-05,
-      "loss": 5.5745,
-      "step": 17683
-    },
-    {
-      "epoch": 9.222425032594524,
-      "grad_norm": 1.4924283027648926,
-      "learning_rate": 8.273467336683417e-05,
-      "loss": 5.5884,
-      "step": 17684
-    },
-    {
-      "epoch": 9.222946544980443,
-      "grad_norm": 1.5309468507766724,
-      "learning_rate": 8.273366834170855e-05,
-      "loss": 5.0803,
-      "step": 17685
-    },
-    {
-      "epoch": 9.223468057366363,
-      "grad_norm": 1.3742297887802124,
-      "learning_rate": 8.273266331658291e-05,
-      "loss": 5.3688,
-      "step": 17686
-    },
-    {
-      "epoch": 9.223989569752282,
-      "grad_norm": 1.4030663967132568,
-      "learning_rate": 8.273165829145729e-05,
-      "loss": 5.4677,
-      "step": 17687
-    },
-    {
-      "epoch": 9.2245110821382,
-      "grad_norm": 1.4440110921859741,
-      "learning_rate": 8.273065326633166e-05,
-      "loss": 5.3977,
-      "step": 17688
-    },
-    {
-      "epoch": 9.22503259452412,
-      "grad_norm": 1.4401971101760864,
-      "learning_rate": 8.272964824120603e-05,
-      "loss": 5.4229,
-      "step": 17689
-    },
-    {
-      "epoch": 9.225554106910039,
-      "grad_norm": 1.55340576171875,
-      "learning_rate": 8.27286432160804e-05,
-      "loss": 4.9109,
-      "step": 17690
-    },
-    {
-      "epoch": 9.226075619295958,
-      "grad_norm": 1.7044262886047363,
-      "learning_rate": 8.272763819095478e-05,
-      "loss": 5.547,
-      "step": 17691
-    },
-    {
-      "epoch": 9.226597131681878,
-      "grad_norm": 1.5394755601882935,
-      "learning_rate": 8.272663316582915e-05,
-      "loss": 5.6826,
-      "step": 17692
-    },
-    {
-      "epoch": 9.227118644067797,
-      "grad_norm": 1.85234534740448,
-      "learning_rate": 8.272562814070353e-05,
-      "loss": 5.19,
-      "step": 17693
-    },
-    {
-      "epoch": 9.227640156453715,
-      "grad_norm": 1.628873348236084,
-      "learning_rate": 8.27246231155779e-05,
-      "loss": 5.614,
-      "step": 17694
-    },
-    {
-      "epoch": 9.228161668839634,
-      "grad_norm": 1.4411709308624268,
-      "learning_rate": 8.272361809045227e-05,
-      "loss": 5.6469,
-      "step": 17695
-    },
-    {
-      "epoch": 9.228683181225554,
-      "grad_norm": 1.5104748010635376,
-      "learning_rate": 8.272261306532664e-05,
-      "loss": 5.5845,
-      "step": 17696
-    },
-    {
-      "epoch": 9.229204693611473,
-      "grad_norm": 1.5508170127868652,
-      "learning_rate": 8.2721608040201e-05,
-      "loss": 5.5121,
-      "step": 17697
-    },
-    {
-      "epoch": 9.229726205997393,
-      "grad_norm": 1.5393257141113281,
-      "learning_rate": 8.272060301507538e-05,
-      "loss": 5.7822,
-      "step": 17698
-    },
-    {
-      "epoch": 9.230247718383312,
-      "grad_norm": 1.4890458583831787,
-      "learning_rate": 8.271959798994975e-05,
-      "loss": 5.1785,
-      "step": 17699
-    },
-    {
-      "epoch": 9.23076923076923,
-      "grad_norm": 1.4417632818222046,
-      "learning_rate": 8.271859296482412e-05,
-      "loss": 5.4486,
-      "step": 17700
-    },
-    {
-      "epoch": 9.23129074315515,
-      "grad_norm": 1.496584415435791,
-      "learning_rate": 8.271758793969849e-05,
-      "loss": 5.3299,
-      "step": 17701
-    },
-    {
-      "epoch": 9.231812255541069,
-      "grad_norm": 1.4363856315612793,
-      "learning_rate": 8.271658291457287e-05,
-      "loss": 5.5088,
-      "step": 17702
-    },
-    {
-      "epoch": 9.232333767926988,
-      "grad_norm": 1.4310808181762695,
-      "learning_rate": 8.271557788944724e-05,
-      "loss": 5.2238,
-      "step": 17703
-    },
-    {
-      "epoch": 9.232855280312908,
-      "grad_norm": 1.5458145141601562,
-      "learning_rate": 8.271457286432162e-05,
-      "loss": 5.2272,
-      "step": 17704
-    },
-    {
-      "epoch": 9.233376792698827,
-      "grad_norm": 1.4364433288574219,
-      "learning_rate": 8.271356783919599e-05,
-      "loss": 5.7206,
-      "step": 17705
-    },
-    {
-      "epoch": 9.233898305084745,
-      "grad_norm": 1.3885798454284668,
-      "learning_rate": 8.271256281407036e-05,
-      "loss": 5.6752,
-      "step": 17706
-    },
-    {
-      "epoch": 9.234419817470664,
-      "grad_norm": 1.825056791305542,
-      "learning_rate": 8.271155778894473e-05,
-      "loss": 4.8135,
-      "step": 17707
-    },
-    {
-      "epoch": 9.234941329856584,
-      "grad_norm": 1.501104712486267,
-      "learning_rate": 8.27105527638191e-05,
-      "loss": 5.6924,
-      "step": 17708
-    },
-    {
-      "epoch": 9.235462842242503,
-      "grad_norm": 1.4547303915023804,
-      "learning_rate": 8.270954773869347e-05,
-      "loss": 5.4298,
-      "step": 17709
-    },
-    {
-      "epoch": 9.235984354628423,
-      "grad_norm": 1.5359396934509277,
-      "learning_rate": 8.270854271356783e-05,
-      "loss": 5.3711,
-      "step": 17710
-    },
-    {
-      "epoch": 9.236505867014342,
-      "grad_norm": 1.6270534992218018,
-      "learning_rate": 8.270753768844221e-05,
-      "loss": 5.1488,
-      "step": 17711
-    },
-    {
-      "epoch": 9.23702737940026,
-      "grad_norm": 1.5204668045043945,
-      "learning_rate": 8.270653266331659e-05,
-      "loss": 5.4652,
-      "step": 17712
-    },
-    {
-      "epoch": 9.23754889178618,
-      "grad_norm": 1.4600553512573242,
-      "learning_rate": 8.270552763819097e-05,
-      "loss": 5.8536,
-      "step": 17713
-    },
-    {
-      "epoch": 9.238070404172099,
-      "grad_norm": 1.570419192314148,
-      "learning_rate": 8.270452261306533e-05,
-      "loss": 5.7535,
-      "step": 17714
-    },
-    {
-      "epoch": 9.238591916558018,
-      "grad_norm": 1.5596985816955566,
-      "learning_rate": 8.270351758793971e-05,
-      "loss": 4.7723,
-      "step": 17715
-    },
-    {
-      "epoch": 9.239113428943938,
-      "grad_norm": 1.4300976991653442,
-      "learning_rate": 8.270251256281407e-05,
-      "loss": 5.869,
-      "step": 17716
-    },
-    {
-      "epoch": 9.239634941329857,
-      "grad_norm": 1.390319585800171,
-      "learning_rate": 8.270150753768845e-05,
-      "loss": 5.884,
-      "step": 17717
-    },
-    {
-      "epoch": 9.240156453715775,
-      "grad_norm": 1.4808224439620972,
-      "learning_rate": 8.270050251256282e-05,
-      "loss": 5.4536,
-      "step": 17718
-    },
-    {
-      "epoch": 9.240677966101694,
-      "grad_norm": 1.3515801429748535,
-      "learning_rate": 8.26994974874372e-05,
-      "loss": 5.4511,
-      "step": 17719
-    },
-    {
-      "epoch": 9.241199478487614,
-      "grad_norm": 1.4516122341156006,
-      "learning_rate": 8.269849246231156e-05,
-      "loss": 5.5167,
-      "step": 17720
-    },
-    {
-      "epoch": 9.241720990873533,
-      "grad_norm": 1.4122556447982788,
-      "learning_rate": 8.269748743718594e-05,
-      "loss": 5.7024,
-      "step": 17721
-    },
-    {
-      "epoch": 9.242242503259453,
-      "grad_norm": 1.501846194267273,
-      "learning_rate": 8.26964824120603e-05,
-      "loss": 5.4392,
-      "step": 17722
-    },
-    {
-      "epoch": 9.242764015645372,
-      "grad_norm": 1.4527744054794312,
-      "learning_rate": 8.269547738693468e-05,
-      "loss": 5.5754,
-      "step": 17723
-    },
-    {
-      "epoch": 9.24328552803129,
-      "grad_norm": 1.4426966905593872,
-      "learning_rate": 8.269447236180906e-05,
-      "loss": 4.868,
-      "step": 17724
-    },
-    {
-      "epoch": 9.24380704041721,
-      "grad_norm": 1.5528970956802368,
-      "learning_rate": 8.269346733668342e-05,
-      "loss": 5.261,
-      "step": 17725
-    },
-    {
-      "epoch": 9.244328552803129,
-      "grad_norm": 1.436089277267456,
-      "learning_rate": 8.26924623115578e-05,
-      "loss": 5.2065,
-      "step": 17726
-    },
-    {
-      "epoch": 9.244850065189048,
-      "grad_norm": 1.4731885194778442,
-      "learning_rate": 8.269145728643216e-05,
-      "loss": 5.2934,
-      "step": 17727
-    },
-    {
-      "epoch": 9.245371577574968,
-      "grad_norm": 1.4337959289550781,
-      "learning_rate": 8.269045226130654e-05,
-      "loss": 5.8395,
-      "step": 17728
-    },
-    {
-      "epoch": 9.245893089960887,
-      "grad_norm": 1.3323200941085815,
-      "learning_rate": 8.26894472361809e-05,
-      "loss": 5.722,
-      "step": 17729
-    },
-    {
-      "epoch": 9.246414602346805,
-      "grad_norm": 1.4669828414916992,
-      "learning_rate": 8.268844221105528e-05,
-      "loss": 5.3796,
-      "step": 17730
-    },
-    {
-      "epoch": 9.246936114732724,
-      "grad_norm": 1.4089971780776978,
-      "learning_rate": 8.268743718592965e-05,
-      "loss": 5.6769,
-      "step": 17731
-    },
-    {
-      "epoch": 9.247457627118644,
-      "grad_norm": 1.4375344514846802,
-      "learning_rate": 8.268643216080402e-05,
-      "loss": 5.766,
-      "step": 17732
-    },
-    {
-      "epoch": 9.247979139504563,
-      "grad_norm": 1.368233561515808,
-      "learning_rate": 8.26854271356784e-05,
-      "loss": 5.6777,
-      "step": 17733
-    },
-    {
-      "epoch": 9.248500651890483,
-      "grad_norm": 1.322202205657959,
-      "learning_rate": 8.268442211055278e-05,
-      "loss": 5.9624,
-      "step": 17734
-    },
-    {
-      "epoch": 9.249022164276402,
-      "grad_norm": 1.480490803718567,
-      "learning_rate": 8.268341708542714e-05,
-      "loss": 5.7003,
-      "step": 17735
-    },
-    {
-      "epoch": 9.24954367666232,
-      "grad_norm": 1.52757728099823,
-      "learning_rate": 8.268241206030151e-05,
-      "loss": 5.4412,
-      "step": 17736
-    },
-    {
-      "epoch": 9.25006518904824,
-      "grad_norm": 1.489312767982483,
-      "learning_rate": 8.268140703517589e-05,
-      "loss": 5.2207,
-      "step": 17737
-    },
-    {
-      "epoch": 9.250586701434159,
-      "grad_norm": 1.373445749282837,
-      "learning_rate": 8.268040201005025e-05,
-      "loss": 5.4687,
-      "step": 17738
-    },
-    {
-      "epoch": 9.251108213820078,
-      "grad_norm": 1.4333490133285522,
-      "learning_rate": 8.267939698492463e-05,
-      "loss": 5.3688,
-      "step": 17739
-    },
-    {
-      "epoch": 9.251629726205998,
-      "grad_norm": 1.6089802980422974,
-      "learning_rate": 8.267839195979899e-05,
-      "loss": 5.4163,
-      "step": 17740
-    },
-    {
-      "epoch": 9.252151238591917,
-      "grad_norm": 1.6360340118408203,
-      "learning_rate": 8.267738693467337e-05,
-      "loss": 5.4082,
-      "step": 17741
-    },
-    {
-      "epoch": 9.252672750977835,
-      "grad_norm": 1.397901177406311,
-      "learning_rate": 8.267638190954773e-05,
-      "loss": 5.6299,
-      "step": 17742
-    },
-    {
-      "epoch": 9.253194263363755,
-      "grad_norm": 1.5862754583358765,
-      "learning_rate": 8.267537688442211e-05,
-      "loss": 5.4751,
-      "step": 17743
-    },
-    {
-      "epoch": 9.253715775749674,
-      "grad_norm": 1.5196529626846313,
-      "learning_rate": 8.267437185929649e-05,
-      "loss": 5.1015,
-      "step": 17744
-    },
-    {
-      "epoch": 9.254237288135593,
-      "grad_norm": 1.425155758857727,
-      "learning_rate": 8.267336683417087e-05,
-      "loss": 5.6747,
-      "step": 17745
-    },
-    {
-      "epoch": 9.254758800521513,
-      "grad_norm": 1.3501734733581543,
-      "learning_rate": 8.267236180904523e-05,
-      "loss": 5.7347,
-      "step": 17746
-    },
-    {
-      "epoch": 9.255280312907432,
-      "grad_norm": 1.3600839376449585,
-      "learning_rate": 8.267135678391961e-05,
-      "loss": 5.1343,
-      "step": 17747
-    },
-    {
-      "epoch": 9.25580182529335,
-      "grad_norm": 1.443397045135498,
-      "learning_rate": 8.267035175879397e-05,
-      "loss": 5.4496,
-      "step": 17748
-    },
-    {
-      "epoch": 9.25632333767927,
-      "grad_norm": 1.40093195438385,
-      "learning_rate": 8.266934673366834e-05,
-      "loss": 5.8704,
-      "step": 17749
-    },
-    {
-      "epoch": 9.256844850065189,
-      "grad_norm": 1.5421353578567505,
-      "learning_rate": 8.266834170854272e-05,
-      "loss": 5.6515,
-      "step": 17750
-    },
-    {
-      "epoch": 9.257366362451108,
-      "grad_norm": 1.4101674556732178,
-      "learning_rate": 8.266733668341708e-05,
-      "loss": 5.2107,
-      "step": 17751
-    },
-    {
-      "epoch": 9.257887874837028,
-      "grad_norm": 1.5319768190383911,
-      "learning_rate": 8.266633165829146e-05,
-      "loss": 5.5386,
-      "step": 17752
-    },
-    {
-      "epoch": 9.258409387222947,
-      "grad_norm": 1.455776333808899,
-      "learning_rate": 8.266532663316582e-05,
-      "loss": 5.768,
-      "step": 17753
-    },
-    {
-      "epoch": 9.258930899608865,
-      "grad_norm": 1.3888945579528809,
-      "learning_rate": 8.26643216080402e-05,
-      "loss": 5.6254,
-      "step": 17754
-    },
-    {
-      "epoch": 9.259452411994785,
-      "grad_norm": 1.4808905124664307,
-      "learning_rate": 8.266331658291458e-05,
-      "loss": 5.6364,
-      "step": 17755
-    },
-    {
-      "epoch": 9.259973924380704,
-      "grad_norm": 1.31795072555542,
-      "learning_rate": 8.266231155778896e-05,
-      "loss": 4.6417,
-      "step": 17756
-    },
-    {
-      "epoch": 9.260495436766623,
-      "grad_norm": 1.4797512292861938,
-      "learning_rate": 8.266130653266332e-05,
-      "loss": 5.6042,
-      "step": 17757
-    },
-    {
-      "epoch": 9.261016949152543,
-      "grad_norm": 1.4107348918914795,
-      "learning_rate": 8.26603015075377e-05,
-      "loss": 5.6475,
-      "step": 17758
-    },
-    {
-      "epoch": 9.261538461538462,
-      "grad_norm": 1.6351252794265747,
-      "learning_rate": 8.265929648241206e-05,
-      "loss": 4.9061,
-      "step": 17759
-    },
-    {
-      "epoch": 9.26205997392438,
-      "grad_norm": 1.4848254919052124,
-      "learning_rate": 8.265829145728644e-05,
-      "loss": 5.3394,
-      "step": 17760
-    },
-    {
-      "epoch": 9.2625814863103,
-      "grad_norm": 1.4435182809829712,
-      "learning_rate": 8.26572864321608e-05,
-      "loss": 5.5685,
-      "step": 17761
-    },
-    {
-      "epoch": 9.263102998696219,
-      "grad_norm": 1.4421178102493286,
-      "learning_rate": 8.265628140703517e-05,
-      "loss": 5.5667,
-      "step": 17762
-    },
-    {
-      "epoch": 9.263624511082138,
-      "grad_norm": 1.3644686937332153,
-      "learning_rate": 8.265527638190955e-05,
-      "loss": 5.8948,
-      "step": 17763
-    },
-    {
-      "epoch": 9.264146023468058,
-      "grad_norm": 1.7325308322906494,
-      "learning_rate": 8.265427135678392e-05,
-      "loss": 5.0516,
-      "step": 17764
-    },
-    {
-      "epoch": 9.264667535853977,
-      "grad_norm": 1.352449655532837,
-      "learning_rate": 8.26532663316583e-05,
-      "loss": 5.4747,
-      "step": 17765
-    },
-    {
-      "epoch": 9.265189048239895,
-      "grad_norm": 1.326794981956482,
-      "learning_rate": 8.265226130653267e-05,
-      "loss": 5.8045,
-      "step": 17766
-    },
-    {
-      "epoch": 9.265710560625815,
-      "grad_norm": 1.4304171800613403,
-      "learning_rate": 8.265125628140704e-05,
-      "loss": 5.5455,
-      "step": 17767
-    },
-    {
-      "epoch": 9.266232073011734,
-      "grad_norm": 1.515570044517517,
-      "learning_rate": 8.265025125628141e-05,
-      "loss": 5.6154,
-      "step": 17768
-    },
-    {
-      "epoch": 9.266753585397653,
-      "grad_norm": 1.5441948175430298,
-      "learning_rate": 8.264924623115579e-05,
-      "loss": 5.3523,
-      "step": 17769
-    },
-    {
-      "epoch": 9.267275097783573,
-      "grad_norm": 1.4001528024673462,
-      "learning_rate": 8.264824120603015e-05,
-      "loss": 5.9866,
-      "step": 17770
-    },
-    {
-      "epoch": 9.26779661016949,
-      "grad_norm": 1.4114130735397339,
-      "learning_rate": 8.264723618090453e-05,
-      "loss": 5.2829,
-      "step": 17771
-    },
-    {
-      "epoch": 9.26831812255541,
-      "grad_norm": 1.2364174127578735,
-      "learning_rate": 8.264623115577889e-05,
-      "loss": 5.9014,
-      "step": 17772
-    },
-    {
-      "epoch": 9.26883963494133,
-      "grad_norm": 1.2977726459503174,
-      "learning_rate": 8.264522613065327e-05,
-      "loss": 5.8592,
-      "step": 17773
-    },
-    {
-      "epoch": 9.269361147327249,
-      "grad_norm": 1.4391499757766724,
-      "learning_rate": 8.264422110552764e-05,
-      "loss": 5.338,
-      "step": 17774
-    },
-    {
-      "epoch": 9.269882659713168,
-      "grad_norm": 1.4657695293426514,
-      "learning_rate": 8.264321608040201e-05,
-      "loss": 5.2253,
-      "step": 17775
-    },
-    {
-      "epoch": 9.270404172099088,
-      "grad_norm": 1.4366235733032227,
-      "learning_rate": 8.264221105527639e-05,
-      "loss": 5.3919,
-      "step": 17776
-    },
-    {
-      "epoch": 9.270925684485007,
-      "grad_norm": 1.4689220190048218,
-      "learning_rate": 8.264120603015075e-05,
-      "loss": 5.1949,
-      "step": 17777
-    },
-    {
-      "epoch": 9.271447196870925,
-      "grad_norm": 1.3017221689224243,
-      "learning_rate": 8.264020100502513e-05,
-      "loss": 5.8293,
-      "step": 17778
-    },
-    {
-      "epoch": 9.271968709256845,
-      "grad_norm": 1.3868951797485352,
-      "learning_rate": 8.26391959798995e-05,
-      "loss": 5.8295,
-      "step": 17779
-    },
-    {
-      "epoch": 9.272490221642764,
-      "grad_norm": 1.4491794109344482,
-      "learning_rate": 8.263819095477387e-05,
-      "loss": 5.615,
-      "step": 17780
-    },
-    {
-      "epoch": 9.273011734028684,
-      "grad_norm": 1.4298210144042969,
-      "learning_rate": 8.263718592964824e-05,
-      "loss": 5.5277,
-      "step": 17781
-    },
-    {
-      "epoch": 9.273533246414603,
-      "grad_norm": 1.3873612880706787,
-      "learning_rate": 8.263618090452262e-05,
-      "loss": 5.9259,
-      "step": 17782
-    },
-    {
-      "epoch": 9.27405475880052,
-      "grad_norm": 1.481906771659851,
-      "learning_rate": 8.263517587939698e-05,
-      "loss": 5.0981,
-      "step": 17783
-    },
-    {
-      "epoch": 9.27457627118644,
-      "grad_norm": 1.4937111139297485,
-      "learning_rate": 8.263417085427136e-05,
-      "loss": 5.7052,
-      "step": 17784
-    },
-    {
-      "epoch": 9.27509778357236,
-      "grad_norm": 1.462133765220642,
-      "learning_rate": 8.263316582914574e-05,
-      "loss": 5.5949,
-      "step": 17785
-    },
-    {
-      "epoch": 9.275619295958279,
-      "grad_norm": 1.4933279752731323,
-      "learning_rate": 8.263216080402011e-05,
-      "loss": 5.1822,
-      "step": 17786
-    },
-    {
-      "epoch": 9.276140808344199,
-      "grad_norm": 1.3937264680862427,
-      "learning_rate": 8.263115577889448e-05,
-      "loss": 5.1483,
-      "step": 17787
-    },
-    {
-      "epoch": 9.276662320730118,
-      "grad_norm": 1.3768932819366455,
-      "learning_rate": 8.263015075376886e-05,
-      "loss": 5.2787,
-      "step": 17788
-    },
-    {
-      "epoch": 9.277183833116036,
-      "grad_norm": 1.4572068452835083,
-      "learning_rate": 8.262914572864322e-05,
-      "loss": 5.3111,
-      "step": 17789
-    },
-    {
-      "epoch": 9.277705345501955,
-      "grad_norm": 1.4442051649093628,
-      "learning_rate": 8.262814070351759e-05,
-      "loss": 5.2887,
-      "step": 17790
-    },
-    {
-      "epoch": 9.278226857887875,
-      "grad_norm": 1.4284992218017578,
-      "learning_rate": 8.262713567839196e-05,
-      "loss": 5.5789,
-      "step": 17791
-    },
-    {
-      "epoch": 9.278748370273794,
-      "grad_norm": 1.4486550092697144,
-      "learning_rate": 8.262613065326633e-05,
-      "loss": 5.6187,
-      "step": 17792
-    },
-    {
-      "epoch": 9.279269882659714,
-      "grad_norm": 1.5003979206085205,
-      "learning_rate": 8.26251256281407e-05,
-      "loss": 5.6175,
-      "step": 17793
-    },
-    {
-      "epoch": 9.279791395045633,
-      "grad_norm": 1.545966625213623,
-      "learning_rate": 8.262412060301507e-05,
-      "loss": 5.4569,
-      "step": 17794
-    },
-    {
-      "epoch": 9.28031290743155,
-      "grad_norm": 1.5083329677581787,
-      "learning_rate": 8.262311557788945e-05,
-      "loss": 5.6087,
-      "step": 17795
-    },
-    {
-      "epoch": 9.28083441981747,
-      "grad_norm": 1.4852930307388306,
-      "learning_rate": 8.262211055276383e-05,
-      "loss": 5.855,
-      "step": 17796
-    },
-    {
-      "epoch": 9.28135593220339,
-      "grad_norm": 1.4113904237747192,
-      "learning_rate": 8.26211055276382e-05,
-      "loss": 5.4847,
-      "step": 17797
-    },
-    {
-      "epoch": 9.281877444589309,
-      "grad_norm": 1.3508943319320679,
-      "learning_rate": 8.262010050251257e-05,
-      "loss": 5.6492,
-      "step": 17798
-    },
-    {
-      "epoch": 9.282398956975229,
-      "grad_norm": 1.3639181852340698,
-      "learning_rate": 8.261909547738695e-05,
-      "loss": 5.8892,
-      "step": 17799
-    },
-    {
-      "epoch": 9.282920469361148,
-      "grad_norm": 1.4258133172988892,
-      "learning_rate": 8.261809045226131e-05,
-      "loss": 5.8121,
-      "step": 17800
-    },
-    {
-      "epoch": 9.283441981747066,
-      "grad_norm": 1.454408884048462,
-      "learning_rate": 8.261708542713569e-05,
-      "loss": 5.39,
-      "step": 17801
-    },
-    {
-      "epoch": 9.283963494132985,
-      "grad_norm": 1.47089684009552,
-      "learning_rate": 8.261608040201005e-05,
-      "loss": 5.2088,
-      "step": 17802
-    },
-    {
-      "epoch": 9.284485006518905,
-      "grad_norm": 1.5482689142227173,
-      "learning_rate": 8.261507537688442e-05,
-      "loss": 5.3165,
-      "step": 17803
-    },
-    {
-      "epoch": 9.285006518904824,
-      "grad_norm": 1.501946210861206,
-      "learning_rate": 8.26140703517588e-05,
-      "loss": 5.8013,
-      "step": 17804
-    },
-    {
-      "epoch": 9.285528031290744,
-      "grad_norm": 1.4821916818618774,
-      "learning_rate": 8.261306532663317e-05,
-      "loss": 5.4193,
-      "step": 17805
-    },
-    {
-      "epoch": 9.286049543676663,
-      "grad_norm": 1.419727087020874,
-      "learning_rate": 8.261206030150755e-05,
-      "loss": 5.7411,
-      "step": 17806
-    },
-    {
-      "epoch": 9.28657105606258,
-      "grad_norm": 1.5094538927078247,
-      "learning_rate": 8.261105527638191e-05,
-      "loss": 5.1663,
-      "step": 17807
-    },
-    {
-      "epoch": 9.2870925684485,
-      "grad_norm": 1.438965082168579,
-      "learning_rate": 8.261005025125629e-05,
-      "loss": 5.6341,
-      "step": 17808
-    },
-    {
-      "epoch": 9.28761408083442,
-      "grad_norm": 1.3845906257629395,
-      "learning_rate": 8.260904522613066e-05,
-      "loss": 5.5458,
-      "step": 17809
-    },
-    {
-      "epoch": 9.288135593220339,
-      "grad_norm": 1.5995172262191772,
-      "learning_rate": 8.260804020100503e-05,
-      "loss": 5.6844,
-      "step": 17810
-    },
-    {
-      "epoch": 9.288657105606259,
-      "grad_norm": 1.4947564601898193,
-      "learning_rate": 8.26070351758794e-05,
-      "loss": 5.4382,
-      "step": 17811
-    },
-    {
-      "epoch": 9.289178617992178,
-      "grad_norm": 1.4404096603393555,
-      "learning_rate": 8.260603015075378e-05,
-      "loss": 5.6746,
-      "step": 17812
-    },
-    {
-      "epoch": 9.289700130378096,
-      "grad_norm": 1.4321869611740112,
-      "learning_rate": 8.260502512562814e-05,
-      "loss": 5.6774,
-      "step": 17813
-    },
-    {
-      "epoch": 9.290221642764015,
-      "grad_norm": 1.4165780544281006,
-      "learning_rate": 8.260402010050252e-05,
-      "loss": 5.7601,
-      "step": 17814
-    },
-    {
-      "epoch": 9.290743155149935,
-      "grad_norm": 1.5595192909240723,
-      "learning_rate": 8.260301507537688e-05,
-      "loss": 5.256,
-      "step": 17815
-    },
-    {
-      "epoch": 9.291264667535854,
-      "grad_norm": 1.3722832202911377,
-      "learning_rate": 8.260201005025126e-05,
-      "loss": 5.4997,
-      "step": 17816
-    },
-    {
-      "epoch": 9.291786179921774,
-      "grad_norm": 1.419309377670288,
-      "learning_rate": 8.260100502512564e-05,
-      "loss": 5.5381,
-      "step": 17817
-    },
-    {
-      "epoch": 9.292307692307693,
-      "grad_norm": 1.4021756649017334,
-      "learning_rate": 8.26e-05,
-      "loss": 5.5132,
-      "step": 17818
-    },
-    {
-      "epoch": 9.29282920469361,
-      "grad_norm": 1.3934431076049805,
-      "learning_rate": 8.259899497487438e-05,
-      "loss": 5.653,
-      "step": 17819
-    },
-    {
-      "epoch": 9.29335071707953,
-      "grad_norm": 1.3754043579101562,
-      "learning_rate": 8.259798994974874e-05,
-      "loss": 5.8093,
-      "step": 17820
-    },
-    {
-      "epoch": 9.29387222946545,
-      "grad_norm": 1.460658073425293,
-      "learning_rate": 8.259698492462312e-05,
-      "loss": 5.8559,
-      "step": 17821
-    },
-    {
-      "epoch": 9.29439374185137,
-      "grad_norm": 1.4343043565750122,
-      "learning_rate": 8.259597989949749e-05,
-      "loss": 5.7651,
-      "step": 17822
-    },
-    {
-      "epoch": 9.294915254237289,
-      "grad_norm": 1.6085783243179321,
-      "learning_rate": 8.259497487437186e-05,
-      "loss": 5.2225,
-      "step": 17823
-    },
-    {
-      "epoch": 9.295436766623208,
-      "grad_norm": 1.387020468711853,
-      "learning_rate": 8.259396984924623e-05,
-      "loss": 4.83,
-      "step": 17824
-    },
-    {
-      "epoch": 9.295958279009126,
-      "grad_norm": 1.401745319366455,
-      "learning_rate": 8.25929648241206e-05,
-      "loss": 5.6094,
-      "step": 17825
-    },
-    {
-      "epoch": 9.296479791395045,
-      "grad_norm": 1.4833757877349854,
-      "learning_rate": 8.259195979899498e-05,
-      "loss": 5.6106,
-      "step": 17826
-    },
-    {
-      "epoch": 9.297001303780965,
-      "grad_norm": 1.5132322311401367,
-      "learning_rate": 8.259095477386936e-05,
-      "loss": 5.8721,
-      "step": 17827
-    },
-    {
-      "epoch": 9.297522816166884,
-      "grad_norm": 1.4045954942703247,
-      "learning_rate": 8.258994974874373e-05,
-      "loss": 5.4682,
-      "step": 17828
-    },
-    {
-      "epoch": 9.298044328552804,
-      "grad_norm": 1.4058681726455688,
-      "learning_rate": 8.258894472361809e-05,
-      "loss": 5.6622,
-      "step": 17829
-    },
-    {
-      "epoch": 9.298565840938723,
-      "grad_norm": 1.424513339996338,
-      "learning_rate": 8.258793969849247e-05,
-      "loss": 5.7598,
-      "step": 17830
-    },
-    {
-      "epoch": 9.29908735332464,
-      "grad_norm": 1.4308576583862305,
-      "learning_rate": 8.258693467336683e-05,
-      "loss": 5.4005,
-      "step": 17831
-    },
-    {
-      "epoch": 9.29960886571056,
-      "grad_norm": 1.5109597444534302,
-      "learning_rate": 8.258592964824121e-05,
-      "loss": 4.7401,
-      "step": 17832
-    },
-    {
-      "epoch": 9.30013037809648,
-      "grad_norm": 1.4433708190917969,
-      "learning_rate": 8.258492462311557e-05,
-      "loss": 5.126,
-      "step": 17833
-    },
-    {
-      "epoch": 9.3006518904824,
-      "grad_norm": 1.3627104759216309,
-      "learning_rate": 8.258391959798995e-05,
-      "loss": 5.9153,
-      "step": 17834
-    },
-    {
-      "epoch": 9.301173402868319,
-      "grad_norm": 1.5091513395309448,
-      "learning_rate": 8.258291457286432e-05,
-      "loss": 5.4223,
-      "step": 17835
-    },
-    {
-      "epoch": 9.301694915254238,
-      "grad_norm": 1.5741223096847534,
-      "learning_rate": 8.25819095477387e-05,
-      "loss": 5.2912,
-      "step": 17836
-    },
-    {
-      "epoch": 9.302216427640156,
-      "grad_norm": 1.3514258861541748,
-      "learning_rate": 8.258090452261307e-05,
-      "loss": 5.8114,
-      "step": 17837
-    },
-    {
-      "epoch": 9.302737940026075,
-      "grad_norm": 1.470290184020996,
-      "learning_rate": 8.257989949748745e-05,
-      "loss": 5.3753,
-      "step": 17838
-    },
-    {
-      "epoch": 9.303259452411995,
-      "grad_norm": 1.4314285516738892,
-      "learning_rate": 8.257889447236181e-05,
-      "loss": 5.6803,
-      "step": 17839
-    },
-    {
-      "epoch": 9.303780964797914,
-      "grad_norm": 1.3847752809524536,
-      "learning_rate": 8.257788944723619e-05,
-      "loss": 5.8366,
-      "step": 17840
-    },
-    {
-      "epoch": 9.304302477183834,
-      "grad_norm": 1.4238336086273193,
-      "learning_rate": 8.257688442211056e-05,
-      "loss": 5.624,
-      "step": 17841
-    },
-    {
-      "epoch": 9.304823989569753,
-      "grad_norm": 1.4281706809997559,
-      "learning_rate": 8.257587939698492e-05,
-      "loss": 6.0283,
-      "step": 17842
-    },
-    {
-      "epoch": 9.30534550195567,
-      "grad_norm": 1.339417576789856,
-      "learning_rate": 8.25748743718593e-05,
-      "loss": 5.9674,
-      "step": 17843
-    },
-    {
-      "epoch": 9.30586701434159,
-      "grad_norm": 1.3529119491577148,
-      "learning_rate": 8.257386934673366e-05,
-      "loss": 5.801,
-      "step": 17844
-    },
-    {
-      "epoch": 9.30638852672751,
-      "grad_norm": 1.4565587043762207,
-      "learning_rate": 8.257286432160804e-05,
-      "loss": 5.1853,
-      "step": 17845
-    },
-    {
-      "epoch": 9.30691003911343,
-      "grad_norm": 1.4607961177825928,
-      "learning_rate": 8.257185929648242e-05,
-      "loss": 5.5579,
-      "step": 17846
-    },
-    {
-      "epoch": 9.307431551499349,
-      "grad_norm": 1.53420889377594,
-      "learning_rate": 8.25708542713568e-05,
-      "loss": 5.8213,
-      "step": 17847
-    },
-    {
-      "epoch": 9.307953063885268,
-      "grad_norm": 1.3008708953857422,
-      "learning_rate": 8.256984924623116e-05,
-      "loss": 5.6503,
-      "step": 17848
-    },
-    {
-      "epoch": 9.308474576271186,
-      "grad_norm": 1.4964479207992554,
-      "learning_rate": 8.256884422110554e-05,
-      "loss": 5.3106,
-      "step": 17849
-    },
-    {
-      "epoch": 9.308996088657105,
-      "grad_norm": 1.3184634447097778,
-      "learning_rate": 8.25678391959799e-05,
-      "loss": 5.7327,
-      "step": 17850
-    },
-    {
-      "epoch": 9.309517601043025,
-      "grad_norm": 1.414868950843811,
-      "learning_rate": 8.256683417085428e-05,
-      "loss": 5.2075,
-      "step": 17851
-    },
-    {
-      "epoch": 9.310039113428944,
-      "grad_norm": 1.4079985618591309,
-      "learning_rate": 8.256582914572864e-05,
-      "loss": 5.424,
-      "step": 17852
-    },
-    {
-      "epoch": 9.310560625814864,
-      "grad_norm": 1.4072798490524292,
-      "learning_rate": 8.256482412060302e-05,
-      "loss": 5.4649,
-      "step": 17853
-    },
-    {
-      "epoch": 9.311082138200783,
-      "grad_norm": 1.432481050491333,
-      "learning_rate": 8.256381909547739e-05,
-      "loss": 5.4919,
-      "step": 17854
-    },
-    {
-      "epoch": 9.3116036505867,
-      "grad_norm": 1.766871452331543,
-      "learning_rate": 8.256281407035176e-05,
-      "loss": 4.6631,
-      "step": 17855
-    },
-    {
-      "epoch": 9.31212516297262,
-      "grad_norm": 1.3734943866729736,
-      "learning_rate": 8.256180904522613e-05,
-      "loss": 5.9,
-      "step": 17856
-    },
-    {
-      "epoch": 9.31264667535854,
-      "grad_norm": 1.4107590913772583,
-      "learning_rate": 8.256080402010051e-05,
-      "loss": 5.4374,
-      "step": 17857
-    },
-    {
-      "epoch": 9.31316818774446,
-      "grad_norm": 1.4185676574707031,
-      "learning_rate": 8.255979899497488e-05,
-      "loss": 5.6754,
-      "step": 17858
-    },
-    {
-      "epoch": 9.313689700130379,
-      "grad_norm": 1.3607779741287231,
-      "learning_rate": 8.255879396984925e-05,
-      "loss": 5.8673,
-      "step": 17859
-    },
-    {
-      "epoch": 9.314211212516298,
-      "grad_norm": 1.4063372611999512,
-      "learning_rate": 8.255778894472363e-05,
-      "loss": 5.8023,
-      "step": 17860
-    },
-    {
-      "epoch": 9.314732724902216,
-      "grad_norm": 1.4814351797103882,
-      "learning_rate": 8.255678391959799e-05,
-      "loss": 5.2821,
-      "step": 17861
-    },
-    {
-      "epoch": 9.315254237288135,
-      "grad_norm": 1.386693000793457,
-      "learning_rate": 8.255577889447237e-05,
-      "loss": 5.421,
-      "step": 17862
-    },
-    {
-      "epoch": 9.315775749674055,
-      "grad_norm": 1.2949022054672241,
-      "learning_rate": 8.255477386934673e-05,
-      "loss": 5.7929,
-      "step": 17863
-    },
-    {
-      "epoch": 9.316297262059974,
-      "grad_norm": 1.4064475297927856,
-      "learning_rate": 8.255376884422111e-05,
-      "loss": 5.6332,
-      "step": 17864
-    },
-    {
-      "epoch": 9.316818774445894,
-      "grad_norm": 1.3515608310699463,
-      "learning_rate": 8.255276381909548e-05,
-      "loss": 5.9011,
-      "step": 17865
-    },
-    {
-      "epoch": 9.317340286831811,
-      "grad_norm": 1.490886926651001,
-      "learning_rate": 8.255175879396985e-05,
-      "loss": 5.5776,
-      "step": 17866
-    },
-    {
-      "epoch": 9.31786179921773,
-      "grad_norm": 1.3623943328857422,
-      "learning_rate": 8.255075376884423e-05,
-      "loss": 5.6724,
-      "step": 17867
-    },
-    {
-      "epoch": 9.31838331160365,
-      "grad_norm": 1.4380289316177368,
-      "learning_rate": 8.254974874371861e-05,
-      "loss": 5.4623,
-      "step": 17868
-    },
-    {
-      "epoch": 9.31890482398957,
-      "grad_norm": 1.5070984363555908,
-      "learning_rate": 8.254874371859297e-05,
-      "loss": 5.3932,
-      "step": 17869
-    },
-    {
-      "epoch": 9.31942633637549,
-      "grad_norm": 1.626399278640747,
-      "learning_rate": 8.254773869346734e-05,
-      "loss": 5.1842,
-      "step": 17870
-    },
-    {
-      "epoch": 9.319947848761409,
-      "grad_norm": 1.7615388631820679,
-      "learning_rate": 8.254673366834172e-05,
-      "loss": 5.3426,
-      "step": 17871
-    },
-    {
-      "epoch": 9.320469361147328,
-      "grad_norm": 1.5111043453216553,
-      "learning_rate": 8.254572864321608e-05,
-      "loss": 5.6556,
-      "step": 17872
-    },
-    {
-      "epoch": 9.320990873533246,
-      "grad_norm": 1.4626452922821045,
-      "learning_rate": 8.254472361809046e-05,
-      "loss": 5.4478,
-      "step": 17873
-    },
-    {
-      "epoch": 9.321512385919165,
-      "grad_norm": 1.5261832475662231,
-      "learning_rate": 8.254371859296482e-05,
-      "loss": 5.2033,
-      "step": 17874
-    },
-    {
-      "epoch": 9.322033898305085,
-      "grad_norm": 1.4829351902008057,
-      "learning_rate": 8.25427135678392e-05,
-      "loss": 5.2263,
-      "step": 17875
-    },
-    {
-      "epoch": 9.322555410691004,
-      "grad_norm": 1.4065306186676025,
-      "learning_rate": 8.254170854271356e-05,
-      "loss": 5.5689,
-      "step": 17876
-    },
-    {
-      "epoch": 9.323076923076924,
-      "grad_norm": 1.4220730066299438,
-      "learning_rate": 8.254070351758794e-05,
-      "loss": 5.6552,
-      "step": 17877
-    },
-    {
-      "epoch": 9.323598435462841,
-      "grad_norm": 1.432881474494934,
-      "learning_rate": 8.253969849246232e-05,
-      "loss": 5.7778,
-      "step": 17878
-    },
-    {
-      "epoch": 9.32411994784876,
-      "grad_norm": 1.5804471969604492,
-      "learning_rate": 8.25386934673367e-05,
-      "loss": 5.5164,
-      "step": 17879
-    },
-    {
-      "epoch": 9.32464146023468,
-      "grad_norm": 1.488150954246521,
-      "learning_rate": 8.253768844221106e-05,
-      "loss": 5.472,
-      "step": 17880
-    },
-    {
-      "epoch": 9.3251629726206,
-      "grad_norm": 1.4708503484725952,
-      "learning_rate": 8.253668341708544e-05,
-      "loss": 5.412,
-      "step": 17881
-    },
-    {
-      "epoch": 9.32568448500652,
-      "grad_norm": 1.4484410285949707,
-      "learning_rate": 8.25356783919598e-05,
-      "loss": 5.7416,
-      "step": 17882
-    },
-    {
-      "epoch": 9.326205997392439,
-      "grad_norm": 1.460995078086853,
-      "learning_rate": 8.253467336683417e-05,
-      "loss": 5.6304,
-      "step": 17883
-    },
-    {
-      "epoch": 9.326727509778356,
-      "grad_norm": 1.4780486822128296,
-      "learning_rate": 8.253366834170855e-05,
-      "loss": 5.4301,
-      "step": 17884
-    },
-    {
-      "epoch": 9.327249022164276,
-      "grad_norm": 1.4285001754760742,
-      "learning_rate": 8.253266331658291e-05,
-      "loss": 5.7071,
-      "step": 17885
-    },
-    {
-      "epoch": 9.327770534550195,
-      "grad_norm": 1.4567639827728271,
-      "learning_rate": 8.253165829145729e-05,
-      "loss": 5.7185,
-      "step": 17886
-    },
-    {
-      "epoch": 9.328292046936115,
-      "grad_norm": 1.35545015335083,
-      "learning_rate": 8.253065326633167e-05,
-      "loss": 5.6046,
-      "step": 17887
-    },
-    {
-      "epoch": 9.328813559322034,
-      "grad_norm": 1.4215872287750244,
-      "learning_rate": 8.252964824120604e-05,
-      "loss": 5.4419,
-      "step": 17888
-    },
-    {
-      "epoch": 9.329335071707954,
-      "grad_norm": 1.3905999660491943,
-      "learning_rate": 8.252864321608041e-05,
-      "loss": 5.5414,
-      "step": 17889
-    },
-    {
-      "epoch": 9.329856584093871,
-      "grad_norm": 1.4365301132202148,
-      "learning_rate": 8.252763819095479e-05,
-      "loss": 5.5871,
-      "step": 17890
-    },
-    {
-      "epoch": 9.33037809647979,
-      "grad_norm": 1.3647773265838623,
-      "learning_rate": 8.252663316582915e-05,
-      "loss": 5.9381,
-      "step": 17891
-    },
-    {
-      "epoch": 9.33089960886571,
-      "grad_norm": 1.3923392295837402,
-      "learning_rate": 8.252562814070353e-05,
-      "loss": 5.7685,
-      "step": 17892
-    },
-    {
-      "epoch": 9.33142112125163,
-      "grad_norm": 1.3768190145492554,
-      "learning_rate": 8.252462311557789e-05,
-      "loss": 5.1359,
-      "step": 17893
-    },
-    {
-      "epoch": 9.33194263363755,
-      "grad_norm": 1.530752182006836,
-      "learning_rate": 8.252361809045227e-05,
-      "loss": 5.4826,
-      "step": 17894
-    },
-    {
-      "epoch": 9.332464146023469,
-      "grad_norm": 1.388155221939087,
-      "learning_rate": 8.252261306532663e-05,
-      "loss": 5.9109,
-      "step": 17895
-    },
-    {
-      "epoch": 9.332985658409386,
-      "grad_norm": 1.4167159795761108,
-      "learning_rate": 8.2521608040201e-05,
-      "loss": 5.3899,
-      "step": 17896
-    },
-    {
-      "epoch": 9.333507170795306,
-      "grad_norm": 1.4357798099517822,
-      "learning_rate": 8.252060301507538e-05,
-      "loss": 5.3507,
-      "step": 17897
-    },
-    {
-      "epoch": 9.334028683181225,
-      "grad_norm": 1.3018168210983276,
-      "learning_rate": 8.251959798994975e-05,
-      "loss": 5.9912,
-      "step": 17898
-    },
-    {
-      "epoch": 9.334550195567145,
-      "grad_norm": 1.4591072797775269,
-      "learning_rate": 8.251859296482413e-05,
-      "loss": 5.5688,
-      "step": 17899
-    },
-    {
-      "epoch": 9.335071707953064,
-      "grad_norm": 1.5286787748336792,
-      "learning_rate": 8.25175879396985e-05,
-      "loss": 5.5641,
-      "step": 17900
-    },
-    {
-      "epoch": 9.335593220338984,
-      "grad_norm": 1.4416577816009521,
-      "learning_rate": 8.251658291457287e-05,
-      "loss": 5.2482,
-      "step": 17901
-    },
-    {
-      "epoch": 9.336114732724901,
-      "grad_norm": 1.3891352415084839,
-      "learning_rate": 8.251557788944724e-05,
-      "loss": 5.7154,
-      "step": 17902
-    },
-    {
-      "epoch": 9.336636245110821,
-      "grad_norm": 1.4562835693359375,
-      "learning_rate": 8.251457286432162e-05,
-      "loss": 5.1169,
-      "step": 17903
-    },
-    {
-      "epoch": 9.33715775749674,
-      "grad_norm": 1.4311975240707397,
-      "learning_rate": 8.251356783919598e-05,
-      "loss": 5.3427,
-      "step": 17904
-    },
-    {
-      "epoch": 9.33767926988266,
-      "grad_norm": 1.2712786197662354,
-      "learning_rate": 8.251256281407036e-05,
-      "loss": 5.716,
-      "step": 17905
-    },
-    {
-      "epoch": 9.33820078226858,
-      "grad_norm": 1.3678568601608276,
-      "learning_rate": 8.251155778894472e-05,
-      "loss": 5.8946,
-      "step": 17906
-    },
-    {
-      "epoch": 9.338722294654499,
-      "grad_norm": 1.40695321559906,
-      "learning_rate": 8.25105527638191e-05,
-      "loss": 5.5554,
-      "step": 17907
-    },
-    {
-      "epoch": 9.339243807040416,
-      "grad_norm": 1.5746636390686035,
-      "learning_rate": 8.250954773869348e-05,
-      "loss": 5.1641,
-      "step": 17908
-    },
-    {
-      "epoch": 9.339765319426336,
-      "grad_norm": 1.4663691520690918,
-      "learning_rate": 8.250854271356784e-05,
-      "loss": 5.1414,
-      "step": 17909
-    },
-    {
-      "epoch": 9.340286831812255,
-      "grad_norm": 1.4852806329727173,
-      "learning_rate": 8.250753768844222e-05,
-      "loss": 5.6245,
-      "step": 17910
-    },
-    {
-      "epoch": 9.340808344198175,
-      "grad_norm": 1.5104782581329346,
-      "learning_rate": 8.250653266331658e-05,
-      "loss": 5.3716,
-      "step": 17911
-    },
-    {
-      "epoch": 9.341329856584094,
-      "grad_norm": 1.5047508478164673,
-      "learning_rate": 8.250552763819096e-05,
-      "loss": 5.7315,
-      "step": 17912
-    },
-    {
-      "epoch": 9.341851368970014,
-      "grad_norm": 1.4777884483337402,
-      "learning_rate": 8.250452261306533e-05,
-      "loss": 5.4365,
-      "step": 17913
-    },
-    {
-      "epoch": 9.342372881355931,
-      "grad_norm": 1.5596725940704346,
-      "learning_rate": 8.25035175879397e-05,
-      "loss": 4.9822,
-      "step": 17914
-    },
-    {
-      "epoch": 9.342894393741851,
-      "grad_norm": 1.4197951555252075,
-      "learning_rate": 8.250251256281407e-05,
-      "loss": 5.7182,
-      "step": 17915
-    },
-    {
-      "epoch": 9.34341590612777,
-      "grad_norm": 1.5719045400619507,
-      "learning_rate": 8.250150753768845e-05,
-      "loss": 5.0499,
-      "step": 17916
-    },
-    {
-      "epoch": 9.34393741851369,
-      "grad_norm": 1.4012089967727661,
-      "learning_rate": 8.250050251256281e-05,
-      "loss": 5.2021,
-      "step": 17917
-    },
-    {
-      "epoch": 9.34445893089961,
-      "grad_norm": 1.6200636625289917,
-      "learning_rate": 8.249949748743719e-05,
-      "loss": 4.6423,
-      "step": 17918
-    },
-    {
-      "epoch": 9.344980443285529,
-      "grad_norm": 1.5283403396606445,
-      "learning_rate": 8.249849246231157e-05,
-      "loss": 5.2189,
-      "step": 17919
-    },
-    {
-      "epoch": 9.345501955671446,
-      "grad_norm": 1.517523169517517,
-      "learning_rate": 8.249748743718594e-05,
-      "loss": 4.9356,
-      "step": 17920
-    },
-    {
-      "epoch": 9.346023468057366,
-      "grad_norm": 1.5763232707977295,
-      "learning_rate": 8.249648241206031e-05,
-      "loss": 4.7929,
-      "step": 17921
-    },
-    {
-      "epoch": 9.346544980443285,
-      "grad_norm": 1.542974829673767,
-      "learning_rate": 8.249547738693467e-05,
-      "loss": 5.6636,
-      "step": 17922
-    },
-    {
-      "epoch": 9.347066492829205,
-      "grad_norm": 1.4555479288101196,
-      "learning_rate": 8.249447236180905e-05,
-      "loss": 5.3015,
-      "step": 17923
-    },
-    {
-      "epoch": 9.347588005215124,
-      "grad_norm": 1.4468481540679932,
-      "learning_rate": 8.249346733668341e-05,
-      "loss": 5.2464,
-      "step": 17924
-    },
-    {
-      "epoch": 9.348109517601044,
-      "grad_norm": 1.4097073078155518,
-      "learning_rate": 8.249246231155779e-05,
-      "loss": 5.0684,
-      "step": 17925
-    },
-    {
-      "epoch": 9.348631029986961,
-      "grad_norm": 1.3771286010742188,
-      "learning_rate": 8.249145728643216e-05,
-      "loss": 5.4908,
-      "step": 17926
-    },
-    {
-      "epoch": 9.349152542372881,
-      "grad_norm": 1.3886290788650513,
-      "learning_rate": 8.249045226130653e-05,
-      "loss": 5.5659,
-      "step": 17927
-    },
-    {
-      "epoch": 9.3496740547588,
-      "grad_norm": 1.3563481569290161,
-      "learning_rate": 8.24894472361809e-05,
-      "loss": 5.9208,
-      "step": 17928
-    },
-    {
-      "epoch": 9.35019556714472,
-      "grad_norm": 1.4084186553955078,
-      "learning_rate": 8.248844221105528e-05,
-      "loss": 5.2676,
-      "step": 17929
-    },
-    {
-      "epoch": 9.35071707953064,
-      "grad_norm": 1.3822731971740723,
-      "learning_rate": 8.248743718592965e-05,
-      "loss": 5.6401,
-      "step": 17930
-    },
-    {
-      "epoch": 9.351238591916559,
-      "grad_norm": 1.3452178239822388,
-      "learning_rate": 8.248643216080403e-05,
-      "loss": 5.644,
-      "step": 17931
-    },
-    {
-      "epoch": 9.351760104302477,
-      "grad_norm": 1.5102779865264893,
-      "learning_rate": 8.24854271356784e-05,
-      "loss": 5.435,
-      "step": 17932
-    },
-    {
-      "epoch": 9.352281616688396,
-      "grad_norm": 1.5017430782318115,
-      "learning_rate": 8.248442211055277e-05,
-      "loss": 4.8151,
-      "step": 17933
-    },
-    {
-      "epoch": 9.352803129074315,
-      "grad_norm": 1.4006116390228271,
-      "learning_rate": 8.248341708542714e-05,
-      "loss": 5.3501,
-      "step": 17934
-    },
-    {
-      "epoch": 9.353324641460235,
-      "grad_norm": 1.424072027206421,
-      "learning_rate": 8.24824120603015e-05,
-      "loss": 5.4814,
-      "step": 17935
-    },
-    {
-      "epoch": 9.353846153846154,
-      "grad_norm": 1.2976264953613281,
-      "learning_rate": 8.248140703517588e-05,
-      "loss": 5.8588,
-      "step": 17936
-    },
-    {
-      "epoch": 9.354367666232074,
-      "grad_norm": 1.42351233959198,
-      "learning_rate": 8.248040201005025e-05,
-      "loss": 5.2779,
-      "step": 17937
-    },
-    {
-      "epoch": 9.354889178617992,
-      "grad_norm": 1.3425321578979492,
-      "learning_rate": 8.247939698492462e-05,
-      "loss": 5.8503,
-      "step": 17938
-    },
-    {
-      "epoch": 9.355410691003911,
-      "grad_norm": 1.427116870880127,
-      "learning_rate": 8.2478391959799e-05,
-      "loss": 5.6053,
-      "step": 17939
-    },
-    {
-      "epoch": 9.35593220338983,
-      "grad_norm": 1.4962180852890015,
-      "learning_rate": 8.247738693467338e-05,
-      "loss": 5.5289,
-      "step": 17940
-    },
-    {
-      "epoch": 9.35645371577575,
-      "grad_norm": 1.5405765771865845,
-      "learning_rate": 8.247638190954774e-05,
-      "loss": 5.1594,
-      "step": 17941
-    },
-    {
-      "epoch": 9.35697522816167,
-      "grad_norm": 1.4033749103546143,
-      "learning_rate": 8.247537688442212e-05,
-      "loss": 4.8032,
-      "step": 17942
-    },
-    {
-      "epoch": 9.357496740547589,
-      "grad_norm": 1.322434902191162,
-      "learning_rate": 8.247437185929649e-05,
-      "loss": 5.8958,
-      "step": 17943
-    },
-    {
-      "epoch": 9.358018252933507,
-      "grad_norm": 1.5230109691619873,
-      "learning_rate": 8.247336683417086e-05,
-      "loss": 5.624,
-      "step": 17944
-    },
-    {
-      "epoch": 9.358539765319426,
-      "grad_norm": 1.7094820737838745,
-      "learning_rate": 8.247236180904523e-05,
-      "loss": 4.8566,
-      "step": 17945
-    },
-    {
-      "epoch": 9.359061277705345,
-      "grad_norm": 1.3719080686569214,
-      "learning_rate": 8.24713567839196e-05,
-      "loss": 5.4225,
-      "step": 17946
-    },
-    {
-      "epoch": 9.359582790091265,
-      "grad_norm": 1.4816004037857056,
-      "learning_rate": 8.247035175879397e-05,
-      "loss": 5.4571,
-      "step": 17947
-    },
-    {
-      "epoch": 9.360104302477184,
-      "grad_norm": 1.7452983856201172,
-      "learning_rate": 8.246934673366835e-05,
-      "loss": 5.2402,
-      "step": 17948
-    },
-    {
-      "epoch": 9.360625814863104,
-      "grad_norm": 2.221337080001831,
-      "learning_rate": 8.246834170854271e-05,
-      "loss": 5.0285,
-      "step": 17949
-    },
-    {
-      "epoch": 9.361147327249022,
-      "grad_norm": 1.3470616340637207,
-      "learning_rate": 8.246733668341709e-05,
-      "loss": 5.8865,
-      "step": 17950
-    },
-    {
-      "epoch": 9.361668839634941,
-      "grad_norm": 1.5290498733520508,
-      "learning_rate": 8.246633165829147e-05,
-      "loss": 5.5062,
-      "step": 17951
-    },
-    {
-      "epoch": 9.36219035202086,
-      "grad_norm": 1.3727655410766602,
-      "learning_rate": 8.246532663316583e-05,
-      "loss": 5.5302,
-      "step": 17952
-    },
-    {
-      "epoch": 9.36271186440678,
-      "grad_norm": 1.3910139799118042,
-      "learning_rate": 8.246432160804021e-05,
-      "loss": 5.4259,
-      "step": 17953
-    },
-    {
-      "epoch": 9.3632333767927,
-      "grad_norm": 1.377794861793518,
-      "learning_rate": 8.246331658291457e-05,
-      "loss": 5.5224,
-      "step": 17954
-    },
-    {
-      "epoch": 9.363754889178619,
-      "grad_norm": 1.4284231662750244,
-      "learning_rate": 8.246231155778895e-05,
-      "loss": 5.0708,
-      "step": 17955
-    },
-    {
-      "epoch": 9.364276401564537,
-      "grad_norm": 1.4817920923233032,
-      "learning_rate": 8.246130653266332e-05,
-      "loss": 5.4352,
-      "step": 17956
-    },
-    {
-      "epoch": 9.364797913950456,
-      "grad_norm": 1.4663662910461426,
-      "learning_rate": 8.24603015075377e-05,
-      "loss": 5.4967,
-      "step": 17957
-    },
-    {
-      "epoch": 9.365319426336375,
-      "grad_norm": 1.4427212476730347,
-      "learning_rate": 8.245929648241206e-05,
-      "loss": 5.6383,
-      "step": 17958
-    },
-    {
-      "epoch": 9.365840938722295,
-      "grad_norm": 1.5546072721481323,
-      "learning_rate": 8.245829145728644e-05,
-      "loss": 5.5745,
-      "step": 17959
-    },
-    {
-      "epoch": 9.366362451108214,
-      "grad_norm": 1.3749091625213623,
-      "learning_rate": 8.245728643216081e-05,
-      "loss": 5.6164,
-      "step": 17960
-    },
-    {
-      "epoch": 9.366883963494132,
-      "grad_norm": 1.3923178911209106,
-      "learning_rate": 8.245628140703519e-05,
-      "loss": 5.5829,
-      "step": 17961
-    },
-    {
-      "epoch": 9.367405475880052,
-      "grad_norm": 1.3997128009796143,
-      "learning_rate": 8.245527638190956e-05,
-      "loss": 5.9493,
-      "step": 17962
-    },
-    {
-      "epoch": 9.367926988265971,
-      "grad_norm": 1.4413673877716064,
-      "learning_rate": 8.245427135678392e-05,
-      "loss": 5.5563,
-      "step": 17963
-    },
-    {
-      "epoch": 9.36844850065189,
-      "grad_norm": 1.4447077512741089,
-      "learning_rate": 8.24532663316583e-05,
-      "loss": 5.7543,
-      "step": 17964
-    },
-    {
-      "epoch": 9.36897001303781,
-      "grad_norm": 1.4972916841506958,
-      "learning_rate": 8.245226130653266e-05,
-      "loss": 5.3976,
-      "step": 17965
-    },
-    {
-      "epoch": 9.36949152542373,
-      "grad_norm": 1.383237361907959,
-      "learning_rate": 8.245125628140704e-05,
-      "loss": 5.7381,
-      "step": 17966
-    },
-    {
-      "epoch": 9.370013037809649,
-      "grad_norm": 1.5408557653427124,
-      "learning_rate": 8.24502512562814e-05,
-      "loss": 5.6729,
-      "step": 17967
-    },
-    {
-      "epoch": 9.370534550195567,
-      "grad_norm": 1.3171483278274536,
-      "learning_rate": 8.244924623115578e-05,
-      "loss": 5.8904,
-      "step": 17968
-    },
-    {
-      "epoch": 9.371056062581486,
-      "grad_norm": 1.3889222145080566,
-      "learning_rate": 8.244824120603015e-05,
-      "loss": 5.5424,
-      "step": 17969
-    },
-    {
-      "epoch": 9.371577574967406,
-      "grad_norm": 1.5315359830856323,
-      "learning_rate": 8.244723618090452e-05,
-      "loss": 5.5191,
-      "step": 17970
-    },
-    {
-      "epoch": 9.372099087353325,
-      "grad_norm": 1.6181244850158691,
-      "learning_rate": 8.24462311557789e-05,
-      "loss": 5.5743,
-      "step": 17971
-    },
-    {
-      "epoch": 9.372620599739244,
-      "grad_norm": 1.56166410446167,
-      "learning_rate": 8.244522613065328e-05,
-      "loss": 4.9858,
-      "step": 17972
-    },
-    {
-      "epoch": 9.373142112125162,
-      "grad_norm": 1.3547282218933105,
-      "learning_rate": 8.244422110552764e-05,
-      "loss": 5.8043,
-      "step": 17973
-    },
-    {
-      "epoch": 9.373663624511082,
-      "grad_norm": 1.454676628112793,
-      "learning_rate": 8.244321608040202e-05,
-      "loss": 5.4957,
-      "step": 17974
-    },
-    {
-      "epoch": 9.374185136897001,
-      "grad_norm": 1.628665566444397,
-      "learning_rate": 8.244221105527639e-05,
-      "loss": 5.1826,
-      "step": 17975
-    },
-    {
-      "epoch": 9.37470664928292,
-      "grad_norm": 1.4482622146606445,
-      "learning_rate": 8.244120603015075e-05,
-      "loss": 5.906,
-      "step": 17976
-    },
-    {
-      "epoch": 9.37522816166884,
-      "grad_norm": 1.3745375871658325,
-      "learning_rate": 8.244020100502513e-05,
-      "loss": 5.5776,
-      "step": 17977
-    },
-    {
-      "epoch": 9.37574967405476,
-      "grad_norm": 1.471303105354309,
-      "learning_rate": 8.243919597989949e-05,
-      "loss": 5.7686,
-      "step": 17978
-    },
-    {
-      "epoch": 9.376271186440677,
-      "grad_norm": 1.3374967575073242,
-      "learning_rate": 8.243819095477387e-05,
-      "loss": 5.665,
-      "step": 17979
-    },
-    {
-      "epoch": 9.376792698826597,
-      "grad_norm": 1.32734215259552,
-      "learning_rate": 8.243718592964825e-05,
-      "loss": 5.81,
-      "step": 17980
-    },
-    {
-      "epoch": 9.377314211212516,
-      "grad_norm": 1.4587576389312744,
-      "learning_rate": 8.243618090452263e-05,
-      "loss": 5.5801,
-      "step": 17981
-    },
-    {
-      "epoch": 9.377835723598436,
-      "grad_norm": 1.3039119243621826,
-      "learning_rate": 8.243517587939699e-05,
-      "loss": 5.9042,
-      "step": 17982
-    },
-    {
-      "epoch": 9.378357235984355,
-      "grad_norm": 1.4757232666015625,
-      "learning_rate": 8.243417085427137e-05,
-      "loss": 5.1175,
-      "step": 17983
-    },
-    {
-      "epoch": 9.378878748370274,
-      "grad_norm": 1.684645175933838,
-      "learning_rate": 8.243316582914573e-05,
-      "loss": 5.6087,
-      "step": 17984
-    },
-    {
-      "epoch": 9.379400260756192,
-      "grad_norm": 1.4444822072982788,
-      "learning_rate": 8.243216080402011e-05,
-      "loss": 5.3423,
-      "step": 17985
-    },
-    {
-      "epoch": 9.379921773142112,
-      "grad_norm": 1.4454703330993652,
-      "learning_rate": 8.243115577889447e-05,
-      "loss": 5.353,
-      "step": 17986
-    },
-    {
-      "epoch": 9.380443285528031,
-      "grad_norm": 1.5121877193450928,
-      "learning_rate": 8.243015075376885e-05,
-      "loss": 5.4985,
-      "step": 17987
-    },
-    {
-      "epoch": 9.38096479791395,
-      "grad_norm": 1.5544407367706299,
-      "learning_rate": 8.242914572864322e-05,
-      "loss": 5.3647,
-      "step": 17988
-    },
-    {
-      "epoch": 9.38148631029987,
-      "grad_norm": 1.5461288690567017,
-      "learning_rate": 8.242814070351758e-05,
-      "loss": 5.2543,
-      "step": 17989
-    },
-    {
-      "epoch": 9.38200782268579,
-      "grad_norm": 1.6166555881500244,
-      "learning_rate": 8.242713567839196e-05,
-      "loss": 5.6636,
-      "step": 17990
-    },
-    {
-      "epoch": 9.382529335071707,
-      "grad_norm": 1.5054370164871216,
-      "learning_rate": 8.242613065326634e-05,
-      "loss": 5.5862,
-      "step": 17991
-    },
-    {
-      "epoch": 9.383050847457627,
-      "grad_norm": 1.4394770860671997,
-      "learning_rate": 8.242512562814071e-05,
-      "loss": 5.2118,
-      "step": 17992
-    },
-    {
-      "epoch": 9.383572359843546,
-      "grad_norm": 1.3795188665390015,
-      "learning_rate": 8.242412060301508e-05,
-      "loss": 5.3331,
-      "step": 17993
-    },
-    {
-      "epoch": 9.384093872229466,
-      "grad_norm": 1.5623884201049805,
-      "learning_rate": 8.242311557788946e-05,
-      "loss": 5.2352,
-      "step": 17994
-    },
-    {
-      "epoch": 9.384615384615385,
-      "grad_norm": 1.4661865234375,
-      "learning_rate": 8.242211055276382e-05,
-      "loss": 5.8629,
-      "step": 17995
-    },
-    {
-      "epoch": 9.385136897001304,
-      "grad_norm": 1.5410175323486328,
-      "learning_rate": 8.24211055276382e-05,
-      "loss": 5.7062,
-      "step": 17996
-    },
-    {
-      "epoch": 9.385658409387222,
-      "grad_norm": 1.4762966632843018,
-      "learning_rate": 8.242010050251256e-05,
-      "loss": 5.5354,
-      "step": 17997
-    },
-    {
-      "epoch": 9.386179921773142,
-      "grad_norm": 1.3615158796310425,
-      "learning_rate": 8.241909547738694e-05,
-      "loss": 5.0909,
-      "step": 17998
-    },
-    {
-      "epoch": 9.386701434159061,
-      "grad_norm": 1.6491873264312744,
-      "learning_rate": 8.24180904522613e-05,
-      "loss": 5.3157,
-      "step": 17999
-    },
-    {
-      "epoch": 9.38722294654498,
-      "grad_norm": 1.5471603870391846,
-      "learning_rate": 8.241708542713568e-05,
-      "loss": 5.1254,
-      "step": 18000
-    },
-    {
-      "epoch": 9.3877444589309,
-      "grad_norm": 1.5019398927688599,
-      "learning_rate": 8.241608040201006e-05,
-      "loss": 5.7954,
-      "step": 18001
-    },
-    {
-      "epoch": 9.38826597131682,
-      "grad_norm": 1.4378749132156372,
-      "learning_rate": 8.241507537688442e-05,
-      "loss": 5.6587,
-      "step": 18002
-    },
-    {
-      "epoch": 9.388787483702737,
-      "grad_norm": 1.4048480987548828,
-      "learning_rate": 8.24140703517588e-05,
-      "loss": 5.3177,
-      "step": 18003
-    },
-    {
-      "epoch": 9.389308996088657,
-      "grad_norm": 1.4480230808258057,
-      "learning_rate": 8.241306532663317e-05,
-      "loss": 5.4942,
-      "step": 18004
-    },
-    {
-      "epoch": 9.389830508474576,
-      "grad_norm": 1.3602466583251953,
-      "learning_rate": 8.241206030150754e-05,
-      "loss": 5.8556,
-      "step": 18005
-    },
-    {
-      "epoch": 9.390352020860496,
-      "grad_norm": 1.4130569696426392,
-      "learning_rate": 8.241105527638191e-05,
-      "loss": 5.1122,
-      "step": 18006
-    },
-    {
-      "epoch": 9.390873533246415,
-      "grad_norm": 1.3938939571380615,
-      "learning_rate": 8.241005025125629e-05,
-      "loss": 5.179,
-      "step": 18007
-    },
-    {
-      "epoch": 9.391395045632335,
-      "grad_norm": 1.584280252456665,
-      "learning_rate": 8.240904522613065e-05,
-      "loss": 5.1791,
-      "step": 18008
-    },
-    {
-      "epoch": 9.391916558018252,
-      "grad_norm": 1.3995248079299927,
-      "learning_rate": 8.240804020100503e-05,
-      "loss": 5.5704,
-      "step": 18009
-    },
-    {
-      "epoch": 9.392438070404172,
-      "grad_norm": 1.5258076190948486,
-      "learning_rate": 8.240703517587939e-05,
-      "loss": 5.2678,
-      "step": 18010
-    },
-    {
-      "epoch": 9.392959582790091,
-      "grad_norm": 1.4317625761032104,
-      "learning_rate": 8.240603015075377e-05,
-      "loss": 5.6957,
-      "step": 18011
-    },
-    {
-      "epoch": 9.39348109517601,
-      "grad_norm": 1.4740370512008667,
-      "learning_rate": 8.240502512562815e-05,
-      "loss": 5.564,
-      "step": 18012
-    },
-    {
-      "epoch": 9.39400260756193,
-      "grad_norm": 1.4794515371322632,
-      "learning_rate": 8.240402010050253e-05,
-      "loss": 5.6956,
-      "step": 18013
-    },
-    {
-      "epoch": 9.39452411994785,
-      "grad_norm": 1.445555329322815,
-      "learning_rate": 8.240301507537689e-05,
-      "loss": 5.3945,
-      "step": 18014
-    },
-    {
-      "epoch": 9.395045632333767,
-      "grad_norm": 1.5151803493499756,
-      "learning_rate": 8.240201005025126e-05,
-      "loss": 5.5269,
-      "step": 18015
-    },
-    {
-      "epoch": 9.395567144719687,
-      "grad_norm": 1.547906756401062,
-      "learning_rate": 8.240100502512563e-05,
-      "loss": 5.3196,
-      "step": 18016
-    },
-    {
-      "epoch": 9.396088657105606,
-      "grad_norm": 1.3441587686538696,
-      "learning_rate": 8.24e-05,
-      "loss": 5.1135,
-      "step": 18017
-    },
-    {
-      "epoch": 9.396610169491526,
-      "grad_norm": 1.4341620206832886,
-      "learning_rate": 8.239899497487437e-05,
-      "loss": 5.0826,
-      "step": 18018
-    },
-    {
-      "epoch": 9.397131681877445,
-      "grad_norm": 1.4498653411865234,
-      "learning_rate": 8.239798994974874e-05,
-      "loss": 5.1991,
-      "step": 18019
-    },
-    {
-      "epoch": 9.397653194263365,
-      "grad_norm": 1.4594868421554565,
-      "learning_rate": 8.239698492462312e-05,
-      "loss": 5.5379,
-      "step": 18020
-    },
-    {
-      "epoch": 9.398174706649282,
-      "grad_norm": 1.5081995725631714,
-      "learning_rate": 8.23959798994975e-05,
-      "loss": 5.3452,
-      "step": 18021
-    },
-    {
-      "epoch": 9.398696219035202,
-      "grad_norm": 1.4254034757614136,
-      "learning_rate": 8.239497487437187e-05,
-      "loss": 5.73,
-      "step": 18022
-    },
-    {
-      "epoch": 9.399217731421121,
-      "grad_norm": 1.512642502784729,
-      "learning_rate": 8.239396984924624e-05,
-      "loss": 5.008,
-      "step": 18023
-    },
-    {
-      "epoch": 9.39973924380704,
-      "grad_norm": 1.362318992614746,
-      "learning_rate": 8.239296482412061e-05,
-      "loss": 5.6289,
-      "step": 18024
-    },
-    {
-      "epoch": 9.40026075619296,
-      "grad_norm": 1.4266718626022339,
-      "learning_rate": 8.239195979899498e-05,
-      "loss": 5.6463,
-      "step": 18025
-    },
-    {
-      "epoch": 9.40078226857888,
-      "grad_norm": 1.5579407215118408,
-      "learning_rate": 8.239095477386936e-05,
-      "loss": 5.0737,
-      "step": 18026
-    },
-    {
-      "epoch": 9.401303780964797,
-      "grad_norm": 1.8256499767303467,
-      "learning_rate": 8.238994974874372e-05,
-      "loss": 5.0753,
-      "step": 18027
-    },
-    {
-      "epoch": 9.401825293350717,
-      "grad_norm": 1.4612624645233154,
-      "learning_rate": 8.238894472361809e-05,
-      "loss": 5.5984,
-      "step": 18028
-    },
-    {
-      "epoch": 9.402346805736636,
-      "grad_norm": 1.4165050983428955,
-      "learning_rate": 8.238793969849246e-05,
-      "loss": 5.4601,
-      "step": 18029
-    },
-    {
-      "epoch": 9.402868318122556,
-      "grad_norm": 1.3717306852340698,
-      "learning_rate": 8.238693467336683e-05,
-      "loss": 5.6591,
-      "step": 18030
-    },
-    {
-      "epoch": 9.403389830508475,
-      "grad_norm": 1.4065914154052734,
-      "learning_rate": 8.23859296482412e-05,
-      "loss": 5.6561,
-      "step": 18031
-    },
-    {
-      "epoch": 9.403911342894395,
-      "grad_norm": 1.488102674484253,
-      "learning_rate": 8.238492462311558e-05,
-      "loss": 5.4781,
-      "step": 18032
-    },
-    {
-      "epoch": 9.404432855280312,
-      "grad_norm": 1.8169970512390137,
-      "learning_rate": 8.238391959798996e-05,
-      "loss": 4.3866,
-      "step": 18033
-    },
-    {
-      "epoch": 9.404954367666232,
-      "grad_norm": 1.3515421152114868,
-      "learning_rate": 8.238291457286433e-05,
-      "loss": 5.9125,
-      "step": 18034
-    },
-    {
-      "epoch": 9.405475880052151,
-      "grad_norm": 1.518065094947815,
-      "learning_rate": 8.23819095477387e-05,
-      "loss": 5.7127,
-      "step": 18035
-    },
-    {
-      "epoch": 9.40599739243807,
-      "grad_norm": 1.429169774055481,
-      "learning_rate": 8.238090452261307e-05,
-      "loss": 5.5387,
-      "step": 18036
-    },
-    {
-      "epoch": 9.40651890482399,
-      "grad_norm": 1.4230960607528687,
-      "learning_rate": 8.237989949748745e-05,
-      "loss": 5.8775,
-      "step": 18037
-    },
-    {
-      "epoch": 9.40704041720991,
-      "grad_norm": 1.5465497970581055,
-      "learning_rate": 8.237889447236181e-05,
-      "loss": 5.559,
-      "step": 18038
-    },
-    {
-      "epoch": 9.407561929595827,
-      "grad_norm": 1.6945523023605347,
-      "learning_rate": 8.237788944723619e-05,
-      "loss": 4.756,
-      "step": 18039
-    },
-    {
-      "epoch": 9.408083441981747,
-      "grad_norm": 1.4151487350463867,
-      "learning_rate": 8.237688442211055e-05,
-      "loss": 5.817,
-      "step": 18040
-    },
-    {
-      "epoch": 9.408604954367666,
-      "grad_norm": 1.3039671182632446,
-      "learning_rate": 8.237587939698493e-05,
-      "loss": 5.8605,
-      "step": 18041
-    },
-    {
-      "epoch": 9.409126466753586,
-      "grad_norm": 1.3569629192352295,
-      "learning_rate": 8.237487437185931e-05,
-      "loss": 5.6501,
-      "step": 18042
-    },
-    {
-      "epoch": 9.409647979139505,
-      "grad_norm": 1.5257023572921753,
-      "learning_rate": 8.237386934673367e-05,
-      "loss": 5.4587,
-      "step": 18043
-    },
-    {
-      "epoch": 9.410169491525423,
-      "grad_norm": 1.5034717321395874,
-      "learning_rate": 8.237286432160805e-05,
-      "loss": 5.3345,
-      "step": 18044
-    },
-    {
-      "epoch": 9.410691003911342,
-      "grad_norm": 1.5466091632843018,
-      "learning_rate": 8.237185929648241e-05,
-      "loss": 4.8443,
-      "step": 18045
-    },
-    {
-      "epoch": 9.411212516297262,
-      "grad_norm": 1.4566874504089355,
-      "learning_rate": 8.237085427135679e-05,
-      "loss": 5.5678,
-      "step": 18046
-    },
-    {
-      "epoch": 9.411734028683181,
-      "grad_norm": 1.510926604270935,
-      "learning_rate": 8.236984924623116e-05,
-      "loss": 5.5622,
-      "step": 18047
-    },
-    {
-      "epoch": 9.4122555410691,
-      "grad_norm": 1.66961669921875,
-      "learning_rate": 8.236884422110553e-05,
-      "loss": 5.0882,
-      "step": 18048
-    },
-    {
-      "epoch": 9.41277705345502,
-      "grad_norm": 1.4918900728225708,
-      "learning_rate": 8.23678391959799e-05,
-      "loss": 5.3934,
-      "step": 18049
-    },
-    {
-      "epoch": 9.41329856584094,
-      "grad_norm": 1.328500747680664,
-      "learning_rate": 8.236683417085428e-05,
-      "loss": 5.118,
-      "step": 18050
-    },
-    {
-      "epoch": 9.413820078226857,
-      "grad_norm": 1.504177212715149,
-      "learning_rate": 8.236582914572864e-05,
-      "loss": 5.1081,
-      "step": 18051
-    },
-    {
-      "epoch": 9.414341590612777,
-      "grad_norm": 1.4242361783981323,
-      "learning_rate": 8.236482412060302e-05,
-      "loss": 5.5008,
-      "step": 18052
-    },
-    {
-      "epoch": 9.414863102998696,
-      "grad_norm": 1.5699717998504639,
-      "learning_rate": 8.23638190954774e-05,
-      "loss": 5.3011,
-      "step": 18053
-    },
-    {
-      "epoch": 9.415384615384616,
-      "grad_norm": 1.5572208166122437,
-      "learning_rate": 8.236281407035177e-05,
-      "loss": 5.1298,
-      "step": 18054
-    },
-    {
-      "epoch": 9.415906127770535,
-      "grad_norm": 1.5199990272521973,
-      "learning_rate": 8.236180904522614e-05,
-      "loss": 5.2991,
-      "step": 18055
-    },
-    {
-      "epoch": 9.416427640156453,
-      "grad_norm": 1.422109842300415,
-      "learning_rate": 8.23608040201005e-05,
-      "loss": 5.1622,
-      "step": 18056
-    },
-    {
-      "epoch": 9.416949152542372,
-      "grad_norm": 1.4713125228881836,
-      "learning_rate": 8.235979899497488e-05,
-      "loss": 5.1302,
-      "step": 18057
-    },
-    {
-      "epoch": 9.417470664928292,
-      "grad_norm": 1.607741355895996,
-      "learning_rate": 8.235879396984924e-05,
-      "loss": 5.0307,
-      "step": 18058
-    },
-    {
-      "epoch": 9.417992177314211,
-      "grad_norm": 1.513523817062378,
-      "learning_rate": 8.235778894472362e-05,
-      "loss": 5.3707,
-      "step": 18059
-    },
-    {
-      "epoch": 9.41851368970013,
-      "grad_norm": 1.473988652229309,
-      "learning_rate": 8.235678391959799e-05,
-      "loss": 5.5948,
-      "step": 18060
-    },
-    {
-      "epoch": 9.41903520208605,
-      "grad_norm": 1.4688174724578857,
-      "learning_rate": 8.235577889447236e-05,
-      "loss": 5.3526,
-      "step": 18061
-    },
-    {
-      "epoch": 9.419556714471968,
-      "grad_norm": 1.426521897315979,
-      "learning_rate": 8.235477386934674e-05,
-      "loss": 5.926,
-      "step": 18062
-    },
-    {
-      "epoch": 9.420078226857887,
-      "grad_norm": 1.4833139181137085,
-      "learning_rate": 8.235376884422112e-05,
-      "loss": 5.5656,
-      "step": 18063
-    },
-    {
-      "epoch": 9.420599739243807,
-      "grad_norm": 1.46029531955719,
-      "learning_rate": 8.235276381909548e-05,
-      "loss": 5.7143,
-      "step": 18064
-    },
-    {
-      "epoch": 9.421121251629726,
-      "grad_norm": 1.4743659496307373,
-      "learning_rate": 8.235175879396986e-05,
-      "loss": 5.2686,
-      "step": 18065
-    },
-    {
-      "epoch": 9.421642764015646,
-      "grad_norm": 1.349268913269043,
-      "learning_rate": 8.235075376884423e-05,
-      "loss": 5.7953,
-      "step": 18066
-    },
-    {
-      "epoch": 9.422164276401565,
-      "grad_norm": 1.3131568431854248,
-      "learning_rate": 8.23497487437186e-05,
-      "loss": 5.6447,
-      "step": 18067
-    },
-    {
-      "epoch": 9.422685788787483,
-      "grad_norm": 1.400826334953308,
-      "learning_rate": 8.234874371859297e-05,
-      "loss": 5.8676,
-      "step": 18068
-    },
-    {
-      "epoch": 9.423207301173402,
-      "grad_norm": 1.463335394859314,
-      "learning_rate": 8.234773869346733e-05,
-      "loss": 5.2802,
-      "step": 18069
-    },
-    {
-      "epoch": 9.423728813559322,
-      "grad_norm": 1.4013656377792358,
-      "learning_rate": 8.234673366834171e-05,
-      "loss": 5.7373,
-      "step": 18070
-    },
-    {
-      "epoch": 9.424250325945241,
-      "grad_norm": 1.4669426679611206,
-      "learning_rate": 8.234572864321607e-05,
-      "loss": 5.3381,
-      "step": 18071
-    },
-    {
-      "epoch": 9.42477183833116,
-      "grad_norm": 1.4457329511642456,
-      "learning_rate": 8.234472361809045e-05,
-      "loss": 5.6334,
-      "step": 18072
-    },
-    {
-      "epoch": 9.42529335071708,
-      "grad_norm": 1.395687222480774,
-      "learning_rate": 8.234371859296483e-05,
-      "loss": 5.3716,
-      "step": 18073
-    },
-    {
-      "epoch": 9.425814863102998,
-      "grad_norm": 1.4185758829116821,
-      "learning_rate": 8.234271356783921e-05,
-      "loss": 5.4904,
-      "step": 18074
-    },
-    {
-      "epoch": 9.426336375488917,
-      "grad_norm": Infinity,
-      "learning_rate": 8.234271356783921e-05,
-      "loss": 5.029,
-      "step": 18075
-    },
-    {
-      "epoch": 9.426857887874837,
-      "grad_norm": 1.4754424095153809,
-      "learning_rate": 8.234170854271357e-05,
-      "loss": 5.425,
-      "step": 18076
-    },
-    {
-      "epoch": 9.427379400260756,
-      "grad_norm": 1.492533564567566,
-      "learning_rate": 8.234070351758795e-05,
-      "loss": 5.3713,
-      "step": 18077
-    },
-    {
-      "epoch": 9.427900912646676,
-      "grad_norm": 1.4642823934555054,
-      "learning_rate": 8.233969849246231e-05,
-      "loss": 5.7868,
-      "step": 18078
-    },
-    {
-      "epoch": 9.428422425032595,
-      "grad_norm": 1.4120131731033325,
-      "learning_rate": 8.233869346733669e-05,
-      "loss": 5.6027,
-      "step": 18079
-    },
-    {
-      "epoch": 9.428943937418513,
-      "grad_norm": 2.0073161125183105,
-      "learning_rate": 8.233768844221106e-05,
-      "loss": 5.2987,
-      "step": 18080
-    },
-    {
-      "epoch": 9.429465449804432,
-      "grad_norm": 1.4899280071258545,
-      "learning_rate": 8.233668341708543e-05,
-      "loss": 5.5589,
-      "step": 18081
-    },
-    {
-      "epoch": 9.429986962190352,
-      "grad_norm": 1.5576510429382324,
-      "learning_rate": 8.23356783919598e-05,
-      "loss": 5.7059,
-      "step": 18082
-    },
-    {
-      "epoch": 9.430508474576271,
-      "grad_norm": 1.4661530256271362,
-      "learning_rate": 8.233467336683416e-05,
-      "loss": 5.511,
-      "step": 18083
-    },
-    {
-      "epoch": 9.43102998696219,
-      "grad_norm": 1.5719908475875854,
-      "learning_rate": 8.233366834170854e-05,
-      "loss": 5.4821,
-      "step": 18084
-    },
-    {
-      "epoch": 9.43155149934811,
-      "grad_norm": 1.341475486755371,
-      "learning_rate": 8.233266331658292e-05,
-      "loss": 6.0162,
-      "step": 18085
-    },
-    {
-      "epoch": 9.432073011734028,
-      "grad_norm": 1.5909852981567383,
-      "learning_rate": 8.23316582914573e-05,
-      "loss": 5.4696,
-      "step": 18086
-    },
-    {
-      "epoch": 9.432594524119947,
-      "grad_norm": 1.4381407499313354,
-      "learning_rate": 8.233065326633166e-05,
-      "loss": 5.5216,
-      "step": 18087
-    },
-    {
-      "epoch": 9.433116036505867,
-      "grad_norm": 1.468895673751831,
-      "learning_rate": 8.232964824120604e-05,
-      "loss": 5.5397,
-      "step": 18088
-    },
-    {
-      "epoch": 9.433637548891786,
-      "grad_norm": 1.5393102169036865,
-      "learning_rate": 8.23286432160804e-05,
-      "loss": 5.2813,
-      "step": 18089
-    },
-    {
-      "epoch": 9.434159061277706,
-      "grad_norm": 1.4526560306549072,
-      "learning_rate": 8.232763819095478e-05,
-      "loss": 5.3149,
-      "step": 18090
-    },
-    {
-      "epoch": 9.434680573663625,
-      "grad_norm": 1.6928762197494507,
-      "learning_rate": 8.232663316582914e-05,
-      "loss": 5.3869,
-      "step": 18091
-    },
-    {
-      "epoch": 9.435202086049543,
-      "grad_norm": 1.4105455875396729,
-      "learning_rate": 8.232562814070352e-05,
-      "loss": 5.1559,
-      "step": 18092
-    },
-    {
-      "epoch": 9.435723598435462,
-      "grad_norm": 1.6322052478790283,
-      "learning_rate": 8.232462311557789e-05,
-      "loss": 5.3534,
-      "step": 18093
-    },
-    {
-      "epoch": 9.436245110821382,
-      "grad_norm": 1.3640223741531372,
-      "learning_rate": 8.232361809045226e-05,
-      "loss": 5.3899,
-      "step": 18094
-    },
-    {
-      "epoch": 9.436766623207301,
-      "grad_norm": 1.4348257780075073,
-      "learning_rate": 8.232261306532664e-05,
-      "loss": 5.2832,
-      "step": 18095
-    },
-    {
-      "epoch": 9.43728813559322,
-      "grad_norm": 1.4504704475402832,
-      "learning_rate": 8.232160804020101e-05,
-      "loss": 5.5772,
-      "step": 18096
-    },
-    {
-      "epoch": 9.43780964797914,
-      "grad_norm": 1.4425619840621948,
-      "learning_rate": 8.232060301507538e-05,
-      "loss": 5.1431,
-      "step": 18097
-    },
-    {
-      "epoch": 9.438331160365058,
-      "grad_norm": 1.4713711738586426,
-      "learning_rate": 8.231959798994975e-05,
-      "loss": 4.998,
-      "step": 18098
-    },
-    {
-      "epoch": 9.438852672750977,
-      "grad_norm": 1.4756457805633545,
-      "learning_rate": 8.231859296482413e-05,
-      "loss": 5.8032,
-      "step": 18099
-    },
-    {
-      "epoch": 9.439374185136897,
-      "grad_norm": 1.6445220708847046,
-      "learning_rate": 8.231758793969849e-05,
-      "loss": 5.4773,
-      "step": 18100
-    },
-    {
-      "epoch": 9.439895697522816,
-      "grad_norm": 1.5312669277191162,
-      "learning_rate": 8.231658291457287e-05,
-      "loss": 4.8865,
-      "step": 18101
-    },
-    {
-      "epoch": 9.440417209908736,
-      "grad_norm": 1.4233381748199463,
-      "learning_rate": 8.231557788944723e-05,
-      "loss": 5.7649,
-      "step": 18102
-    },
-    {
-      "epoch": 9.440938722294655,
-      "grad_norm": 1.312266230583191,
-      "learning_rate": 8.231457286432161e-05,
-      "loss": 5.7553,
-      "step": 18103
-    },
-    {
-      "epoch": 9.441460234680573,
-      "grad_norm": 1.4358023405075073,
-      "learning_rate": 8.231356783919598e-05,
-      "loss": 5.4793,
-      "step": 18104
-    },
-    {
-      "epoch": 9.441981747066492,
-      "grad_norm": 1.5911058187484741,
-      "learning_rate": 8.231256281407035e-05,
-      "loss": 5.3449,
-      "step": 18105
-    },
-    {
-      "epoch": 9.442503259452412,
-      "grad_norm": 1.562393307685852,
-      "learning_rate": 8.231155778894473e-05,
-      "loss": 5.6163,
-      "step": 18106
-    },
-    {
-      "epoch": 9.443024771838331,
-      "grad_norm": 1.4893900156021118,
-      "learning_rate": 8.231055276381911e-05,
-      "loss": 5.6934,
-      "step": 18107
-    },
-    {
-      "epoch": 9.44354628422425,
-      "grad_norm": 1.5460155010223389,
-      "learning_rate": 8.230954773869347e-05,
-      "loss": 5.2206,
-      "step": 18108
-    },
-    {
-      "epoch": 9.44406779661017,
-      "grad_norm": 1.542029857635498,
-      "learning_rate": 8.230854271356784e-05,
-      "loss": 5.3362,
-      "step": 18109
-    },
-    {
-      "epoch": 9.444589308996088,
-      "grad_norm": 1.4582087993621826,
-      "learning_rate": 8.230753768844222e-05,
-      "loss": 5.6703,
-      "step": 18110
-    },
-    {
-      "epoch": 9.445110821382007,
-      "grad_norm": 1.481508731842041,
-      "learning_rate": 8.230653266331658e-05,
-      "loss": 5.005,
-      "step": 18111
-    },
-    {
-      "epoch": 9.445632333767927,
-      "grad_norm": 1.489197850227356,
-      "learning_rate": 8.230552763819096e-05,
-      "loss": 5.6683,
-      "step": 18112
-    },
-    {
-      "epoch": 9.446153846153846,
-      "grad_norm": 1.4981640577316284,
-      "learning_rate": 8.230452261306532e-05,
-      "loss": 5.6698,
-      "step": 18113
-    },
-    {
-      "epoch": 9.446675358539766,
-      "grad_norm": 1.465809941291809,
-      "learning_rate": 8.23035175879397e-05,
-      "loss": 5.5604,
-      "step": 18114
-    },
-    {
-      "epoch": 9.447196870925685,
-      "grad_norm": 1.3653984069824219,
-      "learning_rate": 8.230251256281408e-05,
-      "loss": 5.6712,
-      "step": 18115
-    },
-    {
-      "epoch": 9.447718383311603,
-      "grad_norm": 1.3996353149414062,
-      "learning_rate": 8.230150753768846e-05,
-      "loss": 5.6924,
-      "step": 18116
-    },
-    {
-      "epoch": 9.448239895697522,
-      "grad_norm": 1.3451921939849854,
-      "learning_rate": 8.230050251256282e-05,
-      "loss": 5.6786,
-      "step": 18117
-    },
-    {
-      "epoch": 9.448761408083442,
-      "grad_norm": 1.451648235321045,
-      "learning_rate": 8.22994974874372e-05,
-      "loss": 5.4813,
-      "step": 18118
-    },
-    {
-      "epoch": 9.449282920469361,
-      "grad_norm": 1.7541193962097168,
-      "learning_rate": 8.229849246231156e-05,
-      "loss": 5.4532,
-      "step": 18119
-    },
-    {
-      "epoch": 9.44980443285528,
-      "grad_norm": 1.4935994148254395,
-      "learning_rate": 8.229748743718594e-05,
-      "loss": 5.3215,
-      "step": 18120
-    },
-    {
-      "epoch": 9.4503259452412,
-      "grad_norm": 1.5612107515335083,
-      "learning_rate": 8.22964824120603e-05,
-      "loss": 5.7442,
-      "step": 18121
-    },
-    {
-      "epoch": 9.450847457627118,
-      "grad_norm": 1.2248151302337646,
-      "learning_rate": 8.229547738693467e-05,
-      "loss": 5.3356,
-      "step": 18122
-    },
-    {
-      "epoch": 9.451368970013037,
-      "grad_norm": 1.3826615810394287,
-      "learning_rate": 8.229447236180905e-05,
-      "loss": 5.704,
-      "step": 18123
-    },
-    {
-      "epoch": 9.451890482398957,
-      "grad_norm": 1.496376395225525,
-      "learning_rate": 8.229346733668341e-05,
-      "loss": 5.4235,
-      "step": 18124
-    },
-    {
-      "epoch": 9.452411994784876,
-      "grad_norm": 1.413495421409607,
-      "learning_rate": 8.229246231155779e-05,
-      "loss": 5.2761,
-      "step": 18125
-    },
-    {
-      "epoch": 9.452933507170796,
-      "grad_norm": 1.5687239170074463,
-      "learning_rate": 8.229145728643217e-05,
-      "loss": 5.7749,
-      "step": 18126
-    },
-    {
-      "epoch": 9.453455019556715,
-      "grad_norm": 1.41693913936615,
-      "learning_rate": 8.229045226130654e-05,
-      "loss": 5.6602,
-      "step": 18127
-    },
-    {
-      "epoch": 9.453976531942633,
-      "grad_norm": 1.4881340265274048,
-      "learning_rate": 8.228944723618091e-05,
-      "loss": 5.209,
-      "step": 18128
-    },
-    {
-      "epoch": 9.454498044328552,
-      "grad_norm": 1.4121490716934204,
-      "learning_rate": 8.228844221105529e-05,
-      "loss": 5.9324,
-      "step": 18129
-    },
-    {
-      "epoch": 9.455019556714472,
-      "grad_norm": 1.4613633155822754,
-      "learning_rate": 8.228743718592965e-05,
-      "loss": 5.3094,
-      "step": 18130
-    },
-    {
-      "epoch": 9.455541069100391,
-      "grad_norm": 1.3947190046310425,
-      "learning_rate": 8.228643216080403e-05,
-      "loss": 4.7986,
-      "step": 18131
-    },
-    {
-      "epoch": 9.45606258148631,
-      "grad_norm": 1.319873571395874,
-      "learning_rate": 8.228542713567839e-05,
-      "loss": 5.7077,
-      "step": 18132
-    },
-    {
-      "epoch": 9.45658409387223,
-      "grad_norm": 1.3327840566635132,
-      "learning_rate": 8.228442211055277e-05,
-      "loss": 5.8916,
-      "step": 18133
-    },
-    {
-      "epoch": 9.457105606258148,
-      "grad_norm": 1.391348958015442,
-      "learning_rate": 8.228341708542713e-05,
-      "loss": 5.8923,
-      "step": 18134
-    },
-    {
-      "epoch": 9.457627118644067,
-      "grad_norm": 1.5446748733520508,
-      "learning_rate": 8.228241206030151e-05,
-      "loss": 5.4669,
-      "step": 18135
-    },
-    {
-      "epoch": 9.458148631029987,
-      "grad_norm": 1.49965238571167,
-      "learning_rate": 8.228140703517589e-05,
-      "loss": 4.9423,
-      "step": 18136
-    },
-    {
-      "epoch": 9.458670143415906,
-      "grad_norm": 1.4364827871322632,
-      "learning_rate": 8.228040201005025e-05,
-      "loss": 5.3427,
-      "step": 18137
-    },
-    {
-      "epoch": 9.459191655801826,
-      "grad_norm": 1.3767116069793701,
-      "learning_rate": 8.227939698492463e-05,
-      "loss": 5.4819,
-      "step": 18138
-    },
-    {
-      "epoch": 9.459713168187744,
-      "grad_norm": 1.4229307174682617,
-      "learning_rate": 8.2278391959799e-05,
-      "loss": 5.3037,
-      "step": 18139
-    },
-    {
-      "epoch": 9.460234680573663,
-      "grad_norm": 1.3355556726455688,
-      "learning_rate": 8.227738693467337e-05,
-      "loss": 5.5886,
-      "step": 18140
-    },
-    {
-      "epoch": 9.460756192959582,
-      "grad_norm": 1.5224268436431885,
-      "learning_rate": 8.227638190954774e-05,
-      "loss": 5.3713,
-      "step": 18141
-    },
-    {
-      "epoch": 9.461277705345502,
-      "grad_norm": 1.420087218284607,
-      "learning_rate": 8.227537688442212e-05,
-      "loss": 5.5908,
-      "step": 18142
-    },
-    {
-      "epoch": 9.461799217731421,
-      "grad_norm": 1.5322381258010864,
-      "learning_rate": 8.227437185929648e-05,
-      "loss": 5.3671,
-      "step": 18143
-    },
-    {
-      "epoch": 9.46232073011734,
-      "grad_norm": 1.3624498844146729,
-      "learning_rate": 8.227336683417086e-05,
-      "loss": 5.8578,
-      "step": 18144
-    },
-    {
-      "epoch": 9.46284224250326,
-      "grad_norm": 1.4236645698547363,
-      "learning_rate": 8.227236180904522e-05,
-      "loss": 5.3029,
-      "step": 18145
-    },
-    {
-      "epoch": 9.463363754889178,
-      "grad_norm": 1.392197608947754,
-      "learning_rate": 8.22713567839196e-05,
-      "loss": 5.6463,
-      "step": 18146
-    },
-    {
-      "epoch": 9.463885267275097,
-      "grad_norm": 1.433524250984192,
-      "learning_rate": 8.227035175879398e-05,
-      "loss": 5.3958,
-      "step": 18147
-    },
-    {
-      "epoch": 9.464406779661017,
-      "grad_norm": 1.4635636806488037,
-      "learning_rate": 8.226934673366836e-05,
-      "loss": 5.5263,
-      "step": 18148
-    },
-    {
-      "epoch": 9.464928292046936,
-      "grad_norm": 1.4210084676742554,
-      "learning_rate": 8.226834170854272e-05,
-      "loss": 5.6245,
-      "step": 18149
-    },
-    {
-      "epoch": 9.465449804432856,
-      "grad_norm": 1.423142671585083,
-      "learning_rate": 8.226733668341708e-05,
-      "loss": 5.4601,
-      "step": 18150
-    },
-    {
-      "epoch": 9.465971316818774,
-      "grad_norm": 1.4893046617507935,
-      "learning_rate": 8.226633165829146e-05,
-      "loss": 5.0598,
-      "step": 18151
-    },
-    {
-      "epoch": 9.466492829204693,
-      "grad_norm": 1.4561524391174316,
-      "learning_rate": 8.226532663316583e-05,
-      "loss": 5.2508,
-      "step": 18152
-    },
-    {
-      "epoch": 9.467014341590613,
-      "grad_norm": 1.452970027923584,
-      "learning_rate": 8.22643216080402e-05,
-      "loss": 5.3921,
-      "step": 18153
-    },
-    {
-      "epoch": 9.467535853976532,
-      "grad_norm": 1.411190390586853,
-      "learning_rate": 8.226331658291457e-05,
-      "loss": 5.2647,
-      "step": 18154
-    },
-    {
-      "epoch": 9.468057366362451,
-      "grad_norm": 1.4163627624511719,
-      "learning_rate": 8.226231155778895e-05,
-      "loss": 5.3229,
-      "step": 18155
-    },
-    {
-      "epoch": 9.468578878748371,
-      "grad_norm": 1.3621416091918945,
-      "learning_rate": 8.226130653266332e-05,
-      "loss": 5.8127,
-      "step": 18156
-    },
-    {
-      "epoch": 9.469100391134289,
-      "grad_norm": 1.3748116493225098,
-      "learning_rate": 8.22603015075377e-05,
-      "loss": 5.7932,
-      "step": 18157
-    },
-    {
-      "epoch": 9.469621903520208,
-      "grad_norm": 1.4435878992080688,
-      "learning_rate": 8.225929648241207e-05,
-      "loss": 5.4144,
-      "step": 18158
-    },
-    {
-      "epoch": 9.470143415906128,
-      "grad_norm": 1.4483124017715454,
-      "learning_rate": 8.225829145728644e-05,
-      "loss": 5.357,
-      "step": 18159
-    },
-    {
-      "epoch": 9.470664928292047,
-      "grad_norm": 1.4436863660812378,
-      "learning_rate": 8.225728643216081e-05,
-      "loss": 5.3453,
-      "step": 18160
-    },
-    {
-      "epoch": 9.471186440677966,
-      "grad_norm": 1.498558521270752,
-      "learning_rate": 8.225628140703519e-05,
-      "loss": 5.4217,
-      "step": 18161
-    },
-    {
-      "epoch": 9.471707953063886,
-      "grad_norm": 1.395704984664917,
-      "learning_rate": 8.225527638190955e-05,
-      "loss": 5.2846,
-      "step": 18162
-    },
-    {
-      "epoch": 9.472229465449804,
-      "grad_norm": 1.4260677099227905,
-      "learning_rate": 8.225427135678391e-05,
-      "loss": 5.5014,
-      "step": 18163
-    },
-    {
-      "epoch": 9.472750977835723,
-      "grad_norm": 1.3684805631637573,
-      "learning_rate": 8.225326633165829e-05,
-      "loss": 5.7329,
-      "step": 18164
-    },
-    {
-      "epoch": 9.473272490221643,
-      "grad_norm": 1.3593389987945557,
-      "learning_rate": 8.225226130653266e-05,
-      "loss": 5.5549,
-      "step": 18165
-    },
-    {
-      "epoch": 9.473794002607562,
-      "grad_norm": 1.4801114797592163,
-      "learning_rate": 8.225125628140703e-05,
-      "loss": 5.2773,
-      "step": 18166
-    },
-    {
-      "epoch": 9.474315514993481,
-      "grad_norm": 1.5836405754089355,
-      "learning_rate": 8.225025125628141e-05,
-      "loss": 5.4989,
-      "step": 18167
-    },
-    {
-      "epoch": 9.474837027379401,
-      "grad_norm": 1.4267648458480835,
-      "learning_rate": 8.224924623115579e-05,
-      "loss": 5.5055,
-      "step": 18168
-    },
-    {
-      "epoch": 9.475358539765319,
-      "grad_norm": 1.4449280500411987,
-      "learning_rate": 8.224824120603015e-05,
-      "loss": 5.4869,
-      "step": 18169
-    },
-    {
-      "epoch": 9.475880052151238,
-      "grad_norm": 1.437248945236206,
-      "learning_rate": 8.224723618090453e-05,
-      "loss": 5.5926,
-      "step": 18170
-    },
-    {
-      "epoch": 9.476401564537158,
-      "grad_norm": 1.452779769897461,
-      "learning_rate": 8.22462311557789e-05,
-      "loss": 5.7291,
-      "step": 18171
-    },
-    {
-      "epoch": 9.476923076923077,
-      "grad_norm": 1.5104244947433472,
-      "learning_rate": 8.224522613065327e-05,
-      "loss": 5.5095,
-      "step": 18172
-    },
-    {
-      "epoch": 9.477444589308996,
-      "grad_norm": 1.511814832687378,
-      "learning_rate": 8.224422110552764e-05,
-      "loss": 5.0351,
-      "step": 18173
-    },
-    {
-      "epoch": 9.477966101694916,
-      "grad_norm": 1.4124157428741455,
-      "learning_rate": 8.224321608040202e-05,
-      "loss": 5.8251,
-      "step": 18174
-    },
-    {
-      "epoch": 9.478487614080834,
-      "grad_norm": 1.4250255823135376,
-      "learning_rate": 8.224221105527638e-05,
-      "loss": 5.5644,
-      "step": 18175
-    },
-    {
-      "epoch": 9.479009126466753,
-      "grad_norm": 1.3649638891220093,
-      "learning_rate": 8.224120603015076e-05,
-      "loss": 5.7612,
-      "step": 18176
-    },
-    {
-      "epoch": 9.479530638852673,
-      "grad_norm": 1.401392936706543,
-      "learning_rate": 8.224020100502514e-05,
-      "loss": 5.4735,
-      "step": 18177
-    },
-    {
-      "epoch": 9.480052151238592,
-      "grad_norm": 1.4920768737792969,
-      "learning_rate": 8.22391959798995e-05,
-      "loss": 5.5475,
-      "step": 18178
-    },
-    {
-      "epoch": 9.480573663624511,
-      "grad_norm": 1.4040324687957764,
-      "learning_rate": 8.223819095477388e-05,
-      "loss": 5.4189,
-      "step": 18179
-    },
-    {
-      "epoch": 9.481095176010431,
-      "grad_norm": 1.4657247066497803,
-      "learning_rate": 8.223718592964824e-05,
-      "loss": 5.3389,
-      "step": 18180
-    },
-    {
-      "epoch": 9.481616688396349,
-      "grad_norm": 1.4899739027023315,
-      "learning_rate": 8.223618090452262e-05,
-      "loss": 5.6506,
-      "step": 18181
-    },
-    {
-      "epoch": 9.482138200782268,
-      "grad_norm": 1.3842153549194336,
-      "learning_rate": 8.223517587939699e-05,
-      "loss": 5.6072,
-      "step": 18182
-    },
-    {
-      "epoch": 9.482659713168188,
-      "grad_norm": 1.4851194620132446,
-      "learning_rate": 8.223417085427136e-05,
-      "loss": 5.7379,
-      "step": 18183
-    },
-    {
-      "epoch": 9.483181225554107,
-      "grad_norm": 1.5168735980987549,
-      "learning_rate": 8.223316582914573e-05,
-      "loss": 4.9531,
-      "step": 18184
-    },
-    {
-      "epoch": 9.483702737940026,
-      "grad_norm": 1.2675786018371582,
-      "learning_rate": 8.22321608040201e-05,
-      "loss": 5.6221,
-      "step": 18185
-    },
-    {
-      "epoch": 9.484224250325946,
-      "grad_norm": 1.530930519104004,
-      "learning_rate": 8.223115577889447e-05,
-      "loss": 5.0403,
-      "step": 18186
-    },
-    {
-      "epoch": 9.484745762711864,
-      "grad_norm": 1.5466848611831665,
-      "learning_rate": 8.223015075376885e-05,
-      "loss": 5.4197,
-      "step": 18187
-    },
-    {
-      "epoch": 9.485267275097783,
-      "grad_norm": 1.3801974058151245,
-      "learning_rate": 8.222914572864322e-05,
-      "loss": 4.8385,
-      "step": 18188
-    },
-    {
-      "epoch": 9.485788787483703,
-      "grad_norm": 1.4981963634490967,
-      "learning_rate": 8.222814070351759e-05,
-      "loss": 5.5528,
-      "step": 18189
-    },
-    {
-      "epoch": 9.486310299869622,
-      "grad_norm": 1.5197471380233765,
-      "learning_rate": 8.222713567839197e-05,
-      "loss": 5.5813,
-      "step": 18190
-    },
-    {
-      "epoch": 9.486831812255542,
-      "grad_norm": 1.3611218929290771,
-      "learning_rate": 8.222613065326633e-05,
-      "loss": 5.3733,
-      "step": 18191
-    },
-    {
-      "epoch": 9.487353324641461,
-      "grad_norm": 1.5808380842208862,
-      "learning_rate": 8.222512562814071e-05,
-      "loss": 4.7444,
-      "step": 18192
-    },
-    {
-      "epoch": 9.487874837027379,
-      "grad_norm": 1.9542973041534424,
-      "learning_rate": 8.222412060301507e-05,
-      "loss": 4.9066,
-      "step": 18193
-    },
-    {
-      "epoch": 9.488396349413298,
-      "grad_norm": 1.3597484827041626,
-      "learning_rate": 8.222311557788945e-05,
-      "loss": 5.6176,
-      "step": 18194
-    },
-    {
-      "epoch": 9.488917861799218,
-      "grad_norm": 1.4931716918945312,
-      "learning_rate": 8.222211055276382e-05,
-      "loss": 5.9245,
-      "step": 18195
-    },
-    {
-      "epoch": 9.489439374185137,
-      "grad_norm": 1.4505963325500488,
-      "learning_rate": 8.22211055276382e-05,
-      "loss": 5.478,
-      "step": 18196
-    },
-    {
-      "epoch": 9.489960886571057,
-      "grad_norm": 1.3911272287368774,
-      "learning_rate": 8.222010050251257e-05,
-      "loss": 5.8677,
-      "step": 18197
-    },
-    {
-      "epoch": 9.490482398956976,
-      "grad_norm": 1.4939067363739014,
-      "learning_rate": 8.221909547738695e-05,
-      "loss": 5.9113,
-      "step": 18198
-    },
-    {
-      "epoch": 9.491003911342894,
-      "grad_norm": 1.3979356288909912,
-      "learning_rate": 8.221809045226131e-05,
-      "loss": 5.5785,
-      "step": 18199
-    },
-    {
-      "epoch": 9.491525423728813,
-      "grad_norm": 1.600269079208374,
-      "learning_rate": 8.221708542713569e-05,
-      "loss": 4.367,
-      "step": 18200
-    },
-    {
-      "epoch": 9.492046936114733,
-      "grad_norm": 1.6095668077468872,
-      "learning_rate": 8.221608040201006e-05,
-      "loss": 5.1741,
-      "step": 18201
-    },
-    {
-      "epoch": 9.492568448500652,
-      "grad_norm": 1.4539459943771362,
-      "learning_rate": 8.221507537688442e-05,
-      "loss": 5.8506,
-      "step": 18202
-    },
-    {
-      "epoch": 9.493089960886572,
-      "grad_norm": 1.4142072200775146,
-      "learning_rate": 8.22140703517588e-05,
-      "loss": 5.6662,
-      "step": 18203
-    },
-    {
-      "epoch": 9.493611473272491,
-      "grad_norm": 1.4713304042816162,
-      "learning_rate": 8.221306532663316e-05,
-      "loss": 5.6001,
-      "step": 18204
-    },
-    {
-      "epoch": 9.494132985658409,
-      "grad_norm": 1.4410302639007568,
-      "learning_rate": 8.221206030150754e-05,
-      "loss": 5.3059,
-      "step": 18205
-    },
-    {
-      "epoch": 9.494654498044328,
-      "grad_norm": 1.490399956703186,
-      "learning_rate": 8.22110552763819e-05,
-      "loss": 5.2455,
-      "step": 18206
-    },
-    {
-      "epoch": 9.495176010430248,
-      "grad_norm": 1.3780299425125122,
-      "learning_rate": 8.221005025125628e-05,
-      "loss": 5.8952,
-      "step": 18207
-    },
-    {
-      "epoch": 9.495697522816167,
-      "grad_norm": 1.4944435358047485,
-      "learning_rate": 8.220904522613066e-05,
-      "loss": 5.4811,
-      "step": 18208
-    },
-    {
-      "epoch": 9.496219035202087,
-      "grad_norm": 1.5864391326904297,
-      "learning_rate": 8.220804020100504e-05,
-      "loss": 5.3721,
-      "step": 18209
-    },
-    {
-      "epoch": 9.496740547588006,
-      "grad_norm": 1.5121114253997803,
-      "learning_rate": 8.22070351758794e-05,
-      "loss": 5.9032,
-      "step": 18210
-    },
-    {
-      "epoch": 9.497262059973924,
-      "grad_norm": 1.5650163888931274,
-      "learning_rate": 8.220603015075378e-05,
-      "loss": 4.9563,
-      "step": 18211
-    },
-    {
-      "epoch": 9.497783572359843,
-      "grad_norm": 1.367896556854248,
-      "learning_rate": 8.220502512562814e-05,
-      "loss": 5.0594,
-      "step": 18212
-    },
-    {
-      "epoch": 9.498305084745763,
-      "grad_norm": 1.4310622215270996,
-      "learning_rate": 8.220402010050252e-05,
-      "loss": 5.6884,
-      "step": 18213
-    },
-    {
-      "epoch": 9.498826597131682,
-      "grad_norm": 1.469844102859497,
-      "learning_rate": 8.220301507537689e-05,
-      "loss": 5.4649,
-      "step": 18214
-    },
-    {
-      "epoch": 9.499348109517602,
-      "grad_norm": 1.5925878286361694,
-      "learning_rate": 8.220201005025125e-05,
-      "loss": 5.3178,
-      "step": 18215
-    },
-    {
-      "epoch": 9.499869621903521,
-      "grad_norm": 1.462751865386963,
-      "learning_rate": 8.220100502512563e-05,
-      "loss": 5.6707,
-      "step": 18216
-    },
-    {
-      "epoch": 9.500391134289439,
-      "grad_norm": 1.4164036512374878,
-      "learning_rate": 8.22e-05,
-      "loss": 5.4118,
-      "step": 18217
-    },
-    {
-      "epoch": 9.500912646675358,
-      "grad_norm": 1.429270625114441,
-      "learning_rate": 8.219899497487438e-05,
-      "loss": 5.5524,
-      "step": 18218
-    },
-    {
-      "epoch": 9.501434159061278,
-      "grad_norm": 1.4899104833602905,
-      "learning_rate": 8.219798994974875e-05,
-      "loss": 5.5119,
-      "step": 18219
-    },
-    {
-      "epoch": 9.501955671447197,
-      "grad_norm": 1.4726436138153076,
-      "learning_rate": 8.219698492462313e-05,
-      "loss": 5.4973,
-      "step": 18220
-    },
-    {
-      "epoch": 9.502477183833117,
-      "grad_norm": 1.416284203529358,
-      "learning_rate": 8.219597989949749e-05,
-      "loss": 5.5501,
-      "step": 18221
-    },
-    {
-      "epoch": 9.502998696219036,
-      "grad_norm": 1.3790576457977295,
-      "learning_rate": 8.219497487437187e-05,
-      "loss": 5.5611,
-      "step": 18222
-    },
-    {
-      "epoch": 9.503520208604954,
-      "grad_norm": 1.3395150899887085,
-      "learning_rate": 8.219396984924623e-05,
-      "loss": 5.7675,
-      "step": 18223
-    },
-    {
-      "epoch": 9.504041720990873,
-      "grad_norm": 1.4494237899780273,
-      "learning_rate": 8.219296482412061e-05,
-      "loss": 4.9527,
-      "step": 18224
-    },
-    {
-      "epoch": 9.504563233376793,
-      "grad_norm": 1.4830390214920044,
-      "learning_rate": 8.219195979899497e-05,
-      "loss": 5.2787,
-      "step": 18225
-    },
-    {
-      "epoch": 9.505084745762712,
-      "grad_norm": 1.434812068939209,
-      "learning_rate": 8.219095477386935e-05,
-      "loss": 5.1871,
-      "step": 18226
-    },
-    {
-      "epoch": 9.505606258148632,
-      "grad_norm": 1.4104773998260498,
-      "learning_rate": 8.218994974874372e-05,
-      "loss": 5.9006,
-      "step": 18227
-    },
-    {
-      "epoch": 9.506127770534551,
-      "grad_norm": 1.461693525314331,
-      "learning_rate": 8.21889447236181e-05,
-      "loss": 5.0644,
-      "step": 18228
-    },
-    {
-      "epoch": 9.506649282920469,
-      "grad_norm": 1.6193628311157227,
-      "learning_rate": 8.218793969849247e-05,
-      "loss": 5.5503,
-      "step": 18229
-    },
-    {
-      "epoch": 9.507170795306388,
-      "grad_norm": 1.5623362064361572,
-      "learning_rate": 8.218693467336684e-05,
-      "loss": 5.4542,
-      "step": 18230
-    },
-    {
-      "epoch": 9.507692307692308,
-      "grad_norm": 1.3683351278305054,
-      "learning_rate": 8.218592964824121e-05,
-      "loss": 4.9658,
-      "step": 18231
-    },
-    {
-      "epoch": 9.508213820078227,
-      "grad_norm": 1.4455311298370361,
-      "learning_rate": 8.218492462311558e-05,
-      "loss": 5.3637,
-      "step": 18232
-    },
-    {
-      "epoch": 9.508735332464147,
-      "grad_norm": 1.406025767326355,
-      "learning_rate": 8.218391959798996e-05,
-      "loss": 5.6437,
-      "step": 18233
-    },
-    {
-      "epoch": 9.509256844850064,
-      "grad_norm": 1.3505845069885254,
-      "learning_rate": 8.218291457286432e-05,
-      "loss": 5.3173,
-      "step": 18234
-    },
-    {
-      "epoch": 9.509778357235984,
-      "grad_norm": 1.3794913291931152,
-      "learning_rate": 8.21819095477387e-05,
-      "loss": 5.2234,
-      "step": 18235
-    },
-    {
-      "epoch": 9.510299869621903,
-      "grad_norm": 1.3628244400024414,
-      "learning_rate": 8.218090452261306e-05,
-      "loss": 5.7041,
-      "step": 18236
-    },
-    {
-      "epoch": 9.510821382007823,
-      "grad_norm": 1.593664526939392,
-      "learning_rate": 8.217989949748744e-05,
-      "loss": 5.4752,
-      "step": 18237
-    },
-    {
-      "epoch": 9.511342894393742,
-      "grad_norm": 1.3716124296188354,
-      "learning_rate": 8.217889447236182e-05,
-      "loss": 5.7151,
-      "step": 18238
-    },
-    {
-      "epoch": 9.511864406779662,
-      "grad_norm": 1.363182544708252,
-      "learning_rate": 8.21778894472362e-05,
-      "loss": 5.602,
-      "step": 18239
-    },
-    {
-      "epoch": 9.512385919165581,
-      "grad_norm": 1.3740439414978027,
-      "learning_rate": 8.217688442211056e-05,
-      "loss": 5.5541,
-      "step": 18240
-    },
-    {
-      "epoch": 9.512907431551499,
-      "grad_norm": 1.4734939336776733,
-      "learning_rate": 8.217587939698494e-05,
-      "loss": 5.5309,
-      "step": 18241
-    },
-    {
-      "epoch": 9.513428943937418,
-      "grad_norm": 1.4590164422988892,
-      "learning_rate": 8.21748743718593e-05,
-      "loss": 5.7523,
-      "step": 18242
-    },
-    {
-      "epoch": 9.513950456323338,
-      "grad_norm": 1.4605140686035156,
-      "learning_rate": 8.217386934673367e-05,
-      "loss": 5.3006,
-      "step": 18243
-    },
-    {
-      "epoch": 9.514471968709257,
-      "grad_norm": 1.74355149269104,
-      "learning_rate": 8.217286432160804e-05,
-      "loss": 5.4567,
-      "step": 18244
-    },
-    {
-      "epoch": 9.514993481095177,
-      "grad_norm": 1.3160961866378784,
-      "learning_rate": 8.217185929648241e-05,
-      "loss": 5.5054,
-      "step": 18245
-    },
-    {
-      "epoch": 9.515514993481094,
-      "grad_norm": 1.3497449159622192,
-      "learning_rate": 8.217085427135679e-05,
-      "loss": 5.7948,
-      "step": 18246
-    },
-    {
-      "epoch": 9.516036505867014,
-      "grad_norm": 1.6617319583892822,
-      "learning_rate": 8.216984924623115e-05,
-      "loss": 5.4085,
-      "step": 18247
-    },
-    {
-      "epoch": 9.516558018252933,
-      "grad_norm": 1.53920316696167,
-      "learning_rate": 8.216884422110553e-05,
-      "loss": 5.8398,
-      "step": 18248
-    },
-    {
-      "epoch": 9.517079530638853,
-      "grad_norm": 1.5589096546173096,
-      "learning_rate": 8.21678391959799e-05,
-      "loss": 5.7034,
-      "step": 18249
-    },
-    {
-      "epoch": 9.517601043024772,
-      "grad_norm": 1.4164501428604126,
-      "learning_rate": 8.216683417085428e-05,
-      "loss": 5.0216,
-      "step": 18250
-    },
-    {
-      "epoch": 9.518122555410692,
-      "grad_norm": 1.4184236526489258,
-      "learning_rate": 8.216582914572865e-05,
-      "loss": 5.4354,
-      "step": 18251
-    },
-    {
-      "epoch": 9.518644067796611,
-      "grad_norm": 1.385456919670105,
-      "learning_rate": 8.216482412060303e-05,
-      "loss": 5.6413,
-      "step": 18252
-    },
-    {
-      "epoch": 9.519165580182529,
-      "grad_norm": 1.3513023853302002,
-      "learning_rate": 8.216381909547739e-05,
-      "loss": 5.3303,
-      "step": 18253
-    },
-    {
-      "epoch": 9.519687092568448,
-      "grad_norm": 1.4512683153152466,
-      "learning_rate": 8.216281407035177e-05,
-      "loss": 5.6187,
-      "step": 18254
-    },
-    {
-      "epoch": 9.520208604954368,
-      "grad_norm": 1.455775260925293,
-      "learning_rate": 8.216180904522613e-05,
-      "loss": 5.3244,
-      "step": 18255
-    },
-    {
-      "epoch": 9.520730117340287,
-      "grad_norm": 1.304795265197754,
-      "learning_rate": 8.21608040201005e-05,
-      "loss": 5.2349,
-      "step": 18256
-    },
-    {
-      "epoch": 9.521251629726207,
-      "grad_norm": 1.3285906314849854,
-      "learning_rate": 8.215979899497487e-05,
-      "loss": 5.759,
-      "step": 18257
-    },
-    {
-      "epoch": 9.521773142112124,
-      "grad_norm": 1.4240984916687012,
-      "learning_rate": 8.215879396984924e-05,
-      "loss": 5.79,
-      "step": 18258
-    },
-    {
-      "epoch": 9.522294654498044,
-      "grad_norm": 1.3226473331451416,
-      "learning_rate": 8.215778894472362e-05,
-      "loss": 5.3782,
-      "step": 18259
-    },
-    {
-      "epoch": 9.522816166883963,
-      "grad_norm": 1.4028522968292236,
-      "learning_rate": 8.2156783919598e-05,
-      "loss": 5.5869,
-      "step": 18260
-    },
-    {
-      "epoch": 9.523337679269883,
-      "grad_norm": 1.5244452953338623,
-      "learning_rate": 8.215577889447237e-05,
-      "loss": 5.0069,
-      "step": 18261
-    },
-    {
-      "epoch": 9.523859191655802,
-      "grad_norm": 1.4071040153503418,
-      "learning_rate": 8.215477386934674e-05,
-      "loss": 5.7212,
-      "step": 18262
-    },
-    {
-      "epoch": 9.524380704041722,
-      "grad_norm": 1.6735200881958008,
-      "learning_rate": 8.215376884422111e-05,
-      "loss": 5.3954,
-      "step": 18263
-    },
-    {
-      "epoch": 9.52490221642764,
-      "grad_norm": 1.404481291770935,
-      "learning_rate": 8.215276381909548e-05,
-      "loss": 5.7385,
-      "step": 18264
-    },
-    {
-      "epoch": 9.525423728813559,
-      "grad_norm": 1.5389699935913086,
-      "learning_rate": 8.215175879396986e-05,
-      "loss": 5.0943,
-      "step": 18265
-    },
-    {
-      "epoch": 9.525945241199478,
-      "grad_norm": 1.458540439605713,
-      "learning_rate": 8.215075376884422e-05,
-      "loss": 5.1922,
-      "step": 18266
-    },
-    {
-      "epoch": 9.526466753585398,
-      "grad_norm": 1.4522292613983154,
-      "learning_rate": 8.21497487437186e-05,
-      "loss": 5.5324,
-      "step": 18267
-    },
-    {
-      "epoch": 9.526988265971317,
-      "grad_norm": 1.4940396547317505,
-      "learning_rate": 8.214874371859296e-05,
-      "loss": 5.105,
-      "step": 18268
-    },
-    {
-      "epoch": 9.527509778357237,
-      "grad_norm": 1.4925856590270996,
-      "learning_rate": 8.214773869346734e-05,
-      "loss": 5.0233,
-      "step": 18269
-    },
-    {
-      "epoch": 9.528031290743154,
-      "grad_norm": 1.4090768098831177,
-      "learning_rate": 8.214673366834172e-05,
-      "loss": 5.7091,
-      "step": 18270
-    },
-    {
-      "epoch": 9.528552803129074,
-      "grad_norm": 1.516034722328186,
-      "learning_rate": 8.214572864321608e-05,
-      "loss": 5.5249,
-      "step": 18271
-    },
-    {
-      "epoch": 9.529074315514993,
-      "grad_norm": 1.5842247009277344,
-      "learning_rate": 8.214472361809046e-05,
-      "loss": 5.4615,
-      "step": 18272
-    },
-    {
-      "epoch": 9.529595827900913,
-      "grad_norm": 1.6436148881912231,
-      "learning_rate": 8.214371859296483e-05,
-      "loss": 4.6069,
-      "step": 18273
-    },
-    {
-      "epoch": 9.530117340286832,
-      "grad_norm": 1.4806472063064575,
-      "learning_rate": 8.21427135678392e-05,
-      "loss": 5.4777,
-      "step": 18274
-    },
-    {
-      "epoch": 9.530638852672752,
-      "grad_norm": 1.428080677986145,
-      "learning_rate": 8.214170854271357e-05,
-      "loss": 5.0148,
-      "step": 18275
-    },
-    {
-      "epoch": 9.53116036505867,
-      "grad_norm": 1.465238332748413,
-      "learning_rate": 8.214070351758795e-05,
-      "loss": 5.5715,
-      "step": 18276
-    },
-    {
-      "epoch": 9.531681877444589,
-      "grad_norm": 1.4791090488433838,
-      "learning_rate": 8.213969849246231e-05,
-      "loss": 5.5603,
-      "step": 18277
-    },
-    {
-      "epoch": 9.532203389830508,
-      "grad_norm": 1.5235860347747803,
-      "learning_rate": 8.213869346733669e-05,
-      "loss": 4.9259,
-      "step": 18278
-    },
-    {
-      "epoch": 9.532724902216428,
-      "grad_norm": 1.3658149242401123,
-      "learning_rate": 8.213768844221105e-05,
-      "loss": 5.5824,
-      "step": 18279
-    },
-    {
-      "epoch": 9.533246414602347,
-      "grad_norm": 1.3070549964904785,
-      "learning_rate": 8.213668341708543e-05,
-      "loss": 5.8367,
-      "step": 18280
-    },
-    {
-      "epoch": 9.533767926988267,
-      "grad_norm": 1.3196557760238647,
-      "learning_rate": 8.213567839195981e-05,
-      "loss": 5.6928,
-      "step": 18281
-    },
-    {
-      "epoch": 9.534289439374184,
-      "grad_norm": 1.3971120119094849,
-      "learning_rate": 8.213467336683417e-05,
-      "loss": 5.5049,
-      "step": 18282
-    },
-    {
-      "epoch": 9.534810951760104,
-      "grad_norm": 1.4376627206802368,
-      "learning_rate": 8.213366834170855e-05,
-      "loss": 5.729,
-      "step": 18283
-    },
-    {
-      "epoch": 9.535332464146023,
-      "grad_norm": 1.4513133764266968,
-      "learning_rate": 8.213266331658291e-05,
-      "loss": 5.7432,
-      "step": 18284
-    },
-    {
-      "epoch": 9.535853976531943,
-      "grad_norm": 1.367724061012268,
-      "learning_rate": 8.213165829145729e-05,
-      "loss": 5.4314,
-      "step": 18285
-    },
-    {
-      "epoch": 9.536375488917862,
-      "grad_norm": 1.4014049768447876,
-      "learning_rate": 8.213065326633166e-05,
-      "loss": 4.9823,
-      "step": 18286
-    },
-    {
-      "epoch": 9.536897001303782,
-      "grad_norm": 1.3572008609771729,
-      "learning_rate": 8.212964824120603e-05,
-      "loss": 5.5571,
-      "step": 18287
-    },
-    {
-      "epoch": 9.5374185136897,
-      "grad_norm": 1.4626684188842773,
-      "learning_rate": 8.21286432160804e-05,
-      "loss": 5.1801,
-      "step": 18288
-    },
-    {
-      "epoch": 9.537940026075619,
-      "grad_norm": 1.4876142740249634,
-      "learning_rate": 8.212763819095478e-05,
-      "loss": 5.4644,
-      "step": 18289
-    },
-    {
-      "epoch": 9.538461538461538,
-      "grad_norm": 1.4779225587844849,
-      "learning_rate": 8.212663316582915e-05,
-      "loss": 5.5038,
-      "step": 18290
-    },
-    {
-      "epoch": 9.538983050847458,
-      "grad_norm": 1.3988007307052612,
-      "learning_rate": 8.212562814070353e-05,
-      "loss": 5.3775,
-      "step": 18291
-    },
-    {
-      "epoch": 9.539504563233377,
-      "grad_norm": 1.3744595050811768,
-      "learning_rate": 8.21246231155779e-05,
-      "loss": 5.7703,
-      "step": 18292
-    },
-    {
-      "epoch": 9.540026075619297,
-      "grad_norm": 1.4616116285324097,
-      "learning_rate": 8.212361809045227e-05,
-      "loss": 5.6302,
-      "step": 18293
-    },
-    {
-      "epoch": 9.540547588005214,
-      "grad_norm": 1.4305635690689087,
-      "learning_rate": 8.212261306532664e-05,
-      "loss": 5.4902,
-      "step": 18294
-    },
-    {
-      "epoch": 9.541069100391134,
-      "grad_norm": 1.4078748226165771,
-      "learning_rate": 8.2121608040201e-05,
-      "loss": 5.8846,
-      "step": 18295
-    },
-    {
-      "epoch": 9.541590612777053,
-      "grad_norm": 1.3034636974334717,
-      "learning_rate": 8.212060301507538e-05,
-      "loss": 5.7178,
-      "step": 18296
-    },
-    {
-      "epoch": 9.542112125162973,
-      "grad_norm": 1.4172894954681396,
-      "learning_rate": 8.211959798994974e-05,
-      "loss": 5.5372,
-      "step": 18297
-    },
-    {
-      "epoch": 9.542633637548892,
-      "grad_norm": 1.48302161693573,
-      "learning_rate": 8.211859296482412e-05,
-      "loss": 5.5398,
-      "step": 18298
-    },
-    {
-      "epoch": 9.543155149934812,
-      "grad_norm": 1.2536650896072388,
-      "learning_rate": 8.211758793969849e-05,
-      "loss": 5.8565,
-      "step": 18299
-    },
-    {
-      "epoch": 9.54367666232073,
-      "grad_norm": 1.4930989742279053,
-      "learning_rate": 8.211658291457286e-05,
-      "loss": 5.2333,
-      "step": 18300
-    },
-    {
-      "epoch": 9.544198174706649,
-      "grad_norm": 1.387807011604309,
-      "learning_rate": 8.211557788944724e-05,
-      "loss": 5.5534,
-      "step": 18301
-    },
-    {
-      "epoch": 9.544719687092568,
-      "grad_norm": 1.3794152736663818,
-      "learning_rate": 8.211457286432162e-05,
-      "loss": 5.6356,
-      "step": 18302
-    },
-    {
-      "epoch": 9.545241199478488,
-      "grad_norm": 1.4397284984588623,
-      "learning_rate": 8.211356783919598e-05,
-      "loss": 5.1983,
-      "step": 18303
-    },
-    {
-      "epoch": 9.545762711864407,
-      "grad_norm": 1.4625598192214966,
-      "learning_rate": 8.211256281407036e-05,
-      "loss": 5.0812,
-      "step": 18304
-    },
-    {
-      "epoch": 9.546284224250327,
-      "grad_norm": 1.481059193611145,
-      "learning_rate": 8.211155778894473e-05,
-      "loss": 5.3373,
-      "step": 18305
-    },
-    {
-      "epoch": 9.546805736636244,
-      "grad_norm": 1.3754284381866455,
-      "learning_rate": 8.21105527638191e-05,
-      "loss": 5.0372,
-      "step": 18306
-    },
-    {
-      "epoch": 9.547327249022164,
-      "grad_norm": 1.4203444719314575,
-      "learning_rate": 8.210954773869347e-05,
-      "loss": 5.2696,
-      "step": 18307
-    },
-    {
-      "epoch": 9.547848761408083,
-      "grad_norm": 1.4433382749557495,
-      "learning_rate": 8.210854271356785e-05,
-      "loss": 5.7037,
-      "step": 18308
-    },
-    {
-      "epoch": 9.548370273794003,
-      "grad_norm": 1.4698634147644043,
-      "learning_rate": 8.210753768844221e-05,
-      "loss": 5.5036,
-      "step": 18309
-    },
-    {
-      "epoch": 9.548891786179922,
-      "grad_norm": 1.3343249559402466,
-      "learning_rate": 8.210653266331659e-05,
-      "loss": 5.714,
-      "step": 18310
-    },
-    {
-      "epoch": 9.549413298565842,
-      "grad_norm": 1.594611644744873,
-      "learning_rate": 8.210552763819097e-05,
-      "loss": 5.099,
-      "step": 18311
-    },
-    {
-      "epoch": 9.54993481095176,
-      "grad_norm": 1.299289345741272,
-      "learning_rate": 8.210452261306533e-05,
-      "loss": 5.9767,
-      "step": 18312
-    },
-    {
-      "epoch": 9.550456323337679,
-      "grad_norm": 1.4694918394088745,
-      "learning_rate": 8.210351758793971e-05,
-      "loss": 5.443,
-      "step": 18313
-    },
-    {
-      "epoch": 9.550977835723598,
-      "grad_norm": 1.3582379817962646,
-      "learning_rate": 8.210251256281407e-05,
-      "loss": 5.7773,
-      "step": 18314
-    },
-    {
-      "epoch": 9.551499348109518,
-      "grad_norm": 1.4757329225540161,
-      "learning_rate": 8.210150753768845e-05,
-      "loss": 5.2156,
-      "step": 18315
-    },
-    {
-      "epoch": 9.552020860495437,
-      "grad_norm": 1.4559952020645142,
-      "learning_rate": 8.210050251256281e-05,
-      "loss": 5.7311,
-      "step": 18316
-    },
-    {
-      "epoch": 9.552542372881355,
-      "grad_norm": 1.4397358894348145,
-      "learning_rate": 8.209949748743719e-05,
-      "loss": 4.953,
-      "step": 18317
-    },
-    {
-      "epoch": 9.553063885267274,
-      "grad_norm": 1.4345252513885498,
-      "learning_rate": 8.209849246231156e-05,
-      "loss": 5.4396,
-      "step": 18318
-    },
-    {
-      "epoch": 9.553585397653194,
-      "grad_norm": 1.4411087036132812,
-      "learning_rate": 8.209748743718593e-05,
-      "loss": 5.318,
-      "step": 18319
-    },
-    {
-      "epoch": 9.554106910039113,
-      "grad_norm": 1.359351396560669,
-      "learning_rate": 8.20964824120603e-05,
-      "loss": 5.4927,
-      "step": 18320
-    },
-    {
-      "epoch": 9.554628422425033,
-      "grad_norm": 1.3427972793579102,
-      "learning_rate": 8.209547738693468e-05,
-      "loss": 5.685,
-      "step": 18321
-    },
-    {
-      "epoch": 9.555149934810952,
-      "grad_norm": 1.4132918119430542,
-      "learning_rate": 8.209447236180905e-05,
-      "loss": 5.6539,
-      "step": 18322
-    },
-    {
-      "epoch": 9.555671447196872,
-      "grad_norm": 1.3377482891082764,
-      "learning_rate": 8.209346733668342e-05,
-      "loss": 5.8177,
-      "step": 18323
-    },
-    {
-      "epoch": 9.55619295958279,
-      "grad_norm": 1.4829702377319336,
-      "learning_rate": 8.20924623115578e-05,
-      "loss": 5.1974,
-      "step": 18324
-    },
-    {
-      "epoch": 9.556714471968709,
-      "grad_norm": 1.5048482418060303,
-      "learning_rate": 8.209145728643216e-05,
-      "loss": 5.0607,
-      "step": 18325
-    },
-    {
-      "epoch": 9.557235984354628,
-      "grad_norm": 1.2498520612716675,
-      "learning_rate": 8.209045226130654e-05,
-      "loss": 4.6177,
-      "step": 18326
-    },
-    {
-      "epoch": 9.557757496740548,
-      "grad_norm": 1.4731005430221558,
-      "learning_rate": 8.20894472361809e-05,
-      "loss": 5.2686,
-      "step": 18327
-    },
-    {
-      "epoch": 9.558279009126467,
-      "grad_norm": 1.4990254640579224,
-      "learning_rate": 8.208844221105528e-05,
-      "loss": 5.9133,
-      "step": 18328
-    },
-    {
-      "epoch": 9.558800521512385,
-      "grad_norm": 1.5770225524902344,
-      "learning_rate": 8.208743718592964e-05,
-      "loss": 5.4638,
-      "step": 18329
-    },
-    {
-      "epoch": 9.559322033898304,
-      "grad_norm": 1.3742477893829346,
-      "learning_rate": 8.208643216080402e-05,
-      "loss": 5.5491,
-      "step": 18330
-    },
-    {
-      "epoch": 9.559843546284224,
-      "grad_norm": 1.3710339069366455,
-      "learning_rate": 8.20854271356784e-05,
-      "loss": 5.5765,
-      "step": 18331
-    },
-    {
-      "epoch": 9.560365058670143,
-      "grad_norm": 1.4051966667175293,
-      "learning_rate": 8.208442211055278e-05,
-      "loss": 5.5219,
-      "step": 18332
-    },
-    {
-      "epoch": 9.560886571056063,
-      "grad_norm": 1.5180063247680664,
-      "learning_rate": 8.208341708542714e-05,
-      "loss": 5.4054,
-      "step": 18333
-    },
-    {
-      "epoch": 9.561408083441982,
-      "grad_norm": 1.5427783727645874,
-      "learning_rate": 8.208241206030152e-05,
-      "loss": 5.1961,
-      "step": 18334
-    },
-    {
-      "epoch": 9.561929595827902,
-      "grad_norm": 1.583938479423523,
-      "learning_rate": 8.208140703517588e-05,
-      "loss": 5.4539,
-      "step": 18335
-    },
-    {
-      "epoch": 9.56245110821382,
-      "grad_norm": 1.5169938802719116,
-      "learning_rate": 8.208040201005025e-05,
-      "loss": 5.1016,
-      "step": 18336
-    },
-    {
-      "epoch": 9.562972620599739,
-      "grad_norm": 1.4597307443618774,
-      "learning_rate": 8.207939698492463e-05,
-      "loss": 4.7695,
-      "step": 18337
-    },
-    {
-      "epoch": 9.563494132985658,
-      "grad_norm": 1.4156146049499512,
-      "learning_rate": 8.207839195979899e-05,
-      "loss": 5.5606,
-      "step": 18338
-    },
-    {
-      "epoch": 9.564015645371578,
-      "grad_norm": 1.4167028665542603,
-      "learning_rate": 8.207738693467337e-05,
-      "loss": 5.579,
-      "step": 18339
-    },
-    {
-      "epoch": 9.564537157757497,
-      "grad_norm": 1.5436700582504272,
-      "learning_rate": 8.207638190954773e-05,
-      "loss": 5.5429,
-      "step": 18340
-    },
-    {
-      "epoch": 9.565058670143415,
-      "grad_norm": 1.4117560386657715,
-      "learning_rate": 8.207537688442211e-05,
-      "loss": 5.4547,
-      "step": 18341
-    },
-    {
-      "epoch": 9.565580182529335,
-      "grad_norm": 1.4248008728027344,
-      "learning_rate": 8.207437185929649e-05,
-      "loss": 5.8892,
-      "step": 18342
-    },
-    {
-      "epoch": 9.566101694915254,
-      "grad_norm": 1.4370615482330322,
-      "learning_rate": 8.207336683417087e-05,
-      "loss": 5.3731,
-      "step": 18343
-    },
-    {
-      "epoch": 9.566623207301173,
-      "grad_norm": 1.5552817583084106,
-      "learning_rate": 8.207236180904523e-05,
-      "loss": 4.9971,
-      "step": 18344
-    },
-    {
-      "epoch": 9.567144719687093,
-      "grad_norm": 1.5064730644226074,
-      "learning_rate": 8.207135678391961e-05,
-      "loss": 5.564,
-      "step": 18345
-    },
-    {
-      "epoch": 9.567666232073012,
-      "grad_norm": 1.4385499954223633,
-      "learning_rate": 8.207035175879397e-05,
-      "loss": 5.788,
-      "step": 18346
-    },
-    {
-      "epoch": 9.568187744458932,
-      "grad_norm": 1.4837226867675781,
-      "learning_rate": 8.206934673366835e-05,
-      "loss": 5.4653,
-      "step": 18347
-    },
-    {
-      "epoch": 9.56870925684485,
-      "grad_norm": 1.3416292667388916,
-      "learning_rate": 8.206834170854272e-05,
-      "loss": 5.6925,
-      "step": 18348
-    },
-    {
-      "epoch": 9.569230769230769,
-      "grad_norm": 1.4626269340515137,
-      "learning_rate": 8.206733668341708e-05,
-      "loss": 5.5569,
-      "step": 18349
-    },
-    {
-      "epoch": 9.569752281616688,
-      "grad_norm": 1.382272720336914,
-      "learning_rate": 8.206633165829146e-05,
-      "loss": 5.7215,
-      "step": 18350
-    },
-    {
-      "epoch": 9.570273794002608,
-      "grad_norm": 1.402427077293396,
-      "learning_rate": 8.206532663316584e-05,
-      "loss": 5.6271,
-      "step": 18351
-    },
-    {
-      "epoch": 9.570795306388527,
-      "grad_norm": 1.5238440036773682,
-      "learning_rate": 8.206432160804021e-05,
-      "loss": 5.3069,
-      "step": 18352
-    },
-    {
-      "epoch": 9.571316818774445,
-      "grad_norm": 1.5014384984970093,
-      "learning_rate": 8.206331658291458e-05,
-      "loss": 5.9431,
-      "step": 18353
-    },
-    {
-      "epoch": 9.571838331160365,
-      "grad_norm": 1.4079301357269287,
-      "learning_rate": 8.206231155778896e-05,
-      "loss": 5.4642,
-      "step": 18354
-    },
-    {
-      "epoch": 9.572359843546284,
-      "grad_norm": 1.3998842239379883,
-      "learning_rate": 8.206130653266332e-05,
-      "loss": 5.6276,
-      "step": 18355
-    },
-    {
-      "epoch": 9.572881355932203,
-      "grad_norm": 1.4364532232284546,
-      "learning_rate": 8.20603015075377e-05,
-      "loss": 5.1197,
-      "step": 18356
-    },
-    {
-      "epoch": 9.573402868318123,
-      "grad_norm": 1.3005691766738892,
-      "learning_rate": 8.205929648241206e-05,
-      "loss": 5.8906,
-      "step": 18357
-    },
-    {
-      "epoch": 9.573924380704042,
-      "grad_norm": 1.3533811569213867,
-      "learning_rate": 8.205829145728644e-05,
-      "loss": 5.2059,
-      "step": 18358
-    },
-    {
-      "epoch": 9.57444589308996,
-      "grad_norm": 1.4331055879592896,
-      "learning_rate": 8.20572864321608e-05,
-      "loss": 5.7286,
-      "step": 18359
-    },
-    {
-      "epoch": 9.57496740547588,
-      "grad_norm": 1.4530460834503174,
-      "learning_rate": 8.205628140703518e-05,
-      "loss": 5.5951,
-      "step": 18360
-    },
-    {
-      "epoch": 9.575488917861799,
-      "grad_norm": 1.50508451461792,
-      "learning_rate": 8.205527638190955e-05,
-      "loss": 5.048,
-      "step": 18361
-    },
-    {
-      "epoch": 9.576010430247718,
-      "grad_norm": 1.3186484575271606,
-      "learning_rate": 8.205427135678392e-05,
-      "loss": 5.3076,
-      "step": 18362
-    },
-    {
-      "epoch": 9.576531942633638,
-      "grad_norm": 1.5734187364578247,
-      "learning_rate": 8.20532663316583e-05,
-      "loss": 5.5519,
-      "step": 18363
-    },
-    {
-      "epoch": 9.577053455019557,
-      "grad_norm": 1.4869722127914429,
-      "learning_rate": 8.205226130653267e-05,
-      "loss": 5.2079,
-      "step": 18364
-    },
-    {
-      "epoch": 9.577574967405475,
-      "grad_norm": 1.3278858661651611,
-      "learning_rate": 8.205125628140704e-05,
-      "loss": 5.9114,
-      "step": 18365
-    },
-    {
-      "epoch": 9.578096479791395,
-      "grad_norm": 1.5703504085540771,
-      "learning_rate": 8.205025125628141e-05,
-      "loss": 4.6126,
-      "step": 18366
-    },
-    {
-      "epoch": 9.578617992177314,
-      "grad_norm": 1.4676986932754517,
-      "learning_rate": 8.204924623115579e-05,
-      "loss": 5.323,
-      "step": 18367
-    },
-    {
-      "epoch": 9.579139504563233,
-      "grad_norm": 1.5338784456253052,
-      "learning_rate": 8.204824120603015e-05,
-      "loss": 5.6064,
-      "step": 18368
-    },
-    {
-      "epoch": 9.579661016949153,
-      "grad_norm": 1.3678576946258545,
-      "learning_rate": 8.204723618090453e-05,
-      "loss": 5.7686,
-      "step": 18369
-    },
-    {
-      "epoch": 9.580182529335072,
-      "grad_norm": 1.5025380849838257,
-      "learning_rate": 8.204623115577889e-05,
-      "loss": 5.5066,
-      "step": 18370
-    },
-    {
-      "epoch": 9.58070404172099,
-      "grad_norm": 1.4196665287017822,
-      "learning_rate": 8.204522613065327e-05,
-      "loss": 5.4945,
-      "step": 18371
-    },
-    {
-      "epoch": 9.58122555410691,
-      "grad_norm": 1.513031244277954,
-      "learning_rate": 8.204422110552765e-05,
-      "loss": 4.8683,
-      "step": 18372
-    },
-    {
-      "epoch": 9.581747066492829,
-      "grad_norm": 1.3793237209320068,
-      "learning_rate": 8.204321608040203e-05,
-      "loss": 4.9467,
-      "step": 18373
-    },
-    {
-      "epoch": 9.582268578878748,
-      "grad_norm": 1.5272548198699951,
-      "learning_rate": 8.204221105527639e-05,
-      "loss": 5.6703,
-      "step": 18374
-    },
-    {
-      "epoch": 9.582790091264668,
-      "grad_norm": 1.5520931482315063,
-      "learning_rate": 8.204120603015075e-05,
-      "loss": 5.2037,
-      "step": 18375
-    },
-    {
-      "epoch": 9.583311603650587,
-      "grad_norm": 1.3727627992630005,
-      "learning_rate": 8.204020100502513e-05,
-      "loss": 5.9334,
-      "step": 18376
-    },
-    {
-      "epoch": 9.583833116036505,
-      "grad_norm": 1.329674482345581,
-      "learning_rate": 8.20391959798995e-05,
-      "loss": 5.6482,
-      "step": 18377
-    },
-    {
-      "epoch": 9.584354628422425,
-      "grad_norm": 1.6507521867752075,
-      "learning_rate": 8.203819095477387e-05,
-      "loss": 4.8525,
-      "step": 18378
-    },
-    {
-      "epoch": 9.584876140808344,
-      "grad_norm": 1.470011591911316,
-      "learning_rate": 8.203718592964824e-05,
-      "loss": 5.2684,
-      "step": 18379
-    },
-    {
-      "epoch": 9.585397653194264,
-      "grad_norm": 1.4173240661621094,
-      "learning_rate": 8.203618090452262e-05,
-      "loss": 5.4601,
-      "step": 18380
-    },
-    {
-      "epoch": 9.585919165580183,
-      "grad_norm": 1.3744840621948242,
-      "learning_rate": 8.203517587939698e-05,
-      "loss": 5.2331,
-      "step": 18381
-    },
-    {
-      "epoch": 9.586440677966102,
-      "grad_norm": 1.4921026229858398,
-      "learning_rate": 8.203417085427136e-05,
-      "loss": 4.9676,
-      "step": 18382
-    },
-    {
-      "epoch": 9.58696219035202,
-      "grad_norm": 1.3921645879745483,
-      "learning_rate": 8.203316582914574e-05,
-      "loss": 5.6855,
-      "step": 18383
-    },
-    {
-      "epoch": 9.58748370273794,
-      "grad_norm": 1.4761162996292114,
-      "learning_rate": 8.203216080402011e-05,
-      "loss": 5.2246,
-      "step": 18384
-    },
-    {
-      "epoch": 9.588005215123859,
-      "grad_norm": 1.4205816984176636,
-      "learning_rate": 8.203115577889448e-05,
-      "loss": 5.2386,
-      "step": 18385
-    },
-    {
-      "epoch": 9.588526727509779,
-      "grad_norm": 1.3870853185653687,
-      "learning_rate": 8.203015075376886e-05,
-      "loss": 5.5675,
-      "step": 18386
-    },
-    {
-      "epoch": 9.589048239895698,
-      "grad_norm": 1.4582679271697998,
-      "learning_rate": 8.202914572864322e-05,
-      "loss": 5.3366,
-      "step": 18387
-    },
-    {
-      "epoch": 9.589569752281617,
-      "grad_norm": 1.533099889755249,
-      "learning_rate": 8.202814070351758e-05,
-      "loss": 5.1712,
-      "step": 18388
-    },
-    {
-      "epoch": 9.590091264667535,
-      "grad_norm": 1.7438627481460571,
-      "learning_rate": 8.202713567839196e-05,
-      "loss": 5.0833,
-      "step": 18389
-    },
-    {
-      "epoch": 9.590612777053455,
-      "grad_norm": 1.4177792072296143,
-      "learning_rate": 8.202613065326633e-05,
-      "loss": 5.6486,
-      "step": 18390
-    },
-    {
-      "epoch": 9.591134289439374,
-      "grad_norm": 1.4243345260620117,
-      "learning_rate": 8.20251256281407e-05,
-      "loss": 5.6234,
-      "step": 18391
-    },
-    {
-      "epoch": 9.591655801825294,
-      "grad_norm": 1.468032717704773,
-      "learning_rate": 8.202412060301508e-05,
-      "loss": 5.56,
-      "step": 18392
-    },
-    {
-      "epoch": 9.592177314211213,
-      "grad_norm": 1.4212567806243896,
-      "learning_rate": 8.202311557788946e-05,
-      "loss": 5.8212,
-      "step": 18393
-    },
-    {
-      "epoch": 9.592698826597132,
-      "grad_norm": 1.52122962474823,
-      "learning_rate": 8.202211055276382e-05,
-      "loss": 5.385,
-      "step": 18394
-    },
-    {
-      "epoch": 9.59322033898305,
-      "grad_norm": 1.281960129737854,
-      "learning_rate": 8.20211055276382e-05,
-      "loss": 5.1863,
-      "step": 18395
-    },
-    {
-      "epoch": 9.59374185136897,
-      "grad_norm": 1.6613415479660034,
-      "learning_rate": 8.202010050251257e-05,
-      "loss": 5.2463,
-      "step": 18396
-    },
-    {
-      "epoch": 9.594263363754889,
-      "grad_norm": 1.4577258825302124,
-      "learning_rate": 8.201909547738694e-05,
-      "loss": 5.2644,
-      "step": 18397
-    },
-    {
-      "epoch": 9.594784876140809,
-      "grad_norm": 1.4321560859680176,
-      "learning_rate": 8.201809045226131e-05,
-      "loss": 5.7986,
-      "step": 18398
-    },
-    {
-      "epoch": 9.595306388526728,
-      "grad_norm": 1.3093544244766235,
-      "learning_rate": 8.201708542713569e-05,
-      "loss": 5.6196,
-      "step": 18399
-    },
-    {
-      "epoch": 9.595827900912647,
-      "grad_norm": 1.4659130573272705,
-      "learning_rate": 8.201608040201005e-05,
-      "loss": 5.5407,
-      "step": 18400
-    },
-    {
-      "epoch": 9.596349413298565,
-      "grad_norm": 1.3577601909637451,
-      "learning_rate": 8.201507537688443e-05,
-      "loss": 5.3856,
-      "step": 18401
-    },
-    {
-      "epoch": 9.596870925684485,
-      "grad_norm": 1.4544841051101685,
-      "learning_rate": 8.201407035175879e-05,
-      "loss": 5.2095,
-      "step": 18402
-    },
-    {
-      "epoch": 9.597392438070404,
-      "grad_norm": 1.5011377334594727,
-      "learning_rate": 8.201306532663317e-05,
-      "loss": 5.3268,
-      "step": 18403
-    },
-    {
-      "epoch": 9.597913950456324,
-      "grad_norm": 1.519648790359497,
-      "learning_rate": 8.201206030150755e-05,
-      "loss": 5.0936,
-      "step": 18404
-    },
-    {
-      "epoch": 9.598435462842243,
-      "grad_norm": 1.5134702920913696,
-      "learning_rate": 8.201105527638191e-05,
-      "loss": 5.5277,
-      "step": 18405
-    },
-    {
-      "epoch": 9.598956975228162,
-      "grad_norm": 1.504813551902771,
-      "learning_rate": 8.201005025125629e-05,
-      "loss": 5.6111,
-      "step": 18406
-    },
-    {
-      "epoch": 9.59947848761408,
-      "grad_norm": 1.4620130062103271,
-      "learning_rate": 8.200904522613065e-05,
-      "loss": 4.9746,
-      "step": 18407
-    },
-    {
-      "epoch": 9.6,
-      "grad_norm": 1.426472544670105,
-      "learning_rate": 8.200804020100503e-05,
-      "loss": 5.7596,
-      "step": 18408
-    },
-    {
-      "epoch": 9.600521512385919,
-      "grad_norm": 1.5054049491882324,
-      "learning_rate": 8.20070351758794e-05,
-      "loss": 5.6354,
-      "step": 18409
-    },
-    {
-      "epoch": 9.601043024771839,
-      "grad_norm": 1.5484403371810913,
-      "learning_rate": 8.200603015075377e-05,
-      "loss": 5.0328,
-      "step": 18410
-    },
-    {
-      "epoch": 9.601564537157758,
-      "grad_norm": 1.5226997137069702,
-      "learning_rate": 8.200502512562814e-05,
-      "loss": 4.9849,
-      "step": 18411
-    },
-    {
-      "epoch": 9.602086049543676,
-      "grad_norm": 1.4151394367218018,
-      "learning_rate": 8.200402010050252e-05,
-      "loss": 5.4376,
-      "step": 18412
-    },
-    {
-      "epoch": 9.602607561929595,
-      "grad_norm": 1.5649597644805908,
-      "learning_rate": 8.200301507537688e-05,
-      "loss": 5.4881,
-      "step": 18413
-    },
-    {
-      "epoch": 9.603129074315515,
-      "grad_norm": 1.3760944604873657,
-      "learning_rate": 8.200201005025126e-05,
-      "loss": 5.416,
-      "step": 18414
-    },
-    {
-      "epoch": 9.603650586701434,
-      "grad_norm": 1.530717134475708,
-      "learning_rate": 8.200100502512564e-05,
-      "loss": 5.3066,
-      "step": 18415
-    },
-    {
-      "epoch": 9.604172099087354,
-      "grad_norm": 1.4824564456939697,
-      "learning_rate": 8.2e-05,
-      "loss": 5.533,
-      "step": 18416
-    },
-    {
-      "epoch": 9.604693611473273,
-      "grad_norm": 1.4879517555236816,
-      "learning_rate": 8.199899497487438e-05,
-      "loss": 5.554,
-      "step": 18417
-    },
-    {
-      "epoch": 9.605215123859193,
-      "grad_norm": 1.4429326057434082,
-      "learning_rate": 8.199798994974874e-05,
-      "loss": 5.6076,
-      "step": 18418
-    },
-    {
-      "epoch": 9.60573663624511,
-      "grad_norm": 1.4914982318878174,
-      "learning_rate": 8.199698492462312e-05,
-      "loss": 5.4015,
-      "step": 18419
-    },
-    {
-      "epoch": 9.60625814863103,
-      "grad_norm": 1.3539097309112549,
-      "learning_rate": 8.199597989949749e-05,
-      "loss": 5.6999,
-      "step": 18420
-    },
-    {
-      "epoch": 9.60677966101695,
-      "grad_norm": 1.3722338676452637,
-      "learning_rate": 8.199497487437186e-05,
-      "loss": 5.8473,
-      "step": 18421
-    },
-    {
-      "epoch": 9.607301173402869,
-      "grad_norm": 1.3885574340820312,
-      "learning_rate": 8.199396984924623e-05,
-      "loss": 5.604,
-      "step": 18422
-    },
-    {
-      "epoch": 9.607822685788788,
-      "grad_norm": 1.6428301334381104,
-      "learning_rate": 8.19929648241206e-05,
-      "loss": 5.1716,
-      "step": 18423
-    },
-    {
-      "epoch": 9.608344198174706,
-      "grad_norm": 1.5168668031692505,
-      "learning_rate": 8.199195979899498e-05,
-      "loss": 5.3154,
-      "step": 18424
-    },
-    {
-      "epoch": 9.608865710560625,
-      "grad_norm": 1.4617899656295776,
-      "learning_rate": 8.199095477386936e-05,
-      "loss": 5.5677,
-      "step": 18425
-    },
-    {
-      "epoch": 9.609387222946545,
-      "grad_norm": 1.3344500064849854,
-      "learning_rate": 8.198994974874373e-05,
-      "loss": 5.6638,
-      "step": 18426
-    },
-    {
-      "epoch": 9.609908735332464,
-      "grad_norm": 1.3702986240386963,
-      "learning_rate": 8.19889447236181e-05,
-      "loss": 5.6939,
-      "step": 18427
-    },
-    {
-      "epoch": 9.610430247718384,
-      "grad_norm": 1.5982046127319336,
-      "learning_rate": 8.198793969849247e-05,
-      "loss": 5.3848,
-      "step": 18428
-    },
-    {
-      "epoch": 9.610951760104303,
-      "grad_norm": 1.4961978197097778,
-      "learning_rate": 8.198693467336683e-05,
-      "loss": 5.5576,
-      "step": 18429
-    },
-    {
-      "epoch": 9.611473272490223,
-      "grad_norm": 1.3434443473815918,
-      "learning_rate": 8.198592964824121e-05,
-      "loss": 5.6093,
-      "step": 18430
-    },
-    {
-      "epoch": 9.61199478487614,
-      "grad_norm": 1.3862298727035522,
-      "learning_rate": 8.198492462311557e-05,
-      "loss": 5.7293,
-      "step": 18431
-    },
-    {
-      "epoch": 9.61251629726206,
-      "grad_norm": 1.4626322984695435,
-      "learning_rate": 8.198391959798995e-05,
-      "loss": 5.4167,
-      "step": 18432
-    },
-    {
-      "epoch": 9.61303780964798,
-      "grad_norm": 1.5036150217056274,
-      "learning_rate": 8.198291457286432e-05,
-      "loss": 5.5649,
-      "step": 18433
-    },
-    {
-      "epoch": 9.613559322033899,
-      "grad_norm": 1.624660849571228,
-      "learning_rate": 8.19819095477387e-05,
-      "loss": 4.9833,
-      "step": 18434
-    },
-    {
-      "epoch": 9.614080834419818,
-      "grad_norm": 1.5296603441238403,
-      "learning_rate": 8.198090452261307e-05,
-      "loss": 5.1221,
-      "step": 18435
-    },
-    {
-      "epoch": 9.614602346805736,
-      "grad_norm": 1.4884387254714966,
-      "learning_rate": 8.197989949748745e-05,
-      "loss": 5.6654,
-      "step": 18436
-    },
-    {
-      "epoch": 9.615123859191655,
-      "grad_norm": 1.4879038333892822,
-      "learning_rate": 8.197889447236181e-05,
-      "loss": 5.8457,
-      "step": 18437
-    },
-    {
-      "epoch": 9.615645371577575,
-      "grad_norm": 1.6049697399139404,
-      "learning_rate": 8.197788944723619e-05,
-      "loss": 5.0425,
-      "step": 18438
-    },
-    {
-      "epoch": 9.616166883963494,
-      "grad_norm": 1.4730912446975708,
-      "learning_rate": 8.197688442211056e-05,
-      "loss": 5.6596,
-      "step": 18439
-    },
-    {
-      "epoch": 9.616688396349414,
-      "grad_norm": 1.6819064617156982,
-      "learning_rate": 8.197587939698493e-05,
-      "loss": 5.4894,
-      "step": 18440
-    },
-    {
-      "epoch": 9.617209908735333,
-      "grad_norm": 1.522387146949768,
-      "learning_rate": 8.19748743718593e-05,
-      "loss": 5.6538,
-      "step": 18441
-    },
-    {
-      "epoch": 9.617731421121253,
-      "grad_norm": 1.6732240915298462,
-      "learning_rate": 8.197386934673366e-05,
-      "loss": 5.3023,
-      "step": 18442
-    },
-    {
-      "epoch": 9.61825293350717,
-      "grad_norm": 1.3353803157806396,
-      "learning_rate": 8.197286432160804e-05,
-      "loss": 5.7024,
-      "step": 18443
-    },
-    {
-      "epoch": 9.61877444589309,
-      "grad_norm": 1.3304393291473389,
-      "learning_rate": 8.197185929648242e-05,
-      "loss": 5.4267,
-      "step": 18444
-    },
-    {
-      "epoch": 9.61929595827901,
-      "grad_norm": 1.5635309219360352,
-      "learning_rate": 8.19708542713568e-05,
-      "loss": 4.9325,
-      "step": 18445
-    },
-    {
-      "epoch": 9.619817470664929,
-      "grad_norm": 1.3761059045791626,
-      "learning_rate": 8.196984924623116e-05,
-      "loss": 5.5599,
-      "step": 18446
-    },
-    {
-      "epoch": 9.620338983050848,
-      "grad_norm": 1.47213613986969,
-      "learning_rate": 8.196884422110554e-05,
-      "loss": 4.7837,
-      "step": 18447
-    },
-    {
-      "epoch": 9.620860495436766,
-      "grad_norm": 1.3744595050811768,
-      "learning_rate": 8.19678391959799e-05,
-      "loss": 5.0223,
-      "step": 18448
-    },
-    {
-      "epoch": 9.621382007822685,
-      "grad_norm": 1.358615756034851,
-      "learning_rate": 8.196683417085428e-05,
-      "loss": 5.7578,
-      "step": 18449
-    },
-    {
-      "epoch": 9.621903520208605,
-      "grad_norm": 1.438184142112732,
-      "learning_rate": 8.196582914572864e-05,
-      "loss": 5.2886,
-      "step": 18450
-    },
-    {
-      "epoch": 9.622425032594524,
-      "grad_norm": 1.5022472143173218,
-      "learning_rate": 8.196482412060302e-05,
-      "loss": 5.7655,
-      "step": 18451
-    },
-    {
-      "epoch": 9.622946544980444,
-      "grad_norm": 1.4404679536819458,
-      "learning_rate": 8.196381909547739e-05,
-      "loss": 5.3263,
-      "step": 18452
-    },
-    {
-      "epoch": 9.623468057366363,
-      "grad_norm": 1.4668915271759033,
-      "learning_rate": 8.196281407035176e-05,
-      "loss": 5.3864,
-      "step": 18453
-    },
-    {
-      "epoch": 9.62398956975228,
-      "grad_norm": 1.4519809484481812,
-      "learning_rate": 8.196180904522613e-05,
-      "loss": 5.3109,
-      "step": 18454
-    },
-    {
-      "epoch": 9.6245110821382,
-      "grad_norm": 1.523984670639038,
-      "learning_rate": 8.19608040201005e-05,
-      "loss": 5.4361,
-      "step": 18455
-    },
-    {
-      "epoch": 9.62503259452412,
-      "grad_norm": 1.437371015548706,
-      "learning_rate": 8.195979899497488e-05,
-      "loss": 5.2901,
-      "step": 18456
-    },
-    {
-      "epoch": 9.62555410691004,
-      "grad_norm": 1.426194429397583,
-      "learning_rate": 8.195879396984925e-05,
-      "loss": 5.8952,
-      "step": 18457
-    },
-    {
-      "epoch": 9.626075619295959,
-      "grad_norm": 1.4064124822616577,
-      "learning_rate": 8.195778894472363e-05,
-      "loss": 5.749,
-      "step": 18458
-    },
-    {
-      "epoch": 9.626597131681878,
-      "grad_norm": 1.3884235620498657,
-      "learning_rate": 8.195678391959799e-05,
-      "loss": 5.3946,
-      "step": 18459
-    },
-    {
-      "epoch": 9.627118644067796,
-      "grad_norm": 1.3600341081619263,
-      "learning_rate": 8.195577889447237e-05,
-      "loss": 5.5987,
-      "step": 18460
-    },
-    {
-      "epoch": 9.627640156453715,
-      "grad_norm": 1.3391847610473633,
-      "learning_rate": 8.195477386934673e-05,
-      "loss": 5.5961,
-      "step": 18461
-    },
-    {
-      "epoch": 9.628161668839635,
-      "grad_norm": 1.533144474029541,
-      "learning_rate": 8.195376884422111e-05,
-      "loss": 5.2199,
-      "step": 18462
-    },
-    {
-      "epoch": 9.628683181225554,
-      "grad_norm": 1.5005134344100952,
-      "learning_rate": 8.195276381909547e-05,
-      "loss": 5.7109,
-      "step": 18463
-    },
-    {
-      "epoch": 9.629204693611474,
-      "grad_norm": 1.3814374208450317,
-      "learning_rate": 8.195175879396985e-05,
-      "loss": 5.4567,
-      "step": 18464
-    },
-    {
-      "epoch": 9.629726205997393,
-      "grad_norm": 1.4182343482971191,
-      "learning_rate": 8.195075376884423e-05,
-      "loss": 5.6996,
-      "step": 18465
-    },
-    {
-      "epoch": 9.63024771838331,
-      "grad_norm": 1.408696174621582,
-      "learning_rate": 8.194974874371861e-05,
-      "loss": 5.4403,
-      "step": 18466
-    },
-    {
-      "epoch": 9.63076923076923,
-      "grad_norm": 1.5200049877166748,
-      "learning_rate": 8.194874371859297e-05,
-      "loss": 5.5859,
-      "step": 18467
-    },
-    {
-      "epoch": 9.63129074315515,
-      "grad_norm": 1.5586684942245483,
-      "learning_rate": 8.194773869346734e-05,
-      "loss": 5.139,
-      "step": 18468
-    },
-    {
-      "epoch": 9.63181225554107,
-      "grad_norm": 1.422659158706665,
-      "learning_rate": 8.194673366834171e-05,
-      "loss": 5.3886,
-      "step": 18469
-    },
-    {
-      "epoch": 9.632333767926989,
-      "grad_norm": 1.4464585781097412,
-      "learning_rate": 8.194572864321608e-05,
-      "loss": 5.2581,
-      "step": 18470
-    },
-    {
-      "epoch": 9.632855280312908,
-      "grad_norm": 1.5193934440612793,
-      "learning_rate": 8.194472361809046e-05,
-      "loss": 5.7944,
-      "step": 18471
-    },
-    {
-      "epoch": 9.633376792698826,
-      "grad_norm": 1.4978387355804443,
-      "learning_rate": 8.194371859296482e-05,
-      "loss": 5.314,
-      "step": 18472
-    },
-    {
-      "epoch": 9.633898305084745,
-      "grad_norm": 1.4781062602996826,
-      "learning_rate": 8.19427135678392e-05,
-      "loss": 5.3965,
-      "step": 18473
-    },
-    {
-      "epoch": 9.634419817470665,
-      "grad_norm": 1.39667809009552,
-      "learning_rate": 8.194170854271356e-05,
-      "loss": 5.3114,
-      "step": 18474
-    },
-    {
-      "epoch": 9.634941329856584,
-      "grad_norm": 1.4217413663864136,
-      "learning_rate": 8.194070351758794e-05,
-      "loss": 5.2763,
-      "step": 18475
-    },
-    {
-      "epoch": 9.635462842242504,
-      "grad_norm": 1.406404972076416,
-      "learning_rate": 8.193969849246232e-05,
-      "loss": 5.5626,
-      "step": 18476
-    },
-    {
-      "epoch": 9.635984354628423,
-      "grad_norm": 1.4393328428268433,
-      "learning_rate": 8.19386934673367e-05,
-      "loss": 5.1243,
-      "step": 18477
-    },
-    {
-      "epoch": 9.63650586701434,
-      "grad_norm": 1.3487833738327026,
-      "learning_rate": 8.193768844221106e-05,
-      "loss": 5.4863,
-      "step": 18478
-    },
-    {
-      "epoch": 9.63702737940026,
-      "grad_norm": 1.4196925163269043,
-      "learning_rate": 8.193668341708544e-05,
-      "loss": 5.3122,
-      "step": 18479
-    },
-    {
-      "epoch": 9.63754889178618,
-      "grad_norm": 1.5375620126724243,
-      "learning_rate": 8.19356783919598e-05,
-      "loss": 5.024,
-      "step": 18480
-    },
-    {
-      "epoch": 9.6380704041721,
-      "grad_norm": 1.49607253074646,
-      "learning_rate": 8.193467336683417e-05,
-      "loss": 5.5294,
-      "step": 18481
-    },
-    {
-      "epoch": 9.638591916558019,
-      "grad_norm": 1.4931012392044067,
-      "learning_rate": 8.193366834170854e-05,
-      "loss": 5.086,
-      "step": 18482
-    },
-    {
-      "epoch": 9.639113428943938,
-      "grad_norm": 1.450100302696228,
-      "learning_rate": 8.193266331658291e-05,
-      "loss": 5.4947,
-      "step": 18483
-    },
-    {
-      "epoch": 9.639634941329856,
-      "grad_norm": 1.6052347421646118,
-      "learning_rate": 8.193165829145729e-05,
-      "loss": 5.2095,
-      "step": 18484
-    },
-    {
-      "epoch": 9.640156453715775,
-      "grad_norm": 1.4805704355239868,
-      "learning_rate": 8.193065326633166e-05,
-      "loss": 4.869,
-      "step": 18485
-    },
-    {
-      "epoch": 9.640677966101695,
-      "grad_norm": 1.335795521736145,
-      "learning_rate": 8.192964824120604e-05,
-      "loss": 5.673,
-      "step": 18486
-    },
-    {
-      "epoch": 9.641199478487614,
-      "grad_norm": 1.5425224304199219,
-      "learning_rate": 8.19286432160804e-05,
-      "loss": 5.2155,
-      "step": 18487
-    },
-    {
-      "epoch": 9.641720990873534,
-      "grad_norm": 1.4234529733657837,
-      "learning_rate": 8.192763819095478e-05,
-      "loss": 5.5917,
-      "step": 18488
-    },
-    {
-      "epoch": 9.642242503259453,
-      "grad_norm": 1.4052904844284058,
-      "learning_rate": 8.192663316582915e-05,
-      "loss": 5.7944,
-      "step": 18489
-    },
-    {
-      "epoch": 9.642764015645371,
-      "grad_norm": 1.516837239265442,
-      "learning_rate": 8.192562814070353e-05,
-      "loss": 5.4385,
-      "step": 18490
-    },
-    {
-      "epoch": 9.64328552803129,
-      "grad_norm": 1.4365055561065674,
-      "learning_rate": 8.192462311557789e-05,
-      "loss": 5.0849,
-      "step": 18491
-    },
-    {
-      "epoch": 9.64380704041721,
-      "grad_norm": 1.4088938236236572,
-      "learning_rate": 8.192361809045227e-05,
-      "loss": 5.4662,
-      "step": 18492
-    },
-    {
-      "epoch": 9.64432855280313,
-      "grad_norm": 1.4429852962493896,
-      "learning_rate": 8.192261306532663e-05,
-      "loss": 5.7832,
-      "step": 18493
-    },
-    {
-      "epoch": 9.644850065189049,
-      "grad_norm": 1.3857098817825317,
-      "learning_rate": 8.192160804020101e-05,
-      "loss": 5.206,
-      "step": 18494
-    },
-    {
-      "epoch": 9.645371577574968,
-      "grad_norm": 1.403621792793274,
-      "learning_rate": 8.192060301507538e-05,
-      "loss": 5.3351,
-      "step": 18495
-    },
-    {
-      "epoch": 9.645893089960886,
-      "grad_norm": 1.488184928894043,
-      "learning_rate": 8.191959798994975e-05,
-      "loss": 5.5051,
-      "step": 18496
-    },
-    {
-      "epoch": 9.646414602346805,
-      "grad_norm": 1.3625980615615845,
-      "learning_rate": 8.191859296482413e-05,
-      "loss": 5.4946,
-      "step": 18497
-    },
-    {
-      "epoch": 9.646936114732725,
-      "grad_norm": 1.4375823736190796,
-      "learning_rate": 8.19175879396985e-05,
-      "loss": 5.4029,
-      "step": 18498
-    },
-    {
-      "epoch": 9.647457627118644,
-      "grad_norm": 1.4194555282592773,
-      "learning_rate": 8.191658291457287e-05,
-      "loss": 5.2528,
-      "step": 18499
-    },
-    {
-      "epoch": 9.647979139504564,
-      "grad_norm": 1.4066858291625977,
-      "learning_rate": 8.191557788944724e-05,
-      "loss": 5.333,
-      "step": 18500
-    },
-    {
-      "epoch": 9.648500651890483,
-      "grad_norm": 1.5877779722213745,
-      "learning_rate": 8.191457286432161e-05,
-      "loss": 5.4645,
-      "step": 18501
-    },
-    {
-      "epoch": 9.649022164276401,
-      "grad_norm": 1.4791117906570435,
-      "learning_rate": 8.191356783919598e-05,
-      "loss": 5.476,
-      "step": 18502
-    },
-    {
-      "epoch": 9.64954367666232,
-      "grad_norm": 1.6221725940704346,
-      "learning_rate": 8.191256281407036e-05,
-      "loss": 5.2129,
-      "step": 18503
-    },
-    {
-      "epoch": 9.65006518904824,
-      "grad_norm": 1.4762927293777466,
-      "learning_rate": 8.191155778894472e-05,
-      "loss": 5.4215,
-      "step": 18504
-    },
-    {
-      "epoch": 9.65058670143416,
-      "grad_norm": 1.5427014827728271,
-      "learning_rate": 8.19105527638191e-05,
-      "loss": 5.3017,
-      "step": 18505
-    },
-    {
-      "epoch": 9.651108213820079,
-      "grad_norm": 1.366348147392273,
-      "learning_rate": 8.190954773869348e-05,
-      "loss": 5.4331,
-      "step": 18506
-    },
-    {
-      "epoch": 9.651629726205996,
-      "grad_norm": 1.5387725830078125,
-      "learning_rate": 8.190854271356785e-05,
-      "loss": 5.424,
-      "step": 18507
-    },
-    {
-      "epoch": 9.652151238591916,
-      "grad_norm": 1.43003249168396,
-      "learning_rate": 8.190753768844222e-05,
-      "loss": 5.4255,
-      "step": 18508
-    },
-    {
-      "epoch": 9.652672750977835,
-      "grad_norm": 1.439244270324707,
-      "learning_rate": 8.190653266331658e-05,
-      "loss": 5.5497,
-      "step": 18509
-    },
-    {
-      "epoch": 9.653194263363755,
-      "grad_norm": 1.3890118598937988,
-      "learning_rate": 8.190552763819096e-05,
-      "loss": 5.626,
-      "step": 18510
-    },
-    {
-      "epoch": 9.653715775749674,
-      "grad_norm": 1.3543609380722046,
-      "learning_rate": 8.190452261306533e-05,
-      "loss": 5.5644,
-      "step": 18511
-    },
-    {
-      "epoch": 9.654237288135594,
-      "grad_norm": 1.467510461807251,
-      "learning_rate": 8.19035175879397e-05,
-      "loss": 5.5853,
-      "step": 18512
-    },
-    {
-      "epoch": 9.654758800521513,
-      "grad_norm": 1.4935667514801025,
-      "learning_rate": 8.190251256281407e-05,
-      "loss": 5.0923,
-      "step": 18513
-    },
-    {
-      "epoch": 9.655280312907431,
-      "grad_norm": 1.4377920627593994,
-      "learning_rate": 8.190150753768845e-05,
-      "loss": 5.791,
-      "step": 18514
-    },
-    {
-      "epoch": 9.65580182529335,
-      "grad_norm": 1.4241738319396973,
-      "learning_rate": 8.190050251256281e-05,
-      "loss": 5.5682,
-      "step": 18515
-    },
-    {
-      "epoch": 9.65632333767927,
-      "grad_norm": 1.4348348379135132,
-      "learning_rate": 8.189949748743719e-05,
-      "loss": 5.6983,
-      "step": 18516
-    },
-    {
-      "epoch": 9.65684485006519,
-      "grad_norm": 1.3447948694229126,
-      "learning_rate": 8.189849246231157e-05,
-      "loss": 5.5444,
-      "step": 18517
-    },
-    {
-      "epoch": 9.657366362451109,
-      "grad_norm": 1.3739675283432007,
-      "learning_rate": 8.189748743718594e-05,
-      "loss": 5.695,
-      "step": 18518
-    },
-    {
-      "epoch": 9.657887874837026,
-      "grad_norm": 1.5312697887420654,
-      "learning_rate": 8.189648241206031e-05,
-      "loss": 5.1959,
-      "step": 18519
-    },
-    {
-      "epoch": 9.658409387222946,
-      "grad_norm": 1.495697259902954,
-      "learning_rate": 8.189547738693469e-05,
-      "loss": 5.1828,
-      "step": 18520
-    },
-    {
-      "epoch": 9.658930899608865,
-      "grad_norm": 1.5004196166992188,
-      "learning_rate": 8.189447236180905e-05,
-      "loss": 4.9661,
-      "step": 18521
-    },
-    {
-      "epoch": 9.659452411994785,
-      "grad_norm": 1.4237473011016846,
-      "learning_rate": 8.189346733668341e-05,
-      "loss": 5.5479,
-      "step": 18522
-    },
-    {
-      "epoch": 9.659973924380704,
-      "grad_norm": 1.601123332977295,
-      "learning_rate": 8.189246231155779e-05,
-      "loss": 5.3587,
-      "step": 18523
-    },
-    {
-      "epoch": 9.660495436766624,
-      "grad_norm": 1.5177417993545532,
-      "learning_rate": 8.189145728643216e-05,
-      "loss": 5.503,
-      "step": 18524
-    },
-    {
-      "epoch": 9.661016949152543,
-      "grad_norm": 1.3770344257354736,
-      "learning_rate": 8.189045226130653e-05,
-      "loss": 5.7034,
-      "step": 18525
-    },
-    {
-      "epoch": 9.661538461538461,
-      "grad_norm": 1.377465844154358,
-      "learning_rate": 8.188944723618091e-05,
-      "loss": 4.9787,
-      "step": 18526
-    },
-    {
-      "epoch": 9.66205997392438,
-      "grad_norm": 1.5656682252883911,
-      "learning_rate": 8.188844221105529e-05,
-      "loss": 5.4231,
-      "step": 18527
-    },
-    {
-      "epoch": 9.6625814863103,
-      "grad_norm": 1.5359641313552856,
-      "learning_rate": 8.188743718592965e-05,
-      "loss": 5.3672,
-      "step": 18528
-    },
-    {
-      "epoch": 9.66310299869622,
-      "grad_norm": 1.5246117115020752,
-      "learning_rate": 8.188643216080403e-05,
-      "loss": 5.4485,
-      "step": 18529
-    },
-    {
-      "epoch": 9.663624511082139,
-      "grad_norm": 1.4060789346694946,
-      "learning_rate": 8.18854271356784e-05,
-      "loss": 5.4822,
-      "step": 18530
-    },
-    {
-      "epoch": 9.664146023468057,
-      "grad_norm": 1.41658616065979,
-      "learning_rate": 8.188442211055277e-05,
-      "loss": 5.7391,
-      "step": 18531
-    },
-    {
-      "epoch": 9.664667535853976,
-      "grad_norm": 1.3729561567306519,
-      "learning_rate": 8.188341708542714e-05,
-      "loss": 5.6623,
-      "step": 18532
-    },
-    {
-      "epoch": 9.665189048239895,
-      "grad_norm": 1.4294042587280273,
-      "learning_rate": 8.188241206030152e-05,
-      "loss": 5.6419,
-      "step": 18533
-    },
-    {
-      "epoch": 9.665710560625815,
-      "grad_norm": 1.4671632051467896,
-      "learning_rate": 8.188140703517588e-05,
-      "loss": 5.3906,
-      "step": 18534
-    },
-    {
-      "epoch": 9.666232073011734,
-      "grad_norm": 1.4637659788131714,
-      "learning_rate": 8.188040201005024e-05,
-      "loss": 5.6943,
-      "step": 18535
-    },
-    {
-      "epoch": 9.666753585397654,
-      "grad_norm": 1.3566315174102783,
-      "learning_rate": 8.187939698492462e-05,
-      "loss": 5.6497,
-      "step": 18536
-    },
-    {
-      "epoch": 9.667275097783573,
-      "grad_norm": 1.4262971878051758,
-      "learning_rate": 8.1878391959799e-05,
-      "loss": 5.3377,
-      "step": 18537
-    },
-    {
-      "epoch": 9.667796610169491,
-      "grad_norm": 1.4709491729736328,
-      "learning_rate": 8.187738693467338e-05,
-      "loss": 5.2975,
-      "step": 18538
-    },
-    {
-      "epoch": 9.66831812255541,
-      "grad_norm": 1.3951104879379272,
-      "learning_rate": 8.187638190954774e-05,
-      "loss": 5.3456,
-      "step": 18539
-    },
-    {
-      "epoch": 9.66883963494133,
-      "grad_norm": 1.448448657989502,
-      "learning_rate": 8.187537688442212e-05,
-      "loss": 5.2095,
-      "step": 18540
-    },
-    {
-      "epoch": 9.66936114732725,
-      "grad_norm": 1.4019685983657837,
-      "learning_rate": 8.187437185929648e-05,
-      "loss": 5.2884,
-      "step": 18541
-    },
-    {
-      "epoch": 9.669882659713169,
-      "grad_norm": 1.421449065208435,
-      "learning_rate": 8.187336683417086e-05,
-      "loss": 5.7136,
-      "step": 18542
-    },
-    {
-      "epoch": 9.670404172099087,
-      "grad_norm": 1.3908705711364746,
-      "learning_rate": 8.187236180904523e-05,
-      "loss": 5.4797,
-      "step": 18543
-    },
-    {
-      "epoch": 9.670925684485006,
-      "grad_norm": 1.4989664554595947,
-      "learning_rate": 8.18713567839196e-05,
-      "loss": 5.4516,
-      "step": 18544
-    },
-    {
-      "epoch": 9.671447196870925,
-      "grad_norm": 1.4984453916549683,
-      "learning_rate": 8.187035175879397e-05,
-      "loss": 5.0046,
-      "step": 18545
-    },
-    {
-      "epoch": 9.671968709256845,
-      "grad_norm": 1.4291915893554688,
-      "learning_rate": 8.186934673366835e-05,
-      "loss": 5.673,
-      "step": 18546
-    },
-    {
-      "epoch": 9.672490221642764,
-      "grad_norm": 1.303781270980835,
-      "learning_rate": 8.186834170854272e-05,
-      "loss": 5.6747,
-      "step": 18547
-    },
-    {
-      "epoch": 9.673011734028684,
-      "grad_norm": 1.503077745437622,
-      "learning_rate": 8.186733668341709e-05,
-      "loss": 5.4237,
-      "step": 18548
-    },
-    {
-      "epoch": 9.673533246414602,
-      "grad_norm": 1.4869318008422852,
-      "learning_rate": 8.186633165829147e-05,
-      "loss": 5.1963,
-      "step": 18549
-    },
-    {
-      "epoch": 9.674054758800521,
-      "grad_norm": 1.4966760873794556,
-      "learning_rate": 8.186532663316583e-05,
-      "loss": 5.0418,
-      "step": 18550
-    },
-    {
-      "epoch": 9.67457627118644,
-      "grad_norm": 1.285454273223877,
-      "learning_rate": 8.186432160804021e-05,
-      "loss": 5.3238,
-      "step": 18551
-    },
-    {
-      "epoch": 9.67509778357236,
-      "grad_norm": 1.3157427310943604,
-      "learning_rate": 8.186331658291457e-05,
-      "loss": 5.6321,
-      "step": 18552
-    },
-    {
-      "epoch": 9.67561929595828,
-      "grad_norm": 1.5899189710617065,
-      "learning_rate": 8.186231155778895e-05,
-      "loss": 4.7137,
-      "step": 18553
-    },
-    {
-      "epoch": 9.676140808344199,
-      "grad_norm": 1.3239777088165283,
-      "learning_rate": 8.186130653266331e-05,
-      "loss": 4.8424,
-      "step": 18554
-    },
-    {
-      "epoch": 9.676662320730117,
-      "grad_norm": 1.2444803714752197,
-      "learning_rate": 8.186030150753769e-05,
-      "loss": 4.6692,
-      "step": 18555
-    },
-    {
-      "epoch": 9.677183833116036,
-      "grad_norm": 1.4459562301635742,
-      "learning_rate": 8.185929648241206e-05,
-      "loss": 5.3119,
-      "step": 18556
-    },
-    {
-      "epoch": 9.677705345501955,
-      "grad_norm": 1.5792566537857056,
-      "learning_rate": 8.185829145728643e-05,
-      "loss": 4.3991,
-      "step": 18557
-    },
-    {
-      "epoch": 9.678226857887875,
-      "grad_norm": 1.550134301185608,
-      "learning_rate": 8.185728643216081e-05,
-      "loss": 5.1472,
-      "step": 18558
-    },
-    {
-      "epoch": 9.678748370273794,
-      "grad_norm": 1.4792834520339966,
-      "learning_rate": 8.185628140703519e-05,
-      "loss": 5.2587,
-      "step": 18559
-    },
-    {
-      "epoch": 9.679269882659714,
-      "grad_norm": 1.254046082496643,
-      "learning_rate": 8.185527638190955e-05,
-      "loss": 5.5426,
-      "step": 18560
-    },
-    {
-      "epoch": 9.679791395045632,
-      "grad_norm": 1.423142671585083,
-      "learning_rate": 8.185427135678392e-05,
-      "loss": 5.6034,
-      "step": 18561
-    },
-    {
-      "epoch": 9.680312907431551,
-      "grad_norm": 1.693392276763916,
-      "learning_rate": 8.18532663316583e-05,
-      "loss": 5.4676,
-      "step": 18562
-    },
-    {
-      "epoch": 9.68083441981747,
-      "grad_norm": 1.3050262928009033,
-      "learning_rate": 8.185226130653266e-05,
-      "loss": 5.8323,
-      "step": 18563
-    },
-    {
-      "epoch": 9.68135593220339,
-      "grad_norm": 1.4795236587524414,
-      "learning_rate": 8.185125628140704e-05,
-      "loss": 5.2812,
-      "step": 18564
-    },
-    {
-      "epoch": 9.68187744458931,
-      "grad_norm": 1.5197834968566895,
-      "learning_rate": 8.18502512562814e-05,
-      "loss": 5.2166,
-      "step": 18565
-    },
-    {
-      "epoch": 9.682398956975229,
-      "grad_norm": 1.443449854850769,
-      "learning_rate": 8.184924623115578e-05,
-      "loss": 4.9839,
-      "step": 18566
-    },
-    {
-      "epoch": 9.682920469361147,
-      "grad_norm": 1.4738835096359253,
-      "learning_rate": 8.184824120603016e-05,
-      "loss": 5.5138,
-      "step": 18567
-    },
-    {
-      "epoch": 9.683441981747066,
-      "grad_norm": 1.6017212867736816,
-      "learning_rate": 8.184723618090454e-05,
-      "loss": 5.5475,
-      "step": 18568
-    },
-    {
-      "epoch": 9.683963494132986,
-      "grad_norm": 1.610708236694336,
-      "learning_rate": 8.18462311557789e-05,
-      "loss": 4.7084,
-      "step": 18569
-    },
-    {
-      "epoch": 9.684485006518905,
-      "grad_norm": 1.3483750820159912,
-      "learning_rate": 8.184522613065328e-05,
-      "loss": 5.642,
-      "step": 18570
-    },
-    {
-      "epoch": 9.685006518904824,
-      "grad_norm": 1.3887420892715454,
-      "learning_rate": 8.184422110552764e-05,
-      "loss": 5.6433,
-      "step": 18571
-    },
-    {
-      "epoch": 9.685528031290744,
-      "grad_norm": 1.4449222087860107,
-      "learning_rate": 8.184321608040202e-05,
-      "loss": 5.1799,
-      "step": 18572
-    },
-    {
-      "epoch": 9.686049543676662,
-      "grad_norm": 1.4323101043701172,
-      "learning_rate": 8.184221105527638e-05,
-      "loss": 5.4026,
-      "step": 18573
-    },
-    {
-      "epoch": 9.686571056062581,
-      "grad_norm": 1.399656057357788,
-      "learning_rate": 8.184120603015075e-05,
-      "loss": 5.8414,
-      "step": 18574
-    },
-    {
-      "epoch": 9.6870925684485,
-      "grad_norm": 1.3919442892074585,
-      "learning_rate": 8.184020100502513e-05,
-      "loss": 5.7636,
-      "step": 18575
-    },
-    {
-      "epoch": 9.68761408083442,
-      "grad_norm": 1.544398307800293,
-      "learning_rate": 8.183919597989949e-05,
-      "loss": 5.3126,
-      "step": 18576
-    },
-    {
-      "epoch": 9.68813559322034,
-      "grad_norm": 1.567663550376892,
-      "learning_rate": 8.183819095477387e-05,
-      "loss": 5.1066,
-      "step": 18577
-    },
-    {
-      "epoch": 9.688657105606259,
-      "grad_norm": 1.5866788625717163,
-      "learning_rate": 8.183718592964825e-05,
-      "loss": 5.6647,
-      "step": 18578
-    },
-    {
-      "epoch": 9.689178617992177,
-      "grad_norm": 1.4563877582550049,
-      "learning_rate": 8.183618090452262e-05,
-      "loss": 5.2073,
-      "step": 18579
-    },
-    {
-      "epoch": 9.689700130378096,
-      "grad_norm": 1.367419719696045,
-      "learning_rate": 8.183517587939699e-05,
-      "loss": 5.8106,
-      "step": 18580
-    },
-    {
-      "epoch": 9.690221642764016,
-      "grad_norm": 1.4419505596160889,
-      "learning_rate": 8.183417085427137e-05,
-      "loss": 5.3653,
-      "step": 18581
-    },
-    {
-      "epoch": 9.690743155149935,
-      "grad_norm": 1.481321096420288,
-      "learning_rate": 8.183316582914573e-05,
-      "loss": 5.2723,
-      "step": 18582
-    },
-    {
-      "epoch": 9.691264667535854,
-      "grad_norm": 1.4356770515441895,
-      "learning_rate": 8.183216080402011e-05,
-      "loss": 5.468,
-      "step": 18583
-    },
-    {
-      "epoch": 9.691786179921774,
-      "grad_norm": 1.494615912437439,
-      "learning_rate": 8.183115577889447e-05,
-      "loss": 5.5991,
-      "step": 18584
-    },
-    {
-      "epoch": 9.692307692307692,
-      "grad_norm": 1.3518173694610596,
-      "learning_rate": 8.183015075376885e-05,
-      "loss": 5.7269,
-      "step": 18585
-    },
-    {
-      "epoch": 9.692829204693611,
-      "grad_norm": 1.5440707206726074,
-      "learning_rate": 8.182914572864322e-05,
-      "loss": 5.1808,
-      "step": 18586
-    },
-    {
-      "epoch": 9.69335071707953,
-      "grad_norm": 1.5151623487472534,
-      "learning_rate": 8.182814070351759e-05,
-      "loss": 5.1726,
-      "step": 18587
-    },
-    {
-      "epoch": 9.69387222946545,
-      "grad_norm": 1.4492238759994507,
-      "learning_rate": 8.182713567839196e-05,
-      "loss": 5.5534,
-      "step": 18588
-    },
-    {
-      "epoch": 9.69439374185137,
-      "grad_norm": 1.3935397863388062,
-      "learning_rate": 8.182613065326634e-05,
-      "loss": 5.4794,
-      "step": 18589
-    },
-    {
-      "epoch": 9.694915254237289,
-      "grad_norm": 1.5803651809692383,
-      "learning_rate": 8.182512562814071e-05,
-      "loss": 5.0965,
-      "step": 18590
-    },
-    {
-      "epoch": 9.695436766623207,
-      "grad_norm": 1.482117772102356,
-      "learning_rate": 8.182412060301508e-05,
-      "loss": 5.7144,
-      "step": 18591
-    },
-    {
-      "epoch": 9.695958279009126,
-      "grad_norm": 1.5487957000732422,
-      "learning_rate": 8.182311557788946e-05,
-      "loss": 5.0869,
-      "step": 18592
-    },
-    {
-      "epoch": 9.696479791395046,
-      "grad_norm": 1.4621587991714478,
-      "learning_rate": 8.182211055276382e-05,
-      "loss": 5.4922,
-      "step": 18593
-    },
-    {
-      "epoch": 9.697001303780965,
-      "grad_norm": 1.4516290426254272,
-      "learning_rate": 8.18211055276382e-05,
-      "loss": 5.8441,
-      "step": 18594
-    },
-    {
-      "epoch": 9.697522816166884,
-      "grad_norm": 1.3454692363739014,
-      "learning_rate": 8.182010050251256e-05,
-      "loss": 5.6367,
-      "step": 18595
-    },
-    {
-      "epoch": 9.698044328552804,
-      "grad_norm": 1.4739863872528076,
-      "learning_rate": 8.181909547738694e-05,
-      "loss": 5.4223,
-      "step": 18596
-    },
-    {
-      "epoch": 9.698565840938722,
-      "grad_norm": 1.389207124710083,
-      "learning_rate": 8.18180904522613e-05,
-      "loss": 5.8802,
-      "step": 18597
-    },
-    {
-      "epoch": 9.699087353324641,
-      "grad_norm": 1.4222999811172485,
-      "learning_rate": 8.181708542713568e-05,
-      "loss": 5.4466,
-      "step": 18598
-    },
-    {
-      "epoch": 9.69960886571056,
-      "grad_norm": 1.351042628288269,
-      "learning_rate": 8.181608040201006e-05,
-      "loss": 5.7687,
-      "step": 18599
-    },
-    {
-      "epoch": 9.70013037809648,
-      "grad_norm": 1.5414776802062988,
-      "learning_rate": 8.181507537688444e-05,
-      "loss": 5.0474,
-      "step": 18600
-    },
-    {
-      "epoch": 9.7006518904824,
-      "grad_norm": 1.4343734979629517,
-      "learning_rate": 8.18140703517588e-05,
-      "loss": 5.4148,
-      "step": 18601
-    },
-    {
-      "epoch": 9.701173402868317,
-      "grad_norm": 1.4445431232452393,
-      "learning_rate": 8.181306532663317e-05,
-      "loss": 5.771,
-      "step": 18602
-    },
-    {
-      "epoch": 9.701694915254237,
-      "grad_norm": 1.6484830379486084,
-      "learning_rate": 8.181206030150754e-05,
-      "loss": 4.5569,
-      "step": 18603
-    },
-    {
-      "epoch": 9.702216427640156,
-      "grad_norm": 1.4708679914474487,
-      "learning_rate": 8.181105527638191e-05,
-      "loss": 5.6621,
-      "step": 18604
-    },
-    {
-      "epoch": 9.702737940026076,
-      "grad_norm": 1.5130834579467773,
-      "learning_rate": 8.181005025125629e-05,
-      "loss": 5.1753,
-      "step": 18605
-    },
-    {
-      "epoch": 9.703259452411995,
-      "grad_norm": 1.4642672538757324,
-      "learning_rate": 8.180904522613065e-05,
-      "loss": 5.7014,
-      "step": 18606
-    },
-    {
-      "epoch": 9.703780964797915,
-      "grad_norm": 1.4952397346496582,
-      "learning_rate": 8.180804020100503e-05,
-      "loss": 5.2089,
-      "step": 18607
-    },
-    {
-      "epoch": 9.704302477183834,
-      "grad_norm": 1.5272061824798584,
-      "learning_rate": 8.180703517587939e-05,
-      "loss": 5.4861,
-      "step": 18608
-    },
-    {
-      "epoch": 9.704823989569752,
-      "grad_norm": 1.4700024127960205,
-      "learning_rate": 8.180603015075377e-05,
-      "loss": 5.7819,
-      "step": 18609
-    },
-    {
-      "epoch": 9.705345501955671,
-      "grad_norm": 1.366637110710144,
-      "learning_rate": 8.180502512562815e-05,
-      "loss": 5.4501,
-      "step": 18610
-    },
-    {
-      "epoch": 9.70586701434159,
-      "grad_norm": 1.4005943536758423,
-      "learning_rate": 8.180402010050253e-05,
-      "loss": 5.7976,
-      "step": 18611
-    },
-    {
-      "epoch": 9.70638852672751,
-      "grad_norm": 1.403409481048584,
-      "learning_rate": 8.180301507537689e-05,
-      "loss": 4.8107,
-      "step": 18612
-    },
-    {
-      "epoch": 9.70691003911343,
-      "grad_norm": 1.4385838508605957,
-      "learning_rate": 8.180201005025127e-05,
-      "loss": 4.8594,
-      "step": 18613
-    },
-    {
-      "epoch": 9.707431551499347,
-      "grad_norm": 1.5059614181518555,
-      "learning_rate": 8.180100502512563e-05,
-      "loss": 5.3527,
-      "step": 18614
-    },
-    {
-      "epoch": 9.707953063885267,
-      "grad_norm": 1.3317420482635498,
-      "learning_rate": 8.18e-05,
-      "loss": 5.899,
-      "step": 18615
-    },
-    {
-      "epoch": 9.708474576271186,
-      "grad_norm": 1.4835461378097534,
-      "learning_rate": 8.179899497487437e-05,
-      "loss": 5.8567,
-      "step": 18616
-    },
-    {
-      "epoch": 9.708996088657106,
-      "grad_norm": 1.4382926225662231,
-      "learning_rate": 8.179798994974874e-05,
-      "loss": 5.486,
-      "step": 18617
-    },
-    {
-      "epoch": 9.709517601043025,
-      "grad_norm": 1.438188076019287,
-      "learning_rate": 8.179698492462312e-05,
-      "loss": 4.6588,
-      "step": 18618
-    },
-    {
-      "epoch": 9.710039113428945,
-      "grad_norm": 1.7021583318710327,
-      "learning_rate": 8.17959798994975e-05,
-      "loss": 5.2937,
-      "step": 18619
-    },
-    {
-      "epoch": 9.710560625814864,
-      "grad_norm": 1.44141685962677,
-      "learning_rate": 8.179497487437187e-05,
-      "loss": 5.4536,
-      "step": 18620
-    },
-    {
-      "epoch": 9.711082138200782,
-      "grad_norm": 1.5291990041732788,
-      "learning_rate": 8.179396984924624e-05,
-      "loss": 5.2909,
-      "step": 18621
-    },
-    {
-      "epoch": 9.711603650586701,
-      "grad_norm": 1.3769620656967163,
-      "learning_rate": 8.179296482412061e-05,
-      "loss": 5.5902,
-      "step": 18622
-    },
-    {
-      "epoch": 9.71212516297262,
-      "grad_norm": 1.3233537673950195,
-      "learning_rate": 8.179195979899498e-05,
-      "loss": 5.7766,
-      "step": 18623
-    },
-    {
-      "epoch": 9.71264667535854,
-      "grad_norm": 1.6375271081924438,
-      "learning_rate": 8.179095477386936e-05,
-      "loss": 5.3433,
-      "step": 18624
-    },
-    {
-      "epoch": 9.71316818774446,
-      "grad_norm": 1.5172581672668457,
-      "learning_rate": 8.178994974874372e-05,
-      "loss": 5.039,
-      "step": 18625
-    },
-    {
-      "epoch": 9.713689700130377,
-      "grad_norm": 1.3950598239898682,
-      "learning_rate": 8.17889447236181e-05,
-      "loss": 5.5522,
-      "step": 18626
-    },
-    {
-      "epoch": 9.714211212516297,
-      "grad_norm": 1.4036316871643066,
-      "learning_rate": 8.178793969849246e-05,
-      "loss": 5.3894,
-      "step": 18627
-    },
-    {
-      "epoch": 9.714732724902216,
-      "grad_norm": 1.308119773864746,
-      "learning_rate": 8.178693467336683e-05,
-      "loss": 5.75,
-      "step": 18628
-    },
-    {
-      "epoch": 9.715254237288136,
-      "grad_norm": 1.3728142976760864,
-      "learning_rate": 8.17859296482412e-05,
-      "loss": 5.5433,
-      "step": 18629
-    },
-    {
-      "epoch": 9.715775749674055,
-      "grad_norm": 1.3861063718795776,
-      "learning_rate": 8.178492462311558e-05,
-      "loss": 5.6978,
-      "step": 18630
-    },
-    {
-      "epoch": 9.716297262059975,
-      "grad_norm": 1.3265024423599243,
-      "learning_rate": 8.178391959798996e-05,
-      "loss": 5.6679,
-      "step": 18631
-    },
-    {
-      "epoch": 9.716818774445892,
-      "grad_norm": 1.6424033641815186,
-      "learning_rate": 8.178291457286432e-05,
-      "loss": 5.1796,
-      "step": 18632
-    },
-    {
-      "epoch": 9.717340286831812,
-      "grad_norm": 1.4027554988861084,
-      "learning_rate": 8.17819095477387e-05,
-      "loss": 5.9367,
-      "step": 18633
-    },
-    {
-      "epoch": 9.717861799217731,
-      "grad_norm": 1.5086455345153809,
-      "learning_rate": 8.178090452261307e-05,
-      "loss": 5.4798,
-      "step": 18634
-    },
-    {
-      "epoch": 9.71838331160365,
-      "grad_norm": 1.455838680267334,
-      "learning_rate": 8.177989949748744e-05,
-      "loss": 5.5253,
-      "step": 18635
-    },
-    {
-      "epoch": 9.71890482398957,
-      "grad_norm": 1.4250065088272095,
-      "learning_rate": 8.177889447236181e-05,
-      "loss": 5.3475,
-      "step": 18636
-    },
-    {
-      "epoch": 9.71942633637549,
-      "grad_norm": 1.4521886110305786,
-      "learning_rate": 8.177788944723619e-05,
-      "loss": 5.3148,
-      "step": 18637
-    },
-    {
-      "epoch": 9.719947848761407,
-      "grad_norm": 1.432077407836914,
-      "learning_rate": 8.177688442211055e-05,
-      "loss": 5.6794,
-      "step": 18638
-    },
-    {
-      "epoch": 9.720469361147327,
-      "grad_norm": 1.4427014589309692,
-      "learning_rate": 8.177587939698493e-05,
-      "loss": 5.6651,
-      "step": 18639
-    },
-    {
-      "epoch": 9.720990873533246,
-      "grad_norm": 1.474668025970459,
-      "learning_rate": 8.17748743718593e-05,
-      "loss": 5.6559,
-      "step": 18640
-    },
-    {
-      "epoch": 9.721512385919166,
-      "grad_norm": 1.5356340408325195,
-      "learning_rate": 8.177386934673367e-05,
-      "loss": 5.1265,
-      "step": 18641
-    },
-    {
-      "epoch": 9.722033898305085,
-      "grad_norm": 1.472482681274414,
-      "learning_rate": 8.177286432160805e-05,
-      "loss": 5.8024,
-      "step": 18642
-    },
-    {
-      "epoch": 9.722555410691005,
-      "grad_norm": 1.4466021060943604,
-      "learning_rate": 8.177185929648241e-05,
-      "loss": 5.2977,
-      "step": 18643
-    },
-    {
-      "epoch": 9.723076923076922,
-      "grad_norm": 1.3532615900039673,
-      "learning_rate": 8.177085427135679e-05,
-      "loss": 5.6708,
-      "step": 18644
-    },
-    {
-      "epoch": 9.723598435462842,
-      "grad_norm": 1.4758414030075073,
-      "learning_rate": 8.176984924623115e-05,
-      "loss": 5.4848,
-      "step": 18645
-    },
-    {
-      "epoch": 9.724119947848761,
-      "grad_norm": 1.3391772508621216,
-      "learning_rate": 8.176884422110553e-05,
-      "loss": 5.5246,
-      "step": 18646
-    },
-    {
-      "epoch": 9.72464146023468,
-      "grad_norm": 1.4163191318511963,
-      "learning_rate": 8.17678391959799e-05,
-      "loss": 5.3795,
-      "step": 18647
-    },
-    {
-      "epoch": 9.7251629726206,
-      "grad_norm": 1.4137624502182007,
-      "learning_rate": 8.176683417085427e-05,
-      "loss": 5.4173,
-      "step": 18648
-    },
-    {
-      "epoch": 9.72568448500652,
-      "grad_norm": 1.5122489929199219,
-      "learning_rate": 8.176582914572864e-05,
-      "loss": 5.2407,
-      "step": 18649
-    },
-    {
-      "epoch": 9.726205997392437,
-      "grad_norm": 1.5166511535644531,
-      "learning_rate": 8.176482412060302e-05,
-      "loss": 5.7698,
-      "step": 18650
-    },
-    {
-      "epoch": 9.726727509778357,
-      "grad_norm": 1.531610131263733,
-      "learning_rate": 8.17638190954774e-05,
-      "loss": 5.683,
-      "step": 18651
-    },
-    {
-      "epoch": 9.727249022164276,
-      "grad_norm": 1.3876957893371582,
-      "learning_rate": 8.176281407035177e-05,
-      "loss": 5.3044,
-      "step": 18652
-    },
-    {
-      "epoch": 9.727770534550196,
-      "grad_norm": 1.6202011108398438,
-      "learning_rate": 8.176180904522614e-05,
-      "loss": 5.0108,
-      "step": 18653
-    },
-    {
-      "epoch": 9.728292046936115,
-      "grad_norm": 1.3470455408096313,
-      "learning_rate": 8.17608040201005e-05,
-      "loss": 5.4393,
-      "step": 18654
-    },
-    {
-      "epoch": 9.728813559322035,
-      "grad_norm": 1.4794467687606812,
-      "learning_rate": 8.175979899497488e-05,
-      "loss": 5.0472,
-      "step": 18655
-    },
-    {
-      "epoch": 9.729335071707952,
-      "grad_norm": 1.507926344871521,
-      "learning_rate": 8.175879396984924e-05,
-      "loss": 5.1952,
-      "step": 18656
-    },
-    {
-      "epoch": 9.729856584093872,
-      "grad_norm": 1.3959763050079346,
-      "learning_rate": 8.175778894472362e-05,
-      "loss": 5.9126,
-      "step": 18657
-    },
-    {
-      "epoch": 9.730378096479791,
-      "grad_norm": 1.419403314590454,
-      "learning_rate": 8.175678391959799e-05,
-      "loss": 5.8554,
-      "step": 18658
-    },
-    {
-      "epoch": 9.73089960886571,
-      "grad_norm": 1.4629031419754028,
-      "learning_rate": 8.175577889447236e-05,
-      "loss": 5.2055,
-      "step": 18659
-    },
-    {
-      "epoch": 9.73142112125163,
-      "grad_norm": 1.4522790908813477,
-      "learning_rate": 8.175477386934674e-05,
-      "loss": 4.9952,
-      "step": 18660
-    },
-    {
-      "epoch": 9.73194263363755,
-      "grad_norm": 1.4359452724456787,
-      "learning_rate": 8.175376884422112e-05,
-      "loss": 4.9995,
-      "step": 18661
-    },
-    {
-      "epoch": 9.732464146023467,
-      "grad_norm": 1.4343265295028687,
-      "learning_rate": 8.175276381909548e-05,
-      "loss": 5.6549,
-      "step": 18662
-    },
-    {
-      "epoch": 9.732985658409387,
-      "grad_norm": 1.41818106174469,
-      "learning_rate": 8.175175879396986e-05,
-      "loss": 5.5114,
-      "step": 18663
-    },
-    {
-      "epoch": 9.733507170795306,
-      "grad_norm": 1.5784001350402832,
-      "learning_rate": 8.175075376884423e-05,
-      "loss": 5.1904,
-      "step": 18664
-    },
-    {
-      "epoch": 9.734028683181226,
-      "grad_norm": 1.4167181253433228,
-      "learning_rate": 8.17497487437186e-05,
-      "loss": 5.7491,
-      "step": 18665
-    },
-    {
-      "epoch": 9.734550195567145,
-      "grad_norm": 1.412499189376831,
-      "learning_rate": 8.174874371859297e-05,
-      "loss": 5.0275,
-      "step": 18666
-    },
-    {
-      "epoch": 9.735071707953065,
-      "grad_norm": 1.4488942623138428,
-      "learning_rate": 8.174773869346734e-05,
-      "loss": 5.4696,
-      "step": 18667
-    },
-    {
-      "epoch": 9.735593220338982,
-      "grad_norm": 1.3687667846679688,
-      "learning_rate": 8.174673366834171e-05,
-      "loss": 4.674,
-      "step": 18668
-    },
-    {
-      "epoch": 9.736114732724902,
-      "grad_norm": 1.4834998846054077,
-      "learning_rate": 8.174572864321607e-05,
-      "loss": 5.3708,
-      "step": 18669
-    },
-    {
-      "epoch": 9.736636245110821,
-      "grad_norm": 1.5331450700759888,
-      "learning_rate": 8.174472361809045e-05,
-      "loss": 4.8616,
-      "step": 18670
-    },
-    {
-      "epoch": 9.73715775749674,
-      "grad_norm": 1.5534929037094116,
-      "learning_rate": 8.174371859296483e-05,
-      "loss": 5.228,
-      "step": 18671
-    },
-    {
-      "epoch": 9.73767926988266,
-      "grad_norm": 1.3710709810256958,
-      "learning_rate": 8.174271356783921e-05,
-      "loss": 6.1562,
-      "step": 18672
-    },
-    {
-      "epoch": 9.73820078226858,
-      "grad_norm": 1.4513472318649292,
-      "learning_rate": 8.174170854271357e-05,
-      "loss": 5.6704,
-      "step": 18673
-    },
-    {
-      "epoch": 9.738722294654497,
-      "grad_norm": 1.4319325685501099,
-      "learning_rate": 8.174070351758795e-05,
-      "loss": 5.7313,
-      "step": 18674
-    },
-    {
-      "epoch": 9.739243807040417,
-      "grad_norm": 1.5158483982086182,
-      "learning_rate": 8.173969849246231e-05,
-      "loss": 5.4816,
-      "step": 18675
-    },
-    {
-      "epoch": 9.739765319426336,
-      "grad_norm": 1.444833755493164,
-      "learning_rate": 8.173869346733669e-05,
-      "loss": 5.3023,
-      "step": 18676
-    },
-    {
-      "epoch": 9.740286831812256,
-      "grad_norm": 1.5773440599441528,
-      "learning_rate": 8.173768844221106e-05,
-      "loss": 5.803,
-      "step": 18677
-    },
-    {
-      "epoch": 9.740808344198175,
-      "grad_norm": 1.4418303966522217,
-      "learning_rate": 8.173668341708543e-05,
-      "loss": 5.7438,
-      "step": 18678
-    },
-    {
-      "epoch": 9.741329856584095,
-      "grad_norm": 1.4371198415756226,
-      "learning_rate": 8.17356783919598e-05,
-      "loss": 5.644,
-      "step": 18679
-    },
-    {
-      "epoch": 9.741851368970012,
-      "grad_norm": 1.6133707761764526,
-      "learning_rate": 8.173467336683418e-05,
-      "loss": 5.1333,
-      "step": 18680
-    },
-    {
-      "epoch": 9.742372881355932,
-      "grad_norm": 1.56196129322052,
-      "learning_rate": 8.173366834170855e-05,
-      "loss": 5.6199,
-      "step": 18681
-    },
-    {
-      "epoch": 9.742894393741851,
-      "grad_norm": 1.460421085357666,
-      "learning_rate": 8.173266331658292e-05,
-      "loss": 5.7842,
-      "step": 18682
-    },
-    {
-      "epoch": 9.74341590612777,
-      "grad_norm": 1.53962242603302,
-      "learning_rate": 8.17316582914573e-05,
-      "loss": 5.7664,
-      "step": 18683
-    },
-    {
-      "epoch": 9.74393741851369,
-      "grad_norm": 1.4369391202926636,
-      "learning_rate": 8.173065326633166e-05,
-      "loss": 5.3777,
-      "step": 18684
-    },
-    {
-      "epoch": 9.74445893089961,
-      "grad_norm": 1.4716638326644897,
-      "learning_rate": 8.172964824120604e-05,
-      "loss": 5.6756,
-      "step": 18685
-    },
-    {
-      "epoch": 9.744980443285527,
-      "grad_norm": 1.4272010326385498,
-      "learning_rate": 8.17286432160804e-05,
-      "loss": 5.5638,
-      "step": 18686
-    },
-    {
-      "epoch": 9.745501955671447,
-      "grad_norm": 1.4544620513916016,
-      "learning_rate": 8.172763819095478e-05,
-      "loss": 5.4273,
-      "step": 18687
-    },
-    {
-      "epoch": 9.746023468057366,
-      "grad_norm": 1.345214605331421,
-      "learning_rate": 8.172663316582914e-05,
-      "loss": 5.7922,
-      "step": 18688
-    },
-    {
-      "epoch": 9.746544980443286,
-      "grad_norm": 1.5575881004333496,
-      "learning_rate": 8.172562814070352e-05,
-      "loss": 4.8657,
-      "step": 18689
-    },
-    {
-      "epoch": 9.747066492829205,
-      "grad_norm": 1.45172917842865,
-      "learning_rate": 8.172462311557789e-05,
-      "loss": 5.379,
-      "step": 18690
-    },
-    {
-      "epoch": 9.747588005215125,
-      "grad_norm": 1.420206069946289,
-      "learning_rate": 8.172361809045226e-05,
-      "loss": 5.6552,
-      "step": 18691
-    },
-    {
-      "epoch": 9.748109517601042,
-      "grad_norm": 1.3357425928115845,
-      "learning_rate": 8.172261306532664e-05,
-      "loss": 5.5288,
-      "step": 18692
-    },
-    {
-      "epoch": 9.748631029986962,
-      "grad_norm": 1.6237003803253174,
-      "learning_rate": 8.172160804020102e-05,
-      "loss": 5.0656,
-      "step": 18693
-    },
-    {
-      "epoch": 9.749152542372881,
-      "grad_norm": 1.4047448635101318,
-      "learning_rate": 8.172060301507538e-05,
-      "loss": 5.4136,
-      "step": 18694
-    },
-    {
-      "epoch": 9.7496740547588,
-      "grad_norm": 1.459893822669983,
-      "learning_rate": 8.171959798994975e-05,
-      "loss": 5.2996,
-      "step": 18695
-    },
-    {
-      "epoch": 9.75019556714472,
-      "grad_norm": 1.427721381187439,
-      "learning_rate": 8.171859296482413e-05,
-      "loss": 5.6767,
-      "step": 18696
-    },
-    {
-      "epoch": 9.750717079530638,
-      "grad_norm": 1.4373003244400024,
-      "learning_rate": 8.171758793969849e-05,
-      "loss": 5.4612,
-      "step": 18697
-    },
-    {
-      "epoch": 9.751238591916557,
-      "grad_norm": 1.4318232536315918,
-      "learning_rate": 8.171658291457287e-05,
-      "loss": 5.4256,
-      "step": 18698
-    },
-    {
-      "epoch": 9.751760104302477,
-      "grad_norm": 1.4988802671432495,
-      "learning_rate": 8.171557788944723e-05,
-      "loss": 5.6443,
-      "step": 18699
-    },
-    {
-      "epoch": 9.752281616688396,
-      "grad_norm": 1.4381216764450073,
-      "learning_rate": 8.171457286432161e-05,
-      "loss": 5.6104,
-      "step": 18700
-    },
-    {
-      "epoch": 9.752803129074316,
-      "grad_norm": 1.4902812242507935,
-      "learning_rate": 8.171356783919599e-05,
-      "loss": 5.2999,
-      "step": 18701
-    },
-    {
-      "epoch": 9.753324641460235,
-      "grad_norm": 1.4792976379394531,
-      "learning_rate": 8.171256281407037e-05,
-      "loss": 5.7409,
-      "step": 18702
-    },
-    {
-      "epoch": 9.753846153846155,
-      "grad_norm": 1.424703598022461,
-      "learning_rate": 8.171155778894473e-05,
-      "loss": 4.8409,
-      "step": 18703
-    },
-    {
-      "epoch": 9.754367666232072,
-      "grad_norm": 1.395787239074707,
-      "learning_rate": 8.171055276381911e-05,
-      "loss": 5.4305,
-      "step": 18704
-    },
-    {
-      "epoch": 9.754889178617992,
-      "grad_norm": 1.4718749523162842,
-      "learning_rate": 8.170954773869347e-05,
-      "loss": 5.609,
-      "step": 18705
-    },
-    {
-      "epoch": 9.755410691003911,
-      "grad_norm": 1.542810082435608,
-      "learning_rate": 8.170854271356785e-05,
-      "loss": 5.1836,
-      "step": 18706
-    },
-    {
-      "epoch": 9.75593220338983,
-      "grad_norm": 1.3462250232696533,
-      "learning_rate": 8.170753768844221e-05,
-      "loss": 5.1766,
-      "step": 18707
-    },
-    {
-      "epoch": 9.75645371577575,
-      "grad_norm": 1.4435293674468994,
-      "learning_rate": 8.170653266331658e-05,
-      "loss": 5.6317,
-      "step": 18708
-    },
-    {
-      "epoch": 9.756975228161668,
-      "grad_norm": 1.445448875427246,
-      "learning_rate": 8.170552763819096e-05,
-      "loss": 5.2633,
-      "step": 18709
-    },
-    {
-      "epoch": 9.757496740547587,
-      "grad_norm": 1.5255001783370972,
-      "learning_rate": 8.170452261306532e-05,
-      "loss": 5.0939,
-      "step": 18710
-    },
-    {
-      "epoch": 9.758018252933507,
-      "grad_norm": 1.5364750623703003,
-      "learning_rate": 8.17035175879397e-05,
-      "loss": 5.3467,
-      "step": 18711
-    },
-    {
-      "epoch": 9.758539765319426,
-      "grad_norm": 1.48344886302948,
-      "learning_rate": 8.170251256281408e-05,
-      "loss": 5.7316,
-      "step": 18712
-    },
-    {
-      "epoch": 9.759061277705346,
-      "grad_norm": 1.4243154525756836,
-      "learning_rate": 8.170150753768845e-05,
-      "loss": 5.3532,
-      "step": 18713
-    },
-    {
-      "epoch": 9.759582790091265,
-      "grad_norm": 1.5385390520095825,
-      "learning_rate": 8.170050251256282e-05,
-      "loss": 5.6724,
-      "step": 18714
-    },
-    {
-      "epoch": 9.760104302477185,
-      "grad_norm": 1.5950798988342285,
-      "learning_rate": 8.16994974874372e-05,
-      "loss": 5.3799,
-      "step": 18715
-    },
-    {
-      "epoch": 9.760625814863102,
-      "grad_norm": 1.4778952598571777,
-      "learning_rate": 8.169849246231156e-05,
-      "loss": 5.1686,
-      "step": 18716
-    },
-    {
-      "epoch": 9.761147327249022,
-      "grad_norm": 1.3493316173553467,
-      "learning_rate": 8.169748743718594e-05,
-      "loss": 5.8916,
-      "step": 18717
-    },
-    {
-      "epoch": 9.761668839634941,
-      "grad_norm": 1.4093279838562012,
-      "learning_rate": 8.16964824120603e-05,
-      "loss": 5.5277,
-      "step": 18718
-    },
-    {
-      "epoch": 9.76219035202086,
-      "grad_norm": 1.5444834232330322,
-      "learning_rate": 8.169547738693468e-05,
-      "loss": 5.654,
-      "step": 18719
-    },
-    {
-      "epoch": 9.76271186440678,
-      "grad_norm": 1.4080978631973267,
-      "learning_rate": 8.169447236180904e-05,
-      "loss": 5.4278,
-      "step": 18720
-    },
-    {
-      "epoch": 9.763233376792698,
-      "grad_norm": 1.491479516029358,
-      "learning_rate": 8.169346733668342e-05,
-      "loss": 5.5259,
-      "step": 18721
-    },
-    {
-      "epoch": 9.763754889178617,
-      "grad_norm": 1.353216528892517,
-      "learning_rate": 8.16924623115578e-05,
-      "loss": 5.6442,
-      "step": 18722
-    },
-    {
-      "epoch": 9.764276401564537,
-      "grad_norm": 1.4683020114898682,
-      "learning_rate": 8.169145728643216e-05,
-      "loss": 5.6103,
-      "step": 18723
-    },
-    {
-      "epoch": 9.764797913950456,
-      "grad_norm": 1.49290931224823,
-      "learning_rate": 8.169045226130654e-05,
-      "loss": 4.9945,
-      "step": 18724
-    },
-    {
-      "epoch": 9.765319426336376,
-      "grad_norm": 1.505211353302002,
-      "learning_rate": 8.16894472361809e-05,
-      "loss": 5.2664,
-      "step": 18725
-    },
-    {
-      "epoch": 9.765840938722295,
-      "grad_norm": 1.4182904958724976,
-      "learning_rate": 8.168844221105528e-05,
-      "loss": 5.5644,
-      "step": 18726
-    },
-    {
-      "epoch": 9.766362451108213,
-      "grad_norm": 1.3857450485229492,
-      "learning_rate": 8.168743718592965e-05,
-      "loss": 5.532,
-      "step": 18727
-    },
-    {
-      "epoch": 9.766883963494132,
-      "grad_norm": 1.423683762550354,
-      "learning_rate": 8.168643216080403e-05,
-      "loss": 5.4862,
-      "step": 18728
-    },
-    {
-      "epoch": 9.767405475880052,
-      "grad_norm": 1.4100990295410156,
-      "learning_rate": 8.168542713567839e-05,
-      "loss": 5.5865,
-      "step": 18729
-    },
-    {
-      "epoch": 9.767926988265971,
-      "grad_norm": 1.3633017539978027,
-      "learning_rate": 8.168442211055277e-05,
-      "loss": 5.6582,
-      "step": 18730
-    },
-    {
-      "epoch": 9.76844850065189,
-      "grad_norm": 1.432947039604187,
-      "learning_rate": 8.168341708542713e-05,
-      "loss": 5.3155,
-      "step": 18731
-    },
-    {
-      "epoch": 9.76897001303781,
-      "grad_norm": 1.4410195350646973,
-      "learning_rate": 8.168241206030151e-05,
-      "loss": 4.972,
-      "step": 18732
-    },
-    {
-      "epoch": 9.769491525423728,
-      "grad_norm": 1.383204698562622,
-      "learning_rate": 8.168140703517589e-05,
-      "loss": 5.7137,
-      "step": 18733
-    },
-    {
-      "epoch": 9.770013037809647,
-      "grad_norm": 1.5261754989624023,
-      "learning_rate": 8.168040201005025e-05,
-      "loss": 5.5841,
-      "step": 18734
-    },
-    {
-      "epoch": 9.770534550195567,
-      "grad_norm": 1.4425770044326782,
-      "learning_rate": 8.167939698492463e-05,
-      "loss": 5.2751,
-      "step": 18735
-    },
-    {
-      "epoch": 9.771056062581486,
-      "grad_norm": 1.4786243438720703,
-      "learning_rate": 8.1678391959799e-05,
-      "loss": 5.4452,
-      "step": 18736
-    },
-    {
-      "epoch": 9.771577574967406,
-      "grad_norm": 1.4356626272201538,
-      "learning_rate": 8.167738693467337e-05,
-      "loss": 5.629,
-      "step": 18737
-    },
-    {
-      "epoch": 9.772099087353325,
-      "grad_norm": 1.450703740119934,
-      "learning_rate": 8.167638190954774e-05,
-      "loss": 4.8787,
-      "step": 18738
-    },
-    {
-      "epoch": 9.772620599739243,
-      "grad_norm": 1.58195161819458,
-      "learning_rate": 8.167537688442211e-05,
-      "loss": 5.18,
-      "step": 18739
-    },
-    {
-      "epoch": 9.773142112125162,
-      "grad_norm": 1.5900564193725586,
-      "learning_rate": 8.167437185929648e-05,
-      "loss": 5.1515,
-      "step": 18740
-    },
-    {
-      "epoch": 9.773663624511082,
-      "grad_norm": 1.4741532802581787,
-      "learning_rate": 8.167336683417086e-05,
-      "loss": 5.4508,
-      "step": 18741
-    },
-    {
-      "epoch": 9.774185136897001,
-      "grad_norm": 1.6422709226608276,
-      "learning_rate": 8.167236180904522e-05,
-      "loss": 5.4192,
-      "step": 18742
-    },
-    {
-      "epoch": 9.77470664928292,
-      "grad_norm": 1.3778619766235352,
-      "learning_rate": 8.16713567839196e-05,
-      "loss": 5.9432,
-      "step": 18743
-    },
-    {
-      "epoch": 9.77522816166884,
-      "grad_norm": 1.5278359651565552,
-      "learning_rate": 8.167035175879398e-05,
-      "loss": 5.6264,
-      "step": 18744
-    },
-    {
-      "epoch": 9.775749674054758,
-      "grad_norm": 1.4320108890533447,
-      "learning_rate": 8.166934673366835e-05,
-      "loss": 5.5473,
-      "step": 18745
-    },
-    {
-      "epoch": 9.776271186440677,
-      "grad_norm": 1.3931071758270264,
-      "learning_rate": 8.166834170854272e-05,
-      "loss": 5.6958,
-      "step": 18746
-    },
-    {
-      "epoch": 9.776792698826597,
-      "grad_norm": 1.4775317907333374,
-      "learning_rate": 8.166733668341708e-05,
-      "loss": 5.446,
-      "step": 18747
-    },
-    {
-      "epoch": 9.777314211212516,
-      "grad_norm": 1.413123607635498,
-      "learning_rate": 8.166633165829146e-05,
-      "loss": 5.1131,
-      "step": 18748
-    },
-    {
-      "epoch": 9.777835723598436,
-      "grad_norm": 1.368774652481079,
-      "learning_rate": 8.166532663316583e-05,
-      "loss": 5.3037,
-      "step": 18749
-    },
-    {
-      "epoch": 9.778357235984355,
-      "grad_norm": 1.403531551361084,
-      "learning_rate": 8.16643216080402e-05,
-      "loss": 5.3152,
-      "step": 18750
-    },
-    {
-      "epoch": 9.778878748370273,
-      "grad_norm": 1.403852105140686,
-      "learning_rate": 8.166331658291457e-05,
-      "loss": 5.2872,
-      "step": 18751
-    },
-    {
-      "epoch": 9.779400260756193,
-      "grad_norm": 1.3769657611846924,
-      "learning_rate": 8.166231155778895e-05,
-      "loss": 5.687,
-      "step": 18752
-    },
-    {
-      "epoch": 9.779921773142112,
-      "grad_norm": 1.4277992248535156,
-      "learning_rate": 8.166130653266332e-05,
-      "loss": 5.7698,
-      "step": 18753
-    },
-    {
-      "epoch": 9.780443285528031,
-      "grad_norm": 1.3920516967773438,
-      "learning_rate": 8.16603015075377e-05,
-      "loss": 5.4083,
-      "step": 18754
-    },
-    {
-      "epoch": 9.780964797913951,
-      "grad_norm": 1.63652765750885,
-      "learning_rate": 8.165929648241207e-05,
-      "loss": 5.1056,
-      "step": 18755
-    },
-    {
-      "epoch": 9.78148631029987,
-      "grad_norm": 1.4491411447525024,
-      "learning_rate": 8.165829145728644e-05,
-      "loss": 5.2202,
-      "step": 18756
-    },
-    {
-      "epoch": 9.782007822685788,
-      "grad_norm": 1.5373526811599731,
-      "learning_rate": 8.165728643216081e-05,
-      "loss": 5.3665,
-      "step": 18757
-    },
-    {
-      "epoch": 9.782529335071708,
-      "grad_norm": 1.4949219226837158,
-      "learning_rate": 8.165628140703519e-05,
-      "loss": 5.1171,
-      "step": 18758
-    },
-    {
-      "epoch": 9.783050847457627,
-      "grad_norm": 1.4364678859710693,
-      "learning_rate": 8.165527638190955e-05,
-      "loss": 4.9946,
-      "step": 18759
-    },
-    {
-      "epoch": 9.783572359843546,
-      "grad_norm": 1.4808757305145264,
-      "learning_rate": 8.165427135678393e-05,
-      "loss": 5.5155,
-      "step": 18760
-    },
-    {
-      "epoch": 9.784093872229466,
-      "grad_norm": 1.4968109130859375,
-      "learning_rate": 8.165326633165829e-05,
-      "loss": 5.0605,
-      "step": 18761
-    },
-    {
-      "epoch": 9.784615384615385,
-      "grad_norm": 1.4909348487854004,
-      "learning_rate": 8.165226130653266e-05,
-      "loss": 5.6026,
-      "step": 18762
-    },
-    {
-      "epoch": 9.785136897001303,
-      "grad_norm": 1.5875134468078613,
-      "learning_rate": 8.165125628140703e-05,
-      "loss": 5.2109,
-      "step": 18763
-    },
-    {
-      "epoch": 9.785658409387223,
-      "grad_norm": 1.4120519161224365,
-      "learning_rate": 8.165025125628141e-05,
-      "loss": 5.6189,
-      "step": 18764
-    },
-    {
-      "epoch": 9.786179921773142,
-      "grad_norm": 1.4031940698623657,
-      "learning_rate": 8.164924623115579e-05,
-      "loss": 5.5703,
-      "step": 18765
-    },
-    {
-      "epoch": 9.786701434159061,
-      "grad_norm": 1.4226372241973877,
-      "learning_rate": 8.164824120603015e-05,
-      "loss": 5.3847,
-      "step": 18766
-    },
-    {
-      "epoch": 9.787222946544981,
-      "grad_norm": 1.4345930814743042,
-      "learning_rate": 8.164723618090453e-05,
-      "loss": 5.3802,
-      "step": 18767
-    },
-    {
-      "epoch": 9.7877444589309,
-      "grad_norm": 1.6860346794128418,
-      "learning_rate": 8.16462311557789e-05,
-      "loss": 4.2616,
-      "step": 18768
-    },
-    {
-      "epoch": 9.788265971316818,
-      "grad_norm": 1.481545090675354,
-      "learning_rate": 8.164522613065327e-05,
-      "loss": 5.7262,
-      "step": 18769
-    },
-    {
-      "epoch": 9.788787483702738,
-      "grad_norm": 1.4126156568527222,
-      "learning_rate": 8.164422110552764e-05,
-      "loss": 5.1905,
-      "step": 18770
-    },
-    {
-      "epoch": 9.789308996088657,
-      "grad_norm": 1.4206933975219727,
-      "learning_rate": 8.164321608040202e-05,
-      "loss": 5.4232,
-      "step": 18771
-    },
-    {
-      "epoch": 9.789830508474576,
-      "grad_norm": 1.4722241163253784,
-      "learning_rate": 8.164221105527638e-05,
-      "loss": 5.5743,
-      "step": 18772
-    },
-    {
-      "epoch": 9.790352020860496,
-      "grad_norm": 1.3364452123641968,
-      "learning_rate": 8.164120603015076e-05,
-      "loss": 5.8062,
-      "step": 18773
-    },
-    {
-      "epoch": 9.790873533246415,
-      "grad_norm": 1.3238264322280884,
-      "learning_rate": 8.164020100502514e-05,
-      "loss": 5.7054,
-      "step": 18774
-    },
-    {
-      "epoch": 9.791395045632333,
-      "grad_norm": 1.4345972537994385,
-      "learning_rate": 8.16391959798995e-05,
-      "loss": 5.4621,
-      "step": 18775
-    },
-    {
-      "epoch": 9.791916558018253,
-      "grad_norm": 1.4506257772445679,
-      "learning_rate": 8.163819095477388e-05,
-      "loss": 5.482,
-      "step": 18776
-    },
-    {
-      "epoch": 9.792438070404172,
-      "grad_norm": 1.4590561389923096,
-      "learning_rate": 8.163718592964824e-05,
-      "loss": 5.709,
-      "step": 18777
-    },
-    {
-      "epoch": 9.792959582790091,
-      "grad_norm": 1.4055347442626953,
-      "learning_rate": 8.163618090452262e-05,
-      "loss": 5.7277,
-      "step": 18778
-    },
-    {
-      "epoch": 9.793481095176011,
-      "grad_norm": 1.4934043884277344,
-      "learning_rate": 8.163517587939698e-05,
-      "loss": 5.5902,
-      "step": 18779
-    },
-    {
-      "epoch": 9.79400260756193,
-      "grad_norm": 1.4280189275741577,
-      "learning_rate": 8.163417085427136e-05,
-      "loss": 5.5951,
-      "step": 18780
-    },
-    {
-      "epoch": 9.794524119947848,
-      "grad_norm": 1.4233018159866333,
-      "learning_rate": 8.163316582914573e-05,
-      "loss": 5.5975,
-      "step": 18781
-    },
-    {
-      "epoch": 9.795045632333768,
-      "grad_norm": 1.420574426651001,
-      "learning_rate": 8.16321608040201e-05,
-      "loss": 4.9335,
-      "step": 18782
-    },
-    {
-      "epoch": 9.795567144719687,
-      "grad_norm": 1.4069085121154785,
-      "learning_rate": 8.163115577889447e-05,
-      "loss": 5.7459,
-      "step": 18783
-    },
-    {
-      "epoch": 9.796088657105607,
-      "grad_norm": 1.7237991094589233,
-      "learning_rate": 8.163015075376885e-05,
-      "loss": 5.232,
-      "step": 18784
-    },
-    {
-      "epoch": 9.796610169491526,
-      "grad_norm": 1.396418571472168,
-      "learning_rate": 8.162914572864322e-05,
-      "loss": 5.2956,
-      "step": 18785
-    },
-    {
-      "epoch": 9.797131681877445,
-      "grad_norm": 1.4072816371917725,
-      "learning_rate": 8.16281407035176e-05,
-      "loss": 5.6309,
-      "step": 18786
-    },
-    {
-      "epoch": 9.797653194263363,
-      "grad_norm": 1.359890341758728,
-      "learning_rate": 8.162713567839197e-05,
-      "loss": 5.7561,
-      "step": 18787
-    },
-    {
-      "epoch": 9.798174706649283,
-      "grad_norm": 1.4151477813720703,
-      "learning_rate": 8.162613065326633e-05,
-      "loss": 5.4661,
-      "step": 18788
-    },
-    {
-      "epoch": 9.798696219035202,
-      "grad_norm": 1.3870978355407715,
-      "learning_rate": 8.162512562814071e-05,
-      "loss": 5.382,
-      "step": 18789
-    },
-    {
-      "epoch": 9.799217731421122,
-      "grad_norm": 1.5548274517059326,
-      "learning_rate": 8.162412060301507e-05,
-      "loss": 5.0256,
-      "step": 18790
-    },
-    {
-      "epoch": 9.799739243807041,
-      "grad_norm": 1.6256885528564453,
-      "learning_rate": 8.162311557788945e-05,
-      "loss": 5.3333,
-      "step": 18791
-    },
-    {
-      "epoch": 9.800260756192959,
-      "grad_norm": 1.363315224647522,
-      "learning_rate": 8.162211055276381e-05,
-      "loss": 5.9492,
-      "step": 18792
-    },
-    {
-      "epoch": 9.800782268578878,
-      "grad_norm": 1.3605036735534668,
-      "learning_rate": 8.162110552763819e-05,
-      "loss": 5.7858,
-      "step": 18793
-    },
-    {
-      "epoch": 9.801303780964798,
-      "grad_norm": 1.4177042245864868,
-      "learning_rate": 8.162010050251257e-05,
-      "loss": 5.5254,
-      "step": 18794
-    },
-    {
-      "epoch": 9.801825293350717,
-      "grad_norm": 1.4957058429718018,
-      "learning_rate": 8.161909547738695e-05,
-      "loss": 5.7703,
-      "step": 18795
-    },
-    {
-      "epoch": 9.802346805736637,
-      "grad_norm": 1.3762632608413696,
-      "learning_rate": 8.161809045226131e-05,
-      "loss": 5.2932,
-      "step": 18796
-    },
-    {
-      "epoch": 9.802868318122556,
-      "grad_norm": 1.277329444885254,
-      "learning_rate": 8.161708542713569e-05,
-      "loss": 5.7655,
-      "step": 18797
-    },
-    {
-      "epoch": 9.803389830508475,
-      "grad_norm": 1.4631900787353516,
-      "learning_rate": 8.161608040201005e-05,
-      "loss": 5.2617,
-      "step": 18798
-    },
-    {
-      "epoch": 9.803911342894393,
-      "grad_norm": 1.6917686462402344,
-      "learning_rate": 8.161507537688443e-05,
-      "loss": 5.1344,
-      "step": 18799
-    },
-    {
-      "epoch": 9.804432855280313,
-      "grad_norm": 1.4270349740982056,
-      "learning_rate": 8.16140703517588e-05,
-      "loss": 5.7663,
-      "step": 18800
-    },
-    {
-      "epoch": 9.804954367666232,
-      "grad_norm": 1.4527724981307983,
-      "learning_rate": 8.161306532663316e-05,
-      "loss": 5.6209,
-      "step": 18801
-    },
-    {
-      "epoch": 9.805475880052152,
-      "grad_norm": 1.4133434295654297,
-      "learning_rate": 8.161206030150754e-05,
-      "loss": 5.923,
-      "step": 18802
-    },
-    {
-      "epoch": 9.805997392438071,
-      "grad_norm": 1.418936014175415,
-      "learning_rate": 8.16110552763819e-05,
-      "loss": 5.8359,
-      "step": 18803
-    },
-    {
-      "epoch": 9.806518904823989,
-      "grad_norm": 1.3846182823181152,
-      "learning_rate": 8.161005025125628e-05,
-      "loss": 5.585,
-      "step": 18804
-    },
-    {
-      "epoch": 9.807040417209908,
-      "grad_norm": 1.406559944152832,
-      "learning_rate": 8.160904522613066e-05,
-      "loss": 5.3978,
-      "step": 18805
-    },
-    {
-      "epoch": 9.807561929595828,
-      "grad_norm": 1.467599630355835,
-      "learning_rate": 8.160804020100504e-05,
-      "loss": 5.545,
-      "step": 18806
-    },
-    {
-      "epoch": 9.808083441981747,
-      "grad_norm": 1.3431358337402344,
-      "learning_rate": 8.16070351758794e-05,
-      "loss": 5.0636,
-      "step": 18807
-    },
-    {
-      "epoch": 9.808604954367667,
-      "grad_norm": 1.398579478263855,
-      "learning_rate": 8.160603015075378e-05,
-      "loss": 5.5567,
-      "step": 18808
-    },
-    {
-      "epoch": 9.809126466753586,
-      "grad_norm": 1.3804455995559692,
-      "learning_rate": 8.160502512562814e-05,
-      "loss": 5.449,
-      "step": 18809
-    },
-    {
-      "epoch": 9.809647979139505,
-      "grad_norm": 1.5802336931228638,
-      "learning_rate": 8.160402010050252e-05,
-      "loss": 5.2804,
-      "step": 18810
-    },
-    {
-      "epoch": 9.810169491525423,
-      "grad_norm": 1.5036910772323608,
-      "learning_rate": 8.160301507537688e-05,
-      "loss": 5.104,
-      "step": 18811
-    },
-    {
-      "epoch": 9.810691003911343,
-      "grad_norm": 1.335680365562439,
-      "learning_rate": 8.160201005025126e-05,
-      "loss": 5.1279,
-      "step": 18812
-    },
-    {
-      "epoch": 9.811212516297262,
-      "grad_norm": 1.5513349771499634,
-      "learning_rate": 8.160100502512563e-05,
-      "loss": 5.0275,
-      "step": 18813
-    },
-    {
-      "epoch": 9.811734028683182,
-      "grad_norm": 1.557739019393921,
-      "learning_rate": 8.16e-05,
-      "loss": 5.3305,
-      "step": 18814
-    },
-    {
-      "epoch": 9.812255541069101,
-      "grad_norm": 1.3543133735656738,
-      "learning_rate": 8.159899497487438e-05,
-      "loss": 5.3197,
-      "step": 18815
-    },
-    {
-      "epoch": 9.812777053455019,
-      "grad_norm": 1.380679726600647,
-      "learning_rate": 8.159798994974875e-05,
-      "loss": 5.5661,
-      "step": 18816
-    },
-    {
-      "epoch": 9.813298565840938,
-      "grad_norm": 1.3249982595443726,
-      "learning_rate": 8.159698492462312e-05,
-      "loss": 5.8286,
-      "step": 18817
-    },
-    {
-      "epoch": 9.813820078226858,
-      "grad_norm": 1.5055056810379028,
-      "learning_rate": 8.159597989949749e-05,
-      "loss": 4.5437,
-      "step": 18818
-    },
-    {
-      "epoch": 9.814341590612777,
-      "grad_norm": 1.5723572969436646,
-      "learning_rate": 8.159497487437187e-05,
-      "loss": 5.2732,
-      "step": 18819
-    },
-    {
-      "epoch": 9.814863102998697,
-      "grad_norm": 1.386507511138916,
-      "learning_rate": 8.159396984924623e-05,
-      "loss": 5.8666,
-      "step": 18820
-    },
-    {
-      "epoch": 9.815384615384616,
-      "grad_norm": 1.4611873626708984,
-      "learning_rate": 8.159296482412061e-05,
-      "loss": 5.8214,
-      "step": 18821
-    },
-    {
-      "epoch": 9.815906127770534,
-      "grad_norm": 1.3396035432815552,
-      "learning_rate": 8.159195979899497e-05,
-      "loss": 5.9309,
-      "step": 18822
-    },
-    {
-      "epoch": 9.816427640156453,
-      "grad_norm": 1.283207654953003,
-      "learning_rate": 8.159095477386935e-05,
-      "loss": 5.9526,
-      "step": 18823
-    },
-    {
-      "epoch": 9.816949152542373,
-      "grad_norm": 1.5138282775878906,
-      "learning_rate": 8.158994974874372e-05,
-      "loss": 5.157,
-      "step": 18824
-    },
-    {
-      "epoch": 9.817470664928292,
-      "grad_norm": 1.6958012580871582,
-      "learning_rate": 8.158894472361809e-05,
-      "loss": 5.3487,
-      "step": 18825
-    },
-    {
-      "epoch": 9.817992177314212,
-      "grad_norm": 1.5657508373260498,
-      "learning_rate": 8.158793969849247e-05,
-      "loss": 4.8755,
-      "step": 18826
-    },
-    {
-      "epoch": 9.818513689700131,
-      "grad_norm": 1.4833314418792725,
-      "learning_rate": 8.158693467336684e-05,
-      "loss": 5.0079,
-      "step": 18827
-    },
-    {
-      "epoch": 9.819035202086049,
-      "grad_norm": 1.5305812358856201,
-      "learning_rate": 8.158592964824121e-05,
-      "loss": 5.4943,
-      "step": 18828
-    },
-    {
-      "epoch": 9.819556714471968,
-      "grad_norm": 1.434797763824463,
-      "learning_rate": 8.158492462311558e-05,
-      "loss": 5.5853,
-      "step": 18829
-    },
-    {
-      "epoch": 9.820078226857888,
-      "grad_norm": 1.4151307344436646,
-      "learning_rate": 8.158391959798996e-05,
-      "loss": 5.7899,
-      "step": 18830
-    },
-    {
-      "epoch": 9.820599739243807,
-      "grad_norm": 1.575729250907898,
-      "learning_rate": 8.158291457286432e-05,
-      "loss": 5.0509,
-      "step": 18831
-    },
-    {
-      "epoch": 9.821121251629727,
-      "grad_norm": 1.4846992492675781,
-      "learning_rate": 8.15819095477387e-05,
-      "loss": 5.2576,
-      "step": 18832
-    },
-    {
-      "epoch": 9.821642764015646,
-      "grad_norm": 1.4463894367218018,
-      "learning_rate": 8.158090452261306e-05,
-      "loss": 5.6715,
-      "step": 18833
-    },
-    {
-      "epoch": 9.822164276401564,
-      "grad_norm": 1.5585966110229492,
-      "learning_rate": 8.157989949748744e-05,
-      "loss": 5.6754,
-      "step": 18834
-    },
-    {
-      "epoch": 9.822685788787483,
-      "grad_norm": 1.617010235786438,
-      "learning_rate": 8.157889447236182e-05,
-      "loss": 4.9578,
-      "step": 18835
-    },
-    {
-      "epoch": 9.823207301173403,
-      "grad_norm": 1.4858665466308594,
-      "learning_rate": 8.15778894472362e-05,
-      "loss": 5.4435,
-      "step": 18836
-    },
-    {
-      "epoch": 9.823728813559322,
-      "grad_norm": 1.4326707124710083,
-      "learning_rate": 8.157688442211056e-05,
-      "loss": 5.2799,
-      "step": 18837
-    },
-    {
-      "epoch": 9.824250325945242,
-      "grad_norm": 1.4918309450149536,
-      "learning_rate": 8.157587939698494e-05,
-      "loss": 5.579,
-      "step": 18838
-    },
-    {
-      "epoch": 9.824771838331161,
-      "grad_norm": 1.600682258605957,
-      "learning_rate": 8.15748743718593e-05,
-      "loss": 5.1021,
-      "step": 18839
-    },
-    {
-      "epoch": 9.825293350717079,
-      "grad_norm": 1.4583885669708252,
-      "learning_rate": 8.157386934673367e-05,
-      "loss": 5.341,
-      "step": 18840
-    },
-    {
-      "epoch": 9.825814863102998,
-      "grad_norm": 1.5668346881866455,
-      "learning_rate": 8.157286432160804e-05,
-      "loss": 5.6872,
-      "step": 18841
-    },
-    {
-      "epoch": 9.826336375488918,
-      "grad_norm": 1.375795602798462,
-      "learning_rate": 8.157185929648241e-05,
-      "loss": 5.7434,
-      "step": 18842
-    },
-    {
-      "epoch": 9.826857887874837,
-      "grad_norm": 1.4205663204193115,
-      "learning_rate": 8.157085427135679e-05,
-      "loss": 5.9063,
-      "step": 18843
-    },
-    {
-      "epoch": 9.827379400260757,
-      "grad_norm": 1.510650634765625,
-      "learning_rate": 8.156984924623115e-05,
-      "loss": 5.5451,
-      "step": 18844
-    },
-    {
-      "epoch": 9.827900912646676,
-      "grad_norm": 1.3845927715301514,
-      "learning_rate": 8.156884422110553e-05,
-      "loss": 5.7148,
-      "step": 18845
-    },
-    {
-      "epoch": 9.828422425032594,
-      "grad_norm": 1.3882251977920532,
-      "learning_rate": 8.15678391959799e-05,
-      "loss": 5.2,
-      "step": 18846
-    },
-    {
-      "epoch": 9.828943937418513,
-      "grad_norm": 1.4715758562088013,
-      "learning_rate": 8.156683417085428e-05,
-      "loss": 5.4933,
-      "step": 18847
-    },
-    {
-      "epoch": 9.829465449804433,
-      "grad_norm": 1.342267394065857,
-      "learning_rate": 8.156582914572865e-05,
-      "loss": 5.6232,
-      "step": 18848
-    },
-    {
-      "epoch": 9.829986962190352,
-      "grad_norm": 1.542335033416748,
-      "learning_rate": 8.156482412060303e-05,
-      "loss": 5.7102,
-      "step": 18849
-    },
-    {
-      "epoch": 9.830508474576272,
-      "grad_norm": 1.3038665056228638,
-      "learning_rate": 8.156381909547739e-05,
-      "loss": 5.8304,
-      "step": 18850
-    },
-    {
-      "epoch": 9.831029986962191,
-      "grad_norm": 1.475143313407898,
-      "learning_rate": 8.156281407035177e-05,
-      "loss": 5.1172,
-      "step": 18851
-    },
-    {
-      "epoch": 9.831551499348109,
-      "grad_norm": 1.4131441116333008,
-      "learning_rate": 8.156180904522613e-05,
-      "loss": 5.2947,
-      "step": 18852
-    },
-    {
-      "epoch": 9.832073011734028,
-      "grad_norm": 1.5086700916290283,
-      "learning_rate": 8.156080402010051e-05,
-      "loss": 5.6492,
-      "step": 18853
-    },
-    {
-      "epoch": 9.832594524119948,
-      "grad_norm": 1.3493973016738892,
-      "learning_rate": 8.155979899497487e-05,
-      "loss": 5.7619,
-      "step": 18854
-    },
-    {
-      "epoch": 9.833116036505867,
-      "grad_norm": 1.3690553903579712,
-      "learning_rate": 8.155879396984925e-05,
-      "loss": 5.8245,
-      "step": 18855
-    },
-    {
-      "epoch": 9.833637548891787,
-      "grad_norm": 1.4208611249923706,
-      "learning_rate": 8.155778894472363e-05,
-      "loss": 5.079,
-      "step": 18856
-    },
-    {
-      "epoch": 9.834159061277706,
-      "grad_norm": 1.2651432752609253,
-      "learning_rate": 8.1556783919598e-05,
-      "loss": 5.6992,
-      "step": 18857
-    },
-    {
-      "epoch": 9.834680573663624,
-      "grad_norm": 1.4521236419677734,
-      "learning_rate": 8.155577889447237e-05,
-      "loss": 5.1632,
-      "step": 18858
-    },
-    {
-      "epoch": 9.835202086049543,
-      "grad_norm": 1.4869657754898071,
-      "learning_rate": 8.155477386934674e-05,
-      "loss": 5.1051,
-      "step": 18859
-    },
-    {
-      "epoch": 9.835723598435463,
-      "grad_norm": 1.4910728931427002,
-      "learning_rate": 8.155376884422111e-05,
-      "loss": 5.356,
-      "step": 18860
-    },
-    {
-      "epoch": 9.836245110821382,
-      "grad_norm": 1.47030770778656,
-      "learning_rate": 8.155276381909548e-05,
-      "loss": 5.5372,
-      "step": 18861
-    },
-    {
-      "epoch": 9.836766623207302,
-      "grad_norm": 1.3523238897323608,
-      "learning_rate": 8.155175879396986e-05,
-      "loss": 4.7673,
-      "step": 18862
-    },
-    {
-      "epoch": 9.837288135593221,
-      "grad_norm": 1.3635660409927368,
-      "learning_rate": 8.155075376884422e-05,
-      "loss": 4.6646,
-      "step": 18863
-    },
-    {
-      "epoch": 9.837809647979139,
-      "grad_norm": 1.3605055809020996,
-      "learning_rate": 8.15497487437186e-05,
-      "loss": 5.6889,
-      "step": 18864
-    },
-    {
-      "epoch": 9.838331160365058,
-      "grad_norm": 1.4220993518829346,
-      "learning_rate": 8.154874371859296e-05,
-      "loss": 5.2597,
-      "step": 18865
-    },
-    {
-      "epoch": 9.838852672750978,
-      "grad_norm": 1.4281303882598877,
-      "learning_rate": 8.154773869346734e-05,
-      "loss": 5.7396,
-      "step": 18866
-    },
-    {
-      "epoch": 9.839374185136897,
-      "grad_norm": 1.4710369110107422,
-      "learning_rate": 8.154673366834172e-05,
-      "loss": 5.6739,
-      "step": 18867
-    },
-    {
-      "epoch": 9.839895697522817,
-      "grad_norm": 1.4634591341018677,
-      "learning_rate": 8.154572864321608e-05,
-      "loss": 5.2925,
-      "step": 18868
-    },
-    {
-      "epoch": 9.840417209908736,
-      "grad_norm": 1.4451037645339966,
-      "learning_rate": 8.154472361809046e-05,
-      "loss": 5.0543,
-      "step": 18869
-    },
-    {
-      "epoch": 9.840938722294654,
-      "grad_norm": 1.5098178386688232,
-      "learning_rate": 8.154371859296482e-05,
-      "loss": 5.3023,
-      "step": 18870
-    },
-    {
-      "epoch": 9.841460234680573,
-      "grad_norm": 1.4066853523254395,
-      "learning_rate": 8.15427135678392e-05,
-      "loss": 5.8847,
-      "step": 18871
-    },
-    {
-      "epoch": 9.841981747066493,
-      "grad_norm": 1.3362759351730347,
-      "learning_rate": 8.154170854271357e-05,
-      "loss": 5.8221,
-      "step": 18872
-    },
-    {
-      "epoch": 9.842503259452412,
-      "grad_norm": 1.4905142784118652,
-      "learning_rate": 8.154070351758794e-05,
-      "loss": 4.8676,
-      "step": 18873
-    },
-    {
-      "epoch": 9.843024771838332,
-      "grad_norm": 1.4173389673233032,
-      "learning_rate": 8.153969849246231e-05,
-      "loss": 5.4939,
-      "step": 18874
-    },
-    {
-      "epoch": 9.843546284224251,
-      "grad_norm": 1.3508365154266357,
-      "learning_rate": 8.153869346733669e-05,
-      "loss": 5.5821,
-      "step": 18875
-    },
-    {
-      "epoch": 9.844067796610169,
-      "grad_norm": 1.3303414583206177,
-      "learning_rate": 8.153768844221106e-05,
-      "loss": 5.6829,
-      "step": 18876
-    },
-    {
-      "epoch": 9.844589308996088,
-      "grad_norm": 1.514719843864441,
-      "learning_rate": 8.153668341708544e-05,
-      "loss": 4.7906,
-      "step": 18877
-    },
-    {
-      "epoch": 9.845110821382008,
-      "grad_norm": 1.423122763633728,
-      "learning_rate": 8.15356783919598e-05,
-      "loss": 5.5742,
-      "step": 18878
-    },
-    {
-      "epoch": 9.845632333767927,
-      "grad_norm": 1.4707762002944946,
-      "learning_rate": 8.153467336683418e-05,
-      "loss": 5.3542,
-      "step": 18879
-    },
-    {
-      "epoch": 9.846153846153847,
-      "grad_norm": 1.637756109237671,
-      "learning_rate": 8.153366834170855e-05,
-      "loss": 5.2408,
-      "step": 18880
-    },
-    {
-      "epoch": 9.846675358539766,
-      "grad_norm": 1.491081714630127,
-      "learning_rate": 8.153266331658291e-05,
-      "loss": 5.5207,
-      "step": 18881
-    },
-    {
-      "epoch": 9.847196870925684,
-      "grad_norm": 1.529356598854065,
-      "learning_rate": 8.153165829145729e-05,
-      "loss": 5.5198,
-      "step": 18882
-    },
-    {
-      "epoch": 9.847718383311603,
-      "grad_norm": 1.4644250869750977,
-      "learning_rate": 8.153065326633165e-05,
-      "loss": 5.4591,
-      "step": 18883
-    },
-    {
-      "epoch": 9.848239895697523,
-      "grad_norm": 1.4229185581207275,
-      "learning_rate": 8.152964824120603e-05,
-      "loss": 4.9404,
-      "step": 18884
-    },
-    {
-      "epoch": 9.848761408083442,
-      "grad_norm": 1.3820608854293823,
-      "learning_rate": 8.15286432160804e-05,
-      "loss": 5.6197,
-      "step": 18885
-    },
-    {
-      "epoch": 9.849282920469362,
-      "grad_norm": 1.3401633501052856,
-      "learning_rate": 8.152763819095477e-05,
-      "loss": 5.9148,
-      "step": 18886
-    },
-    {
-      "epoch": 9.84980443285528,
-      "grad_norm": 1.5599145889282227,
-      "learning_rate": 8.152663316582915e-05,
-      "loss": 4.9957,
-      "step": 18887
-    },
-    {
-      "epoch": 9.850325945241199,
-      "grad_norm": 1.4089999198913574,
-      "learning_rate": 8.152562814070353e-05,
-      "loss": 5.125,
-      "step": 18888
-    },
-    {
-      "epoch": 9.850847457627118,
-      "grad_norm": 1.3984298706054688,
-      "learning_rate": 8.15246231155779e-05,
-      "loss": 5.8352,
-      "step": 18889
-    },
-    {
-      "epoch": 9.851368970013038,
-      "grad_norm": 1.3625659942626953,
-      "learning_rate": 8.152361809045227e-05,
-      "loss": 5.4781,
-      "step": 18890
-    },
-    {
-      "epoch": 9.851890482398957,
-      "grad_norm": 1.3269946575164795,
-      "learning_rate": 8.152261306532664e-05,
-      "loss": 5.5337,
-      "step": 18891
-    },
-    {
-      "epoch": 9.852411994784877,
-      "grad_norm": 1.382431983947754,
-      "learning_rate": 8.152160804020101e-05,
-      "loss": 5.6886,
-      "step": 18892
-    },
-    {
-      "epoch": 9.852933507170796,
-      "grad_norm": 1.3401826620101929,
-      "learning_rate": 8.152060301507538e-05,
-      "loss": 5.7318,
-      "step": 18893
-    },
-    {
-      "epoch": 9.853455019556714,
-      "grad_norm": 1.5526294708251953,
-      "learning_rate": 8.151959798994974e-05,
-      "loss": 5.3919,
-      "step": 18894
-    },
-    {
-      "epoch": 9.853976531942633,
-      "grad_norm": 1.318988561630249,
-      "learning_rate": 8.151859296482412e-05,
-      "loss": 5.6052,
-      "step": 18895
-    },
-    {
-      "epoch": 9.854498044328553,
-      "grad_norm": 1.5080275535583496,
-      "learning_rate": 8.15175879396985e-05,
-      "loss": 5.1096,
-      "step": 18896
-    },
-    {
-      "epoch": 9.855019556714472,
-      "grad_norm": 1.4578192234039307,
-      "learning_rate": 8.151658291457288e-05,
-      "loss": 5.2731,
-      "step": 18897
-    },
-    {
-      "epoch": 9.855541069100392,
-      "grad_norm": 1.4876171350479126,
-      "learning_rate": 8.151557788944724e-05,
-      "loss": 4.9786,
-      "step": 18898
-    },
-    {
-      "epoch": 9.85606258148631,
-      "grad_norm": 1.474180817604065,
-      "learning_rate": 8.151457286432162e-05,
-      "loss": 5.4343,
-      "step": 18899
-    },
-    {
-      "epoch": 9.856584093872229,
-      "grad_norm": 1.388884425163269,
-      "learning_rate": 8.151356783919598e-05,
-      "loss": 5.5142,
-      "step": 18900
-    },
-    {
-      "epoch": 9.857105606258148,
-      "grad_norm": 1.4965856075286865,
-      "learning_rate": 8.151256281407036e-05,
-      "loss": 5.6792,
-      "step": 18901
-    },
-    {
-      "epoch": 9.857627118644068,
-      "grad_norm": 1.4323372840881348,
-      "learning_rate": 8.151155778894473e-05,
-      "loss": 5.3885,
-      "step": 18902
-    },
-    {
-      "epoch": 9.858148631029987,
-      "grad_norm": 1.359602689743042,
-      "learning_rate": 8.15105527638191e-05,
-      "loss": 5.6085,
-      "step": 18903
-    },
-    {
-      "epoch": 9.858670143415907,
-      "grad_norm": 1.3607302904129028,
-      "learning_rate": 8.150954773869347e-05,
-      "loss": 5.4653,
-      "step": 18904
-    },
-    {
-      "epoch": 9.859191655801826,
-      "grad_norm": 1.4618481397628784,
-      "learning_rate": 8.150854271356785e-05,
-      "loss": 5.7132,
-      "step": 18905
-    },
-    {
-      "epoch": 9.859713168187744,
-      "grad_norm": 1.419859528541565,
-      "learning_rate": 8.150753768844221e-05,
-      "loss": 5.5395,
-      "step": 18906
-    },
-    {
-      "epoch": 9.860234680573663,
-      "grad_norm": 1.3764078617095947,
-      "learning_rate": 8.150653266331659e-05,
-      "loss": 5.7023,
-      "step": 18907
-    },
-    {
-      "epoch": 9.860756192959583,
-      "grad_norm": 1.3198775053024292,
-      "learning_rate": 8.150552763819096e-05,
-      "loss": 5.4857,
-      "step": 18908
-    },
-    {
-      "epoch": 9.861277705345502,
-      "grad_norm": 1.4312407970428467,
-      "learning_rate": 8.150452261306533e-05,
-      "loss": 5.0223,
-      "step": 18909
-    },
-    {
-      "epoch": 9.861799217731422,
-      "grad_norm": 1.5155631303787231,
-      "learning_rate": 8.150351758793971e-05,
-      "loss": 5.2398,
-      "step": 18910
-    },
-    {
-      "epoch": 9.86232073011734,
-      "grad_norm": 1.5549267530441284,
-      "learning_rate": 8.150251256281407e-05,
-      "loss": 5.0545,
-      "step": 18911
-    },
-    {
-      "epoch": 9.862842242503259,
-      "grad_norm": 1.3913756608963013,
-      "learning_rate": 8.150150753768845e-05,
-      "loss": 5.7832,
-      "step": 18912
-    },
-    {
-      "epoch": 9.863363754889178,
-      "grad_norm": 1.5730245113372803,
-      "learning_rate": 8.150050251256281e-05,
-      "loss": 5.1918,
-      "step": 18913
-    },
-    {
-      "epoch": 9.863885267275098,
-      "grad_norm": 1.479029893875122,
-      "learning_rate": 8.149949748743719e-05,
-      "loss": 5.3568,
-      "step": 18914
-    },
-    {
-      "epoch": 9.864406779661017,
-      "grad_norm": 1.4946562051773071,
-      "learning_rate": 8.149849246231156e-05,
-      "loss": 5.5501,
-      "step": 18915
-    },
-    {
-      "epoch": 9.864928292046937,
-      "grad_norm": 1.527305245399475,
-      "learning_rate": 8.149748743718593e-05,
-      "loss": 5.3042,
-      "step": 18916
-    },
-    {
-      "epoch": 9.865449804432854,
-      "grad_norm": 1.4032378196716309,
-      "learning_rate": 8.14964824120603e-05,
-      "loss": 5.3333,
-      "step": 18917
-    },
-    {
-      "epoch": 9.865971316818774,
-      "grad_norm": 1.4341539144515991,
-      "learning_rate": 8.149547738693468e-05,
-      "loss": 5.2739,
-      "step": 18918
-    },
-    {
-      "epoch": 9.866492829204693,
-      "grad_norm": 1.556546926498413,
-      "learning_rate": 8.149447236180905e-05,
-      "loss": 5.2747,
-      "step": 18919
-    },
-    {
-      "epoch": 9.867014341590613,
-      "grad_norm": 1.3808815479278564,
-      "learning_rate": 8.149346733668342e-05,
-      "loss": 5.2508,
-      "step": 18920
-    },
-    {
-      "epoch": 9.867535853976532,
-      "grad_norm": 1.4103188514709473,
-      "learning_rate": 8.14924623115578e-05,
-      "loss": 5.6544,
-      "step": 18921
-    },
-    {
-      "epoch": 9.868057366362452,
-      "grad_norm": 1.5518792867660522,
-      "learning_rate": 8.149145728643216e-05,
-      "loss": 5.2806,
-      "step": 18922
-    },
-    {
-      "epoch": 9.86857887874837,
-      "grad_norm": 1.5141112804412842,
-      "learning_rate": 8.149045226130654e-05,
-      "loss": 5.3195,
-      "step": 18923
-    },
-    {
-      "epoch": 9.869100391134289,
-      "grad_norm": 1.3024266958236694,
-      "learning_rate": 8.14894472361809e-05,
-      "loss": 5.4753,
-      "step": 18924
-    },
-    {
-      "epoch": 9.869621903520208,
-      "grad_norm": 1.3423711061477661,
-      "learning_rate": 8.148844221105528e-05,
-      "loss": 5.4957,
-      "step": 18925
-    },
-    {
-      "epoch": 9.870143415906128,
-      "grad_norm": 1.5627321004867554,
-      "learning_rate": 8.148743718592964e-05,
-      "loss": 5.2467,
-      "step": 18926
-    },
-    {
-      "epoch": 9.870664928292047,
-      "grad_norm": 1.3703978061676025,
-      "learning_rate": 8.148643216080402e-05,
-      "loss": 5.6933,
-      "step": 18927
-    },
-    {
-      "epoch": 9.871186440677967,
-      "grad_norm": 1.450538992881775,
-      "learning_rate": 8.14854271356784e-05,
-      "loss": 5.6107,
-      "step": 18928
-    },
-    {
-      "epoch": 9.871707953063884,
-      "grad_norm": 1.4881075620651245,
-      "learning_rate": 8.148442211055278e-05,
-      "loss": 5.3103,
-      "step": 18929
-    },
-    {
-      "epoch": 9.872229465449804,
-      "grad_norm": 1.4314340353012085,
-      "learning_rate": 8.148341708542714e-05,
-      "loss": 5.5758,
-      "step": 18930
-    },
-    {
-      "epoch": 9.872750977835723,
-      "grad_norm": 1.535010576248169,
-      "learning_rate": 8.148241206030152e-05,
-      "loss": 5.5448,
-      "step": 18931
-    },
-    {
-      "epoch": 9.873272490221643,
-      "grad_norm": 1.4442405700683594,
-      "learning_rate": 8.148140703517588e-05,
-      "loss": 5.6248,
-      "step": 18932
-    },
-    {
-      "epoch": 9.873794002607562,
-      "grad_norm": 1.33511483669281,
-      "learning_rate": 8.148040201005025e-05,
-      "loss": 5.9814,
-      "step": 18933
-    },
-    {
-      "epoch": 9.874315514993482,
-      "grad_norm": 1.346980094909668,
-      "learning_rate": 8.147939698492463e-05,
-      "loss": 5.7066,
-      "step": 18934
-    },
-    {
-      "epoch": 9.8748370273794,
-      "grad_norm": 1.3210828304290771,
-      "learning_rate": 8.147839195979899e-05,
-      "loss": 5.9721,
-      "step": 18935
-    },
-    {
-      "epoch": 9.875358539765319,
-      "grad_norm": 1.4853821992874146,
-      "learning_rate": 8.147738693467337e-05,
-      "loss": 5.5041,
-      "step": 18936
-    },
-    {
-      "epoch": 9.875880052151238,
-      "grad_norm": 1.4116325378417969,
-      "learning_rate": 8.147638190954773e-05,
-      "loss": 5.3517,
-      "step": 18937
-    },
-    {
-      "epoch": 9.876401564537158,
-      "grad_norm": 1.4929753541946411,
-      "learning_rate": 8.147537688442211e-05,
-      "loss": 5.6231,
-      "step": 18938
-    },
-    {
-      "epoch": 9.876923076923077,
-      "grad_norm": 1.5282824039459229,
-      "learning_rate": 8.147437185929649e-05,
-      "loss": 5.4738,
-      "step": 18939
-    },
-    {
-      "epoch": 9.877444589308997,
-      "grad_norm": 1.4355461597442627,
-      "learning_rate": 8.147336683417087e-05,
-      "loss": 5.4087,
-      "step": 18940
-    },
-    {
-      "epoch": 9.877966101694915,
-      "grad_norm": 1.4045108556747437,
-      "learning_rate": 8.147236180904523e-05,
-      "loss": 5.753,
-      "step": 18941
-    },
-    {
-      "epoch": 9.878487614080834,
-      "grad_norm": 1.3695921897888184,
-      "learning_rate": 8.147135678391961e-05,
-      "loss": 5.7157,
-      "step": 18942
-    },
-    {
-      "epoch": 9.879009126466753,
-      "grad_norm": 1.511897325515747,
-      "learning_rate": 8.147035175879397e-05,
-      "loss": 5.5872,
-      "step": 18943
-    },
-    {
-      "epoch": 9.879530638852673,
-      "grad_norm": 1.4101191759109497,
-      "learning_rate": 8.146934673366835e-05,
-      "loss": 5.7312,
-      "step": 18944
-    },
-    {
-      "epoch": 9.880052151238592,
-      "grad_norm": 1.4767974615097046,
-      "learning_rate": 8.146834170854271e-05,
-      "loss": 5.1691,
-      "step": 18945
-    },
-    {
-      "epoch": 9.880573663624512,
-      "grad_norm": 1.4591628313064575,
-      "learning_rate": 8.146733668341709e-05,
-      "loss": 5.4313,
-      "step": 18946
-    },
-    {
-      "epoch": 9.88109517601043,
-      "grad_norm": 1.3642665147781372,
-      "learning_rate": 8.146633165829146e-05,
-      "loss": 5.5792,
-      "step": 18947
-    },
-    {
-      "epoch": 9.881616688396349,
-      "grad_norm": 1.4090399742126465,
-      "learning_rate": 8.146532663316583e-05,
-      "loss": 5.9452,
-      "step": 18948
-    },
-    {
-      "epoch": 9.882138200782268,
-      "grad_norm": 1.458201289176941,
-      "learning_rate": 8.146432160804021e-05,
-      "loss": 5.1397,
-      "step": 18949
-    },
-    {
-      "epoch": 9.882659713168188,
-      "grad_norm": 1.4682083129882812,
-      "learning_rate": 8.146331658291458e-05,
-      "loss": 4.8965,
-      "step": 18950
-    },
-    {
-      "epoch": 9.883181225554107,
-      "grad_norm": 1.3434338569641113,
-      "learning_rate": 8.146231155778895e-05,
-      "loss": 5.7222,
-      "step": 18951
-    },
-    {
-      "epoch": 9.883702737940027,
-      "grad_norm": 1.4034472703933716,
-      "learning_rate": 8.146130653266332e-05,
-      "loss": 5.2877,
-      "step": 18952
-    },
-    {
-      "epoch": 9.884224250325945,
-      "grad_norm": 1.4746540784835815,
-      "learning_rate": 8.14603015075377e-05,
-      "loss": 5.7083,
-      "step": 18953
-    },
-    {
-      "epoch": 9.884745762711864,
-      "grad_norm": 1.6290390491485596,
-      "learning_rate": 8.145929648241206e-05,
-      "loss": 5.4689,
-      "step": 18954
-    },
-    {
-      "epoch": 9.885267275097783,
-      "grad_norm": 1.4388952255249023,
-      "learning_rate": 8.145829145728644e-05,
-      "loss": 5.5161,
-      "step": 18955
-    },
-    {
-      "epoch": 9.885788787483703,
-      "grad_norm": 1.4674385786056519,
-      "learning_rate": 8.14572864321608e-05,
-      "loss": 5.5271,
-      "step": 18956
-    },
-    {
-      "epoch": 9.886310299869622,
-      "grad_norm": 1.400253415107727,
-      "learning_rate": 8.145628140703518e-05,
-      "loss": 5.4612,
-      "step": 18957
-    },
-    {
-      "epoch": 9.886831812255542,
-      "grad_norm": 1.420029878616333,
-      "learning_rate": 8.145527638190954e-05,
-      "loss": 5.4385,
-      "step": 18958
-    },
-    {
-      "epoch": 9.88735332464146,
-      "grad_norm": 1.3299713134765625,
-      "learning_rate": 8.145427135678392e-05,
-      "loss": 5.6023,
-      "step": 18959
-    },
-    {
-      "epoch": 9.887874837027379,
-      "grad_norm": 1.4464234113693237,
-      "learning_rate": 8.14532663316583e-05,
-      "loss": 5.0186,
-      "step": 18960
-    },
-    {
-      "epoch": 9.888396349413298,
-      "grad_norm": 1.492723822593689,
-      "learning_rate": 8.145226130653266e-05,
-      "loss": 5.6454,
-      "step": 18961
-    },
-    {
-      "epoch": 9.888917861799218,
-      "grad_norm": 1.315596580505371,
-      "learning_rate": 8.145125628140704e-05,
-      "loss": 5.5022,
-      "step": 18962
-    },
-    {
-      "epoch": 9.889439374185137,
-      "grad_norm": 1.3878294229507446,
-      "learning_rate": 8.14502512562814e-05,
-      "loss": 5.5521,
-      "step": 18963
-    },
-    {
-      "epoch": 9.889960886571057,
-      "grad_norm": 1.4243993759155273,
-      "learning_rate": 8.144924623115578e-05,
-      "loss": 5.8142,
-      "step": 18964
-    },
-    {
-      "epoch": 9.890482398956975,
-      "grad_norm": 1.3934370279312134,
-      "learning_rate": 8.144824120603015e-05,
-      "loss": 5.8063,
-      "step": 18965
-    },
-    {
-      "epoch": 9.891003911342894,
-      "grad_norm": 1.5747432708740234,
-      "learning_rate": 8.144723618090453e-05,
-      "loss": 5.1517,
-      "step": 18966
-    },
-    {
-      "epoch": 9.891525423728813,
-      "grad_norm": 1.6129409074783325,
-      "learning_rate": 8.144623115577889e-05,
-      "loss": 5.2656,
-      "step": 18967
-    },
-    {
-      "epoch": 9.892046936114733,
-      "grad_norm": 1.6949797868728638,
-      "learning_rate": 8.144522613065327e-05,
-      "loss": 5.5283,
-      "step": 18968
-    },
-    {
-      "epoch": 9.892568448500652,
-      "grad_norm": 1.5663963556289673,
-      "learning_rate": 8.144422110552765e-05,
-      "loss": 5.3111,
-      "step": 18969
-    },
-    {
-      "epoch": 9.893089960886572,
-      "grad_norm": 1.4378150701522827,
-      "learning_rate": 8.144321608040202e-05,
-      "loss": 5.4715,
-      "step": 18970
-    },
-    {
-      "epoch": 9.89361147327249,
-      "grad_norm": 1.432955265045166,
-      "learning_rate": 8.144221105527639e-05,
-      "loss": 5.4892,
-      "step": 18971
-    },
-    {
-      "epoch": 9.894132985658409,
-      "grad_norm": 1.3503034114837646,
-      "learning_rate": 8.144120603015077e-05,
-      "loss": 5.669,
-      "step": 18972
-    },
-    {
-      "epoch": 9.894654498044329,
-      "grad_norm": 1.3985849618911743,
-      "learning_rate": 8.144020100502513e-05,
-      "loss": 5.7261,
-      "step": 18973
-    },
-    {
-      "epoch": 9.895176010430248,
-      "grad_norm": 1.440733551979065,
-      "learning_rate": 8.14391959798995e-05,
-      "loss": 4.8851,
-      "step": 18974
-    },
-    {
-      "epoch": 9.895697522816167,
-      "grad_norm": 1.4635276794433594,
-      "learning_rate": 8.143819095477387e-05,
-      "loss": 5.7972,
-      "step": 18975
-    },
-    {
-      "epoch": 9.896219035202087,
-      "grad_norm": 1.3380005359649658,
-      "learning_rate": 8.143718592964824e-05,
-      "loss": 5.2589,
-      "step": 18976
-    },
-    {
-      "epoch": 9.896740547588005,
-      "grad_norm": 1.4112027883529663,
-      "learning_rate": 8.143618090452261e-05,
-      "loss": 5.7545,
-      "step": 18977
-    },
-    {
-      "epoch": 9.897262059973924,
-      "grad_norm": 1.5215460062026978,
-      "learning_rate": 8.143517587939698e-05,
-      "loss": 4.6125,
-      "step": 18978
-    },
-    {
-      "epoch": 9.897783572359844,
-      "grad_norm": 1.330257534980774,
-      "learning_rate": 8.143417085427136e-05,
-      "loss": 5.9892,
-      "step": 18979
-    },
-    {
-      "epoch": 9.898305084745763,
-      "grad_norm": 1.8168784379959106,
-      "learning_rate": 8.143316582914573e-05,
-      "loss": 5.2948,
-      "step": 18980
-    },
-    {
-      "epoch": 9.898826597131682,
-      "grad_norm": 1.2780791521072388,
-      "learning_rate": 8.143216080402011e-05,
-      "loss": 5.8721,
-      "step": 18981
-    },
-    {
-      "epoch": 9.8993481095176,
-      "grad_norm": 1.421470046043396,
-      "learning_rate": 8.143115577889448e-05,
-      "loss": 5.2903,
-      "step": 18982
-    },
-    {
-      "epoch": 9.89986962190352,
-      "grad_norm": 1.4439671039581299,
-      "learning_rate": 8.143015075376885e-05,
-      "loss": 5.499,
-      "step": 18983
-    },
-    {
-      "epoch": 9.900391134289439,
-      "grad_norm": 1.71378755569458,
-      "learning_rate": 8.142914572864322e-05,
-      "loss": 5.0259,
-      "step": 18984
-    },
-    {
-      "epoch": 9.900912646675359,
-      "grad_norm": 1.4430296421051025,
-      "learning_rate": 8.14281407035176e-05,
-      "loss": 5.4741,
-      "step": 18985
-    },
-    {
-      "epoch": 9.901434159061278,
-      "grad_norm": 1.4124850034713745,
-      "learning_rate": 8.142713567839196e-05,
-      "loss": 5.6969,
-      "step": 18986
-    },
-    {
-      "epoch": 9.901955671447197,
-      "grad_norm": 1.395836353302002,
-      "learning_rate": 8.142613065326633e-05,
-      "loss": 5.4566,
-      "step": 18987
-    },
-    {
-      "epoch": 9.902477183833117,
-      "grad_norm": 1.2927149534225464,
-      "learning_rate": 8.14251256281407e-05,
-      "loss": 5.8466,
-      "step": 18988
-    },
-    {
-      "epoch": 9.902998696219035,
-      "grad_norm": 1.9621624946594238,
-      "learning_rate": 8.142412060301508e-05,
-      "loss": 5.177,
-      "step": 18989
-    },
-    {
-      "epoch": 9.903520208604954,
-      "grad_norm": 1.398086667060852,
-      "learning_rate": 8.142311557788946e-05,
-      "loss": 5.4037,
-      "step": 18990
-    },
-    {
-      "epoch": 9.904041720990874,
-      "grad_norm": 1.6141467094421387,
-      "learning_rate": 8.142211055276382e-05,
-      "loss": 5.251,
-      "step": 18991
-    },
-    {
-      "epoch": 9.904563233376793,
-      "grad_norm": 1.3788906335830688,
-      "learning_rate": 8.14211055276382e-05,
-      "loss": 5.407,
-      "step": 18992
-    },
-    {
-      "epoch": 9.905084745762712,
-      "grad_norm": 1.5052052736282349,
-      "learning_rate": 8.142010050251257e-05,
-      "loss": 5.4253,
-      "step": 18993
-    },
-    {
-      "epoch": 9.90560625814863,
-      "grad_norm": 1.4209084510803223,
-      "learning_rate": 8.141909547738694e-05,
-      "loss": 5.3344,
-      "step": 18994
-    },
-    {
-      "epoch": 9.90612777053455,
-      "grad_norm": 1.3242473602294922,
-      "learning_rate": 8.141809045226131e-05,
-      "loss": 5.9352,
-      "step": 18995
-    },
-    {
-      "epoch": 9.906649282920469,
-      "grad_norm": 1.426491141319275,
-      "learning_rate": 8.141708542713569e-05,
-      "loss": 5.5666,
-      "step": 18996
-    },
-    {
-      "epoch": 9.907170795306389,
-      "grad_norm": 1.3522502183914185,
-      "learning_rate": 8.141608040201005e-05,
-      "loss": 5.601,
-      "step": 18997
-    },
-    {
-      "epoch": 9.907692307692308,
-      "grad_norm": 1.4910187721252441,
-      "learning_rate": 8.141507537688443e-05,
-      "loss": 5.2339,
-      "step": 18998
-    },
-    {
-      "epoch": 9.908213820078227,
-      "grad_norm": 1.5837610960006714,
-      "learning_rate": 8.141407035175879e-05,
-      "loss": 4.8941,
-      "step": 18999
-    },
-    {
-      "epoch": 9.908735332464147,
-      "grad_norm": 1.4559060335159302,
-      "learning_rate": 8.141306532663317e-05,
-      "loss": 5.6368,
-      "step": 19000
-    },
-    {
-      "epoch": 9.909256844850065,
-      "grad_norm": 1.4467158317565918,
-      "learning_rate": 8.141206030150755e-05,
-      "loss": 5.6214,
-      "step": 19001
-    },
-    {
-      "epoch": 9.909778357235984,
-      "grad_norm": 1.4220423698425293,
-      "learning_rate": 8.141105527638191e-05,
-      "loss": 5.2892,
-      "step": 19002
-    },
-    {
-      "epoch": 9.910299869621904,
-      "grad_norm": 1.4295992851257324,
-      "learning_rate": 8.141005025125629e-05,
-      "loss": 5.2701,
-      "step": 19003
-    },
-    {
-      "epoch": 9.910821382007823,
-      "grad_norm": 1.4433839321136475,
-      "learning_rate": 8.140904522613065e-05,
-      "loss": 5.5522,
-      "step": 19004
-    },
-    {
-      "epoch": 9.911342894393742,
-      "grad_norm": 1.5417948961257935,
-      "learning_rate": 8.140804020100503e-05,
-      "loss": 5.5151,
-      "step": 19005
-    },
-    {
-      "epoch": 9.91186440677966,
-      "grad_norm": 1.5292103290557861,
-      "learning_rate": 8.14070351758794e-05,
-      "loss": 5.0369,
-      "step": 19006
-    },
-    {
-      "epoch": 9.91238591916558,
-      "grad_norm": 1.4945679903030396,
-      "learning_rate": 8.140603015075377e-05,
-      "loss": 5.1697,
-      "step": 19007
-    },
-    {
-      "epoch": 9.9129074315515,
-      "grad_norm": 1.4211195707321167,
-      "learning_rate": 8.140502512562814e-05,
-      "loss": 5.594,
-      "step": 19008
-    },
-    {
-      "epoch": 9.913428943937419,
-      "grad_norm": 1.366811752319336,
-      "learning_rate": 8.140402010050252e-05,
-      "loss": 5.4322,
-      "step": 19009
-    },
-    {
-      "epoch": 9.913950456323338,
-      "grad_norm": 1.3830369710922241,
-      "learning_rate": 8.14030150753769e-05,
-      "loss": 5.9532,
-      "step": 19010
-    },
-    {
-      "epoch": 9.914471968709258,
-      "grad_norm": 1.4110532999038696,
-      "learning_rate": 8.140201005025127e-05,
-      "loss": 5.7933,
-      "step": 19011
-    },
-    {
-      "epoch": 9.914993481095175,
-      "grad_norm": 1.4042266607284546,
-      "learning_rate": 8.140100502512564e-05,
-      "loss": 5.5369,
-      "step": 19012
-    },
-    {
-      "epoch": 9.915514993481095,
-      "grad_norm": 1.3875893354415894,
-      "learning_rate": 8.14e-05,
-      "loss": 5.4546,
-      "step": 19013
-    },
-    {
-      "epoch": 9.916036505867014,
-      "grad_norm": 1.5292747020721436,
-      "learning_rate": 8.139899497487438e-05,
-      "loss": 4.7787,
-      "step": 19014
-    },
-    {
-      "epoch": 9.916558018252934,
-      "grad_norm": 1.4562453031539917,
-      "learning_rate": 8.139798994974874e-05,
-      "loss": 5.776,
-      "step": 19015
-    },
-    {
-      "epoch": 9.917079530638853,
-      "grad_norm": 1.4821418523788452,
-      "learning_rate": 8.139698492462312e-05,
-      "loss": 4.6333,
-      "step": 19016
-    },
-    {
-      "epoch": 9.917601043024773,
-      "grad_norm": 1.5525473356246948,
-      "learning_rate": 8.139597989949748e-05,
-      "loss": 5.5162,
-      "step": 19017
-    },
-    {
-      "epoch": 9.91812255541069,
-      "grad_norm": 1.346979022026062,
-      "learning_rate": 8.139497487437186e-05,
-      "loss": 5.6803,
-      "step": 19018
-    },
-    {
-      "epoch": 9.91864406779661,
-      "grad_norm": 1.3840863704681396,
-      "learning_rate": 8.139396984924623e-05,
-      "loss": 5.4334,
-      "step": 19019
-    },
-    {
-      "epoch": 9.91916558018253,
-      "grad_norm": 1.428637981414795,
-      "learning_rate": 8.13929648241206e-05,
-      "loss": 5.841,
-      "step": 19020
-    },
-    {
-      "epoch": 9.919687092568449,
-      "grad_norm": 1.471103310585022,
-      "learning_rate": 8.139195979899498e-05,
-      "loss": 5.5944,
-      "step": 19021
-    },
-    {
-      "epoch": 9.920208604954368,
-      "grad_norm": 1.3140798807144165,
-      "learning_rate": 8.139095477386936e-05,
-      "loss": 5.7029,
-      "step": 19022
-    },
-    {
-      "epoch": 9.920730117340288,
-      "grad_norm": 1.289729118347168,
-      "learning_rate": 8.138994974874372e-05,
-      "loss": 5.7942,
-      "step": 19023
-    },
-    {
-      "epoch": 9.921251629726205,
-      "grad_norm": 1.3748635053634644,
-      "learning_rate": 8.13889447236181e-05,
-      "loss": 5.6691,
-      "step": 19024
-    },
-    {
-      "epoch": 9.921773142112125,
-      "grad_norm": 1.4134958982467651,
-      "learning_rate": 8.138793969849247e-05,
-      "loss": 5.6204,
-      "step": 19025
-    },
-    {
-      "epoch": 9.922294654498044,
-      "grad_norm": 1.3892827033996582,
-      "learning_rate": 8.138693467336683e-05,
-      "loss": 5.6974,
-      "step": 19026
-    },
-    {
-      "epoch": 9.922816166883964,
-      "grad_norm": 1.4403830766677856,
-      "learning_rate": 8.138592964824121e-05,
-      "loss": 5.5608,
-      "step": 19027
-    },
-    {
-      "epoch": 9.923337679269883,
-      "grad_norm": 1.4532029628753662,
-      "learning_rate": 8.138492462311557e-05,
-      "loss": 5.6023,
-      "step": 19028
-    },
-    {
-      "epoch": 9.923859191655803,
-      "grad_norm": 1.4322328567504883,
-      "learning_rate": 8.138391959798995e-05,
-      "loss": 5.3622,
-      "step": 19029
-    },
-    {
-      "epoch": 9.92438070404172,
-      "grad_norm": 1.4148895740509033,
-      "learning_rate": 8.138291457286433e-05,
-      "loss": 5.5552,
-      "step": 19030
-    },
-    {
-      "epoch": 9.92490221642764,
-      "grad_norm": 1.4331471920013428,
-      "learning_rate": 8.13819095477387e-05,
-      "loss": 5.4715,
-      "step": 19031
-    },
-    {
-      "epoch": 9.92542372881356,
-      "grad_norm": 1.46225905418396,
-      "learning_rate": 8.138090452261307e-05,
-      "loss": 5.6782,
-      "step": 19032
-    },
-    {
-      "epoch": 9.925945241199479,
-      "grad_norm": 1.5385360717773438,
-      "learning_rate": 8.137989949748745e-05,
-      "loss": 5.6772,
-      "step": 19033
-    },
-    {
-      "epoch": 9.926466753585398,
-      "grad_norm": 1.5795198678970337,
-      "learning_rate": 8.137889447236181e-05,
-      "loss": 5.539,
-      "step": 19034
-    },
-    {
-      "epoch": 9.926988265971318,
-      "grad_norm": 1.3662545680999756,
-      "learning_rate": 8.137788944723619e-05,
-      "loss": 5.8319,
-      "step": 19035
-    },
-    {
-      "epoch": 9.927509778357235,
-      "grad_norm": 1.510064959526062,
-      "learning_rate": 8.137688442211055e-05,
-      "loss": 5.4328,
-      "step": 19036
-    },
-    {
-      "epoch": 9.928031290743155,
-      "grad_norm": 1.371141791343689,
-      "learning_rate": 8.137587939698493e-05,
-      "loss": 5.6232,
-      "step": 19037
-    },
-    {
-      "epoch": 9.928552803129074,
-      "grad_norm": 1.355199933052063,
-      "learning_rate": 8.13748743718593e-05,
-      "loss": 5.6193,
-      "step": 19038
-    },
-    {
-      "epoch": 9.929074315514994,
-      "grad_norm": 1.5269793272018433,
-      "learning_rate": 8.137386934673367e-05,
-      "loss": 5.0624,
-      "step": 19039
-    },
-    {
-      "epoch": 9.929595827900913,
-      "grad_norm": 1.4727648496627808,
-      "learning_rate": 8.137286432160804e-05,
-      "loss": 5.772,
-      "step": 19040
-    },
-    {
-      "epoch": 9.930117340286833,
-      "grad_norm": 1.4341942071914673,
-      "learning_rate": 8.137185929648242e-05,
-      "loss": 4.9072,
-      "step": 19041
-    },
-    {
-      "epoch": 9.93063885267275,
-      "grad_norm": 1.525680661201477,
-      "learning_rate": 8.13708542713568e-05,
-      "loss": 5.2146,
-      "step": 19042
-    },
-    {
-      "epoch": 9.93116036505867,
-      "grad_norm": 1.4286155700683594,
-      "learning_rate": 8.136984924623116e-05,
-      "loss": 5.2272,
-      "step": 19043
-    },
-    {
-      "epoch": 9.93168187744459,
-      "grad_norm": 1.4472020864486694,
-      "learning_rate": 8.136884422110554e-05,
-      "loss": 5.0355,
-      "step": 19044
-    },
-    {
-      "epoch": 9.932203389830509,
-      "grad_norm": 1.394162654876709,
-      "learning_rate": 8.13678391959799e-05,
-      "loss": 5.3629,
-      "step": 19045
-    },
-    {
-      "epoch": 9.932724902216428,
-      "grad_norm": 1.4716174602508545,
-      "learning_rate": 8.136683417085428e-05,
-      "loss": 5.4329,
-      "step": 19046
-    },
-    {
-      "epoch": 9.933246414602348,
-      "grad_norm": 1.4157112836837769,
-      "learning_rate": 8.136582914572864e-05,
-      "loss": 5.531,
-      "step": 19047
-    },
-    {
-      "epoch": 9.933767926988265,
-      "grad_norm": 1.4641704559326172,
-      "learning_rate": 8.136482412060302e-05,
-      "loss": 5.4517,
-      "step": 19048
-    },
-    {
-      "epoch": 9.934289439374185,
-      "grad_norm": 1.509461760520935,
-      "learning_rate": 8.136381909547738e-05,
-      "loss": 5.3161,
-      "step": 19049
-    },
-    {
-      "epoch": 9.934810951760104,
-      "grad_norm": 1.3616735935211182,
-      "learning_rate": 8.136281407035176e-05,
-      "loss": 5.8504,
-      "step": 19050
-    },
-    {
-      "epoch": 9.935332464146024,
-      "grad_norm": 1.3911982774734497,
-      "learning_rate": 8.136180904522614e-05,
-      "loss": 5.4846,
-      "step": 19051
-    },
-    {
-      "epoch": 9.935853976531943,
-      "grad_norm": 1.4084901809692383,
-      "learning_rate": 8.136080402010052e-05,
-      "loss": 5.7379,
-      "step": 19052
-    },
-    {
-      "epoch": 9.936375488917863,
-      "grad_norm": 1.4128212928771973,
-      "learning_rate": 8.135979899497488e-05,
-      "loss": 5.6189,
-      "step": 19053
-    },
-    {
-      "epoch": 9.93689700130378,
-      "grad_norm": 1.5238960981369019,
-      "learning_rate": 8.135879396984925e-05,
-      "loss": 5.1966,
-      "step": 19054
-    },
-    {
-      "epoch": 9.9374185136897,
-      "grad_norm": 1.4262605905532837,
-      "learning_rate": 8.135778894472362e-05,
-      "loss": 5.2772,
-      "step": 19055
-    },
-    {
-      "epoch": 9.93794002607562,
-      "grad_norm": 1.5504429340362549,
-      "learning_rate": 8.135678391959799e-05,
-      "loss": 5.4162,
-      "step": 19056
-    },
-    {
-      "epoch": 9.938461538461539,
-      "grad_norm": 1.5035852193832397,
-      "learning_rate": 8.135577889447237e-05,
-      "loss": 5.5948,
-      "step": 19057
-    },
-    {
-      "epoch": 9.938983050847458,
-      "grad_norm": 1.4751832485198975,
-      "learning_rate": 8.135477386934673e-05,
-      "loss": 5.3762,
-      "step": 19058
-    },
-    {
-      "epoch": 9.939504563233378,
-      "grad_norm": 1.5473871231079102,
-      "learning_rate": 8.135376884422111e-05,
-      "loss": 5.3947,
-      "step": 19059
-    },
-    {
-      "epoch": 9.940026075619295,
-      "grad_norm": 1.7073365449905396,
-      "learning_rate": 8.135276381909547e-05,
-      "loss": 5.3955,
-      "step": 19060
-    },
-    {
-      "epoch": 9.940547588005215,
-      "grad_norm": 1.4953656196594238,
-      "learning_rate": 8.135175879396985e-05,
-      "loss": 5.5613,
-      "step": 19061
-    },
-    {
-      "epoch": 9.941069100391134,
-      "grad_norm": 1.52162766456604,
-      "learning_rate": 8.135075376884423e-05,
-      "loss": 5.4654,
-      "step": 19062
-    },
-    {
-      "epoch": 9.941590612777054,
-      "grad_norm": 1.4691964387893677,
-      "learning_rate": 8.13497487437186e-05,
-      "loss": 5.523,
-      "step": 19063
-    },
-    {
-      "epoch": 9.942112125162973,
-      "grad_norm": 1.3093374967575073,
-      "learning_rate": 8.134874371859297e-05,
-      "loss": 5.8557,
-      "step": 19064
-    },
-    {
-      "epoch": 9.94263363754889,
-      "grad_norm": 1.3412202596664429,
-      "learning_rate": 8.134773869346735e-05,
-      "loss": 5.6495,
-      "step": 19065
-    },
-    {
-      "epoch": 9.94315514993481,
-      "grad_norm": 1.3825339078903198,
-      "learning_rate": 8.134673366834171e-05,
-      "loss": 5.512,
-      "step": 19066
-    },
-    {
-      "epoch": 9.94367666232073,
-      "grad_norm": 1.4201782941818237,
-      "learning_rate": 8.134572864321608e-05,
-      "loss": 4.9418,
-      "step": 19067
-    },
-    {
-      "epoch": 9.94419817470665,
-      "grad_norm": 1.3730721473693848,
-      "learning_rate": 8.134472361809046e-05,
-      "loss": 5.4788,
-      "step": 19068
-    },
-    {
-      "epoch": 9.944719687092569,
-      "grad_norm": 1.4039807319641113,
-      "learning_rate": 8.134371859296482e-05,
-      "loss": 5.7617,
-      "step": 19069
-    },
-    {
-      "epoch": 9.945241199478488,
-      "grad_norm": 1.56761634349823,
-      "learning_rate": 8.13427135678392e-05,
-      "loss": 5.2011,
-      "step": 19070
-    },
-    {
-      "epoch": 9.945762711864408,
-      "grad_norm": 1.5166455507278442,
-      "learning_rate": 8.134170854271356e-05,
-      "loss": 5.8872,
-      "step": 19071
-    },
-    {
-      "epoch": 9.946284224250325,
-      "grad_norm": 1.5359549522399902,
-      "learning_rate": 8.134070351758794e-05,
-      "loss": 5.1756,
-      "step": 19072
-    },
-    {
-      "epoch": 9.946805736636245,
-      "grad_norm": 1.5010124444961548,
-      "learning_rate": 8.133969849246232e-05,
-      "loss": 5.6014,
-      "step": 19073
-    },
-    {
-      "epoch": 9.947327249022164,
-      "grad_norm": 1.5179216861724854,
-      "learning_rate": 8.13386934673367e-05,
-      "loss": 4.9029,
-      "step": 19074
-    },
-    {
-      "epoch": 9.947848761408084,
-      "grad_norm": 1.416573166847229,
-      "learning_rate": 8.133768844221106e-05,
-      "loss": 5.483,
-      "step": 19075
-    },
-    {
-      "epoch": 9.948370273794003,
-      "grad_norm": 1.463661789894104,
-      "learning_rate": 8.133668341708544e-05,
-      "loss": 5.3683,
-      "step": 19076
-    },
-    {
-      "epoch": 9.94889178617992,
-      "grad_norm": 1.4426389932632446,
-      "learning_rate": 8.13356783919598e-05,
-      "loss": 5.0996,
-      "step": 19077
-    },
-    {
-      "epoch": 9.94941329856584,
-      "grad_norm": 1.3805052042007446,
-      "learning_rate": 8.133467336683418e-05,
-      "loss": 5.6084,
-      "step": 19078
-    },
-    {
-      "epoch": 9.94993481095176,
-      "grad_norm": 1.3703936338424683,
-      "learning_rate": 8.133366834170854e-05,
-      "loss": 5.4708,
-      "step": 19079
-    },
-    {
-      "epoch": 9.95045632333768,
-      "grad_norm": 1.4453959465026855,
-      "learning_rate": 8.133266331658291e-05,
-      "loss": 5.5801,
-      "step": 19080
-    },
-    {
-      "epoch": 9.950977835723599,
-      "grad_norm": 1.4377617835998535,
-      "learning_rate": 8.133165829145729e-05,
-      "loss": 5.4107,
-      "step": 19081
-    },
-    {
-      "epoch": 9.951499348109518,
-      "grad_norm": 1.5398176908493042,
-      "learning_rate": 8.133065326633166e-05,
-      "loss": 5.36,
-      "step": 19082
-    },
-    {
-      "epoch": 9.952020860495438,
-      "grad_norm": 1.3322324752807617,
-      "learning_rate": 8.132964824120604e-05,
-      "loss": 5.8526,
-      "step": 19083
-    },
-    {
-      "epoch": 9.952542372881355,
-      "grad_norm": 1.4438626766204834,
-      "learning_rate": 8.13286432160804e-05,
-      "loss": 5.248,
-      "step": 19084
-    },
-    {
-      "epoch": 9.953063885267275,
-      "grad_norm": 1.5246882438659668,
-      "learning_rate": 8.132763819095478e-05,
-      "loss": 5.4507,
-      "step": 19085
-    },
-    {
-      "epoch": 9.953585397653194,
-      "grad_norm": 1.5473477840423584,
-      "learning_rate": 8.132663316582915e-05,
-      "loss": 5.057,
-      "step": 19086
-    },
-    {
-      "epoch": 9.954106910039114,
-      "grad_norm": 1.3806309700012207,
-      "learning_rate": 8.132562814070353e-05,
-      "loss": 5.9723,
-      "step": 19087
-    },
-    {
-      "epoch": 9.954628422425033,
-      "grad_norm": 1.5173453092575073,
-      "learning_rate": 8.132462311557789e-05,
-      "loss": 5.0633,
-      "step": 19088
-    },
-    {
-      "epoch": 9.955149934810951,
-      "grad_norm": 1.3269238471984863,
-      "learning_rate": 8.132361809045227e-05,
-      "loss": 5.3715,
-      "step": 19089
-    },
-    {
-      "epoch": 9.95567144719687,
-      "grad_norm": 1.367398977279663,
-      "learning_rate": 8.132261306532663e-05,
-      "loss": 5.333,
-      "step": 19090
-    },
-    {
-      "epoch": 9.95619295958279,
-      "grad_norm": 1.4719958305358887,
-      "learning_rate": 8.132160804020101e-05,
-      "loss": 5.5071,
-      "step": 19091
-    },
-    {
-      "epoch": 9.95671447196871,
-      "grad_norm": 1.3012548685073853,
-      "learning_rate": 8.132060301507537e-05,
-      "loss": 5.599,
-      "step": 19092
-    },
-    {
-      "epoch": 9.957235984354629,
-      "grad_norm": 1.501981258392334,
-      "learning_rate": 8.131959798994975e-05,
-      "loss": 5.3871,
-      "step": 19093
-    },
-    {
-      "epoch": 9.957757496740548,
-      "grad_norm": 1.3773146867752075,
-      "learning_rate": 8.131859296482413e-05,
-      "loss": 5.3722,
-      "step": 19094
-    },
-    {
-      "epoch": 9.958279009126468,
-      "grad_norm": 1.4484068155288696,
-      "learning_rate": 8.13175879396985e-05,
-      "loss": 5.5825,
-      "step": 19095
-    },
-    {
-      "epoch": 9.958800521512385,
-      "grad_norm": 1.4462921619415283,
-      "learning_rate": 8.131658291457287e-05,
-      "loss": 5.8939,
-      "step": 19096
-    },
-    {
-      "epoch": 9.959322033898305,
-      "grad_norm": 1.4065076112747192,
-      "learning_rate": 8.131557788944724e-05,
-      "loss": 5.4214,
-      "step": 19097
-    },
-    {
-      "epoch": 9.959843546284224,
-      "grad_norm": 1.4755072593688965,
-      "learning_rate": 8.131457286432161e-05,
-      "loss": 5.4905,
-      "step": 19098
-    },
-    {
-      "epoch": 9.960365058670144,
-      "grad_norm": 1.3575702905654907,
-      "learning_rate": 8.131356783919598e-05,
-      "loss": 5.7864,
-      "step": 19099
-    },
-    {
-      "epoch": 9.960886571056063,
-      "grad_norm": 1.2982832193374634,
-      "learning_rate": 8.131256281407036e-05,
-      "loss": 5.281,
-      "step": 19100
-    },
-    {
-      "epoch": 9.961408083441981,
-      "grad_norm": 1.420318603515625,
-      "learning_rate": 8.131155778894472e-05,
-      "loss": 5.4519,
-      "step": 19101
-    },
-    {
-      "epoch": 9.9619295958279,
-      "grad_norm": 1.3651236295700073,
-      "learning_rate": 8.13105527638191e-05,
-      "loss": 5.6365,
-      "step": 19102
-    },
-    {
-      "epoch": 9.96245110821382,
-      "grad_norm": 1.3081884384155273,
-      "learning_rate": 8.130954773869348e-05,
-      "loss": 5.8474,
-      "step": 19103
-    },
-    {
-      "epoch": 9.96297262059974,
-      "grad_norm": 1.3793697357177734,
-      "learning_rate": 8.130854271356785e-05,
-      "loss": 5.4978,
-      "step": 19104
-    },
-    {
-      "epoch": 9.963494132985659,
-      "grad_norm": 1.3946070671081543,
-      "learning_rate": 8.130753768844222e-05,
-      "loss": 5.3388,
-      "step": 19105
-    },
-    {
-      "epoch": 9.964015645371578,
-      "grad_norm": 1.403092861175537,
-      "learning_rate": 8.130653266331658e-05,
-      "loss": 5.5323,
-      "step": 19106
-    },
-    {
-      "epoch": 9.964537157757496,
-      "grad_norm": 1.5760133266448975,
-      "learning_rate": 8.130552763819096e-05,
-      "loss": 5.6037,
-      "step": 19107
-    },
-    {
-      "epoch": 9.965058670143415,
-      "grad_norm": 1.443266749382019,
-      "learning_rate": 8.130452261306532e-05,
-      "loss": 5.1146,
-      "step": 19108
-    },
-    {
-      "epoch": 9.965580182529335,
-      "grad_norm": 1.3929591178894043,
-      "learning_rate": 8.13035175879397e-05,
-      "loss": 5.5923,
-      "step": 19109
-    },
-    {
-      "epoch": 9.966101694915254,
-      "grad_norm": 1.5283291339874268,
-      "learning_rate": 8.130251256281407e-05,
-      "loss": 5.2434,
-      "step": 19110
-    },
-    {
-      "epoch": 9.966623207301174,
-      "grad_norm": 1.3588038682937622,
-      "learning_rate": 8.130150753768844e-05,
-      "loss": 5.7242,
-      "step": 19111
-    },
-    {
-      "epoch": 9.967144719687093,
-      "grad_norm": 1.527885913848877,
-      "learning_rate": 8.130050251256281e-05,
-      "loss": 5.2507,
-      "step": 19112
-    },
-    {
-      "epoch": 9.967666232073011,
-      "grad_norm": 1.4081473350524902,
-      "learning_rate": 8.129949748743719e-05,
-      "loss": 5.7167,
-      "step": 19113
-    },
-    {
-      "epoch": 9.96818774445893,
-      "grad_norm": 1.514689326286316,
-      "learning_rate": 8.129849246231156e-05,
-      "loss": 5.3345,
-      "step": 19114
-    },
-    {
-      "epoch": 9.96870925684485,
-      "grad_norm": 1.3903493881225586,
-      "learning_rate": 8.129748743718594e-05,
-      "loss": 5.7088,
-      "step": 19115
-    },
-    {
-      "epoch": 9.96923076923077,
-      "grad_norm": 1.497959852218628,
-      "learning_rate": 8.12964824120603e-05,
-      "loss": 5.4127,
-      "step": 19116
-    },
-    {
-      "epoch": 9.969752281616689,
-      "grad_norm": 1.5525034666061401,
-      "learning_rate": 8.129547738693468e-05,
-      "loss": 5.6004,
-      "step": 19117
-    },
-    {
-      "epoch": 9.970273794002608,
-      "grad_norm": 1.4513038396835327,
-      "learning_rate": 8.129447236180905e-05,
-      "loss": 5.7249,
-      "step": 19118
-    },
-    {
-      "epoch": 9.970795306388526,
-      "grad_norm": 1.483587622642517,
-      "learning_rate": 8.129346733668343e-05,
-      "loss": 5.2588,
-      "step": 19119
-    },
-    {
-      "epoch": 9.971316818774445,
-      "grad_norm": 1.284005045890808,
-      "learning_rate": 8.129246231155779e-05,
-      "loss": 5.8445,
-      "step": 19120
-    },
-    {
-      "epoch": 9.971838331160365,
-      "grad_norm": 1.5742162466049194,
-      "learning_rate": 8.129145728643215e-05,
-      "loss": 4.8139,
-      "step": 19121
-    },
-    {
-      "epoch": 9.972359843546284,
-      "grad_norm": 1.6167311668395996,
-      "learning_rate": 8.129045226130653e-05,
-      "loss": 5.5813,
-      "step": 19122
-    },
-    {
-      "epoch": 9.972881355932204,
-      "grad_norm": 1.394535779953003,
-      "learning_rate": 8.128944723618091e-05,
-      "loss": 5.3446,
-      "step": 19123
-    },
-    {
-      "epoch": 9.973402868318123,
-      "grad_norm": 1.4823797941207886,
-      "learning_rate": 8.128844221105529e-05,
-      "loss": 4.4942,
-      "step": 19124
-    },
-    {
-      "epoch": 9.973924380704041,
-      "grad_norm": 1.344460368156433,
-      "learning_rate": 8.128743718592965e-05,
-      "loss": 5.5226,
-      "step": 19125
-    },
-    {
-      "epoch": 9.97444589308996,
-      "grad_norm": 1.4834285974502563,
-      "learning_rate": 8.128643216080403e-05,
-      "loss": 5.4989,
-      "step": 19126
-    },
-    {
-      "epoch": 9.97496740547588,
-      "grad_norm": 1.3655022382736206,
-      "learning_rate": 8.12854271356784e-05,
-      "loss": 5.1271,
-      "step": 19127
-    },
-    {
-      "epoch": 9.9754889178618,
-      "grad_norm": 1.428627610206604,
-      "learning_rate": 8.128442211055277e-05,
-      "loss": 5.2347,
-      "step": 19128
-    },
-    {
-      "epoch": 9.976010430247719,
-      "grad_norm": 1.4022146463394165,
-      "learning_rate": 8.128341708542714e-05,
-      "loss": 5.8174,
-      "step": 19129
-    },
-    {
-      "epoch": 9.976531942633638,
-      "grad_norm": 1.3465352058410645,
-      "learning_rate": 8.128241206030151e-05,
-      "loss": 5.7361,
-      "step": 19130
-    },
-    {
-      "epoch": 9.977053455019556,
-      "grad_norm": 1.331575870513916,
-      "learning_rate": 8.128140703517588e-05,
-      "loss": 5.7451,
-      "step": 19131
-    },
-    {
-      "epoch": 9.977574967405475,
-      "grad_norm": 1.3546241521835327,
-      "learning_rate": 8.128040201005026e-05,
-      "loss": 5.5175,
-      "step": 19132
-    },
-    {
-      "epoch": 9.978096479791395,
-      "grad_norm": 1.4345918893814087,
-      "learning_rate": 8.127939698492462e-05,
-      "loss": 5.3407,
-      "step": 19133
-    },
-    {
-      "epoch": 9.978617992177314,
-      "grad_norm": 1.356874704360962,
-      "learning_rate": 8.1278391959799e-05,
-      "loss": 5.389,
-      "step": 19134
-    },
-    {
-      "epoch": 9.979139504563234,
-      "grad_norm": 1.4534032344818115,
-      "learning_rate": 8.127738693467338e-05,
-      "loss": 5.4185,
-      "step": 19135
-    },
-    {
-      "epoch": 9.979661016949153,
-      "grad_norm": 1.3718070983886719,
-      "learning_rate": 8.127638190954774e-05,
-      "loss": 5.5759,
-      "step": 19136
-    },
-    {
-      "epoch": 9.980182529335071,
-      "grad_norm": 1.3757226467132568,
-      "learning_rate": 8.127537688442212e-05,
-      "loss": 5.3197,
-      "step": 19137
-    },
-    {
-      "epoch": 9.98070404172099,
-      "grad_norm": 1.4161653518676758,
-      "learning_rate": 8.127437185929648e-05,
-      "loss": 5.5569,
-      "step": 19138
-    },
-    {
-      "epoch": 9.98122555410691,
-      "grad_norm": 1.432983636856079,
-      "learning_rate": 8.127336683417086e-05,
-      "loss": 5.288,
-      "step": 19139
-    },
-    {
-      "epoch": 9.98174706649283,
-      "grad_norm": 1.389546275138855,
-      "learning_rate": 8.127236180904523e-05,
-      "loss": 5.4848,
-      "step": 19140
-    },
-    {
-      "epoch": 9.982268578878749,
-      "grad_norm": 1.4189705848693848,
-      "learning_rate": 8.12713567839196e-05,
-      "loss": 5.5295,
-      "step": 19141
-    },
-    {
-      "epoch": 9.982790091264668,
-      "grad_norm": 1.4098879098892212,
-      "learning_rate": 8.127035175879397e-05,
-      "loss": 5.7444,
-      "step": 19142
-    },
-    {
-      "epoch": 9.983311603650586,
-      "grad_norm": 1.406623363494873,
-      "learning_rate": 8.126934673366835e-05,
-      "loss": 5.4983,
-      "step": 19143
-    },
-    {
-      "epoch": 9.983833116036505,
-      "grad_norm": 1.4388296604156494,
-      "learning_rate": 8.126834170854272e-05,
-      "loss": 5.5782,
-      "step": 19144
-    },
-    {
-      "epoch": 9.984354628422425,
-      "grad_norm": 1.3847720623016357,
-      "learning_rate": 8.12673366834171e-05,
-      "loss": 5.5961,
-      "step": 19145
-    },
-    {
-      "epoch": 9.984876140808344,
-      "grad_norm": 1.3790737390518188,
-      "learning_rate": 8.126633165829146e-05,
-      "loss": 5.3601,
-      "step": 19146
-    },
-    {
-      "epoch": 9.985397653194264,
-      "grad_norm": 1.4118373394012451,
-      "learning_rate": 8.126532663316583e-05,
-      "loss": 5.473,
-      "step": 19147
-    },
-    {
-      "epoch": 9.985919165580183,
-      "grad_norm": 1.3063966035842896,
-      "learning_rate": 8.126432160804021e-05,
-      "loss": 5.806,
-      "step": 19148
-    },
-    {
-      "epoch": 9.986440677966101,
-      "grad_norm": 1.3166313171386719,
-      "learning_rate": 8.126331658291457e-05,
-      "loss": 5.4967,
-      "step": 19149
-    },
-    {
-      "epoch": 9.98696219035202,
-      "grad_norm": 1.5583546161651611,
-      "learning_rate": 8.126231155778895e-05,
-      "loss": 5.1759,
-      "step": 19150
-    },
-    {
-      "epoch": 9.98748370273794,
-      "grad_norm": 1.504077672958374,
-      "learning_rate": 8.126130653266331e-05,
-      "loss": 5.6572,
-      "step": 19151
-    },
-    {
-      "epoch": 9.98800521512386,
-      "grad_norm": 1.3517287969589233,
-      "learning_rate": 8.126030150753769e-05,
-      "loss": 5.6065,
-      "step": 19152
-    },
-    {
-      "epoch": 9.988526727509779,
-      "grad_norm": 1.460100769996643,
-      "learning_rate": 8.125929648241206e-05,
-      "loss": 5.6533,
-      "step": 19153
-    },
-    {
-      "epoch": 9.989048239895698,
-      "grad_norm": 1.320932388305664,
-      "learning_rate": 8.125829145728643e-05,
-      "loss": 5.8114,
-      "step": 19154
-    },
-    {
-      "epoch": 9.989569752281616,
-      "grad_norm": 1.3253446817398071,
-      "learning_rate": 8.125728643216081e-05,
-      "loss": 5.4911,
-      "step": 19155
-    },
-    {
-      "epoch": 9.990091264667535,
-      "grad_norm": 1.4313970804214478,
-      "learning_rate": 8.125628140703519e-05,
-      "loss": 5.6992,
-      "step": 19156
-    },
-    {
-      "epoch": 9.990612777053455,
-      "grad_norm": 1.4552379846572876,
-      "learning_rate": 8.125527638190955e-05,
-      "loss": 5.3911,
-      "step": 19157
-    },
-    {
-      "epoch": 9.991134289439374,
-      "grad_norm": 1.6307376623153687,
-      "learning_rate": 8.125427135678393e-05,
-      "loss": 5.1911,
-      "step": 19158
-    },
-    {
-      "epoch": 9.991655801825294,
-      "grad_norm": 1.400229573249817,
-      "learning_rate": 8.12532663316583e-05,
-      "loss": 5.6716,
-      "step": 19159
-    },
-    {
-      "epoch": 9.992177314211212,
-      "grad_norm": 1.3206816911697388,
-      "learning_rate": 8.125226130653266e-05,
-      "loss": 5.5433,
-      "step": 19160
-    },
-    {
-      "epoch": 9.992698826597131,
-      "grad_norm": 1.6872576475143433,
-      "learning_rate": 8.125125628140704e-05,
-      "loss": 5.1651,
-      "step": 19161
-    },
-    {
-      "epoch": 9.99322033898305,
-      "grad_norm": 1.4041895866394043,
-      "learning_rate": 8.12502512562814e-05,
-      "loss": 5.4849,
-      "step": 19162
-    },
-    {
-      "epoch": 9.99374185136897,
-      "grad_norm": 1.4195914268493652,
-      "learning_rate": 8.124924623115578e-05,
-      "loss": 5.4219,
-      "step": 19163
-    },
-    {
-      "epoch": 9.99426336375489,
-      "grad_norm": 1.5154247283935547,
-      "learning_rate": 8.124824120603016e-05,
-      "loss": 5.6025,
-      "step": 19164
-    },
-    {
-      "epoch": 9.994784876140809,
-      "grad_norm": 1.4071340560913086,
-      "learning_rate": 8.124723618090454e-05,
-      "loss": 4.9277,
-      "step": 19165
-    },
-    {
-      "epoch": 9.995306388526728,
-      "grad_norm": 1.4853787422180176,
-      "learning_rate": 8.12462311557789e-05,
-      "loss": 5.3997,
-      "step": 19166
-    },
-    {
-      "epoch": 9.995827900912646,
-      "grad_norm": 1.794089674949646,
-      "learning_rate": 8.124522613065328e-05,
-      "loss": 4.8875,
-      "step": 19167
-    },
-    {
-      "epoch": 9.996349413298566,
-      "grad_norm": 1.427100658416748,
-      "learning_rate": 8.124422110552764e-05,
-      "loss": 5.7298,
-      "step": 19168
-    },
-    {
-      "epoch": 9.996870925684485,
-      "grad_norm": 1.4734169244766235,
-      "learning_rate": 8.124321608040202e-05,
-      "loss": 5.3671,
-      "step": 19169
-    },
-    {
-      "epoch": 9.997392438070404,
-      "grad_norm": 1.436055302619934,
-      "learning_rate": 8.124221105527638e-05,
-      "loss": 5.5048,
-      "step": 19170
-    },
-    {
-      "epoch": 9.997913950456324,
-      "grad_norm": 1.3756427764892578,
-      "learning_rate": 8.124120603015076e-05,
-      "loss": 5.7793,
-      "step": 19171
-    },
-    {
-      "epoch": 9.998435462842242,
-      "grad_norm": 1.5057133436203003,
-      "learning_rate": 8.124020100502513e-05,
-      "loss": 5.406,
-      "step": 19172
-    },
-    {
-      "epoch": 9.998956975228161,
-      "grad_norm": 1.444242000579834,
-      "learning_rate": 8.123919597989949e-05,
-      "loss": 5.2039,
-      "step": 19173
-    },
-    {
-      "epoch": 9.99947848761408,
-      "grad_norm": 1.3665419816970825,
-      "learning_rate": 8.123819095477387e-05,
-      "loss": 5.6518,
-      "step": 19174
-    },
-    {
-      "epoch": 10.0,
-      "grad_norm": 1.4895446300506592,
-      "learning_rate": 8.123718592964825e-05,
-      "loss": 5.7852,
-      "step": 19175
-    },
-    {
-      "epoch": 10.00052151238592,
-      "grad_norm": 1.3535881042480469,
-      "learning_rate": 8.123618090452262e-05,
-      "loss": 5.4831,
-      "step": 19176
-    },
-    {
-      "epoch": 10.001043024771839,
-      "grad_norm": 1.5101163387298584,
-      "learning_rate": 8.123517587939699e-05,
-      "loss": 4.9482,
-      "step": 19177
-    },
-    {
-      "epoch": 10.001564537157757,
-      "grad_norm": 1.3609122037887573,
-      "learning_rate": 8.123417085427137e-05,
-      "loss": 5.4377,
-      "step": 19178
-    },
-    {
-      "epoch": 10.002086049543676,
-      "grad_norm": 1.4885094165802002,
-      "learning_rate": 8.123316582914573e-05,
-      "loss": 4.9194,
-      "step": 19179
-    },
-    {
-      "epoch": 10.002607561929596,
-      "grad_norm": 1.359678864479065,
-      "learning_rate": 8.123216080402011e-05,
-      "loss": 5.2817,
-      "step": 19180
-    },
-    {
-      "epoch": 10.003129074315515,
-      "grad_norm": 1.4171149730682373,
-      "learning_rate": 8.123115577889447e-05,
-      "loss": 5.2793,
-      "step": 19181
-    },
-    {
-      "epoch": 10.003650586701434,
-      "grad_norm": 1.406046748161316,
-      "learning_rate": 8.123015075376885e-05,
-      "loss": 5.5008,
-      "step": 19182
-    },
-    {
-      "epoch": 10.004172099087354,
-      "grad_norm": 1.5066537857055664,
-      "learning_rate": 8.122914572864321e-05,
-      "loss": 5.5782,
-      "step": 19183
-    },
-    {
-      "epoch": 10.004693611473272,
-      "grad_norm": 1.3261677026748657,
-      "learning_rate": 8.122814070351759e-05,
-      "loss": 5.7969,
-      "step": 19184
-    },
-    {
-      "epoch": 10.005215123859191,
-      "grad_norm": 1.5266515016555786,
-      "learning_rate": 8.122713567839197e-05,
-      "loss": 5.0743,
-      "step": 19185
-    },
-    {
-      "epoch": 10.00573663624511,
-      "grad_norm": 1.331714391708374,
-      "learning_rate": 8.122613065326633e-05,
-      "loss": 5.8745,
-      "step": 19186
-    },
-    {
-      "epoch": 10.00625814863103,
-      "grad_norm": 1.3899743556976318,
-      "learning_rate": 8.122512562814071e-05,
-      "loss": 5.7709,
-      "step": 19187
-    },
-    {
-      "epoch": 10.00677966101695,
-      "grad_norm": 1.5344220399856567,
-      "learning_rate": 8.122412060301508e-05,
-      "loss": 5.0891,
-      "step": 19188
-    },
-    {
-      "epoch": 10.007301173402869,
-      "grad_norm": 1.4051916599273682,
-      "learning_rate": 8.122311557788945e-05,
-      "loss": 5.7676,
-      "step": 19189
-    },
-    {
-      "epoch": 10.007822685788787,
-      "grad_norm": 1.4859658479690552,
-      "learning_rate": 8.122211055276382e-05,
-      "loss": 5.1602,
-      "step": 19190
-    },
-    {
-      "epoch": 10.008344198174706,
-      "grad_norm": 1.427297592163086,
-      "learning_rate": 8.12211055276382e-05,
-      "loss": 5.5231,
-      "step": 19191
-    },
-    {
-      "epoch": 10.008865710560626,
-      "grad_norm": 1.5073884725570679,
-      "learning_rate": 8.122010050251256e-05,
-      "loss": 5.5955,
-      "step": 19192
-    },
-    {
-      "epoch": 10.009387222946545,
-      "grad_norm": 1.3651376962661743,
-      "learning_rate": 8.121909547738694e-05,
-      "loss": 5.4368,
-      "step": 19193
-    },
-    {
-      "epoch": 10.009908735332465,
-      "grad_norm": 1.4432231187820435,
-      "learning_rate": 8.12180904522613e-05,
-      "loss": 5.4201,
-      "step": 19194
-    },
-    {
-      "epoch": 10.010430247718384,
-      "grad_norm": 1.496381163597107,
-      "learning_rate": 8.121708542713568e-05,
-      "loss": 5.1609,
-      "step": 19195
-    },
-    {
-      "epoch": 10.010951760104302,
-      "grad_norm": 1.4189872741699219,
-      "learning_rate": 8.121608040201006e-05,
-      "loss": 5.6764,
-      "step": 19196
-    },
-    {
-      "epoch": 10.011473272490221,
-      "grad_norm": 1.487499713897705,
-      "learning_rate": 8.121507537688444e-05,
-      "loss": 5.4268,
-      "step": 19197
-    },
-    {
-      "epoch": 10.01199478487614,
-      "grad_norm": 1.4959497451782227,
-      "learning_rate": 8.12140703517588e-05,
-      "loss": 5.6745,
-      "step": 19198
-    },
-    {
-      "epoch": 10.01251629726206,
-      "grad_norm": 1.5172396898269653,
-      "learning_rate": 8.121306532663316e-05,
-      "loss": 5.4773,
-      "step": 19199
-    },
-    {
-      "epoch": 10.01303780964798,
-      "grad_norm": 1.5348068475723267,
-      "learning_rate": 8.121206030150754e-05,
-      "loss": 5.5291,
-      "step": 19200
-    },
-    {
-      "epoch": 10.013559322033899,
-      "grad_norm": 1.4773908853530884,
-      "learning_rate": 8.12110552763819e-05,
-      "loss": 5.3614,
-      "step": 19201
-    },
-    {
-      "epoch": 10.014080834419817,
-      "grad_norm": 1.494300365447998,
-      "learning_rate": 8.121005025125628e-05,
-      "loss": 5.3198,
-      "step": 19202
-    },
-    {
-      "epoch": 10.014602346805736,
-      "grad_norm": 1.4780830144882202,
-      "learning_rate": 8.120904522613065e-05,
-      "loss": 5.5639,
-      "step": 19203
-    },
-    {
-      "epoch": 10.015123859191656,
-      "grad_norm": 1.4712142944335938,
-      "learning_rate": 8.120804020100503e-05,
-      "loss": 5.4654,
-      "step": 19204
-    },
-    {
-      "epoch": 10.015645371577575,
-      "grad_norm": 1.4856996536254883,
-      "learning_rate": 8.12070351758794e-05,
-      "loss": 4.9498,
-      "step": 19205
-    },
-    {
-      "epoch": 10.016166883963495,
-      "grad_norm": 1.514612078666687,
-      "learning_rate": 8.120603015075378e-05,
-      "loss": 5.2722,
-      "step": 19206
-    },
-    {
-      "epoch": 10.016688396349414,
-      "grad_norm": 1.4236782789230347,
-      "learning_rate": 8.120502512562815e-05,
-      "loss": 5.5055,
-      "step": 19207
-    },
-    {
-      "epoch": 10.017209908735332,
-      "grad_norm": 1.4451059103012085,
-      "learning_rate": 8.120402010050252e-05,
-      "loss": 5.618,
-      "step": 19208
-    },
-    {
-      "epoch": 10.017731421121251,
-      "grad_norm": 1.487575650215149,
-      "learning_rate": 8.120301507537689e-05,
-      "loss": 5.589,
-      "step": 19209
-    },
-    {
-      "epoch": 10.01825293350717,
-      "grad_norm": 1.370160460472107,
-      "learning_rate": 8.120201005025127e-05,
-      "loss": 5.2558,
-      "step": 19210
-    },
-    {
-      "epoch": 10.01877444589309,
-      "grad_norm": 1.4894109964370728,
-      "learning_rate": 8.120100502512563e-05,
-      "loss": 4.8025,
-      "step": 19211
-    },
-    {
-      "epoch": 10.01929595827901,
-      "grad_norm": 1.4270145893096924,
-      "learning_rate": 8.120000000000001e-05,
-      "loss": 5.447,
-      "step": 19212
-    },
-    {
-      "epoch": 10.019817470664929,
-      "grad_norm": 1.5027819871902466,
-      "learning_rate": 8.119899497487437e-05,
-      "loss": 5.3873,
-      "step": 19213
-    },
-    {
-      "epoch": 10.020338983050847,
-      "grad_norm": 1.4780253171920776,
-      "learning_rate": 8.119798994974874e-05,
-      "loss": 5.4928,
-      "step": 19214
-    },
-    {
-      "epoch": 10.020860495436766,
-      "grad_norm": 1.502392292022705,
-      "learning_rate": 8.119698492462311e-05,
-      "loss": 5.5449,
-      "step": 19215
-    },
-    {
-      "epoch": 10.021382007822686,
-      "grad_norm": 1.3431240320205688,
-      "learning_rate": 8.119597989949749e-05,
-      "loss": 5.7271,
-      "step": 19216
-    },
-    {
-      "epoch": 10.021903520208605,
-      "grad_norm": 1.4712812900543213,
-      "learning_rate": 8.119497487437187e-05,
-      "loss": 5.5106,
-      "step": 19217
-    },
-    {
-      "epoch": 10.022425032594525,
-      "grad_norm": 1.4397385120391846,
-      "learning_rate": 8.119396984924623e-05,
-      "loss": 5.4591,
-      "step": 19218
-    },
-    {
-      "epoch": 10.022946544980444,
-      "grad_norm": 1.3612157106399536,
-      "learning_rate": 8.119296482412061e-05,
-      "loss": 5.6743,
-      "step": 19219
-    },
-    {
-      "epoch": 10.023468057366362,
-      "grad_norm": 1.4182382822036743,
-      "learning_rate": 8.119195979899498e-05,
-      "loss": 5.3529,
-      "step": 19220
-    },
-    {
-      "epoch": 10.023989569752281,
-      "grad_norm": 1.4806327819824219,
-      "learning_rate": 8.119095477386935e-05,
-      "loss": 5.5076,
-      "step": 19221
-    },
-    {
-      "epoch": 10.0245110821382,
-      "grad_norm": 1.3532460927963257,
-      "learning_rate": 8.118994974874372e-05,
-      "loss": 5.7604,
-      "step": 19222
-    },
-    {
-      "epoch": 10.02503259452412,
-      "grad_norm": 1.7349433898925781,
-      "learning_rate": 8.11889447236181e-05,
-      "loss": 5.2048,
-      "step": 19223
-    },
-    {
-      "epoch": 10.02555410691004,
-      "grad_norm": 1.383598804473877,
-      "learning_rate": 8.118793969849246e-05,
-      "loss": 5.8589,
-      "step": 19224
-    },
-    {
-      "epoch": 10.026075619295959,
-      "grad_norm": 1.5332878828048706,
-      "learning_rate": 8.118693467336684e-05,
-      "loss": 5.2024,
-      "step": 19225
-    },
-    {
-      "epoch": 10.026597131681877,
-      "grad_norm": 1.41984224319458,
-      "learning_rate": 8.118592964824122e-05,
-      "loss": 5.4899,
-      "step": 19226
-    },
-    {
-      "epoch": 10.027118644067796,
-      "grad_norm": 1.4189543724060059,
-      "learning_rate": 8.118492462311558e-05,
-      "loss": 5.6564,
-      "step": 19227
-    },
-    {
-      "epoch": 10.027640156453716,
-      "grad_norm": 1.540278434753418,
-      "learning_rate": 8.118391959798996e-05,
-      "loss": 5.0673,
-      "step": 19228
-    },
-    {
-      "epoch": 10.028161668839635,
-      "grad_norm": 1.5051237344741821,
-      "learning_rate": 8.118291457286432e-05,
-      "loss": 5.47,
-      "step": 19229
-    },
-    {
-      "epoch": 10.028683181225555,
-      "grad_norm": 1.3818399906158447,
-      "learning_rate": 8.11819095477387e-05,
-      "loss": 5.5001,
-      "step": 19230
-    },
-    {
-      "epoch": 10.029204693611474,
-      "grad_norm": 1.543975591659546,
-      "learning_rate": 8.118090452261307e-05,
-      "loss": 5.258,
-      "step": 19231
-    },
-    {
-      "epoch": 10.029726205997392,
-      "grad_norm": 1.5582503080368042,
-      "learning_rate": 8.117989949748744e-05,
-      "loss": 5.428,
-      "step": 19232
-    },
-    {
-      "epoch": 10.030247718383311,
-      "grad_norm": 1.5041838884353638,
-      "learning_rate": 8.117889447236181e-05,
-      "loss": 4.9071,
-      "step": 19233
-    },
-    {
-      "epoch": 10.03076923076923,
-      "grad_norm": 1.5186221599578857,
-      "learning_rate": 8.117788944723619e-05,
-      "loss": 5.512,
-      "step": 19234
-    },
-    {
-      "epoch": 10.03129074315515,
-      "grad_norm": 1.4308427572250366,
-      "learning_rate": 8.117688442211055e-05,
-      "loss": 5.3441,
-      "step": 19235
-    },
-    {
-      "epoch": 10.03181225554107,
-      "grad_norm": 1.4917926788330078,
-      "learning_rate": 8.117587939698493e-05,
-      "loss": 4.9616,
-      "step": 19236
-    },
-    {
-      "epoch": 10.032333767926989,
-      "grad_norm": 1.531262993812561,
-      "learning_rate": 8.11748743718593e-05,
-      "loss": 4.8622,
-      "step": 19237
-    },
-    {
-      "epoch": 10.032855280312907,
-      "grad_norm": 1.365200400352478,
-      "learning_rate": 8.117386934673368e-05,
-      "loss": 5.8046,
-      "step": 19238
-    },
-    {
-      "epoch": 10.033376792698826,
-      "grad_norm": 1.3833461999893188,
-      "learning_rate": 8.117286432160805e-05,
-      "loss": 5.3911,
-      "step": 19239
-    },
-    {
-      "epoch": 10.033898305084746,
-      "grad_norm": 1.380707025527954,
-      "learning_rate": 8.117185929648241e-05,
-      "loss": 5.5584,
-      "step": 19240
-    },
-    {
-      "epoch": 10.034419817470665,
-      "grad_norm": 1.441726565361023,
-      "learning_rate": 8.117085427135679e-05,
-      "loss": 5.5312,
-      "step": 19241
-    },
-    {
-      "epoch": 10.034941329856585,
-      "grad_norm": 1.6169283390045166,
-      "learning_rate": 8.116984924623115e-05,
-      "loss": 5.0093,
-      "step": 19242
-    },
-    {
-      "epoch": 10.035462842242504,
-      "grad_norm": 1.451399803161621,
-      "learning_rate": 8.116884422110553e-05,
-      "loss": 5.3974,
-      "step": 19243
-    },
-    {
-      "epoch": 10.035984354628422,
-      "grad_norm": 1.3837239742279053,
-      "learning_rate": 8.11678391959799e-05,
-      "loss": 5.7725,
-      "step": 19244
-    },
-    {
-      "epoch": 10.036505867014341,
-      "grad_norm": 1.3740222454071045,
-      "learning_rate": 8.116683417085427e-05,
-      "loss": 5.2841,
-      "step": 19245
-    },
-    {
-      "epoch": 10.03702737940026,
-      "grad_norm": 1.4473029375076294,
-      "learning_rate": 8.116582914572864e-05,
-      "loss": 5.4854,
-      "step": 19246
-    },
-    {
-      "epoch": 10.03754889178618,
-      "grad_norm": 1.4862998723983765,
-      "learning_rate": 8.116482412060302e-05,
-      "loss": 5.6429,
-      "step": 19247
-    },
-    {
-      "epoch": 10.0380704041721,
-      "grad_norm": 1.5405683517456055,
-      "learning_rate": 8.11638190954774e-05,
-      "loss": 5.6094,
-      "step": 19248
-    },
-    {
-      "epoch": 10.038591916558019,
-      "grad_norm": 1.3177255392074585,
-      "learning_rate": 8.116281407035177e-05,
-      "loss": 5.5173,
-      "step": 19249
-    },
-    {
-      "epoch": 10.039113428943937,
-      "grad_norm": 1.5374847650527954,
-      "learning_rate": 8.116180904522614e-05,
-      "loss": 5.5432,
-      "step": 19250
-    },
-    {
-      "epoch": 10.039634941329856,
-      "grad_norm": 1.396622657775879,
-      "learning_rate": 8.116080402010051e-05,
-      "loss": 5.4305,
-      "step": 19251
-    },
-    {
-      "epoch": 10.040156453715776,
-      "grad_norm": 1.354366421699524,
-      "learning_rate": 8.115979899497488e-05,
-      "loss": 5.8078,
-      "step": 19252
-    },
-    {
-      "epoch": 10.040677966101695,
-      "grad_norm": 1.49080228805542,
-      "learning_rate": 8.115879396984924e-05,
-      "loss": 5.6135,
-      "step": 19253
-    },
-    {
-      "epoch": 10.041199478487615,
-      "grad_norm": 1.6180803775787354,
-      "learning_rate": 8.115778894472362e-05,
-      "loss": 5.2574,
-      "step": 19254
-    },
-    {
-      "epoch": 10.041720990873534,
-      "grad_norm": 1.431665301322937,
-      "learning_rate": 8.115678391959798e-05,
-      "loss": 5.5684,
-      "step": 19255
-    },
-    {
-      "epoch": 10.042242503259452,
-      "grad_norm": 1.4182755947113037,
-      "learning_rate": 8.115577889447236e-05,
-      "loss": 5.2074,
-      "step": 19256
-    },
-    {
-      "epoch": 10.042764015645371,
-      "grad_norm": 1.4100284576416016,
-      "learning_rate": 8.115477386934674e-05,
-      "loss": 5.4743,
-      "step": 19257
-    },
-    {
-      "epoch": 10.04328552803129,
-      "grad_norm": 1.3494096994400024,
-      "learning_rate": 8.115376884422112e-05,
-      "loss": 5.6684,
-      "step": 19258
-    },
-    {
-      "epoch": 10.04380704041721,
-      "grad_norm": 1.4632843732833862,
-      "learning_rate": 8.115276381909548e-05,
-      "loss": 5.6635,
-      "step": 19259
-    },
-    {
-      "epoch": 10.04432855280313,
-      "grad_norm": 1.3621827363967896,
-      "learning_rate": 8.115175879396986e-05,
-      "loss": 5.721,
-      "step": 19260
-    },
-    {
-      "epoch": 10.044850065189049,
-      "grad_norm": 1.4241751432418823,
-      "learning_rate": 8.115075376884422e-05,
-      "loss": 5.2756,
-      "step": 19261
-    },
-    {
-      "epoch": 10.045371577574967,
-      "grad_norm": 1.68697190284729,
-      "learning_rate": 8.11497487437186e-05,
-      "loss": 5.0448,
-      "step": 19262
-    },
-    {
-      "epoch": 10.045893089960886,
-      "grad_norm": 1.4379736185073853,
-      "learning_rate": 8.114874371859297e-05,
-      "loss": 5.7328,
-      "step": 19263
-    },
-    {
-      "epoch": 10.046414602346806,
-      "grad_norm": 1.338120698928833,
-      "learning_rate": 8.114773869346734e-05,
-      "loss": 5.0565,
-      "step": 19264
-    },
-    {
-      "epoch": 10.046936114732725,
-      "grad_norm": 1.503679633140564,
-      "learning_rate": 8.114673366834171e-05,
-      "loss": 5.6368,
-      "step": 19265
-    },
-    {
-      "epoch": 10.047457627118645,
-      "grad_norm": 1.379663348197937,
-      "learning_rate": 8.114572864321607e-05,
-      "loss": 5.5629,
-      "step": 19266
-    },
-    {
-      "epoch": 10.047979139504564,
-      "grad_norm": 1.4055668115615845,
-      "learning_rate": 8.114472361809045e-05,
-      "loss": 5.401,
-      "step": 19267
-    },
-    {
-      "epoch": 10.048500651890482,
-      "grad_norm": 1.4922829866409302,
-      "learning_rate": 8.114371859296483e-05,
-      "loss": 5.3688,
-      "step": 19268
-    },
-    {
-      "epoch": 10.049022164276401,
-      "grad_norm": 1.3867160081863403,
-      "learning_rate": 8.11427135678392e-05,
-      "loss": 5.8862,
-      "step": 19269
-    },
-    {
-      "epoch": 10.04954367666232,
-      "grad_norm": 2.515937089920044,
-      "learning_rate": 8.114170854271357e-05,
-      "loss": 4.9629,
-      "step": 19270
-    },
-    {
-      "epoch": 10.05006518904824,
-      "grad_norm": 1.5366891622543335,
-      "learning_rate": 8.114070351758795e-05,
-      "loss": 5.4601,
-      "step": 19271
-    },
-    {
-      "epoch": 10.05058670143416,
-      "grad_norm": 1.474718689918518,
-      "learning_rate": 8.113969849246231e-05,
-      "loss": 5.4887,
-      "step": 19272
-    },
-    {
-      "epoch": 10.051108213820077,
-      "grad_norm": 1.3567965030670166,
-      "learning_rate": 8.113869346733669e-05,
-      "loss": 5.8179,
-      "step": 19273
-    },
-    {
-      "epoch": 10.051629726205997,
-      "grad_norm": 1.3927943706512451,
-      "learning_rate": 8.113768844221105e-05,
-      "loss": 5.2114,
-      "step": 19274
-    },
-    {
-      "epoch": 10.052151238591916,
-      "grad_norm": 1.4238077402114868,
-      "learning_rate": 8.113668341708543e-05,
-      "loss": 5.1244,
-      "step": 19275
-    },
-    {
-      "epoch": 10.052672750977836,
-      "grad_norm": 1.471169352531433,
-      "learning_rate": 8.11356783919598e-05,
-      "loss": 5.4444,
-      "step": 19276
-    },
-    {
-      "epoch": 10.053194263363755,
-      "grad_norm": 1.5888460874557495,
-      "learning_rate": 8.113467336683417e-05,
-      "loss": 5.4136,
-      "step": 19277
-    },
-    {
-      "epoch": 10.053715775749675,
-      "grad_norm": 1.392857313156128,
-      "learning_rate": 8.113366834170855e-05,
-      "loss": 5.5142,
-      "step": 19278
-    },
-    {
-      "epoch": 10.054237288135592,
-      "grad_norm": 1.7594866752624512,
-      "learning_rate": 8.113266331658292e-05,
-      "loss": 5.1871,
-      "step": 19279
-    },
-    {
-      "epoch": 10.054758800521512,
-      "grad_norm": 1.4361591339111328,
-      "learning_rate": 8.11316582914573e-05,
-      "loss": 5.482,
-      "step": 19280
-    },
-    {
-      "epoch": 10.055280312907431,
-      "grad_norm": 1.4749622344970703,
-      "learning_rate": 8.113065326633166e-05,
-      "loss": 5.6214,
-      "step": 19281
-    },
-    {
-      "epoch": 10.05580182529335,
-      "grad_norm": 1.5904312133789062,
-      "learning_rate": 8.112964824120604e-05,
-      "loss": 5.2324,
-      "step": 19282
-    },
-    {
-      "epoch": 10.05632333767927,
-      "grad_norm": 1.427329659461975,
-      "learning_rate": 8.11286432160804e-05,
-      "loss": 5.6467,
-      "step": 19283
-    },
-    {
-      "epoch": 10.05684485006519,
-      "grad_norm": 1.5107073783874512,
-      "learning_rate": 8.112763819095478e-05,
-      "loss": 5.6049,
-      "step": 19284
-    },
-    {
-      "epoch": 10.057366362451107,
-      "grad_norm": 1.4855140447616577,
-      "learning_rate": 8.112663316582914e-05,
-      "loss": 5.5145,
-      "step": 19285
-    },
-    {
-      "epoch": 10.057887874837027,
-      "grad_norm": 1.4284321069717407,
-      "learning_rate": 8.112562814070352e-05,
-      "loss": 5.6905,
-      "step": 19286
-    },
-    {
-      "epoch": 10.058409387222946,
-      "grad_norm": 1.4676274061203003,
-      "learning_rate": 8.112462311557788e-05,
-      "loss": 5.6387,
-      "step": 19287
-    },
-    {
-      "epoch": 10.058930899608866,
-      "grad_norm": 1.5244499444961548,
-      "learning_rate": 8.112361809045226e-05,
-      "loss": 4.9867,
-      "step": 19288
-    },
-    {
-      "epoch": 10.059452411994785,
-      "grad_norm": 1.493977665901184,
-      "learning_rate": 8.112261306532664e-05,
-      "loss": 5.4306,
-      "step": 19289
-    },
-    {
-      "epoch": 10.059973924380705,
-      "grad_norm": 1.535022258758545,
-      "learning_rate": 8.112160804020102e-05,
-      "loss": 5.3021,
-      "step": 19290
-    },
-    {
-      "epoch": 10.060495436766622,
-      "grad_norm": 1.5175269842147827,
-      "learning_rate": 8.112060301507538e-05,
-      "loss": 5.4837,
-      "step": 19291
-    },
-    {
-      "epoch": 10.061016949152542,
-      "grad_norm": 1.3836230039596558,
-      "learning_rate": 8.111959798994975e-05,
-      "loss": 5.3813,
-      "step": 19292
-    },
-    {
-      "epoch": 10.061538461538461,
-      "grad_norm": 1.4451349973678589,
-      "learning_rate": 8.111859296482412e-05,
-      "loss": 5.7162,
-      "step": 19293
-    },
-    {
-      "epoch": 10.06205997392438,
-      "grad_norm": 1.4583630561828613,
-      "learning_rate": 8.111758793969849e-05,
-      "loss": 5.83,
-      "step": 19294
-    },
-    {
-      "epoch": 10.0625814863103,
-      "grad_norm": 1.5019235610961914,
-      "learning_rate": 8.111658291457287e-05,
-      "loss": 5.0436,
-      "step": 19295
-    },
-    {
-      "epoch": 10.06310299869622,
-      "grad_norm": 1.531646490097046,
-      "learning_rate": 8.111557788944723e-05,
-      "loss": 4.8945,
-      "step": 19296
-    },
-    {
-      "epoch": 10.063624511082137,
-      "grad_norm": 1.3866416215896606,
-      "learning_rate": 8.111457286432161e-05,
-      "loss": 4.9915,
-      "step": 19297
-    },
-    {
-      "epoch": 10.064146023468057,
-      "grad_norm": 1.4777435064315796,
-      "learning_rate": 8.111356783919599e-05,
-      "loss": 5.4067,
-      "step": 19298
-    },
-    {
-      "epoch": 10.064667535853976,
-      "grad_norm": 1.4132122993469238,
-      "learning_rate": 8.111256281407036e-05,
-      "loss": 5.4919,
-      "step": 19299
-    },
-    {
-      "epoch": 10.065189048239896,
-      "grad_norm": 1.4551937580108643,
-      "learning_rate": 8.111155778894473e-05,
-      "loss": 5.8075,
-      "step": 19300
-    },
-    {
-      "epoch": 10.065710560625815,
-      "grad_norm": 1.3863393068313599,
-      "learning_rate": 8.11105527638191e-05,
-      "loss": 5.5971,
-      "step": 19301
-    },
-    {
-      "epoch": 10.066232073011735,
-      "grad_norm": 1.4928009510040283,
-      "learning_rate": 8.110954773869347e-05,
-      "loss": 5.0422,
-      "step": 19302
-    },
-    {
-      "epoch": 10.066753585397652,
-      "grad_norm": 1.4350641965866089,
-      "learning_rate": 8.110854271356785e-05,
-      "loss": 5.8135,
-      "step": 19303
-    },
-    {
-      "epoch": 10.067275097783572,
-      "grad_norm": 1.7269724607467651,
-      "learning_rate": 8.110753768844221e-05,
-      "loss": 5.4302,
-      "step": 19304
-    },
-    {
-      "epoch": 10.067796610169491,
-      "grad_norm": 1.3695640563964844,
-      "learning_rate": 8.110653266331659e-05,
-      "loss": 5.6059,
-      "step": 19305
-    },
-    {
-      "epoch": 10.06831812255541,
-      "grad_norm": 1.4098560810089111,
-      "learning_rate": 8.110552763819096e-05,
-      "loss": 5.7357,
-      "step": 19306
-    },
-    {
-      "epoch": 10.06883963494133,
-      "grad_norm": 1.4084358215332031,
-      "learning_rate": 8.110452261306532e-05,
-      "loss": 5.7217,
-      "step": 19307
-    },
-    {
-      "epoch": 10.06936114732725,
-      "grad_norm": 1.4089970588684082,
-      "learning_rate": 8.11035175879397e-05,
-      "loss": 5.3665,
-      "step": 19308
-    },
-    {
-      "epoch": 10.069882659713167,
-      "grad_norm": 1.421798586845398,
-      "learning_rate": 8.110251256281408e-05,
-      "loss": 5.3983,
-      "step": 19309
-    },
-    {
-      "epoch": 10.070404172099087,
-      "grad_norm": 1.4100911617279053,
-      "learning_rate": 8.110150753768845e-05,
-      "loss": 5.4405,
-      "step": 19310
-    },
-    {
-      "epoch": 10.070925684485006,
-      "grad_norm": 1.4434186220169067,
-      "learning_rate": 8.110050251256282e-05,
-      "loss": 5.7355,
-      "step": 19311
-    },
-    {
-      "epoch": 10.071447196870926,
-      "grad_norm": 1.3049589395523071,
-      "learning_rate": 8.10994974874372e-05,
-      "loss": 5.4985,
-      "step": 19312
-    },
-    {
-      "epoch": 10.071968709256845,
-      "grad_norm": 1.4290828704833984,
-      "learning_rate": 8.109849246231156e-05,
-      "loss": 5.4818,
-      "step": 19313
-    },
-    {
-      "epoch": 10.072490221642765,
-      "grad_norm": 1.4059878587722778,
-      "learning_rate": 8.109748743718594e-05,
-      "loss": 5.6919,
-      "step": 19314
-    },
-    {
-      "epoch": 10.073011734028682,
-      "grad_norm": 1.525236964225769,
-      "learning_rate": 8.10964824120603e-05,
-      "loss": 5.7167,
-      "step": 19315
-    },
-    {
-      "epoch": 10.073533246414602,
-      "grad_norm": 1.4142699241638184,
-      "learning_rate": 8.109547738693468e-05,
-      "loss": 5.5478,
-      "step": 19316
-    },
-    {
-      "epoch": 10.074054758800521,
-      "grad_norm": 1.2994166612625122,
-      "learning_rate": 8.109447236180904e-05,
-      "loss": 5.8627,
-      "step": 19317
-    },
-    {
-      "epoch": 10.07457627118644,
-      "grad_norm": 1.4129185676574707,
-      "learning_rate": 8.109346733668342e-05,
-      "loss": 5.3876,
-      "step": 19318
-    },
-    {
-      "epoch": 10.07509778357236,
-      "grad_norm": 1.5683854818344116,
-      "learning_rate": 8.10924623115578e-05,
-      "loss": 5.1457,
-      "step": 19319
-    },
-    {
-      "epoch": 10.07561929595828,
-      "grad_norm": 1.378893494606018,
-      "learning_rate": 8.109145728643216e-05,
-      "loss": 5.6973,
-      "step": 19320
-    },
-    {
-      "epoch": 10.076140808344197,
-      "grad_norm": 1.4195994138717651,
-      "learning_rate": 8.109045226130654e-05,
-      "loss": 5.7066,
-      "step": 19321
-    },
-    {
-      "epoch": 10.076662320730117,
-      "grad_norm": 1.4260025024414062,
-      "learning_rate": 8.10894472361809e-05,
-      "loss": 5.6352,
-      "step": 19322
-    },
-    {
-      "epoch": 10.077183833116036,
-      "grad_norm": 1.4632985591888428,
-      "learning_rate": 8.108844221105528e-05,
-      "loss": 5.5168,
-      "step": 19323
-    },
-    {
-      "epoch": 10.077705345501956,
-      "grad_norm": 1.6575618982315063,
-      "learning_rate": 8.108743718592965e-05,
-      "loss": 5.1193,
-      "step": 19324
-    },
-    {
-      "epoch": 10.078226857887875,
-      "grad_norm": 1.5155746936798096,
-      "learning_rate": 8.108643216080403e-05,
-      "loss": 5.014,
-      "step": 19325
-    },
-    {
-      "epoch": 10.078748370273795,
-      "grad_norm": 1.4355226755142212,
-      "learning_rate": 8.108542713567839e-05,
-      "loss": 5.4741,
-      "step": 19326
-    },
-    {
-      "epoch": 10.079269882659712,
-      "grad_norm": 1.430546522140503,
-      "learning_rate": 8.108442211055277e-05,
-      "loss": 5.6278,
-      "step": 19327
-    },
-    {
-      "epoch": 10.079791395045632,
-      "grad_norm": 1.4982028007507324,
-      "learning_rate": 8.108341708542713e-05,
-      "loss": 5.4793,
-      "step": 19328
-    },
-    {
-      "epoch": 10.080312907431551,
-      "grad_norm": 1.305800437927246,
-      "learning_rate": 8.108241206030151e-05,
-      "loss": 5.6507,
-      "step": 19329
-    },
-    {
-      "epoch": 10.08083441981747,
-      "grad_norm": 1.5027934312820435,
-      "learning_rate": 8.108140703517589e-05,
-      "loss": 5.4888,
-      "step": 19330
-    },
-    {
-      "epoch": 10.08135593220339,
-      "grad_norm": 1.4169176816940308,
-      "learning_rate": 8.108040201005027e-05,
-      "loss": 5.8532,
-      "step": 19331
-    },
-    {
-      "epoch": 10.08187744458931,
-      "grad_norm": 1.4002432823181152,
-      "learning_rate": 8.107939698492463e-05,
-      "loss": 5.2923,
-      "step": 19332
-    },
-    {
-      "epoch": 10.082398956975227,
-      "grad_norm": 1.5200055837631226,
-      "learning_rate": 8.1078391959799e-05,
-      "loss": 5.4707,
-      "step": 19333
-    },
-    {
-      "epoch": 10.082920469361147,
-      "grad_norm": 1.373326301574707,
-      "learning_rate": 8.107738693467337e-05,
-      "loss": 5.6683,
-      "step": 19334
-    },
-    {
-      "epoch": 10.083441981747066,
-      "grad_norm": 1.4947618246078491,
-      "learning_rate": 8.107638190954774e-05,
-      "loss": 5.5216,
-      "step": 19335
-    },
-    {
-      "epoch": 10.083963494132986,
-      "grad_norm": 1.5954937934875488,
-      "learning_rate": 8.107537688442211e-05,
-      "loss": 5.4502,
-      "step": 19336
-    },
-    {
-      "epoch": 10.084485006518905,
-      "grad_norm": 1.3913668394088745,
-      "learning_rate": 8.107437185929648e-05,
-      "loss": 5.6349,
-      "step": 19337
-    },
-    {
-      "epoch": 10.085006518904825,
-      "grad_norm": 1.3716650009155273,
-      "learning_rate": 8.107336683417086e-05,
-      "loss": 5.5101,
-      "step": 19338
-    },
-    {
-      "epoch": 10.085528031290742,
-      "grad_norm": 1.420180320739746,
-      "learning_rate": 8.107236180904523e-05,
-      "loss": 5.2732,
-      "step": 19339
-    },
-    {
-      "epoch": 10.086049543676662,
-      "grad_norm": 1.3234541416168213,
-      "learning_rate": 8.107135678391961e-05,
-      "loss": 5.8117,
-      "step": 19340
-    },
-    {
-      "epoch": 10.086571056062581,
-      "grad_norm": 1.9023507833480835,
-      "learning_rate": 8.107035175879398e-05,
-      "loss": 5.3586,
-      "step": 19341
-    },
-    {
-      "epoch": 10.0870925684485,
-      "grad_norm": 1.6827796697616577,
-      "learning_rate": 8.106934673366835e-05,
-      "loss": 5.2181,
-      "step": 19342
-    },
-    {
-      "epoch": 10.08761408083442,
-      "grad_norm": 1.3675897121429443,
-      "learning_rate": 8.106834170854272e-05,
-      "loss": 5.5892,
-      "step": 19343
-    },
-    {
-      "epoch": 10.08813559322034,
-      "grad_norm": 1.315682053565979,
-      "learning_rate": 8.10673366834171e-05,
-      "loss": 4.6557,
-      "step": 19344
-    },
-    {
-      "epoch": 10.088657105606258,
-      "grad_norm": 1.3968167304992676,
-      "learning_rate": 8.106633165829146e-05,
-      "loss": 5.5062,
-      "step": 19345
-    },
-    {
-      "epoch": 10.089178617992177,
-      "grad_norm": 1.4380054473876953,
-      "learning_rate": 8.106532663316582e-05,
-      "loss": 5.3305,
-      "step": 19346
-    },
-    {
-      "epoch": 10.089700130378096,
-      "grad_norm": 1.3944514989852905,
-      "learning_rate": 8.10643216080402e-05,
-      "loss": 5.6389,
-      "step": 19347
-    },
-    {
-      "epoch": 10.090221642764016,
-      "grad_norm": 1.4481744766235352,
-      "learning_rate": 8.106331658291457e-05,
-      "loss": 5.0974,
-      "step": 19348
-    },
-    {
-      "epoch": 10.090743155149935,
-      "grad_norm": 1.4218538999557495,
-      "learning_rate": 8.106231155778894e-05,
-      "loss": 5.3883,
-      "step": 19349
-    },
-    {
-      "epoch": 10.091264667535855,
-      "grad_norm": 1.4861276149749756,
-      "learning_rate": 8.106130653266332e-05,
-      "loss": 5.6051,
-      "step": 19350
-    },
-    {
-      "epoch": 10.091786179921773,
-      "grad_norm": 1.5284123420715332,
-      "learning_rate": 8.10603015075377e-05,
-      "loss": 5.2251,
-      "step": 19351
-    },
-    {
-      "epoch": 10.092307692307692,
-      "grad_norm": 1.428809404373169,
-      "learning_rate": 8.105929648241206e-05,
-      "loss": 5.4539,
-      "step": 19352
-    },
-    {
-      "epoch": 10.092829204693611,
-      "grad_norm": 1.4252064228057861,
-      "learning_rate": 8.105829145728644e-05,
-      "loss": 4.7584,
-      "step": 19353
-    },
-    {
-      "epoch": 10.093350717079531,
-      "grad_norm": 1.4737119674682617,
-      "learning_rate": 8.10572864321608e-05,
-      "loss": 5.9467,
-      "step": 19354
-    },
-    {
-      "epoch": 10.09387222946545,
-      "grad_norm": 1.4642868041992188,
-      "learning_rate": 8.105628140703518e-05,
-      "loss": 4.7951,
-      "step": 19355
-    },
-    {
-      "epoch": 10.09439374185137,
-      "grad_norm": 1.4414517879486084,
-      "learning_rate": 8.105527638190955e-05,
-      "loss": 5.6432,
-      "step": 19356
-    },
-    {
-      "epoch": 10.094915254237288,
-      "grad_norm": 1.6741973161697388,
-      "learning_rate": 8.105427135678393e-05,
-      "loss": 4.8376,
-      "step": 19357
-    },
-    {
-      "epoch": 10.095436766623207,
-      "grad_norm": 1.4339593648910522,
-      "learning_rate": 8.105326633165829e-05,
-      "loss": 5.7529,
-      "step": 19358
-    },
-    {
-      "epoch": 10.095958279009126,
-      "grad_norm": 1.4616647958755493,
-      "learning_rate": 8.105226130653267e-05,
-      "loss": 5.6174,
-      "step": 19359
-    },
-    {
-      "epoch": 10.096479791395046,
-      "grad_norm": 1.4643762111663818,
-      "learning_rate": 8.105125628140705e-05,
-      "loss": 5.4295,
-      "step": 19360
-    },
-    {
-      "epoch": 10.097001303780965,
-      "grad_norm": 1.4800091981887817,
-      "learning_rate": 8.105025125628141e-05,
-      "loss": 5.4135,
-      "step": 19361
-    },
-    {
-      "epoch": 10.097522816166883,
-      "grad_norm": 1.598447322845459,
-      "learning_rate": 8.104924623115579e-05,
-      "loss": 4.9903,
-      "step": 19362
-    },
-    {
-      "epoch": 10.098044328552803,
-      "grad_norm": 1.6117134094238281,
-      "learning_rate": 8.104824120603015e-05,
-      "loss": 5.7504,
-      "step": 19363
-    },
-    {
-      "epoch": 10.098565840938722,
-      "grad_norm": 1.470806360244751,
-      "learning_rate": 8.104723618090453e-05,
-      "loss": 6.0269,
-      "step": 19364
-    },
-    {
-      "epoch": 10.099087353324641,
-      "grad_norm": 1.3830100297927856,
-      "learning_rate": 8.10462311557789e-05,
-      "loss": 5.5934,
-      "step": 19365
-    },
-    {
-      "epoch": 10.099608865710561,
-      "grad_norm": 1.4174232482910156,
-      "learning_rate": 8.104522613065327e-05,
-      "loss": 5.7619,
-      "step": 19366
-    },
-    {
-      "epoch": 10.10013037809648,
-      "grad_norm": 1.622283697128296,
-      "learning_rate": 8.104422110552764e-05,
-      "loss": 5.3888,
-      "step": 19367
-    },
-    {
-      "epoch": 10.100651890482398,
-      "grad_norm": 1.6269261837005615,
-      "learning_rate": 8.104321608040201e-05,
-      "loss": 5.6361,
-      "step": 19368
-    },
-    {
-      "epoch": 10.101173402868318,
-      "grad_norm": 1.5204291343688965,
-      "learning_rate": 8.104221105527638e-05,
-      "loss": 5.4566,
-      "step": 19369
-    },
-    {
-      "epoch": 10.101694915254237,
-      "grad_norm": 1.4245600700378418,
-      "learning_rate": 8.104120603015076e-05,
-      "loss": 5.4761,
-      "step": 19370
-    },
-    {
-      "epoch": 10.102216427640156,
-      "grad_norm": 1.4549909830093384,
-      "learning_rate": 8.104020100502513e-05,
-      "loss": 5.2651,
-      "step": 19371
-    },
-    {
-      "epoch": 10.102737940026076,
-      "grad_norm": 1.4305329322814941,
-      "learning_rate": 8.10391959798995e-05,
-      "loss": 5.5125,
-      "step": 19372
-    },
-    {
-      "epoch": 10.103259452411995,
-      "grad_norm": 1.3953747749328613,
-      "learning_rate": 8.103819095477388e-05,
-      "loss": 5.8032,
-      "step": 19373
-    },
-    {
-      "epoch": 10.103780964797913,
-      "grad_norm": 1.3545929193496704,
-      "learning_rate": 8.103718592964824e-05,
-      "loss": 5.8731,
-      "step": 19374
-    },
-    {
-      "epoch": 10.104302477183833,
-      "grad_norm": 1.460200548171997,
-      "learning_rate": 8.103618090452262e-05,
-      "loss": 4.9004,
-      "step": 19375
-    },
-    {
-      "epoch": 10.104823989569752,
-      "grad_norm": 1.399328589439392,
-      "learning_rate": 8.103517587939698e-05,
-      "loss": 5.4886,
-      "step": 19376
-    },
-    {
-      "epoch": 10.105345501955671,
-      "grad_norm": 1.5347579717636108,
-      "learning_rate": 8.103417085427136e-05,
-      "loss": 5.4536,
-      "step": 19377
-    },
-    {
-      "epoch": 10.105867014341591,
-      "grad_norm": 1.4564110040664673,
-      "learning_rate": 8.103316582914573e-05,
-      "loss": 5.2865,
-      "step": 19378
-    },
-    {
-      "epoch": 10.10638852672751,
-      "grad_norm": 1.391337275505066,
-      "learning_rate": 8.10321608040201e-05,
-      "loss": 5.0897,
-      "step": 19379
-    },
-    {
-      "epoch": 10.106910039113428,
-      "grad_norm": 1.4346710443496704,
-      "learning_rate": 8.103115577889448e-05,
-      "loss": 5.3711,
-      "step": 19380
-    },
-    {
-      "epoch": 10.107431551499348,
-      "grad_norm": 1.4485634565353394,
-      "learning_rate": 8.103015075376886e-05,
-      "loss": 5.6235,
-      "step": 19381
-    },
-    {
-      "epoch": 10.107953063885267,
-      "grad_norm": 1.4469274282455444,
-      "learning_rate": 8.102914572864322e-05,
-      "loss": 5.5899,
-      "step": 19382
-    },
-    {
-      "epoch": 10.108474576271187,
-      "grad_norm": 1.2840619087219238,
-      "learning_rate": 8.10281407035176e-05,
-      "loss": 5.5122,
-      "step": 19383
-    },
-    {
-      "epoch": 10.108996088657106,
-      "grad_norm": 1.2605955600738525,
-      "learning_rate": 8.102713567839197e-05,
-      "loss": 5.6477,
-      "step": 19384
-    },
-    {
-      "epoch": 10.109517601043025,
-      "grad_norm": 1.327576756477356,
-      "learning_rate": 8.102613065326633e-05,
-      "loss": 5.4141,
-      "step": 19385
-    },
-    {
-      "epoch": 10.110039113428943,
-      "grad_norm": 1.3796167373657227,
-      "learning_rate": 8.102512562814071e-05,
-      "loss": 5.9073,
-      "step": 19386
-    },
-    {
-      "epoch": 10.110560625814863,
-      "grad_norm": 1.468694806098938,
-      "learning_rate": 8.102412060301507e-05,
-      "loss": 5.129,
-      "step": 19387
-    },
-    {
-      "epoch": 10.111082138200782,
-      "grad_norm": 1.4745725393295288,
-      "learning_rate": 8.102311557788945e-05,
-      "loss": 5.6439,
-      "step": 19388
-    },
-    {
-      "epoch": 10.111603650586702,
-      "grad_norm": 1.5124186277389526,
-      "learning_rate": 8.102211055276381e-05,
-      "loss": 5.0527,
-      "step": 19389
-    },
-    {
-      "epoch": 10.112125162972621,
-      "grad_norm": 1.5504379272460938,
-      "learning_rate": 8.102110552763819e-05,
-      "loss": 5.3652,
-      "step": 19390
-    },
-    {
-      "epoch": 10.11264667535854,
-      "grad_norm": 1.3777881860733032,
-      "learning_rate": 8.102010050251257e-05,
-      "loss": 5.3954,
-      "step": 19391
-    },
-    {
-      "epoch": 10.113168187744458,
-      "grad_norm": 1.3130061626434326,
-      "learning_rate": 8.101909547738695e-05,
-      "loss": 5.512,
-      "step": 19392
-    },
-    {
-      "epoch": 10.113689700130378,
-      "grad_norm": 1.4320799112319946,
-      "learning_rate": 8.101809045226131e-05,
-      "loss": 5.425,
-      "step": 19393
-    },
-    {
-      "epoch": 10.114211212516297,
-      "grad_norm": 1.5394262075424194,
-      "learning_rate": 8.101708542713569e-05,
-      "loss": 5.1248,
-      "step": 19394
-    },
-    {
-      "epoch": 10.114732724902217,
-      "grad_norm": 1.4459471702575684,
-      "learning_rate": 8.101608040201005e-05,
-      "loss": 5.4195,
-      "step": 19395
-    },
-    {
-      "epoch": 10.115254237288136,
-      "grad_norm": 1.4085073471069336,
-      "learning_rate": 8.101507537688443e-05,
-      "loss": 5.5481,
-      "step": 19396
-    },
-    {
-      "epoch": 10.115775749674055,
-      "grad_norm": 1.4302815198898315,
-      "learning_rate": 8.10140703517588e-05,
-      "loss": 5.7015,
-      "step": 19397
-    },
-    {
-      "epoch": 10.116297262059973,
-      "grad_norm": 1.343001365661621,
-      "learning_rate": 8.101306532663317e-05,
-      "loss": 5.6206,
-      "step": 19398
-    },
-    {
-      "epoch": 10.116818774445893,
-      "grad_norm": 1.4386265277862549,
-      "learning_rate": 8.101206030150754e-05,
-      "loss": 5.0602,
-      "step": 19399
-    },
-    {
-      "epoch": 10.117340286831812,
-      "grad_norm": 1.368187665939331,
-      "learning_rate": 8.10110552763819e-05,
-      "loss": 5.7946,
-      "step": 19400
-    },
-    {
-      "epoch": 10.117861799217732,
-      "grad_norm": 1.4549378156661987,
-      "learning_rate": 8.101005025125628e-05,
-      "loss": 5.6113,
-      "step": 19401
-    },
-    {
-      "epoch": 10.118383311603651,
-      "grad_norm": 1.3259813785552979,
-      "learning_rate": 8.100904522613066e-05,
-      "loss": 5.206,
-      "step": 19402
-    },
-    {
-      "epoch": 10.11890482398957,
-      "grad_norm": 1.48300039768219,
-      "learning_rate": 8.100804020100504e-05,
-      "loss": 5.4663,
-      "step": 19403
-    },
-    {
-      "epoch": 10.119426336375488,
-      "grad_norm": 1.3611689805984497,
-      "learning_rate": 8.10070351758794e-05,
-      "loss": 5.3548,
-      "step": 19404
-    },
-    {
-      "epoch": 10.119947848761408,
-      "grad_norm": 1.3667930364608765,
-      "learning_rate": 8.100603015075378e-05,
-      "loss": 5.7014,
-      "step": 19405
-    },
-    {
-      "epoch": 10.120469361147327,
-      "grad_norm": 1.438826560974121,
-      "learning_rate": 8.100502512562814e-05,
-      "loss": 5.3092,
-      "step": 19406
-    },
-    {
-      "epoch": 10.120990873533247,
-      "grad_norm": 1.365665078163147,
-      "learning_rate": 8.100402010050252e-05,
-      "loss": 5.135,
-      "step": 19407
-    },
-    {
-      "epoch": 10.121512385919166,
-      "grad_norm": 1.95602548122406,
-      "learning_rate": 8.100301507537688e-05,
-      "loss": 4.5689,
-      "step": 19408
-    },
-    {
-      "epoch": 10.122033898305085,
-      "grad_norm": 1.7213242053985596,
-      "learning_rate": 8.100201005025126e-05,
-      "loss": 5.4611,
-      "step": 19409
-    },
-    {
-      "epoch": 10.122555410691003,
-      "grad_norm": 1.490323543548584,
-      "learning_rate": 8.100100502512563e-05,
-      "loss": 5.6172,
-      "step": 19410
-    },
-    {
-      "epoch": 10.123076923076923,
-      "grad_norm": 1.3966107368469238,
-      "learning_rate": 8.1e-05,
-      "loss": 5.7484,
-      "step": 19411
-    },
-    {
-      "epoch": 10.123598435462842,
-      "grad_norm": 1.4256106615066528,
-      "learning_rate": 8.099899497487438e-05,
-      "loss": 5.3941,
-      "step": 19412
-    },
-    {
-      "epoch": 10.124119947848762,
-      "grad_norm": 1.477371096611023,
-      "learning_rate": 8.099798994974875e-05,
-      "loss": 5.326,
-      "step": 19413
-    },
-    {
-      "epoch": 10.124641460234681,
-      "grad_norm": 1.534819483757019,
-      "learning_rate": 8.099698492462312e-05,
-      "loss": 5.324,
-      "step": 19414
-    },
-    {
-      "epoch": 10.1251629726206,
-      "grad_norm": 1.40630042552948,
-      "learning_rate": 8.099597989949749e-05,
-      "loss": 5.3949,
-      "step": 19415
-    },
-    {
-      "epoch": 10.125684485006518,
-      "grad_norm": 1.6721776723861694,
-      "learning_rate": 8.099497487437187e-05,
-      "loss": 5.3887,
-      "step": 19416
-    },
-    {
-      "epoch": 10.126205997392438,
-      "grad_norm": 1.4102164506912231,
-      "learning_rate": 8.099396984924623e-05,
-      "loss": 5.5115,
-      "step": 19417
-    },
-    {
-      "epoch": 10.126727509778357,
-      "grad_norm": 1.416919469833374,
-      "learning_rate": 8.099296482412061e-05,
-      "loss": 5.7177,
-      "step": 19418
-    },
-    {
-      "epoch": 10.127249022164277,
-      "grad_norm": 1.4293354749679565,
-      "learning_rate": 8.099195979899497e-05,
-      "loss": 5.6671,
-      "step": 19419
-    },
-    {
-      "epoch": 10.127770534550196,
-      "grad_norm": 1.5438790321350098,
-      "learning_rate": 8.099095477386935e-05,
-      "loss": 5.5132,
-      "step": 19420
-    },
-    {
-      "epoch": 10.128292046936116,
-      "grad_norm": 1.5145270824432373,
-      "learning_rate": 8.098994974874371e-05,
-      "loss": 5.7938,
-      "step": 19421
-    },
-    {
-      "epoch": 10.128813559322033,
-      "grad_norm": 1.338782787322998,
-      "learning_rate": 8.098894472361809e-05,
-      "loss": 5.1055,
-      "step": 19422
-    },
-    {
-      "epoch": 10.129335071707953,
-      "grad_norm": 1.3905164003372192,
-      "learning_rate": 8.098793969849247e-05,
-      "loss": 5.2523,
-      "step": 19423
-    },
-    {
-      "epoch": 10.129856584093872,
-      "grad_norm": 1.5859357118606567,
-      "learning_rate": 8.098693467336685e-05,
-      "loss": 4.6029,
-      "step": 19424
-    },
-    {
-      "epoch": 10.130378096479792,
-      "grad_norm": 1.3598369359970093,
-      "learning_rate": 8.098592964824121e-05,
-      "loss": 5.5411,
-      "step": 19425
-    },
-    {
-      "epoch": 10.130899608865711,
-      "grad_norm": 1.4330435991287231,
-      "learning_rate": 8.098492462311558e-05,
-      "loss": 5.594,
-      "step": 19426
-    },
-    {
-      "epoch": 10.13142112125163,
-      "grad_norm": 1.339848518371582,
-      "learning_rate": 8.098391959798995e-05,
-      "loss": 5.3849,
-      "step": 19427
-    },
-    {
-      "epoch": 10.131942633637548,
-      "grad_norm": 1.4098496437072754,
-      "learning_rate": 8.098291457286432e-05,
-      "loss": 5.018,
-      "step": 19428
-    },
-    {
-      "epoch": 10.132464146023468,
-      "grad_norm": 1.4616820812225342,
-      "learning_rate": 8.09819095477387e-05,
-      "loss": 5.2786,
-      "step": 19429
-    },
-    {
-      "epoch": 10.132985658409387,
-      "grad_norm": 1.5147500038146973,
-      "learning_rate": 8.098090452261306e-05,
-      "loss": 5.4018,
-      "step": 19430
-    },
-    {
-      "epoch": 10.133507170795307,
-      "grad_norm": 1.4410611391067505,
-      "learning_rate": 8.097989949748744e-05,
-      "loss": 5.2911,
-      "step": 19431
-    },
-    {
-      "epoch": 10.134028683181226,
-      "grad_norm": 1.4800357818603516,
-      "learning_rate": 8.097889447236182e-05,
-      "loss": 5.5125,
-      "step": 19432
-    },
-    {
-      "epoch": 10.134550195567146,
-      "grad_norm": 1.4444271326065063,
-      "learning_rate": 8.09778894472362e-05,
-      "loss": 5.2846,
-      "step": 19433
-    },
-    {
-      "epoch": 10.135071707953063,
-      "grad_norm": 1.4188146591186523,
-      "learning_rate": 8.097688442211056e-05,
-      "loss": 5.4669,
-      "step": 19434
-    },
-    {
-      "epoch": 10.135593220338983,
-      "grad_norm": 1.3974123001098633,
-      "learning_rate": 8.097587939698494e-05,
-      "loss": 5.0747,
-      "step": 19435
-    },
-    {
-      "epoch": 10.136114732724902,
-      "grad_norm": 1.412791132926941,
-      "learning_rate": 8.09748743718593e-05,
-      "loss": 5.1656,
-      "step": 19436
-    },
-    {
-      "epoch": 10.136636245110822,
-      "grad_norm": 1.3574947118759155,
-      "learning_rate": 8.097386934673368e-05,
-      "loss": 5.7706,
-      "step": 19437
-    },
-    {
-      "epoch": 10.137157757496741,
-      "grad_norm": 1.3819210529327393,
-      "learning_rate": 8.097286432160804e-05,
-      "loss": 5.5968,
-      "step": 19438
-    },
-    {
-      "epoch": 10.13767926988266,
-      "grad_norm": 1.4285236597061157,
-      "learning_rate": 8.09718592964824e-05,
-      "loss": 5.5119,
-      "step": 19439
-    },
-    {
-      "epoch": 10.138200782268578,
-      "grad_norm": 1.436313509941101,
-      "learning_rate": 8.097085427135678e-05,
-      "loss": 5.4203,
-      "step": 19440
-    },
-    {
-      "epoch": 10.138722294654498,
-      "grad_norm": 1.4157030582427979,
-      "learning_rate": 8.096984924623115e-05,
-      "loss": 5.5211,
-      "step": 19441
-    },
-    {
-      "epoch": 10.139243807040417,
-      "grad_norm": 1.4061200618743896,
-      "learning_rate": 8.096884422110553e-05,
-      "loss": 5.4446,
-      "step": 19442
-    },
-    {
-      "epoch": 10.139765319426337,
-      "grad_norm": 1.3598960638046265,
-      "learning_rate": 8.09678391959799e-05,
-      "loss": 4.9979,
-      "step": 19443
-    },
-    {
-      "epoch": 10.140286831812256,
-      "grad_norm": 1.563914179801941,
-      "learning_rate": 8.096683417085428e-05,
-      "loss": 5.3924,
-      "step": 19444
-    },
-    {
-      "epoch": 10.140808344198176,
-      "grad_norm": 1.3593469858169556,
-      "learning_rate": 8.096582914572865e-05,
-      "loss": 5.6077,
-      "step": 19445
-    },
-    {
-      "epoch": 10.141329856584093,
-      "grad_norm": 1.568639874458313,
-      "learning_rate": 8.096482412060302e-05,
-      "loss": 5.43,
-      "step": 19446
-    },
-    {
-      "epoch": 10.141851368970013,
-      "grad_norm": 1.5309666395187378,
-      "learning_rate": 8.096381909547739e-05,
-      "loss": 5.4006,
-      "step": 19447
-    },
-    {
-      "epoch": 10.142372881355932,
-      "grad_norm": 1.432808756828308,
-      "learning_rate": 8.096281407035177e-05,
-      "loss": 5.5184,
-      "step": 19448
-    },
-    {
-      "epoch": 10.142894393741852,
-      "grad_norm": 1.391912579536438,
-      "learning_rate": 8.096180904522613e-05,
-      "loss": 5.7298,
-      "step": 19449
-    },
-    {
-      "epoch": 10.143415906127771,
-      "grad_norm": 1.4470077753067017,
-      "learning_rate": 8.096080402010051e-05,
-      "loss": 5.5497,
-      "step": 19450
-    },
-    {
-      "epoch": 10.14393741851369,
-      "grad_norm": 1.379650354385376,
-      "learning_rate": 8.095979899497487e-05,
-      "loss": 5.4782,
-      "step": 19451
-    },
-    {
-      "epoch": 10.144458930899608,
-      "grad_norm": 1.406801462173462,
-      "learning_rate": 8.095879396984925e-05,
-      "loss": 5.4487,
-      "step": 19452
-    },
-    {
-      "epoch": 10.144980443285528,
-      "grad_norm": 1.4356049299240112,
-      "learning_rate": 8.095778894472363e-05,
-      "loss": 5.4606,
-      "step": 19453
-    },
-    {
-      "epoch": 10.145501955671447,
-      "grad_norm": 1.5053104162216187,
-      "learning_rate": 8.095678391959799e-05,
-      "loss": 4.8403,
-      "step": 19454
-    },
-    {
-      "epoch": 10.146023468057367,
-      "grad_norm": 1.5434212684631348,
-      "learning_rate": 8.095577889447237e-05,
-      "loss": 5.6106,
-      "step": 19455
-    },
-    {
-      "epoch": 10.146544980443286,
-      "grad_norm": 1.4464401006698608,
-      "learning_rate": 8.095477386934673e-05,
-      "loss": 5.6273,
-      "step": 19456
-    },
-    {
-      "epoch": 10.147066492829204,
-      "grad_norm": 1.5642483234405518,
-      "learning_rate": 8.095376884422111e-05,
-      "loss": 5.4732,
-      "step": 19457
-    },
-    {
-      "epoch": 10.147588005215123,
-      "grad_norm": 1.44374680519104,
-      "learning_rate": 8.095276381909548e-05,
-      "loss": 5.4537,
-      "step": 19458
-    },
-    {
-      "epoch": 10.148109517601043,
-      "grad_norm": 1.4491840600967407,
-      "learning_rate": 8.095175879396985e-05,
-      "loss": 5.5975,
-      "step": 19459
-    },
-    {
-      "epoch": 10.148631029986962,
-      "grad_norm": 1.3132178783416748,
-      "learning_rate": 8.095075376884422e-05,
-      "loss": 5.9017,
-      "step": 19460
-    },
-    {
-      "epoch": 10.149152542372882,
-      "grad_norm": 1.4303269386291504,
-      "learning_rate": 8.09497487437186e-05,
-      "loss": 5.6484,
-      "step": 19461
-    },
-    {
-      "epoch": 10.149674054758801,
-      "grad_norm": 1.4474009275436401,
-      "learning_rate": 8.094874371859296e-05,
-      "loss": 5.2511,
-      "step": 19462
-    },
-    {
-      "epoch": 10.150195567144719,
-      "grad_norm": 1.4012030363082886,
-      "learning_rate": 8.094773869346734e-05,
-      "loss": 5.486,
-      "step": 19463
-    },
-    {
-      "epoch": 10.150717079530638,
-      "grad_norm": 1.488549828529358,
-      "learning_rate": 8.094673366834172e-05,
-      "loss": 5.5351,
-      "step": 19464
-    },
-    {
-      "epoch": 10.151238591916558,
-      "grad_norm": 1.4877772331237793,
-      "learning_rate": 8.094572864321608e-05,
-      "loss": 5.0137,
-      "step": 19465
-    },
-    {
-      "epoch": 10.151760104302477,
-      "grad_norm": 1.348861575126648,
-      "learning_rate": 8.094472361809046e-05,
-      "loss": 5.4438,
-      "step": 19466
-    },
-    {
-      "epoch": 10.152281616688397,
-      "grad_norm": 1.3827354907989502,
-      "learning_rate": 8.094371859296482e-05,
-      "loss": 5.6863,
-      "step": 19467
-    },
-    {
-      "epoch": 10.152803129074316,
-      "grad_norm": 1.4139838218688965,
-      "learning_rate": 8.09427135678392e-05,
-      "loss": 4.9972,
-      "step": 19468
-    },
-    {
-      "epoch": 10.153324641460234,
-      "grad_norm": 1.3846946954727173,
-      "learning_rate": 8.094170854271357e-05,
-      "loss": 5.5418,
-      "step": 19469
-    },
-    {
-      "epoch": 10.153846153846153,
-      "grad_norm": 1.3526767492294312,
-      "learning_rate": 8.094070351758794e-05,
-      "loss": 5.8296,
-      "step": 19470
-    },
-    {
-      "epoch": 10.154367666232073,
-      "grad_norm": 1.424808382987976,
-      "learning_rate": 8.093969849246231e-05,
-      "loss": 4.9865,
-      "step": 19471
-    },
-    {
-      "epoch": 10.154889178617992,
-      "grad_norm": 1.451377034187317,
-      "learning_rate": 8.093869346733669e-05,
-      "loss": 5.3889,
-      "step": 19472
-    },
-    {
-      "epoch": 10.155410691003912,
-      "grad_norm": 1.4603956937789917,
-      "learning_rate": 8.093768844221106e-05,
-      "loss": 5.0889,
-      "step": 19473
-    },
-    {
-      "epoch": 10.155932203389831,
-      "grad_norm": 1.4246495962142944,
-      "learning_rate": 8.093668341708544e-05,
-      "loss": 5.5314,
-      "step": 19474
-    },
-    {
-      "epoch": 10.156453715775749,
-      "grad_norm": 1.5225346088409424,
-      "learning_rate": 8.09356783919598e-05,
-      "loss": 5.4891,
-      "step": 19475
-    },
-    {
-      "epoch": 10.156975228161668,
-      "grad_norm": 1.4308207035064697,
-      "learning_rate": 8.093467336683418e-05,
-      "loss": 5.6255,
-      "step": 19476
-    },
-    {
-      "epoch": 10.157496740547588,
-      "grad_norm": 1.4333186149597168,
-      "learning_rate": 8.093366834170855e-05,
-      "loss": 5.5403,
-      "step": 19477
-    },
-    {
-      "epoch": 10.158018252933507,
-      "grad_norm": 1.4095463752746582,
-      "learning_rate": 8.093266331658293e-05,
-      "loss": 5.7589,
-      "step": 19478
-    },
-    {
-      "epoch": 10.158539765319427,
-      "grad_norm": 1.499704122543335,
-      "learning_rate": 8.093165829145729e-05,
-      "loss": 5.198,
-      "step": 19479
-    },
-    {
-      "epoch": 10.159061277705346,
-      "grad_norm": 1.3963059186935425,
-      "learning_rate": 8.093065326633165e-05,
-      "loss": 4.9333,
-      "step": 19480
-    },
-    {
-      "epoch": 10.159582790091264,
-      "grad_norm": 1.4165339469909668,
-      "learning_rate": 8.092964824120603e-05,
-      "loss": 5.8574,
-      "step": 19481
-    },
-    {
-      "epoch": 10.160104302477183,
-      "grad_norm": 1.505659818649292,
-      "learning_rate": 8.09286432160804e-05,
-      "loss": 5.2383,
-      "step": 19482
-    },
-    {
-      "epoch": 10.160625814863103,
-      "grad_norm": 1.4203904867172241,
-      "learning_rate": 8.092763819095477e-05,
-      "loss": 5.4004,
-      "step": 19483
-    },
-    {
-      "epoch": 10.161147327249022,
-      "grad_norm": 1.4006047248840332,
-      "learning_rate": 8.092663316582915e-05,
-      "loss": 5.3541,
-      "step": 19484
-    },
-    {
-      "epoch": 10.161668839634942,
-      "grad_norm": 1.5044543743133545,
-      "learning_rate": 8.092562814070353e-05,
-      "loss": 5.5395,
-      "step": 19485
-    },
-    {
-      "epoch": 10.162190352020861,
-      "grad_norm": 1.5390909910202026,
-      "learning_rate": 8.09246231155779e-05,
-      "loss": 5.3361,
-      "step": 19486
-    },
-    {
-      "epoch": 10.162711864406779,
-      "grad_norm": 1.608689546585083,
-      "learning_rate": 8.092361809045227e-05,
-      "loss": 5.282,
-      "step": 19487
-    },
-    {
-      "epoch": 10.163233376792698,
-      "grad_norm": 1.4848641157150269,
-      "learning_rate": 8.092261306532664e-05,
-      "loss": 5.425,
-      "step": 19488
-    },
-    {
-      "epoch": 10.163754889178618,
-      "grad_norm": 1.4469681978225708,
-      "learning_rate": 8.092160804020101e-05,
-      "loss": 5.5673,
-      "step": 19489
-    },
-    {
-      "epoch": 10.164276401564537,
-      "grad_norm": 1.508902668952942,
-      "learning_rate": 8.092060301507538e-05,
-      "loss": 5.3573,
-      "step": 19490
-    },
-    {
-      "epoch": 10.164797913950457,
-      "grad_norm": 1.4977362155914307,
-      "learning_rate": 8.091959798994976e-05,
-      "loss": 5.1354,
-      "step": 19491
-    },
-    {
-      "epoch": 10.165319426336376,
-      "grad_norm": 1.5958330631256104,
-      "learning_rate": 8.091859296482412e-05,
-      "loss": 5.0081,
-      "step": 19492
-    },
-    {
-      "epoch": 10.165840938722294,
-      "grad_norm": 1.3918375968933105,
-      "learning_rate": 8.09175879396985e-05,
-      "loss": 5.3736,
-      "step": 19493
-    },
-    {
-      "epoch": 10.166362451108213,
-      "grad_norm": 1.5944808721542358,
-      "learning_rate": 8.091658291457288e-05,
-      "loss": 5.3807,
-      "step": 19494
-    },
-    {
-      "epoch": 10.166883963494133,
-      "grad_norm": 1.4084001779556274,
-      "learning_rate": 8.091557788944724e-05,
-      "loss": 5.3604,
-      "step": 19495
-    },
-    {
-      "epoch": 10.167405475880052,
-      "grad_norm": 1.475034236907959,
-      "learning_rate": 8.091457286432162e-05,
-      "loss": 5.0697,
-      "step": 19496
-    },
-    {
-      "epoch": 10.167926988265972,
-      "grad_norm": 1.4723076820373535,
-      "learning_rate": 8.091356783919598e-05,
-      "loss": 5.5588,
-      "step": 19497
-    },
-    {
-      "epoch": 10.168448500651891,
-      "grad_norm": 1.3276011943817139,
-      "learning_rate": 8.091256281407036e-05,
-      "loss": 4.8597,
-      "step": 19498
-    },
-    {
-      "epoch": 10.168970013037809,
-      "grad_norm": 1.5811959505081177,
-      "learning_rate": 8.091155778894472e-05,
-      "loss": 5.5177,
-      "step": 19499
-    },
-    {
-      "epoch": 10.169491525423728,
-      "grad_norm": 1.60812509059906,
-      "learning_rate": 8.09105527638191e-05,
-      "loss": 5.284,
-      "step": 19500
-    },
-    {
-      "epoch": 10.170013037809648,
-      "grad_norm": 1.5519214868545532,
-      "learning_rate": 8.090954773869347e-05,
-      "loss": 5.6353,
-      "step": 19501
-    },
-    {
-      "epoch": 10.170534550195567,
-      "grad_norm": 1.5051097869873047,
-      "learning_rate": 8.090854271356784e-05,
-      "loss": 5.5308,
-      "step": 19502
-    },
-    {
-      "epoch": 10.171056062581487,
-      "grad_norm": 1.5351275205612183,
-      "learning_rate": 8.090753768844221e-05,
-      "loss": 5.0022,
-      "step": 19503
-    },
-    {
-      "epoch": 10.171577574967406,
-      "grad_norm": 1.3498361110687256,
-      "learning_rate": 8.090653266331659e-05,
-      "loss": 5.7266,
-      "step": 19504
-    },
-    {
-      "epoch": 10.172099087353324,
-      "grad_norm": 1.5015009641647339,
-      "learning_rate": 8.090552763819096e-05,
-      "loss": 5.3658,
-      "step": 19505
-    },
-    {
-      "epoch": 10.172620599739243,
-      "grad_norm": 1.4775804281234741,
-      "learning_rate": 8.090452261306533e-05,
-      "loss": 5.3991,
-      "step": 19506
-    },
-    {
-      "epoch": 10.173142112125163,
-      "grad_norm": 1.4987527132034302,
-      "learning_rate": 8.09035175879397e-05,
-      "loss": 5.1844,
-      "step": 19507
-    },
-    {
-      "epoch": 10.173663624511082,
-      "grad_norm": 1.495638132095337,
-      "learning_rate": 8.090251256281407e-05,
-      "loss": 5.3837,
-      "step": 19508
-    },
-    {
-      "epoch": 10.174185136897002,
-      "grad_norm": 1.4567981958389282,
-      "learning_rate": 8.090150753768845e-05,
-      "loss": 5.7706,
-      "step": 19509
-    },
-    {
-      "epoch": 10.174706649282921,
-      "grad_norm": 1.5451104640960693,
-      "learning_rate": 8.090050251256281e-05,
-      "loss": 5.198,
-      "step": 19510
-    },
-    {
-      "epoch": 10.175228161668839,
-      "grad_norm": 1.3409355878829956,
-      "learning_rate": 8.089949748743719e-05,
-      "loss": 5.4722,
-      "step": 19511
-    },
-    {
-      "epoch": 10.175749674054758,
-      "grad_norm": 1.4394769668579102,
-      "learning_rate": 8.089849246231155e-05,
-      "loss": 5.4397,
-      "step": 19512
-    },
-    {
-      "epoch": 10.176271186440678,
-      "grad_norm": 1.394413948059082,
-      "learning_rate": 8.089748743718593e-05,
-      "loss": 5.6753,
-      "step": 19513
-    },
-    {
-      "epoch": 10.176792698826597,
-      "grad_norm": 1.8432376384735107,
-      "learning_rate": 8.089648241206031e-05,
-      "loss": 5.5957,
-      "step": 19514
-    },
-    {
-      "epoch": 10.177314211212517,
-      "grad_norm": 1.456478238105774,
-      "learning_rate": 8.089547738693469e-05,
-      "loss": 5.6535,
-      "step": 19515
-    },
-    {
-      "epoch": 10.177835723598436,
-      "grad_norm": 1.4209572076797485,
-      "learning_rate": 8.089447236180905e-05,
-      "loss": 5.8665,
-      "step": 19516
-    },
-    {
-      "epoch": 10.178357235984354,
-      "grad_norm": 1.4971332550048828,
-      "learning_rate": 8.089346733668343e-05,
-      "loss": 4.9922,
-      "step": 19517
-    },
-    {
-      "epoch": 10.178878748370273,
-      "grad_norm": 1.5081357955932617,
-      "learning_rate": 8.08924623115578e-05,
-      "loss": 5.5818,
-      "step": 19518
-    },
-    {
-      "epoch": 10.179400260756193,
-      "grad_norm": 1.4449812173843384,
-      "learning_rate": 8.089145728643216e-05,
-      "loss": 5.5052,
-      "step": 19519
-    },
-    {
-      "epoch": 10.179921773142112,
-      "grad_norm": 1.45856773853302,
-      "learning_rate": 8.089045226130654e-05,
-      "loss": 5.8847,
-      "step": 19520
-    },
-    {
-      "epoch": 10.180443285528032,
-      "grad_norm": 1.4284471273422241,
-      "learning_rate": 8.08894472361809e-05,
-      "loss": 5.4045,
-      "step": 19521
-    },
-    {
-      "epoch": 10.180964797913951,
-      "grad_norm": 1.4807220697402954,
-      "learning_rate": 8.088844221105528e-05,
-      "loss": 5.1943,
-      "step": 19522
-    },
-    {
-      "epoch": 10.181486310299869,
-      "grad_norm": 1.4431860446929932,
-      "learning_rate": 8.088743718592964e-05,
-      "loss": 4.7862,
-      "step": 19523
-    },
-    {
-      "epoch": 10.182007822685788,
-      "grad_norm": 1.328526496887207,
-      "learning_rate": 8.088643216080402e-05,
-      "loss": 5.6054,
-      "step": 19524
-    },
-    {
-      "epoch": 10.182529335071708,
-      "grad_norm": 1.472015380859375,
-      "learning_rate": 8.08854271356784e-05,
-      "loss": 5.3308,
-      "step": 19525
-    },
-    {
-      "epoch": 10.183050847457627,
-      "grad_norm": 1.3769311904907227,
-      "learning_rate": 8.088442211055278e-05,
-      "loss": 5.9147,
-      "step": 19526
-    },
-    {
-      "epoch": 10.183572359843547,
-      "grad_norm": 1.3648805618286133,
-      "learning_rate": 8.088341708542714e-05,
-      "loss": 5.5299,
-      "step": 19527
-    },
-    {
-      "epoch": 10.184093872229466,
-      "grad_norm": 1.3849849700927734,
-      "learning_rate": 8.088241206030152e-05,
-      "loss": 5.5687,
-      "step": 19528
-    },
-    {
-      "epoch": 10.184615384615384,
-      "grad_norm": 1.339496374130249,
-      "learning_rate": 8.088140703517588e-05,
-      "loss": 5.8218,
-      "step": 19529
-    },
-    {
-      "epoch": 10.185136897001303,
-      "grad_norm": 1.3970128297805786,
-      "learning_rate": 8.088040201005026e-05,
-      "loss": 5.3434,
-      "step": 19530
-    },
-    {
-      "epoch": 10.185658409387223,
-      "grad_norm": 1.6431965827941895,
-      "learning_rate": 8.087939698492462e-05,
-      "loss": 4.6975,
-      "step": 19531
-    },
-    {
-      "epoch": 10.186179921773142,
-      "grad_norm": 1.4530564546585083,
-      "learning_rate": 8.087839195979899e-05,
-      "loss": 5.6143,
-      "step": 19532
-    },
-    {
-      "epoch": 10.186701434159062,
-      "grad_norm": 1.364723563194275,
-      "learning_rate": 8.087738693467337e-05,
-      "loss": 5.9728,
-      "step": 19533
-    },
-    {
-      "epoch": 10.187222946544981,
-      "grad_norm": 1.397769570350647,
-      "learning_rate": 8.087638190954774e-05,
-      "loss": 5.2697,
-      "step": 19534
-    },
-    {
-      "epoch": 10.187744458930899,
-      "grad_norm": 1.472971796989441,
-      "learning_rate": 8.087537688442212e-05,
-      "loss": 5.5195,
-      "step": 19535
-    },
-    {
-      "epoch": 10.188265971316818,
-      "grad_norm": 1.5955216884613037,
-      "learning_rate": 8.087437185929649e-05,
-      "loss": 5.4901,
-      "step": 19536
-    },
-    {
-      "epoch": 10.188787483702738,
-      "grad_norm": 1.387121558189392,
-      "learning_rate": 8.087336683417086e-05,
-      "loss": 5.573,
-      "step": 19537
-    },
-    {
-      "epoch": 10.189308996088657,
-      "grad_norm": 1.525375247001648,
-      "learning_rate": 8.087236180904523e-05,
-      "loss": 5.535,
-      "step": 19538
-    },
-    {
-      "epoch": 10.189830508474577,
-      "grad_norm": 1.4134432077407837,
-      "learning_rate": 8.087135678391961e-05,
-      "loss": 5.5791,
-      "step": 19539
-    },
-    {
-      "epoch": 10.190352020860496,
-      "grad_norm": 1.3946542739868164,
-      "learning_rate": 8.087035175879397e-05,
-      "loss": 5.5472,
-      "step": 19540
-    },
-    {
-      "epoch": 10.190873533246414,
-      "grad_norm": 1.3503446578979492,
-      "learning_rate": 8.086934673366835e-05,
-      "loss": 5.6657,
-      "step": 19541
-    },
-    {
-      "epoch": 10.191395045632333,
-      "grad_norm": 1.617401361465454,
-      "learning_rate": 8.086834170854271e-05,
-      "loss": 5.303,
-      "step": 19542
-    },
-    {
-      "epoch": 10.191916558018253,
-      "grad_norm": 1.4470224380493164,
-      "learning_rate": 8.086733668341709e-05,
-      "loss": 5.5871,
-      "step": 19543
-    },
-    {
-      "epoch": 10.192438070404172,
-      "grad_norm": 1.5155029296875,
-      "learning_rate": 8.086633165829146e-05,
-      "loss": 5.3056,
-      "step": 19544
-    },
-    {
-      "epoch": 10.192959582790092,
-      "grad_norm": 1.3856877088546753,
-      "learning_rate": 8.086532663316583e-05,
-      "loss": 5.3653,
-      "step": 19545
-    },
-    {
-      "epoch": 10.193481095176011,
-      "grad_norm": 1.6283833980560303,
-      "learning_rate": 8.086432160804021e-05,
-      "loss": 5.465,
-      "step": 19546
-    },
-    {
-      "epoch": 10.194002607561929,
-      "grad_norm": 1.4039561748504639,
-      "learning_rate": 8.086331658291458e-05,
-      "loss": 5.3723,
-      "step": 19547
-    },
-    {
-      "epoch": 10.194524119947848,
-      "grad_norm": 1.4427791833877563,
-      "learning_rate": 8.086231155778895e-05,
-      "loss": 5.4508,
-      "step": 19548
-    },
-    {
-      "epoch": 10.195045632333768,
-      "grad_norm": 1.3689892292022705,
-      "learning_rate": 8.086130653266332e-05,
-      "loss": 5.1913,
-      "step": 19549
-    },
-    {
-      "epoch": 10.195567144719687,
-      "grad_norm": 1.553533673286438,
-      "learning_rate": 8.08603015075377e-05,
-      "loss": 5.1189,
-      "step": 19550
-    },
-    {
-      "epoch": 10.196088657105607,
-      "grad_norm": 1.5530108213424683,
-      "learning_rate": 8.085929648241206e-05,
-      "loss": 5.4885,
-      "step": 19551
-    },
-    {
-      "epoch": 10.196610169491525,
-      "grad_norm": 1.5213204622268677,
-      "learning_rate": 8.085829145728644e-05,
-      "loss": 5.1214,
-      "step": 19552
-    },
-    {
-      "epoch": 10.197131681877444,
-      "grad_norm": 1.4061203002929688,
-      "learning_rate": 8.08572864321608e-05,
-      "loss": 5.5267,
-      "step": 19553
-    },
-    {
-      "epoch": 10.197653194263363,
-      "grad_norm": 1.5255638360977173,
-      "learning_rate": 8.085628140703518e-05,
-      "loss": 5.5033,
-      "step": 19554
-    },
-    {
-      "epoch": 10.198174706649283,
-      "grad_norm": 1.5357197523117065,
-      "learning_rate": 8.085527638190956e-05,
-      "loss": 5.2114,
-      "step": 19555
-    },
-    {
-      "epoch": 10.198696219035202,
-      "grad_norm": 1.5390784740447998,
-      "learning_rate": 8.085427135678393e-05,
-      "loss": 5.2242,
-      "step": 19556
-    },
-    {
-      "epoch": 10.199217731421122,
-      "grad_norm": 1.3924392461776733,
-      "learning_rate": 8.08532663316583e-05,
-      "loss": 5.6394,
-      "step": 19557
-    },
-    {
-      "epoch": 10.19973924380704,
-      "grad_norm": 1.5759093761444092,
-      "learning_rate": 8.085226130653266e-05,
-      "loss": 4.9743,
-      "step": 19558
-    },
-    {
-      "epoch": 10.200260756192959,
-      "grad_norm": 1.426018476486206,
-      "learning_rate": 8.085125628140704e-05,
-      "loss": 5.377,
-      "step": 19559
-    },
-    {
-      "epoch": 10.200782268578878,
-      "grad_norm": 1.4727998971939087,
-      "learning_rate": 8.08502512562814e-05,
-      "loss": 5.7096,
-      "step": 19560
-    },
-    {
-      "epoch": 10.201303780964798,
-      "grad_norm": 1.4071377515792847,
-      "learning_rate": 8.084924623115578e-05,
-      "loss": 5.3999,
-      "step": 19561
-    },
-    {
-      "epoch": 10.201825293350717,
-      "grad_norm": 1.4726241827011108,
-      "learning_rate": 8.084824120603015e-05,
-      "loss": 5.36,
-      "step": 19562
-    },
-    {
-      "epoch": 10.202346805736637,
-      "grad_norm": 1.7984583377838135,
-      "learning_rate": 8.084723618090453e-05,
-      "loss": 5.3368,
-      "step": 19563
-    },
-    {
-      "epoch": 10.202868318122555,
-      "grad_norm": 1.4509174823760986,
-      "learning_rate": 8.084623115577889e-05,
-      "loss": 5.5806,
-      "step": 19564
-    },
-    {
-      "epoch": 10.203389830508474,
-      "grad_norm": 1.4595590829849243,
-      "learning_rate": 8.084522613065327e-05,
-      "loss": 5.1923,
-      "step": 19565
-    },
-    {
-      "epoch": 10.203911342894393,
-      "grad_norm": 1.4845354557037354,
-      "learning_rate": 8.084422110552765e-05,
-      "loss": 5.4969,
-      "step": 19566
-    },
-    {
-      "epoch": 10.204432855280313,
-      "grad_norm": 1.356342077255249,
-      "learning_rate": 8.084321608040202e-05,
-      "loss": 5.8975,
-      "step": 19567
-    },
-    {
-      "epoch": 10.204954367666232,
-      "grad_norm": 1.5728614330291748,
-      "learning_rate": 8.084221105527639e-05,
-      "loss": 5.5541,
-      "step": 19568
-    },
-    {
-      "epoch": 10.205475880052152,
-      "grad_norm": 1.5314064025878906,
-      "learning_rate": 8.084120603015077e-05,
-      "loss": 5.3693,
-      "step": 19569
-    },
-    {
-      "epoch": 10.20599739243807,
-      "grad_norm": 1.5257186889648438,
-      "learning_rate": 8.084020100502513e-05,
-      "loss": 5.6453,
-      "step": 19570
-    },
-    {
-      "epoch": 10.206518904823989,
-      "grad_norm": 1.4200880527496338,
-      "learning_rate": 8.083919597989951e-05,
-      "loss": 5.2554,
-      "step": 19571
-    },
-    {
-      "epoch": 10.207040417209909,
-      "grad_norm": 1.5496748685836792,
-      "learning_rate": 8.083819095477387e-05,
-      "loss": 5.5453,
-      "step": 19572
-    },
-    {
-      "epoch": 10.207561929595828,
-      "grad_norm": 1.4285515546798706,
-      "learning_rate": 8.083718592964824e-05,
-      "loss": 5.4685,
-      "step": 19573
-    },
-    {
-      "epoch": 10.208083441981747,
-      "grad_norm": 1.3518513441085815,
-      "learning_rate": 8.083618090452261e-05,
-      "loss": 5.6231,
-      "step": 19574
-    },
-    {
-      "epoch": 10.208604954367667,
-      "grad_norm": 1.4289746284484863,
-      "learning_rate": 8.083517587939698e-05,
-      "loss": 5.6491,
-      "step": 19575
-    },
-    {
-      "epoch": 10.209126466753585,
-      "grad_norm": 1.4411437511444092,
-      "learning_rate": 8.083417085427136e-05,
-      "loss": 5.303,
-      "step": 19576
-    },
-    {
-      "epoch": 10.209647979139504,
-      "grad_norm": 1.4915893077850342,
-      "learning_rate": 8.083316582914573e-05,
-      "loss": 5.5105,
-      "step": 19577
-    },
-    {
-      "epoch": 10.210169491525424,
-      "grad_norm": 1.4788860082626343,
-      "learning_rate": 8.083216080402011e-05,
-      "loss": 5.5654,
-      "step": 19578
-    },
-    {
-      "epoch": 10.210691003911343,
-      "grad_norm": 1.4436284303665161,
-      "learning_rate": 8.083115577889448e-05,
-      "loss": 5.8927,
-      "step": 19579
-    },
-    {
-      "epoch": 10.211212516297262,
-      "grad_norm": 1.4151679277420044,
-      "learning_rate": 8.083015075376885e-05,
-      "loss": 5.5406,
-      "step": 19580
-    },
-    {
-      "epoch": 10.211734028683182,
-      "grad_norm": 1.408040165901184,
-      "learning_rate": 8.082914572864322e-05,
-      "loss": 5.7614,
-      "step": 19581
-    },
-    {
-      "epoch": 10.2122555410691,
-      "grad_norm": 1.4444472789764404,
-      "learning_rate": 8.08281407035176e-05,
-      "loss": 5.5461,
-      "step": 19582
-    },
-    {
-      "epoch": 10.212777053455019,
-      "grad_norm": 1.4350438117980957,
-      "learning_rate": 8.082713567839196e-05,
-      "loss": 5.509,
-      "step": 19583
-    },
-    {
-      "epoch": 10.213298565840939,
-      "grad_norm": 1.4145528078079224,
-      "learning_rate": 8.082613065326634e-05,
-      "loss": 5.254,
-      "step": 19584
-    },
-    {
-      "epoch": 10.213820078226858,
-      "grad_norm": 1.481654167175293,
-      "learning_rate": 8.08251256281407e-05,
-      "loss": 5.8675,
-      "step": 19585
-    },
-    {
-      "epoch": 10.214341590612777,
-      "grad_norm": 1.5872541666030884,
-      "learning_rate": 8.082412060301508e-05,
-      "loss": 4.7835,
-      "step": 19586
-    },
-    {
-      "epoch": 10.214863102998697,
-      "grad_norm": 1.3086472749710083,
-      "learning_rate": 8.082311557788946e-05,
-      "loss": 5.5659,
-      "step": 19587
-    },
-    {
-      "epoch": 10.215384615384615,
-      "grad_norm": 1.4820972681045532,
-      "learning_rate": 8.082211055276382e-05,
-      "loss": 5.4076,
-      "step": 19588
-    },
-    {
-      "epoch": 10.215906127770534,
-      "grad_norm": 1.45146906375885,
-      "learning_rate": 8.08211055276382e-05,
-      "loss": 5.822,
-      "step": 19589
-    },
-    {
-      "epoch": 10.216427640156454,
-      "grad_norm": 1.4474022388458252,
-      "learning_rate": 8.082010050251256e-05,
-      "loss": 5.3611,
-      "step": 19590
-    },
-    {
-      "epoch": 10.216949152542373,
-      "grad_norm": 1.4980000257492065,
-      "learning_rate": 8.081909547738694e-05,
-      "loss": 5.281,
-      "step": 19591
-    },
-    {
-      "epoch": 10.217470664928292,
-      "grad_norm": 1.3466154336929321,
-      "learning_rate": 8.08180904522613e-05,
-      "loss": 5.7644,
-      "step": 19592
-    },
-    {
-      "epoch": 10.217992177314212,
-      "grad_norm": 1.531679630279541,
-      "learning_rate": 8.081708542713568e-05,
-      "loss": 5.5486,
-      "step": 19593
-    },
-    {
-      "epoch": 10.21851368970013,
-      "grad_norm": 1.5522480010986328,
-      "learning_rate": 8.081608040201005e-05,
-      "loss": 5.0496,
-      "step": 19594
-    },
-    {
-      "epoch": 10.219035202086049,
-      "grad_norm": 1.503346562385559,
-      "learning_rate": 8.081507537688443e-05,
-      "loss": 5.2293,
-      "step": 19595
-    },
-    {
-      "epoch": 10.219556714471969,
-      "grad_norm": 1.3459651470184326,
-      "learning_rate": 8.081407035175879e-05,
-      "loss": 5.4918,
-      "step": 19596
-    },
-    {
-      "epoch": 10.220078226857888,
-      "grad_norm": 1.4573163986206055,
-      "learning_rate": 8.081306532663317e-05,
-      "loss": 5.2729,
-      "step": 19597
-    },
-    {
-      "epoch": 10.220599739243807,
-      "grad_norm": 1.498792052268982,
-      "learning_rate": 8.081206030150755e-05,
-      "loss": 5.2739,
-      "step": 19598
-    },
-    {
-      "epoch": 10.221121251629727,
-      "grad_norm": 1.4183645248413086,
-      "learning_rate": 8.081105527638191e-05,
-      "loss": 5.1902,
-      "step": 19599
-    },
-    {
-      "epoch": 10.221642764015645,
-      "grad_norm": 1.5358985662460327,
-      "learning_rate": 8.081005025125629e-05,
-      "loss": 5.3685,
-      "step": 19600
-    },
-    {
-      "epoch": 10.222164276401564,
-      "grad_norm": 1.5211600065231323,
-      "learning_rate": 8.080904522613065e-05,
-      "loss": 4.8663,
-      "step": 19601
-    },
-    {
-      "epoch": 10.222685788787484,
-      "grad_norm": 1.4565904140472412,
-      "learning_rate": 8.080804020100503e-05,
-      "loss": 5.9083,
-      "step": 19602
-    },
-    {
-      "epoch": 10.223207301173403,
-      "grad_norm": 1.4751838445663452,
-      "learning_rate": 8.08070351758794e-05,
-      "loss": 5.2475,
-      "step": 19603
-    },
-    {
-      "epoch": 10.223728813559323,
-      "grad_norm": 1.4287371635437012,
-      "learning_rate": 8.080603015075377e-05,
-      "loss": 5.6182,
-      "step": 19604
-    },
-    {
-      "epoch": 10.224250325945242,
-      "grad_norm": 1.3804352283477783,
-      "learning_rate": 8.080502512562814e-05,
-      "loss": 5.5716,
-      "step": 19605
-    },
-    {
-      "epoch": 10.22477183833116,
-      "grad_norm": 1.7012817859649658,
-      "learning_rate": 8.080402010050251e-05,
-      "loss": 5.5798,
-      "step": 19606
-    },
-    {
-      "epoch": 10.22529335071708,
-      "grad_norm": 1.3865857124328613,
-      "learning_rate": 8.080301507537689e-05,
-      "loss": 5.8727,
-      "step": 19607
-    },
-    {
-      "epoch": 10.225814863102999,
-      "grad_norm": 1.4692447185516357,
-      "learning_rate": 8.080201005025127e-05,
-      "loss": 5.4642,
-      "step": 19608
-    },
-    {
-      "epoch": 10.226336375488918,
-      "grad_norm": 1.4985402822494507,
-      "learning_rate": 8.080100502512563e-05,
-      "loss": 5.3091,
-      "step": 19609
-    },
-    {
-      "epoch": 10.226857887874838,
-      "grad_norm": 1.4226957559585571,
-      "learning_rate": 8.080000000000001e-05,
-      "loss": 5.4742,
-      "step": 19610
-    },
-    {
-      "epoch": 10.227379400260757,
-      "grad_norm": 1.4808582067489624,
-      "learning_rate": 8.079899497487438e-05,
-      "loss": 5.7134,
-      "step": 19611
-    },
-    {
-      "epoch": 10.227900912646675,
-      "grad_norm": 1.5480968952178955,
-      "learning_rate": 8.079798994974874e-05,
-      "loss": 5.4516,
-      "step": 19612
-    },
-    {
-      "epoch": 10.228422425032594,
-      "grad_norm": 1.4227374792099,
-      "learning_rate": 8.079698492462312e-05,
-      "loss": 5.5684,
-      "step": 19613
-    },
-    {
-      "epoch": 10.228943937418514,
-      "grad_norm": 1.4119012355804443,
-      "learning_rate": 8.079597989949748e-05,
-      "loss": 5.222,
-      "step": 19614
-    },
-    {
-      "epoch": 10.229465449804433,
-      "grad_norm": 1.2883234024047852,
-      "learning_rate": 8.079497487437186e-05,
-      "loss": 5.7549,
-      "step": 19615
-    },
-    {
-      "epoch": 10.229986962190353,
-      "grad_norm": 1.3884540796279907,
-      "learning_rate": 8.079396984924623e-05,
-      "loss": 5.5718,
-      "step": 19616
-    },
-    {
-      "epoch": 10.230508474576272,
-      "grad_norm": 1.4120254516601562,
-      "learning_rate": 8.07929648241206e-05,
-      "loss": 5.2385,
-      "step": 19617
-    },
-    {
-      "epoch": 10.23102998696219,
-      "grad_norm": 1.3745555877685547,
-      "learning_rate": 8.079195979899498e-05,
-      "loss": 5.4634,
-      "step": 19618
-    },
-    {
-      "epoch": 10.23155149934811,
-      "grad_norm": 1.2764778137207031,
-      "learning_rate": 8.079095477386936e-05,
-      "loss": 5.1321,
-      "step": 19619
-    },
-    {
-      "epoch": 10.232073011734029,
-      "grad_norm": 1.6837291717529297,
-      "learning_rate": 8.078994974874372e-05,
-      "loss": 5.1021,
-      "step": 19620
-    },
-    {
-      "epoch": 10.232594524119948,
-      "grad_norm": 1.3928585052490234,
-      "learning_rate": 8.07889447236181e-05,
-      "loss": 5.28,
-      "step": 19621
-    },
-    {
-      "epoch": 10.233116036505868,
-      "grad_norm": 1.4383965730667114,
-      "learning_rate": 8.078793969849247e-05,
-      "loss": 5.3772,
-      "step": 19622
-    },
-    {
-      "epoch": 10.233637548891787,
-      "grad_norm": 1.4495186805725098,
-      "learning_rate": 8.078693467336684e-05,
-      "loss": 5.4689,
-      "step": 19623
-    },
-    {
-      "epoch": 10.234159061277705,
-      "grad_norm": 1.293430209159851,
-      "learning_rate": 8.078592964824121e-05,
-      "loss": 5.4266,
-      "step": 19624
-    },
-    {
-      "epoch": 10.234680573663624,
-      "grad_norm": 1.3780547380447388,
-      "learning_rate": 8.078492462311557e-05,
-      "loss": 5.5844,
-      "step": 19625
-    },
-    {
-      "epoch": 10.235202086049544,
-      "grad_norm": 1.3688702583312988,
-      "learning_rate": 8.078391959798995e-05,
-      "loss": 5.9246,
-      "step": 19626
-    },
-    {
-      "epoch": 10.235723598435463,
-      "grad_norm": 1.4142534732818604,
-      "learning_rate": 8.078291457286433e-05,
-      "loss": 5.5482,
-      "step": 19627
-    },
-    {
-      "epoch": 10.236245110821383,
-      "grad_norm": 1.346163034439087,
-      "learning_rate": 8.07819095477387e-05,
-      "loss": 5.816,
-      "step": 19628
-    },
-    {
-      "epoch": 10.236766623207302,
-      "grad_norm": 1.3786574602127075,
-      "learning_rate": 8.078090452261307e-05,
-      "loss": 5.3948,
-      "step": 19629
-    },
-    {
-      "epoch": 10.23728813559322,
-      "grad_norm": 1.3628560304641724,
-      "learning_rate": 8.077989949748745e-05,
-      "loss": 5.6967,
-      "step": 19630
-    },
-    {
-      "epoch": 10.23780964797914,
-      "grad_norm": 1.3814842700958252,
-      "learning_rate": 8.077889447236181e-05,
-      "loss": 5.7126,
-      "step": 19631
-    },
-    {
-      "epoch": 10.238331160365059,
-      "grad_norm": 1.5733493566513062,
-      "learning_rate": 8.077788944723619e-05,
-      "loss": 4.9158,
-      "step": 19632
-    },
-    {
-      "epoch": 10.238852672750978,
-      "grad_norm": 1.5785692930221558,
-      "learning_rate": 8.077688442211055e-05,
-      "loss": 5.4759,
-      "step": 19633
-    },
-    {
-      "epoch": 10.239374185136898,
-      "grad_norm": 1.3717598915100098,
-      "learning_rate": 8.077587939698493e-05,
-      "loss": 5.5135,
-      "step": 19634
-    },
-    {
-      "epoch": 10.239895697522817,
-      "grad_norm": 1.4811967611312866,
-      "learning_rate": 8.07748743718593e-05,
-      "loss": 5.6997,
-      "step": 19635
-    },
-    {
-      "epoch": 10.240417209908735,
-      "grad_norm": 1.4611668586730957,
-      "learning_rate": 8.077386934673367e-05,
-      "loss": 5.7389,
-      "step": 19636
-    },
-    {
-      "epoch": 10.240938722294654,
-      "grad_norm": 1.3842359781265259,
-      "learning_rate": 8.077286432160804e-05,
-      "loss": 5.4109,
-      "step": 19637
-    },
-    {
-      "epoch": 10.241460234680574,
-      "grad_norm": 1.5225578546524048,
-      "learning_rate": 8.077185929648242e-05,
-      "loss": 5.3212,
-      "step": 19638
-    },
-    {
-      "epoch": 10.241981747066493,
-      "grad_norm": 1.5818392038345337,
-      "learning_rate": 8.07708542713568e-05,
-      "loss": 5.1879,
-      "step": 19639
-    },
-    {
-      "epoch": 10.242503259452413,
-      "grad_norm": 1.403860092163086,
-      "learning_rate": 8.076984924623116e-05,
-      "loss": 5.3438,
-      "step": 19640
-    },
-    {
-      "epoch": 10.243024771838332,
-      "grad_norm": 1.3938579559326172,
-      "learning_rate": 8.076884422110554e-05,
-      "loss": 5.7699,
-      "step": 19641
-    },
-    {
-      "epoch": 10.24354628422425,
-      "grad_norm": 1.543271780014038,
-      "learning_rate": 8.07678391959799e-05,
-      "loss": 4.9279,
-      "step": 19642
-    },
-    {
-      "epoch": 10.24406779661017,
-      "grad_norm": 1.581017017364502,
-      "learning_rate": 8.076683417085428e-05,
-      "loss": 5.259,
-      "step": 19643
-    },
-    {
-      "epoch": 10.244589308996089,
-      "grad_norm": 1.5491999387741089,
-      "learning_rate": 8.076582914572864e-05,
-      "loss": 5.1597,
-      "step": 19644
-    },
-    {
-      "epoch": 10.245110821382008,
-      "grad_norm": 1.6269747018814087,
-      "learning_rate": 8.076482412060302e-05,
-      "loss": 5.7479,
-      "step": 19645
-    },
-    {
-      "epoch": 10.245632333767928,
-      "grad_norm": 1.579291820526123,
-      "learning_rate": 8.076381909547738e-05,
-      "loss": 5.2597,
-      "step": 19646
-    },
-    {
-      "epoch": 10.246153846153845,
-      "grad_norm": 1.5358558893203735,
-      "learning_rate": 8.076281407035176e-05,
-      "loss": 5.698,
-      "step": 19647
-    },
-    {
-      "epoch": 10.246675358539765,
-      "grad_norm": 1.5543749332427979,
-      "learning_rate": 8.076180904522614e-05,
-      "loss": 5.4434,
-      "step": 19648
-    },
-    {
-      "epoch": 10.247196870925684,
-      "grad_norm": 1.3405026197433472,
-      "learning_rate": 8.076080402010052e-05,
-      "loss": 5.6659,
-      "step": 19649
-    },
-    {
-      "epoch": 10.247718383311604,
-      "grad_norm": 1.593376874923706,
-      "learning_rate": 8.075979899497488e-05,
-      "loss": 5.5829,
-      "step": 19650
-    },
-    {
-      "epoch": 10.248239895697523,
-      "grad_norm": 1.425152063369751,
-      "learning_rate": 8.075879396984925e-05,
-      "loss": 5.5907,
-      "step": 19651
-    },
-    {
-      "epoch": 10.248761408083443,
-      "grad_norm": 1.331728219985962,
-      "learning_rate": 8.075778894472362e-05,
-      "loss": 5.167,
-      "step": 19652
-    },
-    {
-      "epoch": 10.24928292046936,
-      "grad_norm": 1.5044746398925781,
-      "learning_rate": 8.075678391959799e-05,
-      "loss": 5.6241,
-      "step": 19653
-    },
-    {
-      "epoch": 10.24980443285528,
-      "grad_norm": 1.4949020147323608,
-      "learning_rate": 8.075577889447237e-05,
-      "loss": 5.5695,
-      "step": 19654
-    },
-    {
-      "epoch": 10.2503259452412,
-      "grad_norm": 1.456380844116211,
-      "learning_rate": 8.075477386934673e-05,
-      "loss": 5.7567,
-      "step": 19655
-    },
-    {
-      "epoch": 10.250847457627119,
-      "grad_norm": 1.4906812906265259,
-      "learning_rate": 8.075376884422111e-05,
-      "loss": 5.7877,
-      "step": 19656
-    },
-    {
-      "epoch": 10.251368970013038,
-      "grad_norm": 1.444018840789795,
-      "learning_rate": 8.075276381909547e-05,
-      "loss": 5.0632,
-      "step": 19657
-    },
-    {
-      "epoch": 10.251890482398958,
-      "grad_norm": 1.4124714136123657,
-      "learning_rate": 8.075175879396985e-05,
-      "loss": 5.1224,
-      "step": 19658
-    },
-    {
-      "epoch": 10.252411994784875,
-      "grad_norm": 1.5122853517532349,
-      "learning_rate": 8.075075376884423e-05,
-      "loss": 5.0807,
-      "step": 19659
-    },
-    {
-      "epoch": 10.252933507170795,
-      "grad_norm": 1.5395748615264893,
-      "learning_rate": 8.07497487437186e-05,
-      "loss": 5.3744,
-      "step": 19660
-    },
-    {
-      "epoch": 10.253455019556714,
-      "grad_norm": 1.4178379774093628,
-      "learning_rate": 8.074874371859297e-05,
-      "loss": 5.6678,
-      "step": 19661
-    },
-    {
-      "epoch": 10.253976531942634,
-      "grad_norm": 1.5267361402511597,
-      "learning_rate": 8.074773869346735e-05,
-      "loss": 5.2803,
-      "step": 19662
-    },
-    {
-      "epoch": 10.254498044328553,
-      "grad_norm": 1.484771966934204,
-      "learning_rate": 8.074673366834171e-05,
-      "loss": 5.4648,
-      "step": 19663
-    },
-    {
-      "epoch": 10.255019556714473,
-      "grad_norm": 1.386078119277954,
-      "learning_rate": 8.074572864321609e-05,
-      "loss": 5.4271,
-      "step": 19664
-    },
-    {
-      "epoch": 10.25554106910039,
-      "grad_norm": 1.5352014303207397,
-      "learning_rate": 8.074472361809045e-05,
-      "loss": 5.249,
-      "step": 19665
-    },
-    {
-      "epoch": 10.25606258148631,
-      "grad_norm": 1.4068042039871216,
-      "learning_rate": 8.074371859296482e-05,
-      "loss": 4.8913,
-      "step": 19666
-    },
-    {
-      "epoch": 10.25658409387223,
-      "grad_norm": 1.3765565156936646,
-      "learning_rate": 8.07427135678392e-05,
-      "loss": 5.7824,
-      "step": 19667
-    },
-    {
-      "epoch": 10.257105606258149,
-      "grad_norm": 1.435583233833313,
-      "learning_rate": 8.074170854271357e-05,
-      "loss": 5.6348,
-      "step": 19668
-    },
-    {
-      "epoch": 10.257627118644068,
-      "grad_norm": 1.4217019081115723,
-      "learning_rate": 8.074070351758795e-05,
-      "loss": 5.8315,
-      "step": 19669
-    },
-    {
-      "epoch": 10.258148631029988,
-      "grad_norm": 1.4041348695755005,
-      "learning_rate": 8.073969849246232e-05,
-      "loss": 5.1795,
-      "step": 19670
-    },
-    {
-      "epoch": 10.258670143415905,
-      "grad_norm": 1.4029031991958618,
-      "learning_rate": 8.07386934673367e-05,
-      "loss": 5.3346,
-      "step": 19671
-    },
-    {
-      "epoch": 10.259191655801825,
-      "grad_norm": 1.457323431968689,
-      "learning_rate": 8.073768844221106e-05,
-      "loss": 5.5564,
-      "step": 19672
-    },
-    {
-      "epoch": 10.259713168187744,
-      "grad_norm": 1.3805011510849,
-      "learning_rate": 8.073668341708544e-05,
-      "loss": 5.6959,
-      "step": 19673
-    },
-    {
-      "epoch": 10.260234680573664,
-      "grad_norm": 1.586300253868103,
-      "learning_rate": 8.07356783919598e-05,
-      "loss": 5.1742,
-      "step": 19674
-    },
-    {
-      "epoch": 10.260756192959583,
-      "grad_norm": 1.426257610321045,
-      "learning_rate": 8.073467336683418e-05,
-      "loss": 5.4104,
-      "step": 19675
-    },
-    {
-      "epoch": 10.261277705345503,
-      "grad_norm": 1.5237494707107544,
-      "learning_rate": 8.073366834170854e-05,
-      "loss": 4.9027,
-      "step": 19676
-    },
-    {
-      "epoch": 10.26179921773142,
-      "grad_norm": 1.5269652605056763,
-      "learning_rate": 8.073266331658292e-05,
-      "loss": 5.621,
-      "step": 19677
-    },
-    {
-      "epoch": 10.26232073011734,
-      "grad_norm": 1.3534530401229858,
-      "learning_rate": 8.073165829145728e-05,
-      "loss": 5.7794,
-      "step": 19678
-    },
-    {
-      "epoch": 10.26284224250326,
-      "grad_norm": 1.4539451599121094,
-      "learning_rate": 8.073065326633166e-05,
-      "loss": 5.3409,
-      "step": 19679
-    },
-    {
-      "epoch": 10.263363754889179,
-      "grad_norm": 1.4593442678451538,
-      "learning_rate": 8.072964824120604e-05,
-      "loss": 5.7777,
-      "step": 19680
-    },
-    {
-      "epoch": 10.263885267275098,
-      "grad_norm": 1.5943520069122314,
-      "learning_rate": 8.07286432160804e-05,
-      "loss": 5.5489,
-      "step": 19681
-    },
-    {
-      "epoch": 10.264406779661018,
-      "grad_norm": 1.513055443763733,
-      "learning_rate": 8.072763819095478e-05,
-      "loss": 5.1005,
-      "step": 19682
-    },
-    {
-      "epoch": 10.264928292046935,
-      "grad_norm": 1.454781174659729,
-      "learning_rate": 8.072663316582915e-05,
-      "loss": 5.7266,
-      "step": 19683
-    },
-    {
-      "epoch": 10.265449804432855,
-      "grad_norm": 1.4382703304290771,
-      "learning_rate": 8.072562814070352e-05,
-      "loss": 5.5601,
-      "step": 19684
-    },
-    {
-      "epoch": 10.265971316818774,
-      "grad_norm": 1.3640079498291016,
-      "learning_rate": 8.072462311557789e-05,
-      "loss": 5.1543,
-      "step": 19685
-    },
-    {
-      "epoch": 10.266492829204694,
-      "grad_norm": 1.3830897808074951,
-      "learning_rate": 8.072361809045227e-05,
-      "loss": 5.7764,
-      "step": 19686
-    },
-    {
-      "epoch": 10.267014341590613,
-      "grad_norm": 1.5197783708572388,
-      "learning_rate": 8.072261306532663e-05,
-      "loss": 5.1717,
-      "step": 19687
-    },
-    {
-      "epoch": 10.267535853976533,
-      "grad_norm": 1.3675800561904907,
-      "learning_rate": 8.072160804020101e-05,
-      "loss": 5.6146,
-      "step": 19688
-    },
-    {
-      "epoch": 10.26805736636245,
-      "grad_norm": 1.3948725461959839,
-      "learning_rate": 8.072060301507539e-05,
-      "loss": 5.4829,
-      "step": 19689
-    },
-    {
-      "epoch": 10.26857887874837,
-      "grad_norm": 1.5496026277542114,
-      "learning_rate": 8.071959798994976e-05,
-      "loss": 5.4633,
-      "step": 19690
-    },
-    {
-      "epoch": 10.26910039113429,
-      "grad_norm": 1.3389567136764526,
-      "learning_rate": 8.071859296482413e-05,
-      "loss": 5.7157,
-      "step": 19691
-    },
-    {
-      "epoch": 10.269621903520209,
-      "grad_norm": 1.4519349336624146,
-      "learning_rate": 8.071758793969849e-05,
-      "loss": 5.744,
-      "step": 19692
-    },
-    {
-      "epoch": 10.270143415906128,
-      "grad_norm": 1.413030982017517,
-      "learning_rate": 8.071658291457287e-05,
-      "loss": 5.866,
-      "step": 19693
-    },
-    {
-      "epoch": 10.270664928292048,
-      "grad_norm": 1.3603932857513428,
-      "learning_rate": 8.071557788944723e-05,
-      "loss": 5.4254,
-      "step": 19694
-    },
-    {
-      "epoch": 10.271186440677965,
-      "grad_norm": 1.5265876054763794,
-      "learning_rate": 8.071457286432161e-05,
-      "loss": 5.2407,
-      "step": 19695
-    },
-    {
-      "epoch": 10.271707953063885,
-      "grad_norm": 1.535760760307312,
-      "learning_rate": 8.071356783919598e-05,
-      "loss": 5.5262,
-      "step": 19696
-    },
-    {
-      "epoch": 10.272229465449804,
-      "grad_norm": 1.4889934062957764,
-      "learning_rate": 8.071256281407035e-05,
-      "loss": 5.5721,
-      "step": 19697
-    },
-    {
-      "epoch": 10.272750977835724,
-      "grad_norm": 1.3927905559539795,
-      "learning_rate": 8.071155778894472e-05,
-      "loss": 5.624,
-      "step": 19698
-    },
-    {
-      "epoch": 10.273272490221643,
-      "grad_norm": 1.4411745071411133,
-      "learning_rate": 8.07105527638191e-05,
-      "loss": 5.1418,
-      "step": 19699
-    },
-    {
-      "epoch": 10.273794002607563,
-      "grad_norm": 1.4094096422195435,
-      "learning_rate": 8.070954773869347e-05,
-      "loss": 5.6996,
-      "step": 19700
-    },
-    {
-      "epoch": 10.27431551499348,
-      "grad_norm": 1.331878423690796,
-      "learning_rate": 8.070854271356785e-05,
-      "loss": 5.4242,
-      "step": 19701
-    },
-    {
-      "epoch": 10.2748370273794,
-      "grad_norm": 1.5839717388153076,
-      "learning_rate": 8.070753768844222e-05,
-      "loss": 4.9366,
-      "step": 19702
-    },
-    {
-      "epoch": 10.27535853976532,
-      "grad_norm": 1.5349148511886597,
-      "learning_rate": 8.07065326633166e-05,
-      "loss": 5.1897,
-      "step": 19703
-    },
-    {
-      "epoch": 10.275880052151239,
-      "grad_norm": 1.4989714622497559,
-      "learning_rate": 8.070552763819096e-05,
-      "loss": 5.6447,
-      "step": 19704
-    },
-    {
-      "epoch": 10.276401564537158,
-      "grad_norm": 1.45926034450531,
-      "learning_rate": 8.070452261306532e-05,
-      "loss": 5.0081,
-      "step": 19705
-    },
-    {
-      "epoch": 10.276923076923078,
-      "grad_norm": 1.3755757808685303,
-      "learning_rate": 8.07035175879397e-05,
-      "loss": 5.262,
-      "step": 19706
-    },
-    {
-      "epoch": 10.277444589308995,
-      "grad_norm": 1.4238673448562622,
-      "learning_rate": 8.070251256281407e-05,
-      "loss": 5.7853,
-      "step": 19707
-    },
-    {
-      "epoch": 10.277966101694915,
-      "grad_norm": 1.5322812795639038,
-      "learning_rate": 8.070150753768844e-05,
-      "loss": 5.0723,
-      "step": 19708
-    },
-    {
-      "epoch": 10.278487614080834,
-      "grad_norm": 1.4861791133880615,
-      "learning_rate": 8.070050251256282e-05,
-      "loss": 5.3556,
-      "step": 19709
-    },
-    {
-      "epoch": 10.279009126466754,
-      "grad_norm": 1.3552556037902832,
-      "learning_rate": 8.06994974874372e-05,
-      "loss": 5.6164,
-      "step": 19710
-    },
-    {
-      "epoch": 10.279530638852673,
-      "grad_norm": 1.383139967918396,
-      "learning_rate": 8.069849246231156e-05,
-      "loss": 5.4888,
-      "step": 19711
-    },
-    {
-      "epoch": 10.280052151238593,
-      "grad_norm": 1.3704948425292969,
-      "learning_rate": 8.069748743718594e-05,
-      "loss": 5.7088,
-      "step": 19712
-    },
-    {
-      "epoch": 10.28057366362451,
-      "grad_norm": 1.482918381690979,
-      "learning_rate": 8.06964824120603e-05,
-      "loss": 5.237,
-      "step": 19713
-    },
-    {
-      "epoch": 10.28109517601043,
-      "grad_norm": 1.3751720190048218,
-      "learning_rate": 8.069547738693468e-05,
-      "loss": 5.6332,
-      "step": 19714
-    },
-    {
-      "epoch": 10.28161668839635,
-      "grad_norm": 1.4367578029632568,
-      "learning_rate": 8.069447236180905e-05,
-      "loss": 5.5835,
-      "step": 19715
-    },
-    {
-      "epoch": 10.282138200782269,
-      "grad_norm": 1.4691919088363647,
-      "learning_rate": 8.069346733668343e-05,
-      "loss": 5.3707,
-      "step": 19716
-    },
-    {
-      "epoch": 10.282659713168188,
-      "grad_norm": 1.379629373550415,
-      "learning_rate": 8.069246231155779e-05,
-      "loss": 5.4364,
-      "step": 19717
-    },
-    {
-      "epoch": 10.283181225554108,
-      "grad_norm": 1.3491318225860596,
-      "learning_rate": 8.069145728643215e-05,
-      "loss": 5.564,
-      "step": 19718
-    },
-    {
-      "epoch": 10.283702737940025,
-      "grad_norm": 1.4381773471832275,
-      "learning_rate": 8.069045226130653e-05,
-      "loss": 5.5097,
-      "step": 19719
-    },
-    {
-      "epoch": 10.284224250325945,
-      "grad_norm": 1.3275089263916016,
-      "learning_rate": 8.068944723618091e-05,
-      "loss": 5.864,
-      "step": 19720
-    },
-    {
-      "epoch": 10.284745762711864,
-      "grad_norm": 1.4705091714859009,
-      "learning_rate": 8.068844221105529e-05,
-      "loss": 5.7192,
-      "step": 19721
-    },
-    {
-      "epoch": 10.285267275097784,
-      "grad_norm": 1.4034209251403809,
-      "learning_rate": 8.068743718592965e-05,
-      "loss": 5.4819,
-      "step": 19722
-    },
-    {
-      "epoch": 10.285788787483703,
-      "grad_norm": 1.5316036939620972,
-      "learning_rate": 8.068643216080403e-05,
-      "loss": 5.3658,
-      "step": 19723
-    },
-    {
-      "epoch": 10.286310299869623,
-      "grad_norm": 1.272879958152771,
-      "learning_rate": 8.06854271356784e-05,
-      "loss": 5.6798,
-      "step": 19724
-    },
-    {
-      "epoch": 10.28683181225554,
-      "grad_norm": 1.457205891609192,
-      "learning_rate": 8.068442211055277e-05,
-      "loss": 5.424,
-      "step": 19725
-    },
-    {
-      "epoch": 10.28735332464146,
-      "grad_norm": 1.5827877521514893,
-      "learning_rate": 8.068341708542714e-05,
-      "loss": 5.038,
-      "step": 19726
-    },
-    {
-      "epoch": 10.28787483702738,
-      "grad_norm": 1.5501630306243896,
-      "learning_rate": 8.068241206030151e-05,
-      "loss": 5.0506,
-      "step": 19727
-    },
-    {
-      "epoch": 10.288396349413299,
-      "grad_norm": 1.5606615543365479,
-      "learning_rate": 8.068140703517588e-05,
-      "loss": 5.526,
-      "step": 19728
-    },
-    {
-      "epoch": 10.288917861799218,
-      "grad_norm": 1.5909795761108398,
-      "learning_rate": 8.068040201005026e-05,
-      "loss": 5.444,
-      "step": 19729
-    },
-    {
-      "epoch": 10.289439374185136,
-      "grad_norm": 1.5725868940353394,
-      "learning_rate": 8.067939698492462e-05,
-      "loss": 5.1615,
-      "step": 19730
-    },
-    {
-      "epoch": 10.289960886571055,
-      "grad_norm": 1.5448347330093384,
-      "learning_rate": 8.0678391959799e-05,
-      "loss": 5.5534,
-      "step": 19731
-    },
-    {
-      "epoch": 10.290482398956975,
-      "grad_norm": 1.4106744527816772,
-      "learning_rate": 8.067738693467338e-05,
-      "loss": 5.2316,
-      "step": 19732
-    },
-    {
-      "epoch": 10.291003911342894,
-      "grad_norm": 1.3785110712051392,
-      "learning_rate": 8.067638190954774e-05,
-      "loss": 5.3182,
-      "step": 19733
-    },
-    {
-      "epoch": 10.291525423728814,
-      "grad_norm": 1.3887274265289307,
-      "learning_rate": 8.067537688442212e-05,
-      "loss": 5.7057,
-      "step": 19734
-    },
-    {
-      "epoch": 10.292046936114733,
-      "grad_norm": 1.396034598350525,
-      "learning_rate": 8.067437185929648e-05,
-      "loss": 5.6827,
-      "step": 19735
-    },
-    {
-      "epoch": 10.292568448500653,
-      "grad_norm": 1.5648188591003418,
-      "learning_rate": 8.067336683417086e-05,
-      "loss": 5.3744,
-      "step": 19736
-    },
-    {
-      "epoch": 10.29308996088657,
-      "grad_norm": 1.3619029521942139,
-      "learning_rate": 8.067236180904522e-05,
-      "loss": 5.7955,
-      "step": 19737
-    },
-    {
-      "epoch": 10.29361147327249,
-      "grad_norm": 1.4577748775482178,
-      "learning_rate": 8.06713567839196e-05,
-      "loss": 5.452,
-      "step": 19738
-    },
-    {
-      "epoch": 10.29413298565841,
-      "grad_norm": 1.3790501356124878,
-      "learning_rate": 8.067035175879397e-05,
-      "loss": 5.284,
-      "step": 19739
-    },
-    {
-      "epoch": 10.294654498044329,
-      "grad_norm": 1.47101891040802,
-      "learning_rate": 8.066934673366834e-05,
-      "loss": 5.5862,
-      "step": 19740
-    },
-    {
-      "epoch": 10.295176010430248,
-      "grad_norm": 1.4468876123428345,
-      "learning_rate": 8.066834170854272e-05,
-      "loss": 5.2988,
-      "step": 19741
-    },
-    {
-      "epoch": 10.295697522816166,
-      "grad_norm": 1.4375320672988892,
-      "learning_rate": 8.06673366834171e-05,
-      "loss": 5.6101,
-      "step": 19742
-    },
-    {
-      "epoch": 10.296219035202085,
-      "grad_norm": 1.4924684762954712,
-      "learning_rate": 8.066633165829146e-05,
-      "loss": 5.1953,
-      "step": 19743
-    },
-    {
-      "epoch": 10.296740547588005,
-      "grad_norm": 1.500489592552185,
-      "learning_rate": 8.066532663316583e-05,
-      "loss": 5.477,
-      "step": 19744
-    },
-    {
-      "epoch": 10.297262059973924,
-      "grad_norm": 1.3847743272781372,
-      "learning_rate": 8.06643216080402e-05,
-      "loss": 5.4902,
-      "step": 19745
-    },
-    {
-      "epoch": 10.297783572359844,
-      "grad_norm": 1.635524034500122,
-      "learning_rate": 8.066331658291457e-05,
-      "loss": 5.2532,
-      "step": 19746
-    },
-    {
-      "epoch": 10.298305084745763,
-      "grad_norm": 1.353298306465149,
-      "learning_rate": 8.066231155778895e-05,
-      "loss": 5.7615,
-      "step": 19747
-    },
-    {
-      "epoch": 10.298826597131681,
-      "grad_norm": 1.3988207578659058,
-      "learning_rate": 8.066130653266331e-05,
-      "loss": 5.6587,
-      "step": 19748
-    },
-    {
-      "epoch": 10.2993481095176,
-      "grad_norm": 1.42936110496521,
-      "learning_rate": 8.066030150753769e-05,
-      "loss": 5.5101,
-      "step": 19749
-    },
-    {
-      "epoch": 10.29986962190352,
-      "grad_norm": 1.389043927192688,
-      "learning_rate": 8.065929648241205e-05,
-      "loss": 5.2493,
-      "step": 19750
-    },
-    {
-      "epoch": 10.30039113428944,
-      "grad_norm": 1.5476787090301514,
-      "learning_rate": 8.065829145728643e-05,
-      "loss": 4.9675,
-      "step": 19751
-    },
-    {
-      "epoch": 10.300912646675359,
-      "grad_norm": 1.5318892002105713,
-      "learning_rate": 8.065728643216081e-05,
-      "loss": 5.127,
-      "step": 19752
-    },
-    {
-      "epoch": 10.301434159061278,
-      "grad_norm": 1.492361307144165,
-      "learning_rate": 8.065628140703519e-05,
-      "loss": 5.099,
-      "step": 19753
-    },
-    {
-      "epoch": 10.301955671447196,
-      "grad_norm": 1.3157211542129517,
-      "learning_rate": 8.065527638190955e-05,
-      "loss": 5.7055,
-      "step": 19754
-    },
-    {
-      "epoch": 10.302477183833116,
-      "grad_norm": 1.41178560256958,
-      "learning_rate": 8.065427135678393e-05,
-      "loss": 5.7273,
-      "step": 19755
-    },
-    {
-      "epoch": 10.302998696219035,
-      "grad_norm": 1.404778003692627,
-      "learning_rate": 8.06532663316583e-05,
-      "loss": 5.7915,
-      "step": 19756
-    },
-    {
-      "epoch": 10.303520208604954,
-      "grad_norm": 1.3424372673034668,
-      "learning_rate": 8.065226130653267e-05,
-      "loss": 5.8097,
-      "step": 19757
-    },
-    {
-      "epoch": 10.304041720990874,
-      "grad_norm": 1.3517769575119019,
-      "learning_rate": 8.065125628140704e-05,
-      "loss": 5.9101,
-      "step": 19758
-    },
-    {
-      "epoch": 10.304563233376793,
-      "grad_norm": 1.4897257089614868,
-      "learning_rate": 8.06502512562814e-05,
-      "loss": 5.5216,
-      "step": 19759
-    },
-    {
-      "epoch": 10.305084745762711,
-      "grad_norm": 1.4176007509231567,
-      "learning_rate": 8.064924623115578e-05,
-      "loss": 5.2794,
-      "step": 19760
-    },
-    {
-      "epoch": 10.30560625814863,
-      "grad_norm": 1.3954837322235107,
-      "learning_rate": 8.064824120603016e-05,
-      "loss": 5.6428,
-      "step": 19761
-    },
-    {
-      "epoch": 10.30612777053455,
-      "grad_norm": 1.5730704069137573,
-      "learning_rate": 8.064723618090453e-05,
-      "loss": 5.3944,
-      "step": 19762
-    },
-    {
-      "epoch": 10.30664928292047,
-      "grad_norm": 1.5818344354629517,
-      "learning_rate": 8.06462311557789e-05,
-      "loss": 5.1065,
-      "step": 19763
-    },
-    {
-      "epoch": 10.307170795306389,
-      "grad_norm": 1.5362111330032349,
-      "learning_rate": 8.064522613065328e-05,
-      "loss": 5.6834,
-      "step": 19764
-    },
-    {
-      "epoch": 10.307692307692308,
-      "grad_norm": 1.4292616844177246,
-      "learning_rate": 8.064422110552764e-05,
-      "loss": 5.3976,
-      "step": 19765
-    },
-    {
-      "epoch": 10.308213820078226,
-      "grad_norm": 1.4488484859466553,
-      "learning_rate": 8.064321608040202e-05,
-      "loss": 5.5285,
-      "step": 19766
-    },
-    {
-      "epoch": 10.308735332464146,
-      "grad_norm": 1.3237862586975098,
-      "learning_rate": 8.064221105527638e-05,
-      "loss": 5.5267,
-      "step": 19767
-    },
-    {
-      "epoch": 10.309256844850065,
-      "grad_norm": 1.552161693572998,
-      "learning_rate": 8.064120603015076e-05,
-      "loss": 5.1688,
-      "step": 19768
-    },
-    {
-      "epoch": 10.309778357235984,
-      "grad_norm": 1.4585472345352173,
-      "learning_rate": 8.064020100502512e-05,
-      "loss": 5.6491,
-      "step": 19769
-    },
-    {
-      "epoch": 10.310299869621904,
-      "grad_norm": 1.4584161043167114,
-      "learning_rate": 8.06391959798995e-05,
-      "loss": 5.373,
-      "step": 19770
-    },
-    {
-      "epoch": 10.310821382007823,
-      "grad_norm": 1.5045256614685059,
-      "learning_rate": 8.063819095477387e-05,
-      "loss": 5.3348,
-      "step": 19771
-    },
-    {
-      "epoch": 10.311342894393741,
-      "grad_norm": 1.413736343383789,
-      "learning_rate": 8.063718592964824e-05,
-      "loss": 5.6683,
-      "step": 19772
-    },
-    {
-      "epoch": 10.31186440677966,
-      "grad_norm": 1.437482476234436,
-      "learning_rate": 8.063618090452262e-05,
-      "loss": 5.4111,
-      "step": 19773
-    },
-    {
-      "epoch": 10.31238591916558,
-      "grad_norm": 1.4075196981430054,
-      "learning_rate": 8.063517587939699e-05,
-      "loss": 5.7813,
-      "step": 19774
-    },
-    {
-      "epoch": 10.3129074315515,
-      "grad_norm": 1.5158138275146484,
-      "learning_rate": 8.063417085427136e-05,
-      "loss": 4.9191,
-      "step": 19775
-    },
-    {
-      "epoch": 10.313428943937419,
-      "grad_norm": 1.4858710765838623,
-      "learning_rate": 8.063316582914573e-05,
-      "loss": 5.2243,
-      "step": 19776
-    },
-    {
-      "epoch": 10.313950456323338,
-      "grad_norm": 1.4248180389404297,
-      "learning_rate": 8.063216080402011e-05,
-      "loss": 5.3247,
-      "step": 19777
-    },
-    {
-      "epoch": 10.314471968709256,
-      "grad_norm": 1.5085856914520264,
-      "learning_rate": 8.063115577889447e-05,
-      "loss": 5.5014,
-      "step": 19778
-    },
-    {
-      "epoch": 10.314993481095176,
-      "grad_norm": 1.4496021270751953,
-      "learning_rate": 8.063015075376885e-05,
-      "loss": 5.6399,
-      "step": 19779
-    },
-    {
-      "epoch": 10.315514993481095,
-      "grad_norm": 1.4402644634246826,
-      "learning_rate": 8.062914572864321e-05,
-      "loss": 5.4492,
-      "step": 19780
-    },
-    {
-      "epoch": 10.316036505867014,
-      "grad_norm": 1.4391881227493286,
-      "learning_rate": 8.062814070351759e-05,
-      "loss": 5.8239,
-      "step": 19781
-    },
-    {
-      "epoch": 10.316558018252934,
-      "grad_norm": 1.4308634996414185,
-      "learning_rate": 8.062713567839197e-05,
-      "loss": 5.5093,
-      "step": 19782
-    },
-    {
-      "epoch": 10.317079530638853,
-      "grad_norm": 1.3870587348937988,
-      "learning_rate": 8.062613065326635e-05,
-      "loss": 5.8527,
-      "step": 19783
-    },
-    {
-      "epoch": 10.317601043024771,
-      "grad_norm": 1.5040020942687988,
-      "learning_rate": 8.062512562814071e-05,
-      "loss": 5.495,
-      "step": 19784
-    },
-    {
-      "epoch": 10.31812255541069,
-      "grad_norm": 1.3588260412216187,
-      "learning_rate": 8.062412060301508e-05,
-      "loss": 5.4612,
-      "step": 19785
-    },
-    {
-      "epoch": 10.31864406779661,
-      "grad_norm": 1.3641928434371948,
-      "learning_rate": 8.062311557788945e-05,
-      "loss": 5.5744,
-      "step": 19786
-    },
-    {
-      "epoch": 10.31916558018253,
-      "grad_norm": 1.4436606168746948,
-      "learning_rate": 8.062211055276382e-05,
-      "loss": 5.0211,
-      "step": 19787
-    },
-    {
-      "epoch": 10.319687092568449,
-      "grad_norm": 1.4276338815689087,
-      "learning_rate": 8.06211055276382e-05,
-      "loss": 5.6513,
-      "step": 19788
-    },
-    {
-      "epoch": 10.320208604954368,
-      "grad_norm": 1.4876201152801514,
-      "learning_rate": 8.062010050251256e-05,
-      "loss": 5.6305,
-      "step": 19789
-    },
-    {
-      "epoch": 10.320730117340286,
-      "grad_norm": 1.5833113193511963,
-      "learning_rate": 8.061909547738694e-05,
-      "loss": 4.6314,
-      "step": 19790
-    },
-    {
-      "epoch": 10.321251629726206,
-      "grad_norm": 1.4114474058151245,
-      "learning_rate": 8.06180904522613e-05,
-      "loss": 5.5418,
-      "step": 19791
-    },
-    {
-      "epoch": 10.321773142112125,
-      "grad_norm": 1.4560524225234985,
-      "learning_rate": 8.061708542713568e-05,
-      "loss": 5.241,
-      "step": 19792
-    },
-    {
-      "epoch": 10.322294654498045,
-      "grad_norm": 1.444079875946045,
-      "learning_rate": 8.061608040201006e-05,
-      "loss": 5.5885,
-      "step": 19793
-    },
-    {
-      "epoch": 10.322816166883964,
-      "grad_norm": 1.3734147548675537,
-      "learning_rate": 8.061507537688444e-05,
-      "loss": 5.5322,
-      "step": 19794
-    },
-    {
-      "epoch": 10.323337679269883,
-      "grad_norm": 1.4911632537841797,
-      "learning_rate": 8.06140703517588e-05,
-      "loss": 5.3889,
-      "step": 19795
-    },
-    {
-      "epoch": 10.323859191655801,
-      "grad_norm": 1.4427682161331177,
-      "learning_rate": 8.061306532663318e-05,
-      "loss": 5.8106,
-      "step": 19796
-    },
-    {
-      "epoch": 10.32438070404172,
-      "grad_norm": 1.4021507501602173,
-      "learning_rate": 8.061206030150754e-05,
-      "loss": 5.5727,
-      "step": 19797
-    },
-    {
-      "epoch": 10.32490221642764,
-      "grad_norm": 1.3106871843338013,
-      "learning_rate": 8.06110552763819e-05,
-      "loss": 5.74,
-      "step": 19798
-    },
-    {
-      "epoch": 10.32542372881356,
-      "grad_norm": 1.3510991334915161,
-      "learning_rate": 8.061005025125628e-05,
-      "loss": 5.5231,
-      "step": 19799
-    },
-    {
-      "epoch": 10.325945241199479,
-      "grad_norm": 1.386620283126831,
-      "learning_rate": 8.060904522613065e-05,
-      "loss": 5.3057,
-      "step": 19800
-    },
-    {
-      "epoch": 10.326466753585398,
-      "grad_norm": 1.3749103546142578,
-      "learning_rate": 8.060804020100503e-05,
-      "loss": 5.7692,
-      "step": 19801
-    },
-    {
-      "epoch": 10.326988265971316,
-      "grad_norm": 1.348341941833496,
-      "learning_rate": 8.06070351758794e-05,
-      "loss": 5.8455,
-      "step": 19802
-    },
-    {
-      "epoch": 10.327509778357236,
-      "grad_norm": 1.565545678138733,
-      "learning_rate": 8.060603015075378e-05,
-      "loss": 5.2218,
-      "step": 19803
-    },
-    {
-      "epoch": 10.328031290743155,
-      "grad_norm": 1.565821647644043,
-      "learning_rate": 8.060502512562815e-05,
-      "loss": 5.4205,
-      "step": 19804
-    },
-    {
-      "epoch": 10.328552803129075,
-      "grad_norm": 1.4983456134796143,
-      "learning_rate": 8.060402010050252e-05,
-      "loss": 5.3835,
-      "step": 19805
-    },
-    {
-      "epoch": 10.329074315514994,
-      "grad_norm": 1.3550137281417847,
-      "learning_rate": 8.060301507537689e-05,
-      "loss": 5.4738,
-      "step": 19806
-    },
-    {
-      "epoch": 10.329595827900913,
-      "grad_norm": 1.4826171398162842,
-      "learning_rate": 8.060201005025127e-05,
-      "loss": 5.1315,
-      "step": 19807
-    },
-    {
-      "epoch": 10.330117340286831,
-      "grad_norm": 1.479589581489563,
-      "learning_rate": 8.060100502512563e-05,
-      "loss": 5.4416,
-      "step": 19808
-    },
-    {
-      "epoch": 10.33063885267275,
-      "grad_norm": 1.4084969758987427,
-      "learning_rate": 8.060000000000001e-05,
-      "loss": 5.3883,
-      "step": 19809
-    },
-    {
-      "epoch": 10.33116036505867,
-      "grad_norm": 1.3796697854995728,
-      "learning_rate": 8.059899497487437e-05,
-      "loss": 5.5951,
-      "step": 19810
-    },
-    {
-      "epoch": 10.33168187744459,
-      "grad_norm": 1.350969910621643,
-      "learning_rate": 8.059798994974874e-05,
-      "loss": 5.5594,
-      "step": 19811
-    },
-    {
-      "epoch": 10.332203389830509,
-      "grad_norm": 1.4563184976577759,
-      "learning_rate": 8.059698492462311e-05,
-      "loss": 5.5697,
-      "step": 19812
-    },
-    {
-      "epoch": 10.332724902216428,
-      "grad_norm": 1.5332075357437134,
-      "learning_rate": 8.059597989949749e-05,
-      "loss": 5.3495,
-      "step": 19813
-    },
-    {
-      "epoch": 10.333246414602346,
-      "grad_norm": 1.4552077054977417,
-      "learning_rate": 8.059497487437187e-05,
-      "loss": 5.3637,
-      "step": 19814
-    },
-    {
-      "epoch": 10.333767926988266,
-      "grad_norm": 1.5139317512512207,
-      "learning_rate": 8.059396984924623e-05,
-      "loss": 5.182,
-      "step": 19815
-    },
-    {
-      "epoch": 10.334289439374185,
-      "grad_norm": 1.5004678964614868,
-      "learning_rate": 8.059296482412061e-05,
-      "loss": 5.203,
-      "step": 19816
-    },
-    {
-      "epoch": 10.334810951760105,
-      "grad_norm": 1.357333779335022,
-      "learning_rate": 8.059195979899498e-05,
-      "loss": 5.6212,
-      "step": 19817
-    },
-    {
-      "epoch": 10.335332464146024,
-      "grad_norm": 1.552425742149353,
-      "learning_rate": 8.059095477386935e-05,
-      "loss": 5.3441,
-      "step": 19818
-    },
-    {
-      "epoch": 10.335853976531943,
-      "grad_norm": 1.4816426038742065,
-      "learning_rate": 8.058994974874372e-05,
-      "loss": 5.4108,
-      "step": 19819
-    },
-    {
-      "epoch": 10.336375488917861,
-      "grad_norm": 1.409996747970581,
-      "learning_rate": 8.05889447236181e-05,
-      "loss": 5.776,
-      "step": 19820
-    },
-    {
-      "epoch": 10.33689700130378,
-      "grad_norm": 1.372625708580017,
-      "learning_rate": 8.058793969849246e-05,
-      "loss": 5.5522,
-      "step": 19821
-    },
-    {
-      "epoch": 10.3374185136897,
-      "grad_norm": 1.458839774131775,
-      "learning_rate": 8.058693467336684e-05,
-      "loss": 5.2303,
-      "step": 19822
-    },
-    {
-      "epoch": 10.33794002607562,
-      "grad_norm": 1.305536150932312,
-      "learning_rate": 8.058592964824122e-05,
-      "loss": 5.5047,
-      "step": 19823
-    },
-    {
-      "epoch": 10.338461538461539,
-      "grad_norm": 1.4422553777694702,
-      "learning_rate": 8.058492462311558e-05,
-      "loss": 5.3836,
-      "step": 19824
-    },
-    {
-      "epoch": 10.338983050847457,
-      "grad_norm": 1.3653644323349,
-      "learning_rate": 8.058391959798996e-05,
-      "loss": 5.9654,
-      "step": 19825
-    },
-    {
-      "epoch": 10.339504563233376,
-      "grad_norm": 1.4360255002975464,
-      "learning_rate": 8.058291457286432e-05,
-      "loss": 5.1319,
-      "step": 19826
-    },
-    {
-      "epoch": 10.340026075619296,
-      "grad_norm": 1.8096814155578613,
-      "learning_rate": 8.05819095477387e-05,
-      "loss": 4.4558,
-      "step": 19827
-    },
-    {
-      "epoch": 10.340547588005215,
-      "grad_norm": 1.394923448562622,
-      "learning_rate": 8.058090452261306e-05,
-      "loss": 5.6642,
-      "step": 19828
-    },
-    {
-      "epoch": 10.341069100391135,
-      "grad_norm": 1.3629857301712036,
-      "learning_rate": 8.057989949748744e-05,
-      "loss": 5.2881,
-      "step": 19829
-    },
-    {
-      "epoch": 10.341590612777054,
-      "grad_norm": 1.4019643068313599,
-      "learning_rate": 8.05788944723618e-05,
-      "loss": 5.653,
-      "step": 19830
-    },
-    {
-      "epoch": 10.342112125162974,
-      "grad_norm": 1.414834976196289,
-      "learning_rate": 8.057788944723618e-05,
-      "loss": 5.3295,
-      "step": 19831
-    },
-    {
-      "epoch": 10.342633637548891,
-      "grad_norm": 1.4326661825180054,
-      "learning_rate": 8.057688442211055e-05,
-      "loss": 5.1506,
-      "step": 19832
-    },
-    {
-      "epoch": 10.34315514993481,
-      "grad_norm": 1.4020349979400635,
-      "learning_rate": 8.057587939698493e-05,
-      "loss": 5.7316,
-      "step": 19833
-    },
-    {
-      "epoch": 10.34367666232073,
-      "grad_norm": 1.2964818477630615,
-      "learning_rate": 8.05748743718593e-05,
-      "loss": 5.1177,
-      "step": 19834
-    },
-    {
-      "epoch": 10.34419817470665,
-      "grad_norm": 1.3350223302841187,
-      "learning_rate": 8.057386934673368e-05,
-      "loss": 5.6078,
-      "step": 19835
-    },
-    {
-      "epoch": 10.344719687092569,
-      "grad_norm": 1.181787133216858,
-      "learning_rate": 8.057286432160805e-05,
-      "loss": 5.2182,
-      "step": 19836
-    },
-    {
-      "epoch": 10.345241199478487,
-      "grad_norm": 1.3475735187530518,
-      "learning_rate": 8.057185929648241e-05,
-      "loss": 5.9375,
-      "step": 19837
-    },
-    {
-      "epoch": 10.345762711864406,
-      "grad_norm": 1.4052367210388184,
-      "learning_rate": 8.057085427135679e-05,
-      "loss": 5.336,
-      "step": 19838
-    },
-    {
-      "epoch": 10.346284224250326,
-      "grad_norm": 1.3503929376602173,
-      "learning_rate": 8.056984924623115e-05,
-      "loss": 5.8518,
-      "step": 19839
-    },
-    {
-      "epoch": 10.346805736636245,
-      "grad_norm": 1.4892840385437012,
-      "learning_rate": 8.056884422110553e-05,
-      "loss": 5.5372,
-      "step": 19840
-    },
-    {
-      "epoch": 10.347327249022165,
-      "grad_norm": 1.33498215675354,
-      "learning_rate": 8.05678391959799e-05,
-      "loss": 4.739,
-      "step": 19841
-    },
-    {
-      "epoch": 10.347848761408084,
-      "grad_norm": 1.4579837322235107,
-      "learning_rate": 8.056683417085427e-05,
-      "loss": 5.2181,
-      "step": 19842
-    },
-    {
-      "epoch": 10.348370273794002,
-      "grad_norm": 1.3395124673843384,
-      "learning_rate": 8.056582914572865e-05,
-      "loss": 5.6773,
-      "step": 19843
-    },
-    {
-      "epoch": 10.348891786179921,
-      "grad_norm": 1.3693982362747192,
-      "learning_rate": 8.056482412060303e-05,
-      "loss": 5.6594,
-      "step": 19844
-    },
-    {
-      "epoch": 10.34941329856584,
-      "grad_norm": 1.4906103610992432,
-      "learning_rate": 8.056381909547739e-05,
-      "loss": 4.9869,
-      "step": 19845
-    },
-    {
-      "epoch": 10.34993481095176,
-      "grad_norm": 1.3108893632888794,
-      "learning_rate": 8.056281407035177e-05,
-      "loss": 5.6334,
-      "step": 19846
-    },
-    {
-      "epoch": 10.35045632333768,
-      "grad_norm": 1.4764999151229858,
-      "learning_rate": 8.056180904522613e-05,
-      "loss": 5.6453,
-      "step": 19847
-    },
-    {
-      "epoch": 10.350977835723599,
-      "grad_norm": 1.5518531799316406,
-      "learning_rate": 8.056080402010051e-05,
-      "loss": 5.1401,
-      "step": 19848
-    },
-    {
-      "epoch": 10.351499348109517,
-      "grad_norm": 1.3917856216430664,
-      "learning_rate": 8.055979899497488e-05,
-      "loss": 5.0446,
-      "step": 19849
-    },
-    {
-      "epoch": 10.352020860495436,
-      "grad_norm": 1.3659136295318604,
-      "learning_rate": 8.055879396984925e-05,
-      "loss": 5.7786,
-      "step": 19850
-    },
-    {
-      "epoch": 10.352542372881356,
-      "grad_norm": 1.457262396812439,
-      "learning_rate": 8.055778894472362e-05,
-      "loss": 5.6548,
-      "step": 19851
-    },
-    {
-      "epoch": 10.353063885267275,
-      "grad_norm": 1.4075727462768555,
-      "learning_rate": 8.055678391959798e-05,
-      "loss": 5.816,
-      "step": 19852
-    },
-    {
-      "epoch": 10.353585397653195,
-      "grad_norm": 1.5356796979904175,
-      "learning_rate": 8.055577889447236e-05,
-      "loss": 5.1764,
-      "step": 19853
-    },
-    {
-      "epoch": 10.354106910039114,
-      "grad_norm": 1.461155891418457,
-      "learning_rate": 8.055477386934674e-05,
-      "loss": 5.6068,
-      "step": 19854
-    },
-    {
-      "epoch": 10.354628422425032,
-      "grad_norm": 1.5037068128585815,
-      "learning_rate": 8.055376884422112e-05,
-      "loss": 5.4061,
-      "step": 19855
-    },
-    {
-      "epoch": 10.355149934810951,
-      "grad_norm": 1.517964243888855,
-      "learning_rate": 8.055276381909548e-05,
-      "loss": 5.6929,
-      "step": 19856
-    },
-    {
-      "epoch": 10.35567144719687,
-      "grad_norm": 1.4980500936508179,
-      "learning_rate": 8.055175879396986e-05,
-      "loss": 5.2745,
-      "step": 19857
-    },
-    {
-      "epoch": 10.35619295958279,
-      "grad_norm": 1.4591717720031738,
-      "learning_rate": 8.055075376884422e-05,
-      "loss": 5.2639,
-      "step": 19858
-    },
-    {
-      "epoch": 10.35671447196871,
-      "grad_norm": 1.3534847497940063,
-      "learning_rate": 8.05497487437186e-05,
-      "loss": 5.6992,
-      "step": 19859
-    },
-    {
-      "epoch": 10.357235984354629,
-      "grad_norm": 1.4396982192993164,
-      "learning_rate": 8.054874371859297e-05,
-      "loss": 5.3898,
-      "step": 19860
-    },
-    {
-      "epoch": 10.357757496740547,
-      "grad_norm": 1.4613100290298462,
-      "learning_rate": 8.054773869346734e-05,
-      "loss": 5.6587,
-      "step": 19861
-    },
-    {
-      "epoch": 10.358279009126466,
-      "grad_norm": 1.4455432891845703,
-      "learning_rate": 8.054673366834171e-05,
-      "loss": 5.317,
-      "step": 19862
-    },
-    {
-      "epoch": 10.358800521512386,
-      "grad_norm": 1.44996976852417,
-      "learning_rate": 8.054572864321609e-05,
-      "loss": 5.5909,
-      "step": 19863
-    },
-    {
-      "epoch": 10.359322033898305,
-      "grad_norm": 1.4759631156921387,
-      "learning_rate": 8.054472361809046e-05,
-      "loss": 5.4615,
-      "step": 19864
-    },
-    {
-      "epoch": 10.359843546284225,
-      "grad_norm": 1.38669753074646,
-      "learning_rate": 8.054371859296483e-05,
-      "loss": 5.6315,
-      "step": 19865
-    },
-    {
-      "epoch": 10.360365058670144,
-      "grad_norm": 1.324921727180481,
-      "learning_rate": 8.05427135678392e-05,
-      "loss": 5.7339,
-      "step": 19866
-    },
-    {
-      "epoch": 10.360886571056062,
-      "grad_norm": 1.3872483968734741,
-      "learning_rate": 8.054170854271357e-05,
-      "loss": 5.4624,
-      "step": 19867
-    },
-    {
-      "epoch": 10.361408083441981,
-      "grad_norm": 1.4025781154632568,
-      "learning_rate": 8.054070351758795e-05,
-      "loss": 5.2241,
-      "step": 19868
-    },
-    {
-      "epoch": 10.3619295958279,
-      "grad_norm": 1.3966808319091797,
-      "learning_rate": 8.053969849246231e-05,
-      "loss": 5.8647,
-      "step": 19869
-    },
-    {
-      "epoch": 10.36245110821382,
-      "grad_norm": 1.3957785367965698,
-      "learning_rate": 8.053869346733669e-05,
-      "loss": 5.5616,
-      "step": 19870
-    },
-    {
-      "epoch": 10.36297262059974,
-      "grad_norm": 1.4674654006958008,
-      "learning_rate": 8.053768844221105e-05,
-      "loss": 5.5028,
-      "step": 19871
-    },
-    {
-      "epoch": 10.36349413298566,
-      "grad_norm": 1.3123106956481934,
-      "learning_rate": 8.053668341708543e-05,
-      "loss": 4.8583,
-      "step": 19872
-    },
-    {
-      "epoch": 10.364015645371577,
-      "grad_norm": 1.3349554538726807,
-      "learning_rate": 8.05356783919598e-05,
-      "loss": 5.5804,
-      "step": 19873
-    },
-    {
-      "epoch": 10.364537157757496,
-      "grad_norm": 1.4158401489257812,
-      "learning_rate": 8.053467336683417e-05,
-      "loss": 5.6288,
-      "step": 19874
-    },
-    {
-      "epoch": 10.365058670143416,
-      "grad_norm": 1.4898449182510376,
-      "learning_rate": 8.053366834170855e-05,
-      "loss": 5.5112,
-      "step": 19875
-    },
-    {
-      "epoch": 10.365580182529335,
-      "grad_norm": 1.4258404970169067,
-      "learning_rate": 8.053266331658293e-05,
-      "loss": 5.4408,
-      "step": 19876
-    },
-    {
-      "epoch": 10.366101694915255,
-      "grad_norm": 1.4582278728485107,
-      "learning_rate": 8.05316582914573e-05,
-      "loss": 5.6328,
-      "step": 19877
-    },
-    {
-      "epoch": 10.366623207301174,
-      "grad_norm": 1.5079933404922485,
-      "learning_rate": 8.053065326633166e-05,
-      "loss": 5.2247,
-      "step": 19878
-    },
-    {
-      "epoch": 10.367144719687092,
-      "grad_norm": 1.3985559940338135,
-      "learning_rate": 8.052964824120604e-05,
-      "loss": 5.364,
-      "step": 19879
-    },
-    {
-      "epoch": 10.367666232073011,
-      "grad_norm": 1.3819801807403564,
-      "learning_rate": 8.05286432160804e-05,
-      "loss": 5.4357,
-      "step": 19880
-    },
-    {
-      "epoch": 10.36818774445893,
-      "grad_norm": 1.5296952724456787,
-      "learning_rate": 8.052763819095478e-05,
-      "loss": 5.3386,
-      "step": 19881
-    },
-    {
-      "epoch": 10.36870925684485,
-      "grad_norm": 1.4211132526397705,
-      "learning_rate": 8.052663316582914e-05,
-      "loss": 5.4523,
-      "step": 19882
-    },
-    {
-      "epoch": 10.36923076923077,
-      "grad_norm": 1.3990495204925537,
-      "learning_rate": 8.052562814070352e-05,
-      "loss": 5.7629,
-      "step": 19883
-    },
-    {
-      "epoch": 10.36975228161669,
-      "grad_norm": 1.321939468383789,
-      "learning_rate": 8.05246231155779e-05,
-      "loss": 5.4636,
-      "step": 19884
-    },
-    {
-      "epoch": 10.370273794002607,
-      "grad_norm": 1.5131045579910278,
-      "learning_rate": 8.052361809045228e-05,
-      "loss": 5.6481,
-      "step": 19885
-    },
-    {
-      "epoch": 10.370795306388526,
-      "grad_norm": 1.4053682088851929,
-      "learning_rate": 8.052261306532664e-05,
-      "loss": 5.6659,
-      "step": 19886
-    },
-    {
-      "epoch": 10.371316818774446,
-      "grad_norm": 1.3573309183120728,
-      "learning_rate": 8.052160804020102e-05,
-      "loss": 5.775,
-      "step": 19887
-    },
-    {
-      "epoch": 10.371838331160365,
-      "grad_norm": 1.5085012912750244,
-      "learning_rate": 8.052060301507538e-05,
-      "loss": 5.328,
-      "step": 19888
-    },
-    {
-      "epoch": 10.372359843546285,
-      "grad_norm": 1.4049532413482666,
-      "learning_rate": 8.051959798994976e-05,
-      "loss": 5.433,
-      "step": 19889
-    },
-    {
-      "epoch": 10.372881355932204,
-      "grad_norm": 1.5856317281723022,
-      "learning_rate": 8.051859296482412e-05,
-      "loss": 5.2516,
-      "step": 19890
-    },
-    {
-      "epoch": 10.373402868318122,
-      "grad_norm": 1.3369239568710327,
-      "learning_rate": 8.051758793969849e-05,
-      "loss": 5.7125,
-      "step": 19891
-    },
-    {
-      "epoch": 10.373924380704041,
-      "grad_norm": 1.4399809837341309,
-      "learning_rate": 8.051658291457287e-05,
-      "loss": 5.509,
-      "step": 19892
-    },
-    {
-      "epoch": 10.37444589308996,
-      "grad_norm": 1.462722897529602,
-      "learning_rate": 8.051557788944723e-05,
-      "loss": 5.5633,
-      "step": 19893
-    },
-    {
-      "epoch": 10.37496740547588,
-      "grad_norm": 1.394222617149353,
-      "learning_rate": 8.051457286432161e-05,
-      "loss": 5.5082,
-      "step": 19894
-    },
-    {
-      "epoch": 10.3754889178618,
-      "grad_norm": 1.3280123472213745,
-      "learning_rate": 8.051356783919599e-05,
-      "loss": 5.6744,
-      "step": 19895
-    },
-    {
-      "epoch": 10.37601043024772,
-      "grad_norm": 1.4408824443817139,
-      "learning_rate": 8.051256281407036e-05,
-      "loss": 5.6258,
-      "step": 19896
-    },
-    {
-      "epoch": 10.376531942633637,
-      "grad_norm": 1.478295922279358,
-      "learning_rate": 8.051155778894473e-05,
-      "loss": 5.0166,
-      "step": 19897
-    },
-    {
-      "epoch": 10.377053455019556,
-      "grad_norm": 1.2992810010910034,
-      "learning_rate": 8.05105527638191e-05,
-      "loss": 5.9811,
-      "step": 19898
-    },
-    {
-      "epoch": 10.377574967405476,
-      "grad_norm": 1.4433223009109497,
-      "learning_rate": 8.050954773869347e-05,
-      "loss": 5.5634,
-      "step": 19899
-    },
-    {
-      "epoch": 10.378096479791395,
-      "grad_norm": 1.4470492601394653,
-      "learning_rate": 8.050854271356785e-05,
-      "loss": 5.4205,
-      "step": 19900
-    },
-    {
-      "epoch": 10.378617992177315,
-      "grad_norm": 1.5078661441802979,
-      "learning_rate": 8.050753768844221e-05,
-      "loss": 5.1658,
-      "step": 19901
-    },
-    {
-      "epoch": 10.379139504563234,
-      "grad_norm": 1.4682739973068237,
-      "learning_rate": 8.050653266331659e-05,
-      "loss": 5.8103,
-      "step": 19902
-    },
-    {
-      "epoch": 10.379661016949152,
-      "grad_norm": 1.437899112701416,
-      "learning_rate": 8.050552763819095e-05,
-      "loss": 5.5797,
-      "step": 19903
-    },
-    {
-      "epoch": 10.380182529335071,
-      "grad_norm": 1.4933797121047974,
-      "learning_rate": 8.050452261306532e-05,
-      "loss": 5.599,
-      "step": 19904
-    },
-    {
-      "epoch": 10.38070404172099,
-      "grad_norm": 1.548477292060852,
-      "learning_rate": 8.05035175879397e-05,
-      "loss": 5.4559,
-      "step": 19905
-    },
-    {
-      "epoch": 10.38122555410691,
-      "grad_norm": 1.4017431735992432,
-      "learning_rate": 8.050251256281407e-05,
-      "loss": 5.5069,
-      "step": 19906
-    },
-    {
-      "epoch": 10.38174706649283,
-      "grad_norm": 1.4197993278503418,
-      "learning_rate": 8.050150753768845e-05,
-      "loss": 5.5367,
-      "step": 19907
-    },
-    {
-      "epoch": 10.38226857887875,
-      "grad_norm": 1.3796544075012207,
-      "learning_rate": 8.050050251256282e-05,
-      "loss": 4.8312,
-      "step": 19908
-    },
-    {
-      "epoch": 10.382790091264667,
-      "grad_norm": 1.4183613061904907,
-      "learning_rate": 8.04994974874372e-05,
-      "loss": 5.6411,
-      "step": 19909
-    },
-    {
-      "epoch": 10.383311603650586,
-      "grad_norm": 1.5794012546539307,
-      "learning_rate": 8.049849246231156e-05,
-      "loss": 5.301,
-      "step": 19910
-    },
-    {
-      "epoch": 10.383833116036506,
-      "grad_norm": 1.3517440557479858,
-      "learning_rate": 8.049748743718594e-05,
-      "loss": 5.8297,
-      "step": 19911
-    },
-    {
-      "epoch": 10.384354628422425,
-      "grad_norm": 1.2583736181259155,
-      "learning_rate": 8.04964824120603e-05,
-      "loss": 6.0487,
-      "step": 19912
-    },
-    {
-      "epoch": 10.384876140808345,
-      "grad_norm": 1.450395107269287,
-      "learning_rate": 8.049547738693468e-05,
-      "loss": 5.3735,
-      "step": 19913
-    },
-    {
-      "epoch": 10.385397653194264,
-      "grad_norm": 1.29973304271698,
-      "learning_rate": 8.049447236180904e-05,
-      "loss": 5.641,
-      "step": 19914
-    },
-    {
-      "epoch": 10.385919165580182,
-      "grad_norm": 1.5069270133972168,
-      "learning_rate": 8.049346733668342e-05,
-      "loss": 5.2526,
-      "step": 19915
-    },
-    {
-      "epoch": 10.386440677966101,
-      "grad_norm": 1.3516182899475098,
-      "learning_rate": 8.04924623115578e-05,
-      "loss": 5.8404,
-      "step": 19916
-    },
-    {
-      "epoch": 10.38696219035202,
-      "grad_norm": 1.3427523374557495,
-      "learning_rate": 8.049145728643216e-05,
-      "loss": 5.8426,
-      "step": 19917
-    },
-    {
-      "epoch": 10.38748370273794,
-      "grad_norm": 1.4065072536468506,
-      "learning_rate": 8.049045226130654e-05,
-      "loss": 5.5529,
-      "step": 19918
-    },
-    {
-      "epoch": 10.38800521512386,
-      "grad_norm": 1.5368915796279907,
-      "learning_rate": 8.04894472361809e-05,
-      "loss": 5.0762,
-      "step": 19919
-    },
-    {
-      "epoch": 10.388526727509777,
-      "grad_norm": 1.504387378692627,
-      "learning_rate": 8.048844221105528e-05,
-      "loss": 5.273,
-      "step": 19920
-    },
-    {
-      "epoch": 10.389048239895697,
-      "grad_norm": 1.5388939380645752,
-      "learning_rate": 8.048743718592965e-05,
-      "loss": 5.2839,
-      "step": 19921
-    },
-    {
-      "epoch": 10.389569752281616,
-      "grad_norm": 1.4012963771820068,
-      "learning_rate": 8.048643216080402e-05,
-      "loss": 5.4168,
-      "step": 19922
-    },
-    {
-      "epoch": 10.390091264667536,
-      "grad_norm": 1.428266167640686,
-      "learning_rate": 8.048542713567839e-05,
-      "loss": 5.4528,
-      "step": 19923
-    },
-    {
-      "epoch": 10.390612777053455,
-      "grad_norm": 1.3810570240020752,
-      "learning_rate": 8.048442211055277e-05,
-      "loss": 5.6033,
-      "step": 19924
-    },
-    {
-      "epoch": 10.391134289439375,
-      "grad_norm": 1.554585337638855,
-      "learning_rate": 8.048341708542713e-05,
-      "loss": 5.5115,
-      "step": 19925
-    },
-    {
-      "epoch": 10.391655801825294,
-      "grad_norm": 1.4755282402038574,
-      "learning_rate": 8.048241206030151e-05,
-      "loss": 5.136,
-      "step": 19926
-    },
-    {
-      "epoch": 10.392177314211212,
-      "grad_norm": 1.4590977430343628,
-      "learning_rate": 8.048140703517589e-05,
-      "loss": 5.5444,
-      "step": 19927
-    },
-    {
-      "epoch": 10.392698826597131,
-      "grad_norm": 1.4041529893875122,
-      "learning_rate": 8.048040201005026e-05,
-      "loss": 5.6541,
-      "step": 19928
-    },
-    {
-      "epoch": 10.39322033898305,
-      "grad_norm": 1.38002347946167,
-      "learning_rate": 8.047939698492463e-05,
-      "loss": 5.7059,
-      "step": 19929
-    },
-    {
-      "epoch": 10.39374185136897,
-      "grad_norm": 1.372519612312317,
-      "learning_rate": 8.0478391959799e-05,
-      "loss": 5.8126,
-      "step": 19930
-    },
-    {
-      "epoch": 10.39426336375489,
-      "grad_norm": 1.43014657497406,
-      "learning_rate": 8.047738693467337e-05,
-      "loss": 5.6474,
-      "step": 19931
-    },
-    {
-      "epoch": 10.394784876140807,
-      "grad_norm": 1.477543830871582,
-      "learning_rate": 8.047638190954774e-05,
-      "loss": 5.3197,
-      "step": 19932
-    },
-    {
-      "epoch": 10.395306388526727,
-      "grad_norm": 1.488138198852539,
-      "learning_rate": 8.047537688442211e-05,
-      "loss": 5.0518,
-      "step": 19933
-    },
-    {
-      "epoch": 10.395827900912646,
-      "grad_norm": 1.35440993309021,
-      "learning_rate": 8.047437185929648e-05,
-      "loss": 5.6389,
-      "step": 19934
-    },
-    {
-      "epoch": 10.396349413298566,
-      "grad_norm": 1.3866780996322632,
-      "learning_rate": 8.047336683417085e-05,
-      "loss": 5.3888,
-      "step": 19935
-    },
-    {
-      "epoch": 10.396870925684485,
-      "grad_norm": 1.4034926891326904,
-      "learning_rate": 8.047236180904523e-05,
-      "loss": 5.6027,
-      "step": 19936
-    },
-    {
-      "epoch": 10.397392438070405,
-      "grad_norm": 1.374549388885498,
-      "learning_rate": 8.047135678391961e-05,
-      "loss": 5.6272,
-      "step": 19937
-    },
-    {
-      "epoch": 10.397913950456322,
-      "grad_norm": 1.4031211137771606,
-      "learning_rate": 8.047035175879397e-05,
-      "loss": 5.3778,
-      "step": 19938
-    },
-    {
-      "epoch": 10.398435462842242,
-      "grad_norm": 2.1654834747314453,
-      "learning_rate": 8.046934673366835e-05,
-      "loss": 5.0983,
-      "step": 19939
-    },
-    {
-      "epoch": 10.398956975228161,
-      "grad_norm": 1.4986932277679443,
-      "learning_rate": 8.046834170854272e-05,
-      "loss": 5.4858,
-      "step": 19940
-    },
-    {
-      "epoch": 10.399478487614081,
-      "grad_norm": 1.480557918548584,
-      "learning_rate": 8.04673366834171e-05,
-      "loss": 5.5685,
-      "step": 19941
-    },
-    {
-      "epoch": 10.4,
-      "grad_norm": 1.4317172765731812,
-      "learning_rate": 8.046633165829146e-05,
-      "loss": 5.2303,
-      "step": 19942
-    },
-    {
-      "epoch": 10.40052151238592,
-      "grad_norm": 1.5317397117614746,
-      "learning_rate": 8.046532663316584e-05,
-      "loss": 4.61,
-      "step": 19943
-    },
-    {
-      "epoch": 10.401043024771838,
-      "grad_norm": 1.4820213317871094,
-      "learning_rate": 8.04643216080402e-05,
-      "loss": 5.1816,
-      "step": 19944
-    },
-    {
-      "epoch": 10.401564537157757,
-      "grad_norm": 1.555635929107666,
-      "learning_rate": 8.046331658291457e-05,
-      "loss": 5.3381,
-      "step": 19945
-    },
-    {
-      "epoch": 10.402086049543676,
-      "grad_norm": 1.419935941696167,
-      "learning_rate": 8.046231155778894e-05,
-      "loss": 5.802,
-      "step": 19946
-    },
-    {
-      "epoch": 10.402607561929596,
-      "grad_norm": 1.4380462169647217,
-      "learning_rate": 8.046130653266332e-05,
-      "loss": 5.3628,
-      "step": 19947
-    },
-    {
-      "epoch": 10.403129074315515,
-      "grad_norm": 1.5624200105667114,
-      "learning_rate": 8.04603015075377e-05,
-      "loss": 5.2221,
-      "step": 19948
-    },
-    {
-      "epoch": 10.403650586701435,
-      "grad_norm": 1.4189268350601196,
-      "learning_rate": 8.045929648241206e-05,
-      "loss": 4.976,
-      "step": 19949
-    },
-    {
-      "epoch": 10.404172099087353,
-      "grad_norm": 1.3355695009231567,
-      "learning_rate": 8.045829145728644e-05,
-      "loss": 5.7669,
-      "step": 19950
-    },
-    {
-      "epoch": 10.404693611473272,
-      "grad_norm": 1.316667079925537,
-      "learning_rate": 8.04572864321608e-05,
-      "loss": 5.7616,
-      "step": 19951
-    },
-    {
-      "epoch": 10.405215123859191,
-      "grad_norm": 1.545291543006897,
-      "learning_rate": 8.045628140703518e-05,
-      "loss": 5.3583,
-      "step": 19952
-    },
-    {
-      "epoch": 10.405736636245111,
-      "grad_norm": 1.4335500001907349,
-      "learning_rate": 8.045527638190955e-05,
-      "loss": 5.5262,
-      "step": 19953
-    },
-    {
-      "epoch": 10.40625814863103,
-      "grad_norm": 1.4660751819610596,
-      "learning_rate": 8.045427135678393e-05,
-      "loss": 5.4173,
-      "step": 19954
-    },
-    {
-      "epoch": 10.40677966101695,
-      "grad_norm": 1.6030346155166626,
-      "learning_rate": 8.045326633165829e-05,
-      "loss": 4.7999,
-      "step": 19955
-    },
-    {
-      "epoch": 10.407301173402868,
-      "grad_norm": 1.493813157081604,
-      "learning_rate": 8.045226130653267e-05,
-      "loss": 5.0958,
-      "step": 19956
-    },
-    {
-      "epoch": 10.407822685788787,
-      "grad_norm": 1.4698222875595093,
-      "learning_rate": 8.045125628140705e-05,
-      "loss": 5.4604,
-      "step": 19957
-    },
-    {
-      "epoch": 10.408344198174706,
-      "grad_norm": 1.450211763381958,
-      "learning_rate": 8.045025125628141e-05,
-      "loss": 5.4088,
-      "step": 19958
-    },
-    {
-      "epoch": 10.408865710560626,
-      "grad_norm": 1.4533228874206543,
-      "learning_rate": 8.044924623115579e-05,
-      "loss": 5.2014,
-      "step": 19959
-    },
-    {
-      "epoch": 10.409387222946545,
-      "grad_norm": 1.3268141746520996,
-      "learning_rate": 8.044824120603015e-05,
-      "loss": 5.735,
-      "step": 19960
-    },
-    {
-      "epoch": 10.409908735332465,
-      "grad_norm": 1.3400477170944214,
-      "learning_rate": 8.044723618090453e-05,
-      "loss": 5.6407,
-      "step": 19961
-    },
-    {
-      "epoch": 10.410430247718383,
-      "grad_norm": 1.3470672369003296,
-      "learning_rate": 8.04462311557789e-05,
-      "loss": 5.8412,
-      "step": 19962
-    },
-    {
-      "epoch": 10.410951760104302,
-      "grad_norm": 1.3763611316680908,
-      "learning_rate": 8.044522613065327e-05,
-      "loss": 5.3828,
-      "step": 19963
-    },
-    {
-      "epoch": 10.411473272490221,
-      "grad_norm": 1.4590768814086914,
-      "learning_rate": 8.044422110552764e-05,
-      "loss": 5.4262,
-      "step": 19964
-    },
-    {
-      "epoch": 10.411994784876141,
-      "grad_norm": 1.4496899843215942,
-      "learning_rate": 8.044321608040201e-05,
-      "loss": 5.334,
-      "step": 19965
-    },
-    {
-      "epoch": 10.41251629726206,
-      "grad_norm": 1.3286716938018799,
-      "learning_rate": 8.044221105527638e-05,
-      "loss": 5.3623,
-      "step": 19966
-    },
-    {
-      "epoch": 10.41303780964798,
-      "grad_norm": 1.3421027660369873,
-      "learning_rate": 8.044120603015076e-05,
-      "loss": 5.7623,
-      "step": 19967
-    },
-    {
-      "epoch": 10.413559322033898,
-      "grad_norm": 1.4689563512802124,
-      "learning_rate": 8.044020100502513e-05,
-      "loss": 5.6136,
-      "step": 19968
-    },
-    {
-      "epoch": 10.414080834419817,
-      "grad_norm": 1.4162120819091797,
-      "learning_rate": 8.043919597989951e-05,
-      "loss": 5.5996,
-      "step": 19969
-    },
-    {
-      "epoch": 10.414602346805736,
-      "grad_norm": 1.5002307891845703,
-      "learning_rate": 8.043819095477388e-05,
-      "loss": 5.2115,
-      "step": 19970
-    },
-    {
-      "epoch": 10.415123859191656,
-      "grad_norm": 1.377344012260437,
-      "learning_rate": 8.043718592964824e-05,
-      "loss": 5.8095,
-      "step": 19971
-    },
-    {
-      "epoch": 10.415645371577575,
-      "grad_norm": 1.451331615447998,
-      "learning_rate": 8.043618090452262e-05,
-      "loss": 5.5103,
-      "step": 19972
-    },
-    {
-      "epoch": 10.416166883963495,
-      "grad_norm": 1.4327746629714966,
-      "learning_rate": 8.043517587939698e-05,
-      "loss": 5.7684,
-      "step": 19973
-    },
-    {
-      "epoch": 10.416688396349413,
-      "grad_norm": 1.4448026418685913,
-      "learning_rate": 8.043417085427136e-05,
-      "loss": 5.6955,
-      "step": 19974
-    },
-    {
-      "epoch": 10.417209908735332,
-      "grad_norm": 1.39994215965271,
-      "learning_rate": 8.043316582914572e-05,
-      "loss": 5.33,
-      "step": 19975
-    },
-    {
-      "epoch": 10.417731421121252,
-      "grad_norm": 1.6093553304672241,
-      "learning_rate": 8.04321608040201e-05,
-      "loss": 5.2564,
-      "step": 19976
-    },
-    {
-      "epoch": 10.418252933507171,
-      "grad_norm": 1.683631420135498,
-      "learning_rate": 8.043115577889448e-05,
-      "loss": 5.3948,
-      "step": 19977
-    },
-    {
-      "epoch": 10.41877444589309,
-      "grad_norm": 1.3730961084365845,
-      "learning_rate": 8.043015075376886e-05,
-      "loss": 5.709,
-      "step": 19978
-    },
-    {
-      "epoch": 10.41929595827901,
-      "grad_norm": 1.523729681968689,
-      "learning_rate": 8.042914572864322e-05,
-      "loss": 5.4784,
-      "step": 19979
-    },
-    {
-      "epoch": 10.419817470664928,
-      "grad_norm": 1.4242973327636719,
-      "learning_rate": 8.04281407035176e-05,
-      "loss": 5.2078,
-      "step": 19980
-    },
-    {
-      "epoch": 10.420338983050847,
-      "grad_norm": 1.529475450515747,
-      "learning_rate": 8.042713567839196e-05,
-      "loss": 5.0303,
-      "step": 19981
-    },
-    {
-      "epoch": 10.420860495436767,
-      "grad_norm": 1.6017335653305054,
-      "learning_rate": 8.042613065326634e-05,
-      "loss": 4.8492,
-      "step": 19982
-    },
-    {
-      "epoch": 10.421382007822686,
-      "grad_norm": 1.5167200565338135,
-      "learning_rate": 8.04251256281407e-05,
-      "loss": 4.7967,
-      "step": 19983
-    },
-    {
-      "epoch": 10.421903520208605,
-      "grad_norm": 1.3874661922454834,
-      "learning_rate": 8.042412060301507e-05,
-      "loss": 5.4967,
-      "step": 19984
-    },
-    {
-      "epoch": 10.422425032594525,
-      "grad_norm": 1.3371829986572266,
-      "learning_rate": 8.042311557788945e-05,
-      "loss": 5.6029,
-      "step": 19985
-    },
-    {
-      "epoch": 10.422946544980443,
-      "grad_norm": 1.4296631813049316,
-      "learning_rate": 8.042211055276381e-05,
-      "loss": 5.321,
-      "step": 19986
-    },
-    {
-      "epoch": 10.423468057366362,
-      "grad_norm": 1.345910668373108,
-      "learning_rate": 8.042110552763819e-05,
-      "loss": 5.575,
-      "step": 19987
-    },
-    {
-      "epoch": 10.423989569752282,
-      "grad_norm": 1.4190165996551514,
-      "learning_rate": 8.042010050251257e-05,
-      "loss": 4.9707,
-      "step": 19988
-    },
-    {
-      "epoch": 10.424511082138201,
-      "grad_norm": 1.3458001613616943,
-      "learning_rate": 8.041909547738695e-05,
-      "loss": 4.5988,
-      "step": 19989
-    },
-    {
-      "epoch": 10.42503259452412,
-      "grad_norm": 1.4076688289642334,
-      "learning_rate": 8.041809045226131e-05,
-      "loss": 5.7325,
-      "step": 19990
-    },
-    {
-      "epoch": 10.42555410691004,
-      "grad_norm": 1.3628008365631104,
-      "learning_rate": 8.041708542713569e-05,
-      "loss": 5.5956,
-      "step": 19991
-    },
-    {
-      "epoch": 10.426075619295958,
-      "grad_norm": 1.5019603967666626,
-      "learning_rate": 8.041608040201005e-05,
-      "loss": 4.9267,
-      "step": 19992
-    },
-    {
-      "epoch": 10.426597131681877,
-      "grad_norm": 1.5173205137252808,
-      "learning_rate": 8.041507537688443e-05,
-      "loss": 5.5556,
-      "step": 19993
-    },
-    {
-      "epoch": 10.427118644067797,
-      "grad_norm": 1.376816987991333,
-      "learning_rate": 8.04140703517588e-05,
-      "loss": 5.8866,
-      "step": 19994
-    },
-    {
-      "epoch": 10.427640156453716,
-      "grad_norm": 1.3881354331970215,
-      "learning_rate": 8.041306532663317e-05,
-      "loss": 5.5236,
-      "step": 19995
-    },
-    {
-      "epoch": 10.428161668839635,
-      "grad_norm": 1.3512701988220215,
-      "learning_rate": 8.041206030150754e-05,
-      "loss": 5.603,
-      "step": 19996
-    },
-    {
-      "epoch": 10.428683181225555,
-      "grad_norm": 1.4895601272583008,
-      "learning_rate": 8.041105527638191e-05,
-      "loss": 5.7758,
-      "step": 19997
-    },
-    {
-      "epoch": 10.429204693611473,
-      "grad_norm": 1.5127331018447876,
-      "learning_rate": 8.041005025125629e-05,
-      "loss": 5.6008,
-      "step": 19998
-    },
-    {
-      "epoch": 10.429726205997392,
-      "grad_norm": 1.3891011476516724,
-      "learning_rate": 8.040904522613066e-05,
-      "loss": 5.5819,
-      "step": 19999
-    },
-    {
-      "epoch": 10.430247718383312,
-      "grad_norm": 1.3282994031906128,
-      "learning_rate": 8.040804020100503e-05,
-      "loss": 5.8975,
-      "step": 20000
-    },
-    {
-      "epoch": 10.430247718383312,
-      "eval_loss": 5.543665885925293,
-      "eval_runtime": 42.6755,
-      "eval_samples_per_second": 28.728,
-      "eval_steps_per_second": 3.609,
-      "step": 20000
-    },
-    {
-      "epoch": 10.430769230769231,
-      "grad_norm": 1.529467225074768,
-      "learning_rate": 8.04070351758794e-05,
-      "loss": 5.1916,
-      "step": 20001
-    },
-    {
-      "epoch": 10.43129074315515,
-      "grad_norm": 1.4803346395492554,
-      "learning_rate": 8.040603015075378e-05,
-      "loss": 5.8947,
-      "step": 20002
-    },
-    {
-      "epoch": 10.43181225554107,
-      "grad_norm": 1.5069351196289062,
-      "learning_rate": 8.040502512562814e-05,
-      "loss": 5.4197,
-      "step": 20003
-    },
-    {
-      "epoch": 10.432333767926988,
-      "grad_norm": 1.2735505104064941,
-      "learning_rate": 8.040402010050252e-05,
-      "loss": 4.9513,
-      "step": 20004
-    },
-    {
-      "epoch": 10.432855280312907,
-      "grad_norm": 1.4791656732559204,
-      "learning_rate": 8.040301507537688e-05,
-      "loss": 5.4412,
-      "step": 20005
-    },
-    {
-      "epoch": 10.433376792698827,
-      "grad_norm": 1.4405359029769897,
-      "learning_rate": 8.040201005025126e-05,
-      "loss": 5.3529,
-      "step": 20006
-    },
-    {
-      "epoch": 10.433898305084746,
-      "grad_norm": 1.4260791540145874,
-      "learning_rate": 8.040100502512562e-05,
-      "loss": 5.6774,
-      "step": 20007
-    },
-    {
-      "epoch": 10.434419817470665,
-      "grad_norm": 1.3430771827697754,
-      "learning_rate": 8.04e-05,
-      "loss": 5.3872,
-      "step": 20008
-    },
-    {
-      "epoch": 10.434941329856585,
-      "grad_norm": 1.5009551048278809,
-      "learning_rate": 8.039899497487438e-05,
-      "loss": 5.656,
-      "step": 20009
-    },
-    {
-      "epoch": 10.435462842242503,
-      "grad_norm": 1.4954805374145508,
-      "learning_rate": 8.039798994974874e-05,
-      "loss": 5.5423,
-      "step": 20010
-    },
-    {
-      "epoch": 10.435984354628422,
-      "grad_norm": 1.4897936582565308,
-      "learning_rate": 8.039698492462312e-05,
-      "loss": 4.7234,
-      "step": 20011
-    },
-    {
-      "epoch": 10.436505867014342,
-      "grad_norm": 1.4408941268920898,
-      "learning_rate": 8.039597989949749e-05,
-      "loss": 5.6228,
-      "step": 20012
-    },
-    {
-      "epoch": 10.437027379400261,
-      "grad_norm": 1.435788631439209,
-      "learning_rate": 8.039497487437186e-05,
-      "loss": 4.7287,
-      "step": 20013
-    },
-    {
-      "epoch": 10.43754889178618,
-      "grad_norm": 1.4675769805908203,
-      "learning_rate": 8.039396984924623e-05,
-      "loss": 5.1364,
-      "step": 20014
-    },
-    {
-      "epoch": 10.438070404172098,
-      "grad_norm": 1.4676421880722046,
-      "learning_rate": 8.039296482412061e-05,
-      "loss": 5.652,
-      "step": 20015
-    },
-    {
-      "epoch": 10.438591916558018,
-      "grad_norm": 1.4743765592575073,
-      "learning_rate": 8.039195979899497e-05,
-      "loss": 5.5859,
-      "step": 20016
-    },
-    {
-      "epoch": 10.439113428943937,
-      "grad_norm": 1.6436294317245483,
-      "learning_rate": 8.039095477386935e-05,
-      "loss": 5.5575,
-      "step": 20017
-    },
-    {
-      "epoch": 10.439634941329857,
-      "grad_norm": 1.536979079246521,
-      "learning_rate": 8.038994974874373e-05,
-      "loss": 5.1156,
-      "step": 20018
-    },
-    {
-      "epoch": 10.440156453715776,
-      "grad_norm": 1.4624965190887451,
-      "learning_rate": 8.03889447236181e-05,
-      "loss": 5.3718,
-      "step": 20019
-    },
-    {
-      "epoch": 10.440677966101696,
-      "grad_norm": 1.5442885160446167,
-      "learning_rate": 8.038793969849247e-05,
-      "loss": 5.5724,
-      "step": 20020
-    },
-    {
-      "epoch": 10.441199478487615,
-      "grad_norm": 1.4140805006027222,
-      "learning_rate": 8.038693467336685e-05,
-      "loss": 5.3503,
-      "step": 20021
-    },
-    {
-      "epoch": 10.441720990873533,
-      "grad_norm": 1.4362179040908813,
-      "learning_rate": 8.038592964824121e-05,
-      "loss": 5.5993,
-      "step": 20022
-    },
-    {
-      "epoch": 10.442242503259452,
-      "grad_norm": 1.4394136667251587,
-      "learning_rate": 8.038492462311559e-05,
-      "loss": 5.1675,
-      "step": 20023
-    },
-    {
-      "epoch": 10.442764015645372,
-      "grad_norm": 1.3118693828582764,
-      "learning_rate": 8.038391959798995e-05,
-      "loss": 5.3249,
-      "step": 20024
-    },
-    {
-      "epoch": 10.443285528031291,
-      "grad_norm": 1.4695910215377808,
-      "learning_rate": 8.038291457286432e-05,
-      "loss": 5.2091,
-      "step": 20025
-    },
-    {
-      "epoch": 10.44380704041721,
-      "grad_norm": 1.607361078262329,
-      "learning_rate": 8.03819095477387e-05,
-      "loss": 4.823,
-      "step": 20026
-    },
-    {
-      "epoch": 10.444328552803128,
-      "grad_norm": 1.382102608680725,
-      "learning_rate": 8.038090452261306e-05,
-      "loss": 5.5601,
-      "step": 20027
-    },
-    {
-      "epoch": 10.444850065189048,
-      "grad_norm": 1.4521409273147583,
-      "learning_rate": 8.037989949748744e-05,
-      "loss": 5.7681,
-      "step": 20028
-    },
-    {
-      "epoch": 10.445371577574967,
-      "grad_norm": 1.4423408508300781,
-      "learning_rate": 8.037889447236182e-05,
-      "loss": 5.6067,
-      "step": 20029
-    },
-    {
-      "epoch": 10.445893089960887,
-      "grad_norm": 1.4133226871490479,
-      "learning_rate": 8.037788944723619e-05,
-      "loss": 5.8418,
-      "step": 20030
-    },
-    {
-      "epoch": 10.446414602346806,
-      "grad_norm": 1.4076740741729736,
-      "learning_rate": 8.037688442211056e-05,
-      "loss": 5.4703,
-      "step": 20031
-    },
-    {
-      "epoch": 10.446936114732726,
-      "grad_norm": 1.3780018091201782,
-      "learning_rate": 8.037587939698494e-05,
-      "loss": 5.4979,
-      "step": 20032
-    },
-    {
-      "epoch": 10.447457627118643,
-      "grad_norm": 1.4381235837936401,
-      "learning_rate": 8.03748743718593e-05,
-      "loss": 5.9097,
-      "step": 20033
-    },
-    {
-      "epoch": 10.447979139504563,
-      "grad_norm": 1.3625723123550415,
-      "learning_rate": 8.037386934673368e-05,
-      "loss": 5.2353,
-      "step": 20034
-    },
-    {
-      "epoch": 10.448500651890482,
-      "grad_norm": 1.3680737018585205,
-      "learning_rate": 8.037286432160804e-05,
-      "loss": 5.9323,
-      "step": 20035
-    },
-    {
-      "epoch": 10.449022164276402,
-      "grad_norm": 1.349200963973999,
-      "learning_rate": 8.037185929648242e-05,
-      "loss": 5.5433,
-      "step": 20036
-    },
-    {
-      "epoch": 10.449543676662321,
-      "grad_norm": 1.517087697982788,
-      "learning_rate": 8.037085427135678e-05,
-      "loss": 5.1883,
-      "step": 20037
-    },
-    {
-      "epoch": 10.45006518904824,
-      "grad_norm": 1.662245750427246,
-      "learning_rate": 8.036984924623116e-05,
-      "loss": 5.1363,
-      "step": 20038
-    },
-    {
-      "epoch": 10.450586701434158,
-      "grad_norm": 1.4189982414245605,
-      "learning_rate": 8.036884422110554e-05,
-      "loss": 5.1179,
-      "step": 20039
-    },
-    {
-      "epoch": 10.451108213820078,
-      "grad_norm": 1.4751296043395996,
-      "learning_rate": 8.03678391959799e-05,
-      "loss": 5.5195,
-      "step": 20040
-    },
-    {
-      "epoch": 10.451629726205997,
-      "grad_norm": 1.4925450086593628,
-      "learning_rate": 8.036683417085428e-05,
-      "loss": 5.5955,
-      "step": 20041
-    },
-    {
-      "epoch": 10.452151238591917,
-      "grad_norm": 1.6336311101913452,
-      "learning_rate": 8.036582914572865e-05,
-      "loss": 5.1711,
-      "step": 20042
-    },
-    {
-      "epoch": 10.452672750977836,
-      "grad_norm": 1.4046108722686768,
-      "learning_rate": 8.036482412060302e-05,
-      "loss": 5.0925,
-      "step": 20043
-    },
-    {
-      "epoch": 10.453194263363756,
-      "grad_norm": 1.3398125171661377,
-      "learning_rate": 8.036381909547739e-05,
-      "loss": 5.7482,
-      "step": 20044
-    },
-    {
-      "epoch": 10.453715775749673,
-      "grad_norm": 1.4135229587554932,
-      "learning_rate": 8.036281407035177e-05,
-      "loss": 5.4997,
-      "step": 20045
-    },
-    {
-      "epoch": 10.454237288135593,
-      "grad_norm": 1.3738007545471191,
-      "learning_rate": 8.036180904522613e-05,
-      "loss": 5.6229,
-      "step": 20046
-    },
-    {
-      "epoch": 10.454758800521512,
-      "grad_norm": 1.3398128747940063,
-      "learning_rate": 8.036080402010051e-05,
-      "loss": 5.8303,
-      "step": 20047
-    },
-    {
-      "epoch": 10.455280312907432,
-      "grad_norm": 1.3650574684143066,
-      "learning_rate": 8.035979899497487e-05,
-      "loss": 5.6673,
-      "step": 20048
-    },
-    {
-      "epoch": 10.455801825293351,
-      "grad_norm": 1.3992582559585571,
-      "learning_rate": 8.035879396984925e-05,
-      "loss": 5.7969,
-      "step": 20049
-    },
-    {
-      "epoch": 10.45632333767927,
-      "grad_norm": 1.5242825746536255,
-      "learning_rate": 8.035778894472363e-05,
-      "loss": 5.5548,
-      "step": 20050
-    },
-    {
-      "epoch": 10.456844850065188,
-      "grad_norm": 1.4391740560531616,
-      "learning_rate": 8.035678391959799e-05,
-      "loss": 5.346,
-      "step": 20051
-    },
-    {
-      "epoch": 10.457366362451108,
-      "grad_norm": 1.3462814092636108,
-      "learning_rate": 8.035577889447237e-05,
-      "loss": 5.3232,
-      "step": 20052
-    },
-    {
-      "epoch": 10.457887874837027,
-      "grad_norm": 1.4395314455032349,
-      "learning_rate": 8.035477386934673e-05,
-      "loss": 5.6247,
-      "step": 20053
-    },
-    {
-      "epoch": 10.458409387222947,
-      "grad_norm": 1.5495858192443848,
-      "learning_rate": 8.035376884422111e-05,
-      "loss": 5.2577,
-      "step": 20054
-    },
-    {
-      "epoch": 10.458930899608866,
-      "grad_norm": 1.5547749996185303,
-      "learning_rate": 8.035276381909548e-05,
-      "loss": 5.3037,
-      "step": 20055
-    },
-    {
-      "epoch": 10.459452411994786,
-      "grad_norm": 1.4934298992156982,
-      "learning_rate": 8.035175879396985e-05,
-      "loss": 5.4626,
-      "step": 20056
-    },
-    {
-      "epoch": 10.459973924380703,
-      "grad_norm": 1.421324372291565,
-      "learning_rate": 8.035075376884422e-05,
-      "loss": 5.7333,
-      "step": 20057
-    },
-    {
-      "epoch": 10.460495436766623,
-      "grad_norm": 1.4620543718338013,
-      "learning_rate": 8.03497487437186e-05,
-      "loss": 5.8741,
-      "step": 20058
-    },
-    {
-      "epoch": 10.461016949152542,
-      "grad_norm": 1.35321044921875,
-      "learning_rate": 8.034874371859297e-05,
-      "loss": 5.4649,
-      "step": 20059
-    },
-    {
-      "epoch": 10.461538461538462,
-      "grad_norm": 1.3660801649093628,
-      "learning_rate": 8.034773869346735e-05,
-      "loss": 5.7613,
-      "step": 20060
-    },
-    {
-      "epoch": 10.462059973924381,
-      "grad_norm": 1.3329601287841797,
-      "learning_rate": 8.034673366834172e-05,
-      "loss": 5.7614,
-      "step": 20061
-    },
-    {
-      "epoch": 10.4625814863103,
-      "grad_norm": 1.40496826171875,
-      "learning_rate": 8.03457286432161e-05,
-      "loss": 5.0708,
-      "step": 20062
-    },
-    {
-      "epoch": 10.463102998696218,
-      "grad_norm": 1.4888346195220947,
-      "learning_rate": 8.034472361809046e-05,
-      "loss": 5.2961,
-      "step": 20063
-    },
-    {
-      "epoch": 10.463624511082138,
-      "grad_norm": 1.4310290813446045,
-      "learning_rate": 8.034371859296482e-05,
-      "loss": 5.5376,
-      "step": 20064
-    },
-    {
-      "epoch": 10.464146023468057,
-      "grad_norm": 1.488019585609436,
-      "learning_rate": 8.03427135678392e-05,
-      "loss": 5.4115,
-      "step": 20065
-    },
-    {
-      "epoch": 10.464667535853977,
-      "grad_norm": 1.447433590888977,
-      "learning_rate": 8.034170854271356e-05,
-      "loss": 5.5074,
-      "step": 20066
-    },
-    {
-      "epoch": 10.465189048239896,
-      "grad_norm": 1.3938887119293213,
-      "learning_rate": 8.034070351758794e-05,
-      "loss": 5.3147,
-      "step": 20067
-    },
-    {
-      "epoch": 10.465710560625816,
-      "grad_norm": 1.459694504737854,
-      "learning_rate": 8.03396984924623e-05,
-      "loss": 4.9825,
-      "step": 20068
-    },
-    {
-      "epoch": 10.466232073011733,
-      "grad_norm": 1.3934952020645142,
-      "learning_rate": 8.033869346733668e-05,
-      "loss": 5.8073,
-      "step": 20069
-    },
-    {
-      "epoch": 10.466753585397653,
-      "grad_norm": 1.4346258640289307,
-      "learning_rate": 8.033768844221106e-05,
-      "loss": 4.9881,
-      "step": 20070
-    },
-    {
-      "epoch": 10.467275097783572,
-      "grad_norm": 1.7041927576065063,
-      "learning_rate": 8.033668341708544e-05,
-      "loss": 4.8441,
-      "step": 20071
-    },
-    {
-      "epoch": 10.467796610169492,
-      "grad_norm": 1.3612425327301025,
-      "learning_rate": 8.03356783919598e-05,
-      "loss": 5.5077,
-      "step": 20072
-    },
-    {
-      "epoch": 10.468318122555411,
-      "grad_norm": 1.420318603515625,
-      "learning_rate": 8.033467336683418e-05,
-      "loss": 5.7527,
-      "step": 20073
-    },
-    {
-      "epoch": 10.46883963494133,
-      "grad_norm": 1.5034620761871338,
-      "learning_rate": 8.033366834170855e-05,
-      "loss": 5.6687,
-      "step": 20074
-    },
-    {
-      "epoch": 10.469361147327248,
-      "grad_norm": 1.3948981761932373,
-      "learning_rate": 8.033266331658292e-05,
-      "loss": 5.4495,
-      "step": 20075
-    },
-    {
-      "epoch": 10.469882659713168,
-      "grad_norm": 1.4296536445617676,
-      "learning_rate": 8.033165829145729e-05,
-      "loss": 5.3257,
-      "step": 20076
-    },
-    {
-      "epoch": 10.470404172099087,
-      "grad_norm": 1.371720552444458,
-      "learning_rate": 8.033065326633165e-05,
-      "loss": 5.5087,
-      "step": 20077
-    },
-    {
-      "epoch": 10.470925684485007,
-      "grad_norm": 1.3871865272521973,
-      "learning_rate": 8.032964824120603e-05,
-      "loss": 5.8193,
-      "step": 20078
-    },
-    {
-      "epoch": 10.471447196870926,
-      "grad_norm": 1.447293758392334,
-      "learning_rate": 8.03286432160804e-05,
-      "loss": 5.1732,
-      "step": 20079
-    },
-    {
-      "epoch": 10.471968709256846,
-      "grad_norm": 1.4432779550552368,
-      "learning_rate": 8.032763819095477e-05,
-      "loss": 5.0305,
-      "step": 20080
-    },
-    {
-      "epoch": 10.472490221642763,
-      "grad_norm": 1.597410798072815,
-      "learning_rate": 8.032663316582915e-05,
-      "loss": 5.6692,
-      "step": 20081
-    },
-    {
-      "epoch": 10.473011734028683,
-      "grad_norm": 1.4556232690811157,
-      "learning_rate": 8.032562814070353e-05,
-      "loss": 5.549,
-      "step": 20082
-    },
-    {
-      "epoch": 10.473533246414602,
-      "grad_norm": 1.5208501815795898,
-      "learning_rate": 8.032462311557789e-05,
-      "loss": 5.3184,
-      "step": 20083
-    },
-    {
-      "epoch": 10.474054758800522,
-      "grad_norm": 1.4683654308319092,
-      "learning_rate": 8.032361809045227e-05,
-      "loss": 5.207,
-      "step": 20084
-    },
-    {
-      "epoch": 10.474576271186441,
-      "grad_norm": 1.6169641017913818,
-      "learning_rate": 8.032261306532663e-05,
-      "loss": 5.5924,
-      "step": 20085
-    },
-    {
-      "epoch": 10.47509778357236,
-      "grad_norm": 1.363630771636963,
-      "learning_rate": 8.032160804020101e-05,
-      "loss": 5.502,
-      "step": 20086
-    },
-    {
-      "epoch": 10.475619295958278,
-      "grad_norm": 1.439030408859253,
-      "learning_rate": 8.032060301507538e-05,
-      "loss": 5.4995,
-      "step": 20087
-    },
-    {
-      "epoch": 10.476140808344198,
-      "grad_norm": 1.3769181966781616,
-      "learning_rate": 8.031959798994975e-05,
-      "loss": 5.8242,
-      "step": 20088
-    },
-    {
-      "epoch": 10.476662320730117,
-      "grad_norm": 1.316177248954773,
-      "learning_rate": 8.031859296482412e-05,
-      "loss": 5.7361,
-      "step": 20089
-    },
-    {
-      "epoch": 10.477183833116037,
-      "grad_norm": 1.4474636316299438,
-      "learning_rate": 8.03175879396985e-05,
-      "loss": 5.3925,
-      "step": 20090
-    },
-    {
-      "epoch": 10.477705345501956,
-      "grad_norm": 1.381136178970337,
-      "learning_rate": 8.031658291457287e-05,
-      "loss": 5.4628,
-      "step": 20091
-    },
-    {
-      "epoch": 10.478226857887876,
-      "grad_norm": 1.589564323425293,
-      "learning_rate": 8.031557788944724e-05,
-      "loss": 5.0696,
-      "step": 20092
-    },
-    {
-      "epoch": 10.478748370273793,
-      "grad_norm": 1.5227071046829224,
-      "learning_rate": 8.031457286432162e-05,
-      "loss": 5.511,
-      "step": 20093
-    },
-    {
-      "epoch": 10.479269882659713,
-      "grad_norm": 1.4458715915679932,
-      "learning_rate": 8.031356783919598e-05,
-      "loss": 4.9456,
-      "step": 20094
-    },
-    {
-      "epoch": 10.479791395045632,
-      "grad_norm": 1.332492470741272,
-      "learning_rate": 8.031256281407036e-05,
-      "loss": 5.4605,
-      "step": 20095
-    },
-    {
-      "epoch": 10.480312907431552,
-      "grad_norm": 1.4579753875732422,
-      "learning_rate": 8.031155778894472e-05,
-      "loss": 5.3445,
-      "step": 20096
-    },
-    {
-      "epoch": 10.480834419817471,
-      "grad_norm": 1.5495436191558838,
-      "learning_rate": 8.03105527638191e-05,
-      "loss": 4.9402,
-      "step": 20097
-    },
-    {
-      "epoch": 10.48135593220339,
-      "grad_norm": 1.3392939567565918,
-      "learning_rate": 8.030954773869347e-05,
-      "loss": 5.7875,
-      "step": 20098
-    },
-    {
-      "epoch": 10.481877444589308,
-      "grad_norm": 1.437435269355774,
-      "learning_rate": 8.030854271356784e-05,
-      "loss": 5.5514,
-      "step": 20099
-    },
-    {
-      "epoch": 10.482398956975228,
-      "grad_norm": 1.5421310663223267,
-      "learning_rate": 8.030753768844221e-05,
-      "loss": 5.3931,
-      "step": 20100
-    },
-    {
-      "epoch": 10.482920469361147,
-      "grad_norm": 1.4569154977798462,
-      "learning_rate": 8.030653266331659e-05,
-      "loss": 5.7439,
-      "step": 20101
-    },
-    {
-      "epoch": 10.483441981747067,
-      "grad_norm": 1.448169231414795,
-      "learning_rate": 8.030552763819096e-05,
-      "loss": 5.4705,
-      "step": 20102
-    },
-    {
-      "epoch": 10.483963494132986,
-      "grad_norm": 1.540806531906128,
-      "learning_rate": 8.030452261306533e-05,
-      "loss": 5.2795,
-      "step": 20103
-    },
-    {
-      "epoch": 10.484485006518906,
-      "grad_norm": 1.6125191450119019,
-      "learning_rate": 8.03035175879397e-05,
-      "loss": 5.1977,
-      "step": 20104
-    },
-    {
-      "epoch": 10.485006518904823,
-      "grad_norm": 1.4899132251739502,
-      "learning_rate": 8.030251256281407e-05,
-      "loss": 5.5326,
-      "step": 20105
-    },
-    {
-      "epoch": 10.485528031290743,
-      "grad_norm": 1.6312202215194702,
-      "learning_rate": 8.030150753768845e-05,
-      "loss": 5.033,
-      "step": 20106
-    },
-    {
-      "epoch": 10.486049543676662,
-      "grad_norm": 1.4905378818511963,
-      "learning_rate": 8.030050251256281e-05,
-      "loss": 5.4171,
-      "step": 20107
-    },
-    {
-      "epoch": 10.486571056062582,
-      "grad_norm": 1.5115697383880615,
-      "learning_rate": 8.029949748743719e-05,
-      "loss": 4.9864,
-      "step": 20108
-    },
-    {
-      "epoch": 10.487092568448501,
-      "grad_norm": 1.4244314432144165,
-      "learning_rate": 8.029849246231155e-05,
-      "loss": 5.1203,
-      "step": 20109
-    },
-    {
-      "epoch": 10.487614080834419,
-      "grad_norm": 1.4433428049087524,
-      "learning_rate": 8.029748743718593e-05,
-      "loss": 4.7104,
-      "step": 20110
-    },
-    {
-      "epoch": 10.488135593220338,
-      "grad_norm": 1.5289520025253296,
-      "learning_rate": 8.029648241206031e-05,
-      "loss": 5.3702,
-      "step": 20111
-    },
-    {
-      "epoch": 10.488657105606258,
-      "grad_norm": 1.4814295768737793,
-      "learning_rate": 8.029547738693469e-05,
-      "loss": 5.4001,
-      "step": 20112
-    },
-    {
-      "epoch": 10.489178617992177,
-      "grad_norm": 1.4356199502944946,
-      "learning_rate": 8.029447236180905e-05,
-      "loss": 5.5284,
-      "step": 20113
-    },
-    {
-      "epoch": 10.489700130378097,
-      "grad_norm": 1.4398802518844604,
-      "learning_rate": 8.029346733668343e-05,
-      "loss": 5.4107,
-      "step": 20114
-    },
-    {
-      "epoch": 10.490221642764016,
-      "grad_norm": 1.5544341802597046,
-      "learning_rate": 8.02924623115578e-05,
-      "loss": 5.1624,
-      "step": 20115
-    },
-    {
-      "epoch": 10.490743155149936,
-      "grad_norm": 1.3757280111312866,
-      "learning_rate": 8.029145728643217e-05,
-      "loss": 5.3625,
-      "step": 20116
-    },
-    {
-      "epoch": 10.491264667535853,
-      "grad_norm": 1.5161798000335693,
-      "learning_rate": 8.029045226130654e-05,
-      "loss": 5.4733,
-      "step": 20117
-    },
-    {
-      "epoch": 10.491786179921773,
-      "grad_norm": 1.4094661474227905,
-      "learning_rate": 8.02894472361809e-05,
-      "loss": 5.8116,
-      "step": 20118
-    },
-    {
-      "epoch": 10.492307692307692,
-      "grad_norm": 1.4476978778839111,
-      "learning_rate": 8.028844221105528e-05,
-      "loss": 5.4741,
-      "step": 20119
-    },
-    {
-      "epoch": 10.492829204693612,
-      "grad_norm": 1.312962532043457,
-      "learning_rate": 8.028743718592964e-05,
-      "loss": 5.5957,
-      "step": 20120
-    },
-    {
-      "epoch": 10.493350717079531,
-      "grad_norm": 1.5023175477981567,
-      "learning_rate": 8.028643216080402e-05,
-      "loss": 5.3582,
-      "step": 20121
-    },
-    {
-      "epoch": 10.493872229465449,
-      "grad_norm": 1.5131431818008423,
-      "learning_rate": 8.02854271356784e-05,
-      "loss": 5.2185,
-      "step": 20122
-    },
-    {
-      "epoch": 10.494393741851368,
-      "grad_norm": 1.4092590808868408,
-      "learning_rate": 8.028442211055278e-05,
-      "loss": 5.5067,
-      "step": 20123
-    },
-    {
-      "epoch": 10.494915254237288,
-      "grad_norm": 1.3934842348098755,
-      "learning_rate": 8.028341708542714e-05,
-      "loss": 5.6525,
-      "step": 20124
-    },
-    {
-      "epoch": 10.495436766623207,
-      "grad_norm": 1.29954993724823,
-      "learning_rate": 8.028241206030152e-05,
-      "loss": 5.8051,
-      "step": 20125
-    },
-    {
-      "epoch": 10.495958279009127,
-      "grad_norm": 1.6116315126419067,
-      "learning_rate": 8.028140703517588e-05,
-      "loss": 4.9881,
-      "step": 20126
-    },
-    {
-      "epoch": 10.496479791395046,
-      "grad_norm": 1.3698092699050903,
-      "learning_rate": 8.028040201005026e-05,
-      "loss": 5.2496,
-      "step": 20127
-    },
-    {
-      "epoch": 10.497001303780964,
-      "grad_norm": 1.3783009052276611,
-      "learning_rate": 8.027939698492462e-05,
-      "loss": 5.7843,
-      "step": 20128
-    },
-    {
-      "epoch": 10.497522816166883,
-      "grad_norm": 1.4377788305282593,
-      "learning_rate": 8.0278391959799e-05,
-      "loss": 5.8222,
-      "step": 20129
-    },
-    {
-      "epoch": 10.498044328552803,
-      "grad_norm": 1.45806884765625,
-      "learning_rate": 8.027738693467337e-05,
-      "loss": 5.6618,
-      "step": 20130
-    },
-    {
-      "epoch": 10.498565840938722,
-      "grad_norm": 1.3856240510940552,
-      "learning_rate": 8.027638190954774e-05,
-      "loss": 5.8062,
-      "step": 20131
-    },
-    {
-      "epoch": 10.499087353324642,
-      "grad_norm": 1.4183002710342407,
-      "learning_rate": 8.027537688442212e-05,
-      "loss": 5.5042,
-      "step": 20132
-    },
-    {
-      "epoch": 10.499608865710561,
-      "grad_norm": 1.3297879695892334,
-      "learning_rate": 8.027437185929649e-05,
-      "loss": 5.0398,
-      "step": 20133
-    },
-    {
-      "epoch": 10.500130378096479,
-      "grad_norm": 1.342708945274353,
-      "learning_rate": 8.027336683417086e-05,
-      "loss": 5.6797,
-      "step": 20134
-    },
-    {
-      "epoch": 10.500651890482398,
-      "grad_norm": 1.3475439548492432,
-      "learning_rate": 8.027236180904523e-05,
-      "loss": 5.9379,
-      "step": 20135
-    },
-    {
-      "epoch": 10.501173402868318,
-      "grad_norm": 1.365881085395813,
-      "learning_rate": 8.02713567839196e-05,
-      "loss": 5.8206,
-      "step": 20136
-    },
-    {
-      "epoch": 10.501694915254237,
-      "grad_norm": 1.3986566066741943,
-      "learning_rate": 8.027035175879397e-05,
-      "loss": 5.1699,
-      "step": 20137
-    },
-    {
-      "epoch": 10.502216427640157,
-      "grad_norm": 1.4930223226547241,
-      "learning_rate": 8.026934673366835e-05,
-      "loss": 5.1956,
-      "step": 20138
-    },
-    {
-      "epoch": 10.502737940026076,
-      "grad_norm": 1.4285295009613037,
-      "learning_rate": 8.026834170854271e-05,
-      "loss": 5.5209,
-      "step": 20139
-    },
-    {
-      "epoch": 10.503259452411994,
-      "grad_norm": 1.431732177734375,
-      "learning_rate": 8.026733668341709e-05,
-      "loss": 5.3187,
-      "step": 20140
-    },
-    {
-      "epoch": 10.503780964797913,
-      "grad_norm": 1.3656694889068604,
-      "learning_rate": 8.026633165829145e-05,
-      "loss": 5.8665,
-      "step": 20141
-    },
-    {
-      "epoch": 10.504302477183833,
-      "grad_norm": 1.4408859014511108,
-      "learning_rate": 8.026532663316583e-05,
-      "loss": 5.8001,
-      "step": 20142
-    },
-    {
-      "epoch": 10.504823989569752,
-      "grad_norm": 1.3333592414855957,
-      "learning_rate": 8.026432160804021e-05,
-      "loss": 5.7296,
-      "step": 20143
-    },
-    {
-      "epoch": 10.505345501955672,
-      "grad_norm": 1.326034426689148,
-      "learning_rate": 8.026331658291457e-05,
-      "loss": 5.3533,
-      "step": 20144
-    },
-    {
-      "epoch": 10.505867014341591,
-      "grad_norm": 1.6271909475326538,
-      "learning_rate": 8.026231155778895e-05,
-      "loss": 5.2683,
-      "step": 20145
-    },
-    {
-      "epoch": 10.506388526727509,
-      "grad_norm": 1.423873782157898,
-      "learning_rate": 8.026130653266332e-05,
-      "loss": 5.1823,
-      "step": 20146
-    },
-    {
-      "epoch": 10.506910039113428,
-      "grad_norm": 1.4594800472259521,
-      "learning_rate": 8.02603015075377e-05,
-      "loss": 5.0716,
-      "step": 20147
-    },
-    {
-      "epoch": 10.507431551499348,
-      "grad_norm": 1.4479162693023682,
-      "learning_rate": 8.025929648241206e-05,
-      "loss": 5.438,
-      "step": 20148
-    },
-    {
-      "epoch": 10.507953063885267,
-      "grad_norm": 1.7412333488464355,
-      "learning_rate": 8.025829145728644e-05,
-      "loss": 4.7908,
-      "step": 20149
-    },
-    {
-      "epoch": 10.508474576271187,
-      "grad_norm": 1.4621638059616089,
-      "learning_rate": 8.02572864321608e-05,
-      "loss": 5.6346,
-      "step": 20150
-    },
-    {
-      "epoch": 10.508996088657106,
-      "grad_norm": 1.4613630771636963,
-      "learning_rate": 8.025628140703518e-05,
-      "loss": 5.6434,
-      "step": 20151
-    },
-    {
-      "epoch": 10.509517601043024,
-      "grad_norm": 1.458966612815857,
-      "learning_rate": 8.025527638190956e-05,
-      "loss": 5.5293,
-      "step": 20152
-    },
-    {
-      "epoch": 10.510039113428943,
-      "grad_norm": 1.5054205656051636,
-      "learning_rate": 8.025427135678393e-05,
-      "loss": 4.3591,
-      "step": 20153
-    },
-    {
-      "epoch": 10.510560625814863,
-      "grad_norm": 1.5300554037094116,
-      "learning_rate": 8.02532663316583e-05,
-      "loss": 5.0629,
-      "step": 20154
-    },
-    {
-      "epoch": 10.511082138200782,
-      "grad_norm": 1.4448838233947754,
-      "learning_rate": 8.025226130653268e-05,
-      "loss": 5.4664,
-      "step": 20155
-    },
-    {
-      "epoch": 10.511603650586702,
-      "grad_norm": 1.4856313467025757,
-      "learning_rate": 8.025125628140704e-05,
-      "loss": 5.104,
-      "step": 20156
-    },
-    {
-      "epoch": 10.512125162972621,
-      "grad_norm": 1.2974441051483154,
-      "learning_rate": 8.02502512562814e-05,
-      "loss": 5.5306,
-      "step": 20157
-    },
-    {
-      "epoch": 10.512646675358539,
-      "grad_norm": 1.3960912227630615,
-      "learning_rate": 8.024924623115578e-05,
-      "loss": 5.4729,
-      "step": 20158
-    },
-    {
-      "epoch": 10.513168187744458,
-      "grad_norm": 1.4792180061340332,
-      "learning_rate": 8.024824120603015e-05,
-      "loss": 5.1307,
-      "step": 20159
-    },
-    {
-      "epoch": 10.513689700130378,
-      "grad_norm": 1.4676841497421265,
-      "learning_rate": 8.024723618090452e-05,
-      "loss": 5.1799,
-      "step": 20160
-    },
-    {
-      "epoch": 10.514211212516297,
-      "grad_norm": 1.419899344444275,
-      "learning_rate": 8.024623115577889e-05,
-      "loss": 4.7953,
-      "step": 20161
-    },
-    {
-      "epoch": 10.514732724902217,
-      "grad_norm": 1.4318413734436035,
-      "learning_rate": 8.024522613065327e-05,
-      "loss": 5.1573,
-      "step": 20162
-    },
-    {
-      "epoch": 10.515254237288136,
-      "grad_norm": 1.450117588043213,
-      "learning_rate": 8.024422110552764e-05,
-      "loss": 5.2572,
-      "step": 20163
-    },
-    {
-      "epoch": 10.515775749674054,
-      "grad_norm": 1.4222686290740967,
-      "learning_rate": 8.024321608040202e-05,
-      "loss": 5.9119,
-      "step": 20164
-    },
-    {
-      "epoch": 10.516297262059974,
-      "grad_norm": 1.4318339824676514,
-      "learning_rate": 8.024221105527639e-05,
-      "loss": 5.7185,
-      "step": 20165
-    },
-    {
-      "epoch": 10.516818774445893,
-      "grad_norm": 1.3940114974975586,
-      "learning_rate": 8.024120603015076e-05,
-      "loss": 5.7025,
-      "step": 20166
-    },
-    {
-      "epoch": 10.517340286831812,
-      "grad_norm": 1.4331717491149902,
-      "learning_rate": 8.024020100502513e-05,
-      "loss": 5.4105,
-      "step": 20167
-    },
-    {
-      "epoch": 10.517861799217732,
-      "grad_norm": 1.5807502269744873,
-      "learning_rate": 8.02391959798995e-05,
-      "loss": 5.3105,
-      "step": 20168
-    },
-    {
-      "epoch": 10.518383311603651,
-      "grad_norm": 1.4683704376220703,
-      "learning_rate": 8.023819095477387e-05,
-      "loss": 5.586,
-      "step": 20169
-    },
-    {
-      "epoch": 10.518904823989569,
-      "grad_norm": 1.5349172353744507,
-      "learning_rate": 8.023718592964824e-05,
-      "loss": 5.5554,
-      "step": 20170
-    },
-    {
-      "epoch": 10.519426336375489,
-      "grad_norm": 1.45823073387146,
-      "learning_rate": 8.023618090452261e-05,
-      "loss": 5.4825,
-      "step": 20171
-    },
-    {
-      "epoch": 10.519947848761408,
-      "grad_norm": 1.3578556776046753,
-      "learning_rate": 8.023517587939699e-05,
-      "loss": 5.7894,
-      "step": 20172
-    },
-    {
-      "epoch": 10.520469361147327,
-      "grad_norm": 1.4050040245056152,
-      "learning_rate": 8.023417085427137e-05,
-      "loss": 5.2001,
-      "step": 20173
-    },
-    {
-      "epoch": 10.520990873533247,
-      "grad_norm": 1.42388117313385,
-      "learning_rate": 8.023316582914573e-05,
-      "loss": 5.6446,
-      "step": 20174
-    },
-    {
-      "epoch": 10.521512385919166,
-      "grad_norm": 1.5353286266326904,
-      "learning_rate": 8.023216080402011e-05,
-      "loss": 5.343,
-      "step": 20175
-    },
-    {
-      "epoch": 10.522033898305084,
-      "grad_norm": 1.4062973260879517,
-      "learning_rate": 8.023115577889447e-05,
-      "loss": 5.5191,
-      "step": 20176
-    },
-    {
-      "epoch": 10.522555410691004,
-      "grad_norm": 1.4024502038955688,
-      "learning_rate": 8.023015075376885e-05,
-      "loss": 5.2884,
-      "step": 20177
-    },
-    {
-      "epoch": 10.523076923076923,
-      "grad_norm": 1.353084683418274,
-      "learning_rate": 8.022914572864322e-05,
-      "loss": 5.7559,
-      "step": 20178
-    },
-    {
-      "epoch": 10.523598435462842,
-      "grad_norm": 1.3515088558197021,
-      "learning_rate": 8.02281407035176e-05,
-      "loss": 4.7229,
-      "step": 20179
-    },
-    {
-      "epoch": 10.524119947848762,
-      "grad_norm": 1.410803198814392,
-      "learning_rate": 8.022713567839196e-05,
-      "loss": 5.7901,
-      "step": 20180
-    },
-    {
-      "epoch": 10.524641460234681,
-      "grad_norm": 1.5132052898406982,
-      "learning_rate": 8.022613065326634e-05,
-      "loss": 5.6676,
-      "step": 20181
-    },
-    {
-      "epoch": 10.525162972620599,
-      "grad_norm": 1.5468604564666748,
-      "learning_rate": 8.02251256281407e-05,
-      "loss": 5.3625,
-      "step": 20182
-    },
-    {
-      "epoch": 10.525684485006519,
-      "grad_norm": 1.5415101051330566,
-      "learning_rate": 8.022412060301508e-05,
-      "loss": 5.4442,
-      "step": 20183
-    },
-    {
-      "epoch": 10.526205997392438,
-      "grad_norm": 1.4193804264068604,
-      "learning_rate": 8.022311557788946e-05,
-      "loss": 5.1048,
-      "step": 20184
-    },
-    {
-      "epoch": 10.526727509778357,
-      "grad_norm": 1.4701184034347534,
-      "learning_rate": 8.022211055276382e-05,
-      "loss": 5.423,
-      "step": 20185
-    },
-    {
-      "epoch": 10.527249022164277,
-      "grad_norm": 1.298774003982544,
-      "learning_rate": 8.02211055276382e-05,
-      "loss": 4.9924,
-      "step": 20186
-    },
-    {
-      "epoch": 10.527770534550196,
-      "grad_norm": 1.4836339950561523,
-      "learning_rate": 8.022010050251256e-05,
-      "loss": 5.3841,
-      "step": 20187
-    },
-    {
-      "epoch": 10.528292046936114,
-      "grad_norm": 1.6865088939666748,
-      "learning_rate": 8.021909547738694e-05,
-      "loss": 5.0764,
-      "step": 20188
-    },
-    {
-      "epoch": 10.528813559322034,
-      "grad_norm": 1.4704179763793945,
-      "learning_rate": 8.02180904522613e-05,
-      "loss": 5.6315,
-      "step": 20189
-    },
-    {
-      "epoch": 10.529335071707953,
-      "grad_norm": 1.4819533824920654,
-      "learning_rate": 8.021708542713568e-05,
-      "loss": 5.3103,
-      "step": 20190
-    },
-    {
-      "epoch": 10.529856584093872,
-      "grad_norm": 1.4180898666381836,
-      "learning_rate": 8.021608040201005e-05,
-      "loss": 5.6733,
-      "step": 20191
-    },
-    {
-      "epoch": 10.530378096479792,
-      "grad_norm": 1.5153095722198486,
-      "learning_rate": 8.021507537688443e-05,
-      "loss": 5.3496,
-      "step": 20192
-    },
-    {
-      "epoch": 10.53089960886571,
-      "grad_norm": 1.4520914554595947,
-      "learning_rate": 8.02140703517588e-05,
-      "loss": 5.5713,
-      "step": 20193
-    },
-    {
-      "epoch": 10.531421121251629,
-      "grad_norm": 1.4028310775756836,
-      "learning_rate": 8.021306532663318e-05,
-      "loss": 5.6173,
-      "step": 20194
-    },
-    {
-      "epoch": 10.531942633637549,
-      "grad_norm": 1.38916015625,
-      "learning_rate": 8.021206030150755e-05,
-      "loss": 5.8968,
-      "step": 20195
-    },
-    {
-      "epoch": 10.532464146023468,
-      "grad_norm": 1.4884041547775269,
-      "learning_rate": 8.021105527638191e-05,
-      "loss": 5.2367,
-      "step": 20196
-    },
-    {
-      "epoch": 10.532985658409387,
-      "grad_norm": 1.3871575593948364,
-      "learning_rate": 8.021005025125629e-05,
-      "loss": 5.6366,
-      "step": 20197
-    },
-    {
-      "epoch": 10.533507170795307,
-      "grad_norm": 1.4404693841934204,
-      "learning_rate": 8.020904522613065e-05,
-      "loss": 5.7333,
-      "step": 20198
-    },
-    {
-      "epoch": 10.534028683181226,
-      "grad_norm": 1.3881118297576904,
-      "learning_rate": 8.020804020100503e-05,
-      "loss": 5.3177,
-      "step": 20199
-    },
-    {
-      "epoch": 10.534550195567144,
-      "grad_norm": 1.4080289602279663,
-      "learning_rate": 8.02070351758794e-05,
-      "loss": 5.7817,
-      "step": 20200
-    },
-    {
-      "epoch": 10.535071707953064,
-      "grad_norm": 1.384378433227539,
-      "learning_rate": 8.020603015075377e-05,
-      "loss": 5.1451,
-      "step": 20201
-    },
-    {
-      "epoch": 10.535593220338983,
-      "grad_norm": 1.496822476387024,
-      "learning_rate": 8.020502512562814e-05,
-      "loss": 5.2427,
-      "step": 20202
-    },
-    {
-      "epoch": 10.536114732724903,
-      "grad_norm": 1.2980034351348877,
-      "learning_rate": 8.020402010050251e-05,
-      "loss": 5.7594,
-      "step": 20203
-    },
-    {
-      "epoch": 10.536636245110822,
-      "grad_norm": 1.4821263551712036,
-      "learning_rate": 8.020301507537689e-05,
-      "loss": 5.7245,
-      "step": 20204
-    },
-    {
-      "epoch": 10.53715775749674,
-      "grad_norm": 1.3586602210998535,
-      "learning_rate": 8.020201005025127e-05,
-      "loss": 5.7562,
-      "step": 20205
-    },
-    {
-      "epoch": 10.53767926988266,
-      "grad_norm": 1.527610421180725,
-      "learning_rate": 8.020100502512563e-05,
-      "loss": 4.9659,
-      "step": 20206
-    },
-    {
-      "epoch": 10.538200782268579,
-      "grad_norm": 1.4380759000778198,
-      "learning_rate": 8.020000000000001e-05,
-      "loss": 5.3976,
-      "step": 20207
-    },
-    {
-      "epoch": 10.538722294654498,
-      "grad_norm": 1.419843316078186,
-      "learning_rate": 8.019899497487438e-05,
-      "loss": 5.2719,
-      "step": 20208
-    },
-    {
-      "epoch": 10.539243807040418,
-      "grad_norm": 1.4401570558547974,
-      "learning_rate": 8.019798994974875e-05,
-      "loss": 5.3882,
-      "step": 20209
-    },
-    {
-      "epoch": 10.539765319426337,
-      "grad_norm": 1.50314462184906,
-      "learning_rate": 8.019698492462312e-05,
-      "loss": 5.0976,
-      "step": 20210
-    },
-    {
-      "epoch": 10.540286831812256,
-      "grad_norm": 1.4578981399536133,
-      "learning_rate": 8.019597989949748e-05,
-      "loss": 4.837,
-      "step": 20211
-    },
-    {
-      "epoch": 10.540808344198174,
-      "grad_norm": 1.5523169040679932,
-      "learning_rate": 8.019497487437186e-05,
-      "loss": 5.4135,
-      "step": 20212
-    },
-    {
-      "epoch": 10.541329856584094,
-      "grad_norm": 1.46126389503479,
-      "learning_rate": 8.019396984924624e-05,
-      "loss": 5.6705,
-      "step": 20213
-    },
-    {
-      "epoch": 10.541851368970013,
-      "grad_norm": 1.4168742895126343,
-      "learning_rate": 8.019296482412062e-05,
-      "loss": 5.1659,
-      "step": 20214
-    },
-    {
-      "epoch": 10.542372881355933,
-      "grad_norm": 1.4067457914352417,
-      "learning_rate": 8.019195979899498e-05,
-      "loss": 5.8521,
-      "step": 20215
-    },
-    {
-      "epoch": 10.542894393741852,
-      "grad_norm": 1.427014708518982,
-      "learning_rate": 8.019095477386936e-05,
-      "loss": 5.6971,
-      "step": 20216
-    },
-    {
-      "epoch": 10.54341590612777,
-      "grad_norm": 1.4897271394729614,
-      "learning_rate": 8.018994974874372e-05,
-      "loss": 4.9842,
-      "step": 20217
-    },
-    {
-      "epoch": 10.54393741851369,
-      "grad_norm": 1.357631802558899,
-      "learning_rate": 8.01889447236181e-05,
-      "loss": 5.5134,
-      "step": 20218
-    },
-    {
-      "epoch": 10.544458930899609,
-      "grad_norm": 1.5416765213012695,
-      "learning_rate": 8.018793969849246e-05,
-      "loss": 5.276,
-      "step": 20219
-    },
-    {
-      "epoch": 10.544980443285528,
-      "grad_norm": 1.373114824295044,
-      "learning_rate": 8.018693467336684e-05,
-      "loss": 5.4876,
-      "step": 20220
-    },
-    {
-      "epoch": 10.545501955671448,
-      "grad_norm": 1.3414688110351562,
-      "learning_rate": 8.01859296482412e-05,
-      "loss": 5.418,
-      "step": 20221
-    },
-    {
-      "epoch": 10.546023468057367,
-      "grad_norm": 1.5266329050064087,
-      "learning_rate": 8.018492462311558e-05,
-      "loss": 4.6372,
-      "step": 20222
-    },
-    {
-      "epoch": 10.546544980443285,
-      "grad_norm": 1.537272572517395,
-      "learning_rate": 8.018391959798995e-05,
-      "loss": 5.7233,
-      "step": 20223
-    },
-    {
-      "epoch": 10.547066492829204,
-      "grad_norm": 1.4588062763214111,
-      "learning_rate": 8.018291457286433e-05,
-      "loss": 5.2741,
-      "step": 20224
-    },
-    {
-      "epoch": 10.547588005215124,
-      "grad_norm": 1.384966254234314,
-      "learning_rate": 8.01819095477387e-05,
-      "loss": 5.5459,
-      "step": 20225
-    },
-    {
-      "epoch": 10.548109517601043,
-      "grad_norm": 1.5126590728759766,
-      "learning_rate": 8.018090452261307e-05,
-      "loss": 5.5557,
-      "step": 20226
-    },
-    {
-      "epoch": 10.548631029986963,
-      "grad_norm": 1.4294350147247314,
-      "learning_rate": 8.017989949748745e-05,
-      "loss": 4.9969,
-      "step": 20227
-    },
-    {
-      "epoch": 10.549152542372882,
-      "grad_norm": 1.415366530418396,
-      "learning_rate": 8.017889447236181e-05,
-      "loss": 5.0759,
-      "step": 20228
-    },
-    {
-      "epoch": 10.5496740547588,
-      "grad_norm": 1.3845386505126953,
-      "learning_rate": 8.017788944723619e-05,
-      "loss": 5.6658,
-      "step": 20229
-    },
-    {
-      "epoch": 10.55019556714472,
-      "grad_norm": 1.3841091394424438,
-      "learning_rate": 8.017688442211055e-05,
-      "loss": 5.5155,
-      "step": 20230
-    },
-    {
-      "epoch": 10.550717079530639,
-      "grad_norm": 1.3541297912597656,
-      "learning_rate": 8.017587939698493e-05,
-      "loss": 5.8013,
-      "step": 20231
-    },
-    {
-      "epoch": 10.551238591916558,
-      "grad_norm": 1.588049292564392,
-      "learning_rate": 8.01748743718593e-05,
-      "loss": 5.4246,
-      "step": 20232
-    },
-    {
-      "epoch": 10.551760104302478,
-      "grad_norm": 1.4129724502563477,
-      "learning_rate": 8.017386934673367e-05,
-      "loss": 5.8969,
-      "step": 20233
-    },
-    {
-      "epoch": 10.552281616688397,
-      "grad_norm": 1.5299749374389648,
-      "learning_rate": 8.017286432160804e-05,
-      "loss": 5.6204,
-      "step": 20234
-    },
-    {
-      "epoch": 10.552803129074315,
-      "grad_norm": 1.466318964958191,
-      "learning_rate": 8.017185929648241e-05,
-      "loss": 5.3806,
-      "step": 20235
-    },
-    {
-      "epoch": 10.553324641460234,
-      "grad_norm": 1.5021966695785522,
-      "learning_rate": 8.017085427135679e-05,
-      "loss": 5.4497,
-      "step": 20236
-    },
-    {
-      "epoch": 10.553846153846154,
-      "grad_norm": 1.3643347024917603,
-      "learning_rate": 8.016984924623116e-05,
-      "loss": 5.5697,
-      "step": 20237
-    },
-    {
-      "epoch": 10.554367666232073,
-      "grad_norm": 1.4493151903152466,
-      "learning_rate": 8.016884422110553e-05,
-      "loss": 5.7334,
-      "step": 20238
-    },
-    {
-      "epoch": 10.554889178617993,
-      "grad_norm": 1.393432855606079,
-      "learning_rate": 8.01678391959799e-05,
-      "loss": 5.58,
-      "step": 20239
-    },
-    {
-      "epoch": 10.555410691003912,
-      "grad_norm": 1.3610363006591797,
-      "learning_rate": 8.016683417085428e-05,
-      "loss": 5.4687,
-      "step": 20240
-    },
-    {
-      "epoch": 10.55593220338983,
-      "grad_norm": 1.7768837213516235,
-      "learning_rate": 8.016582914572864e-05,
-      "loss": 4.8861,
-      "step": 20241
-    },
-    {
-      "epoch": 10.55645371577575,
-      "grad_norm": 1.4048658609390259,
-      "learning_rate": 8.016482412060302e-05,
-      "loss": 4.794,
-      "step": 20242
-    },
-    {
-      "epoch": 10.556975228161669,
-      "grad_norm": 1.4334431886672974,
-      "learning_rate": 8.016381909547738e-05,
-      "loss": 5.6109,
-      "step": 20243
-    },
-    {
-      "epoch": 10.557496740547588,
-      "grad_norm": 1.431370496749878,
-      "learning_rate": 8.016281407035176e-05,
-      "loss": 5.8116,
-      "step": 20244
-    },
-    {
-      "epoch": 10.558018252933508,
-      "grad_norm": 1.3407193422317505,
-      "learning_rate": 8.016180904522614e-05,
-      "loss": 5.9563,
-      "step": 20245
-    },
-    {
-      "epoch": 10.558539765319427,
-      "grad_norm": 1.5004607439041138,
-      "learning_rate": 8.016080402010052e-05,
-      "loss": 5.3835,
-      "step": 20246
-    },
-    {
-      "epoch": 10.559061277705345,
-      "grad_norm": 1.4581056833267212,
-      "learning_rate": 8.015979899497488e-05,
-      "loss": 5.542,
-      "step": 20247
-    },
-    {
-      "epoch": 10.559582790091264,
-      "grad_norm": 1.4441161155700684,
-      "learning_rate": 8.015879396984926e-05,
-      "loss": 5.4108,
-      "step": 20248
-    },
-    {
-      "epoch": 10.560104302477184,
-      "grad_norm": 1.4695091247558594,
-      "learning_rate": 8.015778894472362e-05,
-      "loss": 5.6434,
-      "step": 20249
-    },
-    {
-      "epoch": 10.560625814863103,
-      "grad_norm": 1.4314546585083008,
-      "learning_rate": 8.015678391959799e-05,
-      "loss": 5.2793,
-      "step": 20250
-    },
-    {
-      "epoch": 10.561147327249023,
-      "grad_norm": 1.4139429330825806,
-      "learning_rate": 8.015577889447236e-05,
-      "loss": 5.5812,
-      "step": 20251
-    },
-    {
-      "epoch": 10.561668839634942,
-      "grad_norm": 1.3729063272476196,
-      "learning_rate": 8.015477386934673e-05,
-      "loss": 5.4747,
-      "step": 20252
-    },
-    {
-      "epoch": 10.56219035202086,
-      "grad_norm": 1.408679723739624,
-      "learning_rate": 8.015376884422111e-05,
-      "loss": 5.4106,
-      "step": 20253
-    },
-    {
-      "epoch": 10.56271186440678,
-      "grad_norm": 1.473600149154663,
-      "learning_rate": 8.015276381909547e-05,
-      "loss": 5.7142,
-      "step": 20254
-    },
-    {
-      "epoch": 10.563233376792699,
-      "grad_norm": 1.3662124872207642,
-      "learning_rate": 8.015175879396985e-05,
-      "loss": 5.2257,
-      "step": 20255
-    },
-    {
-      "epoch": 10.563754889178618,
-      "grad_norm": 1.4238380193710327,
-      "learning_rate": 8.015075376884423e-05,
-      "loss": 4.999,
-      "step": 20256
-    },
-    {
-      "epoch": 10.564276401564538,
-      "grad_norm": 1.5282313823699951,
-      "learning_rate": 8.01497487437186e-05,
-      "loss": 5.6665,
-      "step": 20257
-    },
-    {
-      "epoch": 10.564797913950457,
-      "grad_norm": 1.3787691593170166,
-      "learning_rate": 8.014874371859297e-05,
-      "loss": 5.5735,
-      "step": 20258
-    },
-    {
-      "epoch": 10.565319426336375,
-      "grad_norm": 1.398950219154358,
-      "learning_rate": 8.014773869346735e-05,
-      "loss": 5.6961,
-      "step": 20259
-    },
-    {
-      "epoch": 10.565840938722294,
-      "grad_norm": 1.3992815017700195,
-      "learning_rate": 8.014673366834171e-05,
-      "loss": 5.2653,
-      "step": 20260
-    },
-    {
-      "epoch": 10.566362451108214,
-      "grad_norm": 1.4563251733779907,
-      "learning_rate": 8.014572864321609e-05,
-      "loss": 5.2659,
-      "step": 20261
-    },
-    {
-      "epoch": 10.566883963494133,
-      "grad_norm": 1.556904673576355,
-      "learning_rate": 8.014472361809045e-05,
-      "loss": 5.3668,
-      "step": 20262
-    },
-    {
-      "epoch": 10.567405475880053,
-      "grad_norm": 1.3301118612289429,
-      "learning_rate": 8.014371859296482e-05,
-      "loss": 5.6001,
-      "step": 20263
-    },
-    {
-      "epoch": 10.567926988265972,
-      "grad_norm": 1.4295158386230469,
-      "learning_rate": 8.01427135678392e-05,
-      "loss": 5.728,
-      "step": 20264
-    },
-    {
-      "epoch": 10.56844850065189,
-      "grad_norm": 1.6140402555465698,
-      "learning_rate": 8.014170854271357e-05,
-      "loss": 5.2846,
-      "step": 20265
-    },
-    {
-      "epoch": 10.56897001303781,
-      "grad_norm": 1.5145275592803955,
-      "learning_rate": 8.014070351758795e-05,
-      "loss": 5.4039,
-      "step": 20266
-    },
-    {
-      "epoch": 10.569491525423729,
-      "grad_norm": 1.6052517890930176,
-      "learning_rate": 8.013969849246232e-05,
-      "loss": 4.8935,
-      "step": 20267
-    },
-    {
-      "epoch": 10.570013037809648,
-      "grad_norm": 1.4529507160186768,
-      "learning_rate": 8.013869346733669e-05,
-      "loss": 5.91,
-      "step": 20268
-    },
-    {
-      "epoch": 10.570534550195568,
-      "grad_norm": 1.5304895639419556,
-      "learning_rate": 8.013768844221106e-05,
-      "loss": 5.4269,
-      "step": 20269
-    },
-    {
-      "epoch": 10.571056062581487,
-      "grad_norm": 1.3281275033950806,
-      "learning_rate": 8.013668341708544e-05,
-      "loss": 5.5004,
-      "step": 20270
-    },
-    {
-      "epoch": 10.571577574967405,
-      "grad_norm": 1.3979135751724243,
-      "learning_rate": 8.01356783919598e-05,
-      "loss": 5.8135,
-      "step": 20271
-    },
-    {
-      "epoch": 10.572099087353324,
-      "grad_norm": 1.4603631496429443,
-      "learning_rate": 8.013467336683418e-05,
-      "loss": 5.5909,
-      "step": 20272
-    },
-    {
-      "epoch": 10.572620599739244,
-      "grad_norm": 1.3499153852462769,
-      "learning_rate": 8.013366834170854e-05,
-      "loss": 5.9828,
-      "step": 20273
-    },
-    {
-      "epoch": 10.573142112125163,
-      "grad_norm": 1.381571650505066,
-      "learning_rate": 8.013266331658292e-05,
-      "loss": 5.6914,
-      "step": 20274
-    },
-    {
-      "epoch": 10.573663624511083,
-      "grad_norm": 1.3627787828445435,
-      "learning_rate": 8.013165829145728e-05,
-      "loss": 5.8989,
-      "step": 20275
-    },
-    {
-      "epoch": 10.574185136897002,
-      "grad_norm": 1.478264331817627,
-      "learning_rate": 8.013065326633166e-05,
-      "loss": 5.0448,
-      "step": 20276
-    },
-    {
-      "epoch": 10.57470664928292,
-      "grad_norm": 1.4360945224761963,
-      "learning_rate": 8.012964824120604e-05,
-      "loss": 5.2642,
-      "step": 20277
-    },
-    {
-      "epoch": 10.57522816166884,
-      "grad_norm": 1.4869340658187866,
-      "learning_rate": 8.01286432160804e-05,
-      "loss": 5.664,
-      "step": 20278
-    },
-    {
-      "epoch": 10.575749674054759,
-      "grad_norm": 1.4918346405029297,
-      "learning_rate": 8.012763819095478e-05,
-      "loss": 5.3719,
-      "step": 20279
-    },
-    {
-      "epoch": 10.576271186440678,
-      "grad_norm": 1.4212660789489746,
-      "learning_rate": 8.012663316582915e-05,
-      "loss": 5.7056,
-      "step": 20280
-    },
-    {
-      "epoch": 10.576792698826598,
-      "grad_norm": 1.4206234216690063,
-      "learning_rate": 8.012562814070352e-05,
-      "loss": 5.8042,
-      "step": 20281
-    },
-    {
-      "epoch": 10.577314211212517,
-      "grad_norm": 1.4932959079742432,
-      "learning_rate": 8.012462311557789e-05,
-      "loss": 5.1194,
-      "step": 20282
-    },
-    {
-      "epoch": 10.577835723598435,
-      "grad_norm": 1.5154012441635132,
-      "learning_rate": 8.012361809045227e-05,
-      "loss": 5.0379,
-      "step": 20283
-    },
-    {
-      "epoch": 10.578357235984354,
-      "grad_norm": 1.3067644834518433,
-      "learning_rate": 8.012261306532663e-05,
-      "loss": 6.043,
-      "step": 20284
-    },
-    {
-      "epoch": 10.578878748370274,
-      "grad_norm": 1.515801191329956,
-      "learning_rate": 8.012160804020101e-05,
-      "loss": 5.4639,
-      "step": 20285
-    },
-    {
-      "epoch": 10.579400260756193,
-      "grad_norm": 1.4406015872955322,
-      "learning_rate": 8.012060301507539e-05,
-      "loss": 5.692,
-      "step": 20286
-    },
-    {
-      "epoch": 10.579921773142113,
-      "grad_norm": 1.473102331161499,
-      "learning_rate": 8.011959798994976e-05,
-      "loss": 5.1343,
-      "step": 20287
-    },
-    {
-      "epoch": 10.58044328552803,
-      "grad_norm": 1.525311827659607,
-      "learning_rate": 8.011859296482413e-05,
-      "loss": 5.5333,
-      "step": 20288
-    },
-    {
-      "epoch": 10.58096479791395,
-      "grad_norm": 1.479345679283142,
-      "learning_rate": 8.01175879396985e-05,
-      "loss": 5.6151,
-      "step": 20289
-    },
-    {
-      "epoch": 10.58148631029987,
-      "grad_norm": 1.300231695175171,
-      "learning_rate": 8.011658291457287e-05,
-      "loss": 5.9811,
-      "step": 20290
-    },
-    {
-      "epoch": 10.582007822685789,
-      "grad_norm": 1.4784913063049316,
-      "learning_rate": 8.011557788944723e-05,
-      "loss": 5.6193,
-      "step": 20291
-    },
-    {
-      "epoch": 10.582529335071708,
-      "grad_norm": 1.5426132678985596,
-      "learning_rate": 8.011457286432161e-05,
-      "loss": 4.7521,
-      "step": 20292
-    },
-    {
-      "epoch": 10.583050847457628,
-      "grad_norm": 1.479264497756958,
-      "learning_rate": 8.011356783919598e-05,
-      "loss": 5.321,
-      "step": 20293
-    },
-    {
-      "epoch": 10.583572359843547,
-      "grad_norm": 1.3865805864334106,
-      "learning_rate": 8.011256281407035e-05,
-      "loss": 5.8094,
-      "step": 20294
-    },
-    {
-      "epoch": 10.584093872229465,
-      "grad_norm": 1.3486926555633545,
-      "learning_rate": 8.011155778894472e-05,
-      "loss": 5.4826,
-      "step": 20295
-    },
-    {
-      "epoch": 10.584615384615384,
-      "grad_norm": 1.3089896440505981,
-      "learning_rate": 8.01105527638191e-05,
-      "loss": 5.4196,
-      "step": 20296
-    },
-    {
-      "epoch": 10.585136897001304,
-      "grad_norm": 1.4309192895889282,
-      "learning_rate": 8.010954773869347e-05,
-      "loss": 5.2769,
-      "step": 20297
-    },
-    {
-      "epoch": 10.585658409387223,
-      "grad_norm": 1.4082967042922974,
-      "learning_rate": 8.010854271356785e-05,
-      "loss": 5.745,
-      "step": 20298
-    },
-    {
-      "epoch": 10.586179921773143,
-      "grad_norm": 1.472242832183838,
-      "learning_rate": 8.010753768844222e-05,
-      "loss": 5.7669,
-      "step": 20299
-    },
-    {
-      "epoch": 10.58670143415906,
-      "grad_norm": 1.7023063898086548,
-      "learning_rate": 8.01065326633166e-05,
-      "loss": 5.1477,
-      "step": 20300
-    },
-    {
-      "epoch": 10.58722294654498,
-      "grad_norm": 1.4095755815505981,
-      "learning_rate": 8.010552763819096e-05,
-      "loss": 5.7885,
-      "step": 20301
-    },
-    {
-      "epoch": 10.5877444589309,
-      "grad_norm": 1.4386128187179565,
-      "learning_rate": 8.010452261306534e-05,
-      "loss": 5.8386,
-      "step": 20302
-    },
-    {
-      "epoch": 10.588265971316819,
-      "grad_norm": 1.592929720878601,
-      "learning_rate": 8.01035175879397e-05,
-      "loss": 5.5571,
-      "step": 20303
-    },
-    {
-      "epoch": 10.588787483702738,
-      "grad_norm": 1.345913290977478,
-      "learning_rate": 8.010251256281406e-05,
-      "loss": 5.7579,
-      "step": 20304
-    },
-    {
-      "epoch": 10.589308996088658,
-      "grad_norm": 1.4026896953582764,
-      "learning_rate": 8.010150753768844e-05,
-      "loss": 5.6908,
-      "step": 20305
-    },
-    {
-      "epoch": 10.589830508474577,
-      "grad_norm": 1.8465924263000488,
-      "learning_rate": 8.010050251256282e-05,
-      "loss": 5.4552,
-      "step": 20306
-    },
-    {
-      "epoch": 10.590352020860495,
-      "grad_norm": 1.515138864517212,
-      "learning_rate": 8.00994974874372e-05,
-      "loss": 5.738,
-      "step": 20307
-    },
-    {
-      "epoch": 10.590873533246414,
-      "grad_norm": 1.5409685373306274,
-      "learning_rate": 8.009849246231156e-05,
-      "loss": 5.4639,
-      "step": 20308
-    },
-    {
-      "epoch": 10.591395045632334,
-      "grad_norm": 1.3915365934371948,
-      "learning_rate": 8.009748743718594e-05,
-      "loss": 5.2295,
-      "step": 20309
-    },
-    {
-      "epoch": 10.591916558018253,
-      "grad_norm": 1.4868148565292358,
-      "learning_rate": 8.00964824120603e-05,
-      "loss": 6.0164,
-      "step": 20310
-    },
-    {
-      "epoch": 10.592438070404173,
-      "grad_norm": 1.3760885000228882,
-      "learning_rate": 8.009547738693468e-05,
-      "loss": 5.8632,
-      "step": 20311
-    },
-    {
-      "epoch": 10.59295958279009,
-      "grad_norm": 1.7551581859588623,
-      "learning_rate": 8.009447236180905e-05,
-      "loss": 5.1178,
-      "step": 20312
-    },
-    {
-      "epoch": 10.59348109517601,
-      "grad_norm": 1.4474648237228394,
-      "learning_rate": 8.009346733668342e-05,
-      "loss": 5.7155,
-      "step": 20313
-    },
-    {
-      "epoch": 10.59400260756193,
-      "grad_norm": 1.425426721572876,
-      "learning_rate": 8.009246231155779e-05,
-      "loss": 5.5291,
-      "step": 20314
-    },
-    {
-      "epoch": 10.594524119947849,
-      "grad_norm": 1.6412614583969116,
-      "learning_rate": 8.009145728643217e-05,
-      "loss": 5.6707,
-      "step": 20315
-    },
-    {
-      "epoch": 10.595045632333768,
-      "grad_norm": 1.635238528251648,
-      "learning_rate": 8.009045226130653e-05,
-      "loss": 5.2968,
-      "step": 20316
-    },
-    {
-      "epoch": 10.595567144719688,
-      "grad_norm": 1.551027774810791,
-      "learning_rate": 8.008944723618091e-05,
-      "loss": 5.313,
-      "step": 20317
-    },
-    {
-      "epoch": 10.596088657105605,
-      "grad_norm": 1.6399531364440918,
-      "learning_rate": 8.008844221105529e-05,
-      "loss": 5.6269,
-      "step": 20318
-    },
-    {
-      "epoch": 10.596610169491525,
-      "grad_norm": 1.5596975088119507,
-      "learning_rate": 8.008743718592965e-05,
-      "loss": 5.4558,
-      "step": 20319
-    },
-    {
-      "epoch": 10.597131681877444,
-      "grad_norm": 1.5884283781051636,
-      "learning_rate": 8.008643216080403e-05,
-      "loss": 5.5783,
-      "step": 20320
-    },
-    {
-      "epoch": 10.597653194263364,
-      "grad_norm": 1.512988567352295,
-      "learning_rate": 8.008542713567839e-05,
-      "loss": 5.7011,
-      "step": 20321
-    },
-    {
-      "epoch": 10.598174706649283,
-      "grad_norm": 1.5800082683563232,
-      "learning_rate": 8.008442211055277e-05,
-      "loss": 4.8432,
-      "step": 20322
-    },
-    {
-      "epoch": 10.598696219035203,
-      "grad_norm": 1.593319058418274,
-      "learning_rate": 8.008341708542713e-05,
-      "loss": 5.4296,
-      "step": 20323
-    },
-    {
-      "epoch": 10.59921773142112,
-      "grad_norm": 1.4557939767837524,
-      "learning_rate": 8.008241206030151e-05,
-      "loss": 5.1171,
-      "step": 20324
-    },
-    {
-      "epoch": 10.59973924380704,
-      "grad_norm": 1.6261017322540283,
-      "learning_rate": 8.008140703517588e-05,
-      "loss": 5.6028,
-      "step": 20325
-    },
-    {
-      "epoch": 10.60026075619296,
-      "grad_norm": 1.5650697946548462,
-      "learning_rate": 8.008040201005025e-05,
-      "loss": 5.4331,
-      "step": 20326
-    },
-    {
-      "epoch": 10.600782268578879,
-      "grad_norm": 1.5699433088302612,
-      "learning_rate": 8.007939698492463e-05,
-      "loss": 5.6079,
-      "step": 20327
-    },
-    {
-      "epoch": 10.601303780964798,
-      "grad_norm": 1.4536148309707642,
-      "learning_rate": 8.007839195979901e-05,
-      "loss": 5.7236,
-      "step": 20328
-    },
-    {
-      "epoch": 10.601825293350718,
-      "grad_norm": 1.4225445985794067,
-      "learning_rate": 8.007738693467337e-05,
-      "loss": 5.5124,
-      "step": 20329
-    },
-    {
-      "epoch": 10.602346805736635,
-      "grad_norm": 1.5412858724594116,
-      "learning_rate": 8.007638190954774e-05,
-      "loss": 5.7813,
-      "step": 20330
-    },
-    {
-      "epoch": 10.602868318122555,
-      "grad_norm": 1.4570627212524414,
-      "learning_rate": 8.007537688442212e-05,
-      "loss": 5.2537,
-      "step": 20331
-    },
-    {
-      "epoch": 10.603389830508474,
-      "grad_norm": 1.4667084217071533,
-      "learning_rate": 8.007437185929648e-05,
-      "loss": 5.2296,
-      "step": 20332
-    },
-    {
-      "epoch": 10.603911342894394,
-      "grad_norm": 1.458038330078125,
-      "learning_rate": 8.007336683417086e-05,
-      "loss": 5.7835,
-      "step": 20333
-    },
-    {
-      "epoch": 10.604432855280313,
-      "grad_norm": 1.5938169956207275,
-      "learning_rate": 8.007236180904522e-05,
-      "loss": 4.8911,
-      "step": 20334
-    },
-    {
-      "epoch": 10.604954367666233,
-      "grad_norm": 1.4067832231521606,
-      "learning_rate": 8.00713567839196e-05,
-      "loss": 5.0328,
-      "step": 20335
-    },
-    {
-      "epoch": 10.60547588005215,
-      "grad_norm": 1.3974424600601196,
-      "learning_rate": 8.007035175879397e-05,
-      "loss": 5.7257,
-      "step": 20336
-    },
-    {
-      "epoch": 10.60599739243807,
-      "grad_norm": 1.4746146202087402,
-      "learning_rate": 8.006934673366834e-05,
-      "loss": 5.3342,
-      "step": 20337
-    },
-    {
-      "epoch": 10.60651890482399,
-      "grad_norm": 1.385873556137085,
-      "learning_rate": 8.006834170854272e-05,
-      "loss": 5.7591,
-      "step": 20338
-    },
-    {
-      "epoch": 10.607040417209909,
-      "grad_norm": 1.4458677768707275,
-      "learning_rate": 8.00673366834171e-05,
-      "loss": 4.9866,
-      "step": 20339
-    },
-    {
-      "epoch": 10.607561929595828,
-      "grad_norm": 1.53517746925354,
-      "learning_rate": 8.006633165829146e-05,
-      "loss": 5.3816,
-      "step": 20340
-    },
-    {
-      "epoch": 10.608083441981748,
-      "grad_norm": 1.3803837299346924,
-      "learning_rate": 8.006532663316584e-05,
-      "loss": 5.6776,
-      "step": 20341
-    },
-    {
-      "epoch": 10.608604954367665,
-      "grad_norm": 1.5671350955963135,
-      "learning_rate": 8.00643216080402e-05,
-      "loss": 5.5076,
-      "step": 20342
-    },
-    {
-      "epoch": 10.609126466753585,
-      "grad_norm": 1.5558252334594727,
-      "learning_rate": 8.006331658291457e-05,
-      "loss": 5.3208,
-      "step": 20343
-    },
-    {
-      "epoch": 10.609647979139504,
-      "grad_norm": 1.434950828552246,
-      "learning_rate": 8.006231155778895e-05,
-      "loss": 5.0781,
-      "step": 20344
-    },
-    {
-      "epoch": 10.610169491525424,
-      "grad_norm": 1.372941255569458,
-      "learning_rate": 8.006130653266331e-05,
-      "loss": 5.3012,
-      "step": 20345
-    },
-    {
-      "epoch": 10.610691003911343,
-      "grad_norm": 1.4889776706695557,
-      "learning_rate": 8.006030150753769e-05,
-      "loss": 5.4904,
-      "step": 20346
-    },
-    {
-      "epoch": 10.611212516297263,
-      "grad_norm": 1.5329807996749878,
-      "learning_rate": 8.005929648241207e-05,
-      "loss": 5.6627,
-      "step": 20347
-    },
-    {
-      "epoch": 10.61173402868318,
-      "grad_norm": 1.4532989263534546,
-      "learning_rate": 8.005829145728644e-05,
-      "loss": 5.5543,
-      "step": 20348
-    },
-    {
-      "epoch": 10.6122555410691,
-      "grad_norm": 1.5099384784698486,
-      "learning_rate": 8.005728643216081e-05,
-      "loss": 5.351,
-      "step": 20349
-    },
-    {
-      "epoch": 10.61277705345502,
-      "grad_norm": 1.4249534606933594,
-      "learning_rate": 8.005628140703519e-05,
-      "loss": 5.9103,
-      "step": 20350
-    },
-    {
-      "epoch": 10.613298565840939,
-      "grad_norm": 1.4880112409591675,
-      "learning_rate": 8.005527638190955e-05,
-      "loss": 5.5792,
-      "step": 20351
-    },
-    {
-      "epoch": 10.613820078226858,
-      "grad_norm": 1.420192003250122,
-      "learning_rate": 8.005427135678393e-05,
-      "loss": 5.6302,
-      "step": 20352
-    },
-    {
-      "epoch": 10.614341590612778,
-      "grad_norm": 1.6055805683135986,
-      "learning_rate": 8.00532663316583e-05,
-      "loss": 4.8824,
-      "step": 20353
-    },
-    {
-      "epoch": 10.614863102998696,
-      "grad_norm": 1.4716219902038574,
-      "learning_rate": 8.005226130653267e-05,
-      "loss": 5.4559,
-      "step": 20354
-    },
-    {
-      "epoch": 10.615384615384615,
-      "grad_norm": 1.471611738204956,
-      "learning_rate": 8.005125628140704e-05,
-      "loss": 5.7122,
-      "step": 20355
-    },
-    {
-      "epoch": 10.615906127770534,
-      "grad_norm": 1.3853797912597656,
-      "learning_rate": 8.00502512562814e-05,
-      "loss": 5.6336,
-      "step": 20356
-    },
-    {
-      "epoch": 10.616427640156454,
-      "grad_norm": 1.4593647718429565,
-      "learning_rate": 8.004924623115578e-05,
-      "loss": 4.9666,
-      "step": 20357
-    },
-    {
-      "epoch": 10.616949152542373,
-      "grad_norm": 1.5172309875488281,
-      "learning_rate": 8.004824120603016e-05,
-      "loss": 5.2017,
-      "step": 20358
-    },
-    {
-      "epoch": 10.617470664928293,
-      "grad_norm": 1.574142575263977,
-      "learning_rate": 8.004723618090453e-05,
-      "loss": 5.484,
-      "step": 20359
-    },
-    {
-      "epoch": 10.61799217731421,
-      "grad_norm": 1.3219945430755615,
-      "learning_rate": 8.00462311557789e-05,
-      "loss": 5.8777,
-      "step": 20360
-    },
-    {
-      "epoch": 10.61851368970013,
-      "grad_norm": 1.3347351551055908,
-      "learning_rate": 8.004522613065328e-05,
-      "loss": 5.9346,
-      "step": 20361
-    },
-    {
-      "epoch": 10.61903520208605,
-      "grad_norm": 1.3545855283737183,
-      "learning_rate": 8.004422110552764e-05,
-      "loss": 5.5842,
-      "step": 20362
-    },
-    {
-      "epoch": 10.619556714471969,
-      "grad_norm": 1.546891689300537,
-      "learning_rate": 8.004321608040202e-05,
-      "loss": 5.0978,
-      "step": 20363
-    },
-    {
-      "epoch": 10.620078226857888,
-      "grad_norm": 1.5986939668655396,
-      "learning_rate": 8.004221105527638e-05,
-      "loss": 5.3689,
-      "step": 20364
-    },
-    {
-      "epoch": 10.620599739243808,
-      "grad_norm": 1.3462872505187988,
-      "learning_rate": 8.004120603015076e-05,
-      "loss": 5.5303,
-      "step": 20365
-    },
-    {
-      "epoch": 10.621121251629726,
-      "grad_norm": 1.6108083724975586,
-      "learning_rate": 8.004020100502512e-05,
-      "loss": 5.0074,
-      "step": 20366
-    },
-    {
-      "epoch": 10.621642764015645,
-      "grad_norm": 1.5181044340133667,
-      "learning_rate": 8.00391959798995e-05,
-      "loss": 5.3537,
-      "step": 20367
-    },
-    {
-      "epoch": 10.622164276401564,
-      "grad_norm": 1.4695930480957031,
-      "learning_rate": 8.003819095477388e-05,
-      "loss": 5.2942,
-      "step": 20368
-    },
-    {
-      "epoch": 10.622685788787484,
-      "grad_norm": 1.4911702871322632,
-      "learning_rate": 8.003718592964824e-05,
-      "loss": 4.6752,
-      "step": 20369
-    },
-    {
-      "epoch": 10.623207301173403,
-      "grad_norm": 1.3689013719558716,
-      "learning_rate": 8.003618090452262e-05,
-      "loss": 5.7836,
-      "step": 20370
-    },
-    {
-      "epoch": 10.623728813559323,
-      "grad_norm": 1.504179835319519,
-      "learning_rate": 8.003517587939699e-05,
-      "loss": 5.5564,
-      "step": 20371
-    },
-    {
-      "epoch": 10.62425032594524,
-      "grad_norm": 1.4184364080429077,
-      "learning_rate": 8.003417085427136e-05,
-      "loss": 5.202,
-      "step": 20372
-    },
-    {
-      "epoch": 10.62477183833116,
-      "grad_norm": 1.446561336517334,
-      "learning_rate": 8.003316582914573e-05,
-      "loss": 5.2597,
-      "step": 20373
-    },
-    {
-      "epoch": 10.62529335071708,
-      "grad_norm": 1.3740795850753784,
-      "learning_rate": 8.00321608040201e-05,
-      "loss": 5.6263,
-      "step": 20374
-    },
-    {
-      "epoch": 10.625814863102999,
-      "grad_norm": 1.290969967842102,
-      "learning_rate": 8.003115577889447e-05,
-      "loss": 5.7536,
-      "step": 20375
-    },
-    {
-      "epoch": 10.626336375488918,
-      "grad_norm": 1.483827829360962,
-      "learning_rate": 8.003015075376885e-05,
-      "loss": 5.464,
-      "step": 20376
-    },
-    {
-      "epoch": 10.626857887874838,
-      "grad_norm": 1.381015658378601,
-      "learning_rate": 8.002914572864321e-05,
-      "loss": 5.4788,
-      "step": 20377
-    },
-    {
-      "epoch": 10.627379400260756,
-      "grad_norm": 1.4579358100891113,
-      "learning_rate": 8.002814070351759e-05,
-      "loss": 5.3345,
-      "step": 20378
-    },
-    {
-      "epoch": 10.627900912646675,
-      "grad_norm": 1.4005515575408936,
-      "learning_rate": 8.002713567839197e-05,
-      "loss": 5.6301,
-      "step": 20379
-    },
-    {
-      "epoch": 10.628422425032594,
-      "grad_norm": 1.446740984916687,
-      "learning_rate": 8.002613065326635e-05,
-      "loss": 5.5977,
-      "step": 20380
-    },
-    {
-      "epoch": 10.628943937418514,
-      "grad_norm": 1.4647579193115234,
-      "learning_rate": 8.002512562814071e-05,
-      "loss": 4.9954,
-      "step": 20381
-    },
-    {
-      "epoch": 10.629465449804433,
-      "grad_norm": 1.5701813697814941,
-      "learning_rate": 8.002412060301509e-05,
-      "loss": 5.4452,
-      "step": 20382
-    },
-    {
-      "epoch": 10.629986962190351,
-      "grad_norm": 1.4504609107971191,
-      "learning_rate": 8.002311557788945e-05,
-      "loss": 5.7179,
-      "step": 20383
-    },
-    {
-      "epoch": 10.63050847457627,
-      "grad_norm": 1.4959039688110352,
-      "learning_rate": 8.002211055276382e-05,
-      "loss": 5.6526,
-      "step": 20384
-    },
-    {
-      "epoch": 10.63102998696219,
-      "grad_norm": 1.4168250560760498,
-      "learning_rate": 8.00211055276382e-05,
-      "loss": 5.249,
-      "step": 20385
-    },
-    {
-      "epoch": 10.63155149934811,
-      "grad_norm": 1.5743677616119385,
-      "learning_rate": 8.002010050251256e-05,
-      "loss": 5.1759,
-      "step": 20386
-    },
-    {
-      "epoch": 10.632073011734029,
-      "grad_norm": 1.5216364860534668,
-      "learning_rate": 8.001909547738694e-05,
-      "loss": 5.5421,
-      "step": 20387
-    },
-    {
-      "epoch": 10.632594524119948,
-      "grad_norm": 1.4515125751495361,
-      "learning_rate": 8.001809045226131e-05,
-      "loss": 5.226,
-      "step": 20388
-    },
-    {
-      "epoch": 10.633116036505868,
-      "grad_norm": 1.5928610563278198,
-      "learning_rate": 8.001708542713569e-05,
-      "loss": 5.4009,
-      "step": 20389
-    },
-    {
-      "epoch": 10.633637548891786,
-      "grad_norm": 1.4723674058914185,
-      "learning_rate": 8.001608040201006e-05,
-      "loss": 5.5598,
-      "step": 20390
-    },
-    {
-      "epoch": 10.634159061277705,
-      "grad_norm": 1.547851800918579,
-      "learning_rate": 8.001507537688443e-05,
-      "loss": 4.8628,
-      "step": 20391
-    },
-    {
-      "epoch": 10.634680573663625,
-      "grad_norm": 1.625963807106018,
-      "learning_rate": 8.00140703517588e-05,
-      "loss": 5.4782,
-      "step": 20392
-    },
-    {
-      "epoch": 10.635202086049544,
-      "grad_norm": 1.6613879203796387,
-      "learning_rate": 8.001306532663318e-05,
-      "loss": 4.6291,
-      "step": 20393
-    },
-    {
-      "epoch": 10.635723598435463,
-      "grad_norm": 1.5430057048797607,
-      "learning_rate": 8.001206030150754e-05,
-      "loss": 5.7443,
-      "step": 20394
-    },
-    {
-      "epoch": 10.636245110821381,
-      "grad_norm": 1.5002100467681885,
-      "learning_rate": 8.001105527638192e-05,
-      "loss": 5.4551,
-      "step": 20395
-    },
-    {
-      "epoch": 10.6367666232073,
-      "grad_norm": 1.382204294204712,
-      "learning_rate": 8.001005025125628e-05,
-      "loss": 5.7206,
-      "step": 20396
-    },
-    {
-      "epoch": 10.63728813559322,
-      "grad_norm": 1.436213731765747,
-      "learning_rate": 8.000904522613065e-05,
-      "loss": 5.5281,
-      "step": 20397
-    },
-    {
-      "epoch": 10.63780964797914,
-      "grad_norm": 1.3904532194137573,
-      "learning_rate": 8.000804020100502e-05,
-      "loss": 5.7955,
-      "step": 20398
-    },
-    {
-      "epoch": 10.638331160365059,
-      "grad_norm": 1.4920960664749146,
-      "learning_rate": 8.00070351758794e-05,
-      "loss": 5.3288,
-      "step": 20399
-    },
-    {
-      "epoch": 10.638852672750978,
-      "grad_norm": 1.4184107780456543,
-      "learning_rate": 8.000603015075378e-05,
-      "loss": 5.0959,
-      "step": 20400
-    },
-    {
-      "epoch": 10.639374185136898,
-      "grad_norm": 1.3702064752578735,
-      "learning_rate": 8.000502512562814e-05,
-      "loss": 5.4189,
-      "step": 20401
-    },
-    {
-      "epoch": 10.639895697522816,
-      "grad_norm": 1.3656468391418457,
-      "learning_rate": 8.000402010050252e-05,
-      "loss": 5.8103,
-      "step": 20402
-    },
-    {
-      "epoch": 10.640417209908735,
-      "grad_norm": 1.3742328882217407,
-      "learning_rate": 8.000301507537689e-05,
-      "loss": 5.5756,
-      "step": 20403
-    },
-    {
-      "epoch": 10.640938722294655,
-      "grad_norm": 1.394443154335022,
-      "learning_rate": 8.000201005025126e-05,
-      "loss": 5.6556,
-      "step": 20404
-    },
-    {
-      "epoch": 10.641460234680574,
-      "grad_norm": 1.369411826133728,
-      "learning_rate": 8.000100502512563e-05,
-      "loss": 5.3979,
-      "step": 20405
-    },
-    {
-      "epoch": 10.641981747066493,
-      "grad_norm": 1.434195876121521,
-      "learning_rate": 8e-05,
-      "loss": 5.5978,
-      "step": 20406
-    },
-    {
-      "epoch": 10.642503259452411,
-      "grad_norm": 1.2996437549591064,
-      "learning_rate": 7.999899497487437e-05,
-      "loss": 5.6086,
-      "step": 20407
-    },
-    {
-      "epoch": 10.64302477183833,
-      "grad_norm": 1.6014324426651,
-      "learning_rate": 7.999798994974875e-05,
-      "loss": 4.9028,
-      "step": 20408
-    },
-    {
-      "epoch": 10.64354628422425,
-      "grad_norm": 1.6334145069122314,
-      "learning_rate": 7.999698492462311e-05,
-      "loss": 5.1804,
-      "step": 20409
-    },
-    {
-      "epoch": 10.64406779661017,
-      "grad_norm": 1.5978317260742188,
-      "learning_rate": 7.999597989949749e-05,
-      "loss": 5.5301,
-      "step": 20410
-    },
-    {
-      "epoch": 10.644589308996089,
-      "grad_norm": 1.4751754999160767,
-      "learning_rate": 7.999497487437187e-05,
-      "loss": 5.3707,
-      "step": 20411
-    },
-    {
-      "epoch": 10.645110821382008,
-      "grad_norm": 1.4821021556854248,
-      "learning_rate": 7.999396984924623e-05,
-      "loss": 5.1339,
-      "step": 20412
-    },
-    {
-      "epoch": 10.645632333767926,
-      "grad_norm": 1.3982552289962769,
-      "learning_rate": 7.999296482412061e-05,
-      "loss": 5.4253,
-      "step": 20413
-    },
-    {
-      "epoch": 10.646153846153846,
-      "grad_norm": 1.4856362342834473,
-      "learning_rate": 7.999195979899497e-05,
-      "loss": 5.1998,
-      "step": 20414
-    },
-    {
-      "epoch": 10.646675358539765,
-      "grad_norm": 1.4273440837860107,
-      "learning_rate": 7.999095477386935e-05,
-      "loss": 5.3857,
-      "step": 20415
-    },
-    {
-      "epoch": 10.647196870925685,
-      "grad_norm": 1.4154889583587646,
-      "learning_rate": 7.998994974874372e-05,
-      "loss": 5.5954,
-      "step": 20416
-    },
-    {
-      "epoch": 10.647718383311604,
-      "grad_norm": 1.5018383264541626,
-      "learning_rate": 7.99889447236181e-05,
-      "loss": 5.383,
-      "step": 20417
-    },
-    {
-      "epoch": 10.648239895697523,
-      "grad_norm": 1.3882123231887817,
-      "learning_rate": 7.998793969849246e-05,
-      "loss": 5.8197,
-      "step": 20418
-    },
-    {
-      "epoch": 10.648761408083441,
-      "grad_norm": 1.5093927383422852,
-      "learning_rate": 7.998693467336684e-05,
-      "loss": 5.3145,
-      "step": 20419
-    },
-    {
-      "epoch": 10.64928292046936,
-      "grad_norm": 1.5344239473342896,
-      "learning_rate": 7.998592964824121e-05,
-      "loss": 5.7671,
-      "step": 20420
-    },
-    {
-      "epoch": 10.64980443285528,
-      "grad_norm": 1.427409052848816,
-      "learning_rate": 7.998492462311559e-05,
-      "loss": 5.3355,
-      "step": 20421
-    },
-    {
-      "epoch": 10.6503259452412,
-      "grad_norm": 1.4106247425079346,
-      "learning_rate": 7.998391959798996e-05,
-      "loss": 5.0892,
-      "step": 20422
-    },
-    {
-      "epoch": 10.650847457627119,
-      "grad_norm": 1.4413973093032837,
-      "learning_rate": 7.998291457286432e-05,
-      "loss": 5.7152,
-      "step": 20423
-    },
-    {
-      "epoch": 10.651368970013039,
-      "grad_norm": 1.3835229873657227,
-      "learning_rate": 7.99819095477387e-05,
-      "loss": 5.616,
-      "step": 20424
-    },
-    {
-      "epoch": 10.651890482398956,
-      "grad_norm": 1.3275269269943237,
-      "learning_rate": 7.998090452261306e-05,
-      "loss": 5.3928,
-      "step": 20425
-    },
-    {
-      "epoch": 10.652411994784876,
-      "grad_norm": 1.3810938596725464,
-      "learning_rate": 7.997989949748744e-05,
-      "loss": 5.4861,
-      "step": 20426
-    },
-    {
-      "epoch": 10.652933507170795,
-      "grad_norm": 1.594072937965393,
-      "learning_rate": 7.99788944723618e-05,
-      "loss": 5.1142,
-      "step": 20427
-    },
-    {
-      "epoch": 10.653455019556715,
-      "grad_norm": 1.4089927673339844,
-      "learning_rate": 7.997788944723618e-05,
-      "loss": 5.2154,
-      "step": 20428
-    },
-    {
-      "epoch": 10.653976531942634,
-      "grad_norm": 1.4329657554626465,
-      "learning_rate": 7.997688442211055e-05,
-      "loss": 5.8838,
-      "step": 20429
-    },
-    {
-      "epoch": 10.654498044328554,
-      "grad_norm": 1.4430019855499268,
-      "learning_rate": 7.997587939698493e-05,
-      "loss": 5.6904,
-      "step": 20430
-    },
-    {
-      "epoch": 10.655019556714471,
-      "grad_norm": 1.3546912670135498,
-      "learning_rate": 7.99748743718593e-05,
-      "loss": 5.6376,
-      "step": 20431
-    },
-    {
-      "epoch": 10.65554106910039,
-      "grad_norm": 1.4717532396316528,
-      "learning_rate": 7.997386934673368e-05,
-      "loss": 4.9669,
-      "step": 20432
-    },
-    {
-      "epoch": 10.65606258148631,
-      "grad_norm": 1.44791579246521,
-      "learning_rate": 7.997286432160805e-05,
-      "loss": 5.7617,
-      "step": 20433
-    },
-    {
-      "epoch": 10.65658409387223,
-      "grad_norm": 1.539704442024231,
-      "learning_rate": 7.997185929648242e-05,
-      "loss": 5.4685,
-      "step": 20434
-    },
-    {
-      "epoch": 10.657105606258149,
-      "grad_norm": 1.4861712455749512,
-      "learning_rate": 7.997085427135679e-05,
-      "loss": 5.6139,
-      "step": 20435
-    },
-    {
-      "epoch": 10.657627118644069,
-      "grad_norm": 1.431490421295166,
-      "learning_rate": 7.996984924623115e-05,
-      "loss": 5.5992,
-      "step": 20436
-    },
-    {
-      "epoch": 10.658148631029986,
-      "grad_norm": 1.5139541625976562,
-      "learning_rate": 7.996884422110553e-05,
-      "loss": 5.1198,
-      "step": 20437
-    },
-    {
-      "epoch": 10.658670143415906,
-      "grad_norm": 1.5528419017791748,
-      "learning_rate": 7.99678391959799e-05,
-      "loss": 5.2089,
-      "step": 20438
-    },
-    {
-      "epoch": 10.659191655801825,
-      "grad_norm": 1.4837076663970947,
-      "learning_rate": 7.996683417085427e-05,
-      "loss": 5.4449,
-      "step": 20439
-    },
-    {
-      "epoch": 10.659713168187745,
-      "grad_norm": 1.456648588180542,
-      "learning_rate": 7.996582914572865e-05,
-      "loss": 5.4871,
-      "step": 20440
-    },
-    {
-      "epoch": 10.660234680573664,
-      "grad_norm": 1.4770307540893555,
-      "learning_rate": 7.996482412060303e-05,
-      "loss": 4.8401,
-      "step": 20441
-    },
-    {
-      "epoch": 10.660756192959584,
-      "grad_norm": 1.6018253564834595,
-      "learning_rate": 7.996381909547739e-05,
-      "loss": 5.8436,
-      "step": 20442
-    },
-    {
-      "epoch": 10.661277705345501,
-      "grad_norm": 1.4170273542404175,
-      "learning_rate": 7.996281407035177e-05,
-      "loss": 5.5579,
-      "step": 20443
-    },
-    {
-      "epoch": 10.66179921773142,
-      "grad_norm": 1.4857269525527954,
-      "learning_rate": 7.996180904522613e-05,
-      "loss": 5.0157,
-      "step": 20444
-    },
-    {
-      "epoch": 10.66232073011734,
-      "grad_norm": 1.3211464881896973,
-      "learning_rate": 7.996080402010051e-05,
-      "loss": 5.8236,
-      "step": 20445
-    },
-    {
-      "epoch": 10.66284224250326,
-      "grad_norm": 1.3719292879104614,
-      "learning_rate": 7.995979899497488e-05,
-      "loss": 5.5774,
-      "step": 20446
-    },
-    {
-      "epoch": 10.663363754889179,
-      "grad_norm": 1.319704532623291,
-      "learning_rate": 7.995879396984925e-05,
-      "loss": 5.6609,
-      "step": 20447
-    },
-    {
-      "epoch": 10.663885267275099,
-      "grad_norm": 1.4762024879455566,
-      "learning_rate": 7.995778894472362e-05,
-      "loss": 5.5468,
-      "step": 20448
-    },
-    {
-      "epoch": 10.664406779661016,
-      "grad_norm": 1.414969801902771,
-      "learning_rate": 7.995678391959798e-05,
-      "loss": 4.8806,
-      "step": 20449
-    },
-    {
-      "epoch": 10.664928292046936,
-      "grad_norm": 1.430688738822937,
-      "learning_rate": 7.995577889447236e-05,
-      "loss": 5.6722,
-      "step": 20450
-    },
-    {
-      "epoch": 10.665449804432855,
-      "grad_norm": 1.4975852966308594,
-      "learning_rate": 7.995477386934674e-05,
-      "loss": 5.1847,
-      "step": 20451
-    },
-    {
-      "epoch": 10.665971316818775,
-      "grad_norm": 1.4785892963409424,
-      "learning_rate": 7.995376884422112e-05,
-      "loss": 5.4119,
-      "step": 20452
-    },
-    {
-      "epoch": 10.666492829204694,
-      "grad_norm": 1.5793944597244263,
-      "learning_rate": 7.995276381909548e-05,
-      "loss": 5.3995,
-      "step": 20453
-    },
-    {
-      "epoch": 10.667014341590614,
-      "grad_norm": 1.783992886543274,
-      "learning_rate": 7.995175879396986e-05,
-      "loss": 4.6559,
-      "step": 20454
-    },
-    {
-      "epoch": 10.667535853976531,
-      "grad_norm": 1.4308851957321167,
-      "learning_rate": 7.995075376884422e-05,
-      "loss": 5.8921,
-      "step": 20455
-    },
-    {
-      "epoch": 10.66805736636245,
-      "grad_norm": 1.4882891178131104,
-      "learning_rate": 7.99497487437186e-05,
-      "loss": 4.7408,
-      "step": 20456
-    },
-    {
-      "epoch": 10.66857887874837,
-      "grad_norm": 1.6278200149536133,
-      "learning_rate": 7.994874371859296e-05,
-      "loss": 5.2355,
-      "step": 20457
-    },
-    {
-      "epoch": 10.66910039113429,
-      "grad_norm": 1.4345632791519165,
-      "learning_rate": 7.994773869346734e-05,
-      "loss": 4.9048,
-      "step": 20458
-    },
-    {
-      "epoch": 10.66962190352021,
-      "grad_norm": 1.4995640516281128,
-      "learning_rate": 7.99467336683417e-05,
-      "loss": 5.453,
-      "step": 20459
-    },
-    {
-      "epoch": 10.670143415906129,
-      "grad_norm": 1.3975576162338257,
-      "learning_rate": 7.994572864321608e-05,
-      "loss": 5.6245,
-      "step": 20460
-    },
-    {
-      "epoch": 10.670664928292046,
-      "grad_norm": 1.36367928981781,
-      "learning_rate": 7.994472361809046e-05,
-      "loss": 5.4647,
-      "step": 20461
-    },
-    {
-      "epoch": 10.671186440677966,
-      "grad_norm": 1.3793821334838867,
-      "learning_rate": 7.994371859296483e-05,
-      "loss": 5.9298,
-      "step": 20462
-    },
-    {
-      "epoch": 10.671707953063885,
-      "grad_norm": 1.3245853185653687,
-      "learning_rate": 7.99427135678392e-05,
-      "loss": 5.6329,
-      "step": 20463
-    },
-    {
-      "epoch": 10.672229465449805,
-      "grad_norm": 1.4752954244613647,
-      "learning_rate": 7.994170854271357e-05,
-      "loss": 5.422,
-      "step": 20464
-    },
-    {
-      "epoch": 10.672750977835724,
-      "grad_norm": 1.2769120931625366,
-      "learning_rate": 7.994070351758795e-05,
-      "loss": 5.3849,
-      "step": 20465
-    },
-    {
-      "epoch": 10.673272490221644,
-      "grad_norm": 1.4323991537094116,
-      "learning_rate": 7.993969849246231e-05,
-      "loss": 5.4029,
-      "step": 20466
-    },
-    {
-      "epoch": 10.673794002607561,
-      "grad_norm": 1.462304949760437,
-      "learning_rate": 7.993869346733669e-05,
-      "loss": 5.1681,
-      "step": 20467
-    },
-    {
-      "epoch": 10.67431551499348,
-      "grad_norm": 1.4048998355865479,
-      "learning_rate": 7.993768844221105e-05,
-      "loss": 5.5977,
-      "step": 20468
-    },
-    {
-      "epoch": 10.6748370273794,
-      "grad_norm": 1.4725191593170166,
-      "learning_rate": 7.993668341708543e-05,
-      "loss": 5.4992,
-      "step": 20469
-    },
-    {
-      "epoch": 10.67535853976532,
-      "grad_norm": 1.4131604433059692,
-      "learning_rate": 7.99356783919598e-05,
-      "loss": 4.9901,
-      "step": 20470
-    },
-    {
-      "epoch": 10.67588005215124,
-      "grad_norm": 1.3895456790924072,
-      "learning_rate": 7.993467336683417e-05,
-      "loss": 5.1311,
-      "step": 20471
-    },
-    {
-      "epoch": 10.676401564537159,
-      "grad_norm": 1.4617865085601807,
-      "learning_rate": 7.993366834170855e-05,
-      "loss": 5.5898,
-      "step": 20472
-    },
-    {
-      "epoch": 10.676923076923076,
-      "grad_norm": 1.4720485210418701,
-      "learning_rate": 7.993266331658293e-05,
-      "loss": 5.0233,
-      "step": 20473
-    },
-    {
-      "epoch": 10.677444589308996,
-      "grad_norm": 1.3894908428192139,
-      "learning_rate": 7.993165829145729e-05,
-      "loss": 5.8299,
-      "step": 20474
-    },
-    {
-      "epoch": 10.677966101694915,
-      "grad_norm": 1.42914879322052,
-      "learning_rate": 7.993065326633167e-05,
-      "loss": 5.3665,
-      "step": 20475
-    },
-    {
-      "epoch": 10.678487614080835,
-      "grad_norm": 1.4619572162628174,
-      "learning_rate": 7.992964824120603e-05,
-      "loss": 5.764,
-      "step": 20476
-    },
-    {
-      "epoch": 10.679009126466754,
-      "grad_norm": 1.4516571760177612,
-      "learning_rate": 7.99286432160804e-05,
-      "loss": 5.1505,
-      "step": 20477
-    },
-    {
-      "epoch": 10.679530638852672,
-      "grad_norm": 1.561389684677124,
-      "learning_rate": 7.992763819095478e-05,
-      "loss": 5.2136,
-      "step": 20478
-    },
-    {
-      "epoch": 10.680052151238591,
-      "grad_norm": 1.3896256685256958,
-      "learning_rate": 7.992663316582914e-05,
-      "loss": 5.3542,
-      "step": 20479
-    },
-    {
-      "epoch": 10.68057366362451,
-      "grad_norm": 1.445923924446106,
-      "learning_rate": 7.992562814070352e-05,
-      "loss": 5.2904,
-      "step": 20480
-    },
-    {
-      "epoch": 10.68109517601043,
-      "grad_norm": 1.473493218421936,
-      "learning_rate": 7.99246231155779e-05,
-      "loss": 5.6773,
-      "step": 20481
-    },
-    {
-      "epoch": 10.68161668839635,
-      "grad_norm": 1.4576836824417114,
-      "learning_rate": 7.992361809045227e-05,
-      "loss": 5.597,
-      "step": 20482
-    },
-    {
-      "epoch": 10.68213820078227,
-      "grad_norm": 1.3500571250915527,
-      "learning_rate": 7.992261306532664e-05,
-      "loss": 5.8563,
-      "step": 20483
-    },
-    {
-      "epoch": 10.682659713168189,
-      "grad_norm": 1.4064286947250366,
-      "learning_rate": 7.992160804020102e-05,
-      "loss": 5.6306,
-      "step": 20484
-    },
-    {
-      "epoch": 10.683181225554106,
-      "grad_norm": 1.4754688739776611,
-      "learning_rate": 7.992060301507538e-05,
-      "loss": 5.6553,
-      "step": 20485
-    },
-    {
-      "epoch": 10.683702737940026,
-      "grad_norm": 1.4821614027023315,
-      "learning_rate": 7.991959798994976e-05,
-      "loss": 4.665,
-      "step": 20486
-    },
-    {
-      "epoch": 10.684224250325945,
-      "grad_norm": 1.4698485136032104,
-      "learning_rate": 7.991859296482412e-05,
-      "loss": 5.148,
-      "step": 20487
-    },
-    {
-      "epoch": 10.684745762711865,
-      "grad_norm": 1.473509669303894,
-      "learning_rate": 7.99175879396985e-05,
-      "loss": 5.7363,
-      "step": 20488
-    },
-    {
-      "epoch": 10.685267275097784,
-      "grad_norm": 1.3971935510635376,
-      "learning_rate": 7.991658291457286e-05,
-      "loss": 5.5441,
-      "step": 20489
-    },
-    {
-      "epoch": 10.685788787483702,
-      "grad_norm": 1.3447508811950684,
-      "learning_rate": 7.991557788944723e-05,
-      "loss": 5.4796,
-      "step": 20490
-    },
-    {
-      "epoch": 10.686310299869621,
-      "grad_norm": 1.4347686767578125,
-      "learning_rate": 7.991457286432161e-05,
-      "loss": 5.2088,
-      "step": 20491
-    },
-    {
-      "epoch": 10.68683181225554,
-      "grad_norm": 1.4749305248260498,
-      "learning_rate": 7.991356783919598e-05,
-      "loss": 5.5624,
-      "step": 20492
-    },
-    {
-      "epoch": 10.68735332464146,
-      "grad_norm": 1.520880103111267,
-      "learning_rate": 7.991256281407036e-05,
-      "loss": 5.2881,
-      "step": 20493
-    },
-    {
-      "epoch": 10.68787483702738,
-      "grad_norm": 1.3556042909622192,
-      "learning_rate": 7.991155778894473e-05,
-      "loss": 5.4571,
-      "step": 20494
-    },
-    {
-      "epoch": 10.6883963494133,
-      "grad_norm": 1.3905621767044067,
-      "learning_rate": 7.99105527638191e-05,
-      "loss": 5.4627,
-      "step": 20495
-    },
-    {
-      "epoch": 10.688917861799219,
-      "grad_norm": 1.4204280376434326,
-      "learning_rate": 7.990954773869347e-05,
-      "loss": 5.1988,
-      "step": 20496
-    },
-    {
-      "epoch": 10.689439374185136,
-      "grad_norm": 1.3841419219970703,
-      "learning_rate": 7.990854271356785e-05,
-      "loss": 5.7257,
-      "step": 20497
-    },
-    {
-      "epoch": 10.689960886571056,
-      "grad_norm": 1.5838942527770996,
-      "learning_rate": 7.990753768844221e-05,
-      "loss": 4.9205,
-      "step": 20498
-    },
-    {
-      "epoch": 10.690482398956975,
-      "grad_norm": 1.4827369451522827,
-      "learning_rate": 7.990653266331659e-05,
-      "loss": 5.4745,
-      "step": 20499
-    },
-    {
-      "epoch": 10.691003911342895,
-      "grad_norm": 1.6517480611801147,
-      "learning_rate": 7.990552763819095e-05,
-      "loss": 4.6473,
-      "step": 20500
-    },
-    {
-      "epoch": 10.691525423728814,
-      "grad_norm": 1.4296152591705322,
-      "learning_rate": 7.990452261306533e-05,
-      "loss": 5.5354,
-      "step": 20501
-    },
-    {
-      "epoch": 10.692046936114732,
-      "grad_norm": 1.4654333591461182,
-      "learning_rate": 7.990351758793971e-05,
-      "loss": 5.4749,
-      "step": 20502
-    },
-    {
-      "epoch": 10.692568448500651,
-      "grad_norm": 1.370771884918213,
-      "learning_rate": 7.990251256281407e-05,
-      "loss": 5.3795,
-      "step": 20503
-    },
-    {
-      "epoch": 10.69308996088657,
-      "grad_norm": 1.4126360416412354,
-      "learning_rate": 7.990150753768845e-05,
-      "loss": 5.5419,
-      "step": 20504
-    },
-    {
-      "epoch": 10.69361147327249,
-      "grad_norm": 1.541509747505188,
-      "learning_rate": 7.990050251256282e-05,
-      "loss": 4.792,
-      "step": 20505
-    },
-    {
-      "epoch": 10.69413298565841,
-      "grad_norm": 1.4434982538223267,
-      "learning_rate": 7.989949748743719e-05,
-      "loss": 5.5915,
-      "step": 20506
-    },
-    {
-      "epoch": 10.69465449804433,
-      "grad_norm": 1.429824709892273,
-      "learning_rate": 7.989849246231156e-05,
-      "loss": 5.1691,
-      "step": 20507
-    },
-    {
-      "epoch": 10.695176010430247,
-      "grad_norm": 1.3567109107971191,
-      "learning_rate": 7.989748743718594e-05,
-      "loss": 5.686,
-      "step": 20508
-    },
-    {
-      "epoch": 10.695697522816166,
-      "grad_norm": 1.2894232273101807,
-      "learning_rate": 7.98964824120603e-05,
-      "loss": 5.9141,
-      "step": 20509
-    },
-    {
-      "epoch": 10.696219035202086,
-      "grad_norm": 1.4606778621673584,
-      "learning_rate": 7.989547738693468e-05,
-      "loss": 5.4178,
-      "step": 20510
-    },
-    {
-      "epoch": 10.696740547588005,
-      "grad_norm": 1.345837950706482,
-      "learning_rate": 7.989447236180904e-05,
-      "loss": 5.3253,
-      "step": 20511
-    },
-    {
-      "epoch": 10.697262059973925,
-      "grad_norm": 1.3714817762374878,
-      "learning_rate": 7.989346733668342e-05,
-      "loss": 5.3431,
-      "step": 20512
-    },
-    {
-      "epoch": 10.697783572359844,
-      "grad_norm": 1.4228413105010986,
-      "learning_rate": 7.98924623115578e-05,
-      "loss": 5.1586,
-      "step": 20513
-    },
-    {
-      "epoch": 10.698305084745762,
-      "grad_norm": 1.387399435043335,
-      "learning_rate": 7.989145728643217e-05,
-      "loss": 5.5915,
-      "step": 20514
-    },
-    {
-      "epoch": 10.698826597131681,
-      "grad_norm": 1.4857217073440552,
-      "learning_rate": 7.989045226130654e-05,
-      "loss": 5.0827,
-      "step": 20515
-    },
-    {
-      "epoch": 10.6993481095176,
-      "grad_norm": 1.3520209789276123,
-      "learning_rate": 7.98894472361809e-05,
-      "loss": 5.4773,
-      "step": 20516
-    },
-    {
-      "epoch": 10.69986962190352,
-      "grad_norm": 1.5135937929153442,
-      "learning_rate": 7.988844221105528e-05,
-      "loss": 5.2222,
-      "step": 20517
-    },
-    {
-      "epoch": 10.70039113428944,
-      "grad_norm": 1.541866660118103,
-      "learning_rate": 7.988743718592965e-05,
-      "loss": 5.5711,
-      "step": 20518
-    },
-    {
-      "epoch": 10.70091264667536,
-      "grad_norm": 1.5338534116744995,
-      "learning_rate": 7.988643216080402e-05,
-      "loss": 5.0605,
-      "step": 20519
-    },
-    {
-      "epoch": 10.701434159061277,
-      "grad_norm": 1.3591800928115845,
-      "learning_rate": 7.988542713567839e-05,
-      "loss": 5.537,
-      "step": 20520
-    },
-    {
-      "epoch": 10.701955671447196,
-      "grad_norm": 1.4054569005966187,
-      "learning_rate": 7.988442211055277e-05,
-      "loss": 5.7144,
-      "step": 20521
-    },
-    {
-      "epoch": 10.702477183833116,
-      "grad_norm": 1.3315844535827637,
-      "learning_rate": 7.988341708542714e-05,
-      "loss": 5.5304,
-      "step": 20522
-    },
-    {
-      "epoch": 10.702998696219035,
-      "grad_norm": 1.3895169496536255,
-      "learning_rate": 7.988241206030152e-05,
-      "loss": 5.6002,
-      "step": 20523
-    },
-    {
-      "epoch": 10.703520208604955,
-      "grad_norm": 1.370125412940979,
-      "learning_rate": 7.988140703517589e-05,
-      "loss": 5.6228,
-      "step": 20524
-    },
-    {
-      "epoch": 10.704041720990874,
-      "grad_norm": 1.4668270349502563,
-      "learning_rate": 7.988040201005026e-05,
-      "loss": 5.4316,
-      "step": 20525
-    },
-    {
-      "epoch": 10.704563233376792,
-      "grad_norm": 1.458225965499878,
-      "learning_rate": 7.987939698492463e-05,
-      "loss": 5.6003,
-      "step": 20526
-    },
-    {
-      "epoch": 10.705084745762711,
-      "grad_norm": 1.398246169090271,
-      "learning_rate": 7.9878391959799e-05,
-      "loss": 5.6214,
-      "step": 20527
-    },
-    {
-      "epoch": 10.70560625814863,
-      "grad_norm": 1.4032458066940308,
-      "learning_rate": 7.987738693467337e-05,
-      "loss": 5.3571,
-      "step": 20528
-    },
-    {
-      "epoch": 10.70612777053455,
-      "grad_norm": 1.4126532077789307,
-      "learning_rate": 7.987638190954773e-05,
-      "loss": 5.7295,
-      "step": 20529
-    },
-    {
-      "epoch": 10.70664928292047,
-      "grad_norm": 1.3040403127670288,
-      "learning_rate": 7.987537688442211e-05,
-      "loss": 5.6338,
-      "step": 20530
-    },
-    {
-      "epoch": 10.70717079530639,
-      "grad_norm": 1.4223980903625488,
-      "learning_rate": 7.987437185929648e-05,
-      "loss": 5.7874,
-      "step": 20531
-    },
-    {
-      "epoch": 10.707692307692307,
-      "grad_norm": 1.4649335145950317,
-      "learning_rate": 7.987336683417085e-05,
-      "loss": 5.3489,
-      "step": 20532
-    },
-    {
-      "epoch": 10.708213820078226,
-      "grad_norm": 1.415537714958191,
-      "learning_rate": 7.987236180904523e-05,
-      "loss": 5.1988,
-      "step": 20533
-    },
-    {
-      "epoch": 10.708735332464146,
-      "grad_norm": 1.4973971843719482,
-      "learning_rate": 7.987135678391961e-05,
-      "loss": 5.4386,
-      "step": 20534
-    },
-    {
-      "epoch": 10.709256844850065,
-      "grad_norm": 1.326097846031189,
-      "learning_rate": 7.987035175879397e-05,
-      "loss": 5.3591,
-      "step": 20535
-    },
-    {
-      "epoch": 10.709778357235985,
-      "grad_norm": 1.4599062204360962,
-      "learning_rate": 7.986934673366835e-05,
-      "loss": 5.3282,
-      "step": 20536
-    },
-    {
-      "epoch": 10.710299869621904,
-      "grad_norm": 1.5606449842453003,
-      "learning_rate": 7.986834170854272e-05,
-      "loss": 5.7351,
-      "step": 20537
-    },
-    {
-      "epoch": 10.710821382007822,
-      "grad_norm": 1.2671868801116943,
-      "learning_rate": 7.98673366834171e-05,
-      "loss": 4.8856,
-      "step": 20538
-    },
-    {
-      "epoch": 10.711342894393741,
-      "grad_norm": 1.4558982849121094,
-      "learning_rate": 7.986633165829146e-05,
-      "loss": 5.8186,
-      "step": 20539
-    },
-    {
-      "epoch": 10.711864406779661,
-      "grad_norm": 1.5179119110107422,
-      "learning_rate": 7.986532663316584e-05,
-      "loss": 5.5546,
-      "step": 20540
-    },
-    {
-      "epoch": 10.71238591916558,
-      "grad_norm": 1.5825707912445068,
-      "learning_rate": 7.98643216080402e-05,
-      "loss": 5.453,
-      "step": 20541
-    },
-    {
-      "epoch": 10.7129074315515,
-      "grad_norm": 1.4324754476547241,
-      "learning_rate": 7.986331658291458e-05,
-      "loss": 5.0837,
-      "step": 20542
-    },
-    {
-      "epoch": 10.71342894393742,
-      "grad_norm": 1.4329947233200073,
-      "learning_rate": 7.986231155778896e-05,
-      "loss": 5.5977,
-      "step": 20543
-    },
-    {
-      "epoch": 10.713950456323337,
-      "grad_norm": 1.4459701776504517,
-      "learning_rate": 7.986130653266332e-05,
-      "loss": 5.6189,
-      "step": 20544
-    },
-    {
-      "epoch": 10.714471968709256,
-      "grad_norm": 1.4875367879867554,
-      "learning_rate": 7.98603015075377e-05,
-      "loss": 5.2919,
-      "step": 20545
-    },
-    {
-      "epoch": 10.714993481095176,
-      "grad_norm": 1.6733160018920898,
-      "learning_rate": 7.985929648241206e-05,
-      "loss": 5.6982,
-      "step": 20546
-    },
-    {
-      "epoch": 10.715514993481095,
-      "grad_norm": 1.4029656648635864,
-      "learning_rate": 7.985829145728644e-05,
-      "loss": 5.5928,
-      "step": 20547
-    },
-    {
-      "epoch": 10.716036505867015,
-      "grad_norm": 1.440141201019287,
-      "learning_rate": 7.98572864321608e-05,
-      "loss": 5.7757,
-      "step": 20548
-    },
-    {
-      "epoch": 10.716558018252934,
-      "grad_norm": 1.5664933919906616,
-      "learning_rate": 7.985628140703518e-05,
-      "loss": 5.5303,
-      "step": 20549
-    },
-    {
-      "epoch": 10.717079530638852,
-      "grad_norm": 1.6042330265045166,
-      "learning_rate": 7.985527638190955e-05,
-      "loss": 4.6371,
-      "step": 20550
-    },
-    {
-      "epoch": 10.717601043024771,
-      "grad_norm": 1.6687901020050049,
-      "learning_rate": 7.985427135678392e-05,
-      "loss": 5.2901,
-      "step": 20551
-    },
-    {
-      "epoch": 10.718122555410691,
-      "grad_norm": 1.4783134460449219,
-      "learning_rate": 7.985326633165829e-05,
-      "loss": 5.3499,
-      "step": 20552
-    },
-    {
-      "epoch": 10.71864406779661,
-      "grad_norm": 1.5388374328613281,
-      "learning_rate": 7.985226130653267e-05,
-      "loss": 5.5597,
-      "step": 20553
-    },
-    {
-      "epoch": 10.71916558018253,
-      "grad_norm": 1.464840292930603,
-      "learning_rate": 7.985125628140704e-05,
-      "loss": 5.3158,
-      "step": 20554
-    },
-    {
-      "epoch": 10.71968709256845,
-      "grad_norm": 1.5695183277130127,
-      "learning_rate": 7.985025125628141e-05,
-      "loss": 5.1365,
-      "step": 20555
-    },
-    {
-      "epoch": 10.720208604954367,
-      "grad_norm": 1.4815515279769897,
-      "learning_rate": 7.984924623115579e-05,
-      "loss": 5.3227,
-      "step": 20556
-    },
-    {
-      "epoch": 10.720730117340286,
-      "grad_norm": 1.4057945013046265,
-      "learning_rate": 7.984824120603015e-05,
-      "loss": 5.7882,
-      "step": 20557
-    },
-    {
-      "epoch": 10.721251629726206,
-      "grad_norm": 1.5721955299377441,
-      "learning_rate": 7.984723618090453e-05,
-      "loss": 4.9397,
-      "step": 20558
-    },
-    {
-      "epoch": 10.721773142112125,
-      "grad_norm": 1.500988483428955,
-      "learning_rate": 7.984623115577889e-05,
-      "loss": 5.2082,
-      "step": 20559
-    },
-    {
-      "epoch": 10.722294654498045,
-      "grad_norm": 1.5223915576934814,
-      "learning_rate": 7.984522613065327e-05,
-      "loss": 5.356,
-      "step": 20560
-    },
-    {
-      "epoch": 10.722816166883963,
-      "grad_norm": 1.5964893102645874,
-      "learning_rate": 7.984422110552763e-05,
-      "loss": 5.2922,
-      "step": 20561
-    },
-    {
-      "epoch": 10.723337679269882,
-      "grad_norm": 1.3739266395568848,
-      "learning_rate": 7.984321608040201e-05,
-      "loss": 5.8048,
-      "step": 20562
-    },
-    {
-      "epoch": 10.723859191655801,
-      "grad_norm": 1.546673059463501,
-      "learning_rate": 7.984221105527638e-05,
-      "loss": 5.0196,
-      "step": 20563
-    },
-    {
-      "epoch": 10.724380704041721,
-      "grad_norm": 1.407586693763733,
-      "learning_rate": 7.984120603015075e-05,
-      "loss": 5.2983,
-      "step": 20564
-    },
-    {
-      "epoch": 10.72490221642764,
-      "grad_norm": 1.351186752319336,
-      "learning_rate": 7.984020100502513e-05,
-      "loss": 5.1317,
-      "step": 20565
-    },
-    {
-      "epoch": 10.72542372881356,
-      "grad_norm": 1.4932016134262085,
-      "learning_rate": 7.983919597989951e-05,
-      "loss": 5.7853,
-      "step": 20566
-    },
-    {
-      "epoch": 10.72594524119948,
-      "grad_norm": 1.4835835695266724,
-      "learning_rate": 7.983819095477387e-05,
-      "loss": 5.4415,
-      "step": 20567
-    },
-    {
-      "epoch": 10.726466753585397,
-      "grad_norm": 1.4884053468704224,
-      "learning_rate": 7.983718592964825e-05,
-      "loss": 5.5918,
-      "step": 20568
-    },
-    {
-      "epoch": 10.726988265971316,
-      "grad_norm": 1.3119715452194214,
-      "learning_rate": 7.983618090452262e-05,
-      "loss": 5.7149,
-      "step": 20569
-    },
-    {
-      "epoch": 10.727509778357236,
-      "grad_norm": 1.5007261037826538,
-      "learning_rate": 7.983517587939698e-05,
-      "loss": 5.1437,
-      "step": 20570
-    },
-    {
-      "epoch": 10.728031290743155,
-      "grad_norm": 1.5473265647888184,
-      "learning_rate": 7.983417085427136e-05,
-      "loss": 5.7658,
-      "step": 20571
-    },
-    {
-      "epoch": 10.728552803129075,
-      "grad_norm": 1.4065437316894531,
-      "learning_rate": 7.983316582914572e-05,
-      "loss": 5.7831,
-      "step": 20572
-    },
-    {
-      "epoch": 10.729074315514993,
-      "grad_norm": 1.370995283126831,
-      "learning_rate": 7.98321608040201e-05,
-      "loss": 5.4698,
-      "step": 20573
-    },
-    {
-      "epoch": 10.729595827900912,
-      "grad_norm": 1.4958773851394653,
-      "learning_rate": 7.983115577889448e-05,
-      "loss": 5.456,
-      "step": 20574
-    },
-    {
-      "epoch": 10.730117340286832,
-      "grad_norm": 1.531615972518921,
-      "learning_rate": 7.983015075376886e-05,
-      "loss": 5.1943,
-      "step": 20575
-    },
-    {
-      "epoch": 10.730638852672751,
-      "grad_norm": 1.5144647359848022,
-      "learning_rate": 7.982914572864322e-05,
-      "loss": 5.6949,
-      "step": 20576
-    },
-    {
-      "epoch": 10.73116036505867,
-      "grad_norm": 1.727154016494751,
-      "learning_rate": 7.98281407035176e-05,
-      "loss": 4.4463,
-      "step": 20577
-    },
-    {
-      "epoch": 10.73168187744459,
-      "grad_norm": 1.5031750202178955,
-      "learning_rate": 7.982713567839196e-05,
-      "loss": 4.8425,
-      "step": 20578
-    },
-    {
-      "epoch": 10.73220338983051,
-      "grad_norm": 1.4502664804458618,
-      "learning_rate": 7.982613065326634e-05,
-      "loss": 5.8831,
-      "step": 20579
-    },
-    {
-      "epoch": 10.732724902216427,
-      "grad_norm": 1.4416425228118896,
-      "learning_rate": 7.98251256281407e-05,
-      "loss": 5.2087,
-      "step": 20580
-    },
-    {
-      "epoch": 10.733246414602347,
-      "grad_norm": 1.4262460470199585,
-      "learning_rate": 7.982412060301508e-05,
-      "loss": 5.4214,
-      "step": 20581
-    },
-    {
-      "epoch": 10.733767926988266,
-      "grad_norm": 1.4188470840454102,
-      "learning_rate": 7.982311557788945e-05,
-      "loss": 5.267,
-      "step": 20582
-    },
-    {
-      "epoch": 10.734289439374185,
-      "grad_norm": 1.4282373189926147,
-      "learning_rate": 7.982211055276381e-05,
-      "loss": 5.2587,
-      "step": 20583
-    },
-    {
-      "epoch": 10.734810951760105,
-      "grad_norm": 1.474522352218628,
-      "learning_rate": 7.982110552763819e-05,
-      "loss": 5.5145,
-      "step": 20584
-    },
-    {
-      "epoch": 10.735332464146023,
-      "grad_norm": 1.9726178646087646,
-      "learning_rate": 7.982010050251257e-05,
-      "loss": 5.3637,
-      "step": 20585
-    },
-    {
-      "epoch": 10.735853976531942,
-      "grad_norm": 1.3897017240524292,
-      "learning_rate": 7.981909547738694e-05,
-      "loss": 5.5929,
-      "step": 20586
-    },
-    {
-      "epoch": 10.736375488917862,
-      "grad_norm": 1.440308690071106,
-      "learning_rate": 7.981809045226131e-05,
-      "loss": 5.7166,
-      "step": 20587
-    },
-    {
-      "epoch": 10.736897001303781,
-      "grad_norm": 1.404708743095398,
-      "learning_rate": 7.981708542713569e-05,
-      "loss": 5.6116,
-      "step": 20588
-    },
-    {
-      "epoch": 10.7374185136897,
-      "grad_norm": 1.6150121688842773,
-      "learning_rate": 7.981608040201005e-05,
-      "loss": 4.985,
-      "step": 20589
-    },
-    {
-      "epoch": 10.73794002607562,
-      "grad_norm": 1.4066436290740967,
-      "learning_rate": 7.981507537688443e-05,
-      "loss": 5.3529,
-      "step": 20590
-    },
-    {
-      "epoch": 10.73846153846154,
-      "grad_norm": 1.3918448686599731,
-      "learning_rate": 7.98140703517588e-05,
-      "loss": 5.6902,
-      "step": 20591
-    },
-    {
-      "epoch": 10.738983050847457,
-      "grad_norm": 1.4605953693389893,
-      "learning_rate": 7.981306532663317e-05,
-      "loss": 5.3604,
-      "step": 20592
-    },
-    {
-      "epoch": 10.739504563233377,
-      "grad_norm": 1.4968782663345337,
-      "learning_rate": 7.981206030150754e-05,
-      "loss": 5.4065,
-      "step": 20593
-    },
-    {
-      "epoch": 10.740026075619296,
-      "grad_norm": 1.468072772026062,
-      "learning_rate": 7.981105527638191e-05,
-      "loss": 5.3069,
-      "step": 20594
-    },
-    {
-      "epoch": 10.740547588005215,
-      "grad_norm": 1.4255447387695312,
-      "learning_rate": 7.981005025125629e-05,
-      "loss": 5.6418,
-      "step": 20595
-    },
-    {
-      "epoch": 10.741069100391135,
-      "grad_norm": 1.4637668132781982,
-      "learning_rate": 7.980904522613066e-05,
-      "loss": 5.3355,
-      "step": 20596
-    },
-    {
-      "epoch": 10.741590612777053,
-      "grad_norm": 1.397374153137207,
-      "learning_rate": 7.980804020100503e-05,
-      "loss": 5.5698,
-      "step": 20597
-    },
-    {
-      "epoch": 10.742112125162972,
-      "grad_norm": 1.549070119857788,
-      "learning_rate": 7.98070351758794e-05,
-      "loss": 5.0289,
-      "step": 20598
-    },
-    {
-      "epoch": 10.742633637548892,
-      "grad_norm": 1.3298895359039307,
-      "learning_rate": 7.980603015075378e-05,
-      "loss": 5.7493,
-      "step": 20599
-    },
-    {
-      "epoch": 10.743155149934811,
-      "grad_norm": 1.4399086236953735,
-      "learning_rate": 7.980502512562814e-05,
-      "loss": 5.7734,
-      "step": 20600
-    },
-    {
-      "epoch": 10.74367666232073,
-      "grad_norm": 1.46286141872406,
-      "learning_rate": 7.980402010050252e-05,
-      "loss": 5.1524,
-      "step": 20601
-    },
-    {
-      "epoch": 10.74419817470665,
-      "grad_norm": 1.3924506902694702,
-      "learning_rate": 7.980301507537688e-05,
-      "loss": 5.0101,
-      "step": 20602
-    },
-    {
-      "epoch": 10.744719687092568,
-      "grad_norm": 1.382190227508545,
-      "learning_rate": 7.980201005025126e-05,
-      "loss": 5.4776,
-      "step": 20603
-    },
-    {
-      "epoch": 10.745241199478487,
-      "grad_norm": 1.4183335304260254,
-      "learning_rate": 7.980100502512562e-05,
-      "loss": 5.5494,
-      "step": 20604
-    },
-    {
-      "epoch": 10.745762711864407,
-      "grad_norm": 1.3883990049362183,
-      "learning_rate": 7.98e-05,
-      "loss": 5.7059,
-      "step": 20605
-    },
-    {
-      "epoch": 10.746284224250326,
-      "grad_norm": 1.4127143621444702,
-      "learning_rate": 7.979899497487438e-05,
-      "loss": 5.5519,
-      "step": 20606
-    },
-    {
-      "epoch": 10.746805736636245,
-      "grad_norm": 1.3930257558822632,
-      "learning_rate": 7.979798994974876e-05,
-      "loss": 5.7016,
-      "step": 20607
-    },
-    {
-      "epoch": 10.747327249022165,
-      "grad_norm": 1.547387719154358,
-      "learning_rate": 7.979698492462312e-05,
-      "loss": 5.2773,
-      "step": 20608
-    },
-    {
-      "epoch": 10.747848761408083,
-      "grad_norm": 1.4675372838974,
-      "learning_rate": 7.979597989949749e-05,
-      "loss": 5.5175,
-      "step": 20609
-    },
-    {
-      "epoch": 10.748370273794002,
-      "grad_norm": 1.5473538637161255,
-      "learning_rate": 7.979497487437186e-05,
-      "loss": 5.4943,
-      "step": 20610
-    },
-    {
-      "epoch": 10.748891786179922,
-      "grad_norm": 1.530320644378662,
-      "learning_rate": 7.979396984924623e-05,
-      "loss": 5.6468,
-      "step": 20611
-    },
-    {
-      "epoch": 10.749413298565841,
-      "grad_norm": 1.407534122467041,
-      "learning_rate": 7.97929648241206e-05,
-      "loss": 5.4529,
-      "step": 20612
-    },
-    {
-      "epoch": 10.74993481095176,
-      "grad_norm": 1.476136326789856,
-      "learning_rate": 7.979195979899497e-05,
-      "loss": 5.7752,
-      "step": 20613
-    },
-    {
-      "epoch": 10.75045632333768,
-      "grad_norm": 1.4766126871109009,
-      "learning_rate": 7.979095477386935e-05,
-      "loss": 5.2872,
-      "step": 20614
-    },
-    {
-      "epoch": 10.750977835723598,
-      "grad_norm": 1.5243223905563354,
-      "learning_rate": 7.978994974874373e-05,
-      "loss": 5.7856,
-      "step": 20615
-    },
-    {
-      "epoch": 10.751499348109517,
-      "grad_norm": 1.3945000171661377,
-      "learning_rate": 7.97889447236181e-05,
-      "loss": 5.1904,
-      "step": 20616
-    },
-    {
-      "epoch": 10.752020860495437,
-      "grad_norm": 1.5416702032089233,
-      "learning_rate": 7.978793969849247e-05,
-      "loss": 4.7181,
-      "step": 20617
-    },
-    {
-      "epoch": 10.752542372881356,
-      "grad_norm": 1.6678142547607422,
-      "learning_rate": 7.978693467336685e-05,
-      "loss": 4.8833,
-      "step": 20618
-    },
-    {
-      "epoch": 10.753063885267276,
-      "grad_norm": 1.4212430715560913,
-      "learning_rate": 7.978592964824121e-05,
-      "loss": 5.7327,
-      "step": 20619
-    },
-    {
-      "epoch": 10.753585397653195,
-      "grad_norm": 1.544442057609558,
-      "learning_rate": 7.978492462311559e-05,
-      "loss": 5.0351,
-      "step": 20620
-    },
-    {
-      "epoch": 10.754106910039113,
-      "grad_norm": 1.4248130321502686,
-      "learning_rate": 7.978391959798995e-05,
-      "loss": 5.8795,
-      "step": 20621
-    },
-    {
-      "epoch": 10.754628422425032,
-      "grad_norm": 1.496909737586975,
-      "learning_rate": 7.978291457286432e-05,
-      "loss": 5.3171,
-      "step": 20622
-    },
-    {
-      "epoch": 10.755149934810952,
-      "grad_norm": 1.4453941583633423,
-      "learning_rate": 7.97819095477387e-05,
-      "loss": 5.1945,
-      "step": 20623
-    },
-    {
-      "epoch": 10.755671447196871,
-      "grad_norm": 1.563942313194275,
-      "learning_rate": 7.978090452261306e-05,
-      "loss": 5.6247,
-      "step": 20624
-    },
-    {
-      "epoch": 10.75619295958279,
-      "grad_norm": 1.424263596534729,
-      "learning_rate": 7.977989949748744e-05,
-      "loss": 5.4349,
-      "step": 20625
-    },
-    {
-      "epoch": 10.75671447196871,
-      "grad_norm": 1.327027678489685,
-      "learning_rate": 7.977889447236181e-05,
-      "loss": 5.8415,
-      "step": 20626
-    },
-    {
-      "epoch": 10.757235984354628,
-      "grad_norm": 1.448517084121704,
-      "learning_rate": 7.977788944723619e-05,
-      "loss": 5.0308,
-      "step": 20627
-    },
-    {
-      "epoch": 10.757757496740547,
-      "grad_norm": 1.4885210990905762,
-      "learning_rate": 7.977688442211056e-05,
-      "loss": 5.7001,
-      "step": 20628
-    },
-    {
-      "epoch": 10.758279009126467,
-      "grad_norm": 1.4293689727783203,
-      "learning_rate": 7.977587939698493e-05,
-      "loss": 5.3782,
-      "step": 20629
-    },
-    {
-      "epoch": 10.758800521512386,
-      "grad_norm": 1.4227476119995117,
-      "learning_rate": 7.97748743718593e-05,
-      "loss": 5.6843,
-      "step": 20630
-    },
-    {
-      "epoch": 10.759322033898306,
-      "grad_norm": 1.49143648147583,
-      "learning_rate": 7.977386934673368e-05,
-      "loss": 5.318,
-      "step": 20631
-    },
-    {
-      "epoch": 10.759843546284225,
-      "grad_norm": 1.470916986465454,
-      "learning_rate": 7.977286432160804e-05,
-      "loss": 5.8685,
-      "step": 20632
-    },
-    {
-      "epoch": 10.760365058670143,
-      "grad_norm": 1.5921236276626587,
-      "learning_rate": 7.977185929648242e-05,
-      "loss": 4.787,
-      "step": 20633
-    },
-    {
-      "epoch": 10.760886571056062,
-      "grad_norm": 1.4297915697097778,
-      "learning_rate": 7.977085427135678e-05,
-      "loss": 5.5969,
-      "step": 20634
-    },
-    {
-      "epoch": 10.761408083441982,
-      "grad_norm": 1.410170316696167,
-      "learning_rate": 7.976984924623116e-05,
-      "loss": 4.9429,
-      "step": 20635
-    },
-    {
-      "epoch": 10.761929595827901,
-      "grad_norm": 1.3154422044754028,
-      "learning_rate": 7.976884422110554e-05,
-      "loss": 5.8411,
-      "step": 20636
-    },
-    {
-      "epoch": 10.76245110821382,
-      "grad_norm": 1.5135153532028198,
-      "learning_rate": 7.97678391959799e-05,
-      "loss": 5.659,
-      "step": 20637
-    },
-    {
-      "epoch": 10.76297262059974,
-      "grad_norm": 1.5385178327560425,
-      "learning_rate": 7.976683417085428e-05,
-      "loss": 5.3286,
-      "step": 20638
-    },
-    {
-      "epoch": 10.763494132985658,
-      "grad_norm": 1.3901264667510986,
-      "learning_rate": 7.976582914572864e-05,
-      "loss": 5.6606,
-      "step": 20639
-    },
-    {
-      "epoch": 10.764015645371577,
-      "grad_norm": 1.3723020553588867,
-      "learning_rate": 7.976482412060302e-05,
-      "loss": 5.709,
-      "step": 20640
-    },
-    {
-      "epoch": 10.764537157757497,
-      "grad_norm": 1.375578761100769,
-      "learning_rate": 7.976381909547739e-05,
-      "loss": 5.7049,
-      "step": 20641
-    },
-    {
-      "epoch": 10.765058670143416,
-      "grad_norm": 1.4342303276062012,
-      "learning_rate": 7.976281407035176e-05,
-      "loss": 5.3004,
-      "step": 20642
-    },
-    {
-      "epoch": 10.765580182529336,
-      "grad_norm": 1.5188686847686768,
-      "learning_rate": 7.976180904522613e-05,
-      "loss": 5.4535,
-      "step": 20643
-    },
-    {
-      "epoch": 10.766101694915255,
-      "grad_norm": 1.5327662229537964,
-      "learning_rate": 7.97608040201005e-05,
-      "loss": 5.2864,
-      "step": 20644
-    },
-    {
-      "epoch": 10.766623207301173,
-      "grad_norm": 1.3032110929489136,
-      "learning_rate": 7.975979899497487e-05,
-      "loss": 5.5598,
-      "step": 20645
-    },
-    {
-      "epoch": 10.767144719687092,
-      "grad_norm": 1.5523598194122314,
-      "learning_rate": 7.975879396984925e-05,
-      "loss": 5.1685,
-      "step": 20646
-    },
-    {
-      "epoch": 10.767666232073012,
-      "grad_norm": 1.4198572635650635,
-      "learning_rate": 7.975778894472363e-05,
-      "loss": 5.61,
-      "step": 20647
-    },
-    {
-      "epoch": 10.768187744458931,
-      "grad_norm": 1.4331961870193481,
-      "learning_rate": 7.975678391959799e-05,
-      "loss": 5.4318,
-      "step": 20648
-    },
-    {
-      "epoch": 10.76870925684485,
-      "grad_norm": 1.4220598936080933,
-      "learning_rate": 7.975577889447237e-05,
-      "loss": 5.6553,
-      "step": 20649
-    },
-    {
-      "epoch": 10.76923076923077,
-      "grad_norm": 1.413525938987732,
-      "learning_rate": 7.975477386934673e-05,
-      "loss": 5.6348,
-      "step": 20650
-    },
-    {
-      "epoch": 10.769752281616688,
-      "grad_norm": 1.3968886137008667,
-      "learning_rate": 7.975376884422111e-05,
-      "loss": 5.5248,
-      "step": 20651
-    },
-    {
-      "epoch": 10.770273794002607,
-      "grad_norm": 1.478899359703064,
-      "learning_rate": 7.975276381909547e-05,
-      "loss": 5.501,
-      "step": 20652
-    },
-    {
-      "epoch": 10.770795306388527,
-      "grad_norm": 1.445420265197754,
-      "learning_rate": 7.975175879396985e-05,
-      "loss": 5.437,
-      "step": 20653
-    },
-    {
-      "epoch": 10.771316818774446,
-      "grad_norm": 1.3095381259918213,
-      "learning_rate": 7.975075376884422e-05,
-      "loss": 5.6446,
-      "step": 20654
-    },
-    {
-      "epoch": 10.771838331160366,
-      "grad_norm": 1.424641728401184,
-      "learning_rate": 7.97497487437186e-05,
-      "loss": 5.8524,
-      "step": 20655
-    },
-    {
-      "epoch": 10.772359843546283,
-      "grad_norm": 1.4018781185150146,
-      "learning_rate": 7.974874371859297e-05,
-      "loss": 5.8825,
-      "step": 20656
-    },
-    {
-      "epoch": 10.772881355932203,
-      "grad_norm": 1.3746228218078613,
-      "learning_rate": 7.974773869346735e-05,
-      "loss": 5.5888,
-      "step": 20657
-    },
-    {
-      "epoch": 10.773402868318122,
-      "grad_norm": 1.446773648262024,
-      "learning_rate": 7.974673366834171e-05,
-      "loss": 5.3737,
-      "step": 20658
-    },
-    {
-      "epoch": 10.773924380704042,
-      "grad_norm": 1.446924090385437,
-      "learning_rate": 7.974572864321609e-05,
-      "loss": 5.5067,
-      "step": 20659
-    },
-    {
-      "epoch": 10.774445893089961,
-      "grad_norm": 1.6077567338943481,
-      "learning_rate": 7.974472361809046e-05,
-      "loss": 5.5169,
-      "step": 20660
-    },
-    {
-      "epoch": 10.77496740547588,
-      "grad_norm": 1.3975844383239746,
-      "learning_rate": 7.974371859296483e-05,
-      "loss": 5.6503,
-      "step": 20661
-    },
-    {
-      "epoch": 10.7754889178618,
-      "grad_norm": 1.3694440126419067,
-      "learning_rate": 7.97427135678392e-05,
-      "loss": 5.7835,
-      "step": 20662
-    },
-    {
-      "epoch": 10.776010430247718,
-      "grad_norm": 1.4807727336883545,
-      "learning_rate": 7.974170854271356e-05,
-      "loss": 5.1575,
-      "step": 20663
-    },
-    {
-      "epoch": 10.776531942633637,
-      "grad_norm": 1.4640098810195923,
-      "learning_rate": 7.974070351758794e-05,
-      "loss": 5.5631,
-      "step": 20664
-    },
-    {
-      "epoch": 10.777053455019557,
-      "grad_norm": 1.4074827432632446,
-      "learning_rate": 7.97396984924623e-05,
-      "loss": 5.2005,
-      "step": 20665
-    },
-    {
-      "epoch": 10.777574967405476,
-      "grad_norm": 1.6540461778640747,
-      "learning_rate": 7.973869346733668e-05,
-      "loss": 5.5338,
-      "step": 20666
-    },
-    {
-      "epoch": 10.778096479791396,
-      "grad_norm": 1.5071027278900146,
-      "learning_rate": 7.973768844221106e-05,
-      "loss": 5.5616,
-      "step": 20667
-    },
-    {
-      "epoch": 10.778617992177313,
-      "grad_norm": 1.4506456851959229,
-      "learning_rate": 7.973668341708544e-05,
-      "loss": 5.393,
-      "step": 20668
-    },
-    {
-      "epoch": 10.779139504563233,
-      "grad_norm": 1.3780595064163208,
-      "learning_rate": 7.97356783919598e-05,
-      "loss": 5.8274,
-      "step": 20669
-    },
-    {
-      "epoch": 10.779661016949152,
-      "grad_norm": 1.5183708667755127,
-      "learning_rate": 7.973467336683418e-05,
-      "loss": 4.5431,
-      "step": 20670
-    },
-    {
-      "epoch": 10.780182529335072,
-      "grad_norm": 1.3814094066619873,
-      "learning_rate": 7.973366834170855e-05,
-      "loss": 5.7101,
-      "step": 20671
-    },
-    {
-      "epoch": 10.780704041720991,
-      "grad_norm": 1.4854841232299805,
-      "learning_rate": 7.973266331658292e-05,
-      "loss": 5.2639,
-      "step": 20672
-    },
-    {
-      "epoch": 10.78122555410691,
-      "grad_norm": 1.4210858345031738,
-      "learning_rate": 7.973165829145729e-05,
-      "loss": 5.532,
-      "step": 20673
-    },
-    {
-      "epoch": 10.78174706649283,
-      "grad_norm": 1.5298528671264648,
-      "learning_rate": 7.973065326633167e-05,
-      "loss": 5.3229,
-      "step": 20674
-    },
-    {
-      "epoch": 10.782268578878748,
-      "grad_norm": 1.4310117959976196,
-      "learning_rate": 7.972964824120603e-05,
-      "loss": 5.4179,
-      "step": 20675
-    },
-    {
-      "epoch": 10.782790091264667,
-      "grad_norm": 1.3451337814331055,
-      "learning_rate": 7.972864321608041e-05,
-      "loss": 6.0487,
-      "step": 20676
-    },
-    {
-      "epoch": 10.783311603650587,
-      "grad_norm": 1.5247447490692139,
-      "learning_rate": 7.972763819095479e-05,
-      "loss": 5.4684,
-      "step": 20677
-    },
-    {
-      "epoch": 10.783833116036506,
-      "grad_norm": 1.3892323970794678,
-      "learning_rate": 7.972663316582915e-05,
-      "loss": 5.6121,
-      "step": 20678
-    },
-    {
-      "epoch": 10.784354628422426,
-      "grad_norm": 1.4021673202514648,
-      "learning_rate": 7.972562814070353e-05,
-      "loss": 5.6363,
-      "step": 20679
-    },
-    {
-      "epoch": 10.784876140808343,
-      "grad_norm": 1.4437789916992188,
-      "learning_rate": 7.972462311557789e-05,
-      "loss": 5.5443,
-      "step": 20680
-    },
-    {
-      "epoch": 10.785397653194263,
-      "grad_norm": 1.4582810401916504,
-      "learning_rate": 7.972361809045227e-05,
-      "loss": 5.8015,
-      "step": 20681
-    },
-    {
-      "epoch": 10.785919165580182,
-      "grad_norm": 1.4905661344528198,
-      "learning_rate": 7.972261306532663e-05,
-      "loss": 5.371,
-      "step": 20682
-    },
-    {
-      "epoch": 10.786440677966102,
-      "grad_norm": 1.3369591236114502,
-      "learning_rate": 7.972160804020101e-05,
-      "loss": 5.9088,
-      "step": 20683
-    },
-    {
-      "epoch": 10.786962190352021,
-      "grad_norm": 1.423959732055664,
-      "learning_rate": 7.972060301507538e-05,
-      "loss": 4.9422,
-      "step": 20684
-    },
-    {
-      "epoch": 10.78748370273794,
-      "grad_norm": 1.4387743473052979,
-      "learning_rate": 7.971959798994975e-05,
-      "loss": 5.4876,
-      "step": 20685
-    },
-    {
-      "epoch": 10.78800521512386,
-      "grad_norm": 1.4731855392456055,
-      "learning_rate": 7.971859296482412e-05,
-      "loss": 5.5573,
-      "step": 20686
-    },
-    {
-      "epoch": 10.788526727509778,
-      "grad_norm": 1.4022163152694702,
-      "learning_rate": 7.97175879396985e-05,
-      "loss": 5.2713,
-      "step": 20687
-    },
-    {
-      "epoch": 10.789048239895697,
-      "grad_norm": 1.4151439666748047,
-      "learning_rate": 7.971658291457287e-05,
-      "loss": 4.9932,
-      "step": 20688
-    },
-    {
-      "epoch": 10.789569752281617,
-      "grad_norm": 1.446157455444336,
-      "learning_rate": 7.971557788944724e-05,
-      "loss": 5.4095,
-      "step": 20689
-    },
-    {
-      "epoch": 10.790091264667536,
-      "grad_norm": 1.43240487575531,
-      "learning_rate": 7.971457286432162e-05,
-      "loss": 5.2029,
-      "step": 20690
-    },
-    {
-      "epoch": 10.790612777053456,
-      "grad_norm": 1.520775318145752,
-      "learning_rate": 7.971356783919598e-05,
-      "loss": 4.9632,
-      "step": 20691
-    },
-    {
-      "epoch": 10.791134289439373,
-      "grad_norm": 1.4017713069915771,
-      "learning_rate": 7.971256281407036e-05,
-      "loss": 5.5734,
-      "step": 20692
-    },
-    {
-      "epoch": 10.791655801825293,
-      "grad_norm": 1.3306727409362793,
-      "learning_rate": 7.971155778894472e-05,
-      "loss": 6.0564,
-      "step": 20693
-    },
-    {
-      "epoch": 10.792177314211212,
-      "grad_norm": 1.4964497089385986,
-      "learning_rate": 7.97105527638191e-05,
-      "loss": 5.5376,
-      "step": 20694
-    },
-    {
-      "epoch": 10.792698826597132,
-      "grad_norm": 1.4500621557235718,
-      "learning_rate": 7.970954773869346e-05,
-      "loss": 5.8292,
-      "step": 20695
-    },
-    {
-      "epoch": 10.793220338983051,
-      "grad_norm": 1.338826060295105,
-      "learning_rate": 7.970854271356784e-05,
-      "loss": 4.9867,
-      "step": 20696
-    },
-    {
-      "epoch": 10.79374185136897,
-      "grad_norm": 1.378324270248413,
-      "learning_rate": 7.970753768844222e-05,
-      "loss": 5.2013,
-      "step": 20697
-    },
-    {
-      "epoch": 10.794263363754888,
-      "grad_norm": 1.4890415668487549,
-      "learning_rate": 7.97065326633166e-05,
-      "loss": 5.342,
-      "step": 20698
-    },
-    {
-      "epoch": 10.794784876140808,
-      "grad_norm": 1.5941314697265625,
-      "learning_rate": 7.970552763819096e-05,
-      "loss": 5.5614,
-      "step": 20699
-    },
-    {
-      "epoch": 10.795306388526727,
-      "grad_norm": 1.6168421506881714,
-      "learning_rate": 7.970452261306534e-05,
-      "loss": 5.3583,
-      "step": 20700
-    },
-    {
-      "epoch": 10.795827900912647,
-      "grad_norm": 1.4827972650527954,
-      "learning_rate": 7.97035175879397e-05,
-      "loss": 5.3185,
-      "step": 20701
-    },
-    {
-      "epoch": 10.796349413298566,
-      "grad_norm": 1.5068382024765015,
-      "learning_rate": 7.970251256281407e-05,
-      "loss": 5.4768,
-      "step": 20702
-    },
-    {
-      "epoch": 10.796870925684486,
-      "grad_norm": 1.9235774278640747,
-      "learning_rate": 7.970150753768845e-05,
-      "loss": 5.1556,
-      "step": 20703
-    },
-    {
-      "epoch": 10.797392438070403,
-      "grad_norm": 1.4397050142288208,
-      "learning_rate": 7.970050251256281e-05,
-      "loss": 5.4701,
-      "step": 20704
-    },
-    {
-      "epoch": 10.797913950456323,
-      "grad_norm": 1.4578948020935059,
-      "learning_rate": 7.969949748743719e-05,
-      "loss": 4.9752,
-      "step": 20705
-    },
-    {
-      "epoch": 10.798435462842242,
-      "grad_norm": 1.441693663597107,
-      "learning_rate": 7.969849246231155e-05,
-      "loss": 5.2692,
-      "step": 20706
-    },
-    {
-      "epoch": 10.798956975228162,
-      "grad_norm": 1.6203075647354126,
-      "learning_rate": 7.969748743718593e-05,
-      "loss": 5.2555,
-      "step": 20707
-    },
-    {
-      "epoch": 10.799478487614081,
-      "grad_norm": 1.597507119178772,
-      "learning_rate": 7.969648241206031e-05,
-      "loss": 5.3899,
-      "step": 20708
-    },
-    {
-      "epoch": 10.8,
-      "grad_norm": 1.623583197593689,
-      "learning_rate": 7.969547738693469e-05,
-      "loss": 5.0377,
-      "step": 20709
-    },
-    {
-      "epoch": 10.800521512385918,
-      "grad_norm": 1.3858660459518433,
-      "learning_rate": 7.969447236180905e-05,
-      "loss": 5.8976,
-      "step": 20710
-    },
-    {
-      "epoch": 10.801043024771838,
-      "grad_norm": 1.525005578994751,
-      "learning_rate": 7.969346733668343e-05,
-      "loss": 5.0873,
-      "step": 20711
-    },
-    {
-      "epoch": 10.801564537157757,
-      "grad_norm": 1.5472322702407837,
-      "learning_rate": 7.969246231155779e-05,
-      "loss": 5.5273,
-      "step": 20712
-    },
-    {
-      "epoch": 10.802086049543677,
-      "grad_norm": 1.3793222904205322,
-      "learning_rate": 7.969145728643217e-05,
-      "loss": 5.8114,
-      "step": 20713
-    },
-    {
-      "epoch": 10.802607561929596,
-      "grad_norm": 1.4530913829803467,
-      "learning_rate": 7.969045226130653e-05,
-      "loss": 5.3409,
-      "step": 20714
-    },
-    {
-      "epoch": 10.803129074315516,
-      "grad_norm": 1.414528489112854,
-      "learning_rate": 7.96894472361809e-05,
-      "loss": 5.6146,
-      "step": 20715
-    },
-    {
-      "epoch": 10.803650586701433,
-      "grad_norm": 1.4556586742401123,
-      "learning_rate": 7.968844221105528e-05,
-      "loss": 5.4034,
-      "step": 20716
-    },
-    {
-      "epoch": 10.804172099087353,
-      "grad_norm": 1.5128448009490967,
-      "learning_rate": 7.968743718592965e-05,
-      "loss": 5.2992,
-      "step": 20717
-    },
-    {
-      "epoch": 10.804693611473272,
-      "grad_norm": 1.4829140901565552,
-      "learning_rate": 7.968643216080403e-05,
-      "loss": 5.7485,
-      "step": 20718
-    },
-    {
-      "epoch": 10.805215123859192,
-      "grad_norm": 1.4861263036727905,
-      "learning_rate": 7.96854271356784e-05,
-      "loss": 5.6274,
-      "step": 20719
-    },
-    {
-      "epoch": 10.805736636245111,
-      "grad_norm": 1.4723312854766846,
-      "learning_rate": 7.968442211055277e-05,
-      "loss": 5.4937,
-      "step": 20720
-    },
-    {
-      "epoch": 10.80625814863103,
-      "grad_norm": 1.3949435949325562,
-      "learning_rate": 7.968341708542714e-05,
-      "loss": 5.9183,
-      "step": 20721
-    },
-    {
-      "epoch": 10.806779661016948,
-      "grad_norm": 1.441552996635437,
-      "learning_rate": 7.968241206030152e-05,
-      "loss": 5.6363,
-      "step": 20722
-    },
-    {
-      "epoch": 10.807301173402868,
-      "grad_norm": 1.3486603498458862,
-      "learning_rate": 7.968140703517588e-05,
-      "loss": 5.8713,
-      "step": 20723
-    },
-    {
-      "epoch": 10.807822685788787,
-      "grad_norm": 1.3587275743484497,
-      "learning_rate": 7.968040201005026e-05,
-      "loss": 5.4983,
-      "step": 20724
-    },
-    {
-      "epoch": 10.808344198174707,
-      "grad_norm": 1.4620894193649292,
-      "learning_rate": 7.967939698492462e-05,
-      "loss": 5.4494,
-      "step": 20725
-    },
-    {
-      "epoch": 10.808865710560626,
-      "grad_norm": 1.5368926525115967,
-      "learning_rate": 7.9678391959799e-05,
-      "loss": 4.9162,
-      "step": 20726
-    },
-    {
-      "epoch": 10.809387222946546,
-      "grad_norm": 1.5064212083816528,
-      "learning_rate": 7.967738693467336e-05,
-      "loss": 5.261,
-      "step": 20727
-    },
-    {
-      "epoch": 10.809908735332463,
-      "grad_norm": 1.4238957166671753,
-      "learning_rate": 7.967638190954774e-05,
-      "loss": 5.5199,
-      "step": 20728
-    },
-    {
-      "epoch": 10.810430247718383,
-      "grad_norm": 1.4107391834259033,
-      "learning_rate": 7.967537688442212e-05,
-      "loss": 5.738,
-      "step": 20729
-    },
-    {
-      "epoch": 10.810951760104302,
-      "grad_norm": 1.377893328666687,
-      "learning_rate": 7.967437185929648e-05,
-      "loss": 5.7457,
-      "step": 20730
-    },
-    {
-      "epoch": 10.811473272490222,
-      "grad_norm": 1.3903776407241821,
-      "learning_rate": 7.967336683417086e-05,
-      "loss": 5.0453,
-      "step": 20731
-    },
-    {
-      "epoch": 10.811994784876141,
-      "grad_norm": 1.5792397260665894,
-      "learning_rate": 7.967236180904523e-05,
-      "loss": 5.298,
-      "step": 20732
-    },
-    {
-      "epoch": 10.81251629726206,
-      "grad_norm": 1.4612237215042114,
-      "learning_rate": 7.96713567839196e-05,
-      "loss": 5.688,
-      "step": 20733
-    },
-    {
-      "epoch": 10.813037809647978,
-      "grad_norm": 1.410043478012085,
-      "learning_rate": 7.967035175879397e-05,
-      "loss": 5.1718,
-      "step": 20734
-    },
-    {
-      "epoch": 10.813559322033898,
-      "grad_norm": 1.4629569053649902,
-      "learning_rate": 7.966934673366835e-05,
-      "loss": 5.4274,
-      "step": 20735
-    },
-    {
-      "epoch": 10.814080834419817,
-      "grad_norm": 1.3004539012908936,
-      "learning_rate": 7.966834170854271e-05,
-      "loss": 5.0218,
-      "step": 20736
-    },
-    {
-      "epoch": 10.814602346805737,
-      "grad_norm": 1.3722233772277832,
-      "learning_rate": 7.966733668341709e-05,
-      "loss": 5.6449,
-      "step": 20737
-    },
-    {
-      "epoch": 10.815123859191656,
-      "grad_norm": 1.5427888631820679,
-      "learning_rate": 7.966633165829145e-05,
-      "loss": 5.6112,
-      "step": 20738
-    },
-    {
-      "epoch": 10.815645371577576,
-      "grad_norm": 1.6418004035949707,
-      "learning_rate": 7.966532663316583e-05,
-      "loss": 5.1956,
-      "step": 20739
-    },
-    {
-      "epoch": 10.816166883963493,
-      "grad_norm": 1.573679804801941,
-      "learning_rate": 7.966432160804021e-05,
-      "loss": 5.5393,
-      "step": 20740
-    },
-    {
-      "epoch": 10.816688396349413,
-      "grad_norm": 1.592287540435791,
-      "learning_rate": 7.966331658291459e-05,
-      "loss": 5.2725,
-      "step": 20741
-    },
-    {
-      "epoch": 10.817209908735332,
-      "grad_norm": 1.4096673727035522,
-      "learning_rate": 7.966231155778895e-05,
-      "loss": 5.4112,
-      "step": 20742
-    },
-    {
-      "epoch": 10.817731421121252,
-      "grad_norm": 1.4068632125854492,
-      "learning_rate": 7.966130653266332e-05,
-      "loss": 5.0282,
-      "step": 20743
-    },
-    {
-      "epoch": 10.818252933507171,
-      "grad_norm": 1.6458516120910645,
-      "learning_rate": 7.966030150753769e-05,
-      "loss": 5.4022,
-      "step": 20744
-    },
-    {
-      "epoch": 10.81877444589309,
-      "grad_norm": 1.4579191207885742,
-      "learning_rate": 7.965929648241206e-05,
-      "loss": 5.4319,
-      "step": 20745
-    },
-    {
-      "epoch": 10.819295958279008,
-      "grad_norm": 1.497306227684021,
-      "learning_rate": 7.965829145728644e-05,
-      "loss": 5.6255,
-      "step": 20746
-    },
-    {
-      "epoch": 10.819817470664928,
-      "grad_norm": 1.4214626550674438,
-      "learning_rate": 7.96572864321608e-05,
-      "loss": 5.5334,
-      "step": 20747
-    },
-    {
-      "epoch": 10.820338983050847,
-      "grad_norm": 1.4386236667633057,
-      "learning_rate": 7.965628140703518e-05,
-      "loss": 5.381,
-      "step": 20748
-    },
-    {
-      "epoch": 10.820860495436767,
-      "grad_norm": 1.4974688291549683,
-      "learning_rate": 7.965527638190956e-05,
-      "loss": 5.6032,
-      "step": 20749
-    },
-    {
-      "epoch": 10.821382007822686,
-      "grad_norm": 1.3928426504135132,
-      "learning_rate": 7.965427135678393e-05,
-      "loss": 5.5421,
-      "step": 20750
-    },
-    {
-      "epoch": 10.821903520208604,
-      "grad_norm": 1.4738370180130005,
-      "learning_rate": 7.96532663316583e-05,
-      "loss": 5.4332,
-      "step": 20751
-    },
-    {
-      "epoch": 10.822425032594523,
-      "grad_norm": 1.4198094606399536,
-      "learning_rate": 7.965226130653268e-05,
-      "loss": 5.8073,
-      "step": 20752
-    },
-    {
-      "epoch": 10.822946544980443,
-      "grad_norm": 1.3754414319992065,
-      "learning_rate": 7.965125628140704e-05,
-      "loss": 5.6626,
-      "step": 20753
-    },
-    {
-      "epoch": 10.823468057366362,
-      "grad_norm": 1.4757311344146729,
-      "learning_rate": 7.965025125628142e-05,
-      "loss": 5.6423,
-      "step": 20754
-    },
-    {
-      "epoch": 10.823989569752282,
-      "grad_norm": 1.5737947225570679,
-      "learning_rate": 7.964924623115578e-05,
-      "loss": 5.3111,
-      "step": 20755
-    },
-    {
-      "epoch": 10.824511082138201,
-      "grad_norm": 1.4191819429397583,
-      "learning_rate": 7.964824120603015e-05,
-      "loss": 5.6418,
-      "step": 20756
-    },
-    {
-      "epoch": 10.82503259452412,
-      "grad_norm": 1.4644770622253418,
-      "learning_rate": 7.964723618090452e-05,
-      "loss": 4.8861,
-      "step": 20757
-    },
-    {
-      "epoch": 10.825554106910038,
-      "grad_norm": 1.5920915603637695,
-      "learning_rate": 7.964623115577889e-05,
-      "loss": 5.0429,
-      "step": 20758
-    },
-    {
-      "epoch": 10.826075619295958,
-      "grad_norm": 1.499885082244873,
-      "learning_rate": 7.964522613065327e-05,
-      "loss": 5.4004,
-      "step": 20759
-    },
-    {
-      "epoch": 10.826597131681877,
-      "grad_norm": 1.513565182685852,
-      "learning_rate": 7.964422110552764e-05,
-      "loss": 5.1507,
-      "step": 20760
-    },
-    {
-      "epoch": 10.827118644067797,
-      "grad_norm": 1.448095440864563,
-      "learning_rate": 7.964321608040202e-05,
-      "loss": 5.607,
-      "step": 20761
-    },
-    {
-      "epoch": 10.827640156453716,
-      "grad_norm": 1.3437647819519043,
-      "learning_rate": 7.964221105527639e-05,
-      "loss": 5.7006,
-      "step": 20762
-    },
-    {
-      "epoch": 10.828161668839634,
-      "grad_norm": 1.5024175643920898,
-      "learning_rate": 7.964120603015076e-05,
-      "loss": 4.8149,
-      "step": 20763
-    },
-    {
-      "epoch": 10.828683181225554,
-      "grad_norm": 1.465254783630371,
-      "learning_rate": 7.964020100502513e-05,
-      "loss": 5.4909,
-      "step": 20764
-    },
-    {
-      "epoch": 10.829204693611473,
-      "grad_norm": 1.5923737287521362,
-      "learning_rate": 7.96391959798995e-05,
-      "loss": 5.5883,
-      "step": 20765
-    },
-    {
-      "epoch": 10.829726205997392,
-      "grad_norm": 1.5382260084152222,
-      "learning_rate": 7.963819095477387e-05,
-      "loss": 5.0806,
-      "step": 20766
-    },
-    {
-      "epoch": 10.830247718383312,
-      "grad_norm": 1.4349660873413086,
-      "learning_rate": 7.963718592964825e-05,
-      "loss": 5.4018,
-      "step": 20767
-    },
-    {
-      "epoch": 10.830769230769231,
-      "grad_norm": 1.3795368671417236,
-      "learning_rate": 7.963618090452261e-05,
-      "loss": 5.2949,
-      "step": 20768
-    },
-    {
-      "epoch": 10.83129074315515,
-      "grad_norm": 1.4719438552856445,
-      "learning_rate": 7.963517587939699e-05,
-      "loss": 5.475,
-      "step": 20769
-    },
-    {
-      "epoch": 10.831812255541069,
-      "grad_norm": 1.5309430360794067,
-      "learning_rate": 7.963417085427137e-05,
-      "loss": 5.4106,
-      "step": 20770
-    },
-    {
-      "epoch": 10.832333767926988,
-      "grad_norm": 1.3946200609207153,
-      "learning_rate": 7.963316582914573e-05,
-      "loss": 5.3135,
-      "step": 20771
-    },
-    {
-      "epoch": 10.832855280312907,
-      "grad_norm": 1.336147665977478,
-      "learning_rate": 7.963216080402011e-05,
-      "loss": 5.5459,
-      "step": 20772
-    },
-    {
-      "epoch": 10.833376792698827,
-      "grad_norm": 1.5295556783676147,
-      "learning_rate": 7.963115577889447e-05,
-      "loss": 5.3811,
-      "step": 20773
-    },
-    {
-      "epoch": 10.833898305084746,
-      "grad_norm": 1.5538806915283203,
-      "learning_rate": 7.963015075376885e-05,
-      "loss": 5.1544,
-      "step": 20774
-    },
-    {
-      "epoch": 10.834419817470664,
-      "grad_norm": 1.3942598104476929,
-      "learning_rate": 7.962914572864322e-05,
-      "loss": 5.3993,
-      "step": 20775
-    },
-    {
-      "epoch": 10.834941329856584,
-      "grad_norm": 1.4581702947616577,
-      "learning_rate": 7.96281407035176e-05,
-      "loss": 5.6406,
-      "step": 20776
-    },
-    {
-      "epoch": 10.835462842242503,
-      "grad_norm": 1.3644657135009766,
-      "learning_rate": 7.962713567839196e-05,
-      "loss": 4.7246,
-      "step": 20777
-    },
-    {
-      "epoch": 10.835984354628422,
-      "grad_norm": 1.478860855102539,
-      "learning_rate": 7.962613065326634e-05,
-      "loss": 5.5707,
-      "step": 20778
-    },
-    {
-      "epoch": 10.836505867014342,
-      "grad_norm": 1.4410970211029053,
-      "learning_rate": 7.96251256281407e-05,
-      "loss": 5.5544,
-      "step": 20779
-    },
-    {
-      "epoch": 10.837027379400261,
-      "grad_norm": 1.428435206413269,
-      "learning_rate": 7.962412060301508e-05,
-      "loss": 5.5196,
-      "step": 20780
-    },
-    {
-      "epoch": 10.83754889178618,
-      "grad_norm": 1.4673107862472534,
-      "learning_rate": 7.962311557788946e-05,
-      "loss": 4.6384,
-      "step": 20781
-    },
-    {
-      "epoch": 10.838070404172099,
-      "grad_norm": 1.4291393756866455,
-      "learning_rate": 7.962211055276382e-05,
-      "loss": 5.497,
-      "step": 20782
-    },
-    {
-      "epoch": 10.838591916558018,
-      "grad_norm": 1.427751898765564,
-      "learning_rate": 7.96211055276382e-05,
-      "loss": 5.493,
-      "step": 20783
-    },
-    {
-      "epoch": 10.839113428943937,
-      "grad_norm": 1.5893616676330566,
-      "learning_rate": 7.962010050251256e-05,
-      "loss": 5.5176,
-      "step": 20784
-    },
-    {
-      "epoch": 10.839634941329857,
-      "grad_norm": 1.437824010848999,
-      "learning_rate": 7.961909547738694e-05,
-      "loss": 5.5151,
-      "step": 20785
-    },
-    {
-      "epoch": 10.840156453715776,
-      "grad_norm": 1.4060401916503906,
-      "learning_rate": 7.96180904522613e-05,
-      "loss": 5.4245,
-      "step": 20786
-    },
-    {
-      "epoch": 10.840677966101694,
-      "grad_norm": 1.41290283203125,
-      "learning_rate": 7.961708542713568e-05,
-      "loss": 5.6157,
-      "step": 20787
-    },
-    {
-      "epoch": 10.841199478487614,
-      "grad_norm": 1.4320114850997925,
-      "learning_rate": 7.961608040201005e-05,
-      "loss": 5.4521,
-      "step": 20788
-    },
-    {
-      "epoch": 10.841720990873533,
-      "grad_norm": 1.5476760864257812,
-      "learning_rate": 7.961507537688442e-05,
-      "loss": 5.5295,
-      "step": 20789
-    },
-    {
-      "epoch": 10.842242503259452,
-      "grad_norm": 1.3641599416732788,
-      "learning_rate": 7.96140703517588e-05,
-      "loss": 5.1513,
-      "step": 20790
-    },
-    {
-      "epoch": 10.842764015645372,
-      "grad_norm": 1.462260365486145,
-      "learning_rate": 7.961306532663318e-05,
-      "loss": 5.6811,
-      "step": 20791
-    },
-    {
-      "epoch": 10.843285528031291,
-      "grad_norm": 1.4889155626296997,
-      "learning_rate": 7.961206030150754e-05,
-      "loss": 5.2913,
-      "step": 20792
-    },
-    {
-      "epoch": 10.843807040417209,
-      "grad_norm": 1.5695089101791382,
-      "learning_rate": 7.961105527638192e-05,
-      "loss": 5.5161,
-      "step": 20793
-    },
-    {
-      "epoch": 10.844328552803129,
-      "grad_norm": 1.630618691444397,
-      "learning_rate": 7.961005025125629e-05,
-      "loss": 5.3449,
-      "step": 20794
-    },
-    {
-      "epoch": 10.844850065189048,
-      "grad_norm": 1.3879542350769043,
-      "learning_rate": 7.960904522613065e-05,
-      "loss": 5.6203,
-      "step": 20795
-    },
-    {
-      "epoch": 10.845371577574968,
-      "grad_norm": 1.3964046239852905,
-      "learning_rate": 7.960804020100503e-05,
-      "loss": 5.5846,
-      "step": 20796
-    },
-    {
-      "epoch": 10.845893089960887,
-      "grad_norm": 1.3749357461929321,
-      "learning_rate": 7.960703517587939e-05,
-      "loss": 5.6297,
-      "step": 20797
-    },
-    {
-      "epoch": 10.846414602346806,
-      "grad_norm": 1.353702187538147,
-      "learning_rate": 7.960603015075377e-05,
-      "loss": 5.567,
-      "step": 20798
-    },
-    {
-      "epoch": 10.846936114732724,
-      "grad_norm": 1.4541682004928589,
-      "learning_rate": 7.960502512562813e-05,
-      "loss": 5.1598,
-      "step": 20799
-    },
-    {
-      "epoch": 10.847457627118644,
-      "grad_norm": 1.4829739332199097,
-      "learning_rate": 7.960402010050251e-05,
-      "loss": 5.4598,
-      "step": 20800
-    },
-    {
-      "epoch": 10.847979139504563,
-      "grad_norm": 1.4891096353530884,
-      "learning_rate": 7.960301507537689e-05,
-      "loss": 5.5299,
-      "step": 20801
-    },
-    {
-      "epoch": 10.848500651890483,
-      "grad_norm": 1.4753087759017944,
-      "learning_rate": 7.960201005025127e-05,
-      "loss": 5.6889,
-      "step": 20802
-    },
-    {
-      "epoch": 10.849022164276402,
-      "grad_norm": 1.4102638959884644,
-      "learning_rate": 7.960100502512563e-05,
-      "loss": 5.2995,
-      "step": 20803
-    },
-    {
-      "epoch": 10.849543676662321,
-      "grad_norm": 1.611689805984497,
-      "learning_rate": 7.960000000000001e-05,
-      "loss": 4.9359,
-      "step": 20804
-    },
-    {
-      "epoch": 10.85006518904824,
-      "grad_norm": 1.5377171039581299,
-      "learning_rate": 7.959899497487437e-05,
-      "loss": 5.6585,
-      "step": 20805
-    },
-    {
-      "epoch": 10.850586701434159,
-      "grad_norm": 1.4895716905593872,
-      "learning_rate": 7.959798994974875e-05,
-      "loss": 5.0642,
-      "step": 20806
-    },
-    {
-      "epoch": 10.851108213820078,
-      "grad_norm": 1.3922721147537231,
-      "learning_rate": 7.959698492462312e-05,
-      "loss": 5.1579,
-      "step": 20807
-    },
-    {
-      "epoch": 10.851629726205998,
-      "grad_norm": 1.2907953262329102,
-      "learning_rate": 7.959597989949748e-05,
-      "loss": 5.7076,
-      "step": 20808
-    },
-    {
-      "epoch": 10.852151238591917,
-      "grad_norm": 1.5816020965576172,
-      "learning_rate": 7.959497487437186e-05,
-      "loss": 4.9784,
-      "step": 20809
-    },
-    {
-      "epoch": 10.852672750977836,
-      "grad_norm": 1.4898992776870728,
-      "learning_rate": 7.959396984924624e-05,
-      "loss": 5.2614,
-      "step": 20810
-    },
-    {
-      "epoch": 10.853194263363754,
-      "grad_norm": 1.554205298423767,
-      "learning_rate": 7.959296482412061e-05,
-      "loss": 4.8855,
-      "step": 20811
-    },
-    {
-      "epoch": 10.853715775749674,
-      "grad_norm": 1.45045006275177,
-      "learning_rate": 7.959195979899498e-05,
-      "loss": 4.6911,
-      "step": 20812
-    },
-    {
-      "epoch": 10.854237288135593,
-      "grad_norm": 1.5101748704910278,
-      "learning_rate": 7.959095477386936e-05,
-      "loss": 5.2293,
-      "step": 20813
-    },
-    {
-      "epoch": 10.854758800521513,
-      "grad_norm": 1.5312845706939697,
-      "learning_rate": 7.958994974874372e-05,
-      "loss": 5.3077,
-      "step": 20814
-    },
-    {
-      "epoch": 10.855280312907432,
-      "grad_norm": 1.645363211631775,
-      "learning_rate": 7.95889447236181e-05,
-      "loss": 5.3089,
-      "step": 20815
-    },
-    {
-      "epoch": 10.855801825293351,
-      "grad_norm": 1.4280797243118286,
-      "learning_rate": 7.958793969849246e-05,
-      "loss": 5.4614,
-      "step": 20816
-    },
-    {
-      "epoch": 10.85632333767927,
-      "grad_norm": 1.599623441696167,
-      "learning_rate": 7.958693467336684e-05,
-      "loss": 5.4334,
-      "step": 20817
-    },
-    {
-      "epoch": 10.856844850065189,
-      "grad_norm": 1.5686933994293213,
-      "learning_rate": 7.95859296482412e-05,
-      "loss": 4.984,
-      "step": 20818
-    },
-    {
-      "epoch": 10.857366362451108,
-      "grad_norm": 1.4481548070907593,
-      "learning_rate": 7.958492462311558e-05,
-      "loss": 5.3231,
-      "step": 20819
-    },
-    {
-      "epoch": 10.857887874837028,
-      "grad_norm": 1.463856816291809,
-      "learning_rate": 7.958391959798995e-05,
-      "loss": 5.7162,
-      "step": 20820
-    },
-    {
-      "epoch": 10.858409387222947,
-      "grad_norm": 1.5069621801376343,
-      "learning_rate": 7.958291457286433e-05,
-      "loss": 5.3596,
-      "step": 20821
-    },
-    {
-      "epoch": 10.858930899608866,
-      "grad_norm": 1.4765156507492065,
-      "learning_rate": 7.95819095477387e-05,
-      "loss": 5.4304,
-      "step": 20822
-    },
-    {
-      "epoch": 10.859452411994784,
-      "grad_norm": 1.5320043563842773,
-      "learning_rate": 7.958090452261307e-05,
-      "loss": 5.4837,
-      "step": 20823
-    },
-    {
-      "epoch": 10.859973924380704,
-      "grad_norm": 1.481360673904419,
-      "learning_rate": 7.957989949748744e-05,
-      "loss": 5.2685,
-      "step": 20824
-    },
-    {
-      "epoch": 10.860495436766623,
-      "grad_norm": 1.4351886510849,
-      "learning_rate": 7.957889447236181e-05,
-      "loss": 5.6573,
-      "step": 20825
-    },
-    {
-      "epoch": 10.861016949152543,
-      "grad_norm": 1.5245764255523682,
-      "learning_rate": 7.957788944723619e-05,
-      "loss": 5.7187,
-      "step": 20826
-    },
-    {
-      "epoch": 10.861538461538462,
-      "grad_norm": 1.4680283069610596,
-      "learning_rate": 7.957688442211055e-05,
-      "loss": 4.985,
-      "step": 20827
-    },
-    {
-      "epoch": 10.862059973924381,
-      "grad_norm": 1.5380178689956665,
-      "learning_rate": 7.957587939698493e-05,
-      "loss": 5.5582,
-      "step": 20828
-    },
-    {
-      "epoch": 10.8625814863103,
-      "grad_norm": 1.5265454053878784,
-      "learning_rate": 7.95748743718593e-05,
-      "loss": 5.2843,
-      "step": 20829
-    },
-    {
-      "epoch": 10.863102998696219,
-      "grad_norm": 1.479627251625061,
-      "learning_rate": 7.957386934673367e-05,
-      "loss": 5.5416,
-      "step": 20830
-    },
-    {
-      "epoch": 10.863624511082138,
-      "grad_norm": 1.5117517709732056,
-      "learning_rate": 7.957286432160805e-05,
-      "loss": 5.3156,
-      "step": 20831
-    },
-    {
-      "epoch": 10.864146023468058,
-      "grad_norm": 1.4894530773162842,
-      "learning_rate": 7.957185929648243e-05,
-      "loss": 5.5364,
-      "step": 20832
-    },
-    {
-      "epoch": 10.864667535853977,
-      "grad_norm": 1.4878729581832886,
-      "learning_rate": 7.957085427135679e-05,
-      "loss": 5.5564,
-      "step": 20833
-    },
-    {
-      "epoch": 10.865189048239897,
-      "grad_norm": 1.410269021987915,
-      "learning_rate": 7.956984924623117e-05,
-      "loss": 5.6029,
-      "step": 20834
-    },
-    {
-      "epoch": 10.865710560625814,
-      "grad_norm": 1.471883773803711,
-      "learning_rate": 7.956884422110553e-05,
-      "loss": 5.3606,
-      "step": 20835
-    },
-    {
-      "epoch": 10.866232073011734,
-      "grad_norm": 1.5169390439987183,
-      "learning_rate": 7.95678391959799e-05,
-      "loss": 5.4083,
-      "step": 20836
-    },
-    {
-      "epoch": 10.866753585397653,
-      "grad_norm": 1.5776190757751465,
-      "learning_rate": 7.956683417085428e-05,
-      "loss": 4.652,
-      "step": 20837
-    },
-    {
-      "epoch": 10.867275097783573,
-      "grad_norm": 1.5750374794006348,
-      "learning_rate": 7.956582914572864e-05,
-      "loss": 5.1033,
-      "step": 20838
-    },
-    {
-      "epoch": 10.867796610169492,
-      "grad_norm": 1.4520243406295776,
-      "learning_rate": 7.956482412060302e-05,
-      "loss": 5.4374,
-      "step": 20839
-    },
-    {
-      "epoch": 10.868318122555412,
-      "grad_norm": 1.58138108253479,
-      "learning_rate": 7.956381909547738e-05,
-      "loss": 5.4905,
-      "step": 20840
-    },
-    {
-      "epoch": 10.86883963494133,
-      "grad_norm": 1.485060691833496,
-      "learning_rate": 7.956281407035176e-05,
-      "loss": 5.0284,
-      "step": 20841
-    },
-    {
-      "epoch": 10.869361147327249,
-      "grad_norm": 1.5503602027893066,
-      "learning_rate": 7.956180904522614e-05,
-      "loss": 5.6503,
-      "step": 20842
-    },
-    {
-      "epoch": 10.869882659713168,
-      "grad_norm": 1.5430030822753906,
-      "learning_rate": 7.956080402010052e-05,
-      "loss": 5.7022,
-      "step": 20843
-    },
-    {
-      "epoch": 10.870404172099088,
-      "grad_norm": 1.3324098587036133,
-      "learning_rate": 7.955979899497488e-05,
-      "loss": 5.6439,
-      "step": 20844
-    },
-    {
-      "epoch": 10.870925684485007,
-      "grad_norm": 1.431255578994751,
-      "learning_rate": 7.955879396984926e-05,
-      "loss": 5.677,
-      "step": 20845
-    },
-    {
-      "epoch": 10.871447196870925,
-      "grad_norm": 1.4600403308868408,
-      "learning_rate": 7.955778894472362e-05,
-      "loss": 5.4426,
-      "step": 20846
-    },
-    {
-      "epoch": 10.871968709256844,
-      "grad_norm": 1.5493687391281128,
-      "learning_rate": 7.9556783919598e-05,
-      "loss": 5.1807,
-      "step": 20847
-    },
-    {
-      "epoch": 10.872490221642764,
-      "grad_norm": 1.4435315132141113,
-      "learning_rate": 7.955577889447236e-05,
-      "loss": 5.6907,
-      "step": 20848
-    },
-    {
-      "epoch": 10.873011734028683,
-      "grad_norm": 1.457694411277771,
-      "learning_rate": 7.955477386934673e-05,
-      "loss": 5.3472,
-      "step": 20849
-    },
-    {
-      "epoch": 10.873533246414603,
-      "grad_norm": 1.5176588296890259,
-      "learning_rate": 7.95537688442211e-05,
-      "loss": 5.7104,
-      "step": 20850
-    },
-    {
-      "epoch": 10.874054758800522,
-      "grad_norm": 1.5433452129364014,
-      "learning_rate": 7.955276381909548e-05,
-      "loss": 5.6314,
-      "step": 20851
-    },
-    {
-      "epoch": 10.874576271186442,
-      "grad_norm": 1.4065159559249878,
-      "learning_rate": 7.955175879396986e-05,
-      "loss": 4.8043,
-      "step": 20852
-    },
-    {
-      "epoch": 10.87509778357236,
-      "grad_norm": 1.4746954441070557,
-      "learning_rate": 7.955075376884423e-05,
-      "loss": 4.7756,
-      "step": 20853
-    },
-    {
-      "epoch": 10.875619295958279,
-      "grad_norm": 1.5058845281600952,
-      "learning_rate": 7.95497487437186e-05,
-      "loss": 5.355,
-      "step": 20854
-    },
-    {
-      "epoch": 10.876140808344198,
-      "grad_norm": 1.4877228736877441,
-      "learning_rate": 7.954874371859297e-05,
-      "loss": 5.1562,
-      "step": 20855
-    },
-    {
-      "epoch": 10.876662320730118,
-      "grad_norm": 1.4097403287887573,
-      "learning_rate": 7.954773869346735e-05,
-      "loss": 5.624,
-      "step": 20856
-    },
-    {
-      "epoch": 10.877183833116037,
-      "grad_norm": 1.5058139562606812,
-      "learning_rate": 7.954673366834171e-05,
-      "loss": 5.3751,
-      "step": 20857
-    },
-    {
-      "epoch": 10.877705345501955,
-      "grad_norm": 1.3550471067428589,
-      "learning_rate": 7.954572864321609e-05,
-      "loss": 5.7129,
-      "step": 20858
-    },
-    {
-      "epoch": 10.878226857887874,
-      "grad_norm": 1.3287475109100342,
-      "learning_rate": 7.954472361809045e-05,
-      "loss": 5.6979,
-      "step": 20859
-    },
-    {
-      "epoch": 10.878748370273794,
-      "grad_norm": 1.5471543073654175,
-      "learning_rate": 7.954371859296483e-05,
-      "loss": 5.3526,
-      "step": 20860
-    },
-    {
-      "epoch": 10.879269882659713,
-      "grad_norm": 1.6042343378067017,
-      "learning_rate": 7.95427135678392e-05,
-      "loss": 5.1967,
-      "step": 20861
-    },
-    {
-      "epoch": 10.879791395045633,
-      "grad_norm": 1.4433425664901733,
-      "learning_rate": 7.954170854271357e-05,
-      "loss": 5.5429,
-      "step": 20862
-    },
-    {
-      "epoch": 10.880312907431552,
-      "grad_norm": 1.4447836875915527,
-      "learning_rate": 7.954070351758795e-05,
-      "loss": 5.3748,
-      "step": 20863
-    },
-    {
-      "epoch": 10.880834419817472,
-      "grad_norm": 1.4456778764724731,
-      "learning_rate": 7.953969849246231e-05,
-      "loss": 5.4958,
-      "step": 20864
-    },
-    {
-      "epoch": 10.88135593220339,
-      "grad_norm": 1.4391014575958252,
-      "learning_rate": 7.953869346733669e-05,
-      "loss": 5.2561,
-      "step": 20865
-    },
-    {
-      "epoch": 10.881877444589309,
-      "grad_norm": 1.4296648502349854,
-      "learning_rate": 7.953768844221106e-05,
-      "loss": 5.7641,
-      "step": 20866
-    },
-    {
-      "epoch": 10.882398956975228,
-      "grad_norm": 1.4551854133605957,
-      "learning_rate": 7.953668341708543e-05,
-      "loss": 5.6885,
-      "step": 20867
-    },
-    {
-      "epoch": 10.882920469361148,
-      "grad_norm": 1.3159570693969727,
-      "learning_rate": 7.95356783919598e-05,
-      "loss": 5.3096,
-      "step": 20868
-    },
-    {
-      "epoch": 10.883441981747067,
-      "grad_norm": 1.5035128593444824,
-      "learning_rate": 7.953467336683418e-05,
-      "loss": 5.4712,
-      "step": 20869
-    },
-    {
-      "epoch": 10.883963494132985,
-      "grad_norm": 1.434493064880371,
-      "learning_rate": 7.953366834170854e-05,
-      "loss": 5.5397,
-      "step": 20870
-    },
-    {
-      "epoch": 10.884485006518904,
-      "grad_norm": 1.407245397567749,
-      "learning_rate": 7.953266331658292e-05,
-      "loss": 5.2874,
-      "step": 20871
-    },
-    {
-      "epoch": 10.885006518904824,
-      "grad_norm": 1.615294098854065,
-      "learning_rate": 7.95316582914573e-05,
-      "loss": 4.9579,
-      "step": 20872
-    },
-    {
-      "epoch": 10.885528031290743,
-      "grad_norm": 1.3720768690109253,
-      "learning_rate": 7.953065326633167e-05,
-      "loss": 5.9315,
-      "step": 20873
-    },
-    {
-      "epoch": 10.886049543676663,
-      "grad_norm": 1.5831645727157593,
-      "learning_rate": 7.952964824120604e-05,
-      "loss": 5.576,
-      "step": 20874
-    },
-    {
-      "epoch": 10.886571056062582,
-      "grad_norm": 1.3927009105682373,
-      "learning_rate": 7.95286432160804e-05,
-      "loss": 5.8921,
-      "step": 20875
-    },
-    {
-      "epoch": 10.887092568448502,
-      "grad_norm": 1.322348713874817,
-      "learning_rate": 7.952763819095478e-05,
-      "loss": 5.7025,
-      "step": 20876
-    },
-    {
-      "epoch": 10.88761408083442,
-      "grad_norm": 1.3513325452804565,
-      "learning_rate": 7.952663316582914e-05,
-      "loss": 5.6977,
-      "step": 20877
-    },
-    {
-      "epoch": 10.888135593220339,
-      "grad_norm": 1.3339743614196777,
-      "learning_rate": 7.952562814070352e-05,
-      "loss": 5.5695,
-      "step": 20878
-    },
-    {
-      "epoch": 10.888657105606258,
-      "grad_norm": 1.3675743341445923,
-      "learning_rate": 7.952462311557789e-05,
-      "loss": 5.7272,
-      "step": 20879
-    },
-    {
-      "epoch": 10.889178617992178,
-      "grad_norm": 1.5403809547424316,
-      "learning_rate": 7.952361809045226e-05,
-      "loss": 5.5248,
-      "step": 20880
-    },
-    {
-      "epoch": 10.889700130378097,
-      "grad_norm": 1.492254614830017,
-      "learning_rate": 7.952261306532663e-05,
-      "loss": 5.5498,
-      "step": 20881
-    },
-    {
-      "epoch": 10.890221642764015,
-      "grad_norm": 1.4785937070846558,
-      "learning_rate": 7.9521608040201e-05,
-      "loss": 5.3354,
-      "step": 20882
-    },
-    {
-      "epoch": 10.890743155149934,
-      "grad_norm": 1.3871748447418213,
-      "learning_rate": 7.952060301507538e-05,
-      "loss": 4.6877,
-      "step": 20883
-    },
-    {
-      "epoch": 10.891264667535854,
-      "grad_norm": 1.498595118522644,
-      "learning_rate": 7.951959798994976e-05,
-      "loss": 5.508,
-      "step": 20884
-    },
-    {
-      "epoch": 10.891786179921773,
-      "grad_norm": 1.4056581258773804,
-      "learning_rate": 7.951859296482413e-05,
-      "loss": 5.9151,
-      "step": 20885
-    },
-    {
-      "epoch": 10.892307692307693,
-      "grad_norm": 1.5041513442993164,
-      "learning_rate": 7.95175879396985e-05,
-      "loss": 5.442,
-      "step": 20886
-    },
-    {
-      "epoch": 10.892829204693612,
-      "grad_norm": 1.4165537357330322,
-      "learning_rate": 7.951658291457287e-05,
-      "loss": 5.3976,
-      "step": 20887
-    },
-    {
-      "epoch": 10.89335071707953,
-      "grad_norm": 1.4203828573226929,
-      "learning_rate": 7.951557788944723e-05,
-      "loss": 5.6919,
-      "step": 20888
-    },
-    {
-      "epoch": 10.89387222946545,
-      "grad_norm": 1.4517167806625366,
-      "learning_rate": 7.951457286432161e-05,
-      "loss": 5.6548,
-      "step": 20889
-    },
-    {
-      "epoch": 10.894393741851369,
-      "grad_norm": 1.4476011991500854,
-      "learning_rate": 7.951356783919598e-05,
-      "loss": 4.6524,
-      "step": 20890
-    },
-    {
-      "epoch": 10.894915254237288,
-      "grad_norm": 1.422726035118103,
-      "learning_rate": 7.951256281407035e-05,
-      "loss": 5.8377,
-      "step": 20891
-    },
-    {
-      "epoch": 10.895436766623208,
-      "grad_norm": 1.3993656635284424,
-      "learning_rate": 7.951155778894472e-05,
-      "loss": 5.8051,
-      "step": 20892
-    },
-    {
-      "epoch": 10.895958279009127,
-      "grad_norm": 1.452288031578064,
-      "learning_rate": 7.95105527638191e-05,
-      "loss": 5.11,
-      "step": 20893
-    },
-    {
-      "epoch": 10.896479791395045,
-      "grad_norm": 1.5393177270889282,
-      "learning_rate": 7.950954773869347e-05,
-      "loss": 5.5751,
-      "step": 20894
-    },
-    {
-      "epoch": 10.897001303780964,
-      "grad_norm": 1.4200940132141113,
-      "learning_rate": 7.950854271356785e-05,
-      "loss": 5.9156,
-      "step": 20895
-    },
-    {
-      "epoch": 10.897522816166884,
-      "grad_norm": 1.4003561735153198,
-      "learning_rate": 7.950753768844221e-05,
-      "loss": 5.3178,
-      "step": 20896
-    },
-    {
-      "epoch": 10.898044328552803,
-      "grad_norm": 1.364443063735962,
-      "learning_rate": 7.950653266331659e-05,
-      "loss": 5.7249,
-      "step": 20897
-    },
-    {
-      "epoch": 10.898565840938723,
-      "grad_norm": 1.5462408065795898,
-      "learning_rate": 7.950552763819096e-05,
-      "loss": 5.4026,
-      "step": 20898
-    },
-    {
-      "epoch": 10.899087353324642,
-      "grad_norm": 1.6502461433410645,
-      "learning_rate": 7.950452261306533e-05,
-      "loss": 5.2963,
-      "step": 20899
-    },
-    {
-      "epoch": 10.89960886571056,
-      "grad_norm": 1.6449497938156128,
-      "learning_rate": 7.95035175879397e-05,
-      "loss": 4.8947,
-      "step": 20900
-    },
-    {
-      "epoch": 10.90013037809648,
-      "grad_norm": 1.5031843185424805,
-      "learning_rate": 7.950251256281406e-05,
-      "loss": 4.8982,
-      "step": 20901
-    },
-    {
-      "epoch": 10.900651890482399,
-      "grad_norm": 1.5747390985488892,
-      "learning_rate": 7.950150753768844e-05,
-      "loss": 5.1242,
-      "step": 20902
-    },
-    {
-      "epoch": 10.901173402868318,
-      "grad_norm": 1.3169786930084229,
-      "learning_rate": 7.950050251256282e-05,
-      "loss": 5.9117,
-      "step": 20903
-    },
-    {
-      "epoch": 10.901694915254238,
-      "grad_norm": 1.370862364768982,
-      "learning_rate": 7.94994974874372e-05,
-      "loss": 5.6626,
-      "step": 20904
-    },
-    {
-      "epoch": 10.902216427640157,
-      "grad_norm": 1.5698710680007935,
-      "learning_rate": 7.949849246231156e-05,
-      "loss": 5.2537,
-      "step": 20905
-    },
-    {
-      "epoch": 10.902737940026075,
-      "grad_norm": 1.4392200708389282,
-      "learning_rate": 7.949748743718594e-05,
-      "loss": 4.9474,
-      "step": 20906
-    },
-    {
-      "epoch": 10.903259452411994,
-      "grad_norm": 1.4793663024902344,
-      "learning_rate": 7.94964824120603e-05,
-      "loss": 5.1674,
-      "step": 20907
-    },
-    {
-      "epoch": 10.903780964797914,
-      "grad_norm": 1.4574165344238281,
-      "learning_rate": 7.949547738693468e-05,
-      "loss": 5.4334,
-      "step": 20908
-    },
-    {
-      "epoch": 10.904302477183833,
-      "grad_norm": 1.6038566827774048,
-      "learning_rate": 7.949447236180905e-05,
-      "loss": 4.7896,
-      "step": 20909
-    },
-    {
-      "epoch": 10.904823989569753,
-      "grad_norm": 1.488199234008789,
-      "learning_rate": 7.949346733668342e-05,
-      "loss": 5.9032,
-      "step": 20910
-    },
-    {
-      "epoch": 10.905345501955672,
-      "grad_norm": 1.4139366149902344,
-      "learning_rate": 7.949246231155779e-05,
-      "loss": 5.6636,
-      "step": 20911
-    },
-    {
-      "epoch": 10.90586701434159,
-      "grad_norm": 1.4479649066925049,
-      "learning_rate": 7.949145728643217e-05,
-      "loss": 5.0333,
-      "step": 20912
-    },
-    {
-      "epoch": 10.90638852672751,
-      "grad_norm": 1.4658572673797607,
-      "learning_rate": 7.949045226130653e-05,
-      "loss": 5.4264,
-      "step": 20913
-    },
-    {
-      "epoch": 10.906910039113429,
-      "grad_norm": 1.392055630683899,
-      "learning_rate": 7.948944723618091e-05,
-      "loss": 5.6167,
-      "step": 20914
-    },
-    {
-      "epoch": 10.907431551499348,
-      "grad_norm": 1.4329018592834473,
-      "learning_rate": 7.948844221105529e-05,
-      "loss": 5.7922,
-      "step": 20915
-    },
-    {
-      "epoch": 10.907953063885268,
-      "grad_norm": 1.544110894203186,
-      "learning_rate": 7.948743718592965e-05,
-      "loss": 5.4387,
-      "step": 20916
-    },
-    {
-      "epoch": 10.908474576271187,
-      "grad_norm": 1.6874760389328003,
-      "learning_rate": 7.948643216080403e-05,
-      "loss": 5.0029,
-      "step": 20917
-    },
-    {
-      "epoch": 10.908996088657105,
-      "grad_norm": 1.4808834791183472,
-      "learning_rate": 7.948542713567839e-05,
-      "loss": 5.2479,
-      "step": 20918
-    },
-    {
-      "epoch": 10.909517601043024,
-      "grad_norm": 1.557495355606079,
-      "learning_rate": 7.948442211055277e-05,
-      "loss": 5.0523,
-      "step": 20919
-    },
-    {
-      "epoch": 10.910039113428944,
-      "grad_norm": 1.4027904272079468,
-      "learning_rate": 7.948341708542713e-05,
-      "loss": 5.4982,
-      "step": 20920
-    },
-    {
-      "epoch": 10.910560625814863,
-      "grad_norm": 1.418133020401001,
-      "learning_rate": 7.948241206030151e-05,
-      "loss": 5.3878,
-      "step": 20921
-    },
-    {
-      "epoch": 10.911082138200783,
-      "grad_norm": 1.389695644378662,
-      "learning_rate": 7.948140703517588e-05,
-      "loss": 5.5336,
-      "step": 20922
-    },
-    {
-      "epoch": 10.911603650586702,
-      "grad_norm": 1.4180316925048828,
-      "learning_rate": 7.948040201005025e-05,
-      "loss": 5.6098,
-      "step": 20923
-    },
-    {
-      "epoch": 10.91212516297262,
-      "grad_norm": 1.5257996320724487,
-      "learning_rate": 7.947939698492463e-05,
-      "loss": 5.4964,
-      "step": 20924
-    },
-    {
-      "epoch": 10.91264667535854,
-      "grad_norm": 1.4572893381118774,
-      "learning_rate": 7.947839195979901e-05,
-      "loss": 5.2873,
-      "step": 20925
-    },
-    {
-      "epoch": 10.913168187744459,
-      "grad_norm": 1.4000223875045776,
-      "learning_rate": 7.947738693467337e-05,
-      "loss": 5.5371,
-      "step": 20926
-    },
-    {
-      "epoch": 10.913689700130378,
-      "grad_norm": 1.5175187587738037,
-      "learning_rate": 7.947638190954775e-05,
-      "loss": 5.7718,
-      "step": 20927
-    },
-    {
-      "epoch": 10.914211212516298,
-      "grad_norm": 1.417612075805664,
-      "learning_rate": 7.947537688442212e-05,
-      "loss": 5.4433,
-      "step": 20928
-    },
-    {
-      "epoch": 10.914732724902217,
-      "grad_norm": 1.425512433052063,
-      "learning_rate": 7.947437185929648e-05,
-      "loss": 5.0224,
-      "step": 20929
-    },
-    {
-      "epoch": 10.915254237288135,
-      "grad_norm": 1.4505774974822998,
-      "learning_rate": 7.947336683417086e-05,
-      "loss": 5.4547,
-      "step": 20930
-    },
-    {
-      "epoch": 10.915775749674054,
-      "grad_norm": 1.5449849367141724,
-      "learning_rate": 7.947236180904522e-05,
-      "loss": 5.2426,
-      "step": 20931
-    },
-    {
-      "epoch": 10.916297262059974,
-      "grad_norm": 1.5006132125854492,
-      "learning_rate": 7.94713567839196e-05,
-      "loss": 5.3141,
-      "step": 20932
-    },
-    {
-      "epoch": 10.916818774445893,
-      "grad_norm": 1.278379201889038,
-      "learning_rate": 7.947035175879396e-05,
-      "loss": 5.9952,
-      "step": 20933
-    },
-    {
-      "epoch": 10.917340286831813,
-      "grad_norm": 1.4494935274124146,
-      "learning_rate": 7.946934673366834e-05,
-      "loss": 5.2735,
-      "step": 20934
-    },
-    {
-      "epoch": 10.917861799217732,
-      "grad_norm": 1.44352126121521,
-      "learning_rate": 7.946834170854272e-05,
-      "loss": 5.4757,
-      "step": 20935
-    },
-    {
-      "epoch": 10.91838331160365,
-      "grad_norm": 1.499895691871643,
-      "learning_rate": 7.94673366834171e-05,
-      "loss": 5.5513,
-      "step": 20936
-    },
-    {
-      "epoch": 10.91890482398957,
-      "grad_norm": 1.463358998298645,
-      "learning_rate": 7.946633165829146e-05,
-      "loss": 5.6771,
-      "step": 20937
-    },
-    {
-      "epoch": 10.919426336375489,
-      "grad_norm": 1.6340488195419312,
-      "learning_rate": 7.946532663316584e-05,
-      "loss": 5.2406,
-      "step": 20938
-    },
-    {
-      "epoch": 10.919947848761408,
-      "grad_norm": 1.5814470052719116,
-      "learning_rate": 7.94643216080402e-05,
-      "loss": 5.4777,
-      "step": 20939
-    },
-    {
-      "epoch": 10.920469361147328,
-      "grad_norm": 1.531904935836792,
-      "learning_rate": 7.946331658291458e-05,
-      "loss": 5.5558,
-      "step": 20940
-    },
-    {
-      "epoch": 10.920990873533245,
-      "grad_norm": 1.4873367547988892,
-      "learning_rate": 7.946231155778895e-05,
-      "loss": 5.706,
-      "step": 20941
-    },
-    {
-      "epoch": 10.921512385919165,
-      "grad_norm": 1.4803050756454468,
-      "learning_rate": 7.946130653266331e-05,
-      "loss": 4.9068,
-      "step": 20942
-    },
-    {
-      "epoch": 10.922033898305084,
-      "grad_norm": 1.4614310264587402,
-      "learning_rate": 7.946030150753769e-05,
-      "loss": 5.6866,
-      "step": 20943
-    },
-    {
-      "epoch": 10.922555410691004,
-      "grad_norm": 1.5398675203323364,
-      "learning_rate": 7.945929648241207e-05,
-      "loss": 5.3529,
-      "step": 20944
-    },
-    {
-      "epoch": 10.923076923076923,
-      "grad_norm": 1.3800475597381592,
-      "learning_rate": 7.945829145728644e-05,
-      "loss": 5.8213,
-      "step": 20945
-    },
-    {
-      "epoch": 10.923598435462843,
-      "grad_norm": 1.5008275508880615,
-      "learning_rate": 7.945728643216081e-05,
-      "loss": 5.3283,
-      "step": 20946
-    },
-    {
-      "epoch": 10.924119947848762,
-      "grad_norm": 1.5373327732086182,
-      "learning_rate": 7.945628140703519e-05,
-      "loss": 5.1157,
-      "step": 20947
-    },
-    {
-      "epoch": 10.92464146023468,
-      "grad_norm": 1.390774130821228,
-      "learning_rate": 7.945527638190955e-05,
-      "loss": 5.5616,
-      "step": 20948
-    },
-    {
-      "epoch": 10.9251629726206,
-      "grad_norm": 1.4165246486663818,
-      "learning_rate": 7.945427135678393e-05,
-      "loss": 5.5931,
-      "step": 20949
-    },
-    {
-      "epoch": 10.925684485006519,
-      "grad_norm": 1.4504565000534058,
-      "learning_rate": 7.945326633165829e-05,
-      "loss": 5.6173,
-      "step": 20950
-    },
-    {
-      "epoch": 10.926205997392438,
-      "grad_norm": 1.3563454151153564,
-      "learning_rate": 7.945226130653267e-05,
-      "loss": 5.6162,
-      "step": 20951
-    },
-    {
-      "epoch": 10.926727509778358,
-      "grad_norm": 1.5011118650436401,
-      "learning_rate": 7.945125628140703e-05,
-      "loss": 4.8219,
-      "step": 20952
-    },
-    {
-      "epoch": 10.927249022164276,
-      "grad_norm": 1.4267215728759766,
-      "learning_rate": 7.945025125628141e-05,
-      "loss": 5.5383,
-      "step": 20953
-    },
-    {
-      "epoch": 10.927770534550195,
-      "grad_norm": 1.4159268140792847,
-      "learning_rate": 7.944924623115578e-05,
-      "loss": 5.2531,
-      "step": 20954
-    },
-    {
-      "epoch": 10.928292046936114,
-      "grad_norm": 1.3717079162597656,
-      "learning_rate": 7.944824120603015e-05,
-      "loss": 4.8347,
-      "step": 20955
-    },
-    {
-      "epoch": 10.928813559322034,
-      "grad_norm": 1.497125267982483,
-      "learning_rate": 7.944723618090453e-05,
-      "loss": 5.2158,
-      "step": 20956
-    },
-    {
-      "epoch": 10.929335071707953,
-      "grad_norm": 1.5739015340805054,
-      "learning_rate": 7.94462311557789e-05,
-      "loss": 5.1905,
-      "step": 20957
-    },
-    {
-      "epoch": 10.929856584093873,
-      "grad_norm": 1.4662703275680542,
-      "learning_rate": 7.944522613065327e-05,
-      "loss": 5.461,
-      "step": 20958
-    },
-    {
-      "epoch": 10.930378096479792,
-      "grad_norm": 1.454738736152649,
-      "learning_rate": 7.944422110552764e-05,
-      "loss": 5.5569,
-      "step": 20959
-    },
-    {
-      "epoch": 10.93089960886571,
-      "grad_norm": 1.4483228921890259,
-      "learning_rate": 7.944321608040202e-05,
-      "loss": 5.5923,
-      "step": 20960
-    },
-    {
-      "epoch": 10.93142112125163,
-      "grad_norm": 1.5428428649902344,
-      "learning_rate": 7.944221105527638e-05,
-      "loss": 4.9053,
-      "step": 20961
-    },
-    {
-      "epoch": 10.931942633637549,
-      "grad_norm": 1.9260282516479492,
-      "learning_rate": 7.944120603015076e-05,
-      "loss": 4.7062,
-      "step": 20962
-    },
-    {
-      "epoch": 10.932464146023468,
-      "grad_norm": 1.4864997863769531,
-      "learning_rate": 7.944020100502512e-05,
-      "loss": 5.4357,
-      "step": 20963
-    },
-    {
-      "epoch": 10.932985658409388,
-      "grad_norm": 1.4165695905685425,
-      "learning_rate": 7.94391959798995e-05,
-      "loss": 5.0916,
-      "step": 20964
-    },
-    {
-      "epoch": 10.933507170795306,
-      "grad_norm": 1.4304850101470947,
-      "learning_rate": 7.943819095477388e-05,
-      "loss": 5.6316,
-      "step": 20965
-    },
-    {
-      "epoch": 10.934028683181225,
-      "grad_norm": 1.425891637802124,
-      "learning_rate": 7.943718592964826e-05,
-      "loss": 5.6791,
-      "step": 20966
-    },
-    {
-      "epoch": 10.934550195567144,
-      "grad_norm": 1.4568068981170654,
-      "learning_rate": 7.943618090452262e-05,
-      "loss": 5.1236,
-      "step": 20967
-    },
-    {
-      "epoch": 10.935071707953064,
-      "grad_norm": 1.5454902648925781,
-      "learning_rate": 7.943517587939698e-05,
-      "loss": 5.291,
-      "step": 20968
-    },
-    {
-      "epoch": 10.935593220338983,
-      "grad_norm": 1.5041025876998901,
-      "learning_rate": 7.943417085427136e-05,
-      "loss": 5.4392,
-      "step": 20969
-    },
-    {
-      "epoch": 10.936114732724903,
-      "grad_norm": 1.5437592267990112,
-      "learning_rate": 7.943316582914573e-05,
-      "loss": 5.0549,
-      "step": 20970
-    },
-    {
-      "epoch": 10.93663624511082,
-      "grad_norm": 1.3061270713806152,
-      "learning_rate": 7.94321608040201e-05,
-      "loss": 5.7704,
-      "step": 20971
-    },
-    {
-      "epoch": 10.93715775749674,
-      "grad_norm": 1.4510531425476074,
-      "learning_rate": 7.943115577889447e-05,
-      "loss": 5.2776,
-      "step": 20972
-    },
-    {
-      "epoch": 10.93767926988266,
-      "grad_norm": 1.5836305618286133,
-      "learning_rate": 7.943015075376885e-05,
-      "loss": 4.92,
-      "step": 20973
-    },
-    {
-      "epoch": 10.938200782268579,
-      "grad_norm": 1.3863914012908936,
-      "learning_rate": 7.942914572864321e-05,
-      "loss": 5.9593,
-      "step": 20974
-    },
-    {
-      "epoch": 10.938722294654498,
-      "grad_norm": 1.3610464334487915,
-      "learning_rate": 7.942814070351759e-05,
-      "loss": 5.1472,
-      "step": 20975
-    },
-    {
-      "epoch": 10.939243807040418,
-      "grad_norm": 1.3911867141723633,
-      "learning_rate": 7.942713567839197e-05,
-      "loss": 4.6204,
-      "step": 20976
-    },
-    {
-      "epoch": 10.939765319426336,
-      "grad_norm": 1.3874753713607788,
-      "learning_rate": 7.942613065326634e-05,
-      "loss": 5.3447,
-      "step": 20977
-    },
-    {
-      "epoch": 10.940286831812255,
-      "grad_norm": 1.3914989233016968,
-      "learning_rate": 7.942512562814071e-05,
-      "loss": 5.4827,
-      "step": 20978
-    },
-    {
-      "epoch": 10.940808344198174,
-      "grad_norm": 1.5483626127243042,
-      "learning_rate": 7.942412060301509e-05,
-      "loss": 5.237,
-      "step": 20979
-    },
-    {
-      "epoch": 10.941329856584094,
-      "grad_norm": 1.450799584388733,
-      "learning_rate": 7.942311557788945e-05,
-      "loss": 5.4829,
-      "step": 20980
-    },
-    {
-      "epoch": 10.941851368970013,
-      "grad_norm": 1.453369140625,
-      "learning_rate": 7.942211055276382e-05,
-      "loss": 5.4678,
-      "step": 20981
-    },
-    {
-      "epoch": 10.942372881355933,
-      "grad_norm": 1.4180506467819214,
-      "learning_rate": 7.942110552763819e-05,
-      "loss": 5.3243,
-      "step": 20982
-    },
-    {
-      "epoch": 10.94289439374185,
-      "grad_norm": 1.4748303890228271,
-      "learning_rate": 7.942010050251256e-05,
-      "loss": 5.924,
-      "step": 20983
-    },
-    {
-      "epoch": 10.94341590612777,
-      "grad_norm": 1.5197535753250122,
-      "learning_rate": 7.941909547738694e-05,
-      "loss": 5.5981,
-      "step": 20984
-    },
-    {
-      "epoch": 10.94393741851369,
-      "grad_norm": 1.5293771028518677,
-      "learning_rate": 7.941809045226131e-05,
-      "loss": 5.5658,
-      "step": 20985
-    },
-    {
-      "epoch": 10.944458930899609,
-      "grad_norm": 1.4990962743759155,
-      "learning_rate": 7.941708542713569e-05,
-      "loss": 5.4771,
-      "step": 20986
-    },
-    {
-      "epoch": 10.944980443285528,
-      "grad_norm": 1.3228641748428345,
-      "learning_rate": 7.941608040201006e-05,
-      "loss": 5.6902,
-      "step": 20987
-    },
-    {
-      "epoch": 10.945501955671448,
-      "grad_norm": 1.2924830913543701,
-      "learning_rate": 7.941507537688443e-05,
-      "loss": 5.7745,
-      "step": 20988
-    },
-    {
-      "epoch": 10.946023468057366,
-      "grad_norm": 1.3878355026245117,
-      "learning_rate": 7.94140703517588e-05,
-      "loss": 5.1028,
-      "step": 20989
-    },
-    {
-      "epoch": 10.946544980443285,
-      "grad_norm": 1.5296218395233154,
-      "learning_rate": 7.941306532663318e-05,
-      "loss": 4.9931,
-      "step": 20990
-    },
-    {
-      "epoch": 10.947066492829205,
-      "grad_norm": 1.515978217124939,
-      "learning_rate": 7.941206030150754e-05,
-      "loss": 5.6961,
-      "step": 20991
-    },
-    {
-      "epoch": 10.947588005215124,
-      "grad_norm": 1.5519098043441772,
-      "learning_rate": 7.941105527638192e-05,
-      "loss": 5.2795,
-      "step": 20992
-    },
-    {
-      "epoch": 10.948109517601043,
-      "grad_norm": 1.4744489192962646,
-      "learning_rate": 7.941005025125628e-05,
-      "loss": 5.1416,
-      "step": 20993
-    },
-    {
-      "epoch": 10.948631029986963,
-      "grad_norm": 1.4355180263519287,
-      "learning_rate": 7.940904522613065e-05,
-      "loss": 5.584,
-      "step": 20994
-    },
-    {
-      "epoch": 10.94915254237288,
-      "grad_norm": 1.4336366653442383,
-      "learning_rate": 7.940804020100502e-05,
-      "loss": 5.462,
-      "step": 20995
-    },
-    {
-      "epoch": 10.9496740547588,
-      "grad_norm": 1.4444173574447632,
-      "learning_rate": 7.94070351758794e-05,
-      "loss": 4.9244,
-      "step": 20996
-    },
-    {
-      "epoch": 10.95019556714472,
-      "grad_norm": 1.5954136848449707,
-      "learning_rate": 7.940603015075378e-05,
-      "loss": 5.3825,
-      "step": 20997
-    },
-    {
-      "epoch": 10.950717079530639,
-      "grad_norm": 1.4889582395553589,
-      "learning_rate": 7.940502512562814e-05,
-      "loss": 5.1979,
-      "step": 20998
-    },
-    {
-      "epoch": 10.951238591916558,
-      "grad_norm": 1.3851956129074097,
-      "learning_rate": 7.940402010050252e-05,
-      "loss": 5.6736,
-      "step": 20999
-    },
-    {
-      "epoch": 10.951760104302478,
-      "grad_norm": 1.3554573059082031,
-      "learning_rate": 7.940301507537689e-05,
-      "loss": 5.8085,
-      "step": 21000
-    },
-    {
-      "epoch": 10.952281616688396,
-      "grad_norm": 1.399256706237793,
-      "learning_rate": 7.940201005025126e-05,
-      "loss": 5.441,
-      "step": 21001
-    },
-    {
-      "epoch": 10.952803129074315,
-      "grad_norm": 1.4235142469406128,
-      "learning_rate": 7.940100502512563e-05,
-      "loss": 5.0651,
-      "step": 21002
-    },
-    {
-      "epoch": 10.953324641460235,
-      "grad_norm": 1.4815534353256226,
-      "learning_rate": 7.94e-05,
-      "loss": 5.1548,
-      "step": 21003
-    },
-    {
-      "epoch": 10.953846153846154,
-      "grad_norm": 1.3943684101104736,
-      "learning_rate": 7.939899497487437e-05,
-      "loss": 5.4679,
-      "step": 21004
-    },
-    {
-      "epoch": 10.954367666232073,
-      "grad_norm": 1.4294825792312622,
-      "learning_rate": 7.939798994974875e-05,
-      "loss": 5.1398,
-      "step": 21005
-    },
-    {
-      "epoch": 10.954889178617993,
-      "grad_norm": 1.4298815727233887,
-      "learning_rate": 7.939698492462313e-05,
-      "loss": 5.4605,
-      "step": 21006
-    },
-    {
-      "epoch": 10.95541069100391,
-      "grad_norm": 1.507045030593872,
-      "learning_rate": 7.939597989949749e-05,
-      "loss": 5.2197,
-      "step": 21007
-    },
-    {
-      "epoch": 10.95593220338983,
-      "grad_norm": 1.3973795175552368,
-      "learning_rate": 7.939497487437187e-05,
-      "loss": 5.7279,
-      "step": 21008
-    },
-    {
-      "epoch": 10.95645371577575,
-      "grad_norm": 1.367243766784668,
-      "learning_rate": 7.939396984924623e-05,
-      "loss": 5.5702,
-      "step": 21009
-    },
-    {
-      "epoch": 10.956975228161669,
-      "grad_norm": 1.3824946880340576,
-      "learning_rate": 7.939296482412061e-05,
-      "loss": 5.4399,
-      "step": 21010
-    },
-    {
-      "epoch": 10.957496740547588,
-      "grad_norm": 1.3645747900009155,
-      "learning_rate": 7.939195979899497e-05,
-      "loss": 5.673,
-      "step": 21011
-    },
-    {
-      "epoch": 10.958018252933508,
-      "grad_norm": 1.4059242010116577,
-      "learning_rate": 7.939095477386935e-05,
-      "loss": 5.4967,
-      "step": 21012
-    },
-    {
-      "epoch": 10.958539765319426,
-      "grad_norm": 1.4147907495498657,
-      "learning_rate": 7.938994974874372e-05,
-      "loss": 5.2812,
-      "step": 21013
-    },
-    {
-      "epoch": 10.959061277705345,
-      "grad_norm": 1.4081979990005493,
-      "learning_rate": 7.93889447236181e-05,
-      "loss": 5.7041,
-      "step": 21014
-    },
-    {
-      "epoch": 10.959582790091265,
-      "grad_norm": 1.5151044130325317,
-      "learning_rate": 7.938793969849246e-05,
-      "loss": 5.2227,
-      "step": 21015
-    },
-    {
-      "epoch": 10.960104302477184,
-      "grad_norm": 1.4338107109069824,
-      "learning_rate": 7.938693467336684e-05,
-      "loss": 4.716,
-      "step": 21016
-    },
-    {
-      "epoch": 10.960625814863103,
-      "grad_norm": 1.4353386163711548,
-      "learning_rate": 7.938592964824121e-05,
-      "loss": 5.8658,
-      "step": 21017
-    },
-    {
-      "epoch": 10.961147327249023,
-      "grad_norm": 1.5235562324523926,
-      "learning_rate": 7.938492462311559e-05,
-      "loss": 5.2962,
-      "step": 21018
-    },
-    {
-      "epoch": 10.96166883963494,
-      "grad_norm": 1.4557902812957764,
-      "learning_rate": 7.938391959798996e-05,
-      "loss": 5.1995,
-      "step": 21019
-    },
-    {
-      "epoch": 10.96219035202086,
-      "grad_norm": 1.4175407886505127,
-      "learning_rate": 7.938291457286433e-05,
-      "loss": 5.8074,
-      "step": 21020
-    },
-    {
-      "epoch": 10.96271186440678,
-      "grad_norm": 1.4482581615447998,
-      "learning_rate": 7.93819095477387e-05,
-      "loss": 5.7119,
-      "step": 21021
-    },
-    {
-      "epoch": 10.963233376792699,
-      "grad_norm": 1.4694383144378662,
-      "learning_rate": 7.938090452261306e-05,
-      "loss": 5.5255,
-      "step": 21022
-    },
-    {
-      "epoch": 10.963754889178619,
-      "grad_norm": 1.3603373765945435,
-      "learning_rate": 7.937989949748744e-05,
-      "loss": 5.6891,
-      "step": 21023
-    },
-    {
-      "epoch": 10.964276401564538,
-      "grad_norm": 1.5070805549621582,
-      "learning_rate": 7.93788944723618e-05,
-      "loss": 5.3614,
-      "step": 21024
-    },
-    {
-      "epoch": 10.964797913950456,
-      "grad_norm": 1.6231030225753784,
-      "learning_rate": 7.937788944723618e-05,
-      "loss": 5.1995,
-      "step": 21025
-    },
-    {
-      "epoch": 10.965319426336375,
-      "grad_norm": 1.461775302886963,
-      "learning_rate": 7.937688442211056e-05,
-      "loss": 5.6276,
-      "step": 21026
-    },
-    {
-      "epoch": 10.965840938722295,
-      "grad_norm": 1.4600508213043213,
-      "learning_rate": 7.937587939698494e-05,
-      "loss": 5.4838,
-      "step": 21027
-    },
-    {
-      "epoch": 10.966362451108214,
-      "grad_norm": 1.3968610763549805,
-      "learning_rate": 7.93748743718593e-05,
-      "loss": 5.6328,
-      "step": 21028
-    },
-    {
-      "epoch": 10.966883963494134,
-      "grad_norm": 1.4032931327819824,
-      "learning_rate": 7.937386934673368e-05,
-      "loss": 5.7985,
-      "step": 21029
-    },
-    {
-      "epoch": 10.967405475880053,
-      "grad_norm": 1.3797416687011719,
-      "learning_rate": 7.937286432160804e-05,
-      "loss": 5.6328,
-      "step": 21030
-    },
-    {
-      "epoch": 10.96792698826597,
-      "grad_norm": 1.5559303760528564,
-      "learning_rate": 7.937185929648242e-05,
-      "loss": 4.9025,
-      "step": 21031
-    },
-    {
-      "epoch": 10.96844850065189,
-      "grad_norm": 1.435437560081482,
-      "learning_rate": 7.937085427135679e-05,
-      "loss": 5.6093,
-      "step": 21032
-    },
-    {
-      "epoch": 10.96897001303781,
-      "grad_norm": 1.4638452529907227,
-      "learning_rate": 7.936984924623116e-05,
-      "loss": 5.5519,
-      "step": 21033
-    },
-    {
-      "epoch": 10.969491525423729,
-      "grad_norm": 1.4011547565460205,
-      "learning_rate": 7.936884422110553e-05,
-      "loss": 5.3242,
-      "step": 21034
-    },
-    {
-      "epoch": 10.970013037809649,
-      "grad_norm": 1.4581902027130127,
-      "learning_rate": 7.936783919597989e-05,
-      "loss": 5.5338,
-      "step": 21035
-    },
-    {
-      "epoch": 10.970534550195566,
-      "grad_norm": 1.5235960483551025,
-      "learning_rate": 7.936683417085427e-05,
-      "loss": 5.732,
-      "step": 21036
-    },
-    {
-      "epoch": 10.971056062581486,
-      "grad_norm": 1.5757670402526855,
-      "learning_rate": 7.936582914572865e-05,
-      "loss": 4.8605,
-      "step": 21037
-    },
-    {
-      "epoch": 10.971577574967405,
-      "grad_norm": 1.5269334316253662,
-      "learning_rate": 7.936482412060303e-05,
-      "loss": 5.7846,
-      "step": 21038
-    },
-    {
-      "epoch": 10.972099087353325,
-      "grad_norm": 1.397024154663086,
-      "learning_rate": 7.936381909547739e-05,
-      "loss": 5.5091,
-      "step": 21039
-    },
-    {
-      "epoch": 10.972620599739244,
-      "grad_norm": 1.4778767824172974,
-      "learning_rate": 7.936281407035177e-05,
-      "loss": 5.8512,
-      "step": 21040
-    },
-    {
-      "epoch": 10.973142112125164,
-      "grad_norm": 1.529292106628418,
-      "learning_rate": 7.936180904522613e-05,
-      "loss": 5.4362,
-      "step": 21041
-    },
-    {
-      "epoch": 10.973663624511083,
-      "grad_norm": 1.7048566341400146,
-      "learning_rate": 7.936080402010051e-05,
-      "loss": 4.4969,
-      "step": 21042
-    },
-    {
-      "epoch": 10.974185136897,
-      "grad_norm": 1.5877914428710938,
-      "learning_rate": 7.935979899497487e-05,
-      "loss": 5.3893,
-      "step": 21043
-    },
-    {
-      "epoch": 10.97470664928292,
-      "grad_norm": 1.4953404664993286,
-      "learning_rate": 7.935879396984925e-05,
-      "loss": 5.1923,
-      "step": 21044
-    },
-    {
-      "epoch": 10.97522816166884,
-      "grad_norm": 1.6752102375030518,
-      "learning_rate": 7.935778894472362e-05,
-      "loss": 4.6655,
-      "step": 21045
-    },
-    {
-      "epoch": 10.975749674054759,
-      "grad_norm": 1.4276094436645508,
-      "learning_rate": 7.9356783919598e-05,
-      "loss": 5.4718,
-      "step": 21046
-    },
-    {
-      "epoch": 10.976271186440679,
-      "grad_norm": 1.4129613637924194,
-      "learning_rate": 7.935577889447237e-05,
-      "loss": 5.7898,
-      "step": 21047
-    },
-    {
-      "epoch": 10.976792698826596,
-      "grad_norm": 1.5442242622375488,
-      "learning_rate": 7.935477386934674e-05,
-      "loss": 5.1174,
-      "step": 21048
-    },
-    {
-      "epoch": 10.977314211212516,
-      "grad_norm": 1.4921859502792358,
-      "learning_rate": 7.935376884422111e-05,
-      "loss": 5.1287,
-      "step": 21049
-    },
-    {
-      "epoch": 10.977835723598435,
-      "grad_norm": 1.4914977550506592,
-      "learning_rate": 7.935276381909548e-05,
-      "loss": 5.4843,
-      "step": 21050
-    },
-    {
-      "epoch": 10.978357235984355,
-      "grad_norm": 1.5180046558380127,
-      "learning_rate": 7.935175879396986e-05,
-      "loss": 5.4365,
-      "step": 21051
-    },
-    {
-      "epoch": 10.978878748370274,
-      "grad_norm": 1.4861451387405396,
-      "learning_rate": 7.935075376884422e-05,
-      "loss": 5.2494,
-      "step": 21052
-    },
-    {
-      "epoch": 10.979400260756194,
-      "grad_norm": 1.3985316753387451,
-      "learning_rate": 7.93497487437186e-05,
-      "loss": 5.7148,
-      "step": 21053
-    },
-    {
-      "epoch": 10.979921773142113,
-      "grad_norm": 1.4998106956481934,
-      "learning_rate": 7.934874371859296e-05,
-      "loss": 5.1073,
-      "step": 21054
-    },
-    {
-      "epoch": 10.98044328552803,
-      "grad_norm": 1.4142663478851318,
-      "learning_rate": 7.934773869346734e-05,
-      "loss": 5.8018,
-      "step": 21055
-    },
-    {
-      "epoch": 10.98096479791395,
-      "grad_norm": 1.4890509843826294,
-      "learning_rate": 7.93467336683417e-05,
-      "loss": 5.4453,
-      "step": 21056
-    },
-    {
-      "epoch": 10.98148631029987,
-      "grad_norm": 1.448545217514038,
-      "learning_rate": 7.934572864321608e-05,
-      "loss": 5.57,
-      "step": 21057
-    },
-    {
-      "epoch": 10.98200782268579,
-      "grad_norm": 1.3900812864303589,
-      "learning_rate": 7.934472361809046e-05,
-      "loss": 5.2422,
-      "step": 21058
-    },
-    {
-      "epoch": 10.982529335071709,
-      "grad_norm": 1.734815001487732,
-      "learning_rate": 7.934371859296484e-05,
-      "loss": 5.6319,
-      "step": 21059
-    },
-    {
-      "epoch": 10.983050847457626,
-      "grad_norm": 1.5178097486495972,
-      "learning_rate": 7.93427135678392e-05,
-      "loss": 5.6084,
-      "step": 21060
-    },
-    {
-      "epoch": 10.983572359843546,
-      "grad_norm": 1.6060247421264648,
-      "learning_rate": 7.934170854271357e-05,
-      "loss": 5.059,
-      "step": 21061
-    },
-    {
-      "epoch": 10.984093872229465,
-      "grad_norm": 1.590624213218689,
-      "learning_rate": 7.934070351758794e-05,
-      "loss": 5.2193,
-      "step": 21062
-    },
-    {
-      "epoch": 10.984615384615385,
-      "grad_norm": 1.4666028022766113,
-      "learning_rate": 7.933969849246231e-05,
-      "loss": 5.7239,
-      "step": 21063
-    },
-    {
-      "epoch": 10.985136897001304,
-      "grad_norm": 1.6972624063491821,
-      "learning_rate": 7.933869346733669e-05,
-      "loss": 5.0521,
-      "step": 21064
-    },
-    {
-      "epoch": 10.985658409387224,
-      "grad_norm": 1.4879679679870605,
-      "learning_rate": 7.933768844221105e-05,
-      "loss": 5.4594,
-      "step": 21065
-    },
-    {
-      "epoch": 10.986179921773141,
-      "grad_norm": 1.3531349897384644,
-      "learning_rate": 7.933668341708543e-05,
-      "loss": 5.748,
-      "step": 21066
-    },
-    {
-      "epoch": 10.98670143415906,
-      "grad_norm": 1.5164768695831299,
-      "learning_rate": 7.93356783919598e-05,
-      "loss": 5.2773,
-      "step": 21067
-    },
-    {
-      "epoch": 10.98722294654498,
-      "grad_norm": 1.4546769857406616,
-      "learning_rate": 7.933467336683417e-05,
-      "loss": 5.3853,
-      "step": 21068
-    },
-    {
-      "epoch": 10.9877444589309,
-      "grad_norm": 1.4752851724624634,
-      "learning_rate": 7.933366834170855e-05,
-      "loss": 5.2252,
-      "step": 21069
-    },
-    {
-      "epoch": 10.98826597131682,
-      "grad_norm": 1.440137267112732,
-      "learning_rate": 7.933266331658293e-05,
-      "loss": 5.1276,
-      "step": 21070
-    },
-    {
-      "epoch": 10.988787483702739,
-      "grad_norm": 1.489648699760437,
-      "learning_rate": 7.933165829145729e-05,
-      "loss": 5.4491,
-      "step": 21071
-    },
-    {
-      "epoch": 10.989308996088656,
-      "grad_norm": 1.430466890335083,
-      "learning_rate": 7.933065326633167e-05,
-      "loss": 5.4771,
-      "step": 21072
-    },
-    {
-      "epoch": 10.989830508474576,
-      "grad_norm": 1.4354274272918701,
-      "learning_rate": 7.932964824120603e-05,
-      "loss": 5.6375,
-      "step": 21073
-    },
-    {
-      "epoch": 10.990352020860495,
-      "grad_norm": 1.333548665046692,
-      "learning_rate": 7.93286432160804e-05,
-      "loss": 5.5565,
-      "step": 21074
-    },
-    {
-      "epoch": 10.990873533246415,
-      "grad_norm": 1.406813383102417,
-      "learning_rate": 7.932763819095478e-05,
-      "loss": 5.389,
-      "step": 21075
-    },
-    {
-      "epoch": 10.991395045632334,
-      "grad_norm": 1.3964906930923462,
-      "learning_rate": 7.932663316582914e-05,
-      "loss": 5.4897,
-      "step": 21076
-    },
-    {
-      "epoch": 10.991916558018254,
-      "grad_norm": 1.6340597867965698,
-      "learning_rate": 7.932562814070352e-05,
-      "loss": 4.908,
-      "step": 21077
-    },
-    {
-      "epoch": 10.992438070404171,
-      "grad_norm": 1.451432466506958,
-      "learning_rate": 7.93246231155779e-05,
-      "loss": 5.5549,
-      "step": 21078
-    },
-    {
-      "epoch": 10.99295958279009,
-      "grad_norm": 1.4636764526367188,
-      "learning_rate": 7.932361809045227e-05,
-      "loss": 5.729,
-      "step": 21079
-    },
-    {
-      "epoch": 10.99348109517601,
-      "grad_norm": 1.3572015762329102,
-      "learning_rate": 7.932261306532664e-05,
-      "loss": 5.641,
-      "step": 21080
-    },
-    {
-      "epoch": 10.99400260756193,
-      "grad_norm": 1.3986600637435913,
-      "learning_rate": 7.932160804020102e-05,
-      "loss": 5.5881,
-      "step": 21081
-    },
-    {
-      "epoch": 10.99452411994785,
-      "grad_norm": 1.4535712003707886,
-      "learning_rate": 7.932060301507538e-05,
-      "loss": 5.7041,
-      "step": 21082
-    },
-    {
-      "epoch": 10.995045632333769,
-      "grad_norm": 1.4579298496246338,
-      "learning_rate": 7.931959798994976e-05,
-      "loss": 5.5399,
-      "step": 21083
-    },
-    {
-      "epoch": 10.995567144719686,
-      "grad_norm": 1.3273316621780396,
-      "learning_rate": 7.931859296482412e-05,
-      "loss": 5.7706,
-      "step": 21084
-    },
-    {
-      "epoch": 10.996088657105606,
-      "grad_norm": 1.5663350820541382,
-      "learning_rate": 7.93175879396985e-05,
-      "loss": 4.9125,
-      "step": 21085
-    },
-    {
-      "epoch": 10.996610169491525,
-      "grad_norm": 1.3956539630889893,
-      "learning_rate": 7.931658291457286e-05,
-      "loss": 5.7102,
-      "step": 21086
-    },
-    {
-      "epoch": 10.997131681877445,
-      "grad_norm": 1.4856353998184204,
-      "learning_rate": 7.931557788944723e-05,
-      "loss": 5.191,
-      "step": 21087
-    },
-    {
-      "epoch": 10.997653194263364,
-      "grad_norm": 1.5751895904541016,
-      "learning_rate": 7.93145728643216e-05,
-      "loss": 4.8252,
-      "step": 21088
-    },
-    {
-      "epoch": 10.998174706649284,
-      "grad_norm": 1.5057225227355957,
-      "learning_rate": 7.931356783919598e-05,
-      "loss": 5.6352,
-      "step": 21089
-    },
-    {
-      "epoch": 10.998696219035201,
-      "grad_norm": 1.3950471878051758,
-      "learning_rate": 7.931256281407036e-05,
-      "loss": 5.9779,
-      "step": 21090
-    },
-    {
-      "epoch": 10.99921773142112,
-      "grad_norm": 1.4227081537246704,
-      "learning_rate": 7.931155778894473e-05,
-      "loss": 5.5034,
-      "step": 21091
-    },
-    {
-      "epoch": 10.99973924380704,
-      "grad_norm": 1.4018871784210205,
-      "learning_rate": 7.93105527638191e-05,
-      "loss": 4.8098,
-      "step": 21092
-    },
-    {
-      "epoch": 11.00026075619296,
-      "grad_norm": 1.5166515111923218,
-      "learning_rate": 7.930954773869347e-05,
-      "loss": 5.1578,
-      "step": 21093
-    },
-    {
-      "epoch": 11.00078226857888,
-      "grad_norm": 1.328994631767273,
-      "learning_rate": 7.930854271356785e-05,
-      "loss": 5.7272,
-      "step": 21094
-    },
-    {
-      "epoch": 11.001303780964799,
-      "grad_norm": 1.3658723831176758,
-      "learning_rate": 7.930753768844221e-05,
-      "loss": 5.4617,
-      "step": 21095
-    },
-    {
-      "epoch": 11.001825293350716,
-      "grad_norm": 1.4680907726287842,
-      "learning_rate": 7.930653266331659e-05,
-      "loss": 5.0846,
-      "step": 21096
-    },
-    {
-      "epoch": 11.002346805736636,
-      "grad_norm": 1.5040714740753174,
-      "learning_rate": 7.930552763819095e-05,
-      "loss": 5.4929,
-      "step": 21097
-    },
-    {
-      "epoch": 11.002868318122555,
-      "grad_norm": 1.4263101816177368,
-      "learning_rate": 7.930452261306533e-05,
-      "loss": 5.1554,
-      "step": 21098
-    },
-    {
-      "epoch": 11.003389830508475,
-      "grad_norm": 1.4052366018295288,
-      "learning_rate": 7.930351758793971e-05,
-      "loss": 5.7564,
-      "step": 21099
-    },
-    {
-      "epoch": 11.003911342894394,
-      "grad_norm": 1.5011886358261108,
-      "learning_rate": 7.930251256281409e-05,
-      "loss": 5.4001,
-      "step": 21100
-    },
-    {
-      "epoch": 11.004432855280314,
-      "grad_norm": 1.4798582792282104,
-      "learning_rate": 7.930150753768845e-05,
-      "loss": 5.6641,
-      "step": 21101
-    },
-    {
-      "epoch": 11.004954367666231,
-      "grad_norm": 1.4804071187973022,
-      "learning_rate": 7.930050251256281e-05,
-      "loss": 5.5523,
-      "step": 21102
-    },
-    {
-      "epoch": 11.00547588005215,
-      "grad_norm": 1.586796522140503,
-      "learning_rate": 7.929949748743719e-05,
-      "loss": 5.5702,
-      "step": 21103
-    },
-    {
-      "epoch": 11.00599739243807,
-      "grad_norm": 1.487642526626587,
-      "learning_rate": 7.929849246231156e-05,
-      "loss": 5.3448,
-      "step": 21104
-    },
-    {
-      "epoch": 11.00651890482399,
-      "grad_norm": 1.5953071117401123,
-      "learning_rate": 7.929748743718593e-05,
-      "loss": 5.1455,
-      "step": 21105
-    },
-    {
-      "epoch": 11.00704041720991,
-      "grad_norm": 1.4451327323913574,
-      "learning_rate": 7.92964824120603e-05,
-      "loss": 5.666,
-      "step": 21106
-    },
-    {
-      "epoch": 11.007561929595829,
-      "grad_norm": 1.4291576147079468,
-      "learning_rate": 7.929547738693468e-05,
-      "loss": 5.567,
-      "step": 21107
-    },
-    {
-      "epoch": 11.008083441981746,
-      "grad_norm": 1.5038859844207764,
-      "learning_rate": 7.929447236180904e-05,
-      "loss": 5.6229,
-      "step": 21108
-    },
-    {
-      "epoch": 11.008604954367666,
-      "grad_norm": 1.5580543279647827,
-      "learning_rate": 7.929346733668342e-05,
-      "loss": 5.5601,
-      "step": 21109
-    },
-    {
-      "epoch": 11.009126466753585,
-      "grad_norm": 1.417738676071167,
-      "learning_rate": 7.92924623115578e-05,
-      "loss": 5.7592,
-      "step": 21110
-    },
-    {
-      "epoch": 11.009647979139505,
-      "grad_norm": 1.447232723236084,
-      "learning_rate": 7.929145728643217e-05,
-      "loss": 5.2207,
-      "step": 21111
-    },
-    {
-      "epoch": 11.010169491525424,
-      "grad_norm": 1.44540536403656,
-      "learning_rate": 7.929045226130654e-05,
-      "loss": 5.6559,
-      "step": 21112
-    },
-    {
-      "epoch": 11.010691003911344,
-      "grad_norm": 1.4783271551132202,
-      "learning_rate": 7.928944723618092e-05,
-      "loss": 5.428,
-      "step": 21113
-    },
-    {
-      "epoch": 11.011212516297261,
-      "grad_norm": 1.4941426515579224,
-      "learning_rate": 7.928844221105528e-05,
-      "loss": 5.7347,
-      "step": 21114
-    },
-    {
-      "epoch": 11.01173402868318,
-      "grad_norm": 1.5299079418182373,
-      "learning_rate": 7.928743718592964e-05,
-      "loss": 5.4626,
-      "step": 21115
-    },
-    {
-      "epoch": 11.0122555410691,
-      "grad_norm": 1.3269082307815552,
-      "learning_rate": 7.928643216080402e-05,
-      "loss": 5.4847,
-      "step": 21116
-    },
-    {
-      "epoch": 11.01277705345502,
-      "grad_norm": 1.4174437522888184,
-      "learning_rate": 7.928542713567839e-05,
-      "loss": 5.3985,
-      "step": 21117
-    },
-    {
-      "epoch": 11.01329856584094,
-      "grad_norm": 1.3714170455932617,
-      "learning_rate": 7.928442211055276e-05,
-      "loss": 5.7612,
-      "step": 21118
-    },
-    {
-      "epoch": 11.013820078226859,
-      "grad_norm": 1.3990304470062256,
-      "learning_rate": 7.928341708542714e-05,
-      "loss": 5.2894,
-      "step": 21119
-    },
-    {
-      "epoch": 11.014341590612776,
-      "grad_norm": 1.405818223953247,
-      "learning_rate": 7.928241206030152e-05,
-      "loss": 5.7779,
-      "step": 21120
-    },
-    {
-      "epoch": 11.014863102998696,
-      "grad_norm": 1.4300235509872437,
-      "learning_rate": 7.928140703517588e-05,
-      "loss": 5.6564,
-      "step": 21121
-    },
-    {
-      "epoch": 11.015384615384615,
-      "grad_norm": 1.3794262409210205,
-      "learning_rate": 7.928040201005026e-05,
-      "loss": 5.8594,
-      "step": 21122
-    },
-    {
-      "epoch": 11.015906127770535,
-      "grad_norm": 1.4532383680343628,
-      "learning_rate": 7.927939698492463e-05,
-      "loss": 5.6946,
-      "step": 21123
-    },
-    {
-      "epoch": 11.016427640156454,
-      "grad_norm": 1.3015906810760498,
-      "learning_rate": 7.9278391959799e-05,
-      "loss": 5.7648,
-      "step": 21124
-    },
-    {
-      "epoch": 11.016949152542374,
-      "grad_norm": 1.415192723274231,
-      "learning_rate": 7.927738693467337e-05,
-      "loss": 5.459,
-      "step": 21125
-    },
-    {
-      "epoch": 11.017470664928291,
-      "grad_norm": 1.470921277999878,
-      "learning_rate": 7.927638190954775e-05,
-      "loss": 4.8687,
-      "step": 21126
-    },
-    {
-      "epoch": 11.01799217731421,
-      "grad_norm": 1.551245927810669,
-      "learning_rate": 7.927537688442211e-05,
-      "loss": 5.5919,
-      "step": 21127
-    },
-    {
-      "epoch": 11.01851368970013,
-      "grad_norm": 1.8100107908248901,
-      "learning_rate": 7.927437185929648e-05,
-      "loss": 4.523,
-      "step": 21128
-    },
-    {
-      "epoch": 11.01903520208605,
-      "grad_norm": 1.4497419595718384,
-      "learning_rate": 7.927336683417085e-05,
-      "loss": 4.7015,
-      "step": 21129
-    },
-    {
-      "epoch": 11.01955671447197,
-      "grad_norm": 1.5328397750854492,
-      "learning_rate": 7.927236180904523e-05,
-      "loss": 5.3442,
-      "step": 21130
-    },
-    {
-      "epoch": 11.020078226857889,
-      "grad_norm": 1.5297495126724243,
-      "learning_rate": 7.927135678391961e-05,
-      "loss": 5.3453,
-      "step": 21131
-    },
-    {
-      "epoch": 11.020599739243806,
-      "grad_norm": 1.538192629814148,
-      "learning_rate": 7.927035175879397e-05,
-      "loss": 5.1751,
-      "step": 21132
-    },
-    {
-      "epoch": 11.021121251629726,
-      "grad_norm": 1.4744600057601929,
-      "learning_rate": 7.926934673366835e-05,
-      "loss": 5.5567,
-      "step": 21133
-    },
-    {
-      "epoch": 11.021642764015645,
-      "grad_norm": 1.418416976928711,
-      "learning_rate": 7.926834170854271e-05,
-      "loss": 5.5466,
-      "step": 21134
-    },
-    {
-      "epoch": 11.022164276401565,
-      "grad_norm": 1.598259449005127,
-      "learning_rate": 7.926733668341709e-05,
-      "loss": 5.2062,
-      "step": 21135
-    },
-    {
-      "epoch": 11.022685788787484,
-      "grad_norm": 1.3259409666061401,
-      "learning_rate": 7.926633165829146e-05,
-      "loss": 5.8078,
-      "step": 21136
-    },
-    {
-      "epoch": 11.023207301173404,
-      "grad_norm": 1.3556679487228394,
-      "learning_rate": 7.926532663316583e-05,
-      "loss": 5.3739,
-      "step": 21137
-    },
-    {
-      "epoch": 11.023728813559321,
-      "grad_norm": 1.4929907321929932,
-      "learning_rate": 7.92643216080402e-05,
-      "loss": 5.6128,
-      "step": 21138
-    },
-    {
-      "epoch": 11.024250325945241,
-      "grad_norm": 1.5007375478744507,
-      "learning_rate": 7.926331658291458e-05,
-      "loss": 5.2952,
-      "step": 21139
-    },
-    {
-      "epoch": 11.02477183833116,
-      "grad_norm": 1.5095196962356567,
-      "learning_rate": 7.926231155778895e-05,
-      "loss": 5.3569,
-      "step": 21140
-    },
-    {
-      "epoch": 11.02529335071708,
-      "grad_norm": 1.3737893104553223,
-      "learning_rate": 7.926130653266332e-05,
-      "loss": 5.691,
-      "step": 21141
-    },
-    {
-      "epoch": 11.025814863103,
-      "grad_norm": 1.4987726211547852,
-      "learning_rate": 7.92603015075377e-05,
-      "loss": 5.2381,
-      "step": 21142
-    },
-    {
-      "epoch": 11.026336375488917,
-      "grad_norm": 1.4082118272781372,
-      "learning_rate": 7.925929648241206e-05,
-      "loss": 5.5554,
-      "step": 21143
-    },
-    {
-      "epoch": 11.026857887874836,
-      "grad_norm": 1.521992802619934,
-      "learning_rate": 7.925829145728644e-05,
-      "loss": 5.118,
-      "step": 21144
-    },
-    {
-      "epoch": 11.027379400260756,
-      "grad_norm": 1.3802266120910645,
-      "learning_rate": 7.92572864321608e-05,
-      "loss": 5.832,
-      "step": 21145
-    },
-    {
-      "epoch": 11.027900912646675,
-      "grad_norm": 1.4444068670272827,
-      "learning_rate": 7.925628140703518e-05,
-      "loss": 5.5427,
-      "step": 21146
-    },
-    {
-      "epoch": 11.028422425032595,
-      "grad_norm": 1.423485517501831,
-      "learning_rate": 7.925527638190955e-05,
-      "loss": 5.219,
-      "step": 21147
-    },
-    {
-      "epoch": 11.028943937418514,
-      "grad_norm": 1.3630317449569702,
-      "learning_rate": 7.925427135678392e-05,
-      "loss": 5.9402,
-      "step": 21148
-    },
-    {
-      "epoch": 11.029465449804432,
-      "grad_norm": 1.399409532546997,
-      "learning_rate": 7.925326633165829e-05,
-      "loss": 5.5909,
-      "step": 21149
-    },
-    {
-      "epoch": 11.029986962190351,
-      "grad_norm": 1.4210819005966187,
-      "learning_rate": 7.925226130653267e-05,
-      "loss": 5.4409,
-      "step": 21150
-    },
-    {
-      "epoch": 11.030508474576271,
-      "grad_norm": 1.4736242294311523,
-      "learning_rate": 7.925125628140704e-05,
-      "loss": 5.529,
-      "step": 21151
-    },
-    {
-      "epoch": 11.03102998696219,
-      "grad_norm": 1.4509730339050293,
-      "learning_rate": 7.925025125628142e-05,
-      "loss": 5.18,
-      "step": 21152
-    },
-    {
-      "epoch": 11.03155149934811,
-      "grad_norm": 1.457413911819458,
-      "learning_rate": 7.924924623115579e-05,
-      "loss": 5.6129,
-      "step": 21153
-    },
-    {
-      "epoch": 11.03207301173403,
-      "grad_norm": 1.4118088483810425,
-      "learning_rate": 7.924824120603015e-05,
-      "loss": 5.4667,
-      "step": 21154
-    },
-    {
-      "epoch": 11.032594524119947,
-      "grad_norm": 1.4206299781799316,
-      "learning_rate": 7.924723618090453e-05,
-      "loss": 5.4768,
-      "step": 21155
-    },
-    {
-      "epoch": 11.033116036505866,
-      "grad_norm": 1.4658806324005127,
-      "learning_rate": 7.924623115577889e-05,
-      "loss": 5.2773,
-      "step": 21156
-    },
-    {
-      "epoch": 11.033637548891786,
-      "grad_norm": 1.4071263074874878,
-      "learning_rate": 7.924522613065327e-05,
-      "loss": 5.3763,
-      "step": 21157
-    },
-    {
-      "epoch": 11.034159061277705,
-      "grad_norm": 1.4719984531402588,
-      "learning_rate": 7.924422110552763e-05,
-      "loss": 5.766,
-      "step": 21158
-    },
-    {
-      "epoch": 11.034680573663625,
-      "grad_norm": 1.4947247505187988,
-      "learning_rate": 7.924321608040201e-05,
-      "loss": 5.366,
-      "step": 21159
-    },
-    {
-      "epoch": 11.035202086049544,
-      "grad_norm": 1.3157250881195068,
-      "learning_rate": 7.924221105527639e-05,
-      "loss": 5.2314,
-      "step": 21160
-    },
-    {
-      "epoch": 11.035723598435462,
-      "grad_norm": 1.423624873161316,
-      "learning_rate": 7.924120603015077e-05,
-      "loss": 5.4141,
-      "step": 21161
-    },
-    {
-      "epoch": 11.036245110821381,
-      "grad_norm": 1.4969482421875,
-      "learning_rate": 7.924020100502513e-05,
-      "loss": 5.1185,
-      "step": 21162
-    },
-    {
-      "epoch": 11.036766623207301,
-      "grad_norm": 1.5351263284683228,
-      "learning_rate": 7.923919597989951e-05,
-      "loss": 5.0627,
-      "step": 21163
-    },
-    {
-      "epoch": 11.03728813559322,
-      "grad_norm": 1.4909899234771729,
-      "learning_rate": 7.923819095477387e-05,
-      "loss": 5.0985,
-      "step": 21164
-    },
-    {
-      "epoch": 11.03780964797914,
-      "grad_norm": 1.5653775930404663,
-      "learning_rate": 7.923718592964825e-05,
-      "loss": 5.1166,
-      "step": 21165
-    },
-    {
-      "epoch": 11.03833116036506,
-      "grad_norm": 1.464033603668213,
-      "learning_rate": 7.923618090452262e-05,
-      "loss": 5.4518,
-      "step": 21166
-    },
-    {
-      "epoch": 11.038852672750977,
-      "grad_norm": 1.57407808303833,
-      "learning_rate": 7.923517587939698e-05,
-      "loss": 5.3287,
-      "step": 21167
-    },
-    {
-      "epoch": 11.039374185136897,
-      "grad_norm": 1.991546630859375,
-      "learning_rate": 7.923417085427136e-05,
-      "loss": 4.8421,
-      "step": 21168
-    },
-    {
-      "epoch": 11.039895697522816,
-      "grad_norm": 1.4552110433578491,
-      "learning_rate": 7.923316582914572e-05,
-      "loss": 4.7238,
-      "step": 21169
-    },
-    {
-      "epoch": 11.040417209908735,
-      "grad_norm": 1.4475016593933105,
-      "learning_rate": 7.92321608040201e-05,
-      "loss": 5.2892,
-      "step": 21170
-    },
-    {
-      "epoch": 11.040938722294655,
-      "grad_norm": 1.3825756311416626,
-      "learning_rate": 7.923115577889448e-05,
-      "loss": 5.6258,
-      "step": 21171
-    },
-    {
-      "epoch": 11.041460234680574,
-      "grad_norm": 1.5588867664337158,
-      "learning_rate": 7.923015075376886e-05,
-      "loss": 5.5649,
-      "step": 21172
-    },
-    {
-      "epoch": 11.041981747066492,
-      "grad_norm": 1.4414454698562622,
-      "learning_rate": 7.922914572864322e-05,
-      "loss": 5.5498,
-      "step": 21173
-    },
-    {
-      "epoch": 11.042503259452412,
-      "grad_norm": 1.3351343870162964,
-      "learning_rate": 7.92281407035176e-05,
-      "loss": 4.9711,
-      "step": 21174
-    },
-    {
-      "epoch": 11.043024771838331,
-      "grad_norm": 1.4149149656295776,
-      "learning_rate": 7.922713567839196e-05,
-      "loss": 5.406,
-      "step": 21175
-    },
-    {
-      "epoch": 11.04354628422425,
-      "grad_norm": 1.468466877937317,
-      "learning_rate": 7.922613065326634e-05,
-      "loss": 5.0924,
-      "step": 21176
-    },
-    {
-      "epoch": 11.04406779661017,
-      "grad_norm": 1.4128668308258057,
-      "learning_rate": 7.92251256281407e-05,
-      "loss": 5.6233,
-      "step": 21177
-    },
-    {
-      "epoch": 11.04458930899609,
-      "grad_norm": 1.3970285654067993,
-      "learning_rate": 7.922412060301508e-05,
-      "loss": 5.7635,
-      "step": 21178
-    },
-    {
-      "epoch": 11.045110821382007,
-      "grad_norm": 1.3432559967041016,
-      "learning_rate": 7.922311557788945e-05,
-      "loss": 5.7469,
-      "step": 21179
-    },
-    {
-      "epoch": 11.045632333767927,
-      "grad_norm": 1.3531700372695923,
-      "learning_rate": 7.922211055276382e-05,
-      "loss": 5.8838,
-      "step": 21180
-    },
-    {
-      "epoch": 11.046153846153846,
-      "grad_norm": 1.4331624507904053,
-      "learning_rate": 7.92211055276382e-05,
-      "loss": 5.5378,
-      "step": 21181
-    },
-    {
-      "epoch": 11.046675358539765,
-      "grad_norm": 1.4573742151260376,
-      "learning_rate": 7.922010050251257e-05,
-      "loss": 4.933,
-      "step": 21182
-    },
-    {
-      "epoch": 11.047196870925685,
-      "grad_norm": 1.6009808778762817,
-      "learning_rate": 7.921909547738694e-05,
-      "loss": 5.2543,
-      "step": 21183
-    },
-    {
-      "epoch": 11.047718383311604,
-      "grad_norm": 1.501879334449768,
-      "learning_rate": 7.921809045226131e-05,
-      "loss": 5.1047,
-      "step": 21184
-    },
-    {
-      "epoch": 11.048239895697522,
-      "grad_norm": 1.4740437269210815,
-      "learning_rate": 7.921708542713569e-05,
-      "loss": 5.4539,
-      "step": 21185
-    },
-    {
-      "epoch": 11.048761408083442,
-      "grad_norm": 1.4400067329406738,
-      "learning_rate": 7.921608040201005e-05,
-      "loss": 5.2321,
-      "step": 21186
-    },
-    {
-      "epoch": 11.049282920469361,
-      "grad_norm": 1.480890154838562,
-      "learning_rate": 7.921507537688443e-05,
-      "loss": 5.4579,
-      "step": 21187
-    },
-    {
-      "epoch": 11.04980443285528,
-      "grad_norm": 1.4505083560943604,
-      "learning_rate": 7.921407035175879e-05,
-      "loss": 5.7324,
-      "step": 21188
-    },
-    {
-      "epoch": 11.0503259452412,
-      "grad_norm": 1.4105050563812256,
-      "learning_rate": 7.921306532663317e-05,
-      "loss": 5.5168,
-      "step": 21189
-    },
-    {
-      "epoch": 11.05084745762712,
-      "grad_norm": 1.380867600440979,
-      "learning_rate": 7.921206030150753e-05,
-      "loss": 5.6185,
-      "step": 21190
-    },
-    {
-      "epoch": 11.051368970013037,
-      "grad_norm": 1.4515094757080078,
-      "learning_rate": 7.921105527638191e-05,
-      "loss": 4.6706,
-      "step": 21191
-    },
-    {
-      "epoch": 11.051890482398957,
-      "grad_norm": 1.6026372909545898,
-      "learning_rate": 7.921005025125629e-05,
-      "loss": 5.161,
-      "step": 21192
-    },
-    {
-      "epoch": 11.052411994784876,
-      "grad_norm": 1.2773542404174805,
-      "learning_rate": 7.920904522613067e-05,
-      "loss": 5.496,
-      "step": 21193
-    },
-    {
-      "epoch": 11.052933507170795,
-      "grad_norm": 1.406302809715271,
-      "learning_rate": 7.920804020100503e-05,
-      "loss": 5.3697,
-      "step": 21194
-    },
-    {
-      "epoch": 11.053455019556715,
-      "grad_norm": 1.4720832109451294,
-      "learning_rate": 7.92070351758794e-05,
-      "loss": 5.3428,
-      "step": 21195
-    },
-    {
-      "epoch": 11.053976531942634,
-      "grad_norm": 1.4226127862930298,
-      "learning_rate": 7.920603015075377e-05,
-      "loss": 5.5224,
-      "step": 21196
-    },
-    {
-      "epoch": 11.054498044328552,
-      "grad_norm": 1.5327823162078857,
-      "learning_rate": 7.920502512562814e-05,
-      "loss": 5.4728,
-      "step": 21197
-    },
-    {
-      "epoch": 11.055019556714472,
-      "grad_norm": 1.4475775957107544,
-      "learning_rate": 7.920402010050252e-05,
-      "loss": 5.5179,
-      "step": 21198
-    },
-    {
-      "epoch": 11.055541069100391,
-      "grad_norm": 1.3338887691497803,
-      "learning_rate": 7.920301507537688e-05,
-      "loss": 5.6187,
-      "step": 21199
-    },
-    {
-      "epoch": 11.05606258148631,
-      "grad_norm": 1.509534239768982,
-      "learning_rate": 7.920201005025126e-05,
-      "loss": 4.5805,
-      "step": 21200
-    },
-    {
-      "epoch": 11.05658409387223,
-      "grad_norm": 1.4591636657714844,
-      "learning_rate": 7.920100502512564e-05,
-      "loss": 5.7511,
-      "step": 21201
-    },
-    {
-      "epoch": 11.05710560625815,
-      "grad_norm": 1.8478814363479614,
-      "learning_rate": 7.920000000000001e-05,
-      "loss": 4.8851,
-      "step": 21202
-    },
-    {
-      "epoch": 11.057627118644067,
-      "grad_norm": 1.426405668258667,
-      "learning_rate": 7.919899497487438e-05,
-      "loss": 5.5841,
-      "step": 21203
-    },
-    {
-      "epoch": 11.058148631029987,
-      "grad_norm": 1.3223819732666016,
-      "learning_rate": 7.919798994974876e-05,
-      "loss": 5.8779,
-      "step": 21204
-    },
-    {
-      "epoch": 11.058670143415906,
-      "grad_norm": 1.4365882873535156,
-      "learning_rate": 7.919698492462312e-05,
-      "loss": 5.2282,
-      "step": 21205
-    },
-    {
-      "epoch": 11.059191655801826,
-      "grad_norm": 1.5225387811660767,
-      "learning_rate": 7.91959798994975e-05,
-      "loss": 5.3887,
-      "step": 21206
-    },
-    {
-      "epoch": 11.059713168187745,
-      "grad_norm": 1.3905385732650757,
-      "learning_rate": 7.919497487437186e-05,
-      "loss": 5.586,
-      "step": 21207
-    },
-    {
-      "epoch": 11.060234680573664,
-      "grad_norm": 1.4842307567596436,
-      "learning_rate": 7.919396984924623e-05,
-      "loss": 5.4017,
-      "step": 21208
-    },
-    {
-      "epoch": 11.060756192959582,
-      "grad_norm": 1.4051399230957031,
-      "learning_rate": 7.91929648241206e-05,
-      "loss": 5.6332,
-      "step": 21209
-    },
-    {
-      "epoch": 11.061277705345502,
-      "grad_norm": 1.8363021612167358,
-      "learning_rate": 7.919195979899497e-05,
-      "loss": 5.0921,
-      "step": 21210
-    },
-    {
-      "epoch": 11.061799217731421,
-      "grad_norm": 1.4234111309051514,
-      "learning_rate": 7.919095477386935e-05,
-      "loss": 5.8824,
-      "step": 21211
-    },
-    {
-      "epoch": 11.06232073011734,
-      "grad_norm": 1.4245593547821045,
-      "learning_rate": 7.918994974874372e-05,
-      "loss": 5.4575,
-      "step": 21212
-    },
-    {
-      "epoch": 11.06284224250326,
-      "grad_norm": 1.4988527297973633,
-      "learning_rate": 7.91889447236181e-05,
-      "loss": 5.806,
-      "step": 21213
-    },
-    {
-      "epoch": 11.06336375488918,
-      "grad_norm": 1.5243198871612549,
-      "learning_rate": 7.918793969849247e-05,
-      "loss": 5.4911,
-      "step": 21214
-    },
-    {
-      "epoch": 11.063885267275097,
-      "grad_norm": 1.5343937873840332,
-      "learning_rate": 7.918693467336684e-05,
-      "loss": 5.4117,
-      "step": 21215
-    },
-    {
-      "epoch": 11.064406779661017,
-      "grad_norm": 1.3255337476730347,
-      "learning_rate": 7.918592964824121e-05,
-      "loss": 5.7498,
-      "step": 21216
-    },
-    {
-      "epoch": 11.064928292046936,
-      "grad_norm": 1.4247322082519531,
-      "learning_rate": 7.918492462311559e-05,
-      "loss": 5.6449,
-      "step": 21217
-    },
-    {
-      "epoch": 11.065449804432856,
-      "grad_norm": 1.4279701709747314,
-      "learning_rate": 7.918391959798995e-05,
-      "loss": 5.4121,
-      "step": 21218
-    },
-    {
-      "epoch": 11.065971316818775,
-      "grad_norm": 1.4766770601272583,
-      "learning_rate": 7.918291457286433e-05,
-      "loss": 5.0202,
-      "step": 21219
-    },
-    {
-      "epoch": 11.066492829204694,
-      "grad_norm": 1.39676034450531,
-      "learning_rate": 7.918190954773869e-05,
-      "loss": 5.6075,
-      "step": 21220
-    },
-    {
-      "epoch": 11.067014341590612,
-      "grad_norm": 1.4796735048294067,
-      "learning_rate": 7.918090452261306e-05,
-      "loss": 5.5205,
-      "step": 21221
-    },
-    {
-      "epoch": 11.067535853976532,
-      "grad_norm": 1.3716721534729004,
-      "learning_rate": 7.917989949748744e-05,
-      "loss": 5.6054,
-      "step": 21222
-    },
-    {
-      "epoch": 11.068057366362451,
-      "grad_norm": 1.4955058097839355,
-      "learning_rate": 7.917889447236181e-05,
-      "loss": 5.146,
-      "step": 21223
-    },
-    {
-      "epoch": 11.06857887874837,
-      "grad_norm": 1.4187158346176147,
-      "learning_rate": 7.917788944723619e-05,
-      "loss": 5.6838,
-      "step": 21224
-    },
-    {
-      "epoch": 11.06910039113429,
-      "grad_norm": 1.4310904741287231,
-      "learning_rate": 7.917688442211056e-05,
-      "loss": 5.0359,
-      "step": 21225
-    },
-    {
-      "epoch": 11.06962190352021,
-      "grad_norm": 1.4314510822296143,
-      "learning_rate": 7.917587939698493e-05,
-      "loss": 5.334,
-      "step": 21226
-    },
-    {
-      "epoch": 11.070143415906127,
-      "grad_norm": 1.4251179695129395,
-      "learning_rate": 7.91748743718593e-05,
-      "loss": 5.3107,
-      "step": 21227
-    },
-    {
-      "epoch": 11.070664928292047,
-      "grad_norm": 1.4709123373031616,
-      "learning_rate": 7.917386934673368e-05,
-      "loss": 5.4979,
-      "step": 21228
-    },
-    {
-      "epoch": 11.071186440677966,
-      "grad_norm": 1.3267229795455933,
-      "learning_rate": 7.917286432160804e-05,
-      "loss": 5.6282,
-      "step": 21229
-    },
-    {
-      "epoch": 11.071707953063886,
-      "grad_norm": 1.5685628652572632,
-      "learning_rate": 7.917185929648242e-05,
-      "loss": 5.1599,
-      "step": 21230
-    },
-    {
-      "epoch": 11.072229465449805,
-      "grad_norm": 1.4914478063583374,
-      "learning_rate": 7.917085427135678e-05,
-      "loss": 5.2843,
-      "step": 21231
-    },
-    {
-      "epoch": 11.072750977835724,
-      "grad_norm": 1.3930107355117798,
-      "learning_rate": 7.916984924623116e-05,
-      "loss": 5.4282,
-      "step": 21232
-    },
-    {
-      "epoch": 11.073272490221642,
-      "grad_norm": 1.4599641561508179,
-      "learning_rate": 7.916884422110554e-05,
-      "loss": 5.5887,
-      "step": 21233
-    },
-    {
-      "epoch": 11.073794002607562,
-      "grad_norm": 1.371597409248352,
-      "learning_rate": 7.91678391959799e-05,
-      "loss": 5.7628,
-      "step": 21234
-    },
-    {
-      "epoch": 11.074315514993481,
-      "grad_norm": 1.5411545038223267,
-      "learning_rate": 7.916683417085428e-05,
-      "loss": 5.4043,
-      "step": 21235
-    },
-    {
-      "epoch": 11.0748370273794,
-      "grad_norm": 1.5288842916488647,
-      "learning_rate": 7.916582914572864e-05,
-      "loss": 5.2011,
-      "step": 21236
-    },
-    {
-      "epoch": 11.07535853976532,
-      "grad_norm": 1.4874614477157593,
-      "learning_rate": 7.916482412060302e-05,
-      "loss": 5.4894,
-      "step": 21237
-    },
-    {
-      "epoch": 11.075880052151238,
-      "grad_norm": 1.490326166152954,
-      "learning_rate": 7.916381909547739e-05,
-      "loss": 5.4642,
-      "step": 21238
-    },
-    {
-      "epoch": 11.076401564537157,
-      "grad_norm": 1.5152308940887451,
-      "learning_rate": 7.916281407035176e-05,
-      "loss": 5.7777,
-      "step": 21239
-    },
-    {
-      "epoch": 11.076923076923077,
-      "grad_norm": 1.615773320198059,
-      "learning_rate": 7.916180904522613e-05,
-      "loss": 5.1504,
-      "step": 21240
-    },
-    {
-      "epoch": 11.077444589308996,
-      "grad_norm": 1.4543851613998413,
-      "learning_rate": 7.91608040201005e-05,
-      "loss": 5.8302,
-      "step": 21241
-    },
-    {
-      "epoch": 11.077966101694916,
-      "grad_norm": 1.4628264904022217,
-      "learning_rate": 7.915979899497487e-05,
-      "loss": 5.2132,
-      "step": 21242
-    },
-    {
-      "epoch": 11.078487614080835,
-      "grad_norm": 1.3789745569229126,
-      "learning_rate": 7.915879396984925e-05,
-      "loss": 5.7544,
-      "step": 21243
-    },
-    {
-      "epoch": 11.079009126466753,
-      "grad_norm": 1.4564965963363647,
-      "learning_rate": 7.915778894472363e-05,
-      "loss": 5.69,
-      "step": 21244
-    },
-    {
-      "epoch": 11.079530638852672,
-      "grad_norm": 1.3901844024658203,
-      "learning_rate": 7.9156783919598e-05,
-      "loss": 5.565,
-      "step": 21245
-    },
-    {
-      "epoch": 11.080052151238592,
-      "grad_norm": 1.4455740451812744,
-      "learning_rate": 7.915577889447237e-05,
-      "loss": 5.2531,
-      "step": 21246
-    },
-    {
-      "epoch": 11.080573663624511,
-      "grad_norm": 1.4660471677780151,
-      "learning_rate": 7.915477386934673e-05,
-      "loss": 5.6869,
-      "step": 21247
-    },
-    {
-      "epoch": 11.08109517601043,
-      "grad_norm": 1.5206879377365112,
-      "learning_rate": 7.915376884422111e-05,
-      "loss": 5.5161,
-      "step": 21248
-    },
-    {
-      "epoch": 11.08161668839635,
-      "grad_norm": 1.5010071992874146,
-      "learning_rate": 7.915276381909547e-05,
-      "loss": 5.0025,
-      "step": 21249
-    },
-    {
-      "epoch": 11.082138200782268,
-      "grad_norm": 1.6146413087844849,
-      "learning_rate": 7.915175879396985e-05,
-      "loss": 4.8091,
-      "step": 21250
-    },
-    {
-      "epoch": 11.082659713168187,
-      "grad_norm": 1.3685529232025146,
-      "learning_rate": 7.915075376884422e-05,
-      "loss": 5.6754,
-      "step": 21251
-    },
-    {
-      "epoch": 11.083181225554107,
-      "grad_norm": 1.4532442092895508,
-      "learning_rate": 7.91497487437186e-05,
-      "loss": 5.7915,
-      "step": 21252
-    },
-    {
-      "epoch": 11.083702737940026,
-      "grad_norm": 1.593357801437378,
-      "learning_rate": 7.914874371859297e-05,
-      "loss": 5.396,
-      "step": 21253
-    },
-    {
-      "epoch": 11.084224250325946,
-      "grad_norm": 1.6555933952331543,
-      "learning_rate": 7.914773869346735e-05,
-      "loss": 5.0878,
-      "step": 21254
-    },
-    {
-      "epoch": 11.084745762711865,
-      "grad_norm": 1.5196130275726318,
-      "learning_rate": 7.914673366834171e-05,
-      "loss": 5.2046,
-      "step": 21255
-    },
-    {
-      "epoch": 11.085267275097783,
-      "grad_norm": 1.4507348537445068,
-      "learning_rate": 7.914572864321609e-05,
-      "loss": 5.4953,
-      "step": 21256
-    },
-    {
-      "epoch": 11.085788787483702,
-      "grad_norm": 1.4180045127868652,
-      "learning_rate": 7.914472361809046e-05,
-      "loss": 5.8469,
-      "step": 21257
-    },
-    {
-      "epoch": 11.086310299869622,
-      "grad_norm": 1.43502938747406,
-      "learning_rate": 7.914371859296483e-05,
-      "loss": 5.6067,
-      "step": 21258
-    },
-    {
-      "epoch": 11.086831812255541,
-      "grad_norm": 1.4701634645462036,
-      "learning_rate": 7.91427135678392e-05,
-      "loss": 5.6349,
-      "step": 21259
-    },
-    {
-      "epoch": 11.08735332464146,
-      "grad_norm": 1.4130322933197021,
-      "learning_rate": 7.914170854271356e-05,
-      "loss": 5.4914,
-      "step": 21260
-    },
-    {
-      "epoch": 11.08787483702738,
-      "grad_norm": 1.395987868309021,
-      "learning_rate": 7.914070351758794e-05,
-      "loss": 5.2964,
-      "step": 21261
-    },
-    {
-      "epoch": 11.088396349413298,
-      "grad_norm": Infinity,
-      "learning_rate": 7.914070351758794e-05,
-      "loss": 5.7257,
-      "step": 21262
-    },
-    {
-      "epoch": 11.088917861799217,
-      "grad_norm": 1.4860707521438599,
-      "learning_rate": 7.91396984924623e-05,
-      "loss": 5.3845,
-      "step": 21263
-    },
-    {
-      "epoch": 11.089439374185137,
-      "grad_norm": 1.5197169780731201,
-      "learning_rate": 7.913869346733668e-05,
-      "loss": 5.0208,
-      "step": 21264
-    },
-    {
-      "epoch": 11.089960886571056,
-      "grad_norm": 1.5625925064086914,
-      "learning_rate": 7.913768844221106e-05,
-      "loss": 5.0949,
-      "step": 21265
-    },
-    {
-      "epoch": 11.090482398956976,
-      "grad_norm": 1.3764362335205078,
-      "learning_rate": 7.913668341708544e-05,
-      "loss": 5.2299,
-      "step": 21266
-    },
-    {
-      "epoch": 11.091003911342895,
-      "grad_norm": 1.4350699186325073,
-      "learning_rate": 7.91356783919598e-05,
-      "loss": 5.6264,
-      "step": 21267
-    },
-    {
-      "epoch": 11.091525423728813,
-      "grad_norm": 1.4505139589309692,
-      "learning_rate": 7.913467336683418e-05,
-      "loss": 5.5128,
-      "step": 21268
-    },
-    {
-      "epoch": 11.092046936114732,
-      "grad_norm": 1.4239321947097778,
-      "learning_rate": 7.913366834170854e-05,
-      "loss": 5.2717,
-      "step": 21269
-    },
-    {
-      "epoch": 11.092568448500652,
-      "grad_norm": 1.415115475654602,
-      "learning_rate": 7.913266331658292e-05,
-      "loss": 5.6871,
-      "step": 21270
-    },
-    {
-      "epoch": 11.093089960886571,
-      "grad_norm": 1.498635172843933,
-      "learning_rate": 7.913165829145729e-05,
-      "loss": 5.6594,
-      "step": 21271
-    },
-    {
-      "epoch": 11.09361147327249,
-      "grad_norm": 1.5114023685455322,
-      "learning_rate": 7.913065326633166e-05,
-      "loss": 5.5052,
-      "step": 21272
-    },
-    {
-      "epoch": 11.09413298565841,
-      "grad_norm": 1.498643159866333,
-      "learning_rate": 7.912964824120603e-05,
-      "loss": 5.2302,
-      "step": 21273
-    },
-    {
-      "epoch": 11.094654498044328,
-      "grad_norm": 1.4168274402618408,
-      "learning_rate": 7.91286432160804e-05,
-      "loss": 6.0205,
-      "step": 21274
-    },
-    {
-      "epoch": 11.095176010430247,
-      "grad_norm": 1.630904197692871,
-      "learning_rate": 7.912763819095478e-05,
-      "loss": 5.2637,
-      "step": 21275
-    },
-    {
-      "epoch": 11.095697522816167,
-      "grad_norm": 1.3958061933517456,
-      "learning_rate": 7.912663316582915e-05,
-      "loss": 5.6839,
-      "step": 21276
-    },
-    {
-      "epoch": 11.096219035202086,
-      "grad_norm": 1.3279367685317993,
-      "learning_rate": 7.912562814070353e-05,
-      "loss": 5.8795,
-      "step": 21277
-    },
-    {
-      "epoch": 11.096740547588006,
-      "grad_norm": 1.402670979499817,
-      "learning_rate": 7.912462311557789e-05,
-      "loss": 5.4999,
-      "step": 21278
-    },
-    {
-      "epoch": 11.097262059973925,
-      "grad_norm": 1.435163974761963,
-      "learning_rate": 7.912361809045227e-05,
-      "loss": 5.592,
-      "step": 21279
-    },
-    {
-      "epoch": 11.097783572359843,
-      "grad_norm": 1.4810028076171875,
-      "learning_rate": 7.912261306532663e-05,
-      "loss": 5.2114,
-      "step": 21280
-    },
-    {
-      "epoch": 11.098305084745762,
-      "grad_norm": 1.4576942920684814,
-      "learning_rate": 7.912160804020101e-05,
-      "loss": 5.1535,
-      "step": 21281
-    },
-    {
-      "epoch": 11.098826597131682,
-      "grad_norm": 1.4434548616409302,
-      "learning_rate": 7.912060301507537e-05,
-      "loss": 5.3347,
-      "step": 21282
-    },
-    {
-      "epoch": 11.099348109517601,
-      "grad_norm": 1.5359630584716797,
-      "learning_rate": 7.911959798994975e-05,
-      "loss": 5.112,
-      "step": 21283
-    },
-    {
-      "epoch": 11.09986962190352,
-      "grad_norm": 1.468384027481079,
-      "learning_rate": 7.911859296482412e-05,
-      "loss": 5.5106,
-      "step": 21284
-    },
-    {
-      "epoch": 11.10039113428944,
-      "grad_norm": 1.4887703657150269,
-      "learning_rate": 7.91175879396985e-05,
-      "loss": 5.2821,
-      "step": 21285
-    },
-    {
-      "epoch": 11.100912646675358,
-      "grad_norm": 1.3913615942001343,
-      "learning_rate": 7.911658291457287e-05,
-      "loss": 5.1712,
-      "step": 21286
-    },
-    {
-      "epoch": 11.101434159061277,
-      "grad_norm": 1.420173168182373,
-      "learning_rate": 7.911557788944725e-05,
-      "loss": 5.4368,
-      "step": 21287
-    },
-    {
-      "epoch": 11.101955671447197,
-      "grad_norm": 1.4927680492401123,
-      "learning_rate": 7.911457286432161e-05,
-      "loss": 5.4484,
-      "step": 21288
-    },
-    {
-      "epoch": 11.102477183833116,
-      "grad_norm": 1.4720066785812378,
-      "learning_rate": 7.911356783919598e-05,
-      "loss": 5.7663,
-      "step": 21289
-    },
-    {
-      "epoch": 11.102998696219036,
-      "grad_norm": 1.4533277750015259,
-      "learning_rate": 7.911256281407036e-05,
-      "loss": 5.7769,
-      "step": 21290
-    },
-    {
-      "epoch": 11.103520208604955,
-      "grad_norm": 1.5876013040542603,
-      "learning_rate": 7.911155778894472e-05,
-      "loss": 5.1546,
-      "step": 21291
-    },
-    {
-      "epoch": 11.104041720990873,
-      "grad_norm": 1.37773859500885,
-      "learning_rate": 7.91105527638191e-05,
-      "loss": 5.5711,
-      "step": 21292
-    },
-    {
-      "epoch": 11.104563233376792,
-      "grad_norm": 1.3498377799987793,
-      "learning_rate": 7.910954773869346e-05,
-      "loss": 5.5363,
-      "step": 21293
-    },
-    {
-      "epoch": 11.105084745762712,
-      "grad_norm": 1.4859713315963745,
-      "learning_rate": 7.910854271356784e-05,
-      "loss": 5.3093,
-      "step": 21294
-    },
-    {
-      "epoch": 11.105606258148631,
-      "grad_norm": 1.4717473983764648,
-      "learning_rate": 7.910753768844222e-05,
-      "loss": 5.471,
-      "step": 21295
-    },
-    {
-      "epoch": 11.10612777053455,
-      "grad_norm": 1.4986631870269775,
-      "learning_rate": 7.91065326633166e-05,
-      "loss": 5.6772,
-      "step": 21296
-    },
-    {
-      "epoch": 11.10664928292047,
-      "grad_norm": 1.4816921949386597,
-      "learning_rate": 7.910552763819096e-05,
-      "loss": 5.316,
-      "step": 21297
-    },
-    {
-      "epoch": 11.107170795306388,
-      "grad_norm": 1.5816766023635864,
-      "learning_rate": 7.910452261306534e-05,
-      "loss": 5.5382,
-      "step": 21298
-    },
-    {
-      "epoch": 11.107692307692307,
-      "grad_norm": 1.542664885520935,
-      "learning_rate": 7.91035175879397e-05,
-      "loss": 5.1647,
-      "step": 21299
-    },
-    {
-      "epoch": 11.108213820078227,
-      "grad_norm": 1.304136037826538,
-      "learning_rate": 7.910251256281408e-05,
-      "loss": 5.3134,
-      "step": 21300
-    },
-    {
-      "epoch": 11.108735332464146,
-      "grad_norm": 1.442695140838623,
-      "learning_rate": 7.910150753768845e-05,
-      "loss": 5.3785,
-      "step": 21301
-    },
-    {
-      "epoch": 11.109256844850066,
-      "grad_norm": 1.4418717622756958,
-      "learning_rate": 7.910050251256281e-05,
-      "loss": 5.2745,
-      "step": 21302
-    },
-    {
-      "epoch": 11.109778357235985,
-      "grad_norm": 1.3686498403549194,
-      "learning_rate": 7.909949748743719e-05,
-      "loss": 5.9698,
-      "step": 21303
-    },
-    {
-      "epoch": 11.110299869621903,
-      "grad_norm": 1.3885929584503174,
-      "learning_rate": 7.909849246231155e-05,
-      "loss": 5.8157,
-      "step": 21304
-    },
-    {
-      "epoch": 11.110821382007822,
-      "grad_norm": 1.5441056489944458,
-      "learning_rate": 7.909748743718593e-05,
-      "loss": 5.0413,
-      "step": 21305
-    },
-    {
-      "epoch": 11.111342894393742,
-      "grad_norm": 1.4308533668518066,
-      "learning_rate": 7.909648241206031e-05,
-      "loss": 5.5623,
-      "step": 21306
-    },
-    {
-      "epoch": 11.111864406779661,
-      "grad_norm": 1.555012822151184,
-      "learning_rate": 7.909547738693468e-05,
-      "loss": 5.2997,
-      "step": 21307
-    },
-    {
-      "epoch": 11.11238591916558,
-      "grad_norm": 1.4833509922027588,
-      "learning_rate": 7.909447236180905e-05,
-      "loss": 5.687,
-      "step": 21308
-    },
-    {
-      "epoch": 11.1129074315515,
-      "grad_norm": 1.4388275146484375,
-      "learning_rate": 7.909346733668343e-05,
-      "loss": 5.6397,
-      "step": 21309
-    },
-    {
-      "epoch": 11.113428943937418,
-      "grad_norm": 1.5124293565750122,
-      "learning_rate": 7.909246231155779e-05,
-      "loss": 5.5139,
-      "step": 21310
-    },
-    {
-      "epoch": 11.113950456323337,
-      "grad_norm": 1.4840890169143677,
-      "learning_rate": 7.909145728643217e-05,
-      "loss": 5.2564,
-      "step": 21311
-    },
-    {
-      "epoch": 11.114471968709257,
-      "grad_norm": 1.3972749710083008,
-      "learning_rate": 7.909045226130653e-05,
-      "loss": 5.557,
-      "step": 21312
-    },
-    {
-      "epoch": 11.114993481095176,
-      "grad_norm": 1.4226508140563965,
-      "learning_rate": 7.908944723618091e-05,
-      "loss": 5.4972,
-      "step": 21313
-    },
-    {
-      "epoch": 11.115514993481096,
-      "grad_norm": 1.5256398916244507,
-      "learning_rate": 7.908844221105528e-05,
-      "loss": 5.5025,
-      "step": 21314
-    },
-    {
-      "epoch": 11.116036505867015,
-      "grad_norm": 1.3962414264678955,
-      "learning_rate": 7.908743718592965e-05,
-      "loss": 5.6824,
-      "step": 21315
-    },
-    {
-      "epoch": 11.116558018252933,
-      "grad_norm": 1.4412841796875,
-      "learning_rate": 7.908643216080403e-05,
-      "loss": 5.4293,
-      "step": 21316
-    },
-    {
-      "epoch": 11.117079530638852,
-      "grad_norm": 1.4183921813964844,
-      "learning_rate": 7.90854271356784e-05,
-      "loss": 5.1309,
-      "step": 21317
-    },
-    {
-      "epoch": 11.117601043024772,
-      "grad_norm": 1.4119958877563477,
-      "learning_rate": 7.908442211055277e-05,
-      "loss": 5.4706,
-      "step": 21318
-    },
-    {
-      "epoch": 11.118122555410691,
-      "grad_norm": 1.4091073274612427,
-      "learning_rate": 7.908341708542714e-05,
-      "loss": 5.0468,
-      "step": 21319
-    },
-    {
-      "epoch": 11.11864406779661,
-      "grad_norm": 1.403618574142456,
-      "learning_rate": 7.908241206030152e-05,
-      "loss": 5.1639,
-      "step": 21320
-    },
-    {
-      "epoch": 11.11916558018253,
-      "grad_norm": 1.4982681274414062,
-      "learning_rate": 7.908140703517588e-05,
-      "loss": 5.0411,
-      "step": 21321
-    },
-    {
-      "epoch": 11.119687092568448,
-      "grad_norm": 1.6579148769378662,
-      "learning_rate": 7.908040201005026e-05,
-      "loss": 5.2419,
-      "step": 21322
-    },
-    {
-      "epoch": 11.120208604954367,
-      "grad_norm": 1.4114866256713867,
-      "learning_rate": 7.907939698492462e-05,
-      "loss": 5.1317,
-      "step": 21323
-    },
-    {
-      "epoch": 11.120730117340287,
-      "grad_norm": 1.4738303422927856,
-      "learning_rate": 7.9078391959799e-05,
-      "loss": 5.2142,
-      "step": 21324
-    },
-    {
-      "epoch": 11.121251629726206,
-      "grad_norm": 1.3809763193130493,
-      "learning_rate": 7.907738693467336e-05,
-      "loss": 5.7776,
-      "step": 21325
-    },
-    {
-      "epoch": 11.121773142112126,
-      "grad_norm": 1.45130455493927,
-      "learning_rate": 7.907638190954774e-05,
-      "loss": 5.1694,
-      "step": 21326
-    },
-    {
-      "epoch": 11.122294654498045,
-      "grad_norm": 1.4348806142807007,
-      "learning_rate": 7.907537688442212e-05,
-      "loss": 5.5582,
-      "step": 21327
-    },
-    {
-      "epoch": 11.122816166883963,
-      "grad_norm": 1.4996081590652466,
-      "learning_rate": 7.907437185929648e-05,
-      "loss": 5.4694,
-      "step": 21328
-    },
-    {
-      "epoch": 11.123337679269882,
-      "grad_norm": 1.5494446754455566,
-      "learning_rate": 7.907336683417086e-05,
-      "loss": 5.4363,
-      "step": 21329
-    },
-    {
-      "epoch": 11.123859191655802,
-      "grad_norm": 1.4614343643188477,
-      "learning_rate": 7.907236180904523e-05,
-      "loss": 5.332,
-      "step": 21330
-    },
-    {
-      "epoch": 11.124380704041721,
-      "grad_norm": 1.3817330598831177,
-      "learning_rate": 7.90713567839196e-05,
-      "loss": 5.505,
-      "step": 21331
-    },
-    {
-      "epoch": 11.12490221642764,
-      "grad_norm": 1.4325079917907715,
-      "learning_rate": 7.907035175879397e-05,
-      "loss": 5.1667,
-      "step": 21332
-    },
-    {
-      "epoch": 11.125423728813558,
-      "grad_norm": 1.3807177543640137,
-      "learning_rate": 7.906934673366835e-05,
-      "loss": 5.6405,
-      "step": 21333
-    },
-    {
-      "epoch": 11.125945241199478,
-      "grad_norm": 1.494067668914795,
-      "learning_rate": 7.906834170854271e-05,
-      "loss": 5.4016,
-      "step": 21334
-    },
-    {
-      "epoch": 11.126466753585397,
-      "grad_norm": 1.4297558069229126,
-      "learning_rate": 7.906733668341709e-05,
-      "loss": 5.0914,
-      "step": 21335
-    },
-    {
-      "epoch": 11.126988265971317,
-      "grad_norm": 1.7369492053985596,
-      "learning_rate": 7.906633165829147e-05,
-      "loss": 4.714,
-      "step": 21336
-    },
-    {
-      "epoch": 11.127509778357236,
-      "grad_norm": 1.5385626554489136,
-      "learning_rate": 7.906532663316584e-05,
-      "loss": 5.6693,
-      "step": 21337
-    },
-    {
-      "epoch": 11.128031290743156,
-      "grad_norm": 1.4284844398498535,
-      "learning_rate": 7.906432160804021e-05,
-      "loss": 5.6562,
-      "step": 21338
-    },
-    {
-      "epoch": 11.128552803129073,
-      "grad_norm": 1.6436957120895386,
-      "learning_rate": 7.906331658291459e-05,
-      "loss": 5.5379,
-      "step": 21339
-    },
-    {
-      "epoch": 11.129074315514993,
-      "grad_norm": 1.436186671257019,
-      "learning_rate": 7.906231155778895e-05,
-      "loss": 5.345,
-      "step": 21340
-    },
-    {
-      "epoch": 11.129595827900912,
-      "grad_norm": 1.4999115467071533,
-      "learning_rate": 7.906130653266331e-05,
-      "loss": 4.9306,
-      "step": 21341
-    },
-    {
-      "epoch": 11.130117340286832,
-      "grad_norm": 1.4505506753921509,
-      "learning_rate": 7.906030150753769e-05,
-      "loss": 5.1937,
-      "step": 21342
-    },
-    {
-      "epoch": 11.130638852672751,
-      "grad_norm": 1.4610061645507812,
-      "learning_rate": 7.905929648241206e-05,
-      "loss": 5.533,
-      "step": 21343
-    },
-    {
-      "epoch": 11.13116036505867,
-      "grad_norm": 1.4570389986038208,
-      "learning_rate": 7.905829145728643e-05,
-      "loss": 5.2592,
-      "step": 21344
-    },
-    {
-      "epoch": 11.131681877444588,
-      "grad_norm": 1.613128900527954,
-      "learning_rate": 7.90572864321608e-05,
-      "loss": 5.2162,
-      "step": 21345
-    },
-    {
-      "epoch": 11.132203389830508,
-      "grad_norm": 1.397904872894287,
-      "learning_rate": 7.905628140703518e-05,
-      "loss": 5.2699,
-      "step": 21346
-    },
-    {
-      "epoch": 11.132724902216427,
-      "grad_norm": 1.6108289957046509,
-      "learning_rate": 7.905527638190955e-05,
-      "loss": 5.2732,
-      "step": 21347
-    },
-    {
-      "epoch": 11.133246414602347,
-      "grad_norm": 1.4717718362808228,
-      "learning_rate": 7.905427135678393e-05,
-      "loss": 5.6659,
-      "step": 21348
-    },
-    {
-      "epoch": 11.133767926988266,
-      "grad_norm": 1.4224653244018555,
-      "learning_rate": 7.90532663316583e-05,
-      "loss": 5.3043,
-      "step": 21349
-    },
-    {
-      "epoch": 11.134289439374186,
-      "grad_norm": 1.394222378730774,
-      "learning_rate": 7.905226130653267e-05,
-      "loss": 5.5759,
-      "step": 21350
-    },
-    {
-      "epoch": 11.134810951760103,
-      "grad_norm": 1.4094802141189575,
-      "learning_rate": 7.905125628140704e-05,
-      "loss": 5.3345,
-      "step": 21351
-    },
-    {
-      "epoch": 11.135332464146023,
-      "grad_norm": 1.4624834060668945,
-      "learning_rate": 7.905025125628142e-05,
-      "loss": 5.3954,
-      "step": 21352
-    },
-    {
-      "epoch": 11.135853976531942,
-      "grad_norm": 1.5574744939804077,
-      "learning_rate": 7.904924623115578e-05,
-      "loss": 5.4129,
-      "step": 21353
-    },
-    {
-      "epoch": 11.136375488917862,
-      "grad_norm": 1.4083311557769775,
-      "learning_rate": 7.904824120603014e-05,
-      "loss": 5.6245,
-      "step": 21354
-    },
-    {
-      "epoch": 11.136897001303781,
-      "grad_norm": 1.486976981163025,
-      "learning_rate": 7.904723618090452e-05,
-      "loss": 5.0884,
-      "step": 21355
-    },
-    {
-      "epoch": 11.1374185136897,
-      "grad_norm": 1.5600109100341797,
-      "learning_rate": 7.90462311557789e-05,
-      "loss": 5.6982,
-      "step": 21356
-    },
-    {
-      "epoch": 11.137940026075619,
-      "grad_norm": 1.4768006801605225,
-      "learning_rate": 7.904522613065328e-05,
-      "loss": 5.5715,
-      "step": 21357
-    },
-    {
-      "epoch": 11.138461538461538,
-      "grad_norm": 1.51589035987854,
-      "learning_rate": 7.904422110552764e-05,
-      "loss": 5.3454,
-      "step": 21358
-    },
-    {
-      "epoch": 11.138983050847457,
-      "grad_norm": 1.4575436115264893,
-      "learning_rate": 7.904321608040202e-05,
-      "loss": 5.1589,
-      "step": 21359
-    },
-    {
-      "epoch": 11.139504563233377,
-      "grad_norm": 1.401412010192871,
-      "learning_rate": 7.904221105527638e-05,
-      "loss": 5.5201,
-      "step": 21360
-    },
-    {
-      "epoch": 11.140026075619296,
-      "grad_norm": 1.3746368885040283,
-      "learning_rate": 7.904120603015076e-05,
-      "loss": 5.3697,
-      "step": 21361
-    },
-    {
-      "epoch": 11.140547588005216,
-      "grad_norm": 1.4726382493972778,
-      "learning_rate": 7.904020100502513e-05,
-      "loss": 4.7123,
-      "step": 21362
-    },
-    {
-      "epoch": 11.141069100391134,
-      "grad_norm": 1.4307986497879028,
-      "learning_rate": 7.90391959798995e-05,
-      "loss": 5.5249,
-      "step": 21363
-    },
-    {
-      "epoch": 11.141590612777053,
-      "grad_norm": 1.5728458166122437,
-      "learning_rate": 7.903819095477387e-05,
-      "loss": 5.3412,
-      "step": 21364
-    },
-    {
-      "epoch": 11.142112125162972,
-      "grad_norm": 1.357520341873169,
-      "learning_rate": 7.903718592964825e-05,
-      "loss": 5.6215,
-      "step": 21365
-    },
-    {
-      "epoch": 11.142633637548892,
-      "grad_norm": 1.4037086963653564,
-      "learning_rate": 7.903618090452261e-05,
-      "loss": 5.9303,
-      "step": 21366
-    },
-    {
-      "epoch": 11.143155149934811,
-      "grad_norm": 1.530768871307373,
-      "learning_rate": 7.903517587939699e-05,
-      "loss": 5.4398,
-      "step": 21367
-    },
-    {
-      "epoch": 11.14367666232073,
-      "grad_norm": 1.3421982526779175,
-      "learning_rate": 7.903417085427137e-05,
-      "loss": 5.7736,
-      "step": 21368
-    },
-    {
-      "epoch": 11.144198174706649,
-      "grad_norm": 1.3881607055664062,
-      "learning_rate": 7.903316582914573e-05,
-      "loss": 5.3231,
-      "step": 21369
-    },
-    {
-      "epoch": 11.144719687092568,
-      "grad_norm": 1.5353821516036987,
-      "learning_rate": 7.903216080402011e-05,
-      "loss": 5.1406,
-      "step": 21370
-    },
-    {
-      "epoch": 11.145241199478487,
-      "grad_norm": 1.4521106481552124,
-      "learning_rate": 7.903115577889447e-05,
-      "loss": 5.681,
-      "step": 21371
-    },
-    {
-      "epoch": 11.145762711864407,
-      "grad_norm": 1.4113845825195312,
-      "learning_rate": 7.903015075376885e-05,
-      "loss": 5.6468,
-      "step": 21372
-    },
-    {
-      "epoch": 11.146284224250326,
-      "grad_norm": 1.3554868698120117,
-      "learning_rate": 7.902914572864321e-05,
-      "loss": 5.5326,
-      "step": 21373
-    },
-    {
-      "epoch": 11.146805736636246,
-      "grad_norm": 1.4909281730651855,
-      "learning_rate": 7.902814070351759e-05,
-      "loss": 5.0995,
-      "step": 21374
-    },
-    {
-      "epoch": 11.147327249022164,
-      "grad_norm": 1.44448721408844,
-      "learning_rate": 7.902713567839196e-05,
-      "loss": 5.7707,
-      "step": 21375
-    },
-    {
-      "epoch": 11.147848761408083,
-      "grad_norm": 1.5857930183410645,
-      "learning_rate": 7.902613065326633e-05,
-      "loss": 5.0657,
-      "step": 21376
-    },
-    {
-      "epoch": 11.148370273794002,
-      "grad_norm": 1.6383559703826904,
-      "learning_rate": 7.902512562814071e-05,
-      "loss": 5.6202,
-      "step": 21377
-    },
-    {
-      "epoch": 11.148891786179922,
-      "grad_norm": 1.4449540376663208,
-      "learning_rate": 7.902412060301509e-05,
-      "loss": 5.4997,
-      "step": 21378
-    },
-    {
-      "epoch": 11.149413298565841,
-      "grad_norm": 1.4250297546386719,
-      "learning_rate": 7.902311557788945e-05,
-      "loss": 5.4367,
-      "step": 21379
-    },
-    {
-      "epoch": 11.14993481095176,
-      "grad_norm": 1.3871220350265503,
-      "learning_rate": 7.902211055276383e-05,
-      "loss": 5.5282,
-      "step": 21380
-    },
-    {
-      "epoch": 11.150456323337679,
-      "grad_norm": 1.36214280128479,
-      "learning_rate": 7.90211055276382e-05,
-      "loss": 5.8248,
-      "step": 21381
-    },
-    {
-      "epoch": 11.150977835723598,
-      "grad_norm": 1.403998851776123,
-      "learning_rate": 7.902010050251256e-05,
-      "loss": 5.2967,
-      "step": 21382
-    },
-    {
-      "epoch": 11.151499348109517,
-      "grad_norm": 1.639087438583374,
-      "learning_rate": 7.901909547738694e-05,
-      "loss": 5.7568,
-      "step": 21383
-    },
-    {
-      "epoch": 11.152020860495437,
-      "grad_norm": 1.3040581941604614,
-      "learning_rate": 7.90180904522613e-05,
-      "loss": 4.9459,
-      "step": 21384
-    },
-    {
-      "epoch": 11.152542372881356,
-      "grad_norm": 1.278948426246643,
-      "learning_rate": 7.901708542713568e-05,
-      "loss": 5.5898,
-      "step": 21385
-    },
-    {
-      "epoch": 11.153063885267276,
-      "grad_norm": 1.3205883502960205,
-      "learning_rate": 7.901608040201005e-05,
-      "loss": 4.7975,
-      "step": 21386
-    },
-    {
-      "epoch": 11.153585397653194,
-      "grad_norm": 1.4380989074707031,
-      "learning_rate": 7.901507537688442e-05,
-      "loss": 5.5905,
-      "step": 21387
-    },
-    {
-      "epoch": 11.154106910039113,
-      "grad_norm": 1.3458704948425293,
-      "learning_rate": 7.90140703517588e-05,
-      "loss": 4.9709,
-      "step": 21388
-    },
-    {
-      "epoch": 11.154628422425032,
-      "grad_norm": 1.361082911491394,
-      "learning_rate": 7.901306532663318e-05,
-      "loss": 5.1245,
-      "step": 21389
-    },
-    {
-      "epoch": 11.155149934810952,
-      "grad_norm": Infinity,
-      "learning_rate": 7.901306532663318e-05,
-      "loss": 5.8181,
-      "step": 21390
-    },
-    {
-      "epoch": 11.155671447196871,
-      "grad_norm": 1.5814590454101562,
-      "learning_rate": 7.901206030150754e-05,
-      "loss": 5.4896,
-      "step": 21391
-    },
-    {
-      "epoch": 11.156192959582791,
-      "grad_norm": 1.4058235883712769,
-      "learning_rate": 7.901105527638192e-05,
-      "loss": 5.4626,
-      "step": 21392
-    },
-    {
-      "epoch": 11.156714471968709,
-      "grad_norm": 1.3964712619781494,
-      "learning_rate": 7.901005025125629e-05,
-      "loss": 5.7314,
-      "step": 21393
-    },
-    {
-      "epoch": 11.157235984354628,
-      "grad_norm": 1.4859364032745361,
-      "learning_rate": 7.900904522613066e-05,
-      "loss": 5.5918,
-      "step": 21394
-    },
-    {
-      "epoch": 11.157757496740548,
-      "grad_norm": 1.5116446018218994,
-      "learning_rate": 7.900804020100503e-05,
-      "loss": 5.2633,
-      "step": 21395
-    },
-    {
-      "epoch": 11.158279009126467,
-      "grad_norm": 1.4421643018722534,
-      "learning_rate": 7.900703517587939e-05,
-      "loss": 5.8496,
-      "step": 21396
-    },
-    {
-      "epoch": 11.158800521512386,
-      "grad_norm": 1.382397174835205,
-      "learning_rate": 7.900603015075377e-05,
-      "loss": 5.7319,
-      "step": 21397
-    },
-    {
-      "epoch": 11.159322033898306,
-      "grad_norm": 1.5418555736541748,
-      "learning_rate": 7.900502512562813e-05,
-      "loss": 5.4556,
-      "step": 21398
-    },
-    {
-      "epoch": 11.159843546284224,
-      "grad_norm": 1.386739730834961,
-      "learning_rate": 7.900402010050251e-05,
-      "loss": 5.3852,
-      "step": 21399
-    },
-    {
-      "epoch": 11.160365058670143,
-      "grad_norm": 1.4956578016281128,
-      "learning_rate": 7.900301507537689e-05,
-      "loss": 4.8532,
-      "step": 21400
-    },
-    {
-      "epoch": 11.160886571056063,
-      "grad_norm": 1.5478298664093018,
-      "learning_rate": 7.900201005025127e-05,
-      "loss": 5.3722,
-      "step": 21401
-    },
-    {
-      "epoch": 11.161408083441982,
-      "grad_norm": 1.5793441534042358,
-      "learning_rate": 7.900100502512563e-05,
-      "loss": 5.3935,
-      "step": 21402
-    },
-    {
-      "epoch": 11.161929595827901,
-      "grad_norm": 1.5033138990402222,
-      "learning_rate": 7.900000000000001e-05,
-      "loss": 5.5907,
-      "step": 21403
-    },
-    {
-      "epoch": 11.162451108213821,
-      "grad_norm": 1.4276869297027588,
-      "learning_rate": 7.899899497487437e-05,
-      "loss": 4.8596,
-      "step": 21404
-    },
-    {
-      "epoch": 11.162972620599739,
-      "grad_norm": 1.4800950288772583,
-      "learning_rate": 7.899798994974875e-05,
-      "loss": 5.347,
-      "step": 21405
-    },
-    {
-      "epoch": 11.163494132985658,
-      "grad_norm": 1.3886867761611938,
-      "learning_rate": 7.899698492462312e-05,
-      "loss": 5.5428,
-      "step": 21406
-    },
-    {
-      "epoch": 11.164015645371578,
-      "grad_norm": 1.3459917306900024,
-      "learning_rate": 7.89959798994975e-05,
-      "loss": 5.7811,
-      "step": 21407
-    },
-    {
-      "epoch": 11.164537157757497,
-      "grad_norm": 1.4136309623718262,
-      "learning_rate": 7.899497487437186e-05,
-      "loss": 5.7765,
-      "step": 21408
-    },
-    {
-      "epoch": 11.165058670143416,
-      "grad_norm": 1.4757213592529297,
-      "learning_rate": 7.899396984924624e-05,
-      "loss": 5.5431,
-      "step": 21409
-    },
-    {
-      "epoch": 11.165580182529336,
-      "grad_norm": 1.416334867477417,
-      "learning_rate": 7.899296482412061e-05,
-      "loss": 5.4526,
-      "step": 21410
-    },
-    {
-      "epoch": 11.166101694915254,
-      "grad_norm": 1.4864475727081299,
-      "learning_rate": 7.899195979899498e-05,
-      "loss": 5.3419,
-      "step": 21411
-    },
-    {
-      "epoch": 11.166623207301173,
-      "grad_norm": 1.5004969835281372,
-      "learning_rate": 7.899095477386936e-05,
-      "loss": 5.3685,
-      "step": 21412
-    },
-    {
-      "epoch": 11.167144719687093,
-      "grad_norm": 1.4711356163024902,
-      "learning_rate": 7.898994974874372e-05,
-      "loss": 5.3128,
-      "step": 21413
-    },
-    {
-      "epoch": 11.167666232073012,
-      "grad_norm": 1.4084699153900146,
-      "learning_rate": 7.89889447236181e-05,
-      "loss": 5.5163,
-      "step": 21414
-    },
-    {
-      "epoch": 11.168187744458931,
-      "grad_norm": 1.5308912992477417,
-      "learning_rate": 7.898793969849246e-05,
-      "loss": 5.2395,
-      "step": 21415
-    },
-    {
-      "epoch": 11.16870925684485,
-      "grad_norm": 1.468528389930725,
-      "learning_rate": 7.898693467336684e-05,
-      "loss": 5.1627,
-      "step": 21416
-    },
-    {
-      "epoch": 11.169230769230769,
-      "grad_norm": 1.4206342697143555,
-      "learning_rate": 7.89859296482412e-05,
-      "loss": 5.3648,
-      "step": 21417
-    },
-    {
-      "epoch": 11.169752281616688,
-      "grad_norm": 1.3944761753082275,
-      "learning_rate": 7.898492462311558e-05,
-      "loss": 6.0184,
-      "step": 21418
-    },
-    {
-      "epoch": 11.170273794002608,
-      "grad_norm": 1.4147226810455322,
-      "learning_rate": 7.898391959798995e-05,
-      "loss": 5.47,
-      "step": 21419
-    },
-    {
-      "epoch": 11.170795306388527,
-      "grad_norm": 3.2456023693084717,
-      "learning_rate": 7.898291457286432e-05,
-      "loss": 5.7947,
-      "step": 21420
-    },
-    {
-      "epoch": 11.171316818774446,
-      "grad_norm": 1.4755452871322632,
-      "learning_rate": 7.89819095477387e-05,
-      "loss": 5.4626,
-      "step": 21421
-    },
-    {
-      "epoch": 11.171838331160366,
-      "grad_norm": 1.5307981967926025,
-      "learning_rate": 7.898090452261307e-05,
-      "loss": 5.3972,
-      "step": 21422
-    },
-    {
-      "epoch": 11.172359843546284,
-      "grad_norm": 2.9035685062408447,
-      "learning_rate": 7.897989949748744e-05,
-      "loss": 5.531,
-      "step": 21423
-    },
-    {
-      "epoch": 11.172881355932203,
-      "grad_norm": 1.494965672492981,
-      "learning_rate": 7.897889447236181e-05,
-      "loss": 5.274,
-      "step": 21424
-    },
-    {
-      "epoch": 11.173402868318123,
-      "grad_norm": 1.4702191352844238,
-      "learning_rate": 7.897788944723619e-05,
-      "loss": 4.7638,
-      "step": 21425
-    },
-    {
-      "epoch": 11.173924380704042,
-      "grad_norm": 1.4538065195083618,
-      "learning_rate": 7.897688442211055e-05,
-      "loss": 5.9042,
-      "step": 21426
-    },
-    {
-      "epoch": 11.174445893089962,
-      "grad_norm": 1.4500160217285156,
-      "learning_rate": 7.897587939698493e-05,
-      "loss": 5.7525,
-      "step": 21427
-    },
-    {
-      "epoch": 11.17496740547588,
-      "grad_norm": 1.401063084602356,
-      "learning_rate": 7.897487437185929e-05,
-      "loss": 5.2538,
-      "step": 21428
-    },
-    {
-      "epoch": 11.175488917861799,
-      "grad_norm": 1.4053161144256592,
-      "learning_rate": 7.897386934673367e-05,
-      "loss": 5.3865,
-      "step": 21429
-    },
-    {
-      "epoch": 11.176010430247718,
-      "grad_norm": 1.527596116065979,
-      "learning_rate": 7.897286432160805e-05,
-      "loss": 5.289,
-      "step": 21430
-    },
-    {
-      "epoch": 11.176531942633638,
-      "grad_norm": 1.3261667490005493,
-      "learning_rate": 7.897185929648243e-05,
-      "loss": 5.9546,
-      "step": 21431
-    },
-    {
-      "epoch": 11.177053455019557,
-      "grad_norm": 1.4866340160369873,
-      "learning_rate": 7.897085427135679e-05,
-      "loss": 5.8646,
-      "step": 21432
-    },
-    {
-      "epoch": 11.177574967405477,
-      "grad_norm": 1.4691447019577026,
-      "learning_rate": 7.896984924623117e-05,
-      "loss": 5.6522,
-      "step": 21433
-    },
-    {
-      "epoch": 11.178096479791394,
-      "grad_norm": 1.543697714805603,
-      "learning_rate": 7.896884422110553e-05,
-      "loss": 5.3257,
-      "step": 21434
-    },
-    {
-      "epoch": 11.178617992177314,
-      "grad_norm": 1.3420677185058594,
-      "learning_rate": 7.89678391959799e-05,
-      "loss": 5.9075,
-      "step": 21435
-    },
-    {
-      "epoch": 11.179139504563233,
-      "grad_norm": 1.4195939302444458,
-      "learning_rate": 7.896683417085427e-05,
-      "loss": 5.4083,
-      "step": 21436
-    },
-    {
-      "epoch": 11.179661016949153,
-      "grad_norm": 1.4105525016784668,
-      "learning_rate": 7.896582914572864e-05,
-      "loss": 5.5105,
-      "step": 21437
-    },
-    {
-      "epoch": 11.180182529335072,
-      "grad_norm": 1.4549907445907593,
-      "learning_rate": 7.896482412060302e-05,
-      "loss": 5.7765,
-      "step": 21438
-    },
-    {
-      "epoch": 11.180704041720992,
-      "grad_norm": 1.4540163278579712,
-      "learning_rate": 7.896381909547738e-05,
-      "loss": 5.6389,
-      "step": 21439
-    },
-    {
-      "epoch": 11.18122555410691,
-      "grad_norm": 1.4632179737091064,
-      "learning_rate": 7.896281407035176e-05,
-      "loss": 5.4398,
-      "step": 21440
-    },
-    {
-      "epoch": 11.181747066492829,
-      "grad_norm": 1.51848566532135,
-      "learning_rate": 7.896180904522614e-05,
-      "loss": 5.5185,
-      "step": 21441
-    },
-    {
-      "epoch": 11.182268578878748,
-      "grad_norm": 1.4623430967330933,
-      "learning_rate": 7.896080402010051e-05,
-      "loss": 4.9591,
-      "step": 21442
-    },
-    {
-      "epoch": 11.182790091264668,
-      "grad_norm": 1.3230340480804443,
-      "learning_rate": 7.895979899497488e-05,
-      "loss": 4.9826,
-      "step": 21443
-    },
-    {
-      "epoch": 11.183311603650587,
-      "grad_norm": 1.5166327953338623,
-      "learning_rate": 7.895879396984926e-05,
-      "loss": 5.5543,
-      "step": 21444
-    },
-    {
-      "epoch": 11.183833116036507,
-      "grad_norm": 1.4975099563598633,
-      "learning_rate": 7.895778894472362e-05,
-      "loss": 5.4934,
-      "step": 21445
-    },
-    {
-      "epoch": 11.184354628422424,
-      "grad_norm": 1.6669707298278809,
-      "learning_rate": 7.8956783919598e-05,
-      "loss": 5.2082,
-      "step": 21446
-    },
-    {
-      "epoch": 11.184876140808344,
-      "grad_norm": 1.4566094875335693,
-      "learning_rate": 7.895577889447236e-05,
-      "loss": 5.6995,
-      "step": 21447
-    },
-    {
-      "epoch": 11.185397653194263,
-      "grad_norm": 1.5891447067260742,
-      "learning_rate": 7.895477386934673e-05,
-      "loss": 4.9566,
-      "step": 21448
-    },
-    {
-      "epoch": 11.185919165580183,
-      "grad_norm": 1.4323794841766357,
-      "learning_rate": 7.89537688442211e-05,
-      "loss": 5.5042,
-      "step": 21449
-    },
-    {
-      "epoch": 11.186440677966102,
-      "grad_norm": 1.467942714691162,
-      "learning_rate": 7.895276381909548e-05,
-      "loss": 5.5617,
-      "step": 21450
-    },
-    {
-      "epoch": 11.186962190352022,
-      "grad_norm": 1.426840901374817,
-      "learning_rate": 7.895175879396986e-05,
-      "loss": 5.4699,
-      "step": 21451
-    },
-    {
-      "epoch": 11.18748370273794,
-      "grad_norm": 1.440025806427002,
-      "learning_rate": 7.895075376884422e-05,
-      "loss": 5.5095,
-      "step": 21452
-    },
-    {
-      "epoch": 11.188005215123859,
-      "grad_norm": 1.4048981666564941,
-      "learning_rate": 7.89497487437186e-05,
-      "loss": 5.5936,
-      "step": 21453
-    },
-    {
-      "epoch": 11.188526727509778,
-      "grad_norm": 1.489065408706665,
-      "learning_rate": 7.894874371859297e-05,
-      "loss": 5.026,
-      "step": 21454
-    },
-    {
-      "epoch": 11.189048239895698,
-      "grad_norm": 1.4639655351638794,
-      "learning_rate": 7.894773869346734e-05,
-      "loss": 5.6209,
-      "step": 21455
-    },
-    {
-      "epoch": 11.189569752281617,
-      "grad_norm": 1.33316969871521,
-      "learning_rate": 7.894673366834171e-05,
-      "loss": 5.4306,
-      "step": 21456
-    },
-    {
-      "epoch": 11.190091264667537,
-      "grad_norm": 1.4886330366134644,
-      "learning_rate": 7.894572864321609e-05,
-      "loss": 5.479,
-      "step": 21457
-    },
-    {
-      "epoch": 11.190612777053454,
-      "grad_norm": 1.4789910316467285,
-      "learning_rate": 7.894472361809045e-05,
-      "loss": 5.1823,
-      "step": 21458
-    },
-    {
-      "epoch": 11.191134289439374,
-      "grad_norm": 1.5499558448791504,
-      "learning_rate": 7.894371859296483e-05,
-      "loss": 5.1691,
-      "step": 21459
-    },
-    {
-      "epoch": 11.191655801825293,
-      "grad_norm": 1.8193312883377075,
-      "learning_rate": 7.894271356783919e-05,
-      "loss": 5.1624,
-      "step": 21460
-    },
-    {
-      "epoch": 11.192177314211213,
-      "grad_norm": 1.4061253070831299,
-      "learning_rate": 7.894170854271357e-05,
-      "loss": 5.6306,
-      "step": 21461
-    },
-    {
-      "epoch": 11.192698826597132,
-      "grad_norm": 1.492948293685913,
-      "learning_rate": 7.894070351758795e-05,
-      "loss": 5.3694,
-      "step": 21462
-    },
-    {
-      "epoch": 11.193220338983052,
-      "grad_norm": 1.3546621799468994,
-      "learning_rate": 7.893969849246231e-05,
-      "loss": 5.7486,
-      "step": 21463
-    },
-    {
-      "epoch": 11.19374185136897,
-      "grad_norm": 1.536510705947876,
-      "learning_rate": 7.893869346733669e-05,
-      "loss": 5.4113,
-      "step": 21464
-    },
-    {
-      "epoch": 11.194263363754889,
-      "grad_norm": 1.447645902633667,
-      "learning_rate": 7.893768844221106e-05,
-      "loss": 5.5825,
-      "step": 21465
-    },
-    {
-      "epoch": 11.194784876140808,
-      "grad_norm": 1.5211478471755981,
-      "learning_rate": 7.893668341708543e-05,
-      "loss": 5.4131,
-      "step": 21466
-    },
-    {
-      "epoch": 11.195306388526728,
-      "grad_norm": 1.4291818141937256,
-      "learning_rate": 7.89356783919598e-05,
-      "loss": 5.5831,
-      "step": 21467
-    },
-    {
-      "epoch": 11.195827900912647,
-      "grad_norm": 1.3633430004119873,
-      "learning_rate": 7.893467336683418e-05,
-      "loss": 5.8668,
-      "step": 21468
-    },
-    {
-      "epoch": 11.196349413298567,
-      "grad_norm": 1.5488660335540771,
-      "learning_rate": 7.893366834170854e-05,
-      "loss": 5.2447,
-      "step": 21469
-    },
-    {
-      "epoch": 11.196870925684484,
-      "grad_norm": 1.4044426679611206,
-      "learning_rate": 7.893266331658292e-05,
-      "loss": 5.3241,
-      "step": 21470
-    },
-    {
-      "epoch": 11.197392438070404,
-      "grad_norm": 1.3618015050888062,
-      "learning_rate": 7.89316582914573e-05,
-      "loss": 5.6438,
-      "step": 21471
-    },
-    {
-      "epoch": 11.197913950456323,
-      "grad_norm": 1.4643442630767822,
-      "learning_rate": 7.893065326633167e-05,
-      "loss": 5.4334,
-      "step": 21472
-    },
-    {
-      "epoch": 11.198435462842243,
-      "grad_norm": 1.4577127695083618,
-      "learning_rate": 7.892964824120604e-05,
-      "loss": 5.5301,
-      "step": 21473
-    },
-    {
-      "epoch": 11.198956975228162,
-      "grad_norm": 1.4671868085861206,
-      "learning_rate": 7.892864321608041e-05,
-      "loss": 5.4253,
-      "step": 21474
-    },
-    {
-      "epoch": 11.199478487614082,
-      "grad_norm": 1.3992878198623657,
-      "learning_rate": 7.892763819095478e-05,
-      "loss": 5.0573,
-      "step": 21475
-    },
-    {
-      "epoch": 11.2,
-      "grad_norm": 2.1125471591949463,
-      "learning_rate": 7.892663316582914e-05,
-      "loss": 4.505,
-      "step": 21476
-    },
-    {
-      "epoch": 11.200521512385919,
-      "grad_norm": 1.4772303104400635,
-      "learning_rate": 7.892562814070352e-05,
-      "loss": 5.6599,
-      "step": 21477
-    },
-    {
-      "epoch": 11.201043024771838,
-      "grad_norm": 1.5771074295043945,
-      "learning_rate": 7.892462311557789e-05,
-      "loss": 5.3751,
-      "step": 21478
-    },
-    {
-      "epoch": 11.201564537157758,
-      "grad_norm": 1.535473108291626,
-      "learning_rate": 7.892361809045226e-05,
-      "loss": 5.4789,
-      "step": 21479
-    },
-    {
-      "epoch": 11.202086049543677,
-      "grad_norm": 1.5222721099853516,
-      "learning_rate": 7.892261306532663e-05,
-      "loss": 5.4958,
-      "step": 21480
-    },
-    {
-      "epoch": 11.202607561929597,
-      "grad_norm": 1.3667140007019043,
-      "learning_rate": 7.8921608040201e-05,
-      "loss": 5.8466,
-      "step": 21481
-    },
-    {
-      "epoch": 11.203129074315514,
-      "grad_norm": 1.3408379554748535,
-      "learning_rate": 7.892060301507538e-05,
-      "loss": 5.286,
-      "step": 21482
-    },
-    {
-      "epoch": 11.203650586701434,
-      "grad_norm": 1.5271731615066528,
-      "learning_rate": 7.891959798994976e-05,
-      "loss": 5.2534,
-      "step": 21483
-    },
-    {
-      "epoch": 11.204172099087353,
-      "grad_norm": 1.3228726387023926,
-      "learning_rate": 7.891859296482413e-05,
-      "loss": 5.4423,
-      "step": 21484
-    },
-    {
-      "epoch": 11.204693611473273,
-      "grad_norm": 1.3987928628921509,
-      "learning_rate": 7.89175879396985e-05,
-      "loss": 5.3836,
-      "step": 21485
-    },
-    {
-      "epoch": 11.205215123859192,
-      "grad_norm": 1.3483314514160156,
-      "learning_rate": 7.891658291457287e-05,
-      "loss": 5.541,
-      "step": 21486
-    },
-    {
-      "epoch": 11.205736636245112,
-      "grad_norm": 1.4330178499221802,
-      "learning_rate": 7.891557788944725e-05,
-      "loss": 5.9208,
-      "step": 21487
-    },
-    {
-      "epoch": 11.20625814863103,
-      "grad_norm": 1.3872272968292236,
-      "learning_rate": 7.891457286432161e-05,
-      "loss": 5.4811,
-      "step": 21488
-    },
-    {
-      "epoch": 11.206779661016949,
-      "grad_norm": 1.4325693845748901,
-      "learning_rate": 7.891356783919597e-05,
-      "loss": 5.3847,
-      "step": 21489
-    },
-    {
-      "epoch": 11.207301173402868,
-      "grad_norm": 1.4498069286346436,
-      "learning_rate": 7.891256281407035e-05,
-      "loss": 4.9195,
-      "step": 21490
-    },
-    {
-      "epoch": 11.207822685788788,
-      "grad_norm": 1.5049388408660889,
-      "learning_rate": 7.891155778894473e-05,
-      "loss": 5.5336,
-      "step": 21491
-    },
-    {
-      "epoch": 11.208344198174707,
-      "grad_norm": 1.4558861255645752,
-      "learning_rate": 7.891055276381911e-05,
-      "loss": 5.3999,
-      "step": 21492
-    },
-    {
-      "epoch": 11.208865710560627,
-      "grad_norm": 1.4831517934799194,
-      "learning_rate": 7.890954773869347e-05,
-      "loss": 5.5013,
-      "step": 21493
-    },
-    {
-      "epoch": 11.209387222946544,
-      "grad_norm": 2.994799852371216,
-      "learning_rate": 7.890854271356785e-05,
-      "loss": 5.5826,
-      "step": 21494
-    },
-    {
-      "epoch": 11.209908735332464,
-      "grad_norm": 1.4364490509033203,
-      "learning_rate": 7.890753768844221e-05,
-      "loss": 5.5178,
-      "step": 21495
-    },
-    {
-      "epoch": 11.210430247718383,
-      "grad_norm": 1.4685165882110596,
-      "learning_rate": 7.890653266331659e-05,
-      "loss": 5.4407,
-      "step": 21496
-    },
-    {
-      "epoch": 11.210951760104303,
-      "grad_norm": 1.5045666694641113,
-      "learning_rate": 7.890552763819096e-05,
-      "loss": 5.4671,
-      "step": 21497
-    },
-    {
-      "epoch": 11.211473272490222,
-      "grad_norm": 1.4626744985580444,
-      "learning_rate": 7.890452261306533e-05,
-      "loss": 5.6034,
-      "step": 21498
-    },
-    {
-      "epoch": 11.211994784876142,
-      "grad_norm": 1.568367600440979,
-      "learning_rate": 7.89035175879397e-05,
-      "loss": 5.5173,
-      "step": 21499
-    },
-    {
-      "epoch": 11.21251629726206,
-      "grad_norm": 1.5381991863250732,
-      "learning_rate": 7.890251256281408e-05,
-      "loss": 5.233,
-      "step": 21500
-    },
-    {
-      "epoch": 11.213037809647979,
-      "grad_norm": 1.5123845338821411,
-      "learning_rate": 7.890150753768844e-05,
-      "loss": 5.6484,
-      "step": 21501
-    },
-    {
-      "epoch": 11.213559322033898,
-      "grad_norm": 1.4461729526519775,
-      "learning_rate": 7.890050251256282e-05,
-      "loss": 4.9405,
-      "step": 21502
-    },
-    {
-      "epoch": 11.214080834419818,
-      "grad_norm": 1.5219615697860718,
-      "learning_rate": 7.88994974874372e-05,
-      "loss": 5.0119,
-      "step": 21503
-    },
-    {
-      "epoch": 11.214602346805737,
-      "grad_norm": 1.4035308361053467,
-      "learning_rate": 7.889849246231156e-05,
-      "loss": 5.2054,
-      "step": 21504
-    },
-    {
-      "epoch": 11.215123859191657,
-      "grad_norm": 1.4346859455108643,
-      "learning_rate": 7.889748743718594e-05,
-      "loss": 5.1466,
-      "step": 21505
-    },
-    {
-      "epoch": 11.215645371577574,
-      "grad_norm": 1.3500901460647583,
-      "learning_rate": 7.88964824120603e-05,
-      "loss": 5.7903,
-      "step": 21506
-    },
-    {
-      "epoch": 11.216166883963494,
-      "grad_norm": 1.507265567779541,
-      "learning_rate": 7.889547738693468e-05,
-      "loss": 5.3845,
-      "step": 21507
-    },
-    {
-      "epoch": 11.216688396349413,
-      "grad_norm": 1.367097020149231,
-      "learning_rate": 7.889447236180904e-05,
-      "loss": 5.664,
-      "step": 21508
-    },
-    {
-      "epoch": 11.217209908735333,
-      "grad_norm": 1.3703149557113647,
-      "learning_rate": 7.889346733668342e-05,
-      "loss": 5.2935,
-      "step": 21509
-    },
-    {
-      "epoch": 11.217731421121252,
-      "grad_norm": 1.3428248167037964,
-      "learning_rate": 7.889246231155779e-05,
-      "loss": 5.6581,
-      "step": 21510
-    },
-    {
-      "epoch": 11.21825293350717,
-      "grad_norm": 1.4297832250595093,
-      "learning_rate": 7.889145728643216e-05,
-      "loss": 5.6598,
-      "step": 21511
-    },
-    {
-      "epoch": 11.21877444589309,
-      "grad_norm": 1.5338757038116455,
-      "learning_rate": 7.889045226130654e-05,
-      "loss": 5.3501,
-      "step": 21512
-    },
-    {
-      "epoch": 11.219295958279009,
-      "grad_norm": 1.6730594635009766,
-      "learning_rate": 7.888944723618092e-05,
-      "loss": 4.9975,
-      "step": 21513
-    },
-    {
-      "epoch": 11.219817470664928,
-      "grad_norm": 1.3663370609283447,
-      "learning_rate": 7.888844221105528e-05,
-      "loss": 5.6281,
-      "step": 21514
-    },
-    {
-      "epoch": 11.220338983050848,
-      "grad_norm": 1.5572341680526733,
-      "learning_rate": 7.888743718592965e-05,
-      "loss": 5.0866,
-      "step": 21515
-    },
-    {
-      "epoch": 11.220860495436767,
-      "grad_norm": 1.3738493919372559,
-      "learning_rate": 7.888643216080403e-05,
-      "loss": 5.4958,
-      "step": 21516
-    },
-    {
-      "epoch": 11.221382007822687,
-      "grad_norm": 1.4467744827270508,
-      "learning_rate": 7.888542713567839e-05,
-      "loss": 5.3352,
-      "step": 21517
-    },
-    {
-      "epoch": 11.221903520208604,
-      "grad_norm": 1.4745378494262695,
-      "learning_rate": 7.888442211055277e-05,
-      "loss": 5.568,
-      "step": 21518
-    },
-    {
-      "epoch": 11.222425032594524,
-      "grad_norm": 1.412121295928955,
-      "learning_rate": 7.888341708542713e-05,
-      "loss": 5.6383,
-      "step": 21519
-    },
-    {
-      "epoch": 11.222946544980443,
-      "grad_norm": 1.3945833444595337,
-      "learning_rate": 7.888241206030151e-05,
-      "loss": 5.5991,
-      "step": 21520
-    },
-    {
-      "epoch": 11.223468057366363,
-      "grad_norm": 1.450310230255127,
-      "learning_rate": 7.888140703517587e-05,
-      "loss": 5.6653,
-      "step": 21521
-    },
-    {
-      "epoch": 11.223989569752282,
-      "grad_norm": 1.5411611795425415,
-      "learning_rate": 7.888040201005025e-05,
-      "loss": 5.6859,
-      "step": 21522
-    },
-    {
-      "epoch": 11.2245110821382,
-      "grad_norm": 1.3612892627716064,
-      "learning_rate": 7.887939698492463e-05,
-      "loss": 5.5687,
-      "step": 21523
-    },
-    {
-      "epoch": 11.22503259452412,
-      "grad_norm": 1.3785853385925293,
-      "learning_rate": 7.887839195979901e-05,
-      "loss": 5.5774,
-      "step": 21524
-    },
-    {
-      "epoch": 11.225554106910039,
-      "grad_norm": 1.4974998235702515,
-      "learning_rate": 7.887738693467337e-05,
-      "loss": 5.3532,
-      "step": 21525
-    },
-    {
-      "epoch": 11.226075619295958,
-      "grad_norm": 1.465301513671875,
-      "learning_rate": 7.887638190954775e-05,
-      "loss": 5.0989,
-      "step": 21526
-    },
-    {
-      "epoch": 11.226597131681878,
-      "grad_norm": 1.4580435752868652,
-      "learning_rate": 7.887537688442211e-05,
-      "loss": 4.9424,
-      "step": 21527
-    },
-    {
-      "epoch": 11.227118644067797,
-      "grad_norm": 1.5323933362960815,
-      "learning_rate": 7.887437185929648e-05,
-      "loss": 5.7789,
-      "step": 21528
-    },
-    {
-      "epoch": 11.227640156453715,
-      "grad_norm": 1.4701049327850342,
-      "learning_rate": 7.887336683417086e-05,
-      "loss": 5.3661,
-      "step": 21529
-    },
-    {
-      "epoch": 11.228161668839634,
-      "grad_norm": 1.5602619647979736,
-      "learning_rate": 7.887236180904522e-05,
-      "loss": 5.1222,
-      "step": 21530
-    },
-    {
-      "epoch": 11.228683181225554,
-      "grad_norm": 1.447422981262207,
-      "learning_rate": 7.88713567839196e-05,
-      "loss": 5.3228,
-      "step": 21531
-    },
-    {
-      "epoch": 11.229204693611473,
-      "grad_norm": 1.4521474838256836,
-      "learning_rate": 7.887035175879398e-05,
-      "loss": 5.6748,
-      "step": 21532
-    },
-    {
-      "epoch": 11.229726205997393,
-      "grad_norm": 1.4486596584320068,
-      "learning_rate": 7.886934673366835e-05,
-      "loss": 5.1863,
-      "step": 21533
-    },
-    {
-      "epoch": 11.230247718383312,
-      "grad_norm": 1.4275764226913452,
-      "learning_rate": 7.886834170854272e-05,
-      "loss": 5.4241,
-      "step": 21534
-    },
-    {
-      "epoch": 11.23076923076923,
-      "grad_norm": 1.5261961221694946,
-      "learning_rate": 7.88673366834171e-05,
-      "loss": 5.446,
-      "step": 21535
-    },
-    {
-      "epoch": 11.23129074315515,
-      "grad_norm": 1.4586701393127441,
-      "learning_rate": 7.886633165829146e-05,
-      "loss": 5.6475,
-      "step": 21536
-    },
-    {
-      "epoch": 11.231812255541069,
-      "grad_norm": 1.4515610933303833,
-      "learning_rate": 7.886532663316584e-05,
-      "loss": 5.5842,
-      "step": 21537
-    },
-    {
-      "epoch": 11.232333767926988,
-      "grad_norm": 1.6099001169204712,
-      "learning_rate": 7.88643216080402e-05,
-      "loss": 5.3886,
-      "step": 21538
-    },
-    {
-      "epoch": 11.232855280312908,
-      "grad_norm": 1.477060079574585,
-      "learning_rate": 7.886331658291458e-05,
-      "loss": 5.6424,
-      "step": 21539
-    },
-    {
-      "epoch": 11.233376792698827,
-      "grad_norm": 1.465377688407898,
-      "learning_rate": 7.886231155778895e-05,
-      "loss": 5.4489,
-      "step": 21540
-    },
-    {
-      "epoch": 11.233898305084745,
-      "grad_norm": 1.496896505355835,
-      "learning_rate": 7.886130653266331e-05,
-      "loss": 5.2585,
-      "step": 21541
-    },
-    {
-      "epoch": 11.234419817470664,
-      "grad_norm": 1.3949434757232666,
-      "learning_rate": 7.886030150753769e-05,
-      "loss": 4.9599,
-      "step": 21542
-    },
-    {
-      "epoch": 11.234941329856584,
-      "grad_norm": 1.6489702463150024,
-      "learning_rate": 7.885929648241206e-05,
-      "loss": 4.4796,
-      "step": 21543
-    },
-    {
-      "epoch": 11.235462842242503,
-      "grad_norm": 1.6040754318237305,
-      "learning_rate": 7.885829145728644e-05,
-      "loss": 5.4701,
-      "step": 21544
-    },
-    {
-      "epoch": 11.235984354628423,
-      "grad_norm": 1.47360098361969,
-      "learning_rate": 7.885728643216081e-05,
-      "loss": 5.7293,
-      "step": 21545
-    },
-    {
-      "epoch": 11.236505867014342,
-      "grad_norm": 1.5060369968414307,
-      "learning_rate": 7.885628140703518e-05,
-      "loss": 5.4387,
-      "step": 21546
-    },
-    {
-      "epoch": 11.23702737940026,
-      "grad_norm": 1.4256051778793335,
-      "learning_rate": 7.885527638190955e-05,
-      "loss": 5.7548,
-      "step": 21547
-    },
-    {
-      "epoch": 11.23754889178618,
-      "grad_norm": 1.4275476932525635,
-      "learning_rate": 7.885427135678393e-05,
-      "loss": 5.2658,
-      "step": 21548
-    },
-    {
-      "epoch": 11.238070404172099,
-      "grad_norm": 1.5032707452774048,
-      "learning_rate": 7.885326633165829e-05,
-      "loss": 5.4251,
-      "step": 21549
-    },
-    {
-      "epoch": 11.238591916558018,
-      "grad_norm": 1.4898746013641357,
-      "learning_rate": 7.885226130653267e-05,
-      "loss": 5.3745,
-      "step": 21550
-    },
-    {
-      "epoch": 11.239113428943938,
-      "grad_norm": 1.4509886503219604,
-      "learning_rate": 7.885125628140703e-05,
-      "loss": 5.5634,
-      "step": 21551
-    },
-    {
-      "epoch": 11.239634941329857,
-      "grad_norm": 1.5046229362487793,
-      "learning_rate": 7.885025125628141e-05,
-      "loss": 5.1409,
-      "step": 21552
-    },
-    {
-      "epoch": 11.240156453715775,
-      "grad_norm": 1.4850744009017944,
-      "learning_rate": 7.884924623115578e-05,
-      "loss": 5.6722,
-      "step": 21553
-    },
-    {
-      "epoch": 11.240677966101694,
-      "grad_norm": 1.3835476636886597,
-      "learning_rate": 7.884824120603015e-05,
-      "loss": 5.3723,
-      "step": 21554
-    },
-    {
-      "epoch": 11.241199478487614,
-      "grad_norm": 1.496640920639038,
-      "learning_rate": 7.884723618090453e-05,
-      "loss": 5.3677,
-      "step": 21555
-    },
-    {
-      "epoch": 11.241720990873533,
-      "grad_norm": 1.4247251749038696,
-      "learning_rate": 7.88462311557789e-05,
-      "loss": 5.3982,
-      "step": 21556
-    },
-    {
-      "epoch": 11.242242503259453,
-      "grad_norm": 1.4319449663162231,
-      "learning_rate": 7.884522613065327e-05,
-      "loss": 5.5826,
-      "step": 21557
-    },
-    {
-      "epoch": 11.242764015645372,
-      "grad_norm": 1.4314829111099243,
-      "learning_rate": 7.884422110552764e-05,
-      "loss": 5.6213,
-      "step": 21558
-    },
-    {
-      "epoch": 11.24328552803129,
-      "grad_norm": 1.4000111818313599,
-      "learning_rate": 7.884321608040202e-05,
-      "loss": 5.8262,
-      "step": 21559
-    },
-    {
-      "epoch": 11.24380704041721,
-      "grad_norm": 1.375978946685791,
-      "learning_rate": 7.884221105527638e-05,
-      "loss": 5.3902,
-      "step": 21560
-    },
-    {
-      "epoch": 11.244328552803129,
-      "grad_norm": 1.440431833267212,
-      "learning_rate": 7.884120603015076e-05,
-      "loss": 5.0666,
-      "step": 21561
-    },
-    {
-      "epoch": 11.244850065189048,
-      "grad_norm": 1.3646066188812256,
-      "learning_rate": 7.884020100502512e-05,
-      "loss": 5.8275,
-      "step": 21562
-    },
-    {
-      "epoch": 11.245371577574968,
-      "grad_norm": 1.4053401947021484,
-      "learning_rate": 7.88391959798995e-05,
-      "loss": 5.4296,
-      "step": 21563
-    },
-    {
-      "epoch": 11.245893089960887,
-      "grad_norm": 1.365931749343872,
-      "learning_rate": 7.883819095477388e-05,
-      "loss": 5.4847,
-      "step": 21564
-    },
-    {
-      "epoch": 11.246414602346805,
-      "grad_norm": 1.4233733415603638,
-      "learning_rate": 7.883718592964826e-05,
-      "loss": 5.7139,
-      "step": 21565
-    },
-    {
-      "epoch": 11.246936114732724,
-      "grad_norm": 1.493709683418274,
-      "learning_rate": 7.883618090452262e-05,
-      "loss": 5.0806,
-      "step": 21566
-    },
-    {
-      "epoch": 11.247457627118644,
-      "grad_norm": 1.457600474357605,
-      "learning_rate": 7.8835175879397e-05,
-      "loss": 5.5715,
-      "step": 21567
-    },
-    {
-      "epoch": 11.247979139504563,
-      "grad_norm": 1.4536068439483643,
-      "learning_rate": 7.883417085427136e-05,
-      "loss": 5.4161,
-      "step": 21568
-    },
-    {
-      "epoch": 11.248500651890483,
-      "grad_norm": 1.4486724138259888,
-      "learning_rate": 7.883316582914573e-05,
-      "loss": 5.6544,
-      "step": 21569
-    },
-    {
-      "epoch": 11.249022164276402,
-      "grad_norm": 1.4196668863296509,
-      "learning_rate": 7.88321608040201e-05,
-      "loss": 5.4837,
-      "step": 21570
-    },
-    {
-      "epoch": 11.24954367666232,
-      "grad_norm": 1.4879069328308105,
-      "learning_rate": 7.883115577889447e-05,
-      "loss": 5.483,
-      "step": 21571
-    },
-    {
-      "epoch": 11.25006518904824,
-      "grad_norm": 1.5800832509994507,
-      "learning_rate": 7.883015075376885e-05,
-      "loss": 5.6275,
-      "step": 21572
-    },
-    {
-      "epoch": 11.250586701434159,
-      "grad_norm": 1.4933505058288574,
-      "learning_rate": 7.882914572864321e-05,
-      "loss": 4.9533,
-      "step": 21573
-    },
-    {
-      "epoch": 11.251108213820078,
-      "grad_norm": 1.5035470724105835,
-      "learning_rate": 7.882814070351759e-05,
-      "loss": 5.6989,
-      "step": 21574
-    },
-    {
-      "epoch": 11.251629726205998,
-      "grad_norm": 1.3317551612854004,
-      "learning_rate": 7.882713567839197e-05,
-      "loss": 5.2577,
-      "step": 21575
-    },
-    {
-      "epoch": 11.252151238591917,
-      "grad_norm": 1.3526691198349,
-      "learning_rate": 7.882613065326634e-05,
-      "loss": 5.6486,
-      "step": 21576
-    },
-    {
-      "epoch": 11.252672750977835,
-      "grad_norm": 1.5067870616912842,
-      "learning_rate": 7.882512562814071e-05,
-      "loss": 5.0809,
-      "step": 21577
-    },
-    {
-      "epoch": 11.253194263363755,
-      "grad_norm": 1.4263241291046143,
-      "learning_rate": 7.882412060301509e-05,
-      "loss": 5.0907,
-      "step": 21578
-    },
-    {
-      "epoch": 11.253715775749674,
-      "grad_norm": 1.4343351125717163,
-      "learning_rate": 7.882311557788945e-05,
-      "loss": 5.6543,
-      "step": 21579
-    },
-    {
-      "epoch": 11.254237288135593,
-      "grad_norm": 1.4877642393112183,
-      "learning_rate": 7.882211055276383e-05,
-      "loss": 5.0292,
-      "step": 21580
-    },
-    {
-      "epoch": 11.254758800521513,
-      "grad_norm": 1.376044750213623,
-      "learning_rate": 7.882110552763819e-05,
-      "loss": 4.6048,
-      "step": 21581
-    },
-    {
-      "epoch": 11.255280312907432,
-      "grad_norm": 1.3962892293930054,
-      "learning_rate": 7.882010050251256e-05,
-      "loss": 4.9571,
-      "step": 21582
-    },
-    {
-      "epoch": 11.25580182529335,
-      "grad_norm": 1.5281423330307007,
-      "learning_rate": 7.881909547738693e-05,
-      "loss": 5.2469,
-      "step": 21583
-    },
-    {
-      "epoch": 11.25632333767927,
-      "grad_norm": 1.4197388887405396,
-      "learning_rate": 7.881809045226131e-05,
-      "loss": 4.9759,
-      "step": 21584
-    },
-    {
-      "epoch": 11.256844850065189,
-      "grad_norm": 1.4619269371032715,
-      "learning_rate": 7.881708542713569e-05,
-      "loss": 5.3881,
-      "step": 21585
-    },
-    {
-      "epoch": 11.257366362451108,
-      "grad_norm": 1.4233261346817017,
-      "learning_rate": 7.881608040201005e-05,
-      "loss": 5.611,
-      "step": 21586
-    },
-    {
-      "epoch": 11.257887874837028,
-      "grad_norm": 1.5144531726837158,
-      "learning_rate": 7.881507537688443e-05,
-      "loss": 5.1416,
-      "step": 21587
-    },
-    {
-      "epoch": 11.258409387222947,
-      "grad_norm": 1.504586935043335,
-      "learning_rate": 7.88140703517588e-05,
-      "loss": 5.164,
-      "step": 21588
-    },
-    {
-      "epoch": 11.258930899608865,
-      "grad_norm": 1.5594593286514282,
-      "learning_rate": 7.881306532663317e-05,
-      "loss": 5.0528,
-      "step": 21589
-    },
-    {
-      "epoch": 11.259452411994785,
-      "grad_norm": 1.5064724683761597,
-      "learning_rate": 7.881206030150754e-05,
-      "loss": 4.9534,
-      "step": 21590
-    },
-    {
-      "epoch": 11.259973924380704,
-      "grad_norm": 1.5834522247314453,
-      "learning_rate": 7.881105527638192e-05,
-      "loss": 5.3783,
-      "step": 21591
-    },
-    {
-      "epoch": 11.260495436766623,
-      "grad_norm": 1.478816270828247,
-      "learning_rate": 7.881005025125628e-05,
-      "loss": 5.3543,
-      "step": 21592
-    },
-    {
-      "epoch": 11.261016949152543,
-      "grad_norm": 1.4150121212005615,
-      "learning_rate": 7.880904522613066e-05,
-      "loss": 5.6807,
-      "step": 21593
-    },
-    {
-      "epoch": 11.261538461538462,
-      "grad_norm": 1.5920788049697876,
-      "learning_rate": 7.880804020100502e-05,
-      "loss": 5.1905,
-      "step": 21594
-    },
-    {
-      "epoch": 11.26205997392438,
-      "grad_norm": 1.4119662046432495,
-      "learning_rate": 7.88070351758794e-05,
-      "loss": 5.477,
-      "step": 21595
-    },
-    {
-      "epoch": 11.2625814863103,
-      "grad_norm": 1.4847437143325806,
-      "learning_rate": 7.880603015075378e-05,
-      "loss": 5.5603,
-      "step": 21596
-    },
-    {
-      "epoch": 11.263102998696219,
-      "grad_norm": 1.5683306455612183,
-      "learning_rate": 7.880502512562814e-05,
-      "loss": 5.3248,
-      "step": 21597
-    },
-    {
-      "epoch": 11.263624511082138,
-      "grad_norm": 1.4918023347854614,
-      "learning_rate": 7.880402010050252e-05,
-      "loss": 5.7277,
-      "step": 21598
-    },
-    {
-      "epoch": 11.264146023468058,
-      "grad_norm": 1.5125819444656372,
-      "learning_rate": 7.880301507537688e-05,
-      "loss": 5.5612,
-      "step": 21599
-    },
-    {
-      "epoch": 11.264667535853977,
-      "grad_norm": 1.5569976568222046,
-      "learning_rate": 7.880201005025126e-05,
-      "loss": 4.9604,
-      "step": 21600
-    },
-    {
-      "epoch": 11.265189048239895,
-      "grad_norm": 1.4182040691375732,
-      "learning_rate": 7.880100502512563e-05,
-      "loss": 5.6887,
-      "step": 21601
-    },
-    {
-      "epoch": 11.265710560625815,
-      "grad_norm": 1.4610908031463623,
-      "learning_rate": 7.88e-05,
-      "loss": 5.6013,
-      "step": 21602
-    },
-    {
-      "epoch": 11.266232073011734,
-      "grad_norm": 1.5677963495254517,
-      "learning_rate": 7.879899497487437e-05,
-      "loss": 5.2086,
-      "step": 21603
-    },
-    {
-      "epoch": 11.266753585397653,
-      "grad_norm": 1.359065294265747,
-      "learning_rate": 7.879798994974875e-05,
-      "loss": 5.3232,
-      "step": 21604
-    },
-    {
-      "epoch": 11.267275097783573,
-      "grad_norm": 1.460388422012329,
-      "learning_rate": 7.879698492462312e-05,
-      "loss": 5.7164,
-      "step": 21605
-    },
-    {
-      "epoch": 11.26779661016949,
-      "grad_norm": 1.4160175323486328,
-      "learning_rate": 7.87959798994975e-05,
-      "loss": 5.4959,
-      "step": 21606
-    },
-    {
-      "epoch": 11.26831812255541,
-      "grad_norm": 1.4602148532867432,
-      "learning_rate": 7.879497487437187e-05,
-      "loss": 5.616,
-      "step": 21607
-    },
-    {
-      "epoch": 11.26883963494133,
-      "grad_norm": 1.5534533262252808,
-      "learning_rate": 7.879396984924623e-05,
-      "loss": 5.0207,
-      "step": 21608
-    },
-    {
-      "epoch": 11.269361147327249,
-      "grad_norm": 1.473680019378662,
-      "learning_rate": 7.879296482412061e-05,
-      "loss": 5.6233,
-      "step": 21609
-    },
-    {
-      "epoch": 11.269882659713168,
-      "grad_norm": 1.470304250717163,
-      "learning_rate": 7.879195979899497e-05,
-      "loss": 5.7335,
-      "step": 21610
-    },
-    {
-      "epoch": 11.270404172099088,
-      "grad_norm": 1.4981063604354858,
-      "learning_rate": 7.879095477386935e-05,
-      "loss": 5.5175,
-      "step": 21611
-    },
-    {
-      "epoch": 11.270925684485007,
-      "grad_norm": 1.4647741317749023,
-      "learning_rate": 7.878994974874371e-05,
-      "loss": 5.6077,
-      "step": 21612
-    },
-    {
-      "epoch": 11.271447196870925,
-      "grad_norm": 1.43826162815094,
-      "learning_rate": 7.878894472361809e-05,
-      "loss": 5.5871,
-      "step": 21613
-    },
-    {
-      "epoch": 11.271968709256845,
-      "grad_norm": 1.4858157634735107,
-      "learning_rate": 7.878793969849246e-05,
-      "loss": 5.4493,
-      "step": 21614
-    },
-    {
-      "epoch": 11.272490221642764,
-      "grad_norm": 1.437557339668274,
-      "learning_rate": 7.878693467336683e-05,
-      "loss": 5.2311,
-      "step": 21615
-    },
-    {
-      "epoch": 11.273011734028684,
-      "grad_norm": 1.4938292503356934,
-      "learning_rate": 7.878592964824121e-05,
-      "loss": 5.3305,
-      "step": 21616
-    },
-    {
-      "epoch": 11.273533246414603,
-      "grad_norm": 1.4398385286331177,
-      "learning_rate": 7.878492462311559e-05,
-      "loss": 5.1291,
-      "step": 21617
-    },
-    {
-      "epoch": 11.27405475880052,
-      "grad_norm": 1.414345383644104,
-      "learning_rate": 7.878391959798995e-05,
-      "loss": 5.0577,
-      "step": 21618
-    },
-    {
-      "epoch": 11.27457627118644,
-      "grad_norm": 1.390173077583313,
-      "learning_rate": 7.878291457286433e-05,
-      "loss": 4.8144,
-      "step": 21619
-    },
-    {
-      "epoch": 11.27509778357236,
-      "grad_norm": 1.4529165029525757,
-      "learning_rate": 7.87819095477387e-05,
-      "loss": 5.4237,
-      "step": 21620
-    },
-    {
-      "epoch": 11.275619295958279,
-      "grad_norm": 1.5168319940567017,
-      "learning_rate": 7.878090452261306e-05,
-      "loss": 5.2594,
-      "step": 21621
-    },
-    {
-      "epoch": 11.276140808344199,
-      "grad_norm": 1.5305049419403076,
-      "learning_rate": 7.877989949748744e-05,
-      "loss": 5.1227,
-      "step": 21622
-    },
-    {
-      "epoch": 11.276662320730118,
-      "grad_norm": 1.3875600099563599,
-      "learning_rate": 7.87788944723618e-05,
-      "loss": 5.5648,
-      "step": 21623
-    },
-    {
-      "epoch": 11.277183833116036,
-      "grad_norm": 1.3123477697372437,
-      "learning_rate": 7.877788944723618e-05,
-      "loss": 5.6254,
-      "step": 21624
-    },
-    {
-      "epoch": 11.277705345501955,
-      "grad_norm": 1.45758056640625,
-      "learning_rate": 7.877688442211056e-05,
-      "loss": 5.8644,
-      "step": 21625
-    },
-    {
-      "epoch": 11.278226857887875,
-      "grad_norm": 1.4205445051193237,
-      "learning_rate": 7.877587939698494e-05,
-      "loss": 5.7443,
-      "step": 21626
-    },
-    {
-      "epoch": 11.278748370273794,
-      "grad_norm": 1.4454681873321533,
-      "learning_rate": 7.87748743718593e-05,
-      "loss": 5.6216,
-      "step": 21627
-    },
-    {
-      "epoch": 11.279269882659714,
-      "grad_norm": 1.3755699396133423,
-      "learning_rate": 7.877386934673368e-05,
-      "loss": 5.7872,
-      "step": 21628
-    },
-    {
-      "epoch": 11.279791395045633,
-      "grad_norm": 1.4007513523101807,
-      "learning_rate": 7.877286432160804e-05,
-      "loss": 5.535,
-      "step": 21629
-    },
-    {
-      "epoch": 11.28031290743155,
-      "grad_norm": 1.4445738792419434,
-      "learning_rate": 7.877185929648242e-05,
-      "loss": 5.3599,
-      "step": 21630
-    },
-    {
-      "epoch": 11.28083441981747,
-      "grad_norm": 1.4437800645828247,
-      "learning_rate": 7.877085427135679e-05,
-      "loss": 5.6709,
-      "step": 21631
-    },
-    {
-      "epoch": 11.28135593220339,
-      "grad_norm": 1.5837682485580444,
-      "learning_rate": 7.876984924623116e-05,
-      "loss": 5.7918,
-      "step": 21632
-    },
-    {
-      "epoch": 11.281877444589309,
-      "grad_norm": 1.518086552619934,
-      "learning_rate": 7.876884422110553e-05,
-      "loss": 5.2927,
-      "step": 21633
-    },
-    {
-      "epoch": 11.282398956975229,
-      "grad_norm": 1.5282893180847168,
-      "learning_rate": 7.876783919597989e-05,
-      "loss": 5.4521,
-      "step": 21634
-    },
-    {
-      "epoch": 11.282920469361148,
-      "grad_norm": 1.5289322137832642,
-      "learning_rate": 7.876683417085427e-05,
-      "loss": 5.3482,
-      "step": 21635
-    },
-    {
-      "epoch": 11.283441981747066,
-      "grad_norm": 1.4867479801177979,
-      "learning_rate": 7.876582914572865e-05,
-      "loss": 5.4155,
-      "step": 21636
-    },
-    {
-      "epoch": 11.283963494132985,
-      "grad_norm": 1.5066344738006592,
-      "learning_rate": 7.876482412060303e-05,
-      "loss": 5.5038,
-      "step": 21637
-    },
-    {
-      "epoch": 11.284485006518905,
-      "grad_norm": 1.4927717447280884,
-      "learning_rate": 7.876381909547739e-05,
-      "loss": 5.5159,
-      "step": 21638
-    },
-    {
-      "epoch": 11.285006518904824,
-      "grad_norm": 1.7075626850128174,
-      "learning_rate": 7.876281407035177e-05,
-      "loss": 5.051,
-      "step": 21639
-    },
-    {
-      "epoch": 11.285528031290744,
-      "grad_norm": 1.4367772340774536,
-      "learning_rate": 7.876180904522613e-05,
-      "loss": 5.5218,
-      "step": 21640
-    },
-    {
-      "epoch": 11.286049543676663,
-      "grad_norm": 1.5889720916748047,
-      "learning_rate": 7.876080402010051e-05,
-      "loss": 5.2232,
-      "step": 21641
-    },
-    {
-      "epoch": 11.28657105606258,
-      "grad_norm": 1.4641749858856201,
-      "learning_rate": 7.875979899497487e-05,
-      "loss": 5.4305,
-      "step": 21642
-    },
-    {
-      "epoch": 11.2870925684485,
-      "grad_norm": 1.431882381439209,
-      "learning_rate": 7.875879396984925e-05,
-      "loss": 5.5397,
-      "step": 21643
-    },
-    {
-      "epoch": 11.28761408083442,
-      "grad_norm": 1.5783023834228516,
-      "learning_rate": 7.875778894472362e-05,
-      "loss": 5.461,
-      "step": 21644
-    },
-    {
-      "epoch": 11.288135593220339,
-      "grad_norm": 1.3988182544708252,
-      "learning_rate": 7.8756783919598e-05,
-      "loss": 5.7646,
-      "step": 21645
-    },
-    {
-      "epoch": 11.288657105606259,
-      "grad_norm": 1.4139059782028198,
-      "learning_rate": 7.875577889447237e-05,
-      "loss": 4.8788,
-      "step": 21646
-    },
-    {
-      "epoch": 11.289178617992178,
-      "grad_norm": 1.42238450050354,
-      "learning_rate": 7.875477386934675e-05,
-      "loss": 5.659,
-      "step": 21647
-    },
-    {
-      "epoch": 11.289700130378096,
-      "grad_norm": 1.467110276222229,
-      "learning_rate": 7.875376884422111e-05,
-      "loss": 5.1249,
-      "step": 21648
-    },
-    {
-      "epoch": 11.290221642764015,
-      "grad_norm": 1.509706974029541,
-      "learning_rate": 7.875276381909548e-05,
-      "loss": 5.1249,
-      "step": 21649
-    },
-    {
-      "epoch": 11.290743155149935,
-      "grad_norm": 1.587476134300232,
-      "learning_rate": 7.875175879396986e-05,
-      "loss": 5.6022,
-      "step": 21650
-    },
-    {
-      "epoch": 11.291264667535854,
-      "grad_norm": 1.4178283214569092,
-      "learning_rate": 7.875075376884422e-05,
-      "loss": 5.2942,
-      "step": 21651
-    },
-    {
-      "epoch": 11.291786179921774,
-      "grad_norm": 1.5194664001464844,
-      "learning_rate": 7.87497487437186e-05,
-      "loss": 5.5068,
-      "step": 21652
-    },
-    {
-      "epoch": 11.292307692307693,
-      "grad_norm": 1.3547942638397217,
-      "learning_rate": 7.874874371859296e-05,
-      "loss": 5.6458,
-      "step": 21653
-    },
-    {
-      "epoch": 11.29282920469361,
-      "grad_norm": 1.3816196918487549,
-      "learning_rate": 7.874773869346734e-05,
-      "loss": 5.6521,
-      "step": 21654
-    },
-    {
-      "epoch": 11.29335071707953,
-      "grad_norm": 1.677965521812439,
-      "learning_rate": 7.87467336683417e-05,
-      "loss": 5.103,
-      "step": 21655
-    },
-    {
-      "epoch": 11.29387222946545,
-      "grad_norm": 1.3823999166488647,
-      "learning_rate": 7.874572864321608e-05,
-      "loss": 5.6454,
-      "step": 21656
-    },
-    {
-      "epoch": 11.29439374185137,
-      "grad_norm": 1.4373409748077393,
-      "learning_rate": 7.874472361809046e-05,
-      "loss": 5.2443,
-      "step": 21657
-    },
-    {
-      "epoch": 11.294915254237289,
-      "grad_norm": 1.4840341806411743,
-      "learning_rate": 7.874371859296484e-05,
-      "loss": 5.4994,
-      "step": 21658
-    },
-    {
-      "epoch": 11.295436766623208,
-      "grad_norm": 1.4489637613296509,
-      "learning_rate": 7.87427135678392e-05,
-      "loss": 5.5414,
-      "step": 21659
-    },
-    {
-      "epoch": 11.295958279009126,
-      "grad_norm": 1.5394620895385742,
-      "learning_rate": 7.874170854271358e-05,
-      "loss": 5.126,
-      "step": 21660
-    },
-    {
-      "epoch": 11.296479791395045,
-      "grad_norm": 1.5719976425170898,
-      "learning_rate": 7.874070351758794e-05,
-      "loss": 5.0383,
-      "step": 21661
-    },
-    {
-      "epoch": 11.297001303780965,
-      "grad_norm": 1.4003826379776,
-      "learning_rate": 7.873969849246231e-05,
-      "loss": 5.1448,
-      "step": 21662
-    },
-    {
-      "epoch": 11.297522816166884,
-      "grad_norm": 1.5871381759643555,
-      "learning_rate": 7.873869346733669e-05,
-      "loss": 5.7664,
-      "step": 21663
-    },
-    {
-      "epoch": 11.298044328552804,
-      "grad_norm": 1.405902624130249,
-      "learning_rate": 7.873768844221105e-05,
-      "loss": 5.5197,
-      "step": 21664
-    },
-    {
-      "epoch": 11.298565840938723,
-      "grad_norm": 1.4409465789794922,
-      "learning_rate": 7.873668341708543e-05,
-      "loss": 5.3211,
-      "step": 21665
-    },
-    {
-      "epoch": 11.29908735332464,
-      "grad_norm": 1.3689464330673218,
-      "learning_rate": 7.87356783919598e-05,
-      "loss": 5.903,
-      "step": 21666
-    },
-    {
-      "epoch": 11.29960886571056,
-      "grad_norm": 1.3606454133987427,
-      "learning_rate": 7.873467336683418e-05,
-      "loss": 5.9306,
-      "step": 21667
-    },
-    {
-      "epoch": 11.30013037809648,
-      "grad_norm": 1.3385142087936401,
-      "learning_rate": 7.873366834170855e-05,
-      "loss": 5.7855,
-      "step": 21668
-    },
-    {
-      "epoch": 11.3006518904824,
-      "grad_norm": 1.346885085105896,
-      "learning_rate": 7.873266331658293e-05,
-      "loss": 5.6832,
-      "step": 21669
-    },
-    {
-      "epoch": 11.301173402868319,
-      "grad_norm": 1.5122830867767334,
-      "learning_rate": 7.873165829145729e-05,
-      "loss": 5.5178,
-      "step": 21670
-    },
-    {
-      "epoch": 11.301694915254238,
-      "grad_norm": 1.4186575412750244,
-      "learning_rate": 7.873065326633167e-05,
-      "loss": 5.7872,
-      "step": 21671
-    },
-    {
-      "epoch": 11.302216427640156,
-      "grad_norm": 1.4320359230041504,
-      "learning_rate": 7.872964824120603e-05,
-      "loss": 5.4787,
-      "step": 21672
-    },
-    {
-      "epoch": 11.302737940026075,
-      "grad_norm": 1.4661873579025269,
-      "learning_rate": 7.872864321608041e-05,
-      "loss": 5.3163,
-      "step": 21673
-    },
-    {
-      "epoch": 11.303259452411995,
-      "grad_norm": 1.3950302600860596,
-      "learning_rate": 7.872763819095477e-05,
-      "loss": 5.6558,
-      "step": 21674
-    },
-    {
-      "epoch": 11.303780964797914,
-      "grad_norm": 1.5166090726852417,
-      "learning_rate": 7.872663316582914e-05,
-      "loss": 5.4117,
-      "step": 21675
-    },
-    {
-      "epoch": 11.304302477183834,
-      "grad_norm": 1.5395445823669434,
-      "learning_rate": 7.872562814070352e-05,
-      "loss": 5.6975,
-      "step": 21676
-    },
-    {
-      "epoch": 11.304823989569753,
-      "grad_norm": 1.496320128440857,
-      "learning_rate": 7.87246231155779e-05,
-      "loss": 4.9412,
-      "step": 21677
-    },
-    {
-      "epoch": 11.30534550195567,
-      "grad_norm": 1.449505090713501,
-      "learning_rate": 7.872361809045227e-05,
-      "loss": 5.6699,
-      "step": 21678
-    },
-    {
-      "epoch": 11.30586701434159,
-      "grad_norm": 1.4392473697662354,
-      "learning_rate": 7.872261306532664e-05,
-      "loss": 5.4199,
-      "step": 21679
-    },
-    {
-      "epoch": 11.30638852672751,
-      "grad_norm": 1.3558123111724854,
-      "learning_rate": 7.872160804020101e-05,
-      "loss": 5.0571,
-      "step": 21680
-    },
-    {
-      "epoch": 11.30691003911343,
-      "grad_norm": 1.467471957206726,
-      "learning_rate": 7.872060301507538e-05,
-      "loss": 5.3993,
-      "step": 21681
-    },
-    {
-      "epoch": 11.307431551499349,
-      "grad_norm": 1.424646258354187,
-      "learning_rate": 7.871959798994976e-05,
-      "loss": 5.4294,
-      "step": 21682
-    },
-    {
-      "epoch": 11.307953063885268,
-      "grad_norm": 1.536030650138855,
-      "learning_rate": 7.871859296482412e-05,
-      "loss": 4.9608,
-      "step": 21683
-    },
-    {
-      "epoch": 11.308474576271186,
-      "grad_norm": 1.376602292060852,
-      "learning_rate": 7.87175879396985e-05,
-      "loss": 4.9608,
-      "step": 21684
-    },
-    {
-      "epoch": 11.308996088657105,
-      "grad_norm": 1.4355603456497192,
-      "learning_rate": 7.871658291457286e-05,
-      "loss": 5.4988,
-      "step": 21685
-    },
-    {
-      "epoch": 11.309517601043025,
-      "grad_norm": 1.4147093296051025,
-      "learning_rate": 7.871557788944724e-05,
-      "loss": 5.3931,
-      "step": 21686
-    },
-    {
-      "epoch": 11.310039113428944,
-      "grad_norm": 1.4983640909194946,
-      "learning_rate": 7.871457286432162e-05,
-      "loss": 5.3016,
-      "step": 21687
-    },
-    {
-      "epoch": 11.310560625814864,
-      "grad_norm": 1.3394490480422974,
-      "learning_rate": 7.871356783919598e-05,
-      "loss": 5.6722,
-      "step": 21688
-    },
-    {
-      "epoch": 11.311082138200783,
-      "grad_norm": 1.4607887268066406,
-      "learning_rate": 7.871256281407036e-05,
-      "loss": 5.5796,
-      "step": 21689
-    },
-    {
-      "epoch": 11.3116036505867,
-      "grad_norm": 1.5087865591049194,
-      "learning_rate": 7.871155778894472e-05,
-      "loss": 5.0257,
-      "step": 21690
-    },
-    {
-      "epoch": 11.31212516297262,
-      "grad_norm": 1.3865034580230713,
-      "learning_rate": 7.87105527638191e-05,
-      "loss": 5.2286,
-      "step": 21691
-    },
-    {
-      "epoch": 11.31264667535854,
-      "grad_norm": 1.4927884340286255,
-      "learning_rate": 7.870954773869347e-05,
-      "loss": 4.9123,
-      "step": 21692
-    },
-    {
-      "epoch": 11.31316818774446,
-      "grad_norm": 1.4468927383422852,
-      "learning_rate": 7.870854271356784e-05,
-      "loss": 5.5098,
-      "step": 21693
-    },
-    {
-      "epoch": 11.313689700130379,
-      "grad_norm": 1.4657273292541504,
-      "learning_rate": 7.870753768844221e-05,
-      "loss": 5.3475,
-      "step": 21694
-    },
-    {
-      "epoch": 11.314211212516298,
-      "grad_norm": 1.3471827507019043,
-      "learning_rate": 7.870653266331659e-05,
-      "loss": 5.3179,
-      "step": 21695
-    },
-    {
-      "epoch": 11.314732724902216,
-      "grad_norm": 1.4233967065811157,
-      "learning_rate": 7.870552763819095e-05,
-      "loss": 5.7189,
-      "step": 21696
-    },
-    {
-      "epoch": 11.315254237288135,
-      "grad_norm": 1.4787977933883667,
-      "learning_rate": 7.870452261306533e-05,
-      "loss": 5.2952,
-      "step": 21697
-    },
-    {
-      "epoch": 11.315775749674055,
-      "grad_norm": 1.5005193948745728,
-      "learning_rate": 7.87035175879397e-05,
-      "loss": 4.559,
-      "step": 21698
-    },
-    {
-      "epoch": 11.316297262059974,
-      "grad_norm": 1.4352010488510132,
-      "learning_rate": 7.870251256281408e-05,
-      "loss": 5.5067,
-      "step": 21699
-    },
-    {
-      "epoch": 11.316818774445894,
-      "grad_norm": 1.413405179977417,
-      "learning_rate": 7.870150753768845e-05,
-      "loss": 5.224,
-      "step": 21700
-    },
-    {
-      "epoch": 11.317340286831811,
-      "grad_norm": 1.4513120651245117,
-      "learning_rate": 7.870050251256281e-05,
-      "loss": 5.513,
-      "step": 21701
-    },
-    {
-      "epoch": 11.31786179921773,
-      "grad_norm": 1.5047459602355957,
-      "learning_rate": 7.869949748743719e-05,
-      "loss": 4.74,
-      "step": 21702
-    },
-    {
-      "epoch": 11.31838331160365,
-      "grad_norm": 1.3665696382522583,
-      "learning_rate": 7.869849246231156e-05,
-      "loss": 5.8602,
-      "step": 21703
-    },
-    {
-      "epoch": 11.31890482398957,
-      "grad_norm": 1.4468640089035034,
-      "learning_rate": 7.869748743718593e-05,
-      "loss": 5.6038,
-      "step": 21704
-    },
-    {
-      "epoch": 11.31942633637549,
-      "grad_norm": 1.374899983406067,
-      "learning_rate": 7.86964824120603e-05,
-      "loss": 5.7025,
-      "step": 21705
-    },
-    {
-      "epoch": 11.319947848761409,
-      "grad_norm": 1.351311206817627,
-      "learning_rate": 7.869547738693468e-05,
-      "loss": 5.7848,
-      "step": 21706
-    },
-    {
-      "epoch": 11.320469361147328,
-      "grad_norm": 1.4649688005447388,
-      "learning_rate": 7.869447236180905e-05,
-      "loss": 5.7967,
-      "step": 21707
-    },
-    {
-      "epoch": 11.320990873533246,
-      "grad_norm": 1.5314698219299316,
-      "learning_rate": 7.869346733668343e-05,
-      "loss": 5.2417,
-      "step": 21708
-    },
-    {
-      "epoch": 11.321512385919165,
-      "grad_norm": 1.2981817722320557,
-      "learning_rate": 7.86924623115578e-05,
-      "loss": 5.8456,
-      "step": 21709
-    },
-    {
-      "epoch": 11.322033898305085,
-      "grad_norm": 1.3368864059448242,
-      "learning_rate": 7.869145728643217e-05,
-      "loss": 5.6746,
-      "step": 21710
-    },
-    {
-      "epoch": 11.322555410691004,
-      "grad_norm": 1.3727753162384033,
-      "learning_rate": 7.869045226130654e-05,
-      "loss": 5.5538,
-      "step": 21711
-    },
-    {
-      "epoch": 11.323076923076924,
-      "grad_norm": 1.3732481002807617,
-      "learning_rate": 7.868944723618092e-05,
-      "loss": 5.4391,
-      "step": 21712
-    },
-    {
-      "epoch": 11.323598435462841,
-      "grad_norm": 1.5128899812698364,
-      "learning_rate": 7.868844221105528e-05,
-      "loss": 5.4505,
-      "step": 21713
-    },
-    {
-      "epoch": 11.32411994784876,
-      "grad_norm": 1.3202365636825562,
-      "learning_rate": 7.868743718592964e-05,
-      "loss": 5.7668,
-      "step": 21714
-    },
-    {
-      "epoch": 11.32464146023468,
-      "grad_norm": 1.5275702476501465,
-      "learning_rate": 7.868643216080402e-05,
-      "loss": 5.284,
-      "step": 21715
-    },
-    {
-      "epoch": 11.3251629726206,
-      "grad_norm": 1.443802833557129,
-      "learning_rate": 7.868542713567839e-05,
-      "loss": 5.4951,
-      "step": 21716
-    },
-    {
-      "epoch": 11.32568448500652,
-      "grad_norm": 1.5177580118179321,
-      "learning_rate": 7.868442211055276e-05,
-      "loss": 5.4356,
-      "step": 21717
-    },
-    {
-      "epoch": 11.326205997392439,
-      "grad_norm": 1.4694074392318726,
-      "learning_rate": 7.868341708542714e-05,
-      "loss": 5.3902,
-      "step": 21718
-    },
-    {
-      "epoch": 11.326727509778356,
-      "grad_norm": 1.3986836671829224,
-      "learning_rate": 7.868241206030152e-05,
-      "loss": 5.7161,
-      "step": 21719
-    },
-    {
-      "epoch": 11.327249022164276,
-      "grad_norm": 1.4091523885726929,
-      "learning_rate": 7.868140703517588e-05,
-      "loss": 5.0891,
-      "step": 21720
-    },
-    {
-      "epoch": 11.327770534550195,
-      "grad_norm": 1.4325882196426392,
-      "learning_rate": 7.868040201005026e-05,
-      "loss": 5.5224,
-      "step": 21721
-    },
-    {
-      "epoch": 11.328292046936115,
-      "grad_norm": 1.4647177457809448,
-      "learning_rate": 7.867939698492463e-05,
-      "loss": 5.4542,
-      "step": 21722
-    },
-    {
-      "epoch": 11.328813559322034,
-      "grad_norm": 1.5012428760528564,
-      "learning_rate": 7.8678391959799e-05,
-      "loss": 5.6321,
-      "step": 21723
-    },
-    {
-      "epoch": 11.329335071707954,
-      "grad_norm": 1.4447579383850098,
-      "learning_rate": 7.867738693467337e-05,
-      "loss": 5.6305,
-      "step": 21724
-    },
-    {
-      "epoch": 11.329856584093871,
-      "grad_norm": 1.4803937673568726,
-      "learning_rate": 7.867638190954775e-05,
-      "loss": 5.6418,
-      "step": 21725
-    },
-    {
-      "epoch": 11.33037809647979,
-      "grad_norm": 1.354620099067688,
-      "learning_rate": 7.867537688442211e-05,
-      "loss": 5.663,
-      "step": 21726
-    },
-    {
-      "epoch": 11.33089960886571,
-      "grad_norm": 1.5506181716918945,
-      "learning_rate": 7.867437185929647e-05,
-      "loss": 5.4349,
-      "step": 21727
-    },
-    {
-      "epoch": 11.33142112125163,
-      "grad_norm": 1.598701000213623,
-      "learning_rate": 7.867336683417085e-05,
-      "loss": 4.8045,
-      "step": 21728
-    },
-    {
-      "epoch": 11.33194263363755,
-      "grad_norm": 1.4568543434143066,
-      "learning_rate": 7.867236180904523e-05,
-      "loss": 5.1624,
-      "step": 21729
-    },
-    {
-      "epoch": 11.332464146023469,
-      "grad_norm": 1.4934104681015015,
-      "learning_rate": 7.867135678391961e-05,
-      "loss": 5.2705,
-      "step": 21730
-    },
-    {
-      "epoch": 11.332985658409386,
-      "grad_norm": 1.474877119064331,
-      "learning_rate": 7.867035175879397e-05,
-      "loss": 5.958,
-      "step": 21731
-    },
-    {
-      "epoch": 11.333507170795306,
-      "grad_norm": 1.546889066696167,
-      "learning_rate": 7.866934673366835e-05,
-      "loss": 5.0403,
-      "step": 21732
-    },
-    {
-      "epoch": 11.334028683181225,
-      "grad_norm": 1.4867258071899414,
-      "learning_rate": 7.866834170854271e-05,
-      "loss": 5.7254,
-      "step": 21733
-    },
-    {
-      "epoch": 11.334550195567145,
-      "grad_norm": 1.4174695014953613,
-      "learning_rate": 7.866733668341709e-05,
-      "loss": 5.8021,
-      "step": 21734
-    },
-    {
-      "epoch": 11.335071707953064,
-      "grad_norm": 1.4627094268798828,
-      "learning_rate": 7.866633165829146e-05,
-      "loss": 5.4129,
-      "step": 21735
-    },
-    {
-      "epoch": 11.335593220338984,
-      "grad_norm": 1.388047695159912,
-      "learning_rate": 7.866532663316583e-05,
-      "loss": 5.3995,
-      "step": 21736
-    },
-    {
-      "epoch": 11.336114732724901,
-      "grad_norm": 1.5661519765853882,
-      "learning_rate": 7.86643216080402e-05,
-      "loss": 5.5956,
-      "step": 21737
-    },
-    {
-      "epoch": 11.336636245110821,
-      "grad_norm": 1.5054123401641846,
-      "learning_rate": 7.866331658291458e-05,
-      "loss": 5.6972,
-      "step": 21738
-    },
-    {
-      "epoch": 11.33715775749674,
-      "grad_norm": 1.4591230154037476,
-      "learning_rate": 7.866231155778895e-05,
-      "loss": 5.5517,
-      "step": 21739
-    },
-    {
-      "epoch": 11.33767926988266,
-      "grad_norm": 1.5391422510147095,
-      "learning_rate": 7.866130653266333e-05,
-      "loss": 5.5024,
-      "step": 21740
-    },
-    {
-      "epoch": 11.33820078226858,
-      "grad_norm": 1.4838882684707642,
-      "learning_rate": 7.86603015075377e-05,
-      "loss": 5.5934,
-      "step": 21741
-    },
-    {
-      "epoch": 11.338722294654499,
-      "grad_norm": 1.5804290771484375,
-      "learning_rate": 7.865929648241206e-05,
-      "loss": 5.5129,
-      "step": 21742
-    },
-    {
-      "epoch": 11.339243807040416,
-      "grad_norm": 1.4329185485839844,
-      "learning_rate": 7.865829145728644e-05,
-      "loss": 5.4622,
-      "step": 21743
-    },
-    {
-      "epoch": 11.339765319426336,
-      "grad_norm": 1.607994556427002,
-      "learning_rate": 7.86572864321608e-05,
-      "loss": 5.1216,
-      "step": 21744
-    },
-    {
-      "epoch": 11.340286831812255,
-      "grad_norm": 1.56110417842865,
-      "learning_rate": 7.865628140703518e-05,
-      "loss": 5.0403,
-      "step": 21745
-    },
-    {
-      "epoch": 11.340808344198175,
-      "grad_norm": 1.485825777053833,
-      "learning_rate": 7.865527638190954e-05,
-      "loss": 5.1737,
-      "step": 21746
-    },
-    {
-      "epoch": 11.341329856584094,
-      "grad_norm": 1.5019627809524536,
-      "learning_rate": 7.865427135678392e-05,
-      "loss": 5.2306,
-      "step": 21747
-    },
-    {
-      "epoch": 11.341851368970014,
-      "grad_norm": 1.4020031690597534,
-      "learning_rate": 7.865326633165829e-05,
-      "loss": 5.5457,
-      "step": 21748
-    },
-    {
-      "epoch": 11.342372881355931,
-      "grad_norm": 1.5080572366714478,
-      "learning_rate": 7.865226130653266e-05,
-      "loss": 5.1665,
-      "step": 21749
-    },
-    {
-      "epoch": 11.342894393741851,
-      "grad_norm": 1.5478860139846802,
-      "learning_rate": 7.865125628140704e-05,
-      "loss": 5.1837,
-      "step": 21750
-    },
-    {
-      "epoch": 11.34341590612777,
-      "grad_norm": 1.5807262659072876,
-      "learning_rate": 7.865025125628142e-05,
-      "loss": 5.1167,
-      "step": 21751
-    },
-    {
-      "epoch": 11.34393741851369,
-      "grad_norm": 1.3690661191940308,
-      "learning_rate": 7.864924623115578e-05,
-      "loss": 5.7527,
-      "step": 21752
-    },
-    {
-      "epoch": 11.34445893089961,
-      "grad_norm": 1.3948677778244019,
-      "learning_rate": 7.864824120603016e-05,
-      "loss": 5.2899,
-      "step": 21753
-    },
-    {
-      "epoch": 11.344980443285529,
-      "grad_norm": 1.479723572731018,
-      "learning_rate": 7.864723618090453e-05,
-      "loss": 5.2025,
-      "step": 21754
-    },
-    {
-      "epoch": 11.345501955671446,
-      "grad_norm": 1.4443081617355347,
-      "learning_rate": 7.864623115577889e-05,
-      "loss": 5.3808,
-      "step": 21755
-    },
-    {
-      "epoch": 11.346023468057366,
-      "grad_norm": 1.3138716220855713,
-      "learning_rate": 7.864522613065327e-05,
-      "loss": 5.5581,
-      "step": 21756
-    },
-    {
-      "epoch": 11.346544980443285,
-      "grad_norm": 1.4755359888076782,
-      "learning_rate": 7.864422110552763e-05,
-      "loss": 5.3475,
-      "step": 21757
-    },
-    {
-      "epoch": 11.347066492829205,
-      "grad_norm": 1.474910855293274,
-      "learning_rate": 7.864321608040201e-05,
-      "loss": 5.6706,
-      "step": 21758
-    },
-    {
-      "epoch": 11.347588005215124,
-      "grad_norm": 1.5763154029846191,
-      "learning_rate": 7.864221105527639e-05,
-      "loss": 5.3254,
-      "step": 21759
-    },
-    {
-      "epoch": 11.348109517601044,
-      "grad_norm": 1.4158711433410645,
-      "learning_rate": 7.864120603015077e-05,
-      "loss": 5.7263,
-      "step": 21760
-    },
-    {
-      "epoch": 11.348631029986961,
-      "grad_norm": 1.505584716796875,
-      "learning_rate": 7.864020100502513e-05,
-      "loss": 5.2386,
-      "step": 21761
-    },
-    {
-      "epoch": 11.349152542372881,
-      "grad_norm": 1.570313811302185,
-      "learning_rate": 7.863919597989951e-05,
-      "loss": 4.9412,
-      "step": 21762
-    },
-    {
-      "epoch": 11.3496740547588,
-      "grad_norm": 1.4248522520065308,
-      "learning_rate": 7.863819095477387e-05,
-      "loss": 5.5037,
-      "step": 21763
-    },
-    {
-      "epoch": 11.35019556714472,
-      "grad_norm": 1.5762652158737183,
-      "learning_rate": 7.863718592964825e-05,
-      "loss": 5.6628,
-      "step": 21764
-    },
-    {
-      "epoch": 11.35071707953064,
-      "grad_norm": 1.420334815979004,
-      "learning_rate": 7.863618090452261e-05,
-      "loss": 5.3716,
-      "step": 21765
-    },
-    {
-      "epoch": 11.351238591916559,
-      "grad_norm": 1.5772536993026733,
-      "learning_rate": 7.863517587939699e-05,
-      "loss": 4.8715,
-      "step": 21766
-    },
-    {
-      "epoch": 11.351760104302477,
-      "grad_norm": 1.4493656158447266,
-      "learning_rate": 7.863417085427136e-05,
-      "loss": 5.4999,
-      "step": 21767
-    },
-    {
-      "epoch": 11.352281616688396,
-      "grad_norm": 1.503708839416504,
-      "learning_rate": 7.863316582914572e-05,
-      "loss": 5.5539,
-      "step": 21768
-    },
-    {
-      "epoch": 11.352803129074315,
-      "grad_norm": 1.3251762390136719,
-      "learning_rate": 7.86321608040201e-05,
-      "loss": 5.5219,
-      "step": 21769
-    },
-    {
-      "epoch": 11.353324641460235,
-      "grad_norm": 1.4438284635543823,
-      "learning_rate": 7.863115577889448e-05,
-      "loss": 5.6832,
-      "step": 21770
-    },
-    {
-      "epoch": 11.353846153846154,
-      "grad_norm": 1.3287664651870728,
-      "learning_rate": 7.863015075376885e-05,
-      "loss": 5.532,
-      "step": 21771
-    },
-    {
-      "epoch": 11.354367666232074,
-      "grad_norm": 1.435477614402771,
-      "learning_rate": 7.862914572864322e-05,
-      "loss": 5.456,
-      "step": 21772
-    },
-    {
-      "epoch": 11.354889178617992,
-      "grad_norm": 1.4021674394607544,
-      "learning_rate": 7.86281407035176e-05,
-      "loss": 5.4887,
-      "step": 21773
-    },
-    {
-      "epoch": 11.355410691003911,
-      "grad_norm": 1.4358917474746704,
-      "learning_rate": 7.862713567839196e-05,
-      "loss": 5.0422,
-      "step": 21774
-    },
-    {
-      "epoch": 11.35593220338983,
-      "grad_norm": 1.5566003322601318,
-      "learning_rate": 7.862613065326634e-05,
-      "loss": 5.6133,
-      "step": 21775
-    },
-    {
-      "epoch": 11.35645371577575,
-      "grad_norm": 1.541052222251892,
-      "learning_rate": 7.86251256281407e-05,
-      "loss": 5.7871,
-      "step": 21776
-    },
-    {
-      "epoch": 11.35697522816167,
-      "grad_norm": 1.4621762037277222,
-      "learning_rate": 7.862412060301508e-05,
-      "loss": 5.3893,
-      "step": 21777
-    },
-    {
-      "epoch": 11.357496740547589,
-      "grad_norm": 1.3973109722137451,
-      "learning_rate": 7.862311557788945e-05,
-      "loss": 5.664,
-      "step": 21778
-    },
-    {
-      "epoch": 11.358018252933507,
-      "grad_norm": 1.524572491645813,
-      "learning_rate": 7.862211055276382e-05,
-      "loss": 5.1073,
-      "step": 21779
-    },
-    {
-      "epoch": 11.358539765319426,
-      "grad_norm": 1.3713734149932861,
-      "learning_rate": 7.86211055276382e-05,
-      "loss": 5.4258,
-      "step": 21780
-    },
-    {
-      "epoch": 11.359061277705345,
-      "grad_norm": 1.4195119142532349,
-      "learning_rate": 7.862010050251257e-05,
-      "loss": 5.439,
-      "step": 21781
-    },
-    {
-      "epoch": 11.359582790091265,
-      "grad_norm": 1.53103506565094,
-      "learning_rate": 7.861909547738694e-05,
-      "loss": 5.4862,
-      "step": 21782
-    },
-    {
-      "epoch": 11.360104302477184,
-      "grad_norm": 1.4309606552124023,
-      "learning_rate": 7.861809045226131e-05,
-      "loss": 5.3696,
-      "step": 21783
-    },
-    {
-      "epoch": 11.360625814863104,
-      "grad_norm": 1.4213894605636597,
-      "learning_rate": 7.861708542713568e-05,
-      "loss": 5.4576,
-      "step": 21784
-    },
-    {
-      "epoch": 11.361147327249022,
-      "grad_norm": 1.374656081199646,
-      "learning_rate": 7.861608040201005e-05,
-      "loss": 5.7647,
-      "step": 21785
-    },
-    {
-      "epoch": 11.361668839634941,
-      "grad_norm": 1.4988603591918945,
-      "learning_rate": 7.861507537688443e-05,
-      "loss": 5.3268,
-      "step": 21786
-    },
-    {
-      "epoch": 11.36219035202086,
-      "grad_norm": 1.4622688293457031,
-      "learning_rate": 7.861407035175879e-05,
-      "loss": 5.49,
-      "step": 21787
-    },
-    {
-      "epoch": 11.36271186440678,
-      "grad_norm": 1.4135977029800415,
-      "learning_rate": 7.861306532663317e-05,
-      "loss": 5.6199,
-      "step": 21788
-    },
-    {
-      "epoch": 11.3632333767927,
-      "grad_norm": 1.5629712343215942,
-      "learning_rate": 7.861206030150753e-05,
-      "loss": 5.1384,
-      "step": 21789
-    },
-    {
-      "epoch": 11.363754889178619,
-      "grad_norm": 1.5867526531219482,
-      "learning_rate": 7.861105527638191e-05,
-      "loss": 5.4405,
-      "step": 21790
-    },
-    {
-      "epoch": 11.364276401564537,
-      "grad_norm": 1.436922550201416,
-      "learning_rate": 7.861005025125629e-05,
-      "loss": 5.7425,
-      "step": 21791
-    },
-    {
-      "epoch": 11.364797913950456,
-      "grad_norm": 1.553611159324646,
-      "learning_rate": 7.860904522613067e-05,
-      "loss": 5.6427,
-      "step": 21792
-    },
-    {
-      "epoch": 11.365319426336375,
-      "grad_norm": 1.5044598579406738,
-      "learning_rate": 7.860804020100503e-05,
-      "loss": 5.5306,
-      "step": 21793
-    },
-    {
-      "epoch": 11.365840938722295,
-      "grad_norm": 1.4946401119232178,
-      "learning_rate": 7.86070351758794e-05,
-      "loss": 5.1743,
-      "step": 21794
-    },
-    {
-      "epoch": 11.366362451108214,
-      "grad_norm": 1.4074593782424927,
-      "learning_rate": 7.860603015075377e-05,
-      "loss": 5.6987,
-      "step": 21795
-    },
-    {
-      "epoch": 11.366883963494132,
-      "grad_norm": 1.3871705532073975,
-      "learning_rate": 7.860502512562814e-05,
-      "loss": 5.4657,
-      "step": 21796
-    },
-    {
-      "epoch": 11.367405475880052,
-      "grad_norm": 1.6498055458068848,
-      "learning_rate": 7.860402010050252e-05,
-      "loss": 4.6426,
-      "step": 21797
-    },
-    {
-      "epoch": 11.367926988265971,
-      "grad_norm": 1.48219633102417,
-      "learning_rate": 7.860301507537688e-05,
-      "loss": 5.5364,
-      "step": 21798
-    },
-    {
-      "epoch": 11.36844850065189,
-      "grad_norm": 1.5038427114486694,
-      "learning_rate": 7.860201005025126e-05,
-      "loss": 5.696,
-      "step": 21799
-    },
-    {
-      "epoch": 11.36897001303781,
-      "grad_norm": 1.67229163646698,
-      "learning_rate": 7.860100502512564e-05,
-      "loss": 4.5538,
-      "step": 21800
-    },
-    {
-      "epoch": 11.36949152542373,
-      "grad_norm": 1.3851772546768188,
-      "learning_rate": 7.860000000000001e-05,
-      "loss": 5.6611,
-      "step": 21801
-    },
-    {
-      "epoch": 11.370013037809649,
-      "grad_norm": 1.435460090637207,
-      "learning_rate": 7.859899497487438e-05,
-      "loss": 5.5806,
-      "step": 21802
-    },
-    {
-      "epoch": 11.370534550195567,
-      "grad_norm": 1.468352198600769,
-      "learning_rate": 7.859798994974876e-05,
-      "loss": 5.782,
-      "step": 21803
-    },
-    {
-      "epoch": 11.371056062581486,
-      "grad_norm": 1.3749054670333862,
-      "learning_rate": 7.859698492462312e-05,
-      "loss": 5.3778,
-      "step": 21804
-    },
-    {
-      "epoch": 11.371577574967406,
-      "grad_norm": 1.5206177234649658,
-      "learning_rate": 7.85959798994975e-05,
-      "loss": 4.9551,
-      "step": 21805
-    },
-    {
-      "epoch": 11.372099087353325,
-      "grad_norm": 1.7995771169662476,
-      "learning_rate": 7.859497487437186e-05,
-      "loss": 5.3813,
-      "step": 21806
-    },
-    {
-      "epoch": 11.372620599739244,
-      "grad_norm": 1.4341408014297485,
-      "learning_rate": 7.859396984924623e-05,
-      "loss": 5.5137,
-      "step": 21807
-    },
-    {
-      "epoch": 11.373142112125162,
-      "grad_norm": 1.5919218063354492,
-      "learning_rate": 7.85929648241206e-05,
-      "loss": 5.2871,
-      "step": 21808
-    },
-    {
-      "epoch": 11.373663624511082,
-      "grad_norm": 1.3649134635925293,
-      "learning_rate": 7.859195979899497e-05,
-      "loss": 5.7053,
-      "step": 21809
-    },
-    {
-      "epoch": 11.374185136897001,
-      "grad_norm": 1.4382117986679077,
-      "learning_rate": 7.859095477386935e-05,
-      "loss": 5.5924,
-      "step": 21810
-    },
-    {
-      "epoch": 11.37470664928292,
-      "grad_norm": 1.4861161708831787,
-      "learning_rate": 7.858994974874372e-05,
-      "loss": 5.6273,
-      "step": 21811
-    },
-    {
-      "epoch": 11.37522816166884,
-      "grad_norm": 1.454245686531067,
-      "learning_rate": 7.85889447236181e-05,
-      "loss": 5.3579,
-      "step": 21812
-    },
-    {
-      "epoch": 11.37574967405476,
-      "grad_norm": 1.3954452276229858,
-      "learning_rate": 7.858793969849247e-05,
-      "loss": 5.6758,
-      "step": 21813
-    },
-    {
-      "epoch": 11.376271186440677,
-      "grad_norm": 1.3056309223175049,
-      "learning_rate": 7.858693467336684e-05,
-      "loss": 4.8902,
-      "step": 21814
-    },
-    {
-      "epoch": 11.376792698826597,
-      "grad_norm": 1.7619348764419556,
-      "learning_rate": 7.858592964824121e-05,
-      "loss": 4.7097,
-      "step": 21815
-    },
-    {
-      "epoch": 11.377314211212516,
-      "grad_norm": 1.486729383468628,
-      "learning_rate": 7.858492462311559e-05,
-      "loss": 5.2309,
-      "step": 21816
-    },
-    {
-      "epoch": 11.377835723598436,
-      "grad_norm": 1.4645698070526123,
-      "learning_rate": 7.858391959798995e-05,
-      "loss": 5.0711,
-      "step": 21817
-    },
-    {
-      "epoch": 11.378357235984355,
-      "grad_norm": 1.4453046321868896,
-      "learning_rate": 7.858291457286433e-05,
-      "loss": 5.8522,
-      "step": 21818
-    },
-    {
-      "epoch": 11.378878748370274,
-      "grad_norm": 1.4162838459014893,
-      "learning_rate": 7.858190954773869e-05,
-      "loss": 5.2711,
-      "step": 21819
-    },
-    {
-      "epoch": 11.379400260756192,
-      "grad_norm": 1.485290288925171,
-      "learning_rate": 7.858090452261307e-05,
-      "loss": 5.4127,
-      "step": 21820
-    },
-    {
-      "epoch": 11.379921773142112,
-      "grad_norm": 1.431525468826294,
-      "learning_rate": 7.857989949748745e-05,
-      "loss": 5.538,
-      "step": 21821
-    },
-    {
-      "epoch": 11.380443285528031,
-      "grad_norm": 1.4927219152450562,
-      "learning_rate": 7.857889447236181e-05,
-      "loss": 5.3559,
-      "step": 21822
-    },
-    {
-      "epoch": 11.38096479791395,
-      "grad_norm": 1.4848524332046509,
-      "learning_rate": 7.857788944723619e-05,
-      "loss": 5.1884,
-      "step": 21823
-    },
-    {
-      "epoch": 11.38148631029987,
-      "grad_norm": 1.477676510810852,
-      "learning_rate": 7.857688442211055e-05,
-      "loss": 5.468,
-      "step": 21824
-    },
-    {
-      "epoch": 11.38200782268579,
-      "grad_norm": 1.532486915588379,
-      "learning_rate": 7.857587939698493e-05,
-      "loss": 5.3404,
-      "step": 21825
-    },
-    {
-      "epoch": 11.382529335071707,
-      "grad_norm": 1.521554708480835,
-      "learning_rate": 7.85748743718593e-05,
-      "loss": 5.4773,
-      "step": 21826
-    },
-    {
-      "epoch": 11.383050847457627,
-      "grad_norm": 1.3908954858779907,
-      "learning_rate": 7.857386934673367e-05,
-      "loss": 5.3834,
-      "step": 21827
-    },
-    {
-      "epoch": 11.383572359843546,
-      "grad_norm": 1.4814939498901367,
-      "learning_rate": 7.857286432160804e-05,
-      "loss": 5.4129,
-      "step": 21828
-    },
-    {
-      "epoch": 11.384093872229466,
-      "grad_norm": 1.4466344118118286,
-      "learning_rate": 7.857185929648242e-05,
-      "loss": 5.5475,
-      "step": 21829
-    },
-    {
-      "epoch": 11.384615384615385,
-      "grad_norm": 1.3979688882827759,
-      "learning_rate": 7.857085427135678e-05,
-      "loss": 5.4978,
-      "step": 21830
-    },
-    {
-      "epoch": 11.385136897001304,
-      "grad_norm": 1.3965809345245361,
-      "learning_rate": 7.856984924623116e-05,
-      "loss": 5.3038,
-      "step": 21831
-    },
-    {
-      "epoch": 11.385658409387222,
-      "grad_norm": 1.5357191562652588,
-      "learning_rate": 7.856884422110554e-05,
-      "loss": 5.3165,
-      "step": 21832
-    },
-    {
-      "epoch": 11.386179921773142,
-      "grad_norm": 1.5139240026474,
-      "learning_rate": 7.856783919597991e-05,
-      "loss": 5.3878,
-      "step": 21833
-    },
-    {
-      "epoch": 11.386701434159061,
-      "grad_norm": 1.419657588005066,
-      "learning_rate": 7.856683417085428e-05,
-      "loss": 5.2234,
-      "step": 21834
-    },
-    {
-      "epoch": 11.38722294654498,
-      "grad_norm": 1.4303820133209229,
-      "learning_rate": 7.856582914572864e-05,
-      "loss": 5.4119,
-      "step": 21835
-    },
-    {
-      "epoch": 11.3877444589309,
-      "grad_norm": 1.442871332168579,
-      "learning_rate": 7.856482412060302e-05,
-      "loss": 5.762,
-      "step": 21836
-    },
-    {
-      "epoch": 11.38826597131682,
-      "grad_norm": 1.3156871795654297,
-      "learning_rate": 7.856381909547738e-05,
-      "loss": 5.8422,
-      "step": 21837
-    },
-    {
-      "epoch": 11.388787483702737,
-      "grad_norm": 1.4140875339508057,
-      "learning_rate": 7.856281407035176e-05,
-      "loss": 5.3877,
-      "step": 21838
-    },
-    {
-      "epoch": 11.389308996088657,
-      "grad_norm": 1.5652005672454834,
-      "learning_rate": 7.856180904522613e-05,
-      "loss": 5.6054,
-      "step": 21839
-    },
-    {
-      "epoch": 11.389830508474576,
-      "grad_norm": 1.6796746253967285,
-      "learning_rate": 7.85608040201005e-05,
-      "loss": 4.9087,
-      "step": 21840
-    },
-    {
-      "epoch": 11.390352020860496,
-      "grad_norm": 1.5600570440292358,
-      "learning_rate": 7.855979899497488e-05,
-      "loss": 5.1941,
-      "step": 21841
-    },
-    {
-      "epoch": 11.390873533246415,
-      "grad_norm": 1.4330295324325562,
-      "learning_rate": 7.855879396984926e-05,
-      "loss": 5.2039,
-      "step": 21842
-    },
-    {
-      "epoch": 11.391395045632335,
-      "grad_norm": 1.459499716758728,
-      "learning_rate": 7.855778894472362e-05,
-      "loss": 5.3959,
-      "step": 21843
-    },
-    {
-      "epoch": 11.391916558018252,
-      "grad_norm": 1.445096731185913,
-      "learning_rate": 7.8556783919598e-05,
-      "loss": 5.6705,
-      "step": 21844
-    },
-    {
-      "epoch": 11.392438070404172,
-      "grad_norm": 1.4754167795181274,
-      "learning_rate": 7.855577889447237e-05,
-      "loss": 5.7487,
-      "step": 21845
-    },
-    {
-      "epoch": 11.392959582790091,
-      "grad_norm": 1.4088784456253052,
-      "learning_rate": 7.855477386934674e-05,
-      "loss": 5.6059,
-      "step": 21846
-    },
-    {
-      "epoch": 11.39348109517601,
-      "grad_norm": 1.451524019241333,
-      "learning_rate": 7.855376884422111e-05,
-      "loss": 5.5125,
-      "step": 21847
-    },
-    {
-      "epoch": 11.39400260756193,
-      "grad_norm": 1.381650447845459,
-      "learning_rate": 7.855276381909547e-05,
-      "loss": 5.8576,
-      "step": 21848
-    },
-    {
-      "epoch": 11.39452411994785,
-      "grad_norm": 1.431604027748108,
-      "learning_rate": 7.855175879396985e-05,
-      "loss": 5.5235,
-      "step": 21849
-    },
-    {
-      "epoch": 11.395045632333767,
-      "grad_norm": 1.8606265783309937,
-      "learning_rate": 7.855075376884422e-05,
-      "loss": 4.645,
-      "step": 21850
-    },
-    {
-      "epoch": 11.395567144719687,
-      "grad_norm": 1.3166486024856567,
-      "learning_rate": 7.854974874371859e-05,
-      "loss": 5.3334,
-      "step": 21851
-    },
-    {
-      "epoch": 11.396088657105606,
-      "grad_norm": 1.5102750062942505,
-      "learning_rate": 7.854874371859297e-05,
-      "loss": 5.2965,
-      "step": 21852
-    },
-    {
-      "epoch": 11.396610169491526,
-      "grad_norm": 1.6603626012802124,
-      "learning_rate": 7.854773869346735e-05,
-      "loss": 5.3292,
-      "step": 21853
-    },
-    {
-      "epoch": 11.397131681877445,
-      "grad_norm": 1.5495054721832275,
-      "learning_rate": 7.854673366834171e-05,
-      "loss": 4.6729,
-      "step": 21854
-    },
-    {
-      "epoch": 11.397653194263365,
-      "grad_norm": 1.3960696458816528,
-      "learning_rate": 7.854572864321609e-05,
-      "loss": 5.5745,
-      "step": 21855
-    },
-    {
-      "epoch": 11.398174706649282,
-      "grad_norm": 1.5293467044830322,
-      "learning_rate": 7.854472361809045e-05,
-      "loss": 5.0175,
-      "step": 21856
-    },
-    {
-      "epoch": 11.398696219035202,
-      "grad_norm": 1.3938359022140503,
-      "learning_rate": 7.854371859296483e-05,
-      "loss": 5.7647,
-      "step": 21857
-    },
-    {
-      "epoch": 11.399217731421121,
-      "grad_norm": 1.3842154741287231,
-      "learning_rate": 7.85427135678392e-05,
-      "loss": 5.5669,
-      "step": 21858
-    },
-    {
-      "epoch": 11.39973924380704,
-      "grad_norm": 1.4969512224197388,
-      "learning_rate": 7.854170854271357e-05,
-      "loss": 5.4558,
-      "step": 21859
-    },
-    {
-      "epoch": 11.40026075619296,
-      "grad_norm": 1.4498741626739502,
-      "learning_rate": 7.854070351758794e-05,
-      "loss": 5.1766,
-      "step": 21860
-    },
-    {
-      "epoch": 11.40078226857888,
-      "grad_norm": 1.4000742435455322,
-      "learning_rate": 7.853969849246232e-05,
-      "loss": 5.7034,
-      "step": 21861
-    },
-    {
-      "epoch": 11.401303780964797,
-      "grad_norm": 1.4385626316070557,
-      "learning_rate": 7.85386934673367e-05,
-      "loss": 5.0622,
-      "step": 21862
-    },
-    {
-      "epoch": 11.401825293350717,
-      "grad_norm": 1.429851770401001,
-      "learning_rate": 7.853768844221106e-05,
-      "loss": 5.2725,
-      "step": 21863
-    },
-    {
-      "epoch": 11.402346805736636,
-      "grad_norm": 1.3781784772872925,
-      "learning_rate": 7.853668341708544e-05,
-      "loss": 4.6698,
-      "step": 21864
-    },
-    {
-      "epoch": 11.402868318122556,
-      "grad_norm": 1.3589377403259277,
-      "learning_rate": 7.85356783919598e-05,
-      "loss": 5.7674,
-      "step": 21865
-    },
-    {
-      "epoch": 11.403389830508475,
-      "grad_norm": 1.401558518409729,
-      "learning_rate": 7.853467336683418e-05,
-      "loss": 5.6803,
-      "step": 21866
-    },
-    {
-      "epoch": 11.403911342894395,
-      "grad_norm": 1.6059449911117554,
-      "learning_rate": 7.853366834170854e-05,
-      "loss": 5.2274,
-      "step": 21867
-    },
-    {
-      "epoch": 11.404432855280312,
-      "grad_norm": 1.5354362726211548,
-      "learning_rate": 7.853266331658292e-05,
-      "loss": 5.1094,
-      "step": 21868
-    },
-    {
-      "epoch": 11.404954367666232,
-      "grad_norm": 1.4055652618408203,
-      "learning_rate": 7.853165829145729e-05,
-      "loss": 5.5948,
-      "step": 21869
-    },
-    {
-      "epoch": 11.405475880052151,
-      "grad_norm": 1.3795655965805054,
-      "learning_rate": 7.853065326633166e-05,
-      "loss": 5.7219,
-      "step": 21870
-    },
-    {
-      "epoch": 11.40599739243807,
-      "grad_norm": 1.529893398284912,
-      "learning_rate": 7.852964824120603e-05,
-      "loss": 5.1536,
-      "step": 21871
-    },
-    {
-      "epoch": 11.40651890482399,
-      "grad_norm": 1.4744333028793335,
-      "learning_rate": 7.85286432160804e-05,
-      "loss": 5.484,
-      "step": 21872
-    },
-    {
-      "epoch": 11.40704041720991,
-      "grad_norm": 1.5490103960037231,
-      "learning_rate": 7.852763819095478e-05,
-      "loss": 5.4271,
-      "step": 21873
-    },
-    {
-      "epoch": 11.407561929595827,
-      "grad_norm": 1.3909502029418945,
-      "learning_rate": 7.852663316582915e-05,
-      "loss": 5.4237,
-      "step": 21874
-    },
-    {
-      "epoch": 11.408083441981747,
-      "grad_norm": 1.4503576755523682,
-      "learning_rate": 7.852562814070353e-05,
-      "loss": 6.0101,
-      "step": 21875
-    },
-    {
-      "epoch": 11.408604954367666,
-      "grad_norm": 1.3577269315719604,
-      "learning_rate": 7.852462311557789e-05,
-      "loss": 5.8518,
-      "step": 21876
-    },
-    {
-      "epoch": 11.409126466753586,
-      "grad_norm": 1.4512319564819336,
-      "learning_rate": 7.852361809045227e-05,
-      "loss": 5.3647,
-      "step": 21877
-    },
-    {
-      "epoch": 11.409647979139505,
-      "grad_norm": 1.5142520666122437,
-      "learning_rate": 7.852261306532663e-05,
-      "loss": 5.222,
-      "step": 21878
-    },
-    {
-      "epoch": 11.410169491525423,
-      "grad_norm": 1.4484385251998901,
-      "learning_rate": 7.852160804020101e-05,
-      "loss": 5.2293,
-      "step": 21879
-    },
-    {
-      "epoch": 11.410691003911342,
-      "grad_norm": 1.5830174684524536,
-      "learning_rate": 7.852060301507537e-05,
-      "loss": 5.0627,
-      "step": 21880
-    },
-    {
-      "epoch": 11.411212516297262,
-      "grad_norm": 1.4563528299331665,
-      "learning_rate": 7.851959798994975e-05,
-      "loss": 5.2313,
-      "step": 21881
-    },
-    {
-      "epoch": 11.411734028683181,
-      "grad_norm": 1.4964964389801025,
-      "learning_rate": 7.851859296482412e-05,
-      "loss": 5.5,
-      "step": 21882
-    },
-    {
-      "epoch": 11.4122555410691,
-      "grad_norm": 1.5114717483520508,
-      "learning_rate": 7.85175879396985e-05,
-      "loss": 5.3744,
-      "step": 21883
-    },
-    {
-      "epoch": 11.41277705345502,
-      "grad_norm": 1.4424583911895752,
-      "learning_rate": 7.851658291457287e-05,
-      "loss": 5.4012,
-      "step": 21884
-    },
-    {
-      "epoch": 11.41329856584094,
-      "grad_norm": 1.5589587688446045,
-      "learning_rate": 7.851557788944725e-05,
-      "loss": 5.0574,
-      "step": 21885
-    },
-    {
-      "epoch": 11.413820078226857,
-      "grad_norm": 1.456688404083252,
-      "learning_rate": 7.851457286432161e-05,
-      "loss": 5.8742,
-      "step": 21886
-    },
-    {
-      "epoch": 11.414341590612777,
-      "grad_norm": 1.5405702590942383,
-      "learning_rate": 7.851356783919598e-05,
-      "loss": 5.1262,
-      "step": 21887
-    },
-    {
-      "epoch": 11.414863102998696,
-      "grad_norm": 1.4275109767913818,
-      "learning_rate": 7.851256281407036e-05,
-      "loss": 5.5369,
-      "step": 21888
-    },
-    {
-      "epoch": 11.415384615384616,
-      "grad_norm": 1.5039089918136597,
-      "learning_rate": 7.851155778894472e-05,
-      "loss": 5.7804,
-      "step": 21889
-    },
-    {
-      "epoch": 11.415906127770535,
-      "grad_norm": 1.86427640914917,
-      "learning_rate": 7.85105527638191e-05,
-      "loss": 5.408,
-      "step": 21890
-    },
-    {
-      "epoch": 11.416427640156453,
-      "grad_norm": 1.422865390777588,
-      "learning_rate": 7.850954773869346e-05,
-      "loss": 4.9773,
-      "step": 21891
-    },
-    {
-      "epoch": 11.416949152542372,
-      "grad_norm": 1.4872196912765503,
-      "learning_rate": 7.850854271356784e-05,
-      "loss": 5.5224,
-      "step": 21892
-    },
-    {
-      "epoch": 11.417470664928292,
-      "grad_norm": 1.4279751777648926,
-      "learning_rate": 7.850753768844222e-05,
-      "loss": 5.5532,
-      "step": 21893
-    },
-    {
-      "epoch": 11.417992177314211,
-      "grad_norm": 1.4344799518585205,
-      "learning_rate": 7.85065326633166e-05,
-      "loss": 5.7818,
-      "step": 21894
-    },
-    {
-      "epoch": 11.41851368970013,
-      "grad_norm": 1.5251939296722412,
-      "learning_rate": 7.850552763819096e-05,
-      "loss": 5.1791,
-      "step": 21895
-    },
-    {
-      "epoch": 11.41903520208605,
-      "grad_norm": 1.4395220279693604,
-      "learning_rate": 7.850452261306534e-05,
-      "loss": 5.4745,
-      "step": 21896
-    },
-    {
-      "epoch": 11.419556714471968,
-      "grad_norm": 1.3906700611114502,
-      "learning_rate": 7.85035175879397e-05,
-      "loss": 5.7328,
-      "step": 21897
-    },
-    {
-      "epoch": 11.420078226857887,
-      "grad_norm": 1.5754047632217407,
-      "learning_rate": 7.850251256281408e-05,
-      "loss": 4.9641,
-      "step": 21898
-    },
-    {
-      "epoch": 11.420599739243807,
-      "grad_norm": 1.495476245880127,
-      "learning_rate": 7.850150753768844e-05,
-      "loss": 5.65,
-      "step": 21899
-    },
-    {
-      "epoch": 11.421121251629726,
-      "grad_norm": 1.574713110923767,
-      "learning_rate": 7.850050251256281e-05,
-      "loss": 5.4676,
-      "step": 21900
-    },
-    {
-      "epoch": 11.421642764015646,
-      "grad_norm": 1.4176820516586304,
-      "learning_rate": 7.849949748743719e-05,
-      "loss": 5.3262,
-      "step": 21901
-    },
-    {
-      "epoch": 11.422164276401565,
-      "grad_norm": 1.3127020597457886,
-      "learning_rate": 7.849849246231155e-05,
-      "loss": 5.4166,
-      "step": 21902
-    },
-    {
-      "epoch": 11.422685788787483,
-      "grad_norm": 1.3340725898742676,
-      "learning_rate": 7.849748743718593e-05,
-      "loss": 5.672,
-      "step": 21903
-    },
-    {
-      "epoch": 11.423207301173402,
-      "grad_norm": 1.473162293434143,
-      "learning_rate": 7.84964824120603e-05,
-      "loss": 5.5737,
-      "step": 21904
-    },
-    {
-      "epoch": 11.423728813559322,
-      "grad_norm": 1.3909664154052734,
-      "learning_rate": 7.849547738693468e-05,
-      "loss": 5.5645,
-      "step": 21905
-    },
-    {
-      "epoch": 11.424250325945241,
-      "grad_norm": 1.3671730756759644,
-      "learning_rate": 7.849447236180905e-05,
-      "loss": 5.635,
-      "step": 21906
-    },
-    {
-      "epoch": 11.42477183833116,
-      "grad_norm": 1.401138186454773,
-      "learning_rate": 7.849346733668343e-05,
-      "loss": 4.7777,
-      "step": 21907
-    },
-    {
-      "epoch": 11.42529335071708,
-      "grad_norm": 1.3124910593032837,
-      "learning_rate": 7.849246231155779e-05,
-      "loss": 5.6879,
-      "step": 21908
-    },
-    {
-      "epoch": 11.425814863102998,
-      "grad_norm": 1.3680022954940796,
-      "learning_rate": 7.849145728643217e-05,
-      "loss": 5.0888,
-      "step": 21909
-    },
-    {
-      "epoch": 11.426336375488917,
-      "grad_norm": 1.4863232374191284,
-      "learning_rate": 7.849045226130653e-05,
-      "loss": 5.4867,
-      "step": 21910
-    },
-    {
-      "epoch": 11.426857887874837,
-      "grad_norm": 1.4337248802185059,
-      "learning_rate": 7.848944723618091e-05,
-      "loss": 5.6014,
-      "step": 21911
-    },
-    {
-      "epoch": 11.427379400260756,
-      "grad_norm": 1.519982099533081,
-      "learning_rate": 7.848844221105527e-05,
-      "loss": 5.2844,
-      "step": 21912
-    },
-    {
-      "epoch": 11.427900912646676,
-      "grad_norm": 1.5238186120986938,
-      "learning_rate": 7.848743718592965e-05,
-      "loss": 5.4097,
-      "step": 21913
-    },
-    {
-      "epoch": 11.428422425032595,
-      "grad_norm": 1.7384514808654785,
-      "learning_rate": 7.848643216080403e-05,
-      "loss": 4.866,
-      "step": 21914
-    },
-    {
-      "epoch": 11.428943937418513,
-      "grad_norm": 1.3947458267211914,
-      "learning_rate": 7.84854271356784e-05,
-      "loss": 5.7437,
-      "step": 21915
-    },
-    {
-      "epoch": 11.429465449804432,
-      "grad_norm": 1.486433506011963,
-      "learning_rate": 7.848442211055277e-05,
-      "loss": 4.9929,
-      "step": 21916
-    },
-    {
-      "epoch": 11.429986962190352,
-      "grad_norm": 1.4421298503875732,
-      "learning_rate": 7.848341708542714e-05,
-      "loss": 5.6647,
-      "step": 21917
-    },
-    {
-      "epoch": 11.430508474576271,
-      "grad_norm": 1.50239098072052,
-      "learning_rate": 7.848241206030151e-05,
-      "loss": 5.3928,
-      "step": 21918
-    },
-    {
-      "epoch": 11.43102998696219,
-      "grad_norm": 1.48284912109375,
-      "learning_rate": 7.848140703517588e-05,
-      "loss": 5.6424,
-      "step": 21919
-    },
-    {
-      "epoch": 11.43155149934811,
-      "grad_norm": 1.4868388175964355,
-      "learning_rate": 7.848040201005026e-05,
-      "loss": 5.7844,
-      "step": 21920
-    },
-    {
-      "epoch": 11.432073011734028,
-      "grad_norm": 1.4461861848831177,
-      "learning_rate": 7.847939698492462e-05,
-      "loss": 5.4306,
-      "step": 21921
-    },
-    {
-      "epoch": 11.432594524119947,
-      "grad_norm": 1.44652259349823,
-      "learning_rate": 7.8478391959799e-05,
-      "loss": 5.4135,
-      "step": 21922
-    },
-    {
-      "epoch": 11.433116036505867,
-      "grad_norm": 1.4737640619277954,
-      "learning_rate": 7.847738693467336e-05,
-      "loss": 5.2075,
-      "step": 21923
-    },
-    {
-      "epoch": 11.433637548891786,
-      "grad_norm": 1.3885012865066528,
-      "learning_rate": 7.847638190954774e-05,
-      "loss": 5.4222,
-      "step": 21924
-    },
-    {
-      "epoch": 11.434159061277706,
-      "grad_norm": 1.3937768936157227,
-      "learning_rate": 7.847537688442212e-05,
-      "loss": 5.1805,
-      "step": 21925
-    },
-    {
-      "epoch": 11.434680573663625,
-      "grad_norm": 1.5210520029067993,
-      "learning_rate": 7.84743718592965e-05,
-      "loss": 5.5378,
-      "step": 21926
-    },
-    {
-      "epoch": 11.435202086049543,
-      "grad_norm": 1.407549262046814,
-      "learning_rate": 7.847336683417086e-05,
-      "loss": 5.6436,
-      "step": 21927
-    },
-    {
-      "epoch": 11.435723598435462,
-      "grad_norm": 1.499626636505127,
-      "learning_rate": 7.847236180904522e-05,
-      "loss": 4.9547,
-      "step": 21928
-    },
-    {
-      "epoch": 11.436245110821382,
-      "grad_norm": 1.4379938840866089,
-      "learning_rate": 7.84713567839196e-05,
-      "loss": 5.7285,
-      "step": 21929
-    },
-    {
-      "epoch": 11.436766623207301,
-      "grad_norm": 1.5242055654525757,
-      "learning_rate": 7.847035175879397e-05,
-      "loss": 5.1968,
-      "step": 21930
-    },
-    {
-      "epoch": 11.43728813559322,
-      "grad_norm": 1.397607445716858,
-      "learning_rate": 7.846934673366834e-05,
-      "loss": 5.575,
-      "step": 21931
-    },
-    {
-      "epoch": 11.43780964797914,
-      "grad_norm": 1.4581818580627441,
-      "learning_rate": 7.846834170854271e-05,
-      "loss": 5.6431,
-      "step": 21932
-    },
-    {
-      "epoch": 11.438331160365058,
-      "grad_norm": 1.420741081237793,
-      "learning_rate": 7.846733668341709e-05,
-      "loss": 5.6763,
-      "step": 21933
-    },
-    {
-      "epoch": 11.438852672750977,
-      "grad_norm": 1.4887621402740479,
-      "learning_rate": 7.846633165829146e-05,
-      "loss": 5.1734,
-      "step": 21934
-    },
-    {
-      "epoch": 11.439374185136897,
-      "grad_norm": 1.5526282787322998,
-      "learning_rate": 7.846532663316584e-05,
-      "loss": 5.5498,
-      "step": 21935
-    },
-    {
-      "epoch": 11.439895697522816,
-      "grad_norm": 1.5617296695709229,
-      "learning_rate": 7.846432160804021e-05,
-      "loss": 5.272,
-      "step": 21936
-    },
-    {
-      "epoch": 11.440417209908736,
-      "grad_norm": 1.4087352752685547,
-      "learning_rate": 7.846331658291458e-05,
-      "loss": 4.9606,
-      "step": 21937
-    },
-    {
-      "epoch": 11.440938722294655,
-      "grad_norm": 1.5483160018920898,
-      "learning_rate": 7.846231155778895e-05,
-      "loss": 5.1717,
-      "step": 21938
-    },
-    {
-      "epoch": 11.441460234680573,
-      "grad_norm": 1.4929463863372803,
-      "learning_rate": 7.846130653266333e-05,
-      "loss": 5.5545,
-      "step": 21939
-    },
-    {
-      "epoch": 11.441981747066492,
-      "grad_norm": 1.5375516414642334,
-      "learning_rate": 7.846030150753769e-05,
-      "loss": 5.4777,
-      "step": 21940
-    },
-    {
-      "epoch": 11.442503259452412,
-      "grad_norm": 1.5398472547531128,
-      "learning_rate": 7.845929648241206e-05,
-      "loss": 5.268,
-      "step": 21941
-    },
-    {
-      "epoch": 11.443024771838331,
-      "grad_norm": 1.5212663412094116,
-      "learning_rate": 7.845829145728643e-05,
-      "loss": 5.3657,
-      "step": 21942
-    },
-    {
-      "epoch": 11.44354628422425,
-      "grad_norm": 1.4740883111953735,
-      "learning_rate": 7.84572864321608e-05,
-      "loss": 5.2739,
-      "step": 21943
-    },
-    {
-      "epoch": 11.44406779661017,
-      "grad_norm": 1.486876130104065,
-      "learning_rate": 7.845628140703518e-05,
-      "loss": 5.0599,
-      "step": 21944
-    },
-    {
-      "epoch": 11.444589308996088,
-      "grad_norm": 1.3826375007629395,
-      "learning_rate": 7.845527638190955e-05,
-      "loss": 5.5315,
-      "step": 21945
-    },
-    {
-      "epoch": 11.445110821382007,
-      "grad_norm": 1.4849064350128174,
-      "learning_rate": 7.845427135678393e-05,
-      "loss": 5.6089,
-      "step": 21946
-    },
-    {
-      "epoch": 11.445632333767927,
-      "grad_norm": 1.4315204620361328,
-      "learning_rate": 7.84532663316583e-05,
-      "loss": 5.4314,
-      "step": 21947
-    },
-    {
-      "epoch": 11.446153846153846,
-      "grad_norm": 1.5106080770492554,
-      "learning_rate": 7.845226130653267e-05,
-      "loss": 5.0167,
-      "step": 21948
-    },
-    {
-      "epoch": 11.446675358539766,
-      "grad_norm": 1.351318359375,
-      "learning_rate": 7.845125628140704e-05,
-      "loss": 5.6514,
-      "step": 21949
-    },
-    {
-      "epoch": 11.447196870925685,
-      "grad_norm": 1.4929858446121216,
-      "learning_rate": 7.845025125628142e-05,
-      "loss": 5.6561,
-      "step": 21950
-    },
-    {
-      "epoch": 11.447718383311603,
-      "grad_norm": 1.4310219287872314,
-      "learning_rate": 7.844924623115578e-05,
-      "loss": 5.1796,
-      "step": 21951
-    },
-    {
-      "epoch": 11.448239895697522,
-      "grad_norm": 1.431115984916687,
-      "learning_rate": 7.844824120603016e-05,
-      "loss": 5.4759,
-      "step": 21952
-    },
-    {
-      "epoch": 11.448761408083442,
-      "grad_norm": 1.5305322408676147,
-      "learning_rate": 7.844723618090452e-05,
-      "loss": 5.3751,
-      "step": 21953
-    },
-    {
-      "epoch": 11.449282920469361,
-      "grad_norm": 1.6017018556594849,
-      "learning_rate": 7.84462311557789e-05,
-      "loss": 5.2726,
-      "step": 21954
-    },
-    {
-      "epoch": 11.44980443285528,
-      "grad_norm": 1.3883600234985352,
-      "learning_rate": 7.844522613065328e-05,
-      "loss": 5.8804,
-      "step": 21955
-    },
-    {
-      "epoch": 11.4503259452412,
-      "grad_norm": 1.4482166767120361,
-      "learning_rate": 7.844422110552764e-05,
-      "loss": 5.6237,
-      "step": 21956
-    },
-    {
-      "epoch": 11.450847457627118,
-      "grad_norm": 1.577818751335144,
-      "learning_rate": 7.844321608040202e-05,
-      "loss": 5.4012,
-      "step": 21957
-    },
-    {
-      "epoch": 11.451368970013037,
-      "grad_norm": 1.3832038640975952,
-      "learning_rate": 7.844221105527638e-05,
-      "loss": 5.957,
-      "step": 21958
-    },
-    {
-      "epoch": 11.451890482398957,
-      "grad_norm": 1.57763671875,
-      "learning_rate": 7.844120603015076e-05,
-      "loss": 5.1847,
-      "step": 21959
-    },
-    {
-      "epoch": 11.452411994784876,
-      "grad_norm": 1.4948711395263672,
-      "learning_rate": 7.844020100502513e-05,
-      "loss": 5.1108,
-      "step": 21960
-    },
-    {
-      "epoch": 11.452933507170796,
-      "grad_norm": 1.3991780281066895,
-      "learning_rate": 7.84391959798995e-05,
-      "loss": 5.735,
-      "step": 21961
-    },
-    {
-      "epoch": 11.453455019556715,
-      "grad_norm": 1.4345166683197021,
-      "learning_rate": 7.843819095477387e-05,
-      "loss": 5.6279,
-      "step": 21962
-    },
-    {
-      "epoch": 11.453976531942633,
-      "grad_norm": 1.4509143829345703,
-      "learning_rate": 7.843718592964825e-05,
-      "loss": 5.5996,
-      "step": 21963
-    },
-    {
-      "epoch": 11.454498044328552,
-      "grad_norm": 1.4207608699798584,
-      "learning_rate": 7.843618090452261e-05,
-      "loss": 5.5298,
-      "step": 21964
-    },
-    {
-      "epoch": 11.455019556714472,
-      "grad_norm": 1.372391700744629,
-      "learning_rate": 7.843517587939699e-05,
-      "loss": 5.8364,
-      "step": 21965
-    },
-    {
-      "epoch": 11.455541069100391,
-      "grad_norm": 1.3441556692123413,
-      "learning_rate": 7.843417085427137e-05,
-      "loss": 5.7855,
-      "step": 21966
-    },
-    {
-      "epoch": 11.45606258148631,
-      "grad_norm": 1.3344390392303467,
-      "learning_rate": 7.843316582914573e-05,
-      "loss": 5.7719,
-      "step": 21967
-    },
-    {
-      "epoch": 11.45658409387223,
-      "grad_norm": 1.4809998273849487,
-      "learning_rate": 7.843216080402011e-05,
-      "loss": 5.5408,
-      "step": 21968
-    },
-    {
-      "epoch": 11.457105606258148,
-      "grad_norm": 1.4344843626022339,
-      "learning_rate": 7.843115577889447e-05,
-      "loss": 5.1593,
-      "step": 21969
-    },
-    {
-      "epoch": 11.457627118644067,
-      "grad_norm": 1.5232303142547607,
-      "learning_rate": 7.843015075376885e-05,
-      "loss": 5.534,
-      "step": 21970
-    },
-    {
-      "epoch": 11.458148631029987,
-      "grad_norm": 1.638791799545288,
-      "learning_rate": 7.842914572864321e-05,
-      "loss": 5.609,
-      "step": 21971
-    },
-    {
-      "epoch": 11.458670143415906,
-      "grad_norm": 1.4142677783966064,
-      "learning_rate": 7.842814070351759e-05,
-      "loss": 5.4502,
-      "step": 21972
-    },
-    {
-      "epoch": 11.459191655801826,
-      "grad_norm": 1.415374994277954,
-      "learning_rate": 7.842713567839196e-05,
-      "loss": 5.2744,
-      "step": 21973
-    },
-    {
-      "epoch": 11.459713168187744,
-      "grad_norm": 1.3498796224594116,
-      "learning_rate": 7.842613065326633e-05,
-      "loss": 5.5606,
-      "step": 21974
-    },
-    {
-      "epoch": 11.460234680573663,
-      "grad_norm": 1.375058650970459,
-      "learning_rate": 7.842512562814071e-05,
-      "loss": 5.6306,
-      "step": 21975
-    },
-    {
-      "epoch": 11.460756192959582,
-      "grad_norm": 1.4073662757873535,
-      "learning_rate": 7.842412060301509e-05,
-      "loss": 5.9595,
-      "step": 21976
-    },
-    {
-      "epoch": 11.461277705345502,
-      "grad_norm": 1.4929685592651367,
-      "learning_rate": 7.842311557788945e-05,
-      "loss": 5.3015,
-      "step": 21977
-    },
-    {
-      "epoch": 11.461799217731421,
-      "grad_norm": 1.4289606809616089,
-      "learning_rate": 7.842211055276383e-05,
-      "loss": 5.7435,
-      "step": 21978
-    },
-    {
-      "epoch": 11.46232073011734,
-      "grad_norm": 1.362449049949646,
-      "learning_rate": 7.84211055276382e-05,
-      "loss": 5.4133,
-      "step": 21979
-    },
-    {
-      "epoch": 11.46284224250326,
-      "grad_norm": 1.3966174125671387,
-      "learning_rate": 7.842010050251256e-05,
-      "loss": 5.3964,
-      "step": 21980
-    },
-    {
-      "epoch": 11.463363754889178,
-      "grad_norm": 1.4034438133239746,
-      "learning_rate": 7.841909547738694e-05,
-      "loss": 5.5037,
-      "step": 21981
-    },
-    {
-      "epoch": 11.463885267275097,
-      "grad_norm": 1.4697024822235107,
-      "learning_rate": 7.84180904522613e-05,
-      "loss": 5.2215,
-      "step": 21982
-    },
-    {
-      "epoch": 11.464406779661017,
-      "grad_norm": 1.5010675191879272,
-      "learning_rate": 7.841708542713568e-05,
-      "loss": 5.6836,
-      "step": 21983
-    },
-    {
-      "epoch": 11.464928292046936,
-      "grad_norm": 1.4661678075790405,
-      "learning_rate": 7.841608040201004e-05,
-      "loss": 5.1944,
-      "step": 21984
-    },
-    {
-      "epoch": 11.465449804432856,
-      "grad_norm": 1.4996755123138428,
-      "learning_rate": 7.841507537688442e-05,
-      "loss": 5.5073,
-      "step": 21985
-    },
-    {
-      "epoch": 11.465971316818774,
-      "grad_norm": 1.4562228918075562,
-      "learning_rate": 7.84140703517588e-05,
-      "loss": 5.6547,
-      "step": 21986
-    },
-    {
-      "epoch": 11.466492829204693,
-      "grad_norm": 1.537667989730835,
-      "learning_rate": 7.841306532663318e-05,
-      "loss": 5.3058,
-      "step": 21987
-    },
-    {
-      "epoch": 11.467014341590613,
-      "grad_norm": 1.4160059690475464,
-      "learning_rate": 7.841206030150754e-05,
-      "loss": 5.271,
-      "step": 21988
-    },
-    {
-      "epoch": 11.467535853976532,
-      "grad_norm": 1.657823085784912,
-      "learning_rate": 7.841105527638192e-05,
-      "loss": 4.6542,
-      "step": 21989
-    },
-    {
-      "epoch": 11.468057366362451,
-      "grad_norm": 1.599209189414978,
-      "learning_rate": 7.841005025125628e-05,
-      "loss": 5.1911,
-      "step": 21990
-    },
-    {
-      "epoch": 11.468578878748371,
-      "grad_norm": 1.5031826496124268,
-      "learning_rate": 7.840904522613066e-05,
-      "loss": 5.5607,
-      "step": 21991
-    },
-    {
-      "epoch": 11.469100391134289,
-      "grad_norm": 1.3519002199172974,
-      "learning_rate": 7.840804020100503e-05,
-      "loss": 5.4351,
-      "step": 21992
-    },
-    {
-      "epoch": 11.469621903520208,
-      "grad_norm": 1.4788060188293457,
-      "learning_rate": 7.840703517587939e-05,
-      "loss": 4.8278,
-      "step": 21993
-    },
-    {
-      "epoch": 11.470143415906128,
-      "grad_norm": 1.407860517501831,
-      "learning_rate": 7.840603015075377e-05,
-      "loss": 5.2593,
-      "step": 21994
-    },
-    {
-      "epoch": 11.470664928292047,
-      "grad_norm": 1.671431303024292,
-      "learning_rate": 7.840502512562815e-05,
-      "loss": 5.0801,
-      "step": 21995
-    },
-    {
-      "epoch": 11.471186440677966,
-      "grad_norm": 1.5901298522949219,
-      "learning_rate": 7.840402010050252e-05,
-      "loss": 5.3414,
-      "step": 21996
-    },
-    {
-      "epoch": 11.471707953063886,
-      "grad_norm": 1.4328685998916626,
-      "learning_rate": 7.840301507537689e-05,
-      "loss": 5.477,
-      "step": 21997
-    },
-    {
-      "epoch": 11.472229465449804,
-      "grad_norm": 1.444889783859253,
-      "learning_rate": 7.840201005025127e-05,
-      "loss": 5.3084,
-      "step": 21998
-    },
-    {
-      "epoch": 11.472750977835723,
-      "grad_norm": 1.6165482997894287,
-      "learning_rate": 7.840100502512563e-05,
-      "loss": 5.0452,
-      "step": 21999
-    },
-    {
-      "epoch": 11.473272490221643,
-      "grad_norm": 1.3856250047683716,
-      "learning_rate": 7.840000000000001e-05,
-      "loss": 5.9452,
-      "step": 22000
-    },
-    {
-      "epoch": 11.473794002607562,
-      "grad_norm": 1.4822074174880981,
-      "learning_rate": 7.839899497487437e-05,
-      "loss": 5.3831,
-      "step": 22001
-    },
-    {
-      "epoch": 11.474315514993481,
-      "grad_norm": 1.4833163022994995,
-      "learning_rate": 7.839798994974875e-05,
-      "loss": 5.095,
-      "step": 22002
-    },
-    {
-      "epoch": 11.474837027379401,
-      "grad_norm": 1.4961899518966675,
-      "learning_rate": 7.839698492462311e-05,
-      "loss": 5.5564,
-      "step": 22003
-    },
-    {
-      "epoch": 11.475358539765319,
-      "grad_norm": 1.4784750938415527,
-      "learning_rate": 7.839597989949749e-05,
-      "loss": 5.6416,
-      "step": 22004
-    },
-    {
-      "epoch": 11.475880052151238,
-      "grad_norm": 1.4673463106155396,
-      "learning_rate": 7.839497487437186e-05,
-      "loss": 5.7318,
-      "step": 22005
-    },
-    {
-      "epoch": 11.476401564537158,
-      "grad_norm": 1.5036605596542358,
-      "learning_rate": 7.839396984924623e-05,
-      "loss": 5.0609,
-      "step": 22006
-    },
-    {
-      "epoch": 11.476923076923077,
-      "grad_norm": 1.5010902881622314,
-      "learning_rate": 7.839296482412061e-05,
-      "loss": 5.3502,
-      "step": 22007
-    },
-    {
-      "epoch": 11.477444589308996,
-      "grad_norm": 1.4496947526931763,
-      "learning_rate": 7.839195979899498e-05,
-      "loss": 5.0693,
-      "step": 22008
-    },
-    {
-      "epoch": 11.477966101694916,
-      "grad_norm": 1.450539469718933,
-      "learning_rate": 7.839095477386935e-05,
-      "loss": 5.4105,
-      "step": 22009
-    },
-    {
-      "epoch": 11.478487614080834,
-      "grad_norm": 1.397951602935791,
-      "learning_rate": 7.838994974874372e-05,
-      "loss": 5.3455,
-      "step": 22010
-    },
-    {
-      "epoch": 11.479009126466753,
-      "grad_norm": 1.4314615726470947,
-      "learning_rate": 7.83889447236181e-05,
-      "loss": 5.4323,
-      "step": 22011
-    },
-    {
-      "epoch": 11.479530638852673,
-      "grad_norm": 1.4441031217575073,
-      "learning_rate": 7.838793969849246e-05,
-      "loss": 5.6638,
-      "step": 22012
-    },
-    {
-      "epoch": 11.480052151238592,
-      "grad_norm": 1.5298746824264526,
-      "learning_rate": 7.838693467336684e-05,
-      "loss": 5.018,
-      "step": 22013
-    },
-    {
-      "epoch": 11.480573663624511,
-      "grad_norm": 1.5738763809204102,
-      "learning_rate": 7.83859296482412e-05,
-      "loss": 5.2867,
-      "step": 22014
-    },
-    {
-      "epoch": 11.481095176010431,
-      "grad_norm": 1.4805608987808228,
-      "learning_rate": 7.838492462311558e-05,
-      "loss": 5.364,
-      "step": 22015
-    },
-    {
-      "epoch": 11.481616688396349,
-      "grad_norm": 1.334243655204773,
-      "learning_rate": 7.838391959798996e-05,
-      "loss": 5.742,
-      "step": 22016
-    },
-    {
-      "epoch": 11.482138200782268,
-      "grad_norm": 1.6181221008300781,
-      "learning_rate": 7.838291457286434e-05,
-      "loss": 5.2396,
-      "step": 22017
-    },
-    {
-      "epoch": 11.482659713168188,
-      "grad_norm": 1.45015549659729,
-      "learning_rate": 7.83819095477387e-05,
-      "loss": 5.3695,
-      "step": 22018
-    },
-    {
-      "epoch": 11.483181225554107,
-      "grad_norm": 1.5038368701934814,
-      "learning_rate": 7.838090452261308e-05,
-      "loss": 5.6693,
-      "step": 22019
-    },
-    {
-      "epoch": 11.483702737940026,
-      "grad_norm": 1.5261831283569336,
-      "learning_rate": 7.837989949748744e-05,
-      "loss": 5.2663,
-      "step": 22020
-    },
-    {
-      "epoch": 11.484224250325946,
-      "grad_norm": 1.39738130569458,
-      "learning_rate": 7.837889447236181e-05,
-      "loss": 5.7737,
-      "step": 22021
-    },
-    {
-      "epoch": 11.484745762711864,
-      "grad_norm": 1.4783954620361328,
-      "learning_rate": 7.837788944723619e-05,
-      "loss": 5.5251,
-      "step": 22022
-    },
-    {
-      "epoch": 11.485267275097783,
-      "grad_norm": 1.4941200017929077,
-      "learning_rate": 7.837688442211055e-05,
-      "loss": 4.6932,
-      "step": 22023
-    },
-    {
-      "epoch": 11.485788787483703,
-      "grad_norm": 1.3873728513717651,
-      "learning_rate": 7.837587939698493e-05,
-      "loss": 5.6064,
-      "step": 22024
-    },
-    {
-      "epoch": 11.486310299869622,
-      "grad_norm": 1.3849279880523682,
-      "learning_rate": 7.837487437185929e-05,
-      "loss": 5.4478,
-      "step": 22025
-    },
-    {
-      "epoch": 11.486831812255542,
-      "grad_norm": 1.4963833093643188,
-      "learning_rate": 7.837386934673367e-05,
-      "loss": 5.1525,
-      "step": 22026
-    },
-    {
-      "epoch": 11.487353324641461,
-      "grad_norm": 1.3520461320877075,
-      "learning_rate": 7.837286432160805e-05,
-      "loss": 5.4748,
-      "step": 22027
-    },
-    {
-      "epoch": 11.487874837027379,
-      "grad_norm": 1.5138440132141113,
-      "learning_rate": 7.837185929648242e-05,
-      "loss": 5.0788,
-      "step": 22028
-    },
-    {
-      "epoch": 11.488396349413298,
-      "grad_norm": 1.4641120433807373,
-      "learning_rate": 7.837085427135679e-05,
-      "loss": 5.2901,
-      "step": 22029
-    },
-    {
-      "epoch": 11.488917861799218,
-      "grad_norm": 1.4419300556182861,
-      "learning_rate": 7.836984924623117e-05,
-      "loss": 5.2436,
-      "step": 22030
-    },
-    {
-      "epoch": 11.489439374185137,
-      "grad_norm": 1.356374740600586,
-      "learning_rate": 7.836884422110553e-05,
-      "loss": 5.3817,
-      "step": 22031
-    },
-    {
-      "epoch": 11.489960886571057,
-      "grad_norm": 1.3761968612670898,
-      "learning_rate": 7.836783919597991e-05,
-      "loss": 5.5451,
-      "step": 22032
-    },
-    {
-      "epoch": 11.490482398956976,
-      "grad_norm": 1.341571569442749,
-      "learning_rate": 7.836683417085427e-05,
-      "loss": 5.8131,
-      "step": 22033
-    },
-    {
-      "epoch": 11.491003911342894,
-      "grad_norm": 1.4230791330337524,
-      "learning_rate": 7.836582914572864e-05,
-      "loss": 5.4598,
-      "step": 22034
-    },
-    {
-      "epoch": 11.491525423728813,
-      "grad_norm": 1.3997801542282104,
-      "learning_rate": 7.836482412060302e-05,
-      "loss": 5.7011,
-      "step": 22035
-    },
-    {
-      "epoch": 11.492046936114733,
-      "grad_norm": 1.440280556678772,
-      "learning_rate": 7.83638190954774e-05,
-      "loss": 5.5776,
-      "step": 22036
-    },
-    {
-      "epoch": 11.492568448500652,
-      "grad_norm": 1.4204421043395996,
-      "learning_rate": 7.836281407035177e-05,
-      "loss": 5.5262,
-      "step": 22037
-    },
-    {
-      "epoch": 11.493089960886572,
-      "grad_norm": 1.4615310430526733,
-      "learning_rate": 7.836180904522614e-05,
-      "loss": 5.4183,
-      "step": 22038
-    },
-    {
-      "epoch": 11.493611473272491,
-      "grad_norm": 1.4440776109695435,
-      "learning_rate": 7.836080402010051e-05,
-      "loss": 4.8154,
-      "step": 22039
-    },
-    {
-      "epoch": 11.494132985658409,
-      "grad_norm": 1.6582533121109009,
-      "learning_rate": 7.835979899497488e-05,
-      "loss": 5.2021,
-      "step": 22040
-    },
-    {
-      "epoch": 11.494654498044328,
-      "grad_norm": 1.3975924253463745,
-      "learning_rate": 7.835879396984926e-05,
-      "loss": 5.39,
-      "step": 22041
-    },
-    {
-      "epoch": 11.495176010430248,
-      "grad_norm": 1.3990367650985718,
-      "learning_rate": 7.835778894472362e-05,
-      "loss": 5.4398,
-      "step": 22042
-    },
-    {
-      "epoch": 11.495697522816167,
-      "grad_norm": 1.4350630044937134,
-      "learning_rate": 7.8356783919598e-05,
-      "loss": 5.0901,
-      "step": 22043
-    },
-    {
-      "epoch": 11.496219035202087,
-      "grad_norm": 1.456649899482727,
-      "learning_rate": 7.835577889447236e-05,
-      "loss": 5.8809,
-      "step": 22044
-    },
-    {
-      "epoch": 11.496740547588006,
-      "grad_norm": 1.3437873125076294,
-      "learning_rate": 7.835477386934674e-05,
-      "loss": 5.6007,
-      "step": 22045
-    },
-    {
-      "epoch": 11.497262059973924,
-      "grad_norm": 1.3147121667861938,
-      "learning_rate": 7.83537688442211e-05,
-      "loss": 5.6977,
-      "step": 22046
-    },
-    {
-      "epoch": 11.497783572359843,
-      "grad_norm": 1.4249787330627441,
-      "learning_rate": 7.835276381909548e-05,
-      "loss": 5.3814,
-      "step": 22047
-    },
-    {
-      "epoch": 11.498305084745763,
-      "grad_norm": 1.614884614944458,
-      "learning_rate": 7.835175879396986e-05,
-      "loss": 5.1382,
-      "step": 22048
-    },
-    {
-      "epoch": 11.498826597131682,
-      "grad_norm": 1.2952823638916016,
-      "learning_rate": 7.835075376884422e-05,
-      "loss": 5.8547,
-      "step": 22049
-    },
-    {
-      "epoch": 11.499348109517602,
-      "grad_norm": 1.486340045928955,
-      "learning_rate": 7.83497487437186e-05,
-      "loss": 5.383,
-      "step": 22050
-    },
-    {
-      "epoch": 11.499869621903521,
-      "grad_norm": 1.4171990156173706,
-      "learning_rate": 7.834874371859297e-05,
-      "loss": 5.1657,
-      "step": 22051
-    },
-    {
-      "epoch": 11.500391134289439,
-      "grad_norm": 1.6413767337799072,
-      "learning_rate": 7.834773869346734e-05,
-      "loss": 5.0643,
-      "step": 22052
-    },
-    {
-      "epoch": 11.500912646675358,
-      "grad_norm": 1.402390718460083,
-      "learning_rate": 7.834673366834171e-05,
-      "loss": 5.6415,
-      "step": 22053
-    },
-    {
-      "epoch": 11.501434159061278,
-      "grad_norm": 1.4967855215072632,
-      "learning_rate": 7.834572864321609e-05,
-      "loss": 5.6466,
-      "step": 22054
-    },
-    {
-      "epoch": 11.501955671447197,
-      "grad_norm": 1.392391562461853,
-      "learning_rate": 7.834472361809045e-05,
-      "loss": 5.547,
-      "step": 22055
-    },
-    {
-      "epoch": 11.502477183833117,
-      "grad_norm": 1.4362050294876099,
-      "learning_rate": 7.834371859296483e-05,
-      "loss": 5.5457,
-      "step": 22056
-    },
-    {
-      "epoch": 11.502998696219036,
-      "grad_norm": 1.577022671699524,
-      "learning_rate": 7.834271356783919e-05,
-      "loss": 5.3913,
-      "step": 22057
-    },
-    {
-      "epoch": 11.503520208604954,
-      "grad_norm": 1.5109155178070068,
-      "learning_rate": 7.834170854271357e-05,
-      "loss": 5.2979,
-      "step": 22058
-    },
-    {
-      "epoch": 11.504041720990873,
-      "grad_norm": 1.4602346420288086,
-      "learning_rate": 7.834070351758795e-05,
-      "loss": 5.6269,
-      "step": 22059
-    },
-    {
-      "epoch": 11.504563233376793,
-      "grad_norm": 1.7328978776931763,
-      "learning_rate": 7.833969849246231e-05,
-      "loss": 5.1215,
-      "step": 22060
-    },
-    {
-      "epoch": 11.505084745762712,
-      "grad_norm": 1.4055620431900024,
-      "learning_rate": 7.833869346733669e-05,
-      "loss": 4.9465,
-      "step": 22061
-    },
-    {
-      "epoch": 11.505606258148632,
-      "grad_norm": 1.4749118089675903,
-      "learning_rate": 7.833768844221105e-05,
-      "loss": 5.6937,
-      "step": 22062
-    },
-    {
-      "epoch": 11.506127770534551,
-      "grad_norm": 1.3829560279846191,
-      "learning_rate": 7.833668341708543e-05,
-      "loss": 5.6575,
-      "step": 22063
-    },
-    {
-      "epoch": 11.506649282920469,
-      "grad_norm": 1.3687540292739868,
-      "learning_rate": 7.83356783919598e-05,
-      "loss": 4.9315,
-      "step": 22064
-    },
-    {
-      "epoch": 11.507170795306388,
-      "grad_norm": 1.3750133514404297,
-      "learning_rate": 7.833467336683417e-05,
-      "loss": 5.6117,
-      "step": 22065
-    },
-    {
-      "epoch": 11.507692307692308,
-      "grad_norm": 1.4864928722381592,
-      "learning_rate": 7.833366834170854e-05,
-      "loss": 5.7189,
-      "step": 22066
-    },
-    {
-      "epoch": 11.508213820078227,
-      "grad_norm": 1.6073940992355347,
-      "learning_rate": 7.833266331658292e-05,
-      "loss": 5.0151,
-      "step": 22067
-    },
-    {
-      "epoch": 11.508735332464147,
-      "grad_norm": 1.466808557510376,
-      "learning_rate": 7.83316582914573e-05,
-      "loss": 5.3441,
-      "step": 22068
-    },
-    {
-      "epoch": 11.509256844850064,
-      "grad_norm": 1.396406650543213,
-      "learning_rate": 7.833065326633167e-05,
-      "loss": 5.1395,
-      "step": 22069
-    },
-    {
-      "epoch": 11.509778357235984,
-      "grad_norm": 1.4206547737121582,
-      "learning_rate": 7.832964824120604e-05,
-      "loss": 5.7811,
-      "step": 22070
-    },
-    {
-      "epoch": 11.510299869621903,
-      "grad_norm": 1.5092201232910156,
-      "learning_rate": 7.832864321608041e-05,
-      "loss": 5.3575,
-      "step": 22071
-    },
-    {
-      "epoch": 11.510821382007823,
-      "grad_norm": 1.4595863819122314,
-      "learning_rate": 7.832763819095478e-05,
-      "loss": 5.354,
-      "step": 22072
-    },
-    {
-      "epoch": 11.511342894393742,
-      "grad_norm": 1.3813353776931763,
-      "learning_rate": 7.832663316582914e-05,
-      "loss": 5.5364,
-      "step": 22073
-    },
-    {
-      "epoch": 11.511864406779662,
-      "grad_norm": 1.326134204864502,
-      "learning_rate": 7.832562814070352e-05,
-      "loss": 6.0597,
-      "step": 22074
-    },
-    {
-      "epoch": 11.512385919165581,
-      "grad_norm": 1.4541294574737549,
-      "learning_rate": 7.832462311557788e-05,
-      "loss": 5.5939,
-      "step": 22075
-    },
-    {
-      "epoch": 11.512907431551499,
-      "grad_norm": 1.3797980546951294,
-      "learning_rate": 7.832361809045226e-05,
-      "loss": 5.5935,
-      "step": 22076
-    },
-    {
-      "epoch": 11.513428943937418,
-      "grad_norm": 1.4359245300292969,
-      "learning_rate": 7.832261306532663e-05,
-      "loss": 5.4824,
-      "step": 22077
-    },
-    {
-      "epoch": 11.513950456323338,
-      "grad_norm": 1.337112545967102,
-      "learning_rate": 7.8321608040201e-05,
-      "loss": 5.8417,
-      "step": 22078
-    },
-    {
-      "epoch": 11.514471968709257,
-      "grad_norm": 1.4842630624771118,
-      "learning_rate": 7.832060301507538e-05,
-      "loss": 5.5827,
-      "step": 22079
-    },
-    {
-      "epoch": 11.514993481095177,
-      "grad_norm": 1.377435564994812,
-      "learning_rate": 7.831959798994976e-05,
-      "loss": 5.5528,
-      "step": 22080
-    },
-    {
-      "epoch": 11.515514993481094,
-      "grad_norm": 1.3752537965774536,
-      "learning_rate": 7.831859296482412e-05,
-      "loss": 5.5483,
-      "step": 22081
-    },
-    {
-      "epoch": 11.516036505867014,
-      "grad_norm": 1.3932490348815918,
-      "learning_rate": 7.83175879396985e-05,
-      "loss": 5.8189,
-      "step": 22082
-    },
-    {
-      "epoch": 11.516558018252933,
-      "grad_norm": 1.5265852212905884,
-      "learning_rate": 7.831658291457287e-05,
-      "loss": 5.1998,
-      "step": 22083
-    },
-    {
-      "epoch": 11.517079530638853,
-      "grad_norm": 1.4365898370742798,
-      "learning_rate": 7.831557788944724e-05,
-      "loss": 5.2392,
-      "step": 22084
-    },
-    {
-      "epoch": 11.517601043024772,
-      "grad_norm": 1.5252139568328857,
-      "learning_rate": 7.831457286432161e-05,
-      "loss": 5.4448,
-      "step": 22085
-    },
-    {
-      "epoch": 11.518122555410692,
-      "grad_norm": 1.4665093421936035,
-      "learning_rate": 7.831356783919597e-05,
-      "loss": 5.1608,
-      "step": 22086
-    },
-    {
-      "epoch": 11.518644067796611,
-      "grad_norm": 1.5303748846054077,
-      "learning_rate": 7.831256281407035e-05,
-      "loss": 5.4259,
-      "step": 22087
-    },
-    {
-      "epoch": 11.519165580182529,
-      "grad_norm": 1.3533899784088135,
-      "learning_rate": 7.831155778894473e-05,
-      "loss": 5.8171,
-      "step": 22088
-    },
-    {
-      "epoch": 11.519687092568448,
-      "grad_norm": 1.5275877714157104,
-      "learning_rate": 7.83105527638191e-05,
-      "loss": 4.7506,
-      "step": 22089
-    },
-    {
-      "epoch": 11.520208604954368,
-      "grad_norm": 1.4735097885131836,
-      "learning_rate": 7.830954773869347e-05,
-      "loss": 5.7322,
-      "step": 22090
-    },
-    {
-      "epoch": 11.520730117340287,
-      "grad_norm": 1.4973050355911255,
-      "learning_rate": 7.830854271356785e-05,
-      "loss": 5.439,
-      "step": 22091
-    },
-    {
-      "epoch": 11.521251629726207,
-      "grad_norm": 1.3999415636062622,
-      "learning_rate": 7.830753768844221e-05,
-      "loss": 5.1723,
-      "step": 22092
-    },
-    {
-      "epoch": 11.521773142112124,
-      "grad_norm": 1.4076952934265137,
-      "learning_rate": 7.830653266331659e-05,
-      "loss": 5.4478,
-      "step": 22093
-    },
-    {
-      "epoch": 11.522294654498044,
-      "grad_norm": 1.4733366966247559,
-      "learning_rate": 7.830552763819095e-05,
-      "loss": 5.4797,
-      "step": 22094
-    },
-    {
-      "epoch": 11.522816166883963,
-      "grad_norm": 1.4806276559829712,
-      "learning_rate": 7.830452261306533e-05,
-      "loss": 4.9645,
-      "step": 22095
-    },
-    {
-      "epoch": 11.523337679269883,
-      "grad_norm": 1.4671084880828857,
-      "learning_rate": 7.83035175879397e-05,
-      "loss": 5.5842,
-      "step": 22096
-    },
-    {
-      "epoch": 11.523859191655802,
-      "grad_norm": 1.4196441173553467,
-      "learning_rate": 7.830251256281407e-05,
-      "loss": 5.2962,
-      "step": 22097
-    },
-    {
-      "epoch": 11.524380704041722,
-      "grad_norm": 1.4241409301757812,
-      "learning_rate": 7.830150753768844e-05,
-      "loss": 5.617,
-      "step": 22098
-    },
-    {
-      "epoch": 11.52490221642764,
-      "grad_norm": 1.4878774881362915,
-      "learning_rate": 7.830050251256282e-05,
-      "loss": 5.5182,
-      "step": 22099
-    },
-    {
-      "epoch": 11.525423728813559,
-      "grad_norm": 1.537034511566162,
-      "learning_rate": 7.82994974874372e-05,
-      "loss": 5.3022,
-      "step": 22100
-    },
-    {
-      "epoch": 11.525945241199478,
-      "grad_norm": 1.4222759008407593,
-      "learning_rate": 7.829849246231156e-05,
-      "loss": 5.4814,
-      "step": 22101
-    },
-    {
-      "epoch": 11.526466753585398,
-      "grad_norm": 1.509555697441101,
-      "learning_rate": 7.829748743718594e-05,
-      "loss": 5.2602,
-      "step": 22102
-    },
-    {
-      "epoch": 11.526988265971317,
-      "grad_norm": 1.5052516460418701,
-      "learning_rate": 7.82964824120603e-05,
-      "loss": 5.2613,
-      "step": 22103
-    },
-    {
-      "epoch": 11.527509778357237,
-      "grad_norm": 1.4853413105010986,
-      "learning_rate": 7.829547738693468e-05,
-      "loss": 4.7721,
-      "step": 22104
-    },
-    {
-      "epoch": 11.528031290743154,
-      "grad_norm": 1.4062718152999878,
-      "learning_rate": 7.829447236180904e-05,
-      "loss": 5.4749,
-      "step": 22105
-    },
-    {
-      "epoch": 11.528552803129074,
-      "grad_norm": 1.360733985900879,
-      "learning_rate": 7.829346733668342e-05,
-      "loss": 5.6124,
-      "step": 22106
-    },
-    {
-      "epoch": 11.529074315514993,
-      "grad_norm": 1.33681321144104,
-      "learning_rate": 7.829246231155779e-05,
-      "loss": 5.3781,
-      "step": 22107
-    },
-    {
-      "epoch": 11.529595827900913,
-      "grad_norm": 1.626869797706604,
-      "learning_rate": 7.829145728643216e-05,
-      "loss": 4.5528,
-      "step": 22108
-    },
-    {
-      "epoch": 11.530117340286832,
-      "grad_norm": 1.4051096439361572,
-      "learning_rate": 7.829045226130654e-05,
-      "loss": 5.6486,
-      "step": 22109
-    },
-    {
-      "epoch": 11.530638852672752,
-      "grad_norm": 1.4996206760406494,
-      "learning_rate": 7.828944723618092e-05,
-      "loss": 5.049,
-      "step": 22110
-    },
-    {
-      "epoch": 11.53116036505867,
-      "grad_norm": 1.3945491313934326,
-      "learning_rate": 7.828844221105528e-05,
-      "loss": 5.4094,
-      "step": 22111
-    },
-    {
-      "epoch": 11.531681877444589,
-      "grad_norm": 1.5715994834899902,
-      "learning_rate": 7.828743718592966e-05,
-      "loss": 5.2852,
-      "step": 22112
-    },
-    {
-      "epoch": 11.532203389830508,
-      "grad_norm": 1.4541016817092896,
-      "learning_rate": 7.828643216080403e-05,
-      "loss": 5.2281,
-      "step": 22113
-    },
-    {
-      "epoch": 11.532724902216428,
-      "grad_norm": 1.3652236461639404,
-      "learning_rate": 7.828542713567839e-05,
-      "loss": 5.684,
-      "step": 22114
-    },
-    {
-      "epoch": 11.533246414602347,
-      "grad_norm": 1.347825050354004,
-      "learning_rate": 7.828442211055277e-05,
-      "loss": 5.6621,
-      "step": 22115
-    },
-    {
-      "epoch": 11.533767926988267,
-      "grad_norm": 1.3632346391677856,
-      "learning_rate": 7.828341708542713e-05,
-      "loss": 5.6357,
-      "step": 22116
-    },
-    {
-      "epoch": 11.534289439374184,
-      "grad_norm": 1.5317305326461792,
-      "learning_rate": 7.828241206030151e-05,
-      "loss": 5.2768,
-      "step": 22117
-    },
-    {
-      "epoch": 11.534810951760104,
-      "grad_norm": 1.414209008216858,
-      "learning_rate": 7.828140703517587e-05,
-      "loss": 5.2025,
-      "step": 22118
-    },
-    {
-      "epoch": 11.535332464146023,
-      "grad_norm": 1.3407833576202393,
-      "learning_rate": 7.828040201005025e-05,
-      "loss": 5.8454,
-      "step": 22119
-    },
-    {
-      "epoch": 11.535853976531943,
-      "grad_norm": 1.468333125114441,
-      "learning_rate": 7.827939698492463e-05,
-      "loss": 5.6834,
-      "step": 22120
-    },
-    {
-      "epoch": 11.536375488917862,
-      "grad_norm": 1.4168390035629272,
-      "learning_rate": 7.827839195979901e-05,
-      "loss": 5.5851,
-      "step": 22121
-    },
-    {
-      "epoch": 11.536897001303782,
-      "grad_norm": 1.4518572092056274,
-      "learning_rate": 7.827738693467337e-05,
-      "loss": 5.6148,
-      "step": 22122
-    },
-    {
-      "epoch": 11.5374185136897,
-      "grad_norm": 1.420469045639038,
-      "learning_rate": 7.827638190954775e-05,
-      "loss": 4.9092,
-      "step": 22123
-    },
-    {
-      "epoch": 11.537940026075619,
-      "grad_norm": 1.3616431951522827,
-      "learning_rate": 7.827537688442211e-05,
-      "loss": 5.1447,
-      "step": 22124
-    },
-    {
-      "epoch": 11.538461538461538,
-      "grad_norm": 1.452860713005066,
-      "learning_rate": 7.827437185929649e-05,
-      "loss": 5.2975,
-      "step": 22125
-    },
-    {
-      "epoch": 11.538983050847458,
-      "grad_norm": 1.3899126052856445,
-      "learning_rate": 7.827336683417086e-05,
-      "loss": 5.4758,
-      "step": 22126
-    },
-    {
-      "epoch": 11.539504563233377,
-      "grad_norm": 1.5371448993682861,
-      "learning_rate": 7.827236180904522e-05,
-      "loss": 5.528,
-      "step": 22127
-    },
-    {
-      "epoch": 11.540026075619297,
-      "grad_norm": 1.4121392965316772,
-      "learning_rate": 7.82713567839196e-05,
-      "loss": 5.2918,
-      "step": 22128
-    },
-    {
-      "epoch": 11.540547588005214,
-      "grad_norm": 1.4460489749908447,
-      "learning_rate": 7.827035175879398e-05,
-      "loss": 5.6393,
-      "step": 22129
-    },
-    {
-      "epoch": 11.541069100391134,
-      "grad_norm": 1.51017427444458,
-      "learning_rate": 7.826934673366835e-05,
-      "loss": 5.7783,
-      "step": 22130
-    },
-    {
-      "epoch": 11.541590612777053,
-      "grad_norm": 1.5760149955749512,
-      "learning_rate": 7.826834170854272e-05,
-      "loss": 5.3629,
-      "step": 22131
-    },
-    {
-      "epoch": 11.542112125162973,
-      "grad_norm": 1.7335904836654663,
-      "learning_rate": 7.82673366834171e-05,
-      "loss": 4.6991,
-      "step": 22132
-    },
-    {
-      "epoch": 11.542633637548892,
-      "grad_norm": 1.3692766427993774,
-      "learning_rate": 7.826633165829146e-05,
-      "loss": 5.542,
-      "step": 22133
-    },
-    {
-      "epoch": 11.543155149934812,
-      "grad_norm": 1.3693904876708984,
-      "learning_rate": 7.826532663316584e-05,
-      "loss": 5.6318,
-      "step": 22134
-    },
-    {
-      "epoch": 11.54367666232073,
-      "grad_norm": 1.4866416454315186,
-      "learning_rate": 7.82643216080402e-05,
-      "loss": 5.3599,
-      "step": 22135
-    },
-    {
-      "epoch": 11.544198174706649,
-      "grad_norm": 1.3466613292694092,
-      "learning_rate": 7.826331658291458e-05,
-      "loss": 5.7759,
-      "step": 22136
-    },
-    {
-      "epoch": 11.544719687092568,
-      "grad_norm": 1.395326018333435,
-      "learning_rate": 7.826231155778894e-05,
-      "loss": 5.6837,
-      "step": 22137
-    },
-    {
-      "epoch": 11.545241199478488,
-      "grad_norm": 1.4039946794509888,
-      "learning_rate": 7.826130653266332e-05,
-      "loss": 5.1831,
-      "step": 22138
-    },
-    {
-      "epoch": 11.545762711864407,
-      "grad_norm": 1.340196967124939,
-      "learning_rate": 7.826030150753769e-05,
-      "loss": 5.4611,
-      "step": 22139
-    },
-    {
-      "epoch": 11.546284224250327,
-      "grad_norm": 1.4165114164352417,
-      "learning_rate": 7.825929648241206e-05,
-      "loss": 5.5677,
-      "step": 22140
-    },
-    {
-      "epoch": 11.546805736636244,
-      "grad_norm": 1.4705196619033813,
-      "learning_rate": 7.825829145728644e-05,
-      "loss": 5.5672,
-      "step": 22141
-    },
-    {
-      "epoch": 11.547327249022164,
-      "grad_norm": 1.3600423336029053,
-      "learning_rate": 7.82572864321608e-05,
-      "loss": 5.7881,
-      "step": 22142
-    },
-    {
-      "epoch": 11.547848761408083,
-      "grad_norm": 1.3753876686096191,
-      "learning_rate": 7.825628140703518e-05,
-      "loss": 5.4428,
-      "step": 22143
-    },
-    {
-      "epoch": 11.548370273794003,
-      "grad_norm": 1.4449810981750488,
-      "learning_rate": 7.825527638190955e-05,
-      "loss": 5.4687,
-      "step": 22144
-    },
-    {
-      "epoch": 11.548891786179922,
-      "grad_norm": 1.4047844409942627,
-      "learning_rate": 7.825427135678393e-05,
-      "loss": 5.5265,
-      "step": 22145
-    },
-    {
-      "epoch": 11.549413298565842,
-      "grad_norm": 1.4163531064987183,
-      "learning_rate": 7.825326633165829e-05,
-      "loss": 5.4158,
-      "step": 22146
-    },
-    {
-      "epoch": 11.54993481095176,
-      "grad_norm": 1.4283349514007568,
-      "learning_rate": 7.825226130653267e-05,
-      "loss": 5.6068,
-      "step": 22147
-    },
-    {
-      "epoch": 11.550456323337679,
-      "grad_norm": 1.5690315961837769,
-      "learning_rate": 7.825125628140703e-05,
-      "loss": 5.3684,
-      "step": 22148
-    },
-    {
-      "epoch": 11.550977835723598,
-      "grad_norm": 1.3714631795883179,
-      "learning_rate": 7.825025125628141e-05,
-      "loss": 5.8475,
-      "step": 22149
-    },
-    {
-      "epoch": 11.551499348109518,
-      "grad_norm": 1.3561311960220337,
-      "learning_rate": 7.824924623115579e-05,
-      "loss": 5.3824,
-      "step": 22150
-    },
-    {
-      "epoch": 11.552020860495437,
-      "grad_norm": 1.3751566410064697,
-      "learning_rate": 7.824824120603017e-05,
-      "loss": 5.7811,
-      "step": 22151
-    },
-    {
-      "epoch": 11.552542372881355,
-      "grad_norm": 1.5321232080459595,
-      "learning_rate": 7.824723618090453e-05,
-      "loss": 5.2429,
-      "step": 22152
-    },
-    {
-      "epoch": 11.553063885267274,
-      "grad_norm": 1.5304880142211914,
-      "learning_rate": 7.82462311557789e-05,
-      "loss": 5.7509,
-      "step": 22153
-    },
-    {
-      "epoch": 11.553585397653194,
-      "grad_norm": 1.4711540937423706,
-      "learning_rate": 7.824522613065327e-05,
-      "loss": 5.4732,
-      "step": 22154
-    },
-    {
-      "epoch": 11.554106910039113,
-      "grad_norm": 1.559682011604309,
-      "learning_rate": 7.824422110552764e-05,
-      "loss": 5.0288,
-      "step": 22155
-    },
-    {
-      "epoch": 11.554628422425033,
-      "grad_norm": 1.4623527526855469,
-      "learning_rate": 7.824321608040201e-05,
-      "loss": 5.533,
-      "step": 22156
-    },
-    {
-      "epoch": 11.555149934810952,
-      "grad_norm": 1.4227215051651,
-      "learning_rate": 7.824221105527638e-05,
-      "loss": 5.4402,
-      "step": 22157
-    },
-    {
-      "epoch": 11.555671447196872,
-      "grad_norm": 1.4006030559539795,
-      "learning_rate": 7.824120603015076e-05,
-      "loss": 5.6528,
-      "step": 22158
-    },
-    {
-      "epoch": 11.55619295958279,
-      "grad_norm": 1.4222112894058228,
-      "learning_rate": 7.824020100502512e-05,
-      "loss": 5.5376,
-      "step": 22159
-    },
-    {
-      "epoch": 11.556714471968709,
-      "grad_norm": 1.5910199880599976,
-      "learning_rate": 7.82391959798995e-05,
-      "loss": 5.2232,
-      "step": 22160
-    },
-    {
-      "epoch": 11.557235984354628,
-      "grad_norm": 1.489809513092041,
-      "learning_rate": 7.823819095477388e-05,
-      "loss": 5.2796,
-      "step": 22161
-    },
-    {
-      "epoch": 11.557757496740548,
-      "grad_norm": 1.4213695526123047,
-      "learning_rate": 7.823718592964825e-05,
-      "loss": 5.7613,
-      "step": 22162
-    },
-    {
-      "epoch": 11.558279009126467,
-      "grad_norm": 1.4982469081878662,
-      "learning_rate": 7.823618090452262e-05,
-      "loss": 5.256,
-      "step": 22163
-    },
-    {
-      "epoch": 11.558800521512385,
-      "grad_norm": 1.4811235666275024,
-      "learning_rate": 7.8235175879397e-05,
-      "loss": 5.565,
-      "step": 22164
-    },
-    {
-      "epoch": 11.559322033898304,
-      "grad_norm": 1.2767549753189087,
-      "learning_rate": 7.823417085427136e-05,
-      "loss": 5.8773,
-      "step": 22165
-    },
-    {
-      "epoch": 11.559843546284224,
-      "grad_norm": 1.4317289590835571,
-      "learning_rate": 7.823316582914572e-05,
-      "loss": 5.1842,
-      "step": 22166
-    },
-    {
-      "epoch": 11.560365058670143,
-      "grad_norm": 1.5952823162078857,
-      "learning_rate": 7.82321608040201e-05,
-      "loss": 4.9602,
-      "step": 22167
-    },
-    {
-      "epoch": 11.560886571056063,
-      "grad_norm": 1.469472050666809,
-      "learning_rate": 7.823115577889447e-05,
-      "loss": 5.705,
-      "step": 22168
-    },
-    {
-      "epoch": 11.561408083441982,
-      "grad_norm": 1.5273867845535278,
-      "learning_rate": 7.823015075376884e-05,
-      "loss": 5.5779,
-      "step": 22169
-    },
-    {
-      "epoch": 11.561929595827902,
-      "grad_norm": 1.5344551801681519,
-      "learning_rate": 7.822914572864322e-05,
-      "loss": 5.6934,
-      "step": 22170
-    },
-    {
-      "epoch": 11.56245110821382,
-      "grad_norm": 1.4314265251159668,
-      "learning_rate": 7.82281407035176e-05,
-      "loss": 5.2901,
-      "step": 22171
-    },
-    {
-      "epoch": 11.562972620599739,
-      "grad_norm": 1.3685470819473267,
-      "learning_rate": 7.822713567839196e-05,
-      "loss": 5.5199,
-      "step": 22172
-    },
-    {
-      "epoch": 11.563494132985658,
-      "grad_norm": 1.5146431922912598,
-      "learning_rate": 7.822613065326634e-05,
-      "loss": 5.3411,
-      "step": 22173
-    },
-    {
-      "epoch": 11.564015645371578,
-      "grad_norm": 1.3759765625,
-      "learning_rate": 7.822512562814071e-05,
-      "loss": 5.7251,
-      "step": 22174
-    },
-    {
-      "epoch": 11.564537157757497,
-      "grad_norm": 1.4317816495895386,
-      "learning_rate": 7.822412060301508e-05,
-      "loss": 5.6166,
-      "step": 22175
-    },
-    {
-      "epoch": 11.565058670143415,
-      "grad_norm": 1.3486006259918213,
-      "learning_rate": 7.822311557788945e-05,
-      "loss": 5.7524,
-      "step": 22176
-    },
-    {
-      "epoch": 11.565580182529335,
-      "grad_norm": 1.3786684274673462,
-      "learning_rate": 7.822211055276383e-05,
-      "loss": 5.7553,
-      "step": 22177
-    },
-    {
-      "epoch": 11.566101694915254,
-      "grad_norm": 1.4051475524902344,
-      "learning_rate": 7.822110552763819e-05,
-      "loss": 5.6098,
-      "step": 22178
-    },
-    {
-      "epoch": 11.566623207301173,
-      "grad_norm": 1.435900092124939,
-      "learning_rate": 7.822010050251256e-05,
-      "loss": 5.5072,
-      "step": 22179
-    },
-    {
-      "epoch": 11.567144719687093,
-      "grad_norm": 1.5103262662887573,
-      "learning_rate": 7.821909547738693e-05,
-      "loss": 5.0764,
-      "step": 22180
-    },
-    {
-      "epoch": 11.567666232073012,
-      "grad_norm": 1.4245679378509521,
-      "learning_rate": 7.821809045226131e-05,
-      "loss": 5.5014,
-      "step": 22181
-    },
-    {
-      "epoch": 11.568187744458932,
-      "grad_norm": 1.4157688617706299,
-      "learning_rate": 7.821708542713569e-05,
-      "loss": 5.7409,
-      "step": 22182
-    },
-    {
-      "epoch": 11.56870925684485,
-      "grad_norm": 1.5170047283172607,
-      "learning_rate": 7.821608040201005e-05,
-      "loss": 5.0676,
-      "step": 22183
-    },
-    {
-      "epoch": 11.569230769230769,
-      "grad_norm": 1.492152452468872,
-      "learning_rate": 7.821507537688443e-05,
-      "loss": 5.5366,
-      "step": 22184
-    },
-    {
-      "epoch": 11.569752281616688,
-      "grad_norm": 1.394702434539795,
-      "learning_rate": 7.82140703517588e-05,
-      "loss": 5.5894,
-      "step": 22185
-    },
-    {
-      "epoch": 11.570273794002608,
-      "grad_norm": 1.4771320819854736,
-      "learning_rate": 7.821306532663317e-05,
-      "loss": 4.2992,
-      "step": 22186
-    },
-    {
-      "epoch": 11.570795306388527,
-      "grad_norm": 1.5010353326797485,
-      "learning_rate": 7.821206030150754e-05,
-      "loss": 5.343,
-      "step": 22187
-    },
-    {
-      "epoch": 11.571316818774445,
-      "grad_norm": 1.4606380462646484,
-      "learning_rate": 7.821105527638192e-05,
-      "loss": 5.5141,
-      "step": 22188
-    },
-    {
-      "epoch": 11.571838331160365,
-      "grad_norm": 1.5737509727478027,
-      "learning_rate": 7.821005025125628e-05,
-      "loss": 5.0212,
-      "step": 22189
-    },
-    {
-      "epoch": 11.572359843546284,
-      "grad_norm": 1.3477305173873901,
-      "learning_rate": 7.820904522613066e-05,
-      "loss": 5.7367,
-      "step": 22190
-    },
-    {
-      "epoch": 11.572881355932203,
-      "grad_norm": 1.336254358291626,
-      "learning_rate": 7.820804020100504e-05,
-      "loss": 5.5176,
-      "step": 22191
-    },
-    {
-      "epoch": 11.573402868318123,
-      "grad_norm": 1.5757997035980225,
-      "learning_rate": 7.820703517587941e-05,
-      "loss": 5.3041,
-      "step": 22192
-    },
-    {
-      "epoch": 11.573924380704042,
-      "grad_norm": 1.4285372495651245,
-      "learning_rate": 7.820603015075378e-05,
-      "loss": 5.7026,
-      "step": 22193
-    },
-    {
-      "epoch": 11.57444589308996,
-      "grad_norm": 1.510820984840393,
-      "learning_rate": 7.820502512562814e-05,
-      "loss": 5.2068,
-      "step": 22194
-    },
-    {
-      "epoch": 11.57496740547588,
-      "grad_norm": 1.5256520509719849,
-      "learning_rate": 7.820402010050252e-05,
-      "loss": 5.0884,
-      "step": 22195
-    },
-    {
-      "epoch": 11.575488917861799,
-      "grad_norm": 1.5937970876693726,
-      "learning_rate": 7.820301507537688e-05,
-      "loss": 4.8922,
-      "step": 22196
-    },
-    {
-      "epoch": 11.576010430247718,
-      "grad_norm": 1.4253637790679932,
-      "learning_rate": 7.820201005025126e-05,
-      "loss": 5.2783,
-      "step": 22197
-    },
-    {
-      "epoch": 11.576531942633638,
-      "grad_norm": 1.513887643814087,
-      "learning_rate": 7.820100502512563e-05,
-      "loss": 4.9653,
-      "step": 22198
-    },
-    {
-      "epoch": 11.577053455019557,
-      "grad_norm": 1.5422521829605103,
-      "learning_rate": 7.82e-05,
-      "loss": 5.2293,
-      "step": 22199
-    },
-    {
-      "epoch": 11.577574967405475,
-      "grad_norm": 1.387839674949646,
-      "learning_rate": 7.819899497487437e-05,
-      "loss": 4.7921,
-      "step": 22200
-    },
-    {
-      "epoch": 11.578096479791395,
-      "grad_norm": 1.3379532098770142,
-      "learning_rate": 7.819798994974875e-05,
-      "loss": 5.6958,
-      "step": 22201
-    },
-    {
-      "epoch": 11.578617992177314,
-      "grad_norm": 1.357678771018982,
-      "learning_rate": 7.819698492462312e-05,
-      "loss": 5.5383,
-      "step": 22202
-    },
-    {
-      "epoch": 11.579139504563233,
-      "grad_norm": 1.4132375717163086,
-      "learning_rate": 7.81959798994975e-05,
-      "loss": 5.4977,
-      "step": 22203
-    },
-    {
-      "epoch": 11.579661016949153,
-      "grad_norm": 1.4325311183929443,
-      "learning_rate": 7.819497487437187e-05,
-      "loss": 5.6126,
-      "step": 22204
-    },
-    {
-      "epoch": 11.580182529335072,
-      "grad_norm": 1.4130209684371948,
-      "learning_rate": 7.819396984924624e-05,
-      "loss": 5.6472,
-      "step": 22205
-    },
-    {
-      "epoch": 11.58070404172099,
-      "grad_norm": 1.614508867263794,
-      "learning_rate": 7.819296482412061e-05,
-      "loss": 4.8391,
-      "step": 22206
-    },
-    {
-      "epoch": 11.58122555410691,
-      "grad_norm": 1.4867111444473267,
-      "learning_rate": 7.819195979899497e-05,
-      "loss": 5.2927,
-      "step": 22207
-    },
-    {
-      "epoch": 11.581747066492829,
-      "grad_norm": 1.3977354764938354,
-      "learning_rate": 7.819095477386935e-05,
-      "loss": 5.2194,
-      "step": 22208
-    },
-    {
-      "epoch": 11.582268578878748,
-      "grad_norm": 1.4261736869812012,
-      "learning_rate": 7.818994974874371e-05,
-      "loss": 5.316,
-      "step": 22209
-    },
-    {
-      "epoch": 11.582790091264668,
-      "grad_norm": 1.4234219789505005,
-      "learning_rate": 7.818894472361809e-05,
-      "loss": 5.0744,
-      "step": 22210
-    },
-    {
-      "epoch": 11.583311603650587,
-      "grad_norm": 1.4661136865615845,
-      "learning_rate": 7.818793969849246e-05,
-      "loss": 5.8166,
-      "step": 22211
-    },
-    {
-      "epoch": 11.583833116036505,
-      "grad_norm": 1.4551454782485962,
-      "learning_rate": 7.818693467336683e-05,
-      "loss": 5.2902,
-      "step": 22212
-    },
-    {
-      "epoch": 11.584354628422425,
-      "grad_norm": 1.406449317932129,
-      "learning_rate": 7.818592964824121e-05,
-      "loss": 5.8406,
-      "step": 22213
-    },
-    {
-      "epoch": 11.584876140808344,
-      "grad_norm": 1.5082216262817383,
-      "learning_rate": 7.818492462311559e-05,
-      "loss": 5.2295,
-      "step": 22214
-    },
-    {
-      "epoch": 11.585397653194264,
-      "grad_norm": 1.495180368423462,
-      "learning_rate": 7.818391959798995e-05,
-      "loss": 4.3825,
-      "step": 22215
-    },
-    {
-      "epoch": 11.585919165580183,
-      "grad_norm": 1.4745031595230103,
-      "learning_rate": 7.818291457286433e-05,
-      "loss": 5.7416,
-      "step": 22216
-    },
-    {
-      "epoch": 11.586440677966102,
-      "grad_norm": 1.6791819334030151,
-      "learning_rate": 7.81819095477387e-05,
-      "loss": 5.1544,
-      "step": 22217
-    },
-    {
-      "epoch": 11.58696219035202,
-      "grad_norm": 1.478054404258728,
-      "learning_rate": 7.818090452261307e-05,
-      "loss": 5.5913,
-      "step": 22218
-    },
-    {
-      "epoch": 11.58748370273794,
-      "grad_norm": 1.4979058504104614,
-      "learning_rate": 7.817989949748744e-05,
-      "loss": 5.3803,
-      "step": 22219
-    },
-    {
-      "epoch": 11.588005215123859,
-      "grad_norm": 1.4696394205093384,
-      "learning_rate": 7.81788944723618e-05,
-      "loss": 5.3836,
-      "step": 22220
-    },
-    {
-      "epoch": 11.588526727509779,
-      "grad_norm": 1.5257624387741089,
-      "learning_rate": 7.817788944723618e-05,
-      "loss": 5.3322,
-      "step": 22221
-    },
-    {
-      "epoch": 11.589048239895698,
-      "grad_norm": 1.6462492942810059,
-      "learning_rate": 7.817688442211056e-05,
-      "loss": 5.0342,
-      "step": 22222
-    },
-    {
-      "epoch": 11.589569752281617,
-      "grad_norm": 1.5049477815628052,
-      "learning_rate": 7.817587939698494e-05,
-      "loss": 5.5866,
-      "step": 22223
-    },
-    {
-      "epoch": 11.590091264667535,
-      "grad_norm": 1.5852024555206299,
-      "learning_rate": 7.81748743718593e-05,
-      "loss": 5.4811,
-      "step": 22224
-    },
-    {
-      "epoch": 11.590612777053455,
-      "grad_norm": 1.533625841140747,
-      "learning_rate": 7.817386934673368e-05,
-      "loss": 5.5069,
-      "step": 22225
-    },
-    {
-      "epoch": 11.591134289439374,
-      "grad_norm": 1.6040693521499634,
-      "learning_rate": 7.817286432160804e-05,
-      "loss": 5.3857,
-      "step": 22226
-    },
-    {
-      "epoch": 11.591655801825294,
-      "grad_norm": 1.485075831413269,
-      "learning_rate": 7.817185929648242e-05,
-      "loss": 5.1815,
-      "step": 22227
-    },
-    {
-      "epoch": 11.592177314211213,
-      "grad_norm": 1.6008497476577759,
-      "learning_rate": 7.817085427135678e-05,
-      "loss": 5.3485,
-      "step": 22228
-    },
-    {
-      "epoch": 11.592698826597132,
-      "grad_norm": 1.4957135915756226,
-      "learning_rate": 7.816984924623116e-05,
-      "loss": 5.5023,
-      "step": 22229
-    },
-    {
-      "epoch": 11.59322033898305,
-      "grad_norm": 1.3759287595748901,
-      "learning_rate": 7.816884422110553e-05,
-      "loss": 5.6067,
-      "step": 22230
-    },
-    {
-      "epoch": 11.59374185136897,
-      "grad_norm": 1.5668936967849731,
-      "learning_rate": 7.81678391959799e-05,
-      "loss": 5.0459,
-      "step": 22231
-    },
-    {
-      "epoch": 11.594263363754889,
-      "grad_norm": 1.4681673049926758,
-      "learning_rate": 7.816683417085427e-05,
-      "loss": 5.3947,
-      "step": 22232
-    },
-    {
-      "epoch": 11.594784876140809,
-      "grad_norm": 1.4406111240386963,
-      "learning_rate": 7.816582914572865e-05,
-      "loss": 5.4428,
-      "step": 22233
-    },
-    {
-      "epoch": 11.595306388526728,
-      "grad_norm": 1.4699944257736206,
-      "learning_rate": 7.816482412060302e-05,
-      "loss": 5.175,
-      "step": 22234
-    },
-    {
-      "epoch": 11.595827900912647,
-      "grad_norm": 1.393046498298645,
-      "learning_rate": 7.816381909547739e-05,
-      "loss": 5.4501,
-      "step": 22235
-    },
-    {
-      "epoch": 11.596349413298565,
-      "grad_norm": 1.457277774810791,
-      "learning_rate": 7.816281407035177e-05,
-      "loss": 5.3715,
-      "step": 22236
-    },
-    {
-      "epoch": 11.596870925684485,
-      "grad_norm": 1.4374773502349854,
-      "learning_rate": 7.816180904522613e-05,
-      "loss": 5.7428,
-      "step": 22237
-    },
-    {
-      "epoch": 11.597392438070404,
-      "grad_norm": 1.4734430313110352,
-      "learning_rate": 7.816080402010051e-05,
-      "loss": 5.4347,
-      "step": 22238
-    },
-    {
-      "epoch": 11.597913950456324,
-      "grad_norm": 1.4397342205047607,
-      "learning_rate": 7.815979899497487e-05,
-      "loss": 5.4129,
-      "step": 22239
-    },
-    {
-      "epoch": 11.598435462842243,
-      "grad_norm": 1.5833195447921753,
-      "learning_rate": 7.815879396984925e-05,
-      "loss": 5.1467,
-      "step": 22240
-    },
-    {
-      "epoch": 11.598956975228162,
-      "grad_norm": 1.5682786703109741,
-      "learning_rate": 7.815778894472361e-05,
-      "loss": 5.7073,
-      "step": 22241
-    },
-    {
-      "epoch": 11.59947848761408,
-      "grad_norm": 1.5889302492141724,
-      "learning_rate": 7.815678391959799e-05,
-      "loss": 5.4516,
-      "step": 22242
-    },
-    {
-      "epoch": 11.6,
-      "grad_norm": 1.4085415601730347,
-      "learning_rate": 7.815577889447237e-05,
-      "loss": 5.7257,
-      "step": 22243
-    },
-    {
-      "epoch": 11.600521512385919,
-      "grad_norm": 1.5042115449905396,
-      "learning_rate": 7.815477386934675e-05,
-      "loss": 5.2843,
-      "step": 22244
-    },
-    {
-      "epoch": 11.601043024771839,
-      "grad_norm": 1.4303078651428223,
-      "learning_rate": 7.815376884422111e-05,
-      "loss": 5.4366,
-      "step": 22245
-    },
-    {
-      "epoch": 11.601564537157758,
-      "grad_norm": 1.5163918733596802,
-      "learning_rate": 7.815276381909548e-05,
-      "loss": 5.2764,
-      "step": 22246
-    },
-    {
-      "epoch": 11.602086049543676,
-      "grad_norm": 1.5033605098724365,
-      "learning_rate": 7.815175879396985e-05,
-      "loss": 5.7736,
-      "step": 22247
-    },
-    {
-      "epoch": 11.602607561929595,
-      "grad_norm": 1.3807663917541504,
-      "learning_rate": 7.815075376884422e-05,
-      "loss": 4.773,
-      "step": 22248
-    },
-    {
-      "epoch": 11.603129074315515,
-      "grad_norm": 1.5445321798324585,
-      "learning_rate": 7.81497487437186e-05,
-      "loss": 5.6875,
-      "step": 22249
-    },
-    {
-      "epoch": 11.603650586701434,
-      "grad_norm": 1.6280958652496338,
-      "learning_rate": 7.814874371859296e-05,
-      "loss": 4.6392,
-      "step": 22250
-    },
-    {
-      "epoch": 11.604172099087354,
-      "grad_norm": 1.5137485265731812,
-      "learning_rate": 7.814773869346734e-05,
-      "loss": 5.1257,
-      "step": 22251
-    },
-    {
-      "epoch": 11.604693611473273,
-      "grad_norm": 1.5387120246887207,
-      "learning_rate": 7.81467336683417e-05,
-      "loss": 4.966,
-      "step": 22252
-    },
-    {
-      "epoch": 11.605215123859193,
-      "grad_norm": 1.3929563760757446,
-      "learning_rate": 7.814572864321608e-05,
-      "loss": 5.5264,
-      "step": 22253
-    },
-    {
-      "epoch": 11.60573663624511,
-      "grad_norm": 1.372735619544983,
-      "learning_rate": 7.814472361809046e-05,
-      "loss": 5.4844,
-      "step": 22254
-    },
-    {
-      "epoch": 11.60625814863103,
-      "grad_norm": 1.4124233722686768,
-      "learning_rate": 7.814371859296484e-05,
-      "loss": 5.2074,
-      "step": 22255
-    },
-    {
-      "epoch": 11.60677966101695,
-      "grad_norm": 1.4083153009414673,
-      "learning_rate": 7.81427135678392e-05,
-      "loss": 5.0566,
-      "step": 22256
-    },
-    {
-      "epoch": 11.607301173402869,
-      "grad_norm": 1.4077048301696777,
-      "learning_rate": 7.814170854271358e-05,
-      "loss": 5.2411,
-      "step": 22257
-    },
-    {
-      "epoch": 11.607822685788788,
-      "grad_norm": 1.5373262166976929,
-      "learning_rate": 7.814070351758794e-05,
-      "loss": 5.5414,
-      "step": 22258
-    },
-    {
-      "epoch": 11.608344198174706,
-      "grad_norm": 1.491277813911438,
-      "learning_rate": 7.813969849246231e-05,
-      "loss": 5.3129,
-      "step": 22259
-    },
-    {
-      "epoch": 11.608865710560625,
-      "grad_norm": 1.3822565078735352,
-      "learning_rate": 7.813869346733669e-05,
-      "loss": 4.7661,
-      "step": 22260
-    },
-    {
-      "epoch": 11.609387222946545,
-      "grad_norm": 1.4708571434020996,
-      "learning_rate": 7.813768844221105e-05,
-      "loss": 5.5881,
-      "step": 22261
-    },
-    {
-      "epoch": 11.609908735332464,
-      "grad_norm": 1.4857215881347656,
-      "learning_rate": 7.813668341708543e-05,
-      "loss": 5.4841,
-      "step": 22262
-    },
-    {
-      "epoch": 11.610430247718384,
-      "grad_norm": 1.5040340423583984,
-      "learning_rate": 7.81356783919598e-05,
-      "loss": 5.1947,
-      "step": 22263
-    },
-    {
-      "epoch": 11.610951760104303,
-      "grad_norm": 1.5267579555511475,
-      "learning_rate": 7.813467336683418e-05,
-      "loss": 5.4333,
-      "step": 22264
-    },
-    {
-      "epoch": 11.611473272490223,
-      "grad_norm": 1.4858763217926025,
-      "learning_rate": 7.813366834170855e-05,
-      "loss": 5.1045,
-      "step": 22265
-    },
-    {
-      "epoch": 11.61199478487614,
-      "grad_norm": 1.4751107692718506,
-      "learning_rate": 7.813266331658292e-05,
-      "loss": 5.1586,
-      "step": 22266
-    },
-    {
-      "epoch": 11.61251629726206,
-      "grad_norm": 1.4431971311569214,
-      "learning_rate": 7.813165829145729e-05,
-      "loss": 5.7895,
-      "step": 22267
-    },
-    {
-      "epoch": 11.61303780964798,
-      "grad_norm": 1.53630793094635,
-      "learning_rate": 7.813065326633167e-05,
-      "loss": 4.5753,
-      "step": 22268
-    },
-    {
-      "epoch": 11.613559322033899,
-      "grad_norm": 1.3435505628585815,
-      "learning_rate": 7.812964824120603e-05,
-      "loss": 5.6129,
-      "step": 22269
-    },
-    {
-      "epoch": 11.614080834419818,
-      "grad_norm": 1.3443918228149414,
-      "learning_rate": 7.812864321608041e-05,
-      "loss": 5.3278,
-      "step": 22270
-    },
-    {
-      "epoch": 11.614602346805736,
-      "grad_norm": 1.3864680528640747,
-      "learning_rate": 7.812763819095477e-05,
-      "loss": 5.2531,
-      "step": 22271
-    },
-    {
-      "epoch": 11.615123859191655,
-      "grad_norm": 1.5269242525100708,
-      "learning_rate": 7.812663316582914e-05,
-      "loss": 5.2768,
-      "step": 22272
-    },
-    {
-      "epoch": 11.615645371577575,
-      "grad_norm": 1.4812111854553223,
-      "learning_rate": 7.812562814070352e-05,
-      "loss": 5.2434,
-      "step": 22273
-    },
-    {
-      "epoch": 11.616166883963494,
-      "grad_norm": 1.4119127988815308,
-      "learning_rate": 7.81246231155779e-05,
-      "loss": 5.4174,
-      "step": 22274
-    },
-    {
-      "epoch": 11.616688396349414,
-      "grad_norm": 1.3582298755645752,
-      "learning_rate": 7.812361809045227e-05,
-      "loss": 4.9632,
-      "step": 22275
-    },
-    {
-      "epoch": 11.617209908735333,
-      "grad_norm": 1.399513602256775,
-      "learning_rate": 7.812261306532664e-05,
-      "loss": 5.5916,
-      "step": 22276
-    },
-    {
-      "epoch": 11.617731421121253,
-      "grad_norm": 1.398338794708252,
-      "learning_rate": 7.812160804020101e-05,
-      "loss": 5.5129,
-      "step": 22277
-    },
-    {
-      "epoch": 11.61825293350717,
-      "grad_norm": 1.4259601831436157,
-      "learning_rate": 7.812060301507538e-05,
-      "loss": 5.3897,
-      "step": 22278
-    },
-    {
-      "epoch": 11.61877444589309,
-      "grad_norm": 1.4211174249649048,
-      "learning_rate": 7.811959798994976e-05,
-      "loss": 5.264,
-      "step": 22279
-    },
-    {
-      "epoch": 11.61929595827901,
-      "grad_norm": 1.3958979845046997,
-      "learning_rate": 7.811859296482412e-05,
-      "loss": 5.358,
-      "step": 22280
-    },
-    {
-      "epoch": 11.619817470664929,
-      "grad_norm": 1.357175588607788,
-      "learning_rate": 7.81175879396985e-05,
-      "loss": 5.6092,
-      "step": 22281
-    },
-    {
-      "epoch": 11.620338983050848,
-      "grad_norm": 1.41832435131073,
-      "learning_rate": 7.811658291457286e-05,
-      "loss": 5.0918,
-      "step": 22282
-    },
-    {
-      "epoch": 11.620860495436766,
-      "grad_norm": 1.4910151958465576,
-      "learning_rate": 7.811557788944724e-05,
-      "loss": 5.6387,
-      "step": 22283
-    },
-    {
-      "epoch": 11.621382007822685,
-      "grad_norm": 1.4622622728347778,
-      "learning_rate": 7.811457286432162e-05,
-      "loss": 5.549,
-      "step": 22284
-    },
-    {
-      "epoch": 11.621903520208605,
-      "grad_norm": 1.6053091287612915,
-      "learning_rate": 7.8113567839196e-05,
-      "loss": 4.9446,
-      "step": 22285
-    },
-    {
-      "epoch": 11.622425032594524,
-      "grad_norm": 1.412048578262329,
-      "learning_rate": 7.811256281407036e-05,
-      "loss": 5.577,
-      "step": 22286
-    },
-    {
-      "epoch": 11.622946544980444,
-      "grad_norm": 1.48428475856781,
-      "learning_rate": 7.811155778894472e-05,
-      "loss": 5.4005,
-      "step": 22287
-    },
-    {
-      "epoch": 11.623468057366363,
-      "grad_norm": 1.4788563251495361,
-      "learning_rate": 7.81105527638191e-05,
-      "loss": 5.4403,
-      "step": 22288
-    },
-    {
-      "epoch": 11.62398956975228,
-      "grad_norm": 1.4281319379806519,
-      "learning_rate": 7.810954773869347e-05,
-      "loss": 5.5595,
-      "step": 22289
-    },
-    {
-      "epoch": 11.6245110821382,
-      "grad_norm": 1.5275001525878906,
-      "learning_rate": 7.810854271356784e-05,
-      "loss": 5.5628,
-      "step": 22290
-    },
-    {
-      "epoch": 11.62503259452412,
-      "grad_norm": 1.4414547681808472,
-      "learning_rate": 7.810753768844221e-05,
-      "loss": 5.5918,
-      "step": 22291
-    },
-    {
-      "epoch": 11.62555410691004,
-      "grad_norm": 1.4442588090896606,
-      "learning_rate": 7.810653266331659e-05,
-      "loss": 5.333,
-      "step": 22292
-    },
-    {
-      "epoch": 11.626075619295959,
-      "grad_norm": 1.4458249807357788,
-      "learning_rate": 7.810552763819095e-05,
-      "loss": 5.1259,
-      "step": 22293
-    },
-    {
-      "epoch": 11.626597131681878,
-      "grad_norm": 1.5442019701004028,
-      "learning_rate": 7.810452261306533e-05,
-      "loss": 5.5105,
-      "step": 22294
-    },
-    {
-      "epoch": 11.627118644067796,
-      "grad_norm": 1.4487329721450806,
-      "learning_rate": 7.81035175879397e-05,
-      "loss": 5.5016,
-      "step": 22295
-    },
-    {
-      "epoch": 11.627640156453715,
-      "grad_norm": 1.382688045501709,
-      "learning_rate": 7.810251256281408e-05,
-      "loss": 5.9889,
-      "step": 22296
-    },
-    {
-      "epoch": 11.628161668839635,
-      "grad_norm": 1.4224717617034912,
-      "learning_rate": 7.810150753768845e-05,
-      "loss": 5.5586,
-      "step": 22297
-    },
-    {
-      "epoch": 11.628683181225554,
-      "grad_norm": 1.4151055812835693,
-      "learning_rate": 7.810050251256283e-05,
-      "loss": 5.7615,
-      "step": 22298
-    },
-    {
-      "epoch": 11.629204693611474,
-      "grad_norm": 1.626007080078125,
-      "learning_rate": 7.809949748743719e-05,
-      "loss": 4.6048,
-      "step": 22299
-    },
-    {
-      "epoch": 11.629726205997393,
-      "grad_norm": 1.429244875907898,
-      "learning_rate": 7.809849246231155e-05,
-      "loss": 5.3918,
-      "step": 22300
-    },
-    {
-      "epoch": 11.63024771838331,
-      "grad_norm": 1.5034164190292358,
-      "learning_rate": 7.809748743718593e-05,
-      "loss": 5.7503,
-      "step": 22301
-    },
-    {
-      "epoch": 11.63076923076923,
-      "grad_norm": 1.5023574829101562,
-      "learning_rate": 7.80964824120603e-05,
-      "loss": 4.8049,
-      "step": 22302
-    },
-    {
-      "epoch": 11.63129074315515,
-      "grad_norm": 1.521430492401123,
-      "learning_rate": 7.809547738693467e-05,
-      "loss": 5.3649,
-      "step": 22303
-    },
-    {
-      "epoch": 11.63181225554107,
-      "grad_norm": 1.4477745294570923,
-      "learning_rate": 7.809447236180905e-05,
-      "loss": 4.9434,
-      "step": 22304
-    },
-    {
-      "epoch": 11.632333767926989,
-      "grad_norm": 1.36488676071167,
-      "learning_rate": 7.809346733668343e-05,
-      "loss": 5.58,
-      "step": 22305
-    },
-    {
-      "epoch": 11.632855280312908,
-      "grad_norm": 1.4158498048782349,
-      "learning_rate": 7.80924623115578e-05,
-      "loss": 5.7401,
-      "step": 22306
-    },
-    {
-      "epoch": 11.633376792698826,
-      "grad_norm": 1.5280967950820923,
-      "learning_rate": 7.809145728643217e-05,
-      "loss": 5.7082,
-      "step": 22307
-    },
-    {
-      "epoch": 11.633898305084745,
-      "grad_norm": 1.3891371488571167,
-      "learning_rate": 7.809045226130654e-05,
-      "loss": 5.1931,
-      "step": 22308
-    },
-    {
-      "epoch": 11.634419817470665,
-      "grad_norm": 1.3955918550491333,
-      "learning_rate": 7.808944723618091e-05,
-      "loss": 5.7935,
-      "step": 22309
-    },
-    {
-      "epoch": 11.634941329856584,
-      "grad_norm": 1.3654305934906006,
-      "learning_rate": 7.808844221105528e-05,
-      "loss": 5.5312,
-      "step": 22310
-    },
-    {
-      "epoch": 11.635462842242504,
-      "grad_norm": 1.464335560798645,
-      "learning_rate": 7.808743718592966e-05,
-      "loss": 5.5765,
-      "step": 22311
-    },
-    {
-      "epoch": 11.635984354628423,
-      "grad_norm": 1.3539581298828125,
-      "learning_rate": 7.808643216080402e-05,
-      "loss": 5.574,
-      "step": 22312
-    },
-    {
-      "epoch": 11.63650586701434,
-      "grad_norm": 1.4545167684555054,
-      "learning_rate": 7.808542713567838e-05,
-      "loss": 4.9999,
-      "step": 22313
-    },
-    {
-      "epoch": 11.63702737940026,
-      "grad_norm": 1.513115644454956,
-      "learning_rate": 7.808442211055276e-05,
-      "loss": 5.397,
-      "step": 22314
-    },
-    {
-      "epoch": 11.63754889178618,
-      "grad_norm": 1.4858161211013794,
-      "learning_rate": 7.808341708542714e-05,
-      "loss": 5.0545,
-      "step": 22315
-    },
-    {
-      "epoch": 11.6380704041721,
-      "grad_norm": 1.377718210220337,
-      "learning_rate": 7.808241206030152e-05,
-      "loss": 5.8143,
-      "step": 22316
-    },
-    {
-      "epoch": 11.638591916558019,
-      "grad_norm": 1.3916711807250977,
-      "learning_rate": 7.808140703517588e-05,
-      "loss": 5.4029,
-      "step": 22317
-    },
-    {
-      "epoch": 11.639113428943938,
-      "grad_norm": 1.3970084190368652,
-      "learning_rate": 7.808040201005026e-05,
-      "loss": 5.9787,
-      "step": 22318
-    },
-    {
-      "epoch": 11.639634941329856,
-      "grad_norm": 1.6625604629516602,
-      "learning_rate": 7.807939698492462e-05,
-      "loss": 5.1852,
-      "step": 22319
-    },
-    {
-      "epoch": 11.640156453715775,
-      "grad_norm": 1.5090150833129883,
-      "learning_rate": 7.8078391959799e-05,
-      "loss": 5.1911,
-      "step": 22320
-    },
-    {
-      "epoch": 11.640677966101695,
-      "grad_norm": 1.5273280143737793,
-      "learning_rate": 7.807738693467337e-05,
-      "loss": 5.0253,
-      "step": 22321
-    },
-    {
-      "epoch": 11.641199478487614,
-      "grad_norm": 1.638242483139038,
-      "learning_rate": 7.807638190954774e-05,
-      "loss": 5.3564,
-      "step": 22322
-    },
-    {
-      "epoch": 11.641720990873534,
-      "grad_norm": 1.428669810295105,
-      "learning_rate": 7.807537688442211e-05,
-      "loss": 5.8484,
-      "step": 22323
-    },
-    {
-      "epoch": 11.642242503259453,
-      "grad_norm": 1.559396505355835,
-      "learning_rate": 7.807437185929649e-05,
-      "loss": 5.59,
-      "step": 22324
-    },
-    {
-      "epoch": 11.642764015645371,
-      "grad_norm": 1.5145939588546753,
-      "learning_rate": 7.807336683417086e-05,
-      "loss": 5.8569,
-      "step": 22325
-    },
-    {
-      "epoch": 11.64328552803129,
-      "grad_norm": 1.4094789028167725,
-      "learning_rate": 7.807236180904523e-05,
-      "loss": 5.3865,
-      "step": 22326
-    },
-    {
-      "epoch": 11.64380704041721,
-      "grad_norm": 1.4380170106887817,
-      "learning_rate": 7.80713567839196e-05,
-      "loss": 5.6407,
-      "step": 22327
-    },
-    {
-      "epoch": 11.64432855280313,
-      "grad_norm": 1.42546546459198,
-      "learning_rate": 7.807035175879397e-05,
-      "loss": 5.1998,
-      "step": 22328
-    },
-    {
-      "epoch": 11.644850065189049,
-      "grad_norm": 1.4278517961502075,
-      "learning_rate": 7.806934673366835e-05,
-      "loss": 5.5158,
-      "step": 22329
-    },
-    {
-      "epoch": 11.645371577574968,
-      "grad_norm": 1.4560823440551758,
-      "learning_rate": 7.806834170854271e-05,
-      "loss": 5.1295,
-      "step": 22330
-    },
-    {
-      "epoch": 11.645893089960886,
-      "grad_norm": 1.6115690469741821,
-      "learning_rate": 7.806733668341709e-05,
-      "loss": 5.4047,
-      "step": 22331
-    },
-    {
-      "epoch": 11.646414602346805,
-      "grad_norm": 1.4294345378875732,
-      "learning_rate": 7.806633165829145e-05,
-      "loss": 5.6737,
-      "step": 22332
-    },
-    {
-      "epoch": 11.646936114732725,
-      "grad_norm": 1.3632735013961792,
-      "learning_rate": 7.806532663316583e-05,
-      "loss": 5.5422,
-      "step": 22333
-    },
-    {
-      "epoch": 11.647457627118644,
-      "grad_norm": 1.4169162511825562,
-      "learning_rate": 7.80643216080402e-05,
-      "loss": 5.4302,
-      "step": 22334
-    },
-    {
-      "epoch": 11.647979139504564,
-      "grad_norm": 1.417777419090271,
-      "learning_rate": 7.806331658291457e-05,
-      "loss": 5.5383,
-      "step": 22335
-    },
-    {
-      "epoch": 11.648500651890483,
-      "grad_norm": 1.4470220804214478,
-      "learning_rate": 7.806231155778895e-05,
-      "loss": 5.4729,
-      "step": 22336
-    },
-    {
-      "epoch": 11.649022164276401,
-      "grad_norm": 1.533598780632019,
-      "learning_rate": 7.806130653266333e-05,
-      "loss": 5.2597,
-      "step": 22337
-    },
-    {
-      "epoch": 11.64954367666232,
-      "grad_norm": 1.4031347036361694,
-      "learning_rate": 7.80603015075377e-05,
-      "loss": 5.3718,
-      "step": 22338
-    },
-    {
-      "epoch": 11.65006518904824,
-      "grad_norm": 1.5345033407211304,
-      "learning_rate": 7.805929648241206e-05,
-      "loss": 5.5711,
-      "step": 22339
-    },
-    {
-      "epoch": 11.65058670143416,
-      "grad_norm": 1.347015142440796,
-      "learning_rate": 7.805829145728644e-05,
-      "loss": 5.3265,
-      "step": 22340
-    },
-    {
-      "epoch": 11.651108213820079,
-      "grad_norm": 1.3459367752075195,
-      "learning_rate": 7.80572864321608e-05,
-      "loss": 5.0758,
-      "step": 22341
-    },
-    {
-      "epoch": 11.651629726205996,
-      "grad_norm": 1.4291714429855347,
-      "learning_rate": 7.805628140703518e-05,
-      "loss": 5.8195,
-      "step": 22342
-    },
-    {
-      "epoch": 11.652151238591916,
-      "grad_norm": 1.3891022205352783,
-      "learning_rate": 7.805527638190954e-05,
-      "loss": 5.5874,
-      "step": 22343
-    },
-    {
-      "epoch": 11.652672750977835,
-      "grad_norm": 1.4743692874908447,
-      "learning_rate": 7.805427135678392e-05,
-      "loss": 5.3898,
-      "step": 22344
-    },
-    {
-      "epoch": 11.653194263363755,
-      "grad_norm": 1.4610532522201538,
-      "learning_rate": 7.80532663316583e-05,
-      "loss": 5.4898,
-      "step": 22345
-    },
-    {
-      "epoch": 11.653715775749674,
-      "grad_norm": 1.4509382247924805,
-      "learning_rate": 7.805226130653268e-05,
-      "loss": 5.6498,
-      "step": 22346
-    },
-    {
-      "epoch": 11.654237288135594,
-      "grad_norm": 1.513219952583313,
-      "learning_rate": 7.805125628140704e-05,
-      "loss": 5.202,
-      "step": 22347
-    },
-    {
-      "epoch": 11.654758800521513,
-      "grad_norm": 1.4862080812454224,
-      "learning_rate": 7.805025125628142e-05,
-      "loss": 5.6566,
-      "step": 22348
-    },
-    {
-      "epoch": 11.655280312907431,
-      "grad_norm": 1.4414032697677612,
-      "learning_rate": 7.804924623115578e-05,
-      "loss": 5.3113,
-      "step": 22349
-    },
-    {
-      "epoch": 11.65580182529335,
-      "grad_norm": 1.4618940353393555,
-      "learning_rate": 7.804824120603016e-05,
-      "loss": 5.2711,
-      "step": 22350
-    },
-    {
-      "epoch": 11.65632333767927,
-      "grad_norm": 1.3114490509033203,
-      "learning_rate": 7.804723618090453e-05,
-      "loss": 5.1517,
-      "step": 22351
-    },
-    {
-      "epoch": 11.65684485006519,
-      "grad_norm": 1.6304571628570557,
-      "learning_rate": 7.804623115577889e-05,
-      "loss": 5.1509,
-      "step": 22352
-    },
-    {
-      "epoch": 11.657366362451109,
-      "grad_norm": 1.4391857385635376,
-      "learning_rate": 7.804522613065327e-05,
-      "loss": 5.8491,
-      "step": 22353
-    },
-    {
-      "epoch": 11.657887874837026,
-      "grad_norm": 1.4381139278411865,
-      "learning_rate": 7.804422110552763e-05,
-      "loss": 5.0546,
-      "step": 22354
-    },
-    {
-      "epoch": 11.658409387222946,
-      "grad_norm": 1.4368985891342163,
-      "learning_rate": 7.804321608040201e-05,
-      "loss": 5.3484,
-      "step": 22355
-    },
-    {
-      "epoch": 11.658930899608865,
-      "grad_norm": 1.496816873550415,
-      "learning_rate": 7.804221105527639e-05,
-      "loss": 5.3823,
-      "step": 22356
-    },
-    {
-      "epoch": 11.659452411994785,
-      "grad_norm": 1.509801983833313,
-      "learning_rate": 7.804120603015077e-05,
-      "loss": 5.4997,
-      "step": 22357
-    },
-    {
-      "epoch": 11.659973924380704,
-      "grad_norm": 1.3923392295837402,
-      "learning_rate": 7.804020100502513e-05,
-      "loss": 5.3031,
-      "step": 22358
-    },
-    {
-      "epoch": 11.660495436766624,
-      "grad_norm": 1.7121953964233398,
-      "learning_rate": 7.803919597989951e-05,
-      "loss": 4.3741,
-      "step": 22359
-    },
-    {
-      "epoch": 11.661016949152543,
-      "grad_norm": 1.5564768314361572,
-      "learning_rate": 7.803819095477387e-05,
-      "loss": 5.1836,
-      "step": 22360
-    },
-    {
-      "epoch": 11.661538461538461,
-      "grad_norm": 1.4686307907104492,
-      "learning_rate": 7.803718592964825e-05,
-      "loss": 5.4277,
-      "step": 22361
-    },
-    {
-      "epoch": 11.66205997392438,
-      "grad_norm": 1.5234954357147217,
-      "learning_rate": 7.803618090452261e-05,
-      "loss": 5.5649,
-      "step": 22362
-    },
-    {
-      "epoch": 11.6625814863103,
-      "grad_norm": 1.5818192958831787,
-      "learning_rate": 7.803517587939699e-05,
-      "loss": 5.4999,
-      "step": 22363
-    },
-    {
-      "epoch": 11.66310299869622,
-      "grad_norm": 1.4053621292114258,
-      "learning_rate": 7.803417085427136e-05,
-      "loss": 5.4529,
-      "step": 22364
-    },
-    {
-      "epoch": 11.663624511082139,
-      "grad_norm": 1.5030752420425415,
-      "learning_rate": 7.803316582914573e-05,
-      "loss": 5.6779,
-      "step": 22365
-    },
-    {
-      "epoch": 11.664146023468057,
-      "grad_norm": 1.4251352548599243,
-      "learning_rate": 7.803216080402011e-05,
-      "loss": 5.725,
-      "step": 22366
-    },
-    {
-      "epoch": 11.664667535853976,
-      "grad_norm": 1.7698302268981934,
-      "learning_rate": 7.803115577889448e-05,
-      "loss": 4.8946,
-      "step": 22367
-    },
-    {
-      "epoch": 11.665189048239895,
-      "grad_norm": 1.459350347518921,
-      "learning_rate": 7.803015075376885e-05,
-      "loss": 5.1424,
-      "step": 22368
-    },
-    {
-      "epoch": 11.665710560625815,
-      "grad_norm": 1.470205307006836,
-      "learning_rate": 7.802914572864322e-05,
-      "loss": 5.5461,
-      "step": 22369
-    },
-    {
-      "epoch": 11.666232073011734,
-      "grad_norm": 1.5984972715377808,
-      "learning_rate": 7.80281407035176e-05,
-      "loss": 5.2973,
-      "step": 22370
-    },
-    {
-      "epoch": 11.666753585397654,
-      "grad_norm": 1.4555083513259888,
-      "learning_rate": 7.802713567839196e-05,
-      "loss": 5.4886,
-      "step": 22371
-    },
-    {
-      "epoch": 11.667275097783573,
-      "grad_norm": 1.4290682077407837,
-      "learning_rate": 7.802613065326634e-05,
-      "loss": 5.6123,
-      "step": 22372
-    },
-    {
-      "epoch": 11.667796610169491,
-      "grad_norm": 1.393673300743103,
-      "learning_rate": 7.80251256281407e-05,
-      "loss": 5.6438,
-      "step": 22373
-    },
-    {
-      "epoch": 11.66831812255541,
-      "grad_norm": 1.4559621810913086,
-      "learning_rate": 7.802412060301508e-05,
-      "loss": 5.6223,
-      "step": 22374
-    },
-    {
-      "epoch": 11.66883963494133,
-      "grad_norm": 1.534934163093567,
-      "learning_rate": 7.802311557788944e-05,
-      "loss": 5.2654,
-      "step": 22375
-    },
-    {
-      "epoch": 11.66936114732725,
-      "grad_norm": 1.7711553573608398,
-      "learning_rate": 7.802211055276382e-05,
-      "loss": 4.6471,
-      "step": 22376
-    },
-    {
-      "epoch": 11.669882659713169,
-      "grad_norm": 1.4310418367385864,
-      "learning_rate": 7.80211055276382e-05,
-      "loss": 5.6589,
-      "step": 22377
-    },
-    {
-      "epoch": 11.670404172099087,
-      "grad_norm": 1.3637804985046387,
-      "learning_rate": 7.802010050251258e-05,
-      "loss": 5.4824,
-      "step": 22378
-    },
-    {
-      "epoch": 11.670925684485006,
-      "grad_norm": 1.4700206518173218,
-      "learning_rate": 7.801909547738694e-05,
-      "loss": 5.1121,
-      "step": 22379
-    },
-    {
-      "epoch": 11.671447196870925,
-      "grad_norm": 1.4324235916137695,
-      "learning_rate": 7.80180904522613e-05,
-      "loss": 5.3511,
-      "step": 22380
-    },
-    {
-      "epoch": 11.671968709256845,
-      "grad_norm": 1.4991589784622192,
-      "learning_rate": 7.801708542713568e-05,
-      "loss": 5.4234,
-      "step": 22381
-    },
-    {
-      "epoch": 11.672490221642764,
-      "grad_norm": 1.4901080131530762,
-      "learning_rate": 7.801608040201005e-05,
-      "loss": 5.325,
-      "step": 22382
-    },
-    {
-      "epoch": 11.673011734028684,
-      "grad_norm": 1.3198072910308838,
-      "learning_rate": 7.801507537688443e-05,
-      "loss": 5.919,
-      "step": 22383
-    },
-    {
-      "epoch": 11.673533246414602,
-      "grad_norm": 1.3627763986587524,
-      "learning_rate": 7.801407035175879e-05,
-      "loss": 5.8345,
-      "step": 22384
-    },
-    {
-      "epoch": 11.674054758800521,
-      "grad_norm": 1.4137372970581055,
-      "learning_rate": 7.801306532663317e-05,
-      "loss": 5.8474,
-      "step": 22385
-    },
-    {
-      "epoch": 11.67457627118644,
-      "grad_norm": 1.5607151985168457,
-      "learning_rate": 7.801206030150753e-05,
-      "loss": 4.9886,
-      "step": 22386
-    },
-    {
-      "epoch": 11.67509778357236,
-      "grad_norm": 1.4341670274734497,
-      "learning_rate": 7.801105527638191e-05,
-      "loss": 5.6952,
-      "step": 22387
-    },
-    {
-      "epoch": 11.67561929595828,
-      "grad_norm": 1.473534107208252,
-      "learning_rate": 7.801005025125629e-05,
-      "loss": 4.9042,
-      "step": 22388
-    },
-    {
-      "epoch": 11.676140808344199,
-      "grad_norm": 1.366288185119629,
-      "learning_rate": 7.800904522613067e-05,
-      "loss": 5.4932,
-      "step": 22389
-    },
-    {
-      "epoch": 11.676662320730117,
-      "grad_norm": 1.3821322917938232,
-      "learning_rate": 7.800804020100503e-05,
-      "loss": 5.7056,
-      "step": 22390
-    },
-    {
-      "epoch": 11.677183833116036,
-      "grad_norm": 1.4480736255645752,
-      "learning_rate": 7.800703517587941e-05,
-      "loss": 5.5137,
-      "step": 22391
-    },
-    {
-      "epoch": 11.677705345501955,
-      "grad_norm": 1.4653959274291992,
-      "learning_rate": 7.800603015075377e-05,
-      "loss": 5.5651,
-      "step": 22392
-    },
-    {
-      "epoch": 11.678226857887875,
-      "grad_norm": 1.4862428903579712,
-      "learning_rate": 7.800502512562814e-05,
-      "loss": 5.5951,
-      "step": 22393
-    },
-    {
-      "epoch": 11.678748370273794,
-      "grad_norm": 1.5052144527435303,
-      "learning_rate": 7.800402010050251e-05,
-      "loss": 5.3508,
-      "step": 22394
-    },
-    {
-      "epoch": 11.679269882659714,
-      "grad_norm": 1.4957154989242554,
-      "learning_rate": 7.800301507537688e-05,
-      "loss": 5.3551,
-      "step": 22395
-    },
-    {
-      "epoch": 11.679791395045632,
-      "grad_norm": 1.5022441148757935,
-      "learning_rate": 7.800201005025126e-05,
-      "loss": 5.2716,
-      "step": 22396
-    },
-    {
-      "epoch": 11.680312907431551,
-      "grad_norm": 1.4118150472640991,
-      "learning_rate": 7.800100502512563e-05,
-      "loss": 5.4522,
-      "step": 22397
-    },
-    {
-      "epoch": 11.68083441981747,
-      "grad_norm": 1.3767317533493042,
-      "learning_rate": 7.800000000000001e-05,
-      "loss": 5.5111,
-      "step": 22398
-    },
-    {
-      "epoch": 11.68135593220339,
-      "grad_norm": 1.3343117237091064,
-      "learning_rate": 7.799899497487438e-05,
-      "loss": 5.8141,
-      "step": 22399
-    },
-    {
-      "epoch": 11.68187744458931,
-      "grad_norm": 1.414777159690857,
-      "learning_rate": 7.799798994974875e-05,
-      "loss": 5.8817,
-      "step": 22400
-    },
-    {
-      "epoch": 11.682398956975229,
-      "grad_norm": 1.3941991329193115,
-      "learning_rate": 7.799698492462312e-05,
-      "loss": 5.7587,
-      "step": 22401
-    },
-    {
-      "epoch": 11.682920469361147,
-      "grad_norm": 1.441398024559021,
-      "learning_rate": 7.79959798994975e-05,
-      "loss": 5.5157,
-      "step": 22402
-    },
-    {
-      "epoch": 11.683441981747066,
-      "grad_norm": 1.407328486442566,
-      "learning_rate": 7.799497487437186e-05,
-      "loss": 5.4552,
-      "step": 22403
-    },
-    {
-      "epoch": 11.683963494132986,
-      "grad_norm": 1.4890426397323608,
-      "learning_rate": 7.799396984924624e-05,
-      "loss": 5.0828,
-      "step": 22404
-    },
-    {
-      "epoch": 11.684485006518905,
-      "grad_norm": 1.5332647562026978,
-      "learning_rate": 7.79929648241206e-05,
-      "loss": 4.6572,
-      "step": 22405
-    },
-    {
-      "epoch": 11.685006518904824,
-      "grad_norm": 1.6417287588119507,
-      "learning_rate": 7.799195979899497e-05,
-      "loss": 5.4807,
-      "step": 22406
-    },
-    {
-      "epoch": 11.685528031290744,
-      "grad_norm": 1.4415868520736694,
-      "learning_rate": 7.799095477386934e-05,
-      "loss": 5.3368,
-      "step": 22407
-    },
-    {
-      "epoch": 11.686049543676662,
-      "grad_norm": 1.5408474206924438,
-      "learning_rate": 7.798994974874372e-05,
-      "loss": 5.4405,
-      "step": 22408
-    },
-    {
-      "epoch": 11.686571056062581,
-      "grad_norm": 1.4350969791412354,
-      "learning_rate": 7.79889447236181e-05,
-      "loss": 5.4645,
-      "step": 22409
-    },
-    {
-      "epoch": 11.6870925684485,
-      "grad_norm": 1.370524525642395,
-      "learning_rate": 7.798793969849246e-05,
-      "loss": 5.702,
-      "step": 22410
-    },
-    {
-      "epoch": 11.68761408083442,
-      "grad_norm": 1.488935947418213,
-      "learning_rate": 7.798693467336684e-05,
-      "loss": 5.6628,
-      "step": 22411
-    },
-    {
-      "epoch": 11.68813559322034,
-      "grad_norm": 1.4086413383483887,
-      "learning_rate": 7.798592964824121e-05,
-      "loss": 5.748,
-      "step": 22412
-    },
-    {
-      "epoch": 11.688657105606259,
-      "grad_norm": 1.4073898792266846,
-      "learning_rate": 7.798492462311558e-05,
-      "loss": 5.3414,
-      "step": 22413
-    },
-    {
-      "epoch": 11.689178617992177,
-      "grad_norm": 1.4010579586029053,
-      "learning_rate": 7.798391959798995e-05,
-      "loss": 5.0693,
-      "step": 22414
-    },
-    {
-      "epoch": 11.689700130378096,
-      "grad_norm": 1.4029541015625,
-      "learning_rate": 7.798291457286433e-05,
-      "loss": 5.6084,
-      "step": 22415
-    },
-    {
-      "epoch": 11.690221642764016,
-      "grad_norm": 1.422835350036621,
-      "learning_rate": 7.798190954773869e-05,
-      "loss": 5.6635,
-      "step": 22416
-    },
-    {
-      "epoch": 11.690743155149935,
-      "grad_norm": 1.427840232849121,
-      "learning_rate": 7.798090452261307e-05,
-      "loss": 5.663,
-      "step": 22417
-    },
-    {
-      "epoch": 11.691264667535854,
-      "grad_norm": 1.4030089378356934,
-      "learning_rate": 7.797989949748745e-05,
-      "loss": 5.5428,
-      "step": 22418
-    },
-    {
-      "epoch": 11.691786179921774,
-      "grad_norm": 1.4339790344238281,
-      "learning_rate": 7.797889447236181e-05,
-      "loss": 5.1488,
-      "step": 22419
-    },
-    {
-      "epoch": 11.692307692307692,
-      "grad_norm": 1.3914114236831665,
-      "learning_rate": 7.797788944723619e-05,
-      "loss": 5.135,
-      "step": 22420
-    },
-    {
-      "epoch": 11.692829204693611,
-      "grad_norm": 1.3858121633529663,
-      "learning_rate": 7.797688442211055e-05,
-      "loss": 5.6975,
-      "step": 22421
-    },
-    {
-      "epoch": 11.69335071707953,
-      "grad_norm": 1.4575973749160767,
-      "learning_rate": 7.797587939698493e-05,
-      "loss": 5.5233,
-      "step": 22422
-    },
-    {
-      "epoch": 11.69387222946545,
-      "grad_norm": 1.5779324769973755,
-      "learning_rate": 7.79748743718593e-05,
-      "loss": 5.3345,
-      "step": 22423
-    },
-    {
-      "epoch": 11.69439374185137,
-      "grad_norm": 1.538330316543579,
-      "learning_rate": 7.797386934673367e-05,
-      "loss": 4.8423,
-      "step": 22424
-    },
-    {
-      "epoch": 11.694915254237289,
-      "grad_norm": 1.517224907875061,
-      "learning_rate": 7.797286432160804e-05,
-      "loss": 4.5682,
-      "step": 22425
-    },
-    {
-      "epoch": 11.695436766623207,
-      "grad_norm": 1.5725706815719604,
-      "learning_rate": 7.797185929648242e-05,
-      "loss": 5.2264,
-      "step": 22426
-    },
-    {
-      "epoch": 11.695958279009126,
-      "grad_norm": 1.4596368074417114,
-      "learning_rate": 7.797085427135678e-05,
-      "loss": 5.5034,
-      "step": 22427
-    },
-    {
-      "epoch": 11.696479791395046,
-      "grad_norm": 1.4667460918426514,
-      "learning_rate": 7.796984924623116e-05,
-      "loss": 5.4027,
-      "step": 22428
-    },
-    {
-      "epoch": 11.697001303780965,
-      "grad_norm": 1.4085818529129028,
-      "learning_rate": 7.796884422110554e-05,
-      "loss": 5.4322,
-      "step": 22429
-    },
-    {
-      "epoch": 11.697522816166884,
-      "grad_norm": 1.4669299125671387,
-      "learning_rate": 7.796783919597991e-05,
-      "loss": 5.6352,
-      "step": 22430
-    },
-    {
-      "epoch": 11.698044328552804,
-      "grad_norm": 1.3443292379379272,
-      "learning_rate": 7.796683417085428e-05,
-      "loss": 5.7931,
-      "step": 22431
-    },
-    {
-      "epoch": 11.698565840938722,
-      "grad_norm": 1.5046756267547607,
-      "learning_rate": 7.796582914572864e-05,
-      "loss": 5.2159,
-      "step": 22432
-    },
-    {
-      "epoch": 11.699087353324641,
-      "grad_norm": 1.4159889221191406,
-      "learning_rate": 7.796482412060302e-05,
-      "loss": 5.6469,
-      "step": 22433
-    },
-    {
-      "epoch": 11.69960886571056,
-      "grad_norm": 1.429555058479309,
-      "learning_rate": 7.796381909547738e-05,
-      "loss": 5.4423,
-      "step": 22434
-    },
-    {
-      "epoch": 11.70013037809648,
-      "grad_norm": 1.4651635885238647,
-      "learning_rate": 7.796281407035176e-05,
-      "loss": 5.4051,
-      "step": 22435
-    },
-    {
-      "epoch": 11.7006518904824,
-      "grad_norm": 1.6165627241134644,
-      "learning_rate": 7.796180904522613e-05,
-      "loss": 5.1825,
-      "step": 22436
-    },
-    {
-      "epoch": 11.701173402868317,
-      "grad_norm": 1.4805530309677124,
-      "learning_rate": 7.79608040201005e-05,
-      "loss": 5.2124,
-      "step": 22437
-    },
-    {
-      "epoch": 11.701694915254237,
-      "grad_norm": 1.4260494709014893,
-      "learning_rate": 7.795979899497488e-05,
-      "loss": 5.8221,
-      "step": 22438
-    },
-    {
-      "epoch": 11.702216427640156,
-      "grad_norm": 1.4330998659133911,
-      "learning_rate": 7.795879396984926e-05,
-      "loss": 5.2056,
-      "step": 22439
-    },
-    {
-      "epoch": 11.702737940026076,
-      "grad_norm": 1.409659743309021,
-      "learning_rate": 7.795778894472362e-05,
-      "loss": 5.5548,
-      "step": 22440
-    },
-    {
-      "epoch": 11.703259452411995,
-      "grad_norm": 1.48440420627594,
-      "learning_rate": 7.7956783919598e-05,
-      "loss": 5.4245,
-      "step": 22441
-    },
-    {
-      "epoch": 11.703780964797915,
-      "grad_norm": 1.4510475397109985,
-      "learning_rate": 7.795577889447237e-05,
-      "loss": 5.4432,
-      "step": 22442
-    },
-    {
-      "epoch": 11.704302477183834,
-      "grad_norm": 1.7049930095672607,
-      "learning_rate": 7.795477386934674e-05,
-      "loss": 5.0972,
-      "step": 22443
-    },
-    {
-      "epoch": 11.704823989569752,
-      "grad_norm": 1.4148764610290527,
-      "learning_rate": 7.795376884422111e-05,
-      "loss": 5.2654,
-      "step": 22444
-    },
-    {
-      "epoch": 11.705345501955671,
-      "grad_norm": 1.4843679666519165,
-      "learning_rate": 7.795276381909547e-05,
-      "loss": 5.0513,
-      "step": 22445
-    },
-    {
-      "epoch": 11.70586701434159,
-      "grad_norm": 1.4580845832824707,
-      "learning_rate": 7.795175879396985e-05,
-      "loss": 5.5574,
-      "step": 22446
-    },
-    {
-      "epoch": 11.70638852672751,
-      "grad_norm": 1.5612263679504395,
-      "learning_rate": 7.795075376884421e-05,
-      "loss": 5.5625,
-      "step": 22447
-    },
-    {
-      "epoch": 11.70691003911343,
-      "grad_norm": 1.5042870044708252,
-      "learning_rate": 7.794974874371859e-05,
-      "loss": 5.3421,
-      "step": 22448
-    },
-    {
-      "epoch": 11.707431551499347,
-      "grad_norm": 1.4396135807037354,
-      "learning_rate": 7.794874371859297e-05,
-      "loss": 5.4092,
-      "step": 22449
-    },
-    {
-      "epoch": 11.707953063885267,
-      "grad_norm": 1.4393309354782104,
-      "learning_rate": 7.794773869346735e-05,
-      "loss": 5.6972,
-      "step": 22450
-    },
-    {
-      "epoch": 11.708474576271186,
-      "grad_norm": 1.6035959720611572,
-      "learning_rate": 7.794673366834171e-05,
-      "loss": 5.4299,
-      "step": 22451
-    },
-    {
-      "epoch": 11.708996088657106,
-      "grad_norm": 1.3688639402389526,
-      "learning_rate": 7.794572864321609e-05,
-      "loss": 5.4815,
-      "step": 22452
-    },
-    {
-      "epoch": 11.709517601043025,
-      "grad_norm": 1.412873387336731,
-      "learning_rate": 7.794472361809045e-05,
-      "loss": 4.9337,
-      "step": 22453
-    },
-    {
-      "epoch": 11.710039113428945,
-      "grad_norm": 1.4145488739013672,
-      "learning_rate": 7.794371859296483e-05,
-      "loss": 5.3946,
-      "step": 22454
-    },
-    {
-      "epoch": 11.710560625814864,
-      "grad_norm": 1.4828784465789795,
-      "learning_rate": 7.79427135678392e-05,
-      "loss": 5.8173,
-      "step": 22455
-    },
-    {
-      "epoch": 11.711082138200782,
-      "grad_norm": 1.4704004526138306,
-      "learning_rate": 7.794170854271357e-05,
-      "loss": 5.125,
-      "step": 22456
-    },
-    {
-      "epoch": 11.711603650586701,
-      "grad_norm": 1.3961068391799927,
-      "learning_rate": 7.794070351758794e-05,
-      "loss": 5.5263,
-      "step": 22457
-    },
-    {
-      "epoch": 11.71212516297262,
-      "grad_norm": 1.439123511314392,
-      "learning_rate": 7.793969849246232e-05,
-      "loss": 5.4156,
-      "step": 22458
-    },
-    {
-      "epoch": 11.71264667535854,
-      "grad_norm": 1.4569675922393799,
-      "learning_rate": 7.79386934673367e-05,
-      "loss": 5.3458,
-      "step": 22459
-    },
-    {
-      "epoch": 11.71316818774446,
-      "grad_norm": 1.437272071838379,
-      "learning_rate": 7.793768844221106e-05,
-      "loss": 5.4905,
-      "step": 22460
-    },
-    {
-      "epoch": 11.713689700130377,
-      "grad_norm": 1.395545482635498,
-      "learning_rate": 7.793668341708544e-05,
-      "loss": 5.5455,
-      "step": 22461
-    },
-    {
-      "epoch": 11.714211212516297,
-      "grad_norm": 1.509758472442627,
-      "learning_rate": 7.79356783919598e-05,
-      "loss": 5.5208,
-      "step": 22462
-    },
-    {
-      "epoch": 11.714732724902216,
-      "grad_norm": 1.578309178352356,
-      "learning_rate": 7.793467336683418e-05,
-      "loss": 4.9136,
-      "step": 22463
-    },
-    {
-      "epoch": 11.715254237288136,
-      "grad_norm": 1.5560270547866821,
-      "learning_rate": 7.793366834170854e-05,
-      "loss": 5.5725,
-      "step": 22464
-    },
-    {
-      "epoch": 11.715775749674055,
-      "grad_norm": 1.4218329191207886,
-      "learning_rate": 7.793266331658292e-05,
-      "loss": 5.3323,
-      "step": 22465
-    },
-    {
-      "epoch": 11.716297262059975,
-      "grad_norm": 1.5480468273162842,
-      "learning_rate": 7.793165829145728e-05,
-      "loss": 5.2148,
-      "step": 22466
-    },
-    {
-      "epoch": 11.716818774445892,
-      "grad_norm": 1.4353183507919312,
-      "learning_rate": 7.793065326633166e-05,
-      "loss": 5.6021,
-      "step": 22467
-    },
-    {
-      "epoch": 11.717340286831812,
-      "grad_norm": 1.5578943490982056,
-      "learning_rate": 7.792964824120603e-05,
-      "loss": 5.0772,
-      "step": 22468
-    },
-    {
-      "epoch": 11.717861799217731,
-      "grad_norm": 1.5849608182907104,
-      "learning_rate": 7.79286432160804e-05,
-      "loss": 5.0252,
-      "step": 22469
-    },
-    {
-      "epoch": 11.71838331160365,
-      "grad_norm": 1.38527512550354,
-      "learning_rate": 7.792763819095478e-05,
-      "loss": 5.6332,
-      "step": 22470
-    },
-    {
-      "epoch": 11.71890482398957,
-      "grad_norm": 1.4296066761016846,
-      "learning_rate": 7.792663316582916e-05,
-      "loss": 5.4885,
-      "step": 22471
-    },
-    {
-      "epoch": 11.71942633637549,
-      "grad_norm": 1.3703811168670654,
-      "learning_rate": 7.792562814070352e-05,
-      "loss": 5.7659,
-      "step": 22472
-    },
-    {
-      "epoch": 11.719947848761407,
-      "grad_norm": 1.4113913774490356,
-      "learning_rate": 7.792462311557789e-05,
-      "loss": 4.8689,
-      "step": 22473
-    },
-    {
-      "epoch": 11.720469361147327,
-      "grad_norm": 1.4096713066101074,
-      "learning_rate": 7.792361809045227e-05,
-      "loss": 5.6609,
-      "step": 22474
-    },
-    {
-      "epoch": 11.720990873533246,
-      "grad_norm": 1.4305946826934814,
-      "learning_rate": 7.792261306532663e-05,
-      "loss": 5.5221,
-      "step": 22475
-    },
-    {
-      "epoch": 11.721512385919166,
-      "grad_norm": 1.5416173934936523,
-      "learning_rate": 7.792160804020101e-05,
-      "loss": 4.7127,
-      "step": 22476
-    },
-    {
-      "epoch": 11.722033898305085,
-      "grad_norm": 1.4359829425811768,
-      "learning_rate": 7.792060301507537e-05,
-      "loss": 5.266,
-      "step": 22477
-    },
-    {
-      "epoch": 11.722555410691005,
-      "grad_norm": 1.7345619201660156,
-      "learning_rate": 7.791959798994975e-05,
-      "loss": 5.4537,
-      "step": 22478
-    },
-    {
-      "epoch": 11.723076923076922,
-      "grad_norm": 1.3928526639938354,
-      "learning_rate": 7.791859296482413e-05,
-      "loss": 5.4544,
-      "step": 22479
-    },
-    {
-      "epoch": 11.723598435462842,
-      "grad_norm": 1.4959369897842407,
-      "learning_rate": 7.79175879396985e-05,
-      "loss": 5.3098,
-      "step": 22480
-    },
-    {
-      "epoch": 11.724119947848761,
-      "grad_norm": 1.3802310228347778,
-      "learning_rate": 7.791658291457287e-05,
-      "loss": 5.7844,
-      "step": 22481
-    },
-    {
-      "epoch": 11.72464146023468,
-      "grad_norm": 1.4135864973068237,
-      "learning_rate": 7.791557788944725e-05,
-      "loss": 5.6923,
-      "step": 22482
-    },
-    {
-      "epoch": 11.7251629726206,
-      "grad_norm": 1.400306224822998,
-      "learning_rate": 7.791457286432161e-05,
-      "loss": 5.4089,
-      "step": 22483
-    },
-    {
-      "epoch": 11.72568448500652,
-      "grad_norm": 1.374598503112793,
-      "learning_rate": 7.791356783919599e-05,
-      "loss": 5.6734,
-      "step": 22484
-    },
-    {
-      "epoch": 11.726205997392437,
-      "grad_norm": 1.4260706901550293,
-      "learning_rate": 7.791256281407035e-05,
-      "loss": 5.0811,
-      "step": 22485
-    },
-    {
-      "epoch": 11.726727509778357,
-      "grad_norm": 1.4543240070343018,
-      "learning_rate": 7.791155778894472e-05,
-      "loss": 5.4197,
-      "step": 22486
-    },
-    {
-      "epoch": 11.727249022164276,
-      "grad_norm": 1.6673510074615479,
-      "learning_rate": 7.79105527638191e-05,
-      "loss": 5.1242,
-      "step": 22487
-    },
-    {
-      "epoch": 11.727770534550196,
-      "grad_norm": 1.410360336303711,
-      "learning_rate": 7.790954773869346e-05,
-      "loss": 5.5391,
-      "step": 22488
-    },
-    {
-      "epoch": 11.728292046936115,
-      "grad_norm": 1.3380177021026611,
-      "learning_rate": 7.790854271356784e-05,
-      "loss": 5.8526,
-      "step": 22489
-    },
-    {
-      "epoch": 11.728813559322035,
-      "grad_norm": 1.486576795578003,
-      "learning_rate": 7.790753768844222e-05,
-      "loss": 5.1498,
-      "step": 22490
-    },
-    {
-      "epoch": 11.729335071707952,
-      "grad_norm": 1.4250015020370483,
-      "learning_rate": 7.79065326633166e-05,
-      "loss": 4.8451,
-      "step": 22491
-    },
-    {
-      "epoch": 11.729856584093872,
-      "grad_norm": 1.571189045906067,
-      "learning_rate": 7.790552763819096e-05,
-      "loss": 5.1853,
-      "step": 22492
-    },
-    {
-      "epoch": 11.730378096479791,
-      "grad_norm": 1.517397403717041,
-      "learning_rate": 7.790452261306534e-05,
-      "loss": 5.5334,
-      "step": 22493
-    },
-    {
-      "epoch": 11.73089960886571,
-      "grad_norm": 1.4223803281784058,
-      "learning_rate": 7.79035175879397e-05,
-      "loss": 4.8781,
-      "step": 22494
-    },
-    {
-      "epoch": 11.73142112125163,
-      "grad_norm": 1.4819656610488892,
-      "learning_rate": 7.790251256281408e-05,
-      "loss": 5.3218,
-      "step": 22495
-    },
-    {
-      "epoch": 11.73194263363755,
-      "grad_norm": 1.4287105798721313,
-      "learning_rate": 7.790150753768844e-05,
-      "loss": 5.6135,
-      "step": 22496
-    },
-    {
-      "epoch": 11.732464146023467,
-      "grad_norm": 1.4194462299346924,
-      "learning_rate": 7.790050251256282e-05,
-      "loss": 5.6368,
-      "step": 22497
-    },
-    {
-      "epoch": 11.732985658409387,
-      "grad_norm": 1.3945426940917969,
-      "learning_rate": 7.789949748743719e-05,
-      "loss": 5.797,
-      "step": 22498
-    },
-    {
-      "epoch": 11.733507170795306,
-      "grad_norm": 1.5675188302993774,
-      "learning_rate": 7.789849246231156e-05,
-      "loss": 5.5485,
-      "step": 22499
-    },
-    {
-      "epoch": 11.734028683181226,
-      "grad_norm": 1.505635142326355,
-      "learning_rate": 7.789748743718594e-05,
-      "loss": 5.2848,
-      "step": 22500
-    },
-    {
-      "epoch": 11.734550195567145,
-      "grad_norm": 1.4929746389389038,
-      "learning_rate": 7.78964824120603e-05,
-      "loss": 4.7798,
-      "step": 22501
-    },
-    {
-      "epoch": 11.735071707953065,
-      "grad_norm": 1.5554380416870117,
-      "learning_rate": 7.789547738693468e-05,
-      "loss": 4.8184,
-      "step": 22502
-    },
-    {
-      "epoch": 11.735593220338982,
-      "grad_norm": 1.3248733282089233,
-      "learning_rate": 7.789447236180905e-05,
-      "loss": 5.4825,
-      "step": 22503
-    },
-    {
-      "epoch": 11.736114732724902,
-      "grad_norm": 1.4561240673065186,
-      "learning_rate": 7.789346733668342e-05,
-      "loss": 5.7085,
-      "step": 22504
-    },
-    {
-      "epoch": 11.736636245110821,
-      "grad_norm": 1.432570219039917,
-      "learning_rate": 7.789246231155779e-05,
-      "loss": 5.2966,
-      "step": 22505
-    },
-    {
-      "epoch": 11.73715775749674,
-      "grad_norm": 1.481252908706665,
-      "learning_rate": 7.789145728643217e-05,
-      "loss": 5.3442,
-      "step": 22506
-    },
-    {
-      "epoch": 11.73767926988266,
-      "grad_norm": 1.445008397102356,
-      "learning_rate": 7.789045226130653e-05,
-      "loss": 5.3971,
-      "step": 22507
-    },
-    {
-      "epoch": 11.73820078226858,
-      "grad_norm": 1.5979551076889038,
-      "learning_rate": 7.788944723618091e-05,
-      "loss": 5.3204,
-      "step": 22508
-    },
-    {
-      "epoch": 11.738722294654497,
-      "grad_norm": 1.336997389793396,
-      "learning_rate": 7.788844221105527e-05,
-      "loss": 5.366,
-      "step": 22509
-    },
-    {
-      "epoch": 11.739243807040417,
-      "grad_norm": 1.482154130935669,
-      "learning_rate": 7.788743718592965e-05,
-      "loss": 5.454,
-      "step": 22510
-    },
-    {
-      "epoch": 11.739765319426336,
-      "grad_norm": 1.436630368232727,
-      "learning_rate": 7.788643216080403e-05,
-      "loss": 5.4852,
-      "step": 22511
-    },
-    {
-      "epoch": 11.740286831812256,
-      "grad_norm": 1.4218045473098755,
-      "learning_rate": 7.78854271356784e-05,
-      "loss": 5.3147,
-      "step": 22512
-    },
-    {
-      "epoch": 11.740808344198175,
-      "grad_norm": 1.3961760997772217,
-      "learning_rate": 7.788442211055277e-05,
-      "loss": 5.6236,
-      "step": 22513
-    },
-    {
-      "epoch": 11.741329856584095,
-      "grad_norm": 1.4075239896774292,
-      "learning_rate": 7.788341708542714e-05,
-      "loss": 5.5331,
-      "step": 22514
-    },
-    {
-      "epoch": 11.741851368970012,
-      "grad_norm": 1.399818778038025,
-      "learning_rate": 7.788241206030151e-05,
-      "loss": 5.5106,
-      "step": 22515
-    },
-    {
-      "epoch": 11.742372881355932,
-      "grad_norm": 1.4347901344299316,
-      "learning_rate": 7.788140703517588e-05,
-      "loss": 5.2248,
-      "step": 22516
-    },
-    {
-      "epoch": 11.742894393741851,
-      "grad_norm": 1.54386568069458,
-      "learning_rate": 7.788040201005026e-05,
-      "loss": 4.4802,
-      "step": 22517
-    },
-    {
-      "epoch": 11.74341590612777,
-      "grad_norm": 1.548140048980713,
-      "learning_rate": 7.787939698492462e-05,
-      "loss": 5.574,
-      "step": 22518
-    },
-    {
-      "epoch": 11.74393741851369,
-      "grad_norm": 1.3945552110671997,
-      "learning_rate": 7.7878391959799e-05,
-      "loss": 5.8887,
-      "step": 22519
-    },
-    {
-      "epoch": 11.74445893089961,
-      "grad_norm": 1.4446918964385986,
-      "learning_rate": 7.787738693467338e-05,
-      "loss": 5.0808,
-      "step": 22520
-    },
-    {
-      "epoch": 11.744980443285527,
-      "grad_norm": 1.5292549133300781,
-      "learning_rate": 7.787638190954775e-05,
-      "loss": 5.4542,
-      "step": 22521
-    },
-    {
-      "epoch": 11.745501955671447,
-      "grad_norm": 1.4869401454925537,
-      "learning_rate": 7.787537688442212e-05,
-      "loss": 5.051,
-      "step": 22522
-    },
-    {
-      "epoch": 11.746023468057366,
-      "grad_norm": 1.4456251859664917,
-      "learning_rate": 7.78743718592965e-05,
-      "loss": 5.7349,
-      "step": 22523
-    },
-    {
-      "epoch": 11.746544980443286,
-      "grad_norm": 1.3408106565475464,
-      "learning_rate": 7.787336683417086e-05,
-      "loss": 5.5598,
-      "step": 22524
-    },
-    {
-      "epoch": 11.747066492829205,
-      "grad_norm": 1.4064748287200928,
-      "learning_rate": 7.787236180904522e-05,
-      "loss": 5.7426,
-      "step": 22525
-    },
-    {
-      "epoch": 11.747588005215125,
-      "grad_norm": 1.6299433708190918,
-      "learning_rate": 7.78713567839196e-05,
-      "loss": 4.4662,
-      "step": 22526
-    },
-    {
-      "epoch": 11.748109517601042,
-      "grad_norm": 1.538394570350647,
-      "learning_rate": 7.787035175879397e-05,
-      "loss": 5.3228,
-      "step": 22527
-    },
-    {
-      "epoch": 11.748631029986962,
-      "grad_norm": 1.4869658946990967,
-      "learning_rate": 7.786934673366834e-05,
-      "loss": 5.3851,
-      "step": 22528
-    },
-    {
-      "epoch": 11.749152542372881,
-      "grad_norm": 1.3739008903503418,
-      "learning_rate": 7.786834170854271e-05,
-      "loss": 5.6402,
-      "step": 22529
-    },
-    {
-      "epoch": 11.7496740547588,
-      "grad_norm": 1.641646146774292,
-      "learning_rate": 7.786733668341709e-05,
-      "loss": 5.4642,
-      "step": 22530
-    },
-    {
-      "epoch": 11.75019556714472,
-      "grad_norm": 1.5026894807815552,
-      "learning_rate": 7.786633165829146e-05,
-      "loss": 5.6192,
-      "step": 22531
-    },
-    {
-      "epoch": 11.750717079530638,
-      "grad_norm": 1.4282575845718384,
-      "learning_rate": 7.786532663316584e-05,
-      "loss": 5.3258,
-      "step": 22532
-    },
-    {
-      "epoch": 11.751238591916557,
-      "grad_norm": 1.452181100845337,
-      "learning_rate": 7.78643216080402e-05,
-      "loss": 5.6193,
-      "step": 22533
-    },
-    {
-      "epoch": 11.751760104302477,
-      "grad_norm": 1.4432228803634644,
-      "learning_rate": 7.786331658291458e-05,
-      "loss": 4.9001,
-      "step": 22534
-    },
-    {
-      "epoch": 11.752281616688396,
-      "grad_norm": 1.423448085784912,
-      "learning_rate": 7.786231155778895e-05,
-      "loss": 5.641,
-      "step": 22535
-    },
-    {
-      "epoch": 11.752803129074316,
-      "grad_norm": 1.5182878971099854,
-      "learning_rate": 7.786130653266333e-05,
-      "loss": 5.4369,
-      "step": 22536
-    },
-    {
-      "epoch": 11.753324641460235,
-      "grad_norm": 1.6211521625518799,
-      "learning_rate": 7.786030150753769e-05,
-      "loss": 4.8658,
-      "step": 22537
-    },
-    {
-      "epoch": 11.753846153846155,
-      "grad_norm": 1.378760576248169,
-      "learning_rate": 7.785929648241205e-05,
-      "loss": 5.1783,
-      "step": 22538
-    },
-    {
-      "epoch": 11.754367666232072,
-      "grad_norm": 1.516028881072998,
-      "learning_rate": 7.785829145728643e-05,
-      "loss": 5.518,
-      "step": 22539
-    },
-    {
-      "epoch": 11.754889178617992,
-      "grad_norm": 1.4516773223876953,
-      "learning_rate": 7.78572864321608e-05,
-      "loss": 5.5256,
-      "step": 22540
-    },
-    {
-      "epoch": 11.755410691003911,
-      "grad_norm": 1.433862328529358,
-      "learning_rate": 7.785628140703517e-05,
-      "loss": 5.5394,
-      "step": 22541
-    },
-    {
-      "epoch": 11.75593220338983,
-      "grad_norm": 1.5036695003509521,
-      "learning_rate": 7.785527638190955e-05,
-      "loss": 5.2842,
-      "step": 22542
-    },
-    {
-      "epoch": 11.75645371577575,
-      "grad_norm": 1.5545848608016968,
-      "learning_rate": 7.785427135678393e-05,
-      "loss": 5.2328,
-      "step": 22543
-    },
-    {
-      "epoch": 11.756975228161668,
-      "grad_norm": 1.4566547870635986,
-      "learning_rate": 7.78532663316583e-05,
-      "loss": 4.9184,
-      "step": 22544
-    },
-    {
-      "epoch": 11.757496740547587,
-      "grad_norm": 1.3913334608078003,
-      "learning_rate": 7.785226130653267e-05,
-      "loss": 5.497,
-      "step": 22545
-    },
-    {
-      "epoch": 11.758018252933507,
-      "grad_norm": 1.3711093664169312,
-      "learning_rate": 7.785125628140704e-05,
-      "loss": 5.7265,
-      "step": 22546
-    },
-    {
-      "epoch": 11.758539765319426,
-      "grad_norm": 1.3727680444717407,
-      "learning_rate": 7.785025125628141e-05,
-      "loss": 5.5559,
-      "step": 22547
-    },
-    {
-      "epoch": 11.759061277705346,
-      "grad_norm": 1.389897108078003,
-      "learning_rate": 7.784924623115578e-05,
-      "loss": 5.6088,
-      "step": 22548
-    },
-    {
-      "epoch": 11.759582790091265,
-      "grad_norm": 1.3964351415634155,
-      "learning_rate": 7.784824120603016e-05,
-      "loss": 5.6793,
-      "step": 22549
-    },
-    {
-      "epoch": 11.760104302477185,
-      "grad_norm": 1.5080510377883911,
-      "learning_rate": 7.784723618090452e-05,
-      "loss": 4.9174,
-      "step": 22550
-    },
-    {
-      "epoch": 11.760625814863102,
-      "grad_norm": 1.5277127027511597,
-      "learning_rate": 7.78462311557789e-05,
-      "loss": 5.5011,
-      "step": 22551
-    },
-    {
-      "epoch": 11.761147327249022,
-      "grad_norm": 1.4179033041000366,
-      "learning_rate": 7.784522613065328e-05,
-      "loss": 5.314,
-      "step": 22552
-    },
-    {
-      "epoch": 11.761668839634941,
-      "grad_norm": 1.507889986038208,
-      "learning_rate": 7.784422110552764e-05,
-      "loss": 5.474,
-      "step": 22553
-    },
-    {
-      "epoch": 11.76219035202086,
-      "grad_norm": 1.451195478439331,
-      "learning_rate": 7.784321608040202e-05,
-      "loss": 5.6296,
-      "step": 22554
-    },
-    {
-      "epoch": 11.76271186440678,
-      "grad_norm": 1.5349527597427368,
-      "learning_rate": 7.784221105527638e-05,
-      "loss": 5.1075,
-      "step": 22555
-    },
-    {
-      "epoch": 11.763233376792698,
-      "grad_norm": 1.4114254713058472,
-      "learning_rate": 7.784120603015076e-05,
-      "loss": 5.5489,
-      "step": 22556
-    },
-    {
-      "epoch": 11.763754889178617,
-      "grad_norm": 1.39677894115448,
-      "learning_rate": 7.784020100502512e-05,
-      "loss": 5.3971,
-      "step": 22557
-    },
-    {
-      "epoch": 11.764276401564537,
-      "grad_norm": 1.459970235824585,
-      "learning_rate": 7.78391959798995e-05,
-      "loss": 5.2786,
-      "step": 22558
-    },
-    {
-      "epoch": 11.764797913950456,
-      "grad_norm": 1.4387083053588867,
-      "learning_rate": 7.783819095477387e-05,
-      "loss": 5.6503,
-      "step": 22559
-    },
-    {
-      "epoch": 11.765319426336376,
-      "grad_norm": 1.375139832496643,
-      "learning_rate": 7.783718592964824e-05,
-      "loss": 5.4207,
-      "step": 22560
-    },
-    {
-      "epoch": 11.765840938722295,
-      "grad_norm": 1.4097038507461548,
-      "learning_rate": 7.783618090452261e-05,
-      "loss": 5.5737,
-      "step": 22561
-    },
-    {
-      "epoch": 11.766362451108213,
-      "grad_norm": 1.4475419521331787,
-      "learning_rate": 7.783517587939699e-05,
-      "loss": 5.6283,
-      "step": 22562
-    },
-    {
-      "epoch": 11.766883963494132,
-      "grad_norm": 1.5301154851913452,
-      "learning_rate": 7.783417085427136e-05,
-      "loss": 5.0862,
-      "step": 22563
-    },
-    {
-      "epoch": 11.767405475880052,
-      "grad_norm": 1.3411800861358643,
-      "learning_rate": 7.783316582914574e-05,
-      "loss": 5.8724,
-      "step": 22564
-    },
-    {
-      "epoch": 11.767926988265971,
-      "grad_norm": 1.3830626010894775,
-      "learning_rate": 7.78321608040201e-05,
-      "loss": 5.2935,
-      "step": 22565
-    },
-    {
-      "epoch": 11.76844850065189,
-      "grad_norm": 1.4088339805603027,
-      "learning_rate": 7.783115577889447e-05,
-      "loss": 5.6065,
-      "step": 22566
-    },
-    {
-      "epoch": 11.76897001303781,
-      "grad_norm": 1.435139536857605,
-      "learning_rate": 7.783015075376885e-05,
-      "loss": 5.2776,
-      "step": 22567
-    },
-    {
-      "epoch": 11.769491525423728,
-      "grad_norm": 1.633093237876892,
-      "learning_rate": 7.782914572864321e-05,
-      "loss": 4.761,
-      "step": 22568
-    },
-    {
-      "epoch": 11.770013037809647,
-      "grad_norm": 1.4369118213653564,
-      "learning_rate": 7.782814070351759e-05,
-      "loss": 5.6728,
-      "step": 22569
-    },
-    {
-      "epoch": 11.770534550195567,
-      "grad_norm": 1.614274024963379,
-      "learning_rate": 7.782713567839196e-05,
-      "loss": 5.506,
-      "step": 22570
-    },
-    {
-      "epoch": 11.771056062581486,
-      "grad_norm": 1.4434432983398438,
-      "learning_rate": 7.782613065326633e-05,
-      "loss": 5.3088,
-      "step": 22571
-    },
-    {
-      "epoch": 11.771577574967406,
-      "grad_norm": 1.3761581182479858,
-      "learning_rate": 7.782512562814071e-05,
-      "loss": 5.1766,
-      "step": 22572
-    },
-    {
-      "epoch": 11.772099087353325,
-      "grad_norm": 1.6434475183486938,
-      "learning_rate": 7.782412060301509e-05,
-      "loss": 5.3583,
-      "step": 22573
-    },
-    {
-      "epoch": 11.772620599739243,
-      "grad_norm": 1.4031544923782349,
-      "learning_rate": 7.782311557788945e-05,
-      "loss": 5.3243,
-      "step": 22574
-    },
-    {
-      "epoch": 11.773142112125162,
-      "grad_norm": 1.5299957990646362,
-      "learning_rate": 7.782211055276383e-05,
-      "loss": 5.5318,
-      "step": 22575
-    },
-    {
-      "epoch": 11.773663624511082,
-      "grad_norm": 1.4312094449996948,
-      "learning_rate": 7.78211055276382e-05,
-      "loss": 5.3413,
-      "step": 22576
-    },
-    {
-      "epoch": 11.774185136897001,
-      "grad_norm": 1.6112605333328247,
-      "learning_rate": 7.782010050251257e-05,
-      "loss": 4.8223,
-      "step": 22577
-    },
-    {
-      "epoch": 11.77470664928292,
-      "grad_norm": 1.524142861366272,
-      "learning_rate": 7.781909547738694e-05,
-      "loss": 5.452,
-      "step": 22578
-    },
-    {
-      "epoch": 11.77522816166884,
-      "grad_norm": 1.4954161643981934,
-      "learning_rate": 7.78180904522613e-05,
-      "loss": 5.1696,
-      "step": 22579
-    },
-    {
-      "epoch": 11.775749674054758,
-      "grad_norm": 1.5065370798110962,
-      "learning_rate": 7.781708542713568e-05,
-      "loss": 4.8782,
-      "step": 22580
-    },
-    {
-      "epoch": 11.776271186440677,
-      "grad_norm": 1.3939415216445923,
-      "learning_rate": 7.781608040201004e-05,
-      "loss": 5.6884,
-      "step": 22581
-    },
-    {
-      "epoch": 11.776792698826597,
-      "grad_norm": 1.419543981552124,
-      "learning_rate": 7.781507537688442e-05,
-      "loss": 5.3058,
-      "step": 22582
-    },
-    {
-      "epoch": 11.777314211212516,
-      "grad_norm": 1.5199605226516724,
-      "learning_rate": 7.78140703517588e-05,
-      "loss": 5.443,
-      "step": 22583
-    },
-    {
-      "epoch": 11.777835723598436,
-      "grad_norm": 1.4135801792144775,
-      "learning_rate": 7.781306532663318e-05,
-      "loss": 5.8503,
-      "step": 22584
-    },
-    {
-      "epoch": 11.778357235984355,
-      "grad_norm": 1.3290280103683472,
-      "learning_rate": 7.781206030150754e-05,
-      "loss": 5.666,
-      "step": 22585
-    },
-    {
-      "epoch": 11.778878748370273,
-      "grad_norm": 1.3210972547531128,
-      "learning_rate": 7.781105527638192e-05,
-      "loss": 5.68,
-      "step": 22586
-    },
-    {
-      "epoch": 11.779400260756193,
-      "grad_norm": 1.360033631324768,
-      "learning_rate": 7.781005025125628e-05,
-      "loss": 5.4132,
-      "step": 22587
-    },
-    {
-      "epoch": 11.779921773142112,
-      "grad_norm": 1.360029935836792,
-      "learning_rate": 7.780904522613066e-05,
-      "loss": 5.5163,
-      "step": 22588
-    },
-    {
-      "epoch": 11.780443285528031,
-      "grad_norm": 1.5235960483551025,
-      "learning_rate": 7.780804020100503e-05,
-      "loss": 5.0893,
-      "step": 22589
-    },
-    {
-      "epoch": 11.780964797913951,
-      "grad_norm": 1.5871440172195435,
-      "learning_rate": 7.78070351758794e-05,
-      "loss": 4.9198,
-      "step": 22590
-    },
-    {
-      "epoch": 11.78148631029987,
-      "grad_norm": 1.3662500381469727,
-      "learning_rate": 7.780603015075377e-05,
-      "loss": 5.3048,
-      "step": 22591
-    },
-    {
-      "epoch": 11.782007822685788,
-      "grad_norm": 1.5816771984100342,
-      "learning_rate": 7.780502512562815e-05,
-      "loss": 5.1533,
-      "step": 22592
-    },
-    {
-      "epoch": 11.782529335071708,
-      "grad_norm": 1.4431374073028564,
-      "learning_rate": 7.780402010050252e-05,
-      "loss": 4.9745,
-      "step": 22593
-    },
-    {
-      "epoch": 11.783050847457627,
-      "grad_norm": 1.3320066928863525,
-      "learning_rate": 7.780301507537689e-05,
-      "loss": 5.7833,
-      "step": 22594
-    },
-    {
-      "epoch": 11.783572359843546,
-      "grad_norm": 1.4902459383010864,
-      "learning_rate": 7.780201005025127e-05,
-      "loss": 5.4279,
-      "step": 22595
-    },
-    {
-      "epoch": 11.784093872229466,
-      "grad_norm": 1.379684329032898,
-      "learning_rate": 7.780100502512563e-05,
-      "loss": 5.5875,
-      "step": 22596
-    },
-    {
-      "epoch": 11.784615384615385,
-      "grad_norm": 1.3418806791305542,
-      "learning_rate": 7.780000000000001e-05,
-      "loss": 5.7309,
-      "step": 22597
-    },
-    {
-      "epoch": 11.785136897001303,
-      "grad_norm": 1.5177135467529297,
-      "learning_rate": 7.779899497487437e-05,
-      "loss": 5.6287,
-      "step": 22598
-    },
-    {
-      "epoch": 11.785658409387223,
-      "grad_norm": 1.567966341972351,
-      "learning_rate": 7.779798994974875e-05,
-      "loss": 5.2633,
-      "step": 22599
-    },
-    {
-      "epoch": 11.786179921773142,
-      "grad_norm": 1.4194515943527222,
-      "learning_rate": 7.779698492462311e-05,
-      "loss": 5.2517,
-      "step": 22600
-    },
-    {
-      "epoch": 11.786701434159061,
-      "grad_norm": 1.4837363958358765,
-      "learning_rate": 7.779597989949749e-05,
-      "loss": 5.1693,
-      "step": 22601
-    },
-    {
-      "epoch": 11.787222946544981,
-      "grad_norm": 1.324831485748291,
-      "learning_rate": 7.779497487437186e-05,
-      "loss": 5.2914,
-      "step": 22602
-    },
-    {
-      "epoch": 11.7877444589309,
-      "grad_norm": 1.3949034214019775,
-      "learning_rate": 7.779396984924623e-05,
-      "loss": 5.7782,
-      "step": 22603
-    },
-    {
-      "epoch": 11.788265971316818,
-      "grad_norm": 1.390512228012085,
-      "learning_rate": 7.779296482412061e-05,
-      "loss": 5.5478,
-      "step": 22604
-    },
-    {
-      "epoch": 11.788787483702738,
-      "grad_norm": 1.4182865619659424,
-      "learning_rate": 7.779195979899498e-05,
-      "loss": 5.0027,
-      "step": 22605
-    },
-    {
-      "epoch": 11.789308996088657,
-      "grad_norm": 1.4023958444595337,
-      "learning_rate": 7.779095477386935e-05,
-      "loss": 5.801,
-      "step": 22606
-    },
-    {
-      "epoch": 11.789830508474576,
-      "grad_norm": 1.4441583156585693,
-      "learning_rate": 7.778994974874372e-05,
-      "loss": 5.4366,
-      "step": 22607
-    },
-    {
-      "epoch": 11.790352020860496,
-      "grad_norm": 1.3786672353744507,
-      "learning_rate": 7.77889447236181e-05,
-      "loss": 5.6111,
-      "step": 22608
-    },
-    {
-      "epoch": 11.790873533246415,
-      "grad_norm": 1.4142861366271973,
-      "learning_rate": 7.778793969849246e-05,
-      "loss": 4.8512,
-      "step": 22609
-    },
-    {
-      "epoch": 11.791395045632333,
-      "grad_norm": 1.556159496307373,
-      "learning_rate": 7.778693467336684e-05,
-      "loss": 5.3905,
-      "step": 22610
-    },
-    {
-      "epoch": 11.791916558018253,
-      "grad_norm": 1.321419596672058,
-      "learning_rate": 7.77859296482412e-05,
-      "loss": 5.9656,
-      "step": 22611
-    },
-    {
-      "epoch": 11.792438070404172,
-      "grad_norm": 1.5288840532302856,
-      "learning_rate": 7.778492462311558e-05,
-      "loss": 4.8526,
-      "step": 22612
-    },
-    {
-      "epoch": 11.792959582790091,
-      "grad_norm": 1.531825065612793,
-      "learning_rate": 7.778391959798996e-05,
-      "loss": 5.5563,
-      "step": 22613
-    },
-    {
-      "epoch": 11.793481095176011,
-      "grad_norm": 1.3932687044143677,
-      "learning_rate": 7.778291457286434e-05,
-      "loss": 5.5183,
-      "step": 22614
-    },
-    {
-      "epoch": 11.79400260756193,
-      "grad_norm": 1.4611543416976929,
-      "learning_rate": 7.77819095477387e-05,
-      "loss": 5.4018,
-      "step": 22615
-    },
-    {
-      "epoch": 11.794524119947848,
-      "grad_norm": 1.4656734466552734,
-      "learning_rate": 7.778090452261308e-05,
-      "loss": 5.277,
-      "step": 22616
-    },
-    {
-      "epoch": 11.795045632333768,
-      "grad_norm": 1.4397114515304565,
-      "learning_rate": 7.777989949748744e-05,
-      "loss": 5.412,
-      "step": 22617
-    },
-    {
-      "epoch": 11.795567144719687,
-      "grad_norm": 1.3317029476165771,
-      "learning_rate": 7.77788944723618e-05,
-      "loss": 5.7058,
-      "step": 22618
-    },
-    {
-      "epoch": 11.796088657105607,
-      "grad_norm": 1.417751669883728,
-      "learning_rate": 7.777788944723618e-05,
-      "loss": 4.8532,
-      "step": 22619
-    },
-    {
-      "epoch": 11.796610169491526,
-      "grad_norm": 1.4027276039123535,
-      "learning_rate": 7.777688442211055e-05,
-      "loss": 5.4494,
-      "step": 22620
-    },
-    {
-      "epoch": 11.797131681877445,
-      "grad_norm": 1.506284475326538,
-      "learning_rate": 7.777587939698493e-05,
-      "loss": 5.548,
-      "step": 22621
-    },
-    {
-      "epoch": 11.797653194263363,
-      "grad_norm": 1.559933066368103,
-      "learning_rate": 7.777487437185929e-05,
-      "loss": 5.3562,
-      "step": 22622
-    },
-    {
-      "epoch": 11.798174706649283,
-      "grad_norm": 1.5100486278533936,
-      "learning_rate": 7.777386934673367e-05,
-      "loss": 5.5662,
-      "step": 22623
-    },
-    {
-      "epoch": 11.798696219035202,
-      "grad_norm": 1.4609471559524536,
-      "learning_rate": 7.777286432160805e-05,
-      "loss": 5.5185,
-      "step": 22624
-    },
-    {
-      "epoch": 11.799217731421122,
-      "grad_norm": 1.5066847801208496,
-      "learning_rate": 7.777185929648242e-05,
-      "loss": 5.5483,
-      "step": 22625
-    },
-    {
-      "epoch": 11.799739243807041,
-      "grad_norm": 1.5386645793914795,
-      "learning_rate": 7.777085427135679e-05,
-      "loss": 4.8114,
-      "step": 22626
-    },
-    {
-      "epoch": 11.800260756192959,
-      "grad_norm": 1.5480759143829346,
-      "learning_rate": 7.776984924623117e-05,
-      "loss": 5.1437,
-      "step": 22627
-    },
-    {
-      "epoch": 11.800782268578878,
-      "grad_norm": 1.615178108215332,
-      "learning_rate": 7.776884422110553e-05,
-      "loss": 5.4463,
-      "step": 22628
-    },
-    {
-      "epoch": 11.801303780964798,
-      "grad_norm": 1.5255157947540283,
-      "learning_rate": 7.776783919597991e-05,
-      "loss": 5.6629,
-      "step": 22629
-    },
-    {
-      "epoch": 11.801825293350717,
-      "grad_norm": 1.4355379343032837,
-      "learning_rate": 7.776683417085427e-05,
-      "loss": 5.4075,
-      "step": 22630
-    },
-    {
-      "epoch": 11.802346805736637,
-      "grad_norm": 1.455886721611023,
-      "learning_rate": 7.776582914572864e-05,
-      "loss": 5.5523,
-      "step": 22631
-    },
-    {
-      "epoch": 11.802868318122556,
-      "grad_norm": 1.5625265836715698,
-      "learning_rate": 7.776482412060301e-05,
-      "loss": 5.4104,
-      "step": 22632
-    },
-    {
-      "epoch": 11.803389830508475,
-      "grad_norm": 1.5349183082580566,
-      "learning_rate": 7.776381909547739e-05,
-      "loss": 5.732,
-      "step": 22633
-    },
-    {
-      "epoch": 11.803911342894393,
-      "grad_norm": 1.3888195753097534,
-      "learning_rate": 7.776281407035177e-05,
-      "loss": 5.7543,
-      "step": 22634
-    },
-    {
-      "epoch": 11.804432855280313,
-      "grad_norm": 1.380212664604187,
-      "learning_rate": 7.776180904522613e-05,
-      "loss": 5.7872,
-      "step": 22635
-    },
-    {
-      "epoch": 11.804954367666232,
-      "grad_norm": 1.3873121738433838,
-      "learning_rate": 7.776080402010051e-05,
-      "loss": 5.9743,
-      "step": 22636
-    },
-    {
-      "epoch": 11.805475880052152,
-      "grad_norm": 1.4057403802871704,
-      "learning_rate": 7.775979899497488e-05,
-      "loss": 5.47,
-      "step": 22637
-    },
-    {
-      "epoch": 11.805997392438071,
-      "grad_norm": 1.397611379623413,
-      "learning_rate": 7.775879396984925e-05,
-      "loss": 5.6726,
-      "step": 22638
-    },
-    {
-      "epoch": 11.806518904823989,
-      "grad_norm": 1.4628974199295044,
-      "learning_rate": 7.775778894472362e-05,
-      "loss": 5.7401,
-      "step": 22639
-    },
-    {
-      "epoch": 11.807040417209908,
-      "grad_norm": 1.396308422088623,
-      "learning_rate": 7.7756783919598e-05,
-      "loss": 5.6031,
-      "step": 22640
-    },
-    {
-      "epoch": 11.807561929595828,
-      "grad_norm": 1.4852187633514404,
-      "learning_rate": 7.775577889447236e-05,
-      "loss": 5.618,
-      "step": 22641
-    },
-    {
-      "epoch": 11.808083441981747,
-      "grad_norm": 1.308880090713501,
-      "learning_rate": 7.775477386934674e-05,
-      "loss": 5.3694,
-      "step": 22642
-    },
-    {
-      "epoch": 11.808604954367667,
-      "grad_norm": 1.5052186250686646,
-      "learning_rate": 7.77537688442211e-05,
-      "loss": 5.4973,
-      "step": 22643
-    },
-    {
-      "epoch": 11.809126466753586,
-      "grad_norm": 1.4929285049438477,
-      "learning_rate": 7.775276381909548e-05,
-      "loss": 5.4388,
-      "step": 22644
-    },
-    {
-      "epoch": 11.809647979139505,
-      "grad_norm": 1.4425764083862305,
-      "learning_rate": 7.775175879396986e-05,
-      "loss": 5.5968,
-      "step": 22645
-    },
-    {
-      "epoch": 11.810169491525423,
-      "grad_norm": 1.4013690948486328,
-      "learning_rate": 7.775075376884422e-05,
-      "loss": 5.3699,
-      "step": 22646
-    },
-    {
-      "epoch": 11.810691003911343,
-      "grad_norm": 1.5370193719863892,
-      "learning_rate": 7.77497487437186e-05,
-      "loss": 5.4315,
-      "step": 22647
-    },
-    {
-      "epoch": 11.811212516297262,
-      "grad_norm": 1.3738844394683838,
-      "learning_rate": 7.774874371859296e-05,
-      "loss": 5.4048,
-      "step": 22648
-    },
-    {
-      "epoch": 11.811734028683182,
-      "grad_norm": 1.3755747079849243,
-      "learning_rate": 7.774773869346734e-05,
-      "loss": 5.4961,
-      "step": 22649
-    },
-    {
-      "epoch": 11.812255541069101,
-      "grad_norm": 1.3077497482299805,
-      "learning_rate": 7.774673366834171e-05,
-      "loss": 5.7154,
-      "step": 22650
-    },
-    {
-      "epoch": 11.812777053455019,
-      "grad_norm": 1.4507712125778198,
-      "learning_rate": 7.774572864321608e-05,
-      "loss": 4.2373,
-      "step": 22651
-    },
-    {
-      "epoch": 11.813298565840938,
-      "grad_norm": 1.441184401512146,
-      "learning_rate": 7.774472361809045e-05,
-      "loss": 5.6634,
-      "step": 22652
-    },
-    {
-      "epoch": 11.813820078226858,
-      "grad_norm": 1.4922544956207275,
-      "learning_rate": 7.774371859296483e-05,
-      "loss": 5.0742,
-      "step": 22653
-    },
-    {
-      "epoch": 11.814341590612777,
-      "grad_norm": 1.8142421245574951,
-      "learning_rate": 7.77427135678392e-05,
-      "loss": 5.6118,
-      "step": 22654
-    },
-    {
-      "epoch": 11.814863102998697,
-      "grad_norm": 1.4250328540802002,
-      "learning_rate": 7.774170854271358e-05,
-      "loss": 5.5076,
-      "step": 22655
-    },
-    {
-      "epoch": 11.815384615384616,
-      "grad_norm": 1.462842345237732,
-      "learning_rate": 7.774070351758795e-05,
-      "loss": 5.4461,
-      "step": 22656
-    },
-    {
-      "epoch": 11.815906127770534,
-      "grad_norm": 1.4932646751403809,
-      "learning_rate": 7.773969849246232e-05,
-      "loss": 5.8046,
-      "step": 22657
-    },
-    {
-      "epoch": 11.816427640156453,
-      "grad_norm": 1.4065730571746826,
-      "learning_rate": 7.773869346733669e-05,
-      "loss": 5.6839,
-      "step": 22658
-    },
-    {
-      "epoch": 11.816949152542373,
-      "grad_norm": 1.3836950063705444,
-      "learning_rate": 7.773768844221105e-05,
-      "loss": 5.3066,
-      "step": 22659
-    },
-    {
-      "epoch": 11.817470664928292,
-      "grad_norm": 1.4914215803146362,
-      "learning_rate": 7.773668341708543e-05,
-      "loss": 5.6476,
-      "step": 22660
-    },
-    {
-      "epoch": 11.817992177314212,
-      "grad_norm": 1.5420743227005005,
-      "learning_rate": 7.77356783919598e-05,
-      "loss": 5.0739,
-      "step": 22661
-    },
-    {
-      "epoch": 11.818513689700131,
-      "grad_norm": 1.473806381225586,
-      "learning_rate": 7.773467336683417e-05,
-      "loss": 5.3445,
-      "step": 22662
-    },
-    {
-      "epoch": 11.819035202086049,
-      "grad_norm": 1.5632953643798828,
-      "learning_rate": 7.773366834170854e-05,
-      "loss": 5.5676,
-      "step": 22663
-    },
-    {
-      "epoch": 11.819556714471968,
-      "grad_norm": 1.4963737726211548,
-      "learning_rate": 7.773266331658292e-05,
-      "loss": 5.3732,
-      "step": 22664
-    },
-    {
-      "epoch": 11.820078226857888,
-      "grad_norm": 1.5482940673828125,
-      "learning_rate": 7.773165829145729e-05,
-      "loss": 4.7295,
-      "step": 22665
-    },
-    {
-      "epoch": 11.820599739243807,
-      "grad_norm": 1.5018850564956665,
-      "learning_rate": 7.773065326633167e-05,
-      "loss": 5.3457,
-      "step": 22666
-    },
-    {
-      "epoch": 11.821121251629727,
-      "grad_norm": 1.502486228942871,
-      "learning_rate": 7.772964824120604e-05,
-      "loss": 5.3797,
-      "step": 22667
-    },
-    {
-      "epoch": 11.821642764015646,
-      "grad_norm": 1.5150305032730103,
-      "learning_rate": 7.772864321608041e-05,
-      "loss": 5.0542,
-      "step": 22668
-    },
-    {
-      "epoch": 11.822164276401564,
-      "grad_norm": 1.6526433229446411,
-      "learning_rate": 7.772763819095478e-05,
-      "loss": 5.0098,
-      "step": 22669
-    },
-    {
-      "epoch": 11.822685788787483,
-      "grad_norm": 1.467862606048584,
-      "learning_rate": 7.772663316582916e-05,
-      "loss": 4.4549,
-      "step": 22670
-    },
-    {
-      "epoch": 11.823207301173403,
-      "grad_norm": 1.4273595809936523,
-      "learning_rate": 7.772562814070352e-05,
-      "loss": 5.5257,
-      "step": 22671
-    },
-    {
-      "epoch": 11.823728813559322,
-      "grad_norm": 1.436381220817566,
-      "learning_rate": 7.772462311557788e-05,
-      "loss": 5.5541,
-      "step": 22672
-    },
-    {
-      "epoch": 11.824250325945242,
-      "grad_norm": 1.3694087266921997,
-      "learning_rate": 7.772361809045226e-05,
-      "loss": 5.8117,
-      "step": 22673
-    },
-    {
-      "epoch": 11.824771838331161,
-      "grad_norm": 1.4815592765808105,
-      "learning_rate": 7.772261306532664e-05,
-      "loss": 5.7105,
-      "step": 22674
-    },
-    {
-      "epoch": 11.825293350717079,
-      "grad_norm": 1.4112731218338013,
-      "learning_rate": 7.772160804020102e-05,
-      "loss": 5.3238,
-      "step": 22675
-    },
-    {
-      "epoch": 11.825814863102998,
-      "grad_norm": 1.543005108833313,
-      "learning_rate": 7.772060301507538e-05,
-      "loss": 5.5464,
-      "step": 22676
-    },
-    {
-      "epoch": 11.826336375488918,
-      "grad_norm": 1.5518546104431152,
-      "learning_rate": 7.771959798994976e-05,
-      "loss": 5.6072,
-      "step": 22677
-    },
-    {
-      "epoch": 11.826857887874837,
-      "grad_norm": 1.4318475723266602,
-      "learning_rate": 7.771859296482412e-05,
-      "loss": 5.3797,
-      "step": 22678
-    },
-    {
-      "epoch": 11.827379400260757,
-      "grad_norm": 1.4969969987869263,
-      "learning_rate": 7.77175879396985e-05,
-      "loss": 5.5642,
-      "step": 22679
-    },
-    {
-      "epoch": 11.827900912646676,
-      "grad_norm": 1.4877046346664429,
-      "learning_rate": 7.771658291457287e-05,
-      "loss": 4.9703,
-      "step": 22680
-    },
-    {
-      "epoch": 11.828422425032594,
-      "grad_norm": 1.477501630783081,
-      "learning_rate": 7.771557788944724e-05,
-      "loss": 5.3417,
-      "step": 22681
-    },
-    {
-      "epoch": 11.828943937418513,
-      "grad_norm": 1.5503380298614502,
-      "learning_rate": 7.771457286432161e-05,
-      "loss": 5.4458,
-      "step": 22682
-    },
-    {
-      "epoch": 11.829465449804433,
-      "grad_norm": 1.4590044021606445,
-      "learning_rate": 7.771356783919599e-05,
-      "loss": 5.4496,
-      "step": 22683
-    },
-    {
-      "epoch": 11.829986962190352,
-      "grad_norm": 1.5131574869155884,
-      "learning_rate": 7.771256281407035e-05,
-      "loss": 5.5568,
-      "step": 22684
-    },
-    {
-      "epoch": 11.830508474576272,
-      "grad_norm": 1.4273806810379028,
-      "learning_rate": 7.771155778894473e-05,
-      "loss": 5.5218,
-      "step": 22685
-    },
-    {
-      "epoch": 11.831029986962191,
-      "grad_norm": 1.5326489210128784,
-      "learning_rate": 7.77105527638191e-05,
-      "loss": 5.5346,
-      "step": 22686
-    },
-    {
-      "epoch": 11.831551499348109,
-      "grad_norm": 1.5204427242279053,
-      "learning_rate": 7.770954773869347e-05,
-      "loss": 5.513,
-      "step": 22687
-    },
-    {
-      "epoch": 11.832073011734028,
-      "grad_norm": 1.4651696681976318,
-      "learning_rate": 7.770854271356785e-05,
-      "loss": 5.5848,
-      "step": 22688
-    },
-    {
-      "epoch": 11.832594524119948,
-      "grad_norm": 1.5152647495269775,
-      "learning_rate": 7.770753768844221e-05,
-      "loss": 5.6562,
-      "step": 22689
-    },
-    {
-      "epoch": 11.833116036505867,
-      "grad_norm": 1.524372935295105,
-      "learning_rate": 7.770653266331659e-05,
-      "loss": 5.3378,
-      "step": 22690
-    },
-    {
-      "epoch": 11.833637548891787,
-      "grad_norm": 1.3820195198059082,
-      "learning_rate": 7.770552763819095e-05,
-      "loss": 5.2402,
-      "step": 22691
-    },
-    {
-      "epoch": 11.834159061277706,
-      "grad_norm": 1.4915844202041626,
-      "learning_rate": 7.770452261306533e-05,
-      "loss": 5.598,
-      "step": 22692
-    },
-    {
-      "epoch": 11.834680573663624,
-      "grad_norm": 1.3802319765090942,
-      "learning_rate": 7.77035175879397e-05,
-      "loss": 5.7639,
-      "step": 22693
-    },
-    {
-      "epoch": 11.835202086049543,
-      "grad_norm": 1.4876441955566406,
-      "learning_rate": 7.770251256281407e-05,
-      "loss": 5.4509,
-      "step": 22694
-    },
-    {
-      "epoch": 11.835723598435463,
-      "grad_norm": 1.6112655401229858,
-      "learning_rate": 7.770150753768845e-05,
-      "loss": 4.8846,
-      "step": 22695
-    },
-    {
-      "epoch": 11.836245110821382,
-      "grad_norm": 1.4504104852676392,
-      "learning_rate": 7.770050251256283e-05,
-      "loss": 5.5573,
-      "step": 22696
-    },
-    {
-      "epoch": 11.836766623207302,
-      "grad_norm": 1.4873734712600708,
-      "learning_rate": 7.76994974874372e-05,
-      "loss": 5.3434,
-      "step": 22697
-    },
-    {
-      "epoch": 11.837288135593221,
-      "grad_norm": 1.5122020244598389,
-      "learning_rate": 7.769849246231156e-05,
-      "loss": 5.0015,
-      "step": 22698
-    },
-    {
-      "epoch": 11.837809647979139,
-      "grad_norm": 1.3753334283828735,
-      "learning_rate": 7.769748743718594e-05,
-      "loss": 5.6805,
-      "step": 22699
-    },
-    {
-      "epoch": 11.838331160365058,
-      "grad_norm": 1.3953170776367188,
-      "learning_rate": 7.76964824120603e-05,
-      "loss": 5.5356,
-      "step": 22700
-    },
-    {
-      "epoch": 11.838852672750978,
-      "grad_norm": 1.3814905881881714,
-      "learning_rate": 7.769547738693468e-05,
-      "loss": 5.7792,
-      "step": 22701
-    },
-    {
-      "epoch": 11.839374185136897,
-      "grad_norm": 1.3932191133499146,
-      "learning_rate": 7.769447236180904e-05,
-      "loss": 5.6008,
-      "step": 22702
-    },
-    {
-      "epoch": 11.839895697522817,
-      "grad_norm": 1.4615752696990967,
-      "learning_rate": 7.769346733668342e-05,
-      "loss": 5.0847,
-      "step": 22703
-    },
-    {
-      "epoch": 11.840417209908736,
-      "grad_norm": 1.4024382829666138,
-      "learning_rate": 7.769246231155778e-05,
-      "loss": 5.7398,
-      "step": 22704
-    },
-    {
-      "epoch": 11.840938722294654,
-      "grad_norm": 1.3653922080993652,
-      "learning_rate": 7.769145728643216e-05,
-      "loss": 5.4427,
-      "step": 22705
-    },
-    {
-      "epoch": 11.841460234680573,
-      "grad_norm": 1.3664058446884155,
-      "learning_rate": 7.769045226130654e-05,
-      "loss": 5.7548,
-      "step": 22706
-    },
-    {
-      "epoch": 11.841981747066493,
-      "grad_norm": 1.3710386753082275,
-      "learning_rate": 7.768944723618092e-05,
-      "loss": 5.2954,
-      "step": 22707
-    },
-    {
-      "epoch": 11.842503259452412,
-      "grad_norm": 1.6766345500946045,
-      "learning_rate": 7.768844221105528e-05,
-      "loss": 4.7559,
-      "step": 22708
-    },
-    {
-      "epoch": 11.843024771838332,
-      "grad_norm": 1.4093843698501587,
-      "learning_rate": 7.768743718592966e-05,
-      "loss": 5.4424,
-      "step": 22709
-    },
-    {
-      "epoch": 11.843546284224251,
-      "grad_norm": 1.2592823505401611,
-      "learning_rate": 7.768643216080402e-05,
-      "loss": 5.7294,
-      "step": 22710
-    },
-    {
-      "epoch": 11.844067796610169,
-      "grad_norm": 1.4486922025680542,
-      "learning_rate": 7.768542713567839e-05,
-      "loss": 5.7768,
-      "step": 22711
-    },
-    {
-      "epoch": 11.844589308996088,
-      "grad_norm": 1.4419364929199219,
-      "learning_rate": 7.768442211055277e-05,
-      "loss": 5.1853,
-      "step": 22712
-    },
-    {
-      "epoch": 11.845110821382008,
-      "grad_norm": 1.4274532794952393,
-      "learning_rate": 7.768341708542713e-05,
-      "loss": 5.6509,
-      "step": 22713
-    },
-    {
-      "epoch": 11.845632333767927,
-      "grad_norm": 1.49880850315094,
-      "learning_rate": 7.768241206030151e-05,
-      "loss": 5.2686,
-      "step": 22714
-    },
-    {
-      "epoch": 11.846153846153847,
-      "grad_norm": 1.4882546663284302,
-      "learning_rate": 7.768140703517587e-05,
-      "loss": 5.242,
-      "step": 22715
-    },
-    {
-      "epoch": 11.846675358539766,
-      "grad_norm": 1.4689980745315552,
-      "learning_rate": 7.768040201005025e-05,
-      "loss": 5.5217,
-      "step": 22716
-    },
-    {
-      "epoch": 11.847196870925684,
-      "grad_norm": 1.4190237522125244,
-      "learning_rate": 7.767939698492463e-05,
-      "loss": 5.5521,
-      "step": 22717
-    },
-    {
-      "epoch": 11.847718383311603,
-      "grad_norm": 1.3695807456970215,
-      "learning_rate": 7.7678391959799e-05,
-      "loss": 5.6601,
-      "step": 22718
-    },
-    {
-      "epoch": 11.848239895697523,
-      "grad_norm": 1.4892425537109375,
-      "learning_rate": 7.767738693467337e-05,
-      "loss": 5.3858,
-      "step": 22719
-    },
-    {
-      "epoch": 11.848761408083442,
-      "grad_norm": 1.4445255994796753,
-      "learning_rate": 7.767638190954775e-05,
-      "loss": 5.8437,
-      "step": 22720
-    },
-    {
-      "epoch": 11.849282920469362,
-      "grad_norm": 1.4237394332885742,
-      "learning_rate": 7.767537688442211e-05,
-      "loss": 5.1048,
-      "step": 22721
-    },
-    {
-      "epoch": 11.84980443285528,
-      "grad_norm": 1.4066697359085083,
-      "learning_rate": 7.767437185929649e-05,
-      "loss": 5.3903,
-      "step": 22722
-    },
-    {
-      "epoch": 11.850325945241199,
-      "grad_norm": 1.3540749549865723,
-      "learning_rate": 7.767336683417085e-05,
-      "loss": 5.7411,
-      "step": 22723
-    },
-    {
-      "epoch": 11.850847457627118,
-      "grad_norm": 1.441124439239502,
-      "learning_rate": 7.767236180904522e-05,
-      "loss": 5.7049,
-      "step": 22724
-    },
-    {
-      "epoch": 11.851368970013038,
-      "grad_norm": 1.5376390218734741,
-      "learning_rate": 7.76713567839196e-05,
-      "loss": 5.6738,
-      "step": 22725
-    },
-    {
-      "epoch": 11.851890482398957,
-      "grad_norm": 1.5435471534729004,
-      "learning_rate": 7.767035175879397e-05,
-      "loss": 5.0871,
-      "step": 22726
-    },
-    {
-      "epoch": 11.852411994784877,
-      "grad_norm": 1.3801062107086182,
-      "learning_rate": 7.766934673366835e-05,
-      "loss": 5.6365,
-      "step": 22727
-    },
-    {
-      "epoch": 11.852933507170796,
-      "grad_norm": 1.4566866159439087,
-      "learning_rate": 7.766834170854272e-05,
-      "loss": 5.398,
-      "step": 22728
-    },
-    {
-      "epoch": 11.853455019556714,
-      "grad_norm": 1.389764428138733,
-      "learning_rate": 7.76673366834171e-05,
-      "loss": 5.53,
-      "step": 22729
-    },
-    {
-      "epoch": 11.853976531942633,
-      "grad_norm": 1.3744151592254639,
-      "learning_rate": 7.766633165829146e-05,
-      "loss": 5.8851,
-      "step": 22730
-    },
-    {
-      "epoch": 11.854498044328553,
-      "grad_norm": 1.4052224159240723,
-      "learning_rate": 7.766532663316584e-05,
-      "loss": 5.7469,
-      "step": 22731
-    },
-    {
-      "epoch": 11.855019556714472,
-      "grad_norm": 1.7201809883117676,
-      "learning_rate": 7.76643216080402e-05,
-      "loss": 5.3621,
-      "step": 22732
-    },
-    {
-      "epoch": 11.855541069100392,
-      "grad_norm": 1.5014305114746094,
-      "learning_rate": 7.766331658291458e-05,
-      "loss": 5.5024,
-      "step": 22733
-    },
-    {
-      "epoch": 11.85606258148631,
-      "grad_norm": 1.3792978525161743,
-      "learning_rate": 7.766231155778894e-05,
-      "loss": 5.1503,
-      "step": 22734
-    },
-    {
-      "epoch": 11.856584093872229,
-      "grad_norm": 1.5915908813476562,
-      "learning_rate": 7.766130653266332e-05,
-      "loss": 5.5429,
-      "step": 22735
-    },
-    {
-      "epoch": 11.857105606258148,
-      "grad_norm": 1.421342134475708,
-      "learning_rate": 7.766030150753769e-05,
-      "loss": 5.5586,
-      "step": 22736
-    },
-    {
-      "epoch": 11.857627118644068,
-      "grad_norm": 1.5954725742340088,
-      "learning_rate": 7.765929648241206e-05,
-      "loss": 5.6307,
-      "step": 22737
-    },
-    {
-      "epoch": 11.858148631029987,
-      "grad_norm": 1.4558377265930176,
-      "learning_rate": 7.765829145728644e-05,
-      "loss": 5.4019,
-      "step": 22738
-    },
-    {
-      "epoch": 11.858670143415907,
-      "grad_norm": 1.4969691038131714,
-      "learning_rate": 7.76572864321608e-05,
-      "loss": 5.6081,
-      "step": 22739
-    },
-    {
-      "epoch": 11.859191655801826,
-      "grad_norm": 1.45846688747406,
-      "learning_rate": 7.765628140703518e-05,
-      "loss": 4.9655,
-      "step": 22740
-    },
-    {
-      "epoch": 11.859713168187744,
-      "grad_norm": 1.4759550094604492,
-      "learning_rate": 7.765527638190955e-05,
-      "loss": 5.7098,
-      "step": 22741
-    },
-    {
-      "epoch": 11.860234680573663,
-      "grad_norm": 1.3591625690460205,
-      "learning_rate": 7.765427135678392e-05,
-      "loss": 5.7386,
-      "step": 22742
-    },
-    {
-      "epoch": 11.860756192959583,
-      "grad_norm": 1.5931816101074219,
-      "learning_rate": 7.765326633165829e-05,
-      "loss": 5.0178,
-      "step": 22743
-    },
-    {
-      "epoch": 11.861277705345502,
-      "grad_norm": 1.4250962734222412,
-      "learning_rate": 7.765226130653267e-05,
-      "loss": 5.5771,
-      "step": 22744
-    },
-    {
-      "epoch": 11.861799217731422,
-      "grad_norm": 1.2878308296203613,
-      "learning_rate": 7.765125628140703e-05,
-      "loss": 5.3277,
-      "step": 22745
-    },
-    {
-      "epoch": 11.86232073011734,
-      "grad_norm": 1.4066941738128662,
-      "learning_rate": 7.765025125628141e-05,
-      "loss": 4.9266,
-      "step": 22746
-    },
-    {
-      "epoch": 11.862842242503259,
-      "grad_norm": 1.3655264377593994,
-      "learning_rate": 7.764924623115579e-05,
-      "loss": 5.4413,
-      "step": 22747
-    },
-    {
-      "epoch": 11.863363754889178,
-      "grad_norm": 1.3785651922225952,
-      "learning_rate": 7.764824120603016e-05,
-      "loss": 5.6103,
-      "step": 22748
-    },
-    {
-      "epoch": 11.863885267275098,
-      "grad_norm": 1.544202208518982,
-      "learning_rate": 7.764723618090453e-05,
-      "loss": 5.7629,
-      "step": 22749
-    },
-    {
-      "epoch": 11.864406779661017,
-      "grad_norm": 1.5446349382400513,
-      "learning_rate": 7.764623115577891e-05,
-      "loss": 4.9071,
-      "step": 22750
-    },
-    {
-      "epoch": 11.864928292046937,
-      "grad_norm": 1.3299455642700195,
-      "learning_rate": 7.764522613065327e-05,
-      "loss": 5.6792,
-      "step": 22751
-    },
-    {
-      "epoch": 11.865449804432854,
-      "grad_norm": 1.566326379776001,
-      "learning_rate": 7.764422110552764e-05,
-      "loss": 5.6187,
-      "step": 22752
-    },
-    {
-      "epoch": 11.865971316818774,
-      "grad_norm": 1.4433218240737915,
-      "learning_rate": 7.764321608040201e-05,
-      "loss": 5.1613,
-      "step": 22753
-    },
-    {
-      "epoch": 11.866492829204693,
-      "grad_norm": 1.488104224205017,
-      "learning_rate": 7.764221105527638e-05,
-      "loss": 5.1093,
-      "step": 22754
-    },
-    {
-      "epoch": 11.867014341590613,
-      "grad_norm": 1.5377894639968872,
-      "learning_rate": 7.764120603015076e-05,
-      "loss": 5.7258,
-      "step": 22755
-    },
-    {
-      "epoch": 11.867535853976532,
-      "grad_norm": 1.470167875289917,
-      "learning_rate": 7.764020100502512e-05,
-      "loss": 5.5673,
-      "step": 22756
-    },
-    {
-      "epoch": 11.868057366362452,
-      "grad_norm": 1.4235527515411377,
-      "learning_rate": 7.76391959798995e-05,
-      "loss": 5.3679,
-      "step": 22757
-    },
-    {
-      "epoch": 11.86857887874837,
-      "grad_norm": 1.5554255247116089,
-      "learning_rate": 7.763819095477388e-05,
-      "loss": 4.9945,
-      "step": 22758
-    },
-    {
-      "epoch": 11.869100391134289,
-      "grad_norm": 1.3428468704223633,
-      "learning_rate": 7.763718592964825e-05,
-      "loss": 5.6422,
-      "step": 22759
-    },
-    {
-      "epoch": 11.869621903520208,
-      "grad_norm": 1.5373897552490234,
-      "learning_rate": 7.763618090452262e-05,
-      "loss": 4.9548,
-      "step": 22760
-    },
-    {
-      "epoch": 11.870143415906128,
-      "grad_norm": 1.4787404537200928,
-      "learning_rate": 7.7635175879397e-05,
-      "loss": 5.4868,
-      "step": 22761
-    },
-    {
-      "epoch": 11.870664928292047,
-      "grad_norm": 1.615301489830017,
-      "learning_rate": 7.763417085427136e-05,
-      "loss": 5.359,
-      "step": 22762
-    },
-    {
-      "epoch": 11.871186440677967,
-      "grad_norm": 1.3885455131530762,
-      "learning_rate": 7.763316582914574e-05,
-      "loss": 5.6623,
-      "step": 22763
-    },
-    {
-      "epoch": 11.871707953063884,
-      "grad_norm": 1.3734537363052368,
-      "learning_rate": 7.76321608040201e-05,
-      "loss": 5.7026,
-      "step": 22764
-    },
-    {
-      "epoch": 11.872229465449804,
-      "grad_norm": 1.4371575117111206,
-      "learning_rate": 7.763115577889447e-05,
-      "loss": 4.9216,
-      "step": 22765
-    },
-    {
-      "epoch": 11.872750977835723,
-      "grad_norm": 1.3584342002868652,
-      "learning_rate": 7.763015075376884e-05,
-      "loss": 5.4639,
-      "step": 22766
-    },
-    {
-      "epoch": 11.873272490221643,
-      "grad_norm": 1.5779708623886108,
-      "learning_rate": 7.762914572864322e-05,
-      "loss": 5.1834,
-      "step": 22767
-    },
-    {
-      "epoch": 11.873794002607562,
-      "grad_norm": 1.4435522556304932,
-      "learning_rate": 7.76281407035176e-05,
-      "loss": 5.7174,
-      "step": 22768
-    },
-    {
-      "epoch": 11.874315514993482,
-      "grad_norm": 1.399422526359558,
-      "learning_rate": 7.762713567839196e-05,
-      "loss": 5.5019,
-      "step": 22769
-    },
-    {
-      "epoch": 11.8748370273794,
-      "grad_norm": 1.4205807447433472,
-      "learning_rate": 7.762613065326634e-05,
-      "loss": 5.5402,
-      "step": 22770
-    },
-    {
-      "epoch": 11.875358539765319,
-      "grad_norm": 1.4055366516113281,
-      "learning_rate": 7.76251256281407e-05,
-      "loss": 5.5026,
-      "step": 22771
-    },
-    {
-      "epoch": 11.875880052151238,
-      "grad_norm": 1.5622785091400146,
-      "learning_rate": 7.762412060301508e-05,
-      "loss": 5.3571,
-      "step": 22772
-    },
-    {
-      "epoch": 11.876401564537158,
-      "grad_norm": 1.4336961507797241,
-      "learning_rate": 7.762311557788945e-05,
-      "loss": 5.6756,
-      "step": 22773
-    },
-    {
-      "epoch": 11.876923076923077,
-      "grad_norm": 1.5095692873001099,
-      "learning_rate": 7.762211055276383e-05,
-      "loss": 5.5744,
-      "step": 22774
-    },
-    {
-      "epoch": 11.877444589308997,
-      "grad_norm": 1.450853943824768,
-      "learning_rate": 7.762110552763819e-05,
-      "loss": 5.4134,
-      "step": 22775
-    },
-    {
-      "epoch": 11.877966101694915,
-      "grad_norm": 1.4331743717193604,
-      "learning_rate": 7.762010050251257e-05,
-      "loss": 5.6386,
-      "step": 22776
-    },
-    {
-      "epoch": 11.878487614080834,
-      "grad_norm": 1.4617465734481812,
-      "learning_rate": 7.761909547738693e-05,
-      "loss": 5.5952,
-      "step": 22777
-    },
-    {
-      "epoch": 11.879009126466753,
-      "grad_norm": 1.61185622215271,
-      "learning_rate": 7.761809045226131e-05,
-      "loss": 4.8385,
-      "step": 22778
-    },
-    {
-      "epoch": 11.879530638852673,
-      "grad_norm": 1.49136221408844,
-      "learning_rate": 7.761708542713569e-05,
-      "loss": 5.7215,
-      "step": 22779
-    },
-    {
-      "epoch": 11.880052151238592,
-      "grad_norm": 1.435266375541687,
-      "learning_rate": 7.761608040201005e-05,
-      "loss": 5.469,
-      "step": 22780
-    },
-    {
-      "epoch": 11.880573663624512,
-      "grad_norm": 1.4352315664291382,
-      "learning_rate": 7.761507537688443e-05,
-      "loss": 5.2443,
-      "step": 22781
-    },
-    {
-      "epoch": 11.88109517601043,
-      "grad_norm": 1.4091715812683105,
-      "learning_rate": 7.76140703517588e-05,
-      "loss": 5.4824,
-      "step": 22782
-    },
-    {
-      "epoch": 11.881616688396349,
-      "grad_norm": 1.4578218460083008,
-      "learning_rate": 7.761306532663317e-05,
-      "loss": 5.2459,
-      "step": 22783
-    },
-    {
-      "epoch": 11.882138200782268,
-      "grad_norm": 1.4634780883789062,
-      "learning_rate": 7.761206030150754e-05,
-      "loss": 5.3587,
-      "step": 22784
-    },
-    {
-      "epoch": 11.882659713168188,
-      "grad_norm": 1.447589635848999,
-      "learning_rate": 7.761105527638191e-05,
-      "loss": 5.4676,
-      "step": 22785
-    },
-    {
-      "epoch": 11.883181225554107,
-      "grad_norm": 1.4280949831008911,
-      "learning_rate": 7.761005025125628e-05,
-      "loss": 5.1423,
-      "step": 22786
-    },
-    {
-      "epoch": 11.883702737940027,
-      "grad_norm": 1.5410606861114502,
-      "learning_rate": 7.760904522613066e-05,
-      "loss": 4.5789,
-      "step": 22787
-    },
-    {
-      "epoch": 11.884224250325945,
-      "grad_norm": 1.4003599882125854,
-      "learning_rate": 7.760804020100503e-05,
-      "loss": 5.6668,
-      "step": 22788
-    },
-    {
-      "epoch": 11.884745762711864,
-      "grad_norm": 1.3400455713272095,
-      "learning_rate": 7.760703517587941e-05,
-      "loss": 5.3984,
-      "step": 22789
-    },
-    {
-      "epoch": 11.885267275097783,
-      "grad_norm": 1.4327237606048584,
-      "learning_rate": 7.760603015075378e-05,
-      "loss": 5.4558,
-      "step": 22790
-    },
-    {
-      "epoch": 11.885788787483703,
-      "grad_norm": 1.387759804725647,
-      "learning_rate": 7.760502512562814e-05,
-      "loss": 5.7487,
-      "step": 22791
-    },
-    {
-      "epoch": 11.886310299869622,
-      "grad_norm": 1.3164583444595337,
-      "learning_rate": 7.760402010050252e-05,
-      "loss": 5.7215,
-      "step": 22792
-    },
-    {
-      "epoch": 11.886831812255542,
-      "grad_norm": 1.4991700649261475,
-      "learning_rate": 7.760301507537688e-05,
-      "loss": 5.3973,
-      "step": 22793
-    },
-    {
-      "epoch": 11.88735332464146,
-      "grad_norm": 1.348984956741333,
-      "learning_rate": 7.760201005025126e-05,
-      "loss": 5.3126,
-      "step": 22794
-    },
-    {
-      "epoch": 11.887874837027379,
-      "grad_norm": 1.466622233390808,
-      "learning_rate": 7.760100502512562e-05,
-      "loss": 5.583,
-      "step": 22795
-    },
-    {
-      "epoch": 11.888396349413298,
-      "grad_norm": 1.6002801656723022,
-      "learning_rate": 7.76e-05,
-      "loss": 4.7766,
-      "step": 22796
-    },
-    {
-      "epoch": 11.888917861799218,
-      "grad_norm": 1.462462067604065,
-      "learning_rate": 7.759899497487437e-05,
-      "loss": 5.5335,
-      "step": 22797
-    },
-    {
-      "epoch": 11.889439374185137,
-      "grad_norm": 1.622644305229187,
-      "learning_rate": 7.759798994974874e-05,
-      "loss": 5.4052,
-      "step": 22798
-    },
-    {
-      "epoch": 11.889960886571057,
-      "grad_norm": 1.441701889038086,
-      "learning_rate": 7.759698492462312e-05,
-      "loss": 5.7722,
-      "step": 22799
-    },
-    {
-      "epoch": 11.890482398956975,
-      "grad_norm": 1.5509207248687744,
-      "learning_rate": 7.75959798994975e-05,
-      "loss": 5.2706,
-      "step": 22800
-    },
-    {
-      "epoch": 11.891003911342894,
-      "grad_norm": 1.5040091276168823,
-      "learning_rate": 7.759497487437186e-05,
-      "loss": 5.4223,
-      "step": 22801
-    },
-    {
-      "epoch": 11.891525423728813,
-      "grad_norm": 1.3873541355133057,
-      "learning_rate": 7.759396984924624e-05,
-      "loss": 5.6891,
-      "step": 22802
-    },
-    {
-      "epoch": 11.892046936114733,
-      "grad_norm": 1.436231255531311,
-      "learning_rate": 7.75929648241206e-05,
-      "loss": 5.2115,
-      "step": 22803
-    },
-    {
-      "epoch": 11.892568448500652,
-      "grad_norm": 1.3234046697616577,
-      "learning_rate": 7.759195979899497e-05,
-      "loss": 5.5025,
-      "step": 22804
-    },
-    {
-      "epoch": 11.893089960886572,
-      "grad_norm": 1.5106326341629028,
-      "learning_rate": 7.759095477386935e-05,
-      "loss": 5.2402,
-      "step": 22805
-    },
-    {
-      "epoch": 11.89361147327249,
-      "grad_norm": 1.4511653184890747,
-      "learning_rate": 7.758994974874371e-05,
-      "loss": 5.7186,
-      "step": 22806
-    },
-    {
-      "epoch": 11.894132985658409,
-      "grad_norm": 1.402060866355896,
-      "learning_rate": 7.758894472361809e-05,
-      "loss": 5.6111,
-      "step": 22807
-    },
-    {
-      "epoch": 11.894654498044329,
-      "grad_norm": 1.4000260829925537,
-      "learning_rate": 7.758793969849247e-05,
-      "loss": 5.7007,
-      "step": 22808
-    },
-    {
-      "epoch": 11.895176010430248,
-      "grad_norm": 1.5257493257522583,
-      "learning_rate": 7.758693467336685e-05,
-      "loss": 5.0421,
-      "step": 22809
-    },
-    {
-      "epoch": 11.895697522816167,
-      "grad_norm": 1.494583010673523,
-      "learning_rate": 7.758592964824121e-05,
-      "loss": 5.6571,
-      "step": 22810
-    },
-    {
-      "epoch": 11.896219035202087,
-      "grad_norm": 1.4354771375656128,
-      "learning_rate": 7.758492462311559e-05,
-      "loss": 5.6048,
-      "step": 22811
-    },
-    {
-      "epoch": 11.896740547588005,
-      "grad_norm": 1.5401678085327148,
-      "learning_rate": 7.758391959798995e-05,
-      "loss": 5.4956,
-      "step": 22812
-    },
-    {
-      "epoch": 11.897262059973924,
-      "grad_norm": 1.3480920791625977,
-      "learning_rate": 7.758291457286433e-05,
-      "loss": 5.6419,
-      "step": 22813
-    },
-    {
-      "epoch": 11.897783572359844,
-      "grad_norm": 1.4230210781097412,
-      "learning_rate": 7.75819095477387e-05,
-      "loss": 5.3635,
-      "step": 22814
-    },
-    {
-      "epoch": 11.898305084745763,
-      "grad_norm": 1.4187970161437988,
-      "learning_rate": 7.758090452261307e-05,
-      "loss": 5.2659,
-      "step": 22815
-    },
-    {
-      "epoch": 11.898826597131682,
-      "grad_norm": 1.463417887687683,
-      "learning_rate": 7.757989949748744e-05,
-      "loss": 5.5211,
-      "step": 22816
-    },
-    {
-      "epoch": 11.8993481095176,
-      "grad_norm": 1.5144002437591553,
-      "learning_rate": 7.757889447236181e-05,
-      "loss": 4.9418,
-      "step": 22817
-    },
-    {
-      "epoch": 11.89986962190352,
-      "grad_norm": 1.4543139934539795,
-      "learning_rate": 7.757788944723618e-05,
-      "loss": 4.9694,
-      "step": 22818
-    },
-    {
-      "epoch": 11.900391134289439,
-      "grad_norm": 1.514482855796814,
-      "learning_rate": 7.757688442211056e-05,
-      "loss": 5.1917,
-      "step": 22819
-    },
-    {
-      "epoch": 11.900912646675359,
-      "grad_norm": 1.4549249410629272,
-      "learning_rate": 7.757587939698493e-05,
-      "loss": 5.747,
-      "step": 22820
-    },
-    {
-      "epoch": 11.901434159061278,
-      "grad_norm": 1.5126415491104126,
-      "learning_rate": 7.75748743718593e-05,
-      "loss": 5.079,
-      "step": 22821
-    },
-    {
-      "epoch": 11.901955671447197,
-      "grad_norm": 1.6441028118133545,
-      "learning_rate": 7.757386934673368e-05,
-      "loss": 5.105,
-      "step": 22822
-    },
-    {
-      "epoch": 11.902477183833117,
-      "grad_norm": 1.4159923791885376,
-      "learning_rate": 7.757286432160804e-05,
-      "loss": 5.5145,
-      "step": 22823
-    },
-    {
-      "epoch": 11.902998696219035,
-      "grad_norm": 1.548539400100708,
-      "learning_rate": 7.757185929648242e-05,
-      "loss": 5.5158,
-      "step": 22824
-    },
-    {
-      "epoch": 11.903520208604954,
-      "grad_norm": 1.4797399044036865,
-      "learning_rate": 7.757085427135678e-05,
-      "loss": 5.4004,
-      "step": 22825
-    },
-    {
-      "epoch": 11.904041720990874,
-      "grad_norm": 1.321890950202942,
-      "learning_rate": 7.756984924623116e-05,
-      "loss": 5.7552,
-      "step": 22826
-    },
-    {
-      "epoch": 11.904563233376793,
-      "grad_norm": 1.5082499980926514,
-      "learning_rate": 7.756884422110553e-05,
-      "loss": 5.5115,
-      "step": 22827
-    },
-    {
-      "epoch": 11.905084745762712,
-      "grad_norm": 1.4244741201400757,
-      "learning_rate": 7.75678391959799e-05,
-      "loss": 5.8067,
-      "step": 22828
-    },
-    {
-      "epoch": 11.90560625814863,
-      "grad_norm": 1.4083441495895386,
-      "learning_rate": 7.756683417085428e-05,
-      "loss": 5.0828,
-      "step": 22829
-    },
-    {
-      "epoch": 11.90612777053455,
-      "grad_norm": 1.3979841470718384,
-      "learning_rate": 7.756582914572866e-05,
-      "loss": 5.4976,
-      "step": 22830
-    },
-    {
-      "epoch": 11.906649282920469,
-      "grad_norm": 1.4337127208709717,
-      "learning_rate": 7.756482412060302e-05,
-      "loss": 5.6878,
-      "step": 22831
-    },
-    {
-      "epoch": 11.907170795306389,
-      "grad_norm": 1.4998655319213867,
-      "learning_rate": 7.756381909547739e-05,
-      "loss": 5.2629,
-      "step": 22832
-    },
-    {
-      "epoch": 11.907692307692308,
-      "grad_norm": 1.5177509784698486,
-      "learning_rate": 7.756281407035177e-05,
-      "loss": 5.2169,
-      "step": 22833
-    },
-    {
-      "epoch": 11.908213820078227,
-      "grad_norm": 1.465701699256897,
-      "learning_rate": 7.756180904522613e-05,
-      "loss": 5.1916,
-      "step": 22834
-    },
-    {
-      "epoch": 11.908735332464147,
-      "grad_norm": 1.3066152334213257,
-      "learning_rate": 7.756080402010051e-05,
-      "loss": 5.5898,
-      "step": 22835
-    },
-    {
-      "epoch": 11.909256844850065,
-      "grad_norm": 1.512636423110962,
-      "learning_rate": 7.755979899497487e-05,
-      "loss": 5.3812,
-      "step": 22836
-    },
-    {
-      "epoch": 11.909778357235984,
-      "grad_norm": 1.4630831480026245,
-      "learning_rate": 7.755879396984925e-05,
-      "loss": 5.0614,
-      "step": 22837
-    },
-    {
-      "epoch": 11.910299869621904,
-      "grad_norm": 1.4972333908081055,
-      "learning_rate": 7.755778894472361e-05,
-      "loss": 5.4945,
-      "step": 22838
-    },
-    {
-      "epoch": 11.910821382007823,
-      "grad_norm": 1.4103953838348389,
-      "learning_rate": 7.755678391959799e-05,
-      "loss": 5.6814,
-      "step": 22839
-    },
-    {
-      "epoch": 11.911342894393742,
-      "grad_norm": 1.4429258108139038,
-      "learning_rate": 7.755577889447237e-05,
-      "loss": 5.4453,
-      "step": 22840
-    },
-    {
-      "epoch": 11.91186440677966,
-      "grad_norm": 1.3515636920928955,
-      "learning_rate": 7.755477386934675e-05,
-      "loss": 5.4124,
-      "step": 22841
-    },
-    {
-      "epoch": 11.91238591916558,
-      "grad_norm": 1.4410792589187622,
-      "learning_rate": 7.755376884422111e-05,
-      "loss": 5.3818,
-      "step": 22842
-    },
-    {
-      "epoch": 11.9129074315515,
-      "grad_norm": 1.4222732782363892,
-      "learning_rate": 7.755276381909549e-05,
-      "loss": 5.5009,
-      "step": 22843
-    },
-    {
-      "epoch": 11.913428943937419,
-      "grad_norm": 1.4764423370361328,
-      "learning_rate": 7.755175879396985e-05,
-      "loss": 5.3812,
-      "step": 22844
-    },
-    {
-      "epoch": 11.913950456323338,
-      "grad_norm": 1.5048819780349731,
-      "learning_rate": 7.755075376884422e-05,
-      "loss": 5.2435,
-      "step": 22845
-    },
-    {
-      "epoch": 11.914471968709258,
-      "grad_norm": 1.5366404056549072,
-      "learning_rate": 7.75497487437186e-05,
-      "loss": 4.9811,
-      "step": 22846
-    },
-    {
-      "epoch": 11.914993481095175,
-      "grad_norm": 1.4410068988800049,
-      "learning_rate": 7.754874371859296e-05,
-      "loss": 5.2261,
-      "step": 22847
-    },
-    {
-      "epoch": 11.915514993481095,
-      "grad_norm": 1.3997759819030762,
-      "learning_rate": 7.754773869346734e-05,
-      "loss": 5.1885,
-      "step": 22848
-    },
-    {
-      "epoch": 11.916036505867014,
-      "grad_norm": 1.405009150505066,
-      "learning_rate": 7.754673366834172e-05,
-      "loss": 5.4254,
-      "step": 22849
-    },
-    {
-      "epoch": 11.916558018252934,
-      "grad_norm": 1.5465158224105835,
-      "learning_rate": 7.75457286432161e-05,
-      "loss": 5.4577,
-      "step": 22850
-    },
-    {
-      "epoch": 11.917079530638853,
-      "grad_norm": 1.5516701936721802,
-      "learning_rate": 7.754472361809046e-05,
-      "loss": 4.9255,
-      "step": 22851
-    },
-    {
-      "epoch": 11.917601043024773,
-      "grad_norm": 1.398799180984497,
-      "learning_rate": 7.754371859296484e-05,
-      "loss": 5.1286,
-      "step": 22852
-    },
-    {
-      "epoch": 11.91812255541069,
-      "grad_norm": 1.4291921854019165,
-      "learning_rate": 7.75427135678392e-05,
-      "loss": 5.2491,
-      "step": 22853
-    },
-    {
-      "epoch": 11.91864406779661,
-      "grad_norm": 1.3680471181869507,
-      "learning_rate": 7.754170854271358e-05,
-      "loss": 5.651,
-      "step": 22854
-    },
-    {
-      "epoch": 11.91916558018253,
-      "grad_norm": 1.4037113189697266,
-      "learning_rate": 7.754070351758794e-05,
-      "loss": 5.73,
-      "step": 22855
-    },
-    {
-      "epoch": 11.919687092568449,
-      "grad_norm": 1.36825430393219,
-      "learning_rate": 7.753969849246232e-05,
-      "loss": 5.5351,
-      "step": 22856
-    },
-    {
-      "epoch": 11.920208604954368,
-      "grad_norm": 1.4208530187606812,
-      "learning_rate": 7.753869346733668e-05,
-      "loss": 5.6324,
-      "step": 22857
-    },
-    {
-      "epoch": 11.920730117340288,
-      "grad_norm": 1.375556468963623,
-      "learning_rate": 7.753768844221105e-05,
-      "loss": 4.9496,
-      "step": 22858
-    },
-    {
-      "epoch": 11.921251629726205,
-      "grad_norm": 1.5284929275512695,
-      "learning_rate": 7.753668341708543e-05,
-      "loss": 5.7215,
-      "step": 22859
-    },
-    {
-      "epoch": 11.921773142112125,
-      "grad_norm": 1.4598140716552734,
-      "learning_rate": 7.75356783919598e-05,
-      "loss": 5.805,
-      "step": 22860
-    },
-    {
-      "epoch": 11.922294654498044,
-      "grad_norm": 1.6093276739120483,
-      "learning_rate": 7.753467336683418e-05,
-      "loss": 4.8365,
-      "step": 22861
-    },
-    {
-      "epoch": 11.922816166883964,
-      "grad_norm": 1.5204020738601685,
-      "learning_rate": 7.753366834170855e-05,
-      "loss": 5.2214,
-      "step": 22862
-    },
-    {
-      "epoch": 11.923337679269883,
-      "grad_norm": 1.3323092460632324,
-      "learning_rate": 7.753266331658292e-05,
-      "loss": 5.5469,
-      "step": 22863
-    },
-    {
-      "epoch": 11.923859191655803,
-      "grad_norm": 1.4833664894104004,
-      "learning_rate": 7.753165829145729e-05,
-      "loss": 5.2296,
-      "step": 22864
-    },
-    {
-      "epoch": 11.92438070404172,
-      "grad_norm": 1.4347758293151855,
-      "learning_rate": 7.753065326633167e-05,
-      "loss": 5.5105,
-      "step": 22865
-    },
-    {
-      "epoch": 11.92490221642764,
-      "grad_norm": 1.6885360479354858,
-      "learning_rate": 7.752964824120603e-05,
-      "loss": 5.1407,
-      "step": 22866
-    },
-    {
-      "epoch": 11.92542372881356,
-      "grad_norm": 1.5492643117904663,
-      "learning_rate": 7.752864321608041e-05,
-      "loss": 4.9557,
-      "step": 22867
-    },
-    {
-      "epoch": 11.925945241199479,
-      "grad_norm": 1.3427265882492065,
-      "learning_rate": 7.752763819095477e-05,
-      "loss": 5.7423,
-      "step": 22868
-    },
-    {
-      "epoch": 11.926466753585398,
-      "grad_norm": 1.4374852180480957,
-      "learning_rate": 7.752663316582915e-05,
-      "loss": 5.4093,
-      "step": 22869
-    },
-    {
-      "epoch": 11.926988265971318,
-      "grad_norm": 1.4844778776168823,
-      "learning_rate": 7.752562814070351e-05,
-      "loss": 5.2429,
-      "step": 22870
-    },
-    {
-      "epoch": 11.927509778357235,
-      "grad_norm": 1.4306800365447998,
-      "learning_rate": 7.752462311557789e-05,
-      "loss": 5.6985,
-      "step": 22871
-    },
-    {
-      "epoch": 11.928031290743155,
-      "grad_norm": 1.3779678344726562,
-      "learning_rate": 7.752361809045227e-05,
-      "loss": 5.278,
-      "step": 22872
-    },
-    {
-      "epoch": 11.928552803129074,
-      "grad_norm": 1.4241935014724731,
-      "learning_rate": 7.752261306532663e-05,
-      "loss": 4.9573,
-      "step": 22873
-    },
-    {
-      "epoch": 11.929074315514994,
-      "grad_norm": 1.4509252309799194,
-      "learning_rate": 7.752160804020101e-05,
-      "loss": 5.3836,
-      "step": 22874
-    },
-    {
-      "epoch": 11.929595827900913,
-      "grad_norm": 1.45831298828125,
-      "learning_rate": 7.752060301507538e-05,
-      "loss": 5.1369,
-      "step": 22875
-    },
-    {
-      "epoch": 11.930117340286833,
-      "grad_norm": 1.336169719696045,
-      "learning_rate": 7.751959798994975e-05,
-      "loss": 5.7205,
-      "step": 22876
-    },
-    {
-      "epoch": 11.93063885267275,
-      "grad_norm": 1.4821585416793823,
-      "learning_rate": 7.751859296482412e-05,
-      "loss": 5.5223,
-      "step": 22877
-    },
-    {
-      "epoch": 11.93116036505867,
-      "grad_norm": 1.3948441743850708,
-      "learning_rate": 7.75175879396985e-05,
-      "loss": 5.7977,
-      "step": 22878
-    },
-    {
-      "epoch": 11.93168187744459,
-      "grad_norm": 1.427140235900879,
-      "learning_rate": 7.751658291457286e-05,
-      "loss": 5.2725,
-      "step": 22879
-    },
-    {
-      "epoch": 11.932203389830509,
-      "grad_norm": 1.4401754140853882,
-      "learning_rate": 7.751557788944724e-05,
-      "loss": 5.047,
-      "step": 22880
-    },
-    {
-      "epoch": 11.932724902216428,
-      "grad_norm": 1.4448680877685547,
-      "learning_rate": 7.751457286432162e-05,
-      "loss": 5.0463,
-      "step": 22881
-    },
-    {
-      "epoch": 11.933246414602348,
-      "grad_norm": 1.4015953540802002,
-      "learning_rate": 7.7513567839196e-05,
-      "loss": 5.8469,
-      "step": 22882
-    },
-    {
-      "epoch": 11.933767926988265,
-      "grad_norm": 1.4136987924575806,
-      "learning_rate": 7.751256281407036e-05,
-      "loss": 5.5123,
-      "step": 22883
-    },
-    {
-      "epoch": 11.934289439374185,
-      "grad_norm": 1.3676910400390625,
-      "learning_rate": 7.751155778894472e-05,
-      "loss": 5.568,
-      "step": 22884
-    },
-    {
-      "epoch": 11.934810951760104,
-      "grad_norm": 1.512812852859497,
-      "learning_rate": 7.75105527638191e-05,
-      "loss": 4.5273,
-      "step": 22885
-    },
-    {
-      "epoch": 11.935332464146024,
-      "grad_norm": 1.3816696405410767,
-      "learning_rate": 7.750954773869346e-05,
-      "loss": 5.6683,
-      "step": 22886
-    },
-    {
-      "epoch": 11.935853976531943,
-      "grad_norm": 1.6008821725845337,
-      "learning_rate": 7.750854271356784e-05,
-      "loss": 5.052,
-      "step": 22887
-    },
-    {
-      "epoch": 11.936375488917863,
-      "grad_norm": 1.3696075677871704,
-      "learning_rate": 7.750753768844221e-05,
-      "loss": 5.8726,
-      "step": 22888
-    },
-    {
-      "epoch": 11.93689700130378,
-      "grad_norm": 1.3850027322769165,
-      "learning_rate": 7.750653266331658e-05,
-      "loss": 5.5269,
-      "step": 22889
-    },
-    {
-      "epoch": 11.9374185136897,
-      "grad_norm": 1.3178136348724365,
-      "learning_rate": 7.750552763819095e-05,
-      "loss": 5.4601,
-      "step": 22890
-    },
-    {
-      "epoch": 11.93794002607562,
-      "grad_norm": 1.3278322219848633,
-      "learning_rate": 7.750452261306533e-05,
-      "loss": 5.6932,
-      "step": 22891
-    },
-    {
-      "epoch": 11.938461538461539,
-      "grad_norm": 1.6154142618179321,
-      "learning_rate": 7.75035175879397e-05,
-      "loss": 5.0555,
-      "step": 22892
-    },
-    {
-      "epoch": 11.938983050847458,
-      "grad_norm": 1.6157270669937134,
-      "learning_rate": 7.750251256281408e-05,
-      "loss": 5.0396,
-      "step": 22893
-    },
-    {
-      "epoch": 11.939504563233378,
-      "grad_norm": 1.4964320659637451,
-      "learning_rate": 7.750150753768845e-05,
-      "loss": 5.38,
-      "step": 22894
-    },
-    {
-      "epoch": 11.940026075619295,
-      "grad_norm": 1.5105252265930176,
-      "learning_rate": 7.750050251256282e-05,
-      "loss": 5.2109,
-      "step": 22895
-    },
-    {
-      "epoch": 11.940547588005215,
-      "grad_norm": 1.4408823251724243,
-      "learning_rate": 7.749949748743719e-05,
-      "loss": 5.5066,
-      "step": 22896
-    },
-    {
-      "epoch": 11.941069100391134,
-      "grad_norm": 1.358117938041687,
-      "learning_rate": 7.749849246231155e-05,
-      "loss": 5.4111,
-      "step": 22897
-    },
-    {
-      "epoch": 11.941590612777054,
-      "grad_norm": 1.4580084085464478,
-      "learning_rate": 7.749748743718593e-05,
-      "loss": 4.7753,
-      "step": 22898
-    },
-    {
-      "epoch": 11.942112125162973,
-      "grad_norm": 1.4823791980743408,
-      "learning_rate": 7.74964824120603e-05,
-      "loss": 5.5798,
-      "step": 22899
-    },
-    {
-      "epoch": 11.94263363754889,
-      "grad_norm": 1.318467378616333,
-      "learning_rate": 7.749547738693467e-05,
-      "loss": 5.7957,
-      "step": 22900
-    },
-    {
-      "epoch": 11.94315514993481,
-      "grad_norm": 1.4421759843826294,
-      "learning_rate": 7.749447236180905e-05,
-      "loss": 5.3881,
-      "step": 22901
-    },
-    {
-      "epoch": 11.94367666232073,
-      "grad_norm": 1.5247275829315186,
-      "learning_rate": 7.749346733668343e-05,
-      "loss": 4.5736,
-      "step": 22902
-    },
-    {
-      "epoch": 11.94419817470665,
-      "grad_norm": 1.3596153259277344,
-      "learning_rate": 7.749246231155779e-05,
-      "loss": 5.6093,
-      "step": 22903
-    },
-    {
-      "epoch": 11.944719687092569,
-      "grad_norm": 1.572378158569336,
-      "learning_rate": 7.749145728643217e-05,
-      "loss": 5.5783,
-      "step": 22904
-    },
-    {
-      "epoch": 11.945241199478488,
-      "grad_norm": 1.4474315643310547,
-      "learning_rate": 7.749045226130654e-05,
-      "loss": 5.787,
-      "step": 22905
-    },
-    {
-      "epoch": 11.945762711864408,
-      "grad_norm": 1.5192053318023682,
-      "learning_rate": 7.748944723618091e-05,
-      "loss": 4.9053,
-      "step": 22906
-    },
-    {
-      "epoch": 11.946284224250325,
-      "grad_norm": 1.5773200988769531,
-      "learning_rate": 7.748844221105528e-05,
-      "loss": 5.7178,
-      "step": 22907
-    },
-    {
-      "epoch": 11.946805736636245,
-      "grad_norm": 1.4677947759628296,
-      "learning_rate": 7.748743718592966e-05,
-      "loss": 5.1982,
-      "step": 22908
-    },
-    {
-      "epoch": 11.947327249022164,
-      "grad_norm": 1.577429175376892,
-      "learning_rate": 7.748643216080402e-05,
-      "loss": 5.1003,
-      "step": 22909
-    },
-    {
-      "epoch": 11.947848761408084,
-      "grad_norm": 1.485586404800415,
-      "learning_rate": 7.74854271356784e-05,
-      "loss": 5.4856,
-      "step": 22910
-    },
-    {
-      "epoch": 11.948370273794003,
-      "grad_norm": 1.4528512954711914,
-      "learning_rate": 7.748442211055276e-05,
-      "loss": 5.4762,
-      "step": 22911
-    },
-    {
-      "epoch": 11.94889178617992,
-      "grad_norm": 1.4264497756958008,
-      "learning_rate": 7.748341708542714e-05,
-      "loss": 5.165,
-      "step": 22912
-    },
-    {
-      "epoch": 11.94941329856584,
-      "grad_norm": 1.3501282930374146,
-      "learning_rate": 7.748241206030152e-05,
-      "loss": 5.6086,
-      "step": 22913
-    },
-    {
-      "epoch": 11.94993481095176,
-      "grad_norm": 1.413609266281128,
-      "learning_rate": 7.748140703517588e-05,
-      "loss": 5.7157,
-      "step": 22914
-    },
-    {
-      "epoch": 11.95045632333768,
-      "grad_norm": 1.4211093187332153,
-      "learning_rate": 7.748040201005026e-05,
-      "loss": 5.5683,
-      "step": 22915
-    },
-    {
-      "epoch": 11.950977835723599,
-      "grad_norm": 1.4390029907226562,
-      "learning_rate": 7.747939698492462e-05,
-      "loss": 5.0202,
-      "step": 22916
-    },
-    {
-      "epoch": 11.951499348109518,
-      "grad_norm": 1.4236406087875366,
-      "learning_rate": 7.7478391959799e-05,
-      "loss": 5.7061,
-      "step": 22917
-    },
-    {
-      "epoch": 11.952020860495438,
-      "grad_norm": 1.4782980680465698,
-      "learning_rate": 7.747738693467337e-05,
-      "loss": 5.7844,
-      "step": 22918
-    },
-    {
-      "epoch": 11.952542372881355,
-      "grad_norm": 1.6496809720993042,
-      "learning_rate": 7.747638190954774e-05,
-      "loss": 5.45,
-      "step": 22919
-    },
-    {
-      "epoch": 11.953063885267275,
-      "grad_norm": 1.587364673614502,
-      "learning_rate": 7.747537688442211e-05,
-      "loss": 5.5383,
-      "step": 22920
-    },
-    {
-      "epoch": 11.953585397653194,
-      "grad_norm": 1.3951923847198486,
-      "learning_rate": 7.747437185929649e-05,
-      "loss": 5.617,
-      "step": 22921
-    },
-    {
-      "epoch": 11.954106910039114,
-      "grad_norm": 1.4872679710388184,
-      "learning_rate": 7.747336683417086e-05,
-      "loss": 5.3978,
-      "step": 22922
-    },
-    {
-      "epoch": 11.954628422425033,
-      "grad_norm": 1.5248669385910034,
-      "learning_rate": 7.747236180904524e-05,
-      "loss": 5.4296,
-      "step": 22923
-    },
-    {
-      "epoch": 11.955149934810951,
-      "grad_norm": 1.4441609382629395,
-      "learning_rate": 7.74713567839196e-05,
-      "loss": 5.4566,
-      "step": 22924
-    },
-    {
-      "epoch": 11.95567144719687,
-      "grad_norm": 1.6143574714660645,
-      "learning_rate": 7.747035175879397e-05,
-      "loss": 5.4018,
-      "step": 22925
-    },
-    {
-      "epoch": 11.95619295958279,
-      "grad_norm": 1.516904592514038,
-      "learning_rate": 7.746934673366835e-05,
-      "loss": 5.5865,
-      "step": 22926
-    },
-    {
-      "epoch": 11.95671447196871,
-      "grad_norm": 1.6117643117904663,
-      "learning_rate": 7.746834170854271e-05,
-      "loss": 5.324,
-      "step": 22927
-    },
-    {
-      "epoch": 11.957235984354629,
-      "grad_norm": 1.4899641275405884,
-      "learning_rate": 7.746733668341709e-05,
-      "loss": 5.3224,
-      "step": 22928
-    },
-    {
-      "epoch": 11.957757496740548,
-      "grad_norm": 1.378010869026184,
-      "learning_rate": 7.746633165829145e-05,
-      "loss": 5.2692,
-      "step": 22929
-    },
-    {
-      "epoch": 11.958279009126468,
-      "grad_norm": 1.4138506650924683,
-      "learning_rate": 7.746532663316583e-05,
-      "loss": 5.5553,
-      "step": 22930
-    },
-    {
-      "epoch": 11.958800521512385,
-      "grad_norm": 1.4516761302947998,
-      "learning_rate": 7.74643216080402e-05,
-      "loss": 5.3097,
-      "step": 22931
-    },
-    {
-      "epoch": 11.959322033898305,
-      "grad_norm": 1.3401198387145996,
-      "learning_rate": 7.746331658291457e-05,
-      "loss": 5.9274,
-      "step": 22932
-    },
-    {
-      "epoch": 11.959843546284224,
-      "grad_norm": 1.4307774305343628,
-      "learning_rate": 7.746231155778895e-05,
-      "loss": 5.5347,
-      "step": 22933
-    },
-    {
-      "epoch": 11.960365058670144,
-      "grad_norm": 1.419299840927124,
-      "learning_rate": 7.746130653266333e-05,
-      "loss": 5.0269,
-      "step": 22934
-    },
-    {
-      "epoch": 11.960886571056063,
-      "grad_norm": 1.4793978929519653,
-      "learning_rate": 7.74603015075377e-05,
-      "loss": 5.5296,
-      "step": 22935
-    },
-    {
-      "epoch": 11.961408083441981,
-      "grad_norm": 1.2592582702636719,
-      "learning_rate": 7.745929648241207e-05,
-      "loss": 4.9113,
-      "step": 22936
-    },
-    {
-      "epoch": 11.9619295958279,
-      "grad_norm": 1.4450180530548096,
-      "learning_rate": 7.745829145728644e-05,
-      "loss": 5.5124,
-      "step": 22937
-    },
-    {
-      "epoch": 11.96245110821382,
-      "grad_norm": 1.455039620399475,
-      "learning_rate": 7.74572864321608e-05,
-      "loss": 5.5062,
-      "step": 22938
-    },
-    {
-      "epoch": 11.96297262059974,
-      "grad_norm": 1.3687421083450317,
-      "learning_rate": 7.745628140703518e-05,
-      "loss": 4.9313,
-      "step": 22939
-    },
-    {
-      "epoch": 11.963494132985659,
-      "grad_norm": 1.4199001789093018,
-      "learning_rate": 7.745527638190954e-05,
-      "loss": 5.4903,
-      "step": 22940
-    },
-    {
-      "epoch": 11.964015645371578,
-      "grad_norm": 1.3811789751052856,
-      "learning_rate": 7.745427135678392e-05,
-      "loss": 5.5484,
-      "step": 22941
-    },
-    {
-      "epoch": 11.964537157757496,
-      "grad_norm": 1.5089401006698608,
-      "learning_rate": 7.74532663316583e-05,
-      "loss": 5.7181,
-      "step": 22942
-    },
-    {
-      "epoch": 11.965058670143415,
-      "grad_norm": 1.3901058435440063,
-      "learning_rate": 7.745226130653268e-05,
-      "loss": 5.6014,
-      "step": 22943
-    },
-    {
-      "epoch": 11.965580182529335,
-      "grad_norm": 1.4288290739059448,
-      "learning_rate": 7.745125628140704e-05,
-      "loss": 5.4775,
-      "step": 22944
-    },
-    {
-      "epoch": 11.966101694915254,
-      "grad_norm": 1.579694151878357,
-      "learning_rate": 7.745025125628142e-05,
-      "loss": 4.5908,
-      "step": 22945
-    },
-    {
-      "epoch": 11.966623207301174,
-      "grad_norm": 1.4494441747665405,
-      "learning_rate": 7.744924623115578e-05,
-      "loss": 5.1616,
-      "step": 22946
-    },
-    {
-      "epoch": 11.967144719687093,
-      "grad_norm": 1.377129077911377,
-      "learning_rate": 7.744824120603016e-05,
-      "loss": 5.4526,
-      "step": 22947
-    },
-    {
-      "epoch": 11.967666232073011,
-      "grad_norm": 1.545974612236023,
-      "learning_rate": 7.744723618090452e-05,
-      "loss": 5.161,
-      "step": 22948
-    },
-    {
-      "epoch": 11.96818774445893,
-      "grad_norm": 1.471352219581604,
-      "learning_rate": 7.74462311557789e-05,
-      "loss": 5.2824,
-      "step": 22949
-    },
-    {
-      "epoch": 11.96870925684485,
-      "grad_norm": 1.6461633443832397,
-      "learning_rate": 7.744522613065327e-05,
-      "loss": 5.3714,
-      "step": 22950
-    },
-    {
-      "epoch": 11.96923076923077,
-      "grad_norm": 1.549264669418335,
-      "learning_rate": 7.744422110552763e-05,
-      "loss": 5.4444,
-      "step": 22951
-    },
-    {
-      "epoch": 11.969752281616689,
-      "grad_norm": 1.4102518558502197,
-      "learning_rate": 7.744321608040201e-05,
-      "loss": 5.3566,
-      "step": 22952
-    },
-    {
-      "epoch": 11.970273794002608,
-      "grad_norm": 1.4235460758209229,
-      "learning_rate": 7.744221105527639e-05,
-      "loss": 5.5685,
-      "step": 22953
-    },
-    {
-      "epoch": 11.970795306388526,
-      "grad_norm": 1.4383094310760498,
-      "learning_rate": 7.744120603015076e-05,
-      "loss": 5.7895,
-      "step": 22954
-    },
-    {
-      "epoch": 11.971316818774445,
-      "grad_norm": 1.355224609375,
-      "learning_rate": 7.744020100502513e-05,
-      "loss": 5.3349,
-      "step": 22955
-    },
-    {
-      "epoch": 11.971838331160365,
-      "grad_norm": 1.3440808057785034,
-      "learning_rate": 7.74391959798995e-05,
-      "loss": 4.6717,
-      "step": 22956
-    },
-    {
-      "epoch": 11.972359843546284,
-      "grad_norm": 1.4453121423721313,
-      "learning_rate": 7.743819095477387e-05,
-      "loss": 5.3944,
-      "step": 22957
-    },
-    {
-      "epoch": 11.972881355932204,
-      "grad_norm": 1.4313324689865112,
-      "learning_rate": 7.743718592964825e-05,
-      "loss": 5.6264,
-      "step": 22958
-    },
-    {
-      "epoch": 11.973402868318123,
-      "grad_norm": 1.4688475131988525,
-      "learning_rate": 7.743618090452261e-05,
-      "loss": 5.669,
-      "step": 22959
-    },
-    {
-      "epoch": 11.973924380704041,
-      "grad_norm": 1.509272575378418,
-      "learning_rate": 7.743517587939699e-05,
-      "loss": 4.943,
-      "step": 22960
-    },
-    {
-      "epoch": 11.97444589308996,
-      "grad_norm": 1.462412714958191,
-      "learning_rate": 7.743417085427135e-05,
-      "loss": 5.1811,
-      "step": 22961
-    },
-    {
-      "epoch": 11.97496740547588,
-      "grad_norm": 1.3269699811935425,
-      "learning_rate": 7.743316582914573e-05,
-      "loss": 5.1096,
-      "step": 22962
-    },
-    {
-      "epoch": 11.9754889178618,
-      "grad_norm": 1.435983657836914,
-      "learning_rate": 7.743216080402011e-05,
-      "loss": 4.9752,
-      "step": 22963
-    },
-    {
-      "epoch": 11.976010430247719,
-      "grad_norm": 1.4768002033233643,
-      "learning_rate": 7.743115577889447e-05,
-      "loss": 5.4148,
-      "step": 22964
-    },
-    {
-      "epoch": 11.976531942633638,
-      "grad_norm": 1.3879103660583496,
-      "learning_rate": 7.743015075376885e-05,
-      "loss": 5.7411,
-      "step": 22965
-    },
-    {
-      "epoch": 11.977053455019556,
-      "grad_norm": 1.3144611120224,
-      "learning_rate": 7.742914572864322e-05,
-      "loss": 5.5622,
-      "step": 22966
-    },
-    {
-      "epoch": 11.977574967405475,
-      "grad_norm": 1.3055403232574463,
-      "learning_rate": 7.74281407035176e-05,
-      "loss": 5.8914,
-      "step": 22967
-    },
-    {
-      "epoch": 11.978096479791395,
-      "grad_norm": 1.3867517709732056,
-      "learning_rate": 7.742713567839196e-05,
-      "loss": 5.5364,
-      "step": 22968
-    },
-    {
-      "epoch": 11.978617992177314,
-      "grad_norm": 1.4226723909378052,
-      "learning_rate": 7.742613065326634e-05,
-      "loss": 5.385,
-      "step": 22969
-    },
-    {
-      "epoch": 11.979139504563234,
-      "grad_norm": 1.5965116024017334,
-      "learning_rate": 7.74251256281407e-05,
-      "loss": 5.1103,
-      "step": 22970
-    },
-    {
-      "epoch": 11.979661016949153,
-      "grad_norm": 1.4780265092849731,
-      "learning_rate": 7.742412060301508e-05,
-      "loss": 5.4493,
-      "step": 22971
-    },
-    {
-      "epoch": 11.980182529335071,
-      "grad_norm": 1.4549636840820312,
-      "learning_rate": 7.742311557788944e-05,
-      "loss": 5.4608,
-      "step": 22972
-    },
-    {
-      "epoch": 11.98070404172099,
-      "grad_norm": 1.5664701461791992,
-      "learning_rate": 7.742211055276382e-05,
-      "loss": 5.5927,
-      "step": 22973
-    },
-    {
-      "epoch": 11.98122555410691,
-      "grad_norm": 1.4730370044708252,
-      "learning_rate": 7.74211055276382e-05,
-      "loss": 5.505,
-      "step": 22974
-    },
-    {
-      "epoch": 11.98174706649283,
-      "grad_norm": 1.6222821474075317,
-      "learning_rate": 7.742010050251258e-05,
-      "loss": 5.2496,
-      "step": 22975
-    },
-    {
-      "epoch": 11.982268578878749,
-      "grad_norm": 1.451987862586975,
-      "learning_rate": 7.741909547738694e-05,
-      "loss": 5.7032,
-      "step": 22976
-    },
-    {
-      "epoch": 11.982790091264668,
-      "grad_norm": 1.5514923334121704,
-      "learning_rate": 7.74180904522613e-05,
-      "loss": 5.0709,
-      "step": 22977
-    },
-    {
-      "epoch": 11.983311603650586,
-      "grad_norm": 1.5507442951202393,
-      "learning_rate": 7.741708542713568e-05,
-      "loss": 5.2622,
-      "step": 22978
-    },
-    {
-      "epoch": 11.983833116036505,
-      "grad_norm": 1.3387459516525269,
-      "learning_rate": 7.741608040201005e-05,
-      "loss": 5.1802,
-      "step": 22979
-    },
-    {
-      "epoch": 11.984354628422425,
-      "grad_norm": 1.4337706565856934,
-      "learning_rate": 7.741507537688443e-05,
-      "loss": 4.9556,
-      "step": 22980
-    },
-    {
-      "epoch": 11.984876140808344,
-      "grad_norm": 1.4726160764694214,
-      "learning_rate": 7.741407035175879e-05,
-      "loss": 5.2272,
-      "step": 22981
-    },
-    {
-      "epoch": 11.985397653194264,
-      "grad_norm": 1.37355637550354,
-      "learning_rate": 7.741306532663317e-05,
-      "loss": 5.158,
-      "step": 22982
-    },
-    {
-      "epoch": 11.985919165580183,
-      "grad_norm": 1.6005475521087646,
-      "learning_rate": 7.741206030150754e-05,
-      "loss": 5.5226,
-      "step": 22983
-    },
-    {
-      "epoch": 11.986440677966101,
-      "grad_norm": 1.5161288976669312,
-      "learning_rate": 7.741105527638192e-05,
-      "loss": 5.5159,
-      "step": 22984
-    },
-    {
-      "epoch": 11.98696219035202,
-      "grad_norm": 1.4105861186981201,
-      "learning_rate": 7.741005025125629e-05,
-      "loss": 4.5088,
-      "step": 22985
-    },
-    {
-      "epoch": 11.98748370273794,
-      "grad_norm": 1.402233362197876,
-      "learning_rate": 7.740904522613066e-05,
-      "loss": 5.751,
-      "step": 22986
-    },
-    {
-      "epoch": 11.98800521512386,
-      "grad_norm": 1.411699891090393,
-      "learning_rate": 7.740804020100503e-05,
-      "loss": 4.953,
-      "step": 22987
-    },
-    {
-      "epoch": 11.988526727509779,
-      "grad_norm": 1.4951319694519043,
-      "learning_rate": 7.740703517587941e-05,
-      "loss": 5.6998,
-      "step": 22988
-    },
-    {
-      "epoch": 11.989048239895698,
-      "grad_norm": 1.3839956521987915,
-      "learning_rate": 7.740603015075377e-05,
-      "loss": 5.2351,
-      "step": 22989
-    },
-    {
-      "epoch": 11.989569752281616,
-      "grad_norm": 1.5212976932525635,
-      "learning_rate": 7.740502512562814e-05,
-      "loss": 5.2589,
-      "step": 22990
-    },
-    {
-      "epoch": 11.990091264667535,
-      "grad_norm": 1.4692249298095703,
-      "learning_rate": 7.740402010050251e-05,
-      "loss": 5.3548,
-      "step": 22991
-    },
-    {
-      "epoch": 11.990612777053455,
-      "grad_norm": 1.383193016052246,
-      "learning_rate": 7.740301507537688e-05,
-      "loss": 5.5483,
-      "step": 22992
-    },
-    {
-      "epoch": 11.991134289439374,
-      "grad_norm": 1.3485733270645142,
-      "learning_rate": 7.740201005025126e-05,
-      "loss": 5.0923,
-      "step": 22993
-    },
-    {
-      "epoch": 11.991655801825294,
-      "grad_norm": 1.415628433227539,
-      "learning_rate": 7.740100502512563e-05,
-      "loss": 5.7726,
-      "step": 22994
-    },
-    {
-      "epoch": 11.992177314211212,
-      "grad_norm": 1.4167263507843018,
-      "learning_rate": 7.740000000000001e-05,
-      "loss": 5.6277,
-      "step": 22995
-    },
-    {
-      "epoch": 11.992698826597131,
-      "grad_norm": 1.4581527709960938,
-      "learning_rate": 7.739899497487438e-05,
-      "loss": 5.8325,
-      "step": 22996
-    },
-    {
-      "epoch": 11.99322033898305,
-      "grad_norm": 1.4527504444122314,
-      "learning_rate": 7.739798994974875e-05,
-      "loss": 5.7563,
-      "step": 22997
-    },
-    {
-      "epoch": 11.99374185136897,
-      "grad_norm": 1.4024816751480103,
-      "learning_rate": 7.739698492462312e-05,
-      "loss": 5.6077,
-      "step": 22998
-    },
-    {
-      "epoch": 11.99426336375489,
-      "grad_norm": 1.55463445186615,
-      "learning_rate": 7.73959798994975e-05,
-      "loss": 5.3885,
-      "step": 22999
-    },
-    {
-      "epoch": 11.994784876140809,
-      "grad_norm": 1.548906683921814,
-      "learning_rate": 7.739497487437186e-05,
-      "loss": 5.4905,
-      "step": 23000
-    },
-    {
-      "epoch": 11.995306388526728,
-      "grad_norm": 1.4390546083450317,
-      "learning_rate": 7.739396984924624e-05,
-      "loss": 5.4738,
-      "step": 23001
-    },
-    {
-      "epoch": 11.995827900912646,
-      "grad_norm": 1.45039701461792,
-      "learning_rate": 7.73929648241206e-05,
-      "loss": 5.2864,
-      "step": 23002
-    },
-    {
-      "epoch": 11.996349413298566,
-      "grad_norm": 1.3973594903945923,
-      "learning_rate": 7.739195979899498e-05,
-      "loss": 5.6716,
-      "step": 23003
-    },
-    {
-      "epoch": 11.996870925684485,
-      "grad_norm": 1.4853302240371704,
-      "learning_rate": 7.739095477386936e-05,
-      "loss": 5.4836,
-      "step": 23004
-    },
-    {
-      "epoch": 11.997392438070404,
-      "grad_norm": 1.5190021991729736,
-      "learning_rate": 7.738994974874372e-05,
-      "loss": 5.2924,
-      "step": 23005
-    },
-    {
-      "epoch": 11.997913950456324,
-      "grad_norm": 1.6016267538070679,
-      "learning_rate": 7.73889447236181e-05,
-      "loss": 4.7034,
-      "step": 23006
-    },
-    {
-      "epoch": 11.998435462842242,
-      "grad_norm": 1.594670295715332,
-      "learning_rate": 7.738793969849246e-05,
-      "loss": 5.3146,
-      "step": 23007
-    },
-    {
-      "epoch": 11.998956975228161,
-      "grad_norm": 1.558101773262024,
-      "learning_rate": 7.738693467336684e-05,
-      "loss": 5.5484,
-      "step": 23008
-    },
-    {
-      "epoch": 11.99947848761408,
-      "grad_norm": 1.5748473405838013,
-      "learning_rate": 7.73859296482412e-05,
-      "loss": 5.0185,
-      "step": 23009
-    },
-    {
-      "epoch": 12.0,
-      "grad_norm": 1.6429901123046875,
-      "learning_rate": 7.738492462311558e-05,
-      "loss": 5.0599,
-      "step": 23010
-    },
-    {
-      "epoch": 12.00052151238592,
-      "grad_norm": 1.3868403434753418,
-      "learning_rate": 7.738391959798995e-05,
-      "loss": 5.6849,
-      "step": 23011
-    },
-    {
-      "epoch": 12.001043024771839,
-      "grad_norm": 1.55873441696167,
-      "learning_rate": 7.738291457286433e-05,
-      "loss": 5.4073,
-      "step": 23012
-    },
-    {
-      "epoch": 12.001564537157757,
-      "grad_norm": 1.5900051593780518,
-      "learning_rate": 7.738190954773869e-05,
-      "loss": 4.9917,
-      "step": 23013
-    },
-    {
-      "epoch": 12.002086049543676,
-      "grad_norm": 1.488173246383667,
-      "learning_rate": 7.738090452261307e-05,
-      "loss": 4.9508,
-      "step": 23014
-    },
-    {
-      "epoch": 12.002607561929596,
-      "grad_norm": 1.5797466039657593,
-      "learning_rate": 7.737989949748745e-05,
-      "loss": 5.2239,
-      "step": 23015
-    },
-    {
-      "epoch": 12.003129074315515,
-      "grad_norm": 1.5107700824737549,
-      "learning_rate": 7.737889447236182e-05,
-      "loss": 5.61,
-      "step": 23016
-    },
-    {
-      "epoch": 12.003650586701434,
-      "grad_norm": 1.4363987445831299,
-      "learning_rate": 7.737788944723619e-05,
-      "loss": 5.1831,
-      "step": 23017
-    },
-    {
-      "epoch": 12.004172099087354,
-      "grad_norm": 1.5059337615966797,
-      "learning_rate": 7.737688442211055e-05,
-      "loss": 5.5337,
-      "step": 23018
-    },
-    {
-      "epoch": 12.004693611473272,
-      "grad_norm": 1.3895223140716553,
-      "learning_rate": 7.737587939698493e-05,
-      "loss": 5.0801,
-      "step": 23019
-    },
-    {
-      "epoch": 12.005215123859191,
-      "grad_norm": 1.4425482749938965,
-      "learning_rate": 7.73748743718593e-05,
-      "loss": 5.018,
-      "step": 23020
-    },
-    {
-      "epoch": 12.00573663624511,
-      "grad_norm": 1.4788527488708496,
-      "learning_rate": 7.737386934673367e-05,
-      "loss": 5.2914,
-      "step": 23021
-    },
-    {
-      "epoch": 12.00625814863103,
-      "grad_norm": 1.4998434782028198,
-      "learning_rate": 7.737286432160804e-05,
-      "loss": 5.6597,
-      "step": 23022
-    },
-    {
-      "epoch": 12.00677966101695,
-      "grad_norm": 1.3338634967803955,
-      "learning_rate": 7.737185929648241e-05,
-      "loss": 5.0609,
-      "step": 23023
-    },
-    {
-      "epoch": 12.007301173402869,
-      "grad_norm": 1.4579410552978516,
-      "learning_rate": 7.737085427135679e-05,
-      "loss": 5.6053,
-      "step": 23024
-    },
-    {
-      "epoch": 12.007822685788787,
-      "grad_norm": 1.3267433643341064,
-      "learning_rate": 7.736984924623117e-05,
-      "loss": 5.5105,
-      "step": 23025
-    },
-    {
-      "epoch": 12.008344198174706,
-      "grad_norm": 1.4452365636825562,
-      "learning_rate": 7.736884422110553e-05,
-      "loss": 5.2404,
-      "step": 23026
-    },
-    {
-      "epoch": 12.008865710560626,
-      "grad_norm": 1.4934563636779785,
-      "learning_rate": 7.736783919597991e-05,
-      "loss": 5.4014,
-      "step": 23027
-    },
-    {
-      "epoch": 12.009387222946545,
-      "grad_norm": 1.4981578588485718,
-      "learning_rate": 7.736683417085428e-05,
-      "loss": 5.6698,
-      "step": 23028
-    },
-    {
-      "epoch": 12.009908735332465,
-      "grad_norm": 1.568068027496338,
-      "learning_rate": 7.736582914572865e-05,
-      "loss": 5.255,
-      "step": 23029
-    },
-    {
-      "epoch": 12.010430247718384,
-      "grad_norm": 1.7857692241668701,
-      "learning_rate": 7.736482412060302e-05,
-      "loss": 5.0551,
-      "step": 23030
-    },
-    {
-      "epoch": 12.010951760104302,
-      "grad_norm": 1.4389275312423706,
-      "learning_rate": 7.736381909547738e-05,
-      "loss": 5.2006,
-      "step": 23031
-    },
-    {
-      "epoch": 12.011473272490221,
-      "grad_norm": 1.4503684043884277,
-      "learning_rate": 7.736281407035176e-05,
-      "loss": 4.8194,
-      "step": 23032
-    },
-    {
-      "epoch": 12.01199478487614,
-      "grad_norm": 1.3615896701812744,
-      "learning_rate": 7.736180904522612e-05,
-      "loss": 5.8071,
-      "step": 23033
-    },
-    {
-      "epoch": 12.01251629726206,
-      "grad_norm": 1.374568223953247,
-      "learning_rate": 7.73608040201005e-05,
-      "loss": 5.9056,
-      "step": 23034
-    },
-    {
-      "epoch": 12.01303780964798,
-      "grad_norm": 1.4508581161499023,
-      "learning_rate": 7.735979899497488e-05,
-      "loss": 5.711,
-      "step": 23035
-    },
-    {
-      "epoch": 12.013559322033899,
-      "grad_norm": 1.4165550470352173,
-      "learning_rate": 7.735879396984926e-05,
-      "loss": 5.737,
-      "step": 23036
-    },
-    {
-      "epoch": 12.014080834419817,
-      "grad_norm": 1.6466091871261597,
-      "learning_rate": 7.735778894472362e-05,
-      "loss": 5.5826,
-      "step": 23037
-    },
-    {
-      "epoch": 12.014602346805736,
-      "grad_norm": 1.5673227310180664,
-      "learning_rate": 7.7356783919598e-05,
-      "loss": 5.1376,
-      "step": 23038
-    },
-    {
-      "epoch": 12.015123859191656,
-      "grad_norm": 1.4129509925842285,
-      "learning_rate": 7.735577889447236e-05,
-      "loss": 5.5523,
-      "step": 23039
-    },
-    {
-      "epoch": 12.015645371577575,
-      "grad_norm": 1.505897879600525,
-      "learning_rate": 7.735477386934674e-05,
-      "loss": 5.0505,
-      "step": 23040
-    },
-    {
-      "epoch": 12.016166883963495,
-      "grad_norm": 1.424403429031372,
-      "learning_rate": 7.73537688442211e-05,
-      "loss": 5.7055,
-      "step": 23041
-    },
-    {
-      "epoch": 12.016688396349414,
-      "grad_norm": 1.4854938983917236,
-      "learning_rate": 7.735276381909548e-05,
-      "loss": 5.6729,
-      "step": 23042
-    },
-    {
-      "epoch": 12.017209908735332,
-      "grad_norm": 1.5336570739746094,
-      "learning_rate": 7.735175879396985e-05,
-      "loss": 4.9899,
-      "step": 23043
-    },
-    {
-      "epoch": 12.017731421121251,
-      "grad_norm": 1.7156323194503784,
-      "learning_rate": 7.735075376884421e-05,
-      "loss": 5.4018,
-      "step": 23044
-    },
-    {
-      "epoch": 12.01825293350717,
-      "grad_norm": 1.4789094924926758,
-      "learning_rate": 7.734974874371859e-05,
-      "loss": 5.4424,
-      "step": 23045
-    },
-    {
-      "epoch": 12.01877444589309,
-      "grad_norm": 1.52579927444458,
-      "learning_rate": 7.734874371859297e-05,
-      "loss": 5.6689,
-      "step": 23046
-    },
-    {
-      "epoch": 12.01929595827901,
-      "grad_norm": 1.6701877117156982,
-      "learning_rate": 7.734773869346735e-05,
-      "loss": 5.6131,
-      "step": 23047
-    },
-    {
-      "epoch": 12.019817470664929,
-      "grad_norm": 1.4676882028579712,
-      "learning_rate": 7.734673366834171e-05,
-      "loss": 5.4915,
-      "step": 23048
-    },
-    {
-      "epoch": 12.020338983050847,
-      "grad_norm": 1.4788750410079956,
-      "learning_rate": 7.734572864321609e-05,
-      "loss": 5.487,
-      "step": 23049
-    },
-    {
-      "epoch": 12.020860495436766,
-      "grad_norm": 1.5135889053344727,
-      "learning_rate": 7.734472361809045e-05,
-      "loss": 5.4157,
-      "step": 23050
-    },
-    {
-      "epoch": 12.021382007822686,
-      "grad_norm": 1.4289710521697998,
-      "learning_rate": 7.734371859296483e-05,
-      "loss": 5.2048,
-      "step": 23051
-    },
-    {
-      "epoch": 12.021903520208605,
-      "grad_norm": 1.438616156578064,
-      "learning_rate": 7.73427135678392e-05,
-      "loss": 5.4265,
-      "step": 23052
-    },
-    {
-      "epoch": 12.022425032594525,
-      "grad_norm": 1.339760184288025,
-      "learning_rate": 7.734170854271357e-05,
-      "loss": 5.626,
-      "step": 23053
-    },
-    {
-      "epoch": 12.022946544980444,
-      "grad_norm": 1.4843639135360718,
-      "learning_rate": 7.734070351758794e-05,
-      "loss": 5.3591,
-      "step": 23054
-    },
-    {
-      "epoch": 12.023468057366362,
-      "grad_norm": 1.5094364881515503,
-      "learning_rate": 7.733969849246231e-05,
-      "loss": 5.5945,
-      "step": 23055
-    },
-    {
-      "epoch": 12.023989569752281,
-      "grad_norm": 1.5017175674438477,
-      "learning_rate": 7.733869346733669e-05,
-      "loss": 5.0144,
-      "step": 23056
-    },
-    {
-      "epoch": 12.0245110821382,
-      "grad_norm": 1.497463345527649,
-      "learning_rate": 7.733768844221106e-05,
-      "loss": 5.4758,
-      "step": 23057
-    },
-    {
-      "epoch": 12.02503259452412,
-      "grad_norm": 1.4867048263549805,
-      "learning_rate": 7.733668341708543e-05,
-      "loss": 5.4507,
-      "step": 23058
-    },
-    {
-      "epoch": 12.02555410691004,
-      "grad_norm": 1.5620288848876953,
-      "learning_rate": 7.73356783919598e-05,
-      "loss": 5.612,
-      "step": 23059
-    },
-    {
-      "epoch": 12.026075619295959,
-      "grad_norm": 1.4907606840133667,
-      "learning_rate": 7.733467336683418e-05,
-      "loss": 5.6935,
-      "step": 23060
-    },
-    {
-      "epoch": 12.026597131681877,
-      "grad_norm": 1.4035640954971313,
-      "learning_rate": 7.733366834170854e-05,
-      "loss": 5.0586,
-      "step": 23061
-    },
-    {
-      "epoch": 12.027118644067796,
-      "grad_norm": 1.515650749206543,
-      "learning_rate": 7.733266331658292e-05,
-      "loss": 5.6342,
-      "step": 23062
-    },
-    {
-      "epoch": 12.027640156453716,
-      "grad_norm": 1.674403429031372,
-      "learning_rate": 7.733165829145728e-05,
-      "loss": 5.0236,
-      "step": 23063
-    },
-    {
-      "epoch": 12.028161668839635,
-      "grad_norm": 1.5215201377868652,
-      "learning_rate": 7.733065326633166e-05,
-      "loss": 5.7664,
-      "step": 23064
-    },
-    {
-      "epoch": 12.028683181225555,
-      "grad_norm": 1.3893413543701172,
-      "learning_rate": 7.732964824120603e-05,
-      "loss": 5.6832,
-      "step": 23065
-    },
-    {
-      "epoch": 12.029204693611474,
-      "grad_norm": 1.37310791015625,
-      "learning_rate": 7.73286432160804e-05,
-      "loss": 5.5458,
-      "step": 23066
-    },
-    {
-      "epoch": 12.029726205997392,
-      "grad_norm": 1.408949375152588,
-      "learning_rate": 7.732763819095478e-05,
-      "loss": 5.4108,
-      "step": 23067
-    },
-    {
-      "epoch": 12.030247718383311,
-      "grad_norm": 1.455244541168213,
-      "learning_rate": 7.732663316582916e-05,
-      "loss": 5.8502,
-      "step": 23068
-    },
-    {
-      "epoch": 12.03076923076923,
-      "grad_norm": 1.5637232065200806,
-      "learning_rate": 7.732562814070352e-05,
-      "loss": 4.8855,
-      "step": 23069
-    },
-    {
-      "epoch": 12.03129074315515,
-      "grad_norm": 1.5385692119598389,
-      "learning_rate": 7.732462311557789e-05,
-      "loss": 5.0241,
-      "step": 23070
-    },
-    {
-      "epoch": 12.03181225554107,
-      "grad_norm": 1.517449140548706,
-      "learning_rate": 7.732361809045227e-05,
-      "loss": 4.9672,
-      "step": 23071
-    },
-    {
-      "epoch": 12.032333767926989,
-      "grad_norm": 1.5120311975479126,
-      "learning_rate": 7.732261306532663e-05,
-      "loss": 5.4023,
-      "step": 23072
-    },
-    {
-      "epoch": 12.032855280312907,
-      "grad_norm": 1.41050124168396,
-      "learning_rate": 7.732160804020101e-05,
-      "loss": 5.6811,
-      "step": 23073
-    },
-    {
-      "epoch": 12.033376792698826,
-      "grad_norm": 1.492136001586914,
-      "learning_rate": 7.732060301507537e-05,
-      "loss": 4.8404,
-      "step": 23074
-    },
-    {
-      "epoch": 12.033898305084746,
-      "grad_norm": 1.5372898578643799,
-      "learning_rate": 7.731959798994975e-05,
-      "loss": 4.9455,
-      "step": 23075
-    },
-    {
-      "epoch": 12.034419817470665,
-      "grad_norm": 1.401280164718628,
-      "learning_rate": 7.731859296482413e-05,
-      "loss": 5.54,
-      "step": 23076
-    },
-    {
-      "epoch": 12.034941329856585,
-      "grad_norm": 1.3830305337905884,
-      "learning_rate": 7.73175879396985e-05,
-      "loss": 5.7067,
-      "step": 23077
-    },
-    {
-      "epoch": 12.035462842242504,
-      "grad_norm": 1.3954341411590576,
-      "learning_rate": 7.731658291457287e-05,
-      "loss": 5.776,
-      "step": 23078
-    },
-    {
-      "epoch": 12.035984354628422,
-      "grad_norm": 1.4666849374771118,
-      "learning_rate": 7.731557788944725e-05,
-      "loss": 5.5772,
-      "step": 23079
-    },
-    {
-      "epoch": 12.036505867014341,
-      "grad_norm": 1.440699577331543,
-      "learning_rate": 7.731457286432161e-05,
-      "loss": 5.6447,
-      "step": 23080
-    },
-    {
-      "epoch": 12.03702737940026,
-      "grad_norm": 1.4799137115478516,
-      "learning_rate": 7.731356783919599e-05,
-      "loss": 5.4413,
-      "step": 23081
-    },
-    {
-      "epoch": 12.03754889178618,
-      "grad_norm": 1.412797212600708,
-      "learning_rate": 7.731256281407035e-05,
-      "loss": 5.6522,
-      "step": 23082
-    },
-    {
-      "epoch": 12.0380704041721,
-      "grad_norm": 1.506855845451355,
-      "learning_rate": 7.731155778894472e-05,
-      "loss": 5.7743,
-      "step": 23083
-    },
-    {
-      "epoch": 12.038591916558019,
-      "grad_norm": 1.3782081604003906,
-      "learning_rate": 7.73105527638191e-05,
-      "loss": 5.4172,
-      "step": 23084
-    },
-    {
-      "epoch": 12.039113428943937,
-      "grad_norm": 1.4747611284255981,
-      "learning_rate": 7.730954773869346e-05,
-      "loss": 5.5969,
-      "step": 23085
-    },
-    {
-      "epoch": 12.039634941329856,
-      "grad_norm": 1.4648793935775757,
-      "learning_rate": 7.730854271356784e-05,
-      "loss": 5.6073,
-      "step": 23086
-    },
-    {
-      "epoch": 12.040156453715776,
-      "grad_norm": 1.4057538509368896,
-      "learning_rate": 7.730753768844222e-05,
-      "loss": 5.5792,
-      "step": 23087
-    },
-    {
-      "epoch": 12.040677966101695,
-      "grad_norm": 1.5262291431427002,
-      "learning_rate": 7.73065326633166e-05,
-      "loss": 4.8959,
-      "step": 23088
-    },
-    {
-      "epoch": 12.041199478487615,
-      "grad_norm": 1.4551416635513306,
-      "learning_rate": 7.730552763819096e-05,
-      "loss": 5.1434,
-      "step": 23089
-    },
-    {
-      "epoch": 12.041720990873534,
-      "grad_norm": 1.5444854497909546,
-      "learning_rate": 7.730452261306534e-05,
-      "loss": 5.0794,
-      "step": 23090
-    },
-    {
-      "epoch": 12.042242503259452,
-      "grad_norm": 1.5522671937942505,
-      "learning_rate": 7.73035175879397e-05,
-      "loss": 5.4987,
-      "step": 23091
-    },
-    {
-      "epoch": 12.042764015645371,
-      "grad_norm": 1.6167161464691162,
-      "learning_rate": 7.730251256281408e-05,
-      "loss": 5.3182,
-      "step": 23092
-    },
-    {
-      "epoch": 12.04328552803129,
-      "grad_norm": 1.501484751701355,
-      "learning_rate": 7.730150753768844e-05,
-      "loss": 5.16,
-      "step": 23093
-    },
-    {
-      "epoch": 12.04380704041721,
-      "grad_norm": 1.597159504890442,
-      "learning_rate": 7.730050251256282e-05,
-      "loss": 5.6526,
-      "step": 23094
-    },
-    {
-      "epoch": 12.04432855280313,
-      "grad_norm": 1.5498296022415161,
-      "learning_rate": 7.729949748743718e-05,
-      "loss": 5.552,
-      "step": 23095
-    },
-    {
-      "epoch": 12.044850065189049,
-      "grad_norm": 1.4342327117919922,
-      "learning_rate": 7.729849246231156e-05,
-      "loss": 5.3567,
-      "step": 23096
-    },
-    {
-      "epoch": 12.045371577574967,
-      "grad_norm": 1.4757411479949951,
-      "learning_rate": 7.729748743718594e-05,
-      "loss": 5.397,
-      "step": 23097
-    },
-    {
-      "epoch": 12.045893089960886,
-      "grad_norm": 1.5573078393936157,
-      "learning_rate": 7.72964824120603e-05,
-      "loss": 5.4249,
-      "step": 23098
-    },
-    {
-      "epoch": 12.046414602346806,
-      "grad_norm": 1.480603575706482,
-      "learning_rate": 7.729547738693468e-05,
-      "loss": 5.3173,
-      "step": 23099
-    },
-    {
-      "epoch": 12.046936114732725,
-      "grad_norm": 1.4882498979568481,
-      "learning_rate": 7.729447236180905e-05,
-      "loss": 5.1823,
-      "step": 23100
-    },
-    {
-      "epoch": 12.047457627118645,
-      "grad_norm": 1.428487777709961,
-      "learning_rate": 7.729346733668342e-05,
-      "loss": 5.7096,
-      "step": 23101
-    },
-    {
-      "epoch": 12.047979139504564,
-      "grad_norm": 1.6970319747924805,
-      "learning_rate": 7.729246231155779e-05,
-      "loss": 4.7932,
-      "step": 23102
-    },
-    {
-      "epoch": 12.048500651890482,
-      "grad_norm": 1.5958157777786255,
-      "learning_rate": 7.729145728643217e-05,
-      "loss": 5.1405,
-      "step": 23103
-    },
-    {
-      "epoch": 12.049022164276401,
-      "grad_norm": 1.512058138847351,
-      "learning_rate": 7.729045226130653e-05,
-      "loss": 5.255,
-      "step": 23104
-    },
-    {
-      "epoch": 12.04954367666232,
-      "grad_norm": 1.442771315574646,
-      "learning_rate": 7.728944723618091e-05,
-      "loss": 5.4573,
-      "step": 23105
-    },
-    {
-      "epoch": 12.05006518904824,
-      "grad_norm": 1.5278420448303223,
-      "learning_rate": 7.728844221105527e-05,
-      "loss": 5.6034,
-      "step": 23106
-    },
-    {
-      "epoch": 12.05058670143416,
-      "grad_norm": 1.4150595664978027,
-      "learning_rate": 7.728743718592965e-05,
-      "loss": 5.5045,
-      "step": 23107
-    },
-    {
-      "epoch": 12.051108213820077,
-      "grad_norm": 1.6483243703842163,
-      "learning_rate": 7.728643216080403e-05,
-      "loss": 5.4618,
-      "step": 23108
-    },
-    {
-      "epoch": 12.051629726205997,
-      "grad_norm": 1.5049381256103516,
-      "learning_rate": 7.72854271356784e-05,
-      "loss": 5.5043,
-      "step": 23109
-    },
-    {
-      "epoch": 12.052151238591916,
-      "grad_norm": 1.5205689668655396,
-      "learning_rate": 7.728442211055277e-05,
-      "loss": 5.1002,
-      "step": 23110
-    },
-    {
-      "epoch": 12.052672750977836,
-      "grad_norm": 1.4858235120773315,
-      "learning_rate": 7.728341708542713e-05,
-      "loss": 5.9753,
-      "step": 23111
-    },
-    {
-      "epoch": 12.053194263363755,
-      "grad_norm": 1.3633630275726318,
-      "learning_rate": 7.728241206030151e-05,
-      "loss": 5.7302,
-      "step": 23112
-    },
-    {
-      "epoch": 12.053715775749675,
-      "grad_norm": 1.4039028882980347,
-      "learning_rate": 7.728140703517588e-05,
-      "loss": 5.5766,
-      "step": 23113
-    },
-    {
-      "epoch": 12.054237288135592,
-      "grad_norm": 1.515754222869873,
-      "learning_rate": 7.728040201005025e-05,
-      "loss": 5.1773,
-      "step": 23114
-    },
-    {
-      "epoch": 12.054758800521512,
-      "grad_norm": 1.4896279573440552,
-      "learning_rate": 7.727939698492462e-05,
-      "loss": 5.088,
-      "step": 23115
-    },
-    {
-      "epoch": 12.055280312907431,
-      "grad_norm": 1.4344727993011475,
-      "learning_rate": 7.7278391959799e-05,
-      "loss": 5.3875,
-      "step": 23116
-    },
-    {
-      "epoch": 12.05580182529335,
-      "grad_norm": 1.4616960287094116,
-      "learning_rate": 7.727738693467337e-05,
-      "loss": 5.495,
-      "step": 23117
-    },
-    {
-      "epoch": 12.05632333767927,
-      "grad_norm": 1.4395264387130737,
-      "learning_rate": 7.727638190954775e-05,
-      "loss": 5.5669,
-      "step": 23118
-    },
-    {
-      "epoch": 12.05684485006519,
-      "grad_norm": 1.3966515064239502,
-      "learning_rate": 7.727537688442212e-05,
-      "loss": 5.7163,
-      "step": 23119
-    },
-    {
-      "epoch": 12.057366362451107,
-      "grad_norm": 1.3422608375549316,
-      "learning_rate": 7.72743718592965e-05,
-      "loss": 5.6116,
-      "step": 23120
-    },
-    {
-      "epoch": 12.057887874837027,
-      "grad_norm": 1.4429709911346436,
-      "learning_rate": 7.727336683417086e-05,
-      "loss": 5.0624,
-      "step": 23121
-    },
-    {
-      "epoch": 12.058409387222946,
-      "grad_norm": 1.4024609327316284,
-      "learning_rate": 7.727236180904524e-05,
-      "loss": 5.5866,
-      "step": 23122
-    },
-    {
-      "epoch": 12.058930899608866,
-      "grad_norm": 1.3678864240646362,
-      "learning_rate": 7.72713567839196e-05,
-      "loss": 5.4809,
-      "step": 23123
-    },
-    {
-      "epoch": 12.059452411994785,
-      "grad_norm": 1.4872329235076904,
-      "learning_rate": 7.727035175879396e-05,
-      "loss": 4.7608,
-      "step": 23124
-    },
-    {
-      "epoch": 12.059973924380705,
-      "grad_norm": 1.5194522142410278,
-      "learning_rate": 7.726934673366834e-05,
-      "loss": 5.224,
-      "step": 23125
-    },
-    {
-      "epoch": 12.060495436766622,
-      "grad_norm": 1.6315858364105225,
-      "learning_rate": 7.726834170854271e-05,
-      "loss": 5.5373,
-      "step": 23126
-    },
-    {
-      "epoch": 12.061016949152542,
-      "grad_norm": 1.4971426725387573,
-      "learning_rate": 7.726733668341708e-05,
-      "loss": 4.9076,
-      "step": 23127
-    },
-    {
-      "epoch": 12.061538461538461,
-      "grad_norm": 1.614307165145874,
-      "learning_rate": 7.726633165829146e-05,
-      "loss": 5.7007,
-      "step": 23128
-    },
-    {
-      "epoch": 12.06205997392438,
-      "grad_norm": 1.38572096824646,
-      "learning_rate": 7.726532663316584e-05,
-      "loss": 5.5066,
-      "step": 23129
-    },
-    {
-      "epoch": 12.0625814863103,
-      "grad_norm": 1.4711697101593018,
-      "learning_rate": 7.72643216080402e-05,
-      "loss": 5.3969,
-      "step": 23130
-    },
-    {
-      "epoch": 12.06310299869622,
-      "grad_norm": 1.46401846408844,
-      "learning_rate": 7.726331658291458e-05,
-      "loss": 5.3729,
-      "step": 23131
-    },
-    {
-      "epoch": 12.063624511082137,
-      "grad_norm": 1.5941005945205688,
-      "learning_rate": 7.726231155778895e-05,
-      "loss": 5.4361,
-      "step": 23132
-    },
-    {
-      "epoch": 12.064146023468057,
-      "grad_norm": 1.4965379238128662,
-      "learning_rate": 7.726130653266332e-05,
-      "loss": 5.0809,
-      "step": 23133
-    },
-    {
-      "epoch": 12.064667535853976,
-      "grad_norm": 1.382506012916565,
-      "learning_rate": 7.726030150753769e-05,
-      "loss": 5.624,
-      "step": 23134
-    },
-    {
-      "epoch": 12.065189048239896,
-      "grad_norm": 1.5484803915023804,
-      "learning_rate": 7.725929648241207e-05,
-      "loss": 5.3123,
-      "step": 23135
-    },
-    {
-      "epoch": 12.065710560625815,
-      "grad_norm": 1.463524580001831,
-      "learning_rate": 7.725829145728643e-05,
-      "loss": 5.2147,
-      "step": 23136
-    },
-    {
-      "epoch": 12.066232073011735,
-      "grad_norm": 1.3822216987609863,
-      "learning_rate": 7.725728643216081e-05,
-      "loss": 5.6799,
-      "step": 23137
-    },
-    {
-      "epoch": 12.066753585397652,
-      "grad_norm": 1.5488626956939697,
-      "learning_rate": 7.725628140703519e-05,
-      "loss": 5.2549,
-      "step": 23138
-    },
-    {
-      "epoch": 12.067275097783572,
-      "grad_norm": 1.392382264137268,
-      "learning_rate": 7.725527638190955e-05,
-      "loss": 5.6464,
-      "step": 23139
-    },
-    {
-      "epoch": 12.067796610169491,
-      "grad_norm": 1.5545934438705444,
-      "learning_rate": 7.725427135678393e-05,
-      "loss": 5.5901,
-      "step": 23140
-    },
-    {
-      "epoch": 12.06831812255541,
-      "grad_norm": 1.5810672044754028,
-      "learning_rate": 7.725326633165829e-05,
-      "loss": 5.0754,
-      "step": 23141
-    },
-    {
-      "epoch": 12.06883963494133,
-      "grad_norm": 1.479113221168518,
-      "learning_rate": 7.725226130653267e-05,
-      "loss": 5.594,
-      "step": 23142
-    },
-    {
-      "epoch": 12.06936114732725,
-      "grad_norm": 1.5307499170303345,
-      "learning_rate": 7.725125628140704e-05,
-      "loss": 5.7683,
-      "step": 23143
-    },
-    {
-      "epoch": 12.069882659713167,
-      "grad_norm": 1.5482925176620483,
-      "learning_rate": 7.725025125628141e-05,
-      "loss": 5.3627,
-      "step": 23144
-    },
-    {
-      "epoch": 12.070404172099087,
-      "grad_norm": 1.4254382848739624,
-      "learning_rate": 7.724924623115578e-05,
-      "loss": 5.5368,
-      "step": 23145
-    },
-    {
-      "epoch": 12.070925684485006,
-      "grad_norm": 1.354294776916504,
-      "learning_rate": 7.724824120603016e-05,
-      "loss": 5.8986,
-      "step": 23146
-    },
-    {
-      "epoch": 12.071447196870926,
-      "grad_norm": 1.4707800149917603,
-      "learning_rate": 7.724723618090452e-05,
-      "loss": 5.4867,
-      "step": 23147
-    },
-    {
-      "epoch": 12.071968709256845,
-      "grad_norm": 1.5186947584152222,
-      "learning_rate": 7.72462311557789e-05,
-      "loss": 5.3528,
-      "step": 23148
-    },
-    {
-      "epoch": 12.072490221642765,
-      "grad_norm": 1.5172853469848633,
-      "learning_rate": 7.724522613065328e-05,
-      "loss": 5.2225,
-      "step": 23149
-    },
-    {
-      "epoch": 12.073011734028682,
-      "grad_norm": 1.6064374446868896,
-      "learning_rate": 7.724422110552764e-05,
-      "loss": 5.3446,
-      "step": 23150
-    },
-    {
-      "epoch": 12.073533246414602,
-      "grad_norm": 1.401328206062317,
-      "learning_rate": 7.724321608040202e-05,
-      "loss": 5.7057,
-      "step": 23151
-    },
-    {
-      "epoch": 12.074054758800521,
-      "grad_norm": 1.5704351663589478,
-      "learning_rate": 7.724221105527638e-05,
-      "loss": 5.1928,
-      "step": 23152
-    },
-    {
-      "epoch": 12.07457627118644,
-      "grad_norm": 1.5500965118408203,
-      "learning_rate": 7.724120603015076e-05,
-      "loss": 4.631,
-      "step": 23153
-    },
-    {
-      "epoch": 12.07509778357236,
-      "grad_norm": 1.4541308879852295,
-      "learning_rate": 7.724020100502512e-05,
-      "loss": 5.5314,
-      "step": 23154
-    },
-    {
-      "epoch": 12.07561929595828,
-      "grad_norm": 1.3724642992019653,
-      "learning_rate": 7.72391959798995e-05,
-      "loss": 5.7501,
-      "step": 23155
-    },
-    {
-      "epoch": 12.076140808344197,
-      "grad_norm": 1.4142793416976929,
-      "learning_rate": 7.723819095477387e-05,
-      "loss": 5.2688,
-      "step": 23156
-    },
-    {
-      "epoch": 12.076662320730117,
-      "grad_norm": 1.4733211994171143,
-      "learning_rate": 7.723718592964824e-05,
-      "loss": 5.0826,
-      "step": 23157
-    },
-    {
-      "epoch": 12.077183833116036,
-      "grad_norm": 1.4014768600463867,
-      "learning_rate": 7.723618090452262e-05,
-      "loss": 5.4662,
-      "step": 23158
-    },
-    {
-      "epoch": 12.077705345501956,
-      "grad_norm": 1.4360324144363403,
-      "learning_rate": 7.7235175879397e-05,
-      "loss": 5.4341,
-      "step": 23159
-    },
-    {
-      "epoch": 12.078226857887875,
-      "grad_norm": 1.4753880500793457,
-      "learning_rate": 7.723417085427136e-05,
-      "loss": 5.6949,
-      "step": 23160
-    },
-    {
-      "epoch": 12.078748370273795,
-      "grad_norm": 1.476901650428772,
-      "learning_rate": 7.723316582914574e-05,
-      "loss": 5.259,
-      "step": 23161
-    },
-    {
-      "epoch": 12.079269882659712,
-      "grad_norm": 1.4592376947402954,
-      "learning_rate": 7.72321608040201e-05,
-      "loss": 5.1868,
-      "step": 23162
-    },
-    {
-      "epoch": 12.079791395045632,
-      "grad_norm": 1.4176607131958008,
-      "learning_rate": 7.723115577889447e-05,
-      "loss": 5.6763,
-      "step": 23163
-    },
-    {
-      "epoch": 12.080312907431551,
-      "grad_norm": 1.394101619720459,
-      "learning_rate": 7.723015075376885e-05,
-      "loss": 5.428,
-      "step": 23164
-    },
-    {
-      "epoch": 12.08083441981747,
-      "grad_norm": 1.5066335201263428,
-      "learning_rate": 7.722914572864321e-05,
-      "loss": 5.2668,
-      "step": 23165
-    },
-    {
-      "epoch": 12.08135593220339,
-      "grad_norm": 1.4272500276565552,
-      "learning_rate": 7.722814070351759e-05,
-      "loss": 5.6717,
-      "step": 23166
-    },
-    {
-      "epoch": 12.08187744458931,
-      "grad_norm": 1.4151893854141235,
-      "learning_rate": 7.722713567839195e-05,
-      "loss": 5.3976,
-      "step": 23167
-    },
-    {
-      "epoch": 12.082398956975227,
-      "grad_norm": 1.3158751726150513,
-      "learning_rate": 7.722613065326633e-05,
-      "loss": 5.5846,
-      "step": 23168
-    },
-    {
-      "epoch": 12.082920469361147,
-      "grad_norm": 1.4716885089874268,
-      "learning_rate": 7.722512562814071e-05,
-      "loss": 5.342,
-      "step": 23169
-    },
-    {
-      "epoch": 12.083441981747066,
-      "grad_norm": 1.4621703624725342,
-      "learning_rate": 7.722412060301509e-05,
-      "loss": 5.5367,
-      "step": 23170
-    },
-    {
-      "epoch": 12.083963494132986,
-      "grad_norm": 1.3957018852233887,
-      "learning_rate": 7.722311557788945e-05,
-      "loss": 5.58,
-      "step": 23171
-    },
-    {
-      "epoch": 12.084485006518905,
-      "grad_norm": 1.489406943321228,
-      "learning_rate": 7.722211055276383e-05,
-      "loss": 5.8162,
-      "step": 23172
-    },
-    {
-      "epoch": 12.085006518904825,
-      "grad_norm": 1.4156359434127808,
-      "learning_rate": 7.72211055276382e-05,
-      "loss": 5.3326,
-      "step": 23173
-    },
-    {
-      "epoch": 12.085528031290742,
-      "grad_norm": 1.4730842113494873,
-      "learning_rate": 7.722010050251257e-05,
-      "loss": 5.8026,
-      "step": 23174
-    },
-    {
-      "epoch": 12.086049543676662,
-      "grad_norm": 1.493222951889038,
-      "learning_rate": 7.721909547738694e-05,
-      "loss": 5.2776,
-      "step": 23175
-    },
-    {
-      "epoch": 12.086571056062581,
-      "grad_norm": 1.4378817081451416,
-      "learning_rate": 7.721809045226131e-05,
-      "loss": 5.373,
-      "step": 23176
-    },
-    {
-      "epoch": 12.0870925684485,
-      "grad_norm": 1.4092351198196411,
-      "learning_rate": 7.721708542713568e-05,
-      "loss": 5.3503,
-      "step": 23177
-    },
-    {
-      "epoch": 12.08761408083442,
-      "grad_norm": 1.4170786142349243,
-      "learning_rate": 7.721608040201006e-05,
-      "loss": 5.2392,
-      "step": 23178
-    },
-    {
-      "epoch": 12.08813559322034,
-      "grad_norm": 1.6509698629379272,
-      "learning_rate": 7.721507537688443e-05,
-      "loss": 4.934,
-      "step": 23179
-    },
-    {
-      "epoch": 12.088657105606258,
-      "grad_norm": 1.4109649658203125,
-      "learning_rate": 7.72140703517588e-05,
-      "loss": 5.5416,
-      "step": 23180
-    },
-    {
-      "epoch": 12.089178617992177,
-      "grad_norm": 1.4339953660964966,
-      "learning_rate": 7.721306532663318e-05,
-      "loss": 4.9824,
-      "step": 23181
-    },
-    {
-      "epoch": 12.089700130378096,
-      "grad_norm": 1.3650174140930176,
-      "learning_rate": 7.721206030150754e-05,
-      "loss": 5.4538,
-      "step": 23182
-    },
-    {
-      "epoch": 12.090221642764016,
-      "grad_norm": 1.413261890411377,
-      "learning_rate": 7.721105527638192e-05,
-      "loss": 5.2586,
-      "step": 23183
-    },
-    {
-      "epoch": 12.090743155149935,
-      "grad_norm": 1.6468067169189453,
-      "learning_rate": 7.721005025125628e-05,
-      "loss": 5.0641,
-      "step": 23184
-    },
-    {
-      "epoch": 12.091264667535855,
-      "grad_norm": 1.465459942817688,
-      "learning_rate": 7.720904522613066e-05,
-      "loss": 5.508,
-      "step": 23185
-    },
-    {
-      "epoch": 12.091786179921773,
-      "grad_norm": 1.581454873085022,
-      "learning_rate": 7.720804020100502e-05,
-      "loss": 5.1751,
-      "step": 23186
-    },
-    {
-      "epoch": 12.092307692307692,
-      "grad_norm": 1.4704396724700928,
-      "learning_rate": 7.72070351758794e-05,
-      "loss": 5.603,
-      "step": 23187
-    },
-    {
-      "epoch": 12.092829204693611,
-      "grad_norm": 1.613256812095642,
-      "learning_rate": 7.720603015075377e-05,
-      "loss": 5.3642,
-      "step": 23188
-    },
-    {
-      "epoch": 12.093350717079531,
-      "grad_norm": 1.4250587224960327,
-      "learning_rate": 7.720502512562814e-05,
-      "loss": 5.7103,
-      "step": 23189
-    },
-    {
-      "epoch": 12.09387222946545,
-      "grad_norm": 1.4374372959136963,
-      "learning_rate": 7.720402010050252e-05,
-      "loss": 5.4226,
-      "step": 23190
-    },
-    {
-      "epoch": 12.09439374185137,
-      "grad_norm": 1.4899542331695557,
-      "learning_rate": 7.720301507537689e-05,
-      "loss": 5.2381,
-      "step": 23191
-    },
-    {
-      "epoch": 12.094915254237288,
-      "grad_norm": 1.4355809688568115,
-      "learning_rate": 7.720201005025126e-05,
-      "loss": 5.1474,
-      "step": 23192
-    },
-    {
-      "epoch": 12.095436766623207,
-      "grad_norm": 1.558153748512268,
-      "learning_rate": 7.720100502512563e-05,
-      "loss": 4.6743,
-      "step": 23193
-    },
-    {
-      "epoch": 12.095958279009126,
-      "grad_norm": 1.45798921585083,
-      "learning_rate": 7.72e-05,
-      "loss": 4.8972,
-      "step": 23194
-    },
-    {
-      "epoch": 12.096479791395046,
-      "grad_norm": 1.4685579538345337,
-      "learning_rate": 7.719899497487437e-05,
-      "loss": 5.2735,
-      "step": 23195
-    },
-    {
-      "epoch": 12.097001303780965,
-      "grad_norm": 1.6373865604400635,
-      "learning_rate": 7.719798994974875e-05,
-      "loss": 5.5401,
-      "step": 23196
-    },
-    {
-      "epoch": 12.097522816166883,
-      "grad_norm": 1.5046583414077759,
-      "learning_rate": 7.719698492462311e-05,
-      "loss": 5.3829,
-      "step": 23197
-    },
-    {
-      "epoch": 12.098044328552803,
-      "grad_norm": 1.4812555313110352,
-      "learning_rate": 7.719597989949749e-05,
-      "loss": 5.3505,
-      "step": 23198
-    },
-    {
-      "epoch": 12.098565840938722,
-      "grad_norm": 1.3717961311340332,
-      "learning_rate": 7.719497487437185e-05,
-      "loss": 5.6134,
-      "step": 23199
-    },
-    {
-      "epoch": 12.099087353324641,
-      "grad_norm": 1.6426846981048584,
-      "learning_rate": 7.719396984924623e-05,
-      "loss": 5.4598,
-      "step": 23200
-    },
-    {
-      "epoch": 12.099608865710561,
-      "grad_norm": 1.518883228302002,
-      "learning_rate": 7.719296482412061e-05,
-      "loss": 5.2238,
-      "step": 23201
-    },
-    {
-      "epoch": 12.10013037809648,
-      "grad_norm": 1.5520178079605103,
-      "learning_rate": 7.719195979899499e-05,
-      "loss": 5.2854,
-      "step": 23202
-    },
-    {
-      "epoch": 12.100651890482398,
-      "grad_norm": 1.4103515148162842,
-      "learning_rate": 7.719095477386935e-05,
-      "loss": 5.4528,
-      "step": 23203
-    },
-    {
-      "epoch": 12.101173402868318,
-      "grad_norm": 1.4383925199508667,
-      "learning_rate": 7.718994974874372e-05,
-      "loss": 5.8024,
-      "step": 23204
-    },
-    {
-      "epoch": 12.101694915254237,
-      "grad_norm": 1.559143304824829,
-      "learning_rate": 7.71889447236181e-05,
-      "loss": 5.312,
-      "step": 23205
-    },
-    {
-      "epoch": 12.102216427640156,
-      "grad_norm": 1.3964194059371948,
-      "learning_rate": 7.718793969849246e-05,
-      "loss": 5.2673,
-      "step": 23206
-    },
-    {
-      "epoch": 12.102737940026076,
-      "grad_norm": 1.4628323316574097,
-      "learning_rate": 7.718693467336684e-05,
-      "loss": 5.1973,
-      "step": 23207
-    },
-    {
-      "epoch": 12.103259452411995,
-      "grad_norm": 1.4317137002944946,
-      "learning_rate": 7.71859296482412e-05,
-      "loss": 5.4461,
-      "step": 23208
-    },
-    {
-      "epoch": 12.103780964797913,
-      "grad_norm": 1.471629023551941,
-      "learning_rate": 7.718492462311558e-05,
-      "loss": 5.5767,
-      "step": 23209
-    },
-    {
-      "epoch": 12.104302477183833,
-      "grad_norm": 1.4211130142211914,
-      "learning_rate": 7.718391959798996e-05,
-      "loss": 5.1819,
-      "step": 23210
-    },
-    {
-      "epoch": 12.104823989569752,
-      "grad_norm": 1.6804918050765991,
-      "learning_rate": 7.718291457286433e-05,
-      "loss": 5.7377,
-      "step": 23211
-    },
-    {
-      "epoch": 12.105345501955671,
-      "grad_norm": 1.6533293724060059,
-      "learning_rate": 7.71819095477387e-05,
-      "loss": 4.7257,
-      "step": 23212
-    },
-    {
-      "epoch": 12.105867014341591,
-      "grad_norm": 1.4809714555740356,
-      "learning_rate": 7.718090452261308e-05,
-      "loss": 5.6189,
-      "step": 23213
-    },
-    {
-      "epoch": 12.10638852672751,
-      "grad_norm": 1.6182283163070679,
-      "learning_rate": 7.717989949748744e-05,
-      "loss": 5.1938,
-      "step": 23214
-    },
-    {
-      "epoch": 12.106910039113428,
-      "grad_norm": 1.5200282335281372,
-      "learning_rate": 7.717889447236182e-05,
-      "loss": 5.5414,
-      "step": 23215
-    },
-    {
-      "epoch": 12.107431551499348,
-      "grad_norm": 1.4282560348510742,
-      "learning_rate": 7.717788944723618e-05,
-      "loss": 5.4383,
-      "step": 23216
-    },
-    {
-      "epoch": 12.107953063885267,
-      "grad_norm": 1.5305250883102417,
-      "learning_rate": 7.717688442211055e-05,
-      "loss": 4.7738,
-      "step": 23217
-    },
-    {
-      "epoch": 12.108474576271187,
-      "grad_norm": 1.5039963722229004,
-      "learning_rate": 7.717587939698493e-05,
-      "loss": 5.5709,
-      "step": 23218
-    },
-    {
-      "epoch": 12.108996088657106,
-      "grad_norm": 1.4642914533615112,
-      "learning_rate": 7.717487437185929e-05,
-      "loss": 5.6915,
-      "step": 23219
-    },
-    {
-      "epoch": 12.109517601043025,
-      "grad_norm": 1.437987208366394,
-      "learning_rate": 7.717386934673367e-05,
-      "loss": 5.4001,
-      "step": 23220
-    },
-    {
-      "epoch": 12.110039113428943,
-      "grad_norm": 1.4373502731323242,
-      "learning_rate": 7.717286432160804e-05,
-      "loss": 5.3281,
-      "step": 23221
-    },
-    {
-      "epoch": 12.110560625814863,
-      "grad_norm": 1.3864829540252686,
-      "learning_rate": 7.717185929648242e-05,
-      "loss": 5.2867,
-      "step": 23222
-    },
-    {
-      "epoch": 12.111082138200782,
-      "grad_norm": 1.4542267322540283,
-      "learning_rate": 7.717085427135679e-05,
-      "loss": 5.5894,
-      "step": 23223
-    },
-    {
-      "epoch": 12.111603650586702,
-      "grad_norm": 1.5707799196243286,
-      "learning_rate": 7.716984924623116e-05,
-      "loss": 5.1944,
-      "step": 23224
-    },
-    {
-      "epoch": 12.112125162972621,
-      "grad_norm": 1.5086551904678345,
-      "learning_rate": 7.716884422110553e-05,
-      "loss": 5.5396,
-      "step": 23225
-    },
-    {
-      "epoch": 12.11264667535854,
-      "grad_norm": 1.4224663972854614,
-      "learning_rate": 7.716783919597991e-05,
-      "loss": 5.4392,
-      "step": 23226
-    },
-    {
-      "epoch": 12.113168187744458,
-      "grad_norm": 1.4865385293960571,
-      "learning_rate": 7.716683417085427e-05,
-      "loss": 5.5434,
-      "step": 23227
-    },
-    {
-      "epoch": 12.113689700130378,
-      "grad_norm": 1.3933360576629639,
-      "learning_rate": 7.716582914572865e-05,
-      "loss": 5.5136,
-      "step": 23228
-    },
-    {
-      "epoch": 12.114211212516297,
-      "grad_norm": 1.562211513519287,
-      "learning_rate": 7.716482412060301e-05,
-      "loss": 4.8786,
-      "step": 23229
-    },
-    {
-      "epoch": 12.114732724902217,
-      "grad_norm": 1.6270502805709839,
-      "learning_rate": 7.716381909547739e-05,
-      "loss": 5.8043,
-      "step": 23230
-    },
-    {
-      "epoch": 12.115254237288136,
-      "grad_norm": 1.5078649520874023,
-      "learning_rate": 7.716281407035177e-05,
-      "loss": 5.1878,
-      "step": 23231
-    },
-    {
-      "epoch": 12.115775749674055,
-      "grad_norm": 1.5005996227264404,
-      "learning_rate": 7.716180904522613e-05,
-      "loss": 5.4168,
-      "step": 23232
-    },
-    {
-      "epoch": 12.116297262059973,
-      "grad_norm": 1.5568784475326538,
-      "learning_rate": 7.716080402010051e-05,
-      "loss": 5.3491,
-      "step": 23233
-    },
-    {
-      "epoch": 12.116818774445893,
-      "grad_norm": 1.3571484088897705,
-      "learning_rate": 7.715979899497488e-05,
-      "loss": 5.5368,
-      "step": 23234
-    },
-    {
-      "epoch": 12.117340286831812,
-      "grad_norm": 1.4131619930267334,
-      "learning_rate": 7.715879396984925e-05,
-      "loss": 5.3988,
-      "step": 23235
-    },
-    {
-      "epoch": 12.117861799217732,
-      "grad_norm": 1.3802992105484009,
-      "learning_rate": 7.715778894472362e-05,
-      "loss": 5.6483,
-      "step": 23236
-    },
-    {
-      "epoch": 12.118383311603651,
-      "grad_norm": 1.5045723915100098,
-      "learning_rate": 7.7156783919598e-05,
-      "loss": 5.44,
-      "step": 23237
-    },
-    {
-      "epoch": 12.11890482398957,
-      "grad_norm": 1.3542206287384033,
-      "learning_rate": 7.715577889447236e-05,
-      "loss": 5.4072,
-      "step": 23238
-    },
-    {
-      "epoch": 12.119426336375488,
-      "grad_norm": 1.4684351682662964,
-      "learning_rate": 7.715477386934674e-05,
-      "loss": 5.6189,
-      "step": 23239
-    },
-    {
-      "epoch": 12.119947848761408,
-      "grad_norm": 1.3552581071853638,
-      "learning_rate": 7.71537688442211e-05,
-      "loss": 5.8418,
-      "step": 23240
-    },
-    {
-      "epoch": 12.120469361147327,
-      "grad_norm": 1.3855769634246826,
-      "learning_rate": 7.715276381909548e-05,
-      "loss": 5.5141,
-      "step": 23241
-    },
-    {
-      "epoch": 12.120990873533247,
-      "grad_norm": 1.3863250017166138,
-      "learning_rate": 7.715175879396986e-05,
-      "loss": 5.7289,
-      "step": 23242
-    },
-    {
-      "epoch": 12.121512385919166,
-      "grad_norm": 1.4180080890655518,
-      "learning_rate": 7.715075376884422e-05,
-      "loss": 5.2798,
-      "step": 23243
-    },
-    {
-      "epoch": 12.122033898305085,
-      "grad_norm": 1.3867744207382202,
-      "learning_rate": 7.71497487437186e-05,
-      "loss": 5.1754,
-      "step": 23244
-    },
-    {
-      "epoch": 12.122555410691003,
-      "grad_norm": 1.4599356651306152,
-      "learning_rate": 7.714874371859296e-05,
-      "loss": 5.4251,
-      "step": 23245
-    },
-    {
-      "epoch": 12.123076923076923,
-      "grad_norm": 1.4853521585464478,
-      "learning_rate": 7.714773869346734e-05,
-      "loss": 4.8818,
-      "step": 23246
-    },
-    {
-      "epoch": 12.123598435462842,
-      "grad_norm": 1.4472476243972778,
-      "learning_rate": 7.71467336683417e-05,
-      "loss": 6.0063,
-      "step": 23247
-    },
-    {
-      "epoch": 12.124119947848762,
-      "grad_norm": 1.4048186540603638,
-      "learning_rate": 7.714572864321608e-05,
-      "loss": 5.1808,
-      "step": 23248
-    },
-    {
-      "epoch": 12.124641460234681,
-      "grad_norm": 1.510714054107666,
-      "learning_rate": 7.714472361809045e-05,
-      "loss": 5.553,
-      "step": 23249
-    },
-    {
-      "epoch": 12.1251629726206,
-      "grad_norm": 1.4529436826705933,
-      "learning_rate": 7.714371859296483e-05,
-      "loss": 5.5134,
-      "step": 23250
-    },
-    {
-      "epoch": 12.125684485006518,
-      "grad_norm": 1.5888347625732422,
-      "learning_rate": 7.71427135678392e-05,
-      "loss": 5.0681,
-      "step": 23251
-    },
-    {
-      "epoch": 12.126205997392438,
-      "grad_norm": 1.5713202953338623,
-      "learning_rate": 7.714170854271358e-05,
-      "loss": 5.0697,
-      "step": 23252
-    },
-    {
-      "epoch": 12.126727509778357,
-      "grad_norm": 1.456058144569397,
-      "learning_rate": 7.714070351758795e-05,
-      "loss": 5.3732,
-      "step": 23253
-    },
-    {
-      "epoch": 12.127249022164277,
-      "grad_norm": 1.3542330265045166,
-      "learning_rate": 7.713969849246232e-05,
-      "loss": 5.4632,
-      "step": 23254
-    },
-    {
-      "epoch": 12.127770534550196,
-      "grad_norm": 1.4905760288238525,
-      "learning_rate": 7.713869346733669e-05,
-      "loss": 5.2779,
-      "step": 23255
-    },
-    {
-      "epoch": 12.128292046936116,
-      "grad_norm": 1.505298376083374,
-      "learning_rate": 7.713768844221105e-05,
-      "loss": 4.6976,
-      "step": 23256
-    },
-    {
-      "epoch": 12.128813559322033,
-      "grad_norm": 1.4407790899276733,
-      "learning_rate": 7.713668341708543e-05,
-      "loss": 5.0925,
-      "step": 23257
-    },
-    {
-      "epoch": 12.129335071707953,
-      "grad_norm": 1.5303047895431519,
-      "learning_rate": 7.71356783919598e-05,
-      "loss": 5.2716,
-      "step": 23258
-    },
-    {
-      "epoch": 12.129856584093872,
-      "grad_norm": 1.4792834520339966,
-      "learning_rate": 7.713467336683417e-05,
-      "loss": 5.2168,
-      "step": 23259
-    },
-    {
-      "epoch": 12.130378096479792,
-      "grad_norm": 1.4654704332351685,
-      "learning_rate": 7.713366834170854e-05,
-      "loss": 5.3315,
-      "step": 23260
-    },
-    {
-      "epoch": 12.130899608865711,
-      "grad_norm": 1.4507076740264893,
-      "learning_rate": 7.713266331658291e-05,
-      "loss": 5.5352,
-      "step": 23261
-    },
-    {
-      "epoch": 12.13142112125163,
-      "grad_norm": 1.5841705799102783,
-      "learning_rate": 7.713165829145729e-05,
-      "loss": 5.4442,
-      "step": 23262
-    },
-    {
-      "epoch": 12.131942633637548,
-      "grad_norm": 1.5021167993545532,
-      "learning_rate": 7.713065326633167e-05,
-      "loss": 5.5044,
-      "step": 23263
-    },
-    {
-      "epoch": 12.132464146023468,
-      "grad_norm": 1.4707633256912231,
-      "learning_rate": 7.712964824120603e-05,
-      "loss": 5.4177,
-      "step": 23264
-    },
-    {
-      "epoch": 12.132985658409387,
-      "grad_norm": 1.4848772287368774,
-      "learning_rate": 7.712864321608041e-05,
-      "loss": 5.1925,
-      "step": 23265
-    },
-    {
-      "epoch": 12.133507170795307,
-      "grad_norm": 1.356120228767395,
-      "learning_rate": 7.712763819095478e-05,
-      "loss": 5.7198,
-      "step": 23266
-    },
-    {
-      "epoch": 12.134028683181226,
-      "grad_norm": 1.424798846244812,
-      "learning_rate": 7.712663316582915e-05,
-      "loss": 5.5113,
-      "step": 23267
-    },
-    {
-      "epoch": 12.134550195567146,
-      "grad_norm": 1.4528850317001343,
-      "learning_rate": 7.712562814070352e-05,
-      "loss": 5.2151,
-      "step": 23268
-    },
-    {
-      "epoch": 12.135071707953063,
-      "grad_norm": 1.3560599088668823,
-      "learning_rate": 7.71246231155779e-05,
-      "loss": 5.5187,
-      "step": 23269
-    },
-    {
-      "epoch": 12.135593220338983,
-      "grad_norm": 1.4713330268859863,
-      "learning_rate": 7.712361809045226e-05,
-      "loss": 5.2389,
-      "step": 23270
-    },
-    {
-      "epoch": 12.136114732724902,
-      "grad_norm": 1.4393237829208374,
-      "learning_rate": 7.712261306532664e-05,
-      "loss": 5.1559,
-      "step": 23271
-    },
-    {
-      "epoch": 12.136636245110822,
-      "grad_norm": 1.3506579399108887,
-      "learning_rate": 7.712160804020102e-05,
-      "loss": 5.6588,
-      "step": 23272
-    },
-    {
-      "epoch": 12.137157757496741,
-      "grad_norm": 1.4694457054138184,
-      "learning_rate": 7.712060301507538e-05,
-      "loss": 5.572,
-      "step": 23273
-    },
-    {
-      "epoch": 12.13767926988266,
-      "grad_norm": 1.6166958808898926,
-      "learning_rate": 7.711959798994976e-05,
-      "loss": 4.7243,
-      "step": 23274
-    },
-    {
-      "epoch": 12.138200782268578,
-      "grad_norm": 1.451624870300293,
-      "learning_rate": 7.711859296482412e-05,
-      "loss": 5.3111,
-      "step": 23275
-    },
-    {
-      "epoch": 12.138722294654498,
-      "grad_norm": 1.3972625732421875,
-      "learning_rate": 7.71175879396985e-05,
-      "loss": 5.6035,
-      "step": 23276
-    },
-    {
-      "epoch": 12.139243807040417,
-      "grad_norm": 1.5093107223510742,
-      "learning_rate": 7.711658291457286e-05,
-      "loss": 5.2821,
-      "step": 23277
-    },
-    {
-      "epoch": 12.139765319426337,
-      "grad_norm": 1.4752861261367798,
-      "learning_rate": 7.711557788944724e-05,
-      "loss": 5.6475,
-      "step": 23278
-    },
-    {
-      "epoch": 12.140286831812256,
-      "grad_norm": 1.4938879013061523,
-      "learning_rate": 7.71145728643216e-05,
-      "loss": 5.5052,
-      "step": 23279
-    },
-    {
-      "epoch": 12.140808344198176,
-      "grad_norm": 1.5232247114181519,
-      "learning_rate": 7.711356783919598e-05,
-      "loss": 5.3171,
-      "step": 23280
-    },
-    {
-      "epoch": 12.141329856584093,
-      "grad_norm": 1.5013631582260132,
-      "learning_rate": 7.711256281407035e-05,
-      "loss": 5.5011,
-      "step": 23281
-    },
-    {
-      "epoch": 12.141851368970013,
-      "grad_norm": 1.537766695022583,
-      "learning_rate": 7.711155778894473e-05,
-      "loss": 5.4879,
-      "step": 23282
-    },
-    {
-      "epoch": 12.142372881355932,
-      "grad_norm": 1.5254372358322144,
-      "learning_rate": 7.71105527638191e-05,
-      "loss": 5.6191,
-      "step": 23283
-    },
-    {
-      "epoch": 12.142894393741852,
-      "grad_norm": 1.5185115337371826,
-      "learning_rate": 7.710954773869347e-05,
-      "loss": 5.3931,
-      "step": 23284
-    },
-    {
-      "epoch": 12.143415906127771,
-      "grad_norm": 1.497800350189209,
-      "learning_rate": 7.710854271356785e-05,
-      "loss": 5.3358,
-      "step": 23285
-    },
-    {
-      "epoch": 12.14393741851369,
-      "grad_norm": 1.5053385496139526,
-      "learning_rate": 7.710753768844221e-05,
-      "loss": 5.5635,
-      "step": 23286
-    },
-    {
-      "epoch": 12.144458930899608,
-      "grad_norm": 1.9365413188934326,
-      "learning_rate": 7.710653266331659e-05,
-      "loss": 5.2961,
-      "step": 23287
-    },
-    {
-      "epoch": 12.144980443285528,
-      "grad_norm": 1.5243555307388306,
-      "learning_rate": 7.710552763819095e-05,
-      "loss": 5.7222,
-      "step": 23288
-    },
-    {
-      "epoch": 12.145501955671447,
-      "grad_norm": 1.4120352268218994,
-      "learning_rate": 7.710452261306533e-05,
-      "loss": 5.7926,
-      "step": 23289
-    },
-    {
-      "epoch": 12.146023468057367,
-      "grad_norm": 1.4759808778762817,
-      "learning_rate": 7.71035175879397e-05,
-      "loss": 5.3006,
-      "step": 23290
-    },
-    {
-      "epoch": 12.146544980443286,
-      "grad_norm": 1.302177906036377,
-      "learning_rate": 7.710251256281407e-05,
-      "loss": 5.8449,
-      "step": 23291
-    },
-    {
-      "epoch": 12.147066492829204,
-      "grad_norm": 1.472990870475769,
-      "learning_rate": 7.710150753768845e-05,
-      "loss": 5.3852,
-      "step": 23292
-    },
-    {
-      "epoch": 12.147588005215123,
-      "grad_norm": 1.3896753787994385,
-      "learning_rate": 7.710050251256283e-05,
-      "loss": 5.4171,
-      "step": 23293
-    },
-    {
-      "epoch": 12.148109517601043,
-      "grad_norm": 1.5493955612182617,
-      "learning_rate": 7.709949748743719e-05,
-      "loss": 5.4883,
-      "step": 23294
-    },
-    {
-      "epoch": 12.148631029986962,
-      "grad_norm": 1.4997193813323975,
-      "learning_rate": 7.709849246231157e-05,
-      "loss": 5.2254,
-      "step": 23295
-    },
-    {
-      "epoch": 12.149152542372882,
-      "grad_norm": 1.6209088563919067,
-      "learning_rate": 7.709748743718593e-05,
-      "loss": 5.4005,
-      "step": 23296
-    },
-    {
-      "epoch": 12.149674054758801,
-      "grad_norm": 1.5070935487747192,
-      "learning_rate": 7.70964824120603e-05,
-      "loss": 5.3359,
-      "step": 23297
-    },
-    {
-      "epoch": 12.150195567144719,
-      "grad_norm": 1.5476658344268799,
-      "learning_rate": 7.709547738693468e-05,
-      "loss": 5.4895,
-      "step": 23298
-    },
-    {
-      "epoch": 12.150717079530638,
-      "grad_norm": 1.4580047130584717,
-      "learning_rate": 7.709447236180904e-05,
-      "loss": 5.4815,
-      "step": 23299
-    },
-    {
-      "epoch": 12.151238591916558,
-      "grad_norm": 1.4353770017623901,
-      "learning_rate": 7.709346733668342e-05,
-      "loss": 5.6993,
-      "step": 23300
-    },
-    {
-      "epoch": 12.151760104302477,
-      "grad_norm": 1.3868427276611328,
-      "learning_rate": 7.709246231155778e-05,
-      "loss": 5.4958,
-      "step": 23301
-    },
-    {
-      "epoch": 12.152281616688397,
-      "grad_norm": 1.3699768781661987,
-      "learning_rate": 7.709145728643216e-05,
-      "loss": 5.7146,
-      "step": 23302
-    },
-    {
-      "epoch": 12.152803129074316,
-      "grad_norm": 1.6336146593093872,
-      "learning_rate": 7.709045226130654e-05,
-      "loss": 5.2244,
-      "step": 23303
-    },
-    {
-      "epoch": 12.153324641460234,
-      "grad_norm": 1.5282599925994873,
-      "learning_rate": 7.708944723618092e-05,
-      "loss": 5.3147,
-      "step": 23304
-    },
-    {
-      "epoch": 12.153846153846153,
-      "grad_norm": 1.3613404035568237,
-      "learning_rate": 7.708844221105528e-05,
-      "loss": 5.3852,
-      "step": 23305
-    },
-    {
-      "epoch": 12.154367666232073,
-      "grad_norm": 1.5635039806365967,
-      "learning_rate": 7.708743718592966e-05,
-      "loss": 5.4757,
-      "step": 23306
-    },
-    {
-      "epoch": 12.154889178617992,
-      "grad_norm": 1.5353165864944458,
-      "learning_rate": 7.708643216080402e-05,
-      "loss": 5.3019,
-      "step": 23307
-    },
-    {
-      "epoch": 12.155410691003912,
-      "grad_norm": 1.4703949689865112,
-      "learning_rate": 7.70854271356784e-05,
-      "loss": 5.7228,
-      "step": 23308
-    },
-    {
-      "epoch": 12.155932203389831,
-      "grad_norm": 1.4320669174194336,
-      "learning_rate": 7.708442211055277e-05,
-      "loss": 5.4785,
-      "step": 23309
-    },
-    {
-      "epoch": 12.156453715775749,
-      "grad_norm": 1.3571455478668213,
-      "learning_rate": 7.708341708542713e-05,
-      "loss": 5.505,
-      "step": 23310
-    },
-    {
-      "epoch": 12.156975228161668,
-      "grad_norm": 1.6154870986938477,
-      "learning_rate": 7.708241206030151e-05,
-      "loss": 4.6275,
-      "step": 23311
-    },
-    {
-      "epoch": 12.157496740547588,
-      "grad_norm": 1.4904800653457642,
-      "learning_rate": 7.708140703517589e-05,
-      "loss": 4.9685,
-      "step": 23312
-    },
-    {
-      "epoch": 12.158018252933507,
-      "grad_norm": 1.6181299686431885,
-      "learning_rate": 7.708040201005026e-05,
-      "loss": 5.1812,
-      "step": 23313
-    },
-    {
-      "epoch": 12.158539765319427,
-      "grad_norm": 1.513921856880188,
-      "learning_rate": 7.707939698492463e-05,
-      "loss": 5.1277,
-      "step": 23314
-    },
-    {
-      "epoch": 12.159061277705346,
-      "grad_norm": 1.3423831462860107,
-      "learning_rate": 7.7078391959799e-05,
-      "loss": 6.0305,
-      "step": 23315
-    },
-    {
-      "epoch": 12.159582790091264,
-      "grad_norm": 1.4354511499404907,
-      "learning_rate": 7.707738693467337e-05,
-      "loss": 5.1426,
-      "step": 23316
-    },
-    {
-      "epoch": 12.160104302477183,
-      "grad_norm": 1.5266344547271729,
-      "learning_rate": 7.707638190954775e-05,
-      "loss": 4.8829,
-      "step": 23317
-    },
-    {
-      "epoch": 12.160625814863103,
-      "grad_norm": 1.5573450326919556,
-      "learning_rate": 7.707537688442211e-05,
-      "loss": 5.8431,
-      "step": 23318
-    },
-    {
-      "epoch": 12.161147327249022,
-      "grad_norm": 1.4455690383911133,
-      "learning_rate": 7.707437185929649e-05,
-      "loss": 5.3474,
-      "step": 23319
-    },
-    {
-      "epoch": 12.161668839634942,
-      "grad_norm": 1.4828075170516968,
-      "learning_rate": 7.707336683417085e-05,
-      "loss": 5.7715,
-      "step": 23320
-    },
-    {
-      "epoch": 12.162190352020861,
-      "grad_norm": 1.454125165939331,
-      "learning_rate": 7.707236180904523e-05,
-      "loss": 5.8353,
-      "step": 23321
-    },
-    {
-      "epoch": 12.162711864406779,
-      "grad_norm": 1.6863183975219727,
-      "learning_rate": 7.70713567839196e-05,
-      "loss": 5.0384,
-      "step": 23322
-    },
-    {
-      "epoch": 12.163233376792698,
-      "grad_norm": 1.481653094291687,
-      "learning_rate": 7.707035175879397e-05,
-      "loss": 5.3017,
-      "step": 23323
-    },
-    {
-      "epoch": 12.163754889178618,
-      "grad_norm": 1.5463052988052368,
-      "learning_rate": 7.706934673366835e-05,
-      "loss": 5.3041,
-      "step": 23324
-    },
-    {
-      "epoch": 12.164276401564537,
-      "grad_norm": 1.4869391918182373,
-      "learning_rate": 7.706834170854272e-05,
-      "loss": 5.3943,
-      "step": 23325
-    },
-    {
-      "epoch": 12.164797913950457,
-      "grad_norm": 1.5012106895446777,
-      "learning_rate": 7.70673366834171e-05,
-      "loss": 5.4998,
-      "step": 23326
-    },
-    {
-      "epoch": 12.165319426336376,
-      "grad_norm": 1.43594229221344,
-      "learning_rate": 7.706633165829146e-05,
-      "loss": 5.7417,
-      "step": 23327
-    },
-    {
-      "epoch": 12.165840938722294,
-      "grad_norm": 1.489554762840271,
-      "learning_rate": 7.706532663316584e-05,
-      "loss": 5.2817,
-      "step": 23328
-    },
-    {
-      "epoch": 12.166362451108213,
-      "grad_norm": 1.6523733139038086,
-      "learning_rate": 7.70643216080402e-05,
-      "loss": 5.1725,
-      "step": 23329
-    },
-    {
-      "epoch": 12.166883963494133,
-      "grad_norm": 1.5325548648834229,
-      "learning_rate": 7.706331658291458e-05,
-      "loss": 5.2822,
-      "step": 23330
-    },
-    {
-      "epoch": 12.167405475880052,
-      "grad_norm": 1.3747526407241821,
-      "learning_rate": 7.706231155778894e-05,
-      "loss": 5.493,
-      "step": 23331
-    },
-    {
-      "epoch": 12.167926988265972,
-      "grad_norm": 1.4299225807189941,
-      "learning_rate": 7.706130653266332e-05,
-      "loss": 5.2501,
-      "step": 23332
-    },
-    {
-      "epoch": 12.168448500651891,
-      "grad_norm": 1.4282739162445068,
-      "learning_rate": 7.70603015075377e-05,
-      "loss": 5.5737,
-      "step": 23333
-    },
-    {
-      "epoch": 12.168970013037809,
-      "grad_norm": 1.4036917686462402,
-      "learning_rate": 7.705929648241208e-05,
-      "loss": 5.6002,
-      "step": 23334
-    },
-    {
-      "epoch": 12.169491525423728,
-      "grad_norm": 1.515649437904358,
-      "learning_rate": 7.705829145728644e-05,
-      "loss": 5.3753,
-      "step": 23335
-    },
-    {
-      "epoch": 12.170013037809648,
-      "grad_norm": 1.4405518770217896,
-      "learning_rate": 7.70572864321608e-05,
-      "loss": 5.8509,
-      "step": 23336
-    },
-    {
-      "epoch": 12.170534550195567,
-      "grad_norm": 1.3676750659942627,
-      "learning_rate": 7.705628140703518e-05,
-      "loss": 5.3341,
-      "step": 23337
-    },
-    {
-      "epoch": 12.171056062581487,
-      "grad_norm": 1.3554993867874146,
-      "learning_rate": 7.705527638190955e-05,
-      "loss": 5.86,
-      "step": 23338
-    },
-    {
-      "epoch": 12.171577574967406,
-      "grad_norm": 1.556240200996399,
-      "learning_rate": 7.705427135678392e-05,
-      "loss": 5.322,
-      "step": 23339
-    },
-    {
-      "epoch": 12.172099087353324,
-      "grad_norm": 1.4049954414367676,
-      "learning_rate": 7.705326633165829e-05,
-      "loss": 5.263,
-      "step": 23340
-    },
-    {
-      "epoch": 12.172620599739243,
-      "grad_norm": 1.468074083328247,
-      "learning_rate": 7.705226130653267e-05,
-      "loss": 5.0556,
-      "step": 23341
-    },
-    {
-      "epoch": 12.173142112125163,
-      "grad_norm": 1.3800314664840698,
-      "learning_rate": 7.705125628140703e-05,
-      "loss": 5.6102,
-      "step": 23342
-    },
-    {
-      "epoch": 12.173663624511082,
-      "grad_norm": 1.5552878379821777,
-      "learning_rate": 7.705025125628141e-05,
-      "loss": 5.5784,
-      "step": 23343
-    },
-    {
-      "epoch": 12.174185136897002,
-      "grad_norm": 1.5406370162963867,
-      "learning_rate": 7.704924623115579e-05,
-      "loss": 5.3496,
-      "step": 23344
-    },
-    {
-      "epoch": 12.174706649282921,
-      "grad_norm": 1.516253113746643,
-      "learning_rate": 7.704824120603016e-05,
-      "loss": 5.2539,
-      "step": 23345
-    },
-    {
-      "epoch": 12.175228161668839,
-      "grad_norm": 1.4414924383163452,
-      "learning_rate": 7.704723618090453e-05,
-      "loss": 5.5794,
-      "step": 23346
-    },
-    {
-      "epoch": 12.175749674054758,
-      "grad_norm": 1.510703682899475,
-      "learning_rate": 7.70462311557789e-05,
-      "loss": 5.3524,
-      "step": 23347
-    },
-    {
-      "epoch": 12.176271186440678,
-      "grad_norm": 1.4159233570098877,
-      "learning_rate": 7.704522613065327e-05,
-      "loss": 5.7146,
-      "step": 23348
-    },
-    {
-      "epoch": 12.176792698826597,
-      "grad_norm": 1.9038708209991455,
-      "learning_rate": 7.704422110552763e-05,
-      "loss": 5.2348,
-      "step": 23349
-    },
-    {
-      "epoch": 12.177314211212517,
-      "grad_norm": 1.6102526187896729,
-      "learning_rate": 7.704321608040201e-05,
-      "loss": 5.249,
-      "step": 23350
-    },
-    {
-      "epoch": 12.177835723598436,
-      "grad_norm": 1.4696005582809448,
-      "learning_rate": 7.704221105527638e-05,
-      "loss": 5.845,
-      "step": 23351
-    },
-    {
-      "epoch": 12.178357235984354,
-      "grad_norm": 1.581902265548706,
-      "learning_rate": 7.704120603015075e-05,
-      "loss": 5.2773,
-      "step": 23352
-    },
-    {
-      "epoch": 12.178878748370273,
-      "grad_norm": 1.41827392578125,
-      "learning_rate": 7.704020100502513e-05,
-      "loss": 5.6197,
-      "step": 23353
-    },
-    {
-      "epoch": 12.179400260756193,
-      "grad_norm": 1.4019994735717773,
-      "learning_rate": 7.703919597989951e-05,
-      "loss": 5.298,
-      "step": 23354
-    },
-    {
-      "epoch": 12.179921773142112,
-      "grad_norm": 1.4796041250228882,
-      "learning_rate": 7.703819095477387e-05,
-      "loss": 5.8274,
-      "step": 23355
-    },
-    {
-      "epoch": 12.180443285528032,
-      "grad_norm": 1.5686041116714478,
-      "learning_rate": 7.703718592964825e-05,
-      "loss": 5.4601,
-      "step": 23356
-    },
-    {
-      "epoch": 12.180964797913951,
-      "grad_norm": 1.4996767044067383,
-      "learning_rate": 7.703618090452262e-05,
-      "loss": 5.13,
-      "step": 23357
-    },
-    {
-      "epoch": 12.181486310299869,
-      "grad_norm": 1.429119348526001,
-      "learning_rate": 7.7035175879397e-05,
-      "loss": 5.5594,
-      "step": 23358
-    },
-    {
-      "epoch": 12.182007822685788,
-      "grad_norm": 1.4767425060272217,
-      "learning_rate": 7.703417085427136e-05,
-      "loss": 5.4482,
-      "step": 23359
-    },
-    {
-      "epoch": 12.182529335071708,
-      "grad_norm": 1.342010736465454,
-      "learning_rate": 7.703316582914574e-05,
-      "loss": 4.8981,
-      "step": 23360
-    },
-    {
-      "epoch": 12.183050847457627,
-      "grad_norm": 1.6532597541809082,
-      "learning_rate": 7.70321608040201e-05,
-      "loss": 4.9596,
-      "step": 23361
-    },
-    {
-      "epoch": 12.183572359843547,
-      "grad_norm": 1.296413779258728,
-      "learning_rate": 7.703115577889448e-05,
-      "loss": 5.7844,
-      "step": 23362
-    },
-    {
-      "epoch": 12.184093872229466,
-      "grad_norm": 1.324422836303711,
-      "learning_rate": 7.703015075376884e-05,
-      "loss": 4.78,
-      "step": 23363
-    },
-    {
-      "epoch": 12.184615384615384,
-      "grad_norm": 1.5211480855941772,
-      "learning_rate": 7.702914572864322e-05,
-      "loss": 5.3522,
-      "step": 23364
-    },
-    {
-      "epoch": 12.185136897001303,
-      "grad_norm": 1.5446710586547852,
-      "learning_rate": 7.70281407035176e-05,
-      "loss": 4.8354,
-      "step": 23365
-    },
-    {
-      "epoch": 12.185658409387223,
-      "grad_norm": 1.4676069021224976,
-      "learning_rate": 7.702713567839196e-05,
-      "loss": 5.8365,
-      "step": 23366
-    },
-    {
-      "epoch": 12.186179921773142,
-      "grad_norm": 1.5342400074005127,
-      "learning_rate": 7.702613065326634e-05,
-      "loss": 5.7587,
-      "step": 23367
-    },
-    {
-      "epoch": 12.186701434159062,
-      "grad_norm": 1.5539108514785767,
-      "learning_rate": 7.70251256281407e-05,
-      "loss": 5.2583,
-      "step": 23368
-    },
-    {
-      "epoch": 12.187222946544981,
-      "grad_norm": 1.3007473945617676,
-      "learning_rate": 7.702412060301508e-05,
-      "loss": 5.6048,
-      "step": 23369
-    },
-    {
-      "epoch": 12.187744458930899,
-      "grad_norm": 1.3909543752670288,
-      "learning_rate": 7.702311557788945e-05,
-      "loss": 5.616,
-      "step": 23370
-    },
-    {
-      "epoch": 12.188265971316818,
-      "grad_norm": 1.464339017868042,
-      "learning_rate": 7.702211055276382e-05,
-      "loss": 5.1818,
-      "step": 23371
-    },
-    {
-      "epoch": 12.188787483702738,
-      "grad_norm": 1.5235601663589478,
-      "learning_rate": 7.702110552763819e-05,
-      "loss": 5.4435,
-      "step": 23372
-    },
-    {
-      "epoch": 12.189308996088657,
-      "grad_norm": 1.445291519165039,
-      "learning_rate": 7.702010050251257e-05,
-      "loss": 5.6791,
-      "step": 23373
-    },
-    {
-      "epoch": 12.189830508474577,
-      "grad_norm": 1.5655055046081543,
-      "learning_rate": 7.701909547738693e-05,
-      "loss": 5.3876,
-      "step": 23374
-    },
-    {
-      "epoch": 12.190352020860496,
-      "grad_norm": 1.5699520111083984,
-      "learning_rate": 7.701809045226131e-05,
-      "loss": 5.5915,
-      "step": 23375
-    },
-    {
-      "epoch": 12.190873533246414,
-      "grad_norm": 1.473563551902771,
-      "learning_rate": 7.701708542713569e-05,
-      "loss": 5.4007,
-      "step": 23376
-    },
-    {
-      "epoch": 12.191395045632333,
-      "grad_norm": 1.3971940279006958,
-      "learning_rate": 7.701608040201005e-05,
-      "loss": 4.644,
-      "step": 23377
-    },
-    {
-      "epoch": 12.191916558018253,
-      "grad_norm": 1.4787750244140625,
-      "learning_rate": 7.701507537688443e-05,
-      "loss": 5.8785,
-      "step": 23378
-    },
-    {
-      "epoch": 12.192438070404172,
-      "grad_norm": 1.405579924583435,
-      "learning_rate": 7.701407035175879e-05,
-      "loss": 5.2865,
-      "step": 23379
-    },
-    {
-      "epoch": 12.192959582790092,
-      "grad_norm": 1.448868751525879,
-      "learning_rate": 7.701306532663317e-05,
-      "loss": 5.1594,
-      "step": 23380
-    },
-    {
-      "epoch": 12.193481095176011,
-      "grad_norm": 1.5433615446090698,
-      "learning_rate": 7.701206030150754e-05,
-      "loss": 4.8572,
-      "step": 23381
-    },
-    {
-      "epoch": 12.194002607561929,
-      "grad_norm": 1.5075712203979492,
-      "learning_rate": 7.701105527638191e-05,
-      "loss": 5.1814,
-      "step": 23382
-    },
-    {
-      "epoch": 12.194524119947848,
-      "grad_norm": 1.482069730758667,
-      "learning_rate": 7.701005025125628e-05,
-      "loss": 5.4948,
-      "step": 23383
-    },
-    {
-      "epoch": 12.195045632333768,
-      "grad_norm": 1.4244662523269653,
-      "learning_rate": 7.700904522613066e-05,
-      "loss": 4.959,
-      "step": 23384
-    },
-    {
-      "epoch": 12.195567144719687,
-      "grad_norm": 1.4528125524520874,
-      "learning_rate": 7.700804020100503e-05,
-      "loss": 5.5357,
-      "step": 23385
-    },
-    {
-      "epoch": 12.196088657105607,
-      "grad_norm": 1.3969098329544067,
-      "learning_rate": 7.700703517587941e-05,
-      "loss": 5.7862,
-      "step": 23386
-    },
-    {
-      "epoch": 12.196610169491525,
-      "grad_norm": 1.494681477546692,
-      "learning_rate": 7.700603015075378e-05,
-      "loss": 5.4779,
-      "step": 23387
-    },
-    {
-      "epoch": 12.197131681877444,
-      "grad_norm": 1.4752023220062256,
-      "learning_rate": 7.700502512562815e-05,
-      "loss": 5.5634,
-      "step": 23388
-    },
-    {
-      "epoch": 12.197653194263363,
-      "grad_norm": 1.5467218160629272,
-      "learning_rate": 7.700402010050252e-05,
-      "loss": 5.2846,
-      "step": 23389
-    },
-    {
-      "epoch": 12.198174706649283,
-      "grad_norm": 1.444936990737915,
-      "learning_rate": 7.700301507537688e-05,
-      "loss": 5.469,
-      "step": 23390
-    },
-    {
-      "epoch": 12.198696219035202,
-      "grad_norm": 1.4908379316329956,
-      "learning_rate": 7.700201005025126e-05,
-      "loss": 4.9276,
-      "step": 23391
-    },
-    {
-      "epoch": 12.199217731421122,
-      "grad_norm": 1.6406298875808716,
-      "learning_rate": 7.700100502512562e-05,
-      "loss": 5.255,
-      "step": 23392
-    },
-    {
-      "epoch": 12.19973924380704,
-      "grad_norm": 1.3495728969573975,
-      "learning_rate": 7.7e-05,
-      "loss": 5.6502,
-      "step": 23393
-    },
-    {
-      "epoch": 12.200260756192959,
-      "grad_norm": 1.4650436639785767,
-      "learning_rate": 7.699899497487437e-05,
-      "loss": 5.6506,
-      "step": 23394
-    },
-    {
-      "epoch": 12.200782268578878,
-      "grad_norm": 1.6269320249557495,
-      "learning_rate": 7.699798994974874e-05,
-      "loss": 5.2085,
-      "step": 23395
-    },
-    {
-      "epoch": 12.201303780964798,
-      "grad_norm": 1.5121986865997314,
-      "learning_rate": 7.699698492462312e-05,
-      "loss": 5.4597,
-      "step": 23396
-    },
-    {
-      "epoch": 12.201825293350717,
-      "grad_norm": 1.4394415616989136,
-      "learning_rate": 7.69959798994975e-05,
-      "loss": 5.7235,
-      "step": 23397
-    },
-    {
-      "epoch": 12.202346805736637,
-      "grad_norm": 1.3976236581802368,
-      "learning_rate": 7.699497487437186e-05,
-      "loss": 5.5565,
-      "step": 23398
-    },
-    {
-      "epoch": 12.202868318122555,
-      "grad_norm": 1.4677674770355225,
-      "learning_rate": 7.699396984924624e-05,
-      "loss": 5.6801,
-      "step": 23399
-    },
-    {
-      "epoch": 12.203389830508474,
-      "grad_norm": 1.4842697381973267,
-      "learning_rate": 7.69929648241206e-05,
-      "loss": 5.5434,
-      "step": 23400
-    },
-    {
-      "epoch": 12.203911342894393,
-      "grad_norm": 1.4093700647354126,
-      "learning_rate": 7.699195979899498e-05,
-      "loss": 5.5177,
-      "step": 23401
-    },
-    {
-      "epoch": 12.204432855280313,
-      "grad_norm": 1.4265549182891846,
-      "learning_rate": 7.699095477386935e-05,
-      "loss": 5.3644,
-      "step": 23402
-    },
-    {
-      "epoch": 12.204954367666232,
-      "grad_norm": 1.3934346437454224,
-      "learning_rate": 7.698994974874371e-05,
-      "loss": 5.7197,
-      "step": 23403
-    },
-    {
-      "epoch": 12.205475880052152,
-      "grad_norm": 1.568651556968689,
-      "learning_rate": 7.698894472361809e-05,
-      "loss": 5.3019,
-      "step": 23404
-    },
-    {
-      "epoch": 12.20599739243807,
-      "grad_norm": 1.4075555801391602,
-      "learning_rate": 7.698793969849247e-05,
-      "loss": 5.6702,
-      "step": 23405
-    },
-    {
-      "epoch": 12.206518904823989,
-      "grad_norm": 1.6549304723739624,
-      "learning_rate": 7.698693467336685e-05,
-      "loss": 5.516,
-      "step": 23406
-    },
-    {
-      "epoch": 12.207040417209909,
-      "grad_norm": 1.488179087638855,
-      "learning_rate": 7.698592964824121e-05,
-      "loss": 5.2979,
-      "step": 23407
-    },
-    {
-      "epoch": 12.207561929595828,
-      "grad_norm": 1.3907779455184937,
-      "learning_rate": 7.698492462311559e-05,
-      "loss": 5.5283,
-      "step": 23408
-    },
-    {
-      "epoch": 12.208083441981747,
-      "grad_norm": 1.348798155784607,
-      "learning_rate": 7.698391959798995e-05,
-      "loss": 5.9286,
-      "step": 23409
-    },
-    {
-      "epoch": 12.208604954367667,
-      "grad_norm": 1.4377814531326294,
-      "learning_rate": 7.698291457286433e-05,
-      "loss": 5.5903,
-      "step": 23410
-    },
-    {
-      "epoch": 12.209126466753585,
-      "grad_norm": 1.461807131767273,
-      "learning_rate": 7.69819095477387e-05,
-      "loss": 5.2634,
-      "step": 23411
-    },
-    {
-      "epoch": 12.209647979139504,
-      "grad_norm": 1.4917467832565308,
-      "learning_rate": 7.698090452261307e-05,
-      "loss": 5.2107,
-      "step": 23412
-    },
-    {
-      "epoch": 12.210169491525424,
-      "grad_norm": 1.4868557453155518,
-      "learning_rate": 7.697989949748744e-05,
-      "loss": 5.4532,
-      "step": 23413
-    },
-    {
-      "epoch": 12.210691003911343,
-      "grad_norm": 1.475351333618164,
-      "learning_rate": 7.697889447236181e-05,
-      "loss": 5.3339,
-      "step": 23414
-    },
-    {
-      "epoch": 12.211212516297262,
-      "grad_norm": 1.4658844470977783,
-      "learning_rate": 7.697788944723618e-05,
-      "loss": 4.8308,
-      "step": 23415
-    },
-    {
-      "epoch": 12.211734028683182,
-      "grad_norm": 1.484336495399475,
-      "learning_rate": 7.697688442211056e-05,
-      "loss": 5.2462,
-      "step": 23416
-    },
-    {
-      "epoch": 12.2122555410691,
-      "grad_norm": 1.5299919843673706,
-      "learning_rate": 7.697587939698493e-05,
-      "loss": 4.8613,
-      "step": 23417
-    },
-    {
-      "epoch": 12.212777053455019,
-      "grad_norm": 1.369123935699463,
-      "learning_rate": 7.69748743718593e-05,
-      "loss": 5.9503,
-      "step": 23418
-    },
-    {
-      "epoch": 12.213298565840939,
-      "grad_norm": 1.3725521564483643,
-      "learning_rate": 7.697386934673368e-05,
-      "loss": 5.4528,
-      "step": 23419
-    },
-    {
-      "epoch": 12.213820078226858,
-      "grad_norm": 1.421111822128296,
-      "learning_rate": 7.697286432160804e-05,
-      "loss": 4.9316,
-      "step": 23420
-    },
-    {
-      "epoch": 12.214341590612777,
-      "grad_norm": 1.3240050077438354,
-      "learning_rate": 7.697185929648242e-05,
-      "loss": 5.8193,
-      "step": 23421
-    },
-    {
-      "epoch": 12.214863102998697,
-      "grad_norm": 1.3601304292678833,
-      "learning_rate": 7.697085427135678e-05,
-      "loss": 5.9338,
-      "step": 23422
-    },
-    {
-      "epoch": 12.215384615384615,
-      "grad_norm": 1.4981346130371094,
-      "learning_rate": 7.696984924623116e-05,
-      "loss": 4.8694,
-      "step": 23423
-    },
-    {
-      "epoch": 12.215906127770534,
-      "grad_norm": 1.5075773000717163,
-      "learning_rate": 7.696884422110552e-05,
-      "loss": 5.2662,
-      "step": 23424
-    },
-    {
-      "epoch": 12.216427640156454,
-      "grad_norm": 1.3956466913223267,
-      "learning_rate": 7.69678391959799e-05,
-      "loss": 5.5857,
-      "step": 23425
-    },
-    {
-      "epoch": 12.216949152542373,
-      "grad_norm": 1.441998839378357,
-      "learning_rate": 7.696683417085428e-05,
-      "loss": 5.5369,
-      "step": 23426
-    },
-    {
-      "epoch": 12.217470664928292,
-      "grad_norm": 1.4464836120605469,
-      "learning_rate": 7.696582914572866e-05,
-      "loss": 5.6208,
-      "step": 23427
-    },
-    {
-      "epoch": 12.217992177314212,
-      "grad_norm": 1.5827168226242065,
-      "learning_rate": 7.696482412060302e-05,
-      "loss": 5.0801,
-      "step": 23428
-    },
-    {
-      "epoch": 12.21851368970013,
-      "grad_norm": 1.496633529663086,
-      "learning_rate": 7.696381909547739e-05,
-      "loss": 5.4215,
-      "step": 23429
-    },
-    {
-      "epoch": 12.219035202086049,
-      "grad_norm": 1.3296139240264893,
-      "learning_rate": 7.696281407035176e-05,
-      "loss": 5.145,
-      "step": 23430
-    },
-    {
-      "epoch": 12.219556714471969,
-      "grad_norm": 1.5011311769485474,
-      "learning_rate": 7.696180904522613e-05,
-      "loss": 5.5977,
-      "step": 23431
-    },
-    {
-      "epoch": 12.220078226857888,
-      "grad_norm": 1.4469687938690186,
-      "learning_rate": 7.69608040201005e-05,
-      "loss": 5.6083,
-      "step": 23432
-    },
-    {
-      "epoch": 12.220599739243807,
-      "grad_norm": 1.5205358266830444,
-      "learning_rate": 7.695979899497487e-05,
-      "loss": 5.5453,
-      "step": 23433
-    },
-    {
-      "epoch": 12.221121251629727,
-      "grad_norm": 1.426040768623352,
-      "learning_rate": 7.695879396984925e-05,
-      "loss": 5.3442,
-      "step": 23434
-    },
-    {
-      "epoch": 12.221642764015645,
-      "grad_norm": 1.5034383535385132,
-      "learning_rate": 7.695778894472361e-05,
-      "loss": 5.8048,
-      "step": 23435
-    },
-    {
-      "epoch": 12.222164276401564,
-      "grad_norm": 1.3825082778930664,
-      "learning_rate": 7.695678391959799e-05,
-      "loss": 5.2168,
-      "step": 23436
-    },
-    {
-      "epoch": 12.222685788787484,
-      "grad_norm": 1.3344391584396362,
-      "learning_rate": 7.695577889447237e-05,
-      "loss": 5.268,
-      "step": 23437
-    },
-    {
-      "epoch": 12.223207301173403,
-      "grad_norm": 1.468492031097412,
-      "learning_rate": 7.695477386934675e-05,
-      "loss": 5.4971,
-      "step": 23438
-    },
-    {
-      "epoch": 12.223728813559323,
-      "grad_norm": 1.4639095067977905,
-      "learning_rate": 7.695376884422111e-05,
-      "loss": 5.2452,
-      "step": 23439
-    },
-    {
-      "epoch": 12.224250325945242,
-      "grad_norm": 1.5493566989898682,
-      "learning_rate": 7.695276381909549e-05,
-      "loss": 5.6298,
-      "step": 23440
-    },
-    {
-      "epoch": 12.22477183833116,
-      "grad_norm": 1.4565658569335938,
-      "learning_rate": 7.695175879396985e-05,
-      "loss": 5.1287,
-      "step": 23441
-    },
-    {
-      "epoch": 12.22529335071708,
-      "grad_norm": 1.4800050258636475,
-      "learning_rate": 7.695075376884422e-05,
-      "loss": 4.8522,
-      "step": 23442
-    },
-    {
-      "epoch": 12.225814863102999,
-      "grad_norm": 1.4877023696899414,
-      "learning_rate": 7.69497487437186e-05,
-      "loss": 5.3778,
-      "step": 23443
-    },
-    {
-      "epoch": 12.226336375488918,
-      "grad_norm": 1.5119552612304688,
-      "learning_rate": 7.694874371859296e-05,
-      "loss": 5.3344,
-      "step": 23444
-    },
-    {
-      "epoch": 12.226857887874838,
-      "grad_norm": 1.4463722705841064,
-      "learning_rate": 7.694773869346734e-05,
-      "loss": 5.4279,
-      "step": 23445
-    },
-    {
-      "epoch": 12.227379400260757,
-      "grad_norm": 1.481207251548767,
-      "learning_rate": 7.694673366834171e-05,
-      "loss": 5.2532,
-      "step": 23446
-    },
-    {
-      "epoch": 12.227900912646675,
-      "grad_norm": 1.5031452178955078,
-      "learning_rate": 7.694572864321609e-05,
-      "loss": 5.6102,
-      "step": 23447
-    },
-    {
-      "epoch": 12.228422425032594,
-      "grad_norm": 1.6356819868087769,
-      "learning_rate": 7.694472361809046e-05,
-      "loss": 4.9041,
-      "step": 23448
-    },
-    {
-      "epoch": 12.228943937418514,
-      "grad_norm": 1.5679041147232056,
-      "learning_rate": 7.694371859296483e-05,
-      "loss": 5.0112,
-      "step": 23449
-    },
-    {
-      "epoch": 12.229465449804433,
-      "grad_norm": 1.4716136455535889,
-      "learning_rate": 7.69427135678392e-05,
-      "loss": 5.269,
-      "step": 23450
-    },
-    {
-      "epoch": 12.229986962190353,
-      "grad_norm": 1.5324437618255615,
-      "learning_rate": 7.694170854271358e-05,
-      "loss": 5.0637,
-      "step": 23451
-    },
-    {
-      "epoch": 12.230508474576272,
-      "grad_norm": 1.453991413116455,
-      "learning_rate": 7.694070351758794e-05,
-      "loss": 5.2728,
-      "step": 23452
-    },
-    {
-      "epoch": 12.23102998696219,
-      "grad_norm": 1.5617198944091797,
-      "learning_rate": 7.693969849246232e-05,
-      "loss": 5.1903,
-      "step": 23453
-    },
-    {
-      "epoch": 12.23155149934811,
-      "grad_norm": 1.438713788986206,
-      "learning_rate": 7.693869346733668e-05,
-      "loss": 5.7281,
-      "step": 23454
-    },
-    {
-      "epoch": 12.232073011734029,
-      "grad_norm": 1.4736517667770386,
-      "learning_rate": 7.693768844221106e-05,
-      "loss": 4.8486,
-      "step": 23455
-    },
-    {
-      "epoch": 12.232594524119948,
-      "grad_norm": 1.4834736585617065,
-      "learning_rate": 7.693668341708543e-05,
-      "loss": 5.3221,
-      "step": 23456
-    },
-    {
-      "epoch": 12.233116036505868,
-      "grad_norm": 1.3648407459259033,
-      "learning_rate": 7.69356783919598e-05,
-      "loss": 5.3461,
-      "step": 23457
-    },
-    {
-      "epoch": 12.233637548891787,
-      "grad_norm": 1.6084610223770142,
-      "learning_rate": 7.693467336683418e-05,
-      "loss": 4.6965,
-      "step": 23458
-    },
-    {
-      "epoch": 12.234159061277705,
-      "grad_norm": 1.4772447347640991,
-      "learning_rate": 7.693366834170855e-05,
-      "loss": 5.559,
-      "step": 23459
-    },
-    {
-      "epoch": 12.234680573663624,
-      "grad_norm": 1.44584059715271,
-      "learning_rate": 7.693266331658292e-05,
-      "loss": 5.2308,
-      "step": 23460
-    },
-    {
-      "epoch": 12.235202086049544,
-      "grad_norm": 1.4519597291946411,
-      "learning_rate": 7.693165829145729e-05,
-      "loss": 5.5279,
-      "step": 23461
-    },
-    {
-      "epoch": 12.235723598435463,
-      "grad_norm": 1.4691693782806396,
-      "learning_rate": 7.693065326633166e-05,
-      "loss": 5.3888,
-      "step": 23462
-    },
-    {
-      "epoch": 12.236245110821383,
-      "grad_norm": 1.5871127843856812,
-      "learning_rate": 7.692964824120603e-05,
-      "loss": 4.7803,
-      "step": 23463
-    },
-    {
-      "epoch": 12.236766623207302,
-      "grad_norm": 1.4981111288070679,
-      "learning_rate": 7.692864321608041e-05,
-      "loss": 5.356,
-      "step": 23464
-    },
-    {
-      "epoch": 12.23728813559322,
-      "grad_norm": 1.7245451211929321,
-      "learning_rate": 7.692763819095477e-05,
-      "loss": 5.0893,
-      "step": 23465
-    },
-    {
-      "epoch": 12.23780964797914,
-      "grad_norm": 1.3875981569290161,
-      "learning_rate": 7.692663316582915e-05,
-      "loss": 4.9763,
-      "step": 23466
-    },
-    {
-      "epoch": 12.238331160365059,
-      "grad_norm": 1.4713716506958008,
-      "learning_rate": 7.692562814070353e-05,
-      "loss": 5.3436,
-      "step": 23467
-    },
-    {
-      "epoch": 12.238852672750978,
-      "grad_norm": 1.5183501243591309,
-      "learning_rate": 7.69246231155779e-05,
-      "loss": 5.5228,
-      "step": 23468
-    },
-    {
-      "epoch": 12.239374185136898,
-      "grad_norm": 1.4351000785827637,
-      "learning_rate": 7.692361809045227e-05,
-      "loss": 5.7282,
-      "step": 23469
-    },
-    {
-      "epoch": 12.239895697522817,
-      "grad_norm": 1.433763861656189,
-      "learning_rate": 7.692261306532663e-05,
-      "loss": 5.7184,
-      "step": 23470
-    },
-    {
-      "epoch": 12.240417209908735,
-      "grad_norm": 1.4274888038635254,
-      "learning_rate": 7.692160804020101e-05,
-      "loss": 5.4598,
-      "step": 23471
-    },
-    {
-      "epoch": 12.240938722294654,
-      "grad_norm": 1.3823484182357788,
-      "learning_rate": 7.692060301507538e-05,
-      "loss": 5.4318,
-      "step": 23472
-    },
-    {
-      "epoch": 12.241460234680574,
-      "grad_norm": 1.5556508302688599,
-      "learning_rate": 7.691959798994975e-05,
-      "loss": 5.2309,
-      "step": 23473
-    },
-    {
-      "epoch": 12.241981747066493,
-      "grad_norm": 1.4689873456954956,
-      "learning_rate": 7.691859296482412e-05,
-      "loss": 5.4114,
-      "step": 23474
-    },
-    {
-      "epoch": 12.242503259452413,
-      "grad_norm": 1.4011679887771606,
-      "learning_rate": 7.69175879396985e-05,
-      "loss": 5.5183,
-      "step": 23475
-    },
-    {
-      "epoch": 12.243024771838332,
-      "grad_norm": 1.3916503190994263,
-      "learning_rate": 7.691658291457286e-05,
-      "loss": 5.2072,
-      "step": 23476
-    },
-    {
-      "epoch": 12.24354628422425,
-      "grad_norm": 1.4306871891021729,
-      "learning_rate": 7.691557788944724e-05,
-      "loss": 5.4965,
-      "step": 23477
-    },
-    {
-      "epoch": 12.24406779661017,
-      "grad_norm": 1.5122649669647217,
-      "learning_rate": 7.691457286432162e-05,
-      "loss": 5.4916,
-      "step": 23478
-    },
-    {
-      "epoch": 12.244589308996089,
-      "grad_norm": 1.4805233478546143,
-      "learning_rate": 7.691356783919599e-05,
-      "loss": 5.254,
-      "step": 23479
-    },
-    {
-      "epoch": 12.245110821382008,
-      "grad_norm": 1.44832181930542,
-      "learning_rate": 7.691256281407036e-05,
-      "loss": 5.5046,
-      "step": 23480
-    },
-    {
-      "epoch": 12.245632333767928,
-      "grad_norm": 1.51243257522583,
-      "learning_rate": 7.691155778894474e-05,
-      "loss": 5.1793,
-      "step": 23481
-    },
-    {
-      "epoch": 12.246153846153845,
-      "grad_norm": 1.3970242738723755,
-      "learning_rate": 7.69105527638191e-05,
-      "loss": 5.5904,
-      "step": 23482
-    },
-    {
-      "epoch": 12.246675358539765,
-      "grad_norm": 1.5265660285949707,
-      "learning_rate": 7.690954773869346e-05,
-      "loss": 4.9084,
-      "step": 23483
-    },
-    {
-      "epoch": 12.247196870925684,
-      "grad_norm": 1.5538488626480103,
-      "learning_rate": 7.690854271356784e-05,
-      "loss": 5.0855,
-      "step": 23484
-    },
-    {
-      "epoch": 12.247718383311604,
-      "grad_norm": 1.465204119682312,
-      "learning_rate": 7.69075376884422e-05,
-      "loss": 5.4725,
-      "step": 23485
-    },
-    {
-      "epoch": 12.248239895697523,
-      "grad_norm": 1.537168025970459,
-      "learning_rate": 7.690653266331658e-05,
-      "loss": 5.3468,
-      "step": 23486
-    },
-    {
-      "epoch": 12.248761408083443,
-      "grad_norm": 1.479641079902649,
-      "learning_rate": 7.690552763819096e-05,
-      "loss": 5.4702,
-      "step": 23487
-    },
-    {
-      "epoch": 12.24928292046936,
-      "grad_norm": 1.4460856914520264,
-      "learning_rate": 7.690452261306534e-05,
-      "loss": 5.453,
-      "step": 23488
-    },
-    {
-      "epoch": 12.24980443285528,
-      "grad_norm": 1.4692192077636719,
-      "learning_rate": 7.69035175879397e-05,
-      "loss": 5.2307,
-      "step": 23489
-    },
-    {
-      "epoch": 12.2503259452412,
-      "grad_norm": 1.4893276691436768,
-      "learning_rate": 7.690251256281408e-05,
-      "loss": 5.6183,
-      "step": 23490
-    },
-    {
-      "epoch": 12.250847457627119,
-      "grad_norm": 1.4596339464187622,
-      "learning_rate": 7.690150753768845e-05,
-      "loss": 5.2109,
-      "step": 23491
-    },
-    {
-      "epoch": 12.251368970013038,
-      "grad_norm": 1.6318341493606567,
-      "learning_rate": 7.690050251256282e-05,
-      "loss": 5.3446,
-      "step": 23492
-    },
-    {
-      "epoch": 12.251890482398958,
-      "grad_norm": 1.4798396825790405,
-      "learning_rate": 7.689949748743719e-05,
-      "loss": 5.2297,
-      "step": 23493
-    },
-    {
-      "epoch": 12.252411994784875,
-      "grad_norm": 1.426254391670227,
-      "learning_rate": 7.689849246231157e-05,
-      "loss": 5.7886,
-      "step": 23494
-    },
-    {
-      "epoch": 12.252933507170795,
-      "grad_norm": 1.47239351272583,
-      "learning_rate": 7.689748743718593e-05,
-      "loss": 5.4894,
-      "step": 23495
-    },
-    {
-      "epoch": 12.253455019556714,
-      "grad_norm": 1.454176425933838,
-      "learning_rate": 7.68964824120603e-05,
-      "loss": 5.0353,
-      "step": 23496
-    },
-    {
-      "epoch": 12.253976531942634,
-      "grad_norm": 2.3065714836120605,
-      "learning_rate": 7.689547738693467e-05,
-      "loss": 4.9133,
-      "step": 23497
-    },
-    {
-      "epoch": 12.254498044328553,
-      "grad_norm": 1.3974833488464355,
-      "learning_rate": 7.689447236180905e-05,
-      "loss": 5.7384,
-      "step": 23498
-    },
-    {
-      "epoch": 12.255019556714473,
-      "grad_norm": 1.5268901586532593,
-      "learning_rate": 7.689346733668343e-05,
-      "loss": 5.543,
-      "step": 23499
-    },
-    {
-      "epoch": 12.25554106910039,
-      "grad_norm": 1.372110366821289,
-      "learning_rate": 7.689246231155779e-05,
-      "loss": 5.6629,
-      "step": 23500
-    },
-    {
-      "epoch": 12.25606258148631,
-      "grad_norm": 1.3891956806182861,
-      "learning_rate": 7.689145728643217e-05,
-      "loss": 5.3797,
-      "step": 23501
-    },
-    {
-      "epoch": 12.25658409387223,
-      "grad_norm": 1.5568557977676392,
-      "learning_rate": 7.689045226130653e-05,
-      "loss": 5.0984,
-      "step": 23502
-    },
-    {
-      "epoch": 12.257105606258149,
-      "grad_norm": 1.3741075992584229,
-      "learning_rate": 7.688944723618091e-05,
-      "loss": 5.3692,
-      "step": 23503
-    },
-    {
-      "epoch": 12.257627118644068,
-      "grad_norm": 1.4389379024505615,
-      "learning_rate": 7.688844221105528e-05,
-      "loss": 5.3149,
-      "step": 23504
-    },
-    {
-      "epoch": 12.258148631029988,
-      "grad_norm": 2.152184009552002,
-      "learning_rate": 7.688743718592965e-05,
-      "loss": 4.4842,
-      "step": 23505
-    },
-    {
-      "epoch": 12.258670143415905,
-      "grad_norm": 1.5198057889938354,
-      "learning_rate": 7.688643216080402e-05,
-      "loss": 5.5752,
-      "step": 23506
-    },
-    {
-      "epoch": 12.259191655801825,
-      "grad_norm": 1.4948941469192505,
-      "learning_rate": 7.68854271356784e-05,
-      "loss": 5.4062,
-      "step": 23507
-    },
-    {
-      "epoch": 12.259713168187744,
-      "grad_norm": 1.3873770236968994,
-      "learning_rate": 7.688442211055277e-05,
-      "loss": 5.3707,
-      "step": 23508
-    },
-    {
-      "epoch": 12.260234680573664,
-      "grad_norm": 1.433823585510254,
-      "learning_rate": 7.688341708542714e-05,
-      "loss": 5.4269,
-      "step": 23509
-    },
-    {
-      "epoch": 12.260756192959583,
-      "grad_norm": 1.5629554986953735,
-      "learning_rate": 7.688241206030152e-05,
-      "loss": 5.65,
-      "step": 23510
-    },
-    {
-      "epoch": 12.261277705345503,
-      "grad_norm": 1.6349718570709229,
-      "learning_rate": 7.688140703517588e-05,
-      "loss": 5.2877,
-      "step": 23511
-    },
-    {
-      "epoch": 12.26179921773142,
-      "grad_norm": 1.3161325454711914,
-      "learning_rate": 7.688040201005026e-05,
-      "loss": 5.655,
-      "step": 23512
-    },
-    {
-      "epoch": 12.26232073011734,
-      "grad_norm": 1.3688303232192993,
-      "learning_rate": 7.687939698492462e-05,
-      "loss": 5.6284,
-      "step": 23513
-    },
-    {
-      "epoch": 12.26284224250326,
-      "grad_norm": 1.4010018110275269,
-      "learning_rate": 7.6878391959799e-05,
-      "loss": 5.5411,
-      "step": 23514
-    },
-    {
-      "epoch": 12.263363754889179,
-      "grad_norm": 1.3352155685424805,
-      "learning_rate": 7.687738693467336e-05,
-      "loss": 5.6485,
-      "step": 23515
-    },
-    {
-      "epoch": 12.263885267275098,
-      "grad_norm": 1.4744479656219482,
-      "learning_rate": 7.687638190954774e-05,
-      "loss": 5.6441,
-      "step": 23516
-    },
-    {
-      "epoch": 12.264406779661018,
-      "grad_norm": 1.378240704536438,
-      "learning_rate": 7.68753768844221e-05,
-      "loss": 5.363,
-      "step": 23517
-    },
-    {
-      "epoch": 12.264928292046935,
-      "grad_norm": 1.3530656099319458,
-      "learning_rate": 7.687437185929648e-05,
-      "loss": 5.7812,
-      "step": 23518
-    },
-    {
-      "epoch": 12.265449804432855,
-      "grad_norm": 1.4664875268936157,
-      "learning_rate": 7.687336683417086e-05,
-      "loss": 5.3958,
-      "step": 23519
-    },
-    {
-      "epoch": 12.265971316818774,
-      "grad_norm": 1.5010510683059692,
-      "learning_rate": 7.687236180904524e-05,
-      "loss": 5.2494,
-      "step": 23520
-    },
-    {
-      "epoch": 12.266492829204694,
-      "grad_norm": 1.4408434629440308,
-      "learning_rate": 7.68713567839196e-05,
-      "loss": 5.5437,
-      "step": 23521
-    },
-    {
-      "epoch": 12.267014341590613,
-      "grad_norm": 1.484817385673523,
-      "learning_rate": 7.687035175879397e-05,
-      "loss": 5.4117,
-      "step": 23522
-    },
-    {
-      "epoch": 12.267535853976533,
-      "grad_norm": 1.4730479717254639,
-      "learning_rate": 7.686934673366835e-05,
-      "loss": 5.3054,
-      "step": 23523
-    },
-    {
-      "epoch": 12.26805736636245,
-      "grad_norm": 1.4410946369171143,
-      "learning_rate": 7.686834170854271e-05,
-      "loss": 5.5291,
-      "step": 23524
-    },
-    {
-      "epoch": 12.26857887874837,
-      "grad_norm": 1.5051729679107666,
-      "learning_rate": 7.686733668341709e-05,
-      "loss": 5.3207,
-      "step": 23525
-    },
-    {
-      "epoch": 12.26910039113429,
-      "grad_norm": 1.577158808708191,
-      "learning_rate": 7.686633165829145e-05,
-      "loss": 5.3924,
-      "step": 23526
-    },
-    {
-      "epoch": 12.269621903520209,
-      "grad_norm": 1.5518120527267456,
-      "learning_rate": 7.686532663316583e-05,
-      "loss": 5.5911,
-      "step": 23527
-    },
-    {
-      "epoch": 12.270143415906128,
-      "grad_norm": 1.4769539833068848,
-      "learning_rate": 7.68643216080402e-05,
-      "loss": 5.3826,
-      "step": 23528
-    },
-    {
-      "epoch": 12.270664928292048,
-      "grad_norm": 1.4242756366729736,
-      "learning_rate": 7.686331658291457e-05,
-      "loss": 5.7021,
-      "step": 23529
-    },
-    {
-      "epoch": 12.271186440677965,
-      "grad_norm": 1.481825828552246,
-      "learning_rate": 7.686231155778895e-05,
-      "loss": 5.3296,
-      "step": 23530
-    },
-    {
-      "epoch": 12.271707953063885,
-      "grad_norm": 1.4833170175552368,
-      "learning_rate": 7.686130653266333e-05,
-      "loss": 5.5658,
-      "step": 23531
-    },
-    {
-      "epoch": 12.272229465449804,
-      "grad_norm": 1.5606683492660522,
-      "learning_rate": 7.686030150753769e-05,
-      "loss": 5.0721,
-      "step": 23532
-    },
-    {
-      "epoch": 12.272750977835724,
-      "grad_norm": 1.4315614700317383,
-      "learning_rate": 7.685929648241207e-05,
-      "loss": 5.5472,
-      "step": 23533
-    },
-    {
-      "epoch": 12.273272490221643,
-      "grad_norm": 1.4477698802947998,
-      "learning_rate": 7.685829145728643e-05,
-      "loss": 5.5709,
-      "step": 23534
-    },
-    {
-      "epoch": 12.273794002607563,
-      "grad_norm": 1.38666832447052,
-      "learning_rate": 7.68572864321608e-05,
-      "loss": 5.7203,
-      "step": 23535
-    },
-    {
-      "epoch": 12.27431551499348,
-      "grad_norm": 1.5146559476852417,
-      "learning_rate": 7.685628140703518e-05,
-      "loss": 5.414,
-      "step": 23536
-    },
-    {
-      "epoch": 12.2748370273794,
-      "grad_norm": 1.918952226638794,
-      "learning_rate": 7.685527638190954e-05,
-      "loss": 4.9108,
-      "step": 23537
-    },
-    {
-      "epoch": 12.27535853976532,
-      "grad_norm": 1.471876859664917,
-      "learning_rate": 7.685427135678392e-05,
-      "loss": 5.6064,
-      "step": 23538
-    },
-    {
-      "epoch": 12.275880052151239,
-      "grad_norm": 1.584214448928833,
-      "learning_rate": 7.68532663316583e-05,
-      "loss": 5.2214,
-      "step": 23539
-    },
-    {
-      "epoch": 12.276401564537158,
-      "grad_norm": 1.529974102973938,
-      "learning_rate": 7.685226130653267e-05,
-      "loss": 5.2193,
-      "step": 23540
-    },
-    {
-      "epoch": 12.276923076923078,
-      "grad_norm": 1.5200854539871216,
-      "learning_rate": 7.685125628140704e-05,
-      "loss": 5.5041,
-      "step": 23541
-    },
-    {
-      "epoch": 12.277444589308995,
-      "grad_norm": 1.4809788465499878,
-      "learning_rate": 7.685025125628142e-05,
-      "loss": 5.5802,
-      "step": 23542
-    },
-    {
-      "epoch": 12.277966101694915,
-      "grad_norm": 1.4301517009735107,
-      "learning_rate": 7.684924623115578e-05,
-      "loss": 5.4782,
-      "step": 23543
-    },
-    {
-      "epoch": 12.278487614080834,
-      "grad_norm": 1.4741039276123047,
-      "learning_rate": 7.684824120603016e-05,
-      "loss": 5.3981,
-      "step": 23544
-    },
-    {
-      "epoch": 12.279009126466754,
-      "grad_norm": 1.5055347681045532,
-      "learning_rate": 7.684723618090452e-05,
-      "loss": 5.5865,
-      "step": 23545
-    },
-    {
-      "epoch": 12.279530638852673,
-      "grad_norm": 1.4368410110473633,
-      "learning_rate": 7.68462311557789e-05,
-      "loss": 4.9875,
-      "step": 23546
-    },
-    {
-      "epoch": 12.280052151238593,
-      "grad_norm": 1.403823971748352,
-      "learning_rate": 7.684522613065327e-05,
-      "loss": 5.2696,
-      "step": 23547
-    },
-    {
-      "epoch": 12.28057366362451,
-      "grad_norm": 1.4264287948608398,
-      "learning_rate": 7.684422110552764e-05,
-      "loss": 5.3022,
-      "step": 23548
-    },
-    {
-      "epoch": 12.28109517601043,
-      "grad_norm": 1.3024344444274902,
-      "learning_rate": 7.684321608040201e-05,
-      "loss": 5.0571,
-      "step": 23549
-    },
-    {
-      "epoch": 12.28161668839635,
-      "grad_norm": 1.435333490371704,
-      "learning_rate": 7.684221105527639e-05,
-      "loss": 5.0737,
-      "step": 23550
-    },
-    {
-      "epoch": 12.282138200782269,
-      "grad_norm": 1.5997816324234009,
-      "learning_rate": 7.684120603015076e-05,
-      "loss": 5.4165,
-      "step": 23551
-    },
-    {
-      "epoch": 12.282659713168188,
-      "grad_norm": 1.406262755393982,
-      "learning_rate": 7.684020100502513e-05,
-      "loss": 5.4008,
-      "step": 23552
-    },
-    {
-      "epoch": 12.283181225554108,
-      "grad_norm": 1.5628767013549805,
-      "learning_rate": 7.68391959798995e-05,
-      "loss": 5.0606,
-      "step": 23553
-    },
-    {
-      "epoch": 12.283702737940025,
-      "grad_norm": 1.5092662572860718,
-      "learning_rate": 7.683819095477387e-05,
-      "loss": 5.1271,
-      "step": 23554
-    },
-    {
-      "epoch": 12.284224250325945,
-      "grad_norm": 1.5450851917266846,
-      "learning_rate": 7.683718592964825e-05,
-      "loss": 5.5982,
-      "step": 23555
-    },
-    {
-      "epoch": 12.284745762711864,
-      "grad_norm": 1.4596476554870605,
-      "learning_rate": 7.683618090452261e-05,
-      "loss": 5.4758,
-      "step": 23556
-    },
-    {
-      "epoch": 12.285267275097784,
-      "grad_norm": 1.4630751609802246,
-      "learning_rate": 7.683517587939699e-05,
-      "loss": 5.0783,
-      "step": 23557
-    },
-    {
-      "epoch": 12.285788787483703,
-      "grad_norm": 1.401288628578186,
-      "learning_rate": 7.683417085427135e-05,
-      "loss": 5.7694,
-      "step": 23558
-    },
-    {
-      "epoch": 12.286310299869623,
-      "grad_norm": 1.6064871549606323,
-      "learning_rate": 7.683316582914573e-05,
-      "loss": 5.1078,
-      "step": 23559
-    },
-    {
-      "epoch": 12.28683181225554,
-      "grad_norm": 1.4648468494415283,
-      "learning_rate": 7.683216080402011e-05,
-      "loss": 5.64,
-      "step": 23560
-    },
-    {
-      "epoch": 12.28735332464146,
-      "grad_norm": 1.391213059425354,
-      "learning_rate": 7.683115577889449e-05,
-      "loss": 5.8249,
-      "step": 23561
-    },
-    {
-      "epoch": 12.28787483702738,
-      "grad_norm": 1.3655915260314941,
-      "learning_rate": 7.683015075376885e-05,
-      "loss": 5.8063,
-      "step": 23562
-    },
-    {
-      "epoch": 12.288396349413299,
-      "grad_norm": 1.4405900239944458,
-      "learning_rate": 7.682914572864322e-05,
-      "loss": 5.7258,
-      "step": 23563
-    },
-    {
-      "epoch": 12.288917861799218,
-      "grad_norm": 1.5103713274002075,
-      "learning_rate": 7.68281407035176e-05,
-      "loss": 5.0084,
-      "step": 23564
-    },
-    {
-      "epoch": 12.289439374185136,
-      "grad_norm": 1.5414390563964844,
-      "learning_rate": 7.682713567839196e-05,
-      "loss": 5.2937,
-      "step": 23565
-    },
-    {
-      "epoch": 12.289960886571055,
-      "grad_norm": 1.604797124862671,
-      "learning_rate": 7.682613065326634e-05,
-      "loss": 5.2154,
-      "step": 23566
-    },
-    {
-      "epoch": 12.290482398956975,
-      "grad_norm": 1.4621418714523315,
-      "learning_rate": 7.68251256281407e-05,
-      "loss": 5.7437,
-      "step": 23567
-    },
-    {
-      "epoch": 12.291003911342894,
-      "grad_norm": 1.395200252532959,
-      "learning_rate": 7.682412060301508e-05,
-      "loss": 5.5055,
-      "step": 23568
-    },
-    {
-      "epoch": 12.291525423728814,
-      "grad_norm": 1.6037142276763916,
-      "learning_rate": 7.682311557788944e-05,
-      "loss": 4.1766,
-      "step": 23569
-    },
-    {
-      "epoch": 12.292046936114733,
-      "grad_norm": 1.401550531387329,
-      "learning_rate": 7.682211055276382e-05,
-      "loss": 5.5729,
-      "step": 23570
-    },
-    {
-      "epoch": 12.292568448500653,
-      "grad_norm": 1.5912796258926392,
-      "learning_rate": 7.68211055276382e-05,
-      "loss": 5.7182,
-      "step": 23571
-    },
-    {
-      "epoch": 12.29308996088657,
-      "grad_norm": 1.4828217029571533,
-      "learning_rate": 7.682010050251258e-05,
-      "loss": 5.2549,
-      "step": 23572
-    },
-    {
-      "epoch": 12.29361147327249,
-      "grad_norm": 1.4439067840576172,
-      "learning_rate": 7.681909547738694e-05,
-      "loss": 5.217,
-      "step": 23573
-    },
-    {
-      "epoch": 12.29413298565841,
-      "grad_norm": 1.5239394903182983,
-      "learning_rate": 7.681809045226132e-05,
-      "loss": 5.278,
-      "step": 23574
-    },
-    {
-      "epoch": 12.294654498044329,
-      "grad_norm": 1.4097518920898438,
-      "learning_rate": 7.681708542713568e-05,
-      "loss": 5.4917,
-      "step": 23575
-    },
-    {
-      "epoch": 12.295176010430248,
-      "grad_norm": 1.5155402421951294,
-      "learning_rate": 7.681608040201005e-05,
-      "loss": 5.1568,
-      "step": 23576
-    },
-    {
-      "epoch": 12.295697522816166,
-      "grad_norm": 1.4986567497253418,
-      "learning_rate": 7.681507537688442e-05,
-      "loss": 5.3506,
-      "step": 23577
-    },
-    {
-      "epoch": 12.296219035202085,
-      "grad_norm": 1.5848749876022339,
-      "learning_rate": 7.681407035175879e-05,
-      "loss": 5.1226,
-      "step": 23578
-    },
-    {
-      "epoch": 12.296740547588005,
-      "grad_norm": 1.5412745475769043,
-      "learning_rate": 7.681306532663317e-05,
-      "loss": 5.5389,
-      "step": 23579
-    },
-    {
-      "epoch": 12.297262059973924,
-      "grad_norm": 1.4789113998413086,
-      "learning_rate": 7.681206030150754e-05,
-      "loss": 5.548,
-      "step": 23580
-    },
-    {
-      "epoch": 12.297783572359844,
-      "grad_norm": 1.4641568660736084,
-      "learning_rate": 7.681105527638192e-05,
-      "loss": 5.4423,
-      "step": 23581
-    },
-    {
-      "epoch": 12.298305084745763,
-      "grad_norm": 1.5310620069503784,
-      "learning_rate": 7.681005025125629e-05,
-      "loss": 5.0952,
-      "step": 23582
-    },
-    {
-      "epoch": 12.298826597131681,
-      "grad_norm": 1.4498887062072754,
-      "learning_rate": 7.680904522613066e-05,
-      "loss": 5.4163,
-      "step": 23583
-    },
-    {
-      "epoch": 12.2993481095176,
-      "grad_norm": 1.4151877164840698,
-      "learning_rate": 7.680804020100503e-05,
-      "loss": 5.3434,
-      "step": 23584
-    },
-    {
-      "epoch": 12.29986962190352,
-      "grad_norm": 1.7020518779754639,
-      "learning_rate": 7.68070351758794e-05,
-      "loss": 4.9903,
-      "step": 23585
-    },
-    {
-      "epoch": 12.30039113428944,
-      "grad_norm": 1.353913426399231,
-      "learning_rate": 7.680603015075377e-05,
-      "loss": 5.7409,
-      "step": 23586
-    },
-    {
-      "epoch": 12.300912646675359,
-      "grad_norm": 1.5250664949417114,
-      "learning_rate": 7.680502512562815e-05,
-      "loss": 5.5964,
-      "step": 23587
-    },
-    {
-      "epoch": 12.301434159061278,
-      "grad_norm": 1.6111100912094116,
-      "learning_rate": 7.680402010050251e-05,
-      "loss": 5.1991,
-      "step": 23588
-    },
-    {
-      "epoch": 12.301955671447196,
-      "grad_norm": 1.4251304864883423,
-      "learning_rate": 7.680301507537688e-05,
-      "loss": 5.785,
-      "step": 23589
-    },
-    {
-      "epoch": 12.302477183833116,
-      "grad_norm": 1.349582314491272,
-      "learning_rate": 7.680201005025125e-05,
-      "loss": 5.5351,
-      "step": 23590
-    },
-    {
-      "epoch": 12.302998696219035,
-      "grad_norm": 1.5072436332702637,
-      "learning_rate": 7.680100502512563e-05,
-      "loss": 5.3655,
-      "step": 23591
-    },
-    {
-      "epoch": 12.303520208604954,
-      "grad_norm": 1.381162405014038,
-      "learning_rate": 7.680000000000001e-05,
-      "loss": 5.1491,
-      "step": 23592
-    },
-    {
-      "epoch": 12.304041720990874,
-      "grad_norm": 1.5225462913513184,
-      "learning_rate": 7.679899497487437e-05,
-      "loss": 4.7669,
-      "step": 23593
-    },
-    {
-      "epoch": 12.304563233376793,
-      "grad_norm": 1.499366044998169,
-      "learning_rate": 7.679798994974875e-05,
-      "loss": 5.52,
-      "step": 23594
-    },
-    {
-      "epoch": 12.305084745762711,
-      "grad_norm": 1.4931479692459106,
-      "learning_rate": 7.679698492462312e-05,
-      "loss": 5.3771,
-      "step": 23595
-    },
-    {
-      "epoch": 12.30560625814863,
-      "grad_norm": 1.4286094903945923,
-      "learning_rate": 7.67959798994975e-05,
-      "loss": 5.3746,
-      "step": 23596
-    },
-    {
-      "epoch": 12.30612777053455,
-      "grad_norm": 1.4306505918502808,
-      "learning_rate": 7.679497487437186e-05,
-      "loss": 5.3166,
-      "step": 23597
-    },
-    {
-      "epoch": 12.30664928292047,
-      "grad_norm": 1.5476855039596558,
-      "learning_rate": 7.679396984924624e-05,
-      "loss": 4.767,
-      "step": 23598
-    },
-    {
-      "epoch": 12.307170795306389,
-      "grad_norm": 1.4481042623519897,
-      "learning_rate": 7.67929648241206e-05,
-      "loss": 5.6851,
-      "step": 23599
-    },
-    {
-      "epoch": 12.307692307692308,
-      "grad_norm": 1.441007137298584,
-      "learning_rate": 7.679195979899498e-05,
-      "loss": 5.7031,
-      "step": 23600
-    },
-    {
-      "epoch": 12.308213820078226,
-      "grad_norm": 1.6182913780212402,
-      "learning_rate": 7.679095477386936e-05,
-      "loss": 4.9624,
-      "step": 23601
-    },
-    {
-      "epoch": 12.308735332464146,
-      "grad_norm": 1.4677131175994873,
-      "learning_rate": 7.678994974874372e-05,
-      "loss": 5.6574,
-      "step": 23602
-    },
-    {
-      "epoch": 12.309256844850065,
-      "grad_norm": 1.4375437498092651,
-      "learning_rate": 7.67889447236181e-05,
-      "loss": 5.3312,
-      "step": 23603
-    },
-    {
-      "epoch": 12.309778357235984,
-      "grad_norm": 1.4641664028167725,
-      "learning_rate": 7.678793969849246e-05,
-      "loss": 5.3092,
-      "step": 23604
-    },
-    {
-      "epoch": 12.310299869621904,
-      "grad_norm": 1.4859267473220825,
-      "learning_rate": 7.678693467336684e-05,
-      "loss": 5.4255,
-      "step": 23605
-    },
-    {
-      "epoch": 12.310821382007823,
-      "grad_norm": 1.4179842472076416,
-      "learning_rate": 7.67859296482412e-05,
-      "loss": 5.5159,
-      "step": 23606
-    },
-    {
-      "epoch": 12.311342894393741,
-      "grad_norm": 1.450583815574646,
-      "learning_rate": 7.678492462311558e-05,
-      "loss": 5.0103,
-      "step": 23607
-    },
-    {
-      "epoch": 12.31186440677966,
-      "grad_norm": 1.4421262741088867,
-      "learning_rate": 7.678391959798995e-05,
-      "loss": 5.4183,
-      "step": 23608
-    },
-    {
-      "epoch": 12.31238591916558,
-      "grad_norm": 1.7716171741485596,
-      "learning_rate": 7.678291457286432e-05,
-      "loss": 4.7523,
-      "step": 23609
-    },
-    {
-      "epoch": 12.3129074315515,
-      "grad_norm": 1.4396649599075317,
-      "learning_rate": 7.678190954773869e-05,
-      "loss": 5.5994,
-      "step": 23610
-    },
-    {
-      "epoch": 12.313428943937419,
-      "grad_norm": 1.5082061290740967,
-      "learning_rate": 7.678090452261307e-05,
-      "loss": 5.659,
-      "step": 23611
-    },
-    {
-      "epoch": 12.313950456323338,
-      "grad_norm": 1.4844192266464233,
-      "learning_rate": 7.677989949748744e-05,
-      "loss": 5.1157,
-      "step": 23612
-    },
-    {
-      "epoch": 12.314471968709256,
-      "grad_norm": 1.3942553997039795,
-      "learning_rate": 7.677889447236182e-05,
-      "loss": 4.3651,
-      "step": 23613
-    },
-    {
-      "epoch": 12.314993481095176,
-      "grad_norm": 1.4373472929000854,
-      "learning_rate": 7.677788944723619e-05,
-      "loss": 5.6214,
-      "step": 23614
-    },
-    {
-      "epoch": 12.315514993481095,
-      "grad_norm": 1.4922218322753906,
-      "learning_rate": 7.677688442211055e-05,
-      "loss": 5.4275,
-      "step": 23615
-    },
-    {
-      "epoch": 12.316036505867014,
-      "grad_norm": 1.519867181777954,
-      "learning_rate": 7.677587939698493e-05,
-      "loss": 5.3819,
-      "step": 23616
-    },
-    {
-      "epoch": 12.316558018252934,
-      "grad_norm": 1.5627973079681396,
-      "learning_rate": 7.677487437185929e-05,
-      "loss": 5.5159,
-      "step": 23617
-    },
-    {
-      "epoch": 12.317079530638853,
-      "grad_norm": 1.4274423122406006,
-      "learning_rate": 7.677386934673367e-05,
-      "loss": 5.5081,
-      "step": 23618
-    },
-    {
-      "epoch": 12.317601043024771,
-      "grad_norm": 1.4206336736679077,
-      "learning_rate": 7.677286432160804e-05,
-      "loss": 5.4786,
-      "step": 23619
-    },
-    {
-      "epoch": 12.31812255541069,
-      "grad_norm": 1.567037582397461,
-      "learning_rate": 7.677185929648241e-05,
-      "loss": 5.1983,
-      "step": 23620
-    },
-    {
-      "epoch": 12.31864406779661,
-      "grad_norm": 1.6467007398605347,
-      "learning_rate": 7.677085427135679e-05,
-      "loss": 5.0612,
-      "step": 23621
-    },
-    {
-      "epoch": 12.31916558018253,
-      "grad_norm": 1.5112676620483398,
-      "learning_rate": 7.676984924623117e-05,
-      "loss": 5.823,
-      "step": 23622
-    },
-    {
-      "epoch": 12.319687092568449,
-      "grad_norm": 1.4643086194992065,
-      "learning_rate": 7.676884422110553e-05,
-      "loss": 4.8157,
-      "step": 23623
-    },
-    {
-      "epoch": 12.320208604954368,
-      "grad_norm": 1.5048716068267822,
-      "learning_rate": 7.676783919597991e-05,
-      "loss": 5.1658,
-      "step": 23624
-    },
-    {
-      "epoch": 12.320730117340286,
-      "grad_norm": 1.4298841953277588,
-      "learning_rate": 7.676683417085428e-05,
-      "loss": 5.209,
-      "step": 23625
-    },
-    {
-      "epoch": 12.321251629726206,
-      "grad_norm": 1.4957914352416992,
-      "learning_rate": 7.676582914572865e-05,
-      "loss": 5.2019,
-      "step": 23626
-    },
-    {
-      "epoch": 12.321773142112125,
-      "grad_norm": 1.488970160484314,
-      "learning_rate": 7.676482412060302e-05,
-      "loss": 5.2745,
-      "step": 23627
-    },
-    {
-      "epoch": 12.322294654498045,
-      "grad_norm": 1.5350968837738037,
-      "learning_rate": 7.67638190954774e-05,
-      "loss": 5.4405,
-      "step": 23628
-    },
-    {
-      "epoch": 12.322816166883964,
-      "grad_norm": 1.3907203674316406,
-      "learning_rate": 7.676281407035176e-05,
-      "loss": 5.0682,
-      "step": 23629
-    },
-    {
-      "epoch": 12.323337679269883,
-      "grad_norm": 1.3493101596832275,
-      "learning_rate": 7.676180904522612e-05,
-      "loss": 5.4242,
-      "step": 23630
-    },
-    {
-      "epoch": 12.323859191655801,
-      "grad_norm": 1.4014216661453247,
-      "learning_rate": 7.67608040201005e-05,
-      "loss": 5.7191,
-      "step": 23631
-    },
-    {
-      "epoch": 12.32438070404172,
-      "grad_norm": 1.4452831745147705,
-      "learning_rate": 7.675979899497488e-05,
-      "loss": 5.4763,
-      "step": 23632
-    },
-    {
-      "epoch": 12.32490221642764,
-      "grad_norm": 1.471582055091858,
-      "learning_rate": 7.675879396984926e-05,
-      "loss": 5.4928,
-      "step": 23633
-    },
-    {
-      "epoch": 12.32542372881356,
-      "grad_norm": 1.4015012979507446,
-      "learning_rate": 7.675778894472362e-05,
-      "loss": 5.1767,
-      "step": 23634
-    },
-    {
-      "epoch": 12.325945241199479,
-      "grad_norm": 1.3674722909927368,
-      "learning_rate": 7.6756783919598e-05,
-      "loss": 5.4317,
-      "step": 23635
-    },
-    {
-      "epoch": 12.326466753585398,
-      "grad_norm": 1.4732887744903564,
-      "learning_rate": 7.675577889447236e-05,
-      "loss": 5.2916,
-      "step": 23636
-    },
-    {
-      "epoch": 12.326988265971316,
-      "grad_norm": 1.4947097301483154,
-      "learning_rate": 7.675477386934674e-05,
-      "loss": 4.9706,
-      "step": 23637
-    },
-    {
-      "epoch": 12.327509778357236,
-      "grad_norm": 1.4739835262298584,
-      "learning_rate": 7.67537688442211e-05,
-      "loss": 5.4793,
-      "step": 23638
-    },
-    {
-      "epoch": 12.328031290743155,
-      "grad_norm": 1.4755786657333374,
-      "learning_rate": 7.675276381909548e-05,
-      "loss": 5.5868,
-      "step": 23639
-    },
-    {
-      "epoch": 12.328552803129075,
-      "grad_norm": 1.507470965385437,
-      "learning_rate": 7.675175879396985e-05,
-      "loss": 5.4429,
-      "step": 23640
-    },
-    {
-      "epoch": 12.329074315514994,
-      "grad_norm": 1.3887451887130737,
-      "learning_rate": 7.675075376884423e-05,
-      "loss": 4.9723,
-      "step": 23641
-    },
-    {
-      "epoch": 12.329595827900913,
-      "grad_norm": 1.8033865690231323,
-      "learning_rate": 7.67497487437186e-05,
-      "loss": 5.5927,
-      "step": 23642
-    },
-    {
-      "epoch": 12.330117340286831,
-      "grad_norm": 1.5592985153198242,
-      "learning_rate": 7.674874371859297e-05,
-      "loss": 4.9597,
-      "step": 23643
-    },
-    {
-      "epoch": 12.33063885267275,
-      "grad_norm": 1.5028451681137085,
-      "learning_rate": 7.674773869346735e-05,
-      "loss": 5.2987,
-      "step": 23644
-    },
-    {
-      "epoch": 12.33116036505867,
-      "grad_norm": 1.4387253522872925,
-      "learning_rate": 7.674673366834171e-05,
-      "loss": 5.4767,
-      "step": 23645
-    },
-    {
-      "epoch": 12.33168187744459,
-      "grad_norm": 1.5133087635040283,
-      "learning_rate": 7.674572864321609e-05,
-      "loss": 5.2067,
-      "step": 23646
-    },
-    {
-      "epoch": 12.332203389830509,
-      "grad_norm": 1.541138768196106,
-      "learning_rate": 7.674472361809045e-05,
-      "loss": 5.2295,
-      "step": 23647
-    },
-    {
-      "epoch": 12.332724902216428,
-      "grad_norm": 1.4436063766479492,
-      "learning_rate": 7.674371859296483e-05,
-      "loss": 5.6369,
-      "step": 23648
-    },
-    {
-      "epoch": 12.333246414602346,
-      "grad_norm": 1.4134939908981323,
-      "learning_rate": 7.67427135678392e-05,
-      "loss": 5.48,
-      "step": 23649
-    },
-    {
-      "epoch": 12.333767926988266,
-      "grad_norm": 1.429584264755249,
-      "learning_rate": 7.674170854271357e-05,
-      "loss": 5.4802,
-      "step": 23650
-    },
-    {
-      "epoch": 12.334289439374185,
-      "grad_norm": 1.3951953649520874,
-      "learning_rate": 7.674070351758794e-05,
-      "loss": 5.6477,
-      "step": 23651
-    },
-    {
-      "epoch": 12.334810951760105,
-      "grad_norm": 1.3422776460647583,
-      "learning_rate": 7.673969849246231e-05,
-      "loss": 5.4358,
-      "step": 23652
-    },
-    {
-      "epoch": 12.335332464146024,
-      "grad_norm": 1.4771535396575928,
-      "learning_rate": 7.673869346733669e-05,
-      "loss": 5.1673,
-      "step": 23653
-    },
-    {
-      "epoch": 12.335853976531943,
-      "grad_norm": 1.443471074104309,
-      "learning_rate": 7.673768844221107e-05,
-      "loss": 5.2717,
-      "step": 23654
-    },
-    {
-      "epoch": 12.336375488917861,
-      "grad_norm": 1.5715981721878052,
-      "learning_rate": 7.673668341708543e-05,
-      "loss": 5.0272,
-      "step": 23655
-    },
-    {
-      "epoch": 12.33689700130378,
-      "grad_norm": 1.4622397422790527,
-      "learning_rate": 7.67356783919598e-05,
-      "loss": 5.051,
-      "step": 23656
-    },
-    {
-      "epoch": 12.3374185136897,
-      "grad_norm": 1.5086497068405151,
-      "learning_rate": 7.673467336683418e-05,
-      "loss": 5.4506,
-      "step": 23657
-    },
-    {
-      "epoch": 12.33794002607562,
-      "grad_norm": 1.4242887496948242,
-      "learning_rate": 7.673366834170854e-05,
-      "loss": 5.4185,
-      "step": 23658
-    },
-    {
-      "epoch": 12.338461538461539,
-      "grad_norm": 1.4471826553344727,
-      "learning_rate": 7.673266331658292e-05,
-      "loss": 5.539,
-      "step": 23659
-    },
-    {
-      "epoch": 12.338983050847457,
-      "grad_norm": 1.5423641204833984,
-      "learning_rate": 7.673165829145728e-05,
-      "loss": 4.9435,
-      "step": 23660
-    },
-    {
-      "epoch": 12.339504563233376,
-      "grad_norm": 1.3631783723831177,
-      "learning_rate": 7.673065326633166e-05,
-      "loss": 5.5042,
-      "step": 23661
-    },
-    {
-      "epoch": 12.340026075619296,
-      "grad_norm": 1.4827919006347656,
-      "learning_rate": 7.672964824120604e-05,
-      "loss": 5.5358,
-      "step": 23662
-    },
-    {
-      "epoch": 12.340547588005215,
-      "grad_norm": 1.6170425415039062,
-      "learning_rate": 7.672864321608042e-05,
-      "loss": 5.2686,
-      "step": 23663
-    },
-    {
-      "epoch": 12.341069100391135,
-      "grad_norm": 1.4040875434875488,
-      "learning_rate": 7.672763819095478e-05,
-      "loss": 5.7891,
-      "step": 23664
-    },
-    {
-      "epoch": 12.341590612777054,
-      "grad_norm": 1.440994143486023,
-      "learning_rate": 7.672663316582916e-05,
-      "loss": 5.5944,
-      "step": 23665
-    },
-    {
-      "epoch": 12.342112125162974,
-      "grad_norm": 1.3963356018066406,
-      "learning_rate": 7.672562814070352e-05,
-      "loss": 5.3525,
-      "step": 23666
-    },
-    {
-      "epoch": 12.342633637548891,
-      "grad_norm": 1.5071451663970947,
-      "learning_rate": 7.67246231155779e-05,
-      "loss": 5.1959,
-      "step": 23667
-    },
-    {
-      "epoch": 12.34315514993481,
-      "grad_norm": 1.5508922338485718,
-      "learning_rate": 7.672361809045226e-05,
-      "loss": 5.3707,
-      "step": 23668
-    },
-    {
-      "epoch": 12.34367666232073,
-      "grad_norm": 1.5107954740524292,
-      "learning_rate": 7.672261306532663e-05,
-      "loss": 5.6344,
-      "step": 23669
-    },
-    {
-      "epoch": 12.34419817470665,
-      "grad_norm": 1.493680715560913,
-      "learning_rate": 7.6721608040201e-05,
-      "loss": 5.4422,
-      "step": 23670
-    },
-    {
-      "epoch": 12.344719687092569,
-      "grad_norm": 1.4298105239868164,
-      "learning_rate": 7.672060301507537e-05,
-      "loss": 5.6975,
-      "step": 23671
-    },
-    {
-      "epoch": 12.345241199478487,
-      "grad_norm": 1.5458966493606567,
-      "learning_rate": 7.671959798994975e-05,
-      "loss": 5.4024,
-      "step": 23672
-    },
-    {
-      "epoch": 12.345762711864406,
-      "grad_norm": 1.5120104551315308,
-      "learning_rate": 7.671859296482413e-05,
-      "loss": 5.4573,
-      "step": 23673
-    },
-    {
-      "epoch": 12.346284224250326,
-      "grad_norm": 1.5521783828735352,
-      "learning_rate": 7.67175879396985e-05,
-      "loss": 4.5637,
-      "step": 23674
-    },
-    {
-      "epoch": 12.346805736636245,
-      "grad_norm": 1.5491816997528076,
-      "learning_rate": 7.671658291457287e-05,
-      "loss": 5.3356,
-      "step": 23675
-    },
-    {
-      "epoch": 12.347327249022165,
-      "grad_norm": 1.4213879108428955,
-      "learning_rate": 7.671557788944725e-05,
-      "loss": 5.5824,
-      "step": 23676
-    },
-    {
-      "epoch": 12.347848761408084,
-      "grad_norm": 1.4127589464187622,
-      "learning_rate": 7.671457286432161e-05,
-      "loss": 5.2708,
-      "step": 23677
-    },
-    {
-      "epoch": 12.348370273794002,
-      "grad_norm": 1.4159135818481445,
-      "learning_rate": 7.671356783919599e-05,
-      "loss": 5.7077,
-      "step": 23678
-    },
-    {
-      "epoch": 12.348891786179921,
-      "grad_norm": 1.5135833024978638,
-      "learning_rate": 7.671256281407035e-05,
-      "loss": 5.0548,
-      "step": 23679
-    },
-    {
-      "epoch": 12.34941329856584,
-      "grad_norm": 1.4795161485671997,
-      "learning_rate": 7.671155778894473e-05,
-      "loss": 5.6348,
-      "step": 23680
-    },
-    {
-      "epoch": 12.34993481095176,
-      "grad_norm": 1.636663556098938,
-      "learning_rate": 7.67105527638191e-05,
-      "loss": 5.1344,
-      "step": 23681
-    },
-    {
-      "epoch": 12.35045632333768,
-      "grad_norm": 1.4778704643249512,
-      "learning_rate": 7.670954773869347e-05,
-      "loss": 5.2474,
-      "step": 23682
-    },
-    {
-      "epoch": 12.350977835723599,
-      "grad_norm": 1.5429329872131348,
-      "learning_rate": 7.670854271356785e-05,
-      "loss": 5.1915,
-      "step": 23683
-    },
-    {
-      "epoch": 12.351499348109517,
-      "grad_norm": 1.8718957901000977,
-      "learning_rate": 7.670753768844221e-05,
-      "loss": 4.996,
-      "step": 23684
-    },
-    {
-      "epoch": 12.352020860495436,
-      "grad_norm": 1.5296001434326172,
-      "learning_rate": 7.670653266331659e-05,
-      "loss": 5.4835,
-      "step": 23685
-    },
-    {
-      "epoch": 12.352542372881356,
-      "grad_norm": 1.4372080564498901,
-      "learning_rate": 7.670552763819096e-05,
-      "loss": 5.3605,
-      "step": 23686
-    },
-    {
-      "epoch": 12.353063885267275,
-      "grad_norm": 1.4230800867080688,
-      "learning_rate": 7.670452261306533e-05,
-      "loss": 5.2265,
-      "step": 23687
-    },
-    {
-      "epoch": 12.353585397653195,
-      "grad_norm": 1.3419502973556519,
-      "learning_rate": 7.67035175879397e-05,
-      "loss": 5.8891,
-      "step": 23688
-    },
-    {
-      "epoch": 12.354106910039114,
-      "grad_norm": 1.440359354019165,
-      "learning_rate": 7.670251256281408e-05,
-      "loss": 5.0338,
-      "step": 23689
-    },
-    {
-      "epoch": 12.354628422425032,
-      "grad_norm": 1.41267728805542,
-      "learning_rate": 7.670150753768844e-05,
-      "loss": 5.6896,
-      "step": 23690
-    },
-    {
-      "epoch": 12.355149934810951,
-      "grad_norm": 1.4996075630187988,
-      "learning_rate": 7.670050251256282e-05,
-      "loss": 5.4835,
-      "step": 23691
-    },
-    {
-      "epoch": 12.35567144719687,
-      "grad_norm": 1.4483411312103271,
-      "learning_rate": 7.669949748743718e-05,
-      "loss": 5.3165,
-      "step": 23692
-    },
-    {
-      "epoch": 12.35619295958279,
-      "grad_norm": 1.4509235620498657,
-      "learning_rate": 7.669849246231156e-05,
-      "loss": 5.4318,
-      "step": 23693
-    },
-    {
-      "epoch": 12.35671447196871,
-      "grad_norm": 1.4060497283935547,
-      "learning_rate": 7.669748743718594e-05,
-      "loss": 5.6409,
-      "step": 23694
-    },
-    {
-      "epoch": 12.357235984354629,
-      "grad_norm": 1.3695839643478394,
-      "learning_rate": 7.66964824120603e-05,
-      "loss": 5.2531,
-      "step": 23695
-    },
-    {
-      "epoch": 12.357757496740547,
-      "grad_norm": 1.612196683883667,
-      "learning_rate": 7.669547738693468e-05,
-      "loss": 5.4142,
-      "step": 23696
-    },
-    {
-      "epoch": 12.358279009126466,
-      "grad_norm": 1.3526718616485596,
-      "learning_rate": 7.669447236180905e-05,
-      "loss": 5.8948,
-      "step": 23697
-    },
-    {
-      "epoch": 12.358800521512386,
-      "grad_norm": 1.5156418085098267,
-      "learning_rate": 7.669346733668342e-05,
-      "loss": 4.688,
-      "step": 23698
-    },
-    {
-      "epoch": 12.359322033898305,
-      "grad_norm": 1.5916551351547241,
-      "learning_rate": 7.669246231155779e-05,
-      "loss": 5.1795,
-      "step": 23699
-    },
-    {
-      "epoch": 12.359843546284225,
-      "grad_norm": 1.5424823760986328,
-      "learning_rate": 7.669145728643216e-05,
-      "loss": 5.3087,
-      "step": 23700
-    },
-    {
-      "epoch": 12.360365058670144,
-      "grad_norm": 1.5496686697006226,
-      "learning_rate": 7.669045226130653e-05,
-      "loss": 5.2375,
-      "step": 23701
-    },
-    {
-      "epoch": 12.360886571056062,
-      "grad_norm": 1.3954726457595825,
-      "learning_rate": 7.668944723618091e-05,
-      "loss": 5.5602,
-      "step": 23702
-    },
-    {
-      "epoch": 12.361408083441981,
-      "grad_norm": 1.55675208568573,
-      "learning_rate": 7.668844221105527e-05,
-      "loss": 5.4108,
-      "step": 23703
-    },
-    {
-      "epoch": 12.3619295958279,
-      "grad_norm": 1.487550139427185,
-      "learning_rate": 7.668743718592965e-05,
-      "loss": 5.3019,
-      "step": 23704
-    },
-    {
-      "epoch": 12.36245110821382,
-      "grad_norm": 1.4501068592071533,
-      "learning_rate": 7.668643216080403e-05,
-      "loss": 5.2373,
-      "step": 23705
-    },
-    {
-      "epoch": 12.36297262059974,
-      "grad_norm": 1.4231985807418823,
-      "learning_rate": 7.66854271356784e-05,
-      "loss": 5.1564,
-      "step": 23706
-    },
-    {
-      "epoch": 12.36349413298566,
-      "grad_norm": 1.4249277114868164,
-      "learning_rate": 7.668442211055277e-05,
-      "loss": 5.5049,
-      "step": 23707
-    },
-    {
-      "epoch": 12.364015645371577,
-      "grad_norm": 1.4019403457641602,
-      "learning_rate": 7.668341708542713e-05,
-      "loss": 5.5468,
-      "step": 23708
-    },
-    {
-      "epoch": 12.364537157757496,
-      "grad_norm": 1.3663320541381836,
-      "learning_rate": 7.668241206030151e-05,
-      "loss": 5.8751,
-      "step": 23709
-    },
-    {
-      "epoch": 12.365058670143416,
-      "grad_norm": 1.5527546405792236,
-      "learning_rate": 7.668140703517588e-05,
-      "loss": 4.8432,
-      "step": 23710
-    },
-    {
-      "epoch": 12.365580182529335,
-      "grad_norm": 1.4574214220046997,
-      "learning_rate": 7.668040201005025e-05,
-      "loss": 5.351,
-      "step": 23711
-    },
-    {
-      "epoch": 12.366101694915255,
-      "grad_norm": 1.4102742671966553,
-      "learning_rate": 7.667939698492462e-05,
-      "loss": 5.2538,
-      "step": 23712
-    },
-    {
-      "epoch": 12.366623207301174,
-      "grad_norm": 1.433205008506775,
-      "learning_rate": 7.6678391959799e-05,
-      "loss": 5.4038,
-      "step": 23713
-    },
-    {
-      "epoch": 12.367144719687092,
-      "grad_norm": 1.4234426021575928,
-      "learning_rate": 7.667738693467337e-05,
-      "loss": 5.502,
-      "step": 23714
-    },
-    {
-      "epoch": 12.367666232073011,
-      "grad_norm": 1.5028311014175415,
-      "learning_rate": 7.667638190954775e-05,
-      "loss": 5.6442,
-      "step": 23715
-    },
-    {
-      "epoch": 12.36818774445893,
-      "grad_norm": 1.4873422384262085,
-      "learning_rate": 7.667537688442212e-05,
-      "loss": 5.2097,
-      "step": 23716
-    },
-    {
-      "epoch": 12.36870925684485,
-      "grad_norm": 1.392982840538025,
-      "learning_rate": 7.667437185929649e-05,
-      "loss": 5.6197,
-      "step": 23717
-    },
-    {
-      "epoch": 12.36923076923077,
-      "grad_norm": 1.4624534845352173,
-      "learning_rate": 7.667336683417086e-05,
-      "loss": 5.296,
-      "step": 23718
-    },
-    {
-      "epoch": 12.36975228161669,
-      "grad_norm": 1.7282353639602661,
-      "learning_rate": 7.667236180904524e-05,
-      "loss": 5.0758,
-      "step": 23719
-    },
-    {
-      "epoch": 12.370273794002607,
-      "grad_norm": 1.5903007984161377,
-      "learning_rate": 7.66713567839196e-05,
-      "loss": 5.0268,
-      "step": 23720
-    },
-    {
-      "epoch": 12.370795306388526,
-      "grad_norm": 1.443490982055664,
-      "learning_rate": 7.667035175879398e-05,
-      "loss": 4.8423,
-      "step": 23721
-    },
-    {
-      "epoch": 12.371316818774446,
-      "grad_norm": 1.386910080909729,
-      "learning_rate": 7.666934673366834e-05,
-      "loss": 5.343,
-      "step": 23722
-    },
-    {
-      "epoch": 12.371838331160365,
-      "grad_norm": 1.414070963859558,
-      "learning_rate": 7.66683417085427e-05,
-      "loss": 5.8942,
-      "step": 23723
-    },
-    {
-      "epoch": 12.372359843546285,
-      "grad_norm": 1.5113693475723267,
-      "learning_rate": 7.666733668341708e-05,
-      "loss": 5.272,
-      "step": 23724
-    },
-    {
-      "epoch": 12.372881355932204,
-      "grad_norm": 1.4365779161453247,
-      "learning_rate": 7.666633165829146e-05,
-      "loss": 5.3439,
-      "step": 23725
-    },
-    {
-      "epoch": 12.373402868318122,
-      "grad_norm": 1.501918077468872,
-      "learning_rate": 7.666532663316584e-05,
-      "loss": 5.4954,
-      "step": 23726
-    },
-    {
-      "epoch": 12.373924380704041,
-      "grad_norm": 1.4559036493301392,
-      "learning_rate": 7.66643216080402e-05,
-      "loss": 5.1647,
-      "step": 23727
-    },
-    {
-      "epoch": 12.37444589308996,
-      "grad_norm": 1.4163917303085327,
-      "learning_rate": 7.666331658291458e-05,
-      "loss": 5.5564,
-      "step": 23728
-    },
-    {
-      "epoch": 12.37496740547588,
-      "grad_norm": 1.4073452949523926,
-      "learning_rate": 7.666231155778895e-05,
-      "loss": 5.5797,
-      "step": 23729
-    },
-    {
-      "epoch": 12.3754889178618,
-      "grad_norm": 1.3822050094604492,
-      "learning_rate": 7.666130653266332e-05,
-      "loss": 5.5231,
-      "step": 23730
-    },
-    {
-      "epoch": 12.37601043024772,
-      "grad_norm": 1.4354628324508667,
-      "learning_rate": 7.666030150753769e-05,
-      "loss": 5.4379,
-      "step": 23731
-    },
-    {
-      "epoch": 12.376531942633637,
-      "grad_norm": 1.4167670011520386,
-      "learning_rate": 7.665929648241207e-05,
-      "loss": 5.8126,
-      "step": 23732
-    },
-    {
-      "epoch": 12.377053455019556,
-      "grad_norm": 1.6715668439865112,
-      "learning_rate": 7.665829145728643e-05,
-      "loss": 4.9626,
-      "step": 23733
-    },
-    {
-      "epoch": 12.377574967405476,
-      "grad_norm": 1.4501445293426514,
-      "learning_rate": 7.665728643216081e-05,
-      "loss": 5.2342,
-      "step": 23734
-    },
-    {
-      "epoch": 12.378096479791395,
-      "grad_norm": 1.4735110998153687,
-      "learning_rate": 7.665628140703519e-05,
-      "loss": 5.4406,
-      "step": 23735
-    },
-    {
-      "epoch": 12.378617992177315,
-      "grad_norm": 1.6361209154129028,
-      "learning_rate": 7.665527638190955e-05,
-      "loss": 5.3364,
-      "step": 23736
-    },
-    {
-      "epoch": 12.379139504563234,
-      "grad_norm": 1.4286049604415894,
-      "learning_rate": 7.665427135678393e-05,
-      "loss": 5.5233,
-      "step": 23737
-    },
-    {
-      "epoch": 12.379661016949152,
-      "grad_norm": 1.528102993965149,
-      "learning_rate": 7.665326633165829e-05,
-      "loss": 5.3237,
-      "step": 23738
-    },
-    {
-      "epoch": 12.380182529335071,
-      "grad_norm": 1.5193580389022827,
-      "learning_rate": 7.665226130653267e-05,
-      "loss": 5.245,
-      "step": 23739
-    },
-    {
-      "epoch": 12.38070404172099,
-      "grad_norm": 1.5164334774017334,
-      "learning_rate": 7.665125628140703e-05,
-      "loss": 5.2387,
-      "step": 23740
-    },
-    {
-      "epoch": 12.38122555410691,
-      "grad_norm": 1.63292396068573,
-      "learning_rate": 7.665025125628141e-05,
-      "loss": 5.1478,
-      "step": 23741
-    },
-    {
-      "epoch": 12.38174706649283,
-      "grad_norm": 1.5880143642425537,
-      "learning_rate": 7.664924623115578e-05,
-      "loss": 4.6874,
-      "step": 23742
-    },
-    {
-      "epoch": 12.38226857887875,
-      "grad_norm": 1.5438241958618164,
-      "learning_rate": 7.664824120603015e-05,
-      "loss": 4.9601,
-      "step": 23743
-    },
-    {
-      "epoch": 12.382790091264667,
-      "grad_norm": 1.499232292175293,
-      "learning_rate": 7.664723618090452e-05,
-      "loss": 5.3103,
-      "step": 23744
-    },
-    {
-      "epoch": 12.383311603650586,
-      "grad_norm": 1.3890080451965332,
-      "learning_rate": 7.66462311557789e-05,
-      "loss": 5.4212,
-      "step": 23745
-    },
-    {
-      "epoch": 12.383833116036506,
-      "grad_norm": 1.3916972875595093,
-      "learning_rate": 7.664522613065327e-05,
-      "loss": 5.5923,
-      "step": 23746
-    },
-    {
-      "epoch": 12.384354628422425,
-      "grad_norm": 1.4140353202819824,
-      "learning_rate": 7.664422110552765e-05,
-      "loss": 5.6644,
-      "step": 23747
-    },
-    {
-      "epoch": 12.384876140808345,
-      "grad_norm": 1.4810951948165894,
-      "learning_rate": 7.664321608040202e-05,
-      "loss": 5.3137,
-      "step": 23748
-    },
-    {
-      "epoch": 12.385397653194264,
-      "grad_norm": 1.4368773698806763,
-      "learning_rate": 7.664221105527638e-05,
-      "loss": 5.3774,
-      "step": 23749
-    },
-    {
-      "epoch": 12.385919165580182,
-      "grad_norm": 1.454219102859497,
-      "learning_rate": 7.664120603015076e-05,
-      "loss": 5.8086,
-      "step": 23750
-    },
-    {
-      "epoch": 12.386440677966101,
-      "grad_norm": 1.3830466270446777,
-      "learning_rate": 7.664020100502512e-05,
-      "loss": 5.7732,
-      "step": 23751
-    },
-    {
-      "epoch": 12.38696219035202,
-      "grad_norm": 1.5816638469696045,
-      "learning_rate": 7.66391959798995e-05,
-      "loss": 5.0469,
-      "step": 23752
-    },
-    {
-      "epoch": 12.38748370273794,
-      "grad_norm": 1.6049405336380005,
-      "learning_rate": 7.663819095477386e-05,
-      "loss": 5.4273,
-      "step": 23753
-    },
-    {
-      "epoch": 12.38800521512386,
-      "grad_norm": 1.4875094890594482,
-      "learning_rate": 7.663718592964824e-05,
-      "loss": 5.3415,
-      "step": 23754
-    },
-    {
-      "epoch": 12.388526727509777,
-      "grad_norm": 1.4727206230163574,
-      "learning_rate": 7.663618090452262e-05,
-      "loss": 5.5229,
-      "step": 23755
-    },
-    {
-      "epoch": 12.389048239895697,
-      "grad_norm": 1.560967206954956,
-      "learning_rate": 7.6635175879397e-05,
-      "loss": 5.0815,
-      "step": 23756
-    },
-    {
-      "epoch": 12.389569752281616,
-      "grad_norm": 1.5528901815414429,
-      "learning_rate": 7.663417085427136e-05,
-      "loss": 5.3369,
-      "step": 23757
-    },
-    {
-      "epoch": 12.390091264667536,
-      "grad_norm": 1.3624002933502197,
-      "learning_rate": 7.663316582914574e-05,
-      "loss": 5.7886,
-      "step": 23758
-    },
-    {
-      "epoch": 12.390612777053455,
-      "grad_norm": 1.4848289489746094,
-      "learning_rate": 7.66321608040201e-05,
-      "loss": 5.5882,
-      "step": 23759
-    },
-    {
-      "epoch": 12.391134289439375,
-      "grad_norm": 1.476951241493225,
-      "learning_rate": 7.663115577889448e-05,
-      "loss": 5.5261,
-      "step": 23760
-    },
-    {
-      "epoch": 12.391655801825294,
-      "grad_norm": 1.394593358039856,
-      "learning_rate": 7.663015075376885e-05,
-      "loss": 5.8159,
-      "step": 23761
-    },
-    {
-      "epoch": 12.392177314211212,
-      "grad_norm": 2.0089752674102783,
-      "learning_rate": 7.662914572864321e-05,
-      "loss": 4.8575,
-      "step": 23762
-    },
-    {
-      "epoch": 12.392698826597131,
-      "grad_norm": 1.4256051778793335,
-      "learning_rate": 7.662814070351759e-05,
-      "loss": 5.4253,
-      "step": 23763
-    },
-    {
-      "epoch": 12.39322033898305,
-      "grad_norm": 1.578748345375061,
-      "learning_rate": 7.662713567839195e-05,
-      "loss": 5.4285,
-      "step": 23764
-    },
-    {
-      "epoch": 12.39374185136897,
-      "grad_norm": 1.482709527015686,
-      "learning_rate": 7.662613065326633e-05,
-      "loss": 5.4518,
-      "step": 23765
-    },
-    {
-      "epoch": 12.39426336375489,
-      "grad_norm": 1.5098317861557007,
-      "learning_rate": 7.662512562814071e-05,
-      "loss": 5.2705,
-      "step": 23766
-    },
-    {
-      "epoch": 12.394784876140807,
-      "grad_norm": 1.3870972394943237,
-      "learning_rate": 7.662412060301509e-05,
-      "loss": 5.6458,
-      "step": 23767
-    },
-    {
-      "epoch": 12.395306388526727,
-      "grad_norm": 1.3240457773208618,
-      "learning_rate": 7.662311557788945e-05,
-      "loss": 4.8305,
-      "step": 23768
-    },
-    {
-      "epoch": 12.395827900912646,
-      "grad_norm": 1.3355458974838257,
-      "learning_rate": 7.662211055276383e-05,
-      "loss": 5.736,
-      "step": 23769
-    },
-    {
-      "epoch": 12.396349413298566,
-      "grad_norm": 1.4091652631759644,
-      "learning_rate": 7.662110552763819e-05,
-      "loss": 5.5905,
-      "step": 23770
-    },
-    {
-      "epoch": 12.396870925684485,
-      "grad_norm": 1.6322332620620728,
-      "learning_rate": 7.662010050251257e-05,
-      "loss": 4.2978,
-      "step": 23771
-    },
-    {
-      "epoch": 12.397392438070405,
-      "grad_norm": 1.5264809131622314,
-      "learning_rate": 7.661909547738693e-05,
-      "loss": 5.3908,
-      "step": 23772
-    },
-    {
-      "epoch": 12.397913950456322,
-      "grad_norm": 1.480878472328186,
-      "learning_rate": 7.661809045226131e-05,
-      "loss": 5.1181,
-      "step": 23773
-    },
-    {
-      "epoch": 12.398435462842242,
-      "grad_norm": 1.5581618547439575,
-      "learning_rate": 7.661708542713568e-05,
-      "loss": 5.2572,
-      "step": 23774
-    },
-    {
-      "epoch": 12.398956975228161,
-      "grad_norm": 1.4146952629089355,
-      "learning_rate": 7.661608040201005e-05,
-      "loss": 5.1697,
-      "step": 23775
-    },
-    {
-      "epoch": 12.399478487614081,
-      "grad_norm": 1.4930747747421265,
-      "learning_rate": 7.661507537688443e-05,
-      "loss": 5.2314,
-      "step": 23776
-    },
-    {
-      "epoch": 12.4,
-      "grad_norm": 1.3794203996658325,
-      "learning_rate": 7.66140703517588e-05,
-      "loss": 5.5589,
-      "step": 23777
-    },
-    {
-      "epoch": 12.40052151238592,
-      "grad_norm": 1.4893410205841064,
-      "learning_rate": 7.661306532663317e-05,
-      "loss": 5.3771,
-      "step": 23778
-    },
-    {
-      "epoch": 12.401043024771838,
-      "grad_norm": 1.4471783638000488,
-      "learning_rate": 7.661206030150754e-05,
-      "loss": 5.5363,
-      "step": 23779
-    },
-    {
-      "epoch": 12.401564537157757,
-      "grad_norm": 1.506136178970337,
-      "learning_rate": 7.661105527638192e-05,
-      "loss": 4.3449,
-      "step": 23780
-    },
-    {
-      "epoch": 12.402086049543676,
-      "grad_norm": 1.3989169597625732,
-      "learning_rate": 7.661005025125628e-05,
-      "loss": 5.6992,
-      "step": 23781
-    },
-    {
-      "epoch": 12.402607561929596,
-      "grad_norm": 1.443914532661438,
-      "learning_rate": 7.660904522613066e-05,
-      "loss": 5.515,
-      "step": 23782
-    },
-    {
-      "epoch": 12.403129074315515,
-      "grad_norm": 1.4601478576660156,
-      "learning_rate": 7.660804020100502e-05,
-      "loss": 5.7207,
-      "step": 23783
-    },
-    {
-      "epoch": 12.403650586701435,
-      "grad_norm": 1.5320745706558228,
-      "learning_rate": 7.66070351758794e-05,
-      "loss": 5.2528,
-      "step": 23784
-    },
-    {
-      "epoch": 12.404172099087353,
-      "grad_norm": 1.4792823791503906,
-      "learning_rate": 7.660603015075377e-05,
-      "loss": 5.6863,
-      "step": 23785
-    },
-    {
-      "epoch": 12.404693611473272,
-      "grad_norm": 1.4663664102554321,
-      "learning_rate": 7.660502512562814e-05,
-      "loss": 5.3809,
-      "step": 23786
-    },
-    {
-      "epoch": 12.405215123859191,
-      "grad_norm": 1.407945156097412,
-      "learning_rate": 7.660402010050252e-05,
-      "loss": 5.7999,
-      "step": 23787
-    },
-    {
-      "epoch": 12.405736636245111,
-      "grad_norm": 1.4283978939056396,
-      "learning_rate": 7.660301507537689e-05,
-      "loss": 5.4627,
-      "step": 23788
-    },
-    {
-      "epoch": 12.40625814863103,
-      "grad_norm": 1.446805477142334,
-      "learning_rate": 7.660201005025126e-05,
-      "loss": 5.6867,
-      "step": 23789
-    },
-    {
-      "epoch": 12.40677966101695,
-      "grad_norm": 1.4864200353622437,
-      "learning_rate": 7.660100502512563e-05,
-      "loss": 5.608,
-      "step": 23790
-    },
-    {
-      "epoch": 12.407301173402868,
-      "grad_norm": 1.4039350748062134,
-      "learning_rate": 7.66e-05,
-      "loss": 5.7236,
-      "step": 23791
-    },
-    {
-      "epoch": 12.407822685788787,
-      "grad_norm": 1.4183052778244019,
-      "learning_rate": 7.659899497487437e-05,
-      "loss": 5.4293,
-      "step": 23792
-    },
-    {
-      "epoch": 12.408344198174706,
-      "grad_norm": 1.467413306236267,
-      "learning_rate": 7.659798994974875e-05,
-      "loss": 5.7466,
-      "step": 23793
-    },
-    {
-      "epoch": 12.408865710560626,
-      "grad_norm": 1.4488184452056885,
-      "learning_rate": 7.659698492462311e-05,
-      "loss": 5.4932,
-      "step": 23794
-    },
-    {
-      "epoch": 12.409387222946545,
-      "grad_norm": 1.4276982545852661,
-      "learning_rate": 7.659597989949749e-05,
-      "loss": 5.8916,
-      "step": 23795
-    },
-    {
-      "epoch": 12.409908735332465,
-      "grad_norm": 1.397135615348816,
-      "learning_rate": 7.659497487437187e-05,
-      "loss": 5.1112,
-      "step": 23796
-    },
-    {
-      "epoch": 12.410430247718383,
-      "grad_norm": 1.512736439704895,
-      "learning_rate": 7.659396984924625e-05,
-      "loss": 5.1141,
-      "step": 23797
-    },
-    {
-      "epoch": 12.410951760104302,
-      "grad_norm": 1.5082546472549438,
-      "learning_rate": 7.659296482412061e-05,
-      "loss": 5.5431,
-      "step": 23798
-    },
-    {
-      "epoch": 12.411473272490221,
-      "grad_norm": 1.381402850151062,
-      "learning_rate": 7.659195979899499e-05,
-      "loss": 5.4434,
-      "step": 23799
-    },
-    {
-      "epoch": 12.411994784876141,
-      "grad_norm": 1.4765366315841675,
-      "learning_rate": 7.659095477386935e-05,
-      "loss": 4.9069,
-      "step": 23800
-    },
-    {
-      "epoch": 12.41251629726206,
-      "grad_norm": 1.4581681489944458,
-      "learning_rate": 7.658994974874372e-05,
-      "loss": 5.2014,
-      "step": 23801
-    },
-    {
-      "epoch": 12.41303780964798,
-      "grad_norm": 1.4212056398391724,
-      "learning_rate": 7.65889447236181e-05,
-      "loss": 5.5983,
-      "step": 23802
-    },
-    {
-      "epoch": 12.413559322033898,
-      "grad_norm": 1.5099722146987915,
-      "learning_rate": 7.658793969849246e-05,
-      "loss": 4.5845,
-      "step": 23803
-    },
-    {
-      "epoch": 12.414080834419817,
-      "grad_norm": 1.4099961519241333,
-      "learning_rate": 7.658693467336684e-05,
-      "loss": 5.5637,
-      "step": 23804
-    },
-    {
-      "epoch": 12.414602346805736,
-      "grad_norm": 1.442097544670105,
-      "learning_rate": 7.65859296482412e-05,
-      "loss": 4.9078,
-      "step": 23805
-    },
-    {
-      "epoch": 12.415123859191656,
-      "grad_norm": 1.4677298069000244,
-      "learning_rate": 7.658492462311558e-05,
-      "loss": 5.8991,
-      "step": 23806
-    },
-    {
-      "epoch": 12.415645371577575,
-      "grad_norm": 1.5004733800888062,
-      "learning_rate": 7.658391959798996e-05,
-      "loss": 5.3621,
-      "step": 23807
-    },
-    {
-      "epoch": 12.416166883963495,
-      "grad_norm": 1.522870421409607,
-      "learning_rate": 7.658291457286433e-05,
-      "loss": 5.3912,
-      "step": 23808
-    },
-    {
-      "epoch": 12.416688396349413,
-      "grad_norm": 1.4441994428634644,
-      "learning_rate": 7.65819095477387e-05,
-      "loss": 5.321,
-      "step": 23809
-    },
-    {
-      "epoch": 12.417209908735332,
-      "grad_norm": 1.4667942523956299,
-      "learning_rate": 7.658090452261308e-05,
-      "loss": 5.3715,
-      "step": 23810
-    },
-    {
-      "epoch": 12.417731421121252,
-      "grad_norm": 1.401685357093811,
-      "learning_rate": 7.657989949748744e-05,
-      "loss": 5.5363,
-      "step": 23811
-    },
-    {
-      "epoch": 12.418252933507171,
-      "grad_norm": 1.710755467414856,
-      "learning_rate": 7.657889447236182e-05,
-      "loss": 5.3003,
-      "step": 23812
-    },
-    {
-      "epoch": 12.41877444589309,
-      "grad_norm": 1.4062641859054565,
-      "learning_rate": 7.657788944723618e-05,
-      "loss": 5.4908,
-      "step": 23813
-    },
-    {
-      "epoch": 12.41929595827901,
-      "grad_norm": 1.4042850732803345,
-      "learning_rate": 7.657688442211056e-05,
-      "loss": 5.3088,
-      "step": 23814
-    },
-    {
-      "epoch": 12.419817470664928,
-      "grad_norm": 1.543470025062561,
-      "learning_rate": 7.657587939698492e-05,
-      "loss": 5.1533,
-      "step": 23815
-    },
-    {
-      "epoch": 12.420338983050847,
-      "grad_norm": 1.6130540370941162,
-      "learning_rate": 7.65748743718593e-05,
-      "loss": 5.6161,
-      "step": 23816
-    },
-    {
-      "epoch": 12.420860495436767,
-      "grad_norm": 1.4108381271362305,
-      "learning_rate": 7.657386934673368e-05,
-      "loss": 6.0046,
-      "step": 23817
-    },
-    {
-      "epoch": 12.421382007822686,
-      "grad_norm": 1.4215340614318848,
-      "learning_rate": 7.657286432160804e-05,
-      "loss": 5.2988,
-      "step": 23818
-    },
-    {
-      "epoch": 12.421903520208605,
-      "grad_norm": 1.5006026029586792,
-      "learning_rate": 7.657185929648242e-05,
-      "loss": 5.2794,
-      "step": 23819
-    },
-    {
-      "epoch": 12.422425032594525,
-      "grad_norm": 1.3946694135665894,
-      "learning_rate": 7.657085427135679e-05,
-      "loss": 5.4232,
-      "step": 23820
-    },
-    {
-      "epoch": 12.422946544980443,
-      "grad_norm": 1.3157027959823608,
-      "learning_rate": 7.656984924623116e-05,
-      "loss": 5.243,
-      "step": 23821
-    },
-    {
-      "epoch": 12.423468057366362,
-      "grad_norm": 1.5194106101989746,
-      "learning_rate": 7.656884422110553e-05,
-      "loss": 5.2074,
-      "step": 23822
-    },
-    {
-      "epoch": 12.423989569752282,
-      "grad_norm": 1.4282373189926147,
-      "learning_rate": 7.65678391959799e-05,
-      "loss": 5.3838,
-      "step": 23823
-    },
-    {
-      "epoch": 12.424511082138201,
-      "grad_norm": 1.4137600660324097,
-      "learning_rate": 7.656683417085427e-05,
-      "loss": 5.1618,
-      "step": 23824
-    },
-    {
-      "epoch": 12.42503259452412,
-      "grad_norm": 1.4104396104812622,
-      "learning_rate": 7.656582914572865e-05,
-      "loss": 5.6612,
-      "step": 23825
-    },
-    {
-      "epoch": 12.42555410691004,
-      "grad_norm": 1.5428458452224731,
-      "learning_rate": 7.656482412060301e-05,
-      "loss": 5.2756,
-      "step": 23826
-    },
-    {
-      "epoch": 12.426075619295958,
-      "grad_norm": 1.4886205196380615,
-      "learning_rate": 7.656381909547739e-05,
-      "loss": 5.2185,
-      "step": 23827
-    },
-    {
-      "epoch": 12.426597131681877,
-      "grad_norm": 1.5071150064468384,
-      "learning_rate": 7.656281407035177e-05,
-      "loss": 5.5885,
-      "step": 23828
-    },
-    {
-      "epoch": 12.427118644067797,
-      "grad_norm": 1.444319725036621,
-      "learning_rate": 7.656180904522613e-05,
-      "loss": 5.2856,
-      "step": 23829
-    },
-    {
-      "epoch": 12.427640156453716,
-      "grad_norm": 1.4940437078475952,
-      "learning_rate": 7.656080402010051e-05,
-      "loss": 5.2211,
-      "step": 23830
-    },
-    {
-      "epoch": 12.428161668839635,
-      "grad_norm": 1.3786982297897339,
-      "learning_rate": 7.655979899497487e-05,
-      "loss": 5.4861,
-      "step": 23831
-    },
-    {
-      "epoch": 12.428683181225555,
-      "grad_norm": 1.4851627349853516,
-      "learning_rate": 7.655879396984925e-05,
-      "loss": 4.2424,
-      "step": 23832
-    },
-    {
-      "epoch": 12.429204693611473,
-      "grad_norm": 1.4165419340133667,
-      "learning_rate": 7.655778894472362e-05,
-      "loss": 5.5186,
-      "step": 23833
-    },
-    {
-      "epoch": 12.429726205997392,
-      "grad_norm": 1.4130934476852417,
-      "learning_rate": 7.6556783919598e-05,
-      "loss": 5.7447,
-      "step": 23834
-    },
-    {
-      "epoch": 12.430247718383312,
-      "grad_norm": 1.4840353727340698,
-      "learning_rate": 7.655577889447236e-05,
-      "loss": 4.8314,
-      "step": 23835
-    },
-    {
-      "epoch": 12.430769230769231,
-      "grad_norm": 1.3252582550048828,
-      "learning_rate": 7.655477386934674e-05,
-      "loss": 5.8509,
-      "step": 23836
-    },
-    {
-      "epoch": 12.43129074315515,
-      "grad_norm": 1.3644167184829712,
-      "learning_rate": 7.655376884422111e-05,
-      "loss": 5.6896,
-      "step": 23837
-    },
-    {
-      "epoch": 12.43181225554107,
-      "grad_norm": 1.4254096746444702,
-      "learning_rate": 7.655276381909549e-05,
-      "loss": 5.5191,
-      "step": 23838
-    },
-    {
-      "epoch": 12.432333767926988,
-      "grad_norm": 1.4065399169921875,
-      "learning_rate": 7.655175879396986e-05,
-      "loss": 5.3711,
-      "step": 23839
-    },
-    {
-      "epoch": 12.432855280312907,
-      "grad_norm": 1.3681453466415405,
-      "learning_rate": 7.655075376884423e-05,
-      "loss": 4.8664,
-      "step": 23840
-    },
-    {
-      "epoch": 12.433376792698827,
-      "grad_norm": 1.4717154502868652,
-      "learning_rate": 7.65497487437186e-05,
-      "loss": 5.5235,
-      "step": 23841
-    },
-    {
-      "epoch": 12.433898305084746,
-      "grad_norm": 1.5091842412948608,
-      "learning_rate": 7.654874371859296e-05,
-      "loss": 5.4984,
-      "step": 23842
-    },
-    {
-      "epoch": 12.434419817470665,
-      "grad_norm": 1.4147522449493408,
-      "learning_rate": 7.654773869346734e-05,
-      "loss": 5.7321,
-      "step": 23843
-    },
-    {
-      "epoch": 12.434941329856585,
-      "grad_norm": 1.3422292470932007,
-      "learning_rate": 7.65467336683417e-05,
-      "loss": 5.677,
-      "step": 23844
-    },
-    {
-      "epoch": 12.435462842242503,
-      "grad_norm": 1.4607919454574585,
-      "learning_rate": 7.654572864321608e-05,
-      "loss": 5.0811,
-      "step": 23845
-    },
-    {
-      "epoch": 12.435984354628422,
-      "grad_norm": 1.520227313041687,
-      "learning_rate": 7.654472361809045e-05,
-      "loss": 5.1744,
-      "step": 23846
-    },
-    {
-      "epoch": 12.436505867014342,
-      "grad_norm": 1.5551426410675049,
-      "learning_rate": 7.654371859296482e-05,
-      "loss": 5.5663,
-      "step": 23847
-    },
-    {
-      "epoch": 12.437027379400261,
-      "grad_norm": 1.4993596076965332,
-      "learning_rate": 7.65427135678392e-05,
-      "loss": 4.9552,
-      "step": 23848
-    },
-    {
-      "epoch": 12.43754889178618,
-      "grad_norm": 1.540223240852356,
-      "learning_rate": 7.654170854271358e-05,
-      "loss": 5.5629,
-      "step": 23849
-    },
-    {
-      "epoch": 12.438070404172098,
-      "grad_norm": 1.3231174945831299,
-      "learning_rate": 7.654070351758794e-05,
-      "loss": 5.7494,
-      "step": 23850
-    },
-    {
-      "epoch": 12.438591916558018,
-      "grad_norm": 1.4028513431549072,
-      "learning_rate": 7.653969849246232e-05,
-      "loss": 5.6408,
-      "step": 23851
-    },
-    {
-      "epoch": 12.439113428943937,
-      "grad_norm": 1.3757060766220093,
-      "learning_rate": 7.653869346733669e-05,
-      "loss": 4.8704,
-      "step": 23852
-    },
-    {
-      "epoch": 12.439634941329857,
-      "grad_norm": 1.3621817827224731,
-      "learning_rate": 7.653768844221106e-05,
-      "loss": 5.6849,
-      "step": 23853
-    },
-    {
-      "epoch": 12.440156453715776,
-      "grad_norm": 1.476741075515747,
-      "learning_rate": 7.653668341708543e-05,
-      "loss": 5.5682,
-      "step": 23854
-    },
-    {
-      "epoch": 12.440677966101696,
-      "grad_norm": 1.5105949640274048,
-      "learning_rate": 7.653567839195979e-05,
-      "loss": 5.5357,
-      "step": 23855
-    },
-    {
-      "epoch": 12.441199478487615,
-      "grad_norm": 1.5366222858428955,
-      "learning_rate": 7.653467336683417e-05,
-      "loss": 5.0613,
-      "step": 23856
-    },
-    {
-      "epoch": 12.441720990873533,
-      "grad_norm": 1.5610473155975342,
-      "learning_rate": 7.653366834170854e-05,
-      "loss": 5.2283,
-      "step": 23857
-    },
-    {
-      "epoch": 12.442242503259452,
-      "grad_norm": 1.4782990217208862,
-      "learning_rate": 7.653266331658291e-05,
-      "loss": 5.7343,
-      "step": 23858
-    },
-    {
-      "epoch": 12.442764015645372,
-      "grad_norm": 1.3663442134857178,
-      "learning_rate": 7.653165829145729e-05,
-      "loss": 5.3092,
-      "step": 23859
-    },
-    {
-      "epoch": 12.443285528031291,
-      "grad_norm": 1.5326111316680908,
-      "learning_rate": 7.653065326633167e-05,
-      "loss": 4.9813,
-      "step": 23860
-    },
-    {
-      "epoch": 12.44380704041721,
-      "grad_norm": 1.5193885564804077,
-      "learning_rate": 7.652964824120603e-05,
-      "loss": 4.9787,
-      "step": 23861
-    },
-    {
-      "epoch": 12.444328552803128,
-      "grad_norm": 1.4454288482666016,
-      "learning_rate": 7.652864321608041e-05,
-      "loss": 5.7573,
-      "step": 23862
-    },
-    {
-      "epoch": 12.444850065189048,
-      "grad_norm": 1.3996063470840454,
-      "learning_rate": 7.652763819095478e-05,
-      "loss": 5.1412,
-      "step": 23863
-    },
-    {
-      "epoch": 12.445371577574967,
-      "grad_norm": 1.545922040939331,
-      "learning_rate": 7.652663316582915e-05,
-      "loss": 5.609,
-      "step": 23864
-    },
-    {
-      "epoch": 12.445893089960887,
-      "grad_norm": 1.3752044439315796,
-      "learning_rate": 7.652562814070352e-05,
-      "loss": 5.5444,
-      "step": 23865
-    },
-    {
-      "epoch": 12.446414602346806,
-      "grad_norm": 1.3943368196487427,
-      "learning_rate": 7.65246231155779e-05,
-      "loss": 5.384,
-      "step": 23866
-    },
-    {
-      "epoch": 12.446936114732726,
-      "grad_norm": 1.4490538835525513,
-      "learning_rate": 7.652361809045226e-05,
-      "loss": 5.6049,
-      "step": 23867
-    },
-    {
-      "epoch": 12.447457627118643,
-      "grad_norm": 1.377888798713684,
-      "learning_rate": 7.652261306532664e-05,
-      "loss": 5.7445,
-      "step": 23868
-    },
-    {
-      "epoch": 12.447979139504563,
-      "grad_norm": 1.462333083152771,
-      "learning_rate": 7.652160804020102e-05,
-      "loss": 5.1237,
-      "step": 23869
-    },
-    {
-      "epoch": 12.448500651890482,
-      "grad_norm": 1.4214469194412231,
-      "learning_rate": 7.652060301507538e-05,
-      "loss": 5.5969,
-      "step": 23870
-    },
-    {
-      "epoch": 12.449022164276402,
-      "grad_norm": 1.5922590494155884,
-      "learning_rate": 7.651959798994976e-05,
-      "loss": 5.0325,
-      "step": 23871
-    },
-    {
-      "epoch": 12.449543676662321,
-      "grad_norm": 1.505969524383545,
-      "learning_rate": 7.651859296482412e-05,
-      "loss": 5.651,
-      "step": 23872
-    },
-    {
-      "epoch": 12.45006518904824,
-      "grad_norm": 1.4821701049804688,
-      "learning_rate": 7.65175879396985e-05,
-      "loss": 5.5641,
-      "step": 23873
-    },
-    {
-      "epoch": 12.450586701434158,
-      "grad_norm": 1.410272240638733,
-      "learning_rate": 7.651658291457286e-05,
-      "loss": 5.5392,
-      "step": 23874
-    },
-    {
-      "epoch": 12.451108213820078,
-      "grad_norm": 1.487518072128296,
-      "learning_rate": 7.651557788944724e-05,
-      "loss": 5.384,
-      "step": 23875
-    },
-    {
-      "epoch": 12.451629726205997,
-      "grad_norm": 1.4635494947433472,
-      "learning_rate": 7.65145728643216e-05,
-      "loss": 5.3759,
-      "step": 23876
-    },
-    {
-      "epoch": 12.452151238591917,
-      "grad_norm": 1.4112154245376587,
-      "learning_rate": 7.651356783919598e-05,
-      "loss": 5.0038,
-      "step": 23877
-    },
-    {
-      "epoch": 12.452672750977836,
-      "grad_norm": 1.466858148574829,
-      "learning_rate": 7.651256281407035e-05,
-      "loss": 4.9729,
-      "step": 23878
-    },
-    {
-      "epoch": 12.453194263363756,
-      "grad_norm": 1.4696847200393677,
-      "learning_rate": 7.651155778894473e-05,
-      "loss": 4.9088,
-      "step": 23879
-    },
-    {
-      "epoch": 12.453715775749673,
-      "grad_norm": 1.559050440788269,
-      "learning_rate": 7.65105527638191e-05,
-      "loss": 4.7958,
-      "step": 23880
-    },
-    {
-      "epoch": 12.454237288135593,
-      "grad_norm": 1.5882033109664917,
-      "learning_rate": 7.650954773869347e-05,
-      "loss": 5.3624,
-      "step": 23881
-    },
-    {
-      "epoch": 12.454758800521512,
-      "grad_norm": 1.4567261934280396,
-      "learning_rate": 7.650854271356785e-05,
-      "loss": 5.3868,
-      "step": 23882
-    },
-    {
-      "epoch": 12.455280312907432,
-      "grad_norm": 1.5936743021011353,
-      "learning_rate": 7.650753768844221e-05,
-      "loss": 5.2319,
-      "step": 23883
-    },
-    {
-      "epoch": 12.455801825293351,
-      "grad_norm": 1.4073845148086548,
-      "learning_rate": 7.650653266331659e-05,
-      "loss": 5.609,
-      "step": 23884
-    },
-    {
-      "epoch": 12.45632333767927,
-      "grad_norm": 1.4026669263839722,
-      "learning_rate": 7.650552763819095e-05,
-      "loss": 5.5536,
-      "step": 23885
-    },
-    {
-      "epoch": 12.456844850065188,
-      "grad_norm": 1.5067285299301147,
-      "learning_rate": 7.650452261306533e-05,
-      "loss": 5.3227,
-      "step": 23886
-    },
-    {
-      "epoch": 12.457366362451108,
-      "grad_norm": 1.3606045246124268,
-      "learning_rate": 7.65035175879397e-05,
-      "loss": 5.8216,
-      "step": 23887
-    },
-    {
-      "epoch": 12.457887874837027,
-      "grad_norm": 1.4823493957519531,
-      "learning_rate": 7.650251256281407e-05,
-      "loss": 5.4332,
-      "step": 23888
-    },
-    {
-      "epoch": 12.458409387222947,
-      "grad_norm": 1.473591923713684,
-      "learning_rate": 7.650150753768845e-05,
-      "loss": 5.1253,
-      "step": 23889
-    },
-    {
-      "epoch": 12.458930899608866,
-      "grad_norm": 1.4161796569824219,
-      "learning_rate": 7.650050251256283e-05,
-      "loss": 5.5614,
-      "step": 23890
-    },
-    {
-      "epoch": 12.459452411994786,
-      "grad_norm": 1.4895586967468262,
-      "learning_rate": 7.649949748743719e-05,
-      "loss": 5.5382,
-      "step": 23891
-    },
-    {
-      "epoch": 12.459973924380703,
-      "grad_norm": 1.4927915334701538,
-      "learning_rate": 7.649849246231157e-05,
-      "loss": 5.2074,
-      "step": 23892
-    },
-    {
-      "epoch": 12.460495436766623,
-      "grad_norm": 1.4737502336502075,
-      "learning_rate": 7.649748743718593e-05,
-      "loss": 5.8044,
-      "step": 23893
-    },
-    {
-      "epoch": 12.461016949152542,
-      "grad_norm": 1.4642428159713745,
-      "learning_rate": 7.64964824120603e-05,
-      "loss": 5.3438,
-      "step": 23894
-    },
-    {
-      "epoch": 12.461538461538462,
-      "grad_norm": 1.3166215419769287,
-      "learning_rate": 7.649547738693468e-05,
-      "loss": 5.678,
-      "step": 23895
-    },
-    {
-      "epoch": 12.462059973924381,
-      "grad_norm": 1.4340157508850098,
-      "learning_rate": 7.649447236180904e-05,
-      "loss": 5.5556,
-      "step": 23896
-    },
-    {
-      "epoch": 12.4625814863103,
-      "grad_norm": 1.4314852952957153,
-      "learning_rate": 7.649346733668342e-05,
-      "loss": 5.1315,
-      "step": 23897
-    },
-    {
-      "epoch": 12.463102998696218,
-      "grad_norm": 1.5475008487701416,
-      "learning_rate": 7.649246231155778e-05,
-      "loss": 4.9837,
-      "step": 23898
-    },
-    {
-      "epoch": 12.463624511082138,
-      "grad_norm": 1.4987730979919434,
-      "learning_rate": 7.649145728643216e-05,
-      "loss": 5.7163,
-      "step": 23899
-    },
-    {
-      "epoch": 12.464146023468057,
-      "grad_norm": 1.4902095794677734,
-      "learning_rate": 7.649045226130654e-05,
-      "loss": 5.6731,
-      "step": 23900
-    },
-    {
-      "epoch": 12.464667535853977,
-      "grad_norm": 1.381601333618164,
-      "learning_rate": 7.648944723618092e-05,
-      "loss": 5.7253,
-      "step": 23901
-    },
-    {
-      "epoch": 12.465189048239896,
-      "grad_norm": 1.4353667497634888,
-      "learning_rate": 7.648844221105528e-05,
-      "loss": 5.7145,
-      "step": 23902
-    },
-    {
-      "epoch": 12.465710560625816,
-      "grad_norm": 1.5271464586257935,
-      "learning_rate": 7.648743718592966e-05,
-      "loss": 5.2344,
-      "step": 23903
-    },
-    {
-      "epoch": 12.466232073011733,
-      "grad_norm": 1.5510101318359375,
-      "learning_rate": 7.648643216080402e-05,
-      "loss": 5.222,
-      "step": 23904
-    },
-    {
-      "epoch": 12.466753585397653,
-      "grad_norm": 1.4215399026870728,
-      "learning_rate": 7.64854271356784e-05,
-      "loss": 5.731,
-      "step": 23905
-    },
-    {
-      "epoch": 12.467275097783572,
-      "grad_norm": 1.3483597040176392,
-      "learning_rate": 7.648442211055276e-05,
-      "loss": 5.9383,
-      "step": 23906
-    },
-    {
-      "epoch": 12.467796610169492,
-      "grad_norm": 1.4476491212844849,
-      "learning_rate": 7.648341708542714e-05,
-      "loss": 5.5246,
-      "step": 23907
-    },
-    {
-      "epoch": 12.468318122555411,
-      "grad_norm": 1.5162078142166138,
-      "learning_rate": 7.64824120603015e-05,
-      "loss": 5.642,
-      "step": 23908
-    },
-    {
-      "epoch": 12.46883963494133,
-      "grad_norm": 1.6466444730758667,
-      "learning_rate": 7.648140703517588e-05,
-      "loss": 5.2493,
-      "step": 23909
-    },
-    {
-      "epoch": 12.469361147327248,
-      "grad_norm": 1.4474380016326904,
-      "learning_rate": 7.648040201005026e-05,
-      "loss": 5.5958,
-      "step": 23910
-    },
-    {
-      "epoch": 12.469882659713168,
-      "grad_norm": 1.4507007598876953,
-      "learning_rate": 7.647939698492463e-05,
-      "loss": 5.6084,
-      "step": 23911
-    },
-    {
-      "epoch": 12.470404172099087,
-      "grad_norm": 1.5435689687728882,
-      "learning_rate": 7.6478391959799e-05,
-      "loss": 5.146,
-      "step": 23912
-    },
-    {
-      "epoch": 12.470925684485007,
-      "grad_norm": 1.3129143714904785,
-      "learning_rate": 7.647738693467337e-05,
-      "loss": 5.8203,
-      "step": 23913
-    },
-    {
-      "epoch": 12.471447196870926,
-      "grad_norm": 1.48931086063385,
-      "learning_rate": 7.647638190954775e-05,
-      "loss": 5.1595,
-      "step": 23914
-    },
-    {
-      "epoch": 12.471968709256846,
-      "grad_norm": 1.418338418006897,
-      "learning_rate": 7.647537688442211e-05,
-      "loss": 5.3893,
-      "step": 23915
-    },
-    {
-      "epoch": 12.472490221642763,
-      "grad_norm": 1.4487195014953613,
-      "learning_rate": 7.647437185929649e-05,
-      "loss": 5.3915,
-      "step": 23916
-    },
-    {
-      "epoch": 12.473011734028683,
-      "grad_norm": 1.4759330749511719,
-      "learning_rate": 7.647336683417085e-05,
-      "loss": 5.416,
-      "step": 23917
-    },
-    {
-      "epoch": 12.473533246414602,
-      "grad_norm": 1.4914275407791138,
-      "learning_rate": 7.647236180904523e-05,
-      "loss": 5.6065,
-      "step": 23918
-    },
-    {
-      "epoch": 12.474054758800522,
-      "grad_norm": 1.4382572174072266,
-      "learning_rate": 7.64713567839196e-05,
-      "loss": 4.7425,
-      "step": 23919
-    },
-    {
-      "epoch": 12.474576271186441,
-      "grad_norm": 1.3593164682388306,
-      "learning_rate": 7.647035175879397e-05,
-      "loss": 5.7138,
-      "step": 23920
-    },
-    {
-      "epoch": 12.47509778357236,
-      "grad_norm": 1.5651980638504028,
-      "learning_rate": 7.646934673366835e-05,
-      "loss": 5.1222,
-      "step": 23921
-    },
-    {
-      "epoch": 12.475619295958278,
-      "grad_norm": 1.4208489656448364,
-      "learning_rate": 7.646834170854271e-05,
-      "loss": 5.1233,
-      "step": 23922
-    },
-    {
-      "epoch": 12.476140808344198,
-      "grad_norm": 1.446416974067688,
-      "learning_rate": 7.646733668341709e-05,
-      "loss": 5.6009,
-      "step": 23923
-    },
-    {
-      "epoch": 12.476662320730117,
-      "grad_norm": 1.577379822731018,
-      "learning_rate": 7.646633165829146e-05,
-      "loss": 5.351,
-      "step": 23924
-    },
-    {
-      "epoch": 12.477183833116037,
-      "grad_norm": 1.4392658472061157,
-      "learning_rate": 7.646532663316583e-05,
-      "loss": 5.4865,
-      "step": 23925
-    },
-    {
-      "epoch": 12.477705345501956,
-      "grad_norm": 1.4650200605392456,
-      "learning_rate": 7.64643216080402e-05,
-      "loss": 5.0934,
-      "step": 23926
-    },
-    {
-      "epoch": 12.478226857887876,
-      "grad_norm": 1.5383466482162476,
-      "learning_rate": 7.646331658291458e-05,
-      "loss": 5.1927,
-      "step": 23927
-    },
-    {
-      "epoch": 12.478748370273793,
-      "grad_norm": 1.4945564270019531,
-      "learning_rate": 7.646231155778894e-05,
-      "loss": 5.4315,
-      "step": 23928
-    },
-    {
-      "epoch": 12.479269882659713,
-      "grad_norm": 1.5480318069458008,
-      "learning_rate": 7.646130653266332e-05,
-      "loss": 5.3006,
-      "step": 23929
-    },
-    {
-      "epoch": 12.479791395045632,
-      "grad_norm": 1.5284807682037354,
-      "learning_rate": 7.64603015075377e-05,
-      "loss": 5.2525,
-      "step": 23930
-    },
-    {
-      "epoch": 12.480312907431552,
-      "grad_norm": 1.4392942190170288,
-      "learning_rate": 7.645929648241207e-05,
-      "loss": 5.5731,
-      "step": 23931
-    },
-    {
-      "epoch": 12.480834419817471,
-      "grad_norm": 1.4416391849517822,
-      "learning_rate": 7.645829145728644e-05,
-      "loss": 5.5038,
-      "step": 23932
-    },
-    {
-      "epoch": 12.48135593220339,
-      "grad_norm": 1.5883125066757202,
-      "learning_rate": 7.645728643216082e-05,
-      "loss": 5.6548,
-      "step": 23933
-    },
-    {
-      "epoch": 12.481877444589308,
-      "grad_norm": 1.59064519405365,
-      "learning_rate": 7.645628140703518e-05,
-      "loss": 5.7752,
-      "step": 23934
-    },
-    {
-      "epoch": 12.482398956975228,
-      "grad_norm": 1.4246289730072021,
-      "learning_rate": 7.645527638190955e-05,
-      "loss": 5.6915,
-      "step": 23935
-    },
-    {
-      "epoch": 12.482920469361147,
-      "grad_norm": 1.5602648258209229,
-      "learning_rate": 7.645427135678392e-05,
-      "loss": 5.1131,
-      "step": 23936
-    },
-    {
-      "epoch": 12.483441981747067,
-      "grad_norm": 1.308302402496338,
-      "learning_rate": 7.645326633165829e-05,
-      "loss": 5.8375,
-      "step": 23937
-    },
-    {
-      "epoch": 12.483963494132986,
-      "grad_norm": 1.6943588256835938,
-      "learning_rate": 7.645226130653267e-05,
-      "loss": 4.9007,
-      "step": 23938
-    },
-    {
-      "epoch": 12.484485006518906,
-      "grad_norm": 1.6048747301101685,
-      "learning_rate": 7.645125628140703e-05,
-      "loss": 5.5039,
-      "step": 23939
-    },
-    {
-      "epoch": 12.485006518904823,
-      "grad_norm": 1.5620343685150146,
-      "learning_rate": 7.645025125628141e-05,
-      "loss": 5.1718,
-      "step": 23940
-    },
-    {
-      "epoch": 12.485528031290743,
-      "grad_norm": 1.4623560905456543,
-      "learning_rate": 7.644924623115578e-05,
-      "loss": 5.4313,
-      "step": 23941
-    },
-    {
-      "epoch": 12.486049543676662,
-      "grad_norm": 1.442973017692566,
-      "learning_rate": 7.644824120603016e-05,
-      "loss": 5.6549,
-      "step": 23942
-    },
-    {
-      "epoch": 12.486571056062582,
-      "grad_norm": 1.394755244255066,
-      "learning_rate": 7.644723618090453e-05,
-      "loss": 5.3332,
-      "step": 23943
-    },
-    {
-      "epoch": 12.487092568448501,
-      "grad_norm": 1.4716360569000244,
-      "learning_rate": 7.64462311557789e-05,
-      "loss": 5.6972,
-      "step": 23944
-    },
-    {
-      "epoch": 12.487614080834419,
-      "grad_norm": 1.557801365852356,
-      "learning_rate": 7.644522613065327e-05,
-      "loss": 5.0637,
-      "step": 23945
-    },
-    {
-      "epoch": 12.488135593220338,
-      "grad_norm": 1.5008137226104736,
-      "learning_rate": 7.644422110552765e-05,
-      "loss": 5.502,
-      "step": 23946
-    },
-    {
-      "epoch": 12.488657105606258,
-      "grad_norm": 1.6561548709869385,
-      "learning_rate": 7.644321608040201e-05,
-      "loss": 4.9433,
-      "step": 23947
-    },
-    {
-      "epoch": 12.489178617992177,
-      "grad_norm": 1.4328820705413818,
-      "learning_rate": 7.644221105527638e-05,
-      "loss": 5.0531,
-      "step": 23948
-    },
-    {
-      "epoch": 12.489700130378097,
-      "grad_norm": 1.4527838230133057,
-      "learning_rate": 7.644120603015075e-05,
-      "loss": 5.646,
-      "step": 23949
-    },
-    {
-      "epoch": 12.490221642764016,
-      "grad_norm": 1.5010101795196533,
-      "learning_rate": 7.644020100502513e-05,
-      "loss": 5.3127,
-      "step": 23950
-    },
-    {
-      "epoch": 12.490743155149936,
-      "grad_norm": 1.5378553867340088,
-      "learning_rate": 7.643919597989951e-05,
-      "loss": 5.214,
-      "step": 23951
-    },
-    {
-      "epoch": 12.491264667535853,
-      "grad_norm": 1.436081051826477,
-      "learning_rate": 7.643819095477387e-05,
-      "loss": 5.6131,
-      "step": 23952
-    },
-    {
-      "epoch": 12.491786179921773,
-      "grad_norm": 1.431287169456482,
-      "learning_rate": 7.643718592964825e-05,
-      "loss": 5.1902,
-      "step": 23953
-    },
-    {
-      "epoch": 12.492307692307692,
-      "grad_norm": 1.4422441720962524,
-      "learning_rate": 7.643618090452262e-05,
-      "loss": 5.1914,
-      "step": 23954
-    },
-    {
-      "epoch": 12.492829204693612,
-      "grad_norm": 1.6241364479064941,
-      "learning_rate": 7.643517587939699e-05,
-      "loss": 5.0827,
-      "step": 23955
-    },
-    {
-      "epoch": 12.493350717079531,
-      "grad_norm": 1.4172743558883667,
-      "learning_rate": 7.643417085427136e-05,
-      "loss": 5.1411,
-      "step": 23956
-    },
-    {
-      "epoch": 12.493872229465449,
-      "grad_norm": 1.3096743822097778,
-      "learning_rate": 7.643316582914574e-05,
-      "loss": 5.87,
-      "step": 23957
-    },
-    {
-      "epoch": 12.494393741851368,
-      "grad_norm": 1.3828628063201904,
-      "learning_rate": 7.64321608040201e-05,
-      "loss": 5.5224,
-      "step": 23958
-    },
-    {
-      "epoch": 12.494915254237288,
-      "grad_norm": 1.4948405027389526,
-      "learning_rate": 7.643115577889448e-05,
-      "loss": 5.3998,
-      "step": 23959
-    },
-    {
-      "epoch": 12.495436766623207,
-      "grad_norm": 1.349073052406311,
-      "learning_rate": 7.643015075376884e-05,
-      "loss": 5.7904,
-      "step": 23960
-    },
-    {
-      "epoch": 12.495958279009127,
-      "grad_norm": 1.5361419916152954,
-      "learning_rate": 7.642914572864322e-05,
-      "loss": 5.3808,
-      "step": 23961
-    },
-    {
-      "epoch": 12.496479791395046,
-      "grad_norm": 1.447091817855835,
-      "learning_rate": 7.64281407035176e-05,
-      "loss": 4.7208,
-      "step": 23962
-    },
-    {
-      "epoch": 12.497001303780964,
-      "grad_norm": 1.4541609287261963,
-      "learning_rate": 7.642713567839196e-05,
-      "loss": 5.1941,
-      "step": 23963
-    },
-    {
-      "epoch": 12.497522816166883,
-      "grad_norm": 1.4621177911758423,
-      "learning_rate": 7.642613065326634e-05,
-      "loss": 5.4574,
-      "step": 23964
-    },
-    {
-      "epoch": 12.498044328552803,
-      "grad_norm": 1.5041385889053345,
-      "learning_rate": 7.64251256281407e-05,
-      "loss": 5.4247,
-      "step": 23965
-    },
-    {
-      "epoch": 12.498565840938722,
-      "grad_norm": 1.4481472969055176,
-      "learning_rate": 7.642412060301508e-05,
-      "loss": 5.912,
-      "step": 23966
-    },
-    {
-      "epoch": 12.499087353324642,
-      "grad_norm": 1.5137687921524048,
-      "learning_rate": 7.642311557788945e-05,
-      "loss": 5.4625,
-      "step": 23967
-    },
-    {
-      "epoch": 12.499608865710561,
-      "grad_norm": 1.4013690948486328,
-      "learning_rate": 7.642211055276382e-05,
-      "loss": 5.4516,
-      "step": 23968
-    },
-    {
-      "epoch": 12.500130378096479,
-      "grad_norm": 1.470645785331726,
-      "learning_rate": 7.642110552763819e-05,
-      "loss": 5.4676,
-      "step": 23969
-    },
-    {
-      "epoch": 12.500651890482398,
-      "grad_norm": 1.4191772937774658,
-      "learning_rate": 7.642010050251257e-05,
-      "loss": 5.7481,
-      "step": 23970
-    },
-    {
-      "epoch": 12.501173402868318,
-      "grad_norm": 1.3853963613510132,
-      "learning_rate": 7.641909547738694e-05,
-      "loss": 5.5958,
-      "step": 23971
-    },
-    {
-      "epoch": 12.501694915254237,
-      "grad_norm": 1.5253382921218872,
-      "learning_rate": 7.641809045226132e-05,
-      "loss": 5.2485,
-      "step": 23972
-    },
-    {
-      "epoch": 12.502216427640157,
-      "grad_norm": 1.407069206237793,
-      "learning_rate": 7.641708542713569e-05,
-      "loss": 5.8219,
-      "step": 23973
-    },
-    {
-      "epoch": 12.502737940026076,
-      "grad_norm": 1.477566123008728,
-      "learning_rate": 7.641608040201005e-05,
-      "loss": 5.0922,
-      "step": 23974
-    },
-    {
-      "epoch": 12.503259452411994,
-      "grad_norm": 1.5209016799926758,
-      "learning_rate": 7.641507537688443e-05,
-      "loss": 5.5627,
-      "step": 23975
-    },
-    {
-      "epoch": 12.503780964797913,
-      "grad_norm": 1.3633991479873657,
-      "learning_rate": 7.641407035175879e-05,
-      "loss": 5.8327,
-      "step": 23976
-    },
-    {
-      "epoch": 12.504302477183833,
-      "grad_norm": 1.4334821701049805,
-      "learning_rate": 7.641306532663317e-05,
-      "loss": 5.5376,
-      "step": 23977
-    },
-    {
-      "epoch": 12.504823989569752,
-      "grad_norm": 1.4631969928741455,
-      "learning_rate": 7.641206030150753e-05,
-      "loss": 5.2542,
-      "step": 23978
-    },
-    {
-      "epoch": 12.505345501955672,
-      "grad_norm": 1.6088911294937134,
-      "learning_rate": 7.641105527638191e-05,
-      "loss": 5.2371,
-      "step": 23979
-    },
-    {
-      "epoch": 12.505867014341591,
-      "grad_norm": 1.4348424673080444,
-      "learning_rate": 7.641005025125628e-05,
-      "loss": 4.975,
-      "step": 23980
-    },
-    {
-      "epoch": 12.506388526727509,
-      "grad_norm": 1.61935293674469,
-      "learning_rate": 7.640904522613065e-05,
-      "loss": 5.4608,
-      "step": 23981
-    },
-    {
-      "epoch": 12.506910039113428,
-      "grad_norm": 1.4549404382705688,
-      "learning_rate": 7.640804020100503e-05,
-      "loss": 5.3078,
-      "step": 23982
-    },
-    {
-      "epoch": 12.507431551499348,
-      "grad_norm": 1.5335890054702759,
-      "learning_rate": 7.640703517587941e-05,
-      "loss": 5.4256,
-      "step": 23983
-    },
-    {
-      "epoch": 12.507953063885267,
-      "grad_norm": 1.794306755065918,
-      "learning_rate": 7.640603015075377e-05,
-      "loss": 5.4584,
-      "step": 23984
-    },
-    {
-      "epoch": 12.508474576271187,
-      "grad_norm": 1.5369359254837036,
-      "learning_rate": 7.640502512562815e-05,
-      "loss": 5.3379,
-      "step": 23985
-    },
-    {
-      "epoch": 12.508996088657106,
-      "grad_norm": 1.442427396774292,
-      "learning_rate": 7.640402010050252e-05,
-      "loss": 5.6744,
-      "step": 23986
-    },
-    {
-      "epoch": 12.509517601043024,
-      "grad_norm": 1.3501238822937012,
-      "learning_rate": 7.64030150753769e-05,
-      "loss": 5.7066,
-      "step": 23987
-    },
-    {
-      "epoch": 12.510039113428943,
-      "grad_norm": 1.4165916442871094,
-      "learning_rate": 7.640201005025126e-05,
-      "loss": 5.3645,
-      "step": 23988
-    },
-    {
-      "epoch": 12.510560625814863,
-      "grad_norm": 1.65791654586792,
-      "learning_rate": 7.640100502512562e-05,
-      "loss": 5.4648,
-      "step": 23989
-    },
-    {
-      "epoch": 12.511082138200782,
-      "grad_norm": 1.428136944770813,
-      "learning_rate": 7.64e-05,
-      "loss": 5.5408,
-      "step": 23990
-    },
-    {
-      "epoch": 12.511603650586702,
-      "grad_norm": 1.5523273944854736,
-      "learning_rate": 7.639899497487438e-05,
-      "loss": 5.4954,
-      "step": 23991
-    },
-    {
-      "epoch": 12.512125162972621,
-      "grad_norm": 1.485776662826538,
-      "learning_rate": 7.639798994974876e-05,
-      "loss": 5.5272,
-      "step": 23992
-    },
-    {
-      "epoch": 12.512646675358539,
-      "grad_norm": 1.5057960748672485,
-      "learning_rate": 7.639698492462312e-05,
-      "loss": 5.5043,
-      "step": 23993
-    },
-    {
-      "epoch": 12.513168187744458,
-      "grad_norm": 1.5747747421264648,
-      "learning_rate": 7.63959798994975e-05,
-      "loss": 5.3336,
-      "step": 23994
-    },
-    {
-      "epoch": 12.513689700130378,
-      "grad_norm": 1.5737559795379639,
-      "learning_rate": 7.639497487437186e-05,
-      "loss": 5.3503,
-      "step": 23995
-    },
-    {
-      "epoch": 12.514211212516297,
-      "grad_norm": 1.345757246017456,
-      "learning_rate": 7.639396984924624e-05,
-      "loss": 5.3754,
-      "step": 23996
-    },
-    {
-      "epoch": 12.514732724902217,
-      "grad_norm": 1.3995518684387207,
-      "learning_rate": 7.63929648241206e-05,
-      "loss": 5.7906,
-      "step": 23997
-    },
-    {
-      "epoch": 12.515254237288136,
-      "grad_norm": 1.5265735387802124,
-      "learning_rate": 7.639195979899498e-05,
-      "loss": 5.1573,
-      "step": 23998
-    },
-    {
-      "epoch": 12.515775749674054,
-      "grad_norm": 1.5353530645370483,
-      "learning_rate": 7.639095477386935e-05,
-      "loss": 5.4419,
-      "step": 23999
-    },
-    {
-      "epoch": 12.516297262059974,
-      "grad_norm": 1.4458953142166138,
-      "learning_rate": 7.638994974874372e-05,
-      "loss": 5.6868,
-      "step": 24000
-    },
-    {
-      "epoch": 12.516818774445893,
-      "grad_norm": 1.326342225074768,
-      "learning_rate": 7.638894472361809e-05,
-      "loss": 5.685,
-      "step": 24001
-    },
-    {
-      "epoch": 12.517340286831812,
-      "grad_norm": 1.3961377143859863,
-      "learning_rate": 7.638793969849247e-05,
-      "loss": 5.6772,
-      "step": 24002
-    },
-    {
-      "epoch": 12.517861799217732,
-      "grad_norm": 1.3921561241149902,
-      "learning_rate": 7.638693467336684e-05,
-      "loss": 5.2917,
-      "step": 24003
-    },
-    {
-      "epoch": 12.518383311603651,
-      "grad_norm": 1.5722582340240479,
-      "learning_rate": 7.638592964824121e-05,
-      "loss": 4.948,
-      "step": 24004
-    },
-    {
-      "epoch": 12.518904823989569,
-      "grad_norm": 1.373491644859314,
-      "learning_rate": 7.638492462311559e-05,
-      "loss": 5.6027,
-      "step": 24005
-    },
-    {
-      "epoch": 12.519426336375489,
-      "grad_norm": 1.5441336631774902,
-      "learning_rate": 7.638391959798995e-05,
-      "loss": 5.2614,
-      "step": 24006
-    },
-    {
-      "epoch": 12.519947848761408,
-      "grad_norm": 1.7612119913101196,
-      "learning_rate": 7.638291457286433e-05,
-      "loss": 5.4208,
-      "step": 24007
-    },
-    {
-      "epoch": 12.520469361147327,
-      "grad_norm": 1.5488641262054443,
-      "learning_rate": 7.638190954773869e-05,
-      "loss": 5.6309,
-      "step": 24008
-    },
-    {
-      "epoch": 12.520990873533247,
-      "grad_norm": 1.3743047714233398,
-      "learning_rate": 7.638090452261307e-05,
-      "loss": 5.8355,
-      "step": 24009
-    },
-    {
-      "epoch": 12.521512385919166,
-      "grad_norm": 1.5425219535827637,
-      "learning_rate": 7.637989949748743e-05,
-      "loss": 5.2699,
-      "step": 24010
-    },
-    {
-      "epoch": 12.522033898305084,
-      "grad_norm": 1.4721312522888184,
-      "learning_rate": 7.637889447236181e-05,
-      "loss": 5.0939,
-      "step": 24011
-    },
-    {
-      "epoch": 12.522555410691004,
-      "grad_norm": 1.549061894416809,
-      "learning_rate": 7.637788944723619e-05,
-      "loss": 4.9992,
-      "step": 24012
-    },
-    {
-      "epoch": 12.523076923076923,
-      "grad_norm": 1.5016804933547974,
-      "learning_rate": 7.637688442211057e-05,
-      "loss": 5.4633,
-      "step": 24013
-    },
-    {
-      "epoch": 12.523598435462842,
-      "grad_norm": 1.5122665166854858,
-      "learning_rate": 7.637587939698493e-05,
-      "loss": 5.2116,
-      "step": 24014
-    },
-    {
-      "epoch": 12.524119947848762,
-      "grad_norm": 1.4193669557571411,
-      "learning_rate": 7.63748743718593e-05,
-      "loss": 5.2719,
-      "step": 24015
-    },
-    {
-      "epoch": 12.524641460234681,
-      "grad_norm": 1.5420687198638916,
-      "learning_rate": 7.637386934673367e-05,
-      "loss": 4.9329,
-      "step": 24016
-    },
-    {
-      "epoch": 12.525162972620599,
-      "grad_norm": 1.4554176330566406,
-      "learning_rate": 7.637286432160804e-05,
-      "loss": 5.7281,
-      "step": 24017
-    },
-    {
-      "epoch": 12.525684485006519,
-      "grad_norm": 1.5088129043579102,
-      "learning_rate": 7.637185929648242e-05,
-      "loss": 5.3336,
-      "step": 24018
-    },
-    {
-      "epoch": 12.526205997392438,
-      "grad_norm": 1.4380431175231934,
-      "learning_rate": 7.637085427135678e-05,
-      "loss": 5.6608,
-      "step": 24019
-    },
-    {
-      "epoch": 12.526727509778357,
-      "grad_norm": 1.5412853956222534,
-      "learning_rate": 7.636984924623116e-05,
-      "loss": 5.3897,
-      "step": 24020
-    },
-    {
-      "epoch": 12.527249022164277,
-      "grad_norm": 1.4521303176879883,
-      "learning_rate": 7.636884422110552e-05,
-      "loss": 5.2485,
-      "step": 24021
-    },
-    {
-      "epoch": 12.527770534550196,
-      "grad_norm": 1.5471569299697876,
-      "learning_rate": 7.63678391959799e-05,
-      "loss": 5.3092,
-      "step": 24022
-    },
-    {
-      "epoch": 12.528292046936114,
-      "grad_norm": 1.350314736366272,
-      "learning_rate": 7.636683417085428e-05,
-      "loss": 5.0287,
-      "step": 24023
-    },
-    {
-      "epoch": 12.528813559322034,
-      "grad_norm": 1.4373352527618408,
-      "learning_rate": 7.636582914572866e-05,
-      "loss": 5.2534,
-      "step": 24024
-    },
-    {
-      "epoch": 12.529335071707953,
-      "grad_norm": 1.428499460220337,
-      "learning_rate": 7.636482412060302e-05,
-      "loss": 5.3649,
-      "step": 24025
-    },
-    {
-      "epoch": 12.529856584093872,
-      "grad_norm": 1.3979252576828003,
-      "learning_rate": 7.63638190954774e-05,
-      "loss": 5.6563,
-      "step": 24026
-    },
-    {
-      "epoch": 12.530378096479792,
-      "grad_norm": 1.3768976926803589,
-      "learning_rate": 7.636281407035176e-05,
-      "loss": 5.9811,
-      "step": 24027
-    },
-    {
-      "epoch": 12.53089960886571,
-      "grad_norm": 1.4787874221801758,
-      "learning_rate": 7.636180904522613e-05,
-      "loss": 5.4166,
-      "step": 24028
-    },
-    {
-      "epoch": 12.531421121251629,
-      "grad_norm": 1.3674734830856323,
-      "learning_rate": 7.63608040201005e-05,
-      "loss": 5.7595,
-      "step": 24029
-    },
-    {
-      "epoch": 12.531942633637549,
-      "grad_norm": 1.4548860788345337,
-      "learning_rate": 7.635979899497487e-05,
-      "loss": 5.339,
-      "step": 24030
-    },
-    {
-      "epoch": 12.532464146023468,
-      "grad_norm": 1.5006753206253052,
-      "learning_rate": 7.635879396984925e-05,
-      "loss": 5.4244,
-      "step": 24031
-    },
-    {
-      "epoch": 12.532985658409387,
-      "grad_norm": 1.5715025663375854,
-      "learning_rate": 7.635778894472361e-05,
-      "loss": 5.5989,
-      "step": 24032
-    },
-    {
-      "epoch": 12.533507170795307,
-      "grad_norm": 1.319155216217041,
-      "learning_rate": 7.635678391959799e-05,
-      "loss": 4.2722,
-      "step": 24033
-    },
-    {
-      "epoch": 12.534028683181226,
-      "grad_norm": 1.5723530054092407,
-      "learning_rate": 7.635577889447237e-05,
-      "loss": 5.4504,
-      "step": 24034
-    },
-    {
-      "epoch": 12.534550195567144,
-      "grad_norm": 1.464119553565979,
-      "learning_rate": 7.635477386934675e-05,
-      "loss": 5.7512,
-      "step": 24035
-    },
-    {
-      "epoch": 12.535071707953064,
-      "grad_norm": 1.5393651723861694,
-      "learning_rate": 7.635376884422111e-05,
-      "loss": 4.9948,
-      "step": 24036
-    },
-    {
-      "epoch": 12.535593220338983,
-      "grad_norm": 1.4583607912063599,
-      "learning_rate": 7.635276381909549e-05,
-      "loss": 5.6383,
-      "step": 24037
-    },
-    {
-      "epoch": 12.536114732724903,
-      "grad_norm": 1.5604584217071533,
-      "learning_rate": 7.635175879396985e-05,
-      "loss": 5.1859,
-      "step": 24038
-    },
-    {
-      "epoch": 12.536636245110822,
-      "grad_norm": 1.8999650478363037,
-      "learning_rate": 7.635075376884423e-05,
-      "loss": 4.7615,
-      "step": 24039
-    },
-    {
-      "epoch": 12.53715775749674,
-      "grad_norm": 1.4389922618865967,
-      "learning_rate": 7.63497487437186e-05,
-      "loss": 5.4352,
-      "step": 24040
-    },
-    {
-      "epoch": 12.53767926988266,
-      "grad_norm": 1.4579598903656006,
-      "learning_rate": 7.634874371859296e-05,
-      "loss": 5.2612,
-      "step": 24041
-    },
-    {
-      "epoch": 12.538200782268579,
-      "grad_norm": 1.406278133392334,
-      "learning_rate": 7.634773869346734e-05,
-      "loss": 5.8563,
-      "step": 24042
-    },
-    {
-      "epoch": 12.538722294654498,
-      "grad_norm": 1.4361164569854736,
-      "learning_rate": 7.634673366834171e-05,
-      "loss": 5.7577,
-      "step": 24043
-    },
-    {
-      "epoch": 12.539243807040418,
-      "grad_norm": 1.4122111797332764,
-      "learning_rate": 7.634572864321609e-05,
-      "loss": 5.5339,
-      "step": 24044
-    },
-    {
-      "epoch": 12.539765319426337,
-      "grad_norm": 1.5974842309951782,
-      "learning_rate": 7.634472361809046e-05,
-      "loss": 4.9756,
-      "step": 24045
-    },
-    {
-      "epoch": 12.540286831812256,
-      "grad_norm": 1.5577232837677002,
-      "learning_rate": 7.634371859296483e-05,
-      "loss": 5.0403,
-      "step": 24046
-    },
-    {
-      "epoch": 12.540808344198174,
-      "grad_norm": 1.7335478067398071,
-      "learning_rate": 7.63427135678392e-05,
-      "loss": 5.1093,
-      "step": 24047
-    },
-    {
-      "epoch": 12.541329856584094,
-      "grad_norm": 1.455938458442688,
-      "learning_rate": 7.634170854271358e-05,
-      "loss": 5.3593,
-      "step": 24048
-    },
-    {
-      "epoch": 12.541851368970013,
-      "grad_norm": 1.579758882522583,
-      "learning_rate": 7.634070351758794e-05,
-      "loss": 5.2255,
-      "step": 24049
-    },
-    {
-      "epoch": 12.542372881355933,
-      "grad_norm": 1.3703843355178833,
-      "learning_rate": 7.633969849246232e-05,
-      "loss": 5.7102,
-      "step": 24050
-    },
-    {
-      "epoch": 12.542894393741852,
-      "grad_norm": 1.5027378797531128,
-      "learning_rate": 7.633869346733668e-05,
-      "loss": 5.3348,
-      "step": 24051
-    },
-    {
-      "epoch": 12.54341590612777,
-      "grad_norm": 1.5968945026397705,
-      "learning_rate": 7.633768844221106e-05,
-      "loss": 5.0147,
-      "step": 24052
-    },
-    {
-      "epoch": 12.54393741851369,
-      "grad_norm": 1.4413731098175049,
-      "learning_rate": 7.633668341708542e-05,
-      "loss": 5.3156,
-      "step": 24053
-    },
-    {
-      "epoch": 12.544458930899609,
-      "grad_norm": 1.4975122213363647,
-      "learning_rate": 7.63356783919598e-05,
-      "loss": 5.2105,
-      "step": 24054
-    },
-    {
-      "epoch": 12.544980443285528,
-      "grad_norm": 1.6527618169784546,
-      "learning_rate": 7.633467336683418e-05,
-      "loss": 5.4189,
-      "step": 24055
-    },
-    {
-      "epoch": 12.545501955671448,
-      "grad_norm": 1.5941699743270874,
-      "learning_rate": 7.633366834170854e-05,
-      "loss": 5.2952,
-      "step": 24056
-    },
-    {
-      "epoch": 12.546023468057367,
-      "grad_norm": 1.4538296461105347,
-      "learning_rate": 7.633266331658292e-05,
-      "loss": 5.5466,
-      "step": 24057
-    },
-    {
-      "epoch": 12.546544980443285,
-      "grad_norm": 1.4288579225540161,
-      "learning_rate": 7.633165829145729e-05,
-      "loss": 5.2128,
-      "step": 24058
-    },
-    {
-      "epoch": 12.547066492829204,
-      "grad_norm": 1.4165767431259155,
-      "learning_rate": 7.633065326633166e-05,
-      "loss": 5.3245,
-      "step": 24059
-    },
-    {
-      "epoch": 12.547588005215124,
-      "grad_norm": 1.5150104761123657,
-      "learning_rate": 7.632964824120603e-05,
-      "loss": 5.2358,
-      "step": 24060
-    },
-    {
-      "epoch": 12.548109517601043,
-      "grad_norm": 1.3911129236221313,
-      "learning_rate": 7.63286432160804e-05,
-      "loss": 5.0671,
-      "step": 24061
-    },
-    {
-      "epoch": 12.548631029986963,
-      "grad_norm": 1.5053627490997314,
-      "learning_rate": 7.632763819095477e-05,
-      "loss": 5.3879,
-      "step": 24062
-    },
-    {
-      "epoch": 12.549152542372882,
-      "grad_norm": 1.4448109865188599,
-      "learning_rate": 7.632663316582915e-05,
-      "loss": 5.7534,
-      "step": 24063
-    },
-    {
-      "epoch": 12.5496740547588,
-      "grad_norm": 1.4452301263809204,
-      "learning_rate": 7.632562814070353e-05,
-      "loss": 5.4577,
-      "step": 24064
-    },
-    {
-      "epoch": 12.55019556714472,
-      "grad_norm": 1.4025191068649292,
-      "learning_rate": 7.63246231155779e-05,
-      "loss": 5.5259,
-      "step": 24065
-    },
-    {
-      "epoch": 12.550717079530639,
-      "grad_norm": 1.4997824430465698,
-      "learning_rate": 7.632361809045227e-05,
-      "loss": 5.5111,
-      "step": 24066
-    },
-    {
-      "epoch": 12.551238591916558,
-      "grad_norm": 1.4373723268508911,
-      "learning_rate": 7.632261306532663e-05,
-      "loss": 5.6024,
-      "step": 24067
-    },
-    {
-      "epoch": 12.551760104302478,
-      "grad_norm": 1.6126000881195068,
-      "learning_rate": 7.632160804020101e-05,
-      "loss": 5.2945,
-      "step": 24068
-    },
-    {
-      "epoch": 12.552281616688397,
-      "grad_norm": 1.521464467048645,
-      "learning_rate": 7.632060301507537e-05,
-      "loss": 5.0887,
-      "step": 24069
-    },
-    {
-      "epoch": 12.552803129074315,
-      "grad_norm": 1.5646284818649292,
-      "learning_rate": 7.631959798994975e-05,
-      "loss": 5.3947,
-      "step": 24070
-    },
-    {
-      "epoch": 12.553324641460234,
-      "grad_norm": 1.463364601135254,
-      "learning_rate": 7.631859296482412e-05,
-      "loss": 5.6258,
-      "step": 24071
-    },
-    {
-      "epoch": 12.553846153846154,
-      "grad_norm": 1.5667576789855957,
-      "learning_rate": 7.63175879396985e-05,
-      "loss": 5.3443,
-      "step": 24072
-    },
-    {
-      "epoch": 12.554367666232073,
-      "grad_norm": 1.327050805091858,
-      "learning_rate": 7.631658291457286e-05,
-      "loss": 5.692,
-      "step": 24073
-    },
-    {
-      "epoch": 12.554889178617993,
-      "grad_norm": 1.3768302202224731,
-      "learning_rate": 7.631557788944724e-05,
-      "loss": 5.6739,
-      "step": 24074
-    },
-    {
-      "epoch": 12.555410691003912,
-      "grad_norm": 1.4015555381774902,
-      "learning_rate": 7.631457286432161e-05,
-      "loss": 5.7283,
-      "step": 24075
-    },
-    {
-      "epoch": 12.55593220338983,
-      "grad_norm": 1.4619792699813843,
-      "learning_rate": 7.631356783919599e-05,
-      "loss": 5.6194,
-      "step": 24076
-    },
-    {
-      "epoch": 12.55645371577575,
-      "grad_norm": 1.3818918466567993,
-      "learning_rate": 7.631256281407036e-05,
-      "loss": 5.5992,
-      "step": 24077
-    },
-    {
-      "epoch": 12.556975228161669,
-      "grad_norm": 1.4130626916885376,
-      "learning_rate": 7.631155778894473e-05,
-      "loss": 5.845,
-      "step": 24078
-    },
-    {
-      "epoch": 12.557496740547588,
-      "grad_norm": 1.5316866636276245,
-      "learning_rate": 7.63105527638191e-05,
-      "loss": 5.2808,
-      "step": 24079
-    },
-    {
-      "epoch": 12.558018252933508,
-      "grad_norm": 1.4711416959762573,
-      "learning_rate": 7.630954773869348e-05,
-      "loss": 4.8518,
-      "step": 24080
-    },
-    {
-      "epoch": 12.558539765319427,
-      "grad_norm": 1.7254583835601807,
-      "learning_rate": 7.630854271356784e-05,
-      "loss": 5.1424,
-      "step": 24081
-    },
-    {
-      "epoch": 12.559061277705345,
-      "grad_norm": 1.496216893196106,
-      "learning_rate": 7.63075376884422e-05,
-      "loss": 5.3217,
-      "step": 24082
-    },
-    {
-      "epoch": 12.559582790091264,
-      "grad_norm": 1.6095421314239502,
-      "learning_rate": 7.630653266331658e-05,
-      "loss": 5.7075,
-      "step": 24083
-    },
-    {
-      "epoch": 12.560104302477184,
-      "grad_norm": 1.4454114437103271,
-      "learning_rate": 7.630552763819096e-05,
-      "loss": 5.605,
-      "step": 24084
-    },
-    {
-      "epoch": 12.560625814863103,
-      "grad_norm": 1.481951355934143,
-      "learning_rate": 7.630452261306534e-05,
-      "loss": 5.2712,
-      "step": 24085
-    },
-    {
-      "epoch": 12.561147327249023,
-      "grad_norm": 1.4036610126495361,
-      "learning_rate": 7.63035175879397e-05,
-      "loss": 5.4912,
-      "step": 24086
-    },
-    {
-      "epoch": 12.561668839634942,
-      "grad_norm": 1.4880784749984741,
-      "learning_rate": 7.630251256281408e-05,
-      "loss": 5.4649,
-      "step": 24087
-    },
-    {
-      "epoch": 12.56219035202086,
-      "grad_norm": 1.4286760091781616,
-      "learning_rate": 7.630150753768844e-05,
-      "loss": 5.1267,
-      "step": 24088
-    },
-    {
-      "epoch": 12.56271186440678,
-      "grad_norm": 1.4319508075714111,
-      "learning_rate": 7.630050251256282e-05,
-      "loss": 5.6124,
-      "step": 24089
-    },
-    {
-      "epoch": 12.563233376792699,
-      "grad_norm": 1.4441328048706055,
-      "learning_rate": 7.629949748743719e-05,
-      "loss": 5.5669,
-      "step": 24090
-    },
-    {
-      "epoch": 12.563754889178618,
-      "grad_norm": 1.4575923681259155,
-      "learning_rate": 7.629849246231156e-05,
-      "loss": 5.3625,
-      "step": 24091
-    },
-    {
-      "epoch": 12.564276401564538,
-      "grad_norm": 1.4259700775146484,
-      "learning_rate": 7.629748743718593e-05,
-      "loss": 5.6959,
-      "step": 24092
-    },
-    {
-      "epoch": 12.564797913950457,
-      "grad_norm": 1.4856853485107422,
-      "learning_rate": 7.62964824120603e-05,
-      "loss": 5.4471,
-      "step": 24093
-    },
-    {
-      "epoch": 12.565319426336375,
-      "grad_norm": 1.5089404582977295,
-      "learning_rate": 7.629547738693467e-05,
-      "loss": 5.6335,
-      "step": 24094
-    },
-    {
-      "epoch": 12.565840938722294,
-      "grad_norm": 1.4292603731155396,
-      "learning_rate": 7.629447236180905e-05,
-      "loss": 5.81,
-      "step": 24095
-    },
-    {
-      "epoch": 12.566362451108214,
-      "grad_norm": 1.414827823638916,
-      "learning_rate": 7.629346733668343e-05,
-      "loss": 5.2227,
-      "step": 24096
-    },
-    {
-      "epoch": 12.566883963494133,
-      "grad_norm": 1.5426223278045654,
-      "learning_rate": 7.629246231155779e-05,
-      "loss": 5.6166,
-      "step": 24097
-    },
-    {
-      "epoch": 12.567405475880053,
-      "grad_norm": 1.4248595237731934,
-      "learning_rate": 7.629145728643217e-05,
-      "loss": 5.7097,
-      "step": 24098
-    },
-    {
-      "epoch": 12.567926988265972,
-      "grad_norm": 1.4977900981903076,
-      "learning_rate": 7.629045226130653e-05,
-      "loss": 5.6689,
-      "step": 24099
-    },
-    {
-      "epoch": 12.56844850065189,
-      "grad_norm": 1.4485907554626465,
-      "learning_rate": 7.628944723618091e-05,
-      "loss": 5.5703,
-      "step": 24100
-    },
-    {
-      "epoch": 12.56897001303781,
-      "grad_norm": 1.3246238231658936,
-      "learning_rate": 7.628844221105528e-05,
-      "loss": 5.7299,
-      "step": 24101
-    },
-    {
-      "epoch": 12.569491525423729,
-      "grad_norm": 1.5607991218566895,
-      "learning_rate": 7.628743718592965e-05,
-      "loss": 5.0582,
-      "step": 24102
-    },
-    {
-      "epoch": 12.570013037809648,
-      "grad_norm": 1.3766207695007324,
-      "learning_rate": 7.628643216080402e-05,
-      "loss": 5.5735,
-      "step": 24103
-    },
-    {
-      "epoch": 12.570534550195568,
-      "grad_norm": 1.3481364250183105,
-      "learning_rate": 7.62854271356784e-05,
-      "loss": 5.6675,
-      "step": 24104
-    },
-    {
-      "epoch": 12.571056062581487,
-      "grad_norm": 1.4065724611282349,
-      "learning_rate": 7.628442211055277e-05,
-      "loss": 5.339,
-      "step": 24105
-    },
-    {
-      "epoch": 12.571577574967405,
-      "grad_norm": 1.3406007289886475,
-      "learning_rate": 7.628341708542715e-05,
-      "loss": 5.5998,
-      "step": 24106
-    },
-    {
-      "epoch": 12.572099087353324,
-      "grad_norm": 1.3822495937347412,
-      "learning_rate": 7.628241206030152e-05,
-      "loss": 5.6214,
-      "step": 24107
-    },
-    {
-      "epoch": 12.572620599739244,
-      "grad_norm": 1.5310370922088623,
-      "learning_rate": 7.628140703517588e-05,
-      "loss": 4.7087,
-      "step": 24108
-    },
-    {
-      "epoch": 12.573142112125163,
-      "grad_norm": 1.5979671478271484,
-      "learning_rate": 7.628040201005026e-05,
-      "loss": 5.0719,
-      "step": 24109
-    },
-    {
-      "epoch": 12.573663624511083,
-      "grad_norm": 1.4765231609344482,
-      "learning_rate": 7.627939698492462e-05,
-      "loss": 5.4973,
-      "step": 24110
-    },
-    {
-      "epoch": 12.574185136897002,
-      "grad_norm": 1.418381690979004,
-      "learning_rate": 7.6278391959799e-05,
-      "loss": 5.5665,
-      "step": 24111
-    },
-    {
-      "epoch": 12.57470664928292,
-      "grad_norm": 1.3813358545303345,
-      "learning_rate": 7.627738693467336e-05,
-      "loss": 5.7308,
-      "step": 24112
-    },
-    {
-      "epoch": 12.57522816166884,
-      "grad_norm": 1.4717144966125488,
-      "learning_rate": 7.627638190954774e-05,
-      "loss": 5.2331,
-      "step": 24113
-    },
-    {
-      "epoch": 12.575749674054759,
-      "grad_norm": 1.4739454984664917,
-      "learning_rate": 7.62753768844221e-05,
-      "loss": 5.198,
-      "step": 24114
-    },
-    {
-      "epoch": 12.576271186440678,
-      "grad_norm": 1.4204965829849243,
-      "learning_rate": 7.627437185929648e-05,
-      "loss": 5.2445,
-      "step": 24115
-    },
-    {
-      "epoch": 12.576792698826598,
-      "grad_norm": 1.4052542448043823,
-      "learning_rate": 7.627336683417086e-05,
-      "loss": 5.2774,
-      "step": 24116
-    },
-    {
-      "epoch": 12.577314211212517,
-      "grad_norm": 1.5703935623168945,
-      "learning_rate": 7.627236180904524e-05,
-      "loss": 4.8241,
-      "step": 24117
-    },
-    {
-      "epoch": 12.577835723598435,
-      "grad_norm": 1.3988990783691406,
-      "learning_rate": 7.62713567839196e-05,
-      "loss": 5.2675,
-      "step": 24118
-    },
-    {
-      "epoch": 12.578357235984354,
-      "grad_norm": 1.496049404144287,
-      "learning_rate": 7.627035175879398e-05,
-      "loss": 5.4987,
-      "step": 24119
-    },
-    {
-      "epoch": 12.578878748370274,
-      "grad_norm": 1.5364041328430176,
-      "learning_rate": 7.626934673366835e-05,
-      "loss": 5.3965,
-      "step": 24120
-    },
-    {
-      "epoch": 12.579400260756193,
-      "grad_norm": 1.3627725839614868,
-      "learning_rate": 7.626834170854271e-05,
-      "loss": 5.2062,
-      "step": 24121
-    },
-    {
-      "epoch": 12.579921773142113,
-      "grad_norm": 1.5016710758209229,
-      "learning_rate": 7.626733668341709e-05,
-      "loss": 4.7963,
-      "step": 24122
-    },
-    {
-      "epoch": 12.58044328552803,
-      "grad_norm": 1.4281171560287476,
-      "learning_rate": 7.626633165829145e-05,
-      "loss": 5.7446,
-      "step": 24123
-    },
-    {
-      "epoch": 12.58096479791395,
-      "grad_norm": 1.4082679748535156,
-      "learning_rate": 7.626532663316583e-05,
-      "loss": 5.7332,
-      "step": 24124
-    },
-    {
-      "epoch": 12.58148631029987,
-      "grad_norm": 1.4607770442962646,
-      "learning_rate": 7.626432160804021e-05,
-      "loss": 5.6527,
-      "step": 24125
-    },
-    {
-      "epoch": 12.582007822685789,
-      "grad_norm": 1.447011947631836,
-      "learning_rate": 7.626331658291459e-05,
-      "loss": 5.2131,
-      "step": 24126
-    },
-    {
-      "epoch": 12.582529335071708,
-      "grad_norm": 1.355242371559143,
-      "learning_rate": 7.626231155778895e-05,
-      "loss": 5.8521,
-      "step": 24127
-    },
-    {
-      "epoch": 12.583050847457628,
-      "grad_norm": 1.4031840562820435,
-      "learning_rate": 7.626130653266333e-05,
-      "loss": 5.6994,
-      "step": 24128
-    },
-    {
-      "epoch": 12.583572359843547,
-      "grad_norm": 1.4380722045898438,
-      "learning_rate": 7.626030150753769e-05,
-      "loss": 5.2785,
-      "step": 24129
-    },
-    {
-      "epoch": 12.584093872229465,
-      "grad_norm": 1.5059834718704224,
-      "learning_rate": 7.625929648241207e-05,
-      "loss": 5.2479,
-      "step": 24130
-    },
-    {
-      "epoch": 12.584615384615384,
-      "grad_norm": 1.5301505327224731,
-      "learning_rate": 7.625829145728643e-05,
-      "loss": 5.0478,
-      "step": 24131
-    },
-    {
-      "epoch": 12.585136897001304,
-      "grad_norm": 1.5937585830688477,
-      "learning_rate": 7.625728643216081e-05,
-      "loss": 5.5311,
-      "step": 24132
-    },
-    {
-      "epoch": 12.585658409387223,
-      "grad_norm": 1.4250609874725342,
-      "learning_rate": 7.625628140703518e-05,
-      "loss": 5.2276,
-      "step": 24133
-    },
-    {
-      "epoch": 12.586179921773143,
-      "grad_norm": 1.439974308013916,
-      "learning_rate": 7.625527638190954e-05,
-      "loss": 5.6186,
-      "step": 24134
-    },
-    {
-      "epoch": 12.58670143415906,
-      "grad_norm": 1.5494141578674316,
-      "learning_rate": 7.625427135678392e-05,
-      "loss": 5.4496,
-      "step": 24135
-    },
-    {
-      "epoch": 12.58722294654498,
-      "grad_norm": 1.498809576034546,
-      "learning_rate": 7.62532663316583e-05,
-      "loss": 5.0745,
-      "step": 24136
-    },
-    {
-      "epoch": 12.5877444589309,
-      "grad_norm": 1.4965367317199707,
-      "learning_rate": 7.625226130653267e-05,
-      "loss": 5.1224,
-      "step": 24137
-    },
-    {
-      "epoch": 12.588265971316819,
-      "grad_norm": 1.5342637300491333,
-      "learning_rate": 7.625125628140704e-05,
-      "loss": 5.1519,
-      "step": 24138
-    },
-    {
-      "epoch": 12.588787483702738,
-      "grad_norm": 1.491075873374939,
-      "learning_rate": 7.625025125628142e-05,
-      "loss": 5.3477,
-      "step": 24139
-    },
-    {
-      "epoch": 12.589308996088658,
-      "grad_norm": 1.3220551013946533,
-      "learning_rate": 7.624924623115578e-05,
-      "loss": 5.7869,
-      "step": 24140
-    },
-    {
-      "epoch": 12.589830508474577,
-      "grad_norm": 1.4122811555862427,
-      "learning_rate": 7.624824120603016e-05,
-      "loss": 5.226,
-      "step": 24141
-    },
-    {
-      "epoch": 12.590352020860495,
-      "grad_norm": 1.4936463832855225,
-      "learning_rate": 7.624723618090452e-05,
-      "loss": 5.7548,
-      "step": 24142
-    },
-    {
-      "epoch": 12.590873533246414,
-      "grad_norm": 1.5473060607910156,
-      "learning_rate": 7.62462311557789e-05,
-      "loss": 5.0599,
-      "step": 24143
-    },
-    {
-      "epoch": 12.591395045632334,
-      "grad_norm": 1.491936206817627,
-      "learning_rate": 7.624522613065326e-05,
-      "loss": 5.5759,
-      "step": 24144
-    },
-    {
-      "epoch": 12.591916558018253,
-      "grad_norm": 1.4325995445251465,
-      "learning_rate": 7.624422110552764e-05,
-      "loss": 5.3913,
-      "step": 24145
-    },
-    {
-      "epoch": 12.592438070404173,
-      "grad_norm": 1.6516683101654053,
-      "learning_rate": 7.624321608040202e-05,
-      "loss": 4.3424,
-      "step": 24146
-    },
-    {
-      "epoch": 12.59295958279009,
-      "grad_norm": 1.422881007194519,
-      "learning_rate": 7.624221105527638e-05,
-      "loss": 5.312,
-      "step": 24147
-    },
-    {
-      "epoch": 12.59348109517601,
-      "grad_norm": 1.4374313354492188,
-      "learning_rate": 7.624120603015076e-05,
-      "loss": 5.4347,
-      "step": 24148
-    },
-    {
-      "epoch": 12.59400260756193,
-      "grad_norm": 1.5058659315109253,
-      "learning_rate": 7.624020100502513e-05,
-      "loss": 4.9774,
-      "step": 24149
-    },
-    {
-      "epoch": 12.594524119947849,
-      "grad_norm": 1.4113134145736694,
-      "learning_rate": 7.62391959798995e-05,
-      "loss": 5.6875,
-      "step": 24150
-    },
-    {
-      "epoch": 12.595045632333768,
-      "grad_norm": 1.4116578102111816,
-      "learning_rate": 7.623819095477387e-05,
-      "loss": 5.1119,
-      "step": 24151
-    },
-    {
-      "epoch": 12.595567144719688,
-      "grad_norm": 1.4490816593170166,
-      "learning_rate": 7.623718592964825e-05,
-      "loss": 5.3124,
-      "step": 24152
-    },
-    {
-      "epoch": 12.596088657105605,
-      "grad_norm": 1.430983543395996,
-      "learning_rate": 7.623618090452261e-05,
-      "loss": 5.4707,
-      "step": 24153
-    },
-    {
-      "epoch": 12.596610169491525,
-      "grad_norm": 1.5544211864471436,
-      "learning_rate": 7.623517587939699e-05,
-      "loss": 5.1152,
-      "step": 24154
-    },
-    {
-      "epoch": 12.597131681877444,
-      "grad_norm": 1.4214533567428589,
-      "learning_rate": 7.623417085427135e-05,
-      "loss": 5.2389,
-      "step": 24155
-    },
-    {
-      "epoch": 12.597653194263364,
-      "grad_norm": 1.6084270477294922,
-      "learning_rate": 7.623316582914573e-05,
-      "loss": 5.1936,
-      "step": 24156
-    },
-    {
-      "epoch": 12.598174706649283,
-      "grad_norm": 1.411137580871582,
-      "learning_rate": 7.623216080402011e-05,
-      "loss": 5.2389,
-      "step": 24157
-    },
-    {
-      "epoch": 12.598696219035203,
-      "grad_norm": 1.390541911125183,
-      "learning_rate": 7.623115577889449e-05,
-      "loss": 5.0296,
-      "step": 24158
-    },
-    {
-      "epoch": 12.59921773142112,
-      "grad_norm": 1.5025936365127563,
-      "learning_rate": 7.623015075376885e-05,
-      "loss": 5.0984,
-      "step": 24159
-    },
-    {
-      "epoch": 12.59973924380704,
-      "grad_norm": 1.4402844905853271,
-      "learning_rate": 7.622914572864321e-05,
-      "loss": 5.108,
-      "step": 24160
-    },
-    {
-      "epoch": 12.60026075619296,
-      "grad_norm": 1.3617522716522217,
-      "learning_rate": 7.622814070351759e-05,
-      "loss": 5.5347,
-      "step": 24161
-    },
-    {
-      "epoch": 12.600782268578879,
-      "grad_norm": 1.3534892797470093,
-      "learning_rate": 7.622713567839196e-05,
-      "loss": 5.1745,
-      "step": 24162
-    },
-    {
-      "epoch": 12.601303780964798,
-      "grad_norm": 1.454686164855957,
-      "learning_rate": 7.622613065326633e-05,
-      "loss": 5.5468,
-      "step": 24163
-    },
-    {
-      "epoch": 12.601825293350718,
-      "grad_norm": 1.3817729949951172,
-      "learning_rate": 7.62251256281407e-05,
-      "loss": 5.4126,
-      "step": 24164
-    },
-    {
-      "epoch": 12.602346805736635,
-      "grad_norm": 1.4556024074554443,
-      "learning_rate": 7.622412060301508e-05,
-      "loss": 4.6736,
-      "step": 24165
-    },
-    {
-      "epoch": 12.602868318122555,
-      "grad_norm": 1.3483344316482544,
-      "learning_rate": 7.622311557788945e-05,
-      "loss": 5.7941,
-      "step": 24166
-    },
-    {
-      "epoch": 12.603389830508474,
-      "grad_norm": 1.5422853231430054,
-      "learning_rate": 7.622211055276383e-05,
-      "loss": 5.2449,
-      "step": 24167
-    },
-    {
-      "epoch": 12.603911342894394,
-      "grad_norm": 1.4208157062530518,
-      "learning_rate": 7.62211055276382e-05,
-      "loss": 5.4183,
-      "step": 24168
-    },
-    {
-      "epoch": 12.604432855280313,
-      "grad_norm": 1.514634370803833,
-      "learning_rate": 7.622010050251257e-05,
-      "loss": 4.761,
-      "step": 24169
-    },
-    {
-      "epoch": 12.604954367666233,
-      "grad_norm": 1.5216585397720337,
-      "learning_rate": 7.621909547738694e-05,
-      "loss": 4.8503,
-      "step": 24170
-    },
-    {
-      "epoch": 12.60547588005215,
-      "grad_norm": 1.4797309637069702,
-      "learning_rate": 7.621809045226132e-05,
-      "loss": 5.4263,
-      "step": 24171
-    },
-    {
-      "epoch": 12.60599739243807,
-      "grad_norm": 1.4402559995651245,
-      "learning_rate": 7.621708542713568e-05,
-      "loss": 5.6771,
-      "step": 24172
-    },
-    {
-      "epoch": 12.60651890482399,
-      "grad_norm": 1.4043331146240234,
-      "learning_rate": 7.621608040201006e-05,
-      "loss": 5.6299,
-      "step": 24173
-    },
-    {
-      "epoch": 12.607040417209909,
-      "grad_norm": 1.594487190246582,
-      "learning_rate": 7.621507537688442e-05,
-      "loss": 5.0859,
-      "step": 24174
-    },
-    {
-      "epoch": 12.607561929595828,
-      "grad_norm": 1.631951928138733,
-      "learning_rate": 7.621407035175879e-05,
-      "loss": 5.3209,
-      "step": 24175
-    },
-    {
-      "epoch": 12.608083441981748,
-      "grad_norm": 1.357652187347412,
-      "learning_rate": 7.621306532663317e-05,
-      "loss": 5.9911,
-      "step": 24176
-    },
-    {
-      "epoch": 12.608604954367665,
-      "grad_norm": 1.6444132328033447,
-      "learning_rate": 7.621206030150754e-05,
-      "loss": 5.2708,
-      "step": 24177
-    },
-    {
-      "epoch": 12.609126466753585,
-      "grad_norm": 1.508195400238037,
-      "learning_rate": 7.621105527638192e-05,
-      "loss": 5.7682,
-      "step": 24178
-    },
-    {
-      "epoch": 12.609647979139504,
-      "grad_norm": 1.5563690662384033,
-      "learning_rate": 7.621005025125628e-05,
-      "loss": 5.2938,
-      "step": 24179
-    },
-    {
-      "epoch": 12.610169491525424,
-      "grad_norm": 1.2951006889343262,
-      "learning_rate": 7.620904522613066e-05,
-      "loss": 5.3951,
-      "step": 24180
-    },
-    {
-      "epoch": 12.610691003911343,
-      "grad_norm": 1.586787223815918,
-      "learning_rate": 7.620804020100503e-05,
-      "loss": 5.3108,
-      "step": 24181
-    },
-    {
-      "epoch": 12.611212516297263,
-      "grad_norm": 1.4401277303695679,
-      "learning_rate": 7.62070351758794e-05,
-      "loss": 5.3007,
-      "step": 24182
-    },
-    {
-      "epoch": 12.61173402868318,
-      "grad_norm": 1.432283639907837,
-      "learning_rate": 7.620603015075377e-05,
-      "loss": 5.3844,
-      "step": 24183
-    },
-    {
-      "epoch": 12.6122555410691,
-      "grad_norm": 1.4901829957962036,
-      "learning_rate": 7.620502512562815e-05,
-      "loss": 5.3679,
-      "step": 24184
-    },
-    {
-      "epoch": 12.61277705345502,
-      "grad_norm": 1.4038480520248413,
-      "learning_rate": 7.620402010050251e-05,
-      "loss": 5.6167,
-      "step": 24185
-    },
-    {
-      "epoch": 12.613298565840939,
-      "grad_norm": 1.417486548423767,
-      "learning_rate": 7.620301507537689e-05,
-      "loss": 5.289,
-      "step": 24186
-    },
-    {
-      "epoch": 12.613820078226858,
-      "grad_norm": 1.3204275369644165,
-      "learning_rate": 7.620201005025125e-05,
-      "loss": 5.5602,
-      "step": 24187
-    },
-    {
-      "epoch": 12.614341590612778,
-      "grad_norm": 1.4863862991333008,
-      "learning_rate": 7.620100502512563e-05,
-      "loss": 5.427,
-      "step": 24188
-    },
-    {
-      "epoch": 12.614863102998696,
-      "grad_norm": 1.3843861818313599,
-      "learning_rate": 7.620000000000001e-05,
-      "loss": 5.6665,
-      "step": 24189
-    },
-    {
-      "epoch": 12.615384615384615,
-      "grad_norm": 1.4128495454788208,
-      "learning_rate": 7.619899497487437e-05,
-      "loss": 5.6004,
-      "step": 24190
-    },
-    {
-      "epoch": 12.615906127770534,
-      "grad_norm": 1.4508839845657349,
-      "learning_rate": 7.619798994974875e-05,
-      "loss": 5.5374,
-      "step": 24191
-    },
-    {
-      "epoch": 12.616427640156454,
-      "grad_norm": 1.3601515293121338,
-      "learning_rate": 7.619698492462312e-05,
-      "loss": 5.6222,
-      "step": 24192
-    },
-    {
-      "epoch": 12.616949152542373,
-      "grad_norm": 1.5514222383499146,
-      "learning_rate": 7.61959798994975e-05,
-      "loss": 5.2341,
-      "step": 24193
-    },
-    {
-      "epoch": 12.617470664928293,
-      "grad_norm": 1.3885352611541748,
-      "learning_rate": 7.619497487437186e-05,
-      "loss": 5.6994,
-      "step": 24194
-    },
-    {
-      "epoch": 12.61799217731421,
-      "grad_norm": 1.531398892402649,
-      "learning_rate": 7.619396984924624e-05,
-      "loss": 4.3476,
-      "step": 24195
-    },
-    {
-      "epoch": 12.61851368970013,
-      "grad_norm": 1.4559667110443115,
-      "learning_rate": 7.61929648241206e-05,
-      "loss": 5.5837,
-      "step": 24196
-    },
-    {
-      "epoch": 12.61903520208605,
-      "grad_norm": 1.4265440702438354,
-      "learning_rate": 7.619195979899498e-05,
-      "loss": 5.2975,
-      "step": 24197
-    },
-    {
-      "epoch": 12.619556714471969,
-      "grad_norm": 1.3890931606292725,
-      "learning_rate": 7.619095477386936e-05,
-      "loss": 5.7354,
-      "step": 24198
-    },
-    {
-      "epoch": 12.620078226857888,
-      "grad_norm": 1.4274128675460815,
-      "learning_rate": 7.618994974874373e-05,
-      "loss": 5.22,
-      "step": 24199
-    },
-    {
-      "epoch": 12.620599739243808,
-      "grad_norm": 1.4132174253463745,
-      "learning_rate": 7.61889447236181e-05,
-      "loss": 5.011,
-      "step": 24200
-    },
-    {
-      "epoch": 12.621121251629726,
-      "grad_norm": 1.4926615953445435,
-      "learning_rate": 7.618793969849246e-05,
-      "loss": 5.4963,
-      "step": 24201
-    },
-    {
-      "epoch": 12.621642764015645,
-      "grad_norm": 1.6381510496139526,
-      "learning_rate": 7.618693467336684e-05,
-      "loss": 5.3921,
-      "step": 24202
-    },
-    {
-      "epoch": 12.622164276401564,
-      "grad_norm": 1.4361557960510254,
-      "learning_rate": 7.61859296482412e-05,
-      "loss": 5.6453,
-      "step": 24203
-    },
-    {
-      "epoch": 12.622685788787484,
-      "grad_norm": 1.5244101285934448,
-      "learning_rate": 7.618492462311558e-05,
-      "loss": 5.7121,
-      "step": 24204
-    },
-    {
-      "epoch": 12.623207301173403,
-      "grad_norm": 1.4404422044754028,
-      "learning_rate": 7.618391959798995e-05,
-      "loss": 5.801,
-      "step": 24205
-    },
-    {
-      "epoch": 12.623728813559323,
-      "grad_norm": 1.3525457382202148,
-      "learning_rate": 7.618291457286432e-05,
-      "loss": 5.5174,
-      "step": 24206
-    },
-    {
-      "epoch": 12.62425032594524,
-      "grad_norm": 1.478474497795105,
-      "learning_rate": 7.618190954773869e-05,
-      "loss": 5.5023,
-      "step": 24207
-    },
-    {
-      "epoch": 12.62477183833116,
-      "grad_norm": 1.5552101135253906,
-      "learning_rate": 7.618090452261307e-05,
-      "loss": 5.1016,
-      "step": 24208
-    },
-    {
-      "epoch": 12.62529335071708,
-      "grad_norm": 1.5484611988067627,
-      "learning_rate": 7.617989949748744e-05,
-      "loss": 4.7616,
-      "step": 24209
-    },
-    {
-      "epoch": 12.625814863102999,
-      "grad_norm": 1.5170210599899292,
-      "learning_rate": 7.617889447236182e-05,
-      "loss": 5.2322,
-      "step": 24210
-    },
-    {
-      "epoch": 12.626336375488918,
-      "grad_norm": 1.5304886102676392,
-      "learning_rate": 7.617788944723619e-05,
-      "loss": 5.1984,
-      "step": 24211
-    },
-    {
-      "epoch": 12.626857887874838,
-      "grad_norm": 1.483656406402588,
-      "learning_rate": 7.617688442211056e-05,
-      "loss": 5.0748,
-      "step": 24212
-    },
-    {
-      "epoch": 12.627379400260756,
-      "grad_norm": 1.4101771116256714,
-      "learning_rate": 7.617587939698493e-05,
-      "loss": 5.7327,
-      "step": 24213
-    },
-    {
-      "epoch": 12.627900912646675,
-      "grad_norm": 1.4998592138290405,
-      "learning_rate": 7.617487437185929e-05,
-      "loss": 5.3793,
-      "step": 24214
-    },
-    {
-      "epoch": 12.628422425032594,
-      "grad_norm": 1.4013618230819702,
-      "learning_rate": 7.617386934673367e-05,
-      "loss": 5.4732,
-      "step": 24215
-    },
-    {
-      "epoch": 12.628943937418514,
-      "grad_norm": 1.4045625925064087,
-      "learning_rate": 7.617286432160803e-05,
-      "loss": 5.621,
-      "step": 24216
-    },
-    {
-      "epoch": 12.629465449804433,
-      "grad_norm": 1.4038751125335693,
-      "learning_rate": 7.617185929648241e-05,
-      "loss": 5.324,
-      "step": 24217
-    },
-    {
-      "epoch": 12.629986962190351,
-      "grad_norm": 1.4875398874282837,
-      "learning_rate": 7.617085427135679e-05,
-      "loss": 4.9792,
-      "step": 24218
-    },
-    {
-      "epoch": 12.63050847457627,
-      "grad_norm": 1.4283933639526367,
-      "learning_rate": 7.616984924623117e-05,
-      "loss": 5.4929,
-      "step": 24219
-    },
-    {
-      "epoch": 12.63102998696219,
-      "grad_norm": 1.3719485998153687,
-      "learning_rate": 7.616884422110553e-05,
-      "loss": 5.4388,
-      "step": 24220
-    },
-    {
-      "epoch": 12.63155149934811,
-      "grad_norm": 1.5548839569091797,
-      "learning_rate": 7.616783919597991e-05,
-      "loss": 5.3351,
-      "step": 24221
-    },
-    {
-      "epoch": 12.632073011734029,
-      "grad_norm": 1.5367653369903564,
-      "learning_rate": 7.616683417085427e-05,
-      "loss": 5.122,
-      "step": 24222
-    },
-    {
-      "epoch": 12.632594524119948,
-      "grad_norm": 1.6774739027023315,
-      "learning_rate": 7.616582914572865e-05,
-      "loss": 5.4308,
-      "step": 24223
-    },
-    {
-      "epoch": 12.633116036505868,
-      "grad_norm": 1.422676682472229,
-      "learning_rate": 7.616482412060302e-05,
-      "loss": 5.3742,
-      "step": 24224
-    },
-    {
-      "epoch": 12.633637548891786,
-      "grad_norm": 1.4175447225570679,
-      "learning_rate": 7.61638190954774e-05,
-      "loss": 5.5007,
-      "step": 24225
-    },
-    {
-      "epoch": 12.634159061277705,
-      "grad_norm": 1.3326914310455322,
-      "learning_rate": 7.616281407035176e-05,
-      "loss": 5.4993,
-      "step": 24226
-    },
-    {
-      "epoch": 12.634680573663625,
-      "grad_norm": 1.5058164596557617,
-      "learning_rate": 7.616180904522612e-05,
-      "loss": 5.1884,
-      "step": 24227
-    },
-    {
-      "epoch": 12.635202086049544,
-      "grad_norm": 1.7090483903884888,
-      "learning_rate": 7.61608040201005e-05,
-      "loss": 4.4624,
-      "step": 24228
-    },
-    {
-      "epoch": 12.635723598435463,
-      "grad_norm": 1.3659987449645996,
-      "learning_rate": 7.615979899497488e-05,
-      "loss": 5.7738,
-      "step": 24229
-    },
-    {
-      "epoch": 12.636245110821381,
-      "grad_norm": 1.4246097803115845,
-      "learning_rate": 7.615879396984926e-05,
-      "loss": 5.5055,
-      "step": 24230
-    },
-    {
-      "epoch": 12.6367666232073,
-      "grad_norm": 1.437363624572754,
-      "learning_rate": 7.615778894472362e-05,
-      "loss": 5.1862,
-      "step": 24231
-    },
-    {
-      "epoch": 12.63728813559322,
-      "grad_norm": 1.4266610145568848,
-      "learning_rate": 7.6156783919598e-05,
-      "loss": 5.3939,
-      "step": 24232
-    },
-    {
-      "epoch": 12.63780964797914,
-      "grad_norm": 1.3684862852096558,
-      "learning_rate": 7.615577889447236e-05,
-      "loss": 5.3784,
-      "step": 24233
-    },
-    {
-      "epoch": 12.638331160365059,
-      "grad_norm": 1.4450666904449463,
-      "learning_rate": 7.615477386934674e-05,
-      "loss": 5.3802,
-      "step": 24234
-    },
-    {
-      "epoch": 12.638852672750978,
-      "grad_norm": 1.430763602256775,
-      "learning_rate": 7.61537688442211e-05,
-      "loss": 5.3984,
-      "step": 24235
-    },
-    {
-      "epoch": 12.639374185136898,
-      "grad_norm": 1.4839718341827393,
-      "learning_rate": 7.615276381909548e-05,
-      "loss": 5.4469,
-      "step": 24236
-    },
-    {
-      "epoch": 12.639895697522816,
-      "grad_norm": 1.4265727996826172,
-      "learning_rate": 7.615175879396985e-05,
-      "loss": 5.4513,
-      "step": 24237
-    },
-    {
-      "epoch": 12.640417209908735,
-      "grad_norm": 1.4096124172210693,
-      "learning_rate": 7.615075376884422e-05,
-      "loss": 5.3273,
-      "step": 24238
-    },
-    {
-      "epoch": 12.640938722294655,
-      "grad_norm": 1.3463743925094604,
-      "learning_rate": 7.61497487437186e-05,
-      "loss": 5.6453,
-      "step": 24239
-    },
-    {
-      "epoch": 12.641460234680574,
-      "grad_norm": 1.399750828742981,
-      "learning_rate": 7.614874371859297e-05,
-      "loss": 5.2865,
-      "step": 24240
-    },
-    {
-      "epoch": 12.641981747066493,
-      "grad_norm": 1.2925084829330444,
-      "learning_rate": 7.614773869346734e-05,
-      "loss": 4.9909,
-      "step": 24241
-    },
-    {
-      "epoch": 12.642503259452411,
-      "grad_norm": 1.5188477039337158,
-      "learning_rate": 7.614673366834171e-05,
-      "loss": 5.5364,
-      "step": 24242
-    },
-    {
-      "epoch": 12.64302477183833,
-      "grad_norm": 1.3568624258041382,
-      "learning_rate": 7.614572864321609e-05,
-      "loss": 5.6568,
-      "step": 24243
-    },
-    {
-      "epoch": 12.64354628422425,
-      "grad_norm": 1.3039069175720215,
-      "learning_rate": 7.614472361809045e-05,
-      "loss": 5.6859,
-      "step": 24244
-    },
-    {
-      "epoch": 12.64406779661017,
-      "grad_norm": 1.5273234844207764,
-      "learning_rate": 7.614371859296483e-05,
-      "loss": 5.3488,
-      "step": 24245
-    },
-    {
-      "epoch": 12.644589308996089,
-      "grad_norm": 1.4715096950531006,
-      "learning_rate": 7.614271356783919e-05,
-      "loss": 5.1486,
-      "step": 24246
-    },
-    {
-      "epoch": 12.645110821382008,
-      "grad_norm": 1.4775631427764893,
-      "learning_rate": 7.614170854271357e-05,
-      "loss": 5.4399,
-      "step": 24247
-    },
-    {
-      "epoch": 12.645632333767926,
-      "grad_norm": 1.5695005655288696,
-      "learning_rate": 7.614070351758793e-05,
-      "loss": 5.3482,
-      "step": 24248
-    },
-    {
-      "epoch": 12.646153846153846,
-      "grad_norm": 1.4540103673934937,
-      "learning_rate": 7.613969849246231e-05,
-      "loss": 5.7559,
-      "step": 24249
-    },
-    {
-      "epoch": 12.646675358539765,
-      "grad_norm": 1.5203204154968262,
-      "learning_rate": 7.613869346733669e-05,
-      "loss": 5.769,
-      "step": 24250
-    },
-    {
-      "epoch": 12.647196870925685,
-      "grad_norm": 1.4782140254974365,
-      "learning_rate": 7.613768844221107e-05,
-      "loss": 5.5672,
-      "step": 24251
-    },
-    {
-      "epoch": 12.647718383311604,
-      "grad_norm": 1.4632525444030762,
-      "learning_rate": 7.613668341708543e-05,
-      "loss": 5.4666,
-      "step": 24252
-    },
-    {
-      "epoch": 12.648239895697523,
-      "grad_norm": 1.3034495115280151,
-      "learning_rate": 7.61356783919598e-05,
-      "loss": 5.9652,
-      "step": 24253
-    },
-    {
-      "epoch": 12.648761408083441,
-      "grad_norm": 1.5588845014572144,
-      "learning_rate": 7.613467336683417e-05,
-      "loss": 5.5238,
-      "step": 24254
-    },
-    {
-      "epoch": 12.64928292046936,
-      "grad_norm": 1.4894484281539917,
-      "learning_rate": 7.613366834170854e-05,
-      "loss": 5.2221,
-      "step": 24255
-    },
-    {
-      "epoch": 12.64980443285528,
-      "grad_norm": 1.368666172027588,
-      "learning_rate": 7.613266331658292e-05,
-      "loss": 5.8006,
-      "step": 24256
-    },
-    {
-      "epoch": 12.6503259452412,
-      "grad_norm": 1.449100136756897,
-      "learning_rate": 7.613165829145728e-05,
-      "loss": 5.4617,
-      "step": 24257
-    },
-    {
-      "epoch": 12.650847457627119,
-      "grad_norm": 1.4456567764282227,
-      "learning_rate": 7.613065326633166e-05,
-      "loss": 5.7216,
-      "step": 24258
-    },
-    {
-      "epoch": 12.651368970013039,
-      "grad_norm": 1.4415283203125,
-      "learning_rate": 7.612964824120604e-05,
-      "loss": 5.6801,
-      "step": 24259
-    },
-    {
-      "epoch": 12.651890482398956,
-      "grad_norm": 1.4767874479293823,
-      "learning_rate": 7.612864321608041e-05,
-      "loss": 5.0167,
-      "step": 24260
-    },
-    {
-      "epoch": 12.652411994784876,
-      "grad_norm": 1.4466893672943115,
-      "learning_rate": 7.612763819095478e-05,
-      "loss": 4.8332,
-      "step": 24261
-    },
-    {
-      "epoch": 12.652933507170795,
-      "grad_norm": 1.4080402851104736,
-      "learning_rate": 7.612663316582916e-05,
-      "loss": 5.5815,
-      "step": 24262
-    },
-    {
-      "epoch": 12.653455019556715,
-      "grad_norm": 1.4494566917419434,
-      "learning_rate": 7.612562814070352e-05,
-      "loss": 5.0936,
-      "step": 24263
-    },
-    {
-      "epoch": 12.653976531942634,
-      "grad_norm": 1.4056930541992188,
-      "learning_rate": 7.61246231155779e-05,
-      "loss": 5.5623,
-      "step": 24264
-    },
-    {
-      "epoch": 12.654498044328554,
-      "grad_norm": 1.362209439277649,
-      "learning_rate": 7.612361809045226e-05,
-      "loss": 5.0709,
-      "step": 24265
-    },
-    {
-      "epoch": 12.655019556714471,
-      "grad_norm": 1.5332584381103516,
-      "learning_rate": 7.612261306532664e-05,
-      "loss": 5.5733,
-      "step": 24266
-    },
-    {
-      "epoch": 12.65554106910039,
-      "grad_norm": 1.4077486991882324,
-      "learning_rate": 7.6121608040201e-05,
-      "loss": 5.2072,
-      "step": 24267
-    },
-    {
-      "epoch": 12.65606258148631,
-      "grad_norm": 1.5075082778930664,
-      "learning_rate": 7.612060301507537e-05,
-      "loss": 5.2357,
-      "step": 24268
-    },
-    {
-      "epoch": 12.65658409387223,
-      "grad_norm": 1.445037841796875,
-      "learning_rate": 7.611959798994975e-05,
-      "loss": 5.385,
-      "step": 24269
-    },
-    {
-      "epoch": 12.657105606258149,
-      "grad_norm": 1.3983641862869263,
-      "learning_rate": 7.611859296482413e-05,
-      "loss": 5.1941,
-      "step": 24270
-    },
-    {
-      "epoch": 12.657627118644069,
-      "grad_norm": 1.421000361442566,
-      "learning_rate": 7.61175879396985e-05,
-      "loss": 5.3457,
-      "step": 24271
-    },
-    {
-      "epoch": 12.658148631029986,
-      "grad_norm": 1.534568190574646,
-      "learning_rate": 7.611658291457287e-05,
-      "loss": 5.2769,
-      "step": 24272
-    },
-    {
-      "epoch": 12.658670143415906,
-      "grad_norm": 1.410872220993042,
-      "learning_rate": 7.611557788944725e-05,
-      "loss": 5.0188,
-      "step": 24273
-    },
-    {
-      "epoch": 12.659191655801825,
-      "grad_norm": 1.5291202068328857,
-      "learning_rate": 7.611457286432161e-05,
-      "loss": 5.5128,
-      "step": 24274
-    },
-    {
-      "epoch": 12.659713168187745,
-      "grad_norm": 1.5109515190124512,
-      "learning_rate": 7.611356783919599e-05,
-      "loss": 5.331,
-      "step": 24275
-    },
-    {
-      "epoch": 12.660234680573664,
-      "grad_norm": 1.4862571954727173,
-      "learning_rate": 7.611256281407035e-05,
-      "loss": 5.2865,
-      "step": 24276
-    },
-    {
-      "epoch": 12.660756192959584,
-      "grad_norm": 1.642731785774231,
-      "learning_rate": 7.611155778894473e-05,
-      "loss": 5.5275,
-      "step": 24277
-    },
-    {
-      "epoch": 12.661277705345501,
-      "grad_norm": 1.6462132930755615,
-      "learning_rate": 7.61105527638191e-05,
-      "loss": 4.8973,
-      "step": 24278
-    },
-    {
-      "epoch": 12.66179921773142,
-      "grad_norm": 1.3594563007354736,
-      "learning_rate": 7.610954773869347e-05,
-      "loss": 5.7134,
-      "step": 24279
-    },
-    {
-      "epoch": 12.66232073011734,
-      "grad_norm": 1.4011937379837036,
-      "learning_rate": 7.610854271356785e-05,
-      "loss": 5.6743,
-      "step": 24280
-    },
-    {
-      "epoch": 12.66284224250326,
-      "grad_norm": 1.3196862936019897,
-      "learning_rate": 7.610753768844221e-05,
-      "loss": 5.9021,
-      "step": 24281
-    },
-    {
-      "epoch": 12.663363754889179,
-      "grad_norm": 1.422443151473999,
-      "learning_rate": 7.610653266331659e-05,
-      "loss": 5.2851,
-      "step": 24282
-    },
-    {
-      "epoch": 12.663885267275099,
-      "grad_norm": 1.485835075378418,
-      "learning_rate": 7.610552763819096e-05,
-      "loss": 5.1623,
-      "step": 24283
-    },
-    {
-      "epoch": 12.664406779661016,
-      "grad_norm": 1.583411455154419,
-      "learning_rate": 7.610452261306533e-05,
-      "loss": 5.2727,
-      "step": 24284
-    },
-    {
-      "epoch": 12.664928292046936,
-      "grad_norm": 1.4471313953399658,
-      "learning_rate": 7.61035175879397e-05,
-      "loss": 5.3968,
-      "step": 24285
-    },
-    {
-      "epoch": 12.665449804432855,
-      "grad_norm": 1.526421070098877,
-      "learning_rate": 7.610251256281408e-05,
-      "loss": 5.4136,
-      "step": 24286
-    },
-    {
-      "epoch": 12.665971316818775,
-      "grad_norm": 1.6667462587356567,
-      "learning_rate": 7.610150753768844e-05,
-      "loss": 5.5285,
-      "step": 24287
-    },
-    {
-      "epoch": 12.666492829204694,
-      "grad_norm": 1.3714112043380737,
-      "learning_rate": 7.610050251256282e-05,
-      "loss": 5.5651,
-      "step": 24288
-    },
-    {
-      "epoch": 12.667014341590614,
-      "grad_norm": 1.4060344696044922,
-      "learning_rate": 7.609949748743718e-05,
-      "loss": 5.3768,
-      "step": 24289
-    },
-    {
-      "epoch": 12.667535853976531,
-      "grad_norm": 1.5278794765472412,
-      "learning_rate": 7.609849246231156e-05,
-      "loss": 5.612,
-      "step": 24290
-    },
-    {
-      "epoch": 12.66805736636245,
-      "grad_norm": 1.4664849042892456,
-      "learning_rate": 7.609748743718594e-05,
-      "loss": 5.4703,
-      "step": 24291
-    },
-    {
-      "epoch": 12.66857887874837,
-      "grad_norm": 1.540303349494934,
-      "learning_rate": 7.609648241206032e-05,
-      "loss": 5.2535,
-      "step": 24292
-    },
-    {
-      "epoch": 12.66910039113429,
-      "grad_norm": 1.4502711296081543,
-      "learning_rate": 7.609547738693468e-05,
-      "loss": 5.5521,
-      "step": 24293
-    },
-    {
-      "epoch": 12.66962190352021,
-      "grad_norm": 1.422184944152832,
-      "learning_rate": 7.609447236180904e-05,
-      "loss": 5.4656,
-      "step": 24294
-    },
-    {
-      "epoch": 12.670143415906129,
-      "grad_norm": 1.4040900468826294,
-      "learning_rate": 7.609346733668342e-05,
-      "loss": 5.6685,
-      "step": 24295
-    },
-    {
-      "epoch": 12.670664928292046,
-      "grad_norm": 1.3601429462432861,
-      "learning_rate": 7.609246231155779e-05,
-      "loss": 5.7537,
-      "step": 24296
-    },
-    {
-      "epoch": 12.671186440677966,
-      "grad_norm": 1.3180136680603027,
-      "learning_rate": 7.609145728643216e-05,
-      "loss": 5.6173,
-      "step": 24297
-    },
-    {
-      "epoch": 12.671707953063885,
-      "grad_norm": 1.4264917373657227,
-      "learning_rate": 7.609045226130653e-05,
-      "loss": 5.591,
-      "step": 24298
-    },
-    {
-      "epoch": 12.672229465449805,
-      "grad_norm": 1.341850996017456,
-      "learning_rate": 7.60894472361809e-05,
-      "loss": 5.6738,
-      "step": 24299
-    },
-    {
-      "epoch": 12.672750977835724,
-      "grad_norm": 1.4943071603775024,
-      "learning_rate": 7.608844221105528e-05,
-      "loss": 5.3389,
-      "step": 24300
-    },
-    {
-      "epoch": 12.673272490221644,
-      "grad_norm": 1.4800583124160767,
-      "learning_rate": 7.608743718592966e-05,
-      "loss": 5.2033,
-      "step": 24301
-    },
-    {
-      "epoch": 12.673794002607561,
-      "grad_norm": 1.4654072523117065,
-      "learning_rate": 7.608643216080403e-05,
-      "loss": 5.293,
-      "step": 24302
-    },
-    {
-      "epoch": 12.67431551499348,
-      "grad_norm": 1.5404644012451172,
-      "learning_rate": 7.60854271356784e-05,
-      "loss": 5.7291,
-      "step": 24303
-    },
-    {
-      "epoch": 12.6748370273794,
-      "grad_norm": 1.4334239959716797,
-      "learning_rate": 7.608442211055277e-05,
-      "loss": 5.7201,
-      "step": 24304
-    },
-    {
-      "epoch": 12.67535853976532,
-      "grad_norm": 1.5415775775909424,
-      "learning_rate": 7.608341708542715e-05,
-      "loss": 5.176,
-      "step": 24305
-    },
-    {
-      "epoch": 12.67588005215124,
-      "grad_norm": 1.392251968383789,
-      "learning_rate": 7.608241206030151e-05,
-      "loss": 5.2615,
-      "step": 24306
-    },
-    {
-      "epoch": 12.676401564537159,
-      "grad_norm": 1.441267490386963,
-      "learning_rate": 7.608140703517587e-05,
-      "loss": 5.5743,
-      "step": 24307
-    },
-    {
-      "epoch": 12.676923076923076,
-      "grad_norm": 1.4565941095352173,
-      "learning_rate": 7.608040201005025e-05,
-      "loss": 5.8273,
-      "step": 24308
-    },
-    {
-      "epoch": 12.677444589308996,
-      "grad_norm": 1.5276778936386108,
-      "learning_rate": 7.607939698492462e-05,
-      "loss": 5.0985,
-      "step": 24309
-    },
-    {
-      "epoch": 12.677966101694915,
-      "grad_norm": 1.5106818675994873,
-      "learning_rate": 7.6078391959799e-05,
-      "loss": 5.6026,
-      "step": 24310
-    },
-    {
-      "epoch": 12.678487614080835,
-      "grad_norm": 1.4590986967086792,
-      "learning_rate": 7.607738693467337e-05,
-      "loss": 5.7015,
-      "step": 24311
-    },
-    {
-      "epoch": 12.679009126466754,
-      "grad_norm": 1.4629194736480713,
-      "learning_rate": 7.607638190954775e-05,
-      "loss": 5.5036,
-      "step": 24312
-    },
-    {
-      "epoch": 12.679530638852672,
-      "grad_norm": 1.3608206510543823,
-      "learning_rate": 7.607537688442211e-05,
-      "loss": 5.9413,
-      "step": 24313
-    },
-    {
-      "epoch": 12.680052151238591,
-      "grad_norm": 1.6010980606079102,
-      "learning_rate": 7.607437185929649e-05,
-      "loss": 5.2212,
-      "step": 24314
-    },
-    {
-      "epoch": 12.68057366362451,
-      "grad_norm": 1.4699190855026245,
-      "learning_rate": 7.607336683417086e-05,
-      "loss": 5.6009,
-      "step": 24315
-    },
-    {
-      "epoch": 12.68109517601043,
-      "grad_norm": 1.3533704280853271,
-      "learning_rate": 7.607236180904523e-05,
-      "loss": 5.8968,
-      "step": 24316
-    },
-    {
-      "epoch": 12.68161668839635,
-      "grad_norm": 1.510364055633545,
-      "learning_rate": 7.60713567839196e-05,
-      "loss": 5.5867,
-      "step": 24317
-    },
-    {
-      "epoch": 12.68213820078227,
-      "grad_norm": 1.5813214778900146,
-      "learning_rate": 7.607035175879398e-05,
-      "loss": 5.2547,
-      "step": 24318
-    },
-    {
-      "epoch": 12.682659713168189,
-      "grad_norm": 1.4414528608322144,
-      "learning_rate": 7.606934673366834e-05,
-      "loss": 5.7396,
-      "step": 24319
-    },
-    {
-      "epoch": 12.683181225554106,
-      "grad_norm": 1.5756957530975342,
-      "learning_rate": 7.606834170854272e-05,
-      "loss": 5.4813,
-      "step": 24320
-    },
-    {
-      "epoch": 12.683702737940026,
-      "grad_norm": 1.5599771738052368,
-      "learning_rate": 7.60673366834171e-05,
-      "loss": 5.4947,
-      "step": 24321
-    },
-    {
-      "epoch": 12.684224250325945,
-      "grad_norm": 1.5776969194412231,
-      "learning_rate": 7.606633165829146e-05,
-      "loss": 5.1298,
-      "step": 24322
-    },
-    {
-      "epoch": 12.684745762711865,
-      "grad_norm": 1.6920337677001953,
-      "learning_rate": 7.606532663316584e-05,
-      "loss": 5.1769,
-      "step": 24323
-    },
-    {
-      "epoch": 12.685267275097784,
-      "grad_norm": 1.4756355285644531,
-      "learning_rate": 7.60643216080402e-05,
-      "loss": 5.7018,
-      "step": 24324
-    },
-    {
-      "epoch": 12.685788787483702,
-      "grad_norm": 1.5206096172332764,
-      "learning_rate": 7.606331658291458e-05,
-      "loss": 5.205,
-      "step": 24325
-    },
-    {
-      "epoch": 12.686310299869621,
-      "grad_norm": 1.4926693439483643,
-      "learning_rate": 7.606231155778894e-05,
-      "loss": 5.3184,
-      "step": 24326
-    },
-    {
-      "epoch": 12.68683181225554,
-      "grad_norm": 1.3996894359588623,
-      "learning_rate": 7.606130653266332e-05,
-      "loss": 5.2571,
-      "step": 24327
-    },
-    {
-      "epoch": 12.68735332464146,
-      "grad_norm": 1.3785954713821411,
-      "learning_rate": 7.606030150753769e-05,
-      "loss": 5.5346,
-      "step": 24328
-    },
-    {
-      "epoch": 12.68787483702738,
-      "grad_norm": 1.4803731441497803,
-      "learning_rate": 7.605929648241206e-05,
-      "loss": 5.479,
-      "step": 24329
-    },
-    {
-      "epoch": 12.6883963494133,
-      "grad_norm": 1.4773006439208984,
-      "learning_rate": 7.605829145728643e-05,
-      "loss": 5.1236,
-      "step": 24330
-    },
-    {
-      "epoch": 12.688917861799219,
-      "grad_norm": 1.5420628786087036,
-      "learning_rate": 7.605728643216081e-05,
-      "loss": 5.3164,
-      "step": 24331
-    },
-    {
-      "epoch": 12.689439374185136,
-      "grad_norm": 1.3643741607666016,
-      "learning_rate": 7.605628140703518e-05,
-      "loss": 5.732,
-      "step": 24332
-    },
-    {
-      "epoch": 12.689960886571056,
-      "grad_norm": 1.3995733261108398,
-      "learning_rate": 7.605527638190955e-05,
-      "loss": 5.3489,
-      "step": 24333
-    },
-    {
-      "epoch": 12.690482398956975,
-      "grad_norm": 1.5113794803619385,
-      "learning_rate": 7.605427135678393e-05,
-      "loss": 5.0839,
-      "step": 24334
-    },
-    {
-      "epoch": 12.691003911342895,
-      "grad_norm": 1.4621622562408447,
-      "learning_rate": 7.605326633165829e-05,
-      "loss": 5.4285,
-      "step": 24335
-    },
-    {
-      "epoch": 12.691525423728814,
-      "grad_norm": 1.5336389541625977,
-      "learning_rate": 7.605226130653267e-05,
-      "loss": 5.5405,
-      "step": 24336
-    },
-    {
-      "epoch": 12.692046936114732,
-      "grad_norm": 1.4523630142211914,
-      "learning_rate": 7.605125628140703e-05,
-      "loss": 5.4894,
-      "step": 24337
-    },
-    {
-      "epoch": 12.692568448500651,
-      "grad_norm": 1.4674464464187622,
-      "learning_rate": 7.605025125628141e-05,
-      "loss": 5.1259,
-      "step": 24338
-    },
-    {
-      "epoch": 12.69308996088657,
-      "grad_norm": 1.5984127521514893,
-      "learning_rate": 7.604924623115578e-05,
-      "loss": 5.0148,
-      "step": 24339
-    },
-    {
-      "epoch": 12.69361147327249,
-      "grad_norm": 1.4757450819015503,
-      "learning_rate": 7.604824120603015e-05,
-      "loss": 5.5154,
-      "step": 24340
-    },
-    {
-      "epoch": 12.69413298565841,
-      "grad_norm": 1.5860075950622559,
-      "learning_rate": 7.604723618090453e-05,
-      "loss": 5.6452,
-      "step": 24341
-    },
-    {
-      "epoch": 12.69465449804433,
-      "grad_norm": 1.3731824159622192,
-      "learning_rate": 7.604623115577891e-05,
-      "loss": 5.7384,
-      "step": 24342
-    },
-    {
-      "epoch": 12.695176010430247,
-      "grad_norm": 1.5322567224502563,
-      "learning_rate": 7.604522613065327e-05,
-      "loss": 5.1629,
-      "step": 24343
-    },
-    {
-      "epoch": 12.695697522816166,
-      "grad_norm": 1.3123407363891602,
-      "learning_rate": 7.604422110552765e-05,
-      "loss": 4.9578,
-      "step": 24344
-    },
-    {
-      "epoch": 12.696219035202086,
-      "grad_norm": 1.508303165435791,
-      "learning_rate": 7.604321608040202e-05,
-      "loss": 5.8433,
-      "step": 24345
-    },
-    {
-      "epoch": 12.696740547588005,
-      "grad_norm": 1.4722509384155273,
-      "learning_rate": 7.604221105527638e-05,
-      "loss": 5.5069,
-      "step": 24346
-    },
-    {
-      "epoch": 12.697262059973925,
-      "grad_norm": 1.5153048038482666,
-      "learning_rate": 7.604120603015076e-05,
-      "loss": 5.4865,
-      "step": 24347
-    },
-    {
-      "epoch": 12.697783572359844,
-      "grad_norm": 1.566575050354004,
-      "learning_rate": 7.604020100502512e-05,
-      "loss": 4.9991,
-      "step": 24348
-    },
-    {
-      "epoch": 12.698305084745762,
-      "grad_norm": 1.408167839050293,
-      "learning_rate": 7.60391959798995e-05,
-      "loss": 5.5933,
-      "step": 24349
-    },
-    {
-      "epoch": 12.698826597131681,
-      "grad_norm": 1.6111527681350708,
-      "learning_rate": 7.603819095477386e-05,
-      "loss": 4.9464,
-      "step": 24350
-    },
-    {
-      "epoch": 12.6993481095176,
-      "grad_norm": 1.5153045654296875,
-      "learning_rate": 7.603718592964824e-05,
-      "loss": 5.143,
-      "step": 24351
-    },
-    {
-      "epoch": 12.69986962190352,
-      "grad_norm": 1.478361964225769,
-      "learning_rate": 7.603618090452262e-05,
-      "loss": 5.2975,
-      "step": 24352
-    },
-    {
-      "epoch": 12.70039113428944,
-      "grad_norm": 1.3774333000183105,
-      "learning_rate": 7.6035175879397e-05,
-      "loss": 5.7246,
-      "step": 24353
-    },
-    {
-      "epoch": 12.70091264667536,
-      "grad_norm": 1.4211243391036987,
-      "learning_rate": 7.603417085427136e-05,
-      "loss": 5.5883,
-      "step": 24354
-    },
-    {
-      "epoch": 12.701434159061277,
-      "grad_norm": 1.6116986274719238,
-      "learning_rate": 7.603316582914574e-05,
-      "loss": 5.1494,
-      "step": 24355
-    },
-    {
-      "epoch": 12.701955671447196,
-      "grad_norm": 1.473743200302124,
-      "learning_rate": 7.60321608040201e-05,
-      "loss": 5.2769,
-      "step": 24356
-    },
-    {
-      "epoch": 12.702477183833116,
-      "grad_norm": 1.3275612592697144,
-      "learning_rate": 7.603115577889448e-05,
-      "loss": 5.9453,
-      "step": 24357
-    },
-    {
-      "epoch": 12.702998696219035,
-      "grad_norm": 1.3406250476837158,
-      "learning_rate": 7.603015075376885e-05,
-      "loss": 5.6537,
-      "step": 24358
-    },
-    {
-      "epoch": 12.703520208604955,
-      "grad_norm": 1.5148389339447021,
-      "learning_rate": 7.602914572864322e-05,
-      "loss": 5.3399,
-      "step": 24359
-    },
-    {
-      "epoch": 12.704041720990874,
-      "grad_norm": 1.5284837484359741,
-      "learning_rate": 7.602814070351759e-05,
-      "loss": 5.4602,
-      "step": 24360
-    },
-    {
-      "epoch": 12.704563233376792,
-      "grad_norm": 1.5396493673324585,
-      "learning_rate": 7.602713567839195e-05,
-      "loss": 5.1218,
-      "step": 24361
-    },
-    {
-      "epoch": 12.705084745762711,
-      "grad_norm": 1.4979456663131714,
-      "learning_rate": 7.602613065326633e-05,
-      "loss": 5.3511,
-      "step": 24362
-    },
-    {
-      "epoch": 12.70560625814863,
-      "grad_norm": 1.4480172395706177,
-      "learning_rate": 7.602512562814071e-05,
-      "loss": 5.4436,
-      "step": 24363
-    },
-    {
-      "epoch": 12.70612777053455,
-      "grad_norm": 1.4359452724456787,
-      "learning_rate": 7.602412060301509e-05,
-      "loss": 5.2013,
-      "step": 24364
-    },
-    {
-      "epoch": 12.70664928292047,
-      "grad_norm": 1.4739799499511719,
-      "learning_rate": 7.602311557788945e-05,
-      "loss": 5.2594,
-      "step": 24365
-    },
-    {
-      "epoch": 12.70717079530639,
-      "grad_norm": 1.366713047027588,
-      "learning_rate": 7.602211055276383e-05,
-      "loss": 5.8227,
-      "step": 24366
-    },
-    {
-      "epoch": 12.707692307692307,
-      "grad_norm": 1.4721421003341675,
-      "learning_rate": 7.602110552763819e-05,
-      "loss": 5.2837,
-      "step": 24367
-    },
-    {
-      "epoch": 12.708213820078226,
-      "grad_norm": 1.477561354637146,
-      "learning_rate": 7.602010050251257e-05,
-      "loss": 5.0041,
-      "step": 24368
-    },
-    {
-      "epoch": 12.708735332464146,
-      "grad_norm": 1.521535038948059,
-      "learning_rate": 7.601909547738693e-05,
-      "loss": 5.3205,
-      "step": 24369
-    },
-    {
-      "epoch": 12.709256844850065,
-      "grad_norm": 1.5003838539123535,
-      "learning_rate": 7.601809045226131e-05,
-      "loss": 5.5445,
-      "step": 24370
-    },
-    {
-      "epoch": 12.709778357235985,
-      "grad_norm": 1.4894654750823975,
-      "learning_rate": 7.601708542713568e-05,
-      "loss": 5.2131,
-      "step": 24371
-    },
-    {
-      "epoch": 12.710299869621904,
-      "grad_norm": 1.4190751314163208,
-      "learning_rate": 7.601608040201005e-05,
-      "loss": 5.6079,
-      "step": 24372
-    },
-    {
-      "epoch": 12.710821382007822,
-      "grad_norm": 1.4663288593292236,
-      "learning_rate": 7.601507537688443e-05,
-      "loss": 5.2136,
-      "step": 24373
-    },
-    {
-      "epoch": 12.711342894393741,
-      "grad_norm": 1.4758223295211792,
-      "learning_rate": 7.60140703517588e-05,
-      "loss": 5.4003,
-      "step": 24374
-    },
-    {
-      "epoch": 12.711864406779661,
-      "grad_norm": 1.4517029523849487,
-      "learning_rate": 7.601306532663317e-05,
-      "loss": 5.1399,
-      "step": 24375
-    },
-    {
-      "epoch": 12.71238591916558,
-      "grad_norm": 1.4407105445861816,
-      "learning_rate": 7.601206030150754e-05,
-      "loss": 5.4242,
-      "step": 24376
-    },
-    {
-      "epoch": 12.7129074315515,
-      "grad_norm": 1.5691652297973633,
-      "learning_rate": 7.601105527638192e-05,
-      "loss": 5.3879,
-      "step": 24377
-    },
-    {
-      "epoch": 12.71342894393742,
-      "grad_norm": 1.342659831047058,
-      "learning_rate": 7.601005025125628e-05,
-      "loss": 5.6063,
-      "step": 24378
-    },
-    {
-      "epoch": 12.713950456323337,
-      "grad_norm": 1.5279192924499512,
-      "learning_rate": 7.600904522613066e-05,
-      "loss": 5.2738,
-      "step": 24379
-    },
-    {
-      "epoch": 12.714471968709256,
-      "grad_norm": 1.3627943992614746,
-      "learning_rate": 7.600804020100502e-05,
-      "loss": 5.5396,
-      "step": 24380
-    },
-    {
-      "epoch": 12.714993481095176,
-      "grad_norm": 1.3977882862091064,
-      "learning_rate": 7.60070351758794e-05,
-      "loss": 5.5392,
-      "step": 24381
-    },
-    {
-      "epoch": 12.715514993481095,
-      "grad_norm": 1.6020926237106323,
-      "learning_rate": 7.600603015075376e-05,
-      "loss": 5.3559,
-      "step": 24382
-    },
-    {
-      "epoch": 12.716036505867015,
-      "grad_norm": 1.4645613431930542,
-      "learning_rate": 7.600502512562814e-05,
-      "loss": 5.4186,
-      "step": 24383
-    },
-    {
-      "epoch": 12.716558018252934,
-      "grad_norm": 1.4330931901931763,
-      "learning_rate": 7.600402010050252e-05,
-      "loss": 5.1623,
-      "step": 24384
-    },
-    {
-      "epoch": 12.717079530638852,
-      "grad_norm": 1.4565563201904297,
-      "learning_rate": 7.60030150753769e-05,
-      "loss": 5.9705,
-      "step": 24385
-    },
-    {
-      "epoch": 12.717601043024771,
-      "grad_norm": 1.4503005743026733,
-      "learning_rate": 7.600201005025126e-05,
-      "loss": 5.9482,
-      "step": 24386
-    },
-    {
-      "epoch": 12.718122555410691,
-      "grad_norm": 1.749280571937561,
-      "learning_rate": 7.600100502512563e-05,
-      "loss": 5.4631,
-      "step": 24387
-    },
-    {
-      "epoch": 12.71864406779661,
-      "grad_norm": 1.303693413734436,
-      "learning_rate": 7.6e-05,
-      "loss": 5.2168,
-      "step": 24388
-    },
-    {
-      "epoch": 12.71916558018253,
-      "grad_norm": 1.3896561861038208,
-      "learning_rate": 7.599899497487437e-05,
-      "loss": 5.6433,
-      "step": 24389
-    },
-    {
-      "epoch": 12.71968709256845,
-      "grad_norm": 1.3905243873596191,
-      "learning_rate": 7.599798994974875e-05,
-      "loss": 5.6029,
-      "step": 24390
-    },
-    {
-      "epoch": 12.720208604954367,
-      "grad_norm": 1.3509730100631714,
-      "learning_rate": 7.599698492462311e-05,
-      "loss": 5.2764,
-      "step": 24391
-    },
-    {
-      "epoch": 12.720730117340286,
-      "grad_norm": 1.4457157850265503,
-      "learning_rate": 7.599597989949749e-05,
-      "loss": 5.3835,
-      "step": 24392
-    },
-    {
-      "epoch": 12.721251629726206,
-      "grad_norm": 1.4932770729064941,
-      "learning_rate": 7.599497487437187e-05,
-      "loss": 5.5482,
-      "step": 24393
-    },
-    {
-      "epoch": 12.721773142112125,
-      "grad_norm": 1.3761019706726074,
-      "learning_rate": 7.599396984924624e-05,
-      "loss": 5.5008,
-      "step": 24394
-    },
-    {
-      "epoch": 12.722294654498045,
-      "grad_norm": 1.4271925687789917,
-      "learning_rate": 7.599296482412061e-05,
-      "loss": 5.7056,
-      "step": 24395
-    },
-    {
-      "epoch": 12.722816166883963,
-      "grad_norm": 1.3666000366210938,
-      "learning_rate": 7.599195979899499e-05,
-      "loss": 5.7291,
-      "step": 24396
-    },
-    {
-      "epoch": 12.723337679269882,
-      "grad_norm": 1.6009079217910767,
-      "learning_rate": 7.599095477386935e-05,
-      "loss": 5.0507,
-      "step": 24397
-    },
-    {
-      "epoch": 12.723859191655801,
-      "grad_norm": 1.5036541223526,
-      "learning_rate": 7.598994974874373e-05,
-      "loss": 5.0745,
-      "step": 24398
-    },
-    {
-      "epoch": 12.724380704041721,
-      "grad_norm": 1.6462188959121704,
-      "learning_rate": 7.598894472361809e-05,
-      "loss": 5.1669,
-      "step": 24399
-    },
-    {
-      "epoch": 12.72490221642764,
-      "grad_norm": 1.3240766525268555,
-      "learning_rate": 7.598793969849246e-05,
-      "loss": 5.3907,
-      "step": 24400
-    },
-    {
-      "epoch": 12.72542372881356,
-      "grad_norm": 1.5743635892868042,
-      "learning_rate": 7.598693467336683e-05,
-      "loss": 5.1259,
-      "step": 24401
-    },
-    {
-      "epoch": 12.72594524119948,
-      "grad_norm": 1.3634172677993774,
-      "learning_rate": 7.59859296482412e-05,
-      "loss": 5.6334,
-      "step": 24402
-    },
-    {
-      "epoch": 12.726466753585397,
-      "grad_norm": 1.5094478130340576,
-      "learning_rate": 7.598492462311558e-05,
-      "loss": 5.2596,
-      "step": 24403
-    },
-    {
-      "epoch": 12.726988265971316,
-      "grad_norm": 1.4883472919464111,
-      "learning_rate": 7.598391959798995e-05,
-      "loss": 5.6445,
-      "step": 24404
-    },
-    {
-      "epoch": 12.727509778357236,
-      "grad_norm": 1.3886327743530273,
-      "learning_rate": 7.598291457286433e-05,
-      "loss": 5.3604,
-      "step": 24405
-    },
-    {
-      "epoch": 12.728031290743155,
-      "grad_norm": 1.5796140432357788,
-      "learning_rate": 7.59819095477387e-05,
-      "loss": 5.003,
-      "step": 24406
-    },
-    {
-      "epoch": 12.728552803129075,
-      "grad_norm": 1.5153332948684692,
-      "learning_rate": 7.598090452261307e-05,
-      "loss": 5.6326,
-      "step": 24407
-    },
-    {
-      "epoch": 12.729074315514993,
-      "grad_norm": 1.4648395776748657,
-      "learning_rate": 7.597989949748744e-05,
-      "loss": 5.3948,
-      "step": 24408
-    },
-    {
-      "epoch": 12.729595827900912,
-      "grad_norm": 1.4774514436721802,
-      "learning_rate": 7.597889447236182e-05,
-      "loss": 5.271,
-      "step": 24409
-    },
-    {
-      "epoch": 12.730117340286832,
-      "grad_norm": 1.3379167318344116,
-      "learning_rate": 7.597788944723618e-05,
-      "loss": 5.8556,
-      "step": 24410
-    },
-    {
-      "epoch": 12.730638852672751,
-      "grad_norm": 1.5000122785568237,
-      "learning_rate": 7.597688442211056e-05,
-      "loss": 5.9195,
-      "step": 24411
-    },
-    {
-      "epoch": 12.73116036505867,
-      "grad_norm": 1.4926040172576904,
-      "learning_rate": 7.597587939698492e-05,
-      "loss": 5.6912,
-      "step": 24412
-    },
-    {
-      "epoch": 12.73168187744459,
-      "grad_norm": 1.642241358757019,
-      "learning_rate": 7.59748743718593e-05,
-      "loss": 4.8152,
-      "step": 24413
-    },
-    {
-      "epoch": 12.73220338983051,
-      "grad_norm": 1.4520188570022583,
-      "learning_rate": 7.597386934673368e-05,
-      "loss": 5.5295,
-      "step": 24414
-    },
-    {
-      "epoch": 12.732724902216427,
-      "grad_norm": 1.4170966148376465,
-      "learning_rate": 7.597286432160804e-05,
-      "loss": 5.2973,
-      "step": 24415
-    },
-    {
-      "epoch": 12.733246414602347,
-      "grad_norm": 1.5295612812042236,
-      "learning_rate": 7.597185929648242e-05,
-      "loss": 5.3146,
-      "step": 24416
-    },
-    {
-      "epoch": 12.733767926988266,
-      "grad_norm": 1.4901432991027832,
-      "learning_rate": 7.597085427135679e-05,
-      "loss": 5.4067,
-      "step": 24417
-    },
-    {
-      "epoch": 12.734289439374185,
-      "grad_norm": 1.4407120943069458,
-      "learning_rate": 7.596984924623116e-05,
-      "loss": 5.824,
-      "step": 24418
-    },
-    {
-      "epoch": 12.734810951760105,
-      "grad_norm": 1.438481330871582,
-      "learning_rate": 7.596884422110553e-05,
-      "loss": 5.5781,
-      "step": 24419
-    },
-    {
-      "epoch": 12.735332464146023,
-      "grad_norm": 1.4472700357437134,
-      "learning_rate": 7.59678391959799e-05,
-      "loss": 5.5665,
-      "step": 24420
-    },
-    {
-      "epoch": 12.735853976531942,
-      "grad_norm": 1.4392955303192139,
-      "learning_rate": 7.596683417085427e-05,
-      "loss": 5.8269,
-      "step": 24421
-    },
-    {
-      "epoch": 12.736375488917862,
-      "grad_norm": 1.4716408252716064,
-      "learning_rate": 7.596582914572865e-05,
-      "loss": 5.4987,
-      "step": 24422
-    },
-    {
-      "epoch": 12.736897001303781,
-      "grad_norm": 1.4219905138015747,
-      "learning_rate": 7.596482412060301e-05,
-      "loss": 5.3849,
-      "step": 24423
-    },
-    {
-      "epoch": 12.7374185136897,
-      "grad_norm": 2.0042507648468018,
-      "learning_rate": 7.596381909547739e-05,
-      "loss": 5.2999,
-      "step": 24424
-    },
-    {
-      "epoch": 12.73794002607562,
-      "grad_norm": 1.4204896688461304,
-      "learning_rate": 7.596281407035177e-05,
-      "loss": 5.4794,
-      "step": 24425
-    },
-    {
-      "epoch": 12.73846153846154,
-      "grad_norm": 1.567097544670105,
-      "learning_rate": 7.596180904522613e-05,
-      "loss": 5.0879,
-      "step": 24426
-    },
-    {
-      "epoch": 12.738983050847457,
-      "grad_norm": 1.6261464357376099,
-      "learning_rate": 7.596080402010051e-05,
-      "loss": 5.2085,
-      "step": 24427
-    },
-    {
-      "epoch": 12.739504563233377,
-      "grad_norm": 1.5107901096343994,
-      "learning_rate": 7.595979899497487e-05,
-      "loss": 5.4325,
-      "step": 24428
-    },
-    {
-      "epoch": 12.740026075619296,
-      "grad_norm": 1.4872994422912598,
-      "learning_rate": 7.595879396984925e-05,
-      "loss": 5.4289,
-      "step": 24429
-    },
-    {
-      "epoch": 12.740547588005215,
-      "grad_norm": 1.710824728012085,
-      "learning_rate": 7.595778894472362e-05,
-      "loss": 5.6704,
-      "step": 24430
-    },
-    {
-      "epoch": 12.741069100391135,
-      "grad_norm": 1.5858311653137207,
-      "learning_rate": 7.5956783919598e-05,
-      "loss": 5.5291,
-      "step": 24431
-    },
-    {
-      "epoch": 12.741590612777053,
-      "grad_norm": 1.4226577281951904,
-      "learning_rate": 7.595577889447236e-05,
-      "loss": 5.49,
-      "step": 24432
-    },
-    {
-      "epoch": 12.742112125162972,
-      "grad_norm": 1.5252529382705688,
-      "learning_rate": 7.595477386934674e-05,
-      "loss": 4.5934,
-      "step": 24433
-    },
-    {
-      "epoch": 12.742633637548892,
-      "grad_norm": 1.5185630321502686,
-      "learning_rate": 7.595376884422111e-05,
-      "loss": 5.4964,
-      "step": 24434
-    },
-    {
-      "epoch": 12.743155149934811,
-      "grad_norm": 1.3094062805175781,
-      "learning_rate": 7.595276381909549e-05,
-      "loss": 4.762,
-      "step": 24435
-    },
-    {
-      "epoch": 12.74367666232073,
-      "grad_norm": 1.469008207321167,
-      "learning_rate": 7.595175879396986e-05,
-      "loss": 5.5265,
-      "step": 24436
-    },
-    {
-      "epoch": 12.74419817470665,
-      "grad_norm": 1.3892662525177002,
-      "learning_rate": 7.595075376884423e-05,
-      "loss": 5.6677,
-      "step": 24437
-    },
-    {
-      "epoch": 12.744719687092568,
-      "grad_norm": 1.4142245054244995,
-      "learning_rate": 7.59497487437186e-05,
-      "loss": 5.6014,
-      "step": 24438
-    },
-    {
-      "epoch": 12.745241199478487,
-      "grad_norm": 1.8250715732574463,
-      "learning_rate": 7.594874371859298e-05,
-      "loss": 4.9789,
-      "step": 24439
-    },
-    {
-      "epoch": 12.745762711864407,
-      "grad_norm": 1.457497477531433,
-      "learning_rate": 7.594773869346734e-05,
-      "loss": 5.7955,
-      "step": 24440
-    },
-    {
-      "epoch": 12.746284224250326,
-      "grad_norm": 1.483396053314209,
-      "learning_rate": 7.59467336683417e-05,
-      "loss": 5.2935,
-      "step": 24441
-    },
-    {
-      "epoch": 12.746805736636245,
-      "grad_norm": 1.3736979961395264,
-      "learning_rate": 7.594572864321608e-05,
-      "loss": 5.4291,
-      "step": 24442
-    },
-    {
-      "epoch": 12.747327249022165,
-      "grad_norm": 1.3602410554885864,
-      "learning_rate": 7.594472361809045e-05,
-      "loss": 5.5918,
-      "step": 24443
-    },
-    {
-      "epoch": 12.747848761408083,
-      "grad_norm": 1.5721389055252075,
-      "learning_rate": 7.594371859296482e-05,
-      "loss": 5.2534,
-      "step": 24444
-    },
-    {
-      "epoch": 12.748370273794002,
-      "grad_norm": 1.368338942527771,
-      "learning_rate": 7.59427135678392e-05,
-      "loss": 5.4527,
-      "step": 24445
-    },
-    {
-      "epoch": 12.748891786179922,
-      "grad_norm": 1.4265849590301514,
-      "learning_rate": 7.594170854271358e-05,
-      "loss": 5.4541,
-      "step": 24446
-    },
-    {
-      "epoch": 12.749413298565841,
-      "grad_norm": 1.603132963180542,
-      "learning_rate": 7.594070351758794e-05,
-      "loss": 5.1432,
-      "step": 24447
-    },
-    {
-      "epoch": 12.74993481095176,
-      "grad_norm": 1.4352918863296509,
-      "learning_rate": 7.593969849246232e-05,
-      "loss": 5.4494,
-      "step": 24448
-    },
-    {
-      "epoch": 12.75045632333768,
-      "grad_norm": 1.3804314136505127,
-      "learning_rate": 7.593869346733669e-05,
-      "loss": 5.5292,
-      "step": 24449
-    },
-    {
-      "epoch": 12.750977835723598,
-      "grad_norm": 1.5542335510253906,
-      "learning_rate": 7.593768844221106e-05,
-      "loss": 5.4825,
-      "step": 24450
-    },
-    {
-      "epoch": 12.751499348109517,
-      "grad_norm": 1.491312026977539,
-      "learning_rate": 7.593668341708543e-05,
-      "loss": 5.4275,
-      "step": 24451
-    },
-    {
-      "epoch": 12.752020860495437,
-      "grad_norm": 1.3868355751037598,
-      "learning_rate": 7.59356783919598e-05,
-      "loss": 5.2631,
-      "step": 24452
-    },
-    {
-      "epoch": 12.752542372881356,
-      "grad_norm": 1.4047391414642334,
-      "learning_rate": 7.593467336683417e-05,
-      "loss": 5.1587,
-      "step": 24453
-    },
-    {
-      "epoch": 12.753063885267276,
-      "grad_norm": 1.4351847171783447,
-      "learning_rate": 7.593366834170855e-05,
-      "loss": 5.4138,
-      "step": 24454
-    },
-    {
-      "epoch": 12.753585397653195,
-      "grad_norm": 1.4722750186920166,
-      "learning_rate": 7.593266331658293e-05,
-      "loss": 5.7876,
-      "step": 24455
-    },
-    {
-      "epoch": 12.754106910039113,
-      "grad_norm": 1.3697633743286133,
-      "learning_rate": 7.593165829145729e-05,
-      "loss": 5.6563,
-      "step": 24456
-    },
-    {
-      "epoch": 12.754628422425032,
-      "grad_norm": 1.3705617189407349,
-      "learning_rate": 7.593065326633167e-05,
-      "loss": 4.9764,
-      "step": 24457
-    },
-    {
-      "epoch": 12.755149934810952,
-      "grad_norm": 1.3481708765029907,
-      "learning_rate": 7.592964824120603e-05,
-      "loss": 5.1014,
-      "step": 24458
-    },
-    {
-      "epoch": 12.755671447196871,
-      "grad_norm": 1.5050820112228394,
-      "learning_rate": 7.592864321608041e-05,
-      "loss": 5.4981,
-      "step": 24459
-    },
-    {
-      "epoch": 12.75619295958279,
-      "grad_norm": 1.4407942295074463,
-      "learning_rate": 7.592763819095477e-05,
-      "loss": 5.5589,
-      "step": 24460
-    },
-    {
-      "epoch": 12.75671447196871,
-      "grad_norm": 1.5081263780593872,
-      "learning_rate": 7.592663316582915e-05,
-      "loss": 5.4533,
-      "step": 24461
-    },
-    {
-      "epoch": 12.757235984354628,
-      "grad_norm": 1.321273922920227,
-      "learning_rate": 7.592562814070352e-05,
-      "loss": 5.1324,
-      "step": 24462
-    },
-    {
-      "epoch": 12.757757496740547,
-      "grad_norm": 1.4555095434188843,
-      "learning_rate": 7.59246231155779e-05,
-      "loss": 5.3874,
-      "step": 24463
-    },
-    {
-      "epoch": 12.758279009126467,
-      "grad_norm": 1.5075820684432983,
-      "learning_rate": 7.592361809045226e-05,
-      "loss": 5.4763,
-      "step": 24464
-    },
-    {
-      "epoch": 12.758800521512386,
-      "grad_norm": 1.5498849153518677,
-      "learning_rate": 7.592261306532664e-05,
-      "loss": 5.1063,
-      "step": 24465
-    },
-    {
-      "epoch": 12.759322033898306,
-      "grad_norm": 1.5460013151168823,
-      "learning_rate": 7.592160804020101e-05,
-      "loss": 5.4162,
-      "step": 24466
-    },
-    {
-      "epoch": 12.759843546284225,
-      "grad_norm": 1.4269638061523438,
-      "learning_rate": 7.592060301507538e-05,
-      "loss": 5.355,
-      "step": 24467
-    },
-    {
-      "epoch": 12.760365058670143,
-      "grad_norm": 1.4727325439453125,
-      "learning_rate": 7.591959798994976e-05,
-      "loss": 5.2076,
-      "step": 24468
-    },
-    {
-      "epoch": 12.760886571056062,
-      "grad_norm": 1.4000864028930664,
-      "learning_rate": 7.591859296482412e-05,
-      "loss": 5.1894,
-      "step": 24469
-    },
-    {
-      "epoch": 12.761408083441982,
-      "grad_norm": 1.3517787456512451,
-      "learning_rate": 7.59175879396985e-05,
-      "loss": 5.0786,
-      "step": 24470
-    },
-    {
-      "epoch": 12.761929595827901,
-      "grad_norm": 1.4659368991851807,
-      "learning_rate": 7.591658291457286e-05,
-      "loss": 5.1884,
-      "step": 24471
-    },
-    {
-      "epoch": 12.76245110821382,
-      "grad_norm": 1.5233185291290283,
-      "learning_rate": 7.591557788944724e-05,
-      "loss": 5.384,
-      "step": 24472
-    },
-    {
-      "epoch": 12.76297262059974,
-      "grad_norm": 1.5392242670059204,
-      "learning_rate": 7.59145728643216e-05,
-      "loss": 5.4176,
-      "step": 24473
-    },
-    {
-      "epoch": 12.763494132985658,
-      "grad_norm": 1.5026501417160034,
-      "learning_rate": 7.591356783919598e-05,
-      "loss": 5.5277,
-      "step": 24474
-    },
-    {
-      "epoch": 12.764015645371577,
-      "grad_norm": 1.5138453245162964,
-      "learning_rate": 7.591256281407036e-05,
-      "loss": 5.1506,
-      "step": 24475
-    },
-    {
-      "epoch": 12.764537157757497,
-      "grad_norm": 1.4911173582077026,
-      "learning_rate": 7.591155778894474e-05,
-      "loss": 5.646,
-      "step": 24476
-    },
-    {
-      "epoch": 12.765058670143416,
-      "grad_norm": 1.6545356512069702,
-      "learning_rate": 7.59105527638191e-05,
-      "loss": 5.678,
-      "step": 24477
-    },
-    {
-      "epoch": 12.765580182529336,
-      "grad_norm": 1.4842973947525024,
-      "learning_rate": 7.590954773869348e-05,
-      "loss": 5.3708,
-      "step": 24478
-    },
-    {
-      "epoch": 12.766101694915255,
-      "grad_norm": 1.4955476522445679,
-      "learning_rate": 7.590854271356784e-05,
-      "loss": 5.028,
-      "step": 24479
-    },
-    {
-      "epoch": 12.766623207301173,
-      "grad_norm": 1.508634090423584,
-      "learning_rate": 7.590753768844221e-05,
-      "loss": 5.77,
-      "step": 24480
-    },
-    {
-      "epoch": 12.767144719687092,
-      "grad_norm": 1.4501996040344238,
-      "learning_rate": 7.590653266331659e-05,
-      "loss": 5.5554,
-      "step": 24481
-    },
-    {
-      "epoch": 12.767666232073012,
-      "grad_norm": 1.5697319507598877,
-      "learning_rate": 7.590552763819095e-05,
-      "loss": 5.2541,
-      "step": 24482
-    },
-    {
-      "epoch": 12.768187744458931,
-      "grad_norm": 1.634009838104248,
-      "learning_rate": 7.590452261306533e-05,
-      "loss": 5.4781,
-      "step": 24483
-    },
-    {
-      "epoch": 12.76870925684485,
-      "grad_norm": 1.407063603401184,
-      "learning_rate": 7.590351758793969e-05,
-      "loss": 5.7823,
-      "step": 24484
-    },
-    {
-      "epoch": 12.76923076923077,
-      "grad_norm": 1.4577723741531372,
-      "learning_rate": 7.590251256281407e-05,
-      "loss": 5.2127,
-      "step": 24485
-    },
-    {
-      "epoch": 12.769752281616688,
-      "grad_norm": 1.340656042098999,
-      "learning_rate": 7.590150753768845e-05,
-      "loss": 5.5296,
-      "step": 24486
-    },
-    {
-      "epoch": 12.770273794002607,
-      "grad_norm": 1.4389591217041016,
-      "learning_rate": 7.590050251256283e-05,
-      "loss": 5.6096,
-      "step": 24487
-    },
-    {
-      "epoch": 12.770795306388527,
-      "grad_norm": 1.636777400970459,
-      "learning_rate": 7.589949748743719e-05,
-      "loss": 4.9154,
-      "step": 24488
-    },
-    {
-      "epoch": 12.771316818774446,
-      "grad_norm": 1.4798144102096558,
-      "learning_rate": 7.589849246231157e-05,
-      "loss": 5.3381,
-      "step": 24489
-    },
-    {
-      "epoch": 12.771838331160366,
-      "grad_norm": 1.508229374885559,
-      "learning_rate": 7.589748743718593e-05,
-      "loss": 5.3649,
-      "step": 24490
-    },
-    {
-      "epoch": 12.772359843546283,
-      "grad_norm": 1.403053879737854,
-      "learning_rate": 7.589648241206031e-05,
-      "loss": 5.3456,
-      "step": 24491
-    },
-    {
-      "epoch": 12.772881355932203,
-      "grad_norm": 1.3963004350662231,
-      "learning_rate": 7.589547738693467e-05,
-      "loss": 5.6098,
-      "step": 24492
-    },
-    {
-      "epoch": 12.773402868318122,
-      "grad_norm": 1.49527907371521,
-      "learning_rate": 7.589447236180904e-05,
-      "loss": 5.3411,
-      "step": 24493
-    },
-    {
-      "epoch": 12.773924380704042,
-      "grad_norm": 1.3587065935134888,
-      "learning_rate": 7.589346733668342e-05,
-      "loss": 5.7066,
-      "step": 24494
-    },
-    {
-      "epoch": 12.774445893089961,
-      "grad_norm": 1.4166120290756226,
-      "learning_rate": 7.58924623115578e-05,
-      "loss": 5.298,
-      "step": 24495
-    },
-    {
-      "epoch": 12.77496740547588,
-      "grad_norm": 1.4985336065292358,
-      "learning_rate": 7.589145728643217e-05,
-      "loss": 5.5796,
-      "step": 24496
-    },
-    {
-      "epoch": 12.7754889178618,
-      "grad_norm": 1.3810404539108276,
-      "learning_rate": 7.589045226130654e-05,
-      "loss": 5.1456,
-      "step": 24497
-    },
-    {
-      "epoch": 12.776010430247718,
-      "grad_norm": 1.494794249534607,
-      "learning_rate": 7.588944723618091e-05,
-      "loss": 5.127,
-      "step": 24498
-    },
-    {
-      "epoch": 12.776531942633637,
-      "grad_norm": 1.365222454071045,
-      "learning_rate": 7.588844221105528e-05,
-      "loss": 5.1753,
-      "step": 24499
-    },
-    {
-      "epoch": 12.777053455019557,
-      "grad_norm": 1.471259355545044,
-      "learning_rate": 7.588743718592966e-05,
-      "loss": 5.4338,
-      "step": 24500
-    },
-    {
-      "epoch": 12.777574967405476,
-      "grad_norm": 1.4088819026947021,
-      "learning_rate": 7.588643216080402e-05,
-      "loss": 5.475,
-      "step": 24501
-    },
-    {
-      "epoch": 12.778096479791396,
-      "grad_norm": 1.3831071853637695,
-      "learning_rate": 7.58854271356784e-05,
-      "loss": 5.2094,
-      "step": 24502
-    },
-    {
-      "epoch": 12.778617992177313,
-      "grad_norm": 1.5086346864700317,
-      "learning_rate": 7.588442211055276e-05,
-      "loss": 5.1499,
-      "step": 24503
-    },
-    {
-      "epoch": 12.779139504563233,
-      "grad_norm": 1.4440158605575562,
-      "learning_rate": 7.588341708542714e-05,
-      "loss": 5.1501,
-      "step": 24504
-    },
-    {
-      "epoch": 12.779661016949152,
-      "grad_norm": 1.4619547128677368,
-      "learning_rate": 7.58824120603015e-05,
-      "loss": 5.5021,
-      "step": 24505
-    },
-    {
-      "epoch": 12.780182529335072,
-      "grad_norm": 1.4613354206085205,
-      "learning_rate": 7.588140703517588e-05,
-      "loss": 5.6212,
-      "step": 24506
-    },
-    {
-      "epoch": 12.780704041720991,
-      "grad_norm": 1.5420024394989014,
-      "learning_rate": 7.588040201005026e-05,
-      "loss": 5.2748,
-      "step": 24507
-    },
-    {
-      "epoch": 12.78122555410691,
-      "grad_norm": 1.5137053728103638,
-      "learning_rate": 7.587939698492463e-05,
-      "loss": 5.2766,
-      "step": 24508
-    },
-    {
-      "epoch": 12.78174706649283,
-      "grad_norm": 1.7658950090408325,
-      "learning_rate": 7.5878391959799e-05,
-      "loss": 4.9366,
-      "step": 24509
-    },
-    {
-      "epoch": 12.782268578878748,
-      "grad_norm": 1.461885690689087,
-      "learning_rate": 7.587738693467337e-05,
-      "loss": 5.6584,
-      "step": 24510
-    },
-    {
-      "epoch": 12.782790091264667,
-      "grad_norm": 1.4473845958709717,
-      "learning_rate": 7.587638190954775e-05,
-      "loss": 5.4796,
-      "step": 24511
-    },
-    {
-      "epoch": 12.783311603650587,
-      "grad_norm": 1.4687386751174927,
-      "learning_rate": 7.587537688442211e-05,
-      "loss": 5.2223,
-      "step": 24512
-    },
-    {
-      "epoch": 12.783833116036506,
-      "grad_norm": 1.4030112028121948,
-      "learning_rate": 7.587437185929649e-05,
-      "loss": 5.6082,
-      "step": 24513
-    },
-    {
-      "epoch": 12.784354628422426,
-      "grad_norm": 1.4951237440109253,
-      "learning_rate": 7.587336683417085e-05,
-      "loss": 5.5026,
-      "step": 24514
-    },
-    {
-      "epoch": 12.784876140808343,
-      "grad_norm": 1.42330002784729,
-      "learning_rate": 7.587236180904523e-05,
-      "loss": 5.2129,
-      "step": 24515
-    },
-    {
-      "epoch": 12.785397653194263,
-      "grad_norm": 1.3895014524459839,
-      "learning_rate": 7.587135678391961e-05,
-      "loss": 5.157,
-      "step": 24516
-    },
-    {
-      "epoch": 12.785919165580182,
-      "grad_norm": 1.3528995513916016,
-      "learning_rate": 7.587035175879399e-05,
-      "loss": 5.8409,
-      "step": 24517
-    },
-    {
-      "epoch": 12.786440677966102,
-      "grad_norm": 1.4449176788330078,
-      "learning_rate": 7.586934673366835e-05,
-      "loss": 5.4043,
-      "step": 24518
-    },
-    {
-      "epoch": 12.786962190352021,
-      "grad_norm": 1.6309062242507935,
-      "learning_rate": 7.586834170854271e-05,
-      "loss": 5.0753,
-      "step": 24519
-    },
-    {
-      "epoch": 12.78748370273794,
-      "grad_norm": 1.383406162261963,
-      "learning_rate": 7.586733668341709e-05,
-      "loss": 5.7901,
-      "step": 24520
-    },
-    {
-      "epoch": 12.78800521512386,
-      "grad_norm": 1.454232096672058,
-      "learning_rate": 7.586633165829146e-05,
-      "loss": 5.4033,
-      "step": 24521
-    },
-    {
-      "epoch": 12.788526727509778,
-      "grad_norm": 1.4895360469818115,
-      "learning_rate": 7.586532663316583e-05,
-      "loss": 5.2949,
-      "step": 24522
-    },
-    {
-      "epoch": 12.789048239895697,
-      "grad_norm": 1.5504213571548462,
-      "learning_rate": 7.58643216080402e-05,
-      "loss": 5.2135,
-      "step": 24523
-    },
-    {
-      "epoch": 12.789569752281617,
-      "grad_norm": 1.3436044454574585,
-      "learning_rate": 7.586331658291458e-05,
-      "loss": 5.9614,
-      "step": 24524
-    },
-    {
-      "epoch": 12.790091264667536,
-      "grad_norm": 1.5209850072860718,
-      "learning_rate": 7.586231155778894e-05,
-      "loss": 5.5668,
-      "step": 24525
-    },
-    {
-      "epoch": 12.790612777053456,
-      "grad_norm": 1.60165536403656,
-      "learning_rate": 7.586130653266332e-05,
-      "loss": 4.615,
-      "step": 24526
-    },
-    {
-      "epoch": 12.791134289439373,
-      "grad_norm": 1.5910780429840088,
-      "learning_rate": 7.58603015075377e-05,
-      "loss": 5.3874,
-      "step": 24527
-    },
-    {
-      "epoch": 12.791655801825293,
-      "grad_norm": 1.484256386756897,
-      "learning_rate": 7.585929648241207e-05,
-      "loss": 5.3925,
-      "step": 24528
-    },
-    {
-      "epoch": 12.792177314211212,
-      "grad_norm": 1.432582974433899,
-      "learning_rate": 7.585829145728644e-05,
-      "loss": 5.561,
-      "step": 24529
-    },
-    {
-      "epoch": 12.792698826597132,
-      "grad_norm": 1.3557921648025513,
-      "learning_rate": 7.585728643216082e-05,
-      "loss": 5.7018,
-      "step": 24530
-    },
-    {
-      "epoch": 12.793220338983051,
-      "grad_norm": 1.3070831298828125,
-      "learning_rate": 7.585628140703518e-05,
-      "loss": 5.1476,
-      "step": 24531
-    },
-    {
-      "epoch": 12.79374185136897,
-      "grad_norm": 1.5297027826309204,
-      "learning_rate": 7.585527638190956e-05,
-      "loss": 5.1804,
-      "step": 24532
-    },
-    {
-      "epoch": 12.794263363754888,
-      "grad_norm": 1.3827495574951172,
-      "learning_rate": 7.585427135678392e-05,
-      "loss": 5.7287,
-      "step": 24533
-    },
-    {
-      "epoch": 12.794784876140808,
-      "grad_norm": 1.4751304388046265,
-      "learning_rate": 7.585326633165829e-05,
-      "loss": 5.4131,
-      "step": 24534
-    },
-    {
-      "epoch": 12.795306388526727,
-      "grad_norm": 1.4296321868896484,
-      "learning_rate": 7.585226130653266e-05,
-      "loss": 5.4956,
-      "step": 24535
-    },
-    {
-      "epoch": 12.795827900912647,
-      "grad_norm": 1.2894346714019775,
-      "learning_rate": 7.585125628140703e-05,
-      "loss": 5.5422,
-      "step": 24536
-    },
-    {
-      "epoch": 12.796349413298566,
-      "grad_norm": 1.3907595872879028,
-      "learning_rate": 7.58502512562814e-05,
-      "loss": 5.5871,
-      "step": 24537
-    },
-    {
-      "epoch": 12.796870925684486,
-      "grad_norm": 1.3855032920837402,
-      "learning_rate": 7.584924623115578e-05,
-      "loss": 5.7338,
-      "step": 24538
-    },
-    {
-      "epoch": 12.797392438070403,
-      "grad_norm": 1.5798546075820923,
-      "learning_rate": 7.584824120603016e-05,
-      "loss": 5.3166,
-      "step": 24539
-    },
-    {
-      "epoch": 12.797913950456323,
-      "grad_norm": 1.3911021947860718,
-      "learning_rate": 7.584723618090453e-05,
-      "loss": 5.0137,
-      "step": 24540
-    },
-    {
-      "epoch": 12.798435462842242,
-      "grad_norm": 1.4940603971481323,
-      "learning_rate": 7.58462311557789e-05,
-      "loss": 5.1354,
-      "step": 24541
-    },
-    {
-      "epoch": 12.798956975228162,
-      "grad_norm": 1.3526928424835205,
-      "learning_rate": 7.584522613065327e-05,
-      "loss": 5.6737,
-      "step": 24542
-    },
-    {
-      "epoch": 12.799478487614081,
-      "grad_norm": 1.4715166091918945,
-      "learning_rate": 7.584422110552765e-05,
-      "loss": 5.5175,
-      "step": 24543
-    },
-    {
-      "epoch": 12.8,
-      "grad_norm": 1.5627928972244263,
-      "learning_rate": 7.584321608040201e-05,
-      "loss": 5.5459,
-      "step": 24544
-    },
-    {
-      "epoch": 12.800521512385918,
-      "grad_norm": 1.4839084148406982,
-      "learning_rate": 7.584221105527639e-05,
-      "loss": 5.1083,
-      "step": 24545
-    },
-    {
-      "epoch": 12.801043024771838,
-      "grad_norm": 1.4555813074111938,
-      "learning_rate": 7.584120603015075e-05,
-      "loss": 5.2082,
-      "step": 24546
-    },
-    {
-      "epoch": 12.801564537157757,
-      "grad_norm": 1.4859861135482788,
-      "learning_rate": 7.584020100502513e-05,
-      "loss": 5.6679,
-      "step": 24547
-    },
-    {
-      "epoch": 12.802086049543677,
-      "grad_norm": 1.43168306350708,
-      "learning_rate": 7.583919597989951e-05,
-      "loss": 5.4612,
-      "step": 24548
-    },
-    {
-      "epoch": 12.802607561929596,
-      "grad_norm": 1.4660041332244873,
-      "learning_rate": 7.583819095477387e-05,
-      "loss": 5.2578,
-      "step": 24549
-    },
-    {
-      "epoch": 12.803129074315516,
-      "grad_norm": 1.3965574502944946,
-      "learning_rate": 7.583718592964825e-05,
-      "loss": 5.5429,
-      "step": 24550
-    },
-    {
-      "epoch": 12.803650586701433,
-      "grad_norm": 1.3908474445343018,
-      "learning_rate": 7.583618090452261e-05,
-      "loss": 5.2491,
-      "step": 24551
-    },
-    {
-      "epoch": 12.804172099087353,
-      "grad_norm": 1.4310665130615234,
-      "learning_rate": 7.583517587939699e-05,
-      "loss": 5.2824,
-      "step": 24552
-    },
-    {
-      "epoch": 12.804693611473272,
-      "grad_norm": 1.4238231182098389,
-      "learning_rate": 7.583417085427136e-05,
-      "loss": 5.2977,
-      "step": 24553
-    },
-    {
-      "epoch": 12.805215123859192,
-      "grad_norm": 1.4451740980148315,
-      "learning_rate": 7.583316582914573e-05,
-      "loss": 5.4287,
-      "step": 24554
-    },
-    {
-      "epoch": 12.805736636245111,
-      "grad_norm": 1.395241379737854,
-      "learning_rate": 7.58321608040201e-05,
-      "loss": 5.518,
-      "step": 24555
-    },
-    {
-      "epoch": 12.80625814863103,
-      "grad_norm": 1.5774917602539062,
-      "learning_rate": 7.583115577889448e-05,
-      "loss": 4.7822,
-      "step": 24556
-    },
-    {
-      "epoch": 12.806779661016948,
-      "grad_norm": 1.5945773124694824,
-      "learning_rate": 7.583015075376884e-05,
-      "loss": 5.1921,
-      "step": 24557
-    },
-    {
-      "epoch": 12.807301173402868,
-      "grad_norm": 1.505287528038025,
-      "learning_rate": 7.582914572864322e-05,
-      "loss": 4.7526,
-      "step": 24558
-    },
-    {
-      "epoch": 12.807822685788787,
-      "grad_norm": 1.4281015396118164,
-      "learning_rate": 7.58281407035176e-05,
-      "loss": 5.2252,
-      "step": 24559
-    },
-    {
-      "epoch": 12.808344198174707,
-      "grad_norm": 1.511772632598877,
-      "learning_rate": 7.582713567839196e-05,
-      "loss": 5.1247,
-      "step": 24560
-    },
-    {
-      "epoch": 12.808865710560626,
-      "grad_norm": 1.492177128791809,
-      "learning_rate": 7.582613065326634e-05,
-      "loss": 5.5545,
-      "step": 24561
-    },
-    {
-      "epoch": 12.809387222946546,
-      "grad_norm": 1.4830312728881836,
-      "learning_rate": 7.58251256281407e-05,
-      "loss": 5.6446,
-      "step": 24562
-    },
-    {
-      "epoch": 12.809908735332463,
-      "grad_norm": 1.4984049797058105,
-      "learning_rate": 7.582412060301508e-05,
-      "loss": 5.4812,
-      "step": 24563
-    },
-    {
-      "epoch": 12.810430247718383,
-      "grad_norm": 1.5742549896240234,
-      "learning_rate": 7.582311557788944e-05,
-      "loss": 4.9532,
-      "step": 24564
-    },
-    {
-      "epoch": 12.810951760104302,
-      "grad_norm": 1.488010287284851,
-      "learning_rate": 7.582211055276382e-05,
-      "loss": 5.4845,
-      "step": 24565
-    },
-    {
-      "epoch": 12.811473272490222,
-      "grad_norm": 1.4581317901611328,
-      "learning_rate": 7.582110552763819e-05,
-      "loss": 5.5042,
-      "step": 24566
-    },
-    {
-      "epoch": 12.811994784876141,
-      "grad_norm": 1.4735196828842163,
-      "learning_rate": 7.582010050251256e-05,
-      "loss": 5.6493,
-      "step": 24567
-    },
-    {
-      "epoch": 12.81251629726206,
-      "grad_norm": 1.4385771751403809,
-      "learning_rate": 7.581909547738694e-05,
-      "loss": 5.741,
-      "step": 24568
-    },
-    {
-      "epoch": 12.813037809647978,
-      "grad_norm": 1.415540099143982,
-      "learning_rate": 7.581809045226132e-05,
-      "loss": 5.7768,
-      "step": 24569
-    },
-    {
-      "epoch": 12.813559322033898,
-      "grad_norm": 1.4559924602508545,
-      "learning_rate": 7.581708542713568e-05,
-      "loss": 5.509,
-      "step": 24570
-    },
-    {
-      "epoch": 12.814080834419817,
-      "grad_norm": 1.470900058746338,
-      "learning_rate": 7.581608040201006e-05,
-      "loss": 5.3688,
-      "step": 24571
-    },
-    {
-      "epoch": 12.814602346805737,
-      "grad_norm": 1.5045692920684814,
-      "learning_rate": 7.581507537688443e-05,
-      "loss": 5.4026,
-      "step": 24572
-    },
-    {
-      "epoch": 12.815123859191656,
-      "grad_norm": 1.4353454113006592,
-      "learning_rate": 7.581407035175879e-05,
-      "loss": 5.5834,
-      "step": 24573
-    },
-    {
-      "epoch": 12.815645371577576,
-      "grad_norm": 1.4498240947723389,
-      "learning_rate": 7.581306532663317e-05,
-      "loss": 5.8123,
-      "step": 24574
-    },
-    {
-      "epoch": 12.816166883963493,
-      "grad_norm": 1.5003852844238281,
-      "learning_rate": 7.581206030150753e-05,
-      "loss": 5.346,
-      "step": 24575
-    },
-    {
-      "epoch": 12.816688396349413,
-      "grad_norm": 1.4373533725738525,
-      "learning_rate": 7.581105527638191e-05,
-      "loss": 5.7674,
-      "step": 24576
-    },
-    {
-      "epoch": 12.817209908735332,
-      "grad_norm": 1.5096843242645264,
-      "learning_rate": 7.581005025125628e-05,
-      "loss": 5.6016,
-      "step": 24577
-    },
-    {
-      "epoch": 12.817731421121252,
-      "grad_norm": 1.4364408254623413,
-      "learning_rate": 7.580904522613065e-05,
-      "loss": 4.6783,
-      "step": 24578
-    },
-    {
-      "epoch": 12.818252933507171,
-      "grad_norm": 1.4959897994995117,
-      "learning_rate": 7.580804020100503e-05,
-      "loss": 5.0824,
-      "step": 24579
-    },
-    {
-      "epoch": 12.81877444589309,
-      "grad_norm": 1.4754608869552612,
-      "learning_rate": 7.580703517587941e-05,
-      "loss": 5.6108,
-      "step": 24580
-    },
-    {
-      "epoch": 12.819295958279008,
-      "grad_norm": 1.4132837057113647,
-      "learning_rate": 7.580603015075377e-05,
-      "loss": 5.539,
-      "step": 24581
-    },
-    {
-      "epoch": 12.819817470664928,
-      "grad_norm": 1.4900455474853516,
-      "learning_rate": 7.580502512562815e-05,
-      "loss": 5.2632,
-      "step": 24582
-    },
-    {
-      "epoch": 12.820338983050847,
-      "grad_norm": 1.5477955341339111,
-      "learning_rate": 7.580402010050252e-05,
-      "loss": 4.6157,
-      "step": 24583
-    },
-    {
-      "epoch": 12.820860495436767,
-      "grad_norm": 1.5465805530548096,
-      "learning_rate": 7.580301507537689e-05,
-      "loss": 5.1586,
-      "step": 24584
-    },
-    {
-      "epoch": 12.821382007822686,
-      "grad_norm": 1.506121039390564,
-      "learning_rate": 7.580201005025126e-05,
-      "loss": 5.1493,
-      "step": 24585
-    },
-    {
-      "epoch": 12.821903520208604,
-      "grad_norm": 1.4871801137924194,
-      "learning_rate": 7.580100502512562e-05,
-      "loss": 5.6211,
-      "step": 24586
-    },
-    {
-      "epoch": 12.822425032594523,
-      "grad_norm": 1.4952073097229004,
-      "learning_rate": 7.58e-05,
-      "loss": 5.4179,
-      "step": 24587
-    },
-    {
-      "epoch": 12.822946544980443,
-      "grad_norm": 1.3834048509597778,
-      "learning_rate": 7.579899497487438e-05,
-      "loss": 5.0119,
-      "step": 24588
-    },
-    {
-      "epoch": 12.823468057366362,
-      "grad_norm": 1.5951744318008423,
-      "learning_rate": 7.579798994974875e-05,
-      "loss": 4.8816,
-      "step": 24589
-    },
-    {
-      "epoch": 12.823989569752282,
-      "grad_norm": 1.4223023653030396,
-      "learning_rate": 7.579698492462312e-05,
-      "loss": 5.5994,
-      "step": 24590
-    },
-    {
-      "epoch": 12.824511082138201,
-      "grad_norm": 1.4151676893234253,
-      "learning_rate": 7.57959798994975e-05,
-      "loss": 5.5993,
-      "step": 24591
-    },
-    {
-      "epoch": 12.82503259452412,
-      "grad_norm": 1.41693913936615,
-      "learning_rate": 7.579497487437186e-05,
-      "loss": 5.3093,
-      "step": 24592
-    },
-    {
-      "epoch": 12.825554106910038,
-      "grad_norm": 1.494327187538147,
-      "learning_rate": 7.579396984924624e-05,
-      "loss": 5.5578,
-      "step": 24593
-    },
-    {
-      "epoch": 12.826075619295958,
-      "grad_norm": 1.4850293397903442,
-      "learning_rate": 7.57929648241206e-05,
-      "loss": 5.4049,
-      "step": 24594
-    },
-    {
-      "epoch": 12.826597131681877,
-      "grad_norm": 1.375154972076416,
-      "learning_rate": 7.579195979899498e-05,
-      "loss": 5.5599,
-      "step": 24595
-    },
-    {
-      "epoch": 12.827118644067797,
-      "grad_norm": 1.5015965700149536,
-      "learning_rate": 7.579095477386935e-05,
-      "loss": 5.7374,
-      "step": 24596
-    },
-    {
-      "epoch": 12.827640156453716,
-      "grad_norm": 1.6176033020019531,
-      "learning_rate": 7.578994974874372e-05,
-      "loss": 5.2466,
-      "step": 24597
-    },
-    {
-      "epoch": 12.828161668839634,
-      "grad_norm": 1.4474153518676758,
-      "learning_rate": 7.578894472361809e-05,
-      "loss": 5.3839,
-      "step": 24598
-    },
-    {
-      "epoch": 12.828683181225554,
-      "grad_norm": 1.465911865234375,
-      "learning_rate": 7.578793969849247e-05,
-      "loss": 5.2779,
-      "step": 24599
-    },
-    {
-      "epoch": 12.829204693611473,
-      "grad_norm": 1.452130675315857,
-      "learning_rate": 7.578693467336684e-05,
-      "loss": 5.4289,
-      "step": 24600
-    },
-    {
-      "epoch": 12.829726205997392,
-      "grad_norm": 1.3997102975845337,
-      "learning_rate": 7.578592964824121e-05,
-      "loss": 5.6301,
-      "step": 24601
-    },
-    {
-      "epoch": 12.830247718383312,
-      "grad_norm": 1.5454487800598145,
-      "learning_rate": 7.578492462311559e-05,
-      "loss": 5.2869,
-      "step": 24602
-    },
-    {
-      "epoch": 12.830769230769231,
-      "grad_norm": 1.4041472673416138,
-      "learning_rate": 7.578391959798995e-05,
-      "loss": 5.5771,
-      "step": 24603
-    },
-    {
-      "epoch": 12.83129074315515,
-      "grad_norm": 1.428382158279419,
-      "learning_rate": 7.578291457286433e-05,
-      "loss": 5.5219,
-      "step": 24604
-    },
-    {
-      "epoch": 12.831812255541069,
-      "grad_norm": 1.3847894668579102,
-      "learning_rate": 7.578190954773869e-05,
-      "loss": 5.4516,
-      "step": 24605
-    },
-    {
-      "epoch": 12.832333767926988,
-      "grad_norm": 1.3994027376174927,
-      "learning_rate": 7.578090452261307e-05,
-      "loss": 5.6267,
-      "step": 24606
-    },
-    {
-      "epoch": 12.832855280312907,
-      "grad_norm": 1.3892827033996582,
-      "learning_rate": 7.577989949748743e-05,
-      "loss": 5.3245,
-      "step": 24607
-    },
-    {
-      "epoch": 12.833376792698827,
-      "grad_norm": 1.3365362882614136,
-      "learning_rate": 7.577889447236181e-05,
-      "loss": 5.7082,
-      "step": 24608
-    },
-    {
-      "epoch": 12.833898305084746,
-      "grad_norm": 1.393423080444336,
-      "learning_rate": 7.577788944723619e-05,
-      "loss": 5.2247,
-      "step": 24609
-    },
-    {
-      "epoch": 12.834419817470664,
-      "grad_norm": 1.4267898797988892,
-      "learning_rate": 7.577688442211057e-05,
-      "loss": 5.4871,
-      "step": 24610
-    },
-    {
-      "epoch": 12.834941329856584,
-      "grad_norm": 1.383127212524414,
-      "learning_rate": 7.577587939698493e-05,
-      "loss": 5.5478,
-      "step": 24611
-    },
-    {
-      "epoch": 12.835462842242503,
-      "grad_norm": 1.5042227506637573,
-      "learning_rate": 7.57748743718593e-05,
-      "loss": 5.2516,
-      "step": 24612
-    },
-    {
-      "epoch": 12.835984354628422,
-      "grad_norm": 1.3994364738464355,
-      "learning_rate": 7.577386934673367e-05,
-      "loss": 5.3072,
-      "step": 24613
-    },
-    {
-      "epoch": 12.836505867014342,
-      "grad_norm": 1.4603432416915894,
-      "learning_rate": 7.577286432160804e-05,
-      "loss": 5.5117,
-      "step": 24614
-    },
-    {
-      "epoch": 12.837027379400261,
-      "grad_norm": 1.3548238277435303,
-      "learning_rate": 7.577185929648242e-05,
-      "loss": 5.6875,
-      "step": 24615
-    },
-    {
-      "epoch": 12.83754889178618,
-      "grad_norm": 1.4390456676483154,
-      "learning_rate": 7.577085427135678e-05,
-      "loss": 5.4838,
-      "step": 24616
-    },
-    {
-      "epoch": 12.838070404172099,
-      "grad_norm": 1.3834501504898071,
-      "learning_rate": 7.576984924623116e-05,
-      "loss": 5.3341,
-      "step": 24617
-    },
-    {
-      "epoch": 12.838591916558018,
-      "grad_norm": 1.5526729822158813,
-      "learning_rate": 7.576884422110552e-05,
-      "loss": 5.0434,
-      "step": 24618
-    },
-    {
-      "epoch": 12.839113428943937,
-      "grad_norm": 1.6381199359893799,
-      "learning_rate": 7.57678391959799e-05,
-      "loss": 5.147,
-      "step": 24619
-    },
-    {
-      "epoch": 12.839634941329857,
-      "grad_norm": 1.533531904220581,
-      "learning_rate": 7.576683417085428e-05,
-      "loss": 5.7222,
-      "step": 24620
-    },
-    {
-      "epoch": 12.840156453715776,
-      "grad_norm": 1.6220197677612305,
-      "learning_rate": 7.576582914572866e-05,
-      "loss": 4.5642,
-      "step": 24621
-    },
-    {
-      "epoch": 12.840677966101694,
-      "grad_norm": 1.6668591499328613,
-      "learning_rate": 7.576482412060302e-05,
-      "loss": 5.6086,
-      "step": 24622
-    },
-    {
-      "epoch": 12.841199478487614,
-      "grad_norm": 1.5508743524551392,
-      "learning_rate": 7.57638190954774e-05,
-      "loss": 4.7793,
-      "step": 24623
-    },
-    {
-      "epoch": 12.841720990873533,
-      "grad_norm": 1.5960558652877808,
-      "learning_rate": 7.576281407035176e-05,
-      "loss": 5.3743,
-      "step": 24624
-    },
-    {
-      "epoch": 12.842242503259452,
-      "grad_norm": 1.4818769693374634,
-      "learning_rate": 7.576180904522614e-05,
-      "loss": 5.4784,
-      "step": 24625
-    },
-    {
-      "epoch": 12.842764015645372,
-      "grad_norm": 1.5495015382766724,
-      "learning_rate": 7.57608040201005e-05,
-      "loss": 5.104,
-      "step": 24626
-    },
-    {
-      "epoch": 12.843285528031291,
-      "grad_norm": 1.4388843774795532,
-      "learning_rate": 7.575979899497487e-05,
-      "loss": 4.8785,
-      "step": 24627
-    },
-    {
-      "epoch": 12.843807040417209,
-      "grad_norm": 1.2963098287582397,
-      "learning_rate": 7.575879396984925e-05,
-      "loss": 5.8735,
-      "step": 24628
-    },
-    {
-      "epoch": 12.844328552803129,
-      "grad_norm": 1.4902575016021729,
-      "learning_rate": 7.575778894472362e-05,
-      "loss": 5.2214,
-      "step": 24629
-    },
-    {
-      "epoch": 12.844850065189048,
-      "grad_norm": 1.4846484661102295,
-      "learning_rate": 7.5756783919598e-05,
-      "loss": 5.3312,
-      "step": 24630
-    },
-    {
-      "epoch": 12.845371577574968,
-      "grad_norm": 1.3898589611053467,
-      "learning_rate": 7.575577889447237e-05,
-      "loss": 5.5731,
-      "step": 24631
-    },
-    {
-      "epoch": 12.845893089960887,
-      "grad_norm": 1.4773694276809692,
-      "learning_rate": 7.575477386934674e-05,
-      "loss": 5.3644,
-      "step": 24632
-    },
-    {
-      "epoch": 12.846414602346806,
-      "grad_norm": 1.3766512870788574,
-      "learning_rate": 7.575376884422111e-05,
-      "loss": 5.578,
-      "step": 24633
-    },
-    {
-      "epoch": 12.846936114732724,
-      "grad_norm": 1.4852056503295898,
-      "learning_rate": 7.575276381909549e-05,
-      "loss": 5.5608,
-      "step": 24634
-    },
-    {
-      "epoch": 12.847457627118644,
-      "grad_norm": 1.424828290939331,
-      "learning_rate": 7.575175879396985e-05,
-      "loss": 5.6975,
-      "step": 24635
-    },
-    {
-      "epoch": 12.847979139504563,
-      "grad_norm": 1.4773342609405518,
-      "learning_rate": 7.575075376884423e-05,
-      "loss": 5.0337,
-      "step": 24636
-    },
-    {
-      "epoch": 12.848500651890483,
-      "grad_norm": 1.4607136249542236,
-      "learning_rate": 7.574974874371859e-05,
-      "loss": 4.7321,
-      "step": 24637
-    },
-    {
-      "epoch": 12.849022164276402,
-      "grad_norm": 1.364302158355713,
-      "learning_rate": 7.574874371859297e-05,
-      "loss": 5.6721,
-      "step": 24638
-    },
-    {
-      "epoch": 12.849543676662321,
-      "grad_norm": 1.3853726387023926,
-      "learning_rate": 7.574773869346733e-05,
-      "loss": 5.4704,
-      "step": 24639
-    },
-    {
-      "epoch": 12.85006518904824,
-      "grad_norm": 1.3563792705535889,
-      "learning_rate": 7.574673366834171e-05,
-      "loss": 5.7899,
-      "step": 24640
-    },
-    {
-      "epoch": 12.850586701434159,
-      "grad_norm": 1.3884869813919067,
-      "learning_rate": 7.574572864321609e-05,
-      "loss": 5.4705,
-      "step": 24641
-    },
-    {
-      "epoch": 12.851108213820078,
-      "grad_norm": 1.3340996503829956,
-      "learning_rate": 7.574472361809045e-05,
-      "loss": 6.0538,
-      "step": 24642
-    },
-    {
-      "epoch": 12.851629726205998,
-      "grad_norm": 1.4597491025924683,
-      "learning_rate": 7.574371859296483e-05,
-      "loss": 5.3891,
-      "step": 24643
-    },
-    {
-      "epoch": 12.852151238591917,
-      "grad_norm": 1.4486027956008911,
-      "learning_rate": 7.57427135678392e-05,
-      "loss": 5.3116,
-      "step": 24644
-    },
-    {
-      "epoch": 12.852672750977836,
-      "grad_norm": 1.4889715909957886,
-      "learning_rate": 7.574170854271357e-05,
-      "loss": 5.6215,
-      "step": 24645
-    },
-    {
-      "epoch": 12.853194263363754,
-      "grad_norm": 1.5381659269332886,
-      "learning_rate": 7.574070351758794e-05,
-      "loss": 5.0078,
-      "step": 24646
-    },
-    {
-      "epoch": 12.853715775749674,
-      "grad_norm": 1.3356539011001587,
-      "learning_rate": 7.573969849246232e-05,
-      "loss": 5.4502,
-      "step": 24647
-    },
-    {
-      "epoch": 12.854237288135593,
-      "grad_norm": 1.4161639213562012,
-      "learning_rate": 7.573869346733668e-05,
-      "loss": 5.5974,
-      "step": 24648
-    },
-    {
-      "epoch": 12.854758800521513,
-      "grad_norm": 1.4979901313781738,
-      "learning_rate": 7.573768844221106e-05,
-      "loss": 5.2938,
-      "step": 24649
-    },
-    {
-      "epoch": 12.855280312907432,
-      "grad_norm": 1.42640221118927,
-      "learning_rate": 7.573668341708544e-05,
-      "loss": 5.3252,
-      "step": 24650
-    },
-    {
-      "epoch": 12.855801825293351,
-      "grad_norm": 1.4360921382904053,
-      "learning_rate": 7.573567839195981e-05,
-      "loss": 5.4554,
-      "step": 24651
-    },
-    {
-      "epoch": 12.85632333767927,
-      "grad_norm": 1.4213193655014038,
-      "learning_rate": 7.573467336683418e-05,
-      "loss": 5.3807,
-      "step": 24652
-    },
-    {
-      "epoch": 12.856844850065189,
-      "grad_norm": 1.3851252794265747,
-      "learning_rate": 7.573366834170854e-05,
-      "loss": 5.2522,
-      "step": 24653
-    },
-    {
-      "epoch": 12.857366362451108,
-      "grad_norm": 1.6763086318969727,
-      "learning_rate": 7.573266331658292e-05,
-      "loss": 5.4223,
-      "step": 24654
-    },
-    {
-      "epoch": 12.857887874837028,
-      "grad_norm": 1.446797251701355,
-      "learning_rate": 7.573165829145729e-05,
-      "loss": 5.49,
-      "step": 24655
-    },
-    {
-      "epoch": 12.858409387222947,
-      "grad_norm": 1.3818776607513428,
-      "learning_rate": 7.573065326633166e-05,
-      "loss": 5.6658,
-      "step": 24656
-    },
-    {
-      "epoch": 12.858930899608866,
-      "grad_norm": 1.310022234916687,
-      "learning_rate": 7.572964824120603e-05,
-      "loss": 5.9202,
-      "step": 24657
-    },
-    {
-      "epoch": 12.859452411994784,
-      "grad_norm": 1.4335439205169678,
-      "learning_rate": 7.57286432160804e-05,
-      "loss": 5.6191,
-      "step": 24658
-    },
-    {
-      "epoch": 12.859973924380704,
-      "grad_norm": 1.3717665672302246,
-      "learning_rate": 7.572763819095477e-05,
-      "loss": 5.8445,
-      "step": 24659
-    },
-    {
-      "epoch": 12.860495436766623,
-      "grad_norm": 1.4856759309768677,
-      "learning_rate": 7.572663316582915e-05,
-      "loss": 5.3872,
-      "step": 24660
-    },
-    {
-      "epoch": 12.861016949152543,
-      "grad_norm": 1.4241504669189453,
-      "learning_rate": 7.572562814070352e-05,
-      "loss": 5.7959,
-      "step": 24661
-    },
-    {
-      "epoch": 12.861538461538462,
-      "grad_norm": 1.4128996133804321,
-      "learning_rate": 7.57246231155779e-05,
-      "loss": 4.5976,
-      "step": 24662
-    },
-    {
-      "epoch": 12.862059973924381,
-      "grad_norm": 1.416710615158081,
-      "learning_rate": 7.572361809045227e-05,
-      "loss": 5.2748,
-      "step": 24663
-    },
-    {
-      "epoch": 12.8625814863103,
-      "grad_norm": 1.391680121421814,
-      "learning_rate": 7.572261306532664e-05,
-      "loss": 5.1431,
-      "step": 24664
-    },
-    {
-      "epoch": 12.863102998696219,
-      "grad_norm": 1.5697083473205566,
-      "learning_rate": 7.572160804020101e-05,
-      "loss": 5.0868,
-      "step": 24665
-    },
-    {
-      "epoch": 12.863624511082138,
-      "grad_norm": 1.4734899997711182,
-      "learning_rate": 7.572060301507537e-05,
-      "loss": 5.696,
-      "step": 24666
-    },
-    {
-      "epoch": 12.864146023468058,
-      "grad_norm": 1.3817274570465088,
-      "learning_rate": 7.571959798994975e-05,
-      "loss": 5.3746,
-      "step": 24667
-    },
-    {
-      "epoch": 12.864667535853977,
-      "grad_norm": 1.4497634172439575,
-      "learning_rate": 7.571859296482412e-05,
-      "loss": 5.4158,
-      "step": 24668
-    },
-    {
-      "epoch": 12.865189048239897,
-      "grad_norm": 1.335124135017395,
-      "learning_rate": 7.57175879396985e-05,
-      "loss": 5.2901,
-      "step": 24669
-    },
-    {
-      "epoch": 12.865710560625814,
-      "grad_norm": 1.4868993759155273,
-      "learning_rate": 7.571658291457287e-05,
-      "loss": 5.2843,
-      "step": 24670
-    },
-    {
-      "epoch": 12.866232073011734,
-      "grad_norm": 1.4996832609176636,
-      "learning_rate": 7.571557788944725e-05,
-      "loss": 5.1432,
-      "step": 24671
-    },
-    {
-      "epoch": 12.866753585397653,
-      "grad_norm": 1.3443577289581299,
-      "learning_rate": 7.571457286432161e-05,
-      "loss": 5.505,
-      "step": 24672
-    },
-    {
-      "epoch": 12.867275097783573,
-      "grad_norm": 1.4522989988327026,
-      "learning_rate": 7.571356783919599e-05,
-      "loss": 4.8426,
-      "step": 24673
-    },
-    {
-      "epoch": 12.867796610169492,
-      "grad_norm": 1.522265076637268,
-      "learning_rate": 7.571256281407036e-05,
-      "loss": 5.4489,
-      "step": 24674
-    },
-    {
-      "epoch": 12.868318122555412,
-      "grad_norm": 1.5269722938537598,
-      "learning_rate": 7.571155778894473e-05,
-      "loss": 5.2563,
-      "step": 24675
-    },
-    {
-      "epoch": 12.86883963494133,
-      "grad_norm": 1.4577420949935913,
-      "learning_rate": 7.57105527638191e-05,
-      "loss": 5.4515,
-      "step": 24676
-    },
-    {
-      "epoch": 12.869361147327249,
-      "grad_norm": 1.4942537546157837,
-      "learning_rate": 7.570954773869348e-05,
-      "loss": 5.1623,
-      "step": 24677
-    },
-    {
-      "epoch": 12.869882659713168,
-      "grad_norm": 1.4226210117340088,
-      "learning_rate": 7.570854271356784e-05,
-      "loss": 5.1736,
-      "step": 24678
-    },
-    {
-      "epoch": 12.870404172099088,
-      "grad_norm": 1.5292154550552368,
-      "learning_rate": 7.57075376884422e-05,
-      "loss": 5.6541,
-      "step": 24679
-    },
-    {
-      "epoch": 12.870925684485007,
-      "grad_norm": 1.4317790269851685,
-      "learning_rate": 7.570653266331658e-05,
-      "loss": 5.6224,
-      "step": 24680
-    },
-    {
-      "epoch": 12.871447196870925,
-      "grad_norm": 1.4248631000518799,
-      "learning_rate": 7.570552763819096e-05,
-      "loss": 5.3469,
-      "step": 24681
-    },
-    {
-      "epoch": 12.871968709256844,
-      "grad_norm": 1.353656530380249,
-      "learning_rate": 7.570452261306534e-05,
-      "loss": 5.6784,
-      "step": 24682
-    },
-    {
-      "epoch": 12.872490221642764,
-      "grad_norm": 1.4326950311660767,
-      "learning_rate": 7.57035175879397e-05,
-      "loss": 5.5582,
-      "step": 24683
-    },
-    {
-      "epoch": 12.873011734028683,
-      "grad_norm": 1.4663530588150024,
-      "learning_rate": 7.570251256281408e-05,
-      "loss": 4.844,
-      "step": 24684
-    },
-    {
-      "epoch": 12.873533246414603,
-      "grad_norm": 1.4593175649642944,
-      "learning_rate": 7.570150753768844e-05,
-      "loss": 4.8548,
-      "step": 24685
-    },
-    {
-      "epoch": 12.874054758800522,
-      "grad_norm": 1.5612571239471436,
-      "learning_rate": 7.570050251256282e-05,
-      "loss": 4.8014,
-      "step": 24686
-    },
-    {
-      "epoch": 12.874576271186442,
-      "grad_norm": 1.4868346452713013,
-      "learning_rate": 7.569949748743719e-05,
-      "loss": 5.2076,
-      "step": 24687
-    },
-    {
-      "epoch": 12.87509778357236,
-      "grad_norm": 1.3718276023864746,
-      "learning_rate": 7.569849246231156e-05,
-      "loss": 4.9069,
-      "step": 24688
-    },
-    {
-      "epoch": 12.875619295958279,
-      "grad_norm": 1.4213429689407349,
-      "learning_rate": 7.569748743718593e-05,
-      "loss": 4.9285,
-      "step": 24689
-    },
-    {
-      "epoch": 12.876140808344198,
-      "grad_norm": 1.3828840255737305,
-      "learning_rate": 7.56964824120603e-05,
-      "loss": 5.6519,
-      "step": 24690
-    },
-    {
-      "epoch": 12.876662320730118,
-      "grad_norm": 1.3778210878372192,
-      "learning_rate": 7.569547738693467e-05,
-      "loss": 5.5389,
-      "step": 24691
-    },
-    {
-      "epoch": 12.877183833116037,
-      "grad_norm": 1.522128701210022,
-      "learning_rate": 7.569447236180905e-05,
-      "loss": 5.2499,
-      "step": 24692
-    },
-    {
-      "epoch": 12.877705345501955,
-      "grad_norm": 1.3811395168304443,
-      "learning_rate": 7.569346733668343e-05,
-      "loss": 5.6035,
-      "step": 24693
-    },
-    {
-      "epoch": 12.878226857887874,
-      "grad_norm": 1.3600643873214722,
-      "learning_rate": 7.569246231155779e-05,
-      "loss": 5.8185,
-      "step": 24694
-    },
-    {
-      "epoch": 12.878748370273794,
-      "grad_norm": 1.3867524862289429,
-      "learning_rate": 7.569145728643217e-05,
-      "loss": 5.6079,
-      "step": 24695
-    },
-    {
-      "epoch": 12.879269882659713,
-      "grad_norm": 1.435459017753601,
-      "learning_rate": 7.569045226130653e-05,
-      "loss": 4.9936,
-      "step": 24696
-    },
-    {
-      "epoch": 12.879791395045633,
-      "grad_norm": 1.5101405382156372,
-      "learning_rate": 7.568944723618091e-05,
-      "loss": 5.5689,
-      "step": 24697
-    },
-    {
-      "epoch": 12.880312907431552,
-      "grad_norm": 1.4307211637496948,
-      "learning_rate": 7.568844221105527e-05,
-      "loss": 5.3757,
-      "step": 24698
-    },
-    {
-      "epoch": 12.880834419817472,
-      "grad_norm": 1.661227822303772,
-      "learning_rate": 7.568743718592965e-05,
-      "loss": 5.1283,
-      "step": 24699
-    },
-    {
-      "epoch": 12.88135593220339,
-      "grad_norm": 1.4922553300857544,
-      "learning_rate": 7.568643216080402e-05,
-      "loss": 5.1282,
-      "step": 24700
-    },
-    {
-      "epoch": 12.881877444589309,
-      "grad_norm": 1.5142545700073242,
-      "learning_rate": 7.56854271356784e-05,
-      "loss": 5.4349,
-      "step": 24701
-    },
-    {
-      "epoch": 12.882398956975228,
-      "grad_norm": 1.5512373447418213,
-      "learning_rate": 7.568442211055277e-05,
-      "loss": 5.2353,
-      "step": 24702
-    },
-    {
-      "epoch": 12.882920469361148,
-      "grad_norm": 1.57902991771698,
-      "learning_rate": 7.568341708542715e-05,
-      "loss": 5.4021,
-      "step": 24703
-    },
-    {
-      "epoch": 12.883441981747067,
-      "grad_norm": 1.8808839321136475,
-      "learning_rate": 7.568241206030151e-05,
-      "loss": 4.8933,
-      "step": 24704
-    },
-    {
-      "epoch": 12.883963494132985,
-      "grad_norm": 1.4376376867294312,
-      "learning_rate": 7.568140703517588e-05,
-      "loss": 5.8702,
-      "step": 24705
-    },
-    {
-      "epoch": 12.884485006518904,
-      "grad_norm": 1.512863039970398,
-      "learning_rate": 7.568040201005026e-05,
-      "loss": 5.3041,
-      "step": 24706
-    },
-    {
-      "epoch": 12.885006518904824,
-      "grad_norm": 1.391075611114502,
-      "learning_rate": 7.567939698492462e-05,
-      "loss": 5.6659,
-      "step": 24707
-    },
-    {
-      "epoch": 12.885528031290743,
-      "grad_norm": 1.4961916208267212,
-      "learning_rate": 7.5678391959799e-05,
-      "loss": 5.2566,
-      "step": 24708
-    },
-    {
-      "epoch": 12.886049543676663,
-      "grad_norm": 1.4994032382965088,
-      "learning_rate": 7.567738693467336e-05,
-      "loss": 5.5353,
-      "step": 24709
-    },
-    {
-      "epoch": 12.886571056062582,
-      "grad_norm": 1.4890272617340088,
-      "learning_rate": 7.567638190954774e-05,
-      "loss": 4.8463,
-      "step": 24710
-    },
-    {
-      "epoch": 12.887092568448502,
-      "grad_norm": 1.4404600858688354,
-      "learning_rate": 7.56753768844221e-05,
-      "loss": 5.1893,
-      "step": 24711
-    },
-    {
-      "epoch": 12.88761408083442,
-      "grad_norm": 1.529931902885437,
-      "learning_rate": 7.567437185929648e-05,
-      "loss": 5.4085,
-      "step": 24712
-    },
-    {
-      "epoch": 12.888135593220339,
-      "grad_norm": 1.4638240337371826,
-      "learning_rate": 7.567336683417086e-05,
-      "loss": 5.5004,
-      "step": 24713
-    },
-    {
-      "epoch": 12.888657105606258,
-      "grad_norm": 1.5970109701156616,
-      "learning_rate": 7.567236180904524e-05,
-      "loss": 5.5006,
-      "step": 24714
-    },
-    {
-      "epoch": 12.889178617992178,
-      "grad_norm": 1.4922162294387817,
-      "learning_rate": 7.56713567839196e-05,
-      "loss": 5.0644,
-      "step": 24715
-    },
-    {
-      "epoch": 12.889700130378097,
-      "grad_norm": 1.5717778205871582,
-      "learning_rate": 7.567035175879398e-05,
-      "loss": 5.2653,
-      "step": 24716
-    },
-    {
-      "epoch": 12.890221642764015,
-      "grad_norm": 1.7003695964813232,
-      "learning_rate": 7.566934673366834e-05,
-      "loss": 5.3433,
-      "step": 24717
-    },
-    {
-      "epoch": 12.890743155149934,
-      "grad_norm": 1.506618857383728,
-      "learning_rate": 7.566834170854272e-05,
-      "loss": 5.1638,
-      "step": 24718
-    },
-    {
-      "epoch": 12.891264667535854,
-      "grad_norm": 1.5073165893554688,
-      "learning_rate": 7.566733668341709e-05,
-      "loss": 5.5457,
-      "step": 24719
-    },
-    {
-      "epoch": 12.891786179921773,
-      "grad_norm": 1.3132332563400269,
-      "learning_rate": 7.566633165829145e-05,
-      "loss": 5.9267,
-      "step": 24720
-    },
-    {
-      "epoch": 12.892307692307693,
-      "grad_norm": 1.4223763942718506,
-      "learning_rate": 7.566532663316583e-05,
-      "loss": 5.6635,
-      "step": 24721
-    },
-    {
-      "epoch": 12.892829204693612,
-      "grad_norm": 1.4862831830978394,
-      "learning_rate": 7.56643216080402e-05,
-      "loss": 5.3067,
-      "step": 24722
-    },
-    {
-      "epoch": 12.89335071707953,
-      "grad_norm": 1.5342066287994385,
-      "learning_rate": 7.566331658291458e-05,
-      "loss": 4.9901,
-      "step": 24723
-    },
-    {
-      "epoch": 12.89387222946545,
-      "grad_norm": 1.773094654083252,
-      "learning_rate": 7.566231155778895e-05,
-      "loss": 5.005,
-      "step": 24724
-    },
-    {
-      "epoch": 12.894393741851369,
-      "grad_norm": 1.6567087173461914,
-      "learning_rate": 7.566130653266333e-05,
-      "loss": 5.1104,
-      "step": 24725
-    },
-    {
-      "epoch": 12.894915254237288,
-      "grad_norm": 1.5628445148468018,
-      "learning_rate": 7.566030150753769e-05,
-      "loss": 5.7579,
-      "step": 24726
-    },
-    {
-      "epoch": 12.895436766623208,
-      "grad_norm": 1.4400020837783813,
-      "learning_rate": 7.565929648241207e-05,
-      "loss": 5.4541,
-      "step": 24727
-    },
-    {
-      "epoch": 12.895958279009127,
-      "grad_norm": 1.5084190368652344,
-      "learning_rate": 7.565829145728643e-05,
-      "loss": 5.486,
-      "step": 24728
-    },
-    {
-      "epoch": 12.896479791395045,
-      "grad_norm": 1.5837458372116089,
-      "learning_rate": 7.565728643216081e-05,
-      "loss": 5.395,
-      "step": 24729
-    },
-    {
-      "epoch": 12.897001303780964,
-      "grad_norm": 1.6655513048171997,
-      "learning_rate": 7.565628140703517e-05,
-      "loss": 5.7062,
-      "step": 24730
-    },
-    {
-      "epoch": 12.897522816166884,
-      "grad_norm": 1.4584990739822388,
-      "learning_rate": 7.565527638190955e-05,
-      "loss": 5.7195,
-      "step": 24731
-    },
-    {
-      "epoch": 12.898044328552803,
-      "grad_norm": 1.4373011589050293,
-      "learning_rate": 7.565427135678392e-05,
-      "loss": 5.6163,
-      "step": 24732
-    },
-    {
-      "epoch": 12.898565840938723,
-      "grad_norm": 1.4039498567581177,
-      "learning_rate": 7.56532663316583e-05,
-      "loss": 5.2665,
-      "step": 24733
-    },
-    {
-      "epoch": 12.899087353324642,
-      "grad_norm": 1.4629631042480469,
-      "learning_rate": 7.565226130653267e-05,
-      "loss": 5.4701,
-      "step": 24734
-    },
-    {
-      "epoch": 12.89960886571056,
-      "grad_norm": 1.483049750328064,
-      "learning_rate": 7.565125628140704e-05,
-      "loss": 4.3813,
-      "step": 24735
-    },
-    {
-      "epoch": 12.90013037809648,
-      "grad_norm": 1.5531747341156006,
-      "learning_rate": 7.565025125628141e-05,
-      "loss": 5.398,
-      "step": 24736
-    },
-    {
-      "epoch": 12.900651890482399,
-      "grad_norm": 1.4685455560684204,
-      "learning_rate": 7.564924623115578e-05,
-      "loss": 5.5311,
-      "step": 24737
-    },
-    {
-      "epoch": 12.901173402868318,
-      "grad_norm": 1.425740122795105,
-      "learning_rate": 7.564824120603016e-05,
-      "loss": 5.3124,
-      "step": 24738
-    },
-    {
-      "epoch": 12.901694915254238,
-      "grad_norm": 1.323992133140564,
-      "learning_rate": 7.564723618090452e-05,
-      "loss": 5.7898,
-      "step": 24739
-    },
-    {
-      "epoch": 12.902216427640157,
-      "grad_norm": 1.5405912399291992,
-      "learning_rate": 7.56462311557789e-05,
-      "loss": 4.6799,
-      "step": 24740
-    },
-    {
-      "epoch": 12.902737940026075,
-      "grad_norm": 1.4073818922042847,
-      "learning_rate": 7.564522613065326e-05,
-      "loss": 5.7235,
-      "step": 24741
-    },
-    {
-      "epoch": 12.903259452411994,
-      "grad_norm": 1.5196220874786377,
-      "learning_rate": 7.564422110552764e-05,
-      "loss": 5.2907,
-      "step": 24742
-    },
-    {
-      "epoch": 12.903780964797914,
-      "grad_norm": 1.4104396104812622,
-      "learning_rate": 7.564321608040202e-05,
-      "loss": 5.2988,
-      "step": 24743
-    },
-    {
-      "epoch": 12.904302477183833,
-      "grad_norm": 1.4315919876098633,
-      "learning_rate": 7.56422110552764e-05,
-      "loss": 5.8048,
-      "step": 24744
-    },
-    {
-      "epoch": 12.904823989569753,
-      "grad_norm": 1.4866909980773926,
-      "learning_rate": 7.564120603015076e-05,
-      "loss": 5.6088,
-      "step": 24745
-    },
-    {
-      "epoch": 12.905345501955672,
-      "grad_norm": 1.43842613697052,
-      "learning_rate": 7.564020100502513e-05,
-      "loss": 5.254,
-      "step": 24746
-    },
-    {
-      "epoch": 12.90586701434159,
-      "grad_norm": 1.4491745233535767,
-      "learning_rate": 7.56391959798995e-05,
-      "loss": 5.6495,
-      "step": 24747
-    },
-    {
-      "epoch": 12.90638852672751,
-      "grad_norm": 1.3679744005203247,
-      "learning_rate": 7.563819095477387e-05,
-      "loss": 5.6551,
-      "step": 24748
-    },
-    {
-      "epoch": 12.906910039113429,
-      "grad_norm": 1.440392255783081,
-      "learning_rate": 7.563718592964825e-05,
-      "loss": 5.8958,
-      "step": 24749
-    },
-    {
-      "epoch": 12.907431551499348,
-      "grad_norm": 1.4384727478027344,
-      "learning_rate": 7.563618090452261e-05,
-      "loss": 5.4749,
-      "step": 24750
-    },
-    {
-      "epoch": 12.907953063885268,
-      "grad_norm": 1.4295719861984253,
-      "learning_rate": 7.563517587939699e-05,
-      "loss": 5.2561,
-      "step": 24751
-    },
-    {
-      "epoch": 12.908474576271187,
-      "grad_norm": 1.5152682065963745,
-      "learning_rate": 7.563417085427135e-05,
-      "loss": 4.9466,
-      "step": 24752
-    },
-    {
-      "epoch": 12.908996088657105,
-      "grad_norm": 1.4244900941848755,
-      "learning_rate": 7.563316582914573e-05,
-      "loss": 5.4927,
-      "step": 24753
-    },
-    {
-      "epoch": 12.909517601043024,
-      "grad_norm": 1.4309799671173096,
-      "learning_rate": 7.563216080402011e-05,
-      "loss": 5.4996,
-      "step": 24754
-    },
-    {
-      "epoch": 12.910039113428944,
-      "grad_norm": 1.5996869802474976,
-      "learning_rate": 7.563115577889449e-05,
-      "loss": 5.1045,
-      "step": 24755
-    },
-    {
-      "epoch": 12.910560625814863,
-      "grad_norm": 1.6982598304748535,
-      "learning_rate": 7.563015075376885e-05,
-      "loss": 5.3132,
-      "step": 24756
-    },
-    {
-      "epoch": 12.911082138200783,
-      "grad_norm": 1.5868597030639648,
-      "learning_rate": 7.562914572864323e-05,
-      "loss": 5.1141,
-      "step": 24757
-    },
-    {
-      "epoch": 12.911603650586702,
-      "grad_norm": 1.4919596910476685,
-      "learning_rate": 7.562814070351759e-05,
-      "loss": 5.2685,
-      "step": 24758
-    },
-    {
-      "epoch": 12.91212516297262,
-      "grad_norm": 1.4294100999832153,
-      "learning_rate": 7.562713567839196e-05,
-      "loss": 5.2396,
-      "step": 24759
-    },
-    {
-      "epoch": 12.91264667535854,
-      "grad_norm": 1.4403411149978638,
-      "learning_rate": 7.562613065326633e-05,
-      "loss": 5.4755,
-      "step": 24760
-    },
-    {
-      "epoch": 12.913168187744459,
-      "grad_norm": 1.3791896104812622,
-      "learning_rate": 7.56251256281407e-05,
-      "loss": 5.1694,
-      "step": 24761
-    },
-    {
-      "epoch": 12.913689700130378,
-      "grad_norm": 1.4054217338562012,
-      "learning_rate": 7.562412060301508e-05,
-      "loss": 5.2736,
-      "step": 24762
-    },
-    {
-      "epoch": 12.914211212516298,
-      "grad_norm": 1.3709959983825684,
-      "learning_rate": 7.562311557788945e-05,
-      "loss": 5.7586,
-      "step": 24763
-    },
-    {
-      "epoch": 12.914732724902217,
-      "grad_norm": 1.4125761985778809,
-      "learning_rate": 7.562211055276383e-05,
-      "loss": 5.3814,
-      "step": 24764
-    },
-    {
-      "epoch": 12.915254237288135,
-      "grad_norm": 1.5272672176361084,
-      "learning_rate": 7.56211055276382e-05,
-      "loss": 5.2205,
-      "step": 24765
-    },
-    {
-      "epoch": 12.915775749674054,
-      "grad_norm": 1.436463713645935,
-      "learning_rate": 7.562010050251257e-05,
-      "loss": 5.9344,
-      "step": 24766
-    },
-    {
-      "epoch": 12.916297262059974,
-      "grad_norm": 1.3474106788635254,
-      "learning_rate": 7.561909547738694e-05,
-      "loss": 5.451,
-      "step": 24767
-    },
-    {
-      "epoch": 12.916818774445893,
-      "grad_norm": 1.4660958051681519,
-      "learning_rate": 7.561809045226132e-05,
-      "loss": 5.5878,
-      "step": 24768
-    },
-    {
-      "epoch": 12.917340286831813,
-      "grad_norm": 1.4184856414794922,
-      "learning_rate": 7.561708542713568e-05,
-      "loss": 5.5349,
-      "step": 24769
-    },
-    {
-      "epoch": 12.917861799217732,
-      "grad_norm": 1.5144238471984863,
-      "learning_rate": 7.561608040201006e-05,
-      "loss": 5.4453,
-      "step": 24770
-    },
-    {
-      "epoch": 12.91838331160365,
-      "grad_norm": 1.5152181386947632,
-      "learning_rate": 7.561507537688442e-05,
-      "loss": 5.1079,
-      "step": 24771
-    },
-    {
-      "epoch": 12.91890482398957,
-      "grad_norm": 1.4755703210830688,
-      "learning_rate": 7.561407035175879e-05,
-      "loss": 5.3345,
-      "step": 24772
-    },
-    {
-      "epoch": 12.919426336375489,
-      "grad_norm": 1.494856357574463,
-      "learning_rate": 7.561306532663316e-05,
-      "loss": 4.9492,
-      "step": 24773
-    },
-    {
-      "epoch": 12.919947848761408,
-      "grad_norm": 1.443739652633667,
-      "learning_rate": 7.561206030150754e-05,
-      "loss": 5.3487,
-      "step": 24774
-    },
-    {
-      "epoch": 12.920469361147328,
-      "grad_norm": 1.5493388175964355,
-      "learning_rate": 7.561105527638192e-05,
-      "loss": 5.1933,
-      "step": 24775
-    },
-    {
-      "epoch": 12.920990873533245,
-      "grad_norm": 1.4273240566253662,
-      "learning_rate": 7.561005025125628e-05,
-      "loss": 5.8165,
-      "step": 24776
-    },
-    {
-      "epoch": 12.921512385919165,
-      "grad_norm": 1.3853821754455566,
-      "learning_rate": 7.560904522613066e-05,
-      "loss": 5.6322,
-      "step": 24777
-    },
-    {
-      "epoch": 12.922033898305084,
-      "grad_norm": 1.4013075828552246,
-      "learning_rate": 7.560804020100503e-05,
-      "loss": 5.8326,
-      "step": 24778
-    },
-    {
-      "epoch": 12.922555410691004,
-      "grad_norm": 1.5085194110870361,
-      "learning_rate": 7.56070351758794e-05,
-      "loss": 5.5342,
-      "step": 24779
-    },
-    {
-      "epoch": 12.923076923076923,
-      "grad_norm": 1.543379306793213,
-      "learning_rate": 7.560603015075377e-05,
-      "loss": 4.7945,
-      "step": 24780
-    },
-    {
-      "epoch": 12.923598435462843,
-      "grad_norm": 1.560597538948059,
-      "learning_rate": 7.560502512562815e-05,
-      "loss": 5.5286,
-      "step": 24781
-    },
-    {
-      "epoch": 12.924119947848762,
-      "grad_norm": 1.4275848865509033,
-      "learning_rate": 7.560402010050251e-05,
-      "loss": 5.5153,
-      "step": 24782
-    },
-    {
-      "epoch": 12.92464146023468,
-      "grad_norm": 1.4730587005615234,
-      "learning_rate": 7.560301507537689e-05,
-      "loss": 4.9346,
-      "step": 24783
-    },
-    {
-      "epoch": 12.9251629726206,
-      "grad_norm": 1.3778982162475586,
-      "learning_rate": 7.560201005025127e-05,
-      "loss": 5.6533,
-      "step": 24784
-    },
-    {
-      "epoch": 12.925684485006519,
-      "grad_norm": 1.561908483505249,
-      "learning_rate": 7.560100502512563e-05,
-      "loss": 5.1719,
-      "step": 24785
-    },
-    {
-      "epoch": 12.926205997392438,
-      "grad_norm": 1.3793351650238037,
-      "learning_rate": 7.560000000000001e-05,
-      "loss": 5.5108,
-      "step": 24786
-    },
-    {
-      "epoch": 12.926727509778358,
-      "grad_norm": 1.4578075408935547,
-      "learning_rate": 7.559899497487437e-05,
-      "loss": 5.1803,
-      "step": 24787
-    },
-    {
-      "epoch": 12.927249022164276,
-      "grad_norm": 1.4672389030456543,
-      "learning_rate": 7.559798994974875e-05,
-      "loss": 5.1678,
-      "step": 24788
-    },
-    {
-      "epoch": 12.927770534550195,
-      "grad_norm": 1.4986213445663452,
-      "learning_rate": 7.559698492462311e-05,
-      "loss": 5.1918,
-      "step": 24789
-    },
-    {
-      "epoch": 12.928292046936114,
-      "grad_norm": 1.4703031778335571,
-      "learning_rate": 7.559597989949749e-05,
-      "loss": 5.1953,
-      "step": 24790
-    },
-    {
-      "epoch": 12.928813559322034,
-      "grad_norm": 1.4692702293395996,
-      "learning_rate": 7.559497487437186e-05,
-      "loss": 5.4214,
-      "step": 24791
-    },
-    {
-      "epoch": 12.929335071707953,
-      "grad_norm": 1.6389929056167603,
-      "learning_rate": 7.559396984924623e-05,
-      "loss": 5.1505,
-      "step": 24792
-    },
-    {
-      "epoch": 12.929856584093873,
-      "grad_norm": 1.4495924711227417,
-      "learning_rate": 7.55929648241206e-05,
-      "loss": 4.9941,
-      "step": 24793
-    },
-    {
-      "epoch": 12.930378096479792,
-      "grad_norm": 1.4150946140289307,
-      "learning_rate": 7.559195979899498e-05,
-      "loss": 5.5168,
-      "step": 24794
-    },
-    {
-      "epoch": 12.93089960886571,
-      "grad_norm": 1.396484613418579,
-      "learning_rate": 7.559095477386935e-05,
-      "loss": 5.7857,
-      "step": 24795
-    },
-    {
-      "epoch": 12.93142112125163,
-      "grad_norm": 1.595192790031433,
-      "learning_rate": 7.558994974874373e-05,
-      "loss": 4.2998,
-      "step": 24796
-    },
-    {
-      "epoch": 12.931942633637549,
-      "grad_norm": 1.5848087072372437,
-      "learning_rate": 7.55889447236181e-05,
-      "loss": 5.4786,
-      "step": 24797
-    },
-    {
-      "epoch": 12.932464146023468,
-      "grad_norm": 1.5089913606643677,
-      "learning_rate": 7.558793969849247e-05,
-      "loss": 5.5465,
-      "step": 24798
-    },
-    {
-      "epoch": 12.932985658409388,
-      "grad_norm": 1.4019140005111694,
-      "learning_rate": 7.558693467336684e-05,
-      "loss": 4.6754,
-      "step": 24799
-    },
-    {
-      "epoch": 12.933507170795306,
-      "grad_norm": 1.4451634883880615,
-      "learning_rate": 7.55859296482412e-05,
-      "loss": 5.5873,
-      "step": 24800
-    },
-    {
-      "epoch": 12.934028683181225,
-      "grad_norm": 1.4193488359451294,
-      "learning_rate": 7.558492462311558e-05,
-      "loss": 5.5498,
-      "step": 24801
-    },
-    {
-      "epoch": 12.934550195567144,
-      "grad_norm": 1.4507259130477905,
-      "learning_rate": 7.558391959798994e-05,
-      "loss": 5.6571,
-      "step": 24802
-    },
-    {
-      "epoch": 12.935071707953064,
-      "grad_norm": 1.3644264936447144,
-      "learning_rate": 7.558291457286432e-05,
-      "loss": 5.0865,
-      "step": 24803
-    },
-    {
-      "epoch": 12.935593220338983,
-      "grad_norm": 1.4585180282592773,
-      "learning_rate": 7.55819095477387e-05,
-      "loss": 5.5353,
-      "step": 24804
-    },
-    {
-      "epoch": 12.936114732724903,
-      "grad_norm": 1.4994611740112305,
-      "learning_rate": 7.558090452261308e-05,
-      "loss": 5.5891,
-      "step": 24805
-    },
-    {
-      "epoch": 12.93663624511082,
-      "grad_norm": 1.4000085592269897,
-      "learning_rate": 7.557989949748744e-05,
-      "loss": 5.7094,
-      "step": 24806
-    },
-    {
-      "epoch": 12.93715775749674,
-      "grad_norm": 1.3854570388793945,
-      "learning_rate": 7.557889447236182e-05,
-      "loss": 5.9452,
-      "step": 24807
-    },
-    {
-      "epoch": 12.93767926988266,
-      "grad_norm": 1.5004512071609497,
-      "learning_rate": 7.557788944723618e-05,
-      "loss": 5.5201,
-      "step": 24808
-    },
-    {
-      "epoch": 12.938200782268579,
-      "grad_norm": 1.4625142812728882,
-      "learning_rate": 7.557688442211056e-05,
-      "loss": 5.4512,
-      "step": 24809
-    },
-    {
-      "epoch": 12.938722294654498,
-      "grad_norm": 1.396570086479187,
-      "learning_rate": 7.557587939698493e-05,
-      "loss": 5.7577,
-      "step": 24810
-    },
-    {
-      "epoch": 12.939243807040418,
-      "grad_norm": 1.462796926498413,
-      "learning_rate": 7.55748743718593e-05,
-      "loss": 5.7861,
-      "step": 24811
-    },
-    {
-      "epoch": 12.939765319426336,
-      "grad_norm": 1.419164776802063,
-      "learning_rate": 7.557386934673367e-05,
-      "loss": 5.3733,
-      "step": 24812
-    },
-    {
-      "epoch": 12.940286831812255,
-      "grad_norm": 1.421269178390503,
-      "learning_rate": 7.557286432160803e-05,
-      "loss": 5.4047,
-      "step": 24813
-    },
-    {
-      "epoch": 12.940808344198174,
-      "grad_norm": 1.40413498878479,
-      "learning_rate": 7.557185929648241e-05,
-      "loss": 5.6773,
-      "step": 24814
-    },
-    {
-      "epoch": 12.941329856584094,
-      "grad_norm": 1.4439491033554077,
-      "learning_rate": 7.557085427135679e-05,
-      "loss": 5.4638,
-      "step": 24815
-    },
-    {
-      "epoch": 12.941851368970013,
-      "grad_norm": 1.4479589462280273,
-      "learning_rate": 7.556984924623117e-05,
-      "loss": 5.131,
-      "step": 24816
-    },
-    {
-      "epoch": 12.942372881355933,
-      "grad_norm": 1.4161211252212524,
-      "learning_rate": 7.556884422110553e-05,
-      "loss": 5.3297,
-      "step": 24817
-    },
-    {
-      "epoch": 12.94289439374185,
-      "grad_norm": 1.430370569229126,
-      "learning_rate": 7.556783919597991e-05,
-      "loss": 5.3817,
-      "step": 24818
-    },
-    {
-      "epoch": 12.94341590612777,
-      "grad_norm": 1.4038915634155273,
-      "learning_rate": 7.556683417085427e-05,
-      "loss": 5.5971,
-      "step": 24819
-    },
-    {
-      "epoch": 12.94393741851369,
-      "grad_norm": 1.4134514331817627,
-      "learning_rate": 7.556582914572865e-05,
-      "loss": 5.5991,
-      "step": 24820
-    },
-    {
-      "epoch": 12.944458930899609,
-      "grad_norm": 1.538896083831787,
-      "learning_rate": 7.556482412060302e-05,
-      "loss": 5.156,
-      "step": 24821
-    },
-    {
-      "epoch": 12.944980443285528,
-      "grad_norm": 1.4824374914169312,
-      "learning_rate": 7.556381909547739e-05,
-      "loss": 5.4251,
-      "step": 24822
-    },
-    {
-      "epoch": 12.945501955671448,
-      "grad_norm": 1.5658413171768188,
-      "learning_rate": 7.556281407035176e-05,
-      "loss": 5.1457,
-      "step": 24823
-    },
-    {
-      "epoch": 12.946023468057366,
-      "grad_norm": 1.3248335123062134,
-      "learning_rate": 7.556180904522614e-05,
-      "loss": 5.7374,
-      "step": 24824
-    },
-    {
-      "epoch": 12.946544980443285,
-      "grad_norm": 1.4150969982147217,
-      "learning_rate": 7.556080402010051e-05,
-      "loss": 5.6304,
-      "step": 24825
-    },
-    {
-      "epoch": 12.947066492829205,
-      "grad_norm": 1.4679011106491089,
-      "learning_rate": 7.555979899497488e-05,
-      "loss": 5.5264,
-      "step": 24826
-    },
-    {
-      "epoch": 12.947588005215124,
-      "grad_norm": 1.4957494735717773,
-      "learning_rate": 7.555879396984926e-05,
-      "loss": 5.3678,
-      "step": 24827
-    },
-    {
-      "epoch": 12.948109517601043,
-      "grad_norm": 1.327399730682373,
-      "learning_rate": 7.555778894472362e-05,
-      "loss": 5.5366,
-      "step": 24828
-    },
-    {
-      "epoch": 12.948631029986963,
-      "grad_norm": 1.298120379447937,
-      "learning_rate": 7.5556783919598e-05,
-      "loss": 5.6457,
-      "step": 24829
-    },
-    {
-      "epoch": 12.94915254237288,
-      "grad_norm": 1.4371241331100464,
-      "learning_rate": 7.555577889447236e-05,
-      "loss": 5.1917,
-      "step": 24830
-    },
-    {
-      "epoch": 12.9496740547588,
-      "grad_norm": 1.4717812538146973,
-      "learning_rate": 7.555477386934674e-05,
-      "loss": 5.2117,
-      "step": 24831
-    },
-    {
-      "epoch": 12.95019556714472,
-      "grad_norm": 1.3676960468292236,
-      "learning_rate": 7.55537688442211e-05,
-      "loss": 5.6923,
-      "step": 24832
-    },
-    {
-      "epoch": 12.950717079530639,
-      "grad_norm": 1.4917563199996948,
-      "learning_rate": 7.555276381909548e-05,
-      "loss": 4.8156,
-      "step": 24833
-    },
-    {
-      "epoch": 12.951238591916558,
-      "grad_norm": 1.4077788591384888,
-      "learning_rate": 7.555175879396985e-05,
-      "loss": 5.8754,
-      "step": 24834
-    },
-    {
-      "epoch": 12.951760104302478,
-      "grad_norm": 1.412641167640686,
-      "learning_rate": 7.555075376884422e-05,
-      "loss": 5.2142,
-      "step": 24835
-    },
-    {
-      "epoch": 12.952281616688396,
-      "grad_norm": 1.527685284614563,
-      "learning_rate": 7.55497487437186e-05,
-      "loss": 5.4658,
-      "step": 24836
-    },
-    {
-      "epoch": 12.952803129074315,
-      "grad_norm": 1.5200613737106323,
-      "learning_rate": 7.554874371859298e-05,
-      "loss": 5.2716,
-      "step": 24837
-    },
-    {
-      "epoch": 12.953324641460235,
-      "grad_norm": 1.4599411487579346,
-      "learning_rate": 7.554773869346734e-05,
-      "loss": 5.4855,
-      "step": 24838
-    },
-    {
-      "epoch": 12.953846153846154,
-      "grad_norm": 1.5375727415084839,
-      "learning_rate": 7.554673366834171e-05,
-      "loss": 5.2622,
-      "step": 24839
-    },
-    {
-      "epoch": 12.954367666232073,
-      "grad_norm": 1.3947031497955322,
-      "learning_rate": 7.554572864321609e-05,
-      "loss": 5.5609,
-      "step": 24840
-    },
-    {
-      "epoch": 12.954889178617993,
-      "grad_norm": 1.4404702186584473,
-      "learning_rate": 7.554472361809045e-05,
-      "loss": 5.0811,
-      "step": 24841
-    },
-    {
-      "epoch": 12.95541069100391,
-      "grad_norm": 1.3956023454666138,
-      "learning_rate": 7.554371859296483e-05,
-      "loss": 5.4779,
-      "step": 24842
-    },
-    {
-      "epoch": 12.95593220338983,
-      "grad_norm": 1.4340088367462158,
-      "learning_rate": 7.554271356783919e-05,
-      "loss": 5.5091,
-      "step": 24843
-    },
-    {
-      "epoch": 12.95645371577575,
-      "grad_norm": 1.4932284355163574,
-      "learning_rate": 7.554170854271357e-05,
-      "loss": 5.069,
-      "step": 24844
-    },
-    {
-      "epoch": 12.956975228161669,
-      "grad_norm": 1.51760733127594,
-      "learning_rate": 7.554070351758795e-05,
-      "loss": 5.2169,
-      "step": 24845
-    },
-    {
-      "epoch": 12.957496740547588,
-      "grad_norm": 1.412537932395935,
-      "learning_rate": 7.553969849246233e-05,
-      "loss": 5.6892,
-      "step": 24846
-    },
-    {
-      "epoch": 12.958018252933508,
-      "grad_norm": 1.5469245910644531,
-      "learning_rate": 7.553869346733669e-05,
-      "loss": 5.5725,
-      "step": 24847
-    },
-    {
-      "epoch": 12.958539765319426,
-      "grad_norm": 1.3908629417419434,
-      "learning_rate": 7.553768844221107e-05,
-      "loss": 5.5032,
-      "step": 24848
-    },
-    {
-      "epoch": 12.959061277705345,
-      "grad_norm": 1.4897438287734985,
-      "learning_rate": 7.553668341708543e-05,
-      "loss": 5.4558,
-      "step": 24849
-    },
-    {
-      "epoch": 12.959582790091265,
-      "grad_norm": 1.4776113033294678,
-      "learning_rate": 7.553567839195981e-05,
-      "loss": 5.8042,
-      "step": 24850
-    },
-    {
-      "epoch": 12.960104302477184,
-      "grad_norm": 1.353235125541687,
-      "learning_rate": 7.553467336683417e-05,
-      "loss": 5.9104,
-      "step": 24851
-    },
-    {
-      "epoch": 12.960625814863103,
-      "grad_norm": 1.4854031801223755,
-      "learning_rate": 7.553366834170854e-05,
-      "loss": 5.196,
-      "step": 24852
-    },
-    {
-      "epoch": 12.961147327249023,
-      "grad_norm": 1.565555214881897,
-      "learning_rate": 7.553266331658292e-05,
-      "loss": 5.6112,
-      "step": 24853
-    },
-    {
-      "epoch": 12.96166883963494,
-      "grad_norm": 1.610148310661316,
-      "learning_rate": 7.553165829145728e-05,
-      "loss": 5.6234,
-      "step": 24854
-    },
-    {
-      "epoch": 12.96219035202086,
-      "grad_norm": 1.5746228694915771,
-      "learning_rate": 7.553065326633166e-05,
-      "loss": 5.4364,
-      "step": 24855
-    },
-    {
-      "epoch": 12.96271186440678,
-      "grad_norm": 1.3966357707977295,
-      "learning_rate": 7.552964824120604e-05,
-      "loss": 5.7366,
-      "step": 24856
-    },
-    {
-      "epoch": 12.963233376792699,
-      "grad_norm": 1.5200656652450562,
-      "learning_rate": 7.552864321608041e-05,
-      "loss": 5.2514,
-      "step": 24857
-    },
-    {
-      "epoch": 12.963754889178619,
-      "grad_norm": 1.4276994466781616,
-      "learning_rate": 7.552763819095478e-05,
-      "loss": 5.0177,
-      "step": 24858
-    },
-    {
-      "epoch": 12.964276401564538,
-      "grad_norm": 1.4425915479660034,
-      "learning_rate": 7.552663316582916e-05,
-      "loss": 5.4594,
-      "step": 24859
-    },
-    {
-      "epoch": 12.964797913950456,
-      "grad_norm": 1.4052938222885132,
-      "learning_rate": 7.552562814070352e-05,
-      "loss": 5.5855,
-      "step": 24860
-    },
-    {
-      "epoch": 12.965319426336375,
-      "grad_norm": 1.5571333169937134,
-      "learning_rate": 7.55246231155779e-05,
-      "loss": 5.5483,
-      "step": 24861
-    },
-    {
-      "epoch": 12.965840938722295,
-      "grad_norm": 1.4361435174942017,
-      "learning_rate": 7.552361809045226e-05,
-      "loss": 5.0715,
-      "step": 24862
-    },
-    {
-      "epoch": 12.966362451108214,
-      "grad_norm": 1.5516752004623413,
-      "learning_rate": 7.552261306532664e-05,
-      "loss": 5.3161,
-      "step": 24863
-    },
-    {
-      "epoch": 12.966883963494134,
-      "grad_norm": 1.5567405223846436,
-      "learning_rate": 7.5521608040201e-05,
-      "loss": 5.3579,
-      "step": 24864
-    },
-    {
-      "epoch": 12.967405475880053,
-      "grad_norm": 1.4082837104797363,
-      "learning_rate": 7.552060301507537e-05,
-      "loss": 5.1166,
-      "step": 24865
-    },
-    {
-      "epoch": 12.96792698826597,
-      "grad_norm": 1.446543574333191,
-      "learning_rate": 7.551959798994975e-05,
-      "loss": 5.8086,
-      "step": 24866
-    },
-    {
-      "epoch": 12.96844850065189,
-      "grad_norm": 1.4792859554290771,
-      "learning_rate": 7.551859296482412e-05,
-      "loss": 5.2138,
-      "step": 24867
-    },
-    {
-      "epoch": 12.96897001303781,
-      "grad_norm": 1.46998929977417,
-      "learning_rate": 7.55175879396985e-05,
-      "loss": 5.4319,
-      "step": 24868
-    },
-    {
-      "epoch": 12.969491525423729,
-      "grad_norm": 1.503684639930725,
-      "learning_rate": 7.551658291457287e-05,
-      "loss": 5.0115,
-      "step": 24869
-    },
-    {
-      "epoch": 12.970013037809649,
-      "grad_norm": 1.4610308408737183,
-      "learning_rate": 7.551557788944724e-05,
-      "loss": 5.0842,
-      "step": 24870
-    },
-    {
-      "epoch": 12.970534550195566,
-      "grad_norm": 1.397529125213623,
-      "learning_rate": 7.551457286432161e-05,
-      "loss": 5.224,
-      "step": 24871
-    },
-    {
-      "epoch": 12.971056062581486,
-      "grad_norm": 1.414838194847107,
-      "learning_rate": 7.551356783919599e-05,
-      "loss": 5.3745,
-      "step": 24872
-    },
-    {
-      "epoch": 12.971577574967405,
-      "grad_norm": 1.3712496757507324,
-      "learning_rate": 7.551256281407035e-05,
-      "loss": 5.6031,
-      "step": 24873
-    },
-    {
-      "epoch": 12.972099087353325,
-      "grad_norm": 1.615883469581604,
-      "learning_rate": 7.551155778894473e-05,
-      "loss": 5.0794,
-      "step": 24874
-    },
-    {
-      "epoch": 12.972620599739244,
-      "grad_norm": 1.4590216875076294,
-      "learning_rate": 7.551055276381909e-05,
-      "loss": 5.3046,
-      "step": 24875
-    },
-    {
-      "epoch": 12.973142112125164,
-      "grad_norm": 1.5282623767852783,
-      "learning_rate": 7.550954773869347e-05,
-      "loss": 5.1191,
-      "step": 24876
-    },
-    {
-      "epoch": 12.973663624511083,
-      "grad_norm": 1.4989322423934937,
-      "learning_rate": 7.550854271356785e-05,
-      "loss": 5.6628,
-      "step": 24877
-    },
-    {
-      "epoch": 12.974185136897,
-      "grad_norm": 1.4191858768463135,
-      "learning_rate": 7.550753768844221e-05,
-      "loss": 5.368,
-      "step": 24878
-    },
-    {
-      "epoch": 12.97470664928292,
-      "grad_norm": 1.4943689107894897,
-      "learning_rate": 7.550653266331659e-05,
-      "loss": 5.4281,
-      "step": 24879
-    },
-    {
-      "epoch": 12.97522816166884,
-      "grad_norm": 1.5637153387069702,
-      "learning_rate": 7.550552763819095e-05,
-      "loss": 5.2501,
-      "step": 24880
-    },
-    {
-      "epoch": 12.975749674054759,
-      "grad_norm": 2.25223445892334,
-      "learning_rate": 7.550452261306533e-05,
-      "loss": 4.6496,
-      "step": 24881
-    },
-    {
-      "epoch": 12.976271186440679,
-      "grad_norm": 1.3737996816635132,
-      "learning_rate": 7.55035175879397e-05,
-      "loss": 5.6518,
-      "step": 24882
-    },
-    {
-      "epoch": 12.976792698826596,
-      "grad_norm": 1.563730001449585,
-      "learning_rate": 7.550251256281407e-05,
-      "loss": 5.2636,
-      "step": 24883
-    },
-    {
-      "epoch": 12.977314211212516,
-      "grad_norm": 1.4592657089233398,
-      "learning_rate": 7.550150753768844e-05,
-      "loss": 5.4849,
-      "step": 24884
-    },
-    {
-      "epoch": 12.977835723598435,
-      "grad_norm": 1.6189231872558594,
-      "learning_rate": 7.550050251256282e-05,
-      "loss": 5.1686,
-      "step": 24885
-    },
-    {
-      "epoch": 12.978357235984355,
-      "grad_norm": 1.3698781728744507,
-      "learning_rate": 7.549949748743718e-05,
-      "loss": 5.6983,
-      "step": 24886
-    },
-    {
-      "epoch": 12.978878748370274,
-      "grad_norm": 1.4323832988739014,
-      "learning_rate": 7.549849246231156e-05,
-      "loss": 5.8107,
-      "step": 24887
-    },
-    {
-      "epoch": 12.979400260756194,
-      "grad_norm": 1.4170339107513428,
-      "learning_rate": 7.549748743718594e-05,
-      "loss": 5.6339,
-      "step": 24888
-    },
-    {
-      "epoch": 12.979921773142113,
-      "grad_norm": 1.4335356950759888,
-      "learning_rate": 7.549648241206031e-05,
-      "loss": 5.6271,
-      "step": 24889
-    },
-    {
-      "epoch": 12.98044328552803,
-      "grad_norm": 1.45816969871521,
-      "learning_rate": 7.549547738693468e-05,
-      "loss": 5.4166,
-      "step": 24890
-    },
-    {
-      "epoch": 12.98096479791395,
-      "grad_norm": 1.5053993463516235,
-      "learning_rate": 7.549447236180906e-05,
-      "loss": 5.6864,
-      "step": 24891
-    },
-    {
-      "epoch": 12.98148631029987,
-      "grad_norm": 1.3992886543273926,
-      "learning_rate": 7.549346733668342e-05,
-      "loss": 5.5012,
-      "step": 24892
-    },
-    {
-      "epoch": 12.98200782268579,
-      "grad_norm": 1.4984796047210693,
-      "learning_rate": 7.549246231155779e-05,
-      "loss": 5.3648,
-      "step": 24893
-    },
-    {
-      "epoch": 12.982529335071709,
-      "grad_norm": 1.51664137840271,
-      "learning_rate": 7.549145728643216e-05,
-      "loss": 4.8902,
-      "step": 24894
-    },
-    {
-      "epoch": 12.983050847457626,
-      "grad_norm": 1.4392110109329224,
-      "learning_rate": 7.549045226130653e-05,
-      "loss": 5.1816,
-      "step": 24895
-    },
-    {
-      "epoch": 12.983572359843546,
-      "grad_norm": 1.5878565311431885,
-      "learning_rate": 7.54894472361809e-05,
-      "loss": 5.4343,
-      "step": 24896
-    },
-    {
-      "epoch": 12.984093872229465,
-      "grad_norm": 1.399032473564148,
-      "learning_rate": 7.548844221105528e-05,
-      "loss": 5.9353,
-      "step": 24897
-    },
-    {
-      "epoch": 12.984615384615385,
-      "grad_norm": 1.5939244031906128,
-      "learning_rate": 7.548743718592966e-05,
-      "loss": 5.5592,
-      "step": 24898
-    },
-    {
-      "epoch": 12.985136897001304,
-      "grad_norm": 1.4364012479782104,
-      "learning_rate": 7.548643216080402e-05,
-      "loss": 5.9514,
-      "step": 24899
-    },
-    {
-      "epoch": 12.985658409387224,
-      "grad_norm": 1.401566982269287,
-      "learning_rate": 7.54854271356784e-05,
-      "loss": 5.8318,
-      "step": 24900
-    },
-    {
-      "epoch": 12.986179921773141,
-      "grad_norm": 1.4025949239730835,
-      "learning_rate": 7.548442211055277e-05,
-      "loss": 5.6004,
-      "step": 24901
-    },
-    {
-      "epoch": 12.98670143415906,
-      "grad_norm": 1.418747067451477,
-      "learning_rate": 7.548341708542714e-05,
-      "loss": 5.6355,
-      "step": 24902
-    },
-    {
-      "epoch": 12.98722294654498,
-      "grad_norm": 1.360119342803955,
-      "learning_rate": 7.548241206030151e-05,
-      "loss": 5.7012,
-      "step": 24903
-    },
-    {
-      "epoch": 12.9877444589309,
-      "grad_norm": 1.3518282175064087,
-      "learning_rate": 7.548140703517589e-05,
-      "loss": 5.5493,
-      "step": 24904
-    },
-    {
-      "epoch": 12.98826597131682,
-      "grad_norm": 1.4630800485610962,
-      "learning_rate": 7.548040201005025e-05,
-      "loss": 5.07,
-      "step": 24905
-    },
-    {
-      "epoch": 12.988787483702739,
-      "grad_norm": 1.3679165840148926,
-      "learning_rate": 7.547939698492462e-05,
-      "loss": 4.8339,
-      "step": 24906
-    },
-    {
-      "epoch": 12.989308996088656,
-      "grad_norm": 1.4355125427246094,
-      "learning_rate": 7.5478391959799e-05,
-      "loss": 5.1333,
-      "step": 24907
-    },
-    {
-      "epoch": 12.989830508474576,
-      "grad_norm": 1.4147162437438965,
-      "learning_rate": 7.547738693467337e-05,
-      "loss": 5.5698,
-      "step": 24908
-    },
-    {
-      "epoch": 12.990352020860495,
-      "grad_norm": 1.4937529563903809,
-      "learning_rate": 7.547638190954775e-05,
-      "loss": 5.1795,
-      "step": 24909
-    },
-    {
-      "epoch": 12.990873533246415,
-      "grad_norm": 1.434704065322876,
-      "learning_rate": 7.547537688442211e-05,
-      "loss": 5.3368,
-      "step": 24910
-    },
-    {
-      "epoch": 12.991395045632334,
-      "grad_norm": 1.3380588293075562,
-      "learning_rate": 7.547437185929649e-05,
-      "loss": 4.5438,
-      "step": 24911
-    },
-    {
-      "epoch": 12.991916558018254,
-      "grad_norm": 1.4174107313156128,
-      "learning_rate": 7.547336683417086e-05,
-      "loss": 5.2401,
-      "step": 24912
-    },
-    {
-      "epoch": 12.992438070404171,
-      "grad_norm": 1.4838937520980835,
-      "learning_rate": 7.547236180904523e-05,
-      "loss": 5.7458,
-      "step": 24913
-    },
-    {
-      "epoch": 12.99295958279009,
-      "grad_norm": 1.460187315940857,
-      "learning_rate": 7.54713567839196e-05,
-      "loss": 5.571,
-      "step": 24914
-    },
-    {
-      "epoch": 12.99348109517601,
-      "grad_norm": 1.4413350820541382,
-      "learning_rate": 7.547035175879398e-05,
-      "loss": 5.2312,
-      "step": 24915
-    },
-    {
-      "epoch": 12.99400260756193,
-      "grad_norm": 1.404197335243225,
-      "learning_rate": 7.546934673366834e-05,
-      "loss": 4.8674,
-      "step": 24916
-    },
-    {
-      "epoch": 12.99452411994785,
-      "grad_norm": 1.5999112129211426,
-      "learning_rate": 7.546834170854272e-05,
-      "loss": 4.9212,
-      "step": 24917
-    },
-    {
-      "epoch": 12.995045632333769,
-      "grad_norm": 1.9199494123458862,
-      "learning_rate": 7.54673366834171e-05,
-      "loss": 5.1977,
-      "step": 24918
-    },
-    {
-      "epoch": 12.995567144719686,
-      "grad_norm": 1.4525030851364136,
-      "learning_rate": 7.546633165829146e-05,
-      "loss": 5.0752,
-      "step": 24919
-    },
-    {
-      "epoch": 12.996088657105606,
-      "grad_norm": 1.5402097702026367,
-      "learning_rate": 7.546532663316584e-05,
-      "loss": 5.4852,
-      "step": 24920
-    },
-    {
-      "epoch": 12.996610169491525,
-      "grad_norm": 1.5111380815505981,
-      "learning_rate": 7.54643216080402e-05,
-      "loss": 4.8002,
-      "step": 24921
-    },
-    {
-      "epoch": 12.997131681877445,
-      "grad_norm": 1.5582941770553589,
-      "learning_rate": 7.546331658291458e-05,
-      "loss": 5.3745,
-      "step": 24922
-    },
-    {
-      "epoch": 12.997653194263364,
-      "grad_norm": 1.5431370735168457,
-      "learning_rate": 7.546231155778894e-05,
-      "loss": 5.4898,
-      "step": 24923
-    },
-    {
-      "epoch": 12.998174706649284,
-      "grad_norm": 1.4679903984069824,
-      "learning_rate": 7.546130653266332e-05,
-      "loss": 4.9059,
-      "step": 24924
-    },
-    {
-      "epoch": 12.998696219035201,
-      "grad_norm": 1.4424365758895874,
-      "learning_rate": 7.546030150753769e-05,
-      "loss": 5.8304,
-      "step": 24925
-    },
-    {
-      "epoch": 12.99921773142112,
-      "grad_norm": 1.449933409690857,
-      "learning_rate": 7.545929648241206e-05,
-      "loss": 5.4914,
-      "step": 24926
-    },
-    {
-      "epoch": 12.99973924380704,
-      "grad_norm": 1.4250776767730713,
-      "learning_rate": 7.545829145728643e-05,
-      "loss": 5.6755,
-      "step": 24927
-    },
-    {
-      "epoch": 13.00026075619296,
-      "grad_norm": 1.8117423057556152,
-      "learning_rate": 7.54572864321608e-05,
-      "loss": 5.054,
-      "step": 24928
-    },
-    {
-      "epoch": 13.00078226857888,
-      "grad_norm": 1.4200389385223389,
-      "learning_rate": 7.545628140703518e-05,
-      "loss": 5.4375,
-      "step": 24929
-    },
-    {
-      "epoch": 13.001303780964799,
-      "grad_norm": 1.3528181314468384,
-      "learning_rate": 7.545527638190956e-05,
-      "loss": 5.4881,
-      "step": 24930
-    },
-    {
-      "epoch": 13.001825293350716,
-      "grad_norm": 1.4377245903015137,
-      "learning_rate": 7.545427135678393e-05,
-      "loss": 5.7187,
-      "step": 24931
-    },
-    {
-      "epoch": 13.002346805736636,
-      "grad_norm": 1.444352388381958,
-      "learning_rate": 7.545326633165829e-05,
-      "loss": 5.4847,
-      "step": 24932
-    },
-    {
-      "epoch": 13.002868318122555,
-      "grad_norm": 1.4764409065246582,
-      "learning_rate": 7.545226130653267e-05,
-      "loss": 4.8908,
-      "step": 24933
-    },
-    {
-      "epoch": 13.003389830508475,
-      "grad_norm": 1.5274739265441895,
-      "learning_rate": 7.545125628140703e-05,
-      "loss": 5.2774,
-      "step": 24934
-    },
-    {
-      "epoch": 13.003911342894394,
-      "grad_norm": 1.571966290473938,
-      "learning_rate": 7.545025125628141e-05,
-      "loss": 5.018,
-      "step": 24935
-    },
-    {
-      "epoch": 13.004432855280314,
-      "grad_norm": 1.5012880563735962,
-      "learning_rate": 7.544924623115577e-05,
-      "loss": 5.1411,
-      "step": 24936
-    },
-    {
-      "epoch": 13.004954367666231,
-      "grad_norm": 1.4766172170639038,
-      "learning_rate": 7.544824120603015e-05,
-      "loss": 4.8378,
-      "step": 24937
-    },
-    {
-      "epoch": 13.00547588005215,
-      "grad_norm": 1.3836112022399902,
-      "learning_rate": 7.544723618090453e-05,
-      "loss": 5.8167,
-      "step": 24938
-    },
-    {
-      "epoch": 13.00599739243807,
-      "grad_norm": 1.4877537488937378,
-      "learning_rate": 7.544623115577891e-05,
-      "loss": 5.3696,
-      "step": 24939
-    },
-    {
-      "epoch": 13.00651890482399,
-      "grad_norm": 1.4720375537872314,
-      "learning_rate": 7.544522613065327e-05,
-      "loss": 5.3887,
-      "step": 24940
-    },
-    {
-      "epoch": 13.00704041720991,
-      "grad_norm": 1.3576501607894897,
-      "learning_rate": 7.544422110552765e-05,
-      "loss": 5.2396,
-      "step": 24941
-    },
-    {
-      "epoch": 13.007561929595829,
-      "grad_norm": 1.5126111507415771,
-      "learning_rate": 7.544321608040201e-05,
-      "loss": 5.5781,
-      "step": 24942
-    },
-    {
-      "epoch": 13.008083441981746,
-      "grad_norm": 1.4237885475158691,
-      "learning_rate": 7.544221105527639e-05,
-      "loss": 5.3022,
-      "step": 24943
-    },
-    {
-      "epoch": 13.008604954367666,
-      "grad_norm": 1.4641135931015015,
-      "learning_rate": 7.544120603015076e-05,
-      "loss": 5.4139,
-      "step": 24944
-    },
-    {
-      "epoch": 13.009126466753585,
-      "grad_norm": 1.343546748161316,
-      "learning_rate": 7.544020100502512e-05,
-      "loss": 5.3308,
-      "step": 24945
-    },
-    {
-      "epoch": 13.009647979139505,
-      "grad_norm": 1.4830869436264038,
-      "learning_rate": 7.54391959798995e-05,
-      "loss": 5.7493,
-      "step": 24946
-    },
-    {
-      "epoch": 13.010169491525424,
-      "grad_norm": 1.3791488409042358,
-      "learning_rate": 7.543819095477386e-05,
-      "loss": 5.2108,
-      "step": 24947
-    },
-    {
-      "epoch": 13.010691003911344,
-      "grad_norm": 1.5253902673721313,
-      "learning_rate": 7.543718592964824e-05,
-      "loss": 5.1736,
-      "step": 24948
-    },
-    {
-      "epoch": 13.011212516297261,
-      "grad_norm": 1.4549850225448608,
-      "learning_rate": 7.543618090452262e-05,
-      "loss": 5.599,
-      "step": 24949
-    },
-    {
-      "epoch": 13.01173402868318,
-      "grad_norm": 1.3687851428985596,
-      "learning_rate": 7.5435175879397e-05,
-      "loss": 5.2702,
-      "step": 24950
-    },
-    {
-      "epoch": 13.0122555410691,
-      "grad_norm": 1.52349853515625,
-      "learning_rate": 7.543417085427136e-05,
-      "loss": 5.5421,
-      "step": 24951
-    },
-    {
-      "epoch": 13.01277705345502,
-      "grad_norm": 1.5492075681686401,
-      "learning_rate": 7.543316582914574e-05,
-      "loss": 5.4474,
-      "step": 24952
-    },
-    {
-      "epoch": 13.01329856584094,
-      "grad_norm": 1.3883212804794312,
-      "learning_rate": 7.54321608040201e-05,
-      "loss": 5.7876,
-      "step": 24953
-    },
-    {
-      "epoch": 13.013820078226859,
-      "grad_norm": 1.4092342853546143,
-      "learning_rate": 7.543115577889448e-05,
-      "loss": 5.5784,
-      "step": 24954
-    },
-    {
-      "epoch": 13.014341590612776,
-      "grad_norm": 1.574446201324463,
-      "learning_rate": 7.543015075376884e-05,
-      "loss": 5.3385,
-      "step": 24955
-    },
-    {
-      "epoch": 13.014863102998696,
-      "grad_norm": 1.3892382383346558,
-      "learning_rate": 7.542914572864322e-05,
-      "loss": 5.3549,
-      "step": 24956
-    },
-    {
-      "epoch": 13.015384615384615,
-      "grad_norm": 1.4137601852416992,
-      "learning_rate": 7.542814070351759e-05,
-      "loss": 5.4973,
-      "step": 24957
-    },
-    {
-      "epoch": 13.015906127770535,
-      "grad_norm": 1.5254262685775757,
-      "learning_rate": 7.542713567839196e-05,
-      "loss": 5.1363,
-      "step": 24958
-    },
-    {
-      "epoch": 13.016427640156454,
-      "grad_norm": 1.489316701889038,
-      "learning_rate": 7.542613065326634e-05,
-      "loss": 5.514,
-      "step": 24959
-    },
-    {
-      "epoch": 13.016949152542374,
-      "grad_norm": 1.4311922788619995,
-      "learning_rate": 7.54251256281407e-05,
-      "loss": 5.4795,
-      "step": 24960
-    },
-    {
-      "epoch": 13.017470664928291,
-      "grad_norm": 1.4278647899627686,
-      "learning_rate": 7.542412060301508e-05,
-      "loss": 5.5704,
-      "step": 24961
-    },
-    {
-      "epoch": 13.01799217731421,
-      "grad_norm": 1.4737575054168701,
-      "learning_rate": 7.542311557788945e-05,
-      "loss": 5.3254,
-      "step": 24962
-    },
-    {
-      "epoch": 13.01851368970013,
-      "grad_norm": 1.448438048362732,
-      "learning_rate": 7.542211055276383e-05,
-      "loss": 5.3868,
-      "step": 24963
-    },
-    {
-      "epoch": 13.01903520208605,
-      "grad_norm": 1.520855188369751,
-      "learning_rate": 7.542110552763819e-05,
-      "loss": 5.0831,
-      "step": 24964
-    },
-    {
-      "epoch": 13.01955671447197,
-      "grad_norm": 1.5036416053771973,
-      "learning_rate": 7.542010050251257e-05,
-      "loss": 5.6077,
-      "step": 24965
-    },
-    {
-      "epoch": 13.020078226857889,
-      "grad_norm": 1.39872407913208,
-      "learning_rate": 7.541909547738693e-05,
-      "loss": 5.3136,
-      "step": 24966
-    },
-    {
-      "epoch": 13.020599739243806,
-      "grad_norm": 1.4274282455444336,
-      "learning_rate": 7.541809045226131e-05,
-      "loss": 5.7438,
-      "step": 24967
-    },
-    {
-      "epoch": 13.021121251629726,
-      "grad_norm": 1.4124516248703003,
-      "learning_rate": 7.541708542713567e-05,
-      "loss": 5.3484,
-      "step": 24968
-    },
-    {
-      "epoch": 13.021642764015645,
-      "grad_norm": 1.4041868448257446,
-      "learning_rate": 7.541608040201005e-05,
-      "loss": 5.3449,
-      "step": 24969
-    },
-    {
-      "epoch": 13.022164276401565,
-      "grad_norm": 1.3599072694778442,
-      "learning_rate": 7.541507537688443e-05,
-      "loss": 5.6724,
-      "step": 24970
-    },
-    {
-      "epoch": 13.022685788787484,
-      "grad_norm": 1.4578707218170166,
-      "learning_rate": 7.54140703517588e-05,
-      "loss": 5.6556,
-      "step": 24971
-    },
-    {
-      "epoch": 13.023207301173404,
-      "grad_norm": 1.442463994026184,
-      "learning_rate": 7.541306532663317e-05,
-      "loss": 5.4312,
-      "step": 24972
-    },
-    {
-      "epoch": 13.023728813559321,
-      "grad_norm": 1.4686410427093506,
-      "learning_rate": 7.541206030150754e-05,
-      "loss": 5.345,
-      "step": 24973
-    },
-    {
-      "epoch": 13.024250325945241,
-      "grad_norm": 1.5030460357666016,
-      "learning_rate": 7.541105527638191e-05,
-      "loss": 5.3385,
-      "step": 24974
-    },
-    {
-      "epoch": 13.02477183833116,
-      "grad_norm": 1.5403265953063965,
-      "learning_rate": 7.541005025125628e-05,
-      "loss": 4.6231,
-      "step": 24975
-    },
-    {
-      "epoch": 13.02529335071708,
-      "grad_norm": 1.8225224018096924,
-      "learning_rate": 7.540904522613066e-05,
-      "loss": 4.6271,
-      "step": 24976
-    },
-    {
-      "epoch": 13.025814863103,
-      "grad_norm": 1.9311195611953735,
-      "learning_rate": 7.540804020100502e-05,
-      "loss": 5.0377,
-      "step": 24977
-    },
-    {
-      "epoch": 13.026336375488917,
-      "grad_norm": 1.4456654787063599,
-      "learning_rate": 7.54070351758794e-05,
-      "loss": 4.8066,
-      "step": 24978
-    },
-    {
-      "epoch": 13.026857887874836,
-      "grad_norm": 1.4270296096801758,
-      "learning_rate": 7.540603015075378e-05,
-      "loss": 5.5561,
-      "step": 24979
-    },
-    {
-      "epoch": 13.027379400260756,
-      "grad_norm": 1.4739720821380615,
-      "learning_rate": 7.540502512562815e-05,
-      "loss": 4.9893,
-      "step": 24980
-    },
-    {
-      "epoch": 13.027900912646675,
-      "grad_norm": 1.4896410703659058,
-      "learning_rate": 7.540402010050252e-05,
-      "loss": 5.2836,
-      "step": 24981
-    },
-    {
-      "epoch": 13.028422425032595,
-      "grad_norm": 1.4746875762939453,
-      "learning_rate": 7.54030150753769e-05,
-      "loss": 5.5772,
-      "step": 24982
-    },
-    {
-      "epoch": 13.028943937418514,
-      "grad_norm": 1.6914674043655396,
-      "learning_rate": 7.540201005025126e-05,
-      "loss": 4.4894,
-      "step": 24983
-    },
-    {
-      "epoch": 13.029465449804432,
-      "grad_norm": 1.3429515361785889,
-      "learning_rate": 7.540100502512564e-05,
-      "loss": 5.8461,
-      "step": 24984
-    },
-    {
-      "epoch": 13.029986962190351,
-      "grad_norm": 1.4505583047866821,
-      "learning_rate": 7.54e-05,
-      "loss": 5.4922,
-      "step": 24985
-    },
-    {
-      "epoch": 13.030508474576271,
-      "grad_norm": 1.5194478034973145,
-      "learning_rate": 7.539899497487437e-05,
-      "loss": 4.9128,
-      "step": 24986
-    },
-    {
-      "epoch": 13.03102998696219,
-      "grad_norm": 1.4806746244430542,
-      "learning_rate": 7.539798994974875e-05,
-      "loss": 5.2395,
-      "step": 24987
-    },
-    {
-      "epoch": 13.03155149934811,
-      "grad_norm": 1.3566452264785767,
-      "learning_rate": 7.539698492462311e-05,
-      "loss": 5.6985,
-      "step": 24988
-    },
-    {
-      "epoch": 13.03207301173403,
-      "grad_norm": 1.389236569404602,
-      "learning_rate": 7.539597989949749e-05,
-      "loss": 5.4454,
-      "step": 24989
-    },
-    {
-      "epoch": 13.032594524119947,
-      "grad_norm": 1.45772385597229,
-      "learning_rate": 7.539497487437187e-05,
-      "loss": 4.7858,
-      "step": 24990
-    },
-    {
-      "epoch": 13.033116036505866,
-      "grad_norm": 1.4208375215530396,
-      "learning_rate": 7.539396984924624e-05,
-      "loss": 5.8702,
-      "step": 24991
-    },
-    {
-      "epoch": 13.033637548891786,
-      "grad_norm": 1.4874900579452515,
-      "learning_rate": 7.539296482412061e-05,
-      "loss": 5.2632,
-      "step": 24992
-    },
-    {
-      "epoch": 13.034159061277705,
-      "grad_norm": 1.6581958532333374,
-      "learning_rate": 7.539195979899499e-05,
-      "loss": 4.95,
-      "step": 24993
-    },
-    {
-      "epoch": 13.034680573663625,
-      "grad_norm": 1.505218505859375,
-      "learning_rate": 7.539095477386935e-05,
-      "loss": 5.3283,
-      "step": 24994
-    },
-    {
-      "epoch": 13.035202086049544,
-      "grad_norm": 1.5088810920715332,
-      "learning_rate": 7.538994974874373e-05,
-      "loss": 5.2763,
-      "step": 24995
-    },
-    {
-      "epoch": 13.035723598435462,
-      "grad_norm": 1.4907021522521973,
-      "learning_rate": 7.538894472361809e-05,
-      "loss": 5.2277,
-      "step": 24996
-    },
-    {
-      "epoch": 13.036245110821381,
-      "grad_norm": 1.4617527723312378,
-      "learning_rate": 7.538793969849247e-05,
-      "loss": 5.4886,
-      "step": 24997
-    },
-    {
-      "epoch": 13.036766623207301,
-      "grad_norm": 1.5130726099014282,
-      "learning_rate": 7.538693467336683e-05,
-      "loss": 5.1118,
-      "step": 24998
-    },
-    {
-      "epoch": 13.03728813559322,
-      "grad_norm": 1.6083691120147705,
-      "learning_rate": 7.538592964824121e-05,
-      "loss": 5.5955,
-      "step": 24999
-    },
-    {
-      "epoch": 13.03780964797914,
-      "grad_norm": 1.532025694847107,
-      "learning_rate": 7.538492462311559e-05,
-      "loss": 4.9124,
-      "step": 25000
-    },
-    {
-      "epoch": 13.03780964797914,
-      "eval_loss": 5.480602741241455,
-      "eval_runtime": 42.6816,
-      "eval_samples_per_second": 28.724,
-      "eval_steps_per_second": 3.608,
-      "step": 25000
-    },
-    {
-      "epoch": 13.03833116036506,
-      "grad_norm": 1.5681285858154297,
-      "learning_rate": 7.538391959798995e-05,
-      "loss": 5.4305,
-      "step": 25001
-    },
-    {
-      "epoch": 13.038852672750977,
-      "grad_norm": 1.3859326839447021,
-      "learning_rate": 7.538291457286433e-05,
-      "loss": 5.3414,
-      "step": 25002
-    },
-    {
-      "epoch": 13.039374185136897,
-      "grad_norm": 1.4273037910461426,
-      "learning_rate": 7.53819095477387e-05,
-      "loss": 5.1319,
-      "step": 25003
-    },
-    {
-      "epoch": 13.039895697522816,
-      "grad_norm": 1.4868451356887817,
-      "learning_rate": 7.538090452261307e-05,
-      "loss": 5.6115,
-      "step": 25004
-    },
-    {
-      "epoch": 13.040417209908735,
-      "grad_norm": 1.4755724668502808,
-      "learning_rate": 7.537989949748744e-05,
-      "loss": 5.1901,
-      "step": 25005
-    },
-    {
-      "epoch": 13.040938722294655,
-      "grad_norm": 1.3661320209503174,
-      "learning_rate": 7.537889447236182e-05,
-      "loss": 5.532,
-      "step": 25006
-    },
-    {
-      "epoch": 13.041460234680574,
-      "grad_norm": 1.506785273551941,
-      "learning_rate": 7.537788944723618e-05,
-      "loss": 5.6451,
-      "step": 25007
-    },
-    {
-      "epoch": 13.041981747066492,
-      "grad_norm": 1.433674693107605,
-      "learning_rate": 7.537688442211056e-05,
-      "loss": 5.8557,
-      "step": 25008
-    },
-    {
-      "epoch": 13.042503259452412,
-      "grad_norm": 1.4180102348327637,
-      "learning_rate": 7.537587939698492e-05,
-      "loss": 5.3097,
-      "step": 25009
-    },
-    {
-      "epoch": 13.043024771838331,
-      "grad_norm": 1.4880775213241577,
-      "learning_rate": 7.53748743718593e-05,
-      "loss": 5.2601,
-      "step": 25010
-    },
-    {
-      "epoch": 13.04354628422425,
-      "grad_norm": 1.4273602962493896,
-      "learning_rate": 7.537386934673368e-05,
-      "loss": 5.1229,
-      "step": 25011
-    },
-    {
-      "epoch": 13.04406779661017,
-      "grad_norm": 1.4471709728240967,
-      "learning_rate": 7.537286432160804e-05,
-      "loss": 5.2641,
-      "step": 25012
-    },
-    {
-      "epoch": 13.04458930899609,
-      "grad_norm": 1.4549627304077148,
-      "learning_rate": 7.537185929648242e-05,
-      "loss": 5.1747,
-      "step": 25013
-    },
-    {
-      "epoch": 13.045110821382007,
-      "grad_norm": 1.3808501958847046,
-      "learning_rate": 7.537085427135678e-05,
-      "loss": 5.1887,
-      "step": 25014
-    },
-    {
-      "epoch": 13.045632333767927,
-      "grad_norm": 1.4362021684646606,
-      "learning_rate": 7.536984924623116e-05,
-      "loss": 5.7697,
-      "step": 25015
-    },
-    {
-      "epoch": 13.046153846153846,
-      "grad_norm": 1.4059470891952515,
-      "learning_rate": 7.536884422110553e-05,
-      "loss": 5.5131,
-      "step": 25016
-    },
-    {
-      "epoch": 13.046675358539765,
-      "grad_norm": 1.3300868272781372,
-      "learning_rate": 7.53678391959799e-05,
-      "loss": 5.8265,
-      "step": 25017
-    },
-    {
-      "epoch": 13.047196870925685,
-      "grad_norm": 1.4380558729171753,
-      "learning_rate": 7.536683417085427e-05,
-      "loss": 5.283,
-      "step": 25018
-    },
-    {
-      "epoch": 13.047718383311604,
-      "grad_norm": 1.5371414422988892,
-      "learning_rate": 7.536582914572865e-05,
-      "loss": 5.0966,
-      "step": 25019
-    },
-    {
-      "epoch": 13.048239895697522,
-      "grad_norm": 1.8003718852996826,
-      "learning_rate": 7.536482412060301e-05,
-      "loss": 5.1469,
-      "step": 25020
-    },
-    {
-      "epoch": 13.048761408083442,
-      "grad_norm": 1.404632806777954,
-      "learning_rate": 7.536381909547739e-05,
-      "loss": 5.1903,
-      "step": 25021
-    },
-    {
-      "epoch": 13.049282920469361,
-      "grad_norm": 1.5047526359558105,
-      "learning_rate": 7.536281407035177e-05,
-      "loss": 5.1553,
-      "step": 25022
-    },
-    {
-      "epoch": 13.04980443285528,
-      "grad_norm": 1.447920322418213,
-      "learning_rate": 7.536180904522614e-05,
-      "loss": 5.3584,
-      "step": 25023
-    },
-    {
-      "epoch": 13.0503259452412,
-      "grad_norm": 1.5456644296646118,
-      "learning_rate": 7.536080402010051e-05,
-      "loss": 5.202,
-      "step": 25024
-    },
-    {
-      "epoch": 13.05084745762712,
-      "grad_norm": 1.4793750047683716,
-      "learning_rate": 7.535979899497487e-05,
-      "loss": 4.9771,
-      "step": 25025
-    },
-    {
-      "epoch": 13.051368970013037,
-      "grad_norm": 1.5456069707870483,
-      "learning_rate": 7.535879396984925e-05,
-      "loss": 5.7858,
-      "step": 25026
-    },
-    {
-      "epoch": 13.051890482398957,
-      "grad_norm": 1.538077473640442,
-      "learning_rate": 7.535778894472361e-05,
-      "loss": 5.0891,
-      "step": 25027
-    },
-    {
-      "epoch": 13.052411994784876,
-      "grad_norm": 1.4607704877853394,
-      "learning_rate": 7.535678391959799e-05,
-      "loss": 5.4683,
-      "step": 25028
-    },
-    {
-      "epoch": 13.052933507170795,
-      "grad_norm": 1.4046400785446167,
-      "learning_rate": 7.535577889447236e-05,
-      "loss": 5.6173,
-      "step": 25029
-    },
-    {
-      "epoch": 13.053455019556715,
-      "grad_norm": 1.4793953895568848,
-      "learning_rate": 7.535477386934673e-05,
-      "loss": 5.2467,
-      "step": 25030
-    },
-    {
-      "epoch": 13.053976531942634,
-      "grad_norm": 1.3344489336013794,
-      "learning_rate": 7.535376884422111e-05,
-      "loss": 5.549,
-      "step": 25031
-    },
-    {
-      "epoch": 13.054498044328552,
-      "grad_norm": 1.6344150304794312,
-      "learning_rate": 7.535276381909549e-05,
-      "loss": 5.2559,
-      "step": 25032
-    },
-    {
-      "epoch": 13.055019556714472,
-      "grad_norm": 1.4933382272720337,
-      "learning_rate": 7.535175879396985e-05,
-      "loss": 5.2285,
-      "step": 25033
-    },
-    {
-      "epoch": 13.055541069100391,
-      "grad_norm": 1.4500466585159302,
-      "learning_rate": 7.535075376884423e-05,
-      "loss": 5.2132,
-      "step": 25034
-    },
-    {
-      "epoch": 13.05606258148631,
-      "grad_norm": 1.4440773725509644,
-      "learning_rate": 7.53497487437186e-05,
-      "loss": 5.4344,
-      "step": 25035
-    },
-    {
-      "epoch": 13.05658409387223,
-      "grad_norm": 1.4005767107009888,
-      "learning_rate": 7.534874371859297e-05,
-      "loss": 5.6166,
-      "step": 25036
-    },
-    {
-      "epoch": 13.05710560625815,
-      "grad_norm": 1.4739924669265747,
-      "learning_rate": 7.534773869346734e-05,
-      "loss": 5.766,
-      "step": 25037
-    },
-    {
-      "epoch": 13.057627118644067,
-      "grad_norm": 1.4168720245361328,
-      "learning_rate": 7.53467336683417e-05,
-      "loss": 5.4681,
-      "step": 25038
-    },
-    {
-      "epoch": 13.058148631029987,
-      "grad_norm": 1.4937533140182495,
-      "learning_rate": 7.534572864321608e-05,
-      "loss": 5.598,
-      "step": 25039
-    },
-    {
-      "epoch": 13.058670143415906,
-      "grad_norm": 1.4248523712158203,
-      "learning_rate": 7.534472361809044e-05,
-      "loss": 5.347,
-      "step": 25040
-    },
-    {
-      "epoch": 13.059191655801826,
-      "grad_norm": 1.472029209136963,
-      "learning_rate": 7.534371859296482e-05,
-      "loss": 5.6407,
-      "step": 25041
-    },
-    {
-      "epoch": 13.059713168187745,
-      "grad_norm": 1.4908260107040405,
-      "learning_rate": 7.53427135678392e-05,
-      "loss": 5.5702,
-      "step": 25042
-    },
-    {
-      "epoch": 13.060234680573664,
-      "grad_norm": 1.367125153541565,
-      "learning_rate": 7.534170854271358e-05,
-      "loss": 5.2285,
-      "step": 25043
-    },
-    {
-      "epoch": 13.060756192959582,
-      "grad_norm": 1.3811912536621094,
-      "learning_rate": 7.534070351758794e-05,
-      "loss": 5.7186,
-      "step": 25044
-    },
-    {
-      "epoch": 13.061277705345502,
-      "grad_norm": 1.4316325187683105,
-      "learning_rate": 7.533969849246232e-05,
-      "loss": 5.1802,
-      "step": 25045
-    },
-    {
-      "epoch": 13.061799217731421,
-      "grad_norm": 1.4048737287521362,
-      "learning_rate": 7.533869346733668e-05,
-      "loss": 5.4774,
-      "step": 25046
-    },
-    {
-      "epoch": 13.06232073011734,
-      "grad_norm": 1.6188302040100098,
-      "learning_rate": 7.533768844221106e-05,
-      "loss": 4.7949,
-      "step": 25047
-    },
-    {
-      "epoch": 13.06284224250326,
-      "grad_norm": 1.411867380142212,
-      "learning_rate": 7.533668341708543e-05,
-      "loss": 5.5588,
-      "step": 25048
-    },
-    {
-      "epoch": 13.06336375488918,
-      "grad_norm": 1.5180368423461914,
-      "learning_rate": 7.53356783919598e-05,
-      "loss": 5.1152,
-      "step": 25049
-    },
-    {
-      "epoch": 13.063885267275097,
-      "grad_norm": 1.5499504804611206,
-      "learning_rate": 7.533467336683417e-05,
-      "loss": 5.4896,
-      "step": 25050
-    },
-    {
-      "epoch": 13.064406779661017,
-      "grad_norm": 1.3863475322723389,
-      "learning_rate": 7.533366834170855e-05,
-      "loss": 5.6585,
-      "step": 25051
-    },
-    {
-      "epoch": 13.064928292046936,
-      "grad_norm": 1.4281535148620605,
-      "learning_rate": 7.533266331658292e-05,
-      "loss": 5.8017,
-      "step": 25052
-    },
-    {
-      "epoch": 13.065449804432856,
-      "grad_norm": 1.483319878578186,
-      "learning_rate": 7.533165829145729e-05,
-      "loss": 5.3709,
-      "step": 25053
-    },
-    {
-      "epoch": 13.065971316818775,
-      "grad_norm": 1.4394121170043945,
-      "learning_rate": 7.533065326633167e-05,
-      "loss": 5.5648,
-      "step": 25054
-    },
-    {
-      "epoch": 13.066492829204694,
-      "grad_norm": 1.5054552555084229,
-      "learning_rate": 7.532964824120603e-05,
-      "loss": 5.061,
-      "step": 25055
-    },
-    {
-      "epoch": 13.067014341590612,
-      "grad_norm": 1.3593753576278687,
-      "learning_rate": 7.532864321608041e-05,
-      "loss": 5.5065,
-      "step": 25056
-    },
-    {
-      "epoch": 13.067535853976532,
-      "grad_norm": 1.4533852338790894,
-      "learning_rate": 7.532763819095477e-05,
-      "loss": 5.4714,
-      "step": 25057
-    },
-    {
-      "epoch": 13.068057366362451,
-      "grad_norm": 1.375229835510254,
-      "learning_rate": 7.532663316582915e-05,
-      "loss": 5.5585,
-      "step": 25058
-    },
-    {
-      "epoch": 13.06857887874837,
-      "grad_norm": 1.451245665550232,
-      "learning_rate": 7.532562814070352e-05,
-      "loss": 5.2783,
-      "step": 25059
-    },
-    {
-      "epoch": 13.06910039113429,
-      "grad_norm": 1.4193123579025269,
-      "learning_rate": 7.532462311557789e-05,
-      "loss": 5.5125,
-      "step": 25060
-    },
-    {
-      "epoch": 13.06962190352021,
-      "grad_norm": 1.4900869131088257,
-      "learning_rate": 7.532361809045226e-05,
-      "loss": 5.3788,
-      "step": 25061
-    },
-    {
-      "epoch": 13.070143415906127,
-      "grad_norm": 1.488438367843628,
-      "learning_rate": 7.532261306532664e-05,
-      "loss": 5.1694,
-      "step": 25062
-    },
-    {
-      "epoch": 13.070664928292047,
-      "grad_norm": 1.3810878992080688,
-      "learning_rate": 7.532160804020101e-05,
-      "loss": 5.7402,
-      "step": 25063
-    },
-    {
-      "epoch": 13.071186440677966,
-      "grad_norm": 1.3966197967529297,
-      "learning_rate": 7.532060301507538e-05,
-      "loss": 4.9495,
-      "step": 25064
-    },
-    {
-      "epoch": 13.071707953063886,
-      "grad_norm": 1.4670382738113403,
-      "learning_rate": 7.531959798994976e-05,
-      "loss": 5.1611,
-      "step": 25065
-    },
-    {
-      "epoch": 13.072229465449805,
-      "grad_norm": 1.3351085186004639,
-      "learning_rate": 7.531859296482412e-05,
-      "loss": 5.6076,
-      "step": 25066
-    },
-    {
-      "epoch": 13.072750977835724,
-      "grad_norm": 1.4964770078659058,
-      "learning_rate": 7.53175879396985e-05,
-      "loss": 5.1354,
-      "step": 25067
-    },
-    {
-      "epoch": 13.073272490221642,
-      "grad_norm": 1.5482255220413208,
-      "learning_rate": 7.531658291457286e-05,
-      "loss": 5.4422,
-      "step": 25068
-    },
-    {
-      "epoch": 13.073794002607562,
-      "grad_norm": 1.5550713539123535,
-      "learning_rate": 7.531557788944724e-05,
-      "loss": 5.0096,
-      "step": 25069
-    },
-    {
-      "epoch": 13.074315514993481,
-      "grad_norm": 1.3136382102966309,
-      "learning_rate": 7.53145728643216e-05,
-      "loss": 5.4884,
-      "step": 25070
-    },
-    {
-      "epoch": 13.0748370273794,
-      "grad_norm": 1.5356523990631104,
-      "learning_rate": 7.531356783919598e-05,
-      "loss": 5.2229,
-      "step": 25071
-    },
-    {
-      "epoch": 13.07535853976532,
-      "grad_norm": 1.3667304515838623,
-      "learning_rate": 7.531256281407036e-05,
-      "loss": 5.7175,
-      "step": 25072
-    },
-    {
-      "epoch": 13.075880052151238,
-      "grad_norm": 1.4704686403274536,
-      "learning_rate": 7.531155778894474e-05,
-      "loss": 5.6305,
-      "step": 25073
-    },
-    {
-      "epoch": 13.076401564537157,
-      "grad_norm": 1.3831396102905273,
-      "learning_rate": 7.53105527638191e-05,
-      "loss": 5.3539,
-      "step": 25074
-    },
-    {
-      "epoch": 13.076923076923077,
-      "grad_norm": 1.3950331211090088,
-      "learning_rate": 7.530954773869348e-05,
-      "loss": 5.5623,
-      "step": 25075
-    },
-    {
-      "epoch": 13.077444589308996,
-      "grad_norm": 1.4728302955627441,
-      "learning_rate": 7.530854271356784e-05,
-      "loss": 5.7949,
-      "step": 25076
-    },
-    {
-      "epoch": 13.077966101694916,
-      "grad_norm": 1.4279857873916626,
-      "learning_rate": 7.530753768844222e-05,
-      "loss": 5.6959,
-      "step": 25077
-    },
-    {
-      "epoch": 13.078487614080835,
-      "grad_norm": 1.4262562990188599,
-      "learning_rate": 7.530653266331659e-05,
-      "loss": 4.7335,
-      "step": 25078
-    },
-    {
-      "epoch": 13.079009126466753,
-      "grad_norm": 1.4884908199310303,
-      "learning_rate": 7.530552763819095e-05,
-      "loss": 5.054,
-      "step": 25079
-    },
-    {
-      "epoch": 13.079530638852672,
-      "grad_norm": 1.4779795408248901,
-      "learning_rate": 7.530452261306533e-05,
-      "loss": 5.6333,
-      "step": 25080
-    },
-    {
-      "epoch": 13.080052151238592,
-      "grad_norm": 1.4568383693695068,
-      "learning_rate": 7.530351758793969e-05,
-      "loss": 5.7795,
-      "step": 25081
-    },
-    {
-      "epoch": 13.080573663624511,
-      "grad_norm": 1.469770073890686,
-      "learning_rate": 7.530251256281407e-05,
-      "loss": 5.5671,
-      "step": 25082
-    },
-    {
-      "epoch": 13.08109517601043,
-      "grad_norm": 1.5236530303955078,
-      "learning_rate": 7.530150753768845e-05,
-      "loss": 5.6963,
-      "step": 25083
-    },
-    {
-      "epoch": 13.08161668839635,
-      "grad_norm": 1.540531039237976,
-      "learning_rate": 7.530050251256283e-05,
-      "loss": 5.4676,
-      "step": 25084
-    },
-    {
-      "epoch": 13.082138200782268,
-      "grad_norm": 1.5736092329025269,
-      "learning_rate": 7.529949748743719e-05,
-      "loss": 5.3604,
-      "step": 25085
-    },
-    {
-      "epoch": 13.082659713168187,
-      "grad_norm": 1.4377678632736206,
-      "learning_rate": 7.529849246231157e-05,
-      "loss": 4.9539,
-      "step": 25086
-    },
-    {
-      "epoch": 13.083181225554107,
-      "grad_norm": 1.4185621738433838,
-      "learning_rate": 7.529748743718593e-05,
-      "loss": 5.7506,
-      "step": 25087
-    },
-    {
-      "epoch": 13.083702737940026,
-      "grad_norm": 1.419871211051941,
-      "learning_rate": 7.529648241206031e-05,
-      "loss": 5.5598,
-      "step": 25088
-    },
-    {
-      "epoch": 13.084224250325946,
-      "grad_norm": 1.4907103776931763,
-      "learning_rate": 7.529547738693467e-05,
-      "loss": 5.4017,
-      "step": 25089
-    },
-    {
-      "epoch": 13.084745762711865,
-      "grad_norm": 1.472992181777954,
-      "learning_rate": 7.529447236180905e-05,
-      "loss": 5.4528,
-      "step": 25090
-    },
-    {
-      "epoch": 13.085267275097783,
-      "grad_norm": 1.439988374710083,
-      "learning_rate": 7.529346733668342e-05,
-      "loss": 5.3537,
-      "step": 25091
-    },
-    {
-      "epoch": 13.085788787483702,
-      "grad_norm": 1.504423975944519,
-      "learning_rate": 7.52924623115578e-05,
-      "loss": 5.1763,
-      "step": 25092
-    },
-    {
-      "epoch": 13.086310299869622,
-      "grad_norm": 1.5364910364151,
-      "learning_rate": 7.529145728643217e-05,
-      "loss": 5.4415,
-      "step": 25093
-    },
-    {
-      "epoch": 13.086831812255541,
-      "grad_norm": 1.5027742385864258,
-      "learning_rate": 7.529045226130654e-05,
-      "loss": 4.7892,
-      "step": 25094
-    },
-    {
-      "epoch": 13.08735332464146,
-      "grad_norm": 1.5341349840164185,
-      "learning_rate": 7.528944723618091e-05,
-      "loss": 5.343,
-      "step": 25095
-    },
-    {
-      "epoch": 13.08787483702738,
-      "grad_norm": 1.5523431301116943,
-      "learning_rate": 7.528844221105528e-05,
-      "loss": 5.4983,
-      "step": 25096
-    },
-    {
-      "epoch": 13.088396349413298,
-      "grad_norm": 1.6440255641937256,
-      "learning_rate": 7.528743718592966e-05,
-      "loss": 5.2632,
-      "step": 25097
-    },
-    {
-      "epoch": 13.088917861799217,
-      "grad_norm": 1.6088463068008423,
-      "learning_rate": 7.528643216080402e-05,
-      "loss": 5.4873,
-      "step": 25098
-    },
-    {
-      "epoch": 13.089439374185137,
-      "grad_norm": 1.656606674194336,
-      "learning_rate": 7.52854271356784e-05,
-      "loss": 4.9075,
-      "step": 25099
-    },
-    {
-      "epoch": 13.089960886571056,
-      "grad_norm": 1.4864325523376465,
-      "learning_rate": 7.528442211055276e-05,
-      "loss": 5.5789,
-      "step": 25100
-    },
-    {
-      "epoch": 13.090482398956976,
-      "grad_norm": 1.5176934003829956,
-      "learning_rate": 7.528341708542714e-05,
-      "loss": 5.2836,
-      "step": 25101
-    },
-    {
-      "epoch": 13.091003911342895,
-      "grad_norm": 1.4057013988494873,
-      "learning_rate": 7.52824120603015e-05,
-      "loss": 5.5386,
-      "step": 25102
-    },
-    {
-      "epoch": 13.091525423728813,
-      "grad_norm": 1.4734258651733398,
-      "learning_rate": 7.528140703517588e-05,
-      "loss": 5.2982,
-      "step": 25103
-    },
-    {
-      "epoch": 13.092046936114732,
-      "grad_norm": 1.4289977550506592,
-      "learning_rate": 7.528040201005026e-05,
-      "loss": 5.6123,
-      "step": 25104
-    },
-    {
-      "epoch": 13.092568448500652,
-      "grad_norm": 1.5152521133422852,
-      "learning_rate": 7.527939698492462e-05,
-      "loss": 5.5751,
-      "step": 25105
-    },
-    {
-      "epoch": 13.093089960886571,
-      "grad_norm": 1.7377262115478516,
-      "learning_rate": 7.5278391959799e-05,
-      "loss": 5.5361,
-      "step": 25106
-    },
-    {
-      "epoch": 13.09361147327249,
-      "grad_norm": 1.4620219469070435,
-      "learning_rate": 7.527738693467337e-05,
-      "loss": 5.4823,
-      "step": 25107
-    },
-    {
-      "epoch": 13.09413298565841,
-      "grad_norm": 1.376672387123108,
-      "learning_rate": 7.527638190954774e-05,
-      "loss": 5.4084,
-      "step": 25108
-    },
-    {
-      "epoch": 13.094654498044328,
-      "grad_norm": 1.4942291975021362,
-      "learning_rate": 7.527537688442211e-05,
-      "loss": 4.8811,
-      "step": 25109
-    },
-    {
-      "epoch": 13.095176010430247,
-      "grad_norm": 1.4416841268539429,
-      "learning_rate": 7.527437185929649e-05,
-      "loss": 5.336,
-      "step": 25110
-    },
-    {
-      "epoch": 13.095697522816167,
-      "grad_norm": 1.4809435606002808,
-      "learning_rate": 7.527336683417085e-05,
-      "loss": 5.582,
-      "step": 25111
-    },
-    {
-      "epoch": 13.096219035202086,
-      "grad_norm": 1.5408440828323364,
-      "learning_rate": 7.527236180904523e-05,
-      "loss": 5.1958,
-      "step": 25112
-    },
-    {
-      "epoch": 13.096740547588006,
-      "grad_norm": 1.3429187536239624,
-      "learning_rate": 7.52713567839196e-05,
-      "loss": 5.248,
-      "step": 25113
-    },
-    {
-      "epoch": 13.097262059973925,
-      "grad_norm": 1.489094853401184,
-      "learning_rate": 7.527035175879398e-05,
-      "loss": 5.1926,
-      "step": 25114
-    },
-    {
-      "epoch": 13.097783572359843,
-      "grad_norm": 1.5334630012512207,
-      "learning_rate": 7.526934673366835e-05,
-      "loss": 5.5628,
-      "step": 25115
-    },
-    {
-      "epoch": 13.098305084745762,
-      "grad_norm": 1.4755303859710693,
-      "learning_rate": 7.526834170854273e-05,
-      "loss": 5.4832,
-      "step": 25116
-    },
-    {
-      "epoch": 13.098826597131682,
-      "grad_norm": 1.5134422779083252,
-      "learning_rate": 7.526733668341709e-05,
-      "loss": 5.2358,
-      "step": 25117
-    },
-    {
-      "epoch": 13.099348109517601,
-      "grad_norm": 1.5732399225234985,
-      "learning_rate": 7.526633165829145e-05,
-      "loss": 4.5971,
-      "step": 25118
-    },
-    {
-      "epoch": 13.09986962190352,
-      "grad_norm": 1.4462542533874512,
-      "learning_rate": 7.526532663316583e-05,
-      "loss": 5.5894,
-      "step": 25119
-    },
-    {
-      "epoch": 13.10039113428944,
-      "grad_norm": 1.480206847190857,
-      "learning_rate": 7.52643216080402e-05,
-      "loss": 5.3836,
-      "step": 25120
-    },
-    {
-      "epoch": 13.100912646675358,
-      "grad_norm": 1.431334376335144,
-      "learning_rate": 7.526331658291457e-05,
-      "loss": 5.4143,
-      "step": 25121
-    },
-    {
-      "epoch": 13.101434159061277,
-      "grad_norm": 1.4676493406295776,
-      "learning_rate": 7.526231155778894e-05,
-      "loss": 5.2956,
-      "step": 25122
-    },
-    {
-      "epoch": 13.101955671447197,
-      "grad_norm": 1.5867829322814941,
-      "learning_rate": 7.526130653266332e-05,
-      "loss": 5.5868,
-      "step": 25123
-    },
-    {
-      "epoch": 13.102477183833116,
-      "grad_norm": 1.489991545677185,
-      "learning_rate": 7.52603015075377e-05,
-      "loss": 5.2759,
-      "step": 25124
-    },
-    {
-      "epoch": 13.102998696219036,
-      "grad_norm": 1.480041742324829,
-      "learning_rate": 7.525929648241207e-05,
-      "loss": 5.1028,
-      "step": 25125
-    },
-    {
-      "epoch": 13.103520208604955,
-      "grad_norm": 1.3973528146743774,
-      "learning_rate": 7.525829145728644e-05,
-      "loss": 5.4011,
-      "step": 25126
-    },
-    {
-      "epoch": 13.104041720990873,
-      "grad_norm": 1.4877779483795166,
-      "learning_rate": 7.525728643216081e-05,
-      "loss": 4.795,
-      "step": 25127
-    },
-    {
-      "epoch": 13.104563233376792,
-      "grad_norm": 1.6170357465744019,
-      "learning_rate": 7.525628140703518e-05,
-      "loss": 4.7331,
-      "step": 25128
-    },
-    {
-      "epoch": 13.105084745762712,
-      "grad_norm": 1.3885246515274048,
-      "learning_rate": 7.525527638190956e-05,
-      "loss": 5.6365,
-      "step": 25129
-    },
-    {
-      "epoch": 13.105606258148631,
-      "grad_norm": 1.3848153352737427,
-      "learning_rate": 7.525427135678392e-05,
-      "loss": 5.5304,
-      "step": 25130
-    },
-    {
-      "epoch": 13.10612777053455,
-      "grad_norm": 1.4276084899902344,
-      "learning_rate": 7.525326633165829e-05,
-      "loss": 5.598,
-      "step": 25131
-    },
-    {
-      "epoch": 13.10664928292047,
-      "grad_norm": 1.6110529899597168,
-      "learning_rate": 7.525226130653266e-05,
-      "loss": 5.1046,
-      "step": 25132
-    },
-    {
-      "epoch": 13.107170795306388,
-      "grad_norm": 1.3766005039215088,
-      "learning_rate": 7.525125628140704e-05,
-      "loss": 5.5362,
-      "step": 25133
-    },
-    {
-      "epoch": 13.107692307692307,
-      "grad_norm": 1.3922392129898071,
-      "learning_rate": 7.525025125628142e-05,
-      "loss": 5.7649,
-      "step": 25134
-    },
-    {
-      "epoch": 13.108213820078227,
-      "grad_norm": 1.407058835029602,
-      "learning_rate": 7.524924623115578e-05,
-      "loss": 5.4345,
-      "step": 25135
-    },
-    {
-      "epoch": 13.108735332464146,
-      "grad_norm": 1.4444992542266846,
-      "learning_rate": 7.524824120603016e-05,
-      "loss": 4.9895,
-      "step": 25136
-    },
-    {
-      "epoch": 13.109256844850066,
-      "grad_norm": 1.5205063819885254,
-      "learning_rate": 7.524723618090452e-05,
-      "loss": 5.4086,
-      "step": 25137
-    },
-    {
-      "epoch": 13.109778357235985,
-      "grad_norm": 1.4468477964401245,
-      "learning_rate": 7.52462311557789e-05,
-      "loss": 5.8485,
-      "step": 25138
-    },
-    {
-      "epoch": 13.110299869621903,
-      "grad_norm": 1.4355002641677856,
-      "learning_rate": 7.524522613065327e-05,
-      "loss": 5.4352,
-      "step": 25139
-    },
-    {
-      "epoch": 13.110821382007822,
-      "grad_norm": 1.4055415391921997,
-      "learning_rate": 7.524422110552764e-05,
-      "loss": 5.8012,
-      "step": 25140
-    },
-    {
-      "epoch": 13.111342894393742,
-      "grad_norm": 1.4706157445907593,
-      "learning_rate": 7.524321608040201e-05,
-      "loss": 5.4683,
-      "step": 25141
-    },
-    {
-      "epoch": 13.111864406779661,
-      "grad_norm": 1.3362462520599365,
-      "learning_rate": 7.524221105527639e-05,
-      "loss": 5.9222,
-      "step": 25142
-    },
-    {
-      "epoch": 13.11238591916558,
-      "grad_norm": 1.371922492980957,
-      "learning_rate": 7.524120603015075e-05,
-      "loss": 5.6387,
-      "step": 25143
-    },
-    {
-      "epoch": 13.1129074315515,
-      "grad_norm": 1.631081461906433,
-      "learning_rate": 7.524020100502513e-05,
-      "loss": 5.0964,
-      "step": 25144
-    },
-    {
-      "epoch": 13.113428943937418,
-      "grad_norm": 1.402008295059204,
-      "learning_rate": 7.523919597989951e-05,
-      "loss": 5.1552,
-      "step": 25145
-    },
-    {
-      "epoch": 13.113950456323337,
-      "grad_norm": 1.4351791143417358,
-      "learning_rate": 7.523819095477387e-05,
-      "loss": 5.4799,
-      "step": 25146
-    },
-    {
-      "epoch": 13.114471968709257,
-      "grad_norm": 1.4819391965866089,
-      "learning_rate": 7.523718592964825e-05,
-      "loss": 5.4855,
-      "step": 25147
-    },
-    {
-      "epoch": 13.114993481095176,
-      "grad_norm": 1.4288625717163086,
-      "learning_rate": 7.523618090452261e-05,
-      "loss": 5.1918,
-      "step": 25148
-    },
-    {
-      "epoch": 13.115514993481096,
-      "grad_norm": 1.4658851623535156,
-      "learning_rate": 7.523517587939699e-05,
-      "loss": 5.3301,
-      "step": 25149
-    },
-    {
-      "epoch": 13.116036505867015,
-      "grad_norm": 1.4835419654846191,
-      "learning_rate": 7.523417085427136e-05,
-      "loss": 5.3442,
-      "step": 25150
-    },
-    {
-      "epoch": 13.116558018252933,
-      "grad_norm": 1.4519551992416382,
-      "learning_rate": 7.523316582914573e-05,
-      "loss": 5.3356,
-      "step": 25151
-    },
-    {
-      "epoch": 13.117079530638852,
-      "grad_norm": 1.4122859239578247,
-      "learning_rate": 7.52321608040201e-05,
-      "loss": 5.2117,
-      "step": 25152
-    },
-    {
-      "epoch": 13.117601043024772,
-      "grad_norm": 1.409703016281128,
-      "learning_rate": 7.523115577889448e-05,
-      "loss": 5.2204,
-      "step": 25153
-    },
-    {
-      "epoch": 13.118122555410691,
-      "grad_norm": 1.4363669157028198,
-      "learning_rate": 7.523015075376885e-05,
-      "loss": 5.6138,
-      "step": 25154
-    },
-    {
-      "epoch": 13.11864406779661,
-      "grad_norm": 1.5355581045150757,
-      "learning_rate": 7.522914572864323e-05,
-      "loss": 5.0141,
-      "step": 25155
-    },
-    {
-      "epoch": 13.11916558018253,
-      "grad_norm": 1.4587815999984741,
-      "learning_rate": 7.52281407035176e-05,
-      "loss": 5.4516,
-      "step": 25156
-    },
-    {
-      "epoch": 13.119687092568448,
-      "grad_norm": 1.5065200328826904,
-      "learning_rate": 7.522713567839196e-05,
-      "loss": 5.1917,
-      "step": 25157
-    },
-    {
-      "epoch": 13.120208604954367,
-      "grad_norm": 1.281298041343689,
-      "learning_rate": 7.522613065326634e-05,
-      "loss": 5.2311,
-      "step": 25158
-    },
-    {
-      "epoch": 13.120730117340287,
-      "grad_norm": 1.365914225578308,
-      "learning_rate": 7.52251256281407e-05,
-      "loss": 5.395,
-      "step": 25159
-    },
-    {
-      "epoch": 13.121251629726206,
-      "grad_norm": 1.7724839448928833,
-      "learning_rate": 7.522412060301508e-05,
-      "loss": 4.9295,
-      "step": 25160
-    },
-    {
-      "epoch": 13.121773142112126,
-      "grad_norm": 1.4676743745803833,
-      "learning_rate": 7.522311557788944e-05,
-      "loss": 5.0951,
-      "step": 25161
-    },
-    {
-      "epoch": 13.122294654498045,
-      "grad_norm": 1.4871352910995483,
-      "learning_rate": 7.522211055276382e-05,
-      "loss": 5.1357,
-      "step": 25162
-    },
-    {
-      "epoch": 13.122816166883963,
-      "grad_norm": 1.485009789466858,
-      "learning_rate": 7.522110552763819e-05,
-      "loss": 5.4757,
-      "step": 25163
-    },
-    {
-      "epoch": 13.123337679269882,
-      "grad_norm": 1.4658973217010498,
-      "learning_rate": 7.522010050251256e-05,
-      "loss": 5.2949,
-      "step": 25164
-    },
-    {
-      "epoch": 13.123859191655802,
-      "grad_norm": 1.4807689189910889,
-      "learning_rate": 7.521909547738694e-05,
-      "loss": 5.5555,
-      "step": 25165
-    },
-    {
-      "epoch": 13.124380704041721,
-      "grad_norm": 1.4030925035476685,
-      "learning_rate": 7.521809045226132e-05,
-      "loss": 5.5406,
-      "step": 25166
-    },
-    {
-      "epoch": 13.12490221642764,
-      "grad_norm": 1.4906975030899048,
-      "learning_rate": 7.521708542713568e-05,
-      "loss": 5.1398,
-      "step": 25167
-    },
-    {
-      "epoch": 13.125423728813558,
-      "grad_norm": 1.5304378271102905,
-      "learning_rate": 7.521608040201006e-05,
-      "loss": 5.3768,
-      "step": 25168
-    },
-    {
-      "epoch": 13.125945241199478,
-      "grad_norm": 1.4647471904754639,
-      "learning_rate": 7.521507537688443e-05,
-      "loss": 5.4267,
-      "step": 25169
-    },
-    {
-      "epoch": 13.126466753585397,
-      "grad_norm": 1.4680771827697754,
-      "learning_rate": 7.52140703517588e-05,
-      "loss": 5.2132,
-      "step": 25170
-    },
-    {
-      "epoch": 13.126988265971317,
-      "grad_norm": 1.3946707248687744,
-      "learning_rate": 7.521306532663317e-05,
-      "loss": 5.6097,
-      "step": 25171
-    },
-    {
-      "epoch": 13.127509778357236,
-      "grad_norm": 1.4917488098144531,
-      "learning_rate": 7.521206030150753e-05,
-      "loss": 5.4537,
-      "step": 25172
-    },
-    {
-      "epoch": 13.128031290743156,
-      "grad_norm": 1.3621923923492432,
-      "learning_rate": 7.521105527638191e-05,
-      "loss": 5.6704,
-      "step": 25173
-    },
-    {
-      "epoch": 13.128552803129073,
-      "grad_norm": 1.4501692056655884,
-      "learning_rate": 7.521005025125629e-05,
-      "loss": 5.5443,
-      "step": 25174
-    },
-    {
-      "epoch": 13.129074315514993,
-      "grad_norm": 1.5247628688812256,
-      "learning_rate": 7.520904522613067e-05,
-      "loss": 5.2935,
-      "step": 25175
-    },
-    {
-      "epoch": 13.129595827900912,
-      "grad_norm": 1.421472191810608,
-      "learning_rate": 7.520804020100503e-05,
-      "loss": 5.3774,
-      "step": 25176
-    },
-    {
-      "epoch": 13.130117340286832,
-      "grad_norm": 1.4811326265335083,
-      "learning_rate": 7.520703517587941e-05,
-      "loss": 4.9176,
-      "step": 25177
-    },
-    {
-      "epoch": 13.130638852672751,
-      "grad_norm": 1.3955714702606201,
-      "learning_rate": 7.520603015075377e-05,
-      "loss": 5.4488,
-      "step": 25178
-    },
-    {
-      "epoch": 13.13116036505867,
-      "grad_norm": 1.464870572090149,
-      "learning_rate": 7.520502512562815e-05,
-      "loss": 5.255,
-      "step": 25179
-    },
-    {
-      "epoch": 13.131681877444588,
-      "grad_norm": 1.5355099439620972,
-      "learning_rate": 7.520402010050251e-05,
-      "loss": 5.7683,
-      "step": 25180
-    },
-    {
-      "epoch": 13.132203389830508,
-      "grad_norm": 1.6420077085494995,
-      "learning_rate": 7.520301507537689e-05,
-      "loss": 4.7608,
-      "step": 25181
-    },
-    {
-      "epoch": 13.132724902216427,
-      "grad_norm": 1.368011713027954,
-      "learning_rate": 7.520201005025126e-05,
-      "loss": 5.4534,
-      "step": 25182
-    },
-    {
-      "epoch": 13.133246414602347,
-      "grad_norm": 1.3707728385925293,
-      "learning_rate": 7.520100502512563e-05,
-      "loss": 5.4152,
-      "step": 25183
-    },
-    {
-      "epoch": 13.133767926988266,
-      "grad_norm": 1.4228885173797607,
-      "learning_rate": 7.52e-05,
-      "loss": 5.854,
-      "step": 25184
-    },
-    {
-      "epoch": 13.134289439374186,
-      "grad_norm": 1.5545129776000977,
-      "learning_rate": 7.519899497487438e-05,
-      "loss": 5.5057,
-      "step": 25185
-    },
-    {
-      "epoch": 13.134810951760103,
-      "grad_norm": 1.3742518424987793,
-      "learning_rate": 7.519798994974875e-05,
-      "loss": 5.0162,
-      "step": 25186
-    },
-    {
-      "epoch": 13.135332464146023,
-      "grad_norm": 1.6018249988555908,
-      "learning_rate": 7.519698492462312e-05,
-      "loss": 4.8228,
-      "step": 25187
-    },
-    {
-      "epoch": 13.135853976531942,
-      "grad_norm": 1.474749207496643,
-      "learning_rate": 7.51959798994975e-05,
-      "loss": 5.449,
-      "step": 25188
-    },
-    {
-      "epoch": 13.136375488917862,
-      "grad_norm": 1.3731311559677124,
-      "learning_rate": 7.519497487437186e-05,
-      "loss": 5.4235,
-      "step": 25189
-    },
-    {
-      "epoch": 13.136897001303781,
-      "grad_norm": 1.4387822151184082,
-      "learning_rate": 7.519396984924624e-05,
-      "loss": 5.7949,
-      "step": 25190
-    },
-    {
-      "epoch": 13.1374185136897,
-      "grad_norm": 1.5263595581054688,
-      "learning_rate": 7.51929648241206e-05,
-      "loss": 5.4052,
-      "step": 25191
-    },
-    {
-      "epoch": 13.137940026075619,
-      "grad_norm": 1.5006766319274902,
-      "learning_rate": 7.519195979899498e-05,
-      "loss": 5.2654,
-      "step": 25192
-    },
-    {
-      "epoch": 13.138461538461538,
-      "grad_norm": 1.4632511138916016,
-      "learning_rate": 7.519095477386934e-05,
-      "loss": 5.4663,
-      "step": 25193
-    },
-    {
-      "epoch": 13.138983050847457,
-      "grad_norm": 1.5952000617980957,
-      "learning_rate": 7.518994974874372e-05,
-      "loss": 5.6914,
-      "step": 25194
-    },
-    {
-      "epoch": 13.139504563233377,
-      "grad_norm": 1.3941690921783447,
-      "learning_rate": 7.518894472361809e-05,
-      "loss": 5.4478,
-      "step": 25195
-    },
-    {
-      "epoch": 13.140026075619296,
-      "grad_norm": 1.495408296585083,
-      "learning_rate": 7.518793969849246e-05,
-      "loss": 5.4356,
-      "step": 25196
-    },
-    {
-      "epoch": 13.140547588005216,
-      "grad_norm": 1.5246511697769165,
-      "learning_rate": 7.518693467336684e-05,
-      "loss": 5.5778,
-      "step": 25197
-    },
-    {
-      "epoch": 13.141069100391134,
-      "grad_norm": 1.4134644269943237,
-      "learning_rate": 7.51859296482412e-05,
-      "loss": 5.5395,
-      "step": 25198
-    },
-    {
-      "epoch": 13.141590612777053,
-      "grad_norm": 1.4303315877914429,
-      "learning_rate": 7.518492462311558e-05,
-      "loss": 5.0463,
-      "step": 25199
-    },
-    {
-      "epoch": 13.142112125162972,
-      "grad_norm": 1.4206912517547607,
-      "learning_rate": 7.518391959798995e-05,
-      "loss": 5.5757,
-      "step": 25200
-    },
-    {
-      "epoch": 13.142633637548892,
-      "grad_norm": 1.3214962482452393,
-      "learning_rate": 7.518291457286433e-05,
-      "loss": 5.9257,
-      "step": 25201
-    },
-    {
-      "epoch": 13.143155149934811,
-      "grad_norm": 1.439445972442627,
-      "learning_rate": 7.518190954773869e-05,
-      "loss": 4.849,
-      "step": 25202
-    },
-    {
-      "epoch": 13.14367666232073,
-      "grad_norm": 1.5055605173110962,
-      "learning_rate": 7.518090452261307e-05,
-      "loss": 5.3929,
-      "step": 25203
-    },
-    {
-      "epoch": 13.144198174706649,
-      "grad_norm": 1.4331488609313965,
-      "learning_rate": 7.517989949748743e-05,
-      "loss": 5.3445,
-      "step": 25204
-    },
-    {
-      "epoch": 13.144719687092568,
-      "grad_norm": 1.379417896270752,
-      "learning_rate": 7.517889447236181e-05,
-      "loss": 5.2591,
-      "step": 25205
-    },
-    {
-      "epoch": 13.145241199478487,
-      "grad_norm": 1.4135205745697021,
-      "learning_rate": 7.517788944723619e-05,
-      "loss": 5.4163,
-      "step": 25206
-    },
-    {
-      "epoch": 13.145762711864407,
-      "grad_norm": 1.3682451248168945,
-      "learning_rate": 7.517688442211057e-05,
-      "loss": 5.5839,
-      "step": 25207
-    },
-    {
-      "epoch": 13.146284224250326,
-      "grad_norm": 1.4360519647598267,
-      "learning_rate": 7.517587939698493e-05,
-      "loss": 5.5795,
-      "step": 25208
-    },
-    {
-      "epoch": 13.146805736636246,
-      "grad_norm": 1.602061152458191,
-      "learning_rate": 7.517487437185931e-05,
-      "loss": 4.9595,
-      "step": 25209
-    },
-    {
-      "epoch": 13.147327249022164,
-      "grad_norm": 1.447446346282959,
-      "learning_rate": 7.517386934673367e-05,
-      "loss": 5.5864,
-      "step": 25210
-    },
-    {
-      "epoch": 13.147848761408083,
-      "grad_norm": 1.464275598526001,
-      "learning_rate": 7.517286432160804e-05,
-      "loss": 5.2592,
-      "step": 25211
-    },
-    {
-      "epoch": 13.148370273794002,
-      "grad_norm": 1.403616189956665,
-      "learning_rate": 7.517185929648241e-05,
-      "loss": 5.6849,
-      "step": 25212
-    },
-    {
-      "epoch": 13.148891786179922,
-      "grad_norm": 1.4978772401809692,
-      "learning_rate": 7.517085427135678e-05,
-      "loss": 5.0599,
-      "step": 25213
-    },
-    {
-      "epoch": 13.149413298565841,
-      "grad_norm": 1.446114420890808,
-      "learning_rate": 7.516984924623116e-05,
-      "loss": 5.1957,
-      "step": 25214
-    },
-    {
-      "epoch": 13.14993481095176,
-      "grad_norm": 1.3357371091842651,
-      "learning_rate": 7.516884422110552e-05,
-      "loss": 5.2748,
-      "step": 25215
-    },
-    {
-      "epoch": 13.150456323337679,
-      "grad_norm": 1.348413348197937,
-      "learning_rate": 7.51678391959799e-05,
-      "loss": 5.4667,
-      "step": 25216
-    },
-    {
-      "epoch": 13.150977835723598,
-      "grad_norm": 1.4176950454711914,
-      "learning_rate": 7.516683417085428e-05,
-      "loss": 5.6824,
-      "step": 25217
-    },
-    {
-      "epoch": 13.151499348109517,
-      "grad_norm": 1.5069847106933594,
-      "learning_rate": 7.516582914572865e-05,
-      "loss": 5.2248,
-      "step": 25218
-    },
-    {
-      "epoch": 13.152020860495437,
-      "grad_norm": 1.639671802520752,
-      "learning_rate": 7.516482412060302e-05,
-      "loss": 5.3382,
-      "step": 25219
-    },
-    {
-      "epoch": 13.152542372881356,
-      "grad_norm": 1.4268959760665894,
-      "learning_rate": 7.51638190954774e-05,
-      "loss": 4.6472,
-      "step": 25220
-    },
-    {
-      "epoch": 13.153063885267276,
-      "grad_norm": 1.5120863914489746,
-      "learning_rate": 7.516281407035176e-05,
-      "loss": 5.1403,
-      "step": 25221
-    },
-    {
-      "epoch": 13.153585397653194,
-      "grad_norm": 1.3998470306396484,
-      "learning_rate": 7.516180904522614e-05,
-      "loss": 5.1004,
-      "step": 25222
-    },
-    {
-      "epoch": 13.154106910039113,
-      "grad_norm": 1.5750643014907837,
-      "learning_rate": 7.51608040201005e-05,
-      "loss": 5.4682,
-      "step": 25223
-    },
-    {
-      "epoch": 13.154628422425032,
-      "grad_norm": 1.4712989330291748,
-      "learning_rate": 7.515979899497487e-05,
-      "loss": 5.1003,
-      "step": 25224
-    },
-    {
-      "epoch": 13.155149934810952,
-      "grad_norm": 1.5370848178863525,
-      "learning_rate": 7.515879396984925e-05,
-      "loss": 5.2259,
-      "step": 25225
-    },
-    {
-      "epoch": 13.155671447196871,
-      "grad_norm": 1.4099410772323608,
-      "learning_rate": 7.515778894472362e-05,
-      "loss": 5.6413,
-      "step": 25226
-    },
-    {
-      "epoch": 13.156192959582791,
-      "grad_norm": 1.4605542421340942,
-      "learning_rate": 7.5156783919598e-05,
-      "loss": 5.514,
-      "step": 25227
-    },
-    {
-      "epoch": 13.156714471968709,
-      "grad_norm": 1.4837132692337036,
-      "learning_rate": 7.515577889447237e-05,
-      "loss": 5.1766,
-      "step": 25228
-    },
-    {
-      "epoch": 13.157235984354628,
-      "grad_norm": 1.7435322999954224,
-      "learning_rate": 7.515477386934674e-05,
-      "loss": 5.1242,
-      "step": 25229
-    },
-    {
-      "epoch": 13.157757496740548,
-      "grad_norm": 1.446797490119934,
-      "learning_rate": 7.515376884422111e-05,
-      "loss": 5.6827,
-      "step": 25230
-    },
-    {
-      "epoch": 13.158279009126467,
-      "grad_norm": 1.5141786336898804,
-      "learning_rate": 7.515276381909549e-05,
-      "loss": 5.4957,
-      "step": 25231
-    },
-    {
-      "epoch": 13.158800521512386,
-      "grad_norm": 1.5257445573806763,
-      "learning_rate": 7.515175879396985e-05,
-      "loss": 5.4489,
-      "step": 25232
-    },
-    {
-      "epoch": 13.159322033898306,
-      "grad_norm": 1.4821162223815918,
-      "learning_rate": 7.515075376884423e-05,
-      "loss": 4.9729,
-      "step": 25233
-    },
-    {
-      "epoch": 13.159843546284224,
-      "grad_norm": 1.4828994274139404,
-      "learning_rate": 7.514974874371859e-05,
-      "loss": 5.5719,
-      "step": 25234
-    },
-    {
-      "epoch": 13.160365058670143,
-      "grad_norm": 1.4445174932479858,
-      "learning_rate": 7.514874371859297e-05,
-      "loss": 5.5621,
-      "step": 25235
-    },
-    {
-      "epoch": 13.160886571056063,
-      "grad_norm": 1.6115525960922241,
-      "learning_rate": 7.514773869346733e-05,
-      "loss": 4.5553,
-      "step": 25236
-    },
-    {
-      "epoch": 13.161408083441982,
-      "grad_norm": 1.5476309061050415,
-      "learning_rate": 7.514673366834171e-05,
-      "loss": 5.5753,
-      "step": 25237
-    },
-    {
-      "epoch": 13.161929595827901,
-      "grad_norm": 1.367323637008667,
-      "learning_rate": 7.514572864321609e-05,
-      "loss": 5.1424,
-      "step": 25238
-    },
-    {
-      "epoch": 13.162451108213821,
-      "grad_norm": 1.524544358253479,
-      "learning_rate": 7.514472361809045e-05,
-      "loss": 5.5115,
-      "step": 25239
-    },
-    {
-      "epoch": 13.162972620599739,
-      "grad_norm": 1.5763441324234009,
-      "learning_rate": 7.514371859296483e-05,
-      "loss": 5.135,
-      "step": 25240
-    },
-    {
-      "epoch": 13.163494132985658,
-      "grad_norm": 1.365749716758728,
-      "learning_rate": 7.51427135678392e-05,
-      "loss": 5.4766,
-      "step": 25241
-    },
-    {
-      "epoch": 13.164015645371578,
-      "grad_norm": 1.470178484916687,
-      "learning_rate": 7.514170854271357e-05,
-      "loss": 5.5999,
-      "step": 25242
-    },
-    {
-      "epoch": 13.164537157757497,
-      "grad_norm": 1.6216228008270264,
-      "learning_rate": 7.514070351758794e-05,
-      "loss": 5.0079,
-      "step": 25243
-    },
-    {
-      "epoch": 13.165058670143416,
-      "grad_norm": 1.4924181699752808,
-      "learning_rate": 7.513969849246232e-05,
-      "loss": 4.8182,
-      "step": 25244
-    },
-    {
-      "epoch": 13.165580182529336,
-      "grad_norm": 1.4343562126159668,
-      "learning_rate": 7.513869346733668e-05,
-      "loss": 5.2449,
-      "step": 25245
-    },
-    {
-      "epoch": 13.166101694915254,
-      "grad_norm": 1.4974836111068726,
-      "learning_rate": 7.513768844221106e-05,
-      "loss": 5.8123,
-      "step": 25246
-    },
-    {
-      "epoch": 13.166623207301173,
-      "grad_norm": 1.3701577186584473,
-      "learning_rate": 7.513668341708544e-05,
-      "loss": 5.7129,
-      "step": 25247
-    },
-    {
-      "epoch": 13.167144719687093,
-      "grad_norm": 1.423564076423645,
-      "learning_rate": 7.513567839195981e-05,
-      "loss": 5.6504,
-      "step": 25248
-    },
-    {
-      "epoch": 13.167666232073012,
-      "grad_norm": 1.4094010591506958,
-      "learning_rate": 7.513467336683418e-05,
-      "loss": 5.6017,
-      "step": 25249
-    },
-    {
-      "epoch": 13.168187744458931,
-      "grad_norm": 1.47555673122406,
-      "learning_rate": 7.513366834170856e-05,
-      "loss": 5.4155,
-      "step": 25250
-    },
-    {
-      "epoch": 13.16870925684485,
-      "grad_norm": 1.4619511365890503,
-      "learning_rate": 7.513266331658292e-05,
-      "loss": 5.4583,
-      "step": 25251
-    },
-    {
-      "epoch": 13.169230769230769,
-      "grad_norm": 1.468177080154419,
-      "learning_rate": 7.513165829145728e-05,
-      "loss": 5.5679,
-      "step": 25252
-    },
-    {
-      "epoch": 13.169752281616688,
-      "grad_norm": 1.4287903308868408,
-      "learning_rate": 7.513065326633166e-05,
-      "loss": 5.6069,
-      "step": 25253
-    },
-    {
-      "epoch": 13.170273794002608,
-      "grad_norm": 1.3606998920440674,
-      "learning_rate": 7.512964824120603e-05,
-      "loss": 5.5437,
-      "step": 25254
-    },
-    {
-      "epoch": 13.170795306388527,
-      "grad_norm": 1.4613357782363892,
-      "learning_rate": 7.51286432160804e-05,
-      "loss": 5.1932,
-      "step": 25255
-    },
-    {
-      "epoch": 13.171316818774446,
-      "grad_norm": 1.461578369140625,
-      "learning_rate": 7.512763819095477e-05,
-      "loss": 5.3027,
-      "step": 25256
-    },
-    {
-      "epoch": 13.171838331160366,
-      "grad_norm": 1.492972731590271,
-      "learning_rate": 7.512663316582915e-05,
-      "loss": 4.9425,
-      "step": 25257
-    },
-    {
-      "epoch": 13.172359843546284,
-      "grad_norm": 1.5285316705703735,
-      "learning_rate": 7.512562814070352e-05,
-      "loss": 5.5542,
-      "step": 25258
-    },
-    {
-      "epoch": 13.172881355932203,
-      "grad_norm": 1.4932959079742432,
-      "learning_rate": 7.51246231155779e-05,
-      "loss": 4.8506,
-      "step": 25259
-    },
-    {
-      "epoch": 13.173402868318123,
-      "grad_norm": 1.439805269241333,
-      "learning_rate": 7.512361809045227e-05,
-      "loss": 5.4235,
-      "step": 25260
-    },
-    {
-      "epoch": 13.173924380704042,
-      "grad_norm": 1.4999237060546875,
-      "learning_rate": 7.512261306532664e-05,
-      "loss": 5.143,
-      "step": 25261
-    },
-    {
-      "epoch": 13.174445893089962,
-      "grad_norm": 1.5481159687042236,
-      "learning_rate": 7.512160804020101e-05,
-      "loss": 5.6293,
-      "step": 25262
-    },
-    {
-      "epoch": 13.17496740547588,
-      "grad_norm": 1.388663649559021,
-      "learning_rate": 7.512060301507539e-05,
-      "loss": 4.707,
-      "step": 25263
-    },
-    {
-      "epoch": 13.175488917861799,
-      "grad_norm": 1.475832462310791,
-      "learning_rate": 7.511959798994975e-05,
-      "loss": 5.4087,
-      "step": 25264
-    },
-    {
-      "epoch": 13.176010430247718,
-      "grad_norm": 1.4332634210586548,
-      "learning_rate": 7.511859296482411e-05,
-      "loss": 5.2103,
-      "step": 25265
-    },
-    {
-      "epoch": 13.176531942633638,
-      "grad_norm": 1.54741632938385,
-      "learning_rate": 7.511758793969849e-05,
-      "loss": 5.6008,
-      "step": 25266
-    },
-    {
-      "epoch": 13.177053455019557,
-      "grad_norm": 1.554762601852417,
-      "learning_rate": 7.511658291457287e-05,
-      "loss": 5.2034,
-      "step": 25267
-    },
-    {
-      "epoch": 13.177574967405477,
-      "grad_norm": 1.3685431480407715,
-      "learning_rate": 7.511557788944725e-05,
-      "loss": 5.6397,
-      "step": 25268
-    },
-    {
-      "epoch": 13.178096479791394,
-      "grad_norm": 1.3834539651870728,
-      "learning_rate": 7.511457286432161e-05,
-      "loss": 5.5578,
-      "step": 25269
-    },
-    {
-      "epoch": 13.178617992177314,
-      "grad_norm": 1.4845693111419678,
-      "learning_rate": 7.511356783919599e-05,
-      "loss": 5.3625,
-      "step": 25270
-    },
-    {
-      "epoch": 13.179139504563233,
-      "grad_norm": 1.409140706062317,
-      "learning_rate": 7.511256281407035e-05,
-      "loss": 5.4328,
-      "step": 25271
-    },
-    {
-      "epoch": 13.179661016949153,
-      "grad_norm": 1.583441972732544,
-      "learning_rate": 7.511155778894473e-05,
-      "loss": 4.9113,
-      "step": 25272
-    },
-    {
-      "epoch": 13.180182529335072,
-      "grad_norm": 1.3735960721969604,
-      "learning_rate": 7.51105527638191e-05,
-      "loss": 5.5509,
-      "step": 25273
-    },
-    {
-      "epoch": 13.180704041720992,
-      "grad_norm": 1.5326156616210938,
-      "learning_rate": 7.510954773869347e-05,
-      "loss": 5.3682,
-      "step": 25274
-    },
-    {
-      "epoch": 13.18122555410691,
-      "grad_norm": 1.471601128578186,
-      "learning_rate": 7.510854271356784e-05,
-      "loss": 5.291,
-      "step": 25275
-    },
-    {
-      "epoch": 13.181747066492829,
-      "grad_norm": 1.4952102899551392,
-      "learning_rate": 7.510753768844222e-05,
-      "loss": 5.2322,
-      "step": 25276
-    },
-    {
-      "epoch": 13.182268578878748,
-      "grad_norm": 1.4075672626495361,
-      "learning_rate": 7.510653266331658e-05,
-      "loss": 5.5866,
-      "step": 25277
-    },
-    {
-      "epoch": 13.182790091264668,
-      "grad_norm": 1.411667823791504,
-      "learning_rate": 7.510552763819096e-05,
-      "loss": 5.1595,
-      "step": 25278
-    },
-    {
-      "epoch": 13.183311603650587,
-      "grad_norm": 1.453439474105835,
-      "learning_rate": 7.510452261306534e-05,
-      "loss": 5.4423,
-      "step": 25279
-    },
-    {
-      "epoch": 13.183833116036507,
-      "grad_norm": 1.4613054990768433,
-      "learning_rate": 7.51035175879397e-05,
-      "loss": 4.9892,
-      "step": 25280
-    },
-    {
-      "epoch": 13.184354628422424,
-      "grad_norm": 1.3879953622817993,
-      "learning_rate": 7.510251256281408e-05,
-      "loss": 5.6154,
-      "step": 25281
-    },
-    {
-      "epoch": 13.184876140808344,
-      "grad_norm": 1.309019684791565,
-      "learning_rate": 7.510150753768844e-05,
-      "loss": 5.4199,
-      "step": 25282
-    },
-    {
-      "epoch": 13.185397653194263,
-      "grad_norm": 1.4620929956436157,
-      "learning_rate": 7.510050251256282e-05,
-      "loss": 4.8847,
-      "step": 25283
-    },
-    {
-      "epoch": 13.185919165580183,
-      "grad_norm": 1.3729901313781738,
-      "learning_rate": 7.509949748743718e-05,
-      "loss": 5.8207,
-      "step": 25284
-    },
-    {
-      "epoch": 13.186440677966102,
-      "grad_norm": 1.628687858581543,
-      "learning_rate": 7.509849246231156e-05,
-      "loss": 4.8705,
-      "step": 25285
-    },
-    {
-      "epoch": 13.186962190352022,
-      "grad_norm": 1.481257677078247,
-      "learning_rate": 7.509748743718593e-05,
-      "loss": 5.4343,
-      "step": 25286
-    },
-    {
-      "epoch": 13.18748370273794,
-      "grad_norm": 1.4264168739318848,
-      "learning_rate": 7.50964824120603e-05,
-      "loss": 5.2649,
-      "step": 25287
-    },
-    {
-      "epoch": 13.188005215123859,
-      "grad_norm": 1.3662102222442627,
-      "learning_rate": 7.509547738693468e-05,
-      "loss": 5.8929,
-      "step": 25288
-    },
-    {
-      "epoch": 13.188526727509778,
-      "grad_norm": 1.4284721612930298,
-      "learning_rate": 7.509447236180906e-05,
-      "loss": 5.8247,
-      "step": 25289
-    },
-    {
-      "epoch": 13.189048239895698,
-      "grad_norm": 1.5036088228225708,
-      "learning_rate": 7.509346733668342e-05,
-      "loss": 4.8536,
-      "step": 25290
-    },
-    {
-      "epoch": 13.189569752281617,
-      "grad_norm": 1.3249458074569702,
-      "learning_rate": 7.509246231155779e-05,
-      "loss": 5.895,
-      "step": 25291
-    },
-    {
-      "epoch": 13.190091264667537,
-      "grad_norm": 1.520720362663269,
-      "learning_rate": 7.509145728643217e-05,
-      "loss": 5.2981,
-      "step": 25292
-    },
-    {
-      "epoch": 13.190612777053454,
-      "grad_norm": 1.4117666482925415,
-      "learning_rate": 7.509045226130653e-05,
-      "loss": 5.212,
-      "step": 25293
-    },
-    {
-      "epoch": 13.191134289439374,
-      "grad_norm": 1.494834065437317,
-      "learning_rate": 7.508944723618091e-05,
-      "loss": 5.2485,
-      "step": 25294
-    },
-    {
-      "epoch": 13.191655801825293,
-      "grad_norm": 1.5160597562789917,
-      "learning_rate": 7.508844221105527e-05,
-      "loss": 5.4866,
-      "step": 25295
-    },
-    {
-      "epoch": 13.192177314211213,
-      "grad_norm": 1.5150141716003418,
-      "learning_rate": 7.508743718592965e-05,
-      "loss": 5.521,
-      "step": 25296
-    },
-    {
-      "epoch": 13.192698826597132,
-      "grad_norm": 1.477573275566101,
-      "learning_rate": 7.508643216080402e-05,
-      "loss": 5.4645,
-      "step": 25297
-    },
-    {
-      "epoch": 13.193220338983052,
-      "grad_norm": 1.4478883743286133,
-      "learning_rate": 7.508542713567839e-05,
-      "loss": 5.6425,
-      "step": 25298
-    },
-    {
-      "epoch": 13.19374185136897,
-      "grad_norm": 1.4807947874069214,
-      "learning_rate": 7.508442211055277e-05,
-      "loss": 5.3986,
-      "step": 25299
-    },
-    {
-      "epoch": 13.194263363754889,
-      "grad_norm": 1.466870903968811,
-      "learning_rate": 7.508341708542715e-05,
-      "loss": 5.6284,
-      "step": 25300
-    },
-    {
-      "epoch": 13.194784876140808,
-      "grad_norm": 1.518020510673523,
-      "learning_rate": 7.508241206030151e-05,
-      "loss": 5.6894,
-      "step": 25301
-    },
-    {
-      "epoch": 13.195306388526728,
-      "grad_norm": 1.4533865451812744,
-      "learning_rate": 7.508140703517589e-05,
-      "loss": 5.3161,
-      "step": 25302
-    },
-    {
-      "epoch": 13.195827900912647,
-      "grad_norm": 1.4525467157363892,
-      "learning_rate": 7.508040201005026e-05,
-      "loss": 5.6545,
-      "step": 25303
-    },
-    {
-      "epoch": 13.196349413298567,
-      "grad_norm": 1.4165788888931274,
-      "learning_rate": 7.507939698492462e-05,
-      "loss": 5.4238,
-      "step": 25304
-    },
-    {
-      "epoch": 13.196870925684484,
-      "grad_norm": 1.574947476387024,
-      "learning_rate": 7.5078391959799e-05,
-      "loss": 4.7021,
-      "step": 25305
-    },
-    {
-      "epoch": 13.197392438070404,
-      "grad_norm": 1.5506794452667236,
-      "learning_rate": 7.507738693467336e-05,
-      "loss": 4.9751,
-      "step": 25306
-    },
-    {
-      "epoch": 13.197913950456323,
-      "grad_norm": 1.5280377864837646,
-      "learning_rate": 7.507638190954774e-05,
-      "loss": 4.7964,
-      "step": 25307
-    },
-    {
-      "epoch": 13.198435462842243,
-      "grad_norm": 1.5694472789764404,
-      "learning_rate": 7.507537688442212e-05,
-      "loss": 5.2039,
-      "step": 25308
-    },
-    {
-      "epoch": 13.198956975228162,
-      "grad_norm": 1.4281989336013794,
-      "learning_rate": 7.50743718592965e-05,
-      "loss": 5.0129,
-      "step": 25309
-    },
-    {
-      "epoch": 13.199478487614082,
-      "grad_norm": 1.6032114028930664,
-      "learning_rate": 7.507336683417086e-05,
-      "loss": 5.0869,
-      "step": 25310
-    },
-    {
-      "epoch": 13.2,
-      "grad_norm": 1.5611605644226074,
-      "learning_rate": 7.507236180904524e-05,
-      "loss": 5.163,
-      "step": 25311
-    },
-    {
-      "epoch": 13.200521512385919,
-      "grad_norm": 1.5177199840545654,
-      "learning_rate": 7.50713567839196e-05,
-      "loss": 5.7383,
-      "step": 25312
-    },
-    {
-      "epoch": 13.201043024771838,
-      "grad_norm": 1.4981129169464111,
-      "learning_rate": 7.507035175879398e-05,
-      "loss": 5.1907,
-      "step": 25313
-    },
-    {
-      "epoch": 13.201564537157758,
-      "grad_norm": 1.5409106016159058,
-      "learning_rate": 7.506934673366834e-05,
-      "loss": 5.2101,
-      "step": 25314
-    },
-    {
-      "epoch": 13.202086049543677,
-      "grad_norm": 1.5703922510147095,
-      "learning_rate": 7.506834170854272e-05,
-      "loss": 5.1259,
-      "step": 25315
-    },
-    {
-      "epoch": 13.202607561929597,
-      "grad_norm": 1.4342584609985352,
-      "learning_rate": 7.506733668341709e-05,
-      "loss": 5.7659,
-      "step": 25316
-    },
-    {
-      "epoch": 13.203129074315514,
-      "grad_norm": 1.456813931465149,
-      "learning_rate": 7.506633165829145e-05,
-      "loss": 5.2367,
-      "step": 25317
-    },
-    {
-      "epoch": 13.203650586701434,
-      "grad_norm": 1.492476224899292,
-      "learning_rate": 7.506532663316583e-05,
-      "loss": 5.5118,
-      "step": 25318
-    },
-    {
-      "epoch": 13.204172099087353,
-      "grad_norm": 1.454817771911621,
-      "learning_rate": 7.50643216080402e-05,
-      "loss": 5.0384,
-      "step": 25319
-    },
-    {
-      "epoch": 13.204693611473273,
-      "grad_norm": 1.573767066001892,
-      "learning_rate": 7.506331658291458e-05,
-      "loss": 5.2877,
-      "step": 25320
-    },
-    {
-      "epoch": 13.205215123859192,
-      "grad_norm": 1.560809850692749,
-      "learning_rate": 7.506231155778895e-05,
-      "loss": 5.2358,
-      "step": 25321
-    },
-    {
-      "epoch": 13.205736636245112,
-      "grad_norm": 1.4942336082458496,
-      "learning_rate": 7.506130653266333e-05,
-      "loss": 5.0911,
-      "step": 25322
-    },
-    {
-      "epoch": 13.20625814863103,
-      "grad_norm": 1.330315113067627,
-      "learning_rate": 7.506030150753769e-05,
-      "loss": 4.8084,
-      "step": 25323
-    },
-    {
-      "epoch": 13.206779661016949,
-      "grad_norm": 1.4787858724594116,
-      "learning_rate": 7.505929648241207e-05,
-      "loss": 5.35,
-      "step": 25324
-    },
-    {
-      "epoch": 13.207301173402868,
-      "grad_norm": 1.4714666604995728,
-      "learning_rate": 7.505829145728643e-05,
-      "loss": 5.3506,
-      "step": 25325
-    },
-    {
-      "epoch": 13.207822685788788,
-      "grad_norm": 1.3703646659851074,
-      "learning_rate": 7.505728643216081e-05,
-      "loss": 5.4104,
-      "step": 25326
-    },
-    {
-      "epoch": 13.208344198174707,
-      "grad_norm": 1.3625128269195557,
-      "learning_rate": 7.505628140703517e-05,
-      "loss": 5.5092,
-      "step": 25327
-    },
-    {
-      "epoch": 13.208865710560627,
-      "grad_norm": 1.4992117881774902,
-      "learning_rate": 7.505527638190955e-05,
-      "loss": 5.6188,
-      "step": 25328
-    },
-    {
-      "epoch": 13.209387222946544,
-      "grad_norm": 1.4594110250473022,
-      "learning_rate": 7.505427135678393e-05,
-      "loss": 5.2019,
-      "step": 25329
-    },
-    {
-      "epoch": 13.209908735332464,
-      "grad_norm": 1.463136076927185,
-      "learning_rate": 7.50532663316583e-05,
-      "loss": 5.4361,
-      "step": 25330
-    },
-    {
-      "epoch": 13.210430247718383,
-      "grad_norm": 1.4826561212539673,
-      "learning_rate": 7.505226130653267e-05,
-      "loss": 5.4191,
-      "step": 25331
-    },
-    {
-      "epoch": 13.210951760104303,
-      "grad_norm": 1.4466357231140137,
-      "learning_rate": 7.505125628140704e-05,
-      "loss": 4.9051,
-      "step": 25332
-    },
-    {
-      "epoch": 13.211473272490222,
-      "grad_norm": 1.4381699562072754,
-      "learning_rate": 7.505025125628141e-05,
-      "loss": 5.2836,
-      "step": 25333
-    },
-    {
-      "epoch": 13.211994784876142,
-      "grad_norm": 1.4257413148880005,
-      "learning_rate": 7.504924623115578e-05,
-      "loss": 5.6269,
-      "step": 25334
-    },
-    {
-      "epoch": 13.21251629726206,
-      "grad_norm": 1.486267328262329,
-      "learning_rate": 7.504824120603016e-05,
-      "loss": 5.4114,
-      "step": 25335
-    },
-    {
-      "epoch": 13.213037809647979,
-      "grad_norm": 1.4078189134597778,
-      "learning_rate": 7.504723618090452e-05,
-      "loss": 5.4286,
-      "step": 25336
-    },
-    {
-      "epoch": 13.213559322033898,
-      "grad_norm": 1.406980276107788,
-      "learning_rate": 7.50462311557789e-05,
-      "loss": 5.5712,
-      "step": 25337
-    },
-    {
-      "epoch": 13.214080834419818,
-      "grad_norm": 1.371066927909851,
-      "learning_rate": 7.504522613065326e-05,
-      "loss": 5.4071,
-      "step": 25338
-    },
-    {
-      "epoch": 13.214602346805737,
-      "grad_norm": 1.5278620719909668,
-      "learning_rate": 7.504422110552764e-05,
-      "loss": 5.4969,
-      "step": 25339
-    },
-    {
-      "epoch": 13.215123859191657,
-      "grad_norm": 1.5154681205749512,
-      "learning_rate": 7.504321608040202e-05,
-      "loss": 5.5747,
-      "step": 25340
-    },
-    {
-      "epoch": 13.215645371577574,
-      "grad_norm": 1.3747785091400146,
-      "learning_rate": 7.50422110552764e-05,
-      "loss": 5.4525,
-      "step": 25341
-    },
-    {
-      "epoch": 13.216166883963494,
-      "grad_norm": 1.4558794498443604,
-      "learning_rate": 7.504120603015076e-05,
-      "loss": 5.3588,
-      "step": 25342
-    },
-    {
-      "epoch": 13.216688396349413,
-      "grad_norm": 1.4241397380828857,
-      "learning_rate": 7.504020100502514e-05,
-      "loss": 5.3524,
-      "step": 25343
-    },
-    {
-      "epoch": 13.217209908735333,
-      "grad_norm": 1.4441947937011719,
-      "learning_rate": 7.50391959798995e-05,
-      "loss": 5.4163,
-      "step": 25344
-    },
-    {
-      "epoch": 13.217731421121252,
-      "grad_norm": 1.4620333909988403,
-      "learning_rate": 7.503819095477387e-05,
-      "loss": 5.358,
-      "step": 25345
-    },
-    {
-      "epoch": 13.21825293350717,
-      "grad_norm": 1.6816033124923706,
-      "learning_rate": 7.503718592964824e-05,
-      "loss": 4.2642,
-      "step": 25346
-    },
-    {
-      "epoch": 13.21877444589309,
-      "grad_norm": 1.588183045387268,
-      "learning_rate": 7.503618090452261e-05,
-      "loss": 5.2059,
-      "step": 25347
-    },
-    {
-      "epoch": 13.219295958279009,
-      "grad_norm": 1.5693033933639526,
-      "learning_rate": 7.503517587939699e-05,
-      "loss": 5.8044,
-      "step": 25348
-    },
-    {
-      "epoch": 13.219817470664928,
-      "grad_norm": 1.4597482681274414,
-      "learning_rate": 7.503417085427135e-05,
-      "loss": 5.4203,
-      "step": 25349
-    },
-    {
-      "epoch": 13.220338983050848,
-      "grad_norm": 1.4356553554534912,
-      "learning_rate": 7.503316582914573e-05,
-      "loss": 5.4432,
-      "step": 25350
-    },
-    {
-      "epoch": 13.220860495436767,
-      "grad_norm": 1.4457191228866577,
-      "learning_rate": 7.50321608040201e-05,
-      "loss": 5.748,
-      "step": 25351
-    },
-    {
-      "epoch": 13.221382007822687,
-      "grad_norm": 1.3870149850845337,
-      "learning_rate": 7.503115577889448e-05,
-      "loss": 5.4931,
-      "step": 25352
-    },
-    {
-      "epoch": 13.221903520208604,
-      "grad_norm": 1.451483130455017,
-      "learning_rate": 7.503015075376885e-05,
-      "loss": 5.407,
-      "step": 25353
-    },
-    {
-      "epoch": 13.222425032594524,
-      "grad_norm": 1.5064964294433594,
-      "learning_rate": 7.502914572864323e-05,
-      "loss": 5.2266,
-      "step": 25354
-    },
-    {
-      "epoch": 13.222946544980443,
-      "grad_norm": 1.461282730102539,
-      "learning_rate": 7.502814070351759e-05,
-      "loss": 5.193,
-      "step": 25355
-    },
-    {
-      "epoch": 13.223468057366363,
-      "grad_norm": 1.5089267492294312,
-      "learning_rate": 7.502713567839197e-05,
-      "loss": 5.3562,
-      "step": 25356
-    },
-    {
-      "epoch": 13.223989569752282,
-      "grad_norm": 1.3725641965866089,
-      "learning_rate": 7.502613065326633e-05,
-      "loss": 5.6455,
-      "step": 25357
-    },
-    {
-      "epoch": 13.2245110821382,
-      "grad_norm": 1.394415020942688,
-      "learning_rate": 7.50251256281407e-05,
-      "loss": 5.5716,
-      "step": 25358
-    },
-    {
-      "epoch": 13.22503259452412,
-      "grad_norm": 1.3184003829956055,
-      "learning_rate": 7.502412060301507e-05,
-      "loss": 5.9532,
-      "step": 25359
-    },
-    {
-      "epoch": 13.225554106910039,
-      "grad_norm": 1.3919098377227783,
-      "learning_rate": 7.502311557788945e-05,
-      "loss": 5.7278,
-      "step": 25360
-    },
-    {
-      "epoch": 13.226075619295958,
-      "grad_norm": 1.3293943405151367,
-      "learning_rate": 7.502211055276383e-05,
-      "loss": 5.586,
-      "step": 25361
-    },
-    {
-      "epoch": 13.226597131681878,
-      "grad_norm": 1.4981939792633057,
-      "learning_rate": 7.50211055276382e-05,
-      "loss": 4.9895,
-      "step": 25362
-    },
-    {
-      "epoch": 13.227118644067797,
-      "grad_norm": 1.367714762687683,
-      "learning_rate": 7.502010050251257e-05,
-      "loss": 5.589,
-      "step": 25363
-    },
-    {
-      "epoch": 13.227640156453715,
-      "grad_norm": 1.5233795642852783,
-      "learning_rate": 7.501909547738694e-05,
-      "loss": 5.1297,
-      "step": 25364
-    },
-    {
-      "epoch": 13.228161668839634,
-      "grad_norm": 1.3977288007736206,
-      "learning_rate": 7.501809045226131e-05,
-      "loss": 5.3153,
-      "step": 25365
-    },
-    {
-      "epoch": 13.228683181225554,
-      "grad_norm": 1.5007826089859009,
-      "learning_rate": 7.501708542713568e-05,
-      "loss": 5.7703,
-      "step": 25366
-    },
-    {
-      "epoch": 13.229204693611473,
-      "grad_norm": 1.436737060546875,
-      "learning_rate": 7.501608040201006e-05,
-      "loss": 4.8979,
-      "step": 25367
-    },
-    {
-      "epoch": 13.229726205997393,
-      "grad_norm": 1.4515581130981445,
-      "learning_rate": 7.501507537688442e-05,
-      "loss": 5.2758,
-      "step": 25368
-    },
-    {
-      "epoch": 13.230247718383312,
-      "grad_norm": 1.560506820678711,
-      "learning_rate": 7.50140703517588e-05,
-      "loss": 5.3452,
-      "step": 25369
-    },
-    {
-      "epoch": 13.23076923076923,
-      "grad_norm": 1.4635729789733887,
-      "learning_rate": 7.501306532663316e-05,
-      "loss": 5.4294,
-      "step": 25370
-    },
-    {
-      "epoch": 13.23129074315515,
-      "grad_norm": 1.5501474142074585,
-      "learning_rate": 7.501206030150754e-05,
-      "loss": 5.6119,
-      "step": 25371
-    },
-    {
-      "epoch": 13.231812255541069,
-      "grad_norm": 1.4319087266921997,
-      "learning_rate": 7.501105527638192e-05,
-      "loss": 5.5077,
-      "step": 25372
-    },
-    {
-      "epoch": 13.232333767926988,
-      "grad_norm": 1.4985228776931763,
-      "learning_rate": 7.501005025125628e-05,
-      "loss": 4.9088,
-      "step": 25373
-    },
-    {
-      "epoch": 13.232855280312908,
-      "grad_norm": 1.5319523811340332,
-      "learning_rate": 7.500904522613066e-05,
-      "loss": 5.5647,
-      "step": 25374
-    },
-    {
-      "epoch": 13.233376792698827,
-      "grad_norm": 1.5170196294784546,
-      "learning_rate": 7.500804020100503e-05,
-      "loss": 4.7421,
-      "step": 25375
-    },
-    {
-      "epoch": 13.233898305084745,
-      "grad_norm": 1.56182861328125,
-      "learning_rate": 7.50070351758794e-05,
-      "loss": 4.8624,
-      "step": 25376
-    },
-    {
-      "epoch": 13.234419817470664,
-      "grad_norm": 1.3863128423690796,
-      "learning_rate": 7.500603015075377e-05,
-      "loss": 5.0806,
-      "step": 25377
-    },
-    {
-      "epoch": 13.234941329856584,
-      "grad_norm": 1.4402841329574585,
-      "learning_rate": 7.500502512562814e-05,
-      "loss": 5.6636,
-      "step": 25378
-    },
-    {
-      "epoch": 13.235462842242503,
-      "grad_norm": 1.5879734754562378,
-      "learning_rate": 7.500402010050251e-05,
-      "loss": 5.3308,
-      "step": 25379
-    },
-    {
-      "epoch": 13.235984354628423,
-      "grad_norm": 1.4170444011688232,
-      "learning_rate": 7.500301507537689e-05,
-      "loss": 5.7819,
-      "step": 25380
-    },
-    {
-      "epoch": 13.236505867014342,
-      "grad_norm": 1.4053277969360352,
-      "learning_rate": 7.500201005025126e-05,
-      "loss": 5.3469,
-      "step": 25381
-    },
-    {
-      "epoch": 13.23702737940026,
-      "grad_norm": 1.3464709520339966,
-      "learning_rate": 7.500100502512564e-05,
-      "loss": 5.641,
-      "step": 25382
-    },
-    {
-      "epoch": 13.23754889178618,
-      "grad_norm": 1.4172112941741943,
-      "learning_rate": 7.500000000000001e-05,
-      "loss": 5.5381,
-      "step": 25383
-    },
-    {
-      "epoch": 13.238070404172099,
-      "grad_norm": 1.3637399673461914,
-      "learning_rate": 7.499899497487437e-05,
-      "loss": 4.9254,
-      "step": 25384
-    },
-    {
-      "epoch": 13.238591916558018,
-      "grad_norm": 1.4713795185089111,
-      "learning_rate": 7.499798994974875e-05,
-      "loss": 5.7206,
-      "step": 25385
-    },
-    {
-      "epoch": 13.239113428943938,
-      "grad_norm": 1.6072629690170288,
-      "learning_rate": 7.499698492462311e-05,
-      "loss": 5.2498,
-      "step": 25386
-    },
-    {
-      "epoch": 13.239634941329857,
-      "grad_norm": 1.4919612407684326,
-      "learning_rate": 7.499597989949749e-05,
-      "loss": 5.301,
-      "step": 25387
-    },
-    {
-      "epoch": 13.240156453715775,
-      "grad_norm": 1.3701772689819336,
-      "learning_rate": 7.499497487437186e-05,
-      "loss": 5.3621,
-      "step": 25388
-    },
-    {
-      "epoch": 13.240677966101694,
-      "grad_norm": 1.5149915218353271,
-      "learning_rate": 7.499396984924623e-05,
-      "loss": 5.329,
-      "step": 25389
-    },
-    {
-      "epoch": 13.241199478487614,
-      "grad_norm": 1.341346025466919,
-      "learning_rate": 7.49929648241206e-05,
-      "loss": 5.7364,
-      "step": 25390
-    },
-    {
-      "epoch": 13.241720990873533,
-      "grad_norm": 1.3798847198486328,
-      "learning_rate": 7.499195979899498e-05,
-      "loss": 5.0959,
-      "step": 25391
-    },
-    {
-      "epoch": 13.242242503259453,
-      "grad_norm": 1.5736886262893677,
-      "learning_rate": 7.499095477386935e-05,
-      "loss": 5.0503,
-      "step": 25392
-    },
-    {
-      "epoch": 13.242764015645372,
-      "grad_norm": 1.4940001964569092,
-      "learning_rate": 7.498994974874373e-05,
-      "loss": 5.6135,
-      "step": 25393
-    },
-    {
-      "epoch": 13.24328552803129,
-      "grad_norm": 1.4454067945480347,
-      "learning_rate": 7.49889447236181e-05,
-      "loss": 5.3579,
-      "step": 25394
-    },
-    {
-      "epoch": 13.24380704041721,
-      "grad_norm": 1.5157781839370728,
-      "learning_rate": 7.498793969849247e-05,
-      "loss": 5.358,
-      "step": 25395
-    },
-    {
-      "epoch": 13.244328552803129,
-      "grad_norm": 1.4811928272247314,
-      "learning_rate": 7.498693467336684e-05,
-      "loss": 5.2597,
-      "step": 25396
-    },
-    {
-      "epoch": 13.244850065189048,
-      "grad_norm": 1.3369364738464355,
-      "learning_rate": 7.49859296482412e-05,
-      "loss": 5.8848,
-      "step": 25397
-    },
-    {
-      "epoch": 13.245371577574968,
-      "grad_norm": 1.4870104789733887,
-      "learning_rate": 7.498492462311558e-05,
-      "loss": 5.1615,
-      "step": 25398
-    },
-    {
-      "epoch": 13.245893089960887,
-      "grad_norm": 1.4334619045257568,
-      "learning_rate": 7.498391959798994e-05,
-      "loss": 5.412,
-      "step": 25399
-    },
-    {
-      "epoch": 13.246414602346805,
-      "grad_norm": 1.4014244079589844,
-      "learning_rate": 7.498291457286432e-05,
-      "loss": 5.4811,
-      "step": 25400
-    },
-    {
-      "epoch": 13.246936114732724,
-      "grad_norm": 1.560128927230835,
-      "learning_rate": 7.49819095477387e-05,
-      "loss": 5.3214,
-      "step": 25401
-    },
-    {
-      "epoch": 13.247457627118644,
-      "grad_norm": 1.4108891487121582,
-      "learning_rate": 7.498090452261308e-05,
-      "loss": 5.3792,
-      "step": 25402
-    },
-    {
-      "epoch": 13.247979139504563,
-      "grad_norm": 1.4482345581054688,
-      "learning_rate": 7.497989949748744e-05,
-      "loss": 5.5627,
-      "step": 25403
-    },
-    {
-      "epoch": 13.248500651890483,
-      "grad_norm": 1.469772219657898,
-      "learning_rate": 7.497889447236182e-05,
-      "loss": 5.1852,
-      "step": 25404
-    },
-    {
-      "epoch": 13.249022164276402,
-      "grad_norm": 1.4286941289901733,
-      "learning_rate": 7.497788944723618e-05,
-      "loss": 5.3633,
-      "step": 25405
-    },
-    {
-      "epoch": 13.24954367666232,
-      "grad_norm": 1.3124587535858154,
-      "learning_rate": 7.497688442211056e-05,
-      "loss": 5.2422,
-      "step": 25406
-    },
-    {
-      "epoch": 13.25006518904824,
-      "grad_norm": 1.4637466669082642,
-      "learning_rate": 7.497587939698493e-05,
-      "loss": 5.7955,
-      "step": 25407
-    },
-    {
-      "epoch": 13.250586701434159,
-      "grad_norm": 1.2904689311981201,
-      "learning_rate": 7.49748743718593e-05,
-      "loss": 5.6626,
-      "step": 25408
-    },
-    {
-      "epoch": 13.251108213820078,
-      "grad_norm": 1.3540548086166382,
-      "learning_rate": 7.497386934673367e-05,
-      "loss": 5.2314,
-      "step": 25409
-    },
-    {
-      "epoch": 13.251629726205998,
-      "grad_norm": 1.5281566381454468,
-      "learning_rate": 7.497286432160803e-05,
-      "loss": 5.127,
-      "step": 25410
-    },
-    {
-      "epoch": 13.252151238591917,
-      "grad_norm": 1.445682406425476,
-      "learning_rate": 7.497185929648241e-05,
-      "loss": 5.5945,
-      "step": 25411
-    },
-    {
-      "epoch": 13.252672750977835,
-      "grad_norm": 1.490857481956482,
-      "learning_rate": 7.497085427135679e-05,
-      "loss": 4.8647,
-      "step": 25412
-    },
-    {
-      "epoch": 13.253194263363755,
-      "grad_norm": 1.3601360321044922,
-      "learning_rate": 7.496984924623117e-05,
-      "loss": 5.7316,
-      "step": 25413
-    },
-    {
-      "epoch": 13.253715775749674,
-      "grad_norm": 1.5419914722442627,
-      "learning_rate": 7.496884422110553e-05,
-      "loss": 5.0059,
-      "step": 25414
-    },
-    {
-      "epoch": 13.254237288135593,
-      "grad_norm": 1.4919794797897339,
-      "learning_rate": 7.496783919597991e-05,
-      "loss": 4.798,
-      "step": 25415
-    },
-    {
-      "epoch": 13.254758800521513,
-      "grad_norm": 1.4225702285766602,
-      "learning_rate": 7.496683417085427e-05,
-      "loss": 5.6059,
-      "step": 25416
-    },
-    {
-      "epoch": 13.255280312907432,
-      "grad_norm": 1.4600328207015991,
-      "learning_rate": 7.496582914572865e-05,
-      "loss": 5.6937,
-      "step": 25417
-    },
-    {
-      "epoch": 13.25580182529335,
-      "grad_norm": 1.453451156616211,
-      "learning_rate": 7.496482412060301e-05,
-      "loss": 5.4528,
-      "step": 25418
-    },
-    {
-      "epoch": 13.25632333767927,
-      "grad_norm": 1.4664722681045532,
-      "learning_rate": 7.496381909547739e-05,
-      "loss": 5.2624,
-      "step": 25419
-    },
-    {
-      "epoch": 13.256844850065189,
-      "grad_norm": 1.3433711528778076,
-      "learning_rate": 7.496281407035176e-05,
-      "loss": 5.3387,
-      "step": 25420
-    },
-    {
-      "epoch": 13.257366362451108,
-      "grad_norm": 1.5245369672775269,
-      "learning_rate": 7.496180904522613e-05,
-      "loss": 5.3212,
-      "step": 25421
-    },
-    {
-      "epoch": 13.257887874837028,
-      "grad_norm": 1.396591305732727,
-      "learning_rate": 7.496080402010051e-05,
-      "loss": 5.4732,
-      "step": 25422
-    },
-    {
-      "epoch": 13.258409387222947,
-      "grad_norm": 1.4624009132385254,
-      "learning_rate": 7.495979899497488e-05,
-      "loss": 5.3436,
-      "step": 25423
-    },
-    {
-      "epoch": 13.258930899608865,
-      "grad_norm": 1.4881036281585693,
-      "learning_rate": 7.495879396984925e-05,
-      "loss": 5.6723,
-      "step": 25424
-    },
-    {
-      "epoch": 13.259452411994785,
-      "grad_norm": 1.444699764251709,
-      "learning_rate": 7.495778894472362e-05,
-      "loss": 5.5548,
-      "step": 25425
-    },
-    {
-      "epoch": 13.259973924380704,
-      "grad_norm": 1.483657956123352,
-      "learning_rate": 7.4956783919598e-05,
-      "loss": 5.7527,
-      "step": 25426
-    },
-    {
-      "epoch": 13.260495436766623,
-      "grad_norm": 1.5487092733383179,
-      "learning_rate": 7.495577889447236e-05,
-      "loss": 4.8058,
-      "step": 25427
-    },
-    {
-      "epoch": 13.261016949152543,
-      "grad_norm": 1.4776846170425415,
-      "learning_rate": 7.495477386934674e-05,
-      "loss": 5.5708,
-      "step": 25428
-    },
-    {
-      "epoch": 13.261538461538462,
-      "grad_norm": 1.4845448732376099,
-      "learning_rate": 7.49537688442211e-05,
-      "loss": 5.0636,
-      "step": 25429
-    },
-    {
-      "epoch": 13.26205997392438,
-      "grad_norm": 1.4320068359375,
-      "learning_rate": 7.495276381909548e-05,
-      "loss": 5.6014,
-      "step": 25430
-    },
-    {
-      "epoch": 13.2625814863103,
-      "grad_norm": 1.5208157300949097,
-      "learning_rate": 7.495175879396984e-05,
-      "loss": 5.3573,
-      "step": 25431
-    },
-    {
-      "epoch": 13.263102998696219,
-      "grad_norm": 1.4467675685882568,
-      "learning_rate": 7.495075376884422e-05,
-      "loss": 5.3684,
-      "step": 25432
-    },
-    {
-      "epoch": 13.263624511082138,
-      "grad_norm": 1.3831690549850464,
-      "learning_rate": 7.49497487437186e-05,
-      "loss": 5.4093,
-      "step": 25433
-    },
-    {
-      "epoch": 13.264146023468058,
-      "grad_norm": 1.4340338706970215,
-      "learning_rate": 7.494874371859298e-05,
-      "loss": 5.711,
-      "step": 25434
-    },
-    {
-      "epoch": 13.264667535853977,
-      "grad_norm": 1.487748146057129,
-      "learning_rate": 7.494773869346734e-05,
-      "loss": 5.3334,
-      "step": 25435
-    },
-    {
-      "epoch": 13.265189048239895,
-      "grad_norm": 1.6333342790603638,
-      "learning_rate": 7.494673366834172e-05,
-      "loss": 5.3091,
-      "step": 25436
-    },
-    {
-      "epoch": 13.265710560625815,
-      "grad_norm": 1.49601149559021,
-      "learning_rate": 7.494572864321608e-05,
-      "loss": 5.3201,
-      "step": 25437
-    },
-    {
-      "epoch": 13.266232073011734,
-      "grad_norm": 1.358510971069336,
-      "learning_rate": 7.494472361809045e-05,
-      "loss": 5.7732,
-      "step": 25438
-    },
-    {
-      "epoch": 13.266753585397653,
-      "grad_norm": 1.4911569356918335,
-      "learning_rate": 7.494371859296483e-05,
-      "loss": 5.4249,
-      "step": 25439
-    },
-    {
-      "epoch": 13.267275097783573,
-      "grad_norm": 1.6501615047454834,
-      "learning_rate": 7.494271356783919e-05,
-      "loss": 5.2455,
-      "step": 25440
-    },
-    {
-      "epoch": 13.26779661016949,
-      "grad_norm": 1.5945181846618652,
-      "learning_rate": 7.494170854271357e-05,
-      "loss": 5.4265,
-      "step": 25441
-    },
-    {
-      "epoch": 13.26831812255541,
-      "grad_norm": 1.5750492811203003,
-      "learning_rate": 7.494070351758795e-05,
-      "loss": 5.568,
-      "step": 25442
-    },
-    {
-      "epoch": 13.26883963494133,
-      "grad_norm": 1.5138028860092163,
-      "learning_rate": 7.493969849246232e-05,
-      "loss": 5.2356,
-      "step": 25443
-    },
-    {
-      "epoch": 13.269361147327249,
-      "grad_norm": 1.3969258069992065,
-      "learning_rate": 7.493869346733669e-05,
-      "loss": 5.4655,
-      "step": 25444
-    },
-    {
-      "epoch": 13.269882659713168,
-      "grad_norm": 1.5121150016784668,
-      "learning_rate": 7.493768844221107e-05,
-      "loss": 5.1723,
-      "step": 25445
-    },
-    {
-      "epoch": 13.270404172099088,
-      "grad_norm": 1.5728468894958496,
-      "learning_rate": 7.493668341708543e-05,
-      "loss": 5.2414,
-      "step": 25446
-    },
-    {
-      "epoch": 13.270925684485007,
-      "grad_norm": 1.387013554573059,
-      "learning_rate": 7.493567839195981e-05,
-      "loss": 5.7552,
-      "step": 25447
-    },
-    {
-      "epoch": 13.271447196870925,
-      "grad_norm": 1.639012336730957,
-      "learning_rate": 7.493467336683417e-05,
-      "loss": 5.1895,
-      "step": 25448
-    },
-    {
-      "epoch": 13.271968709256845,
-      "grad_norm": 1.4688339233398438,
-      "learning_rate": 7.493366834170855e-05,
-      "loss": 5.6309,
-      "step": 25449
-    },
-    {
-      "epoch": 13.272490221642764,
-      "grad_norm": 1.4740759134292603,
-      "learning_rate": 7.493266331658291e-05,
-      "loss": 5.5697,
-      "step": 25450
-    },
-    {
-      "epoch": 13.273011734028684,
-      "grad_norm": 1.387086272239685,
-      "learning_rate": 7.493165829145728e-05,
-      "loss": 5.1108,
-      "step": 25451
-    },
-    {
-      "epoch": 13.273533246414603,
-      "grad_norm": 1.4997508525848389,
-      "learning_rate": 7.493065326633166e-05,
-      "loss": 4.8153,
-      "step": 25452
-    },
-    {
-      "epoch": 13.27405475880052,
-      "grad_norm": 1.5837382078170776,
-      "learning_rate": 7.492964824120603e-05,
-      "loss": 5.4297,
-      "step": 25453
-    },
-    {
-      "epoch": 13.27457627118644,
-      "grad_norm": 1.3818271160125732,
-      "learning_rate": 7.492864321608041e-05,
-      "loss": 5.5643,
-      "step": 25454
-    },
-    {
-      "epoch": 13.27509778357236,
-      "grad_norm": 1.4828463792800903,
-      "learning_rate": 7.492763819095478e-05,
-      "loss": 5.3296,
-      "step": 25455
-    },
-    {
-      "epoch": 13.275619295958279,
-      "grad_norm": 1.43472421169281,
-      "learning_rate": 7.492663316582915e-05,
-      "loss": 5.6744,
-      "step": 25456
-    },
-    {
-      "epoch": 13.276140808344199,
-      "grad_norm": 1.4476120471954346,
-      "learning_rate": 7.492562814070352e-05,
-      "loss": 5.698,
-      "step": 25457
-    },
-    {
-      "epoch": 13.276662320730118,
-      "grad_norm": 1.4214569330215454,
-      "learning_rate": 7.49246231155779e-05,
-      "loss": 5.7965,
-      "step": 25458
-    },
-    {
-      "epoch": 13.277183833116036,
-      "grad_norm": 1.456536889076233,
-      "learning_rate": 7.492361809045226e-05,
-      "loss": 5.5261,
-      "step": 25459
-    },
-    {
-      "epoch": 13.277705345501955,
-      "grad_norm": 1.5149778127670288,
-      "learning_rate": 7.492261306532664e-05,
-      "loss": 4.8079,
-      "step": 25460
-    },
-    {
-      "epoch": 13.278226857887875,
-      "grad_norm": 1.4437638521194458,
-      "learning_rate": 7.4921608040201e-05,
-      "loss": 5.5186,
-      "step": 25461
-    },
-    {
-      "epoch": 13.278748370273794,
-      "grad_norm": 1.546217441558838,
-      "learning_rate": 7.492060301507538e-05,
-      "loss": 5.2631,
-      "step": 25462
-    },
-    {
-      "epoch": 13.279269882659714,
-      "grad_norm": 1.4815410375595093,
-      "learning_rate": 7.491959798994976e-05,
-      "loss": 4.3787,
-      "step": 25463
-    },
-    {
-      "epoch": 13.279791395045633,
-      "grad_norm": 1.4813884496688843,
-      "learning_rate": 7.491859296482412e-05,
-      "loss": 5.475,
-      "step": 25464
-    },
-    {
-      "epoch": 13.28031290743155,
-      "grad_norm": 1.5217376947402954,
-      "learning_rate": 7.49175879396985e-05,
-      "loss": 5.5502,
-      "step": 25465
-    },
-    {
-      "epoch": 13.28083441981747,
-      "grad_norm": 1.4435014724731445,
-      "learning_rate": 7.491658291457287e-05,
-      "loss": 5.6696,
-      "step": 25466
-    },
-    {
-      "epoch": 13.28135593220339,
-      "grad_norm": 1.399161696434021,
-      "learning_rate": 7.491557788944724e-05,
-      "loss": 5.5473,
-      "step": 25467
-    },
-    {
-      "epoch": 13.281877444589309,
-      "grad_norm": 1.4496208429336548,
-      "learning_rate": 7.491457286432161e-05,
-      "loss": 5.3645,
-      "step": 25468
-    },
-    {
-      "epoch": 13.282398956975229,
-      "grad_norm": 1.4534052610397339,
-      "learning_rate": 7.491356783919599e-05,
-      "loss": 5.688,
-      "step": 25469
-    },
-    {
-      "epoch": 13.282920469361148,
-      "grad_norm": 1.5404130220413208,
-      "learning_rate": 7.491256281407035e-05,
-      "loss": 5.6561,
-      "step": 25470
-    },
-    {
-      "epoch": 13.283441981747066,
-      "grad_norm": 1.6273983716964722,
-      "learning_rate": 7.491155778894473e-05,
-      "loss": 4.5418,
-      "step": 25471
-    },
-    {
-      "epoch": 13.283963494132985,
-      "grad_norm": 1.516241192817688,
-      "learning_rate": 7.491055276381909e-05,
-      "loss": 5.2987,
-      "step": 25472
-    },
-    {
-      "epoch": 13.284485006518905,
-      "grad_norm": 1.514333963394165,
-      "learning_rate": 7.490954773869347e-05,
-      "loss": 5.5389,
-      "step": 25473
-    },
-    {
-      "epoch": 13.285006518904824,
-      "grad_norm": 1.4123773574829102,
-      "learning_rate": 7.490854271356785e-05,
-      "loss": 5.5382,
-      "step": 25474
-    },
-    {
-      "epoch": 13.285528031290744,
-      "grad_norm": 1.5146337747573853,
-      "learning_rate": 7.490753768844223e-05,
-      "loss": 5.3066,
-      "step": 25475
-    },
-    {
-      "epoch": 13.286049543676663,
-      "grad_norm": 1.457753300666809,
-      "learning_rate": 7.490653266331659e-05,
-      "loss": 5.4247,
-      "step": 25476
-    },
-    {
-      "epoch": 13.28657105606258,
-      "grad_norm": 1.562856674194336,
-      "learning_rate": 7.490552763819095e-05,
-      "loss": 5.4172,
-      "step": 25477
-    },
-    {
-      "epoch": 13.2870925684485,
-      "grad_norm": 1.713830828666687,
-      "learning_rate": 7.490452261306533e-05,
-      "loss": 5.4161,
-      "step": 25478
-    },
-    {
-      "epoch": 13.28761408083442,
-      "grad_norm": 1.4195996522903442,
-      "learning_rate": 7.49035175879397e-05,
-      "loss": 5.4058,
-      "step": 25479
-    },
-    {
-      "epoch": 13.288135593220339,
-      "grad_norm": 1.4242713451385498,
-      "learning_rate": 7.490251256281407e-05,
-      "loss": 5.6569,
-      "step": 25480
-    },
-    {
-      "epoch": 13.288657105606259,
-      "grad_norm": 1.403554916381836,
-      "learning_rate": 7.490150753768844e-05,
-      "loss": 5.6674,
-      "step": 25481
-    },
-    {
-      "epoch": 13.289178617992178,
-      "grad_norm": 1.4927210807800293,
-      "learning_rate": 7.490050251256282e-05,
-      "loss": 5.5946,
-      "step": 25482
-    },
-    {
-      "epoch": 13.289700130378096,
-      "grad_norm": 1.5187278985977173,
-      "learning_rate": 7.48994974874372e-05,
-      "loss": 5.312,
-      "step": 25483
-    },
-    {
-      "epoch": 13.290221642764015,
-      "grad_norm": 1.4256985187530518,
-      "learning_rate": 7.489849246231157e-05,
-      "loss": 5.2557,
-      "step": 25484
-    },
-    {
-      "epoch": 13.290743155149935,
-      "grad_norm": 1.4232560396194458,
-      "learning_rate": 7.489748743718594e-05,
-      "loss": 5.6367,
-      "step": 25485
-    },
-    {
-      "epoch": 13.291264667535854,
-      "grad_norm": 1.4602237939834595,
-      "learning_rate": 7.489648241206031e-05,
-      "loss": 5.7261,
-      "step": 25486
-    },
-    {
-      "epoch": 13.291786179921774,
-      "grad_norm": 1.4894514083862305,
-      "learning_rate": 7.489547738693468e-05,
-      "loss": 5.429,
-      "step": 25487
-    },
-    {
-      "epoch": 13.292307692307693,
-      "grad_norm": 1.4317399263381958,
-      "learning_rate": 7.489447236180906e-05,
-      "loss": 5.6069,
-      "step": 25488
-    },
-    {
-      "epoch": 13.29282920469361,
-      "grad_norm": 1.4902106523513794,
-      "learning_rate": 7.489346733668342e-05,
-      "loss": 5.5808,
-      "step": 25489
-    },
-    {
-      "epoch": 13.29335071707953,
-      "grad_norm": 1.5316826105117798,
-      "learning_rate": 7.489246231155778e-05,
-      "loss": 5.651,
-      "step": 25490
-    },
-    {
-      "epoch": 13.29387222946545,
-      "grad_norm": 1.4146475791931152,
-      "learning_rate": 7.489145728643216e-05,
-      "loss": 5.7673,
-      "step": 25491
-    },
-    {
-      "epoch": 13.29439374185137,
-      "grad_norm": 1.4503220319747925,
-      "learning_rate": 7.489045226130653e-05,
-      "loss": 5.8194,
-      "step": 25492
-    },
-    {
-      "epoch": 13.294915254237289,
-      "grad_norm": 1.5079025030136108,
-      "learning_rate": 7.48894472361809e-05,
-      "loss": 5.5096,
-      "step": 25493
-    },
-    {
-      "epoch": 13.295436766623208,
-      "grad_norm": 1.4352809190750122,
-      "learning_rate": 7.488844221105528e-05,
-      "loss": 5.5479,
-      "step": 25494
-    },
-    {
-      "epoch": 13.295958279009126,
-      "grad_norm": 1.3879892826080322,
-      "learning_rate": 7.488743718592966e-05,
-      "loss": 5.4174,
-      "step": 25495
-    },
-    {
-      "epoch": 13.296479791395045,
-      "grad_norm": 1.5129196643829346,
-      "learning_rate": 7.488643216080402e-05,
-      "loss": 4.8313,
-      "step": 25496
-    },
-    {
-      "epoch": 13.297001303780965,
-      "grad_norm": 1.5429164171218872,
-      "learning_rate": 7.48854271356784e-05,
-      "loss": 5.0285,
-      "step": 25497
-    },
-    {
-      "epoch": 13.297522816166884,
-      "grad_norm": 1.3927971124649048,
-      "learning_rate": 7.488442211055277e-05,
-      "loss": 5.1617,
-      "step": 25498
-    },
-    {
-      "epoch": 13.298044328552804,
-      "grad_norm": 1.4243543148040771,
-      "learning_rate": 7.488341708542714e-05,
-      "loss": 5.0987,
-      "step": 25499
-    },
-    {
-      "epoch": 13.298565840938723,
-      "grad_norm": 1.442317247390747,
-      "learning_rate": 7.488241206030151e-05,
-      "loss": 5.7821,
-      "step": 25500
-    },
-    {
-      "epoch": 13.29908735332464,
-      "grad_norm": 1.4412719011306763,
-      "learning_rate": 7.488140703517589e-05,
-      "loss": 5.4878,
-      "step": 25501
-    },
-    {
-      "epoch": 13.29960886571056,
-      "grad_norm": 1.4520676136016846,
-      "learning_rate": 7.488040201005025e-05,
-      "loss": 5.3983,
-      "step": 25502
-    },
-    {
-      "epoch": 13.30013037809648,
-      "grad_norm": 1.5508865118026733,
-      "learning_rate": 7.487939698492463e-05,
-      "loss": 4.9832,
-      "step": 25503
-    },
-    {
-      "epoch": 13.3006518904824,
-      "grad_norm": 1.4898319244384766,
-      "learning_rate": 7.4878391959799e-05,
-      "loss": 4.9675,
-      "step": 25504
-    },
-    {
-      "epoch": 13.301173402868319,
-      "grad_norm": 1.3181625604629517,
-      "learning_rate": 7.487738693467337e-05,
-      "loss": 5.8454,
-      "step": 25505
-    },
-    {
-      "epoch": 13.301694915254238,
-      "grad_norm": 1.4271488189697266,
-      "learning_rate": 7.487638190954775e-05,
-      "loss": 5.6292,
-      "step": 25506
-    },
-    {
-      "epoch": 13.302216427640156,
-      "grad_norm": 1.4042320251464844,
-      "learning_rate": 7.487537688442211e-05,
-      "loss": 5.2591,
-      "step": 25507
-    },
-    {
-      "epoch": 13.302737940026075,
-      "grad_norm": 1.4861184358596802,
-      "learning_rate": 7.487437185929649e-05,
-      "loss": 5.3209,
-      "step": 25508
-    },
-    {
-      "epoch": 13.303259452411995,
-      "grad_norm": 1.5696898698806763,
-      "learning_rate": 7.487336683417085e-05,
-      "loss": 5.1774,
-      "step": 25509
-    },
-    {
-      "epoch": 13.303780964797914,
-      "grad_norm": 1.576454997062683,
-      "learning_rate": 7.487236180904523e-05,
-      "loss": 5.82,
-      "step": 25510
-    },
-    {
-      "epoch": 13.304302477183834,
-      "grad_norm": 1.4434504508972168,
-      "learning_rate": 7.48713567839196e-05,
-      "loss": 5.636,
-      "step": 25511
-    },
-    {
-      "epoch": 13.304823989569753,
-      "grad_norm": 1.4732856750488281,
-      "learning_rate": 7.487035175879397e-05,
-      "loss": 5.419,
-      "step": 25512
-    },
-    {
-      "epoch": 13.30534550195567,
-      "grad_norm": 1.5503524541854858,
-      "learning_rate": 7.486934673366834e-05,
-      "loss": 4.9895,
-      "step": 25513
-    },
-    {
-      "epoch": 13.30586701434159,
-      "grad_norm": 1.5306724309921265,
-      "learning_rate": 7.486834170854272e-05,
-      "loss": 5.0775,
-      "step": 25514
-    },
-    {
-      "epoch": 13.30638852672751,
-      "grad_norm": 1.4281803369522095,
-      "learning_rate": 7.48673366834171e-05,
-      "loss": 5.4563,
-      "step": 25515
-    },
-    {
-      "epoch": 13.30691003911343,
-      "grad_norm": 1.5258958339691162,
-      "learning_rate": 7.486633165829146e-05,
-      "loss": 5.4758,
-      "step": 25516
-    },
-    {
-      "epoch": 13.307431551499349,
-      "grad_norm": 1.337449073791504,
-      "learning_rate": 7.486532663316584e-05,
-      "loss": 4.4649,
-      "step": 25517
-    },
-    {
-      "epoch": 13.307953063885268,
-      "grad_norm": 1.4921514987945557,
-      "learning_rate": 7.48643216080402e-05,
-      "loss": 5.4881,
-      "step": 25518
-    },
-    {
-      "epoch": 13.308474576271186,
-      "grad_norm": 1.5289459228515625,
-      "learning_rate": 7.486331658291458e-05,
-      "loss": 5.0555,
-      "step": 25519
-    },
-    {
-      "epoch": 13.308996088657105,
-      "grad_norm": 1.4499554634094238,
-      "learning_rate": 7.486231155778894e-05,
-      "loss": 5.2757,
-      "step": 25520
-    },
-    {
-      "epoch": 13.309517601043025,
-      "grad_norm": 1.47807776927948,
-      "learning_rate": 7.486130653266332e-05,
-      "loss": 5.4779,
-      "step": 25521
-    },
-    {
-      "epoch": 13.310039113428944,
-      "grad_norm": 1.5748295783996582,
-      "learning_rate": 7.486030150753768e-05,
-      "loss": 5.6045,
-      "step": 25522
-    },
-    {
-      "epoch": 13.310560625814864,
-      "grad_norm": 1.446698546409607,
-      "learning_rate": 7.485929648241206e-05,
-      "loss": 5.596,
-      "step": 25523
-    },
-    {
-      "epoch": 13.311082138200783,
-      "grad_norm": 1.4704374074935913,
-      "learning_rate": 7.485829145728643e-05,
-      "loss": 5.7527,
-      "step": 25524
-    },
-    {
-      "epoch": 13.3116036505867,
-      "grad_norm": 1.5408073663711548,
-      "learning_rate": 7.48572864321608e-05,
-      "loss": 5.7377,
-      "step": 25525
-    },
-    {
-      "epoch": 13.31212516297262,
-      "grad_norm": 1.5659096240997314,
-      "learning_rate": 7.485628140703518e-05,
-      "loss": 5.3096,
-      "step": 25526
-    },
-    {
-      "epoch": 13.31264667535854,
-      "grad_norm": 1.5742545127868652,
-      "learning_rate": 7.485527638190956e-05,
-      "loss": 4.4117,
-      "step": 25527
-    },
-    {
-      "epoch": 13.31316818774446,
-      "grad_norm": 1.4061338901519775,
-      "learning_rate": 7.485427135678392e-05,
-      "loss": 5.4871,
-      "step": 25528
-    },
-    {
-      "epoch": 13.313689700130379,
-      "grad_norm": 1.5263487100601196,
-      "learning_rate": 7.48532663316583e-05,
-      "loss": 5.2008,
-      "step": 25529
-    },
-    {
-      "epoch": 13.314211212516298,
-      "grad_norm": 1.6418848037719727,
-      "learning_rate": 7.485226130653267e-05,
-      "loss": 5.4449,
-      "step": 25530
-    },
-    {
-      "epoch": 13.314732724902216,
-      "grad_norm": 1.5521506071090698,
-      "learning_rate": 7.485125628140703e-05,
-      "loss": 5.4927,
-      "step": 25531
-    },
-    {
-      "epoch": 13.315254237288135,
-      "grad_norm": 1.5622690916061401,
-      "learning_rate": 7.485025125628141e-05,
-      "loss": 5.6648,
-      "step": 25532
-    },
-    {
-      "epoch": 13.315775749674055,
-      "grad_norm": 1.3792476654052734,
-      "learning_rate": 7.484924623115577e-05,
-      "loss": 5.6607,
-      "step": 25533
-    },
-    {
-      "epoch": 13.316297262059974,
-      "grad_norm": 1.4701852798461914,
-      "learning_rate": 7.484824120603015e-05,
-      "loss": 5.5734,
-      "step": 25534
-    },
-    {
-      "epoch": 13.316818774445894,
-      "grad_norm": 1.487239956855774,
-      "learning_rate": 7.484723618090453e-05,
-      "loss": 5.1162,
-      "step": 25535
-    },
-    {
-      "epoch": 13.317340286831811,
-      "grad_norm": 1.7171566486358643,
-      "learning_rate": 7.48462311557789e-05,
-      "loss": 5.0515,
-      "step": 25536
-    },
-    {
-      "epoch": 13.31786179921773,
-      "grad_norm": 1.4166496992111206,
-      "learning_rate": 7.484522613065327e-05,
-      "loss": 5.7802,
-      "step": 25537
-    },
-    {
-      "epoch": 13.31838331160365,
-      "grad_norm": 1.4246013164520264,
-      "learning_rate": 7.484422110552765e-05,
-      "loss": 5.5315,
-      "step": 25538
-    },
-    {
-      "epoch": 13.31890482398957,
-      "grad_norm": 1.438037395477295,
-      "learning_rate": 7.484321608040201e-05,
-      "loss": 5.7734,
-      "step": 25539
-    },
-    {
-      "epoch": 13.31942633637549,
-      "grad_norm": 1.5860764980316162,
-      "learning_rate": 7.484221105527639e-05,
-      "loss": 5.1958,
-      "step": 25540
-    },
-    {
-      "epoch": 13.319947848761409,
-      "grad_norm": 1.4423797130584717,
-      "learning_rate": 7.484120603015076e-05,
-      "loss": 5.0965,
-      "step": 25541
-    },
-    {
-      "epoch": 13.320469361147328,
-      "grad_norm": 1.449641466140747,
-      "learning_rate": 7.484020100502513e-05,
-      "loss": 5.6765,
-      "step": 25542
-    },
-    {
-      "epoch": 13.320990873533246,
-      "grad_norm": 1.4187029600143433,
-      "learning_rate": 7.48391959798995e-05,
-      "loss": 5.3924,
-      "step": 25543
-    },
-    {
-      "epoch": 13.321512385919165,
-      "grad_norm": 1.4197331666946411,
-      "learning_rate": 7.483819095477386e-05,
-      "loss": 5.4747,
-      "step": 25544
-    },
-    {
-      "epoch": 13.322033898305085,
-      "grad_norm": 1.5158134698867798,
-      "learning_rate": 7.483718592964824e-05,
-      "loss": 5.2891,
-      "step": 25545
-    },
-    {
-      "epoch": 13.322555410691004,
-      "grad_norm": 1.570731520652771,
-      "learning_rate": 7.483618090452262e-05,
-      "loss": 4.6639,
-      "step": 25546
-    },
-    {
-      "epoch": 13.323076923076924,
-      "grad_norm": 1.4813400506973267,
-      "learning_rate": 7.4835175879397e-05,
-      "loss": 5.76,
-      "step": 25547
-    },
-    {
-      "epoch": 13.323598435462841,
-      "grad_norm": 1.591463565826416,
-      "learning_rate": 7.483417085427136e-05,
-      "loss": 5.0242,
-      "step": 25548
-    },
-    {
-      "epoch": 13.32411994784876,
-      "grad_norm": 1.4473334550857544,
-      "learning_rate": 7.483316582914574e-05,
-      "loss": 5.4688,
-      "step": 25549
-    },
-    {
-      "epoch": 13.32464146023468,
-      "grad_norm": 1.416987419128418,
-      "learning_rate": 7.48321608040201e-05,
-      "loss": 5.1433,
-      "step": 25550
-    },
-    {
-      "epoch": 13.3251629726206,
-      "grad_norm": 1.4789289236068726,
-      "learning_rate": 7.483115577889448e-05,
-      "loss": 5.5065,
-      "step": 25551
-    },
-    {
-      "epoch": 13.32568448500652,
-      "grad_norm": 1.623256802558899,
-      "learning_rate": 7.483015075376884e-05,
-      "loss": 5.3526,
-      "step": 25552
-    },
-    {
-      "epoch": 13.326205997392439,
-      "grad_norm": 1.3860317468643188,
-      "learning_rate": 7.482914572864322e-05,
-      "loss": 5.7319,
-      "step": 25553
-    },
-    {
-      "epoch": 13.326727509778356,
-      "grad_norm": 1.51033353805542,
-      "learning_rate": 7.482814070351759e-05,
-      "loss": 5.3128,
-      "step": 25554
-    },
-    {
-      "epoch": 13.327249022164276,
-      "grad_norm": 1.5349633693695068,
-      "learning_rate": 7.482713567839196e-05,
-      "loss": 5.1729,
-      "step": 25555
-    },
-    {
-      "epoch": 13.327770534550195,
-      "grad_norm": 1.5134656429290771,
-      "learning_rate": 7.482613065326634e-05,
-      "loss": 5.1573,
-      "step": 25556
-    },
-    {
-      "epoch": 13.328292046936115,
-      "grad_norm": 1.5234713554382324,
-      "learning_rate": 7.48251256281407e-05,
-      "loss": 5.4401,
-      "step": 25557
-    },
-    {
-      "epoch": 13.328813559322034,
-      "grad_norm": 1.4853897094726562,
-      "learning_rate": 7.482412060301508e-05,
-      "loss": 5.4037,
-      "step": 25558
-    },
-    {
-      "epoch": 13.329335071707954,
-      "grad_norm": 1.4234157800674438,
-      "learning_rate": 7.482311557788945e-05,
-      "loss": 5.5121,
-      "step": 25559
-    },
-    {
-      "epoch": 13.329856584093871,
-      "grad_norm": 1.537839651107788,
-      "learning_rate": 7.482211055276383e-05,
-      "loss": 5.1771,
-      "step": 25560
-    },
-    {
-      "epoch": 13.33037809647979,
-      "grad_norm": 1.5234686136245728,
-      "learning_rate": 7.482110552763819e-05,
-      "loss": 5.2573,
-      "step": 25561
-    },
-    {
-      "epoch": 13.33089960886571,
-      "grad_norm": 1.4342267513275146,
-      "learning_rate": 7.482010050251257e-05,
-      "loss": 5.3798,
-      "step": 25562
-    },
-    {
-      "epoch": 13.33142112125163,
-      "grad_norm": 1.4434552192687988,
-      "learning_rate": 7.481909547738693e-05,
-      "loss": 4.9074,
-      "step": 25563
-    },
-    {
-      "epoch": 13.33194263363755,
-      "grad_norm": 1.449726939201355,
-      "learning_rate": 7.481809045226131e-05,
-      "loss": 5.2803,
-      "step": 25564
-    },
-    {
-      "epoch": 13.332464146023469,
-      "grad_norm": 1.5331059694290161,
-      "learning_rate": 7.481708542713567e-05,
-      "loss": 5.0215,
-      "step": 25565
-    },
-    {
-      "epoch": 13.332985658409386,
-      "grad_norm": 1.5103806257247925,
-      "learning_rate": 7.481608040201005e-05,
-      "loss": 5.4239,
-      "step": 25566
-    },
-    {
-      "epoch": 13.333507170795306,
-      "grad_norm": 1.554749608039856,
-      "learning_rate": 7.481507537688443e-05,
-      "loss": 5.1711,
-      "step": 25567
-    },
-    {
-      "epoch": 13.334028683181225,
-      "grad_norm": 1.4405765533447266,
-      "learning_rate": 7.481407035175881e-05,
-      "loss": 5.5412,
-      "step": 25568
-    },
-    {
-      "epoch": 13.334550195567145,
-      "grad_norm": 1.3587651252746582,
-      "learning_rate": 7.481306532663317e-05,
-      "loss": 5.6807,
-      "step": 25569
-    },
-    {
-      "epoch": 13.335071707953064,
-      "grad_norm": 1.4758906364440918,
-      "learning_rate": 7.481206030150754e-05,
-      "loss": 5.7347,
-      "step": 25570
-    },
-    {
-      "epoch": 13.335593220338984,
-      "grad_norm": 1.5429096221923828,
-      "learning_rate": 7.481105527638191e-05,
-      "loss": 5.2594,
-      "step": 25571
-    },
-    {
-      "epoch": 13.336114732724901,
-      "grad_norm": 1.4282653331756592,
-      "learning_rate": 7.481005025125628e-05,
-      "loss": 5.7007,
-      "step": 25572
-    },
-    {
-      "epoch": 13.336636245110821,
-      "grad_norm": 1.3735647201538086,
-      "learning_rate": 7.480904522613066e-05,
-      "loss": 5.6683,
-      "step": 25573
-    },
-    {
-      "epoch": 13.33715775749674,
-      "grad_norm": 1.4209283590316772,
-      "learning_rate": 7.480804020100502e-05,
-      "loss": 4.9736,
-      "step": 25574
-    },
-    {
-      "epoch": 13.33767926988266,
-      "grad_norm": 1.4768247604370117,
-      "learning_rate": 7.48070351758794e-05,
-      "loss": 5.0097,
-      "step": 25575
-    },
-    {
-      "epoch": 13.33820078226858,
-      "grad_norm": 1.4996529817581177,
-      "learning_rate": 7.480603015075378e-05,
-      "loss": 5.2568,
-      "step": 25576
-    },
-    {
-      "epoch": 13.338722294654499,
-      "grad_norm": 1.4720381498336792,
-      "learning_rate": 7.480502512562815e-05,
-      "loss": 5.6309,
-      "step": 25577
-    },
-    {
-      "epoch": 13.339243807040416,
-      "grad_norm": 1.4799480438232422,
-      "learning_rate": 7.480402010050252e-05,
-      "loss": 5.3764,
-      "step": 25578
-    },
-    {
-      "epoch": 13.339765319426336,
-      "grad_norm": 1.5044090747833252,
-      "learning_rate": 7.48030150753769e-05,
-      "loss": 5.5646,
-      "step": 25579
-    },
-    {
-      "epoch": 13.340286831812255,
-      "grad_norm": 1.4270243644714355,
-      "learning_rate": 7.480201005025126e-05,
-      "loss": 5.1767,
-      "step": 25580
-    },
-    {
-      "epoch": 13.340808344198175,
-      "grad_norm": 1.54099440574646,
-      "learning_rate": 7.480100502512564e-05,
-      "loss": 5.2961,
-      "step": 25581
-    },
-    {
-      "epoch": 13.341329856584094,
-      "grad_norm": 1.3753052949905396,
-      "learning_rate": 7.48e-05,
-      "loss": 5.4752,
-      "step": 25582
-    },
-    {
-      "epoch": 13.341851368970014,
-      "grad_norm": 1.4975007772445679,
-      "learning_rate": 7.479899497487437e-05,
-      "loss": 5.6612,
-      "step": 25583
-    },
-    {
-      "epoch": 13.342372881355931,
-      "grad_norm": 1.4001284837722778,
-      "learning_rate": 7.479798994974874e-05,
-      "loss": 5.5031,
-      "step": 25584
-    },
-    {
-      "epoch": 13.342894393741851,
-      "grad_norm": 1.4902957677841187,
-      "learning_rate": 7.479698492462311e-05,
-      "loss": 5.3734,
-      "step": 25585
-    },
-    {
-      "epoch": 13.34341590612777,
-      "grad_norm": 1.6393481492996216,
-      "learning_rate": 7.479597989949749e-05,
-      "loss": 5.0605,
-      "step": 25586
-    },
-    {
-      "epoch": 13.34393741851369,
-      "grad_norm": 1.5268242359161377,
-      "learning_rate": 7.479497487437186e-05,
-      "loss": 5.5633,
-      "step": 25587
-    },
-    {
-      "epoch": 13.34445893089961,
-      "grad_norm": 1.5437623262405396,
-      "learning_rate": 7.479396984924624e-05,
-      "loss": 5.5552,
-      "step": 25588
-    },
-    {
-      "epoch": 13.344980443285529,
-      "grad_norm": 1.3292922973632812,
-      "learning_rate": 7.47929648241206e-05,
-      "loss": 5.3713,
-      "step": 25589
-    },
-    {
-      "epoch": 13.345501955671446,
-      "grad_norm": 1.4286381006240845,
-      "learning_rate": 7.479195979899498e-05,
-      "loss": 5.4373,
-      "step": 25590
-    },
-    {
-      "epoch": 13.346023468057366,
-      "grad_norm": 1.4227012395858765,
-      "learning_rate": 7.479095477386935e-05,
-      "loss": 5.7199,
-      "step": 25591
-    },
-    {
-      "epoch": 13.346544980443285,
-      "grad_norm": 1.351165771484375,
-      "learning_rate": 7.478994974874373e-05,
-      "loss": 5.7236,
-      "step": 25592
-    },
-    {
-      "epoch": 13.347066492829205,
-      "grad_norm": 1.5079376697540283,
-      "learning_rate": 7.478894472361809e-05,
-      "loss": 5.7103,
-      "step": 25593
-    },
-    {
-      "epoch": 13.347588005215124,
-      "grad_norm": 1.469797968864441,
-      "learning_rate": 7.478793969849247e-05,
-      "loss": 5.4346,
-      "step": 25594
-    },
-    {
-      "epoch": 13.348109517601044,
-      "grad_norm": 1.3638031482696533,
-      "learning_rate": 7.478693467336683e-05,
-      "loss": 5.6972,
-      "step": 25595
-    },
-    {
-      "epoch": 13.348631029986961,
-      "grad_norm": 1.4642471075057983,
-      "learning_rate": 7.478592964824121e-05,
-      "loss": 5.3389,
-      "step": 25596
-    },
-    {
-      "epoch": 13.349152542372881,
-      "grad_norm": 1.5090336799621582,
-      "learning_rate": 7.478492462311559e-05,
-      "loss": 5.5247,
-      "step": 25597
-    },
-    {
-      "epoch": 13.3496740547588,
-      "grad_norm": 1.4612860679626465,
-      "learning_rate": 7.478391959798995e-05,
-      "loss": 5.6164,
-      "step": 25598
-    },
-    {
-      "epoch": 13.35019556714472,
-      "grad_norm": 1.4524247646331787,
-      "learning_rate": 7.478291457286433e-05,
-      "loss": 5.401,
-      "step": 25599
-    },
-    {
-      "epoch": 13.35071707953064,
-      "grad_norm": 1.5271186828613281,
-      "learning_rate": 7.47819095477387e-05,
-      "loss": 5.0485,
-      "step": 25600
-    },
-    {
-      "epoch": 13.351238591916559,
-      "grad_norm": 1.4511741399765015,
-      "learning_rate": 7.478090452261307e-05,
-      "loss": 5.1249,
-      "step": 25601
-    },
-    {
-      "epoch": 13.351760104302477,
-      "grad_norm": 1.5485676527023315,
-      "learning_rate": 7.477989949748744e-05,
-      "loss": 5.1945,
-      "step": 25602
-    },
-    {
-      "epoch": 13.352281616688396,
-      "grad_norm": 1.6101113557815552,
-      "learning_rate": 7.477889447236181e-05,
-      "loss": 5.1073,
-      "step": 25603
-    },
-    {
-      "epoch": 13.352803129074315,
-      "grad_norm": 1.4590812921524048,
-      "learning_rate": 7.477788944723618e-05,
-      "loss": 4.8594,
-      "step": 25604
-    },
-    {
-      "epoch": 13.353324641460235,
-      "grad_norm": 1.408347249031067,
-      "learning_rate": 7.477688442211056e-05,
-      "loss": 5.6154,
-      "step": 25605
-    },
-    {
-      "epoch": 13.353846153846154,
-      "grad_norm": 1.3557367324829102,
-      "learning_rate": 7.477587939698492e-05,
-      "loss": 5.6507,
-      "step": 25606
-    },
-    {
-      "epoch": 13.354367666232074,
-      "grad_norm": 1.4424008131027222,
-      "learning_rate": 7.47748743718593e-05,
-      "loss": 5.4778,
-      "step": 25607
-    },
-    {
-      "epoch": 13.354889178617992,
-      "grad_norm": 1.4769909381866455,
-      "learning_rate": 7.477386934673368e-05,
-      "loss": 5.2171,
-      "step": 25608
-    },
-    {
-      "epoch": 13.355410691003911,
-      "grad_norm": 1.5071487426757812,
-      "learning_rate": 7.477286432160805e-05,
-      "loss": 5.4186,
-      "step": 25609
-    },
-    {
-      "epoch": 13.35593220338983,
-      "grad_norm": 1.4086124897003174,
-      "learning_rate": 7.477185929648242e-05,
-      "loss": 5.3972,
-      "step": 25610
-    },
-    {
-      "epoch": 13.35645371577575,
-      "grad_norm": 1.637149453163147,
-      "learning_rate": 7.477085427135678e-05,
-      "loss": 5.4532,
-      "step": 25611
-    },
-    {
-      "epoch": 13.35697522816167,
-      "grad_norm": 1.5026112794876099,
-      "learning_rate": 7.476984924623116e-05,
-      "loss": 5.4415,
-      "step": 25612
-    },
-    {
-      "epoch": 13.357496740547589,
-      "grad_norm": 1.5974375009536743,
-      "learning_rate": 7.476884422110553e-05,
-      "loss": 5.0971,
-      "step": 25613
-    },
-    {
-      "epoch": 13.358018252933507,
-      "grad_norm": 1.4003827571868896,
-      "learning_rate": 7.47678391959799e-05,
-      "loss": 5.7445,
-      "step": 25614
-    },
-    {
-      "epoch": 13.358539765319426,
-      "grad_norm": 1.4238201379776,
-      "learning_rate": 7.476683417085427e-05,
-      "loss": 5.5155,
-      "step": 25615
-    },
-    {
-      "epoch": 13.359061277705345,
-      "grad_norm": 1.4987165927886963,
-      "learning_rate": 7.476582914572864e-05,
-      "loss": 5.5398,
-      "step": 25616
-    },
-    {
-      "epoch": 13.359582790091265,
-      "grad_norm": 1.4484001398086548,
-      "learning_rate": 7.476482412060302e-05,
-      "loss": 5.3879,
-      "step": 25617
-    },
-    {
-      "epoch": 13.360104302477184,
-      "grad_norm": 1.4772412776947021,
-      "learning_rate": 7.47638190954774e-05,
-      "loss": 5.4973,
-      "step": 25618
-    },
-    {
-      "epoch": 13.360625814863104,
-      "grad_norm": 1.3976552486419678,
-      "learning_rate": 7.476281407035176e-05,
-      "loss": 5.856,
-      "step": 25619
-    },
-    {
-      "epoch": 13.361147327249022,
-      "grad_norm": 1.5786577463150024,
-      "learning_rate": 7.476180904522614e-05,
-      "loss": 5.3627,
-      "step": 25620
-    },
-    {
-      "epoch": 13.361668839634941,
-      "grad_norm": 1.4778320789337158,
-      "learning_rate": 7.476080402010051e-05,
-      "loss": 5.3258,
-      "step": 25621
-    },
-    {
-      "epoch": 13.36219035202086,
-      "grad_norm": 1.4685102701187134,
-      "learning_rate": 7.475979899497488e-05,
-      "loss": 5.528,
-      "step": 25622
-    },
-    {
-      "epoch": 13.36271186440678,
-      "grad_norm": 1.416680097579956,
-      "learning_rate": 7.475879396984925e-05,
-      "loss": 5.2404,
-      "step": 25623
-    },
-    {
-      "epoch": 13.3632333767927,
-      "grad_norm": 1.437638282775879,
-      "learning_rate": 7.475778894472361e-05,
-      "loss": 5.6444,
-      "step": 25624
-    },
-    {
-      "epoch": 13.363754889178619,
-      "grad_norm": 1.4124419689178467,
-      "learning_rate": 7.475678391959799e-05,
-      "loss": 5.1119,
-      "step": 25625
-    },
-    {
-      "epoch": 13.364276401564537,
-      "grad_norm": 1.4327555894851685,
-      "learning_rate": 7.475577889447236e-05,
-      "loss": 5.4685,
-      "step": 25626
-    },
-    {
-      "epoch": 13.364797913950456,
-      "grad_norm": 1.3395960330963135,
-      "learning_rate": 7.475477386934673e-05,
-      "loss": 5.4885,
-      "step": 25627
-    },
-    {
-      "epoch": 13.365319426336375,
-      "grad_norm": 1.2821604013442993,
-      "learning_rate": 7.475376884422111e-05,
-      "loss": 5.6666,
-      "step": 25628
-    },
-    {
-      "epoch": 13.365840938722295,
-      "grad_norm": 1.5461915731430054,
-      "learning_rate": 7.475276381909549e-05,
-      "loss": 5.3706,
-      "step": 25629
-    },
-    {
-      "epoch": 13.366362451108214,
-      "grad_norm": 1.4926625490188599,
-      "learning_rate": 7.475175879396985e-05,
-      "loss": 5.2751,
-      "step": 25630
-    },
-    {
-      "epoch": 13.366883963494132,
-      "grad_norm": 1.4054052829742432,
-      "learning_rate": 7.475075376884423e-05,
-      "loss": 5.3923,
-      "step": 25631
-    },
-    {
-      "epoch": 13.367405475880052,
-      "grad_norm": 1.5060489177703857,
-      "learning_rate": 7.47497487437186e-05,
-      "loss": 4.8704,
-      "step": 25632
-    },
-    {
-      "epoch": 13.367926988265971,
-      "grad_norm": 1.5691038370132446,
-      "learning_rate": 7.474874371859297e-05,
-      "loss": 5.1224,
-      "step": 25633
-    },
-    {
-      "epoch": 13.36844850065189,
-      "grad_norm": 1.4810845851898193,
-      "learning_rate": 7.474773869346734e-05,
-      "loss": 5.5049,
-      "step": 25634
-    },
-    {
-      "epoch": 13.36897001303781,
-      "grad_norm": 1.5178236961364746,
-      "learning_rate": 7.474673366834172e-05,
-      "loss": 5.5532,
-      "step": 25635
-    },
-    {
-      "epoch": 13.36949152542373,
-      "grad_norm": 1.3550325632095337,
-      "learning_rate": 7.474572864321608e-05,
-      "loss": 5.3052,
-      "step": 25636
-    },
-    {
-      "epoch": 13.370013037809649,
-      "grad_norm": 1.4664517641067505,
-      "learning_rate": 7.474472361809046e-05,
-      "loss": 5.1136,
-      "step": 25637
-    },
-    {
-      "epoch": 13.370534550195567,
-      "grad_norm": 1.3351376056671143,
-      "learning_rate": 7.474371859296484e-05,
-      "loss": 4.6674,
-      "step": 25638
-    },
-    {
-      "epoch": 13.371056062581486,
-      "grad_norm": 1.3599261045455933,
-      "learning_rate": 7.47427135678392e-05,
-      "loss": 5.6493,
-      "step": 25639
-    },
-    {
-      "epoch": 13.371577574967406,
-      "grad_norm": 1.402835726737976,
-      "learning_rate": 7.474170854271358e-05,
-      "loss": 5.391,
-      "step": 25640
-    },
-    {
-      "epoch": 13.372099087353325,
-      "grad_norm": 1.4513009786605835,
-      "learning_rate": 7.474070351758794e-05,
-      "loss": 4.9485,
-      "step": 25641
-    },
-    {
-      "epoch": 13.372620599739244,
-      "grad_norm": 1.4422439336776733,
-      "learning_rate": 7.473969849246232e-05,
-      "loss": 5.4695,
-      "step": 25642
-    },
-    {
-      "epoch": 13.373142112125162,
-      "grad_norm": 1.4548817873001099,
-      "learning_rate": 7.473869346733668e-05,
-      "loss": 5.2472,
-      "step": 25643
-    },
-    {
-      "epoch": 13.373663624511082,
-      "grad_norm": 1.3729034662246704,
-      "learning_rate": 7.473768844221106e-05,
-      "loss": 5.4184,
-      "step": 25644
-    },
-    {
-      "epoch": 13.374185136897001,
-      "grad_norm": 1.440813422203064,
-      "learning_rate": 7.473668341708543e-05,
-      "loss": 4.9386,
-      "step": 25645
-    },
-    {
-      "epoch": 13.37470664928292,
-      "grad_norm": 1.3942042589187622,
-      "learning_rate": 7.47356783919598e-05,
-      "loss": 5.5756,
-      "step": 25646
-    },
-    {
-      "epoch": 13.37522816166884,
-      "grad_norm": 1.5231152772903442,
-      "learning_rate": 7.473467336683417e-05,
-      "loss": 5.3009,
-      "step": 25647
-    },
-    {
-      "epoch": 13.37574967405476,
-      "grad_norm": 1.554020881652832,
-      "learning_rate": 7.473366834170855e-05,
-      "loss": 5.7521,
-      "step": 25648
-    },
-    {
-      "epoch": 13.376271186440677,
-      "grad_norm": 1.5375339984893799,
-      "learning_rate": 7.473266331658292e-05,
-      "loss": 5.151,
-      "step": 25649
-    },
-    {
-      "epoch": 13.376792698826597,
-      "grad_norm": 1.6193546056747437,
-      "learning_rate": 7.473165829145729e-05,
-      "loss": 5.0716,
-      "step": 25650
-    },
-    {
-      "epoch": 13.377314211212516,
-      "grad_norm": 1.395033597946167,
-      "learning_rate": 7.473065326633167e-05,
-      "loss": 5.5376,
-      "step": 25651
-    },
-    {
-      "epoch": 13.377835723598436,
-      "grad_norm": 1.3955535888671875,
-      "learning_rate": 7.472964824120603e-05,
-      "loss": 5.5587,
-      "step": 25652
-    },
-    {
-      "epoch": 13.378357235984355,
-      "grad_norm": 1.3478915691375732,
-      "learning_rate": 7.472864321608041e-05,
-      "loss": 5.9704,
-      "step": 25653
-    },
-    {
-      "epoch": 13.378878748370274,
-      "grad_norm": 1.477400779724121,
-      "learning_rate": 7.472763819095477e-05,
-      "loss": 5.3968,
-      "step": 25654
-    },
-    {
-      "epoch": 13.379400260756192,
-      "grad_norm": 1.404414176940918,
-      "learning_rate": 7.472663316582915e-05,
-      "loss": 5.7882,
-      "step": 25655
-    },
-    {
-      "epoch": 13.379921773142112,
-      "grad_norm": 1.4042272567749023,
-      "learning_rate": 7.472562814070351e-05,
-      "loss": 5.3531,
-      "step": 25656
-    },
-    {
-      "epoch": 13.380443285528031,
-      "grad_norm": 1.5719094276428223,
-      "learning_rate": 7.472462311557789e-05,
-      "loss": 5.0507,
-      "step": 25657
-    },
-    {
-      "epoch": 13.38096479791395,
-      "grad_norm": 1.410982370376587,
-      "learning_rate": 7.472361809045227e-05,
-      "loss": 5.6696,
-      "step": 25658
-    },
-    {
-      "epoch": 13.38148631029987,
-      "grad_norm": 1.4770731925964355,
-      "learning_rate": 7.472261306532665e-05,
-      "loss": 5.2387,
-      "step": 25659
-    },
-    {
-      "epoch": 13.38200782268579,
-      "grad_norm": 1.4308491945266724,
-      "learning_rate": 7.472160804020101e-05,
-      "loss": 5.7605,
-      "step": 25660
-    },
-    {
-      "epoch": 13.382529335071707,
-      "grad_norm": 1.4518773555755615,
-      "learning_rate": 7.472060301507539e-05,
-      "loss": 5.5057,
-      "step": 25661
-    },
-    {
-      "epoch": 13.383050847457627,
-      "grad_norm": 1.496427297592163,
-      "learning_rate": 7.471959798994975e-05,
-      "loss": 5.4727,
-      "step": 25662
-    },
-    {
-      "epoch": 13.383572359843546,
-      "grad_norm": 1.4180673360824585,
-      "learning_rate": 7.471859296482412e-05,
-      "loss": 5.4186,
-      "step": 25663
-    },
-    {
-      "epoch": 13.384093872229466,
-      "grad_norm": 1.5029807090759277,
-      "learning_rate": 7.47175879396985e-05,
-      "loss": 5.2266,
-      "step": 25664
-    },
-    {
-      "epoch": 13.384615384615385,
-      "grad_norm": 1.4320181608200073,
-      "learning_rate": 7.471658291457286e-05,
-      "loss": 5.5155,
-      "step": 25665
-    },
-    {
-      "epoch": 13.385136897001304,
-      "grad_norm": 1.5789037942886353,
-      "learning_rate": 7.471557788944724e-05,
-      "loss": 5.0076,
-      "step": 25666
-    },
-    {
-      "epoch": 13.385658409387222,
-      "grad_norm": 1.4821689128875732,
-      "learning_rate": 7.47145728643216e-05,
-      "loss": 5.1354,
-      "step": 25667
-    },
-    {
-      "epoch": 13.386179921773142,
-      "grad_norm": 1.4501688480377197,
-      "learning_rate": 7.471356783919598e-05,
-      "loss": 5.2539,
-      "step": 25668
-    },
-    {
-      "epoch": 13.386701434159061,
-      "grad_norm": 1.5472636222839355,
-      "learning_rate": 7.471256281407036e-05,
-      "loss": 5.5932,
-      "step": 25669
-    },
-    {
-      "epoch": 13.38722294654498,
-      "grad_norm": 1.5350309610366821,
-      "learning_rate": 7.471155778894474e-05,
-      "loss": 5.383,
-      "step": 25670
-    },
-    {
-      "epoch": 13.3877444589309,
-      "grad_norm": 1.5583363771438599,
-      "learning_rate": 7.47105527638191e-05,
-      "loss": 5.5647,
-      "step": 25671
-    },
-    {
-      "epoch": 13.38826597131682,
-      "grad_norm": 1.4599146842956543,
-      "learning_rate": 7.470954773869348e-05,
-      "loss": 4.8558,
-      "step": 25672
-    },
-    {
-      "epoch": 13.388787483702737,
-      "grad_norm": 1.5359036922454834,
-      "learning_rate": 7.470854271356784e-05,
-      "loss": 5.116,
-      "step": 25673
-    },
-    {
-      "epoch": 13.389308996088657,
-      "grad_norm": 1.3672057390213013,
-      "learning_rate": 7.470753768844222e-05,
-      "loss": 5.8933,
-      "step": 25674
-    },
-    {
-      "epoch": 13.389830508474576,
-      "grad_norm": 1.3970375061035156,
-      "learning_rate": 7.470653266331658e-05,
-      "loss": 5.2878,
-      "step": 25675
-    },
-    {
-      "epoch": 13.390352020860496,
-      "grad_norm": 1.4483989477157593,
-      "learning_rate": 7.470552763819095e-05,
-      "loss": 5.3768,
-      "step": 25676
-    },
-    {
-      "epoch": 13.390873533246415,
-      "grad_norm": 1.488216519355774,
-      "learning_rate": 7.470452261306533e-05,
-      "loss": 5.46,
-      "step": 25677
-    },
-    {
-      "epoch": 13.391395045632335,
-      "grad_norm": 1.395548939704895,
-      "learning_rate": 7.470351758793969e-05,
-      "loss": 5.6775,
-      "step": 25678
-    },
-    {
-      "epoch": 13.391916558018252,
-      "grad_norm": 1.4001498222351074,
-      "learning_rate": 7.470251256281407e-05,
-      "loss": 5.5676,
-      "step": 25679
-    },
-    {
-      "epoch": 13.392438070404172,
-      "grad_norm": 1.4096919298171997,
-      "learning_rate": 7.470150753768845e-05,
-      "loss": 5.6761,
-      "step": 25680
-    },
-    {
-      "epoch": 13.392959582790091,
-      "grad_norm": 1.5681188106536865,
-      "learning_rate": 7.470050251256282e-05,
-      "loss": 5.0092,
-      "step": 25681
-    },
-    {
-      "epoch": 13.39348109517601,
-      "grad_norm": 1.5594825744628906,
-      "learning_rate": 7.469949748743719e-05,
-      "loss": 5.2187,
-      "step": 25682
-    },
-    {
-      "epoch": 13.39400260756193,
-      "grad_norm": 1.4995973110198975,
-      "learning_rate": 7.469849246231157e-05,
-      "loss": 4.5949,
-      "step": 25683
-    },
-    {
-      "epoch": 13.39452411994785,
-      "grad_norm": 1.3732390403747559,
-      "learning_rate": 7.469748743718593e-05,
-      "loss": 5.5535,
-      "step": 25684
-    },
-    {
-      "epoch": 13.395045632333767,
-      "grad_norm": 1.4590548276901245,
-      "learning_rate": 7.469648241206031e-05,
-      "loss": 5.4645,
-      "step": 25685
-    },
-    {
-      "epoch": 13.395567144719687,
-      "grad_norm": 1.4474213123321533,
-      "learning_rate": 7.469547738693467e-05,
-      "loss": 5.4554,
-      "step": 25686
-    },
-    {
-      "epoch": 13.396088657105606,
-      "grad_norm": 1.5184473991394043,
-      "learning_rate": 7.469447236180905e-05,
-      "loss": 5.2476,
-      "step": 25687
-    },
-    {
-      "epoch": 13.396610169491526,
-      "grad_norm": 1.4180519580841064,
-      "learning_rate": 7.469346733668341e-05,
-      "loss": 5.5725,
-      "step": 25688
-    },
-    {
-      "epoch": 13.397131681877445,
-      "grad_norm": 1.54567289352417,
-      "learning_rate": 7.469246231155779e-05,
-      "loss": 4.9414,
-      "step": 25689
-    },
-    {
-      "epoch": 13.397653194263365,
-      "grad_norm": 1.4667245149612427,
-      "learning_rate": 7.469145728643217e-05,
-      "loss": 5.3278,
-      "step": 25690
-    },
-    {
-      "epoch": 13.398174706649282,
-      "grad_norm": 1.4029037952423096,
-      "learning_rate": 7.469045226130653e-05,
-      "loss": 5.6966,
-      "step": 25691
-    },
-    {
-      "epoch": 13.398696219035202,
-      "grad_norm": 1.4207602739334106,
-      "learning_rate": 7.468944723618091e-05,
-      "loss": 5.6032,
-      "step": 25692
-    },
-    {
-      "epoch": 13.399217731421121,
-      "grad_norm": 1.5073564052581787,
-      "learning_rate": 7.468844221105528e-05,
-      "loss": 5.5909,
-      "step": 25693
-    },
-    {
-      "epoch": 13.39973924380704,
-      "grad_norm": 1.4459848403930664,
-      "learning_rate": 7.468743718592965e-05,
-      "loss": 5.4975,
-      "step": 25694
-    },
-    {
-      "epoch": 13.40026075619296,
-      "grad_norm": 1.4309762716293335,
-      "learning_rate": 7.468643216080402e-05,
-      "loss": 5.5593,
-      "step": 25695
-    },
-    {
-      "epoch": 13.40078226857888,
-      "grad_norm": 1.459652066230774,
-      "learning_rate": 7.46854271356784e-05,
-      "loss": 5.321,
-      "step": 25696
-    },
-    {
-      "epoch": 13.401303780964797,
-      "grad_norm": 1.4077222347259521,
-      "learning_rate": 7.468442211055276e-05,
-      "loss": 5.6692,
-      "step": 25697
-    },
-    {
-      "epoch": 13.401825293350717,
-      "grad_norm": 1.4687877893447876,
-      "learning_rate": 7.468341708542714e-05,
-      "loss": 5.3568,
-      "step": 25698
-    },
-    {
-      "epoch": 13.402346805736636,
-      "grad_norm": 1.3625712394714355,
-      "learning_rate": 7.46824120603015e-05,
-      "loss": 4.3135,
-      "step": 25699
-    },
-    {
-      "epoch": 13.402868318122556,
-      "grad_norm": 1.4091373682022095,
-      "learning_rate": 7.468140703517588e-05,
-      "loss": 5.5904,
-      "step": 25700
-    },
-    {
-      "epoch": 13.403389830508475,
-      "grad_norm": 1.3913195133209229,
-      "learning_rate": 7.468040201005026e-05,
-      "loss": 5.2957,
-      "step": 25701
-    },
-    {
-      "epoch": 13.403911342894395,
-      "grad_norm": 1.4963539838790894,
-      "learning_rate": 7.467939698492464e-05,
-      "loss": 5.3414,
-      "step": 25702
-    },
-    {
-      "epoch": 13.404432855280312,
-      "grad_norm": 1.4907171726226807,
-      "learning_rate": 7.4678391959799e-05,
-      "loss": 4.8583,
-      "step": 25703
-    },
-    {
-      "epoch": 13.404954367666232,
-      "grad_norm": 1.4909913539886475,
-      "learning_rate": 7.467738693467337e-05,
-      "loss": 5.3484,
-      "step": 25704
-    },
-    {
-      "epoch": 13.405475880052151,
-      "grad_norm": 1.3909473419189453,
-      "learning_rate": 7.467638190954774e-05,
-      "loss": 5.5971,
-      "step": 25705
-    },
-    {
-      "epoch": 13.40599739243807,
-      "grad_norm": 1.368985652923584,
-      "learning_rate": 7.467537688442211e-05,
-      "loss": 5.7136,
-      "step": 25706
-    },
-    {
-      "epoch": 13.40651890482399,
-      "grad_norm": 1.4492480754852295,
-      "learning_rate": 7.467437185929649e-05,
-      "loss": 5.0845,
-      "step": 25707
-    },
-    {
-      "epoch": 13.40704041720991,
-      "grad_norm": 1.5859419107437134,
-      "learning_rate": 7.467336683417085e-05,
-      "loss": 4.8755,
-      "step": 25708
-    },
-    {
-      "epoch": 13.407561929595827,
-      "grad_norm": 1.5622738599777222,
-      "learning_rate": 7.467236180904523e-05,
-      "loss": 5.6294,
-      "step": 25709
-    },
-    {
-      "epoch": 13.408083441981747,
-      "grad_norm": 1.4414821863174438,
-      "learning_rate": 7.46713567839196e-05,
-      "loss": 5.6022,
-      "step": 25710
-    },
-    {
-      "epoch": 13.408604954367666,
-      "grad_norm": 1.4327222108840942,
-      "learning_rate": 7.467035175879398e-05,
-      "loss": 5.708,
-      "step": 25711
-    },
-    {
-      "epoch": 13.409126466753586,
-      "grad_norm": 1.4639970064163208,
-      "learning_rate": 7.466934673366835e-05,
-      "loss": 5.1007,
-      "step": 25712
-    },
-    {
-      "epoch": 13.409647979139505,
-      "grad_norm": 1.5644515752792358,
-      "learning_rate": 7.466834170854273e-05,
-      "loss": 5.092,
-      "step": 25713
-    },
-    {
-      "epoch": 13.410169491525423,
-      "grad_norm": 1.4968621730804443,
-      "learning_rate": 7.466733668341709e-05,
-      "loss": 5.7236,
-      "step": 25714
-    },
-    {
-      "epoch": 13.410691003911342,
-      "grad_norm": 1.4468327760696411,
-      "learning_rate": 7.466633165829147e-05,
-      "loss": 5.5253,
-      "step": 25715
-    },
-    {
-      "epoch": 13.411212516297262,
-      "grad_norm": 1.4137842655181885,
-      "learning_rate": 7.466532663316583e-05,
-      "loss": 5.5908,
-      "step": 25716
-    },
-    {
-      "epoch": 13.411734028683181,
-      "grad_norm": 1.612979769706726,
-      "learning_rate": 7.46643216080402e-05,
-      "loss": 5.4851,
-      "step": 25717
-    },
-    {
-      "epoch": 13.4122555410691,
-      "grad_norm": 1.3531816005706787,
-      "learning_rate": 7.466331658291457e-05,
-      "loss": 5.8046,
-      "step": 25718
-    },
-    {
-      "epoch": 13.41277705345502,
-      "grad_norm": 1.413576364517212,
-      "learning_rate": 7.466231155778894e-05,
-      "loss": 5.5687,
-      "step": 25719
-    },
-    {
-      "epoch": 13.41329856584094,
-      "grad_norm": 1.386243224143982,
-      "learning_rate": 7.466130653266332e-05,
-      "loss": 5.6071,
-      "step": 25720
-    },
-    {
-      "epoch": 13.413820078226857,
-      "grad_norm": 1.6928720474243164,
-      "learning_rate": 7.46603015075377e-05,
-      "loss": 4.6581,
-      "step": 25721
-    },
-    {
-      "epoch": 13.414341590612777,
-      "grad_norm": 1.5335665941238403,
-      "learning_rate": 7.465929648241207e-05,
-      "loss": 5.2963,
-      "step": 25722
-    },
-    {
-      "epoch": 13.414863102998696,
-      "grad_norm": 1.4856847524642944,
-      "learning_rate": 7.465829145728644e-05,
-      "loss": 5.9385,
-      "step": 25723
-    },
-    {
-      "epoch": 13.415384615384616,
-      "grad_norm": 1.4318376779556274,
-      "learning_rate": 7.465728643216081e-05,
-      "loss": 5.4577,
-      "step": 25724
-    },
-    {
-      "epoch": 13.415906127770535,
-      "grad_norm": 1.3441425561904907,
-      "learning_rate": 7.465628140703518e-05,
-      "loss": 5.6296,
-      "step": 25725
-    },
-    {
-      "epoch": 13.416427640156453,
-      "grad_norm": 1.420954704284668,
-      "learning_rate": 7.465527638190956e-05,
-      "loss": 5.2107,
-      "step": 25726
-    },
-    {
-      "epoch": 13.416949152542372,
-      "grad_norm": 1.4372855424880981,
-      "learning_rate": 7.465427135678392e-05,
-      "loss": 5.3871,
-      "step": 25727
-    },
-    {
-      "epoch": 13.417470664928292,
-      "grad_norm": 1.4491597414016724,
-      "learning_rate": 7.46532663316583e-05,
-      "loss": 5.4877,
-      "step": 25728
-    },
-    {
-      "epoch": 13.417992177314211,
-      "grad_norm": 1.4583172798156738,
-      "learning_rate": 7.465226130653266e-05,
-      "loss": 5.1172,
-      "step": 25729
-    },
-    {
-      "epoch": 13.41851368970013,
-      "grad_norm": 1.5192793607711792,
-      "learning_rate": 7.465125628140704e-05,
-      "loss": 5.607,
-      "step": 25730
-    },
-    {
-      "epoch": 13.41903520208605,
-      "grad_norm": 1.312098503112793,
-      "learning_rate": 7.465025125628142e-05,
-      "loss": 5.2955,
-      "step": 25731
-    },
-    {
-      "epoch": 13.419556714471968,
-      "grad_norm": 1.6170865297317505,
-      "learning_rate": 7.464924623115578e-05,
-      "loss": 5.4144,
-      "step": 25732
-    },
-    {
-      "epoch": 13.420078226857887,
-      "grad_norm": 1.3526105880737305,
-      "learning_rate": 7.464824120603016e-05,
-      "loss": 5.8623,
-      "step": 25733
-    },
-    {
-      "epoch": 13.420599739243807,
-      "grad_norm": 1.3872977495193481,
-      "learning_rate": 7.464723618090452e-05,
-      "loss": 5.582,
-      "step": 25734
-    },
-    {
-      "epoch": 13.421121251629726,
-      "grad_norm": 1.4210340976715088,
-      "learning_rate": 7.46462311557789e-05,
-      "loss": 5.5156,
-      "step": 25735
-    },
-    {
-      "epoch": 13.421642764015646,
-      "grad_norm": 1.3963134288787842,
-      "learning_rate": 7.464522613065327e-05,
-      "loss": 5.4318,
-      "step": 25736
-    },
-    {
-      "epoch": 13.422164276401565,
-      "grad_norm": 1.506334900856018,
-      "learning_rate": 7.464422110552764e-05,
-      "loss": 5.2795,
-      "step": 25737
-    },
-    {
-      "epoch": 13.422685788787483,
-      "grad_norm": 1.4962852001190186,
-      "learning_rate": 7.464321608040201e-05,
-      "loss": 5.5451,
-      "step": 25738
-    },
-    {
-      "epoch": 13.423207301173402,
-      "grad_norm": 1.4622297286987305,
-      "learning_rate": 7.464221105527639e-05,
-      "loss": 5.794,
-      "step": 25739
-    },
-    {
-      "epoch": 13.423728813559322,
-      "grad_norm": 1.4828394651412964,
-      "learning_rate": 7.464120603015075e-05,
-      "loss": 5.3586,
-      "step": 25740
-    },
-    {
-      "epoch": 13.424250325945241,
-      "grad_norm": 1.446427822113037,
-      "learning_rate": 7.464020100502513e-05,
-      "loss": 5.5556,
-      "step": 25741
-    },
-    {
-      "epoch": 13.42477183833116,
-      "grad_norm": 1.4946023225784302,
-      "learning_rate": 7.46391959798995e-05,
-      "loss": 5.195,
-      "step": 25742
-    },
-    {
-      "epoch": 13.42529335071708,
-      "grad_norm": 1.499858021736145,
-      "learning_rate": 7.463819095477387e-05,
-      "loss": 4.8704,
-      "step": 25743
-    },
-    {
-      "epoch": 13.425814863102998,
-      "grad_norm": 1.4598660469055176,
-      "learning_rate": 7.463718592964825e-05,
-      "loss": 5.2659,
-      "step": 25744
-    },
-    {
-      "epoch": 13.426336375488917,
-      "grad_norm": 1.5045286417007446,
-      "learning_rate": 7.463618090452261e-05,
-      "loss": 5.4861,
-      "step": 25745
-    },
-    {
-      "epoch": 13.426857887874837,
-      "grad_norm": 1.3390766382217407,
-      "learning_rate": 7.463517587939699e-05,
-      "loss": 4.9591,
-      "step": 25746
-    },
-    {
-      "epoch": 13.427379400260756,
-      "grad_norm": 1.3206419944763184,
-      "learning_rate": 7.463417085427135e-05,
-      "loss": 5.4552,
-      "step": 25747
-    },
-    {
-      "epoch": 13.427900912646676,
-      "grad_norm": 1.4373273849487305,
-      "learning_rate": 7.463316582914573e-05,
-      "loss": 5.5481,
-      "step": 25748
-    },
-    {
-      "epoch": 13.428422425032595,
-      "grad_norm": 1.3725289106369019,
-      "learning_rate": 7.46321608040201e-05,
-      "loss": 5.7542,
-      "step": 25749
-    },
-    {
-      "epoch": 13.428943937418513,
-      "grad_norm": 1.3941349983215332,
-      "learning_rate": 7.463115577889447e-05,
-      "loss": 4.9363,
-      "step": 25750
-    },
-    {
-      "epoch": 13.429465449804432,
-      "grad_norm": 1.4652057886123657,
-      "learning_rate": 7.463015075376885e-05,
-      "loss": 5.0224,
-      "step": 25751
-    },
-    {
-      "epoch": 13.429986962190352,
-      "grad_norm": 1.4516547918319702,
-      "learning_rate": 7.462914572864323e-05,
-      "loss": 5.4692,
-      "step": 25752
-    },
-    {
-      "epoch": 13.430508474576271,
-      "grad_norm": 1.4874000549316406,
-      "learning_rate": 7.46281407035176e-05,
-      "loss": 5.3026,
-      "step": 25753
-    },
-    {
-      "epoch": 13.43102998696219,
-      "grad_norm": 1.4034217596054077,
-      "learning_rate": 7.462713567839197e-05,
-      "loss": 5.3594,
-      "step": 25754
-    },
-    {
-      "epoch": 13.43155149934811,
-      "grad_norm": 1.4512810707092285,
-      "learning_rate": 7.462613065326634e-05,
-      "loss": 5.3288,
-      "step": 25755
-    },
-    {
-      "epoch": 13.432073011734028,
-      "grad_norm": 1.417320966720581,
-      "learning_rate": 7.46251256281407e-05,
-      "loss": 4.9557,
-      "step": 25756
-    },
-    {
-      "epoch": 13.432594524119947,
-      "grad_norm": 1.3907581567764282,
-      "learning_rate": 7.462412060301508e-05,
-      "loss": 5.6044,
-      "step": 25757
-    },
-    {
-      "epoch": 13.433116036505867,
-      "grad_norm": 1.3842668533325195,
-      "learning_rate": 7.462311557788944e-05,
-      "loss": 5.414,
-      "step": 25758
-    },
-    {
-      "epoch": 13.433637548891786,
-      "grad_norm": 1.4100455045700073,
-      "learning_rate": 7.462211055276382e-05,
-      "loss": 5.4194,
-      "step": 25759
-    },
-    {
-      "epoch": 13.434159061277706,
-      "grad_norm": 1.539253830909729,
-      "learning_rate": 7.462110552763818e-05,
-      "loss": 5.1229,
-      "step": 25760
-    },
-    {
-      "epoch": 13.434680573663625,
-      "grad_norm": 1.4331300258636475,
-      "learning_rate": 7.462010050251256e-05,
-      "loss": 5.6595,
-      "step": 25761
-    },
-    {
-      "epoch": 13.435202086049543,
-      "grad_norm": 1.5121443271636963,
-      "learning_rate": 7.461909547738694e-05,
-      "loss": 5.4901,
-      "step": 25762
-    },
-    {
-      "epoch": 13.435723598435462,
-      "grad_norm": 1.3742835521697998,
-      "learning_rate": 7.461809045226132e-05,
-      "loss": 5.4985,
-      "step": 25763
-    },
-    {
-      "epoch": 13.436245110821382,
-      "grad_norm": 1.4671132564544678,
-      "learning_rate": 7.461708542713568e-05,
-      "loss": 5.3447,
-      "step": 25764
-    },
-    {
-      "epoch": 13.436766623207301,
-      "grad_norm": 1.4815346002578735,
-      "learning_rate": 7.461608040201006e-05,
-      "loss": 5.7226,
-      "step": 25765
-    },
-    {
-      "epoch": 13.43728813559322,
-      "grad_norm": 1.3589967489242554,
-      "learning_rate": 7.461507537688442e-05,
-      "loss": 5.3828,
-      "step": 25766
-    },
-    {
-      "epoch": 13.43780964797914,
-      "grad_norm": 1.5116108655929565,
-      "learning_rate": 7.46140703517588e-05,
-      "loss": 5.4127,
-      "step": 25767
-    },
-    {
-      "epoch": 13.438331160365058,
-      "grad_norm": 1.502476453781128,
-      "learning_rate": 7.461306532663317e-05,
-      "loss": 5.5417,
-      "step": 25768
-    },
-    {
-      "epoch": 13.438852672750977,
-      "grad_norm": 1.4385216236114502,
-      "learning_rate": 7.461206030150753e-05,
-      "loss": 5.4728,
-      "step": 25769
-    },
-    {
-      "epoch": 13.439374185136897,
-      "grad_norm": 1.3518720865249634,
-      "learning_rate": 7.461105527638191e-05,
-      "loss": 5.6282,
-      "step": 25770
-    },
-    {
-      "epoch": 13.439895697522816,
-      "grad_norm": 1.640574336051941,
-      "learning_rate": 7.461005025125629e-05,
-      "loss": 4.5703,
-      "step": 25771
-    },
-    {
-      "epoch": 13.440417209908736,
-      "grad_norm": 1.4608571529388428,
-      "learning_rate": 7.460904522613066e-05,
-      "loss": 5.3401,
-      "step": 25772
-    },
-    {
-      "epoch": 13.440938722294655,
-      "grad_norm": 1.4023679494857788,
-      "learning_rate": 7.460804020100503e-05,
-      "loss": 5.4572,
-      "step": 25773
-    },
-    {
-      "epoch": 13.441460234680573,
-      "grad_norm": 1.3250154256820679,
-      "learning_rate": 7.46070351758794e-05,
-      "loss": 5.3791,
-      "step": 25774
-    },
-    {
-      "epoch": 13.441981747066492,
-      "grad_norm": 1.5098063945770264,
-      "learning_rate": 7.460603015075377e-05,
-      "loss": 4.9468,
-      "step": 25775
-    },
-    {
-      "epoch": 13.442503259452412,
-      "grad_norm": 1.6293339729309082,
-      "learning_rate": 7.460502512562815e-05,
-      "loss": 5.5215,
-      "step": 25776
-    },
-    {
-      "epoch": 13.443024771838331,
-      "grad_norm": 1.417500615119934,
-      "learning_rate": 7.460402010050251e-05,
-      "loss": 4.9634,
-      "step": 25777
-    },
-    {
-      "epoch": 13.44354628422425,
-      "grad_norm": 1.370314598083496,
-      "learning_rate": 7.460301507537689e-05,
-      "loss": 5.3518,
-      "step": 25778
-    },
-    {
-      "epoch": 13.44406779661017,
-      "grad_norm": 1.463358759880066,
-      "learning_rate": 7.460201005025126e-05,
-      "loss": 5.3308,
-      "step": 25779
-    },
-    {
-      "epoch": 13.444589308996088,
-      "grad_norm": 1.4858185052871704,
-      "learning_rate": 7.460100502512563e-05,
-      "loss": 5.5331,
-      "step": 25780
-    },
-    {
-      "epoch": 13.445110821382007,
-      "grad_norm": 1.423521876335144,
-      "learning_rate": 7.46e-05,
-      "loss": 5.1747,
-      "step": 25781
-    },
-    {
-      "epoch": 13.445632333767927,
-      "grad_norm": 1.3970584869384766,
-      "learning_rate": 7.459899497487438e-05,
-      "loss": 5.4251,
-      "step": 25782
-    },
-    {
-      "epoch": 13.446153846153846,
-      "grad_norm": 1.458516240119934,
-      "learning_rate": 7.459798994974875e-05,
-      "loss": 5.3551,
-      "step": 25783
-    },
-    {
-      "epoch": 13.446675358539766,
-      "grad_norm": 1.4815362691879272,
-      "learning_rate": 7.459698492462312e-05,
-      "loss": 5.6472,
-      "step": 25784
-    },
-    {
-      "epoch": 13.447196870925685,
-      "grad_norm": 1.5546815395355225,
-      "learning_rate": 7.45959798994975e-05,
-      "loss": 5.0378,
-      "step": 25785
-    },
-    {
-      "epoch": 13.447718383311603,
-      "grad_norm": 1.4199405908584595,
-      "learning_rate": 7.459497487437186e-05,
-      "loss": 5.5287,
-      "step": 25786
-    },
-    {
-      "epoch": 13.448239895697522,
-      "grad_norm": 1.3967225551605225,
-      "learning_rate": 7.459396984924624e-05,
-      "loss": 5.3675,
-      "step": 25787
-    },
-    {
-      "epoch": 13.448761408083442,
-      "grad_norm": 1.409605622291565,
-      "learning_rate": 7.45929648241206e-05,
-      "loss": 5.3944,
-      "step": 25788
-    },
-    {
-      "epoch": 13.449282920469361,
-      "grad_norm": 1.427430272102356,
-      "learning_rate": 7.459195979899498e-05,
-      "loss": 5.5058,
-      "step": 25789
-    },
-    {
-      "epoch": 13.44980443285528,
-      "grad_norm": 1.5583319664001465,
-      "learning_rate": 7.459095477386934e-05,
-      "loss": 5.4593,
-      "step": 25790
-    },
-    {
-      "epoch": 13.4503259452412,
-      "grad_norm": 1.5089486837387085,
-      "learning_rate": 7.458994974874372e-05,
-      "loss": 5.2153,
-      "step": 25791
-    },
-    {
-      "epoch": 13.450847457627118,
-      "grad_norm": 1.4515163898468018,
-      "learning_rate": 7.45889447236181e-05,
-      "loss": 5.64,
-      "step": 25792
-    },
-    {
-      "epoch": 13.451368970013037,
-      "grad_norm": 1.5082346200942993,
-      "learning_rate": 7.458793969849248e-05,
-      "loss": 5.4596,
-      "step": 25793
-    },
-    {
-      "epoch": 13.451890482398957,
-      "grad_norm": 1.3669602870941162,
-      "learning_rate": 7.458693467336684e-05,
-      "loss": 5.0045,
-      "step": 25794
-    },
-    {
-      "epoch": 13.452411994784876,
-      "grad_norm": 1.5132296085357666,
-      "learning_rate": 7.458592964824122e-05,
-      "loss": 4.9038,
-      "step": 25795
-    },
-    {
-      "epoch": 13.452933507170796,
-      "grad_norm": 1.403357744216919,
-      "learning_rate": 7.458492462311558e-05,
-      "loss": 5.2711,
-      "step": 25796
-    },
-    {
-      "epoch": 13.453455019556715,
-      "grad_norm": 1.3375333547592163,
-      "learning_rate": 7.458391959798995e-05,
-      "loss": 5.4691,
-      "step": 25797
-    },
-    {
-      "epoch": 13.453976531942633,
-      "grad_norm": 1.4695682525634766,
-      "learning_rate": 7.458291457286433e-05,
-      "loss": 5.3132,
-      "step": 25798
-    },
-    {
-      "epoch": 13.454498044328552,
-      "grad_norm": 1.3572773933410645,
-      "learning_rate": 7.458190954773869e-05,
-      "loss": 5.6159,
-      "step": 25799
-    },
-    {
-      "epoch": 13.455019556714472,
-      "grad_norm": 1.4240750074386597,
-      "learning_rate": 7.458090452261307e-05,
-      "loss": 5.4923,
-      "step": 25800
-    },
-    {
-      "epoch": 13.455541069100391,
-      "grad_norm": 1.3610186576843262,
-      "learning_rate": 7.457989949748743e-05,
-      "loss": 5.7566,
-      "step": 25801
-    },
-    {
-      "epoch": 13.45606258148631,
-      "grad_norm": 1.3745728731155396,
-      "learning_rate": 7.457889447236181e-05,
-      "loss": 5.7702,
-      "step": 25802
-    },
-    {
-      "epoch": 13.45658409387223,
-      "grad_norm": 1.431480050086975,
-      "learning_rate": 7.457788944723619e-05,
-      "loss": 5.4793,
-      "step": 25803
-    },
-    {
-      "epoch": 13.457105606258148,
-      "grad_norm": 1.4504729509353638,
-      "learning_rate": 7.457688442211057e-05,
-      "loss": 5.2754,
-      "step": 25804
-    },
-    {
-      "epoch": 13.457627118644067,
-      "grad_norm": 1.2721911668777466,
-      "learning_rate": 7.457587939698493e-05,
-      "loss": 5.7739,
-      "step": 25805
-    },
-    {
-      "epoch": 13.458148631029987,
-      "grad_norm": 1.5029208660125732,
-      "learning_rate": 7.457487437185931e-05,
-      "loss": 5.0333,
-      "step": 25806
-    },
-    {
-      "epoch": 13.458670143415906,
-      "grad_norm": 1.395711898803711,
-      "learning_rate": 7.457386934673367e-05,
-      "loss": 5.5502,
-      "step": 25807
-    },
-    {
-      "epoch": 13.459191655801826,
-      "grad_norm": 1.4403815269470215,
-      "learning_rate": 7.457286432160805e-05,
-      "loss": 5.3764,
-      "step": 25808
-    },
-    {
-      "epoch": 13.459713168187744,
-      "grad_norm": 1.4121901988983154,
-      "learning_rate": 7.457185929648241e-05,
-      "loss": 5.5473,
-      "step": 25809
-    },
-    {
-      "epoch": 13.460234680573663,
-      "grad_norm": 1.4377591609954834,
-      "learning_rate": 7.457085427135678e-05,
-      "loss": 5.4363,
-      "step": 25810
-    },
-    {
-      "epoch": 13.460756192959582,
-      "grad_norm": 1.5347790718078613,
-      "learning_rate": 7.456984924623116e-05,
-      "loss": 5.0654,
-      "step": 25811
-    },
-    {
-      "epoch": 13.461277705345502,
-      "grad_norm": 1.4654875993728638,
-      "learning_rate": 7.456884422110553e-05,
-      "loss": 5.4834,
-      "step": 25812
-    },
-    {
-      "epoch": 13.461799217731421,
-      "grad_norm": 1.3891584873199463,
-      "learning_rate": 7.456783919597991e-05,
-      "loss": 5.6157,
-      "step": 25813
-    },
-    {
-      "epoch": 13.46232073011734,
-      "grad_norm": 1.3398462533950806,
-      "learning_rate": 7.456683417085428e-05,
-      "loss": 5.6903,
-      "step": 25814
-    },
-    {
-      "epoch": 13.46284224250326,
-      "grad_norm": 1.5170092582702637,
-      "learning_rate": 7.456582914572865e-05,
-      "loss": 5.6203,
-      "step": 25815
-    },
-    {
-      "epoch": 13.463363754889178,
-      "grad_norm": 1.4461970329284668,
-      "learning_rate": 7.456482412060302e-05,
-      "loss": 4.7269,
-      "step": 25816
-    },
-    {
-      "epoch": 13.463885267275097,
-      "grad_norm": 1.4339324235916138,
-      "learning_rate": 7.45638190954774e-05,
-      "loss": 5.6637,
-      "step": 25817
-    },
-    {
-      "epoch": 13.464406779661017,
-      "grad_norm": 1.3859716653823853,
-      "learning_rate": 7.456281407035176e-05,
-      "loss": 5.6405,
-      "step": 25818
-    },
-    {
-      "epoch": 13.464928292046936,
-      "grad_norm": 1.4901890754699707,
-      "learning_rate": 7.456180904522614e-05,
-      "loss": 5.0904,
-      "step": 25819
-    },
-    {
-      "epoch": 13.465449804432856,
-      "grad_norm": 1.5199823379516602,
-      "learning_rate": 7.45608040201005e-05,
-      "loss": 5.5266,
-      "step": 25820
-    },
-    {
-      "epoch": 13.465971316818774,
-      "grad_norm": 1.4449909925460815,
-      "learning_rate": 7.455979899497488e-05,
-      "loss": 5.2191,
-      "step": 25821
-    },
-    {
-      "epoch": 13.466492829204693,
-      "grad_norm": 1.3983596563339233,
-      "learning_rate": 7.455879396984924e-05,
-      "loss": 5.9131,
-      "step": 25822
-    },
-    {
-      "epoch": 13.467014341590613,
-      "grad_norm": 1.4448316097259521,
-      "learning_rate": 7.455778894472362e-05,
-      "loss": 5.4456,
-      "step": 25823
-    },
-    {
-      "epoch": 13.467535853976532,
-      "grad_norm": 1.5587831735610962,
-      "learning_rate": 7.4556783919598e-05,
-      "loss": 5.2695,
-      "step": 25824
-    },
-    {
-      "epoch": 13.468057366362451,
-      "grad_norm": 1.5552594661712646,
-      "learning_rate": 7.455577889447236e-05,
-      "loss": 5.2706,
-      "step": 25825
-    },
-    {
-      "epoch": 13.468578878748371,
-      "grad_norm": 1.5522208213806152,
-      "learning_rate": 7.455477386934674e-05,
-      "loss": 5.2447,
-      "step": 25826
-    },
-    {
-      "epoch": 13.469100391134289,
-      "grad_norm": 1.4148774147033691,
-      "learning_rate": 7.45537688442211e-05,
-      "loss": 5.3895,
-      "step": 25827
-    },
-    {
-      "epoch": 13.469621903520208,
-      "grad_norm": 1.3500038385391235,
-      "learning_rate": 7.455276381909548e-05,
-      "loss": 5.7849,
-      "step": 25828
-    },
-    {
-      "epoch": 13.470143415906128,
-      "grad_norm": 1.352275013923645,
-      "learning_rate": 7.455175879396985e-05,
-      "loss": 4.8507,
-      "step": 25829
-    },
-    {
-      "epoch": 13.470664928292047,
-      "grad_norm": 1.4313663244247437,
-      "learning_rate": 7.455075376884423e-05,
-      "loss": 5.5664,
-      "step": 25830
-    },
-    {
-      "epoch": 13.471186440677966,
-      "grad_norm": 1.4129598140716553,
-      "learning_rate": 7.454974874371859e-05,
-      "loss": 5.6285,
-      "step": 25831
-    },
-    {
-      "epoch": 13.471707953063886,
-      "grad_norm": 1.4581941366195679,
-      "learning_rate": 7.454874371859297e-05,
-      "loss": 5.4077,
-      "step": 25832
-    },
-    {
-      "epoch": 13.472229465449804,
-      "grad_norm": 1.4769515991210938,
-      "learning_rate": 7.454773869346735e-05,
-      "loss": 5.3971,
-      "step": 25833
-    },
-    {
-      "epoch": 13.472750977835723,
-      "grad_norm": 1.4114179611206055,
-      "learning_rate": 7.454673366834172e-05,
-      "loss": 5.6117,
-      "step": 25834
-    },
-    {
-      "epoch": 13.473272490221643,
-      "grad_norm": 1.5007543563842773,
-      "learning_rate": 7.454572864321609e-05,
-      "loss": 5.0904,
-      "step": 25835
-    },
-    {
-      "epoch": 13.473794002607562,
-      "grad_norm": 1.4294143915176392,
-      "learning_rate": 7.454472361809045e-05,
-      "loss": 5.1709,
-      "step": 25836
-    },
-    {
-      "epoch": 13.474315514993481,
-      "grad_norm": 1.4615232944488525,
-      "learning_rate": 7.454371859296483e-05,
-      "loss": 5.5042,
-      "step": 25837
-    },
-    {
-      "epoch": 13.474837027379401,
-      "grad_norm": 1.4054678678512573,
-      "learning_rate": 7.45427135678392e-05,
-      "loss": 5.5467,
-      "step": 25838
-    },
-    {
-      "epoch": 13.475358539765319,
-      "grad_norm": 1.4967602491378784,
-      "learning_rate": 7.454170854271357e-05,
-      "loss": 5.4114,
-      "step": 25839
-    },
-    {
-      "epoch": 13.475880052151238,
-      "grad_norm": 1.4218305349349976,
-      "learning_rate": 7.454070351758794e-05,
-      "loss": 5.7032,
-      "step": 25840
-    },
-    {
-      "epoch": 13.476401564537158,
-      "grad_norm": 1.4123295545578003,
-      "learning_rate": 7.453969849246231e-05,
-      "loss": 5.2423,
-      "step": 25841
-    },
-    {
-      "epoch": 13.476923076923077,
-      "grad_norm": 1.3700181245803833,
-      "learning_rate": 7.453869346733668e-05,
-      "loss": 5.531,
-      "step": 25842
-    },
-    {
-      "epoch": 13.477444589308996,
-      "grad_norm": 1.4921541213989258,
-      "learning_rate": 7.453768844221106e-05,
-      "loss": 5.387,
-      "step": 25843
-    },
-    {
-      "epoch": 13.477966101694916,
-      "grad_norm": 1.4084131717681885,
-      "learning_rate": 7.453668341708543e-05,
-      "loss": 5.7106,
-      "step": 25844
-    },
-    {
-      "epoch": 13.478487614080834,
-      "grad_norm": 1.5282868146896362,
-      "learning_rate": 7.453567839195981e-05,
-      "loss": 5.2512,
-      "step": 25845
-    },
-    {
-      "epoch": 13.479009126466753,
-      "grad_norm": 1.5084800720214844,
-      "learning_rate": 7.453467336683418e-05,
-      "loss": 4.4248,
-      "step": 25846
-    },
-    {
-      "epoch": 13.479530638852673,
-      "grad_norm": 1.469380497932434,
-      "learning_rate": 7.453366834170855e-05,
-      "loss": 5.341,
-      "step": 25847
-    },
-    {
-      "epoch": 13.480052151238592,
-      "grad_norm": 1.4949133396148682,
-      "learning_rate": 7.453266331658292e-05,
-      "loss": 5.0434,
-      "step": 25848
-    },
-    {
-      "epoch": 13.480573663624511,
-      "grad_norm": 1.4612425565719604,
-      "learning_rate": 7.453165829145728e-05,
-      "loss": 5.7477,
-      "step": 25849
-    },
-    {
-      "epoch": 13.481095176010431,
-      "grad_norm": 1.4086154699325562,
-      "learning_rate": 7.453065326633166e-05,
-      "loss": 5.2406,
-      "step": 25850
-    },
-    {
-      "epoch": 13.481616688396349,
-      "grad_norm": 1.4912303686141968,
-      "learning_rate": 7.452964824120603e-05,
-      "loss": 5.5551,
-      "step": 25851
-    },
-    {
-      "epoch": 13.482138200782268,
-      "grad_norm": 1.4745655059814453,
-      "learning_rate": 7.45286432160804e-05,
-      "loss": 5.0839,
-      "step": 25852
-    },
-    {
-      "epoch": 13.482659713168188,
-      "grad_norm": 1.4148823022842407,
-      "learning_rate": 7.452763819095477e-05,
-      "loss": 5.3035,
-      "step": 25853
-    },
-    {
-      "epoch": 13.483181225554107,
-      "grad_norm": 1.4886530637741089,
-      "learning_rate": 7.452663316582915e-05,
-      "loss": 5.3835,
-      "step": 25854
-    },
-    {
-      "epoch": 13.483702737940026,
-      "grad_norm": 1.4907963275909424,
-      "learning_rate": 7.452562814070352e-05,
-      "loss": 5.2467,
-      "step": 25855
-    },
-    {
-      "epoch": 13.484224250325946,
-      "grad_norm": 1.4680252075195312,
-      "learning_rate": 7.45246231155779e-05,
-      "loss": 4.8174,
-      "step": 25856
-    },
-    {
-      "epoch": 13.484745762711864,
-      "grad_norm": 1.5300137996673584,
-      "learning_rate": 7.452361809045226e-05,
-      "loss": 4.9515,
-      "step": 25857
-    },
-    {
-      "epoch": 13.485267275097783,
-      "grad_norm": 1.539993405342102,
-      "learning_rate": 7.452261306532664e-05,
-      "loss": 5.6634,
-      "step": 25858
-    },
-    {
-      "epoch": 13.485788787483703,
-      "grad_norm": 1.51149582862854,
-      "learning_rate": 7.452160804020101e-05,
-      "loss": 5.2168,
-      "step": 25859
-    },
-    {
-      "epoch": 13.486310299869622,
-      "grad_norm": 1.3279024362564087,
-      "learning_rate": 7.452060301507538e-05,
-      "loss": 5.7916,
-      "step": 25860
-    },
-    {
-      "epoch": 13.486831812255542,
-      "grad_norm": 1.4908638000488281,
-      "learning_rate": 7.451959798994975e-05,
-      "loss": 4.9764,
-      "step": 25861
-    },
-    {
-      "epoch": 13.487353324641461,
-      "grad_norm": 1.4590305089950562,
-      "learning_rate": 7.451859296482411e-05,
-      "loss": 5.5383,
-      "step": 25862
-    },
-    {
-      "epoch": 13.487874837027379,
-      "grad_norm": 1.5093717575073242,
-      "learning_rate": 7.451758793969849e-05,
-      "loss": 5.4227,
-      "step": 25863
-    },
-    {
-      "epoch": 13.488396349413298,
-      "grad_norm": 1.4809075593948364,
-      "learning_rate": 7.451658291457287e-05,
-      "loss": 4.8143,
-      "step": 25864
-    },
-    {
-      "epoch": 13.488917861799218,
-      "grad_norm": 1.5347845554351807,
-      "learning_rate": 7.451557788944725e-05,
-      "loss": 5.4025,
-      "step": 25865
-    },
-    {
-      "epoch": 13.489439374185137,
-      "grad_norm": 1.4081379175186157,
-      "learning_rate": 7.451457286432161e-05,
-      "loss": 5.3954,
-      "step": 25866
-    },
-    {
-      "epoch": 13.489960886571057,
-      "grad_norm": 1.5008811950683594,
-      "learning_rate": 7.451356783919599e-05,
-      "loss": 5.6215,
-      "step": 25867
-    },
-    {
-      "epoch": 13.490482398956976,
-      "grad_norm": 1.4237391948699951,
-      "learning_rate": 7.451256281407035e-05,
-      "loss": 5.5102,
-      "step": 25868
-    },
-    {
-      "epoch": 13.491003911342894,
-      "grad_norm": 1.5548973083496094,
-      "learning_rate": 7.451155778894473e-05,
-      "loss": 5.1366,
-      "step": 25869
-    },
-    {
-      "epoch": 13.491525423728813,
-      "grad_norm": 1.4376118183135986,
-      "learning_rate": 7.45105527638191e-05,
-      "loss": 5.4914,
-      "step": 25870
-    },
-    {
-      "epoch": 13.492046936114733,
-      "grad_norm": 1.6306744813919067,
-      "learning_rate": 7.450954773869347e-05,
-      "loss": 4.7485,
-      "step": 25871
-    },
-    {
-      "epoch": 13.492568448500652,
-      "grad_norm": 1.4672616720199585,
-      "learning_rate": 7.450854271356784e-05,
-      "loss": 5.5098,
-      "step": 25872
-    },
-    {
-      "epoch": 13.493089960886572,
-      "grad_norm": 1.471908688545227,
-      "learning_rate": 7.450753768844222e-05,
-      "loss": 5.2155,
-      "step": 25873
-    },
-    {
-      "epoch": 13.493611473272491,
-      "grad_norm": 1.4668617248535156,
-      "learning_rate": 7.450653266331658e-05,
-      "loss": 5.2369,
-      "step": 25874
-    },
-    {
-      "epoch": 13.494132985658409,
-      "grad_norm": 1.474664330482483,
-      "learning_rate": 7.450552763819096e-05,
-      "loss": 5.6196,
-      "step": 25875
-    },
-    {
-      "epoch": 13.494654498044328,
-      "grad_norm": 1.4355820417404175,
-      "learning_rate": 7.450452261306534e-05,
-      "loss": 5.425,
-      "step": 25876
-    },
-    {
-      "epoch": 13.495176010430248,
-      "grad_norm": 1.4470003843307495,
-      "learning_rate": 7.45035175879397e-05,
-      "loss": 5.0652,
-      "step": 25877
-    },
-    {
-      "epoch": 13.495697522816167,
-      "grad_norm": 1.4440104961395264,
-      "learning_rate": 7.450251256281408e-05,
-      "loss": 5.0029,
-      "step": 25878
-    },
-    {
-      "epoch": 13.496219035202087,
-      "grad_norm": 1.4482178688049316,
-      "learning_rate": 7.450150753768844e-05,
-      "loss": 5.2514,
-      "step": 25879
-    },
-    {
-      "epoch": 13.496740547588006,
-      "grad_norm": 1.454601526260376,
-      "learning_rate": 7.450050251256282e-05,
-      "loss": 5.6576,
-      "step": 25880
-    },
-    {
-      "epoch": 13.497262059973924,
-      "grad_norm": 1.4809670448303223,
-      "learning_rate": 7.449949748743718e-05,
-      "loss": 5.494,
-      "step": 25881
-    },
-    {
-      "epoch": 13.497783572359843,
-      "grad_norm": 1.492385745048523,
-      "learning_rate": 7.449849246231156e-05,
-      "loss": 5.1223,
-      "step": 25882
-    },
-    {
-      "epoch": 13.498305084745763,
-      "grad_norm": 1.3625870943069458,
-      "learning_rate": 7.449748743718593e-05,
-      "loss": 5.704,
-      "step": 25883
-    },
-    {
-      "epoch": 13.498826597131682,
-      "grad_norm": 1.4367799758911133,
-      "learning_rate": 7.44964824120603e-05,
-      "loss": 5.4066,
-      "step": 25884
-    },
-    {
-      "epoch": 13.499348109517602,
-      "grad_norm": 1.4163209199905396,
-      "learning_rate": 7.449547738693468e-05,
-      "loss": 5.5574,
-      "step": 25885
-    },
-    {
-      "epoch": 13.499869621903521,
-      "grad_norm": 1.391627550125122,
-      "learning_rate": 7.449447236180906e-05,
-      "loss": 5.5572,
-      "step": 25886
-    },
-    {
-      "epoch": 13.500391134289439,
-      "grad_norm": 1.5171618461608887,
-      "learning_rate": 7.449346733668342e-05,
-      "loss": 5.4884,
-      "step": 25887
-    },
-    {
-      "epoch": 13.500912646675358,
-      "grad_norm": 1.4364819526672363,
-      "learning_rate": 7.44924623115578e-05,
-      "loss": 5.3345,
-      "step": 25888
-    },
-    {
-      "epoch": 13.501434159061278,
-      "grad_norm": 1.4262064695358276,
-      "learning_rate": 7.449145728643217e-05,
-      "loss": 5.7556,
-      "step": 25889
-    },
-    {
-      "epoch": 13.501955671447197,
-      "grad_norm": 1.540083646774292,
-      "learning_rate": 7.449045226130653e-05,
-      "loss": 5.6249,
-      "step": 25890
-    },
-    {
-      "epoch": 13.502477183833117,
-      "grad_norm": 1.4518067836761475,
-      "learning_rate": 7.448944723618091e-05,
-      "loss": 5.5693,
-      "step": 25891
-    },
-    {
-      "epoch": 13.502998696219036,
-      "grad_norm": 1.4950978755950928,
-      "learning_rate": 7.448844221105527e-05,
-      "loss": 5.6473,
-      "step": 25892
-    },
-    {
-      "epoch": 13.503520208604954,
-      "grad_norm": 1.3784857988357544,
-      "learning_rate": 7.448743718592965e-05,
-      "loss": 4.7384,
-      "step": 25893
-    },
-    {
-      "epoch": 13.504041720990873,
-      "grad_norm": 1.405648946762085,
-      "learning_rate": 7.448643216080401e-05,
-      "loss": 5.5047,
-      "step": 25894
-    },
-    {
-      "epoch": 13.504563233376793,
-      "grad_norm": 1.4763556718826294,
-      "learning_rate": 7.448542713567839e-05,
-      "loss": 5.6243,
-      "step": 25895
-    },
-    {
-      "epoch": 13.505084745762712,
-      "grad_norm": 1.3729089498519897,
-      "learning_rate": 7.448442211055277e-05,
-      "loss": 5.6672,
-      "step": 25896
-    },
-    {
-      "epoch": 13.505606258148632,
-      "grad_norm": 1.5258829593658447,
-      "learning_rate": 7.448341708542715e-05,
-      "loss": 5.6468,
-      "step": 25897
-    },
-    {
-      "epoch": 13.506127770534551,
-      "grad_norm": 1.375726580619812,
-      "learning_rate": 7.448241206030151e-05,
-      "loss": 5.5837,
-      "step": 25898
-    },
-    {
-      "epoch": 13.506649282920469,
-      "grad_norm": 1.3957444429397583,
-      "learning_rate": 7.448140703517589e-05,
-      "loss": 5.3249,
-      "step": 25899
-    },
-    {
-      "epoch": 13.507170795306388,
-      "grad_norm": 1.3999435901641846,
-      "learning_rate": 7.448040201005025e-05,
-      "loss": 5.3531,
-      "step": 25900
-    },
-    {
-      "epoch": 13.507692307692308,
-      "grad_norm": 1.4674938917160034,
-      "learning_rate": 7.447939698492463e-05,
-      "loss": 5.2208,
-      "step": 25901
-    },
-    {
-      "epoch": 13.508213820078227,
-      "grad_norm": 1.6256000995635986,
-      "learning_rate": 7.4478391959799e-05,
-      "loss": 5.0913,
-      "step": 25902
-    },
-    {
-      "epoch": 13.508735332464147,
-      "grad_norm": 1.3792133331298828,
-      "learning_rate": 7.447738693467336e-05,
-      "loss": 4.1037,
-      "step": 25903
-    },
-    {
-      "epoch": 13.509256844850064,
-      "grad_norm": 1.4572721719741821,
-      "learning_rate": 7.447638190954774e-05,
-      "loss": 5.3904,
-      "step": 25904
-    },
-    {
-      "epoch": 13.509778357235984,
-      "grad_norm": 1.5057754516601562,
-      "learning_rate": 7.447537688442212e-05,
-      "loss": 5.6223,
-      "step": 25905
-    },
-    {
-      "epoch": 13.510299869621903,
-      "grad_norm": 1.455460548400879,
-      "learning_rate": 7.44743718592965e-05,
-      "loss": 5.7147,
-      "step": 25906
-    },
-    {
-      "epoch": 13.510821382007823,
-      "grad_norm": 1.5177149772644043,
-      "learning_rate": 7.447336683417086e-05,
-      "loss": 5.4809,
-      "step": 25907
-    },
-    {
-      "epoch": 13.511342894393742,
-      "grad_norm": 1.473115086555481,
-      "learning_rate": 7.447236180904524e-05,
-      "loss": 5.288,
-      "step": 25908
-    },
-    {
-      "epoch": 13.511864406779662,
-      "grad_norm": 1.4698179960250854,
-      "learning_rate": 7.44713567839196e-05,
-      "loss": 5.7207,
-      "step": 25909
-    },
-    {
-      "epoch": 13.512385919165581,
-      "grad_norm": 1.609645962715149,
-      "learning_rate": 7.447035175879398e-05,
-      "loss": 5.231,
-      "step": 25910
-    },
-    {
-      "epoch": 13.512907431551499,
-      "grad_norm": 1.3645532131195068,
-      "learning_rate": 7.446934673366834e-05,
-      "loss": 5.5301,
-      "step": 25911
-    },
-    {
-      "epoch": 13.513428943937418,
-      "grad_norm": 1.5437088012695312,
-      "learning_rate": 7.446834170854272e-05,
-      "loss": 4.8578,
-      "step": 25912
-    },
-    {
-      "epoch": 13.513950456323338,
-      "grad_norm": 1.5836268663406372,
-      "learning_rate": 7.446733668341708e-05,
-      "loss": 5.2183,
-      "step": 25913
-    },
-    {
-      "epoch": 13.514471968709257,
-      "grad_norm": 1.4085338115692139,
-      "learning_rate": 7.446633165829146e-05,
-      "loss": 5.119,
-      "step": 25914
-    },
-    {
-      "epoch": 13.514993481095177,
-      "grad_norm": 1.5094655752182007,
-      "learning_rate": 7.446532663316583e-05,
-      "loss": 5.3904,
-      "step": 25915
-    },
-    {
-      "epoch": 13.515514993481094,
-      "grad_norm": 1.6087820529937744,
-      "learning_rate": 7.44643216080402e-05,
-      "loss": 4.8717,
-      "step": 25916
-    },
-    {
-      "epoch": 13.516036505867014,
-      "grad_norm": 1.5217869281768799,
-      "learning_rate": 7.446331658291458e-05,
-      "loss": 5.3468,
-      "step": 25917
-    },
-    {
-      "epoch": 13.516558018252933,
-      "grad_norm": 1.4924753904342651,
-      "learning_rate": 7.446231155778895e-05,
-      "loss": 4.8923,
-      "step": 25918
-    },
-    {
-      "epoch": 13.517079530638853,
-      "grad_norm": 1.3980692625045776,
-      "learning_rate": 7.446130653266332e-05,
-      "loss": 5.2736,
-      "step": 25919
-    },
-    {
-      "epoch": 13.517601043024772,
-      "grad_norm": 1.458905816078186,
-      "learning_rate": 7.446030150753769e-05,
-      "loss": 5.745,
-      "step": 25920
-    },
-    {
-      "epoch": 13.518122555410692,
-      "grad_norm": 1.4303532838821411,
-      "learning_rate": 7.445929648241207e-05,
-      "loss": 5.8091,
-      "step": 25921
-    },
-    {
-      "epoch": 13.518644067796611,
-      "grad_norm": 1.3283510208129883,
-      "learning_rate": 7.445829145728643e-05,
-      "loss": 5.8899,
-      "step": 25922
-    },
-    {
-      "epoch": 13.519165580182529,
-      "grad_norm": 1.4057729244232178,
-      "learning_rate": 7.445728643216081e-05,
-      "loss": 5.6153,
-      "step": 25923
-    },
-    {
-      "epoch": 13.519687092568448,
-      "grad_norm": 1.4421374797821045,
-      "learning_rate": 7.445628140703517e-05,
-      "loss": 5.0896,
-      "step": 25924
-    },
-    {
-      "epoch": 13.520208604954368,
-      "grad_norm": 1.449812650680542,
-      "learning_rate": 7.445527638190955e-05,
-      "loss": 5.7735,
-      "step": 25925
-    },
-    {
-      "epoch": 13.520730117340287,
-      "grad_norm": 1.495416522026062,
-      "learning_rate": 7.445427135678393e-05,
-      "loss": 5.7947,
-      "step": 25926
-    },
-    {
-      "epoch": 13.521251629726207,
-      "grad_norm": 1.3541388511657715,
-      "learning_rate": 7.44532663316583e-05,
-      "loss": 5.4728,
-      "step": 25927
-    },
-    {
-      "epoch": 13.521773142112124,
-      "grad_norm": 1.3956247568130493,
-      "learning_rate": 7.445226130653267e-05,
-      "loss": 5.465,
-      "step": 25928
-    },
-    {
-      "epoch": 13.522294654498044,
-      "grad_norm": 1.5294207334518433,
-      "learning_rate": 7.445125628140703e-05,
-      "loss": 5.2378,
-      "step": 25929
-    },
-    {
-      "epoch": 13.522816166883963,
-      "grad_norm": 1.555928111076355,
-      "learning_rate": 7.445025125628141e-05,
-      "loss": 5.1533,
-      "step": 25930
-    },
-    {
-      "epoch": 13.523337679269883,
-      "grad_norm": 1.605849266052246,
-      "learning_rate": 7.444924623115578e-05,
-      "loss": 5.5037,
-      "step": 25931
-    },
-    {
-      "epoch": 13.523859191655802,
-      "grad_norm": 1.4362053871154785,
-      "learning_rate": 7.444824120603015e-05,
-      "loss": 5.3092,
-      "step": 25932
-    },
-    {
-      "epoch": 13.524380704041722,
-      "grad_norm": 1.5531110763549805,
-      "learning_rate": 7.444723618090452e-05,
-      "loss": 5.1146,
-      "step": 25933
-    },
-    {
-      "epoch": 13.52490221642764,
-      "grad_norm": 1.5433335304260254,
-      "learning_rate": 7.44462311557789e-05,
-      "loss": 5.4141,
-      "step": 25934
-    },
-    {
-      "epoch": 13.525423728813559,
-      "grad_norm": 1.428383231163025,
-      "learning_rate": 7.444522613065326e-05,
-      "loss": 5.8697,
-      "step": 25935
-    },
-    {
-      "epoch": 13.525945241199478,
-      "grad_norm": 1.474097728729248,
-      "learning_rate": 7.444422110552764e-05,
-      "loss": 5.1844,
-      "step": 25936
-    },
-    {
-      "epoch": 13.526466753585398,
-      "grad_norm": 1.5129460096359253,
-      "learning_rate": 7.444321608040202e-05,
-      "loss": 5.4424,
-      "step": 25937
-    },
-    {
-      "epoch": 13.526988265971317,
-      "grad_norm": 1.5997051000595093,
-      "learning_rate": 7.44422110552764e-05,
-      "loss": 4.9856,
-      "step": 25938
-    },
-    {
-      "epoch": 13.527509778357237,
-      "grad_norm": 1.4350281953811646,
-      "learning_rate": 7.444120603015076e-05,
-      "loss": 5.5749,
-      "step": 25939
-    },
-    {
-      "epoch": 13.528031290743154,
-      "grad_norm": 1.448862910270691,
-      "learning_rate": 7.444020100502514e-05,
-      "loss": 4.8669,
-      "step": 25940
-    },
-    {
-      "epoch": 13.528552803129074,
-      "grad_norm": 1.5424367189407349,
-      "learning_rate": 7.44391959798995e-05,
-      "loss": 5.1068,
-      "step": 25941
-    },
-    {
-      "epoch": 13.529074315514993,
-      "grad_norm": 1.4703173637390137,
-      "learning_rate": 7.443819095477387e-05,
-      "loss": 5.5195,
-      "step": 25942
-    },
-    {
-      "epoch": 13.529595827900913,
-      "grad_norm": 1.4776593446731567,
-      "learning_rate": 7.443718592964824e-05,
-      "loss": 5.427,
-      "step": 25943
-    },
-    {
-      "epoch": 13.530117340286832,
-      "grad_norm": 1.4899742603302002,
-      "learning_rate": 7.443618090452261e-05,
-      "loss": 4.5995,
-      "step": 25944
-    },
-    {
-      "epoch": 13.530638852672752,
-      "grad_norm": 1.577268362045288,
-      "learning_rate": 7.443517587939699e-05,
-      "loss": 5.1151,
-      "step": 25945
-    },
-    {
-      "epoch": 13.53116036505867,
-      "grad_norm": 1.6029402017593384,
-      "learning_rate": 7.443417085427136e-05,
-      "loss": 4.8939,
-      "step": 25946
-    },
-    {
-      "epoch": 13.531681877444589,
-      "grad_norm": 1.4128305912017822,
-      "learning_rate": 7.443316582914574e-05,
-      "loss": 5.5488,
-      "step": 25947
-    },
-    {
-      "epoch": 13.532203389830508,
-      "grad_norm": 1.4772982597351074,
-      "learning_rate": 7.44321608040201e-05,
-      "loss": 5.7103,
-      "step": 25948
-    },
-    {
-      "epoch": 13.532724902216428,
-      "grad_norm": 1.4211106300354004,
-      "learning_rate": 7.443115577889448e-05,
-      "loss": 5.3892,
-      "step": 25949
-    },
-    {
-      "epoch": 13.533246414602347,
-      "grad_norm": 1.4242748022079468,
-      "learning_rate": 7.443015075376885e-05,
-      "loss": 5.5066,
-      "step": 25950
-    },
-    {
-      "epoch": 13.533767926988267,
-      "grad_norm": 1.4830518960952759,
-      "learning_rate": 7.442914572864323e-05,
-      "loss": 5.3333,
-      "step": 25951
-    },
-    {
-      "epoch": 13.534289439374184,
-      "grad_norm": 1.561877727508545,
-      "learning_rate": 7.442814070351759e-05,
-      "loss": 4.8842,
-      "step": 25952
-    },
-    {
-      "epoch": 13.534810951760104,
-      "grad_norm": 1.5283575057983398,
-      "learning_rate": 7.442713567839197e-05,
-      "loss": 5.061,
-      "step": 25953
-    },
-    {
-      "epoch": 13.535332464146023,
-      "grad_norm": 1.4257550239562988,
-      "learning_rate": 7.442613065326633e-05,
-      "loss": 5.7237,
-      "step": 25954
-    },
-    {
-      "epoch": 13.535853976531943,
-      "grad_norm": 1.5343230962753296,
-      "learning_rate": 7.44251256281407e-05,
-      "loss": 5.3933,
-      "step": 25955
-    },
-    {
-      "epoch": 13.536375488917862,
-      "grad_norm": 1.4083856344223022,
-      "learning_rate": 7.442412060301507e-05,
-      "loss": 5.2668,
-      "step": 25956
-    },
-    {
-      "epoch": 13.536897001303782,
-      "grad_norm": 1.3650084733963013,
-      "learning_rate": 7.442311557788945e-05,
-      "loss": 5.5279,
-      "step": 25957
-    },
-    {
-      "epoch": 13.5374185136897,
-      "grad_norm": 1.374729037284851,
-      "learning_rate": 7.442211055276383e-05,
-      "loss": 5.3821,
-      "step": 25958
-    },
-    {
-      "epoch": 13.537940026075619,
-      "grad_norm": 1.503204107284546,
-      "learning_rate": 7.44211055276382e-05,
-      "loss": 5.2006,
-      "step": 25959
-    },
-    {
-      "epoch": 13.538461538461538,
-      "grad_norm": 1.4023096561431885,
-      "learning_rate": 7.442010050251257e-05,
-      "loss": 5.6624,
-      "step": 25960
-    },
-    {
-      "epoch": 13.538983050847458,
-      "grad_norm": 1.4590502977371216,
-      "learning_rate": 7.441909547738694e-05,
-      "loss": 5.3592,
-      "step": 25961
-    },
-    {
-      "epoch": 13.539504563233377,
-      "grad_norm": 1.388390302658081,
-      "learning_rate": 7.441809045226131e-05,
-      "loss": 5.764,
-      "step": 25962
-    },
-    {
-      "epoch": 13.540026075619297,
-      "grad_norm": 1.434746503829956,
-      "learning_rate": 7.441708542713568e-05,
-      "loss": 5.6813,
-      "step": 25963
-    },
-    {
-      "epoch": 13.540547588005214,
-      "grad_norm": 1.396411418914795,
-      "learning_rate": 7.441608040201006e-05,
-      "loss": 5.4462,
-      "step": 25964
-    },
-    {
-      "epoch": 13.541069100391134,
-      "grad_norm": 1.2860091924667358,
-      "learning_rate": 7.441507537688442e-05,
-      "loss": 5.6052,
-      "step": 25965
-    },
-    {
-      "epoch": 13.541590612777053,
-      "grad_norm": 1.4205468893051147,
-      "learning_rate": 7.44140703517588e-05,
-      "loss": 5.5528,
-      "step": 25966
-    },
-    {
-      "epoch": 13.542112125162973,
-      "grad_norm": 1.3812165260314941,
-      "learning_rate": 7.441306532663318e-05,
-      "loss": 5.7332,
-      "step": 25967
-    },
-    {
-      "epoch": 13.542633637548892,
-      "grad_norm": 1.4035179615020752,
-      "learning_rate": 7.441206030150754e-05,
-      "loss": 5.6863,
-      "step": 25968
-    },
-    {
-      "epoch": 13.543155149934812,
-      "grad_norm": 1.3843671083450317,
-      "learning_rate": 7.441105527638192e-05,
-      "loss": 5.7062,
-      "step": 25969
-    },
-    {
-      "epoch": 13.54367666232073,
-      "grad_norm": 1.481192946434021,
-      "learning_rate": 7.441005025125628e-05,
-      "loss": 5.4784,
-      "step": 25970
-    },
-    {
-      "epoch": 13.544198174706649,
-      "grad_norm": 1.3759831190109253,
-      "learning_rate": 7.440904522613066e-05,
-      "loss": 5.7803,
-      "step": 25971
-    },
-    {
-      "epoch": 13.544719687092568,
-      "grad_norm": 1.4097306728363037,
-      "learning_rate": 7.440804020100502e-05,
-      "loss": 5.3642,
-      "step": 25972
-    },
-    {
-      "epoch": 13.545241199478488,
-      "grad_norm": 1.600789189338684,
-      "learning_rate": 7.44070351758794e-05,
-      "loss": 5.1343,
-      "step": 25973
-    },
-    {
-      "epoch": 13.545762711864407,
-      "grad_norm": 1.5550453662872314,
-      "learning_rate": 7.440603015075377e-05,
-      "loss": 5.351,
-      "step": 25974
-    },
-    {
-      "epoch": 13.546284224250327,
-      "grad_norm": 1.4413334131240845,
-      "learning_rate": 7.440502512562814e-05,
-      "loss": 5.5572,
-      "step": 25975
-    },
-    {
-      "epoch": 13.546805736636244,
-      "grad_norm": 1.4505242109298706,
-      "learning_rate": 7.440402010050251e-05,
-      "loss": 5.5466,
-      "step": 25976
-    },
-    {
-      "epoch": 13.547327249022164,
-      "grad_norm": 1.6240147352218628,
-      "learning_rate": 7.440301507537689e-05,
-      "loss": 4.8758,
-      "step": 25977
-    },
-    {
-      "epoch": 13.547848761408083,
-      "grad_norm": 1.4561899900436401,
-      "learning_rate": 7.440201005025126e-05,
-      "loss": 5.3084,
-      "step": 25978
-    },
-    {
-      "epoch": 13.548370273794003,
-      "grad_norm": 1.5579581260681152,
-      "learning_rate": 7.440100502512564e-05,
-      "loss": 5.5399,
-      "step": 25979
-    },
-    {
-      "epoch": 13.548891786179922,
-      "grad_norm": 1.5164270401000977,
-      "learning_rate": 7.44e-05,
-      "loss": 5.4031,
-      "step": 25980
-    },
-    {
-      "epoch": 13.549413298565842,
-      "grad_norm": 1.5167006254196167,
-      "learning_rate": 7.439899497487438e-05,
-      "loss": 5.1222,
-      "step": 25981
-    },
-    {
-      "epoch": 13.54993481095176,
-      "grad_norm": 1.4125583171844482,
-      "learning_rate": 7.439798994974875e-05,
-      "loss": 5.4795,
-      "step": 25982
-    },
-    {
-      "epoch": 13.550456323337679,
-      "grad_norm": 1.407762050628662,
-      "learning_rate": 7.439698492462311e-05,
-      "loss": 5.3182,
-      "step": 25983
-    },
-    {
-      "epoch": 13.550977835723598,
-      "grad_norm": 1.5006707906723022,
-      "learning_rate": 7.439597989949749e-05,
-      "loss": 4.8142,
-      "step": 25984
-    },
-    {
-      "epoch": 13.551499348109518,
-      "grad_norm": 1.430970549583435,
-      "learning_rate": 7.439497487437185e-05,
-      "loss": 5.546,
-      "step": 25985
-    },
-    {
-      "epoch": 13.552020860495437,
-      "grad_norm": 1.4947845935821533,
-      "learning_rate": 7.439396984924623e-05,
-      "loss": 5.5284,
-      "step": 25986
-    },
-    {
-      "epoch": 13.552542372881355,
-      "grad_norm": 1.5008940696716309,
-      "learning_rate": 7.439296482412061e-05,
-      "loss": 5.0784,
-      "step": 25987
-    },
-    {
-      "epoch": 13.553063885267274,
-      "grad_norm": 1.4941920042037964,
-      "learning_rate": 7.439195979899499e-05,
-      "loss": 5.2119,
-      "step": 25988
-    },
-    {
-      "epoch": 13.553585397653194,
-      "grad_norm": 1.4802685976028442,
-      "learning_rate": 7.439095477386935e-05,
-      "loss": 5.187,
-      "step": 25989
-    },
-    {
-      "epoch": 13.554106910039113,
-      "grad_norm": 1.4996001720428467,
-      "learning_rate": 7.438994974874373e-05,
-      "loss": 5.2142,
-      "step": 25990
-    },
-    {
-      "epoch": 13.554628422425033,
-      "grad_norm": 1.4696357250213623,
-      "learning_rate": 7.43889447236181e-05,
-      "loss": 5.5947,
-      "step": 25991
-    },
-    {
-      "epoch": 13.555149934810952,
-      "grad_norm": 1.2747399806976318,
-      "learning_rate": 7.438793969849247e-05,
-      "loss": 4.9744,
-      "step": 25992
-    },
-    {
-      "epoch": 13.555671447196872,
-      "grad_norm": 1.4503189325332642,
-      "learning_rate": 7.438693467336684e-05,
-      "loss": 5.2261,
-      "step": 25993
-    },
-    {
-      "epoch": 13.55619295958279,
-      "grad_norm": 1.3531697988510132,
-      "learning_rate": 7.438592964824121e-05,
-      "loss": 5.592,
-      "step": 25994
-    },
-    {
-      "epoch": 13.556714471968709,
-      "grad_norm": 1.600414514541626,
-      "learning_rate": 7.438492462311558e-05,
-      "loss": 4.9958,
-      "step": 25995
-    },
-    {
-      "epoch": 13.557235984354628,
-      "grad_norm": 1.483246922492981,
-      "learning_rate": 7.438391959798994e-05,
-      "loss": 5.454,
-      "step": 25996
-    },
-    {
-      "epoch": 13.557757496740548,
-      "grad_norm": 1.4340903759002686,
-      "learning_rate": 7.438291457286432e-05,
-      "loss": 5.4076,
-      "step": 25997
-    },
-    {
-      "epoch": 13.558279009126467,
-      "grad_norm": 1.4346660375595093,
-      "learning_rate": 7.43819095477387e-05,
-      "loss": 5.5081,
-      "step": 25998
-    },
-    {
-      "epoch": 13.558800521512385,
-      "grad_norm": 1.6049302816390991,
-      "learning_rate": 7.438090452261308e-05,
-      "loss": 4.9579,
-      "step": 25999
-    },
-    {
-      "epoch": 13.559322033898304,
-      "grad_norm": 1.406704306602478,
-      "learning_rate": 7.437989949748744e-05,
-      "loss": 5.4594,
-      "step": 26000
-    },
-    {
-      "epoch": 13.559843546284224,
-      "grad_norm": 1.4006972312927246,
-      "learning_rate": 7.437889447236182e-05,
-      "loss": 5.6853,
-      "step": 26001
-    },
-    {
-      "epoch": 13.560365058670143,
-      "grad_norm": 1.4542219638824463,
-      "learning_rate": 7.437788944723618e-05,
-      "loss": 5.4594,
-      "step": 26002
-    },
-    {
-      "epoch": 13.560886571056063,
-      "grad_norm": 1.4597899913787842,
-      "learning_rate": 7.437688442211056e-05,
-      "loss": 5.8747,
-      "step": 26003
-    },
-    {
-      "epoch": 13.561408083441982,
-      "grad_norm": 1.4665014743804932,
-      "learning_rate": 7.437587939698492e-05,
-      "loss": 5.4685,
-      "step": 26004
-    },
-    {
-      "epoch": 13.561929595827902,
-      "grad_norm": 1.480733036994934,
-      "learning_rate": 7.43748743718593e-05,
-      "loss": 5.6536,
-      "step": 26005
-    },
-    {
-      "epoch": 13.56245110821382,
-      "grad_norm": 1.3248893022537231,
-      "learning_rate": 7.437386934673367e-05,
-      "loss": 5.8838,
-      "step": 26006
-    },
-    {
-      "epoch": 13.562972620599739,
-      "grad_norm": 1.4129865169525146,
-      "learning_rate": 7.437286432160804e-05,
-      "loss": 5.1572,
-      "step": 26007
-    },
-    {
-      "epoch": 13.563494132985658,
-      "grad_norm": 1.5340595245361328,
-      "learning_rate": 7.437185929648241e-05,
-      "loss": 5.5518,
-      "step": 26008
-    },
-    {
-      "epoch": 13.564015645371578,
-      "grad_norm": 1.4122991561889648,
-      "learning_rate": 7.437085427135679e-05,
-      "loss": 5.7039,
-      "step": 26009
-    },
-    {
-      "epoch": 13.564537157757497,
-      "grad_norm": 1.442515254020691,
-      "learning_rate": 7.436984924623116e-05,
-      "loss": 5.6728,
-      "step": 26010
-    },
-    {
-      "epoch": 13.565058670143415,
-      "grad_norm": 1.4544196128845215,
-      "learning_rate": 7.436884422110553e-05,
-      "loss": 5.6622,
-      "step": 26011
-    },
-    {
-      "epoch": 13.565580182529335,
-      "grad_norm": 1.5100643634796143,
-      "learning_rate": 7.43678391959799e-05,
-      "loss": 5.0699,
-      "step": 26012
-    },
-    {
-      "epoch": 13.566101694915254,
-      "grad_norm": 1.4016236066818237,
-      "learning_rate": 7.436683417085427e-05,
-      "loss": 5.2161,
-      "step": 26013
-    },
-    {
-      "epoch": 13.566623207301173,
-      "grad_norm": 1.5255712270736694,
-      "learning_rate": 7.436582914572865e-05,
-      "loss": 5.519,
-      "step": 26014
-    },
-    {
-      "epoch": 13.567144719687093,
-      "grad_norm": 1.4653573036193848,
-      "learning_rate": 7.436482412060301e-05,
-      "loss": 5.3673,
-      "step": 26015
-    },
-    {
-      "epoch": 13.567666232073012,
-      "grad_norm": 1.583132028579712,
-      "learning_rate": 7.436381909547739e-05,
-      "loss": 5.1888,
-      "step": 26016
-    },
-    {
-      "epoch": 13.568187744458932,
-      "grad_norm": 1.6191048622131348,
-      "learning_rate": 7.436281407035176e-05,
-      "loss": 5.5302,
-      "step": 26017
-    },
-    {
-      "epoch": 13.56870925684485,
-      "grad_norm": 1.548568844795227,
-      "learning_rate": 7.436180904522613e-05,
-      "loss": 5.1323,
-      "step": 26018
-    },
-    {
-      "epoch": 13.569230769230769,
-      "grad_norm": 1.449314832687378,
-      "learning_rate": 7.436080402010051e-05,
-      "loss": 5.4913,
-      "step": 26019
-    },
-    {
-      "epoch": 13.569752281616688,
-      "grad_norm": 1.38150954246521,
-      "learning_rate": 7.435979899497489e-05,
-      "loss": 5.589,
-      "step": 26020
-    },
-    {
-      "epoch": 13.570273794002608,
-      "grad_norm": 1.444655179977417,
-      "learning_rate": 7.435879396984925e-05,
-      "loss": 5.7878,
-      "step": 26021
-    },
-    {
-      "epoch": 13.570795306388527,
-      "grad_norm": 1.4486509561538696,
-      "learning_rate": 7.435778894472362e-05,
-      "loss": 5.496,
-      "step": 26022
-    },
-    {
-      "epoch": 13.571316818774445,
-      "grad_norm": 1.4810986518859863,
-      "learning_rate": 7.4356783919598e-05,
-      "loss": 5.5789,
-      "step": 26023
-    },
-    {
-      "epoch": 13.571838331160365,
-      "grad_norm": 1.5056471824645996,
-      "learning_rate": 7.435577889447236e-05,
-      "loss": 5.5904,
-      "step": 26024
-    },
-    {
-      "epoch": 13.572359843546284,
-      "grad_norm": 1.3563398122787476,
-      "learning_rate": 7.435477386934674e-05,
-      "loss": 5.5999,
-      "step": 26025
-    },
-    {
-      "epoch": 13.572881355932203,
-      "grad_norm": 1.3871859312057495,
-      "learning_rate": 7.43537688442211e-05,
-      "loss": 5.422,
-      "step": 26026
-    },
-    {
-      "epoch": 13.573402868318123,
-      "grad_norm": 1.3316458463668823,
-      "learning_rate": 7.435276381909548e-05,
-      "loss": 5.5855,
-      "step": 26027
-    },
-    {
-      "epoch": 13.573924380704042,
-      "grad_norm": 1.4283597469329834,
-      "learning_rate": 7.435175879396984e-05,
-      "loss": 5.5955,
-      "step": 26028
-    },
-    {
-      "epoch": 13.57444589308996,
-      "grad_norm": 1.5681982040405273,
-      "learning_rate": 7.435075376884422e-05,
-      "loss": 5.4473,
-      "step": 26029
-    },
-    {
-      "epoch": 13.57496740547588,
-      "grad_norm": 1.5039085149765015,
-      "learning_rate": 7.43497487437186e-05,
-      "loss": 5.155,
-      "step": 26030
-    },
-    {
-      "epoch": 13.575488917861799,
-      "grad_norm": 1.488642930984497,
-      "learning_rate": 7.434874371859298e-05,
-      "loss": 5.4681,
-      "step": 26031
-    },
-    {
-      "epoch": 13.576010430247718,
-      "grad_norm": 1.3703835010528564,
-      "learning_rate": 7.434773869346734e-05,
-      "loss": 5.1261,
-      "step": 26032
-    },
-    {
-      "epoch": 13.576531942633638,
-      "grad_norm": 1.4371651411056519,
-      "learning_rate": 7.434673366834172e-05,
-      "loss": 5.2824,
-      "step": 26033
-    },
-    {
-      "epoch": 13.577053455019557,
-      "grad_norm": 1.3895591497421265,
-      "learning_rate": 7.434572864321608e-05,
-      "loss": 5.6247,
-      "step": 26034
-    },
-    {
-      "epoch": 13.577574967405475,
-      "grad_norm": 1.4925317764282227,
-      "learning_rate": 7.434472361809045e-05,
-      "loss": 5.4687,
-      "step": 26035
-    },
-    {
-      "epoch": 13.578096479791395,
-      "grad_norm": 1.4482593536376953,
-      "learning_rate": 7.434371859296483e-05,
-      "loss": 4.7855,
-      "step": 26036
-    },
-    {
-      "epoch": 13.578617992177314,
-      "grad_norm": 1.5875540971755981,
-      "learning_rate": 7.434271356783919e-05,
-      "loss": 5.2388,
-      "step": 26037
-    },
-    {
-      "epoch": 13.579139504563233,
-      "grad_norm": 1.4118850231170654,
-      "learning_rate": 7.434170854271357e-05,
-      "loss": 5.3682,
-      "step": 26038
-    },
-    {
-      "epoch": 13.579661016949153,
-      "grad_norm": 1.4788395166397095,
-      "learning_rate": 7.434070351758795e-05,
-      "loss": 5.654,
-      "step": 26039
-    },
-    {
-      "epoch": 13.580182529335072,
-      "grad_norm": 1.4811419248580933,
-      "learning_rate": 7.433969849246232e-05,
-      "loss": 4.9133,
-      "step": 26040
-    },
-    {
-      "epoch": 13.58070404172099,
-      "grad_norm": 1.489540696144104,
-      "learning_rate": 7.433869346733669e-05,
-      "loss": 5.4461,
-      "step": 26041
-    },
-    {
-      "epoch": 13.58122555410691,
-      "grad_norm": 1.4745806455612183,
-      "learning_rate": 7.433768844221107e-05,
-      "loss": 5.2191,
-      "step": 26042
-    },
-    {
-      "epoch": 13.581747066492829,
-      "grad_norm": 1.5811735391616821,
-      "learning_rate": 7.433668341708543e-05,
-      "loss": 5.3493,
-      "step": 26043
-    },
-    {
-      "epoch": 13.582268578878748,
-      "grad_norm": 1.5544021129608154,
-      "learning_rate": 7.433567839195981e-05,
-      "loss": 4.8793,
-      "step": 26044
-    },
-    {
-      "epoch": 13.582790091264668,
-      "grad_norm": 1.3718843460083008,
-      "learning_rate": 7.433467336683417e-05,
-      "loss": 5.8899,
-      "step": 26045
-    },
-    {
-      "epoch": 13.583311603650587,
-      "grad_norm": 1.551510214805603,
-      "learning_rate": 7.433366834170855e-05,
-      "loss": 5.4853,
-      "step": 26046
-    },
-    {
-      "epoch": 13.583833116036505,
-      "grad_norm": 1.3683111667633057,
-      "learning_rate": 7.433266331658291e-05,
-      "loss": 5.4532,
-      "step": 26047
-    },
-    {
-      "epoch": 13.584354628422425,
-      "grad_norm": 1.3143409490585327,
-      "learning_rate": 7.433165829145728e-05,
-      "loss": 5.7177,
-      "step": 26048
-    },
-    {
-      "epoch": 13.584876140808344,
-      "grad_norm": 1.4222733974456787,
-      "learning_rate": 7.433065326633166e-05,
-      "loss": 5.398,
-      "step": 26049
-    },
-    {
-      "epoch": 13.585397653194264,
-      "grad_norm": 1.468855857849121,
-      "learning_rate": 7.432964824120603e-05,
-      "loss": 5.5878,
-      "step": 26050
-    },
-    {
-      "epoch": 13.585919165580183,
-      "grad_norm": 1.3871091604232788,
-      "learning_rate": 7.432864321608041e-05,
-      "loss": 5.7379,
-      "step": 26051
-    },
-    {
-      "epoch": 13.586440677966102,
-      "grad_norm": 1.3512885570526123,
-      "learning_rate": 7.432763819095478e-05,
-      "loss": 5.5,
-      "step": 26052
-    },
-    {
-      "epoch": 13.58696219035202,
-      "grad_norm": 1.5669827461242676,
-      "learning_rate": 7.432663316582915e-05,
-      "loss": 4.1664,
-      "step": 26053
-    },
-    {
-      "epoch": 13.58748370273794,
-      "grad_norm": 1.47391676902771,
-      "learning_rate": 7.432562814070352e-05,
-      "loss": 5.4363,
-      "step": 26054
-    },
-    {
-      "epoch": 13.588005215123859,
-      "grad_norm": 1.4235639572143555,
-      "learning_rate": 7.43246231155779e-05,
-      "loss": 5.6182,
-      "step": 26055
-    },
-    {
-      "epoch": 13.588526727509779,
-      "grad_norm": 1.494527816772461,
-      "learning_rate": 7.432361809045226e-05,
-      "loss": 5.057,
-      "step": 26056
-    },
-    {
-      "epoch": 13.589048239895698,
-      "grad_norm": 1.597481608390808,
-      "learning_rate": 7.432261306532664e-05,
-      "loss": 4.9702,
-      "step": 26057
-    },
-    {
-      "epoch": 13.589569752281617,
-      "grad_norm": 1.4714155197143555,
-      "learning_rate": 7.4321608040201e-05,
-      "loss": 5.7136,
-      "step": 26058
-    },
-    {
-      "epoch": 13.590091264667535,
-      "grad_norm": 1.390350103378296,
-      "learning_rate": 7.432060301507538e-05,
-      "loss": 5.7085,
-      "step": 26059
-    },
-    {
-      "epoch": 13.590612777053455,
-      "grad_norm": 1.5258535146713257,
-      "learning_rate": 7.431959798994976e-05,
-      "loss": 5.2635,
-      "step": 26060
-    },
-    {
-      "epoch": 13.591134289439374,
-      "grad_norm": 1.3271933794021606,
-      "learning_rate": 7.431859296482414e-05,
-      "loss": 5.3009,
-      "step": 26061
-    },
-    {
-      "epoch": 13.591655801825294,
-      "grad_norm": 1.4481569528579712,
-      "learning_rate": 7.43175879396985e-05,
-      "loss": 5.2012,
-      "step": 26062
-    },
-    {
-      "epoch": 13.592177314211213,
-      "grad_norm": 1.433272361755371,
-      "learning_rate": 7.431658291457286e-05,
-      "loss": 4.9715,
-      "step": 26063
-    },
-    {
-      "epoch": 13.592698826597132,
-      "grad_norm": 1.515386939048767,
-      "learning_rate": 7.431557788944724e-05,
-      "loss": 5.1879,
-      "step": 26064
-    },
-    {
-      "epoch": 13.59322033898305,
-      "grad_norm": 1.432094693183899,
-      "learning_rate": 7.43145728643216e-05,
-      "loss": 4.9586,
-      "step": 26065
-    },
-    {
-      "epoch": 13.59374185136897,
-      "grad_norm": 1.38548743724823,
-      "learning_rate": 7.431356783919598e-05,
-      "loss": 5.4431,
-      "step": 26066
-    },
-    {
-      "epoch": 13.594263363754889,
-      "grad_norm": 1.4325021505355835,
-      "learning_rate": 7.431256281407035e-05,
-      "loss": 5.446,
-      "step": 26067
-    },
-    {
-      "epoch": 13.594784876140809,
-      "grad_norm": 1.3506349325180054,
-      "learning_rate": 7.431155778894473e-05,
-      "loss": 5.3938,
-      "step": 26068
-    },
-    {
-      "epoch": 13.595306388526728,
-      "grad_norm": 1.4106082916259766,
-      "learning_rate": 7.431055276381909e-05,
-      "loss": 5.2892,
-      "step": 26069
-    },
-    {
-      "epoch": 13.595827900912647,
-      "grad_norm": 1.4728944301605225,
-      "learning_rate": 7.430954773869347e-05,
-      "loss": 5.3158,
-      "step": 26070
-    },
-    {
-      "epoch": 13.596349413298565,
-      "grad_norm": 1.49068284034729,
-      "learning_rate": 7.430854271356785e-05,
-      "loss": 5.3951,
-      "step": 26071
-    },
-    {
-      "epoch": 13.596870925684485,
-      "grad_norm": 1.4055867195129395,
-      "learning_rate": 7.430753768844222e-05,
-      "loss": 5.6645,
-      "step": 26072
-    },
-    {
-      "epoch": 13.597392438070404,
-      "grad_norm": 1.4547690153121948,
-      "learning_rate": 7.430653266331659e-05,
-      "loss": 5.3575,
-      "step": 26073
-    },
-    {
-      "epoch": 13.597913950456324,
-      "grad_norm": 1.3729548454284668,
-      "learning_rate": 7.430552763819097e-05,
-      "loss": 5.6431,
-      "step": 26074
-    },
-    {
-      "epoch": 13.598435462842243,
-      "grad_norm": 1.4396542310714722,
-      "learning_rate": 7.430452261306533e-05,
-      "loss": 5.5288,
-      "step": 26075
-    },
-    {
-      "epoch": 13.598956975228162,
-      "grad_norm": 1.5782101154327393,
-      "learning_rate": 7.43035175879397e-05,
-      "loss": 4.9493,
-      "step": 26076
-    },
-    {
-      "epoch": 13.59947848761408,
-      "grad_norm": 1.4793769121170044,
-      "learning_rate": 7.430251256281407e-05,
-      "loss": 5.0505,
-      "step": 26077
-    },
-    {
-      "epoch": 13.6,
-      "grad_norm": 1.505983829498291,
-      "learning_rate": 7.430150753768844e-05,
-      "loss": 5.585,
-      "step": 26078
-    },
-    {
-      "epoch": 13.600521512385919,
-      "grad_norm": 1.422743558883667,
-      "learning_rate": 7.430050251256281e-05,
-      "loss": 5.7707,
-      "step": 26079
-    },
-    {
-      "epoch": 13.601043024771839,
-      "grad_norm": 1.7881399393081665,
-      "learning_rate": 7.429949748743719e-05,
-      "loss": 4.5366,
-      "step": 26080
-    },
-    {
-      "epoch": 13.601564537157758,
-      "grad_norm": 1.4123483896255493,
-      "learning_rate": 7.429849246231157e-05,
-      "loss": 5.6908,
-      "step": 26081
-    },
-    {
-      "epoch": 13.602086049543676,
-      "grad_norm": 1.3842480182647705,
-      "learning_rate": 7.429748743718593e-05,
-      "loss": 5.6,
-      "step": 26082
-    },
-    {
-      "epoch": 13.602607561929595,
-      "grad_norm": 1.3630621433258057,
-      "learning_rate": 7.429648241206031e-05,
-      "loss": 5.6323,
-      "step": 26083
-    },
-    {
-      "epoch": 13.603129074315515,
-      "grad_norm": 1.4178394079208374,
-      "learning_rate": 7.429547738693468e-05,
-      "loss": 5.668,
-      "step": 26084
-    },
-    {
-      "epoch": 13.603650586701434,
-      "grad_norm": 1.4774198532104492,
-      "learning_rate": 7.429447236180905e-05,
-      "loss": 5.4208,
-      "step": 26085
-    },
-    {
-      "epoch": 13.604172099087354,
-      "grad_norm": 1.4778708219528198,
-      "learning_rate": 7.429346733668342e-05,
-      "loss": 5.2447,
-      "step": 26086
-    },
-    {
-      "epoch": 13.604693611473273,
-      "grad_norm": 1.4368757009506226,
-      "learning_rate": 7.42924623115578e-05,
-      "loss": 5.1727,
-      "step": 26087
-    },
-    {
-      "epoch": 13.605215123859193,
-      "grad_norm": 1.3840830326080322,
-      "learning_rate": 7.429145728643216e-05,
-      "loss": 5.3899,
-      "step": 26088
-    },
-    {
-      "epoch": 13.60573663624511,
-      "grad_norm": 1.526163935661316,
-      "learning_rate": 7.429045226130653e-05,
-      "loss": 5.5228,
-      "step": 26089
-    },
-    {
-      "epoch": 13.60625814863103,
-      "grad_norm": 1.5301872491836548,
-      "learning_rate": 7.42894472361809e-05,
-      "loss": 5.3923,
-      "step": 26090
-    },
-    {
-      "epoch": 13.60677966101695,
-      "grad_norm": 1.5186803340911865,
-      "learning_rate": 7.428844221105528e-05,
-      "loss": 4.9935,
-      "step": 26091
-    },
-    {
-      "epoch": 13.607301173402869,
-      "grad_norm": 1.5396679639816284,
-      "learning_rate": 7.428743718592966e-05,
-      "loss": 5.3114,
-      "step": 26092
-    },
-    {
-      "epoch": 13.607822685788788,
-      "grad_norm": 1.5623466968536377,
-      "learning_rate": 7.428643216080402e-05,
-      "loss": 4.9858,
-      "step": 26093
-    },
-    {
-      "epoch": 13.608344198174706,
-      "grad_norm": 1.4633643627166748,
-      "learning_rate": 7.42854271356784e-05,
-      "loss": 5.2921,
-      "step": 26094
-    },
-    {
-      "epoch": 13.608865710560625,
-      "grad_norm": 1.4185411930084229,
-      "learning_rate": 7.428442211055276e-05,
-      "loss": 5.8455,
-      "step": 26095
-    },
-    {
-      "epoch": 13.609387222946545,
-      "grad_norm": 1.4201661348342896,
-      "learning_rate": 7.428341708542714e-05,
-      "loss": 5.4255,
-      "step": 26096
-    },
-    {
-      "epoch": 13.609908735332464,
-      "grad_norm": 1.5785168409347534,
-      "learning_rate": 7.428241206030151e-05,
-      "loss": 4.7963,
-      "step": 26097
-    },
-    {
-      "epoch": 13.610430247718384,
-      "grad_norm": 1.4296561479568481,
-      "learning_rate": 7.428140703517588e-05,
-      "loss": 5.4951,
-      "step": 26098
-    },
-    {
-      "epoch": 13.610951760104303,
-      "grad_norm": 1.6388473510742188,
-      "learning_rate": 7.428040201005025e-05,
-      "loss": 5.2203,
-      "step": 26099
-    },
-    {
-      "epoch": 13.611473272490223,
-      "grad_norm": 1.554927945137024,
-      "learning_rate": 7.427939698492463e-05,
-      "loss": 5.3799,
-      "step": 26100
-    },
-    {
-      "epoch": 13.61199478487614,
-      "grad_norm": 1.4610971212387085,
-      "learning_rate": 7.4278391959799e-05,
-      "loss": 5.483,
-      "step": 26101
-    },
-    {
-      "epoch": 13.61251629726206,
-      "grad_norm": 1.4932643175125122,
-      "learning_rate": 7.427738693467337e-05,
-      "loss": 5.1398,
-      "step": 26102
-    },
-    {
-      "epoch": 13.61303780964798,
-      "grad_norm": 1.421945571899414,
-      "learning_rate": 7.427638190954775e-05,
-      "loss": 5.5296,
-      "step": 26103
-    },
-    {
-      "epoch": 13.613559322033899,
-      "grad_norm": 1.4800184965133667,
-      "learning_rate": 7.427537688442211e-05,
-      "loss": 5.3602,
-      "step": 26104
-    },
-    {
-      "epoch": 13.614080834419818,
-      "grad_norm": 1.375724196434021,
-      "learning_rate": 7.427437185929649e-05,
-      "loss": 5.6454,
-      "step": 26105
-    },
-    {
-      "epoch": 13.614602346805736,
-      "grad_norm": 1.3794810771942139,
-      "learning_rate": 7.427336683417085e-05,
-      "loss": 5.5726,
-      "step": 26106
-    },
-    {
-      "epoch": 13.615123859191655,
-      "grad_norm": 1.4910796880722046,
-      "learning_rate": 7.427236180904523e-05,
-      "loss": 5.7046,
-      "step": 26107
-    },
-    {
-      "epoch": 13.615645371577575,
-      "grad_norm": 1.472817301750183,
-      "learning_rate": 7.42713567839196e-05,
-      "loss": 5.4161,
-      "step": 26108
-    },
-    {
-      "epoch": 13.616166883963494,
-      "grad_norm": 1.4683984518051147,
-      "learning_rate": 7.427035175879397e-05,
-      "loss": 5.4105,
-      "step": 26109
-    },
-    {
-      "epoch": 13.616688396349414,
-      "grad_norm": 1.410421371459961,
-      "learning_rate": 7.426934673366834e-05,
-      "loss": 5.5677,
-      "step": 26110
-    },
-    {
-      "epoch": 13.617209908735333,
-      "grad_norm": 1.9965509176254272,
-      "learning_rate": 7.426834170854272e-05,
-      "loss": 4.6763,
-      "step": 26111
-    },
-    {
-      "epoch": 13.617731421121253,
-      "grad_norm": 1.6112295389175415,
-      "learning_rate": 7.426733668341709e-05,
-      "loss": 5.2869,
-      "step": 26112
-    },
-    {
-      "epoch": 13.61825293350717,
-      "grad_norm": 1.4278899431228638,
-      "learning_rate": 7.426633165829147e-05,
-      "loss": 5.3888,
-      "step": 26113
-    },
-    {
-      "epoch": 13.61877444589309,
-      "grad_norm": 1.5451023578643799,
-      "learning_rate": 7.426532663316584e-05,
-      "loss": 5.2448,
-      "step": 26114
-    },
-    {
-      "epoch": 13.61929595827901,
-      "grad_norm": 1.495601773262024,
-      "learning_rate": 7.42643216080402e-05,
-      "loss": 5.5686,
-      "step": 26115
-    },
-    {
-      "epoch": 13.619817470664929,
-      "grad_norm": 1.3815990686416626,
-      "learning_rate": 7.426331658291458e-05,
-      "loss": 5.3057,
-      "step": 26116
-    },
-    {
-      "epoch": 13.620338983050848,
-      "grad_norm": 1.7193833589553833,
-      "learning_rate": 7.426231155778894e-05,
-      "loss": 5.1115,
-      "step": 26117
-    },
-    {
-      "epoch": 13.620860495436766,
-      "grad_norm": 1.385528802871704,
-      "learning_rate": 7.426130653266332e-05,
-      "loss": 5.2837,
-      "step": 26118
-    },
-    {
-      "epoch": 13.621382007822685,
-      "grad_norm": 1.4630603790283203,
-      "learning_rate": 7.426030150753768e-05,
-      "loss": 5.4018,
-      "step": 26119
-    },
-    {
-      "epoch": 13.621903520208605,
-      "grad_norm": 1.4596033096313477,
-      "learning_rate": 7.425929648241206e-05,
-      "loss": 5.4073,
-      "step": 26120
-    },
-    {
-      "epoch": 13.622425032594524,
-      "grad_norm": 1.503363013267517,
-      "learning_rate": 7.425829145728644e-05,
-      "loss": 5.5269,
-      "step": 26121
-    },
-    {
-      "epoch": 13.622946544980444,
-      "grad_norm": 1.3908815383911133,
-      "learning_rate": 7.425728643216082e-05,
-      "loss": 5.8433,
-      "step": 26122
-    },
-    {
-      "epoch": 13.623468057366363,
-      "grad_norm": 1.53335440158844,
-      "learning_rate": 7.425628140703518e-05,
-      "loss": 5.0917,
-      "step": 26123
-    },
-    {
-      "epoch": 13.62398956975228,
-      "grad_norm": 1.3855700492858887,
-      "learning_rate": 7.425527638190956e-05,
-      "loss": 5.5589,
-      "step": 26124
-    },
-    {
-      "epoch": 13.6245110821382,
-      "grad_norm": 1.5878850221633911,
-      "learning_rate": 7.425427135678392e-05,
-      "loss": 5.6055,
-      "step": 26125
-    },
-    {
-      "epoch": 13.62503259452412,
-      "grad_norm": 1.4044032096862793,
-      "learning_rate": 7.42532663316583e-05,
-      "loss": 5.7794,
-      "step": 26126
-    },
-    {
-      "epoch": 13.62555410691004,
-      "grad_norm": 1.4140361547470093,
-      "learning_rate": 7.425226130653267e-05,
-      "loss": 5.6732,
-      "step": 26127
-    },
-    {
-      "epoch": 13.626075619295959,
-      "grad_norm": 1.4245022535324097,
-      "learning_rate": 7.425125628140703e-05,
-      "loss": 5.5075,
-      "step": 26128
-    },
-    {
-      "epoch": 13.626597131681878,
-      "grad_norm": 1.4786078929901123,
-      "learning_rate": 7.425025125628141e-05,
-      "loss": 5.1137,
-      "step": 26129
-    },
-    {
-      "epoch": 13.627118644067796,
-      "grad_norm": 1.474573016166687,
-      "learning_rate": 7.424924623115577e-05,
-      "loss": 5.0354,
-      "step": 26130
-    },
-    {
-      "epoch": 13.627640156453715,
-      "grad_norm": 1.3722200393676758,
-      "learning_rate": 7.424824120603015e-05,
-      "loss": 5.5187,
-      "step": 26131
-    },
-    {
-      "epoch": 13.628161668839635,
-      "grad_norm": 1.5479954481124878,
-      "learning_rate": 7.424723618090453e-05,
-      "loss": 5.4685,
-      "step": 26132
-    },
-    {
-      "epoch": 13.628683181225554,
-      "grad_norm": 1.3789873123168945,
-      "learning_rate": 7.42462311557789e-05,
-      "loss": 5.6692,
-      "step": 26133
-    },
-    {
-      "epoch": 13.629204693611474,
-      "grad_norm": 1.4237197637557983,
-      "learning_rate": 7.424522613065327e-05,
-      "loss": 5.4826,
-      "step": 26134
-    },
-    {
-      "epoch": 13.629726205997393,
-      "grad_norm": 1.6209639310836792,
-      "learning_rate": 7.424422110552765e-05,
-      "loss": 5.4992,
-      "step": 26135
-    },
-    {
-      "epoch": 13.63024771838331,
-      "grad_norm": 1.5788004398345947,
-      "learning_rate": 7.424321608040201e-05,
-      "loss": 5.1654,
-      "step": 26136
-    },
-    {
-      "epoch": 13.63076923076923,
-      "grad_norm": 1.545098900794983,
-      "learning_rate": 7.424221105527639e-05,
-      "loss": 5.6515,
-      "step": 26137
-    },
-    {
-      "epoch": 13.63129074315515,
-      "grad_norm": 1.399854302406311,
-      "learning_rate": 7.424120603015075e-05,
-      "loss": 5.2923,
-      "step": 26138
-    },
-    {
-      "epoch": 13.63181225554107,
-      "grad_norm": 1.5861592292785645,
-      "learning_rate": 7.424020100502513e-05,
-      "loss": 5.591,
-      "step": 26139
-    },
-    {
-      "epoch": 13.632333767926989,
-      "grad_norm": 1.47199547290802,
-      "learning_rate": 7.42391959798995e-05,
-      "loss": 5.5389,
-      "step": 26140
-    },
-    {
-      "epoch": 13.632855280312908,
-      "grad_norm": 1.5486048460006714,
-      "learning_rate": 7.423819095477387e-05,
-      "loss": 5.1669,
-      "step": 26141
-    },
-    {
-      "epoch": 13.633376792698826,
-      "grad_norm": 1.517118215560913,
-      "learning_rate": 7.423718592964825e-05,
-      "loss": 5.251,
-      "step": 26142
-    },
-    {
-      "epoch": 13.633898305084745,
-      "grad_norm": 1.404724359512329,
-      "learning_rate": 7.423618090452262e-05,
-      "loss": 5.6496,
-      "step": 26143
-    },
-    {
-      "epoch": 13.634419817470665,
-      "grad_norm": 1.6063823699951172,
-      "learning_rate": 7.4235175879397e-05,
-      "loss": 5.179,
-      "step": 26144
-    },
-    {
-      "epoch": 13.634941329856584,
-      "grad_norm": 1.5282258987426758,
-      "learning_rate": 7.423417085427136e-05,
-      "loss": 5.6728,
-      "step": 26145
-    },
-    {
-      "epoch": 13.635462842242504,
-      "grad_norm": 1.493125081062317,
-      "learning_rate": 7.423316582914574e-05,
-      "loss": 4.722,
-      "step": 26146
-    },
-    {
-      "epoch": 13.635984354628423,
-      "grad_norm": 1.6052831411361694,
-      "learning_rate": 7.42321608040201e-05,
-      "loss": 5.0067,
-      "step": 26147
-    },
-    {
-      "epoch": 13.63650586701434,
-      "grad_norm": 1.4930953979492188,
-      "learning_rate": 7.423115577889448e-05,
-      "loss": 5.623,
-      "step": 26148
-    },
-    {
-      "epoch": 13.63702737940026,
-      "grad_norm": 1.4622656106948853,
-      "learning_rate": 7.423015075376884e-05,
-      "loss": 4.8117,
-      "step": 26149
-    },
-    {
-      "epoch": 13.63754889178618,
-      "grad_norm": 1.4713860750198364,
-      "learning_rate": 7.422914572864322e-05,
-      "loss": 5.5922,
-      "step": 26150
-    },
-    {
-      "epoch": 13.6380704041721,
-      "grad_norm": 1.4636738300323486,
-      "learning_rate": 7.422814070351758e-05,
-      "loss": 5.721,
-      "step": 26151
-    },
-    {
-      "epoch": 13.638591916558019,
-      "grad_norm": 1.5155296325683594,
-      "learning_rate": 7.422713567839196e-05,
-      "loss": 5.3482,
-      "step": 26152
-    },
-    {
-      "epoch": 13.639113428943938,
-      "grad_norm": 1.4691797494888306,
-      "learning_rate": 7.422613065326634e-05,
-      "loss": 5.5538,
-      "step": 26153
-    },
-    {
-      "epoch": 13.639634941329856,
-      "grad_norm": 1.4502317905426025,
-      "learning_rate": 7.422512562814072e-05,
-      "loss": 4.8343,
-      "step": 26154
-    },
-    {
-      "epoch": 13.640156453715775,
-      "grad_norm": 1.4991663694381714,
-      "learning_rate": 7.422412060301508e-05,
-      "loss": 5.5547,
-      "step": 26155
-    },
-    {
-      "epoch": 13.640677966101695,
-      "grad_norm": 1.488993525505066,
-      "learning_rate": 7.422311557788945e-05,
-      "loss": 5.2664,
-      "step": 26156
-    },
-    {
-      "epoch": 13.641199478487614,
-      "grad_norm": 1.7084804773330688,
-      "learning_rate": 7.422211055276382e-05,
-      "loss": 5.4609,
-      "step": 26157
-    },
-    {
-      "epoch": 13.641720990873534,
-      "grad_norm": 1.445791244506836,
-      "learning_rate": 7.422110552763819e-05,
-      "loss": 5.6316,
-      "step": 26158
-    },
-    {
-      "epoch": 13.642242503259453,
-      "grad_norm": 1.4453641176223755,
-      "learning_rate": 7.422010050251257e-05,
-      "loss": 5.5443,
-      "step": 26159
-    },
-    {
-      "epoch": 13.642764015645371,
-      "grad_norm": 1.5148558616638184,
-      "learning_rate": 7.421909547738693e-05,
-      "loss": 4.7972,
-      "step": 26160
-    },
-    {
-      "epoch": 13.64328552803129,
-      "grad_norm": 2.355069875717163,
-      "learning_rate": 7.421809045226131e-05,
-      "loss": 5.2027,
-      "step": 26161
-    },
-    {
-      "epoch": 13.64380704041721,
-      "grad_norm": 1.4367586374282837,
-      "learning_rate": 7.421708542713569e-05,
-      "loss": 5.6728,
-      "step": 26162
-    },
-    {
-      "epoch": 13.64432855280313,
-      "grad_norm": 1.488897681236267,
-      "learning_rate": 7.421608040201006e-05,
-      "loss": 5.3677,
-      "step": 26163
-    },
-    {
-      "epoch": 13.644850065189049,
-      "grad_norm": 1.4052753448486328,
-      "learning_rate": 7.421507537688443e-05,
-      "loss": 5.0378,
-      "step": 26164
-    },
-    {
-      "epoch": 13.645371577574968,
-      "grad_norm": 1.5003809928894043,
-      "learning_rate": 7.42140703517588e-05,
-      "loss": 5.4462,
-      "step": 26165
-    },
-    {
-      "epoch": 13.645893089960886,
-      "grad_norm": 1.6001914739608765,
-      "learning_rate": 7.421306532663317e-05,
-      "loss": 5.2903,
-      "step": 26166
-    },
-    {
-      "epoch": 13.646414602346805,
-      "grad_norm": 1.5029942989349365,
-      "learning_rate": 7.421206030150755e-05,
-      "loss": 5.6414,
-      "step": 26167
-    },
-    {
-      "epoch": 13.646936114732725,
-      "grad_norm": 1.3701965808868408,
-      "learning_rate": 7.421105527638191e-05,
-      "loss": 5.6194,
-      "step": 26168
-    },
-    {
-      "epoch": 13.647457627118644,
-      "grad_norm": 1.4534550905227661,
-      "learning_rate": 7.421005025125628e-05,
-      "loss": 5.0325,
-      "step": 26169
-    },
-    {
-      "epoch": 13.647979139504564,
-      "grad_norm": 1.2662545442581177,
-      "learning_rate": 7.420904522613065e-05,
-      "loss": 5.9199,
-      "step": 26170
-    },
-    {
-      "epoch": 13.648500651890483,
-      "grad_norm": 1.5127530097961426,
-      "learning_rate": 7.420804020100502e-05,
-      "loss": 5.5336,
-      "step": 26171
-    },
-    {
-      "epoch": 13.649022164276401,
-      "grad_norm": 1.4174138307571411,
-      "learning_rate": 7.42070351758794e-05,
-      "loss": 4.9872,
-      "step": 26172
-    },
-    {
-      "epoch": 13.64954367666232,
-      "grad_norm": 1.5074068307876587,
-      "learning_rate": 7.420603015075377e-05,
-      "loss": 5.0552,
-      "step": 26173
-    },
-    {
-      "epoch": 13.65006518904824,
-      "grad_norm": 1.5188313722610474,
-      "learning_rate": 7.420502512562815e-05,
-      "loss": 5.3935,
-      "step": 26174
-    },
-    {
-      "epoch": 13.65058670143416,
-      "grad_norm": 1.4314706325531006,
-      "learning_rate": 7.420402010050252e-05,
-      "loss": 5.5797,
-      "step": 26175
-    },
-    {
-      "epoch": 13.651108213820079,
-      "grad_norm": 1.4563297033309937,
-      "learning_rate": 7.42030150753769e-05,
-      "loss": 5.6362,
-      "step": 26176
-    },
-    {
-      "epoch": 13.651629726205996,
-      "grad_norm": 1.3818061351776123,
-      "learning_rate": 7.420201005025126e-05,
-      "loss": 5.0999,
-      "step": 26177
-    },
-    {
-      "epoch": 13.652151238591916,
-      "grad_norm": 1.4794371128082275,
-      "learning_rate": 7.420100502512564e-05,
-      "loss": 5.4999,
-      "step": 26178
-    },
-    {
-      "epoch": 13.652672750977835,
-      "grad_norm": 1.4799684286117554,
-      "learning_rate": 7.42e-05,
-      "loss": 5.7729,
-      "step": 26179
-    },
-    {
-      "epoch": 13.653194263363755,
-      "grad_norm": 1.5239866971969604,
-      "learning_rate": 7.419899497487438e-05,
-      "loss": 5.3849,
-      "step": 26180
-    },
-    {
-      "epoch": 13.653715775749674,
-      "grad_norm": 1.4996247291564941,
-      "learning_rate": 7.419798994974874e-05,
-      "loss": 5.2522,
-      "step": 26181
-    },
-    {
-      "epoch": 13.654237288135594,
-      "grad_norm": 1.409036636352539,
-      "learning_rate": 7.419698492462311e-05,
-      "loss": 4.8579,
-      "step": 26182
-    },
-    {
-      "epoch": 13.654758800521513,
-      "grad_norm": 1.4676793813705444,
-      "learning_rate": 7.419597989949749e-05,
-      "loss": 5.4014,
-      "step": 26183
-    },
-    {
-      "epoch": 13.655280312907431,
-      "grad_norm": 1.446311354637146,
-      "learning_rate": 7.419497487437186e-05,
-      "loss": 5.2416,
-      "step": 26184
-    },
-    {
-      "epoch": 13.65580182529335,
-      "grad_norm": 1.4883307218551636,
-      "learning_rate": 7.419396984924624e-05,
-      "loss": 5.8059,
-      "step": 26185
-    },
-    {
-      "epoch": 13.65632333767927,
-      "grad_norm": 1.6310043334960938,
-      "learning_rate": 7.41929648241206e-05,
-      "loss": 5.1787,
-      "step": 26186
-    },
-    {
-      "epoch": 13.65684485006519,
-      "grad_norm": 1.4577267169952393,
-      "learning_rate": 7.419195979899498e-05,
-      "loss": 5.2385,
-      "step": 26187
-    },
-    {
-      "epoch": 13.657366362451109,
-      "grad_norm": 1.6027939319610596,
-      "learning_rate": 7.419095477386935e-05,
-      "loss": 4.9974,
-      "step": 26188
-    },
-    {
-      "epoch": 13.657887874837026,
-      "grad_norm": 1.5702874660491943,
-      "learning_rate": 7.418994974874373e-05,
-      "loss": 5.5085,
-      "step": 26189
-    },
-    {
-      "epoch": 13.658409387222946,
-      "grad_norm": 1.4723410606384277,
-      "learning_rate": 7.418894472361809e-05,
-      "loss": 5.5255,
-      "step": 26190
-    },
-    {
-      "epoch": 13.658930899608865,
-      "grad_norm": 1.5120006799697876,
-      "learning_rate": 7.418793969849247e-05,
-      "loss": 5.8529,
-      "step": 26191
-    },
-    {
-      "epoch": 13.659452411994785,
-      "grad_norm": 1.392195224761963,
-      "learning_rate": 7.418693467336683e-05,
-      "loss": 5.5675,
-      "step": 26192
-    },
-    {
-      "epoch": 13.659973924380704,
-      "grad_norm": 1.4808648824691772,
-      "learning_rate": 7.418592964824121e-05,
-      "loss": 5.7374,
-      "step": 26193
-    },
-    {
-      "epoch": 13.660495436766624,
-      "grad_norm": 1.4854035377502441,
-      "learning_rate": 7.418492462311559e-05,
-      "loss": 5.336,
-      "step": 26194
-    },
-    {
-      "epoch": 13.661016949152543,
-      "grad_norm": 1.506764531135559,
-      "learning_rate": 7.418391959798995e-05,
-      "loss": 5.5313,
-      "step": 26195
-    },
-    {
-      "epoch": 13.661538461538461,
-      "grad_norm": 1.4673322439193726,
-      "learning_rate": 7.418291457286433e-05,
-      "loss": 5.5385,
-      "step": 26196
-    },
-    {
-      "epoch": 13.66205997392438,
-      "grad_norm": 1.5096347332000732,
-      "learning_rate": 7.41819095477387e-05,
-      "loss": 5.4562,
-      "step": 26197
-    },
-    {
-      "epoch": 13.6625814863103,
-      "grad_norm": 2.107394218444824,
-      "learning_rate": 7.418090452261307e-05,
-      "loss": 5.4657,
-      "step": 26198
-    },
-    {
-      "epoch": 13.66310299869622,
-      "grad_norm": 1.4587910175323486,
-      "learning_rate": 7.417989949748744e-05,
-      "loss": 5.5091,
-      "step": 26199
-    },
-    {
-      "epoch": 13.663624511082139,
-      "grad_norm": 1.4861432313919067,
-      "learning_rate": 7.417889447236181e-05,
-      "loss": 5.5744,
-      "step": 26200
-    },
-    {
-      "epoch": 13.664146023468057,
-      "grad_norm": 1.5024362802505493,
-      "learning_rate": 7.417788944723618e-05,
-      "loss": 5.3625,
-      "step": 26201
-    },
-    {
-      "epoch": 13.664667535853976,
-      "grad_norm": 1.5409218072891235,
-      "learning_rate": 7.417688442211056e-05,
-      "loss": 5.3109,
-      "step": 26202
-    },
-    {
-      "epoch": 13.665189048239895,
-      "grad_norm": 1.5802841186523438,
-      "learning_rate": 7.417587939698492e-05,
-      "loss": 5.2909,
-      "step": 26203
-    },
-    {
-      "epoch": 13.665710560625815,
-      "grad_norm": 1.5415915250778198,
-      "learning_rate": 7.41748743718593e-05,
-      "loss": 5.3321,
-      "step": 26204
-    },
-    {
-      "epoch": 13.666232073011734,
-      "grad_norm": 1.3742016553878784,
-      "learning_rate": 7.417386934673368e-05,
-      "loss": 5.4723,
-      "step": 26205
-    },
-    {
-      "epoch": 13.666753585397654,
-      "grad_norm": 1.4965910911560059,
-      "learning_rate": 7.417286432160805e-05,
-      "loss": 5.5317,
-      "step": 26206
-    },
-    {
-      "epoch": 13.667275097783573,
-      "grad_norm": 1.4204915761947632,
-      "learning_rate": 7.417185929648242e-05,
-      "loss": 5.0331,
-      "step": 26207
-    },
-    {
-      "epoch": 13.667796610169491,
-      "grad_norm": 1.4878114461898804,
-      "learning_rate": 7.417085427135678e-05,
-      "loss": 5.1348,
-      "step": 26208
-    },
-    {
-      "epoch": 13.66831812255541,
-      "grad_norm": 1.4559324979782104,
-      "learning_rate": 7.416984924623116e-05,
-      "loss": 4.9563,
-      "step": 26209
-    },
-    {
-      "epoch": 13.66883963494133,
-      "grad_norm": 1.5609060525894165,
-      "learning_rate": 7.416884422110552e-05,
-      "loss": 4.9849,
-      "step": 26210
-    },
-    {
-      "epoch": 13.66936114732725,
-      "grad_norm": 1.381268858909607,
-      "learning_rate": 7.41678391959799e-05,
-      "loss": 5.6484,
-      "step": 26211
-    },
-    {
-      "epoch": 13.669882659713169,
-      "grad_norm": 1.3492250442504883,
-      "learning_rate": 7.416683417085427e-05,
-      "loss": 5.5157,
-      "step": 26212
-    },
-    {
-      "epoch": 13.670404172099087,
-      "grad_norm": 1.4626526832580566,
-      "learning_rate": 7.416582914572864e-05,
-      "loss": 5.5079,
-      "step": 26213
-    },
-    {
-      "epoch": 13.670925684485006,
-      "grad_norm": 1.409006118774414,
-      "learning_rate": 7.416482412060302e-05,
-      "loss": 5.2888,
-      "step": 26214
-    },
-    {
-      "epoch": 13.671447196870925,
-      "grad_norm": 1.436029076576233,
-      "learning_rate": 7.41638190954774e-05,
-      "loss": 4.97,
-      "step": 26215
-    },
-    {
-      "epoch": 13.671968709256845,
-      "grad_norm": 1.354110598564148,
-      "learning_rate": 7.416281407035176e-05,
-      "loss": 5.7618,
-      "step": 26216
-    },
-    {
-      "epoch": 13.672490221642764,
-      "grad_norm": 1.5244249105453491,
-      "learning_rate": 7.416180904522614e-05,
-      "loss": 5.1877,
-      "step": 26217
-    },
-    {
-      "epoch": 13.673011734028684,
-      "grad_norm": 1.4458073377609253,
-      "learning_rate": 7.41608040201005e-05,
-      "loss": 5.3287,
-      "step": 26218
-    },
-    {
-      "epoch": 13.673533246414602,
-      "grad_norm": 1.4134246110916138,
-      "learning_rate": 7.415979899497488e-05,
-      "loss": 5.6795,
-      "step": 26219
-    },
-    {
-      "epoch": 13.674054758800521,
-      "grad_norm": 1.4877550601959229,
-      "learning_rate": 7.415879396984925e-05,
-      "loss": 5.4524,
-      "step": 26220
-    },
-    {
-      "epoch": 13.67457627118644,
-      "grad_norm": 1.5454896688461304,
-      "learning_rate": 7.415778894472361e-05,
-      "loss": 5.106,
-      "step": 26221
-    },
-    {
-      "epoch": 13.67509778357236,
-      "grad_norm": 1.449265956878662,
-      "learning_rate": 7.415678391959799e-05,
-      "loss": 5.4544,
-      "step": 26222
-    },
-    {
-      "epoch": 13.67561929595828,
-      "grad_norm": 1.4541906118392944,
-      "learning_rate": 7.415577889447235e-05,
-      "loss": 5.3673,
-      "step": 26223
-    },
-    {
-      "epoch": 13.676140808344199,
-      "grad_norm": 1.4044163227081299,
-      "learning_rate": 7.415477386934673e-05,
-      "loss": 5.5651,
-      "step": 26224
-    },
-    {
-      "epoch": 13.676662320730117,
-      "grad_norm": 1.3367114067077637,
-      "learning_rate": 7.415376884422111e-05,
-      "loss": 5.7251,
-      "step": 26225
-    },
-    {
-      "epoch": 13.677183833116036,
-      "grad_norm": 1.6734681129455566,
-      "learning_rate": 7.415276381909549e-05,
-      "loss": 4.5713,
-      "step": 26226
-    },
-    {
-      "epoch": 13.677705345501955,
-      "grad_norm": 1.4872862100601196,
-      "learning_rate": 7.415175879396985e-05,
-      "loss": 5.311,
-      "step": 26227
-    },
-    {
-      "epoch": 13.678226857887875,
-      "grad_norm": 1.4825828075408936,
-      "learning_rate": 7.415075376884423e-05,
-      "loss": 5.3993,
-      "step": 26228
-    },
-    {
-      "epoch": 13.678748370273794,
-      "grad_norm": 1.575156807899475,
-      "learning_rate": 7.41497487437186e-05,
-      "loss": 4.687,
-      "step": 26229
-    },
-    {
-      "epoch": 13.679269882659714,
-      "grad_norm": 1.4130020141601562,
-      "learning_rate": 7.414874371859297e-05,
-      "loss": 5.4552,
-      "step": 26230
-    },
-    {
-      "epoch": 13.679791395045632,
-      "grad_norm": 1.438066840171814,
-      "learning_rate": 7.414773869346734e-05,
-      "loss": 5.4389,
-      "step": 26231
-    },
-    {
-      "epoch": 13.680312907431551,
-      "grad_norm": 1.4070160388946533,
-      "learning_rate": 7.414673366834171e-05,
-      "loss": 5.9027,
-      "step": 26232
-    },
-    {
-      "epoch": 13.68083441981747,
-      "grad_norm": 1.4406101703643799,
-      "learning_rate": 7.414572864321608e-05,
-      "loss": 5.5522,
-      "step": 26233
-    },
-    {
-      "epoch": 13.68135593220339,
-      "grad_norm": 1.650753140449524,
-      "learning_rate": 7.414472361809046e-05,
-      "loss": 5.3567,
-      "step": 26234
-    },
-    {
-      "epoch": 13.68187744458931,
-      "grad_norm": 1.3738634586334229,
-      "learning_rate": 7.414371859296483e-05,
-      "loss": 5.5649,
-      "step": 26235
-    },
-    {
-      "epoch": 13.682398956975229,
-      "grad_norm": 1.468075156211853,
-      "learning_rate": 7.41427135678392e-05,
-      "loss": 5.1266,
-      "step": 26236
-    },
-    {
-      "epoch": 13.682920469361147,
-      "grad_norm": 1.5536202192306519,
-      "learning_rate": 7.414170854271358e-05,
-      "loss": 5.3822,
-      "step": 26237
-    },
-    {
-      "epoch": 13.683441981747066,
-      "grad_norm": 1.4811030626296997,
-      "learning_rate": 7.414070351758794e-05,
-      "loss": 5.5061,
-      "step": 26238
-    },
-    {
-      "epoch": 13.683963494132986,
-      "grad_norm": 1.5883451700210571,
-      "learning_rate": 7.413969849246232e-05,
-      "loss": 5.4024,
-      "step": 26239
-    },
-    {
-      "epoch": 13.684485006518905,
-      "grad_norm": 1.5401113033294678,
-      "learning_rate": 7.413869346733668e-05,
-      "loss": 5.5374,
-      "step": 26240
-    },
-    {
-      "epoch": 13.685006518904824,
-      "grad_norm": 1.4195226430892944,
-      "learning_rate": 7.413768844221106e-05,
-      "loss": 4.774,
-      "step": 26241
-    },
-    {
-      "epoch": 13.685528031290744,
-      "grad_norm": 1.4024386405944824,
-      "learning_rate": 7.413668341708542e-05,
-      "loss": 5.3215,
-      "step": 26242
-    },
-    {
-      "epoch": 13.686049543676662,
-      "grad_norm": 1.4251238107681274,
-      "learning_rate": 7.41356783919598e-05,
-      "loss": 5.3823,
-      "step": 26243
-    },
-    {
-      "epoch": 13.686571056062581,
-      "grad_norm": 1.4096269607543945,
-      "learning_rate": 7.413467336683417e-05,
-      "loss": 5.3796,
-      "step": 26244
-    },
-    {
-      "epoch": 13.6870925684485,
-      "grad_norm": 1.420304536819458,
-      "learning_rate": 7.413366834170854e-05,
-      "loss": 5.4005,
-      "step": 26245
-    },
-    {
-      "epoch": 13.68761408083442,
-      "grad_norm": 1.496395230293274,
-      "learning_rate": 7.413266331658292e-05,
-      "loss": 4.9784,
-      "step": 26246
-    },
-    {
-      "epoch": 13.68813559322034,
-      "grad_norm": 1.479042887687683,
-      "learning_rate": 7.41316582914573e-05,
-      "loss": 5.2394,
-      "step": 26247
-    },
-    {
-      "epoch": 13.688657105606259,
-      "grad_norm": 1.4018672704696655,
-      "learning_rate": 7.413065326633166e-05,
-      "loss": 5.3966,
-      "step": 26248
-    },
-    {
-      "epoch": 13.689178617992177,
-      "grad_norm": 1.4291772842407227,
-      "learning_rate": 7.412964824120603e-05,
-      "loss": 5.3348,
-      "step": 26249
-    },
-    {
-      "epoch": 13.689700130378096,
-      "grad_norm": 1.5117086172103882,
-      "learning_rate": 7.41286432160804e-05,
-      "loss": 5.3186,
-      "step": 26250
-    },
-    {
-      "epoch": 13.690221642764016,
-      "grad_norm": 1.394789695739746,
-      "learning_rate": 7.412763819095477e-05,
-      "loss": 5.7351,
-      "step": 26251
-    },
-    {
-      "epoch": 13.690743155149935,
-      "grad_norm": 1.51079523563385,
-      "learning_rate": 7.412663316582915e-05,
-      "loss": 5.4097,
-      "step": 26252
-    },
-    {
-      "epoch": 13.691264667535854,
-      "grad_norm": 1.3716630935668945,
-      "learning_rate": 7.412562814070351e-05,
-      "loss": 5.4202,
-      "step": 26253
-    },
-    {
-      "epoch": 13.691786179921774,
-      "grad_norm": 1.5489226579666138,
-      "learning_rate": 7.412462311557789e-05,
-      "loss": 4.93,
-      "step": 26254
-    },
-    {
-      "epoch": 13.692307692307692,
-      "grad_norm": 1.4554541110992432,
-      "learning_rate": 7.412361809045227e-05,
-      "loss": 5.4049,
-      "step": 26255
-    },
-    {
-      "epoch": 13.692829204693611,
-      "grad_norm": 1.4946415424346924,
-      "learning_rate": 7.412261306532665e-05,
-      "loss": 5.0249,
-      "step": 26256
-    },
-    {
-      "epoch": 13.69335071707953,
-      "grad_norm": 1.3879883289337158,
-      "learning_rate": 7.412160804020101e-05,
-      "loss": 5.4705,
-      "step": 26257
-    },
-    {
-      "epoch": 13.69387222946545,
-      "grad_norm": 1.5199731588363647,
-      "learning_rate": 7.412060301507539e-05,
-      "loss": 5.1561,
-      "step": 26258
-    },
-    {
-      "epoch": 13.69439374185137,
-      "grad_norm": 1.4834749698638916,
-      "learning_rate": 7.411959798994975e-05,
-      "loss": 5.2967,
-      "step": 26259
-    },
-    {
-      "epoch": 13.694915254237289,
-      "grad_norm": 1.4142404794692993,
-      "learning_rate": 7.411859296482413e-05,
-      "loss": 5.4138,
-      "step": 26260
-    },
-    {
-      "epoch": 13.695436766623207,
-      "grad_norm": 1.4084123373031616,
-      "learning_rate": 7.41175879396985e-05,
-      "loss": 5.7512,
-      "step": 26261
-    },
-    {
-      "epoch": 13.695958279009126,
-      "grad_norm": 1.4060670137405396,
-      "learning_rate": 7.411658291457286e-05,
-      "loss": 5.306,
-      "step": 26262
-    },
-    {
-      "epoch": 13.696479791395046,
-      "grad_norm": 1.3916294574737549,
-      "learning_rate": 7.411557788944724e-05,
-      "loss": 5.449,
-      "step": 26263
-    },
-    {
-      "epoch": 13.697001303780965,
-      "grad_norm": 1.3441957235336304,
-      "learning_rate": 7.41145728643216e-05,
-      "loss": 5.5937,
-      "step": 26264
-    },
-    {
-      "epoch": 13.697522816166884,
-      "grad_norm": 1.3253542184829712,
-      "learning_rate": 7.411356783919598e-05,
-      "loss": 5.8398,
-      "step": 26265
-    },
-    {
-      "epoch": 13.698044328552804,
-      "grad_norm": 1.4485461711883545,
-      "learning_rate": 7.411256281407036e-05,
-      "loss": 5.5493,
-      "step": 26266
-    },
-    {
-      "epoch": 13.698565840938722,
-      "grad_norm": 1.3540000915527344,
-      "learning_rate": 7.411155778894473e-05,
-      "loss": 5.8116,
-      "step": 26267
-    },
-    {
-      "epoch": 13.699087353324641,
-      "grad_norm": 1.4409502744674683,
-      "learning_rate": 7.41105527638191e-05,
-      "loss": 5.6855,
-      "step": 26268
-    },
-    {
-      "epoch": 13.69960886571056,
-      "grad_norm": 1.4635642766952515,
-      "learning_rate": 7.410954773869348e-05,
-      "loss": 5.6311,
-      "step": 26269
-    },
-    {
-      "epoch": 13.70013037809648,
-      "grad_norm": 1.4778673648834229,
-      "learning_rate": 7.410854271356784e-05,
-      "loss": 5.1935,
-      "step": 26270
-    },
-    {
-      "epoch": 13.7006518904824,
-      "grad_norm": 1.6008974313735962,
-      "learning_rate": 7.410753768844222e-05,
-      "loss": 5.1809,
-      "step": 26271
-    },
-    {
-      "epoch": 13.701173402868317,
-      "grad_norm": 1.5531880855560303,
-      "learning_rate": 7.410653266331658e-05,
-      "loss": 5.0368,
-      "step": 26272
-    },
-    {
-      "epoch": 13.701694915254237,
-      "grad_norm": 1.4938340187072754,
-      "learning_rate": 7.410552763819096e-05,
-      "loss": 5.2948,
-      "step": 26273
-    },
-    {
-      "epoch": 13.702216427640156,
-      "grad_norm": 1.3586103916168213,
-      "learning_rate": 7.410452261306533e-05,
-      "loss": 5.2642,
-      "step": 26274
-    },
-    {
-      "epoch": 13.702737940026076,
-      "grad_norm": 1.7340881824493408,
-      "learning_rate": 7.41035175879397e-05,
-      "loss": 4.6779,
-      "step": 26275
-    },
-    {
-      "epoch": 13.703259452411995,
-      "grad_norm": 1.3740848302841187,
-      "learning_rate": 7.410251256281408e-05,
-      "loss": 5.7208,
-      "step": 26276
-    },
-    {
-      "epoch": 13.703780964797915,
-      "grad_norm": 1.4257525205612183,
-      "learning_rate": 7.410150753768845e-05,
-      "loss": 5.655,
-      "step": 26277
-    },
-    {
-      "epoch": 13.704302477183834,
-      "grad_norm": 1.3289958238601685,
-      "learning_rate": 7.410050251256282e-05,
-      "loss": 5.0164,
-      "step": 26278
-    },
-    {
-      "epoch": 13.704823989569752,
-      "grad_norm": 1.3960164785385132,
-      "learning_rate": 7.409949748743719e-05,
-      "loss": 5.2162,
-      "step": 26279
-    },
-    {
-      "epoch": 13.705345501955671,
-      "grad_norm": 1.5484191179275513,
-      "learning_rate": 7.409849246231157e-05,
-      "loss": 5.1099,
-      "step": 26280
-    },
-    {
-      "epoch": 13.70586701434159,
-      "grad_norm": 1.546349048614502,
-      "learning_rate": 7.409748743718593e-05,
-      "loss": 5.1393,
-      "step": 26281
-    },
-    {
-      "epoch": 13.70638852672751,
-      "grad_norm": 1.3247578144073486,
-      "learning_rate": 7.409648241206031e-05,
-      "loss": 5.8952,
-      "step": 26282
-    },
-    {
-      "epoch": 13.70691003911343,
-      "grad_norm": 1.3990881443023682,
-      "learning_rate": 7.409547738693467e-05,
-      "loss": 5.4744,
-      "step": 26283
-    },
-    {
-      "epoch": 13.707431551499347,
-      "grad_norm": 1.5416021347045898,
-      "learning_rate": 7.409447236180905e-05,
-      "loss": 5.4675,
-      "step": 26284
-    },
-    {
-      "epoch": 13.707953063885267,
-      "grad_norm": 1.3134530782699585,
-      "learning_rate": 7.409346733668341e-05,
-      "loss": 5.6475,
-      "step": 26285
-    },
-    {
-      "epoch": 13.708474576271186,
-      "grad_norm": 1.3997832536697388,
-      "learning_rate": 7.409246231155779e-05,
-      "loss": 5.3321,
-      "step": 26286
-    },
-    {
-      "epoch": 13.708996088657106,
-      "grad_norm": 1.5522408485412598,
-      "learning_rate": 7.409145728643217e-05,
-      "loss": 5.0478,
-      "step": 26287
-    },
-    {
-      "epoch": 13.709517601043025,
-      "grad_norm": 1.4679230451583862,
-      "learning_rate": 7.409045226130653e-05,
-      "loss": 5.3113,
-      "step": 26288
-    },
-    {
-      "epoch": 13.710039113428945,
-      "grad_norm": 1.5354830026626587,
-      "learning_rate": 7.408944723618091e-05,
-      "loss": 5.533,
-      "step": 26289
-    },
-    {
-      "epoch": 13.710560625814864,
-      "grad_norm": 1.5053530931472778,
-      "learning_rate": 7.408844221105528e-05,
-      "loss": 4.8804,
-      "step": 26290
-    },
-    {
-      "epoch": 13.711082138200782,
-      "grad_norm": 1.4091129302978516,
-      "learning_rate": 7.408743718592965e-05,
-      "loss": 5.7551,
-      "step": 26291
-    },
-    {
-      "epoch": 13.711603650586701,
-      "grad_norm": 1.3390133380889893,
-      "learning_rate": 7.408643216080402e-05,
-      "loss": 4.7371,
-      "step": 26292
-    },
-    {
-      "epoch": 13.71212516297262,
-      "grad_norm": 1.6122639179229736,
-      "learning_rate": 7.40854271356784e-05,
-      "loss": 5.1205,
-      "step": 26293
-    },
-    {
-      "epoch": 13.71264667535854,
-      "grad_norm": 1.4704855680465698,
-      "learning_rate": 7.408442211055276e-05,
-      "loss": 5.6712,
-      "step": 26294
-    },
-    {
-      "epoch": 13.71316818774446,
-      "grad_norm": 1.3971049785614014,
-      "learning_rate": 7.408341708542714e-05,
-      "loss": 5.6084,
-      "step": 26295
-    },
-    {
-      "epoch": 13.713689700130377,
-      "grad_norm": 1.5945837497711182,
-      "learning_rate": 7.408241206030152e-05,
-      "loss": 5.0501,
-      "step": 26296
-    },
-    {
-      "epoch": 13.714211212516297,
-      "grad_norm": 1.5383433103561401,
-      "learning_rate": 7.40814070351759e-05,
-      "loss": 5.4105,
-      "step": 26297
-    },
-    {
-      "epoch": 13.714732724902216,
-      "grad_norm": 1.547374963760376,
-      "learning_rate": 7.408040201005026e-05,
-      "loss": 5.2056,
-      "step": 26298
-    },
-    {
-      "epoch": 13.715254237288136,
-      "grad_norm": 1.5434465408325195,
-      "learning_rate": 7.407939698492464e-05,
-      "loss": 5.1946,
-      "step": 26299
-    },
-    {
-      "epoch": 13.715775749674055,
-      "grad_norm": 1.4849501848220825,
-      "learning_rate": 7.4078391959799e-05,
-      "loss": 5.2289,
-      "step": 26300
-    },
-    {
-      "epoch": 13.716297262059975,
-      "grad_norm": 1.3837039470672607,
-      "learning_rate": 7.407738693467336e-05,
-      "loss": 5.8061,
-      "step": 26301
-    },
-    {
-      "epoch": 13.716818774445892,
-      "grad_norm": 1.6627229452133179,
-      "learning_rate": 7.407638190954774e-05,
-      "loss": 5.1194,
-      "step": 26302
-    },
-    {
-      "epoch": 13.717340286831812,
-      "grad_norm": 1.4262263774871826,
-      "learning_rate": 7.40753768844221e-05,
-      "loss": 5.9162,
-      "step": 26303
-    },
-    {
-      "epoch": 13.717861799217731,
-      "grad_norm": 1.4253407716751099,
-      "learning_rate": 7.407437185929648e-05,
-      "loss": 5.621,
-      "step": 26304
-    },
-    {
-      "epoch": 13.71838331160365,
-      "grad_norm": 1.5102624893188477,
-      "learning_rate": 7.407336683417085e-05,
-      "loss": 5.3633,
-      "step": 26305
-    },
-    {
-      "epoch": 13.71890482398957,
-      "grad_norm": 1.5098985433578491,
-      "learning_rate": 7.407236180904523e-05,
-      "loss": 5.3244,
-      "step": 26306
-    },
-    {
-      "epoch": 13.71942633637549,
-      "grad_norm": 1.4127471446990967,
-      "learning_rate": 7.40713567839196e-05,
-      "loss": 5.6632,
-      "step": 26307
-    },
-    {
-      "epoch": 13.719947848761407,
-      "grad_norm": 1.5025441646575928,
-      "learning_rate": 7.407035175879398e-05,
-      "loss": 5.4477,
-      "step": 26308
-    },
-    {
-      "epoch": 13.720469361147327,
-      "grad_norm": 1.4731711149215698,
-      "learning_rate": 7.406934673366835e-05,
-      "loss": 5.2424,
-      "step": 26309
-    },
-    {
-      "epoch": 13.720990873533246,
-      "grad_norm": 1.407908320426941,
-      "learning_rate": 7.406834170854272e-05,
-      "loss": 5.6632,
-      "step": 26310
-    },
-    {
-      "epoch": 13.721512385919166,
-      "grad_norm": 1.570426344871521,
-      "learning_rate": 7.406733668341709e-05,
-      "loss": 5.2882,
-      "step": 26311
-    },
-    {
-      "epoch": 13.722033898305085,
-      "grad_norm": 1.528321385383606,
-      "learning_rate": 7.406633165829147e-05,
-      "loss": 4.7139,
-      "step": 26312
-    },
-    {
-      "epoch": 13.722555410691005,
-      "grad_norm": 1.3895516395568848,
-      "learning_rate": 7.406532663316583e-05,
-      "loss": 5.5204,
-      "step": 26313
-    },
-    {
-      "epoch": 13.723076923076922,
-      "grad_norm": 1.5471476316452026,
-      "learning_rate": 7.40643216080402e-05,
-      "loss": 5.0588,
-      "step": 26314
-    },
-    {
-      "epoch": 13.723598435462842,
-      "grad_norm": 1.5119361877441406,
-      "learning_rate": 7.406331658291457e-05,
-      "loss": 5.358,
-      "step": 26315
-    },
-    {
-      "epoch": 13.724119947848761,
-      "grad_norm": 1.4902794361114502,
-      "learning_rate": 7.406231155778895e-05,
-      "loss": 4.9237,
-      "step": 26316
-    },
-    {
-      "epoch": 13.72464146023468,
-      "grad_norm": 1.342659831047058,
-      "learning_rate": 7.406130653266333e-05,
-      "loss": 5.852,
-      "step": 26317
-    },
-    {
-      "epoch": 13.7251629726206,
-      "grad_norm": 1.7578521966934204,
-      "learning_rate": 7.406030150753769e-05,
-      "loss": 5.4874,
-      "step": 26318
-    },
-    {
-      "epoch": 13.72568448500652,
-      "grad_norm": 1.5006173849105835,
-      "learning_rate": 7.405929648241207e-05,
-      "loss": 5.4944,
-      "step": 26319
-    },
-    {
-      "epoch": 13.726205997392437,
-      "grad_norm": 1.4205178022384644,
-      "learning_rate": 7.405829145728643e-05,
-      "loss": 5.5384,
-      "step": 26320
-    },
-    {
-      "epoch": 13.726727509778357,
-      "grad_norm": 1.4477325677871704,
-      "learning_rate": 7.405728643216081e-05,
-      "loss": 5.6302,
-      "step": 26321
-    },
-    {
-      "epoch": 13.727249022164276,
-      "grad_norm": 1.4187220335006714,
-      "learning_rate": 7.405628140703518e-05,
-      "loss": 5.484,
-      "step": 26322
-    },
-    {
-      "epoch": 13.727770534550196,
-      "grad_norm": 1.4639639854431152,
-      "learning_rate": 7.405527638190955e-05,
-      "loss": 5.4063,
-      "step": 26323
-    },
-    {
-      "epoch": 13.728292046936115,
-      "grad_norm": 1.4644054174423218,
-      "learning_rate": 7.405427135678392e-05,
-      "loss": 5.2331,
-      "step": 26324
-    },
-    {
-      "epoch": 13.728813559322035,
-      "grad_norm": 1.5963212251663208,
-      "learning_rate": 7.40532663316583e-05,
-      "loss": 4.912,
-      "step": 26325
-    },
-    {
-      "epoch": 13.729335071707952,
-      "grad_norm": 1.501006007194519,
-      "learning_rate": 7.405226130653266e-05,
-      "loss": 5.7069,
-      "step": 26326
-    },
-    {
-      "epoch": 13.729856584093872,
-      "grad_norm": 1.4457000494003296,
-      "learning_rate": 7.405125628140704e-05,
-      "loss": 5.2542,
-      "step": 26327
-    },
-    {
-      "epoch": 13.730378096479791,
-      "grad_norm": 1.4152408838272095,
-      "learning_rate": 7.405025125628142e-05,
-      "loss": 4.7322,
-      "step": 26328
-    },
-    {
-      "epoch": 13.73089960886571,
-      "grad_norm": 1.4741460084915161,
-      "learning_rate": 7.404924623115578e-05,
-      "loss": 5.5146,
-      "step": 26329
-    },
-    {
-      "epoch": 13.73142112125163,
-      "grad_norm": 1.4437048435211182,
-      "learning_rate": 7.404824120603016e-05,
-      "loss": 5.4486,
-      "step": 26330
-    },
-    {
-      "epoch": 13.73194263363755,
-      "grad_norm": 1.5044375658035278,
-      "learning_rate": 7.404723618090452e-05,
-      "loss": 5.2145,
-      "step": 26331
-    },
-    {
-      "epoch": 13.732464146023467,
-      "grad_norm": 1.4556844234466553,
-      "learning_rate": 7.40462311557789e-05,
-      "loss": 5.5573,
-      "step": 26332
-    },
-    {
-      "epoch": 13.732985658409387,
-      "grad_norm": 1.431643009185791,
-      "learning_rate": 7.404522613065327e-05,
-      "loss": 5.5591,
-      "step": 26333
-    },
-    {
-      "epoch": 13.733507170795306,
-      "grad_norm": 1.5585272312164307,
-      "learning_rate": 7.404422110552764e-05,
-      "loss": 4.5254,
-      "step": 26334
-    },
-    {
-      "epoch": 13.734028683181226,
-      "grad_norm": 1.3811918497085571,
-      "learning_rate": 7.404321608040201e-05,
-      "loss": 5.8452,
-      "step": 26335
-    },
-    {
-      "epoch": 13.734550195567145,
-      "grad_norm": 1.4615755081176758,
-      "learning_rate": 7.404221105527638e-05,
-      "loss": 5.1662,
-      "step": 26336
-    },
-    {
-      "epoch": 13.735071707953065,
-      "grad_norm": 1.423498511314392,
-      "learning_rate": 7.404120603015075e-05,
-      "loss": 5.5802,
-      "step": 26337
-    },
-    {
-      "epoch": 13.735593220338982,
-      "grad_norm": 1.4805662631988525,
-      "learning_rate": 7.404020100502513e-05,
-      "loss": 5.4495,
-      "step": 26338
-    },
-    {
-      "epoch": 13.736114732724902,
-      "grad_norm": 1.4652297496795654,
-      "learning_rate": 7.40391959798995e-05,
-      "loss": 5.3078,
-      "step": 26339
-    },
-    {
-      "epoch": 13.736636245110821,
-      "grad_norm": 1.468070387840271,
-      "learning_rate": 7.403819095477388e-05,
-      "loss": 5.3727,
-      "step": 26340
-    },
-    {
-      "epoch": 13.73715775749674,
-      "grad_norm": 1.5637736320495605,
-      "learning_rate": 7.403718592964825e-05,
-      "loss": 5.5458,
-      "step": 26341
-    },
-    {
-      "epoch": 13.73767926988266,
-      "grad_norm": 1.4767038822174072,
-      "learning_rate": 7.403618090452261e-05,
-      "loss": 5.5236,
-      "step": 26342
-    },
-    {
-      "epoch": 13.73820078226858,
-      "grad_norm": 1.3895175457000732,
-      "learning_rate": 7.403517587939699e-05,
-      "loss": 5.8886,
-      "step": 26343
-    },
-    {
-      "epoch": 13.738722294654497,
-      "grad_norm": 1.5004115104675293,
-      "learning_rate": 7.403417085427135e-05,
-      "loss": 5.6567,
-      "step": 26344
-    },
-    {
-      "epoch": 13.739243807040417,
-      "grad_norm": 1.5777119398117065,
-      "learning_rate": 7.403316582914573e-05,
-      "loss": 5.3782,
-      "step": 26345
-    },
-    {
-      "epoch": 13.739765319426336,
-      "grad_norm": 1.526443362236023,
-      "learning_rate": 7.40321608040201e-05,
-      "loss": 5.1267,
-      "step": 26346
-    },
-    {
-      "epoch": 13.740286831812256,
-      "grad_norm": 1.6024081707000732,
-      "learning_rate": 7.403115577889447e-05,
-      "loss": 5.3713,
-      "step": 26347
-    },
-    {
-      "epoch": 13.740808344198175,
-      "grad_norm": 1.530344843864441,
-      "learning_rate": 7.403015075376885e-05,
-      "loss": 5.265,
-      "step": 26348
-    },
-    {
-      "epoch": 13.741329856584095,
-      "grad_norm": 1.4765715599060059,
-      "learning_rate": 7.402914572864323e-05,
-      "loss": 5.3534,
-      "step": 26349
-    },
-    {
-      "epoch": 13.741851368970012,
-      "grad_norm": 1.507709264755249,
-      "learning_rate": 7.40281407035176e-05,
-      "loss": 5.2719,
-      "step": 26350
-    },
-    {
-      "epoch": 13.742372881355932,
-      "grad_norm": 1.5508559942245483,
-      "learning_rate": 7.402713567839197e-05,
-      "loss": 5.2962,
-      "step": 26351
-    },
-    {
-      "epoch": 13.742894393741851,
-      "grad_norm": 1.524999976158142,
-      "learning_rate": 7.402613065326634e-05,
-      "loss": 5.2288,
-      "step": 26352
-    },
-    {
-      "epoch": 13.74341590612777,
-      "grad_norm": 1.482681393623352,
-      "learning_rate": 7.402512562814071e-05,
-      "loss": 5.3317,
-      "step": 26353
-    },
-    {
-      "epoch": 13.74393741851369,
-      "grad_norm": 1.5423485040664673,
-      "learning_rate": 7.402412060301508e-05,
-      "loss": 5.2761,
-      "step": 26354
-    },
-    {
-      "epoch": 13.74445893089961,
-      "grad_norm": 1.5471887588500977,
-      "learning_rate": 7.402311557788944e-05,
-      "loss": 5.6069,
-      "step": 26355
-    },
-    {
-      "epoch": 13.744980443285527,
-      "grad_norm": 1.646363615989685,
-      "learning_rate": 7.402211055276382e-05,
-      "loss": 4.9835,
-      "step": 26356
-    },
-    {
-      "epoch": 13.745501955671447,
-      "grad_norm": 1.4680784940719604,
-      "learning_rate": 7.402110552763818e-05,
-      "loss": 5.463,
-      "step": 26357
-    },
-    {
-      "epoch": 13.746023468057366,
-      "grad_norm": 1.5570931434631348,
-      "learning_rate": 7.402010050251256e-05,
-      "loss": 5.247,
-      "step": 26358
-    },
-    {
-      "epoch": 13.746544980443286,
-      "grad_norm": 1.4824610948562622,
-      "learning_rate": 7.401909547738694e-05,
-      "loss": 5.716,
-      "step": 26359
-    },
-    {
-      "epoch": 13.747066492829205,
-      "grad_norm": 1.5080955028533936,
-      "learning_rate": 7.401809045226132e-05,
-      "loss": 5.1804,
-      "step": 26360
-    },
-    {
-      "epoch": 13.747588005215125,
-      "grad_norm": 1.4445420503616333,
-      "learning_rate": 7.401708542713568e-05,
-      "loss": 4.6785,
-      "step": 26361
-    },
-    {
-      "epoch": 13.748109517601042,
-      "grad_norm": 1.4829580783843994,
-      "learning_rate": 7.401608040201006e-05,
-      "loss": 5.1302,
-      "step": 26362
-    },
-    {
-      "epoch": 13.748631029986962,
-      "grad_norm": 1.5469622611999512,
-      "learning_rate": 7.401507537688442e-05,
-      "loss": 4.6272,
-      "step": 26363
-    },
-    {
-      "epoch": 13.749152542372881,
-      "grad_norm": 1.4333226680755615,
-      "learning_rate": 7.40140703517588e-05,
-      "loss": 5.4068,
-      "step": 26364
-    },
-    {
-      "epoch": 13.7496740547588,
-      "grad_norm": 1.5405176877975464,
-      "learning_rate": 7.401306532663317e-05,
-      "loss": 5.3051,
-      "step": 26365
-    },
-    {
-      "epoch": 13.75019556714472,
-      "grad_norm": 1.436424970626831,
-      "learning_rate": 7.401206030150754e-05,
-      "loss": 5.3225,
-      "step": 26366
-    },
-    {
-      "epoch": 13.750717079530638,
-      "grad_norm": 1.6037594079971313,
-      "learning_rate": 7.401105527638191e-05,
-      "loss": 4.6573,
-      "step": 26367
-    },
-    {
-      "epoch": 13.751238591916557,
-      "grad_norm": 1.5370171070098877,
-      "learning_rate": 7.401005025125629e-05,
-      "loss": 5.1641,
-      "step": 26368
-    },
-    {
-      "epoch": 13.751760104302477,
-      "grad_norm": 1.3944084644317627,
-      "learning_rate": 7.400904522613066e-05,
-      "loss": 5.8155,
-      "step": 26369
-    },
-    {
-      "epoch": 13.752281616688396,
-      "grad_norm": 1.4413623809814453,
-      "learning_rate": 7.400804020100503e-05,
-      "loss": 5.167,
-      "step": 26370
-    },
-    {
-      "epoch": 13.752803129074316,
-      "grad_norm": 1.385704755783081,
-      "learning_rate": 7.40070351758794e-05,
-      "loss": 5.7975,
-      "step": 26371
-    },
-    {
-      "epoch": 13.753324641460235,
-      "grad_norm": 1.5941667556762695,
-      "learning_rate": 7.400603015075377e-05,
-      "loss": 5.3999,
-      "step": 26372
-    },
-    {
-      "epoch": 13.753846153846155,
-      "grad_norm": 1.320093035697937,
-      "learning_rate": 7.400502512562815e-05,
-      "loss": 5.8138,
-      "step": 26373
-    },
-    {
-      "epoch": 13.754367666232072,
-      "grad_norm": 1.4092094898223877,
-      "learning_rate": 7.400402010050251e-05,
-      "loss": 5.4197,
-      "step": 26374
-    },
-    {
-      "epoch": 13.754889178617992,
-      "grad_norm": 1.3717724084854126,
-      "learning_rate": 7.400301507537689e-05,
-      "loss": 5.5153,
-      "step": 26375
-    },
-    {
-      "epoch": 13.755410691003911,
-      "grad_norm": 1.4430638551712036,
-      "learning_rate": 7.400201005025125e-05,
-      "loss": 5.5917,
-      "step": 26376
-    },
-    {
-      "epoch": 13.75593220338983,
-      "grad_norm": 1.3952490091323853,
-      "learning_rate": 7.400100502512563e-05,
-      "loss": 5.6751,
-      "step": 26377
-    },
-    {
-      "epoch": 13.75645371577575,
-      "grad_norm": 1.6894973516464233,
-      "learning_rate": 7.4e-05,
-      "loss": 4.2621,
-      "step": 26378
-    },
-    {
-      "epoch": 13.756975228161668,
-      "grad_norm": 1.7109872102737427,
-      "learning_rate": 7.399899497487437e-05,
-      "loss": 4.6341,
-      "step": 26379
-    },
-    {
-      "epoch": 13.757496740547587,
-      "grad_norm": 1.4606164693832397,
-      "learning_rate": 7.399798994974875e-05,
-      "loss": 5.5662,
-      "step": 26380
-    },
-    {
-      "epoch": 13.758018252933507,
-      "grad_norm": 1.4700651168823242,
-      "learning_rate": 7.399698492462312e-05,
-      "loss": 5.438,
-      "step": 26381
-    },
-    {
-      "epoch": 13.758539765319426,
-      "grad_norm": 1.4906933307647705,
-      "learning_rate": 7.39959798994975e-05,
-      "loss": 5.1271,
-      "step": 26382
-    },
-    {
-      "epoch": 13.759061277705346,
-      "grad_norm": 1.5538970232009888,
-      "learning_rate": 7.399497487437186e-05,
-      "loss": 5.456,
-      "step": 26383
-    },
-    {
-      "epoch": 13.759582790091265,
-      "grad_norm": 1.394538164138794,
-      "learning_rate": 7.399396984924624e-05,
-      "loss": 5.2725,
-      "step": 26384
-    },
-    {
-      "epoch": 13.760104302477185,
-      "grad_norm": 1.5566115379333496,
-      "learning_rate": 7.39929648241206e-05,
-      "loss": 4.9309,
-      "step": 26385
-    },
-    {
-      "epoch": 13.760625814863102,
-      "grad_norm": 1.4490926265716553,
-      "learning_rate": 7.399195979899498e-05,
-      "loss": 5.4758,
-      "step": 26386
-    },
-    {
-      "epoch": 13.761147327249022,
-      "grad_norm": 1.4888676404953003,
-      "learning_rate": 7.399095477386934e-05,
-      "loss": 5.4139,
-      "step": 26387
-    },
-    {
-      "epoch": 13.761668839634941,
-      "grad_norm": 1.5319782495498657,
-      "learning_rate": 7.398994974874372e-05,
-      "loss": 5.254,
-      "step": 26388
-    },
-    {
-      "epoch": 13.76219035202086,
-      "grad_norm": 1.5528079271316528,
-      "learning_rate": 7.39889447236181e-05,
-      "loss": 5.2903,
-      "step": 26389
-    },
-    {
-      "epoch": 13.76271186440678,
-      "grad_norm": 1.4860446453094482,
-      "learning_rate": 7.398793969849248e-05,
-      "loss": 5.1487,
-      "step": 26390
-    },
-    {
-      "epoch": 13.763233376792698,
-      "grad_norm": 1.4436328411102295,
-      "learning_rate": 7.398693467336684e-05,
-      "loss": 5.5762,
-      "step": 26391
-    },
-    {
-      "epoch": 13.763754889178617,
-      "grad_norm": 1.5115957260131836,
-      "learning_rate": 7.398592964824122e-05,
-      "loss": 4.9015,
-      "step": 26392
-    },
-    {
-      "epoch": 13.764276401564537,
-      "grad_norm": 1.5232385396957397,
-      "learning_rate": 7.398492462311558e-05,
-      "loss": 5.2507,
-      "step": 26393
-    },
-    {
-      "epoch": 13.764797913950456,
-      "grad_norm": 1.5084216594696045,
-      "learning_rate": 7.398391959798995e-05,
-      "loss": 5.4185,
-      "step": 26394
-    },
-    {
-      "epoch": 13.765319426336376,
-      "grad_norm": 1.5073093175888062,
-      "learning_rate": 7.398291457286432e-05,
-      "loss": 5.3633,
-      "step": 26395
-    },
-    {
-      "epoch": 13.765840938722295,
-      "grad_norm": 1.5302278995513916,
-      "learning_rate": 7.398190954773869e-05,
-      "loss": 5.4567,
-      "step": 26396
-    },
-    {
-      "epoch": 13.766362451108213,
-      "grad_norm": 1.546541452407837,
-      "learning_rate": 7.398090452261307e-05,
-      "loss": 4.2343,
-      "step": 26397
-    },
-    {
-      "epoch": 13.766883963494132,
-      "grad_norm": 1.4552042484283447,
-      "learning_rate": 7.397989949748743e-05,
-      "loss": 5.4573,
-      "step": 26398
-    },
-    {
-      "epoch": 13.767405475880052,
-      "grad_norm": 1.529159665107727,
-      "learning_rate": 7.397889447236181e-05,
-      "loss": 5.2786,
-      "step": 26399
-    },
-    {
-      "epoch": 13.767926988265971,
-      "grad_norm": 1.416229248046875,
-      "learning_rate": 7.397788944723619e-05,
-      "loss": 5.7447,
-      "step": 26400
-    },
-    {
-      "epoch": 13.76844850065189,
-      "grad_norm": 1.5637726783752441,
-      "learning_rate": 7.397688442211056e-05,
-      "loss": 5.0406,
-      "step": 26401
-    },
-    {
-      "epoch": 13.76897001303781,
-      "grad_norm": 1.4302639961242676,
-      "learning_rate": 7.397587939698493e-05,
-      "loss": 5.7047,
-      "step": 26402
-    },
-    {
-      "epoch": 13.769491525423728,
-      "grad_norm": 1.5397229194641113,
-      "learning_rate": 7.39748743718593e-05,
-      "loss": 5.0968,
-      "step": 26403
-    },
-    {
-      "epoch": 13.770013037809647,
-      "grad_norm": 1.434273600578308,
-      "learning_rate": 7.397386934673367e-05,
-      "loss": 5.5531,
-      "step": 26404
-    },
-    {
-      "epoch": 13.770534550195567,
-      "grad_norm": 1.9008731842041016,
-      "learning_rate": 7.397286432160805e-05,
-      "loss": 4.773,
-      "step": 26405
-    },
-    {
-      "epoch": 13.771056062581486,
-      "grad_norm": 1.4220852851867676,
-      "learning_rate": 7.397185929648241e-05,
-      "loss": 5.2731,
-      "step": 26406
-    },
-    {
-      "epoch": 13.771577574967406,
-      "grad_norm": 1.3954423666000366,
-      "learning_rate": 7.397085427135678e-05,
-      "loss": 5.4523,
-      "step": 26407
-    },
-    {
-      "epoch": 13.772099087353325,
-      "grad_norm": 1.5403112173080444,
-      "learning_rate": 7.396984924623115e-05,
-      "loss": 5.033,
-      "step": 26408
-    },
-    {
-      "epoch": 13.772620599739243,
-      "grad_norm": 1.490450143814087,
-      "learning_rate": 7.396884422110553e-05,
-      "loss": 5.4627,
-      "step": 26409
-    },
-    {
-      "epoch": 13.773142112125162,
-      "grad_norm": 1.5374774932861328,
-      "learning_rate": 7.396783919597991e-05,
-      "loss": 5.3327,
-      "step": 26410
-    },
-    {
-      "epoch": 13.773663624511082,
-      "grad_norm": 1.4153592586517334,
-      "learning_rate": 7.396683417085427e-05,
-      "loss": 5.8982,
-      "step": 26411
-    },
-    {
-      "epoch": 13.774185136897001,
-      "grad_norm": 1.4049193859100342,
-      "learning_rate": 7.396582914572865e-05,
-      "loss": 5.5068,
-      "step": 26412
-    },
-    {
-      "epoch": 13.77470664928292,
-      "grad_norm": 1.4835766553878784,
-      "learning_rate": 7.396482412060302e-05,
-      "loss": 4.6239,
-      "step": 26413
-    },
-    {
-      "epoch": 13.77522816166884,
-      "grad_norm": 1.5727577209472656,
-      "learning_rate": 7.39638190954774e-05,
-      "loss": 5.4333,
-      "step": 26414
-    },
-    {
-      "epoch": 13.775749674054758,
-      "grad_norm": 1.5635651350021362,
-      "learning_rate": 7.396281407035176e-05,
-      "loss": 4.8797,
-      "step": 26415
-    },
-    {
-      "epoch": 13.776271186440677,
-      "grad_norm": 1.4108856916427612,
-      "learning_rate": 7.396180904522614e-05,
-      "loss": 5.5554,
-      "step": 26416
-    },
-    {
-      "epoch": 13.776792698826597,
-      "grad_norm": 1.3529754877090454,
-      "learning_rate": 7.39608040201005e-05,
-      "loss": 5.0123,
-      "step": 26417
-    },
-    {
-      "epoch": 13.777314211212516,
-      "grad_norm": 1.5034209489822388,
-      "learning_rate": 7.395979899497488e-05,
-      "loss": 5.4143,
-      "step": 26418
-    },
-    {
-      "epoch": 13.777835723598436,
-      "grad_norm": 1.463484525680542,
-      "learning_rate": 7.395879396984924e-05,
-      "loss": 5.3837,
-      "step": 26419
-    },
-    {
-      "epoch": 13.778357235984355,
-      "grad_norm": 1.3996230363845825,
-      "learning_rate": 7.395778894472362e-05,
-      "loss": 5.0837,
-      "step": 26420
-    },
-    {
-      "epoch": 13.778878748370273,
-      "grad_norm": 1.3740538358688354,
-      "learning_rate": 7.3956783919598e-05,
-      "loss": 5.0725,
-      "step": 26421
-    },
-    {
-      "epoch": 13.779400260756193,
-      "grad_norm": 1.5479425191879272,
-      "learning_rate": 7.395577889447236e-05,
-      "loss": 5.329,
-      "step": 26422
-    },
-    {
-      "epoch": 13.779921773142112,
-      "grad_norm": 1.4755905866622925,
-      "learning_rate": 7.395477386934674e-05,
-      "loss": 5.3834,
-      "step": 26423
-    },
-    {
-      "epoch": 13.780443285528031,
-      "grad_norm": 1.4012584686279297,
-      "learning_rate": 7.39537688442211e-05,
-      "loss": 4.9891,
-      "step": 26424
-    },
-    {
-      "epoch": 13.780964797913951,
-      "grad_norm": 1.5003336668014526,
-      "learning_rate": 7.395276381909548e-05,
-      "loss": 5.2476,
-      "step": 26425
-    },
-    {
-      "epoch": 13.78148631029987,
-      "grad_norm": 1.4437271356582642,
-      "learning_rate": 7.395175879396985e-05,
-      "loss": 5.5242,
-      "step": 26426
-    },
-    {
-      "epoch": 13.782007822685788,
-      "grad_norm": 1.399918794631958,
-      "learning_rate": 7.395075376884423e-05,
-      "loss": 5.3508,
-      "step": 26427
-    },
-    {
-      "epoch": 13.782529335071708,
-      "grad_norm": 1.4936041831970215,
-      "learning_rate": 7.394974874371859e-05,
-      "loss": 5.0351,
-      "step": 26428
-    },
-    {
-      "epoch": 13.783050847457627,
-      "grad_norm": 1.4257824420928955,
-      "learning_rate": 7.394874371859297e-05,
-      "loss": 5.8234,
-      "step": 26429
-    },
-    {
-      "epoch": 13.783572359843546,
-      "grad_norm": 1.5363036394119263,
-      "learning_rate": 7.394773869346735e-05,
-      "loss": 5.2239,
-      "step": 26430
-    },
-    {
-      "epoch": 13.784093872229466,
-      "grad_norm": 1.4956098794937134,
-      "learning_rate": 7.394673366834172e-05,
-      "loss": 5.4911,
-      "step": 26431
-    },
-    {
-      "epoch": 13.784615384615385,
-      "grad_norm": 1.5966894626617432,
-      "learning_rate": 7.394572864321609e-05,
-      "loss": 4.975,
-      "step": 26432
-    },
-    {
-      "epoch": 13.785136897001303,
-      "grad_norm": 1.553889274597168,
-      "learning_rate": 7.394472361809047e-05,
-      "loss": 4.8457,
-      "step": 26433
-    },
-    {
-      "epoch": 13.785658409387223,
-      "grad_norm": 1.496362328529358,
-      "learning_rate": 7.394371859296483e-05,
-      "loss": 5.0984,
-      "step": 26434
-    },
-    {
-      "epoch": 13.786179921773142,
-      "grad_norm": 1.4430550336837769,
-      "learning_rate": 7.39427135678392e-05,
-      "loss": 5.6545,
-      "step": 26435
-    },
-    {
-      "epoch": 13.786701434159061,
-      "grad_norm": 1.3767973184585571,
-      "learning_rate": 7.394170854271357e-05,
-      "loss": 5.7211,
-      "step": 26436
-    },
-    {
-      "epoch": 13.787222946544981,
-      "grad_norm": 1.4855372905731201,
-      "learning_rate": 7.394070351758794e-05,
-      "loss": 5.5372,
-      "step": 26437
-    },
-    {
-      "epoch": 13.7877444589309,
-      "grad_norm": 1.446744680404663,
-      "learning_rate": 7.393969849246231e-05,
-      "loss": 5.3369,
-      "step": 26438
-    },
-    {
-      "epoch": 13.788265971316818,
-      "grad_norm": 1.4282217025756836,
-      "learning_rate": 7.393869346733668e-05,
-      "loss": 5.3068,
-      "step": 26439
-    },
-    {
-      "epoch": 13.788787483702738,
-      "grad_norm": 1.3969601392745972,
-      "learning_rate": 7.393768844221106e-05,
-      "loss": 5.3525,
-      "step": 26440
-    },
-    {
-      "epoch": 13.789308996088657,
-      "grad_norm": 1.5351313352584839,
-      "learning_rate": 7.393668341708543e-05,
-      "loss": 5.4492,
-      "step": 26441
-    },
-    {
-      "epoch": 13.789830508474576,
-      "grad_norm": 1.5437430143356323,
-      "learning_rate": 7.393567839195981e-05,
-      "loss": 4.9325,
-      "step": 26442
-    },
-    {
-      "epoch": 13.790352020860496,
-      "grad_norm": 1.4813867807388306,
-      "learning_rate": 7.393467336683418e-05,
-      "loss": 5.4112,
-      "step": 26443
-    },
-    {
-      "epoch": 13.790873533246415,
-      "grad_norm": 1.4538313150405884,
-      "learning_rate": 7.393366834170855e-05,
-      "loss": 5.2482,
-      "step": 26444
-    },
-    {
-      "epoch": 13.791395045632333,
-      "grad_norm": 1.517920732498169,
-      "learning_rate": 7.393266331658292e-05,
-      "loss": 5.0593,
-      "step": 26445
-    },
-    {
-      "epoch": 13.791916558018253,
-      "grad_norm": 1.4986337423324585,
-      "learning_rate": 7.39316582914573e-05,
-      "loss": 5.1683,
-      "step": 26446
-    },
-    {
-      "epoch": 13.792438070404172,
-      "grad_norm": 1.3524914979934692,
-      "learning_rate": 7.393065326633166e-05,
-      "loss": 5.911,
-      "step": 26447
-    },
-    {
-      "epoch": 13.792959582790091,
-      "grad_norm": 1.2871067523956299,
-      "learning_rate": 7.392964824120602e-05,
-      "loss": 5.7825,
-      "step": 26448
-    },
-    {
-      "epoch": 13.793481095176011,
-      "grad_norm": 1.3542218208312988,
-      "learning_rate": 7.39286432160804e-05,
-      "loss": 5.8188,
-      "step": 26449
-    },
-    {
-      "epoch": 13.79400260756193,
-      "grad_norm": 1.3945420980453491,
-      "learning_rate": 7.392763819095478e-05,
-      "loss": 5.7745,
-      "step": 26450
-    },
-    {
-      "epoch": 13.794524119947848,
-      "grad_norm": 1.483166217803955,
-      "learning_rate": 7.392663316582916e-05,
-      "loss": 5.2426,
-      "step": 26451
-    },
-    {
-      "epoch": 13.795045632333768,
-      "grad_norm": 1.41875422000885,
-      "learning_rate": 7.392562814070352e-05,
-      "loss": 5.4855,
-      "step": 26452
-    },
-    {
-      "epoch": 13.795567144719687,
-      "grad_norm": 1.4704698324203491,
-      "learning_rate": 7.39246231155779e-05,
-      "loss": 5.0032,
-      "step": 26453
-    },
-    {
-      "epoch": 13.796088657105607,
-      "grad_norm": 1.4065637588500977,
-      "learning_rate": 7.392361809045226e-05,
-      "loss": 5.6701,
-      "step": 26454
-    },
-    {
-      "epoch": 13.796610169491526,
-      "grad_norm": 1.4952120780944824,
-      "learning_rate": 7.392261306532664e-05,
-      "loss": 5.3275,
-      "step": 26455
-    },
-    {
-      "epoch": 13.797131681877445,
-      "grad_norm": 1.369248628616333,
-      "learning_rate": 7.3921608040201e-05,
-      "loss": 5.3767,
-      "step": 26456
-    },
-    {
-      "epoch": 13.797653194263363,
-      "grad_norm": 1.3602946996688843,
-      "learning_rate": 7.392060301507538e-05,
-      "loss": 5.5782,
-      "step": 26457
-    },
-    {
-      "epoch": 13.798174706649283,
-      "grad_norm": 1.5441423654556274,
-      "learning_rate": 7.391959798994975e-05,
-      "loss": 5.0225,
-      "step": 26458
-    },
-    {
-      "epoch": 13.798696219035202,
-      "grad_norm": 1.413991093635559,
-      "learning_rate": 7.391859296482413e-05,
-      "loss": 4.9619,
-      "step": 26459
-    },
-    {
-      "epoch": 13.799217731421122,
-      "grad_norm": 1.4765141010284424,
-      "learning_rate": 7.391758793969849e-05,
-      "loss": 5.4183,
-      "step": 26460
-    },
-    {
-      "epoch": 13.799739243807041,
-      "grad_norm": 1.3677716255187988,
-      "learning_rate": 7.391658291457287e-05,
-      "loss": 5.5329,
-      "step": 26461
-    },
-    {
-      "epoch": 13.800260756192959,
-      "grad_norm": 1.4586433172225952,
-      "learning_rate": 7.391557788944725e-05,
-      "loss": 5.0815,
-      "step": 26462
-    },
-    {
-      "epoch": 13.800782268578878,
-      "grad_norm": 1.3711371421813965,
-      "learning_rate": 7.391457286432161e-05,
-      "loss": 5.4382,
-      "step": 26463
-    },
-    {
-      "epoch": 13.801303780964798,
-      "grad_norm": 1.4584420919418335,
-      "learning_rate": 7.391356783919599e-05,
-      "loss": 5.0904,
-      "step": 26464
-    },
-    {
-      "epoch": 13.801825293350717,
-      "grad_norm": 1.5582571029663086,
-      "learning_rate": 7.391256281407035e-05,
-      "loss": 4.5633,
-      "step": 26465
-    },
-    {
-      "epoch": 13.802346805736637,
-      "grad_norm": 1.4537912607192993,
-      "learning_rate": 7.391155778894473e-05,
-      "loss": 5.4942,
-      "step": 26466
-    },
-    {
-      "epoch": 13.802868318122556,
-      "grad_norm": 1.40284264087677,
-      "learning_rate": 7.39105527638191e-05,
-      "loss": 5.3799,
-      "step": 26467
-    },
-    {
-      "epoch": 13.803389830508475,
-      "grad_norm": 1.4334890842437744,
-      "learning_rate": 7.390954773869347e-05,
-      "loss": 5.9143,
-      "step": 26468
-    },
-    {
-      "epoch": 13.803911342894393,
-      "grad_norm": 1.5458240509033203,
-      "learning_rate": 7.390854271356784e-05,
-      "loss": 5.1991,
-      "step": 26469
-    },
-    {
-      "epoch": 13.804432855280313,
-      "grad_norm": 1.4049530029296875,
-      "learning_rate": 7.390753768844221e-05,
-      "loss": 5.4987,
-      "step": 26470
-    },
-    {
-      "epoch": 13.804954367666232,
-      "grad_norm": 1.6335694789886475,
-      "learning_rate": 7.390653266331659e-05,
-      "loss": 4.9888,
-      "step": 26471
-    },
-    {
-      "epoch": 13.805475880052152,
-      "grad_norm": 1.434862494468689,
-      "learning_rate": 7.390552763819097e-05,
-      "loss": 4.7372,
-      "step": 26472
-    },
-    {
-      "epoch": 13.805997392438071,
-      "grad_norm": 1.4554126262664795,
-      "learning_rate": 7.390452261306533e-05,
-      "loss": 5.2764,
-      "step": 26473
-    },
-    {
-      "epoch": 13.806518904823989,
-      "grad_norm": 1.3685393333435059,
-      "learning_rate": 7.39035175879397e-05,
-      "loss": 5.4117,
-      "step": 26474
-    },
-    {
-      "epoch": 13.807040417209908,
-      "grad_norm": 1.4697195291519165,
-      "learning_rate": 7.390251256281408e-05,
-      "loss": 4.9632,
-      "step": 26475
-    },
-    {
-      "epoch": 13.807561929595828,
-      "grad_norm": 1.3608393669128418,
-      "learning_rate": 7.390150753768844e-05,
-      "loss": 5.3465,
-      "step": 26476
-    },
-    {
-      "epoch": 13.808083441981747,
-      "grad_norm": 1.4113236665725708,
-      "learning_rate": 7.390050251256282e-05,
-      "loss": 5.1723,
-      "step": 26477
-    },
-    {
-      "epoch": 13.808604954367667,
-      "grad_norm": 1.436285376548767,
-      "learning_rate": 7.389949748743718e-05,
-      "loss": 5.3155,
-      "step": 26478
-    },
-    {
-      "epoch": 13.809126466753586,
-      "grad_norm": 1.4620888233184814,
-      "learning_rate": 7.389849246231156e-05,
-      "loss": 4.5991,
-      "step": 26479
-    },
-    {
-      "epoch": 13.809647979139505,
-      "grad_norm": 1.450632929801941,
-      "learning_rate": 7.389748743718592e-05,
-      "loss": 5.2029,
-      "step": 26480
-    },
-    {
-      "epoch": 13.810169491525423,
-      "grad_norm": 1.3802186250686646,
-      "learning_rate": 7.38964824120603e-05,
-      "loss": 5.872,
-      "step": 26481
-    },
-    {
-      "epoch": 13.810691003911343,
-      "grad_norm": 1.368727445602417,
-      "learning_rate": 7.389547738693468e-05,
-      "loss": 5.7114,
-      "step": 26482
-    },
-    {
-      "epoch": 13.811212516297262,
-      "grad_norm": 1.3834394216537476,
-      "learning_rate": 7.389447236180906e-05,
-      "loss": 5.4965,
-      "step": 26483
-    },
-    {
-      "epoch": 13.811734028683182,
-      "grad_norm": 1.4570322036743164,
-      "learning_rate": 7.389346733668342e-05,
-      "loss": 5.7206,
-      "step": 26484
-    },
-    {
-      "epoch": 13.812255541069101,
-      "grad_norm": 1.382216453552246,
-      "learning_rate": 7.38924623115578e-05,
-      "loss": 5.6353,
-      "step": 26485
-    },
-    {
-      "epoch": 13.812777053455019,
-      "grad_norm": 1.5031565427780151,
-      "learning_rate": 7.389145728643216e-05,
-      "loss": 5.2228,
-      "step": 26486
-    },
-    {
-      "epoch": 13.813298565840938,
-      "grad_norm": 1.778857707977295,
-      "learning_rate": 7.389045226130653e-05,
-      "loss": 4.9284,
-      "step": 26487
-    },
-    {
-      "epoch": 13.813820078226858,
-      "grad_norm": 1.5163599252700806,
-      "learning_rate": 7.388944723618091e-05,
-      "loss": 4.9919,
-      "step": 26488
-    },
-    {
-      "epoch": 13.814341590612777,
-      "grad_norm": 1.482047438621521,
-      "learning_rate": 7.388844221105527e-05,
-      "loss": 5.1664,
-      "step": 26489
-    },
-    {
-      "epoch": 13.814863102998697,
-      "grad_norm": 1.4220746755599976,
-      "learning_rate": 7.388743718592965e-05,
-      "loss": 5.4121,
-      "step": 26490
-    },
-    {
-      "epoch": 13.815384615384616,
-      "grad_norm": 1.5180259943008423,
-      "learning_rate": 7.388643216080403e-05,
-      "loss": 4.7004,
-      "step": 26491
-    },
-    {
-      "epoch": 13.815906127770534,
-      "grad_norm": 1.4680911302566528,
-      "learning_rate": 7.38854271356784e-05,
-      "loss": 5.5885,
-      "step": 26492
-    },
-    {
-      "epoch": 13.816427640156453,
-      "grad_norm": 1.3919789791107178,
-      "learning_rate": 7.388442211055277e-05,
-      "loss": 5.7083,
-      "step": 26493
-    },
-    {
-      "epoch": 13.816949152542373,
-      "grad_norm": 1.44468355178833,
-      "learning_rate": 7.388341708542715e-05,
-      "loss": 5.2311,
-      "step": 26494
-    },
-    {
-      "epoch": 13.817470664928292,
-      "grad_norm": 1.4075679779052734,
-      "learning_rate": 7.388241206030151e-05,
-      "loss": 5.5748,
-      "step": 26495
-    },
-    {
-      "epoch": 13.817992177314212,
-      "grad_norm": 1.444176197052002,
-      "learning_rate": 7.388140703517589e-05,
-      "loss": 5.3789,
-      "step": 26496
-    },
-    {
-      "epoch": 13.818513689700131,
-      "grad_norm": 1.4759618043899536,
-      "learning_rate": 7.388040201005025e-05,
-      "loss": 5.7278,
-      "step": 26497
-    },
-    {
-      "epoch": 13.819035202086049,
-      "grad_norm": 1.4898877143859863,
-      "learning_rate": 7.387939698492463e-05,
-      "loss": 5.0296,
-      "step": 26498
-    },
-    {
-      "epoch": 13.819556714471968,
-      "grad_norm": 1.3908525705337524,
-      "learning_rate": 7.3878391959799e-05,
-      "loss": 5.8774,
-      "step": 26499
-    },
-    {
-      "epoch": 13.820078226857888,
-      "grad_norm": 1.4607182741165161,
-      "learning_rate": 7.387738693467336e-05,
-      "loss": 4.9837,
-      "step": 26500
-    },
-    {
-      "epoch": 13.820599739243807,
-      "grad_norm": 1.3323724269866943,
-      "learning_rate": 7.387638190954774e-05,
-      "loss": 5.7404,
-      "step": 26501
-    },
-    {
-      "epoch": 13.821121251629727,
-      "grad_norm": 1.4811209440231323,
-      "learning_rate": 7.387537688442212e-05,
-      "loss": 5.0073,
-      "step": 26502
-    },
-    {
-      "epoch": 13.821642764015646,
-      "grad_norm": 1.404847264289856,
-      "learning_rate": 7.387437185929649e-05,
-      "loss": 5.663,
-      "step": 26503
-    },
-    {
-      "epoch": 13.822164276401564,
-      "grad_norm": 1.3617666959762573,
-      "learning_rate": 7.387336683417086e-05,
-      "loss": 5.7464,
-      "step": 26504
-    },
-    {
-      "epoch": 13.822685788787483,
-      "grad_norm": 1.3998384475708008,
-      "learning_rate": 7.387236180904523e-05,
-      "loss": 5.607,
-      "step": 26505
-    },
-    {
-      "epoch": 13.823207301173403,
-      "grad_norm": 1.4475196599960327,
-      "learning_rate": 7.38713567839196e-05,
-      "loss": 5.0385,
-      "step": 26506
-    },
-    {
-      "epoch": 13.823728813559322,
-      "grad_norm": 1.4412034749984741,
-      "learning_rate": 7.387035175879398e-05,
-      "loss": 5.57,
-      "step": 26507
-    },
-    {
-      "epoch": 13.824250325945242,
-      "grad_norm": 1.5133788585662842,
-      "learning_rate": 7.386934673366834e-05,
-      "loss": 5.4613,
-      "step": 26508
-    },
-    {
-      "epoch": 13.824771838331161,
-      "grad_norm": 1.4022142887115479,
-      "learning_rate": 7.386834170854272e-05,
-      "loss": 5.0918,
-      "step": 26509
-    },
-    {
-      "epoch": 13.825293350717079,
-      "grad_norm": 1.5118639469146729,
-      "learning_rate": 7.386733668341708e-05,
-      "loss": 4.9873,
-      "step": 26510
-    },
-    {
-      "epoch": 13.825814863102998,
-      "grad_norm": 1.420628547668457,
-      "learning_rate": 7.386633165829146e-05,
-      "loss": 5.4842,
-      "step": 26511
-    },
-    {
-      "epoch": 13.826336375488918,
-      "grad_norm": 1.4550787210464478,
-      "learning_rate": 7.386532663316583e-05,
-      "loss": 5.8457,
-      "step": 26512
-    },
-    {
-      "epoch": 13.826857887874837,
-      "grad_norm": 1.3898526430130005,
-      "learning_rate": 7.38643216080402e-05,
-      "loss": 5.6544,
-      "step": 26513
-    },
-    {
-      "epoch": 13.827379400260757,
-      "grad_norm": 1.461527943611145,
-      "learning_rate": 7.386331658291458e-05,
-      "loss": 4.654,
-      "step": 26514
-    },
-    {
-      "epoch": 13.827900912646676,
-      "grad_norm": 1.4114949703216553,
-      "learning_rate": 7.386231155778895e-05,
-      "loss": 5.5261,
-      "step": 26515
-    },
-    {
-      "epoch": 13.828422425032594,
-      "grad_norm": 1.3397787809371948,
-      "learning_rate": 7.386130653266332e-05,
-      "loss": 5.8636,
-      "step": 26516
-    },
-    {
-      "epoch": 13.828943937418513,
-      "grad_norm": 1.4602088928222656,
-      "learning_rate": 7.386030150753769e-05,
-      "loss": 5.4346,
-      "step": 26517
-    },
-    {
-      "epoch": 13.829465449804433,
-      "grad_norm": 1.4274699687957764,
-      "learning_rate": 7.385929648241207e-05,
-      "loss": 5.0238,
-      "step": 26518
-    },
-    {
-      "epoch": 13.829986962190352,
-      "grad_norm": 1.4114453792572021,
-      "learning_rate": 7.385829145728643e-05,
-      "loss": 5.374,
-      "step": 26519
-    },
-    {
-      "epoch": 13.830508474576272,
-      "grad_norm": 1.4334321022033691,
-      "learning_rate": 7.385728643216081e-05,
-      "loss": 5.1367,
-      "step": 26520
-    },
-    {
-      "epoch": 13.831029986962191,
-      "grad_norm": 1.3766230344772339,
-      "learning_rate": 7.385628140703517e-05,
-      "loss": 5.4523,
-      "step": 26521
-    },
-    {
-      "epoch": 13.831551499348109,
-      "grad_norm": 1.425680160522461,
-      "learning_rate": 7.385527638190955e-05,
-      "loss": 5.9825,
-      "step": 26522
-    },
-    {
-      "epoch": 13.832073011734028,
-      "grad_norm": 1.4496358633041382,
-      "learning_rate": 7.385427135678393e-05,
-      "loss": 5.5025,
-      "step": 26523
-    },
-    {
-      "epoch": 13.832594524119948,
-      "grad_norm": 1.4499424695968628,
-      "learning_rate": 7.38532663316583e-05,
-      "loss": 5.3854,
-      "step": 26524
-    },
-    {
-      "epoch": 13.833116036505867,
-      "grad_norm": 1.425278902053833,
-      "learning_rate": 7.385226130653267e-05,
-      "loss": 5.166,
-      "step": 26525
-    },
-    {
-      "epoch": 13.833637548891787,
-      "grad_norm": 1.5153617858886719,
-      "learning_rate": 7.385125628140705e-05,
-      "loss": 5.5275,
-      "step": 26526
-    },
-    {
-      "epoch": 13.834159061277706,
-      "grad_norm": 1.5246303081512451,
-      "learning_rate": 7.385025125628141e-05,
-      "loss": 5.0897,
-      "step": 26527
-    },
-    {
-      "epoch": 13.834680573663624,
-      "grad_norm": 1.5999515056610107,
-      "learning_rate": 7.384924623115578e-05,
-      "loss": 5.1258,
-      "step": 26528
-    },
-    {
-      "epoch": 13.835202086049543,
-      "grad_norm": 1.4156593084335327,
-      "learning_rate": 7.384824120603015e-05,
-      "loss": 5.4218,
-      "step": 26529
-    },
-    {
-      "epoch": 13.835723598435463,
-      "grad_norm": 1.3648791313171387,
-      "learning_rate": 7.384723618090452e-05,
-      "loss": 5.7005,
-      "step": 26530
-    },
-    {
-      "epoch": 13.836245110821382,
-      "grad_norm": 1.4114359617233276,
-      "learning_rate": 7.38462311557789e-05,
-      "loss": 5.7446,
-      "step": 26531
-    },
-    {
-      "epoch": 13.836766623207302,
-      "grad_norm": 1.5978981256484985,
-      "learning_rate": 7.384522613065326e-05,
-      "loss": 5.2427,
-      "step": 26532
-    },
-    {
-      "epoch": 13.837288135593221,
-      "grad_norm": 1.3401691913604736,
-      "learning_rate": 7.384422110552764e-05,
-      "loss": 5.763,
-      "step": 26533
-    },
-    {
-      "epoch": 13.837809647979139,
-      "grad_norm": 1.4701534509658813,
-      "learning_rate": 7.384321608040202e-05,
-      "loss": 5.3439,
-      "step": 26534
-    },
-    {
-      "epoch": 13.838331160365058,
-      "grad_norm": 1.4014315605163574,
-      "learning_rate": 7.38422110552764e-05,
-      "loss": 4.9311,
-      "step": 26535
-    },
-    {
-      "epoch": 13.838852672750978,
-      "grad_norm": 1.4601953029632568,
-      "learning_rate": 7.384120603015076e-05,
-      "loss": 5.0572,
-      "step": 26536
-    },
-    {
-      "epoch": 13.839374185136897,
-      "grad_norm": 1.4669698476791382,
-      "learning_rate": 7.384020100502514e-05,
-      "loss": 5.3103,
-      "step": 26537
-    },
-    {
-      "epoch": 13.839895697522817,
-      "grad_norm": 1.479597568511963,
-      "learning_rate": 7.38391959798995e-05,
-      "loss": 5.3424,
-      "step": 26538
-    },
-    {
-      "epoch": 13.840417209908736,
-      "grad_norm": 1.478246808052063,
-      "learning_rate": 7.383819095477388e-05,
-      "loss": 5.669,
-      "step": 26539
-    },
-    {
-      "epoch": 13.840938722294654,
-      "grad_norm": 1.5578365325927734,
-      "learning_rate": 7.383718592964824e-05,
-      "loss": 5.5864,
-      "step": 26540
-    },
-    {
-      "epoch": 13.841460234680573,
-      "grad_norm": 1.44267737865448,
-      "learning_rate": 7.38361809045226e-05,
-      "loss": 5.4649,
-      "step": 26541
-    },
-    {
-      "epoch": 13.841981747066493,
-      "grad_norm": 1.509235143661499,
-      "learning_rate": 7.383517587939698e-05,
-      "loss": 5.2228,
-      "step": 26542
-    },
-    {
-      "epoch": 13.842503259452412,
-      "grad_norm": 1.5816330909729004,
-      "learning_rate": 7.383417085427136e-05,
-      "loss": 5.4159,
-      "step": 26543
-    },
-    {
-      "epoch": 13.843024771838332,
-      "grad_norm": 1.5089608430862427,
-      "learning_rate": 7.383316582914574e-05,
-      "loss": 5.6484,
-      "step": 26544
-    },
-    {
-      "epoch": 13.843546284224251,
-      "grad_norm": 1.3855961561203003,
-      "learning_rate": 7.38321608040201e-05,
-      "loss": 5.505,
-      "step": 26545
-    },
-    {
-      "epoch": 13.844067796610169,
-      "grad_norm": 1.456153392791748,
-      "learning_rate": 7.383115577889448e-05,
-      "loss": 5.4411,
-      "step": 26546
-    },
-    {
-      "epoch": 13.844589308996088,
-      "grad_norm": 1.4587428569793701,
-      "learning_rate": 7.383015075376885e-05,
-      "loss": 5.4471,
-      "step": 26547
-    },
-    {
-      "epoch": 13.845110821382008,
-      "grad_norm": 1.4712167978286743,
-      "learning_rate": 7.382914572864322e-05,
-      "loss": 5.3967,
-      "step": 26548
-    },
-    {
-      "epoch": 13.845632333767927,
-      "grad_norm": 1.3772804737091064,
-      "learning_rate": 7.382814070351759e-05,
-      "loss": 5.6875,
-      "step": 26549
-    },
-    {
-      "epoch": 13.846153846153847,
-      "grad_norm": 1.5212371349334717,
-      "learning_rate": 7.382713567839197e-05,
-      "loss": 4.8244,
-      "step": 26550
-    },
-    {
-      "epoch": 13.846675358539766,
-      "grad_norm": 1.519421935081482,
-      "learning_rate": 7.382613065326633e-05,
-      "loss": 5.0225,
-      "step": 26551
-    },
-    {
-      "epoch": 13.847196870925684,
-      "grad_norm": 1.409741759300232,
-      "learning_rate": 7.382512562814071e-05,
-      "loss": 5.6238,
-      "step": 26552
-    },
-    {
-      "epoch": 13.847718383311603,
-      "grad_norm": 1.4318193197250366,
-      "learning_rate": 7.382412060301507e-05,
-      "loss": 5.4995,
-      "step": 26553
-    },
-    {
-      "epoch": 13.848239895697523,
-      "grad_norm": 1.4362225532531738,
-      "learning_rate": 7.382311557788945e-05,
-      "loss": 4.9806,
-      "step": 26554
-    },
-    {
-      "epoch": 13.848761408083442,
-      "grad_norm": 1.4308249950408936,
-      "learning_rate": 7.382211055276383e-05,
-      "loss": 5.5641,
-      "step": 26555
-    },
-    {
-      "epoch": 13.849282920469362,
-      "grad_norm": 1.6544915437698364,
-      "learning_rate": 7.382110552763819e-05,
-      "loss": 4.6351,
-      "step": 26556
-    },
-    {
-      "epoch": 13.84980443285528,
-      "grad_norm": 1.5612237453460693,
-      "learning_rate": 7.382010050251257e-05,
-      "loss": 5.0211,
-      "step": 26557
-    },
-    {
-      "epoch": 13.850325945241199,
-      "grad_norm": 1.4336183071136475,
-      "learning_rate": 7.381909547738693e-05,
-      "loss": 5.3679,
-      "step": 26558
-    },
-    {
-      "epoch": 13.850847457627118,
-      "grad_norm": 1.467228651046753,
-      "learning_rate": 7.381809045226131e-05,
-      "loss": 5.4192,
-      "step": 26559
-    },
-    {
-      "epoch": 13.851368970013038,
-      "grad_norm": 1.5703413486480713,
-      "learning_rate": 7.381708542713568e-05,
-      "loss": 5.1837,
-      "step": 26560
-    },
-    {
-      "epoch": 13.851890482398957,
-      "grad_norm": 1.392774224281311,
-      "learning_rate": 7.381608040201005e-05,
-      "loss": 5.5974,
-      "step": 26561
-    },
-    {
-      "epoch": 13.852411994784877,
-      "grad_norm": 1.4373931884765625,
-      "learning_rate": 7.381507537688442e-05,
-      "loss": 5.3749,
-      "step": 26562
-    },
-    {
-      "epoch": 13.852933507170796,
-      "grad_norm": 1.4433844089508057,
-      "learning_rate": 7.38140703517588e-05,
-      "loss": 5.6808,
-      "step": 26563
-    },
-    {
-      "epoch": 13.853455019556714,
-      "grad_norm": 1.4451144933700562,
-      "learning_rate": 7.381306532663317e-05,
-      "loss": 5.2216,
-      "step": 26564
-    },
-    {
-      "epoch": 13.853976531942633,
-      "grad_norm": 1.3851608037948608,
-      "learning_rate": 7.381206030150755e-05,
-      "loss": 5.5259,
-      "step": 26565
-    },
-    {
-      "epoch": 13.854498044328553,
-      "grad_norm": 1.4585822820663452,
-      "learning_rate": 7.381105527638192e-05,
-      "loss": 5.5045,
-      "step": 26566
-    },
-    {
-      "epoch": 13.855019556714472,
-      "grad_norm": 1.503200650215149,
-      "learning_rate": 7.381005025125628e-05,
-      "loss": 5.4925,
-      "step": 26567
-    },
-    {
-      "epoch": 13.855541069100392,
-      "grad_norm": 1.4481111764907837,
-      "learning_rate": 7.380904522613066e-05,
-      "loss": 5.3854,
-      "step": 26568
-    },
-    {
-      "epoch": 13.85606258148631,
-      "grad_norm": 1.511469841003418,
-      "learning_rate": 7.380804020100502e-05,
-      "loss": 5.3785,
-      "step": 26569
-    },
-    {
-      "epoch": 13.856584093872229,
-      "grad_norm": 1.4648741483688354,
-      "learning_rate": 7.38070351758794e-05,
-      "loss": 5.0836,
-      "step": 26570
-    },
-    {
-      "epoch": 13.857105606258148,
-      "grad_norm": 1.4487144947052002,
-      "learning_rate": 7.380603015075377e-05,
-      "loss": 5.5105,
-      "step": 26571
-    },
-    {
-      "epoch": 13.857627118644068,
-      "grad_norm": 1.4355738162994385,
-      "learning_rate": 7.380502512562814e-05,
-      "loss": 5.4768,
-      "step": 26572
-    },
-    {
-      "epoch": 13.858148631029987,
-      "grad_norm": 1.4884871244430542,
-      "learning_rate": 7.380402010050251e-05,
-      "loss": 5.3628,
-      "step": 26573
-    },
-    {
-      "epoch": 13.858670143415907,
-      "grad_norm": 1.4084011316299438,
-      "learning_rate": 7.380301507537688e-05,
-      "loss": 5.4352,
-      "step": 26574
-    },
-    {
-      "epoch": 13.859191655801826,
-      "grad_norm": 1.4333350658416748,
-      "learning_rate": 7.380201005025126e-05,
-      "loss": 5.8894,
-      "step": 26575
-    },
-    {
-      "epoch": 13.859713168187744,
-      "grad_norm": 1.5169113874435425,
-      "learning_rate": 7.380100502512564e-05,
-      "loss": 4.8846,
-      "step": 26576
-    },
-    {
-      "epoch": 13.860234680573663,
-      "grad_norm": 1.3853192329406738,
-      "learning_rate": 7.38e-05,
-      "loss": 5.49,
-      "step": 26577
-    },
-    {
-      "epoch": 13.860756192959583,
-      "grad_norm": 1.4637190103530884,
-      "learning_rate": 7.379899497487438e-05,
-      "loss": 5.5796,
-      "step": 26578
-    },
-    {
-      "epoch": 13.861277705345502,
-      "grad_norm": 1.546027421951294,
-      "learning_rate": 7.379798994974875e-05,
-      "loss": 5.269,
-      "step": 26579
-    },
-    {
-      "epoch": 13.861799217731422,
-      "grad_norm": 1.3591747283935547,
-      "learning_rate": 7.379698492462311e-05,
-      "loss": 5.9401,
-      "step": 26580
-    },
-    {
-      "epoch": 13.86232073011734,
-      "grad_norm": 1.3200143575668335,
-      "learning_rate": 7.379597989949749e-05,
-      "loss": 5.6316,
-      "step": 26581
-    },
-    {
-      "epoch": 13.862842242503259,
-      "grad_norm": 1.3101123571395874,
-      "learning_rate": 7.379497487437185e-05,
-      "loss": 5.2788,
-      "step": 26582
-    },
-    {
-      "epoch": 13.863363754889178,
-      "grad_norm": 1.4713033437728882,
-      "learning_rate": 7.379396984924623e-05,
-      "loss": 5.3545,
-      "step": 26583
-    },
-    {
-      "epoch": 13.863885267275098,
-      "grad_norm": 1.4868366718292236,
-      "learning_rate": 7.379296482412061e-05,
-      "loss": 5.1477,
-      "step": 26584
-    },
-    {
-      "epoch": 13.864406779661017,
-      "grad_norm": 1.383829951286316,
-      "learning_rate": 7.379195979899499e-05,
-      "loss": 5.5549,
-      "step": 26585
-    },
-    {
-      "epoch": 13.864928292046937,
-      "grad_norm": 1.3996530771255493,
-      "learning_rate": 7.379095477386935e-05,
-      "loss": 5.8232,
-      "step": 26586
-    },
-    {
-      "epoch": 13.865449804432854,
-      "grad_norm": 1.4808762073516846,
-      "learning_rate": 7.378994974874373e-05,
-      "loss": 5.2201,
-      "step": 26587
-    },
-    {
-      "epoch": 13.865971316818774,
-      "grad_norm": 1.4125200510025024,
-      "learning_rate": 7.37889447236181e-05,
-      "loss": 5.5388,
-      "step": 26588
-    },
-    {
-      "epoch": 13.866492829204693,
-      "grad_norm": 1.4406697750091553,
-      "learning_rate": 7.378793969849247e-05,
-      "loss": 5.6218,
-      "step": 26589
-    },
-    {
-      "epoch": 13.867014341590613,
-      "grad_norm": 1.3532670736312866,
-      "learning_rate": 7.378693467336684e-05,
-      "loss": 5.6486,
-      "step": 26590
-    },
-    {
-      "epoch": 13.867535853976532,
-      "grad_norm": 1.4575310945510864,
-      "learning_rate": 7.378592964824121e-05,
-      "loss": 4.9575,
-      "step": 26591
-    },
-    {
-      "epoch": 13.868057366362452,
-      "grad_norm": 1.4450225830078125,
-      "learning_rate": 7.378492462311558e-05,
-      "loss": 5.365,
-      "step": 26592
-    },
-    {
-      "epoch": 13.86857887874837,
-      "grad_norm": 1.4031126499176025,
-      "learning_rate": 7.378391959798994e-05,
-      "loss": 5.6124,
-      "step": 26593
-    },
-    {
-      "epoch": 13.869100391134289,
-      "grad_norm": 1.5389498472213745,
-      "learning_rate": 7.378291457286432e-05,
-      "loss": 5.2427,
-      "step": 26594
-    },
-    {
-      "epoch": 13.869621903520208,
-      "grad_norm": 1.3555395603179932,
-      "learning_rate": 7.37819095477387e-05,
-      "loss": 5.0971,
-      "step": 26595
-    },
-    {
-      "epoch": 13.870143415906128,
-      "grad_norm": 1.465713620185852,
-      "learning_rate": 7.378090452261308e-05,
-      "loss": 5.5205,
-      "step": 26596
-    },
-    {
-      "epoch": 13.870664928292047,
-      "grad_norm": 1.4382522106170654,
-      "learning_rate": 7.377989949748744e-05,
-      "loss": 5.1945,
-      "step": 26597
-    },
-    {
-      "epoch": 13.871186440677967,
-      "grad_norm": 1.5855571031570435,
-      "learning_rate": 7.377889447236182e-05,
-      "loss": 5.4128,
-      "step": 26598
-    },
-    {
-      "epoch": 13.871707953063884,
-      "grad_norm": 1.4720386266708374,
-      "learning_rate": 7.377788944723618e-05,
-      "loss": 5.4909,
-      "step": 26599
-    },
-    {
-      "epoch": 13.872229465449804,
-      "grad_norm": 1.404062032699585,
-      "learning_rate": 7.377688442211056e-05,
-      "loss": 5.3406,
-      "step": 26600
-    },
-    {
-      "epoch": 13.872750977835723,
-      "grad_norm": 1.620447039604187,
-      "learning_rate": 7.377587939698492e-05,
-      "loss": 4.5968,
-      "step": 26601
-    },
-    {
-      "epoch": 13.873272490221643,
-      "grad_norm": 1.3941786289215088,
-      "learning_rate": 7.37748743718593e-05,
-      "loss": 5.2166,
-      "step": 26602
-    },
-    {
-      "epoch": 13.873794002607562,
-      "grad_norm": 1.5450648069381714,
-      "learning_rate": 7.377386934673367e-05,
-      "loss": 4.9481,
-      "step": 26603
-    },
-    {
-      "epoch": 13.874315514993482,
-      "grad_norm": 1.5950566530227661,
-      "learning_rate": 7.377286432160804e-05,
-      "loss": 4.8157,
-      "step": 26604
-    },
-    {
-      "epoch": 13.8748370273794,
-      "grad_norm": 1.5029428005218506,
-      "learning_rate": 7.377185929648242e-05,
-      "loss": 5.1117,
-      "step": 26605
-    },
-    {
-      "epoch": 13.875358539765319,
-      "grad_norm": 1.4175562858581543,
-      "learning_rate": 7.37708542713568e-05,
-      "loss": 5.6209,
-      "step": 26606
-    },
-    {
-      "epoch": 13.875880052151238,
-      "grad_norm": 1.4179774522781372,
-      "learning_rate": 7.376984924623116e-05,
-      "loss": 5.302,
-      "step": 26607
-    },
-    {
-      "epoch": 13.876401564537158,
-      "grad_norm": 1.432729959487915,
-      "learning_rate": 7.376884422110553e-05,
-      "loss": 5.6803,
-      "step": 26608
-    },
-    {
-      "epoch": 13.876923076923077,
-      "grad_norm": 1.5391219854354858,
-      "learning_rate": 7.37678391959799e-05,
-      "loss": 5.5157,
-      "step": 26609
-    },
-    {
-      "epoch": 13.877444589308997,
-      "grad_norm": 1.469746708869934,
-      "learning_rate": 7.376683417085427e-05,
-      "loss": 5.5035,
-      "step": 26610
-    },
-    {
-      "epoch": 13.877966101694915,
-      "grad_norm": 1.473394751548767,
-      "learning_rate": 7.376582914572865e-05,
-      "loss": 5.4138,
-      "step": 26611
-    },
-    {
-      "epoch": 13.878487614080834,
-      "grad_norm": 1.5536819696426392,
-      "learning_rate": 7.376482412060301e-05,
-      "loss": 5.176,
-      "step": 26612
-    },
-    {
-      "epoch": 13.879009126466753,
-      "grad_norm": 1.4682466983795166,
-      "learning_rate": 7.376381909547739e-05,
-      "loss": 5.367,
-      "step": 26613
-    },
-    {
-      "epoch": 13.879530638852673,
-      "grad_norm": 1.4385857582092285,
-      "learning_rate": 7.376281407035175e-05,
-      "loss": 5.2971,
-      "step": 26614
-    },
-    {
-      "epoch": 13.880052151238592,
-      "grad_norm": 1.482029676437378,
-      "learning_rate": 7.376180904522613e-05,
-      "loss": 5.1889,
-      "step": 26615
-    },
-    {
-      "epoch": 13.880573663624512,
-      "grad_norm": 1.427282452583313,
-      "learning_rate": 7.376080402010051e-05,
-      "loss": 5.3557,
-      "step": 26616
-    },
-    {
-      "epoch": 13.88109517601043,
-      "grad_norm": 1.4401918649673462,
-      "learning_rate": 7.375979899497489e-05,
-      "loss": 5.5741,
-      "step": 26617
-    },
-    {
-      "epoch": 13.881616688396349,
-      "grad_norm": 1.6312189102172852,
-      "learning_rate": 7.375879396984925e-05,
-      "loss": 5.3206,
-      "step": 26618
-    },
-    {
-      "epoch": 13.882138200782268,
-      "grad_norm": 1.4221608638763428,
-      "learning_rate": 7.375778894472363e-05,
-      "loss": 4.9026,
-      "step": 26619
-    },
-    {
-      "epoch": 13.882659713168188,
-      "grad_norm": 1.4329110383987427,
-      "learning_rate": 7.3756783919598e-05,
-      "loss": 5.3631,
-      "step": 26620
-    },
-    {
-      "epoch": 13.883181225554107,
-      "grad_norm": 1.452627420425415,
-      "learning_rate": 7.375577889447236e-05,
-      "loss": 5.2424,
-      "step": 26621
-    },
-    {
-      "epoch": 13.883702737940027,
-      "grad_norm": 1.323655128479004,
-      "learning_rate": 7.375477386934674e-05,
-      "loss": 5.9703,
-      "step": 26622
-    },
-    {
-      "epoch": 13.884224250325945,
-      "grad_norm": 1.5112659931182861,
-      "learning_rate": 7.37537688442211e-05,
-      "loss": 4.9537,
-      "step": 26623
-    },
-    {
-      "epoch": 13.884745762711864,
-      "grad_norm": 1.4555965662002563,
-      "learning_rate": 7.375276381909548e-05,
-      "loss": 5.4629,
-      "step": 26624
-    },
-    {
-      "epoch": 13.885267275097783,
-      "grad_norm": 1.733452320098877,
-      "learning_rate": 7.375175879396986e-05,
-      "loss": 4.9269,
-      "step": 26625
-    },
-    {
-      "epoch": 13.885788787483703,
-      "grad_norm": 1.4827934503555298,
-      "learning_rate": 7.375075376884423e-05,
-      "loss": 5.1021,
-      "step": 26626
-    },
-    {
-      "epoch": 13.886310299869622,
-      "grad_norm": 1.3795722723007202,
-      "learning_rate": 7.37497487437186e-05,
-      "loss": 5.2372,
-      "step": 26627
-    },
-    {
-      "epoch": 13.886831812255542,
-      "grad_norm": 1.4426764249801636,
-      "learning_rate": 7.374874371859298e-05,
-      "loss": 5.3715,
-      "step": 26628
-    },
-    {
-      "epoch": 13.88735332464146,
-      "grad_norm": 1.4397040605545044,
-      "learning_rate": 7.374773869346734e-05,
-      "loss": 5.6023,
-      "step": 26629
-    },
-    {
-      "epoch": 13.887874837027379,
-      "grad_norm": 1.3743141889572144,
-      "learning_rate": 7.374673366834172e-05,
-      "loss": 4.9576,
-      "step": 26630
-    },
-    {
-      "epoch": 13.888396349413298,
-      "grad_norm": 1.5843952894210815,
-      "learning_rate": 7.374572864321608e-05,
-      "loss": 5.3932,
-      "step": 26631
-    },
-    {
-      "epoch": 13.888917861799218,
-      "grad_norm": 1.4581767320632935,
-      "learning_rate": 7.374472361809046e-05,
-      "loss": 5.1086,
-      "step": 26632
-    },
-    {
-      "epoch": 13.889439374185137,
-      "grad_norm": 1.4186147451400757,
-      "learning_rate": 7.374371859296482e-05,
-      "loss": 5.5278,
-      "step": 26633
-    },
-    {
-      "epoch": 13.889960886571057,
-      "grad_norm": 1.4549742937088013,
-      "learning_rate": 7.374271356783919e-05,
-      "loss": 5.5259,
-      "step": 26634
-    },
-    {
-      "epoch": 13.890482398956975,
-      "grad_norm": 1.5221470594406128,
-      "learning_rate": 7.374170854271357e-05,
-      "loss": 5.3998,
-      "step": 26635
-    },
-    {
-      "epoch": 13.891003911342894,
-      "grad_norm": 1.5695905685424805,
-      "learning_rate": 7.374070351758794e-05,
-      "loss": 5.4041,
-      "step": 26636
-    },
-    {
-      "epoch": 13.891525423728813,
-      "grad_norm": 1.4858338832855225,
-      "learning_rate": 7.373969849246232e-05,
-      "loss": 5.6866,
-      "step": 26637
-    },
-    {
-      "epoch": 13.892046936114733,
-      "grad_norm": 1.5797772407531738,
-      "learning_rate": 7.373869346733669e-05,
-      "loss": 5.0802,
-      "step": 26638
-    },
-    {
-      "epoch": 13.892568448500652,
-      "grad_norm": 1.4593744277954102,
-      "learning_rate": 7.373768844221106e-05,
-      "loss": 5.2031,
-      "step": 26639
-    },
-    {
-      "epoch": 13.893089960886572,
-      "grad_norm": 1.37606680393219,
-      "learning_rate": 7.373668341708543e-05,
-      "loss": 5.6182,
-      "step": 26640
-    },
-    {
-      "epoch": 13.89361147327249,
-      "grad_norm": 1.4816226959228516,
-      "learning_rate": 7.37356783919598e-05,
-      "loss": 5.6605,
-      "step": 26641
-    },
-    {
-      "epoch": 13.894132985658409,
-      "grad_norm": 1.4543999433517456,
-      "learning_rate": 7.373467336683417e-05,
-      "loss": 5.1805,
-      "step": 26642
-    },
-    {
-      "epoch": 13.894654498044329,
-      "grad_norm": 1.4872883558273315,
-      "learning_rate": 7.373366834170855e-05,
-      "loss": 5.437,
-      "step": 26643
-    },
-    {
-      "epoch": 13.895176010430248,
-      "grad_norm": 1.4359242916107178,
-      "learning_rate": 7.373266331658291e-05,
-      "loss": 5.2835,
-      "step": 26644
-    },
-    {
-      "epoch": 13.895697522816167,
-      "grad_norm": 1.4933618307113647,
-      "learning_rate": 7.373165829145729e-05,
-      "loss": 5.3383,
-      "step": 26645
-    },
-    {
-      "epoch": 13.896219035202087,
-      "grad_norm": 1.5379003286361694,
-      "learning_rate": 7.373065326633167e-05,
-      "loss": 5.5382,
-      "step": 26646
-    },
-    {
-      "epoch": 13.896740547588005,
-      "grad_norm": 1.526021122932434,
-      "learning_rate": 7.372964824120603e-05,
-      "loss": 5.5964,
-      "step": 26647
-    },
-    {
-      "epoch": 13.897262059973924,
-      "grad_norm": 1.5674846172332764,
-      "learning_rate": 7.372864321608041e-05,
-      "loss": 5.7654,
-      "step": 26648
-    },
-    {
-      "epoch": 13.897783572359844,
-      "grad_norm": 1.75889253616333,
-      "learning_rate": 7.372763819095477e-05,
-      "loss": 4.7114,
-      "step": 26649
-    },
-    {
-      "epoch": 13.898305084745763,
-      "grad_norm": 1.3132448196411133,
-      "learning_rate": 7.372663316582915e-05,
-      "loss": 5.6897,
-      "step": 26650
-    },
-    {
-      "epoch": 13.898826597131682,
-      "grad_norm": 1.3682386875152588,
-      "learning_rate": 7.372562814070352e-05,
-      "loss": 5.9535,
-      "step": 26651
-    },
-    {
-      "epoch": 13.8993481095176,
-      "grad_norm": 1.4459233283996582,
-      "learning_rate": 7.37246231155779e-05,
-      "loss": 5.6753,
-      "step": 26652
-    },
-    {
-      "epoch": 13.89986962190352,
-      "grad_norm": 1.4184913635253906,
-      "learning_rate": 7.372361809045226e-05,
-      "loss": 5.2294,
-      "step": 26653
-    },
-    {
-      "epoch": 13.900391134289439,
-      "grad_norm": 1.4373600482940674,
-      "learning_rate": 7.372261306532664e-05,
-      "loss": 5.5242,
-      "step": 26654
-    },
-    {
-      "epoch": 13.900912646675359,
-      "grad_norm": 1.4687817096710205,
-      "learning_rate": 7.3721608040201e-05,
-      "loss": 5.2158,
-      "step": 26655
-    },
-    {
-      "epoch": 13.901434159061278,
-      "grad_norm": 1.3878799676895142,
-      "learning_rate": 7.372060301507538e-05,
-      "loss": 5.6253,
-      "step": 26656
-    },
-    {
-      "epoch": 13.901955671447197,
-      "grad_norm": 1.5440422296524048,
-      "learning_rate": 7.371959798994976e-05,
-      "loss": 5.387,
-      "step": 26657
-    },
-    {
-      "epoch": 13.902477183833117,
-      "grad_norm": 1.5765721797943115,
-      "learning_rate": 7.371859296482413e-05,
-      "loss": 5.3325,
-      "step": 26658
-    },
-    {
-      "epoch": 13.902998696219035,
-      "grad_norm": 1.5343388319015503,
-      "learning_rate": 7.37175879396985e-05,
-      "loss": 4.8661,
-      "step": 26659
-    },
-    {
-      "epoch": 13.903520208604954,
-      "grad_norm": 1.5436588525772095,
-      "learning_rate": 7.371658291457286e-05,
-      "loss": 5.3694,
-      "step": 26660
-    },
-    {
-      "epoch": 13.904041720990874,
-      "grad_norm": 1.7210440635681152,
-      "learning_rate": 7.371557788944724e-05,
-      "loss": 5.0818,
-      "step": 26661
-    },
-    {
-      "epoch": 13.904563233376793,
-      "grad_norm": 1.5256341695785522,
-      "learning_rate": 7.37145728643216e-05,
-      "loss": 5.2195,
-      "step": 26662
-    },
-    {
-      "epoch": 13.905084745762712,
-      "grad_norm": 1.4127004146575928,
-      "learning_rate": 7.371356783919598e-05,
-      "loss": 5.7659,
-      "step": 26663
-    },
-    {
-      "epoch": 13.90560625814863,
-      "grad_norm": 1.4423679113388062,
-      "learning_rate": 7.371256281407035e-05,
-      "loss": 5.3089,
-      "step": 26664
-    },
-    {
-      "epoch": 13.90612777053455,
-      "grad_norm": 1.508985996246338,
-      "learning_rate": 7.371155778894473e-05,
-      "loss": 5.4576,
-      "step": 26665
-    },
-    {
-      "epoch": 13.906649282920469,
-      "grad_norm": 1.4204457998275757,
-      "learning_rate": 7.371055276381909e-05,
-      "loss": 5.3685,
-      "step": 26666
-    },
-    {
-      "epoch": 13.907170795306389,
-      "grad_norm": 1.494886875152588,
-      "learning_rate": 7.370954773869347e-05,
-      "loss": 5.2995,
-      "step": 26667
-    },
-    {
-      "epoch": 13.907692307692308,
-      "grad_norm": 1.42082679271698,
-      "learning_rate": 7.370854271356785e-05,
-      "loss": 5.3842,
-      "step": 26668
-    },
-    {
-      "epoch": 13.908213820078227,
-      "grad_norm": 1.4341078996658325,
-      "learning_rate": 7.370753768844222e-05,
-      "loss": 5.5901,
-      "step": 26669
-    },
-    {
-      "epoch": 13.908735332464147,
-      "grad_norm": 1.5035512447357178,
-      "learning_rate": 7.370653266331659e-05,
-      "loss": 5.7014,
-      "step": 26670
-    },
-    {
-      "epoch": 13.909256844850065,
-      "grad_norm": 1.4467698335647583,
-      "learning_rate": 7.370552763819097e-05,
-      "loss": 5.3172,
-      "step": 26671
-    },
-    {
-      "epoch": 13.909778357235984,
-      "grad_norm": 1.6182844638824463,
-      "learning_rate": 7.370452261306533e-05,
-      "loss": 5.3316,
-      "step": 26672
-    },
-    {
-      "epoch": 13.910299869621904,
-      "grad_norm": 1.3526809215545654,
-      "learning_rate": 7.37035175879397e-05,
-      "loss": 5.2553,
-      "step": 26673
-    },
-    {
-      "epoch": 13.910821382007823,
-      "grad_norm": 1.5579122304916382,
-      "learning_rate": 7.370251256281407e-05,
-      "loss": 4.9359,
-      "step": 26674
-    },
-    {
-      "epoch": 13.911342894393742,
-      "grad_norm": 1.5064327716827393,
-      "learning_rate": 7.370150753768844e-05,
-      "loss": 4.5857,
-      "step": 26675
-    },
-    {
-      "epoch": 13.91186440677966,
-      "grad_norm": 1.452307939529419,
-      "learning_rate": 7.370050251256281e-05,
-      "loss": 5.5639,
-      "step": 26676
-    },
-    {
-      "epoch": 13.91238591916558,
-      "grad_norm": 1.479504942893982,
-      "learning_rate": 7.369949748743719e-05,
-      "loss": 5.7205,
-      "step": 26677
-    },
-    {
-      "epoch": 13.9129074315515,
-      "grad_norm": 1.3828312158584595,
-      "learning_rate": 7.369849246231157e-05,
-      "loss": 5.8899,
-      "step": 26678
-    },
-    {
-      "epoch": 13.913428943937419,
-      "grad_norm": 1.4434571266174316,
-      "learning_rate": 7.369748743718593e-05,
-      "loss": 5.5319,
-      "step": 26679
-    },
-    {
-      "epoch": 13.913950456323338,
-      "grad_norm": 1.5244044065475464,
-      "learning_rate": 7.369648241206031e-05,
-      "loss": 5.0372,
-      "step": 26680
-    },
-    {
-      "epoch": 13.914471968709258,
-      "grad_norm": 1.73919677734375,
-      "learning_rate": 7.369547738693468e-05,
-      "loss": 4.6015,
-      "step": 26681
-    },
-    {
-      "epoch": 13.914993481095175,
-      "grad_norm": 1.4555275440216064,
-      "learning_rate": 7.369447236180905e-05,
-      "loss": 5.7343,
-      "step": 26682
-    },
-    {
-      "epoch": 13.915514993481095,
-      "grad_norm": 1.5297082662582397,
-      "learning_rate": 7.369346733668342e-05,
-      "loss": 5.2371,
-      "step": 26683
-    },
-    {
-      "epoch": 13.916036505867014,
-      "grad_norm": 1.4727615118026733,
-      "learning_rate": 7.36924623115578e-05,
-      "loss": 5.3155,
-      "step": 26684
-    },
-    {
-      "epoch": 13.916558018252934,
-      "grad_norm": 1.4350155591964722,
-      "learning_rate": 7.369145728643216e-05,
-      "loss": 5.2208,
-      "step": 26685
-    },
-    {
-      "epoch": 13.917079530638853,
-      "grad_norm": 1.4707037210464478,
-      "learning_rate": 7.369045226130652e-05,
-      "loss": 5.6404,
-      "step": 26686
-    },
-    {
-      "epoch": 13.917601043024773,
-      "grad_norm": 1.3821433782577515,
-      "learning_rate": 7.36894472361809e-05,
-      "loss": 5.3282,
-      "step": 26687
-    },
-    {
-      "epoch": 13.91812255541069,
-      "grad_norm": 1.4766658544540405,
-      "learning_rate": 7.368844221105528e-05,
-      "loss": 5.226,
-      "step": 26688
-    },
-    {
-      "epoch": 13.91864406779661,
-      "grad_norm": 1.45279061794281,
-      "learning_rate": 7.368743718592966e-05,
-      "loss": 5.5047,
-      "step": 26689
-    },
-    {
-      "epoch": 13.91916558018253,
-      "grad_norm": 1.6684800386428833,
-      "learning_rate": 7.368643216080402e-05,
-      "loss": 4.6905,
-      "step": 26690
-    },
-    {
-      "epoch": 13.919687092568449,
-      "grad_norm": 1.4954174757003784,
-      "learning_rate": 7.36854271356784e-05,
-      "loss": 5.4543,
-      "step": 26691
-    },
-    {
-      "epoch": 13.920208604954368,
-      "grad_norm": 1.4659150838851929,
-      "learning_rate": 7.368442211055276e-05,
-      "loss": 5.4317,
-      "step": 26692
-    },
-    {
-      "epoch": 13.920730117340288,
-      "grad_norm": 1.4975770711898804,
-      "learning_rate": 7.368341708542714e-05,
-      "loss": 4.7188,
-      "step": 26693
-    },
-    {
-      "epoch": 13.921251629726205,
-      "grad_norm": 1.443884253501892,
-      "learning_rate": 7.36824120603015e-05,
-      "loss": 5.264,
-      "step": 26694
-    },
-    {
-      "epoch": 13.921773142112125,
-      "grad_norm": 1.4936622381210327,
-      "learning_rate": 7.368140703517588e-05,
-      "loss": 4.5737,
-      "step": 26695
-    },
-    {
-      "epoch": 13.922294654498044,
-      "grad_norm": 1.4416905641555786,
-      "learning_rate": 7.368040201005025e-05,
-      "loss": 5.4038,
-      "step": 26696
-    },
-    {
-      "epoch": 13.922816166883964,
-      "grad_norm": 1.4603726863861084,
-      "learning_rate": 7.367939698492463e-05,
-      "loss": 5.3269,
-      "step": 26697
-    },
-    {
-      "epoch": 13.923337679269883,
-      "grad_norm": 1.415736436843872,
-      "learning_rate": 7.3678391959799e-05,
-      "loss": 5.6547,
-      "step": 26698
-    },
-    {
-      "epoch": 13.923859191655803,
-      "grad_norm": 1.369124174118042,
-      "learning_rate": 7.367738693467338e-05,
-      "loss": 5.216,
-      "step": 26699
-    },
-    {
-      "epoch": 13.92438070404172,
-      "grad_norm": 1.4672560691833496,
-      "learning_rate": 7.367638190954775e-05,
-      "loss": 5.2612,
-      "step": 26700
-    },
-    {
-      "epoch": 13.92490221642764,
-      "grad_norm": 1.4825761318206787,
-      "learning_rate": 7.367537688442211e-05,
-      "loss": 5.6364,
-      "step": 26701
-    },
-    {
-      "epoch": 13.92542372881356,
-      "grad_norm": 1.5184375047683716,
-      "learning_rate": 7.367437185929649e-05,
-      "loss": 5.167,
-      "step": 26702
-    },
-    {
-      "epoch": 13.925945241199479,
-      "grad_norm": 1.2806264162063599,
-      "learning_rate": 7.367336683417085e-05,
-      "loss": 5.7637,
-      "step": 26703
-    },
-    {
-      "epoch": 13.926466753585398,
-      "grad_norm": 1.5501829385757446,
-      "learning_rate": 7.367236180904523e-05,
-      "loss": 5.1534,
-      "step": 26704
-    },
-    {
-      "epoch": 13.926988265971318,
-      "grad_norm": 1.418771505355835,
-      "learning_rate": 7.36713567839196e-05,
-      "loss": 5.4883,
-      "step": 26705
-    },
-    {
-      "epoch": 13.927509778357235,
-      "grad_norm": 1.4486854076385498,
-      "learning_rate": 7.367035175879397e-05,
-      "loss": 5.3028,
-      "step": 26706
-    },
-    {
-      "epoch": 13.928031290743155,
-      "grad_norm": 1.523600697517395,
-      "learning_rate": 7.366934673366834e-05,
-      "loss": 5.1216,
-      "step": 26707
-    },
-    {
-      "epoch": 13.928552803129074,
-      "grad_norm": 1.5072319507598877,
-      "learning_rate": 7.366834170854271e-05,
-      "loss": 5.0388,
-      "step": 26708
-    },
-    {
-      "epoch": 13.929074315514994,
-      "grad_norm": 1.4231427907943726,
-      "learning_rate": 7.366733668341709e-05,
-      "loss": 5.6187,
-      "step": 26709
-    },
-    {
-      "epoch": 13.929595827900913,
-      "grad_norm": 1.4185776710510254,
-      "learning_rate": 7.366633165829147e-05,
-      "loss": 5.6172,
-      "step": 26710
-    },
-    {
-      "epoch": 13.930117340286833,
-      "grad_norm": 1.4876748323440552,
-      "learning_rate": 7.366532663316583e-05,
-      "loss": 5.4194,
-      "step": 26711
-    },
-    {
-      "epoch": 13.93063885267275,
-      "grad_norm": 1.5744749307632446,
-      "learning_rate": 7.366432160804021e-05,
-      "loss": 5.3129,
-      "step": 26712
-    },
-    {
-      "epoch": 13.93116036505867,
-      "grad_norm": 1.3311198949813843,
-      "learning_rate": 7.366331658291458e-05,
-      "loss": 5.8718,
-      "step": 26713
-    },
-    {
-      "epoch": 13.93168187744459,
-      "grad_norm": 1.3622770309448242,
-      "learning_rate": 7.366231155778894e-05,
-      "loss": 5.6396,
-      "step": 26714
-    },
-    {
-      "epoch": 13.932203389830509,
-      "grad_norm": 1.4100186824798584,
-      "learning_rate": 7.366130653266332e-05,
-      "loss": 5.6709,
-      "step": 26715
-    },
-    {
-      "epoch": 13.932724902216428,
-      "grad_norm": 1.4706860780715942,
-      "learning_rate": 7.366030150753768e-05,
-      "loss": 5.4392,
-      "step": 26716
-    },
-    {
-      "epoch": 13.933246414602348,
-      "grad_norm": 1.4863117933273315,
-      "learning_rate": 7.365929648241206e-05,
-      "loss": 5.2469,
-      "step": 26717
-    },
-    {
-      "epoch": 13.933767926988265,
-      "grad_norm": 1.4887430667877197,
-      "learning_rate": 7.365829145728644e-05,
-      "loss": 5.3699,
-      "step": 26718
-    },
-    {
-      "epoch": 13.934289439374185,
-      "grad_norm": 1.4002934694290161,
-      "learning_rate": 7.365728643216082e-05,
-      "loss": 5.4467,
-      "step": 26719
-    },
-    {
-      "epoch": 13.934810951760104,
-      "grad_norm": 1.3688433170318604,
-      "learning_rate": 7.365628140703518e-05,
-      "loss": 5.6979,
-      "step": 26720
-    },
-    {
-      "epoch": 13.935332464146024,
-      "grad_norm": 1.6244536638259888,
-      "learning_rate": 7.365527638190956e-05,
-      "loss": 5.6054,
-      "step": 26721
-    },
-    {
-      "epoch": 13.935853976531943,
-      "grad_norm": 1.5005621910095215,
-      "learning_rate": 7.365427135678392e-05,
-      "loss": 5.5437,
-      "step": 26722
-    },
-    {
-      "epoch": 13.936375488917863,
-      "grad_norm": 1.4373232126235962,
-      "learning_rate": 7.36532663316583e-05,
-      "loss": 5.4355,
-      "step": 26723
-    },
-    {
-      "epoch": 13.93689700130378,
-      "grad_norm": 1.4109933376312256,
-      "learning_rate": 7.365226130653266e-05,
-      "loss": 5.6648,
-      "step": 26724
-    },
-    {
-      "epoch": 13.9374185136897,
-      "grad_norm": 1.5422117710113525,
-      "learning_rate": 7.365125628140704e-05,
-      "loss": 5.7122,
-      "step": 26725
-    },
-    {
-      "epoch": 13.93794002607562,
-      "grad_norm": 1.554109811782837,
-      "learning_rate": 7.365025125628141e-05,
-      "loss": 5.3159,
-      "step": 26726
-    },
-    {
-      "epoch": 13.938461538461539,
-      "grad_norm": 1.569801688194275,
-      "learning_rate": 7.364924623115577e-05,
-      "loss": 5.3349,
-      "step": 26727
-    },
-    {
-      "epoch": 13.938983050847458,
-      "grad_norm": 1.417837381362915,
-      "learning_rate": 7.364824120603015e-05,
-      "loss": 5.4791,
-      "step": 26728
-    },
-    {
-      "epoch": 13.939504563233378,
-      "grad_norm": 1.2948740720748901,
-      "learning_rate": 7.364723618090453e-05,
-      "loss": 6.0277,
-      "step": 26729
-    },
-    {
-      "epoch": 13.940026075619295,
-      "grad_norm": 1.576796054840088,
-      "learning_rate": 7.36462311557789e-05,
-      "loss": 5.3382,
-      "step": 26730
-    },
-    {
-      "epoch": 13.940547588005215,
-      "grad_norm": 1.473318338394165,
-      "learning_rate": 7.364522613065327e-05,
-      "loss": 5.6456,
-      "step": 26731
-    },
-    {
-      "epoch": 13.941069100391134,
-      "grad_norm": 1.545254111289978,
-      "learning_rate": 7.364422110552765e-05,
-      "loss": 5.399,
-      "step": 26732
-    },
-    {
-      "epoch": 13.941590612777054,
-      "grad_norm": 1.5057916641235352,
-      "learning_rate": 7.364321608040201e-05,
-      "loss": 5.2219,
-      "step": 26733
-    },
-    {
-      "epoch": 13.942112125162973,
-      "grad_norm": 1.4794948101043701,
-      "learning_rate": 7.364221105527639e-05,
-      "loss": 5.6555,
-      "step": 26734
-    },
-    {
-      "epoch": 13.94263363754889,
-      "grad_norm": 1.4810484647750854,
-      "learning_rate": 7.364120603015075e-05,
-      "loss": 5.6945,
-      "step": 26735
-    },
-    {
-      "epoch": 13.94315514993481,
-      "grad_norm": 1.4121111631393433,
-      "learning_rate": 7.364020100502513e-05,
-      "loss": 5.1214,
-      "step": 26736
-    },
-    {
-      "epoch": 13.94367666232073,
-      "grad_norm": 1.3786876201629639,
-      "learning_rate": 7.36391959798995e-05,
-      "loss": 5.5324,
-      "step": 26737
-    },
-    {
-      "epoch": 13.94419817470665,
-      "grad_norm": 1.443158507347107,
-      "learning_rate": 7.363819095477387e-05,
-      "loss": 4.6538,
-      "step": 26738
-    },
-    {
-      "epoch": 13.944719687092569,
-      "grad_norm": 1.5720232725143433,
-      "learning_rate": 7.363718592964825e-05,
-      "loss": 5.1649,
-      "step": 26739
-    },
-    {
-      "epoch": 13.945241199478488,
-      "grad_norm": 1.4947254657745361,
-      "learning_rate": 7.363618090452262e-05,
-      "loss": 5.5885,
-      "step": 26740
-    },
-    {
-      "epoch": 13.945762711864408,
-      "grad_norm": 1.5446851253509521,
-      "learning_rate": 7.363517587939699e-05,
-      "loss": 5.1162,
-      "step": 26741
-    },
-    {
-      "epoch": 13.946284224250325,
-      "grad_norm": 1.4884545803070068,
-      "learning_rate": 7.363417085427136e-05,
-      "loss": 5.0191,
-      "step": 26742
-    },
-    {
-      "epoch": 13.946805736636245,
-      "grad_norm": 1.4477216005325317,
-      "learning_rate": 7.363316582914574e-05,
-      "loss": 5.0399,
-      "step": 26743
-    },
-    {
-      "epoch": 13.947327249022164,
-      "grad_norm": 1.45907461643219,
-      "learning_rate": 7.36321608040201e-05,
-      "loss": 5.3398,
-      "step": 26744
-    },
-    {
-      "epoch": 13.947848761408084,
-      "grad_norm": 1.5139464139938354,
-      "learning_rate": 7.363115577889448e-05,
-      "loss": 5.6116,
-      "step": 26745
-    },
-    {
-      "epoch": 13.948370273794003,
-      "grad_norm": 1.4166967868804932,
-      "learning_rate": 7.363015075376884e-05,
-      "loss": 5.7158,
-      "step": 26746
-    },
-    {
-      "epoch": 13.94889178617992,
-      "grad_norm": 1.455305814743042,
-      "learning_rate": 7.362914572864322e-05,
-      "loss": 5.5612,
-      "step": 26747
-    },
-    {
-      "epoch": 13.94941329856584,
-      "grad_norm": 1.5572513341903687,
-      "learning_rate": 7.362814070351758e-05,
-      "loss": 5.2216,
-      "step": 26748
-    },
-    {
-      "epoch": 13.94993481095176,
-      "grad_norm": 1.5442153215408325,
-      "learning_rate": 7.362713567839196e-05,
-      "loss": 5.6762,
-      "step": 26749
-    },
-    {
-      "epoch": 13.95045632333768,
-      "grad_norm": 1.4083895683288574,
-      "learning_rate": 7.362613065326634e-05,
-      "loss": 5.4785,
-      "step": 26750
-    },
-    {
-      "epoch": 13.950977835723599,
-      "grad_norm": 1.4755198955535889,
-      "learning_rate": 7.362512562814072e-05,
-      "loss": 5.0154,
-      "step": 26751
-    },
-    {
-      "epoch": 13.951499348109518,
-      "grad_norm": 1.6276192665100098,
-      "learning_rate": 7.362412060301508e-05,
-      "loss": 4.7048,
-      "step": 26752
-    },
-    {
-      "epoch": 13.952020860495438,
-      "grad_norm": 1.4662811756134033,
-      "learning_rate": 7.362311557788945e-05,
-      "loss": 5.126,
-      "step": 26753
-    },
-    {
-      "epoch": 13.952542372881355,
-      "grad_norm": 1.5883311033248901,
-      "learning_rate": 7.362211055276382e-05,
-      "loss": 5.1382,
-      "step": 26754
-    },
-    {
-      "epoch": 13.953063885267275,
-      "grad_norm": 1.3924087285995483,
-      "learning_rate": 7.362110552763819e-05,
-      "loss": 5.0355,
-      "step": 26755
-    },
-    {
-      "epoch": 13.953585397653194,
-      "grad_norm": 1.526879072189331,
-      "learning_rate": 7.362010050251257e-05,
-      "loss": 4.9037,
-      "step": 26756
-    },
-    {
-      "epoch": 13.954106910039114,
-      "grad_norm": 1.4299628734588623,
-      "learning_rate": 7.361909547738693e-05,
-      "loss": 5.533,
-      "step": 26757
-    },
-    {
-      "epoch": 13.954628422425033,
-      "grad_norm": 1.489676594734192,
-      "learning_rate": 7.361809045226131e-05,
-      "loss": 4.9036,
-      "step": 26758
-    },
-    {
-      "epoch": 13.955149934810951,
-      "grad_norm": 1.375650405883789,
-      "learning_rate": 7.361708542713569e-05,
-      "loss": 5.2429,
-      "step": 26759
-    },
-    {
-      "epoch": 13.95567144719687,
-      "grad_norm": 1.4780923128128052,
-      "learning_rate": 7.361608040201006e-05,
-      "loss": 5.6222,
-      "step": 26760
-    },
-    {
-      "epoch": 13.95619295958279,
-      "grad_norm": 1.4481362104415894,
-      "learning_rate": 7.361507537688443e-05,
-      "loss": 5.3446,
-      "step": 26761
-    },
-    {
-      "epoch": 13.95671447196871,
-      "grad_norm": 1.4543001651763916,
-      "learning_rate": 7.36140703517588e-05,
-      "loss": 5.8181,
-      "step": 26762
-    },
-    {
-      "epoch": 13.957235984354629,
-      "grad_norm": 1.606495976448059,
-      "learning_rate": 7.361306532663317e-05,
-      "loss": 5.1825,
-      "step": 26763
-    },
-    {
-      "epoch": 13.957757496740548,
-      "grad_norm": 1.4093849658966064,
-      "learning_rate": 7.361206030150755e-05,
-      "loss": 5.5873,
-      "step": 26764
-    },
-    {
-      "epoch": 13.958279009126468,
-      "grad_norm": 1.470317006111145,
-      "learning_rate": 7.361105527638191e-05,
-      "loss": 5.5007,
-      "step": 26765
-    },
-    {
-      "epoch": 13.958800521512385,
-      "grad_norm": 1.532192349433899,
-      "learning_rate": 7.361005025125628e-05,
-      "loss": 5.4882,
-      "step": 26766
-    },
-    {
-      "epoch": 13.959322033898305,
-      "grad_norm": 1.3733165264129639,
-      "learning_rate": 7.360904522613065e-05,
-      "loss": 5.7388,
-      "step": 26767
-    },
-    {
-      "epoch": 13.959843546284224,
-      "grad_norm": 1.5592191219329834,
-      "learning_rate": 7.360804020100502e-05,
-      "loss": 5.2777,
-      "step": 26768
-    },
-    {
-      "epoch": 13.960365058670144,
-      "grad_norm": 1.450516700744629,
-      "learning_rate": 7.36070351758794e-05,
-      "loss": 5.4359,
-      "step": 26769
-    },
-    {
-      "epoch": 13.960886571056063,
-      "grad_norm": 1.3780673742294312,
-      "learning_rate": 7.360603015075377e-05,
-      "loss": 5.6496,
-      "step": 26770
-    },
-    {
-      "epoch": 13.961408083441981,
-      "grad_norm": 1.3412578105926514,
-      "learning_rate": 7.360502512562815e-05,
-      "loss": 5.6529,
-      "step": 26771
-    },
-    {
-      "epoch": 13.9619295958279,
-      "grad_norm": 1.409651517868042,
-      "learning_rate": 7.360402010050252e-05,
-      "loss": 5.1035,
-      "step": 26772
-    },
-    {
-      "epoch": 13.96245110821382,
-      "grad_norm": 1.4353704452514648,
-      "learning_rate": 7.36030150753769e-05,
-      "loss": 5.5726,
-      "step": 26773
-    },
-    {
-      "epoch": 13.96297262059974,
-      "grad_norm": 1.4615349769592285,
-      "learning_rate": 7.360201005025126e-05,
-      "loss": 5.3564,
-      "step": 26774
-    },
-    {
-      "epoch": 13.963494132985659,
-      "grad_norm": 1.4256513118743896,
-      "learning_rate": 7.360100502512564e-05,
-      "loss": 5.4341,
-      "step": 26775
-    },
-    {
-      "epoch": 13.964015645371578,
-      "grad_norm": 1.4297988414764404,
-      "learning_rate": 7.36e-05,
-      "loss": 5.7102,
-      "step": 26776
-    },
-    {
-      "epoch": 13.964537157757496,
-      "grad_norm": 1.3601183891296387,
-      "learning_rate": 7.359899497487438e-05,
-      "loss": 5.6929,
-      "step": 26777
-    },
-    {
-      "epoch": 13.965058670143415,
-      "grad_norm": 1.4266067743301392,
-      "learning_rate": 7.359798994974874e-05,
-      "loss": 5.4361,
-      "step": 26778
-    },
-    {
-      "epoch": 13.965580182529335,
-      "grad_norm": 1.4265316724777222,
-      "learning_rate": 7.359698492462312e-05,
-      "loss": 5.0847,
-      "step": 26779
-    },
-    {
-      "epoch": 13.966101694915254,
-      "grad_norm": 1.4128801822662354,
-      "learning_rate": 7.35959798994975e-05,
-      "loss": 4.9898,
-      "step": 26780
-    },
-    {
-      "epoch": 13.966623207301174,
-      "grad_norm": 1.4600292444229126,
-      "learning_rate": 7.359497487437186e-05,
-      "loss": 5.1557,
-      "step": 26781
-    },
-    {
-      "epoch": 13.967144719687093,
-      "grad_norm": 1.4585713148117065,
-      "learning_rate": 7.359396984924624e-05,
-      "loss": 5.4017,
-      "step": 26782
-    },
-    {
-      "epoch": 13.967666232073011,
-      "grad_norm": 1.49857497215271,
-      "learning_rate": 7.35929648241206e-05,
-      "loss": 5.6955,
-      "step": 26783
-    },
-    {
-      "epoch": 13.96818774445893,
-      "grad_norm": 1.5361857414245605,
-      "learning_rate": 7.359195979899498e-05,
-      "loss": 5.3786,
-      "step": 26784
-    },
-    {
-      "epoch": 13.96870925684485,
-      "grad_norm": 1.3994060754776,
-      "learning_rate": 7.359095477386935e-05,
-      "loss": 5.5702,
-      "step": 26785
-    },
-    {
-      "epoch": 13.96923076923077,
-      "grad_norm": 1.4731388092041016,
-      "learning_rate": 7.358994974874372e-05,
-      "loss": 4.9833,
-      "step": 26786
-    },
-    {
-      "epoch": 13.969752281616689,
-      "grad_norm": 1.425644874572754,
-      "learning_rate": 7.358894472361809e-05,
-      "loss": 5.5767,
-      "step": 26787
-    },
-    {
-      "epoch": 13.970273794002608,
-      "grad_norm": 1.3357279300689697,
-      "learning_rate": 7.358793969849247e-05,
-      "loss": 5.3458,
-      "step": 26788
-    },
-    {
-      "epoch": 13.970795306388526,
-      "grad_norm": 1.2940394878387451,
-      "learning_rate": 7.358693467336683e-05,
-      "loss": 5.8294,
-      "step": 26789
-    },
-    {
-      "epoch": 13.971316818774445,
-      "grad_norm": 1.3942792415618896,
-      "learning_rate": 7.358592964824121e-05,
-      "loss": 5.4873,
-      "step": 26790
-    },
-    {
-      "epoch": 13.971838331160365,
-      "grad_norm": 1.3350924253463745,
-      "learning_rate": 7.358492462311559e-05,
-      "loss": 5.4439,
-      "step": 26791
-    },
-    {
-      "epoch": 13.972359843546284,
-      "grad_norm": 1.5189212560653687,
-      "learning_rate": 7.358391959798996e-05,
-      "loss": 5.3676,
-      "step": 26792
-    },
-    {
-      "epoch": 13.972881355932204,
-      "grad_norm": 1.4011837244033813,
-      "learning_rate": 7.358291457286433e-05,
-      "loss": 5.6413,
-      "step": 26793
-    },
-    {
-      "epoch": 13.973402868318123,
-      "grad_norm": 1.5772035121917725,
-      "learning_rate": 7.358190954773869e-05,
-      "loss": 5.3732,
-      "step": 26794
-    },
-    {
-      "epoch": 13.973924380704041,
-      "grad_norm": 1.5933120250701904,
-      "learning_rate": 7.358090452261307e-05,
-      "loss": 5.5655,
-      "step": 26795
-    },
-    {
-      "epoch": 13.97444589308996,
-      "grad_norm": 1.6761027574539185,
-      "learning_rate": 7.357989949748743e-05,
-      "loss": 4.6845,
-      "step": 26796
-    },
-    {
-      "epoch": 13.97496740547588,
-      "grad_norm": 1.4810163974761963,
-      "learning_rate": 7.357889447236181e-05,
-      "loss": 5.5088,
-      "step": 26797
-    },
-    {
-      "epoch": 13.9754889178618,
-      "grad_norm": 1.4113966226577759,
-      "learning_rate": 7.357788944723618e-05,
-      "loss": 5.7016,
-      "step": 26798
-    },
-    {
-      "epoch": 13.976010430247719,
-      "grad_norm": 1.4204628467559814,
-      "learning_rate": 7.357688442211055e-05,
-      "loss": 5.3871,
-      "step": 26799
-    },
-    {
-      "epoch": 13.976531942633638,
-      "grad_norm": 1.398368000984192,
-      "learning_rate": 7.357587939698493e-05,
-      "loss": 5.1118,
-      "step": 26800
-    },
-    {
-      "epoch": 13.977053455019556,
-      "grad_norm": 1.4315775632858276,
-      "learning_rate": 7.357487437185931e-05,
-      "loss": 5.1235,
-      "step": 26801
-    },
-    {
-      "epoch": 13.977574967405475,
-      "grad_norm": 1.500160813331604,
-      "learning_rate": 7.357386934673367e-05,
-      "loss": 4.802,
-      "step": 26802
-    },
-    {
-      "epoch": 13.978096479791395,
-      "grad_norm": 1.5232983827590942,
-      "learning_rate": 7.357286432160805e-05,
-      "loss": 5.4479,
-      "step": 26803
-    },
-    {
-      "epoch": 13.978617992177314,
-      "grad_norm": 1.513954520225525,
-      "learning_rate": 7.357185929648242e-05,
-      "loss": 5.6248,
-      "step": 26804
-    },
-    {
-      "epoch": 13.979139504563234,
-      "grad_norm": 1.526707649230957,
-      "learning_rate": 7.35708542713568e-05,
-      "loss": 5.3217,
-      "step": 26805
-    },
-    {
-      "epoch": 13.979661016949153,
-      "grad_norm": 1.5071063041687012,
-      "learning_rate": 7.356984924623116e-05,
-      "loss": 5.1963,
-      "step": 26806
-    },
-    {
-      "epoch": 13.980182529335071,
-      "grad_norm": 1.4635404348373413,
-      "learning_rate": 7.356884422110552e-05,
-      "loss": 5.5328,
-      "step": 26807
-    },
-    {
-      "epoch": 13.98070404172099,
-      "grad_norm": 1.4978973865509033,
-      "learning_rate": 7.35678391959799e-05,
-      "loss": 5.5122,
-      "step": 26808
-    },
-    {
-      "epoch": 13.98122555410691,
-      "grad_norm": 1.441070318222046,
-      "learning_rate": 7.356683417085427e-05,
-      "loss": 5.2107,
-      "step": 26809
-    },
-    {
-      "epoch": 13.98174706649283,
-      "grad_norm": 1.4981123208999634,
-      "learning_rate": 7.356582914572864e-05,
-      "loss": 5.1212,
-      "step": 26810
-    },
-    {
-      "epoch": 13.982268578878749,
-      "grad_norm": 1.3488099575042725,
-      "learning_rate": 7.356482412060302e-05,
-      "loss": 5.7533,
-      "step": 26811
-    },
-    {
-      "epoch": 13.982790091264668,
-      "grad_norm": 1.738165020942688,
-      "learning_rate": 7.35638190954774e-05,
-      "loss": 4.5023,
-      "step": 26812
-    },
-    {
-      "epoch": 13.983311603650586,
-      "grad_norm": 1.5288413763046265,
-      "learning_rate": 7.356281407035176e-05,
-      "loss": 4.8685,
-      "step": 26813
-    },
-    {
-      "epoch": 13.983833116036505,
-      "grad_norm": 1.494642972946167,
-      "learning_rate": 7.356180904522614e-05,
-      "loss": 5.5253,
-      "step": 26814
-    },
-    {
-      "epoch": 13.984354628422425,
-      "grad_norm": 1.4947139024734497,
-      "learning_rate": 7.35608040201005e-05,
-      "loss": 5.1878,
-      "step": 26815
-    },
-    {
-      "epoch": 13.984876140808344,
-      "grad_norm": 1.3864974975585938,
-      "learning_rate": 7.355979899497488e-05,
-      "loss": 5.6899,
-      "step": 26816
-    },
-    {
-      "epoch": 13.985397653194264,
-      "grad_norm": 1.5101107358932495,
-      "learning_rate": 7.355879396984925e-05,
-      "loss": 4.8469,
-      "step": 26817
-    },
-    {
-      "epoch": 13.985919165580183,
-      "grad_norm": 1.545830488204956,
-      "learning_rate": 7.355778894472362e-05,
-      "loss": 4.8856,
-      "step": 26818
-    },
-    {
-      "epoch": 13.986440677966101,
-      "grad_norm": 1.4211416244506836,
-      "learning_rate": 7.355678391959799e-05,
-      "loss": 5.4441,
-      "step": 26819
-    },
-    {
-      "epoch": 13.98696219035202,
-      "grad_norm": 1.5769248008728027,
-      "learning_rate": 7.355577889447237e-05,
-      "loss": 5.1471,
-      "step": 26820
-    },
-    {
-      "epoch": 13.98748370273794,
-      "grad_norm": 1.4263787269592285,
-      "learning_rate": 7.355477386934674e-05,
-      "loss": 5.8847,
-      "step": 26821
-    },
-    {
-      "epoch": 13.98800521512386,
-      "grad_norm": 1.4535173177719116,
-      "learning_rate": 7.355376884422111e-05,
-      "loss": 5.1169,
-      "step": 26822
-    },
-    {
-      "epoch": 13.988526727509779,
-      "grad_norm": 1.378038763999939,
-      "learning_rate": 7.355276381909549e-05,
-      "loss": 5.845,
-      "step": 26823
-    },
-    {
-      "epoch": 13.989048239895698,
-      "grad_norm": 1.5194714069366455,
-      "learning_rate": 7.355175879396985e-05,
-      "loss": 5.2315,
-      "step": 26824
-    },
-    {
-      "epoch": 13.989569752281616,
-      "grad_norm": 1.5826665163040161,
-      "learning_rate": 7.355075376884423e-05,
-      "loss": 5.3273,
-      "step": 26825
-    },
-    {
-      "epoch": 13.990091264667535,
-      "grad_norm": 1.4472769498825073,
-      "learning_rate": 7.35497487437186e-05,
-      "loss": 5.8343,
-      "step": 26826
-    },
-    {
-      "epoch": 13.990612777053455,
-      "grad_norm": 1.6252646446228027,
-      "learning_rate": 7.354874371859297e-05,
-      "loss": 5.0606,
-      "step": 26827
-    },
-    {
-      "epoch": 13.991134289439374,
-      "grad_norm": 1.4935024976730347,
-      "learning_rate": 7.354773869346734e-05,
-      "loss": 5.8001,
-      "step": 26828
-    },
-    {
-      "epoch": 13.991655801825294,
-      "grad_norm": 1.4202709197998047,
-      "learning_rate": 7.354673366834171e-05,
-      "loss": 5.4945,
-      "step": 26829
-    },
-    {
-      "epoch": 13.992177314211212,
-      "grad_norm": 1.4242334365844727,
-      "learning_rate": 7.354572864321608e-05,
-      "loss": 4.8596,
-      "step": 26830
-    },
-    {
-      "epoch": 13.992698826597131,
-      "grad_norm": 1.4604341983795166,
-      "learning_rate": 7.354472361809046e-05,
-      "loss": 5.1734,
-      "step": 26831
-    },
-    {
-      "epoch": 13.99322033898305,
-      "grad_norm": 1.376772403717041,
-      "learning_rate": 7.354371859296483e-05,
-      "loss": 5.0823,
-      "step": 26832
-    },
-    {
-      "epoch": 13.99374185136897,
-      "grad_norm": 1.376705527305603,
-      "learning_rate": 7.35427135678392e-05,
-      "loss": 5.5001,
-      "step": 26833
-    },
-    {
-      "epoch": 13.99426336375489,
-      "grad_norm": 1.5069876909255981,
-      "learning_rate": 7.354170854271358e-05,
-      "loss": 5.5931,
-      "step": 26834
-    },
-    {
-      "epoch": 13.994784876140809,
-      "grad_norm": 1.488939642906189,
-      "learning_rate": 7.354070351758794e-05,
-      "loss": 4.9725,
-      "step": 26835
-    },
-    {
-      "epoch": 13.995306388526728,
-      "grad_norm": 1.4175881147384644,
-      "learning_rate": 7.353969849246232e-05,
-      "loss": 5.4144,
-      "step": 26836
-    },
-    {
-      "epoch": 13.995827900912646,
-      "grad_norm": 1.463783860206604,
-      "learning_rate": 7.353869346733668e-05,
-      "loss": 5.1871,
-      "step": 26837
-    },
-    {
-      "epoch": 13.996349413298566,
-      "grad_norm": 1.5515037775039673,
-      "learning_rate": 7.353768844221106e-05,
-      "loss": 5.3532,
-      "step": 26838
-    },
-    {
-      "epoch": 13.996870925684485,
-      "grad_norm": 1.4369971752166748,
-      "learning_rate": 7.353668341708542e-05,
-      "loss": 5.2837,
-      "step": 26839
-    },
-    {
-      "epoch": 13.997392438070404,
-      "grad_norm": 1.449455976486206,
-      "learning_rate": 7.35356783919598e-05,
-      "loss": 5.04,
-      "step": 26840
-    },
-    {
-      "epoch": 13.997913950456324,
-      "grad_norm": 1.4751110076904297,
-      "learning_rate": 7.353467336683417e-05,
-      "loss": 4.8468,
-      "step": 26841
-    },
-    {
-      "epoch": 13.998435462842242,
-      "grad_norm": 1.3613299131393433,
-      "learning_rate": 7.353366834170854e-05,
-      "loss": 5.6348,
-      "step": 26842
-    },
-    {
-      "epoch": 13.998956975228161,
-      "grad_norm": 1.4341375827789307,
-      "learning_rate": 7.353266331658292e-05,
-      "loss": 5.4747,
-      "step": 26843
-    },
-    {
-      "epoch": 13.99947848761408,
-      "grad_norm": 1.3866671323776245,
-      "learning_rate": 7.35316582914573e-05,
-      "loss": 5.5033,
-      "step": 26844
-    },
-    {
-      "epoch": 14.0,
-      "grad_norm": 1.6326735019683838,
-      "learning_rate": 7.353065326633166e-05,
-      "loss": 5.1622,
-      "step": 26845
-    },
-    {
-      "epoch": 14.00052151238592,
-      "grad_norm": 1.4589935541152954,
-      "learning_rate": 7.352964824120603e-05,
-      "loss": 4.7231,
-      "step": 26846
-    },
-    {
-      "epoch": 14.001043024771839,
-      "grad_norm": 1.4754462242126465,
-      "learning_rate": 7.35286432160804e-05,
-      "loss": 5.3489,
-      "step": 26847
-    },
-    {
-      "epoch": 14.001564537157757,
-      "grad_norm": 1.3962643146514893,
-      "learning_rate": 7.352763819095477e-05,
-      "loss": 5.5245,
-      "step": 26848
-    },
-    {
-      "epoch": 14.002086049543676,
-      "grad_norm": 1.4796777963638306,
-      "learning_rate": 7.352663316582915e-05,
-      "loss": 5.418,
-      "step": 26849
-    },
-    {
-      "epoch": 14.002607561929596,
-      "grad_norm": 1.4989272356033325,
-      "learning_rate": 7.352562814070351e-05,
-      "loss": 5.4504,
-      "step": 26850
-    },
-    {
-      "epoch": 14.003129074315515,
-      "grad_norm": 1.5473209619522095,
-      "learning_rate": 7.352462311557789e-05,
-      "loss": 5.5937,
-      "step": 26851
-    },
-    {
-      "epoch": 14.003650586701434,
-      "grad_norm": 1.580489993095398,
-      "learning_rate": 7.352361809045227e-05,
-      "loss": 4.5548,
-      "step": 26852
-    },
-    {
-      "epoch": 14.004172099087354,
-      "grad_norm": 1.5011030435562134,
-      "learning_rate": 7.352261306532665e-05,
-      "loss": 5.5232,
-      "step": 26853
-    },
-    {
-      "epoch": 14.004693611473272,
-      "grad_norm": 1.3724726438522339,
-      "learning_rate": 7.352160804020101e-05,
-      "loss": 5.619,
-      "step": 26854
-    },
-    {
-      "epoch": 14.005215123859191,
-      "grad_norm": 1.4519985914230347,
-      "learning_rate": 7.352060301507539e-05,
-      "loss": 5.6839,
-      "step": 26855
-    },
-    {
-      "epoch": 14.00573663624511,
-      "grad_norm": 1.4497004747390747,
-      "learning_rate": 7.351959798994975e-05,
-      "loss": 4.9564,
-      "step": 26856
-    },
-    {
-      "epoch": 14.00625814863103,
-      "grad_norm": 1.7752814292907715,
-      "learning_rate": 7.351859296482413e-05,
-      "loss": 4.8688,
-      "step": 26857
-    },
-    {
-      "epoch": 14.00677966101695,
-      "grad_norm": 1.3969091176986694,
-      "learning_rate": 7.35175879396985e-05,
-      "loss": 5.7232,
-      "step": 26858
-    },
-    {
-      "epoch": 14.007301173402869,
-      "grad_norm": 1.4468841552734375,
-      "learning_rate": 7.351658291457286e-05,
-      "loss": 5.5508,
-      "step": 26859
-    },
-    {
-      "epoch": 14.007822685788787,
-      "grad_norm": 1.3771315813064575,
-      "learning_rate": 7.351557788944724e-05,
-      "loss": 5.4213,
-      "step": 26860
-    },
-    {
-      "epoch": 14.008344198174706,
-      "grad_norm": 1.392442226409912,
-      "learning_rate": 7.35145728643216e-05,
-      "loss": 5.705,
-      "step": 26861
-    },
-    {
-      "epoch": 14.008865710560626,
-      "grad_norm": 1.4163898229599,
-      "learning_rate": 7.351356783919598e-05,
-      "loss": 5.6212,
-      "step": 26862
-    },
-    {
-      "epoch": 14.009387222946545,
-      "grad_norm": 1.455457091331482,
-      "learning_rate": 7.351256281407036e-05,
-      "loss": 5.5424,
-      "step": 26863
-    },
-    {
-      "epoch": 14.009908735332465,
-      "grad_norm": 1.3963004350662231,
-      "learning_rate": 7.351155778894473e-05,
-      "loss": 5.2199,
-      "step": 26864
-    },
-    {
-      "epoch": 14.010430247718384,
-      "grad_norm": 1.3810863494873047,
-      "learning_rate": 7.35105527638191e-05,
-      "loss": 5.539,
-      "step": 26865
-    },
-    {
-      "epoch": 14.010951760104302,
-      "grad_norm": 1.4181419610977173,
-      "learning_rate": 7.350954773869348e-05,
-      "loss": 5.3802,
-      "step": 26866
-    },
-    {
-      "epoch": 14.011473272490221,
-      "grad_norm": 1.458590030670166,
-      "learning_rate": 7.350854271356784e-05,
-      "loss": 5.6133,
-      "step": 26867
-    },
-    {
-      "epoch": 14.01199478487614,
-      "grad_norm": 1.4504090547561646,
-      "learning_rate": 7.350753768844222e-05,
-      "loss": 5.3067,
-      "step": 26868
-    },
-    {
-      "epoch": 14.01251629726206,
-      "grad_norm": 1.3107978105545044,
-      "learning_rate": 7.350653266331658e-05,
-      "loss": 5.945,
-      "step": 26869
-    },
-    {
-      "epoch": 14.01303780964798,
-      "grad_norm": 1.620133638381958,
-      "learning_rate": 7.350552763819096e-05,
-      "loss": 4.925,
-      "step": 26870
-    },
-    {
-      "epoch": 14.013559322033899,
-      "grad_norm": 1.615128993988037,
-      "learning_rate": 7.350452261306532e-05,
-      "loss": 4.9392,
-      "step": 26871
-    },
-    {
-      "epoch": 14.014080834419817,
-      "grad_norm": 1.477499008178711,
-      "learning_rate": 7.35035175879397e-05,
-      "loss": 5.5536,
-      "step": 26872
-    },
-    {
-      "epoch": 14.014602346805736,
-      "grad_norm": 1.7378100156784058,
-      "learning_rate": 7.350251256281408e-05,
-      "loss": 4.9507,
-      "step": 26873
-    },
-    {
-      "epoch": 14.015123859191656,
-      "grad_norm": 1.4452301263809204,
-      "learning_rate": 7.350150753768844e-05,
-      "loss": 5.0245,
-      "step": 26874
-    },
-    {
-      "epoch": 14.015645371577575,
-      "grad_norm": 1.4544755220413208,
-      "learning_rate": 7.350050251256282e-05,
-      "loss": 4.8899,
-      "step": 26875
-    },
-    {
-      "epoch": 14.016166883963495,
-      "grad_norm": 1.617704153060913,
-      "learning_rate": 7.349949748743719e-05,
-      "loss": 5.1211,
-      "step": 26876
-    },
-    {
-      "epoch": 14.016688396349414,
-      "grad_norm": 1.4898455142974854,
-      "learning_rate": 7.349849246231156e-05,
-      "loss": 5.1897,
-      "step": 26877
-    },
-    {
-      "epoch": 14.017209908735332,
-      "grad_norm": 1.4862133264541626,
-      "learning_rate": 7.349748743718593e-05,
-      "loss": 5.4621,
-      "step": 26878
-    },
-    {
-      "epoch": 14.017731421121251,
-      "grad_norm": 1.5044457912445068,
-      "learning_rate": 7.34964824120603e-05,
-      "loss": 5.2371,
-      "step": 26879
-    },
-    {
-      "epoch": 14.01825293350717,
-      "grad_norm": 1.4316030740737915,
-      "learning_rate": 7.349547738693467e-05,
-      "loss": 5.7105,
-      "step": 26880
-    },
-    {
-      "epoch": 14.01877444589309,
-      "grad_norm": 1.6181401014328003,
-      "learning_rate": 7.349447236180905e-05,
-      "loss": 5.4356,
-      "step": 26881
-    },
-    {
-      "epoch": 14.01929595827901,
-      "grad_norm": 1.539232611656189,
-      "learning_rate": 7.349346733668341e-05,
-      "loss": 5.3405,
-      "step": 26882
-    },
-    {
-      "epoch": 14.019817470664929,
-      "grad_norm": 1.4436768293380737,
-      "learning_rate": 7.349246231155779e-05,
-      "loss": 5.5902,
-      "step": 26883
-    },
-    {
-      "epoch": 14.020338983050847,
-      "grad_norm": 1.387526512145996,
-      "learning_rate": 7.349145728643217e-05,
-      "loss": 5.5701,
-      "step": 26884
-    },
-    {
-      "epoch": 14.020860495436766,
-      "grad_norm": 1.3931430578231812,
-      "learning_rate": 7.349045226130655e-05,
-      "loss": 5.6088,
-      "step": 26885
-    },
-    {
-      "epoch": 14.021382007822686,
-      "grad_norm": 1.453688144683838,
-      "learning_rate": 7.348944723618091e-05,
-      "loss": 5.5893,
-      "step": 26886
-    },
-    {
-      "epoch": 14.021903520208605,
-      "grad_norm": 1.371160864830017,
-      "learning_rate": 7.348844221105527e-05,
-      "loss": 4.6221,
-      "step": 26887
-    },
-    {
-      "epoch": 14.022425032594525,
-      "grad_norm": 1.42741858959198,
-      "learning_rate": 7.348743718592965e-05,
-      "loss": 5.5143,
-      "step": 26888
-    },
-    {
-      "epoch": 14.022946544980444,
-      "grad_norm": 1.4866973161697388,
-      "learning_rate": 7.348643216080402e-05,
-      "loss": 5.4048,
-      "step": 26889
-    },
-    {
-      "epoch": 14.023468057366362,
-      "grad_norm": 1.3564447164535522,
-      "learning_rate": 7.34854271356784e-05,
-      "loss": 5.5835,
-      "step": 26890
-    },
-    {
-      "epoch": 14.023989569752281,
-      "grad_norm": 1.4545187950134277,
-      "learning_rate": 7.348442211055276e-05,
-      "loss": 5.4094,
-      "step": 26891
-    },
-    {
-      "epoch": 14.0245110821382,
-      "grad_norm": 1.526732087135315,
-      "learning_rate": 7.348341708542714e-05,
-      "loss": 5.3082,
-      "step": 26892
-    },
-    {
-      "epoch": 14.02503259452412,
-      "grad_norm": 1.4828232526779175,
-      "learning_rate": 7.348241206030151e-05,
-      "loss": 5.2854,
-      "step": 26893
-    },
-    {
-      "epoch": 14.02555410691004,
-      "grad_norm": 1.4959583282470703,
-      "learning_rate": 7.348140703517589e-05,
-      "loss": 5.4689,
-      "step": 26894
-    },
-    {
-      "epoch": 14.026075619295959,
-      "grad_norm": 1.4909706115722656,
-      "learning_rate": 7.348040201005026e-05,
-      "loss": 5.609,
-      "step": 26895
-    },
-    {
-      "epoch": 14.026597131681877,
-      "grad_norm": 1.4228639602661133,
-      "learning_rate": 7.347939698492463e-05,
-      "loss": 5.7586,
-      "step": 26896
-    },
-    {
-      "epoch": 14.027118644067796,
-      "grad_norm": 1.456133246421814,
-      "learning_rate": 7.3478391959799e-05,
-      "loss": 5.5625,
-      "step": 26897
-    },
-    {
-      "epoch": 14.027640156453716,
-      "grad_norm": 1.423514485359192,
-      "learning_rate": 7.347738693467338e-05,
-      "loss": 5.1824,
-      "step": 26898
-    },
-    {
-      "epoch": 14.028161668839635,
-      "grad_norm": 1.41177499294281,
-      "learning_rate": 7.347638190954774e-05,
-      "loss": 5.5657,
-      "step": 26899
-    },
-    {
-      "epoch": 14.028683181225555,
-      "grad_norm": 1.4766080379486084,
-      "learning_rate": 7.34753768844221e-05,
-      "loss": 5.5209,
-      "step": 26900
-    },
-    {
-      "epoch": 14.029204693611474,
-      "grad_norm": 1.455452799797058,
-      "learning_rate": 7.347437185929648e-05,
-      "loss": 4.7298,
-      "step": 26901
-    },
-    {
-      "epoch": 14.029726205997392,
-      "grad_norm": 1.4418785572052002,
-      "learning_rate": 7.347336683417085e-05,
-      "loss": 5.4136,
-      "step": 26902
-    },
-    {
-      "epoch": 14.030247718383311,
-      "grad_norm": 1.3936691284179688,
-      "learning_rate": 7.347236180904523e-05,
-      "loss": 5.2357,
-      "step": 26903
-    },
-    {
-      "epoch": 14.03076923076923,
-      "grad_norm": 1.4701529741287231,
-      "learning_rate": 7.34713567839196e-05,
-      "loss": 5.6636,
-      "step": 26904
-    },
-    {
-      "epoch": 14.03129074315515,
-      "grad_norm": 1.4701812267303467,
-      "learning_rate": 7.347035175879398e-05,
-      "loss": 5.3333,
-      "step": 26905
-    },
-    {
-      "epoch": 14.03181225554107,
-      "grad_norm": 1.4231520891189575,
-      "learning_rate": 7.346934673366835e-05,
-      "loss": 5.7002,
-      "step": 26906
-    },
-    {
-      "epoch": 14.032333767926989,
-      "grad_norm": 1.5873533487319946,
-      "learning_rate": 7.346834170854272e-05,
-      "loss": 4.8229,
-      "step": 26907
-    },
-    {
-      "epoch": 14.032855280312907,
-      "grad_norm": 1.4588545560836792,
-      "learning_rate": 7.346733668341709e-05,
-      "loss": 5.3591,
-      "step": 26908
-    },
-    {
-      "epoch": 14.033376792698826,
-      "grad_norm": 1.4215220212936401,
-      "learning_rate": 7.346633165829147e-05,
-      "loss": 5.8304,
-      "step": 26909
-    },
-    {
-      "epoch": 14.033898305084746,
-      "grad_norm": 1.4498523473739624,
-      "learning_rate": 7.346532663316583e-05,
-      "loss": 5.0686,
-      "step": 26910
-    },
-    {
-      "epoch": 14.034419817470665,
-      "grad_norm": 1.4850677251815796,
-      "learning_rate": 7.346432160804021e-05,
-      "loss": 5.4009,
-      "step": 26911
-    },
-    {
-      "epoch": 14.034941329856585,
-      "grad_norm": 1.529175877571106,
-      "learning_rate": 7.346331658291457e-05,
-      "loss": 5.0259,
-      "step": 26912
-    },
-    {
-      "epoch": 14.035462842242504,
-      "grad_norm": 1.6054964065551758,
-      "learning_rate": 7.346231155778895e-05,
-      "loss": 5.5138,
-      "step": 26913
-    },
-    {
-      "epoch": 14.035984354628422,
-      "grad_norm": 1.5308665037155151,
-      "learning_rate": 7.346130653266333e-05,
-      "loss": 5.4189,
-      "step": 26914
-    },
-    {
-      "epoch": 14.036505867014341,
-      "grad_norm": 1.5901618003845215,
-      "learning_rate": 7.346030150753769e-05,
-      "loss": 5.517,
-      "step": 26915
-    },
-    {
-      "epoch": 14.03702737940026,
-      "grad_norm": 1.518808364868164,
-      "learning_rate": 7.345929648241207e-05,
-      "loss": 5.6443,
-      "step": 26916
-    },
-    {
-      "epoch": 14.03754889178618,
-      "grad_norm": 1.5329805612564087,
-      "learning_rate": 7.345829145728643e-05,
-      "loss": 4.8866,
-      "step": 26917
-    },
-    {
-      "epoch": 14.0380704041721,
-      "grad_norm": 1.5338971614837646,
-      "learning_rate": 7.345728643216081e-05,
-      "loss": 5.6613,
-      "step": 26918
-    },
-    {
-      "epoch": 14.038591916558019,
-      "grad_norm": 1.458968162536621,
-      "learning_rate": 7.345628140703518e-05,
-      "loss": 5.3021,
-      "step": 26919
-    },
-    {
-      "epoch": 14.039113428943937,
-      "grad_norm": 1.3792682886123657,
-      "learning_rate": 7.345527638190955e-05,
-      "loss": 5.7242,
-      "step": 26920
-    },
-    {
-      "epoch": 14.039634941329856,
-      "grad_norm": 1.4627803564071655,
-      "learning_rate": 7.345427135678392e-05,
-      "loss": 5.4116,
-      "step": 26921
-    },
-    {
-      "epoch": 14.040156453715776,
-      "grad_norm": 1.5662236213684082,
-      "learning_rate": 7.34532663316583e-05,
-      "loss": 5.4299,
-      "step": 26922
-    },
-    {
-      "epoch": 14.040677966101695,
-      "grad_norm": 1.3735179901123047,
-      "learning_rate": 7.345226130653266e-05,
-      "loss": 5.0843,
-      "step": 26923
-    },
-    {
-      "epoch": 14.041199478487615,
-      "grad_norm": 1.4943009614944458,
-      "learning_rate": 7.345125628140704e-05,
-      "loss": 5.4719,
-      "step": 26924
-    },
-    {
-      "epoch": 14.041720990873534,
-      "grad_norm": 1.4332152605056763,
-      "learning_rate": 7.345025125628142e-05,
-      "loss": 5.6289,
-      "step": 26925
-    },
-    {
-      "epoch": 14.042242503259452,
-      "grad_norm": 1.4733699560165405,
-      "learning_rate": 7.344924623115578e-05,
-      "loss": 5.3615,
-      "step": 26926
-    },
-    {
-      "epoch": 14.042764015645371,
-      "grad_norm": 1.380283236503601,
-      "learning_rate": 7.344824120603016e-05,
-      "loss": 5.5824,
-      "step": 26927
-    },
-    {
-      "epoch": 14.04328552803129,
-      "grad_norm": 1.629823923110962,
-      "learning_rate": 7.344723618090452e-05,
-      "loss": 5.243,
-      "step": 26928
-    },
-    {
-      "epoch": 14.04380704041721,
-      "grad_norm": 1.5369040966033936,
-      "learning_rate": 7.34462311557789e-05,
-      "loss": 5.4125,
-      "step": 26929
-    },
-    {
-      "epoch": 14.04432855280313,
-      "grad_norm": 1.4611072540283203,
-      "learning_rate": 7.344522613065326e-05,
-      "loss": 5.736,
-      "step": 26930
-    },
-    {
-      "epoch": 14.044850065189049,
-      "grad_norm": 1.578600525856018,
-      "learning_rate": 7.344422110552764e-05,
-      "loss": 4.8902,
-      "step": 26931
-    },
-    {
-      "epoch": 14.045371577574967,
-      "grad_norm": 1.4626238346099854,
-      "learning_rate": 7.3443216080402e-05,
-      "loss": 5.7985,
-      "step": 26932
-    },
-    {
-      "epoch": 14.045893089960886,
-      "grad_norm": 1.4488062858581543,
-      "learning_rate": 7.344221105527638e-05,
-      "loss": 5.4533,
-      "step": 26933
-    },
-    {
-      "epoch": 14.046414602346806,
-      "grad_norm": 1.3640638589859009,
-      "learning_rate": 7.344120603015076e-05,
-      "loss": 5.2329,
-      "step": 26934
-    },
-    {
-      "epoch": 14.046936114732725,
-      "grad_norm": 1.5202834606170654,
-      "learning_rate": 7.344020100502514e-05,
-      "loss": 5.6312,
-      "step": 26935
-    },
-    {
-      "epoch": 14.047457627118645,
-      "grad_norm": 1.466470718383789,
-      "learning_rate": 7.34391959798995e-05,
-      "loss": 5.6888,
-      "step": 26936
-    },
-    {
-      "epoch": 14.047979139504564,
-      "grad_norm": 1.472771167755127,
-      "learning_rate": 7.343819095477388e-05,
-      "loss": 5.5493,
-      "step": 26937
-    },
-    {
-      "epoch": 14.048500651890482,
-      "grad_norm": 1.4768167734146118,
-      "learning_rate": 7.343718592964825e-05,
-      "loss": 5.0305,
-      "step": 26938
-    },
-    {
-      "epoch": 14.049022164276401,
-      "grad_norm": 1.4908044338226318,
-      "learning_rate": 7.343618090452261e-05,
-      "loss": 5.1085,
-      "step": 26939
-    },
-    {
-      "epoch": 14.04954367666232,
-      "grad_norm": 1.4810231924057007,
-      "learning_rate": 7.343517587939699e-05,
-      "loss": 5.6078,
-      "step": 26940
-    },
-    {
-      "epoch": 14.05006518904824,
-      "grad_norm": 1.472760796546936,
-      "learning_rate": 7.343417085427135e-05,
-      "loss": 5.0588,
-      "step": 26941
-    },
-    {
-      "epoch": 14.05058670143416,
-      "grad_norm": 1.5106000900268555,
-      "learning_rate": 7.343316582914573e-05,
-      "loss": 5.3316,
-      "step": 26942
-    },
-    {
-      "epoch": 14.051108213820077,
-      "grad_norm": 1.6306064128875732,
-      "learning_rate": 7.34321608040201e-05,
-      "loss": 4.4783,
-      "step": 26943
-    },
-    {
-      "epoch": 14.051629726205997,
-      "grad_norm": 1.5006804466247559,
-      "learning_rate": 7.343115577889447e-05,
-      "loss": 5.2381,
-      "step": 26944
-    },
-    {
-      "epoch": 14.052151238591916,
-      "grad_norm": 1.566216230392456,
-      "learning_rate": 7.343015075376885e-05,
-      "loss": 5.0441,
-      "step": 26945
-    },
-    {
-      "epoch": 14.052672750977836,
-      "grad_norm": 1.5676963329315186,
-      "learning_rate": 7.342914572864323e-05,
-      "loss": 5.4666,
-      "step": 26946
-    },
-    {
-      "epoch": 14.053194263363755,
-      "grad_norm": 1.4918415546417236,
-      "learning_rate": 7.342814070351759e-05,
-      "loss": 5.9083,
-      "step": 26947
-    },
-    {
-      "epoch": 14.053715775749675,
-      "grad_norm": 1.537062168121338,
-      "learning_rate": 7.342713567839197e-05,
-      "loss": 5.2627,
-      "step": 26948
-    },
-    {
-      "epoch": 14.054237288135592,
-      "grad_norm": 1.420619010925293,
-      "learning_rate": 7.342613065326633e-05,
-      "loss": 5.7095,
-      "step": 26949
-    },
-    {
-      "epoch": 14.054758800521512,
-      "grad_norm": 1.5238409042358398,
-      "learning_rate": 7.342512562814071e-05,
-      "loss": 4.4825,
-      "step": 26950
-    },
-    {
-      "epoch": 14.055280312907431,
-      "grad_norm": 1.4303324222564697,
-      "learning_rate": 7.342412060301508e-05,
-      "loss": 5.4923,
-      "step": 26951
-    },
-    {
-      "epoch": 14.05580182529335,
-      "grad_norm": 1.526898980140686,
-      "learning_rate": 7.342311557788944e-05,
-      "loss": 4.7523,
-      "step": 26952
-    },
-    {
-      "epoch": 14.05632333767927,
-      "grad_norm": 1.4111510515213013,
-      "learning_rate": 7.342211055276382e-05,
-      "loss": 5.5588,
-      "step": 26953
-    },
-    {
-      "epoch": 14.05684485006519,
-      "grad_norm": 1.49845552444458,
-      "learning_rate": 7.34211055276382e-05,
-      "loss": 5.497,
-      "step": 26954
-    },
-    {
-      "epoch": 14.057366362451107,
-      "grad_norm": 1.407240390777588,
-      "learning_rate": 7.342010050251257e-05,
-      "loss": 5.5326,
-      "step": 26955
-    },
-    {
-      "epoch": 14.057887874837027,
-      "grad_norm": 1.5240910053253174,
-      "learning_rate": 7.341909547738694e-05,
-      "loss": 5.3038,
-      "step": 26956
-    },
-    {
-      "epoch": 14.058409387222946,
-      "grad_norm": 1.4933522939682007,
-      "learning_rate": 7.341809045226132e-05,
-      "loss": 5.11,
-      "step": 26957
-    },
-    {
-      "epoch": 14.058930899608866,
-      "grad_norm": 1.6219290494918823,
-      "learning_rate": 7.341708542713568e-05,
-      "loss": 5.0242,
-      "step": 26958
-    },
-    {
-      "epoch": 14.059452411994785,
-      "grad_norm": 1.6733415126800537,
-      "learning_rate": 7.341608040201006e-05,
-      "loss": 4.8423,
-      "step": 26959
-    },
-    {
-      "epoch": 14.059973924380705,
-      "grad_norm": 1.5590277910232544,
-      "learning_rate": 7.341507537688442e-05,
-      "loss": 5.7307,
-      "step": 26960
-    },
-    {
-      "epoch": 14.060495436766622,
-      "grad_norm": 1.3886915445327759,
-      "learning_rate": 7.34140703517588e-05,
-      "loss": 5.0253,
-      "step": 26961
-    },
-    {
-      "epoch": 14.061016949152542,
-      "grad_norm": 1.459175944328308,
-      "learning_rate": 7.341306532663316e-05,
-      "loss": 5.2531,
-      "step": 26962
-    },
-    {
-      "epoch": 14.061538461538461,
-      "grad_norm": 1.6267497539520264,
-      "learning_rate": 7.341206030150754e-05,
-      "loss": 4.8548,
-      "step": 26963
-    },
-    {
-      "epoch": 14.06205997392438,
-      "grad_norm": 1.505282998085022,
-      "learning_rate": 7.341105527638191e-05,
-      "loss": 4.7113,
-      "step": 26964
-    },
-    {
-      "epoch": 14.0625814863103,
-      "grad_norm": 1.5379111766815186,
-      "learning_rate": 7.341005025125628e-05,
-      "loss": 5.4143,
-      "step": 26965
-    },
-    {
-      "epoch": 14.06310299869622,
-      "grad_norm": 1.5097013711929321,
-      "learning_rate": 7.340904522613066e-05,
-      "loss": 5.3186,
-      "step": 26966
-    },
-    {
-      "epoch": 14.063624511082137,
-      "grad_norm": 1.4820483922958374,
-      "learning_rate": 7.340804020100503e-05,
-      "loss": 5.5865,
-      "step": 26967
-    },
-    {
-      "epoch": 14.064146023468057,
-      "grad_norm": 1.5100579261779785,
-      "learning_rate": 7.34070351758794e-05,
-      "loss": 5.347,
-      "step": 26968
-    },
-    {
-      "epoch": 14.064667535853976,
-      "grad_norm": 1.3259782791137695,
-      "learning_rate": 7.340603015075377e-05,
-      "loss": 5.5415,
-      "step": 26969
-    },
-    {
-      "epoch": 14.065189048239896,
-      "grad_norm": 1.3338228464126587,
-      "learning_rate": 7.340502512562815e-05,
-      "loss": 5.5601,
-      "step": 26970
-    },
-    {
-      "epoch": 14.065710560625815,
-      "grad_norm": 1.5008724927902222,
-      "learning_rate": 7.340402010050251e-05,
-      "loss": 5.2101,
-      "step": 26971
-    },
-    {
-      "epoch": 14.066232073011735,
-      "grad_norm": 1.5310075283050537,
-      "learning_rate": 7.340301507537689e-05,
-      "loss": 5.3626,
-      "step": 26972
-    },
-    {
-      "epoch": 14.066753585397652,
-      "grad_norm": 1.4345438480377197,
-      "learning_rate": 7.340201005025125e-05,
-      "loss": 5.1643,
-      "step": 26973
-    },
-    {
-      "epoch": 14.067275097783572,
-      "grad_norm": 1.4885064363479614,
-      "learning_rate": 7.340100502512563e-05,
-      "loss": 5.7009,
-      "step": 26974
-    },
-    {
-      "epoch": 14.067796610169491,
-      "grad_norm": 1.4291306734085083,
-      "learning_rate": 7.340000000000001e-05,
-      "loss": 5.3722,
-      "step": 26975
-    },
-    {
-      "epoch": 14.06831812255541,
-      "grad_norm": 1.4195934534072876,
-      "learning_rate": 7.339899497487439e-05,
-      "loss": 5.5797,
-      "step": 26976
-    },
-    {
-      "epoch": 14.06883963494133,
-      "grad_norm": 1.4768248796463013,
-      "learning_rate": 7.339798994974875e-05,
-      "loss": 4.9403,
-      "step": 26977
-    },
-    {
-      "epoch": 14.06936114732725,
-      "grad_norm": 1.5117374658584595,
-      "learning_rate": 7.339698492462313e-05,
-      "loss": 5.0273,
-      "step": 26978
-    },
-    {
-      "epoch": 14.069882659713167,
-      "grad_norm": 1.624394416809082,
-      "learning_rate": 7.339597989949749e-05,
-      "loss": 5.0392,
-      "step": 26979
-    },
-    {
-      "epoch": 14.070404172099087,
-      "grad_norm": 1.6045887470245361,
-      "learning_rate": 7.339497487437186e-05,
-      "loss": 4.7568,
-      "step": 26980
-    },
-    {
-      "epoch": 14.070925684485006,
-      "grad_norm": 1.4850027561187744,
-      "learning_rate": 7.339396984924624e-05,
-      "loss": 5.3149,
-      "step": 26981
-    },
-    {
-      "epoch": 14.071447196870926,
-      "grad_norm": 1.445961356163025,
-      "learning_rate": 7.33929648241206e-05,
-      "loss": 5.2994,
-      "step": 26982
-    },
-    {
-      "epoch": 14.071968709256845,
-      "grad_norm": 1.5436793565750122,
-      "learning_rate": 7.339195979899498e-05,
-      "loss": 5.6356,
-      "step": 26983
-    },
-    {
-      "epoch": 14.072490221642765,
-      "grad_norm": 1.358197808265686,
-      "learning_rate": 7.339095477386934e-05,
-      "loss": 4.7294,
-      "step": 26984
-    },
-    {
-      "epoch": 14.073011734028682,
-      "grad_norm": 1.4321110248565674,
-      "learning_rate": 7.338994974874372e-05,
-      "loss": 5.1181,
-      "step": 26985
-    },
-    {
-      "epoch": 14.073533246414602,
-      "grad_norm": 1.4105048179626465,
-      "learning_rate": 7.33889447236181e-05,
-      "loss": 5.5061,
-      "step": 26986
-    },
-    {
-      "epoch": 14.074054758800521,
-      "grad_norm": 1.3433825969696045,
-      "learning_rate": 7.338793969849247e-05,
-      "loss": 5.0372,
-      "step": 26987
-    },
-    {
-      "epoch": 14.07457627118644,
-      "grad_norm": 1.4765232801437378,
-      "learning_rate": 7.338693467336684e-05,
-      "loss": 5.5381,
-      "step": 26988
-    },
-    {
-      "epoch": 14.07509778357236,
-      "grad_norm": 1.53202486038208,
-      "learning_rate": 7.338592964824122e-05,
-      "loss": 5.4226,
-      "step": 26989
-    },
-    {
-      "epoch": 14.07561929595828,
-      "grad_norm": 1.353413701057434,
-      "learning_rate": 7.338492462311558e-05,
-      "loss": 5.6609,
-      "step": 26990
-    },
-    {
-      "epoch": 14.076140808344197,
-      "grad_norm": 1.4082856178283691,
-      "learning_rate": 7.338391959798996e-05,
-      "loss": 5.3151,
-      "step": 26991
-    },
-    {
-      "epoch": 14.076662320730117,
-      "grad_norm": 1.5153356790542603,
-      "learning_rate": 7.338291457286432e-05,
-      "loss": 5.1468,
-      "step": 26992
-    },
-    {
-      "epoch": 14.077183833116036,
-      "grad_norm": 1.5493944883346558,
-      "learning_rate": 7.338190954773869e-05,
-      "loss": 5.3373,
-      "step": 26993
-    },
-    {
-      "epoch": 14.077705345501956,
-      "grad_norm": 1.4763147830963135,
-      "learning_rate": 7.338090452261307e-05,
-      "loss": 5.4442,
-      "step": 26994
-    },
-    {
-      "epoch": 14.078226857887875,
-      "grad_norm": 1.4531883001327515,
-      "learning_rate": 7.337989949748743e-05,
-      "loss": 5.1473,
-      "step": 26995
-    },
-    {
-      "epoch": 14.078748370273795,
-      "grad_norm": 1.440550684928894,
-      "learning_rate": 7.337889447236181e-05,
-      "loss": 5.6688,
-      "step": 26996
-    },
-    {
-      "epoch": 14.079269882659712,
-      "grad_norm": 1.3294382095336914,
-      "learning_rate": 7.337788944723619e-05,
-      "loss": 5.3996,
-      "step": 26997
-    },
-    {
-      "epoch": 14.079791395045632,
-      "grad_norm": 1.4270083904266357,
-      "learning_rate": 7.337688442211056e-05,
-      "loss": 5.0207,
-      "step": 26998
-    },
-    {
-      "epoch": 14.080312907431551,
-      "grad_norm": 1.4045442342758179,
-      "learning_rate": 7.337587939698493e-05,
-      "loss": 5.524,
-      "step": 26999
-    },
-    {
-      "epoch": 14.08083441981747,
-      "grad_norm": 1.4707595109939575,
-      "learning_rate": 7.33748743718593e-05,
-      "loss": 5.3835,
-      "step": 27000
-    },
-    {
-      "epoch": 14.08135593220339,
-      "grad_norm": 1.510719895362854,
-      "learning_rate": 7.337386934673367e-05,
-      "loss": 5.2587,
-      "step": 27001
-    },
-    {
-      "epoch": 14.08187744458931,
-      "grad_norm": 1.5471960306167603,
-      "learning_rate": 7.337286432160805e-05,
-      "loss": 4.8126,
-      "step": 27002
-    },
-    {
-      "epoch": 14.082398956975227,
-      "grad_norm": 1.547533631324768,
-      "learning_rate": 7.337185929648241e-05,
-      "loss": 4.9195,
-      "step": 27003
-    },
-    {
-      "epoch": 14.082920469361147,
-      "grad_norm": 1.5873563289642334,
-      "learning_rate": 7.337085427135679e-05,
-      "loss": 5.1004,
-      "step": 27004
-    },
-    {
-      "epoch": 14.083441981747066,
-      "grad_norm": 1.5183711051940918,
-      "learning_rate": 7.336984924623115e-05,
-      "loss": 4.7681,
-      "step": 27005
-    },
-    {
-      "epoch": 14.083963494132986,
-      "grad_norm": 1.5216656923294067,
-      "learning_rate": 7.336884422110553e-05,
-      "loss": 5.5801,
-      "step": 27006
-    },
-    {
-      "epoch": 14.084485006518905,
-      "grad_norm": 1.4117413759231567,
-      "learning_rate": 7.336783919597991e-05,
-      "loss": 5.3513,
-      "step": 27007
-    },
-    {
-      "epoch": 14.085006518904825,
-      "grad_norm": 1.3514903783798218,
-      "learning_rate": 7.336683417085427e-05,
-      "loss": 5.5768,
-      "step": 27008
-    },
-    {
-      "epoch": 14.085528031290742,
-      "grad_norm": 1.4644232988357544,
-      "learning_rate": 7.336582914572865e-05,
-      "loss": 5.4679,
-      "step": 27009
-    },
-    {
-      "epoch": 14.086049543676662,
-      "grad_norm": 1.612889289855957,
-      "learning_rate": 7.336482412060302e-05,
-      "loss": 5.0104,
-      "step": 27010
-    },
-    {
-      "epoch": 14.086571056062581,
-      "grad_norm": 1.7338007688522339,
-      "learning_rate": 7.33638190954774e-05,
-      "loss": 4.8515,
-      "step": 27011
-    },
-    {
-      "epoch": 14.0870925684485,
-      "grad_norm": 1.7143433094024658,
-      "learning_rate": 7.336281407035176e-05,
-      "loss": 4.8111,
-      "step": 27012
-    },
-    {
-      "epoch": 14.08761408083442,
-      "grad_norm": 1.667224645614624,
-      "learning_rate": 7.336180904522614e-05,
-      "loss": 4.7538,
-      "step": 27013
-    },
-    {
-      "epoch": 14.08813559322034,
-      "grad_norm": 1.369659423828125,
-      "learning_rate": 7.33608040201005e-05,
-      "loss": 5.377,
-      "step": 27014
-    },
-    {
-      "epoch": 14.088657105606258,
-      "grad_norm": 1.4101405143737793,
-      "learning_rate": 7.335979899497488e-05,
-      "loss": 5.6342,
-      "step": 27015
-    },
-    {
-      "epoch": 14.089178617992177,
-      "grad_norm": 1.4364200830459595,
-      "learning_rate": 7.335879396984924e-05,
-      "loss": 5.4324,
-      "step": 27016
-    },
-    {
-      "epoch": 14.089700130378096,
-      "grad_norm": 1.4673963785171509,
-      "learning_rate": 7.335778894472362e-05,
-      "loss": 5.4297,
-      "step": 27017
-    },
-    {
-      "epoch": 14.090221642764016,
-      "grad_norm": 1.4626142978668213,
-      "learning_rate": 7.3356783919598e-05,
-      "loss": 5.2554,
-      "step": 27018
-    },
-    {
-      "epoch": 14.090743155149935,
-      "grad_norm": 1.5147954225540161,
-      "learning_rate": 7.335577889447236e-05,
-      "loss": 5.3369,
-      "step": 27019
-    },
-    {
-      "epoch": 14.091264667535855,
-      "grad_norm": 1.457413911819458,
-      "learning_rate": 7.335477386934674e-05,
-      "loss": 4.85,
-      "step": 27020
-    },
-    {
-      "epoch": 14.091786179921773,
-      "grad_norm": 1.4690111875534058,
-      "learning_rate": 7.33537688442211e-05,
-      "loss": 5.3016,
-      "step": 27021
-    },
-    {
-      "epoch": 14.092307692307692,
-      "grad_norm": 1.516996145248413,
-      "learning_rate": 7.335276381909548e-05,
-      "loss": 5.3119,
-      "step": 27022
-    },
-    {
-      "epoch": 14.092829204693611,
-      "grad_norm": 1.3609051704406738,
-      "learning_rate": 7.335175879396985e-05,
-      "loss": 5.2329,
-      "step": 27023
-    },
-    {
-      "epoch": 14.093350717079531,
-      "grad_norm": 1.481511116027832,
-      "learning_rate": 7.335075376884422e-05,
-      "loss": 5.0991,
-      "step": 27024
-    },
-    {
-      "epoch": 14.09387222946545,
-      "grad_norm": 1.4077115058898926,
-      "learning_rate": 7.334974874371859e-05,
-      "loss": 5.5002,
-      "step": 27025
-    },
-    {
-      "epoch": 14.09439374185137,
-      "grad_norm": 1.4823514223098755,
-      "learning_rate": 7.334874371859297e-05,
-      "loss": 5.1459,
-      "step": 27026
-    },
-    {
-      "epoch": 14.094915254237288,
-      "grad_norm": 1.3597832918167114,
-      "learning_rate": 7.334773869346734e-05,
-      "loss": 5.551,
-      "step": 27027
-    },
-    {
-      "epoch": 14.095436766623207,
-      "grad_norm": 1.6007215976715088,
-      "learning_rate": 7.334673366834172e-05,
-      "loss": 4.9726,
-      "step": 27028
-    },
-    {
-      "epoch": 14.095958279009126,
-      "grad_norm": 1.4971541166305542,
-      "learning_rate": 7.334572864321609e-05,
-      "loss": 5.1495,
-      "step": 27029
-    },
-    {
-      "epoch": 14.096479791395046,
-      "grad_norm": 1.5027902126312256,
-      "learning_rate": 7.334472361809046e-05,
-      "loss": 5.0862,
-      "step": 27030
-    },
-    {
-      "epoch": 14.097001303780965,
-      "grad_norm": 1.4562513828277588,
-      "learning_rate": 7.334371859296483e-05,
-      "loss": 5.7233,
-      "step": 27031
-    },
-    {
-      "epoch": 14.097522816166883,
-      "grad_norm": 1.3958607912063599,
-      "learning_rate": 7.334271356783919e-05,
-      "loss": 5.57,
-      "step": 27032
-    },
-    {
-      "epoch": 14.098044328552803,
-      "grad_norm": 1.4371981620788574,
-      "learning_rate": 7.334170854271357e-05,
-      "loss": 5.3901,
-      "step": 27033
-    },
-    {
-      "epoch": 14.098565840938722,
-      "grad_norm": 1.4037402868270874,
-      "learning_rate": 7.334070351758793e-05,
-      "loss": 5.3456,
-      "step": 27034
-    },
-    {
-      "epoch": 14.099087353324641,
-      "grad_norm": 1.4690951108932495,
-      "learning_rate": 7.333969849246231e-05,
-      "loss": 5.4477,
-      "step": 27035
-    },
-    {
-      "epoch": 14.099608865710561,
-      "grad_norm": 1.6597872972488403,
-      "learning_rate": 7.333869346733668e-05,
-      "loss": 5.1899,
-      "step": 27036
-    },
-    {
-      "epoch": 14.10013037809648,
-      "grad_norm": 1.6636402606964111,
-      "learning_rate": 7.333768844221105e-05,
-      "loss": 5.0831,
-      "step": 27037
-    },
-    {
-      "epoch": 14.100651890482398,
-      "grad_norm": 1.4132986068725586,
-      "learning_rate": 7.333668341708543e-05,
-      "loss": 5.6209,
-      "step": 27038
-    },
-    {
-      "epoch": 14.101173402868318,
-      "grad_norm": 1.4757620096206665,
-      "learning_rate": 7.333567839195981e-05,
-      "loss": 5.3182,
-      "step": 27039
-    },
-    {
-      "epoch": 14.101694915254237,
-      "grad_norm": 1.5559535026550293,
-      "learning_rate": 7.333467336683417e-05,
-      "loss": 5.3468,
-      "step": 27040
-    },
-    {
-      "epoch": 14.102216427640156,
-      "grad_norm": 1.4102482795715332,
-      "learning_rate": 7.333366834170855e-05,
-      "loss": 5.4503,
-      "step": 27041
-    },
-    {
-      "epoch": 14.102737940026076,
-      "grad_norm": 1.4898840188980103,
-      "learning_rate": 7.333266331658292e-05,
-      "loss": 5.145,
-      "step": 27042
-    },
-    {
-      "epoch": 14.103259452411995,
-      "grad_norm": 1.4442601203918457,
-      "learning_rate": 7.33316582914573e-05,
-      "loss": 5.2195,
-      "step": 27043
-    },
-    {
-      "epoch": 14.103780964797913,
-      "grad_norm": 1.3884971141815186,
-      "learning_rate": 7.333065326633166e-05,
-      "loss": 5.8017,
-      "step": 27044
-    },
-    {
-      "epoch": 14.104302477183833,
-      "grad_norm": 1.483178734779358,
-      "learning_rate": 7.332964824120602e-05,
-      "loss": 5.2426,
-      "step": 27045
-    },
-    {
-      "epoch": 14.104823989569752,
-      "grad_norm": 1.4444904327392578,
-      "learning_rate": 7.33286432160804e-05,
-      "loss": 5.69,
-      "step": 27046
-    },
-    {
-      "epoch": 14.105345501955671,
-      "grad_norm": 1.5036208629608154,
-      "learning_rate": 7.332763819095478e-05,
-      "loss": 5.2371,
-      "step": 27047
-    },
-    {
-      "epoch": 14.105867014341591,
-      "grad_norm": 1.4785077571868896,
-      "learning_rate": 7.332663316582916e-05,
-      "loss": 5.316,
-      "step": 27048
-    },
-    {
-      "epoch": 14.10638852672751,
-      "grad_norm": 1.48932683467865,
-      "learning_rate": 7.332562814070352e-05,
-      "loss": 5.3433,
-      "step": 27049
-    },
-    {
-      "epoch": 14.106910039113428,
-      "grad_norm": 1.4788758754730225,
-      "learning_rate": 7.33246231155779e-05,
-      "loss": 4.9079,
-      "step": 27050
-    },
-    {
-      "epoch": 14.107431551499348,
-      "grad_norm": 1.4703607559204102,
-      "learning_rate": 7.332361809045226e-05,
-      "loss": 4.9579,
-      "step": 27051
-    },
-    {
-      "epoch": 14.107953063885267,
-      "grad_norm": 1.4287683963775635,
-      "learning_rate": 7.332261306532664e-05,
-      "loss": 5.4374,
-      "step": 27052
-    },
-    {
-      "epoch": 14.108474576271187,
-      "grad_norm": 1.4590232372283936,
-      "learning_rate": 7.3321608040201e-05,
-      "loss": 5.4763,
-      "step": 27053
-    },
-    {
-      "epoch": 14.108996088657106,
-      "grad_norm": 1.5458266735076904,
-      "learning_rate": 7.332060301507538e-05,
-      "loss": 5.5374,
-      "step": 27054
-    },
-    {
-      "epoch": 14.109517601043025,
-      "grad_norm": 1.5580174922943115,
-      "learning_rate": 7.331959798994975e-05,
-      "loss": 4.6293,
-      "step": 27055
-    },
-    {
-      "epoch": 14.110039113428943,
-      "grad_norm": 1.40945565700531,
-      "learning_rate": 7.331859296482412e-05,
-      "loss": 5.2408,
-      "step": 27056
-    },
-    {
-      "epoch": 14.110560625814863,
-      "grad_norm": 1.3677732944488525,
-      "learning_rate": 7.331758793969849e-05,
-      "loss": 5.2794,
-      "step": 27057
-    },
-    {
-      "epoch": 14.111082138200782,
-      "grad_norm": 1.4609769582748413,
-      "learning_rate": 7.331658291457287e-05,
-      "loss": 5.5849,
-      "step": 27058
-    },
-    {
-      "epoch": 14.111603650586702,
-      "grad_norm": 1.5783048868179321,
-      "learning_rate": 7.331557788944724e-05,
-      "loss": 5.425,
-      "step": 27059
-    },
-    {
-      "epoch": 14.112125162972621,
-      "grad_norm": 1.466243028640747,
-      "learning_rate": 7.331457286432161e-05,
-      "loss": 5.0443,
-      "step": 27060
-    },
-    {
-      "epoch": 14.11264667535854,
-      "grad_norm": 1.692743182182312,
-      "learning_rate": 7.331356783919599e-05,
-      "loss": 5.0858,
-      "step": 27061
-    },
-    {
-      "epoch": 14.113168187744458,
-      "grad_norm": 1.5572959184646606,
-      "learning_rate": 7.331256281407035e-05,
-      "loss": 5.2547,
-      "step": 27062
-    },
-    {
-      "epoch": 14.113689700130378,
-      "grad_norm": 1.7037220001220703,
-      "learning_rate": 7.331155778894473e-05,
-      "loss": 5.4348,
-      "step": 27063
-    },
-    {
-      "epoch": 14.114211212516297,
-      "grad_norm": 1.5080400705337524,
-      "learning_rate": 7.33105527638191e-05,
-      "loss": 5.792,
-      "step": 27064
-    },
-    {
-      "epoch": 14.114732724902217,
-      "grad_norm": 1.3987113237380981,
-      "learning_rate": 7.330954773869347e-05,
-      "loss": 5.6384,
-      "step": 27065
-    },
-    {
-      "epoch": 14.115254237288136,
-      "grad_norm": 1.4800809621810913,
-      "learning_rate": 7.330854271356784e-05,
-      "loss": 5.6373,
-      "step": 27066
-    },
-    {
-      "epoch": 14.115775749674055,
-      "grad_norm": 1.6037331819534302,
-      "learning_rate": 7.330753768844221e-05,
-      "loss": 5.3818,
-      "step": 27067
-    },
-    {
-      "epoch": 14.116297262059973,
-      "grad_norm": 1.4308743476867676,
-      "learning_rate": 7.330653266331659e-05,
-      "loss": 5.5627,
-      "step": 27068
-    },
-    {
-      "epoch": 14.116818774445893,
-      "grad_norm": 1.5622793436050415,
-      "learning_rate": 7.330552763819097e-05,
-      "loss": 5.42,
-      "step": 27069
-    },
-    {
-      "epoch": 14.117340286831812,
-      "grad_norm": 1.5118988752365112,
-      "learning_rate": 7.330452261306533e-05,
-      "loss": 5.2206,
-      "step": 27070
-    },
-    {
-      "epoch": 14.117861799217732,
-      "grad_norm": 1.3727740049362183,
-      "learning_rate": 7.330351758793971e-05,
-      "loss": 5.1242,
-      "step": 27071
-    },
-    {
-      "epoch": 14.118383311603651,
-      "grad_norm": 1.504246473312378,
-      "learning_rate": 7.330251256281408e-05,
-      "loss": 5.3947,
-      "step": 27072
-    },
-    {
-      "epoch": 14.11890482398957,
-      "grad_norm": 1.5863585472106934,
-      "learning_rate": 7.330150753768844e-05,
-      "loss": 5.5447,
-      "step": 27073
-    },
-    {
-      "epoch": 14.119426336375488,
-      "grad_norm": 1.935731291770935,
-      "learning_rate": 7.330050251256282e-05,
-      "loss": 5.1874,
-      "step": 27074
-    },
-    {
-      "epoch": 14.119947848761408,
-      "grad_norm": 1.8895453214645386,
-      "learning_rate": 7.329949748743718e-05,
-      "loss": 5.2374,
-      "step": 27075
-    },
-    {
-      "epoch": 14.120469361147327,
-      "grad_norm": 1.5498055219650269,
-      "learning_rate": 7.329849246231156e-05,
-      "loss": 5.2522,
-      "step": 27076
-    },
-    {
-      "epoch": 14.120990873533247,
-      "grad_norm": 1.3865498304367065,
-      "learning_rate": 7.329748743718592e-05,
-      "loss": 5.4769,
-      "step": 27077
-    },
-    {
-      "epoch": 14.121512385919166,
-      "grad_norm": 1.4979712963104248,
-      "learning_rate": 7.32964824120603e-05,
-      "loss": 5.3904,
-      "step": 27078
-    },
-    {
-      "epoch": 14.122033898305085,
-      "grad_norm": 1.3477623462677002,
-      "learning_rate": 7.329547738693468e-05,
-      "loss": 5.0017,
-      "step": 27079
-    },
-    {
-      "epoch": 14.122555410691003,
-      "grad_norm": 1.5227361917495728,
-      "learning_rate": 7.329447236180906e-05,
-      "loss": 4.9792,
-      "step": 27080
-    },
-    {
-      "epoch": 14.123076923076923,
-      "grad_norm": 1.6090432405471802,
-      "learning_rate": 7.329346733668342e-05,
-      "loss": 5.3994,
-      "step": 27081
-    },
-    {
-      "epoch": 14.123598435462842,
-      "grad_norm": 1.3310611248016357,
-      "learning_rate": 7.32924623115578e-05,
-      "loss": 5.0064,
-      "step": 27082
-    },
-    {
-      "epoch": 14.124119947848762,
-      "grad_norm": 1.534145712852478,
-      "learning_rate": 7.329145728643216e-05,
-      "loss": 5.7013,
-      "step": 27083
-    },
-    {
-      "epoch": 14.124641460234681,
-      "grad_norm": 1.4824138879776,
-      "learning_rate": 7.329045226130654e-05,
-      "loss": 5.7482,
-      "step": 27084
-    },
-    {
-      "epoch": 14.1251629726206,
-      "grad_norm": 1.4732816219329834,
-      "learning_rate": 7.32894472361809e-05,
-      "loss": 5.6068,
-      "step": 27085
-    },
-    {
-      "epoch": 14.125684485006518,
-      "grad_norm": 1.4836137294769287,
-      "learning_rate": 7.328844221105527e-05,
-      "loss": 5.4504,
-      "step": 27086
-    },
-    {
-      "epoch": 14.126205997392438,
-      "grad_norm": 1.4181419610977173,
-      "learning_rate": 7.328743718592965e-05,
-      "loss": 5.7184,
-      "step": 27087
-    },
-    {
-      "epoch": 14.126727509778357,
-      "grad_norm": 1.5025229454040527,
-      "learning_rate": 7.328643216080403e-05,
-      "loss": 5.8437,
-      "step": 27088
-    },
-    {
-      "epoch": 14.127249022164277,
-      "grad_norm": 1.5615384578704834,
-      "learning_rate": 7.32854271356784e-05,
-      "loss": 5.3272,
-      "step": 27089
-    },
-    {
-      "epoch": 14.127770534550196,
-      "grad_norm": 1.5412967205047607,
-      "learning_rate": 7.328442211055277e-05,
-      "loss": 5.0856,
-      "step": 27090
-    },
-    {
-      "epoch": 14.128292046936116,
-      "grad_norm": 1.6379923820495605,
-      "learning_rate": 7.328341708542715e-05,
-      "loss": 5.4444,
-      "step": 27091
-    },
-    {
-      "epoch": 14.128813559322033,
-      "grad_norm": 1.45161771774292,
-      "learning_rate": 7.328241206030151e-05,
-      "loss": 5.2338,
-      "step": 27092
-    },
-    {
-      "epoch": 14.129335071707953,
-      "grad_norm": 1.892775058746338,
-      "learning_rate": 7.328140703517589e-05,
-      "loss": 4.9928,
-      "step": 27093
-    },
-    {
-      "epoch": 14.129856584093872,
-      "grad_norm": 1.559345006942749,
-      "learning_rate": 7.328040201005025e-05,
-      "loss": 5.3689,
-      "step": 27094
-    },
-    {
-      "epoch": 14.130378096479792,
-      "grad_norm": 1.3712787628173828,
-      "learning_rate": 7.327939698492463e-05,
-      "loss": 5.7156,
-      "step": 27095
-    },
-    {
-      "epoch": 14.130899608865711,
-      "grad_norm": 1.8041661977767944,
-      "learning_rate": 7.3278391959799e-05,
-      "loss": 4.8902,
-      "step": 27096
-    },
-    {
-      "epoch": 14.13142112125163,
-      "grad_norm": 1.5413931608200073,
-      "learning_rate": 7.327738693467337e-05,
-      "loss": 5.3175,
-      "step": 27097
-    },
-    {
-      "epoch": 14.131942633637548,
-      "grad_norm": 1.5107601881027222,
-      "learning_rate": 7.327638190954774e-05,
-      "loss": 5.1382,
-      "step": 27098
-    },
-    {
-      "epoch": 14.132464146023468,
-      "grad_norm": 1.5019197463989258,
-      "learning_rate": 7.327537688442211e-05,
-      "loss": 5.5119,
-      "step": 27099
-    },
-    {
-      "epoch": 14.132985658409387,
-      "grad_norm": 1.3601258993148804,
-      "learning_rate": 7.327437185929649e-05,
-      "loss": 5.5461,
-      "step": 27100
-    },
-    {
-      "epoch": 14.133507170795307,
-      "grad_norm": 1.5678355693817139,
-      "learning_rate": 7.327336683417086e-05,
-      "loss": 4.9278,
-      "step": 27101
-    },
-    {
-      "epoch": 14.134028683181226,
-      "grad_norm": 1.5653232336044312,
-      "learning_rate": 7.327236180904523e-05,
-      "loss": 5.2102,
-      "step": 27102
-    },
-    {
-      "epoch": 14.134550195567146,
-      "grad_norm": 1.5601693391799927,
-      "learning_rate": 7.32713567839196e-05,
-      "loss": 4.8845,
-      "step": 27103
-    },
-    {
-      "epoch": 14.135071707953063,
-      "grad_norm": 1.4565255641937256,
-      "learning_rate": 7.327035175879398e-05,
-      "loss": 5.7434,
-      "step": 27104
-    },
-    {
-      "epoch": 14.135593220338983,
-      "grad_norm": 1.3909814357757568,
-      "learning_rate": 7.326934673366834e-05,
-      "loss": 5.42,
-      "step": 27105
-    },
-    {
-      "epoch": 14.136114732724902,
-      "grad_norm": 1.438152551651001,
-      "learning_rate": 7.326834170854272e-05,
-      "loss": 5.2644,
-      "step": 27106
-    },
-    {
-      "epoch": 14.136636245110822,
-      "grad_norm": 1.5432438850402832,
-      "learning_rate": 7.326733668341708e-05,
-      "loss": 5.398,
-      "step": 27107
-    },
-    {
-      "epoch": 14.137157757496741,
-      "grad_norm": 1.438012719154358,
-      "learning_rate": 7.326633165829146e-05,
-      "loss": 5.6347,
-      "step": 27108
-    },
-    {
-      "epoch": 14.13767926988266,
-      "grad_norm": 1.3561068773269653,
-      "learning_rate": 7.326532663316584e-05,
-      "loss": 5.5383,
-      "step": 27109
-    },
-    {
-      "epoch": 14.138200782268578,
-      "grad_norm": 1.5052940845489502,
-      "learning_rate": 7.326432160804022e-05,
-      "loss": 5.4658,
-      "step": 27110
-    },
-    {
-      "epoch": 14.138722294654498,
-      "grad_norm": 1.49449622631073,
-      "learning_rate": 7.326331658291458e-05,
-      "loss": 5.6203,
-      "step": 27111
-    },
-    {
-      "epoch": 14.139243807040417,
-      "grad_norm": 1.4658457040786743,
-      "learning_rate": 7.326231155778894e-05,
-      "loss": 5.4846,
-      "step": 27112
-    },
-    {
-      "epoch": 14.139765319426337,
-      "grad_norm": 1.43537437915802,
-      "learning_rate": 7.326130653266332e-05,
-      "loss": 5.0746,
-      "step": 27113
-    },
-    {
-      "epoch": 14.140286831812256,
-      "grad_norm": 1.5695340633392334,
-      "learning_rate": 7.326030150753769e-05,
-      "loss": 4.8783,
-      "step": 27114
-    },
-    {
-      "epoch": 14.140808344198176,
-      "grad_norm": 1.5906258821487427,
-      "learning_rate": 7.325929648241206e-05,
-      "loss": 4.8649,
-      "step": 27115
-    },
-    {
-      "epoch": 14.141329856584093,
-      "grad_norm": 1.5330979824066162,
-      "learning_rate": 7.325829145728643e-05,
-      "loss": 5.6333,
-      "step": 27116
-    },
-    {
-      "epoch": 14.141851368970013,
-      "grad_norm": 1.4666411876678467,
-      "learning_rate": 7.32572864321608e-05,
-      "loss": 5.5278,
-      "step": 27117
-    },
-    {
-      "epoch": 14.142372881355932,
-      "grad_norm": 1.476550579071045,
-      "learning_rate": 7.325628140703517e-05,
-      "loss": 5.3708,
-      "step": 27118
-    },
-    {
-      "epoch": 14.142894393741852,
-      "grad_norm": 1.544860601425171,
-      "learning_rate": 7.325527638190955e-05,
-      "loss": 5.0239,
-      "step": 27119
-    },
-    {
-      "epoch": 14.143415906127771,
-      "grad_norm": 1.5729296207427979,
-      "learning_rate": 7.325427135678393e-05,
-      "loss": 5.0477,
-      "step": 27120
-    },
-    {
-      "epoch": 14.14393741851369,
-      "grad_norm": 1.5710042715072632,
-      "learning_rate": 7.32532663316583e-05,
-      "loss": 5.003,
-      "step": 27121
-    },
-    {
-      "epoch": 14.144458930899608,
-      "grad_norm": 1.501647710800171,
-      "learning_rate": 7.325226130653267e-05,
-      "loss": 5.8492,
-      "step": 27122
-    },
-    {
-      "epoch": 14.144980443285528,
-      "grad_norm": 1.504891037940979,
-      "learning_rate": 7.325125628140705e-05,
-      "loss": 5.2151,
-      "step": 27123
-    },
-    {
-      "epoch": 14.145501955671447,
-      "grad_norm": 1.467157006263733,
-      "learning_rate": 7.325025125628141e-05,
-      "loss": 5.4131,
-      "step": 27124
-    },
-    {
-      "epoch": 14.146023468057367,
-      "grad_norm": 1.5662587881088257,
-      "learning_rate": 7.324924623115577e-05,
-      "loss": 5.2642,
-      "step": 27125
-    },
-    {
-      "epoch": 14.146544980443286,
-      "grad_norm": 1.5353291034698486,
-      "learning_rate": 7.324824120603015e-05,
-      "loss": 5.156,
-      "step": 27126
-    },
-    {
-      "epoch": 14.147066492829204,
-      "grad_norm": 1.4948354959487915,
-      "learning_rate": 7.324723618090452e-05,
-      "loss": 5.4854,
-      "step": 27127
-    },
-    {
-      "epoch": 14.147588005215123,
-      "grad_norm": 1.5600003004074097,
-      "learning_rate": 7.32462311557789e-05,
-      "loss": 4.7412,
-      "step": 27128
-    },
-    {
-      "epoch": 14.148109517601043,
-      "grad_norm": 1.447126030921936,
-      "learning_rate": 7.324522613065327e-05,
-      "loss": 4.8743,
-      "step": 27129
-    },
-    {
-      "epoch": 14.148631029986962,
-      "grad_norm": 1.4712954759597778,
-      "learning_rate": 7.324422110552765e-05,
-      "loss": 5.5138,
-      "step": 27130
-    },
-    {
-      "epoch": 14.149152542372882,
-      "grad_norm": 1.4763023853302002,
-      "learning_rate": 7.324321608040201e-05,
-      "loss": 5.1192,
-      "step": 27131
-    },
-    {
-      "epoch": 14.149674054758801,
-      "grad_norm": 1.3705309629440308,
-      "learning_rate": 7.324221105527639e-05,
-      "loss": 5.7604,
-      "step": 27132
-    },
-    {
-      "epoch": 14.150195567144719,
-      "grad_norm": 1.6101802587509155,
-      "learning_rate": 7.324120603015076e-05,
-      "loss": 5.2628,
-      "step": 27133
-    },
-    {
-      "epoch": 14.150717079530638,
-      "grad_norm": 1.4304100275039673,
-      "learning_rate": 7.324020100502513e-05,
-      "loss": 5.234,
-      "step": 27134
-    },
-    {
-      "epoch": 14.151238591916558,
-      "grad_norm": 1.5048824548721313,
-      "learning_rate": 7.32391959798995e-05,
-      "loss": 4.8807,
-      "step": 27135
-    },
-    {
-      "epoch": 14.151760104302477,
-      "grad_norm": 1.4495106935501099,
-      "learning_rate": 7.323819095477388e-05,
-      "loss": 5.319,
-      "step": 27136
-    },
-    {
-      "epoch": 14.152281616688397,
-      "grad_norm": 1.502549171447754,
-      "learning_rate": 7.323718592964824e-05,
-      "loss": 5.1632,
-      "step": 27137
-    },
-    {
-      "epoch": 14.152803129074316,
-      "grad_norm": 1.5764570236206055,
-      "learning_rate": 7.32361809045226e-05,
-      "loss": 5.4746,
-      "step": 27138
-    },
-    {
-      "epoch": 14.153324641460234,
-      "grad_norm": 1.3918178081512451,
-      "learning_rate": 7.323517587939698e-05,
-      "loss": 5.7829,
-      "step": 27139
-    },
-    {
-      "epoch": 14.153846153846153,
-      "grad_norm": 1.4709405899047852,
-      "learning_rate": 7.323417085427136e-05,
-      "loss": 5.2778,
-      "step": 27140
-    },
-    {
-      "epoch": 14.154367666232073,
-      "grad_norm": 1.333682656288147,
-      "learning_rate": 7.323316582914574e-05,
-      "loss": 5.9167,
-      "step": 27141
-    },
-    {
-      "epoch": 14.154889178617992,
-      "grad_norm": 1.3965404033660889,
-      "learning_rate": 7.32321608040201e-05,
-      "loss": 5.6617,
-      "step": 27142
-    },
-    {
-      "epoch": 14.155410691003912,
-      "grad_norm": 1.566819429397583,
-      "learning_rate": 7.323115577889448e-05,
-      "loss": 5.649,
-      "step": 27143
-    },
-    {
-      "epoch": 14.155932203389831,
-      "grad_norm": 1.4535058736801147,
-      "learning_rate": 7.323015075376885e-05,
-      "loss": 5.3768,
-      "step": 27144
-    },
-    {
-      "epoch": 14.156453715775749,
-      "grad_norm": 1.3459033966064453,
-      "learning_rate": 7.322914572864322e-05,
-      "loss": 5.6616,
-      "step": 27145
-    },
-    {
-      "epoch": 14.156975228161668,
-      "grad_norm": 1.4239634275436401,
-      "learning_rate": 7.322814070351759e-05,
-      "loss": 5.5242,
-      "step": 27146
-    },
-    {
-      "epoch": 14.157496740547588,
-      "grad_norm": 1.4476709365844727,
-      "learning_rate": 7.322713567839197e-05,
-      "loss": 5.2555,
-      "step": 27147
-    },
-    {
-      "epoch": 14.158018252933507,
-      "grad_norm": 1.4330954551696777,
-      "learning_rate": 7.322613065326633e-05,
-      "loss": 5.6404,
-      "step": 27148
-    },
-    {
-      "epoch": 14.158539765319427,
-      "grad_norm": 1.4078445434570312,
-      "learning_rate": 7.322512562814071e-05,
-      "loss": 4.9693,
-      "step": 27149
-    },
-    {
-      "epoch": 14.159061277705346,
-      "grad_norm": 1.513153314590454,
-      "learning_rate": 7.322412060301509e-05,
-      "loss": 4.9786,
-      "step": 27150
-    },
-    {
-      "epoch": 14.159582790091264,
-      "grad_norm": 1.4434455633163452,
-      "learning_rate": 7.322311557788946e-05,
-      "loss": 5.5798,
-      "step": 27151
-    },
-    {
-      "epoch": 14.160104302477183,
-      "grad_norm": 1.3965842723846436,
-      "learning_rate": 7.322211055276383e-05,
-      "loss": 5.6412,
-      "step": 27152
-    },
-    {
-      "epoch": 14.160625814863103,
-      "grad_norm": 1.480010986328125,
-      "learning_rate": 7.322110552763819e-05,
-      "loss": 5.2152,
-      "step": 27153
-    },
-    {
-      "epoch": 14.161147327249022,
-      "grad_norm": 1.5230145454406738,
-      "learning_rate": 7.322010050251257e-05,
-      "loss": 4.7261,
-      "step": 27154
-    },
-    {
-      "epoch": 14.161668839634942,
-      "grad_norm": 1.4433557987213135,
-      "learning_rate": 7.321909547738693e-05,
-      "loss": 5.7253,
-      "step": 27155
-    },
-    {
-      "epoch": 14.162190352020861,
-      "grad_norm": 1.3705871105194092,
-      "learning_rate": 7.321809045226131e-05,
-      "loss": 5.7439,
-      "step": 27156
-    },
-    {
-      "epoch": 14.162711864406779,
-      "grad_norm": 1.5261682271957397,
-      "learning_rate": 7.321708542713568e-05,
-      "loss": 5.4958,
-      "step": 27157
-    },
-    {
-      "epoch": 14.163233376792698,
-      "grad_norm": 1.59833824634552,
-      "learning_rate": 7.321608040201005e-05,
-      "loss": 5.619,
-      "step": 27158
-    },
-    {
-      "epoch": 14.163754889178618,
-      "grad_norm": 1.5576152801513672,
-      "learning_rate": 7.321507537688442e-05,
-      "loss": 4.826,
-      "step": 27159
-    },
-    {
-      "epoch": 14.164276401564537,
-      "grad_norm": 1.4859559535980225,
-      "learning_rate": 7.32140703517588e-05,
-      "loss": 5.5445,
-      "step": 27160
-    },
-    {
-      "epoch": 14.164797913950457,
-      "grad_norm": 1.4559345245361328,
-      "learning_rate": 7.321306532663317e-05,
-      "loss": 5.2723,
-      "step": 27161
-    },
-    {
-      "epoch": 14.165319426336376,
-      "grad_norm": 1.3420257568359375,
-      "learning_rate": 7.321206030150755e-05,
-      "loss": 5.4853,
-      "step": 27162
-    },
-    {
-      "epoch": 14.165840938722294,
-      "grad_norm": 1.489880084991455,
-      "learning_rate": 7.321105527638192e-05,
-      "loss": 5.3863,
-      "step": 27163
-    },
-    {
-      "epoch": 14.166362451108213,
-      "grad_norm": 1.5364068746566772,
-      "learning_rate": 7.32100502512563e-05,
-      "loss": 4.9218,
-      "step": 27164
-    },
-    {
-      "epoch": 14.166883963494133,
-      "grad_norm": 1.301005482673645,
-      "learning_rate": 7.320904522613066e-05,
-      "loss": 5.5706,
-      "step": 27165
-    },
-    {
-      "epoch": 14.167405475880052,
-      "grad_norm": 1.3205894231796265,
-      "learning_rate": 7.320804020100502e-05,
-      "loss": 5.489,
-      "step": 27166
-    },
-    {
-      "epoch": 14.167926988265972,
-      "grad_norm": 1.504630446434021,
-      "learning_rate": 7.32070351758794e-05,
-      "loss": 5.1757,
-      "step": 27167
-    },
-    {
-      "epoch": 14.168448500651891,
-      "grad_norm": 1.406087875366211,
-      "learning_rate": 7.320603015075376e-05,
-      "loss": 5.5894,
-      "step": 27168
-    },
-    {
-      "epoch": 14.168970013037809,
-      "grad_norm": 1.4903349876403809,
-      "learning_rate": 7.320502512562814e-05,
-      "loss": 5.3981,
-      "step": 27169
-    },
-    {
-      "epoch": 14.169491525423728,
-      "grad_norm": 1.4248340129852295,
-      "learning_rate": 7.32040201005025e-05,
-      "loss": 5.5165,
-      "step": 27170
-    },
-    {
-      "epoch": 14.170013037809648,
-      "grad_norm": 1.4388628005981445,
-      "learning_rate": 7.320301507537688e-05,
-      "loss": 5.7663,
-      "step": 27171
-    },
-    {
-      "epoch": 14.170534550195567,
-      "grad_norm": 1.431009292602539,
-      "learning_rate": 7.320201005025126e-05,
-      "loss": 5.4129,
-      "step": 27172
-    },
-    {
-      "epoch": 14.171056062581487,
-      "grad_norm": 1.4947129487991333,
-      "learning_rate": 7.320100502512564e-05,
-      "loss": 5.3228,
-      "step": 27173
-    },
-    {
-      "epoch": 14.171577574967406,
-      "grad_norm": 1.489804744720459,
-      "learning_rate": 7.32e-05,
-      "loss": 5.1572,
-      "step": 27174
-    },
-    {
-      "epoch": 14.172099087353324,
-      "grad_norm": 1.4551734924316406,
-      "learning_rate": 7.319899497487438e-05,
-      "loss": 5.4952,
-      "step": 27175
-    },
-    {
-      "epoch": 14.172620599739243,
-      "grad_norm": 1.4447124004364014,
-      "learning_rate": 7.319798994974875e-05,
-      "loss": 5.1216,
-      "step": 27176
-    },
-    {
-      "epoch": 14.173142112125163,
-      "grad_norm": 1.4400486946105957,
-      "learning_rate": 7.319698492462312e-05,
-      "loss": 5.5357,
-      "step": 27177
-    },
-    {
-      "epoch": 14.173663624511082,
-      "grad_norm": 1.5080485343933105,
-      "learning_rate": 7.319597989949749e-05,
-      "loss": 4.9627,
-      "step": 27178
-    },
-    {
-      "epoch": 14.174185136897002,
-      "grad_norm": 1.5429857969284058,
-      "learning_rate": 7.319497487437185e-05,
-      "loss": 5.3929,
-      "step": 27179
-    },
-    {
-      "epoch": 14.174706649282921,
-      "grad_norm": 1.5156521797180176,
-      "learning_rate": 7.319396984924623e-05,
-      "loss": 5.3672,
-      "step": 27180
-    },
-    {
-      "epoch": 14.175228161668839,
-      "grad_norm": 1.362364411354065,
-      "learning_rate": 7.319296482412061e-05,
-      "loss": 5.7604,
-      "step": 27181
-    },
-    {
-      "epoch": 14.175749674054758,
-      "grad_norm": 1.475502371788025,
-      "learning_rate": 7.319195979899499e-05,
-      "loss": 4.8596,
-      "step": 27182
-    },
-    {
-      "epoch": 14.176271186440678,
-      "grad_norm": 1.4980660676956177,
-      "learning_rate": 7.319095477386935e-05,
-      "loss": 5.4394,
-      "step": 27183
-    },
-    {
-      "epoch": 14.176792698826597,
-      "grad_norm": 1.5411347150802612,
-      "learning_rate": 7.318994974874373e-05,
-      "loss": 4.8863,
-      "step": 27184
-    },
-    {
-      "epoch": 14.177314211212517,
-      "grad_norm": 1.3914109468460083,
-      "learning_rate": 7.318894472361809e-05,
-      "loss": 5.7207,
-      "step": 27185
-    },
-    {
-      "epoch": 14.177835723598436,
-      "grad_norm": 1.4637866020202637,
-      "learning_rate": 7.318793969849247e-05,
-      "loss": 5.4958,
-      "step": 27186
-    },
-    {
-      "epoch": 14.178357235984354,
-      "grad_norm": 1.3790674209594727,
-      "learning_rate": 7.318693467336683e-05,
-      "loss": 5.7422,
-      "step": 27187
-    },
-    {
-      "epoch": 14.178878748370273,
-      "grad_norm": 1.4498777389526367,
-      "learning_rate": 7.318592964824121e-05,
-      "loss": 5.4895,
-      "step": 27188
-    },
-    {
-      "epoch": 14.179400260756193,
-      "grad_norm": 1.5177955627441406,
-      "learning_rate": 7.318492462311558e-05,
-      "loss": 5.0719,
-      "step": 27189
-    },
-    {
-      "epoch": 14.179921773142112,
-      "grad_norm": 1.4832698106765747,
-      "learning_rate": 7.318391959798995e-05,
-      "loss": 5.6235,
-      "step": 27190
-    },
-    {
-      "epoch": 14.180443285528032,
-      "grad_norm": 1.483486533164978,
-      "learning_rate": 7.318291457286432e-05,
-      "loss": 5.487,
-      "step": 27191
-    },
-    {
-      "epoch": 14.180964797913951,
-      "grad_norm": 1.3036777973175049,
-      "learning_rate": 7.31819095477387e-05,
-      "loss": 5.854,
-      "step": 27192
-    },
-    {
-      "epoch": 14.181486310299869,
-      "grad_norm": 1.6212745904922485,
-      "learning_rate": 7.318090452261307e-05,
-      "loss": 5.3741,
-      "step": 27193
-    },
-    {
-      "epoch": 14.182007822685788,
-      "grad_norm": 1.3661625385284424,
-      "learning_rate": 7.317989949748744e-05,
-      "loss": 5.9508,
-      "step": 27194
-    },
-    {
-      "epoch": 14.182529335071708,
-      "grad_norm": 1.4193960428237915,
-      "learning_rate": 7.317889447236182e-05,
-      "loss": 5.4099,
-      "step": 27195
-    },
-    {
-      "epoch": 14.183050847457627,
-      "grad_norm": 1.415579915046692,
-      "learning_rate": 7.317788944723618e-05,
-      "loss": 5.782,
-      "step": 27196
-    },
-    {
-      "epoch": 14.183572359843547,
-      "grad_norm": 1.4814778566360474,
-      "learning_rate": 7.317688442211056e-05,
-      "loss": 5.1448,
-      "step": 27197
-    },
-    {
-      "epoch": 14.184093872229466,
-      "grad_norm": 1.3976147174835205,
-      "learning_rate": 7.317587939698492e-05,
-      "loss": 5.2323,
-      "step": 27198
-    },
-    {
-      "epoch": 14.184615384615384,
-      "grad_norm": 1.4749821424484253,
-      "learning_rate": 7.31748743718593e-05,
-      "loss": 5.5298,
-      "step": 27199
-    },
-    {
-      "epoch": 14.185136897001303,
-      "grad_norm": 1.4186656475067139,
-      "learning_rate": 7.317386934673366e-05,
-      "loss": 5.4437,
-      "step": 27200
-    },
-    {
-      "epoch": 14.185658409387223,
-      "grad_norm": 1.4419045448303223,
-      "learning_rate": 7.317286432160804e-05,
-      "loss": 5.3758,
-      "step": 27201
-    },
-    {
-      "epoch": 14.186179921773142,
-      "grad_norm": 1.3796026706695557,
-      "learning_rate": 7.317185929648242e-05,
-      "loss": 5.533,
-      "step": 27202
-    },
-    {
-      "epoch": 14.186701434159062,
-      "grad_norm": 1.4630990028381348,
-      "learning_rate": 7.31708542713568e-05,
-      "loss": 5.273,
-      "step": 27203
-    },
-    {
-      "epoch": 14.187222946544981,
-      "grad_norm": 1.4841080904006958,
-      "learning_rate": 7.316984924623116e-05,
-      "loss": 5.4995,
-      "step": 27204
-    },
-    {
-      "epoch": 14.187744458930899,
-      "grad_norm": 1.5076159238815308,
-      "learning_rate": 7.316884422110553e-05,
-      "loss": 4.7752,
-      "step": 27205
-    },
-    {
-      "epoch": 14.188265971316818,
-      "grad_norm": 1.5728312730789185,
-      "learning_rate": 7.31678391959799e-05,
-      "loss": 5.565,
-      "step": 27206
-    },
-    {
-      "epoch": 14.188787483702738,
-      "grad_norm": 1.4428430795669556,
-      "learning_rate": 7.316683417085427e-05,
-      "loss": 5.1767,
-      "step": 27207
-    },
-    {
-      "epoch": 14.189308996088657,
-      "grad_norm": 1.519760251045227,
-      "learning_rate": 7.316582914572865e-05,
-      "loss": 5.1376,
-      "step": 27208
-    },
-    {
-      "epoch": 14.189830508474577,
-      "grad_norm": 1.4841769933700562,
-      "learning_rate": 7.316482412060301e-05,
-      "loss": 5.4555,
-      "step": 27209
-    },
-    {
-      "epoch": 14.190352020860496,
-      "grad_norm": 1.4226421117782593,
-      "learning_rate": 7.316381909547739e-05,
-      "loss": 5.6837,
-      "step": 27210
-    },
-    {
-      "epoch": 14.190873533246414,
-      "grad_norm": 1.5850707292556763,
-      "learning_rate": 7.316281407035175e-05,
-      "loss": 5.0193,
-      "step": 27211
-    },
-    {
-      "epoch": 14.191395045632333,
-      "grad_norm": 1.5148652791976929,
-      "learning_rate": 7.316180904522613e-05,
-      "loss": 5.5803,
-      "step": 27212
-    },
-    {
-      "epoch": 14.191916558018253,
-      "grad_norm": 1.4596511125564575,
-      "learning_rate": 7.316080402010051e-05,
-      "loss": 5.6349,
-      "step": 27213
-    },
-    {
-      "epoch": 14.192438070404172,
-      "grad_norm": 1.4664758443832397,
-      "learning_rate": 7.315979899497489e-05,
-      "loss": 5.2277,
-      "step": 27214
-    },
-    {
-      "epoch": 14.192959582790092,
-      "grad_norm": 1.314332365989685,
-      "learning_rate": 7.315879396984925e-05,
-      "loss": 5.2574,
-      "step": 27215
-    },
-    {
-      "epoch": 14.193481095176011,
-      "grad_norm": 1.6141880750656128,
-      "learning_rate": 7.315778894472363e-05,
-      "loss": 5.2735,
-      "step": 27216
-    },
-    {
-      "epoch": 14.194002607561929,
-      "grad_norm": 1.6232140064239502,
-      "learning_rate": 7.315678391959799e-05,
-      "loss": 5.3862,
-      "step": 27217
-    },
-    {
-      "epoch": 14.194524119947848,
-      "grad_norm": 1.467341661453247,
-      "learning_rate": 7.315577889447236e-05,
-      "loss": 5.3952,
-      "step": 27218
-    },
-    {
-      "epoch": 14.195045632333768,
-      "grad_norm": 1.511054277420044,
-      "learning_rate": 7.315477386934674e-05,
-      "loss": 5.61,
-      "step": 27219
-    },
-    {
-      "epoch": 14.195567144719687,
-      "grad_norm": 1.4263521432876587,
-      "learning_rate": 7.31537688442211e-05,
-      "loss": 5.4555,
-      "step": 27220
-    },
-    {
-      "epoch": 14.196088657105607,
-      "grad_norm": 1.4854775667190552,
-      "learning_rate": 7.315276381909548e-05,
-      "loss": 5.1756,
-      "step": 27221
-    },
-    {
-      "epoch": 14.196610169491525,
-      "grad_norm": 1.5135750770568848,
-      "learning_rate": 7.315175879396986e-05,
-      "loss": 5.1588,
-      "step": 27222
-    },
-    {
-      "epoch": 14.197131681877444,
-      "grad_norm": 1.5286197662353516,
-      "learning_rate": 7.315075376884423e-05,
-      "loss": 4.7782,
-      "step": 27223
-    },
-    {
-      "epoch": 14.197653194263363,
-      "grad_norm": 1.4718308448791504,
-      "learning_rate": 7.31497487437186e-05,
-      "loss": 5.6709,
-      "step": 27224
-    },
-    {
-      "epoch": 14.198174706649283,
-      "grad_norm": 1.4337424039840698,
-      "learning_rate": 7.314874371859297e-05,
-      "loss": 5.5513,
-      "step": 27225
-    },
-    {
-      "epoch": 14.198696219035202,
-      "grad_norm": 1.3613030910491943,
-      "learning_rate": 7.314773869346734e-05,
-      "loss": 5.7295,
-      "step": 27226
-    },
-    {
-      "epoch": 14.199217731421122,
-      "grad_norm": 1.3673616647720337,
-      "learning_rate": 7.314673366834172e-05,
-      "loss": 5.751,
-      "step": 27227
-    },
-    {
-      "epoch": 14.19973924380704,
-      "grad_norm": 1.4920179843902588,
-      "learning_rate": 7.314572864321608e-05,
-      "loss": 5.4003,
-      "step": 27228
-    },
-    {
-      "epoch": 14.200260756192959,
-      "grad_norm": 1.490362286567688,
-      "learning_rate": 7.314472361809046e-05,
-      "loss": 5.2547,
-      "step": 27229
-    },
-    {
-      "epoch": 14.200782268578878,
-      "grad_norm": 1.4213807582855225,
-      "learning_rate": 7.314371859296482e-05,
-      "loss": 5.5905,
-      "step": 27230
-    },
-    {
-      "epoch": 14.201303780964798,
-      "grad_norm": 1.5121954679489136,
-      "learning_rate": 7.31427135678392e-05,
-      "loss": 5.5349,
-      "step": 27231
-    },
-    {
-      "epoch": 14.201825293350717,
-      "grad_norm": 1.3352218866348267,
-      "learning_rate": 7.314170854271357e-05,
-      "loss": 5.7394,
-      "step": 27232
-    },
-    {
-      "epoch": 14.202346805736637,
-      "grad_norm": 1.5209839344024658,
-      "learning_rate": 7.314070351758794e-05,
-      "loss": 5.1608,
-      "step": 27233
-    },
-    {
-      "epoch": 14.202868318122555,
-      "grad_norm": 1.5732306241989136,
-      "learning_rate": 7.313969849246232e-05,
-      "loss": 5.1975,
-      "step": 27234
-    },
-    {
-      "epoch": 14.203389830508474,
-      "grad_norm": 1.4817997217178345,
-      "learning_rate": 7.313869346733669e-05,
-      "loss": 5.4027,
-      "step": 27235
-    },
-    {
-      "epoch": 14.203911342894393,
-      "grad_norm": 1.478000521659851,
-      "learning_rate": 7.313768844221106e-05,
-      "loss": 5.6123,
-      "step": 27236
-    },
-    {
-      "epoch": 14.204432855280313,
-      "grad_norm": 1.574125051498413,
-      "learning_rate": 7.313668341708543e-05,
-      "loss": 4.9564,
-      "step": 27237
-    },
-    {
-      "epoch": 14.204954367666232,
-      "grad_norm": 1.444568157196045,
-      "learning_rate": 7.31356783919598e-05,
-      "loss": 5.7197,
-      "step": 27238
-    },
-    {
-      "epoch": 14.205475880052152,
-      "grad_norm": 1.4567217826843262,
-      "learning_rate": 7.313467336683417e-05,
-      "loss": 5.3104,
-      "step": 27239
-    },
-    {
-      "epoch": 14.20599739243807,
-      "grad_norm": 1.386392593383789,
-      "learning_rate": 7.313366834170855e-05,
-      "loss": 5.3624,
-      "step": 27240
-    },
-    {
-      "epoch": 14.206518904823989,
-      "grad_norm": 1.388288140296936,
-      "learning_rate": 7.313266331658291e-05,
-      "loss": 5.3403,
-      "step": 27241
-    },
-    {
-      "epoch": 14.207040417209909,
-      "grad_norm": 1.3962334394454956,
-      "learning_rate": 7.313165829145729e-05,
-      "loss": 5.6319,
-      "step": 27242
-    },
-    {
-      "epoch": 14.207561929595828,
-      "grad_norm": 1.5229636430740356,
-      "learning_rate": 7.313065326633167e-05,
-      "loss": 4.9588,
-      "step": 27243
-    },
-    {
-      "epoch": 14.208083441981747,
-      "grad_norm": 1.4855610132217407,
-      "learning_rate": 7.312964824120605e-05,
-      "loss": 5.3265,
-      "step": 27244
-    },
-    {
-      "epoch": 14.208604954367667,
-      "grad_norm": 1.4759796857833862,
-      "learning_rate": 7.312864321608041e-05,
-      "loss": 5.3967,
-      "step": 27245
-    },
-    {
-      "epoch": 14.209126466753585,
-      "grad_norm": 1.4178539514541626,
-      "learning_rate": 7.312763819095477e-05,
-      "loss": 5.5571,
-      "step": 27246
-    },
-    {
-      "epoch": 14.209647979139504,
-      "grad_norm": 1.5256099700927734,
-      "learning_rate": 7.312663316582915e-05,
-      "loss": 5.2919,
-      "step": 27247
-    },
-    {
-      "epoch": 14.210169491525424,
-      "grad_norm": 1.5339365005493164,
-      "learning_rate": 7.312562814070352e-05,
-      "loss": 5.4475,
-      "step": 27248
-    },
-    {
-      "epoch": 14.210691003911343,
-      "grad_norm": 1.4199373722076416,
-      "learning_rate": 7.31246231155779e-05,
-      "loss": 4.953,
-      "step": 27249
-    },
-    {
-      "epoch": 14.211212516297262,
-      "grad_norm": 1.4339454174041748,
-      "learning_rate": 7.312361809045226e-05,
-      "loss": 5.1769,
-      "step": 27250
-    },
-    {
-      "epoch": 14.211734028683182,
-      "grad_norm": 1.43657386302948,
-      "learning_rate": 7.312261306532664e-05,
-      "loss": 5.0392,
-      "step": 27251
-    },
-    {
-      "epoch": 14.2122555410691,
-      "grad_norm": 1.4352450370788574,
-      "learning_rate": 7.3121608040201e-05,
-      "loss": 5.6362,
-      "step": 27252
-    },
-    {
-      "epoch": 14.212777053455019,
-      "grad_norm": 1.3909823894500732,
-      "learning_rate": 7.312060301507538e-05,
-      "loss": 5.5358,
-      "step": 27253
-    },
-    {
-      "epoch": 14.213298565840939,
-      "grad_norm": 1.590221643447876,
-      "learning_rate": 7.311959798994976e-05,
-      "loss": 5.0028,
-      "step": 27254
-    },
-    {
-      "epoch": 14.213820078226858,
-      "grad_norm": 1.4002997875213623,
-      "learning_rate": 7.311859296482413e-05,
-      "loss": 5.2444,
-      "step": 27255
-    },
-    {
-      "epoch": 14.214341590612777,
-      "grad_norm": 1.3776838779449463,
-      "learning_rate": 7.31175879396985e-05,
-      "loss": 5.5415,
-      "step": 27256
-    },
-    {
-      "epoch": 14.214863102998697,
-      "grad_norm": 1.3634021282196045,
-      "learning_rate": 7.311658291457288e-05,
-      "loss": 5.5866,
-      "step": 27257
-    },
-    {
-      "epoch": 14.215384615384615,
-      "grad_norm": 1.496278166770935,
-      "learning_rate": 7.311557788944724e-05,
-      "loss": 5.8045,
-      "step": 27258
-    },
-    {
-      "epoch": 14.215906127770534,
-      "grad_norm": 1.5851322412490845,
-      "learning_rate": 7.31145728643216e-05,
-      "loss": 5.1982,
-      "step": 27259
-    },
-    {
-      "epoch": 14.216427640156454,
-      "grad_norm": 1.50608491897583,
-      "learning_rate": 7.311356783919598e-05,
-      "loss": 5.4151,
-      "step": 27260
-    },
-    {
-      "epoch": 14.216949152542373,
-      "grad_norm": 1.3819899559020996,
-      "learning_rate": 7.311256281407035e-05,
-      "loss": 5.8766,
-      "step": 27261
-    },
-    {
-      "epoch": 14.217470664928292,
-      "grad_norm": 1.5295580625534058,
-      "learning_rate": 7.311155778894472e-05,
-      "loss": 5.1268,
-      "step": 27262
-    },
-    {
-      "epoch": 14.217992177314212,
-      "grad_norm": 1.4227226972579956,
-      "learning_rate": 7.31105527638191e-05,
-      "loss": 5.5679,
-      "step": 27263
-    },
-    {
-      "epoch": 14.21851368970013,
-      "grad_norm": 1.398829698562622,
-      "learning_rate": 7.310954773869348e-05,
-      "loss": 5.245,
-      "step": 27264
-    },
-    {
-      "epoch": 14.219035202086049,
-      "grad_norm": 1.415742039680481,
-      "learning_rate": 7.310854271356784e-05,
-      "loss": 5.7169,
-      "step": 27265
-    },
-    {
-      "epoch": 14.219556714471969,
-      "grad_norm": 1.364363670349121,
-      "learning_rate": 7.310753768844222e-05,
-      "loss": 5.6408,
-      "step": 27266
-    },
-    {
-      "epoch": 14.220078226857888,
-      "grad_norm": 1.8075542449951172,
-      "learning_rate": 7.310653266331659e-05,
-      "loss": 5.3408,
-      "step": 27267
-    },
-    {
-      "epoch": 14.220599739243807,
-      "grad_norm": 1.3957653045654297,
-      "learning_rate": 7.310552763819096e-05,
-      "loss": 5.3424,
-      "step": 27268
-    },
-    {
-      "epoch": 14.221121251629727,
-      "grad_norm": 1.4244279861450195,
-      "learning_rate": 7.310452261306533e-05,
-      "loss": 5.4808,
-      "step": 27269
-    },
-    {
-      "epoch": 14.221642764015645,
-      "grad_norm": 1.4251354932785034,
-      "learning_rate": 7.31035175879397e-05,
-      "loss": 5.2947,
-      "step": 27270
-    },
-    {
-      "epoch": 14.222164276401564,
-      "grad_norm": 1.4166768789291382,
-      "learning_rate": 7.310251256281407e-05,
-      "loss": 5.7704,
-      "step": 27271
-    },
-    {
-      "epoch": 14.222685788787484,
-      "grad_norm": 1.3820013999938965,
-      "learning_rate": 7.310150753768843e-05,
-      "loss": 5.665,
-      "step": 27272
-    },
-    {
-      "epoch": 14.223207301173403,
-      "grad_norm": 1.4777919054031372,
-      "learning_rate": 7.310050251256281e-05,
-      "loss": 5.6909,
-      "step": 27273
-    },
-    {
-      "epoch": 14.223728813559323,
-      "grad_norm": 1.5357627868652344,
-      "learning_rate": 7.309949748743719e-05,
-      "loss": 5.3044,
-      "step": 27274
-    },
-    {
-      "epoch": 14.224250325945242,
-      "grad_norm": 1.4086380004882812,
-      "learning_rate": 7.309849246231157e-05,
-      "loss": 5.0209,
-      "step": 27275
-    },
-    {
-      "epoch": 14.22477183833116,
-      "grad_norm": 1.4369885921478271,
-      "learning_rate": 7.309748743718593e-05,
-      "loss": 5.62,
-      "step": 27276
-    },
-    {
-      "epoch": 14.22529335071708,
-      "grad_norm": 1.482466220855713,
-      "learning_rate": 7.309648241206031e-05,
-      "loss": 5.6127,
-      "step": 27277
-    },
-    {
-      "epoch": 14.225814863102999,
-      "grad_norm": 1.407321572303772,
-      "learning_rate": 7.309547738693467e-05,
-      "loss": 5.7001,
-      "step": 27278
-    },
-    {
-      "epoch": 14.226336375488918,
-      "grad_norm": 1.3797773122787476,
-      "learning_rate": 7.309447236180905e-05,
-      "loss": 5.2694,
-      "step": 27279
-    },
-    {
-      "epoch": 14.226857887874838,
-      "grad_norm": 1.4066593647003174,
-      "learning_rate": 7.309346733668342e-05,
-      "loss": 4.9554,
-      "step": 27280
-    },
-    {
-      "epoch": 14.227379400260757,
-      "grad_norm": 1.443768858909607,
-      "learning_rate": 7.30924623115578e-05,
-      "loss": 5.4694,
-      "step": 27281
-    },
-    {
-      "epoch": 14.227900912646675,
-      "grad_norm": 1.434364676475525,
-      "learning_rate": 7.309145728643216e-05,
-      "loss": 5.4498,
-      "step": 27282
-    },
-    {
-      "epoch": 14.228422425032594,
-      "grad_norm": 1.5249860286712646,
-      "learning_rate": 7.309045226130654e-05,
-      "loss": 4.902,
-      "step": 27283
-    },
-    {
-      "epoch": 14.228943937418514,
-      "grad_norm": 1.563340425491333,
-      "learning_rate": 7.308944723618091e-05,
-      "loss": 5.4862,
-      "step": 27284
-    },
-    {
-      "epoch": 14.229465449804433,
-      "grad_norm": 1.4683183431625366,
-      "learning_rate": 7.308844221105528e-05,
-      "loss": 5.6635,
-      "step": 27285
-    },
-    {
-      "epoch": 14.229986962190353,
-      "grad_norm": 1.481260061264038,
-      "learning_rate": 7.308743718592966e-05,
-      "loss": 4.458,
-      "step": 27286
-    },
-    {
-      "epoch": 14.230508474576272,
-      "grad_norm": 1.4826347827911377,
-      "learning_rate": 7.308643216080402e-05,
-      "loss": 5.3891,
-      "step": 27287
-    },
-    {
-      "epoch": 14.23102998696219,
-      "grad_norm": 1.4255194664001465,
-      "learning_rate": 7.30854271356784e-05,
-      "loss": 5.6463,
-      "step": 27288
-    },
-    {
-      "epoch": 14.23155149934811,
-      "grad_norm": 1.563452124595642,
-      "learning_rate": 7.308442211055276e-05,
-      "loss": 5.0557,
-      "step": 27289
-    },
-    {
-      "epoch": 14.232073011734029,
-      "grad_norm": 1.444115161895752,
-      "learning_rate": 7.308341708542714e-05,
-      "loss": 5.7019,
-      "step": 27290
-    },
-    {
-      "epoch": 14.232594524119948,
-      "grad_norm": 1.3775157928466797,
-      "learning_rate": 7.30824120603015e-05,
-      "loss": 5.8407,
-      "step": 27291
-    },
-    {
-      "epoch": 14.233116036505868,
-      "grad_norm": 1.6213982105255127,
-      "learning_rate": 7.308140703517588e-05,
-      "loss": 5.668,
-      "step": 27292
-    },
-    {
-      "epoch": 14.233637548891787,
-      "grad_norm": 1.790926218032837,
-      "learning_rate": 7.308040201005025e-05,
-      "loss": 4.6647,
-      "step": 27293
-    },
-    {
-      "epoch": 14.234159061277705,
-      "grad_norm": 1.5044662952423096,
-      "learning_rate": 7.307939698492462e-05,
-      "loss": 5.1062,
-      "step": 27294
-    },
-    {
-      "epoch": 14.234680573663624,
-      "grad_norm": 1.4272797107696533,
-      "learning_rate": 7.3078391959799e-05,
-      "loss": 5.3878,
-      "step": 27295
-    },
-    {
-      "epoch": 14.235202086049544,
-      "grad_norm": 1.5998446941375732,
-      "learning_rate": 7.307738693467338e-05,
-      "loss": 5.3925,
-      "step": 27296
-    },
-    {
-      "epoch": 14.235723598435463,
-      "grad_norm": 1.4371025562286377,
-      "learning_rate": 7.307638190954774e-05,
-      "loss": 5.5894,
-      "step": 27297
-    },
-    {
-      "epoch": 14.236245110821383,
-      "grad_norm": 1.7426761388778687,
-      "learning_rate": 7.307537688442211e-05,
-      "loss": 4.9216,
-      "step": 27298
-    },
-    {
-      "epoch": 14.236766623207302,
-      "grad_norm": 1.3583760261535645,
-      "learning_rate": 7.307437185929649e-05,
-      "loss": 5.4269,
-      "step": 27299
-    },
-    {
-      "epoch": 14.23728813559322,
-      "grad_norm": 1.5230202674865723,
-      "learning_rate": 7.307336683417085e-05,
-      "loss": 5.4224,
-      "step": 27300
-    },
-    {
-      "epoch": 14.23780964797914,
-      "grad_norm": 1.3382148742675781,
-      "learning_rate": 7.307236180904523e-05,
-      "loss": 4.4273,
-      "step": 27301
-    },
-    {
-      "epoch": 14.238331160365059,
-      "grad_norm": 1.4035459756851196,
-      "learning_rate": 7.30713567839196e-05,
-      "loss": 5.5305,
-      "step": 27302
-    },
-    {
-      "epoch": 14.238852672750978,
-      "grad_norm": 1.483431100845337,
-      "learning_rate": 7.307035175879397e-05,
-      "loss": 5.3454,
-      "step": 27303
-    },
-    {
-      "epoch": 14.239374185136898,
-      "grad_norm": 1.4702520370483398,
-      "learning_rate": 7.306934673366835e-05,
-      "loss": 5.302,
-      "step": 27304
-    },
-    {
-      "epoch": 14.239895697522817,
-      "grad_norm": 1.5252269506454468,
-      "learning_rate": 7.306834170854273e-05,
-      "loss": 5.4236,
-      "step": 27305
-    },
-    {
-      "epoch": 14.240417209908735,
-      "grad_norm": 1.4960097074508667,
-      "learning_rate": 7.306733668341709e-05,
-      "loss": 5.4778,
-      "step": 27306
-    },
-    {
-      "epoch": 14.240938722294654,
-      "grad_norm": 1.4212547540664673,
-      "learning_rate": 7.306633165829147e-05,
-      "loss": 5.2442,
-      "step": 27307
-    },
-    {
-      "epoch": 14.241460234680574,
-      "grad_norm": 1.4477211236953735,
-      "learning_rate": 7.306532663316583e-05,
-      "loss": 5.4861,
-      "step": 27308
-    },
-    {
-      "epoch": 14.241981747066493,
-      "grad_norm": 1.5423709154129028,
-      "learning_rate": 7.306432160804021e-05,
-      "loss": 5.102,
-      "step": 27309
-    },
-    {
-      "epoch": 14.242503259452413,
-      "grad_norm": 1.3819835186004639,
-      "learning_rate": 7.306331658291458e-05,
-      "loss": 5.6208,
-      "step": 27310
-    },
-    {
-      "epoch": 14.243024771838332,
-      "grad_norm": 1.3800822496414185,
-      "learning_rate": 7.306231155778894e-05,
-      "loss": 5.5215,
-      "step": 27311
-    },
-    {
-      "epoch": 14.24354628422425,
-      "grad_norm": 1.463265299797058,
-      "learning_rate": 7.306130653266332e-05,
-      "loss": 5.1405,
-      "step": 27312
-    },
-    {
-      "epoch": 14.24406779661017,
-      "grad_norm": 1.4437463283538818,
-      "learning_rate": 7.306030150753768e-05,
-      "loss": 5.2342,
-      "step": 27313
-    },
-    {
-      "epoch": 14.244589308996089,
-      "grad_norm": 1.6281908750534058,
-      "learning_rate": 7.305929648241206e-05,
-      "loss": 4.8531,
-      "step": 27314
-    },
-    {
-      "epoch": 14.245110821382008,
-      "grad_norm": 1.4291805028915405,
-      "learning_rate": 7.305829145728644e-05,
-      "loss": 4.9616,
-      "step": 27315
-    },
-    {
-      "epoch": 14.245632333767928,
-      "grad_norm": 1.5477478504180908,
-      "learning_rate": 7.305728643216082e-05,
-      "loss": 5.3448,
-      "step": 27316
-    },
-    {
-      "epoch": 14.246153846153845,
-      "grad_norm": 1.6123816967010498,
-      "learning_rate": 7.305628140703518e-05,
-      "loss": 5.2226,
-      "step": 27317
-    },
-    {
-      "epoch": 14.246675358539765,
-      "grad_norm": 1.508055567741394,
-      "learning_rate": 7.305527638190956e-05,
-      "loss": 4.6682,
-      "step": 27318
-    },
-    {
-      "epoch": 14.247196870925684,
-      "grad_norm": 1.440315842628479,
-      "learning_rate": 7.305427135678392e-05,
-      "loss": 5.3362,
-      "step": 27319
-    },
-    {
-      "epoch": 14.247718383311604,
-      "grad_norm": 1.4433022737503052,
-      "learning_rate": 7.30532663316583e-05,
-      "loss": 4.8389,
-      "step": 27320
-    },
-    {
-      "epoch": 14.248239895697523,
-      "grad_norm": 1.4750778675079346,
-      "learning_rate": 7.305226130653266e-05,
-      "loss": 5.6085,
-      "step": 27321
-    },
-    {
-      "epoch": 14.248761408083443,
-      "grad_norm": 1.4744148254394531,
-      "learning_rate": 7.305125628140704e-05,
-      "loss": 5.1633,
-      "step": 27322
-    },
-    {
-      "epoch": 14.24928292046936,
-      "grad_norm": 1.3542265892028809,
-      "learning_rate": 7.30502512562814e-05,
-      "loss": 5.6371,
-      "step": 27323
-    },
-    {
-      "epoch": 14.24980443285528,
-      "grad_norm": 1.5286089181900024,
-      "learning_rate": 7.304924623115578e-05,
-      "loss": 5.3784,
-      "step": 27324
-    },
-    {
-      "epoch": 14.2503259452412,
-      "grad_norm": 1.4177786111831665,
-      "learning_rate": 7.304824120603015e-05,
-      "loss": 5.4395,
-      "step": 27325
-    },
-    {
-      "epoch": 14.250847457627119,
-      "grad_norm": 1.440119981765747,
-      "learning_rate": 7.304723618090453e-05,
-      "loss": 5.6359,
-      "step": 27326
-    },
-    {
-      "epoch": 14.251368970013038,
-      "grad_norm": 1.4730457067489624,
-      "learning_rate": 7.30462311557789e-05,
-      "loss": 5.4333,
-      "step": 27327
-    },
-    {
-      "epoch": 14.251890482398958,
-      "grad_norm": 1.3955254554748535,
-      "learning_rate": 7.304522613065327e-05,
-      "loss": 5.2767,
-      "step": 27328
-    },
-    {
-      "epoch": 14.252411994784875,
-      "grad_norm": 1.781304121017456,
-      "learning_rate": 7.304422110552765e-05,
-      "loss": 5.2614,
-      "step": 27329
-    },
-    {
-      "epoch": 14.252933507170795,
-      "grad_norm": 1.559217095375061,
-      "learning_rate": 7.304321608040201e-05,
-      "loss": 5.4996,
-      "step": 27330
-    },
-    {
-      "epoch": 14.253455019556714,
-      "grad_norm": 1.5744870901107788,
-      "learning_rate": 7.304221105527639e-05,
-      "loss": 5.1946,
-      "step": 27331
-    },
-    {
-      "epoch": 14.253976531942634,
-      "grad_norm": 1.5614875555038452,
-      "learning_rate": 7.304120603015075e-05,
-      "loss": 5.4666,
-      "step": 27332
-    },
-    {
-      "epoch": 14.254498044328553,
-      "grad_norm": 1.4541552066802979,
-      "learning_rate": 7.304020100502513e-05,
-      "loss": 5.5444,
-      "step": 27333
-    },
-    {
-      "epoch": 14.255019556714473,
-      "grad_norm": 1.3982900381088257,
-      "learning_rate": 7.30391959798995e-05,
-      "loss": 5.4106,
-      "step": 27334
-    },
-    {
-      "epoch": 14.25554106910039,
-      "grad_norm": 1.4159704446792603,
-      "learning_rate": 7.303819095477387e-05,
-      "loss": 5.2256,
-      "step": 27335
-    },
-    {
-      "epoch": 14.25606258148631,
-      "grad_norm": 1.5504511594772339,
-      "learning_rate": 7.303718592964825e-05,
-      "loss": 4.9876,
-      "step": 27336
-    },
-    {
-      "epoch": 14.25658409387223,
-      "grad_norm": 1.4303057193756104,
-      "learning_rate": 7.303618090452263e-05,
-      "loss": 5.4683,
-      "step": 27337
-    },
-    {
-      "epoch": 14.257105606258149,
-      "grad_norm": 1.593218445777893,
-      "learning_rate": 7.303517587939699e-05,
-      "loss": 5.7061,
-      "step": 27338
-    },
-    {
-      "epoch": 14.257627118644068,
-      "grad_norm": 1.466259241104126,
-      "learning_rate": 7.303417085427136e-05,
-      "loss": 5.232,
-      "step": 27339
-    },
-    {
-      "epoch": 14.258148631029988,
-      "grad_norm": 1.4396835565567017,
-      "learning_rate": 7.303316582914573e-05,
-      "loss": 5.3998,
-      "step": 27340
-    },
-    {
-      "epoch": 14.258670143415905,
-      "grad_norm": 1.5036815404891968,
-      "learning_rate": 7.30321608040201e-05,
-      "loss": 5.3328,
-      "step": 27341
-    },
-    {
-      "epoch": 14.259191655801825,
-      "grad_norm": 1.5330265760421753,
-      "learning_rate": 7.303115577889448e-05,
-      "loss": 5.4874,
-      "step": 27342
-    },
-    {
-      "epoch": 14.259713168187744,
-      "grad_norm": 1.5450650453567505,
-      "learning_rate": 7.303015075376884e-05,
-      "loss": 5.3454,
-      "step": 27343
-    },
-    {
-      "epoch": 14.260234680573664,
-      "grad_norm": 1.4540239572525024,
-      "learning_rate": 7.302914572864322e-05,
-      "loss": 5.2745,
-      "step": 27344
-    },
-    {
-      "epoch": 14.260756192959583,
-      "grad_norm": 1.3969920873641968,
-      "learning_rate": 7.302814070351758e-05,
-      "loss": 5.5401,
-      "step": 27345
-    },
-    {
-      "epoch": 14.261277705345503,
-      "grad_norm": 1.4719725847244263,
-      "learning_rate": 7.302713567839196e-05,
-      "loss": 5.574,
-      "step": 27346
-    },
-    {
-      "epoch": 14.26179921773142,
-      "grad_norm": 1.498020052909851,
-      "learning_rate": 7.302613065326634e-05,
-      "loss": 5.1179,
-      "step": 27347
-    },
-    {
-      "epoch": 14.26232073011734,
-      "grad_norm": 1.498841404914856,
-      "learning_rate": 7.302512562814072e-05,
-      "loss": 5.1371,
-      "step": 27348
-    },
-    {
-      "epoch": 14.26284224250326,
-      "grad_norm": 1.5515542030334473,
-      "learning_rate": 7.302412060301508e-05,
-      "loss": 5.5567,
-      "step": 27349
-    },
-    {
-      "epoch": 14.263363754889179,
-      "grad_norm": 1.446483850479126,
-      "learning_rate": 7.302311557788946e-05,
-      "loss": 5.3859,
-      "step": 27350
-    },
-    {
-      "epoch": 14.263885267275098,
-      "grad_norm": 1.4374140501022339,
-      "learning_rate": 7.302211055276382e-05,
-      "loss": 5.6827,
-      "step": 27351
-    },
-    {
-      "epoch": 14.264406779661018,
-      "grad_norm": 1.5385171175003052,
-      "learning_rate": 7.302110552763819e-05,
-      "loss": 5.2828,
-      "step": 27352
-    },
-    {
-      "epoch": 14.264928292046935,
-      "grad_norm": 1.4892191886901855,
-      "learning_rate": 7.302010050251256e-05,
-      "loss": 5.5085,
-      "step": 27353
-    },
-    {
-      "epoch": 14.265449804432855,
-      "grad_norm": 1.4354424476623535,
-      "learning_rate": 7.301909547738693e-05,
-      "loss": 5.7253,
-      "step": 27354
-    },
-    {
-      "epoch": 14.265971316818774,
-      "grad_norm": 1.4219563007354736,
-      "learning_rate": 7.30180904522613e-05,
-      "loss": 5.6285,
-      "step": 27355
-    },
-    {
-      "epoch": 14.266492829204694,
-      "grad_norm": 1.527944564819336,
-      "learning_rate": 7.301708542713568e-05,
-      "loss": 5.4898,
-      "step": 27356
-    },
-    {
-      "epoch": 14.267014341590613,
-      "grad_norm": 1.3339481353759766,
-      "learning_rate": 7.301608040201006e-05,
-      "loss": 5.5757,
-      "step": 27357
-    },
-    {
-      "epoch": 14.267535853976533,
-      "grad_norm": 1.444425106048584,
-      "learning_rate": 7.301507537688443e-05,
-      "loss": 5.4457,
-      "step": 27358
-    },
-    {
-      "epoch": 14.26805736636245,
-      "grad_norm": 1.5400893688201904,
-      "learning_rate": 7.30140703517588e-05,
-      "loss": 4.9829,
-      "step": 27359
-    },
-    {
-      "epoch": 14.26857887874837,
-      "grad_norm": 1.4485074281692505,
-      "learning_rate": 7.301306532663317e-05,
-      "loss": 5.5972,
-      "step": 27360
-    },
-    {
-      "epoch": 14.26910039113429,
-      "grad_norm": 1.6577684879302979,
-      "learning_rate": 7.301206030150755e-05,
-      "loss": 5.1183,
-      "step": 27361
-    },
-    {
-      "epoch": 14.269621903520209,
-      "grad_norm": 1.5232212543487549,
-      "learning_rate": 7.301105527638191e-05,
-      "loss": 5.4756,
-      "step": 27362
-    },
-    {
-      "epoch": 14.270143415906128,
-      "grad_norm": 1.5928058624267578,
-      "learning_rate": 7.301005025125629e-05,
-      "loss": 5.4435,
-      "step": 27363
-    },
-    {
-      "epoch": 14.270664928292048,
-      "grad_norm": 1.456756830215454,
-      "learning_rate": 7.300904522613065e-05,
-      "loss": 5.4724,
-      "step": 27364
-    },
-    {
-      "epoch": 14.271186440677965,
-      "grad_norm": 1.4948891401290894,
-      "learning_rate": 7.300804020100502e-05,
-      "loss": 5.5486,
-      "step": 27365
-    },
-    {
-      "epoch": 14.271707953063885,
-      "grad_norm": 1.4610286951065063,
-      "learning_rate": 7.30070351758794e-05,
-      "loss": 4.9851,
-      "step": 27366
-    },
-    {
-      "epoch": 14.272229465449804,
-      "grad_norm": 1.5788755416870117,
-      "learning_rate": 7.300603015075377e-05,
-      "loss": 5.2745,
-      "step": 27367
-    },
-    {
-      "epoch": 14.272750977835724,
-      "grad_norm": 1.4676644802093506,
-      "learning_rate": 7.300502512562815e-05,
-      "loss": 5.0938,
-      "step": 27368
-    },
-    {
-      "epoch": 14.273272490221643,
-      "grad_norm": 1.4719160795211792,
-      "learning_rate": 7.300402010050251e-05,
-      "loss": 5.2781,
-      "step": 27369
-    },
-    {
-      "epoch": 14.273794002607563,
-      "grad_norm": 1.4142422676086426,
-      "learning_rate": 7.300301507537689e-05,
-      "loss": 4.7577,
-      "step": 27370
-    },
-    {
-      "epoch": 14.27431551499348,
-      "grad_norm": 1.4793493747711182,
-      "learning_rate": 7.300201005025126e-05,
-      "loss": 5.3576,
-      "step": 27371
-    },
-    {
-      "epoch": 14.2748370273794,
-      "grad_norm": 1.4387786388397217,
-      "learning_rate": 7.300100502512563e-05,
-      "loss": 5.8266,
-      "step": 27372
-    },
-    {
-      "epoch": 14.27535853976532,
-      "grad_norm": 1.5395426750183105,
-      "learning_rate": 7.3e-05,
-      "loss": 5.3784,
-      "step": 27373
-    },
-    {
-      "epoch": 14.275880052151239,
-      "grad_norm": 1.4667255878448486,
-      "learning_rate": 7.299899497487438e-05,
-      "loss": 5.5751,
-      "step": 27374
-    },
-    {
-      "epoch": 14.276401564537158,
-      "grad_norm": 1.5359019041061401,
-      "learning_rate": 7.299798994974874e-05,
-      "loss": 5.6446,
-      "step": 27375
-    },
-    {
-      "epoch": 14.276923076923078,
-      "grad_norm": 1.4388748407363892,
-      "learning_rate": 7.299698492462312e-05,
-      "loss": 5.4415,
-      "step": 27376
-    },
-    {
-      "epoch": 14.277444589308995,
-      "grad_norm": 1.543497920036316,
-      "learning_rate": 7.29959798994975e-05,
-      "loss": 5.1976,
-      "step": 27377
-    },
-    {
-      "epoch": 14.277966101694915,
-      "grad_norm": 1.4516485929489136,
-      "learning_rate": 7.299497487437186e-05,
-      "loss": 5.6099,
-      "step": 27378
-    },
-    {
-      "epoch": 14.278487614080834,
-      "grad_norm": 1.4513481855392456,
-      "learning_rate": 7.299396984924624e-05,
-      "loss": 4.9617,
-      "step": 27379
-    },
-    {
-      "epoch": 14.279009126466754,
-      "grad_norm": 1.4837782382965088,
-      "learning_rate": 7.29929648241206e-05,
-      "loss": 5.2316,
-      "step": 27380
-    },
-    {
-      "epoch": 14.279530638852673,
-      "grad_norm": 1.4634177684783936,
-      "learning_rate": 7.299195979899498e-05,
-      "loss": 5.1693,
-      "step": 27381
-    },
-    {
-      "epoch": 14.280052151238593,
-      "grad_norm": 1.5808354616165161,
-      "learning_rate": 7.299095477386935e-05,
-      "loss": 5.1263,
-      "step": 27382
-    },
-    {
-      "epoch": 14.28057366362451,
-      "grad_norm": 1.4774978160858154,
-      "learning_rate": 7.298994974874372e-05,
-      "loss": 5.3355,
-      "step": 27383
-    },
-    {
-      "epoch": 14.28109517601043,
-      "grad_norm": 1.4695812463760376,
-      "learning_rate": 7.298894472361809e-05,
-      "loss": 4.8298,
-      "step": 27384
-    },
-    {
-      "epoch": 14.28161668839635,
-      "grad_norm": 1.3788570165634155,
-      "learning_rate": 7.298793969849247e-05,
-      "loss": 5.6672,
-      "step": 27385
-    },
-    {
-      "epoch": 14.282138200782269,
-      "grad_norm": 1.3766332864761353,
-      "learning_rate": 7.298693467336683e-05,
-      "loss": 5.5087,
-      "step": 27386
-    },
-    {
-      "epoch": 14.282659713168188,
-      "grad_norm": 1.4634814262390137,
-      "learning_rate": 7.298592964824121e-05,
-      "loss": 5.4449,
-      "step": 27387
-    },
-    {
-      "epoch": 14.283181225554108,
-      "grad_norm": 1.4889185428619385,
-      "learning_rate": 7.298492462311559e-05,
-      "loss": 5.5199,
-      "step": 27388
-    },
-    {
-      "epoch": 14.283702737940025,
-      "grad_norm": 1.3976187705993652,
-      "learning_rate": 7.298391959798996e-05,
-      "loss": 5.766,
-      "step": 27389
-    },
-    {
-      "epoch": 14.284224250325945,
-      "grad_norm": 1.4465142488479614,
-      "learning_rate": 7.298291457286433e-05,
-      "loss": 5.7675,
-      "step": 27390
-    },
-    {
-      "epoch": 14.284745762711864,
-      "grad_norm": 1.5418596267700195,
-      "learning_rate": 7.298190954773869e-05,
-      "loss": 5.0603,
-      "step": 27391
-    },
-    {
-      "epoch": 14.285267275097784,
-      "grad_norm": 1.5821375846862793,
-      "learning_rate": 7.298090452261307e-05,
-      "loss": 4.8062,
-      "step": 27392
-    },
-    {
-      "epoch": 14.285788787483703,
-      "grad_norm": Infinity,
-      "learning_rate": 7.298090452261307e-05,
-      "loss": 5.8957,
-      "step": 27393
-    },
-    {
-      "epoch": 14.286310299869623,
-      "grad_norm": 1.429723858833313,
-      "learning_rate": 7.297989949748743e-05,
-      "loss": 4.5338,
-      "step": 27394
-    },
-    {
-      "epoch": 14.28683181225554,
-      "grad_norm": 1.4929630756378174,
-      "learning_rate": 7.297889447236181e-05,
-      "loss": 5.3426,
-      "step": 27395
-    },
-    {
-      "epoch": 14.28735332464146,
-      "grad_norm": 1.4872736930847168,
-      "learning_rate": 7.297788944723618e-05,
-      "loss": 5.5494,
-      "step": 27396
-    },
-    {
-      "epoch": 14.28787483702738,
-      "grad_norm": 1.4138108491897583,
-      "learning_rate": 7.297688442211055e-05,
-      "loss": 5.6932,
-      "step": 27397
-    },
-    {
-      "epoch": 14.288396349413299,
-      "grad_norm": 1.7120633125305176,
-      "learning_rate": 7.297587939698493e-05,
-      "loss": 5.008,
-      "step": 27398
-    },
-    {
-      "epoch": 14.288917861799218,
-      "grad_norm": 1.3868367671966553,
-      "learning_rate": 7.297487437185931e-05,
-      "loss": 5.2364,
-      "step": 27399
-    },
-    {
-      "epoch": 14.289439374185136,
-      "grad_norm": 1.5802234411239624,
-      "learning_rate": 7.297386934673367e-05,
-      "loss": 5.1803,
-      "step": 27400
-    },
-    {
-      "epoch": 14.289960886571055,
-      "grad_norm": 1.4652255773544312,
-      "learning_rate": 7.297286432160805e-05,
-      "loss": 5.0667,
-      "step": 27401
-    },
-    {
-      "epoch": 14.290482398956975,
-      "grad_norm": 1.4111429452896118,
-      "learning_rate": 7.297185929648242e-05,
-      "loss": 5.5551,
-      "step": 27402
-    },
-    {
-      "epoch": 14.291003911342894,
-      "grad_norm": 1.5009095668792725,
-      "learning_rate": 7.29708542713568e-05,
-      "loss": 5.2578,
-      "step": 27403
-    },
-    {
-      "epoch": 14.291525423728814,
-      "grad_norm": 1.4413847923278809,
-      "learning_rate": 7.296984924623116e-05,
-      "loss": 5.4794,
-      "step": 27404
-    },
-    {
-      "epoch": 14.292046936114733,
-      "grad_norm": 1.590943694114685,
-      "learning_rate": 7.296884422110552e-05,
-      "loss": 4.8703,
-      "step": 27405
-    },
-    {
-      "epoch": 14.292568448500653,
-      "grad_norm": 1.5355826616287231,
-      "learning_rate": 7.29678391959799e-05,
-      "loss": 4.9567,
-      "step": 27406
-    },
-    {
-      "epoch": 14.29308996088657,
-      "grad_norm": 1.5388612747192383,
-      "learning_rate": 7.296683417085426e-05,
-      "loss": 4.5651,
-      "step": 27407
-    },
-    {
-      "epoch": 14.29361147327249,
-      "grad_norm": 1.4259611368179321,
-      "learning_rate": 7.296582914572864e-05,
-      "loss": 5.313,
-      "step": 27408
-    },
-    {
-      "epoch": 14.29413298565841,
-      "grad_norm": 1.4333133697509766,
-      "learning_rate": 7.296482412060302e-05,
-      "loss": 5.5652,
-      "step": 27409
-    },
-    {
-      "epoch": 14.294654498044329,
-      "grad_norm": 1.3938486576080322,
-      "learning_rate": 7.29638190954774e-05,
-      "loss": 5.2156,
-      "step": 27410
-    },
-    {
-      "epoch": 14.295176010430248,
-      "grad_norm": 1.4745677709579468,
-      "learning_rate": 7.296281407035176e-05,
-      "loss": 5.5155,
-      "step": 27411
-    },
-    {
-      "epoch": 14.295697522816166,
-      "grad_norm": 1.3446909189224243,
-      "learning_rate": 7.296180904522614e-05,
-      "loss": 5.62,
-      "step": 27412
-    },
-    {
-      "epoch": 14.296219035202085,
-      "grad_norm": 1.4755216836929321,
-      "learning_rate": 7.29608040201005e-05,
-      "loss": 4.6842,
-      "step": 27413
-    },
-    {
-      "epoch": 14.296740547588005,
-      "grad_norm": 1.4578700065612793,
-      "learning_rate": 7.295979899497488e-05,
-      "loss": 5.2694,
-      "step": 27414
-    },
-    {
-      "epoch": 14.297262059973924,
-      "grad_norm": 1.6066657304763794,
-      "learning_rate": 7.295879396984925e-05,
-      "loss": 5.1118,
-      "step": 27415
-    },
-    {
-      "epoch": 14.297783572359844,
-      "grad_norm": 1.4480838775634766,
-      "learning_rate": 7.295778894472362e-05,
-      "loss": 5.5573,
-      "step": 27416
-    },
-    {
-      "epoch": 14.298305084745763,
-      "grad_norm": 1.5230098962783813,
-      "learning_rate": 7.295678391959799e-05,
-      "loss": 4.9817,
-      "step": 27417
-    },
-    {
-      "epoch": 14.298826597131681,
-      "grad_norm": 1.4324089288711548,
-      "learning_rate": 7.295577889447237e-05,
-      "loss": 5.7656,
-      "step": 27418
-    },
-    {
-      "epoch": 14.2993481095176,
-      "grad_norm": 1.423113226890564,
-      "learning_rate": 7.295477386934674e-05,
-      "loss": 5.7986,
-      "step": 27419
-    },
-    {
-      "epoch": 14.29986962190352,
-      "grad_norm": 1.5300703048706055,
-      "learning_rate": 7.295376884422111e-05,
-      "loss": 5.3327,
-      "step": 27420
-    },
-    {
-      "epoch": 14.30039113428944,
-      "grad_norm": 1.4866975545883179,
-      "learning_rate": 7.295276381909549e-05,
-      "loss": 5.4357,
-      "step": 27421
-    },
-    {
-      "epoch": 14.300912646675359,
-      "grad_norm": 1.5817064046859741,
-      "learning_rate": 7.295175879396985e-05,
-      "loss": 5.3046,
-      "step": 27422
-    },
-    {
-      "epoch": 14.301434159061278,
-      "grad_norm": 1.4705170392990112,
-      "learning_rate": 7.295075376884423e-05,
-      "loss": 5.3935,
-      "step": 27423
-    },
-    {
-      "epoch": 14.301955671447196,
-      "grad_norm": 1.5175158977508545,
-      "learning_rate": 7.294974874371859e-05,
-      "loss": 5.3295,
-      "step": 27424
-    },
-    {
-      "epoch": 14.302477183833116,
-      "grad_norm": 1.4500956535339355,
-      "learning_rate": 7.294874371859297e-05,
-      "loss": 5.071,
-      "step": 27425
-    },
-    {
-      "epoch": 14.302998696219035,
-      "grad_norm": 1.4936840534210205,
-      "learning_rate": 7.294773869346733e-05,
-      "loss": 5.5144,
-      "step": 27426
-    },
-    {
-      "epoch": 14.303520208604954,
-      "grad_norm": 1.461538553237915,
-      "learning_rate": 7.294673366834171e-05,
-      "loss": 5.7307,
-      "step": 27427
-    },
-    {
-      "epoch": 14.304041720990874,
-      "grad_norm": 1.3690671920776367,
-      "learning_rate": 7.294572864321608e-05,
-      "loss": 4.5548,
-      "step": 27428
-    },
-    {
-      "epoch": 14.304563233376793,
-      "grad_norm": 1.5518805980682373,
-      "learning_rate": 7.294472361809045e-05,
-      "loss": 5.6204,
-      "step": 27429
-    },
-    {
-      "epoch": 14.305084745762711,
-      "grad_norm": 1.3631479740142822,
-      "learning_rate": 7.294371859296483e-05,
-      "loss": 5.6889,
-      "step": 27430
-    },
-    {
-      "epoch": 14.30560625814863,
-      "grad_norm": 1.4839521646499634,
-      "learning_rate": 7.294271356783921e-05,
-      "loss": 5.8543,
-      "step": 27431
-    },
-    {
-      "epoch": 14.30612777053455,
-      "grad_norm": 1.4553381204605103,
-      "learning_rate": 7.294170854271357e-05,
-      "loss": 4.9997,
-      "step": 27432
-    },
-    {
-      "epoch": 14.30664928292047,
-      "grad_norm": 1.4323261976242065,
-      "learning_rate": 7.294070351758794e-05,
-      "loss": 5.2329,
-      "step": 27433
-    },
-    {
-      "epoch": 14.307170795306389,
-      "grad_norm": 1.451188087463379,
-      "learning_rate": 7.293969849246232e-05,
-      "loss": 5.55,
-      "step": 27434
-    },
-    {
-      "epoch": 14.307692307692308,
-      "grad_norm": 1.3303563594818115,
-      "learning_rate": 7.293869346733668e-05,
-      "loss": 5.9226,
-      "step": 27435
-    },
-    {
-      "epoch": 14.308213820078226,
-      "grad_norm": 1.4706032276153564,
-      "learning_rate": 7.293768844221106e-05,
-      "loss": 5.2359,
-      "step": 27436
-    },
-    {
-      "epoch": 14.308735332464146,
-      "grad_norm": 1.4650769233703613,
-      "learning_rate": 7.293668341708542e-05,
-      "loss": 5.2461,
-      "step": 27437
-    },
-    {
-      "epoch": 14.309256844850065,
-      "grad_norm": 1.5248686075210571,
-      "learning_rate": 7.29356783919598e-05,
-      "loss": 5.5307,
-      "step": 27438
-    },
-    {
-      "epoch": 14.309778357235984,
-      "grad_norm": 1.4946792125701904,
-      "learning_rate": 7.293467336683418e-05,
-      "loss": 5.0137,
-      "step": 27439
-    },
-    {
-      "epoch": 14.310299869621904,
-      "grad_norm": 1.6461896896362305,
-      "learning_rate": 7.293366834170856e-05,
-      "loss": 4.8529,
-      "step": 27440
-    },
-    {
-      "epoch": 14.310821382007823,
-      "grad_norm": 1.4075337648391724,
-      "learning_rate": 7.293266331658292e-05,
-      "loss": 5.4802,
-      "step": 27441
-    },
-    {
-      "epoch": 14.311342894393741,
-      "grad_norm": 1.5433053970336914,
-      "learning_rate": 7.29316582914573e-05,
-      "loss": 5.3915,
-      "step": 27442
-    },
-    {
-      "epoch": 14.31186440677966,
-      "grad_norm": 1.578058123588562,
-      "learning_rate": 7.293065326633166e-05,
-      "loss": 5.2357,
-      "step": 27443
-    },
-    {
-      "epoch": 14.31238591916558,
-      "grad_norm": 1.4983476400375366,
-      "learning_rate": 7.292964824120604e-05,
-      "loss": 5.078,
-      "step": 27444
-    },
-    {
-      "epoch": 14.3129074315515,
-      "grad_norm": 1.4702011346817017,
-      "learning_rate": 7.29286432160804e-05,
-      "loss": 5.3803,
-      "step": 27445
-    },
-    {
-      "epoch": 14.313428943937419,
-      "grad_norm": 1.4907379150390625,
-      "learning_rate": 7.292763819095477e-05,
-      "loss": 5.4359,
-      "step": 27446
-    },
-    {
-      "epoch": 14.313950456323338,
-      "grad_norm": 1.3938888311386108,
-      "learning_rate": 7.292663316582915e-05,
-      "loss": 5.6552,
-      "step": 27447
-    },
-    {
-      "epoch": 14.314471968709256,
-      "grad_norm": 1.4328768253326416,
-      "learning_rate": 7.292562814070351e-05,
-      "loss": 5.1002,
-      "step": 27448
-    },
-    {
-      "epoch": 14.314993481095176,
-      "grad_norm": 1.429824709892273,
-      "learning_rate": 7.292462311557789e-05,
-      "loss": 5.5015,
-      "step": 27449
-    },
-    {
-      "epoch": 14.315514993481095,
-      "grad_norm": 1.4663156270980835,
-      "learning_rate": 7.292361809045227e-05,
-      "loss": 5.5363,
-      "step": 27450
-    },
-    {
-      "epoch": 14.316036505867014,
-      "grad_norm": 1.827901005744934,
-      "learning_rate": 7.292261306532664e-05,
-      "loss": 5.2194,
-      "step": 27451
-    },
-    {
-      "epoch": 14.316558018252934,
-      "grad_norm": 1.622591257095337,
-      "learning_rate": 7.292160804020101e-05,
-      "loss": 5.0618,
-      "step": 27452
-    },
-    {
-      "epoch": 14.317079530638853,
-      "grad_norm": 1.586893081665039,
-      "learning_rate": 7.292060301507539e-05,
-      "loss": 5.3969,
-      "step": 27453
-    },
-    {
-      "epoch": 14.317601043024771,
-      "grad_norm": 1.5023853778839111,
-      "learning_rate": 7.291959798994975e-05,
-      "loss": 5.1119,
-      "step": 27454
-    },
-    {
-      "epoch": 14.31812255541069,
-      "grad_norm": 1.409922480583191,
-      "learning_rate": 7.291859296482413e-05,
-      "loss": 5.1566,
-      "step": 27455
-    },
-    {
-      "epoch": 14.31864406779661,
-      "grad_norm": 1.5489652156829834,
-      "learning_rate": 7.291758793969849e-05,
-      "loss": 5.5798,
-      "step": 27456
-    },
-    {
-      "epoch": 14.31916558018253,
-      "grad_norm": 1.538278341293335,
-      "learning_rate": 7.291658291457287e-05,
-      "loss": 5.0624,
-      "step": 27457
-    },
-    {
-      "epoch": 14.319687092568449,
-      "grad_norm": 1.545274019241333,
-      "learning_rate": 7.291557788944724e-05,
-      "loss": 5.4518,
-      "step": 27458
-    },
-    {
-      "epoch": 14.320208604954368,
-      "grad_norm": 1.5135568380355835,
-      "learning_rate": 7.291457286432161e-05,
-      "loss": 4.8557,
-      "step": 27459
-    },
-    {
-      "epoch": 14.320730117340286,
-      "grad_norm": 1.5176647901535034,
-      "learning_rate": 7.291356783919599e-05,
-      "loss": 5.1626,
-      "step": 27460
-    },
-    {
-      "epoch": 14.321251629726206,
-      "grad_norm": 1.4832367897033691,
-      "learning_rate": 7.291256281407036e-05,
-      "loss": 4.8578,
-      "step": 27461
-    },
-    {
-      "epoch": 14.321773142112125,
-      "grad_norm": 1.5253783464431763,
-      "learning_rate": 7.291155778894473e-05,
-      "loss": 5.4127,
-      "step": 27462
-    },
-    {
-      "epoch": 14.322294654498045,
-      "grad_norm": 1.4345839023590088,
-      "learning_rate": 7.29105527638191e-05,
-      "loss": 5.7144,
-      "step": 27463
-    },
-    {
-      "epoch": 14.322816166883964,
-      "grad_norm": 1.4969979524612427,
-      "learning_rate": 7.290954773869347e-05,
-      "loss": 5.7417,
-      "step": 27464
-    },
-    {
-      "epoch": 14.323337679269883,
-      "grad_norm": 1.6141074895858765,
-      "learning_rate": 7.290854271356784e-05,
-      "loss": 4.8035,
-      "step": 27465
-    },
-    {
-      "epoch": 14.323859191655801,
-      "grad_norm": 1.488839864730835,
-      "learning_rate": 7.290753768844222e-05,
-      "loss": 5.4275,
-      "step": 27466
-    },
-    {
-      "epoch": 14.32438070404172,
-      "grad_norm": 1.479506015777588,
-      "learning_rate": 7.290653266331658e-05,
-      "loss": 5.2807,
-      "step": 27467
-    },
-    {
-      "epoch": 14.32490221642764,
-      "grad_norm": 1.498016595840454,
-      "learning_rate": 7.290552763819096e-05,
-      "loss": 5.3197,
-      "step": 27468
-    },
-    {
-      "epoch": 14.32542372881356,
-      "grad_norm": 1.457808494567871,
-      "learning_rate": 7.290452261306532e-05,
-      "loss": 4.8532,
-      "step": 27469
-    },
-    {
-      "epoch": 14.325945241199479,
-      "grad_norm": 1.4311151504516602,
-      "learning_rate": 7.29035175879397e-05,
-      "loss": 5.9212,
-      "step": 27470
-    },
-    {
-      "epoch": 14.326466753585398,
-      "grad_norm": 1.4903500080108643,
-      "learning_rate": 7.290251256281408e-05,
-      "loss": 5.3105,
-      "step": 27471
-    },
-    {
-      "epoch": 14.326988265971316,
-      "grad_norm": 1.509214997291565,
-      "learning_rate": 7.290150753768844e-05,
-      "loss": 5.3877,
-      "step": 27472
-    },
-    {
-      "epoch": 14.327509778357236,
-      "grad_norm": 1.3766292333602905,
-      "learning_rate": 7.290050251256282e-05,
-      "loss": 5.4578,
-      "step": 27473
-    },
-    {
-      "epoch": 14.328031290743155,
-      "grad_norm": 1.365463376045227,
-      "learning_rate": 7.289949748743719e-05,
-      "loss": 5.6792,
-      "step": 27474
-    },
-    {
-      "epoch": 14.328552803129075,
-      "grad_norm": 1.3935855627059937,
-      "learning_rate": 7.289849246231156e-05,
-      "loss": 5.5715,
-      "step": 27475
-    },
-    {
-      "epoch": 14.329074315514994,
-      "grad_norm": 1.4434998035430908,
-      "learning_rate": 7.289748743718593e-05,
-      "loss": 5.431,
-      "step": 27476
-    },
-    {
-      "epoch": 14.329595827900913,
-      "grad_norm": 1.3183585405349731,
-      "learning_rate": 7.28964824120603e-05,
-      "loss": 5.6513,
-      "step": 27477
-    },
-    {
-      "epoch": 14.330117340286831,
-      "grad_norm": 1.519308090209961,
-      "learning_rate": 7.289547738693467e-05,
-      "loss": 5.0445,
-      "step": 27478
-    },
-    {
-      "epoch": 14.33063885267275,
-      "grad_norm": 1.454529881477356,
-      "learning_rate": 7.289447236180905e-05,
-      "loss": 5.4571,
-      "step": 27479
-    },
-    {
-      "epoch": 14.33116036505867,
-      "grad_norm": 1.371730089187622,
-      "learning_rate": 7.289346733668343e-05,
-      "loss": 5.4375,
-      "step": 27480
-    },
-    {
-      "epoch": 14.33168187744459,
-      "grad_norm": 1.4735101461410522,
-      "learning_rate": 7.28924623115578e-05,
-      "loss": 5.2869,
-      "step": 27481
-    },
-    {
-      "epoch": 14.332203389830509,
-      "grad_norm": 1.4615155458450317,
-      "learning_rate": 7.289145728643217e-05,
-      "loss": 5.3303,
-      "step": 27482
-    },
-    {
-      "epoch": 14.332724902216428,
-      "grad_norm": 1.3872510194778442,
-      "learning_rate": 7.289045226130655e-05,
-      "loss": 5.6194,
-      "step": 27483
-    },
-    {
-      "epoch": 14.333246414602346,
-      "grad_norm": 1.4903813600540161,
-      "learning_rate": 7.288944723618091e-05,
-      "loss": 5.2321,
-      "step": 27484
-    },
-    {
-      "epoch": 14.333767926988266,
-      "grad_norm": 1.5123133659362793,
-      "learning_rate": 7.288844221105527e-05,
-      "loss": 5.5484,
-      "step": 27485
-    },
-    {
-      "epoch": 14.334289439374185,
-      "grad_norm": 1.626212239265442,
-      "learning_rate": 7.288743718592965e-05,
-      "loss": 5.5409,
-      "step": 27486
-    },
-    {
-      "epoch": 14.334810951760105,
-      "grad_norm": 1.425682783126831,
-      "learning_rate": 7.288643216080402e-05,
-      "loss": 5.5576,
-      "step": 27487
-    },
-    {
-      "epoch": 14.335332464146024,
-      "grad_norm": 1.4601831436157227,
-      "learning_rate": 7.28854271356784e-05,
-      "loss": 5.2305,
-      "step": 27488
-    },
-    {
-      "epoch": 14.335853976531943,
-      "grad_norm": 1.6676387786865234,
-      "learning_rate": 7.288442211055276e-05,
-      "loss": 5.4868,
-      "step": 27489
-    },
-    {
-      "epoch": 14.336375488917861,
-      "grad_norm": 1.478464126586914,
-      "learning_rate": 7.288341708542714e-05,
-      "loss": 5.0543,
-      "step": 27490
-    },
-    {
-      "epoch": 14.33689700130378,
-      "grad_norm": 1.5951961278915405,
-      "learning_rate": 7.288241206030151e-05,
-      "loss": 5.4372,
-      "step": 27491
-    },
-    {
-      "epoch": 14.3374185136897,
-      "grad_norm": 1.4223047494888306,
-      "learning_rate": 7.288140703517589e-05,
-      "loss": 5.3052,
-      "step": 27492
-    },
-    {
-      "epoch": 14.33794002607562,
-      "grad_norm": 1.5344531536102295,
-      "learning_rate": 7.288040201005026e-05,
-      "loss": 5.7386,
-      "step": 27493
-    },
-    {
-      "epoch": 14.338461538461539,
-      "grad_norm": 1.3639038801193237,
-      "learning_rate": 7.287939698492463e-05,
-      "loss": 5.6676,
-      "step": 27494
-    },
-    {
-      "epoch": 14.338983050847457,
-      "grad_norm": 1.3661466836929321,
-      "learning_rate": 7.2878391959799e-05,
-      "loss": 5.5521,
-      "step": 27495
-    },
-    {
-      "epoch": 14.339504563233376,
-      "grad_norm": 1.4507789611816406,
-      "learning_rate": 7.287738693467338e-05,
-      "loss": 5.5652,
-      "step": 27496
-    },
-    {
-      "epoch": 14.340026075619296,
-      "grad_norm": 1.3794257640838623,
-      "learning_rate": 7.287638190954774e-05,
-      "loss": 5.1714,
-      "step": 27497
-    },
-    {
-      "epoch": 14.340547588005215,
-      "grad_norm": 1.5496329069137573,
-      "learning_rate": 7.28753768844221e-05,
-      "loss": 4.9416,
-      "step": 27498
-    },
-    {
-      "epoch": 14.341069100391135,
-      "grad_norm": 1.4350783824920654,
-      "learning_rate": 7.287437185929648e-05,
-      "loss": 5.2649,
-      "step": 27499
-    },
-    {
-      "epoch": 14.341590612777054,
-      "grad_norm": 1.4015618562698364,
-      "learning_rate": 7.287336683417085e-05,
-      "loss": 5.4201,
-      "step": 27500
-    },
-    {
-      "epoch": 14.342112125162974,
-      "grad_norm": 1.4831870794296265,
-      "learning_rate": 7.287236180904522e-05,
-      "loss": 5.4297,
-      "step": 27501
-    },
-    {
-      "epoch": 14.342633637548891,
-      "grad_norm": 1.5947537422180176,
-      "learning_rate": 7.28713567839196e-05,
-      "loss": 5.2367,
-      "step": 27502
-    },
-    {
-      "epoch": 14.34315514993481,
-      "grad_norm": 1.4614640474319458,
-      "learning_rate": 7.287035175879398e-05,
-      "loss": 5.1747,
-      "step": 27503
-    },
-    {
-      "epoch": 14.34367666232073,
-      "grad_norm": 1.3973021507263184,
-      "learning_rate": 7.286934673366834e-05,
-      "loss": 5.7724,
-      "step": 27504
-    },
-    {
-      "epoch": 14.34419817470665,
-      "grad_norm": 1.549750804901123,
-      "learning_rate": 7.286834170854272e-05,
-      "loss": 5.1747,
-      "step": 27505
-    },
-    {
-      "epoch": 14.344719687092569,
-      "grad_norm": 1.4175962209701538,
-      "learning_rate": 7.286733668341709e-05,
-      "loss": 5.4403,
-      "step": 27506
-    },
-    {
-      "epoch": 14.345241199478487,
-      "grad_norm": 1.4120837450027466,
-      "learning_rate": 7.286633165829146e-05,
-      "loss": 5.5791,
-      "step": 27507
-    },
-    {
-      "epoch": 14.345762711864406,
-      "grad_norm": 1.5028983354568481,
-      "learning_rate": 7.286532663316583e-05,
-      "loss": 5.1646,
-      "step": 27508
-    },
-    {
-      "epoch": 14.346284224250326,
-      "grad_norm": 1.467103123664856,
-      "learning_rate": 7.28643216080402e-05,
-      "loss": 5.6387,
-      "step": 27509
-    },
-    {
-      "epoch": 14.346805736636245,
-      "grad_norm": 1.6227364540100098,
-      "learning_rate": 7.286331658291457e-05,
-      "loss": 5.0672,
-      "step": 27510
-    },
-    {
-      "epoch": 14.347327249022165,
-      "grad_norm": 1.4558281898498535,
-      "learning_rate": 7.286231155778895e-05,
-      "loss": 5.2751,
-      "step": 27511
-    },
-    {
-      "epoch": 14.347848761408084,
-      "grad_norm": 1.3199410438537598,
-      "learning_rate": 7.286130653266333e-05,
-      "loss": 5.8937,
-      "step": 27512
-    },
-    {
-      "epoch": 14.348370273794002,
-      "grad_norm": 1.5531054735183716,
-      "learning_rate": 7.286030150753769e-05,
-      "loss": 5.2239,
-      "step": 27513
-    },
-    {
-      "epoch": 14.348891786179921,
-      "grad_norm": 1.5203312635421753,
-      "learning_rate": 7.285929648241207e-05,
-      "loss": 5.3984,
-      "step": 27514
-    },
-    {
-      "epoch": 14.34941329856584,
-      "grad_norm": 1.5603739023208618,
-      "learning_rate": 7.285829145728643e-05,
-      "loss": 5.1807,
-      "step": 27515
-    },
-    {
-      "epoch": 14.34993481095176,
-      "grad_norm": 1.48141348361969,
-      "learning_rate": 7.285728643216081e-05,
-      "loss": 5.1576,
-      "step": 27516
-    },
-    {
-      "epoch": 14.35045632333768,
-      "grad_norm": 1.3907461166381836,
-      "learning_rate": 7.285628140703517e-05,
-      "loss": 5.6045,
-      "step": 27517
-    },
-    {
-      "epoch": 14.350977835723599,
-      "grad_norm": 1.4803475141525269,
-      "learning_rate": 7.285527638190955e-05,
-      "loss": 5.4596,
-      "step": 27518
-    },
-    {
-      "epoch": 14.351499348109517,
-      "grad_norm": 1.5468178987503052,
-      "learning_rate": 7.285427135678392e-05,
-      "loss": 5.8264,
-      "step": 27519
-    },
-    {
-      "epoch": 14.352020860495436,
-      "grad_norm": 1.394327998161316,
-      "learning_rate": 7.28532663316583e-05,
-      "loss": 4.6619,
-      "step": 27520
-    },
-    {
-      "epoch": 14.352542372881356,
-      "grad_norm": 1.4731104373931885,
-      "learning_rate": 7.285226130653266e-05,
-      "loss": 5.1754,
-      "step": 27521
-    },
-    {
-      "epoch": 14.353063885267275,
-      "grad_norm": 1.5594784021377563,
-      "learning_rate": 7.285125628140704e-05,
-      "loss": 5.2541,
-      "step": 27522
-    },
-    {
-      "epoch": 14.353585397653195,
-      "grad_norm": 1.51421058177948,
-      "learning_rate": 7.285025125628141e-05,
-      "loss": 5.2421,
-      "step": 27523
-    },
-    {
-      "epoch": 14.354106910039114,
-      "grad_norm": 1.5547796487808228,
-      "learning_rate": 7.284924623115579e-05,
-      "loss": 5.1559,
-      "step": 27524
-    },
-    {
-      "epoch": 14.354628422425032,
-      "grad_norm": 1.439234733581543,
-      "learning_rate": 7.284824120603016e-05,
-      "loss": 5.6476,
-      "step": 27525
-    },
-    {
-      "epoch": 14.355149934810951,
-      "grad_norm": 1.5041128396987915,
-      "learning_rate": 7.284723618090452e-05,
-      "loss": 5.4048,
-      "step": 27526
-    },
-    {
-      "epoch": 14.35567144719687,
-      "grad_norm": 1.4282424449920654,
-      "learning_rate": 7.28462311557789e-05,
-      "loss": 4.9093,
-      "step": 27527
-    },
-    {
-      "epoch": 14.35619295958279,
-      "grad_norm": 1.970528483390808,
-      "learning_rate": 7.284522613065326e-05,
-      "loss": 5.3817,
-      "step": 27528
-    },
-    {
-      "epoch": 14.35671447196871,
-      "grad_norm": 1.387695074081421,
-      "learning_rate": 7.284422110552764e-05,
-      "loss": 5.6351,
-      "step": 27529
-    },
-    {
-      "epoch": 14.357235984354629,
-      "grad_norm": 1.4363348484039307,
-      "learning_rate": 7.2843216080402e-05,
-      "loss": 5.083,
-      "step": 27530
-    },
-    {
-      "epoch": 14.357757496740547,
-      "grad_norm": 1.3983062505722046,
-      "learning_rate": 7.284221105527638e-05,
-      "loss": 5.615,
-      "step": 27531
-    },
-    {
-      "epoch": 14.358279009126466,
-      "grad_norm": 1.4668991565704346,
-      "learning_rate": 7.284120603015076e-05,
-      "loss": 5.3182,
-      "step": 27532
-    },
-    {
-      "epoch": 14.358800521512386,
-      "grad_norm": 1.4466382265090942,
-      "learning_rate": 7.284020100502514e-05,
-      "loss": 5.2277,
-      "step": 27533
-    },
-    {
-      "epoch": 14.359322033898305,
-      "grad_norm": 1.5520015954971313,
-      "learning_rate": 7.28391959798995e-05,
-      "loss": 5.5486,
-      "step": 27534
-    },
-    {
-      "epoch": 14.359843546284225,
-      "grad_norm": 1.4535212516784668,
-      "learning_rate": 7.283819095477388e-05,
-      "loss": 5.2633,
-      "step": 27535
-    },
-    {
-      "epoch": 14.360365058670144,
-      "grad_norm": 1.4340425729751587,
-      "learning_rate": 7.283718592964824e-05,
-      "loss": 5.7504,
-      "step": 27536
-    },
-    {
-      "epoch": 14.360886571056062,
-      "grad_norm": 1.3371672630310059,
-      "learning_rate": 7.283618090452262e-05,
-      "loss": 5.4765,
-      "step": 27537
-    },
-    {
-      "epoch": 14.361408083441981,
-      "grad_norm": 1.5550291538238525,
-      "learning_rate": 7.283517587939699e-05,
-      "loss": 4.9285,
-      "step": 27538
-    },
-    {
-      "epoch": 14.3619295958279,
-      "grad_norm": 1.5957330465316772,
-      "learning_rate": 7.283417085427135e-05,
-      "loss": 4.7543,
-      "step": 27539
-    },
-    {
-      "epoch": 14.36245110821382,
-      "grad_norm": 1.4267586469650269,
-      "learning_rate": 7.283316582914573e-05,
-      "loss": 5.5164,
-      "step": 27540
-    },
-    {
-      "epoch": 14.36297262059974,
-      "grad_norm": 1.5070831775665283,
-      "learning_rate": 7.28321608040201e-05,
-      "loss": 5.3444,
-      "step": 27541
-    },
-    {
-      "epoch": 14.36349413298566,
-      "grad_norm": 1.6141208410263062,
-      "learning_rate": 7.283115577889447e-05,
-      "loss": 5.4401,
-      "step": 27542
-    },
-    {
-      "epoch": 14.364015645371577,
-      "grad_norm": 1.394382119178772,
-      "learning_rate": 7.283015075376885e-05,
-      "loss": 5.7596,
-      "step": 27543
-    },
-    {
-      "epoch": 14.364537157757496,
-      "grad_norm": 1.5034767389297485,
-      "learning_rate": 7.282914572864323e-05,
-      "loss": 5.3179,
-      "step": 27544
-    },
-    {
-      "epoch": 14.365058670143416,
-      "grad_norm": 1.5045875310897827,
-      "learning_rate": 7.282814070351759e-05,
-      "loss": 5.1891,
-      "step": 27545
-    },
-    {
-      "epoch": 14.365580182529335,
-      "grad_norm": 1.3071645498275757,
-      "learning_rate": 7.282713567839197e-05,
-      "loss": 5.1045,
-      "step": 27546
-    },
-    {
-      "epoch": 14.366101694915255,
-      "grad_norm": 1.3774950504302979,
-      "learning_rate": 7.282613065326633e-05,
-      "loss": 5.7486,
-      "step": 27547
-    },
-    {
-      "epoch": 14.366623207301174,
-      "grad_norm": 1.5498045682907104,
-      "learning_rate": 7.282512562814071e-05,
-      "loss": 5.0776,
-      "step": 27548
-    },
-    {
-      "epoch": 14.367144719687092,
-      "grad_norm": 1.5274643898010254,
-      "learning_rate": 7.282412060301508e-05,
-      "loss": 5.1576,
-      "step": 27549
-    },
-    {
-      "epoch": 14.367666232073011,
-      "grad_norm": 1.5538990497589111,
-      "learning_rate": 7.282311557788945e-05,
-      "loss": 4.9033,
-      "step": 27550
-    },
-    {
-      "epoch": 14.36818774445893,
-      "grad_norm": 1.4743868112564087,
-      "learning_rate": 7.282211055276382e-05,
-      "loss": 5.5234,
-      "step": 27551
-    },
-    {
-      "epoch": 14.36870925684485,
-      "grad_norm": 1.4400460720062256,
-      "learning_rate": 7.28211055276382e-05,
-      "loss": 5.6224,
-      "step": 27552
-    },
-    {
-      "epoch": 14.36923076923077,
-      "grad_norm": 1.470577597618103,
-      "learning_rate": 7.282010050251257e-05,
-      "loss": 5.4475,
-      "step": 27553
-    },
-    {
-      "epoch": 14.36975228161669,
-      "grad_norm": 1.4079021215438843,
-      "learning_rate": 7.281909547738694e-05,
-      "loss": 5.3497,
-      "step": 27554
-    },
-    {
-      "epoch": 14.370273794002607,
-      "grad_norm": 1.455165982246399,
-      "learning_rate": 7.281809045226132e-05,
-      "loss": 5.4691,
-      "step": 27555
-    },
-    {
-      "epoch": 14.370795306388526,
-      "grad_norm": 1.6710424423217773,
-      "learning_rate": 7.281708542713568e-05,
-      "loss": 4.7776,
-      "step": 27556
-    },
-    {
-      "epoch": 14.371316818774446,
-      "grad_norm": 1.60850191116333,
-      "learning_rate": 7.281608040201006e-05,
-      "loss": 4.7814,
-      "step": 27557
-    },
-    {
-      "epoch": 14.371838331160365,
-      "grad_norm": 1.511191487312317,
-      "learning_rate": 7.281507537688442e-05,
-      "loss": 5.3523,
-      "step": 27558
-    },
-    {
-      "epoch": 14.372359843546285,
-      "grad_norm": 1.4886810779571533,
-      "learning_rate": 7.28140703517588e-05,
-      "loss": 5.817,
-      "step": 27559
-    },
-    {
-      "epoch": 14.372881355932204,
-      "grad_norm": 1.4136666059494019,
-      "learning_rate": 7.281306532663316e-05,
-      "loss": 5.3974,
-      "step": 27560
-    },
-    {
-      "epoch": 14.373402868318122,
-      "grad_norm": 1.4605038166046143,
-      "learning_rate": 7.281206030150754e-05,
-      "loss": 5.069,
-      "step": 27561
-    },
-    {
-      "epoch": 14.373924380704041,
-      "grad_norm": 1.4737998247146606,
-      "learning_rate": 7.28110552763819e-05,
-      "loss": 5.6547,
-      "step": 27562
-    },
-    {
-      "epoch": 14.37444589308996,
-      "grad_norm": 1.4427639245986938,
-      "learning_rate": 7.281005025125628e-05,
-      "loss": 4.9386,
-      "step": 27563
-    },
-    {
-      "epoch": 14.37496740547588,
-      "grad_norm": 1.3967047929763794,
-      "learning_rate": 7.280904522613066e-05,
-      "loss": 5.5618,
-      "step": 27564
-    },
-    {
-      "epoch": 14.3754889178618,
-      "grad_norm": 1.5285439491271973,
-      "learning_rate": 7.280804020100503e-05,
-      "loss": 5.3108,
-      "step": 27565
-    },
-    {
-      "epoch": 14.37601043024772,
-      "grad_norm": 1.4670512676239014,
-      "learning_rate": 7.28070351758794e-05,
-      "loss": 4.8815,
-      "step": 27566
-    },
-    {
-      "epoch": 14.376531942633637,
-      "grad_norm": 1.4069863557815552,
-      "learning_rate": 7.280603015075377e-05,
-      "loss": 5.4389,
-      "step": 27567
-    },
-    {
-      "epoch": 14.377053455019556,
-      "grad_norm": 1.3812483549118042,
-      "learning_rate": 7.280502512562815e-05,
-      "loss": 5.239,
-      "step": 27568
-    },
-    {
-      "epoch": 14.377574967405476,
-      "grad_norm": 1.37523353099823,
-      "learning_rate": 7.280402010050251e-05,
-      "loss": 5.703,
-      "step": 27569
-    },
-    {
-      "epoch": 14.378096479791395,
-      "grad_norm": 1.4019538164138794,
-      "learning_rate": 7.280301507537689e-05,
-      "loss": 5.4652,
-      "step": 27570
-    },
-    {
-      "epoch": 14.378617992177315,
-      "grad_norm": 1.489944577217102,
-      "learning_rate": 7.280201005025125e-05,
-      "loss": 5.0633,
-      "step": 27571
-    },
-    {
-      "epoch": 14.379139504563234,
-      "grad_norm": 1.3382173776626587,
-      "learning_rate": 7.280100502512563e-05,
-      "loss": 5.4976,
-      "step": 27572
-    },
-    {
-      "epoch": 14.379661016949152,
-      "grad_norm": 1.4841772317886353,
-      "learning_rate": 7.280000000000001e-05,
-      "loss": 5.2958,
-      "step": 27573
-    },
-    {
-      "epoch": 14.380182529335071,
-      "grad_norm": 1.5747919082641602,
-      "learning_rate": 7.279899497487439e-05,
-      "loss": 5.0564,
-      "step": 27574
-    },
-    {
-      "epoch": 14.38070404172099,
-      "grad_norm": 1.512978196144104,
-      "learning_rate": 7.279798994974875e-05,
-      "loss": 5.179,
-      "step": 27575
-    },
-    {
-      "epoch": 14.38122555410691,
-      "grad_norm": 1.4918808937072754,
-      "learning_rate": 7.279698492462313e-05,
-      "loss": 5.1061,
-      "step": 27576
-    },
-    {
-      "epoch": 14.38174706649283,
-      "grad_norm": 1.4735060930252075,
-      "learning_rate": 7.279597989949749e-05,
-      "loss": 5.526,
-      "step": 27577
-    },
-    {
-      "epoch": 14.38226857887875,
-      "grad_norm": 1.419295310974121,
-      "learning_rate": 7.279497487437186e-05,
-      "loss": 5.3387,
-      "step": 27578
-    },
-    {
-      "epoch": 14.382790091264667,
-      "grad_norm": 1.5368515253067017,
-      "learning_rate": 7.279396984924623e-05,
-      "loss": 5.5125,
-      "step": 27579
-    },
-    {
-      "epoch": 14.383311603650586,
-      "grad_norm": 1.4590179920196533,
-      "learning_rate": 7.27929648241206e-05,
-      "loss": 5.7358,
-      "step": 27580
-    },
-    {
-      "epoch": 14.383833116036506,
-      "grad_norm": 1.4521738290786743,
-      "learning_rate": 7.279195979899498e-05,
-      "loss": 4.9136,
-      "step": 27581
-    },
-    {
-      "epoch": 14.384354628422425,
-      "grad_norm": 1.4377857446670532,
-      "learning_rate": 7.279095477386934e-05,
-      "loss": 5.7515,
-      "step": 27582
-    },
-    {
-      "epoch": 14.384876140808345,
-      "grad_norm": 1.4311844110488892,
-      "learning_rate": 7.278994974874372e-05,
-      "loss": 5.0427,
-      "step": 27583
-    },
-    {
-      "epoch": 14.385397653194264,
-      "grad_norm": 1.542914628982544,
-      "learning_rate": 7.27889447236181e-05,
-      "loss": 5.2271,
-      "step": 27584
-    },
-    {
-      "epoch": 14.385919165580182,
-      "grad_norm": 1.4536592960357666,
-      "learning_rate": 7.278793969849247e-05,
-      "loss": 5.829,
-      "step": 27585
-    },
-    {
-      "epoch": 14.386440677966101,
-      "grad_norm": 1.3909837007522583,
-      "learning_rate": 7.278693467336684e-05,
-      "loss": 5.3742,
-      "step": 27586
-    },
-    {
-      "epoch": 14.38696219035202,
-      "grad_norm": 1.4866034984588623,
-      "learning_rate": 7.278592964824122e-05,
-      "loss": 5.3987,
-      "step": 27587
-    },
-    {
-      "epoch": 14.38748370273794,
-      "grad_norm": 1.9148943424224854,
-      "learning_rate": 7.278492462311558e-05,
-      "loss": 4.9646,
-      "step": 27588
-    },
-    {
-      "epoch": 14.38800521512386,
-      "grad_norm": 1.397819995880127,
-      "learning_rate": 7.278391959798996e-05,
-      "loss": 5.779,
-      "step": 27589
-    },
-    {
-      "epoch": 14.388526727509777,
-      "grad_norm": 1.5458022356033325,
-      "learning_rate": 7.278291457286432e-05,
-      "loss": 4.9984,
-      "step": 27590
-    },
-    {
-      "epoch": 14.389048239895697,
-      "grad_norm": 1.4759503602981567,
-      "learning_rate": 7.278190954773869e-05,
-      "loss": 5.6125,
-      "step": 27591
-    },
-    {
-      "epoch": 14.389569752281616,
-      "grad_norm": 1.4542094469070435,
-      "learning_rate": 7.278090452261306e-05,
-      "loss": 5.3708,
-      "step": 27592
-    },
-    {
-      "epoch": 14.390091264667536,
-      "grad_norm": 1.465996265411377,
-      "learning_rate": 7.277989949748744e-05,
-      "loss": 5.226,
-      "step": 27593
-    },
-    {
-      "epoch": 14.390612777053455,
-      "grad_norm": 1.3688136339187622,
-      "learning_rate": 7.277889447236182e-05,
-      "loss": 5.6193,
-      "step": 27594
-    },
-    {
-      "epoch": 14.391134289439375,
-      "grad_norm": 1.5073856115341187,
-      "learning_rate": 7.277788944723618e-05,
-      "loss": 4.7274,
-      "step": 27595
-    },
-    {
-      "epoch": 14.391655801825294,
-      "grad_norm": 1.504684567451477,
-      "learning_rate": 7.277688442211056e-05,
-      "loss": 5.2116,
-      "step": 27596
-    },
-    {
-      "epoch": 14.392177314211212,
-      "grad_norm": 1.4186524152755737,
-      "learning_rate": 7.277587939698493e-05,
-      "loss": 5.3435,
-      "step": 27597
-    },
-    {
-      "epoch": 14.392698826597131,
-      "grad_norm": 1.4407153129577637,
-      "learning_rate": 7.27748743718593e-05,
-      "loss": 5.4847,
-      "step": 27598
-    },
-    {
-      "epoch": 14.39322033898305,
-      "grad_norm": 1.4323970079421997,
-      "learning_rate": 7.277386934673367e-05,
-      "loss": 5.5475,
-      "step": 27599
-    },
-    {
-      "epoch": 14.39374185136897,
-      "grad_norm": 1.4050661325454712,
-      "learning_rate": 7.277286432160805e-05,
-      "loss": 5.5076,
-      "step": 27600
-    },
-    {
-      "epoch": 14.39426336375489,
-      "grad_norm": 1.4234699010849,
-      "learning_rate": 7.277185929648241e-05,
-      "loss": 5.1492,
-      "step": 27601
-    },
-    {
-      "epoch": 14.394784876140807,
-      "grad_norm": 1.6301525831222534,
-      "learning_rate": 7.277085427135679e-05,
-      "loss": 4.7719,
-      "step": 27602
-    },
-    {
-      "epoch": 14.395306388526727,
-      "grad_norm": 1.6184967756271362,
-      "learning_rate": 7.276984924623115e-05,
-      "loss": 5.5347,
-      "step": 27603
-    },
-    {
-      "epoch": 14.395827900912646,
-      "grad_norm": 1.4663903713226318,
-      "learning_rate": 7.276884422110553e-05,
-      "loss": 5.0145,
-      "step": 27604
-    },
-    {
-      "epoch": 14.396349413298566,
-      "grad_norm": 1.7975527048110962,
-      "learning_rate": 7.276783919597991e-05,
-      "loss": 5.0368,
-      "step": 27605
-    },
-    {
-      "epoch": 14.396870925684485,
-      "grad_norm": 1.4398642778396606,
-      "learning_rate": 7.276683417085427e-05,
-      "loss": 5.4723,
-      "step": 27606
-    },
-    {
-      "epoch": 14.397392438070405,
-      "grad_norm": 1.5565979480743408,
-      "learning_rate": 7.276582914572865e-05,
-      "loss": 5.5326,
-      "step": 27607
-    },
-    {
-      "epoch": 14.397913950456322,
-      "grad_norm": 1.5106472969055176,
-      "learning_rate": 7.276482412060301e-05,
-      "loss": 5.2922,
-      "step": 27608
-    },
-    {
-      "epoch": 14.398435462842242,
-      "grad_norm": 1.5161550045013428,
-      "learning_rate": 7.276381909547739e-05,
-      "loss": 5.1155,
-      "step": 27609
-    },
-    {
-      "epoch": 14.398956975228161,
-      "grad_norm": 1.3271260261535645,
-      "learning_rate": 7.276281407035176e-05,
-      "loss": 5.65,
-      "step": 27610
-    },
-    {
-      "epoch": 14.399478487614081,
-      "grad_norm": 1.4156737327575684,
-      "learning_rate": 7.276180904522613e-05,
-      "loss": 5.1066,
-      "step": 27611
-    },
-    {
-      "epoch": 14.4,
-      "grad_norm": 1.4634207487106323,
-      "learning_rate": 7.27608040201005e-05,
-      "loss": 5.4179,
-      "step": 27612
-    },
-    {
-      "epoch": 14.40052151238592,
-      "grad_norm": 1.4780020713806152,
-      "learning_rate": 7.275979899497488e-05,
-      "loss": 5.6949,
-      "step": 27613
-    },
-    {
-      "epoch": 14.401043024771838,
-      "grad_norm": 1.5179702043533325,
-      "learning_rate": 7.275879396984925e-05,
-      "loss": 4.8475,
-      "step": 27614
-    },
-    {
-      "epoch": 14.401564537157757,
-      "grad_norm": 1.4587126970291138,
-      "learning_rate": 7.275778894472363e-05,
-      "loss": 5.6396,
-      "step": 27615
-    },
-    {
-      "epoch": 14.402086049543676,
-      "grad_norm": 1.5007832050323486,
-      "learning_rate": 7.2756783919598e-05,
-      "loss": 5.4136,
-      "step": 27616
-    },
-    {
-      "epoch": 14.402607561929596,
-      "grad_norm": 1.5162914991378784,
-      "learning_rate": 7.275577889447237e-05,
-      "loss": 5.3061,
-      "step": 27617
-    },
-    {
-      "epoch": 14.403129074315515,
-      "grad_norm": 1.358654499053955,
-      "learning_rate": 7.275477386934674e-05,
-      "loss": 5.8459,
-      "step": 27618
-    },
-    {
-      "epoch": 14.403650586701435,
-      "grad_norm": 1.4220609664916992,
-      "learning_rate": 7.27537688442211e-05,
-      "loss": 5.0533,
-      "step": 27619
-    },
-    {
-      "epoch": 14.404172099087353,
-      "grad_norm": 1.4709721803665161,
-      "learning_rate": 7.275276381909548e-05,
-      "loss": 5.0402,
-      "step": 27620
-    },
-    {
-      "epoch": 14.404693611473272,
-      "grad_norm": 1.576439619064331,
-      "learning_rate": 7.275175879396985e-05,
-      "loss": 5.1991,
-      "step": 27621
-    },
-    {
-      "epoch": 14.405215123859191,
-      "grad_norm": 1.3414099216461182,
-      "learning_rate": 7.275075376884422e-05,
-      "loss": 5.8174,
-      "step": 27622
-    },
-    {
-      "epoch": 14.405736636245111,
-      "grad_norm": 1.4068207740783691,
-      "learning_rate": 7.274974874371859e-05,
-      "loss": 5.5249,
-      "step": 27623
-    },
-    {
-      "epoch": 14.40625814863103,
-      "grad_norm": 1.419748067855835,
-      "learning_rate": 7.274874371859297e-05,
-      "loss": 5.5023,
-      "step": 27624
-    },
-    {
-      "epoch": 14.40677966101695,
-      "grad_norm": 1.4400179386138916,
-      "learning_rate": 7.274773869346734e-05,
-      "loss": 5.0627,
-      "step": 27625
-    },
-    {
-      "epoch": 14.407301173402868,
-      "grad_norm": 1.5194300413131714,
-      "learning_rate": 7.274673366834172e-05,
-      "loss": 5.4659,
-      "step": 27626
-    },
-    {
-      "epoch": 14.407822685788787,
-      "grad_norm": 1.5091147422790527,
-      "learning_rate": 7.274572864321609e-05,
-      "loss": 5.3377,
-      "step": 27627
-    },
-    {
-      "epoch": 14.408344198174706,
-      "grad_norm": 1.3843814134597778,
-      "learning_rate": 7.274472361809046e-05,
-      "loss": 5.5531,
-      "step": 27628
-    },
-    {
-      "epoch": 14.408865710560626,
-      "grad_norm": 1.4678709506988525,
-      "learning_rate": 7.274371859296483e-05,
-      "loss": 5.5789,
-      "step": 27629
-    },
-    {
-      "epoch": 14.409387222946545,
-      "grad_norm": 1.4849708080291748,
-      "learning_rate": 7.27427135678392e-05,
-      "loss": 5.5739,
-      "step": 27630
-    },
-    {
-      "epoch": 14.409908735332465,
-      "grad_norm": 1.5949519872665405,
-      "learning_rate": 7.274170854271357e-05,
-      "loss": 5.0549,
-      "step": 27631
-    },
-    {
-      "epoch": 14.410430247718383,
-      "grad_norm": 1.4980106353759766,
-      "learning_rate": 7.274070351758793e-05,
-      "loss": 5.0586,
-      "step": 27632
-    },
-    {
-      "epoch": 14.410951760104302,
-      "grad_norm": 1.3864704370498657,
-      "learning_rate": 7.273969849246231e-05,
-      "loss": 5.7898,
-      "step": 27633
-    },
-    {
-      "epoch": 14.411473272490221,
-      "grad_norm": 1.4171779155731201,
-      "learning_rate": 7.273869346733669e-05,
-      "loss": 4.6579,
-      "step": 27634
-    },
-    {
-      "epoch": 14.411994784876141,
-      "grad_norm": 1.5034793615341187,
-      "learning_rate": 7.273768844221107e-05,
-      "loss": 5.6096,
-      "step": 27635
-    },
-    {
-      "epoch": 14.41251629726206,
-      "grad_norm": 1.5430938005447388,
-      "learning_rate": 7.273668341708543e-05,
-      "loss": 4.7775,
-      "step": 27636
-    },
-    {
-      "epoch": 14.41303780964798,
-      "grad_norm": 1.440457820892334,
-      "learning_rate": 7.273567839195981e-05,
-      "loss": 4.9259,
-      "step": 27637
-    },
-    {
-      "epoch": 14.413559322033898,
-      "grad_norm": 1.4758105278015137,
-      "learning_rate": 7.273467336683417e-05,
-      "loss": 5.3109,
-      "step": 27638
-    },
-    {
-      "epoch": 14.414080834419817,
-      "grad_norm": 1.5456050634384155,
-      "learning_rate": 7.273366834170855e-05,
-      "loss": 5.1605,
-      "step": 27639
-    },
-    {
-      "epoch": 14.414602346805736,
-      "grad_norm": 1.5639119148254395,
-      "learning_rate": 7.273266331658292e-05,
-      "loss": 5.4143,
-      "step": 27640
-    },
-    {
-      "epoch": 14.415123859191656,
-      "grad_norm": 1.4406700134277344,
-      "learning_rate": 7.27316582914573e-05,
-      "loss": 4.9901,
-      "step": 27641
-    },
-    {
-      "epoch": 14.415645371577575,
-      "grad_norm": 1.467941164970398,
-      "learning_rate": 7.273065326633166e-05,
-      "loss": 5.6716,
-      "step": 27642
-    },
-    {
-      "epoch": 14.416166883963495,
-      "grad_norm": 1.4577616453170776,
-      "learning_rate": 7.272964824120604e-05,
-      "loss": 5.7237,
-      "step": 27643
-    },
-    {
-      "epoch": 14.416688396349413,
-      "grad_norm": 1.3786224126815796,
-      "learning_rate": 7.27286432160804e-05,
-      "loss": 5.494,
-      "step": 27644
-    },
-    {
-      "epoch": 14.417209908735332,
-      "grad_norm": 1.5208021402359009,
-      "learning_rate": 7.272763819095478e-05,
-      "loss": 5.1983,
-      "step": 27645
-    },
-    {
-      "epoch": 14.417731421121252,
-      "grad_norm": 1.5820552110671997,
-      "learning_rate": 7.272663316582916e-05,
-      "loss": 5.0089,
-      "step": 27646
-    },
-    {
-      "epoch": 14.418252933507171,
-      "grad_norm": 1.4551130533218384,
-      "learning_rate": 7.272562814070352e-05,
-      "loss": 5.3849,
-      "step": 27647
-    },
-    {
-      "epoch": 14.41877444589309,
-      "grad_norm": 1.447651743888855,
-      "learning_rate": 7.27246231155779e-05,
-      "loss": 5.4428,
-      "step": 27648
-    },
-    {
-      "epoch": 14.41929595827901,
-      "grad_norm": 1.5171215534210205,
-      "learning_rate": 7.272361809045226e-05,
-      "loss": 5.1578,
-      "step": 27649
-    },
-    {
-      "epoch": 14.419817470664928,
-      "grad_norm": 1.4382719993591309,
-      "learning_rate": 7.272261306532664e-05,
-      "loss": 5.8548,
-      "step": 27650
-    },
-    {
-      "epoch": 14.420338983050847,
-      "grad_norm": 1.5202951431274414,
-      "learning_rate": 7.2721608040201e-05,
-      "loss": 5.4418,
-      "step": 27651
-    },
-    {
-      "epoch": 14.420860495436767,
-      "grad_norm": 1.4068336486816406,
-      "learning_rate": 7.272060301507538e-05,
-      "loss": 5.4614,
-      "step": 27652
-    },
-    {
-      "epoch": 14.421382007822686,
-      "grad_norm": 1.525843620300293,
-      "learning_rate": 7.271959798994975e-05,
-      "loss": 5.2123,
-      "step": 27653
-    },
-    {
-      "epoch": 14.421903520208605,
-      "grad_norm": 1.5651401281356812,
-      "learning_rate": 7.271859296482412e-05,
-      "loss": 5.1019,
-      "step": 27654
-    },
-    {
-      "epoch": 14.422425032594525,
-      "grad_norm": 1.5088211297988892,
-      "learning_rate": 7.271758793969849e-05,
-      "loss": 5.1983,
-      "step": 27655
-    },
-    {
-      "epoch": 14.422946544980443,
-      "grad_norm": 1.521807312965393,
-      "learning_rate": 7.271658291457287e-05,
-      "loss": 5.8394,
-      "step": 27656
-    },
-    {
-      "epoch": 14.423468057366362,
-      "grad_norm": 1.5285305976867676,
-      "learning_rate": 7.271557788944724e-05,
-      "loss": 5.1647,
-      "step": 27657
-    },
-    {
-      "epoch": 14.423989569752282,
-      "grad_norm": 1.4264870882034302,
-      "learning_rate": 7.271457286432161e-05,
-      "loss": 5.6697,
-      "step": 27658
-    },
-    {
-      "epoch": 14.424511082138201,
-      "grad_norm": 1.3591221570968628,
-      "learning_rate": 7.271356783919599e-05,
-      "loss": 5.4783,
-      "step": 27659
-    },
-    {
-      "epoch": 14.42503259452412,
-      "grad_norm": 1.569525957107544,
-      "learning_rate": 7.271256281407035e-05,
-      "loss": 4.8244,
-      "step": 27660
-    },
-    {
-      "epoch": 14.42555410691004,
-      "grad_norm": 1.4206935167312622,
-      "learning_rate": 7.271155778894473e-05,
-      "loss": 5.3439,
-      "step": 27661
-    },
-    {
-      "epoch": 14.426075619295958,
-      "grad_norm": 1.724382758140564,
-      "learning_rate": 7.271055276381909e-05,
-      "loss": 5.296,
-      "step": 27662
-    },
-    {
-      "epoch": 14.426597131681877,
-      "grad_norm": 1.4348890781402588,
-      "learning_rate": 7.270954773869347e-05,
-      "loss": 5.7335,
-      "step": 27663
-    },
-    {
-      "epoch": 14.427118644067797,
-      "grad_norm": 1.4241927862167358,
-      "learning_rate": 7.270854271356783e-05,
-      "loss": 5.7369,
-      "step": 27664
-    },
-    {
-      "epoch": 14.427640156453716,
-      "grad_norm": 1.4064114093780518,
-      "learning_rate": 7.270753768844221e-05,
-      "loss": 5.8702,
-      "step": 27665
-    },
-    {
-      "epoch": 14.428161668839635,
-      "grad_norm": 1.5673658847808838,
-      "learning_rate": 7.270653266331659e-05,
-      "loss": 5.1883,
-      "step": 27666
-    },
-    {
-      "epoch": 14.428683181225555,
-      "grad_norm": 1.424349308013916,
-      "learning_rate": 7.270552763819097e-05,
-      "loss": 5.5786,
-      "step": 27667
-    },
-    {
-      "epoch": 14.429204693611473,
-      "grad_norm": 1.5355361700057983,
-      "learning_rate": 7.270452261306533e-05,
-      "loss": 5.2969,
-      "step": 27668
-    },
-    {
-      "epoch": 14.429726205997392,
-      "grad_norm": 1.5192137956619263,
-      "learning_rate": 7.270351758793971e-05,
-      "loss": 5.368,
-      "step": 27669
-    },
-    {
-      "epoch": 14.430247718383312,
-      "grad_norm": 1.5318351984024048,
-      "learning_rate": 7.270251256281407e-05,
-      "loss": 5.0327,
-      "step": 27670
-    },
-    {
-      "epoch": 14.430769230769231,
-      "grad_norm": 1.4656283855438232,
-      "learning_rate": 7.270150753768844e-05,
-      "loss": 5.5624,
-      "step": 27671
-    },
-    {
-      "epoch": 14.43129074315515,
-      "grad_norm": 1.5081496238708496,
-      "learning_rate": 7.270050251256282e-05,
-      "loss": 4.8374,
-      "step": 27672
-    },
-    {
-      "epoch": 14.43181225554107,
-      "grad_norm": 1.4863418340682983,
-      "learning_rate": 7.269949748743718e-05,
-      "loss": 5.5563,
-      "step": 27673
-    },
-    {
-      "epoch": 14.432333767926988,
-      "grad_norm": 1.4961192607879639,
-      "learning_rate": 7.269849246231156e-05,
-      "loss": 5.2957,
-      "step": 27674
-    },
-    {
-      "epoch": 14.432855280312907,
-      "grad_norm": 1.5365610122680664,
-      "learning_rate": 7.269748743718592e-05,
-      "loss": 5.1054,
-      "step": 27675
-    },
-    {
-      "epoch": 14.433376792698827,
-      "grad_norm": 1.471153736114502,
-      "learning_rate": 7.26964824120603e-05,
-      "loss": 5.0909,
-      "step": 27676
-    },
-    {
-      "epoch": 14.433898305084746,
-      "grad_norm": 1.3564966917037964,
-      "learning_rate": 7.269547738693468e-05,
-      "loss": 5.7511,
-      "step": 27677
-    },
-    {
-      "epoch": 14.434419817470665,
-      "grad_norm": 1.439067006111145,
-      "learning_rate": 7.269447236180906e-05,
-      "loss": 5.6651,
-      "step": 27678
-    },
-    {
-      "epoch": 14.434941329856585,
-      "grad_norm": 1.504670262336731,
-      "learning_rate": 7.269346733668342e-05,
-      "loss": 5.3846,
-      "step": 27679
-    },
-    {
-      "epoch": 14.435462842242503,
-      "grad_norm": 1.5574694871902466,
-      "learning_rate": 7.26924623115578e-05,
-      "loss": 5.4486,
-      "step": 27680
-    },
-    {
-      "epoch": 14.435984354628422,
-      "grad_norm": 1.5556209087371826,
-      "learning_rate": 7.269145728643216e-05,
-      "loss": 5.0894,
-      "step": 27681
-    },
-    {
-      "epoch": 14.436505867014342,
-      "grad_norm": 1.4284785985946655,
-      "learning_rate": 7.269045226130654e-05,
-      "loss": 5.7373,
-      "step": 27682
-    },
-    {
-      "epoch": 14.437027379400261,
-      "grad_norm": 1.4737839698791504,
-      "learning_rate": 7.26894472361809e-05,
-      "loss": 5.189,
-      "step": 27683
-    },
-    {
-      "epoch": 14.43754889178618,
-      "grad_norm": 1.528730869293213,
-      "learning_rate": 7.268844221105528e-05,
-      "loss": 5.0716,
-      "step": 27684
-    },
-    {
-      "epoch": 14.438070404172098,
-      "grad_norm": 1.3834675550460815,
-      "learning_rate": 7.268743718592965e-05,
-      "loss": 5.5831,
-      "step": 27685
-    },
-    {
-      "epoch": 14.438591916558018,
-      "grad_norm": 1.4192559719085693,
-      "learning_rate": 7.268643216080402e-05,
-      "loss": 5.358,
-      "step": 27686
-    },
-    {
-      "epoch": 14.439113428943937,
-      "grad_norm": 1.3601224422454834,
-      "learning_rate": 7.26854271356784e-05,
-      "loss": 5.6423,
-      "step": 27687
-    },
-    {
-      "epoch": 14.439634941329857,
-      "grad_norm": 1.3295645713806152,
-      "learning_rate": 7.268442211055277e-05,
-      "loss": 5.5172,
-      "step": 27688
-    },
-    {
-      "epoch": 14.440156453715776,
-      "grad_norm": 1.4197274446487427,
-      "learning_rate": 7.268341708542714e-05,
-      "loss": 4.945,
-      "step": 27689
-    },
-    {
-      "epoch": 14.440677966101696,
-      "grad_norm": 1.4548983573913574,
-      "learning_rate": 7.268241206030151e-05,
-      "loss": 5.8542,
-      "step": 27690
-    },
-    {
-      "epoch": 14.441199478487615,
-      "grad_norm": 1.4680837392807007,
-      "learning_rate": 7.268140703517589e-05,
-      "loss": 4.7951,
-      "step": 27691
-    },
-    {
-      "epoch": 14.441720990873533,
-      "grad_norm": 1.4789537191390991,
-      "learning_rate": 7.268040201005025e-05,
-      "loss": 5.1471,
-      "step": 27692
-    },
-    {
-      "epoch": 14.442242503259452,
-      "grad_norm": 1.3876022100448608,
-      "learning_rate": 7.267939698492463e-05,
-      "loss": 5.5583,
-      "step": 27693
-    },
-    {
-      "epoch": 14.442764015645372,
-      "grad_norm": 1.445919394493103,
-      "learning_rate": 7.267839195979899e-05,
-      "loss": 5.5631,
-      "step": 27694
-    },
-    {
-      "epoch": 14.443285528031291,
-      "grad_norm": 1.418931245803833,
-      "learning_rate": 7.267738693467337e-05,
-      "loss": 5.682,
-      "step": 27695
-    },
-    {
-      "epoch": 14.44380704041721,
-      "grad_norm": 1.4336682558059692,
-      "learning_rate": 7.267638190954774e-05,
-      "loss": 5.7093,
-      "step": 27696
-    },
-    {
-      "epoch": 14.444328552803128,
-      "grad_norm": 1.5327438116073608,
-      "learning_rate": 7.267537688442211e-05,
-      "loss": 5.1836,
-      "step": 27697
-    },
-    {
-      "epoch": 14.444850065189048,
-      "grad_norm": 1.5527621507644653,
-      "learning_rate": 7.267437185929649e-05,
-      "loss": 5.4599,
-      "step": 27698
-    },
-    {
-      "epoch": 14.445371577574967,
-      "grad_norm": 1.4755051136016846,
-      "learning_rate": 7.267336683417086e-05,
-      "loss": 5.2244,
-      "step": 27699
-    },
-    {
-      "epoch": 14.445893089960887,
-      "grad_norm": 1.5321329832077026,
-      "learning_rate": 7.267236180904523e-05,
-      "loss": 5.4423,
-      "step": 27700
-    },
-    {
-      "epoch": 14.446414602346806,
-      "grad_norm": 1.436448574066162,
-      "learning_rate": 7.26713567839196e-05,
-      "loss": 5.5744,
-      "step": 27701
-    },
-    {
-      "epoch": 14.446936114732726,
-      "grad_norm": 1.5852630138397217,
-      "learning_rate": 7.267035175879398e-05,
-      "loss": 5.4965,
-      "step": 27702
-    },
-    {
-      "epoch": 14.447457627118643,
-      "grad_norm": 1.543637990951538,
-      "learning_rate": 7.266934673366834e-05,
-      "loss": 5.3288,
-      "step": 27703
-    },
-    {
-      "epoch": 14.447979139504563,
-      "grad_norm": 1.5154595375061035,
-      "learning_rate": 7.266834170854272e-05,
-      "loss": 5.1237,
-      "step": 27704
-    },
-    {
-      "epoch": 14.448500651890482,
-      "grad_norm": 1.550469994544983,
-      "learning_rate": 7.266733668341708e-05,
-      "loss": 5.1771,
-      "step": 27705
-    },
-    {
-      "epoch": 14.449022164276402,
-      "grad_norm": 1.368315577507019,
-      "learning_rate": 7.266633165829146e-05,
-      "loss": 5.4059,
-      "step": 27706
-    },
-    {
-      "epoch": 14.449543676662321,
-      "grad_norm": 1.5344254970550537,
-      "learning_rate": 7.266532663316584e-05,
-      "loss": 5.5442,
-      "step": 27707
-    },
-    {
-      "epoch": 14.45006518904824,
-      "grad_norm": 1.4184614419937134,
-      "learning_rate": 7.266432160804021e-05,
-      "loss": 5.4949,
-      "step": 27708
-    },
-    {
-      "epoch": 14.450586701434158,
-      "grad_norm": 1.3642655611038208,
-      "learning_rate": 7.266331658291458e-05,
-      "loss": 5.6513,
-      "step": 27709
-    },
-    {
-      "epoch": 14.451108213820078,
-      "grad_norm": 1.4591944217681885,
-      "learning_rate": 7.266231155778896e-05,
-      "loss": 5.231,
-      "step": 27710
-    },
-    {
-      "epoch": 14.451629726205997,
-      "grad_norm": 1.5610713958740234,
-      "learning_rate": 7.266130653266332e-05,
-      "loss": 5.566,
-      "step": 27711
-    },
-    {
-      "epoch": 14.452151238591917,
-      "grad_norm": 1.4501713514328003,
-      "learning_rate": 7.266030150753769e-05,
-      "loss": 5.6188,
-      "step": 27712
-    },
-    {
-      "epoch": 14.452672750977836,
-      "grad_norm": 1.4586360454559326,
-      "learning_rate": 7.265929648241206e-05,
-      "loss": 5.523,
-      "step": 27713
-    },
-    {
-      "epoch": 14.453194263363756,
-      "grad_norm": 1.555533766746521,
-      "learning_rate": 7.265829145728643e-05,
-      "loss": 5.1676,
-      "step": 27714
-    },
-    {
-      "epoch": 14.453715775749673,
-      "grad_norm": 1.4611865282058716,
-      "learning_rate": 7.26572864321608e-05,
-      "loss": 5.4788,
-      "step": 27715
-    },
-    {
-      "epoch": 14.454237288135593,
-      "grad_norm": 1.4902434349060059,
-      "learning_rate": 7.265628140703517e-05,
-      "loss": 5.3554,
-      "step": 27716
-    },
-    {
-      "epoch": 14.454758800521512,
-      "grad_norm": 1.3879120349884033,
-      "learning_rate": 7.265527638190955e-05,
-      "loss": 5.7732,
-      "step": 27717
-    },
-    {
-      "epoch": 14.455280312907432,
-      "grad_norm": 1.660056710243225,
-      "learning_rate": 7.265427135678393e-05,
-      "loss": 4.7633,
-      "step": 27718
-    },
-    {
-      "epoch": 14.455801825293351,
-      "grad_norm": 1.444818377494812,
-      "learning_rate": 7.26532663316583e-05,
-      "loss": 5.4992,
-      "step": 27719
-    },
-    {
-      "epoch": 14.45632333767927,
-      "grad_norm": 1.4109668731689453,
-      "learning_rate": 7.265226130653267e-05,
-      "loss": 5.7103,
-      "step": 27720
-    },
-    {
-      "epoch": 14.456844850065188,
-      "grad_norm": 1.4111560583114624,
-      "learning_rate": 7.265125628140705e-05,
-      "loss": 5.2959,
-      "step": 27721
-    },
-    {
-      "epoch": 14.457366362451108,
-      "grad_norm": 1.4248099327087402,
-      "learning_rate": 7.265025125628141e-05,
-      "loss": 5.2787,
-      "step": 27722
-    },
-    {
-      "epoch": 14.457887874837027,
-      "grad_norm": 1.4393807649612427,
-      "learning_rate": 7.264924623115579e-05,
-      "loss": 5.6183,
-      "step": 27723
-    },
-    {
-      "epoch": 14.458409387222947,
-      "grad_norm": 1.5288970470428467,
-      "learning_rate": 7.264824120603015e-05,
-      "loss": 5.0213,
-      "step": 27724
-    },
-    {
-      "epoch": 14.458930899608866,
-      "grad_norm": 1.428405523300171,
-      "learning_rate": 7.264723618090452e-05,
-      "loss": 4.9787,
-      "step": 27725
-    },
-    {
-      "epoch": 14.459452411994786,
-      "grad_norm": 1.3811813592910767,
-      "learning_rate": 7.26462311557789e-05,
-      "loss": 5.3301,
-      "step": 27726
-    },
-    {
-      "epoch": 14.459973924380703,
-      "grad_norm": 1.4768126010894775,
-      "learning_rate": 7.264522613065327e-05,
-      "loss": 5.2472,
-      "step": 27727
-    },
-    {
-      "epoch": 14.460495436766623,
-      "grad_norm": 1.4634824991226196,
-      "learning_rate": 7.264422110552765e-05,
-      "loss": 5.5592,
-      "step": 27728
-    },
-    {
-      "epoch": 14.461016949152542,
-      "grad_norm": 1.4900482892990112,
-      "learning_rate": 7.264321608040201e-05,
-      "loss": 5.3206,
-      "step": 27729
-    },
-    {
-      "epoch": 14.461538461538462,
-      "grad_norm": 1.3889341354370117,
-      "learning_rate": 7.264221105527639e-05,
-      "loss": 5.4339,
-      "step": 27730
-    },
-    {
-      "epoch": 14.462059973924381,
-      "grad_norm": 1.4030961990356445,
-      "learning_rate": 7.264120603015076e-05,
-      "loss": 5.5131,
-      "step": 27731
-    },
-    {
-      "epoch": 14.4625814863103,
-      "grad_norm": 1.463734745979309,
-      "learning_rate": 7.264020100502513e-05,
-      "loss": 5.3997,
-      "step": 27732
-    },
-    {
-      "epoch": 14.463102998696218,
-      "grad_norm": 1.4711803197860718,
-      "learning_rate": 7.26391959798995e-05,
-      "loss": 5.4106,
-      "step": 27733
-    },
-    {
-      "epoch": 14.463624511082138,
-      "grad_norm": 1.6288126707077026,
-      "learning_rate": 7.263819095477388e-05,
-      "loss": 5.0328,
-      "step": 27734
-    },
-    {
-      "epoch": 14.464146023468057,
-      "grad_norm": 1.3677035570144653,
-      "learning_rate": 7.263718592964824e-05,
-      "loss": 5.6758,
-      "step": 27735
-    },
-    {
-      "epoch": 14.464667535853977,
-      "grad_norm": 1.4606349468231201,
-      "learning_rate": 7.263618090452262e-05,
-      "loss": 5.2466,
-      "step": 27736
-    },
-    {
-      "epoch": 14.465189048239896,
-      "grad_norm": 1.45546555519104,
-      "learning_rate": 7.263517587939698e-05,
-      "loss": 5.3705,
-      "step": 27737
-    },
-    {
-      "epoch": 14.465710560625816,
-      "grad_norm": 1.3759706020355225,
-      "learning_rate": 7.263417085427136e-05,
-      "loss": 5.5812,
-      "step": 27738
-    },
-    {
-      "epoch": 14.466232073011733,
-      "grad_norm": 1.4078596830368042,
-      "learning_rate": 7.263316582914574e-05,
-      "loss": 5.3421,
-      "step": 27739
-    },
-    {
-      "epoch": 14.466753585397653,
-      "grad_norm": 1.5589983463287354,
-      "learning_rate": 7.26321608040201e-05,
-      "loss": 5.2831,
-      "step": 27740
-    },
-    {
-      "epoch": 14.467275097783572,
-      "grad_norm": 1.4708397388458252,
-      "learning_rate": 7.263115577889448e-05,
-      "loss": 5.5382,
-      "step": 27741
-    },
-    {
-      "epoch": 14.467796610169492,
-      "grad_norm": 1.4526363611221313,
-      "learning_rate": 7.263015075376884e-05,
-      "loss": 5.6068,
-      "step": 27742
-    },
-    {
-      "epoch": 14.468318122555411,
-      "grad_norm": 1.4551074504852295,
-      "learning_rate": 7.262914572864322e-05,
-      "loss": 5.2787,
-      "step": 27743
-    },
-    {
-      "epoch": 14.46883963494133,
-      "grad_norm": 1.4054229259490967,
-      "learning_rate": 7.262814070351759e-05,
-      "loss": 5.4746,
-      "step": 27744
-    },
-    {
-      "epoch": 14.469361147327248,
-      "grad_norm": 1.4147849082946777,
-      "learning_rate": 7.262713567839196e-05,
-      "loss": 5.4136,
-      "step": 27745
-    },
-    {
-      "epoch": 14.469882659713168,
-      "grad_norm": 1.4099295139312744,
-      "learning_rate": 7.262613065326633e-05,
-      "loss": 5.307,
-      "step": 27746
-    },
-    {
-      "epoch": 14.470404172099087,
-      "grad_norm": 1.4026758670806885,
-      "learning_rate": 7.26251256281407e-05,
-      "loss": 5.8539,
-      "step": 27747
-    },
-    {
-      "epoch": 14.470925684485007,
-      "grad_norm": 1.529367446899414,
-      "learning_rate": 7.262412060301508e-05,
-      "loss": 5.426,
-      "step": 27748
-    },
-    {
-      "epoch": 14.471447196870926,
-      "grad_norm": 1.5660845041275024,
-      "learning_rate": 7.262311557788946e-05,
-      "loss": 5.493,
-      "step": 27749
-    },
-    {
-      "epoch": 14.471968709256846,
-      "grad_norm": 1.4588487148284912,
-      "learning_rate": 7.262211055276383e-05,
-      "loss": 5.2929,
-      "step": 27750
-    },
-    {
-      "epoch": 14.472490221642763,
-      "grad_norm": 1.5473473072052002,
-      "learning_rate": 7.262110552763819e-05,
-      "loss": 5.8364,
-      "step": 27751
-    },
-    {
-      "epoch": 14.473011734028683,
-      "grad_norm": 1.4805186986923218,
-      "learning_rate": 7.262010050251257e-05,
-      "loss": 4.9987,
-      "step": 27752
-    },
-    {
-      "epoch": 14.473533246414602,
-      "grad_norm": 1.5037000179290771,
-      "learning_rate": 7.261909547738693e-05,
-      "loss": 5.4912,
-      "step": 27753
-    },
-    {
-      "epoch": 14.474054758800522,
-      "grad_norm": 1.5688174962997437,
-      "learning_rate": 7.261809045226131e-05,
-      "loss": 5.0506,
-      "step": 27754
-    },
-    {
-      "epoch": 14.474576271186441,
-      "grad_norm": 1.4966380596160889,
-      "learning_rate": 7.261708542713567e-05,
-      "loss": 5.0404,
-      "step": 27755
-    },
-    {
-      "epoch": 14.47509778357236,
-      "grad_norm": 1.4708398580551147,
-      "learning_rate": 7.261608040201005e-05,
-      "loss": 5.583,
-      "step": 27756
-    },
-    {
-      "epoch": 14.475619295958278,
-      "grad_norm": 1.4749945402145386,
-      "learning_rate": 7.261507537688442e-05,
-      "loss": 5.2941,
-      "step": 27757
-    },
-    {
-      "epoch": 14.476140808344198,
-      "grad_norm": 1.4622371196746826,
-      "learning_rate": 7.26140703517588e-05,
-      "loss": 5.468,
-      "step": 27758
-    },
-    {
-      "epoch": 14.476662320730117,
-      "grad_norm": 1.4731217622756958,
-      "learning_rate": 7.261306532663317e-05,
-      "loss": 5.6188,
-      "step": 27759
-    },
-    {
-      "epoch": 14.477183833116037,
-      "grad_norm": 1.5235589742660522,
-      "learning_rate": 7.261206030150755e-05,
-      "loss": 4.9841,
-      "step": 27760
-    },
-    {
-      "epoch": 14.477705345501956,
-      "grad_norm": 1.4284309148788452,
-      "learning_rate": 7.261105527638191e-05,
-      "loss": 5.6334,
-      "step": 27761
-    },
-    {
-      "epoch": 14.478226857887876,
-      "grad_norm": 1.5896315574645996,
-      "learning_rate": 7.261005025125629e-05,
-      "loss": 4.7828,
-      "step": 27762
-    },
-    {
-      "epoch": 14.478748370273793,
-      "grad_norm": 1.6022361516952515,
-      "learning_rate": 7.260904522613066e-05,
-      "loss": 5.1194,
-      "step": 27763
-    },
-    {
-      "epoch": 14.479269882659713,
-      "grad_norm": 1.4188021421432495,
-      "learning_rate": 7.260804020100502e-05,
-      "loss": 5.3678,
-      "step": 27764
-    },
-    {
-      "epoch": 14.479791395045632,
-      "grad_norm": 1.530936598777771,
-      "learning_rate": 7.26070351758794e-05,
-      "loss": 4.9948,
-      "step": 27765
-    },
-    {
-      "epoch": 14.480312907431552,
-      "grad_norm": 1.3751217126846313,
-      "learning_rate": 7.260603015075376e-05,
-      "loss": 5.559,
-      "step": 27766
-    },
-    {
-      "epoch": 14.480834419817471,
-      "grad_norm": 1.4443784952163696,
-      "learning_rate": 7.260502512562814e-05,
-      "loss": 5.2153,
-      "step": 27767
-    },
-    {
-      "epoch": 14.48135593220339,
-      "grad_norm": 1.4627926349639893,
-      "learning_rate": 7.260402010050252e-05,
-      "loss": 5.2011,
-      "step": 27768
-    },
-    {
-      "epoch": 14.481877444589308,
-      "grad_norm": 1.4500964879989624,
-      "learning_rate": 7.26030150753769e-05,
-      "loss": 5.3407,
-      "step": 27769
-    },
-    {
-      "epoch": 14.482398956975228,
-      "grad_norm": 1.522821068763733,
-      "learning_rate": 7.260201005025126e-05,
-      "loss": 5.1913,
-      "step": 27770
-    },
-    {
-      "epoch": 14.482920469361147,
-      "grad_norm": 1.4358763694763184,
-      "learning_rate": 7.260100502512564e-05,
-      "loss": 5.0352,
-      "step": 27771
-    },
-    {
-      "epoch": 14.483441981747067,
-      "grad_norm": 1.5939900875091553,
-      "learning_rate": 7.26e-05,
-      "loss": 5.0295,
-      "step": 27772
-    },
-    {
-      "epoch": 14.483963494132986,
-      "grad_norm": 1.5440109968185425,
-      "learning_rate": 7.259899497487438e-05,
-      "loss": 4.9022,
-      "step": 27773
-    },
-    {
-      "epoch": 14.484485006518906,
-      "grad_norm": 1.519138216972351,
-      "learning_rate": 7.259798994974874e-05,
-      "loss": 5.1652,
-      "step": 27774
-    },
-    {
-      "epoch": 14.485006518904823,
-      "grad_norm": 1.5002444982528687,
-      "learning_rate": 7.259698492462312e-05,
-      "loss": 5.5496,
-      "step": 27775
-    },
-    {
-      "epoch": 14.485528031290743,
-      "grad_norm": 1.3176140785217285,
-      "learning_rate": 7.259597989949749e-05,
-      "loss": 5.9834,
-      "step": 27776
-    },
-    {
-      "epoch": 14.486049543676662,
-      "grad_norm": 1.5341355800628662,
-      "learning_rate": 7.259497487437186e-05,
-      "loss": 4.6882,
-      "step": 27777
-    },
-    {
-      "epoch": 14.486571056062582,
-      "grad_norm": 1.4376860857009888,
-      "learning_rate": 7.259396984924623e-05,
-      "loss": 5.4669,
-      "step": 27778
-    },
-    {
-      "epoch": 14.487092568448501,
-      "grad_norm": 1.4716298580169678,
-      "learning_rate": 7.259296482412061e-05,
-      "loss": 5.8602,
-      "step": 27779
-    },
-    {
-      "epoch": 14.487614080834419,
-      "grad_norm": 1.4544252157211304,
-      "learning_rate": 7.259195979899498e-05,
-      "loss": 4.8648,
-      "step": 27780
-    },
-    {
-      "epoch": 14.488135593220338,
-      "grad_norm": 1.4903558492660522,
-      "learning_rate": 7.259095477386935e-05,
-      "loss": 5.4138,
-      "step": 27781
-    },
-    {
-      "epoch": 14.488657105606258,
-      "grad_norm": 1.400352120399475,
-      "learning_rate": 7.258994974874373e-05,
-      "loss": 5.4725,
-      "step": 27782
-    },
-    {
-      "epoch": 14.489178617992177,
-      "grad_norm": 1.5347641706466675,
-      "learning_rate": 7.258894472361809e-05,
-      "loss": 5.2969,
-      "step": 27783
-    },
-    {
-      "epoch": 14.489700130378097,
-      "grad_norm": 1.4404325485229492,
-      "learning_rate": 7.258793969849247e-05,
-      "loss": 5.5077,
-      "step": 27784
-    },
-    {
-      "epoch": 14.490221642764016,
-      "grad_norm": 1.4458062648773193,
-      "learning_rate": 7.258693467336683e-05,
-      "loss": 5.0213,
-      "step": 27785
-    },
-    {
-      "epoch": 14.490743155149936,
-      "grad_norm": 1.39250910282135,
-      "learning_rate": 7.258592964824121e-05,
-      "loss": 5.5233,
-      "step": 27786
-    },
-    {
-      "epoch": 14.491264667535853,
-      "grad_norm": 1.4490833282470703,
-      "learning_rate": 7.258492462311558e-05,
-      "loss": 5.2382,
-      "step": 27787
-    },
-    {
-      "epoch": 14.491786179921773,
-      "grad_norm": 1.4546334743499756,
-      "learning_rate": 7.258391959798995e-05,
-      "loss": 5.005,
-      "step": 27788
-    },
-    {
-      "epoch": 14.492307692307692,
-      "grad_norm": 1.3636709451675415,
-      "learning_rate": 7.258291457286433e-05,
-      "loss": 5.5393,
-      "step": 27789
-    },
-    {
-      "epoch": 14.492829204693612,
-      "grad_norm": 1.3715928792953491,
-      "learning_rate": 7.258190954773871e-05,
-      "loss": 5.4288,
-      "step": 27790
-    },
-    {
-      "epoch": 14.493350717079531,
-      "grad_norm": 1.5435729026794434,
-      "learning_rate": 7.258090452261307e-05,
-      "loss": 4.9821,
-      "step": 27791
-    },
-    {
-      "epoch": 14.493872229465449,
-      "grad_norm": 1.3986783027648926,
-      "learning_rate": 7.257989949748744e-05,
-      "loss": 5.4613,
-      "step": 27792
-    },
-    {
-      "epoch": 14.494393741851368,
-      "grad_norm": 1.4486676454544067,
-      "learning_rate": 7.257889447236182e-05,
-      "loss": 5.3538,
-      "step": 27793
-    },
-    {
-      "epoch": 14.494915254237288,
-      "grad_norm": 1.3294181823730469,
-      "learning_rate": 7.257788944723618e-05,
-      "loss": 4.7557,
-      "step": 27794
-    },
-    {
-      "epoch": 14.495436766623207,
-      "grad_norm": 1.4566243886947632,
-      "learning_rate": 7.257688442211056e-05,
-      "loss": 5.624,
-      "step": 27795
-    },
-    {
-      "epoch": 14.495958279009127,
-      "grad_norm": 1.4289017915725708,
-      "learning_rate": 7.257587939698492e-05,
-      "loss": 5.4874,
-      "step": 27796
-    },
-    {
-      "epoch": 14.496479791395046,
-      "grad_norm": 1.475429654121399,
-      "learning_rate": 7.25748743718593e-05,
-      "loss": 5.2971,
-      "step": 27797
-    },
-    {
-      "epoch": 14.497001303780964,
-      "grad_norm": 1.4173730611801147,
-      "learning_rate": 7.257386934673366e-05,
-      "loss": 5.5276,
-      "step": 27798
-    },
-    {
-      "epoch": 14.497522816166883,
-      "grad_norm": 1.5801352262496948,
-      "learning_rate": 7.257286432160804e-05,
-      "loss": 5.0413,
-      "step": 27799
-    },
-    {
-      "epoch": 14.498044328552803,
-      "grad_norm": 1.50397789478302,
-      "learning_rate": 7.257185929648242e-05,
-      "loss": 5.4377,
-      "step": 27800
-    },
-    {
-      "epoch": 14.498565840938722,
-      "grad_norm": 1.3630328178405762,
-      "learning_rate": 7.25708542713568e-05,
-      "loss": 5.6084,
-      "step": 27801
-    },
-    {
-      "epoch": 14.499087353324642,
-      "grad_norm": 1.5982531309127808,
-      "learning_rate": 7.256984924623116e-05,
-      "loss": 5.2986,
-      "step": 27802
-    },
-    {
-      "epoch": 14.499608865710561,
-      "grad_norm": 1.5371696949005127,
-      "learning_rate": 7.256884422110554e-05,
-      "loss": 5.079,
-      "step": 27803
-    },
-    {
-      "epoch": 14.500130378096479,
-      "grad_norm": 1.4569350481033325,
-      "learning_rate": 7.25678391959799e-05,
-      "loss": 4.8015,
-      "step": 27804
-    },
-    {
-      "epoch": 14.500651890482398,
-      "grad_norm": 1.5000020265579224,
-      "learning_rate": 7.256683417085427e-05,
-      "loss": 5.4879,
-      "step": 27805
-    },
-    {
-      "epoch": 14.501173402868318,
-      "grad_norm": 1.4567855596542358,
-      "learning_rate": 7.256582914572865e-05,
-      "loss": 5.8752,
-      "step": 27806
-    },
-    {
-      "epoch": 14.501694915254237,
-      "grad_norm": 1.4723840951919556,
-      "learning_rate": 7.256482412060301e-05,
-      "loss": 4.8792,
-      "step": 27807
-    },
-    {
-      "epoch": 14.502216427640157,
-      "grad_norm": 1.4552149772644043,
-      "learning_rate": 7.256381909547739e-05,
-      "loss": 5.5449,
-      "step": 27808
-    },
-    {
-      "epoch": 14.502737940026076,
-      "grad_norm": 1.67232346534729,
-      "learning_rate": 7.256281407035177e-05,
-      "loss": 4.5688,
-      "step": 27809
-    },
-    {
-      "epoch": 14.503259452411994,
-      "grad_norm": 1.5669032335281372,
-      "learning_rate": 7.256180904522614e-05,
-      "loss": 5.0531,
-      "step": 27810
-    },
-    {
-      "epoch": 14.503780964797913,
-      "grad_norm": 1.624675989151001,
-      "learning_rate": 7.256080402010051e-05,
-      "loss": 5.0836,
-      "step": 27811
-    },
-    {
-      "epoch": 14.504302477183833,
-      "grad_norm": 1.6131874322891235,
-      "learning_rate": 7.255979899497489e-05,
-      "loss": 4.8233,
-      "step": 27812
-    },
-    {
-      "epoch": 14.504823989569752,
-      "grad_norm": 1.8054813146591187,
-      "learning_rate": 7.255879396984925e-05,
-      "loss": 5.3462,
-      "step": 27813
-    },
-    {
-      "epoch": 14.505345501955672,
-      "grad_norm": 1.4859895706176758,
-      "learning_rate": 7.255778894472363e-05,
-      "loss": 5.3245,
-      "step": 27814
-    },
-    {
-      "epoch": 14.505867014341591,
-      "grad_norm": 1.5362536907196045,
-      "learning_rate": 7.255678391959799e-05,
-      "loss": 5.3637,
-      "step": 27815
-    },
-    {
-      "epoch": 14.506388526727509,
-      "grad_norm": 1.4109423160552979,
-      "learning_rate": 7.255577889447237e-05,
-      "loss": 4.4423,
-      "step": 27816
-    },
-    {
-      "epoch": 14.506910039113428,
-      "grad_norm": 1.3675345182418823,
-      "learning_rate": 7.255477386934673e-05,
-      "loss": 5.6,
-      "step": 27817
-    },
-    {
-      "epoch": 14.507431551499348,
-      "grad_norm": 1.4735568761825562,
-      "learning_rate": 7.25537688442211e-05,
-      "loss": 5.1116,
-      "step": 27818
-    },
-    {
-      "epoch": 14.507953063885267,
-      "grad_norm": 1.4379968643188477,
-      "learning_rate": 7.255276381909548e-05,
-      "loss": 5.5676,
-      "step": 27819
-    },
-    {
-      "epoch": 14.508474576271187,
-      "grad_norm": 1.4421945810317993,
-      "learning_rate": 7.255175879396985e-05,
-      "loss": 5.6092,
-      "step": 27820
-    },
-    {
-      "epoch": 14.508996088657106,
-      "grad_norm": 1.5194189548492432,
-      "learning_rate": 7.255075376884423e-05,
-      "loss": 4.6623,
-      "step": 27821
-    },
-    {
-      "epoch": 14.509517601043024,
-      "grad_norm": 1.51508629322052,
-      "learning_rate": 7.25497487437186e-05,
-      "loss": 5.0365,
-      "step": 27822
-    },
-    {
-      "epoch": 14.510039113428943,
-      "grad_norm": 1.3899062871932983,
-      "learning_rate": 7.254874371859297e-05,
-      "loss": 5.75,
-      "step": 27823
-    },
-    {
-      "epoch": 14.510560625814863,
-      "grad_norm": 1.4113346338272095,
-      "learning_rate": 7.254773869346734e-05,
-      "loss": 5.5303,
-      "step": 27824
-    },
-    {
-      "epoch": 14.511082138200782,
-      "grad_norm": 1.4003528356552124,
-      "learning_rate": 7.254673366834172e-05,
-      "loss": 5.6023,
-      "step": 27825
-    },
-    {
-      "epoch": 14.511603650586702,
-      "grad_norm": 1.4544216394424438,
-      "learning_rate": 7.254572864321608e-05,
-      "loss": 5.5838,
-      "step": 27826
-    },
-    {
-      "epoch": 14.512125162972621,
-      "grad_norm": 1.4981330633163452,
-      "learning_rate": 7.254472361809046e-05,
-      "loss": 5.3384,
-      "step": 27827
-    },
-    {
-      "epoch": 14.512646675358539,
-      "grad_norm": 1.5760358572006226,
-      "learning_rate": 7.254371859296482e-05,
-      "loss": 5.0716,
-      "step": 27828
-    },
-    {
-      "epoch": 14.513168187744458,
-      "grad_norm": 1.6229082345962524,
-      "learning_rate": 7.25427135678392e-05,
-      "loss": 4.9964,
-      "step": 27829
-    },
-    {
-      "epoch": 14.513689700130378,
-      "grad_norm": 1.446816325187683,
-      "learning_rate": 7.254170854271356e-05,
-      "loss": 5.4483,
-      "step": 27830
-    },
-    {
-      "epoch": 14.514211212516297,
-      "grad_norm": 1.5357820987701416,
-      "learning_rate": 7.254070351758794e-05,
-      "loss": 5.3868,
-      "step": 27831
-    },
-    {
-      "epoch": 14.514732724902217,
-      "grad_norm": 1.4244036674499512,
-      "learning_rate": 7.253969849246232e-05,
-      "loss": 5.6825,
-      "step": 27832
-    },
-    {
-      "epoch": 14.515254237288136,
-      "grad_norm": 1.4912716150283813,
-      "learning_rate": 7.253869346733668e-05,
-      "loss": 5.4263,
-      "step": 27833
-    },
-    {
-      "epoch": 14.515775749674054,
-      "grad_norm": 1.598034381866455,
-      "learning_rate": 7.253768844221106e-05,
-      "loss": 4.7447,
-      "step": 27834
-    },
-    {
-      "epoch": 14.516297262059974,
-      "grad_norm": 1.5911602973937988,
-      "learning_rate": 7.253668341708543e-05,
-      "loss": 5.3774,
-      "step": 27835
-    },
-    {
-      "epoch": 14.516818774445893,
-      "grad_norm": 1.3794102668762207,
-      "learning_rate": 7.25356783919598e-05,
-      "loss": 5.7125,
-      "step": 27836
-    },
-    {
-      "epoch": 14.517340286831812,
-      "grad_norm": 1.4820152521133423,
-      "learning_rate": 7.253467336683417e-05,
-      "loss": 5.3123,
-      "step": 27837
-    },
-    {
-      "epoch": 14.517861799217732,
-      "grad_norm": 1.3321194648742676,
-      "learning_rate": 7.253366834170855e-05,
-      "loss": 5.9709,
-      "step": 27838
-    },
-    {
-      "epoch": 14.518383311603651,
-      "grad_norm": 1.422397255897522,
-      "learning_rate": 7.253266331658291e-05,
-      "loss": 5.601,
-      "step": 27839
-    },
-    {
-      "epoch": 14.518904823989569,
-      "grad_norm": 1.490161418914795,
-      "learning_rate": 7.253165829145729e-05,
-      "loss": 4.9152,
-      "step": 27840
-    },
-    {
-      "epoch": 14.519426336375489,
-      "grad_norm": 1.3901324272155762,
-      "learning_rate": 7.253065326633167e-05,
-      "loss": 5.7987,
-      "step": 27841
-    },
-    {
-      "epoch": 14.519947848761408,
-      "grad_norm": 1.433100938796997,
-      "learning_rate": 7.252964824120604e-05,
-      "loss": 4.9897,
-      "step": 27842
-    },
-    {
-      "epoch": 14.520469361147327,
-      "grad_norm": 1.4769761562347412,
-      "learning_rate": 7.252864321608041e-05,
-      "loss": 5.6884,
-      "step": 27843
-    },
-    {
-      "epoch": 14.520990873533247,
-      "grad_norm": 1.503115177154541,
-      "learning_rate": 7.252763819095477e-05,
-      "loss": 5.5578,
-      "step": 27844
-    },
-    {
-      "epoch": 14.521512385919166,
-      "grad_norm": 1.3669610023498535,
-      "learning_rate": 7.252663316582915e-05,
-      "loss": 5.6578,
-      "step": 27845
-    },
-    {
-      "epoch": 14.522033898305084,
-      "grad_norm": 1.4712785482406616,
-      "learning_rate": 7.252562814070351e-05,
-      "loss": 5.5068,
-      "step": 27846
-    },
-    {
-      "epoch": 14.522555410691004,
-      "grad_norm": 1.4427900314331055,
-      "learning_rate": 7.252462311557789e-05,
-      "loss": 5.7548,
-      "step": 27847
-    },
-    {
-      "epoch": 14.523076923076923,
-      "grad_norm": 1.4779796600341797,
-      "learning_rate": 7.252361809045226e-05,
-      "loss": 5.3225,
-      "step": 27848
-    },
-    {
-      "epoch": 14.523598435462842,
-      "grad_norm": 1.3925886154174805,
-      "learning_rate": 7.252261306532663e-05,
-      "loss": 5.3753,
-      "step": 27849
-    },
-    {
-      "epoch": 14.524119947848762,
-      "grad_norm": 1.4911679029464722,
-      "learning_rate": 7.2521608040201e-05,
-      "loss": 4.8647,
-      "step": 27850
-    },
-    {
-      "epoch": 14.524641460234681,
-      "grad_norm": 1.3647923469543457,
-      "learning_rate": 7.252060301507538e-05,
-      "loss": 5.3891,
-      "step": 27851
-    },
-    {
-      "epoch": 14.525162972620599,
-      "grad_norm": 1.4012115001678467,
-      "learning_rate": 7.251959798994975e-05,
-      "loss": 5.3568,
-      "step": 27852
-    },
-    {
-      "epoch": 14.525684485006519,
-      "grad_norm": 1.6014699935913086,
-      "learning_rate": 7.251859296482413e-05,
-      "loss": 5.3243,
-      "step": 27853
-    },
-    {
-      "epoch": 14.526205997392438,
-      "grad_norm": 1.5075442790985107,
-      "learning_rate": 7.25175879396985e-05,
-      "loss": 5.085,
-      "step": 27854
-    },
-    {
-      "epoch": 14.526727509778357,
-      "grad_norm": 1.4349801540374756,
-      "learning_rate": 7.251658291457287e-05,
-      "loss": 5.7092,
-      "step": 27855
-    },
-    {
-      "epoch": 14.527249022164277,
-      "grad_norm": 1.4905165433883667,
-      "learning_rate": 7.251557788944724e-05,
-      "loss": 5.3671,
-      "step": 27856
-    },
-    {
-      "epoch": 14.527770534550196,
-      "grad_norm": 1.3983147144317627,
-      "learning_rate": 7.25145728643216e-05,
-      "loss": 5.5352,
-      "step": 27857
-    },
-    {
-      "epoch": 14.528292046936114,
-      "grad_norm": 1.4585809707641602,
-      "learning_rate": 7.251356783919598e-05,
-      "loss": 5.6875,
-      "step": 27858
-    },
-    {
-      "epoch": 14.528813559322034,
-      "grad_norm": 1.388298511505127,
-      "learning_rate": 7.251256281407035e-05,
-      "loss": 5.7522,
-      "step": 27859
-    },
-    {
-      "epoch": 14.529335071707953,
-      "grad_norm": 1.3781304359436035,
-      "learning_rate": 7.251155778894472e-05,
-      "loss": 5.6294,
-      "step": 27860
-    },
-    {
-      "epoch": 14.529856584093872,
-      "grad_norm": 1.3912336826324463,
-      "learning_rate": 7.25105527638191e-05,
-      "loss": 5.5743,
-      "step": 27861
-    },
-    {
-      "epoch": 14.530378096479792,
-      "grad_norm": 1.3602464199066162,
-      "learning_rate": 7.250954773869348e-05,
-      "loss": 5.5523,
-      "step": 27862
-    },
-    {
-      "epoch": 14.53089960886571,
-      "grad_norm": 1.44057297706604,
-      "learning_rate": 7.250854271356784e-05,
-      "loss": 5.3145,
-      "step": 27863
-    },
-    {
-      "epoch": 14.531421121251629,
-      "grad_norm": 1.3622037172317505,
-      "learning_rate": 7.250753768844222e-05,
-      "loss": 5.3622,
-      "step": 27864
-    },
-    {
-      "epoch": 14.531942633637549,
-      "grad_norm": 1.4691115617752075,
-      "learning_rate": 7.250653266331659e-05,
-      "loss": 5.128,
-      "step": 27865
-    },
-    {
-      "epoch": 14.532464146023468,
-      "grad_norm": 1.517576813697815,
-      "learning_rate": 7.250552763819096e-05,
-      "loss": 5.3568,
-      "step": 27866
-    },
-    {
-      "epoch": 14.532985658409387,
-      "grad_norm": 1.517526388168335,
-      "learning_rate": 7.250452261306533e-05,
-      "loss": 5.0675,
-      "step": 27867
-    },
-    {
-      "epoch": 14.533507170795307,
-      "grad_norm": 1.4617775678634644,
-      "learning_rate": 7.25035175879397e-05,
-      "loss": 5.5341,
-      "step": 27868
-    },
-    {
-      "epoch": 14.534028683181226,
-      "grad_norm": 1.3911468982696533,
-      "learning_rate": 7.250251256281407e-05,
-      "loss": 5.5231,
-      "step": 27869
-    },
-    {
-      "epoch": 14.534550195567144,
-      "grad_norm": 1.455308437347412,
-      "learning_rate": 7.250150753768845e-05,
-      "loss": 5.4666,
-      "step": 27870
-    },
-    {
-      "epoch": 14.535071707953064,
-      "grad_norm": 1.491706132888794,
-      "learning_rate": 7.250050251256281e-05,
-      "loss": 4.9408,
-      "step": 27871
-    },
-    {
-      "epoch": 14.535593220338983,
-      "grad_norm": 1.7374087572097778,
-      "learning_rate": 7.249949748743719e-05,
-      "loss": 5.1516,
-      "step": 27872
-    },
-    {
-      "epoch": 14.536114732724903,
-      "grad_norm": 1.5148465633392334,
-      "learning_rate": 7.249849246231157e-05,
-      "loss": 5.6063,
-      "step": 27873
-    },
-    {
-      "epoch": 14.536636245110822,
-      "grad_norm": 1.473503828048706,
-      "learning_rate": 7.249748743718593e-05,
-      "loss": 5.3915,
-      "step": 27874
-    },
-    {
-      "epoch": 14.53715775749674,
-      "grad_norm": 1.6457650661468506,
-      "learning_rate": 7.249648241206031e-05,
-      "loss": 4.393,
-      "step": 27875
-    },
-    {
-      "epoch": 14.53767926988266,
-      "grad_norm": 1.3603637218475342,
-      "learning_rate": 7.249547738693467e-05,
-      "loss": 5.8615,
-      "step": 27876
-    },
-    {
-      "epoch": 14.538200782268579,
-      "grad_norm": 1.487370252609253,
-      "learning_rate": 7.249447236180905e-05,
-      "loss": 5.1896,
-      "step": 27877
-    },
-    {
-      "epoch": 14.538722294654498,
-      "grad_norm": 1.465965747833252,
-      "learning_rate": 7.249346733668342e-05,
-      "loss": 5.1027,
-      "step": 27878
-    },
-    {
-      "epoch": 14.539243807040418,
-      "grad_norm": 1.4736028909683228,
-      "learning_rate": 7.24924623115578e-05,
-      "loss": 5.6157,
-      "step": 27879
-    },
-    {
-      "epoch": 14.539765319426337,
-      "grad_norm": 1.4438899755477905,
-      "learning_rate": 7.249145728643216e-05,
-      "loss": 5.7896,
-      "step": 27880
-    },
-    {
-      "epoch": 14.540286831812256,
-      "grad_norm": 1.4497333765029907,
-      "learning_rate": 7.249045226130654e-05,
-      "loss": 5.4304,
-      "step": 27881
-    },
-    {
-      "epoch": 14.540808344198174,
-      "grad_norm": 1.4916462898254395,
-      "learning_rate": 7.248944723618091e-05,
-      "loss": 4.9061,
-      "step": 27882
-    },
-    {
-      "epoch": 14.541329856584094,
-      "grad_norm": 1.5112353563308716,
-      "learning_rate": 7.248844221105529e-05,
-      "loss": 4.361,
-      "step": 27883
-    },
-    {
-      "epoch": 14.541851368970013,
-      "grad_norm": 1.4771455526351929,
-      "learning_rate": 7.248743718592966e-05,
-      "loss": 5.015,
-      "step": 27884
-    },
-    {
-      "epoch": 14.542372881355933,
-      "grad_norm": 1.4780690670013428,
-      "learning_rate": 7.248643216080402e-05,
-      "loss": 5.028,
-      "step": 27885
-    },
-    {
-      "epoch": 14.542894393741852,
-      "grad_norm": 1.5131094455718994,
-      "learning_rate": 7.24854271356784e-05,
-      "loss": 5.5264,
-      "step": 27886
-    },
-    {
-      "epoch": 14.54341590612777,
-      "grad_norm": 1.3705509901046753,
-      "learning_rate": 7.248442211055276e-05,
-      "loss": 4.5952,
-      "step": 27887
-    },
-    {
-      "epoch": 14.54393741851369,
-      "grad_norm": 1.5237494707107544,
-      "learning_rate": 7.248341708542714e-05,
-      "loss": 5.5806,
-      "step": 27888
-    },
-    {
-      "epoch": 14.544458930899609,
-      "grad_norm": 1.4257622957229614,
-      "learning_rate": 7.24824120603015e-05,
-      "loss": 5.7374,
-      "step": 27889
-    },
-    {
-      "epoch": 14.544980443285528,
-      "grad_norm": 1.4877480268478394,
-      "learning_rate": 7.248140703517588e-05,
-      "loss": 5.2382,
-      "step": 27890
-    },
-    {
-      "epoch": 14.545501955671448,
-      "grad_norm": 1.4090967178344727,
-      "learning_rate": 7.248040201005025e-05,
-      "loss": 5.6638,
-      "step": 27891
-    },
-    {
-      "epoch": 14.546023468057367,
-      "grad_norm": 1.396342158317566,
-      "learning_rate": 7.247939698492462e-05,
-      "loss": 5.3175,
-      "step": 27892
-    },
-    {
-      "epoch": 14.546544980443285,
-      "grad_norm": 1.5543386936187744,
-      "learning_rate": 7.2478391959799e-05,
-      "loss": 5.3827,
-      "step": 27893
-    },
-    {
-      "epoch": 14.547066492829204,
-      "grad_norm": 1.338243842124939,
-      "learning_rate": 7.247738693467338e-05,
-      "loss": 5.4582,
-      "step": 27894
-    },
-    {
-      "epoch": 14.547588005215124,
-      "grad_norm": 1.363004207611084,
-      "learning_rate": 7.247638190954774e-05,
-      "loss": 4.5047,
-      "step": 27895
-    },
-    {
-      "epoch": 14.548109517601043,
-      "grad_norm": 1.4882783889770508,
-      "learning_rate": 7.247537688442212e-05,
-      "loss": 5.3893,
-      "step": 27896
-    },
-    {
-      "epoch": 14.548631029986963,
-      "grad_norm": 1.4327970743179321,
-      "learning_rate": 7.247437185929649e-05,
-      "loss": 5.5993,
-      "step": 27897
-    },
-    {
-      "epoch": 14.549152542372882,
-      "grad_norm": 1.4961345195770264,
-      "learning_rate": 7.247336683417085e-05,
-      "loss": 5.3277,
-      "step": 27898
-    },
-    {
-      "epoch": 14.5496740547588,
-      "grad_norm": 1.5322325229644775,
-      "learning_rate": 7.247236180904523e-05,
-      "loss": 5.5578,
-      "step": 27899
-    },
-    {
-      "epoch": 14.55019556714472,
-      "grad_norm": 1.392107605934143,
-      "learning_rate": 7.247135678391959e-05,
-      "loss": 5.3594,
-      "step": 27900
-    },
-    {
-      "epoch": 14.550717079530639,
-      "grad_norm": 1.3935805559158325,
-      "learning_rate": 7.247035175879397e-05,
-      "loss": 5.2738,
-      "step": 27901
-    },
-    {
-      "epoch": 14.551238591916558,
-      "grad_norm": 1.33814537525177,
-      "learning_rate": 7.246934673366835e-05,
-      "loss": 5.8979,
-      "step": 27902
-    },
-    {
-      "epoch": 14.551760104302478,
-      "grad_norm": 1.4824793338775635,
-      "learning_rate": 7.246834170854273e-05,
-      "loss": 4.9924,
-      "step": 27903
-    },
-    {
-      "epoch": 14.552281616688397,
-      "grad_norm": 1.558030605316162,
-      "learning_rate": 7.246733668341709e-05,
-      "loss": 4.7875,
-      "step": 27904
-    },
-    {
-      "epoch": 14.552803129074315,
-      "grad_norm": 1.576809048652649,
-      "learning_rate": 7.246633165829147e-05,
-      "loss": 4.8173,
-      "step": 27905
-    },
-    {
-      "epoch": 14.553324641460234,
-      "grad_norm": 1.461898684501648,
-      "learning_rate": 7.246532663316583e-05,
-      "loss": 5.3971,
-      "step": 27906
-    },
-    {
-      "epoch": 14.553846153846154,
-      "grad_norm": 1.554019808769226,
-      "learning_rate": 7.246432160804021e-05,
-      "loss": 5.6169,
-      "step": 27907
-    },
-    {
-      "epoch": 14.554367666232073,
-      "grad_norm": 1.4505081176757812,
-      "learning_rate": 7.246331658291457e-05,
-      "loss": 5.617,
-      "step": 27908
-    },
-    {
-      "epoch": 14.554889178617993,
-      "grad_norm": 1.6023842096328735,
-      "learning_rate": 7.246231155778895e-05,
-      "loss": 5.3943,
-      "step": 27909
-    },
-    {
-      "epoch": 14.555410691003912,
-      "grad_norm": 1.5015760660171509,
-      "learning_rate": 7.246130653266332e-05,
-      "loss": 5.5402,
-      "step": 27910
-    },
-    {
-      "epoch": 14.55593220338983,
-      "grad_norm": 1.3955868482589722,
-      "learning_rate": 7.246030150753768e-05,
-      "loss": 5.6496,
-      "step": 27911
-    },
-    {
-      "epoch": 14.55645371577575,
-      "grad_norm": 1.4908267259597778,
-      "learning_rate": 7.245929648241206e-05,
-      "loss": 5.5018,
-      "step": 27912
-    },
-    {
-      "epoch": 14.556975228161669,
-      "grad_norm": 1.5049296617507935,
-      "learning_rate": 7.245829145728644e-05,
-      "loss": 5.5048,
-      "step": 27913
-    },
-    {
-      "epoch": 14.557496740547588,
-      "grad_norm": 1.4449880123138428,
-      "learning_rate": 7.245728643216081e-05,
-      "loss": 5.2626,
-      "step": 27914
-    },
-    {
-      "epoch": 14.558018252933508,
-      "grad_norm": 1.5681626796722412,
-      "learning_rate": 7.245628140703518e-05,
-      "loss": 5.206,
-      "step": 27915
-    },
-    {
-      "epoch": 14.558539765319427,
-      "grad_norm": 1.46823251247406,
-      "learning_rate": 7.245527638190956e-05,
-      "loss": 4.668,
-      "step": 27916
-    },
-    {
-      "epoch": 14.559061277705345,
-      "grad_norm": 1.4520922899246216,
-      "learning_rate": 7.245427135678392e-05,
-      "loss": 5.4373,
-      "step": 27917
-    },
-    {
-      "epoch": 14.559582790091264,
-      "grad_norm": 1.5355932712554932,
-      "learning_rate": 7.24532663316583e-05,
-      "loss": 5.1388,
-      "step": 27918
-    },
-    {
-      "epoch": 14.560104302477184,
-      "grad_norm": 1.5628015995025635,
-      "learning_rate": 7.245226130653266e-05,
-      "loss": 5.2129,
-      "step": 27919
-    },
-    {
-      "epoch": 14.560625814863103,
-      "grad_norm": 1.5367755889892578,
-      "learning_rate": 7.245125628140704e-05,
-      "loss": 5.3881,
-      "step": 27920
-    },
-    {
-      "epoch": 14.561147327249023,
-      "grad_norm": 1.5808110237121582,
-      "learning_rate": 7.24502512562814e-05,
-      "loss": 5.1993,
-      "step": 27921
-    },
-    {
-      "epoch": 14.561668839634942,
-      "grad_norm": 1.5087239742279053,
-      "learning_rate": 7.244924623115578e-05,
-      "loss": 4.9104,
-      "step": 27922
-    },
-    {
-      "epoch": 14.56219035202086,
-      "grad_norm": 1.5801717042922974,
-      "learning_rate": 7.244824120603016e-05,
-      "loss": 5.0842,
-      "step": 27923
-    },
-    {
-      "epoch": 14.56271186440678,
-      "grad_norm": 1.4876720905303955,
-      "learning_rate": 7.244723618090452e-05,
-      "loss": 5.3273,
-      "step": 27924
-    },
-    {
-      "epoch": 14.563233376792699,
-      "grad_norm": 1.3276501893997192,
-      "learning_rate": 7.24462311557789e-05,
-      "loss": 5.6848,
-      "step": 27925
-    },
-    {
-      "epoch": 14.563754889178618,
-      "grad_norm": 1.508569359779358,
-      "learning_rate": 7.244522613065327e-05,
-      "loss": 5.2429,
-      "step": 27926
-    },
-    {
-      "epoch": 14.564276401564538,
-      "grad_norm": 1.599200963973999,
-      "learning_rate": 7.244422110552764e-05,
-      "loss": 4.6345,
-      "step": 27927
-    },
-    {
-      "epoch": 14.564797913950457,
-      "grad_norm": 1.6112041473388672,
-      "learning_rate": 7.244321608040201e-05,
-      "loss": 5.6246,
-      "step": 27928
-    },
-    {
-      "epoch": 14.565319426336375,
-      "grad_norm": 1.4979783296585083,
-      "learning_rate": 7.244221105527639e-05,
-      "loss": 5.3348,
-      "step": 27929
-    },
-    {
-      "epoch": 14.565840938722294,
-      "grad_norm": 1.4365742206573486,
-      "learning_rate": 7.244120603015075e-05,
-      "loss": 5.3314,
-      "step": 27930
-    },
-    {
-      "epoch": 14.566362451108214,
-      "grad_norm": 1.4465360641479492,
-      "learning_rate": 7.244020100502513e-05,
-      "loss": 5.34,
-      "step": 27931
-    },
-    {
-      "epoch": 14.566883963494133,
-      "grad_norm": 1.3453075885772705,
-      "learning_rate": 7.243919597989949e-05,
-      "loss": 5.1277,
-      "step": 27932
-    },
-    {
-      "epoch": 14.567405475880053,
-      "grad_norm": 1.4929431676864624,
-      "learning_rate": 7.243819095477387e-05,
-      "loss": 5.0886,
-      "step": 27933
-    },
-    {
-      "epoch": 14.567926988265972,
-      "grad_norm": 1.4714179039001465,
-      "learning_rate": 7.243718592964825e-05,
-      "loss": 5.7767,
-      "step": 27934
-    },
-    {
-      "epoch": 14.56844850065189,
-      "grad_norm": 1.4937664270401,
-      "learning_rate": 7.243618090452263e-05,
-      "loss": 5.5076,
-      "step": 27935
-    },
-    {
-      "epoch": 14.56897001303781,
-      "grad_norm": 1.4846196174621582,
-      "learning_rate": 7.243517587939699e-05,
-      "loss": 5.0914,
-      "step": 27936
-    },
-    {
-      "epoch": 14.569491525423729,
-      "grad_norm": 1.4035298824310303,
-      "learning_rate": 7.243417085427136e-05,
-      "loss": 5.4061,
-      "step": 27937
-    },
-    {
-      "epoch": 14.570013037809648,
-      "grad_norm": 1.4894604682922363,
-      "learning_rate": 7.243316582914573e-05,
-      "loss": 5.2096,
-      "step": 27938
-    },
-    {
-      "epoch": 14.570534550195568,
-      "grad_norm": 1.3579086065292358,
-      "learning_rate": 7.24321608040201e-05,
-      "loss": 5.5129,
-      "step": 27939
-    },
-    {
-      "epoch": 14.571056062581487,
-      "grad_norm": 1.377989649772644,
-      "learning_rate": 7.243115577889448e-05,
-      "loss": 5.3849,
-      "step": 27940
-    },
-    {
-      "epoch": 14.571577574967405,
-      "grad_norm": 1.4203096628189087,
-      "learning_rate": 7.243015075376884e-05,
-      "loss": 5.2194,
-      "step": 27941
-    },
-    {
-      "epoch": 14.572099087353324,
-      "grad_norm": 1.523910641670227,
-      "learning_rate": 7.242914572864322e-05,
-      "loss": 5.6717,
-      "step": 27942
-    },
-    {
-      "epoch": 14.572620599739244,
-      "grad_norm": 1.5797836780548096,
-      "learning_rate": 7.24281407035176e-05,
-      "loss": 5.4903,
-      "step": 27943
-    },
-    {
-      "epoch": 14.573142112125163,
-      "grad_norm": 1.4015170335769653,
-      "learning_rate": 7.242713567839197e-05,
-      "loss": 5.3102,
-      "step": 27944
-    },
-    {
-      "epoch": 14.573663624511083,
-      "grad_norm": 1.5849860906600952,
-      "learning_rate": 7.242613065326634e-05,
-      "loss": 5.1985,
-      "step": 27945
-    },
-    {
-      "epoch": 14.574185136897002,
-      "grad_norm": 1.4416043758392334,
-      "learning_rate": 7.242512562814071e-05,
-      "loss": 5.4247,
-      "step": 27946
-    },
-    {
-      "epoch": 14.57470664928292,
-      "grad_norm": 1.418518304824829,
-      "learning_rate": 7.242412060301508e-05,
-      "loss": 5.5171,
-      "step": 27947
-    },
-    {
-      "epoch": 14.57522816166884,
-      "grad_norm": 1.352144718170166,
-      "learning_rate": 7.242311557788946e-05,
-      "loss": 5.6436,
-      "step": 27948
-    },
-    {
-      "epoch": 14.575749674054759,
-      "grad_norm": 1.4197378158569336,
-      "learning_rate": 7.242211055276382e-05,
-      "loss": 5.5356,
-      "step": 27949
-    },
-    {
-      "epoch": 14.576271186440678,
-      "grad_norm": 1.3870290517807007,
-      "learning_rate": 7.242110552763819e-05,
-      "loss": 5.6428,
-      "step": 27950
-    },
-    {
-      "epoch": 14.576792698826598,
-      "grad_norm": 1.4848483800888062,
-      "learning_rate": 7.242010050251256e-05,
-      "loss": 5.5096,
-      "step": 27951
-    },
-    {
-      "epoch": 14.577314211212517,
-      "grad_norm": 1.5048930644989014,
-      "learning_rate": 7.241909547738693e-05,
-      "loss": 5.238,
-      "step": 27952
-    },
-    {
-      "epoch": 14.577835723598435,
-      "grad_norm": 1.433355450630188,
-      "learning_rate": 7.24180904522613e-05,
-      "loss": 5.1509,
-      "step": 27953
-    },
-    {
-      "epoch": 14.578357235984354,
-      "grad_norm": 1.510019063949585,
-      "learning_rate": 7.241708542713568e-05,
-      "loss": 4.6101,
-      "step": 27954
-    },
-    {
-      "epoch": 14.578878748370274,
-      "grad_norm": 1.5131990909576416,
-      "learning_rate": 7.241608040201006e-05,
-      "loss": 5.5674,
-      "step": 27955
-    },
-    {
-      "epoch": 14.579400260756193,
-      "grad_norm": 1.5114942789077759,
-      "learning_rate": 7.241507537688443e-05,
-      "loss": 5.7286,
-      "step": 27956
-    },
-    {
-      "epoch": 14.579921773142113,
-      "grad_norm": 1.5068049430847168,
-      "learning_rate": 7.24140703517588e-05,
-      "loss": 5.3119,
-      "step": 27957
-    },
-    {
-      "epoch": 14.58044328552803,
-      "grad_norm": 1.4744279384613037,
-      "learning_rate": 7.241306532663317e-05,
-      "loss": 4.969,
-      "step": 27958
-    },
-    {
-      "epoch": 14.58096479791395,
-      "grad_norm": 1.3491156101226807,
-      "learning_rate": 7.241206030150755e-05,
-      "loss": 5.4698,
-      "step": 27959
-    },
-    {
-      "epoch": 14.58148631029987,
-      "grad_norm": 1.4405107498168945,
-      "learning_rate": 7.241105527638191e-05,
-      "loss": 5.4082,
-      "step": 27960
-    },
-    {
-      "epoch": 14.582007822685789,
-      "grad_norm": 1.4494799375534058,
-      "learning_rate": 7.241005025125629e-05,
-      "loss": 5.5944,
-      "step": 27961
-    },
-    {
-      "epoch": 14.582529335071708,
-      "grad_norm": 1.5520588159561157,
-      "learning_rate": 7.240904522613065e-05,
-      "loss": 5.1959,
-      "step": 27962
-    },
-    {
-      "epoch": 14.583050847457628,
-      "grad_norm": 1.4303780794143677,
-      "learning_rate": 7.240804020100503e-05,
-      "loss": 5.5314,
-      "step": 27963
-    },
-    {
-      "epoch": 14.583572359843547,
-      "grad_norm": 1.3896713256835938,
-      "learning_rate": 7.240703517587941e-05,
-      "loss": 5.5401,
-      "step": 27964
-    },
-    {
-      "epoch": 14.584093872229465,
-      "grad_norm": 1.3579075336456299,
-      "learning_rate": 7.240603015075377e-05,
-      "loss": 5.5182,
-      "step": 27965
-    },
-    {
-      "epoch": 14.584615384615384,
-      "grad_norm": 1.407112956047058,
-      "learning_rate": 7.240502512562815e-05,
-      "loss": 5.2261,
-      "step": 27966
-    },
-    {
-      "epoch": 14.585136897001304,
-      "grad_norm": 1.5164244174957275,
-      "learning_rate": 7.240402010050251e-05,
-      "loss": 5.3393,
-      "step": 27967
-    },
-    {
-      "epoch": 14.585658409387223,
-      "grad_norm": 2.1050806045532227,
-      "learning_rate": 7.240301507537689e-05,
-      "loss": 4.6748,
-      "step": 27968
-    },
-    {
-      "epoch": 14.586179921773143,
-      "grad_norm": 1.4323612451553345,
-      "learning_rate": 7.240201005025126e-05,
-      "loss": 5.537,
-      "step": 27969
-    },
-    {
-      "epoch": 14.58670143415906,
-      "grad_norm": 1.5500739812850952,
-      "learning_rate": 7.240100502512563e-05,
-      "loss": 4.8334,
-      "step": 27970
-    },
-    {
-      "epoch": 14.58722294654498,
-      "grad_norm": 1.4492448568344116,
-      "learning_rate": 7.24e-05,
-      "loss": 5.1291,
-      "step": 27971
-    },
-    {
-      "epoch": 14.5877444589309,
-      "grad_norm": 1.4046571254730225,
-      "learning_rate": 7.239899497487438e-05,
-      "loss": 5.5494,
-      "step": 27972
-    },
-    {
-      "epoch": 14.588265971316819,
-      "grad_norm": 1.5623537302017212,
-      "learning_rate": 7.239798994974874e-05,
-      "loss": 5.3821,
-      "step": 27973
-    },
-    {
-      "epoch": 14.588787483702738,
-      "grad_norm": 1.654383897781372,
-      "learning_rate": 7.239698492462312e-05,
-      "loss": 4.9454,
-      "step": 27974
-    },
-    {
-      "epoch": 14.589308996088658,
-      "grad_norm": 1.543915033340454,
-      "learning_rate": 7.23959798994975e-05,
-      "loss": 5.3267,
-      "step": 27975
-    },
-    {
-      "epoch": 14.589830508474577,
-      "grad_norm": 1.6614588499069214,
-      "learning_rate": 7.239497487437187e-05,
-      "loss": 4.6466,
-      "step": 27976
-    },
-    {
-      "epoch": 14.590352020860495,
-      "grad_norm": 1.5977033376693726,
-      "learning_rate": 7.239396984924624e-05,
-      "loss": 5.103,
-      "step": 27977
-    },
-    {
-      "epoch": 14.590873533246414,
-      "grad_norm": 1.4357144832611084,
-      "learning_rate": 7.23929648241206e-05,
-      "loss": 5.5372,
-      "step": 27978
-    },
-    {
-      "epoch": 14.591395045632334,
-      "grad_norm": 1.4727895259857178,
-      "learning_rate": 7.239195979899498e-05,
-      "loss": 5.6543,
-      "step": 27979
-    },
-    {
-      "epoch": 14.591916558018253,
-      "grad_norm": 1.4912731647491455,
-      "learning_rate": 7.239095477386934e-05,
-      "loss": 5.2778,
-      "step": 27980
-    },
-    {
-      "epoch": 14.592438070404173,
-      "grad_norm": 1.6156896352767944,
-      "learning_rate": 7.238994974874372e-05,
-      "loss": 4.9152,
-      "step": 27981
-    },
-    {
-      "epoch": 14.59295958279009,
-      "grad_norm": 1.6242166757583618,
-      "learning_rate": 7.238894472361809e-05,
-      "loss": 5.3867,
-      "step": 27982
-    },
-    {
-      "epoch": 14.59348109517601,
-      "grad_norm": 1.664057970046997,
-      "learning_rate": 7.238793969849246e-05,
-      "loss": 5.0071,
-      "step": 27983
-    },
-    {
-      "epoch": 14.59400260756193,
-      "grad_norm": 1.5329540967941284,
-      "learning_rate": 7.238693467336683e-05,
-      "loss": 5.2059,
-      "step": 27984
-    },
-    {
-      "epoch": 14.594524119947849,
-      "grad_norm": 1.5543251037597656,
-      "learning_rate": 7.23859296482412e-05,
-      "loss": 5.368,
-      "step": 27985
-    },
-    {
-      "epoch": 14.595045632333768,
-      "grad_norm": 1.5007320642471313,
-      "learning_rate": 7.238492462311558e-05,
-      "loss": 5.2482,
-      "step": 27986
-    },
-    {
-      "epoch": 14.595567144719688,
-      "grad_norm": 1.360280156135559,
-      "learning_rate": 7.238391959798996e-05,
-      "loss": 5.0333,
-      "step": 27987
-    },
-    {
-      "epoch": 14.596088657105605,
-      "grad_norm": 1.4257696866989136,
-      "learning_rate": 7.238291457286433e-05,
-      "loss": 5.1999,
-      "step": 27988
-    },
-    {
-      "epoch": 14.596610169491525,
-      "grad_norm": 1.5119869709014893,
-      "learning_rate": 7.23819095477387e-05,
-      "loss": 5.3479,
-      "step": 27989
-    },
-    {
-      "epoch": 14.597131681877444,
-      "grad_norm": 1.5788111686706543,
-      "learning_rate": 7.238090452261307e-05,
-      "loss": 5.5469,
-      "step": 27990
-    },
-    {
-      "epoch": 14.597653194263364,
-      "grad_norm": 1.430322527885437,
-      "learning_rate": 7.237989949748743e-05,
-      "loss": 5.5571,
-      "step": 27991
-    },
-    {
-      "epoch": 14.598174706649283,
-      "grad_norm": 1.4739853143692017,
-      "learning_rate": 7.237889447236181e-05,
-      "loss": 5.5132,
-      "step": 27992
-    },
-    {
-      "epoch": 14.598696219035203,
-      "grad_norm": 1.4764485359191895,
-      "learning_rate": 7.237788944723617e-05,
-      "loss": 5.4843,
-      "step": 27993
-    },
-    {
-      "epoch": 14.59921773142112,
-      "grad_norm": 1.467971920967102,
-      "learning_rate": 7.237688442211055e-05,
-      "loss": 5.6304,
-      "step": 27994
-    },
-    {
-      "epoch": 14.59973924380704,
-      "grad_norm": 1.476412057876587,
-      "learning_rate": 7.237587939698493e-05,
-      "loss": 5.7819,
-      "step": 27995
-    },
-    {
-      "epoch": 14.60026075619296,
-      "grad_norm": 1.605830192565918,
-      "learning_rate": 7.237487437185931e-05,
-      "loss": 5.0201,
-      "step": 27996
-    },
-    {
-      "epoch": 14.600782268578879,
-      "grad_norm": 1.4195753335952759,
-      "learning_rate": 7.237386934673367e-05,
-      "loss": 5.7424,
-      "step": 27997
-    },
-    {
-      "epoch": 14.601303780964798,
-      "grad_norm": 1.4435501098632812,
-      "learning_rate": 7.237286432160805e-05,
-      "loss": 4.9928,
-      "step": 27998
-    },
-    {
-      "epoch": 14.601825293350718,
-      "grad_norm": 1.5351214408874512,
-      "learning_rate": 7.237185929648241e-05,
-      "loss": 5.4706,
-      "step": 27999
-    },
-    {
-      "epoch": 14.602346805736635,
-      "grad_norm": 1.594736933708191,
-      "learning_rate": 7.237085427135679e-05,
-      "loss": 5.0199,
-      "step": 28000
-    },
-    {
-      "epoch": 14.602868318122555,
-      "grad_norm": 1.4189056158065796,
-      "learning_rate": 7.236984924623116e-05,
-      "loss": 5.5337,
-      "step": 28001
-    },
-    {
-      "epoch": 14.603389830508474,
-      "grad_norm": 1.446407437324524,
-      "learning_rate": 7.236884422110553e-05,
-      "loss": 5.4512,
-      "step": 28002
-    },
-    {
-      "epoch": 14.603911342894394,
-      "grad_norm": 1.5739566087722778,
-      "learning_rate": 7.23678391959799e-05,
-      "loss": 5.176,
-      "step": 28003
-    },
-    {
-      "epoch": 14.604432855280313,
-      "grad_norm": 1.5108952522277832,
-      "learning_rate": 7.236683417085426e-05,
-      "loss": 5.7527,
-      "step": 28004
-    },
-    {
-      "epoch": 14.604954367666233,
-      "grad_norm": 1.3973886966705322,
-      "learning_rate": 7.236582914572864e-05,
-      "loss": 5.2288,
-      "step": 28005
-    },
-    {
-      "epoch": 14.60547588005215,
-      "grad_norm": 1.4772357940673828,
-      "learning_rate": 7.236482412060302e-05,
-      "loss": 5.4711,
-      "step": 28006
-    },
-    {
-      "epoch": 14.60599739243807,
-      "grad_norm": 1.5998121500015259,
-      "learning_rate": 7.23638190954774e-05,
-      "loss": 5.323,
-      "step": 28007
-    },
-    {
-      "epoch": 14.60651890482399,
-      "grad_norm": 1.3847894668579102,
-      "learning_rate": 7.236281407035176e-05,
-      "loss": 5.6462,
-      "step": 28008
-    },
-    {
-      "epoch": 14.607040417209909,
-      "grad_norm": 1.4223586320877075,
-      "learning_rate": 7.236180904522614e-05,
-      "loss": 5.5919,
-      "step": 28009
-    },
-    {
-      "epoch": 14.607561929595828,
-      "grad_norm": 1.541958212852478,
-      "learning_rate": 7.23608040201005e-05,
-      "loss": 5.7123,
-      "step": 28010
-    },
-    {
-      "epoch": 14.608083441981748,
-      "grad_norm": 1.9094480276107788,
-      "learning_rate": 7.235979899497488e-05,
-      "loss": 5.4288,
-      "step": 28011
-    },
-    {
-      "epoch": 14.608604954367665,
-      "grad_norm": 1.5376721620559692,
-      "learning_rate": 7.235879396984924e-05,
-      "loss": 5.0312,
-      "step": 28012
-    },
-    {
-      "epoch": 14.609126466753585,
-      "grad_norm": 1.412558913230896,
-      "learning_rate": 7.235778894472362e-05,
-      "loss": 5.5718,
-      "step": 28013
-    },
-    {
-      "epoch": 14.609647979139504,
-      "grad_norm": 1.4906580448150635,
-      "learning_rate": 7.235678391959799e-05,
-      "loss": 5.5313,
-      "step": 28014
-    },
-    {
-      "epoch": 14.610169491525424,
-      "grad_norm": 1.4512227773666382,
-      "learning_rate": 7.235577889447236e-05,
-      "loss": 5.6135,
-      "step": 28015
-    },
-    {
-      "epoch": 14.610691003911343,
-      "grad_norm": 1.3743159770965576,
-      "learning_rate": 7.235477386934674e-05,
-      "loss": 5.8599,
-      "step": 28016
-    },
-    {
-      "epoch": 14.611212516297263,
-      "grad_norm": 1.5411745309829712,
-      "learning_rate": 7.235376884422111e-05,
-      "loss": 4.6358,
-      "step": 28017
-    },
-    {
-      "epoch": 14.61173402868318,
-      "grad_norm": 1.54520845413208,
-      "learning_rate": 7.235276381909548e-05,
-      "loss": 5.4504,
-      "step": 28018
-    },
-    {
-      "epoch": 14.6122555410691,
-      "grad_norm": 1.542235016822815,
-      "learning_rate": 7.235175879396985e-05,
-      "loss": 5.1504,
-      "step": 28019
-    },
-    {
-      "epoch": 14.61277705345502,
-      "grad_norm": 1.4786159992218018,
-      "learning_rate": 7.235075376884423e-05,
-      "loss": 5.3895,
-      "step": 28020
-    },
-    {
-      "epoch": 14.613298565840939,
-      "grad_norm": 1.4350595474243164,
-      "learning_rate": 7.234974874371859e-05,
-      "loss": 5.183,
-      "step": 28021
-    },
-    {
-      "epoch": 14.613820078226858,
-      "grad_norm": 1.527738094329834,
-      "learning_rate": 7.234874371859297e-05,
-      "loss": 5.3716,
-      "step": 28022
-    },
-    {
-      "epoch": 14.614341590612778,
-      "grad_norm": 1.5216654539108276,
-      "learning_rate": 7.234773869346733e-05,
-      "loss": 5.2096,
-      "step": 28023
-    },
-    {
-      "epoch": 14.614863102998696,
-      "grad_norm": 1.4527535438537598,
-      "learning_rate": 7.234673366834171e-05,
-      "loss": 5.6906,
-      "step": 28024
-    },
-    {
-      "epoch": 14.615384615384615,
-      "grad_norm": 1.5394896268844604,
-      "learning_rate": 7.234572864321608e-05,
-      "loss": 5.2779,
-      "step": 28025
-    },
-    {
-      "epoch": 14.615906127770534,
-      "grad_norm": 1.5099927186965942,
-      "learning_rate": 7.234472361809045e-05,
-      "loss": 5.1625,
-      "step": 28026
-    },
-    {
-      "epoch": 14.616427640156454,
-      "grad_norm": 1.4793306589126587,
-      "learning_rate": 7.234371859296483e-05,
-      "loss": 5.5822,
-      "step": 28027
-    },
-    {
-      "epoch": 14.616949152542373,
-      "grad_norm": 1.4475871324539185,
-      "learning_rate": 7.234271356783921e-05,
-      "loss": 5.626,
-      "step": 28028
-    },
-    {
-      "epoch": 14.617470664928293,
-      "grad_norm": 1.5126404762268066,
-      "learning_rate": 7.234170854271357e-05,
-      "loss": 5.6785,
-      "step": 28029
-    },
-    {
-      "epoch": 14.61799217731421,
-      "grad_norm": 1.502301812171936,
-      "learning_rate": 7.234070351758794e-05,
-      "loss": 5.3574,
-      "step": 28030
-    },
-    {
-      "epoch": 14.61851368970013,
-      "grad_norm": 1.5177226066589355,
-      "learning_rate": 7.233969849246232e-05,
-      "loss": 5.4452,
-      "step": 28031
-    },
-    {
-      "epoch": 14.61903520208605,
-      "grad_norm": 1.4015069007873535,
-      "learning_rate": 7.233869346733668e-05,
-      "loss": 5.3573,
-      "step": 28032
-    },
-    {
-      "epoch": 14.619556714471969,
-      "grad_norm": 1.5564148426055908,
-      "learning_rate": 7.233768844221106e-05,
-      "loss": 5.2268,
-      "step": 28033
-    },
-    {
-      "epoch": 14.620078226857888,
-      "grad_norm": 1.4391882419586182,
-      "learning_rate": 7.233668341708542e-05,
-      "loss": 5.4005,
-      "step": 28034
-    },
-    {
-      "epoch": 14.620599739243808,
-      "grad_norm": 1.5460859537124634,
-      "learning_rate": 7.23356783919598e-05,
-      "loss": 4.8861,
-      "step": 28035
-    },
-    {
-      "epoch": 14.621121251629726,
-      "grad_norm": 1.3467835187911987,
-      "learning_rate": 7.233467336683418e-05,
-      "loss": 5.8648,
-      "step": 28036
-    },
-    {
-      "epoch": 14.621642764015645,
-      "grad_norm": 1.5800251960754395,
-      "learning_rate": 7.233366834170856e-05,
-      "loss": 5.4541,
-      "step": 28037
-    },
-    {
-      "epoch": 14.622164276401564,
-      "grad_norm": 1.444820523262024,
-      "learning_rate": 7.233266331658292e-05,
-      "loss": 5.567,
-      "step": 28038
-    },
-    {
-      "epoch": 14.622685788787484,
-      "grad_norm": 1.620198130607605,
-      "learning_rate": 7.23316582914573e-05,
-      "loss": 5.0313,
-      "step": 28039
-    },
-    {
-      "epoch": 14.623207301173403,
-      "grad_norm": 1.534997820854187,
-      "learning_rate": 7.233065326633166e-05,
-      "loss": 5.3123,
-      "step": 28040
-    },
-    {
-      "epoch": 14.623728813559323,
-      "grad_norm": 1.4978152513504028,
-      "learning_rate": 7.232964824120604e-05,
-      "loss": 5.4966,
-      "step": 28041
-    },
-    {
-      "epoch": 14.62425032594524,
-      "grad_norm": 1.494978904724121,
-      "learning_rate": 7.23286432160804e-05,
-      "loss": 5.3385,
-      "step": 28042
-    },
-    {
-      "epoch": 14.62477183833116,
-      "grad_norm": 1.4366554021835327,
-      "learning_rate": 7.232763819095478e-05,
-      "loss": 5.8428,
-      "step": 28043
-    },
-    {
-      "epoch": 14.62529335071708,
-      "grad_norm": 1.4840329885482788,
-      "learning_rate": 7.232663316582915e-05,
-      "loss": 5.6972,
-      "step": 28044
-    },
-    {
-      "epoch": 14.625814863102999,
-      "grad_norm": 1.5645627975463867,
-      "learning_rate": 7.232562814070351e-05,
-      "loss": 5.2837,
-      "step": 28045
-    },
-    {
-      "epoch": 14.626336375488918,
-      "grad_norm": 1.5362759828567505,
-      "learning_rate": 7.232462311557789e-05,
-      "loss": 5.2555,
-      "step": 28046
-    },
-    {
-      "epoch": 14.626857887874838,
-      "grad_norm": 1.59690260887146,
-      "learning_rate": 7.232361809045227e-05,
-      "loss": 4.962,
-      "step": 28047
-    },
-    {
-      "epoch": 14.627379400260756,
-      "grad_norm": 1.4377580881118774,
-      "learning_rate": 7.232261306532664e-05,
-      "loss": 5.5033,
-      "step": 28048
-    },
-    {
-      "epoch": 14.627900912646675,
-      "grad_norm": 1.499528169631958,
-      "learning_rate": 7.232160804020101e-05,
-      "loss": 5.1934,
-      "step": 28049
-    },
-    {
-      "epoch": 14.628422425032594,
-      "grad_norm": 1.4355498552322388,
-      "learning_rate": 7.232060301507539e-05,
-      "loss": 5.1571,
-      "step": 28050
-    },
-    {
-      "epoch": 14.628943937418514,
-      "grad_norm": 1.447056531906128,
-      "learning_rate": 7.231959798994975e-05,
-      "loss": 4.8334,
-      "step": 28051
-    },
-    {
-      "epoch": 14.629465449804433,
-      "grad_norm": 1.552718997001648,
-      "learning_rate": 7.231859296482413e-05,
-      "loss": 5.2716,
-      "step": 28052
-    },
-    {
-      "epoch": 14.629986962190351,
-      "grad_norm": 1.4355862140655518,
-      "learning_rate": 7.231758793969849e-05,
-      "loss": 5.3548,
-      "step": 28053
-    },
-    {
-      "epoch": 14.63050847457627,
-      "grad_norm": 1.350435495376587,
-      "learning_rate": 7.231658291457287e-05,
-      "loss": 5.5829,
-      "step": 28054
-    },
-    {
-      "epoch": 14.63102998696219,
-      "grad_norm": 1.515970230102539,
-      "learning_rate": 7.231557788944723e-05,
-      "loss": 5.1625,
-      "step": 28055
-    },
-    {
-      "epoch": 14.63155149934811,
-      "grad_norm": 1.5217311382293701,
-      "learning_rate": 7.231457286432161e-05,
-      "loss": 5.2319,
-      "step": 28056
-    },
-    {
-      "epoch": 14.632073011734029,
-      "grad_norm": 1.5027939081192017,
-      "learning_rate": 7.231356783919599e-05,
-      "loss": 5.2898,
-      "step": 28057
-    },
-    {
-      "epoch": 14.632594524119948,
-      "grad_norm": 1.64187753200531,
-      "learning_rate": 7.231256281407035e-05,
-      "loss": 4.8136,
-      "step": 28058
-    },
-    {
-      "epoch": 14.633116036505868,
-      "grad_norm": 1.454702377319336,
-      "learning_rate": 7.231155778894473e-05,
-      "loss": 5.4222,
-      "step": 28059
-    },
-    {
-      "epoch": 14.633637548891786,
-      "grad_norm": 1.4725525379180908,
-      "learning_rate": 7.23105527638191e-05,
-      "loss": 5.4143,
-      "step": 28060
-    },
-    {
-      "epoch": 14.634159061277705,
-      "grad_norm": 1.418113350868225,
-      "learning_rate": 7.230954773869347e-05,
-      "loss": 5.6502,
-      "step": 28061
-    },
-    {
-      "epoch": 14.634680573663625,
-      "grad_norm": 1.3824782371520996,
-      "learning_rate": 7.230854271356784e-05,
-      "loss": 5.3221,
-      "step": 28062
-    },
-    {
-      "epoch": 14.635202086049544,
-      "grad_norm": 1.5377013683319092,
-      "learning_rate": 7.230753768844222e-05,
-      "loss": 5.0215,
-      "step": 28063
-    },
-    {
-      "epoch": 14.635723598435463,
-      "grad_norm": 1.4318524599075317,
-      "learning_rate": 7.230653266331658e-05,
-      "loss": 4.8036,
-      "step": 28064
-    },
-    {
-      "epoch": 14.636245110821381,
-      "grad_norm": 1.4695144891738892,
-      "learning_rate": 7.230552763819096e-05,
-      "loss": 5.4955,
-      "step": 28065
-    },
-    {
-      "epoch": 14.6367666232073,
-      "grad_norm": 1.4555621147155762,
-      "learning_rate": 7.230452261306532e-05,
-      "loss": 5.7144,
-      "step": 28066
-    },
-    {
-      "epoch": 14.63728813559322,
-      "grad_norm": 1.4066228866577148,
-      "learning_rate": 7.23035175879397e-05,
-      "loss": 5.5651,
-      "step": 28067
-    },
-    {
-      "epoch": 14.63780964797914,
-      "grad_norm": 1.4220725297927856,
-      "learning_rate": 7.230251256281408e-05,
-      "loss": 4.5415,
-      "step": 28068
-    },
-    {
-      "epoch": 14.638331160365059,
-      "grad_norm": 1.4254662990570068,
-      "learning_rate": 7.230150753768846e-05,
-      "loss": 5.804,
-      "step": 28069
-    },
-    {
-      "epoch": 14.638852672750978,
-      "grad_norm": 1.4819610118865967,
-      "learning_rate": 7.230050251256282e-05,
-      "loss": 5.5196,
-      "step": 28070
-    },
-    {
-      "epoch": 14.639374185136898,
-      "grad_norm": 1.616431713104248,
-      "learning_rate": 7.229949748743718e-05,
-      "loss": 4.9181,
-      "step": 28071
-    },
-    {
-      "epoch": 14.639895697522816,
-      "grad_norm": 1.4027271270751953,
-      "learning_rate": 7.229849246231156e-05,
-      "loss": 5.8865,
-      "step": 28072
-    },
-    {
-      "epoch": 14.640417209908735,
-      "grad_norm": 1.479526162147522,
-      "learning_rate": 7.229748743718593e-05,
-      "loss": 5.2014,
-      "step": 28073
-    },
-    {
-      "epoch": 14.640938722294655,
-      "grad_norm": 1.3837097883224487,
-      "learning_rate": 7.22964824120603e-05,
-      "loss": 5.6253,
-      "step": 28074
-    },
-    {
-      "epoch": 14.641460234680574,
-      "grad_norm": 1.5124224424362183,
-      "learning_rate": 7.229547738693467e-05,
-      "loss": 5.6861,
-      "step": 28075
-    },
-    {
-      "epoch": 14.641981747066493,
-      "grad_norm": 1.4464548826217651,
-      "learning_rate": 7.229447236180905e-05,
-      "loss": 5.6364,
-      "step": 28076
-    },
-    {
-      "epoch": 14.642503259452411,
-      "grad_norm": 1.4539713859558105,
-      "learning_rate": 7.229346733668342e-05,
-      "loss": 5.7719,
-      "step": 28077
-    },
-    {
-      "epoch": 14.64302477183833,
-      "grad_norm": 1.5073076486587524,
-      "learning_rate": 7.22924623115578e-05,
-      "loss": 5.2897,
-      "step": 28078
-    },
-    {
-      "epoch": 14.64354628422425,
-      "grad_norm": 1.4004369974136353,
-      "learning_rate": 7.229145728643217e-05,
-      "loss": 5.5817,
-      "step": 28079
-    },
-    {
-      "epoch": 14.64406779661017,
-      "grad_norm": 1.358492374420166,
-      "learning_rate": 7.229045226130654e-05,
-      "loss": 5.0299,
-      "step": 28080
-    },
-    {
-      "epoch": 14.644589308996089,
-      "grad_norm": 1.5196316242218018,
-      "learning_rate": 7.228944723618091e-05,
-      "loss": 5.3023,
-      "step": 28081
-    },
-    {
-      "epoch": 14.645110821382008,
-      "grad_norm": 1.5064361095428467,
-      "learning_rate": 7.228844221105529e-05,
-      "loss": 4.9911,
-      "step": 28082
-    },
-    {
-      "epoch": 14.645632333767926,
-      "grad_norm": 1.4914406538009644,
-      "learning_rate": 7.228743718592965e-05,
-      "loss": 5.3503,
-      "step": 28083
-    },
-    {
-      "epoch": 14.646153846153846,
-      "grad_norm": 1.4332497119903564,
-      "learning_rate": 7.228643216080401e-05,
-      "loss": 5.3728,
-      "step": 28084
-    },
-    {
-      "epoch": 14.646675358539765,
-      "grad_norm": 1.4163998365402222,
-      "learning_rate": 7.228542713567839e-05,
-      "loss": 5.482,
-      "step": 28085
-    },
-    {
-      "epoch": 14.647196870925685,
-      "grad_norm": 1.478324055671692,
-      "learning_rate": 7.228442211055276e-05,
-      "loss": 5.7025,
-      "step": 28086
-    },
-    {
-      "epoch": 14.647718383311604,
-      "grad_norm": 1.5797703266143799,
-      "learning_rate": 7.228341708542713e-05,
-      "loss": 5.0705,
-      "step": 28087
-    },
-    {
-      "epoch": 14.648239895697523,
-      "grad_norm": 1.598757028579712,
-      "learning_rate": 7.228241206030151e-05,
-      "loss": 4.949,
-      "step": 28088
-    },
-    {
-      "epoch": 14.648761408083441,
-      "grad_norm": 1.5320329666137695,
-      "learning_rate": 7.228140703517589e-05,
-      "loss": 5.3692,
-      "step": 28089
-    },
-    {
-      "epoch": 14.64928292046936,
-      "grad_norm": 1.4285638332366943,
-      "learning_rate": 7.228040201005025e-05,
-      "loss": 5.3843,
-      "step": 28090
-    },
-    {
-      "epoch": 14.64980443285528,
-      "grad_norm": 1.445030927658081,
-      "learning_rate": 7.227939698492463e-05,
-      "loss": 5.8227,
-      "step": 28091
-    },
-    {
-      "epoch": 14.6503259452412,
-      "grad_norm": 1.5767546892166138,
-      "learning_rate": 7.2278391959799e-05,
-      "loss": 4.8972,
-      "step": 28092
-    },
-    {
-      "epoch": 14.650847457627119,
-      "grad_norm": 1.4194116592407227,
-      "learning_rate": 7.227738693467337e-05,
-      "loss": 5.3037,
-      "step": 28093
-    },
-    {
-      "epoch": 14.651368970013039,
-      "grad_norm": 1.462957739830017,
-      "learning_rate": 7.227638190954774e-05,
-      "loss": 4.9469,
-      "step": 28094
-    },
-    {
-      "epoch": 14.651890482398956,
-      "grad_norm": 1.3990936279296875,
-      "learning_rate": 7.227537688442212e-05,
-      "loss": 5.4519,
-      "step": 28095
-    },
-    {
-      "epoch": 14.652411994784876,
-      "grad_norm": 2.222689390182495,
-      "learning_rate": 7.227437185929648e-05,
-      "loss": 4.5882,
-      "step": 28096
-    },
-    {
-      "epoch": 14.652933507170795,
-      "grad_norm": 1.4791616201400757,
-      "learning_rate": 7.227336683417086e-05,
-      "loss": 5.1946,
-      "step": 28097
-    },
-    {
-      "epoch": 14.653455019556715,
-      "grad_norm": 1.3871874809265137,
-      "learning_rate": 7.227236180904524e-05,
-      "loss": 5.4466,
-      "step": 28098
-    },
-    {
-      "epoch": 14.653976531942634,
-      "grad_norm": 1.3762584924697876,
-      "learning_rate": 7.22713567839196e-05,
-      "loss": 5.7819,
-      "step": 28099
-    },
-    {
-      "epoch": 14.654498044328554,
-      "grad_norm": 1.536110758781433,
-      "learning_rate": 7.227035175879398e-05,
-      "loss": 5.478,
-      "step": 28100
-    },
-    {
-      "epoch": 14.655019556714471,
-      "grad_norm": 1.4418363571166992,
-      "learning_rate": 7.226934673366834e-05,
-      "loss": 5.3224,
-      "step": 28101
-    },
-    {
-      "epoch": 14.65554106910039,
-      "grad_norm": 1.403907299041748,
-      "learning_rate": 7.226834170854272e-05,
-      "loss": 5.3702,
-      "step": 28102
-    },
-    {
-      "epoch": 14.65606258148631,
-      "grad_norm": 1.466815710067749,
-      "learning_rate": 7.226733668341709e-05,
-      "loss": 5.5951,
-      "step": 28103
-    },
-    {
-      "epoch": 14.65658409387223,
-      "grad_norm": 1.5876373052597046,
-      "learning_rate": 7.226633165829146e-05,
-      "loss": 5.2101,
-      "step": 28104
-    },
-    {
-      "epoch": 14.657105606258149,
-      "grad_norm": 1.6359694004058838,
-      "learning_rate": 7.226532663316583e-05,
-      "loss": 5.3103,
-      "step": 28105
-    },
-    {
-      "epoch": 14.657627118644069,
-      "grad_norm": 1.4625928401947021,
-      "learning_rate": 7.22643216080402e-05,
-      "loss": 5.0327,
-      "step": 28106
-    },
-    {
-      "epoch": 14.658148631029986,
-      "grad_norm": 1.4175279140472412,
-      "learning_rate": 7.226331658291457e-05,
-      "loss": 5.9718,
-      "step": 28107
-    },
-    {
-      "epoch": 14.658670143415906,
-      "grad_norm": 1.4957834482192993,
-      "learning_rate": 7.226231155778895e-05,
-      "loss": 5.6512,
-      "step": 28108
-    },
-    {
-      "epoch": 14.659191655801825,
-      "grad_norm": 1.4261925220489502,
-      "learning_rate": 7.226130653266333e-05,
-      "loss": 5.571,
-      "step": 28109
-    },
-    {
-      "epoch": 14.659713168187745,
-      "grad_norm": 1.4031920433044434,
-      "learning_rate": 7.226030150753769e-05,
-      "loss": 5.4135,
-      "step": 28110
-    },
-    {
-      "epoch": 14.660234680573664,
-      "grad_norm": 1.3680752515792847,
-      "learning_rate": 7.225929648241207e-05,
-      "loss": 5.7767,
-      "step": 28111
-    },
-    {
-      "epoch": 14.660756192959584,
-      "grad_norm": 1.5410380363464355,
-      "learning_rate": 7.225829145728643e-05,
-      "loss": 5.5554,
-      "step": 28112
-    },
-    {
-      "epoch": 14.661277705345501,
-      "grad_norm": 1.4223462343215942,
-      "learning_rate": 7.225728643216081e-05,
-      "loss": 5.5536,
-      "step": 28113
-    },
-    {
-      "epoch": 14.66179921773142,
-      "grad_norm": 1.541030764579773,
-      "learning_rate": 7.225628140703517e-05,
-      "loss": 5.5858,
-      "step": 28114
-    },
-    {
-      "epoch": 14.66232073011734,
-      "grad_norm": 1.4607298374176025,
-      "learning_rate": 7.225527638190955e-05,
-      "loss": 5.551,
-      "step": 28115
-    },
-    {
-      "epoch": 14.66284224250326,
-      "grad_norm": 1.3986035585403442,
-      "learning_rate": 7.225427135678392e-05,
-      "loss": 5.556,
-      "step": 28116
-    },
-    {
-      "epoch": 14.663363754889179,
-      "grad_norm": 1.5912995338439941,
-      "learning_rate": 7.22532663316583e-05,
-      "loss": 5.2795,
-      "step": 28117
-    },
-    {
-      "epoch": 14.663885267275099,
-      "grad_norm": 1.4686657190322876,
-      "learning_rate": 7.225226130653267e-05,
-      "loss": 5.3268,
-      "step": 28118
-    },
-    {
-      "epoch": 14.664406779661016,
-      "grad_norm": 1.4905542135238647,
-      "learning_rate": 7.225125628140705e-05,
-      "loss": 4.9626,
-      "step": 28119
-    },
-    {
-      "epoch": 14.664928292046936,
-      "grad_norm": 1.356906771659851,
-      "learning_rate": 7.225025125628141e-05,
-      "loss": 5.6143,
-      "step": 28120
-    },
-    {
-      "epoch": 14.665449804432855,
-      "grad_norm": 1.4076216220855713,
-      "learning_rate": 7.224924623115579e-05,
-      "loss": 5.476,
-      "step": 28121
-    },
-    {
-      "epoch": 14.665971316818775,
-      "grad_norm": 1.3987082242965698,
-      "learning_rate": 7.224824120603016e-05,
-      "loss": 5.3922,
-      "step": 28122
-    },
-    {
-      "epoch": 14.666492829204694,
-      "grad_norm": 1.4661831855773926,
-      "learning_rate": 7.224723618090452e-05,
-      "loss": 5.7243,
-      "step": 28123
-    },
-    {
-      "epoch": 14.667014341590614,
-      "grad_norm": 1.460618257522583,
-      "learning_rate": 7.22462311557789e-05,
-      "loss": 5.5515,
-      "step": 28124
-    },
-    {
-      "epoch": 14.667535853976531,
-      "grad_norm": 1.3748971223831177,
-      "learning_rate": 7.224522613065326e-05,
-      "loss": 5.5255,
-      "step": 28125
-    },
-    {
-      "epoch": 14.66805736636245,
-      "grad_norm": 1.4378528594970703,
-      "learning_rate": 7.224422110552764e-05,
-      "loss": 5.2647,
-      "step": 28126
-    },
-    {
-      "epoch": 14.66857887874837,
-      "grad_norm": 1.386738896369934,
-      "learning_rate": 7.2243216080402e-05,
-      "loss": 4.9842,
-      "step": 28127
-    },
-    {
-      "epoch": 14.66910039113429,
-      "grad_norm": 1.4819945096969604,
-      "learning_rate": 7.224221105527638e-05,
-      "loss": 5.1302,
-      "step": 28128
-    },
-    {
-      "epoch": 14.66962190352021,
-      "grad_norm": 1.4874398708343506,
-      "learning_rate": 7.224120603015076e-05,
-      "loss": 5.8266,
-      "step": 28129
-    },
-    {
-      "epoch": 14.670143415906129,
-      "grad_norm": 1.4627703428268433,
-      "learning_rate": 7.224020100502514e-05,
-      "loss": 5.4983,
-      "step": 28130
-    },
-    {
-      "epoch": 14.670664928292046,
-      "grad_norm": 1.4871090650558472,
-      "learning_rate": 7.22391959798995e-05,
-      "loss": 5.3223,
-      "step": 28131
-    },
-    {
-      "epoch": 14.671186440677966,
-      "grad_norm": 1.4351727962493896,
-      "learning_rate": 7.223819095477388e-05,
-      "loss": 5.6504,
-      "step": 28132
-    },
-    {
-      "epoch": 14.671707953063885,
-      "grad_norm": 1.5209295749664307,
-      "learning_rate": 7.223718592964824e-05,
-      "loss": 5.1513,
-      "step": 28133
-    },
-    {
-      "epoch": 14.672229465449805,
-      "grad_norm": 1.4914367198944092,
-      "learning_rate": 7.223618090452262e-05,
-      "loss": 5.6763,
-      "step": 28134
-    },
-    {
-      "epoch": 14.672750977835724,
-      "grad_norm": 1.693715214729309,
-      "learning_rate": 7.223517587939699e-05,
-      "loss": 5.1142,
-      "step": 28135
-    },
-    {
-      "epoch": 14.673272490221644,
-      "grad_norm": 1.5869464874267578,
-      "learning_rate": 7.223417085427136e-05,
-      "loss": 5.0296,
-      "step": 28136
-    },
-    {
-      "epoch": 14.673794002607561,
-      "grad_norm": 1.565017580986023,
-      "learning_rate": 7.223316582914573e-05,
-      "loss": 5.4273,
-      "step": 28137
-    },
-    {
-      "epoch": 14.67431551499348,
-      "grad_norm": 1.5046159029006958,
-      "learning_rate": 7.22321608040201e-05,
-      "loss": 5.6526,
-      "step": 28138
-    },
-    {
-      "epoch": 14.6748370273794,
-      "grad_norm": 1.362112283706665,
-      "learning_rate": 7.223115577889448e-05,
-      "loss": 5.6844,
-      "step": 28139
-    },
-    {
-      "epoch": 14.67535853976532,
-      "grad_norm": 1.423667311668396,
-      "learning_rate": 7.223015075376885e-05,
-      "loss": 5.741,
-      "step": 28140
-    },
-    {
-      "epoch": 14.67588005215124,
-      "grad_norm": 1.4391424655914307,
-      "learning_rate": 7.222914572864323e-05,
-      "loss": 5.0209,
-      "step": 28141
-    },
-    {
-      "epoch": 14.676401564537159,
-      "grad_norm": 1.4779729843139648,
-      "learning_rate": 7.222814070351759e-05,
-      "loss": 5.0977,
-      "step": 28142
-    },
-    {
-      "epoch": 14.676923076923076,
-      "grad_norm": 1.4692564010620117,
-      "learning_rate": 7.222713567839197e-05,
-      "loss": 5.241,
-      "step": 28143
-    },
-    {
-      "epoch": 14.677444589308996,
-      "grad_norm": 1.5144970417022705,
-      "learning_rate": 7.222613065326633e-05,
-      "loss": 5.0863,
-      "step": 28144
-    },
-    {
-      "epoch": 14.677966101694915,
-      "grad_norm": 1.4394055604934692,
-      "learning_rate": 7.222512562814071e-05,
-      "loss": 5.5017,
-      "step": 28145
-    },
-    {
-      "epoch": 14.678487614080835,
-      "grad_norm": 1.5744074583053589,
-      "learning_rate": 7.222412060301507e-05,
-      "loss": 5.2759,
-      "step": 28146
-    },
-    {
-      "epoch": 14.679009126466754,
-      "grad_norm": 1.4384647607803345,
-      "learning_rate": 7.222311557788945e-05,
-      "loss": 5.3746,
-      "step": 28147
-    },
-    {
-      "epoch": 14.679530638852672,
-      "grad_norm": 1.484196662902832,
-      "learning_rate": 7.222211055276382e-05,
-      "loss": 5.212,
-      "step": 28148
-    },
-    {
-      "epoch": 14.680052151238591,
-      "grad_norm": 1.4263708591461182,
-      "learning_rate": 7.22211055276382e-05,
-      "loss": 4.6512,
-      "step": 28149
-    },
-    {
-      "epoch": 14.68057366362451,
-      "grad_norm": 1.4028003215789795,
-      "learning_rate": 7.222010050251257e-05,
-      "loss": 5.3045,
-      "step": 28150
-    },
-    {
-      "epoch": 14.68109517601043,
-      "grad_norm": 1.527626633644104,
-      "learning_rate": 7.221909547738694e-05,
-      "loss": 5.5244,
-      "step": 28151
-    },
-    {
-      "epoch": 14.68161668839635,
-      "grad_norm": 1.3645013570785522,
-      "learning_rate": 7.221809045226131e-05,
-      "loss": 5.7756,
-      "step": 28152
-    },
-    {
-      "epoch": 14.68213820078227,
-      "grad_norm": 1.5036036968231201,
-      "learning_rate": 7.221708542713568e-05,
-      "loss": 5.3177,
-      "step": 28153
-    },
-    {
-      "epoch": 14.682659713168189,
-      "grad_norm": 1.5129085779190063,
-      "learning_rate": 7.221608040201006e-05,
-      "loss": 5.2543,
-      "step": 28154
-    },
-    {
-      "epoch": 14.683181225554106,
-      "grad_norm": 1.6177122592926025,
-      "learning_rate": 7.221507537688442e-05,
-      "loss": 5.2857,
-      "step": 28155
-    },
-    {
-      "epoch": 14.683702737940026,
-      "grad_norm": 1.4480351209640503,
-      "learning_rate": 7.22140703517588e-05,
-      "loss": 5.1973,
-      "step": 28156
-    },
-    {
-      "epoch": 14.684224250325945,
-      "grad_norm": 2.2204015254974365,
-      "learning_rate": 7.221306532663316e-05,
-      "loss": 4.868,
-      "step": 28157
-    },
-    {
-      "epoch": 14.684745762711865,
-      "grad_norm": 1.3705087900161743,
-      "learning_rate": 7.221206030150754e-05,
-      "loss": 5.7051,
-      "step": 28158
-    },
-    {
-      "epoch": 14.685267275097784,
-      "grad_norm": 1.4836175441741943,
-      "learning_rate": 7.22110552763819e-05,
-      "loss": 5.3287,
-      "step": 28159
-    },
-    {
-      "epoch": 14.685788787483702,
-      "grad_norm": 1.4403913021087646,
-      "learning_rate": 7.221005025125628e-05,
-      "loss": 5.4785,
-      "step": 28160
-    },
-    {
-      "epoch": 14.686310299869621,
-      "grad_norm": 1.5646923780441284,
-      "learning_rate": 7.220904522613066e-05,
-      "loss": 5.4514,
-      "step": 28161
-    },
-    {
-      "epoch": 14.68683181225554,
-      "grad_norm": 1.4896273612976074,
-      "learning_rate": 7.220804020100504e-05,
-      "loss": 5.3447,
-      "step": 28162
-    },
-    {
-      "epoch": 14.68735332464146,
-      "grad_norm": 1.5137661695480347,
-      "learning_rate": 7.22070351758794e-05,
-      "loss": 5.4338,
-      "step": 28163
-    },
-    {
-      "epoch": 14.68787483702738,
-      "grad_norm": 1.457275390625,
-      "learning_rate": 7.220603015075377e-05,
-      "loss": 4.974,
-      "step": 28164
-    },
-    {
-      "epoch": 14.6883963494133,
-      "grad_norm": 1.4073340892791748,
-      "learning_rate": 7.220502512562814e-05,
-      "loss": 5.8568,
-      "step": 28165
-    },
-    {
-      "epoch": 14.688917861799219,
-      "grad_norm": 1.4968607425689697,
-      "learning_rate": 7.220402010050251e-05,
-      "loss": 5.499,
-      "step": 28166
-    },
-    {
-      "epoch": 14.689439374185136,
-      "grad_norm": 1.4751383066177368,
-      "learning_rate": 7.220301507537689e-05,
-      "loss": 5.3574,
-      "step": 28167
-    },
-    {
-      "epoch": 14.689960886571056,
-      "grad_norm": 1.4026161432266235,
-      "learning_rate": 7.220201005025125e-05,
-      "loss": 5.4814,
-      "step": 28168
-    },
-    {
-      "epoch": 14.690482398956975,
-      "grad_norm": 1.489212989807129,
-      "learning_rate": 7.220100502512563e-05,
-      "loss": 5.085,
-      "step": 28169
-    },
-    {
-      "epoch": 14.691003911342895,
-      "grad_norm": 1.4058120250701904,
-      "learning_rate": 7.22e-05,
-      "loss": 5.0366,
-      "step": 28170
-    },
-    {
-      "epoch": 14.691525423728814,
-      "grad_norm": 1.4367023706436157,
-      "learning_rate": 7.219899497487438e-05,
-      "loss": 5.5788,
-      "step": 28171
-    },
-    {
-      "epoch": 14.692046936114732,
-      "grad_norm": 1.460575819015503,
-      "learning_rate": 7.219798994974875e-05,
-      "loss": 5.5572,
-      "step": 28172
-    },
-    {
-      "epoch": 14.692568448500651,
-      "grad_norm": 1.54339599609375,
-      "learning_rate": 7.219698492462313e-05,
-      "loss": 5.2483,
-      "step": 28173
-    },
-    {
-      "epoch": 14.69308996088657,
-      "grad_norm": 1.4549883604049683,
-      "learning_rate": 7.219597989949749e-05,
-      "loss": 5.6114,
-      "step": 28174
-    },
-    {
-      "epoch": 14.69361147327249,
-      "grad_norm": 1.3347749710083008,
-      "learning_rate": 7.219497487437187e-05,
-      "loss": 5.0343,
-      "step": 28175
-    },
-    {
-      "epoch": 14.69413298565841,
-      "grad_norm": 1.441857933998108,
-      "learning_rate": 7.219396984924623e-05,
-      "loss": 4.9028,
-      "step": 28176
-    },
-    {
-      "epoch": 14.69465449804433,
-      "grad_norm": 1.4734344482421875,
-      "learning_rate": 7.21929648241206e-05,
-      "loss": 5.2901,
-      "step": 28177
-    },
-    {
-      "epoch": 14.695176010430247,
-      "grad_norm": 1.758072018623352,
-      "learning_rate": 7.219195979899498e-05,
-      "loss": 4.7856,
-      "step": 28178
-    },
-    {
-      "epoch": 14.695697522816166,
-      "grad_norm": 1.6053146123886108,
-      "learning_rate": 7.219095477386934e-05,
-      "loss": 5.0003,
-      "step": 28179
-    },
-    {
-      "epoch": 14.696219035202086,
-      "grad_norm": 1.6110655069351196,
-      "learning_rate": 7.218994974874372e-05,
-      "loss": 5.3736,
-      "step": 28180
-    },
-    {
-      "epoch": 14.696740547588005,
-      "grad_norm": 1.5266666412353516,
-      "learning_rate": 7.21889447236181e-05,
-      "loss": 5.7229,
-      "step": 28181
-    },
-    {
-      "epoch": 14.697262059973925,
-      "grad_norm": 1.4786460399627686,
-      "learning_rate": 7.218793969849247e-05,
-      "loss": 5.4951,
-      "step": 28182
-    },
-    {
-      "epoch": 14.697783572359844,
-      "grad_norm": 1.4230722188949585,
-      "learning_rate": 7.218693467336684e-05,
-      "loss": 5.4579,
-      "step": 28183
-    },
-    {
-      "epoch": 14.698305084745762,
-      "grad_norm": 1.535028100013733,
-      "learning_rate": 7.218592964824121e-05,
-      "loss": 5.0438,
-      "step": 28184
-    },
-    {
-      "epoch": 14.698826597131681,
-      "grad_norm": 1.5759721994400024,
-      "learning_rate": 7.218492462311558e-05,
-      "loss": 5.4524,
-      "step": 28185
-    },
-    {
-      "epoch": 14.6993481095176,
-      "grad_norm": 1.5206711292266846,
-      "learning_rate": 7.218391959798996e-05,
-      "loss": 5.4994,
-      "step": 28186
-    },
-    {
-      "epoch": 14.69986962190352,
-      "grad_norm": 1.4305146932601929,
-      "learning_rate": 7.218291457286432e-05,
-      "loss": 5.454,
-      "step": 28187
-    },
-    {
-      "epoch": 14.70039113428944,
-      "grad_norm": 1.6186145544052124,
-      "learning_rate": 7.21819095477387e-05,
-      "loss": 4.8757,
-      "step": 28188
-    },
-    {
-      "epoch": 14.70091264667536,
-      "grad_norm": 1.5152559280395508,
-      "learning_rate": 7.218090452261306e-05,
-      "loss": 5.7097,
-      "step": 28189
-    },
-    {
-      "epoch": 14.701434159061277,
-      "grad_norm": 1.4308090209960938,
-      "learning_rate": 7.217989949748744e-05,
-      "loss": 5.3886,
-      "step": 28190
-    },
-    {
-      "epoch": 14.701955671447196,
-      "grad_norm": 1.4486464262008667,
-      "learning_rate": 7.217889447236182e-05,
-      "loss": 5.2343,
-      "step": 28191
-    },
-    {
-      "epoch": 14.702477183833116,
-      "grad_norm": 1.4027429819107056,
-      "learning_rate": 7.217788944723618e-05,
-      "loss": 5.1541,
-      "step": 28192
-    },
-    {
-      "epoch": 14.702998696219035,
-      "grad_norm": 1.4367965459823608,
-      "learning_rate": 7.217688442211056e-05,
-      "loss": 5.6494,
-      "step": 28193
-    },
-    {
-      "epoch": 14.703520208604955,
-      "grad_norm": 1.443812608718872,
-      "learning_rate": 7.217587939698493e-05,
-      "loss": 5.1523,
-      "step": 28194
-    },
-    {
-      "epoch": 14.704041720990874,
-      "grad_norm": 1.4655039310455322,
-      "learning_rate": 7.21748743718593e-05,
-      "loss": 5.2704,
-      "step": 28195
-    },
-    {
-      "epoch": 14.704563233376792,
-      "grad_norm": 1.5851361751556396,
-      "learning_rate": 7.217386934673367e-05,
-      "loss": 5.2637,
-      "step": 28196
-    },
-    {
-      "epoch": 14.705084745762711,
-      "grad_norm": 1.4109044075012207,
-      "learning_rate": 7.217286432160805e-05,
-      "loss": 5.5971,
-      "step": 28197
-    },
-    {
-      "epoch": 14.70560625814863,
-      "grad_norm": 1.4803260564804077,
-      "learning_rate": 7.217185929648241e-05,
-      "loss": 5.2132,
-      "step": 28198
-    },
-    {
-      "epoch": 14.70612777053455,
-      "grad_norm": 1.421075701713562,
-      "learning_rate": 7.217085427135679e-05,
-      "loss": 5.6154,
-      "step": 28199
-    },
-    {
-      "epoch": 14.70664928292047,
-      "grad_norm": 1.4420520067214966,
-      "learning_rate": 7.216984924623115e-05,
-      "loss": 5.2614,
-      "step": 28200
-    },
-    {
-      "epoch": 14.70717079530639,
-      "grad_norm": 1.4863632917404175,
-      "learning_rate": 7.216884422110553e-05,
-      "loss": 5.6562,
-      "step": 28201
-    },
-    {
-      "epoch": 14.707692307692307,
-      "grad_norm": 1.4684005975723267,
-      "learning_rate": 7.216783919597991e-05,
-      "loss": 4.9777,
-      "step": 28202
-    },
-    {
-      "epoch": 14.708213820078226,
-      "grad_norm": 1.5669481754302979,
-      "learning_rate": 7.216683417085427e-05,
-      "loss": 5.2824,
-      "step": 28203
-    },
-    {
-      "epoch": 14.708735332464146,
-      "grad_norm": 1.3931195735931396,
-      "learning_rate": 7.216582914572865e-05,
-      "loss": 5.7996,
-      "step": 28204
-    },
-    {
-      "epoch": 14.709256844850065,
-      "grad_norm": 1.4823716878890991,
-      "learning_rate": 7.216482412060301e-05,
-      "loss": 5.2216,
-      "step": 28205
-    },
-    {
-      "epoch": 14.709778357235985,
-      "grad_norm": 1.4135807752609253,
-      "learning_rate": 7.216381909547739e-05,
-      "loss": 5.3725,
-      "step": 28206
-    },
-    {
-      "epoch": 14.710299869621904,
-      "grad_norm": 1.5749331712722778,
-      "learning_rate": 7.216281407035176e-05,
-      "loss": 4.6094,
-      "step": 28207
-    },
-    {
-      "epoch": 14.710821382007822,
-      "grad_norm": 1.4896013736724854,
-      "learning_rate": 7.216180904522613e-05,
-      "loss": 5.2563,
-      "step": 28208
-    },
-    {
-      "epoch": 14.711342894393741,
-      "grad_norm": 1.4666322469711304,
-      "learning_rate": 7.21608040201005e-05,
-      "loss": 5.2081,
-      "step": 28209
-    },
-    {
-      "epoch": 14.711864406779661,
-      "grad_norm": 1.4001638889312744,
-      "learning_rate": 7.215979899497488e-05,
-      "loss": 5.7995,
-      "step": 28210
-    },
-    {
-      "epoch": 14.71238591916558,
-      "grad_norm": 1.4972490072250366,
-      "learning_rate": 7.215879396984925e-05,
-      "loss": 5.2838,
-      "step": 28211
-    },
-    {
-      "epoch": 14.7129074315515,
-      "grad_norm": 1.499792218208313,
-      "learning_rate": 7.215778894472363e-05,
-      "loss": 5.3936,
-      "step": 28212
-    },
-    {
-      "epoch": 14.71342894393742,
-      "grad_norm": 1.3145692348480225,
-      "learning_rate": 7.2156783919598e-05,
-      "loss": 5.9015,
-      "step": 28213
-    },
-    {
-      "epoch": 14.713950456323337,
-      "grad_norm": 1.448065996170044,
-      "learning_rate": 7.215577889447237e-05,
-      "loss": 5.2874,
-      "step": 28214
-    },
-    {
-      "epoch": 14.714471968709256,
-      "grad_norm": 1.4744722843170166,
-      "learning_rate": 7.215477386934674e-05,
-      "loss": 5.6032,
-      "step": 28215
-    },
-    {
-      "epoch": 14.714993481095176,
-      "grad_norm": 1.522020697593689,
-      "learning_rate": 7.21537688442211e-05,
-      "loss": 4.9345,
-      "step": 28216
-    },
-    {
-      "epoch": 14.715514993481095,
-      "grad_norm": 1.4535695314407349,
-      "learning_rate": 7.215276381909548e-05,
-      "loss": 5.499,
-      "step": 28217
-    },
-    {
-      "epoch": 14.716036505867015,
-      "grad_norm": 1.5238364934921265,
-      "learning_rate": 7.215175879396984e-05,
-      "loss": 5.3798,
-      "step": 28218
-    },
-    {
-      "epoch": 14.716558018252934,
-      "grad_norm": 1.5729619264602661,
-      "learning_rate": 7.215075376884422e-05,
-      "loss": 5.2841,
-      "step": 28219
-    },
-    {
-      "epoch": 14.717079530638852,
-      "grad_norm": 1.3386579751968384,
-      "learning_rate": 7.214974874371859e-05,
-      "loss": 5.4626,
-      "step": 28220
-    },
-    {
-      "epoch": 14.717601043024771,
-      "grad_norm": 1.4146227836608887,
-      "learning_rate": 7.214874371859296e-05,
-      "loss": 5.603,
-      "step": 28221
-    },
-    {
-      "epoch": 14.718122555410691,
-      "grad_norm": 1.5771195888519287,
-      "learning_rate": 7.214773869346734e-05,
-      "loss": 5.0536,
-      "step": 28222
-    },
-    {
-      "epoch": 14.71864406779661,
-      "grad_norm": 1.503506064414978,
-      "learning_rate": 7.214673366834172e-05,
-      "loss": 5.5354,
-      "step": 28223
-    },
-    {
-      "epoch": 14.71916558018253,
-      "grad_norm": 1.456629753112793,
-      "learning_rate": 7.214572864321608e-05,
-      "loss": 4.7953,
-      "step": 28224
-    },
-    {
-      "epoch": 14.71968709256845,
-      "grad_norm": 1.4596984386444092,
-      "learning_rate": 7.214472361809046e-05,
-      "loss": 5.5949,
-      "step": 28225
-    },
-    {
-      "epoch": 14.720208604954367,
-      "grad_norm": 1.3697729110717773,
-      "learning_rate": 7.214371859296483e-05,
-      "loss": 5.8502,
-      "step": 28226
-    },
-    {
-      "epoch": 14.720730117340286,
-      "grad_norm": 1.5031572580337524,
-      "learning_rate": 7.21427135678392e-05,
-      "loss": 5.2277,
-      "step": 28227
-    },
-    {
-      "epoch": 14.721251629726206,
-      "grad_norm": 1.4293179512023926,
-      "learning_rate": 7.214170854271357e-05,
-      "loss": 5.619,
-      "step": 28228
-    },
-    {
-      "epoch": 14.721773142112125,
-      "grad_norm": 1.4785178899765015,
-      "learning_rate": 7.214070351758795e-05,
-      "loss": 5.1229,
-      "step": 28229
-    },
-    {
-      "epoch": 14.722294654498045,
-      "grad_norm": 1.5023335218429565,
-      "learning_rate": 7.213969849246231e-05,
-      "loss": 5.1625,
-      "step": 28230
-    },
-    {
-      "epoch": 14.722816166883963,
-      "grad_norm": 1.478812575340271,
-      "learning_rate": 7.213869346733669e-05,
-      "loss": 5.0986,
-      "step": 28231
-    },
-    {
-      "epoch": 14.723337679269882,
-      "grad_norm": 1.5936671495437622,
-      "learning_rate": 7.213768844221107e-05,
-      "loss": 5.1756,
-      "step": 28232
-    },
-    {
-      "epoch": 14.723859191655801,
-      "grad_norm": 1.5258030891418457,
-      "learning_rate": 7.213668341708543e-05,
-      "loss": 5.2876,
-      "step": 28233
-    },
-    {
-      "epoch": 14.724380704041721,
-      "grad_norm": 1.458328366279602,
-      "learning_rate": 7.213567839195981e-05,
-      "loss": 5.3506,
-      "step": 28234
-    },
-    {
-      "epoch": 14.72490221642764,
-      "grad_norm": 1.3459807634353638,
-      "learning_rate": 7.213467336683417e-05,
-      "loss": 4.8225,
-      "step": 28235
-    },
-    {
-      "epoch": 14.72542372881356,
-      "grad_norm": 1.4427366256713867,
-      "learning_rate": 7.213366834170855e-05,
-      "loss": 5.6207,
-      "step": 28236
-    },
-    {
-      "epoch": 14.72594524119948,
-      "grad_norm": 1.3654327392578125,
-      "learning_rate": 7.213266331658291e-05,
-      "loss": 5.6421,
-      "step": 28237
-    },
-    {
-      "epoch": 14.726466753585397,
-      "grad_norm": 1.6562551259994507,
-      "learning_rate": 7.213165829145729e-05,
-      "loss": 4.7617,
-      "step": 28238
-    },
-    {
-      "epoch": 14.726988265971316,
-      "grad_norm": 1.4640525579452515,
-      "learning_rate": 7.213065326633166e-05,
-      "loss": 5.3036,
-      "step": 28239
-    },
-    {
-      "epoch": 14.727509778357236,
-      "grad_norm": 1.4097212553024292,
-      "learning_rate": 7.212964824120603e-05,
-      "loss": 5.4425,
-      "step": 28240
-    },
-    {
-      "epoch": 14.728031290743155,
-      "grad_norm": 1.445619821548462,
-      "learning_rate": 7.21286432160804e-05,
-      "loss": 5.4059,
-      "step": 28241
-    },
-    {
-      "epoch": 14.728552803129075,
-      "grad_norm": 1.5069063901901245,
-      "learning_rate": 7.212763819095478e-05,
-      "loss": 5.1564,
-      "step": 28242
-    },
-    {
-      "epoch": 14.729074315514993,
-      "grad_norm": 1.6884853839874268,
-      "learning_rate": 7.212663316582915e-05,
-      "loss": 5.6337,
-      "step": 28243
-    },
-    {
-      "epoch": 14.729595827900912,
-      "grad_norm": 1.524701476097107,
-      "learning_rate": 7.212562814070352e-05,
-      "loss": 5.7377,
-      "step": 28244
-    },
-    {
-      "epoch": 14.730117340286832,
-      "grad_norm": 1.4497270584106445,
-      "learning_rate": 7.21246231155779e-05,
-      "loss": 5.6796,
-      "step": 28245
-    },
-    {
-      "epoch": 14.730638852672751,
-      "grad_norm": 1.439222812652588,
-      "learning_rate": 7.212361809045226e-05,
-      "loss": 5.69,
-      "step": 28246
-    },
-    {
-      "epoch": 14.73116036505867,
-      "grad_norm": 1.3670276403427124,
-      "learning_rate": 7.212261306532664e-05,
-      "loss": 5.8648,
-      "step": 28247
-    },
-    {
-      "epoch": 14.73168187744459,
-      "grad_norm": 1.4953721761703491,
-      "learning_rate": 7.2121608040201e-05,
-      "loss": 5.5673,
-      "step": 28248
-    },
-    {
-      "epoch": 14.73220338983051,
-      "grad_norm": 1.47298002243042,
-      "learning_rate": 7.212060301507538e-05,
-      "loss": 5.3708,
-      "step": 28249
-    },
-    {
-      "epoch": 14.732724902216427,
-      "grad_norm": 1.3440204858779907,
-      "learning_rate": 7.211959798994975e-05,
-      "loss": 5.6782,
-      "step": 28250
-    },
-    {
-      "epoch": 14.733246414602347,
-      "grad_norm": 1.3914991617202759,
-      "learning_rate": 7.211859296482412e-05,
-      "loss": 5.2406,
-      "step": 28251
-    },
-    {
-      "epoch": 14.733767926988266,
-      "grad_norm": 1.468814730644226,
-      "learning_rate": 7.21175879396985e-05,
-      "loss": 5.4138,
-      "step": 28252
-    },
-    {
-      "epoch": 14.734289439374185,
-      "grad_norm": 1.4674304723739624,
-      "learning_rate": 7.211658291457288e-05,
-      "loss": 5.685,
-      "step": 28253
-    },
-    {
-      "epoch": 14.734810951760105,
-      "grad_norm": 1.396816372871399,
-      "learning_rate": 7.211557788944724e-05,
-      "loss": 5.0301,
-      "step": 28254
-    },
-    {
-      "epoch": 14.735332464146023,
-      "grad_norm": 1.5079962015151978,
-      "learning_rate": 7.211457286432162e-05,
-      "loss": 5.2401,
-      "step": 28255
-    },
-    {
-      "epoch": 14.735853976531942,
-      "grad_norm": 1.468716025352478,
-      "learning_rate": 7.211356783919598e-05,
-      "loss": 5.484,
-      "step": 28256
-    },
-    {
-      "epoch": 14.736375488917862,
-      "grad_norm": 1.408874750137329,
-      "learning_rate": 7.211256281407035e-05,
-      "loss": 5.3789,
-      "step": 28257
-    },
-    {
-      "epoch": 14.736897001303781,
-      "grad_norm": 1.3703978061676025,
-      "learning_rate": 7.211155778894473e-05,
-      "loss": 5.6283,
-      "step": 28258
-    },
-    {
-      "epoch": 14.7374185136897,
-      "grad_norm": 1.3983272314071655,
-      "learning_rate": 7.211055276381909e-05,
-      "loss": 5.5287,
-      "step": 28259
-    },
-    {
-      "epoch": 14.73794002607562,
-      "grad_norm": 1.3526605367660522,
-      "learning_rate": 7.210954773869347e-05,
-      "loss": 5.5813,
-      "step": 28260
-    },
-    {
-      "epoch": 14.73846153846154,
-      "grad_norm": 1.5043106079101562,
-      "learning_rate": 7.210854271356783e-05,
-      "loss": 4.8432,
-      "step": 28261
-    },
-    {
-      "epoch": 14.738983050847457,
-      "grad_norm": 1.4631733894348145,
-      "learning_rate": 7.210753768844221e-05,
-      "loss": 5.2374,
-      "step": 28262
-    },
-    {
-      "epoch": 14.739504563233377,
-      "grad_norm": 1.6288541555404663,
-      "learning_rate": 7.210653266331659e-05,
-      "loss": 5.0913,
-      "step": 28263
-    },
-    {
-      "epoch": 14.740026075619296,
-      "grad_norm": 1.492732048034668,
-      "learning_rate": 7.210552763819097e-05,
-      "loss": 4.8816,
-      "step": 28264
-    },
-    {
-      "epoch": 14.740547588005215,
-      "grad_norm": 1.4587335586547852,
-      "learning_rate": 7.210452261306533e-05,
-      "loss": 5.4865,
-      "step": 28265
-    },
-    {
-      "epoch": 14.741069100391135,
-      "grad_norm": 1.4465841054916382,
-      "learning_rate": 7.210351758793971e-05,
-      "loss": 5.3657,
-      "step": 28266
-    },
-    {
-      "epoch": 14.741590612777053,
-      "grad_norm": 1.4896210432052612,
-      "learning_rate": 7.210251256281407e-05,
-      "loss": 5.2985,
-      "step": 28267
-    },
-    {
-      "epoch": 14.742112125162972,
-      "grad_norm": 1.473684310913086,
-      "learning_rate": 7.210150753768845e-05,
-      "loss": 4.4418,
-      "step": 28268
-    },
-    {
-      "epoch": 14.742633637548892,
-      "grad_norm": 1.350284218788147,
-      "learning_rate": 7.210050251256282e-05,
-      "loss": 5.8448,
-      "step": 28269
-    },
-    {
-      "epoch": 14.743155149934811,
-      "grad_norm": 1.3703519105911255,
-      "learning_rate": 7.209949748743718e-05,
-      "loss": 5.952,
-      "step": 28270
-    },
-    {
-      "epoch": 14.74367666232073,
-      "grad_norm": 1.5316839218139648,
-      "learning_rate": 7.209849246231156e-05,
-      "loss": 5.2272,
-      "step": 28271
-    },
-    {
-      "epoch": 14.74419817470665,
-      "grad_norm": 1.3518047332763672,
-      "learning_rate": 7.209748743718594e-05,
-      "loss": 5.7838,
-      "step": 28272
-    },
-    {
-      "epoch": 14.744719687092568,
-      "grad_norm": 1.5159269571304321,
-      "learning_rate": 7.209648241206031e-05,
-      "loss": 5.4492,
-      "step": 28273
-    },
-    {
-      "epoch": 14.745241199478487,
-      "grad_norm": 1.3966705799102783,
-      "learning_rate": 7.209547738693468e-05,
-      "loss": 5.4034,
-      "step": 28274
-    },
-    {
-      "epoch": 14.745762711864407,
-      "grad_norm": 1.4833393096923828,
-      "learning_rate": 7.209447236180906e-05,
-      "loss": 5.4608,
-      "step": 28275
-    },
-    {
-      "epoch": 14.746284224250326,
-      "grad_norm": 1.5949362516403198,
-      "learning_rate": 7.209346733668342e-05,
-      "loss": 5.2461,
-      "step": 28276
-    },
-    {
-      "epoch": 14.746805736636245,
-      "grad_norm": 1.4408074617385864,
-      "learning_rate": 7.20924623115578e-05,
-      "loss": 5.8062,
-      "step": 28277
-    },
-    {
-      "epoch": 14.747327249022165,
-      "grad_norm": 1.4951478242874146,
-      "learning_rate": 7.209145728643216e-05,
-      "loss": 5.3169,
-      "step": 28278
-    },
-    {
-      "epoch": 14.747848761408083,
-      "grad_norm": 1.3904082775115967,
-      "learning_rate": 7.209045226130654e-05,
-      "loss": 5.3637,
-      "step": 28279
-    },
-    {
-      "epoch": 14.748370273794002,
-      "grad_norm": 1.5419079065322876,
-      "learning_rate": 7.20894472361809e-05,
-      "loss": 5.5449,
-      "step": 28280
-    },
-    {
-      "epoch": 14.748891786179922,
-      "grad_norm": 1.4475973844528198,
-      "learning_rate": 7.208844221105528e-05,
-      "loss": 5.503,
-      "step": 28281
-    },
-    {
-      "epoch": 14.749413298565841,
-      "grad_norm": 1.4799203872680664,
-      "learning_rate": 7.208743718592965e-05,
-      "loss": 5.2097,
-      "step": 28282
-    },
-    {
-      "epoch": 14.74993481095176,
-      "grad_norm": 1.5139479637145996,
-      "learning_rate": 7.208643216080402e-05,
-      "loss": 5.6974,
-      "step": 28283
-    },
-    {
-      "epoch": 14.75045632333768,
-      "grad_norm": 1.370806336402893,
-      "learning_rate": 7.20854271356784e-05,
-      "loss": 5.6556,
-      "step": 28284
-    },
-    {
-      "epoch": 14.750977835723598,
-      "grad_norm": 1.4623289108276367,
-      "learning_rate": 7.208442211055277e-05,
-      "loss": 5.3927,
-      "step": 28285
-    },
-    {
-      "epoch": 14.751499348109517,
-      "grad_norm": 1.593437671661377,
-      "learning_rate": 7.208341708542714e-05,
-      "loss": 5.1311,
-      "step": 28286
-    },
-    {
-      "epoch": 14.752020860495437,
-      "grad_norm": 1.4435275793075562,
-      "learning_rate": 7.208241206030151e-05,
-      "loss": 5.7737,
-      "step": 28287
-    },
-    {
-      "epoch": 14.752542372881356,
-      "grad_norm": 1.4458776712417603,
-      "learning_rate": 7.208140703517589e-05,
-      "loss": 5.8312,
-      "step": 28288
-    },
-    {
-      "epoch": 14.753063885267276,
-      "grad_norm": 1.4766039848327637,
-      "learning_rate": 7.208040201005025e-05,
-      "loss": 5.165,
-      "step": 28289
-    },
-    {
-      "epoch": 14.753585397653195,
-      "grad_norm": 1.3518368005752563,
-      "learning_rate": 7.207939698492463e-05,
-      "loss": 5.7555,
-      "step": 28290
-    },
-    {
-      "epoch": 14.754106910039113,
-      "grad_norm": 1.3766859769821167,
-      "learning_rate": 7.207839195979899e-05,
-      "loss": 5.5264,
-      "step": 28291
-    },
-    {
-      "epoch": 14.754628422425032,
-      "grad_norm": 1.465477705001831,
-      "learning_rate": 7.207738693467337e-05,
-      "loss": 5.4147,
-      "step": 28292
-    },
-    {
-      "epoch": 14.755149934810952,
-      "grad_norm": 1.4935150146484375,
-      "learning_rate": 7.207638190954775e-05,
-      "loss": 5.1194,
-      "step": 28293
-    },
-    {
-      "epoch": 14.755671447196871,
-      "grad_norm": 1.6477320194244385,
-      "learning_rate": 7.207537688442213e-05,
-      "loss": 4.9388,
-      "step": 28294
-    },
-    {
-      "epoch": 14.75619295958279,
-      "grad_norm": 1.5052202939987183,
-      "learning_rate": 7.207437185929649e-05,
-      "loss": 5.4763,
-      "step": 28295
-    },
-    {
-      "epoch": 14.75671447196871,
-      "grad_norm": 1.6609479188919067,
-      "learning_rate": 7.207336683417085e-05,
-      "loss": 5.3993,
-      "step": 28296
-    },
-    {
-      "epoch": 14.757235984354628,
-      "grad_norm": 1.418387770652771,
-      "learning_rate": 7.207236180904523e-05,
-      "loss": 5.2504,
-      "step": 28297
-    },
-    {
-      "epoch": 14.757757496740547,
-      "grad_norm": 1.366052269935608,
-      "learning_rate": 7.20713567839196e-05,
-      "loss": 5.7176,
-      "step": 28298
-    },
-    {
-      "epoch": 14.758279009126467,
-      "grad_norm": 1.4041014909744263,
-      "learning_rate": 7.207035175879397e-05,
-      "loss": 5.0628,
-      "step": 28299
-    },
-    {
-      "epoch": 14.758800521512386,
-      "grad_norm": 1.445055603981018,
-      "learning_rate": 7.206934673366834e-05,
-      "loss": 5.2906,
-      "step": 28300
-    },
-    {
-      "epoch": 14.759322033898306,
-      "grad_norm": 1.4559251070022583,
-      "learning_rate": 7.206834170854272e-05,
-      "loss": 5.7374,
-      "step": 28301
-    },
-    {
-      "epoch": 14.759843546284225,
-      "grad_norm": 1.4874825477600098,
-      "learning_rate": 7.206733668341708e-05,
-      "loss": 5.5361,
-      "step": 28302
-    },
-    {
-      "epoch": 14.760365058670143,
-      "grad_norm": 1.5163094997406006,
-      "learning_rate": 7.206633165829146e-05,
-      "loss": 5.6911,
-      "step": 28303
-    },
-    {
-      "epoch": 14.760886571056062,
-      "grad_norm": 1.4885907173156738,
-      "learning_rate": 7.206532663316584e-05,
-      "loss": 5.5716,
-      "step": 28304
-    },
-    {
-      "epoch": 14.761408083441982,
-      "grad_norm": 1.5022963285446167,
-      "learning_rate": 7.206432160804021e-05,
-      "loss": 5.0909,
-      "step": 28305
-    },
-    {
-      "epoch": 14.761929595827901,
-      "grad_norm": 1.446107268333435,
-      "learning_rate": 7.206331658291458e-05,
-      "loss": 5.6658,
-      "step": 28306
-    },
-    {
-      "epoch": 14.76245110821382,
-      "grad_norm": 1.5042088031768799,
-      "learning_rate": 7.206231155778896e-05,
-      "loss": 5.3198,
-      "step": 28307
-    },
-    {
-      "epoch": 14.76297262059974,
-      "grad_norm": 1.3899598121643066,
-      "learning_rate": 7.206130653266332e-05,
-      "loss": 5.5407,
-      "step": 28308
-    },
-    {
-      "epoch": 14.763494132985658,
-      "grad_norm": 1.4568006992340088,
-      "learning_rate": 7.206030150753768e-05,
-      "loss": 5.4505,
-      "step": 28309
-    },
-    {
-      "epoch": 14.764015645371577,
-      "grad_norm": 1.3055572509765625,
-      "learning_rate": 7.205929648241206e-05,
-      "loss": 5.6943,
-      "step": 28310
-    },
-    {
-      "epoch": 14.764537157757497,
-      "grad_norm": 1.4284460544586182,
-      "learning_rate": 7.205829145728643e-05,
-      "loss": 4.9162,
-      "step": 28311
-    },
-    {
-      "epoch": 14.765058670143416,
-      "grad_norm": 1.3989551067352295,
-      "learning_rate": 7.20572864321608e-05,
-      "loss": 5.2262,
-      "step": 28312
-    },
-    {
-      "epoch": 14.765580182529336,
-      "grad_norm": 1.4879775047302246,
-      "learning_rate": 7.205628140703517e-05,
-      "loss": 5.427,
-      "step": 28313
-    },
-    {
-      "epoch": 14.766101694915255,
-      "grad_norm": 1.4066416025161743,
-      "learning_rate": 7.205527638190955e-05,
-      "loss": 5.1668,
-      "step": 28314
-    },
-    {
-      "epoch": 14.766623207301173,
-      "grad_norm": 1.7043136358261108,
-      "learning_rate": 7.205427135678392e-05,
-      "loss": 4.8676,
-      "step": 28315
-    },
-    {
-      "epoch": 14.767144719687092,
-      "grad_norm": 1.424278974533081,
-      "learning_rate": 7.20532663316583e-05,
-      "loss": 5.5553,
-      "step": 28316
-    },
-    {
-      "epoch": 14.767666232073012,
-      "grad_norm": 1.4645445346832275,
-      "learning_rate": 7.205226130653267e-05,
-      "loss": 5.299,
-      "step": 28317
-    },
-    {
-      "epoch": 14.768187744458931,
-      "grad_norm": 1.4382243156433105,
-      "learning_rate": 7.205125628140704e-05,
-      "loss": 4.7832,
-      "step": 28318
-    },
-    {
-      "epoch": 14.76870925684485,
-      "grad_norm": 1.355554461479187,
-      "learning_rate": 7.205025125628141e-05,
-      "loss": 5.153,
-      "step": 28319
-    },
-    {
-      "epoch": 14.76923076923077,
-      "grad_norm": 1.4121661186218262,
-      "learning_rate": 7.204924623115579e-05,
-      "loss": 5.7511,
-      "step": 28320
-    },
-    {
-      "epoch": 14.769752281616688,
-      "grad_norm": 1.5884666442871094,
-      "learning_rate": 7.204824120603015e-05,
-      "loss": 5.1549,
-      "step": 28321
-    },
-    {
-      "epoch": 14.770273794002607,
-      "grad_norm": 1.4780855178833008,
-      "learning_rate": 7.204723618090453e-05,
-      "loss": 5.1998,
-      "step": 28322
-    },
-    {
-      "epoch": 14.770795306388527,
-      "grad_norm": 1.6719478368759155,
-      "learning_rate": 7.204623115577889e-05,
-      "loss": 4.5059,
-      "step": 28323
-    },
-    {
-      "epoch": 14.771316818774446,
-      "grad_norm": 1.4553958177566528,
-      "learning_rate": 7.204522613065327e-05,
-      "loss": 5.2536,
-      "step": 28324
-    },
-    {
-      "epoch": 14.771838331160366,
-      "grad_norm": 1.476692795753479,
-      "learning_rate": 7.204422110552765e-05,
-      "loss": 5.7004,
-      "step": 28325
-    },
-    {
-      "epoch": 14.772359843546283,
-      "grad_norm": 1.4955296516418457,
-      "learning_rate": 7.204321608040201e-05,
-      "loss": 5.2673,
-      "step": 28326
-    },
-    {
-      "epoch": 14.772881355932203,
-      "grad_norm": 1.4121158123016357,
-      "learning_rate": 7.204221105527639e-05,
-      "loss": 5.394,
-      "step": 28327
-    },
-    {
-      "epoch": 14.773402868318122,
-      "grad_norm": 1.5238326787948608,
-      "learning_rate": 7.204120603015075e-05,
-      "loss": 5.2971,
-      "step": 28328
-    },
-    {
-      "epoch": 14.773924380704042,
-      "grad_norm": 1.478186011314392,
-      "learning_rate": 7.204020100502513e-05,
-      "loss": 5.2873,
-      "step": 28329
-    },
-    {
-      "epoch": 14.774445893089961,
-      "grad_norm": 1.4535528421401978,
-      "learning_rate": 7.20391959798995e-05,
-      "loss": 5.2777,
-      "step": 28330
-    },
-    {
-      "epoch": 14.77496740547588,
-      "grad_norm": 1.4435721635818481,
-      "learning_rate": 7.203819095477387e-05,
-      "loss": 5.5335,
-      "step": 28331
-    },
-    {
-      "epoch": 14.7754889178618,
-      "grad_norm": 1.4555764198303223,
-      "learning_rate": 7.203718592964824e-05,
-      "loss": 5.599,
-      "step": 28332
-    },
-    {
-      "epoch": 14.776010430247718,
-      "grad_norm": 1.4865517616271973,
-      "learning_rate": 7.203618090452262e-05,
-      "loss": 5.8387,
-      "step": 28333
-    },
-    {
-      "epoch": 14.776531942633637,
-      "grad_norm": 1.49631667137146,
-      "learning_rate": 7.203517587939698e-05,
-      "loss": 5.4954,
-      "step": 28334
-    },
-    {
-      "epoch": 14.777053455019557,
-      "grad_norm": 1.5333492755889893,
-      "learning_rate": 7.203417085427136e-05,
-      "loss": 5.3561,
-      "step": 28335
-    },
-    {
-      "epoch": 14.777574967405476,
-      "grad_norm": 1.323593258857727,
-      "learning_rate": 7.203316582914574e-05,
-      "loss": 5.5609,
-      "step": 28336
-    },
-    {
-      "epoch": 14.778096479791396,
-      "grad_norm": 1.4091992378234863,
-      "learning_rate": 7.20321608040201e-05,
-      "loss": 5.2189,
-      "step": 28337
-    },
-    {
-      "epoch": 14.778617992177313,
-      "grad_norm": 1.5417485237121582,
-      "learning_rate": 7.203115577889448e-05,
-      "loss": 4.4859,
-      "step": 28338
-    },
-    {
-      "epoch": 14.779139504563233,
-      "grad_norm": 1.4363986253738403,
-      "learning_rate": 7.203015075376884e-05,
-      "loss": 5.7113,
-      "step": 28339
-    },
-    {
-      "epoch": 14.779661016949152,
-      "grad_norm": 1.4685544967651367,
-      "learning_rate": 7.202914572864322e-05,
-      "loss": 5.7984,
-      "step": 28340
-    },
-    {
-      "epoch": 14.780182529335072,
-      "grad_norm": 1.5190253257751465,
-      "learning_rate": 7.202814070351759e-05,
-      "loss": 5.5261,
-      "step": 28341
-    },
-    {
-      "epoch": 14.780704041720991,
-      "grad_norm": 1.394889235496521,
-      "learning_rate": 7.202713567839196e-05,
-      "loss": 5.5064,
-      "step": 28342
-    },
-    {
-      "epoch": 14.78122555410691,
-      "grad_norm": 1.3970905542373657,
-      "learning_rate": 7.202613065326633e-05,
-      "loss": 5.5503,
-      "step": 28343
-    },
-    {
-      "epoch": 14.78174706649283,
-      "grad_norm": 1.434092402458191,
-      "learning_rate": 7.20251256281407e-05,
-      "loss": 5.2179,
-      "step": 28344
-    },
-    {
-      "epoch": 14.782268578878748,
-      "grad_norm": 1.4187980890274048,
-      "learning_rate": 7.202412060301508e-05,
-      "loss": 5.5165,
-      "step": 28345
-    },
-    {
-      "epoch": 14.782790091264667,
-      "grad_norm": 1.5131964683532715,
-      "learning_rate": 7.202311557788946e-05,
-      "loss": 5.0614,
-      "step": 28346
-    },
-    {
-      "epoch": 14.783311603650587,
-      "grad_norm": 1.5799658298492432,
-      "learning_rate": 7.202211055276383e-05,
-      "loss": 5.4775,
-      "step": 28347
-    },
-    {
-      "epoch": 14.783833116036506,
-      "grad_norm": 1.49422025680542,
-      "learning_rate": 7.20211055276382e-05,
-      "loss": 5.3311,
-      "step": 28348
-    },
-    {
-      "epoch": 14.784354628422426,
-      "grad_norm": 1.441029667854309,
-      "learning_rate": 7.202010050251257e-05,
-      "loss": 4.7424,
-      "step": 28349
-    },
-    {
-      "epoch": 14.784876140808343,
-      "grad_norm": 1.609880805015564,
-      "learning_rate": 7.201909547738693e-05,
-      "loss": 5.5226,
-      "step": 28350
-    },
-    {
-      "epoch": 14.785397653194263,
-      "grad_norm": 1.5126771926879883,
-      "learning_rate": 7.201809045226131e-05,
-      "loss": 5.3379,
-      "step": 28351
-    },
-    {
-      "epoch": 14.785919165580182,
-      "grad_norm": 1.391235589981079,
-      "learning_rate": 7.201708542713567e-05,
-      "loss": 5.4559,
-      "step": 28352
-    },
-    {
-      "epoch": 14.786440677966102,
-      "grad_norm": 1.435037612915039,
-      "learning_rate": 7.201608040201005e-05,
-      "loss": 5.6031,
-      "step": 28353
-    },
-    {
-      "epoch": 14.786962190352021,
-      "grad_norm": 1.3886114358901978,
-      "learning_rate": 7.201507537688442e-05,
-      "loss": 5.786,
-      "step": 28354
-    },
-    {
-      "epoch": 14.78748370273794,
-      "grad_norm": 1.6178139448165894,
-      "learning_rate": 7.20140703517588e-05,
-      "loss": 4.9623,
-      "step": 28355
-    },
-    {
-      "epoch": 14.78800521512386,
-      "grad_norm": 1.51095449924469,
-      "learning_rate": 7.201306532663317e-05,
-      "loss": 5.1405,
-      "step": 28356
-    },
-    {
-      "epoch": 14.788526727509778,
-      "grad_norm": 1.510780692100525,
-      "learning_rate": 7.201206030150755e-05,
-      "loss": 5.5524,
-      "step": 28357
-    },
-    {
-      "epoch": 14.789048239895697,
-      "grad_norm": 1.5358844995498657,
-      "learning_rate": 7.201105527638191e-05,
-      "loss": 5.3524,
-      "step": 28358
-    },
-    {
-      "epoch": 14.789569752281617,
-      "grad_norm": 1.4407707452774048,
-      "learning_rate": 7.201005025125629e-05,
-      "loss": 5.6449,
-      "step": 28359
-    },
-    {
-      "epoch": 14.790091264667536,
-      "grad_norm": 1.582916259765625,
-      "learning_rate": 7.200904522613066e-05,
-      "loss": 5.4356,
-      "step": 28360
-    },
-    {
-      "epoch": 14.790612777053456,
-      "grad_norm": 1.5410966873168945,
-      "learning_rate": 7.200804020100503e-05,
-      "loss": 5.3754,
-      "step": 28361
-    },
-    {
-      "epoch": 14.791134289439373,
-      "grad_norm": 1.4317045211791992,
-      "learning_rate": 7.20070351758794e-05,
-      "loss": 5.6931,
-      "step": 28362
-    },
-    {
-      "epoch": 14.791655801825293,
-      "grad_norm": 1.4341527223587036,
-      "learning_rate": 7.200603015075376e-05,
-      "loss": 5.2815,
-      "step": 28363
-    },
-    {
-      "epoch": 14.792177314211212,
-      "grad_norm": 1.5376836061477661,
-      "learning_rate": 7.200502512562814e-05,
-      "loss": 5.2624,
-      "step": 28364
-    },
-    {
-      "epoch": 14.792698826597132,
-      "grad_norm": 1.4844456911087036,
-      "learning_rate": 7.200402010050252e-05,
-      "loss": 5.4455,
-      "step": 28365
-    },
-    {
-      "epoch": 14.793220338983051,
-      "grad_norm": 1.4864697456359863,
-      "learning_rate": 7.20030150753769e-05,
-      "loss": 5.8301,
-      "step": 28366
-    },
-    {
-      "epoch": 14.79374185136897,
-      "grad_norm": 1.4701523780822754,
-      "learning_rate": 7.200201005025126e-05,
-      "loss": 5.625,
-      "step": 28367
-    },
-    {
-      "epoch": 14.794263363754888,
-      "grad_norm": 1.5076736211776733,
-      "learning_rate": 7.200100502512564e-05,
-      "loss": 5.1025,
-      "step": 28368
-    },
-    {
-      "epoch": 14.794784876140808,
-      "grad_norm": 1.4905478954315186,
-      "learning_rate": 7.2e-05,
-      "loss": 5.1958,
-      "step": 28369
-    },
-    {
-      "epoch": 14.795306388526727,
-      "grad_norm": 1.4511357545852661,
-      "learning_rate": 7.199899497487438e-05,
-      "loss": 5.0876,
-      "step": 28370
-    },
-    {
-      "epoch": 14.795827900912647,
-      "grad_norm": 1.5931291580200195,
-      "learning_rate": 7.199798994974874e-05,
-      "loss": 5.1619,
-      "step": 28371
-    },
-    {
-      "epoch": 14.796349413298566,
-      "grad_norm": 1.5082045793533325,
-      "learning_rate": 7.199698492462312e-05,
-      "loss": 5.1331,
-      "step": 28372
-    },
-    {
-      "epoch": 14.796870925684486,
-      "grad_norm": 1.428027868270874,
-      "learning_rate": 7.199597989949749e-05,
-      "loss": 5.1827,
-      "step": 28373
-    },
-    {
-      "epoch": 14.797392438070403,
-      "grad_norm": 1.5083097219467163,
-      "learning_rate": 7.199497487437186e-05,
-      "loss": 5.1981,
-      "step": 28374
-    },
-    {
-      "epoch": 14.797913950456323,
-      "grad_norm": 1.557309627532959,
-      "learning_rate": 7.199396984924623e-05,
-      "loss": 5.4803,
-      "step": 28375
-    },
-    {
-      "epoch": 14.798435462842242,
-      "grad_norm": 1.4711641073226929,
-      "learning_rate": 7.19929648241206e-05,
-      "loss": 5.3354,
-      "step": 28376
-    },
-    {
-      "epoch": 14.798956975228162,
-      "grad_norm": 1.447561502456665,
-      "learning_rate": 7.199195979899498e-05,
-      "loss": 5.436,
-      "step": 28377
-    },
-    {
-      "epoch": 14.799478487614081,
-      "grad_norm": 1.4566764831542969,
-      "learning_rate": 7.199095477386935e-05,
-      "loss": 5.0752,
-      "step": 28378
-    },
-    {
-      "epoch": 14.8,
-      "grad_norm": 1.4838379621505737,
-      "learning_rate": 7.198994974874373e-05,
-      "loss": 5.5133,
-      "step": 28379
-    },
-    {
-      "epoch": 14.800521512385918,
-      "grad_norm": 1.6570483446121216,
-      "learning_rate": 7.198894472361809e-05,
-      "loss": 4.7989,
-      "step": 28380
-    },
-    {
-      "epoch": 14.801043024771838,
-      "grad_norm": 1.5289421081542969,
-      "learning_rate": 7.198793969849247e-05,
-      "loss": 5.2668,
-      "step": 28381
-    },
-    {
-      "epoch": 14.801564537157757,
-      "grad_norm": 1.4748972654342651,
-      "learning_rate": 7.198693467336683e-05,
-      "loss": 5.2157,
-      "step": 28382
-    },
-    {
-      "epoch": 14.802086049543677,
-      "grad_norm": 1.4037212133407593,
-      "learning_rate": 7.198592964824121e-05,
-      "loss": 5.3171,
-      "step": 28383
-    },
-    {
-      "epoch": 14.802607561929596,
-      "grad_norm": 1.5365548133850098,
-      "learning_rate": 7.198492462311557e-05,
-      "loss": 5.1475,
-      "step": 28384
-    },
-    {
-      "epoch": 14.803129074315516,
-      "grad_norm": 1.3242210149765015,
-      "learning_rate": 7.198391959798995e-05,
-      "loss": 5.6616,
-      "step": 28385
-    },
-    {
-      "epoch": 14.803650586701433,
-      "grad_norm": 1.4808800220489502,
-      "learning_rate": 7.198291457286433e-05,
-      "loss": 5.3096,
-      "step": 28386
-    },
-    {
-      "epoch": 14.804172099087353,
-      "grad_norm": 1.8534784317016602,
-      "learning_rate": 7.198190954773871e-05,
-      "loss": 4.8502,
-      "step": 28387
-    },
-    {
-      "epoch": 14.804693611473272,
-      "grad_norm": 1.3904591798782349,
-      "learning_rate": 7.198090452261307e-05,
-      "loss": 5.4003,
-      "step": 28388
-    },
-    {
-      "epoch": 14.805215123859192,
-      "grad_norm": 1.394013524055481,
-      "learning_rate": 7.197989949748744e-05,
-      "loss": 5.4192,
-      "step": 28389
-    },
-    {
-      "epoch": 14.805736636245111,
-      "grad_norm": 1.357876181602478,
-      "learning_rate": 7.197889447236181e-05,
-      "loss": 5.3615,
-      "step": 28390
-    },
-    {
-      "epoch": 14.80625814863103,
-      "grad_norm": 1.4564399719238281,
-      "learning_rate": 7.197788944723618e-05,
-      "loss": 5.2074,
-      "step": 28391
-    },
-    {
-      "epoch": 14.806779661016948,
-      "grad_norm": 1.4632742404937744,
-      "learning_rate": 7.197688442211056e-05,
-      "loss": 5.4333,
-      "step": 28392
-    },
-    {
-      "epoch": 14.807301173402868,
-      "grad_norm": 1.5250595808029175,
-      "learning_rate": 7.197587939698492e-05,
-      "loss": 5.1162,
-      "step": 28393
-    },
-    {
-      "epoch": 14.807822685788787,
-      "grad_norm": 1.6097304821014404,
-      "learning_rate": 7.19748743718593e-05,
-      "loss": 5.1175,
-      "step": 28394
-    },
-    {
-      "epoch": 14.808344198174707,
-      "grad_norm": 1.4334156513214111,
-      "learning_rate": 7.197386934673366e-05,
-      "loss": 5.3288,
-      "step": 28395
-    },
-    {
-      "epoch": 14.808865710560626,
-      "grad_norm": 1.479110836982727,
-      "learning_rate": 7.197286432160804e-05,
-      "loss": 4.779,
-      "step": 28396
-    },
-    {
-      "epoch": 14.809387222946546,
-      "grad_norm": 1.4070950746536255,
-      "learning_rate": 7.197185929648242e-05,
-      "loss": 5.4859,
-      "step": 28397
-    },
-    {
-      "epoch": 14.809908735332463,
-      "grad_norm": 1.41886305809021,
-      "learning_rate": 7.19708542713568e-05,
-      "loss": 5.5098,
-      "step": 28398
-    },
-    {
-      "epoch": 14.810430247718383,
-      "grad_norm": 1.4982377290725708,
-      "learning_rate": 7.196984924623116e-05,
-      "loss": 5.5956,
-      "step": 28399
-    },
-    {
-      "epoch": 14.810951760104302,
-      "grad_norm": 1.4552727937698364,
-      "learning_rate": 7.196884422110554e-05,
-      "loss": 5.4708,
-      "step": 28400
-    },
-    {
-      "epoch": 14.811473272490222,
-      "grad_norm": 1.3452742099761963,
-      "learning_rate": 7.19678391959799e-05,
-      "loss": 5.5445,
-      "step": 28401
-    },
-    {
-      "epoch": 14.811994784876141,
-      "grad_norm": 1.5315724611282349,
-      "learning_rate": 7.196683417085427e-05,
-      "loss": 5.2928,
-      "step": 28402
-    },
-    {
-      "epoch": 14.81251629726206,
-      "grad_norm": 1.5228047370910645,
-      "learning_rate": 7.196582914572864e-05,
-      "loss": 4.9602,
-      "step": 28403
-    },
-    {
-      "epoch": 14.813037809647978,
-      "grad_norm": 1.4356403350830078,
-      "learning_rate": 7.196482412060301e-05,
-      "loss": 5.302,
-      "step": 28404
-    },
-    {
-      "epoch": 14.813559322033898,
-      "grad_norm": 1.3993754386901855,
-      "learning_rate": 7.196381909547739e-05,
-      "loss": 4.9415,
-      "step": 28405
-    },
-    {
-      "epoch": 14.814080834419817,
-      "grad_norm": 1.4802138805389404,
-      "learning_rate": 7.196281407035176e-05,
-      "loss": 4.922,
-      "step": 28406
-    },
-    {
-      "epoch": 14.814602346805737,
-      "grad_norm": 1.468979835510254,
-      "learning_rate": 7.196180904522614e-05,
-      "loss": 5.6573,
-      "step": 28407
-    },
-    {
-      "epoch": 14.815123859191656,
-      "grad_norm": 1.5335102081298828,
-      "learning_rate": 7.19608040201005e-05,
-      "loss": 4.9407,
-      "step": 28408
-    },
-    {
-      "epoch": 14.815645371577576,
-      "grad_norm": 1.5413364171981812,
-      "learning_rate": 7.195979899497488e-05,
-      "loss": 5.4383,
-      "step": 28409
-    },
-    {
-      "epoch": 14.816166883963493,
-      "grad_norm": 1.4566144943237305,
-      "learning_rate": 7.195879396984925e-05,
-      "loss": 5.4406,
-      "step": 28410
-    },
-    {
-      "epoch": 14.816688396349413,
-      "grad_norm": 1.570470929145813,
-      "learning_rate": 7.195778894472363e-05,
-      "loss": 4.8597,
-      "step": 28411
-    },
-    {
-      "epoch": 14.817209908735332,
-      "grad_norm": 1.4882588386535645,
-      "learning_rate": 7.195678391959799e-05,
-      "loss": 4.9482,
-      "step": 28412
-    },
-    {
-      "epoch": 14.817731421121252,
-      "grad_norm": 1.4591947793960571,
-      "learning_rate": 7.195577889447237e-05,
-      "loss": 4.8319,
-      "step": 28413
-    },
-    {
-      "epoch": 14.818252933507171,
-      "grad_norm": 1.4119561910629272,
-      "learning_rate": 7.195477386934673e-05,
-      "loss": 5.4829,
-      "step": 28414
-    },
-    {
-      "epoch": 14.81877444589309,
-      "grad_norm": 1.493438482284546,
-      "learning_rate": 7.195376884422111e-05,
-      "loss": 5.8317,
-      "step": 28415
-    },
-    {
-      "epoch": 14.819295958279008,
-      "grad_norm": 1.5846713781356812,
-      "learning_rate": 7.195276381909548e-05,
-      "loss": 4.6832,
-      "step": 28416
-    },
-    {
-      "epoch": 14.819817470664928,
-      "grad_norm": 1.5117803812026978,
-      "learning_rate": 7.195175879396985e-05,
-      "loss": 5.1369,
-      "step": 28417
-    },
-    {
-      "epoch": 14.820338983050847,
-      "grad_norm": 1.5690916776657104,
-      "learning_rate": 7.195075376884423e-05,
-      "loss": 5.1886,
-      "step": 28418
-    },
-    {
-      "epoch": 14.820860495436767,
-      "grad_norm": 1.364010214805603,
-      "learning_rate": 7.19497487437186e-05,
-      "loss": 5.9098,
-      "step": 28419
-    },
-    {
-      "epoch": 14.821382007822686,
-      "grad_norm": 1.5129034519195557,
-      "learning_rate": 7.194874371859297e-05,
-      "loss": 5.7335,
-      "step": 28420
-    },
-    {
-      "epoch": 14.821903520208604,
-      "grad_norm": 1.3470282554626465,
-      "learning_rate": 7.194773869346734e-05,
-      "loss": 5.4748,
-      "step": 28421
-    },
-    {
-      "epoch": 14.822425032594523,
-      "grad_norm": 1.5953823328018188,
-      "learning_rate": 7.194673366834171e-05,
-      "loss": 5.1413,
-      "step": 28422
-    },
-    {
-      "epoch": 14.822946544980443,
-      "grad_norm": 1.3719133138656616,
-      "learning_rate": 7.194572864321608e-05,
-      "loss": 5.3976,
-      "step": 28423
-    },
-    {
-      "epoch": 14.823468057366362,
-      "grad_norm": 1.4776089191436768,
-      "learning_rate": 7.194472361809046e-05,
-      "loss": 5.5218,
-      "step": 28424
-    },
-    {
-      "epoch": 14.823989569752282,
-      "grad_norm": 1.5377651453018188,
-      "learning_rate": 7.194371859296482e-05,
-      "loss": 5.13,
-      "step": 28425
-    },
-    {
-      "epoch": 14.824511082138201,
-      "grad_norm": 1.3713260889053345,
-      "learning_rate": 7.19427135678392e-05,
-      "loss": 5.4168,
-      "step": 28426
-    },
-    {
-      "epoch": 14.82503259452412,
-      "grad_norm": 1.419805645942688,
-      "learning_rate": 7.194170854271358e-05,
-      "loss": 5.7095,
-      "step": 28427
-    },
-    {
-      "epoch": 14.825554106910038,
-      "grad_norm": 1.399126648902893,
-      "learning_rate": 7.194070351758795e-05,
-      "loss": 5.3711,
-      "step": 28428
-    },
-    {
-      "epoch": 14.826075619295958,
-      "grad_norm": 1.5632078647613525,
-      "learning_rate": 7.193969849246232e-05,
-      "loss": 5.0673,
-      "step": 28429
-    },
-    {
-      "epoch": 14.826597131681877,
-      "grad_norm": 1.4745502471923828,
-      "learning_rate": 7.193869346733668e-05,
-      "loss": 5.6218,
-      "step": 28430
-    },
-    {
-      "epoch": 14.827118644067797,
-      "grad_norm": 1.5655299425125122,
-      "learning_rate": 7.193768844221106e-05,
-      "loss": 5.1065,
-      "step": 28431
-    },
-    {
-      "epoch": 14.827640156453716,
-      "grad_norm": 1.6170392036437988,
-      "learning_rate": 7.193668341708543e-05,
-      "loss": 5.3579,
-      "step": 28432
-    },
-    {
-      "epoch": 14.828161668839634,
-      "grad_norm": 1.551432490348816,
-      "learning_rate": 7.19356783919598e-05,
-      "loss": 5.1874,
-      "step": 28433
-    },
-    {
-      "epoch": 14.828683181225554,
-      "grad_norm": 1.5570260286331177,
-      "learning_rate": 7.193467336683417e-05,
-      "loss": 5.2596,
-      "step": 28434
-    },
-    {
-      "epoch": 14.829204693611473,
-      "grad_norm": 1.5406066179275513,
-      "learning_rate": 7.193366834170855e-05,
-      "loss": 5.4944,
-      "step": 28435
-    },
-    {
-      "epoch": 14.829726205997392,
-      "grad_norm": 1.499942421913147,
-      "learning_rate": 7.193266331658291e-05,
-      "loss": 5.4868,
-      "step": 28436
-    },
-    {
-      "epoch": 14.830247718383312,
-      "grad_norm": 1.4518357515335083,
-      "learning_rate": 7.193165829145729e-05,
-      "loss": 5.8492,
-      "step": 28437
-    },
-    {
-      "epoch": 14.830769230769231,
-      "grad_norm": 1.4348112344741821,
-      "learning_rate": 7.193065326633167e-05,
-      "loss": 5.9227,
-      "step": 28438
-    },
-    {
-      "epoch": 14.83129074315515,
-      "grad_norm": 1.4345792531967163,
-      "learning_rate": 7.192964824120604e-05,
-      "loss": 4.8307,
-      "step": 28439
-    },
-    {
-      "epoch": 14.831812255541069,
-      "grad_norm": 1.475465178489685,
-      "learning_rate": 7.192864321608041e-05,
-      "loss": 5.2229,
-      "step": 28440
-    },
-    {
-      "epoch": 14.832333767926988,
-      "grad_norm": 1.4975261688232422,
-      "learning_rate": 7.192763819095479e-05,
-      "loss": 4.8877,
-      "step": 28441
-    },
-    {
-      "epoch": 14.832855280312907,
-      "grad_norm": 1.599745273590088,
-      "learning_rate": 7.192663316582915e-05,
-      "loss": 5.386,
-      "step": 28442
-    },
-    {
-      "epoch": 14.833376792698827,
-      "grad_norm": 1.5644596815109253,
-      "learning_rate": 7.192562814070351e-05,
-      "loss": 4.8309,
-      "step": 28443
-    },
-    {
-      "epoch": 14.833898305084746,
-      "grad_norm": 1.4847742319107056,
-      "learning_rate": 7.192462311557789e-05,
-      "loss": 4.7934,
-      "step": 28444
-    },
-    {
-      "epoch": 14.834419817470664,
-      "grad_norm": 1.4732805490493774,
-      "learning_rate": 7.192361809045226e-05,
-      "loss": 5.3213,
-      "step": 28445
-    },
-    {
-      "epoch": 14.834941329856584,
-      "grad_norm": 1.6270469427108765,
-      "learning_rate": 7.192261306532663e-05,
-      "loss": 4.852,
-      "step": 28446
-    },
-    {
-      "epoch": 14.835462842242503,
-      "grad_norm": 1.5830719470977783,
-      "learning_rate": 7.192160804020101e-05,
-      "loss": 5.6551,
-      "step": 28447
-    },
-    {
-      "epoch": 14.835984354628422,
-      "grad_norm": 1.5434421300888062,
-      "learning_rate": 7.192060301507539e-05,
-      "loss": 5.3986,
-      "step": 28448
-    },
-    {
-      "epoch": 14.836505867014342,
-      "grad_norm": 1.5566529035568237,
-      "learning_rate": 7.191959798994975e-05,
-      "loss": 5.0085,
-      "step": 28449
-    },
-    {
-      "epoch": 14.837027379400261,
-      "grad_norm": 1.4844553470611572,
-      "learning_rate": 7.191859296482413e-05,
-      "loss": 4.5127,
-      "step": 28450
-    },
-    {
-      "epoch": 14.83754889178618,
-      "grad_norm": 1.530661702156067,
-      "learning_rate": 7.19175879396985e-05,
-      "loss": 4.7483,
-      "step": 28451
-    },
-    {
-      "epoch": 14.838070404172099,
-      "grad_norm": 1.5277894735336304,
-      "learning_rate": 7.191658291457287e-05,
-      "loss": 5.042,
-      "step": 28452
-    },
-    {
-      "epoch": 14.838591916558018,
-      "grad_norm": 1.551499843597412,
-      "learning_rate": 7.191557788944724e-05,
-      "loss": 5.145,
-      "step": 28453
-    },
-    {
-      "epoch": 14.839113428943937,
-      "grad_norm": 1.3951458930969238,
-      "learning_rate": 7.191457286432162e-05,
-      "loss": 4.9487,
-      "step": 28454
-    },
-    {
-      "epoch": 14.839634941329857,
-      "grad_norm": 1.373840570449829,
-      "learning_rate": 7.191356783919598e-05,
-      "loss": 5.5079,
-      "step": 28455
-    },
-    {
-      "epoch": 14.840156453715776,
-      "grad_norm": 1.4616882801055908,
-      "learning_rate": 7.191256281407034e-05,
-      "loss": 5.3934,
-      "step": 28456
-    },
-    {
-      "epoch": 14.840677966101694,
-      "grad_norm": 1.5003257989883423,
-      "learning_rate": 7.191155778894472e-05,
-      "loss": 4.8351,
-      "step": 28457
-    },
-    {
-      "epoch": 14.841199478487614,
-      "grad_norm": 1.5650326013565063,
-      "learning_rate": 7.19105527638191e-05,
-      "loss": 5.4433,
-      "step": 28458
-    },
-    {
-      "epoch": 14.841720990873533,
-      "grad_norm": 1.5418931245803833,
-      "learning_rate": 7.190954773869348e-05,
-      "loss": 5.3534,
-      "step": 28459
-    },
-    {
-      "epoch": 14.842242503259452,
-      "grad_norm": 1.4036118984222412,
-      "learning_rate": 7.190854271356784e-05,
-      "loss": 5.5627,
-      "step": 28460
-    },
-    {
-      "epoch": 14.842764015645372,
-      "grad_norm": 1.4264036417007446,
-      "learning_rate": 7.190753768844222e-05,
-      "loss": 5.3587,
-      "step": 28461
-    },
-    {
-      "epoch": 14.843285528031291,
-      "grad_norm": 1.3498971462249756,
-      "learning_rate": 7.190653266331658e-05,
-      "loss": 5.8245,
-      "step": 28462
-    },
-    {
-      "epoch": 14.843807040417209,
-      "grad_norm": 1.4877630472183228,
-      "learning_rate": 7.190552763819096e-05,
-      "loss": 5.2044,
-      "step": 28463
-    },
-    {
-      "epoch": 14.844328552803129,
-      "grad_norm": 1.4211270809173584,
-      "learning_rate": 7.190452261306533e-05,
-      "loss": 4.981,
-      "step": 28464
-    },
-    {
-      "epoch": 14.844850065189048,
-      "grad_norm": 1.5325255393981934,
-      "learning_rate": 7.19035175879397e-05,
-      "loss": 5.4268,
-      "step": 28465
-    },
-    {
-      "epoch": 14.845371577574968,
-      "grad_norm": 1.595265507698059,
-      "learning_rate": 7.190251256281407e-05,
-      "loss": 5.0014,
-      "step": 28466
-    },
-    {
-      "epoch": 14.845893089960887,
-      "grad_norm": 1.553394079208374,
-      "learning_rate": 7.190150753768845e-05,
-      "loss": 4.919,
-      "step": 28467
-    },
-    {
-      "epoch": 14.846414602346806,
-      "grad_norm": 1.4917656183242798,
-      "learning_rate": 7.190050251256282e-05,
-      "loss": 4.5267,
-      "step": 28468
-    },
-    {
-      "epoch": 14.846936114732724,
-      "grad_norm": 1.5339040756225586,
-      "learning_rate": 7.189949748743719e-05,
-      "loss": 5.317,
-      "step": 28469
-    },
-    {
-      "epoch": 14.847457627118644,
-      "grad_norm": 1.4710890054702759,
-      "learning_rate": 7.189849246231157e-05,
-      "loss": 4.9503,
-      "step": 28470
-    },
-    {
-      "epoch": 14.847979139504563,
-      "grad_norm": 1.5651543140411377,
-      "learning_rate": 7.189748743718593e-05,
-      "loss": 5.4159,
-      "step": 28471
-    },
-    {
-      "epoch": 14.848500651890483,
-      "grad_norm": 1.4557793140411377,
-      "learning_rate": 7.189648241206031e-05,
-      "loss": 5.4221,
-      "step": 28472
-    },
-    {
-      "epoch": 14.849022164276402,
-      "grad_norm": 1.467147946357727,
-      "learning_rate": 7.189547738693467e-05,
-      "loss": 5.4341,
-      "step": 28473
-    },
-    {
-      "epoch": 14.849543676662321,
-      "grad_norm": 1.5413587093353271,
-      "learning_rate": 7.189447236180905e-05,
-      "loss": 5.1178,
-      "step": 28474
-    },
-    {
-      "epoch": 14.85006518904824,
-      "grad_norm": 1.4408968687057495,
-      "learning_rate": 7.189346733668341e-05,
-      "loss": 5.8745,
-      "step": 28475
-    },
-    {
-      "epoch": 14.850586701434159,
-      "grad_norm": 1.4764072895050049,
-      "learning_rate": 7.189246231155779e-05,
-      "loss": 5.6623,
-      "step": 28476
-    },
-    {
-      "epoch": 14.851108213820078,
-      "grad_norm": 1.479858160018921,
-      "learning_rate": 7.189145728643216e-05,
-      "loss": 5.366,
-      "step": 28477
-    },
-    {
-      "epoch": 14.851629726205998,
-      "grad_norm": 1.5269756317138672,
-      "learning_rate": 7.189045226130653e-05,
-      "loss": 5.239,
-      "step": 28478
-    },
-    {
-      "epoch": 14.852151238591917,
-      "grad_norm": 1.5052521228790283,
-      "learning_rate": 7.188944723618091e-05,
-      "loss": 5.5053,
-      "step": 28479
-    },
-    {
-      "epoch": 14.852672750977836,
-      "grad_norm": 1.466199278831482,
-      "learning_rate": 7.188844221105529e-05,
-      "loss": 5.7134,
-      "step": 28480
-    },
-    {
-      "epoch": 14.853194263363754,
-      "grad_norm": 1.4813376665115356,
-      "learning_rate": 7.188743718592965e-05,
-      "loss": 5.4507,
-      "step": 28481
-    },
-    {
-      "epoch": 14.853715775749674,
-      "grad_norm": 1.4107283353805542,
-      "learning_rate": 7.188643216080402e-05,
-      "loss": 5.9255,
-      "step": 28482
-    },
-    {
-      "epoch": 14.854237288135593,
-      "grad_norm": 1.393752932548523,
-      "learning_rate": 7.18854271356784e-05,
-      "loss": 5.4088,
-      "step": 28483
-    },
-    {
-      "epoch": 14.854758800521513,
-      "grad_norm": 1.506169319152832,
-      "learning_rate": 7.188442211055276e-05,
-      "loss": 5.6168,
-      "step": 28484
-    },
-    {
-      "epoch": 14.855280312907432,
-      "grad_norm": 1.6763156652450562,
-      "learning_rate": 7.188341708542714e-05,
-      "loss": 5.3946,
-      "step": 28485
-    },
-    {
-      "epoch": 14.855801825293351,
-      "grad_norm": 1.3866115808486938,
-      "learning_rate": 7.18824120603015e-05,
-      "loss": 5.3661,
-      "step": 28486
-    },
-    {
-      "epoch": 14.85632333767927,
-      "grad_norm": 1.4301475286483765,
-      "learning_rate": 7.188140703517588e-05,
-      "loss": 5.3326,
-      "step": 28487
-    },
-    {
-      "epoch": 14.856844850065189,
-      "grad_norm": 1.5268535614013672,
-      "learning_rate": 7.188040201005025e-05,
-      "loss": 5.2061,
-      "step": 28488
-    },
-    {
-      "epoch": 14.857366362451108,
-      "grad_norm": 1.4306271076202393,
-      "learning_rate": 7.187939698492462e-05,
-      "loss": 5.188,
-      "step": 28489
-    },
-    {
-      "epoch": 14.857887874837028,
-      "grad_norm": 1.4544148445129395,
-      "learning_rate": 7.1878391959799e-05,
-      "loss": 5.0992,
-      "step": 28490
-    },
-    {
-      "epoch": 14.858409387222947,
-      "grad_norm": 1.4755271673202515,
-      "learning_rate": 7.187738693467338e-05,
-      "loss": 5.2298,
-      "step": 28491
-    },
-    {
-      "epoch": 14.858930899608866,
-      "grad_norm": 1.5356642007827759,
-      "learning_rate": 7.187638190954774e-05,
-      "loss": 5.5798,
-      "step": 28492
-    },
-    {
-      "epoch": 14.859452411994784,
-      "grad_norm": 1.4326049089431763,
-      "learning_rate": 7.187537688442212e-05,
-      "loss": 5.6761,
-      "step": 28493
-    },
-    {
-      "epoch": 14.859973924380704,
-      "grad_norm": 1.5045281648635864,
-      "learning_rate": 7.187437185929648e-05,
-      "loss": 5.4778,
-      "step": 28494
-    },
-    {
-      "epoch": 14.860495436766623,
-      "grad_norm": 1.52534019947052,
-      "learning_rate": 7.187336683417086e-05,
-      "loss": 5.1587,
-      "step": 28495
-    },
-    {
-      "epoch": 14.861016949152543,
-      "grad_norm": 1.5338770151138306,
-      "learning_rate": 7.187236180904523e-05,
-      "loss": 5.1416,
-      "step": 28496
-    },
-    {
-      "epoch": 14.861538461538462,
-      "grad_norm": 1.4782872200012207,
-      "learning_rate": 7.187135678391959e-05,
-      "loss": 5.2463,
-      "step": 28497
-    },
-    {
-      "epoch": 14.862059973924381,
-      "grad_norm": 1.52986741065979,
-      "learning_rate": 7.187035175879397e-05,
-      "loss": 5.4209,
-      "step": 28498
-    },
-    {
-      "epoch": 14.8625814863103,
-      "grad_norm": 1.4439809322357178,
-      "learning_rate": 7.186934673366835e-05,
-      "loss": 5.474,
-      "step": 28499
-    },
-    {
-      "epoch": 14.863102998696219,
-      "grad_norm": 1.4654728174209595,
-      "learning_rate": 7.186834170854272e-05,
-      "loss": 5.6038,
-      "step": 28500
-    },
-    {
-      "epoch": 14.863624511082138,
-      "grad_norm": 1.4442986249923706,
-      "learning_rate": 7.186733668341709e-05,
-      "loss": 5.5965,
-      "step": 28501
-    },
-    {
-      "epoch": 14.864146023468058,
-      "grad_norm": 1.444387674331665,
-      "learning_rate": 7.186633165829147e-05,
-      "loss": 5.6047,
-      "step": 28502
-    },
-    {
-      "epoch": 14.864667535853977,
-      "grad_norm": 1.3230631351470947,
-      "learning_rate": 7.186532663316583e-05,
-      "loss": 5.4063,
-      "step": 28503
-    },
-    {
-      "epoch": 14.865189048239897,
-      "grad_norm": 1.5281953811645508,
-      "learning_rate": 7.186432160804021e-05,
-      "loss": 5.1765,
-      "step": 28504
-    },
-    {
-      "epoch": 14.865710560625814,
-      "grad_norm": 1.514102578163147,
-      "learning_rate": 7.186331658291457e-05,
-      "loss": 4.942,
-      "step": 28505
-    },
-    {
-      "epoch": 14.866232073011734,
-      "grad_norm": 1.4939452409744263,
-      "learning_rate": 7.186231155778895e-05,
-      "loss": 5.6459,
-      "step": 28506
-    },
-    {
-      "epoch": 14.866753585397653,
-      "grad_norm": 1.4178060293197632,
-      "learning_rate": 7.186130653266332e-05,
-      "loss": 5.6221,
-      "step": 28507
-    },
-    {
-      "epoch": 14.867275097783573,
-      "grad_norm": 1.3868366479873657,
-      "learning_rate": 7.186030150753769e-05,
-      "loss": 5.2898,
-      "step": 28508
-    },
-    {
-      "epoch": 14.867796610169492,
-      "grad_norm": 1.4257410764694214,
-      "learning_rate": 7.185929648241206e-05,
-      "loss": 5.29,
-      "step": 28509
-    },
-    {
-      "epoch": 14.868318122555412,
-      "grad_norm": 1.3984729051589966,
-      "learning_rate": 7.185829145728644e-05,
-      "loss": 5.48,
-      "step": 28510
-    },
-    {
-      "epoch": 14.86883963494133,
-      "grad_norm": 1.4240092039108276,
-      "learning_rate": 7.185728643216081e-05,
-      "loss": 5.5168,
-      "step": 28511
-    },
-    {
-      "epoch": 14.869361147327249,
-      "grad_norm": 1.4490954875946045,
-      "learning_rate": 7.185628140703518e-05,
-      "loss": 5.3373,
-      "step": 28512
-    },
-    {
-      "epoch": 14.869882659713168,
-      "grad_norm": 1.47867751121521,
-      "learning_rate": 7.185527638190956e-05,
-      "loss": 5.6068,
-      "step": 28513
-    },
-    {
-      "epoch": 14.870404172099088,
-      "grad_norm": 1.4989025592803955,
-      "learning_rate": 7.185427135678392e-05,
-      "loss": 5.4276,
-      "step": 28514
-    },
-    {
-      "epoch": 14.870925684485007,
-      "grad_norm": 1.4092265367507935,
-      "learning_rate": 7.18532663316583e-05,
-      "loss": 5.4641,
-      "step": 28515
-    },
-    {
-      "epoch": 14.871447196870925,
-      "grad_norm": 1.4016475677490234,
-      "learning_rate": 7.185226130653266e-05,
-      "loss": 5.6234,
-      "step": 28516
-    },
-    {
-      "epoch": 14.871968709256844,
-      "grad_norm": 1.4328488111495972,
-      "learning_rate": 7.185125628140704e-05,
-      "loss": 5.249,
-      "step": 28517
-    },
-    {
-      "epoch": 14.872490221642764,
-      "grad_norm": 1.4412909746170044,
-      "learning_rate": 7.18502512562814e-05,
-      "loss": 5.411,
-      "step": 28518
-    },
-    {
-      "epoch": 14.873011734028683,
-      "grad_norm": 1.504004955291748,
-      "learning_rate": 7.184924623115578e-05,
-      "loss": 5.3801,
-      "step": 28519
-    },
-    {
-      "epoch": 14.873533246414603,
-      "grad_norm": 1.4553966522216797,
-      "learning_rate": 7.184824120603016e-05,
-      "loss": 5.4607,
-      "step": 28520
-    },
-    {
-      "epoch": 14.874054758800522,
-      "grad_norm": 1.4020143747329712,
-      "learning_rate": 7.184723618090454e-05,
-      "loss": 4.9835,
-      "step": 28521
-    },
-    {
-      "epoch": 14.874576271186442,
-      "grad_norm": 1.434863567352295,
-      "learning_rate": 7.18462311557789e-05,
-      "loss": 5.3752,
-      "step": 28522
-    },
-    {
-      "epoch": 14.87509778357236,
-      "grad_norm": 1.5193005800247192,
-      "learning_rate": 7.184522613065327e-05,
-      "loss": 5.4591,
-      "step": 28523
-    },
-    {
-      "epoch": 14.875619295958279,
-      "grad_norm": 1.5169168710708618,
-      "learning_rate": 7.184422110552764e-05,
-      "loss": 5.2539,
-      "step": 28524
-    },
-    {
-      "epoch": 14.876140808344198,
-      "grad_norm": 1.5206594467163086,
-      "learning_rate": 7.184321608040201e-05,
-      "loss": 5.2334,
-      "step": 28525
-    },
-    {
-      "epoch": 14.876662320730118,
-      "grad_norm": 1.441630482673645,
-      "learning_rate": 7.184221105527639e-05,
-      "loss": 5.7879,
-      "step": 28526
-    },
-    {
-      "epoch": 14.877183833116037,
-      "grad_norm": 1.5630656480789185,
-      "learning_rate": 7.184120603015075e-05,
-      "loss": 5.464,
-      "step": 28527
-    },
-    {
-      "epoch": 14.877705345501955,
-      "grad_norm": 1.582668423652649,
-      "learning_rate": 7.184020100502513e-05,
-      "loss": 5.5892,
-      "step": 28528
-    },
-    {
-      "epoch": 14.878226857887874,
-      "grad_norm": 1.4438753128051758,
-      "learning_rate": 7.183919597989949e-05,
-      "loss": 5.4492,
-      "step": 28529
-    },
-    {
-      "epoch": 14.878748370273794,
-      "grad_norm": 1.4977989196777344,
-      "learning_rate": 7.183819095477387e-05,
-      "loss": 4.9181,
-      "step": 28530
-    },
-    {
-      "epoch": 14.879269882659713,
-      "grad_norm": 1.6182748079299927,
-      "learning_rate": 7.183718592964825e-05,
-      "loss": 5.1795,
-      "step": 28531
-    },
-    {
-      "epoch": 14.879791395045633,
-      "grad_norm": 1.3695989847183228,
-      "learning_rate": 7.183618090452263e-05,
-      "loss": 5.7331,
-      "step": 28532
-    },
-    {
-      "epoch": 14.880312907431552,
-      "grad_norm": 1.4448599815368652,
-      "learning_rate": 7.183517587939699e-05,
-      "loss": 5.0581,
-      "step": 28533
-    },
-    {
-      "epoch": 14.880834419817472,
-      "grad_norm": 1.42927086353302,
-      "learning_rate": 7.183417085427137e-05,
-      "loss": 5.0363,
-      "step": 28534
-    },
-    {
-      "epoch": 14.88135593220339,
-      "grad_norm": 1.4527174234390259,
-      "learning_rate": 7.183316582914573e-05,
-      "loss": 5.4772,
-      "step": 28535
-    },
-    {
-      "epoch": 14.881877444589309,
-      "grad_norm": 1.4350348711013794,
-      "learning_rate": 7.18321608040201e-05,
-      "loss": 5.7193,
-      "step": 28536
-    },
-    {
-      "epoch": 14.882398956975228,
-      "grad_norm": 1.442254662513733,
-      "learning_rate": 7.183115577889447e-05,
-      "loss": 5.6013,
-      "step": 28537
-    },
-    {
-      "epoch": 14.882920469361148,
-      "grad_norm": 1.4083036184310913,
-      "learning_rate": 7.183015075376884e-05,
-      "loss": 5.5739,
-      "step": 28538
-    },
-    {
-      "epoch": 14.883441981747067,
-      "grad_norm": 1.5321266651153564,
-      "learning_rate": 7.182914572864322e-05,
-      "loss": 4.7514,
-      "step": 28539
-    },
-    {
-      "epoch": 14.883963494132985,
-      "grad_norm": 1.414580225944519,
-      "learning_rate": 7.18281407035176e-05,
-      "loss": 4.9258,
-      "step": 28540
-    },
-    {
-      "epoch": 14.884485006518904,
-      "grad_norm": 1.4018856287002563,
-      "learning_rate": 7.182713567839197e-05,
-      "loss": 5.6742,
-      "step": 28541
-    },
-    {
-      "epoch": 14.885006518904824,
-      "grad_norm": 1.4193462133407593,
-      "learning_rate": 7.182613065326634e-05,
-      "loss": 5.8229,
-      "step": 28542
-    },
-    {
-      "epoch": 14.885528031290743,
-      "grad_norm": 1.4967334270477295,
-      "learning_rate": 7.182512562814071e-05,
-      "loss": 5.4188,
-      "step": 28543
-    },
-    {
-      "epoch": 14.886049543676663,
-      "grad_norm": 1.5790925025939941,
-      "learning_rate": 7.182412060301508e-05,
-      "loss": 5.3826,
-      "step": 28544
-    },
-    {
-      "epoch": 14.886571056062582,
-      "grad_norm": 1.3971916437149048,
-      "learning_rate": 7.182311557788946e-05,
-      "loss": 5.7249,
-      "step": 28545
-    },
-    {
-      "epoch": 14.887092568448502,
-      "grad_norm": 1.434005618095398,
-      "learning_rate": 7.182211055276382e-05,
-      "loss": 5.399,
-      "step": 28546
-    },
-    {
-      "epoch": 14.88761408083442,
-      "grad_norm": 1.4434638023376465,
-      "learning_rate": 7.18211055276382e-05,
-      "loss": 5.0261,
-      "step": 28547
-    },
-    {
-      "epoch": 14.888135593220339,
-      "grad_norm": 1.5755486488342285,
-      "learning_rate": 7.182010050251256e-05,
-      "loss": 5.0893,
-      "step": 28548
-    },
-    {
-      "epoch": 14.888657105606258,
-      "grad_norm": 1.4088103771209717,
-      "learning_rate": 7.181909547738693e-05,
-      "loss": 5.5614,
-      "step": 28549
-    },
-    {
-      "epoch": 14.889178617992178,
-      "grad_norm": 1.3984018564224243,
-      "learning_rate": 7.18180904522613e-05,
-      "loss": 5.7938,
-      "step": 28550
-    },
-    {
-      "epoch": 14.889700130378097,
-      "grad_norm": 1.454939365386963,
-      "learning_rate": 7.181708542713568e-05,
-      "loss": 5.6006,
-      "step": 28551
-    },
-    {
-      "epoch": 14.890221642764015,
-      "grad_norm": 1.5137314796447754,
-      "learning_rate": 7.181608040201006e-05,
-      "loss": 5.4549,
-      "step": 28552
-    },
-    {
-      "epoch": 14.890743155149934,
-      "grad_norm": 1.3212642669677734,
-      "learning_rate": 7.181507537688442e-05,
-      "loss": 5.8129,
-      "step": 28553
-    },
-    {
-      "epoch": 14.891264667535854,
-      "grad_norm": 1.4778705835342407,
-      "learning_rate": 7.18140703517588e-05,
-      "loss": 5.1765,
-      "step": 28554
-    },
-    {
-      "epoch": 14.891786179921773,
-      "grad_norm": 1.452956199645996,
-      "learning_rate": 7.181306532663317e-05,
-      "loss": 5.8042,
-      "step": 28555
-    },
-    {
-      "epoch": 14.892307692307693,
-      "grad_norm": 1.4149595499038696,
-      "learning_rate": 7.181206030150754e-05,
-      "loss": 5.391,
-      "step": 28556
-    },
-    {
-      "epoch": 14.892829204693612,
-      "grad_norm": 1.4447195529937744,
-      "learning_rate": 7.181105527638191e-05,
-      "loss": 5.4602,
-      "step": 28557
-    },
-    {
-      "epoch": 14.89335071707953,
-      "grad_norm": 1.4163416624069214,
-      "learning_rate": 7.181005025125629e-05,
-      "loss": 5.6654,
-      "step": 28558
-    },
-    {
-      "epoch": 14.89387222946545,
-      "grad_norm": 1.42918860912323,
-      "learning_rate": 7.180904522613065e-05,
-      "loss": 5.4031,
-      "step": 28559
-    },
-    {
-      "epoch": 14.894393741851369,
-      "grad_norm": 1.4167267084121704,
-      "learning_rate": 7.180804020100503e-05,
-      "loss": 5.3974,
-      "step": 28560
-    },
-    {
-      "epoch": 14.894915254237288,
-      "grad_norm": 1.4058295488357544,
-      "learning_rate": 7.18070351758794e-05,
-      "loss": 5.596,
-      "step": 28561
-    },
-    {
-      "epoch": 14.895436766623208,
-      "grad_norm": 1.3447445631027222,
-      "learning_rate": 7.180603015075377e-05,
-      "loss": 5.7504,
-      "step": 28562
-    },
-    {
-      "epoch": 14.895958279009127,
-      "grad_norm": 1.4927042722702026,
-      "learning_rate": 7.180502512562815e-05,
-      "loss": 5.4397,
-      "step": 28563
-    },
-    {
-      "epoch": 14.896479791395045,
-      "grad_norm": 1.430709958076477,
-      "learning_rate": 7.180402010050251e-05,
-      "loss": 5.7546,
-      "step": 28564
-    },
-    {
-      "epoch": 14.897001303780964,
-      "grad_norm": 1.4624061584472656,
-      "learning_rate": 7.180301507537689e-05,
-      "loss": 5.0466,
-      "step": 28565
-    },
-    {
-      "epoch": 14.897522816166884,
-      "grad_norm": 1.4894922971725464,
-      "learning_rate": 7.180201005025125e-05,
-      "loss": 5.3254,
-      "step": 28566
-    },
-    {
-      "epoch": 14.898044328552803,
-      "grad_norm": 1.4027808904647827,
-      "learning_rate": 7.180100502512563e-05,
-      "loss": 5.3938,
-      "step": 28567
-    },
-    {
-      "epoch": 14.898565840938723,
-      "grad_norm": 1.4182857275009155,
-      "learning_rate": 7.18e-05,
-      "loss": 5.4782,
-      "step": 28568
-    },
-    {
-      "epoch": 14.899087353324642,
-      "grad_norm": 1.488807201385498,
-      "learning_rate": 7.179899497487437e-05,
-      "loss": 5.5033,
-      "step": 28569
-    },
-    {
-      "epoch": 14.89960886571056,
-      "grad_norm": 1.5593899488449097,
-      "learning_rate": 7.179798994974874e-05,
-      "loss": 4.9237,
-      "step": 28570
-    },
-    {
-      "epoch": 14.90013037809648,
-      "grad_norm": 1.4154565334320068,
-      "learning_rate": 7.179698492462312e-05,
-      "loss": 5.5958,
-      "step": 28571
-    },
-    {
-      "epoch": 14.900651890482399,
-      "grad_norm": 1.507839560508728,
-      "learning_rate": 7.17959798994975e-05,
-      "loss": 5.2761,
-      "step": 28572
-    },
-    {
-      "epoch": 14.901173402868318,
-      "grad_norm": 1.470206379890442,
-      "learning_rate": 7.179497487437187e-05,
-      "loss": 5.0811,
-      "step": 28573
-    },
-    {
-      "epoch": 14.901694915254238,
-      "grad_norm": 1.3512649536132812,
-      "learning_rate": 7.179396984924624e-05,
-      "loss": 5.6868,
-      "step": 28574
-    },
-    {
-      "epoch": 14.902216427640157,
-      "grad_norm": 1.4668861627578735,
-      "learning_rate": 7.17929648241206e-05,
-      "loss": 5.4509,
-      "step": 28575
-    },
-    {
-      "epoch": 14.902737940026075,
-      "grad_norm": 1.5129486322402954,
-      "learning_rate": 7.179195979899498e-05,
-      "loss": 5.3588,
-      "step": 28576
-    },
-    {
-      "epoch": 14.903259452411994,
-      "grad_norm": 1.4918949604034424,
-      "learning_rate": 7.179095477386934e-05,
-      "loss": 5.6984,
-      "step": 28577
-    },
-    {
-      "epoch": 14.903780964797914,
-      "grad_norm": 1.4602915048599243,
-      "learning_rate": 7.178994974874372e-05,
-      "loss": 5.6435,
-      "step": 28578
-    },
-    {
-      "epoch": 14.904302477183833,
-      "grad_norm": 1.3953052759170532,
-      "learning_rate": 7.178894472361809e-05,
-      "loss": 5.3959,
-      "step": 28579
-    },
-    {
-      "epoch": 14.904823989569753,
-      "grad_norm": 1.4091367721557617,
-      "learning_rate": 7.178793969849246e-05,
-      "loss": 5.2371,
-      "step": 28580
-    },
-    {
-      "epoch": 14.905345501955672,
-      "grad_norm": 1.455794334411621,
-      "learning_rate": 7.178693467336684e-05,
-      "loss": 5.1858,
-      "step": 28581
-    },
-    {
-      "epoch": 14.90586701434159,
-      "grad_norm": 1.5143053531646729,
-      "learning_rate": 7.178592964824122e-05,
-      "loss": 5.2005,
-      "step": 28582
-    },
-    {
-      "epoch": 14.90638852672751,
-      "grad_norm": 1.5901391506195068,
-      "learning_rate": 7.178492462311558e-05,
-      "loss": 5.3154,
-      "step": 28583
-    },
-    {
-      "epoch": 14.906910039113429,
-      "grad_norm": 1.6330052614212036,
-      "learning_rate": 7.178391959798996e-05,
-      "loss": 4.9081,
-      "step": 28584
-    },
-    {
-      "epoch": 14.907431551499348,
-      "grad_norm": 1.4562150239944458,
-      "learning_rate": 7.178291457286433e-05,
-      "loss": 4.9312,
-      "step": 28585
-    },
-    {
-      "epoch": 14.907953063885268,
-      "grad_norm": 1.4999668598175049,
-      "learning_rate": 7.17819095477387e-05,
-      "loss": 5.5974,
-      "step": 28586
-    },
-    {
-      "epoch": 14.908474576271187,
-      "grad_norm": 1.6403470039367676,
-      "learning_rate": 7.178090452261307e-05,
-      "loss": 5.606,
-      "step": 28587
-    },
-    {
-      "epoch": 14.908996088657105,
-      "grad_norm": 1.4517935514450073,
-      "learning_rate": 7.177989949748745e-05,
-      "loss": 5.2713,
-      "step": 28588
-    },
-    {
-      "epoch": 14.909517601043024,
-      "grad_norm": 1.402818202972412,
-      "learning_rate": 7.177889447236181e-05,
-      "loss": 5.7872,
-      "step": 28589
-    },
-    {
-      "epoch": 14.910039113428944,
-      "grad_norm": 1.4280301332473755,
-      "learning_rate": 7.177788944723617e-05,
-      "loss": 5.2489,
-      "step": 28590
-    },
-    {
-      "epoch": 14.910560625814863,
-      "grad_norm": 1.4343523979187012,
-      "learning_rate": 7.177688442211055e-05,
-      "loss": 5.6438,
-      "step": 28591
-    },
-    {
-      "epoch": 14.911082138200783,
-      "grad_norm": 1.4379621744155884,
-      "learning_rate": 7.177587939698493e-05,
-      "loss": 5.5759,
-      "step": 28592
-    },
-    {
-      "epoch": 14.911603650586702,
-      "grad_norm": 1.4925847053527832,
-      "learning_rate": 7.177487437185931e-05,
-      "loss": 4.8752,
-      "step": 28593
-    },
-    {
-      "epoch": 14.91212516297262,
-      "grad_norm": 1.5479967594146729,
-      "learning_rate": 7.177386934673367e-05,
-      "loss": 5.2418,
-      "step": 28594
-    },
-    {
-      "epoch": 14.91264667535854,
-      "grad_norm": 1.5298296213150024,
-      "learning_rate": 7.177286432160805e-05,
-      "loss": 4.9118,
-      "step": 28595
-    },
-    {
-      "epoch": 14.913168187744459,
-      "grad_norm": 1.497225046157837,
-      "learning_rate": 7.177185929648241e-05,
-      "loss": 5.2403,
-      "step": 28596
-    },
-    {
-      "epoch": 14.913689700130378,
-      "grad_norm": 1.5244134664535522,
-      "learning_rate": 7.177085427135679e-05,
-      "loss": 5.1606,
-      "step": 28597
-    },
-    {
-      "epoch": 14.914211212516298,
-      "grad_norm": 1.4511218070983887,
-      "learning_rate": 7.176984924623116e-05,
-      "loss": 5.7562,
-      "step": 28598
-    },
-    {
-      "epoch": 14.914732724902217,
-      "grad_norm": 1.4166908264160156,
-      "learning_rate": 7.176884422110553e-05,
-      "loss": 4.9948,
-      "step": 28599
-    },
-    {
-      "epoch": 14.915254237288135,
-      "grad_norm": 1.6513055562973022,
-      "learning_rate": 7.17678391959799e-05,
-      "loss": 5.1704,
-      "step": 28600
-    },
-    {
-      "epoch": 14.915775749674054,
-      "grad_norm": 1.4948807954788208,
-      "learning_rate": 7.176683417085428e-05,
-      "loss": 5.5428,
-      "step": 28601
-    },
-    {
-      "epoch": 14.916297262059974,
-      "grad_norm": 1.4504224061965942,
-      "learning_rate": 7.176582914572865e-05,
-      "loss": 5.2379,
-      "step": 28602
-    },
-    {
-      "epoch": 14.916818774445893,
-      "grad_norm": 1.4944734573364258,
-      "learning_rate": 7.176482412060302e-05,
-      "loss": 5.191,
-      "step": 28603
-    },
-    {
-      "epoch": 14.917340286831813,
-      "grad_norm": 1.4826029539108276,
-      "learning_rate": 7.17638190954774e-05,
-      "loss": 5.7172,
-      "step": 28604
-    },
-    {
-      "epoch": 14.917861799217732,
-      "grad_norm": 1.5684783458709717,
-      "learning_rate": 7.176281407035176e-05,
-      "loss": 5.448,
-      "step": 28605
-    },
-    {
-      "epoch": 14.91838331160365,
-      "grad_norm": 1.5861703157424927,
-      "learning_rate": 7.176180904522614e-05,
-      "loss": 4.3413,
-      "step": 28606
-    },
-    {
-      "epoch": 14.91890482398957,
-      "grad_norm": 1.4855420589447021,
-      "learning_rate": 7.17608040201005e-05,
-      "loss": 4.8423,
-      "step": 28607
-    },
-    {
-      "epoch": 14.919426336375489,
-      "grad_norm": 1.4785687923431396,
-      "learning_rate": 7.175979899497488e-05,
-      "loss": 5.3938,
-      "step": 28608
-    },
-    {
-      "epoch": 14.919947848761408,
-      "grad_norm": 1.5513410568237305,
-      "learning_rate": 7.175879396984924e-05,
-      "loss": 5.4591,
-      "step": 28609
-    },
-    {
-      "epoch": 14.920469361147328,
-      "grad_norm": 1.4795007705688477,
-      "learning_rate": 7.175778894472362e-05,
-      "loss": 5.3527,
-      "step": 28610
-    },
-    {
-      "epoch": 14.920990873533245,
-      "grad_norm": 1.4792543649673462,
-      "learning_rate": 7.175678391959799e-05,
-      "loss": 5.0938,
-      "step": 28611
-    },
-    {
-      "epoch": 14.921512385919165,
-      "grad_norm": 1.3774155378341675,
-      "learning_rate": 7.175577889447236e-05,
-      "loss": 5.8174,
-      "step": 28612
-    },
-    {
-      "epoch": 14.922033898305084,
-      "grad_norm": 1.481491208076477,
-      "learning_rate": 7.175477386934674e-05,
-      "loss": 5.4991,
-      "step": 28613
-    },
-    {
-      "epoch": 14.922555410691004,
-      "grad_norm": 1.5664291381835938,
-      "learning_rate": 7.175376884422112e-05,
-      "loss": 4.9276,
-      "step": 28614
-    },
-    {
-      "epoch": 14.923076923076923,
-      "grad_norm": 1.4393399953842163,
-      "learning_rate": 7.175276381909548e-05,
-      "loss": 5.4528,
-      "step": 28615
-    },
-    {
-      "epoch": 14.923598435462843,
-      "grad_norm": 1.5230170488357544,
-      "learning_rate": 7.175175879396985e-05,
-      "loss": 5.551,
-      "step": 28616
-    },
-    {
-      "epoch": 14.924119947848762,
-      "grad_norm": 1.597256064414978,
-      "learning_rate": 7.175075376884423e-05,
-      "loss": 4.686,
-      "step": 28617
-    },
-    {
-      "epoch": 14.92464146023468,
-      "grad_norm": 1.4927494525909424,
-      "learning_rate": 7.174974874371859e-05,
-      "loss": 5.5701,
-      "step": 28618
-    },
-    {
-      "epoch": 14.9251629726206,
-      "grad_norm": 1.5902197360992432,
-      "learning_rate": 7.174874371859297e-05,
-      "loss": 5.5252,
-      "step": 28619
-    },
-    {
-      "epoch": 14.925684485006519,
-      "grad_norm": 1.4441150426864624,
-      "learning_rate": 7.174773869346733e-05,
-      "loss": 5.2231,
-      "step": 28620
-    },
-    {
-      "epoch": 14.926205997392438,
-      "grad_norm": 1.3923417329788208,
-      "learning_rate": 7.174673366834171e-05,
-      "loss": 5.5077,
-      "step": 28621
-    },
-    {
-      "epoch": 14.926727509778358,
-      "grad_norm": 1.5291404724121094,
-      "learning_rate": 7.174572864321609e-05,
-      "loss": 5.4399,
-      "step": 28622
-    },
-    {
-      "epoch": 14.927249022164276,
-      "grad_norm": 1.4962871074676514,
-      "learning_rate": 7.174472361809047e-05,
-      "loss": 5.5526,
-      "step": 28623
-    },
-    {
-      "epoch": 14.927770534550195,
-      "grad_norm": 1.5145909786224365,
-      "learning_rate": 7.174371859296483e-05,
-      "loss": 5.5887,
-      "step": 28624
-    },
-    {
-      "epoch": 14.928292046936114,
-      "grad_norm": 1.3450857400894165,
-      "learning_rate": 7.174271356783921e-05,
-      "loss": 5.7805,
-      "step": 28625
-    },
-    {
-      "epoch": 14.928813559322034,
-      "grad_norm": 1.3729137182235718,
-      "learning_rate": 7.174170854271357e-05,
-      "loss": 5.6355,
-      "step": 28626
-    },
-    {
-      "epoch": 14.929335071707953,
-      "grad_norm": 1.4083070755004883,
-      "learning_rate": 7.174070351758795e-05,
-      "loss": 5.5669,
-      "step": 28627
-    },
-    {
-      "epoch": 14.929856584093873,
-      "grad_norm": 1.5167216062545776,
-      "learning_rate": 7.173969849246231e-05,
-      "loss": 5.4578,
-      "step": 28628
-    },
-    {
-      "epoch": 14.930378096479792,
-      "grad_norm": 1.4094116687774658,
-      "learning_rate": 7.173869346733668e-05,
-      "loss": 5.3137,
-      "step": 28629
-    },
-    {
-      "epoch": 14.93089960886571,
-      "grad_norm": 1.4761216640472412,
-      "learning_rate": 7.173768844221106e-05,
-      "loss": 5.1205,
-      "step": 28630
-    },
-    {
-      "epoch": 14.93142112125163,
-      "grad_norm": 1.552886962890625,
-      "learning_rate": 7.173668341708542e-05,
-      "loss": 4.7728,
-      "step": 28631
-    },
-    {
-      "epoch": 14.931942633637549,
-      "grad_norm": 1.547668695449829,
-      "learning_rate": 7.17356783919598e-05,
-      "loss": 5.3969,
-      "step": 28632
-    },
-    {
-      "epoch": 14.932464146023468,
-      "grad_norm": 1.5625660419464111,
-      "learning_rate": 7.173467336683418e-05,
-      "loss": 5.1843,
-      "step": 28633
-    },
-    {
-      "epoch": 14.932985658409388,
-      "grad_norm": 1.4544841051101685,
-      "learning_rate": 7.173366834170855e-05,
-      "loss": 5.5686,
-      "step": 28634
-    },
-    {
-      "epoch": 14.933507170795306,
-      "grad_norm": 1.4401129484176636,
-      "learning_rate": 7.173266331658292e-05,
-      "loss": 5.1295,
-      "step": 28635
-    },
-    {
-      "epoch": 14.934028683181225,
-      "grad_norm": 1.4355723857879639,
-      "learning_rate": 7.17316582914573e-05,
-      "loss": 5.6651,
-      "step": 28636
-    },
-    {
-      "epoch": 14.934550195567144,
-      "grad_norm": 1.485883116722107,
-      "learning_rate": 7.173065326633166e-05,
-      "loss": 5.2165,
-      "step": 28637
-    },
-    {
-      "epoch": 14.935071707953064,
-      "grad_norm": 1.425299048423767,
-      "learning_rate": 7.172964824120604e-05,
-      "loss": 5.6276,
-      "step": 28638
-    },
-    {
-      "epoch": 14.935593220338983,
-      "grad_norm": 1.4804247617721558,
-      "learning_rate": 7.17286432160804e-05,
-      "loss": 4.9658,
-      "step": 28639
-    },
-    {
-      "epoch": 14.936114732724903,
-      "grad_norm": 1.492874026298523,
-      "learning_rate": 7.172763819095478e-05,
-      "loss": 5.4473,
-      "step": 28640
-    },
-    {
-      "epoch": 14.93663624511082,
-      "grad_norm": 1.4532079696655273,
-      "learning_rate": 7.172663316582914e-05,
-      "loss": 5.6644,
-      "step": 28641
-    },
-    {
-      "epoch": 14.93715775749674,
-      "grad_norm": 1.604250431060791,
-      "learning_rate": 7.172562814070352e-05,
-      "loss": 4.8808,
-      "step": 28642
-    },
-    {
-      "epoch": 14.93767926988266,
-      "grad_norm": 1.4377472400665283,
-      "learning_rate": 7.17246231155779e-05,
-      "loss": 5.6787,
-      "step": 28643
-    },
-    {
-      "epoch": 14.938200782268579,
-      "grad_norm": 1.4968500137329102,
-      "learning_rate": 7.172361809045226e-05,
-      "loss": 5.2968,
-      "step": 28644
-    },
-    {
-      "epoch": 14.938722294654498,
-      "grad_norm": 1.5776867866516113,
-      "learning_rate": 7.172261306532664e-05,
-      "loss": 5.0384,
-      "step": 28645
-    },
-    {
-      "epoch": 14.939243807040418,
-      "grad_norm": 1.498242974281311,
-      "learning_rate": 7.1721608040201e-05,
-      "loss": 5.7009,
-      "step": 28646
-    },
-    {
-      "epoch": 14.939765319426336,
-      "grad_norm": 1.4323538541793823,
-      "learning_rate": 7.172060301507538e-05,
-      "loss": 5.5764,
-      "step": 28647
-    },
-    {
-      "epoch": 14.940286831812255,
-      "grad_norm": 1.532089114189148,
-      "learning_rate": 7.171959798994975e-05,
-      "loss": 5.2511,
-      "step": 28648
-    },
-    {
-      "epoch": 14.940808344198174,
-      "grad_norm": 1.5790698528289795,
-      "learning_rate": 7.171859296482413e-05,
-      "loss": 5.3183,
-      "step": 28649
-    },
-    {
-      "epoch": 14.941329856584094,
-      "grad_norm": 1.5105369091033936,
-      "learning_rate": 7.171758793969849e-05,
-      "loss": 4.8529,
-      "step": 28650
-    },
-    {
-      "epoch": 14.941851368970013,
-      "grad_norm": 1.4573839902877808,
-      "learning_rate": 7.171658291457287e-05,
-      "loss": 4.4762,
-      "step": 28651
-    },
-    {
-      "epoch": 14.942372881355933,
-      "grad_norm": 1.3724313974380493,
-      "learning_rate": 7.171557788944723e-05,
-      "loss": 5.4987,
-      "step": 28652
-    },
-    {
-      "epoch": 14.94289439374185,
-      "grad_norm": 1.438246488571167,
-      "learning_rate": 7.171457286432161e-05,
-      "loss": 5.8217,
-      "step": 28653
-    },
-    {
-      "epoch": 14.94341590612777,
-      "grad_norm": 1.5645272731781006,
-      "learning_rate": 7.171356783919599e-05,
-      "loss": 4.77,
-      "step": 28654
-    },
-    {
-      "epoch": 14.94393741851369,
-      "grad_norm": 1.6361523866653442,
-      "learning_rate": 7.171256281407035e-05,
-      "loss": 4.5458,
-      "step": 28655
-    },
-    {
-      "epoch": 14.944458930899609,
-      "grad_norm": 1.441245436668396,
-      "learning_rate": 7.171155778894473e-05,
-      "loss": 5.2776,
-      "step": 28656
-    },
-    {
-      "epoch": 14.944980443285528,
-      "grad_norm": 1.4473341703414917,
-      "learning_rate": 7.17105527638191e-05,
-      "loss": 5.0768,
-      "step": 28657
-    },
-    {
-      "epoch": 14.945501955671448,
-      "grad_norm": 1.4376143217086792,
-      "learning_rate": 7.170954773869347e-05,
-      "loss": 5.4546,
-      "step": 28658
-    },
-    {
-      "epoch": 14.946023468057366,
-      "grad_norm": 1.449034571647644,
-      "learning_rate": 7.170854271356784e-05,
-      "loss": 4.9609,
-      "step": 28659
-    },
-    {
-      "epoch": 14.946544980443285,
-      "grad_norm": 1.3720182180404663,
-      "learning_rate": 7.170753768844222e-05,
-      "loss": 5.8586,
-      "step": 28660
-    },
-    {
-      "epoch": 14.947066492829205,
-      "grad_norm": 1.3695019483566284,
-      "learning_rate": 7.170653266331658e-05,
-      "loss": 5.4249,
-      "step": 28661
-    },
-    {
-      "epoch": 14.947588005215124,
-      "grad_norm": 1.4081997871398926,
-      "learning_rate": 7.170552763819096e-05,
-      "loss": 5.7932,
-      "step": 28662
-    },
-    {
-      "epoch": 14.948109517601043,
-      "grad_norm": 1.3486179113388062,
-      "learning_rate": 7.170452261306532e-05,
-      "loss": 5.7482,
-      "step": 28663
-    },
-    {
-      "epoch": 14.948631029986963,
-      "grad_norm": 1.4321311712265015,
-      "learning_rate": 7.17035175879397e-05,
-      "loss": 5.247,
-      "step": 28664
-    },
-    {
-      "epoch": 14.94915254237288,
-      "grad_norm": 1.3675172328948975,
-      "learning_rate": 7.170251256281408e-05,
-      "loss": 5.5435,
-      "step": 28665
-    },
-    {
-      "epoch": 14.9496740547588,
-      "grad_norm": 1.5517513751983643,
-      "learning_rate": 7.170150753768845e-05,
-      "loss": 5.5078,
-      "step": 28666
-    },
-    {
-      "epoch": 14.95019556714472,
-      "grad_norm": 1.4665303230285645,
-      "learning_rate": 7.170050251256282e-05,
-      "loss": 5.1849,
-      "step": 28667
-    },
-    {
-      "epoch": 14.950717079530639,
-      "grad_norm": 1.5832356214523315,
-      "learning_rate": 7.169949748743718e-05,
-      "loss": 5.2357,
-      "step": 28668
-    },
-    {
-      "epoch": 14.951238591916558,
-      "grad_norm": 1.4969896078109741,
-      "learning_rate": 7.169849246231156e-05,
-      "loss": 4.9454,
-      "step": 28669
-    },
-    {
-      "epoch": 14.951760104302478,
-      "grad_norm": 1.4712553024291992,
-      "learning_rate": 7.169748743718593e-05,
-      "loss": 5.358,
-      "step": 28670
-    },
-    {
-      "epoch": 14.952281616688396,
-      "grad_norm": 1.4642703533172607,
-      "learning_rate": 7.16964824120603e-05,
-      "loss": 5.4412,
-      "step": 28671
-    },
-    {
-      "epoch": 14.952803129074315,
-      "grad_norm": 1.5053106546401978,
-      "learning_rate": 7.169547738693467e-05,
-      "loss": 5.6361,
-      "step": 28672
-    },
-    {
-      "epoch": 14.953324641460235,
-      "grad_norm": 1.5268771648406982,
-      "learning_rate": 7.169447236180905e-05,
-      "loss": 5.1085,
-      "step": 28673
-    },
-    {
-      "epoch": 14.953846153846154,
-      "grad_norm": 1.498722791671753,
-      "learning_rate": 7.169346733668342e-05,
-      "loss": 4.9676,
-      "step": 28674
-    },
-    {
-      "epoch": 14.954367666232073,
-      "grad_norm": 1.4016809463500977,
-      "learning_rate": 7.16924623115578e-05,
-      "loss": 5.4918,
-      "step": 28675
-    },
-    {
-      "epoch": 14.954889178617993,
-      "grad_norm": 1.513654112815857,
-      "learning_rate": 7.169145728643217e-05,
-      "loss": 4.7108,
-      "step": 28676
-    },
-    {
-      "epoch": 14.95541069100391,
-      "grad_norm": 1.5641177892684937,
-      "learning_rate": 7.169045226130654e-05,
-      "loss": 4.6352,
-      "step": 28677
-    },
-    {
-      "epoch": 14.95593220338983,
-      "grad_norm": 1.4486216306686401,
-      "learning_rate": 7.168944723618091e-05,
-      "loss": 5.468,
-      "step": 28678
-    },
-    {
-      "epoch": 14.95645371577575,
-      "grad_norm": 1.4731370210647583,
-      "learning_rate": 7.168844221105529e-05,
-      "loss": 5.5751,
-      "step": 28679
-    },
-    {
-      "epoch": 14.956975228161669,
-      "grad_norm": 1.4224133491516113,
-      "learning_rate": 7.168743718592965e-05,
-      "loss": 5.6919,
-      "step": 28680
-    },
-    {
-      "epoch": 14.957496740547588,
-      "grad_norm": 1.4449156522750854,
-      "learning_rate": 7.168643216080403e-05,
-      "loss": 5.4323,
-      "step": 28681
-    },
-    {
-      "epoch": 14.958018252933508,
-      "grad_norm": 1.4401071071624756,
-      "learning_rate": 7.168542713567839e-05,
-      "loss": 5.4474,
-      "step": 28682
-    },
-    {
-      "epoch": 14.958539765319426,
-      "grad_norm": 1.4329952001571655,
-      "learning_rate": 7.168442211055276e-05,
-      "loss": 5.2515,
-      "step": 28683
-    },
-    {
-      "epoch": 14.959061277705345,
-      "grad_norm": 1.3728079795837402,
-      "learning_rate": 7.168341708542713e-05,
-      "loss": 5.3729,
-      "step": 28684
-    },
-    {
-      "epoch": 14.959582790091265,
-      "grad_norm": 1.388664722442627,
-      "learning_rate": 7.168241206030151e-05,
-      "loss": 5.4933,
-      "step": 28685
-    },
-    {
-      "epoch": 14.960104302477184,
-      "grad_norm": 1.4649243354797363,
-      "learning_rate": 7.168140703517589e-05,
-      "loss": 5.1329,
-      "step": 28686
-    },
-    {
-      "epoch": 14.960625814863103,
-      "grad_norm": 1.4373661279678345,
-      "learning_rate": 7.168040201005025e-05,
-      "loss": 5.4094,
-      "step": 28687
-    },
-    {
-      "epoch": 14.961147327249023,
-      "grad_norm": 1.4167628288269043,
-      "learning_rate": 7.167939698492463e-05,
-      "loss": 5.6761,
-      "step": 28688
-    },
-    {
-      "epoch": 14.96166883963494,
-      "grad_norm": 1.5517792701721191,
-      "learning_rate": 7.1678391959799e-05,
-      "loss": 5.111,
-      "step": 28689
-    },
-    {
-      "epoch": 14.96219035202086,
-      "grad_norm": 1.5337998867034912,
-      "learning_rate": 7.167738693467337e-05,
-      "loss": 5.1616,
-      "step": 28690
-    },
-    {
-      "epoch": 14.96271186440678,
-      "grad_norm": 1.4408414363861084,
-      "learning_rate": 7.167638190954774e-05,
-      "loss": 5.3281,
-      "step": 28691
-    },
-    {
-      "epoch": 14.963233376792699,
-      "grad_norm": 1.3482372760772705,
-      "learning_rate": 7.167537688442212e-05,
-      "loss": 5.6694,
-      "step": 28692
-    },
-    {
-      "epoch": 14.963754889178619,
-      "grad_norm": 1.4203884601593018,
-      "learning_rate": 7.167437185929648e-05,
-      "loss": 5.2107,
-      "step": 28693
-    },
-    {
-      "epoch": 14.964276401564538,
-      "grad_norm": 1.4743424654006958,
-      "learning_rate": 7.167336683417086e-05,
-      "loss": 5.434,
-      "step": 28694
-    },
-    {
-      "epoch": 14.964797913950456,
-      "grad_norm": 1.4434056282043457,
-      "learning_rate": 7.167236180904524e-05,
-      "loss": 5.7652,
-      "step": 28695
-    },
-    {
-      "epoch": 14.965319426336375,
-      "grad_norm": 1.4742836952209473,
-      "learning_rate": 7.16713567839196e-05,
-      "loss": 5.5115,
-      "step": 28696
-    },
-    {
-      "epoch": 14.965840938722295,
-      "grad_norm": 1.4921780824661255,
-      "learning_rate": 7.167035175879398e-05,
-      "loss": 5.6294,
-      "step": 28697
-    },
-    {
-      "epoch": 14.966362451108214,
-      "grad_norm": 1.3196179866790771,
-      "learning_rate": 7.166934673366834e-05,
-      "loss": 5.7158,
-      "step": 28698
-    },
-    {
-      "epoch": 14.966883963494134,
-      "grad_norm": 1.3442649841308594,
-      "learning_rate": 7.166834170854272e-05,
-      "loss": 4.7841,
-      "step": 28699
-    },
-    {
-      "epoch": 14.967405475880053,
-      "grad_norm": 1.3880915641784668,
-      "learning_rate": 7.166733668341708e-05,
-      "loss": 5.2986,
-      "step": 28700
-    },
-    {
-      "epoch": 14.96792698826597,
-      "grad_norm": 1.3242672681808472,
-      "learning_rate": 7.166633165829146e-05,
-      "loss": 5.4306,
-      "step": 28701
-    },
-    {
-      "epoch": 14.96844850065189,
-      "grad_norm": 1.410508155822754,
-      "learning_rate": 7.166532663316583e-05,
-      "loss": 5.4946,
-      "step": 28702
-    },
-    {
-      "epoch": 14.96897001303781,
-      "grad_norm": 1.3651875257492065,
-      "learning_rate": 7.16643216080402e-05,
-      "loss": 5.7432,
-      "step": 28703
-    },
-    {
-      "epoch": 14.969491525423729,
-      "grad_norm": 1.4595333337783813,
-      "learning_rate": 7.166331658291457e-05,
-      "loss": 5.4658,
-      "step": 28704
-    },
-    {
-      "epoch": 14.970013037809649,
-      "grad_norm": 1.3509241342544556,
-      "learning_rate": 7.166231155778895e-05,
-      "loss": 5.1201,
-      "step": 28705
-    },
-    {
-      "epoch": 14.970534550195566,
-      "grad_norm": 1.6749680042266846,
-      "learning_rate": 7.166130653266332e-05,
-      "loss": 5.6887,
-      "step": 28706
-    },
-    {
-      "epoch": 14.971056062581486,
-      "grad_norm": 1.3925038576126099,
-      "learning_rate": 7.16603015075377e-05,
-      "loss": 5.3005,
-      "step": 28707
-    },
-    {
-      "epoch": 14.971577574967405,
-      "grad_norm": 1.4161747694015503,
-      "learning_rate": 7.165929648241207e-05,
-      "loss": 5.6963,
-      "step": 28708
-    },
-    {
-      "epoch": 14.972099087353325,
-      "grad_norm": 1.4100614786148071,
-      "learning_rate": 7.165829145728643e-05,
-      "loss": 5.5372,
-      "step": 28709
-    },
-    {
-      "epoch": 14.972620599739244,
-      "grad_norm": 1.4832944869995117,
-      "learning_rate": 7.165728643216081e-05,
-      "loss": 5.4686,
-      "step": 28710
-    },
-    {
-      "epoch": 14.973142112125164,
-      "grad_norm": 1.4336020946502686,
-      "learning_rate": 7.165628140703517e-05,
-      "loss": 5.741,
-      "step": 28711
-    },
-    {
-      "epoch": 14.973663624511083,
-      "grad_norm": 1.4400634765625,
-      "learning_rate": 7.165527638190955e-05,
-      "loss": 5.5362,
-      "step": 28712
-    },
-    {
-      "epoch": 14.974185136897,
-      "grad_norm": 1.369765281677246,
-      "learning_rate": 7.165427135678391e-05,
-      "loss": 5.3163,
-      "step": 28713
-    },
-    {
-      "epoch": 14.97470664928292,
-      "grad_norm": 1.5993236303329468,
-      "learning_rate": 7.165326633165829e-05,
-      "loss": 5.3853,
-      "step": 28714
-    },
-    {
-      "epoch": 14.97522816166884,
-      "grad_norm": 1.5911661386489868,
-      "learning_rate": 7.165226130653267e-05,
-      "loss": 5.0187,
-      "step": 28715
-    },
-    {
-      "epoch": 14.975749674054759,
-      "grad_norm": 1.5312795639038086,
-      "learning_rate": 7.165125628140705e-05,
-      "loss": 5.6851,
-      "step": 28716
-    },
-    {
-      "epoch": 14.976271186440679,
-      "grad_norm": 1.4324754476547241,
-      "learning_rate": 7.165025125628141e-05,
-      "loss": 5.635,
-      "step": 28717
-    },
-    {
-      "epoch": 14.976792698826596,
-      "grad_norm": 1.5472065210342407,
-      "learning_rate": 7.164924623115579e-05,
-      "loss": 5.2438,
-      "step": 28718
-    },
-    {
-      "epoch": 14.977314211212516,
-      "grad_norm": 1.403737187385559,
-      "learning_rate": 7.164824120603015e-05,
-      "loss": 5.7839,
-      "step": 28719
-    },
-    {
-      "epoch": 14.977835723598435,
-      "grad_norm": 1.4763922691345215,
-      "learning_rate": 7.164723618090453e-05,
-      "loss": 5.4063,
-      "step": 28720
-    },
-    {
-      "epoch": 14.978357235984355,
-      "grad_norm": 1.6391302347183228,
-      "learning_rate": 7.16462311557789e-05,
-      "loss": 4.8056,
-      "step": 28721
-    },
-    {
-      "epoch": 14.978878748370274,
-      "grad_norm": 1.4879814386367798,
-      "learning_rate": 7.164522613065326e-05,
-      "loss": 5.4352,
-      "step": 28722
-    },
-    {
-      "epoch": 14.979400260756194,
-      "grad_norm": 1.5302475690841675,
-      "learning_rate": 7.164422110552764e-05,
-      "loss": 4.8108,
-      "step": 28723
-    },
-    {
-      "epoch": 14.979921773142113,
-      "grad_norm": 1.5435317754745483,
-      "learning_rate": 7.1643216080402e-05,
-      "loss": 5.5326,
-      "step": 28724
-    },
-    {
-      "epoch": 14.98044328552803,
-      "grad_norm": 1.58011794090271,
-      "learning_rate": 7.164221105527638e-05,
-      "loss": 5.333,
-      "step": 28725
-    },
-    {
-      "epoch": 14.98096479791395,
-      "grad_norm": 1.5476481914520264,
-      "learning_rate": 7.164120603015076e-05,
-      "loss": 4.3638,
-      "step": 28726
-    },
-    {
-      "epoch": 14.98148631029987,
-      "grad_norm": 1.5259873867034912,
-      "learning_rate": 7.164020100502514e-05,
-      "loss": 5.6969,
-      "step": 28727
-    },
-    {
-      "epoch": 14.98200782268579,
-      "grad_norm": 1.4289370775222778,
-      "learning_rate": 7.16391959798995e-05,
-      "loss": 5.2811,
-      "step": 28728
-    },
-    {
-      "epoch": 14.982529335071709,
-      "grad_norm": 1.4673734903335571,
-      "learning_rate": 7.163819095477388e-05,
-      "loss": 5.4536,
-      "step": 28729
-    },
-    {
-      "epoch": 14.983050847457626,
-      "grad_norm": 1.4638323783874512,
-      "learning_rate": 7.163718592964824e-05,
-      "loss": 5.8664,
-      "step": 28730
-    },
-    {
-      "epoch": 14.983572359843546,
-      "grad_norm": 1.538002610206604,
-      "learning_rate": 7.163618090452262e-05,
-      "loss": 5.3419,
-      "step": 28731
-    },
-    {
-      "epoch": 14.984093872229465,
-      "grad_norm": 1.4747270345687866,
-      "learning_rate": 7.163517587939698e-05,
-      "loss": 4.7022,
-      "step": 28732
-    },
-    {
-      "epoch": 14.984615384615385,
-      "grad_norm": 1.5165743827819824,
-      "learning_rate": 7.163417085427136e-05,
-      "loss": 5.4074,
-      "step": 28733
-    },
-    {
-      "epoch": 14.985136897001304,
-      "grad_norm": 1.460039496421814,
-      "learning_rate": 7.163316582914573e-05,
-      "loss": 5.4383,
-      "step": 28734
-    },
-    {
-      "epoch": 14.985658409387224,
-      "grad_norm": 1.5856627225875854,
-      "learning_rate": 7.16321608040201e-05,
-      "loss": 5.2963,
-      "step": 28735
-    },
-    {
-      "epoch": 14.986179921773141,
-      "grad_norm": 1.442042589187622,
-      "learning_rate": 7.163115577889448e-05,
-      "loss": 5.4237,
-      "step": 28736
-    },
-    {
-      "epoch": 14.98670143415906,
-      "grad_norm": 1.4758368730545044,
-      "learning_rate": 7.163015075376885e-05,
-      "loss": 5.3933,
-      "step": 28737
-    },
-    {
-      "epoch": 14.98722294654498,
-      "grad_norm": 1.4830396175384521,
-      "learning_rate": 7.162914572864322e-05,
-      "loss": 5.3216,
-      "step": 28738
-    },
-    {
-      "epoch": 14.9877444589309,
-      "grad_norm": 1.3438105583190918,
-      "learning_rate": 7.162814070351759e-05,
-      "loss": 5.2257,
-      "step": 28739
-    },
-    {
-      "epoch": 14.98826597131682,
-      "grad_norm": 1.478973150253296,
-      "learning_rate": 7.162713567839197e-05,
-      "loss": 5.4409,
-      "step": 28740
-    },
-    {
-      "epoch": 14.988787483702739,
-      "grad_norm": 1.5200358629226685,
-      "learning_rate": 7.162613065326633e-05,
-      "loss": 5.2947,
-      "step": 28741
-    },
-    {
-      "epoch": 14.989308996088656,
-      "grad_norm": 1.534571886062622,
-      "learning_rate": 7.162512562814071e-05,
-      "loss": 5.2428,
-      "step": 28742
-    },
-    {
-      "epoch": 14.989830508474576,
-      "grad_norm": 1.499707579612732,
-      "learning_rate": 7.162412060301507e-05,
-      "loss": 5.501,
-      "step": 28743
-    },
-    {
-      "epoch": 14.990352020860495,
-      "grad_norm": 1.3675246238708496,
-      "learning_rate": 7.162311557788945e-05,
-      "loss": 5.8222,
-      "step": 28744
-    },
-    {
-      "epoch": 14.990873533246415,
-      "grad_norm": 1.430826187133789,
-      "learning_rate": 7.162211055276382e-05,
-      "loss": 5.4385,
-      "step": 28745
-    },
-    {
-      "epoch": 14.991395045632334,
-      "grad_norm": 1.5606554746627808,
-      "learning_rate": 7.16211055276382e-05,
-      "loss": 5.1306,
-      "step": 28746
-    },
-    {
-      "epoch": 14.991916558018254,
-      "grad_norm": 1.5068706274032593,
-      "learning_rate": 7.162010050251257e-05,
-      "loss": 5.0742,
-      "step": 28747
-    },
-    {
-      "epoch": 14.992438070404171,
-      "grad_norm": 1.600827932357788,
-      "learning_rate": 7.161909547738694e-05,
-      "loss": 5.6606,
-      "step": 28748
-    },
-    {
-      "epoch": 14.99295958279009,
-      "grad_norm": 1.5486794710159302,
-      "learning_rate": 7.161809045226131e-05,
-      "loss": 5.1085,
-      "step": 28749
-    },
-    {
-      "epoch": 14.99348109517601,
-      "grad_norm": 1.5057594776153564,
-      "learning_rate": 7.161708542713568e-05,
-      "loss": 5.1392,
-      "step": 28750
-    },
-    {
-      "epoch": 14.99400260756193,
-      "grad_norm": 1.4344435930252075,
-      "learning_rate": 7.161608040201006e-05,
-      "loss": 5.1967,
-      "step": 28751
-    },
-    {
-      "epoch": 14.99452411994785,
-      "grad_norm": 1.4849990606307983,
-      "learning_rate": 7.161507537688442e-05,
-      "loss": 5.6232,
-      "step": 28752
-    },
-    {
-      "epoch": 14.995045632333769,
-      "grad_norm": 1.313296914100647,
-      "learning_rate": 7.16140703517588e-05,
-      "loss": 5.5417,
-      "step": 28753
-    },
-    {
-      "epoch": 14.995567144719686,
-      "grad_norm": 1.447009563446045,
-      "learning_rate": 7.161306532663316e-05,
-      "loss": 4.8909,
-      "step": 28754
-    },
-    {
-      "epoch": 14.996088657105606,
-      "grad_norm": 1.429837942123413,
-      "learning_rate": 7.161206030150754e-05,
-      "loss": 5.1868,
-      "step": 28755
-    },
-    {
-      "epoch": 14.996610169491525,
-      "grad_norm": 1.432717204093933,
-      "learning_rate": 7.161105527638192e-05,
-      "loss": 5.465,
-      "step": 28756
-    },
-    {
-      "epoch": 14.997131681877445,
-      "grad_norm": 1.508255958557129,
-      "learning_rate": 7.16100502512563e-05,
-      "loss": 5.409,
-      "step": 28757
-    },
-    {
-      "epoch": 14.997653194263364,
-      "grad_norm": 1.331375002861023,
-      "learning_rate": 7.160904522613066e-05,
-      "loss": 5.7425,
-      "step": 28758
-    },
-    {
-      "epoch": 14.998174706649284,
-      "grad_norm": 1.39933180809021,
-      "learning_rate": 7.160804020100504e-05,
-      "loss": 5.716,
-      "step": 28759
-    },
-    {
-      "epoch": 14.998696219035201,
-      "grad_norm": 1.4322574138641357,
-      "learning_rate": 7.16070351758794e-05,
-      "loss": 5.3355,
-      "step": 28760
-    },
-    {
-      "epoch": 14.99921773142112,
-      "grad_norm": 1.4264367818832397,
-      "learning_rate": 7.160603015075377e-05,
-      "loss": 5.1726,
-      "step": 28761
-    },
-    {
-      "epoch": 14.99973924380704,
-      "grad_norm": 1.40431809425354,
-      "learning_rate": 7.160502512562814e-05,
-      "loss": 5.657,
-      "step": 28762
-    },
-    {
-      "epoch": 15.00026075619296,
-      "grad_norm": 1.6525298357009888,
-      "learning_rate": 7.160402010050251e-05,
-      "loss": 5.3913,
-      "step": 28763
-    },
-    {
-      "epoch": 15.00078226857888,
-      "grad_norm": 1.5092744827270508,
-      "learning_rate": 7.160301507537689e-05,
-      "loss": 4.9462,
-      "step": 28764
-    },
-    {
-      "epoch": 15.001303780964799,
-      "grad_norm": 1.532710075378418,
-      "learning_rate": 7.160201005025125e-05,
-      "loss": 5.0459,
-      "step": 28765
-    },
-    {
-      "epoch": 15.001825293350716,
-      "grad_norm": 1.4828742742538452,
-      "learning_rate": 7.160100502512563e-05,
-      "loss": 5.3585,
-      "step": 28766
-    },
-    {
-      "epoch": 15.002346805736636,
-      "grad_norm": 1.6898969411849976,
-      "learning_rate": 7.16e-05,
-      "loss": 5.0414,
-      "step": 28767
-    },
-    {
-      "epoch": 15.002868318122555,
-      "grad_norm": 1.4259090423583984,
-      "learning_rate": 7.159899497487438e-05,
-      "loss": 5.4224,
-      "step": 28768
-    },
-    {
-      "epoch": 15.003389830508475,
-      "grad_norm": 1.6078959703445435,
-      "learning_rate": 7.159798994974875e-05,
-      "loss": 4.9443,
-      "step": 28769
-    },
-    {
-      "epoch": 15.003911342894394,
-      "grad_norm": 1.4462345838546753,
-      "learning_rate": 7.159698492462313e-05,
-      "loss": 5.259,
-      "step": 28770
-    },
-    {
-      "epoch": 15.004432855280314,
-      "grad_norm": 1.4513009786605835,
-      "learning_rate": 7.159597989949749e-05,
-      "loss": 5.5397,
-      "step": 28771
-    },
-    {
-      "epoch": 15.004954367666231,
-      "grad_norm": 1.4159588813781738,
-      "learning_rate": 7.159497487437187e-05,
-      "loss": 4.8246,
-      "step": 28772
-    },
-    {
-      "epoch": 15.00547588005215,
-      "grad_norm": 1.6263099908828735,
-      "learning_rate": 7.159396984924623e-05,
-      "loss": 5.3909,
-      "step": 28773
-    },
-    {
-      "epoch": 15.00599739243807,
-      "grad_norm": 1.4494686126708984,
-      "learning_rate": 7.159296482412061e-05,
-      "loss": 5.779,
-      "step": 28774
-    },
-    {
-      "epoch": 15.00651890482399,
-      "grad_norm": 1.505715012550354,
-      "learning_rate": 7.159195979899497e-05,
-      "loss": 5.3623,
-      "step": 28775
-    },
-    {
-      "epoch": 15.00704041720991,
-      "grad_norm": 1.3422138690948486,
-      "learning_rate": 7.159095477386935e-05,
-      "loss": 5.6364,
-      "step": 28776
-    },
-    {
-      "epoch": 15.007561929595829,
-      "grad_norm": 1.5295192003250122,
-      "learning_rate": 7.158994974874373e-05,
-      "loss": 5.0962,
-      "step": 28777
-    },
-    {
-      "epoch": 15.008083441981746,
-      "grad_norm": 1.3739656209945679,
-      "learning_rate": 7.15889447236181e-05,
-      "loss": 5.5243,
-      "step": 28778
-    },
-    {
-      "epoch": 15.008604954367666,
-      "grad_norm": 1.358380675315857,
-      "learning_rate": 7.158793969849247e-05,
-      "loss": 5.6585,
-      "step": 28779
-    },
-    {
-      "epoch": 15.009126466753585,
-      "grad_norm": 1.663046956062317,
-      "learning_rate": 7.158693467336684e-05,
-      "loss": 5.1932,
-      "step": 28780
-    },
-    {
-      "epoch": 15.009647979139505,
-      "grad_norm": 1.538774013519287,
-      "learning_rate": 7.158592964824121e-05,
-      "loss": 5.0794,
-      "step": 28781
-    },
-    {
-      "epoch": 15.010169491525424,
-      "grad_norm": 1.4424406290054321,
-      "learning_rate": 7.158492462311558e-05,
-      "loss": 5.5451,
-      "step": 28782
-    },
-    {
-      "epoch": 15.010691003911344,
-      "grad_norm": 1.6574249267578125,
-      "learning_rate": 7.158391959798996e-05,
-      "loss": 4.7721,
-      "step": 28783
-    },
-    {
-      "epoch": 15.011212516297261,
-      "grad_norm": 1.3812527656555176,
-      "learning_rate": 7.158291457286432e-05,
-      "loss": 5.8809,
-      "step": 28784
-    },
-    {
-      "epoch": 15.01173402868318,
-      "grad_norm": 1.3632599115371704,
-      "learning_rate": 7.15819095477387e-05,
-      "loss": 5.6386,
-      "step": 28785
-    },
-    {
-      "epoch": 15.0122555410691,
-      "grad_norm": 1.421990156173706,
-      "learning_rate": 7.158090452261306e-05,
-      "loss": 5.84,
-      "step": 28786
-    },
-    {
-      "epoch": 15.01277705345502,
-      "grad_norm": 1.4958935976028442,
-      "learning_rate": 7.157989949748744e-05,
-      "loss": 5.5306,
-      "step": 28787
-    },
-    {
-      "epoch": 15.01329856584094,
-      "grad_norm": 1.575251817703247,
-      "learning_rate": 7.157889447236182e-05,
-      "loss": 5.0612,
-      "step": 28788
-    },
-    {
-      "epoch": 15.013820078226859,
-      "grad_norm": 1.5203949213027954,
-      "learning_rate": 7.157788944723618e-05,
-      "loss": 5.3774,
-      "step": 28789
-    },
-    {
-      "epoch": 15.014341590612776,
-      "grad_norm": 1.4855563640594482,
-      "learning_rate": 7.157688442211056e-05,
-      "loss": 4.6897,
-      "step": 28790
-    },
-    {
-      "epoch": 15.014863102998696,
-      "grad_norm": 1.359682559967041,
-      "learning_rate": 7.157587939698492e-05,
-      "loss": 5.9481,
-      "step": 28791
-    },
-    {
-      "epoch": 15.015384615384615,
-      "grad_norm": 1.4472228288650513,
-      "learning_rate": 7.15748743718593e-05,
-      "loss": 5.2749,
-      "step": 28792
-    },
-    {
-      "epoch": 15.015906127770535,
-      "grad_norm": 1.5162242650985718,
-      "learning_rate": 7.157386934673367e-05,
-      "loss": 5.5406,
-      "step": 28793
-    },
-    {
-      "epoch": 15.016427640156454,
-      "grad_norm": 1.7267752885818481,
-      "learning_rate": 7.157286432160804e-05,
-      "loss": 4.129,
-      "step": 28794
-    },
-    {
-      "epoch": 15.016949152542374,
-      "grad_norm": 1.4421755075454712,
-      "learning_rate": 7.157185929648241e-05,
-      "loss": 5.0618,
-      "step": 28795
-    },
-    {
-      "epoch": 15.017470664928291,
-      "grad_norm": 1.6397647857666016,
-      "learning_rate": 7.157085427135679e-05,
-      "loss": 5.1878,
-      "step": 28796
-    },
-    {
-      "epoch": 15.01799217731421,
-      "grad_norm": 1.479836344718933,
-      "learning_rate": 7.156984924623116e-05,
-      "loss": 5.1582,
-      "step": 28797
-    },
-    {
-      "epoch": 15.01851368970013,
-      "grad_norm": 1.469161868095398,
-      "learning_rate": 7.156884422110554e-05,
-      "loss": 5.5185,
-      "step": 28798
-    },
-    {
-      "epoch": 15.01903520208605,
-      "grad_norm": 1.543969988822937,
-      "learning_rate": 7.15678391959799e-05,
-      "loss": 4.8048,
-      "step": 28799
-    },
-    {
-      "epoch": 15.01955671447197,
-      "grad_norm": 1.4900189638137817,
-      "learning_rate": 7.156683417085428e-05,
-      "loss": 5.6398,
-      "step": 28800
-    },
-    {
-      "epoch": 15.020078226857889,
-      "grad_norm": 1.339597463607788,
-      "learning_rate": 7.156582914572865e-05,
-      "loss": 5.4221,
-      "step": 28801
-    },
-    {
-      "epoch": 15.020599739243806,
-      "grad_norm": 1.3894256353378296,
-      "learning_rate": 7.156482412060301e-05,
-      "loss": 5.6948,
-      "step": 28802
-    },
-    {
-      "epoch": 15.021121251629726,
-      "grad_norm": 1.6065609455108643,
-      "learning_rate": 7.156381909547739e-05,
-      "loss": 5.394,
-      "step": 28803
-    },
-    {
-      "epoch": 15.021642764015645,
-      "grad_norm": 1.4424705505371094,
-      "learning_rate": 7.156281407035175e-05,
-      "loss": 5.8872,
-      "step": 28804
-    },
-    {
-      "epoch": 15.022164276401565,
-      "grad_norm": 1.4750677347183228,
-      "learning_rate": 7.156180904522613e-05,
-      "loss": 5.7291,
-      "step": 28805
-    },
-    {
-      "epoch": 15.022685788787484,
-      "grad_norm": 1.637956142425537,
-      "learning_rate": 7.15608040201005e-05,
-      "loss": 5.4882,
-      "step": 28806
-    },
-    {
-      "epoch": 15.023207301173404,
-      "grad_norm": 1.4494270086288452,
-      "learning_rate": 7.155979899497487e-05,
-      "loss": 5.3475,
-      "step": 28807
-    },
-    {
-      "epoch": 15.023728813559321,
-      "grad_norm": 1.4985331296920776,
-      "learning_rate": 7.155879396984925e-05,
-      "loss": 5.1292,
-      "step": 28808
-    },
-    {
-      "epoch": 15.024250325945241,
-      "grad_norm": 1.4529738426208496,
-      "learning_rate": 7.155778894472363e-05,
-      "loss": 5.627,
-      "step": 28809
-    },
-    {
-      "epoch": 15.02477183833116,
-      "grad_norm": 1.441886305809021,
-      "learning_rate": 7.1556783919598e-05,
-      "loss": 5.4407,
-      "step": 28810
-    },
-    {
-      "epoch": 15.02529335071708,
-      "grad_norm": 1.5129750967025757,
-      "learning_rate": 7.155577889447237e-05,
-      "loss": 5.3869,
-      "step": 28811
-    },
-    {
-      "epoch": 15.025814863103,
-      "grad_norm": 1.476235032081604,
-      "learning_rate": 7.155477386934674e-05,
-      "loss": 5.0708,
-      "step": 28812
-    },
-    {
-      "epoch": 15.026336375488917,
-      "grad_norm": 1.5118719339370728,
-      "learning_rate": 7.155376884422111e-05,
-      "loss": 5.1149,
-      "step": 28813
-    },
-    {
-      "epoch": 15.026857887874836,
-      "grad_norm": 1.4996546506881714,
-      "learning_rate": 7.155276381909548e-05,
-      "loss": 5.6304,
-      "step": 28814
-    },
-    {
-      "epoch": 15.027379400260756,
-      "grad_norm": 1.4705839157104492,
-      "learning_rate": 7.155175879396984e-05,
-      "loss": 4.6083,
-      "step": 28815
-    },
-    {
-      "epoch": 15.027900912646675,
-      "grad_norm": 1.4936978816986084,
-      "learning_rate": 7.155075376884422e-05,
-      "loss": 5.0401,
-      "step": 28816
-    },
-    {
-      "epoch": 15.028422425032595,
-      "grad_norm": 1.8294819593429565,
-      "learning_rate": 7.154974874371859e-05,
-      "loss": 5.2383,
-      "step": 28817
-    },
-    {
-      "epoch": 15.028943937418514,
-      "grad_norm": 1.439602017402649,
-      "learning_rate": 7.154874371859296e-05,
-      "loss": 5.7824,
-      "step": 28818
-    },
-    {
-      "epoch": 15.029465449804432,
-      "grad_norm": 1.4859435558319092,
-      "learning_rate": 7.154773869346734e-05,
-      "loss": 5.2004,
-      "step": 28819
-    },
-    {
-      "epoch": 15.029986962190351,
-      "grad_norm": 1.4607715606689453,
-      "learning_rate": 7.154673366834172e-05,
-      "loss": 5.3147,
-      "step": 28820
-    },
-    {
-      "epoch": 15.030508474576271,
-      "grad_norm": 1.3796449899673462,
-      "learning_rate": 7.154572864321608e-05,
-      "loss": 4.3401,
-      "step": 28821
-    },
-    {
-      "epoch": 15.03102998696219,
-      "grad_norm": 1.4791381359100342,
-      "learning_rate": 7.154472361809046e-05,
-      "loss": 4.943,
-      "step": 28822
-    },
-    {
-      "epoch": 15.03155149934811,
-      "grad_norm": 1.479760766029358,
-      "learning_rate": 7.154371859296483e-05,
-      "loss": 4.7303,
-      "step": 28823
-    },
-    {
-      "epoch": 15.03207301173403,
-      "grad_norm": 1.5271408557891846,
-      "learning_rate": 7.15427135678392e-05,
-      "loss": 5.4615,
-      "step": 28824
-    },
-    {
-      "epoch": 15.032594524119947,
-      "grad_norm": 1.3575702905654907,
-      "learning_rate": 7.154170854271357e-05,
-      "loss": 5.6705,
-      "step": 28825
-    },
-    {
-      "epoch": 15.033116036505866,
-      "grad_norm": 1.4282563924789429,
-      "learning_rate": 7.154070351758795e-05,
-      "loss": 5.5813,
-      "step": 28826
-    },
-    {
-      "epoch": 15.033637548891786,
-      "grad_norm": 1.5819123983383179,
-      "learning_rate": 7.153969849246231e-05,
-      "loss": 5.2249,
-      "step": 28827
-    },
-    {
-      "epoch": 15.034159061277705,
-      "grad_norm": 1.490268349647522,
-      "learning_rate": 7.153869346733669e-05,
-      "loss": 5.1746,
-      "step": 28828
-    },
-    {
-      "epoch": 15.034680573663625,
-      "grad_norm": 1.4435425996780396,
-      "learning_rate": 7.153768844221107e-05,
-      "loss": 5.2465,
-      "step": 28829
-    },
-    {
-      "epoch": 15.035202086049544,
-      "grad_norm": 1.497698426246643,
-      "learning_rate": 7.153668341708543e-05,
-      "loss": 5.5486,
-      "step": 28830
-    },
-    {
-      "epoch": 15.035723598435462,
-      "grad_norm": 1.5131421089172363,
-      "learning_rate": 7.153567839195981e-05,
-      "loss": 4.6749,
-      "step": 28831
-    },
-    {
-      "epoch": 15.036245110821381,
-      "grad_norm": 1.4053012132644653,
-      "learning_rate": 7.153467336683417e-05,
-      "loss": 5.6059,
-      "step": 28832
-    },
-    {
-      "epoch": 15.036766623207301,
-      "grad_norm": 1.473907232284546,
-      "learning_rate": 7.153366834170855e-05,
-      "loss": 5.1365,
-      "step": 28833
-    },
-    {
-      "epoch": 15.03728813559322,
-      "grad_norm": 1.4321892261505127,
-      "learning_rate": 7.153266331658291e-05,
-      "loss": 5.2932,
-      "step": 28834
-    },
-    {
-      "epoch": 15.03780964797914,
-      "grad_norm": 1.546321153640747,
-      "learning_rate": 7.153165829145729e-05,
-      "loss": 5.1277,
-      "step": 28835
-    },
-    {
-      "epoch": 15.03833116036506,
-      "grad_norm": 1.4344371557235718,
-      "learning_rate": 7.153065326633166e-05,
-      "loss": 5.2466,
-      "step": 28836
-    },
-    {
-      "epoch": 15.038852672750977,
-      "grad_norm": 1.4754090309143066,
-      "learning_rate": 7.152964824120603e-05,
-      "loss": 5.4899,
-      "step": 28837
-    },
-    {
-      "epoch": 15.039374185136897,
-      "grad_norm": 1.4008880853652954,
-      "learning_rate": 7.15286432160804e-05,
-      "loss": 5.7937,
-      "step": 28838
-    },
-    {
-      "epoch": 15.039895697522816,
-      "grad_norm": 1.4668387174606323,
-      "learning_rate": 7.152763819095478e-05,
-      "loss": 5.5429,
-      "step": 28839
-    },
-    {
-      "epoch": 15.040417209908735,
-      "grad_norm": 1.5064904689788818,
-      "learning_rate": 7.152663316582915e-05,
-      "loss": 4.8388,
-      "step": 28840
-    },
-    {
-      "epoch": 15.040938722294655,
-      "grad_norm": 1.5402803421020508,
-      "learning_rate": 7.152562814070352e-05,
-      "loss": 5.095,
-      "step": 28841
-    },
-    {
-      "epoch": 15.041460234680574,
-      "grad_norm": 1.577242136001587,
-      "learning_rate": 7.15246231155779e-05,
-      "loss": 5.291,
-      "step": 28842
-    },
-    {
-      "epoch": 15.041981747066492,
-      "grad_norm": 1.3930110931396484,
-      "learning_rate": 7.152361809045226e-05,
-      "loss": 5.346,
-      "step": 28843
-    },
-    {
-      "epoch": 15.042503259452412,
-      "grad_norm": 1.4565554857254028,
-      "learning_rate": 7.152261306532664e-05,
-      "loss": 5.6947,
-      "step": 28844
-    },
-    {
-      "epoch": 15.043024771838331,
-      "grad_norm": 1.3500627279281616,
-      "learning_rate": 7.1521608040201e-05,
-      "loss": 5.4896,
-      "step": 28845
-    },
-    {
-      "epoch": 15.04354628422425,
-      "grad_norm": 1.3910415172576904,
-      "learning_rate": 7.152060301507538e-05,
-      "loss": 5.4508,
-      "step": 28846
-    },
-    {
-      "epoch": 15.04406779661017,
-      "grad_norm": 1.5313643217086792,
-      "learning_rate": 7.151959798994974e-05,
-      "loss": 5.4264,
-      "step": 28847
-    },
-    {
-      "epoch": 15.04458930899609,
-      "grad_norm": 1.455739140510559,
-      "learning_rate": 7.151859296482412e-05,
-      "loss": 5.4513,
-      "step": 28848
-    },
-    {
-      "epoch": 15.045110821382007,
-      "grad_norm": 1.384447455406189,
-      "learning_rate": 7.15175879396985e-05,
-      "loss": 5.3703,
-      "step": 28849
-    },
-    {
-      "epoch": 15.045632333767927,
-      "grad_norm": 1.5418479442596436,
-      "learning_rate": 7.151658291457288e-05,
-      "loss": 4.592,
-      "step": 28850
-    },
-    {
-      "epoch": 15.046153846153846,
-      "grad_norm": 1.5046991109848022,
-      "learning_rate": 7.151557788944724e-05,
-      "loss": 5.2092,
-      "step": 28851
-    },
-    {
-      "epoch": 15.046675358539765,
-      "grad_norm": 1.514387607574463,
-      "learning_rate": 7.151457286432162e-05,
-      "loss": 5.7226,
-      "step": 28852
-    },
-    {
-      "epoch": 15.047196870925685,
-      "grad_norm": 1.4052135944366455,
-      "learning_rate": 7.151356783919598e-05,
-      "loss": 4.3011,
-      "step": 28853
-    },
-    {
-      "epoch": 15.047718383311604,
-      "grad_norm": 1.5513349771499634,
-      "learning_rate": 7.151256281407036e-05,
-      "loss": 5.6014,
-      "step": 28854
-    },
-    {
-      "epoch": 15.048239895697522,
-      "grad_norm": 1.5160729885101318,
-      "learning_rate": 7.151155778894473e-05,
-      "loss": 4.8852,
-      "step": 28855
-    },
-    {
-      "epoch": 15.048761408083442,
-      "grad_norm": 1.4477425813674927,
-      "learning_rate": 7.151055276381909e-05,
-      "loss": 5.1647,
-      "step": 28856
-    },
-    {
-      "epoch": 15.049282920469361,
-      "grad_norm": 1.6007227897644043,
-      "learning_rate": 7.150954773869347e-05,
-      "loss": 5.4262,
-      "step": 28857
-    },
-    {
-      "epoch": 15.04980443285528,
-      "grad_norm": 1.5553786754608154,
-      "learning_rate": 7.150854271356783e-05,
-      "loss": 5.0283,
-      "step": 28858
-    },
-    {
-      "epoch": 15.0503259452412,
-      "grad_norm": 1.4909988641738892,
-      "learning_rate": 7.150753768844221e-05,
-      "loss": 5.2264,
-      "step": 28859
-    },
-    {
-      "epoch": 15.05084745762712,
-      "grad_norm": 1.4256234169006348,
-      "learning_rate": 7.150653266331659e-05,
-      "loss": 5.3761,
-      "step": 28860
-    },
-    {
-      "epoch": 15.051368970013037,
-      "grad_norm": 1.478050947189331,
-      "learning_rate": 7.150552763819097e-05,
-      "loss": 5.4356,
-      "step": 28861
-    },
-    {
-      "epoch": 15.051890482398957,
-      "grad_norm": 1.566428303718567,
-      "learning_rate": 7.150452261306533e-05,
-      "loss": 5.4581,
-      "step": 28862
-    },
-    {
-      "epoch": 15.052411994784876,
-      "grad_norm": 1.5278123617172241,
-      "learning_rate": 7.150351758793971e-05,
-      "loss": 5.5649,
-      "step": 28863
-    },
-    {
-      "epoch": 15.052933507170795,
-      "grad_norm": 1.4733365774154663,
-      "learning_rate": 7.150251256281407e-05,
-      "loss": 4.9919,
-      "step": 28864
-    },
-    {
-      "epoch": 15.053455019556715,
-      "grad_norm": 1.423715591430664,
-      "learning_rate": 7.150150753768845e-05,
-      "loss": 5.8353,
-      "step": 28865
-    },
-    {
-      "epoch": 15.053976531942634,
-      "grad_norm": 1.4385682344436646,
-      "learning_rate": 7.150050251256281e-05,
-      "loss": 5.5519,
-      "step": 28866
-    },
-    {
-      "epoch": 15.054498044328552,
-      "grad_norm": 1.4246594905853271,
-      "learning_rate": 7.149949748743719e-05,
-      "loss": 5.3503,
-      "step": 28867
-    },
-    {
-      "epoch": 15.055019556714472,
-      "grad_norm": 1.3816472291946411,
-      "learning_rate": 7.149849246231156e-05,
-      "loss": 5.5198,
-      "step": 28868
-    },
-    {
-      "epoch": 15.055541069100391,
-      "grad_norm": 1.3497544527053833,
-      "learning_rate": 7.149748743718593e-05,
-      "loss": 5.716,
-      "step": 28869
-    },
-    {
-      "epoch": 15.05606258148631,
-      "grad_norm": 1.419115424156189,
-      "learning_rate": 7.149648241206031e-05,
-      "loss": 5.6264,
-      "step": 28870
-    },
-    {
-      "epoch": 15.05658409387223,
-      "grad_norm": 1.4303467273712158,
-      "learning_rate": 7.149547738693468e-05,
-      "loss": 5.0302,
-      "step": 28871
-    },
-    {
-      "epoch": 15.05710560625815,
-      "grad_norm": 1.438164472579956,
-      "learning_rate": 7.149447236180905e-05,
-      "loss": 5.5741,
-      "step": 28872
-    },
-    {
-      "epoch": 15.057627118644067,
-      "grad_norm": 1.4476560354232788,
-      "learning_rate": 7.149346733668342e-05,
-      "loss": 5.3074,
-      "step": 28873
-    },
-    {
-      "epoch": 15.058148631029987,
-      "grad_norm": 1.5245305299758911,
-      "learning_rate": 7.14924623115578e-05,
-      "loss": 5.6523,
-      "step": 28874
-    },
-    {
-      "epoch": 15.058670143415906,
-      "grad_norm": 1.3458268642425537,
-      "learning_rate": 7.149145728643216e-05,
-      "loss": 5.2648,
-      "step": 28875
-    },
-    {
-      "epoch": 15.059191655801826,
-      "grad_norm": 1.510672926902771,
-      "learning_rate": 7.149045226130654e-05,
-      "loss": 5.6866,
-      "step": 28876
-    },
-    {
-      "epoch": 15.059713168187745,
-      "grad_norm": 1.5758062601089478,
-      "learning_rate": 7.14894472361809e-05,
-      "loss": 5.6382,
-      "step": 28877
-    },
-    {
-      "epoch": 15.060234680573664,
-      "grad_norm": 1.596310019493103,
-      "learning_rate": 7.148844221105528e-05,
-      "loss": 4.7085,
-      "step": 28878
-    },
-    {
-      "epoch": 15.060756192959582,
-      "grad_norm": 1.5607863664627075,
-      "learning_rate": 7.148743718592964e-05,
-      "loss": 4.436,
-      "step": 28879
-    },
-    {
-      "epoch": 15.061277705345502,
-      "grad_norm": 1.5862423181533813,
-      "learning_rate": 7.148643216080402e-05,
-      "loss": 5.407,
-      "step": 28880
-    },
-    {
-      "epoch": 15.061799217731421,
-      "grad_norm": 1.4584648609161377,
-      "learning_rate": 7.14854271356784e-05,
-      "loss": 5.1606,
-      "step": 28881
-    },
-    {
-      "epoch": 15.06232073011734,
-      "grad_norm": 1.5613422393798828,
-      "learning_rate": 7.148442211055276e-05,
-      "loss": 4.7233,
-      "step": 28882
-    },
-    {
-      "epoch": 15.06284224250326,
-      "grad_norm": 1.5399249792099,
-      "learning_rate": 7.148341708542714e-05,
-      "loss": 5.2613,
-      "step": 28883
-    },
-    {
-      "epoch": 15.06336375488918,
-      "grad_norm": 1.5518499612808228,
-      "learning_rate": 7.148241206030151e-05,
-      "loss": 5.4214,
-      "step": 28884
-    },
-    {
-      "epoch": 15.063885267275097,
-      "grad_norm": 1.588444471359253,
-      "learning_rate": 7.148140703517588e-05,
-      "loss": 5.1463,
-      "step": 28885
-    },
-    {
-      "epoch": 15.064406779661017,
-      "grad_norm": 1.5751436948776245,
-      "learning_rate": 7.148040201005025e-05,
-      "loss": 5.2571,
-      "step": 28886
-    },
-    {
-      "epoch": 15.064928292046936,
-      "grad_norm": 1.5078915357589722,
-      "learning_rate": 7.147939698492463e-05,
-      "loss": 5.4705,
-      "step": 28887
-    },
-    {
-      "epoch": 15.065449804432856,
-      "grad_norm": 1.4844189882278442,
-      "learning_rate": 7.147839195979899e-05,
-      "loss": 5.3055,
-      "step": 28888
-    },
-    {
-      "epoch": 15.065971316818775,
-      "grad_norm": 1.4628113508224487,
-      "learning_rate": 7.147738693467337e-05,
-      "loss": 5.5058,
-      "step": 28889
-    },
-    {
-      "epoch": 15.066492829204694,
-      "grad_norm": 1.3131061792373657,
-      "learning_rate": 7.147638190954775e-05,
-      "loss": 5.2995,
-      "step": 28890
-    },
-    {
-      "epoch": 15.067014341590612,
-      "grad_norm": 1.406394362449646,
-      "learning_rate": 7.147537688442212e-05,
-      "loss": 5.8129,
-      "step": 28891
-    },
-    {
-      "epoch": 15.067535853976532,
-      "grad_norm": 1.4251024723052979,
-      "learning_rate": 7.147437185929649e-05,
-      "loss": 5.4811,
-      "step": 28892
-    },
-    {
-      "epoch": 15.068057366362451,
-      "grad_norm": 1.5555423498153687,
-      "learning_rate": 7.147336683417087e-05,
-      "loss": 5.3117,
-      "step": 28893
-    },
-    {
-      "epoch": 15.06857887874837,
-      "grad_norm": 1.4877212047576904,
-      "learning_rate": 7.147236180904523e-05,
-      "loss": 5.0694,
-      "step": 28894
-    },
-    {
-      "epoch": 15.06910039113429,
-      "grad_norm": 1.4975770711898804,
-      "learning_rate": 7.14713567839196e-05,
-      "loss": 5.1578,
-      "step": 28895
-    },
-    {
-      "epoch": 15.06962190352021,
-      "grad_norm": 1.5107280015945435,
-      "learning_rate": 7.147035175879397e-05,
-      "loss": 5.4219,
-      "step": 28896
-    },
-    {
-      "epoch": 15.070143415906127,
-      "grad_norm": 1.425217628479004,
-      "learning_rate": 7.146934673366834e-05,
-      "loss": 5.6853,
-      "step": 28897
-    },
-    {
-      "epoch": 15.070664928292047,
-      "grad_norm": 1.4032572507858276,
-      "learning_rate": 7.146834170854272e-05,
-      "loss": 5.4431,
-      "step": 28898
-    },
-    {
-      "epoch": 15.071186440677966,
-      "grad_norm": 1.5202369689941406,
-      "learning_rate": 7.146733668341708e-05,
-      "loss": 5.2559,
-      "step": 28899
-    },
-    {
-      "epoch": 15.071707953063886,
-      "grad_norm": 1.4753310680389404,
-      "learning_rate": 7.146633165829146e-05,
-      "loss": 4.696,
-      "step": 28900
-    },
-    {
-      "epoch": 15.072229465449805,
-      "grad_norm": 1.5914028882980347,
-      "learning_rate": 7.146532663316583e-05,
-      "loss": 5.525,
-      "step": 28901
-    },
-    {
-      "epoch": 15.072750977835724,
-      "grad_norm": 1.5695390701293945,
-      "learning_rate": 7.146432160804021e-05,
-      "loss": 5.0267,
-      "step": 28902
-    },
-    {
-      "epoch": 15.073272490221642,
-      "grad_norm": 1.4306888580322266,
-      "learning_rate": 7.146331658291458e-05,
-      "loss": 5.5274,
-      "step": 28903
-    },
-    {
-      "epoch": 15.073794002607562,
-      "grad_norm": 1.4211045503616333,
-      "learning_rate": 7.146231155778895e-05,
-      "loss": 5.8594,
-      "step": 28904
-    },
-    {
-      "epoch": 15.074315514993481,
-      "grad_norm": 1.335365891456604,
-      "learning_rate": 7.146130653266332e-05,
-      "loss": 5.0985,
-      "step": 28905
-    },
-    {
-      "epoch": 15.0748370273794,
-      "grad_norm": 1.5315121412277222,
-      "learning_rate": 7.14603015075377e-05,
-      "loss": 5.2639,
-      "step": 28906
-    },
-    {
-      "epoch": 15.07535853976532,
-      "grad_norm": 1.4786714315414429,
-      "learning_rate": 7.145929648241206e-05,
-      "loss": 5.6138,
-      "step": 28907
-    },
-    {
-      "epoch": 15.075880052151238,
-      "grad_norm": 1.4788589477539062,
-      "learning_rate": 7.145829145728643e-05,
-      "loss": 5.49,
-      "step": 28908
-    },
-    {
-      "epoch": 15.076401564537157,
-      "grad_norm": 1.410479187965393,
-      "learning_rate": 7.14572864321608e-05,
-      "loss": 5.4614,
-      "step": 28909
-    },
-    {
-      "epoch": 15.076923076923077,
-      "grad_norm": 1.6251330375671387,
-      "learning_rate": 7.145628140703518e-05,
-      "loss": 4.6833,
-      "step": 28910
-    },
-    {
-      "epoch": 15.077444589308996,
-      "grad_norm": 1.5600355863571167,
-      "learning_rate": 7.145527638190956e-05,
-      "loss": 5.316,
-      "step": 28911
-    },
-    {
-      "epoch": 15.077966101694916,
-      "grad_norm": 1.4925483465194702,
-      "learning_rate": 7.145427135678392e-05,
-      "loss": 5.4552,
-      "step": 28912
-    },
-    {
-      "epoch": 15.078487614080835,
-      "grad_norm": 1.5392584800720215,
-      "learning_rate": 7.14532663316583e-05,
-      "loss": 5.2715,
-      "step": 28913
-    },
-    {
-      "epoch": 15.079009126466753,
-      "grad_norm": 1.5024949312210083,
-      "learning_rate": 7.145226130653267e-05,
-      "loss": 5.5497,
-      "step": 28914
-    },
-    {
-      "epoch": 15.079530638852672,
-      "grad_norm": 1.3707231283187866,
-      "learning_rate": 7.145125628140704e-05,
-      "loss": 5.5869,
-      "step": 28915
-    },
-    {
-      "epoch": 15.080052151238592,
-      "grad_norm": 1.5117924213409424,
-      "learning_rate": 7.145025125628141e-05,
-      "loss": 5.3891,
-      "step": 28916
-    },
-    {
-      "epoch": 15.080573663624511,
-      "grad_norm": 1.5013914108276367,
-      "learning_rate": 7.144924623115579e-05,
-      "loss": 5.6617,
-      "step": 28917
-    },
-    {
-      "epoch": 15.08109517601043,
-      "grad_norm": 1.5045989751815796,
-      "learning_rate": 7.144824120603015e-05,
-      "loss": 4.8163,
-      "step": 28918
-    },
-    {
-      "epoch": 15.08161668839635,
-      "grad_norm": 1.4401882886886597,
-      "learning_rate": 7.144723618090453e-05,
-      "loss": 5.0921,
-      "step": 28919
-    },
-    {
-      "epoch": 15.082138200782268,
-      "grad_norm": 1.478485345840454,
-      "learning_rate": 7.144623115577889e-05,
-      "loss": 5.677,
-      "step": 28920
-    },
-    {
-      "epoch": 15.082659713168187,
-      "grad_norm": 1.6924904584884644,
-      "learning_rate": 7.144522613065327e-05,
-      "loss": 4.5949,
-      "step": 28921
-    },
-    {
-      "epoch": 15.083181225554107,
-      "grad_norm": 1.490372896194458,
-      "learning_rate": 7.144422110552765e-05,
-      "loss": 5.4182,
-      "step": 28922
-    },
-    {
-      "epoch": 15.083702737940026,
-      "grad_norm": 1.3259464502334595,
-      "learning_rate": 7.144321608040201e-05,
-      "loss": 5.6872,
-      "step": 28923
-    },
-    {
-      "epoch": 15.084224250325946,
-      "grad_norm": 1.435198426246643,
-      "learning_rate": 7.144221105527639e-05,
-      "loss": 5.3351,
-      "step": 28924
-    },
-    {
-      "epoch": 15.084745762711865,
-      "grad_norm": 1.6211390495300293,
-      "learning_rate": 7.144120603015075e-05,
-      "loss": 5.2755,
-      "step": 28925
-    },
-    {
-      "epoch": 15.085267275097783,
-      "grad_norm": 1.499839425086975,
-      "learning_rate": 7.144020100502513e-05,
-      "loss": 5.3371,
-      "step": 28926
-    },
-    {
-      "epoch": 15.085788787483702,
-      "grad_norm": 1.3940523862838745,
-      "learning_rate": 7.14391959798995e-05,
-      "loss": 5.6963,
-      "step": 28927
-    },
-    {
-      "epoch": 15.086310299869622,
-      "grad_norm": 1.5216419696807861,
-      "learning_rate": 7.143819095477387e-05,
-      "loss": 5.1705,
-      "step": 28928
-    },
-    {
-      "epoch": 15.086831812255541,
-      "grad_norm": 1.40998375415802,
-      "learning_rate": 7.143718592964824e-05,
-      "loss": 5.6686,
-      "step": 28929
-    },
-    {
-      "epoch": 15.08735332464146,
-      "grad_norm": 1.4843734502792358,
-      "learning_rate": 7.143618090452262e-05,
-      "loss": 5.4397,
-      "step": 28930
-    },
-    {
-      "epoch": 15.08787483702738,
-      "grad_norm": 1.4487470388412476,
-      "learning_rate": 7.1435175879397e-05,
-      "loss": 5.6565,
-      "step": 28931
-    },
-    {
-      "epoch": 15.088396349413298,
-      "grad_norm": 1.4440490007400513,
-      "learning_rate": 7.143417085427137e-05,
-      "loss": 5.6362,
-      "step": 28932
-    },
-    {
-      "epoch": 15.088917861799217,
-      "grad_norm": 1.3741968870162964,
-      "learning_rate": 7.143316582914574e-05,
-      "loss": 5.7426,
-      "step": 28933
-    },
-    {
-      "epoch": 15.089439374185137,
-      "grad_norm": 1.3843262195587158,
-      "learning_rate": 7.14321608040201e-05,
-      "loss": 5.709,
-      "step": 28934
-    },
-    {
-      "epoch": 15.089960886571056,
-      "grad_norm": 1.4738620519638062,
-      "learning_rate": 7.143115577889448e-05,
-      "loss": 5.4669,
-      "step": 28935
-    },
-    {
-      "epoch": 15.090482398956976,
-      "grad_norm": 1.555630087852478,
-      "learning_rate": 7.143015075376884e-05,
-      "loss": 5.4933,
-      "step": 28936
-    },
-    {
-      "epoch": 15.091003911342895,
-      "grad_norm": 1.5510210990905762,
-      "learning_rate": 7.142914572864322e-05,
-      "loss": 5.1984,
-      "step": 28937
-    },
-    {
-      "epoch": 15.091525423728813,
-      "grad_norm": 1.4729846715927124,
-      "learning_rate": 7.142814070351758e-05,
-      "loss": 4.7357,
-      "step": 28938
-    },
-    {
-      "epoch": 15.092046936114732,
-      "grad_norm": 1.746331810951233,
-      "learning_rate": 7.142713567839196e-05,
-      "loss": 4.6816,
-      "step": 28939
-    },
-    {
-      "epoch": 15.092568448500652,
-      "grad_norm": 1.5910457372665405,
-      "learning_rate": 7.142613065326633e-05,
-      "loss": 5.1432,
-      "step": 28940
-    },
-    {
-      "epoch": 15.093089960886571,
-      "grad_norm": 1.5701767206192017,
-      "learning_rate": 7.14251256281407e-05,
-      "loss": 5.4806,
-      "step": 28941
-    },
-    {
-      "epoch": 15.09361147327249,
-      "grad_norm": 1.4326491355895996,
-      "learning_rate": 7.142412060301508e-05,
-      "loss": 5.2868,
-      "step": 28942
-    },
-    {
-      "epoch": 15.09413298565841,
-      "grad_norm": 1.5629723072052002,
-      "learning_rate": 7.142311557788946e-05,
-      "loss": 4.6854,
-      "step": 28943
-    },
-    {
-      "epoch": 15.094654498044328,
-      "grad_norm": 1.415118932723999,
-      "learning_rate": 7.142211055276382e-05,
-      "loss": 5.3194,
-      "step": 28944
-    },
-    {
-      "epoch": 15.095176010430247,
-      "grad_norm": 1.4162962436676025,
-      "learning_rate": 7.14211055276382e-05,
-      "loss": 5.3412,
-      "step": 28945
-    },
-    {
-      "epoch": 15.095697522816167,
-      "grad_norm": 1.4055882692337036,
-      "learning_rate": 7.142010050251257e-05,
-      "loss": 5.3284,
-      "step": 28946
-    },
-    {
-      "epoch": 15.096219035202086,
-      "grad_norm": 1.56875741481781,
-      "learning_rate": 7.141909547738694e-05,
-      "loss": 5.1753,
-      "step": 28947
-    },
-    {
-      "epoch": 15.096740547588006,
-      "grad_norm": 1.3937442302703857,
-      "learning_rate": 7.141809045226131e-05,
-      "loss": 5.4329,
-      "step": 28948
-    },
-    {
-      "epoch": 15.097262059973925,
-      "grad_norm": 1.4763462543487549,
-      "learning_rate": 7.141708542713567e-05,
-      "loss": 5.3916,
-      "step": 28949
-    },
-    {
-      "epoch": 15.097783572359843,
-      "grad_norm": 1.4998615980148315,
-      "learning_rate": 7.141608040201005e-05,
-      "loss": 5.3771,
-      "step": 28950
-    },
-    {
-      "epoch": 15.098305084745762,
-      "grad_norm": 1.408146619796753,
-      "learning_rate": 7.141507537688443e-05,
-      "loss": 5.2073,
-      "step": 28951
-    },
-    {
-      "epoch": 15.098826597131682,
-      "grad_norm": 1.494038462638855,
-      "learning_rate": 7.14140703517588e-05,
-      "loss": 5.5704,
-      "step": 28952
-    },
-    {
-      "epoch": 15.099348109517601,
-      "grad_norm": 1.5848718881607056,
-      "learning_rate": 7.141306532663317e-05,
-      "loss": 5.1567,
-      "step": 28953
-    },
-    {
-      "epoch": 15.09986962190352,
-      "grad_norm": 1.5129927396774292,
-      "learning_rate": 7.141206030150755e-05,
-      "loss": 5.4619,
-      "step": 28954
-    },
-    {
-      "epoch": 15.10039113428944,
-      "grad_norm": 1.4144443273544312,
-      "learning_rate": 7.141105527638191e-05,
-      "loss": 5.2435,
-      "step": 28955
-    },
-    {
-      "epoch": 15.100912646675358,
-      "grad_norm": 1.4986590147018433,
-      "learning_rate": 7.141005025125629e-05,
-      "loss": 5.4601,
-      "step": 28956
-    },
-    {
-      "epoch": 15.101434159061277,
-      "grad_norm": 1.4569087028503418,
-      "learning_rate": 7.140904522613065e-05,
-      "loss": 5.5611,
-      "step": 28957
-    },
-    {
-      "epoch": 15.101955671447197,
-      "grad_norm": 1.5059967041015625,
-      "learning_rate": 7.140804020100503e-05,
-      "loss": 5.8556,
-      "step": 28958
-    },
-    {
-      "epoch": 15.102477183833116,
-      "grad_norm": 1.4549798965454102,
-      "learning_rate": 7.14070351758794e-05,
-      "loss": 5.6016,
-      "step": 28959
-    },
-    {
-      "epoch": 15.102998696219036,
-      "grad_norm": 1.4329584836959839,
-      "learning_rate": 7.140603015075377e-05,
-      "loss": 5.3923,
-      "step": 28960
-    },
-    {
-      "epoch": 15.103520208604955,
-      "grad_norm": 1.3929848670959473,
-      "learning_rate": 7.140502512562814e-05,
-      "loss": 5.5664,
-      "step": 28961
-    },
-    {
-      "epoch": 15.104041720990873,
-      "grad_norm": 1.49989652633667,
-      "learning_rate": 7.140402010050252e-05,
-      "loss": 5.2271,
-      "step": 28962
-    },
-    {
-      "epoch": 15.104563233376792,
-      "grad_norm": 1.5136001110076904,
-      "learning_rate": 7.14030150753769e-05,
-      "loss": 5.2107,
-      "step": 28963
-    },
-    {
-      "epoch": 15.105084745762712,
-      "grad_norm": 1.9796257019042969,
-      "learning_rate": 7.140201005025126e-05,
-      "loss": 5.0144,
-      "step": 28964
-    },
-    {
-      "epoch": 15.105606258148631,
-      "grad_norm": 1.4628428220748901,
-      "learning_rate": 7.140100502512564e-05,
-      "loss": 5.0226,
-      "step": 28965
-    },
-    {
-      "epoch": 15.10612777053455,
-      "grad_norm": 1.341151237487793,
-      "learning_rate": 7.14e-05,
-      "loss": 5.3666,
-      "step": 28966
-    },
-    {
-      "epoch": 15.10664928292047,
-      "grad_norm": 1.4637469053268433,
-      "learning_rate": 7.139899497487438e-05,
-      "loss": 5.2468,
-      "step": 28967
-    },
-    {
-      "epoch": 15.107170795306388,
-      "grad_norm": 1.6683974266052246,
-      "learning_rate": 7.139798994974874e-05,
-      "loss": 4.8547,
-      "step": 28968
-    },
-    {
-      "epoch": 15.107692307692307,
-      "grad_norm": 1.6331274509429932,
-      "learning_rate": 7.139698492462312e-05,
-      "loss": 5.1775,
-      "step": 28969
-    },
-    {
-      "epoch": 15.108213820078227,
-      "grad_norm": 1.548012614250183,
-      "learning_rate": 7.139597989949748e-05,
-      "loss": 5.4758,
-      "step": 28970
-    },
-    {
-      "epoch": 15.108735332464146,
-      "grad_norm": 1.5210987329483032,
-      "learning_rate": 7.139497487437186e-05,
-      "loss": 5.0987,
-      "step": 28971
-    },
-    {
-      "epoch": 15.109256844850066,
-      "grad_norm": 1.4844462871551514,
-      "learning_rate": 7.139396984924624e-05,
-      "loss": 5.3036,
-      "step": 28972
-    },
-    {
-      "epoch": 15.109778357235985,
-      "grad_norm": 1.3759374618530273,
-      "learning_rate": 7.139296482412062e-05,
-      "loss": 5.6432,
-      "step": 28973
-    },
-    {
-      "epoch": 15.110299869621903,
-      "grad_norm": 1.4730433225631714,
-      "learning_rate": 7.139195979899498e-05,
-      "loss": 5.4567,
-      "step": 28974
-    },
-    {
-      "epoch": 15.110821382007822,
-      "grad_norm": 1.5208078622817993,
-      "learning_rate": 7.139095477386935e-05,
-      "loss": 5.554,
-      "step": 28975
-    },
-    {
-      "epoch": 15.111342894393742,
-      "grad_norm": 1.529284119606018,
-      "learning_rate": 7.138994974874372e-05,
-      "loss": 5.4257,
-      "step": 28976
-    },
-    {
-      "epoch": 15.111864406779661,
-      "grad_norm": 1.5474213361740112,
-      "learning_rate": 7.138894472361809e-05,
-      "loss": 5.2039,
-      "step": 28977
-    },
-    {
-      "epoch": 15.11238591916558,
-      "grad_norm": 1.4905202388763428,
-      "learning_rate": 7.138793969849247e-05,
-      "loss": 5.5938,
-      "step": 28978
-    },
-    {
-      "epoch": 15.1129074315515,
-      "grad_norm": 1.3677691221237183,
-      "learning_rate": 7.138693467336683e-05,
-      "loss": 5.6089,
-      "step": 28979
-    },
-    {
-      "epoch": 15.113428943937418,
-      "grad_norm": 1.4773911237716675,
-      "learning_rate": 7.138592964824121e-05,
-      "loss": 5.4169,
-      "step": 28980
-    },
-    {
-      "epoch": 15.113950456323337,
-      "grad_norm": 1.4439160823822021,
-      "learning_rate": 7.138492462311557e-05,
-      "loss": 5.5175,
-      "step": 28981
-    },
-    {
-      "epoch": 15.114471968709257,
-      "grad_norm": 1.3880728483200073,
-      "learning_rate": 7.138391959798995e-05,
-      "loss": 5.6056,
-      "step": 28982
-    },
-    {
-      "epoch": 15.114993481095176,
-      "grad_norm": 1.3507097959518433,
-      "learning_rate": 7.138291457286433e-05,
-      "loss": 5.5776,
-      "step": 28983
-    },
-    {
-      "epoch": 15.115514993481096,
-      "grad_norm": 1.4659535884857178,
-      "learning_rate": 7.138190954773871e-05,
-      "loss": 5.0224,
-      "step": 28984
-    },
-    {
-      "epoch": 15.116036505867015,
-      "grad_norm": 1.4414665699005127,
-      "learning_rate": 7.138090452261307e-05,
-      "loss": 5.1551,
-      "step": 28985
-    },
-    {
-      "epoch": 15.116558018252933,
-      "grad_norm": 1.5073434114456177,
-      "learning_rate": 7.137989949748745e-05,
-      "loss": 5.3196,
-      "step": 28986
-    },
-    {
-      "epoch": 15.117079530638852,
-      "grad_norm": 1.4076799154281616,
-      "learning_rate": 7.137889447236181e-05,
-      "loss": 5.376,
-      "step": 28987
-    },
-    {
-      "epoch": 15.117601043024772,
-      "grad_norm": Infinity,
-      "learning_rate": 7.137889447236181e-05,
-      "loss": 4.6917,
-      "step": 28988
-    },
-    {
-      "epoch": 15.118122555410691,
-      "grad_norm": 1.4623253345489502,
-      "learning_rate": 7.137788944723618e-05,
-      "loss": 5.6143,
-      "step": 28989
-    },
-    {
-      "epoch": 15.11864406779661,
-      "grad_norm": 1.5261727571487427,
-      "learning_rate": 7.137688442211056e-05,
-      "loss": 5.4663,
-      "step": 28990
-    },
-    {
-      "epoch": 15.11916558018253,
-      "grad_norm": 1.4280831813812256,
-      "learning_rate": 7.137587939698492e-05,
-      "loss": 5.1533,
-      "step": 28991
-    },
-    {
-      "epoch": 15.119687092568448,
-      "grad_norm": 1.4373342990875244,
-      "learning_rate": 7.13748743718593e-05,
-      "loss": 5.2576,
-      "step": 28992
-    },
-    {
-      "epoch": 15.120208604954367,
-      "grad_norm": 1.5543891191482544,
-      "learning_rate": 7.137386934673366e-05,
-      "loss": 5.3232,
-      "step": 28993
-    },
-    {
-      "epoch": 15.120730117340287,
-      "grad_norm": 1.4455937147140503,
-      "learning_rate": 7.137286432160804e-05,
-      "loss": 5.238,
-      "step": 28994
-    },
-    {
-      "epoch": 15.121251629726206,
-      "grad_norm": 1.4428068399429321,
-      "learning_rate": 7.137185929648242e-05,
-      "loss": 5.4613,
-      "step": 28995
-    },
-    {
-      "epoch": 15.121773142112126,
-      "grad_norm": 1.524838924407959,
-      "learning_rate": 7.13708542713568e-05,
-      "loss": 4.9826,
-      "step": 28996
-    },
-    {
-      "epoch": 15.122294654498045,
-      "grad_norm": 1.4304856061935425,
-      "learning_rate": 7.136984924623116e-05,
-      "loss": 5.7034,
-      "step": 28997
-    },
-    {
-      "epoch": 15.122816166883963,
-      "grad_norm": 1.4452167749404907,
-      "learning_rate": 7.136884422110554e-05,
-      "loss": 5.433,
-      "step": 28998
-    },
-    {
-      "epoch": 15.123337679269882,
-      "grad_norm": 1.342434287071228,
-      "learning_rate": 7.13678391959799e-05,
-      "loss": 5.6212,
-      "step": 28999
-    },
-    {
-      "epoch": 15.123859191655802,
-      "grad_norm": 1.4309464693069458,
-      "learning_rate": 7.136683417085428e-05,
-      "loss": 5.5933,
-      "step": 29000
-    },
-    {
-      "epoch": 15.124380704041721,
-      "grad_norm": 1.4471975564956665,
-      "learning_rate": 7.136582914572864e-05,
-      "loss": 5.2764,
-      "step": 29001
-    },
-    {
-      "epoch": 15.12490221642764,
-      "grad_norm": 1.5104763507843018,
-      "learning_rate": 7.136482412060301e-05,
-      "loss": 5.0428,
-      "step": 29002
-    },
-    {
-      "epoch": 15.125423728813558,
-      "grad_norm": 1.5696898698806763,
-      "learning_rate": 7.136381909547739e-05,
-      "loss": 5.3143,
-      "step": 29003
-    },
-    {
-      "epoch": 15.125945241199478,
-      "grad_norm": 1.4075406789779663,
-      "learning_rate": 7.136281407035176e-05,
-      "loss": 5.6435,
-      "step": 29004
-    },
-    {
-      "epoch": 15.126466753585397,
-      "grad_norm": 1.435833215713501,
-      "learning_rate": 7.136180904522614e-05,
-      "loss": 5.5908,
-      "step": 29005
-    },
-    {
-      "epoch": 15.126988265971317,
-      "grad_norm": 1.3716201782226562,
-      "learning_rate": 7.13608040201005e-05,
-      "loss": 5.4821,
-      "step": 29006
-    },
-    {
-      "epoch": 15.127509778357236,
-      "grad_norm": 1.4538233280181885,
-      "learning_rate": 7.135979899497488e-05,
-      "loss": 5.6437,
-      "step": 29007
-    },
-    {
-      "epoch": 15.128031290743156,
-      "grad_norm": 1.539462924003601,
-      "learning_rate": 7.135879396984925e-05,
-      "loss": 5.4521,
-      "step": 29008
-    },
-    {
-      "epoch": 15.128552803129073,
-      "grad_norm": 1.6915168762207031,
-      "learning_rate": 7.135778894472363e-05,
-      "loss": 4.924,
-      "step": 29009
-    },
-    {
-      "epoch": 15.129074315514993,
-      "grad_norm": 1.4632387161254883,
-      "learning_rate": 7.135678391959799e-05,
-      "loss": 5.5095,
-      "step": 29010
-    },
-    {
-      "epoch": 15.129595827900912,
-      "grad_norm": 1.4300282001495361,
-      "learning_rate": 7.135577889447237e-05,
-      "loss": 5.4327,
-      "step": 29011
-    },
-    {
-      "epoch": 15.130117340286832,
-      "grad_norm": 1.4685933589935303,
-      "learning_rate": 7.135477386934673e-05,
-      "loss": 5.4492,
-      "step": 29012
-    },
-    {
-      "epoch": 15.130638852672751,
-      "grad_norm": 1.4691903591156006,
-      "learning_rate": 7.135376884422111e-05,
-      "loss": 5.4468,
-      "step": 29013
-    },
-    {
-      "epoch": 15.13116036505867,
-      "grad_norm": 1.4280394315719604,
-      "learning_rate": 7.135276381909547e-05,
-      "loss": 5.1211,
-      "step": 29014
-    },
-    {
-      "epoch": 15.131681877444588,
-      "grad_norm": 1.4426642656326294,
-      "learning_rate": 7.135175879396985e-05,
-      "loss": 5.3176,
-      "step": 29015
-    },
-    {
-      "epoch": 15.132203389830508,
-      "grad_norm": 1.4958908557891846,
-      "learning_rate": 7.135075376884423e-05,
-      "loss": 4.9278,
-      "step": 29016
-    },
-    {
-      "epoch": 15.132724902216427,
-      "grad_norm": 1.5318886041641235,
-      "learning_rate": 7.13497487437186e-05,
-      "loss": 5.2839,
-      "step": 29017
-    },
-    {
-      "epoch": 15.133246414602347,
-      "grad_norm": 1.3786710500717163,
-      "learning_rate": 7.134874371859297e-05,
-      "loss": 5.5724,
-      "step": 29018
-    },
-    {
-      "epoch": 15.133767926988266,
-      "grad_norm": 1.4286922216415405,
-      "learning_rate": 7.134773869346734e-05,
-      "loss": 5.3679,
-      "step": 29019
-    },
-    {
-      "epoch": 15.134289439374186,
-      "grad_norm": 1.4419300556182861,
-      "learning_rate": 7.134673366834171e-05,
-      "loss": 5.7136,
-      "step": 29020
-    },
-    {
-      "epoch": 15.134810951760103,
-      "grad_norm": 1.4136335849761963,
-      "learning_rate": 7.134572864321608e-05,
-      "loss": 5.5669,
-      "step": 29021
-    },
-    {
-      "epoch": 15.135332464146023,
-      "grad_norm": 1.4925001859664917,
-      "learning_rate": 7.134472361809046e-05,
-      "loss": 5.6006,
-      "step": 29022
-    },
-    {
-      "epoch": 15.135853976531942,
-      "grad_norm": 1.4636709690093994,
-      "learning_rate": 7.134371859296482e-05,
-      "loss": 5.6432,
-      "step": 29023
-    },
-    {
-      "epoch": 15.136375488917862,
-      "grad_norm": 1.4070241451263428,
-      "learning_rate": 7.13427135678392e-05,
-      "loss": 5.5244,
-      "step": 29024
-    },
-    {
-      "epoch": 15.136897001303781,
-      "grad_norm": 1.5891765356063843,
-      "learning_rate": 7.134170854271358e-05,
-      "loss": 4.7607,
-      "step": 29025
-    },
-    {
-      "epoch": 15.1374185136897,
-      "grad_norm": 1.458716869354248,
-      "learning_rate": 7.134070351758795e-05,
-      "loss": 5.5585,
-      "step": 29026
-    },
-    {
-      "epoch": 15.137940026075619,
-      "grad_norm": 1.3933122158050537,
-      "learning_rate": 7.133969849246232e-05,
-      "loss": 5.5731,
-      "step": 29027
-    },
-    {
-      "epoch": 15.138461538461538,
-      "grad_norm": 1.391535758972168,
-      "learning_rate": 7.133869346733668e-05,
-      "loss": 5.5988,
-      "step": 29028
-    },
-    {
-      "epoch": 15.138983050847457,
-      "grad_norm": 1.3724972009658813,
-      "learning_rate": 7.133768844221106e-05,
-      "loss": 5.1434,
-      "step": 29029
-    },
-    {
-      "epoch": 15.139504563233377,
-      "grad_norm": 1.3958393335342407,
-      "learning_rate": 7.133668341708542e-05,
-      "loss": 5.6245,
-      "step": 29030
-    },
-    {
-      "epoch": 15.140026075619296,
-      "grad_norm": 1.3763427734375,
-      "learning_rate": 7.13356783919598e-05,
-      "loss": 5.5614,
-      "step": 29031
-    },
-    {
-      "epoch": 15.140547588005216,
-      "grad_norm": 1.397996187210083,
-      "learning_rate": 7.133467336683417e-05,
-      "loss": 5.3803,
-      "step": 29032
-    },
-    {
-      "epoch": 15.141069100391134,
-      "grad_norm": 1.4286565780639648,
-      "learning_rate": 7.133366834170854e-05,
-      "loss": 5.4641,
-      "step": 29033
-    },
-    {
-      "epoch": 15.141590612777053,
-      "grad_norm": 1.4353885650634766,
-      "learning_rate": 7.133266331658291e-05,
-      "loss": 5.6271,
-      "step": 29034
-    },
-    {
-      "epoch": 15.142112125162972,
-      "grad_norm": 1.5728447437286377,
-      "learning_rate": 7.133165829145729e-05,
-      "loss": 5.1306,
-      "step": 29035
-    },
-    {
-      "epoch": 15.142633637548892,
-      "grad_norm": 1.4712942838668823,
-      "learning_rate": 7.133065326633166e-05,
-      "loss": 5.4033,
-      "step": 29036
-    },
-    {
-      "epoch": 15.143155149934811,
-      "grad_norm": 1.3912672996520996,
-      "learning_rate": 7.132964824120604e-05,
-      "loss": 5.4614,
-      "step": 29037
-    },
-    {
-      "epoch": 15.14367666232073,
-      "grad_norm": 1.4003973007202148,
-      "learning_rate": 7.13286432160804e-05,
-      "loss": 5.5948,
-      "step": 29038
-    },
-    {
-      "epoch": 15.144198174706649,
-      "grad_norm": 1.4671008586883545,
-      "learning_rate": 7.132763819095478e-05,
-      "loss": 4.8619,
-      "step": 29039
-    },
-    {
-      "epoch": 15.144719687092568,
-      "grad_norm": 1.4142231941223145,
-      "learning_rate": 7.132663316582915e-05,
-      "loss": 5.4722,
-      "step": 29040
-    },
-    {
-      "epoch": 15.145241199478487,
-      "grad_norm": 1.561867356300354,
-      "learning_rate": 7.132562814070353e-05,
-      "loss": 4.9734,
-      "step": 29041
-    },
-    {
-      "epoch": 15.145762711864407,
-      "grad_norm": 1.426042914390564,
-      "learning_rate": 7.132462311557789e-05,
-      "loss": 5.7915,
-      "step": 29042
-    },
-    {
-      "epoch": 15.146284224250326,
-      "grad_norm": 1.3952854871749878,
-      "learning_rate": 7.132361809045225e-05,
-      "loss": 5.441,
-      "step": 29043
-    },
-    {
-      "epoch": 15.146805736636246,
-      "grad_norm": 1.4686346054077148,
-      "learning_rate": 7.132261306532663e-05,
-      "loss": 4.9324,
-      "step": 29044
-    },
-    {
-      "epoch": 15.147327249022164,
-      "grad_norm": 1.4551379680633545,
-      "learning_rate": 7.132160804020101e-05,
-      "loss": 5.6299,
-      "step": 29045
-    },
-    {
-      "epoch": 15.147848761408083,
-      "grad_norm": 1.4839674234390259,
-      "learning_rate": 7.132060301507539e-05,
-      "loss": 5.5071,
-      "step": 29046
-    },
-    {
-      "epoch": 15.148370273794002,
-      "grad_norm": 1.4997496604919434,
-      "learning_rate": 7.131959798994975e-05,
-      "loss": 5.8538,
-      "step": 29047
-    },
-    {
-      "epoch": 15.148891786179922,
-      "grad_norm": 1.4840813875198364,
-      "learning_rate": 7.131859296482413e-05,
-      "loss": 4.9332,
-      "step": 29048
-    },
-    {
-      "epoch": 15.149413298565841,
-      "grad_norm": 1.3952629566192627,
-      "learning_rate": 7.13175879396985e-05,
-      "loss": 5.512,
-      "step": 29049
-    },
-    {
-      "epoch": 15.14993481095176,
-      "grad_norm": 1.3677921295166016,
-      "learning_rate": 7.131658291457287e-05,
-      "loss": 5.0031,
-      "step": 29050
-    },
-    {
-      "epoch": 15.150456323337679,
-      "grad_norm": 1.5143532752990723,
-      "learning_rate": 7.131557788944724e-05,
-      "loss": 5.4013,
-      "step": 29051
-    },
-    {
-      "epoch": 15.150977835723598,
-      "grad_norm": 1.490962028503418,
-      "learning_rate": 7.131457286432161e-05,
-      "loss": 5.3656,
-      "step": 29052
-    },
-    {
-      "epoch": 15.151499348109517,
-      "grad_norm": 1.4711412191390991,
-      "learning_rate": 7.131356783919598e-05,
-      "loss": 5.4535,
-      "step": 29053
-    },
-    {
-      "epoch": 15.152020860495437,
-      "grad_norm": 1.5155580043792725,
-      "learning_rate": 7.131256281407036e-05,
-      "loss": 4.7827,
-      "step": 29054
-    },
-    {
-      "epoch": 15.152542372881356,
-      "grad_norm": 1.5409817695617676,
-      "learning_rate": 7.131155778894472e-05,
-      "loss": 5.1288,
-      "step": 29055
-    },
-    {
-      "epoch": 15.153063885267276,
-      "grad_norm": 1.4916616678237915,
-      "learning_rate": 7.13105527638191e-05,
-      "loss": 5.3661,
-      "step": 29056
-    },
-    {
-      "epoch": 15.153585397653194,
-      "grad_norm": 1.469427227973938,
-      "learning_rate": 7.130954773869348e-05,
-      "loss": 5.2616,
-      "step": 29057
-    },
-    {
-      "epoch": 15.154106910039113,
-      "grad_norm": 1.4927456378936768,
-      "learning_rate": 7.130854271356784e-05,
-      "loss": 5.3439,
-      "step": 29058
-    },
-    {
-      "epoch": 15.154628422425032,
-      "grad_norm": 1.381862998008728,
-      "learning_rate": 7.130753768844222e-05,
-      "loss": 5.0484,
-      "step": 29059
-    },
-    {
-      "epoch": 15.155149934810952,
-      "grad_norm": 1.444161295890808,
-      "learning_rate": 7.130653266331658e-05,
-      "loss": 4.2875,
-      "step": 29060
-    },
-    {
-      "epoch": 15.155671447196871,
-      "grad_norm": 1.4299434423446655,
-      "learning_rate": 7.130552763819096e-05,
-      "loss": 5.5152,
-      "step": 29061
-    },
-    {
-      "epoch": 15.156192959582791,
-      "grad_norm": 1.4419199228286743,
-      "learning_rate": 7.130452261306533e-05,
-      "loss": 4.9734,
-      "step": 29062
-    },
-    {
-      "epoch": 15.156714471968709,
-      "grad_norm": 1.4400181770324707,
-      "learning_rate": 7.13035175879397e-05,
-      "loss": 5.2818,
-      "step": 29063
-    },
-    {
-      "epoch": 15.157235984354628,
-      "grad_norm": 1.522199273109436,
-      "learning_rate": 7.130251256281407e-05,
-      "loss": 5.142,
-      "step": 29064
-    },
-    {
-      "epoch": 15.157757496740548,
-      "grad_norm": 1.5233980417251587,
-      "learning_rate": 7.130150753768845e-05,
-      "loss": 4.8287,
-      "step": 29065
-    },
-    {
-      "epoch": 15.158279009126467,
-      "grad_norm": 1.3926461935043335,
-      "learning_rate": 7.130050251256282e-05,
-      "loss": 5.5495,
-      "step": 29066
-    },
-    {
-      "epoch": 15.158800521512386,
-      "grad_norm": 1.424590826034546,
-      "learning_rate": 7.12994974874372e-05,
-      "loss": 5.4608,
-      "step": 29067
-    },
-    {
-      "epoch": 15.159322033898306,
-      "grad_norm": 1.5626611709594727,
-      "learning_rate": 7.129849246231157e-05,
-      "loss": 4.8506,
-      "step": 29068
-    },
-    {
-      "epoch": 15.159843546284224,
-      "grad_norm": 1.4464964866638184,
-      "learning_rate": 7.129748743718593e-05,
-      "loss": 5.5378,
-      "step": 29069
-    },
-    {
-      "epoch": 15.160365058670143,
-      "grad_norm": 1.3634130954742432,
-      "learning_rate": 7.129648241206031e-05,
-      "loss": 5.6078,
-      "step": 29070
-    },
-    {
-      "epoch": 15.160886571056063,
-      "grad_norm": 1.4363592863082886,
-      "learning_rate": 7.129547738693467e-05,
-      "loss": 5.6336,
-      "step": 29071
-    },
-    {
-      "epoch": 15.161408083441982,
-      "grad_norm": 1.4152342081069946,
-      "learning_rate": 7.129447236180905e-05,
-      "loss": 5.3938,
-      "step": 29072
-    },
-    {
-      "epoch": 15.161929595827901,
-      "grad_norm": 1.4939249753952026,
-      "learning_rate": 7.129346733668341e-05,
-      "loss": 5.1971,
-      "step": 29073
-    },
-    {
-      "epoch": 15.162451108213821,
-      "grad_norm": 1.510880947113037,
-      "learning_rate": 7.129246231155779e-05,
-      "loss": 4.8044,
-      "step": 29074
-    },
-    {
-      "epoch": 15.162972620599739,
-      "grad_norm": 1.4806228876113892,
-      "learning_rate": 7.129145728643216e-05,
-      "loss": 5.2847,
-      "step": 29075
-    },
-    {
-      "epoch": 15.163494132985658,
-      "grad_norm": 1.5433557033538818,
-      "learning_rate": 7.129045226130653e-05,
-      "loss": 5.2917,
-      "step": 29076
-    },
-    {
-      "epoch": 15.164015645371578,
-      "grad_norm": 1.5955464839935303,
-      "learning_rate": 7.128944723618091e-05,
-      "loss": 5.6031,
-      "step": 29077
-    },
-    {
-      "epoch": 15.164537157757497,
-      "grad_norm": 1.4277589321136475,
-      "learning_rate": 7.128844221105529e-05,
-      "loss": 5.5899,
-      "step": 29078
-    },
-    {
-      "epoch": 15.165058670143416,
-      "grad_norm": 1.5461323261260986,
-      "learning_rate": 7.128743718592965e-05,
-      "loss": 4.5077,
-      "step": 29079
-    },
-    {
-      "epoch": 15.165580182529336,
-      "grad_norm": 1.4442739486694336,
-      "learning_rate": 7.128643216080403e-05,
-      "loss": 5.3417,
-      "step": 29080
-    },
-    {
-      "epoch": 15.166101694915254,
-      "grad_norm": 1.4870091676712036,
-      "learning_rate": 7.12854271356784e-05,
-      "loss": 5.1711,
-      "step": 29081
-    },
-    {
-      "epoch": 15.166623207301173,
-      "grad_norm": 1.4961525201797485,
-      "learning_rate": 7.128442211055276e-05,
-      "loss": 4.8405,
-      "step": 29082
-    },
-    {
-      "epoch": 15.167144719687093,
-      "grad_norm": 1.4067193269729614,
-      "learning_rate": 7.128341708542714e-05,
-      "loss": 4.4454,
-      "step": 29083
-    },
-    {
-      "epoch": 15.167666232073012,
-      "grad_norm": 1.561736822128296,
-      "learning_rate": 7.12824120603015e-05,
-      "loss": 5.3293,
-      "step": 29084
-    },
-    {
-      "epoch": 15.168187744458931,
-      "grad_norm": 1.6595354080200195,
-      "learning_rate": 7.128140703517588e-05,
-      "loss": 5.3938,
-      "step": 29085
-    },
-    {
-      "epoch": 15.16870925684485,
-      "grad_norm": 1.586471438407898,
-      "learning_rate": 7.128040201005026e-05,
-      "loss": 5.2605,
-      "step": 29086
-    },
-    {
-      "epoch": 15.169230769230769,
-      "grad_norm": 1.458974838256836,
-      "learning_rate": 7.127939698492464e-05,
-      "loss": 5.1633,
-      "step": 29087
-    },
-    {
-      "epoch": 15.169752281616688,
-      "grad_norm": 1.4685744047164917,
-      "learning_rate": 7.1278391959799e-05,
-      "loss": 5.3059,
-      "step": 29088
-    },
-    {
-      "epoch": 15.170273794002608,
-      "grad_norm": 1.3904953002929688,
-      "learning_rate": 7.127738693467338e-05,
-      "loss": 5.1445,
-      "step": 29089
-    },
-    {
-      "epoch": 15.170795306388527,
-      "grad_norm": 1.5248311758041382,
-      "learning_rate": 7.127638190954774e-05,
-      "loss": 5.4002,
-      "step": 29090
-    },
-    {
-      "epoch": 15.171316818774446,
-      "grad_norm": 1.5510433912277222,
-      "learning_rate": 7.127537688442212e-05,
-      "loss": 5.3188,
-      "step": 29091
-    },
-    {
-      "epoch": 15.171838331160366,
-      "grad_norm": 1.4637070894241333,
-      "learning_rate": 7.127437185929648e-05,
-      "loss": 5.5029,
-      "step": 29092
-    },
-    {
-      "epoch": 15.172359843546284,
-      "grad_norm": 1.4515470266342163,
-      "learning_rate": 7.127336683417086e-05,
-      "loss": 5.3481,
-      "step": 29093
-    },
-    {
-      "epoch": 15.172881355932203,
-      "grad_norm": 1.493395209312439,
-      "learning_rate": 7.127236180904523e-05,
-      "loss": 5.4914,
-      "step": 29094
-    },
-    {
-      "epoch": 15.173402868318123,
-      "grad_norm": 1.318751573562622,
-      "learning_rate": 7.127135678391959e-05,
-      "loss": 5.6605,
-      "step": 29095
-    },
-    {
-      "epoch": 15.173924380704042,
-      "grad_norm": 1.4380578994750977,
-      "learning_rate": 7.127035175879397e-05,
-      "loss": 5.0638,
-      "step": 29096
-    },
-    {
-      "epoch": 15.174445893089962,
-      "grad_norm": 1.4950019121170044,
-      "learning_rate": 7.126934673366835e-05,
-      "loss": 5.4821,
-      "step": 29097
-    },
-    {
-      "epoch": 15.17496740547588,
-      "grad_norm": 1.3410086631774902,
-      "learning_rate": 7.126834170854272e-05,
-      "loss": 4.8569,
-      "step": 29098
-    },
-    {
-      "epoch": 15.175488917861799,
-      "grad_norm": 1.5146808624267578,
-      "learning_rate": 7.126733668341709e-05,
-      "loss": 4.8845,
-      "step": 29099
-    },
-    {
-      "epoch": 15.176010430247718,
-      "grad_norm": 1.5112154483795166,
-      "learning_rate": 7.126633165829147e-05,
-      "loss": 5.1368,
-      "step": 29100
-    },
-    {
-      "epoch": 15.176531942633638,
-      "grad_norm": 1.5761516094207764,
-      "learning_rate": 7.126532663316583e-05,
-      "loss": 5.2325,
-      "step": 29101
-    },
-    {
-      "epoch": 15.177053455019557,
-      "grad_norm": 1.6685019731521606,
-      "learning_rate": 7.126432160804021e-05,
-      "loss": 5.3374,
-      "step": 29102
-    },
-    {
-      "epoch": 15.177574967405477,
-      "grad_norm": 1.5778214931488037,
-      "learning_rate": 7.126331658291457e-05,
-      "loss": 4.7999,
-      "step": 29103
-    },
-    {
-      "epoch": 15.178096479791394,
-      "grad_norm": 1.432312250137329,
-      "learning_rate": 7.126231155778895e-05,
-      "loss": 5.5594,
-      "step": 29104
-    },
-    {
-      "epoch": 15.178617992177314,
-      "grad_norm": 1.4785010814666748,
-      "learning_rate": 7.126130653266331e-05,
-      "loss": 5.3495,
-      "step": 29105
-    },
-    {
-      "epoch": 15.179139504563233,
-      "grad_norm": 1.4270398616790771,
-      "learning_rate": 7.126030150753769e-05,
-      "loss": 5.4846,
-      "step": 29106
-    },
-    {
-      "epoch": 15.179661016949153,
-      "grad_norm": 1.5278854370117188,
-      "learning_rate": 7.125929648241207e-05,
-      "loss": 5.6027,
-      "step": 29107
-    },
-    {
-      "epoch": 15.180182529335072,
-      "grad_norm": 1.5614577531814575,
-      "learning_rate": 7.125829145728643e-05,
-      "loss": 5.1034,
-      "step": 29108
-    },
-    {
-      "epoch": 15.180704041720992,
-      "grad_norm": 1.4690035581588745,
-      "learning_rate": 7.125728643216081e-05,
-      "loss": 5.6178,
-      "step": 29109
-    },
-    {
-      "epoch": 15.18122555410691,
-      "grad_norm": 1.5037943124771118,
-      "learning_rate": 7.125628140703518e-05,
-      "loss": 5.2741,
-      "step": 29110
-    },
-    {
-      "epoch": 15.181747066492829,
-      "grad_norm": 1.442018747329712,
-      "learning_rate": 7.125527638190955e-05,
-      "loss": 5.457,
-      "step": 29111
-    },
-    {
-      "epoch": 15.182268578878748,
-      "grad_norm": 1.564052700996399,
-      "learning_rate": 7.125427135678392e-05,
-      "loss": 5.3729,
-      "step": 29112
-    },
-    {
-      "epoch": 15.182790091264668,
-      "grad_norm": 1.441823959350586,
-      "learning_rate": 7.12532663316583e-05,
-      "loss": 5.2737,
-      "step": 29113
-    },
-    {
-      "epoch": 15.183311603650587,
-      "grad_norm": 1.4704399108886719,
-      "learning_rate": 7.125226130653266e-05,
-      "loss": 5.7685,
-      "step": 29114
-    },
-    {
-      "epoch": 15.183833116036507,
-      "grad_norm": 1.4440233707427979,
-      "learning_rate": 7.125125628140704e-05,
-      "loss": 5.434,
-      "step": 29115
-    },
-    {
-      "epoch": 15.184354628422424,
-      "grad_norm": 1.4333513975143433,
-      "learning_rate": 7.12502512562814e-05,
-      "loss": 5.059,
-      "step": 29116
-    },
-    {
-      "epoch": 15.184876140808344,
-      "grad_norm": 1.591092824935913,
-      "learning_rate": 7.124924623115578e-05,
-      "loss": 5.1417,
-      "step": 29117
-    },
-    {
-      "epoch": 15.185397653194263,
-      "grad_norm": 1.4443446397781372,
-      "learning_rate": 7.124824120603016e-05,
-      "loss": 5.7497,
-      "step": 29118
-    },
-    {
-      "epoch": 15.185919165580183,
-      "grad_norm": 1.513028621673584,
-      "learning_rate": 7.124723618090454e-05,
-      "loss": 5.5676,
-      "step": 29119
-    },
-    {
-      "epoch": 15.186440677966102,
-      "grad_norm": 1.3937820196151733,
-      "learning_rate": 7.12462311557789e-05,
-      "loss": 5.322,
-      "step": 29120
-    },
-    {
-      "epoch": 15.186962190352022,
-      "grad_norm": 1.438401699066162,
-      "learning_rate": 7.124522613065326e-05,
-      "loss": 5.6547,
-      "step": 29121
-    },
-    {
-      "epoch": 15.18748370273794,
-      "grad_norm": 1.4555140733718872,
-      "learning_rate": 7.124422110552764e-05,
-      "loss": 5.2826,
-      "step": 29122
-    },
-    {
-      "epoch": 15.188005215123859,
-      "grad_norm": 1.4398791790008545,
-      "learning_rate": 7.124321608040201e-05,
-      "loss": 5.7094,
-      "step": 29123
-    },
-    {
-      "epoch": 15.188526727509778,
-      "grad_norm": 1.4832066297531128,
-      "learning_rate": 7.124221105527638e-05,
-      "loss": 5.1163,
-      "step": 29124
-    },
-    {
-      "epoch": 15.189048239895698,
-      "grad_norm": 1.4738657474517822,
-      "learning_rate": 7.124120603015075e-05,
-      "loss": 5.6767,
-      "step": 29125
-    },
-    {
-      "epoch": 15.189569752281617,
-      "grad_norm": 1.456805944442749,
-      "learning_rate": 7.124020100502513e-05,
-      "loss": 4.9472,
-      "step": 29126
-    },
-    {
-      "epoch": 15.190091264667537,
-      "grad_norm": 1.5146116018295288,
-      "learning_rate": 7.12391959798995e-05,
-      "loss": 5.1556,
-      "step": 29127
-    },
-    {
-      "epoch": 15.190612777053454,
-      "grad_norm": 1.4952772855758667,
-      "learning_rate": 7.123819095477388e-05,
-      "loss": 4.9264,
-      "step": 29128
-    },
-    {
-      "epoch": 15.191134289439374,
-      "grad_norm": 1.5599579811096191,
-      "learning_rate": 7.123718592964825e-05,
-      "loss": 5.2785,
-      "step": 29129
-    },
-    {
-      "epoch": 15.191655801825293,
-      "grad_norm": 1.5764679908752441,
-      "learning_rate": 7.123618090452262e-05,
-      "loss": 5.3397,
-      "step": 29130
-    },
-    {
-      "epoch": 15.192177314211213,
-      "grad_norm": 1.3770923614501953,
-      "learning_rate": 7.123517587939699e-05,
-      "loss": 5.4137,
-      "step": 29131
-    },
-    {
-      "epoch": 15.192698826597132,
-      "grad_norm": 1.5447611808776855,
-      "learning_rate": 7.123417085427137e-05,
-      "loss": 5.1089,
-      "step": 29132
-    },
-    {
-      "epoch": 15.193220338983052,
-      "grad_norm": 1.4251041412353516,
-      "learning_rate": 7.123316582914573e-05,
-      "loss": 5.6005,
-      "step": 29133
-    },
-    {
-      "epoch": 15.19374185136897,
-      "grad_norm": 1.4695042371749878,
-      "learning_rate": 7.123216080402011e-05,
-      "loss": 5.5516,
-      "step": 29134
-    },
-    {
-      "epoch": 15.194263363754889,
-      "grad_norm": 1.5098638534545898,
-      "learning_rate": 7.123115577889447e-05,
-      "loss": 5.1196,
-      "step": 29135
-    },
-    {
-      "epoch": 15.194784876140808,
-      "grad_norm": 1.5099310874938965,
-      "learning_rate": 7.123015075376884e-05,
-      "loss": 5.3962,
-      "step": 29136
-    },
-    {
-      "epoch": 15.195306388526728,
-      "grad_norm": 1.5616341829299927,
-      "learning_rate": 7.122914572864322e-05,
-      "loss": 5.1171,
-      "step": 29137
-    },
-    {
-      "epoch": 15.195827900912647,
-      "grad_norm": 1.4255425930023193,
-      "learning_rate": 7.122814070351759e-05,
-      "loss": 5.5468,
-      "step": 29138
-    },
-    {
-      "epoch": 15.196349413298567,
-      "grad_norm": 1.5018121004104614,
-      "learning_rate": 7.122713567839197e-05,
-      "loss": 4.8438,
-      "step": 29139
-    },
-    {
-      "epoch": 15.196870925684484,
-      "grad_norm": 1.408457636833191,
-      "learning_rate": 7.122613065326634e-05,
-      "loss": 5.7668,
-      "step": 29140
-    },
-    {
-      "epoch": 15.197392438070404,
-      "grad_norm": 1.4518191814422607,
-      "learning_rate": 7.122512562814071e-05,
-      "loss": 5.2227,
-      "step": 29141
-    },
-    {
-      "epoch": 15.197913950456323,
-      "grad_norm": 1.388339877128601,
-      "learning_rate": 7.122412060301508e-05,
-      "loss": 4.5772,
-      "step": 29142
-    },
-    {
-      "epoch": 15.198435462842243,
-      "grad_norm": 1.5343230962753296,
-      "learning_rate": 7.122311557788945e-05,
-      "loss": 5.3298,
-      "step": 29143
-    },
-    {
-      "epoch": 15.198956975228162,
-      "grad_norm": 1.4954180717468262,
-      "learning_rate": 7.122211055276382e-05,
-      "loss": 5.4006,
-      "step": 29144
-    },
-    {
-      "epoch": 15.199478487614082,
-      "grad_norm": 1.4627164602279663,
-      "learning_rate": 7.12211055276382e-05,
-      "loss": 5.3932,
-      "step": 29145
-    },
-    {
-      "epoch": 15.2,
-      "grad_norm": 1.4242260456085205,
-      "learning_rate": 7.122010050251256e-05,
-      "loss": 5.4733,
-      "step": 29146
-    },
-    {
-      "epoch": 15.200521512385919,
-      "grad_norm": 1.5321979522705078,
-      "learning_rate": 7.121909547738694e-05,
-      "loss": 4.9814,
-      "step": 29147
-    },
-    {
-      "epoch": 15.201043024771838,
-      "grad_norm": 1.4426547288894653,
-      "learning_rate": 7.12180904522613e-05,
-      "loss": 5.42,
-      "step": 29148
-    },
-    {
-      "epoch": 15.201564537157758,
-      "grad_norm": 1.4291707277297974,
-      "learning_rate": 7.121708542713568e-05,
-      "loss": 5.4717,
-      "step": 29149
-    },
-    {
-      "epoch": 15.202086049543677,
-      "grad_norm": 1.4988031387329102,
-      "learning_rate": 7.121608040201006e-05,
-      "loss": 4.8904,
-      "step": 29150
-    },
-    {
-      "epoch": 15.202607561929597,
-      "grad_norm": 1.3476476669311523,
-      "learning_rate": 7.121507537688442e-05,
-      "loss": 5.7488,
-      "step": 29151
-    },
-    {
-      "epoch": 15.203129074315514,
-      "grad_norm": 1.4425764083862305,
-      "learning_rate": 7.12140703517588e-05,
-      "loss": 5.3255,
-      "step": 29152
-    },
-    {
-      "epoch": 15.203650586701434,
-      "grad_norm": 1.4146543741226196,
-      "learning_rate": 7.121306532663317e-05,
-      "loss": 5.4096,
-      "step": 29153
-    },
-    {
-      "epoch": 15.204172099087353,
-      "grad_norm": 1.4881154298782349,
-      "learning_rate": 7.121206030150754e-05,
-      "loss": 5.3535,
-      "step": 29154
-    },
-    {
-      "epoch": 15.204693611473273,
-      "grad_norm": 1.367303490638733,
-      "learning_rate": 7.121105527638191e-05,
-      "loss": 5.5083,
-      "step": 29155
-    },
-    {
-      "epoch": 15.205215123859192,
-      "grad_norm": 1.4239705801010132,
-      "learning_rate": 7.121005025125629e-05,
-      "loss": 5.645,
-      "step": 29156
-    },
-    {
-      "epoch": 15.205736636245112,
-      "grad_norm": 1.4227157831192017,
-      "learning_rate": 7.120904522613065e-05,
-      "loss": 5.5217,
-      "step": 29157
-    },
-    {
-      "epoch": 15.20625814863103,
-      "grad_norm": 1.5438764095306396,
-      "learning_rate": 7.120804020100503e-05,
-      "loss": 5.3585,
-      "step": 29158
-    },
-    {
-      "epoch": 15.206779661016949,
-      "grad_norm": 1.4319339990615845,
-      "learning_rate": 7.12070351758794e-05,
-      "loss": 5.2934,
-      "step": 29159
-    },
-    {
-      "epoch": 15.207301173402868,
-      "grad_norm": 1.4700630903244019,
-      "learning_rate": 7.120603015075378e-05,
-      "loss": 5.3573,
-      "step": 29160
-    },
-    {
-      "epoch": 15.207822685788788,
-      "grad_norm": 1.4925076961517334,
-      "learning_rate": 7.120502512562815e-05,
-      "loss": 4.9017,
-      "step": 29161
-    },
-    {
-      "epoch": 15.208344198174707,
-      "grad_norm": 1.3639377355575562,
-      "learning_rate": 7.120402010050251e-05,
-      "loss": 5.325,
-      "step": 29162
-    },
-    {
-      "epoch": 15.208865710560627,
-      "grad_norm": 1.4609572887420654,
-      "learning_rate": 7.120301507537689e-05,
-      "loss": 5.4757,
-      "step": 29163
-    },
-    {
-      "epoch": 15.209387222946544,
-      "grad_norm": 1.4827523231506348,
-      "learning_rate": 7.120201005025125e-05,
-      "loss": 5.4588,
-      "step": 29164
-    },
-    {
-      "epoch": 15.209908735332464,
-      "grad_norm": 1.422369122505188,
-      "learning_rate": 7.120100502512563e-05,
-      "loss": 5.6691,
-      "step": 29165
-    },
-    {
-      "epoch": 15.210430247718383,
-      "grad_norm": 1.429643988609314,
-      "learning_rate": 7.12e-05,
-      "loss": 5.0762,
-      "step": 29166
-    },
-    {
-      "epoch": 15.210951760104303,
-      "grad_norm": 1.5050907135009766,
-      "learning_rate": 7.119899497487437e-05,
-      "loss": 5.1685,
-      "step": 29167
-    },
-    {
-      "epoch": 15.211473272490222,
-      "grad_norm": 1.6214591264724731,
-      "learning_rate": 7.119798994974874e-05,
-      "loss": 5.2729,
-      "step": 29168
-    },
-    {
-      "epoch": 15.211994784876142,
-      "grad_norm": 1.498030185699463,
-      "learning_rate": 7.119698492462312e-05,
-      "loss": 5.361,
-      "step": 29169
-    },
-    {
-      "epoch": 15.21251629726206,
-      "grad_norm": 1.5234391689300537,
-      "learning_rate": 7.11959798994975e-05,
-      "loss": 5.6009,
-      "step": 29170
-    },
-    {
-      "epoch": 15.213037809647979,
-      "grad_norm": 1.3983094692230225,
-      "learning_rate": 7.119497487437187e-05,
-      "loss": 5.6097,
-      "step": 29171
-    },
-    {
-      "epoch": 15.213559322033898,
-      "grad_norm": 1.4478343725204468,
-      "learning_rate": 7.119396984924624e-05,
-      "loss": 4.8635,
-      "step": 29172
-    },
-    {
-      "epoch": 15.214080834419818,
-      "grad_norm": 1.500260829925537,
-      "learning_rate": 7.119296482412061e-05,
-      "loss": 5.0577,
-      "step": 29173
-    },
-    {
-      "epoch": 15.214602346805737,
-      "grad_norm": 1.316672921180725,
-      "learning_rate": 7.119195979899498e-05,
-      "loss": 5.6918,
-      "step": 29174
-    },
-    {
-      "epoch": 15.215123859191657,
-      "grad_norm": 1.434207558631897,
-      "learning_rate": 7.119095477386934e-05,
-      "loss": 5.4138,
-      "step": 29175
-    },
-    {
-      "epoch": 15.215645371577574,
-      "grad_norm": 1.4147169589996338,
-      "learning_rate": 7.118994974874372e-05,
-      "loss": 5.4568,
-      "step": 29176
-    },
-    {
-      "epoch": 15.216166883963494,
-      "grad_norm": 1.4941564798355103,
-      "learning_rate": 7.118894472361808e-05,
-      "loss": 5.0491,
-      "step": 29177
-    },
-    {
-      "epoch": 15.216688396349413,
-      "grad_norm": 1.4595413208007812,
-      "learning_rate": 7.118793969849246e-05,
-      "loss": 5.522,
-      "step": 29178
-    },
-    {
-      "epoch": 15.217209908735333,
-      "grad_norm": 1.4628609418869019,
-      "learning_rate": 7.118693467336684e-05,
-      "loss": 5.6684,
-      "step": 29179
-    },
-    {
-      "epoch": 15.217731421121252,
-      "grad_norm": 1.4377925395965576,
-      "learning_rate": 7.118592964824122e-05,
-      "loss": 5.4969,
-      "step": 29180
-    },
-    {
-      "epoch": 15.21825293350717,
-      "grad_norm": 1.4442487955093384,
-      "learning_rate": 7.118492462311558e-05,
-      "loss": 5.7352,
-      "step": 29181
-    },
-    {
-      "epoch": 15.21877444589309,
-      "grad_norm": 1.4414522647857666,
-      "learning_rate": 7.118391959798996e-05,
-      "loss": 5.7265,
-      "step": 29182
-    },
-    {
-      "epoch": 15.219295958279009,
-      "grad_norm": 1.4220292568206787,
-      "learning_rate": 7.118291457286432e-05,
-      "loss": 5.5706,
-      "step": 29183
-    },
-    {
-      "epoch": 15.219817470664928,
-      "grad_norm": 1.412300944328308,
-      "learning_rate": 7.11819095477387e-05,
-      "loss": 5.7376,
-      "step": 29184
-    },
-    {
-      "epoch": 15.220338983050848,
-      "grad_norm": 1.3802800178527832,
-      "learning_rate": 7.118090452261307e-05,
-      "loss": 5.8763,
-      "step": 29185
-    },
-    {
-      "epoch": 15.220860495436767,
-      "grad_norm": 1.5501171350479126,
-      "learning_rate": 7.117989949748744e-05,
-      "loss": 5.3293,
-      "step": 29186
-    },
-    {
-      "epoch": 15.221382007822687,
-      "grad_norm": 1.4845623970031738,
-      "learning_rate": 7.117889447236181e-05,
-      "loss": 5.6513,
-      "step": 29187
-    },
-    {
-      "epoch": 15.221903520208604,
-      "grad_norm": 1.5787614583969116,
-      "learning_rate": 7.117788944723617e-05,
-      "loss": 5.0471,
-      "step": 29188
-    },
-    {
-      "epoch": 15.222425032594524,
-      "grad_norm": 1.7037742137908936,
-      "learning_rate": 7.117688442211055e-05,
-      "loss": 4.5109,
-      "step": 29189
-    },
-    {
-      "epoch": 15.222946544980443,
-      "grad_norm": 1.389772891998291,
-      "learning_rate": 7.117587939698493e-05,
-      "loss": 5.4648,
-      "step": 29190
-    },
-    {
-      "epoch": 15.223468057366363,
-      "grad_norm": 1.3605092763900757,
-      "learning_rate": 7.11748743718593e-05,
-      "loss": 5.6748,
-      "step": 29191
-    },
-    {
-      "epoch": 15.223989569752282,
-      "grad_norm": 1.3752412796020508,
-      "learning_rate": 7.117386934673367e-05,
-      "loss": 5.1244,
-      "step": 29192
-    },
-    {
-      "epoch": 15.2245110821382,
-      "grad_norm": 1.431384801864624,
-      "learning_rate": 7.117286432160805e-05,
-      "loss": 5.4054,
-      "step": 29193
-    },
-    {
-      "epoch": 15.22503259452412,
-      "grad_norm": 1.4368969202041626,
-      "learning_rate": 7.117185929648241e-05,
-      "loss": 5.6442,
-      "step": 29194
-    },
-    {
-      "epoch": 15.225554106910039,
-      "grad_norm": 1.44606351852417,
-      "learning_rate": 7.117085427135679e-05,
-      "loss": 5.4648,
-      "step": 29195
-    },
-    {
-      "epoch": 15.226075619295958,
-      "grad_norm": 1.4192358255386353,
-      "learning_rate": 7.116984924623115e-05,
-      "loss": 4.9874,
-      "step": 29196
-    },
-    {
-      "epoch": 15.226597131681878,
-      "grad_norm": 1.4571563005447388,
-      "learning_rate": 7.116884422110553e-05,
-      "loss": 5.3315,
-      "step": 29197
-    },
-    {
-      "epoch": 15.227118644067797,
-      "grad_norm": 1.5193946361541748,
-      "learning_rate": 7.11678391959799e-05,
-      "loss": 5.375,
-      "step": 29198
-    },
-    {
-      "epoch": 15.227640156453715,
-      "grad_norm": 1.4320123195648193,
-      "learning_rate": 7.116683417085427e-05,
-      "loss": 5.4761,
-      "step": 29199
-    },
-    {
-      "epoch": 15.228161668839634,
-      "grad_norm": 1.718379020690918,
-      "learning_rate": 7.116582914572865e-05,
-      "loss": 4.9132,
-      "step": 29200
-    },
-    {
-      "epoch": 15.228683181225554,
-      "grad_norm": 1.420807123184204,
-      "learning_rate": 7.116482412060302e-05,
-      "loss": 5.6224,
-      "step": 29201
-    },
-    {
-      "epoch": 15.229204693611473,
-      "grad_norm": 1.4570138454437256,
-      "learning_rate": 7.11638190954774e-05,
-      "loss": 5.3667,
-      "step": 29202
-    },
-    {
-      "epoch": 15.229726205997393,
-      "grad_norm": 1.5001001358032227,
-      "learning_rate": 7.116281407035176e-05,
-      "loss": 5.2659,
-      "step": 29203
-    },
-    {
-      "epoch": 15.230247718383312,
-      "grad_norm": 1.5709587335586548,
-      "learning_rate": 7.116180904522614e-05,
-      "loss": 5.0525,
-      "step": 29204
-    },
-    {
-      "epoch": 15.23076923076923,
-      "grad_norm": 1.4856010675430298,
-      "learning_rate": 7.11608040201005e-05,
-      "loss": 5.3558,
-      "step": 29205
-    },
-    {
-      "epoch": 15.23129074315515,
-      "grad_norm": 1.4104492664337158,
-      "learning_rate": 7.115979899497488e-05,
-      "loss": 5.4161,
-      "step": 29206
-    },
-    {
-      "epoch": 15.231812255541069,
-      "grad_norm": 1.4696192741394043,
-      "learning_rate": 7.115879396984924e-05,
-      "loss": 5.4708,
-      "step": 29207
-    },
-    {
-      "epoch": 15.232333767926988,
-      "grad_norm": 1.5627691745758057,
-      "learning_rate": 7.115778894472362e-05,
-      "loss": 4.8357,
-      "step": 29208
-    },
-    {
-      "epoch": 15.232855280312908,
-      "grad_norm": 1.4541746377944946,
-      "learning_rate": 7.115678391959799e-05,
-      "loss": 5.8107,
-      "step": 29209
-    },
-    {
-      "epoch": 15.233376792698827,
-      "grad_norm": 1.5496139526367188,
-      "learning_rate": 7.115577889447236e-05,
-      "loss": 5.2817,
-      "step": 29210
-    },
-    {
-      "epoch": 15.233898305084745,
-      "grad_norm": 1.4731202125549316,
-      "learning_rate": 7.115477386934674e-05,
-      "loss": 4.6094,
-      "step": 29211
-    },
-    {
-      "epoch": 15.234419817470664,
-      "grad_norm": 1.4015226364135742,
-      "learning_rate": 7.115376884422112e-05,
-      "loss": 5.4218,
-      "step": 29212
-    },
-    {
-      "epoch": 15.234941329856584,
-      "grad_norm": 1.5217863321304321,
-      "learning_rate": 7.115276381909548e-05,
-      "loss": 5.0055,
-      "step": 29213
-    },
-    {
-      "epoch": 15.235462842242503,
-      "grad_norm": 1.5536103248596191,
-      "learning_rate": 7.115175879396985e-05,
-      "loss": 5.6109,
-      "step": 29214
-    },
-    {
-      "epoch": 15.235984354628423,
-      "grad_norm": 1.4529712200164795,
-      "learning_rate": 7.115075376884422e-05,
-      "loss": 5.1843,
-      "step": 29215
-    },
-    {
-      "epoch": 15.236505867014342,
-      "grad_norm": 1.4078140258789062,
-      "learning_rate": 7.114974874371859e-05,
-      "loss": 5.3024,
-      "step": 29216
-    },
-    {
-      "epoch": 15.23702737940026,
-      "grad_norm": 1.6242976188659668,
-      "learning_rate": 7.114874371859297e-05,
-      "loss": 4.9754,
-      "step": 29217
-    },
-    {
-      "epoch": 15.23754889178618,
-      "grad_norm": 1.4981985092163086,
-      "learning_rate": 7.114773869346733e-05,
-      "loss": 5.41,
-      "step": 29218
-    },
-    {
-      "epoch": 15.238070404172099,
-      "grad_norm": 1.4471553564071655,
-      "learning_rate": 7.114673366834171e-05,
-      "loss": 5.5814,
-      "step": 29219
-    },
-    {
-      "epoch": 15.238591916558018,
-      "grad_norm": 1.4787113666534424,
-      "learning_rate": 7.114572864321609e-05,
-      "loss": 5.4081,
-      "step": 29220
-    },
-    {
-      "epoch": 15.239113428943938,
-      "grad_norm": 1.5602024793624878,
-      "learning_rate": 7.114472361809046e-05,
-      "loss": 5.5021,
-      "step": 29221
-    },
-    {
-      "epoch": 15.239634941329857,
-      "grad_norm": 1.4478280544281006,
-      "learning_rate": 7.114371859296483e-05,
-      "loss": 5.3845,
-      "step": 29222
-    },
-    {
-      "epoch": 15.240156453715775,
-      "grad_norm": 1.4182497262954712,
-      "learning_rate": 7.114271356783921e-05,
-      "loss": 5.3386,
-      "step": 29223
-    },
-    {
-      "epoch": 15.240677966101694,
-      "grad_norm": 1.5345991849899292,
-      "learning_rate": 7.114170854271357e-05,
-      "loss": 5.0601,
-      "step": 29224
-    },
-    {
-      "epoch": 15.241199478487614,
-      "grad_norm": 1.5500789880752563,
-      "learning_rate": 7.114070351758795e-05,
-      "loss": 5.3619,
-      "step": 29225
-    },
-    {
-      "epoch": 15.241720990873533,
-      "grad_norm": 1.6311488151550293,
-      "learning_rate": 7.113969849246231e-05,
-      "loss": 4.9737,
-      "step": 29226
-    },
-    {
-      "epoch": 15.242242503259453,
-      "grad_norm": 1.5715339183807373,
-      "learning_rate": 7.113869346733669e-05,
-      "loss": 5.4178,
-      "step": 29227
-    },
-    {
-      "epoch": 15.242764015645372,
-      "grad_norm": 1.4906986951828003,
-      "learning_rate": 7.113768844221106e-05,
-      "loss": 5.7101,
-      "step": 29228
-    },
-    {
-      "epoch": 15.24328552803129,
-      "grad_norm": 1.519370436668396,
-      "learning_rate": 7.113668341708542e-05,
-      "loss": 5.2404,
-      "step": 29229
-    },
-    {
-      "epoch": 15.24380704041721,
-      "grad_norm": 1.4089536666870117,
-      "learning_rate": 7.11356783919598e-05,
-      "loss": 5.6062,
-      "step": 29230
-    },
-    {
-      "epoch": 15.244328552803129,
-      "grad_norm": 1.4868764877319336,
-      "learning_rate": 7.113467336683418e-05,
-      "loss": 5.1736,
-      "step": 29231
-    },
-    {
-      "epoch": 15.244850065189048,
-      "grad_norm": 1.467621922492981,
-      "learning_rate": 7.113366834170855e-05,
-      "loss": 5.2968,
-      "step": 29232
-    },
-    {
-      "epoch": 15.245371577574968,
-      "grad_norm": 1.4765105247497559,
-      "learning_rate": 7.113266331658292e-05,
-      "loss": 5.4868,
-      "step": 29233
-    },
-    {
-      "epoch": 15.245893089960887,
-      "grad_norm": 1.5225013494491577,
-      "learning_rate": 7.11316582914573e-05,
-      "loss": 5.211,
-      "step": 29234
-    },
-    {
-      "epoch": 15.246414602346805,
-      "grad_norm": 1.4680237770080566,
-      "learning_rate": 7.113065326633166e-05,
-      "loss": 5.5159,
-      "step": 29235
-    },
-    {
-      "epoch": 15.246936114732724,
-      "grad_norm": 1.4028443098068237,
-      "learning_rate": 7.112964824120604e-05,
-      "loss": 4.6197,
-      "step": 29236
-    },
-    {
-      "epoch": 15.247457627118644,
-      "grad_norm": 1.474017858505249,
-      "learning_rate": 7.11286432160804e-05,
-      "loss": 5.0051,
-      "step": 29237
-    },
-    {
-      "epoch": 15.247979139504563,
-      "grad_norm": 1.508884310722351,
-      "learning_rate": 7.112763819095478e-05,
-      "loss": 4.9394,
-      "step": 29238
-    },
-    {
-      "epoch": 15.248500651890483,
-      "grad_norm": 1.6887837648391724,
-      "learning_rate": 7.112663316582914e-05,
-      "loss": 5.036,
-      "step": 29239
-    },
-    {
-      "epoch": 15.249022164276402,
-      "grad_norm": 1.6257001161575317,
-      "learning_rate": 7.112562814070352e-05,
-      "loss": 5.1522,
-      "step": 29240
-    },
-    {
-      "epoch": 15.24954367666232,
-      "grad_norm": 1.3846627473831177,
-      "learning_rate": 7.11246231155779e-05,
-      "loss": 5.5358,
-      "step": 29241
-    },
-    {
-      "epoch": 15.25006518904824,
-      "grad_norm": 1.537520408630371,
-      "learning_rate": 7.112361809045226e-05,
-      "loss": 5.3058,
-      "step": 29242
-    },
-    {
-      "epoch": 15.250586701434159,
-      "grad_norm": 1.5530458688735962,
-      "learning_rate": 7.112261306532664e-05,
-      "loss": 5.4403,
-      "step": 29243
-    },
-    {
-      "epoch": 15.251108213820078,
-      "grad_norm": 1.5129646062850952,
-      "learning_rate": 7.1121608040201e-05,
-      "loss": 5.1782,
-      "step": 29244
-    },
-    {
-      "epoch": 15.251629726205998,
-      "grad_norm": 1.4050081968307495,
-      "learning_rate": 7.112060301507538e-05,
-      "loss": 5.3821,
-      "step": 29245
-    },
-    {
-      "epoch": 15.252151238591917,
-      "grad_norm": 1.4168224334716797,
-      "learning_rate": 7.111959798994975e-05,
-      "loss": 5.7571,
-      "step": 29246
-    },
-    {
-      "epoch": 15.252672750977835,
-      "grad_norm": 1.354550838470459,
-      "learning_rate": 7.111859296482413e-05,
-      "loss": 5.594,
-      "step": 29247
-    },
-    {
-      "epoch": 15.253194263363755,
-      "grad_norm": 1.5746275186538696,
-      "learning_rate": 7.111758793969849e-05,
-      "loss": 5.1267,
-      "step": 29248
-    },
-    {
-      "epoch": 15.253715775749674,
-      "grad_norm": 1.4472053050994873,
-      "learning_rate": 7.111658291457287e-05,
-      "loss": 5.2917,
-      "step": 29249
-    },
-    {
-      "epoch": 15.254237288135593,
-      "grad_norm": 1.4611178636550903,
-      "learning_rate": 7.111557788944723e-05,
-      "loss": 5.4026,
-      "step": 29250
-    },
-    {
-      "epoch": 15.254758800521513,
-      "grad_norm": 1.3689262866973877,
-      "learning_rate": 7.111457286432161e-05,
-      "loss": 5.6506,
-      "step": 29251
-    },
-    {
-      "epoch": 15.255280312907432,
-      "grad_norm": 1.4092766046524048,
-      "learning_rate": 7.111356783919599e-05,
-      "loss": 5.2928,
-      "step": 29252
-    },
-    {
-      "epoch": 15.25580182529335,
-      "grad_norm": 1.5629675388336182,
-      "learning_rate": 7.111256281407037e-05,
-      "loss": 5.4194,
-      "step": 29253
-    },
-    {
-      "epoch": 15.25632333767927,
-      "grad_norm": 1.394439697265625,
-      "learning_rate": 7.111155778894473e-05,
-      "loss": 5.652,
-      "step": 29254
-    },
-    {
-      "epoch": 15.256844850065189,
-      "grad_norm": 1.447618007659912,
-      "learning_rate": 7.11105527638191e-05,
-      "loss": 4.6829,
-      "step": 29255
-    },
-    {
-      "epoch": 15.257366362451108,
-      "grad_norm": 1.419729232788086,
-      "learning_rate": 7.110954773869347e-05,
-      "loss": 5.4535,
-      "step": 29256
-    },
-    {
-      "epoch": 15.257887874837028,
-      "grad_norm": 1.558736801147461,
-      "learning_rate": 7.110854271356784e-05,
-      "loss": 5.2032,
-      "step": 29257
-    },
-    {
-      "epoch": 15.258409387222947,
-      "grad_norm": 1.50729238986969,
-      "learning_rate": 7.110753768844221e-05,
-      "loss": 5.5548,
-      "step": 29258
-    },
-    {
-      "epoch": 15.258930899608865,
-      "grad_norm": 1.5075544118881226,
-      "learning_rate": 7.110653266331658e-05,
-      "loss": 5.1194,
-      "step": 29259
-    },
-    {
-      "epoch": 15.259452411994785,
-      "grad_norm": 3.02138614654541,
-      "learning_rate": 7.110552763819096e-05,
-      "loss": 5.3687,
-      "step": 29260
-    },
-    {
-      "epoch": 15.259973924380704,
-      "grad_norm": 1.3951187133789062,
-      "learning_rate": 7.110452261306533e-05,
-      "loss": 5.686,
-      "step": 29261
-    },
-    {
-      "epoch": 15.260495436766623,
-      "grad_norm": 1.5155194997787476,
-      "learning_rate": 7.110351758793971e-05,
-      "loss": 5.1391,
-      "step": 29262
-    },
-    {
-      "epoch": 15.261016949152543,
-      "grad_norm": 1.5972416400909424,
-      "learning_rate": 7.110251256281408e-05,
-      "loss": 5.3235,
-      "step": 29263
-    },
-    {
-      "epoch": 15.261538461538462,
-      "grad_norm": 1.3795753717422485,
-      "learning_rate": 7.110150753768845e-05,
-      "loss": 5.4485,
-      "step": 29264
-    },
-    {
-      "epoch": 15.26205997392438,
-      "grad_norm": 1.5859390497207642,
-      "learning_rate": 7.110050251256282e-05,
-      "loss": 5.115,
-      "step": 29265
-    },
-    {
-      "epoch": 15.2625814863103,
-      "grad_norm": 1.583330512046814,
-      "learning_rate": 7.10994974874372e-05,
-      "loss": 5.5108,
-      "step": 29266
-    },
-    {
-      "epoch": 15.263102998696219,
-      "grad_norm": 1.4834638833999634,
-      "learning_rate": 7.109849246231156e-05,
-      "loss": 5.269,
-      "step": 29267
-    },
-    {
-      "epoch": 15.263624511082138,
-      "grad_norm": 1.4970821142196655,
-      "learning_rate": 7.109748743718592e-05,
-      "loss": 5.5431,
-      "step": 29268
-    },
-    {
-      "epoch": 15.264146023468058,
-      "grad_norm": 1.4608657360076904,
-      "learning_rate": 7.10964824120603e-05,
-      "loss": 5.5366,
-      "step": 29269
-    },
-    {
-      "epoch": 15.264667535853977,
-      "grad_norm": 1.7062788009643555,
-      "learning_rate": 7.109547738693467e-05,
-      "loss": 5.2902,
-      "step": 29270
-    },
-    {
-      "epoch": 15.265189048239895,
-      "grad_norm": 1.5142731666564941,
-      "learning_rate": 7.109447236180904e-05,
-      "loss": 5.138,
-      "step": 29271
-    },
-    {
-      "epoch": 15.265710560625815,
-      "grad_norm": 1.6138337850570679,
-      "learning_rate": 7.109346733668342e-05,
-      "loss": 4.7861,
-      "step": 29272
-    },
-    {
-      "epoch": 15.266232073011734,
-      "grad_norm": 1.5144842863082886,
-      "learning_rate": 7.10924623115578e-05,
-      "loss": 5.0427,
-      "step": 29273
-    },
-    {
-      "epoch": 15.266753585397653,
-      "grad_norm": 1.430625081062317,
-      "learning_rate": 7.109145728643216e-05,
-      "loss": 5.4273,
-      "step": 29274
-    },
-    {
-      "epoch": 15.267275097783573,
-      "grad_norm": 1.5087437629699707,
-      "learning_rate": 7.109045226130654e-05,
-      "loss": 5.2919,
-      "step": 29275
-    },
-    {
-      "epoch": 15.26779661016949,
-      "grad_norm": 1.497135877609253,
-      "learning_rate": 7.10894472361809e-05,
-      "loss": 5.5167,
-      "step": 29276
-    },
-    {
-      "epoch": 15.26831812255541,
-      "grad_norm": 1.492318868637085,
-      "learning_rate": 7.108844221105528e-05,
-      "loss": 5.2947,
-      "step": 29277
-    },
-    {
-      "epoch": 15.26883963494133,
-      "grad_norm": 1.5044126510620117,
-      "learning_rate": 7.108743718592965e-05,
-      "loss": 5.0413,
-      "step": 29278
-    },
-    {
-      "epoch": 15.269361147327249,
-      "grad_norm": 1.4316357374191284,
-      "learning_rate": 7.108643216080403e-05,
-      "loss": 5.6448,
-      "step": 29279
-    },
-    {
-      "epoch": 15.269882659713168,
-      "grad_norm": 1.4216103553771973,
-      "learning_rate": 7.108542713567839e-05,
-      "loss": 5.1982,
-      "step": 29280
-    },
-    {
-      "epoch": 15.270404172099088,
-      "grad_norm": 1.4883488416671753,
-      "learning_rate": 7.108442211055277e-05,
-      "loss": 5.1059,
-      "step": 29281
-    },
-    {
-      "epoch": 15.270925684485007,
-      "grad_norm": 1.5367505550384521,
-      "learning_rate": 7.108341708542715e-05,
-      "loss": 5.0148,
-      "step": 29282
-    },
-    {
-      "epoch": 15.271447196870925,
-      "grad_norm": 1.5230896472930908,
-      "learning_rate": 7.108241206030151e-05,
-      "loss": 5.0652,
-      "step": 29283
-    },
-    {
-      "epoch": 15.271968709256845,
-      "grad_norm": 1.63201904296875,
-      "learning_rate": 7.108140703517589e-05,
-      "loss": 4.5893,
-      "step": 29284
-    },
-    {
-      "epoch": 15.272490221642764,
-      "grad_norm": 1.490744948387146,
-      "learning_rate": 7.108040201005025e-05,
-      "loss": 5.7108,
-      "step": 29285
-    },
-    {
-      "epoch": 15.273011734028684,
-      "grad_norm": 1.468396782875061,
-      "learning_rate": 7.107939698492463e-05,
-      "loss": 5.4588,
-      "step": 29286
-    },
-    {
-      "epoch": 15.273533246414603,
-      "grad_norm": 1.4380784034729004,
-      "learning_rate": 7.1078391959799e-05,
-      "loss": 5.566,
-      "step": 29287
-    },
-    {
-      "epoch": 15.27405475880052,
-      "grad_norm": 1.6346771717071533,
-      "learning_rate": 7.107738693467337e-05,
-      "loss": 5.1915,
-      "step": 29288
-    },
-    {
-      "epoch": 15.27457627118644,
-      "grad_norm": 1.4833974838256836,
-      "learning_rate": 7.107638190954774e-05,
-      "loss": 5.3341,
-      "step": 29289
-    },
-    {
-      "epoch": 15.27509778357236,
-      "grad_norm": 1.5055322647094727,
-      "learning_rate": 7.107537688442211e-05,
-      "loss": 4.8328,
-      "step": 29290
-    },
-    {
-      "epoch": 15.275619295958279,
-      "grad_norm": 1.4856635332107544,
-      "learning_rate": 7.107437185929648e-05,
-      "loss": 5.3008,
-      "step": 29291
-    },
-    {
-      "epoch": 15.276140808344199,
-      "grad_norm": 1.5556002855300903,
-      "learning_rate": 7.107336683417086e-05,
-      "loss": 4.6245,
-      "step": 29292
-    },
-    {
-      "epoch": 15.276662320730118,
-      "grad_norm": 1.43301522731781,
-      "learning_rate": 7.107236180904523e-05,
-      "loss": 5.7796,
-      "step": 29293
-    },
-    {
-      "epoch": 15.277183833116036,
-      "grad_norm": 1.4609949588775635,
-      "learning_rate": 7.10713567839196e-05,
-      "loss": 5.5224,
-      "step": 29294
-    },
-    {
-      "epoch": 15.277705345501955,
-      "grad_norm": 1.6017494201660156,
-      "learning_rate": 7.107035175879398e-05,
-      "loss": 5.3137,
-      "step": 29295
-    },
-    {
-      "epoch": 15.278226857887875,
-      "grad_norm": 1.5692193508148193,
-      "learning_rate": 7.106934673366834e-05,
-      "loss": 5.5621,
-      "step": 29296
-    },
-    {
-      "epoch": 15.278748370273794,
-      "grad_norm": 1.3760876655578613,
-      "learning_rate": 7.106834170854272e-05,
-      "loss": 5.4452,
-      "step": 29297
-    },
-    {
-      "epoch": 15.279269882659714,
-      "grad_norm": 1.3698985576629639,
-      "learning_rate": 7.106733668341708e-05,
-      "loss": 5.0637,
-      "step": 29298
-    },
-    {
-      "epoch": 15.279791395045633,
-      "grad_norm": 1.5468356609344482,
-      "learning_rate": 7.106633165829146e-05,
-      "loss": 5.443,
-      "step": 29299
-    },
-    {
-      "epoch": 15.28031290743155,
-      "grad_norm": 1.4911890029907227,
-      "learning_rate": 7.106532663316583e-05,
-      "loss": 5.3503,
-      "step": 29300
-    },
-    {
-      "epoch": 15.28083441981747,
-      "grad_norm": 1.4646217823028564,
-      "learning_rate": 7.10643216080402e-05,
-      "loss": 5.1669,
-      "step": 29301
-    },
-    {
-      "epoch": 15.28135593220339,
-      "grad_norm": 1.485007643699646,
-      "learning_rate": 7.106331658291458e-05,
-      "loss": 5.436,
-      "step": 29302
-    },
-    {
-      "epoch": 15.281877444589309,
-      "grad_norm": 1.4798847436904907,
-      "learning_rate": 7.106231155778896e-05,
-      "loss": 5.4015,
-      "step": 29303
-    },
-    {
-      "epoch": 15.282398956975229,
-      "grad_norm": 1.4003758430480957,
-      "learning_rate": 7.106130653266332e-05,
-      "loss": 5.6618,
-      "step": 29304
-    },
-    {
-      "epoch": 15.282920469361148,
-      "grad_norm": 1.667283058166504,
-      "learning_rate": 7.10603015075377e-05,
-      "loss": 4.7938,
-      "step": 29305
-    },
-    {
-      "epoch": 15.283441981747066,
-      "grad_norm": 1.4326856136322021,
-      "learning_rate": 7.105929648241207e-05,
-      "loss": 5.4574,
-      "step": 29306
-    },
-    {
-      "epoch": 15.283963494132985,
-      "grad_norm": 1.439226508140564,
-      "learning_rate": 7.105829145728644e-05,
-      "loss": 5.1851,
-      "step": 29307
-    },
-    {
-      "epoch": 15.284485006518905,
-      "grad_norm": 1.36747407913208,
-      "learning_rate": 7.105728643216081e-05,
-      "loss": 5.6768,
-      "step": 29308
-    },
-    {
-      "epoch": 15.285006518904824,
-      "grad_norm": 1.4422301054000854,
-      "learning_rate": 7.105628140703517e-05,
-      "loss": 5.5489,
-      "step": 29309
-    },
-    {
-      "epoch": 15.285528031290744,
-      "grad_norm": 1.4019887447357178,
-      "learning_rate": 7.105527638190955e-05,
-      "loss": 5.667,
-      "step": 29310
-    },
-    {
-      "epoch": 15.286049543676663,
-      "grad_norm": 1.4363765716552734,
-      "learning_rate": 7.105427135678391e-05,
-      "loss": 5.3976,
-      "step": 29311
-    },
-    {
-      "epoch": 15.28657105606258,
-      "grad_norm": 1.501641035079956,
-      "learning_rate": 7.105326633165829e-05,
-      "loss": 5.1065,
-      "step": 29312
-    },
-    {
-      "epoch": 15.2870925684485,
-      "grad_norm": 1.4707731008529663,
-      "learning_rate": 7.105226130653267e-05,
-      "loss": 5.6657,
-      "step": 29313
-    },
-    {
-      "epoch": 15.28761408083442,
-      "grad_norm": 1.3933601379394531,
-      "learning_rate": 7.105125628140705e-05,
-      "loss": 5.389,
-      "step": 29314
-    },
-    {
-      "epoch": 15.288135593220339,
-      "grad_norm": 1.5441107749938965,
-      "learning_rate": 7.105025125628141e-05,
-      "loss": 5.1169,
-      "step": 29315
-    },
-    {
-      "epoch": 15.288657105606259,
-      "grad_norm": 1.4338984489440918,
-      "learning_rate": 7.104924623115579e-05,
-      "loss": 5.4118,
-      "step": 29316
-    },
-    {
-      "epoch": 15.289178617992178,
-      "grad_norm": 1.4471893310546875,
-      "learning_rate": 7.104824120603015e-05,
-      "loss": 5.1153,
-      "step": 29317
-    },
-    {
-      "epoch": 15.289700130378096,
-      "grad_norm": 1.4199351072311401,
-      "learning_rate": 7.104723618090453e-05,
-      "loss": 5.2912,
-      "step": 29318
-    },
-    {
-      "epoch": 15.290221642764015,
-      "grad_norm": 1.5104995965957642,
-      "learning_rate": 7.10462311557789e-05,
-      "loss": 5.3142,
-      "step": 29319
-    },
-    {
-      "epoch": 15.290743155149935,
-      "grad_norm": 1.546210765838623,
-      "learning_rate": 7.104522613065327e-05,
-      "loss": 5.4852,
-      "step": 29320
-    },
-    {
-      "epoch": 15.291264667535854,
-      "grad_norm": 1.4927351474761963,
-      "learning_rate": 7.104422110552764e-05,
-      "loss": 5.3964,
-      "step": 29321
-    },
-    {
-      "epoch": 15.291786179921774,
-      "grad_norm": 1.4840331077575684,
-      "learning_rate": 7.1043216080402e-05,
-      "loss": 5.2647,
-      "step": 29322
-    },
-    {
-      "epoch": 15.292307692307693,
-      "grad_norm": 1.4876492023468018,
-      "learning_rate": 7.104221105527638e-05,
-      "loss": 4.6135,
-      "step": 29323
-    },
-    {
-      "epoch": 15.29282920469361,
-      "grad_norm": 1.4561541080474854,
-      "learning_rate": 7.104120603015076e-05,
-      "loss": 4.9457,
-      "step": 29324
-    },
-    {
-      "epoch": 15.29335071707953,
-      "grad_norm": 1.3925868272781372,
-      "learning_rate": 7.104020100502514e-05,
-      "loss": 5.7239,
-      "step": 29325
-    },
-    {
-      "epoch": 15.29387222946545,
-      "grad_norm": 1.5435882806777954,
-      "learning_rate": 7.10391959798995e-05,
-      "loss": 5.6045,
-      "step": 29326
-    },
-    {
-      "epoch": 15.29439374185137,
-      "grad_norm": 1.5216354131698608,
-      "learning_rate": 7.103819095477388e-05,
-      "loss": 5.4493,
-      "step": 29327
-    },
-    {
-      "epoch": 15.294915254237289,
-      "grad_norm": 1.5252598524093628,
-      "learning_rate": 7.103718592964824e-05,
-      "loss": 5.538,
-      "step": 29328
-    },
-    {
-      "epoch": 15.295436766623208,
-      "grad_norm": 1.574095368385315,
-      "learning_rate": 7.103618090452262e-05,
-      "loss": 5.2045,
-      "step": 29329
-    },
-    {
-      "epoch": 15.295958279009126,
-      "grad_norm": 1.3456166982650757,
-      "learning_rate": 7.103517587939698e-05,
-      "loss": 5.6847,
-      "step": 29330
-    },
-    {
-      "epoch": 15.296479791395045,
-      "grad_norm": 1.518509030342102,
-      "learning_rate": 7.103417085427136e-05,
-      "loss": 5.47,
-      "step": 29331
-    },
-    {
-      "epoch": 15.297001303780965,
-      "grad_norm": 1.3422966003417969,
-      "learning_rate": 7.103316582914573e-05,
-      "loss": 5.8555,
-      "step": 29332
-    },
-    {
-      "epoch": 15.297522816166884,
-      "grad_norm": 1.498883605003357,
-      "learning_rate": 7.10321608040201e-05,
-      "loss": 5.6202,
-      "step": 29333
-    },
-    {
-      "epoch": 15.298044328552804,
-      "grad_norm": 1.4869309663772583,
-      "learning_rate": 7.103115577889448e-05,
-      "loss": 5.2633,
-      "step": 29334
-    },
-    {
-      "epoch": 15.298565840938723,
-      "grad_norm": 1.427056908607483,
-      "learning_rate": 7.103015075376885e-05,
-      "loss": 5.3598,
-      "step": 29335
-    },
-    {
-      "epoch": 15.29908735332464,
-      "grad_norm": 1.4353587627410889,
-      "learning_rate": 7.102914572864322e-05,
-      "loss": 4.7462,
-      "step": 29336
-    },
-    {
-      "epoch": 15.29960886571056,
-      "grad_norm": 1.4003654718399048,
-      "learning_rate": 7.102814070351759e-05,
-      "loss": 5.5436,
-      "step": 29337
-    },
-    {
-      "epoch": 15.30013037809648,
-      "grad_norm": 1.4537805318832397,
-      "learning_rate": 7.102713567839197e-05,
-      "loss": 5.4403,
-      "step": 29338
-    },
-    {
-      "epoch": 15.3006518904824,
-      "grad_norm": 1.5028003454208374,
-      "learning_rate": 7.102613065326633e-05,
-      "loss": 5.3887,
-      "step": 29339
-    },
-    {
-      "epoch": 15.301173402868319,
-      "grad_norm": 1.4674171209335327,
-      "learning_rate": 7.102512562814071e-05,
-      "loss": 5.395,
-      "step": 29340
-    },
-    {
-      "epoch": 15.301694915254238,
-      "grad_norm": 1.4670584201812744,
-      "learning_rate": 7.102412060301507e-05,
-      "loss": 5.1683,
-      "step": 29341
-    },
-    {
-      "epoch": 15.302216427640156,
-      "grad_norm": 1.4277158975601196,
-      "learning_rate": 7.102311557788945e-05,
-      "loss": 5.5403,
-      "step": 29342
-    },
-    {
-      "epoch": 15.302737940026075,
-      "grad_norm": 1.4988266229629517,
-      "learning_rate": 7.102211055276381e-05,
-      "loss": 4.6656,
-      "step": 29343
-    },
-    {
-      "epoch": 15.303259452411995,
-      "grad_norm": 1.4049707651138306,
-      "learning_rate": 7.102110552763819e-05,
-      "loss": 5.5765,
-      "step": 29344
-    },
-    {
-      "epoch": 15.303780964797914,
-      "grad_norm": 1.4259719848632812,
-      "learning_rate": 7.102010050251257e-05,
-      "loss": 5.631,
-      "step": 29345
-    },
-    {
-      "epoch": 15.304302477183834,
-      "grad_norm": 1.3259862661361694,
-      "learning_rate": 7.101909547738695e-05,
-      "loss": 5.6112,
-      "step": 29346
-    },
-    {
-      "epoch": 15.304823989569753,
-      "grad_norm": 1.312058448791504,
-      "learning_rate": 7.101809045226131e-05,
-      "loss": 5.6319,
-      "step": 29347
-    },
-    {
-      "epoch": 15.30534550195567,
-      "grad_norm": 1.5411399602890015,
-      "learning_rate": 7.101708542713568e-05,
-      "loss": 4.9062,
-      "step": 29348
-    },
-    {
-      "epoch": 15.30586701434159,
-      "grad_norm": 1.4422537088394165,
-      "learning_rate": 7.101608040201005e-05,
-      "loss": 5.0459,
-      "step": 29349
-    },
-    {
-      "epoch": 15.30638852672751,
-      "grad_norm": 1.5396050214767456,
-      "learning_rate": 7.101507537688442e-05,
-      "loss": 5.1656,
-      "step": 29350
-    },
-    {
-      "epoch": 15.30691003911343,
-      "grad_norm": 1.4763191938400269,
-      "learning_rate": 7.10140703517588e-05,
-      "loss": 5.4675,
-      "step": 29351
-    },
-    {
-      "epoch": 15.307431551499349,
-      "grad_norm": 1.3536689281463623,
-      "learning_rate": 7.101306532663316e-05,
-      "loss": 5.5382,
-      "step": 29352
-    },
-    {
-      "epoch": 15.307953063885268,
-      "grad_norm": 1.4902926683425903,
-      "learning_rate": 7.101206030150754e-05,
-      "loss": 5.4117,
-      "step": 29353
-    },
-    {
-      "epoch": 15.308474576271186,
-      "grad_norm": 1.519805669784546,
-      "learning_rate": 7.101105527638192e-05,
-      "loss": 5.5216,
-      "step": 29354
-    },
-    {
-      "epoch": 15.308996088657105,
-      "grad_norm": 1.430617094039917,
-      "learning_rate": 7.10100502512563e-05,
-      "loss": 5.5178,
-      "step": 29355
-    },
-    {
-      "epoch": 15.309517601043025,
-      "grad_norm": 1.474892020225525,
-      "learning_rate": 7.100904522613066e-05,
-      "loss": 5.4783,
-      "step": 29356
-    },
-    {
-      "epoch": 15.310039113428944,
-      "grad_norm": 1.5215966701507568,
-      "learning_rate": 7.100804020100504e-05,
-      "loss": 5.1554,
-      "step": 29357
-    },
-    {
-      "epoch": 15.310560625814864,
-      "grad_norm": 1.5608253479003906,
-      "learning_rate": 7.10070351758794e-05,
-      "loss": 4.9089,
-      "step": 29358
-    },
-    {
-      "epoch": 15.311082138200783,
-      "grad_norm": 1.5770255327224731,
-      "learning_rate": 7.100603015075378e-05,
-      "loss": 5.5241,
-      "step": 29359
-    },
-    {
-      "epoch": 15.3116036505867,
-      "grad_norm": 1.5551069974899292,
-      "learning_rate": 7.100502512562814e-05,
-      "loss": 5.5673,
-      "step": 29360
-    },
-    {
-      "epoch": 15.31212516297262,
-      "grad_norm": 1.4639376401901245,
-      "learning_rate": 7.100402010050251e-05,
-      "loss": 5.3974,
-      "step": 29361
-    },
-    {
-      "epoch": 15.31264667535854,
-      "grad_norm": 1.6201626062393188,
-      "learning_rate": 7.100301507537688e-05,
-      "loss": 4.8468,
-      "step": 29362
-    },
-    {
-      "epoch": 15.31316818774446,
-      "grad_norm": 1.4381189346313477,
-      "learning_rate": 7.100201005025125e-05,
-      "loss": 5.3553,
-      "step": 29363
-    },
-    {
-      "epoch": 15.313689700130379,
-      "grad_norm": 1.5057883262634277,
-      "learning_rate": 7.100100502512563e-05,
-      "loss": 5.2718,
-      "step": 29364
-    },
-    {
-      "epoch": 15.314211212516298,
-      "grad_norm": 1.3808417320251465,
-      "learning_rate": 7.1e-05,
-      "loss": 4.8633,
-      "step": 29365
-    },
-    {
-      "epoch": 15.314732724902216,
-      "grad_norm": 1.4758027791976929,
-      "learning_rate": 7.099899497487438e-05,
-      "loss": 5.5415,
-      "step": 29366
-    },
-    {
-      "epoch": 15.315254237288135,
-      "grad_norm": 1.6319524049758911,
-      "learning_rate": 7.099798994974875e-05,
-      "loss": 4.9194,
-      "step": 29367
-    },
-    {
-      "epoch": 15.315775749674055,
-      "grad_norm": 1.4908320903778076,
-      "learning_rate": 7.099698492462312e-05,
-      "loss": 5.2264,
-      "step": 29368
-    },
-    {
-      "epoch": 15.316297262059974,
-      "grad_norm": 1.5670061111450195,
-      "learning_rate": 7.099597989949749e-05,
-      "loss": 4.7307,
-      "step": 29369
-    },
-    {
-      "epoch": 15.316818774445894,
-      "grad_norm": 1.4453691244125366,
-      "learning_rate": 7.099497487437187e-05,
-      "loss": 5.5637,
-      "step": 29370
-    },
-    {
-      "epoch": 15.317340286831811,
-      "grad_norm": 1.3977118730545044,
-      "learning_rate": 7.099396984924623e-05,
-      "loss": 5.7228,
-      "step": 29371
-    },
-    {
-      "epoch": 15.31786179921773,
-      "grad_norm": 1.3598848581314087,
-      "learning_rate": 7.099296482412061e-05,
-      "loss": 5.5986,
-      "step": 29372
-    },
-    {
-      "epoch": 15.31838331160365,
-      "grad_norm": 1.4755065441131592,
-      "learning_rate": 7.099195979899497e-05,
-      "loss": 5.3857,
-      "step": 29373
-    },
-    {
-      "epoch": 15.31890482398957,
-      "grad_norm": 1.4938299655914307,
-      "learning_rate": 7.099095477386935e-05,
-      "loss": 5.4077,
-      "step": 29374
-    },
-    {
-      "epoch": 15.31942633637549,
-      "grad_norm": 1.4962185621261597,
-      "learning_rate": 7.098994974874373e-05,
-      "loss": 5.0862,
-      "step": 29375
-    },
-    {
-      "epoch": 15.319947848761409,
-      "grad_norm": 1.5540679693222046,
-      "learning_rate": 7.098894472361809e-05,
-      "loss": 5.2589,
-      "step": 29376
-    },
-    {
-      "epoch": 15.320469361147328,
-      "grad_norm": 1.4976190328598022,
-      "learning_rate": 7.098793969849247e-05,
-      "loss": 5.4112,
-      "step": 29377
-    },
-    {
-      "epoch": 15.320990873533246,
-      "grad_norm": 1.4402002096176147,
-      "learning_rate": 7.098693467336684e-05,
-      "loss": 5.5369,
-      "step": 29378
-    },
-    {
-      "epoch": 15.321512385919165,
-      "grad_norm": 1.4116286039352417,
-      "learning_rate": 7.098592964824121e-05,
-      "loss": 5.5559,
-      "step": 29379
-    },
-    {
-      "epoch": 15.322033898305085,
-      "grad_norm": 1.476027250289917,
-      "learning_rate": 7.098492462311558e-05,
-      "loss": 5.496,
-      "step": 29380
-    },
-    {
-      "epoch": 15.322555410691004,
-      "grad_norm": 1.592615008354187,
-      "learning_rate": 7.098391959798996e-05,
-      "loss": 5.5799,
-      "step": 29381
-    },
-    {
-      "epoch": 15.323076923076924,
-      "grad_norm": 1.4436957836151123,
-      "learning_rate": 7.098291457286432e-05,
-      "loss": 5.7652,
-      "step": 29382
-    },
-    {
-      "epoch": 15.323598435462841,
-      "grad_norm": 1.5761618614196777,
-      "learning_rate": 7.09819095477387e-05,
-      "loss": 5.1641,
-      "step": 29383
-    },
-    {
-      "epoch": 15.32411994784876,
-      "grad_norm": 1.5226973295211792,
-      "learning_rate": 7.098090452261306e-05,
-      "loss": 5.04,
-      "step": 29384
-    },
-    {
-      "epoch": 15.32464146023468,
-      "grad_norm": 1.4692763090133667,
-      "learning_rate": 7.097989949748744e-05,
-      "loss": 4.9619,
-      "step": 29385
-    },
-    {
-      "epoch": 15.3251629726206,
-      "grad_norm": 1.6267428398132324,
-      "learning_rate": 7.097889447236182e-05,
-      "loss": 4.735,
-      "step": 29386
-    },
-    {
-      "epoch": 15.32568448500652,
-      "grad_norm": 1.5578140020370483,
-      "learning_rate": 7.097788944723618e-05,
-      "loss": 4.8068,
-      "step": 29387
-    },
-    {
-      "epoch": 15.326205997392439,
-      "grad_norm": 1.556387186050415,
-      "learning_rate": 7.097688442211056e-05,
-      "loss": 5.2255,
-      "step": 29388
-    },
-    {
-      "epoch": 15.326727509778356,
-      "grad_norm": 1.435271143913269,
-      "learning_rate": 7.097587939698492e-05,
-      "loss": 5.5468,
-      "step": 29389
-    },
-    {
-      "epoch": 15.327249022164276,
-      "grad_norm": 1.4457241296768188,
-      "learning_rate": 7.09748743718593e-05,
-      "loss": 4.9683,
-      "step": 29390
-    },
-    {
-      "epoch": 15.327770534550195,
-      "grad_norm": 1.4462552070617676,
-      "learning_rate": 7.097386934673367e-05,
-      "loss": 5.5302,
-      "step": 29391
-    },
-    {
-      "epoch": 15.328292046936115,
-      "grad_norm": 1.515506386756897,
-      "learning_rate": 7.097286432160804e-05,
-      "loss": 5.4462,
-      "step": 29392
-    },
-    {
-      "epoch": 15.328813559322034,
-      "grad_norm": 1.5399415493011475,
-      "learning_rate": 7.097185929648241e-05,
-      "loss": 5.257,
-      "step": 29393
-    },
-    {
-      "epoch": 15.329335071707954,
-      "grad_norm": 1.425667405128479,
-      "learning_rate": 7.097085427135679e-05,
-      "loss": 5.3426,
-      "step": 29394
-    },
-    {
-      "epoch": 15.329856584093871,
-      "grad_norm": 1.4208918809890747,
-      "learning_rate": 7.096984924623116e-05,
-      "loss": 5.7774,
-      "step": 29395
-    },
-    {
-      "epoch": 15.33037809647979,
-      "grad_norm": 1.4325352907180786,
-      "learning_rate": 7.096884422110554e-05,
-      "loss": 5.5942,
-      "step": 29396
-    },
-    {
-      "epoch": 15.33089960886571,
-      "grad_norm": 1.6812721490859985,
-      "learning_rate": 7.09678391959799e-05,
-      "loss": 4.2323,
-      "step": 29397
-    },
-    {
-      "epoch": 15.33142112125163,
-      "grad_norm": 1.6203092336654663,
-      "learning_rate": 7.096683417085428e-05,
-      "loss": 4.9308,
-      "step": 29398
-    },
-    {
-      "epoch": 15.33194263363755,
-      "grad_norm": 1.4274495840072632,
-      "learning_rate": 7.096582914572865e-05,
-      "loss": 5.2838,
-      "step": 29399
-    },
-    {
-      "epoch": 15.332464146023469,
-      "grad_norm": 1.3630969524383545,
-      "learning_rate": 7.096482412060303e-05,
-      "loss": 5.3499,
-      "step": 29400
-    },
-    {
-      "epoch": 15.332985658409386,
-      "grad_norm": 1.5411399602890015,
-      "learning_rate": 7.096381909547739e-05,
-      "loss": 5.1631,
-      "step": 29401
-    },
-    {
-      "epoch": 15.333507170795306,
-      "grad_norm": 1.5021252632141113,
-      "learning_rate": 7.096281407035175e-05,
-      "loss": 5.6213,
-      "step": 29402
-    },
-    {
-      "epoch": 15.334028683181225,
-      "grad_norm": 1.569925308227539,
-      "learning_rate": 7.096180904522613e-05,
-      "loss": 5.0738,
-      "step": 29403
-    },
-    {
-      "epoch": 15.334550195567145,
-      "grad_norm": 1.5451292991638184,
-      "learning_rate": 7.09608040201005e-05,
-      "loss": 4.8355,
-      "step": 29404
-    },
-    {
-      "epoch": 15.335071707953064,
-      "grad_norm": 1.6290525197982788,
-      "learning_rate": 7.095979899497487e-05,
-      "loss": 5.388,
-      "step": 29405
-    },
-    {
-      "epoch": 15.335593220338984,
-      "grad_norm": 1.5279568433761597,
-      "learning_rate": 7.095879396984925e-05,
-      "loss": 4.5987,
-      "step": 29406
-    },
-    {
-      "epoch": 15.336114732724901,
-      "grad_norm": 1.4263486862182617,
-      "learning_rate": 7.095778894472363e-05,
-      "loss": 5.3906,
-      "step": 29407
-    },
-    {
-      "epoch": 15.336636245110821,
-      "grad_norm": 1.6105337142944336,
-      "learning_rate": 7.0956783919598e-05,
-      "loss": 5.1548,
-      "step": 29408
-    },
-    {
-      "epoch": 15.33715775749674,
-      "grad_norm": 1.576453447341919,
-      "learning_rate": 7.095577889447237e-05,
-      "loss": 4.8734,
-      "step": 29409
-    },
-    {
-      "epoch": 15.33767926988266,
-      "grad_norm": 1.4381731748580933,
-      "learning_rate": 7.095477386934674e-05,
-      "loss": 5.5272,
-      "step": 29410
-    },
-    {
-      "epoch": 15.33820078226858,
-      "grad_norm": 1.5227190256118774,
-      "learning_rate": 7.095376884422111e-05,
-      "loss": 5.0092,
-      "step": 29411
-    },
-    {
-      "epoch": 15.338722294654499,
-      "grad_norm": 1.5497066974639893,
-      "learning_rate": 7.095276381909548e-05,
-      "loss": 4.7989,
-      "step": 29412
-    },
-    {
-      "epoch": 15.339243807040416,
-      "grad_norm": 1.4424240589141846,
-      "learning_rate": 7.095175879396986e-05,
-      "loss": 5.4107,
-      "step": 29413
-    },
-    {
-      "epoch": 15.339765319426336,
-      "grad_norm": 1.4284729957580566,
-      "learning_rate": 7.095075376884422e-05,
-      "loss": 5.6906,
-      "step": 29414
-    },
-    {
-      "epoch": 15.340286831812255,
-      "grad_norm": 1.5288091897964478,
-      "learning_rate": 7.09497487437186e-05,
-      "loss": 5.656,
-      "step": 29415
-    },
-    {
-      "epoch": 15.340808344198175,
-      "grad_norm": 1.439754843711853,
-      "learning_rate": 7.094874371859298e-05,
-      "loss": 5.3044,
-      "step": 29416
-    },
-    {
-      "epoch": 15.341329856584094,
-      "grad_norm": 1.4828888177871704,
-      "learning_rate": 7.094773869346734e-05,
-      "loss": 5.6407,
-      "step": 29417
-    },
-    {
-      "epoch": 15.341851368970014,
-      "grad_norm": 1.53744637966156,
-      "learning_rate": 7.094673366834172e-05,
-      "loss": 5.568,
-      "step": 29418
-    },
-    {
-      "epoch": 15.342372881355931,
-      "grad_norm": 1.5117498636245728,
-      "learning_rate": 7.094572864321608e-05,
-      "loss": 4.6729,
-      "step": 29419
-    },
-    {
-      "epoch": 15.342894393741851,
-      "grad_norm": 1.4338788986206055,
-      "learning_rate": 7.094472361809046e-05,
-      "loss": 5.709,
-      "step": 29420
-    },
-    {
-      "epoch": 15.34341590612777,
-      "grad_norm": 1.5683404207229614,
-      "learning_rate": 7.094371859296482e-05,
-      "loss": 5.3939,
-      "step": 29421
-    },
-    {
-      "epoch": 15.34393741851369,
-      "grad_norm": 1.5334486961364746,
-      "learning_rate": 7.09427135678392e-05,
-      "loss": 4.9948,
-      "step": 29422
-    },
-    {
-      "epoch": 15.34445893089961,
-      "grad_norm": 1.4369977712631226,
-      "learning_rate": 7.094170854271357e-05,
-      "loss": 5.3662,
-      "step": 29423
-    },
-    {
-      "epoch": 15.344980443285529,
-      "grad_norm": 1.4915363788604736,
-      "learning_rate": 7.094070351758794e-05,
-      "loss": 4.7705,
-      "step": 29424
-    },
-    {
-      "epoch": 15.345501955671446,
-      "grad_norm": 1.5413792133331299,
-      "learning_rate": 7.093969849246231e-05,
-      "loss": 5.4536,
-      "step": 29425
-    },
-    {
-      "epoch": 15.346023468057366,
-      "grad_norm": 1.4257376194000244,
-      "learning_rate": 7.093869346733669e-05,
-      "loss": 5.6482,
-      "step": 29426
-    },
-    {
-      "epoch": 15.346544980443285,
-      "grad_norm": 1.4750468730926514,
-      "learning_rate": 7.093768844221106e-05,
-      "loss": 5.6041,
-      "step": 29427
-    },
-    {
-      "epoch": 15.347066492829205,
-      "grad_norm": 1.4488298892974854,
-      "learning_rate": 7.093668341708543e-05,
-      "loss": 5.5926,
-      "step": 29428
-    },
-    {
-      "epoch": 15.347588005215124,
-      "grad_norm": 1.466354489326477,
-      "learning_rate": 7.09356783919598e-05,
-      "loss": 5.3742,
-      "step": 29429
-    },
-    {
-      "epoch": 15.348109517601044,
-      "grad_norm": 1.3946584463119507,
-      "learning_rate": 7.093467336683417e-05,
-      "loss": 5.7022,
-      "step": 29430
-    },
-    {
-      "epoch": 15.348631029986961,
-      "grad_norm": 1.294505000114441,
-      "learning_rate": 7.093366834170855e-05,
-      "loss": 5.6307,
-      "step": 29431
-    },
-    {
-      "epoch": 15.349152542372881,
-      "grad_norm": 1.5080928802490234,
-      "learning_rate": 7.093266331658291e-05,
-      "loss": 5.2056,
-      "step": 29432
-    },
-    {
-      "epoch": 15.3496740547588,
-      "grad_norm": 1.5540968179702759,
-      "learning_rate": 7.093165829145729e-05,
-      "loss": 5.2871,
-      "step": 29433
-    },
-    {
-      "epoch": 15.35019556714472,
-      "grad_norm": 1.5119614601135254,
-      "learning_rate": 7.093065326633165e-05,
-      "loss": 5.4781,
-      "step": 29434
-    },
-    {
-      "epoch": 15.35071707953064,
-      "grad_norm": 1.5070730447769165,
-      "learning_rate": 7.092964824120603e-05,
-      "loss": 4.2775,
-      "step": 29435
-    },
-    {
-      "epoch": 15.351238591916559,
-      "grad_norm": 1.5155965089797974,
-      "learning_rate": 7.092864321608041e-05,
-      "loss": 5.4793,
-      "step": 29436
-    },
-    {
-      "epoch": 15.351760104302477,
-      "grad_norm": 1.511824369430542,
-      "learning_rate": 7.092763819095479e-05,
-      "loss": 5.1527,
-      "step": 29437
-    },
-    {
-      "epoch": 15.352281616688396,
-      "grad_norm": 1.510451078414917,
-      "learning_rate": 7.092663316582915e-05,
-      "loss": 5.2706,
-      "step": 29438
-    },
-    {
-      "epoch": 15.352803129074315,
-      "grad_norm": 1.3575663566589355,
-      "learning_rate": 7.092562814070353e-05,
-      "loss": 5.2428,
-      "step": 29439
-    },
-    {
-      "epoch": 15.353324641460235,
-      "grad_norm": 1.5115175247192383,
-      "learning_rate": 7.09246231155779e-05,
-      "loss": 5.4968,
-      "step": 29440
-    },
-    {
-      "epoch": 15.353846153846154,
-      "grad_norm": 1.4243559837341309,
-      "learning_rate": 7.092361809045226e-05,
-      "loss": 4.8828,
-      "step": 29441
-    },
-    {
-      "epoch": 15.354367666232074,
-      "grad_norm": 1.4339983463287354,
-      "learning_rate": 7.092261306532664e-05,
-      "loss": 5.1558,
-      "step": 29442
-    },
-    {
-      "epoch": 15.354889178617992,
-      "grad_norm": 1.469846248626709,
-      "learning_rate": 7.0921608040201e-05,
-      "loss": 5.2444,
-      "step": 29443
-    },
-    {
-      "epoch": 15.355410691003911,
-      "grad_norm": 1.359991431236267,
-      "learning_rate": 7.092060301507538e-05,
-      "loss": 5.4705,
-      "step": 29444
-    },
-    {
-      "epoch": 15.35593220338983,
-      "grad_norm": 1.5310231447219849,
-      "learning_rate": 7.091959798994974e-05,
-      "loss": 5.3698,
-      "step": 29445
-    },
-    {
-      "epoch": 15.35645371577575,
-      "grad_norm": 1.5244665145874023,
-      "learning_rate": 7.091859296482412e-05,
-      "loss": 5.6668,
-      "step": 29446
-    },
-    {
-      "epoch": 15.35697522816167,
-      "grad_norm": 1.416869044303894,
-      "learning_rate": 7.09175879396985e-05,
-      "loss": 5.3848,
-      "step": 29447
-    },
-    {
-      "epoch": 15.357496740547589,
-      "grad_norm": 1.531835675239563,
-      "learning_rate": 7.091658291457288e-05,
-      "loss": 4.8804,
-      "step": 29448
-    },
-    {
-      "epoch": 15.358018252933507,
-      "grad_norm": 1.4638569355010986,
-      "learning_rate": 7.091557788944724e-05,
-      "loss": 5.3677,
-      "step": 29449
-    },
-    {
-      "epoch": 15.358539765319426,
-      "grad_norm": 1.4765348434448242,
-      "learning_rate": 7.091457286432162e-05,
-      "loss": 5.5592,
-      "step": 29450
-    },
-    {
-      "epoch": 15.359061277705345,
-      "grad_norm": 1.5918322801589966,
-      "learning_rate": 7.091356783919598e-05,
-      "loss": 5.231,
-      "step": 29451
-    },
-    {
-      "epoch": 15.359582790091265,
-      "grad_norm": 1.5324763059616089,
-      "learning_rate": 7.091256281407036e-05,
-      "loss": 5.5651,
-      "step": 29452
-    },
-    {
-      "epoch": 15.360104302477184,
-      "grad_norm": 1.5340152978897095,
-      "learning_rate": 7.091155778894472e-05,
-      "loss": 5.2571,
-      "step": 29453
-    },
-    {
-      "epoch": 15.360625814863104,
-      "grad_norm": 1.4719208478927612,
-      "learning_rate": 7.091055276381909e-05,
-      "loss": 5.5596,
-      "step": 29454
-    },
-    {
-      "epoch": 15.361147327249022,
-      "grad_norm": 1.5601240396499634,
-      "learning_rate": 7.090954773869347e-05,
-      "loss": 4.797,
-      "step": 29455
-    },
-    {
-      "epoch": 15.361668839634941,
-      "grad_norm": 1.4410814046859741,
-      "learning_rate": 7.090854271356784e-05,
-      "loss": 5.6716,
-      "step": 29456
-    },
-    {
-      "epoch": 15.36219035202086,
-      "grad_norm": 1.5836313962936401,
-      "learning_rate": 7.090753768844222e-05,
-      "loss": 5.3112,
-      "step": 29457
-    },
-    {
-      "epoch": 15.36271186440678,
-      "grad_norm": 1.7886006832122803,
-      "learning_rate": 7.090653266331659e-05,
-      "loss": 5.1771,
-      "step": 29458
-    },
-    {
-      "epoch": 15.3632333767927,
-      "grad_norm": 1.5515748262405396,
-      "learning_rate": 7.090552763819096e-05,
-      "loss": 5.0074,
-      "step": 29459
-    },
-    {
-      "epoch": 15.363754889178619,
-      "grad_norm": 1.468180775642395,
-      "learning_rate": 7.090452261306533e-05,
-      "loss": 5.6685,
-      "step": 29460
-    },
-    {
-      "epoch": 15.364276401564537,
-      "grad_norm": 1.6270809173583984,
-      "learning_rate": 7.090351758793971e-05,
-      "loss": 4.7666,
-      "step": 29461
-    },
-    {
-      "epoch": 15.364797913950456,
-      "grad_norm": 1.5983784198760986,
-      "learning_rate": 7.090251256281407e-05,
-      "loss": 5.5874,
-      "step": 29462
-    },
-    {
-      "epoch": 15.365319426336375,
-      "grad_norm": 1.4494624137878418,
-      "learning_rate": 7.090150753768845e-05,
-      "loss": 5.4473,
-      "step": 29463
-    },
-    {
-      "epoch": 15.365840938722295,
-      "grad_norm": 1.581477165222168,
-      "learning_rate": 7.090050251256281e-05,
-      "loss": 5.1623,
-      "step": 29464
-    },
-    {
-      "epoch": 15.366362451108214,
-      "grad_norm": 1.3875422477722168,
-      "learning_rate": 7.089949748743719e-05,
-      "loss": 5.4901,
-      "step": 29465
-    },
-    {
-      "epoch": 15.366883963494132,
-      "grad_norm": 1.515095829963684,
-      "learning_rate": 7.089849246231156e-05,
-      "loss": 4.9265,
-      "step": 29466
-    },
-    {
-      "epoch": 15.367405475880052,
-      "grad_norm": 1.4531443119049072,
-      "learning_rate": 7.089748743718593e-05,
-      "loss": 5.3408,
-      "step": 29467
-    },
-    {
-      "epoch": 15.367926988265971,
-      "grad_norm": 1.4652445316314697,
-      "learning_rate": 7.089648241206031e-05,
-      "loss": 5.2715,
-      "step": 29468
-    },
-    {
-      "epoch": 15.36844850065189,
-      "grad_norm": 1.3688033819198608,
-      "learning_rate": 7.089547738693468e-05,
-      "loss": 5.6096,
-      "step": 29469
-    },
-    {
-      "epoch": 15.36897001303781,
-      "grad_norm": 1.6245063543319702,
-      "learning_rate": 7.089447236180905e-05,
-      "loss": 5.2802,
-      "step": 29470
-    },
-    {
-      "epoch": 15.36949152542373,
-      "grad_norm": 1.4012175798416138,
-      "learning_rate": 7.089346733668342e-05,
-      "loss": 5.5854,
-      "step": 29471
-    },
-    {
-      "epoch": 15.370013037809649,
-      "grad_norm": 1.5007888078689575,
-      "learning_rate": 7.08924623115578e-05,
-      "loss": 5.5889,
-      "step": 29472
-    },
-    {
-      "epoch": 15.370534550195567,
-      "grad_norm": 1.3954178094863892,
-      "learning_rate": 7.089145728643216e-05,
-      "loss": 5.6065,
-      "step": 29473
-    },
-    {
-      "epoch": 15.371056062581486,
-      "grad_norm": 1.5731464624404907,
-      "learning_rate": 7.089045226130654e-05,
-      "loss": 4.8426,
-      "step": 29474
-    },
-    {
-      "epoch": 15.371577574967406,
-      "grad_norm": 1.4906718730926514,
-      "learning_rate": 7.08894472361809e-05,
-      "loss": 5.0653,
-      "step": 29475
-    },
-    {
-      "epoch": 15.372099087353325,
-      "grad_norm": 1.4771842956542969,
-      "learning_rate": 7.088844221105528e-05,
-      "loss": 5.7005,
-      "step": 29476
-    },
-    {
-      "epoch": 15.372620599739244,
-      "grad_norm": 1.5862138271331787,
-      "learning_rate": 7.088743718592964e-05,
-      "loss": 5.4729,
-      "step": 29477
-    },
-    {
-      "epoch": 15.373142112125162,
-      "grad_norm": 1.4941807985305786,
-      "learning_rate": 7.088643216080402e-05,
-      "loss": 5.4593,
-      "step": 29478
-    },
-    {
-      "epoch": 15.373663624511082,
-      "grad_norm": 1.342432975769043,
-      "learning_rate": 7.08854271356784e-05,
-      "loss": 5.3999,
-      "step": 29479
-    },
-    {
-      "epoch": 15.374185136897001,
-      "grad_norm": 1.413649320602417,
-      "learning_rate": 7.088442211055276e-05,
-      "loss": 5.019,
-      "step": 29480
-    },
-    {
-      "epoch": 15.37470664928292,
-      "grad_norm": 1.5225557088851929,
-      "learning_rate": 7.088341708542714e-05,
-      "loss": 5.3928,
-      "step": 29481
-    },
-    {
-      "epoch": 15.37522816166884,
-      "grad_norm": 1.5688589811325073,
-      "learning_rate": 7.08824120603015e-05,
-      "loss": 5.6442,
-      "step": 29482
-    },
-    {
-      "epoch": 15.37574967405476,
-      "grad_norm": 1.3705451488494873,
-      "learning_rate": 7.088140703517588e-05,
-      "loss": 5.4481,
-      "step": 29483
-    },
-    {
-      "epoch": 15.376271186440677,
-      "grad_norm": 1.377962350845337,
-      "learning_rate": 7.088040201005025e-05,
-      "loss": 5.1338,
-      "step": 29484
-    },
-    {
-      "epoch": 15.376792698826597,
-      "grad_norm": 1.548869252204895,
-      "learning_rate": 7.087939698492463e-05,
-      "loss": 5.1079,
-      "step": 29485
-    },
-    {
-      "epoch": 15.377314211212516,
-      "grad_norm": 1.532320261001587,
-      "learning_rate": 7.087839195979899e-05,
-      "loss": 5.2124,
-      "step": 29486
-    },
-    {
-      "epoch": 15.377835723598436,
-      "grad_norm": 1.5276405811309814,
-      "learning_rate": 7.087738693467337e-05,
-      "loss": 5.1563,
-      "step": 29487
-    },
-    {
-      "epoch": 15.378357235984355,
-      "grad_norm": 1.3906761407852173,
-      "learning_rate": 7.087638190954775e-05,
-      "loss": 5.637,
-      "step": 29488
-    },
-    {
-      "epoch": 15.378878748370274,
-      "grad_norm": 1.4978618621826172,
-      "learning_rate": 7.087537688442212e-05,
-      "loss": 5.6066,
-      "step": 29489
-    },
-    {
-      "epoch": 15.379400260756192,
-      "grad_norm": 1.417959213256836,
-      "learning_rate": 7.087437185929649e-05,
-      "loss": 5.6684,
-      "step": 29490
-    },
-    {
-      "epoch": 15.379921773142112,
-      "grad_norm": 1.450194001197815,
-      "learning_rate": 7.087336683417087e-05,
-      "loss": 5.7553,
-      "step": 29491
-    },
-    {
-      "epoch": 15.380443285528031,
-      "grad_norm": 1.4563955068588257,
-      "learning_rate": 7.087236180904523e-05,
-      "loss": 5.0688,
-      "step": 29492
-    },
-    {
-      "epoch": 15.38096479791395,
-      "grad_norm": 1.469448447227478,
-      "learning_rate": 7.087135678391961e-05,
-      "loss": 5.5425,
-      "step": 29493
-    },
-    {
-      "epoch": 15.38148631029987,
-      "grad_norm": 2.136993885040283,
-      "learning_rate": 7.087035175879397e-05,
-      "loss": 4.4531,
-      "step": 29494
-    },
-    {
-      "epoch": 15.38200782268579,
-      "grad_norm": 1.4247077703475952,
-      "learning_rate": 7.086934673366834e-05,
-      "loss": 5.5385,
-      "step": 29495
-    },
-    {
-      "epoch": 15.382529335071707,
-      "grad_norm": 1.5677019357681274,
-      "learning_rate": 7.086834170854271e-05,
-      "loss": 4.6737,
-      "step": 29496
-    },
-    {
-      "epoch": 15.383050847457627,
-      "grad_norm": 1.4631104469299316,
-      "learning_rate": 7.086733668341708e-05,
-      "loss": 5.4546,
-      "step": 29497
-    },
-    {
-      "epoch": 15.383572359843546,
-      "grad_norm": 1.4256298542022705,
-      "learning_rate": 7.086633165829146e-05,
-      "loss": 5.4939,
-      "step": 29498
-    },
-    {
-      "epoch": 15.384093872229466,
-      "grad_norm": 1.3910245895385742,
-      "learning_rate": 7.086532663316583e-05,
-      "loss": 5.0762,
-      "step": 29499
-    },
-    {
-      "epoch": 15.384615384615385,
-      "grad_norm": 1.5248229503631592,
-      "learning_rate": 7.086432160804021e-05,
-      "loss": 5.7321,
-      "step": 29500
-    },
-    {
-      "epoch": 15.385136897001304,
-      "grad_norm": 1.443608283996582,
-      "learning_rate": 7.086331658291458e-05,
-      "loss": 5.6693,
-      "step": 29501
-    },
-    {
-      "epoch": 15.385658409387222,
-      "grad_norm": 1.4844220876693726,
-      "learning_rate": 7.086231155778895e-05,
-      "loss": 5.4416,
-      "step": 29502
-    },
-    {
-      "epoch": 15.386179921773142,
-      "grad_norm": 1.5295706987380981,
-      "learning_rate": 7.086130653266332e-05,
-      "loss": 4.7886,
-      "step": 29503
-    },
-    {
-      "epoch": 15.386701434159061,
-      "grad_norm": 1.4381541013717651,
-      "learning_rate": 7.08603015075377e-05,
-      "loss": 5.1235,
-      "step": 29504
-    },
-    {
-      "epoch": 15.38722294654498,
-      "grad_norm": 1.505983829498291,
-      "learning_rate": 7.085929648241206e-05,
-      "loss": 5.24,
-      "step": 29505
-    },
-    {
-      "epoch": 15.3877444589309,
-      "grad_norm": 1.505308985710144,
-      "learning_rate": 7.085829145728644e-05,
-      "loss": 5.0087,
-      "step": 29506
-    },
-    {
-      "epoch": 15.38826597131682,
-      "grad_norm": 1.5623494386672974,
-      "learning_rate": 7.08572864321608e-05,
-      "loss": 5.3989,
-      "step": 29507
-    },
-    {
-      "epoch": 15.388787483702737,
-      "grad_norm": 1.4522091150283813,
-      "learning_rate": 7.085628140703518e-05,
-      "loss": 5.478,
-      "step": 29508
-    },
-    {
-      "epoch": 15.389308996088657,
-      "grad_norm": 1.559999704360962,
-      "learning_rate": 7.085527638190956e-05,
-      "loss": 5.4437,
-      "step": 29509
-    },
-    {
-      "epoch": 15.389830508474576,
-      "grad_norm": 1.4526773691177368,
-      "learning_rate": 7.085427135678392e-05,
-      "loss": 5.3177,
-      "step": 29510
-    },
-    {
-      "epoch": 15.390352020860496,
-      "grad_norm": 1.4847043752670288,
-      "learning_rate": 7.08532663316583e-05,
-      "loss": 5.4479,
-      "step": 29511
-    },
-    {
-      "epoch": 15.390873533246415,
-      "grad_norm": 1.37345290184021,
-      "learning_rate": 7.085226130653266e-05,
-      "loss": 5.6504,
-      "step": 29512
-    },
-    {
-      "epoch": 15.391395045632335,
-      "grad_norm": 1.4435133934020996,
-      "learning_rate": 7.085125628140704e-05,
-      "loss": 5.4697,
-      "step": 29513
-    },
-    {
-      "epoch": 15.391916558018252,
-      "grad_norm": 1.5623966455459595,
-      "learning_rate": 7.08502512562814e-05,
-      "loss": 5.1465,
-      "step": 29514
-    },
-    {
-      "epoch": 15.392438070404172,
-      "grad_norm": 1.6322609186172485,
-      "learning_rate": 7.084924623115578e-05,
-      "loss": 5.4124,
-      "step": 29515
-    },
-    {
-      "epoch": 15.392959582790091,
-      "grad_norm": 1.4241400957107544,
-      "learning_rate": 7.084824120603015e-05,
-      "loss": 5.5326,
-      "step": 29516
-    },
-    {
-      "epoch": 15.39348109517601,
-      "grad_norm": 1.5854367017745972,
-      "learning_rate": 7.084723618090453e-05,
-      "loss": 5.5423,
-      "step": 29517
-    },
-    {
-      "epoch": 15.39400260756193,
-      "grad_norm": 1.5972968339920044,
-      "learning_rate": 7.084623115577889e-05,
-      "loss": 4.9141,
-      "step": 29518
-    },
-    {
-      "epoch": 15.39452411994785,
-      "grad_norm": 1.4286434650421143,
-      "learning_rate": 7.084522613065327e-05,
-      "loss": 5.4271,
-      "step": 29519
-    },
-    {
-      "epoch": 15.395045632333767,
-      "grad_norm": 1.5237102508544922,
-      "learning_rate": 7.084422110552765e-05,
-      "loss": 5.1988,
-      "step": 29520
-    },
-    {
-      "epoch": 15.395567144719687,
-      "grad_norm": 1.4271405935287476,
-      "learning_rate": 7.084321608040201e-05,
-      "loss": 5.3587,
-      "step": 29521
-    },
-    {
-      "epoch": 15.396088657105606,
-      "grad_norm": 1.541608214378357,
-      "learning_rate": 7.084221105527639e-05,
-      "loss": 5.3198,
-      "step": 29522
-    },
-    {
-      "epoch": 15.396610169491526,
-      "grad_norm": 1.5112498998641968,
-      "learning_rate": 7.084120603015075e-05,
-      "loss": 5.0447,
-      "step": 29523
-    },
-    {
-      "epoch": 15.397131681877445,
-      "grad_norm": 1.475095510482788,
-      "learning_rate": 7.084020100502513e-05,
-      "loss": 5.5876,
-      "step": 29524
-    },
-    {
-      "epoch": 15.397653194263365,
-      "grad_norm": 1.4374611377716064,
-      "learning_rate": 7.08391959798995e-05,
-      "loss": 5.3859,
-      "step": 29525
-    },
-    {
-      "epoch": 15.398174706649282,
-      "grad_norm": 1.5243589878082275,
-      "learning_rate": 7.083819095477387e-05,
-      "loss": 4.8675,
-      "step": 29526
-    },
-    {
-      "epoch": 15.398696219035202,
-      "grad_norm": 1.4520447254180908,
-      "learning_rate": 7.083718592964824e-05,
-      "loss": 5.4058,
-      "step": 29527
-    },
-    {
-      "epoch": 15.399217731421121,
-      "grad_norm": 1.484083890914917,
-      "learning_rate": 7.083618090452261e-05,
-      "loss": 4.9759,
-      "step": 29528
-    },
-    {
-      "epoch": 15.39973924380704,
-      "grad_norm": 1.614174485206604,
-      "learning_rate": 7.083517587939699e-05,
-      "loss": 5.0371,
-      "step": 29529
-    },
-    {
-      "epoch": 15.40026075619296,
-      "grad_norm": 1.4630359411239624,
-      "learning_rate": 7.083417085427137e-05,
-      "loss": 5.8009,
-      "step": 29530
-    },
-    {
-      "epoch": 15.40078226857888,
-      "grad_norm": 1.5003620386123657,
-      "learning_rate": 7.083316582914573e-05,
-      "loss": 5.1703,
-      "step": 29531
-    },
-    {
-      "epoch": 15.401303780964797,
-      "grad_norm": 1.421350121498108,
-      "learning_rate": 7.083216080402011e-05,
-      "loss": 4.9362,
-      "step": 29532
-    },
-    {
-      "epoch": 15.401825293350717,
-      "grad_norm": 1.6195141077041626,
-      "learning_rate": 7.083115577889448e-05,
-      "loss": 5.1746,
-      "step": 29533
-    },
-    {
-      "epoch": 15.402346805736636,
-      "grad_norm": 1.432075023651123,
-      "learning_rate": 7.083015075376884e-05,
-      "loss": 5.5218,
-      "step": 29534
-    },
-    {
-      "epoch": 15.402868318122556,
-      "grad_norm": 1.4234788417816162,
-      "learning_rate": 7.082914572864322e-05,
-      "loss": 5.5878,
-      "step": 29535
-    },
-    {
-      "epoch": 15.403389830508475,
-      "grad_norm": 1.3575215339660645,
-      "learning_rate": 7.082814070351758e-05,
-      "loss": 4.8386,
-      "step": 29536
-    },
-    {
-      "epoch": 15.403911342894395,
-      "grad_norm": 1.3796855211257935,
-      "learning_rate": 7.082713567839196e-05,
-      "loss": 5.6088,
-      "step": 29537
-    },
-    {
-      "epoch": 15.404432855280312,
-      "grad_norm": 1.5134105682373047,
-      "learning_rate": 7.082613065326633e-05,
-      "loss": 5.3274,
-      "step": 29538
-    },
-    {
-      "epoch": 15.404954367666232,
-      "grad_norm": 1.4971941709518433,
-      "learning_rate": 7.08251256281407e-05,
-      "loss": 5.7142,
-      "step": 29539
-    },
-    {
-      "epoch": 15.405475880052151,
-      "grad_norm": 1.4265614748001099,
-      "learning_rate": 7.082412060301508e-05,
-      "loss": 5.6567,
-      "step": 29540
-    },
-    {
-      "epoch": 15.40599739243807,
-      "grad_norm": 1.4602537155151367,
-      "learning_rate": 7.082311557788946e-05,
-      "loss": 5.4656,
-      "step": 29541
-    },
-    {
-      "epoch": 15.40651890482399,
-      "grad_norm": 1.5047004222869873,
-      "learning_rate": 7.082211055276382e-05,
-      "loss": 5.5282,
-      "step": 29542
-    },
-    {
-      "epoch": 15.40704041720991,
-      "grad_norm": 1.5932377576828003,
-      "learning_rate": 7.08211055276382e-05,
-      "loss": 4.8816,
-      "step": 29543
-    },
-    {
-      "epoch": 15.407561929595827,
-      "grad_norm": 1.6398977041244507,
-      "learning_rate": 7.082010050251257e-05,
-      "loss": 5.345,
-      "step": 29544
-    },
-    {
-      "epoch": 15.408083441981747,
-      "grad_norm": 1.5432499647140503,
-      "learning_rate": 7.081909547738694e-05,
-      "loss": 5.1098,
-      "step": 29545
-    },
-    {
-      "epoch": 15.408604954367666,
-      "grad_norm": 1.575682282447815,
-      "learning_rate": 7.081809045226131e-05,
-      "loss": 4.7624,
-      "step": 29546
-    },
-    {
-      "epoch": 15.409126466753586,
-      "grad_norm": 1.4707602262496948,
-      "learning_rate": 7.081708542713567e-05,
-      "loss": 5.368,
-      "step": 29547
-    },
-    {
-      "epoch": 15.409647979139505,
-      "grad_norm": 2.2985758781433105,
-      "learning_rate": 7.081608040201005e-05,
-      "loss": 5.2952,
-      "step": 29548
-    },
-    {
-      "epoch": 15.410169491525423,
-      "grad_norm": 1.550214409828186,
-      "learning_rate": 7.081507537688443e-05,
-      "loss": 5.4906,
-      "step": 29549
-    },
-    {
-      "epoch": 15.410691003911342,
-      "grad_norm": 1.5360442399978638,
-      "learning_rate": 7.08140703517588e-05,
-      "loss": 5.3567,
-      "step": 29550
-    },
-    {
-      "epoch": 15.411212516297262,
-      "grad_norm": 1.447292685508728,
-      "learning_rate": 7.081306532663317e-05,
-      "loss": 4.1785,
-      "step": 29551
-    },
-    {
-      "epoch": 15.411734028683181,
-      "grad_norm": 1.448920488357544,
-      "learning_rate": 7.081206030150755e-05,
-      "loss": 5.6326,
-      "step": 29552
-    },
-    {
-      "epoch": 15.4122555410691,
-      "grad_norm": 1.470446228981018,
-      "learning_rate": 7.081105527638191e-05,
-      "loss": 5.1297,
-      "step": 29553
-    },
-    {
-      "epoch": 15.41277705345502,
-      "grad_norm": 1.486828327178955,
-      "learning_rate": 7.081005025125629e-05,
-      "loss": 5.6922,
-      "step": 29554
-    },
-    {
-      "epoch": 15.41329856584094,
-      "grad_norm": 1.4561619758605957,
-      "learning_rate": 7.080904522613065e-05,
-      "loss": 5.7488,
-      "step": 29555
-    },
-    {
-      "epoch": 15.413820078226857,
-      "grad_norm": 1.4942246675491333,
-      "learning_rate": 7.080804020100503e-05,
-      "loss": 5.3823,
-      "step": 29556
-    },
-    {
-      "epoch": 15.414341590612777,
-      "grad_norm": 1.5058679580688477,
-      "learning_rate": 7.08070351758794e-05,
-      "loss": 5.2124,
-      "step": 29557
-    },
-    {
-      "epoch": 15.414863102998696,
-      "grad_norm": 1.4378795623779297,
-      "learning_rate": 7.080603015075377e-05,
-      "loss": 5.2824,
-      "step": 29558
-    },
-    {
-      "epoch": 15.415384615384616,
-      "grad_norm": 1.357006549835205,
-      "learning_rate": 7.080502512562814e-05,
-      "loss": 5.0268,
-      "step": 29559
-    },
-    {
-      "epoch": 15.415906127770535,
-      "grad_norm": 1.548862338066101,
-      "learning_rate": 7.080402010050252e-05,
-      "loss": 5.0198,
-      "step": 29560
-    },
-    {
-      "epoch": 15.416427640156453,
-      "grad_norm": 1.6583184003829956,
-      "learning_rate": 7.08030150753769e-05,
-      "loss": 5.2708,
-      "step": 29561
-    },
-    {
-      "epoch": 15.416949152542372,
-      "grad_norm": 1.3646725416183472,
-      "learning_rate": 7.080201005025126e-05,
-      "loss": 5.5889,
-      "step": 29562
-    },
-    {
-      "epoch": 15.417470664928292,
-      "grad_norm": 1.4311424493789673,
-      "learning_rate": 7.080100502512564e-05,
-      "loss": 5.469,
-      "step": 29563
-    },
-    {
-      "epoch": 15.417992177314211,
-      "grad_norm": 1.4635313749313354,
-      "learning_rate": 7.08e-05,
-      "loss": 5.2255,
-      "step": 29564
-    },
-    {
-      "epoch": 15.41851368970013,
-      "grad_norm": 1.4622552394866943,
-      "learning_rate": 7.079899497487438e-05,
-      "loss": 5.4022,
-      "step": 29565
-    },
-    {
-      "epoch": 15.41903520208605,
-      "grad_norm": 1.351656198501587,
-      "learning_rate": 7.079798994974874e-05,
-      "loss": 5.5965,
-      "step": 29566
-    },
-    {
-      "epoch": 15.419556714471968,
-      "grad_norm": 1.449845790863037,
-      "learning_rate": 7.079698492462312e-05,
-      "loss": 5.2911,
-      "step": 29567
-    },
-    {
-      "epoch": 15.420078226857887,
-      "grad_norm": 1.5032999515533447,
-      "learning_rate": 7.079597989949748e-05,
-      "loss": 5.5233,
-      "step": 29568
-    },
-    {
-      "epoch": 15.420599739243807,
-      "grad_norm": 1.5406436920166016,
-      "learning_rate": 7.079497487437186e-05,
-      "loss": 5.4829,
-      "step": 29569
-    },
-    {
-      "epoch": 15.421121251629726,
-      "grad_norm": 1.4365732669830322,
-      "learning_rate": 7.079396984924624e-05,
-      "loss": 5.5003,
-      "step": 29570
-    },
-    {
-      "epoch": 15.421642764015646,
-      "grad_norm": 1.3835101127624512,
-      "learning_rate": 7.079296482412062e-05,
-      "loss": 5.7901,
-      "step": 29571
-    },
-    {
-      "epoch": 15.422164276401565,
-      "grad_norm": 1.5694085359573364,
-      "learning_rate": 7.079195979899498e-05,
-      "loss": 5.4459,
-      "step": 29572
-    },
-    {
-      "epoch": 15.422685788787483,
-      "grad_norm": 1.45012629032135,
-      "learning_rate": 7.079095477386935e-05,
-      "loss": 5.1625,
-      "step": 29573
-    },
-    {
-      "epoch": 15.423207301173402,
-      "grad_norm": 1.5276774168014526,
-      "learning_rate": 7.078994974874372e-05,
-      "loss": 5.3154,
-      "step": 29574
-    },
-    {
-      "epoch": 15.423728813559322,
-      "grad_norm": 1.3844096660614014,
-      "learning_rate": 7.078894472361809e-05,
-      "loss": 5.5015,
-      "step": 29575
-    },
-    {
-      "epoch": 15.424250325945241,
-      "grad_norm": 1.5624725818634033,
-      "learning_rate": 7.078793969849247e-05,
-      "loss": 5.3282,
-      "step": 29576
-    },
-    {
-      "epoch": 15.42477183833116,
-      "grad_norm": 1.6406066417694092,
-      "learning_rate": 7.078693467336683e-05,
-      "loss": 5.1687,
-      "step": 29577
-    },
-    {
-      "epoch": 15.42529335071708,
-      "grad_norm": 1.4789659976959229,
-      "learning_rate": 7.078592964824121e-05,
-      "loss": 5.5453,
-      "step": 29578
-    },
-    {
-      "epoch": 15.425814863102998,
-      "grad_norm": 1.4823461771011353,
-      "learning_rate": 7.078492462311557e-05,
-      "loss": 5.308,
-      "step": 29579
-    },
-    {
-      "epoch": 15.426336375488917,
-      "grad_norm": 1.4986494779586792,
-      "learning_rate": 7.078391959798995e-05,
-      "loss": 4.8214,
-      "step": 29580
-    },
-    {
-      "epoch": 15.426857887874837,
-      "grad_norm": 1.4134584665298462,
-      "learning_rate": 7.078291457286433e-05,
-      "loss": 5.7069,
-      "step": 29581
-    },
-    {
-      "epoch": 15.427379400260756,
-      "grad_norm": 1.5302245616912842,
-      "learning_rate": 7.07819095477387e-05,
-      "loss": 5.199,
-      "step": 29582
-    },
-    {
-      "epoch": 15.427900912646676,
-      "grad_norm": 1.405515432357788,
-      "learning_rate": 7.078090452261307e-05,
-      "loss": 5.1832,
-      "step": 29583
-    },
-    {
-      "epoch": 15.428422425032595,
-      "grad_norm": 1.3691824674606323,
-      "learning_rate": 7.077989949748745e-05,
-      "loss": 5.2823,
-      "step": 29584
-    },
-    {
-      "epoch": 15.428943937418513,
-      "grad_norm": 1.2872645854949951,
-      "learning_rate": 7.077889447236181e-05,
-      "loss": 4.8932,
-      "step": 29585
-    },
-    {
-      "epoch": 15.429465449804432,
-      "grad_norm": 1.4782060384750366,
-      "learning_rate": 7.077788944723619e-05,
-      "loss": 5.7427,
-      "step": 29586
-    },
-    {
-      "epoch": 15.429986962190352,
-      "grad_norm": 1.4718554019927979,
-      "learning_rate": 7.077688442211055e-05,
-      "loss": 5.6756,
-      "step": 29587
-    },
-    {
-      "epoch": 15.430508474576271,
-      "grad_norm": 1.4041998386383057,
-      "learning_rate": 7.077587939698492e-05,
-      "loss": 5.2644,
-      "step": 29588
-    },
-    {
-      "epoch": 15.43102998696219,
-      "grad_norm": 1.5449012517929077,
-      "learning_rate": 7.07748743718593e-05,
-      "loss": 5.3102,
-      "step": 29589
-    },
-    {
-      "epoch": 15.43155149934811,
-      "grad_norm": 1.367257833480835,
-      "learning_rate": 7.077386934673367e-05,
-      "loss": 5.6383,
-      "step": 29590
-    },
-    {
-      "epoch": 15.432073011734028,
-      "grad_norm": 1.4013272523880005,
-      "learning_rate": 7.077286432160805e-05,
-      "loss": 4.9891,
-      "step": 29591
-    },
-    {
-      "epoch": 15.432594524119947,
-      "grad_norm": 1.441127896308899,
-      "learning_rate": 7.077185929648242e-05,
-      "loss": 5.0508,
-      "step": 29592
-    },
-    {
-      "epoch": 15.433116036505867,
-      "grad_norm": 1.378674864768982,
-      "learning_rate": 7.07708542713568e-05,
-      "loss": 5.6718,
-      "step": 29593
-    },
-    {
-      "epoch": 15.433637548891786,
-      "grad_norm": 1.3998055458068848,
-      "learning_rate": 7.076984924623116e-05,
-      "loss": 4.8891,
-      "step": 29594
-    },
-    {
-      "epoch": 15.434159061277706,
-      "grad_norm": 1.4207749366760254,
-      "learning_rate": 7.076884422110554e-05,
-      "loss": 5.9114,
-      "step": 29595
-    },
-    {
-      "epoch": 15.434680573663625,
-      "grad_norm": 1.4738154411315918,
-      "learning_rate": 7.07678391959799e-05,
-      "loss": 5.3692,
-      "step": 29596
-    },
-    {
-      "epoch": 15.435202086049543,
-      "grad_norm": 1.4386310577392578,
-      "learning_rate": 7.076683417085428e-05,
-      "loss": 5.4765,
-      "step": 29597
-    },
-    {
-      "epoch": 15.435723598435462,
-      "grad_norm": 1.3407903909683228,
-      "learning_rate": 7.076582914572864e-05,
-      "loss": 5.0232,
-      "step": 29598
-    },
-    {
-      "epoch": 15.436245110821382,
-      "grad_norm": 1.5039657354354858,
-      "learning_rate": 7.076482412060302e-05,
-      "loss": 5.5895,
-      "step": 29599
-    },
-    {
-      "epoch": 15.436766623207301,
-      "grad_norm": 1.520312786102295,
-      "learning_rate": 7.076381909547738e-05,
-      "loss": 5.345,
-      "step": 29600
-    },
-    {
-      "epoch": 15.43728813559322,
-      "grad_norm": 1.3942300081253052,
-      "learning_rate": 7.076281407035176e-05,
-      "loss": 5.6043,
-      "step": 29601
-    },
-    {
-      "epoch": 15.43780964797914,
-      "grad_norm": 1.4001415967941284,
-      "learning_rate": 7.076180904522614e-05,
-      "loss": 5.6145,
-      "step": 29602
-    },
-    {
-      "epoch": 15.438331160365058,
-      "grad_norm": 1.3699923753738403,
-      "learning_rate": 7.07608040201005e-05,
-      "loss": 5.4561,
-      "step": 29603
-    },
-    {
-      "epoch": 15.438852672750977,
-      "grad_norm": 1.5702497959136963,
-      "learning_rate": 7.075979899497488e-05,
-      "loss": 5.2234,
-      "step": 29604
-    },
-    {
-      "epoch": 15.439374185136897,
-      "grad_norm": 1.4122146368026733,
-      "learning_rate": 7.075879396984925e-05,
-      "loss": 5.4918,
-      "step": 29605
-    },
-    {
-      "epoch": 15.439895697522816,
-      "grad_norm": 1.3964678049087524,
-      "learning_rate": 7.075778894472362e-05,
-      "loss": 4.864,
-      "step": 29606
-    },
-    {
-      "epoch": 15.440417209908736,
-      "grad_norm": 1.5024381875991821,
-      "learning_rate": 7.075678391959799e-05,
-      "loss": 5.2271,
-      "step": 29607
-    },
-    {
-      "epoch": 15.440938722294655,
-      "grad_norm": 1.6814385652542114,
-      "learning_rate": 7.075577889447237e-05,
-      "loss": 4.9364,
-      "step": 29608
-    },
-    {
-      "epoch": 15.441460234680573,
-      "grad_norm": 1.5830061435699463,
-      "learning_rate": 7.075477386934673e-05,
-      "loss": 5.2519,
-      "step": 29609
-    },
-    {
-      "epoch": 15.441981747066492,
-      "grad_norm": 1.6121630668640137,
-      "learning_rate": 7.075376884422111e-05,
-      "loss": 5.1507,
-      "step": 29610
-    },
-    {
-      "epoch": 15.442503259452412,
-      "grad_norm": 1.4428331851959229,
-      "learning_rate": 7.075276381909549e-05,
-      "loss": 5.1669,
-      "step": 29611
-    },
-    {
-      "epoch": 15.443024771838331,
-      "grad_norm": 1.497091293334961,
-      "learning_rate": 7.075175879396986e-05,
-      "loss": 5.2744,
-      "step": 29612
-    },
-    {
-      "epoch": 15.44354628422425,
-      "grad_norm": 1.4903266429901123,
-      "learning_rate": 7.075075376884423e-05,
-      "loss": 5.1646,
-      "step": 29613
-    },
-    {
-      "epoch": 15.44406779661017,
-      "grad_norm": 1.5185997486114502,
-      "learning_rate": 7.074974874371859e-05,
-      "loss": 5.3623,
-      "step": 29614
-    },
-    {
-      "epoch": 15.444589308996088,
-      "grad_norm": 1.4407308101654053,
-      "learning_rate": 7.074874371859297e-05,
-      "loss": 5.535,
-      "step": 29615
-    },
-    {
-      "epoch": 15.445110821382007,
-      "grad_norm": 1.437556266784668,
-      "learning_rate": 7.074773869346734e-05,
-      "loss": 5.3431,
-      "step": 29616
-    },
-    {
-      "epoch": 15.445632333767927,
-      "grad_norm": 1.5059400796890259,
-      "learning_rate": 7.074673366834171e-05,
-      "loss": 5.4475,
-      "step": 29617
-    },
-    {
-      "epoch": 15.446153846153846,
-      "grad_norm": 1.608389139175415,
-      "learning_rate": 7.074572864321608e-05,
-      "loss": 5.171,
-      "step": 29618
-    },
-    {
-      "epoch": 15.446675358539766,
-      "grad_norm": 1.4422017335891724,
-      "learning_rate": 7.074472361809046e-05,
-      "loss": 5.2614,
-      "step": 29619
-    },
-    {
-      "epoch": 15.447196870925685,
-      "grad_norm": 1.5849167108535767,
-      "learning_rate": 7.074371859296482e-05,
-      "loss": 5.3063,
-      "step": 29620
-    },
-    {
-      "epoch": 15.447718383311603,
-      "grad_norm": 1.4444866180419922,
-      "learning_rate": 7.07427135678392e-05,
-      "loss": 4.7844,
-      "step": 29621
-    },
-    {
-      "epoch": 15.448239895697522,
-      "grad_norm": 1.5534603595733643,
-      "learning_rate": 7.074170854271357e-05,
-      "loss": 5.5338,
-      "step": 29622
-    },
-    {
-      "epoch": 15.448761408083442,
-      "grad_norm": 1.5264501571655273,
-      "learning_rate": 7.074070351758795e-05,
-      "loss": 5.059,
-      "step": 29623
-    },
-    {
-      "epoch": 15.449282920469361,
-      "grad_norm": 1.5578356981277466,
-      "learning_rate": 7.073969849246232e-05,
-      "loss": 5.1032,
-      "step": 29624
-    },
-    {
-      "epoch": 15.44980443285528,
-      "grad_norm": 1.5481518507003784,
-      "learning_rate": 7.07386934673367e-05,
-      "loss": 5.2292,
-      "step": 29625
-    },
-    {
-      "epoch": 15.4503259452412,
-      "grad_norm": 1.4260507822036743,
-      "learning_rate": 7.073768844221106e-05,
-      "loss": 5.2819,
-      "step": 29626
-    },
-    {
-      "epoch": 15.450847457627118,
-      "grad_norm": 1.4907326698303223,
-      "learning_rate": 7.073668341708542e-05,
-      "loss": 5.141,
-      "step": 29627
-    },
-    {
-      "epoch": 15.451368970013037,
-      "grad_norm": 1.4179877042770386,
-      "learning_rate": 7.07356783919598e-05,
-      "loss": 5.4145,
-      "step": 29628
-    },
-    {
-      "epoch": 15.451890482398957,
-      "grad_norm": 1.4200284481048584,
-      "learning_rate": 7.073467336683417e-05,
-      "loss": 5.565,
-      "step": 29629
-    },
-    {
-      "epoch": 15.452411994784876,
-      "grad_norm": 1.4873936176300049,
-      "learning_rate": 7.073366834170854e-05,
-      "loss": 5.5989,
-      "step": 29630
-    },
-    {
-      "epoch": 15.452933507170796,
-      "grad_norm": 1.4622176885604858,
-      "learning_rate": 7.073266331658292e-05,
-      "loss": 5.4698,
-      "step": 29631
-    },
-    {
-      "epoch": 15.453455019556715,
-      "grad_norm": 1.4618054628372192,
-      "learning_rate": 7.07316582914573e-05,
-      "loss": 5.5624,
-      "step": 29632
-    },
-    {
-      "epoch": 15.453976531942633,
-      "grad_norm": 1.501267671585083,
-      "learning_rate": 7.073065326633166e-05,
-      "loss": 5.3128,
-      "step": 29633
-    },
-    {
-      "epoch": 15.454498044328552,
-      "grad_norm": 1.5188133716583252,
-      "learning_rate": 7.072964824120604e-05,
-      "loss": 5.1546,
-      "step": 29634
-    },
-    {
-      "epoch": 15.455019556714472,
-      "grad_norm": 1.4764472246170044,
-      "learning_rate": 7.07286432160804e-05,
-      "loss": 4.7659,
-      "step": 29635
-    },
-    {
-      "epoch": 15.455541069100391,
-      "grad_norm": 1.500624656677246,
-      "learning_rate": 7.072763819095478e-05,
-      "loss": 4.9909,
-      "step": 29636
-    },
-    {
-      "epoch": 15.45606258148631,
-      "grad_norm": 1.4550176858901978,
-      "learning_rate": 7.072663316582915e-05,
-      "loss": 5.6096,
-      "step": 29637
-    },
-    {
-      "epoch": 15.45658409387223,
-      "grad_norm": 1.4890352487564087,
-      "learning_rate": 7.072562814070353e-05,
-      "loss": 5.5865,
-      "step": 29638
-    },
-    {
-      "epoch": 15.457105606258148,
-      "grad_norm": 1.410120964050293,
-      "learning_rate": 7.072462311557789e-05,
-      "loss": 5.564,
-      "step": 29639
-    },
-    {
-      "epoch": 15.457627118644067,
-      "grad_norm": 1.3942935466766357,
-      "learning_rate": 7.072361809045225e-05,
-      "loss": 5.7211,
-      "step": 29640
-    },
-    {
-      "epoch": 15.458148631029987,
-      "grad_norm": 1.5763134956359863,
-      "learning_rate": 7.072261306532663e-05,
-      "loss": 5.0607,
-      "step": 29641
-    },
-    {
-      "epoch": 15.458670143415906,
-      "grad_norm": 1.5173691511154175,
-      "learning_rate": 7.072160804020101e-05,
-      "loss": 5.3327,
-      "step": 29642
-    },
-    {
-      "epoch": 15.459191655801826,
-      "grad_norm": 1.5600018501281738,
-      "learning_rate": 7.072060301507539e-05,
-      "loss": 5.2901,
-      "step": 29643
-    },
-    {
-      "epoch": 15.459713168187744,
-      "grad_norm": 1.4765982627868652,
-      "learning_rate": 7.071959798994975e-05,
-      "loss": 5.4088,
-      "step": 29644
-    },
-    {
-      "epoch": 15.460234680573663,
-      "grad_norm": 1.4144312143325806,
-      "learning_rate": 7.071859296482413e-05,
-      "loss": 5.3649,
-      "step": 29645
-    },
-    {
-      "epoch": 15.460756192959582,
-      "grad_norm": 1.2828383445739746,
-      "learning_rate": 7.07175879396985e-05,
-      "loss": 5.8385,
-      "step": 29646
-    },
-    {
-      "epoch": 15.461277705345502,
-      "grad_norm": 1.4504365921020508,
-      "learning_rate": 7.071658291457287e-05,
-      "loss": 5.7333,
-      "step": 29647
-    },
-    {
-      "epoch": 15.461799217731421,
-      "grad_norm": 1.4201005697250366,
-      "learning_rate": 7.071557788944724e-05,
-      "loss": 5.1866,
-      "step": 29648
-    },
-    {
-      "epoch": 15.46232073011734,
-      "grad_norm": 1.3870352506637573,
-      "learning_rate": 7.071457286432161e-05,
-      "loss": 5.4895,
-      "step": 29649
-    },
-    {
-      "epoch": 15.46284224250326,
-      "grad_norm": 1.5437941551208496,
-      "learning_rate": 7.071356783919598e-05,
-      "loss": 5.7382,
-      "step": 29650
-    },
-    {
-      "epoch": 15.463363754889178,
-      "grad_norm": 1.513773798942566,
-      "learning_rate": 7.071256281407036e-05,
-      "loss": 5.1005,
-      "step": 29651
-    },
-    {
-      "epoch": 15.463885267275097,
-      "grad_norm": 1.4854575395584106,
-      "learning_rate": 7.071155778894472e-05,
-      "loss": 5.5668,
-      "step": 29652
-    },
-    {
-      "epoch": 15.464406779661017,
-      "grad_norm": 1.5179544687271118,
-      "learning_rate": 7.07105527638191e-05,
-      "loss": 5.0194,
-      "step": 29653
-    },
-    {
-      "epoch": 15.464928292046936,
-      "grad_norm": 1.4691311120986938,
-      "learning_rate": 7.070954773869348e-05,
-      "loss": 5.3163,
-      "step": 29654
-    },
-    {
-      "epoch": 15.465449804432856,
-      "grad_norm": 1.4095406532287598,
-      "learning_rate": 7.070854271356784e-05,
-      "loss": 5.4378,
-      "step": 29655
-    },
-    {
-      "epoch": 15.465971316818774,
-      "grad_norm": 1.5820074081420898,
-      "learning_rate": 7.070753768844222e-05,
-      "loss": 5.4457,
-      "step": 29656
-    },
-    {
-      "epoch": 15.466492829204693,
-      "grad_norm": 1.4046874046325684,
-      "learning_rate": 7.070653266331658e-05,
-      "loss": 5.7129,
-      "step": 29657
-    },
-    {
-      "epoch": 15.467014341590613,
-      "grad_norm": 1.4070909023284912,
-      "learning_rate": 7.070552763819096e-05,
-      "loss": 5.6835,
-      "step": 29658
-    },
-    {
-      "epoch": 15.467535853976532,
-      "grad_norm": 1.4900997877120972,
-      "learning_rate": 7.070452261306532e-05,
-      "loss": 5.2144,
-      "step": 29659
-    },
-    {
-      "epoch": 15.468057366362451,
-      "grad_norm": 1.5742108821868896,
-      "learning_rate": 7.07035175879397e-05,
-      "loss": 5.6229,
-      "step": 29660
-    },
-    {
-      "epoch": 15.468578878748371,
-      "grad_norm": 1.4454482793807983,
-      "learning_rate": 7.070251256281407e-05,
-      "loss": 5.5305,
-      "step": 29661
-    },
-    {
-      "epoch": 15.469100391134289,
-      "grad_norm": 1.509254813194275,
-      "learning_rate": 7.070150753768844e-05,
-      "loss": 5.4487,
-      "step": 29662
-    },
-    {
-      "epoch": 15.469621903520208,
-      "grad_norm": 1.4824298620224,
-      "learning_rate": 7.070050251256282e-05,
-      "loss": 5.6882,
-      "step": 29663
-    },
-    {
-      "epoch": 15.470143415906128,
-      "grad_norm": 1.4978344440460205,
-      "learning_rate": 7.06994974874372e-05,
-      "loss": 5.8415,
-      "step": 29664
-    },
-    {
-      "epoch": 15.470664928292047,
-      "grad_norm": 1.5148247480392456,
-      "learning_rate": 7.069849246231156e-05,
-      "loss": 5.4826,
-      "step": 29665
-    },
-    {
-      "epoch": 15.471186440677966,
-      "grad_norm": 1.4739954471588135,
-      "learning_rate": 7.069748743718593e-05,
-      "loss": 5.0488,
-      "step": 29666
-    },
-    {
-      "epoch": 15.471707953063886,
-      "grad_norm": 1.6225553750991821,
-      "learning_rate": 7.06964824120603e-05,
-      "loss": 4.4646,
-      "step": 29667
-    },
-    {
-      "epoch": 15.472229465449804,
-      "grad_norm": 1.4783589839935303,
-      "learning_rate": 7.069547738693467e-05,
-      "loss": 5.7432,
-      "step": 29668
-    },
-    {
-      "epoch": 15.472750977835723,
-      "grad_norm": 1.5139974355697632,
-      "learning_rate": 7.069447236180905e-05,
-      "loss": 5.1828,
-      "step": 29669
-    },
-    {
-      "epoch": 15.473272490221643,
-      "grad_norm": 1.531670093536377,
-      "learning_rate": 7.069346733668341e-05,
-      "loss": 5.1001,
-      "step": 29670
-    },
-    {
-      "epoch": 15.473794002607562,
-      "grad_norm": 1.5472794771194458,
-      "learning_rate": 7.069246231155779e-05,
-      "loss": 5.2673,
-      "step": 29671
-    },
-    {
-      "epoch": 15.474315514993481,
-      "grad_norm": 1.5028196573257446,
-      "learning_rate": 7.069145728643215e-05,
-      "loss": 5.2526,
-      "step": 29672
-    },
-    {
-      "epoch": 15.474837027379401,
-      "grad_norm": 1.458742380142212,
-      "learning_rate": 7.069045226130653e-05,
-      "loss": 5.6484,
-      "step": 29673
-    },
-    {
-      "epoch": 15.475358539765319,
-      "grad_norm": 1.5127419233322144,
-      "learning_rate": 7.068944723618091e-05,
-      "loss": 5.0239,
-      "step": 29674
-    },
-    {
-      "epoch": 15.475880052151238,
-      "grad_norm": 1.377558708190918,
-      "learning_rate": 7.068844221105529e-05,
-      "loss": 5.6177,
-      "step": 29675
-    },
-    {
-      "epoch": 15.476401564537158,
-      "grad_norm": 1.3718022108078003,
-      "learning_rate": 7.068743718592965e-05,
-      "loss": 5.3764,
-      "step": 29676
-    },
-    {
-      "epoch": 15.476923076923077,
-      "grad_norm": 1.3848201036453247,
-      "learning_rate": 7.068643216080403e-05,
-      "loss": 5.6315,
-      "step": 29677
-    },
-    {
-      "epoch": 15.477444589308996,
-      "grad_norm": 1.433778166770935,
-      "learning_rate": 7.06854271356784e-05,
-      "loss": 5.3604,
-      "step": 29678
-    },
-    {
-      "epoch": 15.477966101694916,
-      "grad_norm": 1.437677264213562,
-      "learning_rate": 7.068442211055277e-05,
-      "loss": 5.463,
-      "step": 29679
-    },
-    {
-      "epoch": 15.478487614080834,
-      "grad_norm": 1.4279510974884033,
-      "learning_rate": 7.068341708542714e-05,
-      "loss": 5.3963,
-      "step": 29680
-    },
-    {
-      "epoch": 15.479009126466753,
-      "grad_norm": 1.5222307443618774,
-      "learning_rate": 7.06824120603015e-05,
-      "loss": 5.3391,
-      "step": 29681
-    },
-    {
-      "epoch": 15.479530638852673,
-      "grad_norm": 1.515159010887146,
-      "learning_rate": 7.068140703517588e-05,
-      "loss": 5.0775,
-      "step": 29682
-    },
-    {
-      "epoch": 15.480052151238592,
-      "grad_norm": 1.3471401929855347,
-      "learning_rate": 7.068040201005026e-05,
-      "loss": 5.7657,
-      "step": 29683
-    },
-    {
-      "epoch": 15.480573663624511,
-      "grad_norm": 1.6295406818389893,
-      "learning_rate": 7.067939698492463e-05,
-      "loss": 5.2212,
-      "step": 29684
-    },
-    {
-      "epoch": 15.481095176010431,
-      "grad_norm": 1.4862338304519653,
-      "learning_rate": 7.0678391959799e-05,
-      "loss": 5.3154,
-      "step": 29685
-    },
-    {
-      "epoch": 15.481616688396349,
-      "grad_norm": 1.3492958545684814,
-      "learning_rate": 7.067738693467338e-05,
-      "loss": 5.7642,
-      "step": 29686
-    },
-    {
-      "epoch": 15.482138200782268,
-      "grad_norm": 1.6241532564163208,
-      "learning_rate": 7.067638190954774e-05,
-      "loss": 4.9871,
-      "step": 29687
-    },
-    {
-      "epoch": 15.482659713168188,
-      "grad_norm": 1.4315733909606934,
-      "learning_rate": 7.067537688442212e-05,
-      "loss": 5.6198,
-      "step": 29688
-    },
-    {
-      "epoch": 15.483181225554107,
-      "grad_norm": 1.4734489917755127,
-      "learning_rate": 7.067437185929648e-05,
-      "loss": 5.3938,
-      "step": 29689
-    },
-    {
-      "epoch": 15.483702737940026,
-      "grad_norm": 1.4832732677459717,
-      "learning_rate": 7.067336683417086e-05,
-      "loss": 5.4417,
-      "step": 29690
-    },
-    {
-      "epoch": 15.484224250325946,
-      "grad_norm": 1.436396837234497,
-      "learning_rate": 7.067236180904522e-05,
-      "loss": 5.409,
-      "step": 29691
-    },
-    {
-      "epoch": 15.484745762711864,
-      "grad_norm": 1.4367002248764038,
-      "learning_rate": 7.06713567839196e-05,
-      "loss": 5.5854,
-      "step": 29692
-    },
-    {
-      "epoch": 15.485267275097783,
-      "grad_norm": 1.401820182800293,
-      "learning_rate": 7.067035175879397e-05,
-      "loss": 5.4685,
-      "step": 29693
-    },
-    {
-      "epoch": 15.485788787483703,
-      "grad_norm": 1.7150884866714478,
-      "learning_rate": 7.066934673366834e-05,
-      "loss": 4.7839,
-      "step": 29694
-    },
-    {
-      "epoch": 15.486310299869622,
-      "grad_norm": 1.508100152015686,
-      "learning_rate": 7.066834170854272e-05,
-      "loss": 5.0688,
-      "step": 29695
-    },
-    {
-      "epoch": 15.486831812255542,
-      "grad_norm": 1.4716459512710571,
-      "learning_rate": 7.066733668341709e-05,
-      "loss": 4.888,
-      "step": 29696
-    },
-    {
-      "epoch": 15.487353324641461,
-      "grad_norm": 1.4668211936950684,
-      "learning_rate": 7.066633165829146e-05,
-      "loss": 4.4758,
-      "step": 29697
-    },
-    {
-      "epoch": 15.487874837027379,
-      "grad_norm": 1.447083592414856,
-      "learning_rate": 7.066532663316583e-05,
-      "loss": 5.4312,
-      "step": 29698
-    },
-    {
-      "epoch": 15.488396349413298,
-      "grad_norm": 1.5265862941741943,
-      "learning_rate": 7.066432160804021e-05,
-      "loss": 5.0184,
-      "step": 29699
-    },
-    {
-      "epoch": 15.488917861799218,
-      "grad_norm": 1.4284898042678833,
-      "learning_rate": 7.066331658291457e-05,
-      "loss": 4.9711,
-      "step": 29700
-    },
-    {
-      "epoch": 15.489439374185137,
-      "grad_norm": 1.5861599445343018,
-      "learning_rate": 7.066231155778895e-05,
-      "loss": 4.7551,
-      "step": 29701
-    },
-    {
-      "epoch": 15.489960886571057,
-      "grad_norm": 1.484700083732605,
-      "learning_rate": 7.066130653266331e-05,
-      "loss": 4.9657,
-      "step": 29702
-    },
-    {
-      "epoch": 15.490482398956976,
-      "grad_norm": 1.486958384513855,
-      "learning_rate": 7.066030150753769e-05,
-      "loss": 5.6232,
-      "step": 29703
-    },
-    {
-      "epoch": 15.491003911342894,
-      "grad_norm": 1.399328351020813,
-      "learning_rate": 7.065929648241207e-05,
-      "loss": 5.5511,
-      "step": 29704
-    },
-    {
-      "epoch": 15.491525423728813,
-      "grad_norm": 1.4129737615585327,
-      "learning_rate": 7.065829145728645e-05,
-      "loss": 5.5657,
-      "step": 29705
-    },
-    {
-      "epoch": 15.492046936114733,
-      "grad_norm": 1.422866702079773,
-      "learning_rate": 7.065728643216081e-05,
-      "loss": 5.2962,
-      "step": 29706
-    },
-    {
-      "epoch": 15.492568448500652,
-      "grad_norm": 1.3733655214309692,
-      "learning_rate": 7.065628140703518e-05,
-      "loss": 5.4949,
-      "step": 29707
-    },
-    {
-      "epoch": 15.493089960886572,
-      "grad_norm": 1.417636513710022,
-      "learning_rate": 7.065527638190955e-05,
-      "loss": 5.2925,
-      "step": 29708
-    },
-    {
-      "epoch": 15.493611473272491,
-      "grad_norm": 1.568176031112671,
-      "learning_rate": 7.065427135678392e-05,
-      "loss": 5.0327,
-      "step": 29709
-    },
-    {
-      "epoch": 15.494132985658409,
-      "grad_norm": 1.4728968143463135,
-      "learning_rate": 7.06532663316583e-05,
-      "loss": 5.7222,
-      "step": 29710
-    },
-    {
-      "epoch": 15.494654498044328,
-      "grad_norm": 1.4307410717010498,
-      "learning_rate": 7.065226130653266e-05,
-      "loss": 5.3377,
-      "step": 29711
-    },
-    {
-      "epoch": 15.495176010430248,
-      "grad_norm": 1.4904263019561768,
-      "learning_rate": 7.065125628140704e-05,
-      "loss": 5.5696,
-      "step": 29712
-    },
-    {
-      "epoch": 15.495697522816167,
-      "grad_norm": 1.493096947669983,
-      "learning_rate": 7.06502512562814e-05,
-      "loss": 5.5689,
-      "step": 29713
-    },
-    {
-      "epoch": 15.496219035202087,
-      "grad_norm": 1.4421977996826172,
-      "learning_rate": 7.064924623115578e-05,
-      "loss": 4.7959,
-      "step": 29714
-    },
-    {
-      "epoch": 15.496740547588006,
-      "grad_norm": 1.5103121995925903,
-      "learning_rate": 7.064824120603016e-05,
-      "loss": 5.4254,
-      "step": 29715
-    },
-    {
-      "epoch": 15.497262059973924,
-      "grad_norm": 1.4910341501235962,
-      "learning_rate": 7.064723618090454e-05,
-      "loss": 5.5567,
-      "step": 29716
-    },
-    {
-      "epoch": 15.497783572359843,
-      "grad_norm": 1.5045862197875977,
-      "learning_rate": 7.06462311557789e-05,
-      "loss": 5.3533,
-      "step": 29717
-    },
-    {
-      "epoch": 15.498305084745763,
-      "grad_norm": 1.3986520767211914,
-      "learning_rate": 7.064522613065328e-05,
-      "loss": 5.267,
-      "step": 29718
-    },
-    {
-      "epoch": 15.498826597131682,
-      "grad_norm": 1.4453613758087158,
-      "learning_rate": 7.064422110552764e-05,
-      "loss": 5.3499,
-      "step": 29719
-    },
-    {
-      "epoch": 15.499348109517602,
-      "grad_norm": 1.4646681547164917,
-      "learning_rate": 7.0643216080402e-05,
-      "loss": 5.2901,
-      "step": 29720
-    },
-    {
-      "epoch": 15.499869621903521,
-      "grad_norm": 1.588362455368042,
-      "learning_rate": 7.064221105527638e-05,
-      "loss": 4.8759,
-      "step": 29721
-    },
-    {
-      "epoch": 15.500391134289439,
-      "grad_norm": 1.487717628479004,
-      "learning_rate": 7.064120603015075e-05,
-      "loss": 5.5464,
-      "step": 29722
-    },
-    {
-      "epoch": 15.500912646675358,
-      "grad_norm": 1.7198896408081055,
-      "learning_rate": 7.064020100502513e-05,
-      "loss": 5.5057,
-      "step": 29723
-    },
-    {
-      "epoch": 15.501434159061278,
-      "grad_norm": 1.4188807010650635,
-      "learning_rate": 7.06391959798995e-05,
-      "loss": 5.548,
-      "step": 29724
-    },
-    {
-      "epoch": 15.501955671447197,
-      "grad_norm": 1.5355890989303589,
-      "learning_rate": 7.063819095477388e-05,
-      "loss": 5.4785,
-      "step": 29725
-    },
-    {
-      "epoch": 15.502477183833117,
-      "grad_norm": 1.484499216079712,
-      "learning_rate": 7.063718592964825e-05,
-      "loss": 5.4568,
-      "step": 29726
-    },
-    {
-      "epoch": 15.502998696219036,
-      "grad_norm": 1.4065556526184082,
-      "learning_rate": 7.063618090452262e-05,
-      "loss": 5.6368,
-      "step": 29727
-    },
-    {
-      "epoch": 15.503520208604954,
-      "grad_norm": 1.6094199419021606,
-      "learning_rate": 7.063517587939699e-05,
-      "loss": 5.688,
-      "step": 29728
-    },
-    {
-      "epoch": 15.504041720990873,
-      "grad_norm": 1.4081048965454102,
-      "learning_rate": 7.063417085427137e-05,
-      "loss": 5.1858,
-      "step": 29729
-    },
-    {
-      "epoch": 15.504563233376793,
-      "grad_norm": 1.514070987701416,
-      "learning_rate": 7.063316582914573e-05,
-      "loss": 5.4571,
-      "step": 29730
-    },
-    {
-      "epoch": 15.505084745762712,
-      "grad_norm": 1.483888030052185,
-      "learning_rate": 7.063216080402011e-05,
-      "loss": 5.2293,
-      "step": 29731
-    },
-    {
-      "epoch": 15.505606258148632,
-      "grad_norm": 1.4952235221862793,
-      "learning_rate": 7.063115577889447e-05,
-      "loss": 5.2144,
-      "step": 29732
-    },
-    {
-      "epoch": 15.506127770534551,
-      "grad_norm": 1.4558608531951904,
-      "learning_rate": 7.063015075376884e-05,
-      "loss": 5.8951,
-      "step": 29733
-    },
-    {
-      "epoch": 15.506649282920469,
-      "grad_norm": 1.534998893737793,
-      "learning_rate": 7.062914572864321e-05,
-      "loss": 4.9245,
-      "step": 29734
-    },
-    {
-      "epoch": 15.507170795306388,
-      "grad_norm": 1.428452730178833,
-      "learning_rate": 7.062814070351759e-05,
-      "loss": 5.7881,
-      "step": 29735
-    },
-    {
-      "epoch": 15.507692307692308,
-      "grad_norm": 1.4388796091079712,
-      "learning_rate": 7.062713567839197e-05,
-      "loss": 5.6742,
-      "step": 29736
-    },
-    {
-      "epoch": 15.508213820078227,
-      "grad_norm": 1.6274371147155762,
-      "learning_rate": 7.062613065326633e-05,
-      "loss": 4.8436,
-      "step": 29737
-    },
-    {
-      "epoch": 15.508735332464147,
-      "grad_norm": 1.4667145013809204,
-      "learning_rate": 7.062512562814071e-05,
-      "loss": 5.4955,
-      "step": 29738
-    },
-    {
-      "epoch": 15.509256844850064,
-      "grad_norm": 1.5027046203613281,
-      "learning_rate": 7.062412060301508e-05,
-      "loss": 5.3555,
-      "step": 29739
-    },
-    {
-      "epoch": 15.509778357235984,
-      "grad_norm": 1.515792965888977,
-      "learning_rate": 7.062311557788945e-05,
-      "loss": 5.7367,
-      "step": 29740
-    },
-    {
-      "epoch": 15.510299869621903,
-      "grad_norm": 1.5283528566360474,
-      "learning_rate": 7.062211055276382e-05,
-      "loss": 4.8268,
-      "step": 29741
-    },
-    {
-      "epoch": 15.510821382007823,
-      "grad_norm": 1.3812004327774048,
-      "learning_rate": 7.06211055276382e-05,
-      "loss": 5.5084,
-      "step": 29742
-    },
-    {
-      "epoch": 15.511342894393742,
-      "grad_norm": 1.5578882694244385,
-      "learning_rate": 7.062010050251256e-05,
-      "loss": 5.3913,
-      "step": 29743
-    },
-    {
-      "epoch": 15.511864406779662,
-      "grad_norm": 1.4343864917755127,
-      "learning_rate": 7.061909547738694e-05,
-      "loss": 5.5955,
-      "step": 29744
-    },
-    {
-      "epoch": 15.512385919165581,
-      "grad_norm": 1.5252264738082886,
-      "learning_rate": 7.061809045226132e-05,
-      "loss": 5.4692,
-      "step": 29745
-    },
-    {
-      "epoch": 15.512907431551499,
-      "grad_norm": 1.5099170207977295,
-      "learning_rate": 7.061708542713568e-05,
-      "loss": 5.3558,
-      "step": 29746
-    },
-    {
-      "epoch": 15.513428943937418,
-      "grad_norm": 1.4658761024475098,
-      "learning_rate": 7.061608040201006e-05,
-      "loss": 5.3545,
-      "step": 29747
-    },
-    {
-      "epoch": 15.513950456323338,
-      "grad_norm": 1.3477787971496582,
-      "learning_rate": 7.061507537688442e-05,
-      "loss": 5.434,
-      "step": 29748
-    },
-    {
-      "epoch": 15.514471968709257,
-      "grad_norm": 1.4713538885116577,
-      "learning_rate": 7.06140703517588e-05,
-      "loss": 5.1124,
-      "step": 29749
-    },
-    {
-      "epoch": 15.514993481095177,
-      "grad_norm": 1.4272589683532715,
-      "learning_rate": 7.061306532663316e-05,
-      "loss": 5.7358,
-      "step": 29750
-    },
-    {
-      "epoch": 15.515514993481094,
-      "grad_norm": 1.4134178161621094,
-      "learning_rate": 7.061206030150754e-05,
-      "loss": 5.9954,
-      "step": 29751
-    },
-    {
-      "epoch": 15.516036505867014,
-      "grad_norm": 1.40263831615448,
-      "learning_rate": 7.06110552763819e-05,
-      "loss": 5.3596,
-      "step": 29752
-    },
-    {
-      "epoch": 15.516558018252933,
-      "grad_norm": 1.4973150491714478,
-      "learning_rate": 7.061005025125628e-05,
-      "loss": 5.1567,
-      "step": 29753
-    },
-    {
-      "epoch": 15.517079530638853,
-      "grad_norm": 1.4745291471481323,
-      "learning_rate": 7.060904522613065e-05,
-      "loss": 5.0277,
-      "step": 29754
-    },
-    {
-      "epoch": 15.517601043024772,
-      "grad_norm": 1.5274412631988525,
-      "learning_rate": 7.060804020100503e-05,
-      "loss": 5.5163,
-      "step": 29755
-    },
-    {
-      "epoch": 15.518122555410692,
-      "grad_norm": 1.4831572771072388,
-      "learning_rate": 7.06070351758794e-05,
-      "loss": 5.4218,
-      "step": 29756
-    },
-    {
-      "epoch": 15.518644067796611,
-      "grad_norm": 1.5726218223571777,
-      "learning_rate": 7.060603015075378e-05,
-      "loss": 5.3881,
-      "step": 29757
-    },
-    {
-      "epoch": 15.519165580182529,
-      "grad_norm": 1.539580225944519,
-      "learning_rate": 7.060502512562815e-05,
-      "loss": 4.5145,
-      "step": 29758
-    },
-    {
-      "epoch": 15.519687092568448,
-      "grad_norm": 1.4816638231277466,
-      "learning_rate": 7.060402010050252e-05,
-      "loss": 5.3556,
-      "step": 29759
-    },
-    {
-      "epoch": 15.520208604954368,
-      "grad_norm": 1.6098450422286987,
-      "learning_rate": 7.060301507537689e-05,
-      "loss": 5.2538,
-      "step": 29760
-    },
-    {
-      "epoch": 15.520730117340287,
-      "grad_norm": 1.5723119974136353,
-      "learning_rate": 7.060201005025125e-05,
-      "loss": 4.6901,
-      "step": 29761
-    },
-    {
-      "epoch": 15.521251629726207,
-      "grad_norm": 1.4748954772949219,
-      "learning_rate": 7.060100502512563e-05,
-      "loss": 5.3719,
-      "step": 29762
-    },
-    {
-      "epoch": 15.521773142112124,
-      "grad_norm": 1.5230860710144043,
-      "learning_rate": 7.06e-05,
-      "loss": 5.4467,
-      "step": 29763
-    },
-    {
-      "epoch": 15.522294654498044,
-      "grad_norm": 1.6751346588134766,
-      "learning_rate": 7.059899497487437e-05,
-      "loss": 4.6599,
-      "step": 29764
-    },
-    {
-      "epoch": 15.522816166883963,
-      "grad_norm": 1.7366809844970703,
-      "learning_rate": 7.059798994974875e-05,
-      "loss": 4.9636,
-      "step": 29765
-    },
-    {
-      "epoch": 15.523337679269883,
-      "grad_norm": 1.345672607421875,
-      "learning_rate": 7.059698492462313e-05,
-      "loss": 5.3736,
-      "step": 29766
-    },
-    {
-      "epoch": 15.523859191655802,
-      "grad_norm": 1.4626842737197876,
-      "learning_rate": 7.059597989949749e-05,
-      "loss": 5.6302,
-      "step": 29767
-    },
-    {
-      "epoch": 15.524380704041722,
-      "grad_norm": 1.4428528547286987,
-      "learning_rate": 7.059497487437187e-05,
-      "loss": 5.595,
-      "step": 29768
-    },
-    {
-      "epoch": 15.52490221642764,
-      "grad_norm": 1.5451743602752686,
-      "learning_rate": 7.059396984924623e-05,
-      "loss": 5.2927,
-      "step": 29769
-    },
-    {
-      "epoch": 15.525423728813559,
-      "grad_norm": 1.4627478122711182,
-      "learning_rate": 7.059296482412061e-05,
-      "loss": 5.3079,
-      "step": 29770
-    },
-    {
-      "epoch": 15.525945241199478,
-      "grad_norm": 1.502657413482666,
-      "learning_rate": 7.059195979899498e-05,
-      "loss": 5.4731,
-      "step": 29771
-    },
-    {
-      "epoch": 15.526466753585398,
-      "grad_norm": 1.4977943897247314,
-      "learning_rate": 7.059095477386935e-05,
-      "loss": 4.7859,
-      "step": 29772
-    },
-    {
-      "epoch": 15.526988265971317,
-      "grad_norm": 1.4842380285263062,
-      "learning_rate": 7.058994974874372e-05,
-      "loss": 5.1261,
-      "step": 29773
-    },
-    {
-      "epoch": 15.527509778357237,
-      "grad_norm": 1.5325905084609985,
-      "learning_rate": 7.058894472361808e-05,
-      "loss": 5.0485,
-      "step": 29774
-    },
-    {
-      "epoch": 15.528031290743154,
-      "grad_norm": 1.5889365673065186,
-      "learning_rate": 7.058793969849246e-05,
-      "loss": 5.6515,
-      "step": 29775
-    },
-    {
-      "epoch": 15.528552803129074,
-      "grad_norm": 1.463250756263733,
-      "learning_rate": 7.058693467336684e-05,
-      "loss": 5.3048,
-      "step": 29776
-    },
-    {
-      "epoch": 15.529074315514993,
-      "grad_norm": 1.4086146354675293,
-      "learning_rate": 7.058592964824122e-05,
-      "loss": 5.5337,
-      "step": 29777
-    },
-    {
-      "epoch": 15.529595827900913,
-      "grad_norm": 1.450734257698059,
-      "learning_rate": 7.058492462311558e-05,
-      "loss": 5.2074,
-      "step": 29778
-    },
-    {
-      "epoch": 15.530117340286832,
-      "grad_norm": 1.3955553770065308,
-      "learning_rate": 7.058391959798996e-05,
-      "loss": 5.5409,
-      "step": 29779
-    },
-    {
-      "epoch": 15.530638852672752,
-      "grad_norm": 1.455403208732605,
-      "learning_rate": 7.058291457286432e-05,
-      "loss": 5.4548,
-      "step": 29780
-    },
-    {
-      "epoch": 15.53116036505867,
-      "grad_norm": 1.5893361568450928,
-      "learning_rate": 7.05819095477387e-05,
-      "loss": 5.0385,
-      "step": 29781
-    },
-    {
-      "epoch": 15.531681877444589,
-      "grad_norm": 1.5814707279205322,
-      "learning_rate": 7.058090452261307e-05,
-      "loss": 5.1502,
-      "step": 29782
-    },
-    {
-      "epoch": 15.532203389830508,
-      "grad_norm": 1.39535653591156,
-      "learning_rate": 7.057989949748744e-05,
-      "loss": 5.5591,
-      "step": 29783
-    },
-    {
-      "epoch": 15.532724902216428,
-      "grad_norm": 1.3255698680877686,
-      "learning_rate": 7.057889447236181e-05,
-      "loss": 5.8529,
-      "step": 29784
-    },
-    {
-      "epoch": 15.533246414602347,
-      "grad_norm": 1.5013705492019653,
-      "learning_rate": 7.057788944723619e-05,
-      "loss": 5.443,
-      "step": 29785
-    },
-    {
-      "epoch": 15.533767926988267,
-      "grad_norm": 1.4662271738052368,
-      "learning_rate": 7.057688442211056e-05,
-      "loss": 5.676,
-      "step": 29786
-    },
-    {
-      "epoch": 15.534289439374184,
-      "grad_norm": 1.3984819650650024,
-      "learning_rate": 7.057587939698493e-05,
-      "loss": 5.164,
-      "step": 29787
-    },
-    {
-      "epoch": 15.534810951760104,
-      "grad_norm": 1.4018551111221313,
-      "learning_rate": 7.05748743718593e-05,
-      "loss": 5.5252,
-      "step": 29788
-    },
-    {
-      "epoch": 15.535332464146023,
-      "grad_norm": 1.5171080827713013,
-      "learning_rate": 7.057386934673367e-05,
-      "loss": 5.3123,
-      "step": 29789
-    },
-    {
-      "epoch": 15.535853976531943,
-      "grad_norm": 1.5248485803604126,
-      "learning_rate": 7.057286432160805e-05,
-      "loss": 5.3584,
-      "step": 29790
-    },
-    {
-      "epoch": 15.536375488917862,
-      "grad_norm": 1.4990285634994507,
-      "learning_rate": 7.057185929648241e-05,
-      "loss": 5.1415,
-      "step": 29791
-    },
-    {
-      "epoch": 15.536897001303782,
-      "grad_norm": 1.4484447240829468,
-      "learning_rate": 7.057085427135679e-05,
-      "loss": 5.7667,
-      "step": 29792
-    },
-    {
-      "epoch": 15.5374185136897,
-      "grad_norm": 1.4058154821395874,
-      "learning_rate": 7.056984924623115e-05,
-      "loss": 4.9864,
-      "step": 29793
-    },
-    {
-      "epoch": 15.537940026075619,
-      "grad_norm": 1.4378197193145752,
-      "learning_rate": 7.056884422110553e-05,
-      "loss": 5.6076,
-      "step": 29794
-    },
-    {
-      "epoch": 15.538461538461538,
-      "grad_norm": 1.5059949159622192,
-      "learning_rate": 7.05678391959799e-05,
-      "loss": 5.6208,
-      "step": 29795
-    },
-    {
-      "epoch": 15.538983050847458,
-      "grad_norm": 1.4939093589782715,
-      "learning_rate": 7.056683417085427e-05,
-      "loss": 4.4723,
-      "step": 29796
-    },
-    {
-      "epoch": 15.539504563233377,
-      "grad_norm": 1.4853445291519165,
-      "learning_rate": 7.056582914572865e-05,
-      "loss": 5.6973,
-      "step": 29797
-    },
-    {
-      "epoch": 15.540026075619297,
-      "grad_norm": 1.54734468460083,
-      "learning_rate": 7.056482412060303e-05,
-      "loss": 5.2089,
-      "step": 29798
-    },
-    {
-      "epoch": 15.540547588005214,
-      "grad_norm": 1.4218579530715942,
-      "learning_rate": 7.05638190954774e-05,
-      "loss": 5.3431,
-      "step": 29799
-    },
-    {
-      "epoch": 15.541069100391134,
-      "grad_norm": 1.4414992332458496,
-      "learning_rate": 7.056281407035176e-05,
-      "loss": 5.4397,
-      "step": 29800
-    },
-    {
-      "epoch": 15.541590612777053,
-      "grad_norm": 1.4398777484893799,
-      "learning_rate": 7.056180904522614e-05,
-      "loss": 5.4143,
-      "step": 29801
-    },
-    {
-      "epoch": 15.542112125162973,
-      "grad_norm": 1.513824224472046,
-      "learning_rate": 7.05608040201005e-05,
-      "loss": 5.2995,
-      "step": 29802
-    },
-    {
-      "epoch": 15.542633637548892,
-      "grad_norm": 1.5201908349990845,
-      "learning_rate": 7.055979899497488e-05,
-      "loss": 5.6828,
-      "step": 29803
-    },
-    {
-      "epoch": 15.543155149934812,
-      "grad_norm": 1.6158503293991089,
-      "learning_rate": 7.055879396984924e-05,
-      "loss": 5.413,
-      "step": 29804
-    },
-    {
-      "epoch": 15.54367666232073,
-      "grad_norm": 1.5046491622924805,
-      "learning_rate": 7.055778894472362e-05,
-      "loss": 5.4036,
-      "step": 29805
-    },
-    {
-      "epoch": 15.544198174706649,
-      "grad_norm": 1.9255393743515015,
-      "learning_rate": 7.055678391959798e-05,
-      "loss": 4.7942,
-      "step": 29806
-    },
-    {
-      "epoch": 15.544719687092568,
-      "grad_norm": 1.5638474225997925,
-      "learning_rate": 7.055577889447236e-05,
-      "loss": 5.1776,
-      "step": 29807
-    },
-    {
-      "epoch": 15.545241199478488,
-      "grad_norm": 1.507620096206665,
-      "learning_rate": 7.055477386934674e-05,
-      "loss": 5.4937,
-      "step": 29808
-    },
-    {
-      "epoch": 15.545762711864407,
-      "grad_norm": 1.3841540813446045,
-      "learning_rate": 7.055376884422112e-05,
-      "loss": 5.369,
-      "step": 29809
-    },
-    {
-      "epoch": 15.546284224250327,
-      "grad_norm": 1.3752565383911133,
-      "learning_rate": 7.055276381909548e-05,
-      "loss": 5.2216,
-      "step": 29810
-    },
-    {
-      "epoch": 15.546805736636244,
-      "grad_norm": 1.6476832628250122,
-      "learning_rate": 7.055175879396986e-05,
-      "loss": 5.3487,
-      "step": 29811
-    },
-    {
-      "epoch": 15.547327249022164,
-      "grad_norm": 1.5123642683029175,
-      "learning_rate": 7.055075376884422e-05,
-      "loss": 5.0146,
-      "step": 29812
-    },
-    {
-      "epoch": 15.547848761408083,
-      "grad_norm": 1.5649917125701904,
-      "learning_rate": 7.054974874371859e-05,
-      "loss": 5.4072,
-      "step": 29813
-    },
-    {
-      "epoch": 15.548370273794003,
-      "grad_norm": 1.3651069402694702,
-      "learning_rate": 7.054874371859297e-05,
-      "loss": 5.8063,
-      "step": 29814
-    },
-    {
-      "epoch": 15.548891786179922,
-      "grad_norm": 1.4728069305419922,
-      "learning_rate": 7.054773869346733e-05,
-      "loss": 5.4221,
-      "step": 29815
-    },
-    {
-      "epoch": 15.549413298565842,
-      "grad_norm": 1.571724534034729,
-      "learning_rate": 7.054673366834171e-05,
-      "loss": 5.8254,
-      "step": 29816
-    },
-    {
-      "epoch": 15.54993481095176,
-      "grad_norm": 1.9240257740020752,
-      "learning_rate": 7.054572864321609e-05,
-      "loss": 5.2448,
-      "step": 29817
-    },
-    {
-      "epoch": 15.550456323337679,
-      "grad_norm": 1.4654467105865479,
-      "learning_rate": 7.054472361809046e-05,
-      "loss": 5.7089,
-      "step": 29818
-    },
-    {
-      "epoch": 15.550977835723598,
-      "grad_norm": 1.5009195804595947,
-      "learning_rate": 7.054371859296483e-05,
-      "loss": 5.5112,
-      "step": 29819
-    },
-    {
-      "epoch": 15.551499348109518,
-      "grad_norm": 1.5191709995269775,
-      "learning_rate": 7.05427135678392e-05,
-      "loss": 5.3179,
-      "step": 29820
-    },
-    {
-      "epoch": 15.552020860495437,
-      "grad_norm": 1.5352131128311157,
-      "learning_rate": 7.054170854271357e-05,
-      "loss": 4.9761,
-      "step": 29821
-    },
-    {
-      "epoch": 15.552542372881355,
-      "grad_norm": 1.4045053720474243,
-      "learning_rate": 7.054070351758795e-05,
-      "loss": 5.1678,
-      "step": 29822
-    },
-    {
-      "epoch": 15.553063885267274,
-      "grad_norm": 1.3714873790740967,
-      "learning_rate": 7.053969849246231e-05,
-      "loss": 5.7452,
-      "step": 29823
-    },
-    {
-      "epoch": 15.553585397653194,
-      "grad_norm": 1.7013514041900635,
-      "learning_rate": 7.053869346733669e-05,
-      "loss": 4.9531,
-      "step": 29824
-    },
-    {
-      "epoch": 15.554106910039113,
-      "grad_norm": 1.3601326942443848,
-      "learning_rate": 7.053768844221105e-05,
-      "loss": 5.6883,
-      "step": 29825
-    },
-    {
-      "epoch": 15.554628422425033,
-      "grad_norm": 1.4602686166763306,
-      "learning_rate": 7.053668341708542e-05,
-      "loss": 5.164,
-      "step": 29826
-    },
-    {
-      "epoch": 15.555149934810952,
-      "grad_norm": 1.4759551286697388,
-      "learning_rate": 7.05356783919598e-05,
-      "loss": 5.2029,
-      "step": 29827
-    },
-    {
-      "epoch": 15.555671447196872,
-      "grad_norm": 1.5075558423995972,
-      "learning_rate": 7.053467336683417e-05,
-      "loss": 5.4587,
-      "step": 29828
-    },
-    {
-      "epoch": 15.55619295958279,
-      "grad_norm": 1.5216131210327148,
-      "learning_rate": 7.053366834170855e-05,
-      "loss": 5.561,
-      "step": 29829
-    },
-    {
-      "epoch": 15.556714471968709,
-      "grad_norm": 1.5742125511169434,
-      "learning_rate": 7.053266331658292e-05,
-      "loss": 5.3917,
-      "step": 29830
-    },
-    {
-      "epoch": 15.557235984354628,
-      "grad_norm": 1.4046748876571655,
-      "learning_rate": 7.05316582914573e-05,
-      "loss": 5.7231,
-      "step": 29831
-    },
-    {
-      "epoch": 15.557757496740548,
-      "grad_norm": 1.3964786529541016,
-      "learning_rate": 7.053065326633166e-05,
-      "loss": 5.4838,
-      "step": 29832
-    },
-    {
-      "epoch": 15.558279009126467,
-      "grad_norm": 1.474071979522705,
-      "learning_rate": 7.052964824120604e-05,
-      "loss": 5.1342,
-      "step": 29833
-    },
-    {
-      "epoch": 15.558800521512385,
-      "grad_norm": 1.6422621011734009,
-      "learning_rate": 7.05286432160804e-05,
-      "loss": 5.1839,
-      "step": 29834
-    },
-    {
-      "epoch": 15.559322033898304,
-      "grad_norm": 1.5394941568374634,
-      "learning_rate": 7.052763819095478e-05,
-      "loss": 5.0178,
-      "step": 29835
-    },
-    {
-      "epoch": 15.559843546284224,
-      "grad_norm": 1.3236088752746582,
-      "learning_rate": 7.052663316582914e-05,
-      "loss": 4.8579,
-      "step": 29836
-    },
-    {
-      "epoch": 15.560365058670143,
-      "grad_norm": 1.4940462112426758,
-      "learning_rate": 7.052562814070352e-05,
-      "loss": 5.6638,
-      "step": 29837
-    },
-    {
-      "epoch": 15.560886571056063,
-      "grad_norm": 1.37168550491333,
-      "learning_rate": 7.05246231155779e-05,
-      "loss": 5.6013,
-      "step": 29838
-    },
-    {
-      "epoch": 15.561408083441982,
-      "grad_norm": 1.4741586446762085,
-      "learning_rate": 7.052361809045226e-05,
-      "loss": 5.3339,
-      "step": 29839
-    },
-    {
-      "epoch": 15.561929595827902,
-      "grad_norm": 1.4865258932113647,
-      "learning_rate": 7.052261306532664e-05,
-      "loss": 5.5279,
-      "step": 29840
-    },
-    {
-      "epoch": 15.56245110821382,
-      "grad_norm": 1.4546748399734497,
-      "learning_rate": 7.0521608040201e-05,
-      "loss": 4.793,
-      "step": 29841
-    },
-    {
-      "epoch": 15.562972620599739,
-      "grad_norm": 1.5693137645721436,
-      "learning_rate": 7.052060301507538e-05,
-      "loss": 5.6034,
-      "step": 29842
-    },
-    {
-      "epoch": 15.563494132985658,
-      "grad_norm": 1.513153314590454,
-      "learning_rate": 7.051959798994975e-05,
-      "loss": 5.1504,
-      "step": 29843
-    },
-    {
-      "epoch": 15.564015645371578,
-      "grad_norm": 1.545957326889038,
-      "learning_rate": 7.051859296482412e-05,
-      "loss": 5.4439,
-      "step": 29844
-    },
-    {
-      "epoch": 15.564537157757497,
-      "grad_norm": 1.6157091856002808,
-      "learning_rate": 7.051758793969849e-05,
-      "loss": 5.4018,
-      "step": 29845
-    },
-    {
-      "epoch": 15.565058670143415,
-      "grad_norm": 1.5408800840377808,
-      "learning_rate": 7.051658291457287e-05,
-      "loss": 5.4966,
-      "step": 29846
-    },
-    {
-      "epoch": 15.565580182529335,
-      "grad_norm": 1.4139115810394287,
-      "learning_rate": 7.051557788944723e-05,
-      "loss": 5.5746,
-      "step": 29847
-    },
-    {
-      "epoch": 15.566101694915254,
-      "grad_norm": 1.633948564529419,
-      "learning_rate": 7.051457286432161e-05,
-      "loss": 5.387,
-      "step": 29848
-    },
-    {
-      "epoch": 15.566623207301173,
-      "grad_norm": 1.5489752292633057,
-      "learning_rate": 7.051356783919599e-05,
-      "loss": 5.129,
-      "step": 29849
-    },
-    {
-      "epoch": 15.567144719687093,
-      "grad_norm": 1.6512142419815063,
-      "learning_rate": 7.051256281407036e-05,
-      "loss": 4.9282,
-      "step": 29850
-    },
-    {
-      "epoch": 15.567666232073012,
-      "grad_norm": 1.5166614055633545,
-      "learning_rate": 7.051155778894473e-05,
-      "loss": 5.3985,
-      "step": 29851
-    },
-    {
-      "epoch": 15.568187744458932,
-      "grad_norm": 1.4935499429702759,
-      "learning_rate": 7.05105527638191e-05,
-      "loss": 5.6653,
-      "step": 29852
-    },
-    {
-      "epoch": 15.56870925684485,
-      "grad_norm": 1.3796736001968384,
-      "learning_rate": 7.050954773869347e-05,
-      "loss": 5.265,
-      "step": 29853
-    },
-    {
-      "epoch": 15.569230769230769,
-      "grad_norm": 1.658752202987671,
-      "learning_rate": 7.050854271356784e-05,
-      "loss": 4.9538,
-      "step": 29854
-    },
-    {
-      "epoch": 15.569752281616688,
-      "grad_norm": 1.6187547445297241,
-      "learning_rate": 7.050753768844221e-05,
-      "loss": 4.9371,
-      "step": 29855
-    },
-    {
-      "epoch": 15.570273794002608,
-      "grad_norm": 1.4898301362991333,
-      "learning_rate": 7.050653266331658e-05,
-      "loss": 5.2277,
-      "step": 29856
-    },
-    {
-      "epoch": 15.570795306388527,
-      "grad_norm": 1.5860871076583862,
-      "learning_rate": 7.050552763819096e-05,
-      "loss": 5.4895,
-      "step": 29857
-    },
-    {
-      "epoch": 15.571316818774445,
-      "grad_norm": 1.4545207023620605,
-      "learning_rate": 7.050452261306533e-05,
-      "loss": 5.7731,
-      "step": 29858
-    },
-    {
-      "epoch": 15.571838331160365,
-      "grad_norm": 1.4629108905792236,
-      "learning_rate": 7.050351758793971e-05,
-      "loss": 4.9581,
-      "step": 29859
-    },
-    {
-      "epoch": 15.572359843546284,
-      "grad_norm": 1.4972574710845947,
-      "learning_rate": 7.050251256281408e-05,
-      "loss": 5.473,
-      "step": 29860
-    },
-    {
-      "epoch": 15.572881355932203,
-      "grad_norm": 1.406599521636963,
-      "learning_rate": 7.050150753768845e-05,
-      "loss": 4.6265,
-      "step": 29861
-    },
-    {
-      "epoch": 15.573402868318123,
-      "grad_norm": 1.5145223140716553,
-      "learning_rate": 7.050050251256282e-05,
-      "loss": 5.585,
-      "step": 29862
-    },
-    {
-      "epoch": 15.573924380704042,
-      "grad_norm": 1.7047865390777588,
-      "learning_rate": 7.04994974874372e-05,
-      "loss": 4.5371,
-      "step": 29863
-    },
-    {
-      "epoch": 15.57444589308996,
-      "grad_norm": 1.4536340236663818,
-      "learning_rate": 7.049849246231156e-05,
-      "loss": 5.6756,
-      "step": 29864
-    },
-    {
-      "epoch": 15.57496740547588,
-      "grad_norm": 1.3580827713012695,
-      "learning_rate": 7.049748743718594e-05,
-      "loss": 5.2571,
-      "step": 29865
-    },
-    {
-      "epoch": 15.575488917861799,
-      "grad_norm": 1.460701823234558,
-      "learning_rate": 7.04964824120603e-05,
-      "loss": 5.7662,
-      "step": 29866
-    },
-    {
-      "epoch": 15.576010430247718,
-      "grad_norm": 1.4651075601577759,
-      "learning_rate": 7.049547738693467e-05,
-      "loss": 5.7616,
-      "step": 29867
-    },
-    {
-      "epoch": 15.576531942633638,
-      "grad_norm": 1.643721103668213,
-      "learning_rate": 7.049447236180904e-05,
-      "loss": 4.5942,
-      "step": 29868
-    },
-    {
-      "epoch": 15.577053455019557,
-      "grad_norm": 1.545576810836792,
-      "learning_rate": 7.049346733668342e-05,
-      "loss": 5.3161,
-      "step": 29869
-    },
-    {
-      "epoch": 15.577574967405475,
-      "grad_norm": 1.4163509607315063,
-      "learning_rate": 7.04924623115578e-05,
-      "loss": 5.2046,
-      "step": 29870
-    },
-    {
-      "epoch": 15.578096479791395,
-      "grad_norm": 1.4633320569992065,
-      "learning_rate": 7.049145728643216e-05,
-      "loss": 4.7622,
-      "step": 29871
-    },
-    {
-      "epoch": 15.578617992177314,
-      "grad_norm": 1.4803954362869263,
-      "learning_rate": 7.049045226130654e-05,
-      "loss": 5.4683,
-      "step": 29872
-    },
-    {
-      "epoch": 15.579139504563233,
-      "grad_norm": 1.5595331192016602,
-      "learning_rate": 7.04894472361809e-05,
-      "loss": 5.155,
-      "step": 29873
-    },
-    {
-      "epoch": 15.579661016949153,
-      "grad_norm": 1.5382615327835083,
-      "learning_rate": 7.048844221105528e-05,
-      "loss": 4.9215,
-      "step": 29874
-    },
-    {
-      "epoch": 15.580182529335072,
-      "grad_norm": 1.5639911890029907,
-      "learning_rate": 7.048743718592965e-05,
-      "loss": 5.1985,
-      "step": 29875
-    },
-    {
-      "epoch": 15.58070404172099,
-      "grad_norm": 1.555514931678772,
-      "learning_rate": 7.048643216080403e-05,
-      "loss": 5.2237,
-      "step": 29876
-    },
-    {
-      "epoch": 15.58122555410691,
-      "grad_norm": 1.431175947189331,
-      "learning_rate": 7.048542713567839e-05,
-      "loss": 5.1746,
-      "step": 29877
-    },
-    {
-      "epoch": 15.581747066492829,
-      "grad_norm": 1.5792667865753174,
-      "learning_rate": 7.048442211055277e-05,
-      "loss": 5.3228,
-      "step": 29878
-    },
-    {
-      "epoch": 15.582268578878748,
-      "grad_norm": 1.5162075757980347,
-      "learning_rate": 7.048341708542715e-05,
-      "loss": 5.5811,
-      "step": 29879
-    },
-    {
-      "epoch": 15.582790091264668,
-      "grad_norm": 1.4221224784851074,
-      "learning_rate": 7.048241206030151e-05,
-      "loss": 4.6255,
-      "step": 29880
-    },
-    {
-      "epoch": 15.583311603650587,
-      "grad_norm": 1.467383861541748,
-      "learning_rate": 7.048140703517589e-05,
-      "loss": 5.2627,
-      "step": 29881
-    },
-    {
-      "epoch": 15.583833116036505,
-      "grad_norm": 1.5884525775909424,
-      "learning_rate": 7.048040201005025e-05,
-      "loss": 5.2404,
-      "step": 29882
-    },
-    {
-      "epoch": 15.584354628422425,
-      "grad_norm": 1.479169487953186,
-      "learning_rate": 7.047939698492463e-05,
-      "loss": 5.2846,
-      "step": 29883
-    },
-    {
-      "epoch": 15.584876140808344,
-      "grad_norm": 1.506241798400879,
-      "learning_rate": 7.0478391959799e-05,
-      "loss": 5.0177,
-      "step": 29884
-    },
-    {
-      "epoch": 15.585397653194264,
-      "grad_norm": 1.3723769187927246,
-      "learning_rate": 7.047738693467337e-05,
-      "loss": 5.346,
-      "step": 29885
-    },
-    {
-      "epoch": 15.585919165580183,
-      "grad_norm": 1.533069372177124,
-      "learning_rate": 7.047638190954774e-05,
-      "loss": 5.3081,
-      "step": 29886
-    },
-    {
-      "epoch": 15.586440677966102,
-      "grad_norm": 1.4475507736206055,
-      "learning_rate": 7.047537688442211e-05,
-      "loss": 5.3327,
-      "step": 29887
-    },
-    {
-      "epoch": 15.58696219035202,
-      "grad_norm": 1.4839651584625244,
-      "learning_rate": 7.047437185929648e-05,
-      "loss": 5.5337,
-      "step": 29888
-    },
-    {
-      "epoch": 15.58748370273794,
-      "grad_norm": 1.4660507440567017,
-      "learning_rate": 7.047336683417086e-05,
-      "loss": 5.4932,
-      "step": 29889
-    },
-    {
-      "epoch": 15.588005215123859,
-      "grad_norm": 1.5035451650619507,
-      "learning_rate": 7.047236180904523e-05,
-      "loss": 5.3973,
-      "step": 29890
-    },
-    {
-      "epoch": 15.588526727509779,
-      "grad_norm": 1.5033352375030518,
-      "learning_rate": 7.047135678391961e-05,
-      "loss": 5.4778,
-      "step": 29891
-    },
-    {
-      "epoch": 15.589048239895698,
-      "grad_norm": 1.4247105121612549,
-      "learning_rate": 7.047035175879398e-05,
-      "loss": 4.7245,
-      "step": 29892
-    },
-    {
-      "epoch": 15.589569752281617,
-      "grad_norm": 1.6096715927124023,
-      "learning_rate": 7.046934673366834e-05,
-      "loss": 4.9723,
-      "step": 29893
-    },
-    {
-      "epoch": 15.590091264667535,
-      "grad_norm": 1.391749620437622,
-      "learning_rate": 7.046834170854272e-05,
-      "loss": 5.4632,
-      "step": 29894
-    },
-    {
-      "epoch": 15.590612777053455,
-      "grad_norm": 1.534865140914917,
-      "learning_rate": 7.046733668341708e-05,
-      "loss": 5.1682,
-      "step": 29895
-    },
-    {
-      "epoch": 15.591134289439374,
-      "grad_norm": 1.460228681564331,
-      "learning_rate": 7.046633165829146e-05,
-      "loss": 5.4874,
-      "step": 29896
-    },
-    {
-      "epoch": 15.591655801825294,
-      "grad_norm": 1.5613734722137451,
-      "learning_rate": 7.046532663316582e-05,
-      "loss": 5.0661,
-      "step": 29897
-    },
-    {
-      "epoch": 15.592177314211213,
-      "grad_norm": 1.36672043800354,
-      "learning_rate": 7.04643216080402e-05,
-      "loss": 5.1435,
-      "step": 29898
-    },
-    {
-      "epoch": 15.592698826597132,
-      "grad_norm": 1.4785797595977783,
-      "learning_rate": 7.046331658291458e-05,
-      "loss": 5.4307,
-      "step": 29899
-    },
-    {
-      "epoch": 15.59322033898305,
-      "grad_norm": 1.4932688474655151,
-      "learning_rate": 7.046231155778896e-05,
-      "loss": 5.3701,
-      "step": 29900
-    },
-    {
-      "epoch": 15.59374185136897,
-      "grad_norm": 1.4935871362686157,
-      "learning_rate": 7.046130653266332e-05,
-      "loss": 5.4197,
-      "step": 29901
-    },
-    {
-      "epoch": 15.594263363754889,
-      "grad_norm": 1.4162698984146118,
-      "learning_rate": 7.04603015075377e-05,
-      "loss": 4.9563,
-      "step": 29902
-    },
-    {
-      "epoch": 15.594784876140809,
-      "grad_norm": 1.5761297941207886,
-      "learning_rate": 7.045929648241206e-05,
-      "loss": 5.3335,
-      "step": 29903
-    },
-    {
-      "epoch": 15.595306388526728,
-      "grad_norm": 1.519510269165039,
-      "learning_rate": 7.045829145728644e-05,
-      "loss": 5.3549,
-      "step": 29904
-    },
-    {
-      "epoch": 15.595827900912647,
-      "grad_norm": 1.4000787734985352,
-      "learning_rate": 7.04572864321608e-05,
-      "loss": 4.852,
-      "step": 29905
-    },
-    {
-      "epoch": 15.596349413298565,
-      "grad_norm": 1.4698258638381958,
-      "learning_rate": 7.045628140703517e-05,
-      "loss": 5.4642,
-      "step": 29906
-    },
-    {
-      "epoch": 15.596870925684485,
-      "grad_norm": 1.474560260772705,
-      "learning_rate": 7.045527638190955e-05,
-      "loss": 5.2063,
-      "step": 29907
-    },
-    {
-      "epoch": 15.597392438070404,
-      "grad_norm": 1.5502727031707764,
-      "learning_rate": 7.045427135678391e-05,
-      "loss": 5.2943,
-      "step": 29908
-    },
-    {
-      "epoch": 15.597913950456324,
-      "grad_norm": 1.366568922996521,
-      "learning_rate": 7.045326633165829e-05,
-      "loss": 5.6402,
-      "step": 29909
-    },
-    {
-      "epoch": 15.598435462842243,
-      "grad_norm": 1.414322018623352,
-      "learning_rate": 7.045226130653267e-05,
-      "loss": 5.3582,
-      "step": 29910
-    },
-    {
-      "epoch": 15.598956975228162,
-      "grad_norm": 1.5684572458267212,
-      "learning_rate": 7.045125628140705e-05,
-      "loss": 5.3808,
-      "step": 29911
-    },
-    {
-      "epoch": 15.59947848761408,
-      "grad_norm": 1.6201958656311035,
-      "learning_rate": 7.045025125628141e-05,
-      "loss": 5.006,
-      "step": 29912
-    },
-    {
-      "epoch": 15.6,
-      "grad_norm": 1.4744551181793213,
-      "learning_rate": 7.044924623115579e-05,
-      "loss": 5.6099,
-      "step": 29913
-    },
-    {
-      "epoch": 15.600521512385919,
-      "grad_norm": 1.5173089504241943,
-      "learning_rate": 7.044824120603015e-05,
-      "loss": 5.5942,
-      "step": 29914
-    },
-    {
-      "epoch": 15.601043024771839,
-      "grad_norm": 1.4416636228561401,
-      "learning_rate": 7.044723618090453e-05,
-      "loss": 5.6945,
-      "step": 29915
-    },
-    {
-      "epoch": 15.601564537157758,
-      "grad_norm": 1.392016887664795,
-      "learning_rate": 7.04462311557789e-05,
-      "loss": 5.6053,
-      "step": 29916
-    },
-    {
-      "epoch": 15.602086049543676,
-      "grad_norm": 1.6438734531402588,
-      "learning_rate": 7.044522613065327e-05,
-      "loss": 5.1342,
-      "step": 29917
-    },
-    {
-      "epoch": 15.602607561929595,
-      "grad_norm": 1.4649128913879395,
-      "learning_rate": 7.044422110552764e-05,
-      "loss": 5.683,
-      "step": 29918
-    },
-    {
-      "epoch": 15.603129074315515,
-      "grad_norm": 1.6050852537155151,
-      "learning_rate": 7.044321608040201e-05,
-      "loss": 5.6314,
-      "step": 29919
-    },
-    {
-      "epoch": 15.603650586701434,
-      "grad_norm": 1.3818148374557495,
-      "learning_rate": 7.044221105527639e-05,
-      "loss": 5.708,
-      "step": 29920
-    },
-    {
-      "epoch": 15.604172099087354,
-      "grad_norm": 1.5242812633514404,
-      "learning_rate": 7.044120603015076e-05,
-      "loss": 5.078,
-      "step": 29921
-    },
-    {
-      "epoch": 15.604693611473273,
-      "grad_norm": 1.536713719367981,
-      "learning_rate": 7.044020100502513e-05,
-      "loss": 5.4355,
-      "step": 29922
-    },
-    {
-      "epoch": 15.605215123859193,
-      "grad_norm": 1.4669018983840942,
-      "learning_rate": 7.04391959798995e-05,
-      "loss": 5.2589,
-      "step": 29923
-    },
-    {
-      "epoch": 15.60573663624511,
-      "grad_norm": 1.4739636182785034,
-      "learning_rate": 7.043819095477388e-05,
-      "loss": 5.4632,
-      "step": 29924
-    },
-    {
-      "epoch": 15.60625814863103,
-      "grad_norm": 1.4418977499008179,
-      "learning_rate": 7.043718592964824e-05,
-      "loss": 5.5346,
-      "step": 29925
-    },
-    {
-      "epoch": 15.60677966101695,
-      "grad_norm": 1.4832143783569336,
-      "learning_rate": 7.043618090452262e-05,
-      "loss": 5.3099,
-      "step": 29926
-    },
-    {
-      "epoch": 15.607301173402869,
-      "grad_norm": 1.4884930849075317,
-      "learning_rate": 7.043517587939698e-05,
-      "loss": 5.7038,
-      "step": 29927
-    },
-    {
-      "epoch": 15.607822685788788,
-      "grad_norm": 1.4383413791656494,
-      "learning_rate": 7.043417085427136e-05,
-      "loss": 5.5042,
-      "step": 29928
-    },
-    {
-      "epoch": 15.608344198174706,
-      "grad_norm": 1.4774445295333862,
-      "learning_rate": 7.043316582914573e-05,
-      "loss": 5.4597,
-      "step": 29929
-    },
-    {
-      "epoch": 15.608865710560625,
-      "grad_norm": 1.4659324884414673,
-      "learning_rate": 7.04321608040201e-05,
-      "loss": 5.2099,
-      "step": 29930
-    },
-    {
-      "epoch": 15.609387222946545,
-      "grad_norm": 1.5013165473937988,
-      "learning_rate": 7.043115577889448e-05,
-      "loss": 5.1303,
-      "step": 29931
-    },
-    {
-      "epoch": 15.609908735332464,
-      "grad_norm": 1.407395362854004,
-      "learning_rate": 7.043015075376884e-05,
-      "loss": 5.0874,
-      "step": 29932
-    },
-    {
-      "epoch": 15.610430247718384,
-      "grad_norm": 1.39747154712677,
-      "learning_rate": 7.042914572864322e-05,
-      "loss": 5.5611,
-      "step": 29933
-    },
-    {
-      "epoch": 15.610951760104303,
-      "grad_norm": 1.5613833665847778,
-      "learning_rate": 7.042814070351759e-05,
-      "loss": 4.9609,
-      "step": 29934
-    },
-    {
-      "epoch": 15.611473272490223,
-      "grad_norm": 1.4261175394058228,
-      "learning_rate": 7.042713567839196e-05,
-      "loss": 4.3914,
-      "step": 29935
-    },
-    {
-      "epoch": 15.61199478487614,
-      "grad_norm": 1.5079468488693237,
-      "learning_rate": 7.042613065326633e-05,
-      "loss": 5.1145,
-      "step": 29936
-    },
-    {
-      "epoch": 15.61251629726206,
-      "grad_norm": 1.526282787322998,
-      "learning_rate": 7.042512562814071e-05,
-      "loss": 4.8837,
-      "step": 29937
-    },
-    {
-      "epoch": 15.61303780964798,
-      "grad_norm": 1.4468494653701782,
-      "learning_rate": 7.042412060301507e-05,
-      "loss": 5.6409,
-      "step": 29938
-    },
-    {
-      "epoch": 15.613559322033899,
-      "grad_norm": 1.7262965440750122,
-      "learning_rate": 7.042311557788945e-05,
-      "loss": 4.9102,
-      "step": 29939
-    },
-    {
-      "epoch": 15.614080834419818,
-      "grad_norm": 1.44370436668396,
-      "learning_rate": 7.042211055276383e-05,
-      "loss": 5.1259,
-      "step": 29940
-    },
-    {
-      "epoch": 15.614602346805736,
-      "grad_norm": 1.4548479318618774,
-      "learning_rate": 7.04211055276382e-05,
-      "loss": 5.7495,
-      "step": 29941
-    },
-    {
-      "epoch": 15.615123859191655,
-      "grad_norm": 1.5498237609863281,
-      "learning_rate": 7.042010050251257e-05,
-      "loss": 5.5402,
-      "step": 29942
-    },
-    {
-      "epoch": 15.615645371577575,
-      "grad_norm": 1.5302948951721191,
-      "learning_rate": 7.041909547738695e-05,
-      "loss": 5.1612,
-      "step": 29943
-    },
-    {
-      "epoch": 15.616166883963494,
-      "grad_norm": 1.5376778841018677,
-      "learning_rate": 7.041809045226131e-05,
-      "loss": 5.4204,
-      "step": 29944
-    },
-    {
-      "epoch": 15.616688396349414,
-      "grad_norm": 1.4004099369049072,
-      "learning_rate": 7.041708542713569e-05,
-      "loss": 5.1446,
-      "step": 29945
-    },
-    {
-      "epoch": 15.617209908735333,
-      "grad_norm": 1.3878200054168701,
-      "learning_rate": 7.041608040201005e-05,
-      "loss": 5.2523,
-      "step": 29946
-    },
-    {
-      "epoch": 15.617731421121253,
-      "grad_norm": 1.4165589809417725,
-      "learning_rate": 7.041507537688442e-05,
-      "loss": 5.4634,
-      "step": 29947
-    },
-    {
-      "epoch": 15.61825293350717,
-      "grad_norm": 1.432448148727417,
-      "learning_rate": 7.04140703517588e-05,
-      "loss": 5.5763,
-      "step": 29948
-    },
-    {
-      "epoch": 15.61877444589309,
-      "grad_norm": 1.5806833505630493,
-      "learning_rate": 7.041306532663316e-05,
-      "loss": 4.6218,
-      "step": 29949
-    },
-    {
-      "epoch": 15.61929595827901,
-      "grad_norm": 1.4274758100509644,
-      "learning_rate": 7.041206030150754e-05,
-      "loss": 5.3154,
-      "step": 29950
-    },
-    {
-      "epoch": 15.619817470664929,
-      "grad_norm": 1.5513089895248413,
-      "learning_rate": 7.041105527638192e-05,
-      "loss": 5.1289,
-      "step": 29951
-    },
-    {
-      "epoch": 15.620338983050848,
-      "grad_norm": 1.4450125694274902,
-      "learning_rate": 7.041005025125629e-05,
-      "loss": 5.5936,
-      "step": 29952
-    },
-    {
-      "epoch": 15.620860495436766,
-      "grad_norm": 1.4358642101287842,
-      "learning_rate": 7.040904522613066e-05,
-      "loss": 5.5387,
-      "step": 29953
-    },
-    {
-      "epoch": 15.621382007822685,
-      "grad_norm": 1.4499359130859375,
-      "learning_rate": 7.040804020100504e-05,
-      "loss": 5.2146,
-      "step": 29954
-    },
-    {
-      "epoch": 15.621903520208605,
-      "grad_norm": 1.4747754335403442,
-      "learning_rate": 7.04070351758794e-05,
-      "loss": 5.5442,
-      "step": 29955
-    },
-    {
-      "epoch": 15.622425032594524,
-      "grad_norm": 1.5905131101608276,
-      "learning_rate": 7.040603015075378e-05,
-      "loss": 5.0565,
-      "step": 29956
-    },
-    {
-      "epoch": 15.622946544980444,
-      "grad_norm": 1.44203519821167,
-      "learning_rate": 7.040502512562814e-05,
-      "loss": 5.2919,
-      "step": 29957
-    },
-    {
-      "epoch": 15.623468057366363,
-      "grad_norm": 1.5652529001235962,
-      "learning_rate": 7.040402010050252e-05,
-      "loss": 5.3853,
-      "step": 29958
-    },
-    {
-      "epoch": 15.62398956975228,
-      "grad_norm": 1.4141149520874023,
-      "learning_rate": 7.040301507537688e-05,
-      "loss": 5.3913,
-      "step": 29959
-    },
-    {
-      "epoch": 15.6245110821382,
-      "grad_norm": 1.524466633796692,
-      "learning_rate": 7.040201005025126e-05,
-      "loss": 4.9386,
-      "step": 29960
-    },
-    {
-      "epoch": 15.62503259452412,
-      "grad_norm": 1.350555658340454,
-      "learning_rate": 7.040100502512564e-05,
-      "loss": 4.8937,
-      "step": 29961
-    },
-    {
-      "epoch": 15.62555410691004,
-      "grad_norm": 1.6304889917373657,
-      "learning_rate": 7.04e-05,
-      "loss": 5.0244,
-      "step": 29962
-    },
-    {
-      "epoch": 15.626075619295959,
-      "grad_norm": 1.462827205657959,
-      "learning_rate": 7.039899497487438e-05,
-      "loss": 5.257,
-      "step": 29963
-    },
-    {
-      "epoch": 15.626597131681878,
-      "grad_norm": 1.3728485107421875,
-      "learning_rate": 7.039798994974875e-05,
-      "loss": 5.6536,
-      "step": 29964
-    },
-    {
-      "epoch": 15.627118644067796,
-      "grad_norm": 1.3408362865447998,
-      "learning_rate": 7.039698492462312e-05,
-      "loss": 5.8831,
-      "step": 29965
-    },
-    {
-      "epoch": 15.627640156453715,
-      "grad_norm": 1.429350733757019,
-      "learning_rate": 7.039597989949749e-05,
-      "loss": 5.2888,
-      "step": 29966
-    },
-    {
-      "epoch": 15.628161668839635,
-      "grad_norm": 1.4728062152862549,
-      "learning_rate": 7.039497487437187e-05,
-      "loss": 4.9436,
-      "step": 29967
-    },
-    {
-      "epoch": 15.628683181225554,
-      "grad_norm": 1.4137309789657593,
-      "learning_rate": 7.039396984924623e-05,
-      "loss": 5.6368,
-      "step": 29968
-    },
-    {
-      "epoch": 15.629204693611474,
-      "grad_norm": 1.5079853534698486,
-      "learning_rate": 7.039296482412061e-05,
-      "loss": 5.2758,
-      "step": 29969
-    },
-    {
-      "epoch": 15.629726205997393,
-      "grad_norm": 1.662076711654663,
-      "learning_rate": 7.039195979899497e-05,
-      "loss": 5.1512,
-      "step": 29970
-    },
-    {
-      "epoch": 15.63024771838331,
-      "grad_norm": 1.5657864809036255,
-      "learning_rate": 7.039095477386935e-05,
-      "loss": 5.7042,
-      "step": 29971
-    },
-    {
-      "epoch": 15.63076923076923,
-      "grad_norm": 1.4936258792877197,
-      "learning_rate": 7.038994974874373e-05,
-      "loss": 5.1216,
-      "step": 29972
-    },
-    {
-      "epoch": 15.63129074315515,
-      "grad_norm": 1.4902905225753784,
-      "learning_rate": 7.038894472361809e-05,
-      "loss": 5.4731,
-      "step": 29973
-    },
-    {
-      "epoch": 15.63181225554107,
-      "grad_norm": 1.3768261671066284,
-      "learning_rate": 7.038793969849247e-05,
-      "loss": 5.9513,
-      "step": 29974
-    },
-    {
-      "epoch": 15.632333767926989,
-      "grad_norm": 1.4883761405944824,
-      "learning_rate": 7.038693467336683e-05,
-      "loss": 5.5183,
-      "step": 29975
-    },
-    {
-      "epoch": 15.632855280312908,
-      "grad_norm": 1.5111212730407715,
-      "learning_rate": 7.038592964824121e-05,
-      "loss": 5.285,
-      "step": 29976
-    },
-    {
-      "epoch": 15.633376792698826,
-      "grad_norm": 1.566623330116272,
-      "learning_rate": 7.038492462311558e-05,
-      "loss": 5.4991,
-      "step": 29977
-    },
-    {
-      "epoch": 15.633898305084745,
-      "grad_norm": 1.4468623399734497,
-      "learning_rate": 7.038391959798995e-05,
-      "loss": 5.6106,
-      "step": 29978
-    },
-    {
-      "epoch": 15.634419817470665,
-      "grad_norm": 1.48420250415802,
-      "learning_rate": 7.038291457286432e-05,
-      "loss": 5.3263,
-      "step": 29979
-    },
-    {
-      "epoch": 15.634941329856584,
-      "grad_norm": 1.487633466720581,
-      "learning_rate": 7.03819095477387e-05,
-      "loss": 5.111,
-      "step": 29980
-    },
-    {
-      "epoch": 15.635462842242504,
-      "grad_norm": 1.549830675125122,
-      "learning_rate": 7.038090452261306e-05,
-      "loss": 5.2857,
-      "step": 29981
-    },
-    {
-      "epoch": 15.635984354628423,
-      "grad_norm": 1.4480869770050049,
-      "learning_rate": 7.037989949748744e-05,
-      "loss": 5.3869,
-      "step": 29982
-    },
-    {
-      "epoch": 15.63650586701434,
-      "grad_norm": 1.6220782995224,
-      "learning_rate": 7.037889447236182e-05,
-      "loss": 5.3479,
-      "step": 29983
-    },
-    {
-      "epoch": 15.63702737940026,
-      "grad_norm": 1.5577313899993896,
-      "learning_rate": 7.03778894472362e-05,
-      "loss": 4.7684,
-      "step": 29984
-    },
-    {
-      "epoch": 15.63754889178618,
-      "grad_norm": 1.4343575239181519,
-      "learning_rate": 7.037688442211056e-05,
-      "loss": 5.6796,
-      "step": 29985
-    },
-    {
-      "epoch": 15.6380704041721,
-      "grad_norm": 1.5110361576080322,
-      "learning_rate": 7.037587939698492e-05,
-      "loss": 5.0033,
-      "step": 29986
-    },
-    {
-      "epoch": 15.638591916558019,
-      "grad_norm": 1.631688117980957,
-      "learning_rate": 7.03748743718593e-05,
-      "loss": 4.8838,
-      "step": 29987
-    },
-    {
-      "epoch": 15.639113428943938,
-      "grad_norm": 1.5702265501022339,
-      "learning_rate": 7.037386934673366e-05,
-      "loss": 4.6927,
-      "step": 29988
-    },
-    {
-      "epoch": 15.639634941329856,
-      "grad_norm": 1.6152788400650024,
-      "learning_rate": 7.037286432160804e-05,
-      "loss": 5.1932,
-      "step": 29989
-    },
-    {
-      "epoch": 15.640156453715775,
-      "grad_norm": 1.5983808040618896,
-      "learning_rate": 7.03718592964824e-05,
-      "loss": 4.8768,
-      "step": 29990
-    },
-    {
-      "epoch": 15.640677966101695,
-      "grad_norm": 1.4208405017852783,
-      "learning_rate": 7.037085427135678e-05,
-      "loss": 4.9428,
-      "step": 29991
-    },
-    {
-      "epoch": 15.641199478487614,
-      "grad_norm": 1.5560824871063232,
-      "learning_rate": 7.036984924623116e-05,
-      "loss": 5.261,
-      "step": 29992
-    },
-    {
-      "epoch": 15.641720990873534,
-      "grad_norm": 1.3879438638687134,
-      "learning_rate": 7.036884422110554e-05,
-      "loss": 5.5944,
-      "step": 29993
-    },
-    {
-      "epoch": 15.642242503259453,
-      "grad_norm": 1.468414306640625,
-      "learning_rate": 7.03678391959799e-05,
-      "loss": 5.3608,
-      "step": 29994
-    },
-    {
-      "epoch": 15.642764015645371,
-      "grad_norm": 1.3937424421310425,
-      "learning_rate": 7.036683417085428e-05,
-      "loss": 5.316,
-      "step": 29995
-    },
-    {
-      "epoch": 15.64328552803129,
-      "grad_norm": 1.456793189048767,
-      "learning_rate": 7.036582914572865e-05,
-      "loss": 5.7648,
-      "step": 29996
-    },
-    {
-      "epoch": 15.64380704041721,
-      "grad_norm": 1.5667072534561157,
-      "learning_rate": 7.036482412060302e-05,
-      "loss": 4.9671,
-      "step": 29997
-    },
-    {
-      "epoch": 15.64432855280313,
-      "grad_norm": 1.564516544342041,
-      "learning_rate": 7.036381909547739e-05,
-      "loss": 5.2179,
-      "step": 29998
-    },
-    {
-      "epoch": 15.644850065189049,
-      "grad_norm": 1.4960548877716064,
-      "learning_rate": 7.036281407035175e-05,
-      "loss": 5.0949,
-      "step": 29999
-    },
-    {
-      "epoch": 15.645371577574968,
-      "grad_norm": 1.4661023616790771,
-      "learning_rate": 7.036180904522613e-05,
-      "loss": 5.3394,
-      "step": 30000
-    },
-    {
-      "epoch": 15.645371577574968,
-      "eval_loss": 5.427452087402344,
-      "eval_runtime": 42.67,
-      "eval_samples_per_second": 28.732,
-      "eval_steps_per_second": 3.609,
-      "step": 30000
-    },
-    {
-      "epoch": 15.645893089960886,
-      "grad_norm": 1.5810165405273438,
-      "learning_rate": 7.03608040201005e-05,
-      "loss": 4.7133,
-      "step": 30001
-    },
-    {
-      "epoch": 15.646414602346805,
-      "grad_norm": 1.5174697637557983,
-      "learning_rate": 7.035979899497487e-05,
-      "loss": 5.1704,
-      "step": 30002
-    },
-    {
-      "epoch": 15.646936114732725,
-      "grad_norm": 1.4082316160202026,
-      "learning_rate": 7.035879396984925e-05,
-      "loss": 5.2054,
-      "step": 30003
-    },
-    {
-      "epoch": 15.647457627118644,
-      "grad_norm": 1.4817538261413574,
-      "learning_rate": 7.035778894472363e-05,
-      "loss": 5.0175,
-      "step": 30004
-    },
-    {
-      "epoch": 15.647979139504564,
-      "grad_norm": 1.4781922101974487,
-      "learning_rate": 7.035678391959799e-05,
-      "loss": 5.6151,
-      "step": 30005
-    },
-    {
-      "epoch": 15.648500651890483,
-      "grad_norm": 1.4207042455673218,
-      "learning_rate": 7.035577889447237e-05,
-      "loss": 5.2115,
-      "step": 30006
-    },
-    {
-      "epoch": 15.649022164276401,
-      "grad_norm": 1.3230258226394653,
-      "learning_rate": 7.035477386934673e-05,
-      "loss": 5.8577,
-      "step": 30007
-    },
-    {
-      "epoch": 15.64954367666232,
-      "grad_norm": 1.5758445262908936,
-      "learning_rate": 7.035376884422111e-05,
-      "loss": 5.2338,
-      "step": 30008
-    },
-    {
-      "epoch": 15.65006518904824,
-      "grad_norm": 1.5014439821243286,
-      "learning_rate": 7.035276381909548e-05,
-      "loss": 5.0921,
-      "step": 30009
-    },
-    {
-      "epoch": 15.65058670143416,
-      "grad_norm": 1.5421696901321411,
-      "learning_rate": 7.035175879396985e-05,
-      "loss": 5.3465,
-      "step": 30010
-    },
-    {
-      "epoch": 15.651108213820079,
-      "grad_norm": 1.4673150777816772,
-      "learning_rate": 7.035075376884422e-05,
-      "loss": 5.6166,
-      "step": 30011
-    },
-    {
-      "epoch": 15.651629726205996,
-      "grad_norm": 1.5223835706710815,
-      "learning_rate": 7.03497487437186e-05,
-      "loss": 4.7889,
-      "step": 30012
-    },
-    {
-      "epoch": 15.652151238591916,
-      "grad_norm": 1.3539470434188843,
-      "learning_rate": 7.034874371859297e-05,
-      "loss": 5.7861,
-      "step": 30013
-    },
-    {
-      "epoch": 15.652672750977835,
-      "grad_norm": 1.5034499168395996,
-      "learning_rate": 7.034773869346734e-05,
-      "loss": 5.0802,
-      "step": 30014
-    },
-    {
-      "epoch": 15.653194263363755,
-      "grad_norm": 1.5000252723693848,
-      "learning_rate": 7.034673366834172e-05,
-      "loss": 5.3096,
-      "step": 30015
-    },
-    {
-      "epoch": 15.653715775749674,
-      "grad_norm": 1.3573576211929321,
-      "learning_rate": 7.034572864321608e-05,
-      "loss": 5.1289,
-      "step": 30016
-    },
-    {
-      "epoch": 15.654237288135594,
-      "grad_norm": 1.5119774341583252,
-      "learning_rate": 7.034472361809046e-05,
-      "loss": 5.0959,
-      "step": 30017
-    },
-    {
-      "epoch": 15.654758800521513,
-      "grad_norm": 1.4813077449798584,
-      "learning_rate": 7.034371859296482e-05,
-      "loss": 5.4628,
-      "step": 30018
-    },
-    {
-      "epoch": 15.655280312907431,
-      "grad_norm": 1.5783803462982178,
-      "learning_rate": 7.03427135678392e-05,
-      "loss": 5.2695,
-      "step": 30019
-    },
-    {
-      "epoch": 15.65580182529335,
-      "grad_norm": 1.5176588296890259,
-      "learning_rate": 7.034170854271357e-05,
-      "loss": 5.3559,
-      "step": 30020
-    },
-    {
-      "epoch": 15.65632333767927,
-      "grad_norm": 1.440812349319458,
-      "learning_rate": 7.034070351758794e-05,
-      "loss": 5.4147,
-      "step": 30021
-    },
-    {
-      "epoch": 15.65684485006519,
-      "grad_norm": 1.381279468536377,
-      "learning_rate": 7.033969849246231e-05,
-      "loss": 4.8754,
-      "step": 30022
-    },
-    {
-      "epoch": 15.657366362451109,
-      "grad_norm": 1.4483320713043213,
-      "learning_rate": 7.033869346733669e-05,
-      "loss": 5.7707,
-      "step": 30023
-    },
-    {
-      "epoch": 15.657887874837026,
-      "grad_norm": 1.2826277017593384,
-      "learning_rate": 7.033768844221106e-05,
-      "loss": 4.3647,
-      "step": 30024
-    },
-    {
-      "epoch": 15.658409387222946,
-      "grad_norm": 1.6106362342834473,
-      "learning_rate": 7.033668341708543e-05,
-      "loss": 5.127,
-      "step": 30025
-    },
-    {
-      "epoch": 15.658930899608865,
-      "grad_norm": 1.4412578344345093,
-      "learning_rate": 7.03356783919598e-05,
-      "loss": 5.7045,
-      "step": 30026
-    },
-    {
-      "epoch": 15.659452411994785,
-      "grad_norm": 1.5877286195755005,
-      "learning_rate": 7.033467336683417e-05,
-      "loss": 5.5094,
-      "step": 30027
-    },
-    {
-      "epoch": 15.659973924380704,
-      "grad_norm": 1.38240385055542,
-      "learning_rate": 7.033366834170855e-05,
-      "loss": 5.493,
-      "step": 30028
-    },
-    {
-      "epoch": 15.660495436766624,
-      "grad_norm": 1.405619502067566,
-      "learning_rate": 7.033266331658291e-05,
-      "loss": 5.6836,
-      "step": 30029
-    },
-    {
-      "epoch": 15.661016949152543,
-      "grad_norm": 1.4457433223724365,
-      "learning_rate": 7.033165829145729e-05,
-      "loss": 5.5485,
-      "step": 30030
-    },
-    {
-      "epoch": 15.661538461538461,
-      "grad_norm": 1.459743857383728,
-      "learning_rate": 7.033065326633165e-05,
-      "loss": 5.3881,
-      "step": 30031
-    },
-    {
-      "epoch": 15.66205997392438,
-      "grad_norm": 1.5080881118774414,
-      "learning_rate": 7.032964824120603e-05,
-      "loss": 5.2047,
-      "step": 30032
-    },
-    {
-      "epoch": 15.6625814863103,
-      "grad_norm": 1.384797215461731,
-      "learning_rate": 7.032864321608041e-05,
-      "loss": 5.7627,
-      "step": 30033
-    },
-    {
-      "epoch": 15.66310299869622,
-      "grad_norm": 1.4035999774932861,
-      "learning_rate": 7.032763819095479e-05,
-      "loss": 5.3854,
-      "step": 30034
-    },
-    {
-      "epoch": 15.663624511082139,
-      "grad_norm": 1.3466969728469849,
-      "learning_rate": 7.032663316582915e-05,
-      "loss": 5.1963,
-      "step": 30035
-    },
-    {
-      "epoch": 15.664146023468057,
-      "grad_norm": 1.3868387937545776,
-      "learning_rate": 7.032562814070353e-05,
-      "loss": 4.9735,
-      "step": 30036
-    },
-    {
-      "epoch": 15.664667535853976,
-      "grad_norm": 1.4800297021865845,
-      "learning_rate": 7.03246231155779e-05,
-      "loss": 5.4404,
-      "step": 30037
-    },
-    {
-      "epoch": 15.665189048239895,
-      "grad_norm": 1.3516308069229126,
-      "learning_rate": 7.032361809045227e-05,
-      "loss": 5.6714,
-      "step": 30038
-    },
-    {
-      "epoch": 15.665710560625815,
-      "grad_norm": 1.5503063201904297,
-      "learning_rate": 7.032261306532664e-05,
-      "loss": 5.4439,
-      "step": 30039
-    },
-    {
-      "epoch": 15.666232073011734,
-      "grad_norm": 1.5086185932159424,
-      "learning_rate": 7.0321608040201e-05,
-      "loss": 5.0039,
-      "step": 30040
-    },
-    {
-      "epoch": 15.666753585397654,
-      "grad_norm": 1.3921794891357422,
-      "learning_rate": 7.032060301507538e-05,
-      "loss": 5.6902,
-      "step": 30041
-    },
-    {
-      "epoch": 15.667275097783573,
-      "grad_norm": 1.4566251039505005,
-      "learning_rate": 7.031959798994974e-05,
-      "loss": 5.3333,
-      "step": 30042
-    },
-    {
-      "epoch": 15.667796610169491,
-      "grad_norm": 1.4193425178527832,
-      "learning_rate": 7.031859296482412e-05,
-      "loss": 5.7759,
-      "step": 30043
-    },
-    {
-      "epoch": 15.66831812255541,
-      "grad_norm": 1.3709461688995361,
-      "learning_rate": 7.03175879396985e-05,
-      "loss": 5.7155,
-      "step": 30044
-    },
-    {
-      "epoch": 15.66883963494133,
-      "grad_norm": 1.3801155090332031,
-      "learning_rate": 7.031658291457288e-05,
-      "loss": 5.63,
-      "step": 30045
-    },
-    {
-      "epoch": 15.66936114732725,
-      "grad_norm": 1.451909065246582,
-      "learning_rate": 7.031557788944724e-05,
-      "loss": 5.5007,
-      "step": 30046
-    },
-    {
-      "epoch": 15.669882659713169,
-      "grad_norm": 1.4451613426208496,
-      "learning_rate": 7.031457286432162e-05,
-      "loss": 5.4585,
-      "step": 30047
-    },
-    {
-      "epoch": 15.670404172099087,
-      "grad_norm": 1.4272741079330444,
-      "learning_rate": 7.031356783919598e-05,
-      "loss": 5.7499,
-      "step": 30048
-    },
-    {
-      "epoch": 15.670925684485006,
-      "grad_norm": 1.3623384237289429,
-      "learning_rate": 7.031256281407036e-05,
-      "loss": 5.7812,
-      "step": 30049
-    },
-    {
-      "epoch": 15.671447196870925,
-      "grad_norm": 1.5727695226669312,
-      "learning_rate": 7.031155778894472e-05,
-      "loss": 5.0478,
-      "step": 30050
-    },
-    {
-      "epoch": 15.671968709256845,
-      "grad_norm": 1.4335187673568726,
-      "learning_rate": 7.03105527638191e-05,
-      "loss": 5.6912,
-      "step": 30051
-    },
-    {
-      "epoch": 15.672490221642764,
-      "grad_norm": 1.4679251909255981,
-      "learning_rate": 7.030954773869347e-05,
-      "loss": 5.8462,
-      "step": 30052
-    },
-    {
-      "epoch": 15.673011734028684,
-      "grad_norm": 1.4858371019363403,
-      "learning_rate": 7.030854271356784e-05,
-      "loss": 5.268,
-      "step": 30053
-    },
-    {
-      "epoch": 15.673533246414602,
-      "grad_norm": 1.565849781036377,
-      "learning_rate": 7.030753768844222e-05,
-      "loss": 4.897,
-      "step": 30054
-    },
-    {
-      "epoch": 15.674054758800521,
-      "grad_norm": 1.4003568887710571,
-      "learning_rate": 7.030653266331659e-05,
-      "loss": 5.8525,
-      "step": 30055
-    },
-    {
-      "epoch": 15.67457627118644,
-      "grad_norm": 1.5551717281341553,
-      "learning_rate": 7.030552763819096e-05,
-      "loss": 5.2306,
-      "step": 30056
-    },
-    {
-      "epoch": 15.67509778357236,
-      "grad_norm": 1.4572714567184448,
-      "learning_rate": 7.030452261306533e-05,
-      "loss": 5.17,
-      "step": 30057
-    },
-    {
-      "epoch": 15.67561929595828,
-      "grad_norm": 1.478036642074585,
-      "learning_rate": 7.03035175879397e-05,
-      "loss": 5.2421,
-      "step": 30058
-    },
-    {
-      "epoch": 15.676140808344199,
-      "grad_norm": 1.5803945064544678,
-      "learning_rate": 7.030251256281407e-05,
-      "loss": 5.1637,
-      "step": 30059
-    },
-    {
-      "epoch": 15.676662320730117,
-      "grad_norm": 1.4095864295959473,
-      "learning_rate": 7.030150753768845e-05,
-      "loss": 5.145,
-      "step": 30060
-    },
-    {
-      "epoch": 15.677183833116036,
-      "grad_norm": 1.631874680519104,
-      "learning_rate": 7.030050251256281e-05,
-      "loss": 4.9089,
-      "step": 30061
-    },
-    {
-      "epoch": 15.677705345501955,
-      "grad_norm": 1.5062364339828491,
-      "learning_rate": 7.029949748743719e-05,
-      "loss": 5.4427,
-      "step": 30062
-    },
-    {
-      "epoch": 15.678226857887875,
-      "grad_norm": 1.4489744901657104,
-      "learning_rate": 7.029849246231155e-05,
-      "loss": 5.37,
-      "step": 30063
-    },
-    {
-      "epoch": 15.678748370273794,
-      "grad_norm": 1.4426242113113403,
-      "learning_rate": 7.029748743718593e-05,
-      "loss": 5.3656,
-      "step": 30064
-    },
-    {
-      "epoch": 15.679269882659714,
-      "grad_norm": 1.4968575239181519,
-      "learning_rate": 7.029648241206031e-05,
-      "loss": 5.1945,
-      "step": 30065
-    },
-    {
-      "epoch": 15.679791395045632,
-      "grad_norm": 1.4142004251480103,
-      "learning_rate": 7.029547738693467e-05,
-      "loss": 5.7279,
-      "step": 30066
-    },
-    {
-      "epoch": 15.680312907431551,
-      "grad_norm": 1.4767961502075195,
-      "learning_rate": 7.029447236180905e-05,
-      "loss": 5.3014,
-      "step": 30067
-    },
-    {
-      "epoch": 15.68083441981747,
-      "grad_norm": 1.4306780099868774,
-      "learning_rate": 7.029346733668342e-05,
-      "loss": 5.2876,
-      "step": 30068
-    },
-    {
-      "epoch": 15.68135593220339,
-      "grad_norm": 1.476169466972351,
-      "learning_rate": 7.02924623115578e-05,
-      "loss": 5.2493,
-      "step": 30069
-    },
-    {
-      "epoch": 15.68187744458931,
-      "grad_norm": 1.5283794403076172,
-      "learning_rate": 7.029145728643216e-05,
-      "loss": 5.1722,
-      "step": 30070
-    },
-    {
-      "epoch": 15.682398956975229,
-      "grad_norm": 1.491104245185852,
-      "learning_rate": 7.029045226130654e-05,
-      "loss": 5.1334,
-      "step": 30071
-    },
-    {
-      "epoch": 15.682920469361147,
-      "grad_norm": 1.72818922996521,
-      "learning_rate": 7.02894472361809e-05,
-      "loss": 5.3454,
-      "step": 30072
-    },
-    {
-      "epoch": 15.683441981747066,
-      "grad_norm": 1.4341708421707153,
-      "learning_rate": 7.028844221105528e-05,
-      "loss": 4.7769,
-      "step": 30073
-    },
-    {
-      "epoch": 15.683963494132986,
-      "grad_norm": 1.4728963375091553,
-      "learning_rate": 7.028743718592966e-05,
-      "loss": 4.9205,
-      "step": 30074
-    },
-    {
-      "epoch": 15.684485006518905,
-      "grad_norm": 1.5121079683303833,
-      "learning_rate": 7.028643216080403e-05,
-      "loss": 5.5626,
-      "step": 30075
-    },
-    {
-      "epoch": 15.685006518904824,
-      "grad_norm": 1.4936178922653198,
-      "learning_rate": 7.02854271356784e-05,
-      "loss": 5.5618,
-      "step": 30076
-    },
-    {
-      "epoch": 15.685528031290744,
-      "grad_norm": 1.515897512435913,
-      "learning_rate": 7.028442211055278e-05,
-      "loss": 5.4672,
-      "step": 30077
-    },
-    {
-      "epoch": 15.686049543676662,
-      "grad_norm": 2.133845806121826,
-      "learning_rate": 7.028341708542714e-05,
-      "loss": 5.0213,
-      "step": 30078
-    },
-    {
-      "epoch": 15.686571056062581,
-      "grad_norm": 1.446158528327942,
-      "learning_rate": 7.02824120603015e-05,
-      "loss": 5.3292,
-      "step": 30079
-    },
-    {
-      "epoch": 15.6870925684485,
-      "grad_norm": 1.6821244955062866,
-      "learning_rate": 7.028140703517588e-05,
-      "loss": 5.1643,
-      "step": 30080
-    },
-    {
-      "epoch": 15.68761408083442,
-      "grad_norm": 1.4575139284133911,
-      "learning_rate": 7.028040201005025e-05,
-      "loss": 5.3318,
-      "step": 30081
-    },
-    {
-      "epoch": 15.68813559322034,
-      "grad_norm": 1.480717658996582,
-      "learning_rate": 7.027939698492462e-05,
-      "loss": 5.3023,
-      "step": 30082
-    },
-    {
-      "epoch": 15.688657105606259,
-      "grad_norm": 1.512333869934082,
-      "learning_rate": 7.027839195979899e-05,
-      "loss": 5.2625,
-      "step": 30083
-    },
-    {
-      "epoch": 15.689178617992177,
-      "grad_norm": 1.5238534212112427,
-      "learning_rate": 7.027738693467337e-05,
-      "loss": 5.4947,
-      "step": 30084
-    },
-    {
-      "epoch": 15.689700130378096,
-      "grad_norm": 1.4411053657531738,
-      "learning_rate": 7.027638190954774e-05,
-      "loss": 4.9276,
-      "step": 30085
-    },
-    {
-      "epoch": 15.690221642764016,
-      "grad_norm": 1.5673589706420898,
-      "learning_rate": 7.027537688442212e-05,
-      "loss": 5.4591,
-      "step": 30086
-    },
-    {
-      "epoch": 15.690743155149935,
-      "grad_norm": 1.546901822090149,
-      "learning_rate": 7.027437185929649e-05,
-      "loss": 5.3921,
-      "step": 30087
-    },
-    {
-      "epoch": 15.691264667535854,
-      "grad_norm": 1.443274974822998,
-      "learning_rate": 7.027336683417086e-05,
-      "loss": 5.4802,
-      "step": 30088
-    },
-    {
-      "epoch": 15.691786179921774,
-      "grad_norm": 1.6030664443969727,
-      "learning_rate": 7.027236180904523e-05,
-      "loss": 5.5033,
-      "step": 30089
-    },
-    {
-      "epoch": 15.692307692307692,
-      "grad_norm": 1.4636894464492798,
-      "learning_rate": 7.02713567839196e-05,
-      "loss": 5.1709,
-      "step": 30090
-    },
-    {
-      "epoch": 15.692829204693611,
-      "grad_norm": 1.4852144718170166,
-      "learning_rate": 7.027035175879397e-05,
-      "loss": 5.3215,
-      "step": 30091
-    },
-    {
-      "epoch": 15.69335071707953,
-      "grad_norm": 1.4966167211532593,
-      "learning_rate": 7.026934673366834e-05,
-      "loss": 5.5281,
-      "step": 30092
-    },
-    {
-      "epoch": 15.69387222946545,
-      "grad_norm": 1.5151972770690918,
-      "learning_rate": 7.026834170854271e-05,
-      "loss": 5.2597,
-      "step": 30093
-    },
-    {
-      "epoch": 15.69439374185137,
-      "grad_norm": 1.4905990362167358,
-      "learning_rate": 7.026733668341709e-05,
-      "loss": 5.5019,
-      "step": 30094
-    },
-    {
-      "epoch": 15.694915254237289,
-      "grad_norm": 1.3769769668579102,
-      "learning_rate": 7.026633165829147e-05,
-      "loss": 5.4832,
-      "step": 30095
-    },
-    {
-      "epoch": 15.695436766623207,
-      "grad_norm": 1.3926029205322266,
-      "learning_rate": 7.026532663316583e-05,
-      "loss": 5.5165,
-      "step": 30096
-    },
-    {
-      "epoch": 15.695958279009126,
-      "grad_norm": 1.3695495128631592,
-      "learning_rate": 7.026432160804021e-05,
-      "loss": 5.3511,
-      "step": 30097
-    },
-    {
-      "epoch": 15.696479791395046,
-      "grad_norm": 1.3098804950714111,
-      "learning_rate": 7.026331658291458e-05,
-      "loss": 5.3845,
-      "step": 30098
-    },
-    {
-      "epoch": 15.697001303780965,
-      "grad_norm": 1.4412856101989746,
-      "learning_rate": 7.026231155778895e-05,
-      "loss": 5.3102,
-      "step": 30099
-    },
-    {
-      "epoch": 15.697522816166884,
-      "grad_norm": 1.56721031665802,
-      "learning_rate": 7.026130653266332e-05,
-      "loss": 5.2577,
-      "step": 30100
-    },
-    {
-      "epoch": 15.698044328552804,
-      "grad_norm": 1.6231889724731445,
-      "learning_rate": 7.02603015075377e-05,
-      "loss": 4.7194,
-      "step": 30101
-    },
-    {
-      "epoch": 15.698565840938722,
-      "grad_norm": 1.4875494241714478,
-      "learning_rate": 7.025929648241206e-05,
-      "loss": 5.2453,
-      "step": 30102
-    },
-    {
-      "epoch": 15.699087353324641,
-      "grad_norm": 1.4730647802352905,
-      "learning_rate": 7.025829145728644e-05,
-      "loss": 5.741,
-      "step": 30103
-    },
-    {
-      "epoch": 15.69960886571056,
-      "grad_norm": 1.4078725576400757,
-      "learning_rate": 7.02572864321608e-05,
-      "loss": 5.9139,
-      "step": 30104
-    },
-    {
-      "epoch": 15.70013037809648,
-      "grad_norm": 1.502928376197815,
-      "learning_rate": 7.025628140703518e-05,
-      "loss": 5.3416,
-      "step": 30105
-    },
-    {
-      "epoch": 15.7006518904824,
-      "grad_norm": 1.4487597942352295,
-      "learning_rate": 7.025527638190956e-05,
-      "loss": 5.7301,
-      "step": 30106
-    },
-    {
-      "epoch": 15.701173402868317,
-      "grad_norm": 1.526648998260498,
-      "learning_rate": 7.025427135678392e-05,
-      "loss": 5.0563,
-      "step": 30107
-    },
-    {
-      "epoch": 15.701694915254237,
-      "grad_norm": 1.458340048789978,
-      "learning_rate": 7.02532663316583e-05,
-      "loss": 5.3989,
-      "step": 30108
-    },
-    {
-      "epoch": 15.702216427640156,
-      "grad_norm": 1.5044630765914917,
-      "learning_rate": 7.025226130653266e-05,
-      "loss": 5.5301,
-      "step": 30109
-    },
-    {
-      "epoch": 15.702737940026076,
-      "grad_norm": 1.3620439767837524,
-      "learning_rate": 7.025125628140704e-05,
-      "loss": 5.7877,
-      "step": 30110
-    },
-    {
-      "epoch": 15.703259452411995,
-      "grad_norm": 1.4689435958862305,
-      "learning_rate": 7.02502512562814e-05,
-      "loss": 5.4506,
-      "step": 30111
-    },
-    {
-      "epoch": 15.703780964797915,
-      "grad_norm": 1.4611653089523315,
-      "learning_rate": 7.024924623115578e-05,
-      "loss": 5.1924,
-      "step": 30112
-    },
-    {
-      "epoch": 15.704302477183834,
-      "grad_norm": 1.5253775119781494,
-      "learning_rate": 7.024824120603015e-05,
-      "loss": 5.2959,
-      "step": 30113
-    },
-    {
-      "epoch": 15.704823989569752,
-      "grad_norm": 1.468432068824768,
-      "learning_rate": 7.024723618090453e-05,
-      "loss": 5.2127,
-      "step": 30114
-    },
-    {
-      "epoch": 15.705345501955671,
-      "grad_norm": 1.5194251537322998,
-      "learning_rate": 7.02462311557789e-05,
-      "loss": 5.3093,
-      "step": 30115
-    },
-    {
-      "epoch": 15.70586701434159,
-      "grad_norm": 1.569901704788208,
-      "learning_rate": 7.024522613065328e-05,
-      "loss": 5.6463,
-      "step": 30116
-    },
-    {
-      "epoch": 15.70638852672751,
-      "grad_norm": 1.4999823570251465,
-      "learning_rate": 7.024422110552765e-05,
-      "loss": 5.6806,
-      "step": 30117
-    },
-    {
-      "epoch": 15.70691003911343,
-      "grad_norm": 1.480406641960144,
-      "learning_rate": 7.024321608040202e-05,
-      "loss": 5.3573,
-      "step": 30118
-    },
-    {
-      "epoch": 15.707431551499347,
-      "grad_norm": 1.4196051359176636,
-      "learning_rate": 7.024221105527639e-05,
-      "loss": 5.4391,
-      "step": 30119
-    },
-    {
-      "epoch": 15.707953063885267,
-      "grad_norm": 1.4991166591644287,
-      "learning_rate": 7.024120603015075e-05,
-      "loss": 5.3981,
-      "step": 30120
-    },
-    {
-      "epoch": 15.708474576271186,
-      "grad_norm": 1.5334124565124512,
-      "learning_rate": 7.024020100502513e-05,
-      "loss": 5.1219,
-      "step": 30121
-    },
-    {
-      "epoch": 15.708996088657106,
-      "grad_norm": 1.4536302089691162,
-      "learning_rate": 7.02391959798995e-05,
-      "loss": 5.4865,
-      "step": 30122
-    },
-    {
-      "epoch": 15.709517601043025,
-      "grad_norm": 1.5017125606536865,
-      "learning_rate": 7.023819095477387e-05,
-      "loss": 4.9673,
-      "step": 30123
-    },
-    {
-      "epoch": 15.710039113428945,
-      "grad_norm": 1.4831936359405518,
-      "learning_rate": 7.023718592964824e-05,
-      "loss": 5.1858,
-      "step": 30124
-    },
-    {
-      "epoch": 15.710560625814864,
-      "grad_norm": 1.379374384880066,
-      "learning_rate": 7.023618090452261e-05,
-      "loss": 5.5918,
-      "step": 30125
-    },
-    {
-      "epoch": 15.711082138200782,
-      "grad_norm": 1.5081045627593994,
-      "learning_rate": 7.023517587939699e-05,
-      "loss": 5.1555,
-      "step": 30126
-    },
-    {
-      "epoch": 15.711603650586701,
-      "grad_norm": 1.414910078048706,
-      "learning_rate": 7.023417085427137e-05,
-      "loss": 5.789,
-      "step": 30127
-    },
-    {
-      "epoch": 15.71212516297262,
-      "grad_norm": 1.4945827722549438,
-      "learning_rate": 7.023316582914573e-05,
-      "loss": 5.3886,
-      "step": 30128
-    },
-    {
-      "epoch": 15.71264667535854,
-      "grad_norm": 1.4862589836120605,
-      "learning_rate": 7.023216080402011e-05,
-      "loss": 5.1549,
-      "step": 30129
-    },
-    {
-      "epoch": 15.71316818774446,
-      "grad_norm": 1.437087059020996,
-      "learning_rate": 7.023115577889448e-05,
-      "loss": 5.8176,
-      "step": 30130
-    },
-    {
-      "epoch": 15.713689700130377,
-      "grad_norm": 1.4808441400527954,
-      "learning_rate": 7.023015075376885e-05,
-      "loss": 5.5085,
-      "step": 30131
-    },
-    {
-      "epoch": 15.714211212516297,
-      "grad_norm": 1.364935040473938,
-      "learning_rate": 7.022914572864322e-05,
-      "loss": 5.461,
-      "step": 30132
-    },
-    {
-      "epoch": 15.714732724902216,
-      "grad_norm": 1.5012834072113037,
-      "learning_rate": 7.022814070351758e-05,
-      "loss": 5.357,
-      "step": 30133
-    },
-    {
-      "epoch": 15.715254237288136,
-      "grad_norm": 1.3821066617965698,
-      "learning_rate": 7.022713567839196e-05,
-      "loss": 5.3101,
-      "step": 30134
-    },
-    {
-      "epoch": 15.715775749674055,
-      "grad_norm": 1.281141996383667,
-      "learning_rate": 7.022613065326632e-05,
-      "loss": 5.901,
-      "step": 30135
-    },
-    {
-      "epoch": 15.716297262059975,
-      "grad_norm": 1.3726869821548462,
-      "learning_rate": 7.02251256281407e-05,
-      "loss": 5.5845,
-      "step": 30136
-    },
-    {
-      "epoch": 15.716818774445892,
-      "grad_norm": 1.4535654783248901,
-      "learning_rate": 7.022412060301508e-05,
-      "loss": 5.5668,
-      "step": 30137
-    },
-    {
-      "epoch": 15.717340286831812,
-      "grad_norm": 1.486040472984314,
-      "learning_rate": 7.022311557788946e-05,
-      "loss": 5.7832,
-      "step": 30138
-    },
-    {
-      "epoch": 15.717861799217731,
-      "grad_norm": 1.5427513122558594,
-      "learning_rate": 7.022211055276382e-05,
-      "loss": 5.0528,
-      "step": 30139
-    },
-    {
-      "epoch": 15.71838331160365,
-      "grad_norm": 1.3634629249572754,
-      "learning_rate": 7.02211055276382e-05,
-      "loss": 4.7972,
-      "step": 30140
-    },
-    {
-      "epoch": 15.71890482398957,
-      "grad_norm": 1.4050172567367554,
-      "learning_rate": 7.022010050251256e-05,
-      "loss": 5.7308,
-      "step": 30141
-    },
-    {
-      "epoch": 15.71942633637549,
-      "grad_norm": 1.4565869569778442,
-      "learning_rate": 7.021909547738694e-05,
-      "loss": 5.2719,
-      "step": 30142
-    },
-    {
-      "epoch": 15.719947848761407,
-      "grad_norm": 1.4980944395065308,
-      "learning_rate": 7.02180904522613e-05,
-      "loss": 5.7321,
-      "step": 30143
-    },
-    {
-      "epoch": 15.720469361147327,
-      "grad_norm": 1.4566271305084229,
-      "learning_rate": 7.021708542713568e-05,
-      "loss": 5.284,
-      "step": 30144
-    },
-    {
-      "epoch": 15.720990873533246,
-      "grad_norm": 1.5803946256637573,
-      "learning_rate": 7.021608040201005e-05,
-      "loss": 5.2503,
-      "step": 30145
-    },
-    {
-      "epoch": 15.721512385919166,
-      "grad_norm": 1.444300651550293,
-      "learning_rate": 7.021507537688443e-05,
-      "loss": 5.74,
-      "step": 30146
-    },
-    {
-      "epoch": 15.722033898305085,
-      "grad_norm": 1.4602336883544922,
-      "learning_rate": 7.02140703517588e-05,
-      "loss": 5.9577,
-      "step": 30147
-    },
-    {
-      "epoch": 15.722555410691005,
-      "grad_norm": 1.4776623249053955,
-      "learning_rate": 7.021306532663317e-05,
-      "loss": 5.7194,
-      "step": 30148
-    },
-    {
-      "epoch": 15.723076923076922,
-      "grad_norm": 1.5597072839736938,
-      "learning_rate": 7.021206030150755e-05,
-      "loss": 5.1718,
-      "step": 30149
-    },
-    {
-      "epoch": 15.723598435462842,
-      "grad_norm": 1.5934633016586304,
-      "learning_rate": 7.021105527638191e-05,
-      "loss": 5.3503,
-      "step": 30150
-    },
-    {
-      "epoch": 15.724119947848761,
-      "grad_norm": 1.513995885848999,
-      "learning_rate": 7.021005025125629e-05,
-      "loss": 5.1708,
-      "step": 30151
-    },
-    {
-      "epoch": 15.72464146023468,
-      "grad_norm": 1.4050647020339966,
-      "learning_rate": 7.020904522613065e-05,
-      "loss": 5.5868,
-      "step": 30152
-    },
-    {
-      "epoch": 15.7251629726206,
-      "grad_norm": 1.4873361587524414,
-      "learning_rate": 7.020804020100503e-05,
-      "loss": 5.2993,
-      "step": 30153
-    },
-    {
-      "epoch": 15.72568448500652,
-      "grad_norm": 1.5016478300094604,
-      "learning_rate": 7.02070351758794e-05,
-      "loss": 5.6047,
-      "step": 30154
-    },
-    {
-      "epoch": 15.726205997392437,
-      "grad_norm": 1.4087773561477661,
-      "learning_rate": 7.020603015075377e-05,
-      "loss": 5.6429,
-      "step": 30155
-    },
-    {
-      "epoch": 15.726727509778357,
-      "grad_norm": 1.486862301826477,
-      "learning_rate": 7.020502512562814e-05,
-      "loss": 5.0503,
-      "step": 30156
-    },
-    {
-      "epoch": 15.727249022164276,
-      "grad_norm": 1.3419684171676636,
-      "learning_rate": 7.020402010050251e-05,
-      "loss": 5.8188,
-      "step": 30157
-    },
-    {
-      "epoch": 15.727770534550196,
-      "grad_norm": 1.3484305143356323,
-      "learning_rate": 7.020301507537689e-05,
-      "loss": 5.7336,
-      "step": 30158
-    },
-    {
-      "epoch": 15.728292046936115,
-      "grad_norm": 1.397395372390747,
-      "learning_rate": 7.020201005025126e-05,
-      "loss": 5.7866,
-      "step": 30159
-    },
-    {
-      "epoch": 15.728813559322035,
-      "grad_norm": 1.437585711479187,
-      "learning_rate": 7.020100502512563e-05,
-      "loss": 5.0045,
-      "step": 30160
-    },
-    {
-      "epoch": 15.729335071707952,
-      "grad_norm": 1.348577618598938,
-      "learning_rate": 7.02e-05,
-      "loss": 5.7505,
-      "step": 30161
-    },
-    {
-      "epoch": 15.729856584093872,
-      "grad_norm": 1.6531016826629639,
-      "learning_rate": 7.019899497487438e-05,
-      "loss": 4.7938,
-      "step": 30162
-    },
-    {
-      "epoch": 15.730378096479791,
-      "grad_norm": 1.3978906869888306,
-      "learning_rate": 7.019798994974874e-05,
-      "loss": 5.3425,
-      "step": 30163
-    },
-    {
-      "epoch": 15.73089960886571,
-      "grad_norm": 1.4985617399215698,
-      "learning_rate": 7.019698492462312e-05,
-      "loss": 5.2903,
-      "step": 30164
-    },
-    {
-      "epoch": 15.73142112125163,
-      "grad_norm": 1.495753526687622,
-      "learning_rate": 7.019597989949748e-05,
-      "loss": 5.2949,
-      "step": 30165
-    },
-    {
-      "epoch": 15.73194263363755,
-      "grad_norm": 1.5717226266860962,
-      "learning_rate": 7.019497487437186e-05,
-      "loss": 5.5327,
-      "step": 30166
-    },
-    {
-      "epoch": 15.732464146023467,
-      "grad_norm": 1.5810070037841797,
-      "learning_rate": 7.019396984924624e-05,
-      "loss": 5.518,
-      "step": 30167
-    },
-    {
-      "epoch": 15.732985658409387,
-      "grad_norm": 1.4294079542160034,
-      "learning_rate": 7.019296482412062e-05,
-      "loss": 5.162,
-      "step": 30168
-    },
-    {
-      "epoch": 15.733507170795306,
-      "grad_norm": 1.4951432943344116,
-      "learning_rate": 7.019195979899498e-05,
-      "loss": 5.3655,
-      "step": 30169
-    },
-    {
-      "epoch": 15.734028683181226,
-      "grad_norm": 1.3723925352096558,
-      "learning_rate": 7.019095477386936e-05,
-      "loss": 5.7716,
-      "step": 30170
-    },
-    {
-      "epoch": 15.734550195567145,
-      "grad_norm": 1.3757320642471313,
-      "learning_rate": 7.018994974874372e-05,
-      "loss": 5.5961,
-      "step": 30171
-    },
-    {
-      "epoch": 15.735071707953065,
-      "grad_norm": 1.527732491493225,
-      "learning_rate": 7.018894472361809e-05,
-      "loss": 5.2398,
-      "step": 30172
-    },
-    {
-      "epoch": 15.735593220338982,
-      "grad_norm": 1.4175294637680054,
-      "learning_rate": 7.018793969849246e-05,
-      "loss": 5.2482,
-      "step": 30173
-    },
-    {
-      "epoch": 15.736114732724902,
-      "grad_norm": 1.5223379135131836,
-      "learning_rate": 7.018693467336683e-05,
-      "loss": 4.8564,
-      "step": 30174
-    },
-    {
-      "epoch": 15.736636245110821,
-      "grad_norm": 1.5282152891159058,
-      "learning_rate": 7.018592964824121e-05,
-      "loss": 5.3646,
-      "step": 30175
-    },
-    {
-      "epoch": 15.73715775749674,
-      "grad_norm": 1.4716777801513672,
-      "learning_rate": 7.018492462311557e-05,
-      "loss": 5.467,
-      "step": 30176
-    },
-    {
-      "epoch": 15.73767926988266,
-      "grad_norm": 1.4521106481552124,
-      "learning_rate": 7.018391959798995e-05,
-      "loss": 5.671,
-      "step": 30177
-    },
-    {
-      "epoch": 15.73820078226858,
-      "grad_norm": 1.5095267295837402,
-      "learning_rate": 7.018291457286433e-05,
-      "loss": 4.8514,
-      "step": 30178
-    },
-    {
-      "epoch": 15.738722294654497,
-      "grad_norm": 1.5374425649642944,
-      "learning_rate": 7.01819095477387e-05,
-      "loss": 5.4933,
-      "step": 30179
-    },
-    {
-      "epoch": 15.739243807040417,
-      "grad_norm": 1.4597798585891724,
-      "learning_rate": 7.018090452261307e-05,
-      "loss": 5.2971,
-      "step": 30180
-    },
-    {
-      "epoch": 15.739765319426336,
-      "grad_norm": 1.4977372884750366,
-      "learning_rate": 7.017989949748745e-05,
-      "loss": 5.3486,
-      "step": 30181
-    },
-    {
-      "epoch": 15.740286831812256,
-      "grad_norm": 1.4383758306503296,
-      "learning_rate": 7.017889447236181e-05,
-      "loss": 5.6902,
-      "step": 30182
-    },
-    {
-      "epoch": 15.740808344198175,
-      "grad_norm": 1.5290420055389404,
-      "learning_rate": 7.017788944723619e-05,
-      "loss": 5.3454,
-      "step": 30183
-    },
-    {
-      "epoch": 15.741329856584095,
-      "grad_norm": 1.5572789907455444,
-      "learning_rate": 7.017688442211055e-05,
-      "loss": 5.4163,
-      "step": 30184
-    },
-    {
-      "epoch": 15.741851368970012,
-      "grad_norm": 1.4298185110092163,
-      "learning_rate": 7.017587939698492e-05,
-      "loss": 5.234,
-      "step": 30185
-    },
-    {
-      "epoch": 15.742372881355932,
-      "grad_norm": 1.3295146226882935,
-      "learning_rate": 7.01748743718593e-05,
-      "loss": 5.9275,
-      "step": 30186
-    },
-    {
-      "epoch": 15.742894393741851,
-      "grad_norm": 1.4323266744613647,
-      "learning_rate": 7.017386934673367e-05,
-      "loss": 5.6231,
-      "step": 30187
-    },
-    {
-      "epoch": 15.74341590612777,
-      "grad_norm": 1.5963854789733887,
-      "learning_rate": 7.017286432160805e-05,
-      "loss": 4.8824,
-      "step": 30188
-    },
-    {
-      "epoch": 15.74393741851369,
-      "grad_norm": 1.3458635807037354,
-      "learning_rate": 7.017185929648242e-05,
-      "loss": 5.7491,
-      "step": 30189
-    },
-    {
-      "epoch": 15.74445893089961,
-      "grad_norm": 1.4752349853515625,
-      "learning_rate": 7.017085427135679e-05,
-      "loss": 5.607,
-      "step": 30190
-    },
-    {
-      "epoch": 15.744980443285527,
-      "grad_norm": 1.4376788139343262,
-      "learning_rate": 7.016984924623116e-05,
-      "loss": 5.6167,
-      "step": 30191
-    },
-    {
-      "epoch": 15.745501955671447,
-      "grad_norm": 1.5190824270248413,
-      "learning_rate": 7.016884422110554e-05,
-      "loss": 5.164,
-      "step": 30192
-    },
-    {
-      "epoch": 15.746023468057366,
-      "grad_norm": 1.416258454322815,
-      "learning_rate": 7.01678391959799e-05,
-      "loss": 5.3686,
-      "step": 30193
-    },
-    {
-      "epoch": 15.746544980443286,
-      "grad_norm": 1.5090930461883545,
-      "learning_rate": 7.016683417085428e-05,
-      "loss": 5.5298,
-      "step": 30194
-    },
-    {
-      "epoch": 15.747066492829205,
-      "grad_norm": 1.510217308998108,
-      "learning_rate": 7.016582914572864e-05,
-      "loss": 4.8194,
-      "step": 30195
-    },
-    {
-      "epoch": 15.747588005215125,
-      "grad_norm": 1.4666091203689575,
-      "learning_rate": 7.016482412060302e-05,
-      "loss": 5.7695,
-      "step": 30196
-    },
-    {
-      "epoch": 15.748109517601042,
-      "grad_norm": 1.5803765058517456,
-      "learning_rate": 7.016381909547738e-05,
-      "loss": 4.9248,
-      "step": 30197
-    },
-    {
-      "epoch": 15.748631029986962,
-      "grad_norm": 1.4112569093704224,
-      "learning_rate": 7.016281407035176e-05,
-      "loss": 5.0087,
-      "step": 30198
-    },
-    {
-      "epoch": 15.749152542372881,
-      "grad_norm": 1.4698779582977295,
-      "learning_rate": 7.016180904522614e-05,
-      "loss": 5.3277,
-      "step": 30199
-    },
-    {
-      "epoch": 15.7496740547588,
-      "grad_norm": 1.4512596130371094,
-      "learning_rate": 7.01608040201005e-05,
-      "loss": 5.5048,
-      "step": 30200
-    },
-    {
-      "epoch": 15.75019556714472,
-      "grad_norm": 1.3966971635818481,
-      "learning_rate": 7.015979899497488e-05,
-      "loss": 5.6992,
-      "step": 30201
-    },
-    {
-      "epoch": 15.750717079530638,
-      "grad_norm": 1.435630202293396,
-      "learning_rate": 7.015879396984925e-05,
-      "loss": 5.341,
-      "step": 30202
-    },
-    {
-      "epoch": 15.751238591916557,
-      "grad_norm": 1.3827801942825317,
-      "learning_rate": 7.015778894472362e-05,
-      "loss": 5.1577,
-      "step": 30203
-    },
-    {
-      "epoch": 15.751760104302477,
-      "grad_norm": 1.4235506057739258,
-      "learning_rate": 7.015678391959799e-05,
-      "loss": 5.4024,
-      "step": 30204
-    },
-    {
-      "epoch": 15.752281616688396,
-      "grad_norm": 1.44083833694458,
-      "learning_rate": 7.015577889447237e-05,
-      "loss": 4.8322,
-      "step": 30205
-    },
-    {
-      "epoch": 15.752803129074316,
-      "grad_norm": 1.4391957521438599,
-      "learning_rate": 7.015477386934673e-05,
-      "loss": 4.8388,
-      "step": 30206
-    },
-    {
-      "epoch": 15.753324641460235,
-      "grad_norm": 1.4313545227050781,
-      "learning_rate": 7.015376884422111e-05,
-      "loss": 5.0861,
-      "step": 30207
-    },
-    {
-      "epoch": 15.753846153846155,
-      "grad_norm": 1.5501798391342163,
-      "learning_rate": 7.015276381909549e-05,
-      "loss": 4.9923,
-      "step": 30208
-    },
-    {
-      "epoch": 15.754367666232072,
-      "grad_norm": 1.4115206003189087,
-      "learning_rate": 7.015175879396986e-05,
-      "loss": 5.3569,
-      "step": 30209
-    },
-    {
-      "epoch": 15.754889178617992,
-      "grad_norm": 1.3633978366851807,
-      "learning_rate": 7.015075376884423e-05,
-      "loss": 5.4291,
-      "step": 30210
-    },
-    {
-      "epoch": 15.755410691003911,
-      "grad_norm": 1.4322537183761597,
-      "learning_rate": 7.01497487437186e-05,
-      "loss": 4.844,
-      "step": 30211
-    },
-    {
-      "epoch": 15.75593220338983,
-      "grad_norm": 1.370767593383789,
-      "learning_rate": 7.014874371859297e-05,
-      "loss": 5.597,
-      "step": 30212
-    },
-    {
-      "epoch": 15.75645371577575,
-      "grad_norm": 1.5235148668289185,
-      "learning_rate": 7.014773869346733e-05,
-      "loss": 5.3128,
-      "step": 30213
-    },
-    {
-      "epoch": 15.756975228161668,
-      "grad_norm": 1.5164039134979248,
-      "learning_rate": 7.014673366834171e-05,
-      "loss": 5.1904,
-      "step": 30214
-    },
-    {
-      "epoch": 15.757496740547587,
-      "grad_norm": 1.426954984664917,
-      "learning_rate": 7.014572864321608e-05,
-      "loss": 5.1578,
-      "step": 30215
-    },
-    {
-      "epoch": 15.758018252933507,
-      "grad_norm": 1.465244174003601,
-      "learning_rate": 7.014472361809045e-05,
-      "loss": 5.2268,
-      "step": 30216
-    },
-    {
-      "epoch": 15.758539765319426,
-      "grad_norm": 1.4287030696868896,
-      "learning_rate": 7.014371859296482e-05,
-      "loss": 5.0801,
-      "step": 30217
-    },
-    {
-      "epoch": 15.759061277705346,
-      "grad_norm": 1.5752320289611816,
-      "learning_rate": 7.01427135678392e-05,
-      "loss": 5.6727,
-      "step": 30218
-    },
-    {
-      "epoch": 15.759582790091265,
-      "grad_norm": 1.5236029624938965,
-      "learning_rate": 7.014170854271357e-05,
-      "loss": 5.4103,
-      "step": 30219
-    },
-    {
-      "epoch": 15.760104302477185,
-      "grad_norm": 1.4693915843963623,
-      "learning_rate": 7.014070351758795e-05,
-      "loss": 5.785,
-      "step": 30220
-    },
-    {
-      "epoch": 15.760625814863102,
-      "grad_norm": 1.4478628635406494,
-      "learning_rate": 7.013969849246232e-05,
-      "loss": 5.7221,
-      "step": 30221
-    },
-    {
-      "epoch": 15.761147327249022,
-      "grad_norm": 1.5109699964523315,
-      "learning_rate": 7.01386934673367e-05,
-      "loss": 5.4545,
-      "step": 30222
-    },
-    {
-      "epoch": 15.761668839634941,
-      "grad_norm": 1.5151166915893555,
-      "learning_rate": 7.013768844221106e-05,
-      "loss": 5.2182,
-      "step": 30223
-    },
-    {
-      "epoch": 15.76219035202086,
-      "grad_norm": 1.4266761541366577,
-      "learning_rate": 7.013668341708544e-05,
-      "loss": 5.2744,
-      "step": 30224
-    },
-    {
-      "epoch": 15.76271186440678,
-      "grad_norm": 1.4167400598526,
-      "learning_rate": 7.01356783919598e-05,
-      "loss": 5.2544,
-      "step": 30225
-    },
-    {
-      "epoch": 15.763233376792698,
-      "grad_norm": 1.5913041830062866,
-      "learning_rate": 7.013467336683416e-05,
-      "loss": 5.3824,
-      "step": 30226
-    },
-    {
-      "epoch": 15.763754889178617,
-      "grad_norm": 1.4887992143630981,
-      "learning_rate": 7.013366834170854e-05,
-      "loss": 5.6152,
-      "step": 30227
-    },
-    {
-      "epoch": 15.764276401564537,
-      "grad_norm": 1.49791419506073,
-      "learning_rate": 7.013266331658292e-05,
-      "loss": 5.4871,
-      "step": 30228
-    },
-    {
-      "epoch": 15.764797913950456,
-      "grad_norm": 1.4867668151855469,
-      "learning_rate": 7.01316582914573e-05,
-      "loss": 5.2592,
-      "step": 30229
-    },
-    {
-      "epoch": 15.765319426336376,
-      "grad_norm": 1.5742599964141846,
-      "learning_rate": 7.013065326633166e-05,
-      "loss": 5.2856,
-      "step": 30230
-    },
-    {
-      "epoch": 15.765840938722295,
-      "grad_norm": 1.4172331094741821,
-      "learning_rate": 7.012964824120604e-05,
-      "loss": 5.6812,
-      "step": 30231
-    },
-    {
-      "epoch": 15.766362451108213,
-      "grad_norm": 1.443945288658142,
-      "learning_rate": 7.01286432160804e-05,
-      "loss": 5.4395,
-      "step": 30232
-    },
-    {
-      "epoch": 15.766883963494132,
-      "grad_norm": 1.770742654800415,
-      "learning_rate": 7.012763819095478e-05,
-      "loss": 5.3092,
-      "step": 30233
-    },
-    {
-      "epoch": 15.767405475880052,
-      "grad_norm": 1.455727458000183,
-      "learning_rate": 7.012663316582915e-05,
-      "loss": 5.6324,
-      "step": 30234
-    },
-    {
-      "epoch": 15.767926988265971,
-      "grad_norm": 1.4393950700759888,
-      "learning_rate": 7.012562814070352e-05,
-      "loss": 5.4127,
-      "step": 30235
-    },
-    {
-      "epoch": 15.76844850065189,
-      "grad_norm": 1.4052542448043823,
-      "learning_rate": 7.012462311557789e-05,
-      "loss": 5.6617,
-      "step": 30236
-    },
-    {
-      "epoch": 15.76897001303781,
-      "grad_norm": 1.4439090490341187,
-      "learning_rate": 7.012361809045227e-05,
-      "loss": 5.6982,
-      "step": 30237
-    },
-    {
-      "epoch": 15.769491525423728,
-      "grad_norm": 1.579169511795044,
-      "learning_rate": 7.012261306532663e-05,
-      "loss": 4.7904,
-      "step": 30238
-    },
-    {
-      "epoch": 15.770013037809647,
-      "grad_norm": 1.7143194675445557,
-      "learning_rate": 7.012160804020101e-05,
-      "loss": 5.1403,
-      "step": 30239
-    },
-    {
-      "epoch": 15.770534550195567,
-      "grad_norm": 1.6813476085662842,
-      "learning_rate": 7.012060301507539e-05,
-      "loss": 5.4355,
-      "step": 30240
-    },
-    {
-      "epoch": 15.771056062581486,
-      "grad_norm": 1.523077368736267,
-      "learning_rate": 7.011959798994975e-05,
-      "loss": 5.1652,
-      "step": 30241
-    },
-    {
-      "epoch": 15.771577574967406,
-      "grad_norm": 1.5805588960647583,
-      "learning_rate": 7.011859296482413e-05,
-      "loss": 5.2963,
-      "step": 30242
-    },
-    {
-      "epoch": 15.772099087353325,
-      "grad_norm": 1.6018493175506592,
-      "learning_rate": 7.011758793969849e-05,
-      "loss": 5.2192,
-      "step": 30243
-    },
-    {
-      "epoch": 15.772620599739243,
-      "grad_norm": 1.6076209545135498,
-      "learning_rate": 7.011658291457287e-05,
-      "loss": 5.6409,
-      "step": 30244
-    },
-    {
-      "epoch": 15.773142112125162,
-      "grad_norm": 1.5822476148605347,
-      "learning_rate": 7.011557788944723e-05,
-      "loss": 5.0721,
-      "step": 30245
-    },
-    {
-      "epoch": 15.773663624511082,
-      "grad_norm": 1.4780622720718384,
-      "learning_rate": 7.011457286432161e-05,
-      "loss": 5.1289,
-      "step": 30246
-    },
-    {
-      "epoch": 15.774185136897001,
-      "grad_norm": 1.583699345588684,
-      "learning_rate": 7.011356783919598e-05,
-      "loss": 5.3503,
-      "step": 30247
-    },
-    {
-      "epoch": 15.77470664928292,
-      "grad_norm": 1.3723703622817993,
-      "learning_rate": 7.011256281407035e-05,
-      "loss": 5.8964,
-      "step": 30248
-    },
-    {
-      "epoch": 15.77522816166884,
-      "grad_norm": 1.5845354795455933,
-      "learning_rate": 7.011155778894473e-05,
-      "loss": 5.2421,
-      "step": 30249
-    },
-    {
-      "epoch": 15.775749674054758,
-      "grad_norm": 1.4369146823883057,
-      "learning_rate": 7.011055276381911e-05,
-      "loss": 5.6531,
-      "step": 30250
-    },
-    {
-      "epoch": 15.776271186440677,
-      "grad_norm": 1.5074659585952759,
-      "learning_rate": 7.010954773869347e-05,
-      "loss": 5.274,
-      "step": 30251
-    },
-    {
-      "epoch": 15.776792698826597,
-      "grad_norm": 1.5115385055541992,
-      "learning_rate": 7.010854271356784e-05,
-      "loss": 5.2251,
-      "step": 30252
-    },
-    {
-      "epoch": 15.777314211212516,
-      "grad_norm": 1.4266650676727295,
-      "learning_rate": 7.010753768844222e-05,
-      "loss": 5.2512,
-      "step": 30253
-    },
-    {
-      "epoch": 15.777835723598436,
-      "grad_norm": 1.4803398847579956,
-      "learning_rate": 7.010653266331658e-05,
-      "loss": 5.5569,
-      "step": 30254
-    },
-    {
-      "epoch": 15.778357235984355,
-      "grad_norm": 1.4643216133117676,
-      "learning_rate": 7.010552763819096e-05,
-      "loss": 5.5092,
-      "step": 30255
-    },
-    {
-      "epoch": 15.778878748370273,
-      "grad_norm": 1.4863308668136597,
-      "learning_rate": 7.010452261306532e-05,
-      "loss": 5.3568,
-      "step": 30256
-    },
-    {
-      "epoch": 15.779400260756193,
-      "grad_norm": 1.4342734813690186,
-      "learning_rate": 7.01035175879397e-05,
-      "loss": 5.7933,
-      "step": 30257
-    },
-    {
-      "epoch": 15.779921773142112,
-      "grad_norm": 1.4146156311035156,
-      "learning_rate": 7.010251256281407e-05,
-      "loss": 5.8327,
-      "step": 30258
-    },
-    {
-      "epoch": 15.780443285528031,
-      "grad_norm": 1.4502081871032715,
-      "learning_rate": 7.010150753768844e-05,
-      "loss": 5.606,
-      "step": 30259
-    },
-    {
-      "epoch": 15.780964797913951,
-      "grad_norm": 1.4534488916397095,
-      "learning_rate": 7.010050251256282e-05,
-      "loss": 5.5607,
-      "step": 30260
-    },
-    {
-      "epoch": 15.78148631029987,
-      "grad_norm": 1.457743525505066,
-      "learning_rate": 7.00994974874372e-05,
-      "loss": 5.5754,
-      "step": 30261
-    },
-    {
-      "epoch": 15.782007822685788,
-      "grad_norm": 1.4803214073181152,
-      "learning_rate": 7.009849246231156e-05,
-      "loss": 5.47,
-      "step": 30262
-    },
-    {
-      "epoch": 15.782529335071708,
-      "grad_norm": 1.5743012428283691,
-      "learning_rate": 7.009748743718594e-05,
-      "loss": 5.3209,
-      "step": 30263
-    },
-    {
-      "epoch": 15.783050847457627,
-      "grad_norm": 1.4814375638961792,
-      "learning_rate": 7.00964824120603e-05,
-      "loss": 5.7683,
-      "step": 30264
-    },
-    {
-      "epoch": 15.783572359843546,
-      "grad_norm": 1.4395512342453003,
-      "learning_rate": 7.009547738693467e-05,
-      "loss": 5.2735,
-      "step": 30265
-    },
-    {
-      "epoch": 15.784093872229466,
-      "grad_norm": 1.5958936214447021,
-      "learning_rate": 7.009447236180905e-05,
-      "loss": 4.9997,
-      "step": 30266
-    },
-    {
-      "epoch": 15.784615384615385,
-      "grad_norm": 1.4827110767364502,
-      "learning_rate": 7.009346733668341e-05,
-      "loss": 5.1913,
-      "step": 30267
-    },
-    {
-      "epoch": 15.785136897001303,
-      "grad_norm": 1.4316502809524536,
-      "learning_rate": 7.009246231155779e-05,
-      "loss": 5.1399,
-      "step": 30268
-    },
-    {
-      "epoch": 15.785658409387223,
-      "grad_norm": 1.4640116691589355,
-      "learning_rate": 7.009145728643217e-05,
-      "loss": 5.7357,
-      "step": 30269
-    },
-    {
-      "epoch": 15.786179921773142,
-      "grad_norm": 1.3993936777114868,
-      "learning_rate": 7.009045226130655e-05,
-      "loss": 5.2903,
-      "step": 30270
-    },
-    {
-      "epoch": 15.786701434159061,
-      "grad_norm": 1.4506365060806274,
-      "learning_rate": 7.008944723618091e-05,
-      "loss": 5.2692,
-      "step": 30271
-    },
-    {
-      "epoch": 15.787222946544981,
-      "grad_norm": 1.4127753973007202,
-      "learning_rate": 7.008844221105529e-05,
-      "loss": 5.5118,
-      "step": 30272
-    },
-    {
-      "epoch": 15.7877444589309,
-      "grad_norm": 1.3579820394515991,
-      "learning_rate": 7.008743718592965e-05,
-      "loss": 5.153,
-      "step": 30273
-    },
-    {
-      "epoch": 15.788265971316818,
-      "grad_norm": 1.4421052932739258,
-      "learning_rate": 7.008643216080403e-05,
-      "loss": 4.9209,
-      "step": 30274
-    },
-    {
-      "epoch": 15.788787483702738,
-      "grad_norm": 1.3605066537857056,
-      "learning_rate": 7.00854271356784e-05,
-      "loss": 5.5654,
-      "step": 30275
-    },
-    {
-      "epoch": 15.789308996088657,
-      "grad_norm": 1.4109781980514526,
-      "learning_rate": 7.008442211055277e-05,
-      "loss": 5.3318,
-      "step": 30276
-    },
-    {
-      "epoch": 15.789830508474576,
-      "grad_norm": 1.5437045097351074,
-      "learning_rate": 7.008341708542714e-05,
-      "loss": 5.061,
-      "step": 30277
-    },
-    {
-      "epoch": 15.790352020860496,
-      "grad_norm": 1.4617958068847656,
-      "learning_rate": 7.00824120603015e-05,
-      "loss": 5.4884,
-      "step": 30278
-    },
-    {
-      "epoch": 15.790873533246415,
-      "grad_norm": 1.4619839191436768,
-      "learning_rate": 7.008140703517588e-05,
-      "loss": 5.3513,
-      "step": 30279
-    },
-    {
-      "epoch": 15.791395045632333,
-      "grad_norm": 1.4338233470916748,
-      "learning_rate": 7.008040201005026e-05,
-      "loss": 5.2349,
-      "step": 30280
-    },
-    {
-      "epoch": 15.791916558018253,
-      "grad_norm": 1.573731780052185,
-      "learning_rate": 7.007939698492463e-05,
-      "loss": 4.7327,
-      "step": 30281
-    },
-    {
-      "epoch": 15.792438070404172,
-      "grad_norm": 1.5923352241516113,
-      "learning_rate": 7.0078391959799e-05,
-      "loss": 5.4223,
-      "step": 30282
-    },
-    {
-      "epoch": 15.792959582790091,
-      "grad_norm": 1.5130772590637207,
-      "learning_rate": 7.007738693467338e-05,
-      "loss": 4.702,
-      "step": 30283
-    },
-    {
-      "epoch": 15.793481095176011,
-      "grad_norm": 1.4688860177993774,
-      "learning_rate": 7.007638190954774e-05,
-      "loss": 5.3455,
-      "step": 30284
-    },
-    {
-      "epoch": 15.79400260756193,
-      "grad_norm": 1.4514305591583252,
-      "learning_rate": 7.007537688442212e-05,
-      "loss": 5.5233,
-      "step": 30285
-    },
-    {
-      "epoch": 15.794524119947848,
-      "grad_norm": 1.5359165668487549,
-      "learning_rate": 7.007437185929648e-05,
-      "loss": 5.3806,
-      "step": 30286
-    },
-    {
-      "epoch": 15.795045632333768,
-      "grad_norm": 1.5345733165740967,
-      "learning_rate": 7.007336683417086e-05,
-      "loss": 4.7766,
-      "step": 30287
-    },
-    {
-      "epoch": 15.795567144719687,
-      "grad_norm": 1.4962658882141113,
-      "learning_rate": 7.007236180904522e-05,
-      "loss": 5.6989,
-      "step": 30288
-    },
-    {
-      "epoch": 15.796088657105607,
-      "grad_norm": 1.4354151487350464,
-      "learning_rate": 7.00713567839196e-05,
-      "loss": 5.7055,
-      "step": 30289
-    },
-    {
-      "epoch": 15.796610169491526,
-      "grad_norm": 1.4488197565078735,
-      "learning_rate": 7.007035175879398e-05,
-      "loss": 5.5342,
-      "step": 30290
-    },
-    {
-      "epoch": 15.797131681877445,
-      "grad_norm": 1.547020673751831,
-      "learning_rate": 7.006934673366834e-05,
-      "loss": 5.0435,
-      "step": 30291
-    },
-    {
-      "epoch": 15.797653194263363,
-      "grad_norm": 1.4433211088180542,
-      "learning_rate": 7.006834170854272e-05,
-      "loss": 5.1811,
-      "step": 30292
-    },
-    {
-      "epoch": 15.798174706649283,
-      "grad_norm": 1.5391932725906372,
-      "learning_rate": 7.006733668341709e-05,
-      "loss": 5.3777,
-      "step": 30293
-    },
-    {
-      "epoch": 15.798696219035202,
-      "grad_norm": 1.5522042512893677,
-      "learning_rate": 7.006633165829146e-05,
-      "loss": 5.3493,
-      "step": 30294
-    },
-    {
-      "epoch": 15.799217731421122,
-      "grad_norm": 1.396227478981018,
-      "learning_rate": 7.006532663316583e-05,
-      "loss": 5.1243,
-      "step": 30295
-    },
-    {
-      "epoch": 15.799739243807041,
-      "grad_norm": 1.583461046218872,
-      "learning_rate": 7.00643216080402e-05,
-      "loss": 5.2669,
-      "step": 30296
-    },
-    {
-      "epoch": 15.800260756192959,
-      "grad_norm": 1.489331841468811,
-      "learning_rate": 7.006331658291457e-05,
-      "loss": 5.3373,
-      "step": 30297
-    },
-    {
-      "epoch": 15.800782268578878,
-      "grad_norm": 1.3951588869094849,
-      "learning_rate": 7.006231155778895e-05,
-      "loss": 5.582,
-      "step": 30298
-    },
-    {
-      "epoch": 15.801303780964798,
-      "grad_norm": 1.397723913192749,
-      "learning_rate": 7.006130653266331e-05,
-      "loss": 5.4156,
-      "step": 30299
-    },
-    {
-      "epoch": 15.801825293350717,
-      "grad_norm": 1.5721948146820068,
-      "learning_rate": 7.006030150753769e-05,
-      "loss": 5.3557,
-      "step": 30300
-    },
-    {
-      "epoch": 15.802346805736637,
-      "grad_norm": 1.460800290107727,
-      "learning_rate": 7.005929648241207e-05,
-      "loss": 5.4339,
-      "step": 30301
-    },
-    {
-      "epoch": 15.802868318122556,
-      "grad_norm": 1.3733633756637573,
-      "learning_rate": 7.005829145728645e-05,
-      "loss": 5.776,
-      "step": 30302
-    },
-    {
-      "epoch": 15.803389830508475,
-      "grad_norm": 1.366523265838623,
-      "learning_rate": 7.005728643216081e-05,
-      "loss": 5.6456,
-      "step": 30303
-    },
-    {
-      "epoch": 15.803911342894393,
-      "grad_norm": 1.5622129440307617,
-      "learning_rate": 7.005628140703519e-05,
-      "loss": 5.3028,
-      "step": 30304
-    },
-    {
-      "epoch": 15.804432855280313,
-      "grad_norm": 1.323290228843689,
-      "learning_rate": 7.005527638190955e-05,
-      "loss": 5.6053,
-      "step": 30305
-    },
-    {
-      "epoch": 15.804954367666232,
-      "grad_norm": 1.6503397226333618,
-      "learning_rate": 7.005427135678392e-05,
-      "loss": 4.9613,
-      "step": 30306
-    },
-    {
-      "epoch": 15.805475880052152,
-      "grad_norm": 1.5183258056640625,
-      "learning_rate": 7.00532663316583e-05,
-      "loss": 5.176,
-      "step": 30307
-    },
-    {
-      "epoch": 15.805997392438071,
-      "grad_norm": 1.4939725399017334,
-      "learning_rate": 7.005226130653266e-05,
-      "loss": 5.2501,
-      "step": 30308
-    },
-    {
-      "epoch": 15.806518904823989,
-      "grad_norm": 1.5180881023406982,
-      "learning_rate": 7.005125628140704e-05,
-      "loss": 5.6875,
-      "step": 30309
-    },
-    {
-      "epoch": 15.807040417209908,
-      "grad_norm": 1.5132060050964355,
-      "learning_rate": 7.00502512562814e-05,
-      "loss": 5.9217,
-      "step": 30310
-    },
-    {
-      "epoch": 15.807561929595828,
-      "grad_norm": 1.427817702293396,
-      "learning_rate": 7.004924623115578e-05,
-      "loss": 5.5494,
-      "step": 30311
-    },
-    {
-      "epoch": 15.808083441981747,
-      "grad_norm": 1.5039196014404297,
-      "learning_rate": 7.004824120603016e-05,
-      "loss": 5.0666,
-      "step": 30312
-    },
-    {
-      "epoch": 15.808604954367667,
-      "grad_norm": 1.46401846408844,
-      "learning_rate": 7.004723618090453e-05,
-      "loss": 4.9176,
-      "step": 30313
-    },
-    {
-      "epoch": 15.809126466753586,
-      "grad_norm": 1.4994993209838867,
-      "learning_rate": 7.00462311557789e-05,
-      "loss": 5.2735,
-      "step": 30314
-    },
-    {
-      "epoch": 15.809647979139505,
-      "grad_norm": 1.496349573135376,
-      "learning_rate": 7.004522613065328e-05,
-      "loss": 5.1483,
-      "step": 30315
-    },
-    {
-      "epoch": 15.810169491525423,
-      "grad_norm": 1.4826056957244873,
-      "learning_rate": 7.004422110552764e-05,
-      "loss": 5.2919,
-      "step": 30316
-    },
-    {
-      "epoch": 15.810691003911343,
-      "grad_norm": 1.3570637702941895,
-      "learning_rate": 7.004321608040202e-05,
-      "loss": 5.1636,
-      "step": 30317
-    },
-    {
-      "epoch": 15.811212516297262,
-      "grad_norm": 1.5748114585876465,
-      "learning_rate": 7.004221105527638e-05,
-      "loss": 5.3891,
-      "step": 30318
-    },
-    {
-      "epoch": 15.811734028683182,
-      "grad_norm": 1.402992606163025,
-      "learning_rate": 7.004120603015075e-05,
-      "loss": 5.6146,
-      "step": 30319
-    },
-    {
-      "epoch": 15.812255541069101,
-      "grad_norm": 1.4344443082809448,
-      "learning_rate": 7.004020100502512e-05,
-      "loss": 4.8849,
-      "step": 30320
-    },
-    {
-      "epoch": 15.812777053455019,
-      "grad_norm": 1.518977403640747,
-      "learning_rate": 7.00391959798995e-05,
-      "loss": 5.667,
-      "step": 30321
-    },
-    {
-      "epoch": 15.813298565840938,
-      "grad_norm": 1.4516323804855347,
-      "learning_rate": 7.003819095477388e-05,
-      "loss": 5.0646,
-      "step": 30322
-    },
-    {
-      "epoch": 15.813820078226858,
-      "grad_norm": 1.407728672027588,
-      "learning_rate": 7.003718592964824e-05,
-      "loss": 5.6019,
-      "step": 30323
-    },
-    {
-      "epoch": 15.814341590612777,
-      "grad_norm": 1.5073602199554443,
-      "learning_rate": 7.003618090452262e-05,
-      "loss": 5.2507,
-      "step": 30324
-    },
-    {
-      "epoch": 15.814863102998697,
-      "grad_norm": 1.479079246520996,
-      "learning_rate": 7.003517587939699e-05,
-      "loss": 5.2116,
-      "step": 30325
-    },
-    {
-      "epoch": 15.815384615384616,
-      "grad_norm": 1.486585259437561,
-      "learning_rate": 7.003417085427136e-05,
-      "loss": 5.5101,
-      "step": 30326
-    },
-    {
-      "epoch": 15.815906127770534,
-      "grad_norm": 1.4673086404800415,
-      "learning_rate": 7.003316582914573e-05,
-      "loss": 5.7109,
-      "step": 30327
-    },
-    {
-      "epoch": 15.816427640156453,
-      "grad_norm": 1.4464459419250488,
-      "learning_rate": 7.00321608040201e-05,
-      "loss": 5.2859,
-      "step": 30328
-    },
-    {
-      "epoch": 15.816949152542373,
-      "grad_norm": 1.7978161573410034,
-      "learning_rate": 7.003115577889447e-05,
-      "loss": 4.7755,
-      "step": 30329
-    },
-    {
-      "epoch": 15.817470664928292,
-      "grad_norm": 1.3725334405899048,
-      "learning_rate": 7.003015075376885e-05,
-      "loss": 5.2073,
-      "step": 30330
-    },
-    {
-      "epoch": 15.817992177314212,
-      "grad_norm": 1.5304675102233887,
-      "learning_rate": 7.002914572864321e-05,
-      "loss": 5.0979,
-      "step": 30331
-    },
-    {
-      "epoch": 15.818513689700131,
-      "grad_norm": 1.4424408674240112,
-      "learning_rate": 7.002814070351759e-05,
-      "loss": 5.1601,
-      "step": 30332
-    },
-    {
-      "epoch": 15.819035202086049,
-      "grad_norm": 1.3884862661361694,
-      "learning_rate": 7.002713567839197e-05,
-      "loss": 5.3976,
-      "step": 30333
-    },
-    {
-      "epoch": 15.819556714471968,
-      "grad_norm": 1.3795528411865234,
-      "learning_rate": 7.002613065326633e-05,
-      "loss": 5.8865,
-      "step": 30334
-    },
-    {
-      "epoch": 15.820078226857888,
-      "grad_norm": 1.3934326171875,
-      "learning_rate": 7.002512562814071e-05,
-      "loss": 5.7565,
-      "step": 30335
-    },
-    {
-      "epoch": 15.820599739243807,
-      "grad_norm": 1.6211481094360352,
-      "learning_rate": 7.002412060301508e-05,
-      "loss": 4.8788,
-      "step": 30336
-    },
-    {
-      "epoch": 15.821121251629727,
-      "grad_norm": 1.4858163595199585,
-      "learning_rate": 7.002311557788945e-05,
-      "loss": 5.5005,
-      "step": 30337
-    },
-    {
-      "epoch": 15.821642764015646,
-      "grad_norm": 1.4018003940582275,
-      "learning_rate": 7.002211055276382e-05,
-      "loss": 5.3265,
-      "step": 30338
-    },
-    {
-      "epoch": 15.822164276401564,
-      "grad_norm": 1.449812412261963,
-      "learning_rate": 7.00211055276382e-05,
-      "loss": 5.5353,
-      "step": 30339
-    },
-    {
-      "epoch": 15.822685788787483,
-      "grad_norm": 1.4659650325775146,
-      "learning_rate": 7.002010050251256e-05,
-      "loss": 5.0264,
-      "step": 30340
-    },
-    {
-      "epoch": 15.823207301173403,
-      "grad_norm": 1.4892125129699707,
-      "learning_rate": 7.001909547738694e-05,
-      "loss": 5.1945,
-      "step": 30341
-    },
-    {
-      "epoch": 15.823728813559322,
-      "grad_norm": 1.3576672077178955,
-      "learning_rate": 7.001809045226131e-05,
-      "loss": 5.8044,
-      "step": 30342
-    },
-    {
-      "epoch": 15.824250325945242,
-      "grad_norm": 1.4073001146316528,
-      "learning_rate": 7.001708542713569e-05,
-      "loss": 5.4917,
-      "step": 30343
-    },
-    {
-      "epoch": 15.824771838331161,
-      "grad_norm": 1.5428497791290283,
-      "learning_rate": 7.001608040201006e-05,
-      "loss": 4.9806,
-      "step": 30344
-    },
-    {
-      "epoch": 15.825293350717079,
-      "grad_norm": 1.4903943538665771,
-      "learning_rate": 7.001507537688442e-05,
-      "loss": 5.4686,
-      "step": 30345
-    },
-    {
-      "epoch": 15.825814863102998,
-      "grad_norm": 1.5884997844696045,
-      "learning_rate": 7.00140703517588e-05,
-      "loss": 5.3543,
-      "step": 30346
-    },
-    {
-      "epoch": 15.826336375488918,
-      "grad_norm": 1.5219172239303589,
-      "learning_rate": 7.001306532663316e-05,
-      "loss": 5.4876,
-      "step": 30347
-    },
-    {
-      "epoch": 15.826857887874837,
-      "grad_norm": 1.500691533088684,
-      "learning_rate": 7.001206030150754e-05,
-      "loss": 5.8044,
-      "step": 30348
-    },
-    {
-      "epoch": 15.827379400260757,
-      "grad_norm": 1.4560953378677368,
-      "learning_rate": 7.00110552763819e-05,
-      "loss": 5.0561,
-      "step": 30349
-    },
-    {
-      "epoch": 15.827900912646676,
-      "grad_norm": 1.508933663368225,
-      "learning_rate": 7.001005025125628e-05,
-      "loss": 5.5507,
-      "step": 30350
-    },
-    {
-      "epoch": 15.828422425032594,
-      "grad_norm": 1.6466952562332153,
-      "learning_rate": 7.000904522613065e-05,
-      "loss": 5.3459,
-      "step": 30351
-    },
-    {
-      "epoch": 15.828943937418513,
-      "grad_norm": 1.5006712675094604,
-      "learning_rate": 7.000804020100503e-05,
-      "loss": 5.1996,
-      "step": 30352
-    },
-    {
-      "epoch": 15.829465449804433,
-      "grad_norm": 1.4051467180252075,
-      "learning_rate": 7.00070351758794e-05,
-      "loss": 5.5875,
-      "step": 30353
-    },
-    {
-      "epoch": 15.829986962190352,
-      "grad_norm": 1.4293581247329712,
-      "learning_rate": 7.000603015075378e-05,
-      "loss": 4.8604,
-      "step": 30354
-    },
-    {
-      "epoch": 15.830508474576272,
-      "grad_norm": 1.599920630455017,
-      "learning_rate": 7.000502512562815e-05,
-      "loss": 4.8441,
-      "step": 30355
-    },
-    {
-      "epoch": 15.831029986962191,
-      "grad_norm": 1.6312376260757446,
-      "learning_rate": 7.000402010050252e-05,
-      "loss": 5.4411,
-      "step": 30356
-    },
-    {
-      "epoch": 15.831551499348109,
-      "grad_norm": 1.4122059345245361,
-      "learning_rate": 7.000301507537689e-05,
-      "loss": 4.9716,
-      "step": 30357
-    },
-    {
-      "epoch": 15.832073011734028,
-      "grad_norm": 1.5736958980560303,
-      "learning_rate": 7.000201005025125e-05,
-      "loss": 5.1735,
-      "step": 30358
-    },
-    {
-      "epoch": 15.832594524119948,
-      "grad_norm": 1.4192613363265991,
-      "learning_rate": 7.000100502512563e-05,
-      "loss": 5.5552,
-      "step": 30359
-    },
-    {
-      "epoch": 15.833116036505867,
-      "grad_norm": 1.4193211793899536,
-      "learning_rate": 7e-05,
-      "loss": 5.3639,
-      "step": 30360
-    },
-    {
-      "epoch": 15.833637548891787,
-      "grad_norm": 1.498274564743042,
-      "learning_rate": 6.999899497487437e-05,
-      "loss": 5.3534,
-      "step": 30361
-    },
-    {
-      "epoch": 15.834159061277706,
-      "grad_norm": 1.4515576362609863,
-      "learning_rate": 6.999798994974875e-05,
-      "loss": 4.7174,
-      "step": 30362
-    },
-    {
-      "epoch": 15.834680573663624,
-      "grad_norm": 1.5957297086715698,
-      "learning_rate": 6.999698492462313e-05,
-      "loss": 5.1258,
-      "step": 30363
-    },
-    {
-      "epoch": 15.835202086049543,
-      "grad_norm": 1.5344537496566772,
-      "learning_rate": 6.999597989949749e-05,
-      "loss": 5.6413,
-      "step": 30364
-    },
-    {
-      "epoch": 15.835723598435463,
-      "grad_norm": 1.4449163675308228,
-      "learning_rate": 6.999497487437187e-05,
-      "loss": 5.3065,
-      "step": 30365
-    },
-    {
-      "epoch": 15.836245110821382,
-      "grad_norm": 1.6206272840499878,
-      "learning_rate": 6.999396984924623e-05,
-      "loss": 5.2545,
-      "step": 30366
-    },
-    {
-      "epoch": 15.836766623207302,
-      "grad_norm": 1.4722568988800049,
-      "learning_rate": 6.999296482412061e-05,
-      "loss": 4.6085,
-      "step": 30367
-    },
-    {
-      "epoch": 15.837288135593221,
-      "grad_norm": 1.4736316204071045,
-      "learning_rate": 6.999195979899498e-05,
-      "loss": 5.0595,
-      "step": 30368
-    },
-    {
-      "epoch": 15.837809647979139,
-      "grad_norm": 1.5801132917404175,
-      "learning_rate": 6.999095477386935e-05,
-      "loss": 5.4711,
-      "step": 30369
-    },
-    {
-      "epoch": 15.838331160365058,
-      "grad_norm": 1.3795987367630005,
-      "learning_rate": 6.998994974874372e-05,
-      "loss": 5.7658,
-      "step": 30370
-    },
-    {
-      "epoch": 15.838852672750978,
-      "grad_norm": 1.4492318630218506,
-      "learning_rate": 6.998894472361808e-05,
-      "loss": 5.4105,
-      "step": 30371
-    },
-    {
-      "epoch": 15.839374185136897,
-      "grad_norm": 1.7159430980682373,
-      "learning_rate": 6.998793969849246e-05,
-      "loss": 5.3871,
-      "step": 30372
-    },
-    {
-      "epoch": 15.839895697522817,
-      "grad_norm": 1.4795138835906982,
-      "learning_rate": 6.998693467336684e-05,
-      "loss": 5.3272,
-      "step": 30373
-    },
-    {
-      "epoch": 15.840417209908736,
-      "grad_norm": 1.4877936840057373,
-      "learning_rate": 6.998592964824122e-05,
-      "loss": 5.245,
-      "step": 30374
-    },
-    {
-      "epoch": 15.840938722294654,
-      "grad_norm": 1.506435751914978,
-      "learning_rate": 6.998492462311558e-05,
-      "loss": 5.4601,
-      "step": 30375
-    },
-    {
-      "epoch": 15.841460234680573,
-      "grad_norm": 1.3766324520111084,
-      "learning_rate": 6.998391959798996e-05,
-      "loss": 5.6365,
-      "step": 30376
-    },
-    {
-      "epoch": 15.841981747066493,
-      "grad_norm": 1.4744007587432861,
-      "learning_rate": 6.998291457286432e-05,
-      "loss": 5.6981,
-      "step": 30377
-    },
-    {
-      "epoch": 15.842503259452412,
-      "grad_norm": 1.608835220336914,
-      "learning_rate": 6.99819095477387e-05,
-      "loss": 4.8463,
-      "step": 30378
-    },
-    {
-      "epoch": 15.843024771838332,
-      "grad_norm": 1.6056876182556152,
-      "learning_rate": 6.998090452261306e-05,
-      "loss": 5.3291,
-      "step": 30379
-    },
-    {
-      "epoch": 15.843546284224251,
-      "grad_norm": 1.473455786705017,
-      "learning_rate": 6.997989949748744e-05,
-      "loss": 5.5035,
-      "step": 30380
-    },
-    {
-      "epoch": 15.844067796610169,
-      "grad_norm": 1.5780917406082153,
-      "learning_rate": 6.99788944723618e-05,
-      "loss": 5.3531,
-      "step": 30381
-    },
-    {
-      "epoch": 15.844589308996088,
-      "grad_norm": 1.4314134120941162,
-      "learning_rate": 6.997788944723618e-05,
-      "loss": 5.2002,
-      "step": 30382
-    },
-    {
-      "epoch": 15.845110821382008,
-      "grad_norm": 1.4993404150009155,
-      "learning_rate": 6.997688442211056e-05,
-      "loss": 5.0333,
-      "step": 30383
-    },
-    {
-      "epoch": 15.845632333767927,
-      "grad_norm": 1.4315739870071411,
-      "learning_rate": 6.997587939698493e-05,
-      "loss": 5.6636,
-      "step": 30384
-    },
-    {
-      "epoch": 15.846153846153847,
-      "grad_norm": 1.52027428150177,
-      "learning_rate": 6.99748743718593e-05,
-      "loss": 5.5697,
-      "step": 30385
-    },
-    {
-      "epoch": 15.846675358539766,
-      "grad_norm": 1.5204343795776367,
-      "learning_rate": 6.997386934673367e-05,
-      "loss": 5.7566,
-      "step": 30386
-    },
-    {
-      "epoch": 15.847196870925684,
-      "grad_norm": 1.5659565925598145,
-      "learning_rate": 6.997286432160805e-05,
-      "loss": 5.4243,
-      "step": 30387
-    },
-    {
-      "epoch": 15.847718383311603,
-      "grad_norm": 1.5209474563598633,
-      "learning_rate": 6.997185929648241e-05,
-      "loss": 5.0206,
-      "step": 30388
-    },
-    {
-      "epoch": 15.848239895697523,
-      "grad_norm": 1.5513736009597778,
-      "learning_rate": 6.997085427135679e-05,
-      "loss": 5.1445,
-      "step": 30389
-    },
-    {
-      "epoch": 15.848761408083442,
-      "grad_norm": 1.3855254650115967,
-      "learning_rate": 6.996984924623115e-05,
-      "loss": 5.7442,
-      "step": 30390
-    },
-    {
-      "epoch": 15.849282920469362,
-      "grad_norm": 1.5939538478851318,
-      "learning_rate": 6.996884422110553e-05,
-      "loss": 5.2,
-      "step": 30391
-    },
-    {
-      "epoch": 15.84980443285528,
-      "grad_norm": 1.4568800926208496,
-      "learning_rate": 6.99678391959799e-05,
-      "loss": 5.523,
-      "step": 30392
-    },
-    {
-      "epoch": 15.850325945241199,
-      "grad_norm": 1.4494556188583374,
-      "learning_rate": 6.996683417085427e-05,
-      "loss": 5.0846,
-      "step": 30393
-    },
-    {
-      "epoch": 15.850847457627118,
-      "grad_norm": 1.4347728490829468,
-      "learning_rate": 6.996582914572865e-05,
-      "loss": 5.635,
-      "step": 30394
-    },
-    {
-      "epoch": 15.851368970013038,
-      "grad_norm": 1.5396414995193481,
-      "learning_rate": 6.996482412060303e-05,
-      "loss": 5.2459,
-      "step": 30395
-    },
-    {
-      "epoch": 15.851890482398957,
-      "grad_norm": 1.4824477434158325,
-      "learning_rate": 6.996381909547739e-05,
-      "loss": 5.4001,
-      "step": 30396
-    },
-    {
-      "epoch": 15.852411994784877,
-      "grad_norm": 1.4852873086929321,
-      "learning_rate": 6.996281407035177e-05,
-      "loss": 5.5832,
-      "step": 30397
-    },
-    {
-      "epoch": 15.852933507170796,
-      "grad_norm": 1.5934408903121948,
-      "learning_rate": 6.996180904522613e-05,
-      "loss": 5.563,
-      "step": 30398
-    },
-    {
-      "epoch": 15.853455019556714,
-      "grad_norm": 1.5287351608276367,
-      "learning_rate": 6.99608040201005e-05,
-      "loss": 5.241,
-      "step": 30399
-    },
-    {
-      "epoch": 15.853976531942633,
-      "grad_norm": 1.5906566381454468,
-      "learning_rate": 6.995979899497488e-05,
-      "loss": 5.1592,
-      "step": 30400
-    },
-    {
-      "epoch": 15.854498044328553,
-      "grad_norm": 1.4209178686141968,
-      "learning_rate": 6.995879396984924e-05,
-      "loss": 5.5639,
-      "step": 30401
-    },
-    {
-      "epoch": 15.855019556714472,
-      "grad_norm": 1.4845441579818726,
-      "learning_rate": 6.995778894472362e-05,
-      "loss": 5.3534,
-      "step": 30402
-    },
-    {
-      "epoch": 15.855541069100392,
-      "grad_norm": 1.4019620418548584,
-      "learning_rate": 6.9956783919598e-05,
-      "loss": 5.6128,
-      "step": 30403
-    },
-    {
-      "epoch": 15.85606258148631,
-      "grad_norm": 1.466338872909546,
-      "learning_rate": 6.995577889447237e-05,
-      "loss": 5.295,
-      "step": 30404
-    },
-    {
-      "epoch": 15.856584093872229,
-      "grad_norm": 1.467340111732483,
-      "learning_rate": 6.995477386934674e-05,
-      "loss": 5.6049,
-      "step": 30405
-    },
-    {
-      "epoch": 15.857105606258148,
-      "grad_norm": 1.6030455827713013,
-      "learning_rate": 6.995376884422112e-05,
-      "loss": 5.1364,
-      "step": 30406
-    },
-    {
-      "epoch": 15.857627118644068,
-      "grad_norm": 1.5240498781204224,
-      "learning_rate": 6.995276381909548e-05,
-      "loss": 5.3148,
-      "step": 30407
-    },
-    {
-      "epoch": 15.858148631029987,
-      "grad_norm": 1.4664510488510132,
-      "learning_rate": 6.995175879396986e-05,
-      "loss": 5.3868,
-      "step": 30408
-    },
-    {
-      "epoch": 15.858670143415907,
-      "grad_norm": 1.476261854171753,
-      "learning_rate": 6.995075376884422e-05,
-      "loss": 5.4668,
-      "step": 30409
-    },
-    {
-      "epoch": 15.859191655801826,
-      "grad_norm": 1.4599480628967285,
-      "learning_rate": 6.99497487437186e-05,
-      "loss": 5.4515,
-      "step": 30410
-    },
-    {
-      "epoch": 15.859713168187744,
-      "grad_norm": 1.4538235664367676,
-      "learning_rate": 6.994874371859296e-05,
-      "loss": 5.4526,
-      "step": 30411
-    },
-    {
-      "epoch": 15.860234680573663,
-      "grad_norm": 1.4472113847732544,
-      "learning_rate": 6.994773869346733e-05,
-      "loss": 5.5173,
-      "step": 30412
-    },
-    {
-      "epoch": 15.860756192959583,
-      "grad_norm": 1.4640308618545532,
-      "learning_rate": 6.994673366834171e-05,
-      "loss": 5.3521,
-      "step": 30413
-    },
-    {
-      "epoch": 15.861277705345502,
-      "grad_norm": 1.4631510972976685,
-      "learning_rate": 6.994572864321608e-05,
-      "loss": 5.2826,
-      "step": 30414
-    },
-    {
-      "epoch": 15.861799217731422,
-      "grad_norm": 1.4297021627426147,
-      "learning_rate": 6.994472361809046e-05,
-      "loss": 5.5989,
-      "step": 30415
-    },
-    {
-      "epoch": 15.86232073011734,
-      "grad_norm": 1.5185526609420776,
-      "learning_rate": 6.994371859296483e-05,
-      "loss": 5.1739,
-      "step": 30416
-    },
-    {
-      "epoch": 15.862842242503259,
-      "grad_norm": 1.5432533025741577,
-      "learning_rate": 6.99427135678392e-05,
-      "loss": 5.2642,
-      "step": 30417
-    },
-    {
-      "epoch": 15.863363754889178,
-      "grad_norm": 1.6125239133834839,
-      "learning_rate": 6.994170854271357e-05,
-      "loss": 4.7749,
-      "step": 30418
-    },
-    {
-      "epoch": 15.863885267275098,
-      "grad_norm": 1.445418357849121,
-      "learning_rate": 6.994070351758795e-05,
-      "loss": 5.4322,
-      "step": 30419
-    },
-    {
-      "epoch": 15.864406779661017,
-      "grad_norm": 1.4589712619781494,
-      "learning_rate": 6.993969849246231e-05,
-      "loss": 5.4965,
-      "step": 30420
-    },
-    {
-      "epoch": 15.864928292046937,
-      "grad_norm": 1.4509190320968628,
-      "learning_rate": 6.993869346733669e-05,
-      "loss": 4.8855,
-      "step": 30421
-    },
-    {
-      "epoch": 15.865449804432854,
-      "grad_norm": 1.4240574836730957,
-      "learning_rate": 6.993768844221105e-05,
-      "loss": 5.5068,
-      "step": 30422
-    },
-    {
-      "epoch": 15.865971316818774,
-      "grad_norm": 1.485298752784729,
-      "learning_rate": 6.993668341708543e-05,
-      "loss": 5.4887,
-      "step": 30423
-    },
-    {
-      "epoch": 15.866492829204693,
-      "grad_norm": 1.5533255338668823,
-      "learning_rate": 6.993567839195981e-05,
-      "loss": 5.4599,
-      "step": 30424
-    },
-    {
-      "epoch": 15.867014341590613,
-      "grad_norm": 1.5429524183273315,
-      "learning_rate": 6.993467336683417e-05,
-      "loss": 5.1763,
-      "step": 30425
-    },
-    {
-      "epoch": 15.867535853976532,
-      "grad_norm": 1.4788702726364136,
-      "learning_rate": 6.993366834170855e-05,
-      "loss": 5.1141,
-      "step": 30426
-    },
-    {
-      "epoch": 15.868057366362452,
-      "grad_norm": 1.5168298482894897,
-      "learning_rate": 6.993266331658292e-05,
-      "loss": 5.4283,
-      "step": 30427
-    },
-    {
-      "epoch": 15.86857887874837,
-      "grad_norm": 1.4471083879470825,
-      "learning_rate": 6.993165829145729e-05,
-      "loss": 5.691,
-      "step": 30428
-    },
-    {
-      "epoch": 15.869100391134289,
-      "grad_norm": 1.7111717462539673,
-      "learning_rate": 6.993065326633166e-05,
-      "loss": 4.7194,
-      "step": 30429
-    },
-    {
-      "epoch": 15.869621903520208,
-      "grad_norm": 1.5433863401412964,
-      "learning_rate": 6.992964824120604e-05,
-      "loss": 5.3917,
-      "step": 30430
-    },
-    {
-      "epoch": 15.870143415906128,
-      "grad_norm": 1.45786452293396,
-      "learning_rate": 6.99286432160804e-05,
-      "loss": 5.7097,
-      "step": 30431
-    },
-    {
-      "epoch": 15.870664928292047,
-      "grad_norm": 1.5138249397277832,
-      "learning_rate": 6.992763819095478e-05,
-      "loss": 5.5106,
-      "step": 30432
-    },
-    {
-      "epoch": 15.871186440677967,
-      "grad_norm": 1.423102855682373,
-      "learning_rate": 6.992663316582914e-05,
-      "loss": 5.0892,
-      "step": 30433
-    },
-    {
-      "epoch": 15.871707953063884,
-      "grad_norm": 1.4221900701522827,
-      "learning_rate": 6.992562814070352e-05,
-      "loss": 5.7819,
-      "step": 30434
-    },
-    {
-      "epoch": 15.872229465449804,
-      "grad_norm": 1.4909576177597046,
-      "learning_rate": 6.99246231155779e-05,
-      "loss": 5.397,
-      "step": 30435
-    },
-    {
-      "epoch": 15.872750977835723,
-      "grad_norm": 1.4447462558746338,
-      "learning_rate": 6.992361809045228e-05,
-      "loss": 5.4251,
-      "step": 30436
-    },
-    {
-      "epoch": 15.873272490221643,
-      "grad_norm": 1.5446988344192505,
-      "learning_rate": 6.992261306532664e-05,
-      "loss": 5.6919,
-      "step": 30437
-    },
-    {
-      "epoch": 15.873794002607562,
-      "grad_norm": 1.47120201587677,
-      "learning_rate": 6.9921608040201e-05,
-      "loss": 5.2147,
-      "step": 30438
-    },
-    {
-      "epoch": 15.874315514993482,
-      "grad_norm": 1.4714611768722534,
-      "learning_rate": 6.992060301507538e-05,
-      "loss": 4.6976,
-      "step": 30439
-    },
-    {
-      "epoch": 15.8748370273794,
-      "grad_norm": 1.6353895664215088,
-      "learning_rate": 6.991959798994975e-05,
-      "loss": 4.7489,
-      "step": 30440
-    },
-    {
-      "epoch": 15.875358539765319,
-      "grad_norm": 1.4875680208206177,
-      "learning_rate": 6.991859296482412e-05,
-      "loss": 5.6537,
-      "step": 30441
-    },
-    {
-      "epoch": 15.875880052151238,
-      "grad_norm": 1.4406567811965942,
-      "learning_rate": 6.991758793969849e-05,
-      "loss": 5.1543,
-      "step": 30442
-    },
-    {
-      "epoch": 15.876401564537158,
-      "grad_norm": 1.4801162481307983,
-      "learning_rate": 6.991658291457287e-05,
-      "loss": 5.1982,
-      "step": 30443
-    },
-    {
-      "epoch": 15.876923076923077,
-      "grad_norm": 1.4139851331710815,
-      "learning_rate": 6.991557788944724e-05,
-      "loss": 5.3198,
-      "step": 30444
-    },
-    {
-      "epoch": 15.877444589308997,
-      "grad_norm": 1.3892194032669067,
-      "learning_rate": 6.991457286432162e-05,
-      "loss": 5.4081,
-      "step": 30445
-    },
-    {
-      "epoch": 15.877966101694915,
-      "grad_norm": 1.4366719722747803,
-      "learning_rate": 6.991356783919599e-05,
-      "loss": 5.481,
-      "step": 30446
-    },
-    {
-      "epoch": 15.878487614080834,
-      "grad_norm": 1.4273765087127686,
-      "learning_rate": 6.991256281407036e-05,
-      "loss": 5.7242,
-      "step": 30447
-    },
-    {
-      "epoch": 15.879009126466753,
-      "grad_norm": 1.5065226554870605,
-      "learning_rate": 6.991155778894473e-05,
-      "loss": 5.3761,
-      "step": 30448
-    },
-    {
-      "epoch": 15.879530638852673,
-      "grad_norm": 1.5708869695663452,
-      "learning_rate": 6.99105527638191e-05,
-      "loss": 5.2311,
-      "step": 30449
-    },
-    {
-      "epoch": 15.880052151238592,
-      "grad_norm": 1.6013001203536987,
-      "learning_rate": 6.990954773869347e-05,
-      "loss": 5.2526,
-      "step": 30450
-    },
-    {
-      "epoch": 15.880573663624512,
-      "grad_norm": 1.5219001770019531,
-      "learning_rate": 6.990854271356783e-05,
-      "loss": 5.2146,
-      "step": 30451
-    },
-    {
-      "epoch": 15.88109517601043,
-      "grad_norm": 1.4283350706100464,
-      "learning_rate": 6.990753768844221e-05,
-      "loss": 5.5954,
-      "step": 30452
-    },
-    {
-      "epoch": 15.881616688396349,
-      "grad_norm": 1.5214449167251587,
-      "learning_rate": 6.990653266331658e-05,
-      "loss": 5.7109,
-      "step": 30453
-    },
-    {
-      "epoch": 15.882138200782268,
-      "grad_norm": 1.5626554489135742,
-      "learning_rate": 6.990552763819095e-05,
-      "loss": 5.5268,
-      "step": 30454
-    },
-    {
-      "epoch": 15.882659713168188,
-      "grad_norm": 1.4695441722869873,
-      "learning_rate": 6.990452261306533e-05,
-      "loss": 5.4198,
-      "step": 30455
-    },
-    {
-      "epoch": 15.883181225554107,
-      "grad_norm": 1.4451723098754883,
-      "learning_rate": 6.990351758793971e-05,
-      "loss": 4.7547,
-      "step": 30456
-    },
-    {
-      "epoch": 15.883702737940027,
-      "grad_norm": 1.4845136404037476,
-      "learning_rate": 6.990251256281407e-05,
-      "loss": 5.424,
-      "step": 30457
-    },
-    {
-      "epoch": 15.884224250325945,
-      "grad_norm": 1.5134435892105103,
-      "learning_rate": 6.990150753768845e-05,
-      "loss": 5.6631,
-      "step": 30458
-    },
-    {
-      "epoch": 15.884745762711864,
-      "grad_norm": 1.4327776432037354,
-      "learning_rate": 6.990050251256282e-05,
-      "loss": 5.7436,
-      "step": 30459
-    },
-    {
-      "epoch": 15.885267275097783,
-      "grad_norm": 1.536957859992981,
-      "learning_rate": 6.98994974874372e-05,
-      "loss": 4.7989,
-      "step": 30460
-    },
-    {
-      "epoch": 15.885788787483703,
-      "grad_norm": 1.50222647190094,
-      "learning_rate": 6.989849246231156e-05,
-      "loss": 5.4608,
-      "step": 30461
-    },
-    {
-      "epoch": 15.886310299869622,
-      "grad_norm": 1.5650010108947754,
-      "learning_rate": 6.989748743718594e-05,
-      "loss": 4.9168,
-      "step": 30462
-    },
-    {
-      "epoch": 15.886831812255542,
-      "grad_norm": 1.5856962203979492,
-      "learning_rate": 6.98964824120603e-05,
-      "loss": 5.3551,
-      "step": 30463
-    },
-    {
-      "epoch": 15.88735332464146,
-      "grad_norm": 1.4486721754074097,
-      "learning_rate": 6.989547738693466e-05,
-      "loss": 5.5541,
-      "step": 30464
-    },
-    {
-      "epoch": 15.887874837027379,
-      "grad_norm": 1.488252878189087,
-      "learning_rate": 6.989447236180904e-05,
-      "loss": 5.313,
-      "step": 30465
-    },
-    {
-      "epoch": 15.888396349413298,
-      "grad_norm": 1.3538436889648438,
-      "learning_rate": 6.989346733668342e-05,
-      "loss": 5.8893,
-      "step": 30466
-    },
-    {
-      "epoch": 15.888917861799218,
-      "grad_norm": 1.537735104560852,
-      "learning_rate": 6.98924623115578e-05,
-      "loss": 4.864,
-      "step": 30467
-    },
-    {
-      "epoch": 15.889439374185137,
-      "grad_norm": 1.4194749593734741,
-      "learning_rate": 6.989145728643216e-05,
-      "loss": 5.1102,
-      "step": 30468
-    },
-    {
-      "epoch": 15.889960886571057,
-      "grad_norm": 1.4197466373443604,
-      "learning_rate": 6.989045226130654e-05,
-      "loss": 5.3251,
-      "step": 30469
-    },
-    {
-      "epoch": 15.890482398956975,
-      "grad_norm": 1.4738084077835083,
-      "learning_rate": 6.98894472361809e-05,
-      "loss": 4.8791,
-      "step": 30470
-    },
-    {
-      "epoch": 15.891003911342894,
-      "grad_norm": 1.510414958000183,
-      "learning_rate": 6.988844221105528e-05,
-      "loss": 4.7777,
-      "step": 30471
-    },
-    {
-      "epoch": 15.891525423728813,
-      "grad_norm": 1.5517263412475586,
-      "learning_rate": 6.988743718592965e-05,
-      "loss": 5.3337,
-      "step": 30472
-    },
-    {
-      "epoch": 15.892046936114733,
-      "grad_norm": 1.4085667133331299,
-      "learning_rate": 6.988643216080402e-05,
-      "loss": 5.4821,
-      "step": 30473
-    },
-    {
-      "epoch": 15.892568448500652,
-      "grad_norm": 1.7838771343231201,
-      "learning_rate": 6.988542713567839e-05,
-      "loss": 5.3606,
-      "step": 30474
-    },
-    {
-      "epoch": 15.893089960886572,
-      "grad_norm": 1.3910187482833862,
-      "learning_rate": 6.988442211055277e-05,
-      "loss": 5.5465,
-      "step": 30475
-    },
-    {
-      "epoch": 15.89361147327249,
-      "grad_norm": 1.500244379043579,
-      "learning_rate": 6.988341708542714e-05,
-      "loss": 5.2499,
-      "step": 30476
-    },
-    {
-      "epoch": 15.894132985658409,
-      "grad_norm": 1.4267665147781372,
-      "learning_rate": 6.988241206030151e-05,
-      "loss": 5.4144,
-      "step": 30477
-    },
-    {
-      "epoch": 15.894654498044329,
-      "grad_norm": 1.6558616161346436,
-      "learning_rate": 6.988140703517589e-05,
-      "loss": 4.9437,
-      "step": 30478
-    },
-    {
-      "epoch": 15.895176010430248,
-      "grad_norm": 1.5679224729537964,
-      "learning_rate": 6.988040201005025e-05,
-      "loss": 4.9519,
-      "step": 30479
-    },
-    {
-      "epoch": 15.895697522816167,
-      "grad_norm": 1.5100799798965454,
-      "learning_rate": 6.987939698492463e-05,
-      "loss": 5.5371,
-      "step": 30480
-    },
-    {
-      "epoch": 15.896219035202087,
-      "grad_norm": 1.4489526748657227,
-      "learning_rate": 6.987839195979899e-05,
-      "loss": 5.7444,
-      "step": 30481
-    },
-    {
-      "epoch": 15.896740547588005,
-      "grad_norm": 1.3624541759490967,
-      "learning_rate": 6.987738693467337e-05,
-      "loss": 5.8007,
-      "step": 30482
-    },
-    {
-      "epoch": 15.897262059973924,
-      "grad_norm": 1.487384557723999,
-      "learning_rate": 6.987638190954773e-05,
-      "loss": 5.7486,
-      "step": 30483
-    },
-    {
-      "epoch": 15.897783572359844,
-      "grad_norm": 1.4637142419815063,
-      "learning_rate": 6.987537688442211e-05,
-      "loss": 4.6277,
-      "step": 30484
-    },
-    {
-      "epoch": 15.898305084745763,
-      "grad_norm": 1.3824745416641235,
-      "learning_rate": 6.987437185929648e-05,
-      "loss": 5.833,
-      "step": 30485
-    },
-    {
-      "epoch": 15.898826597131682,
-      "grad_norm": 1.4130640029907227,
-      "learning_rate": 6.987336683417085e-05,
-      "loss": 5.6217,
-      "step": 30486
-    },
-    {
-      "epoch": 15.8993481095176,
-      "grad_norm": 1.3449180126190186,
-      "learning_rate": 6.987236180904523e-05,
-      "loss": 5.7016,
-      "step": 30487
-    },
-    {
-      "epoch": 15.89986962190352,
-      "grad_norm": 1.4384335279464722,
-      "learning_rate": 6.987135678391961e-05,
-      "loss": 5.0689,
-      "step": 30488
-    },
-    {
-      "epoch": 15.900391134289439,
-      "grad_norm": 1.4603253602981567,
-      "learning_rate": 6.987035175879397e-05,
-      "loss": 5.2696,
-      "step": 30489
-    },
-    {
-      "epoch": 15.900912646675359,
-      "grad_norm": 1.6167163848876953,
-      "learning_rate": 6.986934673366835e-05,
-      "loss": 5.0485,
-      "step": 30490
-    },
-    {
-      "epoch": 15.901434159061278,
-      "grad_norm": 1.4862834215164185,
-      "learning_rate": 6.986834170854272e-05,
-      "loss": 5.2703,
-      "step": 30491
-    },
-    {
-      "epoch": 15.901955671447197,
-      "grad_norm": 1.5175169706344604,
-      "learning_rate": 6.986733668341708e-05,
-      "loss": 5.8584,
-      "step": 30492
-    },
-    {
-      "epoch": 15.902477183833117,
-      "grad_norm": 1.5673614740371704,
-      "learning_rate": 6.986633165829146e-05,
-      "loss": 5.4398,
-      "step": 30493
-    },
-    {
-      "epoch": 15.902998696219035,
-      "grad_norm": 1.4730908870697021,
-      "learning_rate": 6.986532663316582e-05,
-      "loss": 5.3463,
-      "step": 30494
-    },
-    {
-      "epoch": 15.903520208604954,
-      "grad_norm": 1.5304926633834839,
-      "learning_rate": 6.98643216080402e-05,
-      "loss": 5.5915,
-      "step": 30495
-    },
-    {
-      "epoch": 15.904041720990874,
-      "grad_norm": 1.5485382080078125,
-      "learning_rate": 6.986331658291458e-05,
-      "loss": 5.2899,
-      "step": 30496
-    },
-    {
-      "epoch": 15.904563233376793,
-      "grad_norm": 1.4278963804244995,
-      "learning_rate": 6.986231155778896e-05,
-      "loss": 4.8495,
-      "step": 30497
-    },
-    {
-      "epoch": 15.905084745762712,
-      "grad_norm": 1.5443533658981323,
-      "learning_rate": 6.986130653266332e-05,
-      "loss": 5.1062,
-      "step": 30498
-    },
-    {
-      "epoch": 15.90560625814863,
-      "grad_norm": 1.6409554481506348,
-      "learning_rate": 6.98603015075377e-05,
-      "loss": 4.2532,
-      "step": 30499
-    },
-    {
-      "epoch": 15.90612777053455,
-      "grad_norm": 1.407578468322754,
-      "learning_rate": 6.985929648241206e-05,
-      "loss": 5.0452,
-      "step": 30500
-    },
-    {
-      "epoch": 15.906649282920469,
-      "grad_norm": 1.4793956279754639,
-      "learning_rate": 6.985829145728644e-05,
-      "loss": 5.4016,
-      "step": 30501
-    },
-    {
-      "epoch": 15.907170795306389,
-      "grad_norm": 1.45779287815094,
-      "learning_rate": 6.98572864321608e-05,
-      "loss": 5.1559,
-      "step": 30502
-    },
-    {
-      "epoch": 15.907692307692308,
-      "grad_norm": 1.5360219478607178,
-      "learning_rate": 6.985628140703518e-05,
-      "loss": 5.3559,
-      "step": 30503
-    },
-    {
-      "epoch": 15.908213820078227,
-      "grad_norm": 1.6594562530517578,
-      "learning_rate": 6.985527638190955e-05,
-      "loss": 5.4329,
-      "step": 30504
-    },
-    {
-      "epoch": 15.908735332464147,
-      "grad_norm": 1.4395743608474731,
-      "learning_rate": 6.985427135678391e-05,
-      "loss": 5.8561,
-      "step": 30505
-    },
-    {
-      "epoch": 15.909256844850065,
-      "grad_norm": 1.4825384616851807,
-      "learning_rate": 6.985326633165829e-05,
-      "loss": 5.1941,
-      "step": 30506
-    },
-    {
-      "epoch": 15.909778357235984,
-      "grad_norm": 1.5066946744918823,
-      "learning_rate": 6.985226130653267e-05,
-      "loss": 5.4394,
-      "step": 30507
-    },
-    {
-      "epoch": 15.910299869621904,
-      "grad_norm": 1.4870461225509644,
-      "learning_rate": 6.985125628140705e-05,
-      "loss": 5.2497,
-      "step": 30508
-    },
-    {
-      "epoch": 15.910821382007823,
-      "grad_norm": 1.4796022176742554,
-      "learning_rate": 6.985025125628141e-05,
-      "loss": 5.4006,
-      "step": 30509
-    },
-    {
-      "epoch": 15.911342894393742,
-      "grad_norm": 1.5225374698638916,
-      "learning_rate": 6.984924623115579e-05,
-      "loss": 5.0695,
-      "step": 30510
-    },
-    {
-      "epoch": 15.91186440677966,
-      "grad_norm": 1.3843533992767334,
-      "learning_rate": 6.984824120603015e-05,
-      "loss": 5.5079,
-      "step": 30511
-    },
-    {
-      "epoch": 15.91238591916558,
-      "grad_norm": 1.554186224937439,
-      "learning_rate": 6.984723618090453e-05,
-      "loss": 4.9157,
-      "step": 30512
-    },
-    {
-      "epoch": 15.9129074315515,
-      "grad_norm": 1.5130693912506104,
-      "learning_rate": 6.98462311557789e-05,
-      "loss": 5.2974,
-      "step": 30513
-    },
-    {
-      "epoch": 15.913428943937419,
-      "grad_norm": 1.472098708152771,
-      "learning_rate": 6.984522613065327e-05,
-      "loss": 5.427,
-      "step": 30514
-    },
-    {
-      "epoch": 15.913950456323338,
-      "grad_norm": 1.4889572858810425,
-      "learning_rate": 6.984422110552764e-05,
-      "loss": 5.2998,
-      "step": 30515
-    },
-    {
-      "epoch": 15.914471968709258,
-      "grad_norm": 1.5021134614944458,
-      "learning_rate": 6.984321608040201e-05,
-      "loss": 4.8687,
-      "step": 30516
-    },
-    {
-      "epoch": 15.914993481095175,
-      "grad_norm": 1.5493137836456299,
-      "learning_rate": 6.984221105527639e-05,
-      "loss": 5.4167,
-      "step": 30517
-    },
-    {
-      "epoch": 15.915514993481095,
-      "grad_norm": 1.552289605140686,
-      "learning_rate": 6.984120603015076e-05,
-      "loss": 5.3652,
-      "step": 30518
-    },
-    {
-      "epoch": 15.916036505867014,
-      "grad_norm": 1.51752507686615,
-      "learning_rate": 6.984020100502513e-05,
-      "loss": 5.2375,
-      "step": 30519
-    },
-    {
-      "epoch": 15.916558018252934,
-      "grad_norm": 1.5014289617538452,
-      "learning_rate": 6.98391959798995e-05,
-      "loss": 5.5662,
-      "step": 30520
-    },
-    {
-      "epoch": 15.917079530638853,
-      "grad_norm": 1.506789207458496,
-      "learning_rate": 6.983819095477388e-05,
-      "loss": 5.2389,
-      "step": 30521
-    },
-    {
-      "epoch": 15.917601043024773,
-      "grad_norm": 1.5766680240631104,
-      "learning_rate": 6.983718592964824e-05,
-      "loss": 5.5072,
-      "step": 30522
-    },
-    {
-      "epoch": 15.91812255541069,
-      "grad_norm": 1.5283777713775635,
-      "learning_rate": 6.983618090452262e-05,
-      "loss": 5.5723,
-      "step": 30523
-    },
-    {
-      "epoch": 15.91864406779661,
-      "grad_norm": 1.4786508083343506,
-      "learning_rate": 6.983517587939698e-05,
-      "loss": 5.1099,
-      "step": 30524
-    },
-    {
-      "epoch": 15.91916558018253,
-      "grad_norm": 1.5761826038360596,
-      "learning_rate": 6.983417085427136e-05,
-      "loss": 5.2308,
-      "step": 30525
-    },
-    {
-      "epoch": 15.919687092568449,
-      "grad_norm": 1.4257081747055054,
-      "learning_rate": 6.983316582914572e-05,
-      "loss": 5.5923,
-      "step": 30526
-    },
-    {
-      "epoch": 15.920208604954368,
-      "grad_norm": 1.4256097078323364,
-      "learning_rate": 6.98321608040201e-05,
-      "loss": 5.6298,
-      "step": 30527
-    },
-    {
-      "epoch": 15.920730117340288,
-      "grad_norm": 1.4084041118621826,
-      "learning_rate": 6.983115577889448e-05,
-      "loss": 5.6928,
-      "step": 30528
-    },
-    {
-      "epoch": 15.921251629726205,
-      "grad_norm": 1.4722398519515991,
-      "learning_rate": 6.983015075376886e-05,
-      "loss": 5.465,
-      "step": 30529
-    },
-    {
-      "epoch": 15.921773142112125,
-      "grad_norm": 1.4442715644836426,
-      "learning_rate": 6.982914572864322e-05,
-      "loss": 5.3873,
-      "step": 30530
-    },
-    {
-      "epoch": 15.922294654498044,
-      "grad_norm": 1.6161645650863647,
-      "learning_rate": 6.982814070351759e-05,
-      "loss": 4.9658,
-      "step": 30531
-    },
-    {
-      "epoch": 15.922816166883964,
-      "grad_norm": 1.4163345098495483,
-      "learning_rate": 6.982713567839196e-05,
-      "loss": 5.7547,
-      "step": 30532
-    },
-    {
-      "epoch": 15.923337679269883,
-      "grad_norm": 1.5296803712844849,
-      "learning_rate": 6.982613065326633e-05,
-      "loss": 5.4842,
-      "step": 30533
-    },
-    {
-      "epoch": 15.923859191655803,
-      "grad_norm": 1.4596439599990845,
-      "learning_rate": 6.98251256281407e-05,
-      "loss": 5.6049,
-      "step": 30534
-    },
-    {
-      "epoch": 15.92438070404172,
-      "grad_norm": 1.4150499105453491,
-      "learning_rate": 6.982412060301507e-05,
-      "loss": 5.5159,
-      "step": 30535
-    },
-    {
-      "epoch": 15.92490221642764,
-      "grad_norm": 1.405438780784607,
-      "learning_rate": 6.982311557788945e-05,
-      "loss": 5.7522,
-      "step": 30536
-    },
-    {
-      "epoch": 15.92542372881356,
-      "grad_norm": 1.4298430681228638,
-      "learning_rate": 6.982211055276383e-05,
-      "loss": 5.439,
-      "step": 30537
-    },
-    {
-      "epoch": 15.925945241199479,
-      "grad_norm": 1.618770956993103,
-      "learning_rate": 6.98211055276382e-05,
-      "loss": 4.7703,
-      "step": 30538
-    },
-    {
-      "epoch": 15.926466753585398,
-      "grad_norm": 1.4642211198806763,
-      "learning_rate": 6.982010050251257e-05,
-      "loss": 5.2547,
-      "step": 30539
-    },
-    {
-      "epoch": 15.926988265971318,
-      "grad_norm": 1.4510276317596436,
-      "learning_rate": 6.981909547738695e-05,
-      "loss": 5.4716,
-      "step": 30540
-    },
-    {
-      "epoch": 15.927509778357235,
-      "grad_norm": 1.3594450950622559,
-      "learning_rate": 6.981809045226131e-05,
-      "loss": 5.6783,
-      "step": 30541
-    },
-    {
-      "epoch": 15.928031290743155,
-      "grad_norm": 1.5075780153274536,
-      "learning_rate": 6.981708542713569e-05,
-      "loss": 5.2341,
-      "step": 30542
-    },
-    {
-      "epoch": 15.928552803129074,
-      "grad_norm": 1.5501152276992798,
-      "learning_rate": 6.981608040201005e-05,
-      "loss": 5.5084,
-      "step": 30543
-    },
-    {
-      "epoch": 15.929074315514994,
-      "grad_norm": 1.492543339729309,
-      "learning_rate": 6.981507537688442e-05,
-      "loss": 5.1334,
-      "step": 30544
-    },
-    {
-      "epoch": 15.929595827900913,
-      "grad_norm": 1.5194264650344849,
-      "learning_rate": 6.98140703517588e-05,
-      "loss": 4.8223,
-      "step": 30545
-    },
-    {
-      "epoch": 15.930117340286833,
-      "grad_norm": 1.5439493656158447,
-      "learning_rate": 6.981306532663316e-05,
-      "loss": 5.0145,
-      "step": 30546
-    },
-    {
-      "epoch": 15.93063885267275,
-      "grad_norm": 1.459333896636963,
-      "learning_rate": 6.981206030150754e-05,
-      "loss": 5.3605,
-      "step": 30547
-    },
-    {
-      "epoch": 15.93116036505867,
-      "grad_norm": 1.473097324371338,
-      "learning_rate": 6.981105527638191e-05,
-      "loss": 5.4467,
-      "step": 30548
-    },
-    {
-      "epoch": 15.93168187744459,
-      "grad_norm": 1.5625369548797607,
-      "learning_rate": 6.981005025125629e-05,
-      "loss": 5.2001,
-      "step": 30549
-    },
-    {
-      "epoch": 15.932203389830509,
-      "grad_norm": 1.3918023109436035,
-      "learning_rate": 6.980904522613066e-05,
-      "loss": 5.6362,
-      "step": 30550
-    },
-    {
-      "epoch": 15.932724902216428,
-      "grad_norm": 1.4947805404663086,
-      "learning_rate": 6.980804020100503e-05,
-      "loss": 4.9681,
-      "step": 30551
-    },
-    {
-      "epoch": 15.933246414602348,
-      "grad_norm": 1.4852923154830933,
-      "learning_rate": 6.98070351758794e-05,
-      "loss": 5.1923,
-      "step": 30552
-    },
-    {
-      "epoch": 15.933767926988265,
-      "grad_norm": 1.5104726552963257,
-      "learning_rate": 6.980603015075378e-05,
-      "loss": 5.5183,
-      "step": 30553
-    },
-    {
-      "epoch": 15.934289439374185,
-      "grad_norm": 1.4657177925109863,
-      "learning_rate": 6.980502512562814e-05,
-      "loss": 5.6832,
-      "step": 30554
-    },
-    {
-      "epoch": 15.934810951760104,
-      "grad_norm": 1.5006704330444336,
-      "learning_rate": 6.980402010050252e-05,
-      "loss": 5.4792,
-      "step": 30555
-    },
-    {
-      "epoch": 15.935332464146024,
-      "grad_norm": 1.343180775642395,
-      "learning_rate": 6.980301507537688e-05,
-      "loss": 5.66,
-      "step": 30556
-    },
-    {
-      "epoch": 15.935853976531943,
-      "grad_norm": 1.483411431312561,
-      "learning_rate": 6.980201005025126e-05,
-      "loss": 5.3568,
-      "step": 30557
-    },
-    {
-      "epoch": 15.936375488917863,
-      "grad_norm": 1.5658529996871948,
-      "learning_rate": 6.980100502512564e-05,
-      "loss": 4.793,
-      "step": 30558
-    },
-    {
-      "epoch": 15.93689700130378,
-      "grad_norm": 1.6157095432281494,
-      "learning_rate": 6.98e-05,
-      "loss": 5.1744,
-      "step": 30559
-    },
-    {
-      "epoch": 15.9374185136897,
-      "grad_norm": 1.421636700630188,
-      "learning_rate": 6.979899497487438e-05,
-      "loss": 5.5081,
-      "step": 30560
-    },
-    {
-      "epoch": 15.93794002607562,
-      "grad_norm": 1.4810876846313477,
-      "learning_rate": 6.979798994974874e-05,
-      "loss": 5.4907,
-      "step": 30561
-    },
-    {
-      "epoch": 15.938461538461539,
-      "grad_norm": 1.4467607736587524,
-      "learning_rate": 6.979698492462312e-05,
-      "loss": 5.2786,
-      "step": 30562
-    },
-    {
-      "epoch": 15.938983050847458,
-      "grad_norm": 1.5469354391098022,
-      "learning_rate": 6.979597989949749e-05,
-      "loss": 5.5457,
-      "step": 30563
-    },
-    {
-      "epoch": 15.939504563233378,
-      "grad_norm": 1.6746810674667358,
-      "learning_rate": 6.979497487437186e-05,
-      "loss": 5.2981,
-      "step": 30564
-    },
-    {
-      "epoch": 15.940026075619295,
-      "grad_norm": 1.4682585000991821,
-      "learning_rate": 6.979396984924623e-05,
-      "loss": 5.2259,
-      "step": 30565
-    },
-    {
-      "epoch": 15.940547588005215,
-      "grad_norm": 1.5570967197418213,
-      "learning_rate": 6.97929648241206e-05,
-      "loss": 5.1437,
-      "step": 30566
-    },
-    {
-      "epoch": 15.941069100391134,
-      "grad_norm": 1.7059297561645508,
-      "learning_rate": 6.979195979899497e-05,
-      "loss": 5.1725,
-      "step": 30567
-    },
-    {
-      "epoch": 15.941590612777054,
-      "grad_norm": 1.4981975555419922,
-      "learning_rate": 6.979095477386935e-05,
-      "loss": 5.6455,
-      "step": 30568
-    },
-    {
-      "epoch": 15.942112125162973,
-      "grad_norm": 1.5595357418060303,
-      "learning_rate": 6.978994974874373e-05,
-      "loss": 5.3786,
-      "step": 30569
-    },
-    {
-      "epoch": 15.94263363754889,
-      "grad_norm": 1.6488229036331177,
-      "learning_rate": 6.97889447236181e-05,
-      "loss": 5.5236,
-      "step": 30570
-    },
-    {
-      "epoch": 15.94315514993481,
-      "grad_norm": 1.6151862144470215,
-      "learning_rate": 6.978793969849247e-05,
-      "loss": 5.1955,
-      "step": 30571
-    },
-    {
-      "epoch": 15.94367666232073,
-      "grad_norm": 1.4873361587524414,
-      "learning_rate": 6.978693467336683e-05,
-      "loss": 5.311,
-      "step": 30572
-    },
-    {
-      "epoch": 15.94419817470665,
-      "grad_norm": 1.5375458002090454,
-      "learning_rate": 6.978592964824121e-05,
-      "loss": 5.1063,
-      "step": 30573
-    },
-    {
-      "epoch": 15.944719687092569,
-      "grad_norm": 1.500336766242981,
-      "learning_rate": 6.978492462311558e-05,
-      "loss": 5.6715,
-      "step": 30574
-    },
-    {
-      "epoch": 15.945241199478488,
-      "grad_norm": 1.517096996307373,
-      "learning_rate": 6.978391959798995e-05,
-      "loss": 4.8429,
-      "step": 30575
-    },
-    {
-      "epoch": 15.945762711864408,
-      "grad_norm": 1.4847476482391357,
-      "learning_rate": 6.978291457286432e-05,
-      "loss": 5.0792,
-      "step": 30576
-    },
-    {
-      "epoch": 15.946284224250325,
-      "grad_norm": 1.334309458732605,
-      "learning_rate": 6.97819095477387e-05,
-      "loss": 5.4444,
-      "step": 30577
-    },
-    {
-      "epoch": 15.946805736636245,
-      "grad_norm": 1.3794628381729126,
-      "learning_rate": 6.978090452261307e-05,
-      "loss": 5.1903,
-      "step": 30578
-    },
-    {
-      "epoch": 15.947327249022164,
-      "grad_norm": 1.4924976825714111,
-      "learning_rate": 6.977989949748745e-05,
-      "loss": 5.2437,
-      "step": 30579
-    },
-    {
-      "epoch": 15.947848761408084,
-      "grad_norm": 1.3449240922927856,
-      "learning_rate": 6.977889447236181e-05,
-      "loss": 5.5655,
-      "step": 30580
-    },
-    {
-      "epoch": 15.948370273794003,
-      "grad_norm": 1.4085640907287598,
-      "learning_rate": 6.977788944723619e-05,
-      "loss": 5.8722,
-      "step": 30581
-    },
-    {
-      "epoch": 15.94889178617992,
-      "grad_norm": 1.4542884826660156,
-      "learning_rate": 6.977688442211056e-05,
-      "loss": 4.9206,
-      "step": 30582
-    },
-    {
-      "epoch": 15.94941329856584,
-      "grad_norm": 1.5679001808166504,
-      "learning_rate": 6.977587939698493e-05,
-      "loss": 5.366,
-      "step": 30583
-    },
-    {
-      "epoch": 15.94993481095176,
-      "grad_norm": 1.4923242330551147,
-      "learning_rate": 6.97748743718593e-05,
-      "loss": 5.1141,
-      "step": 30584
-    },
-    {
-      "epoch": 15.95045632333768,
-      "grad_norm": 1.4613264799118042,
-      "learning_rate": 6.977386934673366e-05,
-      "loss": 5.0599,
-      "step": 30585
-    },
-    {
-      "epoch": 15.950977835723599,
-      "grad_norm": 1.4012250900268555,
-      "learning_rate": 6.977286432160804e-05,
-      "loss": 4.8523,
-      "step": 30586
-    },
-    {
-      "epoch": 15.951499348109518,
-      "grad_norm": 1.436112880706787,
-      "learning_rate": 6.97718592964824e-05,
-      "loss": 5.3189,
-      "step": 30587
-    },
-    {
-      "epoch": 15.952020860495438,
-      "grad_norm": 1.4335441589355469,
-      "learning_rate": 6.977085427135678e-05,
-      "loss": 5.7938,
-      "step": 30588
-    },
-    {
-      "epoch": 15.952542372881355,
-      "grad_norm": 1.5905483961105347,
-      "learning_rate": 6.976984924623116e-05,
-      "loss": 5.2278,
-      "step": 30589
-    },
-    {
-      "epoch": 15.953063885267275,
-      "grad_norm": 1.5670089721679688,
-      "learning_rate": 6.976884422110554e-05,
-      "loss": 5.2917,
-      "step": 30590
-    },
-    {
-      "epoch": 15.953585397653194,
-      "grad_norm": 1.4105052947998047,
-      "learning_rate": 6.97678391959799e-05,
-      "loss": 5.4735,
-      "step": 30591
-    },
-    {
-      "epoch": 15.954106910039114,
-      "grad_norm": 1.4165291786193848,
-      "learning_rate": 6.976683417085428e-05,
-      "loss": 5.76,
-      "step": 30592
-    },
-    {
-      "epoch": 15.954628422425033,
-      "grad_norm": 1.4285995960235596,
-      "learning_rate": 6.976582914572865e-05,
-      "loss": 5.6249,
-      "step": 30593
-    },
-    {
-      "epoch": 15.955149934810951,
-      "grad_norm": 1.5160346031188965,
-      "learning_rate": 6.976482412060302e-05,
-      "loss": 5.5453,
-      "step": 30594
-    },
-    {
-      "epoch": 15.95567144719687,
-      "grad_norm": 1.464688777923584,
-      "learning_rate": 6.976381909547739e-05,
-      "loss": 5.4471,
-      "step": 30595
-    },
-    {
-      "epoch": 15.95619295958279,
-      "grad_norm": 1.4814977645874023,
-      "learning_rate": 6.976281407035177e-05,
-      "loss": 5.2563,
-      "step": 30596
-    },
-    {
-      "epoch": 15.95671447196871,
-      "grad_norm": 1.6632636785507202,
-      "learning_rate": 6.976180904522613e-05,
-      "loss": 5.1775,
-      "step": 30597
-    },
-    {
-      "epoch": 15.957235984354629,
-      "grad_norm": 1.4240761995315552,
-      "learning_rate": 6.976080402010051e-05,
-      "loss": 5.2918,
-      "step": 30598
-    },
-    {
-      "epoch": 15.957757496740548,
-      "grad_norm": 1.5652005672454834,
-      "learning_rate": 6.975979899497489e-05,
-      "loss": 5.1628,
-      "step": 30599
-    },
-    {
-      "epoch": 15.958279009126468,
-      "grad_norm": 1.4978325366973877,
-      "learning_rate": 6.975879396984925e-05,
-      "loss": 5.5343,
-      "step": 30600
-    },
-    {
-      "epoch": 15.958800521512385,
-      "grad_norm": 1.4447420835494995,
-      "learning_rate": 6.975778894472363e-05,
-      "loss": 5.5724,
-      "step": 30601
-    },
-    {
-      "epoch": 15.959322033898305,
-      "grad_norm": 1.3589071035385132,
-      "learning_rate": 6.975678391959799e-05,
-      "loss": 5.7952,
-      "step": 30602
-    },
-    {
-      "epoch": 15.959843546284224,
-      "grad_norm": 1.54403817653656,
-      "learning_rate": 6.975577889447237e-05,
-      "loss": 4.9543,
-      "step": 30603
-    },
-    {
-      "epoch": 15.960365058670144,
-      "grad_norm": 1.4937928915023804,
-      "learning_rate": 6.975477386934673e-05,
-      "loss": 5.4136,
-      "step": 30604
-    },
-    {
-      "epoch": 15.960886571056063,
-      "grad_norm": 1.552374005317688,
-      "learning_rate": 6.975376884422111e-05,
-      "loss": 5.4055,
-      "step": 30605
-    },
-    {
-      "epoch": 15.961408083441981,
-      "grad_norm": 1.551119089126587,
-      "learning_rate": 6.975276381909548e-05,
-      "loss": 5.3265,
-      "step": 30606
-    },
-    {
-      "epoch": 15.9619295958279,
-      "grad_norm": 1.3994163274765015,
-      "learning_rate": 6.975175879396985e-05,
-      "loss": 5.5571,
-      "step": 30607
-    },
-    {
-      "epoch": 15.96245110821382,
-      "grad_norm": 1.338018536567688,
-      "learning_rate": 6.975075376884422e-05,
-      "loss": 5.8717,
-      "step": 30608
-    },
-    {
-      "epoch": 15.96297262059974,
-      "grad_norm": 1.410319447517395,
-      "learning_rate": 6.97497487437186e-05,
-      "loss": 5.6026,
-      "step": 30609
-    },
-    {
-      "epoch": 15.963494132985659,
-      "grad_norm": 1.4660770893096924,
-      "learning_rate": 6.974874371859297e-05,
-      "loss": 5.2006,
-      "step": 30610
-    },
-    {
-      "epoch": 15.964015645371578,
-      "grad_norm": 1.577864646911621,
-      "learning_rate": 6.974773869346734e-05,
-      "loss": 5.1585,
-      "step": 30611
-    },
-    {
-      "epoch": 15.964537157757496,
-      "grad_norm": 1.5334296226501465,
-      "learning_rate": 6.974673366834172e-05,
-      "loss": 5.6559,
-      "step": 30612
-    },
-    {
-      "epoch": 15.965058670143415,
-      "grad_norm": 1.5045878887176514,
-      "learning_rate": 6.974572864321608e-05,
-      "loss": 5.7478,
-      "step": 30613
-    },
-    {
-      "epoch": 15.965580182529335,
-      "grad_norm": 1.3998953104019165,
-      "learning_rate": 6.974472361809046e-05,
-      "loss": 5.4755,
-      "step": 30614
-    },
-    {
-      "epoch": 15.966101694915254,
-      "grad_norm": 1.3904056549072266,
-      "learning_rate": 6.974371859296482e-05,
-      "loss": 5.5681,
-      "step": 30615
-    },
-    {
-      "epoch": 15.966623207301174,
-      "grad_norm": 1.5973035097122192,
-      "learning_rate": 6.97427135678392e-05,
-      "loss": 5.4518,
-      "step": 30616
-    },
-    {
-      "epoch": 15.967144719687093,
-      "grad_norm": 1.4743354320526123,
-      "learning_rate": 6.974170854271356e-05,
-      "loss": 5.6529,
-      "step": 30617
-    },
-    {
-      "epoch": 15.967666232073011,
-      "grad_norm": 1.4561185836791992,
-      "learning_rate": 6.974070351758794e-05,
-      "loss": 5.1494,
-      "step": 30618
-    },
-    {
-      "epoch": 15.96818774445893,
-      "grad_norm": 1.3941494226455688,
-      "learning_rate": 6.973969849246232e-05,
-      "loss": 5.4776,
-      "step": 30619
-    },
-    {
-      "epoch": 15.96870925684485,
-      "grad_norm": 1.3494110107421875,
-      "learning_rate": 6.97386934673367e-05,
-      "loss": 5.5545,
-      "step": 30620
-    },
-    {
-      "epoch": 15.96923076923077,
-      "grad_norm": 1.4165136814117432,
-      "learning_rate": 6.973768844221106e-05,
-      "loss": 5.3575,
-      "step": 30621
-    },
-    {
-      "epoch": 15.969752281616689,
-      "grad_norm": 1.4111732244491577,
-      "learning_rate": 6.973668341708544e-05,
-      "loss": 5.2062,
-      "step": 30622
-    },
-    {
-      "epoch": 15.970273794002608,
-      "grad_norm": 1.4120473861694336,
-      "learning_rate": 6.97356783919598e-05,
-      "loss": 5.8121,
-      "step": 30623
-    },
-    {
-      "epoch": 15.970795306388526,
-      "grad_norm": 1.4634065628051758,
-      "learning_rate": 6.973467336683417e-05,
-      "loss": 5.5427,
-      "step": 30624
-    },
-    {
-      "epoch": 15.971316818774445,
-      "grad_norm": 1.5014352798461914,
-      "learning_rate": 6.973366834170855e-05,
-      "loss": 5.0224,
-      "step": 30625
-    },
-    {
-      "epoch": 15.971838331160365,
-      "grad_norm": 1.4198403358459473,
-      "learning_rate": 6.973266331658291e-05,
-      "loss": 5.453,
-      "step": 30626
-    },
-    {
-      "epoch": 15.972359843546284,
-      "grad_norm": 1.4322142601013184,
-      "learning_rate": 6.973165829145729e-05,
-      "loss": 5.3581,
-      "step": 30627
-    },
-    {
-      "epoch": 15.972881355932204,
-      "grad_norm": 1.65389883518219,
-      "learning_rate": 6.973065326633165e-05,
-      "loss": 4.8223,
-      "step": 30628
-    },
-    {
-      "epoch": 15.973402868318123,
-      "grad_norm": 1.4208296537399292,
-      "learning_rate": 6.972964824120603e-05,
-      "loss": 5.3786,
-      "step": 30629
-    },
-    {
-      "epoch": 15.973924380704041,
-      "grad_norm": 1.503032922744751,
-      "learning_rate": 6.972864321608041e-05,
-      "loss": 5.4808,
-      "step": 30630
-    },
-    {
-      "epoch": 15.97444589308996,
-      "grad_norm": 1.4899228811264038,
-      "learning_rate": 6.972763819095479e-05,
-      "loss": 5.3266,
-      "step": 30631
-    },
-    {
-      "epoch": 15.97496740547588,
-      "grad_norm": 1.5279884338378906,
-      "learning_rate": 6.972663316582915e-05,
-      "loss": 5.0554,
-      "step": 30632
-    },
-    {
-      "epoch": 15.9754889178618,
-      "grad_norm": 1.4246957302093506,
-      "learning_rate": 6.972562814070353e-05,
-      "loss": 5.6907,
-      "step": 30633
-    },
-    {
-      "epoch": 15.976010430247719,
-      "grad_norm": 1.5697423219680786,
-      "learning_rate": 6.972462311557789e-05,
-      "loss": 5.1341,
-      "step": 30634
-    },
-    {
-      "epoch": 15.976531942633638,
-      "grad_norm": 1.4729206562042236,
-      "learning_rate": 6.972361809045227e-05,
-      "loss": 5.3915,
-      "step": 30635
-    },
-    {
-      "epoch": 15.977053455019556,
-      "grad_norm": 1.4184612035751343,
-      "learning_rate": 6.972261306532663e-05,
-      "loss": 5.6344,
-      "step": 30636
-    },
-    {
-      "epoch": 15.977574967405475,
-      "grad_norm": 1.5247238874435425,
-      "learning_rate": 6.9721608040201e-05,
-      "loss": 5.3438,
-      "step": 30637
-    },
-    {
-      "epoch": 15.978096479791395,
-      "grad_norm": 1.5522558689117432,
-      "learning_rate": 6.972060301507538e-05,
-      "loss": 5.0635,
-      "step": 30638
-    },
-    {
-      "epoch": 15.978617992177314,
-      "grad_norm": 1.473021388053894,
-      "learning_rate": 6.971959798994974e-05,
-      "loss": 5.6355,
-      "step": 30639
-    },
-    {
-      "epoch": 15.979139504563234,
-      "grad_norm": 1.4672517776489258,
-      "learning_rate": 6.971859296482412e-05,
-      "loss": 5.5308,
-      "step": 30640
-    },
-    {
-      "epoch": 15.979661016949153,
-      "grad_norm": 1.4990756511688232,
-      "learning_rate": 6.97175879396985e-05,
-      "loss": 5.5324,
-      "step": 30641
-    },
-    {
-      "epoch": 15.980182529335071,
-      "grad_norm": 1.4561917781829834,
-      "learning_rate": 6.971658291457287e-05,
-      "loss": 5.1707,
-      "step": 30642
-    },
-    {
-      "epoch": 15.98070404172099,
-      "grad_norm": 1.4909234046936035,
-      "learning_rate": 6.971557788944724e-05,
-      "loss": 5.3533,
-      "step": 30643
-    },
-    {
-      "epoch": 15.98122555410691,
-      "grad_norm": 1.6253206729888916,
-      "learning_rate": 6.971457286432162e-05,
-      "loss": 5.2287,
-      "step": 30644
-    },
-    {
-      "epoch": 15.98174706649283,
-      "grad_norm": 1.5763137340545654,
-      "learning_rate": 6.971356783919598e-05,
-      "loss": 4.8747,
-      "step": 30645
-    },
-    {
-      "epoch": 15.982268578878749,
-      "grad_norm": 1.439733624458313,
-      "learning_rate": 6.971256281407036e-05,
-      "loss": 5.489,
-      "step": 30646
-    },
-    {
-      "epoch": 15.982790091264668,
-      "grad_norm": 1.405836820602417,
-      "learning_rate": 6.971155778894472e-05,
-      "loss": 5.5814,
-      "step": 30647
-    },
-    {
-      "epoch": 15.983311603650586,
-      "grad_norm": 1.4562348127365112,
-      "learning_rate": 6.97105527638191e-05,
-      "loss": 5.2316,
-      "step": 30648
-    },
-    {
-      "epoch": 15.983833116036505,
-      "grad_norm": 1.5395832061767578,
-      "learning_rate": 6.970954773869346e-05,
-      "loss": 5.5234,
-      "step": 30649
-    },
-    {
-      "epoch": 15.984354628422425,
-      "grad_norm": 1.5050523281097412,
-      "learning_rate": 6.970854271356784e-05,
-      "loss": 5.4651,
-      "step": 30650
-    },
-    {
-      "epoch": 15.984876140808344,
-      "grad_norm": 1.458001732826233,
-      "learning_rate": 6.970753768844222e-05,
-      "loss": 5.7403,
-      "step": 30651
-    },
-    {
-      "epoch": 15.985397653194264,
-      "grad_norm": 1.4809937477111816,
-      "learning_rate": 6.970653266331658e-05,
-      "loss": 5.4148,
-      "step": 30652
-    },
-    {
-      "epoch": 15.985919165580183,
-      "grad_norm": 1.439950704574585,
-      "learning_rate": 6.970552763819096e-05,
-      "loss": 5.7742,
-      "step": 30653
-    },
-    {
-      "epoch": 15.986440677966101,
-      "grad_norm": 1.495163083076477,
-      "learning_rate": 6.970452261306533e-05,
-      "loss": 5.4237,
-      "step": 30654
-    },
-    {
-      "epoch": 15.98696219035202,
-      "grad_norm": 1.4540765285491943,
-      "learning_rate": 6.97035175879397e-05,
-      "loss": 5.4744,
-      "step": 30655
-    },
-    {
-      "epoch": 15.98748370273794,
-      "grad_norm": 1.5739954710006714,
-      "learning_rate": 6.970251256281407e-05,
-      "loss": 5.2633,
-      "step": 30656
-    },
-    {
-      "epoch": 15.98800521512386,
-      "grad_norm": 1.36847722530365,
-      "learning_rate": 6.970150753768845e-05,
-      "loss": 5.2676,
-      "step": 30657
-    },
-    {
-      "epoch": 15.988526727509779,
-      "grad_norm": 1.4379595518112183,
-      "learning_rate": 6.970050251256281e-05,
-      "loss": 5.4446,
-      "step": 30658
-    },
-    {
-      "epoch": 15.989048239895698,
-      "grad_norm": 1.5577547550201416,
-      "learning_rate": 6.969949748743719e-05,
-      "loss": 5.3093,
-      "step": 30659
-    },
-    {
-      "epoch": 15.989569752281616,
-      "grad_norm": 1.4520848989486694,
-      "learning_rate": 6.969849246231155e-05,
-      "loss": 5.5338,
-      "step": 30660
-    },
-    {
-      "epoch": 15.990091264667535,
-      "grad_norm": 1.5237507820129395,
-      "learning_rate": 6.969748743718593e-05,
-      "loss": 5.1066,
-      "step": 30661
-    },
-    {
-      "epoch": 15.990612777053455,
-      "grad_norm": 1.3882689476013184,
-      "learning_rate": 6.969648241206031e-05,
-      "loss": 5.5834,
-      "step": 30662
-    },
-    {
-      "epoch": 15.991134289439374,
-      "grad_norm": 1.4231932163238525,
-      "learning_rate": 6.969547738693469e-05,
-      "loss": 5.3751,
-      "step": 30663
-    },
-    {
-      "epoch": 15.991655801825294,
-      "grad_norm": 1.5476806163787842,
-      "learning_rate": 6.969447236180905e-05,
-      "loss": 4.8789,
-      "step": 30664
-    },
-    {
-      "epoch": 15.992177314211212,
-      "grad_norm": 1.3835195302963257,
-      "learning_rate": 6.969346733668342e-05,
-      "loss": 5.6326,
-      "step": 30665
-    },
-    {
-      "epoch": 15.992698826597131,
-      "grad_norm": 1.4711713790893555,
-      "learning_rate": 6.969246231155779e-05,
-      "loss": 5.2101,
-      "step": 30666
-    },
-    {
-      "epoch": 15.99322033898305,
-      "grad_norm": 1.4689857959747314,
-      "learning_rate": 6.969145728643216e-05,
-      "loss": 5.4065,
-      "step": 30667
-    },
-    {
-      "epoch": 15.99374185136897,
-      "grad_norm": 1.4382920265197754,
-      "learning_rate": 6.969045226130654e-05,
-      "loss": 5.516,
-      "step": 30668
-    },
-    {
-      "epoch": 15.99426336375489,
-      "grad_norm": 1.571627140045166,
-      "learning_rate": 6.96894472361809e-05,
-      "loss": 5.024,
-      "step": 30669
-    },
-    {
-      "epoch": 15.994784876140809,
-      "grad_norm": 1.5832043886184692,
-      "learning_rate": 6.968844221105528e-05,
-      "loss": 4.9485,
-      "step": 30670
-    },
-    {
-      "epoch": 15.995306388526728,
-      "grad_norm": 1.4603594541549683,
-      "learning_rate": 6.968743718592966e-05,
-      "loss": 5.4435,
-      "step": 30671
-    },
-    {
-      "epoch": 15.995827900912646,
-      "grad_norm": 1.5879589319229126,
-      "learning_rate": 6.968643216080403e-05,
-      "loss": 4.913,
-      "step": 30672
-    },
-    {
-      "epoch": 15.996349413298566,
-      "grad_norm": 1.282865047454834,
-      "learning_rate": 6.96854271356784e-05,
-      "loss": 5.8756,
-      "step": 30673
-    },
-    {
-      "epoch": 15.996870925684485,
-      "grad_norm": 1.5468186140060425,
-      "learning_rate": 6.968442211055278e-05,
-      "loss": 5.5638,
-      "step": 30674
-    },
-    {
-      "epoch": 15.997392438070404,
-      "grad_norm": 1.4741789102554321,
-      "learning_rate": 6.968341708542714e-05,
-      "loss": 5.8262,
-      "step": 30675
-    },
-    {
-      "epoch": 15.997913950456324,
-      "grad_norm": 1.3789113759994507,
-      "learning_rate": 6.968241206030152e-05,
-      "loss": 5.7342,
-      "step": 30676
-    },
-    {
-      "epoch": 15.998435462842242,
-      "grad_norm": 1.506566047668457,
-      "learning_rate": 6.968140703517588e-05,
-      "loss": 4.7364,
-      "step": 30677
-    },
-    {
-      "epoch": 15.998956975228161,
-      "grad_norm": 1.5108981132507324,
-      "learning_rate": 6.968040201005025e-05,
-      "loss": 5.0951,
-      "step": 30678
-    },
-    {
-      "epoch": 15.99947848761408,
-      "grad_norm": 1.6002683639526367,
-      "learning_rate": 6.967939698492462e-05,
-      "loss": 5.0293,
-      "step": 30679
-    },
-    {
-      "epoch": 16.0,
-      "grad_norm": 1.6395447254180908,
-      "learning_rate": 6.967839195979899e-05,
-      "loss": 5.3493,
-      "step": 30680
-    },
-    {
-      "epoch": 16.00052151238592,
-      "grad_norm": 1.4383502006530762,
-      "learning_rate": 6.967738693467337e-05,
-      "loss": 5.4975,
-      "step": 30681
-    },
-    {
-      "epoch": 16.00104302477184,
-      "grad_norm": 1.4164187908172607,
-      "learning_rate": 6.967638190954774e-05,
-      "loss": 4.9437,
-      "step": 30682
-    },
-    {
-      "epoch": 16.00156453715776,
-      "grad_norm": 1.4440916776657104,
-      "learning_rate": 6.967537688442212e-05,
-      "loss": 5.4525,
-      "step": 30683
-    },
-    {
-      "epoch": 16.002086049543678,
-      "grad_norm": 1.5494117736816406,
-      "learning_rate": 6.967437185929649e-05,
-      "loss": 5.6075,
-      "step": 30684
-    },
-    {
-      "epoch": 16.002607561929597,
-      "grad_norm": 1.5211825370788574,
-      "learning_rate": 6.967336683417086e-05,
-      "loss": 5.3034,
-      "step": 30685
-    },
-    {
-      "epoch": 16.003129074315513,
-      "grad_norm": 1.4596824645996094,
-      "learning_rate": 6.967236180904523e-05,
-      "loss": 5.6188,
-      "step": 30686
-    },
-    {
-      "epoch": 16.003650586701433,
-      "grad_norm": 1.4196031093597412,
-      "learning_rate": 6.96713567839196e-05,
-      "loss": 5.4304,
-      "step": 30687
-    },
-    {
-      "epoch": 16.004172099087352,
-      "grad_norm": 1.4587527513504028,
-      "learning_rate": 6.967035175879397e-05,
-      "loss": 5.4926,
-      "step": 30688
-    },
-    {
-      "epoch": 16.00469361147327,
-      "grad_norm": 1.62749445438385,
-      "learning_rate": 6.966934673366835e-05,
-      "loss": 5.1088,
-      "step": 30689
-    },
-    {
-      "epoch": 16.00521512385919,
-      "grad_norm": 1.4530503749847412,
-      "learning_rate": 6.966834170854271e-05,
-      "loss": 5.4619,
-      "step": 30690
-    },
-    {
-      "epoch": 16.00573663624511,
-      "grad_norm": 1.6244885921478271,
-      "learning_rate": 6.966733668341709e-05,
-      "loss": 4.9491,
-      "step": 30691
-    },
-    {
-      "epoch": 16.00625814863103,
-      "grad_norm": 1.4052766561508179,
-      "learning_rate": 6.966633165829147e-05,
-      "loss": 5.3446,
-      "step": 30692
-    },
-    {
-      "epoch": 16.00677966101695,
-      "grad_norm": 1.492931842803955,
-      "learning_rate": 6.966532663316583e-05,
-      "loss": 5.0921,
-      "step": 30693
-    },
-    {
-      "epoch": 16.00730117340287,
-      "grad_norm": 1.4445297718048096,
-      "learning_rate": 6.966432160804021e-05,
-      "loss": 4.983,
-      "step": 30694
-    },
-    {
-      "epoch": 16.00782268578879,
-      "grad_norm": 1.5709081888198853,
-      "learning_rate": 6.966331658291457e-05,
-      "loss": 5.2576,
-      "step": 30695
-    },
-    {
-      "epoch": 16.008344198174708,
-      "grad_norm": 1.4803013801574707,
-      "learning_rate": 6.966231155778895e-05,
-      "loss": 5.4899,
-      "step": 30696
-    },
-    {
-      "epoch": 16.008865710560627,
-      "grad_norm": 1.5680888891220093,
-      "learning_rate": 6.966130653266332e-05,
-      "loss": 5.2439,
-      "step": 30697
-    },
-    {
-      "epoch": 16.009387222946543,
-      "grad_norm": 1.5051196813583374,
-      "learning_rate": 6.96603015075377e-05,
-      "loss": 5.573,
-      "step": 30698
-    },
-    {
-      "epoch": 16.009908735332463,
-      "grad_norm": 1.5143128633499146,
-      "learning_rate": 6.965929648241206e-05,
-      "loss": 5.3146,
-      "step": 30699
-    },
-    {
-      "epoch": 16.010430247718382,
-      "grad_norm": 1.5312607288360596,
-      "learning_rate": 6.965829145728644e-05,
-      "loss": 4.7296,
-      "step": 30700
-    },
-    {
-      "epoch": 16.0109517601043,
-      "grad_norm": 1.3519586324691772,
-      "learning_rate": 6.96572864321608e-05,
-      "loss": 5.3393,
-      "step": 30701
-    },
-    {
-      "epoch": 16.01147327249022,
-      "grad_norm": 1.4549139738082886,
-      "learning_rate": 6.965628140703518e-05,
-      "loss": 5.4058,
-      "step": 30702
-    },
-    {
-      "epoch": 16.01199478487614,
-      "grad_norm": 1.598228931427002,
-      "learning_rate": 6.965527638190956e-05,
-      "loss": 5.465,
-      "step": 30703
-    },
-    {
-      "epoch": 16.01251629726206,
-      "grad_norm": 1.4141230583190918,
-      "learning_rate": 6.965427135678392e-05,
-      "loss": 5.3034,
-      "step": 30704
-    },
-    {
-      "epoch": 16.01303780964798,
-      "grad_norm": 1.5903054475784302,
-      "learning_rate": 6.96532663316583e-05,
-      "loss": 4.9786,
-      "step": 30705
-    },
-    {
-      "epoch": 16.0135593220339,
-      "grad_norm": 1.4474236965179443,
-      "learning_rate": 6.965226130653266e-05,
-      "loss": 5.3327,
-      "step": 30706
-    },
-    {
-      "epoch": 16.01408083441982,
-      "grad_norm": 1.463449478149414,
-      "learning_rate": 6.965125628140704e-05,
-      "loss": 5.4723,
-      "step": 30707
-    },
-    {
-      "epoch": 16.014602346805738,
-      "grad_norm": 1.4115749597549438,
-      "learning_rate": 6.96502512562814e-05,
-      "loss": 5.6797,
-      "step": 30708
-    },
-    {
-      "epoch": 16.015123859191657,
-      "grad_norm": 1.4481877088546753,
-      "learning_rate": 6.964924623115578e-05,
-      "loss": 5.3943,
-      "step": 30709
-    },
-    {
-      "epoch": 16.015645371577573,
-      "grad_norm": 1.4080175161361694,
-      "learning_rate": 6.964824120603015e-05,
-      "loss": 5.3358,
-      "step": 30710
-    },
-    {
-      "epoch": 16.016166883963493,
-      "grad_norm": 1.3580437898635864,
-      "learning_rate": 6.964723618090452e-05,
-      "loss": 5.5235,
-      "step": 30711
-    },
-    {
-      "epoch": 16.016688396349412,
-      "grad_norm": 1.396807312965393,
-      "learning_rate": 6.96462311557789e-05,
-      "loss": 5.8003,
-      "step": 30712
-    },
-    {
-      "epoch": 16.01720990873533,
-      "grad_norm": 1.4874883890151978,
-      "learning_rate": 6.964522613065328e-05,
-      "loss": 5.3258,
-      "step": 30713
-    },
-    {
-      "epoch": 16.01773142112125,
-      "grad_norm": 1.4225165843963623,
-      "learning_rate": 6.964422110552764e-05,
-      "loss": 5.4203,
-      "step": 30714
-    },
-    {
-      "epoch": 16.01825293350717,
-      "grad_norm": 1.3836719989776611,
-      "learning_rate": 6.964321608040202e-05,
-      "loss": 5.505,
-      "step": 30715
-    },
-    {
-      "epoch": 16.01877444589309,
-      "grad_norm": 1.4745255708694458,
-      "learning_rate": 6.964221105527639e-05,
-      "loss": 5.4366,
-      "step": 30716
-    },
-    {
-      "epoch": 16.01929595827901,
-      "grad_norm": 1.3978497982025146,
-      "learning_rate": 6.964120603015075e-05,
-      "loss": 5.6013,
-      "step": 30717
-    },
-    {
-      "epoch": 16.01981747066493,
-      "grad_norm": 1.6980338096618652,
-      "learning_rate": 6.964020100502513e-05,
-      "loss": 5.1502,
-      "step": 30718
-    },
-    {
-      "epoch": 16.02033898305085,
-      "grad_norm": 1.4987436532974243,
-      "learning_rate": 6.963919597989949e-05,
-      "loss": 5.1799,
-      "step": 30719
-    },
-    {
-      "epoch": 16.020860495436768,
-      "grad_norm": 1.480669379234314,
-      "learning_rate": 6.963819095477387e-05,
-      "loss": 5.1352,
-      "step": 30720
-    },
-    {
-      "epoch": 16.021382007822687,
-      "grad_norm": 1.5366579294204712,
-      "learning_rate": 6.963718592964823e-05,
-      "loss": 4.5396,
-      "step": 30721
-    },
-    {
-      "epoch": 16.021903520208603,
-      "grad_norm": 1.5701473951339722,
-      "learning_rate": 6.963618090452261e-05,
-      "loss": 4.9298,
-      "step": 30722
-    },
-    {
-      "epoch": 16.022425032594523,
-      "grad_norm": 1.6364071369171143,
-      "learning_rate": 6.963517587939699e-05,
-      "loss": 5.2456,
-      "step": 30723
-    },
-    {
-      "epoch": 16.022946544980442,
-      "grad_norm": 1.3644506931304932,
-      "learning_rate": 6.963417085427137e-05,
-      "loss": 5.0971,
-      "step": 30724
-    },
-    {
-      "epoch": 16.02346805736636,
-      "grad_norm": 1.473729133605957,
-      "learning_rate": 6.963316582914573e-05,
-      "loss": 5.2716,
-      "step": 30725
-    },
-    {
-      "epoch": 16.02398956975228,
-      "grad_norm": 1.4491453170776367,
-      "learning_rate": 6.963216080402011e-05,
-      "loss": 5.7795,
-      "step": 30726
-    },
-    {
-      "epoch": 16.0245110821382,
-      "grad_norm": 1.4092381000518799,
-      "learning_rate": 6.963115577889447e-05,
-      "loss": 5.6047,
-      "step": 30727
-    },
-    {
-      "epoch": 16.02503259452412,
-      "grad_norm": 1.6308404207229614,
-      "learning_rate": 6.963015075376885e-05,
-      "loss": 4.6785,
-      "step": 30728
-    },
-    {
-      "epoch": 16.02555410691004,
-      "grad_norm": 1.5323253870010376,
-      "learning_rate": 6.962914572864322e-05,
-      "loss": 5.3031,
-      "step": 30729
-    },
-    {
-      "epoch": 16.02607561929596,
-      "grad_norm": 1.4357200860977173,
-      "learning_rate": 6.962814070351758e-05,
-      "loss": 5.362,
-      "step": 30730
-    },
-    {
-      "epoch": 16.02659713168188,
-      "grad_norm": 1.5199280977249146,
-      "learning_rate": 6.962713567839196e-05,
-      "loss": 5.3031,
-      "step": 30731
-    },
-    {
-      "epoch": 16.027118644067798,
-      "grad_norm": 1.4421982765197754,
-      "learning_rate": 6.962613065326634e-05,
-      "loss": 5.4466,
-      "step": 30732
-    },
-    {
-      "epoch": 16.027640156453717,
-      "grad_norm": 1.4634263515472412,
-      "learning_rate": 6.962512562814071e-05,
-      "loss": 4.8971,
-      "step": 30733
-    },
-    {
-      "epoch": 16.028161668839633,
-      "grad_norm": 1.4031203985214233,
-      "learning_rate": 6.962412060301508e-05,
-      "loss": 5.439,
-      "step": 30734
-    },
-    {
-      "epoch": 16.028683181225553,
-      "grad_norm": 1.5594736337661743,
-      "learning_rate": 6.962311557788946e-05,
-      "loss": 4.6326,
-      "step": 30735
-    },
-    {
-      "epoch": 16.029204693611472,
-      "grad_norm": 1.5561511516571045,
-      "learning_rate": 6.962211055276382e-05,
-      "loss": 5.2251,
-      "step": 30736
-    },
-    {
-      "epoch": 16.02972620599739,
-      "grad_norm": 1.508277177810669,
-      "learning_rate": 6.96211055276382e-05,
-      "loss": 5.2005,
-      "step": 30737
-    },
-    {
-      "epoch": 16.03024771838331,
-      "grad_norm": 1.4919604063034058,
-      "learning_rate": 6.962010050251256e-05,
-      "loss": 5.4909,
-      "step": 30738
-    },
-    {
-      "epoch": 16.03076923076923,
-      "grad_norm": 1.524104356765747,
-      "learning_rate": 6.961909547738694e-05,
-      "loss": 5.2384,
-      "step": 30739
-    },
-    {
-      "epoch": 16.03129074315515,
-      "grad_norm": 1.5942885875701904,
-      "learning_rate": 6.96180904522613e-05,
-      "loss": 4.9453,
-      "step": 30740
-    },
-    {
-      "epoch": 16.03181225554107,
-      "grad_norm": 1.545280933380127,
-      "learning_rate": 6.961708542713568e-05,
-      "loss": 5.125,
-      "step": 30741
-    },
-    {
-      "epoch": 16.03233376792699,
-      "grad_norm": 1.372029185295105,
-      "learning_rate": 6.961608040201005e-05,
-      "loss": 5.4816,
-      "step": 30742
-    },
-    {
-      "epoch": 16.03285528031291,
-      "grad_norm": 1.3651143312454224,
-      "learning_rate": 6.961507537688443e-05,
-      "loss": 5.7728,
-      "step": 30743
-    },
-    {
-      "epoch": 16.033376792698828,
-      "grad_norm": 1.4839656352996826,
-      "learning_rate": 6.96140703517588e-05,
-      "loss": 5.4413,
-      "step": 30744
-    },
-    {
-      "epoch": 16.033898305084747,
-      "grad_norm": 1.4083585739135742,
-      "learning_rate": 6.961306532663317e-05,
-      "loss": 5.0156,
-      "step": 30745
-    },
-    {
-      "epoch": 16.034419817470663,
-      "grad_norm": 1.410679817199707,
-      "learning_rate": 6.961206030150755e-05,
-      "loss": 5.2947,
-      "step": 30746
-    },
-    {
-      "epoch": 16.034941329856583,
-      "grad_norm": 1.5891562700271606,
-      "learning_rate": 6.961105527638191e-05,
-      "loss": 5.1398,
-      "step": 30747
-    },
-    {
-      "epoch": 16.035462842242502,
-      "grad_norm": 1.3878774642944336,
-      "learning_rate": 6.961005025125629e-05,
-      "loss": 5.1432,
-      "step": 30748
-    },
-    {
-      "epoch": 16.03598435462842,
-      "grad_norm": 1.3956979513168335,
-      "learning_rate": 6.960904522613065e-05,
-      "loss": 5.6003,
-      "step": 30749
-    },
-    {
-      "epoch": 16.03650586701434,
-      "grad_norm": 1.4173128604888916,
-      "learning_rate": 6.960804020100503e-05,
-      "loss": 5.6588,
-      "step": 30750
-    },
-    {
-      "epoch": 16.03702737940026,
-      "grad_norm": 1.5231291055679321,
-      "learning_rate": 6.96070351758794e-05,
-      "loss": 5.4948,
-      "step": 30751
-    },
-    {
-      "epoch": 16.03754889178618,
-      "grad_norm": 1.5192276239395142,
-      "learning_rate": 6.960603015075377e-05,
-      "loss": 5.6147,
-      "step": 30752
-    },
-    {
-      "epoch": 16.0380704041721,
-      "grad_norm": 1.5405147075653076,
-      "learning_rate": 6.960502512562815e-05,
-      "loss": 5.1765,
-      "step": 30753
-    },
-    {
-      "epoch": 16.03859191655802,
-      "grad_norm": 1.430746078491211,
-      "learning_rate": 6.960402010050253e-05,
-      "loss": 5.5227,
-      "step": 30754
-    },
-    {
-      "epoch": 16.03911342894394,
-      "grad_norm": 1.572342872619629,
-      "learning_rate": 6.960301507537689e-05,
-      "loss": 5.3694,
-      "step": 30755
-    },
-    {
-      "epoch": 16.039634941329858,
-      "grad_norm": 1.5026111602783203,
-      "learning_rate": 6.960201005025127e-05,
-      "loss": 5.3678,
-      "step": 30756
-    },
-    {
-      "epoch": 16.040156453715777,
-      "grad_norm": 1.5530282258987427,
-      "learning_rate": 6.960100502512563e-05,
-      "loss": 4.7465,
-      "step": 30757
-    },
-    {
-      "epoch": 16.040677966101693,
-      "grad_norm": 1.5196616649627686,
-      "learning_rate": 6.96e-05,
-      "loss": 4.7238,
-      "step": 30758
-    },
-    {
-      "epoch": 16.041199478487613,
-      "grad_norm": 1.5245023965835571,
-      "learning_rate": 6.959899497487438e-05,
-      "loss": 5.3098,
-      "step": 30759
-    },
-    {
-      "epoch": 16.041720990873532,
-      "grad_norm": 1.5207396745681763,
-      "learning_rate": 6.959798994974874e-05,
-      "loss": 5.3664,
-      "step": 30760
-    },
-    {
-      "epoch": 16.042242503259452,
-      "grad_norm": 1.515708565711975,
-      "learning_rate": 6.959698492462312e-05,
-      "loss": 5.1696,
-      "step": 30761
-    },
-    {
-      "epoch": 16.04276401564537,
-      "grad_norm": 1.5146576166152954,
-      "learning_rate": 6.959597989949748e-05,
-      "loss": 5.5416,
-      "step": 30762
-    },
-    {
-      "epoch": 16.04328552803129,
-      "grad_norm": 1.4251456260681152,
-      "learning_rate": 6.959497487437186e-05,
-      "loss": 5.5477,
-      "step": 30763
-    },
-    {
-      "epoch": 16.04380704041721,
-      "grad_norm": 1.4102466106414795,
-      "learning_rate": 6.959396984924624e-05,
-      "loss": 5.6486,
-      "step": 30764
-    },
-    {
-      "epoch": 16.04432855280313,
-      "grad_norm": 1.579641580581665,
-      "learning_rate": 6.959296482412062e-05,
-      "loss": 5.125,
-      "step": 30765
-    },
-    {
-      "epoch": 16.04485006518905,
-      "grad_norm": 1.4488950967788696,
-      "learning_rate": 6.959195979899498e-05,
-      "loss": 5.2877,
-      "step": 30766
-    },
-    {
-      "epoch": 16.04537157757497,
-      "grad_norm": 1.6173425912857056,
-      "learning_rate": 6.959095477386936e-05,
-      "loss": 4.7283,
-      "step": 30767
-    },
-    {
-      "epoch": 16.045893089960888,
-      "grad_norm": 1.4309731721878052,
-      "learning_rate": 6.958994974874372e-05,
-      "loss": 5.3824,
-      "step": 30768
-    },
-    {
-      "epoch": 16.046414602346807,
-      "grad_norm": 1.4833085536956787,
-      "learning_rate": 6.95889447236181e-05,
-      "loss": 4.993,
-      "step": 30769
-    },
-    {
-      "epoch": 16.046936114732723,
-      "grad_norm": 1.6186977624893188,
-      "learning_rate": 6.958793969849246e-05,
-      "loss": 5.1882,
-      "step": 30770
-    },
-    {
-      "epoch": 16.047457627118643,
-      "grad_norm": 1.495772123336792,
-      "learning_rate": 6.958693467336683e-05,
-      "loss": 5.6713,
-      "step": 30771
-    },
-    {
-      "epoch": 16.047979139504562,
-      "grad_norm": 1.4899786710739136,
-      "learning_rate": 6.95859296482412e-05,
-      "loss": 4.9879,
-      "step": 30772
-    },
-    {
-      "epoch": 16.048500651890482,
-      "grad_norm": 1.4436489343643188,
-      "learning_rate": 6.958492462311558e-05,
-      "loss": 5.2566,
-      "step": 30773
-    },
-    {
-      "epoch": 16.0490221642764,
-      "grad_norm": 1.536820411682129,
-      "learning_rate": 6.958391959798996e-05,
-      "loss": 5.066,
-      "step": 30774
-    },
-    {
-      "epoch": 16.04954367666232,
-      "grad_norm": 1.6025874614715576,
-      "learning_rate": 6.958291457286433e-05,
-      "loss": 5.4341,
-      "step": 30775
-    },
-    {
-      "epoch": 16.05006518904824,
-      "grad_norm": 1.5532057285308838,
-      "learning_rate": 6.95819095477387e-05,
-      "loss": 5.2187,
-      "step": 30776
-    },
-    {
-      "epoch": 16.05058670143416,
-      "grad_norm": 1.387904405593872,
-      "learning_rate": 6.958090452261307e-05,
-      "loss": 5.5832,
-      "step": 30777
-    },
-    {
-      "epoch": 16.05110821382008,
-      "grad_norm": 1.5432369709014893,
-      "learning_rate": 6.957989949748745e-05,
-      "loss": 5.0217,
-      "step": 30778
-    },
-    {
-      "epoch": 16.051629726206,
-      "grad_norm": 1.4339786767959595,
-      "learning_rate": 6.957889447236181e-05,
-      "loss": 5.3005,
-      "step": 30779
-    },
-    {
-      "epoch": 16.052151238591918,
-      "grad_norm": 1.4091196060180664,
-      "learning_rate": 6.957788944723619e-05,
-      "loss": 5.377,
-      "step": 30780
-    },
-    {
-      "epoch": 16.052672750977834,
-      "grad_norm": 1.4629555940628052,
-      "learning_rate": 6.957688442211055e-05,
-      "loss": 5.2593,
-      "step": 30781
-    },
-    {
-      "epoch": 16.053194263363753,
-      "grad_norm": 1.4373677968978882,
-      "learning_rate": 6.957587939698493e-05,
-      "loss": 5.5236,
-      "step": 30782
-    },
-    {
-      "epoch": 16.053715775749673,
-      "grad_norm": 1.4925565719604492,
-      "learning_rate": 6.95748743718593e-05,
-      "loss": 5.4025,
-      "step": 30783
-    },
-    {
-      "epoch": 16.054237288135592,
-      "grad_norm": 1.4511865377426147,
-      "learning_rate": 6.957386934673367e-05,
-      "loss": 5.4201,
-      "step": 30784
-    },
-    {
-      "epoch": 16.054758800521512,
-      "grad_norm": 1.5480666160583496,
-      "learning_rate": 6.957286432160805e-05,
-      "loss": 4.7855,
-      "step": 30785
-    },
-    {
-      "epoch": 16.05528031290743,
-      "grad_norm": 1.5205507278442383,
-      "learning_rate": 6.957185929648241e-05,
-      "loss": 5.0553,
-      "step": 30786
-    },
-    {
-      "epoch": 16.05580182529335,
-      "grad_norm": 1.4433856010437012,
-      "learning_rate": 6.957085427135679e-05,
-      "loss": 5.3648,
-      "step": 30787
-    },
-    {
-      "epoch": 16.05632333767927,
-      "grad_norm": 1.5701196193695068,
-      "learning_rate": 6.956984924623116e-05,
-      "loss": 4.8602,
-      "step": 30788
-    },
-    {
-      "epoch": 16.05684485006519,
-      "grad_norm": 1.552069067955017,
-      "learning_rate": 6.956884422110553e-05,
-      "loss": 5.0333,
-      "step": 30789
-    },
-    {
-      "epoch": 16.05736636245111,
-      "grad_norm": 1.454399585723877,
-      "learning_rate": 6.95678391959799e-05,
-      "loss": 5.5901,
-      "step": 30790
-    },
-    {
-      "epoch": 16.05788787483703,
-      "grad_norm": 1.445487380027771,
-      "learning_rate": 6.956683417085428e-05,
-      "loss": 5.3426,
-      "step": 30791
-    },
-    {
-      "epoch": 16.058409387222948,
-      "grad_norm": 1.4732953310012817,
-      "learning_rate": 6.956582914572864e-05,
-      "loss": 4.86,
-      "step": 30792
-    },
-    {
-      "epoch": 16.058930899608864,
-      "grad_norm": 1.4283004999160767,
-      "learning_rate": 6.956482412060302e-05,
-      "loss": 5.4564,
-      "step": 30793
-    },
-    {
-      "epoch": 16.059452411994783,
-      "grad_norm": 1.4534945487976074,
-      "learning_rate": 6.956381909547738e-05,
-      "loss": 5.4108,
-      "step": 30794
-    },
-    {
-      "epoch": 16.059973924380703,
-      "grad_norm": 1.4278135299682617,
-      "learning_rate": 6.956281407035176e-05,
-      "loss": 5.595,
-      "step": 30795
-    },
-    {
-      "epoch": 16.060495436766622,
-      "grad_norm": 1.446648359298706,
-      "learning_rate": 6.956180904522614e-05,
-      "loss": 5.2716,
-      "step": 30796
-    },
-    {
-      "epoch": 16.061016949152542,
-      "grad_norm": 1.4647767543792725,
-      "learning_rate": 6.95608040201005e-05,
-      "loss": 5.1787,
-      "step": 30797
-    },
-    {
-      "epoch": 16.06153846153846,
-      "grad_norm": 1.4536501169204712,
-      "learning_rate": 6.955979899497488e-05,
-      "loss": 5.4565,
-      "step": 30798
-    },
-    {
-      "epoch": 16.06205997392438,
-      "grad_norm": 1.533584713935852,
-      "learning_rate": 6.955879396984924e-05,
-      "loss": 5.6364,
-      "step": 30799
-    },
-    {
-      "epoch": 16.0625814863103,
-      "grad_norm": 1.474989414215088,
-      "learning_rate": 6.955778894472362e-05,
-      "loss": 5.1992,
-      "step": 30800
-    },
-    {
-      "epoch": 16.06310299869622,
-      "grad_norm": 1.4761842489242554,
-      "learning_rate": 6.955678391959799e-05,
-      "loss": 5.5837,
-      "step": 30801
-    },
-    {
-      "epoch": 16.06362451108214,
-      "grad_norm": 1.454040765762329,
-      "learning_rate": 6.955577889447236e-05,
-      "loss": 5.4461,
-      "step": 30802
-    },
-    {
-      "epoch": 16.06414602346806,
-      "grad_norm": 1.5145896673202515,
-      "learning_rate": 6.955477386934673e-05,
-      "loss": 4.9639,
-      "step": 30803
-    },
-    {
-      "epoch": 16.064667535853978,
-      "grad_norm": 1.465885043144226,
-      "learning_rate": 6.95537688442211e-05,
-      "loss": 5.1112,
-      "step": 30804
-    },
-    {
-      "epoch": 16.065189048239894,
-      "grad_norm": 1.4393587112426758,
-      "learning_rate": 6.955276381909548e-05,
-      "loss": 5.2282,
-      "step": 30805
-    },
-    {
-      "epoch": 16.065710560625813,
-      "grad_norm": 1.4786702394485474,
-      "learning_rate": 6.955175879396986e-05,
-      "loss": 5.4531,
-      "step": 30806
-    },
-    {
-      "epoch": 16.066232073011733,
-      "grad_norm": 1.3969494104385376,
-      "learning_rate": 6.955075376884423e-05,
-      "loss": 5.4046,
-      "step": 30807
-    },
-    {
-      "epoch": 16.066753585397652,
-      "grad_norm": 1.5604004859924316,
-      "learning_rate": 6.95497487437186e-05,
-      "loss": 4.4956,
-      "step": 30808
-    },
-    {
-      "epoch": 16.067275097783572,
-      "grad_norm": 1.6156693696975708,
-      "learning_rate": 6.954874371859297e-05,
-      "loss": 4.9775,
-      "step": 30809
-    },
-    {
-      "epoch": 16.06779661016949,
-      "grad_norm": 1.5552117824554443,
-      "learning_rate": 6.954773869346733e-05,
-      "loss": 5.5124,
-      "step": 30810
-    },
-    {
-      "epoch": 16.06831812255541,
-      "grad_norm": 1.5377196073532104,
-      "learning_rate": 6.954673366834171e-05,
-      "loss": 5.3951,
-      "step": 30811
-    },
-    {
-      "epoch": 16.06883963494133,
-      "grad_norm": 1.4350299835205078,
-      "learning_rate": 6.954572864321608e-05,
-      "loss": 5.7282,
-      "step": 30812
-    },
-    {
-      "epoch": 16.06936114732725,
-      "grad_norm": 1.5535482168197632,
-      "learning_rate": 6.954472361809045e-05,
-      "loss": 5.2758,
-      "step": 30813
-    },
-    {
-      "epoch": 16.06988265971317,
-      "grad_norm": 1.4437572956085205,
-      "learning_rate": 6.954371859296482e-05,
-      "loss": 5.4664,
-      "step": 30814
-    },
-    {
-      "epoch": 16.07040417209909,
-      "grad_norm": 1.4444600343704224,
-      "learning_rate": 6.95427135678392e-05,
-      "loss": 5.6582,
-      "step": 30815
-    },
-    {
-      "epoch": 16.070925684485008,
-      "grad_norm": 1.4941636323928833,
-      "learning_rate": 6.954170854271357e-05,
-      "loss": 5.4987,
-      "step": 30816
-    },
-    {
-      "epoch": 16.071447196870924,
-      "grad_norm": 1.460801601409912,
-      "learning_rate": 6.954070351758795e-05,
-      "loss": 5.7744,
-      "step": 30817
-    },
-    {
-      "epoch": 16.071968709256844,
-      "grad_norm": 1.4305002689361572,
-      "learning_rate": 6.953969849246232e-05,
-      "loss": 5.193,
-      "step": 30818
-    },
-    {
-      "epoch": 16.072490221642763,
-      "grad_norm": 1.4666286706924438,
-      "learning_rate": 6.953869346733669e-05,
-      "loss": 5.299,
-      "step": 30819
-    },
-    {
-      "epoch": 16.073011734028682,
-      "grad_norm": 1.448548674583435,
-      "learning_rate": 6.953768844221106e-05,
-      "loss": 5.6526,
-      "step": 30820
-    },
-    {
-      "epoch": 16.073533246414602,
-      "grad_norm": 1.4030627012252808,
-      "learning_rate": 6.953668341708543e-05,
-      "loss": 5.3281,
-      "step": 30821
-    },
-    {
-      "epoch": 16.07405475880052,
-      "grad_norm": 1.3624361753463745,
-      "learning_rate": 6.95356783919598e-05,
-      "loss": 5.1978,
-      "step": 30822
-    },
-    {
-      "epoch": 16.07457627118644,
-      "grad_norm": 1.4505882263183594,
-      "learning_rate": 6.953467336683416e-05,
-      "loss": 5.3194,
-      "step": 30823
-    },
-    {
-      "epoch": 16.07509778357236,
-      "grad_norm": 1.4724020957946777,
-      "learning_rate": 6.953366834170854e-05,
-      "loss": 5.1698,
-      "step": 30824
-    },
-    {
-      "epoch": 16.07561929595828,
-      "grad_norm": 1.3356804847717285,
-      "learning_rate": 6.953266331658292e-05,
-      "loss": 4.446,
-      "step": 30825
-    },
-    {
-      "epoch": 16.0761408083442,
-      "grad_norm": 1.4150460958480835,
-      "learning_rate": 6.95316582914573e-05,
-      "loss": 5.5115,
-      "step": 30826
-    },
-    {
-      "epoch": 16.07666232073012,
-      "grad_norm": 1.7111420631408691,
-      "learning_rate": 6.953065326633166e-05,
-      "loss": 5.0623,
-      "step": 30827
-    },
-    {
-      "epoch": 16.077183833116038,
-      "grad_norm": 1.5638868808746338,
-      "learning_rate": 6.952964824120604e-05,
-      "loss": 5.1468,
-      "step": 30828
-    },
-    {
-      "epoch": 16.077705345501954,
-      "grad_norm": 1.3936872482299805,
-      "learning_rate": 6.95286432160804e-05,
-      "loss": 5.2815,
-      "step": 30829
-    },
-    {
-      "epoch": 16.078226857887874,
-      "grad_norm": 1.4560637474060059,
-      "learning_rate": 6.952763819095478e-05,
-      "loss": 5.6044,
-      "step": 30830
-    },
-    {
-      "epoch": 16.078748370273793,
-      "grad_norm": 1.5335006713867188,
-      "learning_rate": 6.952663316582915e-05,
-      "loss": 4.9223,
-      "step": 30831
-    },
-    {
-      "epoch": 16.079269882659712,
-      "grad_norm": 1.3708525896072388,
-      "learning_rate": 6.952562814070352e-05,
-      "loss": 5.7652,
-      "step": 30832
-    },
-    {
-      "epoch": 16.079791395045632,
-      "grad_norm": 1.451716423034668,
-      "learning_rate": 6.952462311557789e-05,
-      "loss": 5.2468,
-      "step": 30833
-    },
-    {
-      "epoch": 16.08031290743155,
-      "grad_norm": 1.4747563600540161,
-      "learning_rate": 6.952361809045227e-05,
-      "loss": 5.1203,
-      "step": 30834
-    },
-    {
-      "epoch": 16.08083441981747,
-      "grad_norm": 1.8183717727661133,
-      "learning_rate": 6.952261306532663e-05,
-      "loss": 4.413,
-      "step": 30835
-    },
-    {
-      "epoch": 16.08135593220339,
-      "grad_norm": 1.525620937347412,
-      "learning_rate": 6.952160804020101e-05,
-      "loss": 4.991,
-      "step": 30836
-    },
-    {
-      "epoch": 16.08187744458931,
-      "grad_norm": 1.4709784984588623,
-      "learning_rate": 6.952060301507539e-05,
-      "loss": 5.3778,
-      "step": 30837
-    },
-    {
-      "epoch": 16.08239895697523,
-      "grad_norm": 1.4669103622436523,
-      "learning_rate": 6.951959798994975e-05,
-      "loss": 5.048,
-      "step": 30838
-    },
-    {
-      "epoch": 16.08292046936115,
-      "grad_norm": 1.5365883111953735,
-      "learning_rate": 6.951859296482413e-05,
-      "loss": 4.8725,
-      "step": 30839
-    },
-    {
-      "epoch": 16.083441981747068,
-      "grad_norm": 1.4692875146865845,
-      "learning_rate": 6.951758793969849e-05,
-      "loss": 5.0296,
-      "step": 30840
-    },
-    {
-      "epoch": 16.083963494132984,
-      "grad_norm": 1.5271166563034058,
-      "learning_rate": 6.951658291457287e-05,
-      "loss": 5.6641,
-      "step": 30841
-    },
-    {
-      "epoch": 16.084485006518904,
-      "grad_norm": 1.489038109779358,
-      "learning_rate": 6.951557788944723e-05,
-      "loss": 5.356,
-      "step": 30842
-    },
-    {
-      "epoch": 16.085006518904823,
-      "grad_norm": 1.479081392288208,
-      "learning_rate": 6.951457286432161e-05,
-      "loss": 5.4339,
-      "step": 30843
-    },
-    {
-      "epoch": 16.085528031290742,
-      "grad_norm": 1.411893606185913,
-      "learning_rate": 6.951356783919598e-05,
-      "loss": 5.5943,
-      "step": 30844
-    },
-    {
-      "epoch": 16.086049543676662,
-      "grad_norm": 1.4794155359268188,
-      "learning_rate": 6.951256281407035e-05,
-      "loss": 5.4013,
-      "step": 30845
-    },
-    {
-      "epoch": 16.08657105606258,
-      "grad_norm": 1.391089677810669,
-      "learning_rate": 6.951155778894473e-05,
-      "loss": 5.5906,
-      "step": 30846
-    },
-    {
-      "epoch": 16.0870925684485,
-      "grad_norm": 1.456228256225586,
-      "learning_rate": 6.951055276381911e-05,
-      "loss": 5.1769,
-      "step": 30847
-    },
-    {
-      "epoch": 16.08761408083442,
-      "grad_norm": 1.529538869857788,
-      "learning_rate": 6.950954773869347e-05,
-      "loss": 5.4167,
-      "step": 30848
-    },
-    {
-      "epoch": 16.08813559322034,
-      "grad_norm": 1.5385688543319702,
-      "learning_rate": 6.950854271356785e-05,
-      "loss": 5.0542,
-      "step": 30849
-    },
-    {
-      "epoch": 16.08865710560626,
-      "grad_norm": 1.464777946472168,
-      "learning_rate": 6.950753768844222e-05,
-      "loss": 4.6664,
-      "step": 30850
-    },
-    {
-      "epoch": 16.08917861799218,
-      "grad_norm": 1.449520468711853,
-      "learning_rate": 6.950653266331658e-05,
-      "loss": 5.1737,
-      "step": 30851
-    },
-    {
-      "epoch": 16.089700130378098,
-      "grad_norm": 1.4036203622817993,
-      "learning_rate": 6.950552763819096e-05,
-      "loss": 5.5393,
-      "step": 30852
-    },
-    {
-      "epoch": 16.090221642764014,
-      "grad_norm": 1.5774061679840088,
-      "learning_rate": 6.950452261306532e-05,
-      "loss": 5.2717,
-      "step": 30853
-    },
-    {
-      "epoch": 16.090743155149934,
-      "grad_norm": 1.5585153102874756,
-      "learning_rate": 6.95035175879397e-05,
-      "loss": 5.2473,
-      "step": 30854
-    },
-    {
-      "epoch": 16.091264667535853,
-      "grad_norm": 1.5414174795150757,
-      "learning_rate": 6.950251256281406e-05,
-      "loss": 5.1009,
-      "step": 30855
-    },
-    {
-      "epoch": 16.091786179921773,
-      "grad_norm": 1.6362191438674927,
-      "learning_rate": 6.950150753768844e-05,
-      "loss": 5.2775,
-      "step": 30856
-    },
-    {
-      "epoch": 16.092307692307692,
-      "grad_norm": 1.4857923984527588,
-      "learning_rate": 6.950050251256282e-05,
-      "loss": 5.5684,
-      "step": 30857
-    },
-    {
-      "epoch": 16.09282920469361,
-      "grad_norm": 1.3971922397613525,
-      "learning_rate": 6.94994974874372e-05,
-      "loss": 5.2494,
-      "step": 30858
-    },
-    {
-      "epoch": 16.09335071707953,
-      "grad_norm": 1.5428810119628906,
-      "learning_rate": 6.949849246231156e-05,
-      "loss": 4.7019,
-      "step": 30859
-    },
-    {
-      "epoch": 16.09387222946545,
-      "grad_norm": 1.491753101348877,
-      "learning_rate": 6.949748743718594e-05,
-      "loss": 5.2434,
-      "step": 30860
-    },
-    {
-      "epoch": 16.09439374185137,
-      "grad_norm": 1.6509264707565308,
-      "learning_rate": 6.94964824120603e-05,
-      "loss": 5.3474,
-      "step": 30861
-    },
-    {
-      "epoch": 16.09491525423729,
-      "grad_norm": 1.9592207670211792,
-      "learning_rate": 6.949547738693468e-05,
-      "loss": 5.2409,
-      "step": 30862
-    },
-    {
-      "epoch": 16.09543676662321,
-      "grad_norm": 1.4251818656921387,
-      "learning_rate": 6.949447236180905e-05,
-      "loss": 5.3696,
-      "step": 30863
-    },
-    {
-      "epoch": 16.09595827900913,
-      "grad_norm": 1.4616318941116333,
-      "learning_rate": 6.949346733668341e-05,
-      "loss": 5.5675,
-      "step": 30864
-    },
-    {
-      "epoch": 16.096479791395044,
-      "grad_norm": 1.376705527305603,
-      "learning_rate": 6.949246231155779e-05,
-      "loss": 5.4529,
-      "step": 30865
-    },
-    {
-      "epoch": 16.097001303780964,
-      "grad_norm": 1.4795376062393188,
-      "learning_rate": 6.949145728643217e-05,
-      "loss": 5.3177,
-      "step": 30866
-    },
-    {
-      "epoch": 16.097522816166883,
-      "grad_norm": 1.6817528009414673,
-      "learning_rate": 6.949045226130654e-05,
-      "loss": 5.3335,
-      "step": 30867
-    },
-    {
-      "epoch": 16.098044328552803,
-      "grad_norm": 1.4022592306137085,
-      "learning_rate": 6.948944723618091e-05,
-      "loss": 5.4305,
-      "step": 30868
-    },
-    {
-      "epoch": 16.098565840938722,
-      "grad_norm": 1.4894222021102905,
-      "learning_rate": 6.948844221105529e-05,
-      "loss": 4.7963,
-      "step": 30869
-    },
-    {
-      "epoch": 16.09908735332464,
-      "grad_norm": 1.5019277334213257,
-      "learning_rate": 6.948743718592965e-05,
-      "loss": 5.6161,
-      "step": 30870
-    },
-    {
-      "epoch": 16.09960886571056,
-      "grad_norm": 1.5007206201553345,
-      "learning_rate": 6.948643216080403e-05,
-      "loss": 5.1349,
-      "step": 30871
-    },
-    {
-      "epoch": 16.10013037809648,
-      "grad_norm": 1.554226040840149,
-      "learning_rate": 6.948542713567839e-05,
-      "loss": 5.6397,
-      "step": 30872
-    },
-    {
-      "epoch": 16.1006518904824,
-      "grad_norm": 1.5866203308105469,
-      "learning_rate": 6.948442211055277e-05,
-      "loss": 5.0422,
-      "step": 30873
-    },
-    {
-      "epoch": 16.10117340286832,
-      "grad_norm": 1.5310615301132202,
-      "learning_rate": 6.948341708542713e-05,
-      "loss": 5.063,
-      "step": 30874
-    },
-    {
-      "epoch": 16.10169491525424,
-      "grad_norm": 1.42096745967865,
-      "learning_rate": 6.948241206030151e-05,
-      "loss": 5.7081,
-      "step": 30875
-    },
-    {
-      "epoch": 16.102216427640155,
-      "grad_norm": 1.3846937417984009,
-      "learning_rate": 6.948140703517588e-05,
-      "loss": 5.6365,
-      "step": 30876
-    },
-    {
-      "epoch": 16.102737940026074,
-      "grad_norm": 1.4241719245910645,
-      "learning_rate": 6.948040201005025e-05,
-      "loss": 5.6761,
-      "step": 30877
-    },
-    {
-      "epoch": 16.103259452411994,
-      "grad_norm": 1.4752211570739746,
-      "learning_rate": 6.947939698492463e-05,
-      "loss": 5.2547,
-      "step": 30878
-    },
-    {
-      "epoch": 16.103780964797913,
-      "grad_norm": 1.5101789236068726,
-      "learning_rate": 6.9478391959799e-05,
-      "loss": 5.5612,
-      "step": 30879
-    },
-    {
-      "epoch": 16.104302477183833,
-      "grad_norm": 1.4149963855743408,
-      "learning_rate": 6.947738693467337e-05,
-      "loss": 5.3995,
-      "step": 30880
-    },
-    {
-      "epoch": 16.104823989569752,
-      "grad_norm": 1.4295270442962646,
-      "learning_rate": 6.947638190954774e-05,
-      "loss": 5.5771,
-      "step": 30881
-    },
-    {
-      "epoch": 16.10534550195567,
-      "grad_norm": 1.3884633779525757,
-      "learning_rate": 6.947537688442212e-05,
-      "loss": 4.7332,
-      "step": 30882
-    },
-    {
-      "epoch": 16.10586701434159,
-      "grad_norm": 1.4202823638916016,
-      "learning_rate": 6.947437185929648e-05,
-      "loss": 5.6213,
-      "step": 30883
-    },
-    {
-      "epoch": 16.10638852672751,
-      "grad_norm": 1.4770076274871826,
-      "learning_rate": 6.947336683417086e-05,
-      "loss": 5.2598,
-      "step": 30884
-    },
-    {
-      "epoch": 16.10691003911343,
-      "grad_norm": 1.522358775138855,
-      "learning_rate": 6.947236180904522e-05,
-      "loss": 5.3174,
-      "step": 30885
-    },
-    {
-      "epoch": 16.10743155149935,
-      "grad_norm": 1.4711663722991943,
-      "learning_rate": 6.94713567839196e-05,
-      "loss": 5.3898,
-      "step": 30886
-    },
-    {
-      "epoch": 16.10795306388527,
-      "grad_norm": 1.4753514528274536,
-      "learning_rate": 6.947035175879398e-05,
-      "loss": 4.8557,
-      "step": 30887
-    },
-    {
-      "epoch": 16.108474576271185,
-      "grad_norm": 1.395708680152893,
-      "learning_rate": 6.946934673366836e-05,
-      "loss": 5.3784,
-      "step": 30888
-    },
-    {
-      "epoch": 16.108996088657104,
-      "grad_norm": 1.3904067277908325,
-      "learning_rate": 6.946834170854272e-05,
-      "loss": 5.6937,
-      "step": 30889
-    },
-    {
-      "epoch": 16.109517601043024,
-      "grad_norm": 1.3827471733093262,
-      "learning_rate": 6.946733668341708e-05,
-      "loss": 5.3526,
-      "step": 30890
-    },
-    {
-      "epoch": 16.110039113428943,
-      "grad_norm": 1.398087739944458,
-      "learning_rate": 6.946633165829146e-05,
-      "loss": 4.6958,
-      "step": 30891
-    },
-    {
-      "epoch": 16.110560625814863,
-      "grad_norm": 1.5586256980895996,
-      "learning_rate": 6.946532663316583e-05,
-      "loss": 4.9995,
-      "step": 30892
-    },
-    {
-      "epoch": 16.111082138200782,
-      "grad_norm": 1.4514992237091064,
-      "learning_rate": 6.94643216080402e-05,
-      "loss": 5.6206,
-      "step": 30893
-    },
-    {
-      "epoch": 16.1116036505867,
-      "grad_norm": 1.4911390542984009,
-      "learning_rate": 6.946331658291457e-05,
-      "loss": 5.0377,
-      "step": 30894
-    },
-    {
-      "epoch": 16.11212516297262,
-      "grad_norm": 1.3648226261138916,
-      "learning_rate": 6.946231155778895e-05,
-      "loss": 5.6885,
-      "step": 30895
-    },
-    {
-      "epoch": 16.11264667535854,
-      "grad_norm": 1.5237088203430176,
-      "learning_rate": 6.946130653266331e-05,
-      "loss": 5.0068,
-      "step": 30896
-    },
-    {
-      "epoch": 16.11316818774446,
-      "grad_norm": 1.6293805837631226,
-      "learning_rate": 6.946030150753769e-05,
-      "loss": 5.1995,
-      "step": 30897
-    },
-    {
-      "epoch": 16.11368970013038,
-      "grad_norm": 1.6354091167449951,
-      "learning_rate": 6.945929648241207e-05,
-      "loss": 4.6147,
-      "step": 30898
-    },
-    {
-      "epoch": 16.1142112125163,
-      "grad_norm": 1.5455366373062134,
-      "learning_rate": 6.945829145728644e-05,
-      "loss": 5.6476,
-      "step": 30899
-    },
-    {
-      "epoch": 16.114732724902215,
-      "grad_norm": 1.561474323272705,
-      "learning_rate": 6.945728643216081e-05,
-      "loss": 5.0705,
-      "step": 30900
-    },
-    {
-      "epoch": 16.115254237288134,
-      "grad_norm": 1.4214694499969482,
-      "learning_rate": 6.945628140703519e-05,
-      "loss": 5.3772,
-      "step": 30901
-    },
-    {
-      "epoch": 16.115775749674054,
-      "grad_norm": 1.5568349361419678,
-      "learning_rate": 6.945527638190955e-05,
-      "loss": 4.9576,
-      "step": 30902
-    },
-    {
-      "epoch": 16.116297262059973,
-      "grad_norm": 1.608067274093628,
-      "learning_rate": 6.945427135678392e-05,
-      "loss": 5.0342,
-      "step": 30903
-    },
-    {
-      "epoch": 16.116818774445893,
-      "grad_norm": 1.479416012763977,
-      "learning_rate": 6.945326633165829e-05,
-      "loss": 5.1131,
-      "step": 30904
-    },
-    {
-      "epoch": 16.117340286831812,
-      "grad_norm": 1.4494872093200684,
-      "learning_rate": 6.945226130653266e-05,
-      "loss": 5.6296,
-      "step": 30905
-    },
-    {
-      "epoch": 16.11786179921773,
-      "grad_norm": 1.4571239948272705,
-      "learning_rate": 6.945125628140704e-05,
-      "loss": 5.3533,
-      "step": 30906
-    },
-    {
-      "epoch": 16.11838331160365,
-      "grad_norm": 1.4834619760513306,
-      "learning_rate": 6.945025125628141e-05,
-      "loss": 5.6539,
-      "step": 30907
-    },
-    {
-      "epoch": 16.11890482398957,
-      "grad_norm": 1.8632839918136597,
-      "learning_rate": 6.944924623115579e-05,
-      "loss": 4.5655,
-      "step": 30908
-    },
-    {
-      "epoch": 16.11942633637549,
-      "grad_norm": 1.5673547983169556,
-      "learning_rate": 6.944824120603016e-05,
-      "loss": 5.3448,
-      "step": 30909
-    },
-    {
-      "epoch": 16.11994784876141,
-      "grad_norm": 1.4211896657943726,
-      "learning_rate": 6.944723618090453e-05,
-      "loss": 4.957,
-      "step": 30910
-    },
-    {
-      "epoch": 16.12046936114733,
-      "grad_norm": 1.5035418272018433,
-      "learning_rate": 6.94462311557789e-05,
-      "loss": 5.3665,
-      "step": 30911
-    },
-    {
-      "epoch": 16.120990873533245,
-      "grad_norm": 1.5876163244247437,
-      "learning_rate": 6.944522613065328e-05,
-      "loss": 5.4336,
-      "step": 30912
-    },
-    {
-      "epoch": 16.121512385919164,
-      "grad_norm": 1.491905927658081,
-      "learning_rate": 6.944422110552764e-05,
-      "loss": 5.5141,
-      "step": 30913
-    },
-    {
-      "epoch": 16.122033898305084,
-      "grad_norm": 1.4798401594161987,
-      "learning_rate": 6.944321608040202e-05,
-      "loss": 5.275,
-      "step": 30914
-    },
-    {
-      "epoch": 16.122555410691003,
-      "grad_norm": 1.5043474435806274,
-      "learning_rate": 6.944221105527638e-05,
-      "loss": 5.0673,
-      "step": 30915
-    },
-    {
-      "epoch": 16.123076923076923,
-      "grad_norm": 1.4829188585281372,
-      "learning_rate": 6.944120603015075e-05,
-      "loss": 5.589,
-      "step": 30916
-    },
-    {
-      "epoch": 16.123598435462842,
-      "grad_norm": 1.6205490827560425,
-      "learning_rate": 6.944020100502512e-05,
-      "loss": 5.4894,
-      "step": 30917
-    },
-    {
-      "epoch": 16.12411994784876,
-      "grad_norm": 1.5188641548156738,
-      "learning_rate": 6.94391959798995e-05,
-      "loss": 5.2745,
-      "step": 30918
-    },
-    {
-      "epoch": 16.12464146023468,
-      "grad_norm": 1.5637530088424683,
-      "learning_rate": 6.943819095477388e-05,
-      "loss": 5.6209,
-      "step": 30919
-    },
-    {
-      "epoch": 16.1251629726206,
-      "grad_norm": 1.4995582103729248,
-      "learning_rate": 6.943718592964824e-05,
-      "loss": 5.5466,
-      "step": 30920
-    },
-    {
-      "epoch": 16.12568448500652,
-      "grad_norm": 1.3802902698516846,
-      "learning_rate": 6.943618090452262e-05,
-      "loss": 5.61,
-      "step": 30921
-    },
-    {
-      "epoch": 16.12620599739244,
-      "grad_norm": 1.5374958515167236,
-      "learning_rate": 6.943517587939699e-05,
-      "loss": 4.9995,
-      "step": 30922
-    },
-    {
-      "epoch": 16.12672750977836,
-      "grad_norm": 1.446828007698059,
-      "learning_rate": 6.943417085427136e-05,
-      "loss": 5.5109,
-      "step": 30923
-    },
-    {
-      "epoch": 16.127249022164275,
-      "grad_norm": 1.517698049545288,
-      "learning_rate": 6.943316582914573e-05,
-      "loss": 5.2388,
-      "step": 30924
-    },
-    {
-      "epoch": 16.127770534550194,
-      "grad_norm": 1.506032109260559,
-      "learning_rate": 6.94321608040201e-05,
-      "loss": 5.1549,
-      "step": 30925
-    },
-    {
-      "epoch": 16.128292046936114,
-      "grad_norm": 1.4762961864471436,
-      "learning_rate": 6.943115577889447e-05,
-      "loss": 5.8069,
-      "step": 30926
-    },
-    {
-      "epoch": 16.128813559322033,
-      "grad_norm": 1.5680407285690308,
-      "learning_rate": 6.943015075376885e-05,
-      "loss": 5.2227,
-      "step": 30927
-    },
-    {
-      "epoch": 16.129335071707953,
-      "grad_norm": 1.452688217163086,
-      "learning_rate": 6.942914572864323e-05,
-      "loss": 5.1458,
-      "step": 30928
-    },
-    {
-      "epoch": 16.129856584093872,
-      "grad_norm": 1.4103760719299316,
-      "learning_rate": 6.94281407035176e-05,
-      "loss": 5.4656,
-      "step": 30929
-    },
-    {
-      "epoch": 16.13037809647979,
-      "grad_norm": 1.4444475173950195,
-      "learning_rate": 6.942713567839197e-05,
-      "loss": 5.7472,
-      "step": 30930
-    },
-    {
-      "epoch": 16.13089960886571,
-      "grad_norm": 1.454750657081604,
-      "learning_rate": 6.942613065326633e-05,
-      "loss": 5.2178,
-      "step": 30931
-    },
-    {
-      "epoch": 16.13142112125163,
-      "grad_norm": 1.6125377416610718,
-      "learning_rate": 6.942512562814071e-05,
-      "loss": 5.059,
-      "step": 30932
-    },
-    {
-      "epoch": 16.13194263363755,
-      "grad_norm": 1.6152496337890625,
-      "learning_rate": 6.942412060301507e-05,
-      "loss": 5.2947,
-      "step": 30933
-    },
-    {
-      "epoch": 16.13246414602347,
-      "grad_norm": 1.4299639463424683,
-      "learning_rate": 6.942311557788945e-05,
-      "loss": 5.6048,
-      "step": 30934
-    },
-    {
-      "epoch": 16.13298565840939,
-      "grad_norm": 1.64900803565979,
-      "learning_rate": 6.942211055276382e-05,
-      "loss": 5.2609,
-      "step": 30935
-    },
-    {
-      "epoch": 16.133507170795305,
-      "grad_norm": 1.4914608001708984,
-      "learning_rate": 6.94211055276382e-05,
-      "loss": 5.6914,
-      "step": 30936
-    },
-    {
-      "epoch": 16.134028683181224,
-      "grad_norm": 1.452596664428711,
-      "learning_rate": 6.942010050251256e-05,
-      "loss": 5.5591,
-      "step": 30937
-    },
-    {
-      "epoch": 16.134550195567144,
-      "grad_norm": 1.421158790588379,
-      "learning_rate": 6.941909547738694e-05,
-      "loss": 5.015,
-      "step": 30938
-    },
-    {
-      "epoch": 16.135071707953063,
-      "grad_norm": 1.443095088005066,
-      "learning_rate": 6.941809045226131e-05,
-      "loss": 5.7045,
-      "step": 30939
-    },
-    {
-      "epoch": 16.135593220338983,
-      "grad_norm": 1.5500558614730835,
-      "learning_rate": 6.941708542713569e-05,
-      "loss": 5.115,
-      "step": 30940
-    },
-    {
-      "epoch": 16.136114732724902,
-      "grad_norm": 1.448595643043518,
-      "learning_rate": 6.941608040201006e-05,
-      "loss": 5.5953,
-      "step": 30941
-    },
-    {
-      "epoch": 16.13663624511082,
-      "grad_norm": 1.432407021522522,
-      "learning_rate": 6.941507537688443e-05,
-      "loss": 5.5767,
-      "step": 30942
-    },
-    {
-      "epoch": 16.13715775749674,
-      "grad_norm": 1.5897576808929443,
-      "learning_rate": 6.94140703517588e-05,
-      "loss": 5.4488,
-      "step": 30943
-    },
-    {
-      "epoch": 16.13767926988266,
-      "grad_norm": 1.4501134157180786,
-      "learning_rate": 6.941306532663316e-05,
-      "loss": 5.1651,
-      "step": 30944
-    },
-    {
-      "epoch": 16.13820078226858,
-      "grad_norm": 1.7125779390335083,
-      "learning_rate": 6.941206030150754e-05,
-      "loss": 5.2522,
-      "step": 30945
-    },
-    {
-      "epoch": 16.1387222946545,
-      "grad_norm": 1.4800430536270142,
-      "learning_rate": 6.94110552763819e-05,
-      "loss": 5.4693,
-      "step": 30946
-    },
-    {
-      "epoch": 16.13924380704042,
-      "grad_norm": 1.4681727886199951,
-      "learning_rate": 6.941005025125628e-05,
-      "loss": 5.2779,
-      "step": 30947
-    },
-    {
-      "epoch": 16.139765319426335,
-      "grad_norm": 1.5413490533828735,
-      "learning_rate": 6.940904522613066e-05,
-      "loss": 5.1265,
-      "step": 30948
-    },
-    {
-      "epoch": 16.140286831812254,
-      "grad_norm": 1.4120099544525146,
-      "learning_rate": 6.940804020100504e-05,
-      "loss": 5.5996,
-      "step": 30949
-    },
-    {
-      "epoch": 16.140808344198174,
-      "grad_norm": 1.40192449092865,
-      "learning_rate": 6.94070351758794e-05,
-      "loss": 5.6911,
-      "step": 30950
-    },
-    {
-      "epoch": 16.141329856584093,
-      "grad_norm": 1.3953180313110352,
-      "learning_rate": 6.940603015075378e-05,
-      "loss": 4.9823,
-      "step": 30951
-    },
-    {
-      "epoch": 16.141851368970013,
-      "grad_norm": 1.4177546501159668,
-      "learning_rate": 6.940502512562814e-05,
-      "loss": 5.0636,
-      "step": 30952
-    },
-    {
-      "epoch": 16.142372881355932,
-      "grad_norm": 1.5062530040740967,
-      "learning_rate": 6.940402010050252e-05,
-      "loss": 5.4872,
-      "step": 30953
-    },
-    {
-      "epoch": 16.14289439374185,
-      "grad_norm": 1.530922293663025,
-      "learning_rate": 6.940301507537689e-05,
-      "loss": 5.2536,
-      "step": 30954
-    },
-    {
-      "epoch": 16.14341590612777,
-      "grad_norm": 1.6139620542526245,
-      "learning_rate": 6.940201005025126e-05,
-      "loss": 5.3911,
-      "step": 30955
-    },
-    {
-      "epoch": 16.14393741851369,
-      "grad_norm": 1.5162869691848755,
-      "learning_rate": 6.940100502512563e-05,
-      "loss": 5.3082,
-      "step": 30956
-    },
-    {
-      "epoch": 16.14445893089961,
-      "grad_norm": 1.5649179220199585,
-      "learning_rate": 6.939999999999999e-05,
-      "loss": 5.1973,
-      "step": 30957
-    },
-    {
-      "epoch": 16.14498044328553,
-      "grad_norm": 1.6975173950195312,
-      "learning_rate": 6.939899497487437e-05,
-      "loss": 4.9799,
-      "step": 30958
-    },
-    {
-      "epoch": 16.14550195567145,
-      "grad_norm": 1.4940794706344604,
-      "learning_rate": 6.939798994974875e-05,
-      "loss": 5.6167,
-      "step": 30959
-    },
-    {
-      "epoch": 16.146023468057365,
-      "grad_norm": 1.5439335107803345,
-      "learning_rate": 6.939698492462313e-05,
-      "loss": 5.3682,
-      "step": 30960
-    },
-    {
-      "epoch": 16.146544980443284,
-      "grad_norm": 1.441157341003418,
-      "learning_rate": 6.939597989949749e-05,
-      "loss": 5.4158,
-      "step": 30961
-    },
-    {
-      "epoch": 16.147066492829204,
-      "grad_norm": 1.4640307426452637,
-      "learning_rate": 6.939497487437187e-05,
-      "loss": 5.3336,
-      "step": 30962
-    },
-    {
-      "epoch": 16.147588005215123,
-      "grad_norm": 1.5664806365966797,
-      "learning_rate": 6.939396984924623e-05,
-      "loss": 5.0662,
-      "step": 30963
-    },
-    {
-      "epoch": 16.148109517601043,
-      "grad_norm": 1.5051360130310059,
-      "learning_rate": 6.939296482412061e-05,
-      "loss": 4.8969,
-      "step": 30964
-    },
-    {
-      "epoch": 16.148631029986962,
-      "grad_norm": 1.6253182888031006,
-      "learning_rate": 6.939195979899497e-05,
-      "loss": 5.1532,
-      "step": 30965
-    },
-    {
-      "epoch": 16.14915254237288,
-      "grad_norm": 1.4935022592544556,
-      "learning_rate": 6.939095477386935e-05,
-      "loss": 5.5081,
-      "step": 30966
-    },
-    {
-      "epoch": 16.1496740547588,
-      "grad_norm": 1.574904203414917,
-      "learning_rate": 6.938994974874372e-05,
-      "loss": 5.1349,
-      "step": 30967
-    },
-    {
-      "epoch": 16.15019556714472,
-      "grad_norm": 1.483721137046814,
-      "learning_rate": 6.93889447236181e-05,
-      "loss": 5.2587,
-      "step": 30968
-    },
-    {
-      "epoch": 16.15071707953064,
-      "grad_norm": 1.5140016078948975,
-      "learning_rate": 6.938793969849246e-05,
-      "loss": 5.2813,
-      "step": 30969
-    },
-    {
-      "epoch": 16.15123859191656,
-      "grad_norm": 1.5099924802780151,
-      "learning_rate": 6.938693467336684e-05,
-      "loss": 5.4512,
-      "step": 30970
-    },
-    {
-      "epoch": 16.151760104302475,
-      "grad_norm": 1.5085556507110596,
-      "learning_rate": 6.938592964824121e-05,
-      "loss": 5.5783,
-      "step": 30971
-    },
-    {
-      "epoch": 16.152281616688395,
-      "grad_norm": 1.4936567544937134,
-      "learning_rate": 6.938492462311558e-05,
-      "loss": 5.2276,
-      "step": 30972
-    },
-    {
-      "epoch": 16.152803129074314,
-      "grad_norm": 1.3854644298553467,
-      "learning_rate": 6.938391959798996e-05,
-      "loss": 5.5905,
-      "step": 30973
-    },
-    {
-      "epoch": 16.153324641460234,
-      "grad_norm": 1.505040168762207,
-      "learning_rate": 6.938291457286432e-05,
-      "loss": 5.3087,
-      "step": 30974
-    },
-    {
-      "epoch": 16.153846153846153,
-      "grad_norm": 1.5396931171417236,
-      "learning_rate": 6.93819095477387e-05,
-      "loss": 5.4977,
-      "step": 30975
-    },
-    {
-      "epoch": 16.154367666232073,
-      "grad_norm": 1.480103611946106,
-      "learning_rate": 6.938090452261306e-05,
-      "loss": 5.3681,
-      "step": 30976
-    },
-    {
-      "epoch": 16.154889178617992,
-      "grad_norm": 1.6618238687515259,
-      "learning_rate": 6.937989949748744e-05,
-      "loss": 5.2108,
-      "step": 30977
-    },
-    {
-      "epoch": 16.15541069100391,
-      "grad_norm": 1.3472121953964233,
-      "learning_rate": 6.93788944723618e-05,
-      "loss": 5.5637,
-      "step": 30978
-    },
-    {
-      "epoch": 16.15593220338983,
-      "grad_norm": 1.523094892501831,
-      "learning_rate": 6.937788944723618e-05,
-      "loss": 5.7395,
-      "step": 30979
-    },
-    {
-      "epoch": 16.15645371577575,
-      "grad_norm": 1.5398280620574951,
-      "learning_rate": 6.937688442211056e-05,
-      "loss": 5.3711,
-      "step": 30980
-    },
-    {
-      "epoch": 16.15697522816167,
-      "grad_norm": 1.453578233718872,
-      "learning_rate": 6.937587939698494e-05,
-      "loss": 5.6975,
-      "step": 30981
-    },
-    {
-      "epoch": 16.15749674054759,
-      "grad_norm": 1.5759199857711792,
-      "learning_rate": 6.93748743718593e-05,
-      "loss": 5.0156,
-      "step": 30982
-    },
-    {
-      "epoch": 16.158018252933505,
-      "grad_norm": 1.4901840686798096,
-      "learning_rate": 6.937386934673367e-05,
-      "loss": 5.0428,
-      "step": 30983
-    },
-    {
-      "epoch": 16.158539765319425,
-      "grad_norm": 1.4848973751068115,
-      "learning_rate": 6.937286432160805e-05,
-      "loss": 5.3937,
-      "step": 30984
-    },
-    {
-      "epoch": 16.159061277705344,
-      "grad_norm": 1.5060938596725464,
-      "learning_rate": 6.937185929648241e-05,
-      "loss": 5.5763,
-      "step": 30985
-    },
-    {
-      "epoch": 16.159582790091264,
-      "grad_norm": 1.4460431337356567,
-      "learning_rate": 6.937085427135679e-05,
-      "loss": 5.5586,
-      "step": 30986
-    },
-    {
-      "epoch": 16.160104302477183,
-      "grad_norm": 1.4057544469833374,
-      "learning_rate": 6.936984924623115e-05,
-      "loss": 3.734,
-      "step": 30987
-    },
-    {
-      "epoch": 16.160625814863103,
-      "grad_norm": 1.4940873384475708,
-      "learning_rate": 6.936884422110553e-05,
-      "loss": 5.6839,
-      "step": 30988
-    },
-    {
-      "epoch": 16.161147327249022,
-      "grad_norm": 1.5541502237319946,
-      "learning_rate": 6.93678391959799e-05,
-      "loss": 4.8537,
-      "step": 30989
-    },
-    {
-      "epoch": 16.16166883963494,
-      "grad_norm": 1.4994029998779297,
-      "learning_rate": 6.936683417085427e-05,
-      "loss": 5.2429,
-      "step": 30990
-    },
-    {
-      "epoch": 16.16219035202086,
-      "grad_norm": 1.3941715955734253,
-      "learning_rate": 6.936582914572865e-05,
-      "loss": 5.7115,
-      "step": 30991
-    },
-    {
-      "epoch": 16.16271186440678,
-      "grad_norm": 1.399877667427063,
-      "learning_rate": 6.936482412060303e-05,
-      "loss": 5.4179,
-      "step": 30992
-    },
-    {
-      "epoch": 16.1632333767927,
-      "grad_norm": 1.536030650138855,
-      "learning_rate": 6.936381909547739e-05,
-      "loss": 5.1837,
-      "step": 30993
-    },
-    {
-      "epoch": 16.16375488917862,
-      "grad_norm": 1.6162053346633911,
-      "learning_rate": 6.936281407035177e-05,
-      "loss": 5.3524,
-      "step": 30994
-    },
-    {
-      "epoch": 16.164276401564535,
-      "grad_norm": 1.6573278903961182,
-      "learning_rate": 6.936180904522613e-05,
-      "loss": 4.9049,
-      "step": 30995
-    },
-    {
-      "epoch": 16.164797913950455,
-      "grad_norm": 1.5498918294906616,
-      "learning_rate": 6.93608040201005e-05,
-      "loss": 5.3106,
-      "step": 30996
-    },
-    {
-      "epoch": 16.165319426336374,
-      "grad_norm": 1.3725453615188599,
-      "learning_rate": 6.935979899497488e-05,
-      "loss": 5.5318,
-      "step": 30997
-    },
-    {
-      "epoch": 16.165840938722294,
-      "grad_norm": 1.5398285388946533,
-      "learning_rate": 6.935879396984924e-05,
-      "loss": 5.0013,
-      "step": 30998
-    },
-    {
-      "epoch": 16.166362451108213,
-      "grad_norm": 1.4664270877838135,
-      "learning_rate": 6.935778894472362e-05,
-      "loss": 5.2751,
-      "step": 30999
-    },
-    {
-      "epoch": 16.166883963494133,
-      "grad_norm": 1.367972731590271,
-      "learning_rate": 6.9356783919598e-05,
-      "loss": 5.5325,
-      "step": 31000
-    },
-    {
-      "epoch": 16.167405475880052,
-      "grad_norm": 1.4945025444030762,
-      "learning_rate": 6.935577889447237e-05,
-      "loss": 5.0126,
-      "step": 31001
-    },
-    {
-      "epoch": 16.16792698826597,
-      "grad_norm": 1.4393197298049927,
-      "learning_rate": 6.935477386934674e-05,
-      "loss": 4.7882,
-      "step": 31002
-    },
-    {
-      "epoch": 16.16844850065189,
-      "grad_norm": 1.5353190898895264,
-      "learning_rate": 6.935376884422112e-05,
-      "loss": 5.1167,
-      "step": 31003
-    },
-    {
-      "epoch": 16.16897001303781,
-      "grad_norm": 1.4792746305465698,
-      "learning_rate": 6.935276381909548e-05,
-      "loss": 5.3125,
-      "step": 31004
-    },
-    {
-      "epoch": 16.16949152542373,
-      "grad_norm": 1.433932900428772,
-      "learning_rate": 6.935175879396986e-05,
-      "loss": 5.4604,
-      "step": 31005
-    },
-    {
-      "epoch": 16.17001303780965,
-      "grad_norm": 1.4247996807098389,
-      "learning_rate": 6.935075376884422e-05,
-      "loss": 5.5028,
-      "step": 31006
-    },
-    {
-      "epoch": 16.170534550195566,
-      "grad_norm": 1.4772855043411255,
-      "learning_rate": 6.93497487437186e-05,
-      "loss": 4.9803,
-      "step": 31007
-    },
-    {
-      "epoch": 16.171056062581485,
-      "grad_norm": 1.540046215057373,
-      "learning_rate": 6.934874371859296e-05,
-      "loss": 5.3829,
-      "step": 31008
-    },
-    {
-      "epoch": 16.171577574967404,
-      "grad_norm": 1.4577100276947021,
-      "learning_rate": 6.934773869346733e-05,
-      "loss": 5.2757,
-      "step": 31009
-    },
-    {
-      "epoch": 16.172099087353324,
-      "grad_norm": 1.4610847234725952,
-      "learning_rate": 6.93467336683417e-05,
-      "loss": 5.3101,
-      "step": 31010
-    },
-    {
-      "epoch": 16.172620599739243,
-      "grad_norm": 1.5822687149047852,
-      "learning_rate": 6.934572864321608e-05,
-      "loss": 5.3218,
-      "step": 31011
-    },
-    {
-      "epoch": 16.173142112125163,
-      "grad_norm": 1.471044659614563,
-      "learning_rate": 6.934472361809046e-05,
-      "loss": 5.3674,
-      "step": 31012
-    },
-    {
-      "epoch": 16.173663624511082,
-      "grad_norm": 1.5042380094528198,
-      "learning_rate": 6.934371859296483e-05,
-      "loss": 5.4827,
-      "step": 31013
-    },
-    {
-      "epoch": 16.174185136897,
-      "grad_norm": 1.4603718519210815,
-      "learning_rate": 6.93427135678392e-05,
-      "loss": 5.6867,
-      "step": 31014
-    },
-    {
-      "epoch": 16.17470664928292,
-      "grad_norm": 1.460034966468811,
-      "learning_rate": 6.934170854271357e-05,
-      "loss": 5.2185,
-      "step": 31015
-    },
-    {
-      "epoch": 16.17522816166884,
-      "grad_norm": 1.402159333229065,
-      "learning_rate": 6.934070351758795e-05,
-      "loss": 5.4275,
-      "step": 31016
-    },
-    {
-      "epoch": 16.17574967405476,
-      "grad_norm": 1.4719126224517822,
-      "learning_rate": 6.933969849246231e-05,
-      "loss": 5.3275,
-      "step": 31017
-    },
-    {
-      "epoch": 16.17627118644068,
-      "grad_norm": 1.4382659196853638,
-      "learning_rate": 6.933869346733669e-05,
-      "loss": 5.3734,
-      "step": 31018
-    },
-    {
-      "epoch": 16.176792698826596,
-      "grad_norm": 1.4584487676620483,
-      "learning_rate": 6.933768844221105e-05,
-      "loss": 5.6709,
-      "step": 31019
-    },
-    {
-      "epoch": 16.177314211212515,
-      "grad_norm": 1.5017905235290527,
-      "learning_rate": 6.933668341708543e-05,
-      "loss": 5.1959,
-      "step": 31020
-    },
-    {
-      "epoch": 16.177835723598434,
-      "grad_norm": 1.4500471353530884,
-      "learning_rate": 6.933567839195981e-05,
-      "loss": 5.3786,
-      "step": 31021
-    },
-    {
-      "epoch": 16.178357235984354,
-      "grad_norm": 1.5153350830078125,
-      "learning_rate": 6.933467336683419e-05,
-      "loss": 5.4702,
-      "step": 31022
-    },
-    {
-      "epoch": 16.178878748370273,
-      "grad_norm": 1.4129160642623901,
-      "learning_rate": 6.933366834170855e-05,
-      "loss": 4.8131,
-      "step": 31023
-    },
-    {
-      "epoch": 16.179400260756193,
-      "grad_norm": 1.5010240077972412,
-      "learning_rate": 6.933266331658291e-05,
-      "loss": 5.4957,
-      "step": 31024
-    },
-    {
-      "epoch": 16.179921773142112,
-      "grad_norm": 1.4935449361801147,
-      "learning_rate": 6.933165829145729e-05,
-      "loss": 4.6558,
-      "step": 31025
-    },
-    {
-      "epoch": 16.180443285528032,
-      "grad_norm": 1.496087908744812,
-      "learning_rate": 6.933065326633166e-05,
-      "loss": 5.576,
-      "step": 31026
-    },
-    {
-      "epoch": 16.18096479791395,
-      "grad_norm": 1.5449767112731934,
-      "learning_rate": 6.932964824120603e-05,
-      "loss": 5.3195,
-      "step": 31027
-    },
-    {
-      "epoch": 16.18148631029987,
-      "grad_norm": 1.4260962009429932,
-      "learning_rate": 6.93286432160804e-05,
-      "loss": 5.4215,
-      "step": 31028
-    },
-    {
-      "epoch": 16.18200782268579,
-      "grad_norm": 1.4913897514343262,
-      "learning_rate": 6.932763819095478e-05,
-      "loss": 5.3332,
-      "step": 31029
-    },
-    {
-      "epoch": 16.18252933507171,
-      "grad_norm": 1.4770723581314087,
-      "learning_rate": 6.932663316582914e-05,
-      "loss": 5.4414,
-      "step": 31030
-    },
-    {
-      "epoch": 16.183050847457626,
-      "grad_norm": 1.488770604133606,
-      "learning_rate": 6.932562814070352e-05,
-      "loss": 5.3183,
-      "step": 31031
-    },
-    {
-      "epoch": 16.183572359843545,
-      "grad_norm": 1.514716625213623,
-      "learning_rate": 6.93246231155779e-05,
-      "loss": 5.1597,
-      "step": 31032
-    },
-    {
-      "epoch": 16.184093872229464,
-      "grad_norm": 1.4983950853347778,
-      "learning_rate": 6.932361809045227e-05,
-      "loss": 5.7753,
-      "step": 31033
-    },
-    {
-      "epoch": 16.184615384615384,
-      "grad_norm": 1.462860345840454,
-      "learning_rate": 6.932261306532664e-05,
-      "loss": 5.6212,
-      "step": 31034
-    },
-    {
-      "epoch": 16.185136897001303,
-      "grad_norm": 1.3769932985305786,
-      "learning_rate": 6.932160804020102e-05,
-      "loss": 5.8616,
-      "step": 31035
-    },
-    {
-      "epoch": 16.185658409387223,
-      "grad_norm": 1.4208359718322754,
-      "learning_rate": 6.932060301507538e-05,
-      "loss": 5.3965,
-      "step": 31036
-    },
-    {
-      "epoch": 16.186179921773142,
-      "grad_norm": 1.3949577808380127,
-      "learning_rate": 6.931959798994974e-05,
-      "loss": 5.3119,
-      "step": 31037
-    },
-    {
-      "epoch": 16.186701434159062,
-      "grad_norm": 1.4958423376083374,
-      "learning_rate": 6.931859296482412e-05,
-      "loss": 5.2557,
-      "step": 31038
-    },
-    {
-      "epoch": 16.18722294654498,
-      "grad_norm": 1.4033560752868652,
-      "learning_rate": 6.931758793969849e-05,
-      "loss": 5.3622,
-      "step": 31039
-    },
-    {
-      "epoch": 16.1877444589309,
-      "grad_norm": 2.0020644664764404,
-      "learning_rate": 6.931658291457286e-05,
-      "loss": 5.1232,
-      "step": 31040
-    },
-    {
-      "epoch": 16.18826597131682,
-      "grad_norm": 1.3849378824234009,
-      "learning_rate": 6.931557788944724e-05,
-      "loss": 5.2298,
-      "step": 31041
-    },
-    {
-      "epoch": 16.18878748370274,
-      "grad_norm": 1.433516025543213,
-      "learning_rate": 6.931457286432162e-05,
-      "loss": 5.7402,
-      "step": 31042
-    },
-    {
-      "epoch": 16.189308996088656,
-      "grad_norm": 1.4487532377243042,
-      "learning_rate": 6.931356783919598e-05,
-      "loss": 5.4034,
-      "step": 31043
-    },
-    {
-      "epoch": 16.189830508474575,
-      "grad_norm": 1.5431175231933594,
-      "learning_rate": 6.931256281407036e-05,
-      "loss": 4.9047,
-      "step": 31044
-    },
-    {
-      "epoch": 16.190352020860495,
-      "grad_norm": 1.491159200668335,
-      "learning_rate": 6.931155778894473e-05,
-      "loss": 5.4132,
-      "step": 31045
-    },
-    {
-      "epoch": 16.190873533246414,
-      "grad_norm": 1.4863322973251343,
-      "learning_rate": 6.93105527638191e-05,
-      "loss": 5.0194,
-      "step": 31046
-    },
-    {
-      "epoch": 16.191395045632333,
-      "grad_norm": 1.4713115692138672,
-      "learning_rate": 6.930954773869347e-05,
-      "loss": 5.3142,
-      "step": 31047
-    },
-    {
-      "epoch": 16.191916558018253,
-      "grad_norm": 1.509300947189331,
-      "learning_rate": 6.930854271356785e-05,
-      "loss": 5.1429,
-      "step": 31048
-    },
-    {
-      "epoch": 16.192438070404172,
-      "grad_norm": 1.3954026699066162,
-      "learning_rate": 6.930753768844221e-05,
-      "loss": 5.4507,
-      "step": 31049
-    },
-    {
-      "epoch": 16.192959582790092,
-      "grad_norm": 1.4906244277954102,
-      "learning_rate": 6.930653266331658e-05,
-      "loss": 5.5347,
-      "step": 31050
-    },
-    {
-      "epoch": 16.19348109517601,
-      "grad_norm": 1.4364922046661377,
-      "learning_rate": 6.930552763819095e-05,
-      "loss": 5.4764,
-      "step": 31051
-    },
-    {
-      "epoch": 16.19400260756193,
-      "grad_norm": 1.473679542541504,
-      "learning_rate": 6.930452261306533e-05,
-      "loss": 4.6905,
-      "step": 31052
-    },
-    {
-      "epoch": 16.19452411994785,
-      "grad_norm": 1.4882627725601196,
-      "learning_rate": 6.930351758793971e-05,
-      "loss": 5.1453,
-      "step": 31053
-    },
-    {
-      "epoch": 16.195045632333766,
-      "grad_norm": 1.4740679264068604,
-      "learning_rate": 6.930251256281407e-05,
-      "loss": 5.5045,
-      "step": 31054
-    },
-    {
-      "epoch": 16.195567144719686,
-      "grad_norm": 1.5473634004592896,
-      "learning_rate": 6.930150753768845e-05,
-      "loss": 5.2574,
-      "step": 31055
-    },
-    {
-      "epoch": 16.196088657105605,
-      "grad_norm": 1.6102575063705444,
-      "learning_rate": 6.930050251256282e-05,
-      "loss": 5.1624,
-      "step": 31056
-    },
-    {
-      "epoch": 16.196610169491525,
-      "grad_norm": 1.9668787717819214,
-      "learning_rate": 6.929949748743719e-05,
-      "loss": 4.8126,
-      "step": 31057
-    },
-    {
-      "epoch": 16.197131681877444,
-      "grad_norm": 1.5714538097381592,
-      "learning_rate": 6.929849246231156e-05,
-      "loss": 5.4216,
-      "step": 31058
-    },
-    {
-      "epoch": 16.197653194263363,
-      "grad_norm": 1.511050820350647,
-      "learning_rate": 6.929748743718593e-05,
-      "loss": 5.441,
-      "step": 31059
-    },
-    {
-      "epoch": 16.198174706649283,
-      "grad_norm": 1.48995840549469,
-      "learning_rate": 6.92964824120603e-05,
-      "loss": 5.227,
-      "step": 31060
-    },
-    {
-      "epoch": 16.198696219035202,
-      "grad_norm": 1.4933931827545166,
-      "learning_rate": 6.929547738693468e-05,
-      "loss": 5.6681,
-      "step": 31061
-    },
-    {
-      "epoch": 16.199217731421122,
-      "grad_norm": 1.4405485391616821,
-      "learning_rate": 6.929447236180905e-05,
-      "loss": 5.2446,
-      "step": 31062
-    },
-    {
-      "epoch": 16.19973924380704,
-      "grad_norm": 1.641489028930664,
-      "learning_rate": 6.929346733668342e-05,
-      "loss": 4.6417,
-      "step": 31063
-    },
-    {
-      "epoch": 16.20026075619296,
-      "grad_norm": 1.6750874519348145,
-      "learning_rate": 6.92924623115578e-05,
-      "loss": 5.2454,
-      "step": 31064
-    },
-    {
-      "epoch": 16.20078226857888,
-      "grad_norm": 1.3739182949066162,
-      "learning_rate": 6.929145728643216e-05,
-      "loss": 4.9979,
-      "step": 31065
-    },
-    {
-      "epoch": 16.201303780964796,
-      "grad_norm": 1.5971739292144775,
-      "learning_rate": 6.929045226130654e-05,
-      "loss": 5.3022,
-      "step": 31066
-    },
-    {
-      "epoch": 16.201825293350716,
-      "grad_norm": 1.446584939956665,
-      "learning_rate": 6.92894472361809e-05,
-      "loss": 5.4874,
-      "step": 31067
-    },
-    {
-      "epoch": 16.202346805736635,
-      "grad_norm": 1.5779374837875366,
-      "learning_rate": 6.928844221105528e-05,
-      "loss": 5.0957,
-      "step": 31068
-    },
-    {
-      "epoch": 16.202868318122555,
-      "grad_norm": 1.5435078144073486,
-      "learning_rate": 6.928743718592965e-05,
-      "loss": 5.0175,
-      "step": 31069
-    },
-    {
-      "epoch": 16.203389830508474,
-      "grad_norm": 1.612547516822815,
-      "learning_rate": 6.928643216080402e-05,
-      "loss": 4.5034,
-      "step": 31070
-    },
-    {
-      "epoch": 16.203911342894393,
-      "grad_norm": 1.4271867275238037,
-      "learning_rate": 6.928542713567839e-05,
-      "loss": 4.8651,
-      "step": 31071
-    },
-    {
-      "epoch": 16.204432855280313,
-      "grad_norm": 1.4698894023895264,
-      "learning_rate": 6.928442211055277e-05,
-      "loss": 5.0419,
-      "step": 31072
-    },
-    {
-      "epoch": 16.204954367666232,
-      "grad_norm": 1.4936611652374268,
-      "learning_rate": 6.928341708542714e-05,
-      "loss": 5.3123,
-      "step": 31073
-    },
-    {
-      "epoch": 16.205475880052152,
-      "grad_norm": 1.6179996728897095,
-      "learning_rate": 6.928241206030152e-05,
-      "loss": 4.8703,
-      "step": 31074
-    },
-    {
-      "epoch": 16.20599739243807,
-      "grad_norm": 1.5059226751327515,
-      "learning_rate": 6.928140703517589e-05,
-      "loss": 5.3944,
-      "step": 31075
-    },
-    {
-      "epoch": 16.20651890482399,
-      "grad_norm": 1.4225107431411743,
-      "learning_rate": 6.928040201005025e-05,
-      "loss": 5.5073,
-      "step": 31076
-    },
-    {
-      "epoch": 16.20704041720991,
-      "grad_norm": 1.5877107381820679,
-      "learning_rate": 6.927939698492463e-05,
-      "loss": 5.0435,
-      "step": 31077
-    },
-    {
-      "epoch": 16.207561929595826,
-      "grad_norm": 1.5257872343063354,
-      "learning_rate": 6.927839195979899e-05,
-      "loss": 5.5803,
-      "step": 31078
-    },
-    {
-      "epoch": 16.208083441981746,
-      "grad_norm": 1.42218816280365,
-      "learning_rate": 6.927738693467337e-05,
-      "loss": 5.3323,
-      "step": 31079
-    },
-    {
-      "epoch": 16.208604954367665,
-      "grad_norm": 1.4694528579711914,
-      "learning_rate": 6.927638190954773e-05,
-      "loss": 5.6669,
-      "step": 31080
-    },
-    {
-      "epoch": 16.209126466753585,
-      "grad_norm": 1.4443883895874023,
-      "learning_rate": 6.927537688442211e-05,
-      "loss": 5.1932,
-      "step": 31081
-    },
-    {
-      "epoch": 16.209647979139504,
-      "grad_norm": 1.548561692237854,
-      "learning_rate": 6.927437185929649e-05,
-      "loss": 5.4565,
-      "step": 31082
-    },
-    {
-      "epoch": 16.210169491525424,
-      "grad_norm": 1.537735104560852,
-      "learning_rate": 6.927336683417087e-05,
-      "loss": 5.1837,
-      "step": 31083
-    },
-    {
-      "epoch": 16.210691003911343,
-      "grad_norm": 1.4334149360656738,
-      "learning_rate": 6.927236180904523e-05,
-      "loss": 5.7532,
-      "step": 31084
-    },
-    {
-      "epoch": 16.211212516297262,
-      "grad_norm": 1.4960253238677979,
-      "learning_rate": 6.927135678391961e-05,
-      "loss": 4.8881,
-      "step": 31085
-    },
-    {
-      "epoch": 16.211734028683182,
-      "grad_norm": 1.4487791061401367,
-      "learning_rate": 6.927035175879397e-05,
-      "loss": 4.8722,
-      "step": 31086
-    },
-    {
-      "epoch": 16.2122555410691,
-      "grad_norm": 1.480347752571106,
-      "learning_rate": 6.926934673366835e-05,
-      "loss": 5.6693,
-      "step": 31087
-    },
-    {
-      "epoch": 16.21277705345502,
-      "grad_norm": 1.6001046895980835,
-      "learning_rate": 6.926834170854272e-05,
-      "loss": 5.1815,
-      "step": 31088
-    },
-    {
-      "epoch": 16.21329856584094,
-      "grad_norm": 1.4910993576049805,
-      "learning_rate": 6.926733668341708e-05,
-      "loss": 5.2773,
-      "step": 31089
-    },
-    {
-      "epoch": 16.213820078226856,
-      "grad_norm": 1.5827549695968628,
-      "learning_rate": 6.926633165829146e-05,
-      "loss": 5.1379,
-      "step": 31090
-    },
-    {
-      "epoch": 16.214341590612776,
-      "grad_norm": 1.4520812034606934,
-      "learning_rate": 6.926532663316582e-05,
-      "loss": 5.7729,
-      "step": 31091
-    },
-    {
-      "epoch": 16.214863102998695,
-      "grad_norm": 1.3357058763504028,
-      "learning_rate": 6.92643216080402e-05,
-      "loss": 5.7751,
-      "step": 31092
-    },
-    {
-      "epoch": 16.215384615384615,
-      "grad_norm": 1.538414716720581,
-      "learning_rate": 6.926331658291458e-05,
-      "loss": 5.0563,
-      "step": 31093
-    },
-    {
-      "epoch": 16.215906127770534,
-      "grad_norm": 1.4891408681869507,
-      "learning_rate": 6.926231155778896e-05,
-      "loss": 5.1795,
-      "step": 31094
-    },
-    {
-      "epoch": 16.216427640156454,
-      "grad_norm": 1.5099854469299316,
-      "learning_rate": 6.926130653266332e-05,
-      "loss": 5.4529,
-      "step": 31095
-    },
-    {
-      "epoch": 16.216949152542373,
-      "grad_norm": 1.4191516637802124,
-      "learning_rate": 6.92603015075377e-05,
-      "loss": 5.4716,
-      "step": 31096
-    },
-    {
-      "epoch": 16.217470664928292,
-      "grad_norm": 1.5285416841506958,
-      "learning_rate": 6.925929648241206e-05,
-      "loss": 4.7333,
-      "step": 31097
-    },
-    {
-      "epoch": 16.217992177314212,
-      "grad_norm": 1.4937270879745483,
-      "learning_rate": 6.925829145728644e-05,
-      "loss": 5.3786,
-      "step": 31098
-    },
-    {
-      "epoch": 16.21851368970013,
-      "grad_norm": 1.4908466339111328,
-      "learning_rate": 6.92572864321608e-05,
-      "loss": 5.1821,
-      "step": 31099
-    },
-    {
-      "epoch": 16.21903520208605,
-      "grad_norm": 1.620050311088562,
-      "learning_rate": 6.925628140703518e-05,
-      "loss": 5.4333,
-      "step": 31100
-    },
-    {
-      "epoch": 16.21955671447197,
-      "grad_norm": 1.4987469911575317,
-      "learning_rate": 6.925527638190955e-05,
-      "loss": 5.6617,
-      "step": 31101
-    },
-    {
-      "epoch": 16.220078226857886,
-      "grad_norm": 1.502287745475769,
-      "learning_rate": 6.925427135678392e-05,
-      "loss": 5.3741,
-      "step": 31102
-    },
-    {
-      "epoch": 16.220599739243806,
-      "grad_norm": 1.5648324489593506,
-      "learning_rate": 6.92532663316583e-05,
-      "loss": 5.4678,
-      "step": 31103
-    },
-    {
-      "epoch": 16.221121251629725,
-      "grad_norm": 1.4424359798431396,
-      "learning_rate": 6.925226130653267e-05,
-      "loss": 5.4778,
-      "step": 31104
-    },
-    {
-      "epoch": 16.221642764015645,
-      "grad_norm": 1.491478681564331,
-      "learning_rate": 6.925125628140704e-05,
-      "loss": 5.4773,
-      "step": 31105
-    },
-    {
-      "epoch": 16.222164276401564,
-      "grad_norm": 1.5185734033584595,
-      "learning_rate": 6.925025125628141e-05,
-      "loss": 5.2158,
-      "step": 31106
-    },
-    {
-      "epoch": 16.222685788787484,
-      "grad_norm": 1.5501961708068848,
-      "learning_rate": 6.924924623115579e-05,
-      "loss": 5.4024,
-      "step": 31107
-    },
-    {
-      "epoch": 16.223207301173403,
-      "grad_norm": 1.419853687286377,
-      "learning_rate": 6.924824120603015e-05,
-      "loss": 5.4851,
-      "step": 31108
-    },
-    {
-      "epoch": 16.223728813559323,
-      "grad_norm": 1.6301162242889404,
-      "learning_rate": 6.924723618090453e-05,
-      "loss": 5.3487,
-      "step": 31109
-    },
-    {
-      "epoch": 16.224250325945242,
-      "grad_norm": 1.5110739469528198,
-      "learning_rate": 6.924623115577889e-05,
-      "loss": 5.2956,
-      "step": 31110
-    },
-    {
-      "epoch": 16.22477183833116,
-      "grad_norm": 1.4063645601272583,
-      "learning_rate": 6.924522613065327e-05,
-      "loss": 5.5912,
-      "step": 31111
-    },
-    {
-      "epoch": 16.22529335071708,
-      "grad_norm": 1.5101561546325684,
-      "learning_rate": 6.924422110552763e-05,
-      "loss": 5.6429,
-      "step": 31112
-    },
-    {
-      "epoch": 16.225814863103,
-      "grad_norm": 1.5045820474624634,
-      "learning_rate": 6.924321608040201e-05,
-      "loss": 5.2703,
-      "step": 31113
-    },
-    {
-      "epoch": 16.226336375488916,
-      "grad_norm": 1.6005206108093262,
-      "learning_rate": 6.924221105527639e-05,
-      "loss": 5.023,
-      "step": 31114
-    },
-    {
-      "epoch": 16.226857887874836,
-      "grad_norm": 1.6010268926620483,
-      "learning_rate": 6.924120603015077e-05,
-      "loss": 5.0502,
-      "step": 31115
-    },
-    {
-      "epoch": 16.227379400260755,
-      "grad_norm": 1.5830497741699219,
-      "learning_rate": 6.924020100502513e-05,
-      "loss": 5.4597,
-      "step": 31116
-    },
-    {
-      "epoch": 16.227900912646675,
-      "grad_norm": 1.533555030822754,
-      "learning_rate": 6.92391959798995e-05,
-      "loss": 5.199,
-      "step": 31117
-    },
-    {
-      "epoch": 16.228422425032594,
-      "grad_norm": 1.4514802694320679,
-      "learning_rate": 6.923819095477387e-05,
-      "loss": 5.7166,
-      "step": 31118
-    },
-    {
-      "epoch": 16.228943937418514,
-      "grad_norm": 1.459532380104065,
-      "learning_rate": 6.923718592964824e-05,
-      "loss": 4.8776,
-      "step": 31119
-    },
-    {
-      "epoch": 16.229465449804433,
-      "grad_norm": 1.4712469577789307,
-      "learning_rate": 6.923618090452262e-05,
-      "loss": 5.151,
-      "step": 31120
-    },
-    {
-      "epoch": 16.229986962190353,
-      "grad_norm": 1.4706616401672363,
-      "learning_rate": 6.923517587939698e-05,
-      "loss": 5.6858,
-      "step": 31121
-    },
-    {
-      "epoch": 16.230508474576272,
-      "grad_norm": 1.4809606075286865,
-      "learning_rate": 6.923417085427136e-05,
-      "loss": 5.3276,
-      "step": 31122
-    },
-    {
-      "epoch": 16.23102998696219,
-      "grad_norm": 1.5044983625411987,
-      "learning_rate": 6.923316582914572e-05,
-      "loss": 5.1247,
-      "step": 31123
-    },
-    {
-      "epoch": 16.23155149934811,
-      "grad_norm": 1.4824814796447754,
-      "learning_rate": 6.92321608040201e-05,
-      "loss": 5.5281,
-      "step": 31124
-    },
-    {
-      "epoch": 16.23207301173403,
-      "grad_norm": 1.3190925121307373,
-      "learning_rate": 6.923115577889448e-05,
-      "loss": 5.4757,
-      "step": 31125
-    },
-    {
-      "epoch": 16.232594524119946,
-      "grad_norm": 1.4410829544067383,
-      "learning_rate": 6.923015075376886e-05,
-      "loss": 5.6442,
-      "step": 31126
-    },
-    {
-      "epoch": 16.233116036505866,
-      "grad_norm": 1.530422568321228,
-      "learning_rate": 6.922914572864322e-05,
-      "loss": 5.4087,
-      "step": 31127
-    },
-    {
-      "epoch": 16.233637548891785,
-      "grad_norm": 1.4779512882232666,
-      "learning_rate": 6.92281407035176e-05,
-      "loss": 5.4477,
-      "step": 31128
-    },
-    {
-      "epoch": 16.234159061277705,
-      "grad_norm": 1.5835564136505127,
-      "learning_rate": 6.922713567839196e-05,
-      "loss": 4.9246,
-      "step": 31129
-    },
-    {
-      "epoch": 16.234680573663624,
-      "grad_norm": 1.5404410362243652,
-      "learning_rate": 6.922613065326633e-05,
-      "loss": 5.1954,
-      "step": 31130
-    },
-    {
-      "epoch": 16.235202086049544,
-      "grad_norm": 1.4374843835830688,
-      "learning_rate": 6.92251256281407e-05,
-      "loss": 5.0901,
-      "step": 31131
-    },
-    {
-      "epoch": 16.235723598435463,
-      "grad_norm": 1.4760843515396118,
-      "learning_rate": 6.922412060301507e-05,
-      "loss": 5.3988,
-      "step": 31132
-    },
-    {
-      "epoch": 16.236245110821383,
-      "grad_norm": 1.459097146987915,
-      "learning_rate": 6.922311557788945e-05,
-      "loss": 5.4148,
-      "step": 31133
-    },
-    {
-      "epoch": 16.236766623207302,
-      "grad_norm": 1.4868260622024536,
-      "learning_rate": 6.922211055276382e-05,
-      "loss": 5.31,
-      "step": 31134
-    },
-    {
-      "epoch": 16.23728813559322,
-      "grad_norm": 1.4610999822616577,
-      "learning_rate": 6.92211055276382e-05,
-      "loss": 5.7052,
-      "step": 31135
-    },
-    {
-      "epoch": 16.23780964797914,
-      "grad_norm": 1.4065773487091064,
-      "learning_rate": 6.922010050251257e-05,
-      "loss": 5.4779,
-      "step": 31136
-    },
-    {
-      "epoch": 16.23833116036506,
-      "grad_norm": 1.6316365003585815,
-      "learning_rate": 6.921909547738694e-05,
-      "loss": 4.9049,
-      "step": 31137
-    },
-    {
-      "epoch": 16.238852672750976,
-      "grad_norm": 1.4870314598083496,
-      "learning_rate": 6.921809045226131e-05,
-      "loss": 5.5419,
-      "step": 31138
-    },
-    {
-      "epoch": 16.239374185136896,
-      "grad_norm": 1.4268778562545776,
-      "learning_rate": 6.921708542713569e-05,
-      "loss": 5.5851,
-      "step": 31139
-    },
-    {
-      "epoch": 16.239895697522815,
-      "grad_norm": 1.4559993743896484,
-      "learning_rate": 6.921608040201005e-05,
-      "loss": 5.1562,
-      "step": 31140
-    },
-    {
-      "epoch": 16.240417209908735,
-      "grad_norm": 1.4792026281356812,
-      "learning_rate": 6.921507537688443e-05,
-      "loss": 5.6098,
-      "step": 31141
-    },
-    {
-      "epoch": 16.240938722294654,
-      "grad_norm": 1.4512722492218018,
-      "learning_rate": 6.92140703517588e-05,
-      "loss": 5.7943,
-      "step": 31142
-    },
-    {
-      "epoch": 16.241460234680574,
-      "grad_norm": 1.686238169670105,
-      "learning_rate": 6.921306532663316e-05,
-      "loss": 4.812,
-      "step": 31143
-    },
-    {
-      "epoch": 16.241981747066493,
-      "grad_norm": 1.422866702079773,
-      "learning_rate": 6.921206030150754e-05,
-      "loss": 5.1444,
-      "step": 31144
-    },
-    {
-      "epoch": 16.242503259452413,
-      "grad_norm": 1.5243974924087524,
-      "learning_rate": 6.921105527638191e-05,
-      "loss": 5.1518,
-      "step": 31145
-    },
-    {
-      "epoch": 16.243024771838332,
-      "grad_norm": 1.578454613685608,
-      "learning_rate": 6.921005025125629e-05,
-      "loss": 5.1257,
-      "step": 31146
-    },
-    {
-      "epoch": 16.24354628422425,
-      "grad_norm": 1.5294760465621948,
-      "learning_rate": 6.920904522613066e-05,
-      "loss": 5.5434,
-      "step": 31147
-    },
-    {
-      "epoch": 16.24406779661017,
-      "grad_norm": 1.4657998085021973,
-      "learning_rate": 6.920804020100503e-05,
-      "loss": 5.3318,
-      "step": 31148
-    },
-    {
-      "epoch": 16.24458930899609,
-      "grad_norm": 1.4548890590667725,
-      "learning_rate": 6.92070351758794e-05,
-      "loss": 5.4532,
-      "step": 31149
-    },
-    {
-      "epoch": 16.245110821382006,
-      "grad_norm": 1.3977382183074951,
-      "learning_rate": 6.920603015075378e-05,
-      "loss": 5.4701,
-      "step": 31150
-    },
-    {
-      "epoch": 16.245632333767926,
-      "grad_norm": 1.5284249782562256,
-      "learning_rate": 6.920502512562814e-05,
-      "loss": 5.183,
-      "step": 31151
-    },
-    {
-      "epoch": 16.246153846153845,
-      "grad_norm": 1.4206089973449707,
-      "learning_rate": 6.920402010050252e-05,
-      "loss": 5.5516,
-      "step": 31152
-    },
-    {
-      "epoch": 16.246675358539765,
-      "grad_norm": 1.4672527313232422,
-      "learning_rate": 6.920301507537688e-05,
-      "loss": 5.457,
-      "step": 31153
-    },
-    {
-      "epoch": 16.247196870925684,
-      "grad_norm": 1.5014852285385132,
-      "learning_rate": 6.920201005025126e-05,
-      "loss": 4.8887,
-      "step": 31154
-    },
-    {
-      "epoch": 16.247718383311604,
-      "grad_norm": 1.5018209218978882,
-      "learning_rate": 6.920100502512564e-05,
-      "loss": 5.0426,
-      "step": 31155
-    },
-    {
-      "epoch": 16.248239895697523,
-      "grad_norm": 1.445389986038208,
-      "learning_rate": 6.92e-05,
-      "loss": 5.4669,
-      "step": 31156
-    },
-    {
-      "epoch": 16.248761408083443,
-      "grad_norm": 1.5328763723373413,
-      "learning_rate": 6.919899497487438e-05,
-      "loss": 5.1019,
-      "step": 31157
-    },
-    {
-      "epoch": 16.249282920469362,
-      "grad_norm": 1.5564296245574951,
-      "learning_rate": 6.919798994974874e-05,
-      "loss": 5.1448,
-      "step": 31158
-    },
-    {
-      "epoch": 16.24980443285528,
-      "grad_norm": 1.5688992738723755,
-      "learning_rate": 6.919698492462312e-05,
-      "loss": 4.8091,
-      "step": 31159
-    },
-    {
-      "epoch": 16.2503259452412,
-      "grad_norm": 1.3605997562408447,
-      "learning_rate": 6.919597989949749e-05,
-      "loss": 5.8587,
-      "step": 31160
-    },
-    {
-      "epoch": 16.250847457627117,
-      "grad_norm": 1.3417707681655884,
-      "learning_rate": 6.919497487437186e-05,
-      "loss": 5.5647,
-      "step": 31161
-    },
-    {
-      "epoch": 16.251368970013036,
-      "grad_norm": 1.493355393409729,
-      "learning_rate": 6.919396984924623e-05,
-      "loss": 5.2846,
-      "step": 31162
-    },
-    {
-      "epoch": 16.251890482398956,
-      "grad_norm": 1.4408022165298462,
-      "learning_rate": 6.91929648241206e-05,
-      "loss": 5.0119,
-      "step": 31163
-    },
-    {
-      "epoch": 16.252411994784875,
-      "grad_norm": 1.4317432641983032,
-      "learning_rate": 6.919195979899497e-05,
-      "loss": 5.2902,
-      "step": 31164
-    },
-    {
-      "epoch": 16.252933507170795,
-      "grad_norm": 1.4374948740005493,
-      "learning_rate": 6.919095477386935e-05,
-      "loss": 5.1085,
-      "step": 31165
-    },
-    {
-      "epoch": 16.253455019556714,
-      "grad_norm": 1.5186349153518677,
-      "learning_rate": 6.918994974874373e-05,
-      "loss": 5.54,
-      "step": 31166
-    },
-    {
-      "epoch": 16.253976531942634,
-      "grad_norm": 1.492283582687378,
-      "learning_rate": 6.91889447236181e-05,
-      "loss": 5.6786,
-      "step": 31167
-    },
-    {
-      "epoch": 16.254498044328553,
-      "grad_norm": 1.4649333953857422,
-      "learning_rate": 6.918793969849247e-05,
-      "loss": 5.3592,
-      "step": 31168
-    },
-    {
-      "epoch": 16.255019556714473,
-      "grad_norm": 1.4844154119491577,
-      "learning_rate": 6.918693467336683e-05,
-      "loss": 5.4314,
-      "step": 31169
-    },
-    {
-      "epoch": 16.255541069100392,
-      "grad_norm": 1.3289729356765747,
-      "learning_rate": 6.918592964824121e-05,
-      "loss": 5.7101,
-      "step": 31170
-    },
-    {
-      "epoch": 16.25606258148631,
-      "grad_norm": 1.4748516082763672,
-      "learning_rate": 6.918492462311557e-05,
-      "loss": 4.9995,
-      "step": 31171
-    },
-    {
-      "epoch": 16.25658409387223,
-      "grad_norm": 1.4208893775939941,
-      "learning_rate": 6.918391959798995e-05,
-      "loss": 5.1643,
-      "step": 31172
-    },
-    {
-      "epoch": 16.257105606258147,
-      "grad_norm": 1.4513126611709595,
-      "learning_rate": 6.918291457286432e-05,
-      "loss": 5.3287,
-      "step": 31173
-    },
-    {
-      "epoch": 16.257627118644066,
-      "grad_norm": 1.51252281665802,
-      "learning_rate": 6.91819095477387e-05,
-      "loss": 5.0338,
-      "step": 31174
-    },
-    {
-      "epoch": 16.258148631029986,
-      "grad_norm": 1.4612311124801636,
-      "learning_rate": 6.918090452261307e-05,
-      "loss": 5.7926,
-      "step": 31175
-    },
-    {
-      "epoch": 16.258670143415905,
-      "grad_norm": 1.555660367012024,
-      "learning_rate": 6.917989949748745e-05,
-      "loss": 4.8385,
-      "step": 31176
-    },
-    {
-      "epoch": 16.259191655801825,
-      "grad_norm": 1.5507440567016602,
-      "learning_rate": 6.917889447236181e-05,
-      "loss": 5.4225,
-      "step": 31177
-    },
-    {
-      "epoch": 16.259713168187744,
-      "grad_norm": 1.5529522895812988,
-      "learning_rate": 6.917788944723619e-05,
-      "loss": 5.5473,
-      "step": 31178
-    },
-    {
-      "epoch": 16.260234680573664,
-      "grad_norm": 1.522326111793518,
-      "learning_rate": 6.917688442211056e-05,
-      "loss": 5.3007,
-      "step": 31179
-    },
-    {
-      "epoch": 16.260756192959583,
-      "grad_norm": 1.5410088300704956,
-      "learning_rate": 6.917587939698493e-05,
-      "loss": 5.436,
-      "step": 31180
-    },
-    {
-      "epoch": 16.261277705345503,
-      "grad_norm": 1.4564471244812012,
-      "learning_rate": 6.91748743718593e-05,
-      "loss": 5.7276,
-      "step": 31181
-    },
-    {
-      "epoch": 16.261799217731422,
-      "grad_norm": 1.4776140451431274,
-      "learning_rate": 6.917386934673366e-05,
-      "loss": 5.5467,
-      "step": 31182
-    },
-    {
-      "epoch": 16.26232073011734,
-      "grad_norm": 1.4679334163665771,
-      "learning_rate": 6.917286432160804e-05,
-      "loss": 5.6026,
-      "step": 31183
-    },
-    {
-      "epoch": 16.26284224250326,
-      "grad_norm": 1.3723526000976562,
-      "learning_rate": 6.91718592964824e-05,
-      "loss": 5.0661,
-      "step": 31184
-    },
-    {
-      "epoch": 16.263363754889177,
-      "grad_norm": 1.4877872467041016,
-      "learning_rate": 6.917085427135678e-05,
-      "loss": 5.3253,
-      "step": 31185
-    },
-    {
-      "epoch": 16.263885267275096,
-      "grad_norm": 1.551662802696228,
-      "learning_rate": 6.916984924623116e-05,
-      "loss": 5.1842,
-      "step": 31186
-    },
-    {
-      "epoch": 16.264406779661016,
-      "grad_norm": 1.4470540285110474,
-      "learning_rate": 6.916884422110554e-05,
-      "loss": 5.4921,
-      "step": 31187
-    },
-    {
-      "epoch": 16.264928292046935,
-      "grad_norm": 1.4354807138442993,
-      "learning_rate": 6.91678391959799e-05,
-      "loss": 5.3626,
-      "step": 31188
-    },
-    {
-      "epoch": 16.265449804432855,
-      "grad_norm": 1.657882571220398,
-      "learning_rate": 6.916683417085428e-05,
-      "loss": 5.0075,
-      "step": 31189
-    },
-    {
-      "epoch": 16.265971316818774,
-      "grad_norm": 1.4575449228286743,
-      "learning_rate": 6.916582914572864e-05,
-      "loss": 5.3677,
-      "step": 31190
-    },
-    {
-      "epoch": 16.266492829204694,
-      "grad_norm": 1.4526653289794922,
-      "learning_rate": 6.916482412060302e-05,
-      "loss": 5.4621,
-      "step": 31191
-    },
-    {
-      "epoch": 16.267014341590613,
-      "grad_norm": 1.443003535270691,
-      "learning_rate": 6.916381909547739e-05,
-      "loss": 5.4653,
-      "step": 31192
-    },
-    {
-      "epoch": 16.267535853976533,
-      "grad_norm": 1.5030765533447266,
-      "learning_rate": 6.916281407035176e-05,
-      "loss": 5.5544,
-      "step": 31193
-    },
-    {
-      "epoch": 16.268057366362452,
-      "grad_norm": 1.452349066734314,
-      "learning_rate": 6.916180904522613e-05,
-      "loss": 5.5443,
-      "step": 31194
-    },
-    {
-      "epoch": 16.26857887874837,
-      "grad_norm": 1.468144178390503,
-      "learning_rate": 6.91608040201005e-05,
-      "loss": 5.5023,
-      "step": 31195
-    },
-    {
-      "epoch": 16.26910039113429,
-      "grad_norm": 1.3157298564910889,
-      "learning_rate": 6.915979899497488e-05,
-      "loss": 5.8094,
-      "step": 31196
-    },
-    {
-      "epoch": 16.269621903520207,
-      "grad_norm": 1.5338844060897827,
-      "learning_rate": 6.915879396984925e-05,
-      "loss": 5.1388,
-      "step": 31197
-    },
-    {
-      "epoch": 16.270143415906126,
-      "grad_norm": 1.568088412284851,
-      "learning_rate": 6.915778894472363e-05,
-      "loss": 5.5713,
-      "step": 31198
-    },
-    {
-      "epoch": 16.270664928292046,
-      "grad_norm": 1.3681912422180176,
-      "learning_rate": 6.915678391959799e-05,
-      "loss": 5.5144,
-      "step": 31199
-    },
-    {
-      "epoch": 16.271186440677965,
-      "grad_norm": 1.4830693006515503,
-      "learning_rate": 6.915577889447237e-05,
-      "loss": 5.324,
-      "step": 31200
-    },
-    {
-      "epoch": 16.271707953063885,
-      "grad_norm": 1.351847767829895,
-      "learning_rate": 6.915477386934673e-05,
-      "loss": 5.706,
-      "step": 31201
-    },
-    {
-      "epoch": 16.272229465449804,
-      "grad_norm": 1.523917555809021,
-      "learning_rate": 6.915376884422111e-05,
-      "loss": 5.4364,
-      "step": 31202
-    },
-    {
-      "epoch": 16.272750977835724,
-      "grad_norm": 1.4081720113754272,
-      "learning_rate": 6.915276381909547e-05,
-      "loss": 5.4004,
-      "step": 31203
-    },
-    {
-      "epoch": 16.273272490221643,
-      "grad_norm": 1.4545812606811523,
-      "learning_rate": 6.915175879396985e-05,
-      "loss": 5.5107,
-      "step": 31204
-    },
-    {
-      "epoch": 16.273794002607563,
-      "grad_norm": 1.5296237468719482,
-      "learning_rate": 6.915075376884422e-05,
-      "loss": 5.1814,
-      "step": 31205
-    },
-    {
-      "epoch": 16.274315514993482,
-      "grad_norm": 1.4336802959442139,
-      "learning_rate": 6.91497487437186e-05,
-      "loss": 5.5067,
-      "step": 31206
-    },
-    {
-      "epoch": 16.2748370273794,
-      "grad_norm": 1.534881353378296,
-      "learning_rate": 6.914874371859297e-05,
-      "loss": 5.2452,
-      "step": 31207
-    },
-    {
-      "epoch": 16.27535853976532,
-      "grad_norm": 1.3093602657318115,
-      "learning_rate": 6.914773869346735e-05,
-      "loss": 4.904,
-      "step": 31208
-    },
-    {
-      "epoch": 16.275880052151237,
-      "grad_norm": 1.5695298910140991,
-      "learning_rate": 6.914673366834171e-05,
-      "loss": 5.1777,
-      "step": 31209
-    },
-    {
-      "epoch": 16.276401564537156,
-      "grad_norm": 1.4136409759521484,
-      "learning_rate": 6.914572864321608e-05,
-      "loss": 5.6982,
-      "step": 31210
-    },
-    {
-      "epoch": 16.276923076923076,
-      "grad_norm": 1.4662851095199585,
-      "learning_rate": 6.914472361809046e-05,
-      "loss": 4.8913,
-      "step": 31211
-    },
-    {
-      "epoch": 16.277444589308995,
-      "grad_norm": 1.4879060983657837,
-      "learning_rate": 6.914371859296482e-05,
-      "loss": 5.2742,
-      "step": 31212
-    },
-    {
-      "epoch": 16.277966101694915,
-      "grad_norm": 1.578859567642212,
-      "learning_rate": 6.91427135678392e-05,
-      "loss": 5.267,
-      "step": 31213
-    },
-    {
-      "epoch": 16.278487614080834,
-      "grad_norm": 1.5114257335662842,
-      "learning_rate": 6.914170854271356e-05,
-      "loss": 4.9727,
-      "step": 31214
-    },
-    {
-      "epoch": 16.279009126466754,
-      "grad_norm": 1.419581651687622,
-      "learning_rate": 6.914070351758794e-05,
-      "loss": 5.7282,
-      "step": 31215
-    },
-    {
-      "epoch": 16.279530638852673,
-      "grad_norm": 1.5299999713897705,
-      "learning_rate": 6.913969849246232e-05,
-      "loss": 4.8939,
-      "step": 31216
-    },
-    {
-      "epoch": 16.280052151238593,
-      "grad_norm": 1.4603723287582397,
-      "learning_rate": 6.91386934673367e-05,
-      "loss": 5.5721,
-      "step": 31217
-    },
-    {
-      "epoch": 16.280573663624512,
-      "grad_norm": 1.4693877696990967,
-      "learning_rate": 6.913768844221106e-05,
-      "loss": 5.5191,
-      "step": 31218
-    },
-    {
-      "epoch": 16.28109517601043,
-      "grad_norm": 1.5239485502243042,
-      "learning_rate": 6.913668341708544e-05,
-      "loss": 4.9307,
-      "step": 31219
-    },
-    {
-      "epoch": 16.28161668839635,
-      "grad_norm": 1.5082708597183228,
-      "learning_rate": 6.91356783919598e-05,
-      "loss": 5.4136,
-      "step": 31220
-    },
-    {
-      "epoch": 16.282138200782267,
-      "grad_norm": 1.465074062347412,
-      "learning_rate": 6.913467336683418e-05,
-      "loss": 5.4136,
-      "step": 31221
-    },
-    {
-      "epoch": 16.282659713168186,
-      "grad_norm": 1.5279515981674194,
-      "learning_rate": 6.913366834170855e-05,
-      "loss": 5.145,
-      "step": 31222
-    },
-    {
-      "epoch": 16.283181225554106,
-      "grad_norm": 1.454884648323059,
-      "learning_rate": 6.913266331658291e-05,
-      "loss": 5.6819,
-      "step": 31223
-    },
-    {
-      "epoch": 16.283702737940025,
-      "grad_norm": 1.5164965391159058,
-      "learning_rate": 6.913165829145729e-05,
-      "loss": 5.4672,
-      "step": 31224
-    },
-    {
-      "epoch": 16.284224250325945,
-      "grad_norm": 1.3841710090637207,
-      "learning_rate": 6.913065326633165e-05,
-      "loss": 5.514,
-      "step": 31225
-    },
-    {
-      "epoch": 16.284745762711864,
-      "grad_norm": 1.4138258695602417,
-      "learning_rate": 6.912964824120603e-05,
-      "loss": 5.7555,
-      "step": 31226
-    },
-    {
-      "epoch": 16.285267275097784,
-      "grad_norm": 1.4587318897247314,
-      "learning_rate": 6.912864321608041e-05,
-      "loss": 5.132,
-      "step": 31227
-    },
-    {
-      "epoch": 16.285788787483703,
-      "grad_norm": 1.4891409873962402,
-      "learning_rate": 6.912763819095479e-05,
-      "loss": 5.5914,
-      "step": 31228
-    },
-    {
-      "epoch": 16.286310299869623,
-      "grad_norm": 1.4580469131469727,
-      "learning_rate": 6.912663316582915e-05,
-      "loss": 5.6618,
-      "step": 31229
-    },
-    {
-      "epoch": 16.286831812255542,
-      "grad_norm": 1.5582109689712524,
-      "learning_rate": 6.912562814070353e-05,
-      "loss": 5.0162,
-      "step": 31230
-    },
-    {
-      "epoch": 16.28735332464146,
-      "grad_norm": 1.4468975067138672,
-      "learning_rate": 6.912462311557789e-05,
-      "loss": 5.4752,
-      "step": 31231
-    },
-    {
-      "epoch": 16.28787483702738,
-      "grad_norm": 1.5847190618515015,
-      "learning_rate": 6.912361809045227e-05,
-      "loss": 4.7837,
-      "step": 31232
-    },
-    {
-      "epoch": 16.288396349413297,
-      "grad_norm": 1.4403737783432007,
-      "learning_rate": 6.912261306532663e-05,
-      "loss": 5.5088,
-      "step": 31233
-    },
-    {
-      "epoch": 16.288917861799217,
-      "grad_norm": 1.5605820417404175,
-      "learning_rate": 6.912160804020101e-05,
-      "loss": 4.9008,
-      "step": 31234
-    },
-    {
-      "epoch": 16.289439374185136,
-      "grad_norm": 1.5190587043762207,
-      "learning_rate": 6.912060301507538e-05,
-      "loss": 5.1531,
-      "step": 31235
-    },
-    {
-      "epoch": 16.289960886571055,
-      "grad_norm": 1.4247583150863647,
-      "learning_rate": 6.911959798994975e-05,
-      "loss": 5.6407,
-      "step": 31236
-    },
-    {
-      "epoch": 16.290482398956975,
-      "grad_norm": 1.5523450374603271,
-      "learning_rate": 6.911859296482413e-05,
-      "loss": 4.8086,
-      "step": 31237
-    },
-    {
-      "epoch": 16.291003911342894,
-      "grad_norm": 1.5023653507232666,
-      "learning_rate": 6.91175879396985e-05,
-      "loss": 5.4157,
-      "step": 31238
-    },
-    {
-      "epoch": 16.291525423728814,
-      "grad_norm": 1.4596022367477417,
-      "learning_rate": 6.911658291457287e-05,
-      "loss": 5.7254,
-      "step": 31239
-    },
-    {
-      "epoch": 16.292046936114733,
-      "grad_norm": 1.4000180959701538,
-      "learning_rate": 6.911557788944724e-05,
-      "loss": 5.6521,
-      "step": 31240
-    },
-    {
-      "epoch": 16.292568448500653,
-      "grad_norm": 1.400017499923706,
-      "learning_rate": 6.911457286432162e-05,
-      "loss": 5.6176,
-      "step": 31241
-    },
-    {
-      "epoch": 16.293089960886572,
-      "grad_norm": 1.5628305673599243,
-      "learning_rate": 6.911356783919598e-05,
-      "loss": 5.4478,
-      "step": 31242
-    },
-    {
-      "epoch": 16.29361147327249,
-      "grad_norm": 1.424174427986145,
-      "learning_rate": 6.911256281407036e-05,
-      "loss": 5.6502,
-      "step": 31243
-    },
-    {
-      "epoch": 16.294132985658408,
-      "grad_norm": 1.417543649673462,
-      "learning_rate": 6.911155778894472e-05,
-      "loss": 5.1974,
-      "step": 31244
-    },
-    {
-      "epoch": 16.294654498044327,
-      "grad_norm": 1.5960663557052612,
-      "learning_rate": 6.91105527638191e-05,
-      "loss": 5.4081,
-      "step": 31245
-    },
-    {
-      "epoch": 16.295176010430247,
-      "grad_norm": 1.5144634246826172,
-      "learning_rate": 6.910954773869346e-05,
-      "loss": 5.348,
-      "step": 31246
-    },
-    {
-      "epoch": 16.295697522816166,
-      "grad_norm": 1.5649641752243042,
-      "learning_rate": 6.910854271356784e-05,
-      "loss": 4.9098,
-      "step": 31247
-    },
-    {
-      "epoch": 16.296219035202085,
-      "grad_norm": 1.610914945602417,
-      "learning_rate": 6.910753768844222e-05,
-      "loss": 5.3447,
-      "step": 31248
-    },
-    {
-      "epoch": 16.296740547588005,
-      "grad_norm": 1.556652545928955,
-      "learning_rate": 6.910653266331658e-05,
-      "loss": 5.228,
-      "step": 31249
-    },
-    {
-      "epoch": 16.297262059973924,
-      "grad_norm": 1.4710885286331177,
-      "learning_rate": 6.910552763819096e-05,
-      "loss": 5.504,
-      "step": 31250
-    },
-    {
-      "epoch": 16.297783572359844,
-      "grad_norm": 1.5255745649337769,
-      "learning_rate": 6.910452261306533e-05,
-      "loss": 5.1473,
-      "step": 31251
-    },
-    {
-      "epoch": 16.298305084745763,
-      "grad_norm": 1.504607915878296,
-      "learning_rate": 6.91035175879397e-05,
-      "loss": 5.1172,
-      "step": 31252
-    },
-    {
-      "epoch": 16.298826597131683,
-      "grad_norm": 1.646968126296997,
-      "learning_rate": 6.910251256281407e-05,
-      "loss": 4.73,
-      "step": 31253
-    },
-    {
-      "epoch": 16.299348109517602,
-      "grad_norm": 1.4869636297225952,
-      "learning_rate": 6.910150753768845e-05,
-      "loss": 5.5766,
-      "step": 31254
-    },
-    {
-      "epoch": 16.29986962190352,
-      "grad_norm": 1.406297206878662,
-      "learning_rate": 6.910050251256281e-05,
-      "loss": 4.8736,
-      "step": 31255
-    },
-    {
-      "epoch": 16.300391134289438,
-      "grad_norm": 1.5708647966384888,
-      "learning_rate": 6.909949748743719e-05,
-      "loss": 4.6274,
-      "step": 31256
-    },
-    {
-      "epoch": 16.300912646675357,
-      "grad_norm": 1.5810109376907349,
-      "learning_rate": 6.909849246231157e-05,
-      "loss": 5.0635,
-      "step": 31257
-    },
-    {
-      "epoch": 16.301434159061277,
-      "grad_norm": 1.518337607383728,
-      "learning_rate": 6.909748743718594e-05,
-      "loss": 5.4153,
-      "step": 31258
-    },
-    {
-      "epoch": 16.301955671447196,
-      "grad_norm": 1.465943694114685,
-      "learning_rate": 6.909648241206031e-05,
-      "loss": 5.4488,
-      "step": 31259
-    },
-    {
-      "epoch": 16.302477183833116,
-      "grad_norm": 1.3215973377227783,
-      "learning_rate": 6.909547738693469e-05,
-      "loss": 5.5549,
-      "step": 31260
-    },
-    {
-      "epoch": 16.302998696219035,
-      "grad_norm": 1.3173540830612183,
-      "learning_rate": 6.909447236180905e-05,
-      "loss": 5.701,
-      "step": 31261
-    },
-    {
-      "epoch": 16.303520208604954,
-      "grad_norm": 1.62455415725708,
-      "learning_rate": 6.909346733668341e-05,
-      "loss": 4.9435,
-      "step": 31262
-    },
-    {
-      "epoch": 16.304041720990874,
-      "grad_norm": 1.5001991987228394,
-      "learning_rate": 6.909246231155779e-05,
-      "loss": 5.3616,
-      "step": 31263
-    },
-    {
-      "epoch": 16.304563233376793,
-      "grad_norm": 1.4514966011047363,
-      "learning_rate": 6.909145728643216e-05,
-      "loss": 5.4493,
-      "step": 31264
-    },
-    {
-      "epoch": 16.305084745762713,
-      "grad_norm": 1.5461254119873047,
-      "learning_rate": 6.909045226130653e-05,
-      "loss": 5.6012,
-      "step": 31265
-    },
-    {
-      "epoch": 16.305606258148632,
-      "grad_norm": 1.4280036687850952,
-      "learning_rate": 6.90894472361809e-05,
-      "loss": 5.0724,
-      "step": 31266
-    },
-    {
-      "epoch": 16.30612777053455,
-      "grad_norm": 1.4334359169006348,
-      "learning_rate": 6.908844221105528e-05,
-      "loss": 5.3279,
-      "step": 31267
-    },
-    {
-      "epoch": 16.306649282920468,
-      "grad_norm": 1.395806908607483,
-      "learning_rate": 6.908743718592965e-05,
-      "loss": 5.7179,
-      "step": 31268
-    },
-    {
-      "epoch": 16.307170795306387,
-      "grad_norm": 1.4503413438796997,
-      "learning_rate": 6.908643216080403e-05,
-      "loss": 5.7819,
-      "step": 31269
-    },
-    {
-      "epoch": 16.307692307692307,
-      "grad_norm": 1.54376220703125,
-      "learning_rate": 6.90854271356784e-05,
-      "loss": 5.4849,
-      "step": 31270
-    },
-    {
-      "epoch": 16.308213820078226,
-      "grad_norm": 1.4756830930709839,
-      "learning_rate": 6.908442211055277e-05,
-      "loss": 5.4708,
-      "step": 31271
-    },
-    {
-      "epoch": 16.308735332464146,
-      "grad_norm": 1.4799803495407104,
-      "learning_rate": 6.908341708542714e-05,
-      "loss": 5.1967,
-      "step": 31272
-    },
-    {
-      "epoch": 16.309256844850065,
-      "grad_norm": 1.5036073923110962,
-      "learning_rate": 6.908241206030152e-05,
-      "loss": 5.4543,
-      "step": 31273
-    },
-    {
-      "epoch": 16.309778357235984,
-      "grad_norm": 1.4551055431365967,
-      "learning_rate": 6.908140703517588e-05,
-      "loss": 5.3086,
-      "step": 31274
-    },
-    {
-      "epoch": 16.310299869621904,
-      "grad_norm": 1.4543179273605347,
-      "learning_rate": 6.908040201005024e-05,
-      "loss": 5.5706,
-      "step": 31275
-    },
-    {
-      "epoch": 16.310821382007823,
-      "grad_norm": 1.4630993604660034,
-      "learning_rate": 6.907939698492462e-05,
-      "loss": 5.1712,
-      "step": 31276
-    },
-    {
-      "epoch": 16.311342894393743,
-      "grad_norm": 1.4701939821243286,
-      "learning_rate": 6.9078391959799e-05,
-      "loss": 5.3825,
-      "step": 31277
-    },
-    {
-      "epoch": 16.311864406779662,
-      "grad_norm": 1.4726046323776245,
-      "learning_rate": 6.907738693467338e-05,
-      "loss": 5.0443,
-      "step": 31278
-    },
-    {
-      "epoch": 16.312385919165582,
-      "grad_norm": 1.4095540046691895,
-      "learning_rate": 6.907638190954774e-05,
-      "loss": 5.4287,
-      "step": 31279
-    },
-    {
-      "epoch": 16.312907431551498,
-      "grad_norm": 1.4833619594573975,
-      "learning_rate": 6.907537688442212e-05,
-      "loss": 5.4447,
-      "step": 31280
-    },
-    {
-      "epoch": 16.313428943937417,
-      "grad_norm": 1.5654155015945435,
-      "learning_rate": 6.907437185929648e-05,
-      "loss": 5.6097,
-      "step": 31281
-    },
-    {
-      "epoch": 16.313950456323337,
-      "grad_norm": 1.5312641859054565,
-      "learning_rate": 6.907336683417086e-05,
-      "loss": 5.4514,
-      "step": 31282
-    },
-    {
-      "epoch": 16.314471968709256,
-      "grad_norm": 1.559386134147644,
-      "learning_rate": 6.907236180904523e-05,
-      "loss": 5.6064,
-      "step": 31283
-    },
-    {
-      "epoch": 16.314993481095176,
-      "grad_norm": 1.537150502204895,
-      "learning_rate": 6.90713567839196e-05,
-      "loss": 5.331,
-      "step": 31284
-    },
-    {
-      "epoch": 16.315514993481095,
-      "grad_norm": 1.4087860584259033,
-      "learning_rate": 6.907035175879397e-05,
-      "loss": 5.5754,
-      "step": 31285
-    },
-    {
-      "epoch": 16.316036505867014,
-      "grad_norm": 1.439221739768982,
-      "learning_rate": 6.906934673366835e-05,
-      "loss": 5.5437,
-      "step": 31286
-    },
-    {
-      "epoch": 16.316558018252934,
-      "grad_norm": 1.4157744646072388,
-      "learning_rate": 6.906834170854271e-05,
-      "loss": 5.1712,
-      "step": 31287
-    },
-    {
-      "epoch": 16.317079530638853,
-      "grad_norm": 1.4544053077697754,
-      "learning_rate": 6.906733668341709e-05,
-      "loss": 5.7976,
-      "step": 31288
-    },
-    {
-      "epoch": 16.317601043024773,
-      "grad_norm": 1.5816882848739624,
-      "learning_rate": 6.906633165829147e-05,
-      "loss": 4.9377,
-      "step": 31289
-    },
-    {
-      "epoch": 16.318122555410692,
-      "grad_norm": 1.54128897190094,
-      "learning_rate": 6.906532663316583e-05,
-      "loss": 4.9852,
-      "step": 31290
-    },
-    {
-      "epoch": 16.318644067796612,
-      "grad_norm": 1.5138211250305176,
-      "learning_rate": 6.906432160804021e-05,
-      "loss": 5.5167,
-      "step": 31291
-    },
-    {
-      "epoch": 16.319165580182528,
-      "grad_norm": 1.4980460405349731,
-      "learning_rate": 6.906331658291457e-05,
-      "loss": 5.3663,
-      "step": 31292
-    },
-    {
-      "epoch": 16.319687092568447,
-      "grad_norm": 1.435627818107605,
-      "learning_rate": 6.906231155778895e-05,
-      "loss": 5.0088,
-      "step": 31293
-    },
-    {
-      "epoch": 16.320208604954367,
-      "grad_norm": 1.512392282485962,
-      "learning_rate": 6.906130653266332e-05,
-      "loss": 4.8489,
-      "step": 31294
-    },
-    {
-      "epoch": 16.320730117340286,
-      "grad_norm": 1.5622304677963257,
-      "learning_rate": 6.906030150753769e-05,
-      "loss": 5.0977,
-      "step": 31295
-    },
-    {
-      "epoch": 16.321251629726206,
-      "grad_norm": 1.5381481647491455,
-      "learning_rate": 6.905929648241206e-05,
-      "loss": 5.0898,
-      "step": 31296
-    },
-    {
-      "epoch": 16.321773142112125,
-      "grad_norm": 1.4878348112106323,
-      "learning_rate": 6.905829145728644e-05,
-      "loss": 5.0534,
-      "step": 31297
-    },
-    {
-      "epoch": 16.322294654498045,
-      "grad_norm": 1.5919172763824463,
-      "learning_rate": 6.90572864321608e-05,
-      "loss": 5.2938,
-      "step": 31298
-    },
-    {
-      "epoch": 16.322816166883964,
-      "grad_norm": 1.5337930917739868,
-      "learning_rate": 6.905628140703518e-05,
-      "loss": 5.3141,
-      "step": 31299
-    },
-    {
-      "epoch": 16.323337679269883,
-      "grad_norm": 1.4267339706420898,
-      "learning_rate": 6.905527638190955e-05,
-      "loss": 5.647,
-      "step": 31300
-    },
-    {
-      "epoch": 16.323859191655803,
-      "grad_norm": 1.506076455116272,
-      "learning_rate": 6.905427135678393e-05,
-      "loss": 5.4333,
-      "step": 31301
-    },
-    {
-      "epoch": 16.324380704041722,
-      "grad_norm": 1.7269736528396606,
-      "learning_rate": 6.90532663316583e-05,
-      "loss": 4.8798,
-      "step": 31302
-    },
-    {
-      "epoch": 16.324902216427642,
-      "grad_norm": 1.5383551120758057,
-      "learning_rate": 6.905226130653266e-05,
-      "loss": 5.0544,
-      "step": 31303
-    },
-    {
-      "epoch": 16.325423728813558,
-      "grad_norm": 1.5923889875411987,
-      "learning_rate": 6.905125628140704e-05,
-      "loss": 5.4424,
-      "step": 31304
-    },
-    {
-      "epoch": 16.325945241199477,
-      "grad_norm": 1.5872708559036255,
-      "learning_rate": 6.90502512562814e-05,
-      "loss": 4.9055,
-      "step": 31305
-    },
-    {
-      "epoch": 16.326466753585397,
-      "grad_norm": 1.8498812913894653,
-      "learning_rate": 6.904924623115578e-05,
-      "loss": 5.4021,
-      "step": 31306
-    },
-    {
-      "epoch": 16.326988265971316,
-      "grad_norm": 1.6337701082229614,
-      "learning_rate": 6.904824120603015e-05,
-      "loss": 4.8566,
-      "step": 31307
-    },
-    {
-      "epoch": 16.327509778357236,
-      "grad_norm": 1.5288203954696655,
-      "learning_rate": 6.904723618090452e-05,
-      "loss": 5.6325,
-      "step": 31308
-    },
-    {
-      "epoch": 16.328031290743155,
-      "grad_norm": 1.472981572151184,
-      "learning_rate": 6.90462311557789e-05,
-      "loss": 5.6517,
-      "step": 31309
-    },
-    {
-      "epoch": 16.328552803129075,
-      "grad_norm": 1.5649245977401733,
-      "learning_rate": 6.904522613065328e-05,
-      "loss": 5.0182,
-      "step": 31310
-    },
-    {
-      "epoch": 16.329074315514994,
-      "grad_norm": 1.5460681915283203,
-      "learning_rate": 6.904422110552764e-05,
-      "loss": 5.1401,
-      "step": 31311
-    },
-    {
-      "epoch": 16.329595827900913,
-      "grad_norm": 1.409224033355713,
-      "learning_rate": 6.904321608040202e-05,
-      "loss": 5.8093,
-      "step": 31312
-    },
-    {
-      "epoch": 16.330117340286833,
-      "grad_norm": 1.4421751499176025,
-      "learning_rate": 6.904221105527639e-05,
-      "loss": 5.4908,
-      "step": 31313
-    },
-    {
-      "epoch": 16.330638852672752,
-      "grad_norm": 1.496168613433838,
-      "learning_rate": 6.904120603015076e-05,
-      "loss": 5.613,
-      "step": 31314
-    },
-    {
-      "epoch": 16.331160365058672,
-      "grad_norm": 1.5203732252120972,
-      "learning_rate": 6.904020100502513e-05,
-      "loss": 5.1162,
-      "step": 31315
-    },
-    {
-      "epoch": 16.331681877444588,
-      "grad_norm": 1.4174896478652954,
-      "learning_rate": 6.903919597989949e-05,
-      "loss": 5.0389,
-      "step": 31316
-    },
-    {
-      "epoch": 16.332203389830507,
-      "grad_norm": 1.4883846044540405,
-      "learning_rate": 6.903819095477387e-05,
-      "loss": 5.0027,
-      "step": 31317
-    },
-    {
-      "epoch": 16.332724902216427,
-      "grad_norm": 1.4776694774627686,
-      "learning_rate": 6.903718592964823e-05,
-      "loss": 5.7521,
-      "step": 31318
-    },
-    {
-      "epoch": 16.333246414602346,
-      "grad_norm": 1.4567378759384155,
-      "learning_rate": 6.903618090452261e-05,
-      "loss": 5.216,
-      "step": 31319
-    },
-    {
-      "epoch": 16.333767926988266,
-      "grad_norm": 1.4200201034545898,
-      "learning_rate": 6.903517587939699e-05,
-      "loss": 5.473,
-      "step": 31320
-    },
-    {
-      "epoch": 16.334289439374185,
-      "grad_norm": 1.6450248956680298,
-      "learning_rate": 6.903417085427137e-05,
-      "loss": 4.9807,
-      "step": 31321
-    },
-    {
-      "epoch": 16.334810951760105,
-      "grad_norm": 1.4071166515350342,
-      "learning_rate": 6.903316582914573e-05,
-      "loss": 5.7406,
-      "step": 31322
-    },
-    {
-      "epoch": 16.335332464146024,
-      "grad_norm": 1.5345059633255005,
-      "learning_rate": 6.903216080402011e-05,
-      "loss": 5.189,
-      "step": 31323
-    },
-    {
-      "epoch": 16.335853976531943,
-      "grad_norm": 1.4464019536972046,
-      "learning_rate": 6.903115577889447e-05,
-      "loss": 5.6992,
-      "step": 31324
-    },
-    {
-      "epoch": 16.336375488917863,
-      "grad_norm": 1.4235539436340332,
-      "learning_rate": 6.903015075376885e-05,
-      "loss": 5.3245,
-      "step": 31325
-    },
-    {
-      "epoch": 16.336897001303782,
-      "grad_norm": 1.42299485206604,
-      "learning_rate": 6.902914572864322e-05,
-      "loss": 5.8712,
-      "step": 31326
-    },
-    {
-      "epoch": 16.3374185136897,
-      "grad_norm": 1.4761815071105957,
-      "learning_rate": 6.90281407035176e-05,
-      "loss": 5.8271,
-      "step": 31327
-    },
-    {
-      "epoch": 16.337940026075618,
-      "grad_norm": 1.4862767457962036,
-      "learning_rate": 6.902713567839196e-05,
-      "loss": 5.2627,
-      "step": 31328
-    },
-    {
-      "epoch": 16.338461538461537,
-      "grad_norm": 1.4116549491882324,
-      "learning_rate": 6.902613065326634e-05,
-      "loss": 5.4658,
-      "step": 31329
-    },
-    {
-      "epoch": 16.338983050847457,
-      "grad_norm": 1.4406287670135498,
-      "learning_rate": 6.902512562814071e-05,
-      "loss": 4.6741,
-      "step": 31330
-    },
-    {
-      "epoch": 16.339504563233376,
-      "grad_norm": 1.5698314905166626,
-      "learning_rate": 6.902412060301508e-05,
-      "loss": 4.7017,
-      "step": 31331
-    },
-    {
-      "epoch": 16.340026075619296,
-      "grad_norm": 1.514410138130188,
-      "learning_rate": 6.902311557788946e-05,
-      "loss": 5.5212,
-      "step": 31332
-    },
-    {
-      "epoch": 16.340547588005215,
-      "grad_norm": 1.41310715675354,
-      "learning_rate": 6.902211055276382e-05,
-      "loss": 5.3261,
-      "step": 31333
-    },
-    {
-      "epoch": 16.341069100391135,
-      "grad_norm": 1.525736689567566,
-      "learning_rate": 6.90211055276382e-05,
-      "loss": 5.2288,
-      "step": 31334
-    },
-    {
-      "epoch": 16.341590612777054,
-      "grad_norm": 1.5990067720413208,
-      "learning_rate": 6.902010050251256e-05,
-      "loss": 5.1003,
-      "step": 31335
-    },
-    {
-      "epoch": 16.342112125162974,
-      "grad_norm": 1.425395131111145,
-      "learning_rate": 6.901909547738694e-05,
-      "loss": 5.7655,
-      "step": 31336
-    },
-    {
-      "epoch": 16.342633637548893,
-      "grad_norm": 1.4302016496658325,
-      "learning_rate": 6.90180904522613e-05,
-      "loss": 5.1695,
-      "step": 31337
-    },
-    {
-      "epoch": 16.343155149934812,
-      "grad_norm": 1.4791518449783325,
-      "learning_rate": 6.901708542713568e-05,
-      "loss": 5.1392,
-      "step": 31338
-    },
-    {
-      "epoch": 16.343676662320732,
-      "grad_norm": 1.5140976905822754,
-      "learning_rate": 6.901608040201005e-05,
-      "loss": 5.4317,
-      "step": 31339
-    },
-    {
-      "epoch": 16.344198174706648,
-      "grad_norm": 1.5036382675170898,
-      "learning_rate": 6.901507537688442e-05,
-      "loss": 5.6927,
-      "step": 31340
-    },
-    {
-      "epoch": 16.344719687092567,
-      "grad_norm": 1.4393466711044312,
-      "learning_rate": 6.90140703517588e-05,
-      "loss": 5.6112,
-      "step": 31341
-    },
-    {
-      "epoch": 16.345241199478487,
-      "grad_norm": 1.5401616096496582,
-      "learning_rate": 6.901306532663317e-05,
-      "loss": 5.3054,
-      "step": 31342
-    },
-    {
-      "epoch": 16.345762711864406,
-      "grad_norm": 1.4413738250732422,
-      "learning_rate": 6.901206030150754e-05,
-      "loss": 5.1656,
-      "step": 31343
-    },
-    {
-      "epoch": 16.346284224250326,
-      "grad_norm": 1.5150539875030518,
-      "learning_rate": 6.901105527638191e-05,
-      "loss": 5.4422,
-      "step": 31344
-    },
-    {
-      "epoch": 16.346805736636245,
-      "grad_norm": 1.4767959117889404,
-      "learning_rate": 6.901005025125629e-05,
-      "loss": 5.6114,
-      "step": 31345
-    },
-    {
-      "epoch": 16.347327249022165,
-      "grad_norm": 1.6474640369415283,
-      "learning_rate": 6.900904522613065e-05,
-      "loss": 4.8659,
-      "step": 31346
-    },
-    {
-      "epoch": 16.347848761408084,
-      "grad_norm": 1.4681963920593262,
-      "learning_rate": 6.900804020100503e-05,
-      "loss": 5.3085,
-      "step": 31347
-    },
-    {
-      "epoch": 16.348370273794004,
-      "grad_norm": 1.511553406715393,
-      "learning_rate": 6.900703517587939e-05,
-      "loss": 5.3556,
-      "step": 31348
-    },
-    {
-      "epoch": 16.348891786179923,
-      "grad_norm": 1.6291661262512207,
-      "learning_rate": 6.900603015075377e-05,
-      "loss": 5.2552,
-      "step": 31349
-    },
-    {
-      "epoch": 16.349413298565842,
-      "grad_norm": 1.399874210357666,
-      "learning_rate": 6.900502512562815e-05,
-      "loss": 5.3926,
-      "step": 31350
-    },
-    {
-      "epoch": 16.34993481095176,
-      "grad_norm": 1.4337115287780762,
-      "learning_rate": 6.900402010050253e-05,
-      "loss": 5.3546,
-      "step": 31351
-    },
-    {
-      "epoch": 16.350456323337678,
-      "grad_norm": 1.5041087865829468,
-      "learning_rate": 6.900301507537689e-05,
-      "loss": 5.4248,
-      "step": 31352
-    },
-    {
-      "epoch": 16.350977835723597,
-      "grad_norm": 1.4396296739578247,
-      "learning_rate": 6.900201005025127e-05,
-      "loss": 5.4866,
-      "step": 31353
-    },
-    {
-      "epoch": 16.351499348109517,
-      "grad_norm": 1.4640523195266724,
-      "learning_rate": 6.900100502512563e-05,
-      "loss": 5.456,
-      "step": 31354
-    },
-    {
-      "epoch": 16.352020860495436,
-      "grad_norm": 1.6403312683105469,
-      "learning_rate": 6.9e-05,
-      "loss": 4.4875,
-      "step": 31355
-    },
-    {
-      "epoch": 16.352542372881356,
-      "grad_norm": 1.3589069843292236,
-      "learning_rate": 6.899899497487437e-05,
-      "loss": 5.5492,
-      "step": 31356
-    },
-    {
-      "epoch": 16.353063885267275,
-      "grad_norm": 1.6231597661972046,
-      "learning_rate": 6.899798994974874e-05,
-      "loss": 5.5151,
-      "step": 31357
-    },
-    {
-      "epoch": 16.353585397653195,
-      "grad_norm": 1.5152119398117065,
-      "learning_rate": 6.899698492462312e-05,
-      "loss": 4.7883,
-      "step": 31358
-    },
-    {
-      "epoch": 16.354106910039114,
-      "grad_norm": 1.4809434413909912,
-      "learning_rate": 6.899597989949748e-05,
-      "loss": 5.3221,
-      "step": 31359
-    },
-    {
-      "epoch": 16.354628422425034,
-      "grad_norm": 1.5151695013046265,
-      "learning_rate": 6.899497487437186e-05,
-      "loss": 5.2874,
-      "step": 31360
-    },
-    {
-      "epoch": 16.355149934810953,
-      "grad_norm": 1.4297206401824951,
-      "learning_rate": 6.899396984924624e-05,
-      "loss": 5.456,
-      "step": 31361
-    },
-    {
-      "epoch": 16.355671447196872,
-      "grad_norm": 1.4868309497833252,
-      "learning_rate": 6.899296482412061e-05,
-      "loss": 5.1897,
-      "step": 31362
-    },
-    {
-      "epoch": 16.35619295958279,
-      "grad_norm": 1.5012940168380737,
-      "learning_rate": 6.899195979899498e-05,
-      "loss": 5.2518,
-      "step": 31363
-    },
-    {
-      "epoch": 16.356714471968708,
-      "grad_norm": 1.4325059652328491,
-      "learning_rate": 6.899095477386936e-05,
-      "loss": 4.8565,
-      "step": 31364
-    },
-    {
-      "epoch": 16.357235984354627,
-      "grad_norm": 1.4928745031356812,
-      "learning_rate": 6.898994974874372e-05,
-      "loss": 5.5845,
-      "step": 31365
-    },
-    {
-      "epoch": 16.357757496740547,
-      "grad_norm": 1.5800449848175049,
-      "learning_rate": 6.89889447236181e-05,
-      "loss": 5.0335,
-      "step": 31366
-    },
-    {
-      "epoch": 16.358279009126466,
-      "grad_norm": 1.435771107673645,
-      "learning_rate": 6.898793969849246e-05,
-      "loss": 5.558,
-      "step": 31367
-    },
-    {
-      "epoch": 16.358800521512386,
-      "grad_norm": 1.590396523475647,
-      "learning_rate": 6.898693467336683e-05,
-      "loss": 5.4759,
-      "step": 31368
-    },
-    {
-      "epoch": 16.359322033898305,
-      "grad_norm": 1.5736823081970215,
-      "learning_rate": 6.89859296482412e-05,
-      "loss": 5.4333,
-      "step": 31369
-    },
-    {
-      "epoch": 16.359843546284225,
-      "grad_norm": 1.4675753116607666,
-      "learning_rate": 6.898492462311558e-05,
-      "loss": 5.6642,
-      "step": 31370
-    },
-    {
-      "epoch": 16.360365058670144,
-      "grad_norm": 1.4114046096801758,
-      "learning_rate": 6.898391959798996e-05,
-      "loss": 5.7537,
-      "step": 31371
-    },
-    {
-      "epoch": 16.360886571056064,
-      "grad_norm": 1.5001496076583862,
-      "learning_rate": 6.898291457286432e-05,
-      "loss": 5.4254,
-      "step": 31372
-    },
-    {
-      "epoch": 16.361408083441983,
-      "grad_norm": 1.5954521894454956,
-      "learning_rate": 6.89819095477387e-05,
-      "loss": 5.1979,
-      "step": 31373
-    },
-    {
-      "epoch": 16.361929595827903,
-      "grad_norm": 1.5034642219543457,
-      "learning_rate": 6.898090452261307e-05,
-      "loss": 5.6631,
-      "step": 31374
-    },
-    {
-      "epoch": 16.36245110821382,
-      "grad_norm": 1.5352011919021606,
-      "learning_rate": 6.897989949748744e-05,
-      "loss": 5.3832,
-      "step": 31375
-    },
-    {
-      "epoch": 16.362972620599738,
-      "grad_norm": 1.393612265586853,
-      "learning_rate": 6.897889447236181e-05,
-      "loss": 5.7706,
-      "step": 31376
-    },
-    {
-      "epoch": 16.363494132985657,
-      "grad_norm": 1.3504966497421265,
-      "learning_rate": 6.897788944723619e-05,
-      "loss": 5.1577,
-      "step": 31377
-    },
-    {
-      "epoch": 16.364015645371577,
-      "grad_norm": 1.4589227437973022,
-      "learning_rate": 6.897688442211055e-05,
-      "loss": 5.7165,
-      "step": 31378
-    },
-    {
-      "epoch": 16.364537157757496,
-      "grad_norm": 1.5199185609817505,
-      "learning_rate": 6.897587939698493e-05,
-      "loss": 5.3611,
-      "step": 31379
-    },
-    {
-      "epoch": 16.365058670143416,
-      "grad_norm": 1.5511219501495361,
-      "learning_rate": 6.89748743718593e-05,
-      "loss": 5.7015,
-      "step": 31380
-    },
-    {
-      "epoch": 16.365580182529335,
-      "grad_norm": 1.473868727684021,
-      "learning_rate": 6.897386934673367e-05,
-      "loss": 5.1109,
-      "step": 31381
-    },
-    {
-      "epoch": 16.366101694915255,
-      "grad_norm": 1.3872687816619873,
-      "learning_rate": 6.897286432160805e-05,
-      "loss": 5.7459,
-      "step": 31382
-    },
-    {
-      "epoch": 16.366623207301174,
-      "grad_norm": 1.4107608795166016,
-      "learning_rate": 6.897185929648241e-05,
-      "loss": 5.5284,
-      "step": 31383
-    },
-    {
-      "epoch": 16.367144719687094,
-      "grad_norm": 1.5353628396987915,
-      "learning_rate": 6.897085427135679e-05,
-      "loss": 5.1862,
-      "step": 31384
-    },
-    {
-      "epoch": 16.367666232073013,
-      "grad_norm": 1.4989163875579834,
-      "learning_rate": 6.896984924623116e-05,
-      "loss": 5.3022,
-      "step": 31385
-    },
-    {
-      "epoch": 16.368187744458933,
-      "grad_norm": 1.5068252086639404,
-      "learning_rate": 6.896884422110553e-05,
-      "loss": 5.3046,
-      "step": 31386
-    },
-    {
-      "epoch": 16.36870925684485,
-      "grad_norm": 1.4152882099151611,
-      "learning_rate": 6.89678391959799e-05,
-      "loss": 5.537,
-      "step": 31387
-    },
-    {
-      "epoch": 16.369230769230768,
-      "grad_norm": 1.4673343896865845,
-      "learning_rate": 6.896683417085428e-05,
-      "loss": 5.4425,
-      "step": 31388
-    },
-    {
-      "epoch": 16.369752281616687,
-      "grad_norm": 1.686995506286621,
-      "learning_rate": 6.896582914572864e-05,
-      "loss": 5.5093,
-      "step": 31389
-    },
-    {
-      "epoch": 16.370273794002607,
-      "grad_norm": 1.5616475343704224,
-      "learning_rate": 6.896482412060302e-05,
-      "loss": 5.2094,
-      "step": 31390
-    },
-    {
-      "epoch": 16.370795306388526,
-      "grad_norm": 1.412637710571289,
-      "learning_rate": 6.89638190954774e-05,
-      "loss": 5.6569,
-      "step": 31391
-    },
-    {
-      "epoch": 16.371316818774446,
-      "grad_norm": 1.6278115510940552,
-      "learning_rate": 6.896281407035177e-05,
-      "loss": 5.1612,
-      "step": 31392
-    },
-    {
-      "epoch": 16.371838331160365,
-      "grad_norm": 1.5247102975845337,
-      "learning_rate": 6.896180904522614e-05,
-      "loss": 4.6558,
-      "step": 31393
-    },
-    {
-      "epoch": 16.372359843546285,
-      "grad_norm": 1.4950047731399536,
-      "learning_rate": 6.896080402010052e-05,
-      "loss": 4.9089,
-      "step": 31394
-    },
-    {
-      "epoch": 16.372881355932204,
-      "grad_norm": 1.4651354551315308,
-      "learning_rate": 6.895979899497488e-05,
-      "loss": 5.3474,
-      "step": 31395
-    },
-    {
-      "epoch": 16.373402868318124,
-      "grad_norm": 1.5753048658370972,
-      "learning_rate": 6.895879396984924e-05,
-      "loss": 5.4624,
-      "step": 31396
-    },
-    {
-      "epoch": 16.373924380704043,
-      "grad_norm": 1.5409475564956665,
-      "learning_rate": 6.895778894472362e-05,
-      "loss": 5.441,
-      "step": 31397
-    },
-    {
-      "epoch": 16.374445893089963,
-      "grad_norm": 1.583217740058899,
-      "learning_rate": 6.895678391959799e-05,
-      "loss": 5.0521,
-      "step": 31398
-    },
-    {
-      "epoch": 16.37496740547588,
-      "grad_norm": 1.517371654510498,
-      "learning_rate": 6.895577889447236e-05,
-      "loss": 5.1928,
-      "step": 31399
-    },
-    {
-      "epoch": 16.375488917861798,
-      "grad_norm": 1.4381568431854248,
-      "learning_rate": 6.895477386934673e-05,
-      "loss": 5.4591,
-      "step": 31400
-    },
-    {
-      "epoch": 16.376010430247717,
-      "grad_norm": 1.605342149734497,
-      "learning_rate": 6.89537688442211e-05,
-      "loss": 5.3103,
-      "step": 31401
-    },
-    {
-      "epoch": 16.376531942633637,
-      "grad_norm": 1.516343355178833,
-      "learning_rate": 6.895276381909548e-05,
-      "loss": 5.1973,
-      "step": 31402
-    },
-    {
-      "epoch": 16.377053455019556,
-      "grad_norm": 1.3715795278549194,
-      "learning_rate": 6.895175879396986e-05,
-      "loss": 5.7206,
-      "step": 31403
-    },
-    {
-      "epoch": 16.377574967405476,
-      "grad_norm": 1.5293593406677246,
-      "learning_rate": 6.895075376884423e-05,
-      "loss": 5.1948,
-      "step": 31404
-    },
-    {
-      "epoch": 16.378096479791395,
-      "grad_norm": 1.4066822528839111,
-      "learning_rate": 6.89497487437186e-05,
-      "loss": 5.4395,
-      "step": 31405
-    },
-    {
-      "epoch": 16.378617992177315,
-      "grad_norm": 1.5539097785949707,
-      "learning_rate": 6.894874371859297e-05,
-      "loss": 5.1586,
-      "step": 31406
-    },
-    {
-      "epoch": 16.379139504563234,
-      "grad_norm": 1.551988124847412,
-      "learning_rate": 6.894773869346735e-05,
-      "loss": 5.2223,
-      "step": 31407
-    },
-    {
-      "epoch": 16.379661016949154,
-      "grad_norm": 1.5352592468261719,
-      "learning_rate": 6.894673366834171e-05,
-      "loss": 5.1946,
-      "step": 31408
-    },
-    {
-      "epoch": 16.380182529335073,
-      "grad_norm": 1.414577841758728,
-      "learning_rate": 6.894572864321607e-05,
-      "loss": 5.5527,
-      "step": 31409
-    },
-    {
-      "epoch": 16.380704041720993,
-      "grad_norm": 1.4834299087524414,
-      "learning_rate": 6.894472361809045e-05,
-      "loss": 5.0849,
-      "step": 31410
-    },
-    {
-      "epoch": 16.38122555410691,
-      "grad_norm": 1.4784510135650635,
-      "learning_rate": 6.894371859296483e-05,
-      "loss": 5.3561,
-      "step": 31411
-    },
-    {
-      "epoch": 16.381747066492828,
-      "grad_norm": 1.4112548828125,
-      "learning_rate": 6.894271356783921e-05,
-      "loss": 5.5563,
-      "step": 31412
-    },
-    {
-      "epoch": 16.382268578878747,
-      "grad_norm": 1.5239983797073364,
-      "learning_rate": 6.894170854271357e-05,
-      "loss": 5.0105,
-      "step": 31413
-    },
-    {
-      "epoch": 16.382790091264667,
-      "grad_norm": 1.468925952911377,
-      "learning_rate": 6.894070351758795e-05,
-      "loss": 4.8571,
-      "step": 31414
-    },
-    {
-      "epoch": 16.383311603650586,
-      "grad_norm": 1.5784952640533447,
-      "learning_rate": 6.893969849246231e-05,
-      "loss": 5.5561,
-      "step": 31415
-    },
-    {
-      "epoch": 16.383833116036506,
-      "grad_norm": 1.476928472518921,
-      "learning_rate": 6.893869346733669e-05,
-      "loss": 5.7645,
-      "step": 31416
-    },
-    {
-      "epoch": 16.384354628422425,
-      "grad_norm": 1.6360756158828735,
-      "learning_rate": 6.893768844221106e-05,
-      "loss": 5.4363,
-      "step": 31417
-    },
-    {
-      "epoch": 16.384876140808345,
-      "grad_norm": 1.6301159858703613,
-      "learning_rate": 6.893668341708543e-05,
-      "loss": 4.9735,
-      "step": 31418
-    },
-    {
-      "epoch": 16.385397653194264,
-      "grad_norm": 1.6102248430252075,
-      "learning_rate": 6.89356783919598e-05,
-      "loss": 4.9493,
-      "step": 31419
-    },
-    {
-      "epoch": 16.385919165580184,
-      "grad_norm": 1.5386136770248413,
-      "learning_rate": 6.893467336683418e-05,
-      "loss": 5.6489,
-      "step": 31420
-    },
-    {
-      "epoch": 16.386440677966103,
-      "grad_norm": 1.2797887325286865,
-      "learning_rate": 6.893366834170854e-05,
-      "loss": 5.309,
-      "step": 31421
-    },
-    {
-      "epoch": 16.386962190352023,
-      "grad_norm": 1.583454966545105,
-      "learning_rate": 6.893266331658292e-05,
-      "loss": 4.9988,
-      "step": 31422
-    },
-    {
-      "epoch": 16.38748370273794,
-      "grad_norm": 1.712294340133667,
-      "learning_rate": 6.89316582914573e-05,
-      "loss": 4.8986,
-      "step": 31423
-    },
-    {
-      "epoch": 16.388005215123858,
-      "grad_norm": 1.5195013284683228,
-      "learning_rate": 6.893065326633166e-05,
-      "loss": 5.0471,
-      "step": 31424
-    },
-    {
-      "epoch": 16.388526727509777,
-      "grad_norm": 1.3878997564315796,
-      "learning_rate": 6.892964824120604e-05,
-      "loss": 5.5395,
-      "step": 31425
-    },
-    {
-      "epoch": 16.389048239895697,
-      "grad_norm": 1.3800928592681885,
-      "learning_rate": 6.89286432160804e-05,
-      "loss": 5.6013,
-      "step": 31426
-    },
-    {
-      "epoch": 16.389569752281616,
-      "grad_norm": 1.5995478630065918,
-      "learning_rate": 6.892763819095478e-05,
-      "loss": 4.9784,
-      "step": 31427
-    },
-    {
-      "epoch": 16.390091264667536,
-      "grad_norm": 1.3562729358673096,
-      "learning_rate": 6.892663316582914e-05,
-      "loss": 5.0309,
-      "step": 31428
-    },
-    {
-      "epoch": 16.390612777053455,
-      "grad_norm": 1.5492995977401733,
-      "learning_rate": 6.892562814070352e-05,
-      "loss": 5.2819,
-      "step": 31429
-    },
-    {
-      "epoch": 16.391134289439375,
-      "grad_norm": 1.4751415252685547,
-      "learning_rate": 6.892462311557789e-05,
-      "loss": 5.2379,
-      "step": 31430
-    },
-    {
-      "epoch": 16.391655801825294,
-      "grad_norm": 1.4896831512451172,
-      "learning_rate": 6.892361809045226e-05,
-      "loss": 5.256,
-      "step": 31431
-    },
-    {
-      "epoch": 16.392177314211214,
-      "grad_norm": 1.4806556701660156,
-      "learning_rate": 6.892261306532664e-05,
-      "loss": 5.7211,
-      "step": 31432
-    },
-    {
-      "epoch": 16.392698826597133,
-      "grad_norm": 1.4367865324020386,
-      "learning_rate": 6.892160804020102e-05,
-      "loss": 5.3799,
-      "step": 31433
-    },
-    {
-      "epoch": 16.39322033898305,
-      "grad_norm": 1.418871521949768,
-      "learning_rate": 6.892060301507538e-05,
-      "loss": 5.7837,
-      "step": 31434
-    },
-    {
-      "epoch": 16.39374185136897,
-      "grad_norm": 1.6299155950546265,
-      "learning_rate": 6.891959798994975e-05,
-      "loss": 4.8882,
-      "step": 31435
-    },
-    {
-      "epoch": 16.394263363754888,
-      "grad_norm": 1.4075229167938232,
-      "learning_rate": 6.891859296482413e-05,
-      "loss": 5.7941,
-      "step": 31436
-    },
-    {
-      "epoch": 16.394784876140807,
-      "grad_norm": 1.3940610885620117,
-      "learning_rate": 6.891758793969849e-05,
-      "loss": 5.6929,
-      "step": 31437
-    },
-    {
-      "epoch": 16.395306388526727,
-      "grad_norm": 1.3852134943008423,
-      "learning_rate": 6.891658291457287e-05,
-      "loss": 5.4437,
-      "step": 31438
-    },
-    {
-      "epoch": 16.395827900912646,
-      "grad_norm": 1.3431754112243652,
-      "learning_rate": 6.891557788944723e-05,
-      "loss": 5.7947,
-      "step": 31439
-    },
-    {
-      "epoch": 16.396349413298566,
-      "grad_norm": 1.6359868049621582,
-      "learning_rate": 6.891457286432161e-05,
-      "loss": 5.1503,
-      "step": 31440
-    },
-    {
-      "epoch": 16.396870925684485,
-      "grad_norm": 1.4573370218276978,
-      "learning_rate": 6.891356783919597e-05,
-      "loss": 5.7116,
-      "step": 31441
-    },
-    {
-      "epoch": 16.397392438070405,
-      "grad_norm": 1.5254086256027222,
-      "learning_rate": 6.891256281407035e-05,
-      "loss": 4.6615,
-      "step": 31442
-    },
-    {
-      "epoch": 16.397913950456324,
-      "grad_norm": 1.5613716840744019,
-      "learning_rate": 6.891155778894473e-05,
-      "loss": 5.4738,
-      "step": 31443
-    },
-    {
-      "epoch": 16.398435462842244,
-      "grad_norm": 1.6009904146194458,
-      "learning_rate": 6.891055276381911e-05,
-      "loss": 4.7937,
-      "step": 31444
-    },
-    {
-      "epoch": 16.398956975228163,
-      "grad_norm": 1.5517284870147705,
-      "learning_rate": 6.890954773869347e-05,
-      "loss": 4.9301,
-      "step": 31445
-    },
-    {
-      "epoch": 16.39947848761408,
-      "grad_norm": 1.5141171216964722,
-      "learning_rate": 6.890854271356785e-05,
-      "loss": 4.9411,
-      "step": 31446
-    },
-    {
-      "epoch": 16.4,
-      "grad_norm": 1.571169137954712,
-      "learning_rate": 6.890753768844221e-05,
-      "loss": 5.3504,
-      "step": 31447
-    },
-    {
-      "epoch": 16.400521512385918,
-      "grad_norm": 1.451866626739502,
-      "learning_rate": 6.890653266331658e-05,
-      "loss": 5.5038,
-      "step": 31448
-    },
-    {
-      "epoch": 16.401043024771838,
-      "grad_norm": 1.521067500114441,
-      "learning_rate": 6.890552763819096e-05,
-      "loss": 5.4683,
-      "step": 31449
-    },
-    {
-      "epoch": 16.401564537157757,
-      "grad_norm": 1.4772270917892456,
-      "learning_rate": 6.890452261306532e-05,
-      "loss": 5.6837,
-      "step": 31450
-    },
-    {
-      "epoch": 16.402086049543676,
-      "grad_norm": 1.4776654243469238,
-      "learning_rate": 6.89035175879397e-05,
-      "loss": 5.5615,
-      "step": 31451
-    },
-    {
-      "epoch": 16.402607561929596,
-      "grad_norm": 1.711431622505188,
-      "learning_rate": 6.890251256281406e-05,
-      "loss": 5.1658,
-      "step": 31452
-    },
-    {
-      "epoch": 16.403129074315515,
-      "grad_norm": 1.5539606809616089,
-      "learning_rate": 6.890150753768844e-05,
-      "loss": 5.3862,
-      "step": 31453
-    },
-    {
-      "epoch": 16.403650586701435,
-      "grad_norm": 1.3859986066818237,
-      "learning_rate": 6.890050251256282e-05,
-      "loss": 5.5677,
-      "step": 31454
-    },
-    {
-      "epoch": 16.404172099087354,
-      "grad_norm": 1.4754540920257568,
-      "learning_rate": 6.88994974874372e-05,
-      "loss": 5.6371,
-      "step": 31455
-    },
-    {
-      "epoch": 16.404693611473274,
-      "grad_norm": 1.5024789571762085,
-      "learning_rate": 6.889849246231156e-05,
-      "loss": 5.5614,
-      "step": 31456
-    },
-    {
-      "epoch": 16.405215123859193,
-      "grad_norm": 1.418412446975708,
-      "learning_rate": 6.889748743718594e-05,
-      "loss": 5.2999,
-      "step": 31457
-    },
-    {
-      "epoch": 16.40573663624511,
-      "grad_norm": 1.7404556274414062,
-      "learning_rate": 6.88964824120603e-05,
-      "loss": 5.1911,
-      "step": 31458
-    },
-    {
-      "epoch": 16.40625814863103,
-      "grad_norm": 1.6103993654251099,
-      "learning_rate": 6.889547738693468e-05,
-      "loss": 5.0801,
-      "step": 31459
-    },
-    {
-      "epoch": 16.406779661016948,
-      "grad_norm": 1.5212409496307373,
-      "learning_rate": 6.889447236180905e-05,
-      "loss": 5.2465,
-      "step": 31460
-    },
-    {
-      "epoch": 16.407301173402868,
-      "grad_norm": 1.6127387285232544,
-      "learning_rate": 6.889346733668341e-05,
-      "loss": 4.6567,
-      "step": 31461
-    },
-    {
-      "epoch": 16.407822685788787,
-      "grad_norm": 1.4135453701019287,
-      "learning_rate": 6.889246231155779e-05,
-      "loss": 5.605,
-      "step": 31462
-    },
-    {
-      "epoch": 16.408344198174706,
-      "grad_norm": 1.6409058570861816,
-      "learning_rate": 6.889145728643217e-05,
-      "loss": 5.256,
-      "step": 31463
-    },
-    {
-      "epoch": 16.408865710560626,
-      "grad_norm": 1.487801432609558,
-      "learning_rate": 6.889045226130654e-05,
-      "loss": 5.2905,
-      "step": 31464
-    },
-    {
-      "epoch": 16.409387222946545,
-      "grad_norm": 1.4383563995361328,
-      "learning_rate": 6.888944723618091e-05,
-      "loss": 5.1185,
-      "step": 31465
-    },
-    {
-      "epoch": 16.409908735332465,
-      "grad_norm": 1.4105162620544434,
-      "learning_rate": 6.888844221105529e-05,
-      "loss": 5.7692,
-      "step": 31466
-    },
-    {
-      "epoch": 16.410430247718384,
-      "grad_norm": 1.4209829568862915,
-      "learning_rate": 6.888743718592965e-05,
-      "loss": 5.6434,
-      "step": 31467
-    },
-    {
-      "epoch": 16.410951760104304,
-      "grad_norm": 1.5452382564544678,
-      "learning_rate": 6.888643216080403e-05,
-      "loss": 5.3587,
-      "step": 31468
-    },
-    {
-      "epoch": 16.411473272490223,
-      "grad_norm": 1.4586119651794434,
-      "learning_rate": 6.888542713567839e-05,
-      "loss": 5.1039,
-      "step": 31469
-    },
-    {
-      "epoch": 16.41199478487614,
-      "grad_norm": 1.6741055250167847,
-      "learning_rate": 6.888442211055277e-05,
-      "loss": 5.4133,
-      "step": 31470
-    },
-    {
-      "epoch": 16.41251629726206,
-      "grad_norm": 1.5768665075302124,
-      "learning_rate": 6.888341708542713e-05,
-      "loss": 4.9325,
-      "step": 31471
-    },
-    {
-      "epoch": 16.413037809647978,
-      "grad_norm": 1.415107011795044,
-      "learning_rate": 6.888241206030151e-05,
-      "loss": 5.8905,
-      "step": 31472
-    },
-    {
-      "epoch": 16.413559322033898,
-      "grad_norm": 1.3954713344573975,
-      "learning_rate": 6.888140703517588e-05,
-      "loss": 5.6566,
-      "step": 31473
-    },
-    {
-      "epoch": 16.414080834419817,
-      "grad_norm": 1.372789978981018,
-      "learning_rate": 6.888040201005025e-05,
-      "loss": 5.2211,
-      "step": 31474
-    },
-    {
-      "epoch": 16.414602346805736,
-      "grad_norm": 1.552462100982666,
-      "learning_rate": 6.887939698492463e-05,
-      "loss": 5.346,
-      "step": 31475
-    },
-    {
-      "epoch": 16.415123859191656,
-      "grad_norm": 1.5293842554092407,
-      "learning_rate": 6.8878391959799e-05,
-      "loss": 5.4013,
-      "step": 31476
-    },
-    {
-      "epoch": 16.415645371577575,
-      "grad_norm": 1.515419602394104,
-      "learning_rate": 6.887738693467337e-05,
-      "loss": 5.5721,
-      "step": 31477
-    },
-    {
-      "epoch": 16.416166883963495,
-      "grad_norm": 1.4344475269317627,
-      "learning_rate": 6.887638190954774e-05,
-      "loss": 5.0241,
-      "step": 31478
-    },
-    {
-      "epoch": 16.416688396349414,
-      "grad_norm": 1.5662590265274048,
-      "learning_rate": 6.887537688442212e-05,
-      "loss": 5.1152,
-      "step": 31479
-    },
-    {
-      "epoch": 16.417209908735334,
-      "grad_norm": 1.4616304636001587,
-      "learning_rate": 6.887437185929648e-05,
-      "loss": 5.1677,
-      "step": 31480
-    },
-    {
-      "epoch": 16.417731421121253,
-      "grad_norm": 1.4107712507247925,
-      "learning_rate": 6.887336683417086e-05,
-      "loss": 5.4475,
-      "step": 31481
-    },
-    {
-      "epoch": 16.41825293350717,
-      "grad_norm": 1.4588507413864136,
-      "learning_rate": 6.887236180904522e-05,
-      "loss": 5.5389,
-      "step": 31482
-    },
-    {
-      "epoch": 16.41877444589309,
-      "grad_norm": 1.4740833044052124,
-      "learning_rate": 6.88713567839196e-05,
-      "loss": 5.6419,
-      "step": 31483
-    },
-    {
-      "epoch": 16.419295958279008,
-      "grad_norm": 1.5671931505203247,
-      "learning_rate": 6.887035175879398e-05,
-      "loss": 5.5535,
-      "step": 31484
-    },
-    {
-      "epoch": 16.419817470664928,
-      "grad_norm": 1.4411901235580444,
-      "learning_rate": 6.886934673366836e-05,
-      "loss": 5.3024,
-      "step": 31485
-    },
-    {
-      "epoch": 16.420338983050847,
-      "grad_norm": 1.4808135032653809,
-      "learning_rate": 6.886834170854272e-05,
-      "loss": 5.4568,
-      "step": 31486
-    },
-    {
-      "epoch": 16.420860495436767,
-      "grad_norm": 1.4360417127609253,
-      "learning_rate": 6.88673366834171e-05,
-      "loss": 5.414,
-      "step": 31487
-    },
-    {
-      "epoch": 16.421382007822686,
-      "grad_norm": 1.5471729040145874,
-      "learning_rate": 6.886633165829146e-05,
-      "loss": 5.4119,
-      "step": 31488
-    },
-    {
-      "epoch": 16.421903520208605,
-      "grad_norm": 1.4682108163833618,
-      "learning_rate": 6.886532663316583e-05,
-      "loss": 5.5782,
-      "step": 31489
-    },
-    {
-      "epoch": 16.422425032594525,
-      "grad_norm": 1.5334874391555786,
-      "learning_rate": 6.88643216080402e-05,
-      "loss": 5.3761,
-      "step": 31490
-    },
-    {
-      "epoch": 16.422946544980444,
-      "grad_norm": 1.433405876159668,
-      "learning_rate": 6.886331658291457e-05,
-      "loss": 5.3645,
-      "step": 31491
-    },
-    {
-      "epoch": 16.423468057366364,
-      "grad_norm": 1.5292413234710693,
-      "learning_rate": 6.886231155778895e-05,
-      "loss": 4.7656,
-      "step": 31492
-    },
-    {
-      "epoch": 16.423989569752283,
-      "grad_norm": 1.3334449529647827,
-      "learning_rate": 6.886130653266331e-05,
-      "loss": 4.7488,
-      "step": 31493
-    },
-    {
-      "epoch": 16.4245110821382,
-      "grad_norm": 1.5337753295898438,
-      "learning_rate": 6.886030150753769e-05,
-      "loss": 5.2678,
-      "step": 31494
-    },
-    {
-      "epoch": 16.42503259452412,
-      "grad_norm": 1.4876986742019653,
-      "learning_rate": 6.885929648241207e-05,
-      "loss": 5.0128,
-      "step": 31495
-    },
-    {
-      "epoch": 16.425554106910038,
-      "grad_norm": 1.3829442262649536,
-      "learning_rate": 6.885829145728644e-05,
-      "loss": 5.0891,
-      "step": 31496
-    },
-    {
-      "epoch": 16.426075619295958,
-      "grad_norm": 1.4518061876296997,
-      "learning_rate": 6.885728643216081e-05,
-      "loss": 5.3715,
-      "step": 31497
-    },
-    {
-      "epoch": 16.426597131681877,
-      "grad_norm": 1.4462257623672485,
-      "learning_rate": 6.885628140703519e-05,
-      "loss": 5.269,
-      "step": 31498
-    },
-    {
-      "epoch": 16.427118644067797,
-      "grad_norm": 1.5157294273376465,
-      "learning_rate": 6.885527638190955e-05,
-      "loss": 5.0394,
-      "step": 31499
-    },
-    {
-      "epoch": 16.427640156453716,
-      "grad_norm": 1.5038143396377563,
-      "learning_rate": 6.885427135678393e-05,
-      "loss": 5.3398,
-      "step": 31500
-    },
-    {
-      "epoch": 16.428161668839635,
-      "grad_norm": 1.4919135570526123,
-      "learning_rate": 6.885326633165829e-05,
-      "loss": 5.6225,
-      "step": 31501
-    },
-    {
-      "epoch": 16.428683181225555,
-      "grad_norm": 1.521572470664978,
-      "learning_rate": 6.885226130653266e-05,
-      "loss": 5.0184,
-      "step": 31502
-    },
-    {
-      "epoch": 16.429204693611474,
-      "grad_norm": 1.44117271900177,
-      "learning_rate": 6.885125628140703e-05,
-      "loss": 5.6783,
-      "step": 31503
-    },
-    {
-      "epoch": 16.429726205997394,
-      "grad_norm": 1.4048638343811035,
-      "learning_rate": 6.885025125628141e-05,
-      "loss": 5.509,
-      "step": 31504
-    },
-    {
-      "epoch": 16.430247718383313,
-      "grad_norm": 1.4751418828964233,
-      "learning_rate": 6.884924623115579e-05,
-      "loss": 5.6158,
-      "step": 31505
-    },
-    {
-      "epoch": 16.43076923076923,
-      "grad_norm": 1.458517074584961,
-      "learning_rate": 6.884824120603015e-05,
-      "loss": 5.2798,
-      "step": 31506
-    },
-    {
-      "epoch": 16.43129074315515,
-      "grad_norm": 1.4607861042022705,
-      "learning_rate": 6.884723618090453e-05,
-      "loss": 5.2851,
-      "step": 31507
-    },
-    {
-      "epoch": 16.431812255541068,
-      "grad_norm": 1.5317684412002563,
-      "learning_rate": 6.88462311557789e-05,
-      "loss": 5.2471,
-      "step": 31508
-    },
-    {
-      "epoch": 16.432333767926988,
-      "grad_norm": 1.369711995124817,
-      "learning_rate": 6.884522613065327e-05,
-      "loss": 5.4488,
-      "step": 31509
-    },
-    {
-      "epoch": 16.432855280312907,
-      "grad_norm": 1.482417106628418,
-      "learning_rate": 6.884422110552764e-05,
-      "loss": 5.3076,
-      "step": 31510
-    },
-    {
-      "epoch": 16.433376792698827,
-      "grad_norm": 1.4428211450576782,
-      "learning_rate": 6.884321608040202e-05,
-      "loss": 5.5823,
-      "step": 31511
-    },
-    {
-      "epoch": 16.433898305084746,
-      "grad_norm": 1.3518507480621338,
-      "learning_rate": 6.884221105527638e-05,
-      "loss": 5.1812,
-      "step": 31512
-    },
-    {
-      "epoch": 16.434419817470665,
-      "grad_norm": 1.415927529335022,
-      "learning_rate": 6.884120603015076e-05,
-      "loss": 5.2734,
-      "step": 31513
-    },
-    {
-      "epoch": 16.434941329856585,
-      "grad_norm": 1.5061949491500854,
-      "learning_rate": 6.884020100502512e-05,
-      "loss": 5.3903,
-      "step": 31514
-    },
-    {
-      "epoch": 16.435462842242504,
-      "grad_norm": 1.4957412481307983,
-      "learning_rate": 6.88391959798995e-05,
-      "loss": 5.3112,
-      "step": 31515
-    },
-    {
-      "epoch": 16.435984354628424,
-      "grad_norm": 1.5200008153915405,
-      "learning_rate": 6.883819095477388e-05,
-      "loss": 4.8825,
-      "step": 31516
-    },
-    {
-      "epoch": 16.43650586701434,
-      "grad_norm": 1.4909648895263672,
-      "learning_rate": 6.883718592964824e-05,
-      "loss": 5.3189,
-      "step": 31517
-    },
-    {
-      "epoch": 16.43702737940026,
-      "grad_norm": 1.4690457582473755,
-      "learning_rate": 6.883618090452262e-05,
-      "loss": 5.1718,
-      "step": 31518
-    },
-    {
-      "epoch": 16.43754889178618,
-      "grad_norm": 1.480153203010559,
-      "learning_rate": 6.883517587939698e-05,
-      "loss": 5.2545,
-      "step": 31519
-    },
-    {
-      "epoch": 16.438070404172098,
-      "grad_norm": 1.497173547744751,
-      "learning_rate": 6.883417085427136e-05,
-      "loss": 5.3692,
-      "step": 31520
-    },
-    {
-      "epoch": 16.438591916558018,
-      "grad_norm": 1.5277981758117676,
-      "learning_rate": 6.883316582914573e-05,
-      "loss": 5.4073,
-      "step": 31521
-    },
-    {
-      "epoch": 16.439113428943937,
-      "grad_norm": 1.4134913682937622,
-      "learning_rate": 6.88321608040201e-05,
-      "loss": 5.6578,
-      "step": 31522
-    },
-    {
-      "epoch": 16.439634941329857,
-      "grad_norm": 1.5068495273590088,
-      "learning_rate": 6.883115577889447e-05,
-      "loss": 5.1101,
-      "step": 31523
-    },
-    {
-      "epoch": 16.440156453715776,
-      "grad_norm": 1.4709407091140747,
-      "learning_rate": 6.883015075376885e-05,
-      "loss": 5.037,
-      "step": 31524
-    },
-    {
-      "epoch": 16.440677966101696,
-      "grad_norm": 1.849890947341919,
-      "learning_rate": 6.882914572864322e-05,
-      "loss": 5.1029,
-      "step": 31525
-    },
-    {
-      "epoch": 16.441199478487615,
-      "grad_norm": 1.775009036064148,
-      "learning_rate": 6.88281407035176e-05,
-      "loss": 5.2951,
-      "step": 31526
-    },
-    {
-      "epoch": 16.441720990873534,
-      "grad_norm": 1.4482699632644653,
-      "learning_rate": 6.882713567839197e-05,
-      "loss": 5.2993,
-      "step": 31527
-    },
-    {
-      "epoch": 16.442242503259454,
-      "grad_norm": 1.449783205986023,
-      "learning_rate": 6.882613065326633e-05,
-      "loss": 5.3403,
-      "step": 31528
-    },
-    {
-      "epoch": 16.442764015645373,
-      "grad_norm": 1.3784775733947754,
-      "learning_rate": 6.882512562814071e-05,
-      "loss": 5.4974,
-      "step": 31529
-    },
-    {
-      "epoch": 16.44328552803129,
-      "grad_norm": 1.4608654975891113,
-      "learning_rate": 6.882412060301507e-05,
-      "loss": 5.0796,
-      "step": 31530
-    },
-    {
-      "epoch": 16.44380704041721,
-      "grad_norm": 1.6479482650756836,
-      "learning_rate": 6.882311557788945e-05,
-      "loss": 4.964,
-      "step": 31531
-    },
-    {
-      "epoch": 16.444328552803128,
-      "grad_norm": 1.5592666864395142,
-      "learning_rate": 6.882211055276382e-05,
-      "loss": 5.2746,
-      "step": 31532
-    },
-    {
-      "epoch": 16.444850065189048,
-      "grad_norm": 1.4192198514938354,
-      "learning_rate": 6.882110552763819e-05,
-      "loss": 5.6051,
-      "step": 31533
-    },
-    {
-      "epoch": 16.445371577574967,
-      "grad_norm": 1.4706310033798218,
-      "learning_rate": 6.882010050251256e-05,
-      "loss": 5.3567,
-      "step": 31534
-    },
-    {
-      "epoch": 16.445893089960887,
-      "grad_norm": 1.5065410137176514,
-      "learning_rate": 6.881909547738694e-05,
-      "loss": 4.942,
-      "step": 31535
-    },
-    {
-      "epoch": 16.446414602346806,
-      "grad_norm": 1.901411533355713,
-      "learning_rate": 6.881809045226131e-05,
-      "loss": 5.5669,
-      "step": 31536
-    },
-    {
-      "epoch": 16.446936114732726,
-      "grad_norm": 1.504901647567749,
-      "learning_rate": 6.881708542713569e-05,
-      "loss": 5.6176,
-      "step": 31537
-    },
-    {
-      "epoch": 16.447457627118645,
-      "grad_norm": 1.476576566696167,
-      "learning_rate": 6.881608040201005e-05,
-      "loss": 5.4429,
-      "step": 31538
-    },
-    {
-      "epoch": 16.447979139504564,
-      "grad_norm": 1.427633285522461,
-      "learning_rate": 6.881507537688443e-05,
-      "loss": 5.0558,
-      "step": 31539
-    },
-    {
-      "epoch": 16.448500651890484,
-      "grad_norm": 1.6396561861038208,
-      "learning_rate": 6.88140703517588e-05,
-      "loss": 5.291,
-      "step": 31540
-    },
-    {
-      "epoch": 16.4490221642764,
-      "grad_norm": 1.4416791200637817,
-      "learning_rate": 6.881306532663316e-05,
-      "loss": 5.5062,
-      "step": 31541
-    },
-    {
-      "epoch": 16.44954367666232,
-      "grad_norm": 1.4100072383880615,
-      "learning_rate": 6.881206030150754e-05,
-      "loss": 5.4616,
-      "step": 31542
-    },
-    {
-      "epoch": 16.45006518904824,
-      "grad_norm": 1.4648839235305786,
-      "learning_rate": 6.88110552763819e-05,
-      "loss": 5.6051,
-      "step": 31543
-    },
-    {
-      "epoch": 16.45058670143416,
-      "grad_norm": 1.5893884897232056,
-      "learning_rate": 6.881005025125628e-05,
-      "loss": 5.4,
-      "step": 31544
-    },
-    {
-      "epoch": 16.451108213820078,
-      "grad_norm": 1.5761593580245972,
-      "learning_rate": 6.880904522613066e-05,
-      "loss": 5.6205,
-      "step": 31545
-    },
-    {
-      "epoch": 16.451629726205997,
-      "grad_norm": 1.454721450805664,
-      "learning_rate": 6.880804020100504e-05,
-      "loss": 5.6521,
-      "step": 31546
-    },
-    {
-      "epoch": 16.452151238591917,
-      "grad_norm": 1.478990077972412,
-      "learning_rate": 6.88070351758794e-05,
-      "loss": 5.4025,
-      "step": 31547
-    },
-    {
-      "epoch": 16.452672750977836,
-      "grad_norm": 1.4893959760665894,
-      "learning_rate": 6.880603015075378e-05,
-      "loss": 5.2702,
-      "step": 31548
-    },
-    {
-      "epoch": 16.453194263363756,
-      "grad_norm": 1.5743579864501953,
-      "learning_rate": 6.880502512562814e-05,
-      "loss": 5.5533,
-      "step": 31549
-    },
-    {
-      "epoch": 16.453715775749675,
-      "grad_norm": 1.4575482606887817,
-      "learning_rate": 6.880402010050252e-05,
-      "loss": 5.2077,
-      "step": 31550
-    },
-    {
-      "epoch": 16.454237288135594,
-      "grad_norm": 1.4635769128799438,
-      "learning_rate": 6.880301507537689e-05,
-      "loss": 5.4693,
-      "step": 31551
-    },
-    {
-      "epoch": 16.454758800521514,
-      "grad_norm": 1.5621908903121948,
-      "learning_rate": 6.880201005025126e-05,
-      "loss": 5.3861,
-      "step": 31552
-    },
-    {
-      "epoch": 16.45528031290743,
-      "grad_norm": 1.5089327096939087,
-      "learning_rate": 6.880100502512563e-05,
-      "loss": 4.9017,
-      "step": 31553
-    },
-    {
-      "epoch": 16.45580182529335,
-      "grad_norm": 1.6257036924362183,
-      "learning_rate": 6.879999999999999e-05,
-      "loss": 4.569,
-      "step": 31554
-    },
-    {
-      "epoch": 16.45632333767927,
-      "grad_norm": 1.484552025794983,
-      "learning_rate": 6.879899497487437e-05,
-      "loss": 5.1339,
-      "step": 31555
-    },
-    {
-      "epoch": 16.45684485006519,
-      "grad_norm": 1.5543324947357178,
-      "learning_rate": 6.879798994974875e-05,
-      "loss": 4.6298,
-      "step": 31556
-    },
-    {
-      "epoch": 16.457366362451108,
-      "grad_norm": 1.6917449235916138,
-      "learning_rate": 6.879698492462313e-05,
-      "loss": 5.2793,
-      "step": 31557
-    },
-    {
-      "epoch": 16.457887874837027,
-      "grad_norm": 1.4682215452194214,
-      "learning_rate": 6.879597989949749e-05,
-      "loss": 5.5908,
-      "step": 31558
-    },
-    {
-      "epoch": 16.458409387222947,
-      "grad_norm": 1.4689205884933472,
-      "learning_rate": 6.879497487437187e-05,
-      "loss": 5.5199,
-      "step": 31559
-    },
-    {
-      "epoch": 16.458930899608866,
-      "grad_norm": 1.529730200767517,
-      "learning_rate": 6.879396984924623e-05,
-      "loss": 5.353,
-      "step": 31560
-    },
-    {
-      "epoch": 16.459452411994786,
-      "grad_norm": 1.4763551950454712,
-      "learning_rate": 6.879296482412061e-05,
-      "loss": 4.8555,
-      "step": 31561
-    },
-    {
-      "epoch": 16.459973924380705,
-      "grad_norm": 1.4406198263168335,
-      "learning_rate": 6.879195979899497e-05,
-      "loss": 5.6944,
-      "step": 31562
-    },
-    {
-      "epoch": 16.460495436766625,
-      "grad_norm": 1.5634723901748657,
-      "learning_rate": 6.879095477386935e-05,
-      "loss": 5.3357,
-      "step": 31563
-    },
-    {
-      "epoch": 16.461016949152544,
-      "grad_norm": 1.5407304763793945,
-      "learning_rate": 6.878994974874372e-05,
-      "loss": 4.6626,
-      "step": 31564
-    },
-    {
-      "epoch": 16.46153846153846,
-      "grad_norm": 1.4920364618301392,
-      "learning_rate": 6.87889447236181e-05,
-      "loss": 5.1402,
-      "step": 31565
-    },
-    {
-      "epoch": 16.46205997392438,
-      "grad_norm": 1.4621562957763672,
-      "learning_rate": 6.878793969849247e-05,
-      "loss": 5.7839,
-      "step": 31566
-    },
-    {
-      "epoch": 16.4625814863103,
-      "grad_norm": 1.511030912399292,
-      "learning_rate": 6.878693467336685e-05,
-      "loss": 4.8457,
-      "step": 31567
-    },
-    {
-      "epoch": 16.46310299869622,
-      "grad_norm": 1.4316483736038208,
-      "learning_rate": 6.878592964824121e-05,
-      "loss": 5.3682,
-      "step": 31568
-    },
-    {
-      "epoch": 16.463624511082138,
-      "grad_norm": 1.436259150505066,
-      "learning_rate": 6.878492462311558e-05,
-      "loss": 5.5691,
-      "step": 31569
-    },
-    {
-      "epoch": 16.464146023468057,
-      "grad_norm": 1.4188871383666992,
-      "learning_rate": 6.878391959798996e-05,
-      "loss": 5.322,
-      "step": 31570
-    },
-    {
-      "epoch": 16.464667535853977,
-      "grad_norm": 1.469642162322998,
-      "learning_rate": 6.878291457286432e-05,
-      "loss": 5.3124,
-      "step": 31571
-    },
-    {
-      "epoch": 16.465189048239896,
-      "grad_norm": 1.3788864612579346,
-      "learning_rate": 6.87819095477387e-05,
-      "loss": 5.4108,
-      "step": 31572
-    },
-    {
-      "epoch": 16.465710560625816,
-      "grad_norm": 1.413347601890564,
-      "learning_rate": 6.878090452261306e-05,
-      "loss": 5.392,
-      "step": 31573
-    },
-    {
-      "epoch": 16.466232073011735,
-      "grad_norm": 1.5033146142959595,
-      "learning_rate": 6.877989949748744e-05,
-      "loss": 5.6446,
-      "step": 31574
-    },
-    {
-      "epoch": 16.466753585397655,
-      "grad_norm": 1.4008598327636719,
-      "learning_rate": 6.87788944723618e-05,
-      "loss": 5.468,
-      "step": 31575
-    },
-    {
-      "epoch": 16.467275097783574,
-      "grad_norm": 1.3982207775115967,
-      "learning_rate": 6.877788944723618e-05,
-      "loss": 5.7116,
-      "step": 31576
-    },
-    {
-      "epoch": 16.46779661016949,
-      "grad_norm": 1.687286376953125,
-      "learning_rate": 6.877688442211056e-05,
-      "loss": 4.5928,
-      "step": 31577
-    },
-    {
-      "epoch": 16.46831812255541,
-      "grad_norm": 1.5650418996810913,
-      "learning_rate": 6.877587939698494e-05,
-      "loss": 4.7744,
-      "step": 31578
-    },
-    {
-      "epoch": 16.46883963494133,
-      "grad_norm": 1.534508466720581,
-      "learning_rate": 6.87748743718593e-05,
-      "loss": 5.6012,
-      "step": 31579
-    },
-    {
-      "epoch": 16.46936114732725,
-      "grad_norm": 1.5389155149459839,
-      "learning_rate": 6.877386934673368e-05,
-      "loss": 5.225,
-      "step": 31580
-    },
-    {
-      "epoch": 16.469882659713168,
-      "grad_norm": 1.5734031200408936,
-      "learning_rate": 6.877286432160804e-05,
-      "loss": 4.7741,
-      "step": 31581
-    },
-    {
-      "epoch": 16.470404172099087,
-      "grad_norm": 1.4974159002304077,
-      "learning_rate": 6.877185929648241e-05,
-      "loss": 5.8797,
-      "step": 31582
-    },
-    {
-      "epoch": 16.470925684485007,
-      "grad_norm": 1.5481483936309814,
-      "learning_rate": 6.877085427135679e-05,
-      "loss": 5.3109,
-      "step": 31583
-    },
-    {
-      "epoch": 16.471447196870926,
-      "grad_norm": 1.4955310821533203,
-      "learning_rate": 6.876984924623115e-05,
-      "loss": 4.8503,
-      "step": 31584
-    },
-    {
-      "epoch": 16.471968709256846,
-      "grad_norm": 1.4631552696228027,
-      "learning_rate": 6.876884422110553e-05,
-      "loss": 4.8486,
-      "step": 31585
-    },
-    {
-      "epoch": 16.472490221642765,
-      "grad_norm": 1.4453177452087402,
-      "learning_rate": 6.87678391959799e-05,
-      "loss": 5.6215,
-      "step": 31586
-    },
-    {
-      "epoch": 16.473011734028685,
-      "grad_norm": 1.4452402591705322,
-      "learning_rate": 6.876683417085428e-05,
-      "loss": 5.3852,
-      "step": 31587
-    },
-    {
-      "epoch": 16.473533246414604,
-      "grad_norm": 1.63565993309021,
-      "learning_rate": 6.876582914572865e-05,
-      "loss": 5.0481,
-      "step": 31588
-    },
-    {
-      "epoch": 16.47405475880052,
-      "grad_norm": 1.3847382068634033,
-      "learning_rate": 6.876482412060303e-05,
-      "loss": 5.4186,
-      "step": 31589
-    },
-    {
-      "epoch": 16.47457627118644,
-      "grad_norm": 1.3949429988861084,
-      "learning_rate": 6.876381909547739e-05,
-      "loss": 5.5296,
-      "step": 31590
-    },
-    {
-      "epoch": 16.47509778357236,
-      "grad_norm": 1.39295494556427,
-      "learning_rate": 6.876281407035177e-05,
-      "loss": 5.3235,
-      "step": 31591
-    },
-    {
-      "epoch": 16.47561929595828,
-      "grad_norm": 1.4669668674468994,
-      "learning_rate": 6.876180904522613e-05,
-      "loss": 5.5396,
-      "step": 31592
-    },
-    {
-      "epoch": 16.476140808344198,
-      "grad_norm": 1.4244694709777832,
-      "learning_rate": 6.876080402010051e-05,
-      "loss": 5.7812,
-      "step": 31593
-    },
-    {
-      "epoch": 16.476662320730117,
-      "grad_norm": 1.4643081426620483,
-      "learning_rate": 6.875979899497487e-05,
-      "loss": 5.0858,
-      "step": 31594
-    },
-    {
-      "epoch": 16.477183833116037,
-      "grad_norm": 1.4949363470077515,
-      "learning_rate": 6.875879396984924e-05,
-      "loss": 5.53,
-      "step": 31595
-    },
-    {
-      "epoch": 16.477705345501956,
-      "grad_norm": 1.5006904602050781,
-      "learning_rate": 6.875778894472362e-05,
-      "loss": 5.5472,
-      "step": 31596
-    },
-    {
-      "epoch": 16.478226857887876,
-      "grad_norm": 1.476503610610962,
-      "learning_rate": 6.8756783919598e-05,
-      "loss": 5.2905,
-      "step": 31597
-    },
-    {
-      "epoch": 16.478748370273795,
-      "grad_norm": 1.4219965934753418,
-      "learning_rate": 6.875577889447237e-05,
-      "loss": 5.1907,
-      "step": 31598
-    },
-    {
-      "epoch": 16.479269882659715,
-      "grad_norm": 1.3973084688186646,
-      "learning_rate": 6.875477386934674e-05,
-      "loss": 5.7347,
-      "step": 31599
-    },
-    {
-      "epoch": 16.479791395045634,
-      "grad_norm": 1.47309410572052,
-      "learning_rate": 6.875376884422111e-05,
-      "loss": 5.6656,
-      "step": 31600
-    },
-    {
-      "epoch": 16.48031290743155,
-      "grad_norm": 1.490279197692871,
-      "learning_rate": 6.875276381909548e-05,
-      "loss": 5.2151,
-      "step": 31601
-    },
-    {
-      "epoch": 16.48083441981747,
-      "grad_norm": 1.5514237880706787,
-      "learning_rate": 6.875175879396986e-05,
-      "loss": 5.463,
-      "step": 31602
-    },
-    {
-      "epoch": 16.48135593220339,
-      "grad_norm": 1.5942668914794922,
-      "learning_rate": 6.875075376884422e-05,
-      "loss": 5.5689,
-      "step": 31603
-    },
-    {
-      "epoch": 16.48187744458931,
-      "grad_norm": 1.45650053024292,
-      "learning_rate": 6.87497487437186e-05,
-      "loss": 5.5334,
-      "step": 31604
-    },
-    {
-      "epoch": 16.482398956975228,
-      "grad_norm": 1.5800390243530273,
-      "learning_rate": 6.874874371859296e-05,
-      "loss": 4.9292,
-      "step": 31605
-    },
-    {
-      "epoch": 16.482920469361147,
-      "grad_norm": 1.4555641412734985,
-      "learning_rate": 6.874773869346734e-05,
-      "loss": 5.7411,
-      "step": 31606
-    },
-    {
-      "epoch": 16.483441981747067,
-      "grad_norm": 1.4024330377578735,
-      "learning_rate": 6.874673366834172e-05,
-      "loss": 5.8458,
-      "step": 31607
-    },
-    {
-      "epoch": 16.483963494132986,
-      "grad_norm": 1.5250940322875977,
-      "learning_rate": 6.874572864321608e-05,
-      "loss": 5.4737,
-      "step": 31608
-    },
-    {
-      "epoch": 16.484485006518906,
-      "grad_norm": 1.3987886905670166,
-      "learning_rate": 6.874472361809046e-05,
-      "loss": 5.6416,
-      "step": 31609
-    },
-    {
-      "epoch": 16.485006518904825,
-      "grad_norm": 1.3914990425109863,
-      "learning_rate": 6.874371859296482e-05,
-      "loss": 5.5346,
-      "step": 31610
-    },
-    {
-      "epoch": 16.485528031290745,
-      "grad_norm": 1.4400948286056519,
-      "learning_rate": 6.87427135678392e-05,
-      "loss": 5.4628,
-      "step": 31611
-    },
-    {
-      "epoch": 16.486049543676664,
-      "grad_norm": 1.6058839559555054,
-      "learning_rate": 6.874170854271357e-05,
-      "loss": 5.2398,
-      "step": 31612
-    },
-    {
-      "epoch": 16.48657105606258,
-      "grad_norm": 1.4019441604614258,
-      "learning_rate": 6.874070351758794e-05,
-      "loss": 5.2063,
-      "step": 31613
-    },
-    {
-      "epoch": 16.4870925684485,
-      "grad_norm": 1.5777907371520996,
-      "learning_rate": 6.873969849246231e-05,
-      "loss": 5.6346,
-      "step": 31614
-    },
-    {
-      "epoch": 16.48761408083442,
-      "grad_norm": 1.4229921102523804,
-      "learning_rate": 6.873869346733669e-05,
-      "loss": 5.1604,
-      "step": 31615
-    },
-    {
-      "epoch": 16.48813559322034,
-      "grad_norm": 1.5164068937301636,
-      "learning_rate": 6.873768844221105e-05,
-      "loss": 5.2713,
-      "step": 31616
-    },
-    {
-      "epoch": 16.488657105606258,
-      "grad_norm": 1.4925905466079712,
-      "learning_rate": 6.873668341708543e-05,
-      "loss": 5.5712,
-      "step": 31617
-    },
-    {
-      "epoch": 16.489178617992177,
-      "grad_norm": 1.559501051902771,
-      "learning_rate": 6.873567839195981e-05,
-      "loss": 5.3,
-      "step": 31618
-    },
-    {
-      "epoch": 16.489700130378097,
-      "grad_norm": 1.5119789838790894,
-      "learning_rate": 6.873467336683418e-05,
-      "loss": 4.8141,
-      "step": 31619
-    },
-    {
-      "epoch": 16.490221642764016,
-      "grad_norm": 1.4569865465164185,
-      "learning_rate": 6.873366834170855e-05,
-      "loss": 5.4556,
-      "step": 31620
-    },
-    {
-      "epoch": 16.490743155149936,
-      "grad_norm": 1.5367742776870728,
-      "learning_rate": 6.873266331658291e-05,
-      "loss": 5.4558,
-      "step": 31621
-    },
-    {
-      "epoch": 16.491264667535855,
-      "grad_norm": 1.5353158712387085,
-      "learning_rate": 6.873165829145729e-05,
-      "loss": 5.5721,
-      "step": 31622
-    },
-    {
-      "epoch": 16.491786179921775,
-      "grad_norm": 1.468582034111023,
-      "learning_rate": 6.873065326633166e-05,
-      "loss": 5.6102,
-      "step": 31623
-    },
-    {
-      "epoch": 16.49230769230769,
-      "grad_norm": 1.4470860958099365,
-      "learning_rate": 6.872964824120603e-05,
-      "loss": 5.4036,
-      "step": 31624
-    },
-    {
-      "epoch": 16.49282920469361,
-      "grad_norm": 1.4472681283950806,
-      "learning_rate": 6.87286432160804e-05,
-      "loss": 5.1613,
-      "step": 31625
-    },
-    {
-      "epoch": 16.49335071707953,
-      "grad_norm": 1.4498213529586792,
-      "learning_rate": 6.872763819095478e-05,
-      "loss": 5.4749,
-      "step": 31626
-    },
-    {
-      "epoch": 16.49387222946545,
-      "grad_norm": 1.4506962299346924,
-      "learning_rate": 6.872663316582914e-05,
-      "loss": 5.349,
-      "step": 31627
-    },
-    {
-      "epoch": 16.49439374185137,
-      "grad_norm": 1.6514379978179932,
-      "learning_rate": 6.872562814070352e-05,
-      "loss": 5.1324,
-      "step": 31628
-    },
-    {
-      "epoch": 16.494915254237288,
-      "grad_norm": 1.5837290287017822,
-      "learning_rate": 6.87246231155779e-05,
-      "loss": 5.0866,
-      "step": 31629
-    },
-    {
-      "epoch": 16.495436766623207,
-      "grad_norm": 1.587170958518982,
-      "learning_rate": 6.872361809045227e-05,
-      "loss": 5.3105,
-      "step": 31630
-    },
-    {
-      "epoch": 16.495958279009127,
-      "grad_norm": 1.4991482496261597,
-      "learning_rate": 6.872261306532664e-05,
-      "loss": 5.7244,
-      "step": 31631
-    },
-    {
-      "epoch": 16.496479791395046,
-      "grad_norm": 1.4571247100830078,
-      "learning_rate": 6.872160804020102e-05,
-      "loss": 5.2473,
-      "step": 31632
-    },
-    {
-      "epoch": 16.497001303780966,
-      "grad_norm": 1.486035943031311,
-      "learning_rate": 6.872060301507538e-05,
-      "loss": 5.7817,
-      "step": 31633
-    },
-    {
-      "epoch": 16.497522816166885,
-      "grad_norm": 1.519545555114746,
-      "learning_rate": 6.871959798994974e-05,
-      "loss": 5.2991,
-      "step": 31634
-    },
-    {
-      "epoch": 16.498044328552805,
-      "grad_norm": 1.5607945919036865,
-      "learning_rate": 6.871859296482412e-05,
-      "loss": 5.5814,
-      "step": 31635
-    },
-    {
-      "epoch": 16.49856584093872,
-      "grad_norm": 1.6242576837539673,
-      "learning_rate": 6.871758793969849e-05,
-      "loss": 5.298,
-      "step": 31636
-    },
-    {
-      "epoch": 16.49908735332464,
-      "grad_norm": 1.6042903661727905,
-      "learning_rate": 6.871658291457286e-05,
-      "loss": 5.102,
-      "step": 31637
-    },
-    {
-      "epoch": 16.49960886571056,
-      "grad_norm": 1.515305757522583,
-      "learning_rate": 6.871557788944724e-05,
-      "loss": 5.5231,
-      "step": 31638
-    },
-    {
-      "epoch": 16.50013037809648,
-      "grad_norm": 1.3865524530410767,
-      "learning_rate": 6.871457286432162e-05,
-      "loss": 5.2012,
-      "step": 31639
-    },
-    {
-      "epoch": 16.5006518904824,
-      "grad_norm": 1.4839369058609009,
-      "learning_rate": 6.871356783919598e-05,
-      "loss": 5.0312,
-      "step": 31640
-    },
-    {
-      "epoch": 16.501173402868318,
-      "grad_norm": 1.5297791957855225,
-      "learning_rate": 6.871256281407036e-05,
-      "loss": 5.2636,
-      "step": 31641
-    },
-    {
-      "epoch": 16.501694915254237,
-      "grad_norm": 1.4687018394470215,
-      "learning_rate": 6.871155778894473e-05,
-      "loss": 5.5641,
-      "step": 31642
-    },
-    {
-      "epoch": 16.502216427640157,
-      "grad_norm": 1.4543172121047974,
-      "learning_rate": 6.87105527638191e-05,
-      "loss": 5.5943,
-      "step": 31643
-    },
-    {
-      "epoch": 16.502737940026076,
-      "grad_norm": 1.4142889976501465,
-      "learning_rate": 6.870954773869347e-05,
-      "loss": 5.82,
-      "step": 31644
-    },
-    {
-      "epoch": 16.503259452411996,
-      "grad_norm": 1.508291482925415,
-      "learning_rate": 6.870854271356785e-05,
-      "loss": 5.2766,
-      "step": 31645
-    },
-    {
-      "epoch": 16.503780964797915,
-      "grad_norm": 1.4976019859313965,
-      "learning_rate": 6.870753768844221e-05,
-      "loss": 5.4104,
-      "step": 31646
-    },
-    {
-      "epoch": 16.504302477183835,
-      "grad_norm": 1.324669361114502,
-      "learning_rate": 6.870653266331657e-05,
-      "loss": 5.8712,
-      "step": 31647
-    },
-    {
-      "epoch": 16.50482398956975,
-      "grad_norm": 1.5160701274871826,
-      "learning_rate": 6.870552763819095e-05,
-      "loss": 5.4888,
-      "step": 31648
-    },
-    {
-      "epoch": 16.50534550195567,
-      "grad_norm": 1.4812867641448975,
-      "learning_rate": 6.870452261306533e-05,
-      "loss": 4.9726,
-      "step": 31649
-    },
-    {
-      "epoch": 16.50586701434159,
-      "grad_norm": 1.542826771736145,
-      "learning_rate": 6.870351758793971e-05,
-      "loss": 5.1768,
-      "step": 31650
-    },
-    {
-      "epoch": 16.50638852672751,
-      "grad_norm": 1.3804315328598022,
-      "learning_rate": 6.870251256281407e-05,
-      "loss": 5.819,
-      "step": 31651
-    },
-    {
-      "epoch": 16.50691003911343,
-      "grad_norm": 1.6290278434753418,
-      "learning_rate": 6.870150753768845e-05,
-      "loss": 4.9031,
-      "step": 31652
-    },
-    {
-      "epoch": 16.507431551499348,
-      "grad_norm": 1.4632799625396729,
-      "learning_rate": 6.870050251256281e-05,
-      "loss": 4.7473,
-      "step": 31653
-    },
-    {
-      "epoch": 16.507953063885267,
-      "grad_norm": 1.371263861656189,
-      "learning_rate": 6.869949748743719e-05,
-      "loss": 5.7699,
-      "step": 31654
-    },
-    {
-      "epoch": 16.508474576271187,
-      "grad_norm": 1.4458378553390503,
-      "learning_rate": 6.869849246231156e-05,
-      "loss": 4.9653,
-      "step": 31655
-    },
-    {
-      "epoch": 16.508996088657106,
-      "grad_norm": 1.578955054283142,
-      "learning_rate": 6.869748743718593e-05,
-      "loss": 5.3714,
-      "step": 31656
-    },
-    {
-      "epoch": 16.509517601043026,
-      "grad_norm": 1.540718913078308,
-      "learning_rate": 6.86964824120603e-05,
-      "loss": 5.1384,
-      "step": 31657
-    },
-    {
-      "epoch": 16.510039113428945,
-      "grad_norm": 1.3718857765197754,
-      "learning_rate": 6.869547738693468e-05,
-      "loss": 5.486,
-      "step": 31658
-    },
-    {
-      "epoch": 16.510560625814865,
-      "grad_norm": 1.5581218004226685,
-      "learning_rate": 6.869447236180905e-05,
-      "loss": 5.2964,
-      "step": 31659
-    },
-    {
-      "epoch": 16.51108213820078,
-      "grad_norm": 1.4355064630508423,
-      "learning_rate": 6.869346733668343e-05,
-      "loss": 5.7848,
-      "step": 31660
-    },
-    {
-      "epoch": 16.5116036505867,
-      "grad_norm": 1.4028218984603882,
-      "learning_rate": 6.86924623115578e-05,
-      "loss": 5.5471,
-      "step": 31661
-    },
-    {
-      "epoch": 16.51212516297262,
-      "grad_norm": 1.448853850364685,
-      "learning_rate": 6.869145728643216e-05,
-      "loss": 5.1281,
-      "step": 31662
-    },
-    {
-      "epoch": 16.51264667535854,
-      "grad_norm": 1.516817569732666,
-      "learning_rate": 6.869045226130654e-05,
-      "loss": 5.5699,
-      "step": 31663
-    },
-    {
-      "epoch": 16.51316818774446,
-      "grad_norm": 1.427974820137024,
-      "learning_rate": 6.86894472361809e-05,
-      "loss": 5.3756,
-      "step": 31664
-    },
-    {
-      "epoch": 16.513689700130378,
-      "grad_norm": 1.7629843950271606,
-      "learning_rate": 6.868844221105528e-05,
-      "loss": 4.2512,
-      "step": 31665
-    },
-    {
-      "epoch": 16.514211212516297,
-      "grad_norm": 1.4452588558197021,
-      "learning_rate": 6.868743718592964e-05,
-      "loss": 5.838,
-      "step": 31666
-    },
-    {
-      "epoch": 16.514732724902217,
-      "grad_norm": 1.4854373931884766,
-      "learning_rate": 6.868643216080402e-05,
-      "loss": 5.5963,
-      "step": 31667
-    },
-    {
-      "epoch": 16.515254237288136,
-      "grad_norm": 1.4263315200805664,
-      "learning_rate": 6.868542713567839e-05,
-      "loss": 5.5701,
-      "step": 31668
-    },
-    {
-      "epoch": 16.515775749674056,
-      "grad_norm": 1.4765028953552246,
-      "learning_rate": 6.868442211055276e-05,
-      "loss": 4.9583,
-      "step": 31669
-    },
-    {
-      "epoch": 16.516297262059975,
-      "grad_norm": 1.428579568862915,
-      "learning_rate": 6.868341708542714e-05,
-      "loss": 5.4824,
-      "step": 31670
-    },
-    {
-      "epoch": 16.516818774445895,
-      "grad_norm": 1.4761269092559814,
-      "learning_rate": 6.868241206030152e-05,
-      "loss": 5.4125,
-      "step": 31671
-    },
-    {
-      "epoch": 16.51734028683181,
-      "grad_norm": 1.4993075132369995,
-      "learning_rate": 6.868140703517588e-05,
-      "loss": 5.4718,
-      "step": 31672
-    },
-    {
-      "epoch": 16.51786179921773,
-      "grad_norm": 1.4499928951263428,
-      "learning_rate": 6.868040201005026e-05,
-      "loss": 5.6031,
-      "step": 31673
-    },
-    {
-      "epoch": 16.51838331160365,
-      "grad_norm": 1.531223177909851,
-      "learning_rate": 6.867939698492463e-05,
-      "loss": 5.2911,
-      "step": 31674
-    },
-    {
-      "epoch": 16.51890482398957,
-      "grad_norm": 1.598676085472107,
-      "learning_rate": 6.867839195979899e-05,
-      "loss": 5.3124,
-      "step": 31675
-    },
-    {
-      "epoch": 16.51942633637549,
-      "grad_norm": 1.5528815984725952,
-      "learning_rate": 6.867738693467337e-05,
-      "loss": 5.0958,
-      "step": 31676
-    },
-    {
-      "epoch": 16.519947848761408,
-      "grad_norm": 1.3907850980758667,
-      "learning_rate": 6.867638190954773e-05,
-      "loss": 5.0246,
-      "step": 31677
-    },
-    {
-      "epoch": 16.520469361147327,
-      "grad_norm": 1.472508430480957,
-      "learning_rate": 6.867537688442211e-05,
-      "loss": 5.3032,
-      "step": 31678
-    },
-    {
-      "epoch": 16.520990873533247,
-      "grad_norm": 1.495861530303955,
-      "learning_rate": 6.867437185929649e-05,
-      "loss": 5.312,
-      "step": 31679
-    },
-    {
-      "epoch": 16.521512385919166,
-      "grad_norm": 1.441819190979004,
-      "learning_rate": 6.867336683417087e-05,
-      "loss": 5.3273,
-      "step": 31680
-    },
-    {
-      "epoch": 16.522033898305086,
-      "grad_norm": 1.4437059164047241,
-      "learning_rate": 6.867236180904523e-05,
-      "loss": 5.2403,
-      "step": 31681
-    },
-    {
-      "epoch": 16.522555410691005,
-      "grad_norm": 1.392138957977295,
-      "learning_rate": 6.867135678391961e-05,
-      "loss": 5.4369,
-      "step": 31682
-    },
-    {
-      "epoch": 16.523076923076925,
-      "grad_norm": 1.5668007135391235,
-      "learning_rate": 6.867035175879397e-05,
-      "loss": 5.1764,
-      "step": 31683
-    },
-    {
-      "epoch": 16.52359843546284,
-      "grad_norm": 1.4668787717819214,
-      "learning_rate": 6.866934673366835e-05,
-      "loss": 5.6985,
-      "step": 31684
-    },
-    {
-      "epoch": 16.52411994784876,
-      "grad_norm": 1.4631197452545166,
-      "learning_rate": 6.866834170854271e-05,
-      "loss": 5.3862,
-      "step": 31685
-    },
-    {
-      "epoch": 16.52464146023468,
-      "grad_norm": 1.4383333921432495,
-      "learning_rate": 6.866733668341709e-05,
-      "loss": 5.2176,
-      "step": 31686
-    },
-    {
-      "epoch": 16.5251629726206,
-      "grad_norm": 1.436447262763977,
-      "learning_rate": 6.866633165829146e-05,
-      "loss": 5.3326,
-      "step": 31687
-    },
-    {
-      "epoch": 16.52568448500652,
-      "grad_norm": 1.4406789541244507,
-      "learning_rate": 6.866532663316582e-05,
-      "loss": 5.1135,
-      "step": 31688
-    },
-    {
-      "epoch": 16.526205997392438,
-      "grad_norm": 1.4602856636047363,
-      "learning_rate": 6.86643216080402e-05,
-      "loss": 5.529,
-      "step": 31689
-    },
-    {
-      "epoch": 16.526727509778357,
-      "grad_norm": 1.4394398927688599,
-      "learning_rate": 6.866331658291458e-05,
-      "loss": 5.5929,
-      "step": 31690
-    },
-    {
-      "epoch": 16.527249022164277,
-      "grad_norm": 1.4382332563400269,
-      "learning_rate": 6.866231155778895e-05,
-      "loss": 5.321,
-      "step": 31691
-    },
-    {
-      "epoch": 16.527770534550196,
-      "grad_norm": 1.4467424154281616,
-      "learning_rate": 6.866130653266332e-05,
-      "loss": 5.231,
-      "step": 31692
-    },
-    {
-      "epoch": 16.528292046936116,
-      "grad_norm": 1.4332661628723145,
-      "learning_rate": 6.86603015075377e-05,
-      "loss": 5.4074,
-      "step": 31693
-    },
-    {
-      "epoch": 16.528813559322035,
-      "grad_norm": 1.4333457946777344,
-      "learning_rate": 6.865929648241206e-05,
-      "loss": 5.7832,
-      "step": 31694
-    },
-    {
-      "epoch": 16.529335071707955,
-      "grad_norm": 1.48317551612854,
-      "learning_rate": 6.865829145728644e-05,
-      "loss": 5.2843,
-      "step": 31695
-    },
-    {
-      "epoch": 16.52985658409387,
-      "grad_norm": 1.5993813276290894,
-      "learning_rate": 6.86572864321608e-05,
-      "loss": 4.8812,
-      "step": 31696
-    },
-    {
-      "epoch": 16.53037809647979,
-      "grad_norm": 1.410029411315918,
-      "learning_rate": 6.865628140703518e-05,
-      "loss": 5.5033,
-      "step": 31697
-    },
-    {
-      "epoch": 16.53089960886571,
-      "grad_norm": 1.4207673072814941,
-      "learning_rate": 6.865527638190955e-05,
-      "loss": 5.4977,
-      "step": 31698
-    },
-    {
-      "epoch": 16.53142112125163,
-      "grad_norm": 1.4541505575180054,
-      "learning_rate": 6.865427135678392e-05,
-      "loss": 5.2732,
-      "step": 31699
-    },
-    {
-      "epoch": 16.53194263363755,
-      "grad_norm": 1.4071919918060303,
-      "learning_rate": 6.86532663316583e-05,
-      "loss": 5.2229,
-      "step": 31700
-    },
-    {
-      "epoch": 16.532464146023468,
-      "grad_norm": 1.4518100023269653,
-      "learning_rate": 6.865226130653267e-05,
-      "loss": 5.4892,
-      "step": 31701
-    },
-    {
-      "epoch": 16.532985658409387,
-      "grad_norm": 1.5936201810836792,
-      "learning_rate": 6.865125628140704e-05,
-      "loss": 5.3814,
-      "step": 31702
-    },
-    {
-      "epoch": 16.533507170795307,
-      "grad_norm": 1.5361013412475586,
-      "learning_rate": 6.865025125628141e-05,
-      "loss": 5.3024,
-      "step": 31703
-    },
-    {
-      "epoch": 16.534028683181226,
-      "grad_norm": 1.519648551940918,
-      "learning_rate": 6.864924623115579e-05,
-      "loss": 5.388,
-      "step": 31704
-    },
-    {
-      "epoch": 16.534550195567146,
-      "grad_norm": 1.5176059007644653,
-      "learning_rate": 6.864824120603015e-05,
-      "loss": 5.115,
-      "step": 31705
-    },
-    {
-      "epoch": 16.535071707953065,
-      "grad_norm": 1.396774411201477,
-      "learning_rate": 6.864723618090453e-05,
-      "loss": 5.4013,
-      "step": 31706
-    },
-    {
-      "epoch": 16.53559322033898,
-      "grad_norm": 1.5269083976745605,
-      "learning_rate": 6.864623115577889e-05,
-      "loss": 5.1149,
-      "step": 31707
-    },
-    {
-      "epoch": 16.5361147327249,
-      "grad_norm": 1.428453803062439,
-      "learning_rate": 6.864522613065327e-05,
-      "loss": 5.1302,
-      "step": 31708
-    },
-    {
-      "epoch": 16.53663624511082,
-      "grad_norm": 1.5300451517105103,
-      "learning_rate": 6.864422110552763e-05,
-      "loss": 5.2371,
-      "step": 31709
-    },
-    {
-      "epoch": 16.53715775749674,
-      "grad_norm": 1.4250613451004028,
-      "learning_rate": 6.864321608040201e-05,
-      "loss": 5.5342,
-      "step": 31710
-    },
-    {
-      "epoch": 16.53767926988266,
-      "grad_norm": 1.4323896169662476,
-      "learning_rate": 6.864221105527639e-05,
-      "loss": 5.5604,
-      "step": 31711
-    },
-    {
-      "epoch": 16.53820078226858,
-      "grad_norm": 1.5507025718688965,
-      "learning_rate": 6.864120603015077e-05,
-      "loss": 5.4034,
-      "step": 31712
-    },
-    {
-      "epoch": 16.538722294654498,
-      "grad_norm": 1.5455169677734375,
-      "learning_rate": 6.864020100502513e-05,
-      "loss": 5.4965,
-      "step": 31713
-    },
-    {
-      "epoch": 16.539243807040418,
-      "grad_norm": 1.4942984580993652,
-      "learning_rate": 6.86391959798995e-05,
-      "loss": 5.1636,
-      "step": 31714
-    },
-    {
-      "epoch": 16.539765319426337,
-      "grad_norm": 1.3936381340026855,
-      "learning_rate": 6.863819095477387e-05,
-      "loss": 5.2961,
-      "step": 31715
-    },
-    {
-      "epoch": 16.540286831812256,
-      "grad_norm": 1.4135288000106812,
-      "learning_rate": 6.863718592964824e-05,
-      "loss": 5.8134,
-      "step": 31716
-    },
-    {
-      "epoch": 16.540808344198176,
-      "grad_norm": 1.4060463905334473,
-      "learning_rate": 6.863618090452262e-05,
-      "loss": 5.7609,
-      "step": 31717
-    },
-    {
-      "epoch": 16.541329856584095,
-      "grad_norm": 1.4403984546661377,
-      "learning_rate": 6.863517587939698e-05,
-      "loss": 5.2868,
-      "step": 31718
-    },
-    {
-      "epoch": 16.541851368970015,
-      "grad_norm": 1.4015544652938843,
-      "learning_rate": 6.863417085427136e-05,
-      "loss": 5.2716,
-      "step": 31719
-    },
-    {
-      "epoch": 16.54237288135593,
-      "grad_norm": Infinity,
-      "learning_rate": 6.863417085427136e-05,
-      "loss": 5.1176,
-      "step": 31720
-    },
-    {
-      "epoch": 16.54289439374185,
-      "grad_norm": 1.5108873844146729,
-      "learning_rate": 6.863316582914574e-05,
-      "loss": 5.2165,
-      "step": 31721
-    },
-    {
-      "epoch": 16.54341590612777,
-      "grad_norm": 1.3917577266693115,
-      "learning_rate": 6.863216080402011e-05,
-      "loss": 5.7737,
-      "step": 31722
-    },
-    {
-      "epoch": 16.54393741851369,
-      "grad_norm": 1.4226608276367188,
-      "learning_rate": 6.863115577889448e-05,
-      "loss": 5.5183,
-      "step": 31723
-    },
-    {
-      "epoch": 16.54445893089961,
-      "grad_norm": 1.4402449131011963,
-      "learning_rate": 6.863015075376886e-05,
-      "loss": 5.4168,
-      "step": 31724
-    },
-    {
-      "epoch": 16.544980443285528,
-      "grad_norm": 1.5147724151611328,
-      "learning_rate": 6.862914572864322e-05,
-      "loss": 5.1407,
-      "step": 31725
-    },
-    {
-      "epoch": 16.545501955671448,
-      "grad_norm": 1.4132556915283203,
-      "learning_rate": 6.86281407035176e-05,
-      "loss": 4.8144,
-      "step": 31726
-    },
-    {
-      "epoch": 16.546023468057367,
-      "grad_norm": 1.4122813940048218,
-      "learning_rate": 6.862713567839196e-05,
-      "loss": 5.3953,
-      "step": 31727
-    },
-    {
-      "epoch": 16.546544980443286,
-      "grad_norm": 1.4687728881835938,
-      "learning_rate": 6.862613065326633e-05,
-      "loss": 5.2523,
-      "step": 31728
-    },
-    {
-      "epoch": 16.547066492829206,
-      "grad_norm": 1.4519565105438232,
-      "learning_rate": 6.86251256281407e-05,
-      "loss": 5.4453,
-      "step": 31729
-    },
-    {
-      "epoch": 16.547588005215125,
-      "grad_norm": 1.4250752925872803,
-      "learning_rate": 6.862412060301507e-05,
-      "loss": 5.2829,
-      "step": 31730
-    },
-    {
-      "epoch": 16.54810951760104,
-      "grad_norm": 1.4443093538284302,
-      "learning_rate": 6.862311557788945e-05,
-      "loss": 5.2219,
-      "step": 31731
-    },
-    {
-      "epoch": 16.54863102998696,
-      "grad_norm": 1.3728580474853516,
-      "learning_rate": 6.862211055276382e-05,
-      "loss": 4.782,
-      "step": 31732
-    },
-    {
-      "epoch": 16.54915254237288,
-      "grad_norm": 1.443861484527588,
-      "learning_rate": 6.86211055276382e-05,
-      "loss": 5.5466,
-      "step": 31733
-    },
-    {
-      "epoch": 16.5496740547588,
-      "grad_norm": 1.4279615879058838,
-      "learning_rate": 6.862010050251257e-05,
-      "loss": 5.5486,
-      "step": 31734
-    },
-    {
-      "epoch": 16.55019556714472,
-      "grad_norm": 1.4932620525360107,
-      "learning_rate": 6.861909547738694e-05,
-      "loss": 5.5355,
-      "step": 31735
-    },
-    {
-      "epoch": 16.55071707953064,
-      "grad_norm": 1.6095308065414429,
-      "learning_rate": 6.861809045226131e-05,
-      "loss": 5.1158,
-      "step": 31736
-    },
-    {
-      "epoch": 16.551238591916558,
-      "grad_norm": 1.4348841905593872,
-      "learning_rate": 6.861708542713569e-05,
-      "loss": 5.373,
-      "step": 31737
-    },
-    {
-      "epoch": 16.551760104302478,
-      "grad_norm": 1.5494601726531982,
-      "learning_rate": 6.861608040201005e-05,
-      "loss": 5.0645,
-      "step": 31738
-    },
-    {
-      "epoch": 16.552281616688397,
-      "grad_norm": 1.4508315324783325,
-      "learning_rate": 6.861507537688443e-05,
-      "loss": 5.7718,
-      "step": 31739
-    },
-    {
-      "epoch": 16.552803129074317,
-      "grad_norm": 1.4338085651397705,
-      "learning_rate": 6.861407035175879e-05,
-      "loss": 4.6435,
-      "step": 31740
-    },
-    {
-      "epoch": 16.553324641460236,
-      "grad_norm": 1.5107218027114868,
-      "learning_rate": 6.861306532663317e-05,
-      "loss": 5.7011,
-      "step": 31741
-    },
-    {
-      "epoch": 16.553846153846155,
-      "grad_norm": 1.4490885734558105,
-      "learning_rate": 6.861206030150755e-05,
-      "loss": 5.6455,
-      "step": 31742
-    },
-    {
-      "epoch": 16.55436766623207,
-      "grad_norm": 1.3639578819274902,
-      "learning_rate": 6.861105527638191e-05,
-      "loss": 5.404,
-      "step": 31743
-    },
-    {
-      "epoch": 16.55488917861799,
-      "grad_norm": 1.4840247631072998,
-      "learning_rate": 6.861005025125629e-05,
-      "loss": 5.6851,
-      "step": 31744
-    },
-    {
-      "epoch": 16.55541069100391,
-      "grad_norm": 1.4888404607772827,
-      "learning_rate": 6.860904522613065e-05,
-      "loss": 5.2509,
-      "step": 31745
-    },
-    {
-      "epoch": 16.55593220338983,
-      "grad_norm": 1.3964686393737793,
-      "learning_rate": 6.860804020100503e-05,
-      "loss": 5.5534,
-      "step": 31746
-    },
-    {
-      "epoch": 16.55645371577575,
-      "grad_norm": 1.5131151676177979,
-      "learning_rate": 6.86070351758794e-05,
-      "loss": 5.3472,
-      "step": 31747
-    },
-    {
-      "epoch": 16.55697522816167,
-      "grad_norm": 1.5055270195007324,
-      "learning_rate": 6.860603015075377e-05,
-      "loss": 5.476,
-      "step": 31748
-    },
-    {
-      "epoch": 16.557496740547588,
-      "grad_norm": 1.4821758270263672,
-      "learning_rate": 6.860502512562814e-05,
-      "loss": 5.2252,
-      "step": 31749
-    },
-    {
-      "epoch": 16.558018252933508,
-      "grad_norm": 1.5010623931884766,
-      "learning_rate": 6.860402010050252e-05,
-      "loss": 5.5167,
-      "step": 31750
-    },
-    {
-      "epoch": 16.558539765319427,
-      "grad_norm": 1.4742720127105713,
-      "learning_rate": 6.860301507537688e-05,
-      "loss": 5.1283,
-      "step": 31751
-    },
-    {
-      "epoch": 16.559061277705347,
-      "grad_norm": 1.3924617767333984,
-      "learning_rate": 6.860201005025126e-05,
-      "loss": 5.4063,
-      "step": 31752
-    },
-    {
-      "epoch": 16.559582790091266,
-      "grad_norm": 1.415818214416504,
-      "learning_rate": 6.860100502512564e-05,
-      "loss": 5.3374,
-      "step": 31753
-    },
-    {
-      "epoch": 16.560104302477185,
-      "grad_norm": 1.4692749977111816,
-      "learning_rate": 6.860000000000001e-05,
-      "loss": 5.4593,
-      "step": 31754
-    },
-    {
-      "epoch": 16.5606258148631,
-      "grad_norm": 1.503350019454956,
-      "learning_rate": 6.859899497487438e-05,
-      "loss": 5.3759,
-      "step": 31755
-    },
-    {
-      "epoch": 16.56114732724902,
-      "grad_norm": 1.5380640029907227,
-      "learning_rate": 6.859798994974874e-05,
-      "loss": 5.1153,
-      "step": 31756
-    },
-    {
-      "epoch": 16.56166883963494,
-      "grad_norm": 1.4745564460754395,
-      "learning_rate": 6.859698492462312e-05,
-      "loss": 5.1736,
-      "step": 31757
-    },
-    {
-      "epoch": 16.56219035202086,
-      "grad_norm": 1.3950718641281128,
-      "learning_rate": 6.859597989949748e-05,
-      "loss": 5.6979,
-      "step": 31758
-    },
-    {
-      "epoch": 16.56271186440678,
-      "grad_norm": 1.5142191648483276,
-      "learning_rate": 6.859497487437186e-05,
-      "loss": 5.5974,
-      "step": 31759
-    },
-    {
-      "epoch": 16.5632333767927,
-      "grad_norm": 1.4515448808670044,
-      "learning_rate": 6.859396984924623e-05,
-      "loss": 5.5234,
-      "step": 31760
-    },
-    {
-      "epoch": 16.563754889178618,
-      "grad_norm": 1.5559196472167969,
-      "learning_rate": 6.85929648241206e-05,
-      "loss": 5.1787,
-      "step": 31761
-    },
-    {
-      "epoch": 16.564276401564538,
-      "grad_norm": 1.508660078048706,
-      "learning_rate": 6.859195979899498e-05,
-      "loss": 5.0737,
-      "step": 31762
-    },
-    {
-      "epoch": 16.564797913950457,
-      "grad_norm": 1.4668874740600586,
-      "learning_rate": 6.859095477386936e-05,
-      "loss": 5.3153,
-      "step": 31763
-    },
-    {
-      "epoch": 16.565319426336377,
-      "grad_norm": 1.5736151933670044,
-      "learning_rate": 6.858994974874372e-05,
-      "loss": 4.6752,
-      "step": 31764
-    },
-    {
-      "epoch": 16.565840938722296,
-      "grad_norm": 1.3454478979110718,
-      "learning_rate": 6.85889447236181e-05,
-      "loss": 5.2309,
-      "step": 31765
-    },
-    {
-      "epoch": 16.566362451108215,
-      "grad_norm": 1.4460217952728271,
-      "learning_rate": 6.858793969849247e-05,
-      "loss": 5.7528,
-      "step": 31766
-    },
-    {
-      "epoch": 16.56688396349413,
-      "grad_norm": 1.5007386207580566,
-      "learning_rate": 6.858693467336684e-05,
-      "loss": 5.3206,
-      "step": 31767
-    },
-    {
-      "epoch": 16.56740547588005,
-      "grad_norm": 1.4924925565719604,
-      "learning_rate": 6.858592964824121e-05,
-      "loss": 5.3539,
-      "step": 31768
-    },
-    {
-      "epoch": 16.56792698826597,
-      "grad_norm": 1.3316951990127563,
-      "learning_rate": 6.858492462311557e-05,
-      "loss": 5.4666,
-      "step": 31769
-    },
-    {
-      "epoch": 16.56844850065189,
-      "grad_norm": 1.6352949142456055,
-      "learning_rate": 6.858391959798995e-05,
-      "loss": 5.0374,
-      "step": 31770
-    },
-    {
-      "epoch": 16.56897001303781,
-      "grad_norm": 1.4097951650619507,
-      "learning_rate": 6.858291457286432e-05,
-      "loss": 5.5644,
-      "step": 31771
-    },
-    {
-      "epoch": 16.56949152542373,
-      "grad_norm": 1.6226999759674072,
-      "learning_rate": 6.858190954773869e-05,
-      "loss": 5.4272,
-      "step": 31772
-    },
-    {
-      "epoch": 16.570013037809648,
-      "grad_norm": 1.523787260055542,
-      "learning_rate": 6.858090452261307e-05,
-      "loss": 5.4011,
-      "step": 31773
-    },
-    {
-      "epoch": 16.570534550195568,
-      "grad_norm": 1.495197057723999,
-      "learning_rate": 6.857989949748745e-05,
-      "loss": 5.469,
-      "step": 31774
-    },
-    {
-      "epoch": 16.571056062581487,
-      "grad_norm": 1.4608339071273804,
-      "learning_rate": 6.857889447236181e-05,
-      "loss": 5.2652,
-      "step": 31775
-    },
-    {
-      "epoch": 16.571577574967407,
-      "grad_norm": 1.38694167137146,
-      "learning_rate": 6.857788944723619e-05,
-      "loss": 5.7193,
-      "step": 31776
-    },
-    {
-      "epoch": 16.572099087353326,
-      "grad_norm": 1.4634335041046143,
-      "learning_rate": 6.857688442211056e-05,
-      "loss": 5.2301,
-      "step": 31777
-    },
-    {
-      "epoch": 16.572620599739246,
-      "grad_norm": 1.7041175365447998,
-      "learning_rate": 6.857587939698493e-05,
-      "loss": 4.8583,
-      "step": 31778
-    },
-    {
-      "epoch": 16.57314211212516,
-      "grad_norm": 1.4305096864700317,
-      "learning_rate": 6.85748743718593e-05,
-      "loss": 5.3696,
-      "step": 31779
-    },
-    {
-      "epoch": 16.57366362451108,
-      "grad_norm": 1.3821494579315186,
-      "learning_rate": 6.857386934673367e-05,
-      "loss": 5.7692,
-      "step": 31780
-    },
-    {
-      "epoch": 16.574185136897,
-      "grad_norm": 1.4929012060165405,
-      "learning_rate": 6.857286432160804e-05,
-      "loss": 5.4272,
-      "step": 31781
-    },
-    {
-      "epoch": 16.57470664928292,
-      "grad_norm": 1.5246950387954712,
-      "learning_rate": 6.85718592964824e-05,
-      "loss": 4.9595,
-      "step": 31782
-    },
-    {
-      "epoch": 16.57522816166884,
-      "grad_norm": 1.493899941444397,
-      "learning_rate": 6.857085427135678e-05,
-      "loss": 4.7122,
-      "step": 31783
-    },
-    {
-      "epoch": 16.57574967405476,
-      "grad_norm": 1.5034503936767578,
-      "learning_rate": 6.856984924623116e-05,
-      "loss": 4.8799,
-      "step": 31784
-    },
-    {
-      "epoch": 16.576271186440678,
-      "grad_norm": 1.612138271331787,
-      "learning_rate": 6.856884422110554e-05,
-      "loss": 4.921,
-      "step": 31785
-    },
-    {
-      "epoch": 16.576792698826598,
-      "grad_norm": 2.118063449859619,
-      "learning_rate": 6.85678391959799e-05,
-      "loss": 5.0651,
-      "step": 31786
-    },
-    {
-      "epoch": 16.577314211212517,
-      "grad_norm": 1.4699867963790894,
-      "learning_rate": 6.856683417085428e-05,
-      "loss": 5.3677,
-      "step": 31787
-    },
-    {
-      "epoch": 16.577835723598437,
-      "grad_norm": 1.4850114583969116,
-      "learning_rate": 6.856582914572864e-05,
-      "loss": 5.5366,
-      "step": 31788
-    },
-    {
-      "epoch": 16.578357235984356,
-      "grad_norm": 1.4850448369979858,
-      "learning_rate": 6.856482412060302e-05,
-      "loss": 5.2406,
-      "step": 31789
-    },
-    {
-      "epoch": 16.578878748370272,
-      "grad_norm": 1.7758609056472778,
-      "learning_rate": 6.856381909547739e-05,
-      "loss": 5.3107,
-      "step": 31790
-    },
-    {
-      "epoch": 16.57940026075619,
-      "grad_norm": 1.4418960809707642,
-      "learning_rate": 6.856281407035176e-05,
-      "loss": 4.9315,
-      "step": 31791
-    },
-    {
-      "epoch": 16.57992177314211,
-      "grad_norm": 1.474555253982544,
-      "learning_rate": 6.856180904522613e-05,
-      "loss": 5.2297,
-      "step": 31792
-    },
-    {
-      "epoch": 16.58044328552803,
-      "grad_norm": 1.578627586364746,
-      "learning_rate": 6.85608040201005e-05,
-      "loss": 5.4818,
-      "step": 31793
-    },
-    {
-      "epoch": 16.58096479791395,
-      "grad_norm": 1.5467692613601685,
-      "learning_rate": 6.855979899497488e-05,
-      "loss": 4.8844,
-      "step": 31794
-    },
-    {
-      "epoch": 16.58148631029987,
-      "grad_norm": 1.4057316780090332,
-      "learning_rate": 6.855879396984925e-05,
-      "loss": 5.5666,
-      "step": 31795
-    },
-    {
-      "epoch": 16.58200782268579,
-      "grad_norm": 1.4855581521987915,
-      "learning_rate": 6.855778894472363e-05,
-      "loss": 5.443,
-      "step": 31796
-    },
-    {
-      "epoch": 16.58252933507171,
-      "grad_norm": 1.623557448387146,
-      "learning_rate": 6.855678391959799e-05,
-      "loss": 5.1603,
-      "step": 31797
-    },
-    {
-      "epoch": 16.583050847457628,
-      "grad_norm": 1.5114301443099976,
-      "learning_rate": 6.855577889447237e-05,
-      "loss": 5.4549,
-      "step": 31798
-    },
-    {
-      "epoch": 16.583572359843547,
-      "grad_norm": 1.5366238355636597,
-      "learning_rate": 6.855477386934673e-05,
-      "loss": 5.2634,
-      "step": 31799
-    },
-    {
-      "epoch": 16.584093872229467,
-      "grad_norm": 1.384951114654541,
-      "learning_rate": 6.855376884422111e-05,
-      "loss": 5.6739,
-      "step": 31800
-    },
-    {
-      "epoch": 16.584615384615386,
-      "grad_norm": 1.550540566444397,
-      "learning_rate": 6.855276381909547e-05,
-      "loss": 5.6141,
-      "step": 31801
-    },
-    {
-      "epoch": 16.585136897001306,
-      "grad_norm": 1.404815435409546,
-      "learning_rate": 6.855175879396985e-05,
-      "loss": 5.5869,
-      "step": 31802
-    },
-    {
-      "epoch": 16.58565840938722,
-      "grad_norm": 1.5107088088989258,
-      "learning_rate": 6.855075376884422e-05,
-      "loss": 5.2047,
-      "step": 31803
-    },
-    {
-      "epoch": 16.58617992177314,
-      "grad_norm": 1.4218575954437256,
-      "learning_rate": 6.85497487437186e-05,
-      "loss": 5.7635,
-      "step": 31804
-    },
-    {
-      "epoch": 16.58670143415906,
-      "grad_norm": 1.4871801137924194,
-      "learning_rate": 6.854874371859297e-05,
-      "loss": 5.681,
-      "step": 31805
-    },
-    {
-      "epoch": 16.58722294654498,
-      "grad_norm": 1.5488145351409912,
-      "learning_rate": 6.854773869346735e-05,
-      "loss": 4.7306,
-      "step": 31806
-    },
-    {
-      "epoch": 16.5877444589309,
-      "grad_norm": 1.5847187042236328,
-      "learning_rate": 6.854673366834171e-05,
-      "loss": 5.3348,
-      "step": 31807
-    },
-    {
-      "epoch": 16.58826597131682,
-      "grad_norm": 1.498538851737976,
-      "learning_rate": 6.854572864321608e-05,
-      "loss": 4.8845,
-      "step": 31808
-    },
-    {
-      "epoch": 16.58878748370274,
-      "grad_norm": 1.5194839239120483,
-      "learning_rate": 6.854472361809046e-05,
-      "loss": 5.3274,
-      "step": 31809
-    },
-    {
-      "epoch": 16.589308996088658,
-      "grad_norm": 1.4038681983947754,
-      "learning_rate": 6.854371859296482e-05,
-      "loss": 5.5458,
-      "step": 31810
-    },
-    {
-      "epoch": 16.589830508474577,
-      "grad_norm": 1.4093307256698608,
-      "learning_rate": 6.85427135678392e-05,
-      "loss": 5.5869,
-      "step": 31811
-    },
-    {
-      "epoch": 16.590352020860497,
-      "grad_norm": 1.4300788640975952,
-      "learning_rate": 6.854170854271356e-05,
-      "loss": 5.7271,
-      "step": 31812
-    },
-    {
-      "epoch": 16.590873533246416,
-      "grad_norm": 1.4762706756591797,
-      "learning_rate": 6.854070351758794e-05,
-      "loss": 5.7298,
-      "step": 31813
-    },
-    {
-      "epoch": 16.591395045632332,
-      "grad_norm": 1.5354472398757935,
-      "learning_rate": 6.853969849246232e-05,
-      "loss": 5.1725,
-      "step": 31814
-    },
-    {
-      "epoch": 16.59191655801825,
-      "grad_norm": 1.4573270082473755,
-      "learning_rate": 6.85386934673367e-05,
-      "loss": 5.1419,
-      "step": 31815
-    },
-    {
-      "epoch": 16.59243807040417,
-      "grad_norm": 1.5633689165115356,
-      "learning_rate": 6.853768844221106e-05,
-      "loss": 4.5874,
-      "step": 31816
-    },
-    {
-      "epoch": 16.59295958279009,
-      "grad_norm": 1.4894908666610718,
-      "learning_rate": 6.853668341708544e-05,
-      "loss": 5.6387,
-      "step": 31817
-    },
-    {
-      "epoch": 16.59348109517601,
-      "grad_norm": 1.450129508972168,
-      "learning_rate": 6.85356783919598e-05,
-      "loss": 5.2175,
-      "step": 31818
-    },
-    {
-      "epoch": 16.59400260756193,
-      "grad_norm": 1.4928871393203735,
-      "learning_rate": 6.853467336683418e-05,
-      "loss": 5.6575,
-      "step": 31819
-    },
-    {
-      "epoch": 16.59452411994785,
-      "grad_norm": 1.539548397064209,
-      "learning_rate": 6.853366834170854e-05,
-      "loss": 5.3415,
-      "step": 31820
-    },
-    {
-      "epoch": 16.59504563233377,
-      "grad_norm": 1.454768419265747,
-      "learning_rate": 6.853266331658291e-05,
-      "loss": 5.5495,
-      "step": 31821
-    },
-    {
-      "epoch": 16.595567144719688,
-      "grad_norm": 1.416604995727539,
-      "learning_rate": 6.853165829145729e-05,
-      "loss": 5.4415,
-      "step": 31822
-    },
-    {
-      "epoch": 16.596088657105607,
-      "grad_norm": 1.5218690633773804,
-      "learning_rate": 6.853065326633165e-05,
-      "loss": 5.3862,
-      "step": 31823
-    },
-    {
-      "epoch": 16.596610169491527,
-      "grad_norm": 1.491566777229309,
-      "learning_rate": 6.852964824120603e-05,
-      "loss": 5.5571,
-      "step": 31824
-    },
-    {
-      "epoch": 16.597131681877446,
-      "grad_norm": 1.4965934753417969,
-      "learning_rate": 6.85286432160804e-05,
-      "loss": 5.2095,
-      "step": 31825
-    },
-    {
-      "epoch": 16.597653194263362,
-      "grad_norm": 1.4559623003005981,
-      "learning_rate": 6.852763819095478e-05,
-      "loss": 5.0099,
-      "step": 31826
-    },
-    {
-      "epoch": 16.59817470664928,
-      "grad_norm": 1.516855239868164,
-      "learning_rate": 6.852663316582915e-05,
-      "loss": 4.9343,
-      "step": 31827
-    },
-    {
-      "epoch": 16.5986962190352,
-      "grad_norm": 1.47390615940094,
-      "learning_rate": 6.852562814070353e-05,
-      "loss": 5.2223,
-      "step": 31828
-    },
-    {
-      "epoch": 16.59921773142112,
-      "grad_norm": 1.5157479047775269,
-      "learning_rate": 6.852462311557789e-05,
-      "loss": 5.4117,
-      "step": 31829
-    },
-    {
-      "epoch": 16.59973924380704,
-      "grad_norm": 1.399772047996521,
-      "learning_rate": 6.852361809045227e-05,
-      "loss": 5.5028,
-      "step": 31830
-    },
-    {
-      "epoch": 16.60026075619296,
-      "grad_norm": 1.5147318840026855,
-      "learning_rate": 6.852261306532663e-05,
-      "loss": 5.3494,
-      "step": 31831
-    },
-    {
-      "epoch": 16.60078226857888,
-      "grad_norm": 1.4527376890182495,
-      "learning_rate": 6.852160804020101e-05,
-      "loss": 5.5278,
-      "step": 31832
-    },
-    {
-      "epoch": 16.6013037809648,
-      "grad_norm": 1.4492485523223877,
-      "learning_rate": 6.852060301507537e-05,
-      "loss": 5.2507,
-      "step": 31833
-    },
-    {
-      "epoch": 16.601825293350718,
-      "grad_norm": 1.4392709732055664,
-      "learning_rate": 6.851959798994975e-05,
-      "loss": 5.828,
-      "step": 31834
-    },
-    {
-      "epoch": 16.602346805736637,
-      "grad_norm": 1.471445918083191,
-      "learning_rate": 6.851859296482413e-05,
-      "loss": 4.6498,
-      "step": 31835
-    },
-    {
-      "epoch": 16.602868318122557,
-      "grad_norm": 1.6285513639450073,
-      "learning_rate": 6.85175879396985e-05,
-      "loss": 5.1865,
-      "step": 31836
-    },
-    {
-      "epoch": 16.603389830508476,
-      "grad_norm": 1.511054277420044,
-      "learning_rate": 6.851658291457287e-05,
-      "loss": 5.4284,
-      "step": 31837
-    },
-    {
-      "epoch": 16.603911342894392,
-      "grad_norm": 1.5129042863845825,
-      "learning_rate": 6.851557788944724e-05,
-      "loss": 5.0763,
-      "step": 31838
-    },
-    {
-      "epoch": 16.60443285528031,
-      "grad_norm": 1.4672644138336182,
-      "learning_rate": 6.851457286432161e-05,
-      "loss": 5.404,
-      "step": 31839
-    },
-    {
-      "epoch": 16.60495436766623,
-      "grad_norm": 1.4345884323120117,
-      "learning_rate": 6.851356783919598e-05,
-      "loss": 5.5415,
-      "step": 31840
-    },
-    {
-      "epoch": 16.60547588005215,
-      "grad_norm": 1.5286099910736084,
-      "learning_rate": 6.851256281407036e-05,
-      "loss": 5.283,
-      "step": 31841
-    },
-    {
-      "epoch": 16.60599739243807,
-      "grad_norm": 1.470200777053833,
-      "learning_rate": 6.851155778894472e-05,
-      "loss": 5.5557,
-      "step": 31842
-    },
-    {
-      "epoch": 16.60651890482399,
-      "grad_norm": 1.6150331497192383,
-      "learning_rate": 6.85105527638191e-05,
-      "loss": 5.3101,
-      "step": 31843
-    },
-    {
-      "epoch": 16.60704041720991,
-      "grad_norm": 1.3887778520584106,
-      "learning_rate": 6.850954773869346e-05,
-      "loss": 5.8323,
-      "step": 31844
-    },
-    {
-      "epoch": 16.60756192959583,
-      "grad_norm": 1.4820208549499512,
-      "learning_rate": 6.850854271356784e-05,
-      "loss": 5.2583,
-      "step": 31845
-    },
-    {
-      "epoch": 16.608083441981748,
-      "grad_norm": 1.459841251373291,
-      "learning_rate": 6.850753768844222e-05,
-      "loss": 5.004,
-      "step": 31846
-    },
-    {
-      "epoch": 16.608604954367667,
-      "grad_norm": 1.490854024887085,
-      "learning_rate": 6.85065326633166e-05,
-      "loss": 5.502,
-      "step": 31847
-    },
-    {
-      "epoch": 16.609126466753587,
-      "grad_norm": 1.5440709590911865,
-      "learning_rate": 6.850552763819096e-05,
-      "loss": 4.3563,
-      "step": 31848
-    },
-    {
-      "epoch": 16.609647979139506,
-      "grad_norm": 1.536443829536438,
-      "learning_rate": 6.850452261306532e-05,
-      "loss": 5.4576,
-      "step": 31849
-    },
-    {
-      "epoch": 16.610169491525422,
-      "grad_norm": 1.4842041730880737,
-      "learning_rate": 6.85035175879397e-05,
-      "loss": 5.1197,
-      "step": 31850
-    },
-    {
-      "epoch": 16.61069100391134,
-      "grad_norm": 1.5766043663024902,
-      "learning_rate": 6.850251256281407e-05,
-      "loss": 4.8271,
-      "step": 31851
-    },
-    {
-      "epoch": 16.61121251629726,
-      "grad_norm": 1.506008267402649,
-      "learning_rate": 6.850150753768844e-05,
-      "loss": 4.9587,
-      "step": 31852
-    },
-    {
-      "epoch": 16.61173402868318,
-      "grad_norm": 1.5134464502334595,
-      "learning_rate": 6.850050251256281e-05,
-      "loss": 5.3567,
-      "step": 31853
-    },
-    {
-      "epoch": 16.6122555410691,
-      "grad_norm": 1.6432462930679321,
-      "learning_rate": 6.849949748743719e-05,
-      "loss": 5.0022,
-      "step": 31854
-    },
-    {
-      "epoch": 16.61277705345502,
-      "grad_norm": 1.501944661140442,
-      "learning_rate": 6.849849246231156e-05,
-      "loss": 5.2709,
-      "step": 31855
-    },
-    {
-      "epoch": 16.61329856584094,
-      "grad_norm": 1.5880593061447144,
-      "learning_rate": 6.849748743718594e-05,
-      "loss": 5.2623,
-      "step": 31856
-    },
-    {
-      "epoch": 16.61382007822686,
-      "grad_norm": 1.462196946144104,
-      "learning_rate": 6.849648241206031e-05,
-      "loss": 4.9677,
-      "step": 31857
-    },
-    {
-      "epoch": 16.614341590612778,
-      "grad_norm": 1.4702727794647217,
-      "learning_rate": 6.849547738693468e-05,
-      "loss": 5.424,
-      "step": 31858
-    },
-    {
-      "epoch": 16.614863102998697,
-      "grad_norm": 1.458552360534668,
-      "learning_rate": 6.849447236180905e-05,
-      "loss": 5.0189,
-      "step": 31859
-    },
-    {
-      "epoch": 16.615384615384617,
-      "grad_norm": 1.441962480545044,
-      "learning_rate": 6.849346733668343e-05,
-      "loss": 5.1317,
-      "step": 31860
-    },
-    {
-      "epoch": 16.615906127770536,
-      "grad_norm": 1.5440874099731445,
-      "learning_rate": 6.849246231155779e-05,
-      "loss": 5.0432,
-      "step": 31861
-    },
-    {
-      "epoch": 16.616427640156452,
-      "grad_norm": 1.5573375225067139,
-      "learning_rate": 6.849145728643216e-05,
-      "loss": 5.4556,
-      "step": 31862
-    },
-    {
-      "epoch": 16.61694915254237,
-      "grad_norm": 1.5333975553512573,
-      "learning_rate": 6.849045226130653e-05,
-      "loss": 5.4828,
-      "step": 31863
-    },
-    {
-      "epoch": 16.61747066492829,
-      "grad_norm": 1.6344773769378662,
-      "learning_rate": 6.84894472361809e-05,
-      "loss": 5.0109,
-      "step": 31864
-    },
-    {
-      "epoch": 16.61799217731421,
-      "grad_norm": 1.434840202331543,
-      "learning_rate": 6.848844221105528e-05,
-      "loss": 5.7956,
-      "step": 31865
-    },
-    {
-      "epoch": 16.61851368970013,
-      "grad_norm": 1.527025580406189,
-      "learning_rate": 6.848743718592965e-05,
-      "loss": 5.3184,
-      "step": 31866
-    },
-    {
-      "epoch": 16.61903520208605,
-      "grad_norm": 1.4150981903076172,
-      "learning_rate": 6.848643216080403e-05,
-      "loss": 5.6621,
-      "step": 31867
-    },
-    {
-      "epoch": 16.61955671447197,
-      "grad_norm": 1.416407585144043,
-      "learning_rate": 6.84854271356784e-05,
-      "loss": 5.5301,
-      "step": 31868
-    },
-    {
-      "epoch": 16.62007822685789,
-      "grad_norm": 1.4617884159088135,
-      "learning_rate": 6.848442211055277e-05,
-      "loss": 5.5013,
-      "step": 31869
-    },
-    {
-      "epoch": 16.620599739243808,
-      "grad_norm": 1.4561933279037476,
-      "learning_rate": 6.848341708542714e-05,
-      "loss": 5.4533,
-      "step": 31870
-    },
-    {
-      "epoch": 16.621121251629727,
-      "grad_norm": 1.4716582298278809,
-      "learning_rate": 6.848241206030152e-05,
-      "loss": 5.2156,
-      "step": 31871
-    },
-    {
-      "epoch": 16.621642764015647,
-      "grad_norm": 1.4080970287322998,
-      "learning_rate": 6.848140703517588e-05,
-      "loss": 5.5056,
-      "step": 31872
-    },
-    {
-      "epoch": 16.622164276401566,
-      "grad_norm": 1.4663978815078735,
-      "learning_rate": 6.848040201005026e-05,
-      "loss": 4.9122,
-      "step": 31873
-    },
-    {
-      "epoch": 16.622685788787482,
-      "grad_norm": 1.5482224225997925,
-      "learning_rate": 6.847939698492462e-05,
-      "loss": 5.6171,
-      "step": 31874
-    },
-    {
-      "epoch": 16.6232073011734,
-      "grad_norm": 1.4037301540374756,
-      "learning_rate": 6.8478391959799e-05,
-      "loss": 5.4715,
-      "step": 31875
-    },
-    {
-      "epoch": 16.62372881355932,
-      "grad_norm": 1.5918887853622437,
-      "learning_rate": 6.847738693467338e-05,
-      "loss": 4.9029,
-      "step": 31876
-    },
-    {
-      "epoch": 16.62425032594524,
-      "grad_norm": 1.5359920263290405,
-      "learning_rate": 6.847638190954774e-05,
-      "loss": 4.9092,
-      "step": 31877
-    },
-    {
-      "epoch": 16.62477183833116,
-      "grad_norm": 1.516543984413147,
-      "learning_rate": 6.847537688442212e-05,
-      "loss": 5.1078,
-      "step": 31878
-    },
-    {
-      "epoch": 16.62529335071708,
-      "grad_norm": 1.5235021114349365,
-      "learning_rate": 6.847437185929648e-05,
-      "loss": 5.5048,
-      "step": 31879
-    },
-    {
-      "epoch": 16.625814863103,
-      "grad_norm": 1.4682456254959106,
-      "learning_rate": 6.847336683417086e-05,
-      "loss": 5.5521,
-      "step": 31880
-    },
-    {
-      "epoch": 16.62633637548892,
-      "grad_norm": 1.591693639755249,
-      "learning_rate": 6.847236180904523e-05,
-      "loss": 5.4981,
-      "step": 31881
-    },
-    {
-      "epoch": 16.626857887874838,
-      "grad_norm": 1.2662756443023682,
-      "learning_rate": 6.84713567839196e-05,
-      "loss": 5.9601,
-      "step": 31882
-    },
-    {
-      "epoch": 16.627379400260757,
-      "grad_norm": 1.5226033926010132,
-      "learning_rate": 6.847035175879397e-05,
-      "loss": 5.486,
-      "step": 31883
-    },
-    {
-      "epoch": 16.627900912646677,
-      "grad_norm": 1.4315450191497803,
-      "learning_rate": 6.846934673366835e-05,
-      "loss": 5.4798,
-      "step": 31884
-    },
-    {
-      "epoch": 16.628422425032596,
-      "grad_norm": 1.4928046464920044,
-      "learning_rate": 6.846834170854271e-05,
-      "loss": 5.186,
-      "step": 31885
-    },
-    {
-      "epoch": 16.628943937418512,
-      "grad_norm": 1.557258129119873,
-      "learning_rate": 6.846733668341709e-05,
-      "loss": 5.3907,
-      "step": 31886
-    },
-    {
-      "epoch": 16.62946544980443,
-      "grad_norm": 1.6724591255187988,
-      "learning_rate": 6.846633165829147e-05,
-      "loss": 4.934,
-      "step": 31887
-    },
-    {
-      "epoch": 16.62998696219035,
-      "grad_norm": 1.4886794090270996,
-      "learning_rate": 6.846532663316583e-05,
-      "loss": 5.0048,
-      "step": 31888
-    },
-    {
-      "epoch": 16.63050847457627,
-      "grad_norm": 1.5613658428192139,
-      "learning_rate": 6.846432160804021e-05,
-      "loss": 4.6686,
-      "step": 31889
-    },
-    {
-      "epoch": 16.63102998696219,
-      "grad_norm": 1.431496262550354,
-      "learning_rate": 6.846331658291457e-05,
-      "loss": 5.5014,
-      "step": 31890
-    },
-    {
-      "epoch": 16.63155149934811,
-      "grad_norm": 1.5014609098434448,
-      "learning_rate": 6.846231155778895e-05,
-      "loss": 5.0991,
-      "step": 31891
-    },
-    {
-      "epoch": 16.63207301173403,
-      "grad_norm": 1.6725143194198608,
-      "learning_rate": 6.846130653266331e-05,
-      "loss": 4.6843,
-      "step": 31892
-    },
-    {
-      "epoch": 16.63259452411995,
-      "grad_norm": 1.658812403678894,
-      "learning_rate": 6.846030150753769e-05,
-      "loss": 4.8975,
-      "step": 31893
-    },
-    {
-      "epoch": 16.633116036505868,
-      "grad_norm": 1.6173521280288696,
-      "learning_rate": 6.845929648241206e-05,
-      "loss": 5.3553,
-      "step": 31894
-    },
-    {
-      "epoch": 16.633637548891787,
-      "grad_norm": 1.4931799173355103,
-      "learning_rate": 6.845829145728643e-05,
-      "loss": 5.2964,
-      "step": 31895
-    },
-    {
-      "epoch": 16.634159061277707,
-      "grad_norm": 1.5480148792266846,
-      "learning_rate": 6.845728643216081e-05,
-      "loss": 5.2093,
-      "step": 31896
-    },
-    {
-      "epoch": 16.634680573663623,
-      "grad_norm": 1.4691095352172852,
-      "learning_rate": 6.845628140703519e-05,
-      "loss": 5.4832,
-      "step": 31897
-    },
-    {
-      "epoch": 16.635202086049542,
-      "grad_norm": 1.4638532400131226,
-      "learning_rate": 6.845527638190955e-05,
-      "loss": 5.5678,
-      "step": 31898
-    },
-    {
-      "epoch": 16.63572359843546,
-      "grad_norm": 1.4019490480422974,
-      "learning_rate": 6.845427135678393e-05,
-      "loss": 5.1125,
-      "step": 31899
-    },
-    {
-      "epoch": 16.63624511082138,
-      "grad_norm": 1.611497402191162,
-      "learning_rate": 6.84532663316583e-05,
-      "loss": 4.758,
-      "step": 31900
-    },
-    {
-      "epoch": 16.6367666232073,
-      "grad_norm": 1.567062258720398,
-      "learning_rate": 6.845226130653266e-05,
-      "loss": 4.9861,
-      "step": 31901
-    },
-    {
-      "epoch": 16.63728813559322,
-      "grad_norm": 1.543563723564148,
-      "learning_rate": 6.845125628140704e-05,
-      "loss": 5.1787,
-      "step": 31902
-    },
-    {
-      "epoch": 16.63780964797914,
-      "grad_norm": 1.4323350191116333,
-      "learning_rate": 6.84502512562814e-05,
-      "loss": 5.6205,
-      "step": 31903
-    },
-    {
-      "epoch": 16.63833116036506,
-      "grad_norm": 1.5520819425582886,
-      "learning_rate": 6.844924623115578e-05,
-      "loss": 5.5593,
-      "step": 31904
-    },
-    {
-      "epoch": 16.63885267275098,
-      "grad_norm": 1.4478727579116821,
-      "learning_rate": 6.844824120603014e-05,
-      "loss": 5.2705,
-      "step": 31905
-    },
-    {
-      "epoch": 16.639374185136898,
-      "grad_norm": 1.4533765316009521,
-      "learning_rate": 6.844723618090452e-05,
-      "loss": 5.5308,
-      "step": 31906
-    },
-    {
-      "epoch": 16.639895697522817,
-      "grad_norm": 1.53481924533844,
-      "learning_rate": 6.84462311557789e-05,
-      "loss": 5.3357,
-      "step": 31907
-    },
-    {
-      "epoch": 16.640417209908737,
-      "grad_norm": 1.5570785999298096,
-      "learning_rate": 6.844522613065328e-05,
-      "loss": 5.1334,
-      "step": 31908
-    },
-    {
-      "epoch": 16.640938722294656,
-      "grad_norm": 1.5452288389205933,
-      "learning_rate": 6.844422110552764e-05,
-      "loss": 5.6066,
-      "step": 31909
-    },
-    {
-      "epoch": 16.641460234680572,
-      "grad_norm": 1.5492029190063477,
-      "learning_rate": 6.844321608040202e-05,
-      "loss": 5.3216,
-      "step": 31910
-    },
-    {
-      "epoch": 16.64198174706649,
-      "grad_norm": 1.483486533164978,
-      "learning_rate": 6.844221105527638e-05,
-      "loss": 5.4735,
-      "step": 31911
-    },
-    {
-      "epoch": 16.64250325945241,
-      "grad_norm": 1.4873508214950562,
-      "learning_rate": 6.844120603015076e-05,
-      "loss": 5.5659,
-      "step": 31912
-    },
-    {
-      "epoch": 16.64302477183833,
-      "grad_norm": 1.4417425394058228,
-      "learning_rate": 6.844020100502513e-05,
-      "loss": 5.6749,
-      "step": 31913
-    },
-    {
-      "epoch": 16.64354628422425,
-      "grad_norm": 1.4467220306396484,
-      "learning_rate": 6.843919597989949e-05,
-      "loss": 4.8219,
-      "step": 31914
-    },
-    {
-      "epoch": 16.64406779661017,
-      "grad_norm": 1.6462583541870117,
-      "learning_rate": 6.843819095477387e-05,
-      "loss": 5.1864,
-      "step": 31915
-    },
-    {
-      "epoch": 16.64458930899609,
-      "grad_norm": 1.5797569751739502,
-      "learning_rate": 6.843718592964825e-05,
-      "loss": 4.7511,
-      "step": 31916
-    },
-    {
-      "epoch": 16.64511082138201,
-      "grad_norm": 1.548746943473816,
-      "learning_rate": 6.843618090452262e-05,
-      "loss": 4.9183,
-      "step": 31917
-    },
-    {
-      "epoch": 16.645632333767928,
-      "grad_norm": 1.5686547756195068,
-      "learning_rate": 6.843517587939699e-05,
-      "loss": 5.2533,
-      "step": 31918
-    },
-    {
-      "epoch": 16.646153846153847,
-      "grad_norm": 1.4747228622436523,
-      "learning_rate": 6.843417085427137e-05,
-      "loss": 5.0514,
-      "step": 31919
-    },
-    {
-      "epoch": 16.646675358539767,
-      "grad_norm": 1.5008950233459473,
-      "learning_rate": 6.843316582914573e-05,
-      "loss": 5.4627,
-      "step": 31920
-    },
-    {
-      "epoch": 16.647196870925683,
-      "grad_norm": 1.5788356065750122,
-      "learning_rate": 6.843216080402011e-05,
-      "loss": 5.3666,
-      "step": 31921
-    },
-    {
-      "epoch": 16.647718383311602,
-      "grad_norm": 1.4959403276443481,
-      "learning_rate": 6.843115577889447e-05,
-      "loss": 4.5811,
-      "step": 31922
-    },
-    {
-      "epoch": 16.64823989569752,
-      "grad_norm": 1.6255667209625244,
-      "learning_rate": 6.843015075376885e-05,
-      "loss": 5.4326,
-      "step": 31923
-    },
-    {
-      "epoch": 16.64876140808344,
-      "grad_norm": 1.4038654565811157,
-      "learning_rate": 6.842914572864321e-05,
-      "loss": 5.3508,
-      "step": 31924
-    },
-    {
-      "epoch": 16.64928292046936,
-      "grad_norm": 1.5200715065002441,
-      "learning_rate": 6.842814070351759e-05,
-      "loss": 4.6382,
-      "step": 31925
-    },
-    {
-      "epoch": 16.64980443285528,
-      "grad_norm": 1.3741405010223389,
-      "learning_rate": 6.842713567839196e-05,
-      "loss": 5.2967,
-      "step": 31926
-    },
-    {
-      "epoch": 16.6503259452412,
-      "grad_norm": 1.4211163520812988,
-      "learning_rate": 6.842613065326633e-05,
-      "loss": 5.131,
-      "step": 31927
-    },
-    {
-      "epoch": 16.65084745762712,
-      "grad_norm": 1.4884880781173706,
-      "learning_rate": 6.842512562814071e-05,
-      "loss": 5.4846,
-      "step": 31928
-    },
-    {
-      "epoch": 16.65136897001304,
-      "grad_norm": 1.3818668127059937,
-      "learning_rate": 6.842412060301508e-05,
-      "loss": 5.0727,
-      "step": 31929
-    },
-    {
-      "epoch": 16.651890482398958,
-      "grad_norm": 1.3905547857284546,
-      "learning_rate": 6.842311557788945e-05,
-      "loss": 5.7815,
-      "step": 31930
-    },
-    {
-      "epoch": 16.652411994784877,
-      "grad_norm": 1.5259240865707397,
-      "learning_rate": 6.842211055276382e-05,
-      "loss": 5.0637,
-      "step": 31931
-    },
-    {
-      "epoch": 16.652933507170797,
-      "grad_norm": 1.4397190809249878,
-      "learning_rate": 6.84211055276382e-05,
-      "loss": 5.6769,
-      "step": 31932
-    },
-    {
-      "epoch": 16.653455019556713,
-      "grad_norm": 1.4992903470993042,
-      "learning_rate": 6.842010050251256e-05,
-      "loss": 4.7232,
-      "step": 31933
-    },
-    {
-      "epoch": 16.653976531942632,
-      "grad_norm": 1.388408899307251,
-      "learning_rate": 6.841909547738694e-05,
-      "loss": 5.6064,
-      "step": 31934
-    },
-    {
-      "epoch": 16.65449804432855,
-      "grad_norm": 1.4343599081039429,
-      "learning_rate": 6.84180904522613e-05,
-      "loss": 5.7613,
-      "step": 31935
-    },
-    {
-      "epoch": 16.65501955671447,
-      "grad_norm": 1.4131217002868652,
-      "learning_rate": 6.841708542713568e-05,
-      "loss": 5.6839,
-      "step": 31936
-    },
-    {
-      "epoch": 16.65554106910039,
-      "grad_norm": 1.4816687107086182,
-      "learning_rate": 6.841608040201006e-05,
-      "loss": 5.0489,
-      "step": 31937
-    },
-    {
-      "epoch": 16.65606258148631,
-      "grad_norm": 1.4833767414093018,
-      "learning_rate": 6.841507537688444e-05,
-      "loss": 5.5992,
-      "step": 31938
-    },
-    {
-      "epoch": 16.65658409387223,
-      "grad_norm": 1.5674302577972412,
-      "learning_rate": 6.84140703517588e-05,
-      "loss": 5.617,
-      "step": 31939
-    },
-    {
-      "epoch": 16.65710560625815,
-      "grad_norm": 1.3806949853897095,
-      "learning_rate": 6.841306532663318e-05,
-      "loss": 5.3472,
-      "step": 31940
-    },
-    {
-      "epoch": 16.65762711864407,
-      "grad_norm": 1.42318594455719,
-      "learning_rate": 6.841206030150754e-05,
-      "loss": 5.202,
-      "step": 31941
-    },
-    {
-      "epoch": 16.658148631029988,
-      "grad_norm": 1.446267008781433,
-      "learning_rate": 6.841105527638191e-05,
-      "loss": 5.4369,
-      "step": 31942
-    },
-    {
-      "epoch": 16.658670143415907,
-      "grad_norm": 1.4571664333343506,
-      "learning_rate": 6.841005025125629e-05,
-      "loss": 5.3911,
-      "step": 31943
-    },
-    {
-      "epoch": 16.659191655801827,
-      "grad_norm": 1.3605232238769531,
-      "learning_rate": 6.840904522613065e-05,
-      "loss": 4.948,
-      "step": 31944
-    },
-    {
-      "epoch": 16.659713168187743,
-      "grad_norm": 1.7863389253616333,
-      "learning_rate": 6.840804020100503e-05,
-      "loss": 4.9962,
-      "step": 31945
-    },
-    {
-      "epoch": 16.660234680573662,
-      "grad_norm": 1.455582618713379,
-      "learning_rate": 6.840703517587939e-05,
-      "loss": 5.7022,
-      "step": 31946
-    },
-    {
-      "epoch": 16.66075619295958,
-      "grad_norm": 1.357267141342163,
-      "learning_rate": 6.840603015075377e-05,
-      "loss": 5.2918,
-      "step": 31947
-    },
-    {
-      "epoch": 16.6612777053455,
-      "grad_norm": 1.469768762588501,
-      "learning_rate": 6.840502512562815e-05,
-      "loss": 5.2693,
-      "step": 31948
-    },
-    {
-      "epoch": 16.66179921773142,
-      "grad_norm": 1.4580644369125366,
-      "learning_rate": 6.840402010050252e-05,
-      "loss": 4.9595,
-      "step": 31949
-    },
-    {
-      "epoch": 16.66232073011734,
-      "grad_norm": 1.4501991271972656,
-      "learning_rate": 6.840301507537689e-05,
-      "loss": 5.604,
-      "step": 31950
-    },
-    {
-      "epoch": 16.66284224250326,
-      "grad_norm": 1.5719468593597412,
-      "learning_rate": 6.840201005025127e-05,
-      "loss": 4.8334,
-      "step": 31951
-    },
-    {
-      "epoch": 16.66336375488918,
-      "grad_norm": 1.8010900020599365,
-      "learning_rate": 6.840100502512563e-05,
-      "loss": 5.4659,
-      "step": 31952
-    },
-    {
-      "epoch": 16.6638852672751,
-      "grad_norm": 1.5608073472976685,
-      "learning_rate": 6.840000000000001e-05,
-      "loss": 5.6785,
-      "step": 31953
-    },
-    {
-      "epoch": 16.664406779661018,
-      "grad_norm": 1.5459074974060059,
-      "learning_rate": 6.839899497487437e-05,
-      "loss": 5.1916,
-      "step": 31954
-    },
-    {
-      "epoch": 16.664928292046937,
-      "grad_norm": 1.4627587795257568,
-      "learning_rate": 6.839798994974874e-05,
-      "loss": 4.9402,
-      "step": 31955
-    },
-    {
-      "epoch": 16.665449804432857,
-      "grad_norm": 1.506183385848999,
-      "learning_rate": 6.839698492462312e-05,
-      "loss": 5.5794,
-      "step": 31956
-    },
-    {
-      "epoch": 16.665971316818773,
-      "grad_norm": 1.4704970121383667,
-      "learning_rate": 6.839597989949748e-05,
-      "loss": 5.1187,
-      "step": 31957
-    },
-    {
-      "epoch": 16.666492829204692,
-      "grad_norm": 1.6960220336914062,
-      "learning_rate": 6.839497487437186e-05,
-      "loss": 4.9431,
-      "step": 31958
-    },
-    {
-      "epoch": 16.667014341590612,
-      "grad_norm": 1.632696509361267,
-      "learning_rate": 6.839396984924624e-05,
-      "loss": 5.119,
-      "step": 31959
-    },
-    {
-      "epoch": 16.66753585397653,
-      "grad_norm": 1.5150455236434937,
-      "learning_rate": 6.839296482412061e-05,
-      "loss": 5.4251,
-      "step": 31960
-    },
-    {
-      "epoch": 16.66805736636245,
-      "grad_norm": 1.4508358240127563,
-      "learning_rate": 6.839195979899498e-05,
-      "loss": 5.0038,
-      "step": 31961
-    },
-    {
-      "epoch": 16.66857887874837,
-      "grad_norm": 1.459735631942749,
-      "learning_rate": 6.839095477386936e-05,
-      "loss": 5.5181,
-      "step": 31962
-    },
-    {
-      "epoch": 16.66910039113429,
-      "grad_norm": 1.4761251211166382,
-      "learning_rate": 6.838994974874372e-05,
-      "loss": 5.0162,
-      "step": 31963
-    },
-    {
-      "epoch": 16.66962190352021,
-      "grad_norm": 1.431638479232788,
-      "learning_rate": 6.83889447236181e-05,
-      "loss": 4.8434,
-      "step": 31964
-    },
-    {
-      "epoch": 16.67014341590613,
-      "grad_norm": 1.5756245851516724,
-      "learning_rate": 6.838793969849246e-05,
-      "loss": 5.2848,
-      "step": 31965
-    },
-    {
-      "epoch": 16.670664928292048,
-      "grad_norm": 1.5461002588272095,
-      "learning_rate": 6.838693467336684e-05,
-      "loss": 4.9217,
-      "step": 31966
-    },
-    {
-      "epoch": 16.671186440677968,
-      "grad_norm": 1.3792314529418945,
-      "learning_rate": 6.83859296482412e-05,
-      "loss": 5.2632,
-      "step": 31967
-    },
-    {
-      "epoch": 16.671707953063887,
-      "grad_norm": 1.4643452167510986,
-      "learning_rate": 6.838492462311558e-05,
-      "loss": 4.9405,
-      "step": 31968
-    },
-    {
-      "epoch": 16.672229465449803,
-      "grad_norm": 1.4723987579345703,
-      "learning_rate": 6.838391959798996e-05,
-      "loss": 5.1936,
-      "step": 31969
-    },
-    {
-      "epoch": 16.672750977835722,
-      "grad_norm": 1.4659594297409058,
-      "learning_rate": 6.838291457286432e-05,
-      "loss": 5.1202,
-      "step": 31970
-    },
-    {
-      "epoch": 16.673272490221642,
-      "grad_norm": 1.4708499908447266,
-      "learning_rate": 6.83819095477387e-05,
-      "loss": 5.3598,
-      "step": 31971
-    },
-    {
-      "epoch": 16.67379400260756,
-      "grad_norm": 1.6018763780593872,
-      "learning_rate": 6.838090452261307e-05,
-      "loss": 5.2556,
-      "step": 31972
-    },
-    {
-      "epoch": 16.67431551499348,
-      "grad_norm": 1.5926071405410767,
-      "learning_rate": 6.837989949748744e-05,
-      "loss": 5.1462,
-      "step": 31973
-    },
-    {
-      "epoch": 16.6748370273794,
-      "grad_norm": 1.5126168727874756,
-      "learning_rate": 6.837889447236181e-05,
-      "loss": 5.473,
-      "step": 31974
-    },
-    {
-      "epoch": 16.67535853976532,
-      "grad_norm": 1.3723220825195312,
-      "learning_rate": 6.837788944723619e-05,
-      "loss": 5.4722,
-      "step": 31975
-    },
-    {
-      "epoch": 16.67588005215124,
-      "grad_norm": 1.4753799438476562,
-      "learning_rate": 6.837688442211055e-05,
-      "loss": 5.2273,
-      "step": 31976
-    },
-    {
-      "epoch": 16.67640156453716,
-      "grad_norm": 1.6215736865997314,
-      "learning_rate": 6.837587939698493e-05,
-      "loss": 4.846,
-      "step": 31977
-    },
-    {
-      "epoch": 16.676923076923078,
-      "grad_norm": 1.4044663906097412,
-      "learning_rate": 6.837487437185929e-05,
-      "loss": 4.4987,
-      "step": 31978
-    },
-    {
-      "epoch": 16.677444589308998,
-      "grad_norm": 1.4604347944259644,
-      "learning_rate": 6.837386934673367e-05,
-      "loss": 5.6386,
-      "step": 31979
-    },
-    {
-      "epoch": 16.677966101694913,
-      "grad_norm": 1.4653055667877197,
-      "learning_rate": 6.837286432160805e-05,
-      "loss": 5.4327,
-      "step": 31980
-    },
-    {
-      "epoch": 16.678487614080833,
-      "grad_norm": 1.5169107913970947,
-      "learning_rate": 6.837185929648241e-05,
-      "loss": 5.0267,
-      "step": 31981
-    },
-    {
-      "epoch": 16.679009126466752,
-      "grad_norm": 1.5345293283462524,
-      "learning_rate": 6.837085427135679e-05,
-      "loss": 5.3701,
-      "step": 31982
-    },
-    {
-      "epoch": 16.679530638852672,
-      "grad_norm": 1.5204527378082275,
-      "learning_rate": 6.836984924623115e-05,
-      "loss": 4.9365,
-      "step": 31983
-    },
-    {
-      "epoch": 16.68005215123859,
-      "grad_norm": 1.5136934518814087,
-      "learning_rate": 6.836884422110553e-05,
-      "loss": 5.3612,
-      "step": 31984
-    },
-    {
-      "epoch": 16.68057366362451,
-      "grad_norm": 1.5329545736312866,
-      "learning_rate": 6.83678391959799e-05,
-      "loss": 5.0328,
-      "step": 31985
-    },
-    {
-      "epoch": 16.68109517601043,
-      "grad_norm": 1.458855390548706,
-      "learning_rate": 6.836683417085427e-05,
-      "loss": 5.4262,
-      "step": 31986
-    },
-    {
-      "epoch": 16.68161668839635,
-      "grad_norm": 1.5793513059616089,
-      "learning_rate": 6.836582914572864e-05,
-      "loss": 5.2297,
-      "step": 31987
-    },
-    {
-      "epoch": 16.68213820078227,
-      "grad_norm": 1.553560733795166,
-      "learning_rate": 6.836482412060302e-05,
-      "loss": 5.343,
-      "step": 31988
-    },
-    {
-      "epoch": 16.68265971316819,
-      "grad_norm": 1.5456444025039673,
-      "learning_rate": 6.83638190954774e-05,
-      "loss": 4.4416,
-      "step": 31989
-    },
-    {
-      "epoch": 16.683181225554108,
-      "grad_norm": 1.5801692008972168,
-      "learning_rate": 6.836281407035177e-05,
-      "loss": 5.6659,
-      "step": 31990
-    },
-    {
-      "epoch": 16.683702737940028,
-      "grad_norm": 1.5634796619415283,
-      "learning_rate": 6.836180904522614e-05,
-      "loss": 4.7547,
-      "step": 31991
-    },
-    {
-      "epoch": 16.684224250325947,
-      "grad_norm": 1.505982518196106,
-      "learning_rate": 6.836080402010051e-05,
-      "loss": 5.3472,
-      "step": 31992
-    },
-    {
-      "epoch": 16.684745762711863,
-      "grad_norm": 1.51729154586792,
-      "learning_rate": 6.835979899497488e-05,
-      "loss": 5.2128,
-      "step": 31993
-    },
-    {
-      "epoch": 16.685267275097782,
-      "grad_norm": 1.4766892194747925,
-      "learning_rate": 6.835879396984924e-05,
-      "loss": 5.3472,
-      "step": 31994
-    },
-    {
-      "epoch": 16.685788787483702,
-      "grad_norm": 1.5803604125976562,
-      "learning_rate": 6.835778894472362e-05,
-      "loss": 5.1877,
-      "step": 31995
-    },
-    {
-      "epoch": 16.68631029986962,
-      "grad_norm": 1.4269686937332153,
-      "learning_rate": 6.835678391959798e-05,
-      "loss": 5.5263,
-      "step": 31996
-    },
-    {
-      "epoch": 16.68683181225554,
-      "grad_norm": 1.4586241245269775,
-      "learning_rate": 6.835577889447236e-05,
-      "loss": 5.6591,
-      "step": 31997
-    },
-    {
-      "epoch": 16.68735332464146,
-      "grad_norm": 1.5935258865356445,
-      "learning_rate": 6.835477386934673e-05,
-      "loss": 4.847,
-      "step": 31998
-    },
-    {
-      "epoch": 16.68787483702738,
-      "grad_norm": 1.5129070281982422,
-      "learning_rate": 6.83537688442211e-05,
-      "loss": 5.1345,
-      "step": 31999
-    },
-    {
-      "epoch": 16.6883963494133,
-      "grad_norm": 1.6582225561141968,
-      "learning_rate": 6.835276381909548e-05,
-      "loss": 5.0564,
-      "step": 32000
-    },
-    {
-      "epoch": 16.68891786179922,
-      "grad_norm": 1.6449384689331055,
-      "learning_rate": 6.835175879396986e-05,
-      "loss": 5.2391,
-      "step": 32001
-    },
-    {
-      "epoch": 16.689439374185138,
-      "grad_norm": 1.54106605052948,
-      "learning_rate": 6.835075376884422e-05,
-      "loss": 5.4856,
-      "step": 32002
-    },
-    {
-      "epoch": 16.689960886571058,
-      "grad_norm": 1.4999771118164062,
-      "learning_rate": 6.83497487437186e-05,
-      "loss": 5.5691,
-      "step": 32003
-    },
-    {
-      "epoch": 16.690482398956973,
-      "grad_norm": 1.5358470678329468,
-      "learning_rate": 6.834874371859297e-05,
-      "loss": 4.9833,
-      "step": 32004
-    },
-    {
-      "epoch": 16.691003911342893,
-      "grad_norm": 1.423054575920105,
-      "learning_rate": 6.834773869346734e-05,
-      "loss": 5.5241,
-      "step": 32005
-    },
-    {
-      "epoch": 16.691525423728812,
-      "grad_norm": 1.4493337869644165,
-      "learning_rate": 6.834673366834171e-05,
-      "loss": 5.2989,
-      "step": 32006
-    },
-    {
-      "epoch": 16.692046936114732,
-      "grad_norm": 1.5359723567962646,
-      "learning_rate": 6.834572864321607e-05,
-      "loss": 5.279,
-      "step": 32007
-    },
-    {
-      "epoch": 16.69256844850065,
-      "grad_norm": 1.4938026666641235,
-      "learning_rate": 6.834472361809045e-05,
-      "loss": 4.6092,
-      "step": 32008
-    },
-    {
-      "epoch": 16.69308996088657,
-      "grad_norm": 1.3895890712738037,
-      "learning_rate": 6.834371859296483e-05,
-      "loss": 5.2818,
-      "step": 32009
-    },
-    {
-      "epoch": 16.69361147327249,
-      "grad_norm": 1.4527480602264404,
-      "learning_rate": 6.83427135678392e-05,
-      "loss": 5.2511,
-      "step": 32010
-    },
-    {
-      "epoch": 16.69413298565841,
-      "grad_norm": 1.5115550756454468,
-      "learning_rate": 6.834170854271357e-05,
-      "loss": 5.1459,
-      "step": 32011
-    },
-    {
-      "epoch": 16.69465449804433,
-      "grad_norm": 1.4334189891815186,
-      "learning_rate": 6.834070351758795e-05,
-      "loss": 5.7758,
-      "step": 32012
-    },
-    {
-      "epoch": 16.69517601043025,
-      "grad_norm": 1.4479719400405884,
-      "learning_rate": 6.833969849246231e-05,
-      "loss": 5.33,
-      "step": 32013
-    },
-    {
-      "epoch": 16.695697522816168,
-      "grad_norm": 1.4262350797653198,
-      "learning_rate": 6.833869346733669e-05,
-      "loss": 5.5394,
-      "step": 32014
-    },
-    {
-      "epoch": 16.696219035202088,
-      "grad_norm": 1.6067804098129272,
-      "learning_rate": 6.833768844221106e-05,
-      "loss": 5.0638,
-      "step": 32015
-    },
-    {
-      "epoch": 16.696740547588004,
-      "grad_norm": 1.419331669807434,
-      "learning_rate": 6.833668341708543e-05,
-      "loss": 5.5816,
-      "step": 32016
-    },
-    {
-      "epoch": 16.697262059973923,
-      "grad_norm": 1.5293445587158203,
-      "learning_rate": 6.83356783919598e-05,
-      "loss": 5.2251,
-      "step": 32017
-    },
-    {
-      "epoch": 16.697783572359842,
-      "grad_norm": 1.5045663118362427,
-      "learning_rate": 6.833467336683417e-05,
-      "loss": 5.1528,
-      "step": 32018
-    },
-    {
-      "epoch": 16.698305084745762,
-      "grad_norm": 1.4855873584747314,
-      "learning_rate": 6.833366834170854e-05,
-      "loss": 5.4556,
-      "step": 32019
-    },
-    {
-      "epoch": 16.69882659713168,
-      "grad_norm": 1.4492416381835938,
-      "learning_rate": 6.833266331658292e-05,
-      "loss": 5.7569,
-      "step": 32020
-    },
-    {
-      "epoch": 16.6993481095176,
-      "grad_norm": 1.4488780498504639,
-      "learning_rate": 6.83316582914573e-05,
-      "loss": 5.3055,
-      "step": 32021
-    },
-    {
-      "epoch": 16.69986962190352,
-      "grad_norm": 1.5006591081619263,
-      "learning_rate": 6.833065326633166e-05,
-      "loss": 5.1597,
-      "step": 32022
-    },
-    {
-      "epoch": 16.70039113428944,
-      "grad_norm": 1.4456722736358643,
-      "learning_rate": 6.832964824120604e-05,
-      "loss": 5.1311,
-      "step": 32023
-    },
-    {
-      "epoch": 16.70091264667536,
-      "grad_norm": 1.4838725328445435,
-      "learning_rate": 6.83286432160804e-05,
-      "loss": 5.1238,
-      "step": 32024
-    },
-    {
-      "epoch": 16.70143415906128,
-      "grad_norm": 1.446617841720581,
-      "learning_rate": 6.832763819095478e-05,
-      "loss": 5.656,
-      "step": 32025
-    },
-    {
-      "epoch": 16.701955671447198,
-      "grad_norm": 1.4924781322479248,
-      "learning_rate": 6.832663316582914e-05,
-      "loss": 5.2886,
-      "step": 32026
-    },
-    {
-      "epoch": 16.702477183833118,
-      "grad_norm": 1.4063756465911865,
-      "learning_rate": 6.832562814070352e-05,
-      "loss": 5.3434,
-      "step": 32027
-    },
-    {
-      "epoch": 16.702998696219034,
-      "grad_norm": 1.5528002977371216,
-      "learning_rate": 6.832462311557789e-05,
-      "loss": 5.5495,
-      "step": 32028
-    },
-    {
-      "epoch": 16.703520208604953,
-      "grad_norm": 1.4852066040039062,
-      "learning_rate": 6.832361809045226e-05,
-      "loss": 5.0583,
-      "step": 32029
-    },
-    {
-      "epoch": 16.704041720990872,
-      "grad_norm": 1.5015804767608643,
-      "learning_rate": 6.832261306532664e-05,
-      "loss": 5.7979,
-      "step": 32030
-    },
-    {
-      "epoch": 16.704563233376792,
-      "grad_norm": 1.5225262641906738,
-      "learning_rate": 6.832160804020102e-05,
-      "loss": 5.3236,
-      "step": 32031
-    },
-    {
-      "epoch": 16.70508474576271,
-      "grad_norm": 1.4748858213424683,
-      "learning_rate": 6.832060301507538e-05,
-      "loss": 5.4485,
-      "step": 32032
-    },
-    {
-      "epoch": 16.70560625814863,
-      "grad_norm": 1.3844953775405884,
-      "learning_rate": 6.831959798994976e-05,
-      "loss": 5.4748,
-      "step": 32033
-    },
-    {
-      "epoch": 16.70612777053455,
-      "grad_norm": 1.4835209846496582,
-      "learning_rate": 6.831859296482413e-05,
-      "loss": 5.4723,
-      "step": 32034
-    },
-    {
-      "epoch": 16.70664928292047,
-      "grad_norm": 1.5100219249725342,
-      "learning_rate": 6.831758793969849e-05,
-      "loss": 5.3815,
-      "step": 32035
-    },
-    {
-      "epoch": 16.70717079530639,
-      "grad_norm": 1.4313688278198242,
-      "learning_rate": 6.831658291457287e-05,
-      "loss": 5.5927,
-      "step": 32036
-    },
-    {
-      "epoch": 16.70769230769231,
-      "grad_norm": 1.5883454084396362,
-      "learning_rate": 6.831557788944723e-05,
-      "loss": 4.6124,
-      "step": 32037
-    },
-    {
-      "epoch": 16.708213820078228,
-      "grad_norm": 1.521105408668518,
-      "learning_rate": 6.831457286432161e-05,
-      "loss": 5.4857,
-      "step": 32038
-    },
-    {
-      "epoch": 16.708735332464148,
-      "grad_norm": 1.6098339557647705,
-      "learning_rate": 6.831356783919597e-05,
-      "loss": 4.9547,
-      "step": 32039
-    },
-    {
-      "epoch": 16.709256844850064,
-      "grad_norm": 1.5180692672729492,
-      "learning_rate": 6.831256281407035e-05,
-      "loss": 5.277,
-      "step": 32040
-    },
-    {
-      "epoch": 16.709778357235983,
-      "grad_norm": 1.4843460321426392,
-      "learning_rate": 6.831155778894473e-05,
-      "loss": 5.3106,
-      "step": 32041
-    },
-    {
-      "epoch": 16.710299869621903,
-      "grad_norm": 1.5732197761535645,
-      "learning_rate": 6.831055276381911e-05,
-      "loss": 5.2819,
-      "step": 32042
-    },
-    {
-      "epoch": 16.710821382007822,
-      "grad_norm": 1.3269795179367065,
-      "learning_rate": 6.830954773869347e-05,
-      "loss": 5.6678,
-      "step": 32043
-    },
-    {
-      "epoch": 16.71134289439374,
-      "grad_norm": 1.4089423418045044,
-      "learning_rate": 6.830854271356785e-05,
-      "loss": 5.3553,
-      "step": 32044
-    },
-    {
-      "epoch": 16.71186440677966,
-      "grad_norm": 1.4202280044555664,
-      "learning_rate": 6.830753768844221e-05,
-      "loss": 5.7832,
-      "step": 32045
-    },
-    {
-      "epoch": 16.71238591916558,
-      "grad_norm": 1.5122604370117188,
-      "learning_rate": 6.830653266331659e-05,
-      "loss": 5.229,
-      "step": 32046
-    },
-    {
-      "epoch": 16.7129074315515,
-      "grad_norm": 1.5257827043533325,
-      "learning_rate": 6.830552763819096e-05,
-      "loss": 5.3114,
-      "step": 32047
-    },
-    {
-      "epoch": 16.71342894393742,
-      "grad_norm": 1.420432209968567,
-      "learning_rate": 6.830452261306532e-05,
-      "loss": 5.1025,
-      "step": 32048
-    },
-    {
-      "epoch": 16.71395045632334,
-      "grad_norm": 1.5778098106384277,
-      "learning_rate": 6.83035175879397e-05,
-      "loss": 4.5882,
-      "step": 32049
-    },
-    {
-      "epoch": 16.714471968709258,
-      "grad_norm": 1.4636292457580566,
-      "learning_rate": 6.830251256281408e-05,
-      "loss": 5.6923,
-      "step": 32050
-    },
-    {
-      "epoch": 16.714993481095178,
-      "grad_norm": 1.6086596250534058,
-      "learning_rate": 6.830150753768845e-05,
-      "loss": 5.2698,
-      "step": 32051
-    },
-    {
-      "epoch": 16.715514993481094,
-      "grad_norm": 1.5577521324157715,
-      "learning_rate": 6.830050251256282e-05,
-      "loss": 5.6871,
-      "step": 32052
-    },
-    {
-      "epoch": 16.716036505867013,
-      "grad_norm": 1.4369738101959229,
-      "learning_rate": 6.82994974874372e-05,
-      "loss": 5.7495,
-      "step": 32053
-    },
-    {
-      "epoch": 16.716558018252933,
-      "grad_norm": 1.4845877885818481,
-      "learning_rate": 6.829849246231156e-05,
-      "loss": 5.3827,
-      "step": 32054
-    },
-    {
-      "epoch": 16.717079530638852,
-      "grad_norm": 1.454537272453308,
-      "learning_rate": 6.829748743718594e-05,
-      "loss": 5.3787,
-      "step": 32055
-    },
-    {
-      "epoch": 16.71760104302477,
-      "grad_norm": 1.5892754793167114,
-      "learning_rate": 6.82964824120603e-05,
-      "loss": 5.2177,
-      "step": 32056
-    },
-    {
-      "epoch": 16.71812255541069,
-      "grad_norm": 1.588005781173706,
-      "learning_rate": 6.829547738693468e-05,
-      "loss": 5.4405,
-      "step": 32057
-    },
-    {
-      "epoch": 16.71864406779661,
-      "grad_norm": 1.462024211883545,
-      "learning_rate": 6.829447236180904e-05,
-      "loss": 5.4795,
-      "step": 32058
-    },
-    {
-      "epoch": 16.71916558018253,
-      "grad_norm": 1.503610610961914,
-      "learning_rate": 6.829346733668342e-05,
-      "loss": 5.0622,
-      "step": 32059
-    },
-    {
-      "epoch": 16.71968709256845,
-      "grad_norm": 1.5807548761367798,
-      "learning_rate": 6.829246231155779e-05,
-      "loss": 5.0496,
-      "step": 32060
-    },
-    {
-      "epoch": 16.72020860495437,
-      "grad_norm": 1.3107032775878906,
-      "learning_rate": 6.829145728643216e-05,
-      "loss": 5.7079,
-      "step": 32061
-    },
-    {
-      "epoch": 16.72073011734029,
-      "grad_norm": 1.5398658514022827,
-      "learning_rate": 6.829045226130654e-05,
-      "loss": 5.0112,
-      "step": 32062
-    },
-    {
-      "epoch": 16.721251629726208,
-      "grad_norm": 1.5147812366485596,
-      "learning_rate": 6.82894472361809e-05,
-      "loss": 5.0293,
-      "step": 32063
-    },
-    {
-      "epoch": 16.721773142112124,
-      "grad_norm": 1.4100427627563477,
-      "learning_rate": 6.828844221105528e-05,
-      "loss": 5.7493,
-      "step": 32064
-    },
-    {
-      "epoch": 16.722294654498043,
-      "grad_norm": 1.5007672309875488,
-      "learning_rate": 6.828743718592965e-05,
-      "loss": 5.5515,
-      "step": 32065
-    },
-    {
-      "epoch": 16.722816166883963,
-      "grad_norm": 1.3731693029403687,
-      "learning_rate": 6.828643216080403e-05,
-      "loss": 5.1137,
-      "step": 32066
-    },
-    {
-      "epoch": 16.723337679269882,
-      "grad_norm": 1.468955159187317,
-      "learning_rate": 6.828542713567839e-05,
-      "loss": 5.1844,
-      "step": 32067
-    },
-    {
-      "epoch": 16.7238591916558,
-      "grad_norm": 1.483717918395996,
-      "learning_rate": 6.828442211055277e-05,
-      "loss": 5.5012,
-      "step": 32068
-    },
-    {
-      "epoch": 16.72438070404172,
-      "grad_norm": 1.6542272567749023,
-      "learning_rate": 6.828341708542713e-05,
-      "loss": 4.5993,
-      "step": 32069
-    },
-    {
-      "epoch": 16.72490221642764,
-      "grad_norm": 1.4523786306381226,
-      "learning_rate": 6.828241206030151e-05,
-      "loss": 5.4939,
-      "step": 32070
-    },
-    {
-      "epoch": 16.72542372881356,
-      "grad_norm": 1.4198640584945679,
-      "learning_rate": 6.828140703517589e-05,
-      "loss": 5.4627,
-      "step": 32071
-    },
-    {
-      "epoch": 16.72594524119948,
-      "grad_norm": 1.5143439769744873,
-      "learning_rate": 6.828040201005027e-05,
-      "loss": 5.5713,
-      "step": 32072
-    },
-    {
-      "epoch": 16.7264667535854,
-      "grad_norm": 1.636934757232666,
-      "learning_rate": 6.827939698492463e-05,
-      "loss": 4.6676,
-      "step": 32073
-    },
-    {
-      "epoch": 16.72698826597132,
-      "grad_norm": 1.459311604499817,
-      "learning_rate": 6.8278391959799e-05,
-      "loss": 5.3797,
-      "step": 32074
-    },
-    {
-      "epoch": 16.727509778357238,
-      "grad_norm": 1.544616937637329,
-      "learning_rate": 6.827738693467337e-05,
-      "loss": 5.028,
-      "step": 32075
-    },
-    {
-      "epoch": 16.728031290743154,
-      "grad_norm": 1.3669708967208862,
-      "learning_rate": 6.827638190954774e-05,
-      "loss": 5.5136,
-      "step": 32076
-    },
-    {
-      "epoch": 16.728552803129073,
-      "grad_norm": 1.5376347303390503,
-      "learning_rate": 6.827537688442211e-05,
-      "loss": 5.035,
-      "step": 32077
-    },
-    {
-      "epoch": 16.729074315514993,
-      "grad_norm": 1.515159249305725,
-      "learning_rate": 6.827437185929648e-05,
-      "loss": 5.096,
-      "step": 32078
-    },
-    {
-      "epoch": 16.729595827900912,
-      "grad_norm": 1.5381991863250732,
-      "learning_rate": 6.827336683417086e-05,
-      "loss": 5.2108,
-      "step": 32079
-    },
-    {
-      "epoch": 16.73011734028683,
-      "grad_norm": 1.4692661762237549,
-      "learning_rate": 6.827236180904522e-05,
-      "loss": 5.2953,
-      "step": 32080
-    },
-    {
-      "epoch": 16.73063885267275,
-      "grad_norm": 1.7885565757751465,
-      "learning_rate": 6.82713567839196e-05,
-      "loss": 4.577,
-      "step": 32081
-    },
-    {
-      "epoch": 16.73116036505867,
-      "grad_norm": 1.5153635740280151,
-      "learning_rate": 6.827035175879398e-05,
-      "loss": 4.8147,
-      "step": 32082
-    },
-    {
-      "epoch": 16.73168187744459,
-      "grad_norm": 1.442362904548645,
-      "learning_rate": 6.826934673366835e-05,
-      "loss": 4.9,
-      "step": 32083
-    },
-    {
-      "epoch": 16.73220338983051,
-      "grad_norm": 1.5675828456878662,
-      "learning_rate": 6.826834170854272e-05,
-      "loss": 4.9915,
-      "step": 32084
-    },
-    {
-      "epoch": 16.73272490221643,
-      "grad_norm": 1.4527041912078857,
-      "learning_rate": 6.82673366834171e-05,
-      "loss": 5.4928,
-      "step": 32085
-    },
-    {
-      "epoch": 16.73324641460235,
-      "grad_norm": 1.5254976749420166,
-      "learning_rate": 6.826633165829146e-05,
-      "loss": 4.3864,
-      "step": 32086
-    },
-    {
-      "epoch": 16.733767926988264,
-      "grad_norm": 1.4401723146438599,
-      "learning_rate": 6.826532663316582e-05,
-      "loss": 5.1929,
-      "step": 32087
-    },
-    {
-      "epoch": 16.734289439374184,
-      "grad_norm": 1.5075974464416504,
-      "learning_rate": 6.82643216080402e-05,
-      "loss": 4.896,
-      "step": 32088
-    },
-    {
-      "epoch": 16.734810951760103,
-      "grad_norm": 1.5956988334655762,
-      "learning_rate": 6.826331658291457e-05,
-      "loss": 5.0649,
-      "step": 32089
-    },
-    {
-      "epoch": 16.735332464146023,
-      "grad_norm": 1.5353254079818726,
-      "learning_rate": 6.826231155778894e-05,
-      "loss": 5.0025,
-      "step": 32090
-    },
-    {
-      "epoch": 16.735853976531942,
-      "grad_norm": 1.5404528379440308,
-      "learning_rate": 6.826130653266332e-05,
-      "loss": 5.3508,
-      "step": 32091
-    },
-    {
-      "epoch": 16.73637548891786,
-      "grad_norm": 1.391295075416565,
-      "learning_rate": 6.82603015075377e-05,
-      "loss": 5.5469,
-      "step": 32092
-    },
-    {
-      "epoch": 16.73689700130378,
-      "grad_norm": 1.4764683246612549,
-      "learning_rate": 6.825929648241206e-05,
-      "loss": 5.1145,
-      "step": 32093
-    },
-    {
-      "epoch": 16.7374185136897,
-      "grad_norm": 1.486870288848877,
-      "learning_rate": 6.825829145728644e-05,
-      "loss": 5.0594,
-      "step": 32094
-    },
-    {
-      "epoch": 16.73794002607562,
-      "grad_norm": 1.59083092212677,
-      "learning_rate": 6.825728643216081e-05,
-      "loss": 5.0461,
-      "step": 32095
-    },
-    {
-      "epoch": 16.73846153846154,
-      "grad_norm": 1.441457748413086,
-      "learning_rate": 6.825628140703518e-05,
-      "loss": 5.5432,
-      "step": 32096
-    },
-    {
-      "epoch": 16.73898305084746,
-      "grad_norm": 1.488957166671753,
-      "learning_rate": 6.825527638190955e-05,
-      "loss": 5.4442,
-      "step": 32097
-    },
-    {
-      "epoch": 16.73950456323338,
-      "grad_norm": 1.4642499685287476,
-      "learning_rate": 6.825427135678393e-05,
-      "loss": 5.5825,
-      "step": 32098
-    },
-    {
-      "epoch": 16.740026075619298,
-      "grad_norm": 1.4976849555969238,
-      "learning_rate": 6.825326633165829e-05,
-      "loss": 5.3616,
-      "step": 32099
-    },
-    {
-      "epoch": 16.740547588005214,
-      "grad_norm": 1.7170675992965698,
-      "learning_rate": 6.825226130653266e-05,
-      "loss": 4.7723,
-      "step": 32100
-    },
-    {
-      "epoch": 16.741069100391133,
-      "grad_norm": 1.477333426475525,
-      "learning_rate": 6.825125628140703e-05,
-      "loss": 5.649,
-      "step": 32101
-    },
-    {
-      "epoch": 16.741590612777053,
-      "grad_norm": 1.5829054117202759,
-      "learning_rate": 6.825025125628141e-05,
-      "loss": 5.2818,
-      "step": 32102
-    },
-    {
-      "epoch": 16.742112125162972,
-      "grad_norm": 1.4638057947158813,
-      "learning_rate": 6.824924623115579e-05,
-      "loss": 5.1177,
-      "step": 32103
-    },
-    {
-      "epoch": 16.74263363754889,
-      "grad_norm": 1.469643473625183,
-      "learning_rate": 6.824824120603015e-05,
-      "loss": 5.3045,
-      "step": 32104
-    },
-    {
-      "epoch": 16.74315514993481,
-      "grad_norm": 1.5875632762908936,
-      "learning_rate": 6.824723618090453e-05,
-      "loss": 5.0973,
-      "step": 32105
-    },
-    {
-      "epoch": 16.74367666232073,
-      "grad_norm": 1.503047227859497,
-      "learning_rate": 6.82462311557789e-05,
-      "loss": 5.5399,
-      "step": 32106
-    },
-    {
-      "epoch": 16.74419817470665,
-      "grad_norm": 1.4575330018997192,
-      "learning_rate": 6.824522613065327e-05,
-      "loss": 5.7665,
-      "step": 32107
-    },
-    {
-      "epoch": 16.74471968709257,
-      "grad_norm": 1.5517851114273071,
-      "learning_rate": 6.824422110552764e-05,
-      "loss": 5.1736,
-      "step": 32108
-    },
-    {
-      "epoch": 16.74524119947849,
-      "grad_norm": 1.4675567150115967,
-      "learning_rate": 6.824321608040202e-05,
-      "loss": 5.2823,
-      "step": 32109
-    },
-    {
-      "epoch": 16.74576271186441,
-      "grad_norm": 1.4470287561416626,
-      "learning_rate": 6.824221105527638e-05,
-      "loss": 5.679,
-      "step": 32110
-    },
-    {
-      "epoch": 16.746284224250324,
-      "grad_norm": 1.561443567276001,
-      "learning_rate": 6.824120603015076e-05,
-      "loss": 5.4162,
-      "step": 32111
-    },
-    {
-      "epoch": 16.746805736636244,
-      "grad_norm": 1.4478700160980225,
-      "learning_rate": 6.824020100502512e-05,
-      "loss": 5.4495,
-      "step": 32112
-    },
-    {
-      "epoch": 16.747327249022163,
-      "grad_norm": 1.476166009902954,
-      "learning_rate": 6.82391959798995e-05,
-      "loss": 5.6406,
-      "step": 32113
-    },
-    {
-      "epoch": 16.747848761408083,
-      "grad_norm": 1.5210410356521606,
-      "learning_rate": 6.823819095477388e-05,
-      "loss": 5.3948,
-      "step": 32114
-    },
-    {
-      "epoch": 16.748370273794002,
-      "grad_norm": 1.5466742515563965,
-      "learning_rate": 6.823718592964824e-05,
-      "loss": 5.3212,
-      "step": 32115
-    },
-    {
-      "epoch": 16.74889178617992,
-      "grad_norm": 1.4556159973144531,
-      "learning_rate": 6.823618090452262e-05,
-      "loss": 5.2083,
-      "step": 32116
-    },
-    {
-      "epoch": 16.74941329856584,
-      "grad_norm": 1.4717140197753906,
-      "learning_rate": 6.823517587939698e-05,
-      "loss": 5.3185,
-      "step": 32117
-    },
-    {
-      "epoch": 16.74993481095176,
-      "grad_norm": 1.3798179626464844,
-      "learning_rate": 6.823417085427136e-05,
-      "loss": 5.6402,
-      "step": 32118
-    },
-    {
-      "epoch": 16.75045632333768,
-      "grad_norm": 1.376600980758667,
-      "learning_rate": 6.823316582914573e-05,
-      "loss": 5.0926,
-      "step": 32119
-    },
-    {
-      "epoch": 16.7509778357236,
-      "grad_norm": 1.403267502784729,
-      "learning_rate": 6.82321608040201e-05,
-      "loss": 5.4897,
-      "step": 32120
-    },
-    {
-      "epoch": 16.75149934810952,
-      "grad_norm": 1.5837265253067017,
-      "learning_rate": 6.823115577889447e-05,
-      "loss": 5.289,
-      "step": 32121
-    },
-    {
-      "epoch": 16.75202086049544,
-      "grad_norm": 1.574583649635315,
-      "learning_rate": 6.823015075376885e-05,
-      "loss": 5.4006,
-      "step": 32122
-    },
-    {
-      "epoch": 16.752542372881354,
-      "grad_norm": 1.4509285688400269,
-      "learning_rate": 6.822914572864322e-05,
-      "loss": 5.7117,
-      "step": 32123
-    },
-    {
-      "epoch": 16.753063885267274,
-      "grad_norm": 1.4814643859863281,
-      "learning_rate": 6.82281407035176e-05,
-      "loss": 5.506,
-      "step": 32124
-    },
-    {
-      "epoch": 16.753585397653193,
-      "grad_norm": 1.4786255359649658,
-      "learning_rate": 6.822713567839197e-05,
-      "loss": 5.6403,
-      "step": 32125
-    },
-    {
-      "epoch": 16.754106910039113,
-      "grad_norm": 1.5042423009872437,
-      "learning_rate": 6.822613065326634e-05,
-      "loss": 5.1768,
-      "step": 32126
-    },
-    {
-      "epoch": 16.754628422425032,
-      "grad_norm": 1.4331161975860596,
-      "learning_rate": 6.822512562814071e-05,
-      "loss": 5.4156,
-      "step": 32127
-    },
-    {
-      "epoch": 16.75514993481095,
-      "grad_norm": 1.5388671159744263,
-      "learning_rate": 6.822412060301507e-05,
-      "loss": 5.4051,
-      "step": 32128
-    },
-    {
-      "epoch": 16.75567144719687,
-      "grad_norm": 1.4540215730667114,
-      "learning_rate": 6.822311557788945e-05,
-      "loss": 5.4767,
-      "step": 32129
-    },
-    {
-      "epoch": 16.75619295958279,
-      "grad_norm": 1.3237954378128052,
-      "learning_rate": 6.822211055276381e-05,
-      "loss": 5.6123,
-      "step": 32130
-    },
-    {
-      "epoch": 16.75671447196871,
-      "grad_norm": 1.4257392883300781,
-      "learning_rate": 6.822110552763819e-05,
-      "loss": 5.2283,
-      "step": 32131
-    },
-    {
-      "epoch": 16.75723598435463,
-      "grad_norm": 1.4451539516448975,
-      "learning_rate": 6.822010050251256e-05,
-      "loss": 5.5549,
-      "step": 32132
-    },
-    {
-      "epoch": 16.75775749674055,
-      "grad_norm": 1.4449673891067505,
-      "learning_rate": 6.821909547738693e-05,
-      "loss": 5.2065,
-      "step": 32133
-    },
-    {
-      "epoch": 16.75827900912647,
-      "grad_norm": 1.3807724714279175,
-      "learning_rate": 6.821809045226131e-05,
-      "loss": 5.4929,
-      "step": 32134
-    },
-    {
-      "epoch": 16.758800521512384,
-      "grad_norm": 1.622580647468567,
-      "learning_rate": 6.821708542713569e-05,
-      "loss": 5.1972,
-      "step": 32135
-    },
-    {
-      "epoch": 16.759322033898304,
-      "grad_norm": 1.4418615102767944,
-      "learning_rate": 6.821608040201005e-05,
-      "loss": 5.5234,
-      "step": 32136
-    },
-    {
-      "epoch": 16.759843546284223,
-      "grad_norm": 1.500157117843628,
-      "learning_rate": 6.821507537688443e-05,
-      "loss": 5.5876,
-      "step": 32137
-    },
-    {
-      "epoch": 16.760365058670143,
-      "grad_norm": 1.5800188779830933,
-      "learning_rate": 6.82140703517588e-05,
-      "loss": 5.1733,
-      "step": 32138
-    },
-    {
-      "epoch": 16.760886571056062,
-      "grad_norm": 1.4927252531051636,
-      "learning_rate": 6.821306532663317e-05,
-      "loss": 5.3077,
-      "step": 32139
-    },
-    {
-      "epoch": 16.76140808344198,
-      "grad_norm": 1.6553534269332886,
-      "learning_rate": 6.821206030150754e-05,
-      "loss": 4.5784,
-      "step": 32140
-    },
-    {
-      "epoch": 16.7619295958279,
-      "grad_norm": 1.5384092330932617,
-      "learning_rate": 6.82110552763819e-05,
-      "loss": 5.1933,
-      "step": 32141
-    },
-    {
-      "epoch": 16.76245110821382,
-      "grad_norm": 1.480545997619629,
-      "learning_rate": 6.821005025125628e-05,
-      "loss": 5.459,
-      "step": 32142
-    },
-    {
-      "epoch": 16.76297262059974,
-      "grad_norm": 1.4364148378372192,
-      "learning_rate": 6.820904522613066e-05,
-      "loss": 5.4782,
-      "step": 32143
-    },
-    {
-      "epoch": 16.76349413298566,
-      "grad_norm": 1.4268953800201416,
-      "learning_rate": 6.820804020100504e-05,
-      "loss": 5.6595,
-      "step": 32144
-    },
-    {
-      "epoch": 16.76401564537158,
-      "grad_norm": 1.5145984888076782,
-      "learning_rate": 6.82070351758794e-05,
-      "loss": 5.3405,
-      "step": 32145
-    },
-    {
-      "epoch": 16.7645371577575,
-      "grad_norm": 1.4740502834320068,
-      "learning_rate": 6.820603015075378e-05,
-      "loss": 5.5224,
-      "step": 32146
-    },
-    {
-      "epoch": 16.765058670143414,
-      "grad_norm": 1.525972843170166,
-      "learning_rate": 6.820502512562814e-05,
-      "loss": 5.0288,
-      "step": 32147
-    },
-    {
-      "epoch": 16.765580182529334,
-      "grad_norm": 1.4325059652328491,
-      "learning_rate": 6.820402010050252e-05,
-      "loss": 5.0563,
-      "step": 32148
-    },
-    {
-      "epoch": 16.766101694915253,
-      "grad_norm": 1.4547858238220215,
-      "learning_rate": 6.820301507537688e-05,
-      "loss": 5.6503,
-      "step": 32149
-    },
-    {
-      "epoch": 16.766623207301173,
-      "grad_norm": 1.4337540864944458,
-      "learning_rate": 6.820201005025126e-05,
-      "loss": 5.752,
-      "step": 32150
-    },
-    {
-      "epoch": 16.767144719687092,
-      "grad_norm": 1.4925023317337036,
-      "learning_rate": 6.820100502512563e-05,
-      "loss": 5.1012,
-      "step": 32151
-    },
-    {
-      "epoch": 16.76766623207301,
-      "grad_norm": 1.4462909698486328,
-      "learning_rate": 6.82e-05,
-      "loss": 5.8546,
-      "step": 32152
-    },
-    {
-      "epoch": 16.76818774445893,
-      "grad_norm": 1.5949749946594238,
-      "learning_rate": 6.819899497487437e-05,
-      "loss": 4.3826,
-      "step": 32153
-    },
-    {
-      "epoch": 16.76870925684485,
-      "grad_norm": 1.4138106107711792,
-      "learning_rate": 6.819798994974875e-05,
-      "loss": 5.7878,
-      "step": 32154
-    },
-    {
-      "epoch": 16.76923076923077,
-      "grad_norm": 1.456467866897583,
-      "learning_rate": 6.819698492462312e-05,
-      "loss": 5.4614,
-      "step": 32155
-    },
-    {
-      "epoch": 16.76975228161669,
-      "grad_norm": 1.4989033937454224,
-      "learning_rate": 6.819597989949749e-05,
-      "loss": 5.5149,
-      "step": 32156
-    },
-    {
-      "epoch": 16.77027379400261,
-      "grad_norm": 1.5333174467086792,
-      "learning_rate": 6.819497487437187e-05,
-      "loss": 5.1161,
-      "step": 32157
-    },
-    {
-      "epoch": 16.77079530638853,
-      "grad_norm": 1.4783287048339844,
-      "learning_rate": 6.819396984924623e-05,
-      "loss": 5.4088,
-      "step": 32158
-    },
-    {
-      "epoch": 16.771316818774444,
-      "grad_norm": 1.4731082916259766,
-      "learning_rate": 6.819296482412061e-05,
-      "loss": 5.2375,
-      "step": 32159
-    },
-    {
-      "epoch": 16.771838331160364,
-      "grad_norm": 1.514878273010254,
-      "learning_rate": 6.819195979899497e-05,
-      "loss": 5.1704,
-      "step": 32160
-    },
-    {
-      "epoch": 16.772359843546283,
-      "grad_norm": 1.530027151107788,
-      "learning_rate": 6.819095477386935e-05,
-      "loss": 5.0929,
-      "step": 32161
-    },
-    {
-      "epoch": 16.772881355932203,
-      "grad_norm": 1.3581746816635132,
-      "learning_rate": 6.818994974874371e-05,
-      "loss": 5.7854,
-      "step": 32162
-    },
-    {
-      "epoch": 16.773402868318122,
-      "grad_norm": 1.423215389251709,
-      "learning_rate": 6.818894472361809e-05,
-      "loss": 5.1918,
-      "step": 32163
-    },
-    {
-      "epoch": 16.77392438070404,
-      "grad_norm": 1.373935580253601,
-      "learning_rate": 6.818793969849247e-05,
-      "loss": 5.18,
-      "step": 32164
-    },
-    {
-      "epoch": 16.77444589308996,
-      "grad_norm": 1.4696706533432007,
-      "learning_rate": 6.818693467336685e-05,
-      "loss": 5.4242,
-      "step": 32165
-    },
-    {
-      "epoch": 16.77496740547588,
-      "grad_norm": 1.4267797470092773,
-      "learning_rate": 6.818592964824121e-05,
-      "loss": 5.4275,
-      "step": 32166
-    },
-    {
-      "epoch": 16.7754889178618,
-      "grad_norm": 1.5421521663665771,
-      "learning_rate": 6.818492462311558e-05,
-      "loss": 4.8055,
-      "step": 32167
-    },
-    {
-      "epoch": 16.77601043024772,
-      "grad_norm": 1.459506869316101,
-      "learning_rate": 6.818391959798995e-05,
-      "loss": 5.0041,
-      "step": 32168
-    },
-    {
-      "epoch": 16.77653194263364,
-      "grad_norm": 1.4908419847488403,
-      "learning_rate": 6.818291457286432e-05,
-      "loss": 5.8071,
-      "step": 32169
-    },
-    {
-      "epoch": 16.777053455019555,
-      "grad_norm": 1.5660477876663208,
-      "learning_rate": 6.81819095477387e-05,
-      "loss": 5.2161,
-      "step": 32170
-    },
-    {
-      "epoch": 16.777574967405474,
-      "grad_norm": 1.525856375694275,
-      "learning_rate": 6.818090452261306e-05,
-      "loss": 5.703,
-      "step": 32171
-    },
-    {
-      "epoch": 16.778096479791394,
-      "grad_norm": 1.4893053770065308,
-      "learning_rate": 6.817989949748744e-05,
-      "loss": 5.5031,
-      "step": 32172
-    },
-    {
-      "epoch": 16.778617992177313,
-      "grad_norm": 1.56302809715271,
-      "learning_rate": 6.81788944723618e-05,
-      "loss": 5.1926,
-      "step": 32173
-    },
-    {
-      "epoch": 16.779139504563233,
-      "grad_norm": 1.4364020824432373,
-      "learning_rate": 6.817788944723618e-05,
-      "loss": 5.1163,
-      "step": 32174
-    },
-    {
-      "epoch": 16.779661016949152,
-      "grad_norm": 1.4721269607543945,
-      "learning_rate": 6.817688442211056e-05,
-      "loss": 5.3535,
-      "step": 32175
-    },
-    {
-      "epoch": 16.78018252933507,
-      "grad_norm": 1.5558595657348633,
-      "learning_rate": 6.817587939698494e-05,
-      "loss": 5.1643,
-      "step": 32176
-    },
-    {
-      "epoch": 16.78070404172099,
-      "grad_norm": 1.3646856546401978,
-      "learning_rate": 6.81748743718593e-05,
-      "loss": 5.4677,
-      "step": 32177
-    },
-    {
-      "epoch": 16.78122555410691,
-      "grad_norm": 1.471401572227478,
-      "learning_rate": 6.817386934673368e-05,
-      "loss": 5.5488,
-      "step": 32178
-    },
-    {
-      "epoch": 16.78174706649283,
-      "grad_norm": 1.4115381240844727,
-      "learning_rate": 6.817286432160804e-05,
-      "loss": 5.3739,
-      "step": 32179
-    },
-    {
-      "epoch": 16.78226857887875,
-      "grad_norm": 1.461485505104065,
-      "learning_rate": 6.817185929648241e-05,
-      "loss": 5.7767,
-      "step": 32180
-    },
-    {
-      "epoch": 16.78279009126467,
-      "grad_norm": 1.4169188737869263,
-      "learning_rate": 6.817085427135679e-05,
-      "loss": 5.527,
-      "step": 32181
-    },
-    {
-      "epoch": 16.78331160365059,
-      "grad_norm": 1.5787427425384521,
-      "learning_rate": 6.816984924623115e-05,
-      "loss": 4.9011,
-      "step": 32182
-    },
-    {
-      "epoch": 16.783833116036504,
-      "grad_norm": 1.3940954208374023,
-      "learning_rate": 6.816884422110553e-05,
-      "loss": 5.7509,
-      "step": 32183
-    },
-    {
-      "epoch": 16.784354628422424,
-      "grad_norm": 1.597353458404541,
-      "learning_rate": 6.81678391959799e-05,
-      "loss": 5.1301,
-      "step": 32184
-    },
-    {
-      "epoch": 16.784876140808343,
-      "grad_norm": 1.4991800785064697,
-      "learning_rate": 6.816683417085428e-05,
-      "loss": 5.2222,
-      "step": 32185
-    },
-    {
-      "epoch": 16.785397653194263,
-      "grad_norm": 1.5732795000076294,
-      "learning_rate": 6.816582914572865e-05,
-      "loss": 5.3281,
-      "step": 32186
-    },
-    {
-      "epoch": 16.785919165580182,
-      "grad_norm": 1.4665584564208984,
-      "learning_rate": 6.816482412060303e-05,
-      "loss": 4.7169,
-      "step": 32187
-    },
-    {
-      "epoch": 16.7864406779661,
-      "grad_norm": 1.464357614517212,
-      "learning_rate": 6.816381909547739e-05,
-      "loss": 5.5783,
-      "step": 32188
-    },
-    {
-      "epoch": 16.78696219035202,
-      "grad_norm": 1.4957258701324463,
-      "learning_rate": 6.816281407035177e-05,
-      "loss": 4.7752,
-      "step": 32189
-    },
-    {
-      "epoch": 16.78748370273794,
-      "grad_norm": 1.4736332893371582,
-      "learning_rate": 6.816180904522613e-05,
-      "loss": 5.5465,
-      "step": 32190
-    },
-    {
-      "epoch": 16.78800521512386,
-      "grad_norm": 1.5552585124969482,
-      "learning_rate": 6.816080402010051e-05,
-      "loss": 5.4028,
-      "step": 32191
-    },
-    {
-      "epoch": 16.78852672750978,
-      "grad_norm": 1.3648723363876343,
-      "learning_rate": 6.815979899497487e-05,
-      "loss": 5.5991,
-      "step": 32192
-    },
-    {
-      "epoch": 16.7890482398957,
-      "grad_norm": 1.5452773571014404,
-      "learning_rate": 6.815879396984925e-05,
-      "loss": 5.3919,
-      "step": 32193
-    },
-    {
-      "epoch": 16.789569752281615,
-      "grad_norm": 1.4785019159317017,
-      "learning_rate": 6.815778894472362e-05,
-      "loss": 5.259,
-      "step": 32194
-    },
-    {
-      "epoch": 16.790091264667534,
-      "grad_norm": 1.6074814796447754,
-      "learning_rate": 6.8156783919598e-05,
-      "loss": 5.3228,
-      "step": 32195
-    },
-    {
-      "epoch": 16.790612777053454,
-      "grad_norm": 1.6142023801803589,
-      "learning_rate": 6.815577889447237e-05,
-      "loss": 4.876,
-      "step": 32196
-    },
-    {
-      "epoch": 16.791134289439373,
-      "grad_norm": 1.654758095741272,
-      "learning_rate": 6.815477386934674e-05,
-      "loss": 5.301,
-      "step": 32197
-    },
-    {
-      "epoch": 16.791655801825293,
-      "grad_norm": 1.5032848119735718,
-      "learning_rate": 6.815376884422111e-05,
-      "loss": 5.715,
-      "step": 32198
-    },
-    {
-      "epoch": 16.792177314211212,
-      "grad_norm": 1.5121691226959229,
-      "learning_rate": 6.815276381909548e-05,
-      "loss": 5.2213,
-      "step": 32199
-    },
-    {
-      "epoch": 16.79269882659713,
-      "grad_norm": 1.4614659547805786,
-      "learning_rate": 6.815175879396986e-05,
-      "loss": 5.4412,
-      "step": 32200
-    },
-    {
-      "epoch": 16.79322033898305,
-      "grad_norm": 1.5777974128723145,
-      "learning_rate": 6.815075376884422e-05,
-      "loss": 5.1107,
-      "step": 32201
-    },
-    {
-      "epoch": 16.79374185136897,
-      "grad_norm": 1.6537309885025024,
-      "learning_rate": 6.81497487437186e-05,
-      "loss": 4.4437,
-      "step": 32202
-    },
-    {
-      "epoch": 16.79426336375489,
-      "grad_norm": 1.432127833366394,
-      "learning_rate": 6.814874371859296e-05,
-      "loss": 5.7576,
-      "step": 32203
-    },
-    {
-      "epoch": 16.79478487614081,
-      "grad_norm": 1.5189236402511597,
-      "learning_rate": 6.814773869346734e-05,
-      "loss": 5.2628,
-      "step": 32204
-    },
-    {
-      "epoch": 16.79530638852673,
-      "grad_norm": 1.5158084630966187,
-      "learning_rate": 6.814673366834172e-05,
-      "loss": 5.1339,
-      "step": 32205
-    },
-    {
-      "epoch": 16.795827900912645,
-      "grad_norm": 1.518705129623413,
-      "learning_rate": 6.81457286432161e-05,
-      "loss": 5.0622,
-      "step": 32206
-    },
-    {
-      "epoch": 16.796349413298564,
-      "grad_norm": 1.4828519821166992,
-      "learning_rate": 6.814472361809046e-05,
-      "loss": 5.4874,
-      "step": 32207
-    },
-    {
-      "epoch": 16.796870925684484,
-      "grad_norm": 1.5688568353652954,
-      "learning_rate": 6.814371859296482e-05,
-      "loss": 5.4659,
-      "step": 32208
-    },
-    {
-      "epoch": 16.797392438070403,
-      "grad_norm": 1.4416998624801636,
-      "learning_rate": 6.81427135678392e-05,
-      "loss": 5.5827,
-      "step": 32209
-    },
-    {
-      "epoch": 16.797913950456323,
-      "grad_norm": 1.6394208669662476,
-      "learning_rate": 6.814170854271357e-05,
-      "loss": 4.8147,
-      "step": 32210
-    },
-    {
-      "epoch": 16.798435462842242,
-      "grad_norm": 1.3674403429031372,
-      "learning_rate": 6.814070351758794e-05,
-      "loss": 5.8231,
-      "step": 32211
-    },
-    {
-      "epoch": 16.798956975228162,
-      "grad_norm": 1.4532010555267334,
-      "learning_rate": 6.813969849246231e-05,
-      "loss": 5.5624,
-      "step": 32212
-    },
-    {
-      "epoch": 16.79947848761408,
-      "grad_norm": 1.550743818283081,
-      "learning_rate": 6.813869346733669e-05,
-      "loss": 5.2968,
-      "step": 32213
-    },
-    {
-      "epoch": 16.8,
-      "grad_norm": 1.5190784931182861,
-      "learning_rate": 6.813768844221105e-05,
-      "loss": 5.3274,
-      "step": 32214
-    },
-    {
-      "epoch": 16.80052151238592,
-      "grad_norm": 1.59329092502594,
-      "learning_rate": 6.813668341708543e-05,
-      "loss": 5.4457,
-      "step": 32215
-    },
-    {
-      "epoch": 16.80104302477184,
-      "grad_norm": 1.5243127346038818,
-      "learning_rate": 6.81356783919598e-05,
-      "loss": 4.6692,
-      "step": 32216
-    },
-    {
-      "epoch": 16.80156453715776,
-      "grad_norm": 1.5941133499145508,
-      "learning_rate": 6.813467336683418e-05,
-      "loss": 5.2067,
-      "step": 32217
-    },
-    {
-      "epoch": 16.802086049543675,
-      "grad_norm": 1.618507981300354,
-      "learning_rate": 6.813366834170855e-05,
-      "loss": 5.1478,
-      "step": 32218
-    },
-    {
-      "epoch": 16.802607561929594,
-      "grad_norm": 1.4811009168624878,
-      "learning_rate": 6.813266331658293e-05,
-      "loss": 5.576,
-      "step": 32219
-    },
-    {
-      "epoch": 16.803129074315514,
-      "grad_norm": 1.4906744956970215,
-      "learning_rate": 6.813165829145729e-05,
-      "loss": 5.5973,
-      "step": 32220
-    },
-    {
-      "epoch": 16.803650586701433,
-      "grad_norm": 1.5996023416519165,
-      "learning_rate": 6.813065326633165e-05,
-      "loss": 5.5371,
-      "step": 32221
-    },
-    {
-      "epoch": 16.804172099087353,
-      "grad_norm": 1.4527695178985596,
-      "learning_rate": 6.812964824120603e-05,
-      "loss": 5.2659,
-      "step": 32222
-    },
-    {
-      "epoch": 16.804693611473272,
-      "grad_norm": 1.4184855222702026,
-      "learning_rate": 6.81286432160804e-05,
-      "loss": 5.5417,
-      "step": 32223
-    },
-    {
-      "epoch": 16.805215123859192,
-      "grad_norm": 1.4366674423217773,
-      "learning_rate": 6.812763819095477e-05,
-      "loss": 5.5835,
-      "step": 32224
-    },
-    {
-      "epoch": 16.80573663624511,
-      "grad_norm": 1.4732416868209839,
-      "learning_rate": 6.812663316582915e-05,
-      "loss": 5.4492,
-      "step": 32225
-    },
-    {
-      "epoch": 16.80625814863103,
-      "grad_norm": 1.459498643875122,
-      "learning_rate": 6.812562814070353e-05,
-      "loss": 5.2794,
-      "step": 32226
-    },
-    {
-      "epoch": 16.80677966101695,
-      "grad_norm": 1.4049906730651855,
-      "learning_rate": 6.81246231155779e-05,
-      "loss": 5.5433,
-      "step": 32227
-    },
-    {
-      "epoch": 16.80730117340287,
-      "grad_norm": 1.4600849151611328,
-      "learning_rate": 6.812361809045227e-05,
-      "loss": 5.4881,
-      "step": 32228
-    },
-    {
-      "epoch": 16.80782268578879,
-      "grad_norm": 1.5257198810577393,
-      "learning_rate": 6.812261306532664e-05,
-      "loss": 5.4153,
-      "step": 32229
-    },
-    {
-      "epoch": 16.808344198174705,
-      "grad_norm": 1.4667681455612183,
-      "learning_rate": 6.812160804020101e-05,
-      "loss": 5.3627,
-      "step": 32230
-    },
-    {
-      "epoch": 16.808865710560625,
-      "grad_norm": 1.4093654155731201,
-      "learning_rate": 6.812060301507538e-05,
-      "loss": 5.255,
-      "step": 32231
-    },
-    {
-      "epoch": 16.809387222946544,
-      "grad_norm": 1.4354609251022339,
-      "learning_rate": 6.811959798994976e-05,
-      "loss": 5.409,
-      "step": 32232
-    },
-    {
-      "epoch": 16.809908735332463,
-      "grad_norm": 1.4511804580688477,
-      "learning_rate": 6.811859296482412e-05,
-      "loss": 5.2829,
-      "step": 32233
-    },
-    {
-      "epoch": 16.810430247718383,
-      "grad_norm": 1.631495475769043,
-      "learning_rate": 6.811758793969848e-05,
-      "loss": 4.9771,
-      "step": 32234
-    },
-    {
-      "epoch": 16.810951760104302,
-      "grad_norm": 1.4254612922668457,
-      "learning_rate": 6.811658291457286e-05,
-      "loss": 5.5648,
-      "step": 32235
-    },
-    {
-      "epoch": 16.811473272490222,
-      "grad_norm": 1.4607248306274414,
-      "learning_rate": 6.811557788944724e-05,
-      "loss": 5.1943,
-      "step": 32236
-    },
-    {
-      "epoch": 16.81199478487614,
-      "grad_norm": 1.4698255062103271,
-      "learning_rate": 6.811457286432162e-05,
-      "loss": 5.2882,
-      "step": 32237
-    },
-    {
-      "epoch": 16.81251629726206,
-      "grad_norm": 1.430342674255371,
-      "learning_rate": 6.811356783919598e-05,
-      "loss": 5.0732,
-      "step": 32238
-    },
-    {
-      "epoch": 16.81303780964798,
-      "grad_norm": 1.530485987663269,
-      "learning_rate": 6.811256281407036e-05,
-      "loss": 5.3367,
-      "step": 32239
-    },
-    {
-      "epoch": 16.8135593220339,
-      "grad_norm": 1.4782066345214844,
-      "learning_rate": 6.811155778894472e-05,
-      "loss": 5.3626,
-      "step": 32240
-    },
-    {
-      "epoch": 16.81408083441982,
-      "grad_norm": 1.5078420639038086,
-      "learning_rate": 6.81105527638191e-05,
-      "loss": 4.5488,
-      "step": 32241
-    },
-    {
-      "epoch": 16.814602346805735,
-      "grad_norm": 1.3685628175735474,
-      "learning_rate": 6.810954773869347e-05,
-      "loss": 5.7081,
-      "step": 32242
-    },
-    {
-      "epoch": 16.815123859191655,
-      "grad_norm": 1.5352864265441895,
-      "learning_rate": 6.810854271356784e-05,
-      "loss": 5.4687,
-      "step": 32243
-    },
-    {
-      "epoch": 16.815645371577574,
-      "grad_norm": 1.4922908544540405,
-      "learning_rate": 6.810753768844221e-05,
-      "loss": 5.3035,
-      "step": 32244
-    },
-    {
-      "epoch": 16.816166883963493,
-      "grad_norm": 1.6320977210998535,
-      "learning_rate": 6.810653266331659e-05,
-      "loss": 4.9977,
-      "step": 32245
-    },
-    {
-      "epoch": 16.816688396349413,
-      "grad_norm": 1.4462674856185913,
-      "learning_rate": 6.810552763819096e-05,
-      "loss": 4.7986,
-      "step": 32246
-    },
-    {
-      "epoch": 16.817209908735332,
-      "grad_norm": 1.4855844974517822,
-      "learning_rate": 6.810452261306533e-05,
-      "loss": 5.1601,
-      "step": 32247
-    },
-    {
-      "epoch": 16.817731421121252,
-      "grad_norm": 1.5015418529510498,
-      "learning_rate": 6.81035175879397e-05,
-      "loss": 4.8939,
-      "step": 32248
-    },
-    {
-      "epoch": 16.81825293350717,
-      "grad_norm": 1.4044885635375977,
-      "learning_rate": 6.810251256281407e-05,
-      "loss": 5.5958,
-      "step": 32249
-    },
-    {
-      "epoch": 16.81877444589309,
-      "grad_norm": 1.5564625263214111,
-      "learning_rate": 6.810150753768845e-05,
-      "loss": 5.1493,
-      "step": 32250
-    },
-    {
-      "epoch": 16.81929595827901,
-      "grad_norm": 1.462203860282898,
-      "learning_rate": 6.810050251256281e-05,
-      "loss": 5.6381,
-      "step": 32251
-    },
-    {
-      "epoch": 16.81981747066493,
-      "grad_norm": 1.4473809003829956,
-      "learning_rate": 6.809949748743719e-05,
-      "loss": 4.9622,
-      "step": 32252
-    },
-    {
-      "epoch": 16.820338983050846,
-      "grad_norm": 1.4460340738296509,
-      "learning_rate": 6.809849246231156e-05,
-      "loss": 5.5568,
-      "step": 32253
-    },
-    {
-      "epoch": 16.820860495436765,
-      "grad_norm": 1.7312417030334473,
-      "learning_rate": 6.809748743718593e-05,
-      "loss": 4.3886,
-      "step": 32254
-    },
-    {
-      "epoch": 16.821382007822685,
-      "grad_norm": 1.464523434638977,
-      "learning_rate": 6.80964824120603e-05,
-      "loss": 5.8156,
-      "step": 32255
-    },
-    {
-      "epoch": 16.821903520208604,
-      "grad_norm": 1.4283992052078247,
-      "learning_rate": 6.809547738693468e-05,
-      "loss": 5.4809,
-      "step": 32256
-    },
-    {
-      "epoch": 16.822425032594523,
-      "grad_norm": 1.3495358228683472,
-      "learning_rate": 6.809447236180905e-05,
-      "loss": 5.6428,
-      "step": 32257
-    },
-    {
-      "epoch": 16.822946544980443,
-      "grad_norm": 1.57086181640625,
-      "learning_rate": 6.809346733668343e-05,
-      "loss": 5.3073,
-      "step": 32258
-    },
-    {
-      "epoch": 16.823468057366362,
-      "grad_norm": 1.4902983903884888,
-      "learning_rate": 6.80924623115578e-05,
-      "loss": 5.068,
-      "step": 32259
-    },
-    {
-      "epoch": 16.823989569752282,
-      "grad_norm": 1.3659354448318481,
-      "learning_rate": 6.809145728643216e-05,
-      "loss": 5.3183,
-      "step": 32260
-    },
-    {
-      "epoch": 16.8245110821382,
-      "grad_norm": 1.4588451385498047,
-      "learning_rate": 6.809045226130654e-05,
-      "loss": 5.338,
-      "step": 32261
-    },
-    {
-      "epoch": 16.82503259452412,
-      "grad_norm": 1.4269752502441406,
-      "learning_rate": 6.80894472361809e-05,
-      "loss": 5.5153,
-      "step": 32262
-    },
-    {
-      "epoch": 16.82555410691004,
-      "grad_norm": 1.5318281650543213,
-      "learning_rate": 6.808844221105528e-05,
-      "loss": 5.4477,
-      "step": 32263
-    },
-    {
-      "epoch": 16.82607561929596,
-      "grad_norm": 1.4717127084732056,
-      "learning_rate": 6.808743718592964e-05,
-      "loss": 5.1391,
-      "step": 32264
-    },
-    {
-      "epoch": 16.82659713168188,
-      "grad_norm": 1.3730554580688477,
-      "learning_rate": 6.808643216080402e-05,
-      "loss": 5.8,
-      "step": 32265
-    },
-    {
-      "epoch": 16.827118644067795,
-      "grad_norm": 1.4311480522155762,
-      "learning_rate": 6.80854271356784e-05,
-      "loss": 5.3432,
-      "step": 32266
-    },
-    {
-      "epoch": 16.827640156453715,
-      "grad_norm": 1.4840831756591797,
-      "learning_rate": 6.808442211055278e-05,
-      "loss": 5.3272,
-      "step": 32267
-    },
-    {
-      "epoch": 16.828161668839634,
-      "grad_norm": 1.5875296592712402,
-      "learning_rate": 6.808341708542714e-05,
-      "loss": 5.3552,
-      "step": 32268
-    },
-    {
-      "epoch": 16.828683181225554,
-      "grad_norm": 1.4675390720367432,
-      "learning_rate": 6.808241206030152e-05,
-      "loss": 5.5292,
-      "step": 32269
-    },
-    {
-      "epoch": 16.829204693611473,
-      "grad_norm": 1.487448811531067,
-      "learning_rate": 6.808140703517588e-05,
-      "loss": 5.655,
-      "step": 32270
-    },
-    {
-      "epoch": 16.829726205997392,
-      "grad_norm": 1.6335320472717285,
-      "learning_rate": 6.808040201005026e-05,
-      "loss": 5.4538,
-      "step": 32271
-    },
-    {
-      "epoch": 16.830247718383312,
-      "grad_norm": 1.5432835817337036,
-      "learning_rate": 6.807939698492463e-05,
-      "loss": 5.1874,
-      "step": 32272
-    },
-    {
-      "epoch": 16.83076923076923,
-      "grad_norm": 1.6164653301239014,
-      "learning_rate": 6.807839195979899e-05,
-      "loss": 4.7992,
-      "step": 32273
-    },
-    {
-      "epoch": 16.83129074315515,
-      "grad_norm": 1.5519556999206543,
-      "learning_rate": 6.807738693467337e-05,
-      "loss": 5.6583,
-      "step": 32274
-    },
-    {
-      "epoch": 16.83181225554107,
-      "grad_norm": 1.4818847179412842,
-      "learning_rate": 6.807638190954773e-05,
-      "loss": 5.1017,
-      "step": 32275
-    },
-    {
-      "epoch": 16.83233376792699,
-      "grad_norm": 1.4086637496948242,
-      "learning_rate": 6.807537688442211e-05,
-      "loss": 5.2852,
-      "step": 32276
-    },
-    {
-      "epoch": 16.832855280312906,
-      "grad_norm": 1.4884984493255615,
-      "learning_rate": 6.807437185929649e-05,
-      "loss": 5.5994,
-      "step": 32277
-    },
-    {
-      "epoch": 16.833376792698825,
-      "grad_norm": 1.402036190032959,
-      "learning_rate": 6.807336683417087e-05,
-      "loss": 5.5945,
-      "step": 32278
-    },
-    {
-      "epoch": 16.833898305084745,
-      "grad_norm": 1.3730614185333252,
-      "learning_rate": 6.807236180904523e-05,
-      "loss": 5.6016,
-      "step": 32279
-    },
-    {
-      "epoch": 16.834419817470664,
-      "grad_norm": 1.4380699396133423,
-      "learning_rate": 6.807135678391961e-05,
-      "loss": 5.6197,
-      "step": 32280
-    },
-    {
-      "epoch": 16.834941329856584,
-      "grad_norm": 1.4394521713256836,
-      "learning_rate": 6.807035175879397e-05,
-      "loss": 5.7669,
-      "step": 32281
-    },
-    {
-      "epoch": 16.835462842242503,
-      "grad_norm": 1.666674017906189,
-      "learning_rate": 6.806934673366835e-05,
-      "loss": 5.1824,
-      "step": 32282
-    },
-    {
-      "epoch": 16.835984354628422,
-      "grad_norm": 1.5262519121170044,
-      "learning_rate": 6.806834170854271e-05,
-      "loss": 5.6693,
-      "step": 32283
-    },
-    {
-      "epoch": 16.836505867014342,
-      "grad_norm": 1.408313512802124,
-      "learning_rate": 6.806733668341709e-05,
-      "loss": 5.3513,
-      "step": 32284
-    },
-    {
-      "epoch": 16.83702737940026,
-      "grad_norm": 1.4788936376571655,
-      "learning_rate": 6.806633165829146e-05,
-      "loss": 5.236,
-      "step": 32285
-    },
-    {
-      "epoch": 16.83754889178618,
-      "grad_norm": 1.5548593997955322,
-      "learning_rate": 6.806532663316583e-05,
-      "loss": 5.3283,
-      "step": 32286
-    },
-    {
-      "epoch": 16.8380704041721,
-      "grad_norm": 1.4647866487503052,
-      "learning_rate": 6.80643216080402e-05,
-      "loss": 5.5431,
-      "step": 32287
-    },
-    {
-      "epoch": 16.83859191655802,
-      "grad_norm": 1.8245185613632202,
-      "learning_rate": 6.806331658291458e-05,
-      "loss": 4.4265,
-      "step": 32288
-    },
-    {
-      "epoch": 16.839113428943936,
-      "grad_norm": 1.6040338277816772,
-      "learning_rate": 6.806231155778895e-05,
-      "loss": 5.1472,
-      "step": 32289
-    },
-    {
-      "epoch": 16.839634941329855,
-      "grad_norm": 1.5158417224884033,
-      "learning_rate": 6.806130653266332e-05,
-      "loss": 5.5369,
-      "step": 32290
-    },
-    {
-      "epoch": 16.840156453715775,
-      "grad_norm": 1.4937772750854492,
-      "learning_rate": 6.80603015075377e-05,
-      "loss": 5.5289,
-      "step": 32291
-    },
-    {
-      "epoch": 16.840677966101694,
-      "grad_norm": 1.5024772882461548,
-      "learning_rate": 6.805929648241206e-05,
-      "loss": 5.3196,
-      "step": 32292
-    },
-    {
-      "epoch": 16.841199478487614,
-      "grad_norm": 1.4447509050369263,
-      "learning_rate": 6.805829145728644e-05,
-      "loss": 5.7246,
-      "step": 32293
-    },
-    {
-      "epoch": 16.841720990873533,
-      "grad_norm": 1.552960991859436,
-      "learning_rate": 6.80572864321608e-05,
-      "loss": 5.2145,
-      "step": 32294
-    },
-    {
-      "epoch": 16.842242503259452,
-      "grad_norm": 1.5452227592468262,
-      "learning_rate": 6.805628140703518e-05,
-      "loss": 4.8153,
-      "step": 32295
-    },
-    {
-      "epoch": 16.842764015645372,
-      "grad_norm": 1.6260719299316406,
-      "learning_rate": 6.805527638190954e-05,
-      "loss": 4.9387,
-      "step": 32296
-    },
-    {
-      "epoch": 16.84328552803129,
-      "grad_norm": 1.5294049978256226,
-      "learning_rate": 6.805427135678392e-05,
-      "loss": 4.9174,
-      "step": 32297
-    },
-    {
-      "epoch": 16.84380704041721,
-      "grad_norm": 1.4775010347366333,
-      "learning_rate": 6.80532663316583e-05,
-      "loss": 5.4644,
-      "step": 32298
-    },
-    {
-      "epoch": 16.84432855280313,
-      "grad_norm": 1.3698045015335083,
-      "learning_rate": 6.805226130653268e-05,
-      "loss": 5.6984,
-      "step": 32299
-    },
-    {
-      "epoch": 16.84485006518905,
-      "grad_norm": 1.4460890293121338,
-      "learning_rate": 6.805125628140704e-05,
-      "loss": 5.526,
-      "step": 32300
-    },
-    {
-      "epoch": 16.845371577574966,
-      "grad_norm": 1.4469985961914062,
-      "learning_rate": 6.80502512562814e-05,
-      "loss": 5.3451,
-      "step": 32301
-    },
-    {
-      "epoch": 16.845893089960885,
-      "grad_norm": 1.4390296936035156,
-      "learning_rate": 6.804924623115578e-05,
-      "loss": 5.4194,
-      "step": 32302
-    },
-    {
-      "epoch": 16.846414602346805,
-      "grad_norm": 1.5403863191604614,
-      "learning_rate": 6.804824120603015e-05,
-      "loss": 5.2988,
-      "step": 32303
-    },
-    {
-      "epoch": 16.846936114732724,
-      "grad_norm": 1.4167232513427734,
-      "learning_rate": 6.804723618090453e-05,
-      "loss": 5.5034,
-      "step": 32304
-    },
-    {
-      "epoch": 16.847457627118644,
-      "grad_norm": 1.3592467308044434,
-      "learning_rate": 6.804623115577889e-05,
-      "loss": 5.4431,
-      "step": 32305
-    },
-    {
-      "epoch": 16.847979139504563,
-      "grad_norm": 1.4841742515563965,
-      "learning_rate": 6.804522613065327e-05,
-      "loss": 5.5803,
-      "step": 32306
-    },
-    {
-      "epoch": 16.848500651890483,
-      "grad_norm": 1.4531583786010742,
-      "learning_rate": 6.804422110552763e-05,
-      "loss": 5.0393,
-      "step": 32307
-    },
-    {
-      "epoch": 16.849022164276402,
-      "grad_norm": 1.4909645318984985,
-      "learning_rate": 6.804321608040201e-05,
-      "loss": 5.2716,
-      "step": 32308
-    },
-    {
-      "epoch": 16.84954367666232,
-      "grad_norm": 1.438641905784607,
-      "learning_rate": 6.804221105527639e-05,
-      "loss": 5.451,
-      "step": 32309
-    },
-    {
-      "epoch": 16.85006518904824,
-      "grad_norm": 1.450132966041565,
-      "learning_rate": 6.804120603015077e-05,
-      "loss": 5.0258,
-      "step": 32310
-    },
-    {
-      "epoch": 16.85058670143416,
-      "grad_norm": 1.4128117561340332,
-      "learning_rate": 6.804020100502513e-05,
-      "loss": 5.6001,
-      "step": 32311
-    },
-    {
-      "epoch": 16.85110821382008,
-      "grad_norm": 1.4158730506896973,
-      "learning_rate": 6.803919597989951e-05,
-      "loss": 5.2885,
-      "step": 32312
-    },
-    {
-      "epoch": 16.851629726205996,
-      "grad_norm": 1.517215609550476,
-      "learning_rate": 6.803819095477387e-05,
-      "loss": 5.1261,
-      "step": 32313
-    },
-    {
-      "epoch": 16.852151238591915,
-      "grad_norm": 1.461003303527832,
-      "learning_rate": 6.803718592964824e-05,
-      "loss": 5.5772,
-      "step": 32314
-    },
-    {
-      "epoch": 16.852672750977835,
-      "grad_norm": 1.5857937335968018,
-      "learning_rate": 6.803618090452261e-05,
-      "loss": 5.1324,
-      "step": 32315
-    },
-    {
-      "epoch": 16.853194263363754,
-      "grad_norm": 1.4597573280334473,
-      "learning_rate": 6.803517587939698e-05,
-      "loss": 5.5591,
-      "step": 32316
-    },
-    {
-      "epoch": 16.853715775749674,
-      "grad_norm": 1.64104425907135,
-      "learning_rate": 6.803417085427136e-05,
-      "loss": 4.8194,
-      "step": 32317
-    },
-    {
-      "epoch": 16.854237288135593,
-      "grad_norm": 1.4670978784561157,
-      "learning_rate": 6.803316582914573e-05,
-      "loss": 5.3145,
-      "step": 32318
-    },
-    {
-      "epoch": 16.854758800521513,
-      "grad_norm": 1.5061482191085815,
-      "learning_rate": 6.803216080402011e-05,
-      "loss": 5.5036,
-      "step": 32319
-    },
-    {
-      "epoch": 16.855280312907432,
-      "grad_norm": 1.5042190551757812,
-      "learning_rate": 6.803115577889448e-05,
-      "loss": 4.8788,
-      "step": 32320
-    },
-    {
-      "epoch": 16.85580182529335,
-      "grad_norm": 1.4566477537155151,
-      "learning_rate": 6.803015075376885e-05,
-      "loss": 5.5967,
-      "step": 32321
-    },
-    {
-      "epoch": 16.85632333767927,
-      "grad_norm": 1.5195703506469727,
-      "learning_rate": 6.802914572864322e-05,
-      "loss": 5.5489,
-      "step": 32322
-    },
-    {
-      "epoch": 16.85684485006519,
-      "grad_norm": 1.4620803594589233,
-      "learning_rate": 6.80281407035176e-05,
-      "loss": 5.4269,
-      "step": 32323
-    },
-    {
-      "epoch": 16.85736636245111,
-      "grad_norm": 1.5707426071166992,
-      "learning_rate": 6.802713567839196e-05,
-      "loss": 5.123,
-      "step": 32324
-    },
-    {
-      "epoch": 16.857887874837026,
-      "grad_norm": 1.450875163078308,
-      "learning_rate": 6.802613065326634e-05,
-      "loss": 5.3864,
-      "step": 32325
-    },
-    {
-      "epoch": 16.858409387222945,
-      "grad_norm": 1.4025403261184692,
-      "learning_rate": 6.80251256281407e-05,
-      "loss": 5.7229,
-      "step": 32326
-    },
-    {
-      "epoch": 16.858930899608865,
-      "grad_norm": 1.6100084781646729,
-      "learning_rate": 6.802412060301507e-05,
-      "loss": 5.0825,
-      "step": 32327
-    },
-    {
-      "epoch": 16.859452411994784,
-      "grad_norm": 1.498835563659668,
-      "learning_rate": 6.802311557788944e-05,
-      "loss": 5.3448,
-      "step": 32328
-    },
-    {
-      "epoch": 16.859973924380704,
-      "grad_norm": 1.4917734861373901,
-      "learning_rate": 6.802211055276382e-05,
-      "loss": 5.1804,
-      "step": 32329
-    },
-    {
-      "epoch": 16.860495436766623,
-      "grad_norm": 1.5474884510040283,
-      "learning_rate": 6.80211055276382e-05,
-      "loss": 4.9646,
-      "step": 32330
-    },
-    {
-      "epoch": 16.861016949152543,
-      "grad_norm": 1.5999699831008911,
-      "learning_rate": 6.802010050251256e-05,
-      "loss": 5.3083,
-      "step": 32331
-    },
-    {
-      "epoch": 16.861538461538462,
-      "grad_norm": 1.524290919303894,
-      "learning_rate": 6.801909547738694e-05,
-      "loss": 5.7092,
-      "step": 32332
-    },
-    {
-      "epoch": 16.86205997392438,
-      "grad_norm": 1.4705445766448975,
-      "learning_rate": 6.801809045226131e-05,
-      "loss": 5.32,
-      "step": 32333
-    },
-    {
-      "epoch": 16.8625814863103,
-      "grad_norm": 1.531370997428894,
-      "learning_rate": 6.801708542713568e-05,
-      "loss": 5.0256,
-      "step": 32334
-    },
-    {
-      "epoch": 16.86310299869622,
-      "grad_norm": 1.4705036878585815,
-      "learning_rate": 6.801608040201005e-05,
-      "loss": 5.2815,
-      "step": 32335
-    },
-    {
-      "epoch": 16.86362451108214,
-      "grad_norm": 1.4154504537582397,
-      "learning_rate": 6.801507537688443e-05,
-      "loss": 5.7015,
-      "step": 32336
-    },
-    {
-      "epoch": 16.864146023468056,
-      "grad_norm": 1.4681625366210938,
-      "learning_rate": 6.801407035175879e-05,
-      "loss": 5.0413,
-      "step": 32337
-    },
-    {
-      "epoch": 16.864667535853975,
-      "grad_norm": 1.5421240329742432,
-      "learning_rate": 6.801306532663317e-05,
-      "loss": 5.3368,
-      "step": 32338
-    },
-    {
-      "epoch": 16.865189048239895,
-      "grad_norm": 1.5045236349105835,
-      "learning_rate": 6.801206030150755e-05,
-      "loss": 5.4863,
-      "step": 32339
-    },
-    {
-      "epoch": 16.865710560625814,
-      "grad_norm": 1.504016637802124,
-      "learning_rate": 6.801105527638191e-05,
-      "loss": 5.5495,
-      "step": 32340
-    },
-    {
-      "epoch": 16.866232073011734,
-      "grad_norm": 1.4321191310882568,
-      "learning_rate": 6.801005025125629e-05,
-      "loss": 5.5052,
-      "step": 32341
-    },
-    {
-      "epoch": 16.866753585397653,
-      "grad_norm": 1.4413570165634155,
-      "learning_rate": 6.800904522613065e-05,
-      "loss": 5.5977,
-      "step": 32342
-    },
-    {
-      "epoch": 16.867275097783573,
-      "grad_norm": 1.4629067182540894,
-      "learning_rate": 6.800804020100503e-05,
-      "loss": 5.2107,
-      "step": 32343
-    },
-    {
-      "epoch": 16.867796610169492,
-      "grad_norm": 1.4168341159820557,
-      "learning_rate": 6.80070351758794e-05,
-      "loss": 5.6439,
-      "step": 32344
-    },
-    {
-      "epoch": 16.86831812255541,
-      "grad_norm": 1.4293676614761353,
-      "learning_rate": 6.800603015075377e-05,
-      "loss": 5.6695,
-      "step": 32345
-    },
-    {
-      "epoch": 16.86883963494133,
-      "grad_norm": 1.4909430742263794,
-      "learning_rate": 6.800502512562814e-05,
-      "loss": 5.4145,
-      "step": 32346
-    },
-    {
-      "epoch": 16.86936114732725,
-      "grad_norm": 1.5386234521865845,
-      "learning_rate": 6.800402010050252e-05,
-      "loss": 5.3797,
-      "step": 32347
-    },
-    {
-      "epoch": 16.86988265971317,
-      "grad_norm": 1.4167425632476807,
-      "learning_rate": 6.800301507537688e-05,
-      "loss": 5.313,
-      "step": 32348
-    },
-    {
-      "epoch": 16.870404172099086,
-      "grad_norm": 1.4678267240524292,
-      "learning_rate": 6.800201005025126e-05,
-      "loss": 5.2998,
-      "step": 32349
-    },
-    {
-      "epoch": 16.870925684485005,
-      "grad_norm": 1.3805166482925415,
-      "learning_rate": 6.800100502512564e-05,
-      "loss": 5.5877,
-      "step": 32350
-    },
-    {
-      "epoch": 16.871447196870925,
-      "grad_norm": 1.5353916883468628,
-      "learning_rate": 6.800000000000001e-05,
-      "loss": 5.0314,
-      "step": 32351
-    },
-    {
-      "epoch": 16.871968709256844,
-      "grad_norm": 1.4218164682388306,
-      "learning_rate": 6.799899497487438e-05,
-      "loss": 5.3752,
-      "step": 32352
-    },
-    {
-      "epoch": 16.872490221642764,
-      "grad_norm": 1.5239075422286987,
-      "learning_rate": 6.799798994974874e-05,
-      "loss": 5.5058,
-      "step": 32353
-    },
-    {
-      "epoch": 16.873011734028683,
-      "grad_norm": 1.4978108406066895,
-      "learning_rate": 6.799698492462312e-05,
-      "loss": 5.4224,
-      "step": 32354
-    },
-    {
-      "epoch": 16.873533246414603,
-      "grad_norm": 1.5130846500396729,
-      "learning_rate": 6.799597989949748e-05,
-      "loss": 5.5021,
-      "step": 32355
-    },
-    {
-      "epoch": 16.874054758800522,
-      "grad_norm": 1.3958134651184082,
-      "learning_rate": 6.799497487437186e-05,
-      "loss": 5.2377,
-      "step": 32356
-    },
-    {
-      "epoch": 16.87457627118644,
-      "grad_norm": 1.5282251834869385,
-      "learning_rate": 6.799396984924623e-05,
-      "loss": 4.5441,
-      "step": 32357
-    },
-    {
-      "epoch": 16.87509778357236,
-      "grad_norm": 1.4294945001602173,
-      "learning_rate": 6.79929648241206e-05,
-      "loss": 5.8107,
-      "step": 32358
-    },
-    {
-      "epoch": 16.87561929595828,
-      "grad_norm": 1.5477867126464844,
-      "learning_rate": 6.799195979899498e-05,
-      "loss": 5.2826,
-      "step": 32359
-    },
-    {
-      "epoch": 16.876140808344196,
-      "grad_norm": 1.4709736108779907,
-      "learning_rate": 6.799095477386936e-05,
-      "loss": 5.34,
-      "step": 32360
-    },
-    {
-      "epoch": 16.876662320730116,
-      "grad_norm": 1.5822782516479492,
-      "learning_rate": 6.798994974874372e-05,
-      "loss": 5.1394,
-      "step": 32361
-    },
-    {
-      "epoch": 16.877183833116035,
-      "grad_norm": 1.4112669229507446,
-      "learning_rate": 6.79889447236181e-05,
-      "loss": 5.4785,
-      "step": 32362
-    },
-    {
-      "epoch": 16.877705345501955,
-      "grad_norm": 1.5523914098739624,
-      "learning_rate": 6.798793969849247e-05,
-      "loss": 5.1709,
-      "step": 32363
-    },
-    {
-      "epoch": 16.878226857887874,
-      "grad_norm": 1.41044282913208,
-      "learning_rate": 6.798693467336684e-05,
-      "loss": 5.6549,
-      "step": 32364
-    },
-    {
-      "epoch": 16.878748370273794,
-      "grad_norm": 1.4799189567565918,
-      "learning_rate": 6.798592964824121e-05,
-      "loss": 5.4895,
-      "step": 32365
-    },
-    {
-      "epoch": 16.879269882659713,
-      "grad_norm": 1.428674340248108,
-      "learning_rate": 6.798492462311557e-05,
-      "loss": 5.5695,
-      "step": 32366
-    },
-    {
-      "epoch": 16.879791395045633,
-      "grad_norm": 1.3389296531677246,
-      "learning_rate": 6.798391959798995e-05,
-      "loss": 5.7833,
-      "step": 32367
-    },
-    {
-      "epoch": 16.880312907431552,
-      "grad_norm": 1.4089906215667725,
-      "learning_rate": 6.798291457286431e-05,
-      "loss": 5.1084,
-      "step": 32368
-    },
-    {
-      "epoch": 16.88083441981747,
-      "grad_norm": 1.5674139261245728,
-      "learning_rate": 6.798190954773869e-05,
-      "loss": 5.8572,
-      "step": 32369
-    },
-    {
-      "epoch": 16.88135593220339,
-      "grad_norm": 1.4656537771224976,
-      "learning_rate": 6.798090452261307e-05,
-      "loss": 5.4551,
-      "step": 32370
-    },
-    {
-      "epoch": 16.88187744458931,
-      "grad_norm": 1.4765628576278687,
-      "learning_rate": 6.797989949748745e-05,
-      "loss": 5.3568,
-      "step": 32371
-    },
-    {
-      "epoch": 16.88239895697523,
-      "grad_norm": 1.5358625650405884,
-      "learning_rate": 6.797889447236181e-05,
-      "loss": 5.1087,
-      "step": 32372
-    },
-    {
-      "epoch": 16.882920469361146,
-      "grad_norm": 1.393275260925293,
-      "learning_rate": 6.797788944723619e-05,
-      "loss": 5.8279,
-      "step": 32373
-    },
-    {
-      "epoch": 16.883441981747065,
-      "grad_norm": 1.5468108654022217,
-      "learning_rate": 6.797688442211055e-05,
-      "loss": 5.0335,
-      "step": 32374
-    },
-    {
-      "epoch": 16.883963494132985,
-      "grad_norm": 1.4667243957519531,
-      "learning_rate": 6.797587939698493e-05,
-      "loss": 4.9925,
-      "step": 32375
-    },
-    {
-      "epoch": 16.884485006518904,
-      "grad_norm": 1.502632737159729,
-      "learning_rate": 6.79748743718593e-05,
-      "loss": 5.0556,
-      "step": 32376
-    },
-    {
-      "epoch": 16.885006518904824,
-      "grad_norm": 1.4619288444519043,
-      "learning_rate": 6.797386934673367e-05,
-      "loss": 5.7896,
-      "step": 32377
-    },
-    {
-      "epoch": 16.885528031290743,
-      "grad_norm": 1.46075439453125,
-      "learning_rate": 6.797286432160804e-05,
-      "loss": 5.3907,
-      "step": 32378
-    },
-    {
-      "epoch": 16.886049543676663,
-      "grad_norm": 1.5204224586486816,
-      "learning_rate": 6.797185929648242e-05,
-      "loss": 5.2373,
-      "step": 32379
-    },
-    {
-      "epoch": 16.886571056062582,
-      "grad_norm": 1.5632343292236328,
-      "learning_rate": 6.79708542713568e-05,
-      "loss": 5.467,
-      "step": 32380
-    },
-    {
-      "epoch": 16.8870925684485,
-      "grad_norm": 1.563793659210205,
-      "learning_rate": 6.796984924623116e-05,
-      "loss": 5.5479,
-      "step": 32381
-    },
-    {
-      "epoch": 16.88761408083442,
-      "grad_norm": 1.5477232933044434,
-      "learning_rate": 6.796884422110554e-05,
-      "loss": 5.3936,
-      "step": 32382
-    },
-    {
-      "epoch": 16.88813559322034,
-      "grad_norm": 1.4253052473068237,
-      "learning_rate": 6.79678391959799e-05,
-      "loss": 5.4299,
-      "step": 32383
-    },
-    {
-      "epoch": 16.888657105606256,
-      "grad_norm": 1.4054404497146606,
-      "learning_rate": 6.796683417085428e-05,
-      "loss": 5.505,
-      "step": 32384
-    },
-    {
-      "epoch": 16.889178617992176,
-      "grad_norm": 1.4547144174575806,
-      "learning_rate": 6.796582914572864e-05,
-      "loss": 5.3758,
-      "step": 32385
-    },
-    {
-      "epoch": 16.889700130378095,
-      "grad_norm": 1.3562346696853638,
-      "learning_rate": 6.796482412060302e-05,
-      "loss": 5.4331,
-      "step": 32386
-    },
-    {
-      "epoch": 16.890221642764015,
-      "grad_norm": 1.5622116327285767,
-      "learning_rate": 6.796381909547738e-05,
-      "loss": 5.2867,
-      "step": 32387
-    },
-    {
-      "epoch": 16.890743155149934,
-      "grad_norm": 1.6270904541015625,
-      "learning_rate": 6.796281407035176e-05,
-      "loss": 5.2917,
-      "step": 32388
-    },
-    {
-      "epoch": 16.891264667535854,
-      "grad_norm": 1.52131986618042,
-      "learning_rate": 6.796180904522613e-05,
-      "loss": 5.3574,
-      "step": 32389
-    },
-    {
-      "epoch": 16.891786179921773,
-      "grad_norm": 1.5438766479492188,
-      "learning_rate": 6.79608040201005e-05,
-      "loss": 4.9773,
-      "step": 32390
-    },
-    {
-      "epoch": 16.892307692307693,
-      "grad_norm": 1.4843555688858032,
-      "learning_rate": 6.795979899497488e-05,
-      "loss": 5.2643,
-      "step": 32391
-    },
-    {
-      "epoch": 16.892829204693612,
-      "grad_norm": 1.5478272438049316,
-      "learning_rate": 6.795879396984926e-05,
-      "loss": 4.9886,
-      "step": 32392
-    },
-    {
-      "epoch": 16.89335071707953,
-      "grad_norm": 1.5448975563049316,
-      "learning_rate": 6.795778894472362e-05,
-      "loss": 5.5898,
-      "step": 32393
-    },
-    {
-      "epoch": 16.89387222946545,
-      "grad_norm": 1.4098066091537476,
-      "learning_rate": 6.795678391959799e-05,
-      "loss": 5.4836,
-      "step": 32394
-    },
-    {
-      "epoch": 16.89439374185137,
-      "grad_norm": 1.670881986618042,
-      "learning_rate": 6.795577889447237e-05,
-      "loss": 5.0142,
-      "step": 32395
-    },
-    {
-      "epoch": 16.894915254237286,
-      "grad_norm": 1.634682536125183,
-      "learning_rate": 6.795477386934673e-05,
-      "loss": 4.7982,
-      "step": 32396
-    },
-    {
-      "epoch": 16.895436766623206,
-      "grad_norm": 1.5118247270584106,
-      "learning_rate": 6.795376884422111e-05,
-      "loss": 5.4586,
-      "step": 32397
-    },
-    {
-      "epoch": 16.895958279009125,
-      "grad_norm": 1.450194001197815,
-      "learning_rate": 6.795276381909547e-05,
-      "loss": 5.1116,
-      "step": 32398
-    },
-    {
-      "epoch": 16.896479791395045,
-      "grad_norm": 1.4802435636520386,
-      "learning_rate": 6.795175879396985e-05,
-      "loss": 5.1491,
-      "step": 32399
-    },
-    {
-      "epoch": 16.897001303780964,
-      "grad_norm": 1.4285637140274048,
-      "learning_rate": 6.795075376884423e-05,
-      "loss": 5.1806,
-      "step": 32400
-    },
-    {
-      "epoch": 16.897522816166884,
-      "grad_norm": 1.5475327968597412,
-      "learning_rate": 6.79497487437186e-05,
-      "loss": 5.1658,
-      "step": 32401
-    },
-    {
-      "epoch": 16.898044328552803,
-      "grad_norm": 1.5188158750534058,
-      "learning_rate": 6.794874371859297e-05,
-      "loss": 5.2091,
-      "step": 32402
-    },
-    {
-      "epoch": 16.898565840938723,
-      "grad_norm": 1.4493523836135864,
-      "learning_rate": 6.794773869346735e-05,
-      "loss": 4.792,
-      "step": 32403
-    },
-    {
-      "epoch": 16.899087353324642,
-      "grad_norm": 1.564681053161621,
-      "learning_rate": 6.794673366834171e-05,
-      "loss": 4.2918,
-      "step": 32404
-    },
-    {
-      "epoch": 16.89960886571056,
-      "grad_norm": 1.5450899600982666,
-      "learning_rate": 6.794572864321609e-05,
-      "loss": 4.6961,
-      "step": 32405
-    },
-    {
-      "epoch": 16.90013037809648,
-      "grad_norm": 1.53985595703125,
-      "learning_rate": 6.794472361809045e-05,
-      "loss": 4.5179,
-      "step": 32406
-    },
-    {
-      "epoch": 16.9006518904824,
-      "grad_norm": 1.6042312383651733,
-      "learning_rate": 6.794371859296482e-05,
-      "loss": 5.0994,
-      "step": 32407
-    },
-    {
-      "epoch": 16.901173402868316,
-      "grad_norm": 1.499404788017273,
-      "learning_rate": 6.79427135678392e-05,
-      "loss": 4.9225,
-      "step": 32408
-    },
-    {
-      "epoch": 16.901694915254236,
-      "grad_norm": 1.4318393468856812,
-      "learning_rate": 6.794170854271356e-05,
-      "loss": 4.8485,
-      "step": 32409
-    },
-    {
-      "epoch": 16.902216427640155,
-      "grad_norm": 1.575245976448059,
-      "learning_rate": 6.794070351758794e-05,
-      "loss": 4.7498,
-      "step": 32410
-    },
-    {
-      "epoch": 16.902737940026075,
-      "grad_norm": 1.523076057434082,
-      "learning_rate": 6.793969849246232e-05,
-      "loss": 4.8459,
-      "step": 32411
-    },
-    {
-      "epoch": 16.903259452411994,
-      "grad_norm": 1.3801064491271973,
-      "learning_rate": 6.79386934673367e-05,
-      "loss": 5.6031,
-      "step": 32412
-    },
-    {
-      "epoch": 16.903780964797914,
-      "grad_norm": 1.5999196767807007,
-      "learning_rate": 6.793768844221106e-05,
-      "loss": 5.4116,
-      "step": 32413
-    },
-    {
-      "epoch": 16.904302477183833,
-      "grad_norm": 1.5234853029251099,
-      "learning_rate": 6.793668341708544e-05,
-      "loss": 5.3604,
-      "step": 32414
-    },
-    {
-      "epoch": 16.904823989569753,
-      "grad_norm": 1.4929800033569336,
-      "learning_rate": 6.79356783919598e-05,
-      "loss": 4.963,
-      "step": 32415
-    },
-    {
-      "epoch": 16.905345501955672,
-      "grad_norm": 1.4898545742034912,
-      "learning_rate": 6.793467336683418e-05,
-      "loss": 5.3165,
-      "step": 32416
-    },
-    {
-      "epoch": 16.90586701434159,
-      "grad_norm": 1.5031137466430664,
-      "learning_rate": 6.793366834170854e-05,
-      "loss": 5.2267,
-      "step": 32417
-    },
-    {
-      "epoch": 16.90638852672751,
-      "grad_norm": 1.5518840551376343,
-      "learning_rate": 6.793266331658292e-05,
-      "loss": 5.5924,
-      "step": 32418
-    },
-    {
-      "epoch": 16.90691003911343,
-      "grad_norm": 1.4476866722106934,
-      "learning_rate": 6.793165829145729e-05,
-      "loss": 5.6544,
-      "step": 32419
-    },
-    {
-      "epoch": 16.907431551499347,
-      "grad_norm": 1.3894962072372437,
-      "learning_rate": 6.793065326633166e-05,
-      "loss": 5.3898,
-      "step": 32420
-    },
-    {
-      "epoch": 16.907953063885266,
-      "grad_norm": 1.4671924114227295,
-      "learning_rate": 6.792964824120604e-05,
-      "loss": 5.3894,
-      "step": 32421
-    },
-    {
-      "epoch": 16.908474576271185,
-      "grad_norm": 1.380654215812683,
-      "learning_rate": 6.79286432160804e-05,
-      "loss": 5.1498,
-      "step": 32422
-    },
-    {
-      "epoch": 16.908996088657105,
-      "grad_norm": 1.6227922439575195,
-      "learning_rate": 6.792763819095478e-05,
-      "loss": 4.993,
-      "step": 32423
-    },
-    {
-      "epoch": 16.909517601043024,
-      "grad_norm": 1.4840996265411377,
-      "learning_rate": 6.792663316582915e-05,
-      "loss": 4.6775,
-      "step": 32424
-    },
-    {
-      "epoch": 16.910039113428944,
-      "grad_norm": 1.6057088375091553,
-      "learning_rate": 6.792562814070353e-05,
-      "loss": 5.5235,
-      "step": 32425
-    },
-    {
-      "epoch": 16.910560625814863,
-      "grad_norm": 1.524800181388855,
-      "learning_rate": 6.792462311557789e-05,
-      "loss": 5.0843,
-      "step": 32426
-    },
-    {
-      "epoch": 16.911082138200783,
-      "grad_norm": 1.4572134017944336,
-      "learning_rate": 6.792361809045227e-05,
-      "loss": 5.5798,
-      "step": 32427
-    },
-    {
-      "epoch": 16.911603650586702,
-      "grad_norm": 1.4275416135787964,
-      "learning_rate": 6.792261306532663e-05,
-      "loss": 5.6972,
-      "step": 32428
-    },
-    {
-      "epoch": 16.91212516297262,
-      "grad_norm": 1.4458866119384766,
-      "learning_rate": 6.792160804020101e-05,
-      "loss": 5.2846,
-      "step": 32429
-    },
-    {
-      "epoch": 16.91264667535854,
-      "grad_norm": 1.3837119340896606,
-      "learning_rate": 6.792060301507537e-05,
-      "loss": 5.5417,
-      "step": 32430
-    },
-    {
-      "epoch": 16.91316818774446,
-      "grad_norm": 1.4822882413864136,
-      "learning_rate": 6.791959798994975e-05,
-      "loss": 5.3521,
-      "step": 32431
-    },
-    {
-      "epoch": 16.913689700130377,
-      "grad_norm": 1.5889843702316284,
-      "learning_rate": 6.791859296482413e-05,
-      "loss": 4.9364,
-      "step": 32432
-    },
-    {
-      "epoch": 16.914211212516296,
-      "grad_norm": 1.3969556093215942,
-      "learning_rate": 6.79175879396985e-05,
-      "loss": 5.4135,
-      "step": 32433
-    },
-    {
-      "epoch": 16.914732724902215,
-      "grad_norm": 1.4677510261535645,
-      "learning_rate": 6.791658291457287e-05,
-      "loss": 4.9149,
-      "step": 32434
-    },
-    {
-      "epoch": 16.915254237288135,
-      "grad_norm": 1.4858578443527222,
-      "learning_rate": 6.791557788944724e-05,
-      "loss": 5.4426,
-      "step": 32435
-    },
-    {
-      "epoch": 16.915775749674054,
-      "grad_norm": 1.5129386186599731,
-      "learning_rate": 6.791457286432161e-05,
-      "loss": 5.1704,
-      "step": 32436
-    },
-    {
-      "epoch": 16.916297262059974,
-      "grad_norm": 1.4411780834197998,
-      "learning_rate": 6.791356783919598e-05,
-      "loss": 5.5826,
-      "step": 32437
-    },
-    {
-      "epoch": 16.916818774445893,
-      "grad_norm": 1.6230751276016235,
-      "learning_rate": 6.791256281407036e-05,
-      "loss": 5.1289,
-      "step": 32438
-    },
-    {
-      "epoch": 16.917340286831813,
-      "grad_norm": 1.428541898727417,
-      "learning_rate": 6.791155778894472e-05,
-      "loss": 5.628,
-      "step": 32439
-    },
-    {
-      "epoch": 16.917861799217732,
-      "grad_norm": 1.3998253345489502,
-      "learning_rate": 6.79105527638191e-05,
-      "loss": 5.7819,
-      "step": 32440
-    },
-    {
-      "epoch": 16.91838331160365,
-      "grad_norm": 1.402811050415039,
-      "learning_rate": 6.790954773869346e-05,
-      "loss": 5.2663,
-      "step": 32441
-    },
-    {
-      "epoch": 16.91890482398957,
-      "grad_norm": 1.4845298528671265,
-      "learning_rate": 6.790854271356784e-05,
-      "loss": 4.8641,
-      "step": 32442
-    },
-    {
-      "epoch": 16.919426336375487,
-      "grad_norm": 1.4814549684524536,
-      "learning_rate": 6.790753768844222e-05,
-      "loss": 5.4626,
-      "step": 32443
-    },
-    {
-      "epoch": 16.919947848761407,
-      "grad_norm": 1.3921464681625366,
-      "learning_rate": 6.79065326633166e-05,
-      "loss": 5.7124,
-      "step": 32444
-    },
-    {
-      "epoch": 16.920469361147326,
-      "grad_norm": 1.3104885816574097,
-      "learning_rate": 6.790552763819096e-05,
-      "loss": 5.759,
-      "step": 32445
-    },
-    {
-      "epoch": 16.920990873533245,
-      "grad_norm": 1.507103681564331,
-      "learning_rate": 6.790452261306532e-05,
-      "loss": 5.0896,
-      "step": 32446
-    },
-    {
-      "epoch": 16.921512385919165,
-      "grad_norm": 1.6472971439361572,
-      "learning_rate": 6.79035175879397e-05,
-      "loss": 5.159,
-      "step": 32447
-    },
-    {
-      "epoch": 16.922033898305084,
-      "grad_norm": 1.6453166007995605,
-      "learning_rate": 6.790251256281407e-05,
-      "loss": 5.6292,
-      "step": 32448
-    },
-    {
-      "epoch": 16.922555410691004,
-      "grad_norm": 1.6022409200668335,
-      "learning_rate": 6.790150753768844e-05,
-      "loss": 5.1275,
-      "step": 32449
-    },
-    {
-      "epoch": 16.923076923076923,
-      "grad_norm": 1.5402127504348755,
-      "learning_rate": 6.790050251256281e-05,
-      "loss": 5.3025,
-      "step": 32450
-    },
-    {
-      "epoch": 16.923598435462843,
-      "grad_norm": 1.6186796426773071,
-      "learning_rate": 6.789949748743719e-05,
-      "loss": 5.0808,
-      "step": 32451
-    },
-    {
-      "epoch": 16.924119947848762,
-      "grad_norm": 1.5059268474578857,
-      "learning_rate": 6.789849246231156e-05,
-      "loss": 5.5736,
-      "step": 32452
-    },
-    {
-      "epoch": 16.92464146023468,
-      "grad_norm": 1.47416090965271,
-      "learning_rate": 6.789748743718594e-05,
-      "loss": 5.4597,
-      "step": 32453
-    },
-    {
-      "epoch": 16.9251629726206,
-      "grad_norm": 1.5530303716659546,
-      "learning_rate": 6.78964824120603e-05,
-      "loss": 5.3305,
-      "step": 32454
-    },
-    {
-      "epoch": 16.92568448500652,
-      "grad_norm": 1.4411883354187012,
-      "learning_rate": 6.789547738693468e-05,
-      "loss": 5.6362,
-      "step": 32455
-    },
-    {
-      "epoch": 16.926205997392437,
-      "grad_norm": 1.4205687046051025,
-      "learning_rate": 6.789447236180905e-05,
-      "loss": 5.583,
-      "step": 32456
-    },
-    {
-      "epoch": 16.926727509778356,
-      "grad_norm": 1.5574069023132324,
-      "learning_rate": 6.789346733668343e-05,
-      "loss": 5.441,
-      "step": 32457
-    },
-    {
-      "epoch": 16.927249022164276,
-      "grad_norm": 1.4625929594039917,
-      "learning_rate": 6.789246231155779e-05,
-      "loss": 5.5049,
-      "step": 32458
-    },
-    {
-      "epoch": 16.927770534550195,
-      "grad_norm": 1.3688398599624634,
-      "learning_rate": 6.789145728643215e-05,
-      "loss": 5.557,
-      "step": 32459
-    },
-    {
-      "epoch": 16.928292046936114,
-      "grad_norm": 1.4979876279830933,
-      "learning_rate": 6.789045226130653e-05,
-      "loss": 5.0321,
-      "step": 32460
-    },
-    {
-      "epoch": 16.928813559322034,
-      "grad_norm": 1.5938234329223633,
-      "learning_rate": 6.78894472361809e-05,
-      "loss": 4.8214,
-      "step": 32461
-    },
-    {
-      "epoch": 16.929335071707953,
-      "grad_norm": 1.4277130365371704,
-      "learning_rate": 6.788844221105527e-05,
-      "loss": 5.6107,
-      "step": 32462
-    },
-    {
-      "epoch": 16.929856584093873,
-      "grad_norm": 1.4846256971359253,
-      "learning_rate": 6.788743718592965e-05,
-      "loss": 5.551,
-      "step": 32463
-    },
-    {
-      "epoch": 16.930378096479792,
-      "grad_norm": 1.4373703002929688,
-      "learning_rate": 6.788643216080403e-05,
-      "loss": 5.6521,
-      "step": 32464
-    },
-    {
-      "epoch": 16.93089960886571,
-      "grad_norm": 1.4478901624679565,
-      "learning_rate": 6.78854271356784e-05,
-      "loss": 5.5588,
-      "step": 32465
-    },
-    {
-      "epoch": 16.93142112125163,
-      "grad_norm": 1.5088919401168823,
-      "learning_rate": 6.788442211055277e-05,
-      "loss": 5.2031,
-      "step": 32466
-    },
-    {
-      "epoch": 16.931942633637547,
-      "grad_norm": 1.6543411016464233,
-      "learning_rate": 6.788341708542714e-05,
-      "loss": 4.6802,
-      "step": 32467
-    },
-    {
-      "epoch": 16.932464146023467,
-      "grad_norm": 1.4922834634780884,
-      "learning_rate": 6.788241206030151e-05,
-      "loss": 5.3909,
-      "step": 32468
-    },
-    {
-      "epoch": 16.932985658409386,
-      "grad_norm": 1.4276663064956665,
-      "learning_rate": 6.788140703517588e-05,
-      "loss": 5.5653,
-      "step": 32469
-    },
-    {
-      "epoch": 16.933507170795306,
-      "grad_norm": 1.6414177417755127,
-      "learning_rate": 6.788040201005026e-05,
-      "loss": 5.0289,
-      "step": 32470
-    },
-    {
-      "epoch": 16.934028683181225,
-      "grad_norm": 1.3962595462799072,
-      "learning_rate": 6.787939698492462e-05,
-      "loss": 5.4565,
-      "step": 32471
-    },
-    {
-      "epoch": 16.934550195567144,
-      "grad_norm": 1.5477362871170044,
-      "learning_rate": 6.7878391959799e-05,
-      "loss": 5.0547,
-      "step": 32472
-    },
-    {
-      "epoch": 16.935071707953064,
-      "grad_norm": 1.51752769947052,
-      "learning_rate": 6.787738693467338e-05,
-      "loss": 5.2489,
-      "step": 32473
-    },
-    {
-      "epoch": 16.935593220338983,
-      "grad_norm": 1.4608659744262695,
-      "learning_rate": 6.787638190954774e-05,
-      "loss": 5.6711,
-      "step": 32474
-    },
-    {
-      "epoch": 16.936114732724903,
-      "grad_norm": 1.3969889879226685,
-      "learning_rate": 6.787537688442212e-05,
-      "loss": 5.6891,
-      "step": 32475
-    },
-    {
-      "epoch": 16.936636245110822,
-      "grad_norm": 1.4909021854400635,
-      "learning_rate": 6.787437185929648e-05,
-      "loss": 5.5743,
-      "step": 32476
-    },
-    {
-      "epoch": 16.937157757496742,
-      "grad_norm": 1.5796184539794922,
-      "learning_rate": 6.787336683417086e-05,
-      "loss": 5.3939,
-      "step": 32477
-    },
-    {
-      "epoch": 16.93767926988266,
-      "grad_norm": 1.506629228591919,
-      "learning_rate": 6.787236180904522e-05,
-      "loss": 5.3945,
-      "step": 32478
-    },
-    {
-      "epoch": 16.938200782268577,
-      "grad_norm": 1.454669713973999,
-      "learning_rate": 6.78713567839196e-05,
-      "loss": 5.5389,
-      "step": 32479
-    },
-    {
-      "epoch": 16.938722294654497,
-      "grad_norm": 1.384960651397705,
-      "learning_rate": 6.787035175879397e-05,
-      "loss": 5.3641,
-      "step": 32480
-    },
-    {
-      "epoch": 16.939243807040416,
-      "grad_norm": 1.3818535804748535,
-      "learning_rate": 6.786934673366834e-05,
-      "loss": 5.3624,
-      "step": 32481
-    },
-    {
-      "epoch": 16.939765319426336,
-      "grad_norm": 1.7369868755340576,
-      "learning_rate": 6.786834170854271e-05,
-      "loss": 5.0155,
-      "step": 32482
-    },
-    {
-      "epoch": 16.940286831812255,
-      "grad_norm": 1.3935303688049316,
-      "learning_rate": 6.786733668341709e-05,
-      "loss": 5.8009,
-      "step": 32483
-    },
-    {
-      "epoch": 16.940808344198174,
-      "grad_norm": 1.5462543964385986,
-      "learning_rate": 6.786633165829146e-05,
-      "loss": 5.194,
-      "step": 32484
-    },
-    {
-      "epoch": 16.941329856584094,
-      "grad_norm": 1.3278051614761353,
-      "learning_rate": 6.786532663316584e-05,
-      "loss": 5.8406,
-      "step": 32485
-    },
-    {
-      "epoch": 16.941851368970013,
-      "grad_norm": 1.5530050992965698,
-      "learning_rate": 6.78643216080402e-05,
-      "loss": 5.4171,
-      "step": 32486
-    },
-    {
-      "epoch": 16.942372881355933,
-      "grad_norm": 1.5688409805297852,
-      "learning_rate": 6.786331658291457e-05,
-      "loss": 5.4099,
-      "step": 32487
-    },
-    {
-      "epoch": 16.942894393741852,
-      "grad_norm": 1.7084956169128418,
-      "learning_rate": 6.786231155778895e-05,
-      "loss": 4.8713,
-      "step": 32488
-    },
-    {
-      "epoch": 16.943415906127772,
-      "grad_norm": 1.4010425806045532,
-      "learning_rate": 6.786130653266331e-05,
-      "loss": 5.5267,
-      "step": 32489
-    },
-    {
-      "epoch": 16.94393741851369,
-      "grad_norm": 1.5105950832366943,
-      "learning_rate": 6.786030150753769e-05,
-      "loss": 5.6022,
-      "step": 32490
-    },
-    {
-      "epoch": 16.944458930899607,
-      "grad_norm": 1.536176085472107,
-      "learning_rate": 6.785929648241206e-05,
-      "loss": 5.2281,
-      "step": 32491
-    },
-    {
-      "epoch": 16.944980443285527,
-      "grad_norm": 1.4717206954956055,
-      "learning_rate": 6.785829145728643e-05,
-      "loss": 5.3244,
-      "step": 32492
-    },
-    {
-      "epoch": 16.945501955671446,
-      "grad_norm": 1.4649834632873535,
-      "learning_rate": 6.785728643216081e-05,
-      "loss": 4.979,
-      "step": 32493
-    },
-    {
-      "epoch": 16.946023468057366,
-      "grad_norm": 1.569586992263794,
-      "learning_rate": 6.785628140703519e-05,
-      "loss": 5.109,
-      "step": 32494
-    },
-    {
-      "epoch": 16.946544980443285,
-      "grad_norm": 1.5353845357894897,
-      "learning_rate": 6.785527638190955e-05,
-      "loss": 5.5801,
-      "step": 32495
-    },
-    {
-      "epoch": 16.947066492829205,
-      "grad_norm": 1.583788275718689,
-      "learning_rate": 6.785427135678393e-05,
-      "loss": 5.4363,
-      "step": 32496
-    },
-    {
-      "epoch": 16.947588005215124,
-      "grad_norm": 1.5092730522155762,
-      "learning_rate": 6.78532663316583e-05,
-      "loss": 5.5068,
-      "step": 32497
-    },
-    {
-      "epoch": 16.948109517601043,
-      "grad_norm": 1.515787959098816,
-      "learning_rate": 6.785226130653267e-05,
-      "loss": 5.4917,
-      "step": 32498
-    },
-    {
-      "epoch": 16.948631029986963,
-      "grad_norm": 1.4549691677093506,
-      "learning_rate": 6.785125628140704e-05,
-      "loss": 5.3303,
-      "step": 32499
-    },
-    {
-      "epoch": 16.949152542372882,
-      "grad_norm": 1.4001710414886475,
-      "learning_rate": 6.78502512562814e-05,
-      "loss": 5.65,
-      "step": 32500
-    },
-    {
-      "epoch": 16.949674054758802,
-      "grad_norm": 1.6653918027877808,
-      "learning_rate": 6.784924623115578e-05,
-      "loss": 5.0974,
-      "step": 32501
-    },
-    {
-      "epoch": 16.95019556714472,
-      "grad_norm": 1.4158194065093994,
-      "learning_rate": 6.784824120603014e-05,
-      "loss": 5.6024,
-      "step": 32502
-    },
-    {
-      "epoch": 16.950717079530637,
-      "grad_norm": 1.4104406833648682,
-      "learning_rate": 6.784723618090452e-05,
-      "loss": 5.5611,
-      "step": 32503
-    },
-    {
-      "epoch": 16.951238591916557,
-      "grad_norm": 1.5437138080596924,
-      "learning_rate": 6.78462311557789e-05,
-      "loss": 5.1861,
-      "step": 32504
-    },
-    {
-      "epoch": 16.951760104302476,
-      "grad_norm": 1.5005415678024292,
-      "learning_rate": 6.784522613065328e-05,
-      "loss": 4.9843,
-      "step": 32505
-    },
-    {
-      "epoch": 16.952281616688396,
-      "grad_norm": 1.4291985034942627,
-      "learning_rate": 6.784422110552764e-05,
-      "loss": 5.4527,
-      "step": 32506
-    },
-    {
-      "epoch": 16.952803129074315,
-      "grad_norm": 1.512047529220581,
-      "learning_rate": 6.784321608040202e-05,
-      "loss": 5.0667,
-      "step": 32507
-    },
-    {
-      "epoch": 16.953324641460235,
-      "grad_norm": 1.4717110395431519,
-      "learning_rate": 6.784221105527638e-05,
-      "loss": 5.2194,
-      "step": 32508
-    },
-    {
-      "epoch": 16.953846153846154,
-      "grad_norm": 1.4788250923156738,
-      "learning_rate": 6.784120603015076e-05,
-      "loss": 4.8092,
-      "step": 32509
-    },
-    {
-      "epoch": 16.954367666232073,
-      "grad_norm": 1.4280763864517212,
-      "learning_rate": 6.784020100502513e-05,
-      "loss": 5.744,
-      "step": 32510
-    },
-    {
-      "epoch": 16.954889178617993,
-      "grad_norm": 1.4562805891036987,
-      "learning_rate": 6.78391959798995e-05,
-      "loss": 5.3434,
-      "step": 32511
-    },
-    {
-      "epoch": 16.955410691003912,
-      "grad_norm": 1.5127441883087158,
-      "learning_rate": 6.783819095477387e-05,
-      "loss": 5.4389,
-      "step": 32512
-    },
-    {
-      "epoch": 16.955932203389832,
-      "grad_norm": 1.4529786109924316,
-      "learning_rate": 6.783718592964825e-05,
-      "loss": 5.8001,
-      "step": 32513
-    },
-    {
-      "epoch": 16.95645371577575,
-      "grad_norm": 1.5164039134979248,
-      "learning_rate": 6.783618090452262e-05,
-      "loss": 5.0905,
-      "step": 32514
-    },
-    {
-      "epoch": 16.956975228161667,
-      "grad_norm": 1.543525218963623,
-      "learning_rate": 6.783517587939699e-05,
-      "loss": 5.0411,
-      "step": 32515
-    },
-    {
-      "epoch": 16.957496740547587,
-      "grad_norm": 1.50077486038208,
-      "learning_rate": 6.783417085427137e-05,
-      "loss": 5.1831,
-      "step": 32516
-    },
-    {
-      "epoch": 16.958018252933506,
-      "grad_norm": 1.497809886932373,
-      "learning_rate": 6.783316582914573e-05,
-      "loss": 5.6489,
-      "step": 32517
-    },
-    {
-      "epoch": 16.958539765319426,
-      "grad_norm": 1.4587361812591553,
-      "learning_rate": 6.783216080402011e-05,
-      "loss": 5.2305,
-      "step": 32518
-    },
-    {
-      "epoch": 16.959061277705345,
-      "grad_norm": 1.5162676572799683,
-      "learning_rate": 6.783115577889447e-05,
-      "loss": 5.5771,
-      "step": 32519
-    },
-    {
-      "epoch": 16.959582790091265,
-      "grad_norm": 1.3966959714889526,
-      "learning_rate": 6.783015075376885e-05,
-      "loss": 5.6773,
-      "step": 32520
-    },
-    {
-      "epoch": 16.960104302477184,
-      "grad_norm": 1.513175368309021,
-      "learning_rate": 6.782914572864321e-05,
-      "loss": 5.1244,
-      "step": 32521
-    },
-    {
-      "epoch": 16.960625814863103,
-      "grad_norm": 1.63186514377594,
-      "learning_rate": 6.782814070351759e-05,
-      "loss": 4.8423,
-      "step": 32522
-    },
-    {
-      "epoch": 16.961147327249023,
-      "grad_norm": 1.5202438831329346,
-      "learning_rate": 6.782713567839196e-05,
-      "loss": 4.8323,
-      "step": 32523
-    },
-    {
-      "epoch": 16.961668839634942,
-      "grad_norm": 1.5268582105636597,
-      "learning_rate": 6.782613065326633e-05,
-      "loss": 5.4536,
-      "step": 32524
-    },
-    {
-      "epoch": 16.962190352020862,
-      "grad_norm": 1.4350453615188599,
-      "learning_rate": 6.782512562814071e-05,
-      "loss": 5.5015,
-      "step": 32525
-    },
-    {
-      "epoch": 16.96271186440678,
-      "grad_norm": 1.4945721626281738,
-      "learning_rate": 6.782412060301508e-05,
-      "loss": 5.6765,
-      "step": 32526
-    },
-    {
-      "epoch": 16.963233376792697,
-      "grad_norm": 1.450631856918335,
-      "learning_rate": 6.782311557788945e-05,
-      "loss": 5.4143,
-      "step": 32527
-    },
-    {
-      "epoch": 16.963754889178617,
-      "grad_norm": 1.5579357147216797,
-      "learning_rate": 6.782211055276382e-05,
-      "loss": 5.1957,
-      "step": 32528
-    },
-    {
-      "epoch": 16.964276401564536,
-      "grad_norm": 1.4341241121292114,
-      "learning_rate": 6.78211055276382e-05,
-      "loss": 5.6335,
-      "step": 32529
-    },
-    {
-      "epoch": 16.964797913950456,
-      "grad_norm": 1.431963562965393,
-      "learning_rate": 6.782010050251256e-05,
-      "loss": 5.4324,
-      "step": 32530
-    },
-    {
-      "epoch": 16.965319426336375,
-      "grad_norm": 1.5753487348556519,
-      "learning_rate": 6.781909547738694e-05,
-      "loss": 5.2753,
-      "step": 32531
-    },
-    {
-      "epoch": 16.965840938722295,
-      "grad_norm": 1.5286152362823486,
-      "learning_rate": 6.78180904522613e-05,
-      "loss": 5.2421,
-      "step": 32532
-    },
-    {
-      "epoch": 16.966362451108214,
-      "grad_norm": 1.4572099447250366,
-      "learning_rate": 6.781708542713568e-05,
-      "loss": 5.3035,
-      "step": 32533
-    },
-    {
-      "epoch": 16.966883963494134,
-      "grad_norm": 1.5085810422897339,
-      "learning_rate": 6.781608040201006e-05,
-      "loss": 5.0268,
-      "step": 32534
-    },
-    {
-      "epoch": 16.967405475880053,
-      "grad_norm": 1.4976749420166016,
-      "learning_rate": 6.781507537688444e-05,
-      "loss": 5.0977,
-      "step": 32535
-    },
-    {
-      "epoch": 16.967926988265972,
-      "grad_norm": 1.4044971466064453,
-      "learning_rate": 6.78140703517588e-05,
-      "loss": 5.8287,
-      "step": 32536
-    },
-    {
-      "epoch": 16.968448500651892,
-      "grad_norm": 1.4583863019943237,
-      "learning_rate": 6.781306532663318e-05,
-      "loss": 4.9915,
-      "step": 32537
-    },
-    {
-      "epoch": 16.96897001303781,
-      "grad_norm": 1.5127754211425781,
-      "learning_rate": 6.781206030150754e-05,
-      "loss": 5.4736,
-      "step": 32538
-    },
-    {
-      "epoch": 16.969491525423727,
-      "grad_norm": 1.4333122968673706,
-      "learning_rate": 6.78110552763819e-05,
-      "loss": 4.8429,
-      "step": 32539
-    },
-    {
-      "epoch": 16.970013037809647,
-      "grad_norm": 1.4185161590576172,
-      "learning_rate": 6.781005025125628e-05,
-      "loss": 5.6353,
-      "step": 32540
-    },
-    {
-      "epoch": 16.970534550195566,
-      "grad_norm": 1.563348650932312,
-      "learning_rate": 6.780904522613065e-05,
-      "loss": 5.3613,
-      "step": 32541
-    },
-    {
-      "epoch": 16.971056062581486,
-      "grad_norm": 1.541674017906189,
-      "learning_rate": 6.780804020100503e-05,
-      "loss": 4.9248,
-      "step": 32542
-    },
-    {
-      "epoch": 16.971577574967405,
-      "grad_norm": 1.4914451837539673,
-      "learning_rate": 6.780703517587939e-05,
-      "loss": 5.4834,
-      "step": 32543
-    },
-    {
-      "epoch": 16.972099087353325,
-      "grad_norm": 1.5243611335754395,
-      "learning_rate": 6.780603015075377e-05,
-      "loss": 5.1817,
-      "step": 32544
-    },
-    {
-      "epoch": 16.972620599739244,
-      "grad_norm": 1.3938015699386597,
-      "learning_rate": 6.780502512562815e-05,
-      "loss": 5.5486,
-      "step": 32545
-    },
-    {
-      "epoch": 16.973142112125164,
-      "grad_norm": 1.7328482866287231,
-      "learning_rate": 6.780402010050252e-05,
-      "loss": 4.7936,
-      "step": 32546
-    },
-    {
-      "epoch": 16.973663624511083,
-      "grad_norm": 1.5499812364578247,
-      "learning_rate": 6.780301507537689e-05,
-      "loss": 5.2962,
-      "step": 32547
-    },
-    {
-      "epoch": 16.974185136897002,
-      "grad_norm": 1.503544569015503,
-      "learning_rate": 6.780201005025127e-05,
-      "loss": 4.9077,
-      "step": 32548
-    },
-    {
-      "epoch": 16.974706649282922,
-      "grad_norm": 1.590570330619812,
-      "learning_rate": 6.780100502512563e-05,
-      "loss": 4.3892,
-      "step": 32549
-    },
-    {
-      "epoch": 16.975228161668838,
-      "grad_norm": 1.6227085590362549,
-      "learning_rate": 6.780000000000001e-05,
-      "loss": 4.8424,
-      "step": 32550
-    },
-    {
-      "epoch": 16.975749674054757,
-      "grad_norm": 1.6105303764343262,
-      "learning_rate": 6.779899497487437e-05,
-      "loss": 5.0224,
-      "step": 32551
-    },
-    {
-      "epoch": 16.976271186440677,
-      "grad_norm": 1.442943811416626,
-      "learning_rate": 6.779798994974875e-05,
-      "loss": 5.1003,
-      "step": 32552
-    },
-    {
-      "epoch": 16.976792698826596,
-      "grad_norm": 1.5783569812774658,
-      "learning_rate": 6.779698492462311e-05,
-      "loss": 5.1149,
-      "step": 32553
-    },
-    {
-      "epoch": 16.977314211212516,
-      "grad_norm": 1.4210820198059082,
-      "learning_rate": 6.779597989949749e-05,
-      "loss": 5.5537,
-      "step": 32554
-    },
-    {
-      "epoch": 16.977835723598435,
-      "grad_norm": 1.5846909284591675,
-      "learning_rate": 6.779497487437187e-05,
-      "loss": 5.2629,
-      "step": 32555
-    },
-    {
-      "epoch": 16.978357235984355,
-      "grad_norm": 1.5144847631454468,
-      "learning_rate": 6.779396984924623e-05,
-      "loss": 4.8956,
-      "step": 32556
-    },
-    {
-      "epoch": 16.978878748370274,
-      "grad_norm": 1.4400230646133423,
-      "learning_rate": 6.779296482412061e-05,
-      "loss": 5.7281,
-      "step": 32557
-    },
-    {
-      "epoch": 16.979400260756194,
-      "grad_norm": 1.504347324371338,
-      "learning_rate": 6.779195979899498e-05,
-      "loss": 4.6207,
-      "step": 32558
-    },
-    {
-      "epoch": 16.979921773142113,
-      "grad_norm": 1.5274969339370728,
-      "learning_rate": 6.779095477386935e-05,
-      "loss": 5.1705,
-      "step": 32559
-    },
-    {
-      "epoch": 16.980443285528033,
-      "grad_norm": 1.4403496980667114,
-      "learning_rate": 6.778994974874372e-05,
-      "loss": 5.3427,
-      "step": 32560
-    },
-    {
-      "epoch": 16.980964797913952,
-      "grad_norm": 1.493645429611206,
-      "learning_rate": 6.77889447236181e-05,
-      "loss": 5.0072,
-      "step": 32561
-    },
-    {
-      "epoch": 16.98148631029987,
-      "grad_norm": 1.5158759355545044,
-      "learning_rate": 6.778793969849246e-05,
-      "loss": 5.1369,
-      "step": 32562
-    },
-    {
-      "epoch": 16.982007822685787,
-      "grad_norm": 1.5361461639404297,
-      "learning_rate": 6.778693467336684e-05,
-      "loss": 4.9126,
-      "step": 32563
-    },
-    {
-      "epoch": 16.982529335071707,
-      "grad_norm": 1.6894100904464722,
-      "learning_rate": 6.77859296482412e-05,
-      "loss": 4.8708,
-      "step": 32564
-    },
-    {
-      "epoch": 16.983050847457626,
-      "grad_norm": 1.4957232475280762,
-      "learning_rate": 6.778492462311558e-05,
-      "loss": 5.3323,
-      "step": 32565
-    },
-    {
-      "epoch": 16.983572359843546,
-      "grad_norm": 1.6225494146347046,
-      "learning_rate": 6.778391959798996e-05,
-      "loss": 5.5759,
-      "step": 32566
-    },
-    {
-      "epoch": 16.984093872229465,
-      "grad_norm": 1.4962069988250732,
-      "learning_rate": 6.778291457286432e-05,
-      "loss": 5.3217,
-      "step": 32567
-    },
-    {
-      "epoch": 16.984615384615385,
-      "grad_norm": 1.4344429969787598,
-      "learning_rate": 6.77819095477387e-05,
-      "loss": 5.487,
-      "step": 32568
-    },
-    {
-      "epoch": 16.985136897001304,
-      "grad_norm": 1.4702370166778564,
-      "learning_rate": 6.778090452261306e-05,
-      "loss": 5.1864,
-      "step": 32569
-    },
-    {
-      "epoch": 16.985658409387224,
-      "grad_norm": 1.547245979309082,
-      "learning_rate": 6.777989949748744e-05,
-      "loss": 5.3752,
-      "step": 32570
-    },
-    {
-      "epoch": 16.986179921773143,
-      "grad_norm": 1.5361382961273193,
-      "learning_rate": 6.777889447236181e-05,
-      "loss": 5.141,
-      "step": 32571
-    },
-    {
-      "epoch": 16.986701434159063,
-      "grad_norm": 1.3966479301452637,
-      "learning_rate": 6.777788944723618e-05,
-      "loss": 5.3503,
-      "step": 32572
-    },
-    {
-      "epoch": 16.987222946544982,
-      "grad_norm": 1.5505366325378418,
-      "learning_rate": 6.777688442211055e-05,
-      "loss": 5.2526,
-      "step": 32573
-    },
-    {
-      "epoch": 16.987744458930898,
-      "grad_norm": 1.4683705568313599,
-      "learning_rate": 6.777587939698493e-05,
-      "loss": 5.2492,
-      "step": 32574
-    },
-    {
-      "epoch": 16.988265971316817,
-      "grad_norm": 1.403518557548523,
-      "learning_rate": 6.77748743718593e-05,
-      "loss": 5.7132,
-      "step": 32575
-    },
-    {
-      "epoch": 16.988787483702737,
-      "grad_norm": 1.46780526638031,
-      "learning_rate": 6.777386934673368e-05,
-      "loss": 5.5735,
-      "step": 32576
-    },
-    {
-      "epoch": 16.989308996088656,
-      "grad_norm": 1.4418790340423584,
-      "learning_rate": 6.777286432160805e-05,
-      "loss": 5.2221,
-      "step": 32577
-    },
-    {
-      "epoch": 16.989830508474576,
-      "grad_norm": 1.4316335916519165,
-      "learning_rate": 6.777185929648242e-05,
-      "loss": 5.6872,
-      "step": 32578
-    },
-    {
-      "epoch": 16.990352020860495,
-      "grad_norm": 1.4341421127319336,
-      "learning_rate": 6.777085427135679e-05,
-      "loss": 5.8249,
-      "step": 32579
-    },
-    {
-      "epoch": 16.990873533246415,
-      "grad_norm": 1.5960557460784912,
-      "learning_rate": 6.776984924623115e-05,
-      "loss": 5.0144,
-      "step": 32580
-    },
-    {
-      "epoch": 16.991395045632334,
-      "grad_norm": 1.5283814668655396,
-      "learning_rate": 6.776884422110553e-05,
-      "loss": 5.6476,
-      "step": 32581
-    },
-    {
-      "epoch": 16.991916558018254,
-      "grad_norm": 1.6673715114593506,
-      "learning_rate": 6.77678391959799e-05,
-      "loss": 4.8409,
-      "step": 32582
-    },
-    {
-      "epoch": 16.992438070404173,
-      "grad_norm": 1.5923595428466797,
-      "learning_rate": 6.776683417085427e-05,
-      "loss": 5.3573,
-      "step": 32583
-    },
-    {
-      "epoch": 16.992959582790093,
-      "grad_norm": 1.4287127256393433,
-      "learning_rate": 6.776582914572864e-05,
-      "loss": 5.4149,
-      "step": 32584
-    },
-    {
-      "epoch": 16.993481095176012,
-      "grad_norm": 1.4457000494003296,
-      "learning_rate": 6.776482412060302e-05,
-      "loss": 5.6279,
-      "step": 32585
-    },
-    {
-      "epoch": 16.994002607561928,
-      "grad_norm": 1.5080779790878296,
-      "learning_rate": 6.776381909547739e-05,
-      "loss": 5.0433,
-      "step": 32586
-    },
-    {
-      "epoch": 16.994524119947847,
-      "grad_norm": 1.4879759550094604,
-      "learning_rate": 6.776281407035177e-05,
-      "loss": 5.1047,
-      "step": 32587
-    },
-    {
-      "epoch": 16.995045632333767,
-      "grad_norm": 1.4900236129760742,
-      "learning_rate": 6.776180904522614e-05,
-      "loss": 5.2436,
-      "step": 32588
-    },
-    {
-      "epoch": 16.995567144719686,
-      "grad_norm": 1.633459210395813,
-      "learning_rate": 6.776080402010051e-05,
-      "loss": 4.7248,
-      "step": 32589
-    },
-    {
-      "epoch": 16.996088657105606,
-      "grad_norm": 1.4386694431304932,
-      "learning_rate": 6.775979899497488e-05,
-      "loss": 5.2788,
-      "step": 32590
-    },
-    {
-      "epoch": 16.996610169491525,
-      "grad_norm": 1.4044768810272217,
-      "learning_rate": 6.775879396984926e-05,
-      "loss": 5.5121,
-      "step": 32591
-    },
-    {
-      "epoch": 16.997131681877445,
-      "grad_norm": 1.5841816663742065,
-      "learning_rate": 6.775778894472362e-05,
-      "loss": 5.2981,
-      "step": 32592
-    },
-    {
-      "epoch": 16.997653194263364,
-      "grad_norm": 1.3984915018081665,
-      "learning_rate": 6.775678391959798e-05,
-      "loss": 5.674,
-      "step": 32593
-    },
-    {
-      "epoch": 16.998174706649284,
-      "grad_norm": 1.5158419609069824,
-      "learning_rate": 6.775577889447236e-05,
-      "loss": 5.5053,
-      "step": 32594
-    },
-    {
-      "epoch": 16.998696219035203,
-      "grad_norm": 1.4182237386703491,
-      "learning_rate": 6.775477386934674e-05,
-      "loss": 5.4576,
-      "step": 32595
-    },
-    {
-      "epoch": 16.999217731421123,
-      "grad_norm": 1.425229787826538,
-      "learning_rate": 6.775376884422112e-05,
-      "loss": 5.5624,
-      "step": 32596
-    },
-    {
-      "epoch": 16.999739243807042,
-      "grad_norm": 1.5428942441940308,
-      "learning_rate": 6.775276381909548e-05,
-      "loss": 5.1246,
-      "step": 32597
-    },
-    {
-      "epoch": 17.000260756192958,
-      "grad_norm": 1.5670382976531982,
-      "learning_rate": 6.775175879396986e-05,
-      "loss": 5.4251,
-      "step": 32598
-    },
-    {
-      "epoch": 17.000782268578877,
-      "grad_norm": 1.5354171991348267,
-      "learning_rate": 6.775075376884422e-05,
-      "loss": 5.1945,
-      "step": 32599
-    },
-    {
-      "epoch": 17.001303780964797,
-      "grad_norm": 1.522490382194519,
-      "learning_rate": 6.77497487437186e-05,
-      "loss": 5.4182,
-      "step": 32600
-    },
-    {
-      "epoch": 17.001825293350716,
-      "grad_norm": 1.6278958320617676,
-      "learning_rate": 6.774874371859297e-05,
-      "loss": 4.8089,
-      "step": 32601
-    },
-    {
-      "epoch": 17.002346805736636,
-      "grad_norm": 1.4539995193481445,
-      "learning_rate": 6.774773869346734e-05,
-      "loss": 5.0148,
-      "step": 32602
-    },
-    {
-      "epoch": 17.002868318122555,
-      "grad_norm": 1.6232582330703735,
-      "learning_rate": 6.774673366834171e-05,
-      "loss": 4.9318,
-      "step": 32603
-    },
-    {
-      "epoch": 17.003389830508475,
-      "grad_norm": 1.519358515739441,
-      "learning_rate": 6.774572864321609e-05,
-      "loss": 5.1992,
-      "step": 32604
-    },
-    {
-      "epoch": 17.003911342894394,
-      "grad_norm": 1.586458444595337,
-      "learning_rate": 6.774472361809045e-05,
-      "loss": 5.3822,
-      "step": 32605
-    },
-    {
-      "epoch": 17.004432855280314,
-      "grad_norm": 1.4602147340774536,
-      "learning_rate": 6.774371859296483e-05,
-      "loss": 5.6345,
-      "step": 32606
-    },
-    {
-      "epoch": 17.004954367666233,
-      "grad_norm": 1.7083770036697388,
-      "learning_rate": 6.77427135678392e-05,
-      "loss": 4.621,
-      "step": 32607
-    },
-    {
-      "epoch": 17.005475880052153,
-      "grad_norm": 1.4725090265274048,
-      "learning_rate": 6.774170854271357e-05,
-      "loss": 4.8784,
-      "step": 32608
-    },
-    {
-      "epoch": 17.005997392438072,
-      "grad_norm": 1.5603736639022827,
-      "learning_rate": 6.774070351758795e-05,
-      "loss": 5.6224,
-      "step": 32609
-    },
-    {
-      "epoch": 17.006518904823988,
-      "grad_norm": 1.568301796913147,
-      "learning_rate": 6.773969849246231e-05,
-      "loss": 5.349,
-      "step": 32610
-    },
-    {
-      "epoch": 17.007040417209907,
-      "grad_norm": 1.4682643413543701,
-      "learning_rate": 6.773869346733669e-05,
-      "loss": 5.4426,
-      "step": 32611
-    },
-    {
-      "epoch": 17.007561929595827,
-      "grad_norm": 1.3911458253860474,
-      "learning_rate": 6.773768844221105e-05,
-      "loss": 5.5832,
-      "step": 32612
-    },
-    {
-      "epoch": 17.008083441981746,
-      "grad_norm": 1.457090139389038,
-      "learning_rate": 6.773668341708543e-05,
-      "loss": 5.5655,
-      "step": 32613
-    },
-    {
-      "epoch": 17.008604954367666,
-      "grad_norm": 1.5606069564819336,
-      "learning_rate": 6.77356783919598e-05,
-      "loss": 5.4483,
-      "step": 32614
-    },
-    {
-      "epoch": 17.009126466753585,
-      "grad_norm": 1.5327119827270508,
-      "learning_rate": 6.773467336683417e-05,
-      "loss": 5.3619,
-      "step": 32615
-    },
-    {
-      "epoch": 17.009647979139505,
-      "grad_norm": 1.5005731582641602,
-      "learning_rate": 6.773366834170854e-05,
-      "loss": 4.9634,
-      "step": 32616
-    },
-    {
-      "epoch": 17.010169491525424,
-      "grad_norm": 1.5539613962173462,
-      "learning_rate": 6.773266331658292e-05,
-      "loss": 5.4253,
-      "step": 32617
-    },
-    {
-      "epoch": 17.010691003911344,
-      "grad_norm": 1.4329841136932373,
-      "learning_rate": 6.77316582914573e-05,
-      "loss": 5.5229,
-      "step": 32618
-    },
-    {
-      "epoch": 17.011212516297263,
-      "grad_norm": 1.5503770112991333,
-      "learning_rate": 6.773065326633166e-05,
-      "loss": 5.6102,
-      "step": 32619
-    },
-    {
-      "epoch": 17.011734028683183,
-      "grad_norm": 1.4633795022964478,
-      "learning_rate": 6.772964824120604e-05,
-      "loss": 5.5581,
-      "step": 32620
-    },
-    {
-      "epoch": 17.012255541069102,
-      "grad_norm": 1.458422064781189,
-      "learning_rate": 6.77286432160804e-05,
-      "loss": 5.1167,
-      "step": 32621
-    },
-    {
-      "epoch": 17.012777053455018,
-      "grad_norm": 1.5293638706207275,
-      "learning_rate": 6.772763819095478e-05,
-      "loss": 5.1624,
-      "step": 32622
-    },
-    {
-      "epoch": 17.013298565840937,
-      "grad_norm": 1.5398755073547363,
-      "learning_rate": 6.772663316582914e-05,
-      "loss": 5.1968,
-      "step": 32623
-    },
-    {
-      "epoch": 17.013820078226857,
-      "grad_norm": 1.4639934301376343,
-      "learning_rate": 6.772562814070352e-05,
-      "loss": 5.3432,
-      "step": 32624
-    },
-    {
-      "epoch": 17.014341590612776,
-      "grad_norm": 1.4043803215026855,
-      "learning_rate": 6.772462311557788e-05,
-      "loss": 5.3961,
-      "step": 32625
-    },
-    {
-      "epoch": 17.014863102998696,
-      "grad_norm": 1.5539071559906006,
-      "learning_rate": 6.772361809045226e-05,
-      "loss": 5.2406,
-      "step": 32626
-    },
-    {
-      "epoch": 17.015384615384615,
-      "grad_norm": 1.4409493207931519,
-      "learning_rate": 6.772261306532664e-05,
-      "loss": 5.2237,
-      "step": 32627
-    },
-    {
-      "epoch": 17.015906127770535,
-      "grad_norm": 1.3490709066390991,
-      "learning_rate": 6.772160804020102e-05,
-      "loss": 5.5171,
-      "step": 32628
-    },
-    {
-      "epoch": 17.016427640156454,
-      "grad_norm": 1.5530462265014648,
-      "learning_rate": 6.772060301507538e-05,
-      "loss": 5.1377,
-      "step": 32629
-    },
-    {
-      "epoch": 17.016949152542374,
-      "grad_norm": 1.6804955005645752,
-      "learning_rate": 6.771959798994976e-05,
-      "loss": 4.8724,
-      "step": 32630
-    },
-    {
-      "epoch": 17.017470664928293,
-      "grad_norm": 1.6028428077697754,
-      "learning_rate": 6.771859296482412e-05,
-      "loss": 4.7949,
-      "step": 32631
-    },
-    {
-      "epoch": 17.017992177314213,
-      "grad_norm": 1.5633089542388916,
-      "learning_rate": 6.771758793969849e-05,
-      "loss": 5.1963,
-      "step": 32632
-    },
-    {
-      "epoch": 17.018513689700132,
-      "grad_norm": 1.562380313873291,
-      "learning_rate": 6.771658291457287e-05,
-      "loss": 5.3035,
-      "step": 32633
-    },
-    {
-      "epoch": 17.019035202086048,
-      "grad_norm": 1.6581627130508423,
-      "learning_rate": 6.771557788944723e-05,
-      "loss": 5.0388,
-      "step": 32634
-    },
-    {
-      "epoch": 17.019556714471967,
-      "grad_norm": 1.5315324068069458,
-      "learning_rate": 6.771457286432161e-05,
-      "loss": 5.6578,
-      "step": 32635
-    },
-    {
-      "epoch": 17.020078226857887,
-      "grad_norm": 1.5168949365615845,
-      "learning_rate": 6.771356783919597e-05,
-      "loss": 5.2445,
-      "step": 32636
-    },
-    {
-      "epoch": 17.020599739243806,
-      "grad_norm": 1.4864188432693481,
-      "learning_rate": 6.771256281407035e-05,
-      "loss": 5.685,
-      "step": 32637
-    },
-    {
-      "epoch": 17.021121251629726,
-      "grad_norm": 1.5467888116836548,
-      "learning_rate": 6.771155778894473e-05,
-      "loss": 5.2636,
-      "step": 32638
-    },
-    {
-      "epoch": 17.021642764015645,
-      "grad_norm": 1.3261765241622925,
-      "learning_rate": 6.77105527638191e-05,
-      "loss": 5.7187,
-      "step": 32639
-    },
-    {
-      "epoch": 17.022164276401565,
-      "grad_norm": 1.4192267656326294,
-      "learning_rate": 6.770954773869347e-05,
-      "loss": 5.4172,
-      "step": 32640
-    },
-    {
-      "epoch": 17.022685788787484,
-      "grad_norm": 1.3727198839187622,
-      "learning_rate": 6.770854271356785e-05,
-      "loss": 5.7233,
-      "step": 32641
-    },
-    {
-      "epoch": 17.023207301173404,
-      "grad_norm": 1.3780335187911987,
-      "learning_rate": 6.770753768844221e-05,
-      "loss": 5.4708,
-      "step": 32642
-    },
-    {
-      "epoch": 17.023728813559323,
-      "grad_norm": 1.4693769216537476,
-      "learning_rate": 6.770653266331659e-05,
-      "loss": 5.233,
-      "step": 32643
-    },
-    {
-      "epoch": 17.024250325945243,
-      "grad_norm": 1.5226670503616333,
-      "learning_rate": 6.770552763819095e-05,
-      "loss": 5.6181,
-      "step": 32644
-    },
-    {
-      "epoch": 17.02477183833116,
-      "grad_norm": 1.4405134916305542,
-      "learning_rate": 6.770452261306533e-05,
-      "loss": 5.5426,
-      "step": 32645
-    },
-    {
-      "epoch": 17.025293350717078,
-      "grad_norm": 1.4239792823791504,
-      "learning_rate": 6.77035175879397e-05,
-      "loss": 5.4821,
-      "step": 32646
-    },
-    {
-      "epoch": 17.025814863102998,
-      "grad_norm": 1.4079293012619019,
-      "learning_rate": 6.770251256281407e-05,
-      "loss": 5.6757,
-      "step": 32647
-    },
-    {
-      "epoch": 17.026336375488917,
-      "grad_norm": 1.458410382270813,
-      "learning_rate": 6.770150753768845e-05,
-      "loss": 5.264,
-      "step": 32648
-    },
-    {
-      "epoch": 17.026857887874836,
-      "grad_norm": 1.4556620121002197,
-      "learning_rate": 6.770050251256282e-05,
-      "loss": 5.4918,
-      "step": 32649
-    },
-    {
-      "epoch": 17.027379400260756,
-      "grad_norm": 1.4818741083145142,
-      "learning_rate": 6.76994974874372e-05,
-      "loss": 5.4618,
-      "step": 32650
-    },
-    {
-      "epoch": 17.027900912646675,
-      "grad_norm": 1.427308201789856,
-      "learning_rate": 6.769849246231156e-05,
-      "loss": 5.5824,
-      "step": 32651
-    },
-    {
-      "epoch": 17.028422425032595,
-      "grad_norm": 1.5307976007461548,
-      "learning_rate": 6.769748743718594e-05,
-      "loss": 5.3834,
-      "step": 32652
-    },
-    {
-      "epoch": 17.028943937418514,
-      "grad_norm": 1.4232450723648071,
-      "learning_rate": 6.76964824120603e-05,
-      "loss": 5.0242,
-      "step": 32653
-    },
-    {
-      "epoch": 17.029465449804434,
-      "grad_norm": 1.506750464439392,
-      "learning_rate": 6.769547738693468e-05,
-      "loss": 5.1017,
-      "step": 32654
-    },
-    {
-      "epoch": 17.029986962190353,
-      "grad_norm": 1.495019555091858,
-      "learning_rate": 6.769447236180904e-05,
-      "loss": 5.1299,
-      "step": 32655
-    },
-    {
-      "epoch": 17.030508474576273,
-      "grad_norm": 1.5037857294082642,
-      "learning_rate": 6.769346733668342e-05,
-      "loss": 5.1428,
-      "step": 32656
-    },
-    {
-      "epoch": 17.03102998696219,
-      "grad_norm": 1.382068157196045,
-      "learning_rate": 6.769246231155779e-05,
-      "loss": 5.3528,
-      "step": 32657
-    },
-    {
-      "epoch": 17.031551499348108,
-      "grad_norm": 1.5090495347976685,
-      "learning_rate": 6.769145728643216e-05,
-      "loss": 4.8587,
-      "step": 32658
-    },
-    {
-      "epoch": 17.032073011734028,
-      "grad_norm": 1.5756263732910156,
-      "learning_rate": 6.769045226130654e-05,
-      "loss": 5.2772,
-      "step": 32659
-    },
-    {
-      "epoch": 17.032594524119947,
-      "grad_norm": 1.432529330253601,
-      "learning_rate": 6.76894472361809e-05,
-      "loss": 5.2049,
-      "step": 32660
-    },
-    {
-      "epoch": 17.033116036505866,
-      "grad_norm": 1.6055145263671875,
-      "learning_rate": 6.768844221105528e-05,
-      "loss": 4.8575,
-      "step": 32661
-    },
-    {
-      "epoch": 17.033637548891786,
-      "grad_norm": 1.5106993913650513,
-      "learning_rate": 6.768743718592965e-05,
-      "loss": 5.4161,
-      "step": 32662
-    },
-    {
-      "epoch": 17.034159061277705,
-      "grad_norm": 1.5019325017929077,
-      "learning_rate": 6.768643216080403e-05,
-      "loss": 4.9665,
-      "step": 32663
-    },
-    {
-      "epoch": 17.034680573663625,
-      "grad_norm": 1.482788324356079,
-      "learning_rate": 6.768542713567839e-05,
-      "loss": 5.2746,
-      "step": 32664
-    },
-    {
-      "epoch": 17.035202086049544,
-      "grad_norm": 1.4645472764968872,
-      "learning_rate": 6.768442211055277e-05,
-      "loss": 4.9096,
-      "step": 32665
-    },
-    {
-      "epoch": 17.035723598435464,
-      "grad_norm": 1.536759376525879,
-      "learning_rate": 6.768341708542713e-05,
-      "loss": 5.1574,
-      "step": 32666
-    },
-    {
-      "epoch": 17.036245110821383,
-      "grad_norm": 1.5662240982055664,
-      "learning_rate": 6.768241206030151e-05,
-      "loss": 5.1911,
-      "step": 32667
-    },
-    {
-      "epoch": 17.036766623207303,
-      "grad_norm": 1.5822181701660156,
-      "learning_rate": 6.768140703517589e-05,
-      "loss": 5.0747,
-      "step": 32668
-    },
-    {
-      "epoch": 17.03728813559322,
-      "grad_norm": 1.4780006408691406,
-      "learning_rate": 6.768040201005026e-05,
-      "loss": 5.1508,
-      "step": 32669
-    },
-    {
-      "epoch": 17.037809647979138,
-      "grad_norm": 1.522508978843689,
-      "learning_rate": 6.767939698492463e-05,
-      "loss": 5.4859,
-      "step": 32670
-    },
-    {
-      "epoch": 17.038331160365058,
-      "grad_norm": 1.5402580499649048,
-      "learning_rate": 6.767839195979901e-05,
-      "loss": 5.2505,
-      "step": 32671
-    },
-    {
-      "epoch": 17.038852672750977,
-      "grad_norm": 1.5036377906799316,
-      "learning_rate": 6.767738693467337e-05,
-      "loss": 5.2457,
-      "step": 32672
-    },
-    {
-      "epoch": 17.039374185136897,
-      "grad_norm": 1.5307968854904175,
-      "learning_rate": 6.767638190954774e-05,
-      "loss": 5.2834,
-      "step": 32673
-    },
-    {
-      "epoch": 17.039895697522816,
-      "grad_norm": 1.4937106370925903,
-      "learning_rate": 6.767537688442211e-05,
-      "loss": 5.245,
-      "step": 32674
-    },
-    {
-      "epoch": 17.040417209908735,
-      "grad_norm": 1.3737032413482666,
-      "learning_rate": 6.767437185929648e-05,
-      "loss": 5.0971,
-      "step": 32675
-    },
-    {
-      "epoch": 17.040938722294655,
-      "grad_norm": 1.5311421155929565,
-      "learning_rate": 6.767336683417086e-05,
-      "loss": 5.1244,
-      "step": 32676
-    },
-    {
-      "epoch": 17.041460234680574,
-      "grad_norm": 1.4968583583831787,
-      "learning_rate": 6.767236180904522e-05,
-      "loss": 5.5738,
-      "step": 32677
-    },
-    {
-      "epoch": 17.041981747066494,
-      "grad_norm": 1.38335120677948,
-      "learning_rate": 6.76713567839196e-05,
-      "loss": 5.6969,
-      "step": 32678
-    },
-    {
-      "epoch": 17.042503259452413,
-      "grad_norm": 1.6288938522338867,
-      "learning_rate": 6.767035175879398e-05,
-      "loss": 5.2654,
-      "step": 32679
-    },
-    {
-      "epoch": 17.043024771838333,
-      "grad_norm": 1.5604724884033203,
-      "learning_rate": 6.766934673366835e-05,
-      "loss": 4.5048,
-      "step": 32680
-    },
-    {
-      "epoch": 17.04354628422425,
-      "grad_norm": 1.5755701065063477,
-      "learning_rate": 6.766834170854272e-05,
-      "loss": 4.9701,
-      "step": 32681
-    },
-    {
-      "epoch": 17.044067796610168,
-      "grad_norm": 1.5454376935958862,
-      "learning_rate": 6.76673366834171e-05,
-      "loss": 4.8289,
-      "step": 32682
-    },
-    {
-      "epoch": 17.044589308996088,
-      "grad_norm": 1.6236348152160645,
-      "learning_rate": 6.766633165829146e-05,
-      "loss": 4.7291,
-      "step": 32683
-    },
-    {
-      "epoch": 17.045110821382007,
-      "grad_norm": 1.4493699073791504,
-      "learning_rate": 6.766532663316584e-05,
-      "loss": 5.1784,
-      "step": 32684
-    },
-    {
-      "epoch": 17.045632333767927,
-      "grad_norm": 1.5046217441558838,
-      "learning_rate": 6.76643216080402e-05,
-      "loss": 5.361,
-      "step": 32685
-    },
-    {
-      "epoch": 17.046153846153846,
-      "grad_norm": 1.4661073684692383,
-      "learning_rate": 6.766331658291457e-05,
-      "loss": 5.584,
-      "step": 32686
-    },
-    {
-      "epoch": 17.046675358539765,
-      "grad_norm": 1.5264085531234741,
-      "learning_rate": 6.766231155778894e-05,
-      "loss": 5.313,
-      "step": 32687
-    },
-    {
-      "epoch": 17.047196870925685,
-      "grad_norm": 1.6515746116638184,
-      "learning_rate": 6.766130653266332e-05,
-      "loss": 4.8227,
-      "step": 32688
-    },
-    {
-      "epoch": 17.047718383311604,
-      "grad_norm": 1.4653654098510742,
-      "learning_rate": 6.76603015075377e-05,
-      "loss": 5.3798,
-      "step": 32689
-    },
-    {
-      "epoch": 17.048239895697524,
-      "grad_norm": 1.4337238073349,
-      "learning_rate": 6.765929648241206e-05,
-      "loss": 5.4721,
-      "step": 32690
-    },
-    {
-      "epoch": 17.048761408083443,
-      "grad_norm": 1.4874672889709473,
-      "learning_rate": 6.765829145728644e-05,
-      "loss": 5.5412,
-      "step": 32691
-    },
-    {
-      "epoch": 17.049282920469363,
-      "grad_norm": 1.4808942079544067,
-      "learning_rate": 6.76572864321608e-05,
-      "loss": 5.2112,
-      "step": 32692
-    },
-    {
-      "epoch": 17.04980443285528,
-      "grad_norm": 1.4875415563583374,
-      "learning_rate": 6.765628140703518e-05,
-      "loss": 4.5986,
-      "step": 32693
-    },
-    {
-      "epoch": 17.050325945241198,
-      "grad_norm": 2.0480470657348633,
-      "learning_rate": 6.765527638190955e-05,
-      "loss": 5.1848,
-      "step": 32694
-    },
-    {
-      "epoch": 17.050847457627118,
-      "grad_norm": 1.696158528327942,
-      "learning_rate": 6.765427135678393e-05,
-      "loss": 5.1928,
-      "step": 32695
-    },
-    {
-      "epoch": 17.051368970013037,
-      "grad_norm": 1.4214346408843994,
-      "learning_rate": 6.765326633165829e-05,
-      "loss": 5.2864,
-      "step": 32696
-    },
-    {
-      "epoch": 17.051890482398957,
-      "grad_norm": 1.5173922777175903,
-      "learning_rate": 6.765226130653267e-05,
-      "loss": 5.3088,
-      "step": 32697
-    },
-    {
-      "epoch": 17.052411994784876,
-      "grad_norm": 1.4901471138000488,
-      "learning_rate": 6.765125628140703e-05,
-      "loss": 5.0739,
-      "step": 32698
-    },
-    {
-      "epoch": 17.052933507170795,
-      "grad_norm": 1.4665706157684326,
-      "learning_rate": 6.765025125628141e-05,
-      "loss": 4.9309,
-      "step": 32699
-    },
-    {
-      "epoch": 17.053455019556715,
-      "grad_norm": 1.4632797241210938,
-      "learning_rate": 6.764924623115579e-05,
-      "loss": 5.4377,
-      "step": 32700
-    },
-    {
-      "epoch": 17.053976531942634,
-      "grad_norm": 1.39524507522583,
-      "learning_rate": 6.764824120603015e-05,
-      "loss": 5.6188,
-      "step": 32701
-    },
-    {
-      "epoch": 17.054498044328554,
-      "grad_norm": 1.512126088142395,
-      "learning_rate": 6.764723618090453e-05,
-      "loss": 4.7002,
-      "step": 32702
-    },
-    {
-      "epoch": 17.055019556714473,
-      "grad_norm": 1.4467215538024902,
-      "learning_rate": 6.76462311557789e-05,
-      "loss": 5.5849,
-      "step": 32703
-    },
-    {
-      "epoch": 17.055541069100393,
-      "grad_norm": 1.5909221172332764,
-      "learning_rate": 6.764522613065327e-05,
-      "loss": 4.8495,
-      "step": 32704
-    },
-    {
-      "epoch": 17.05606258148631,
-      "grad_norm": 1.435481309890747,
-      "learning_rate": 6.764422110552764e-05,
-      "loss": 5.7049,
-      "step": 32705
-    },
-    {
-      "epoch": 17.056584093872228,
-      "grad_norm": 1.5251350402832031,
-      "learning_rate": 6.764321608040201e-05,
-      "loss": 5.5684,
-      "step": 32706
-    },
-    {
-      "epoch": 17.057105606258148,
-      "grad_norm": 1.549417495727539,
-      "learning_rate": 6.764221105527638e-05,
-      "loss": 5.1989,
-      "step": 32707
-    },
-    {
-      "epoch": 17.057627118644067,
-      "grad_norm": 1.6187587976455688,
-      "learning_rate": 6.764120603015076e-05,
-      "loss": 5.1165,
-      "step": 32708
-    },
-    {
-      "epoch": 17.058148631029987,
-      "grad_norm": 1.523267388343811,
-      "learning_rate": 6.764020100502513e-05,
-      "loss": 5.23,
-      "step": 32709
-    },
-    {
-      "epoch": 17.058670143415906,
-      "grad_norm": 1.4713726043701172,
-      "learning_rate": 6.763919597989951e-05,
-      "loss": 5.7277,
-      "step": 32710
-    },
-    {
-      "epoch": 17.059191655801826,
-      "grad_norm": 1.6037746667861938,
-      "learning_rate": 6.763819095477388e-05,
-      "loss": 5.0004,
-      "step": 32711
-    },
-    {
-      "epoch": 17.059713168187745,
-      "grad_norm": 1.4857662916183472,
-      "learning_rate": 6.763718592964824e-05,
-      "loss": 5.5369,
-      "step": 32712
-    },
-    {
-      "epoch": 17.060234680573664,
-      "grad_norm": 1.4755051136016846,
-      "learning_rate": 6.763618090452262e-05,
-      "loss": 5.1794,
-      "step": 32713
-    },
-    {
-      "epoch": 17.060756192959584,
-      "grad_norm": 1.531886100769043,
-      "learning_rate": 6.763517587939698e-05,
-      "loss": 5.3273,
-      "step": 32714
-    },
-    {
-      "epoch": 17.061277705345503,
-      "grad_norm": 1.5029942989349365,
-      "learning_rate": 6.763417085427136e-05,
-      "loss": 5.4988,
-      "step": 32715
-    },
-    {
-      "epoch": 17.061799217731423,
-      "grad_norm": 1.4419468641281128,
-      "learning_rate": 6.763316582914572e-05,
-      "loss": 5.1225,
-      "step": 32716
-    },
-    {
-      "epoch": 17.06232073011734,
-      "grad_norm": 1.5093393325805664,
-      "learning_rate": 6.76321608040201e-05,
-      "loss": 5.6438,
-      "step": 32717
-    },
-    {
-      "epoch": 17.062842242503258,
-      "grad_norm": 1.5515681505203247,
-      "learning_rate": 6.763115577889447e-05,
-      "loss": 5.1907,
-      "step": 32718
-    },
-    {
-      "epoch": 17.063363754889178,
-      "grad_norm": 1.3721911907196045,
-      "learning_rate": 6.763015075376884e-05,
-      "loss": 4.8127,
-      "step": 32719
-    },
-    {
-      "epoch": 17.063885267275097,
-      "grad_norm": 1.464516520500183,
-      "learning_rate": 6.762914572864322e-05,
-      "loss": 5.6205,
-      "step": 32720
-    },
-    {
-      "epoch": 17.064406779661017,
-      "grad_norm": 1.5770237445831299,
-      "learning_rate": 6.76281407035176e-05,
-      "loss": 5.3732,
-      "step": 32721
-    },
-    {
-      "epoch": 17.064928292046936,
-      "grad_norm": 1.4601962566375732,
-      "learning_rate": 6.762713567839196e-05,
-      "loss": 5.4901,
-      "step": 32722
-    },
-    {
-      "epoch": 17.065449804432856,
-      "grad_norm": 1.5009537935256958,
-      "learning_rate": 6.762613065326634e-05,
-      "loss": 5.3904,
-      "step": 32723
-    },
-    {
-      "epoch": 17.065971316818775,
-      "grad_norm": 1.5564525127410889,
-      "learning_rate": 6.76251256281407e-05,
-      "loss": 4.9784,
-      "step": 32724
-    },
-    {
-      "epoch": 17.066492829204694,
-      "grad_norm": 1.7365459203720093,
-      "learning_rate": 6.762412060301507e-05,
-      "loss": 4.6995,
-      "step": 32725
-    },
-    {
-      "epoch": 17.067014341590614,
-      "grad_norm": 1.4277204275131226,
-      "learning_rate": 6.762311557788945e-05,
-      "loss": 5.4229,
-      "step": 32726
-    },
-    {
-      "epoch": 17.067535853976533,
-      "grad_norm": 1.5390962362289429,
-      "learning_rate": 6.762211055276381e-05,
-      "loss": 5.0059,
-      "step": 32727
-    },
-    {
-      "epoch": 17.068057366362453,
-      "grad_norm": 1.6140438318252563,
-      "learning_rate": 6.762110552763819e-05,
-      "loss": 4.7953,
-      "step": 32728
-    },
-    {
-      "epoch": 17.06857887874837,
-      "grad_norm": 1.5227715969085693,
-      "learning_rate": 6.762010050251257e-05,
-      "loss": 5.314,
-      "step": 32729
-    },
-    {
-      "epoch": 17.06910039113429,
-      "grad_norm": 1.567001223564148,
-      "learning_rate": 6.761909547738695e-05,
-      "loss": 5.3642,
-      "step": 32730
-    },
-    {
-      "epoch": 17.069621903520208,
-      "grad_norm": 1.4918063879013062,
-      "learning_rate": 6.761809045226131e-05,
-      "loss": 5.0481,
-      "step": 32731
-    },
-    {
-      "epoch": 17.070143415906127,
-      "grad_norm": 1.3693894147872925,
-      "learning_rate": 6.761708542713569e-05,
-      "loss": 5.4249,
-      "step": 32732
-    },
-    {
-      "epoch": 17.070664928292047,
-      "grad_norm": 1.4580495357513428,
-      "learning_rate": 6.761608040201005e-05,
-      "loss": 5.7355,
-      "step": 32733
-    },
-    {
-      "epoch": 17.071186440677966,
-      "grad_norm": 1.4347749948501587,
-      "learning_rate": 6.761507537688443e-05,
-      "loss": 5.5107,
-      "step": 32734
-    },
-    {
-      "epoch": 17.071707953063886,
-      "grad_norm": 1.3903610706329346,
-      "learning_rate": 6.76140703517588e-05,
-      "loss": 5.5455,
-      "step": 32735
-    },
-    {
-      "epoch": 17.072229465449805,
-      "grad_norm": 1.4497184753417969,
-      "learning_rate": 6.761306532663317e-05,
-      "loss": 5.2837,
-      "step": 32736
-    },
-    {
-      "epoch": 17.072750977835724,
-      "grad_norm": 1.4266901016235352,
-      "learning_rate": 6.761206030150754e-05,
-      "loss": 5.693,
-      "step": 32737
-    },
-    {
-      "epoch": 17.073272490221644,
-      "grad_norm": 1.4772578477859497,
-      "learning_rate": 6.761105527638191e-05,
-      "loss": 5.5469,
-      "step": 32738
-    },
-    {
-      "epoch": 17.073794002607563,
-      "grad_norm": 1.592820644378662,
-      "learning_rate": 6.761005025125628e-05,
-      "loss": 4.814,
-      "step": 32739
-    },
-    {
-      "epoch": 17.07431551499348,
-      "grad_norm": 1.4884593486785889,
-      "learning_rate": 6.760904522613066e-05,
-      "loss": 4.9805,
-      "step": 32740
-    },
-    {
-      "epoch": 17.0748370273794,
-      "grad_norm": 1.4994062185287476,
-      "learning_rate": 6.760804020100503e-05,
-      "loss": 5.6004,
-      "step": 32741
-    },
-    {
-      "epoch": 17.07535853976532,
-      "grad_norm": 1.4601818323135376,
-      "learning_rate": 6.76070351758794e-05,
-      "loss": 5.4466,
-      "step": 32742
-    },
-    {
-      "epoch": 17.075880052151238,
-      "grad_norm": 1.5023047924041748,
-      "learning_rate": 6.760603015075378e-05,
-      "loss": 5.5298,
-      "step": 32743
-    },
-    {
-      "epoch": 17.076401564537157,
-      "grad_norm": 1.4750057458877563,
-      "learning_rate": 6.760502512562814e-05,
-      "loss": 5.5194,
-      "step": 32744
-    },
-    {
-      "epoch": 17.076923076923077,
-      "grad_norm": 1.5165528059005737,
-      "learning_rate": 6.760402010050252e-05,
-      "loss": 5.8886,
-      "step": 32745
-    },
-    {
-      "epoch": 17.077444589308996,
-      "grad_norm": 1.5506043434143066,
-      "learning_rate": 6.760301507537688e-05,
-      "loss": 4.8954,
-      "step": 32746
-    },
-    {
-      "epoch": 17.077966101694916,
-      "grad_norm": 1.5712453126907349,
-      "learning_rate": 6.760201005025126e-05,
-      "loss": 5.424,
-      "step": 32747
-    },
-    {
-      "epoch": 17.078487614080835,
-      "grad_norm": 1.5930473804473877,
-      "learning_rate": 6.760100502512563e-05,
-      "loss": 5.4255,
-      "step": 32748
-    },
-    {
-      "epoch": 17.079009126466755,
-      "grad_norm": 1.539815068244934,
-      "learning_rate": 6.76e-05,
-      "loss": 5.2409,
-      "step": 32749
-    },
-    {
-      "epoch": 17.079530638852674,
-      "grad_norm": 1.4858801364898682,
-      "learning_rate": 6.759899497487438e-05,
-      "loss": 5.3951,
-      "step": 32750
-    },
-    {
-      "epoch": 17.080052151238593,
-      "grad_norm": 1.4898788928985596,
-      "learning_rate": 6.759798994974876e-05,
-      "loss": 5.3002,
-      "step": 32751
-    },
-    {
-      "epoch": 17.08057366362451,
-      "grad_norm": 1.6627202033996582,
-      "learning_rate": 6.759698492462312e-05,
-      "loss": 4.9211,
-      "step": 32752
-    },
-    {
-      "epoch": 17.08109517601043,
-      "grad_norm": 1.5020124912261963,
-      "learning_rate": 6.759597989949749e-05,
-      "loss": 5.3173,
-      "step": 32753
-    },
-    {
-      "epoch": 17.08161668839635,
-      "grad_norm": 1.4598956108093262,
-      "learning_rate": 6.759497487437187e-05,
-      "loss": 5.7208,
-      "step": 32754
-    },
-    {
-      "epoch": 17.082138200782268,
-      "grad_norm": 1.4948495626449585,
-      "learning_rate": 6.759396984924623e-05,
-      "loss": 5.3327,
-      "step": 32755
-    },
-    {
-      "epoch": 17.082659713168187,
-      "grad_norm": 1.5163607597351074,
-      "learning_rate": 6.759296482412061e-05,
-      "loss": 5.3067,
-      "step": 32756
-    },
-    {
-      "epoch": 17.083181225554107,
-      "grad_norm": 1.4442147016525269,
-      "learning_rate": 6.759195979899497e-05,
-      "loss": 5.3452,
-      "step": 32757
-    },
-    {
-      "epoch": 17.083702737940026,
-      "grad_norm": 1.590556263923645,
-      "learning_rate": 6.759095477386935e-05,
-      "loss": 5.5242,
-      "step": 32758
-    },
-    {
-      "epoch": 17.084224250325946,
-      "grad_norm": 1.453879714012146,
-      "learning_rate": 6.758994974874371e-05,
-      "loss": 5.1154,
-      "step": 32759
-    },
-    {
-      "epoch": 17.084745762711865,
-      "grad_norm": 1.5159367322921753,
-      "learning_rate": 6.758894472361809e-05,
-      "loss": 5.4297,
-      "step": 32760
-    },
-    {
-      "epoch": 17.085267275097785,
-      "grad_norm": 1.5488022565841675,
-      "learning_rate": 6.758793969849247e-05,
-      "loss": 5.1387,
-      "step": 32761
-    },
-    {
-      "epoch": 17.085788787483704,
-      "grad_norm": 1.5027401447296143,
-      "learning_rate": 6.758693467336685e-05,
-      "loss": 5.2209,
-      "step": 32762
-    },
-    {
-      "epoch": 17.086310299869623,
-      "grad_norm": 1.4556102752685547,
-      "learning_rate": 6.758592964824121e-05,
-      "loss": 5.4966,
-      "step": 32763
-    },
-    {
-      "epoch": 17.08683181225554,
-      "grad_norm": 1.6420769691467285,
-      "learning_rate": 6.758492462311559e-05,
-      "loss": 4.8161,
-      "step": 32764
-    },
-    {
-      "epoch": 17.08735332464146,
-      "grad_norm": 1.593593716621399,
-      "learning_rate": 6.758391959798995e-05,
-      "loss": 5.3397,
-      "step": 32765
-    },
-    {
-      "epoch": 17.08787483702738,
-      "grad_norm": 1.5082271099090576,
-      "learning_rate": 6.758291457286432e-05,
-      "loss": 5.633,
-      "step": 32766
-    },
-    {
-      "epoch": 17.088396349413298,
-      "grad_norm": 1.4972529411315918,
-      "learning_rate": 6.75819095477387e-05,
-      "loss": 5.3673,
-      "step": 32767
-    },
-    {
-      "epoch": 17.088917861799217,
-      "grad_norm": 1.5016146898269653,
-      "learning_rate": 6.758090452261306e-05,
-      "loss": 5.1869,
-      "step": 32768
-    },
-    {
-      "epoch": 17.089439374185137,
-      "grad_norm": 1.4750423431396484,
-      "learning_rate": 6.757989949748744e-05,
-      "loss": 5.3986,
-      "step": 32769
-    },
-    {
-      "epoch": 17.089960886571056,
-      "grad_norm": 1.467110276222229,
-      "learning_rate": 6.75788944723618e-05,
-      "loss": 5.5351,
-      "step": 32770
-    },
-    {
-      "epoch": 17.090482398956976,
-      "grad_norm": 1.5313583612442017,
-      "learning_rate": 6.757788944723618e-05,
-      "loss": 5.5083,
-      "step": 32771
-    },
-    {
-      "epoch": 17.091003911342895,
-      "grad_norm": 1.411472201347351,
-      "learning_rate": 6.757688442211056e-05,
-      "loss": 5.0354,
-      "step": 32772
-    },
-    {
-      "epoch": 17.091525423728815,
-      "grad_norm": 1.4584510326385498,
-      "learning_rate": 6.757587939698494e-05,
-      "loss": 5.4222,
-      "step": 32773
-    },
-    {
-      "epoch": 17.092046936114734,
-      "grad_norm": 1.5042393207550049,
-      "learning_rate": 6.75748743718593e-05,
-      "loss": 5.2893,
-      "step": 32774
-    },
-    {
-      "epoch": 17.092568448500653,
-      "grad_norm": 1.5010935068130493,
-      "learning_rate": 6.757386934673368e-05,
-      "loss": 5.4366,
-      "step": 32775
-    },
-    {
-      "epoch": 17.09308996088657,
-      "grad_norm": 1.5393823385238647,
-      "learning_rate": 6.757286432160804e-05,
-      "loss": 5.3788,
-      "step": 32776
-    },
-    {
-      "epoch": 17.09361147327249,
-      "grad_norm": 1.5449455976486206,
-      "learning_rate": 6.757185929648242e-05,
-      "loss": 5.0745,
-      "step": 32777
-    },
-    {
-      "epoch": 17.09413298565841,
-      "grad_norm": 1.4584803581237793,
-      "learning_rate": 6.757085427135678e-05,
-      "loss": 5.262,
-      "step": 32778
-    },
-    {
-      "epoch": 17.094654498044328,
-      "grad_norm": 1.4740959405899048,
-      "learning_rate": 6.756984924623115e-05,
-      "loss": 5.6124,
-      "step": 32779
-    },
-    {
-      "epoch": 17.095176010430247,
-      "grad_norm": 1.6076184511184692,
-      "learning_rate": 6.756884422110553e-05,
-      "loss": 4.9143,
-      "step": 32780
-    },
-    {
-      "epoch": 17.095697522816167,
-      "grad_norm": 1.4670206308364868,
-      "learning_rate": 6.75678391959799e-05,
-      "loss": 5.4286,
-      "step": 32781
-    },
-    {
-      "epoch": 17.096219035202086,
-      "grad_norm": 1.415895700454712,
-      "learning_rate": 6.756683417085428e-05,
-      "loss": 5.3001,
-      "step": 32782
-    },
-    {
-      "epoch": 17.096740547588006,
-      "grad_norm": 1.4286445379257202,
-      "learning_rate": 6.756582914572865e-05,
-      "loss": 5.4492,
-      "step": 32783
-    },
-    {
-      "epoch": 17.097262059973925,
-      "grad_norm": 1.4280061721801758,
-      "learning_rate": 6.756482412060302e-05,
-      "loss": 5.6781,
-      "step": 32784
-    },
-    {
-      "epoch": 17.097783572359845,
-      "grad_norm": 1.4306566715240479,
-      "learning_rate": 6.756381909547739e-05,
-      "loss": 5.4856,
-      "step": 32785
-    },
-    {
-      "epoch": 17.098305084745764,
-      "grad_norm": 1.5090337991714478,
-      "learning_rate": 6.756281407035177e-05,
-      "loss": 5.1862,
-      "step": 32786
-    },
-    {
-      "epoch": 17.098826597131684,
-      "grad_norm": 1.4397791624069214,
-      "learning_rate": 6.756180904522613e-05,
-      "loss": 5.3678,
-      "step": 32787
-    },
-    {
-      "epoch": 17.0993481095176,
-      "grad_norm": 1.4638454914093018,
-      "learning_rate": 6.756080402010051e-05,
-      "loss": 5.8525,
-      "step": 32788
-    },
-    {
-      "epoch": 17.09986962190352,
-      "grad_norm": 1.4718399047851562,
-      "learning_rate": 6.755979899497487e-05,
-      "loss": 5.0696,
-      "step": 32789
-    },
-    {
-      "epoch": 17.10039113428944,
-      "grad_norm": 1.487788438796997,
-      "learning_rate": 6.755879396984925e-05,
-      "loss": 5.1477,
-      "step": 32790
-    },
-    {
-      "epoch": 17.100912646675358,
-      "grad_norm": 1.4335778951644897,
-      "learning_rate": 6.755778894472361e-05,
-      "loss": 5.7454,
-      "step": 32791
-    },
-    {
-      "epoch": 17.101434159061277,
-      "grad_norm": 1.8780076503753662,
-      "learning_rate": 6.755678391959799e-05,
-      "loss": 5.1329,
-      "step": 32792
-    },
-    {
-      "epoch": 17.101955671447197,
-      "grad_norm": 1.4681329727172852,
-      "learning_rate": 6.755577889447237e-05,
-      "loss": 5.4907,
-      "step": 32793
-    },
-    {
-      "epoch": 17.102477183833116,
-      "grad_norm": 1.3760312795639038,
-      "learning_rate": 6.755477386934673e-05,
-      "loss": 4.5476,
-      "step": 32794
-    },
-    {
-      "epoch": 17.102998696219036,
-      "grad_norm": 1.4816341400146484,
-      "learning_rate": 6.755376884422111e-05,
-      "loss": 5.2057,
-      "step": 32795
-    },
-    {
-      "epoch": 17.103520208604955,
-      "grad_norm": 1.4538564682006836,
-      "learning_rate": 6.755276381909548e-05,
-      "loss": 5.2927,
-      "step": 32796
-    },
-    {
-      "epoch": 17.104041720990875,
-      "grad_norm": 1.3845133781433105,
-      "learning_rate": 6.755175879396985e-05,
-      "loss": 5.7526,
-      "step": 32797
-    },
-    {
-      "epoch": 17.104563233376794,
-      "grad_norm": 1.5140972137451172,
-      "learning_rate": 6.755075376884422e-05,
-      "loss": 4.9992,
-      "step": 32798
-    },
-    {
-      "epoch": 17.105084745762714,
-      "grad_norm": 1.3504012823104858,
-      "learning_rate": 6.75497487437186e-05,
-      "loss": 5.3655,
-      "step": 32799
-    },
-    {
-      "epoch": 17.10560625814863,
-      "grad_norm": 1.520885705947876,
-      "learning_rate": 6.754874371859296e-05,
-      "loss": 5.5593,
-      "step": 32800
-    },
-    {
-      "epoch": 17.10612777053455,
-      "grad_norm": 1.5622756481170654,
-      "learning_rate": 6.754773869346734e-05,
-      "loss": 5.5413,
-      "step": 32801
-    },
-    {
-      "epoch": 17.10664928292047,
-      "grad_norm": 1.5471041202545166,
-      "learning_rate": 6.754673366834172e-05,
-      "loss": 4.9693,
-      "step": 32802
-    },
-    {
-      "epoch": 17.107170795306388,
-      "grad_norm": 1.4357013702392578,
-      "learning_rate": 6.75457286432161e-05,
-      "loss": 5.3392,
-      "step": 32803
-    },
-    {
-      "epoch": 17.107692307692307,
-      "grad_norm": 1.4703725576400757,
-      "learning_rate": 6.754472361809046e-05,
-      "loss": 4.8009,
-      "step": 32804
-    },
-    {
-      "epoch": 17.108213820078227,
-      "grad_norm": 1.4283376932144165,
-      "learning_rate": 6.754371859296482e-05,
-      "loss": 5.8543,
-      "step": 32805
-    },
-    {
-      "epoch": 17.108735332464146,
-      "grad_norm": 1.4282352924346924,
-      "learning_rate": 6.75427135678392e-05,
-      "loss": 4.7559,
-      "step": 32806
-    },
-    {
-      "epoch": 17.109256844850066,
-      "grad_norm": 1.4978197813034058,
-      "learning_rate": 6.754170854271356e-05,
-      "loss": 5.4388,
-      "step": 32807
-    },
-    {
-      "epoch": 17.109778357235985,
-      "grad_norm": 1.5037217140197754,
-      "learning_rate": 6.754070351758794e-05,
-      "loss": 4.7253,
-      "step": 32808
-    },
-    {
-      "epoch": 17.110299869621905,
-      "grad_norm": 1.6101045608520508,
-      "learning_rate": 6.753969849246231e-05,
-      "loss": 5.0356,
-      "step": 32809
-    },
-    {
-      "epoch": 17.110821382007824,
-      "grad_norm": 1.5568077564239502,
-      "learning_rate": 6.753869346733668e-05,
-      "loss": 5.1991,
-      "step": 32810
-    },
-    {
-      "epoch": 17.111342894393744,
-      "grad_norm": 1.4889644384384155,
-      "learning_rate": 6.753768844221105e-05,
-      "loss": 5.3513,
-      "step": 32811
-    },
-    {
-      "epoch": 17.11186440677966,
-      "grad_norm": 1.6457443237304688,
-      "learning_rate": 6.753668341708543e-05,
-      "loss": 5.0389,
-      "step": 32812
-    },
-    {
-      "epoch": 17.11238591916558,
-      "grad_norm": 1.5254173278808594,
-      "learning_rate": 6.75356783919598e-05,
-      "loss": 5.0333,
-      "step": 32813
-    },
-    {
-      "epoch": 17.1129074315515,
-      "grad_norm": 1.5038663148880005,
-      "learning_rate": 6.753467336683418e-05,
-      "loss": 5.3343,
-      "step": 32814
-    },
-    {
-      "epoch": 17.113428943937418,
-      "grad_norm": 1.4917720556259155,
-      "learning_rate": 6.753366834170855e-05,
-      "loss": 5.2058,
-      "step": 32815
-    },
-    {
-      "epoch": 17.113950456323337,
-      "grad_norm": 1.589188575744629,
-      "learning_rate": 6.753266331658292e-05,
-      "loss": 5.2679,
-      "step": 32816
-    },
-    {
-      "epoch": 17.114471968709257,
-      "grad_norm": 1.554328203201294,
-      "learning_rate": 6.753165829145729e-05,
-      "loss": 5.386,
-      "step": 32817
-    },
-    {
-      "epoch": 17.114993481095176,
-      "grad_norm": 1.6161556243896484,
-      "learning_rate": 6.753065326633165e-05,
-      "loss": 5.5792,
-      "step": 32818
-    },
-    {
-      "epoch": 17.115514993481096,
-      "grad_norm": 1.5397775173187256,
-      "learning_rate": 6.752964824120603e-05,
-      "loss": 5.3347,
-      "step": 32819
-    },
-    {
-      "epoch": 17.116036505867015,
-      "grad_norm": 1.511048674583435,
-      "learning_rate": 6.75286432160804e-05,
-      "loss": 5.3704,
-      "step": 32820
-    },
-    {
-      "epoch": 17.116558018252935,
-      "grad_norm": 1.4122600555419922,
-      "learning_rate": 6.752763819095477e-05,
-      "loss": 5.5296,
-      "step": 32821
-    },
-    {
-      "epoch": 17.117079530638854,
-      "grad_norm": 1.5023800134658813,
-      "learning_rate": 6.752663316582915e-05,
-      "loss": 5.1898,
-      "step": 32822
-    },
-    {
-      "epoch": 17.117601043024774,
-      "grad_norm": 1.3877084255218506,
-      "learning_rate": 6.752562814070353e-05,
-      "loss": 5.4509,
-      "step": 32823
-    },
-    {
-      "epoch": 17.11812255541069,
-      "grad_norm": 1.5519635677337646,
-      "learning_rate": 6.752462311557789e-05,
-      "loss": 5.2064,
-      "step": 32824
-    },
-    {
-      "epoch": 17.11864406779661,
-      "grad_norm": 1.5306978225708008,
-      "learning_rate": 6.752361809045227e-05,
-      "loss": 5.3802,
-      "step": 32825
-    },
-    {
-      "epoch": 17.11916558018253,
-      "grad_norm": 1.488799810409546,
-      "learning_rate": 6.752261306532664e-05,
-      "loss": 5.3087,
-      "step": 32826
-    },
-    {
-      "epoch": 17.119687092568448,
-      "grad_norm": 1.4921048879623413,
-      "learning_rate": 6.752160804020101e-05,
-      "loss": 5.3779,
-      "step": 32827
-    },
-    {
-      "epoch": 17.120208604954367,
-      "grad_norm": 1.4637751579284668,
-      "learning_rate": 6.752060301507538e-05,
-      "loss": 5.6631,
-      "step": 32828
-    },
-    {
-      "epoch": 17.120730117340287,
-      "grad_norm": 1.5233283042907715,
-      "learning_rate": 6.751959798994976e-05,
-      "loss": 5.3544,
-      "step": 32829
-    },
-    {
-      "epoch": 17.121251629726206,
-      "grad_norm": 1.4425312280654907,
-      "learning_rate": 6.751859296482412e-05,
-      "loss": 5.3871,
-      "step": 32830
-    },
-    {
-      "epoch": 17.121773142112126,
-      "grad_norm": 1.3826619386672974,
-      "learning_rate": 6.75175879396985e-05,
-      "loss": 5.7644,
-      "step": 32831
-    },
-    {
-      "epoch": 17.122294654498045,
-      "grad_norm": 1.6309010982513428,
-      "learning_rate": 6.751658291457286e-05,
-      "loss": 5.1088,
-      "step": 32832
-    },
-    {
-      "epoch": 17.122816166883965,
-      "grad_norm": 1.4904911518096924,
-      "learning_rate": 6.751557788944724e-05,
-      "loss": 4.7848,
-      "step": 32833
-    },
-    {
-      "epoch": 17.123337679269884,
-      "grad_norm": 1.5239274501800537,
-      "learning_rate": 6.751457286432162e-05,
-      "loss": 5.687,
-      "step": 32834
-    },
-    {
-      "epoch": 17.1238591916558,
-      "grad_norm": 1.4433038234710693,
-      "learning_rate": 6.751356783919598e-05,
-      "loss": 5.3633,
-      "step": 32835
-    },
-    {
-      "epoch": 17.12438070404172,
-      "grad_norm": 1.3901371955871582,
-      "learning_rate": 6.751256281407036e-05,
-      "loss": 5.4589,
-      "step": 32836
-    },
-    {
-      "epoch": 17.12490221642764,
-      "grad_norm": 1.4332612752914429,
-      "learning_rate": 6.751155778894472e-05,
-      "loss": 5.7135,
-      "step": 32837
-    },
-    {
-      "epoch": 17.12542372881356,
-      "grad_norm": 1.5117411613464355,
-      "learning_rate": 6.75105527638191e-05,
-      "loss": 5.6013,
-      "step": 32838
-    },
-    {
-      "epoch": 17.125945241199478,
-      "grad_norm": 1.41855788230896,
-      "learning_rate": 6.750954773869347e-05,
-      "loss": 4.719,
-      "step": 32839
-    },
-    {
-      "epoch": 17.126466753585397,
-      "grad_norm": 1.4722753763198853,
-      "learning_rate": 6.750854271356784e-05,
-      "loss": 5.1567,
-      "step": 32840
-    },
-    {
-      "epoch": 17.126988265971317,
-      "grad_norm": 1.4507497549057007,
-      "learning_rate": 6.750753768844221e-05,
-      "loss": 4.9548,
-      "step": 32841
-    },
-    {
-      "epoch": 17.127509778357236,
-      "grad_norm": 1.5878984928131104,
-      "learning_rate": 6.750653266331659e-05,
-      "loss": 5.1994,
-      "step": 32842
-    },
-    {
-      "epoch": 17.128031290743156,
-      "grad_norm": 1.3931608200073242,
-      "learning_rate": 6.750552763819096e-05,
-      "loss": 5.4339,
-      "step": 32843
-    },
-    {
-      "epoch": 17.128552803129075,
-      "grad_norm": 1.6434316635131836,
-      "learning_rate": 6.750452261306534e-05,
-      "loss": 4.5363,
-      "step": 32844
-    },
-    {
-      "epoch": 17.129074315514995,
-      "grad_norm": 1.5650259256362915,
-      "learning_rate": 6.75035175879397e-05,
-      "loss": 5.3401,
-      "step": 32845
-    },
-    {
-      "epoch": 17.129595827900914,
-      "grad_norm": 1.3892371654510498,
-      "learning_rate": 6.750251256281407e-05,
-      "loss": 5.7455,
-      "step": 32846
-    },
-    {
-      "epoch": 17.13011734028683,
-      "grad_norm": 1.6764248609542847,
-      "learning_rate": 6.750150753768845e-05,
-      "loss": 4.9104,
-      "step": 32847
-    },
-    {
-      "epoch": 17.13063885267275,
-      "grad_norm": 1.4890100955963135,
-      "learning_rate": 6.750050251256281e-05,
-      "loss": 4.8331,
-      "step": 32848
-    },
-    {
-      "epoch": 17.13116036505867,
-      "grad_norm": 1.4367995262145996,
-      "learning_rate": 6.749949748743719e-05,
-      "loss": 5.5359,
-      "step": 32849
-    },
-    {
-      "epoch": 17.13168187744459,
-      "grad_norm": 1.5571117401123047,
-      "learning_rate": 6.749849246231155e-05,
-      "loss": 5.4754,
-      "step": 32850
-    },
-    {
-      "epoch": 17.132203389830508,
-      "grad_norm": 1.4580280780792236,
-      "learning_rate": 6.749748743718593e-05,
-      "loss": 5.6168,
-      "step": 32851
-    },
-    {
-      "epoch": 17.132724902216427,
-      "grad_norm": 1.4597853422164917,
-      "learning_rate": 6.74964824120603e-05,
-      "loss": 5.5372,
-      "step": 32852
-    },
-    {
-      "epoch": 17.133246414602347,
-      "grad_norm": 2.064976215362549,
-      "learning_rate": 6.749547738693467e-05,
-      "loss": 5.1256,
-      "step": 32853
-    },
-    {
-      "epoch": 17.133767926988266,
-      "grad_norm": 1.5614848136901855,
-      "learning_rate": 6.749447236180905e-05,
-      "loss": 5.2288,
-      "step": 32854
-    },
-    {
-      "epoch": 17.134289439374186,
-      "grad_norm": 1.5833646059036255,
-      "learning_rate": 6.749346733668343e-05,
-      "loss": 5.1776,
-      "step": 32855
-    },
-    {
-      "epoch": 17.134810951760105,
-      "grad_norm": 1.4972188472747803,
-      "learning_rate": 6.74924623115578e-05,
-      "loss": 5.3338,
-      "step": 32856
-    },
-    {
-      "epoch": 17.135332464146025,
-      "grad_norm": 1.5132651329040527,
-      "learning_rate": 6.749145728643217e-05,
-      "loss": 5.3822,
-      "step": 32857
-    },
-    {
-      "epoch": 17.135853976531944,
-      "grad_norm": 1.5867760181427002,
-      "learning_rate": 6.749045226130654e-05,
-      "loss": 4.8915,
-      "step": 32858
-    },
-    {
-      "epoch": 17.13637548891786,
-      "grad_norm": 1.5057703256607056,
-      "learning_rate": 6.74894472361809e-05,
-      "loss": 5.3337,
-      "step": 32859
-    },
-    {
-      "epoch": 17.13689700130378,
-      "grad_norm": 1.6862560510635376,
-      "learning_rate": 6.748844221105528e-05,
-      "loss": 4.5638,
-      "step": 32860
-    },
-    {
-      "epoch": 17.1374185136897,
-      "grad_norm": 1.4507765769958496,
-      "learning_rate": 6.748743718592964e-05,
-      "loss": 5.3378,
-      "step": 32861
-    },
-    {
-      "epoch": 17.13794002607562,
-      "grad_norm": 1.4768725633621216,
-      "learning_rate": 6.748643216080402e-05,
-      "loss": 5.6065,
-      "step": 32862
-    },
-    {
-      "epoch": 17.138461538461538,
-      "grad_norm": 1.4905898571014404,
-      "learning_rate": 6.74854271356784e-05,
-      "loss": 5.334,
-      "step": 32863
-    },
-    {
-      "epoch": 17.138983050847457,
-      "grad_norm": 1.457514762878418,
-      "learning_rate": 6.748442211055278e-05,
-      "loss": 5.5363,
-      "step": 32864
-    },
-    {
-      "epoch": 17.139504563233377,
-      "grad_norm": 1.4460097551345825,
-      "learning_rate": 6.748341708542714e-05,
-      "loss": 5.0871,
-      "step": 32865
-    },
-    {
-      "epoch": 17.140026075619296,
-      "grad_norm": 1.582625389099121,
-      "learning_rate": 6.748241206030152e-05,
-      "loss": 5.2255,
-      "step": 32866
-    },
-    {
-      "epoch": 17.140547588005216,
-      "grad_norm": 1.665555477142334,
-      "learning_rate": 6.748140703517588e-05,
-      "loss": 5.021,
-      "step": 32867
-    },
-    {
-      "epoch": 17.141069100391135,
-      "grad_norm": 1.5599218606948853,
-      "learning_rate": 6.748040201005026e-05,
-      "loss": 5.4604,
-      "step": 32868
-    },
-    {
-      "epoch": 17.141590612777055,
-      "grad_norm": 1.4208279848098755,
-      "learning_rate": 6.747939698492462e-05,
-      "loss": 5.7108,
-      "step": 32869
-    },
-    {
-      "epoch": 17.142112125162974,
-      "grad_norm": 1.4826055765151978,
-      "learning_rate": 6.7478391959799e-05,
-      "loss": 5.5218,
-      "step": 32870
-    },
-    {
-      "epoch": 17.14263363754889,
-      "grad_norm": 1.455627202987671,
-      "learning_rate": 6.747738693467337e-05,
-      "loss": 5.7385,
-      "step": 32871
-    },
-    {
-      "epoch": 17.14315514993481,
-      "grad_norm": 1.4555275440216064,
-      "learning_rate": 6.747638190954773e-05,
-      "loss": 5.4537,
-      "step": 32872
-    },
-    {
-      "epoch": 17.14367666232073,
-      "grad_norm": 1.5068496465682983,
-      "learning_rate": 6.747537688442211e-05,
-      "loss": 5.5484,
-      "step": 32873
-    },
-    {
-      "epoch": 17.14419817470665,
-      "grad_norm": 1.4454697370529175,
-      "learning_rate": 6.747437185929649e-05,
-      "loss": 5.4672,
-      "step": 32874
-    },
-    {
-      "epoch": 17.144719687092568,
-      "grad_norm": 1.4496955871582031,
-      "learning_rate": 6.747336683417086e-05,
-      "loss": 5.1031,
-      "step": 32875
-    },
-    {
-      "epoch": 17.145241199478487,
-      "grad_norm": 1.554413914680481,
-      "learning_rate": 6.747236180904523e-05,
-      "loss": 5.3543,
-      "step": 32876
-    },
-    {
-      "epoch": 17.145762711864407,
-      "grad_norm": 1.6058363914489746,
-      "learning_rate": 6.74713567839196e-05,
-      "loss": 4.9776,
-      "step": 32877
-    },
-    {
-      "epoch": 17.146284224250326,
-      "grad_norm": 1.410929560661316,
-      "learning_rate": 6.747035175879397e-05,
-      "loss": 5.3268,
-      "step": 32878
-    },
-    {
-      "epoch": 17.146805736636246,
-      "grad_norm": 1.3689547777175903,
-      "learning_rate": 6.746934673366835e-05,
-      "loss": 5.3072,
-      "step": 32879
-    },
-    {
-      "epoch": 17.147327249022165,
-      "grad_norm": 1.4115556478500366,
-      "learning_rate": 6.746834170854271e-05,
-      "loss": 4.9705,
-      "step": 32880
-    },
-    {
-      "epoch": 17.147848761408085,
-      "grad_norm": 1.4599069356918335,
-      "learning_rate": 6.746733668341709e-05,
-      "loss": 5.3587,
-      "step": 32881
-    },
-    {
-      "epoch": 17.148370273794004,
-      "grad_norm": 1.469376802444458,
-      "learning_rate": 6.746633165829145e-05,
-      "loss": 5.4348,
-      "step": 32882
-    },
-    {
-      "epoch": 17.14889178617992,
-      "grad_norm": 1.4895715713500977,
-      "learning_rate": 6.746532663316583e-05,
-      "loss": 5.2161,
-      "step": 32883
-    },
-    {
-      "epoch": 17.14941329856584,
-      "grad_norm": 1.5645273923873901,
-      "learning_rate": 6.746432160804021e-05,
-      "loss": 5.542,
-      "step": 32884
-    },
-    {
-      "epoch": 17.14993481095176,
-      "grad_norm": 1.512799620628357,
-      "learning_rate": 6.746331658291457e-05,
-      "loss": 5.4569,
-      "step": 32885
-    },
-    {
-      "epoch": 17.15045632333768,
-      "grad_norm": 1.4767218828201294,
-      "learning_rate": 6.746231155778895e-05,
-      "loss": 5.4797,
-      "step": 32886
-    },
-    {
-      "epoch": 17.150977835723598,
-      "grad_norm": 1.4763649702072144,
-      "learning_rate": 6.746130653266332e-05,
-      "loss": 4.9479,
-      "step": 32887
-    },
-    {
-      "epoch": 17.151499348109517,
-      "grad_norm": 1.5591144561767578,
-      "learning_rate": 6.74603015075377e-05,
-      "loss": 5.3848,
-      "step": 32888
-    },
-    {
-      "epoch": 17.152020860495437,
-      "grad_norm": 1.5869780778884888,
-      "learning_rate": 6.745929648241206e-05,
-      "loss": 4.7545,
-      "step": 32889
-    },
-    {
-      "epoch": 17.152542372881356,
-      "grad_norm": 1.6871610879898071,
-      "learning_rate": 6.745829145728644e-05,
-      "loss": 4.9892,
-      "step": 32890
-    },
-    {
-      "epoch": 17.153063885267276,
-      "grad_norm": 1.5329129695892334,
-      "learning_rate": 6.74572864321608e-05,
-      "loss": 5.2002,
-      "step": 32891
-    },
-    {
-      "epoch": 17.153585397653195,
-      "grad_norm": 1.5414183139801025,
-      "learning_rate": 6.745628140703518e-05,
-      "loss": 5.1994,
-      "step": 32892
-    },
-    {
-      "epoch": 17.154106910039115,
-      "grad_norm": 1.7387856245040894,
-      "learning_rate": 6.745527638190954e-05,
-      "loss": 4.8237,
-      "step": 32893
-    },
-    {
-      "epoch": 17.154628422425034,
-      "grad_norm": 1.6126216650009155,
-      "learning_rate": 6.745427135678392e-05,
-      "loss": 5.2741,
-      "step": 32894
-    },
-    {
-      "epoch": 17.15514993481095,
-      "grad_norm": 1.4583334922790527,
-      "learning_rate": 6.74532663316583e-05,
-      "loss": 5.3261,
-      "step": 32895
-    },
-    {
-      "epoch": 17.15567144719687,
-      "grad_norm": 1.3954870700836182,
-      "learning_rate": 6.745226130653268e-05,
-      "loss": 5.0533,
-      "step": 32896
-    },
-    {
-      "epoch": 17.15619295958279,
-      "grad_norm": 1.517376184463501,
-      "learning_rate": 6.745125628140704e-05,
-      "loss": 5.6509,
-      "step": 32897
-    },
-    {
-      "epoch": 17.15671447196871,
-      "grad_norm": 1.4280763864517212,
-      "learning_rate": 6.74502512562814e-05,
-      "loss": 5.1056,
-      "step": 32898
-    },
-    {
-      "epoch": 17.157235984354628,
-      "grad_norm": 1.416228175163269,
-      "learning_rate": 6.744924623115578e-05,
-      "loss": 5.1582,
-      "step": 32899
-    },
-    {
-      "epoch": 17.157757496740548,
-      "grad_norm": 1.4682676792144775,
-      "learning_rate": 6.744824120603015e-05,
-      "loss": 5.5805,
-      "step": 32900
-    },
-    {
-      "epoch": 17.158279009126467,
-      "grad_norm": 1.4207981824874878,
-      "learning_rate": 6.744723618090453e-05,
-      "loss": 5.2059,
-      "step": 32901
-    },
-    {
-      "epoch": 17.158800521512386,
-      "grad_norm": 1.447361946105957,
-      "learning_rate": 6.744623115577889e-05,
-      "loss": 5.0032,
-      "step": 32902
-    },
-    {
-      "epoch": 17.159322033898306,
-      "grad_norm": 1.5244979858398438,
-      "learning_rate": 6.744522613065327e-05,
-      "loss": 4.8954,
-      "step": 32903
-    },
-    {
-      "epoch": 17.159843546284225,
-      "grad_norm": 1.5691324472427368,
-      "learning_rate": 6.744422110552765e-05,
-      "loss": 5.4251,
-      "step": 32904
-    },
-    {
-      "epoch": 17.160365058670145,
-      "grad_norm": 1.5234642028808594,
-      "learning_rate": 6.744321608040202e-05,
-      "loss": 5.1315,
-      "step": 32905
-    },
-    {
-      "epoch": 17.160886571056064,
-      "grad_norm": 1.5097606182098389,
-      "learning_rate": 6.744221105527639e-05,
-      "loss": 5.4229,
-      "step": 32906
-    },
-    {
-      "epoch": 17.16140808344198,
-      "grad_norm": 1.416056513786316,
-      "learning_rate": 6.744120603015076e-05,
-      "loss": 4.92,
-      "step": 32907
-    },
-    {
-      "epoch": 17.1619295958279,
-      "grad_norm": 1.5040582418441772,
-      "learning_rate": 6.744020100502513e-05,
-      "loss": 5.1045,
-      "step": 32908
-    },
-    {
-      "epoch": 17.16245110821382,
-      "grad_norm": 1.5126540660858154,
-      "learning_rate": 6.743919597989951e-05,
-      "loss": 5.5679,
-      "step": 32909
-    },
-    {
-      "epoch": 17.16297262059974,
-      "grad_norm": 1.5757725238800049,
-      "learning_rate": 6.743819095477387e-05,
-      "loss": 5.4934,
-      "step": 32910
-    },
-    {
-      "epoch": 17.163494132985658,
-      "grad_norm": 1.4590747356414795,
-      "learning_rate": 6.743718592964824e-05,
-      "loss": 5.0341,
-      "step": 32911
-    },
-    {
-      "epoch": 17.164015645371578,
-      "grad_norm": 1.4553097486495972,
-      "learning_rate": 6.743618090452261e-05,
-      "loss": 5.7921,
-      "step": 32912
-    },
-    {
-      "epoch": 17.164537157757497,
-      "grad_norm": 1.4581011533737183,
-      "learning_rate": 6.743517587939698e-05,
-      "loss": 5.0727,
-      "step": 32913
-    },
-    {
-      "epoch": 17.165058670143416,
-      "grad_norm": 1.6015175580978394,
-      "learning_rate": 6.743417085427136e-05,
-      "loss": 5.4312,
-      "step": 32914
-    },
-    {
-      "epoch": 17.165580182529336,
-      "grad_norm": 1.557355523109436,
-      "learning_rate": 6.743316582914573e-05,
-      "loss": 5.0298,
-      "step": 32915
-    },
-    {
-      "epoch": 17.166101694915255,
-      "grad_norm": 1.591027855873108,
-      "learning_rate": 6.743216080402011e-05,
-      "loss": 5.5812,
-      "step": 32916
-    },
-    {
-      "epoch": 17.166623207301175,
-      "grad_norm": 1.5983645915985107,
-      "learning_rate": 6.743115577889448e-05,
-      "loss": 5.1457,
-      "step": 32917
-    },
-    {
-      "epoch": 17.167144719687094,
-      "grad_norm": 1.4609453678131104,
-      "learning_rate": 6.743015075376885e-05,
-      "loss": 5.645,
-      "step": 32918
-    },
-    {
-      "epoch": 17.16766623207301,
-      "grad_norm": 1.4419039487838745,
-      "learning_rate": 6.742914572864322e-05,
-      "loss": 4.784,
-      "step": 32919
-    },
-    {
-      "epoch": 17.16818774445893,
-      "grad_norm": 1.4688886404037476,
-      "learning_rate": 6.74281407035176e-05,
-      "loss": 5.4446,
-      "step": 32920
-    },
-    {
-      "epoch": 17.16870925684485,
-      "grad_norm": 1.4568467140197754,
-      "learning_rate": 6.742713567839196e-05,
-      "loss": 4.9233,
-      "step": 32921
-    },
-    {
-      "epoch": 17.16923076923077,
-      "grad_norm": 1.619380235671997,
-      "learning_rate": 6.742613065326634e-05,
-      "loss": 4.872,
-      "step": 32922
-    },
-    {
-      "epoch": 17.169752281616688,
-      "grad_norm": 1.3487818241119385,
-      "learning_rate": 6.74251256281407e-05,
-      "loss": 5.4562,
-      "step": 32923
-    },
-    {
-      "epoch": 17.170273794002608,
-      "grad_norm": 1.5321142673492432,
-      "learning_rate": 6.742412060301508e-05,
-      "loss": 5.0061,
-      "step": 32924
-    },
-    {
-      "epoch": 17.170795306388527,
-      "grad_norm": 1.3323802947998047,
-      "learning_rate": 6.742311557788946e-05,
-      "loss": 5.4478,
-      "step": 32925
-    },
-    {
-      "epoch": 17.171316818774446,
-      "grad_norm": 1.4525063037872314,
-      "learning_rate": 6.742211055276382e-05,
-      "loss": 5.3856,
-      "step": 32926
-    },
-    {
-      "epoch": 17.171838331160366,
-      "grad_norm": 1.4502235651016235,
-      "learning_rate": 6.74211055276382e-05,
-      "loss": 5.6462,
-      "step": 32927
-    },
-    {
-      "epoch": 17.172359843546285,
-      "grad_norm": 1.394148349761963,
-      "learning_rate": 6.742010050251256e-05,
-      "loss": 5.2367,
-      "step": 32928
-    },
-    {
-      "epoch": 17.172881355932205,
-      "grad_norm": 1.4198122024536133,
-      "learning_rate": 6.741909547738694e-05,
-      "loss": 5.3237,
-      "step": 32929
-    },
-    {
-      "epoch": 17.17340286831812,
-      "grad_norm": 1.5226898193359375,
-      "learning_rate": 6.74180904522613e-05,
-      "loss": 5.5998,
-      "step": 32930
-    },
-    {
-      "epoch": 17.17392438070404,
-      "grad_norm": 1.5790024995803833,
-      "learning_rate": 6.741708542713568e-05,
-      "loss": 4.9491,
-      "step": 32931
-    },
-    {
-      "epoch": 17.17444589308996,
-      "grad_norm": 1.542162299156189,
-      "learning_rate": 6.741608040201005e-05,
-      "loss": 5.06,
-      "step": 32932
-    },
-    {
-      "epoch": 17.17496740547588,
-      "grad_norm": 1.5510246753692627,
-      "learning_rate": 6.741507537688443e-05,
-      "loss": 5.2562,
-      "step": 32933
-    },
-    {
-      "epoch": 17.1754889178618,
-      "grad_norm": 1.4274033308029175,
-      "learning_rate": 6.741407035175879e-05,
-      "loss": 5.2639,
-      "step": 32934
-    },
-    {
-      "epoch": 17.176010430247718,
-      "grad_norm": 1.4763379096984863,
-      "learning_rate": 6.741306532663317e-05,
-      "loss": 5.0229,
-      "step": 32935
-    },
-    {
-      "epoch": 17.176531942633638,
-      "grad_norm": 1.4923748970031738,
-      "learning_rate": 6.741206030150755e-05,
-      "loss": 5.4986,
-      "step": 32936
-    },
-    {
-      "epoch": 17.177053455019557,
-      "grad_norm": 1.3514550924301147,
-      "learning_rate": 6.741105527638192e-05,
-      "loss": 5.0466,
-      "step": 32937
-    },
-    {
-      "epoch": 17.177574967405477,
-      "grad_norm": 1.5190341472625732,
-      "learning_rate": 6.741005025125629e-05,
-      "loss": 5.0656,
-      "step": 32938
-    },
-    {
-      "epoch": 17.178096479791396,
-      "grad_norm": 1.5011537075042725,
-      "learning_rate": 6.740904522613065e-05,
-      "loss": 5.4953,
-      "step": 32939
-    },
-    {
-      "epoch": 17.178617992177315,
-      "grad_norm": 1.4527158737182617,
-      "learning_rate": 6.740804020100503e-05,
-      "loss": 5.4899,
-      "step": 32940
-    },
-    {
-      "epoch": 17.179139504563235,
-      "grad_norm": 1.5406252145767212,
-      "learning_rate": 6.74070351758794e-05,
-      "loss": 4.5985,
-      "step": 32941
-    },
-    {
-      "epoch": 17.17966101694915,
-      "grad_norm": 1.5671554803848267,
-      "learning_rate": 6.740603015075377e-05,
-      "loss": 5.4745,
-      "step": 32942
-    },
-    {
-      "epoch": 17.18018252933507,
-      "grad_norm": 1.600188136100769,
-      "learning_rate": 6.740502512562814e-05,
-      "loss": 5.2246,
-      "step": 32943
-    },
-    {
-      "epoch": 17.18070404172099,
-      "grad_norm": 1.4949333667755127,
-      "learning_rate": 6.740402010050251e-05,
-      "loss": 5.3168,
-      "step": 32944
-    },
-    {
-      "epoch": 17.18122555410691,
-      "grad_norm": 1.397896647453308,
-      "learning_rate": 6.740301507537688e-05,
-      "loss": 5.4146,
-      "step": 32945
-    },
-    {
-      "epoch": 17.18174706649283,
-      "grad_norm": 1.5179786682128906,
-      "learning_rate": 6.740201005025126e-05,
-      "loss": 4.9287,
-      "step": 32946
-    },
-    {
-      "epoch": 17.182268578878748,
-      "grad_norm": 1.389293909072876,
-      "learning_rate": 6.740100502512563e-05,
-      "loss": 5.5433,
-      "step": 32947
-    },
-    {
-      "epoch": 17.182790091264668,
-      "grad_norm": 1.5862549543380737,
-      "learning_rate": 6.740000000000001e-05,
-      "loss": 4.6919,
-      "step": 32948
-    },
-    {
-      "epoch": 17.183311603650587,
-      "grad_norm": 1.495482325553894,
-      "learning_rate": 6.739899497487438e-05,
-      "loss": 5.1954,
-      "step": 32949
-    },
-    {
-      "epoch": 17.183833116036507,
-      "grad_norm": 1.6546146869659424,
-      "learning_rate": 6.739798994974875e-05,
-      "loss": 5.1123,
-      "step": 32950
-    },
-    {
-      "epoch": 17.184354628422426,
-      "grad_norm": 1.4995152950286865,
-      "learning_rate": 6.739698492462312e-05,
-      "loss": 5.3682,
-      "step": 32951
-    },
-    {
-      "epoch": 17.184876140808345,
-      "grad_norm": 1.443284034729004,
-      "learning_rate": 6.739597989949748e-05,
-      "loss": 5.3948,
-      "step": 32952
-    },
-    {
-      "epoch": 17.185397653194265,
-      "grad_norm": 1.4093999862670898,
-      "learning_rate": 6.739497487437186e-05,
-      "loss": 5.6315,
-      "step": 32953
-    },
-    {
-      "epoch": 17.18591916558018,
-      "grad_norm": 1.6893444061279297,
-      "learning_rate": 6.739396984924622e-05,
-      "loss": 5.0516,
-      "step": 32954
-    },
-    {
-      "epoch": 17.1864406779661,
-      "grad_norm": 1.5030590295791626,
-      "learning_rate": 6.73929648241206e-05,
-      "loss": 4.9421,
-      "step": 32955
-    },
-    {
-      "epoch": 17.18696219035202,
-      "grad_norm": 1.4450631141662598,
-      "learning_rate": 6.739195979899498e-05,
-      "loss": 5.6119,
-      "step": 32956
-    },
-    {
-      "epoch": 17.18748370273794,
-      "grad_norm": 1.5362632274627686,
-      "learning_rate": 6.739095477386936e-05,
-      "loss": 5.0703,
-      "step": 32957
-    },
-    {
-      "epoch": 17.18800521512386,
-      "grad_norm": 1.5217534303665161,
-      "learning_rate": 6.738994974874372e-05,
-      "loss": 5.2104,
-      "step": 32958
-    },
-    {
-      "epoch": 17.188526727509778,
-      "grad_norm": 1.4878727197647095,
-      "learning_rate": 6.73889447236181e-05,
-      "loss": 5.0472,
-      "step": 32959
-    },
-    {
-      "epoch": 17.189048239895698,
-      "grad_norm": 1.429185390472412,
-      "learning_rate": 6.738793969849246e-05,
-      "loss": 5.3472,
-      "step": 32960
-    },
-    {
-      "epoch": 17.189569752281617,
-      "grad_norm": 1.5073646306991577,
-      "learning_rate": 6.738693467336684e-05,
-      "loss": 5.4654,
-      "step": 32961
-    },
-    {
-      "epoch": 17.190091264667537,
-      "grad_norm": 1.4468257427215576,
-      "learning_rate": 6.73859296482412e-05,
-      "loss": 5.1945,
-      "step": 32962
-    },
-    {
-      "epoch": 17.190612777053456,
-      "grad_norm": 1.4590836763381958,
-      "learning_rate": 6.738492462311558e-05,
-      "loss": 5.5116,
-      "step": 32963
-    },
-    {
-      "epoch": 17.191134289439375,
-      "grad_norm": 1.462514877319336,
-      "learning_rate": 6.738391959798995e-05,
-      "loss": 5.4036,
-      "step": 32964
-    },
-    {
-      "epoch": 17.191655801825295,
-      "grad_norm": 1.539069652557373,
-      "learning_rate": 6.738291457286431e-05,
-      "loss": 5.2431,
-      "step": 32965
-    },
-    {
-      "epoch": 17.19217731421121,
-      "grad_norm": 1.5758827924728394,
-      "learning_rate": 6.738190954773869e-05,
-      "loss": 5.0424,
-      "step": 32966
-    },
-    {
-      "epoch": 17.19269882659713,
-      "grad_norm": 1.5653212070465088,
-      "learning_rate": 6.738090452261307e-05,
-      "loss": 5.1583,
-      "step": 32967
-    },
-    {
-      "epoch": 17.19322033898305,
-      "grad_norm": 1.4738274812698364,
-      "learning_rate": 6.737989949748745e-05,
-      "loss": 5.0173,
-      "step": 32968
-    },
-    {
-      "epoch": 17.19374185136897,
-      "grad_norm": 1.5968079566955566,
-      "learning_rate": 6.737889447236181e-05,
-      "loss": 5.4328,
-      "step": 32969
-    },
-    {
-      "epoch": 17.19426336375489,
-      "grad_norm": 1.5280812978744507,
-      "learning_rate": 6.737788944723619e-05,
-      "loss": 5.6135,
-      "step": 32970
-    },
-    {
-      "epoch": 17.194784876140808,
-      "grad_norm": 1.4790552854537964,
-      "learning_rate": 6.737688442211055e-05,
-      "loss": 5.4406,
-      "step": 32971
-    },
-    {
-      "epoch": 17.195306388526728,
-      "grad_norm": 1.4183121919631958,
-      "learning_rate": 6.737587939698493e-05,
-      "loss": 5.9783,
-      "step": 32972
-    },
-    {
-      "epoch": 17.195827900912647,
-      "grad_norm": 1.614562749862671,
-      "learning_rate": 6.73748743718593e-05,
-      "loss": 4.8376,
-      "step": 32973
-    },
-    {
-      "epoch": 17.196349413298567,
-      "grad_norm": 1.410207986831665,
-      "learning_rate": 6.737386934673367e-05,
-      "loss": 5.3791,
-      "step": 32974
-    },
-    {
-      "epoch": 17.196870925684486,
-      "grad_norm": 1.4985867738723755,
-      "learning_rate": 6.737286432160804e-05,
-      "loss": 5.522,
-      "step": 32975
-    },
-    {
-      "epoch": 17.197392438070406,
-      "grad_norm": 1.4103248119354248,
-      "learning_rate": 6.737185929648241e-05,
-      "loss": 5.0306,
-      "step": 32976
-    },
-    {
-      "epoch": 17.197913950456325,
-      "grad_norm": 1.516087532043457,
-      "learning_rate": 6.737085427135679e-05,
-      "loss": 5.0275,
-      "step": 32977
-    },
-    {
-      "epoch": 17.19843546284224,
-      "grad_norm": 1.4929207563400269,
-      "learning_rate": 6.736984924623116e-05,
-      "loss": 5.1097,
-      "step": 32978
-    },
-    {
-      "epoch": 17.19895697522816,
-      "grad_norm": 1.4872592687606812,
-      "learning_rate": 6.736884422110553e-05,
-      "loss": 5.5908,
-      "step": 32979
-    },
-    {
-      "epoch": 17.19947848761408,
-      "grad_norm": 1.6106736660003662,
-      "learning_rate": 6.73678391959799e-05,
-      "loss": 5.2366,
-      "step": 32980
-    },
-    {
-      "epoch": 17.2,
-      "grad_norm": 1.5341112613677979,
-      "learning_rate": 6.736683417085428e-05,
-      "loss": 5.2078,
-      "step": 32981
-    },
-    {
-      "epoch": 17.20052151238592,
-      "grad_norm": 1.4327343702316284,
-      "learning_rate": 6.736582914572864e-05,
-      "loss": 5.4407,
-      "step": 32982
-    },
-    {
-      "epoch": 17.201043024771838,
-      "grad_norm": 1.5070419311523438,
-      "learning_rate": 6.736482412060302e-05,
-      "loss": 5.5602,
-      "step": 32983
-    },
-    {
-      "epoch": 17.201564537157758,
-      "grad_norm": 1.4391369819641113,
-      "learning_rate": 6.736381909547738e-05,
-      "loss": 5.6127,
-      "step": 32984
-    },
-    {
-      "epoch": 17.202086049543677,
-      "grad_norm": 1.6736966371536255,
-      "learning_rate": 6.736281407035176e-05,
-      "loss": 5.1789,
-      "step": 32985
-    },
-    {
-      "epoch": 17.202607561929597,
-      "grad_norm": 1.5302575826644897,
-      "learning_rate": 6.736180904522613e-05,
-      "loss": 5.552,
-      "step": 32986
-    },
-    {
-      "epoch": 17.203129074315516,
-      "grad_norm": 1.4787553548812866,
-      "learning_rate": 6.73608040201005e-05,
-      "loss": 5.3203,
-      "step": 32987
-    },
-    {
-      "epoch": 17.203650586701436,
-      "grad_norm": 1.4686959981918335,
-      "learning_rate": 6.735979899497488e-05,
-      "loss": 5.6412,
-      "step": 32988
-    },
-    {
-      "epoch": 17.204172099087355,
-      "grad_norm": 1.432977318763733,
-      "learning_rate": 6.735879396984926e-05,
-      "loss": 5.3903,
-      "step": 32989
-    },
-    {
-      "epoch": 17.20469361147327,
-      "grad_norm": 1.6083422899246216,
-      "learning_rate": 6.735778894472362e-05,
-      "loss": 4.789,
-      "step": 32990
-    },
-    {
-      "epoch": 17.20521512385919,
-      "grad_norm": 1.5510125160217285,
-      "learning_rate": 6.735678391959799e-05,
-      "loss": 5.6311,
-      "step": 32991
-    },
-    {
-      "epoch": 17.20573663624511,
-      "grad_norm": 1.5995690822601318,
-      "learning_rate": 6.735577889447237e-05,
-      "loss": 4.8253,
-      "step": 32992
-    },
-    {
-      "epoch": 17.20625814863103,
-      "grad_norm": 1.4871851205825806,
-      "learning_rate": 6.735477386934673e-05,
-      "loss": 5.1167,
-      "step": 32993
-    },
-    {
-      "epoch": 17.20677966101695,
-      "grad_norm": 1.4823243618011475,
-      "learning_rate": 6.735376884422111e-05,
-      "loss": 5.3402,
-      "step": 32994
-    },
-    {
-      "epoch": 17.20730117340287,
-      "grad_norm": 1.4872583150863647,
-      "learning_rate": 6.735276381909547e-05,
-      "loss": 5.5683,
-      "step": 32995
-    },
-    {
-      "epoch": 17.207822685788788,
-      "grad_norm": 1.4722498655319214,
-      "learning_rate": 6.735175879396985e-05,
-      "loss": 5.4301,
-      "step": 32996
-    },
-    {
-      "epoch": 17.208344198174707,
-      "grad_norm": 1.5104588270187378,
-      "learning_rate": 6.735075376884423e-05,
-      "loss": 5.2326,
-      "step": 32997
-    },
-    {
-      "epoch": 17.208865710560627,
-      "grad_norm": 1.416021466255188,
-      "learning_rate": 6.73497487437186e-05,
-      "loss": 5.1825,
-      "step": 32998
-    },
-    {
-      "epoch": 17.209387222946546,
-      "grad_norm": 1.6700868606567383,
-      "learning_rate": 6.734874371859297e-05,
-      "loss": 5.3422,
-      "step": 32999
-    },
-    {
-      "epoch": 17.209908735332466,
-      "grad_norm": 1.5616004467010498,
-      "learning_rate": 6.734773869346735e-05,
-      "loss": 5.4967,
-      "step": 33000
-    },
-    {
-      "epoch": 17.210430247718385,
-      "grad_norm": 1.460205078125,
-      "learning_rate": 6.734673366834171e-05,
-      "loss": 5.2729,
-      "step": 33001
-    },
-    {
-      "epoch": 17.2109517601043,
-      "grad_norm": 1.4771742820739746,
-      "learning_rate": 6.734572864321609e-05,
-      "loss": 5.2296,
-      "step": 33002
-    },
-    {
-      "epoch": 17.21147327249022,
-      "grad_norm": 1.4089467525482178,
-      "learning_rate": 6.734472361809045e-05,
-      "loss": 5.6486,
-      "step": 33003
-    },
-    {
-      "epoch": 17.21199478487614,
-      "grad_norm": 1.3384062051773071,
-      "learning_rate": 6.734371859296483e-05,
-      "loss": 5.7257,
-      "step": 33004
-    },
-    {
-      "epoch": 17.21251629726206,
-      "grad_norm": 1.4010330438613892,
-      "learning_rate": 6.73427135678392e-05,
-      "loss": 5.7266,
-      "step": 33005
-    },
-    {
-      "epoch": 17.21303780964798,
-      "grad_norm": 1.5016255378723145,
-      "learning_rate": 6.734170854271356e-05,
-      "loss": 5.3597,
-      "step": 33006
-    },
-    {
-      "epoch": 17.2135593220339,
-      "grad_norm": 1.600670576095581,
-      "learning_rate": 6.734070351758794e-05,
-      "loss": 5.0756,
-      "step": 33007
-    },
-    {
-      "epoch": 17.214080834419818,
-      "grad_norm": 1.3905823230743408,
-      "learning_rate": 6.733969849246232e-05,
-      "loss": 5.815,
-      "step": 33008
-    },
-    {
-      "epoch": 17.214602346805737,
-      "grad_norm": 1.4914847612380981,
-      "learning_rate": 6.73386934673367e-05,
-      "loss": 5.3183,
-      "step": 33009
-    },
-    {
-      "epoch": 17.215123859191657,
-      "grad_norm": 1.8379583358764648,
-      "learning_rate": 6.733768844221106e-05,
-      "loss": 4.621,
-      "step": 33010
-    },
-    {
-      "epoch": 17.215645371577576,
-      "grad_norm": 1.593700885772705,
-      "learning_rate": 6.733668341708544e-05,
-      "loss": 5.4909,
-      "step": 33011
-    },
-    {
-      "epoch": 17.216166883963496,
-      "grad_norm": 1.599282145500183,
-      "learning_rate": 6.73356783919598e-05,
-      "loss": 5.3386,
-      "step": 33012
-    },
-    {
-      "epoch": 17.216688396349415,
-      "grad_norm": 1.4589089155197144,
-      "learning_rate": 6.733467336683418e-05,
-      "loss": 5.6896,
-      "step": 33013
-    },
-    {
-      "epoch": 17.21720990873533,
-      "grad_norm": 1.5570796728134155,
-      "learning_rate": 6.733366834170854e-05,
-      "loss": 5.2422,
-      "step": 33014
-    },
-    {
-      "epoch": 17.21773142112125,
-      "grad_norm": 1.4735710620880127,
-      "learning_rate": 6.733266331658292e-05,
-      "loss": 5.0286,
-      "step": 33015
-    },
-    {
-      "epoch": 17.21825293350717,
-      "grad_norm": 1.5435670614242554,
-      "learning_rate": 6.733165829145728e-05,
-      "loss": 5.2346,
-      "step": 33016
-    },
-    {
-      "epoch": 17.21877444589309,
-      "grad_norm": 1.4969120025634766,
-      "learning_rate": 6.733065326633166e-05,
-      "loss": 5.2332,
-      "step": 33017
-    },
-    {
-      "epoch": 17.21929595827901,
-      "grad_norm": 1.5742106437683105,
-      "learning_rate": 6.732964824120604e-05,
-      "loss": 5.465,
-      "step": 33018
-    },
-    {
-      "epoch": 17.21981747066493,
-      "grad_norm": 1.5299242734909058,
-      "learning_rate": 6.73286432160804e-05,
-      "loss": 4.8944,
-      "step": 33019
-    },
-    {
-      "epoch": 17.220338983050848,
-      "grad_norm": 1.469718337059021,
-      "learning_rate": 6.732763819095478e-05,
-      "loss": 5.5035,
-      "step": 33020
-    },
-    {
-      "epoch": 17.220860495436767,
-      "grad_norm": 1.5592552423477173,
-      "learning_rate": 6.732663316582915e-05,
-      "loss": 4.7859,
-      "step": 33021
-    },
-    {
-      "epoch": 17.221382007822687,
-      "grad_norm": 1.411392092704773,
-      "learning_rate": 6.732562814070352e-05,
-      "loss": 5.5939,
-      "step": 33022
-    },
-    {
-      "epoch": 17.221903520208606,
-      "grad_norm": 1.4857913255691528,
-      "learning_rate": 6.732462311557789e-05,
-      "loss": 5.4589,
-      "step": 33023
-    },
-    {
-      "epoch": 17.222425032594526,
-      "grad_norm": 1.519877314567566,
-      "learning_rate": 6.732361809045227e-05,
-      "loss": 5.3058,
-      "step": 33024
-    },
-    {
-      "epoch": 17.22294654498044,
-      "grad_norm": 1.6130982637405396,
-      "learning_rate": 6.732261306532663e-05,
-      "loss": 5.4047,
-      "step": 33025
-    },
-    {
-      "epoch": 17.22346805736636,
-      "grad_norm": 1.6669623851776123,
-      "learning_rate": 6.732160804020101e-05,
-      "loss": 5.0843,
-      "step": 33026
-    },
-    {
-      "epoch": 17.22398956975228,
-      "grad_norm": 1.542178750038147,
-      "learning_rate": 6.732060301507537e-05,
-      "loss": 4.9327,
-      "step": 33027
-    },
-    {
-      "epoch": 17.2245110821382,
-      "grad_norm": 1.5398890972137451,
-      "learning_rate": 6.731959798994975e-05,
-      "loss": 5.2641,
-      "step": 33028
-    },
-    {
-      "epoch": 17.22503259452412,
-      "grad_norm": 1.564886450767517,
-      "learning_rate": 6.731859296482413e-05,
-      "loss": 5.2447,
-      "step": 33029
-    },
-    {
-      "epoch": 17.22555410691004,
-      "grad_norm": 1.4806571006774902,
-      "learning_rate": 6.73175879396985e-05,
-      "loss": 5.4506,
-      "step": 33030
-    },
-    {
-      "epoch": 17.22607561929596,
-      "grad_norm": 1.513751745223999,
-      "learning_rate": 6.731658291457287e-05,
-      "loss": 5.2669,
-      "step": 33031
-    },
-    {
-      "epoch": 17.226597131681878,
-      "grad_norm": 1.4606151580810547,
-      "learning_rate": 6.731557788944723e-05,
-      "loss": 5.3583,
-      "step": 33032
-    },
-    {
-      "epoch": 17.227118644067797,
-      "grad_norm": 1.5345149040222168,
-      "learning_rate": 6.731457286432161e-05,
-      "loss": 5.5676,
-      "step": 33033
-    },
-    {
-      "epoch": 17.227640156453717,
-      "grad_norm": 1.569079875946045,
-      "learning_rate": 6.731356783919598e-05,
-      "loss": 5.0151,
-      "step": 33034
-    },
-    {
-      "epoch": 17.228161668839636,
-      "grad_norm": 1.4353673458099365,
-      "learning_rate": 6.731256281407035e-05,
-      "loss": 5.5836,
-      "step": 33035
-    },
-    {
-      "epoch": 17.228683181225556,
-      "grad_norm": 1.5186940431594849,
-      "learning_rate": 6.731155778894472e-05,
-      "loss": 5.2923,
-      "step": 33036
-    },
-    {
-      "epoch": 17.22920469361147,
-      "grad_norm": 1.5227168798446655,
-      "learning_rate": 6.73105527638191e-05,
-      "loss": 5.2045,
-      "step": 33037
-    },
-    {
-      "epoch": 17.22972620599739,
-      "grad_norm": 1.5670908689498901,
-      "learning_rate": 6.730954773869347e-05,
-      "loss": 5.3307,
-      "step": 33038
-    },
-    {
-      "epoch": 17.23024771838331,
-      "grad_norm": 1.5845212936401367,
-      "learning_rate": 6.730854271356785e-05,
-      "loss": 5.5631,
-      "step": 33039
-    },
-    {
-      "epoch": 17.23076923076923,
-      "grad_norm": 1.5276299715042114,
-      "learning_rate": 6.730753768844222e-05,
-      "loss": 4.9217,
-      "step": 33040
-    },
-    {
-      "epoch": 17.23129074315515,
-      "grad_norm": 1.451278805732727,
-      "learning_rate": 6.73065326633166e-05,
-      "loss": 5.1013,
-      "step": 33041
-    },
-    {
-      "epoch": 17.23181225554107,
-      "grad_norm": 1.4404869079589844,
-      "learning_rate": 6.730552763819096e-05,
-      "loss": 5.1484,
-      "step": 33042
-    },
-    {
-      "epoch": 17.23233376792699,
-      "grad_norm": 1.5303912162780762,
-      "learning_rate": 6.730452261306534e-05,
-      "loss": 4.9234,
-      "step": 33043
-    },
-    {
-      "epoch": 17.232855280312908,
-      "grad_norm": 1.5982868671417236,
-      "learning_rate": 6.73035175879397e-05,
-      "loss": 5.0624,
-      "step": 33044
-    },
-    {
-      "epoch": 17.233376792698827,
-      "grad_norm": 1.5918546915054321,
-      "learning_rate": 6.730251256281406e-05,
-      "loss": 5.2703,
-      "step": 33045
-    },
-    {
-      "epoch": 17.233898305084747,
-      "grad_norm": 1.5596381425857544,
-      "learning_rate": 6.730150753768844e-05,
-      "loss": 5.6807,
-      "step": 33046
-    },
-    {
-      "epoch": 17.234419817470666,
-      "grad_norm": 1.387915849685669,
-      "learning_rate": 6.730050251256281e-05,
-      "loss": 5.7737,
-      "step": 33047
-    },
-    {
-      "epoch": 17.234941329856586,
-      "grad_norm": 1.5169873237609863,
-      "learning_rate": 6.729949748743718e-05,
-      "loss": 5.2902,
-      "step": 33048
-    },
-    {
-      "epoch": 17.2354628422425,
-      "grad_norm": 1.3742892742156982,
-      "learning_rate": 6.729849246231156e-05,
-      "loss": 5.6768,
-      "step": 33049
-    },
-    {
-      "epoch": 17.23598435462842,
-      "grad_norm": 1.4264678955078125,
-      "learning_rate": 6.729748743718594e-05,
-      "loss": 5.4213,
-      "step": 33050
-    },
-    {
-      "epoch": 17.23650586701434,
-      "grad_norm": 1.6054863929748535,
-      "learning_rate": 6.72964824120603e-05,
-      "loss": 4.9492,
-      "step": 33051
-    },
-    {
-      "epoch": 17.23702737940026,
-      "grad_norm": 1.4192224740982056,
-      "learning_rate": 6.729547738693468e-05,
-      "loss": 5.1617,
-      "step": 33052
-    },
-    {
-      "epoch": 17.23754889178618,
-      "grad_norm": 1.5397831201553345,
-      "learning_rate": 6.729447236180905e-05,
-      "loss": 5.5401,
-      "step": 33053
-    },
-    {
-      "epoch": 17.2380704041721,
-      "grad_norm": 1.678920865058899,
-      "learning_rate": 6.729346733668342e-05,
-      "loss": 5.1019,
-      "step": 33054
-    },
-    {
-      "epoch": 17.23859191655802,
-      "grad_norm": 1.5028958320617676,
-      "learning_rate": 6.729246231155779e-05,
-      "loss": 5.1992,
-      "step": 33055
-    },
-    {
-      "epoch": 17.239113428943938,
-      "grad_norm": 1.434629201889038,
-      "learning_rate": 6.729145728643217e-05,
-      "loss": 5.2653,
-      "step": 33056
-    },
-    {
-      "epoch": 17.239634941329857,
-      "grad_norm": 1.583923578262329,
-      "learning_rate": 6.729045226130653e-05,
-      "loss": 4.7602,
-      "step": 33057
-    },
-    {
-      "epoch": 17.240156453715777,
-      "grad_norm": 1.4719865322113037,
-      "learning_rate": 6.728944723618091e-05,
-      "loss": 5.7704,
-      "step": 33058
-    },
-    {
-      "epoch": 17.240677966101696,
-      "grad_norm": 1.5597851276397705,
-      "learning_rate": 6.728844221105529e-05,
-      "loss": 5.614,
-      "step": 33059
-    },
-    {
-      "epoch": 17.241199478487616,
-      "grad_norm": 1.4889317750930786,
-      "learning_rate": 6.728743718592965e-05,
-      "loss": 5.5494,
-      "step": 33060
-    },
-    {
-      "epoch": 17.24172099087353,
-      "grad_norm": 1.3837171792984009,
-      "learning_rate": 6.728643216080403e-05,
-      "loss": 5.5966,
-      "step": 33061
-    },
-    {
-      "epoch": 17.24224250325945,
-      "grad_norm": 1.4758646488189697,
-      "learning_rate": 6.728542713567839e-05,
-      "loss": 5.4834,
-      "step": 33062
-    },
-    {
-      "epoch": 17.24276401564537,
-      "grad_norm": 1.519622564315796,
-      "learning_rate": 6.728442211055277e-05,
-      "loss": 5.1589,
-      "step": 33063
-    },
-    {
-      "epoch": 17.24328552803129,
-      "grad_norm": 1.5133532285690308,
-      "learning_rate": 6.728341708542714e-05,
-      "loss": 5.2835,
-      "step": 33064
-    },
-    {
-      "epoch": 17.24380704041721,
-      "grad_norm": 1.5009416341781616,
-      "learning_rate": 6.728241206030151e-05,
-      "loss": 5.5358,
-      "step": 33065
-    },
-    {
-      "epoch": 17.24432855280313,
-      "grad_norm": 1.500701904296875,
-      "learning_rate": 6.728140703517588e-05,
-      "loss": 5.1598,
-      "step": 33066
-    },
-    {
-      "epoch": 17.24485006518905,
-      "grad_norm": 1.5384846925735474,
-      "learning_rate": 6.728040201005026e-05,
-      "loss": 5.147,
-      "step": 33067
-    },
-    {
-      "epoch": 17.245371577574968,
-      "grad_norm": 1.5217021703720093,
-      "learning_rate": 6.727939698492462e-05,
-      "loss": 5.0667,
-      "step": 33068
-    },
-    {
-      "epoch": 17.245893089960887,
-      "grad_norm": 1.624153971672058,
-      "learning_rate": 6.7278391959799e-05,
-      "loss": 5.4558,
-      "step": 33069
-    },
-    {
-      "epoch": 17.246414602346807,
-      "grad_norm": 1.5640376806259155,
-      "learning_rate": 6.727738693467338e-05,
-      "loss": 5.1108,
-      "step": 33070
-    },
-    {
-      "epoch": 17.246936114732726,
-      "grad_norm": 1.4836621284484863,
-      "learning_rate": 6.727638190954774e-05,
-      "loss": 5.097,
-      "step": 33071
-    },
-    {
-      "epoch": 17.247457627118646,
-      "grad_norm": 1.447358250617981,
-      "learning_rate": 6.727537688442212e-05,
-      "loss": 5.501,
-      "step": 33072
-    },
-    {
-      "epoch": 17.24797913950456,
-      "grad_norm": 1.4946554899215698,
-      "learning_rate": 6.727437185929648e-05,
-      "loss": 5.418,
-      "step": 33073
-    },
-    {
-      "epoch": 17.24850065189048,
-      "grad_norm": 1.4443635940551758,
-      "learning_rate": 6.727336683417086e-05,
-      "loss": 5.5153,
-      "step": 33074
-    },
-    {
-      "epoch": 17.2490221642764,
-      "grad_norm": 1.5026990175247192,
-      "learning_rate": 6.727236180904522e-05,
-      "loss": 5.544,
-      "step": 33075
-    },
-    {
-      "epoch": 17.24954367666232,
-      "grad_norm": 1.4703046083450317,
-      "learning_rate": 6.72713567839196e-05,
-      "loss": 5.6513,
-      "step": 33076
-    },
-    {
-      "epoch": 17.25006518904824,
-      "grad_norm": 1.5484355688095093,
-      "learning_rate": 6.727035175879397e-05,
-      "loss": 4.8765,
-      "step": 33077
-    },
-    {
-      "epoch": 17.25058670143416,
-      "grad_norm": 1.3599857091903687,
-      "learning_rate": 6.726934673366834e-05,
-      "loss": 5.4428,
-      "step": 33078
-    },
-    {
-      "epoch": 17.25110821382008,
-      "grad_norm": 1.3206490278244019,
-      "learning_rate": 6.726834170854272e-05,
-      "loss": 4.6021,
-      "step": 33079
-    },
-    {
-      "epoch": 17.251629726205998,
-      "grad_norm": 1.498935341835022,
-      "learning_rate": 6.72673366834171e-05,
-      "loss": 5.4066,
-      "step": 33080
-    },
-    {
-      "epoch": 17.252151238591917,
-      "grad_norm": 1.537130355834961,
-      "learning_rate": 6.726633165829146e-05,
-      "loss": 5.607,
-      "step": 33081
-    },
-    {
-      "epoch": 17.252672750977837,
-      "grad_norm": 1.6036616563796997,
-      "learning_rate": 6.726532663316584e-05,
-      "loss": 4.3552,
-      "step": 33082
-    },
-    {
-      "epoch": 17.253194263363756,
-      "grad_norm": 1.571157693862915,
-      "learning_rate": 6.72643216080402e-05,
-      "loss": 5.1717,
-      "step": 33083
-    },
-    {
-      "epoch": 17.253715775749676,
-      "grad_norm": 1.4563874006271362,
-      "learning_rate": 6.726331658291457e-05,
-      "loss": 5.3483,
-      "step": 33084
-    },
-    {
-      "epoch": 17.25423728813559,
-      "grad_norm": 1.4639335870742798,
-      "learning_rate": 6.726231155778895e-05,
-      "loss": 5.4082,
-      "step": 33085
-    },
-    {
-      "epoch": 17.25475880052151,
-      "grad_norm": 1.5027341842651367,
-      "learning_rate": 6.726130653266331e-05,
-      "loss": 5.2037,
-      "step": 33086
-    },
-    {
-      "epoch": 17.25528031290743,
-      "grad_norm": 1.6195493936538696,
-      "learning_rate": 6.726030150753769e-05,
-      "loss": 5.5882,
-      "step": 33087
-    },
-    {
-      "epoch": 17.25580182529335,
-      "grad_norm": 1.470340609550476,
-      "learning_rate": 6.725929648241205e-05,
-      "loss": 5.2375,
-      "step": 33088
-    },
-    {
-      "epoch": 17.25632333767927,
-      "grad_norm": 1.4694015979766846,
-      "learning_rate": 6.725829145728643e-05,
-      "loss": 5.4597,
-      "step": 33089
-    },
-    {
-      "epoch": 17.25684485006519,
-      "grad_norm": 1.4675050973892212,
-      "learning_rate": 6.725728643216081e-05,
-      "loss": 5.1999,
-      "step": 33090
-    },
-    {
-      "epoch": 17.25736636245111,
-      "grad_norm": 1.4413299560546875,
-      "learning_rate": 6.725628140703519e-05,
-      "loss": 5.5012,
-      "step": 33091
-    },
-    {
-      "epoch": 17.257887874837028,
-      "grad_norm": 1.450779914855957,
-      "learning_rate": 6.725527638190955e-05,
-      "loss": 5.6489,
-      "step": 33092
-    },
-    {
-      "epoch": 17.258409387222947,
-      "grad_norm": 1.3951330184936523,
-      "learning_rate": 6.725427135678393e-05,
-      "loss": 5.6849,
-      "step": 33093
-    },
-    {
-      "epoch": 17.258930899608867,
-      "grad_norm": 1.4722871780395508,
-      "learning_rate": 6.72532663316583e-05,
-      "loss": 5.5087,
-      "step": 33094
-    },
-    {
-      "epoch": 17.259452411994786,
-      "grad_norm": 1.42384934425354,
-      "learning_rate": 6.725226130653267e-05,
-      "loss": 5.4987,
-      "step": 33095
-    },
-    {
-      "epoch": 17.259973924380706,
-      "grad_norm": 1.4726128578186035,
-      "learning_rate": 6.725125628140704e-05,
-      "loss": 5.3534,
-      "step": 33096
-    },
-    {
-      "epoch": 17.26049543676662,
-      "grad_norm": 1.3934811353683472,
-      "learning_rate": 6.725025125628141e-05,
-      "loss": 5.2719,
-      "step": 33097
-    },
-    {
-      "epoch": 17.26101694915254,
-      "grad_norm": 1.6118996143341064,
-      "learning_rate": 6.724924623115578e-05,
-      "loss": 4.9341,
-      "step": 33098
-    },
-    {
-      "epoch": 17.26153846153846,
-      "grad_norm": 1.5051982402801514,
-      "learning_rate": 6.724824120603016e-05,
-      "loss": 4.9932,
-      "step": 33099
-    },
-    {
-      "epoch": 17.26205997392438,
-      "grad_norm": 1.5010708570480347,
-      "learning_rate": 6.724723618090453e-05,
-      "loss": 5.5156,
-      "step": 33100
-    },
-    {
-      "epoch": 17.2625814863103,
-      "grad_norm": 1.4464389085769653,
-      "learning_rate": 6.72462311557789e-05,
-      "loss": 5.5339,
-      "step": 33101
-    },
-    {
-      "epoch": 17.26310299869622,
-      "grad_norm": 1.452811360359192,
-      "learning_rate": 6.724522613065328e-05,
-      "loss": 5.3761,
-      "step": 33102
-    },
-    {
-      "epoch": 17.26362451108214,
-      "grad_norm": 1.4266804456710815,
-      "learning_rate": 6.724422110552764e-05,
-      "loss": 5.4284,
-      "step": 33103
-    },
-    {
-      "epoch": 17.264146023468058,
-      "grad_norm": 1.4337427616119385,
-      "learning_rate": 6.724321608040202e-05,
-      "loss": 5.4107,
-      "step": 33104
-    },
-    {
-      "epoch": 17.264667535853977,
-      "grad_norm": 1.3999773263931274,
-      "learning_rate": 6.724221105527638e-05,
-      "loss": 5.181,
-      "step": 33105
-    },
-    {
-      "epoch": 17.265189048239897,
-      "grad_norm": 1.4316717386245728,
-      "learning_rate": 6.724120603015076e-05,
-      "loss": 5.7232,
-      "step": 33106
-    },
-    {
-      "epoch": 17.265710560625816,
-      "grad_norm": 1.5130348205566406,
-      "learning_rate": 6.724020100502512e-05,
-      "loss": 5.6844,
-      "step": 33107
-    },
-    {
-      "epoch": 17.266232073011736,
-      "grad_norm": 1.507133960723877,
-      "learning_rate": 6.72391959798995e-05,
-      "loss": 5.0935,
-      "step": 33108
-    },
-    {
-      "epoch": 17.26675358539765,
-      "grad_norm": 1.3988518714904785,
-      "learning_rate": 6.723819095477387e-05,
-      "loss": 5.3655,
-      "step": 33109
-    },
-    {
-      "epoch": 17.26727509778357,
-      "grad_norm": 1.4704922437667847,
-      "learning_rate": 6.723718592964824e-05,
-      "loss": 5.4655,
-      "step": 33110
-    },
-    {
-      "epoch": 17.26779661016949,
-      "grad_norm": 1.5172432661056519,
-      "learning_rate": 6.723618090452262e-05,
-      "loss": 5.5026,
-      "step": 33111
-    },
-    {
-      "epoch": 17.26831812255541,
-      "grad_norm": 1.5471817255020142,
-      "learning_rate": 6.723517587939699e-05,
-      "loss": 5.2943,
-      "step": 33112
-    },
-    {
-      "epoch": 17.26883963494133,
-      "grad_norm": 1.468357801437378,
-      "learning_rate": 6.723417085427136e-05,
-      "loss": 5.4901,
-      "step": 33113
-    },
-    {
-      "epoch": 17.26936114732725,
-      "grad_norm": 1.3486647605895996,
-      "learning_rate": 6.723316582914573e-05,
-      "loss": 4.85,
-      "step": 33114
-    },
-    {
-      "epoch": 17.26988265971317,
-      "grad_norm": 1.4096494913101196,
-      "learning_rate": 6.72321608040201e-05,
-      "loss": 5.4003,
-      "step": 33115
-    },
-    {
-      "epoch": 17.270404172099088,
-      "grad_norm": 1.6548469066619873,
-      "learning_rate": 6.723115577889447e-05,
-      "loss": 5.4647,
-      "step": 33116
-    },
-    {
-      "epoch": 17.270925684485007,
-      "grad_norm": 1.4954118728637695,
-      "learning_rate": 6.723015075376885e-05,
-      "loss": 5.6496,
-      "step": 33117
-    },
-    {
-      "epoch": 17.271447196870927,
-      "grad_norm": 1.5319968461990356,
-      "learning_rate": 6.722914572864321e-05,
-      "loss": 4.7432,
-      "step": 33118
-    },
-    {
-      "epoch": 17.271968709256846,
-      "grad_norm": 1.5681389570236206,
-      "learning_rate": 6.722814070351759e-05,
-      "loss": 5.2887,
-      "step": 33119
-    },
-    {
-      "epoch": 17.272490221642762,
-      "grad_norm": 1.5804660320281982,
-      "learning_rate": 6.722713567839195e-05,
-      "loss": 5.4533,
-      "step": 33120
-    },
-    {
-      "epoch": 17.27301173402868,
-      "grad_norm": 1.593161702156067,
-      "learning_rate": 6.722613065326633e-05,
-      "loss": 4.7542,
-      "step": 33121
-    },
-    {
-      "epoch": 17.2735332464146,
-      "grad_norm": 1.577492117881775,
-      "learning_rate": 6.722512562814071e-05,
-      "loss": 5.1895,
-      "step": 33122
-    },
-    {
-      "epoch": 17.27405475880052,
-      "grad_norm": 1.5749127864837646,
-      "learning_rate": 6.722412060301509e-05,
-      "loss": 5.0513,
-      "step": 33123
-    },
-    {
-      "epoch": 17.27457627118644,
-      "grad_norm": 1.4444574117660522,
-      "learning_rate": 6.722311557788945e-05,
-      "loss": 5.2353,
-      "step": 33124
-    },
-    {
-      "epoch": 17.27509778357236,
-      "grad_norm": 1.4535558223724365,
-      "learning_rate": 6.722211055276382e-05,
-      "loss": 5.4569,
-      "step": 33125
-    },
-    {
-      "epoch": 17.27561929595828,
-      "grad_norm": 1.5269397497177124,
-      "learning_rate": 6.72211055276382e-05,
-      "loss": 5.5832,
-      "step": 33126
-    },
-    {
-      "epoch": 17.2761408083442,
-      "grad_norm": 1.4970554113388062,
-      "learning_rate": 6.722010050251256e-05,
-      "loss": 5.2811,
-      "step": 33127
-    },
-    {
-      "epoch": 17.276662320730118,
-      "grad_norm": 1.4499982595443726,
-      "learning_rate": 6.721909547738694e-05,
-      "loss": 5.112,
-      "step": 33128
-    },
-    {
-      "epoch": 17.277183833116037,
-      "grad_norm": 1.5835902690887451,
-      "learning_rate": 6.72180904522613e-05,
-      "loss": 5.5609,
-      "step": 33129
-    },
-    {
-      "epoch": 17.277705345501957,
-      "grad_norm": 1.4528114795684814,
-      "learning_rate": 6.721708542713568e-05,
-      "loss": 5.4103,
-      "step": 33130
-    },
-    {
-      "epoch": 17.278226857887876,
-      "grad_norm": 1.4258383512496948,
-      "learning_rate": 6.721608040201006e-05,
-      "loss": 5.5491,
-      "step": 33131
-    },
-    {
-      "epoch": 17.278748370273792,
-      "grad_norm": 1.4439412355422974,
-      "learning_rate": 6.721507537688443e-05,
-      "loss": 5.6825,
-      "step": 33132
-    },
-    {
-      "epoch": 17.27926988265971,
-      "grad_norm": 1.5801351070404053,
-      "learning_rate": 6.72140703517588e-05,
-      "loss": 4.8651,
-      "step": 33133
-    },
-    {
-      "epoch": 17.27979139504563,
-      "grad_norm": 1.485285758972168,
-      "learning_rate": 6.721306532663318e-05,
-      "loss": 4.4824,
-      "step": 33134
-    },
-    {
-      "epoch": 17.28031290743155,
-      "grad_norm": 1.4575996398925781,
-      "learning_rate": 6.721206030150754e-05,
-      "loss": 5.4403,
-      "step": 33135
-    },
-    {
-      "epoch": 17.28083441981747,
-      "grad_norm": 1.5435138940811157,
-      "learning_rate": 6.721105527638192e-05,
-      "loss": 5.1151,
-      "step": 33136
-    },
-    {
-      "epoch": 17.28135593220339,
-      "grad_norm": 1.4718621969223022,
-      "learning_rate": 6.721005025125628e-05,
-      "loss": 5.0595,
-      "step": 33137
-    },
-    {
-      "epoch": 17.28187744458931,
-      "grad_norm": 1.5386159420013428,
-      "learning_rate": 6.720904522613065e-05,
-      "loss": 5.1874,
-      "step": 33138
-    },
-    {
-      "epoch": 17.28239895697523,
-      "grad_norm": 1.4540226459503174,
-      "learning_rate": 6.720804020100503e-05,
-      "loss": 5.1577,
-      "step": 33139
-    },
-    {
-      "epoch": 17.282920469361148,
-      "grad_norm": 1.475630521774292,
-      "learning_rate": 6.720703517587939e-05,
-      "loss": 4.8422,
-      "step": 33140
-    },
-    {
-      "epoch": 17.283441981747067,
-      "grad_norm": 1.6125669479370117,
-      "learning_rate": 6.720603015075377e-05,
-      "loss": 4.4939,
-      "step": 33141
-    },
-    {
-      "epoch": 17.283963494132987,
-      "grad_norm": 1.4516061544418335,
-      "learning_rate": 6.720502512562815e-05,
-      "loss": 5.4468,
-      "step": 33142
-    },
-    {
-      "epoch": 17.284485006518906,
-      "grad_norm": 1.5348562002182007,
-      "learning_rate": 6.720402010050252e-05,
-      "loss": 5.2769,
-      "step": 33143
-    },
-    {
-      "epoch": 17.285006518904822,
-      "grad_norm": 1.4671552181243896,
-      "learning_rate": 6.720301507537689e-05,
-      "loss": 5.545,
-      "step": 33144
-    },
-    {
-      "epoch": 17.285528031290742,
-      "grad_norm": 1.4223171472549438,
-      "learning_rate": 6.720201005025127e-05,
-      "loss": 5.734,
-      "step": 33145
-    },
-    {
-      "epoch": 17.28604954367666,
-      "grad_norm": 1.5763932466506958,
-      "learning_rate": 6.720100502512563e-05,
-      "loss": 5.3947,
-      "step": 33146
-    },
-    {
-      "epoch": 17.28657105606258,
-      "grad_norm": 1.5405575037002563,
-      "learning_rate": 6.720000000000001e-05,
-      "loss": 5.3235,
-      "step": 33147
-    },
-    {
-      "epoch": 17.2870925684485,
-      "grad_norm": 1.5947425365447998,
-      "learning_rate": 6.719899497487437e-05,
-      "loss": 5.3613,
-      "step": 33148
-    },
-    {
-      "epoch": 17.28761408083442,
-      "grad_norm": 1.4649916887283325,
-      "learning_rate": 6.719798994974875e-05,
-      "loss": 5.5909,
-      "step": 33149
-    },
-    {
-      "epoch": 17.28813559322034,
-      "grad_norm": 1.4919161796569824,
-      "learning_rate": 6.719698492462311e-05,
-      "loss": 5.4265,
-      "step": 33150
-    },
-    {
-      "epoch": 17.28865710560626,
-      "grad_norm": 1.5208284854888916,
-      "learning_rate": 6.719597989949749e-05,
-      "loss": 5.2367,
-      "step": 33151
-    },
-    {
-      "epoch": 17.289178617992178,
-      "grad_norm": 1.462421178817749,
-      "learning_rate": 6.719497487437187e-05,
-      "loss": 5.0673,
-      "step": 33152
-    },
-    {
-      "epoch": 17.289700130378097,
-      "grad_norm": 1.5937293767929077,
-      "learning_rate": 6.719396984924623e-05,
-      "loss": 4.9681,
-      "step": 33153
-    },
-    {
-      "epoch": 17.290221642764017,
-      "grad_norm": 1.513319492340088,
-      "learning_rate": 6.719296482412061e-05,
-      "loss": 5.2845,
-      "step": 33154
-    },
-    {
-      "epoch": 17.290743155149936,
-      "grad_norm": 1.570441722869873,
-      "learning_rate": 6.719195979899498e-05,
-      "loss": 5.2329,
-      "step": 33155
-    },
-    {
-      "epoch": 17.291264667535852,
-      "grad_norm": 1.4280297756195068,
-      "learning_rate": 6.719095477386935e-05,
-      "loss": 5.57,
-      "step": 33156
-    },
-    {
-      "epoch": 17.291786179921772,
-      "grad_norm": 1.588638186454773,
-      "learning_rate": 6.718994974874372e-05,
-      "loss": 5.0189,
-      "step": 33157
-    },
-    {
-      "epoch": 17.29230769230769,
-      "grad_norm": 1.5735361576080322,
-      "learning_rate": 6.71889447236181e-05,
-      "loss": 5.2035,
-      "step": 33158
-    },
-    {
-      "epoch": 17.29282920469361,
-      "grad_norm": 1.6605991125106812,
-      "learning_rate": 6.718793969849246e-05,
-      "loss": 5.3493,
-      "step": 33159
-    },
-    {
-      "epoch": 17.29335071707953,
-      "grad_norm": 1.5678064823150635,
-      "learning_rate": 6.718693467336684e-05,
-      "loss": 5.3412,
-      "step": 33160
-    },
-    {
-      "epoch": 17.29387222946545,
-      "grad_norm": 1.3842918872833252,
-      "learning_rate": 6.71859296482412e-05,
-      "loss": 4.9148,
-      "step": 33161
-    },
-    {
-      "epoch": 17.29439374185137,
-      "grad_norm": 1.4697318077087402,
-      "learning_rate": 6.718492462311558e-05,
-      "loss": 5.0472,
-      "step": 33162
-    },
-    {
-      "epoch": 17.29491525423729,
-      "grad_norm": 1.5619704723358154,
-      "learning_rate": 6.718391959798996e-05,
-      "loss": 4.5757,
-      "step": 33163
-    },
-    {
-      "epoch": 17.295436766623208,
-      "grad_norm": 1.530806303024292,
-      "learning_rate": 6.718291457286432e-05,
-      "loss": 5.1879,
-      "step": 33164
-    },
-    {
-      "epoch": 17.295958279009128,
-      "grad_norm": 1.4949417114257812,
-      "learning_rate": 6.71819095477387e-05,
-      "loss": 4.8458,
-      "step": 33165
-    },
-    {
-      "epoch": 17.296479791395047,
-      "grad_norm": 1.6334277391433716,
-      "learning_rate": 6.718090452261306e-05,
-      "loss": 5.4912,
-      "step": 33166
-    },
-    {
-      "epoch": 17.297001303780966,
-      "grad_norm": 1.492998480796814,
-      "learning_rate": 6.717989949748744e-05,
-      "loss": 5.406,
-      "step": 33167
-    },
-    {
-      "epoch": 17.297522816166882,
-      "grad_norm": 1.4616643190383911,
-      "learning_rate": 6.71788944723618e-05,
-      "loss": 5.2367,
-      "step": 33168
-    },
-    {
-      "epoch": 17.298044328552802,
-      "grad_norm": 1.3032419681549072,
-      "learning_rate": 6.717788944723618e-05,
-      "loss": 4.8502,
-      "step": 33169
-    },
-    {
-      "epoch": 17.29856584093872,
-      "grad_norm": 1.5514568090438843,
-      "learning_rate": 6.717688442211055e-05,
-      "loss": 5.2081,
-      "step": 33170
-    },
-    {
-      "epoch": 17.29908735332464,
-      "grad_norm": 1.6480389833450317,
-      "learning_rate": 6.717587939698493e-05,
-      "loss": 5.2,
-      "step": 33171
-    },
-    {
-      "epoch": 17.29960886571056,
-      "grad_norm": 1.4397997856140137,
-      "learning_rate": 6.71748743718593e-05,
-      "loss": 5.6867,
-      "step": 33172
-    },
-    {
-      "epoch": 17.30013037809648,
-      "grad_norm": 1.515960454940796,
-      "learning_rate": 6.717386934673368e-05,
-      "loss": 5.4481,
-      "step": 33173
-    },
-    {
-      "epoch": 17.3006518904824,
-      "grad_norm": 1.4769114255905151,
-      "learning_rate": 6.717286432160805e-05,
-      "loss": 5.2841,
-      "step": 33174
-    },
-    {
-      "epoch": 17.30117340286832,
-      "grad_norm": 1.6379542350769043,
-      "learning_rate": 6.717185929648242e-05,
-      "loss": 4.9416,
-      "step": 33175
-    },
-    {
-      "epoch": 17.301694915254238,
-      "grad_norm": 1.3936405181884766,
-      "learning_rate": 6.717085427135679e-05,
-      "loss": 5.6795,
-      "step": 33176
-    },
-    {
-      "epoch": 17.302216427640158,
-      "grad_norm": 1.4576603174209595,
-      "learning_rate": 6.716984924623115e-05,
-      "loss": 4.9436,
-      "step": 33177
-    },
-    {
-      "epoch": 17.302737940026077,
-      "grad_norm": 1.5046980381011963,
-      "learning_rate": 6.716884422110553e-05,
-      "loss": 5.4864,
-      "step": 33178
-    },
-    {
-      "epoch": 17.303259452411996,
-      "grad_norm": 1.546419382095337,
-      "learning_rate": 6.71678391959799e-05,
-      "loss": 4.5243,
-      "step": 33179
-    },
-    {
-      "epoch": 17.303780964797912,
-      "grad_norm": 1.6308013200759888,
-      "learning_rate": 6.716683417085427e-05,
-      "loss": 5.6363,
-      "step": 33180
-    },
-    {
-      "epoch": 17.304302477183832,
-      "grad_norm": 1.5208423137664795,
-      "learning_rate": 6.716582914572864e-05,
-      "loss": 5.4821,
-      "step": 33181
-    },
-    {
-      "epoch": 17.30482398956975,
-      "grad_norm": 1.4783250093460083,
-      "learning_rate": 6.716482412060301e-05,
-      "loss": 5.5555,
-      "step": 33182
-    },
-    {
-      "epoch": 17.30534550195567,
-      "grad_norm": 1.5645936727523804,
-      "learning_rate": 6.716381909547739e-05,
-      "loss": 5.428,
-      "step": 33183
-    },
-    {
-      "epoch": 17.30586701434159,
-      "grad_norm": 1.579830288887024,
-      "learning_rate": 6.716281407035177e-05,
-      "loss": 5.6782,
-      "step": 33184
-    },
-    {
-      "epoch": 17.30638852672751,
-      "grad_norm": 1.605069637298584,
-      "learning_rate": 6.716180904522613e-05,
-      "loss": 4.8672,
-      "step": 33185
-    },
-    {
-      "epoch": 17.30691003911343,
-      "grad_norm": 1.4614670276641846,
-      "learning_rate": 6.716080402010051e-05,
-      "loss": 5.2024,
-      "step": 33186
-    },
-    {
-      "epoch": 17.30743155149935,
-      "grad_norm": 1.3892738819122314,
-      "learning_rate": 6.715979899497488e-05,
-      "loss": 5.7253,
-      "step": 33187
-    },
-    {
-      "epoch": 17.307953063885268,
-      "grad_norm": 1.490455985069275,
-      "learning_rate": 6.715879396984925e-05,
-      "loss": 5.0238,
-      "step": 33188
-    },
-    {
-      "epoch": 17.308474576271188,
-      "grad_norm": 1.476836085319519,
-      "learning_rate": 6.715778894472362e-05,
-      "loss": 5.3054,
-      "step": 33189
-    },
-    {
-      "epoch": 17.308996088657107,
-      "grad_norm": 1.5626593828201294,
-      "learning_rate": 6.7156783919598e-05,
-      "loss": 5.1933,
-      "step": 33190
-    },
-    {
-      "epoch": 17.309517601043027,
-      "grad_norm": 1.4611499309539795,
-      "learning_rate": 6.715577889447236e-05,
-      "loss": 5.4942,
-      "step": 33191
-    },
-    {
-      "epoch": 17.310039113428942,
-      "grad_norm": 1.4913069009780884,
-      "learning_rate": 6.715477386934674e-05,
-      "loss": 5.6672,
-      "step": 33192
-    },
-    {
-      "epoch": 17.310560625814862,
-      "grad_norm": 1.617273211479187,
-      "learning_rate": 6.715376884422112e-05,
-      "loss": 5.1062,
-      "step": 33193
-    },
-    {
-      "epoch": 17.31108213820078,
-      "grad_norm": 1.4564327001571655,
-      "learning_rate": 6.715276381909548e-05,
-      "loss": 5.2344,
-      "step": 33194
-    },
-    {
-      "epoch": 17.3116036505867,
-      "grad_norm": 1.599757194519043,
-      "learning_rate": 6.715175879396986e-05,
-      "loss": 5.462,
-      "step": 33195
-    },
-    {
-      "epoch": 17.31212516297262,
-      "grad_norm": 1.5887442827224731,
-      "learning_rate": 6.715075376884422e-05,
-      "loss": 4.7409,
-      "step": 33196
-    },
-    {
-      "epoch": 17.31264667535854,
-      "grad_norm": 1.4789084196090698,
-      "learning_rate": 6.71497487437186e-05,
-      "loss": 5.8818,
-      "step": 33197
-    },
-    {
-      "epoch": 17.31316818774446,
-      "grad_norm": 1.5181585550308228,
-      "learning_rate": 6.714874371859296e-05,
-      "loss": 4.9669,
-      "step": 33198
-    },
-    {
-      "epoch": 17.31368970013038,
-      "grad_norm": 1.5422979593276978,
-      "learning_rate": 6.714773869346734e-05,
-      "loss": 5.3478,
-      "step": 33199
-    },
-    {
-      "epoch": 17.314211212516298,
-      "grad_norm": 1.5148438215255737,
-      "learning_rate": 6.71467336683417e-05,
-      "loss": 4.943,
-      "step": 33200
-    },
-    {
-      "epoch": 17.314732724902218,
-      "grad_norm": 1.4661107063293457,
-      "learning_rate": 6.714572864321608e-05,
-      "loss": 5.4885,
-      "step": 33201
-    },
-    {
-      "epoch": 17.315254237288137,
-      "grad_norm": 1.4110069274902344,
-      "learning_rate": 6.714472361809045e-05,
-      "loss": 5.403,
-      "step": 33202
-    },
-    {
-      "epoch": 17.315775749674053,
-      "grad_norm": 1.3942290544509888,
-      "learning_rate": 6.714371859296483e-05,
-      "loss": 5.5083,
-      "step": 33203
-    },
-    {
-      "epoch": 17.316297262059972,
-      "grad_norm": 1.659695029258728,
-      "learning_rate": 6.71427135678392e-05,
-      "loss": 5.3198,
-      "step": 33204
-    },
-    {
-      "epoch": 17.316818774445892,
-      "grad_norm": 1.5211834907531738,
-      "learning_rate": 6.714170854271357e-05,
-      "loss": 5.5309,
-      "step": 33205
-    },
-    {
-      "epoch": 17.31734028683181,
-      "grad_norm": 1.4818496704101562,
-      "learning_rate": 6.714070351758795e-05,
-      "loss": 5.6084,
-      "step": 33206
-    },
-    {
-      "epoch": 17.31786179921773,
-      "grad_norm": 1.4255462884902954,
-      "learning_rate": 6.713969849246231e-05,
-      "loss": 5.6652,
-      "step": 33207
-    },
-    {
-      "epoch": 17.31838331160365,
-      "grad_norm": 1.5564340353012085,
-      "learning_rate": 6.713869346733669e-05,
-      "loss": 5.3376,
-      "step": 33208
-    },
-    {
-      "epoch": 17.31890482398957,
-      "grad_norm": 1.4262070655822754,
-      "learning_rate": 6.713768844221105e-05,
-      "loss": 5.5301,
-      "step": 33209
-    },
-    {
-      "epoch": 17.31942633637549,
-      "grad_norm": 1.4322673082351685,
-      "learning_rate": 6.713668341708543e-05,
-      "loss": 5.5563,
-      "step": 33210
-    },
-    {
-      "epoch": 17.31994784876141,
-      "grad_norm": 1.450416922569275,
-      "learning_rate": 6.71356783919598e-05,
-      "loss": 5.3191,
-      "step": 33211
-    },
-    {
-      "epoch": 17.320469361147328,
-      "grad_norm": 1.4206316471099854,
-      "learning_rate": 6.713467336683417e-05,
-      "loss": 5.2099,
-      "step": 33212
-    },
-    {
-      "epoch": 17.320990873533248,
-      "grad_norm": 1.3772287368774414,
-      "learning_rate": 6.713366834170855e-05,
-      "loss": 5.8706,
-      "step": 33213
-    },
-    {
-      "epoch": 17.321512385919167,
-      "grad_norm": 1.4742846488952637,
-      "learning_rate": 6.713266331658293e-05,
-      "loss": 5.4429,
-      "step": 33214
-    },
-    {
-      "epoch": 17.322033898305083,
-      "grad_norm": 1.4918235540390015,
-      "learning_rate": 6.713165829145729e-05,
-      "loss": 5.1708,
-      "step": 33215
-    },
-    {
-      "epoch": 17.322555410691002,
-      "grad_norm": 1.4645241498947144,
-      "learning_rate": 6.713065326633167e-05,
-      "loss": 5.6119,
-      "step": 33216
-    },
-    {
-      "epoch": 17.323076923076922,
-      "grad_norm": 1.4755951166152954,
-      "learning_rate": 6.712964824120603e-05,
-      "loss": 5.665,
-      "step": 33217
-    },
-    {
-      "epoch": 17.32359843546284,
-      "grad_norm": 1.4705549478530884,
-      "learning_rate": 6.71286432160804e-05,
-      "loss": 5.5169,
-      "step": 33218
-    },
-    {
-      "epoch": 17.32411994784876,
-      "grad_norm": 1.5725562572479248,
-      "learning_rate": 6.712763819095478e-05,
-      "loss": 4.893,
-      "step": 33219
-    },
-    {
-      "epoch": 17.32464146023468,
-      "grad_norm": 1.4593894481658936,
-      "learning_rate": 6.712663316582914e-05,
-      "loss": 5.9263,
-      "step": 33220
-    },
-    {
-      "epoch": 17.3251629726206,
-      "grad_norm": 1.4249464273452759,
-      "learning_rate": 6.712562814070352e-05,
-      "loss": 5.1909,
-      "step": 33221
-    },
-    {
-      "epoch": 17.32568448500652,
-      "grad_norm": 1.4653352499008179,
-      "learning_rate": 6.712462311557788e-05,
-      "loss": 4.9265,
-      "step": 33222
-    },
-    {
-      "epoch": 17.32620599739244,
-      "grad_norm": 1.4481717348098755,
-      "learning_rate": 6.712361809045226e-05,
-      "loss": 5.4974,
-      "step": 33223
-    },
-    {
-      "epoch": 17.326727509778358,
-      "grad_norm": 1.5504660606384277,
-      "learning_rate": 6.712261306532664e-05,
-      "loss": 5.4364,
-      "step": 33224
-    },
-    {
-      "epoch": 17.327249022164278,
-      "grad_norm": 1.452991247177124,
-      "learning_rate": 6.712160804020102e-05,
-      "loss": 5.2479,
-      "step": 33225
-    },
-    {
-      "epoch": 17.327770534550197,
-      "grad_norm": 1.496942162513733,
-      "learning_rate": 6.712060301507538e-05,
-      "loss": 5.4417,
-      "step": 33226
-    },
-    {
-      "epoch": 17.328292046936113,
-      "grad_norm": 1.5147522687911987,
-      "learning_rate": 6.711959798994976e-05,
-      "loss": 5.3122,
-      "step": 33227
-    },
-    {
-      "epoch": 17.328813559322032,
-      "grad_norm": 1.490186333656311,
-      "learning_rate": 6.711859296482412e-05,
-      "loss": 5.4852,
-      "step": 33228
-    },
-    {
-      "epoch": 17.329335071707952,
-      "grad_norm": 1.4018722772598267,
-      "learning_rate": 6.71175879396985e-05,
-      "loss": 5.7745,
-      "step": 33229
-    },
-    {
-      "epoch": 17.32985658409387,
-      "grad_norm": 1.5242911577224731,
-      "learning_rate": 6.711658291457287e-05,
-      "loss": 5.1237,
-      "step": 33230
-    },
-    {
-      "epoch": 17.33037809647979,
-      "grad_norm": 1.3994967937469482,
-      "learning_rate": 6.711557788944723e-05,
-      "loss": 4.971,
-      "step": 33231
-    },
-    {
-      "epoch": 17.33089960886571,
-      "grad_norm": 1.6286417245864868,
-      "learning_rate": 6.711457286432161e-05,
-      "loss": 5.2154,
-      "step": 33232
-    },
-    {
-      "epoch": 17.33142112125163,
-      "grad_norm": 1.5904589891433716,
-      "learning_rate": 6.711356783919599e-05,
-      "loss": 5.648,
-      "step": 33233
-    },
-    {
-      "epoch": 17.33194263363755,
-      "grad_norm": 1.531175136566162,
-      "learning_rate": 6.711256281407036e-05,
-      "loss": 5.2367,
-      "step": 33234
-    },
-    {
-      "epoch": 17.33246414602347,
-      "grad_norm": 1.4969462156295776,
-      "learning_rate": 6.711155778894473e-05,
-      "loss": 5.3353,
-      "step": 33235
-    },
-    {
-      "epoch": 17.332985658409388,
-      "grad_norm": 1.494615077972412,
-      "learning_rate": 6.71105527638191e-05,
-      "loss": 5.4642,
-      "step": 33236
-    },
-    {
-      "epoch": 17.333507170795308,
-      "grad_norm": 1.4615050554275513,
-      "learning_rate": 6.710954773869347e-05,
-      "loss": 5.5628,
-      "step": 33237
-    },
-    {
-      "epoch": 17.334028683181227,
-      "grad_norm": 1.5205121040344238,
-      "learning_rate": 6.710854271356785e-05,
-      "loss": 5.2111,
-      "step": 33238
-    },
-    {
-      "epoch": 17.334550195567143,
-      "grad_norm": 1.4522955417633057,
-      "learning_rate": 6.710753768844221e-05,
-      "loss": 5.0421,
-      "step": 33239
-    },
-    {
-      "epoch": 17.335071707953063,
-      "grad_norm": 1.5782256126403809,
-      "learning_rate": 6.710653266331659e-05,
-      "loss": 5.0426,
-      "step": 33240
-    },
-    {
-      "epoch": 17.335593220338982,
-      "grad_norm": 1.4247515201568604,
-      "learning_rate": 6.710552763819095e-05,
-      "loss": 5.2889,
-      "step": 33241
-    },
-    {
-      "epoch": 17.3361147327249,
-      "grad_norm": 1.37389075756073,
-      "learning_rate": 6.710452261306533e-05,
-      "loss": 5.4888,
-      "step": 33242
-    },
-    {
-      "epoch": 17.33663624511082,
-      "grad_norm": 1.4760550260543823,
-      "learning_rate": 6.71035175879397e-05,
-      "loss": 4.9202,
-      "step": 33243
-    },
-    {
-      "epoch": 17.33715775749674,
-      "grad_norm": 1.549900770187378,
-      "learning_rate": 6.710251256281407e-05,
-      "loss": 4.7648,
-      "step": 33244
-    },
-    {
-      "epoch": 17.33767926988266,
-      "grad_norm": 1.4825389385223389,
-      "learning_rate": 6.710150753768845e-05,
-      "loss": 5.5162,
-      "step": 33245
-    },
-    {
-      "epoch": 17.33820078226858,
-      "grad_norm": 1.4502508640289307,
-      "learning_rate": 6.710050251256282e-05,
-      "loss": 5.4234,
-      "step": 33246
-    },
-    {
-      "epoch": 17.3387222946545,
-      "grad_norm": 1.4650189876556396,
-      "learning_rate": 6.70994974874372e-05,
-      "loss": 5.312,
-      "step": 33247
-    },
-    {
-      "epoch": 17.33924380704042,
-      "grad_norm": 1.440493106842041,
-      "learning_rate": 6.709849246231156e-05,
-      "loss": 4.8152,
-      "step": 33248
-    },
-    {
-      "epoch": 17.339765319426338,
-      "grad_norm": 1.4682263135910034,
-      "learning_rate": 6.709748743718594e-05,
-      "loss": 5.5043,
-      "step": 33249
-    },
-    {
-      "epoch": 17.340286831812257,
-      "grad_norm": 1.5719629526138306,
-      "learning_rate": 6.70964824120603e-05,
-      "loss": 5.193,
-      "step": 33250
-    },
-    {
-      "epoch": 17.340808344198173,
-      "grad_norm": 1.420749545097351,
-      "learning_rate": 6.709547738693468e-05,
-      "loss": 5.5757,
-      "step": 33251
-    },
-    {
-      "epoch": 17.341329856584093,
-      "grad_norm": 1.524451732635498,
-      "learning_rate": 6.709447236180904e-05,
-      "loss": 5.0751,
-      "step": 33252
-    },
-    {
-      "epoch": 17.341851368970012,
-      "grad_norm": 1.5972613096237183,
-      "learning_rate": 6.709346733668342e-05,
-      "loss": 4.8599,
-      "step": 33253
-    },
-    {
-      "epoch": 17.34237288135593,
-      "grad_norm": 1.5457258224487305,
-      "learning_rate": 6.70924623115578e-05,
-      "loss": 5.3253,
-      "step": 33254
-    },
-    {
-      "epoch": 17.34289439374185,
-      "grad_norm": 1.469899296760559,
-      "learning_rate": 6.709145728643218e-05,
-      "loss": 5.3576,
-      "step": 33255
-    },
-    {
-      "epoch": 17.34341590612777,
-      "grad_norm": 1.4682813882827759,
-      "learning_rate": 6.709045226130654e-05,
-      "loss": 5.5739,
-      "step": 33256
-    },
-    {
-      "epoch": 17.34393741851369,
-      "grad_norm": 1.4549336433410645,
-      "learning_rate": 6.70894472361809e-05,
-      "loss": 4.8927,
-      "step": 33257
-    },
-    {
-      "epoch": 17.34445893089961,
-      "grad_norm": 1.4896405935287476,
-      "learning_rate": 6.708844221105528e-05,
-      "loss": 5.1153,
-      "step": 33258
-    },
-    {
-      "epoch": 17.34498044328553,
-      "grad_norm": 1.4678148031234741,
-      "learning_rate": 6.708743718592965e-05,
-      "loss": 5.5373,
-      "step": 33259
-    },
-    {
-      "epoch": 17.34550195567145,
-      "grad_norm": 1.540353775024414,
-      "learning_rate": 6.708643216080402e-05,
-      "loss": 5.0122,
-      "step": 33260
-    },
-    {
-      "epoch": 17.346023468057368,
-      "grad_norm": 1.4068691730499268,
-      "learning_rate": 6.708542713567839e-05,
-      "loss": 5.7613,
-      "step": 33261
-    },
-    {
-      "epoch": 17.346544980443287,
-      "grad_norm": 1.5258499383926392,
-      "learning_rate": 6.708442211055277e-05,
-      "loss": 5.6161,
-      "step": 33262
-    },
-    {
-      "epoch": 17.347066492829203,
-      "grad_norm": 1.4104093313217163,
-      "learning_rate": 6.708341708542713e-05,
-      "loss": 5.5217,
-      "step": 33263
-    },
-    {
-      "epoch": 17.347588005215123,
-      "grad_norm": 1.4445139169692993,
-      "learning_rate": 6.708241206030151e-05,
-      "loss": 5.2846,
-      "step": 33264
-    },
-    {
-      "epoch": 17.348109517601042,
-      "grad_norm": 1.3789421319961548,
-      "learning_rate": 6.708140703517589e-05,
-      "loss": 5.4516,
-      "step": 33265
-    },
-    {
-      "epoch": 17.34863102998696,
-      "grad_norm": 1.4387985467910767,
-      "learning_rate": 6.708040201005026e-05,
-      "loss": 5.3738,
-      "step": 33266
-    },
-    {
-      "epoch": 17.34915254237288,
-      "grad_norm": 1.5338661670684814,
-      "learning_rate": 6.707939698492463e-05,
-      "loss": 5.4728,
-      "step": 33267
-    },
-    {
-      "epoch": 17.3496740547588,
-      "grad_norm": 1.5096923112869263,
-      "learning_rate": 6.7078391959799e-05,
-      "loss": 5.0071,
-      "step": 33268
-    },
-    {
-      "epoch": 17.35019556714472,
-      "grad_norm": 1.5378172397613525,
-      "learning_rate": 6.707738693467337e-05,
-      "loss": 5.3125,
-      "step": 33269
-    },
-    {
-      "epoch": 17.35071707953064,
-      "grad_norm": 1.4495646953582764,
-      "learning_rate": 6.707638190954773e-05,
-      "loss": 5.6163,
-      "step": 33270
-    },
-    {
-      "epoch": 17.35123859191656,
-      "grad_norm": 1.5388805866241455,
-      "learning_rate": 6.707537688442211e-05,
-      "loss": 5.5217,
-      "step": 33271
-    },
-    {
-      "epoch": 17.35176010430248,
-      "grad_norm": 1.5537958145141602,
-      "learning_rate": 6.707437185929648e-05,
-      "loss": 5.036,
-      "step": 33272
-    },
-    {
-      "epoch": 17.352281616688398,
-      "grad_norm": 1.508648157119751,
-      "learning_rate": 6.707336683417085e-05,
-      "loss": 4.7866,
-      "step": 33273
-    },
-    {
-      "epoch": 17.352803129074317,
-      "grad_norm": 1.6014647483825684,
-      "learning_rate": 6.707236180904522e-05,
-      "loss": 5.2154,
-      "step": 33274
-    },
-    {
-      "epoch": 17.353324641460233,
-      "grad_norm": 1.486573576927185,
-      "learning_rate": 6.70713567839196e-05,
-      "loss": 5.3862,
-      "step": 33275
-    },
-    {
-      "epoch": 17.353846153846153,
-      "grad_norm": 1.4437408447265625,
-      "learning_rate": 6.707035175879397e-05,
-      "loss": 4.9848,
-      "step": 33276
-    },
-    {
-      "epoch": 17.354367666232072,
-      "grad_norm": 1.5180683135986328,
-      "learning_rate": 6.706934673366835e-05,
-      "loss": 4.8545,
-      "step": 33277
-    },
-    {
-      "epoch": 17.35488917861799,
-      "grad_norm": 1.5918524265289307,
-      "learning_rate": 6.706834170854272e-05,
-      "loss": 5.0233,
-      "step": 33278
-    },
-    {
-      "epoch": 17.35541069100391,
-      "grad_norm": 1.616398572921753,
-      "learning_rate": 6.70673366834171e-05,
-      "loss": 5.2491,
-      "step": 33279
-    },
-    {
-      "epoch": 17.35593220338983,
-      "grad_norm": 1.4517589807510376,
-      "learning_rate": 6.706633165829146e-05,
-      "loss": 5.5103,
-      "step": 33280
-    },
-    {
-      "epoch": 17.35645371577575,
-      "grad_norm": 1.4538401365280151,
-      "learning_rate": 6.706532663316584e-05,
-      "loss": 5.5721,
-      "step": 33281
-    },
-    {
-      "epoch": 17.35697522816167,
-      "grad_norm": 1.5214110612869263,
-      "learning_rate": 6.70643216080402e-05,
-      "loss": 5.0706,
-      "step": 33282
-    },
-    {
-      "epoch": 17.35749674054759,
-      "grad_norm": 1.416880488395691,
-      "learning_rate": 6.706331658291458e-05,
-      "loss": 5.7093,
-      "step": 33283
-    },
-    {
-      "epoch": 17.35801825293351,
-      "grad_norm": 1.510191798210144,
-      "learning_rate": 6.706231155778894e-05,
-      "loss": 5.043,
-      "step": 33284
-    },
-    {
-      "epoch": 17.358539765319428,
-      "grad_norm": 1.4894198179244995,
-      "learning_rate": 6.706130653266332e-05,
-      "loss": 5.545,
-      "step": 33285
-    },
-    {
-      "epoch": 17.359061277705347,
-      "grad_norm": 1.5111842155456543,
-      "learning_rate": 6.70603015075377e-05,
-      "loss": 5.2214,
-      "step": 33286
-    },
-    {
-      "epoch": 17.359582790091263,
-      "grad_norm": 1.4494503736495972,
-      "learning_rate": 6.705929648241206e-05,
-      "loss": 5.4127,
-      "step": 33287
-    },
-    {
-      "epoch": 17.360104302477183,
-      "grad_norm": 1.5593048334121704,
-      "learning_rate": 6.705829145728644e-05,
-      "loss": 5.1627,
-      "step": 33288
-    },
-    {
-      "epoch": 17.360625814863102,
-      "grad_norm": 1.528159737586975,
-      "learning_rate": 6.70572864321608e-05,
-      "loss": 5.4747,
-      "step": 33289
-    },
-    {
-      "epoch": 17.36114732724902,
-      "grad_norm": 1.5171750783920288,
-      "learning_rate": 6.705628140703518e-05,
-      "loss": 5.5792,
-      "step": 33290
-    },
-    {
-      "epoch": 17.36166883963494,
-      "grad_norm": 1.5074782371520996,
-      "learning_rate": 6.705527638190955e-05,
-      "loss": 5.158,
-      "step": 33291
-    },
-    {
-      "epoch": 17.36219035202086,
-      "grad_norm": 1.4312578439712524,
-      "learning_rate": 6.705427135678392e-05,
-      "loss": 5.337,
-      "step": 33292
-    },
-    {
-      "epoch": 17.36271186440678,
-      "grad_norm": 1.5035767555236816,
-      "learning_rate": 6.705326633165829e-05,
-      "loss": 5.4988,
-      "step": 33293
-    },
-    {
-      "epoch": 17.3632333767927,
-      "grad_norm": 1.5542173385620117,
-      "learning_rate": 6.705226130653267e-05,
-      "loss": 5.3462,
-      "step": 33294
-    },
-    {
-      "epoch": 17.36375488917862,
-      "grad_norm": 1.4892315864562988,
-      "learning_rate": 6.705125628140703e-05,
-      "loss": 5.6049,
-      "step": 33295
-    },
-    {
-      "epoch": 17.36427640156454,
-      "grad_norm": 1.4602221250534058,
-      "learning_rate": 6.705025125628141e-05,
-      "loss": 5.5096,
-      "step": 33296
-    },
-    {
-      "epoch": 17.364797913950458,
-      "grad_norm": 1.6699146032333374,
-      "learning_rate": 6.704924623115579e-05,
-      "loss": 5.3029,
-      "step": 33297
-    },
-    {
-      "epoch": 17.365319426336377,
-      "grad_norm": 1.510675311088562,
-      "learning_rate": 6.704824120603015e-05,
-      "loss": 5.2456,
-      "step": 33298
-    },
-    {
-      "epoch": 17.365840938722293,
-      "grad_norm": 1.4095947742462158,
-      "learning_rate": 6.704723618090453e-05,
-      "loss": 4.7944,
-      "step": 33299
-    },
-    {
-      "epoch": 17.366362451108213,
-      "grad_norm": 1.4368574619293213,
-      "learning_rate": 6.70462311557789e-05,
-      "loss": 5.3269,
-      "step": 33300
-    },
-    {
-      "epoch": 17.366883963494132,
-      "grad_norm": 1.416461706161499,
-      "learning_rate": 6.704522613065327e-05,
-      "loss": 5.0957,
-      "step": 33301
-    },
-    {
-      "epoch": 17.36740547588005,
-      "grad_norm": 1.4320602416992188,
-      "learning_rate": 6.704422110552764e-05,
-      "loss": 5.3166,
-      "step": 33302
-    },
-    {
-      "epoch": 17.36792698826597,
-      "grad_norm": 1.3926359415054321,
-      "learning_rate": 6.704321608040201e-05,
-      "loss": 5.7496,
-      "step": 33303
-    },
-    {
-      "epoch": 17.36844850065189,
-      "grad_norm": 1.3798729181289673,
-      "learning_rate": 6.704221105527638e-05,
-      "loss": 5.8203,
-      "step": 33304
-    },
-    {
-      "epoch": 17.36897001303781,
-      "grad_norm": 1.5494803190231323,
-      "learning_rate": 6.704120603015076e-05,
-      "loss": 5.2629,
-      "step": 33305
-    },
-    {
-      "epoch": 17.36949152542373,
-      "grad_norm": 1.5308970212936401,
-      "learning_rate": 6.704020100502513e-05,
-      "loss": 4.972,
-      "step": 33306
-    },
-    {
-      "epoch": 17.37001303780965,
-      "grad_norm": 1.4289151430130005,
-      "learning_rate": 6.703919597989951e-05,
-      "loss": 5.7106,
-      "step": 33307
-    },
-    {
-      "epoch": 17.37053455019557,
-      "grad_norm": 1.5010827779769897,
-      "learning_rate": 6.703819095477388e-05,
-      "loss": 5.2,
-      "step": 33308
-    },
-    {
-      "epoch": 17.371056062581488,
-      "grad_norm": 1.6940085887908936,
-      "learning_rate": 6.703718592964825e-05,
-      "loss": 4.7772,
-      "step": 33309
-    },
-    {
-      "epoch": 17.371577574967404,
-      "grad_norm": 1.5264337062835693,
-      "learning_rate": 6.703618090452262e-05,
-      "loss": 4.7844,
-      "step": 33310
-    },
-    {
-      "epoch": 17.372099087353323,
-      "grad_norm": 1.49103844165802,
-      "learning_rate": 6.703517587939698e-05,
-      "loss": 5.3426,
-      "step": 33311
-    },
-    {
-      "epoch": 17.372620599739243,
-      "grad_norm": 1.5894649028778076,
-      "learning_rate": 6.703417085427136e-05,
-      "loss": 5.1003,
-      "step": 33312
-    },
-    {
-      "epoch": 17.373142112125162,
-      "grad_norm": 1.5057048797607422,
-      "learning_rate": 6.703316582914572e-05,
-      "loss": 5.8182,
-      "step": 33313
-    },
-    {
-      "epoch": 17.37366362451108,
-      "grad_norm": 1.402477741241455,
-      "learning_rate": 6.70321608040201e-05,
-      "loss": 5.3998,
-      "step": 33314
-    },
-    {
-      "epoch": 17.374185136897,
-      "grad_norm": 1.494332194328308,
-      "learning_rate": 6.703115577889447e-05,
-      "loss": 5.4828,
-      "step": 33315
-    },
-    {
-      "epoch": 17.37470664928292,
-      "grad_norm": 1.3700257539749146,
-      "learning_rate": 6.703015075376884e-05,
-      "loss": 5.5576,
-      "step": 33316
-    },
-    {
-      "epoch": 17.37522816166884,
-      "grad_norm": 1.5250554084777832,
-      "learning_rate": 6.702914572864322e-05,
-      "loss": 5.2957,
-      "step": 33317
-    },
-    {
-      "epoch": 17.37574967405476,
-      "grad_norm": 1.5218008756637573,
-      "learning_rate": 6.70281407035176e-05,
-      "loss": 4.9751,
-      "step": 33318
-    },
-    {
-      "epoch": 17.37627118644068,
-      "grad_norm": 1.5109528303146362,
-      "learning_rate": 6.702713567839196e-05,
-      "loss": 5.1179,
-      "step": 33319
-    },
-    {
-      "epoch": 17.3767926988266,
-      "grad_norm": 1.6224979162216187,
-      "learning_rate": 6.702613065326634e-05,
-      "loss": 4.7204,
-      "step": 33320
-    },
-    {
-      "epoch": 17.377314211212518,
-      "grad_norm": 1.4317749738693237,
-      "learning_rate": 6.70251256281407e-05,
-      "loss": 5.4391,
-      "step": 33321
-    },
-    {
-      "epoch": 17.377835723598434,
-      "grad_norm": 1.350717306137085,
-      "learning_rate": 6.702412060301508e-05,
-      "loss": 5.4098,
-      "step": 33322
-    },
-    {
-      "epoch": 17.378357235984353,
-      "grad_norm": 1.5577723979949951,
-      "learning_rate": 6.702311557788945e-05,
-      "loss": 5.2119,
-      "step": 33323
-    },
-    {
-      "epoch": 17.378878748370273,
-      "grad_norm": 1.5646779537200928,
-      "learning_rate": 6.702211055276381e-05,
-      "loss": 5.4156,
-      "step": 33324
-    },
-    {
-      "epoch": 17.379400260756192,
-      "grad_norm": 1.5695018768310547,
-      "learning_rate": 6.702110552763819e-05,
-      "loss": 5.178,
-      "step": 33325
-    },
-    {
-      "epoch": 17.37992177314211,
-      "grad_norm": 1.4942240715026855,
-      "learning_rate": 6.702010050251257e-05,
-      "loss": 4.8478,
-      "step": 33326
-    },
-    {
-      "epoch": 17.38044328552803,
-      "grad_norm": 1.5628708600997925,
-      "learning_rate": 6.701909547738695e-05,
-      "loss": 5.6197,
-      "step": 33327
-    },
-    {
-      "epoch": 17.38096479791395,
-      "grad_norm": 1.4393888711929321,
-      "learning_rate": 6.701809045226131e-05,
-      "loss": 5.1081,
-      "step": 33328
-    },
-    {
-      "epoch": 17.38148631029987,
-      "grad_norm": 1.569771409034729,
-      "learning_rate": 6.701708542713569e-05,
-      "loss": 5.3521,
-      "step": 33329
-    },
-    {
-      "epoch": 17.38200782268579,
-      "grad_norm": 1.3715975284576416,
-      "learning_rate": 6.701608040201005e-05,
-      "loss": 5.2551,
-      "step": 33330
-    },
-    {
-      "epoch": 17.38252933507171,
-      "grad_norm": 1.3442103862762451,
-      "learning_rate": 6.701507537688443e-05,
-      "loss": 4.8815,
-      "step": 33331
-    },
-    {
-      "epoch": 17.38305084745763,
-      "grad_norm": 1.5002572536468506,
-      "learning_rate": 6.70140703517588e-05,
-      "loss": 5.2933,
-      "step": 33332
-    },
-    {
-      "epoch": 17.383572359843548,
-      "grad_norm": 1.4916445016860962,
-      "learning_rate": 6.701306532663317e-05,
-      "loss": 5.4251,
-      "step": 33333
-    },
-    {
-      "epoch": 17.384093872229464,
-      "grad_norm": 1.4967323541641235,
-      "learning_rate": 6.701206030150754e-05,
-      "loss": 5.3184,
-      "step": 33334
-    },
-    {
-      "epoch": 17.384615384615383,
-      "grad_norm": 1.4301798343658447,
-      "learning_rate": 6.701105527638191e-05,
-      "loss": 5.6372,
-      "step": 33335
-    },
-    {
-      "epoch": 17.385136897001303,
-      "grad_norm": 1.597723364830017,
-      "learning_rate": 6.701005025125628e-05,
-      "loss": 5.1683,
-      "step": 33336
-    },
-    {
-      "epoch": 17.385658409387222,
-      "grad_norm": 1.5463420152664185,
-      "learning_rate": 6.700904522613066e-05,
-      "loss": 5.1107,
-      "step": 33337
-    },
-    {
-      "epoch": 17.38617992177314,
-      "grad_norm": 1.4795109033584595,
-      "learning_rate": 6.700804020100503e-05,
-      "loss": 5.2401,
-      "step": 33338
-    },
-    {
-      "epoch": 17.38670143415906,
-      "grad_norm": 1.5979831218719482,
-      "learning_rate": 6.70070351758794e-05,
-      "loss": 5.147,
-      "step": 33339
-    },
-    {
-      "epoch": 17.38722294654498,
-      "grad_norm": 1.446309208869934,
-      "learning_rate": 6.700603015075378e-05,
-      "loss": 4.321,
-      "step": 33340
-    },
-    {
-      "epoch": 17.3877444589309,
-      "grad_norm": 1.4715323448181152,
-      "learning_rate": 6.700502512562814e-05,
-      "loss": 5.6124,
-      "step": 33341
-    },
-    {
-      "epoch": 17.38826597131682,
-      "grad_norm": 1.5259476900100708,
-      "learning_rate": 6.700402010050252e-05,
-      "loss": 5.5421,
-      "step": 33342
-    },
-    {
-      "epoch": 17.38878748370274,
-      "grad_norm": 1.5545669794082642,
-      "learning_rate": 6.700301507537688e-05,
-      "loss": 5.4783,
-      "step": 33343
-    },
-    {
-      "epoch": 17.38930899608866,
-      "grad_norm": 1.478026270866394,
-      "learning_rate": 6.700201005025126e-05,
-      "loss": 5.4344,
-      "step": 33344
-    },
-    {
-      "epoch": 17.389830508474578,
-      "grad_norm": 1.5149037837982178,
-      "learning_rate": 6.700100502512562e-05,
-      "loss": 5.5141,
-      "step": 33345
-    },
-    {
-      "epoch": 17.390352020860494,
-      "grad_norm": 1.447161316871643,
-      "learning_rate": 6.7e-05,
-      "loss": 5.6699,
-      "step": 33346
-    },
-    {
-      "epoch": 17.390873533246413,
-      "grad_norm": 1.4778060913085938,
-      "learning_rate": 6.699899497487438e-05,
-      "loss": 5.1445,
-      "step": 33347
-    },
-    {
-      "epoch": 17.391395045632333,
-      "grad_norm": 1.4476680755615234,
-      "learning_rate": 6.699798994974876e-05,
-      "loss": 4.9997,
-      "step": 33348
-    },
-    {
-      "epoch": 17.391916558018252,
-      "grad_norm": 1.4786076545715332,
-      "learning_rate": 6.699698492462312e-05,
-      "loss": 5.6836,
-      "step": 33349
-    },
-    {
-      "epoch": 17.39243807040417,
-      "grad_norm": 1.4523953199386597,
-      "learning_rate": 6.699597989949749e-05,
-      "loss": 5.3827,
-      "step": 33350
-    },
-    {
-      "epoch": 17.39295958279009,
-      "grad_norm": 1.5675325393676758,
-      "learning_rate": 6.699497487437186e-05,
-      "loss": 4.7767,
-      "step": 33351
-    },
-    {
-      "epoch": 17.39348109517601,
-      "grad_norm": 1.4686639308929443,
-      "learning_rate": 6.699396984924623e-05,
-      "loss": 5.5464,
-      "step": 33352
-    },
-    {
-      "epoch": 17.39400260756193,
-      "grad_norm": 1.6835224628448486,
-      "learning_rate": 6.69929648241206e-05,
-      "loss": 5.4047,
-      "step": 33353
-    },
-    {
-      "epoch": 17.39452411994785,
-      "grad_norm": 1.407196283340454,
-      "learning_rate": 6.699195979899497e-05,
-      "loss": 5.0106,
-      "step": 33354
-    },
-    {
-      "epoch": 17.39504563233377,
-      "grad_norm": 1.5315074920654297,
-      "learning_rate": 6.699095477386935e-05,
-      "loss": 5.4374,
-      "step": 33355
-    },
-    {
-      "epoch": 17.39556714471969,
-      "grad_norm": 1.5192471742630005,
-      "learning_rate": 6.698994974874371e-05,
-      "loss": 5.2218,
-      "step": 33356
-    },
-    {
-      "epoch": 17.396088657105608,
-      "grad_norm": 1.5894134044647217,
-      "learning_rate": 6.698894472361809e-05,
-      "loss": 4.8999,
-      "step": 33357
-    },
-    {
-      "epoch": 17.396610169491524,
-      "grad_norm": 1.3911851644515991,
-      "learning_rate": 6.698793969849247e-05,
-      "loss": 5.5698,
-      "step": 33358
-    },
-    {
-      "epoch": 17.397131681877443,
-      "grad_norm": 1.4347749948501587,
-      "learning_rate": 6.698693467336685e-05,
-      "loss": 5.5104,
-      "step": 33359
-    },
-    {
-      "epoch": 17.397653194263363,
-      "grad_norm": 1.5029618740081787,
-      "learning_rate": 6.698592964824121e-05,
-      "loss": 5.3089,
-      "step": 33360
-    },
-    {
-      "epoch": 17.398174706649282,
-      "grad_norm": 1.4655876159667969,
-      "learning_rate": 6.698492462311559e-05,
-      "loss": 5.265,
-      "step": 33361
-    },
-    {
-      "epoch": 17.3986962190352,
-      "grad_norm": 1.3924915790557861,
-      "learning_rate": 6.698391959798995e-05,
-      "loss": 5.5763,
-      "step": 33362
-    },
-    {
-      "epoch": 17.39921773142112,
-      "grad_norm": 1.4144177436828613,
-      "learning_rate": 6.698291457286433e-05,
-      "loss": 5.1312,
-      "step": 33363
-    },
-    {
-      "epoch": 17.39973924380704,
-      "grad_norm": 1.4794998168945312,
-      "learning_rate": 6.69819095477387e-05,
-      "loss": 5.0788,
-      "step": 33364
-    },
-    {
-      "epoch": 17.40026075619296,
-      "grad_norm": 1.3483731746673584,
-      "learning_rate": 6.698090452261306e-05,
-      "loss": 5.4778,
-      "step": 33365
-    },
-    {
-      "epoch": 17.40078226857888,
-      "grad_norm": 1.5851805210113525,
-      "learning_rate": 6.697989949748744e-05,
-      "loss": 5.1411,
-      "step": 33366
-    },
-    {
-      "epoch": 17.4013037809648,
-      "grad_norm": 1.439660668373108,
-      "learning_rate": 6.697889447236181e-05,
-      "loss": 5.359,
-      "step": 33367
-    },
-    {
-      "epoch": 17.40182529335072,
-      "grad_norm": 1.4661223888397217,
-      "learning_rate": 6.697788944723619e-05,
-      "loss": 5.4059,
-      "step": 33368
-    },
-    {
-      "epoch": 17.402346805736638,
-      "grad_norm": 1.4442590475082397,
-      "learning_rate": 6.697688442211056e-05,
-      "loss": 5.5956,
-      "step": 33369
-    },
-    {
-      "epoch": 17.402868318122554,
-      "grad_norm": 1.535129189491272,
-      "learning_rate": 6.697587939698493e-05,
-      "loss": 4.8672,
-      "step": 33370
-    },
-    {
-      "epoch": 17.403389830508473,
-      "grad_norm": 1.4925034046173096,
-      "learning_rate": 6.69748743718593e-05,
-      "loss": 5.4492,
-      "step": 33371
-    },
-    {
-      "epoch": 17.403911342894393,
-      "grad_norm": 1.4940823316574097,
-      "learning_rate": 6.697386934673368e-05,
-      "loss": 5.3661,
-      "step": 33372
-    },
-    {
-      "epoch": 17.404432855280312,
-      "grad_norm": 1.4812257289886475,
-      "learning_rate": 6.697286432160804e-05,
-      "loss": 4.8743,
-      "step": 33373
-    },
-    {
-      "epoch": 17.40495436766623,
-      "grad_norm": 1.5105160474777222,
-      "learning_rate": 6.697185929648242e-05,
-      "loss": 5.4142,
-      "step": 33374
-    },
-    {
-      "epoch": 17.40547588005215,
-      "grad_norm": 1.591681957244873,
-      "learning_rate": 6.697085427135678e-05,
-      "loss": 4.7602,
-      "step": 33375
-    },
-    {
-      "epoch": 17.40599739243807,
-      "grad_norm": 1.4522515535354614,
-      "learning_rate": 6.696984924623116e-05,
-      "loss": 5.5426,
-      "step": 33376
-    },
-    {
-      "epoch": 17.40651890482399,
-      "grad_norm": 1.4685931205749512,
-      "learning_rate": 6.696884422110553e-05,
-      "loss": 5.32,
-      "step": 33377
-    },
-    {
-      "epoch": 17.40704041720991,
-      "grad_norm": 1.304081916809082,
-      "learning_rate": 6.69678391959799e-05,
-      "loss": 5.5398,
-      "step": 33378
-    },
-    {
-      "epoch": 17.40756192959583,
-      "grad_norm": 1.4807417392730713,
-      "learning_rate": 6.696683417085428e-05,
-      "loss": 5.0898,
-      "step": 33379
-    },
-    {
-      "epoch": 17.40808344198175,
-      "grad_norm": 1.4419134855270386,
-      "learning_rate": 6.696582914572865e-05,
-      "loss": 5.1452,
-      "step": 33380
-    },
-    {
-      "epoch": 17.408604954367668,
-      "grad_norm": 1.4522358179092407,
-      "learning_rate": 6.696482412060302e-05,
-      "loss": 5.3972,
-      "step": 33381
-    },
-    {
-      "epoch": 17.409126466753584,
-      "grad_norm": 1.4647465944290161,
-      "learning_rate": 6.696381909547739e-05,
-      "loss": 5.4331,
-      "step": 33382
-    },
-    {
-      "epoch": 17.409647979139503,
-      "grad_norm": 1.506130576133728,
-      "learning_rate": 6.696281407035177e-05,
-      "loss": 5.4074,
-      "step": 33383
-    },
-    {
-      "epoch": 17.410169491525423,
-      "grad_norm": 1.394020438194275,
-      "learning_rate": 6.696180904522613e-05,
-      "loss": 5.7874,
-      "step": 33384
-    },
-    {
-      "epoch": 17.410691003911342,
-      "grad_norm": 1.4645332098007202,
-      "learning_rate": 6.696080402010051e-05,
-      "loss": 5.8053,
-      "step": 33385
-    },
-    {
-      "epoch": 17.41121251629726,
-      "grad_norm": 1.552578330039978,
-      "learning_rate": 6.695979899497487e-05,
-      "loss": 5.3209,
-      "step": 33386
-    },
-    {
-      "epoch": 17.41173402868318,
-      "grad_norm": 1.600856900215149,
-      "learning_rate": 6.695879396984925e-05,
-      "loss": 5.0006,
-      "step": 33387
-    },
-    {
-      "epoch": 17.4122555410691,
-      "grad_norm": 1.506874918937683,
-      "learning_rate": 6.695778894472363e-05,
-      "loss": 5.2559,
-      "step": 33388
-    },
-    {
-      "epoch": 17.41277705345502,
-      "grad_norm": 1.5019475221633911,
-      "learning_rate": 6.6956783919598e-05,
-      "loss": 5.1934,
-      "step": 33389
-    },
-    {
-      "epoch": 17.41329856584094,
-      "grad_norm": 1.4603872299194336,
-      "learning_rate": 6.695577889447237e-05,
-      "loss": 5.6321,
-      "step": 33390
-    },
-    {
-      "epoch": 17.41382007822686,
-      "grad_norm": 1.4812606573104858,
-      "learning_rate": 6.695477386934673e-05,
-      "loss": 5.4891,
-      "step": 33391
-    },
-    {
-      "epoch": 17.41434159061278,
-      "grad_norm": 1.4687554836273193,
-      "learning_rate": 6.695376884422111e-05,
-      "loss": 5.3157,
-      "step": 33392
-    },
-    {
-      "epoch": 17.414863102998694,
-      "grad_norm": 1.5121663808822632,
-      "learning_rate": 6.695276381909548e-05,
-      "loss": 5.5617,
-      "step": 33393
-    },
-    {
-      "epoch": 17.415384615384614,
-      "grad_norm": 1.4370312690734863,
-      "learning_rate": 6.695175879396985e-05,
-      "loss": 5.2216,
-      "step": 33394
-    },
-    {
-      "epoch": 17.415906127770533,
-      "grad_norm": 1.4511982202529907,
-      "learning_rate": 6.695075376884422e-05,
-      "loss": 5.7065,
-      "step": 33395
-    },
-    {
-      "epoch": 17.416427640156453,
-      "grad_norm": 1.5566942691802979,
-      "learning_rate": 6.69497487437186e-05,
-      "loss": 5.2102,
-      "step": 33396
-    },
-    {
-      "epoch": 17.416949152542372,
-      "grad_norm": 1.4614759683609009,
-      "learning_rate": 6.694874371859296e-05,
-      "loss": 5.0203,
-      "step": 33397
-    },
-    {
-      "epoch": 17.41747066492829,
-      "grad_norm": 1.4133148193359375,
-      "learning_rate": 6.694773869346734e-05,
-      "loss": 5.0788,
-      "step": 33398
-    },
-    {
-      "epoch": 17.41799217731421,
-      "grad_norm": 1.5128679275512695,
-      "learning_rate": 6.694673366834172e-05,
-      "loss": 5.0232,
-      "step": 33399
-    },
-    {
-      "epoch": 17.41851368970013,
-      "grad_norm": 1.3955103158950806,
-      "learning_rate": 6.69457286432161e-05,
-      "loss": 5.8089,
-      "step": 33400
-    },
-    {
-      "epoch": 17.41903520208605,
-      "grad_norm": 1.7728441953659058,
-      "learning_rate": 6.694472361809046e-05,
-      "loss": 5.2146,
-      "step": 33401
-    },
-    {
-      "epoch": 17.41955671447197,
-      "grad_norm": 1.5319862365722656,
-      "learning_rate": 6.694371859296484e-05,
-      "loss": 5.1129,
-      "step": 33402
-    },
-    {
-      "epoch": 17.42007822685789,
-      "grad_norm": 1.4230257272720337,
-      "learning_rate": 6.69427135678392e-05,
-      "loss": 5.6501,
-      "step": 33403
-    },
-    {
-      "epoch": 17.42059973924381,
-      "grad_norm": 1.5069741010665894,
-      "learning_rate": 6.694170854271356e-05,
-      "loss": 5.5578,
-      "step": 33404
-    },
-    {
-      "epoch": 17.421121251629724,
-      "grad_norm": 1.4823344945907593,
-      "learning_rate": 6.694070351758794e-05,
-      "loss": 5.3896,
-      "step": 33405
-    },
-    {
-      "epoch": 17.421642764015644,
-      "grad_norm": 1.586643099784851,
-      "learning_rate": 6.69396984924623e-05,
-      "loss": 5.0997,
-      "step": 33406
-    },
-    {
-      "epoch": 17.422164276401563,
-      "grad_norm": 1.5451557636260986,
-      "learning_rate": 6.693869346733668e-05,
-      "loss": 5.0284,
-      "step": 33407
-    },
-    {
-      "epoch": 17.422685788787483,
-      "grad_norm": 1.419773817062378,
-      "learning_rate": 6.693768844221106e-05,
-      "loss": 5.3897,
-      "step": 33408
-    },
-    {
-      "epoch": 17.423207301173402,
-      "grad_norm": 1.4194345474243164,
-      "learning_rate": 6.693668341708544e-05,
-      "loss": 5.5917,
-      "step": 33409
-    },
-    {
-      "epoch": 17.423728813559322,
-      "grad_norm": 1.4225023984909058,
-      "learning_rate": 6.69356783919598e-05,
-      "loss": 5.3747,
-      "step": 33410
-    },
-    {
-      "epoch": 17.42425032594524,
-      "grad_norm": 1.4442819356918335,
-      "learning_rate": 6.693467336683418e-05,
-      "loss": 5.3892,
-      "step": 33411
-    },
-    {
-      "epoch": 17.42477183833116,
-      "grad_norm": 1.4609419107437134,
-      "learning_rate": 6.693366834170855e-05,
-      "loss": 5.4657,
-      "step": 33412
-    },
-    {
-      "epoch": 17.42529335071708,
-      "grad_norm": 1.4340753555297852,
-      "learning_rate": 6.693266331658292e-05,
-      "loss": 5.4272,
-      "step": 33413
-    },
-    {
-      "epoch": 17.425814863103,
-      "grad_norm": 1.461288332939148,
-      "learning_rate": 6.693165829145729e-05,
-      "loss": 5.6586,
-      "step": 33414
-    },
-    {
-      "epoch": 17.42633637548892,
-      "grad_norm": 1.4543256759643555,
-      "learning_rate": 6.693065326633167e-05,
-      "loss": 5.1243,
-      "step": 33415
-    },
-    {
-      "epoch": 17.42685788787484,
-      "grad_norm": 1.591681957244873,
-      "learning_rate": 6.692964824120603e-05,
-      "loss": 5.0296,
-      "step": 33416
-    },
-    {
-      "epoch": 17.427379400260754,
-      "grad_norm": 1.4125231504440308,
-      "learning_rate": 6.69286432160804e-05,
-      "loss": 5.7179,
-      "step": 33417
-    },
-    {
-      "epoch": 17.427900912646674,
-      "grad_norm": 1.5724565982818604,
-      "learning_rate": 6.692763819095477e-05,
-      "loss": 5.0699,
-      "step": 33418
-    },
-    {
-      "epoch": 17.428422425032593,
-      "grad_norm": 1.4777381420135498,
-      "learning_rate": 6.692663316582915e-05,
-      "loss": 5.6745,
-      "step": 33419
-    },
-    {
-      "epoch": 17.428943937418513,
-      "grad_norm": 1.4690351486206055,
-      "learning_rate": 6.692562814070353e-05,
-      "loss": 5.2018,
-      "step": 33420
-    },
-    {
-      "epoch": 17.429465449804432,
-      "grad_norm": 1.5413535833358765,
-      "learning_rate": 6.692462311557789e-05,
-      "loss": 5.3191,
-      "step": 33421
-    },
-    {
-      "epoch": 17.429986962190352,
-      "grad_norm": 1.5026785135269165,
-      "learning_rate": 6.692361809045227e-05,
-      "loss": 5.6279,
-      "step": 33422
-    },
-    {
-      "epoch": 17.43050847457627,
-      "grad_norm": 1.6291320323944092,
-      "learning_rate": 6.692261306532663e-05,
-      "loss": 5.0699,
-      "step": 33423
-    },
-    {
-      "epoch": 17.43102998696219,
-      "grad_norm": 1.5108582973480225,
-      "learning_rate": 6.692160804020101e-05,
-      "loss": 5.2898,
-      "step": 33424
-    },
-    {
-      "epoch": 17.43155149934811,
-      "grad_norm": 1.5346630811691284,
-      "learning_rate": 6.692060301507538e-05,
-      "loss": 5.3551,
-      "step": 33425
-    },
-    {
-      "epoch": 17.43207301173403,
-      "grad_norm": 1.5172280073165894,
-      "learning_rate": 6.691959798994975e-05,
-      "loss": 5.1765,
-      "step": 33426
-    },
-    {
-      "epoch": 17.43259452411995,
-      "grad_norm": 1.5742782354354858,
-      "learning_rate": 6.691859296482412e-05,
-      "loss": 5.4742,
-      "step": 33427
-    },
-    {
-      "epoch": 17.43311603650587,
-      "grad_norm": 1.3925607204437256,
-      "learning_rate": 6.69175879396985e-05,
-      "loss": 5.3325,
-      "step": 33428
-    },
-    {
-      "epoch": 17.433637548891785,
-      "grad_norm": 1.482879877090454,
-      "learning_rate": 6.691658291457287e-05,
-      "loss": 5.4536,
-      "step": 33429
-    },
-    {
-      "epoch": 17.434159061277704,
-      "grad_norm": 1.5765290260314941,
-      "learning_rate": 6.691557788944724e-05,
-      "loss": 5.1975,
-      "step": 33430
-    },
-    {
-      "epoch": 17.434680573663623,
-      "grad_norm": 1.445927619934082,
-      "learning_rate": 6.691457286432162e-05,
-      "loss": 5.328,
-      "step": 33431
-    },
-    {
-      "epoch": 17.435202086049543,
-      "grad_norm": 1.5870797634124756,
-      "learning_rate": 6.691356783919598e-05,
-      "loss": 5.31,
-      "step": 33432
-    },
-    {
-      "epoch": 17.435723598435462,
-      "grad_norm": 1.4691380262374878,
-      "learning_rate": 6.691256281407036e-05,
-      "loss": 5.6373,
-      "step": 33433
-    },
-    {
-      "epoch": 17.436245110821382,
-      "grad_norm": 1.5584250688552856,
-      "learning_rate": 6.691155778894472e-05,
-      "loss": 5.1729,
-      "step": 33434
-    },
-    {
-      "epoch": 17.4367666232073,
-      "grad_norm": 1.406686782836914,
-      "learning_rate": 6.69105527638191e-05,
-      "loss": 5.5716,
-      "step": 33435
-    },
-    {
-      "epoch": 17.43728813559322,
-      "grad_norm": 1.439328908920288,
-      "learning_rate": 6.690954773869346e-05,
-      "loss": 5.349,
-      "step": 33436
-    },
-    {
-      "epoch": 17.43780964797914,
-      "grad_norm": 1.5305980443954468,
-      "learning_rate": 6.690854271356784e-05,
-      "loss": 5.2621,
-      "step": 33437
-    },
-    {
-      "epoch": 17.43833116036506,
-      "grad_norm": 1.4466997385025024,
-      "learning_rate": 6.690753768844221e-05,
-      "loss": 5.5381,
-      "step": 33438
-    },
-    {
-      "epoch": 17.43885267275098,
-      "grad_norm": 1.4293235540390015,
-      "learning_rate": 6.690653266331658e-05,
-      "loss": 5.397,
-      "step": 33439
-    },
-    {
-      "epoch": 17.4393741851369,
-      "grad_norm": 1.3869141340255737,
-      "learning_rate": 6.690552763819096e-05,
-      "loss": 5.0727,
-      "step": 33440
-    },
-    {
-      "epoch": 17.439895697522815,
-      "grad_norm": 1.4625264406204224,
-      "learning_rate": 6.690452261306534e-05,
-      "loss": 4.8833,
-      "step": 33441
-    },
-    {
-      "epoch": 17.440417209908734,
-      "grad_norm": 1.4170868396759033,
-      "learning_rate": 6.69035175879397e-05,
-      "loss": 4.9248,
-      "step": 33442
-    },
-    {
-      "epoch": 17.440938722294653,
-      "grad_norm": 1.4564176797866821,
-      "learning_rate": 6.690251256281407e-05,
-      "loss": 5.3085,
-      "step": 33443
-    },
-    {
-      "epoch": 17.441460234680573,
-      "grad_norm": 1.532454013824463,
-      "learning_rate": 6.690150753768845e-05,
-      "loss": 5.4331,
-      "step": 33444
-    },
-    {
-      "epoch": 17.441981747066492,
-      "grad_norm": 1.485054612159729,
-      "learning_rate": 6.690050251256281e-05,
-      "loss": 5.3198,
-      "step": 33445
-    },
-    {
-      "epoch": 17.442503259452412,
-      "grad_norm": 1.4993305206298828,
-      "learning_rate": 6.689949748743719e-05,
-      "loss": 5.3497,
-      "step": 33446
-    },
-    {
-      "epoch": 17.44302477183833,
-      "grad_norm": 1.559287428855896,
-      "learning_rate": 6.689849246231155e-05,
-      "loss": 5.0438,
-      "step": 33447
-    },
-    {
-      "epoch": 17.44354628422425,
-      "grad_norm": 1.4844106435775757,
-      "learning_rate": 6.689748743718593e-05,
-      "loss": 5.2888,
-      "step": 33448
-    },
-    {
-      "epoch": 17.44406779661017,
-      "grad_norm": 1.450250267982483,
-      "learning_rate": 6.68964824120603e-05,
-      "loss": 5.3924,
-      "step": 33449
-    },
-    {
-      "epoch": 17.44458930899609,
-      "grad_norm": 1.5296880006790161,
-      "learning_rate": 6.689547738693467e-05,
-      "loss": 5.2523,
-      "step": 33450
-    },
-    {
-      "epoch": 17.44511082138201,
-      "grad_norm": 1.5038554668426514,
-      "learning_rate": 6.689447236180905e-05,
-      "loss": 5.2753,
-      "step": 33451
-    },
-    {
-      "epoch": 17.44563233376793,
-      "grad_norm": 1.6183041334152222,
-      "learning_rate": 6.689346733668343e-05,
-      "loss": 5.3079,
-      "step": 33452
-    },
-    {
-      "epoch": 17.446153846153845,
-      "grad_norm": 1.4974493980407715,
-      "learning_rate": 6.689246231155779e-05,
-      "loss": 5.1431,
-      "step": 33453
-    },
-    {
-      "epoch": 17.446675358539764,
-      "grad_norm": 1.4454435110092163,
-      "learning_rate": 6.689145728643217e-05,
-      "loss": 5.6045,
-      "step": 33454
-    },
-    {
-      "epoch": 17.447196870925683,
-      "grad_norm": 1.4477466344833374,
-      "learning_rate": 6.689045226130653e-05,
-      "loss": 5.7088,
-      "step": 33455
-    },
-    {
-      "epoch": 17.447718383311603,
-      "grad_norm": 1.5472747087478638,
-      "learning_rate": 6.688944723618091e-05,
-      "loss": 5.3334,
-      "step": 33456
-    },
-    {
-      "epoch": 17.448239895697522,
-      "grad_norm": 1.5871977806091309,
-      "learning_rate": 6.688844221105528e-05,
-      "loss": 5.0604,
-      "step": 33457
-    },
-    {
-      "epoch": 17.448761408083442,
-      "grad_norm": 1.3523386716842651,
-      "learning_rate": 6.688743718592964e-05,
-      "loss": 5.6409,
-      "step": 33458
-    },
-    {
-      "epoch": 17.44928292046936,
-      "grad_norm": 1.341492772102356,
-      "learning_rate": 6.688643216080402e-05,
-      "loss": 5.6424,
-      "step": 33459
-    },
-    {
-      "epoch": 17.44980443285528,
-      "grad_norm": 1.5359411239624023,
-      "learning_rate": 6.68854271356784e-05,
-      "loss": 5.2636,
-      "step": 33460
-    },
-    {
-      "epoch": 17.4503259452412,
-      "grad_norm": 1.4575995206832886,
-      "learning_rate": 6.688442211055277e-05,
-      "loss": 5.5405,
-      "step": 33461
-    },
-    {
-      "epoch": 17.45084745762712,
-      "grad_norm": 1.4974840879440308,
-      "learning_rate": 6.688341708542714e-05,
-      "loss": 4.8354,
-      "step": 33462
-    },
-    {
-      "epoch": 17.45136897001304,
-      "grad_norm": 1.9118720293045044,
-      "learning_rate": 6.688241206030152e-05,
-      "loss": 5.0692,
-      "step": 33463
-    },
-    {
-      "epoch": 17.45189048239896,
-      "grad_norm": 1.517846941947937,
-      "learning_rate": 6.688140703517588e-05,
-      "loss": 5.4709,
-      "step": 33464
-    },
-    {
-      "epoch": 17.452411994784875,
-      "grad_norm": 1.5454707145690918,
-      "learning_rate": 6.688040201005026e-05,
-      "loss": 5.0994,
-      "step": 33465
-    },
-    {
-      "epoch": 17.452933507170794,
-      "grad_norm": 1.5073370933532715,
-      "learning_rate": 6.687939698492462e-05,
-      "loss": 5.5712,
-      "step": 33466
-    },
-    {
-      "epoch": 17.453455019556714,
-      "grad_norm": 1.469252109527588,
-      "learning_rate": 6.6878391959799e-05,
-      "loss": 5.1803,
-      "step": 33467
-    },
-    {
-      "epoch": 17.453976531942633,
-      "grad_norm": 1.681057095527649,
-      "learning_rate": 6.687738693467337e-05,
-      "loss": 4.7956,
-      "step": 33468
-    },
-    {
-      "epoch": 17.454498044328552,
-      "grad_norm": 1.5131666660308838,
-      "learning_rate": 6.687638190954774e-05,
-      "loss": 4.9839,
-      "step": 33469
-    },
-    {
-      "epoch": 17.455019556714472,
-      "grad_norm": 1.4671692848205566,
-      "learning_rate": 6.687537688442211e-05,
-      "loss": 5.2324,
-      "step": 33470
-    },
-    {
-      "epoch": 17.45554106910039,
-      "grad_norm": 1.4797028303146362,
-      "learning_rate": 6.687437185929649e-05,
-      "loss": 5.7024,
-      "step": 33471
-    },
-    {
-      "epoch": 17.45606258148631,
-      "grad_norm": 1.4669514894485474,
-      "learning_rate": 6.687336683417086e-05,
-      "loss": 5.7934,
-      "step": 33472
-    },
-    {
-      "epoch": 17.45658409387223,
-      "grad_norm": 1.63804292678833,
-      "learning_rate": 6.687236180904523e-05,
-      "loss": 5.2055,
-      "step": 33473
-    },
-    {
-      "epoch": 17.45710560625815,
-      "grad_norm": 1.4778004884719849,
-      "learning_rate": 6.68713567839196e-05,
-      "loss": 5.1445,
-      "step": 33474
-    },
-    {
-      "epoch": 17.45762711864407,
-      "grad_norm": 1.4870740175247192,
-      "learning_rate": 6.687035175879397e-05,
-      "loss": 5.5789,
-      "step": 33475
-    },
-    {
-      "epoch": 17.45814863102999,
-      "grad_norm": 1.5420019626617432,
-      "learning_rate": 6.686934673366835e-05,
-      "loss": 4.8956,
-      "step": 33476
-    },
-    {
-      "epoch": 17.458670143415905,
-      "grad_norm": 1.5009231567382812,
-      "learning_rate": 6.686834170854271e-05,
-      "loss": 5.6337,
-      "step": 33477
-    },
-    {
-      "epoch": 17.459191655801824,
-      "grad_norm": 1.476914882659912,
-      "learning_rate": 6.686733668341709e-05,
-      "loss": 5.1692,
-      "step": 33478
-    },
-    {
-      "epoch": 17.459713168187744,
-      "grad_norm": 1.3784942626953125,
-      "learning_rate": 6.686633165829145e-05,
-      "loss": 5.3096,
-      "step": 33479
-    },
-    {
-      "epoch": 17.460234680573663,
-      "grad_norm": 1.4420616626739502,
-      "learning_rate": 6.686532663316583e-05,
-      "loss": 5.5707,
-      "step": 33480
-    },
-    {
-      "epoch": 17.460756192959582,
-      "grad_norm": 1.438058853149414,
-      "learning_rate": 6.686432160804021e-05,
-      "loss": 5.1297,
-      "step": 33481
-    },
-    {
-      "epoch": 17.461277705345502,
-      "grad_norm": 1.5284255743026733,
-      "learning_rate": 6.686331658291459e-05,
-      "loss": 5.0355,
-      "step": 33482
-    },
-    {
-      "epoch": 17.46179921773142,
-      "grad_norm": 1.6728577613830566,
-      "learning_rate": 6.686231155778895e-05,
-      "loss": 4.6254,
-      "step": 33483
-    },
-    {
-      "epoch": 17.46232073011734,
-      "grad_norm": 1.3940612077713013,
-      "learning_rate": 6.686130653266332e-05,
-      "loss": 5.6026,
-      "step": 33484
-    },
-    {
-      "epoch": 17.46284224250326,
-      "grad_norm": 1.4673092365264893,
-      "learning_rate": 6.68603015075377e-05,
-      "loss": 5.7139,
-      "step": 33485
-    },
-    {
-      "epoch": 17.46336375488918,
-      "grad_norm": 1.4897104501724243,
-      "learning_rate": 6.685929648241206e-05,
-      "loss": 5.247,
-      "step": 33486
-    },
-    {
-      "epoch": 17.4638852672751,
-      "grad_norm": 1.565875768661499,
-      "learning_rate": 6.685829145728644e-05,
-      "loss": 5.4224,
-      "step": 33487
-    },
-    {
-      "epoch": 17.46440677966102,
-      "grad_norm": 1.4719622135162354,
-      "learning_rate": 6.68572864321608e-05,
-      "loss": 5.4058,
-      "step": 33488
-    },
-    {
-      "epoch": 17.464928292046935,
-      "grad_norm": 1.4342900514602661,
-      "learning_rate": 6.685628140703518e-05,
-      "loss": 5.3396,
-      "step": 33489
-    },
-    {
-      "epoch": 17.465449804432854,
-      "grad_norm": 1.4089603424072266,
-      "learning_rate": 6.685527638190954e-05,
-      "loss": 5.8537,
-      "step": 33490
-    },
-    {
-      "epoch": 17.465971316818774,
-      "grad_norm": 1.4695247411727905,
-      "learning_rate": 6.685427135678392e-05,
-      "loss": 4.9305,
-      "step": 33491
-    },
-    {
-      "epoch": 17.466492829204693,
-      "grad_norm": 1.419036865234375,
-      "learning_rate": 6.68532663316583e-05,
-      "loss": 5.3849,
-      "step": 33492
-    },
-    {
-      "epoch": 17.467014341590613,
-      "grad_norm": 1.4535226821899414,
-      "learning_rate": 6.685226130653268e-05,
-      "loss": 5.4034,
-      "step": 33493
-    },
-    {
-      "epoch": 17.467535853976532,
-      "grad_norm": 1.4767887592315674,
-      "learning_rate": 6.685125628140704e-05,
-      "loss": 5.3438,
-      "step": 33494
-    },
-    {
-      "epoch": 17.46805736636245,
-      "grad_norm": 1.4260567426681519,
-      "learning_rate": 6.685025125628142e-05,
-      "loss": 5.5399,
-      "step": 33495
-    },
-    {
-      "epoch": 17.46857887874837,
-      "grad_norm": 1.4666277170181274,
-      "learning_rate": 6.684924623115578e-05,
-      "loss": 5.426,
-      "step": 33496
-    },
-    {
-      "epoch": 17.46910039113429,
-      "grad_norm": 1.5771691799163818,
-      "learning_rate": 6.684824120603015e-05,
-      "loss": 4.918,
-      "step": 33497
-    },
-    {
-      "epoch": 17.46962190352021,
-      "grad_norm": 1.535502552986145,
-      "learning_rate": 6.684723618090452e-05,
-      "loss": 5.0628,
-      "step": 33498
-    },
-    {
-      "epoch": 17.47014341590613,
-      "grad_norm": 1.5861425399780273,
-      "learning_rate": 6.684623115577889e-05,
-      "loss": 5.1203,
-      "step": 33499
-    },
-    {
-      "epoch": 17.470664928292045,
-      "grad_norm": 1.508304238319397,
-      "learning_rate": 6.684522613065327e-05,
-      "loss": 5.1636,
-      "step": 33500
-    },
-    {
-      "epoch": 17.471186440677965,
-      "grad_norm": 1.519816517829895,
-      "learning_rate": 6.684422110552764e-05,
-      "loss": 5.4293,
-      "step": 33501
-    },
-    {
-      "epoch": 17.471707953063884,
-      "grad_norm": 1.4848240613937378,
-      "learning_rate": 6.684321608040202e-05,
-      "loss": 5.6302,
-      "step": 33502
-    },
-    {
-      "epoch": 17.472229465449804,
-      "grad_norm": 1.5345518589019775,
-      "learning_rate": 6.684221105527639e-05,
-      "loss": 4.8362,
-      "step": 33503
-    },
-    {
-      "epoch": 17.472750977835723,
-      "grad_norm": 1.4793061017990112,
-      "learning_rate": 6.684120603015076e-05,
-      "loss": 5.3737,
-      "step": 33504
-    },
-    {
-      "epoch": 17.473272490221643,
-      "grad_norm": 1.3838433027267456,
-      "learning_rate": 6.684020100502513e-05,
-      "loss": 5.5717,
-      "step": 33505
-    },
-    {
-      "epoch": 17.473794002607562,
-      "grad_norm": 1.5499197244644165,
-      "learning_rate": 6.68391959798995e-05,
-      "loss": 5.3729,
-      "step": 33506
-    },
-    {
-      "epoch": 17.47431551499348,
-      "grad_norm": 1.5816171169281006,
-      "learning_rate": 6.683819095477387e-05,
-      "loss": 5.6421,
-      "step": 33507
-    },
-    {
-      "epoch": 17.4748370273794,
-      "grad_norm": 1.4248586893081665,
-      "learning_rate": 6.683718592964825e-05,
-      "loss": 5.1068,
-      "step": 33508
-    },
-    {
-      "epoch": 17.47535853976532,
-      "grad_norm": 1.4413464069366455,
-      "learning_rate": 6.683618090452261e-05,
-      "loss": 5.6054,
-      "step": 33509
-    },
-    {
-      "epoch": 17.47588005215124,
-      "grad_norm": 1.478460669517517,
-      "learning_rate": 6.683517587939698e-05,
-      "loss": 5.5342,
-      "step": 33510
-    },
-    {
-      "epoch": 17.47640156453716,
-      "grad_norm": 1.5583748817443848,
-      "learning_rate": 6.683417085427135e-05,
-      "loss": 4.9795,
-      "step": 33511
-    },
-    {
-      "epoch": 17.476923076923075,
-      "grad_norm": 1.4952244758605957,
-      "learning_rate": 6.683316582914573e-05,
-      "loss": 5.6182,
-      "step": 33512
-    },
-    {
-      "epoch": 17.477444589308995,
-      "grad_norm": 1.4532502889633179,
-      "learning_rate": 6.683216080402011e-05,
-      "loss": 5.4582,
-      "step": 33513
-    },
-    {
-      "epoch": 17.477966101694914,
-      "grad_norm": 1.5680408477783203,
-      "learning_rate": 6.683115577889447e-05,
-      "loss": 4.8428,
-      "step": 33514
-    },
-    {
-      "epoch": 17.478487614080834,
-      "grad_norm": 1.3751276731491089,
-      "learning_rate": 6.683015075376885e-05,
-      "loss": 5.697,
-      "step": 33515
-    },
-    {
-      "epoch": 17.479009126466753,
-      "grad_norm": 1.539723515510559,
-      "learning_rate": 6.682914572864322e-05,
-      "loss": 5.222,
-      "step": 33516
-    },
-    {
-      "epoch": 17.479530638852673,
-      "grad_norm": 1.5359681844711304,
-      "learning_rate": 6.68281407035176e-05,
-      "loss": 5.4974,
-      "step": 33517
-    },
-    {
-      "epoch": 17.480052151238592,
-      "grad_norm": 1.5961744785308838,
-      "learning_rate": 6.682713567839196e-05,
-      "loss": 5.2092,
-      "step": 33518
-    },
-    {
-      "epoch": 17.48057366362451,
-      "grad_norm": 1.51680326461792,
-      "learning_rate": 6.682613065326634e-05,
-      "loss": 5.4328,
-      "step": 33519
-    },
-    {
-      "epoch": 17.48109517601043,
-      "grad_norm": 1.49820077419281,
-      "learning_rate": 6.68251256281407e-05,
-      "loss": 5.3644,
-      "step": 33520
-    },
-    {
-      "epoch": 17.48161668839635,
-      "grad_norm": 1.5129673480987549,
-      "learning_rate": 6.682412060301508e-05,
-      "loss": 5.4348,
-      "step": 33521
-    },
-    {
-      "epoch": 17.48213820078227,
-      "grad_norm": 1.564743161201477,
-      "learning_rate": 6.682311557788946e-05,
-      "loss": 5.2788,
-      "step": 33522
-    },
-    {
-      "epoch": 17.48265971316819,
-      "grad_norm": 1.4221248626708984,
-      "learning_rate": 6.682211055276382e-05,
-      "loss": 5.3871,
-      "step": 33523
-    },
-    {
-      "epoch": 17.483181225554105,
-      "grad_norm": 1.522007703781128,
-      "learning_rate": 6.68211055276382e-05,
-      "loss": 5.6406,
-      "step": 33524
-    },
-    {
-      "epoch": 17.483702737940025,
-      "grad_norm": 1.487996220588684,
-      "learning_rate": 6.682010050251256e-05,
-      "loss": 5.4128,
-      "step": 33525
-    },
-    {
-      "epoch": 17.484224250325944,
-      "grad_norm": 1.5830345153808594,
-      "learning_rate": 6.681909547738694e-05,
-      "loss": 4.6409,
-      "step": 33526
-    },
-    {
-      "epoch": 17.484745762711864,
-      "grad_norm": 1.43160879611969,
-      "learning_rate": 6.68180904522613e-05,
-      "loss": 5.6553,
-      "step": 33527
-    },
-    {
-      "epoch": 17.485267275097783,
-      "grad_norm": 1.5266607999801636,
-      "learning_rate": 6.681708542713568e-05,
-      "loss": 5.1455,
-      "step": 33528
-    },
-    {
-      "epoch": 17.485788787483703,
-      "grad_norm": 1.4455790519714355,
-      "learning_rate": 6.681608040201005e-05,
-      "loss": 5.4504,
-      "step": 33529
-    },
-    {
-      "epoch": 17.486310299869622,
-      "grad_norm": 1.3521860837936401,
-      "learning_rate": 6.681507537688442e-05,
-      "loss": 5.2934,
-      "step": 33530
-    },
-    {
-      "epoch": 17.48683181225554,
-      "grad_norm": 1.3788613080978394,
-      "learning_rate": 6.681407035175879e-05,
-      "loss": 5.4867,
-      "step": 33531
-    },
-    {
-      "epoch": 17.48735332464146,
-      "grad_norm": 1.4932641983032227,
-      "learning_rate": 6.681306532663317e-05,
-      "loss": 4.8435,
-      "step": 33532
-    },
-    {
-      "epoch": 17.48787483702738,
-      "grad_norm": 1.4172641038894653,
-      "learning_rate": 6.681206030150754e-05,
-      "loss": 5.6636,
-      "step": 33533
-    },
-    {
-      "epoch": 17.4883963494133,
-      "grad_norm": 1.3882536888122559,
-      "learning_rate": 6.681105527638192e-05,
-      "loss": 5.7971,
-      "step": 33534
-    },
-    {
-      "epoch": 17.48891786179922,
-      "grad_norm": 1.548936367034912,
-      "learning_rate": 6.681005025125629e-05,
-      "loss": 4.9391,
-      "step": 33535
-    },
-    {
-      "epoch": 17.489439374185135,
-      "grad_norm": 1.4632668495178223,
-      "learning_rate": 6.680904522613065e-05,
-      "loss": 5.7876,
-      "step": 33536
-    },
-    {
-      "epoch": 17.489960886571055,
-      "grad_norm": 1.5263110399246216,
-      "learning_rate": 6.680804020100503e-05,
-      "loss": 4.9114,
-      "step": 33537
-    },
-    {
-      "epoch": 17.490482398956974,
-      "grad_norm": 1.5349773168563843,
-      "learning_rate": 6.68070351758794e-05,
-      "loss": 5.3014,
-      "step": 33538
-    },
-    {
-      "epoch": 17.491003911342894,
-      "grad_norm": 1.4554063081741333,
-      "learning_rate": 6.680603015075377e-05,
-      "loss": 5.4729,
-      "step": 33539
-    },
-    {
-      "epoch": 17.491525423728813,
-      "grad_norm": 1.4763081073760986,
-      "learning_rate": 6.680502512562814e-05,
-      "loss": 5.5152,
-      "step": 33540
-    },
-    {
-      "epoch": 17.492046936114733,
-      "grad_norm": 1.452770709991455,
-      "learning_rate": 6.680402010050251e-05,
-      "loss": 5.4009,
-      "step": 33541
-    },
-    {
-      "epoch": 17.492568448500652,
-      "grad_norm": 1.3868911266326904,
-      "learning_rate": 6.680301507537689e-05,
-      "loss": 5.4403,
-      "step": 33542
-    },
-    {
-      "epoch": 17.49308996088657,
-      "grad_norm": 1.515199065208435,
-      "learning_rate": 6.680201005025127e-05,
-      "loss": 5.1869,
-      "step": 33543
-    },
-    {
-      "epoch": 17.49361147327249,
-      "grad_norm": 1.5757421255111694,
-      "learning_rate": 6.680100502512563e-05,
-      "loss": 5.6033,
-      "step": 33544
-    },
-    {
-      "epoch": 17.49413298565841,
-      "grad_norm": 1.5854285955429077,
-      "learning_rate": 6.680000000000001e-05,
-      "loss": 5.023,
-      "step": 33545
-    },
-    {
-      "epoch": 17.49465449804433,
-      "grad_norm": 1.5450571775436401,
-      "learning_rate": 6.679899497487438e-05,
-      "loss": 4.7958,
-      "step": 33546
-    },
-    {
-      "epoch": 17.49517601043025,
-      "grad_norm": 1.4891716241836548,
-      "learning_rate": 6.679798994974875e-05,
-      "loss": 5.7182,
-      "step": 33547
-    },
-    {
-      "epoch": 17.495697522816165,
-      "grad_norm": 1.582636833190918,
-      "learning_rate": 6.679698492462312e-05,
-      "loss": 5.2431,
-      "step": 33548
-    },
-    {
-      "epoch": 17.496219035202085,
-      "grad_norm": 1.514701008796692,
-      "learning_rate": 6.67959798994975e-05,
-      "loss": 5.7405,
-      "step": 33549
-    },
-    {
-      "epoch": 17.496740547588004,
-      "grad_norm": 1.5487916469573975,
-      "learning_rate": 6.679497487437186e-05,
-      "loss": 5.2309,
-      "step": 33550
-    },
-    {
-      "epoch": 17.497262059973924,
-      "grad_norm": 1.4171514511108398,
-      "learning_rate": 6.679396984924622e-05,
-      "loss": 4.8839,
-      "step": 33551
-    },
-    {
-      "epoch": 17.497783572359843,
-      "grad_norm": 1.3275659084320068,
-      "learning_rate": 6.67929648241206e-05,
-      "loss": 5.6371,
-      "step": 33552
-    },
-    {
-      "epoch": 17.498305084745763,
-      "grad_norm": 1.5101946592330933,
-      "learning_rate": 6.679195979899498e-05,
-      "loss": 5.2267,
-      "step": 33553
-    },
-    {
-      "epoch": 17.498826597131682,
-      "grad_norm": 1.4753997325897217,
-      "learning_rate": 6.679095477386936e-05,
-      "loss": 5.5606,
-      "step": 33554
-    },
-    {
-      "epoch": 17.4993481095176,
-      "grad_norm": 1.5931370258331299,
-      "learning_rate": 6.678994974874372e-05,
-      "loss": 5.5364,
-      "step": 33555
-    },
-    {
-      "epoch": 17.49986962190352,
-      "grad_norm": 1.5078400373458862,
-      "learning_rate": 6.67889447236181e-05,
-      "loss": 5.5726,
-      "step": 33556
-    },
-    {
-      "epoch": 17.50039113428944,
-      "grad_norm": 1.7046397924423218,
-      "learning_rate": 6.678793969849246e-05,
-      "loss": 4.4278,
-      "step": 33557
-    },
-    {
-      "epoch": 17.50091264667536,
-      "grad_norm": 1.5490761995315552,
-      "learning_rate": 6.678693467336684e-05,
-      "loss": 4.4357,
-      "step": 33558
-    },
-    {
-      "epoch": 17.50143415906128,
-      "grad_norm": 1.5909647941589355,
-      "learning_rate": 6.67859296482412e-05,
-      "loss": 5.1081,
-      "step": 33559
-    },
-    {
-      "epoch": 17.501955671447195,
-      "grad_norm": 1.5092285871505737,
-      "learning_rate": 6.678492462311558e-05,
-      "loss": 5.7141,
-      "step": 33560
-    },
-    {
-      "epoch": 17.502477183833115,
-      "grad_norm": 1.5692945718765259,
-      "learning_rate": 6.678391959798995e-05,
-      "loss": 5.025,
-      "step": 33561
-    },
-    {
-      "epoch": 17.502998696219034,
-      "grad_norm": 1.6386104822158813,
-      "learning_rate": 6.678291457286433e-05,
-      "loss": 4.9107,
-      "step": 33562
-    },
-    {
-      "epoch": 17.503520208604954,
-      "grad_norm": 1.6459720134735107,
-      "learning_rate": 6.67819095477387e-05,
-      "loss": 5.0206,
-      "step": 33563
-    },
-    {
-      "epoch": 17.504041720990873,
-      "grad_norm": 1.469666600227356,
-      "learning_rate": 6.678090452261307e-05,
-      "loss": 5.5328,
-      "step": 33564
-    },
-    {
-      "epoch": 17.504563233376793,
-      "grad_norm": 1.498950481414795,
-      "learning_rate": 6.677989949748745e-05,
-      "loss": 5.468,
-      "step": 33565
-    },
-    {
-      "epoch": 17.505084745762712,
-      "grad_norm": 1.4868378639221191,
-      "learning_rate": 6.677889447236181e-05,
-      "loss": 5.3955,
-      "step": 33566
-    },
-    {
-      "epoch": 17.50560625814863,
-      "grad_norm": 1.4994423389434814,
-      "learning_rate": 6.677788944723619e-05,
-      "loss": 5.6555,
-      "step": 33567
-    },
-    {
-      "epoch": 17.50612777053455,
-      "grad_norm": 1.5074726343154907,
-      "learning_rate": 6.677688442211055e-05,
-      "loss": 5.2838,
-      "step": 33568
-    },
-    {
-      "epoch": 17.50664928292047,
-      "grad_norm": 1.4139057397842407,
-      "learning_rate": 6.677587939698493e-05,
-      "loss": 5.3472,
-      "step": 33569
-    },
-    {
-      "epoch": 17.50717079530639,
-      "grad_norm": 1.629477620124817,
-      "learning_rate": 6.67748743718593e-05,
-      "loss": 5.022,
-      "step": 33570
-    },
-    {
-      "epoch": 17.50769230769231,
-      "grad_norm": 1.5378221273422241,
-      "learning_rate": 6.677386934673367e-05,
-      "loss": 5.1959,
-      "step": 33571
-    },
-    {
-      "epoch": 17.508213820078225,
-      "grad_norm": 1.4986706972122192,
-      "learning_rate": 6.677286432160804e-05,
-      "loss": 5.0108,
-      "step": 33572
-    },
-    {
-      "epoch": 17.508735332464145,
-      "grad_norm": 1.516480565071106,
-      "learning_rate": 6.677185929648241e-05,
-      "loss": 5.4887,
-      "step": 33573
-    },
-    {
-      "epoch": 17.509256844850064,
-      "grad_norm": 1.4564619064331055,
-      "learning_rate": 6.677085427135679e-05,
-      "loss": 5.5599,
-      "step": 33574
-    },
-    {
-      "epoch": 17.509778357235984,
-      "grad_norm": 1.6503117084503174,
-      "learning_rate": 6.676984924623117e-05,
-      "loss": 4.4819,
-      "step": 33575
-    },
-    {
-      "epoch": 17.510299869621903,
-      "grad_norm": 1.5032923221588135,
-      "learning_rate": 6.676884422110553e-05,
-      "loss": 4.9698,
-      "step": 33576
-    },
-    {
-      "epoch": 17.510821382007823,
-      "grad_norm": 1.454858422279358,
-      "learning_rate": 6.67678391959799e-05,
-      "loss": 5.3213,
-      "step": 33577
-    },
-    {
-      "epoch": 17.511342894393742,
-      "grad_norm": 1.4162172079086304,
-      "learning_rate": 6.676683417085428e-05,
-      "loss": 5.5863,
-      "step": 33578
-    },
-    {
-      "epoch": 17.51186440677966,
-      "grad_norm": 1.5438073873519897,
-      "learning_rate": 6.676582914572864e-05,
-      "loss": 5.1078,
-      "step": 33579
-    },
-    {
-      "epoch": 17.51238591916558,
-      "grad_norm": 1.501839280128479,
-      "learning_rate": 6.676482412060302e-05,
-      "loss": 5.4797,
-      "step": 33580
-    },
-    {
-      "epoch": 17.5129074315515,
-      "grad_norm": 1.5382083654403687,
-      "learning_rate": 6.676381909547738e-05,
-      "loss": 5.3909,
-      "step": 33581
-    },
-    {
-      "epoch": 17.51342894393742,
-      "grad_norm": 1.6107404232025146,
-      "learning_rate": 6.676281407035176e-05,
-      "loss": 5.1726,
-      "step": 33582
-    },
-    {
-      "epoch": 17.513950456323336,
-      "grad_norm": 1.4606279134750366,
-      "learning_rate": 6.676180904522614e-05,
-      "loss": 5.0782,
-      "step": 33583
-    },
-    {
-      "epoch": 17.514471968709255,
-      "grad_norm": 1.4372286796569824,
-      "learning_rate": 6.676080402010052e-05,
-      "loss": 5.485,
-      "step": 33584
-    },
-    {
-      "epoch": 17.514993481095175,
-      "grad_norm": 1.5472735166549683,
-      "learning_rate": 6.675979899497488e-05,
-      "loss": 5.4155,
-      "step": 33585
-    },
-    {
-      "epoch": 17.515514993481094,
-      "grad_norm": 1.3883097171783447,
-      "learning_rate": 6.675879396984926e-05,
-      "loss": 5.7492,
-      "step": 33586
-    },
-    {
-      "epoch": 17.516036505867014,
-      "grad_norm": 1.4100937843322754,
-      "learning_rate": 6.675778894472362e-05,
-      "loss": 4.9691,
-      "step": 33587
-    },
-    {
-      "epoch": 17.516558018252933,
-      "grad_norm": 1.4207518100738525,
-      "learning_rate": 6.6756783919598e-05,
-      "loss": 5.6347,
-      "step": 33588
-    },
-    {
-      "epoch": 17.517079530638853,
-      "grad_norm": 1.4281411170959473,
-      "learning_rate": 6.675577889447236e-05,
-      "loss": 5.8982,
-      "step": 33589
-    },
-    {
-      "epoch": 17.517601043024772,
-      "grad_norm": 1.4364161491394043,
-      "learning_rate": 6.675477386934673e-05,
-      "loss": 5.793,
-      "step": 33590
-    },
-    {
-      "epoch": 17.51812255541069,
-      "grad_norm": 1.5742452144622803,
-      "learning_rate": 6.67537688442211e-05,
-      "loss": 5.0185,
-      "step": 33591
-    },
-    {
-      "epoch": 17.51864406779661,
-      "grad_norm": 1.3904318809509277,
-      "learning_rate": 6.675276381909547e-05,
-      "loss": 5.5428,
-      "step": 33592
-    },
-    {
-      "epoch": 17.51916558018253,
-      "grad_norm": 1.4912211894989014,
-      "learning_rate": 6.675175879396985e-05,
-      "loss": 5.5415,
-      "step": 33593
-    },
-    {
-      "epoch": 17.51968709256845,
-      "grad_norm": 1.5904203653335571,
-      "learning_rate": 6.675075376884423e-05,
-      "loss": 4.9657,
-      "step": 33594
-    },
-    {
-      "epoch": 17.52020860495437,
-      "grad_norm": 1.543765664100647,
-      "learning_rate": 6.67497487437186e-05,
-      "loss": 5.0986,
-      "step": 33595
-    },
-    {
-      "epoch": 17.520730117340285,
-      "grad_norm": 1.5020555257797241,
-      "learning_rate": 6.674874371859297e-05,
-      "loss": 5.0867,
-      "step": 33596
-    },
-    {
-      "epoch": 17.521251629726205,
-      "grad_norm": 1.4749630689620972,
-      "learning_rate": 6.674773869346735e-05,
-      "loss": 5.1648,
-      "step": 33597
-    },
-    {
-      "epoch": 17.521773142112124,
-      "grad_norm": 1.4686397314071655,
-      "learning_rate": 6.674673366834171e-05,
-      "loss": 5.2412,
-      "step": 33598
-    },
-    {
-      "epoch": 17.522294654498044,
-      "grad_norm": 1.490548849105835,
-      "learning_rate": 6.674572864321609e-05,
-      "loss": 5.5105,
-      "step": 33599
-    },
-    {
-      "epoch": 17.522816166883963,
-      "grad_norm": 1.5316661596298218,
-      "learning_rate": 6.674472361809045e-05,
-      "loss": 5.3177,
-      "step": 33600
-    },
-    {
-      "epoch": 17.523337679269883,
-      "grad_norm": 1.4707355499267578,
-      "learning_rate": 6.674371859296483e-05,
-      "loss": 5.5006,
-      "step": 33601
-    },
-    {
-      "epoch": 17.523859191655802,
-      "grad_norm": 1.445454716682434,
-      "learning_rate": 6.67427135678392e-05,
-      "loss": 5.0029,
-      "step": 33602
-    },
-    {
-      "epoch": 17.52438070404172,
-      "grad_norm": 1.415960669517517,
-      "learning_rate": 6.674170854271356e-05,
-      "loss": 5.7087,
-      "step": 33603
-    },
-    {
-      "epoch": 17.52490221642764,
-      "grad_norm": 1.4890999794006348,
-      "learning_rate": 6.674070351758794e-05,
-      "loss": 5.4903,
-      "step": 33604
-    },
-    {
-      "epoch": 17.52542372881356,
-      "grad_norm": 1.5075147151947021,
-      "learning_rate": 6.673969849246231e-05,
-      "loss": 5.202,
-      "step": 33605
-    },
-    {
-      "epoch": 17.52594524119948,
-      "grad_norm": 1.4761990308761597,
-      "learning_rate": 6.673869346733669e-05,
-      "loss": 5.0487,
-      "step": 33606
-    },
-    {
-      "epoch": 17.526466753585396,
-      "grad_norm": 1.3872963190078735,
-      "learning_rate": 6.673768844221106e-05,
-      "loss": 5.2417,
-      "step": 33607
-    },
-    {
-      "epoch": 17.526988265971315,
-      "grad_norm": 1.717200517654419,
-      "learning_rate": 6.673668341708543e-05,
-      "loss": 5.1929,
-      "step": 33608
-    },
-    {
-      "epoch": 17.527509778357235,
-      "grad_norm": 1.5887441635131836,
-      "learning_rate": 6.67356783919598e-05,
-      "loss": 5.0995,
-      "step": 33609
-    },
-    {
-      "epoch": 17.528031290743154,
-      "grad_norm": 1.567001223564148,
-      "learning_rate": 6.673467336683418e-05,
-      "loss": 5.6378,
-      "step": 33610
-    },
-    {
-      "epoch": 17.528552803129074,
-      "grad_norm": 1.524603009223938,
-      "learning_rate": 6.673366834170854e-05,
-      "loss": 4.9643,
-      "step": 33611
-    },
-    {
-      "epoch": 17.529074315514993,
-      "grad_norm": 1.4020081758499146,
-      "learning_rate": 6.673266331658292e-05,
-      "loss": 5.3662,
-      "step": 33612
-    },
-    {
-      "epoch": 17.529595827900913,
-      "grad_norm": 1.5260341167449951,
-      "learning_rate": 6.673165829145728e-05,
-      "loss": 5.1805,
-      "step": 33613
-    },
-    {
-      "epoch": 17.530117340286832,
-      "grad_norm": 1.514857292175293,
-      "learning_rate": 6.673065326633166e-05,
-      "loss": 5.655,
-      "step": 33614
-    },
-    {
-      "epoch": 17.53063885267275,
-      "grad_norm": 1.4716734886169434,
-      "learning_rate": 6.672964824120604e-05,
-      "loss": 5.2438,
-      "step": 33615
-    },
-    {
-      "epoch": 17.53116036505867,
-      "grad_norm": 1.7494533061981201,
-      "learning_rate": 6.67286432160804e-05,
-      "loss": 4.3148,
-      "step": 33616
-    },
-    {
-      "epoch": 17.53168187744459,
-      "grad_norm": 1.4963124990463257,
-      "learning_rate": 6.672763819095478e-05,
-      "loss": 4.317,
-      "step": 33617
-    },
-    {
-      "epoch": 17.53220338983051,
-      "grad_norm": 1.479438066482544,
-      "learning_rate": 6.672663316582915e-05,
-      "loss": 5.3836,
-      "step": 33618
-    },
-    {
-      "epoch": 17.532724902216426,
-      "grad_norm": 1.5507454872131348,
-      "learning_rate": 6.672562814070352e-05,
-      "loss": 5.028,
-      "step": 33619
-    },
-    {
-      "epoch": 17.533246414602345,
-      "grad_norm": 1.4634819030761719,
-      "learning_rate": 6.672462311557789e-05,
-      "loss": 5.407,
-      "step": 33620
-    },
-    {
-      "epoch": 17.533767926988265,
-      "grad_norm": 1.5122042894363403,
-      "learning_rate": 6.672361809045227e-05,
-      "loss": 5.2915,
-      "step": 33621
-    },
-    {
-      "epoch": 17.534289439374184,
-      "grad_norm": 1.4723451137542725,
-      "learning_rate": 6.672261306532663e-05,
-      "loss": 5.3107,
-      "step": 33622
-    },
-    {
-      "epoch": 17.534810951760104,
-      "grad_norm": 1.5636420249938965,
-      "learning_rate": 6.672160804020101e-05,
-      "loss": 5.0189,
-      "step": 33623
-    },
-    {
-      "epoch": 17.535332464146023,
-      "grad_norm": 1.5002046823501587,
-      "learning_rate": 6.672060301507537e-05,
-      "loss": 4.6782,
-      "step": 33624
-    },
-    {
-      "epoch": 17.535853976531943,
-      "grad_norm": 1.4689173698425293,
-      "learning_rate": 6.671959798994975e-05,
-      "loss": 5.3595,
-      "step": 33625
-    },
-    {
-      "epoch": 17.536375488917862,
-      "grad_norm": 1.4571763277053833,
-      "learning_rate": 6.671859296482413e-05,
-      "loss": 5.277,
-      "step": 33626
-    },
-    {
-      "epoch": 17.53689700130378,
-      "grad_norm": 1.3910627365112305,
-      "learning_rate": 6.67175879396985e-05,
-      "loss": 5.493,
-      "step": 33627
-    },
-    {
-      "epoch": 17.5374185136897,
-      "grad_norm": 1.4405733346939087,
-      "learning_rate": 6.671658291457287e-05,
-      "loss": 5.2626,
-      "step": 33628
-    },
-    {
-      "epoch": 17.53794002607562,
-      "grad_norm": 1.5706112384796143,
-      "learning_rate": 6.671557788944723e-05,
-      "loss": 5.0165,
-      "step": 33629
-    },
-    {
-      "epoch": 17.53846153846154,
-      "grad_norm": 1.4711456298828125,
-      "learning_rate": 6.671457286432161e-05,
-      "loss": 5.5881,
-      "step": 33630
-    },
-    {
-      "epoch": 17.538983050847456,
-      "grad_norm": 1.4380877017974854,
-      "learning_rate": 6.671356783919598e-05,
-      "loss": 5.3846,
-      "step": 33631
-    },
-    {
-      "epoch": 17.539504563233375,
-      "grad_norm": 1.5314782857894897,
-      "learning_rate": 6.671256281407035e-05,
-      "loss": 4.8974,
-      "step": 33632
-    },
-    {
-      "epoch": 17.540026075619295,
-      "grad_norm": 1.4686771631240845,
-      "learning_rate": 6.671155778894472e-05,
-      "loss": 5.3661,
-      "step": 33633
-    },
-    {
-      "epoch": 17.540547588005214,
-      "grad_norm": 1.5379958152770996,
-      "learning_rate": 6.67105527638191e-05,
-      "loss": 5.2994,
-      "step": 33634
-    },
-    {
-      "epoch": 17.541069100391134,
-      "grad_norm": 1.4266431331634521,
-      "learning_rate": 6.670954773869347e-05,
-      "loss": 5.2493,
-      "step": 33635
-    },
-    {
-      "epoch": 17.541590612777053,
-      "grad_norm": 1.4842463731765747,
-      "learning_rate": 6.670854271356785e-05,
-      "loss": 5.2025,
-      "step": 33636
-    },
-    {
-      "epoch": 17.542112125162973,
-      "grad_norm": 1.6071094274520874,
-      "learning_rate": 6.670753768844222e-05,
-      "loss": 5.0505,
-      "step": 33637
-    },
-    {
-      "epoch": 17.542633637548892,
-      "grad_norm": 1.4466701745986938,
-      "learning_rate": 6.67065326633166e-05,
-      "loss": 5.5867,
-      "step": 33638
-    },
-    {
-      "epoch": 17.54315514993481,
-      "grad_norm": 1.4442840814590454,
-      "learning_rate": 6.670552763819096e-05,
-      "loss": 5.3218,
-      "step": 33639
-    },
-    {
-      "epoch": 17.54367666232073,
-      "grad_norm": 1.5439329147338867,
-      "learning_rate": 6.670452261306534e-05,
-      "loss": 4.9258,
-      "step": 33640
-    },
-    {
-      "epoch": 17.54419817470665,
-      "grad_norm": 1.5394783020019531,
-      "learning_rate": 6.67035175879397e-05,
-      "loss": 5.1029,
-      "step": 33641
-    },
-    {
-      "epoch": 17.54471968709257,
-      "grad_norm": 1.4558593034744263,
-      "learning_rate": 6.670251256281408e-05,
-      "loss": 5.6461,
-      "step": 33642
-    },
-    {
-      "epoch": 17.545241199478486,
-      "grad_norm": 1.5075854063034058,
-      "learning_rate": 6.670150753768844e-05,
-      "loss": 5.3562,
-      "step": 33643
-    },
-    {
-      "epoch": 17.545762711864406,
-      "grad_norm": 1.5236035585403442,
-      "learning_rate": 6.67005025125628e-05,
-      "loss": 5.211,
-      "step": 33644
-    },
-    {
-      "epoch": 17.546284224250325,
-      "grad_norm": 1.4719504117965698,
-      "learning_rate": 6.669949748743718e-05,
-      "loss": 5.7418,
-      "step": 33645
-    },
-    {
-      "epoch": 17.546805736636244,
-      "grad_norm": 1.4759455919265747,
-      "learning_rate": 6.669849246231156e-05,
-      "loss": 5.4494,
-      "step": 33646
-    },
-    {
-      "epoch": 17.547327249022164,
-      "grad_norm": 1.5810338258743286,
-      "learning_rate": 6.669748743718594e-05,
-      "loss": 5.3793,
-      "step": 33647
-    },
-    {
-      "epoch": 17.547848761408083,
-      "grad_norm": 1.5113155841827393,
-      "learning_rate": 6.66964824120603e-05,
-      "loss": 5.1755,
-      "step": 33648
-    },
-    {
-      "epoch": 17.548370273794003,
-      "grad_norm": 1.4526629447937012,
-      "learning_rate": 6.669547738693468e-05,
-      "loss": 5.7052,
-      "step": 33649
-    },
-    {
-      "epoch": 17.548891786179922,
-      "grad_norm": 1.5794836282730103,
-      "learning_rate": 6.669447236180905e-05,
-      "loss": 5.4931,
-      "step": 33650
-    },
-    {
-      "epoch": 17.54941329856584,
-      "grad_norm": 1.4392940998077393,
-      "learning_rate": 6.669346733668342e-05,
-      "loss": 5.5864,
-      "step": 33651
-    },
-    {
-      "epoch": 17.54993481095176,
-      "grad_norm": 1.5202467441558838,
-      "learning_rate": 6.669246231155779e-05,
-      "loss": 5.4677,
-      "step": 33652
-    },
-    {
-      "epoch": 17.55045632333768,
-      "grad_norm": 1.519304633140564,
-      "learning_rate": 6.669145728643217e-05,
-      "loss": 4.8759,
-      "step": 33653
-    },
-    {
-      "epoch": 17.5509778357236,
-      "grad_norm": 1.4954559803009033,
-      "learning_rate": 6.669045226130653e-05,
-      "loss": 5.1882,
-      "step": 33654
-    },
-    {
-      "epoch": 17.551499348109516,
-      "grad_norm": 1.469001293182373,
-      "learning_rate": 6.668944723618091e-05,
-      "loss": 5.3801,
-      "step": 33655
-    },
-    {
-      "epoch": 17.552020860495436,
-      "grad_norm": 1.4140528440475464,
-      "learning_rate": 6.668844221105529e-05,
-      "loss": 5.3347,
-      "step": 33656
-    },
-    {
-      "epoch": 17.552542372881355,
-      "grad_norm": 1.3711529970169067,
-      "learning_rate": 6.668743718592965e-05,
-      "loss": 5.4455,
-      "step": 33657
-    },
-    {
-      "epoch": 17.553063885267274,
-      "grad_norm": 1.3301188945770264,
-      "learning_rate": 6.668643216080403e-05,
-      "loss": 5.63,
-      "step": 33658
-    },
-    {
-      "epoch": 17.553585397653194,
-      "grad_norm": 1.5203075408935547,
-      "learning_rate": 6.668542713567839e-05,
-      "loss": 5.5518,
-      "step": 33659
-    },
-    {
-      "epoch": 17.554106910039113,
-      "grad_norm": 1.5048304796218872,
-      "learning_rate": 6.668442211055277e-05,
-      "loss": 5.2824,
-      "step": 33660
-    },
-    {
-      "epoch": 17.554628422425033,
-      "grad_norm": 1.5061664581298828,
-      "learning_rate": 6.668341708542713e-05,
-      "loss": 4.7134,
-      "step": 33661
-    },
-    {
-      "epoch": 17.555149934810952,
-      "grad_norm": 1.5149650573730469,
-      "learning_rate": 6.668241206030151e-05,
-      "loss": 5.2948,
-      "step": 33662
-    },
-    {
-      "epoch": 17.555671447196872,
-      "grad_norm": 1.4104220867156982,
-      "learning_rate": 6.668140703517588e-05,
-      "loss": 5.52,
-      "step": 33663
-    },
-    {
-      "epoch": 17.55619295958279,
-      "grad_norm": 1.5604751110076904,
-      "learning_rate": 6.668040201005025e-05,
-      "loss": 5.6831,
-      "step": 33664
-    },
-    {
-      "epoch": 17.55671447196871,
-      "grad_norm": 1.5430617332458496,
-      "learning_rate": 6.667939698492462e-05,
-      "loss": 5.694,
-      "step": 33665
-    },
-    {
-      "epoch": 17.557235984354627,
-      "grad_norm": 1.4992448091506958,
-      "learning_rate": 6.6678391959799e-05,
-      "loss": 5.6967,
-      "step": 33666
-    },
-    {
-      "epoch": 17.557757496740546,
-      "grad_norm": 1.7155048847198486,
-      "learning_rate": 6.667738693467337e-05,
-      "loss": 4.9862,
-      "step": 33667
-    },
-    {
-      "epoch": 17.558279009126466,
-      "grad_norm": 1.4662829637527466,
-      "learning_rate": 6.667638190954775e-05,
-      "loss": 5.2965,
-      "step": 33668
-    },
-    {
-      "epoch": 17.558800521512385,
-      "grad_norm": 1.5863327980041504,
-      "learning_rate": 6.667537688442212e-05,
-      "loss": 4.7476,
-      "step": 33669
-    },
-    {
-      "epoch": 17.559322033898304,
-      "grad_norm": 1.5220351219177246,
-      "learning_rate": 6.667437185929648e-05,
-      "loss": 5.355,
-      "step": 33670
-    },
-    {
-      "epoch": 17.559843546284224,
-      "grad_norm": 1.4135812520980835,
-      "learning_rate": 6.667336683417086e-05,
-      "loss": 5.1745,
-      "step": 33671
-    },
-    {
-      "epoch": 17.560365058670143,
-      "grad_norm": 1.5421332120895386,
-      "learning_rate": 6.667236180904522e-05,
-      "loss": 4.8216,
-      "step": 33672
-    },
-    {
-      "epoch": 17.560886571056063,
-      "grad_norm": 1.551348328590393,
-      "learning_rate": 6.66713567839196e-05,
-      "loss": 5.6274,
-      "step": 33673
-    },
-    {
-      "epoch": 17.561408083441982,
-      "grad_norm": 1.5774986743927002,
-      "learning_rate": 6.667035175879396e-05,
-      "loss": 5.2699,
-      "step": 33674
-    },
-    {
-      "epoch": 17.561929595827902,
-      "grad_norm": 1.5038342475891113,
-      "learning_rate": 6.666934673366834e-05,
-      "loss": 5.3302,
-      "step": 33675
-    },
-    {
-      "epoch": 17.56245110821382,
-      "grad_norm": 1.4705525636672974,
-      "learning_rate": 6.666834170854272e-05,
-      "loss": 5.3696,
-      "step": 33676
-    },
-    {
-      "epoch": 17.56297262059974,
-      "grad_norm": 1.4745301008224487,
-      "learning_rate": 6.66673366834171e-05,
-      "loss": 5.3296,
-      "step": 33677
-    },
-    {
-      "epoch": 17.56349413298566,
-      "grad_norm": 1.47714364528656,
-      "learning_rate": 6.666633165829146e-05,
-      "loss": 5.3645,
-      "step": 33678
-    },
-    {
-      "epoch": 17.564015645371576,
-      "grad_norm": 1.5404696464538574,
-      "learning_rate": 6.666532663316584e-05,
-      "loss": 5.1053,
-      "step": 33679
-    },
-    {
-      "epoch": 17.564537157757496,
-      "grad_norm": 1.5076885223388672,
-      "learning_rate": 6.66643216080402e-05,
-      "loss": 5.5291,
-      "step": 33680
-    },
-    {
-      "epoch": 17.565058670143415,
-      "grad_norm": 1.570454478263855,
-      "learning_rate": 6.666331658291458e-05,
-      "loss": 4.4599,
-      "step": 33681
-    },
-    {
-      "epoch": 17.565580182529335,
-      "grad_norm": 1.468189001083374,
-      "learning_rate": 6.666231155778895e-05,
-      "loss": 5.1342,
-      "step": 33682
-    },
-    {
-      "epoch": 17.566101694915254,
-      "grad_norm": 1.565236210823059,
-      "learning_rate": 6.666130653266331e-05,
-      "loss": 5.1939,
-      "step": 33683
-    },
-    {
-      "epoch": 17.566623207301173,
-      "grad_norm": 1.4437369108200073,
-      "learning_rate": 6.666030150753769e-05,
-      "loss": 5.4193,
-      "step": 33684
-    },
-    {
-      "epoch": 17.567144719687093,
-      "grad_norm": 1.52808678150177,
-      "learning_rate": 6.665929648241205e-05,
-      "loss": 5.729,
-      "step": 33685
-    },
-    {
-      "epoch": 17.567666232073012,
-      "grad_norm": 1.5112067461013794,
-      "learning_rate": 6.665829145728643e-05,
-      "loss": 5.6051,
-      "step": 33686
-    },
-    {
-      "epoch": 17.568187744458932,
-      "grad_norm": 1.5656968355178833,
-      "learning_rate": 6.665728643216081e-05,
-      "loss": 4.6542,
-      "step": 33687
-    },
-    {
-      "epoch": 17.56870925684485,
-      "grad_norm": 1.4708634614944458,
-      "learning_rate": 6.665628140703519e-05,
-      "loss": 5.3319,
-      "step": 33688
-    },
-    {
-      "epoch": 17.56923076923077,
-      "grad_norm": 1.4556068181991577,
-      "learning_rate": 6.665527638190955e-05,
-      "loss": 5.4015,
-      "step": 33689
-    },
-    {
-      "epoch": 17.569752281616687,
-      "grad_norm": 1.493565559387207,
-      "learning_rate": 6.665427135678393e-05,
-      "loss": 5.3053,
-      "step": 33690
-    },
-    {
-      "epoch": 17.570273794002606,
-      "grad_norm": 1.6009663343429565,
-      "learning_rate": 6.665326633165829e-05,
-      "loss": 5.3105,
-      "step": 33691
-    },
-    {
-      "epoch": 17.570795306388526,
-      "grad_norm": 1.543242335319519,
-      "learning_rate": 6.665226130653267e-05,
-      "loss": 5.2045,
-      "step": 33692
-    },
-    {
-      "epoch": 17.571316818774445,
-      "grad_norm": 1.416947603225708,
-      "learning_rate": 6.665125628140704e-05,
-      "loss": 5.354,
-      "step": 33693
-    },
-    {
-      "epoch": 17.571838331160365,
-      "grad_norm": 1.5502263307571411,
-      "learning_rate": 6.665025125628141e-05,
-      "loss": 5.404,
-      "step": 33694
-    },
-    {
-      "epoch": 17.572359843546284,
-      "grad_norm": 1.5582188367843628,
-      "learning_rate": 6.664924623115578e-05,
-      "loss": 5.5557,
-      "step": 33695
-    },
-    {
-      "epoch": 17.572881355932203,
-      "grad_norm": 1.634811282157898,
-      "learning_rate": 6.664824120603015e-05,
-      "loss": 5.4009,
-      "step": 33696
-    },
-    {
-      "epoch": 17.573402868318123,
-      "grad_norm": 1.5006394386291504,
-      "learning_rate": 6.664723618090453e-05,
-      "loss": 5.3488,
-      "step": 33697
-    },
-    {
-      "epoch": 17.573924380704042,
-      "grad_norm": 1.470304250717163,
-      "learning_rate": 6.66462311557789e-05,
-      "loss": 5.2474,
-      "step": 33698
-    },
-    {
-      "epoch": 17.574445893089962,
-      "grad_norm": 1.5794919729232788,
-      "learning_rate": 6.664522613065327e-05,
-      "loss": 5.7548,
-      "step": 33699
-    },
-    {
-      "epoch": 17.57496740547588,
-      "grad_norm": 1.4947253465652466,
-      "learning_rate": 6.664422110552764e-05,
-      "loss": 5.8375,
-      "step": 33700
-    },
-    {
-      "epoch": 17.5754889178618,
-      "grad_norm": 1.640751838684082,
-      "learning_rate": 6.664321608040202e-05,
-      "loss": 4.6767,
-      "step": 33701
-    },
-    {
-      "epoch": 17.576010430247717,
-      "grad_norm": 1.4546747207641602,
-      "learning_rate": 6.664221105527638e-05,
-      "loss": 5.5339,
-      "step": 33702
-    },
-    {
-      "epoch": 17.576531942633636,
-      "grad_norm": 1.5838881731033325,
-      "learning_rate": 6.664120603015076e-05,
-      "loss": 5.333,
-      "step": 33703
-    },
-    {
-      "epoch": 17.577053455019556,
-      "grad_norm": 1.49124276638031,
-      "learning_rate": 6.664020100502512e-05,
-      "loss": 5.5637,
-      "step": 33704
-    },
-    {
-      "epoch": 17.577574967405475,
-      "grad_norm": 1.516244888305664,
-      "learning_rate": 6.66391959798995e-05,
-      "loss": 5.0493,
-      "step": 33705
-    },
-    {
-      "epoch": 17.578096479791395,
-      "grad_norm": 1.6370948553085327,
-      "learning_rate": 6.663819095477387e-05,
-      "loss": 5.3325,
-      "step": 33706
-    },
-    {
-      "epoch": 17.578617992177314,
-      "grad_norm": 1.573280692100525,
-      "learning_rate": 6.663718592964824e-05,
-      "loss": 5.5464,
-      "step": 33707
-    },
-    {
-      "epoch": 17.579139504563233,
-      "grad_norm": 1.4545621871948242,
-      "learning_rate": 6.663618090452262e-05,
-      "loss": 5.4381,
-      "step": 33708
-    },
-    {
-      "epoch": 17.579661016949153,
-      "grad_norm": 1.536813735961914,
-      "learning_rate": 6.663517587939699e-05,
-      "loss": 5.105,
-      "step": 33709
-    },
-    {
-      "epoch": 17.580182529335072,
-      "grad_norm": 1.5321704149246216,
-      "learning_rate": 6.663417085427136e-05,
-      "loss": 5.2441,
-      "step": 33710
-    },
-    {
-      "epoch": 17.580704041720992,
-      "grad_norm": 1.4790980815887451,
-      "learning_rate": 6.663316582914573e-05,
-      "loss": 5.4938,
-      "step": 33711
-    },
-    {
-      "epoch": 17.58122555410691,
-      "grad_norm": 1.491873860359192,
-      "learning_rate": 6.66321608040201e-05,
-      "loss": 5.7178,
-      "step": 33712
-    },
-    {
-      "epoch": 17.58174706649283,
-      "grad_norm": 1.5181962251663208,
-      "learning_rate": 6.663115577889447e-05,
-      "loss": 5.413,
-      "step": 33713
-    },
-    {
-      "epoch": 17.582268578878747,
-      "grad_norm": 1.6142183542251587,
-      "learning_rate": 6.663015075376885e-05,
-      "loss": 5.0219,
-      "step": 33714
-    },
-    {
-      "epoch": 17.582790091264666,
-      "grad_norm": 1.4339420795440674,
-      "learning_rate": 6.662914572864321e-05,
-      "loss": 5.5702,
-      "step": 33715
-    },
-    {
-      "epoch": 17.583311603650586,
-      "grad_norm": 1.433240532875061,
-      "learning_rate": 6.662814070351759e-05,
-      "loss": 5.7526,
-      "step": 33716
-    },
-    {
-      "epoch": 17.583833116036505,
-      "grad_norm": 1.4996200799942017,
-      "learning_rate": 6.662713567839197e-05,
-      "loss": 5.1479,
-      "step": 33717
-    },
-    {
-      "epoch": 17.584354628422425,
-      "grad_norm": 1.6656213998794556,
-      "learning_rate": 6.662613065326635e-05,
-      "loss": 5.115,
-      "step": 33718
-    },
-    {
-      "epoch": 17.584876140808344,
-      "grad_norm": 1.421303153038025,
-      "learning_rate": 6.662512562814071e-05,
-      "loss": 5.0037,
-      "step": 33719
-    },
-    {
-      "epoch": 17.585397653194264,
-      "grad_norm": 1.7406703233718872,
-      "learning_rate": 6.662412060301509e-05,
-      "loss": 4.337,
-      "step": 33720
-    },
-    {
-      "epoch": 17.585919165580183,
-      "grad_norm": 1.533309817314148,
-      "learning_rate": 6.662311557788945e-05,
-      "loss": 5.4689,
-      "step": 33721
-    },
-    {
-      "epoch": 17.586440677966102,
-      "grad_norm": 1.382738471031189,
-      "learning_rate": 6.662211055276382e-05,
-      "loss": 4.8571,
-      "step": 33722
-    },
-    {
-      "epoch": 17.586962190352022,
-      "grad_norm": 1.5203123092651367,
-      "learning_rate": 6.66211055276382e-05,
-      "loss": 5.6117,
-      "step": 33723
-    },
-    {
-      "epoch": 17.58748370273794,
-      "grad_norm": 1.6312896013259888,
-      "learning_rate": 6.662010050251256e-05,
-      "loss": 5.3036,
-      "step": 33724
-    },
-    {
-      "epoch": 17.58800521512386,
-      "grad_norm": 1.5435765981674194,
-      "learning_rate": 6.661909547738694e-05,
-      "loss": 4.8804,
-      "step": 33725
-    },
-    {
-      "epoch": 17.588526727509777,
-      "grad_norm": 1.529320240020752,
-      "learning_rate": 6.66180904522613e-05,
-      "loss": 5.6202,
-      "step": 33726
-    },
-    {
-      "epoch": 17.589048239895696,
-      "grad_norm": 1.4398858547210693,
-      "learning_rate": 6.661708542713568e-05,
-      "loss": 5.3196,
-      "step": 33727
-    },
-    {
-      "epoch": 17.589569752281616,
-      "grad_norm": 1.573088526725769,
-      "learning_rate": 6.661608040201006e-05,
-      "loss": 5.1118,
-      "step": 33728
-    },
-    {
-      "epoch": 17.590091264667535,
-      "grad_norm": 1.626304268836975,
-      "learning_rate": 6.661507537688443e-05,
-      "loss": 5.3575,
-      "step": 33729
-    },
-    {
-      "epoch": 17.590612777053455,
-      "grad_norm": 1.4716562032699585,
-      "learning_rate": 6.66140703517588e-05,
-      "loss": 5.056,
-      "step": 33730
-    },
-    {
-      "epoch": 17.591134289439374,
-      "grad_norm": 1.5783181190490723,
-      "learning_rate": 6.661306532663318e-05,
-      "loss": 5.4429,
-      "step": 33731
-    },
-    {
-      "epoch": 17.591655801825294,
-      "grad_norm": 1.417047142982483,
-      "learning_rate": 6.661206030150754e-05,
-      "loss": 5.7955,
-      "step": 33732
-    },
-    {
-      "epoch": 17.592177314211213,
-      "grad_norm": 1.4808237552642822,
-      "learning_rate": 6.661105527638192e-05,
-      "loss": 5.1222,
-      "step": 33733
-    },
-    {
-      "epoch": 17.592698826597132,
-      "grad_norm": 1.5179468393325806,
-      "learning_rate": 6.661005025125628e-05,
-      "loss": 4.9214,
-      "step": 33734
-    },
-    {
-      "epoch": 17.593220338983052,
-      "grad_norm": 1.4209370613098145,
-      "learning_rate": 6.660904522613066e-05,
-      "loss": 5.4295,
-      "step": 33735
-    },
-    {
-      "epoch": 17.59374185136897,
-      "grad_norm": 1.4800810813903809,
-      "learning_rate": 6.660804020100502e-05,
-      "loss": 5.6346,
-      "step": 33736
-    },
-    {
-      "epoch": 17.59426336375489,
-      "grad_norm": 1.5786670446395874,
-      "learning_rate": 6.66070351758794e-05,
-      "loss": 4.7558,
-      "step": 33737
-    },
-    {
-      "epoch": 17.594784876140807,
-      "grad_norm": 1.5224394798278809,
-      "learning_rate": 6.660603015075378e-05,
-      "loss": 5.4386,
-      "step": 33738
-    },
-    {
-      "epoch": 17.595306388526726,
-      "grad_norm": 1.4232096672058105,
-      "learning_rate": 6.660502512562814e-05,
-      "loss": 5.6777,
-      "step": 33739
-    },
-    {
-      "epoch": 17.595827900912646,
-      "grad_norm": 1.5230361223220825,
-      "learning_rate": 6.660402010050252e-05,
-      "loss": 5.2491,
-      "step": 33740
-    },
-    {
-      "epoch": 17.596349413298565,
-      "grad_norm": 1.4490054845809937,
-      "learning_rate": 6.660301507537689e-05,
-      "loss": 5.2858,
-      "step": 33741
-    },
-    {
-      "epoch": 17.596870925684485,
-      "grad_norm": 1.4635945558547974,
-      "learning_rate": 6.660201005025126e-05,
-      "loss": 4.8286,
-      "step": 33742
-    },
-    {
-      "epoch": 17.597392438070404,
-      "grad_norm": 1.4475090503692627,
-      "learning_rate": 6.660100502512563e-05,
-      "loss": 5.3005,
-      "step": 33743
-    },
-    {
-      "epoch": 17.597913950456324,
-      "grad_norm": 1.625119686126709,
-      "learning_rate": 6.66e-05,
-      "loss": 5.0339,
-      "step": 33744
-    },
-    {
-      "epoch": 17.598435462842243,
-      "grad_norm": 1.4772495031356812,
-      "learning_rate": 6.659899497487437e-05,
-      "loss": 5.6718,
-      "step": 33745
-    },
-    {
-      "epoch": 17.598956975228162,
-      "grad_norm": 1.3886334896087646,
-      "learning_rate": 6.659798994974875e-05,
-      "loss": 5.815,
-      "step": 33746
-    },
-    {
-      "epoch": 17.599478487614082,
-      "grad_norm": 1.4445483684539795,
-      "learning_rate": 6.659698492462311e-05,
-      "loss": 5.2054,
-      "step": 33747
-    },
-    {
-      "epoch": 17.6,
-      "grad_norm": 1.4985733032226562,
-      "learning_rate": 6.659597989949749e-05,
-      "loss": 5.0516,
-      "step": 33748
-    },
-    {
-      "epoch": 17.60052151238592,
-      "grad_norm": 1.343237042427063,
-      "learning_rate": 6.659497487437187e-05,
-      "loss": 4.6615,
-      "step": 33749
-    },
-    {
-      "epoch": 17.601043024771837,
-      "grad_norm": 1.4917412996292114,
-      "learning_rate": 6.659396984924623e-05,
-      "loss": 5.3123,
-      "step": 33750
-    },
-    {
-      "epoch": 17.601564537157756,
-      "grad_norm": 1.5030463933944702,
-      "learning_rate": 6.659296482412061e-05,
-      "loss": 5.0499,
-      "step": 33751
-    },
-    {
-      "epoch": 17.602086049543676,
-      "grad_norm": 1.6039479970932007,
-      "learning_rate": 6.659195979899497e-05,
-      "loss": 4.57,
-      "step": 33752
-    },
-    {
-      "epoch": 17.602607561929595,
-      "grad_norm": 1.5410643815994263,
-      "learning_rate": 6.659095477386935e-05,
-      "loss": 5.5559,
-      "step": 33753
-    },
-    {
-      "epoch": 17.603129074315515,
-      "grad_norm": 1.5341347455978394,
-      "learning_rate": 6.658994974874372e-05,
-      "loss": 5.1812,
-      "step": 33754
-    },
-    {
-      "epoch": 17.603650586701434,
-      "grad_norm": 1.4772324562072754,
-      "learning_rate": 6.65889447236181e-05,
-      "loss": 5.6167,
-      "step": 33755
-    },
-    {
-      "epoch": 17.604172099087354,
-      "grad_norm": 1.4263862371444702,
-      "learning_rate": 6.658793969849246e-05,
-      "loss": 4.7809,
-      "step": 33756
-    },
-    {
-      "epoch": 17.604693611473273,
-      "grad_norm": 1.6111737489700317,
-      "learning_rate": 6.658693467336684e-05,
-      "loss": 5.2436,
-      "step": 33757
-    },
-    {
-      "epoch": 17.605215123859193,
-      "grad_norm": 1.4762698411941528,
-      "learning_rate": 6.658592964824121e-05,
-      "loss": 5.412,
-      "step": 33758
-    },
-    {
-      "epoch": 17.605736636245112,
-      "grad_norm": 1.4229278564453125,
-      "learning_rate": 6.658492462311559e-05,
-      "loss": 5.7681,
-      "step": 33759
-    },
-    {
-      "epoch": 17.60625814863103,
-      "grad_norm": 1.4975672960281372,
-      "learning_rate": 6.658391959798996e-05,
-      "loss": 5.1172,
-      "step": 33760
-    },
-    {
-      "epoch": 17.60677966101695,
-      "grad_norm": 1.5242888927459717,
-      "learning_rate": 6.658291457286433e-05,
-      "loss": 4.9736,
-      "step": 33761
-    },
-    {
-      "epoch": 17.607301173402867,
-      "grad_norm": 1.4693795442581177,
-      "learning_rate": 6.65819095477387e-05,
-      "loss": 5.6989,
-      "step": 33762
-    },
-    {
-      "epoch": 17.607822685788786,
-      "grad_norm": 1.4379431009292603,
-      "learning_rate": 6.658090452261306e-05,
-      "loss": 5.7966,
-      "step": 33763
-    },
-    {
-      "epoch": 17.608344198174706,
-      "grad_norm": 1.38744056224823,
-      "learning_rate": 6.657989949748744e-05,
-      "loss": 5.5788,
-      "step": 33764
-    },
-    {
-      "epoch": 17.608865710560625,
-      "grad_norm": 1.3683186769485474,
-      "learning_rate": 6.65788944723618e-05,
-      "loss": 5.749,
-      "step": 33765
-    },
-    {
-      "epoch": 17.609387222946545,
-      "grad_norm": 1.5162110328674316,
-      "learning_rate": 6.657788944723618e-05,
-      "loss": 5.4777,
-      "step": 33766
-    },
-    {
-      "epoch": 17.609908735332464,
-      "grad_norm": 1.5629044771194458,
-      "learning_rate": 6.657688442211055e-05,
-      "loss": 5.4041,
-      "step": 33767
-    },
-    {
-      "epoch": 17.610430247718384,
-      "grad_norm": 1.5068835020065308,
-      "learning_rate": 6.657587939698492e-05,
-      "loss": 5.6522,
-      "step": 33768
-    },
-    {
-      "epoch": 17.610951760104303,
-      "grad_norm": 1.4826602935791016,
-      "learning_rate": 6.65748743718593e-05,
-      "loss": 5.122,
-      "step": 33769
-    },
-    {
-      "epoch": 17.611473272490223,
-      "grad_norm": 1.4078724384307861,
-      "learning_rate": 6.657386934673368e-05,
-      "loss": 5.7527,
-      "step": 33770
-    },
-    {
-      "epoch": 17.611994784876142,
-      "grad_norm": 1.460314154624939,
-      "learning_rate": 6.657286432160804e-05,
-      "loss": 5.5369,
-      "step": 33771
-    },
-    {
-      "epoch": 17.61251629726206,
-      "grad_norm": 1.4740344285964966,
-      "learning_rate": 6.657185929648242e-05,
-      "loss": 5.5821,
-      "step": 33772
-    },
-    {
-      "epoch": 17.613037809647977,
-      "grad_norm": 1.4589219093322754,
-      "learning_rate": 6.657085427135679e-05,
-      "loss": 5.3075,
-      "step": 33773
-    },
-    {
-      "epoch": 17.613559322033897,
-      "grad_norm": 1.4922641515731812,
-      "learning_rate": 6.656984924623116e-05,
-      "loss": 5.3301,
-      "step": 33774
-    },
-    {
-      "epoch": 17.614080834419816,
-      "grad_norm": 1.5355814695358276,
-      "learning_rate": 6.656884422110553e-05,
-      "loss": 4.8599,
-      "step": 33775
-    },
-    {
-      "epoch": 17.614602346805736,
-      "grad_norm": 1.449438214302063,
-      "learning_rate": 6.65678391959799e-05,
-      "loss": 5.562,
-      "step": 33776
-    },
-    {
-      "epoch": 17.615123859191655,
-      "grad_norm": 1.5283610820770264,
-      "learning_rate": 6.656683417085427e-05,
-      "loss": 4.9283,
-      "step": 33777
-    },
-    {
-      "epoch": 17.615645371577575,
-      "grad_norm": 1.506994605064392,
-      "learning_rate": 6.656582914572864e-05,
-      "loss": 5.3903,
-      "step": 33778
-    },
-    {
-      "epoch": 17.616166883963494,
-      "grad_norm": 1.556771993637085,
-      "learning_rate": 6.656482412060301e-05,
-      "loss": 5.2816,
-      "step": 33779
-    },
-    {
-      "epoch": 17.616688396349414,
-      "grad_norm": 1.5162992477416992,
-      "learning_rate": 6.656381909547739e-05,
-      "loss": 5.4064,
-      "step": 33780
-    },
-    {
-      "epoch": 17.617209908735333,
-      "grad_norm": 1.5130563974380493,
-      "learning_rate": 6.656281407035177e-05,
-      "loss": 5.271,
-      "step": 33781
-    },
-    {
-      "epoch": 17.617731421121253,
-      "grad_norm": 1.5461342334747314,
-      "learning_rate": 6.656180904522613e-05,
-      "loss": 5.5198,
-      "step": 33782
-    },
-    {
-      "epoch": 17.618252933507172,
-      "grad_norm": 1.493440866470337,
-      "learning_rate": 6.656080402010051e-05,
-      "loss": 5.0257,
-      "step": 33783
-    },
-    {
-      "epoch": 17.61877444589309,
-      "grad_norm": 1.4186168909072876,
-      "learning_rate": 6.655979899497488e-05,
-      "loss": 5.6198,
-      "step": 33784
-    },
-    {
-      "epoch": 17.619295958279007,
-      "grad_norm": 1.5349383354187012,
-      "learning_rate": 6.655879396984925e-05,
-      "loss": 5.4171,
-      "step": 33785
-    },
-    {
-      "epoch": 17.619817470664927,
-      "grad_norm": 1.4538053274154663,
-      "learning_rate": 6.655778894472362e-05,
-      "loss": 5.2591,
-      "step": 33786
-    },
-    {
-      "epoch": 17.620338983050846,
-      "grad_norm": 1.6187940835952759,
-      "learning_rate": 6.6556783919598e-05,
-      "loss": 4.5785,
-      "step": 33787
-    },
-    {
-      "epoch": 17.620860495436766,
-      "grad_norm": 1.4495912790298462,
-      "learning_rate": 6.655577889447236e-05,
-      "loss": 4.9868,
-      "step": 33788
-    },
-    {
-      "epoch": 17.621382007822685,
-      "grad_norm": 1.4841169118881226,
-      "learning_rate": 6.655477386934674e-05,
-      "loss": 5.5063,
-      "step": 33789
-    },
-    {
-      "epoch": 17.621903520208605,
-      "grad_norm": 1.446797490119934,
-      "learning_rate": 6.655376884422112e-05,
-      "loss": 5.2298,
-      "step": 33790
-    },
-    {
-      "epoch": 17.622425032594524,
-      "grad_norm": 1.5024536848068237,
-      "learning_rate": 6.655276381909548e-05,
-      "loss": 5.2061,
-      "step": 33791
-    },
-    {
-      "epoch": 17.622946544980444,
-      "grad_norm": 1.4112651348114014,
-      "learning_rate": 6.655175879396986e-05,
-      "loss": 5.6213,
-      "step": 33792
-    },
-    {
-      "epoch": 17.623468057366363,
-      "grad_norm": 1.4014456272125244,
-      "learning_rate": 6.655075376884422e-05,
-      "loss": 5.589,
-      "step": 33793
-    },
-    {
-      "epoch": 17.623989569752283,
-      "grad_norm": 1.5408058166503906,
-      "learning_rate": 6.65497487437186e-05,
-      "loss": 5.2963,
-      "step": 33794
-    },
-    {
-      "epoch": 17.624511082138202,
-      "grad_norm": 1.5169403553009033,
-      "learning_rate": 6.654874371859296e-05,
-      "loss": 5.2042,
-      "step": 33795
-    },
-    {
-      "epoch": 17.62503259452412,
-      "grad_norm": 1.359486699104309,
-      "learning_rate": 6.654773869346734e-05,
-      "loss": 4.5958,
-      "step": 33796
-    },
-    {
-      "epoch": 17.625554106910037,
-      "grad_norm": 1.3975286483764648,
-      "learning_rate": 6.65467336683417e-05,
-      "loss": 5.0587,
-      "step": 33797
-    },
-    {
-      "epoch": 17.626075619295957,
-      "grad_norm": 1.4145923852920532,
-      "learning_rate": 6.654572864321608e-05,
-      "loss": 5.576,
-      "step": 33798
-    },
-    {
-      "epoch": 17.626597131681876,
-      "grad_norm": 1.4635958671569824,
-      "learning_rate": 6.654472361809045e-05,
-      "loss": 5.3606,
-      "step": 33799
-    },
-    {
-      "epoch": 17.627118644067796,
-      "grad_norm": 1.4422733783721924,
-      "learning_rate": 6.654371859296483e-05,
-      "loss": 5.3383,
-      "step": 33800
-    },
-    {
-      "epoch": 17.627640156453715,
-      "grad_norm": 1.4125279188156128,
-      "learning_rate": 6.65427135678392e-05,
-      "loss": 5.7087,
-      "step": 33801
-    },
-    {
-      "epoch": 17.628161668839635,
-      "grad_norm": 1.5115190744400024,
-      "learning_rate": 6.654170854271357e-05,
-      "loss": 5.5336,
-      "step": 33802
-    },
-    {
-      "epoch": 17.628683181225554,
-      "grad_norm": 1.4467918872833252,
-      "learning_rate": 6.654070351758795e-05,
-      "loss": 5.4829,
-      "step": 33803
-    },
-    {
-      "epoch": 17.629204693611474,
-      "grad_norm": 1.4647235870361328,
-      "learning_rate": 6.653969849246231e-05,
-      "loss": 5.6157,
-      "step": 33804
-    },
-    {
-      "epoch": 17.629726205997393,
-      "grad_norm": 1.5650229454040527,
-      "learning_rate": 6.653869346733669e-05,
-      "loss": 4.8279,
-      "step": 33805
-    },
-    {
-      "epoch": 17.630247718383313,
-      "grad_norm": 1.6451488733291626,
-      "learning_rate": 6.653768844221105e-05,
-      "loss": 5.0798,
-      "step": 33806
-    },
-    {
-      "epoch": 17.630769230769232,
-      "grad_norm": 1.6121498346328735,
-      "learning_rate": 6.653668341708543e-05,
-      "loss": 5.0958,
-      "step": 33807
-    },
-    {
-      "epoch": 17.63129074315515,
-      "grad_norm": 1.5337408781051636,
-      "learning_rate": 6.65356783919598e-05,
-      "loss": 5.5043,
-      "step": 33808
-    },
-    {
-      "epoch": 17.631812255541067,
-      "grad_norm": 1.5322473049163818,
-      "learning_rate": 6.653467336683417e-05,
-      "loss": 5.4167,
-      "step": 33809
-    },
-    {
-      "epoch": 17.632333767926987,
-      "grad_norm": 1.513409972190857,
-      "learning_rate": 6.653366834170855e-05,
-      "loss": 5.469,
-      "step": 33810
-    },
-    {
-      "epoch": 17.632855280312906,
-      "grad_norm": 1.4426590204238892,
-      "learning_rate": 6.653266331658293e-05,
-      "loss": 5.3086,
-      "step": 33811
-    },
-    {
-      "epoch": 17.633376792698826,
-      "grad_norm": Infinity,
-      "learning_rate": 6.653266331658293e-05,
-      "loss": 4.8068,
-      "step": 33812
-    },
-    {
-      "epoch": 17.633898305084745,
-      "grad_norm": 1.4448779821395874,
-      "learning_rate": 6.653165829145729e-05,
-      "loss": 5.5443,
-      "step": 33813
-    },
-    {
-      "epoch": 17.634419817470665,
-      "grad_norm": 1.5438796281814575,
-      "learning_rate": 6.653065326633167e-05,
-      "loss": 4.8116,
-      "step": 33814
-    },
-    {
-      "epoch": 17.634941329856584,
-      "grad_norm": 1.4495450258255005,
-      "learning_rate": 6.652964824120603e-05,
-      "loss": 5.1423,
-      "step": 33815
-    },
-    {
-      "epoch": 17.635462842242504,
-      "grad_norm": 1.5102604627609253,
-      "learning_rate": 6.652864321608041e-05,
-      "loss": 5.2453,
-      "step": 33816
-    },
-    {
-      "epoch": 17.635984354628423,
-      "grad_norm": 1.6123223304748535,
-      "learning_rate": 6.652763819095478e-05,
-      "loss": 5.366,
-      "step": 33817
-    },
-    {
-      "epoch": 17.636505867014343,
-      "grad_norm": 1.5805411338806152,
-      "learning_rate": 6.652663316582914e-05,
-      "loss": 5.0092,
-      "step": 33818
-    },
-    {
-      "epoch": 17.637027379400262,
-      "grad_norm": 1.4305447340011597,
-      "learning_rate": 6.652562814070352e-05,
-      "loss": 5.4225,
-      "step": 33819
-    },
-    {
-      "epoch": 17.63754889178618,
-      "grad_norm": 1.5143747329711914,
-      "learning_rate": 6.652462311557788e-05,
-      "loss": 5.3565,
-      "step": 33820
-    },
-    {
-      "epoch": 17.638070404172097,
-      "grad_norm": 1.4664747714996338,
-      "learning_rate": 6.652361809045226e-05,
-      "loss": 5.2885,
-      "step": 33821
-    },
-    {
-      "epoch": 17.638591916558017,
-      "grad_norm": 1.4960533380508423,
-      "learning_rate": 6.652261306532664e-05,
-      "loss": 5.2998,
-      "step": 33822
-    },
-    {
-      "epoch": 17.639113428943936,
-      "grad_norm": 1.4765934944152832,
-      "learning_rate": 6.652160804020102e-05,
-      "loss": 5.3076,
-      "step": 33823
-    },
-    {
-      "epoch": 17.639634941329856,
-      "grad_norm": 1.4125553369522095,
-      "learning_rate": 6.652060301507538e-05,
-      "loss": 5.4089,
-      "step": 33824
-    },
-    {
-      "epoch": 17.640156453715775,
-      "grad_norm": 1.5316276550292969,
-      "learning_rate": 6.651959798994976e-05,
-      "loss": 5.1998,
-      "step": 33825
-    },
-    {
-      "epoch": 17.640677966101695,
-      "grad_norm": 1.4357267618179321,
-      "learning_rate": 6.651859296482412e-05,
-      "loss": 5.4879,
-      "step": 33826
-    },
-    {
-      "epoch": 17.641199478487614,
-      "grad_norm": 1.4521565437316895,
-      "learning_rate": 6.65175879396985e-05,
-      "loss": 5.0126,
-      "step": 33827
-    },
-    {
-      "epoch": 17.641720990873534,
-      "grad_norm": 1.5098233222961426,
-      "learning_rate": 6.651658291457286e-05,
-      "loss": 5.4746,
-      "step": 33828
-    },
-    {
-      "epoch": 17.642242503259453,
-      "grad_norm": 1.4050415754318237,
-      "learning_rate": 6.651557788944724e-05,
-      "loss": 4.963,
-      "step": 33829
-    },
-    {
-      "epoch": 17.642764015645373,
-      "grad_norm": 1.4532086849212646,
-      "learning_rate": 6.65145728643216e-05,
-      "loss": 5.5061,
-      "step": 33830
-    },
-    {
-      "epoch": 17.643285528031292,
-      "grad_norm": 1.458490014076233,
-      "learning_rate": 6.651356783919598e-05,
-      "loss": 5.2467,
-      "step": 33831
-    },
-    {
-      "epoch": 17.64380704041721,
-      "grad_norm": 1.401644229888916,
-      "learning_rate": 6.651256281407036e-05,
-      "loss": 5.7751,
-      "step": 33832
-    },
-    {
-      "epoch": 17.644328552803128,
-      "grad_norm": 1.4745346307754517,
-      "learning_rate": 6.651155778894473e-05,
-      "loss": 5.0297,
-      "step": 33833
-    },
-    {
-      "epoch": 17.644850065189047,
-      "grad_norm": 1.5701780319213867,
-      "learning_rate": 6.65105527638191e-05,
-      "loss": 5.2247,
-      "step": 33834
-    },
-    {
-      "epoch": 17.645371577574966,
-      "grad_norm": 1.446345329284668,
-      "learning_rate": 6.650954773869347e-05,
-      "loss": 5.6006,
-      "step": 33835
-    },
-    {
-      "epoch": 17.645893089960886,
-      "grad_norm": 1.4327127933502197,
-      "learning_rate": 6.650854271356785e-05,
-      "loss": 5.5744,
-      "step": 33836
-    },
-    {
-      "epoch": 17.646414602346805,
-      "grad_norm": 1.4238333702087402,
-      "learning_rate": 6.650753768844221e-05,
-      "loss": 5.4359,
-      "step": 33837
-    },
-    {
-      "epoch": 17.646936114732725,
-      "grad_norm": 1.53669011592865,
-      "learning_rate": 6.650653266331659e-05,
-      "loss": 5.3665,
-      "step": 33838
-    },
-    {
-      "epoch": 17.647457627118644,
-      "grad_norm": 1.5340012311935425,
-      "learning_rate": 6.650552763819095e-05,
-      "loss": 5.0239,
-      "step": 33839
-    },
-    {
-      "epoch": 17.647979139504564,
-      "grad_norm": 1.449972152709961,
-      "learning_rate": 6.650452261306533e-05,
-      "loss": 5.3324,
-      "step": 33840
-    },
-    {
-      "epoch": 17.648500651890483,
-      "grad_norm": 1.6092926263809204,
-      "learning_rate": 6.65035175879397e-05,
-      "loss": 5.0087,
-      "step": 33841
-    },
-    {
-      "epoch": 17.649022164276403,
-      "grad_norm": 1.451465129852295,
-      "learning_rate": 6.650251256281407e-05,
-      "loss": 5.4494,
-      "step": 33842
-    },
-    {
-      "epoch": 17.649543676662322,
-      "grad_norm": 1.4876304864883423,
-      "learning_rate": 6.650150753768845e-05,
-      "loss": 5.4229,
-      "step": 33843
-    },
-    {
-      "epoch": 17.65006518904824,
-      "grad_norm": 1.3650668859481812,
-      "learning_rate": 6.650050251256281e-05,
-      "loss": 5.4394,
-      "step": 33844
-    },
-    {
-      "epoch": 17.650586701434158,
-      "grad_norm": 1.3925808668136597,
-      "learning_rate": 6.649949748743719e-05,
-      "loss": 5.7454,
-      "step": 33845
-    },
-    {
-      "epoch": 17.651108213820077,
-      "grad_norm": 1.3614554405212402,
-      "learning_rate": 6.649849246231156e-05,
-      "loss": 5.6958,
-      "step": 33846
-    },
-    {
-      "epoch": 17.651629726205996,
-      "grad_norm": 1.4332492351531982,
-      "learning_rate": 6.649748743718593e-05,
-      "loss": 5.6354,
-      "step": 33847
-    },
-    {
-      "epoch": 17.652151238591916,
-      "grad_norm": 1.456476092338562,
-      "learning_rate": 6.64964824120603e-05,
-      "loss": 5.7842,
-      "step": 33848
-    },
-    {
-      "epoch": 17.652672750977835,
-      "grad_norm": 1.494954228401184,
-      "learning_rate": 6.649547738693468e-05,
-      "loss": 5.399,
-      "step": 33849
-    },
-    {
-      "epoch": 17.653194263363755,
-      "grad_norm": 1.4491567611694336,
-      "learning_rate": 6.649447236180904e-05,
-      "loss": 5.5687,
-      "step": 33850
-    },
-    {
-      "epoch": 17.653715775749674,
-      "grad_norm": 1.4236538410186768,
-      "learning_rate": 6.649346733668342e-05,
-      "loss": 5.5914,
-      "step": 33851
-    },
-    {
-      "epoch": 17.654237288135594,
-      "grad_norm": 1.4422253370285034,
-      "learning_rate": 6.64924623115578e-05,
-      "loss": 5.5362,
-      "step": 33852
-    },
-    {
-      "epoch": 17.654758800521513,
-      "grad_norm": 1.4629368782043457,
-      "learning_rate": 6.649145728643217e-05,
-      "loss": 5.5788,
-      "step": 33853
-    },
-    {
-      "epoch": 17.655280312907433,
-      "grad_norm": 1.4266023635864258,
-      "learning_rate": 6.649045226130654e-05,
-      "loss": 5.3217,
-      "step": 33854
-    },
-    {
-      "epoch": 17.655801825293352,
-      "grad_norm": 1.4266482591629028,
-      "learning_rate": 6.648944723618092e-05,
-      "loss": 5.4001,
-      "step": 33855
-    },
-    {
-      "epoch": 17.656323337679268,
-      "grad_norm": 1.477031946182251,
-      "learning_rate": 6.648844221105528e-05,
-      "loss": 5.2308,
-      "step": 33856
-    },
-    {
-      "epoch": 17.656844850065188,
-      "grad_norm": 1.466384768486023,
-      "learning_rate": 6.648743718592965e-05,
-      "loss": 5.5058,
-      "step": 33857
-    },
-    {
-      "epoch": 17.657366362451107,
-      "grad_norm": 1.3970139026641846,
-      "learning_rate": 6.648643216080402e-05,
-      "loss": 5.8083,
-      "step": 33858
-    },
-    {
-      "epoch": 17.657887874837026,
-      "grad_norm": 1.4201661348342896,
-      "learning_rate": 6.648542713567839e-05,
-      "loss": 5.2677,
-      "step": 33859
-    },
-    {
-      "epoch": 17.658409387222946,
-      "grad_norm": 1.4833662509918213,
-      "learning_rate": 6.648442211055277e-05,
-      "loss": 5.3023,
-      "step": 33860
-    },
-    {
-      "epoch": 17.658930899608865,
-      "grad_norm": 1.3651678562164307,
-      "learning_rate": 6.648341708542713e-05,
-      "loss": 5.4226,
-      "step": 33861
-    },
-    {
-      "epoch": 17.659452411994785,
-      "grad_norm": 1.3860450983047485,
-      "learning_rate": 6.648241206030151e-05,
-      "loss": 5.4318,
-      "step": 33862
-    },
-    {
-      "epoch": 17.659973924380704,
-      "grad_norm": 1.4952644109725952,
-      "learning_rate": 6.648140703517589e-05,
-      "loss": 5.1795,
-      "step": 33863
-    },
-    {
-      "epoch": 17.660495436766624,
-      "grad_norm": 1.484712839126587,
-      "learning_rate": 6.648040201005026e-05,
-      "loss": 5.4457,
-      "step": 33864
-    },
-    {
-      "epoch": 17.661016949152543,
-      "grad_norm": 1.6080079078674316,
-      "learning_rate": 6.647939698492463e-05,
-      "loss": 4.7939,
-      "step": 33865
-    },
-    {
-      "epoch": 17.661538461538463,
-      "grad_norm": 1.4690585136413574,
-      "learning_rate": 6.6478391959799e-05,
-      "loss": 5.1035,
-      "step": 33866
-    },
-    {
-      "epoch": 17.662059973924382,
-      "grad_norm": 1.4821181297302246,
-      "learning_rate": 6.647738693467337e-05,
-      "loss": 5.1969,
-      "step": 33867
-    },
-    {
-      "epoch": 17.6625814863103,
-      "grad_norm": 1.412678837776184,
-      "learning_rate": 6.647638190954775e-05,
-      "loss": 5.8397,
-      "step": 33868
-    },
-    {
-      "epoch": 17.663102998696218,
-      "grad_norm": 1.3921016454696655,
-      "learning_rate": 6.647537688442211e-05,
-      "loss": 5.4288,
-      "step": 33869
-    },
-    {
-      "epoch": 17.663624511082137,
-      "grad_norm": 1.5297200679779053,
-      "learning_rate": 6.647437185929648e-05,
-      "loss": 5.6231,
-      "step": 33870
-    },
-    {
-      "epoch": 17.664146023468057,
-      "grad_norm": 1.4491889476776123,
-      "learning_rate": 6.647336683417085e-05,
-      "loss": 5.3223,
-      "step": 33871
-    },
-    {
-      "epoch": 17.664667535853976,
-      "grad_norm": 1.6279913187026978,
-      "learning_rate": 6.647236180904523e-05,
-      "loss": 5.179,
-      "step": 33872
-    },
-    {
-      "epoch": 17.665189048239895,
-      "grad_norm": 1.521485447883606,
-      "learning_rate": 6.647135678391961e-05,
-      "loss": 4.9657,
-      "step": 33873
-    },
-    {
-      "epoch": 17.665710560625815,
-      "grad_norm": 1.503202199935913,
-      "learning_rate": 6.647035175879397e-05,
-      "loss": 5.2446,
-      "step": 33874
-    },
-    {
-      "epoch": 17.666232073011734,
-      "grad_norm": 1.4568456411361694,
-      "learning_rate": 6.646934673366835e-05,
-      "loss": 5.2215,
-      "step": 33875
-    },
-    {
-      "epoch": 17.666753585397654,
-      "grad_norm": 1.3720821142196655,
-      "learning_rate": 6.646834170854272e-05,
-      "loss": 5.6146,
-      "step": 33876
-    },
-    {
-      "epoch": 17.667275097783573,
-      "grad_norm": 2.3169100284576416,
-      "learning_rate": 6.64673366834171e-05,
-      "loss": 4.9517,
-      "step": 33877
-    },
-    {
-      "epoch": 17.667796610169493,
-      "grad_norm": 1.475656270980835,
-      "learning_rate": 6.646633165829146e-05,
-      "loss": 5.371,
-      "step": 33878
-    },
-    {
-      "epoch": 17.668318122555412,
-      "grad_norm": 1.4679182767868042,
-      "learning_rate": 6.646532663316584e-05,
-      "loss": 5.5775,
-      "step": 33879
-    },
-    {
-      "epoch": 17.668839634941328,
-      "grad_norm": 1.5183557271957397,
-      "learning_rate": 6.64643216080402e-05,
-      "loss": 5.0717,
-      "step": 33880
-    },
-    {
-      "epoch": 17.669361147327248,
-      "grad_norm": 1.4274855852127075,
-      "learning_rate": 6.646331658291458e-05,
-      "loss": 5.4067,
-      "step": 33881
-    },
-    {
-      "epoch": 17.669882659713167,
-      "grad_norm": 1.5732702016830444,
-      "learning_rate": 6.646231155778894e-05,
-      "loss": 5.2105,
-      "step": 33882
-    },
-    {
-      "epoch": 17.670404172099087,
-      "grad_norm": 1.493910312652588,
-      "learning_rate": 6.646130653266332e-05,
-      "loss": 5.5915,
-      "step": 33883
-    },
-    {
-      "epoch": 17.670925684485006,
-      "grad_norm": 1.6213699579238892,
-      "learning_rate": 6.64603015075377e-05,
-      "loss": 5.1789,
-      "step": 33884
-    },
-    {
-      "epoch": 17.671447196870925,
-      "grad_norm": 1.6003233194351196,
-      "learning_rate": 6.645929648241206e-05,
-      "loss": 5.3253,
-      "step": 33885
-    },
-    {
-      "epoch": 17.671968709256845,
-      "grad_norm": 1.6233675479888916,
-      "learning_rate": 6.645829145728644e-05,
-      "loss": 5.2121,
-      "step": 33886
-    },
-    {
-      "epoch": 17.672490221642764,
-      "grad_norm": 1.5195295810699463,
-      "learning_rate": 6.64572864321608e-05,
-      "loss": 5.1551,
-      "step": 33887
-    },
-    {
-      "epoch": 17.673011734028684,
-      "grad_norm": 1.4437282085418701,
-      "learning_rate": 6.645628140703518e-05,
-      "loss": 5.4272,
-      "step": 33888
-    },
-    {
-      "epoch": 17.673533246414603,
-      "grad_norm": 1.4974250793457031,
-      "learning_rate": 6.645527638190955e-05,
-      "loss": 5.1523,
-      "step": 33889
-    },
-    {
-      "epoch": 17.674054758800523,
-      "grad_norm": 1.5292688608169556,
-      "learning_rate": 6.645427135678392e-05,
-      "loss": 5.2353,
-      "step": 33890
-    },
-    {
-      "epoch": 17.674576271186442,
-      "grad_norm": 1.431607723236084,
-      "learning_rate": 6.645326633165829e-05,
-      "loss": 5.3304,
-      "step": 33891
-    },
-    {
-      "epoch": 17.675097783572358,
-      "grad_norm": 1.4660619497299194,
-      "learning_rate": 6.645226130653267e-05,
-      "loss": 5.6571,
-      "step": 33892
-    },
-    {
-      "epoch": 17.675619295958278,
-      "grad_norm": 1.4676519632339478,
-      "learning_rate": 6.645125628140704e-05,
-      "loss": 5.3781,
-      "step": 33893
-    },
-    {
-      "epoch": 17.676140808344197,
-      "grad_norm": 1.4927010536193848,
-      "learning_rate": 6.645025125628142e-05,
-      "loss": 5.5255,
-      "step": 33894
-    },
-    {
-      "epoch": 17.676662320730117,
-      "grad_norm": 1.4480130672454834,
-      "learning_rate": 6.644924623115579e-05,
-      "loss": 5.6221,
-      "step": 33895
-    },
-    {
-      "epoch": 17.677183833116036,
-      "grad_norm": 1.4894375801086426,
-      "learning_rate": 6.644824120603015e-05,
-      "loss": 5.4627,
-      "step": 33896
-    },
-    {
-      "epoch": 17.677705345501955,
-      "grad_norm": 1.5619641542434692,
-      "learning_rate": 6.644723618090453e-05,
-      "loss": 5.4296,
-      "step": 33897
-    },
-    {
-      "epoch": 17.678226857887875,
-      "grad_norm": 1.5440536737442017,
-      "learning_rate": 6.644623115577889e-05,
-      "loss": 5.445,
-      "step": 33898
-    },
-    {
-      "epoch": 17.678748370273794,
-      "grad_norm": 1.5323095321655273,
-      "learning_rate": 6.644522613065327e-05,
-      "loss": 5.3187,
-      "step": 33899
-    },
-    {
-      "epoch": 17.679269882659714,
-      "grad_norm": 1.5132790803909302,
-      "learning_rate": 6.644422110552763e-05,
-      "loss": 5.613,
-      "step": 33900
-    },
-    {
-      "epoch": 17.679791395045633,
-      "grad_norm": 1.6001657247543335,
-      "learning_rate": 6.644321608040201e-05,
-      "loss": 4.9786,
-      "step": 33901
-    },
-    {
-      "epoch": 17.680312907431553,
-      "grad_norm": 1.5349171161651611,
-      "learning_rate": 6.644221105527638e-05,
-      "loss": 5.2091,
-      "step": 33902
-    },
-    {
-      "epoch": 17.680834419817472,
-      "grad_norm": 1.5584683418273926,
-      "learning_rate": 6.644120603015075e-05,
-      "loss": 4.927,
-      "step": 33903
-    },
-    {
-      "epoch": 17.681355932203388,
-      "grad_norm": 1.443010687828064,
-      "learning_rate": 6.644020100502513e-05,
-      "loss": 4.8382,
-      "step": 33904
-    },
-    {
-      "epoch": 17.681877444589308,
-      "grad_norm": 1.4086185693740845,
-      "learning_rate": 6.643919597989951e-05,
-      "loss": 5.3392,
-      "step": 33905
-    },
-    {
-      "epoch": 17.682398956975227,
-      "grad_norm": 1.454107642173767,
-      "learning_rate": 6.643819095477387e-05,
-      "loss": 5.2746,
-      "step": 33906
-    },
-    {
-      "epoch": 17.682920469361147,
-      "grad_norm": 1.4081306457519531,
-      "learning_rate": 6.643718592964825e-05,
-      "loss": 5.1607,
-      "step": 33907
-    },
-    {
-      "epoch": 17.683441981747066,
-      "grad_norm": 1.4575961828231812,
-      "learning_rate": 6.643618090452262e-05,
-      "loss": 5.5815,
-      "step": 33908
-    },
-    {
-      "epoch": 17.683963494132986,
-      "grad_norm": 1.4838870763778687,
-      "learning_rate": 6.6435175879397e-05,
-      "loss": 5.239,
-      "step": 33909
-    },
-    {
-      "epoch": 17.684485006518905,
-      "grad_norm": 1.3310797214508057,
-      "learning_rate": 6.643417085427136e-05,
-      "loss": 5.9214,
-      "step": 33910
-    },
-    {
-      "epoch": 17.685006518904824,
-      "grad_norm": 1.4356521368026733,
-      "learning_rate": 6.643316582914572e-05,
-      "loss": 5.4567,
-      "step": 33911
-    },
-    {
-      "epoch": 17.685528031290744,
-      "grad_norm": 1.4327622652053833,
-      "learning_rate": 6.64321608040201e-05,
-      "loss": 4.9907,
-      "step": 33912
-    },
-    {
-      "epoch": 17.686049543676663,
-      "grad_norm": 1.5596448183059692,
-      "learning_rate": 6.643115577889448e-05,
-      "loss": 5.248,
-      "step": 33913
-    },
-    {
-      "epoch": 17.686571056062583,
-      "grad_norm": 1.4986680746078491,
-      "learning_rate": 6.643015075376886e-05,
-      "loss": 5.2715,
-      "step": 33914
-    },
-    {
-      "epoch": 17.687092568448502,
-      "grad_norm": 1.4337475299835205,
-      "learning_rate": 6.642914572864322e-05,
-      "loss": 5.3356,
-      "step": 33915
-    },
-    {
-      "epoch": 17.687614080834418,
-      "grad_norm": 1.3556197881698608,
-      "learning_rate": 6.64281407035176e-05,
-      "loss": 5.6934,
-      "step": 33916
-    },
-    {
-      "epoch": 17.688135593220338,
-      "grad_norm": 1.4887497425079346,
-      "learning_rate": 6.642713567839196e-05,
-      "loss": 5.4294,
-      "step": 33917
-    },
-    {
-      "epoch": 17.688657105606257,
-      "grad_norm": 1.4488849639892578,
-      "learning_rate": 6.642613065326634e-05,
-      "loss": 4.9872,
-      "step": 33918
-    },
-    {
-      "epoch": 17.689178617992177,
-      "grad_norm": 1.4602442979812622,
-      "learning_rate": 6.64251256281407e-05,
-      "loss": 5.2515,
-      "step": 33919
-    },
-    {
-      "epoch": 17.689700130378096,
-      "grad_norm": 1.4292163848876953,
-      "learning_rate": 6.642412060301508e-05,
-      "loss": 5.8074,
-      "step": 33920
-    },
-    {
-      "epoch": 17.690221642764016,
-      "grad_norm": 1.3874404430389404,
-      "learning_rate": 6.642311557788945e-05,
-      "loss": 5.6882,
-      "step": 33921
-    },
-    {
-      "epoch": 17.690743155149935,
-      "grad_norm": 1.546589732170105,
-      "learning_rate": 6.642211055276382e-05,
-      "loss": 4.545,
-      "step": 33922
-    },
-    {
-      "epoch": 17.691264667535854,
-      "grad_norm": 1.4791722297668457,
-      "learning_rate": 6.642110552763819e-05,
-      "loss": 5.3847,
-      "step": 33923
-    },
-    {
-      "epoch": 17.691786179921774,
-      "grad_norm": 1.4502544403076172,
-      "learning_rate": 6.642010050251257e-05,
-      "loss": 5.519,
-      "step": 33924
-    },
-    {
-      "epoch": 17.692307692307693,
-      "grad_norm": 1.47048819065094,
-      "learning_rate": 6.641909547738694e-05,
-      "loss": 5.526,
-      "step": 33925
-    },
-    {
-      "epoch": 17.692829204693613,
-      "grad_norm": 1.4873194694519043,
-      "learning_rate": 6.641809045226131e-05,
-      "loss": 5.2702,
-      "step": 33926
-    },
-    {
-      "epoch": 17.693350717079532,
-      "grad_norm": 1.5172910690307617,
-      "learning_rate": 6.641708542713569e-05,
-      "loss": 5.4209,
-      "step": 33927
-    },
-    {
-      "epoch": 17.69387222946545,
-      "grad_norm": 1.6104000806808472,
-      "learning_rate": 6.641608040201005e-05,
-      "loss": 5.0777,
-      "step": 33928
-    },
-    {
-      "epoch": 17.694393741851368,
-      "grad_norm": 1.5024957656860352,
-      "learning_rate": 6.641507537688443e-05,
-      "loss": 5.2314,
-      "step": 33929
-    },
-    {
-      "epoch": 17.694915254237287,
-      "grad_norm": 1.4198925495147705,
-      "learning_rate": 6.641407035175879e-05,
-      "loss": 4.2081,
-      "step": 33930
-    },
-    {
-      "epoch": 17.695436766623207,
-      "grad_norm": 1.4405436515808105,
-      "learning_rate": 6.641306532663317e-05,
-      "loss": 4.6771,
-      "step": 33931
-    },
-    {
-      "epoch": 17.695958279009126,
-      "grad_norm": 1.4244331121444702,
-      "learning_rate": 6.641206030150754e-05,
-      "loss": 5.7691,
-      "step": 33932
-    },
-    {
-      "epoch": 17.696479791395046,
-      "grad_norm": 1.4938485622406006,
-      "learning_rate": 6.641105527638191e-05,
-      "loss": 5.0114,
-      "step": 33933
-    },
-    {
-      "epoch": 17.697001303780965,
-      "grad_norm": 1.5359841585159302,
-      "learning_rate": 6.641005025125628e-05,
-      "loss": 5.1511,
-      "step": 33934
-    },
-    {
-      "epoch": 17.697522816166884,
-      "grad_norm": 1.4772708415985107,
-      "learning_rate": 6.640904522613065e-05,
-      "loss": 5.4924,
-      "step": 33935
-    },
-    {
-      "epoch": 17.698044328552804,
-      "grad_norm": 1.5132079124450684,
-      "learning_rate": 6.640804020100503e-05,
-      "loss": 5.0828,
-      "step": 33936
-    },
-    {
-      "epoch": 17.698565840938723,
-      "grad_norm": 1.5220805406570435,
-      "learning_rate": 6.64070351758794e-05,
-      "loss": 5.1595,
-      "step": 33937
-    },
-    {
-      "epoch": 17.699087353324643,
-      "grad_norm": 1.5240825414657593,
-      "learning_rate": 6.640603015075377e-05,
-      "loss": 5.2319,
-      "step": 33938
-    },
-    {
-      "epoch": 17.69960886571056,
-      "grad_norm": 1.4871296882629395,
-      "learning_rate": 6.640502512562814e-05,
-      "loss": 5.3343,
-      "step": 33939
-    },
-    {
-      "epoch": 17.70013037809648,
-      "grad_norm": 1.6370012760162354,
-      "learning_rate": 6.640402010050252e-05,
-      "loss": 4.7728,
-      "step": 33940
-    },
-    {
-      "epoch": 17.700651890482398,
-      "grad_norm": 1.4474058151245117,
-      "learning_rate": 6.640301507537688e-05,
-      "loss": 5.7329,
-      "step": 33941
-    },
-    {
-      "epoch": 17.701173402868317,
-      "grad_norm": 1.4073457717895508,
-      "learning_rate": 6.640201005025126e-05,
-      "loss": 5.2567,
-      "step": 33942
-    },
-    {
-      "epoch": 17.701694915254237,
-      "grad_norm": 1.4269779920578003,
-      "learning_rate": 6.640100502512562e-05,
-      "loss": 5.0222,
-      "step": 33943
-    },
-    {
-      "epoch": 17.702216427640156,
-      "grad_norm": 1.5119004249572754,
-      "learning_rate": 6.64e-05,
-      "loss": 5.3722,
-      "step": 33944
-    },
-    {
-      "epoch": 17.702737940026076,
-      "grad_norm": 1.5115355253219604,
-      "learning_rate": 6.639899497487438e-05,
-      "loss": 5.1401,
-      "step": 33945
-    },
-    {
-      "epoch": 17.703259452411995,
-      "grad_norm": 1.4069793224334717,
-      "learning_rate": 6.639798994974876e-05,
-      "loss": 4.9536,
-      "step": 33946
-    },
-    {
-      "epoch": 17.703780964797915,
-      "grad_norm": 1.4682292938232422,
-      "learning_rate": 6.639698492462312e-05,
-      "loss": 5.181,
-      "step": 33947
-    },
-    {
-      "epoch": 17.704302477183834,
-      "grad_norm": 1.5309064388275146,
-      "learning_rate": 6.63959798994975e-05,
-      "loss": 5.1188,
-      "step": 33948
-    },
-    {
-      "epoch": 17.704823989569753,
-      "grad_norm": 1.5165921449661255,
-      "learning_rate": 6.639497487437186e-05,
-      "loss": 5.3185,
-      "step": 33949
-    },
-    {
-      "epoch": 17.705345501955673,
-      "grad_norm": 1.5475646257400513,
-      "learning_rate": 6.639396984924623e-05,
-      "loss": 5.5732,
-      "step": 33950
-    },
-    {
-      "epoch": 17.705867014341592,
-      "grad_norm": 1.5823655128479004,
-      "learning_rate": 6.63929648241206e-05,
-      "loss": 5.1366,
-      "step": 33951
-    },
-    {
-      "epoch": 17.70638852672751,
-      "grad_norm": 1.523358941078186,
-      "learning_rate": 6.639195979899497e-05,
-      "loss": 5.1217,
-      "step": 33952
-    },
-    {
-      "epoch": 17.706910039113428,
-      "grad_norm": 1.4531477689743042,
-      "learning_rate": 6.639095477386935e-05,
-      "loss": 5.2526,
-      "step": 33953
-    },
-    {
-      "epoch": 17.707431551499347,
-      "grad_norm": 1.4679646492004395,
-      "learning_rate": 6.638994974874371e-05,
-      "loss": 5.1577,
-      "step": 33954
-    },
-    {
-      "epoch": 17.707953063885267,
-      "grad_norm": 1.584568738937378,
-      "learning_rate": 6.638894472361809e-05,
-      "loss": 5.4387,
-      "step": 33955
-    },
-    {
-      "epoch": 17.708474576271186,
-      "grad_norm": 1.4758305549621582,
-      "learning_rate": 6.638793969849247e-05,
-      "loss": 5.1792,
-      "step": 33956
-    },
-    {
-      "epoch": 17.708996088657106,
-      "grad_norm": 1.5371962785720825,
-      "learning_rate": 6.638693467336685e-05,
-      "loss": 4.9521,
-      "step": 33957
-    },
-    {
-      "epoch": 17.709517601043025,
-      "grad_norm": 1.5461273193359375,
-      "learning_rate": 6.638592964824121e-05,
-      "loss": 5.164,
-      "step": 33958
-    },
-    {
-      "epoch": 17.710039113428945,
-      "grad_norm": 1.4966219663619995,
-      "learning_rate": 6.638492462311559e-05,
-      "loss": 5.6426,
-      "step": 33959
-    },
-    {
-      "epoch": 17.710560625814864,
-      "grad_norm": 1.3943805694580078,
-      "learning_rate": 6.638391959798995e-05,
-      "loss": 5.7572,
-      "step": 33960
-    },
-    {
-      "epoch": 17.711082138200783,
-      "grad_norm": 1.5811325311660767,
-      "learning_rate": 6.638291457286433e-05,
-      "loss": 5.1492,
-      "step": 33961
-    },
-    {
-      "epoch": 17.711603650586703,
-      "grad_norm": 1.54188871383667,
-      "learning_rate": 6.63819095477387e-05,
-      "loss": 5.3182,
-      "step": 33962
-    },
-    {
-      "epoch": 17.71212516297262,
-      "grad_norm": 1.768578290939331,
-      "learning_rate": 6.638090452261306e-05,
-      "loss": 4.8658,
-      "step": 33963
-    },
-    {
-      "epoch": 17.71264667535854,
-      "grad_norm": 1.583997368812561,
-      "learning_rate": 6.637989949748744e-05,
-      "loss": 5.3972,
-      "step": 33964
-    },
-    {
-      "epoch": 17.713168187744458,
-      "grad_norm": 1.5256150960922241,
-      "learning_rate": 6.637889447236181e-05,
-      "loss": 5.3428,
-      "step": 33965
-    },
-    {
-      "epoch": 17.713689700130377,
-      "grad_norm": 1.6452378034591675,
-      "learning_rate": 6.637788944723619e-05,
-      "loss": 5.1093,
-      "step": 33966
-    },
-    {
-      "epoch": 17.714211212516297,
-      "grad_norm": 1.4007986783981323,
-      "learning_rate": 6.637688442211056e-05,
-      "loss": 5.4862,
-      "step": 33967
-    },
-    {
-      "epoch": 17.714732724902216,
-      "grad_norm": 1.4481621980667114,
-      "learning_rate": 6.637587939698493e-05,
-      "loss": 5.6233,
-      "step": 33968
-    },
-    {
-      "epoch": 17.715254237288136,
-      "grad_norm": 1.5582596063613892,
-      "learning_rate": 6.63748743718593e-05,
-      "loss": 5.2842,
-      "step": 33969
-    },
-    {
-      "epoch": 17.715775749674055,
-      "grad_norm": 1.5077180862426758,
-      "learning_rate": 6.637386934673368e-05,
-      "loss": 5.1885,
-      "step": 33970
-    },
-    {
-      "epoch": 17.716297262059975,
-      "grad_norm": 1.4837169647216797,
-      "learning_rate": 6.637286432160804e-05,
-      "loss": 5.4177,
-      "step": 33971
-    },
-    {
-      "epoch": 17.716818774445894,
-      "grad_norm": 1.4707247018814087,
-      "learning_rate": 6.637185929648242e-05,
-      "loss": 5.1212,
-      "step": 33972
-    },
-    {
-      "epoch": 17.717340286831814,
-      "grad_norm": 1.5598130226135254,
-      "learning_rate": 6.637085427135678e-05,
-      "loss": 4.7059,
-      "step": 33973
-    },
-    {
-      "epoch": 17.717861799217733,
-      "grad_norm": 1.4122270345687866,
-      "learning_rate": 6.636984924623116e-05,
-      "loss": 5.567,
-      "step": 33974
-    },
-    {
-      "epoch": 17.71838331160365,
-      "grad_norm": 1.4736398458480835,
-      "learning_rate": 6.636884422110552e-05,
-      "loss": 5.5962,
-      "step": 33975
-    },
-    {
-      "epoch": 17.71890482398957,
-      "grad_norm": 1.4897257089614868,
-      "learning_rate": 6.63678391959799e-05,
-      "loss": 5.5385,
-      "step": 33976
-    },
-    {
-      "epoch": 17.719426336375488,
-      "grad_norm": 1.3955022096633911,
-      "learning_rate": 6.636683417085428e-05,
-      "loss": 5.4917,
-      "step": 33977
-    },
-    {
-      "epoch": 17.719947848761407,
-      "grad_norm": 1.296190857887268,
-      "learning_rate": 6.636582914572864e-05,
-      "loss": 4.9687,
-      "step": 33978
-    },
-    {
-      "epoch": 17.720469361147327,
-      "grad_norm": 1.4385679960250854,
-      "learning_rate": 6.636482412060302e-05,
-      "loss": 5.411,
-      "step": 33979
-    },
-    {
-      "epoch": 17.720990873533246,
-      "grad_norm": 1.4437214136123657,
-      "learning_rate": 6.636381909547739e-05,
-      "loss": 5.4045,
-      "step": 33980
-    },
-    {
-      "epoch": 17.721512385919166,
-      "grad_norm": 1.5357213020324707,
-      "learning_rate": 6.636281407035176e-05,
-      "loss": 5.1633,
-      "step": 33981
-    },
-    {
-      "epoch": 17.722033898305085,
-      "grad_norm": 1.5092896223068237,
-      "learning_rate": 6.636180904522613e-05,
-      "loss": 5.2877,
-      "step": 33982
-    },
-    {
-      "epoch": 17.722555410691005,
-      "grad_norm": 1.53424072265625,
-      "learning_rate": 6.63608040201005e-05,
-      "loss": 5.1418,
-      "step": 33983
-    },
-    {
-      "epoch": 17.723076923076924,
-      "grad_norm": 1.4948614835739136,
-      "learning_rate": 6.635979899497487e-05,
-      "loss": 5.4838,
-      "step": 33984
-    },
-    {
-      "epoch": 17.723598435462844,
-      "grad_norm": 1.5220615863800049,
-      "learning_rate": 6.635879396984925e-05,
-      "loss": 5.2881,
-      "step": 33985
-    },
-    {
-      "epoch": 17.724119947848763,
-      "grad_norm": 1.5628411769866943,
-      "learning_rate": 6.635778894472363e-05,
-      "loss": 5.1657,
-      "step": 33986
-    },
-    {
-      "epoch": 17.72464146023468,
-      "grad_norm": 1.518025517463684,
-      "learning_rate": 6.6356783919598e-05,
-      "loss": 5.1628,
-      "step": 33987
-    },
-    {
-      "epoch": 17.7251629726206,
-      "grad_norm": 1.4870736598968506,
-      "learning_rate": 6.635577889447237e-05,
-      "loss": 5.262,
-      "step": 33988
-    },
-    {
-      "epoch": 17.725684485006518,
-      "grad_norm": 1.4452954530715942,
-      "learning_rate": 6.635477386934673e-05,
-      "loss": 5.4286,
-      "step": 33989
-    },
-    {
-      "epoch": 17.726205997392437,
-      "grad_norm": 1.511130928993225,
-      "learning_rate": 6.635376884422111e-05,
-      "loss": 5.3685,
-      "step": 33990
-    },
-    {
-      "epoch": 17.726727509778357,
-      "grad_norm": 1.4646263122558594,
-      "learning_rate": 6.635276381909547e-05,
-      "loss": 5.63,
-      "step": 33991
-    },
-    {
-      "epoch": 17.727249022164276,
-      "grad_norm": 1.4366792440414429,
-      "learning_rate": 6.635175879396985e-05,
-      "loss": 5.4624,
-      "step": 33992
-    },
-    {
-      "epoch": 17.727770534550196,
-      "grad_norm": 1.5110094547271729,
-      "learning_rate": 6.635075376884422e-05,
-      "loss": 5.326,
-      "step": 33993
-    },
-    {
-      "epoch": 17.728292046936115,
-      "grad_norm": 1.5028971433639526,
-      "learning_rate": 6.63497487437186e-05,
-      "loss": 5.508,
-      "step": 33994
-    },
-    {
-      "epoch": 17.728813559322035,
-      "grad_norm": 1.5843050479888916,
-      "learning_rate": 6.634874371859296e-05,
-      "loss": 4.9835,
-      "step": 33995
-    },
-    {
-      "epoch": 17.729335071707954,
-      "grad_norm": 1.4506911039352417,
-      "learning_rate": 6.634773869346734e-05,
-      "loss": 5.5396,
-      "step": 33996
-    },
-    {
-      "epoch": 17.729856584093874,
-      "grad_norm": 1.4093742370605469,
-      "learning_rate": 6.634673366834171e-05,
-      "loss": 5.3419,
-      "step": 33997
-    },
-    {
-      "epoch": 17.730378096479793,
-      "grad_norm": 1.4181358814239502,
-      "learning_rate": 6.634572864321609e-05,
-      "loss": 5.2916,
-      "step": 33998
-    },
-    {
-      "epoch": 17.73089960886571,
-      "grad_norm": 1.4684287309646606,
-      "learning_rate": 6.634472361809046e-05,
-      "loss": 5.5967,
-      "step": 33999
-    },
-    {
-      "epoch": 17.73142112125163,
-      "grad_norm": 1.5258405208587646,
-      "learning_rate": 6.634371859296483e-05,
-      "loss": 5.3881,
-      "step": 34000
-    },
-    {
-      "epoch": 17.731942633637548,
-      "grad_norm": 1.5774190425872803,
-      "learning_rate": 6.63427135678392e-05,
-      "loss": 5.2617,
-      "step": 34001
-    },
-    {
-      "epoch": 17.732464146023467,
-      "grad_norm": 1.6051017045974731,
-      "learning_rate": 6.634170854271358e-05,
-      "loss": 4.7995,
-      "step": 34002
-    },
-    {
-      "epoch": 17.732985658409387,
-      "grad_norm": 1.6786003112792969,
-      "learning_rate": 6.634070351758794e-05,
-      "loss": 5.1396,
-      "step": 34003
-    },
-    {
-      "epoch": 17.733507170795306,
-      "grad_norm": 1.4347326755523682,
-      "learning_rate": 6.63396984924623e-05,
-      "loss": 5.3605,
-      "step": 34004
-    },
-    {
-      "epoch": 17.734028683181226,
-      "grad_norm": 1.3448669910430908,
-      "learning_rate": 6.633869346733668e-05,
-      "loss": 5.3233,
-      "step": 34005
-    },
-    {
-      "epoch": 17.734550195567145,
-      "grad_norm": 1.4563310146331787,
-      "learning_rate": 6.633768844221106e-05,
-      "loss": 5.6125,
-      "step": 34006
-    },
-    {
-      "epoch": 17.735071707953065,
-      "grad_norm": 1.4407594203948975,
-      "learning_rate": 6.633668341708544e-05,
-      "loss": 5.4792,
-      "step": 34007
-    },
-    {
-      "epoch": 17.735593220338984,
-      "grad_norm": 1.5474656820297241,
-      "learning_rate": 6.63356783919598e-05,
-      "loss": 5.3747,
-      "step": 34008
-    },
-    {
-      "epoch": 17.736114732724904,
-      "grad_norm": 1.5788911581039429,
-      "learning_rate": 6.633467336683418e-05,
-      "loss": 4.9608,
-      "step": 34009
-    },
-    {
-      "epoch": 17.736636245110823,
-      "grad_norm": 1.584781289100647,
-      "learning_rate": 6.633366834170854e-05,
-      "loss": 5.238,
-      "step": 34010
-    },
-    {
-      "epoch": 17.73715775749674,
-      "grad_norm": 1.44032621383667,
-      "learning_rate": 6.633266331658292e-05,
-      "loss": 5.5393,
-      "step": 34011
-    },
-    {
-      "epoch": 17.73767926988266,
-      "grad_norm": 1.4892401695251465,
-      "learning_rate": 6.633165829145729e-05,
-      "loss": 5.1824,
-      "step": 34012
-    },
-    {
-      "epoch": 17.738200782268578,
-      "grad_norm": 1.5925078392028809,
-      "learning_rate": 6.633065326633166e-05,
-      "loss": 4.8028,
-      "step": 34013
-    },
-    {
-      "epoch": 17.738722294654497,
-      "grad_norm": 1.4390100240707397,
-      "learning_rate": 6.632964824120603e-05,
-      "loss": 5.6108,
-      "step": 34014
-    },
-    {
-      "epoch": 17.739243807040417,
-      "grad_norm": 1.498437762260437,
-      "learning_rate": 6.632864321608041e-05,
-      "loss": 5.3658,
-      "step": 34015
-    },
-    {
-      "epoch": 17.739765319426336,
-      "grad_norm": 1.5017285346984863,
-      "learning_rate": 6.632763819095477e-05,
-      "loss": 5.2876,
-      "step": 34016
-    },
-    {
-      "epoch": 17.740286831812256,
-      "grad_norm": 1.3372889757156372,
-      "learning_rate": 6.632663316582915e-05,
-      "loss": 4.9044,
-      "step": 34017
-    },
-    {
-      "epoch": 17.740808344198175,
-      "grad_norm": 1.4238203763961792,
-      "learning_rate": 6.632562814070353e-05,
-      "loss": 5.4756,
-      "step": 34018
-    },
-    {
-      "epoch": 17.741329856584095,
-      "grad_norm": 1.5349379777908325,
-      "learning_rate": 6.632462311557789e-05,
-      "loss": 5.4139,
-      "step": 34019
-    },
-    {
-      "epoch": 17.741851368970014,
-      "grad_norm": 1.5294885635375977,
-      "learning_rate": 6.632361809045227e-05,
-      "loss": 5.3083,
-      "step": 34020
-    },
-    {
-      "epoch": 17.742372881355934,
-      "grad_norm": 1.5907020568847656,
-      "learning_rate": 6.632261306532663e-05,
-      "loss": 5.4858,
-      "step": 34021
-    },
-    {
-      "epoch": 17.742894393741853,
-      "grad_norm": 1.4713807106018066,
-      "learning_rate": 6.632160804020101e-05,
-      "loss": 5.2902,
-      "step": 34022
-    },
-    {
-      "epoch": 17.74341590612777,
-      "grad_norm": 1.386351227760315,
-      "learning_rate": 6.632060301507538e-05,
-      "loss": 5.7511,
-      "step": 34023
-    },
-    {
-      "epoch": 17.74393741851369,
-      "grad_norm": 1.48618745803833,
-      "learning_rate": 6.631959798994975e-05,
-      "loss": 5.4261,
-      "step": 34024
-    },
-    {
-      "epoch": 17.744458930899608,
-      "grad_norm": 1.4960335493087769,
-      "learning_rate": 6.631859296482412e-05,
-      "loss": 5.6106,
-      "step": 34025
-    },
-    {
-      "epoch": 17.744980443285527,
-      "grad_norm": 1.4809279441833496,
-      "learning_rate": 6.63175879396985e-05,
-      "loss": 5.2712,
-      "step": 34026
-    },
-    {
-      "epoch": 17.745501955671447,
-      "grad_norm": 1.5545375347137451,
-      "learning_rate": 6.631658291457287e-05,
-      "loss": 5.1411,
-      "step": 34027
-    },
-    {
-      "epoch": 17.746023468057366,
-      "grad_norm": 1.4441710710525513,
-      "learning_rate": 6.631557788944725e-05,
-      "loss": 5.5798,
-      "step": 34028
-    },
-    {
-      "epoch": 17.746544980443286,
-      "grad_norm": 1.6738885641098022,
-      "learning_rate": 6.631457286432162e-05,
-      "loss": 4.7865,
-      "step": 34029
-    },
-    {
-      "epoch": 17.747066492829205,
-      "grad_norm": 1.3681094646453857,
-      "learning_rate": 6.631356783919598e-05,
-      "loss": 5.6539,
-      "step": 34030
-    },
-    {
-      "epoch": 17.747588005215125,
-      "grad_norm": 1.6925127506256104,
-      "learning_rate": 6.631256281407036e-05,
-      "loss": 5.002,
-      "step": 34031
-    },
-    {
-      "epoch": 17.748109517601044,
-      "grad_norm": 1.5873538255691528,
-      "learning_rate": 6.631155778894472e-05,
-      "loss": 5.3818,
-      "step": 34032
-    },
-    {
-      "epoch": 17.748631029986964,
-      "grad_norm": 1.524775505065918,
-      "learning_rate": 6.63105527638191e-05,
-      "loss": 5.767,
-      "step": 34033
-    },
-    {
-      "epoch": 17.749152542372883,
-      "grad_norm": 1.4148629903793335,
-      "learning_rate": 6.630954773869346e-05,
-      "loss": 5.3173,
-      "step": 34034
-    },
-    {
-      "epoch": 17.7496740547588,
-      "grad_norm": 1.5661494731903076,
-      "learning_rate": 6.630854271356784e-05,
-      "loss": 4.8247,
-      "step": 34035
-    },
-    {
-      "epoch": 17.75019556714472,
-      "grad_norm": 1.5565531253814697,
-      "learning_rate": 6.63075376884422e-05,
-      "loss": 5.1137,
-      "step": 34036
-    },
-    {
-      "epoch": 17.750717079530638,
-      "grad_norm": 1.617518424987793,
-      "learning_rate": 6.630653266331658e-05,
-      "loss": 5.4182,
-      "step": 34037
-    },
-    {
-      "epoch": 17.751238591916557,
-      "grad_norm": 1.4515782594680786,
-      "learning_rate": 6.630552763819096e-05,
-      "loss": 5.5014,
-      "step": 34038
-    },
-    {
-      "epoch": 17.751760104302477,
-      "grad_norm": 1.442868709564209,
-      "learning_rate": 6.630452261306534e-05,
-      "loss": 5.4058,
-      "step": 34039
-    },
-    {
-      "epoch": 17.752281616688396,
-      "grad_norm": 1.501365303993225,
-      "learning_rate": 6.63035175879397e-05,
-      "loss": 4.9482,
-      "step": 34040
-    },
-    {
-      "epoch": 17.752803129074316,
-      "grad_norm": 1.458452820777893,
-      "learning_rate": 6.630251256281408e-05,
-      "loss": 4.9089,
-      "step": 34041
-    },
-    {
-      "epoch": 17.753324641460235,
-      "grad_norm": 1.446770191192627,
-      "learning_rate": 6.630150753768845e-05,
-      "loss": 5.3489,
-      "step": 34042
-    },
-    {
-      "epoch": 17.753846153846155,
-      "grad_norm": 1.604756474494934,
-      "learning_rate": 6.630050251256281e-05,
-      "loss": 5.0985,
-      "step": 34043
-    },
-    {
-      "epoch": 17.754367666232074,
-      "grad_norm": 1.5058866739273071,
-      "learning_rate": 6.629949748743719e-05,
-      "loss": 5.6879,
-      "step": 34044
-    },
-    {
-      "epoch": 17.754889178617994,
-      "grad_norm": 1.5655930042266846,
-      "learning_rate": 6.629849246231155e-05,
-      "loss": 5.3617,
-      "step": 34045
-    },
-    {
-      "epoch": 17.75541069100391,
-      "grad_norm": 1.4845702648162842,
-      "learning_rate": 6.629748743718593e-05,
-      "loss": 5.5881,
-      "step": 34046
-    },
-    {
-      "epoch": 17.75593220338983,
-      "grad_norm": 1.4271787405014038,
-      "learning_rate": 6.629648241206031e-05,
-      "loss": 5.604,
-      "step": 34047
-    },
-    {
-      "epoch": 17.75645371577575,
-      "grad_norm": 1.526403546333313,
-      "learning_rate": 6.629547738693469e-05,
-      "loss": 4.8927,
-      "step": 34048
-    },
-    {
-      "epoch": 17.756975228161668,
-      "grad_norm": 1.5337697267532349,
-      "learning_rate": 6.629447236180905e-05,
-      "loss": 5.3205,
-      "step": 34049
-    },
-    {
-      "epoch": 17.757496740547587,
-      "grad_norm": 1.4769103527069092,
-      "learning_rate": 6.629346733668343e-05,
-      "loss": 5.1893,
-      "step": 34050
-    },
-    {
-      "epoch": 17.758018252933507,
-      "grad_norm": 1.6541460752487183,
-      "learning_rate": 6.629246231155779e-05,
-      "loss": 5.0246,
-      "step": 34051
-    },
-    {
-      "epoch": 17.758539765319426,
-      "grad_norm": 1.5017657279968262,
-      "learning_rate": 6.629145728643217e-05,
-      "loss": 5.5085,
-      "step": 34052
-    },
-    {
-      "epoch": 17.759061277705346,
-      "grad_norm": 1.3816566467285156,
-      "learning_rate": 6.629045226130653e-05,
-      "loss": 5.4103,
-      "step": 34053
-    },
-    {
-      "epoch": 17.759582790091265,
-      "grad_norm": 1.464143991470337,
-      "learning_rate": 6.628944723618091e-05,
-      "loss": 5.8631,
-      "step": 34054
-    },
-    {
-      "epoch": 17.760104302477185,
-      "grad_norm": 1.5266371965408325,
-      "learning_rate": 6.628844221105528e-05,
-      "loss": 5.5927,
-      "step": 34055
-    },
-    {
-      "epoch": 17.760625814863104,
-      "grad_norm": 1.4708077907562256,
-      "learning_rate": 6.628743718592964e-05,
-      "loss": 5.679,
-      "step": 34056
-    },
-    {
-      "epoch": 17.761147327249024,
-      "grad_norm": 1.5320889949798584,
-      "learning_rate": 6.628643216080402e-05,
-      "loss": 5.6736,
-      "step": 34057
-    },
-    {
-      "epoch": 17.761668839634943,
-      "grad_norm": 1.392321228981018,
-      "learning_rate": 6.62854271356784e-05,
-      "loss": 5.4249,
-      "step": 34058
-    },
-    {
-      "epoch": 17.76219035202086,
-      "grad_norm": 1.4703823328018188,
-      "learning_rate": 6.628442211055277e-05,
-      "loss": 5.3536,
-      "step": 34059
-    },
-    {
-      "epoch": 17.76271186440678,
-      "grad_norm": 1.4343229532241821,
-      "learning_rate": 6.628341708542714e-05,
-      "loss": 5.6212,
-      "step": 34060
-    },
-    {
-      "epoch": 17.763233376792698,
-      "grad_norm": 1.447316288948059,
-      "learning_rate": 6.628241206030152e-05,
-      "loss": 5.6336,
-      "step": 34061
-    },
-    {
-      "epoch": 17.763754889178617,
-      "grad_norm": 1.4628902673721313,
-      "learning_rate": 6.628140703517588e-05,
-      "loss": 5.7239,
-      "step": 34062
-    },
-    {
-      "epoch": 17.764276401564537,
-      "grad_norm": 1.4735627174377441,
-      "learning_rate": 6.628040201005026e-05,
-      "loss": 5.317,
-      "step": 34063
-    },
-    {
-      "epoch": 17.764797913950456,
-      "grad_norm": 1.541876196861267,
-      "learning_rate": 6.627939698492462e-05,
-      "loss": 5.5583,
-      "step": 34064
-    },
-    {
-      "epoch": 17.765319426336376,
-      "grad_norm": 1.5245206356048584,
-      "learning_rate": 6.6278391959799e-05,
-      "loss": 5.3153,
-      "step": 34065
-    },
-    {
-      "epoch": 17.765840938722295,
-      "grad_norm": 1.5133216381072998,
-      "learning_rate": 6.627738693467336e-05,
-      "loss": 5.3205,
-      "step": 34066
-    },
-    {
-      "epoch": 17.766362451108215,
-      "grad_norm": 1.4117406606674194,
-      "learning_rate": 6.627638190954774e-05,
-      "loss": 5.1594,
-      "step": 34067
-    },
-    {
-      "epoch": 17.766883963494134,
-      "grad_norm": 1.6245986223220825,
-      "learning_rate": 6.627537688442212e-05,
-      "loss": 5.2965,
-      "step": 34068
-    },
-    {
-      "epoch": 17.767405475880054,
-      "grad_norm": 1.5102895498275757,
-      "learning_rate": 6.627437185929648e-05,
-      "loss": 4.8158,
-      "step": 34069
-    },
-    {
-      "epoch": 17.76792698826597,
-      "grad_norm": 1.4122233390808105,
-      "learning_rate": 6.627336683417086e-05,
-      "loss": 4.8252,
-      "step": 34070
-    },
-    {
-      "epoch": 17.76844850065189,
-      "grad_norm": 1.4739338159561157,
-      "learning_rate": 6.627236180904523e-05,
-      "loss": 5.4903,
-      "step": 34071
-    },
-    {
-      "epoch": 17.76897001303781,
-      "grad_norm": 1.4737794399261475,
-      "learning_rate": 6.62713567839196e-05,
-      "loss": 5.394,
-      "step": 34072
-    },
-    {
-      "epoch": 17.769491525423728,
-      "grad_norm": 1.4276959896087646,
-      "learning_rate": 6.627035175879397e-05,
-      "loss": 5.1141,
-      "step": 34073
-    },
-    {
-      "epoch": 17.770013037809647,
-      "grad_norm": 1.4630588293075562,
-      "learning_rate": 6.626934673366835e-05,
-      "loss": 5.6762,
-      "step": 34074
-    },
-    {
-      "epoch": 17.770534550195567,
-      "grad_norm": 1.5317208766937256,
-      "learning_rate": 6.626834170854271e-05,
-      "loss": 5.1598,
-      "step": 34075
-    },
-    {
-      "epoch": 17.771056062581486,
-      "grad_norm": 1.4011359214782715,
-      "learning_rate": 6.626733668341709e-05,
-      "loss": 5.5981,
-      "step": 34076
-    },
-    {
-      "epoch": 17.771577574967406,
-      "grad_norm": 1.4914757013320923,
-      "learning_rate": 6.626633165829145e-05,
-      "loss": 5.7157,
-      "step": 34077
-    },
-    {
-      "epoch": 17.772099087353325,
-      "grad_norm": 1.5700732469558716,
-      "learning_rate": 6.626532663316583e-05,
-      "loss": 5.2566,
-      "step": 34078
-    },
-    {
-      "epoch": 17.772620599739245,
-      "grad_norm": 1.659585952758789,
-      "learning_rate": 6.626432160804021e-05,
-      "loss": 5.3603,
-      "step": 34079
-    },
-    {
-      "epoch": 17.773142112125164,
-      "grad_norm": 1.417339563369751,
-      "learning_rate": 6.626331658291459e-05,
-      "loss": 5.6049,
-      "step": 34080
-    },
-    {
-      "epoch": 17.773663624511084,
-      "grad_norm": 1.5806554555892944,
-      "learning_rate": 6.626231155778895e-05,
-      "loss": 5.142,
-      "step": 34081
-    },
-    {
-      "epoch": 17.774185136897,
-      "grad_norm": 1.5389351844787598,
-      "learning_rate": 6.626130653266331e-05,
-      "loss": 5.3522,
-      "step": 34082
-    },
-    {
-      "epoch": 17.77470664928292,
-      "grad_norm": 1.5545969009399414,
-      "learning_rate": 6.626030150753769e-05,
-      "loss": 4.6723,
-      "step": 34083
-    },
-    {
-      "epoch": 17.77522816166884,
-      "grad_norm": 1.5485390424728394,
-      "learning_rate": 6.625929648241206e-05,
-      "loss": 5.6948,
-      "step": 34084
-    },
-    {
-      "epoch": 17.775749674054758,
-      "grad_norm": 1.5437757968902588,
-      "learning_rate": 6.625829145728643e-05,
-      "loss": 5.2709,
-      "step": 34085
-    },
-    {
-      "epoch": 17.776271186440677,
-      "grad_norm": 1.4965580701828003,
-      "learning_rate": 6.62572864321608e-05,
-      "loss": 5.1718,
-      "step": 34086
-    },
-    {
-      "epoch": 17.776792698826597,
-      "grad_norm": 1.5819000005722046,
-      "learning_rate": 6.625628140703518e-05,
-      "loss": 5.1059,
-      "step": 34087
-    },
-    {
-      "epoch": 17.777314211212516,
-      "grad_norm": 1.4199316501617432,
-      "learning_rate": 6.625527638190955e-05,
-      "loss": 5.185,
-      "step": 34088
-    },
-    {
-      "epoch": 17.777835723598436,
-      "grad_norm": 1.3812531232833862,
-      "learning_rate": 6.625427135678393e-05,
-      "loss": 5.1865,
-      "step": 34089
-    },
-    {
-      "epoch": 17.778357235984355,
-      "grad_norm": 1.4293076992034912,
-      "learning_rate": 6.62532663316583e-05,
-      "loss": 4.9047,
-      "step": 34090
-    },
-    {
-      "epoch": 17.778878748370275,
-      "grad_norm": 1.3740917444229126,
-      "learning_rate": 6.625226130653267e-05,
-      "loss": 5.6706,
-      "step": 34091
-    },
-    {
-      "epoch": 17.779400260756194,
-      "grad_norm": 1.6101309061050415,
-      "learning_rate": 6.625125628140704e-05,
-      "loss": 5.0422,
-      "step": 34092
-    },
-    {
-      "epoch": 17.779921773142114,
-      "grad_norm": 1.6216946840286255,
-      "learning_rate": 6.625025125628142e-05,
-      "loss": 5.3548,
-      "step": 34093
-    },
-    {
-      "epoch": 17.78044328552803,
-      "grad_norm": 1.4708176851272583,
-      "learning_rate": 6.624924623115578e-05,
-      "loss": 5.0864,
-      "step": 34094
-    },
-    {
-      "epoch": 17.78096479791395,
-      "grad_norm": 1.5334604978561401,
-      "learning_rate": 6.624824120603016e-05,
-      "loss": 5.1087,
-      "step": 34095
-    },
-    {
-      "epoch": 17.78148631029987,
-      "grad_norm": 1.5152472257614136,
-      "learning_rate": 6.624723618090452e-05,
-      "loss": 5.181,
-      "step": 34096
-    },
-    {
-      "epoch": 17.782007822685788,
-      "grad_norm": 1.4161919355392456,
-      "learning_rate": 6.624623115577889e-05,
-      "loss": 5.3471,
-      "step": 34097
-    },
-    {
-      "epoch": 17.782529335071708,
-      "grad_norm": 1.6311007738113403,
-      "learning_rate": 6.624522613065327e-05,
-      "loss": 5.4504,
-      "step": 34098
-    },
-    {
-      "epoch": 17.783050847457627,
-      "grad_norm": 1.6146471500396729,
-      "learning_rate": 6.624422110552764e-05,
-      "loss": 5.123,
-      "step": 34099
-    },
-    {
-      "epoch": 17.783572359843546,
-      "grad_norm": 1.5950535535812378,
-      "learning_rate": 6.624321608040202e-05,
-      "loss": 5.4654,
-      "step": 34100
-    },
-    {
-      "epoch": 17.784093872229466,
-      "grad_norm": 1.5153263807296753,
-      "learning_rate": 6.624221105527639e-05,
-      "loss": 5.2997,
-      "step": 34101
-    },
-    {
-      "epoch": 17.784615384615385,
-      "grad_norm": 1.6389403343200684,
-      "learning_rate": 6.624120603015076e-05,
-      "loss": 4.8207,
-      "step": 34102
-    },
-    {
-      "epoch": 17.785136897001305,
-      "grad_norm": 1.5448570251464844,
-      "learning_rate": 6.624020100502513e-05,
-      "loss": 5.0458,
-      "step": 34103
-    },
-    {
-      "epoch": 17.785658409387224,
-      "grad_norm": 1.466753363609314,
-      "learning_rate": 6.62391959798995e-05,
-      "loss": 5.4656,
-      "step": 34104
-    },
-    {
-      "epoch": 17.786179921773144,
-      "grad_norm": 1.4027987718582153,
-      "learning_rate": 6.623819095477387e-05,
-      "loss": 5.5565,
-      "step": 34105
-    },
-    {
-      "epoch": 17.78670143415906,
-      "grad_norm": 1.4886294603347778,
-      "learning_rate": 6.623718592964825e-05,
-      "loss": 5.4626,
-      "step": 34106
-    },
-    {
-      "epoch": 17.78722294654498,
-      "grad_norm": 1.5093833208084106,
-      "learning_rate": 6.623618090452261e-05,
-      "loss": 5.5134,
-      "step": 34107
-    },
-    {
-      "epoch": 17.7877444589309,
-      "grad_norm": 1.5809425115585327,
-      "learning_rate": 6.623517587939699e-05,
-      "loss": 5.1823,
-      "step": 34108
-    },
-    {
-      "epoch": 17.788265971316818,
-      "grad_norm": 1.548783779144287,
-      "learning_rate": 6.623417085427135e-05,
-      "loss": 5.0598,
-      "step": 34109
-    },
-    {
-      "epoch": 17.788787483702738,
-      "grad_norm": 1.4810954332351685,
-      "learning_rate": 6.623316582914573e-05,
-      "loss": 5.1004,
-      "step": 34110
-    },
-    {
-      "epoch": 17.789308996088657,
-      "grad_norm": 1.514634132385254,
-      "learning_rate": 6.623216080402011e-05,
-      "loss": 5.5315,
-      "step": 34111
-    },
-    {
-      "epoch": 17.789830508474576,
-      "grad_norm": 1.5924838781356812,
-      "learning_rate": 6.623115577889447e-05,
-      "loss": 4.5643,
-      "step": 34112
-    },
-    {
-      "epoch": 17.790352020860496,
-      "grad_norm": 1.5920065641403198,
-      "learning_rate": 6.623015075376885e-05,
-      "loss": 5.6889,
-      "step": 34113
-    },
-    {
-      "epoch": 17.790873533246415,
-      "grad_norm": 1.5513619184494019,
-      "learning_rate": 6.622914572864322e-05,
-      "loss": 5.036,
-      "step": 34114
-    },
-    {
-      "epoch": 17.791395045632335,
-      "grad_norm": 1.455264687538147,
-      "learning_rate": 6.62281407035176e-05,
-      "loss": 5.2154,
-      "step": 34115
-    },
-    {
-      "epoch": 17.791916558018254,
-      "grad_norm": 1.4811134338378906,
-      "learning_rate": 6.622713567839196e-05,
-      "loss": 5.4472,
-      "step": 34116
-    },
-    {
-      "epoch": 17.792438070404174,
-      "grad_norm": 1.5069597959518433,
-      "learning_rate": 6.622613065326634e-05,
-      "loss": 5.4174,
-      "step": 34117
-    },
-    {
-      "epoch": 17.79295958279009,
-      "grad_norm": 1.400193691253662,
-      "learning_rate": 6.62251256281407e-05,
-      "loss": 5.6491,
-      "step": 34118
-    },
-    {
-      "epoch": 17.79348109517601,
-      "grad_norm": 1.4338458776474,
-      "learning_rate": 6.622412060301508e-05,
-      "loss": 4.9351,
-      "step": 34119
-    },
-    {
-      "epoch": 17.79400260756193,
-      "grad_norm": 1.5411001443862915,
-      "learning_rate": 6.622311557788946e-05,
-      "loss": 5.0466,
-      "step": 34120
-    },
-    {
-      "epoch": 17.794524119947848,
-      "grad_norm": 1.4247519969940186,
-      "learning_rate": 6.622211055276383e-05,
-      "loss": 5.6181,
-      "step": 34121
-    },
-    {
-      "epoch": 17.795045632333768,
-      "grad_norm": 1.3847562074661255,
-      "learning_rate": 6.62211055276382e-05,
-      "loss": 5.4639,
-      "step": 34122
-    },
-    {
-      "epoch": 17.795567144719687,
-      "grad_norm": 1.5717153549194336,
-      "learning_rate": 6.622010050251256e-05,
-      "loss": 5.1208,
-      "step": 34123
-    },
-    {
-      "epoch": 17.796088657105607,
-      "grad_norm": 1.4580587148666382,
-      "learning_rate": 6.621909547738694e-05,
-      "loss": 5.6597,
-      "step": 34124
-    },
-    {
-      "epoch": 17.796610169491526,
-      "grad_norm": 1.4694101810455322,
-      "learning_rate": 6.62180904522613e-05,
-      "loss": 5.2462,
-      "step": 34125
-    },
-    {
-      "epoch": 17.797131681877445,
-      "grad_norm": 1.33254873752594,
-      "learning_rate": 6.621708542713568e-05,
-      "loss": 4.8385,
-      "step": 34126
-    },
-    {
-      "epoch": 17.797653194263365,
-      "grad_norm": 1.5079607963562012,
-      "learning_rate": 6.621608040201005e-05,
-      "loss": 4.7002,
-      "step": 34127
-    },
-    {
-      "epoch": 17.798174706649284,
-      "grad_norm": 1.5180739164352417,
-      "learning_rate": 6.621507537688442e-05,
-      "loss": 5.3188,
-      "step": 34128
-    },
-    {
-      "epoch": 17.7986962190352,
-      "grad_norm": 1.4293779134750366,
-      "learning_rate": 6.621407035175879e-05,
-      "loss": 5.2707,
-      "step": 34129
-    },
-    {
-      "epoch": 17.79921773142112,
-      "grad_norm": 1.4428153038024902,
-      "learning_rate": 6.621306532663317e-05,
-      "loss": 5.5586,
-      "step": 34130
-    },
-    {
-      "epoch": 17.79973924380704,
-      "grad_norm": 1.3518590927124023,
-      "learning_rate": 6.621206030150754e-05,
-      "loss": 5.3518,
-      "step": 34131
-    },
-    {
-      "epoch": 17.80026075619296,
-      "grad_norm": 1.4788901805877686,
-      "learning_rate": 6.621105527638192e-05,
-      "loss": 5.2263,
-      "step": 34132
-    },
-    {
-      "epoch": 17.800782268578878,
-      "grad_norm": 1.4910988807678223,
-      "learning_rate": 6.621005025125629e-05,
-      "loss": 5.3799,
-      "step": 34133
-    },
-    {
-      "epoch": 17.801303780964798,
-      "grad_norm": 1.4465559720993042,
-      "learning_rate": 6.620904522613066e-05,
-      "loss": 5.5595,
-      "step": 34134
-    },
-    {
-      "epoch": 17.801825293350717,
-      "grad_norm": 1.666839599609375,
-      "learning_rate": 6.620804020100503e-05,
-      "loss": 5.204,
-      "step": 34135
-    },
-    {
-      "epoch": 17.802346805736637,
-      "grad_norm": 1.605843424797058,
-      "learning_rate": 6.620703517587939e-05,
-      "loss": 5.4852,
-      "step": 34136
-    },
-    {
-      "epoch": 17.802868318122556,
-      "grad_norm": 1.4259769916534424,
-      "learning_rate": 6.620603015075377e-05,
-      "loss": 5.6396,
-      "step": 34137
-    },
-    {
-      "epoch": 17.803389830508475,
-      "grad_norm": 1.552046537399292,
-      "learning_rate": 6.620502512562813e-05,
-      "loss": 4.9099,
-      "step": 34138
-    },
-    {
-      "epoch": 17.803911342894395,
-      "grad_norm": 1.401660680770874,
-      "learning_rate": 6.620402010050251e-05,
-      "loss": 5.3042,
-      "step": 34139
-    },
-    {
-      "epoch": 17.804432855280314,
-      "grad_norm": 1.6784077882766724,
-      "learning_rate": 6.620301507537689e-05,
-      "loss": 4.7102,
-      "step": 34140
-    },
-    {
-      "epoch": 17.804954367666234,
-      "grad_norm": 1.4585713148117065,
-      "learning_rate": 6.620201005025127e-05,
-      "loss": 4.952,
-      "step": 34141
-    },
-    {
-      "epoch": 17.80547588005215,
-      "grad_norm": 1.512412428855896,
-      "learning_rate": 6.620100502512563e-05,
-      "loss": 5.1914,
-      "step": 34142
-    },
-    {
-      "epoch": 17.80599739243807,
-      "grad_norm": 1.5335512161254883,
-      "learning_rate": 6.620000000000001e-05,
-      "loss": 5.3396,
-      "step": 34143
-    },
-    {
-      "epoch": 17.80651890482399,
-      "grad_norm": 1.4567714929580688,
-      "learning_rate": 6.619899497487437e-05,
-      "loss": 5.6313,
-      "step": 34144
-    },
-    {
-      "epoch": 17.807040417209908,
-      "grad_norm": 1.5505163669586182,
-      "learning_rate": 6.619798994974875e-05,
-      "loss": 5.2584,
-      "step": 34145
-    },
-    {
-      "epoch": 17.807561929595828,
-      "grad_norm": 1.5707868337631226,
-      "learning_rate": 6.619698492462312e-05,
-      "loss": 5.1976,
-      "step": 34146
-    },
-    {
-      "epoch": 17.808083441981747,
-      "grad_norm": 1.5511524677276611,
-      "learning_rate": 6.61959798994975e-05,
-      "loss": 5.419,
-      "step": 34147
-    },
-    {
-      "epoch": 17.808604954367667,
-      "grad_norm": 1.5905958414077759,
-      "learning_rate": 6.619497487437186e-05,
-      "loss": 5.2128,
-      "step": 34148
-    },
-    {
-      "epoch": 17.809126466753586,
-      "grad_norm": 1.648732304573059,
-      "learning_rate": 6.619396984924622e-05,
-      "loss": 4.9393,
-      "step": 34149
-    },
-    {
-      "epoch": 17.809647979139505,
-      "grad_norm": 1.60069739818573,
-      "learning_rate": 6.61929648241206e-05,
-      "loss": 5.2973,
-      "step": 34150
-    },
-    {
-      "epoch": 17.810169491525425,
-      "grad_norm": 1.559180736541748,
-      "learning_rate": 6.619195979899498e-05,
-      "loss": 5.1804,
-      "step": 34151
-    },
-    {
-      "epoch": 17.810691003911344,
-      "grad_norm": 1.461225152015686,
-      "learning_rate": 6.619095477386936e-05,
-      "loss": 5.6263,
-      "step": 34152
-    },
-    {
-      "epoch": 17.81121251629726,
-      "grad_norm": 1.5481172800064087,
-      "learning_rate": 6.618994974874372e-05,
-      "loss": 5.2492,
-      "step": 34153
-    },
-    {
-      "epoch": 17.81173402868318,
-      "grad_norm": 1.4384114742279053,
-      "learning_rate": 6.61889447236181e-05,
-      "loss": 5.6567,
-      "step": 34154
-    },
-    {
-      "epoch": 17.8122555410691,
-      "grad_norm": 1.4931539297103882,
-      "learning_rate": 6.618793969849246e-05,
-      "loss": 5.2217,
-      "step": 34155
-    },
-    {
-      "epoch": 17.81277705345502,
-      "grad_norm": 1.3814979791641235,
-      "learning_rate": 6.618693467336684e-05,
-      "loss": 5.2595,
-      "step": 34156
-    },
-    {
-      "epoch": 17.813298565840938,
-      "grad_norm": 1.474827527999878,
-      "learning_rate": 6.61859296482412e-05,
-      "loss": 5.3994,
-      "step": 34157
-    },
-    {
-      "epoch": 17.813820078226858,
-      "grad_norm": 1.5105987787246704,
-      "learning_rate": 6.618492462311558e-05,
-      "loss": 5.4001,
-      "step": 34158
-    },
-    {
-      "epoch": 17.814341590612777,
-      "grad_norm": 1.426007628440857,
-      "learning_rate": 6.618391959798995e-05,
-      "loss": 5.6906,
-      "step": 34159
-    },
-    {
-      "epoch": 17.814863102998697,
-      "grad_norm": 1.5066930055618286,
-      "learning_rate": 6.618291457286432e-05,
-      "loss": 5.4432,
-      "step": 34160
-    },
-    {
-      "epoch": 17.815384615384616,
-      "grad_norm": 1.5136616230010986,
-      "learning_rate": 6.61819095477387e-05,
-      "loss": 5.5499,
-      "step": 34161
-    },
-    {
-      "epoch": 17.815906127770536,
-      "grad_norm": 1.5226930379867554,
-      "learning_rate": 6.618090452261307e-05,
-      "loss": 5.4659,
-      "step": 34162
-    },
-    {
-      "epoch": 17.816427640156455,
-      "grad_norm": 1.5295212268829346,
-      "learning_rate": 6.617989949748744e-05,
-      "loss": 5.2271,
-      "step": 34163
-    },
-    {
-      "epoch": 17.816949152542374,
-      "grad_norm": 1.5003018379211426,
-      "learning_rate": 6.617889447236181e-05,
-      "loss": 5.3546,
-      "step": 34164
-    },
-    {
-      "epoch": 17.81747066492829,
-      "grad_norm": 1.852987289428711,
-      "learning_rate": 6.617788944723619e-05,
-      "loss": 5.0304,
-      "step": 34165
-    },
-    {
-      "epoch": 17.81799217731421,
-      "grad_norm": 1.3983187675476074,
-      "learning_rate": 6.617688442211055e-05,
-      "loss": 5.7503,
-      "step": 34166
-    },
-    {
-      "epoch": 17.81851368970013,
-      "grad_norm": 1.5318094491958618,
-      "learning_rate": 6.617587939698493e-05,
-      "loss": 4.9793,
-      "step": 34167
-    },
-    {
-      "epoch": 17.81903520208605,
-      "grad_norm": 1.5291134119033813,
-      "learning_rate": 6.617487437185929e-05,
-      "loss": 4.8036,
-      "step": 34168
-    },
-    {
-      "epoch": 17.819556714471968,
-      "grad_norm": 1.5256781578063965,
-      "learning_rate": 6.617386934673367e-05,
-      "loss": 5.487,
-      "step": 34169
-    },
-    {
-      "epoch": 17.820078226857888,
-      "grad_norm": 1.4821258783340454,
-      "learning_rate": 6.617286432160804e-05,
-      "loss": 5.0929,
-      "step": 34170
-    },
-    {
-      "epoch": 17.820599739243807,
-      "grad_norm": 1.5533053874969482,
-      "learning_rate": 6.617185929648241e-05,
-      "loss": 5.1573,
-      "step": 34171
-    },
-    {
-      "epoch": 17.821121251629727,
-      "grad_norm": 1.5665349960327148,
-      "learning_rate": 6.617085427135679e-05,
-      "loss": 4.977,
-      "step": 34172
-    },
-    {
-      "epoch": 17.821642764015646,
-      "grad_norm": 1.4240509271621704,
-      "learning_rate": 6.616984924623117e-05,
-      "loss": 5.4387,
-      "step": 34173
-    },
-    {
-      "epoch": 17.822164276401566,
-      "grad_norm": 1.453436255455017,
-      "learning_rate": 6.616884422110553e-05,
-      "loss": 5.5724,
-      "step": 34174
-    },
-    {
-      "epoch": 17.822685788787485,
-      "grad_norm": 1.4486750364303589,
-      "learning_rate": 6.616783919597991e-05,
-      "loss": 5.2324,
-      "step": 34175
-    },
-    {
-      "epoch": 17.823207301173404,
-      "grad_norm": 1.4306252002716064,
-      "learning_rate": 6.616683417085427e-05,
-      "loss": 5.6746,
-      "step": 34176
-    },
-    {
-      "epoch": 17.82372881355932,
-      "grad_norm": 1.5455257892608643,
-      "learning_rate": 6.616582914572864e-05,
-      "loss": 5.6496,
-      "step": 34177
-    },
-    {
-      "epoch": 17.82425032594524,
-      "grad_norm": 1.4717196226119995,
-      "learning_rate": 6.616482412060302e-05,
-      "loss": 5.6606,
-      "step": 34178
-    },
-    {
-      "epoch": 17.82477183833116,
-      "grad_norm": 1.4590615034103394,
-      "learning_rate": 6.616381909547738e-05,
-      "loss": 5.5873,
-      "step": 34179
-    },
-    {
-      "epoch": 17.82529335071708,
-      "grad_norm": 1.4441953897476196,
-      "learning_rate": 6.616281407035176e-05,
-      "loss": 5.3262,
-      "step": 34180
-    },
-    {
-      "epoch": 17.825814863103,
-      "grad_norm": 1.5191038846969604,
-      "learning_rate": 6.616180904522614e-05,
-      "loss": 5.4373,
-      "step": 34181
-    },
-    {
-      "epoch": 17.826336375488918,
-      "grad_norm": 1.4156248569488525,
-      "learning_rate": 6.616080402010051e-05,
-      "loss": 5.7195,
-      "step": 34182
-    },
-    {
-      "epoch": 17.826857887874837,
-      "grad_norm": 1.5161134004592896,
-      "learning_rate": 6.615979899497488e-05,
-      "loss": 5.3318,
-      "step": 34183
-    },
-    {
-      "epoch": 17.827379400260757,
-      "grad_norm": 1.5400136709213257,
-      "learning_rate": 6.615879396984926e-05,
-      "loss": 5.2867,
-      "step": 34184
-    },
-    {
-      "epoch": 17.827900912646676,
-      "grad_norm": 1.351003885269165,
-      "learning_rate": 6.615778894472362e-05,
-      "loss": 5.3927,
-      "step": 34185
-    },
-    {
-      "epoch": 17.828422425032596,
-      "grad_norm": 1.5704249143600464,
-      "learning_rate": 6.6156783919598e-05,
-      "loss": 4.9257,
-      "step": 34186
-    },
-    {
-      "epoch": 17.828943937418515,
-      "grad_norm": 1.4762811660766602,
-      "learning_rate": 6.615577889447236e-05,
-      "loss": 5.3818,
-      "step": 34187
-    },
-    {
-      "epoch": 17.829465449804434,
-      "grad_norm": 1.4651257991790771,
-      "learning_rate": 6.615477386934674e-05,
-      "loss": 5.2169,
-      "step": 34188
-    },
-    {
-      "epoch": 17.82998696219035,
-      "grad_norm": 1.584814429283142,
-      "learning_rate": 6.61537688442211e-05,
-      "loss": 5.172,
-      "step": 34189
-    },
-    {
-      "epoch": 17.83050847457627,
-      "grad_norm": 1.4631388187408447,
-      "learning_rate": 6.615276381909547e-05,
-      "loss": 5.4748,
-      "step": 34190
-    },
-    {
-      "epoch": 17.83102998696219,
-      "grad_norm": 1.5782526731491089,
-      "learning_rate": 6.615175879396985e-05,
-      "loss": 5.1783,
-      "step": 34191
-    },
-    {
-      "epoch": 17.83155149934811,
-      "grad_norm": 1.426953911781311,
-      "learning_rate": 6.615075376884423e-05,
-      "loss": 5.2351,
-      "step": 34192
-    },
-    {
-      "epoch": 17.83207301173403,
-      "grad_norm": 1.4784955978393555,
-      "learning_rate": 6.61497487437186e-05,
-      "loss": 5.3305,
-      "step": 34193
-    },
-    {
-      "epoch": 17.832594524119948,
-      "grad_norm": 1.5964387655258179,
-      "learning_rate": 6.614874371859297e-05,
-      "loss": 4.9802,
-      "step": 34194
-    },
-    {
-      "epoch": 17.833116036505867,
-      "grad_norm": 1.5952011346817017,
-      "learning_rate": 6.614773869346735e-05,
-      "loss": 4.6885,
-      "step": 34195
-    },
-    {
-      "epoch": 17.833637548891787,
-      "grad_norm": 1.5009689331054688,
-      "learning_rate": 6.614673366834171e-05,
-      "loss": 4.9583,
-      "step": 34196
-    },
-    {
-      "epoch": 17.834159061277706,
-      "grad_norm": 1.442628264427185,
-      "learning_rate": 6.614572864321609e-05,
-      "loss": 5.3865,
-      "step": 34197
-    },
-    {
-      "epoch": 17.834680573663626,
-      "grad_norm": 1.62046480178833,
-      "learning_rate": 6.614472361809045e-05,
-      "loss": 4.909,
-      "step": 34198
-    },
-    {
-      "epoch": 17.835202086049545,
-      "grad_norm": 1.5355955362319946,
-      "learning_rate": 6.614371859296483e-05,
-      "loss": 5.2237,
-      "step": 34199
-    },
-    {
-      "epoch": 17.835723598435465,
-      "grad_norm": 1.5196722745895386,
-      "learning_rate": 6.61427135678392e-05,
-      "loss": 5.3342,
-      "step": 34200
-    },
-    {
-      "epoch": 17.83624511082138,
-      "grad_norm": 1.4143521785736084,
-      "learning_rate": 6.614170854271357e-05,
-      "loss": 5.627,
-      "step": 34201
-    },
-    {
-      "epoch": 17.8367666232073,
-      "grad_norm": 1.4236629009246826,
-      "learning_rate": 6.614070351758795e-05,
-      "loss": 5.5351,
-      "step": 34202
-    },
-    {
-      "epoch": 17.83728813559322,
-      "grad_norm": 1.4348660707473755,
-      "learning_rate": 6.613969849246231e-05,
-      "loss": 5.4767,
-      "step": 34203
-    },
-    {
-      "epoch": 17.83780964797914,
-      "grad_norm": 1.4315319061279297,
-      "learning_rate": 6.613869346733669e-05,
-      "loss": 5.6239,
-      "step": 34204
-    },
-    {
-      "epoch": 17.83833116036506,
-      "grad_norm": 1.440545916557312,
-      "learning_rate": 6.613768844221106e-05,
-      "loss": 5.3444,
-      "step": 34205
-    },
-    {
-      "epoch": 17.838852672750978,
-      "grad_norm": 1.455419898033142,
-      "learning_rate": 6.613668341708543e-05,
-      "loss": 5.4415,
-      "step": 34206
-    },
-    {
-      "epoch": 17.839374185136897,
-      "grad_norm": 1.8297001123428345,
-      "learning_rate": 6.61356783919598e-05,
-      "loss": 5.5604,
-      "step": 34207
-    },
-    {
-      "epoch": 17.839895697522817,
-      "grad_norm": 1.4895025491714478,
-      "learning_rate": 6.613467336683418e-05,
-      "loss": 5.5389,
-      "step": 34208
-    },
-    {
-      "epoch": 17.840417209908736,
-      "grad_norm": 1.5638506412506104,
-      "learning_rate": 6.613366834170854e-05,
-      "loss": 5.2634,
-      "step": 34209
-    },
-    {
-      "epoch": 17.840938722294656,
-      "grad_norm": 1.4221298694610596,
-      "learning_rate": 6.613266331658292e-05,
-      "loss": 5.2549,
-      "step": 34210
-    },
-    {
-      "epoch": 17.841460234680575,
-      "grad_norm": 1.448691725730896,
-      "learning_rate": 6.613165829145728e-05,
-      "loss": 5.6286,
-      "step": 34211
-    },
-    {
-      "epoch": 17.841981747066495,
-      "grad_norm": 1.5602831840515137,
-      "learning_rate": 6.613065326633166e-05,
-      "loss": 5.2697,
-      "step": 34212
-    },
-    {
-      "epoch": 17.84250325945241,
-      "grad_norm": 1.3921191692352295,
-      "learning_rate": 6.612964824120604e-05,
-      "loss": 5.5511,
-      "step": 34213
-    },
-    {
-      "epoch": 17.84302477183833,
-      "grad_norm": 1.4501956701278687,
-      "learning_rate": 6.612864321608042e-05,
-      "loss": 5.3092,
-      "step": 34214
-    },
-    {
-      "epoch": 17.84354628422425,
-      "grad_norm": 1.4256457090377808,
-      "learning_rate": 6.612763819095478e-05,
-      "loss": 5.248,
-      "step": 34215
-    },
-    {
-      "epoch": 17.84406779661017,
-      "grad_norm": 1.6580601930618286,
-      "learning_rate": 6.612663316582914e-05,
-      "loss": 5.0267,
-      "step": 34216
-    },
-    {
-      "epoch": 17.84458930899609,
-      "grad_norm": 1.71620774269104,
-      "learning_rate": 6.612562814070352e-05,
-      "loss": 5.0205,
-      "step": 34217
-    },
-    {
-      "epoch": 17.845110821382008,
-      "grad_norm": 1.4386459589004517,
-      "learning_rate": 6.612462311557789e-05,
-      "loss": 5.3301,
-      "step": 34218
-    },
-    {
-      "epoch": 17.845632333767927,
-      "grad_norm": 1.405339002609253,
-      "learning_rate": 6.612361809045226e-05,
-      "loss": 5.361,
-      "step": 34219
-    },
-    {
-      "epoch": 17.846153846153847,
-      "grad_norm": 1.5633009672164917,
-      "learning_rate": 6.612261306532663e-05,
-      "loss": 5.1953,
-      "step": 34220
-    },
-    {
-      "epoch": 17.846675358539766,
-      "grad_norm": 1.4191161394119263,
-      "learning_rate": 6.6121608040201e-05,
-      "loss": 5.4512,
-      "step": 34221
-    },
-    {
-      "epoch": 17.847196870925686,
-      "grad_norm": 1.7495454549789429,
-      "learning_rate": 6.612060301507538e-05,
-      "loss": 5.464,
-      "step": 34222
-    },
-    {
-      "epoch": 17.847718383311605,
-      "grad_norm": 1.5335725545883179,
-      "learning_rate": 6.611959798994976e-05,
-      "loss": 5.1904,
-      "step": 34223
-    },
-    {
-      "epoch": 17.848239895697525,
-      "grad_norm": 1.614998459815979,
-      "learning_rate": 6.611859296482413e-05,
-      "loss": 5.2055,
-      "step": 34224
-    },
-    {
-      "epoch": 17.84876140808344,
-      "grad_norm": 1.5965970754623413,
-      "learning_rate": 6.61175879396985e-05,
-      "loss": 5.152,
-      "step": 34225
-    },
-    {
-      "epoch": 17.84928292046936,
-      "grad_norm": 1.5307005643844604,
-      "learning_rate": 6.611658291457287e-05,
-      "loss": 5.3936,
-      "step": 34226
-    },
-    {
-      "epoch": 17.84980443285528,
-      "grad_norm": 1.5912830829620361,
-      "learning_rate": 6.611557788944725e-05,
-      "loss": 5.1247,
-      "step": 34227
-    },
-    {
-      "epoch": 17.8503259452412,
-      "grad_norm": 1.4778438806533813,
-      "learning_rate": 6.611457286432161e-05,
-      "loss": 5.5917,
-      "step": 34228
-    },
-    {
-      "epoch": 17.85084745762712,
-      "grad_norm": 1.4843554496765137,
-      "learning_rate": 6.611356783919597e-05,
-      "loss": 5.2679,
-      "step": 34229
-    },
-    {
-      "epoch": 17.851368970013038,
-      "grad_norm": 1.5290446281433105,
-      "learning_rate": 6.611256281407035e-05,
-      "loss": 5.0466,
-      "step": 34230
-    },
-    {
-      "epoch": 17.851890482398957,
-      "grad_norm": 1.6280843019485474,
-      "learning_rate": 6.611155778894472e-05,
-      "loss": 5.3551,
-      "step": 34231
-    },
-    {
-      "epoch": 17.852411994784877,
-      "grad_norm": 1.5427229404449463,
-      "learning_rate": 6.61105527638191e-05,
-      "loss": 5.2826,
-      "step": 34232
-    },
-    {
-      "epoch": 17.852933507170796,
-      "grad_norm": 1.3898035287857056,
-      "learning_rate": 6.610954773869347e-05,
-      "loss": 5.5814,
-      "step": 34233
-    },
-    {
-      "epoch": 17.853455019556716,
-      "grad_norm": 1.409363031387329,
-      "learning_rate": 6.610854271356785e-05,
-      "loss": 5.5992,
-      "step": 34234
-    },
-    {
-      "epoch": 17.853976531942635,
-      "grad_norm": 1.4842053651809692,
-      "learning_rate": 6.610753768844221e-05,
-      "loss": 5.6128,
-      "step": 34235
-    },
-    {
-      "epoch": 17.85449804432855,
-      "grad_norm": 1.5171129703521729,
-      "learning_rate": 6.610653266331659e-05,
-      "loss": 5.0907,
-      "step": 34236
-    },
-    {
-      "epoch": 17.85501955671447,
-      "grad_norm": 1.5887953042984009,
-      "learning_rate": 6.610552763819096e-05,
-      "loss": 5.2574,
-      "step": 34237
-    },
-    {
-      "epoch": 17.85554106910039,
-      "grad_norm": 1.5280817747116089,
-      "learning_rate": 6.610452261306533e-05,
-      "loss": 5.4802,
-      "step": 34238
-    },
-    {
-      "epoch": 17.85606258148631,
-      "grad_norm": 1.3522779941558838,
-      "learning_rate": 6.61035175879397e-05,
-      "loss": 5.8811,
-      "step": 34239
-    },
-    {
-      "epoch": 17.85658409387223,
-      "grad_norm": 1.4000474214553833,
-      "learning_rate": 6.610251256281408e-05,
-      "loss": 5.6149,
-      "step": 34240
-    },
-    {
-      "epoch": 17.85710560625815,
-      "grad_norm": 1.5271005630493164,
-      "learning_rate": 6.610150753768844e-05,
-      "loss": 5.2299,
-      "step": 34241
-    },
-    {
-      "epoch": 17.857627118644068,
-      "grad_norm": 1.5306812524795532,
-      "learning_rate": 6.610050251256282e-05,
-      "loss": 5.1524,
-      "step": 34242
-    },
-    {
-      "epoch": 17.858148631029987,
-      "grad_norm": 1.5865569114685059,
-      "learning_rate": 6.60994974874372e-05,
-      "loss": 4.8119,
-      "step": 34243
-    },
-    {
-      "epoch": 17.858670143415907,
-      "grad_norm": 1.5330123901367188,
-      "learning_rate": 6.609849246231156e-05,
-      "loss": 5.6831,
-      "step": 34244
-    },
-    {
-      "epoch": 17.859191655801826,
-      "grad_norm": 1.5897490978240967,
-      "learning_rate": 6.609748743718594e-05,
-      "loss": 4.8735,
-      "step": 34245
-    },
-    {
-      "epoch": 17.859713168187746,
-      "grad_norm": 1.3767491579055786,
-      "learning_rate": 6.60964824120603e-05,
-      "loss": 5.5944,
-      "step": 34246
-    },
-    {
-      "epoch": 17.860234680573665,
-      "grad_norm": 1.501340627670288,
-      "learning_rate": 6.609547738693468e-05,
-      "loss": 5.7418,
-      "step": 34247
-    },
-    {
-      "epoch": 17.860756192959585,
-      "grad_norm": 1.5212697982788086,
-      "learning_rate": 6.609447236180904e-05,
-      "loss": 5.1356,
-      "step": 34248
-    },
-    {
-      "epoch": 17.8612777053455,
-      "grad_norm": 1.5443267822265625,
-      "learning_rate": 6.609346733668342e-05,
-      "loss": 5.4035,
-      "step": 34249
-    },
-    {
-      "epoch": 17.86179921773142,
-      "grad_norm": 1.4423816204071045,
-      "learning_rate": 6.609246231155779e-05,
-      "loss": 5.6803,
-      "step": 34250
-    },
-    {
-      "epoch": 17.86232073011734,
-      "grad_norm": 1.5504013299942017,
-      "learning_rate": 6.609145728643216e-05,
-      "loss": 5.4724,
-      "step": 34251
-    },
-    {
-      "epoch": 17.86284224250326,
-      "grad_norm": 1.4825057983398438,
-      "learning_rate": 6.609045226130653e-05,
-      "loss": 5.4633,
-      "step": 34252
-    },
-    {
-      "epoch": 17.86336375488918,
-      "grad_norm": 1.5905287265777588,
-      "learning_rate": 6.608944723618091e-05,
-      "loss": 4.8042,
-      "step": 34253
-    },
-    {
-      "epoch": 17.863885267275098,
-      "grad_norm": 1.644788146018982,
-      "learning_rate": 6.608844221105528e-05,
-      "loss": 5.2166,
-      "step": 34254
-    },
-    {
-      "epoch": 17.864406779661017,
-      "grad_norm": 1.5036271810531616,
-      "learning_rate": 6.608743718592965e-05,
-      "loss": 4.7817,
-      "step": 34255
-    },
-    {
-      "epoch": 17.864928292046937,
-      "grad_norm": 1.5006879568099976,
-      "learning_rate": 6.608643216080403e-05,
-      "loss": 5.3809,
-      "step": 34256
-    },
-    {
-      "epoch": 17.865449804432856,
-      "grad_norm": 1.6307622194290161,
-      "learning_rate": 6.608542713567839e-05,
-      "loss": 4.8047,
-      "step": 34257
-    },
-    {
-      "epoch": 17.865971316818776,
-      "grad_norm": 1.7038416862487793,
-      "learning_rate": 6.608442211055277e-05,
-      "loss": 4.8781,
-      "step": 34258
-    },
-    {
-      "epoch": 17.866492829204695,
-      "grad_norm": 1.5110116004943848,
-      "learning_rate": 6.608341708542713e-05,
-      "loss": 5.3178,
-      "step": 34259
-    },
-    {
-      "epoch": 17.86701434159061,
-      "grad_norm": 1.5513111352920532,
-      "learning_rate": 6.608241206030151e-05,
-      "loss": 5.2132,
-      "step": 34260
-    },
-    {
-      "epoch": 17.86753585397653,
-      "grad_norm": 1.484167218208313,
-      "learning_rate": 6.608140703517588e-05,
-      "loss": 5.394,
-      "step": 34261
-    },
-    {
-      "epoch": 17.86805736636245,
-      "grad_norm": 1.4068678617477417,
-      "learning_rate": 6.608040201005025e-05,
-      "loss": 5.244,
-      "step": 34262
-    },
-    {
-      "epoch": 17.86857887874837,
-      "grad_norm": 1.4469465017318726,
-      "learning_rate": 6.607939698492462e-05,
-      "loss": 5.0151,
-      "step": 34263
-    },
-    {
-      "epoch": 17.86910039113429,
-      "grad_norm": 1.479839563369751,
-      "learning_rate": 6.6078391959799e-05,
-      "loss": 5.507,
-      "step": 34264
-    },
-    {
-      "epoch": 17.86962190352021,
-      "grad_norm": 1.5507227182388306,
-      "learning_rate": 6.607738693467337e-05,
-      "loss": 5.1836,
-      "step": 34265
-    },
-    {
-      "epoch": 17.870143415906128,
-      "grad_norm": 1.3847005367279053,
-      "learning_rate": 6.607638190954775e-05,
-      "loss": 5.6938,
-      "step": 34266
-    },
-    {
-      "epoch": 17.870664928292047,
-      "grad_norm": 1.4785410165786743,
-      "learning_rate": 6.607537688442212e-05,
-      "loss": 5.4739,
-      "step": 34267
-    },
-    {
-      "epoch": 17.871186440677967,
-      "grad_norm": 1.5126827955245972,
-      "learning_rate": 6.607437185929649e-05,
-      "loss": 5.401,
-      "step": 34268
-    },
-    {
-      "epoch": 17.871707953063886,
-      "grad_norm": 1.4535393714904785,
-      "learning_rate": 6.607336683417086e-05,
-      "loss": 5.3828,
-      "step": 34269
-    },
-    {
-      "epoch": 17.872229465449806,
-      "grad_norm": 1.57003915309906,
-      "learning_rate": 6.607236180904522e-05,
-      "loss": 4.995,
-      "step": 34270
-    },
-    {
-      "epoch": 17.872750977835725,
-      "grad_norm": 1.4679940938949585,
-      "learning_rate": 6.60713567839196e-05,
-      "loss": 5.1758,
-      "step": 34271
-    },
-    {
-      "epoch": 17.87327249022164,
-      "grad_norm": 1.6485371589660645,
-      "learning_rate": 6.607035175879396e-05,
-      "loss": 5.1855,
-      "step": 34272
-    },
-    {
-      "epoch": 17.87379400260756,
-      "grad_norm": 1.5646096467971802,
-      "learning_rate": 6.606934673366834e-05,
-      "loss": 4.9806,
-      "step": 34273
-    },
-    {
-      "epoch": 17.87431551499348,
-      "grad_norm": 1.646519660949707,
-      "learning_rate": 6.606834170854272e-05,
-      "loss": 5.5725,
-      "step": 34274
-    },
-    {
-      "epoch": 17.8748370273794,
-      "grad_norm": 1.467394232749939,
-      "learning_rate": 6.60673366834171e-05,
-      "loss": 5.2025,
-      "step": 34275
-    },
-    {
-      "epoch": 17.87535853976532,
-      "grad_norm": 1.5502254962921143,
-      "learning_rate": 6.606633165829146e-05,
-      "loss": 5.1567,
-      "step": 34276
-    },
-    {
-      "epoch": 17.87588005215124,
-      "grad_norm": 1.545447587966919,
-      "learning_rate": 6.606532663316584e-05,
-      "loss": 5.2575,
-      "step": 34277
-    },
-    {
-      "epoch": 17.876401564537158,
-      "grad_norm": 1.4885976314544678,
-      "learning_rate": 6.60643216080402e-05,
-      "loss": 5.2117,
-      "step": 34278
-    },
-    {
-      "epoch": 17.876923076923077,
-      "grad_norm": 1.3436576128005981,
-      "learning_rate": 6.606331658291458e-05,
-      "loss": 5.69,
-      "step": 34279
-    },
-    {
-      "epoch": 17.877444589308997,
-      "grad_norm": 1.5994162559509277,
-      "learning_rate": 6.606231155778895e-05,
-      "loss": 5.28,
-      "step": 34280
-    },
-    {
-      "epoch": 17.877966101694916,
-      "grad_norm": 1.5656797885894775,
-      "learning_rate": 6.606130653266332e-05,
-      "loss": 5.2027,
-      "step": 34281
-    },
-    {
-      "epoch": 17.878487614080836,
-      "grad_norm": 1.5898449420928955,
-      "learning_rate": 6.606030150753769e-05,
-      "loss": 4.9238,
-      "step": 34282
-    },
-    {
-      "epoch": 17.879009126466755,
-      "grad_norm": 1.3857134580612183,
-      "learning_rate": 6.605929648241205e-05,
-      "loss": 5.5318,
-      "step": 34283
-    },
-    {
-      "epoch": 17.87953063885267,
-      "grad_norm": 1.4773529767990112,
-      "learning_rate": 6.605829145728643e-05,
-      "loss": 4.7084,
-      "step": 34284
-    },
-    {
-      "epoch": 17.88005215123859,
-      "grad_norm": 1.4176915884017944,
-      "learning_rate": 6.605728643216081e-05,
-      "loss": 4.9889,
-      "step": 34285
-    },
-    {
-      "epoch": 17.88057366362451,
-      "grad_norm": 1.5539906024932861,
-      "learning_rate": 6.605628140703519e-05,
-      "loss": 5.4707,
-      "step": 34286
-    },
-    {
-      "epoch": 17.88109517601043,
-      "grad_norm": 1.457578182220459,
-      "learning_rate": 6.605527638190955e-05,
-      "loss": 5.1965,
-      "step": 34287
-    },
-    {
-      "epoch": 17.88161668839635,
-      "grad_norm": 1.4837476015090942,
-      "learning_rate": 6.605427135678393e-05,
-      "loss": 5.6014,
-      "step": 34288
-    },
-    {
-      "epoch": 17.88213820078227,
-      "grad_norm": 1.4923933744430542,
-      "learning_rate": 6.605326633165829e-05,
-      "loss": 5.6288,
-      "step": 34289
-    },
-    {
-      "epoch": 17.882659713168188,
-      "grad_norm": 1.4792006015777588,
-      "learning_rate": 6.605226130653267e-05,
-      "loss": 5.2508,
-      "step": 34290
-    },
-    {
-      "epoch": 17.883181225554107,
-      "grad_norm": 1.4770957231521606,
-      "learning_rate": 6.605125628140703e-05,
-      "loss": 4.68,
-      "step": 34291
-    },
-    {
-      "epoch": 17.883702737940027,
-      "grad_norm": 1.4279297590255737,
-      "learning_rate": 6.605025125628141e-05,
-      "loss": 5.4131,
-      "step": 34292
-    },
-    {
-      "epoch": 17.884224250325946,
-      "grad_norm": 1.4511011838912964,
-      "learning_rate": 6.604924623115578e-05,
-      "loss": 5.6574,
-      "step": 34293
-    },
-    {
-      "epoch": 17.884745762711866,
-      "grad_norm": 1.6181857585906982,
-      "learning_rate": 6.604824120603015e-05,
-      "loss": 5.2491,
-      "step": 34294
-    },
-    {
-      "epoch": 17.885267275097785,
-      "grad_norm": 1.5415500402450562,
-      "learning_rate": 6.604723618090453e-05,
-      "loss": 5.0471,
-      "step": 34295
-    },
-    {
-      "epoch": 17.8857887874837,
-      "grad_norm": 1.475870132446289,
-      "learning_rate": 6.60462311557789e-05,
-      "loss": 5.2694,
-      "step": 34296
-    },
-    {
-      "epoch": 17.88631029986962,
-      "grad_norm": 1.546311616897583,
-      "learning_rate": 6.604522613065327e-05,
-      "loss": 5.3764,
-      "step": 34297
-    },
-    {
-      "epoch": 17.88683181225554,
-      "grad_norm": 1.5309628248214722,
-      "learning_rate": 6.604422110552764e-05,
-      "loss": 5.1227,
-      "step": 34298
-    },
-    {
-      "epoch": 17.88735332464146,
-      "grad_norm": 1.5206081867218018,
-      "learning_rate": 6.604321608040202e-05,
-      "loss": 5.5534,
-      "step": 34299
-    },
-    {
-      "epoch": 17.88787483702738,
-      "grad_norm": 1.305435299873352,
-      "learning_rate": 6.604221105527638e-05,
-      "loss": 4.8824,
-      "step": 34300
-    },
-    {
-      "epoch": 17.8883963494133,
-      "grad_norm": 1.4808733463287354,
-      "learning_rate": 6.604120603015076e-05,
-      "loss": 5.3793,
-      "step": 34301
-    },
-    {
-      "epoch": 17.888917861799218,
-      "grad_norm": 1.5884743928909302,
-      "learning_rate": 6.604020100502512e-05,
-      "loss": 4.5152,
-      "step": 34302
-    },
-    {
-      "epoch": 17.889439374185137,
-      "grad_norm": 1.5793156623840332,
-      "learning_rate": 6.60391959798995e-05,
-      "loss": 5.1627,
-      "step": 34303
-    },
-    {
-      "epoch": 17.889960886571057,
-      "grad_norm": 1.4587676525115967,
-      "learning_rate": 6.603819095477386e-05,
-      "loss": 5.1129,
-      "step": 34304
-    },
-    {
-      "epoch": 17.890482398956976,
-      "grad_norm": 1.6242718696594238,
-      "learning_rate": 6.603718592964824e-05,
-      "loss": 5.2755,
-      "step": 34305
-    },
-    {
-      "epoch": 17.891003911342896,
-      "grad_norm": 1.5037269592285156,
-      "learning_rate": 6.603618090452262e-05,
-      "loss": 5.4531,
-      "step": 34306
-    },
-    {
-      "epoch": 17.891525423728815,
-      "grad_norm": 1.7668569087982178,
-      "learning_rate": 6.6035175879397e-05,
-      "loss": 4.8486,
-      "step": 34307
-    },
-    {
-      "epoch": 17.89204693611473,
-      "grad_norm": 1.4393941164016724,
-      "learning_rate": 6.603417085427136e-05,
-      "loss": 4.9392,
-      "step": 34308
-    },
-    {
-      "epoch": 17.89256844850065,
-      "grad_norm": 1.5389803647994995,
-      "learning_rate": 6.603316582914573e-05,
-      "loss": 5.3733,
-      "step": 34309
-    },
-    {
-      "epoch": 17.89308996088657,
-      "grad_norm": 1.504636287689209,
-      "learning_rate": 6.60321608040201e-05,
-      "loss": 5.3089,
-      "step": 34310
-    },
-    {
-      "epoch": 17.89361147327249,
-      "grad_norm": 1.4715403318405151,
-      "learning_rate": 6.603115577889447e-05,
-      "loss": 5.288,
-      "step": 34311
-    },
-    {
-      "epoch": 17.89413298565841,
-      "grad_norm": 1.571930170059204,
-      "learning_rate": 6.603015075376885e-05,
-      "loss": 5.2378,
-      "step": 34312
-    },
-    {
-      "epoch": 17.89465449804433,
-      "grad_norm": 1.5191696882247925,
-      "learning_rate": 6.602914572864321e-05,
-      "loss": 5.3735,
-      "step": 34313
-    },
-    {
-      "epoch": 17.895176010430248,
-      "grad_norm": 1.438056230545044,
-      "learning_rate": 6.602814070351759e-05,
-      "loss": 5.5818,
-      "step": 34314
-    },
-    {
-      "epoch": 17.895697522816167,
-      "grad_norm": 1.4287487268447876,
-      "learning_rate": 6.602713567839197e-05,
-      "loss": 4.8871,
-      "step": 34315
-    },
-    {
-      "epoch": 17.896219035202087,
-      "grad_norm": 1.6304067373275757,
-      "learning_rate": 6.602613065326634e-05,
-      "loss": 4.9426,
-      "step": 34316
-    },
-    {
-      "epoch": 17.896740547588006,
-      "grad_norm": 1.4841417074203491,
-      "learning_rate": 6.602512562814071e-05,
-      "loss": 5.62,
-      "step": 34317
-    },
-    {
-      "epoch": 17.897262059973926,
-      "grad_norm": 1.5853943824768066,
-      "learning_rate": 6.602412060301509e-05,
-      "loss": 4.9016,
-      "step": 34318
-    },
-    {
-      "epoch": 17.89778357235984,
-      "grad_norm": 1.5410048961639404,
-      "learning_rate": 6.602311557788945e-05,
-      "loss": 5.7982,
-      "step": 34319
-    },
-    {
-      "epoch": 17.89830508474576,
-      "grad_norm": 1.4535164833068848,
-      "learning_rate": 6.602211055276383e-05,
-      "loss": 5.204,
-      "step": 34320
-    },
-    {
-      "epoch": 17.89882659713168,
-      "grad_norm": 1.477927803993225,
-      "learning_rate": 6.602110552763819e-05,
-      "loss": 4.9445,
-      "step": 34321
-    },
-    {
-      "epoch": 17.8993481095176,
-      "grad_norm": 1.4723994731903076,
-      "learning_rate": 6.602010050251256e-05,
-      "loss": 5.2182,
-      "step": 34322
-    },
-    {
-      "epoch": 17.89986962190352,
-      "grad_norm": 1.5383005142211914,
-      "learning_rate": 6.601909547738693e-05,
-      "loss": 5.1965,
-      "step": 34323
-    },
-    {
-      "epoch": 17.90039113428944,
-      "grad_norm": 1.4806584119796753,
-      "learning_rate": 6.60180904522613e-05,
-      "loss": 5.5848,
-      "step": 34324
-    },
-    {
-      "epoch": 17.90091264667536,
-      "grad_norm": 1.5270910263061523,
-      "learning_rate": 6.601708542713568e-05,
-      "loss": 4.8923,
-      "step": 34325
-    },
-    {
-      "epoch": 17.901434159061278,
-      "grad_norm": 1.4437390565872192,
-      "learning_rate": 6.601608040201005e-05,
-      "loss": 5.5137,
-      "step": 34326
-    },
-    {
-      "epoch": 17.901955671447197,
-      "grad_norm": 1.3967571258544922,
-      "learning_rate": 6.601507537688443e-05,
-      "loss": 5.7095,
-      "step": 34327
-    },
-    {
-      "epoch": 17.902477183833117,
-      "grad_norm": 1.466966986656189,
-      "learning_rate": 6.60140703517588e-05,
-      "loss": 5.2091,
-      "step": 34328
-    },
-    {
-      "epoch": 17.902998696219036,
-      "grad_norm": 1.5472118854522705,
-      "learning_rate": 6.601306532663317e-05,
-      "loss": 5.5125,
-      "step": 34329
-    },
-    {
-      "epoch": 17.903520208604956,
-      "grad_norm": 1.4890903234481812,
-      "learning_rate": 6.601206030150754e-05,
-      "loss": 5.1248,
-      "step": 34330
-    },
-    {
-      "epoch": 17.904041720990875,
-      "grad_norm": 1.4807450771331787,
-      "learning_rate": 6.601105527638192e-05,
-      "loss": 5.6028,
-      "step": 34331
-    },
-    {
-      "epoch": 17.90456323337679,
-      "grad_norm": 1.4130516052246094,
-      "learning_rate": 6.601005025125628e-05,
-      "loss": 4.68,
-      "step": 34332
-    },
-    {
-      "epoch": 17.90508474576271,
-      "grad_norm": 1.6209840774536133,
-      "learning_rate": 6.600904522613066e-05,
-      "loss": 4.883,
-      "step": 34333
-    },
-    {
-      "epoch": 17.90560625814863,
-      "grad_norm": 1.5422109365463257,
-      "learning_rate": 6.600804020100502e-05,
-      "loss": 5.2461,
-      "step": 34334
-    },
-    {
-      "epoch": 17.90612777053455,
-      "grad_norm": 1.5997483730316162,
-      "learning_rate": 6.60070351758794e-05,
-      "loss": 5.5061,
-      "step": 34335
-    },
-    {
-      "epoch": 17.90664928292047,
-      "grad_norm": 1.549616813659668,
-      "learning_rate": 6.600603015075378e-05,
-      "loss": 5.1344,
-      "step": 34336
-    },
-    {
-      "epoch": 17.90717079530639,
-      "grad_norm": 1.4406410455703735,
-      "learning_rate": 6.600502512562814e-05,
-      "loss": 4.6982,
-      "step": 34337
-    },
-    {
-      "epoch": 17.907692307692308,
-      "grad_norm": 1.4506278038024902,
-      "learning_rate": 6.600402010050252e-05,
-      "loss": 5.6724,
-      "step": 34338
-    },
-    {
-      "epoch": 17.908213820078227,
-      "grad_norm": 1.4972416162490845,
-      "learning_rate": 6.600301507537689e-05,
-      "loss": 5.3252,
-      "step": 34339
-    },
-    {
-      "epoch": 17.908735332464147,
-      "grad_norm": 1.525876522064209,
-      "learning_rate": 6.600201005025126e-05,
-      "loss": 5.3308,
-      "step": 34340
-    },
-    {
-      "epoch": 17.909256844850066,
-      "grad_norm": 1.4741392135620117,
-      "learning_rate": 6.600100502512563e-05,
-      "loss": 5.336,
-      "step": 34341
-    },
-    {
-      "epoch": 17.909778357235986,
-      "grad_norm": 1.383142113685608,
-      "learning_rate": 6.6e-05,
-      "loss": 5.6342,
-      "step": 34342
-    },
-    {
-      "epoch": 17.910299869621902,
-      "grad_norm": 1.5795763731002808,
-      "learning_rate": 6.599899497487437e-05,
-      "loss": 5.3659,
-      "step": 34343
-    },
-    {
-      "epoch": 17.91082138200782,
-      "grad_norm": 1.6061278581619263,
-      "learning_rate": 6.599798994974875e-05,
-      "loss": 4.8205,
-      "step": 34344
-    },
-    {
-      "epoch": 17.91134289439374,
-      "grad_norm": 1.5197378396987915,
-      "learning_rate": 6.599698492462311e-05,
-      "loss": 5.072,
-      "step": 34345
-    },
-    {
-      "epoch": 17.91186440677966,
-      "grad_norm": 1.4489303827285767,
-      "learning_rate": 6.599597989949749e-05,
-      "loss": 5.2671,
-      "step": 34346
-    },
-    {
-      "epoch": 17.91238591916558,
-      "grad_norm": 1.5640021562576294,
-      "learning_rate": 6.599497487437187e-05,
-      "loss": 5.0491,
-      "step": 34347
-    },
-    {
-      "epoch": 17.9129074315515,
-      "grad_norm": 1.4317678213119507,
-      "learning_rate": 6.599396984924623e-05,
-      "loss": 5.4981,
-      "step": 34348
-    },
-    {
-      "epoch": 17.91342894393742,
-      "grad_norm": 1.3751047849655151,
-      "learning_rate": 6.599296482412061e-05,
-      "loss": 5.2875,
-      "step": 34349
-    },
-    {
-      "epoch": 17.913950456323338,
-      "grad_norm": 1.4384973049163818,
-      "learning_rate": 6.599195979899497e-05,
-      "loss": 5.344,
-      "step": 34350
-    },
-    {
-      "epoch": 17.914471968709258,
-      "grad_norm": 1.5145318508148193,
-      "learning_rate": 6.599095477386935e-05,
-      "loss": 5.1262,
-      "step": 34351
-    },
-    {
-      "epoch": 17.914993481095177,
-      "grad_norm": 1.4930682182312012,
-      "learning_rate": 6.598994974874372e-05,
-      "loss": 5.2098,
-      "step": 34352
-    },
-    {
-      "epoch": 17.915514993481096,
-      "grad_norm": 1.473304033279419,
-      "learning_rate": 6.59889447236181e-05,
-      "loss": 5.4621,
-      "step": 34353
-    },
-    {
-      "epoch": 17.916036505867016,
-      "grad_norm": 1.450059175491333,
-      "learning_rate": 6.598793969849246e-05,
-      "loss": 5.723,
-      "step": 34354
-    },
-    {
-      "epoch": 17.916558018252932,
-      "grad_norm": 1.5123645067214966,
-      "learning_rate": 6.598693467336684e-05,
-      "loss": 5.2225,
-      "step": 34355
-    },
-    {
-      "epoch": 17.91707953063885,
-      "grad_norm": 1.4717576503753662,
-      "learning_rate": 6.598592964824121e-05,
-      "loss": 5.4957,
-      "step": 34356
-    },
-    {
-      "epoch": 17.91760104302477,
-      "grad_norm": 1.5175846815109253,
-      "learning_rate": 6.598492462311559e-05,
-      "loss": 4.6423,
-      "step": 34357
-    },
-    {
-      "epoch": 17.91812255541069,
-      "grad_norm": 1.5194475650787354,
-      "learning_rate": 6.598391959798996e-05,
-      "loss": 5.451,
-      "step": 34358
-    },
-    {
-      "epoch": 17.91864406779661,
-      "grad_norm": 1.444828987121582,
-      "learning_rate": 6.598291457286433e-05,
-      "loss": 5.2631,
-      "step": 34359
-    },
-    {
-      "epoch": 17.91916558018253,
-      "grad_norm": 1.4019463062286377,
-      "learning_rate": 6.59819095477387e-05,
-      "loss": 5.3991,
-      "step": 34360
-    },
-    {
-      "epoch": 17.91968709256845,
-      "grad_norm": 1.4619684219360352,
-      "learning_rate": 6.598090452261308e-05,
-      "loss": 5.4606,
-      "step": 34361
-    },
-    {
-      "epoch": 17.920208604954368,
-      "grad_norm": 1.4813050031661987,
-      "learning_rate": 6.597989949748744e-05,
-      "loss": 5.4336,
-      "step": 34362
-    },
-    {
-      "epoch": 17.920730117340288,
-      "grad_norm": 1.4892215728759766,
-      "learning_rate": 6.59788944723618e-05,
-      "loss": 5.1176,
-      "step": 34363
-    },
-    {
-      "epoch": 17.921251629726207,
-      "grad_norm": 1.3983666896820068,
-      "learning_rate": 6.597788944723618e-05,
-      "loss": 5.6227,
-      "step": 34364
-    },
-    {
-      "epoch": 17.921773142112126,
-      "grad_norm": 1.4170631170272827,
-      "learning_rate": 6.597688442211055e-05,
-      "loss": 5.3566,
-      "step": 34365
-    },
-    {
-      "epoch": 17.922294654498046,
-      "grad_norm": 1.4657639265060425,
-      "learning_rate": 6.597587939698492e-05,
-      "loss": 5.2947,
-      "step": 34366
-    },
-    {
-      "epoch": 17.922816166883962,
-      "grad_norm": 1.7334684133529663,
-      "learning_rate": 6.59748743718593e-05,
-      "loss": 4.9584,
-      "step": 34367
-    },
-    {
-      "epoch": 17.92333767926988,
-      "grad_norm": 1.6044389009475708,
-      "learning_rate": 6.597386934673368e-05,
-      "loss": 5.1436,
-      "step": 34368
-    },
-    {
-      "epoch": 17.9238591916558,
-      "grad_norm": 1.4594050645828247,
-      "learning_rate": 6.597286432160804e-05,
-      "loss": 5.6422,
-      "step": 34369
-    },
-    {
-      "epoch": 17.92438070404172,
-      "grad_norm": 1.5653172731399536,
-      "learning_rate": 6.597185929648242e-05,
-      "loss": 5.2493,
-      "step": 34370
-    },
-    {
-      "epoch": 17.92490221642764,
-      "grad_norm": 1.4057302474975586,
-      "learning_rate": 6.597085427135679e-05,
-      "loss": 5.6529,
-      "step": 34371
-    },
-    {
-      "epoch": 17.92542372881356,
-      "grad_norm": 1.3936306238174438,
-      "learning_rate": 6.596984924623116e-05,
-      "loss": 5.3327,
-      "step": 34372
-    },
-    {
-      "epoch": 17.92594524119948,
-      "grad_norm": 1.4910236597061157,
-      "learning_rate": 6.596884422110553e-05,
-      "loss": 5.3041,
-      "step": 34373
-    },
-    {
-      "epoch": 17.926466753585398,
-      "grad_norm": 1.5565458536148071,
-      "learning_rate": 6.59678391959799e-05,
-      "loss": 5.1271,
-      "step": 34374
-    },
-    {
-      "epoch": 17.926988265971318,
-      "grad_norm": 1.4892969131469727,
-      "learning_rate": 6.596683417085427e-05,
-      "loss": 5.5242,
-      "step": 34375
-    },
-    {
-      "epoch": 17.927509778357237,
-      "grad_norm": 1.438807725906372,
-      "learning_rate": 6.596582914572865e-05,
-      "loss": 5.5105,
-      "step": 34376
-    },
-    {
-      "epoch": 17.928031290743156,
-      "grad_norm": 1.6103761196136475,
-      "learning_rate": 6.596482412060303e-05,
-      "loss": 5.1425,
-      "step": 34377
-    },
-    {
-      "epoch": 17.928552803129076,
-      "grad_norm": 1.4833029508590698,
-      "learning_rate": 6.596381909547739e-05,
-      "loss": 5.5946,
-      "step": 34378
-    },
-    {
-      "epoch": 17.929074315514992,
-      "grad_norm": 1.445305347442627,
-      "learning_rate": 6.596281407035177e-05,
-      "loss": 5.6615,
-      "step": 34379
-    },
-    {
-      "epoch": 17.92959582790091,
-      "grad_norm": 1.556133508682251,
-      "learning_rate": 6.596180904522613e-05,
-      "loss": 5.048,
-      "step": 34380
-    },
-    {
-      "epoch": 17.93011734028683,
-      "grad_norm": 1.5786863565444946,
-      "learning_rate": 6.596080402010051e-05,
-      "loss": 4.9046,
-      "step": 34381
-    },
-    {
-      "epoch": 17.93063885267275,
-      "grad_norm": 1.5990296602249146,
-      "learning_rate": 6.595979899497487e-05,
-      "loss": 4.73,
-      "step": 34382
-    },
-    {
-      "epoch": 17.93116036505867,
-      "grad_norm": 1.4951269626617432,
-      "learning_rate": 6.595879396984925e-05,
-      "loss": 5.5411,
-      "step": 34383
-    },
-    {
-      "epoch": 17.93168187744459,
-      "grad_norm": 1.5027637481689453,
-      "learning_rate": 6.595778894472362e-05,
-      "loss": 5.2993,
-      "step": 34384
-    },
-    {
-      "epoch": 17.93220338983051,
-      "grad_norm": 1.5903836488723755,
-      "learning_rate": 6.5956783919598e-05,
-      "loss": 5.1278,
-      "step": 34385
-    },
-    {
-      "epoch": 17.932724902216428,
-      "grad_norm": 1.4844489097595215,
-      "learning_rate": 6.595577889447236e-05,
-      "loss": 5.1519,
-      "step": 34386
-    },
-    {
-      "epoch": 17.933246414602348,
-      "grad_norm": 1.5740009546279907,
-      "learning_rate": 6.595477386934674e-05,
-      "loss": 5.332,
-      "step": 34387
-    },
-    {
-      "epoch": 17.933767926988267,
-      "grad_norm": 1.5752532482147217,
-      "learning_rate": 6.595376884422111e-05,
-      "loss": 5.5796,
-      "step": 34388
-    },
-    {
-      "epoch": 17.934289439374187,
-      "grad_norm": 1.6262003183364868,
-      "learning_rate": 6.595276381909548e-05,
-      "loss": 5.1002,
-      "step": 34389
-    },
-    {
-      "epoch": 17.934810951760106,
-      "grad_norm": 1.558205485343933,
-      "learning_rate": 6.595175879396986e-05,
-      "loss": 5.14,
-      "step": 34390
-    },
-    {
-      "epoch": 17.935332464146022,
-      "grad_norm": 1.520970344543457,
-      "learning_rate": 6.595075376884422e-05,
-      "loss": 5.729,
-      "step": 34391
-    },
-    {
-      "epoch": 17.93585397653194,
-      "grad_norm": 1.5807749032974243,
-      "learning_rate": 6.59497487437186e-05,
-      "loss": 5.4206,
-      "step": 34392
-    },
-    {
-      "epoch": 17.93637548891786,
-      "grad_norm": 1.468370795249939,
-      "learning_rate": 6.594874371859296e-05,
-      "loss": 5.3128,
-      "step": 34393
-    },
-    {
-      "epoch": 17.93689700130378,
-      "grad_norm": 1.4803082942962646,
-      "learning_rate": 6.594773869346734e-05,
-      "loss": 5.4769,
-      "step": 34394
-    },
-    {
-      "epoch": 17.9374185136897,
-      "grad_norm": 1.4807759523391724,
-      "learning_rate": 6.59467336683417e-05,
-      "loss": 5.4717,
-      "step": 34395
-    },
-    {
-      "epoch": 17.93794002607562,
-      "grad_norm": 1.4631083011627197,
-      "learning_rate": 6.594572864321608e-05,
-      "loss": 5.6441,
-      "step": 34396
-    },
-    {
-      "epoch": 17.93846153846154,
-      "grad_norm": 1.7901641130447388,
-      "learning_rate": 6.594472361809046e-05,
-      "loss": 5.0724,
-      "step": 34397
-    },
-    {
-      "epoch": 17.938983050847458,
-      "grad_norm": 1.708248496055603,
-      "learning_rate": 6.594371859296484e-05,
-      "loss": 4.6501,
-      "step": 34398
-    },
-    {
-      "epoch": 17.939504563233378,
-      "grad_norm": 1.558355689048767,
-      "learning_rate": 6.59427135678392e-05,
-      "loss": 5.0007,
-      "step": 34399
-    },
-    {
-      "epoch": 17.940026075619297,
-      "grad_norm": 1.4739186763763428,
-      "learning_rate": 6.594170854271358e-05,
-      "loss": 5.6871,
-      "step": 34400
-    },
-    {
-      "epoch": 17.940547588005217,
-      "grad_norm": 1.566986322402954,
-      "learning_rate": 6.594070351758794e-05,
-      "loss": 5.3276,
-      "step": 34401
-    },
-    {
-      "epoch": 17.941069100391136,
-      "grad_norm": 1.5484529733657837,
-      "learning_rate": 6.593969849246231e-05,
-      "loss": 5.3662,
-      "step": 34402
-    },
-    {
-      "epoch": 17.941590612777052,
-      "grad_norm": 1.4450653791427612,
-      "learning_rate": 6.593869346733669e-05,
-      "loss": 5.3392,
-      "step": 34403
-    },
-    {
-      "epoch": 17.94211212516297,
-      "grad_norm": 1.424186110496521,
-      "learning_rate": 6.593768844221105e-05,
-      "loss": 5.3798,
-      "step": 34404
-    },
-    {
-      "epoch": 17.94263363754889,
-      "grad_norm": 1.4306763410568237,
-      "learning_rate": 6.593668341708543e-05,
-      "loss": 5.571,
-      "step": 34405
-    },
-    {
-      "epoch": 17.94315514993481,
-      "grad_norm": 1.423705816268921,
-      "learning_rate": 6.593567839195979e-05,
-      "loss": 5.7543,
-      "step": 34406
-    },
-    {
-      "epoch": 17.94367666232073,
-      "grad_norm": 1.462921380996704,
-      "learning_rate": 6.593467336683417e-05,
-      "loss": 5.2965,
-      "step": 34407
-    },
-    {
-      "epoch": 17.94419817470665,
-      "grad_norm": 1.4385993480682373,
-      "learning_rate": 6.593366834170855e-05,
-      "loss": 4.5254,
-      "step": 34408
-    },
-    {
-      "epoch": 17.94471968709257,
-      "grad_norm": 1.4086865186691284,
-      "learning_rate": 6.593266331658293e-05,
-      "loss": 5.2288,
-      "step": 34409
-    },
-    {
-      "epoch": 17.945241199478488,
-      "grad_norm": 1.443440318107605,
-      "learning_rate": 6.593165829145729e-05,
-      "loss": 5.2817,
-      "step": 34410
-    },
-    {
-      "epoch": 17.945762711864408,
-      "grad_norm": 1.417703628540039,
-      "learning_rate": 6.593065326633167e-05,
-      "loss": 5.4867,
-      "step": 34411
-    },
-    {
-      "epoch": 17.946284224250327,
-      "grad_norm": 1.394863247871399,
-      "learning_rate": 6.592964824120603e-05,
-      "loss": 5.5516,
-      "step": 34412
-    },
-    {
-      "epoch": 17.946805736636247,
-      "grad_norm": 1.509758710861206,
-      "learning_rate": 6.592864321608041e-05,
-      "loss": 4.4577,
-      "step": 34413
-    },
-    {
-      "epoch": 17.947327249022166,
-      "grad_norm": 1.4747332334518433,
-      "learning_rate": 6.592763819095477e-05,
-      "loss": 5.5588,
-      "step": 34414
-    },
-    {
-      "epoch": 17.947848761408082,
-      "grad_norm": 1.4741194248199463,
-      "learning_rate": 6.592663316582914e-05,
-      "loss": 5.5316,
-      "step": 34415
-    },
-    {
-      "epoch": 17.948370273794,
-      "grad_norm": 1.452453374862671,
-      "learning_rate": 6.592562814070352e-05,
-      "loss": 4.9555,
-      "step": 34416
-    },
-    {
-      "epoch": 17.94889178617992,
-      "grad_norm": 1.4843635559082031,
-      "learning_rate": 6.59246231155779e-05,
-      "loss": 4.9671,
-      "step": 34417
-    },
-    {
-      "epoch": 17.94941329856584,
-      "grad_norm": 1.475831389427185,
-      "learning_rate": 6.592361809045227e-05,
-      "loss": 4.971,
-      "step": 34418
-    },
-    {
-      "epoch": 17.94993481095176,
-      "grad_norm": 1.4550071954727173,
-      "learning_rate": 6.592261306532664e-05,
-      "loss": 5.0599,
-      "step": 34419
-    },
-    {
-      "epoch": 17.95045632333768,
-      "grad_norm": 1.414887547492981,
-      "learning_rate": 6.592160804020101e-05,
-      "loss": 5.196,
-      "step": 34420
-    },
-    {
-      "epoch": 17.9509778357236,
-      "grad_norm": 1.5802339315414429,
-      "learning_rate": 6.592060301507538e-05,
-      "loss": 5.207,
-      "step": 34421
-    },
-    {
-      "epoch": 17.951499348109518,
-      "grad_norm": 1.4896469116210938,
-      "learning_rate": 6.591959798994976e-05,
-      "loss": 5.2987,
-      "step": 34422
-    },
-    {
-      "epoch": 17.952020860495438,
-      "grad_norm": 1.4796608686447144,
-      "learning_rate": 6.591859296482412e-05,
-      "loss": 5.2857,
-      "step": 34423
-    },
-    {
-      "epoch": 17.952542372881357,
-      "grad_norm": 1.4489291906356812,
-      "learning_rate": 6.59175879396985e-05,
-      "loss": 5.1175,
-      "step": 34424
-    },
-    {
-      "epoch": 17.953063885267277,
-      "grad_norm": 1.3709566593170166,
-      "learning_rate": 6.591658291457286e-05,
-      "loss": 5.6324,
-      "step": 34425
-    },
-    {
-      "epoch": 17.953585397653193,
-      "grad_norm": 1.511335849761963,
-      "learning_rate": 6.591557788944724e-05,
-      "loss": 5.4055,
-      "step": 34426
-    },
-    {
-      "epoch": 17.954106910039112,
-      "grad_norm": 1.529376745223999,
-      "learning_rate": 6.59145728643216e-05,
-      "loss": 5.0681,
-      "step": 34427
-    },
-    {
-      "epoch": 17.95462842242503,
-      "grad_norm": 1.6351845264434814,
-      "learning_rate": 6.591356783919598e-05,
-      "loss": 5.1163,
-      "step": 34428
-    },
-    {
-      "epoch": 17.95514993481095,
-      "grad_norm": 1.4918276071548462,
-      "learning_rate": 6.591256281407036e-05,
-      "loss": 4.6064,
-      "step": 34429
-    },
-    {
-      "epoch": 17.95567144719687,
-      "grad_norm": 1.4514238834381104,
-      "learning_rate": 6.591155778894473e-05,
-      "loss": 5.4008,
-      "step": 34430
-    },
-    {
-      "epoch": 17.95619295958279,
-      "grad_norm": 1.5568983554840088,
-      "learning_rate": 6.59105527638191e-05,
-      "loss": 5.6156,
-      "step": 34431
-    },
-    {
-      "epoch": 17.95671447196871,
-      "grad_norm": 1.4124045372009277,
-      "learning_rate": 6.590954773869347e-05,
-      "loss": 5.15,
-      "step": 34432
-    },
-    {
-      "epoch": 17.95723598435463,
-      "grad_norm": 1.6143410205841064,
-      "learning_rate": 6.590854271356785e-05,
-      "loss": 5.3033,
-      "step": 34433
-    },
-    {
-      "epoch": 17.957757496740548,
-      "grad_norm": 1.4711146354675293,
-      "learning_rate": 6.590753768844221e-05,
-      "loss": 5.1967,
-      "step": 34434
-    },
-    {
-      "epoch": 17.958279009126468,
-      "grad_norm": 1.3660919666290283,
-      "learning_rate": 6.590653266331659e-05,
-      "loss": 4.7957,
-      "step": 34435
-    },
-    {
-      "epoch": 17.958800521512387,
-      "grad_norm": 1.4333277940750122,
-      "learning_rate": 6.590552763819095e-05,
-      "loss": 5.4178,
-      "step": 34436
-    },
-    {
-      "epoch": 17.959322033898307,
-      "grad_norm": 1.4448387622833252,
-      "learning_rate": 6.590452261306533e-05,
-      "loss": 5.4867,
-      "step": 34437
-    },
-    {
-      "epoch": 17.959843546284226,
-      "grad_norm": 1.556132197380066,
-      "learning_rate": 6.59035175879397e-05,
-      "loss": 5.5271,
-      "step": 34438
-    },
-    {
-      "epoch": 17.960365058670142,
-      "grad_norm": 1.5021482706069946,
-      "learning_rate": 6.590251256281407e-05,
-      "loss": 4.9552,
-      "step": 34439
-    },
-    {
-      "epoch": 17.96088657105606,
-      "grad_norm": 1.3885306119918823,
-      "learning_rate": 6.590150753768845e-05,
-      "loss": 5.7048,
-      "step": 34440
-    },
-    {
-      "epoch": 17.96140808344198,
-      "grad_norm": 1.4817050695419312,
-      "learning_rate": 6.590050251256281e-05,
-      "loss": 5.4861,
-      "step": 34441
-    },
-    {
-      "epoch": 17.9619295958279,
-      "grad_norm": 1.4771947860717773,
-      "learning_rate": 6.589949748743719e-05,
-      "loss": 5.8164,
-      "step": 34442
-    },
-    {
-      "epoch": 17.96245110821382,
-      "grad_norm": 1.427932620048523,
-      "learning_rate": 6.589849246231156e-05,
-      "loss": 4.9818,
-      "step": 34443
-    },
-    {
-      "epoch": 17.96297262059974,
-      "grad_norm": 1.4942162036895752,
-      "learning_rate": 6.589748743718593e-05,
-      "loss": 5.1518,
-      "step": 34444
-    },
-    {
-      "epoch": 17.96349413298566,
-      "grad_norm": 1.4958653450012207,
-      "learning_rate": 6.58964824120603e-05,
-      "loss": 5.401,
-      "step": 34445
-    },
-    {
-      "epoch": 17.96401564537158,
-      "grad_norm": 1.6489388942718506,
-      "learning_rate": 6.589547738693468e-05,
-      "loss": 4.5806,
-      "step": 34446
-    },
-    {
-      "epoch": 17.964537157757498,
-      "grad_norm": 2.018339157104492,
-      "learning_rate": 6.589447236180904e-05,
-      "loss": 4.3656,
-      "step": 34447
-    },
-    {
-      "epoch": 17.965058670143417,
-      "grad_norm": 1.5560673475265503,
-      "learning_rate": 6.589346733668342e-05,
-      "loss": 5.3096,
-      "step": 34448
-    },
-    {
-      "epoch": 17.965580182529337,
-      "grad_norm": 1.4995695352554321,
-      "learning_rate": 6.58924623115578e-05,
-      "loss": 5.0362,
-      "step": 34449
-    },
-    {
-      "epoch": 17.966101694915253,
-      "grad_norm": 1.547061800956726,
-      "learning_rate": 6.589145728643217e-05,
-      "loss": 4.7571,
-      "step": 34450
-    },
-    {
-      "epoch": 17.966623207301172,
-      "grad_norm": 1.4818675518035889,
-      "learning_rate": 6.589045226130654e-05,
-      "loss": 5.5762,
-      "step": 34451
-    },
-    {
-      "epoch": 17.96714471968709,
-      "grad_norm": 1.5967682600021362,
-      "learning_rate": 6.588944723618092e-05,
-      "loss": 4.9524,
-      "step": 34452
-    },
-    {
-      "epoch": 17.96766623207301,
-      "grad_norm": 1.4125715494155884,
-      "learning_rate": 6.588844221105528e-05,
-      "loss": 5.1431,
-      "step": 34453
-    },
-    {
-      "epoch": 17.96818774445893,
-      "grad_norm": 1.3975938558578491,
-      "learning_rate": 6.588743718592966e-05,
-      "loss": 5.5986,
-      "step": 34454
-    },
-    {
-      "epoch": 17.96870925684485,
-      "grad_norm": 1.4413443803787231,
-      "learning_rate": 6.588643216080402e-05,
-      "loss": 5.5699,
-      "step": 34455
-    },
-    {
-      "epoch": 17.96923076923077,
-      "grad_norm": 1.505035400390625,
-      "learning_rate": 6.588542713567839e-05,
-      "loss": 4.9469,
-      "step": 34456
-    },
-    {
-      "epoch": 17.96975228161669,
-      "grad_norm": 1.6174718141555786,
-      "learning_rate": 6.588442211055276e-05,
-      "loss": 5.0715,
-      "step": 34457
-    },
-    {
-      "epoch": 17.97027379400261,
-      "grad_norm": 1.4666956663131714,
-      "learning_rate": 6.588341708542713e-05,
-      "loss": 4.8327,
-      "step": 34458
-    },
-    {
-      "epoch": 17.970795306388528,
-      "grad_norm": 1.6872104406356812,
-      "learning_rate": 6.58824120603015e-05,
-      "loss": 5.0049,
-      "step": 34459
-    },
-    {
-      "epoch": 17.971316818774447,
-      "grad_norm": 1.4665522575378418,
-      "learning_rate": 6.588140703517588e-05,
-      "loss": 5.1514,
-      "step": 34460
-    },
-    {
-      "epoch": 17.971838331160367,
-      "grad_norm": 1.442374587059021,
-      "learning_rate": 6.588040201005026e-05,
-      "loss": 4.9683,
-      "step": 34461
-    },
-    {
-      "epoch": 17.972359843546283,
-      "grad_norm": 1.5501861572265625,
-      "learning_rate": 6.587939698492463e-05,
-      "loss": 5.0206,
-      "step": 34462
-    },
-    {
-      "epoch": 17.972881355932202,
-      "grad_norm": 1.5179953575134277,
-      "learning_rate": 6.5878391959799e-05,
-      "loss": 5.3598,
-      "step": 34463
-    },
-    {
-      "epoch": 17.97340286831812,
-      "grad_norm": 1.5812222957611084,
-      "learning_rate": 6.587738693467337e-05,
-      "loss": 4.858,
-      "step": 34464
-    },
-    {
-      "epoch": 17.97392438070404,
-      "grad_norm": 1.5433059930801392,
-      "learning_rate": 6.587638190954775e-05,
-      "loss": 5.112,
-      "step": 34465
-    },
-    {
-      "epoch": 17.97444589308996,
-      "grad_norm": 1.4881670475006104,
-      "learning_rate": 6.587537688442211e-05,
-      "loss": 5.2423,
-      "step": 34466
-    },
-    {
-      "epoch": 17.97496740547588,
-      "grad_norm": 1.480857014656067,
-      "learning_rate": 6.587437185929649e-05,
-      "loss": 5.2695,
-      "step": 34467
-    },
-    {
-      "epoch": 17.9754889178618,
-      "grad_norm": 1.5090144872665405,
-      "learning_rate": 6.587336683417085e-05,
-      "loss": 5.5395,
-      "step": 34468
-    },
-    {
-      "epoch": 17.97601043024772,
-      "grad_norm": 1.6066844463348389,
-      "learning_rate": 6.587236180904523e-05,
-      "loss": 5.0667,
-      "step": 34469
-    },
-    {
-      "epoch": 17.97653194263364,
-      "grad_norm": 1.5069674253463745,
-      "learning_rate": 6.587135678391961e-05,
-      "loss": 5.3961,
-      "step": 34470
-    },
-    {
-      "epoch": 17.977053455019558,
-      "grad_norm": 1.6355369091033936,
-      "learning_rate": 6.587035175879397e-05,
-      "loss": 4.2723,
-      "step": 34471
-    },
-    {
-      "epoch": 17.977574967405477,
-      "grad_norm": 1.4382572174072266,
-      "learning_rate": 6.586934673366835e-05,
-      "loss": 4.7113,
-      "step": 34472
-    },
-    {
-      "epoch": 17.978096479791397,
-      "grad_norm": 1.3749853372573853,
-      "learning_rate": 6.586834170854271e-05,
-      "loss": 5.6679,
-      "step": 34473
-    },
-    {
-      "epoch": 17.978617992177313,
-      "grad_norm": 1.6042677164077759,
-      "learning_rate": 6.586733668341709e-05,
-      "loss": 5.1639,
-      "step": 34474
-    },
-    {
-      "epoch": 17.979139504563232,
-      "grad_norm": 1.4687360525131226,
-      "learning_rate": 6.586633165829146e-05,
-      "loss": 5.4139,
-      "step": 34475
-    },
-    {
-      "epoch": 17.97966101694915,
-      "grad_norm": 1.4729738235473633,
-      "learning_rate": 6.586532663316583e-05,
-      "loss": 5.3963,
-      "step": 34476
-    },
-    {
-      "epoch": 17.98018252933507,
-      "grad_norm": 1.4514473676681519,
-      "learning_rate": 6.58643216080402e-05,
-      "loss": 5.5556,
-      "step": 34477
-    },
-    {
-      "epoch": 17.98070404172099,
-      "grad_norm": 1.4739985466003418,
-      "learning_rate": 6.586331658291458e-05,
-      "loss": 5.3095,
-      "step": 34478
-    },
-    {
-      "epoch": 17.98122555410691,
-      "grad_norm": 1.5298302173614502,
-      "learning_rate": 6.586231155778894e-05,
-      "loss": 5.3047,
-      "step": 34479
-    },
-    {
-      "epoch": 17.98174706649283,
-      "grad_norm": 1.4847583770751953,
-      "learning_rate": 6.586130653266332e-05,
-      "loss": 5.5505,
-      "step": 34480
-    },
-    {
-      "epoch": 17.98226857887875,
-      "grad_norm": 1.4698156118392944,
-      "learning_rate": 6.58603015075377e-05,
-      "loss": 5.1537,
-      "step": 34481
-    },
-    {
-      "epoch": 17.98279009126467,
-      "grad_norm": 1.6611104011535645,
-      "learning_rate": 6.585929648241206e-05,
-      "loss": 5.0434,
-      "step": 34482
-    },
-    {
-      "epoch": 17.983311603650588,
-      "grad_norm": 1.413561463356018,
-      "learning_rate": 6.585829145728644e-05,
-      "loss": 5.5948,
-      "step": 34483
-    },
-    {
-      "epoch": 17.983833116036507,
-      "grad_norm": 1.6320198774337769,
-      "learning_rate": 6.58572864321608e-05,
-      "loss": 5.3974,
-      "step": 34484
-    },
-    {
-      "epoch": 17.984354628422427,
-      "grad_norm": 1.5658491849899292,
-      "learning_rate": 6.585628140703518e-05,
-      "loss": 5.1025,
-      "step": 34485
-    },
-    {
-      "epoch": 17.984876140808343,
-      "grad_norm": 1.4075894355773926,
-      "learning_rate": 6.585527638190954e-05,
-      "loss": 5.5595,
-      "step": 34486
-    },
-    {
-      "epoch": 17.985397653194262,
-      "grad_norm": 1.4674696922302246,
-      "learning_rate": 6.585427135678392e-05,
-      "loss": 5.7313,
-      "step": 34487
-    },
-    {
-      "epoch": 17.98591916558018,
-      "grad_norm": 1.478952169418335,
-      "learning_rate": 6.585326633165829e-05,
-      "loss": 5.3665,
-      "step": 34488
-    },
-    {
-      "epoch": 17.9864406779661,
-      "grad_norm": 1.503686547279358,
-      "learning_rate": 6.585226130653266e-05,
-      "loss": 5.2726,
-      "step": 34489
-    },
-    {
-      "epoch": 17.98696219035202,
-      "grad_norm": 1.5387918949127197,
-      "learning_rate": 6.585125628140704e-05,
-      "loss": 5.4846,
-      "step": 34490
-    },
-    {
-      "epoch": 17.98748370273794,
-      "grad_norm": 1.6141655445098877,
-      "learning_rate": 6.585025125628142e-05,
-      "loss": 5.0813,
-      "step": 34491
-    },
-    {
-      "epoch": 17.98800521512386,
-      "grad_norm": 1.4289445877075195,
-      "learning_rate": 6.584924623115578e-05,
-      "loss": 5.5869,
-      "step": 34492
-    },
-    {
-      "epoch": 17.98852672750978,
-      "grad_norm": 1.3425426483154297,
-      "learning_rate": 6.584824120603016e-05,
-      "loss": 5.4578,
-      "step": 34493
-    },
-    {
-      "epoch": 17.9890482398957,
-      "grad_norm": 1.4658864736557007,
-      "learning_rate": 6.584723618090453e-05,
-      "loss": 4.7531,
-      "step": 34494
-    },
-    {
-      "epoch": 17.989569752281618,
-      "grad_norm": 1.4604839086532593,
-      "learning_rate": 6.584623115577889e-05,
-      "loss": 5.3061,
-      "step": 34495
-    },
-    {
-      "epoch": 17.990091264667537,
-      "grad_norm": 1.4926074743270874,
-      "learning_rate": 6.584522613065327e-05,
-      "loss": 5.0724,
-      "step": 34496
-    },
-    {
-      "epoch": 17.990612777053457,
-      "grad_norm": 1.4288212060928345,
-      "learning_rate": 6.584422110552763e-05,
-      "loss": 4.904,
-      "step": 34497
-    },
-    {
-      "epoch": 17.991134289439373,
-      "grad_norm": 1.5141072273254395,
-      "learning_rate": 6.584321608040201e-05,
-      "loss": 5.4866,
-      "step": 34498
-    },
-    {
-      "epoch": 17.991655801825292,
-      "grad_norm": 1.503017544746399,
-      "learning_rate": 6.584221105527638e-05,
-      "loss": 5.4056,
-      "step": 34499
-    },
-    {
-      "epoch": 17.99217731421121,
-      "grad_norm": 1.513900637626648,
-      "learning_rate": 6.584120603015075e-05,
-      "loss": 5.3111,
-      "step": 34500
-    },
-    {
-      "epoch": 17.99269882659713,
-      "grad_norm": 1.5654958486557007,
-      "learning_rate": 6.584020100502513e-05,
-      "loss": 5.1362,
-      "step": 34501
-    },
-    {
-      "epoch": 17.99322033898305,
-      "grad_norm": 1.4661364555358887,
-      "learning_rate": 6.583919597989951e-05,
-      "loss": 5.3756,
-      "step": 34502
-    },
-    {
-      "epoch": 17.99374185136897,
-      "grad_norm": 1.4246188402175903,
-      "learning_rate": 6.583819095477387e-05,
-      "loss": 5.4712,
-      "step": 34503
-    },
-    {
-      "epoch": 17.99426336375489,
-      "grad_norm": 1.4901608228683472,
-      "learning_rate": 6.583718592964825e-05,
-      "loss": 5.2447,
-      "step": 34504
-    },
-    {
-      "epoch": 17.99478487614081,
-      "grad_norm": 1.4214166402816772,
-      "learning_rate": 6.583618090452262e-05,
-      "loss": 5.5812,
-      "step": 34505
-    },
-    {
-      "epoch": 17.99530638852673,
-      "grad_norm": 1.4669297933578491,
-      "learning_rate": 6.583517587939699e-05,
-      "loss": 5.6186,
-      "step": 34506
-    },
-    {
-      "epoch": 17.995827900912648,
-      "grad_norm": 1.5792323350906372,
-      "learning_rate": 6.583417085427136e-05,
-      "loss": 5.7059,
-      "step": 34507
-    },
-    {
-      "epoch": 17.996349413298567,
-      "grad_norm": 1.542504072189331,
-      "learning_rate": 6.583316582914572e-05,
-      "loss": 5.6651,
-      "step": 34508
-    },
-    {
-      "epoch": 17.996870925684483,
-      "grad_norm": 1.3837231397628784,
-      "learning_rate": 6.58321608040201e-05,
-      "loss": 5.4374,
-      "step": 34509
-    },
-    {
-      "epoch": 17.997392438070403,
-      "grad_norm": 1.4747819900512695,
-      "learning_rate": 6.583115577889448e-05,
-      "loss": 5.2826,
-      "step": 34510
-    },
-    {
-      "epoch": 17.997913950456322,
-      "grad_norm": 1.5706905126571655,
-      "learning_rate": 6.583015075376886e-05,
-      "loss": 5.6431,
-      "step": 34511
-    },
-    {
-      "epoch": 17.99843546284224,
-      "grad_norm": 1.495792269706726,
-      "learning_rate": 6.582914572864322e-05,
-      "loss": 5.5918,
-      "step": 34512
-    },
-    {
-      "epoch": 17.99895697522816,
-      "grad_norm": 1.5378512144088745,
-      "learning_rate": 6.58281407035176e-05,
-      "loss": 5.5235,
-      "step": 34513
-    },
-    {
-      "epoch": 17.99947848761408,
-      "grad_norm": 1.5648638010025024,
-      "learning_rate": 6.582713567839196e-05,
-      "loss": 5.0495,
-      "step": 34514
-    },
-    {
-      "epoch": 18.0,
-      "grad_norm": 1.7369029521942139,
-      "learning_rate": 6.582613065326634e-05,
-      "loss": 4.642,
-      "step": 34515
-    },
-    {
-      "epoch": 18.00052151238592,
-      "grad_norm": 1.535120964050293,
-      "learning_rate": 6.58251256281407e-05,
-      "loss": 5.4772,
-      "step": 34516
-    },
-    {
-      "epoch": 18.00104302477184,
-      "grad_norm": 1.4991706609725952,
-      "learning_rate": 6.582412060301508e-05,
-      "loss": 4.7877,
-      "step": 34517
-    },
-    {
-      "epoch": 18.00156453715776,
-      "grad_norm": 1.4832313060760498,
-      "learning_rate": 6.582311557788945e-05,
-      "loss": 5.1773,
-      "step": 34518
-    },
-    {
-      "epoch": 18.002086049543678,
-      "grad_norm": 1.4328093528747559,
-      "learning_rate": 6.582211055276382e-05,
-      "loss": 5.7447,
-      "step": 34519
-    },
-    {
-      "epoch": 18.002607561929597,
-      "grad_norm": 1.4386639595031738,
-      "learning_rate": 6.582110552763819e-05,
-      "loss": 5.6265,
-      "step": 34520
-    },
-    {
-      "epoch": 18.003129074315513,
-      "grad_norm": 1.546228289604187,
-      "learning_rate": 6.582010050251257e-05,
-      "loss": 5.7554,
-      "step": 34521
-    },
-    {
-      "epoch": 18.003650586701433,
-      "grad_norm": 1.3976783752441406,
-      "learning_rate": 6.581909547738694e-05,
-      "loss": 5.6754,
-      "step": 34522
-    },
-    {
-      "epoch": 18.004172099087352,
-      "grad_norm": 1.477152705192566,
-      "learning_rate": 6.581809045226131e-05,
-      "loss": 5.3829,
-      "step": 34523
-    },
-    {
-      "epoch": 18.00469361147327,
-      "grad_norm": 1.3992375135421753,
-      "learning_rate": 6.581708542713569e-05,
-      "loss": 5.6274,
-      "step": 34524
-    },
-    {
-      "epoch": 18.00521512385919,
-      "grad_norm": 1.6800516843795776,
-      "learning_rate": 6.581608040201005e-05,
-      "loss": 5.2934,
-      "step": 34525
-    },
-    {
-      "epoch": 18.00573663624511,
-      "grad_norm": 1.4416236877441406,
-      "learning_rate": 6.581507537688443e-05,
-      "loss": 5.7029,
-      "step": 34526
-    },
-    {
-      "epoch": 18.00625814863103,
-      "grad_norm": 1.5177481174468994,
-      "learning_rate": 6.581407035175879e-05,
-      "loss": 5.4666,
-      "step": 34527
-    },
-    {
-      "epoch": 18.00677966101695,
-      "grad_norm": 1.5266560316085815,
-      "learning_rate": 6.581306532663317e-05,
-      "loss": 5.0459,
-      "step": 34528
-    },
-    {
-      "epoch": 18.00730117340287,
-      "grad_norm": 1.4611364603042603,
-      "learning_rate": 6.581206030150753e-05,
-      "loss": 5.114,
-      "step": 34529
-    },
-    {
-      "epoch": 18.00782268578879,
-      "grad_norm": 1.6384785175323486,
-      "learning_rate": 6.581105527638191e-05,
-      "loss": 4.9461,
-      "step": 34530
-    },
-    {
-      "epoch": 18.008344198174708,
-      "grad_norm": 1.5622379779815674,
-      "learning_rate": 6.581005025125629e-05,
-      "loss": 5.1997,
-      "step": 34531
-    },
-    {
-      "epoch": 18.008865710560627,
-      "grad_norm": 1.5216342210769653,
-      "learning_rate": 6.580904522613067e-05,
-      "loss": 5.3772,
-      "step": 34532
-    },
-    {
-      "epoch": 18.009387222946543,
-      "grad_norm": 1.5181798934936523,
-      "learning_rate": 6.580804020100503e-05,
-      "loss": 5.6937,
-      "step": 34533
-    },
-    {
-      "epoch": 18.009908735332463,
-      "grad_norm": 1.4834717512130737,
-      "learning_rate": 6.58070351758794e-05,
-      "loss": 5.6316,
-      "step": 34534
-    },
-    {
-      "epoch": 18.010430247718382,
-      "grad_norm": 1.5943796634674072,
-      "learning_rate": 6.580603015075377e-05,
-      "loss": 4.9181,
-      "step": 34535
-    },
-    {
-      "epoch": 18.0109517601043,
-      "grad_norm": 1.5342224836349487,
-      "learning_rate": 6.580502512562814e-05,
-      "loss": 5.6433,
-      "step": 34536
-    },
-    {
-      "epoch": 18.01147327249022,
-      "grad_norm": 1.4739564657211304,
-      "learning_rate": 6.580402010050252e-05,
-      "loss": 5.6559,
-      "step": 34537
-    },
-    {
-      "epoch": 18.01199478487614,
-      "grad_norm": 1.4124770164489746,
-      "learning_rate": 6.580301507537688e-05,
-      "loss": 5.7322,
-      "step": 34538
-    },
-    {
-      "epoch": 18.01251629726206,
-      "grad_norm": 1.5247951745986938,
-      "learning_rate": 6.580201005025126e-05,
-      "loss": 5.2637,
-      "step": 34539
-    },
-    {
-      "epoch": 18.01303780964798,
-      "grad_norm": 1.7152971029281616,
-      "learning_rate": 6.580100502512562e-05,
-      "loss": 5.2072,
-      "step": 34540
-    },
-    {
-      "epoch": 18.0135593220339,
-      "grad_norm": 1.5179792642593384,
-      "learning_rate": 6.58e-05,
-      "loss": 5.5805,
-      "step": 34541
-    },
-    {
-      "epoch": 18.01408083441982,
-      "grad_norm": 1.5255911350250244,
-      "learning_rate": 6.579899497487438e-05,
-      "loss": 5.1937,
-      "step": 34542
-    },
-    {
-      "epoch": 18.014602346805738,
-      "grad_norm": 1.4795796871185303,
-      "learning_rate": 6.579798994974876e-05,
-      "loss": 5.7144,
-      "step": 34543
-    },
-    {
-      "epoch": 18.015123859191657,
-      "grad_norm": 1.5454121828079224,
-      "learning_rate": 6.579698492462312e-05,
-      "loss": 5.1325,
-      "step": 34544
-    },
-    {
-      "epoch": 18.015645371577573,
-      "grad_norm": 1.5040340423583984,
-      "learning_rate": 6.57959798994975e-05,
-      "loss": 4.9165,
-      "step": 34545
-    },
-    {
-      "epoch": 18.016166883963493,
-      "grad_norm": 1.562425136566162,
-      "learning_rate": 6.579497487437186e-05,
-      "loss": 4.7336,
-      "step": 34546
-    },
-    {
-      "epoch": 18.016688396349412,
-      "grad_norm": 1.492027997970581,
-      "learning_rate": 6.579396984924624e-05,
-      "loss": 4.866,
-      "step": 34547
-    },
-    {
-      "epoch": 18.01720990873533,
-      "grad_norm": 1.5027825832366943,
-      "learning_rate": 6.57929648241206e-05,
-      "loss": 5.172,
-      "step": 34548
-    },
-    {
-      "epoch": 18.01773142112125,
-      "grad_norm": 1.3953830003738403,
-      "learning_rate": 6.579195979899497e-05,
-      "loss": 5.7471,
-      "step": 34549
-    },
-    {
-      "epoch": 18.01825293350717,
-      "grad_norm": 1.4958491325378418,
-      "learning_rate": 6.579095477386935e-05,
-      "loss": 5.4687,
-      "step": 34550
-    },
-    {
-      "epoch": 18.01877444589309,
-      "grad_norm": 1.5545681715011597,
-      "learning_rate": 6.578994974874372e-05,
-      "loss": 5.4791,
-      "step": 34551
-    },
-    {
-      "epoch": 18.01929595827901,
-      "grad_norm": 1.6165564060211182,
-      "learning_rate": 6.57889447236181e-05,
-      "loss": 5.125,
-      "step": 34552
-    },
-    {
-      "epoch": 18.01981747066493,
-      "grad_norm": 1.4830657243728638,
-      "learning_rate": 6.578793969849247e-05,
-      "loss": 5.327,
-      "step": 34553
-    },
-    {
-      "epoch": 18.02033898305085,
-      "grad_norm": 1.5338672399520874,
-      "learning_rate": 6.578693467336684e-05,
-      "loss": 5.3565,
-      "step": 34554
-    },
-    {
-      "epoch": 18.020860495436768,
-      "grad_norm": 1.3909610509872437,
-      "learning_rate": 6.578592964824121e-05,
-      "loss": 4.6126,
-      "step": 34555
-    },
-    {
-      "epoch": 18.021382007822687,
-      "grad_norm": 1.539527416229248,
-      "learning_rate": 6.578492462311559e-05,
-      "loss": 5.5139,
-      "step": 34556
-    },
-    {
-      "epoch": 18.021903520208603,
-      "grad_norm": 1.5561306476593018,
-      "learning_rate": 6.578391959798995e-05,
-      "loss": 5.0995,
-      "step": 34557
-    },
-    {
-      "epoch": 18.022425032594523,
-      "grad_norm": 1.4354084730148315,
-      "learning_rate": 6.578291457286433e-05,
-      "loss": 5.7345,
-      "step": 34558
-    },
-    {
-      "epoch": 18.022946544980442,
-      "grad_norm": 1.4375252723693848,
-      "learning_rate": 6.578190954773869e-05,
-      "loss": 5.3047,
-      "step": 34559
-    },
-    {
-      "epoch": 18.02346805736636,
-      "grad_norm": 1.4918491840362549,
-      "learning_rate": 6.578090452261307e-05,
-      "loss": 5.2178,
-      "step": 34560
-    },
-    {
-      "epoch": 18.02398956975228,
-      "grad_norm": 1.5885992050170898,
-      "learning_rate": 6.577989949748743e-05,
-      "loss": 5.0412,
-      "step": 34561
-    },
-    {
-      "epoch": 18.0245110821382,
-      "grad_norm": 1.483041763305664,
-      "learning_rate": 6.577889447236181e-05,
-      "loss": 5.2471,
-      "step": 34562
-    },
-    {
-      "epoch": 18.02503259452412,
-      "grad_norm": 1.4388571977615356,
-      "learning_rate": 6.577788944723619e-05,
-      "loss": 5.0853,
-      "step": 34563
-    },
-    {
-      "epoch": 18.02555410691004,
-      "grad_norm": 1.5534882545471191,
-      "learning_rate": 6.577688442211055e-05,
-      "loss": 5.0776,
-      "step": 34564
-    },
-    {
-      "epoch": 18.02607561929596,
-      "grad_norm": 1.5898467302322388,
-      "learning_rate": 6.577587939698493e-05,
-      "loss": 4.8765,
-      "step": 34565
-    },
-    {
-      "epoch": 18.02659713168188,
-      "grad_norm": 1.4725756645202637,
-      "learning_rate": 6.57748743718593e-05,
-      "loss": 5.3694,
-      "step": 34566
-    },
-    {
-      "epoch": 18.027118644067798,
-      "grad_norm": 1.3981398344039917,
-      "learning_rate": 6.577386934673367e-05,
-      "loss": 5.5904,
-      "step": 34567
-    },
-    {
-      "epoch": 18.027640156453717,
-      "grad_norm": 1.4878056049346924,
-      "learning_rate": 6.577286432160804e-05,
-      "loss": 5.2232,
-      "step": 34568
-    },
-    {
-      "epoch": 18.028161668839633,
-      "grad_norm": 1.5437144041061401,
-      "learning_rate": 6.577185929648242e-05,
-      "loss": 5.2577,
-      "step": 34569
-    },
-    {
-      "epoch": 18.028683181225553,
-      "grad_norm": 1.4957984685897827,
-      "learning_rate": 6.577085427135678e-05,
-      "loss": 5.4858,
-      "step": 34570
-    },
-    {
-      "epoch": 18.029204693611472,
-      "grad_norm": 1.512265920639038,
-      "learning_rate": 6.576984924623116e-05,
-      "loss": 5.3819,
-      "step": 34571
-    },
-    {
-      "epoch": 18.02972620599739,
-      "grad_norm": 1.5195114612579346,
-      "learning_rate": 6.576884422110554e-05,
-      "loss": 5.248,
-      "step": 34572
-    },
-    {
-      "epoch": 18.03024771838331,
-      "grad_norm": 1.518963098526001,
-      "learning_rate": 6.576783919597991e-05,
-      "loss": 5.6539,
-      "step": 34573
-    },
-    {
-      "epoch": 18.03076923076923,
-      "grad_norm": 1.4354969263076782,
-      "learning_rate": 6.576683417085428e-05,
-      "loss": 5.7035,
-      "step": 34574
-    },
-    {
-      "epoch": 18.03129074315515,
-      "grad_norm": 1.6240463256835938,
-      "learning_rate": 6.576582914572864e-05,
-      "loss": 4.9394,
-      "step": 34575
-    },
-    {
-      "epoch": 18.03181225554107,
-      "grad_norm": 1.5011909008026123,
-      "learning_rate": 6.576482412060302e-05,
-      "loss": 5.1787,
-      "step": 34576
-    },
-    {
-      "epoch": 18.03233376792699,
-      "grad_norm": 1.503113865852356,
-      "learning_rate": 6.576381909547739e-05,
-      "loss": 5.5052,
-      "step": 34577
-    },
-    {
-      "epoch": 18.03285528031291,
-      "grad_norm": 1.4374912977218628,
-      "learning_rate": 6.576281407035176e-05,
-      "loss": 5.1866,
-      "step": 34578
-    },
-    {
-      "epoch": 18.033376792698828,
-      "grad_norm": 1.781506061553955,
-      "learning_rate": 6.576180904522613e-05,
-      "loss": 5.4219,
-      "step": 34579
-    },
-    {
-      "epoch": 18.033898305084747,
-      "grad_norm": 1.5514981746673584,
-      "learning_rate": 6.57608040201005e-05,
-      "loss": 5.4388,
-      "step": 34580
-    },
-    {
-      "epoch": 18.034419817470663,
-      "grad_norm": 1.5693433284759521,
-      "learning_rate": 6.575979899497487e-05,
-      "loss": 5.1358,
-      "step": 34581
-    },
-    {
-      "epoch": 18.034941329856583,
-      "grad_norm": 1.472252607345581,
-      "learning_rate": 6.575879396984925e-05,
-      "loss": 5.3515,
-      "step": 34582
-    },
-    {
-      "epoch": 18.035462842242502,
-      "grad_norm": 1.4153048992156982,
-      "learning_rate": 6.575778894472363e-05,
-      "loss": 5.4055,
-      "step": 34583
-    },
-    {
-      "epoch": 18.03598435462842,
-      "grad_norm": 1.5130982398986816,
-      "learning_rate": 6.5756783919598e-05,
-      "loss": 5.4355,
-      "step": 34584
-    },
-    {
-      "epoch": 18.03650586701434,
-      "grad_norm": 1.599351167678833,
-      "learning_rate": 6.575577889447237e-05,
-      "loss": 5.3867,
-      "step": 34585
-    },
-    {
-      "epoch": 18.03702737940026,
-      "grad_norm": 1.6010029315948486,
-      "learning_rate": 6.575477386934674e-05,
-      "loss": 5.5749,
-      "step": 34586
-    },
-    {
-      "epoch": 18.03754889178618,
-      "grad_norm": 1.6208842992782593,
-      "learning_rate": 6.575376884422111e-05,
-      "loss": 4.8557,
-      "step": 34587
-    },
-    {
-      "epoch": 18.0380704041721,
-      "grad_norm": 1.6073371171951294,
-      "learning_rate": 6.575276381909547e-05,
-      "loss": 5.1565,
-      "step": 34588
-    },
-    {
-      "epoch": 18.03859191655802,
-      "grad_norm": 1.5532559156417847,
-      "learning_rate": 6.575175879396985e-05,
-      "loss": 5.3807,
-      "step": 34589
-    },
-    {
-      "epoch": 18.03911342894394,
-      "grad_norm": 1.5103278160095215,
-      "learning_rate": 6.575075376884422e-05,
-      "loss": 5.2288,
-      "step": 34590
-    },
-    {
-      "epoch": 18.039634941329858,
-      "grad_norm": 1.5695115327835083,
-      "learning_rate": 6.57497487437186e-05,
-      "loss": 5.665,
-      "step": 34591
-    },
-    {
-      "epoch": 18.040156453715777,
-      "grad_norm": 1.639782428741455,
-      "learning_rate": 6.574874371859296e-05,
-      "loss": 5.1793,
-      "step": 34592
-    },
-    {
-      "epoch": 18.040677966101693,
-      "grad_norm": 1.5369418859481812,
-      "learning_rate": 6.574773869346734e-05,
-      "loss": 5.306,
-      "step": 34593
-    },
-    {
-      "epoch": 18.041199478487613,
-      "grad_norm": 1.6074708700180054,
-      "learning_rate": 6.574673366834171e-05,
-      "loss": 5.3109,
-      "step": 34594
-    },
-    {
-      "epoch": 18.041720990873532,
-      "grad_norm": 1.5674928426742554,
-      "learning_rate": 6.574572864321609e-05,
-      "loss": 5.149,
-      "step": 34595
-    },
-    {
-      "epoch": 18.042242503259452,
-      "grad_norm": 1.508589744567871,
-      "learning_rate": 6.574472361809046e-05,
-      "loss": 5.0721,
-      "step": 34596
-    },
-    {
-      "epoch": 18.04276401564537,
-      "grad_norm": 1.5352027416229248,
-      "learning_rate": 6.574371859296483e-05,
-      "loss": 5.5259,
-      "step": 34597
-    },
-    {
-      "epoch": 18.04328552803129,
-      "grad_norm": 1.5572913885116577,
-      "learning_rate": 6.57427135678392e-05,
-      "loss": 5.1563,
-      "step": 34598
-    },
-    {
-      "epoch": 18.04380704041721,
-      "grad_norm": 1.4075987339019775,
-      "learning_rate": 6.574170854271358e-05,
-      "loss": 5.4353,
-      "step": 34599
-    },
-    {
-      "epoch": 18.04432855280313,
-      "grad_norm": 1.464223027229309,
-      "learning_rate": 6.574070351758794e-05,
-      "loss": 5.4848,
-      "step": 34600
-    },
-    {
-      "epoch": 18.04485006518905,
-      "grad_norm": 1.3523657321929932,
-      "learning_rate": 6.57396984924623e-05,
-      "loss": 5.8222,
-      "step": 34601
-    },
-    {
-      "epoch": 18.04537157757497,
-      "grad_norm": 1.4958513975143433,
-      "learning_rate": 6.573869346733668e-05,
-      "loss": 5.2151,
-      "step": 34602
-    },
-    {
-      "epoch": 18.045893089960888,
-      "grad_norm": 1.4361563920974731,
-      "learning_rate": 6.573768844221106e-05,
-      "loss": 5.3382,
-      "step": 34603
-    },
-    {
-      "epoch": 18.046414602346807,
-      "grad_norm": 1.503666639328003,
-      "learning_rate": 6.573668341708544e-05,
-      "loss": 5.2727,
-      "step": 34604
-    },
-    {
-      "epoch": 18.046936114732723,
-      "grad_norm": 1.5056921243667603,
-      "learning_rate": 6.57356783919598e-05,
-      "loss": 5.6894,
-      "step": 34605
-    },
-    {
-      "epoch": 18.047457627118643,
-      "grad_norm": 1.5336576700210571,
-      "learning_rate": 6.573467336683418e-05,
-      "loss": 5.4023,
-      "step": 34606
-    },
-    {
-      "epoch": 18.047979139504562,
-      "grad_norm": 1.4512757062911987,
-      "learning_rate": 6.573366834170854e-05,
-      "loss": 5.4506,
-      "step": 34607
-    },
-    {
-      "epoch": 18.048500651890482,
-      "grad_norm": 1.478692650794983,
-      "learning_rate": 6.573266331658292e-05,
-      "loss": 5.2107,
-      "step": 34608
-    },
-    {
-      "epoch": 18.0490221642764,
-      "grad_norm": 1.4543477296829224,
-      "learning_rate": 6.573165829145729e-05,
-      "loss": 5.499,
-      "step": 34609
-    },
-    {
-      "epoch": 18.04954367666232,
-      "grad_norm": 1.4903199672698975,
-      "learning_rate": 6.573065326633166e-05,
-      "loss": 5.3322,
-      "step": 34610
-    },
-    {
-      "epoch": 18.05006518904824,
-      "grad_norm": 1.5416743755340576,
-      "learning_rate": 6.572964824120603e-05,
-      "loss": 5.0808,
-      "step": 34611
-    },
-    {
-      "epoch": 18.05058670143416,
-      "grad_norm": 1.4331419467926025,
-      "learning_rate": 6.57286432160804e-05,
-      "loss": 5.5223,
-      "step": 34612
-    },
-    {
-      "epoch": 18.05110821382008,
-      "grad_norm": 1.5897860527038574,
-      "learning_rate": 6.572763819095477e-05,
-      "loss": 5.3195,
-      "step": 34613
-    },
-    {
-      "epoch": 18.051629726206,
-      "grad_norm": 1.4615106582641602,
-      "learning_rate": 6.572663316582915e-05,
-      "loss": 5.2033,
-      "step": 34614
-    },
-    {
-      "epoch": 18.052151238591918,
-      "grad_norm": 1.5043411254882812,
-      "learning_rate": 6.572562814070353e-05,
-      "loss": 5.2385,
-      "step": 34615
-    },
-    {
-      "epoch": 18.052672750977834,
-      "grad_norm": 1.5116440057754517,
-      "learning_rate": 6.572462311557789e-05,
-      "loss": 5.4361,
-      "step": 34616
-    },
-    {
-      "epoch": 18.053194263363753,
-      "grad_norm": 1.4523956775665283,
-      "learning_rate": 6.572361809045227e-05,
-      "loss": 5.5032,
-      "step": 34617
-    },
-    {
-      "epoch": 18.053715775749673,
-      "grad_norm": 1.448911428451538,
-      "learning_rate": 6.572261306532663e-05,
-      "loss": 5.2659,
-      "step": 34618
-    },
-    {
-      "epoch": 18.054237288135592,
-      "grad_norm": 1.5118372440338135,
-      "learning_rate": 6.572160804020101e-05,
-      "loss": 5.626,
-      "step": 34619
-    },
-    {
-      "epoch": 18.054758800521512,
-      "grad_norm": 1.5289254188537598,
-      "learning_rate": 6.572060301507537e-05,
-      "loss": 4.9954,
-      "step": 34620
-    },
-    {
-      "epoch": 18.05528031290743,
-      "grad_norm": 1.469541072845459,
-      "learning_rate": 6.571959798994975e-05,
-      "loss": 5.5143,
-      "step": 34621
-    },
-    {
-      "epoch": 18.05580182529335,
-      "grad_norm": 1.5532082319259644,
-      "learning_rate": 6.571859296482412e-05,
-      "loss": 5.4735,
-      "step": 34622
-    },
-    {
-      "epoch": 18.05632333767927,
-      "grad_norm": 1.4450894594192505,
-      "learning_rate": 6.57175879396985e-05,
-      "loss": 5.2722,
-      "step": 34623
-    },
-    {
-      "epoch": 18.05684485006519,
-      "grad_norm": 1.5134862661361694,
-      "learning_rate": 6.571658291457287e-05,
-      "loss": 5.2481,
-      "step": 34624
-    },
-    {
-      "epoch": 18.05736636245111,
-      "grad_norm": 1.5844768285751343,
-      "learning_rate": 6.571557788944725e-05,
-      "loss": 5.6222,
-      "step": 34625
-    },
-    {
-      "epoch": 18.05788787483703,
-      "grad_norm": 1.6586672067642212,
-      "learning_rate": 6.571457286432161e-05,
-      "loss": 5.3307,
-      "step": 34626
-    },
-    {
-      "epoch": 18.058409387222948,
-      "grad_norm": 1.6478641033172607,
-      "learning_rate": 6.571356783919599e-05,
-      "loss": 5.512,
-      "step": 34627
-    },
-    {
-      "epoch": 18.058930899608864,
-      "grad_norm": 1.580997109413147,
-      "learning_rate": 6.571256281407036e-05,
-      "loss": 4.8734,
-      "step": 34628
-    },
-    {
-      "epoch": 18.059452411994783,
-      "grad_norm": 1.4740139245986938,
-      "learning_rate": 6.571155778894472e-05,
-      "loss": 5.4667,
-      "step": 34629
-    },
-    {
-      "epoch": 18.059973924380703,
-      "grad_norm": 1.4440287351608276,
-      "learning_rate": 6.57105527638191e-05,
-      "loss": 5.8202,
-      "step": 34630
-    },
-    {
-      "epoch": 18.060495436766622,
-      "grad_norm": 1.3620930910110474,
-      "learning_rate": 6.570954773869346e-05,
-      "loss": 5.7675,
-      "step": 34631
-    },
-    {
-      "epoch": 18.061016949152542,
-      "grad_norm": 1.5663306713104248,
-      "learning_rate": 6.570854271356784e-05,
-      "loss": 5.1806,
-      "step": 34632
-    },
-    {
-      "epoch": 18.06153846153846,
-      "grad_norm": 1.587952733039856,
-      "learning_rate": 6.57075376884422e-05,
-      "loss": 5.0576,
-      "step": 34633
-    },
-    {
-      "epoch": 18.06205997392438,
-      "grad_norm": 1.5675674676895142,
-      "learning_rate": 6.570653266331658e-05,
-      "loss": 5.3668,
-      "step": 34634
-    },
-    {
-      "epoch": 18.0625814863103,
-      "grad_norm": 1.4197580814361572,
-      "learning_rate": 6.570552763819096e-05,
-      "loss": 5.7254,
-      "step": 34635
-    },
-    {
-      "epoch": 18.06310299869622,
-      "grad_norm": 1.4922350645065308,
-      "learning_rate": 6.570452261306534e-05,
-      "loss": 5.0202,
-      "step": 34636
-    },
-    {
-      "epoch": 18.06362451108214,
-      "grad_norm": 1.5960174798965454,
-      "learning_rate": 6.57035175879397e-05,
-      "loss": 5.2306,
-      "step": 34637
-    },
-    {
-      "epoch": 18.06414602346806,
-      "grad_norm": 1.4535239934921265,
-      "learning_rate": 6.570251256281408e-05,
-      "loss": 5.5079,
-      "step": 34638
-    },
-    {
-      "epoch": 18.064667535853978,
-      "grad_norm": 1.4340791702270508,
-      "learning_rate": 6.570150753768844e-05,
-      "loss": 5.3525,
-      "step": 34639
-    },
-    {
-      "epoch": 18.065189048239894,
-      "grad_norm": 1.5600826740264893,
-      "learning_rate": 6.570050251256282e-05,
-      "loss": 5.1839,
-      "step": 34640
-    },
-    {
-      "epoch": 18.065710560625813,
-      "grad_norm": 1.4844518899917603,
-      "learning_rate": 6.569949748743719e-05,
-      "loss": 5.583,
-      "step": 34641
-    },
-    {
-      "epoch": 18.066232073011733,
-      "grad_norm": 1.5528546571731567,
-      "learning_rate": 6.569849246231155e-05,
-      "loss": 5.384,
-      "step": 34642
-    },
-    {
-      "epoch": 18.066753585397652,
-      "grad_norm": 1.468652606010437,
-      "learning_rate": 6.569748743718593e-05,
-      "loss": 5.2699,
-      "step": 34643
-    },
-    {
-      "epoch": 18.067275097783572,
-      "grad_norm": 1.5190958976745605,
-      "learning_rate": 6.56964824120603e-05,
-      "loss": 5.4236,
-      "step": 34644
-    },
-    {
-      "epoch": 18.06779661016949,
-      "grad_norm": 1.5703768730163574,
-      "learning_rate": 6.569547738693468e-05,
-      "loss": 4.9771,
-      "step": 34645
-    },
-    {
-      "epoch": 18.06831812255541,
-      "grad_norm": 1.6199052333831787,
-      "learning_rate": 6.569447236180905e-05,
-      "loss": 5.3266,
-      "step": 34646
-    },
-    {
-      "epoch": 18.06883963494133,
-      "grad_norm": 1.5301989316940308,
-      "learning_rate": 6.569346733668343e-05,
-      "loss": 5.0812,
-      "step": 34647
-    },
-    {
-      "epoch": 18.06936114732725,
-      "grad_norm": 1.481387972831726,
-      "learning_rate": 6.569246231155779e-05,
-      "loss": 5.1372,
-      "step": 34648
-    },
-    {
-      "epoch": 18.06988265971317,
-      "grad_norm": 1.6264225244522095,
-      "learning_rate": 6.569145728643217e-05,
-      "loss": 5.154,
-      "step": 34649
-    },
-    {
-      "epoch": 18.07040417209909,
-      "grad_norm": 1.5865304470062256,
-      "learning_rate": 6.569045226130653e-05,
-      "loss": 5.4165,
-      "step": 34650
-    },
-    {
-      "epoch": 18.070925684485008,
-      "grad_norm": 1.4768562316894531,
-      "learning_rate": 6.568944723618091e-05,
-      "loss": 5.2928,
-      "step": 34651
-    },
-    {
-      "epoch": 18.071447196870924,
-      "grad_norm": 1.6357741355895996,
-      "learning_rate": 6.568844221105528e-05,
-      "loss": 5.0816,
-      "step": 34652
-    },
-    {
-      "epoch": 18.071968709256844,
-      "grad_norm": 1.5055735111236572,
-      "learning_rate": 6.568743718592965e-05,
-      "loss": 5.0562,
-      "step": 34653
-    },
-    {
-      "epoch": 18.072490221642763,
-      "grad_norm": 1.5756511688232422,
-      "learning_rate": 6.568643216080402e-05,
-      "loss": 4.7764,
-      "step": 34654
-    },
-    {
-      "epoch": 18.073011734028682,
-      "grad_norm": 1.5165220499038696,
-      "learning_rate": 6.56854271356784e-05,
-      "loss": 5.2413,
-      "step": 34655
-    },
-    {
-      "epoch": 18.073533246414602,
-      "grad_norm": 1.6151864528656006,
-      "learning_rate": 6.568442211055277e-05,
-      "loss": 5.5243,
-      "step": 34656
-    },
-    {
-      "epoch": 18.07405475880052,
-      "grad_norm": 1.5366954803466797,
-      "learning_rate": 6.568341708542714e-05,
-      "loss": 4.672,
-      "step": 34657
-    },
-    {
-      "epoch": 18.07457627118644,
-      "grad_norm": 1.454894781112671,
-      "learning_rate": 6.568241206030151e-05,
-      "loss": 5.7864,
-      "step": 34658
-    },
-    {
-      "epoch": 18.07509778357236,
-      "grad_norm": 1.467299461364746,
-      "learning_rate": 6.568140703517588e-05,
-      "loss": 5.2823,
-      "step": 34659
-    },
-    {
-      "epoch": 18.07561929595828,
-      "grad_norm": 1.7381941080093384,
-      "learning_rate": 6.568040201005026e-05,
-      "loss": 4.52,
-      "step": 34660
-    },
-    {
-      "epoch": 18.0761408083442,
-      "grad_norm": 1.442779302597046,
-      "learning_rate": 6.567939698492462e-05,
-      "loss": 5.5065,
-      "step": 34661
-    },
-    {
-      "epoch": 18.07666232073012,
-      "grad_norm": 1.5210611820220947,
-      "learning_rate": 6.5678391959799e-05,
-      "loss": 4.8792,
-      "step": 34662
-    },
-    {
-      "epoch": 18.077183833116038,
-      "grad_norm": 1.5103472471237183,
-      "learning_rate": 6.567738693467336e-05,
-      "loss": 5.3209,
-      "step": 34663
-    },
-    {
-      "epoch": 18.077705345501954,
-      "grad_norm": 1.501133680343628,
-      "learning_rate": 6.567638190954774e-05,
-      "loss": 5.354,
-      "step": 34664
-    },
-    {
-      "epoch": 18.078226857887874,
-      "grad_norm": 1.5217347145080566,
-      "learning_rate": 6.567537688442212e-05,
-      "loss": 5.5244,
-      "step": 34665
-    },
-    {
-      "epoch": 18.078748370273793,
-      "grad_norm": 1.4500813484191895,
-      "learning_rate": 6.56743718592965e-05,
-      "loss": 5.2248,
-      "step": 34666
-    },
-    {
-      "epoch": 18.079269882659712,
-      "grad_norm": 1.5539237260818481,
-      "learning_rate": 6.567336683417086e-05,
-      "loss": 5.467,
-      "step": 34667
-    },
-    {
-      "epoch": 18.079791395045632,
-      "grad_norm": 1.5291252136230469,
-      "learning_rate": 6.567236180904523e-05,
-      "loss": 5.4098,
-      "step": 34668
-    },
-    {
-      "epoch": 18.08031290743155,
-      "grad_norm": 1.3955698013305664,
-      "learning_rate": 6.56713567839196e-05,
-      "loss": 5.5886,
-      "step": 34669
-    },
-    {
-      "epoch": 18.08083441981747,
-      "grad_norm": 1.548580288887024,
-      "learning_rate": 6.567035175879397e-05,
-      "loss": 5.1342,
-      "step": 34670
-    },
-    {
-      "epoch": 18.08135593220339,
-      "grad_norm": 1.4891387224197388,
-      "learning_rate": 6.566934673366835e-05,
-      "loss": 5.2413,
-      "step": 34671
-    },
-    {
-      "epoch": 18.08187744458931,
-      "grad_norm": 1.4824316501617432,
-      "learning_rate": 6.566834170854271e-05,
-      "loss": 5.5576,
-      "step": 34672
-    },
-    {
-      "epoch": 18.08239895697523,
-      "grad_norm": 1.5341542959213257,
-      "learning_rate": 6.566733668341709e-05,
-      "loss": 5.4321,
-      "step": 34673
-    },
-    {
-      "epoch": 18.08292046936115,
-      "grad_norm": 1.3576890230178833,
-      "learning_rate": 6.566633165829145e-05,
-      "loss": 5.3471,
-      "step": 34674
-    },
-    {
-      "epoch": 18.083441981747068,
-      "grad_norm": 1.4840565919876099,
-      "learning_rate": 6.566532663316583e-05,
-      "loss": 5.1134,
-      "step": 34675
-    },
-    {
-      "epoch": 18.083963494132984,
-      "grad_norm": 1.4401267766952515,
-      "learning_rate": 6.566432160804021e-05,
-      "loss": 5.6173,
-      "step": 34676
-    },
-    {
-      "epoch": 18.084485006518904,
-      "grad_norm": 1.4396402835845947,
-      "learning_rate": 6.566331658291459e-05,
-      "loss": 5.5454,
-      "step": 34677
-    },
-    {
-      "epoch": 18.085006518904823,
-      "grad_norm": 1.4900788068771362,
-      "learning_rate": 6.566231155778895e-05,
-      "loss": 5.0067,
-      "step": 34678
-    },
-    {
-      "epoch": 18.085528031290742,
-      "grad_norm": 1.5447666645050049,
-      "learning_rate": 6.566130653266333e-05,
-      "loss": 5.3049,
-      "step": 34679
-    },
-    {
-      "epoch": 18.086049543676662,
-      "grad_norm": 1.4292688369750977,
-      "learning_rate": 6.566030150753769e-05,
-      "loss": 5.4874,
-      "step": 34680
-    },
-    {
-      "epoch": 18.08657105606258,
-      "grad_norm": 1.5779412984848022,
-      "learning_rate": 6.565929648241206e-05,
-      "loss": 5.3022,
-      "step": 34681
-    },
-    {
-      "epoch": 18.0870925684485,
-      "grad_norm": 1.5314241647720337,
-      "learning_rate": 6.565829145728643e-05,
-      "loss": 5.1228,
-      "step": 34682
-    },
-    {
-      "epoch": 18.08761408083442,
-      "grad_norm": 1.550503134727478,
-      "learning_rate": 6.56572864321608e-05,
-      "loss": 5.6373,
-      "step": 34683
-    },
-    {
-      "epoch": 18.08813559322034,
-      "grad_norm": 1.4603941440582275,
-      "learning_rate": 6.565628140703518e-05,
-      "loss": 5.6687,
-      "step": 34684
-    },
-    {
-      "epoch": 18.08865710560626,
-      "grad_norm": 1.536291480064392,
-      "learning_rate": 6.565527638190955e-05,
-      "loss": 5.3633,
-      "step": 34685
-    },
-    {
-      "epoch": 18.08917861799218,
-      "grad_norm": 1.487256646156311,
-      "learning_rate": 6.565427135678393e-05,
-      "loss": 5.1725,
-      "step": 34686
-    },
-    {
-      "epoch": 18.089700130378098,
-      "grad_norm": 1.5196473598480225,
-      "learning_rate": 6.56532663316583e-05,
-      "loss": 5.0312,
-      "step": 34687
-    },
-    {
-      "epoch": 18.090221642764014,
-      "grad_norm": 1.6211988925933838,
-      "learning_rate": 6.565226130653267e-05,
-      "loss": 5.1287,
-      "step": 34688
-    },
-    {
-      "epoch": 18.090743155149934,
-      "grad_norm": 1.490683913230896,
-      "learning_rate": 6.565125628140704e-05,
-      "loss": 5.4484,
-      "step": 34689
-    },
-    {
-      "epoch": 18.091264667535853,
-      "grad_norm": 1.5342042446136475,
-      "learning_rate": 6.565025125628142e-05,
-      "loss": 5.0193,
-      "step": 34690
-    },
-    {
-      "epoch": 18.091786179921773,
-      "grad_norm": 1.5123234987258911,
-      "learning_rate": 6.564924623115578e-05,
-      "loss": 4.9961,
-      "step": 34691
-    },
-    {
-      "epoch": 18.092307692307692,
-      "grad_norm": 1.5340791940689087,
-      "learning_rate": 6.564824120603016e-05,
-      "loss": 5.0627,
-      "step": 34692
-    },
-    {
-      "epoch": 18.09282920469361,
-      "grad_norm": 1.5588375329971313,
-      "learning_rate": 6.564723618090452e-05,
-      "loss": 5.0206,
-      "step": 34693
-    },
-    {
-      "epoch": 18.09335071707953,
-      "grad_norm": 1.6013777256011963,
-      "learning_rate": 6.564623115577889e-05,
-      "loss": 5.3824,
-      "step": 34694
-    },
-    {
-      "epoch": 18.09387222946545,
-      "grad_norm": 1.5170643329620361,
-      "learning_rate": 6.564522613065326e-05,
-      "loss": 5.3069,
-      "step": 34695
-    },
-    {
-      "epoch": 18.09439374185137,
-      "grad_norm": 1.5521310567855835,
-      "learning_rate": 6.564422110552764e-05,
-      "loss": 5.4812,
-      "step": 34696
-    },
-    {
-      "epoch": 18.09491525423729,
-      "grad_norm": 1.5121878385543823,
-      "learning_rate": 6.564321608040202e-05,
-      "loss": 4.9242,
-      "step": 34697
-    },
-    {
-      "epoch": 18.09543676662321,
-      "grad_norm": 1.5706411600112915,
-      "learning_rate": 6.564221105527638e-05,
-      "loss": 4.4191,
-      "step": 34698
-    },
-    {
-      "epoch": 18.09595827900913,
-      "grad_norm": 1.531265139579773,
-      "learning_rate": 6.564120603015076e-05,
-      "loss": 5.3713,
-      "step": 34699
-    },
-    {
-      "epoch": 18.096479791395044,
-      "grad_norm": 1.4260010719299316,
-      "learning_rate": 6.564020100502513e-05,
-      "loss": 5.6904,
-      "step": 34700
-    },
-    {
-      "epoch": 18.097001303780964,
-      "grad_norm": 1.5078775882720947,
-      "learning_rate": 6.56391959798995e-05,
-      "loss": 5.3762,
-      "step": 34701
-    },
-    {
-      "epoch": 18.097522816166883,
-      "grad_norm": 1.4709781408309937,
-      "learning_rate": 6.563819095477387e-05,
-      "loss": 5.0558,
-      "step": 34702
-    },
-    {
-      "epoch": 18.098044328552803,
-      "grad_norm": 1.5705885887145996,
-      "learning_rate": 6.563718592964825e-05,
-      "loss": 5.0662,
-      "step": 34703
-    },
-    {
-      "epoch": 18.098565840938722,
-      "grad_norm": 1.5149872303009033,
-      "learning_rate": 6.563618090452261e-05,
-      "loss": 4.705,
-      "step": 34704
-    },
-    {
-      "epoch": 18.09908735332464,
-      "grad_norm": 1.4991281032562256,
-      "learning_rate": 6.563517587939699e-05,
-      "loss": 5.6538,
-      "step": 34705
-    },
-    {
-      "epoch": 18.09960886571056,
-      "grad_norm": 1.5288684368133545,
-      "learning_rate": 6.563417085427137e-05,
-      "loss": 5.0997,
-      "step": 34706
-    },
-    {
-      "epoch": 18.10013037809648,
-      "grad_norm": 1.5611311197280884,
-      "learning_rate": 6.563316582914573e-05,
-      "loss": 5.0954,
-      "step": 34707
-    },
-    {
-      "epoch": 18.1006518904824,
-      "grad_norm": 1.5838377475738525,
-      "learning_rate": 6.563216080402011e-05,
-      "loss": 5.4684,
-      "step": 34708
-    },
-    {
-      "epoch": 18.10117340286832,
-      "grad_norm": 1.5602283477783203,
-      "learning_rate": 6.563115577889447e-05,
-      "loss": 5.1135,
-      "step": 34709
-    },
-    {
-      "epoch": 18.10169491525424,
-      "grad_norm": 1.5354801416397095,
-      "learning_rate": 6.563015075376885e-05,
-      "loss": 5.0882,
-      "step": 34710
-    },
-    {
-      "epoch": 18.102216427640155,
-      "grad_norm": 1.4216192960739136,
-      "learning_rate": 6.562914572864321e-05,
-      "loss": 5.4656,
-      "step": 34711
-    },
-    {
-      "epoch": 18.102737940026074,
-      "grad_norm": 1.3924976587295532,
-      "learning_rate": 6.562814070351759e-05,
-      "loss": 5.6316,
-      "step": 34712
-    },
-    {
-      "epoch": 18.103259452411994,
-      "grad_norm": 1.3719847202301025,
-      "learning_rate": 6.562713567839196e-05,
-      "loss": 4.708,
-      "step": 34713
-    },
-    {
-      "epoch": 18.103780964797913,
-      "grad_norm": 1.4247363805770874,
-      "learning_rate": 6.562613065326633e-05,
-      "loss": 5.4642,
-      "step": 34714
-    },
-    {
-      "epoch": 18.104302477183833,
-      "grad_norm": 1.434099555015564,
-      "learning_rate": 6.56251256281407e-05,
-      "loss": 5.6319,
-      "step": 34715
-    },
-    {
-      "epoch": 18.104823989569752,
-      "grad_norm": 1.486373782157898,
-      "learning_rate": 6.562412060301508e-05,
-      "loss": 5.4936,
-      "step": 34716
-    },
-    {
-      "epoch": 18.10534550195567,
-      "grad_norm": 1.4984190464019775,
-      "learning_rate": 6.562311557788945e-05,
-      "loss": 5.252,
-      "step": 34717
-    },
-    {
-      "epoch": 18.10586701434159,
-      "grad_norm": 1.5200908184051514,
-      "learning_rate": 6.562211055276383e-05,
-      "loss": 5.3486,
-      "step": 34718
-    },
-    {
-      "epoch": 18.10638852672751,
-      "grad_norm": 1.5217214822769165,
-      "learning_rate": 6.56211055276382e-05,
-      "loss": 5.2665,
-      "step": 34719
-    },
-    {
-      "epoch": 18.10691003911343,
-      "grad_norm": 1.4659438133239746,
-      "learning_rate": 6.562010050251257e-05,
-      "loss": 5.2444,
-      "step": 34720
-    },
-    {
-      "epoch": 18.10743155149935,
-      "grad_norm": 1.489905595779419,
-      "learning_rate": 6.561909547738694e-05,
-      "loss": 5.5706,
-      "step": 34721
-    },
-    {
-      "epoch": 18.10795306388527,
-      "grad_norm": 1.5095866918563843,
-      "learning_rate": 6.56180904522613e-05,
-      "loss": 5.4657,
-      "step": 34722
-    },
-    {
-      "epoch": 18.108474576271185,
-      "grad_norm": 1.4741783142089844,
-      "learning_rate": 6.561708542713568e-05,
-      "loss": 5.578,
-      "step": 34723
-    },
-    {
-      "epoch": 18.108996088657104,
-      "grad_norm": 1.3754218816757202,
-      "learning_rate": 6.561608040201004e-05,
-      "loss": 5.6938,
-      "step": 34724
-    },
-    {
-      "epoch": 18.109517601043024,
-      "grad_norm": 1.5434041023254395,
-      "learning_rate": 6.561507537688442e-05,
-      "loss": 5.1074,
-      "step": 34725
-    },
-    {
-      "epoch": 18.110039113428943,
-      "grad_norm": 1.4458158016204834,
-      "learning_rate": 6.56140703517588e-05,
-      "loss": 5.6353,
-      "step": 34726
-    },
-    {
-      "epoch": 18.110560625814863,
-      "grad_norm": 1.5700881481170654,
-      "learning_rate": 6.561306532663318e-05,
-      "loss": 5.1557,
-      "step": 34727
-    },
-    {
-      "epoch": 18.111082138200782,
-      "grad_norm": 1.551288366317749,
-      "learning_rate": 6.561206030150754e-05,
-      "loss": 5.3828,
-      "step": 34728
-    },
-    {
-      "epoch": 18.1116036505867,
-      "grad_norm": 1.4929635524749756,
-      "learning_rate": 6.561105527638192e-05,
-      "loss": 4.8684,
-      "step": 34729
-    },
-    {
-      "epoch": 18.11212516297262,
-      "grad_norm": 1.4571599960327148,
-      "learning_rate": 6.561005025125628e-05,
-      "loss": 5.5977,
-      "step": 34730
-    },
-    {
-      "epoch": 18.11264667535854,
-      "grad_norm": 1.4532625675201416,
-      "learning_rate": 6.560904522613066e-05,
-      "loss": 5.0411,
-      "step": 34731
-    },
-    {
-      "epoch": 18.11316818774446,
-      "grad_norm": 1.4782838821411133,
-      "learning_rate": 6.560804020100503e-05,
-      "loss": 5.5598,
-      "step": 34732
-    },
-    {
-      "epoch": 18.11368970013038,
-      "grad_norm": 1.5391632318496704,
-      "learning_rate": 6.56070351758794e-05,
-      "loss": 5.2991,
-      "step": 34733
-    },
-    {
-      "epoch": 18.1142112125163,
-      "grad_norm": 1.535821795463562,
-      "learning_rate": 6.560603015075377e-05,
-      "loss": 4.8132,
-      "step": 34734
-    },
-    {
-      "epoch": 18.114732724902215,
-      "grad_norm": 1.415401577949524,
-      "learning_rate": 6.560502512562813e-05,
-      "loss": 5.4479,
-      "step": 34735
-    },
-    {
-      "epoch": 18.115254237288134,
-      "grad_norm": 1.4523394107818604,
-      "learning_rate": 6.560402010050251e-05,
-      "loss": 5.835,
-      "step": 34736
-    },
-    {
-      "epoch": 18.115775749674054,
-      "grad_norm": 1.603868007659912,
-      "learning_rate": 6.560301507537689e-05,
-      "loss": 5.401,
-      "step": 34737
-    },
-    {
-      "epoch": 18.116297262059973,
-      "grad_norm": 1.5369523763656616,
-      "learning_rate": 6.560201005025127e-05,
-      "loss": 5.2468,
-      "step": 34738
-    },
-    {
-      "epoch": 18.116818774445893,
-      "grad_norm": 1.5394657850265503,
-      "learning_rate": 6.560100502512563e-05,
-      "loss": 5.2049,
-      "step": 34739
-    },
-    {
-      "epoch": 18.117340286831812,
-      "grad_norm": 1.5186843872070312,
-      "learning_rate": 6.560000000000001e-05,
-      "loss": 5.5596,
-      "step": 34740
-    },
-    {
-      "epoch": 18.11786179921773,
-      "grad_norm": 1.5636845827102661,
-      "learning_rate": 6.559899497487437e-05,
-      "loss": 5.2315,
-      "step": 34741
-    },
-    {
-      "epoch": 18.11838331160365,
-      "grad_norm": 1.5405263900756836,
-      "learning_rate": 6.559798994974875e-05,
-      "loss": 5.6729,
-      "step": 34742
-    },
-    {
-      "epoch": 18.11890482398957,
-      "grad_norm": 1.5722442865371704,
-      "learning_rate": 6.559698492462312e-05,
-      "loss": 5.1296,
-      "step": 34743
-    },
-    {
-      "epoch": 18.11942633637549,
-      "grad_norm": 1.557265281677246,
-      "learning_rate": 6.559597989949749e-05,
-      "loss": 4.9102,
-      "step": 34744
-    },
-    {
-      "epoch": 18.11994784876141,
-      "grad_norm": 1.5416128635406494,
-      "learning_rate": 6.559497487437186e-05,
-      "loss": 5.5092,
-      "step": 34745
-    },
-    {
-      "epoch": 18.12046936114733,
-      "grad_norm": 1.5286929607391357,
-      "learning_rate": 6.559396984924624e-05,
-      "loss": 5.6569,
-      "step": 34746
-    },
-    {
-      "epoch": 18.120990873533245,
-      "grad_norm": 1.4312111139297485,
-      "learning_rate": 6.559296482412061e-05,
-      "loss": 5.4523,
-      "step": 34747
-    },
-    {
-      "epoch": 18.121512385919164,
-      "grad_norm": 1.503358244895935,
-      "learning_rate": 6.559195979899498e-05,
-      "loss": 5.488,
-      "step": 34748
-    },
-    {
-      "epoch": 18.122033898305084,
-      "grad_norm": 1.6116706132888794,
-      "learning_rate": 6.559095477386936e-05,
-      "loss": 4.6766,
-      "step": 34749
-    },
-    {
-      "epoch": 18.122555410691003,
-      "grad_norm": 1.5582810640335083,
-      "learning_rate": 6.558994974874372e-05,
-      "loss": 5.288,
-      "step": 34750
-    },
-    {
-      "epoch": 18.123076923076923,
-      "grad_norm": 1.4830929040908813,
-      "learning_rate": 6.55889447236181e-05,
-      "loss": 4.614,
-      "step": 34751
-    },
-    {
-      "epoch": 18.123598435462842,
-      "grad_norm": 1.5543071031570435,
-      "learning_rate": 6.558793969849246e-05,
-      "loss": 5.1815,
-      "step": 34752
-    },
-    {
-      "epoch": 18.12411994784876,
-      "grad_norm": 1.4314194917678833,
-      "learning_rate": 6.558693467336684e-05,
-      "loss": 5.5246,
-      "step": 34753
-    },
-    {
-      "epoch": 18.12464146023468,
-      "grad_norm": 1.4149539470672607,
-      "learning_rate": 6.55859296482412e-05,
-      "loss": 4.9751,
-      "step": 34754
-    },
-    {
-      "epoch": 18.1251629726206,
-      "grad_norm": 1.4905495643615723,
-      "learning_rate": 6.558492462311558e-05,
-      "loss": 5.3322,
-      "step": 34755
-    },
-    {
-      "epoch": 18.12568448500652,
-      "grad_norm": 1.5954179763793945,
-      "learning_rate": 6.558391959798995e-05,
-      "loss": 4.996,
-      "step": 34756
-    },
-    {
-      "epoch": 18.12620599739244,
-      "grad_norm": 1.4741817712783813,
-      "learning_rate": 6.558291457286432e-05,
-      "loss": 5.4455,
-      "step": 34757
-    },
-    {
-      "epoch": 18.12672750977836,
-      "grad_norm": 1.4491560459136963,
-      "learning_rate": 6.55819095477387e-05,
-      "loss": 5.4061,
-      "step": 34758
-    },
-    {
-      "epoch": 18.127249022164275,
-      "grad_norm": 1.458097219467163,
-      "learning_rate": 6.558090452261308e-05,
-      "loss": 5.5017,
-      "step": 34759
-    },
-    {
-      "epoch": 18.127770534550194,
-      "grad_norm": 1.5456874370574951,
-      "learning_rate": 6.557989949748744e-05,
-      "loss": 5.1826,
-      "step": 34760
-    },
-    {
-      "epoch": 18.128292046936114,
-      "grad_norm": 1.4613863229751587,
-      "learning_rate": 6.557889447236181e-05,
-      "loss": 5.2529,
-      "step": 34761
-    },
-    {
-      "epoch": 18.128813559322033,
-      "grad_norm": 1.490883469581604,
-      "learning_rate": 6.557788944723619e-05,
-      "loss": 5.4128,
-      "step": 34762
-    },
-    {
-      "epoch": 18.129335071707953,
-      "grad_norm": 1.5510258674621582,
-      "learning_rate": 6.557688442211055e-05,
-      "loss": 5.0327,
-      "step": 34763
-    },
-    {
-      "epoch": 18.129856584093872,
-      "grad_norm": 1.6998423337936401,
-      "learning_rate": 6.557587939698493e-05,
-      "loss": 5.353,
-      "step": 34764
-    },
-    {
-      "epoch": 18.13037809647979,
-      "grad_norm": 1.536425232887268,
-      "learning_rate": 6.557487437185929e-05,
-      "loss": 5.1388,
-      "step": 34765
-    },
-    {
-      "epoch": 18.13089960886571,
-      "grad_norm": 1.5780502557754517,
-      "learning_rate": 6.557386934673367e-05,
-      "loss": 5.1475,
-      "step": 34766
-    },
-    {
-      "epoch": 18.13142112125163,
-      "grad_norm": 1.4200111627578735,
-      "learning_rate": 6.557286432160803e-05,
-      "loss": 5.4249,
-      "step": 34767
-    },
-    {
-      "epoch": 18.13194263363755,
-      "grad_norm": 1.576996922492981,
-      "learning_rate": 6.557185929648241e-05,
-      "loss": 5.1582,
-      "step": 34768
-    },
-    {
-      "epoch": 18.13246414602347,
-      "grad_norm": 1.4303525686264038,
-      "learning_rate": 6.557085427135679e-05,
-      "loss": 5.3291,
-      "step": 34769
-    },
-    {
-      "epoch": 18.13298565840939,
-      "grad_norm": 1.5240280628204346,
-      "learning_rate": 6.556984924623117e-05,
-      "loss": 5.145,
-      "step": 34770
-    },
-    {
-      "epoch": 18.133507170795305,
-      "grad_norm": 1.5700992345809937,
-      "learning_rate": 6.556884422110553e-05,
-      "loss": 5.0276,
-      "step": 34771
-    },
-    {
-      "epoch": 18.134028683181224,
-      "grad_norm": 1.4257880449295044,
-      "learning_rate": 6.556783919597991e-05,
-      "loss": 5.6319,
-      "step": 34772
-    },
-    {
-      "epoch": 18.134550195567144,
-      "grad_norm": 1.4043596982955933,
-      "learning_rate": 6.556683417085427e-05,
-      "loss": 5.516,
-      "step": 34773
-    },
-    {
-      "epoch": 18.135071707953063,
-      "grad_norm": 1.4611157178878784,
-      "learning_rate": 6.556582914572864e-05,
-      "loss": 4.9041,
-      "step": 34774
-    },
-    {
-      "epoch": 18.135593220338983,
-      "grad_norm": 1.4195982217788696,
-      "learning_rate": 6.556482412060302e-05,
-      "loss": 5.2208,
-      "step": 34775
-    },
-    {
-      "epoch": 18.136114732724902,
-      "grad_norm": 1.669226050376892,
-      "learning_rate": 6.556381909547738e-05,
-      "loss": 4.9217,
-      "step": 34776
-    },
-    {
-      "epoch": 18.13663624511082,
-      "grad_norm": 1.4859085083007812,
-      "learning_rate": 6.556281407035176e-05,
-      "loss": 5.5533,
-      "step": 34777
-    },
-    {
-      "epoch": 18.13715775749674,
-      "grad_norm": 1.444925308227539,
-      "learning_rate": 6.556180904522614e-05,
-      "loss": 5.3828,
-      "step": 34778
-    },
-    {
-      "epoch": 18.13767926988266,
-      "grad_norm": 1.4358723163604736,
-      "learning_rate": 6.556080402010051e-05,
-      "loss": 4.9216,
-      "step": 34779
-    },
-    {
-      "epoch": 18.13820078226858,
-      "grad_norm": 1.5191117525100708,
-      "learning_rate": 6.555979899497488e-05,
-      "loss": 5.0092,
-      "step": 34780
-    },
-    {
-      "epoch": 18.1387222946545,
-      "grad_norm": 1.5146043300628662,
-      "learning_rate": 6.555879396984926e-05,
-      "loss": 5.2127,
-      "step": 34781
-    },
-    {
-      "epoch": 18.13924380704042,
-      "grad_norm": 1.4268568754196167,
-      "learning_rate": 6.555778894472362e-05,
-      "loss": 5.266,
-      "step": 34782
-    },
-    {
-      "epoch": 18.139765319426335,
-      "grad_norm": 1.5672972202301025,
-      "learning_rate": 6.5556783919598e-05,
-      "loss": 4.8385,
-      "step": 34783
-    },
-    {
-      "epoch": 18.140286831812254,
-      "grad_norm": 1.5006475448608398,
-      "learning_rate": 6.555577889447236e-05,
-      "loss": 4.6466,
-      "step": 34784
-    },
-    {
-      "epoch": 18.140808344198174,
-      "grad_norm": 1.7708674669265747,
-      "learning_rate": 6.555477386934674e-05,
-      "loss": 4.6203,
-      "step": 34785
-    },
-    {
-      "epoch": 18.141329856584093,
-      "grad_norm": 1.4927825927734375,
-      "learning_rate": 6.55537688442211e-05,
-      "loss": 5.5905,
-      "step": 34786
-    },
-    {
-      "epoch": 18.141851368970013,
-      "grad_norm": 1.5249882936477661,
-      "learning_rate": 6.555276381909547e-05,
-      "loss": 5.6026,
-      "step": 34787
-    },
-    {
-      "epoch": 18.142372881355932,
-      "grad_norm": 1.5089328289031982,
-      "learning_rate": 6.555175879396985e-05,
-      "loss": 4.7361,
-      "step": 34788
-    },
-    {
-      "epoch": 18.14289439374185,
-      "grad_norm": 1.4815776348114014,
-      "learning_rate": 6.555075376884422e-05,
-      "loss": 5.2353,
-      "step": 34789
-    },
-    {
-      "epoch": 18.14341590612777,
-      "grad_norm": 1.5689575672149658,
-      "learning_rate": 6.55497487437186e-05,
-      "loss": 5.1548,
-      "step": 34790
-    },
-    {
-      "epoch": 18.14393741851369,
-      "grad_norm": 1.4064397811889648,
-      "learning_rate": 6.554874371859297e-05,
-      "loss": 5.477,
-      "step": 34791
-    },
-    {
-      "epoch": 18.14445893089961,
-      "grad_norm": 1.438963770866394,
-      "learning_rate": 6.554773869346734e-05,
-      "loss": 5.5769,
-      "step": 34792
-    },
-    {
-      "epoch": 18.14498044328553,
-      "grad_norm": 1.4705028533935547,
-      "learning_rate": 6.554673366834171e-05,
-      "loss": 5.2047,
-      "step": 34793
-    },
-    {
-      "epoch": 18.14550195567145,
-      "grad_norm": 1.6379690170288086,
-      "learning_rate": 6.554572864321609e-05,
-      "loss": 5.208,
-      "step": 34794
-    },
-    {
-      "epoch": 18.146023468057365,
-      "grad_norm": 1.703682780265808,
-      "learning_rate": 6.554472361809045e-05,
-      "loss": 5.1877,
-      "step": 34795
-    },
-    {
-      "epoch": 18.146544980443284,
-      "grad_norm": 1.564548373222351,
-      "learning_rate": 6.554371859296483e-05,
-      "loss": 5.3858,
-      "step": 34796
-    },
-    {
-      "epoch": 18.147066492829204,
-      "grad_norm": 1.5450574159622192,
-      "learning_rate": 6.554271356783919e-05,
-      "loss": 5.2099,
-      "step": 34797
-    },
-    {
-      "epoch": 18.147588005215123,
-      "grad_norm": 1.4630632400512695,
-      "learning_rate": 6.554170854271357e-05,
-      "loss": 4.8974,
-      "step": 34798
-    },
-    {
-      "epoch": 18.148109517601043,
-      "grad_norm": 1.522316336631775,
-      "learning_rate": 6.554070351758795e-05,
-      "loss": 5.138,
-      "step": 34799
-    },
-    {
-      "epoch": 18.148631029986962,
-      "grad_norm": 1.5295764207839966,
-      "learning_rate": 6.553969849246231e-05,
-      "loss": 4.9352,
-      "step": 34800
-    },
-    {
-      "epoch": 18.14915254237288,
-      "grad_norm": 1.441697120666504,
-      "learning_rate": 6.553869346733669e-05,
-      "loss": 5.7123,
-      "step": 34801
-    },
-    {
-      "epoch": 18.1496740547588,
-      "grad_norm": 1.4220669269561768,
-      "learning_rate": 6.553768844221105e-05,
-      "loss": 5.5786,
-      "step": 34802
-    },
-    {
-      "epoch": 18.15019556714472,
-      "grad_norm": 1.414972186088562,
-      "learning_rate": 6.553668341708543e-05,
-      "loss": 5.8154,
-      "step": 34803
-    },
-    {
-      "epoch": 18.15071707953064,
-      "grad_norm": 1.4886680841445923,
-      "learning_rate": 6.55356783919598e-05,
-      "loss": 5.6275,
-      "step": 34804
-    },
-    {
-      "epoch": 18.15123859191656,
-      "grad_norm": 1.4749085903167725,
-      "learning_rate": 6.553467336683417e-05,
-      "loss": 5.3677,
-      "step": 34805
-    },
-    {
-      "epoch": 18.151760104302475,
-      "grad_norm": 1.5885953903198242,
-      "learning_rate": 6.553366834170854e-05,
-      "loss": 5.2762,
-      "step": 34806
-    },
-    {
-      "epoch": 18.152281616688395,
-      "grad_norm": 1.691893458366394,
-      "learning_rate": 6.553266331658292e-05,
-      "loss": 4.6892,
-      "step": 34807
-    },
-    {
-      "epoch": 18.152803129074314,
-      "grad_norm": 1.3592748641967773,
-      "learning_rate": 6.553165829145728e-05,
-      "loss": 5.5479,
-      "step": 34808
-    },
-    {
-      "epoch": 18.153324641460234,
-      "grad_norm": 1.5086095333099365,
-      "learning_rate": 6.553065326633166e-05,
-      "loss": 5.5778,
-      "step": 34809
-    },
-    {
-      "epoch": 18.153846153846153,
-      "grad_norm": 1.6411833763122559,
-      "learning_rate": 6.552964824120604e-05,
-      "loss": 5.2297,
-      "step": 34810
-    },
-    {
-      "epoch": 18.154367666232073,
-      "grad_norm": 1.4941132068634033,
-      "learning_rate": 6.552864321608041e-05,
-      "loss": 5.5801,
-      "step": 34811
-    },
-    {
-      "epoch": 18.154889178617992,
-      "grad_norm": 1.6270151138305664,
-      "learning_rate": 6.552763819095478e-05,
-      "loss": 5.5104,
-      "step": 34812
-    },
-    {
-      "epoch": 18.15541069100391,
-      "grad_norm": 1.6110855340957642,
-      "learning_rate": 6.552663316582916e-05,
-      "loss": 5.2098,
-      "step": 34813
-    },
-    {
-      "epoch": 18.15593220338983,
-      "grad_norm": 1.5446337461471558,
-      "learning_rate": 6.552562814070352e-05,
-      "loss": 4.9073,
-      "step": 34814
-    },
-    {
-      "epoch": 18.15645371577575,
-      "grad_norm": 1.5922949314117432,
-      "learning_rate": 6.552462311557789e-05,
-      "loss": 4.6378,
-      "step": 34815
-    },
-    {
-      "epoch": 18.15697522816167,
-      "grad_norm": 1.4414341449737549,
-      "learning_rate": 6.552361809045226e-05,
-      "loss": 5.5202,
-      "step": 34816
-    },
-    {
-      "epoch": 18.15749674054759,
-      "grad_norm": 1.5039165019989014,
-      "learning_rate": 6.552261306532663e-05,
-      "loss": 5.0949,
-      "step": 34817
-    },
-    {
-      "epoch": 18.158018252933505,
-      "grad_norm": 1.5626249313354492,
-      "learning_rate": 6.5521608040201e-05,
-      "loss": 5.063,
-      "step": 34818
-    },
-    {
-      "epoch": 18.158539765319425,
-      "grad_norm": 1.4279576539993286,
-      "learning_rate": 6.552060301507538e-05,
-      "loss": 5.4147,
-      "step": 34819
-    },
-    {
-      "epoch": 18.159061277705344,
-      "grad_norm": 1.4760702848434448,
-      "learning_rate": 6.551959798994976e-05,
-      "loss": 5.3884,
-      "step": 34820
-    },
-    {
-      "epoch": 18.159582790091264,
-      "grad_norm": 1.4193261861801147,
-      "learning_rate": 6.551859296482413e-05,
-      "loss": 5.4898,
-      "step": 34821
-    },
-    {
-      "epoch": 18.160104302477183,
-      "grad_norm": 1.4374274015426636,
-      "learning_rate": 6.55175879396985e-05,
-      "loss": 5.3901,
-      "step": 34822
-    },
-    {
-      "epoch": 18.160625814863103,
-      "grad_norm": 1.558549404144287,
-      "learning_rate": 6.551658291457287e-05,
-      "loss": 4.9953,
-      "step": 34823
-    },
-    {
-      "epoch": 18.161147327249022,
-      "grad_norm": 1.5603439807891846,
-      "learning_rate": 6.551557788944724e-05,
-      "loss": 5.2498,
-      "step": 34824
-    },
-    {
-      "epoch": 18.16166883963494,
-      "grad_norm": 1.5648305416107178,
-      "learning_rate": 6.551457286432161e-05,
-      "loss": 5.5491,
-      "step": 34825
-    },
-    {
-      "epoch": 18.16219035202086,
-      "grad_norm": 1.36715829372406,
-      "learning_rate": 6.551356783919599e-05,
-      "loss": 5.3994,
-      "step": 34826
-    },
-    {
-      "epoch": 18.16271186440678,
-      "grad_norm": 1.484755516052246,
-      "learning_rate": 6.551256281407035e-05,
-      "loss": 5.2725,
-      "step": 34827
-    },
-    {
-      "epoch": 18.1632333767927,
-      "grad_norm": 1.4975780248641968,
-      "learning_rate": 6.551155778894472e-05,
-      "loss": 5.0361,
-      "step": 34828
-    },
-    {
-      "epoch": 18.16375488917862,
-      "grad_norm": 1.9857290983200073,
-      "learning_rate": 6.55105527638191e-05,
-      "loss": 5.5029,
-      "step": 34829
-    },
-    {
-      "epoch": 18.164276401564535,
-      "grad_norm": 1.5992182493209839,
-      "learning_rate": 6.550954773869347e-05,
-      "loss": 5.1238,
-      "step": 34830
-    },
-    {
-      "epoch": 18.164797913950455,
-      "grad_norm": 1.4990993738174438,
-      "learning_rate": 6.550854271356785e-05,
-      "loss": 5.3654,
-      "step": 34831
-    },
-    {
-      "epoch": 18.165319426336374,
-      "grad_norm": 1.5324193239212036,
-      "learning_rate": 6.550753768844221e-05,
-      "loss": 5.1376,
-      "step": 34832
-    },
-    {
-      "epoch": 18.165840938722294,
-      "grad_norm": 1.504879355430603,
-      "learning_rate": 6.550653266331659e-05,
-      "loss": 5.4207,
-      "step": 34833
-    },
-    {
-      "epoch": 18.166362451108213,
-      "grad_norm": 1.577104926109314,
-      "learning_rate": 6.550552763819096e-05,
-      "loss": 5.0177,
-      "step": 34834
-    },
-    {
-      "epoch": 18.166883963494133,
-      "grad_norm": 1.4338055849075317,
-      "learning_rate": 6.550452261306533e-05,
-      "loss": 5.6527,
-      "step": 34835
-    },
-    {
-      "epoch": 18.167405475880052,
-      "grad_norm": 1.4941486120224,
-      "learning_rate": 6.55035175879397e-05,
-      "loss": 5.2445,
-      "step": 34836
-    },
-    {
-      "epoch": 18.16792698826597,
-      "grad_norm": 1.595936894416809,
-      "learning_rate": 6.550251256281408e-05,
-      "loss": 5.4488,
-      "step": 34837
-    },
-    {
-      "epoch": 18.16844850065189,
-      "grad_norm": 1.5036816596984863,
-      "learning_rate": 6.550150753768844e-05,
-      "loss": 5.6253,
-      "step": 34838
-    },
-    {
-      "epoch": 18.16897001303781,
-      "grad_norm": 1.4664523601531982,
-      "learning_rate": 6.550050251256282e-05,
-      "loss": 5.0494,
-      "step": 34839
-    },
-    {
-      "epoch": 18.16949152542373,
-      "grad_norm": 1.5147820711135864,
-      "learning_rate": 6.54994974874372e-05,
-      "loss": 5.2882,
-      "step": 34840
-    },
-    {
-      "epoch": 18.17001303780965,
-      "grad_norm": 1.4081207513809204,
-      "learning_rate": 6.549849246231156e-05,
-      "loss": 5.5182,
-      "step": 34841
-    },
-    {
-      "epoch": 18.170534550195566,
-      "grad_norm": 1.5154842138290405,
-      "learning_rate": 6.549748743718594e-05,
-      "loss": 5.4794,
-      "step": 34842
-    },
-    {
-      "epoch": 18.171056062581485,
-      "grad_norm": 1.4003199338912964,
-      "learning_rate": 6.54964824120603e-05,
-      "loss": 5.3934,
-      "step": 34843
-    },
-    {
-      "epoch": 18.171577574967404,
-      "grad_norm": 1.4939452409744263,
-      "learning_rate": 6.549547738693468e-05,
-      "loss": 5.3668,
-      "step": 34844
-    },
-    {
-      "epoch": 18.172099087353324,
-      "grad_norm": 1.5782525539398193,
-      "learning_rate": 6.549447236180904e-05,
-      "loss": 5.5585,
-      "step": 34845
-    },
-    {
-      "epoch": 18.172620599739243,
-      "grad_norm": 1.6183388233184814,
-      "learning_rate": 6.549346733668342e-05,
-      "loss": 5.0846,
-      "step": 34846
-    },
-    {
-      "epoch": 18.173142112125163,
-      "grad_norm": 1.4439342021942139,
-      "learning_rate": 6.549246231155779e-05,
-      "loss": 5.4851,
-      "step": 34847
-    },
-    {
-      "epoch": 18.173663624511082,
-      "grad_norm": 1.5396722555160522,
-      "learning_rate": 6.549145728643216e-05,
-      "loss": 5.0394,
-      "step": 34848
-    },
-    {
-      "epoch": 18.174185136897,
-      "grad_norm": 1.549690842628479,
-      "learning_rate": 6.549045226130653e-05,
-      "loss": 5.1813,
-      "step": 34849
-    },
-    {
-      "epoch": 18.17470664928292,
-      "grad_norm": 1.4745891094207764,
-      "learning_rate": 6.54894472361809e-05,
-      "loss": 5.2724,
-      "step": 34850
-    },
-    {
-      "epoch": 18.17522816166884,
-      "grad_norm": 1.4597036838531494,
-      "learning_rate": 6.548844221105528e-05,
-      "loss": 5.6108,
-      "step": 34851
-    },
-    {
-      "epoch": 18.17574967405476,
-      "grad_norm": 1.4187307357788086,
-      "learning_rate": 6.548743718592966e-05,
-      "loss": 5.6991,
-      "step": 34852
-    },
-    {
-      "epoch": 18.17627118644068,
-      "grad_norm": 1.4298580884933472,
-      "learning_rate": 6.548643216080403e-05,
-      "loss": 5.3496,
-      "step": 34853
-    },
-    {
-      "epoch": 18.176792698826596,
-      "grad_norm": 1.548215389251709,
-      "learning_rate": 6.548542713567839e-05,
-      "loss": 4.8909,
-      "step": 34854
-    },
-    {
-      "epoch": 18.177314211212515,
-      "grad_norm": 1.4845858812332153,
-      "learning_rate": 6.548442211055277e-05,
-      "loss": 5.1576,
-      "step": 34855
-    },
-    {
-      "epoch": 18.177835723598434,
-      "grad_norm": 1.492318868637085,
-      "learning_rate": 6.548341708542713e-05,
-      "loss": 4.8227,
-      "step": 34856
-    },
-    {
-      "epoch": 18.178357235984354,
-      "grad_norm": 1.4679827690124512,
-      "learning_rate": 6.548241206030151e-05,
-      "loss": 5.2984,
-      "step": 34857
-    },
-    {
-      "epoch": 18.178878748370273,
-      "grad_norm": 1.3732436895370483,
-      "learning_rate": 6.548140703517587e-05,
-      "loss": 5.5394,
-      "step": 34858
-    },
-    {
-      "epoch": 18.179400260756193,
-      "grad_norm": 1.5245345830917358,
-      "learning_rate": 6.548040201005025e-05,
-      "loss": 5.2395,
-      "step": 34859
-    },
-    {
-      "epoch": 18.179921773142112,
-      "grad_norm": 1.454908013343811,
-      "learning_rate": 6.547939698492463e-05,
-      "loss": 5.6756,
-      "step": 34860
-    },
-    {
-      "epoch": 18.180443285528032,
-      "grad_norm": 1.5772024393081665,
-      "learning_rate": 6.547839195979901e-05,
-      "loss": 5.3701,
-      "step": 34861
-    },
-    {
-      "epoch": 18.18096479791395,
-      "grad_norm": 1.4381818771362305,
-      "learning_rate": 6.547738693467337e-05,
-      "loss": 5.4803,
-      "step": 34862
-    },
-    {
-      "epoch": 18.18148631029987,
-      "grad_norm": 1.529289722442627,
-      "learning_rate": 6.547638190954775e-05,
-      "loss": 5.4907,
-      "step": 34863
-    },
-    {
-      "epoch": 18.18200782268579,
-      "grad_norm": 1.5543384552001953,
-      "learning_rate": 6.547537688442211e-05,
-      "loss": 5.3666,
-      "step": 34864
-    },
-    {
-      "epoch": 18.18252933507171,
-      "grad_norm": 1.6258039474487305,
-      "learning_rate": 6.547437185929649e-05,
-      "loss": 5.0658,
-      "step": 34865
-    },
-    {
-      "epoch": 18.183050847457626,
-      "grad_norm": 1.5146974325180054,
-      "learning_rate": 6.547336683417086e-05,
-      "loss": 4.8877,
-      "step": 34866
-    },
-    {
-      "epoch": 18.183572359843545,
-      "grad_norm": 1.5242600440979004,
-      "learning_rate": 6.547236180904522e-05,
-      "loss": 5.8429,
-      "step": 34867
-    },
-    {
-      "epoch": 18.184093872229464,
-      "grad_norm": 1.8700261116027832,
-      "learning_rate": 6.54713567839196e-05,
-      "loss": 4.7958,
-      "step": 34868
-    },
-    {
-      "epoch": 18.184615384615384,
-      "grad_norm": 1.44826340675354,
-      "learning_rate": 6.547035175879396e-05,
-      "loss": 5.4566,
-      "step": 34869
-    },
-    {
-      "epoch": 18.185136897001303,
-      "grad_norm": 1.536860704421997,
-      "learning_rate": 6.546934673366834e-05,
-      "loss": 5.2347,
-      "step": 34870
-    },
-    {
-      "epoch": 18.185658409387223,
-      "grad_norm": 1.4871758222579956,
-      "learning_rate": 6.546834170854272e-05,
-      "loss": 4.9105,
-      "step": 34871
-    },
-    {
-      "epoch": 18.186179921773142,
-      "grad_norm": 1.555174469947815,
-      "learning_rate": 6.54673366834171e-05,
-      "loss": 5.0048,
-      "step": 34872
-    },
-    {
-      "epoch": 18.186701434159062,
-      "grad_norm": 1.4288769960403442,
-      "learning_rate": 6.546633165829146e-05,
-      "loss": 5.4919,
-      "step": 34873
-    },
-    {
-      "epoch": 18.18722294654498,
-      "grad_norm": 1.5099208354949951,
-      "learning_rate": 6.546532663316584e-05,
-      "loss": 4.9789,
-      "step": 34874
-    },
-    {
-      "epoch": 18.1877444589309,
-      "grad_norm": 1.408856749534607,
-      "learning_rate": 6.54643216080402e-05,
-      "loss": 5.4968,
-      "step": 34875
-    },
-    {
-      "epoch": 18.18826597131682,
-      "grad_norm": 1.4189766645431519,
-      "learning_rate": 6.546331658291458e-05,
-      "loss": 5.9798,
-      "step": 34876
-    },
-    {
-      "epoch": 18.18878748370274,
-      "grad_norm": 1.5772254467010498,
-      "learning_rate": 6.546231155778894e-05,
-      "loss": 5.4231,
-      "step": 34877
-    },
-    {
-      "epoch": 18.189308996088656,
-      "grad_norm": 1.5328242778778076,
-      "learning_rate": 6.546130653266332e-05,
-      "loss": 5.3255,
-      "step": 34878
-    },
-    {
-      "epoch": 18.189830508474575,
-      "grad_norm": 1.3975861072540283,
-      "learning_rate": 6.546030150753769e-05,
-      "loss": 5.0906,
-      "step": 34879
-    },
-    {
-      "epoch": 18.190352020860495,
-      "grad_norm": 1.54291832447052,
-      "learning_rate": 6.545929648241206e-05,
-      "loss": 5.7682,
-      "step": 34880
-    },
-    {
-      "epoch": 18.190873533246414,
-      "grad_norm": 1.4059083461761475,
-      "learning_rate": 6.545829145728644e-05,
-      "loss": 5.6107,
-      "step": 34881
-    },
-    {
-      "epoch": 18.191395045632333,
-      "grad_norm": 1.4131250381469727,
-      "learning_rate": 6.54572864321608e-05,
-      "loss": 6.0269,
-      "step": 34882
-    },
-    {
-      "epoch": 18.191916558018253,
-      "grad_norm": 1.5313061475753784,
-      "learning_rate": 6.545628140703518e-05,
-      "loss": 4.904,
-      "step": 34883
-    },
-    {
-      "epoch": 18.192438070404172,
-      "grad_norm": 1.4025726318359375,
-      "learning_rate": 6.545527638190955e-05,
-      "loss": 5.6513,
-      "step": 34884
-    },
-    {
-      "epoch": 18.192959582790092,
-      "grad_norm": 1.5110735893249512,
-      "learning_rate": 6.545427135678393e-05,
-      "loss": 5.4355,
-      "step": 34885
-    },
-    {
-      "epoch": 18.19348109517601,
-      "grad_norm": 1.4641635417938232,
-      "learning_rate": 6.545326633165829e-05,
-      "loss": 5.3327,
-      "step": 34886
-    },
-    {
-      "epoch": 18.19400260756193,
-      "grad_norm": 1.3486363887786865,
-      "learning_rate": 6.545226130653267e-05,
-      "loss": 4.8032,
-      "step": 34887
-    },
-    {
-      "epoch": 18.19452411994785,
-      "grad_norm": 1.432440161705017,
-      "learning_rate": 6.545125628140703e-05,
-      "loss": 5.4287,
-      "step": 34888
-    },
-    {
-      "epoch": 18.195045632333766,
-      "grad_norm": 1.5041319131851196,
-      "learning_rate": 6.545025125628141e-05,
-      "loss": 5.8263,
-      "step": 34889
-    },
-    {
-      "epoch": 18.195567144719686,
-      "grad_norm": 1.626662015914917,
-      "learning_rate": 6.544924623115578e-05,
-      "loss": 4.9387,
-      "step": 34890
-    },
-    {
-      "epoch": 18.196088657105605,
-      "grad_norm": 1.5161420106887817,
-      "learning_rate": 6.544824120603015e-05,
-      "loss": 5.394,
-      "step": 34891
-    },
-    {
-      "epoch": 18.196610169491525,
-      "grad_norm": 1.6143486499786377,
-      "learning_rate": 6.544723618090453e-05,
-      "loss": 5.4262,
-      "step": 34892
-    },
-    {
-      "epoch": 18.197131681877444,
-      "grad_norm": 1.7086371183395386,
-      "learning_rate": 6.54462311557789e-05,
-      "loss": 4.7685,
-      "step": 34893
-    },
-    {
-      "epoch": 18.197653194263363,
-      "grad_norm": 1.5377107858657837,
-      "learning_rate": 6.544522613065327e-05,
-      "loss": 4.9892,
-      "step": 34894
-    },
-    {
-      "epoch": 18.198174706649283,
-      "grad_norm": 1.4700813293457031,
-      "learning_rate": 6.544422110552764e-05,
-      "loss": 5.7902,
-      "step": 34895
-    },
-    {
-      "epoch": 18.198696219035202,
-      "grad_norm": 1.5194870233535767,
-      "learning_rate": 6.544321608040201e-05,
-      "loss": 5.3173,
-      "step": 34896
-    },
-    {
-      "epoch": 18.199217731421122,
-      "grad_norm": 1.4761676788330078,
-      "learning_rate": 6.544221105527638e-05,
-      "loss": 5.1231,
-      "step": 34897
-    },
-    {
-      "epoch": 18.19973924380704,
-      "grad_norm": 1.5172455310821533,
-      "learning_rate": 6.544120603015076e-05,
-      "loss": 5.2889,
-      "step": 34898
-    },
-    {
-      "epoch": 18.20026075619296,
-      "grad_norm": 1.5345903635025024,
-      "learning_rate": 6.544020100502512e-05,
-      "loss": 5.1841,
-      "step": 34899
-    },
-    {
-      "epoch": 18.20078226857888,
-      "grad_norm": 1.39176607131958,
-      "learning_rate": 6.54391959798995e-05,
-      "loss": 5.3672,
-      "step": 34900
-    },
-    {
-      "epoch": 18.201303780964796,
-      "grad_norm": 1.470988392829895,
-      "learning_rate": 6.543819095477388e-05,
-      "loss": 5.1991,
-      "step": 34901
-    },
-    {
-      "epoch": 18.201825293350716,
-      "grad_norm": 1.4883195161819458,
-      "learning_rate": 6.543718592964825e-05,
-      "loss": 5.2369,
-      "step": 34902
-    },
-    {
-      "epoch": 18.202346805736635,
-      "grad_norm": 1.4885742664337158,
-      "learning_rate": 6.543618090452262e-05,
-      "loss": 4.8383,
-      "step": 34903
-    },
-    {
-      "epoch": 18.202868318122555,
-      "grad_norm": 1.5049000978469849,
-      "learning_rate": 6.5435175879397e-05,
-      "loss": 4.538,
-      "step": 34904
-    },
-    {
-      "epoch": 18.203389830508474,
-      "grad_norm": 1.4914672374725342,
-      "learning_rate": 6.543417085427136e-05,
-      "loss": 5.2485,
-      "step": 34905
-    },
-    {
-      "epoch": 18.203911342894393,
-      "grad_norm": 1.3827276229858398,
-      "learning_rate": 6.543316582914574e-05,
-      "loss": 5.4979,
-      "step": 34906
-    },
-    {
-      "epoch": 18.204432855280313,
-      "grad_norm": 1.4302587509155273,
-      "learning_rate": 6.54321608040201e-05,
-      "loss": 5.6093,
-      "step": 34907
-    },
-    {
-      "epoch": 18.204954367666232,
-      "grad_norm": 1.3937782049179077,
-      "learning_rate": 6.543115577889447e-05,
-      "loss": 5.6041,
-      "step": 34908
-    },
-    {
-      "epoch": 18.205475880052152,
-      "grad_norm": 1.589322805404663,
-      "learning_rate": 6.543015075376885e-05,
-      "loss": 5.1134,
-      "step": 34909
-    },
-    {
-      "epoch": 18.20599739243807,
-      "grad_norm": 1.413887619972229,
-      "learning_rate": 6.542914572864321e-05,
-      "loss": 4.9726,
-      "step": 34910
-    },
-    {
-      "epoch": 18.20651890482399,
-      "grad_norm": 1.5907660722732544,
-      "learning_rate": 6.542814070351759e-05,
-      "loss": 4.744,
-      "step": 34911
-    },
-    {
-      "epoch": 18.20704041720991,
-      "grad_norm": 1.653412938117981,
-      "learning_rate": 6.542713567839197e-05,
-      "loss": 4.5515,
-      "step": 34912
-    },
-    {
-      "epoch": 18.207561929595826,
-      "grad_norm": 1.493399739265442,
-      "learning_rate": 6.542613065326634e-05,
-      "loss": 5.2282,
-      "step": 34913
-    },
-    {
-      "epoch": 18.208083441981746,
-      "grad_norm": 1.5394561290740967,
-      "learning_rate": 6.542512562814071e-05,
-      "loss": 5.3628,
-      "step": 34914
-    },
-    {
-      "epoch": 18.208604954367665,
-      "grad_norm": 1.549996256828308,
-      "learning_rate": 6.542412060301509e-05,
-      "loss": 4.5695,
-      "step": 34915
-    },
-    {
-      "epoch": 18.209126466753585,
-      "grad_norm": 1.6083146333694458,
-      "learning_rate": 6.542311557788945e-05,
-      "loss": 5.2977,
-      "step": 34916
-    },
-    {
-      "epoch": 18.209647979139504,
-      "grad_norm": 1.5066291093826294,
-      "learning_rate": 6.542211055276383e-05,
-      "loss": 5.3512,
-      "step": 34917
-    },
-    {
-      "epoch": 18.210169491525424,
-      "grad_norm": 1.6417279243469238,
-      "learning_rate": 6.542110552763819e-05,
-      "loss": 4.5723,
-      "step": 34918
-    },
-    {
-      "epoch": 18.210691003911343,
-      "grad_norm": 1.4482747316360474,
-      "learning_rate": 6.542010050251257e-05,
-      "loss": 4.2383,
-      "step": 34919
-    },
-    {
-      "epoch": 18.211212516297262,
-      "grad_norm": 1.6391679048538208,
-      "learning_rate": 6.541909547738693e-05,
-      "loss": 5.3475,
-      "step": 34920
-    },
-    {
-      "epoch": 18.211734028683182,
-      "grad_norm": 1.5966691970825195,
-      "learning_rate": 6.54180904522613e-05,
-      "loss": 4.5324,
-      "step": 34921
-    },
-    {
-      "epoch": 18.2122555410691,
-      "grad_norm": 1.5638182163238525,
-      "learning_rate": 6.541708542713568e-05,
-      "loss": 5.2908,
-      "step": 34922
-    },
-    {
-      "epoch": 18.21277705345502,
-      "grad_norm": 1.4671496152877808,
-      "learning_rate": 6.541608040201005e-05,
-      "loss": 5.349,
-      "step": 34923
-    },
-    {
-      "epoch": 18.21329856584094,
-      "grad_norm": 1.4323121309280396,
-      "learning_rate": 6.541507537688443e-05,
-      "loss": 5.4634,
-      "step": 34924
-    },
-    {
-      "epoch": 18.213820078226856,
-      "grad_norm": 1.4304111003875732,
-      "learning_rate": 6.54140703517588e-05,
-      "loss": 5.3667,
-      "step": 34925
-    },
-    {
-      "epoch": 18.214341590612776,
-      "grad_norm": 1.5397720336914062,
-      "learning_rate": 6.541306532663317e-05,
-      "loss": 5.4003,
-      "step": 34926
-    },
-    {
-      "epoch": 18.214863102998695,
-      "grad_norm": 1.5086771249771118,
-      "learning_rate": 6.541206030150754e-05,
-      "loss": 4.8089,
-      "step": 34927
-    },
-    {
-      "epoch": 18.215384615384615,
-      "grad_norm": 1.9160431623458862,
-      "learning_rate": 6.541105527638192e-05,
-      "loss": 4.9003,
-      "step": 34928
-    },
-    {
-      "epoch": 18.215906127770534,
-      "grad_norm": 1.5085468292236328,
-      "learning_rate": 6.541005025125628e-05,
-      "loss": 5.5162,
-      "step": 34929
-    },
-    {
-      "epoch": 18.216427640156454,
-      "grad_norm": 1.4818345308303833,
-      "learning_rate": 6.540904522613066e-05,
-      "loss": 5.0746,
-      "step": 34930
-    },
-    {
-      "epoch": 18.216949152542373,
-      "grad_norm": 1.4832135438919067,
-      "learning_rate": 6.540804020100502e-05,
-      "loss": 5.2095,
-      "step": 34931
-    },
-    {
-      "epoch": 18.217470664928292,
-      "grad_norm": 1.4490822553634644,
-      "learning_rate": 6.54070351758794e-05,
-      "loss": 5.7031,
-      "step": 34932
-    },
-    {
-      "epoch": 18.217992177314212,
-      "grad_norm": 1.4371428489685059,
-      "learning_rate": 6.540603015075378e-05,
-      "loss": 5.84,
-      "step": 34933
-    },
-    {
-      "epoch": 18.21851368970013,
-      "grad_norm": 1.4846360683441162,
-      "learning_rate": 6.540502512562814e-05,
-      "loss": 5.2933,
-      "step": 34934
-    },
-    {
-      "epoch": 18.21903520208605,
-      "grad_norm": 1.3966889381408691,
-      "learning_rate": 6.540402010050252e-05,
-      "loss": 5.3803,
-      "step": 34935
-    },
-    {
-      "epoch": 18.21955671447197,
-      "grad_norm": 1.4686295986175537,
-      "learning_rate": 6.540301507537688e-05,
-      "loss": 4.9963,
-      "step": 34936
-    },
-    {
-      "epoch": 18.220078226857886,
-      "grad_norm": 1.4865243434906006,
-      "learning_rate": 6.540201005025126e-05,
-      "loss": 5.3597,
-      "step": 34937
-    },
-    {
-      "epoch": 18.220599739243806,
-      "grad_norm": 1.4970570802688599,
-      "learning_rate": 6.540100502512563e-05,
-      "loss": 5.1109,
-      "step": 34938
-    },
-    {
-      "epoch": 18.221121251629725,
-      "grad_norm": 1.5543824434280396,
-      "learning_rate": 6.54e-05,
-      "loss": 5.633,
-      "step": 34939
-    },
-    {
-      "epoch": 18.221642764015645,
-      "grad_norm": 1.444128394126892,
-      "learning_rate": 6.539899497487437e-05,
-      "loss": 5.4458,
-      "step": 34940
-    },
-    {
-      "epoch": 18.222164276401564,
-      "grad_norm": 1.373309850692749,
-      "learning_rate": 6.539798994974875e-05,
-      "loss": 5.5022,
-      "step": 34941
-    },
-    {
-      "epoch": 18.222685788787484,
-      "grad_norm": 1.6551028490066528,
-      "learning_rate": 6.539698492462311e-05,
-      "loss": 4.9663,
-      "step": 34942
-    },
-    {
-      "epoch": 18.223207301173403,
-      "grad_norm": 1.4227426052093506,
-      "learning_rate": 6.539597989949749e-05,
-      "loss": 5.7229,
-      "step": 34943
-    },
-    {
-      "epoch": 18.223728813559323,
-      "grad_norm": 1.4319027662277222,
-      "learning_rate": 6.539497487437187e-05,
-      "loss": 5.4229,
-      "step": 34944
-    },
-    {
-      "epoch": 18.224250325945242,
-      "grad_norm": 1.489188313484192,
-      "learning_rate": 6.539396984924624e-05,
-      "loss": 5.5823,
-      "step": 34945
-    },
-    {
-      "epoch": 18.22477183833116,
-      "grad_norm": 1.490453839302063,
-      "learning_rate": 6.539296482412061e-05,
-      "loss": 5.3553,
-      "step": 34946
-    },
-    {
-      "epoch": 18.22529335071708,
-      "grad_norm": 1.4017893075942993,
-      "learning_rate": 6.539195979899497e-05,
-      "loss": 5.2463,
-      "step": 34947
-    },
-    {
-      "epoch": 18.225814863103,
-      "grad_norm": 1.5393755435943604,
-      "learning_rate": 6.539095477386935e-05,
-      "loss": 5.6232,
-      "step": 34948
-    },
-    {
-      "epoch": 18.226336375488916,
-      "grad_norm": 1.4903268814086914,
-      "learning_rate": 6.538994974874371e-05,
-      "loss": 5.2934,
-      "step": 34949
-    },
-    {
-      "epoch": 18.226857887874836,
-      "grad_norm": 1.4268423318862915,
-      "learning_rate": 6.538894472361809e-05,
-      "loss": 5.7189,
-      "step": 34950
-    },
-    {
-      "epoch": 18.227379400260755,
-      "grad_norm": 1.5113489627838135,
-      "learning_rate": 6.538793969849246e-05,
-      "loss": 5.3047,
-      "step": 34951
-    },
-    {
-      "epoch": 18.227900912646675,
-      "grad_norm": 1.667169213294983,
-      "learning_rate": 6.538693467336683e-05,
-      "loss": 5.182,
-      "step": 34952
-    },
-    {
-      "epoch": 18.228422425032594,
-      "grad_norm": 1.5584176778793335,
-      "learning_rate": 6.538592964824121e-05,
-      "loss": 4.8539,
-      "step": 34953
-    },
-    {
-      "epoch": 18.228943937418514,
-      "grad_norm": 1.6588366031646729,
-      "learning_rate": 6.538492462311559e-05,
-      "loss": 5.1831,
-      "step": 34954
-    },
-    {
-      "epoch": 18.229465449804433,
-      "grad_norm": 1.480884313583374,
-      "learning_rate": 6.538391959798995e-05,
-      "loss": 5.1815,
-      "step": 34955
-    },
-    {
-      "epoch": 18.229986962190353,
-      "grad_norm": 1.6008315086364746,
-      "learning_rate": 6.538291457286433e-05,
-      "loss": 4.819,
-      "step": 34956
-    },
-    {
-      "epoch": 18.230508474576272,
-      "grad_norm": 1.616977572441101,
-      "learning_rate": 6.53819095477387e-05,
-      "loss": 5.3242,
-      "step": 34957
-    },
-    {
-      "epoch": 18.23102998696219,
-      "grad_norm": 1.4958577156066895,
-      "learning_rate": 6.538090452261307e-05,
-      "loss": 4.8731,
-      "step": 34958
-    },
-    {
-      "epoch": 18.23155149934811,
-      "grad_norm": 1.6139802932739258,
-      "learning_rate": 6.537989949748744e-05,
-      "loss": 5.3742,
-      "step": 34959
-    },
-    {
-      "epoch": 18.23207301173403,
-      "grad_norm": 1.5121251344680786,
-      "learning_rate": 6.53788944723618e-05,
-      "loss": 5.1271,
-      "step": 34960
-    },
-    {
-      "epoch": 18.232594524119946,
-      "grad_norm": 1.5334523916244507,
-      "learning_rate": 6.537788944723618e-05,
-      "loss": 4.9628,
-      "step": 34961
-    },
-    {
-      "epoch": 18.233116036505866,
-      "grad_norm": 1.5490156412124634,
-      "learning_rate": 6.537688442211054e-05,
-      "loss": 5.1167,
-      "step": 34962
-    },
-    {
-      "epoch": 18.233637548891785,
-      "grad_norm": 1.4508469104766846,
-      "learning_rate": 6.537587939698492e-05,
-      "loss": 5.6574,
-      "step": 34963
-    },
-    {
-      "epoch": 18.234159061277705,
-      "grad_norm": 1.4784696102142334,
-      "learning_rate": 6.53748743718593e-05,
-      "loss": 5.3553,
-      "step": 34964
-    },
-    {
-      "epoch": 18.234680573663624,
-      "grad_norm": 1.5792509317398071,
-      "learning_rate": 6.537386934673368e-05,
-      "loss": 5.1535,
-      "step": 34965
-    },
-    {
-      "epoch": 18.235202086049544,
-      "grad_norm": 1.5628973245620728,
-      "learning_rate": 6.537286432160804e-05,
-      "loss": 5.3379,
-      "step": 34966
-    },
-    {
-      "epoch": 18.235723598435463,
-      "grad_norm": 1.428718090057373,
-      "learning_rate": 6.537185929648242e-05,
-      "loss": 5.2931,
-      "step": 34967
-    },
-    {
-      "epoch": 18.236245110821383,
-      "grad_norm": 1.4996012449264526,
-      "learning_rate": 6.537085427135678e-05,
-      "loss": 5.3151,
-      "step": 34968
-    },
-    {
-      "epoch": 18.236766623207302,
-      "grad_norm": 1.4652284383773804,
-      "learning_rate": 6.536984924623116e-05,
-      "loss": 5.2653,
-      "step": 34969
-    },
-    {
-      "epoch": 18.23728813559322,
-      "grad_norm": 1.4968318939208984,
-      "learning_rate": 6.536884422110553e-05,
-      "loss": 5.2984,
-      "step": 34970
-    },
-    {
-      "epoch": 18.23780964797914,
-      "grad_norm": 1.414518117904663,
-      "learning_rate": 6.53678391959799e-05,
-      "loss": 5.187,
-      "step": 34971
-    },
-    {
-      "epoch": 18.23833116036506,
-      "grad_norm": 1.383997917175293,
-      "learning_rate": 6.536683417085427e-05,
-      "loss": 5.5866,
-      "step": 34972
-    },
-    {
-      "epoch": 18.238852672750976,
-      "grad_norm": 1.3529374599456787,
-      "learning_rate": 6.536582914572865e-05,
-      "loss": 5.7564,
-      "step": 34973
-    },
-    {
-      "epoch": 18.239374185136896,
-      "grad_norm": 1.5095305442810059,
-      "learning_rate": 6.536482412060302e-05,
-      "loss": 5.5031,
-      "step": 34974
-    },
-    {
-      "epoch": 18.239895697522815,
-      "grad_norm": 1.564564824104309,
-      "learning_rate": 6.536381909547739e-05,
-      "loss": 4.8906,
-      "step": 34975
-    },
-    {
-      "epoch": 18.240417209908735,
-      "grad_norm": 1.481202483177185,
-      "learning_rate": 6.536281407035177e-05,
-      "loss": 5.0803,
-      "step": 34976
-    },
-    {
-      "epoch": 18.240938722294654,
-      "grad_norm": 1.4377580881118774,
-      "learning_rate": 6.536180904522613e-05,
-      "loss": 5.6235,
-      "step": 34977
-    },
-    {
-      "epoch": 18.241460234680574,
-      "grad_norm": 1.5630308389663696,
-      "learning_rate": 6.536080402010051e-05,
-      "loss": 5.1241,
-      "step": 34978
-    },
-    {
-      "epoch": 18.241981747066493,
-      "grad_norm": 1.5421499013900757,
-      "learning_rate": 6.535979899497487e-05,
-      "loss": 5.3461,
-      "step": 34979
-    },
-    {
-      "epoch": 18.242503259452413,
-      "grad_norm": 1.4625457525253296,
-      "learning_rate": 6.535879396984925e-05,
-      "loss": 5.1414,
-      "step": 34980
-    },
-    {
-      "epoch": 18.243024771838332,
-      "grad_norm": 1.4876269102096558,
-      "learning_rate": 6.535778894472362e-05,
-      "loss": 5.542,
-      "step": 34981
-    },
-    {
-      "epoch": 18.24354628422425,
-      "grad_norm": 1.6286735534667969,
-      "learning_rate": 6.535678391959799e-05,
-      "loss": 5.1831,
-      "step": 34982
-    },
-    {
-      "epoch": 18.24406779661017,
-      "grad_norm": 1.5157155990600586,
-      "learning_rate": 6.535577889447236e-05,
-      "loss": 5.4166,
-      "step": 34983
-    },
-    {
-      "epoch": 18.24458930899609,
-      "grad_norm": 1.5572270154953003,
-      "learning_rate": 6.535477386934674e-05,
-      "loss": 4.8215,
-      "step": 34984
-    },
-    {
-      "epoch": 18.245110821382006,
-      "grad_norm": 1.6048136949539185,
-      "learning_rate": 6.535376884422111e-05,
-      "loss": 4.9738,
-      "step": 34985
-    },
-    {
-      "epoch": 18.245632333767926,
-      "grad_norm": 1.459631323814392,
-      "learning_rate": 6.535276381909549e-05,
-      "loss": 5.3908,
-      "step": 34986
-    },
-    {
-      "epoch": 18.246153846153845,
-      "grad_norm": 1.38510000705719,
-      "learning_rate": 6.535175879396986e-05,
-      "loss": 5.4302,
-      "step": 34987
-    },
-    {
-      "epoch": 18.246675358539765,
-      "grad_norm": 1.457414150238037,
-      "learning_rate": 6.535075376884422e-05,
-      "loss": 5.4087,
-      "step": 34988
-    },
-    {
-      "epoch": 18.247196870925684,
-      "grad_norm": 1.522373914718628,
-      "learning_rate": 6.53497487437186e-05,
-      "loss": 5.2812,
-      "step": 34989
-    },
-    {
-      "epoch": 18.247718383311604,
-      "grad_norm": 1.6016508340835571,
-      "learning_rate": 6.534874371859296e-05,
-      "loss": 5.1616,
-      "step": 34990
-    },
-    {
-      "epoch": 18.248239895697523,
-      "grad_norm": 1.5636253356933594,
-      "learning_rate": 6.534773869346734e-05,
-      "loss": 5.4699,
-      "step": 34991
-    },
-    {
-      "epoch": 18.248761408083443,
-      "grad_norm": 1.3814116716384888,
-      "learning_rate": 6.53467336683417e-05,
-      "loss": 5.6258,
-      "step": 34992
-    },
-    {
-      "epoch": 18.249282920469362,
-      "grad_norm": 1.4744046926498413,
-      "learning_rate": 6.534572864321608e-05,
-      "loss": 5.1171,
-      "step": 34993
-    },
-    {
-      "epoch": 18.24980443285528,
-      "grad_norm": 1.4139642715454102,
-      "learning_rate": 6.534472361809046e-05,
-      "loss": 5.5253,
-      "step": 34994
-    },
-    {
-      "epoch": 18.2503259452412,
-      "grad_norm": 1.4739009141921997,
-      "learning_rate": 6.534371859296484e-05,
-      "loss": 5.2792,
-      "step": 34995
-    },
-    {
-      "epoch": 18.250847457627117,
-      "grad_norm": 1.4067747592926025,
-      "learning_rate": 6.53427135678392e-05,
-      "loss": 5.6618,
-      "step": 34996
-    },
-    {
-      "epoch": 18.251368970013036,
-      "grad_norm": 1.541038990020752,
-      "learning_rate": 6.534170854271358e-05,
-      "loss": 5.1972,
-      "step": 34997
-    },
-    {
-      "epoch": 18.251890482398956,
-      "grad_norm": 1.5615936517715454,
-      "learning_rate": 6.534070351758794e-05,
-      "loss": 5.0422,
-      "step": 34998
-    },
-    {
-      "epoch": 18.252411994784875,
-      "grad_norm": 2.518430233001709,
-      "learning_rate": 6.533969849246232e-05,
-      "loss": 4.4554,
-      "step": 34999
-    },
-    {
-      "epoch": 18.252933507170795,
-      "grad_norm": 1.599814772605896,
-      "learning_rate": 6.533869346733669e-05,
-      "loss": 4.9589,
-      "step": 35000
-    },
-    {
-      "epoch": 18.252933507170795,
-      "eval_loss": 5.407883644104004,
-      "eval_runtime": 42.6695,
-      "eval_samples_per_second": 28.732,
-      "eval_steps_per_second": 3.609,
-      "step": 35000
-    },
-    {
-      "epoch": 18.253455019556714,
-      "grad_norm": 1.4744583368301392,
-      "learning_rate": 6.533768844221105e-05,
-      "loss": 5.6146,
-      "step": 35001
-    },
-    {
-      "epoch": 18.253976531942634,
-      "grad_norm": 1.5093780755996704,
-      "learning_rate": 6.533668341708543e-05,
-      "loss": 5.5566,
-      "step": 35002
-    },
-    {
-      "epoch": 18.254498044328553,
-      "grad_norm": 1.4635560512542725,
-      "learning_rate": 6.533567839195979e-05,
-      "loss": 4.7602,
-      "step": 35003
-    },
-    {
-      "epoch": 18.255019556714473,
-      "grad_norm": 1.4679769277572632,
-      "learning_rate": 6.533467336683417e-05,
-      "loss": 5.2049,
-      "step": 35004
-    },
-    {
-      "epoch": 18.255541069100392,
-      "grad_norm": 1.4327971935272217,
-      "learning_rate": 6.533366834170855e-05,
-      "loss": 5.5009,
-      "step": 35005
-    },
-    {
-      "epoch": 18.25606258148631,
-      "grad_norm": 1.4482378959655762,
-      "learning_rate": 6.533266331658293e-05,
-      "loss": 5.3461,
-      "step": 35006
-    },
-    {
-      "epoch": 18.25658409387223,
-      "grad_norm": 1.4593786001205444,
-      "learning_rate": 6.533165829145729e-05,
-      "loss": 5.2237,
-      "step": 35007
-    },
-    {
-      "epoch": 18.257105606258147,
-      "grad_norm": 1.5640251636505127,
-      "learning_rate": 6.533065326633167e-05,
-      "loss": 5.0657,
-      "step": 35008
-    },
-    {
-      "epoch": 18.257627118644066,
-      "grad_norm": 1.5037530660629272,
-      "learning_rate": 6.532964824120603e-05,
-      "loss": 4.9187,
-      "step": 35009
-    },
-    {
-      "epoch": 18.258148631029986,
-      "grad_norm": 1.3998653888702393,
-      "learning_rate": 6.532864321608041e-05,
-      "loss": 5.3342,
-      "step": 35010
-    },
-    {
-      "epoch": 18.258670143415905,
-      "grad_norm": 1.4953688383102417,
-      "learning_rate": 6.532763819095477e-05,
-      "loss": 5.0601,
-      "step": 35011
-    },
-    {
-      "epoch": 18.259191655801825,
-      "grad_norm": 1.463054895401001,
-      "learning_rate": 6.532663316582915e-05,
-      "loss": 5.3887,
-      "step": 35012
-    },
-    {
-      "epoch": 18.259713168187744,
-      "grad_norm": 1.44282066822052,
-      "learning_rate": 6.532562814070352e-05,
-      "loss": 5.3184,
-      "step": 35013
-    },
-    {
-      "epoch": 18.260234680573664,
-      "grad_norm": 1.399288535118103,
-      "learning_rate": 6.53246231155779e-05,
-      "loss": 5.563,
-      "step": 35014
-    },
-    {
-      "epoch": 18.260756192959583,
-      "grad_norm": 1.474032998085022,
-      "learning_rate": 6.532361809045227e-05,
-      "loss": 5.0232,
-      "step": 35015
-    },
-    {
-      "epoch": 18.261277705345503,
-      "grad_norm": 1.4506990909576416,
-      "learning_rate": 6.532261306532664e-05,
-      "loss": 5.6039,
-      "step": 35016
-    },
-    {
-      "epoch": 18.261799217731422,
-      "grad_norm": 1.3858733177185059,
-      "learning_rate": 6.532160804020101e-05,
-      "loss": 5.6229,
-      "step": 35017
-    },
-    {
-      "epoch": 18.26232073011734,
-      "grad_norm": 1.4483445882797241,
-      "learning_rate": 6.532060301507538e-05,
-      "loss": 5.1575,
-      "step": 35018
-    },
-    {
-      "epoch": 18.26284224250326,
-      "grad_norm": 1.5089054107666016,
-      "learning_rate": 6.531959798994976e-05,
-      "loss": 5.6749,
-      "step": 35019
-    },
-    {
-      "epoch": 18.263363754889177,
-      "grad_norm": 1.5345643758773804,
-      "learning_rate": 6.531859296482412e-05,
-      "loss": 4.9062,
-      "step": 35020
-    },
-    {
-      "epoch": 18.263885267275096,
-      "grad_norm": 1.5761226415634155,
-      "learning_rate": 6.53175879396985e-05,
-      "loss": 5.0435,
-      "step": 35021
-    },
-    {
-      "epoch": 18.264406779661016,
-      "grad_norm": 1.5919557809829712,
-      "learning_rate": 6.531658291457286e-05,
-      "loss": 5.1726,
-      "step": 35022
-    },
-    {
-      "epoch": 18.264928292046935,
-      "grad_norm": 1.417968988418579,
-      "learning_rate": 6.531557788944724e-05,
-      "loss": 5.4721,
-      "step": 35023
-    },
-    {
-      "epoch": 18.265449804432855,
-      "grad_norm": 1.5379531383514404,
-      "learning_rate": 6.53145728643216e-05,
-      "loss": 5.6888,
-      "step": 35024
-    },
-    {
-      "epoch": 18.265971316818774,
-      "grad_norm": 1.709830403327942,
-      "learning_rate": 6.531356783919598e-05,
-      "loss": 5.115,
-      "step": 35025
-    },
-    {
-      "epoch": 18.266492829204694,
-      "grad_norm": 1.543633222579956,
-      "learning_rate": 6.531256281407036e-05,
-      "loss": 4.8994,
-      "step": 35026
-    },
-    {
-      "epoch": 18.267014341590613,
-      "grad_norm": 1.597963571548462,
-      "learning_rate": 6.531155778894472e-05,
-      "loss": 5.0651,
-      "step": 35027
-    },
-    {
-      "epoch": 18.267535853976533,
-      "grad_norm": 1.4522123336791992,
-      "learning_rate": 6.53105527638191e-05,
-      "loss": 5.102,
-      "step": 35028
-    },
-    {
-      "epoch": 18.268057366362452,
-      "grad_norm": 1.4743984937667847,
-      "learning_rate": 6.530954773869347e-05,
-      "loss": 5.4168,
-      "step": 35029
-    },
-    {
-      "epoch": 18.26857887874837,
-      "grad_norm": 1.477513074874878,
-      "learning_rate": 6.530854271356784e-05,
-      "loss": 5.2536,
-      "step": 35030
-    },
-    {
-      "epoch": 18.26910039113429,
-      "grad_norm": 1.4317578077316284,
-      "learning_rate": 6.530753768844221e-05,
-      "loss": 5.0996,
-      "step": 35031
-    },
-    {
-      "epoch": 18.269621903520207,
-      "grad_norm": 1.428789496421814,
-      "learning_rate": 6.530653266331659e-05,
-      "loss": 5.117,
-      "step": 35032
-    },
-    {
-      "epoch": 18.270143415906126,
-      "grad_norm": 1.5860494375228882,
-      "learning_rate": 6.530552763819095e-05,
-      "loss": 4.809,
-      "step": 35033
-    },
-    {
-      "epoch": 18.270664928292046,
-      "grad_norm": 1.4391672611236572,
-      "learning_rate": 6.530452261306533e-05,
-      "loss": 4.9687,
-      "step": 35034
-    },
-    {
-      "epoch": 18.271186440677965,
-      "grad_norm": 1.4418835639953613,
-      "learning_rate": 6.53035175879397e-05,
-      "loss": 5.528,
-      "step": 35035
-    },
-    {
-      "epoch": 18.271707953063885,
-      "grad_norm": 1.4729626178741455,
-      "learning_rate": 6.530251256281408e-05,
-      "loss": 5.2565,
-      "step": 35036
-    },
-    {
-      "epoch": 18.272229465449804,
-      "grad_norm": 1.4033286571502686,
-      "learning_rate": 6.530150753768845e-05,
-      "loss": 4.9116,
-      "step": 35037
-    },
-    {
-      "epoch": 18.272750977835724,
-      "grad_norm": 1.4527310132980347,
-      "learning_rate": 6.530050251256283e-05,
-      "loss": 5.459,
-      "step": 35038
-    },
-    {
-      "epoch": 18.273272490221643,
-      "grad_norm": 1.5872374773025513,
-      "learning_rate": 6.529949748743719e-05,
-      "loss": 5.334,
-      "step": 35039
-    },
-    {
-      "epoch": 18.273794002607563,
-      "grad_norm": 1.3629285097122192,
-      "learning_rate": 6.529849246231155e-05,
-      "loss": 5.4291,
-      "step": 35040
-    },
-    {
-      "epoch": 18.274315514993482,
-      "grad_norm": 1.4781479835510254,
-      "learning_rate": 6.529748743718593e-05,
-      "loss": 5.0934,
-      "step": 35041
-    },
-    {
-      "epoch": 18.2748370273794,
-      "grad_norm": 1.5002739429473877,
-      "learning_rate": 6.52964824120603e-05,
-      "loss": 4.7202,
-      "step": 35042
-    },
-    {
-      "epoch": 18.27535853976532,
-      "grad_norm": 1.6170668601989746,
-      "learning_rate": 6.529547738693467e-05,
-      "loss": 5.4367,
-      "step": 35043
-    },
-    {
-      "epoch": 18.275880052151237,
-      "grad_norm": 1.4809002876281738,
-      "learning_rate": 6.529447236180904e-05,
-      "loss": 5.6315,
-      "step": 35044
-    },
-    {
-      "epoch": 18.276401564537156,
-      "grad_norm": 1.5057294368743896,
-      "learning_rate": 6.529346733668342e-05,
-      "loss": 5.1497,
-      "step": 35045
-    },
-    {
-      "epoch": 18.276923076923076,
-      "grad_norm": 1.4007129669189453,
-      "learning_rate": 6.52924623115578e-05,
-      "loss": 5.6067,
-      "step": 35046
-    },
-    {
-      "epoch": 18.277444589308995,
-      "grad_norm": 1.524448275566101,
-      "learning_rate": 6.529145728643217e-05,
-      "loss": 5.753,
-      "step": 35047
-    },
-    {
-      "epoch": 18.277966101694915,
-      "grad_norm": 1.59407639503479,
-      "learning_rate": 6.529045226130654e-05,
-      "loss": 5.0088,
-      "step": 35048
-    },
-    {
-      "epoch": 18.278487614080834,
-      "grad_norm": 1.3730294704437256,
-      "learning_rate": 6.528944723618091e-05,
-      "loss": 5.6526,
-      "step": 35049
-    },
-    {
-      "epoch": 18.279009126466754,
-      "grad_norm": 1.4471385478973389,
-      "learning_rate": 6.528844221105528e-05,
-      "loss": 5.2925,
-      "step": 35050
-    },
-    {
-      "epoch": 18.279530638852673,
-      "grad_norm": 1.5030394792556763,
-      "learning_rate": 6.528743718592966e-05,
-      "loss": 5.6132,
-      "step": 35051
-    },
-    {
-      "epoch": 18.280052151238593,
-      "grad_norm": 1.5707652568817139,
-      "learning_rate": 6.528643216080402e-05,
-      "loss": 5.1568,
-      "step": 35052
-    },
-    {
-      "epoch": 18.280573663624512,
-      "grad_norm": 1.4697586297988892,
-      "learning_rate": 6.528542713567839e-05,
-      "loss": 4.9857,
-      "step": 35053
-    },
-    {
-      "epoch": 18.28109517601043,
-      "grad_norm": 1.4924544095993042,
-      "learning_rate": 6.528442211055276e-05,
-      "loss": 5.4866,
-      "step": 35054
-    },
-    {
-      "epoch": 18.28161668839635,
-      "grad_norm": 1.5782535076141357,
-      "learning_rate": 6.528341708542714e-05,
-      "loss": 4.8994,
-      "step": 35055
-    },
-    {
-      "epoch": 18.282138200782267,
-      "grad_norm": 1.4191405773162842,
-      "learning_rate": 6.528241206030152e-05,
-      "loss": 5.4888,
-      "step": 35056
-    },
-    {
-      "epoch": 18.282659713168186,
-      "grad_norm": 1.4657505750656128,
-      "learning_rate": 6.528140703517588e-05,
-      "loss": 5.043,
-      "step": 35057
-    },
-    {
-      "epoch": 18.283181225554106,
-      "grad_norm": 1.4569674730300903,
-      "learning_rate": 6.528040201005026e-05,
-      "loss": 5.4564,
-      "step": 35058
-    },
-    {
-      "epoch": 18.283702737940025,
-      "grad_norm": 1.6470770835876465,
-      "learning_rate": 6.527939698492463e-05,
-      "loss": 5.2388,
-      "step": 35059
-    },
-    {
-      "epoch": 18.284224250325945,
-      "grad_norm": 1.5584067106246948,
-      "learning_rate": 6.5278391959799e-05,
-      "loss": 5.1508,
-      "step": 35060
-    },
-    {
-      "epoch": 18.284745762711864,
-      "grad_norm": 1.3580663204193115,
-      "learning_rate": 6.527738693467337e-05,
-      "loss": 5.9433,
-      "step": 35061
-    },
-    {
-      "epoch": 18.285267275097784,
-      "grad_norm": 1.402835488319397,
-      "learning_rate": 6.527638190954775e-05,
-      "loss": 5.6628,
-      "step": 35062
-    },
-    {
-      "epoch": 18.285788787483703,
-      "grad_norm": 1.4722617864608765,
-      "learning_rate": 6.527537688442211e-05,
-      "loss": 5.2383,
-      "step": 35063
-    },
-    {
-      "epoch": 18.286310299869623,
-      "grad_norm": 1.444960594177246,
-      "learning_rate": 6.527437185929649e-05,
-      "loss": 5.3096,
-      "step": 35064
-    },
-    {
-      "epoch": 18.286831812255542,
-      "grad_norm": 1.541443109512329,
-      "learning_rate": 6.527336683417085e-05,
-      "loss": 5.2715,
-      "step": 35065
-    },
-    {
-      "epoch": 18.28735332464146,
-      "grad_norm": 1.509670615196228,
-      "learning_rate": 6.527236180904523e-05,
-      "loss": 4.8869,
-      "step": 35066
-    },
-    {
-      "epoch": 18.28787483702738,
-      "grad_norm": 1.3948801755905151,
-      "learning_rate": 6.527135678391961e-05,
-      "loss": 5.5176,
-      "step": 35067
-    },
-    {
-      "epoch": 18.288396349413297,
-      "grad_norm": 1.4195419549942017,
-      "learning_rate": 6.527035175879397e-05,
-      "loss": 5.3927,
-      "step": 35068
-    },
-    {
-      "epoch": 18.288917861799217,
-      "grad_norm": 1.9095288515090942,
-      "learning_rate": 6.526934673366835e-05,
-      "loss": 4.3727,
-      "step": 35069
-    },
-    {
-      "epoch": 18.289439374185136,
-      "grad_norm": 1.4740098714828491,
-      "learning_rate": 6.526834170854271e-05,
-      "loss": 5.4788,
-      "step": 35070
-    },
-    {
-      "epoch": 18.289960886571055,
-      "grad_norm": 1.5836005210876465,
-      "learning_rate": 6.526733668341709e-05,
-      "loss": 5.2222,
-      "step": 35071
-    },
-    {
-      "epoch": 18.290482398956975,
-      "grad_norm": 1.4758213758468628,
-      "learning_rate": 6.526633165829146e-05,
-      "loss": 5.1607,
-      "step": 35072
-    },
-    {
-      "epoch": 18.291003911342894,
-      "grad_norm": 1.5381370782852173,
-      "learning_rate": 6.526532663316583e-05,
-      "loss": 5.0456,
-      "step": 35073
-    },
-    {
-      "epoch": 18.291525423728814,
-      "grad_norm": 1.5346479415893555,
-      "learning_rate": 6.52643216080402e-05,
-      "loss": 5.3359,
-      "step": 35074
-    },
-    {
-      "epoch": 18.292046936114733,
-      "grad_norm": 1.5240410566329956,
-      "learning_rate": 6.526331658291458e-05,
-      "loss": 4.9471,
-      "step": 35075
-    },
-    {
-      "epoch": 18.292568448500653,
-      "grad_norm": 1.4305683374404907,
-      "learning_rate": 6.526231155778895e-05,
-      "loss": 4.8442,
-      "step": 35076
-    },
-    {
-      "epoch": 18.293089960886572,
-      "grad_norm": 1.5130236148834229,
-      "learning_rate": 6.526130653266333e-05,
-      "loss": 5.5346,
-      "step": 35077
-    },
-    {
-      "epoch": 18.29361147327249,
-      "grad_norm": 1.5122520923614502,
-      "learning_rate": 6.52603015075377e-05,
-      "loss": 5.576,
-      "step": 35078
-    },
-    {
-      "epoch": 18.294132985658408,
-      "grad_norm": 1.3756455183029175,
-      "learning_rate": 6.525929648241207e-05,
-      "loss": 5.5373,
-      "step": 35079
-    },
-    {
-      "epoch": 18.294654498044327,
-      "grad_norm": 1.4993983507156372,
-      "learning_rate": 6.525829145728644e-05,
-      "loss": 5.5226,
-      "step": 35080
-    },
-    {
-      "epoch": 18.295176010430247,
-      "grad_norm": 1.4946441650390625,
-      "learning_rate": 6.52572864321608e-05,
-      "loss": 5.3777,
-      "step": 35081
-    },
-    {
-      "epoch": 18.295697522816166,
-      "grad_norm": 1.5364803075790405,
-      "learning_rate": 6.525628140703518e-05,
-      "loss": 5.2701,
-      "step": 35082
-    },
-    {
-      "epoch": 18.296219035202085,
-      "grad_norm": 1.4952648878097534,
-      "learning_rate": 6.525527638190954e-05,
-      "loss": 5.4548,
-      "step": 35083
-    },
-    {
-      "epoch": 18.296740547588005,
-      "grad_norm": 1.432321310043335,
-      "learning_rate": 6.525427135678392e-05,
-      "loss": 5.8266,
-      "step": 35084
-    },
-    {
-      "epoch": 18.297262059973924,
-      "grad_norm": 1.5266563892364502,
-      "learning_rate": 6.525326633165829e-05,
-      "loss": 5.1661,
-      "step": 35085
-    },
-    {
-      "epoch": 18.297783572359844,
-      "grad_norm": 1.5056899785995483,
-      "learning_rate": 6.525226130653266e-05,
-      "loss": 5.6401,
-      "step": 35086
-    },
-    {
-      "epoch": 18.298305084745763,
-      "grad_norm": 1.594820261001587,
-      "learning_rate": 6.525125628140704e-05,
-      "loss": 5.449,
-      "step": 35087
-    },
-    {
-      "epoch": 18.298826597131683,
-      "grad_norm": 1.616301417350769,
-      "learning_rate": 6.525025125628142e-05,
-      "loss": 5.1367,
-      "step": 35088
-    },
-    {
-      "epoch": 18.299348109517602,
-      "grad_norm": 1.5118192434310913,
-      "learning_rate": 6.524924623115578e-05,
-      "loss": 4.9854,
-      "step": 35089
-    },
-    {
-      "epoch": 18.29986962190352,
-      "grad_norm": 1.4813098907470703,
-      "learning_rate": 6.524824120603016e-05,
-      "loss": 5.2121,
-      "step": 35090
-    },
-    {
-      "epoch": 18.300391134289438,
-      "grad_norm": 1.492374062538147,
-      "learning_rate": 6.524723618090453e-05,
-      "loss": 4.7307,
-      "step": 35091
-    },
-    {
-      "epoch": 18.300912646675357,
-      "grad_norm": 1.51215660572052,
-      "learning_rate": 6.52462311557789e-05,
-      "loss": 5.6453,
-      "step": 35092
-    },
-    {
-      "epoch": 18.301434159061277,
-      "grad_norm": 1.5311630964279175,
-      "learning_rate": 6.524522613065327e-05,
-      "loss": 5.7132,
-      "step": 35093
-    },
-    {
-      "epoch": 18.301955671447196,
-      "grad_norm": 1.5042660236358643,
-      "learning_rate": 6.524422110552763e-05,
-      "loss": 5.3321,
-      "step": 35094
-    },
-    {
-      "epoch": 18.302477183833116,
-      "grad_norm": 1.3796346187591553,
-      "learning_rate": 6.524321608040201e-05,
-      "loss": 5.4953,
-      "step": 35095
-    },
-    {
-      "epoch": 18.302998696219035,
-      "grad_norm": 1.4991424083709717,
-      "learning_rate": 6.524221105527637e-05,
-      "loss": 5.2295,
-      "step": 35096
-    },
-    {
-      "epoch": 18.303520208604954,
-      "grad_norm": 1.557072401046753,
-      "learning_rate": 6.524120603015075e-05,
-      "loss": 5.3073,
-      "step": 35097
-    },
-    {
-      "epoch": 18.304041720990874,
-      "grad_norm": 1.4419294595718384,
-      "learning_rate": 6.524020100502513e-05,
-      "loss": 5.7083,
-      "step": 35098
-    },
-    {
-      "epoch": 18.304563233376793,
-      "grad_norm": 1.5642204284667969,
-      "learning_rate": 6.523919597989951e-05,
-      "loss": 4.8507,
-      "step": 35099
-    },
-    {
-      "epoch": 18.305084745762713,
-      "grad_norm": 1.6245651245117188,
-      "learning_rate": 6.523819095477387e-05,
-      "loss": 5.0838,
-      "step": 35100
-    },
-    {
-      "epoch": 18.305606258148632,
-      "grad_norm": 1.489628791809082,
-      "learning_rate": 6.523718592964825e-05,
-      "loss": 5.614,
-      "step": 35101
-    },
-    {
-      "epoch": 18.30612777053455,
-      "grad_norm": 1.5807197093963623,
-      "learning_rate": 6.523618090452261e-05,
-      "loss": 5.003,
-      "step": 35102
-    },
-    {
-      "epoch": 18.306649282920468,
-      "grad_norm": 1.660123586654663,
-      "learning_rate": 6.523517587939699e-05,
-      "loss": 5.0949,
-      "step": 35103
-    },
-    {
-      "epoch": 18.307170795306387,
-      "grad_norm": 1.5544617176055908,
-      "learning_rate": 6.523417085427136e-05,
-      "loss": 4.9605,
-      "step": 35104
-    },
-    {
-      "epoch": 18.307692307692307,
-      "grad_norm": 1.5405970811843872,
-      "learning_rate": 6.523316582914573e-05,
-      "loss": 5.1883,
-      "step": 35105
-    },
-    {
-      "epoch": 18.308213820078226,
-      "grad_norm": 1.5640230178833008,
-      "learning_rate": 6.52321608040201e-05,
-      "loss": 4.8407,
-      "step": 35106
-    },
-    {
-      "epoch": 18.308735332464146,
-      "grad_norm": 1.4821656942367554,
-      "learning_rate": 6.523115577889448e-05,
-      "loss": 5.2668,
-      "step": 35107
-    },
-    {
-      "epoch": 18.309256844850065,
-      "grad_norm": 1.521856665611267,
-      "learning_rate": 6.523015075376885e-05,
-      "loss": 5.5759,
-      "step": 35108
-    },
-    {
-      "epoch": 18.309778357235984,
-      "grad_norm": 1.3172255754470825,
-      "learning_rate": 6.522914572864322e-05,
-      "loss": 5.4853,
-      "step": 35109
-    },
-    {
-      "epoch": 18.310299869621904,
-      "grad_norm": 1.457727313041687,
-      "learning_rate": 6.52281407035176e-05,
-      "loss": 5.345,
-      "step": 35110
-    },
-    {
-      "epoch": 18.310821382007823,
-      "grad_norm": 1.4871201515197754,
-      "learning_rate": 6.522713567839196e-05,
-      "loss": 4.9154,
-      "step": 35111
-    },
-    {
-      "epoch": 18.311342894393743,
-      "grad_norm": 1.4918783903121948,
-      "learning_rate": 6.522613065326634e-05,
-      "loss": 5.4383,
-      "step": 35112
-    },
-    {
-      "epoch": 18.311864406779662,
-      "grad_norm": 1.4231157302856445,
-      "learning_rate": 6.52251256281407e-05,
-      "loss": 5.0122,
-      "step": 35113
-    },
-    {
-      "epoch": 18.312385919165582,
-      "grad_norm": 1.5278922319412231,
-      "learning_rate": 6.522412060301508e-05,
-      "loss": 5.5652,
-      "step": 35114
-    },
-    {
-      "epoch": 18.312907431551498,
-      "grad_norm": 1.7249114513397217,
-      "learning_rate": 6.522311557788944e-05,
-      "loss": 4.6803,
-      "step": 35115
-    },
-    {
-      "epoch": 18.313428943937417,
-      "grad_norm": 1.52769935131073,
-      "learning_rate": 6.522211055276382e-05,
-      "loss": 5.3064,
-      "step": 35116
-    },
-    {
-      "epoch": 18.313950456323337,
-      "grad_norm": 1.5816044807434082,
-      "learning_rate": 6.522110552763819e-05,
-      "loss": 5.1862,
-      "step": 35117
-    },
-    {
-      "epoch": 18.314471968709256,
-      "grad_norm": 1.4829825162887573,
-      "learning_rate": 6.522010050251256e-05,
-      "loss": 5.6199,
-      "step": 35118
-    },
-    {
-      "epoch": 18.314993481095176,
-      "grad_norm": 1.4761645793914795,
-      "learning_rate": 6.521909547738694e-05,
-      "loss": 5.2687,
-      "step": 35119
-    },
-    {
-      "epoch": 18.315514993481095,
-      "grad_norm": 1.425998330116272,
-      "learning_rate": 6.52180904522613e-05,
-      "loss": 4.9591,
-      "step": 35120
-    },
-    {
-      "epoch": 18.316036505867014,
-      "grad_norm": 1.5633008480072021,
-      "learning_rate": 6.521708542713568e-05,
-      "loss": 5.3384,
-      "step": 35121
-    },
-    {
-      "epoch": 18.316558018252934,
-      "grad_norm": 1.5759329795837402,
-      "learning_rate": 6.521608040201005e-05,
-      "loss": 5.0275,
-      "step": 35122
-    },
-    {
-      "epoch": 18.317079530638853,
-      "grad_norm": 1.451280951499939,
-      "learning_rate": 6.521507537688443e-05,
-      "loss": 4.906,
-      "step": 35123
-    },
-    {
-      "epoch": 18.317601043024773,
-      "grad_norm": 1.5154730081558228,
-      "learning_rate": 6.521407035175879e-05,
-      "loss": 5.5236,
-      "step": 35124
-    },
-    {
-      "epoch": 18.318122555410692,
-      "grad_norm": 1.4819393157958984,
-      "learning_rate": 6.521306532663317e-05,
-      "loss": 5.4434,
-      "step": 35125
-    },
-    {
-      "epoch": 18.318644067796612,
-      "grad_norm": 1.6486972570419312,
-      "learning_rate": 6.521206030150753e-05,
-      "loss": 4.8115,
-      "step": 35126
-    },
-    {
-      "epoch": 18.319165580182528,
-      "grad_norm": 1.51009202003479,
-      "learning_rate": 6.521105527638191e-05,
-      "loss": 5.51,
-      "step": 35127
-    },
-    {
-      "epoch": 18.319687092568447,
-      "grad_norm": 1.591661810874939,
-      "learning_rate": 6.521005025125629e-05,
-      "loss": 4.784,
-      "step": 35128
-    },
-    {
-      "epoch": 18.320208604954367,
-      "grad_norm": 1.5596026182174683,
-      "learning_rate": 6.520904522613067e-05,
-      "loss": 4.8631,
-      "step": 35129
-    },
-    {
-      "epoch": 18.320730117340286,
-      "grad_norm": 1.5759214162826538,
-      "learning_rate": 6.520804020100503e-05,
-      "loss": 5.3815,
-      "step": 35130
-    },
-    {
-      "epoch": 18.321251629726206,
-      "grad_norm": 1.4733295440673828,
-      "learning_rate": 6.520703517587941e-05,
-      "loss": 5.177,
-      "step": 35131
-    },
-    {
-      "epoch": 18.321773142112125,
-      "grad_norm": 1.6423019170761108,
-      "learning_rate": 6.520603015075377e-05,
-      "loss": 5.2366,
-      "step": 35132
-    },
-    {
-      "epoch": 18.322294654498045,
-      "grad_norm": 1.4718286991119385,
-      "learning_rate": 6.520502512562814e-05,
-      "loss": 5.5968,
-      "step": 35133
-    },
-    {
-      "epoch": 18.322816166883964,
-      "grad_norm": 1.524812936782837,
-      "learning_rate": 6.520402010050251e-05,
-      "loss": 5.1604,
-      "step": 35134
-    },
-    {
-      "epoch": 18.323337679269883,
-      "grad_norm": 1.4660134315490723,
-      "learning_rate": 6.520301507537688e-05,
-      "loss": 5.3487,
-      "step": 35135
-    },
-    {
-      "epoch": 18.323859191655803,
-      "grad_norm": 1.5009397268295288,
-      "learning_rate": 6.520201005025126e-05,
-      "loss": 5.6135,
-      "step": 35136
-    },
-    {
-      "epoch": 18.324380704041722,
-      "grad_norm": 1.463448405265808,
-      "learning_rate": 6.520100502512562e-05,
-      "loss": 5.4579,
-      "step": 35137
-    },
-    {
-      "epoch": 18.324902216427642,
-      "grad_norm": 1.4928309917449951,
-      "learning_rate": 6.52e-05,
-      "loss": 5.0648,
-      "step": 35138
-    },
-    {
-      "epoch": 18.325423728813558,
-      "grad_norm": 1.6146224737167358,
-      "learning_rate": 6.519899497487438e-05,
-      "loss": 5.3982,
-      "step": 35139
-    },
-    {
-      "epoch": 18.325945241199477,
-      "grad_norm": 1.488438367843628,
-      "learning_rate": 6.519798994974875e-05,
-      "loss": 5.439,
-      "step": 35140
-    },
-    {
-      "epoch": 18.326466753585397,
-      "grad_norm": 1.4773155450820923,
-      "learning_rate": 6.519698492462312e-05,
-      "loss": 4.8582,
-      "step": 35141
-    },
-    {
-      "epoch": 18.326988265971316,
-      "grad_norm": 1.4975755214691162,
-      "learning_rate": 6.51959798994975e-05,
-      "loss": 5.6043,
-      "step": 35142
-    },
-    {
-      "epoch": 18.327509778357236,
-      "grad_norm": 1.5166904926300049,
-      "learning_rate": 6.519497487437186e-05,
-      "loss": 5.2688,
-      "step": 35143
-    },
-    {
-      "epoch": 18.328031290743155,
-      "grad_norm": 1.5336179733276367,
-      "learning_rate": 6.519396984924624e-05,
-      "loss": 5.3195,
-      "step": 35144
-    },
-    {
-      "epoch": 18.328552803129075,
-      "grad_norm": 1.5348587036132812,
-      "learning_rate": 6.51929648241206e-05,
-      "loss": 5.4915,
-      "step": 35145
-    },
-    {
-      "epoch": 18.329074315514994,
-      "grad_norm": 1.5124449729919434,
-      "learning_rate": 6.519195979899497e-05,
-      "loss": 5.049,
-      "step": 35146
-    },
-    {
-      "epoch": 18.329595827900913,
-      "grad_norm": 1.4915082454681396,
-      "learning_rate": 6.519095477386935e-05,
-      "loss": 5.498,
-      "step": 35147
-    },
-    {
-      "epoch": 18.330117340286833,
-      "grad_norm": 1.4981940984725952,
-      "learning_rate": 6.518994974874372e-05,
-      "loss": 5.4586,
-      "step": 35148
-    },
-    {
-      "epoch": 18.330638852672752,
-      "grad_norm": 1.544198751449585,
-      "learning_rate": 6.51889447236181e-05,
-      "loss": 5.1423,
-      "step": 35149
-    },
-    {
-      "epoch": 18.331160365058672,
-      "grad_norm": 1.4918171167373657,
-      "learning_rate": 6.518793969849247e-05,
-      "loss": 5.2995,
-      "step": 35150
-    },
-    {
-      "epoch": 18.331681877444588,
-      "grad_norm": 1.6736890077590942,
-      "learning_rate": 6.518693467336684e-05,
-      "loss": 5.5729,
-      "step": 35151
-    },
-    {
-      "epoch": 18.332203389830507,
-      "grad_norm": 1.4865087270736694,
-      "learning_rate": 6.518592964824121e-05,
-      "loss": 5.2274,
-      "step": 35152
-    },
-    {
-      "epoch": 18.332724902216427,
-      "grad_norm": 1.4728646278381348,
-      "learning_rate": 6.518492462311559e-05,
-      "loss": 5.0253,
-      "step": 35153
-    },
-    {
-      "epoch": 18.333246414602346,
-      "grad_norm": 1.4921391010284424,
-      "learning_rate": 6.518391959798995e-05,
-      "loss": 5.1798,
-      "step": 35154
-    },
-    {
-      "epoch": 18.333767926988266,
-      "grad_norm": 1.6297333240509033,
-      "learning_rate": 6.518291457286433e-05,
-      "loss": 4.8407,
-      "step": 35155
-    },
-    {
-      "epoch": 18.334289439374185,
-      "grad_norm": 1.4869496822357178,
-      "learning_rate": 6.518190954773869e-05,
-      "loss": 5.5811,
-      "step": 35156
-    },
-    {
-      "epoch": 18.334810951760105,
-      "grad_norm": 1.544710636138916,
-      "learning_rate": 6.518090452261307e-05,
-      "loss": 4.7884,
-      "step": 35157
-    },
-    {
-      "epoch": 18.335332464146024,
-      "grad_norm": 1.5675817728042603,
-      "learning_rate": 6.517989949748743e-05,
-      "loss": 5.4433,
-      "step": 35158
-    },
-    {
-      "epoch": 18.335853976531943,
-      "grad_norm": 1.5041277408599854,
-      "learning_rate": 6.517889447236181e-05,
-      "loss": 5.1943,
-      "step": 35159
-    },
-    {
-      "epoch": 18.336375488917863,
-      "grad_norm": 1.5333248376846313,
-      "learning_rate": 6.517788944723619e-05,
-      "loss": 5.3319,
-      "step": 35160
-    },
-    {
-      "epoch": 18.336897001303782,
-      "grad_norm": 1.5351759195327759,
-      "learning_rate": 6.517688442211055e-05,
-      "loss": 5.3293,
-      "step": 35161
-    },
-    {
-      "epoch": 18.3374185136897,
-      "grad_norm": 1.5114445686340332,
-      "learning_rate": 6.517587939698493e-05,
-      "loss": 5.4527,
-      "step": 35162
-    },
-    {
-      "epoch": 18.337940026075618,
-      "grad_norm": 1.5147082805633545,
-      "learning_rate": 6.51748743718593e-05,
-      "loss": 5.5439,
-      "step": 35163
-    },
-    {
-      "epoch": 18.338461538461537,
-      "grad_norm": 1.598482370376587,
-      "learning_rate": 6.517386934673367e-05,
-      "loss": 5.1629,
-      "step": 35164
-    },
-    {
-      "epoch": 18.338983050847457,
-      "grad_norm": 1.5303378105163574,
-      "learning_rate": 6.517286432160804e-05,
-      "loss": 5.0091,
-      "step": 35165
-    },
-    {
-      "epoch": 18.339504563233376,
-      "grad_norm": 1.6546776294708252,
-      "learning_rate": 6.517185929648242e-05,
-      "loss": 4.9706,
-      "step": 35166
-    },
-    {
-      "epoch": 18.340026075619296,
-      "grad_norm": 1.5059641599655151,
-      "learning_rate": 6.517085427135678e-05,
-      "loss": 5.0661,
-      "step": 35167
-    },
-    {
-      "epoch": 18.340547588005215,
-      "grad_norm": 1.449028730392456,
-      "learning_rate": 6.516984924623116e-05,
-      "loss": 5.6342,
-      "step": 35168
-    },
-    {
-      "epoch": 18.341069100391135,
-      "grad_norm": 1.4861338138580322,
-      "learning_rate": 6.516884422110554e-05,
-      "loss": 4.6347,
-      "step": 35169
-    },
-    {
-      "epoch": 18.341590612777054,
-      "grad_norm": 1.4365317821502686,
-      "learning_rate": 6.516783919597991e-05,
-      "loss": 5.3018,
-      "step": 35170
-    },
-    {
-      "epoch": 18.342112125162974,
-      "grad_norm": 1.9669784307479858,
-      "learning_rate": 6.516683417085428e-05,
-      "loss": 4.9078,
-      "step": 35171
-    },
-    {
-      "epoch": 18.342633637548893,
-      "grad_norm": 1.4798266887664795,
-      "learning_rate": 6.516582914572866e-05,
-      "loss": 5.2858,
-      "step": 35172
-    },
-    {
-      "epoch": 18.343155149934812,
-      "grad_norm": 1.4691230058670044,
-      "learning_rate": 6.516482412060302e-05,
-      "loss": 5.5603,
-      "step": 35173
-    },
-    {
-      "epoch": 18.343676662320732,
-      "grad_norm": 1.5285770893096924,
-      "learning_rate": 6.516381909547738e-05,
-      "loss": 5.2734,
-      "step": 35174
-    },
-    {
-      "epoch": 18.344198174706648,
-      "grad_norm": 1.438864827156067,
-      "learning_rate": 6.516281407035176e-05,
-      "loss": 5.5221,
-      "step": 35175
-    },
-    {
-      "epoch": 18.344719687092567,
-      "grad_norm": 1.5140881538391113,
-      "learning_rate": 6.516180904522613e-05,
-      "loss": 5.5927,
-      "step": 35176
-    },
-    {
-      "epoch": 18.345241199478487,
-      "grad_norm": 1.5184677839279175,
-      "learning_rate": 6.51608040201005e-05,
-      "loss": 5.6412,
-      "step": 35177
-    },
-    {
-      "epoch": 18.345762711864406,
-      "grad_norm": 1.644676685333252,
-      "learning_rate": 6.515979899497487e-05,
-      "loss": 5.5343,
-      "step": 35178
-    },
-    {
-      "epoch": 18.346284224250326,
-      "grad_norm": 1.6149630546569824,
-      "learning_rate": 6.515879396984925e-05,
-      "loss": 5.0041,
-      "step": 35179
-    },
-    {
-      "epoch": 18.346805736636245,
-      "grad_norm": 1.5279104709625244,
-      "learning_rate": 6.515778894472362e-05,
-      "loss": 5.2118,
-      "step": 35180
-    },
-    {
-      "epoch": 18.347327249022165,
-      "grad_norm": 1.440938115119934,
-      "learning_rate": 6.5156783919598e-05,
-      "loss": 5.6172,
-      "step": 35181
-    },
-    {
-      "epoch": 18.347848761408084,
-      "grad_norm": 1.5831209421157837,
-      "learning_rate": 6.515577889447237e-05,
-      "loss": 4.6745,
-      "step": 35182
-    },
-    {
-      "epoch": 18.348370273794004,
-      "grad_norm": 1.6178001165390015,
-      "learning_rate": 6.515477386934674e-05,
-      "loss": 4.9988,
-      "step": 35183
-    },
-    {
-      "epoch": 18.348891786179923,
-      "grad_norm": 1.5885592699050903,
-      "learning_rate": 6.515376884422111e-05,
-      "loss": 5.3116,
-      "step": 35184
-    },
-    {
-      "epoch": 18.349413298565842,
-      "grad_norm": 1.516729712486267,
-      "learning_rate": 6.515276381909549e-05,
-      "loss": 5.42,
-      "step": 35185
-    },
-    {
-      "epoch": 18.34993481095176,
-      "grad_norm": 1.4849730730056763,
-      "learning_rate": 6.515175879396985e-05,
-      "loss": 5.3537,
-      "step": 35186
-    },
-    {
-      "epoch": 18.350456323337678,
-      "grad_norm": 1.6143810749053955,
-      "learning_rate": 6.515075376884421e-05,
-      "loss": 5.4134,
-      "step": 35187
-    },
-    {
-      "epoch": 18.350977835723597,
-      "grad_norm": 1.534790277481079,
-      "learning_rate": 6.514974874371859e-05,
-      "loss": 5.0033,
-      "step": 35188
-    },
-    {
-      "epoch": 18.351499348109517,
-      "grad_norm": 1.7004790306091309,
-      "learning_rate": 6.514874371859297e-05,
-      "loss": 5.0305,
-      "step": 35189
-    },
-    {
-      "epoch": 18.352020860495436,
-      "grad_norm": 1.513424038887024,
-      "learning_rate": 6.514773869346735e-05,
-      "loss": 5.1107,
-      "step": 35190
-    },
-    {
-      "epoch": 18.352542372881356,
-      "grad_norm": 1.425889492034912,
-      "learning_rate": 6.514673366834171e-05,
-      "loss": 4.942,
-      "step": 35191
-    },
-    {
-      "epoch": 18.353063885267275,
-      "grad_norm": 1.4612765312194824,
-      "learning_rate": 6.514572864321609e-05,
-      "loss": 5.2527,
-      "step": 35192
-    },
-    {
-      "epoch": 18.353585397653195,
-      "grad_norm": 1.5380311012268066,
-      "learning_rate": 6.514472361809045e-05,
-      "loss": 5.0215,
-      "step": 35193
-    },
-    {
-      "epoch": 18.354106910039114,
-      "grad_norm": 1.5127739906311035,
-      "learning_rate": 6.514371859296483e-05,
-      "loss": 5.4098,
-      "step": 35194
-    },
-    {
-      "epoch": 18.354628422425034,
-      "grad_norm": 1.4190622568130493,
-      "learning_rate": 6.51427135678392e-05,
-      "loss": 5.6236,
-      "step": 35195
-    },
-    {
-      "epoch": 18.355149934810953,
-      "grad_norm": 1.5216056108474731,
-      "learning_rate": 6.514170854271357e-05,
-      "loss": 5.133,
-      "step": 35196
-    },
-    {
-      "epoch": 18.355671447196872,
-      "grad_norm": 1.4676140546798706,
-      "learning_rate": 6.514070351758794e-05,
-      "loss": 5.5132,
-      "step": 35197
-    },
-    {
-      "epoch": 18.35619295958279,
-      "grad_norm": 1.5536983013153076,
-      "learning_rate": 6.513969849246232e-05,
-      "loss": 4.7484,
-      "step": 35198
-    },
-    {
-      "epoch": 18.356714471968708,
-      "grad_norm": 1.588032603263855,
-      "learning_rate": 6.513869346733668e-05,
-      "loss": 4.976,
-      "step": 35199
-    },
-    {
-      "epoch": 18.357235984354627,
-      "grad_norm": 1.468733549118042,
-      "learning_rate": 6.513768844221106e-05,
-      "loss": 5.6676,
-      "step": 35200
-    },
-    {
-      "epoch": 18.357757496740547,
-      "grad_norm": 1.4783326387405396,
-      "learning_rate": 6.513668341708544e-05,
-      "loss": 5.2403,
-      "step": 35201
-    },
-    {
-      "epoch": 18.358279009126466,
-      "grad_norm": 1.6664625406265259,
-      "learning_rate": 6.51356783919598e-05,
-      "loss": 5.0735,
-      "step": 35202
-    },
-    {
-      "epoch": 18.358800521512386,
-      "grad_norm": 1.547019362449646,
-      "learning_rate": 6.513467336683418e-05,
-      "loss": 5.2245,
-      "step": 35203
-    },
-    {
-      "epoch": 18.359322033898305,
-      "grad_norm": 1.615593433380127,
-      "learning_rate": 6.513366834170854e-05,
-      "loss": 4.5041,
-      "step": 35204
-    },
-    {
-      "epoch": 18.359843546284225,
-      "grad_norm": 1.4702873229980469,
-      "learning_rate": 6.513266331658292e-05,
-      "loss": 5.25,
-      "step": 35205
-    },
-    {
-      "epoch": 18.360365058670144,
-      "grad_norm": 1.5272537469863892,
-      "learning_rate": 6.513165829145728e-05,
-      "loss": 5.3529,
-      "step": 35206
-    },
-    {
-      "epoch": 18.360886571056064,
-      "grad_norm": 1.4862070083618164,
-      "learning_rate": 6.513065326633166e-05,
-      "loss": 5.7859,
-      "step": 35207
-    },
-    {
-      "epoch": 18.361408083441983,
-      "grad_norm": 1.4013558626174927,
-      "learning_rate": 6.512964824120603e-05,
-      "loss": 5.6293,
-      "step": 35208
-    },
-    {
-      "epoch": 18.361929595827903,
-      "grad_norm": 1.678311824798584,
-      "learning_rate": 6.51286432160804e-05,
-      "loss": 5.1165,
-      "step": 35209
-    },
-    {
-      "epoch": 18.36245110821382,
-      "grad_norm": 1.4972341060638428,
-      "learning_rate": 6.512763819095478e-05,
-      "loss": 5.3459,
-      "step": 35210
-    },
-    {
-      "epoch": 18.362972620599738,
-      "grad_norm": 1.4877958297729492,
-      "learning_rate": 6.512663316582916e-05,
-      "loss": 5.2868,
-      "step": 35211
-    },
-    {
-      "epoch": 18.363494132985657,
-      "grad_norm": 1.600679636001587,
-      "learning_rate": 6.512562814070352e-05,
-      "loss": 4.9573,
-      "step": 35212
-    },
-    {
-      "epoch": 18.364015645371577,
-      "grad_norm": 1.6013747453689575,
-      "learning_rate": 6.512462311557789e-05,
-      "loss": 5.1246,
-      "step": 35213
-    },
-    {
-      "epoch": 18.364537157757496,
-      "grad_norm": 1.4502426385879517,
-      "learning_rate": 6.512361809045227e-05,
-      "loss": 5.3007,
-      "step": 35214
-    },
-    {
-      "epoch": 18.365058670143416,
-      "grad_norm": 1.5188653469085693,
-      "learning_rate": 6.512261306532663e-05,
-      "loss": 5.8121,
-      "step": 35215
-    },
-    {
-      "epoch": 18.365580182529335,
-      "grad_norm": 1.6589570045471191,
-      "learning_rate": 6.512160804020101e-05,
-      "loss": 5.2324,
-      "step": 35216
-    },
-    {
-      "epoch": 18.366101694915255,
-      "grad_norm": 1.7599300146102905,
-      "learning_rate": 6.512060301507537e-05,
-      "loss": 5.1238,
-      "step": 35217
-    },
-    {
-      "epoch": 18.366623207301174,
-      "grad_norm": 1.5365235805511475,
-      "learning_rate": 6.511959798994975e-05,
-      "loss": 4.8656,
-      "step": 35218
-    },
-    {
-      "epoch": 18.367144719687094,
-      "grad_norm": 1.5722875595092773,
-      "learning_rate": 6.511859296482412e-05,
-      "loss": 4.9021,
-      "step": 35219
-    },
-    {
-      "epoch": 18.367666232073013,
-      "grad_norm": 1.3917123079299927,
-      "learning_rate": 6.511758793969849e-05,
-      "loss": 5.5974,
-      "step": 35220
-    },
-    {
-      "epoch": 18.368187744458933,
-      "grad_norm": 1.4243500232696533,
-      "learning_rate": 6.511658291457287e-05,
-      "loss": 5.5722,
-      "step": 35221
-    },
-    {
-      "epoch": 18.36870925684485,
-      "grad_norm": 1.4543038606643677,
-      "learning_rate": 6.511557788944725e-05,
-      "loss": 5.3488,
-      "step": 35222
-    },
-    {
-      "epoch": 18.369230769230768,
-      "grad_norm": 1.4754700660705566,
-      "learning_rate": 6.511457286432161e-05,
-      "loss": 5.2769,
-      "step": 35223
-    },
-    {
-      "epoch": 18.369752281616687,
-      "grad_norm": 1.427228331565857,
-      "learning_rate": 6.511356783919599e-05,
-      "loss": 5.4373,
-      "step": 35224
-    },
-    {
-      "epoch": 18.370273794002607,
-      "grad_norm": 1.4878312349319458,
-      "learning_rate": 6.511256281407036e-05,
-      "loss": 5.1905,
-      "step": 35225
-    },
-    {
-      "epoch": 18.370795306388526,
-      "grad_norm": 1.4094617366790771,
-      "learning_rate": 6.511155778894472e-05,
-      "loss": 5.2975,
-      "step": 35226
-    },
-    {
-      "epoch": 18.371316818774446,
-      "grad_norm": 1.4788113832473755,
-      "learning_rate": 6.51105527638191e-05,
-      "loss": 5.7644,
-      "step": 35227
-    },
-    {
-      "epoch": 18.371838331160365,
-      "grad_norm": 1.34215247631073,
-      "learning_rate": 6.510954773869346e-05,
-      "loss": 5.6988,
-      "step": 35228
-    },
-    {
-      "epoch": 18.372359843546285,
-      "grad_norm": 1.4588916301727295,
-      "learning_rate": 6.510854271356784e-05,
-      "loss": 5.3242,
-      "step": 35229
-    },
-    {
-      "epoch": 18.372881355932204,
-      "grad_norm": 1.528038740158081,
-      "learning_rate": 6.510753768844222e-05,
-      "loss": 5.3378,
-      "step": 35230
-    },
-    {
-      "epoch": 18.373402868318124,
-      "grad_norm": 1.4804543256759644,
-      "learning_rate": 6.51065326633166e-05,
-      "loss": 5.1651,
-      "step": 35231
-    },
-    {
-      "epoch": 18.373924380704043,
-      "grad_norm": 1.5230283737182617,
-      "learning_rate": 6.510552763819096e-05,
-      "loss": 5.029,
-      "step": 35232
-    },
-    {
-      "epoch": 18.374445893089963,
-      "grad_norm": 1.5226306915283203,
-      "learning_rate": 6.510452261306534e-05,
-      "loss": 5.6098,
-      "step": 35233
-    },
-    {
-      "epoch": 18.37496740547588,
-      "grad_norm": 1.435072422027588,
-      "learning_rate": 6.51035175879397e-05,
-      "loss": 5.1473,
-      "step": 35234
-    },
-    {
-      "epoch": 18.375488917861798,
-      "grad_norm": 1.4489738941192627,
-      "learning_rate": 6.510251256281408e-05,
-      "loss": 4.9171,
-      "step": 35235
-    },
-    {
-      "epoch": 18.376010430247717,
-      "grad_norm": 1.5729494094848633,
-      "learning_rate": 6.510150753768844e-05,
-      "loss": 5.4577,
-      "step": 35236
-    },
-    {
-      "epoch": 18.376531942633637,
-      "grad_norm": 1.5707871913909912,
-      "learning_rate": 6.510050251256282e-05,
-      "loss": 5.6443,
-      "step": 35237
-    },
-    {
-      "epoch": 18.377053455019556,
-      "grad_norm": 1.555337905883789,
-      "learning_rate": 6.509949748743719e-05,
-      "loss": 5.2593,
-      "step": 35238
-    },
-    {
-      "epoch": 18.377574967405476,
-      "grad_norm": 1.600921630859375,
-      "learning_rate": 6.509849246231155e-05,
-      "loss": 4.8542,
-      "step": 35239
-    },
-    {
-      "epoch": 18.378096479791395,
-      "grad_norm": 1.6057699918746948,
-      "learning_rate": 6.509748743718593e-05,
-      "loss": 5.3746,
-      "step": 35240
-    },
-    {
-      "epoch": 18.378617992177315,
-      "grad_norm": 1.6773979663848877,
-      "learning_rate": 6.50964824120603e-05,
-      "loss": 5.1454,
-      "step": 35241
-    },
-    {
-      "epoch": 18.379139504563234,
-      "grad_norm": 1.441369891166687,
-      "learning_rate": 6.509547738693468e-05,
-      "loss": 5.3756,
-      "step": 35242
-    },
-    {
-      "epoch": 18.379661016949154,
-      "grad_norm": 1.4600188732147217,
-      "learning_rate": 6.509447236180905e-05,
-      "loss": 5.2631,
-      "step": 35243
-    },
-    {
-      "epoch": 18.380182529335073,
-      "grad_norm": 1.449599027633667,
-      "learning_rate": 6.509346733668343e-05,
-      "loss": 5.4044,
-      "step": 35244
-    },
-    {
-      "epoch": 18.380704041720993,
-      "grad_norm": 1.607900619506836,
-      "learning_rate": 6.509246231155779e-05,
-      "loss": 5.5113,
-      "step": 35245
-    },
-    {
-      "epoch": 18.38122555410691,
-      "grad_norm": 1.5998753309249878,
-      "learning_rate": 6.509145728643217e-05,
-      "loss": 4.9982,
-      "step": 35246
-    },
-    {
-      "epoch": 18.381747066492828,
-      "grad_norm": 1.419301986694336,
-      "learning_rate": 6.509045226130653e-05,
-      "loss": 5.174,
-      "step": 35247
-    },
-    {
-      "epoch": 18.382268578878747,
-      "grad_norm": 1.4373308420181274,
-      "learning_rate": 6.508944723618091e-05,
-      "loss": 5.3976,
-      "step": 35248
-    },
-    {
-      "epoch": 18.382790091264667,
-      "grad_norm": 1.8252958059310913,
-      "learning_rate": 6.508844221105527e-05,
-      "loss": 4.9375,
-      "step": 35249
-    },
-    {
-      "epoch": 18.383311603650586,
-      "grad_norm": 1.4874064922332764,
-      "learning_rate": 6.508743718592965e-05,
-      "loss": 5.4297,
-      "step": 35250
-    },
-    {
-      "epoch": 18.383833116036506,
-      "grad_norm": 1.4181715250015259,
-      "learning_rate": 6.508643216080402e-05,
-      "loss": 5.5747,
-      "step": 35251
-    },
-    {
-      "epoch": 18.384354628422425,
-      "grad_norm": 1.4135088920593262,
-      "learning_rate": 6.50854271356784e-05,
-      "loss": 4.9607,
-      "step": 35252
-    },
-    {
-      "epoch": 18.384876140808345,
-      "grad_norm": 1.6797513961791992,
-      "learning_rate": 6.508442211055277e-05,
-      "loss": 5.2303,
-      "step": 35253
-    },
-    {
-      "epoch": 18.385397653194264,
-      "grad_norm": 1.49441397190094,
-      "learning_rate": 6.508341708542714e-05,
-      "loss": 5.446,
-      "step": 35254
-    },
-    {
-      "epoch": 18.385919165580184,
-      "grad_norm": 1.5303007364273071,
-      "learning_rate": 6.508241206030151e-05,
-      "loss": 4.9379,
-      "step": 35255
-    },
-    {
-      "epoch": 18.386440677966103,
-      "grad_norm": 1.5456660985946655,
-      "learning_rate": 6.508140703517588e-05,
-      "loss": 5.3577,
-      "step": 35256
-    },
-    {
-      "epoch": 18.386962190352023,
-      "grad_norm": 1.534050464630127,
-      "learning_rate": 6.508040201005026e-05,
-      "loss": 5.0826,
-      "step": 35257
-    },
-    {
-      "epoch": 18.38748370273794,
-      "grad_norm": 1.5652581453323364,
-      "learning_rate": 6.507939698492462e-05,
-      "loss": 5.0116,
-      "step": 35258
-    },
-    {
-      "epoch": 18.388005215123858,
-      "grad_norm": 1.5942519903182983,
-      "learning_rate": 6.5078391959799e-05,
-      "loss": 5.5833,
-      "step": 35259
-    },
-    {
-      "epoch": 18.388526727509777,
-      "grad_norm": 1.4917103052139282,
-      "learning_rate": 6.507738693467336e-05,
-      "loss": 4.8085,
-      "step": 35260
-    },
-    {
-      "epoch": 18.389048239895697,
-      "grad_norm": 1.506113052368164,
-      "learning_rate": 6.507638190954774e-05,
-      "loss": 5.5569,
-      "step": 35261
-    },
-    {
-      "epoch": 18.389569752281616,
-      "grad_norm": 1.4957773685455322,
-      "learning_rate": 6.507537688442212e-05,
-      "loss": 5.0019,
-      "step": 35262
-    },
-    {
-      "epoch": 18.390091264667536,
-      "grad_norm": 1.501278281211853,
-      "learning_rate": 6.50743718592965e-05,
-      "loss": 5.6986,
-      "step": 35263
-    },
-    {
-      "epoch": 18.390612777053455,
-      "grad_norm": 1.5174301862716675,
-      "learning_rate": 6.507336683417086e-05,
-      "loss": 5.0302,
-      "step": 35264
-    },
-    {
-      "epoch": 18.391134289439375,
-      "grad_norm": 1.4908496141433716,
-      "learning_rate": 6.507236180904524e-05,
-      "loss": 5.4177,
-      "step": 35265
-    },
-    {
-      "epoch": 18.391655801825294,
-      "grad_norm": 1.4700311422348022,
-      "learning_rate": 6.50713567839196e-05,
-      "loss": 5.3105,
-      "step": 35266
-    },
-    {
-      "epoch": 18.392177314211214,
-      "grad_norm": 1.4361238479614258,
-      "learning_rate": 6.507035175879397e-05,
-      "loss": 5.1158,
-      "step": 35267
-    },
-    {
-      "epoch": 18.392698826597133,
-      "grad_norm": 1.5050125122070312,
-      "learning_rate": 6.506934673366834e-05,
-      "loss": 5.3698,
-      "step": 35268
-    },
-    {
-      "epoch": 18.39322033898305,
-      "grad_norm": 1.5255146026611328,
-      "learning_rate": 6.506834170854271e-05,
-      "loss": 5.0041,
-      "step": 35269
-    },
-    {
-      "epoch": 18.39374185136897,
-      "grad_norm": 1.3809155225753784,
-      "learning_rate": 6.506733668341709e-05,
-      "loss": 5.1002,
-      "step": 35270
-    },
-    {
-      "epoch": 18.394263363754888,
-      "grad_norm": 1.5239441394805908,
-      "learning_rate": 6.506633165829145e-05,
-      "loss": 4.9689,
-      "step": 35271
-    },
-    {
-      "epoch": 18.394784876140807,
-      "grad_norm": 1.4743595123291016,
-      "learning_rate": 6.506532663316583e-05,
-      "loss": 5.7889,
-      "step": 35272
-    },
-    {
-      "epoch": 18.395306388526727,
-      "grad_norm": 1.5137022733688354,
-      "learning_rate": 6.50643216080402e-05,
-      "loss": 5.3488,
-      "step": 35273
-    },
-    {
-      "epoch": 18.395827900912646,
-      "grad_norm": 1.4848301410675049,
-      "learning_rate": 6.506331658291458e-05,
-      "loss": 5.4443,
-      "step": 35274
-    },
-    {
-      "epoch": 18.396349413298566,
-      "grad_norm": 1.462069034576416,
-      "learning_rate": 6.506231155778895e-05,
-      "loss": 5.7674,
-      "step": 35275
-    },
-    {
-      "epoch": 18.396870925684485,
-      "grad_norm": 1.4704978466033936,
-      "learning_rate": 6.506130653266333e-05,
-      "loss": 5.2562,
-      "step": 35276
-    },
-    {
-      "epoch": 18.397392438070405,
-      "grad_norm": 1.490984559059143,
-      "learning_rate": 6.506030150753769e-05,
-      "loss": 5.2877,
-      "step": 35277
-    },
-    {
-      "epoch": 18.397913950456324,
-      "grad_norm": 1.5030500888824463,
-      "learning_rate": 6.505929648241207e-05,
-      "loss": 5.55,
-      "step": 35278
-    },
-    {
-      "epoch": 18.398435462842244,
-      "grad_norm": 1.5085588693618774,
-      "learning_rate": 6.505829145728643e-05,
-      "loss": 5.1266,
-      "step": 35279
-    },
-    {
-      "epoch": 18.398956975228163,
-      "grad_norm": 1.4518979787826538,
-      "learning_rate": 6.50572864321608e-05,
-      "loss": 5.179,
-      "step": 35280
-    },
-    {
-      "epoch": 18.39947848761408,
-      "grad_norm": 1.4150406122207642,
-      "learning_rate": 6.505628140703517e-05,
-      "loss": 5.5495,
-      "step": 35281
-    },
-    {
-      "epoch": 18.4,
-      "grad_norm": 1.4545100927352905,
-      "learning_rate": 6.505527638190955e-05,
-      "loss": 5.7078,
-      "step": 35282
-    },
-    {
-      "epoch": 18.400521512385918,
-      "grad_norm": 1.4610793590545654,
-      "learning_rate": 6.505427135678393e-05,
-      "loss": 5.5073,
-      "step": 35283
-    },
-    {
-      "epoch": 18.401043024771838,
-      "grad_norm": 1.5517970323562622,
-      "learning_rate": 6.50532663316583e-05,
-      "loss": 5.2492,
-      "step": 35284
-    },
-    {
-      "epoch": 18.401564537157757,
-      "grad_norm": 1.5593278408050537,
-      "learning_rate": 6.505226130653267e-05,
-      "loss": 5.3431,
-      "step": 35285
-    },
-    {
-      "epoch": 18.402086049543676,
-      "grad_norm": 1.496038556098938,
-      "learning_rate": 6.505125628140704e-05,
-      "loss": 4.8134,
-      "step": 35286
-    },
-    {
-      "epoch": 18.402607561929596,
-      "grad_norm": 1.461105465888977,
-      "learning_rate": 6.505025125628141e-05,
-      "loss": 5.3797,
-      "step": 35287
-    },
-    {
-      "epoch": 18.403129074315515,
-      "grad_norm": 1.5215188264846802,
-      "learning_rate": 6.504924623115578e-05,
-      "loss": 5.5118,
-      "step": 35288
-    },
-    {
-      "epoch": 18.403650586701435,
-      "grad_norm": 1.519545316696167,
-      "learning_rate": 6.504824120603016e-05,
-      "loss": 5.0402,
-      "step": 35289
-    },
-    {
-      "epoch": 18.404172099087354,
-      "grad_norm": 1.4486206769943237,
-      "learning_rate": 6.504723618090452e-05,
-      "loss": 5.4524,
-      "step": 35290
-    },
-    {
-      "epoch": 18.404693611473274,
-      "grad_norm": 1.5370471477508545,
-      "learning_rate": 6.50462311557789e-05,
-      "loss": 5.2161,
-      "step": 35291
-    },
-    {
-      "epoch": 18.405215123859193,
-      "grad_norm": 1.5663869380950928,
-      "learning_rate": 6.504522613065326e-05,
-      "loss": 5.285,
-      "step": 35292
-    },
-    {
-      "epoch": 18.40573663624511,
-      "grad_norm": 1.6275314092636108,
-      "learning_rate": 6.504422110552764e-05,
-      "loss": 4.4001,
-      "step": 35293
-    },
-    {
-      "epoch": 18.40625814863103,
-      "grad_norm": 1.4769930839538574,
-      "learning_rate": 6.504321608040202e-05,
-      "loss": 5.482,
-      "step": 35294
-    },
-    {
-      "epoch": 18.406779661016948,
-      "grad_norm": 1.4571819305419922,
-      "learning_rate": 6.504221105527638e-05,
-      "loss": 5.5107,
-      "step": 35295
-    },
-    {
-      "epoch": 18.407301173402868,
-      "grad_norm": 1.4298310279846191,
-      "learning_rate": 6.504120603015076e-05,
-      "loss": 5.6881,
-      "step": 35296
-    },
-    {
-      "epoch": 18.407822685788787,
-      "grad_norm": 1.5141212940216064,
-      "learning_rate": 6.504020100502513e-05,
-      "loss": 5.2055,
-      "step": 35297
-    },
-    {
-      "epoch": 18.408344198174706,
-      "grad_norm": 1.4593467712402344,
-      "learning_rate": 6.50391959798995e-05,
-      "loss": 5.7636,
-      "step": 35298
-    },
-    {
-      "epoch": 18.408865710560626,
-      "grad_norm": 1.5348776578903198,
-      "learning_rate": 6.503819095477387e-05,
-      "loss": 5.1077,
-      "step": 35299
-    },
-    {
-      "epoch": 18.409387222946545,
-      "grad_norm": 1.6128573417663574,
-      "learning_rate": 6.503718592964825e-05,
-      "loss": 4.9064,
-      "step": 35300
-    },
-    {
-      "epoch": 18.409908735332465,
-      "grad_norm": 1.7701175212860107,
-      "learning_rate": 6.503618090452261e-05,
-      "loss": 4.4341,
-      "step": 35301
-    },
-    {
-      "epoch": 18.410430247718384,
-      "grad_norm": 1.4944113492965698,
-      "learning_rate": 6.503517587939699e-05,
-      "loss": 5.2663,
-      "step": 35302
-    },
-    {
-      "epoch": 18.410951760104304,
-      "grad_norm": 1.6672803163528442,
-      "learning_rate": 6.503417085427136e-05,
-      "loss": 5.2584,
-      "step": 35303
-    },
-    {
-      "epoch": 18.411473272490223,
-      "grad_norm": 1.586869478225708,
-      "learning_rate": 6.503316582914574e-05,
-      "loss": 5.0572,
-      "step": 35304
-    },
-    {
-      "epoch": 18.41199478487614,
-      "grad_norm": 1.427100419998169,
-      "learning_rate": 6.503216080402011e-05,
-      "loss": 5.6722,
-      "step": 35305
-    },
-    {
-      "epoch": 18.41251629726206,
-      "grad_norm": 1.4616844654083252,
-      "learning_rate": 6.503115577889447e-05,
-      "loss": 5.5391,
-      "step": 35306
-    },
-    {
-      "epoch": 18.413037809647978,
-      "grad_norm": 1.4312878847122192,
-      "learning_rate": 6.503015075376885e-05,
-      "loss": 5.2727,
-      "step": 35307
-    },
-    {
-      "epoch": 18.413559322033898,
-      "grad_norm": 1.4902793169021606,
-      "learning_rate": 6.502914572864321e-05,
-      "loss": 5.5041,
-      "step": 35308
-    },
-    {
-      "epoch": 18.414080834419817,
-      "grad_norm": 1.493479609489441,
-      "learning_rate": 6.502814070351759e-05,
-      "loss": 5.294,
-      "step": 35309
-    },
-    {
-      "epoch": 18.414602346805736,
-      "grad_norm": 1.519619107246399,
-      "learning_rate": 6.502713567839196e-05,
-      "loss": 5.171,
-      "step": 35310
-    },
-    {
-      "epoch": 18.415123859191656,
-      "grad_norm": 1.5770891904830933,
-      "learning_rate": 6.502613065326633e-05,
-      "loss": 5.1954,
-      "step": 35311
-    },
-    {
-      "epoch": 18.415645371577575,
-      "grad_norm": 1.3564246892929077,
-      "learning_rate": 6.50251256281407e-05,
-      "loss": 5.2473,
-      "step": 35312
-    },
-    {
-      "epoch": 18.416166883963495,
-      "grad_norm": 1.5160160064697266,
-      "learning_rate": 6.502412060301508e-05,
-      "loss": 5.1289,
-      "step": 35313
-    },
-    {
-      "epoch": 18.416688396349414,
-      "grad_norm": 1.4503087997436523,
-      "learning_rate": 6.502311557788945e-05,
-      "loss": 5.1837,
-      "step": 35314
-    },
-    {
-      "epoch": 18.417209908735334,
-      "grad_norm": 1.4963157176971436,
-      "learning_rate": 6.502211055276383e-05,
-      "loss": 5.743,
-      "step": 35315
-    },
-    {
-      "epoch": 18.417731421121253,
-      "grad_norm": 1.448970913887024,
-      "learning_rate": 6.50211055276382e-05,
-      "loss": 5.6131,
-      "step": 35316
-    },
-    {
-      "epoch": 18.41825293350717,
-      "grad_norm": 1.5873438119888306,
-      "learning_rate": 6.502010050251257e-05,
-      "loss": 5.1507,
-      "step": 35317
-    },
-    {
-      "epoch": 18.41877444589309,
-      "grad_norm": 1.4814525842666626,
-      "learning_rate": 6.501909547738694e-05,
-      "loss": 5.3282,
-      "step": 35318
-    },
-    {
-      "epoch": 18.419295958279008,
-      "grad_norm": 1.6601386070251465,
-      "learning_rate": 6.50180904522613e-05,
-      "loss": 4.3975,
-      "step": 35319
-    },
-    {
-      "epoch": 18.419817470664928,
-      "grad_norm": 1.5429420471191406,
-      "learning_rate": 6.501708542713568e-05,
-      "loss": 5.6248,
-      "step": 35320
-    },
-    {
-      "epoch": 18.420338983050847,
-      "grad_norm": 1.5141254663467407,
-      "learning_rate": 6.501608040201004e-05,
-      "loss": 5.2372,
-      "step": 35321
-    },
-    {
-      "epoch": 18.420860495436767,
-      "grad_norm": 1.4814646244049072,
-      "learning_rate": 6.501507537688442e-05,
-      "loss": 5.1427,
-      "step": 35322
-    },
-    {
-      "epoch": 18.421382007822686,
-      "grad_norm": 1.4261384010314941,
-      "learning_rate": 6.50140703517588e-05,
-      "loss": 5.6754,
-      "step": 35323
-    },
-    {
-      "epoch": 18.421903520208605,
-      "grad_norm": 1.4649937152862549,
-      "learning_rate": 6.501306532663318e-05,
-      "loss": 5.6532,
-      "step": 35324
-    },
-    {
-      "epoch": 18.422425032594525,
-      "grad_norm": 1.6732724905014038,
-      "learning_rate": 6.501206030150754e-05,
-      "loss": 4.7962,
-      "step": 35325
-    },
-    {
-      "epoch": 18.422946544980444,
-      "grad_norm": 1.4529629945755005,
-      "learning_rate": 6.501105527638192e-05,
-      "loss": 5.4328,
-      "step": 35326
-    },
-    {
-      "epoch": 18.423468057366364,
-      "grad_norm": 1.4552761316299438,
-      "learning_rate": 6.501005025125628e-05,
-      "loss": 5.4543,
-      "step": 35327
-    },
-    {
-      "epoch": 18.423989569752283,
-      "grad_norm": 1.551533579826355,
-      "learning_rate": 6.500904522613066e-05,
-      "loss": 5.053,
-      "step": 35328
-    },
-    {
-      "epoch": 18.4245110821382,
-      "grad_norm": 1.5224037170410156,
-      "learning_rate": 6.500804020100503e-05,
-      "loss": 5.4789,
-      "step": 35329
-    },
-    {
-      "epoch": 18.42503259452412,
-      "grad_norm": 1.4882076978683472,
-      "learning_rate": 6.50070351758794e-05,
-      "loss": 5.3402,
-      "step": 35330
-    },
-    {
-      "epoch": 18.425554106910038,
-      "grad_norm": 1.6162794828414917,
-      "learning_rate": 6.500603015075377e-05,
-      "loss": 4.9185,
-      "step": 35331
-    },
-    {
-      "epoch": 18.426075619295958,
-      "grad_norm": 1.4745005369186401,
-      "learning_rate": 6.500502512562813e-05,
-      "loss": 5.422,
-      "step": 35332
-    },
-    {
-      "epoch": 18.426597131681877,
-      "grad_norm": 1.5176873207092285,
-      "learning_rate": 6.500402010050251e-05,
-      "loss": 5.3841,
-      "step": 35333
-    },
-    {
-      "epoch": 18.427118644067797,
-      "grad_norm": 1.3660645484924316,
-      "learning_rate": 6.500301507537689e-05,
-      "loss": 5.867,
-      "step": 35334
-    },
-    {
-      "epoch": 18.427640156453716,
-      "grad_norm": 1.6045658588409424,
-      "learning_rate": 6.500201005025127e-05,
-      "loss": 4.8194,
-      "step": 35335
-    },
-    {
-      "epoch": 18.428161668839635,
-      "grad_norm": 1.5610361099243164,
-      "learning_rate": 6.500100502512563e-05,
-      "loss": 5.2886,
-      "step": 35336
-    },
-    {
-      "epoch": 18.428683181225555,
-      "grad_norm": 1.614809274673462,
-      "learning_rate": 6.500000000000001e-05,
-      "loss": 5.0943,
-      "step": 35337
-    },
-    {
-      "epoch": 18.429204693611474,
-      "grad_norm": 1.4449462890625,
-      "learning_rate": 6.499899497487437e-05,
-      "loss": 5.5332,
-      "step": 35338
-    },
-    {
-      "epoch": 18.429726205997394,
-      "grad_norm": 1.394823670387268,
-      "learning_rate": 6.499798994974875e-05,
-      "loss": 5.6594,
-      "step": 35339
-    },
-    {
-      "epoch": 18.430247718383313,
-      "grad_norm": 1.4832926988601685,
-      "learning_rate": 6.499698492462311e-05,
-      "loss": 5.5112,
-      "step": 35340
-    },
-    {
-      "epoch": 18.43076923076923,
-      "grad_norm": 1.4940158128738403,
-      "learning_rate": 6.499597989949749e-05,
-      "loss": 4.9424,
-      "step": 35341
-    },
-    {
-      "epoch": 18.43129074315515,
-      "grad_norm": 1.5216963291168213,
-      "learning_rate": 6.499497487437186e-05,
-      "loss": 5.3823,
-      "step": 35342
-    },
-    {
-      "epoch": 18.431812255541068,
-      "grad_norm": 1.4266589879989624,
-      "learning_rate": 6.499396984924623e-05,
-      "loss": 5.7538,
-      "step": 35343
-    },
-    {
-      "epoch": 18.432333767926988,
-      "grad_norm": 1.6211663484573364,
-      "learning_rate": 6.499296482412061e-05,
-      "loss": 5.1607,
-      "step": 35344
-    },
-    {
-      "epoch": 18.432855280312907,
-      "grad_norm": 1.4464757442474365,
-      "learning_rate": 6.499195979899498e-05,
-      "loss": 4.7223,
-      "step": 35345
-    },
-    {
-      "epoch": 18.433376792698827,
-      "grad_norm": 1.4377882480621338,
-      "learning_rate": 6.499095477386935e-05,
-      "loss": 5.6721,
-      "step": 35346
-    },
-    {
-      "epoch": 18.433898305084746,
-      "grad_norm": 1.47059965133667,
-      "learning_rate": 6.498994974874372e-05,
-      "loss": 5.567,
-      "step": 35347
-    },
-    {
-      "epoch": 18.434419817470665,
-      "grad_norm": 1.5374125242233276,
-      "learning_rate": 6.49889447236181e-05,
-      "loss": 5.026,
-      "step": 35348
-    },
-    {
-      "epoch": 18.434941329856585,
-      "grad_norm": 1.520053744316101,
-      "learning_rate": 6.498793969849246e-05,
-      "loss": 5.662,
-      "step": 35349
-    },
-    {
-      "epoch": 18.435462842242504,
-      "grad_norm": 1.554174542427063,
-      "learning_rate": 6.498693467336684e-05,
-      "loss": 4.7984,
-      "step": 35350
-    },
-    {
-      "epoch": 18.435984354628424,
-      "grad_norm": 1.4888135194778442,
-      "learning_rate": 6.49859296482412e-05,
-      "loss": 5.654,
-      "step": 35351
-    },
-    {
-      "epoch": 18.43650586701434,
-      "grad_norm": 1.4486197233200073,
-      "learning_rate": 6.498492462311558e-05,
-      "loss": 5.4912,
-      "step": 35352
-    },
-    {
-      "epoch": 18.43702737940026,
-      "grad_norm": 1.4718955755233765,
-      "learning_rate": 6.498391959798994e-05,
-      "loss": 5.4259,
-      "step": 35353
-    },
-    {
-      "epoch": 18.43754889178618,
-      "grad_norm": 1.5083993673324585,
-      "learning_rate": 6.498291457286432e-05,
-      "loss": 4.7653,
-      "step": 35354
-    },
-    {
-      "epoch": 18.438070404172098,
-      "grad_norm": 1.5512456893920898,
-      "learning_rate": 6.49819095477387e-05,
-      "loss": 5.5338,
-      "step": 35355
-    },
-    {
-      "epoch": 18.438591916558018,
-      "grad_norm": 1.5420596599578857,
-      "learning_rate": 6.498090452261308e-05,
-      "loss": 5.3927,
-      "step": 35356
-    },
-    {
-      "epoch": 18.439113428943937,
-      "grad_norm": 1.65750253200531,
-      "learning_rate": 6.497989949748744e-05,
-      "loss": 5.3095,
-      "step": 35357
-    },
-    {
-      "epoch": 18.439634941329857,
-      "grad_norm": 1.4846305847167969,
-      "learning_rate": 6.497889447236182e-05,
-      "loss": 5.3905,
-      "step": 35358
-    },
-    {
-      "epoch": 18.440156453715776,
-      "grad_norm": 1.5360530614852905,
-      "learning_rate": 6.497788944723618e-05,
-      "loss": 4.9349,
-      "step": 35359
-    },
-    {
-      "epoch": 18.440677966101696,
-      "grad_norm": 1.5036289691925049,
-      "learning_rate": 6.497688442211055e-05,
-      "loss": 5.0728,
-      "step": 35360
-    },
-    {
-      "epoch": 18.441199478487615,
-      "grad_norm": 1.5356239080429077,
-      "learning_rate": 6.497587939698493e-05,
-      "loss": 5.6877,
-      "step": 35361
-    },
-    {
-      "epoch": 18.441720990873534,
-      "grad_norm": 1.5183510780334473,
-      "learning_rate": 6.497487437185929e-05,
-      "loss": 5.3498,
-      "step": 35362
-    },
-    {
-      "epoch": 18.442242503259454,
-      "grad_norm": 1.5315098762512207,
-      "learning_rate": 6.497386934673367e-05,
-      "loss": 5.2903,
-      "step": 35363
-    },
-    {
-      "epoch": 18.442764015645373,
-      "grad_norm": 1.5531846284866333,
-      "learning_rate": 6.497286432160805e-05,
-      "loss": 5.2089,
-      "step": 35364
-    },
-    {
-      "epoch": 18.44328552803129,
-      "grad_norm": 1.4574592113494873,
-      "learning_rate": 6.497185929648242e-05,
-      "loss": 5.72,
-      "step": 35365
-    },
-    {
-      "epoch": 18.44380704041721,
-      "grad_norm": 1.4751760959625244,
-      "learning_rate": 6.497085427135679e-05,
-      "loss": 5.3536,
-      "step": 35366
-    },
-    {
-      "epoch": 18.444328552803128,
-      "grad_norm": 1.452718734741211,
-      "learning_rate": 6.496984924623117e-05,
-      "loss": 4.902,
-      "step": 35367
-    },
-    {
-      "epoch": 18.444850065189048,
-      "grad_norm": 1.4632970094680786,
-      "learning_rate": 6.496884422110553e-05,
-      "loss": 5.1402,
-      "step": 35368
-    },
-    {
-      "epoch": 18.445371577574967,
-      "grad_norm": 1.4100139141082764,
-      "learning_rate": 6.496783919597991e-05,
-      "loss": 5.1328,
-      "step": 35369
-    },
-    {
-      "epoch": 18.445893089960887,
-      "grad_norm": 1.5236485004425049,
-      "learning_rate": 6.496683417085427e-05,
-      "loss": 4.778,
-      "step": 35370
-    },
-    {
-      "epoch": 18.446414602346806,
-      "grad_norm": 1.6898021697998047,
-      "learning_rate": 6.496582914572865e-05,
-      "loss": 4.5007,
-      "step": 35371
-    },
-    {
-      "epoch": 18.446936114732726,
-      "grad_norm": 1.4275513887405396,
-      "learning_rate": 6.496482412060301e-05,
-      "loss": 5.5937,
-      "step": 35372
-    },
-    {
-      "epoch": 18.447457627118645,
-      "grad_norm": 1.4258006811141968,
-      "learning_rate": 6.496381909547738e-05,
-      "loss": 5.6156,
-      "step": 35373
-    },
-    {
-      "epoch": 18.447979139504564,
-      "grad_norm": 1.8907052278518677,
-      "learning_rate": 6.496281407035176e-05,
-      "loss": 4.7683,
-      "step": 35374
-    },
-    {
-      "epoch": 18.448500651890484,
-      "grad_norm": 1.4136124849319458,
-      "learning_rate": 6.496180904522613e-05,
-      "loss": 5.559,
-      "step": 35375
-    },
-    {
-      "epoch": 18.4490221642764,
-      "grad_norm": 1.500968098640442,
-      "learning_rate": 6.496080402010051e-05,
-      "loss": 4.9055,
-      "step": 35376
-    },
-    {
-      "epoch": 18.44954367666232,
-      "grad_norm": 1.4806249141693115,
-      "learning_rate": 6.495979899497488e-05,
-      "loss": 5.5007,
-      "step": 35377
-    },
-    {
-      "epoch": 18.45006518904824,
-      "grad_norm": 1.4415949583053589,
-      "learning_rate": 6.495879396984925e-05,
-      "loss": 5.6106,
-      "step": 35378
-    },
-    {
-      "epoch": 18.45058670143416,
-      "grad_norm": 1.6371302604675293,
-      "learning_rate": 6.495778894472362e-05,
-      "loss": 5.0532,
-      "step": 35379
-    },
-    {
-      "epoch": 18.451108213820078,
-      "grad_norm": 1.4344274997711182,
-      "learning_rate": 6.4956783919598e-05,
-      "loss": 5.3266,
-      "step": 35380
-    },
-    {
-      "epoch": 18.451629726205997,
-      "grad_norm": 1.3941668272018433,
-      "learning_rate": 6.495577889447236e-05,
-      "loss": 5.7062,
-      "step": 35381
-    },
-    {
-      "epoch": 18.452151238591917,
-      "grad_norm": 1.5668227672576904,
-      "learning_rate": 6.495477386934674e-05,
-      "loss": 5.7404,
-      "step": 35382
-    },
-    {
-      "epoch": 18.452672750977836,
-      "grad_norm": 1.5487717390060425,
-      "learning_rate": 6.49537688442211e-05,
-      "loss": 5.2419,
-      "step": 35383
-    },
-    {
-      "epoch": 18.453194263363756,
-      "grad_norm": 1.4686191082000732,
-      "learning_rate": 6.495276381909548e-05,
-      "loss": 5.3603,
-      "step": 35384
-    },
-    {
-      "epoch": 18.453715775749675,
-      "grad_norm": 1.994951605796814,
-      "learning_rate": 6.495175879396986e-05,
-      "loss": 4.6264,
-      "step": 35385
-    },
-    {
-      "epoch": 18.454237288135594,
-      "grad_norm": 1.4223012924194336,
-      "learning_rate": 6.495075376884422e-05,
-      "loss": 5.7303,
-      "step": 35386
-    },
-    {
-      "epoch": 18.454758800521514,
-      "grad_norm": 1.5756138563156128,
-      "learning_rate": 6.49497487437186e-05,
-      "loss": 5.3568,
-      "step": 35387
-    },
-    {
-      "epoch": 18.45528031290743,
-      "grad_norm": 1.5661418437957764,
-      "learning_rate": 6.494874371859297e-05,
-      "loss": 5.1827,
-      "step": 35388
-    },
-    {
-      "epoch": 18.45580182529335,
-      "grad_norm": 1.4975249767303467,
-      "learning_rate": 6.494773869346734e-05,
-      "loss": 5.2714,
-      "step": 35389
-    },
-    {
-      "epoch": 18.45632333767927,
-      "grad_norm": 1.5480414628982544,
-      "learning_rate": 6.494673366834171e-05,
-      "loss": 5.1693,
-      "step": 35390
-    },
-    {
-      "epoch": 18.45684485006519,
-      "grad_norm": 1.3362300395965576,
-      "learning_rate": 6.494572864321609e-05,
-      "loss": 5.0241,
-      "step": 35391
-    },
-    {
-      "epoch": 18.457366362451108,
-      "grad_norm": 1.4789241552352905,
-      "learning_rate": 6.494472361809045e-05,
-      "loss": 5.3813,
-      "step": 35392
-    },
-    {
-      "epoch": 18.457887874837027,
-      "grad_norm": 1.4637984037399292,
-      "learning_rate": 6.494371859296483e-05,
-      "loss": 5.5245,
-      "step": 35393
-    },
-    {
-      "epoch": 18.458409387222947,
-      "grad_norm": 1.5049444437026978,
-      "learning_rate": 6.494271356783919e-05,
-      "loss": 5.3576,
-      "step": 35394
-    },
-    {
-      "epoch": 18.458930899608866,
-      "grad_norm": 1.4522918462753296,
-      "learning_rate": 6.494170854271357e-05,
-      "loss": 5.1529,
-      "step": 35395
-    },
-    {
-      "epoch": 18.459452411994786,
-      "grad_norm": 1.533366084098816,
-      "learning_rate": 6.494070351758795e-05,
-      "loss": 5.3989,
-      "step": 35396
-    },
-    {
-      "epoch": 18.459973924380705,
-      "grad_norm": 1.4364441633224487,
-      "learning_rate": 6.493969849246233e-05,
-      "loss": 5.5818,
-      "step": 35397
-    },
-    {
-      "epoch": 18.460495436766625,
-      "grad_norm": 1.4394077062606812,
-      "learning_rate": 6.493869346733669e-05,
-      "loss": 5.4892,
-      "step": 35398
-    },
-    {
-      "epoch": 18.461016949152544,
-      "grad_norm": 1.9129582643508911,
-      "learning_rate": 6.493768844221105e-05,
-      "loss": 4.8712,
-      "step": 35399
-    },
-    {
-      "epoch": 18.46153846153846,
-      "grad_norm": 1.5635312795639038,
-      "learning_rate": 6.493668341708543e-05,
-      "loss": 5.2926,
-      "step": 35400
-    },
-    {
-      "epoch": 18.46205997392438,
-      "grad_norm": 1.6109960079193115,
-      "learning_rate": 6.49356783919598e-05,
-      "loss": 5.3274,
-      "step": 35401
-    },
-    {
-      "epoch": 18.4625814863103,
-      "grad_norm": 1.443228006362915,
-      "learning_rate": 6.493467336683417e-05,
-      "loss": 5.3535,
-      "step": 35402
-    },
-    {
-      "epoch": 18.46310299869622,
-      "grad_norm": 1.5739573240280151,
-      "learning_rate": 6.493366834170854e-05,
-      "loss": 4.8543,
-      "step": 35403
-    },
-    {
-      "epoch": 18.463624511082138,
-      "grad_norm": 1.463592290878296,
-      "learning_rate": 6.493266331658292e-05,
-      "loss": 4.5751,
-      "step": 35404
-    },
-    {
-      "epoch": 18.464146023468057,
-      "grad_norm": 1.6104319095611572,
-      "learning_rate": 6.49316582914573e-05,
-      "loss": 5.4133,
-      "step": 35405
-    },
-    {
-      "epoch": 18.464667535853977,
-      "grad_norm": 1.4912549257278442,
-      "learning_rate": 6.493065326633167e-05,
-      "loss": 5.2075,
-      "step": 35406
-    },
-    {
-      "epoch": 18.465189048239896,
-      "grad_norm": 1.5088138580322266,
-      "learning_rate": 6.492964824120604e-05,
-      "loss": 5.5302,
-      "step": 35407
-    },
-    {
-      "epoch": 18.465710560625816,
-      "grad_norm": 1.4186854362487793,
-      "learning_rate": 6.492864321608041e-05,
-      "loss": 5.865,
-      "step": 35408
-    },
-    {
-      "epoch": 18.466232073011735,
-      "grad_norm": 1.5519641637802124,
-      "learning_rate": 6.492763819095478e-05,
-      "loss": 5.3448,
-      "step": 35409
-    },
-    {
-      "epoch": 18.466753585397655,
-      "grad_norm": 1.4350141286849976,
-      "learning_rate": 6.492663316582916e-05,
-      "loss": 5.0024,
-      "step": 35410
-    },
-    {
-      "epoch": 18.467275097783574,
-      "grad_norm": 1.6955196857452393,
-      "learning_rate": 6.492562814070352e-05,
-      "loss": 4.4941,
-      "step": 35411
-    },
-    {
-      "epoch": 18.46779661016949,
-      "grad_norm": 1.4555683135986328,
-      "learning_rate": 6.492462311557788e-05,
-      "loss": 5.4659,
-      "step": 35412
-    },
-    {
-      "epoch": 18.46831812255541,
-      "grad_norm": 1.5380946397781372,
-      "learning_rate": 6.492361809045226e-05,
-      "loss": 5.2456,
-      "step": 35413
-    },
-    {
-      "epoch": 18.46883963494133,
-      "grad_norm": 1.585091471672058,
-      "learning_rate": 6.492261306532663e-05,
-      "loss": 5.0918,
-      "step": 35414
-    },
-    {
-      "epoch": 18.46936114732725,
-      "grad_norm": 1.5110771656036377,
-      "learning_rate": 6.4921608040201e-05,
-      "loss": 5.6213,
-      "step": 35415
-    },
-    {
-      "epoch": 18.469882659713168,
-      "grad_norm": 1.6065860986709595,
-      "learning_rate": 6.492060301507538e-05,
-      "loss": 5.3905,
-      "step": 35416
-    },
-    {
-      "epoch": 18.470404172099087,
-      "grad_norm": 1.4187955856323242,
-      "learning_rate": 6.491959798994976e-05,
-      "loss": 5.5393,
-      "step": 35417
-    },
-    {
-      "epoch": 18.470925684485007,
-      "grad_norm": 1.46259605884552,
-      "learning_rate": 6.491859296482412e-05,
-      "loss": 5.5303,
-      "step": 35418
-    },
-    {
-      "epoch": 18.471447196870926,
-      "grad_norm": 1.4321526288986206,
-      "learning_rate": 6.49175879396985e-05,
-      "loss": 5.4976,
-      "step": 35419
-    },
-    {
-      "epoch": 18.471968709256846,
-      "grad_norm": 1.3810888528823853,
-      "learning_rate": 6.491658291457287e-05,
-      "loss": 5.3922,
-      "step": 35420
-    },
-    {
-      "epoch": 18.472490221642765,
-      "grad_norm": 1.5117462873458862,
-      "learning_rate": 6.491557788944724e-05,
-      "loss": 5.2183,
-      "step": 35421
-    },
-    {
-      "epoch": 18.473011734028685,
-      "grad_norm": 1.389697551727295,
-      "learning_rate": 6.491457286432161e-05,
-      "loss": 5.5407,
-      "step": 35422
-    },
-    {
-      "epoch": 18.473533246414604,
-      "grad_norm": 1.342136263847351,
-      "learning_rate": 6.491356783919599e-05,
-      "loss": 5.8574,
-      "step": 35423
-    },
-    {
-      "epoch": 18.47405475880052,
-      "grad_norm": 1.5424485206604004,
-      "learning_rate": 6.491256281407035e-05,
-      "loss": 5.1128,
-      "step": 35424
-    },
-    {
-      "epoch": 18.47457627118644,
-      "grad_norm": 1.5053235292434692,
-      "learning_rate": 6.491155778894471e-05,
-      "loss": 5.6577,
-      "step": 35425
-    },
-    {
-      "epoch": 18.47509778357236,
-      "grad_norm": 1.4760935306549072,
-      "learning_rate": 6.491055276381909e-05,
-      "loss": 5.5241,
-      "step": 35426
-    },
-    {
-      "epoch": 18.47561929595828,
-      "grad_norm": 1.5390331745147705,
-      "learning_rate": 6.490954773869347e-05,
-      "loss": 5.3238,
-      "step": 35427
-    },
-    {
-      "epoch": 18.476140808344198,
-      "grad_norm": 1.4704557657241821,
-      "learning_rate": 6.490854271356785e-05,
-      "loss": 5.1908,
-      "step": 35428
-    },
-    {
-      "epoch": 18.476662320730117,
-      "grad_norm": 1.4879099130630493,
-      "learning_rate": 6.490753768844221e-05,
-      "loss": 5.4092,
-      "step": 35429
-    },
-    {
-      "epoch": 18.477183833116037,
-      "grad_norm": 1.4675546884536743,
-      "learning_rate": 6.490653266331659e-05,
-      "loss": 5.7129,
-      "step": 35430
-    },
-    {
-      "epoch": 18.477705345501956,
-      "grad_norm": 1.493099570274353,
-      "learning_rate": 6.490552763819095e-05,
-      "loss": 5.5864,
-      "step": 35431
-    },
-    {
-      "epoch": 18.478226857887876,
-      "grad_norm": 1.5453606843948364,
-      "learning_rate": 6.490452261306533e-05,
-      "loss": 5.1499,
-      "step": 35432
-    },
-    {
-      "epoch": 18.478748370273795,
-      "grad_norm": 1.5509778261184692,
-      "learning_rate": 6.49035175879397e-05,
-      "loss": 4.963,
-      "step": 35433
-    },
-    {
-      "epoch": 18.479269882659715,
-      "grad_norm": 1.5753614902496338,
-      "learning_rate": 6.490251256281407e-05,
-      "loss": 5.4965,
-      "step": 35434
-    },
-    {
-      "epoch": 18.479791395045634,
-      "grad_norm": 1.4161410331726074,
-      "learning_rate": 6.490150753768844e-05,
-      "loss": 5.7232,
-      "step": 35435
-    },
-    {
-      "epoch": 18.48031290743155,
-      "grad_norm": 1.5373061895370483,
-      "learning_rate": 6.490050251256282e-05,
-      "loss": 5.1043,
-      "step": 35436
-    },
-    {
-      "epoch": 18.48083441981747,
-      "grad_norm": 1.5283647775650024,
-      "learning_rate": 6.48994974874372e-05,
-      "loss": 5.5081,
-      "step": 35437
-    },
-    {
-      "epoch": 18.48135593220339,
-      "grad_norm": 1.6201008558273315,
-      "learning_rate": 6.489849246231157e-05,
-      "loss": 5.0912,
-      "step": 35438
-    },
-    {
-      "epoch": 18.48187744458931,
-      "grad_norm": 1.4762250185012817,
-      "learning_rate": 6.489748743718594e-05,
-      "loss": 5.3188,
-      "step": 35439
-    },
-    {
-      "epoch": 18.482398956975228,
-      "grad_norm": 1.3981010913848877,
-      "learning_rate": 6.48964824120603e-05,
-      "loss": 5.4128,
-      "step": 35440
-    },
-    {
-      "epoch": 18.482920469361147,
-      "grad_norm": 1.4846478700637817,
-      "learning_rate": 6.489547738693468e-05,
-      "loss": 5.5662,
-      "step": 35441
-    },
-    {
-      "epoch": 18.483441981747067,
-      "grad_norm": 1.5230200290679932,
-      "learning_rate": 6.489447236180904e-05,
-      "loss": 5.1054,
-      "step": 35442
-    },
-    {
-      "epoch": 18.483963494132986,
-      "grad_norm": 1.551759123802185,
-      "learning_rate": 6.489346733668342e-05,
-      "loss": 5.0776,
-      "step": 35443
-    },
-    {
-      "epoch": 18.484485006518906,
-      "grad_norm": 1.5195060968399048,
-      "learning_rate": 6.489246231155778e-05,
-      "loss": 5.6296,
-      "step": 35444
-    },
-    {
-      "epoch": 18.485006518904825,
-      "grad_norm": 1.5474330186843872,
-      "learning_rate": 6.489145728643216e-05,
-      "loss": 5.4563,
-      "step": 35445
-    },
-    {
-      "epoch": 18.485528031290745,
-      "grad_norm": 1.5079458951950073,
-      "learning_rate": 6.489045226130653e-05,
-      "loss": 5.4901,
-      "step": 35446
-    },
-    {
-      "epoch": 18.486049543676664,
-      "grad_norm": 1.5125809907913208,
-      "learning_rate": 6.48894472361809e-05,
-      "loss": 5.3729,
-      "step": 35447
-    },
-    {
-      "epoch": 18.48657105606258,
-      "grad_norm": 1.5129650831222534,
-      "learning_rate": 6.488844221105528e-05,
-      "loss": 5.0258,
-      "step": 35448
-    },
-    {
-      "epoch": 18.4870925684485,
-      "grad_norm": 1.4959920644760132,
-      "learning_rate": 6.488743718592966e-05,
-      "loss": 5.571,
-      "step": 35449
-    },
-    {
-      "epoch": 18.48761408083442,
-      "grad_norm": 1.5242972373962402,
-      "learning_rate": 6.488643216080402e-05,
-      "loss": 4.815,
-      "step": 35450
-    },
-    {
-      "epoch": 18.48813559322034,
-      "grad_norm": 1.495603084564209,
-      "learning_rate": 6.48854271356784e-05,
-      "loss": 5.4233,
-      "step": 35451
-    },
-    {
-      "epoch": 18.488657105606258,
-      "grad_norm": 1.5951530933380127,
-      "learning_rate": 6.488442211055277e-05,
-      "loss": 5.4481,
-      "step": 35452
-    },
-    {
-      "epoch": 18.489178617992177,
-      "grad_norm": 1.5429401397705078,
-      "learning_rate": 6.488341708542713e-05,
-      "loss": 5.0311,
-      "step": 35453
-    },
-    {
-      "epoch": 18.489700130378097,
-      "grad_norm": 1.567263126373291,
-      "learning_rate": 6.488241206030151e-05,
-      "loss": 5.3736,
-      "step": 35454
-    },
-    {
-      "epoch": 18.490221642764016,
-      "grad_norm": 1.5790342092514038,
-      "learning_rate": 6.488140703517587e-05,
-      "loss": 5.2556,
-      "step": 35455
-    },
-    {
-      "epoch": 18.490743155149936,
-      "grad_norm": 1.5973724126815796,
-      "learning_rate": 6.488040201005025e-05,
-      "loss": 5.0487,
-      "step": 35456
-    },
-    {
-      "epoch": 18.491264667535855,
-      "grad_norm": 1.541680097579956,
-      "learning_rate": 6.487939698492463e-05,
-      "loss": 5.0512,
-      "step": 35457
-    },
-    {
-      "epoch": 18.491786179921775,
-      "grad_norm": 1.3658303022384644,
-      "learning_rate": 6.4878391959799e-05,
-      "loss": 5.7681,
-      "step": 35458
-    },
-    {
-      "epoch": 18.49230769230769,
-      "grad_norm": 1.540747046470642,
-      "learning_rate": 6.487738693467337e-05,
-      "loss": 5.7482,
-      "step": 35459
-    },
-    {
-      "epoch": 18.49282920469361,
-      "grad_norm": 1.6963053941726685,
-      "learning_rate": 6.487638190954775e-05,
-      "loss": 4.9647,
-      "step": 35460
-    },
-    {
-      "epoch": 18.49335071707953,
-      "grad_norm": 1.4181220531463623,
-      "learning_rate": 6.487537688442211e-05,
-      "loss": 5.2612,
-      "step": 35461
-    },
-    {
-      "epoch": 18.49387222946545,
-      "grad_norm": 1.4932340383529663,
-      "learning_rate": 6.487437185929649e-05,
-      "loss": 5.025,
-      "step": 35462
-    },
-    {
-      "epoch": 18.49439374185137,
-      "grad_norm": 1.4796345233917236,
-      "learning_rate": 6.487336683417086e-05,
-      "loss": 5.3683,
-      "step": 35463
-    },
-    {
-      "epoch": 18.494915254237288,
-      "grad_norm": 1.3907248973846436,
-      "learning_rate": 6.487236180904523e-05,
-      "loss": 5.8619,
-      "step": 35464
-    },
-    {
-      "epoch": 18.495436766623207,
-      "grad_norm": 1.4548245668411255,
-      "learning_rate": 6.48713567839196e-05,
-      "loss": 5.1801,
-      "step": 35465
-    },
-    {
-      "epoch": 18.495958279009127,
-      "grad_norm": 1.5921149253845215,
-      "learning_rate": 6.487035175879396e-05,
-      "loss": 5.0937,
-      "step": 35466
-    },
-    {
-      "epoch": 18.496479791395046,
-      "grad_norm": 1.5150690078735352,
-      "learning_rate": 6.486934673366834e-05,
-      "loss": 5.1891,
-      "step": 35467
-    },
-    {
-      "epoch": 18.497001303780966,
-      "grad_norm": 1.598847508430481,
-      "learning_rate": 6.486834170854272e-05,
-      "loss": 5.609,
-      "step": 35468
-    },
-    {
-      "epoch": 18.497522816166885,
-      "grad_norm": 1.4594804048538208,
-      "learning_rate": 6.48673366834171e-05,
-      "loss": 5.1867,
-      "step": 35469
-    },
-    {
-      "epoch": 18.498044328552805,
-      "grad_norm": 1.4736852645874023,
-      "learning_rate": 6.486633165829146e-05,
-      "loss": 5.5875,
-      "step": 35470
-    },
-    {
-      "epoch": 18.49856584093872,
-      "grad_norm": 1.5286160707473755,
-      "learning_rate": 6.486532663316584e-05,
-      "loss": 5.3453,
-      "step": 35471
-    },
-    {
-      "epoch": 18.49908735332464,
-      "grad_norm": 1.4607089757919312,
-      "learning_rate": 6.48643216080402e-05,
-      "loss": 5.0675,
-      "step": 35472
-    },
-    {
-      "epoch": 18.49960886571056,
-      "grad_norm": 1.4371898174285889,
-      "learning_rate": 6.486331658291458e-05,
-      "loss": 5.5728,
-      "step": 35473
-    },
-    {
-      "epoch": 18.50013037809648,
-      "grad_norm": 1.5173791646957397,
-      "learning_rate": 6.486231155778894e-05,
-      "loss": 5.0847,
-      "step": 35474
-    },
-    {
-      "epoch": 18.5006518904824,
-      "grad_norm": 1.483396053314209,
-      "learning_rate": 6.486130653266332e-05,
-      "loss": 5.6492,
-      "step": 35475
-    },
-    {
-      "epoch": 18.501173402868318,
-      "grad_norm": 1.5216718912124634,
-      "learning_rate": 6.486030150753769e-05,
-      "loss": 4.988,
-      "step": 35476
-    },
-    {
-      "epoch": 18.501694915254237,
-      "grad_norm": 1.6171027421951294,
-      "learning_rate": 6.485929648241206e-05,
-      "loss": 5.0624,
-      "step": 35477
-    },
-    {
-      "epoch": 18.502216427640157,
-      "grad_norm": 1.410894751548767,
-      "learning_rate": 6.485829145728644e-05,
-      "loss": 5.6148,
-      "step": 35478
-    },
-    {
-      "epoch": 18.502737940026076,
-      "grad_norm": 1.5431923866271973,
-      "learning_rate": 6.48572864321608e-05,
-      "loss": 5.5598,
-      "step": 35479
-    },
-    {
-      "epoch": 18.503259452411996,
-      "grad_norm": 1.459151029586792,
-      "learning_rate": 6.485628140703518e-05,
-      "loss": 5.0877,
-      "step": 35480
-    },
-    {
-      "epoch": 18.503780964797915,
-      "grad_norm": 1.4986822605133057,
-      "learning_rate": 6.485527638190955e-05,
-      "loss": 5.5132,
-      "step": 35481
-    },
-    {
-      "epoch": 18.504302477183835,
-      "grad_norm": 1.4528541564941406,
-      "learning_rate": 6.485427135678393e-05,
-      "loss": 5.1542,
-      "step": 35482
-    },
-    {
-      "epoch": 18.50482398956975,
-      "grad_norm": 1.4425474405288696,
-      "learning_rate": 6.485326633165829e-05,
-      "loss": 4.9311,
-      "step": 35483
-    },
-    {
-      "epoch": 18.50534550195567,
-      "grad_norm": 1.5117945671081543,
-      "learning_rate": 6.485226130653267e-05,
-      "loss": 4.7707,
-      "step": 35484
-    },
-    {
-      "epoch": 18.50586701434159,
-      "grad_norm": 1.47257661819458,
-      "learning_rate": 6.485125628140703e-05,
-      "loss": 5.7236,
-      "step": 35485
-    },
-    {
-      "epoch": 18.50638852672751,
-      "grad_norm": 1.4577170610427856,
-      "learning_rate": 6.485025125628141e-05,
-      "loss": 5.7277,
-      "step": 35486
-    },
-    {
-      "epoch": 18.50691003911343,
-      "grad_norm": 1.464302659034729,
-      "learning_rate": 6.484924623115577e-05,
-      "loss": 5.2901,
-      "step": 35487
-    },
-    {
-      "epoch": 18.507431551499348,
-      "grad_norm": 1.4555132389068604,
-      "learning_rate": 6.484824120603015e-05,
-      "loss": 5.3892,
-      "step": 35488
-    },
-    {
-      "epoch": 18.507953063885267,
-      "grad_norm": 1.491204023361206,
-      "learning_rate": 6.484723618090453e-05,
-      "loss": 5.3865,
-      "step": 35489
-    },
-    {
-      "epoch": 18.508474576271187,
-      "grad_norm": 1.5877994298934937,
-      "learning_rate": 6.484623115577891e-05,
-      "loss": 4.993,
-      "step": 35490
-    },
-    {
-      "epoch": 18.508996088657106,
-      "grad_norm": 1.4624348878860474,
-      "learning_rate": 6.484522613065327e-05,
-      "loss": 5.1995,
-      "step": 35491
-    },
-    {
-      "epoch": 18.509517601043026,
-      "grad_norm": 1.446710228919983,
-      "learning_rate": 6.484422110552764e-05,
-      "loss": 5.2028,
-      "step": 35492
-    },
-    {
-      "epoch": 18.510039113428945,
-      "grad_norm": 1.626200795173645,
-      "learning_rate": 6.484321608040201e-05,
-      "loss": 5.1737,
-      "step": 35493
-    },
-    {
-      "epoch": 18.510560625814865,
-      "grad_norm": 1.464115858078003,
-      "learning_rate": 6.484221105527638e-05,
-      "loss": 4.9925,
-      "step": 35494
-    },
-    {
-      "epoch": 18.51108213820078,
-      "grad_norm": 1.4619052410125732,
-      "learning_rate": 6.484120603015076e-05,
-      "loss": 5.4528,
-      "step": 35495
-    },
-    {
-      "epoch": 18.5116036505867,
-      "grad_norm": 1.5576401948928833,
-      "learning_rate": 6.484020100502512e-05,
-      "loss": 5.1253,
-      "step": 35496
-    },
-    {
-      "epoch": 18.51212516297262,
-      "grad_norm": 1.4755072593688965,
-      "learning_rate": 6.48391959798995e-05,
-      "loss": 5.503,
-      "step": 35497
-    },
-    {
-      "epoch": 18.51264667535854,
-      "grad_norm": 1.4425896406173706,
-      "learning_rate": 6.483819095477388e-05,
-      "loss": 5.631,
-      "step": 35498
-    },
-    {
-      "epoch": 18.51316818774446,
-      "grad_norm": 1.4946441650390625,
-      "learning_rate": 6.483718592964825e-05,
-      "loss": 5.0423,
-      "step": 35499
-    },
-    {
-      "epoch": 18.513689700130378,
-      "grad_norm": 1.4918228387832642,
-      "learning_rate": 6.483618090452262e-05,
-      "loss": 5.131,
-      "step": 35500
-    },
-    {
-      "epoch": 18.514211212516297,
-      "grad_norm": 1.6058791875839233,
-      "learning_rate": 6.4835175879397e-05,
-      "loss": 5.1695,
-      "step": 35501
-    },
-    {
-      "epoch": 18.514732724902217,
-      "grad_norm": 1.5280064344406128,
-      "learning_rate": 6.483417085427136e-05,
-      "loss": 5.6078,
-      "step": 35502
-    },
-    {
-      "epoch": 18.515254237288136,
-      "grad_norm": 1.5316718816757202,
-      "learning_rate": 6.483316582914574e-05,
-      "loss": 5.4516,
-      "step": 35503
-    },
-    {
-      "epoch": 18.515775749674056,
-      "grad_norm": 1.6612154245376587,
-      "learning_rate": 6.48321608040201e-05,
-      "loss": 5.0401,
-      "step": 35504
-    },
-    {
-      "epoch": 18.516297262059975,
-      "grad_norm": 1.480027675628662,
-      "learning_rate": 6.483115577889447e-05,
-      "loss": 5.1944,
-      "step": 35505
-    },
-    {
-      "epoch": 18.516818774445895,
-      "grad_norm": 1.4312092065811157,
-      "learning_rate": 6.483015075376884e-05,
-      "loss": 5.1764,
-      "step": 35506
-    },
-    {
-      "epoch": 18.51734028683181,
-      "grad_norm": 1.3732590675354004,
-      "learning_rate": 6.482914572864321e-05,
-      "loss": 5.5659,
-      "step": 35507
-    },
-    {
-      "epoch": 18.51786179921773,
-      "grad_norm": 1.6740384101867676,
-      "learning_rate": 6.482814070351759e-05,
-      "loss": 5.1727,
-      "step": 35508
-    },
-    {
-      "epoch": 18.51838331160365,
-      "grad_norm": 1.5454546213150024,
-      "learning_rate": 6.482713567839196e-05,
-      "loss": 5.1987,
-      "step": 35509
-    },
-    {
-      "epoch": 18.51890482398957,
-      "grad_norm": 1.507055640220642,
-      "learning_rate": 6.482613065326634e-05,
-      "loss": 5.4581,
-      "step": 35510
-    },
-    {
-      "epoch": 18.51942633637549,
-      "grad_norm": 1.3497278690338135,
-      "learning_rate": 6.48251256281407e-05,
-      "loss": 5.5685,
-      "step": 35511
-    },
-    {
-      "epoch": 18.519947848761408,
-      "grad_norm": 1.5519951581954956,
-      "learning_rate": 6.482412060301508e-05,
-      "loss": 4.9936,
-      "step": 35512
-    },
-    {
-      "epoch": 18.520469361147327,
-      "grad_norm": 1.4718763828277588,
-      "learning_rate": 6.482311557788945e-05,
-      "loss": 5.5069,
-      "step": 35513
-    },
-    {
-      "epoch": 18.520990873533247,
-      "grad_norm": 1.5588761568069458,
-      "learning_rate": 6.482211055276383e-05,
-      "loss": 4.9248,
-      "step": 35514
-    },
-    {
-      "epoch": 18.521512385919166,
-      "grad_norm": 1.436373233795166,
-      "learning_rate": 6.482110552763819e-05,
-      "loss": 5.3652,
-      "step": 35515
-    },
-    {
-      "epoch": 18.522033898305086,
-      "grad_norm": 1.5915199518203735,
-      "learning_rate": 6.482010050251257e-05,
-      "loss": 5.4193,
-      "step": 35516
-    },
-    {
-      "epoch": 18.522555410691005,
-      "grad_norm": 1.5224963426589966,
-      "learning_rate": 6.481909547738693e-05,
-      "loss": 5.2056,
-      "step": 35517
-    },
-    {
-      "epoch": 18.523076923076925,
-      "grad_norm": 1.5738803148269653,
-      "learning_rate": 6.481809045226131e-05,
-      "loss": 5.0447,
-      "step": 35518
-    },
-    {
-      "epoch": 18.52359843546284,
-      "grad_norm": 1.4802074432373047,
-      "learning_rate": 6.481708542713569e-05,
-      "loss": 5.4094,
-      "step": 35519
-    },
-    {
-      "epoch": 18.52411994784876,
-      "grad_norm": 1.5555838346481323,
-      "learning_rate": 6.481608040201005e-05,
-      "loss": 5.3868,
-      "step": 35520
-    },
-    {
-      "epoch": 18.52464146023468,
-      "grad_norm": 1.4200795888900757,
-      "learning_rate": 6.481507537688443e-05,
-      "loss": 5.8335,
-      "step": 35521
-    },
-    {
-      "epoch": 18.5251629726206,
-      "grad_norm": 1.4517983198165894,
-      "learning_rate": 6.48140703517588e-05,
-      "loss": 5.2918,
-      "step": 35522
-    },
-    {
-      "epoch": 18.52568448500652,
-      "grad_norm": 1.5185976028442383,
-      "learning_rate": 6.481306532663317e-05,
-      "loss": 5.2263,
-      "step": 35523
-    },
-    {
-      "epoch": 18.526205997392438,
-      "grad_norm": 1.382768154144287,
-      "learning_rate": 6.481206030150754e-05,
-      "loss": 5.0499,
-      "step": 35524
-    },
-    {
-      "epoch": 18.526727509778357,
-      "grad_norm": 1.565409779548645,
-      "learning_rate": 6.481105527638191e-05,
-      "loss": 4.9161,
-      "step": 35525
-    },
-    {
-      "epoch": 18.527249022164277,
-      "grad_norm": 1.5516471862792969,
-      "learning_rate": 6.481005025125628e-05,
-      "loss": 5.7253,
-      "step": 35526
-    },
-    {
-      "epoch": 18.527770534550196,
-      "grad_norm": 1.5027854442596436,
-      "learning_rate": 6.480904522613066e-05,
-      "loss": 5.4209,
-      "step": 35527
-    },
-    {
-      "epoch": 18.528292046936116,
-      "grad_norm": 1.4789053201675415,
-      "learning_rate": 6.480804020100502e-05,
-      "loss": 5.5322,
-      "step": 35528
-    },
-    {
-      "epoch": 18.528813559322035,
-      "grad_norm": 1.4199090003967285,
-      "learning_rate": 6.48070351758794e-05,
-      "loss": 5.1008,
-      "step": 35529
-    },
-    {
-      "epoch": 18.529335071707955,
-      "grad_norm": 1.4920481443405151,
-      "learning_rate": 6.480603015075378e-05,
-      "loss": 5.5376,
-      "step": 35530
-    },
-    {
-      "epoch": 18.52985658409387,
-      "grad_norm": 1.4184472560882568,
-      "learning_rate": 6.480502512562815e-05,
-      "loss": 5.4905,
-      "step": 35531
-    },
-    {
-      "epoch": 18.53037809647979,
-      "grad_norm": 1.5036133527755737,
-      "learning_rate": 6.480402010050252e-05,
-      "loss": 5.4261,
-      "step": 35532
-    },
-    {
-      "epoch": 18.53089960886571,
-      "grad_norm": 1.5692049264907837,
-      "learning_rate": 6.480301507537688e-05,
-      "loss": 5.4678,
-      "step": 35533
-    },
-    {
-      "epoch": 18.53142112125163,
-      "grad_norm": 1.4867217540740967,
-      "learning_rate": 6.480201005025126e-05,
-      "loss": 5.2224,
-      "step": 35534
-    },
-    {
-      "epoch": 18.53194263363755,
-      "grad_norm": 1.4854336977005005,
-      "learning_rate": 6.480100502512563e-05,
-      "loss": 5.421,
-      "step": 35535
-    },
-    {
-      "epoch": 18.532464146023468,
-      "grad_norm": 1.4389443397521973,
-      "learning_rate": 6.48e-05,
-      "loss": 5.6175,
-      "step": 35536
-    },
-    {
-      "epoch": 18.532985658409387,
-      "grad_norm": 1.4565304517745972,
-      "learning_rate": 6.479899497487437e-05,
-      "loss": 5.0368,
-      "step": 35537
-    },
-    {
-      "epoch": 18.533507170795307,
-      "grad_norm": 1.6402233839035034,
-      "learning_rate": 6.479798994974875e-05,
-      "loss": 4.7655,
-      "step": 35538
-    },
-    {
-      "epoch": 18.534028683181226,
-      "grad_norm": 1.5245435237884521,
-      "learning_rate": 6.479698492462312e-05,
-      "loss": 5.4058,
-      "step": 35539
-    },
-    {
-      "epoch": 18.534550195567146,
-      "grad_norm": 1.437264084815979,
-      "learning_rate": 6.47959798994975e-05,
-      "loss": 5.2931,
-      "step": 35540
-    },
-    {
-      "epoch": 18.535071707953065,
-      "grad_norm": 1.5119725465774536,
-      "learning_rate": 6.479497487437187e-05,
-      "loss": 5.4938,
-      "step": 35541
-    },
-    {
-      "epoch": 18.53559322033898,
-      "grad_norm": 1.4952521324157715,
-      "learning_rate": 6.479396984924624e-05,
-      "loss": 5.4068,
-      "step": 35542
-    },
-    {
-      "epoch": 18.5361147327249,
-      "grad_norm": 1.504055380821228,
-      "learning_rate": 6.479296482412061e-05,
-      "loss": 5.4531,
-      "step": 35543
-    },
-    {
-      "epoch": 18.53663624511082,
-      "grad_norm": 1.534546136856079,
-      "learning_rate": 6.479195979899498e-05,
-      "loss": 5.3904,
-      "step": 35544
-    },
-    {
-      "epoch": 18.53715775749674,
-      "grad_norm": 1.5656923055648804,
-      "learning_rate": 6.479095477386935e-05,
-      "loss": 5.0032,
-      "step": 35545
-    },
-    {
-      "epoch": 18.53767926988266,
-      "grad_norm": 1.4622232913970947,
-      "learning_rate": 6.478994974874371e-05,
-      "loss": 5.5152,
-      "step": 35546
-    },
-    {
-      "epoch": 18.53820078226858,
-      "grad_norm": 1.573738932609558,
-      "learning_rate": 6.478894472361809e-05,
-      "loss": 5.2028,
-      "step": 35547
-    },
-    {
-      "epoch": 18.538722294654498,
-      "grad_norm": 1.4976190328598022,
-      "learning_rate": 6.478793969849246e-05,
-      "loss": 5.4686,
-      "step": 35548
-    },
-    {
-      "epoch": 18.539243807040418,
-      "grad_norm": 1.4427175521850586,
-      "learning_rate": 6.478693467336683e-05,
-      "loss": 5.4573,
-      "step": 35549
-    },
-    {
-      "epoch": 18.539765319426337,
-      "grad_norm": 1.4563417434692383,
-      "learning_rate": 6.478592964824121e-05,
-      "loss": 5.3768,
-      "step": 35550
-    },
-    {
-      "epoch": 18.540286831812256,
-      "grad_norm": 1.4356193542480469,
-      "learning_rate": 6.478492462311559e-05,
-      "loss": 5.4663,
-      "step": 35551
-    },
-    {
-      "epoch": 18.540808344198176,
-      "grad_norm": 1.5954298973083496,
-      "learning_rate": 6.478391959798995e-05,
-      "loss": 4.8864,
-      "step": 35552
-    },
-    {
-      "epoch": 18.541329856584095,
-      "grad_norm": 1.5576984882354736,
-      "learning_rate": 6.478291457286433e-05,
-      "loss": 5.1557,
-      "step": 35553
-    },
-    {
-      "epoch": 18.541851368970015,
-      "grad_norm": 1.4916774034500122,
-      "learning_rate": 6.47819095477387e-05,
-      "loss": 5.4654,
-      "step": 35554
-    },
-    {
-      "epoch": 18.54237288135593,
-      "grad_norm": 1.4641087055206299,
-      "learning_rate": 6.478090452261307e-05,
-      "loss": 5.2205,
-      "step": 35555
-    },
-    {
-      "epoch": 18.54289439374185,
-      "grad_norm": 1.5349711179733276,
-      "learning_rate": 6.477989949748744e-05,
-      "loss": 4.9694,
-      "step": 35556
-    },
-    {
-      "epoch": 18.54341590612777,
-      "grad_norm": 1.3819282054901123,
-      "learning_rate": 6.477889447236182e-05,
-      "loss": 5.0068,
-      "step": 35557
-    },
-    {
-      "epoch": 18.54393741851369,
-      "grad_norm": 1.5060696601867676,
-      "learning_rate": 6.477788944723618e-05,
-      "loss": 5.4714,
-      "step": 35558
-    },
-    {
-      "epoch": 18.54445893089961,
-      "grad_norm": 1.713645100593567,
-      "learning_rate": 6.477688442211056e-05,
-      "loss": 5.1647,
-      "step": 35559
-    },
-    {
-      "epoch": 18.544980443285528,
-      "grad_norm": 1.5000172853469849,
-      "learning_rate": 6.477587939698494e-05,
-      "loss": 5.2198,
-      "step": 35560
-    },
-    {
-      "epoch": 18.545501955671448,
-      "grad_norm": 1.512481451034546,
-      "learning_rate": 6.47748743718593e-05,
-      "loss": 4.9982,
-      "step": 35561
-    },
-    {
-      "epoch": 18.546023468057367,
-      "grad_norm": 1.477190613746643,
-      "learning_rate": 6.477386934673368e-05,
-      "loss": 5.4669,
-      "step": 35562
-    },
-    {
-      "epoch": 18.546544980443286,
-      "grad_norm": 1.4803235530853271,
-      "learning_rate": 6.477286432160804e-05,
-      "loss": 5.3387,
-      "step": 35563
-    },
-    {
-      "epoch": 18.547066492829206,
-      "grad_norm": 1.485863208770752,
-      "learning_rate": 6.477185929648242e-05,
-      "loss": 4.661,
-      "step": 35564
-    },
-    {
-      "epoch": 18.547588005215125,
-      "grad_norm": 1.606426477432251,
-      "learning_rate": 6.477085427135678e-05,
-      "loss": 4.9948,
-      "step": 35565
-    },
-    {
-      "epoch": 18.54810951760104,
-      "grad_norm": 1.438741683959961,
-      "learning_rate": 6.476984924623116e-05,
-      "loss": 5.6791,
-      "step": 35566
-    },
-    {
-      "epoch": 18.54863102998696,
-      "grad_norm": 1.409909725189209,
-      "learning_rate": 6.476884422110553e-05,
-      "loss": 5.4527,
-      "step": 35567
-    },
-    {
-      "epoch": 18.54915254237288,
-      "grad_norm": 1.4105454683303833,
-      "learning_rate": 6.47678391959799e-05,
-      "loss": 5.5576,
-      "step": 35568
-    },
-    {
-      "epoch": 18.5496740547588,
-      "grad_norm": 1.404860496520996,
-      "learning_rate": 6.476683417085427e-05,
-      "loss": 5.0654,
-      "step": 35569
-    },
-    {
-      "epoch": 18.55019556714472,
-      "grad_norm": 1.5405018329620361,
-      "learning_rate": 6.476582914572865e-05,
-      "loss": 5.0055,
-      "step": 35570
-    },
-    {
-      "epoch": 18.55071707953064,
-      "grad_norm": 1.5294269323349,
-      "learning_rate": 6.476482412060302e-05,
-      "loss": 5.3264,
-      "step": 35571
-    },
-    {
-      "epoch": 18.551238591916558,
-      "grad_norm": 1.4796496629714966,
-      "learning_rate": 6.476381909547739e-05,
-      "loss": 5.1241,
-      "step": 35572
-    },
-    {
-      "epoch": 18.551760104302478,
-      "grad_norm": 1.5621120929718018,
-      "learning_rate": 6.476281407035177e-05,
-      "loss": 5.2591,
-      "step": 35573
-    },
-    {
-      "epoch": 18.552281616688397,
-      "grad_norm": 1.4804742336273193,
-      "learning_rate": 6.476180904522613e-05,
-      "loss": 4.9445,
-      "step": 35574
-    },
-    {
-      "epoch": 18.552803129074317,
-      "grad_norm": 1.321298599243164,
-      "learning_rate": 6.476080402010051e-05,
-      "loss": 5.2601,
-      "step": 35575
-    },
-    {
-      "epoch": 18.553324641460236,
-      "grad_norm": 1.4469120502471924,
-      "learning_rate": 6.475979899497487e-05,
-      "loss": 5.4447,
-      "step": 35576
-    },
-    {
-      "epoch": 18.553846153846155,
-      "grad_norm": 1.4912959337234497,
-      "learning_rate": 6.475879396984925e-05,
-      "loss": 4.9665,
-      "step": 35577
-    },
-    {
-      "epoch": 18.55436766623207,
-      "grad_norm": 1.4086633920669556,
-      "learning_rate": 6.475778894472361e-05,
-      "loss": 5.6664,
-      "step": 35578
-    },
-    {
-      "epoch": 18.55488917861799,
-      "grad_norm": 1.5440562963485718,
-      "learning_rate": 6.475678391959799e-05,
-      "loss": 5.1566,
-      "step": 35579
-    },
-    {
-      "epoch": 18.55541069100391,
-      "grad_norm": 1.43160879611969,
-      "learning_rate": 6.475577889447236e-05,
-      "loss": 4.9944,
-      "step": 35580
-    },
-    {
-      "epoch": 18.55593220338983,
-      "grad_norm": 1.49031662940979,
-      "learning_rate": 6.475477386934673e-05,
-      "loss": 5.2582,
-      "step": 35581
-    },
-    {
-      "epoch": 18.55645371577575,
-      "grad_norm": 1.5763808488845825,
-      "learning_rate": 6.475376884422111e-05,
-      "loss": 5.6148,
-      "step": 35582
-    },
-    {
-      "epoch": 18.55697522816167,
-      "grad_norm": 1.4925205707550049,
-      "learning_rate": 6.475276381909549e-05,
-      "loss": 5.0341,
-      "step": 35583
-    },
-    {
-      "epoch": 18.557496740547588,
-      "grad_norm": 1.408478021621704,
-      "learning_rate": 6.475175879396985e-05,
-      "loss": 5.6439,
-      "step": 35584
-    },
-    {
-      "epoch": 18.558018252933508,
-      "grad_norm": 1.4134036302566528,
-      "learning_rate": 6.475075376884422e-05,
-      "loss": 5.4019,
-      "step": 35585
-    },
-    {
-      "epoch": 18.558539765319427,
-      "grad_norm": 1.4603469371795654,
-      "learning_rate": 6.47497487437186e-05,
-      "loss": 5.2438,
-      "step": 35586
-    },
-    {
-      "epoch": 18.559061277705347,
-      "grad_norm": 1.5013766288757324,
-      "learning_rate": 6.474874371859296e-05,
-      "loss": 5.1946,
-      "step": 35587
-    },
-    {
-      "epoch": 18.559582790091266,
-      "grad_norm": 1.4313961267471313,
-      "learning_rate": 6.474773869346734e-05,
-      "loss": 5.1489,
-      "step": 35588
-    },
-    {
-      "epoch": 18.560104302477185,
-      "grad_norm": 1.6062086820602417,
-      "learning_rate": 6.47467336683417e-05,
-      "loss": 5.0221,
-      "step": 35589
-    },
-    {
-      "epoch": 18.5606258148631,
-      "grad_norm": 1.4948956966400146,
-      "learning_rate": 6.474572864321608e-05,
-      "loss": 5.3412,
-      "step": 35590
-    },
-    {
-      "epoch": 18.56114732724902,
-      "grad_norm": 1.4754451513290405,
-      "learning_rate": 6.474472361809046e-05,
-      "loss": 5.204,
-      "step": 35591
-    },
-    {
-      "epoch": 18.56166883963494,
-      "grad_norm": 1.4590420722961426,
-      "learning_rate": 6.474371859296484e-05,
-      "loss": 5.0794,
-      "step": 35592
-    },
-    {
-      "epoch": 18.56219035202086,
-      "grad_norm": 1.4891713857650757,
-      "learning_rate": 6.47427135678392e-05,
-      "loss": 5.5038,
-      "step": 35593
-    },
-    {
-      "epoch": 18.56271186440678,
-      "grad_norm": 1.519974708557129,
-      "learning_rate": 6.474170854271358e-05,
-      "loss": 5.1938,
-      "step": 35594
-    },
-    {
-      "epoch": 18.5632333767927,
-      "grad_norm": 1.5187040567398071,
-      "learning_rate": 6.474070351758794e-05,
-      "loss": 4.9275,
-      "step": 35595
-    },
-    {
-      "epoch": 18.563754889178618,
-      "grad_norm": 1.4492067098617554,
-      "learning_rate": 6.473969849246232e-05,
-      "loss": 5.1744,
-      "step": 35596
-    },
-    {
-      "epoch": 18.564276401564538,
-      "grad_norm": 1.523616075515747,
-      "learning_rate": 6.473869346733668e-05,
-      "loss": 5.2482,
-      "step": 35597
-    },
-    {
-      "epoch": 18.564797913950457,
-      "grad_norm": 1.5563427209854126,
-      "learning_rate": 6.473768844221105e-05,
-      "loss": 4.7921,
-      "step": 35598
-    },
-    {
-      "epoch": 18.565319426336377,
-      "grad_norm": 1.571441411972046,
-      "learning_rate": 6.473668341708543e-05,
-      "loss": 5.3674,
-      "step": 35599
-    },
-    {
-      "epoch": 18.565840938722296,
-      "grad_norm": 1.5123358964920044,
-      "learning_rate": 6.473567839195979e-05,
-      "loss": 5.4699,
-      "step": 35600
-    },
-    {
-      "epoch": 18.566362451108215,
-      "grad_norm": 1.5096668004989624,
-      "learning_rate": 6.473467336683417e-05,
-      "loss": 5.2786,
-      "step": 35601
-    },
-    {
-      "epoch": 18.56688396349413,
-      "grad_norm": 1.491349220275879,
-      "learning_rate": 6.473366834170855e-05,
-      "loss": 5.1126,
-      "step": 35602
-    },
-    {
-      "epoch": 18.56740547588005,
-      "grad_norm": 1.4231665134429932,
-      "learning_rate": 6.473266331658292e-05,
-      "loss": 5.8665,
-      "step": 35603
-    },
-    {
-      "epoch": 18.56792698826597,
-      "grad_norm": 1.5278055667877197,
-      "learning_rate": 6.473165829145729e-05,
-      "loss": 5.5027,
-      "step": 35604
-    },
-    {
-      "epoch": 18.56844850065189,
-      "grad_norm": 1.3443926572799683,
-      "learning_rate": 6.473065326633167e-05,
-      "loss": 4.8755,
-      "step": 35605
-    },
-    {
-      "epoch": 18.56897001303781,
-      "grad_norm": 1.5191975831985474,
-      "learning_rate": 6.472964824120603e-05,
-      "loss": 4.6944,
-      "step": 35606
-    },
-    {
-      "epoch": 18.56949152542373,
-      "grad_norm": 1.4903696775436401,
-      "learning_rate": 6.472864321608041e-05,
-      "loss": 5.1913,
-      "step": 35607
-    },
-    {
-      "epoch": 18.570013037809648,
-      "grad_norm": 1.5194417238235474,
-      "learning_rate": 6.472763819095477e-05,
-      "loss": 5.054,
-      "step": 35608
-    },
-    {
-      "epoch": 18.570534550195568,
-      "grad_norm": 1.4624252319335938,
-      "learning_rate": 6.472663316582915e-05,
-      "loss": 5.3377,
-      "step": 35609
-    },
-    {
-      "epoch": 18.571056062581487,
-      "grad_norm": 1.4921990633010864,
-      "learning_rate": 6.472562814070352e-05,
-      "loss": 4.8174,
-      "step": 35610
-    },
-    {
-      "epoch": 18.571577574967407,
-      "grad_norm": 1.5260392427444458,
-      "learning_rate": 6.472462311557789e-05,
-      "loss": 5.1678,
-      "step": 35611
-    },
-    {
-      "epoch": 18.572099087353326,
-      "grad_norm": 1.4944303035736084,
-      "learning_rate": 6.472361809045227e-05,
-      "loss": 5.4034,
-      "step": 35612
-    },
-    {
-      "epoch": 18.572620599739246,
-      "grad_norm": 1.4951120615005493,
-      "learning_rate": 6.472261306532663e-05,
-      "loss": 5.143,
-      "step": 35613
-    },
-    {
-      "epoch": 18.57314211212516,
-      "grad_norm": 1.5664386749267578,
-      "learning_rate": 6.472160804020101e-05,
-      "loss": 5.4,
-      "step": 35614
-    },
-    {
-      "epoch": 18.57366362451108,
-      "grad_norm": 1.5019649267196655,
-      "learning_rate": 6.472060301507538e-05,
-      "loss": 5.3375,
-      "step": 35615
-    },
-    {
-      "epoch": 18.574185136897,
-      "grad_norm": 1.5554442405700684,
-      "learning_rate": 6.471959798994975e-05,
-      "loss": 5.0062,
-      "step": 35616
-    },
-    {
-      "epoch": 18.57470664928292,
-      "grad_norm": 1.6129951477050781,
-      "learning_rate": 6.471859296482412e-05,
-      "loss": 5.1277,
-      "step": 35617
-    },
-    {
-      "epoch": 18.57522816166884,
-      "grad_norm": 1.5531551837921143,
-      "learning_rate": 6.47175879396985e-05,
-      "loss": 5.8224,
-      "step": 35618
-    },
-    {
-      "epoch": 18.57574967405476,
-      "grad_norm": 1.3367098569869995,
-      "learning_rate": 6.471658291457286e-05,
-      "loss": 5.6557,
-      "step": 35619
-    },
-    {
-      "epoch": 18.576271186440678,
-      "grad_norm": 1.5051144361495972,
-      "learning_rate": 6.471557788944724e-05,
-      "loss": 4.995,
-      "step": 35620
-    },
-    {
-      "epoch": 18.576792698826598,
-      "grad_norm": 1.556246280670166,
-      "learning_rate": 6.47145728643216e-05,
-      "loss": 5.3415,
-      "step": 35621
-    },
-    {
-      "epoch": 18.577314211212517,
-      "grad_norm": 1.4459372758865356,
-      "learning_rate": 6.471356783919598e-05,
-      "loss": 5.6476,
-      "step": 35622
-    },
-    {
-      "epoch": 18.577835723598437,
-      "grad_norm": 1.3812949657440186,
-      "learning_rate": 6.471256281407036e-05,
-      "loss": 5.6466,
-      "step": 35623
-    },
-    {
-      "epoch": 18.578357235984356,
-      "grad_norm": 1.3929649591445923,
-      "learning_rate": 6.471155778894474e-05,
-      "loss": 5.4131,
-      "step": 35624
-    },
-    {
-      "epoch": 18.578878748370272,
-      "grad_norm": 1.516778826713562,
-      "learning_rate": 6.47105527638191e-05,
-      "loss": 5.3772,
-      "step": 35625
-    },
-    {
-      "epoch": 18.57940026075619,
-      "grad_norm": 1.4588415622711182,
-      "learning_rate": 6.470954773869347e-05,
-      "loss": 5.1021,
-      "step": 35626
-    },
-    {
-      "epoch": 18.57992177314211,
-      "grad_norm": 1.5326430797576904,
-      "learning_rate": 6.470854271356784e-05,
-      "loss": 5.336,
-      "step": 35627
-    },
-    {
-      "epoch": 18.58044328552803,
-      "grad_norm": 1.4224209785461426,
-      "learning_rate": 6.470753768844221e-05,
-      "loss": 5.3616,
-      "step": 35628
-    },
-    {
-      "epoch": 18.58096479791395,
-      "grad_norm": 1.4800015687942505,
-      "learning_rate": 6.470653266331659e-05,
-      "loss": 5.6189,
-      "step": 35629
-    },
-    {
-      "epoch": 18.58148631029987,
-      "grad_norm": 1.4834058284759521,
-      "learning_rate": 6.470552763819095e-05,
-      "loss": 5.2629,
-      "step": 35630
-    },
-    {
-      "epoch": 18.58200782268579,
-      "grad_norm": 1.4612185955047607,
-      "learning_rate": 6.470452261306533e-05,
-      "loss": 5.2755,
-      "step": 35631
-    },
-    {
-      "epoch": 18.58252933507171,
-      "grad_norm": 1.642467975616455,
-      "learning_rate": 6.47035175879397e-05,
-      "loss": 5.251,
-      "step": 35632
-    },
-    {
-      "epoch": 18.583050847457628,
-      "grad_norm": 1.5199463367462158,
-      "learning_rate": 6.470251256281408e-05,
-      "loss": 5.2039,
-      "step": 35633
-    },
-    {
-      "epoch": 18.583572359843547,
-      "grad_norm": 1.502884030342102,
-      "learning_rate": 6.470150753768845e-05,
-      "loss": 5.4855,
-      "step": 35634
-    },
-    {
-      "epoch": 18.584093872229467,
-      "grad_norm": 1.439645528793335,
-      "learning_rate": 6.470050251256283e-05,
-      "loss": 5.6512,
-      "step": 35635
-    },
-    {
-      "epoch": 18.584615384615386,
-      "grad_norm": 1.5089534521102905,
-      "learning_rate": 6.469949748743719e-05,
-      "loss": 4.8927,
-      "step": 35636
-    },
-    {
-      "epoch": 18.585136897001306,
-      "grad_norm": 1.485379934310913,
-      "learning_rate": 6.469849246231157e-05,
-      "loss": 5.6404,
-      "step": 35637
-    },
-    {
-      "epoch": 18.58565840938722,
-      "grad_norm": 1.520887017250061,
-      "learning_rate": 6.469748743718593e-05,
-      "loss": 5.4014,
-      "step": 35638
-    },
-    {
-      "epoch": 18.58617992177314,
-      "grad_norm": 1.4602385759353638,
-      "learning_rate": 6.46964824120603e-05,
-      "loss": 5.1413,
-      "step": 35639
-    },
-    {
-      "epoch": 18.58670143415906,
-      "grad_norm": 1.437320590019226,
-      "learning_rate": 6.469547738693467e-05,
-      "loss": 4.6434,
-      "step": 35640
-    },
-    {
-      "epoch": 18.58722294654498,
-      "grad_norm": 1.4553226232528687,
-      "learning_rate": 6.469447236180904e-05,
-      "loss": 5.4599,
-      "step": 35641
-    },
-    {
-      "epoch": 18.5877444589309,
-      "grad_norm": 1.6096646785736084,
-      "learning_rate": 6.469346733668342e-05,
-      "loss": 5.5541,
-      "step": 35642
-    },
-    {
-      "epoch": 18.58826597131682,
-      "grad_norm": 1.6640825271606445,
-      "learning_rate": 6.46924623115578e-05,
-      "loss": 5.3014,
-      "step": 35643
-    },
-    {
-      "epoch": 18.58878748370274,
-      "grad_norm": 1.4291355609893799,
-      "learning_rate": 6.469145728643217e-05,
-      "loss": 5.8131,
-      "step": 35644
-    },
-    {
-      "epoch": 18.589308996088658,
-      "grad_norm": 1.4709298610687256,
-      "learning_rate": 6.469045226130654e-05,
-      "loss": 5.1337,
-      "step": 35645
-    },
-    {
-      "epoch": 18.589830508474577,
-      "grad_norm": 1.5349431037902832,
-      "learning_rate": 6.468944723618091e-05,
-      "loss": 5.3621,
-      "step": 35646
-    },
-    {
-      "epoch": 18.590352020860497,
-      "grad_norm": 1.4615540504455566,
-      "learning_rate": 6.468844221105528e-05,
-      "loss": 5.4458,
-      "step": 35647
-    },
-    {
-      "epoch": 18.590873533246416,
-      "grad_norm": 1.4666386842727661,
-      "learning_rate": 6.468743718592966e-05,
-      "loss": 5.1142,
-      "step": 35648
-    },
-    {
-      "epoch": 18.591395045632332,
-      "grad_norm": 1.5116899013519287,
-      "learning_rate": 6.468643216080402e-05,
-      "loss": 5.5801,
-      "step": 35649
-    },
-    {
-      "epoch": 18.59191655801825,
-      "grad_norm": 1.4550970792770386,
-      "learning_rate": 6.46854271356784e-05,
-      "loss": 5.6166,
-      "step": 35650
-    },
-    {
-      "epoch": 18.59243807040417,
-      "grad_norm": 1.4989086389541626,
-      "learning_rate": 6.468442211055276e-05,
-      "loss": 5.4623,
-      "step": 35651
-    },
-    {
-      "epoch": 18.59295958279009,
-      "grad_norm": 1.4663373231887817,
-      "learning_rate": 6.468341708542714e-05,
-      "loss": 5.11,
-      "step": 35652
-    },
-    {
-      "epoch": 18.59348109517601,
-      "grad_norm": 1.4460784196853638,
-      "learning_rate": 6.468241206030152e-05,
-      "loss": 5.7037,
-      "step": 35653
-    },
-    {
-      "epoch": 18.59400260756193,
-      "grad_norm": 1.3866466283798218,
-      "learning_rate": 6.468140703517588e-05,
-      "loss": 5.6477,
-      "step": 35654
-    },
-    {
-      "epoch": 18.59452411994785,
-      "grad_norm": 1.504508376121521,
-      "learning_rate": 6.468040201005026e-05,
-      "loss": 5.0559,
-      "step": 35655
-    },
-    {
-      "epoch": 18.59504563233377,
-      "grad_norm": 1.4549195766448975,
-      "learning_rate": 6.467939698492462e-05,
-      "loss": 5.4825,
-      "step": 35656
-    },
-    {
-      "epoch": 18.595567144719688,
-      "grad_norm": 1.5147217512130737,
-      "learning_rate": 6.4678391959799e-05,
-      "loss": 5.0815,
-      "step": 35657
-    },
-    {
-      "epoch": 18.596088657105607,
-      "grad_norm": 1.5177127122879028,
-      "learning_rate": 6.467738693467337e-05,
-      "loss": 5.6046,
-      "step": 35658
-    },
-    {
-      "epoch": 18.596610169491527,
-      "grad_norm": 1.556951880455017,
-      "learning_rate": 6.467638190954774e-05,
-      "loss": 4.6431,
-      "step": 35659
-    },
-    {
-      "epoch": 18.597131681877446,
-      "grad_norm": 1.526574730873108,
-      "learning_rate": 6.467537688442211e-05,
-      "loss": 5.2371,
-      "step": 35660
-    },
-    {
-      "epoch": 18.597653194263362,
-      "grad_norm": 1.4985172748565674,
-      "learning_rate": 6.467437185929649e-05,
-      "loss": 5.5868,
-      "step": 35661
-    },
-    {
-      "epoch": 18.59817470664928,
-      "grad_norm": 1.6932499408721924,
-      "learning_rate": 6.467336683417085e-05,
-      "loss": 5.7849,
-      "step": 35662
-    },
-    {
-      "epoch": 18.5986962190352,
-      "grad_norm": 1.4690706729888916,
-      "learning_rate": 6.467236180904523e-05,
-      "loss": 5.6624,
-      "step": 35663
-    },
-    {
-      "epoch": 18.59921773142112,
-      "grad_norm": 1.5380942821502686,
-      "learning_rate": 6.46713567839196e-05,
-      "loss": 5.2306,
-      "step": 35664
-    },
-    {
-      "epoch": 18.59973924380704,
-      "grad_norm": 1.5486863851547241,
-      "learning_rate": 6.467035175879397e-05,
-      "loss": 5.2321,
-      "step": 35665
-    },
-    {
-      "epoch": 18.60026075619296,
-      "grad_norm": 1.4356722831726074,
-      "learning_rate": 6.466934673366835e-05,
-      "loss": 5.4949,
-      "step": 35666
-    },
-    {
-      "epoch": 18.60078226857888,
-      "grad_norm": 1.4122802019119263,
-      "learning_rate": 6.466834170854271e-05,
-      "loss": 5.8333,
-      "step": 35667
-    },
-    {
-      "epoch": 18.6013037809648,
-      "grad_norm": 1.5342355966567993,
-      "learning_rate": 6.466733668341709e-05,
-      "loss": 5.2377,
-      "step": 35668
-    },
-    {
-      "epoch": 18.601825293350718,
-      "grad_norm": 1.5379911661148071,
-      "learning_rate": 6.466633165829145e-05,
-      "loss": 5.1378,
-      "step": 35669
-    },
-    {
-      "epoch": 18.602346805736637,
-      "grad_norm": 1.5277397632598877,
-      "learning_rate": 6.466532663316583e-05,
-      "loss": 5.5737,
-      "step": 35670
-    },
-    {
-      "epoch": 18.602868318122557,
-      "grad_norm": 1.4944877624511719,
-      "learning_rate": 6.46643216080402e-05,
-      "loss": 5.3562,
-      "step": 35671
-    },
-    {
-      "epoch": 18.603389830508476,
-      "grad_norm": 1.6642173528671265,
-      "learning_rate": 6.466331658291457e-05,
-      "loss": 5.2812,
-      "step": 35672
-    },
-    {
-      "epoch": 18.603911342894392,
-      "grad_norm": 1.5946834087371826,
-      "learning_rate": 6.466231155778895e-05,
-      "loss": 5.2073,
-      "step": 35673
-    },
-    {
-      "epoch": 18.60443285528031,
-      "grad_norm": 1.565137505531311,
-      "learning_rate": 6.466130653266333e-05,
-      "loss": 4.7934,
-      "step": 35674
-    },
-    {
-      "epoch": 18.60495436766623,
-      "grad_norm": 1.613046407699585,
-      "learning_rate": 6.46603015075377e-05,
-      "loss": 5.1202,
-      "step": 35675
-    },
-    {
-      "epoch": 18.60547588005215,
-      "grad_norm": 1.461641788482666,
-      "learning_rate": 6.465929648241207e-05,
-      "loss": 5.2907,
-      "step": 35676
-    },
-    {
-      "epoch": 18.60599739243807,
-      "grad_norm": 1.5208925008773804,
-      "learning_rate": 6.465829145728644e-05,
-      "loss": 5.4786,
-      "step": 35677
-    },
-    {
-      "epoch": 18.60651890482399,
-      "grad_norm": 1.5136189460754395,
-      "learning_rate": 6.46572864321608e-05,
-      "loss": 5.0732,
-      "step": 35678
-    },
-    {
-      "epoch": 18.60704041720991,
-      "grad_norm": 1.402705430984497,
-      "learning_rate": 6.465628140703518e-05,
-      "loss": 5.7,
-      "step": 35679
-    },
-    {
-      "epoch": 18.60756192959583,
-      "grad_norm": 1.5396026372909546,
-      "learning_rate": 6.465527638190954e-05,
-      "loss": 5.2216,
-      "step": 35680
-    },
-    {
-      "epoch": 18.608083441981748,
-      "grad_norm": 1.6411998271942139,
-      "learning_rate": 6.465427135678392e-05,
-      "loss": 5.3357,
-      "step": 35681
-    },
-    {
-      "epoch": 18.608604954367667,
-      "grad_norm": 1.4079591035842896,
-      "learning_rate": 6.465326633165828e-05,
-      "loss": 5.7004,
-      "step": 35682
-    },
-    {
-      "epoch": 18.609126466753587,
-      "grad_norm": 1.4946318864822388,
-      "learning_rate": 6.465226130653266e-05,
-      "loss": 5.3283,
-      "step": 35683
-    },
-    {
-      "epoch": 18.609647979139506,
-      "grad_norm": 1.5205613374710083,
-      "learning_rate": 6.465125628140704e-05,
-      "loss": 5.419,
-      "step": 35684
-    },
-    {
-      "epoch": 18.610169491525422,
-      "grad_norm": 1.369079351425171,
-      "learning_rate": 6.465025125628142e-05,
-      "loss": 4.9419,
-      "step": 35685
-    },
-    {
-      "epoch": 18.61069100391134,
-      "grad_norm": 1.4976699352264404,
-      "learning_rate": 6.464924623115578e-05,
-      "loss": 5.144,
-      "step": 35686
-    },
-    {
-      "epoch": 18.61121251629726,
-      "grad_norm": 1.4628801345825195,
-      "learning_rate": 6.464824120603016e-05,
-      "loss": 5.1055,
-      "step": 35687
-    },
-    {
-      "epoch": 18.61173402868318,
-      "grad_norm": 1.5604889392852783,
-      "learning_rate": 6.464723618090452e-05,
-      "loss": 5.4881,
-      "step": 35688
-    },
-    {
-      "epoch": 18.6122555410691,
-      "grad_norm": 1.5043004751205444,
-      "learning_rate": 6.46462311557789e-05,
-      "loss": 5.5629,
-      "step": 35689
-    },
-    {
-      "epoch": 18.61277705345502,
-      "grad_norm": 1.5009984970092773,
-      "learning_rate": 6.464522613065327e-05,
-      "loss": 5.3097,
-      "step": 35690
-    },
-    {
-      "epoch": 18.61329856584094,
-      "grad_norm": 1.4242819547653198,
-      "learning_rate": 6.464422110552763e-05,
-      "loss": 5.6209,
-      "step": 35691
-    },
-    {
-      "epoch": 18.61382007822686,
-      "grad_norm": 1.6416399478912354,
-      "learning_rate": 6.464321608040201e-05,
-      "loss": 5.1948,
-      "step": 35692
-    },
-    {
-      "epoch": 18.614341590612778,
-      "grad_norm": 1.4774624109268188,
-      "learning_rate": 6.464221105527639e-05,
-      "loss": 4.8668,
-      "step": 35693
-    },
-    {
-      "epoch": 18.614863102998697,
-      "grad_norm": 1.4252318143844604,
-      "learning_rate": 6.464120603015076e-05,
-      "loss": 5.75,
-      "step": 35694
-    },
-    {
-      "epoch": 18.615384615384617,
-      "grad_norm": 1.5123910903930664,
-      "learning_rate": 6.464020100502513e-05,
-      "loss": 5.1209,
-      "step": 35695
-    },
-    {
-      "epoch": 18.615906127770536,
-      "grad_norm": 1.38404381275177,
-      "learning_rate": 6.463919597989951e-05,
-      "loss": 5.8219,
-      "step": 35696
-    },
-    {
-      "epoch": 18.616427640156452,
-      "grad_norm": 1.5775744915008545,
-      "learning_rate": 6.463819095477387e-05,
-      "loss": 5.3173,
-      "step": 35697
-    },
-    {
-      "epoch": 18.61694915254237,
-      "grad_norm": 1.53171706199646,
-      "learning_rate": 6.463718592964825e-05,
-      "loss": 5.5244,
-      "step": 35698
-    },
-    {
-      "epoch": 18.61747066492829,
-      "grad_norm": 1.5157113075256348,
-      "learning_rate": 6.463618090452261e-05,
-      "loss": 5.4933,
-      "step": 35699
-    },
-    {
-      "epoch": 18.61799217731421,
-      "grad_norm": 1.4244669675827026,
-      "learning_rate": 6.463517587939699e-05,
-      "loss": 5.5625,
-      "step": 35700
-    },
-    {
-      "epoch": 18.61851368970013,
-      "grad_norm": 1.5404554605484009,
-      "learning_rate": 6.463417085427136e-05,
-      "loss": 4.9083,
-      "step": 35701
-    },
-    {
-      "epoch": 18.61903520208605,
-      "grad_norm": 1.511816143989563,
-      "learning_rate": 6.463316582914573e-05,
-      "loss": 5.408,
-      "step": 35702
-    },
-    {
-      "epoch": 18.61955671447197,
-      "grad_norm": 1.6500056982040405,
-      "learning_rate": 6.46321608040201e-05,
-      "loss": 4.9072,
-      "step": 35703
-    },
-    {
-      "epoch": 18.62007822685789,
-      "grad_norm": 1.4363733530044556,
-      "learning_rate": 6.463115577889448e-05,
-      "loss": 5.2935,
-      "step": 35704
-    },
-    {
-      "epoch": 18.620599739243808,
-      "grad_norm": 1.4194766283035278,
-      "learning_rate": 6.463015075376885e-05,
-      "loss": 5.5554,
-      "step": 35705
-    },
-    {
-      "epoch": 18.621121251629727,
-      "grad_norm": 1.431024193763733,
-      "learning_rate": 6.462914572864322e-05,
-      "loss": 5.423,
-      "step": 35706
-    },
-    {
-      "epoch": 18.621642764015647,
-      "grad_norm": 1.6386945247650146,
-      "learning_rate": 6.46281407035176e-05,
-      "loss": 5.0412,
-      "step": 35707
-    },
-    {
-      "epoch": 18.622164276401566,
-      "grad_norm": 1.5235464572906494,
-      "learning_rate": 6.462713567839196e-05,
-      "loss": 5.4026,
-      "step": 35708
-    },
-    {
-      "epoch": 18.622685788787482,
-      "grad_norm": 1.6194361448287964,
-      "learning_rate": 6.462613065326634e-05,
-      "loss": 4.6249,
-      "step": 35709
-    },
-    {
-      "epoch": 18.6232073011734,
-      "grad_norm": 1.5814553499221802,
-      "learning_rate": 6.46251256281407e-05,
-      "loss": 4.9726,
-      "step": 35710
-    },
-    {
-      "epoch": 18.62372881355932,
-      "grad_norm": 1.6135741472244263,
-      "learning_rate": 6.462412060301508e-05,
-      "loss": 5.2646,
-      "step": 35711
-    },
-    {
-      "epoch": 18.62425032594524,
-      "grad_norm": 1.6042035818099976,
-      "learning_rate": 6.462311557788944e-05,
-      "loss": 5.278,
-      "step": 35712
-    },
-    {
-      "epoch": 18.62477183833116,
-      "grad_norm": 1.6219844818115234,
-      "learning_rate": 6.462211055276382e-05,
-      "loss": 4.9808,
-      "step": 35713
-    },
-    {
-      "epoch": 18.62529335071708,
-      "grad_norm": 1.4816056489944458,
-      "learning_rate": 6.46211055276382e-05,
-      "loss": 5.4879,
-      "step": 35714
-    },
-    {
-      "epoch": 18.625814863103,
-      "grad_norm": 1.5391602516174316,
-      "learning_rate": 6.462010050251258e-05,
-      "loss": 5.3604,
-      "step": 35715
-    },
-    {
-      "epoch": 18.62633637548892,
-      "grad_norm": 1.5850592851638794,
-      "learning_rate": 6.461909547738694e-05,
-      "loss": 5.4922,
-      "step": 35716
-    },
-    {
-      "epoch": 18.626857887874838,
-      "grad_norm": 1.5333458185195923,
-      "learning_rate": 6.461809045226132e-05,
-      "loss": 5.6645,
-      "step": 35717
-    },
-    {
-      "epoch": 18.627379400260757,
-      "grad_norm": 1.4506787061691284,
-      "learning_rate": 6.461708542713568e-05,
-      "loss": 5.6141,
-      "step": 35718
-    },
-    {
-      "epoch": 18.627900912646677,
-      "grad_norm": 1.560341238975525,
-      "learning_rate": 6.461608040201005e-05,
-      "loss": 5.1058,
-      "step": 35719
-    },
-    {
-      "epoch": 18.628422425032596,
-      "grad_norm": 1.5256457328796387,
-      "learning_rate": 6.461507537688443e-05,
-      "loss": 5.0338,
-      "step": 35720
-    },
-    {
-      "epoch": 18.628943937418512,
-      "grad_norm": 1.544498324394226,
-      "learning_rate": 6.461407035175879e-05,
-      "loss": 5.2393,
-      "step": 35721
-    },
-    {
-      "epoch": 18.62946544980443,
-      "grad_norm": 1.4463434219360352,
-      "learning_rate": 6.461306532663317e-05,
-      "loss": 4.3408,
-      "step": 35722
-    },
-    {
-      "epoch": 18.62998696219035,
-      "grad_norm": 1.5248974561691284,
-      "learning_rate": 6.461206030150753e-05,
-      "loss": 5.1851,
-      "step": 35723
-    },
-    {
-      "epoch": 18.63050847457627,
-      "grad_norm": 1.4676568508148193,
-      "learning_rate": 6.461105527638191e-05,
-      "loss": 5.5468,
-      "step": 35724
-    },
-    {
-      "epoch": 18.63102998696219,
-      "grad_norm": 1.448968529701233,
-      "learning_rate": 6.461005025125629e-05,
-      "loss": 4.7021,
-      "step": 35725
-    },
-    {
-      "epoch": 18.63155149934811,
-      "grad_norm": 1.4675486087799072,
-      "learning_rate": 6.460904522613067e-05,
-      "loss": 5.4268,
-      "step": 35726
-    },
-    {
-      "epoch": 18.63207301173403,
-      "grad_norm": 1.551528811454773,
-      "learning_rate": 6.460804020100503e-05,
-      "loss": 4.8253,
-      "step": 35727
-    },
-    {
-      "epoch": 18.63259452411995,
-      "grad_norm": 1.5216543674468994,
-      "learning_rate": 6.460703517587941e-05,
-      "loss": 5.5637,
-      "step": 35728
-    },
-    {
-      "epoch": 18.633116036505868,
-      "grad_norm": 1.447148323059082,
-      "learning_rate": 6.460603015075377e-05,
-      "loss": 5.1024,
-      "step": 35729
-    },
-    {
-      "epoch": 18.633637548891787,
-      "grad_norm": 1.4456433057785034,
-      "learning_rate": 6.460502512562815e-05,
-      "loss": 5.6436,
-      "step": 35730
-    },
-    {
-      "epoch": 18.634159061277707,
-      "grad_norm": 1.518314242362976,
-      "learning_rate": 6.460402010050251e-05,
-      "loss": 4.7601,
-      "step": 35731
-    },
-    {
-      "epoch": 18.634680573663623,
-      "grad_norm": 1.5895026922225952,
-      "learning_rate": 6.460301507537688e-05,
-      "loss": 5.2427,
-      "step": 35732
-    },
-    {
-      "epoch": 18.635202086049542,
-      "grad_norm": 1.4320745468139648,
-      "learning_rate": 6.460201005025126e-05,
-      "loss": 5.6824,
-      "step": 35733
-    },
-    {
-      "epoch": 18.63572359843546,
-      "grad_norm": 1.590407371520996,
-      "learning_rate": 6.460100502512563e-05,
-      "loss": 4.0585,
-      "step": 35734
-    },
-    {
-      "epoch": 18.63624511082138,
-      "grad_norm": 1.6626660823822021,
-      "learning_rate": 6.460000000000001e-05,
-      "loss": 5.2052,
-      "step": 35735
-    },
-    {
-      "epoch": 18.6367666232073,
-      "grad_norm": 1.5386260747909546,
-      "learning_rate": 6.459899497487438e-05,
-      "loss": 4.9133,
-      "step": 35736
-    },
-    {
-      "epoch": 18.63728813559322,
-      "grad_norm": 1.3685499429702759,
-      "learning_rate": 6.459798994974875e-05,
-      "loss": 5.8768,
-      "step": 35737
-    },
-    {
-      "epoch": 18.63780964797914,
-      "grad_norm": 1.5671584606170654,
-      "learning_rate": 6.459698492462312e-05,
-      "loss": 4.8872,
-      "step": 35738
-    },
-    {
-      "epoch": 18.63833116036506,
-      "grad_norm": 1.6712560653686523,
-      "learning_rate": 6.45959798994975e-05,
-      "loss": 4.5802,
-      "step": 35739
-    },
-    {
-      "epoch": 18.63885267275098,
-      "grad_norm": 1.5600121021270752,
-      "learning_rate": 6.459497487437186e-05,
-      "loss": 5.3992,
-      "step": 35740
-    },
-    {
-      "epoch": 18.639374185136898,
-      "grad_norm": 1.4680752754211426,
-      "learning_rate": 6.459396984924624e-05,
-      "loss": 5.5519,
-      "step": 35741
-    },
-    {
-      "epoch": 18.639895697522817,
-      "grad_norm": 1.5317106246948242,
-      "learning_rate": 6.45929648241206e-05,
-      "loss": 5.2691,
-      "step": 35742
-    },
-    {
-      "epoch": 18.640417209908737,
-      "grad_norm": 1.4375470876693726,
-      "learning_rate": 6.459195979899498e-05,
-      "loss": 5.1133,
-      "step": 35743
-    },
-    {
-      "epoch": 18.640938722294656,
-      "grad_norm": 1.5963140726089478,
-      "learning_rate": 6.459095477386934e-05,
-      "loss": 5.2218,
-      "step": 35744
-    },
-    {
-      "epoch": 18.641460234680572,
-      "grad_norm": 1.4930917024612427,
-      "learning_rate": 6.458994974874372e-05,
-      "loss": 5.4117,
-      "step": 35745
-    },
-    {
-      "epoch": 18.64198174706649,
-      "grad_norm": 1.5607619285583496,
-      "learning_rate": 6.45889447236181e-05,
-      "loss": 5.1923,
-      "step": 35746
-    },
-    {
-      "epoch": 18.64250325945241,
-      "grad_norm": 1.4626344442367554,
-      "learning_rate": 6.458793969849246e-05,
-      "loss": 5.5924,
-      "step": 35747
-    },
-    {
-      "epoch": 18.64302477183833,
-      "grad_norm": 1.4379467964172363,
-      "learning_rate": 6.458693467336684e-05,
-      "loss": 5.3966,
-      "step": 35748
-    },
-    {
-      "epoch": 18.64354628422425,
-      "grad_norm": 1.4184541702270508,
-      "learning_rate": 6.45859296482412e-05,
-      "loss": 5.8415,
-      "step": 35749
-    },
-    {
-      "epoch": 18.64406779661017,
-      "grad_norm": 1.500709891319275,
-      "learning_rate": 6.458492462311558e-05,
-      "loss": 5.0863,
-      "step": 35750
-    },
-    {
-      "epoch": 18.64458930899609,
-      "grad_norm": 1.3310719728469849,
-      "learning_rate": 6.458391959798995e-05,
-      "loss": 5.6213,
-      "step": 35751
-    },
-    {
-      "epoch": 18.64511082138201,
-      "grad_norm": 1.4872798919677734,
-      "learning_rate": 6.458291457286433e-05,
-      "loss": 5.5593,
-      "step": 35752
-    },
-    {
-      "epoch": 18.645632333767928,
-      "grad_norm": 1.587708592414856,
-      "learning_rate": 6.458190954773869e-05,
-      "loss": 5.3322,
-      "step": 35753
-    },
-    {
-      "epoch": 18.646153846153847,
-      "grad_norm": 1.401796817779541,
-      "learning_rate": 6.458090452261307e-05,
-      "loss": 5.5252,
-      "step": 35754
-    },
-    {
-      "epoch": 18.646675358539767,
-      "grad_norm": 1.4356905221939087,
-      "learning_rate": 6.457989949748743e-05,
-      "loss": 5.2006,
-      "step": 35755
-    },
-    {
-      "epoch": 18.647196870925683,
-      "grad_norm": 1.4411189556121826,
-      "learning_rate": 6.457889447236181e-05,
-      "loss": 5.6274,
-      "step": 35756
-    },
-    {
-      "epoch": 18.647718383311602,
-      "grad_norm": 1.4890936613082886,
-      "learning_rate": 6.457788944723619e-05,
-      "loss": 4.2956,
-      "step": 35757
-    },
-    {
-      "epoch": 18.64823989569752,
-      "grad_norm": 1.516455888748169,
-      "learning_rate": 6.457688442211055e-05,
-      "loss": 5.1727,
-      "step": 35758
-    },
-    {
-      "epoch": 18.64876140808344,
-      "grad_norm": 1.4456534385681152,
-      "learning_rate": 6.457587939698493e-05,
-      "loss": 4.9408,
-      "step": 35759
-    },
-    {
-      "epoch": 18.64928292046936,
-      "grad_norm": 1.503006100654602,
-      "learning_rate": 6.45748743718593e-05,
-      "loss": 4.8298,
-      "step": 35760
-    },
-    {
-      "epoch": 18.64980443285528,
-      "grad_norm": 1.4841952323913574,
-      "learning_rate": 6.457386934673367e-05,
-      "loss": 5.1959,
-      "step": 35761
-    },
-    {
-      "epoch": 18.6503259452412,
-      "grad_norm": 1.4460926055908203,
-      "learning_rate": 6.457286432160804e-05,
-      "loss": 5.0603,
-      "step": 35762
-    },
-    {
-      "epoch": 18.65084745762712,
-      "grad_norm": 1.4572546482086182,
-      "learning_rate": 6.457185929648241e-05,
-      "loss": 5.3988,
-      "step": 35763
-    },
-    {
-      "epoch": 18.65136897001304,
-      "grad_norm": 1.5059317350387573,
-      "learning_rate": 6.457085427135678e-05,
-      "loss": 4.9519,
-      "step": 35764
-    },
-    {
-      "epoch": 18.651890482398958,
-      "grad_norm": 1.4994629621505737,
-      "learning_rate": 6.456984924623116e-05,
-      "loss": 5.282,
-      "step": 35765
-    },
-    {
-      "epoch": 18.652411994784877,
-      "grad_norm": 1.5077558755874634,
-      "learning_rate": 6.456884422110553e-05,
-      "loss": 5.3685,
-      "step": 35766
-    },
-    {
-      "epoch": 18.652933507170797,
-      "grad_norm": 1.5312817096710205,
-      "learning_rate": 6.456783919597991e-05,
-      "loss": 5.5764,
-      "step": 35767
-    },
-    {
-      "epoch": 18.653455019556713,
-      "grad_norm": 1.5252366065979004,
-      "learning_rate": 6.456683417085428e-05,
-      "loss": 5.3226,
-      "step": 35768
-    },
-    {
-      "epoch": 18.653976531942632,
-      "grad_norm": 1.3596715927124023,
-      "learning_rate": 6.456582914572865e-05,
-      "loss": 5.691,
-      "step": 35769
-    },
-    {
-      "epoch": 18.65449804432855,
-      "grad_norm": 1.4140753746032715,
-      "learning_rate": 6.456482412060302e-05,
-      "loss": 5.9027,
-      "step": 35770
-    },
-    {
-      "epoch": 18.65501955671447,
-      "grad_norm": 1.514192819595337,
-      "learning_rate": 6.456381909547738e-05,
-      "loss": 5.4687,
-      "step": 35771
-    },
-    {
-      "epoch": 18.65554106910039,
-      "grad_norm": 1.4447715282440186,
-      "learning_rate": 6.456281407035176e-05,
-      "loss": 5.5384,
-      "step": 35772
-    },
-    {
-      "epoch": 18.65606258148631,
-      "grad_norm": 1.5164729356765747,
-      "learning_rate": 6.456180904522613e-05,
-      "loss": 5.6949,
-      "step": 35773
-    },
-    {
-      "epoch": 18.65658409387223,
-      "grad_norm": 1.4898416996002197,
-      "learning_rate": 6.45608040201005e-05,
-      "loss": 5.2513,
-      "step": 35774
-    },
-    {
-      "epoch": 18.65710560625815,
-      "grad_norm": 1.6705743074417114,
-      "learning_rate": 6.455979899497487e-05,
-      "loss": 5.2746,
-      "step": 35775
-    },
-    {
-      "epoch": 18.65762711864407,
-      "grad_norm": 1.5918811559677124,
-      "learning_rate": 6.455879396984925e-05,
-      "loss": 5.0631,
-      "step": 35776
-    },
-    {
-      "epoch": 18.658148631029988,
-      "grad_norm": 1.5300389528274536,
-      "learning_rate": 6.455778894472362e-05,
-      "loss": 5.1861,
-      "step": 35777
-    },
-    {
-      "epoch": 18.658670143415907,
-      "grad_norm": 1.5191282033920288,
-      "learning_rate": 6.4556783919598e-05,
-      "loss": 5.0186,
-      "step": 35778
-    },
-    {
-      "epoch": 18.659191655801827,
-      "grad_norm": 1.5296348333358765,
-      "learning_rate": 6.455577889447237e-05,
-      "loss": 5.4964,
-      "step": 35779
-    },
-    {
-      "epoch": 18.659713168187743,
-      "grad_norm": 1.552445888519287,
-      "learning_rate": 6.455477386934674e-05,
-      "loss": 5.1891,
-      "step": 35780
-    },
-    {
-      "epoch": 18.660234680573662,
-      "grad_norm": 1.6074129343032837,
-      "learning_rate": 6.455376884422111e-05,
-      "loss": 5.1237,
-      "step": 35781
-    },
-    {
-      "epoch": 18.66075619295958,
-      "grad_norm": 1.5651556253433228,
-      "learning_rate": 6.455276381909548e-05,
-      "loss": 5.4602,
-      "step": 35782
-    },
-    {
-      "epoch": 18.6612777053455,
-      "grad_norm": 1.4472140073776245,
-      "learning_rate": 6.455175879396985e-05,
-      "loss": 5.3256,
-      "step": 35783
-    },
-    {
-      "epoch": 18.66179921773142,
-      "grad_norm": 1.5698143243789673,
-      "learning_rate": 6.455075376884421e-05,
-      "loss": 5.9515,
-      "step": 35784
-    },
-    {
-      "epoch": 18.66232073011734,
-      "grad_norm": 1.5940662622451782,
-      "learning_rate": 6.454974874371859e-05,
-      "loss": 5.0846,
-      "step": 35785
-    },
-    {
-      "epoch": 18.66284224250326,
-      "grad_norm": 1.4774161577224731,
-      "learning_rate": 6.454874371859297e-05,
-      "loss": 5.4936,
-      "step": 35786
-    },
-    {
-      "epoch": 18.66336375488918,
-      "grad_norm": 1.499632477760315,
-      "learning_rate": 6.454773869346735e-05,
-      "loss": 5.398,
-      "step": 35787
-    },
-    {
-      "epoch": 18.6638852672751,
-      "grad_norm": 1.4496902227401733,
-      "learning_rate": 6.454673366834171e-05,
-      "loss": 5.4012,
-      "step": 35788
-    },
-    {
-      "epoch": 18.664406779661018,
-      "grad_norm": 1.5756906270980835,
-      "learning_rate": 6.454572864321609e-05,
-      "loss": 5.1978,
-      "step": 35789
-    },
-    {
-      "epoch": 18.664928292046937,
-      "grad_norm": 1.5237839221954346,
-      "learning_rate": 6.454472361809045e-05,
-      "loss": 5.5109,
-      "step": 35790
-    },
-    {
-      "epoch": 18.665449804432857,
-      "grad_norm": 1.494284987449646,
-      "learning_rate": 6.454371859296483e-05,
-      "loss": 5.1215,
-      "step": 35791
-    },
-    {
-      "epoch": 18.665971316818773,
-      "grad_norm": 1.5743721723556519,
-      "learning_rate": 6.45427135678392e-05,
-      "loss": 5.106,
-      "step": 35792
-    },
-    {
-      "epoch": 18.666492829204692,
-      "grad_norm": 1.548122763633728,
-      "learning_rate": 6.454170854271357e-05,
-      "loss": 4.8674,
-      "step": 35793
-    },
-    {
-      "epoch": 18.667014341590612,
-      "grad_norm": 1.5801239013671875,
-      "learning_rate": 6.454070351758794e-05,
-      "loss": 4.8568,
-      "step": 35794
-    },
-    {
-      "epoch": 18.66753585397653,
-      "grad_norm": 1.7037566900253296,
-      "learning_rate": 6.453969849246232e-05,
-      "loss": 5.6387,
-      "step": 35795
-    },
-    {
-      "epoch": 18.66805736636245,
-      "grad_norm": 1.6037253141403198,
-      "learning_rate": 6.453869346733668e-05,
-      "loss": 5.3808,
-      "step": 35796
-    },
-    {
-      "epoch": 18.66857887874837,
-      "grad_norm": 1.4332889318466187,
-      "learning_rate": 6.453768844221106e-05,
-      "loss": 4.8014,
-      "step": 35797
-    },
-    {
-      "epoch": 18.66910039113429,
-      "grad_norm": 1.535971760749817,
-      "learning_rate": 6.453668341708544e-05,
-      "loss": 5.2199,
-      "step": 35798
-    },
-    {
-      "epoch": 18.66962190352021,
-      "grad_norm": 1.5308573246002197,
-      "learning_rate": 6.45356783919598e-05,
-      "loss": 5.2749,
-      "step": 35799
-    },
-    {
-      "epoch": 18.67014341590613,
-      "grad_norm": 1.5455416440963745,
-      "learning_rate": 6.453467336683418e-05,
-      "loss": 5.3524,
-      "step": 35800
-    },
-    {
-      "epoch": 18.670664928292048,
-      "grad_norm": 1.5711168050765991,
-      "learning_rate": 6.453366834170854e-05,
-      "loss": 5.0899,
-      "step": 35801
-    },
-    {
-      "epoch": 18.671186440677968,
-      "grad_norm": 1.5153120756149292,
-      "learning_rate": 6.453266331658292e-05,
-      "loss": 5.2082,
-      "step": 35802
-    },
-    {
-      "epoch": 18.671707953063887,
-      "grad_norm": 1.5068107843399048,
-      "learning_rate": 6.453165829145728e-05,
-      "loss": 5.2647,
-      "step": 35803
-    },
-    {
-      "epoch": 18.672229465449803,
-      "grad_norm": 1.542945384979248,
-      "learning_rate": 6.453065326633166e-05,
-      "loss": 5.39,
-      "step": 35804
-    },
-    {
-      "epoch": 18.672750977835722,
-      "grad_norm": 1.4741911888122559,
-      "learning_rate": 6.452964824120603e-05,
-      "loss": 4.944,
-      "step": 35805
-    },
-    {
-      "epoch": 18.673272490221642,
-      "grad_norm": 1.69082510471344,
-      "learning_rate": 6.45286432160804e-05,
-      "loss": 5.0588,
-      "step": 35806
-    },
-    {
-      "epoch": 18.67379400260756,
-      "grad_norm": 1.5544863939285278,
-      "learning_rate": 6.452763819095478e-05,
-      "loss": 5.4232,
-      "step": 35807
-    },
-    {
-      "epoch": 18.67431551499348,
-      "grad_norm": 1.5561741590499878,
-      "learning_rate": 6.452663316582916e-05,
-      "loss": 5.571,
-      "step": 35808
-    },
-    {
-      "epoch": 18.6748370273794,
-      "grad_norm": 1.4946218729019165,
-      "learning_rate": 6.452562814070352e-05,
-      "loss": 5.3819,
-      "step": 35809
-    },
-    {
-      "epoch": 18.67535853976532,
-      "grad_norm": 1.4943900108337402,
-      "learning_rate": 6.45246231155779e-05,
-      "loss": 5.3449,
-      "step": 35810
-    },
-    {
-      "epoch": 18.67588005215124,
-      "grad_norm": 1.4619845151901245,
-      "learning_rate": 6.452361809045227e-05,
-      "loss": 5.1864,
-      "step": 35811
-    },
-    {
-      "epoch": 18.67640156453716,
-      "grad_norm": 1.610729455947876,
-      "learning_rate": 6.452261306532663e-05,
-      "loss": 5.2922,
-      "step": 35812
-    },
-    {
-      "epoch": 18.676923076923078,
-      "grad_norm": 1.5920456647872925,
-      "learning_rate": 6.452160804020101e-05,
-      "loss": 4.7326,
-      "step": 35813
-    },
-    {
-      "epoch": 18.677444589308998,
-      "grad_norm": 1.4704784154891968,
-      "learning_rate": 6.452060301507537e-05,
-      "loss": 5.3544,
-      "step": 35814
-    },
-    {
-      "epoch": 18.677966101694913,
-      "grad_norm": 1.4732096195220947,
-      "learning_rate": 6.451959798994975e-05,
-      "loss": 5.5782,
-      "step": 35815
-    },
-    {
-      "epoch": 18.678487614080833,
-      "grad_norm": 1.6220703125,
-      "learning_rate": 6.451859296482411e-05,
-      "loss": 4.6908,
-      "step": 35816
-    },
-    {
-      "epoch": 18.679009126466752,
-      "grad_norm": 1.4792379140853882,
-      "learning_rate": 6.451758793969849e-05,
-      "loss": 5.1441,
-      "step": 35817
-    },
-    {
-      "epoch": 18.679530638852672,
-      "grad_norm": 1.550882339477539,
-      "learning_rate": 6.451658291457287e-05,
-      "loss": 4.8606,
-      "step": 35818
-    },
-    {
-      "epoch": 18.68005215123859,
-      "grad_norm": 1.5322669744491577,
-      "learning_rate": 6.451557788944725e-05,
-      "loss": 5.408,
-      "step": 35819
-    },
-    {
-      "epoch": 18.68057366362451,
-      "grad_norm": 1.508384346961975,
-      "learning_rate": 6.451457286432161e-05,
-      "loss": 5.418,
-      "step": 35820
-    },
-    {
-      "epoch": 18.68109517601043,
-      "grad_norm": 1.496322751045227,
-      "learning_rate": 6.451356783919599e-05,
-      "loss": 5.5945,
-      "step": 35821
-    },
-    {
-      "epoch": 18.68161668839635,
-      "grad_norm": 1.4665178060531616,
-      "learning_rate": 6.451256281407035e-05,
-      "loss": 5.4386,
-      "step": 35822
-    },
-    {
-      "epoch": 18.68213820078227,
-      "grad_norm": 1.4962414503097534,
-      "learning_rate": 6.451155778894473e-05,
-      "loss": 5.5519,
-      "step": 35823
-    },
-    {
-      "epoch": 18.68265971316819,
-      "grad_norm": 1.465462327003479,
-      "learning_rate": 6.45105527638191e-05,
-      "loss": 5.5345,
-      "step": 35824
-    },
-    {
-      "epoch": 18.683181225554108,
-      "grad_norm": 1.5143828392028809,
-      "learning_rate": 6.450954773869346e-05,
-      "loss": 5.6081,
-      "step": 35825
-    },
-    {
-      "epoch": 18.683702737940028,
-      "grad_norm": 1.6393991708755493,
-      "learning_rate": 6.450854271356784e-05,
-      "loss": 5.1193,
-      "step": 35826
-    },
-    {
-      "epoch": 18.684224250325947,
-      "grad_norm": 1.409268856048584,
-      "learning_rate": 6.450753768844222e-05,
-      "loss": 5.3908,
-      "step": 35827
-    },
-    {
-      "epoch": 18.684745762711863,
-      "grad_norm": 1.5955629348754883,
-      "learning_rate": 6.45065326633166e-05,
-      "loss": 5.5188,
-      "step": 35828
-    },
-    {
-      "epoch": 18.685267275097782,
-      "grad_norm": 1.5691931247711182,
-      "learning_rate": 6.450552763819096e-05,
-      "loss": 5.5986,
-      "step": 35829
-    },
-    {
-      "epoch": 18.685788787483702,
-      "grad_norm": 1.5182652473449707,
-      "learning_rate": 6.450452261306534e-05,
-      "loss": 5.2076,
-      "step": 35830
-    },
-    {
-      "epoch": 18.68631029986962,
-      "grad_norm": 1.515506386756897,
-      "learning_rate": 6.45035175879397e-05,
-      "loss": 5.7458,
-      "step": 35831
-    },
-    {
-      "epoch": 18.68683181225554,
-      "grad_norm": 1.483201503753662,
-      "learning_rate": 6.450251256281408e-05,
-      "loss": 5.3837,
-      "step": 35832
-    },
-    {
-      "epoch": 18.68735332464146,
-      "grad_norm": 1.503985047340393,
-      "learning_rate": 6.450150753768844e-05,
-      "loss": 5.5059,
-      "step": 35833
-    },
-    {
-      "epoch": 18.68787483702738,
-      "grad_norm": 1.4424192905426025,
-      "learning_rate": 6.450050251256282e-05,
-      "loss": 5.5429,
-      "step": 35834
-    },
-    {
-      "epoch": 18.6883963494133,
-      "grad_norm": 1.4684611558914185,
-      "learning_rate": 6.449949748743718e-05,
-      "loss": 5.5643,
-      "step": 35835
-    },
-    {
-      "epoch": 18.68891786179922,
-      "grad_norm": 1.6057968139648438,
-      "learning_rate": 6.449849246231156e-05,
-      "loss": 4.6394,
-      "step": 35836
-    },
-    {
-      "epoch": 18.689439374185138,
-      "grad_norm": 1.4764420986175537,
-      "learning_rate": 6.449748743718593e-05,
-      "loss": 5.3702,
-      "step": 35837
-    },
-    {
-      "epoch": 18.689960886571058,
-      "grad_norm": 1.584904670715332,
-      "learning_rate": 6.44964824120603e-05,
-      "loss": 5.1189,
-      "step": 35838
-    },
-    {
-      "epoch": 18.690482398956973,
-      "grad_norm": 1.5156574249267578,
-      "learning_rate": 6.449547738693468e-05,
-      "loss": 5.2726,
-      "step": 35839
-    },
-    {
-      "epoch": 18.691003911342893,
-      "grad_norm": 1.587459921836853,
-      "learning_rate": 6.449447236180905e-05,
-      "loss": 5.3815,
-      "step": 35840
-    },
-    {
-      "epoch": 18.691525423728812,
-      "grad_norm": 1.5527766942977905,
-      "learning_rate": 6.449346733668342e-05,
-      "loss": 5.427,
-      "step": 35841
-    },
-    {
-      "epoch": 18.692046936114732,
-      "grad_norm": 1.412217378616333,
-      "learning_rate": 6.449246231155779e-05,
-      "loss": 5.6249,
-      "step": 35842
-    },
-    {
-      "epoch": 18.69256844850065,
-      "grad_norm": 1.498233437538147,
-      "learning_rate": 6.449145728643217e-05,
-      "loss": 5.2736,
-      "step": 35843
-    },
-    {
-      "epoch": 18.69308996088657,
-      "grad_norm": 1.467262625694275,
-      "learning_rate": 6.449045226130653e-05,
-      "loss": 5.7096,
-      "step": 35844
-    },
-    {
-      "epoch": 18.69361147327249,
-      "grad_norm": 1.629478096961975,
-      "learning_rate": 6.448944723618091e-05,
-      "loss": 5.2919,
-      "step": 35845
-    },
-    {
-      "epoch": 18.69413298565841,
-      "grad_norm": 1.5553669929504395,
-      "learning_rate": 6.448844221105527e-05,
-      "loss": 4.806,
-      "step": 35846
-    },
-    {
-      "epoch": 18.69465449804433,
-      "grad_norm": 1.4115216732025146,
-      "learning_rate": 6.448743718592965e-05,
-      "loss": 5.396,
-      "step": 35847
-    },
-    {
-      "epoch": 18.69517601043025,
-      "grad_norm": 1.467923879623413,
-      "learning_rate": 6.448643216080403e-05,
-      "loss": 5.2609,
-      "step": 35848
-    },
-    {
-      "epoch": 18.695697522816168,
-      "grad_norm": 1.53794527053833,
-      "learning_rate": 6.44854271356784e-05,
-      "loss": 5.3818,
-      "step": 35849
-    },
-    {
-      "epoch": 18.696219035202088,
-      "grad_norm": 1.4258854389190674,
-      "learning_rate": 6.448442211055277e-05,
-      "loss": 5.5802,
-      "step": 35850
-    },
-    {
-      "epoch": 18.696740547588004,
-      "grad_norm": 1.5675604343414307,
-      "learning_rate": 6.448341708542713e-05,
-      "loss": 4.7035,
-      "step": 35851
-    },
-    {
-      "epoch": 18.697262059973923,
-      "grad_norm": 1.5047305822372437,
-      "learning_rate": 6.448241206030151e-05,
-      "loss": 5.5237,
-      "step": 35852
-    },
-    {
-      "epoch": 18.697783572359842,
-      "grad_norm": 1.460528016090393,
-      "learning_rate": 6.448140703517588e-05,
-      "loss": 5.5958,
-      "step": 35853
-    },
-    {
-      "epoch": 18.698305084745762,
-      "grad_norm": 1.5070680379867554,
-      "learning_rate": 6.448040201005025e-05,
-      "loss": 5.4623,
-      "step": 35854
-    },
-    {
-      "epoch": 18.69882659713168,
-      "grad_norm": 1.4966051578521729,
-      "learning_rate": 6.447939698492462e-05,
-      "loss": 5.1239,
-      "step": 35855
-    },
-    {
-      "epoch": 18.6993481095176,
-      "grad_norm": 1.4453483819961548,
-      "learning_rate": 6.4478391959799e-05,
-      "loss": 5.1837,
-      "step": 35856
-    },
-    {
-      "epoch": 18.69986962190352,
-      "grad_norm": 1.5746926069259644,
-      "learning_rate": 6.447738693467336e-05,
-      "loss": 4.7999,
-      "step": 35857
-    },
-    {
-      "epoch": 18.70039113428944,
-      "grad_norm": 1.473960041999817,
-      "learning_rate": 6.447638190954774e-05,
-      "loss": 5.5455,
-      "step": 35858
-    },
-    {
-      "epoch": 18.70091264667536,
-      "grad_norm": 1.5786218643188477,
-      "learning_rate": 6.447537688442212e-05,
-      "loss": 4.5817,
-      "step": 35859
-    },
-    {
-      "epoch": 18.70143415906128,
-      "grad_norm": 1.434946894645691,
-      "learning_rate": 6.44743718592965e-05,
-      "loss": 5.5126,
-      "step": 35860
-    },
-    {
-      "epoch": 18.701955671447198,
-      "grad_norm": 1.4372460842132568,
-      "learning_rate": 6.447336683417086e-05,
-      "loss": 5.6747,
-      "step": 35861
-    },
-    {
-      "epoch": 18.702477183833118,
-      "grad_norm": 1.4418162107467651,
-      "learning_rate": 6.447236180904524e-05,
-      "loss": 5.4345,
-      "step": 35862
-    },
-    {
-      "epoch": 18.702998696219034,
-      "grad_norm": 1.539210557937622,
-      "learning_rate": 6.44713567839196e-05,
-      "loss": 4.9327,
-      "step": 35863
-    },
-    {
-      "epoch": 18.703520208604953,
-      "grad_norm": 1.581088900566101,
-      "learning_rate": 6.447035175879397e-05,
-      "loss": 5.1434,
-      "step": 35864
-    },
-    {
-      "epoch": 18.704041720990872,
-      "grad_norm": 1.559638500213623,
-      "learning_rate": 6.446934673366834e-05,
-      "loss": 5.2323,
-      "step": 35865
-    },
-    {
-      "epoch": 18.704563233376792,
-      "grad_norm": 1.4857606887817383,
-      "learning_rate": 6.446834170854271e-05,
-      "loss": 5.1399,
-      "step": 35866
-    },
-    {
-      "epoch": 18.70508474576271,
-      "grad_norm": 1.5214449167251587,
-      "learning_rate": 6.446733668341709e-05,
-      "loss": 5.093,
-      "step": 35867
-    },
-    {
-      "epoch": 18.70560625814863,
-      "grad_norm": 1.612060785293579,
-      "learning_rate": 6.446633165829146e-05,
-      "loss": 4.7051,
-      "step": 35868
-    },
-    {
-      "epoch": 18.70612777053455,
-      "grad_norm": 1.4832314252853394,
-      "learning_rate": 6.446532663316584e-05,
-      "loss": 5.4856,
-      "step": 35869
-    },
-    {
-      "epoch": 18.70664928292047,
-      "grad_norm": 1.464050054550171,
-      "learning_rate": 6.44643216080402e-05,
-      "loss": 5.5356,
-      "step": 35870
-    },
-    {
-      "epoch": 18.70717079530639,
-      "grad_norm": 1.510913610458374,
-      "learning_rate": 6.446331658291458e-05,
-      "loss": 5.4065,
-      "step": 35871
-    },
-    {
-      "epoch": 18.70769230769231,
-      "grad_norm": 1.4881433248519897,
-      "learning_rate": 6.446231155778895e-05,
-      "loss": 5.671,
-      "step": 35872
-    },
-    {
-      "epoch": 18.708213820078228,
-      "grad_norm": 1.5408648252487183,
-      "learning_rate": 6.446130653266333e-05,
-      "loss": 5.3345,
-      "step": 35873
-    },
-    {
-      "epoch": 18.708735332464148,
-      "grad_norm": 1.5621751546859741,
-      "learning_rate": 6.446030150753769e-05,
-      "loss": 5.1055,
-      "step": 35874
-    },
-    {
-      "epoch": 18.709256844850064,
-      "grad_norm": 1.6275897026062012,
-      "learning_rate": 6.445929648241207e-05,
-      "loss": 4.4156,
-      "step": 35875
-    },
-    {
-      "epoch": 18.709778357235983,
-      "grad_norm": 1.5488805770874023,
-      "learning_rate": 6.445829145728643e-05,
-      "loss": 5.2016,
-      "step": 35876
-    },
-    {
-      "epoch": 18.710299869621903,
-      "grad_norm": 1.416773796081543,
-      "learning_rate": 6.44572864321608e-05,
-      "loss": 5.5555,
-      "step": 35877
-    },
-    {
-      "epoch": 18.710821382007822,
-      "grad_norm": 1.5676122903823853,
-      "learning_rate": 6.445628140703517e-05,
-      "loss": 5.4308,
-      "step": 35878
-    },
-    {
-      "epoch": 18.71134289439374,
-      "grad_norm": 1.4382609128952026,
-      "learning_rate": 6.445527638190955e-05,
-      "loss": 4.8076,
-      "step": 35879
-    },
-    {
-      "epoch": 18.71186440677966,
-      "grad_norm": 1.58767569065094,
-      "learning_rate": 6.445427135678393e-05,
-      "loss": 5.0166,
-      "step": 35880
-    },
-    {
-      "epoch": 18.71238591916558,
-      "grad_norm": 1.406355381011963,
-      "learning_rate": 6.44532663316583e-05,
-      "loss": 5.5933,
-      "step": 35881
-    },
-    {
-      "epoch": 18.7129074315515,
-      "grad_norm": 1.4473050832748413,
-      "learning_rate": 6.445226130653267e-05,
-      "loss": 5.1367,
-      "step": 35882
-    },
-    {
-      "epoch": 18.71342894393742,
-      "grad_norm": 1.3843144178390503,
-      "learning_rate": 6.445125628140704e-05,
-      "loss": 5.4483,
-      "step": 35883
-    },
-    {
-      "epoch": 18.71395045632334,
-      "grad_norm": 1.4565273523330688,
-      "learning_rate": 6.445025125628141e-05,
-      "loss": 5.4261,
-      "step": 35884
-    },
-    {
-      "epoch": 18.714471968709258,
-      "grad_norm": 1.4990873336791992,
-      "learning_rate": 6.444924623115578e-05,
-      "loss": 5.1823,
-      "step": 35885
-    },
-    {
-      "epoch": 18.714993481095178,
-      "grad_norm": 1.5340697765350342,
-      "learning_rate": 6.444824120603016e-05,
-      "loss": 5.4962,
-      "step": 35886
-    },
-    {
-      "epoch": 18.715514993481094,
-      "grad_norm": 1.3688395023345947,
-      "learning_rate": 6.444723618090452e-05,
-      "loss": 5.1884,
-      "step": 35887
-    },
-    {
-      "epoch": 18.716036505867013,
-      "grad_norm": 1.5042369365692139,
-      "learning_rate": 6.44462311557789e-05,
-      "loss": 5.3413,
-      "step": 35888
-    },
-    {
-      "epoch": 18.716558018252933,
-      "grad_norm": 1.5028636455535889,
-      "learning_rate": 6.444522613065328e-05,
-      "loss": 4.8842,
-      "step": 35889
-    },
-    {
-      "epoch": 18.717079530638852,
-      "grad_norm": 1.5264544486999512,
-      "learning_rate": 6.444422110552765e-05,
-      "loss": 4.9118,
-      "step": 35890
-    },
-    {
-      "epoch": 18.71760104302477,
-      "grad_norm": 1.4580771923065186,
-      "learning_rate": 6.444321608040202e-05,
-      "loss": 5.2195,
-      "step": 35891
-    },
-    {
-      "epoch": 18.71812255541069,
-      "grad_norm": 1.4936691522598267,
-      "learning_rate": 6.444221105527638e-05,
-      "loss": 4.8478,
-      "step": 35892
-    },
-    {
-      "epoch": 18.71864406779661,
-      "grad_norm": 1.5331989526748657,
-      "learning_rate": 6.444120603015076e-05,
-      "loss": 5.5033,
-      "step": 35893
-    },
-    {
-      "epoch": 18.71916558018253,
-      "grad_norm": 1.5714527368545532,
-      "learning_rate": 6.444020100502512e-05,
-      "loss": 5.6474,
-      "step": 35894
-    },
-    {
-      "epoch": 18.71968709256845,
-      "grad_norm": 1.5396742820739746,
-      "learning_rate": 6.44391959798995e-05,
-      "loss": 5.3222,
-      "step": 35895
-    },
-    {
-      "epoch": 18.72020860495437,
-      "grad_norm": 1.5309150218963623,
-      "learning_rate": 6.443819095477387e-05,
-      "loss": 5.1256,
-      "step": 35896
-    },
-    {
-      "epoch": 18.72073011734029,
-      "grad_norm": 1.538632869720459,
-      "learning_rate": 6.443718592964824e-05,
-      "loss": 5.451,
-      "step": 35897
-    },
-    {
-      "epoch": 18.721251629726208,
-      "grad_norm": 1.4195797443389893,
-      "learning_rate": 6.443618090452261e-05,
-      "loss": 5.5289,
-      "step": 35898
-    },
-    {
-      "epoch": 18.721773142112124,
-      "grad_norm": 1.565274953842163,
-      "learning_rate": 6.443517587939699e-05,
-      "loss": 5.5902,
-      "step": 35899
-    },
-    {
-      "epoch": 18.722294654498043,
-      "grad_norm": 1.4483401775360107,
-      "learning_rate": 6.443417085427136e-05,
-      "loss": 5.5122,
-      "step": 35900
-    },
-    {
-      "epoch": 18.722816166883963,
-      "grad_norm": 1.4502251148223877,
-      "learning_rate": 6.443316582914574e-05,
-      "loss": 5.0233,
-      "step": 35901
-    },
-    {
-      "epoch": 18.723337679269882,
-      "grad_norm": 1.4447886943817139,
-      "learning_rate": 6.44321608040201e-05,
-      "loss": 5.5414,
-      "step": 35902
-    },
-    {
-      "epoch": 18.7238591916558,
-      "grad_norm": 1.502145528793335,
-      "learning_rate": 6.443115577889448e-05,
-      "loss": 5.1978,
-      "step": 35903
-    },
-    {
-      "epoch": 18.72438070404172,
-      "grad_norm": 1.3791937828063965,
-      "learning_rate": 6.443015075376885e-05,
-      "loss": 5.7487,
-      "step": 35904
-    },
-    {
-      "epoch": 18.72490221642764,
-      "grad_norm": 1.5792959928512573,
-      "learning_rate": 6.442914572864321e-05,
-      "loss": 5.1401,
-      "step": 35905
-    },
-    {
-      "epoch": 18.72542372881356,
-      "grad_norm": 1.5177900791168213,
-      "learning_rate": 6.442814070351759e-05,
-      "loss": 5.2445,
-      "step": 35906
-    },
-    {
-      "epoch": 18.72594524119948,
-      "grad_norm": 1.4903252124786377,
-      "learning_rate": 6.442713567839195e-05,
-      "loss": 4.7671,
-      "step": 35907
-    },
-    {
-      "epoch": 18.7264667535854,
-      "grad_norm": 1.5932835340499878,
-      "learning_rate": 6.442613065326633e-05,
-      "loss": 5.1247,
-      "step": 35908
-    },
-    {
-      "epoch": 18.72698826597132,
-      "grad_norm": 1.6513744592666626,
-      "learning_rate": 6.44251256281407e-05,
-      "loss": 5.3099,
-      "step": 35909
-    },
-    {
-      "epoch": 18.727509778357238,
-      "grad_norm": 1.5338518619537354,
-      "learning_rate": 6.442412060301507e-05,
-      "loss": 5.397,
-      "step": 35910
-    },
-    {
-      "epoch": 18.728031290743154,
-      "grad_norm": 1.580959439277649,
-      "learning_rate": 6.442311557788945e-05,
-      "loss": 4.9072,
-      "step": 35911
-    },
-    {
-      "epoch": 18.728552803129073,
-      "grad_norm": 1.5294533967971802,
-      "learning_rate": 6.442211055276383e-05,
-      "loss": 5.3437,
-      "step": 35912
-    },
-    {
-      "epoch": 18.729074315514993,
-      "grad_norm": 1.4677773714065552,
-      "learning_rate": 6.44211055276382e-05,
-      "loss": 5.2594,
-      "step": 35913
-    },
-    {
-      "epoch": 18.729595827900912,
-      "grad_norm": 1.4779075384140015,
-      "learning_rate": 6.442010050251257e-05,
-      "loss": 4.7745,
-      "step": 35914
-    },
-    {
-      "epoch": 18.73011734028683,
-      "grad_norm": 1.4289888143539429,
-      "learning_rate": 6.441909547738694e-05,
-      "loss": 5.6996,
-      "step": 35915
-    },
-    {
-      "epoch": 18.73063885267275,
-      "grad_norm": 1.40090811252594,
-      "learning_rate": 6.441809045226131e-05,
-      "loss": 5.7723,
-      "step": 35916
-    },
-    {
-      "epoch": 18.73116036505867,
-      "grad_norm": 1.579135775566101,
-      "learning_rate": 6.441708542713568e-05,
-      "loss": 5.2758,
-      "step": 35917
-    },
-    {
-      "epoch": 18.73168187744459,
-      "grad_norm": 1.5520106554031372,
-      "learning_rate": 6.441608040201004e-05,
-      "loss": 5.1556,
-      "step": 35918
-    },
-    {
-      "epoch": 18.73220338983051,
-      "grad_norm": 1.5295913219451904,
-      "learning_rate": 6.441507537688442e-05,
-      "loss": 5.3044,
-      "step": 35919
-    },
-    {
-      "epoch": 18.73272490221643,
-      "grad_norm": 1.4292582273483276,
-      "learning_rate": 6.44140703517588e-05,
-      "loss": 5.7273,
-      "step": 35920
-    },
-    {
-      "epoch": 18.73324641460235,
-      "grad_norm": 1.5562400817871094,
-      "learning_rate": 6.441306532663318e-05,
-      "loss": 5.1322,
-      "step": 35921
-    },
-    {
-      "epoch": 18.733767926988264,
-      "grad_norm": 1.413487195968628,
-      "learning_rate": 6.441206030150754e-05,
-      "loss": 5.4699,
-      "step": 35922
-    },
-    {
-      "epoch": 18.734289439374184,
-      "grad_norm": 1.5121992826461792,
-      "learning_rate": 6.441105527638192e-05,
-      "loss": 5.4812,
-      "step": 35923
-    },
-    {
-      "epoch": 18.734810951760103,
-      "grad_norm": 1.4073642492294312,
-      "learning_rate": 6.441005025125628e-05,
-      "loss": 5.3461,
-      "step": 35924
-    },
-    {
-      "epoch": 18.735332464146023,
-      "grad_norm": 1.479889154434204,
-      "learning_rate": 6.440904522613066e-05,
-      "loss": 5.448,
-      "step": 35925
-    },
-    {
-      "epoch": 18.735853976531942,
-      "grad_norm": 1.4538196325302124,
-      "learning_rate": 6.440804020100502e-05,
-      "loss": 5.5391,
-      "step": 35926
-    },
-    {
-      "epoch": 18.73637548891786,
-      "grad_norm": 1.4855526685714722,
-      "learning_rate": 6.44070351758794e-05,
-      "loss": 5.1415,
-      "step": 35927
-    },
-    {
-      "epoch": 18.73689700130378,
-      "grad_norm": 1.4763131141662598,
-      "learning_rate": 6.440603015075377e-05,
-      "loss": 5.5997,
-      "step": 35928
-    },
-    {
-      "epoch": 18.7374185136897,
-      "grad_norm": 1.4911435842514038,
-      "learning_rate": 6.440502512562814e-05,
-      "loss": 5.2742,
-      "step": 35929
-    },
-    {
-      "epoch": 18.73794002607562,
-      "grad_norm": 1.4628106355667114,
-      "learning_rate": 6.440402010050251e-05,
-      "loss": 5.4732,
-      "step": 35930
-    },
-    {
-      "epoch": 18.73846153846154,
-      "grad_norm": 1.504129409790039,
-      "learning_rate": 6.440301507537689e-05,
-      "loss": 4.6388,
-      "step": 35931
-    },
-    {
-      "epoch": 18.73898305084746,
-      "grad_norm": 1.4977718591690063,
-      "learning_rate": 6.440201005025126e-05,
-      "loss": 5.5933,
-      "step": 35932
-    },
-    {
-      "epoch": 18.73950456323338,
-      "grad_norm": 1.5340303182601929,
-      "learning_rate": 6.440100502512563e-05,
-      "loss": 5.4104,
-      "step": 35933
-    },
-    {
-      "epoch": 18.740026075619298,
-      "grad_norm": 1.5769332647323608,
-      "learning_rate": 6.440000000000001e-05,
-      "loss": 5.3708,
-      "step": 35934
-    },
-    {
-      "epoch": 18.740547588005214,
-      "grad_norm": 1.4378165006637573,
-      "learning_rate": 6.439899497487437e-05,
-      "loss": 4.9009,
-      "step": 35935
-    },
-    {
-      "epoch": 18.741069100391133,
-      "grad_norm": 1.5983855724334717,
-      "learning_rate": 6.439798994974875e-05,
-      "loss": 5.3889,
-      "step": 35936
-    },
-    {
-      "epoch": 18.741590612777053,
-      "grad_norm": 1.5478078126907349,
-      "learning_rate": 6.439698492462311e-05,
-      "loss": 5.0589,
-      "step": 35937
-    },
-    {
-      "epoch": 18.742112125162972,
-      "grad_norm": 1.4716432094573975,
-      "learning_rate": 6.439597989949749e-05,
-      "loss": 5.5704,
-      "step": 35938
-    },
-    {
-      "epoch": 18.74263363754889,
-      "grad_norm": 1.477070927619934,
-      "learning_rate": 6.439497487437186e-05,
-      "loss": 5.6062,
-      "step": 35939
-    },
-    {
-      "epoch": 18.74315514993481,
-      "grad_norm": 1.4988014698028564,
-      "learning_rate": 6.439396984924623e-05,
-      "loss": 5.7342,
-      "step": 35940
-    },
-    {
-      "epoch": 18.74367666232073,
-      "grad_norm": 1.5080150365829468,
-      "learning_rate": 6.439296482412061e-05,
-      "loss": 5.0092,
-      "step": 35941
-    },
-    {
-      "epoch": 18.74419817470665,
-      "grad_norm": 1.396349549293518,
-      "learning_rate": 6.439195979899499e-05,
-      "loss": 5.3552,
-      "step": 35942
-    },
-    {
-      "epoch": 18.74471968709257,
-      "grad_norm": 1.493714451789856,
-      "learning_rate": 6.439095477386935e-05,
-      "loss": 5.1735,
-      "step": 35943
-    },
-    {
-      "epoch": 18.74524119947849,
-      "grad_norm": 1.4889450073242188,
-      "learning_rate": 6.438994974874372e-05,
-      "loss": 5.518,
-      "step": 35944
-    },
-    {
-      "epoch": 18.74576271186441,
-      "grad_norm": 1.3912254571914673,
-      "learning_rate": 6.43889447236181e-05,
-      "loss": 5.7902,
-      "step": 35945
-    },
-    {
-      "epoch": 18.746284224250324,
-      "grad_norm": 1.4617716073989868,
-      "learning_rate": 6.438793969849246e-05,
-      "loss": 5.747,
-      "step": 35946
-    },
-    {
-      "epoch": 18.746805736636244,
-      "grad_norm": 1.4233908653259277,
-      "learning_rate": 6.438693467336684e-05,
-      "loss": 5.6954,
-      "step": 35947
-    },
-    {
-      "epoch": 18.747327249022163,
-      "grad_norm": 1.5063837766647339,
-      "learning_rate": 6.43859296482412e-05,
-      "loss": 5.0712,
-      "step": 35948
-    },
-    {
-      "epoch": 18.747848761408083,
-      "grad_norm": 1.3843729496002197,
-      "learning_rate": 6.438492462311558e-05,
-      "loss": 5.5407,
-      "step": 35949
-    },
-    {
-      "epoch": 18.748370273794002,
-      "grad_norm": 1.4871999025344849,
-      "learning_rate": 6.438391959798994e-05,
-      "loss": 5.5751,
-      "step": 35950
-    },
-    {
-      "epoch": 18.74889178617992,
-      "grad_norm": 1.5002293586730957,
-      "learning_rate": 6.438291457286432e-05,
-      "loss": 5.4059,
-      "step": 35951
-    },
-    {
-      "epoch": 18.74941329856584,
-      "grad_norm": 1.4471940994262695,
-      "learning_rate": 6.43819095477387e-05,
-      "loss": 5.2221,
-      "step": 35952
-    },
-    {
-      "epoch": 18.74993481095176,
-      "grad_norm": 1.4587483406066895,
-      "learning_rate": 6.438090452261308e-05,
-      "loss": 5.3665,
-      "step": 35953
-    },
-    {
-      "epoch": 18.75045632333768,
-      "grad_norm": 1.454163908958435,
-      "learning_rate": 6.437989949748744e-05,
-      "loss": 5.5564,
-      "step": 35954
-    },
-    {
-      "epoch": 18.7509778357236,
-      "grad_norm": 1.5216068029403687,
-      "learning_rate": 6.437889447236182e-05,
-      "loss": 5.4751,
-      "step": 35955
-    },
-    {
-      "epoch": 18.75149934810952,
-      "grad_norm": 1.7101479768753052,
-      "learning_rate": 6.437788944723618e-05,
-      "loss": 4.3614,
-      "step": 35956
-    },
-    {
-      "epoch": 18.75202086049544,
-      "grad_norm": 1.5059953927993774,
-      "learning_rate": 6.437688442211055e-05,
-      "loss": 5.4161,
-      "step": 35957
-    },
-    {
-      "epoch": 18.752542372881354,
-      "grad_norm": 1.4646583795547485,
-      "learning_rate": 6.437587939698493e-05,
-      "loss": 5.1433,
-      "step": 35958
-    },
-    {
-      "epoch": 18.753063885267274,
-      "grad_norm": 1.5954256057739258,
-      "learning_rate": 6.437487437185929e-05,
-      "loss": 4.8411,
-      "step": 35959
-    },
-    {
-      "epoch": 18.753585397653193,
-      "grad_norm": 1.462733268737793,
-      "learning_rate": 6.437386934673367e-05,
-      "loss": 5.5399,
-      "step": 35960
-    },
-    {
-      "epoch": 18.754106910039113,
-      "grad_norm": 1.3803651332855225,
-      "learning_rate": 6.437286432160805e-05,
-      "loss": 5.6655,
-      "step": 35961
-    },
-    {
-      "epoch": 18.754628422425032,
-      "grad_norm": 1.4766440391540527,
-      "learning_rate": 6.437185929648242e-05,
-      "loss": 5.3898,
-      "step": 35962
-    },
-    {
-      "epoch": 18.75514993481095,
-      "grad_norm": 1.5125455856323242,
-      "learning_rate": 6.437085427135679e-05,
-      "loss": 5.4823,
-      "step": 35963
-    },
-    {
-      "epoch": 18.75567144719687,
-      "grad_norm": 1.4845635890960693,
-      "learning_rate": 6.436984924623117e-05,
-      "loss": 5.2042,
-      "step": 35964
-    },
-    {
-      "epoch": 18.75619295958279,
-      "grad_norm": 1.5359193086624146,
-      "learning_rate": 6.436884422110553e-05,
-      "loss": 4.9357,
-      "step": 35965
-    },
-    {
-      "epoch": 18.75671447196871,
-      "grad_norm": 1.5510168075561523,
-      "learning_rate": 6.436783919597991e-05,
-      "loss": 4.8557,
-      "step": 35966
-    },
-    {
-      "epoch": 18.75723598435463,
-      "grad_norm": 1.52571439743042,
-      "learning_rate": 6.436683417085427e-05,
-      "loss": 5.4275,
-      "step": 35967
-    },
-    {
-      "epoch": 18.75775749674055,
-      "grad_norm": 1.5027728080749512,
-      "learning_rate": 6.436582914572865e-05,
-      "loss": 5.2151,
-      "step": 35968
-    },
-    {
-      "epoch": 18.75827900912647,
-      "grad_norm": 1.4956024885177612,
-      "learning_rate": 6.436482412060301e-05,
-      "loss": 5.7419,
-      "step": 35969
-    },
-    {
-      "epoch": 18.758800521512384,
-      "grad_norm": 1.4039554595947266,
-      "learning_rate": 6.436381909547738e-05,
-      "loss": 5.3327,
-      "step": 35970
-    },
-    {
-      "epoch": 18.759322033898304,
-      "grad_norm": 1.4389750957489014,
-      "learning_rate": 6.436281407035176e-05,
-      "loss": 5.5641,
-      "step": 35971
-    },
-    {
-      "epoch": 18.759843546284223,
-      "grad_norm": 1.479677438735962,
-      "learning_rate": 6.436180904522613e-05,
-      "loss": 5.4728,
-      "step": 35972
-    },
-    {
-      "epoch": 18.760365058670143,
-      "grad_norm": 1.440555453300476,
-      "learning_rate": 6.436080402010051e-05,
-      "loss": 5.5631,
-      "step": 35973
-    },
-    {
-      "epoch": 18.760886571056062,
-      "grad_norm": 1.4581302404403687,
-      "learning_rate": 6.435979899497488e-05,
-      "loss": 5.2504,
-      "step": 35974
-    },
-    {
-      "epoch": 18.76140808344198,
-      "grad_norm": 1.4555846452713013,
-      "learning_rate": 6.435879396984925e-05,
-      "loss": 5.5736,
-      "step": 35975
-    },
-    {
-      "epoch": 18.7619295958279,
-      "grad_norm": 1.464991569519043,
-      "learning_rate": 6.435778894472362e-05,
-      "loss": 5.7302,
-      "step": 35976
-    },
-    {
-      "epoch": 18.76245110821382,
-      "grad_norm": 1.4912514686584473,
-      "learning_rate": 6.4356783919598e-05,
-      "loss": 4.9978,
-      "step": 35977
-    },
-    {
-      "epoch": 18.76297262059974,
-      "grad_norm": 1.5738859176635742,
-      "learning_rate": 6.435577889447236e-05,
-      "loss": 5.1236,
-      "step": 35978
-    },
-    {
-      "epoch": 18.76349413298566,
-      "grad_norm": 1.416872262954712,
-      "learning_rate": 6.435477386934674e-05,
-      "loss": 5.732,
-      "step": 35979
-    },
-    {
-      "epoch": 18.76401564537158,
-      "grad_norm": 1.5594245195388794,
-      "learning_rate": 6.43537688442211e-05,
-      "loss": 5.7775,
-      "step": 35980
-    },
-    {
-      "epoch": 18.7645371577575,
-      "grad_norm": 1.5753251314163208,
-      "learning_rate": 6.435276381909548e-05,
-      "loss": 5.057,
-      "step": 35981
-    },
-    {
-      "epoch": 18.765058670143414,
-      "grad_norm": 1.4689850807189941,
-      "learning_rate": 6.435175879396986e-05,
-      "loss": 5.5669,
-      "step": 35982
-    },
-    {
-      "epoch": 18.765580182529334,
-      "grad_norm": 1.50339937210083,
-      "learning_rate": 6.435075376884424e-05,
-      "loss": 5.4985,
-      "step": 35983
-    },
-    {
-      "epoch": 18.766101694915253,
-      "grad_norm": 1.5625276565551758,
-      "learning_rate": 6.43497487437186e-05,
-      "loss": 4.8275,
-      "step": 35984
-    },
-    {
-      "epoch": 18.766623207301173,
-      "grad_norm": 1.6479352712631226,
-      "learning_rate": 6.434874371859296e-05,
-      "loss": 5.0595,
-      "step": 35985
-    },
-    {
-      "epoch": 18.767144719687092,
-      "grad_norm": 1.5480881929397583,
-      "learning_rate": 6.434773869346734e-05,
-      "loss": 5.4488,
-      "step": 35986
-    },
-    {
-      "epoch": 18.76766623207301,
-      "grad_norm": 1.5219255685806274,
-      "learning_rate": 6.43467336683417e-05,
-      "loss": 5.1787,
-      "step": 35987
-    },
-    {
-      "epoch": 18.76818774445893,
-      "grad_norm": 1.4788928031921387,
-      "learning_rate": 6.434572864321608e-05,
-      "loss": 5.2386,
-      "step": 35988
-    },
-    {
-      "epoch": 18.76870925684485,
-      "grad_norm": 1.417788028717041,
-      "learning_rate": 6.434472361809045e-05,
-      "loss": 5.6073,
-      "step": 35989
-    },
-    {
-      "epoch": 18.76923076923077,
-      "grad_norm": 1.4561620950698853,
-      "learning_rate": 6.434371859296483e-05,
-      "loss": 5.3993,
-      "step": 35990
-    },
-    {
-      "epoch": 18.76975228161669,
-      "grad_norm": 1.5165493488311768,
-      "learning_rate": 6.434271356783919e-05,
-      "loss": 5.3941,
-      "step": 35991
-    },
-    {
-      "epoch": 18.77027379400261,
-      "grad_norm": 1.5922296047210693,
-      "learning_rate": 6.434170854271357e-05,
-      "loss": 5.0538,
-      "step": 35992
-    },
-    {
-      "epoch": 18.77079530638853,
-      "grad_norm": 1.4400945901870728,
-      "learning_rate": 6.434070351758795e-05,
-      "loss": 5.8271,
-      "step": 35993
-    },
-    {
-      "epoch": 18.771316818774444,
-      "grad_norm": 1.5726474523544312,
-      "learning_rate": 6.433969849246232e-05,
-      "loss": 4.7619,
-      "step": 35994
-    },
-    {
-      "epoch": 18.771838331160364,
-      "grad_norm": 1.5216251611709595,
-      "learning_rate": 6.433869346733669e-05,
-      "loss": 5.2877,
-      "step": 35995
-    },
-    {
-      "epoch": 18.772359843546283,
-      "grad_norm": 1.5327459573745728,
-      "learning_rate": 6.433768844221107e-05,
-      "loss": 5.3404,
-      "step": 35996
-    },
-    {
-      "epoch": 18.772881355932203,
-      "grad_norm": 1.4364293813705444,
-      "learning_rate": 6.433668341708543e-05,
-      "loss": 5.3564,
-      "step": 35997
-    },
-    {
-      "epoch": 18.773402868318122,
-      "grad_norm": 1.5556209087371826,
-      "learning_rate": 6.43356783919598e-05,
-      "loss": 5.2787,
-      "step": 35998
-    },
-    {
-      "epoch": 18.77392438070404,
-      "grad_norm": 1.5219001770019531,
-      "learning_rate": 6.433467336683417e-05,
-      "loss": 5.1653,
-      "step": 35999
-    },
-    {
-      "epoch": 18.77444589308996,
-      "grad_norm": 1.5169719457626343,
-      "learning_rate": 6.433366834170854e-05,
-      "loss": 5.6007,
-      "step": 36000
-    },
-    {
-      "epoch": 18.77496740547588,
-      "grad_norm": 1.446241855621338,
-      "learning_rate": 6.433266331658291e-05,
-      "loss": 5.5105,
-      "step": 36001
-    },
-    {
-      "epoch": 18.7754889178618,
-      "grad_norm": 1.365213394165039,
-      "learning_rate": 6.433165829145729e-05,
-      "loss": 5.625,
-      "step": 36002
-    },
-    {
-      "epoch": 18.77601043024772,
-      "grad_norm": 1.5360474586486816,
-      "learning_rate": 6.433065326633167e-05,
-      "loss": 5.0326,
-      "step": 36003
-    },
-    {
-      "epoch": 18.77653194263364,
-      "grad_norm": 1.5679458379745483,
-      "learning_rate": 6.432964824120603e-05,
-      "loss": 5.1289,
-      "step": 36004
-    },
-    {
-      "epoch": 18.777053455019555,
-      "grad_norm": 1.4371522665023804,
-      "learning_rate": 6.432864321608041e-05,
-      "loss": 5.1662,
-      "step": 36005
-    },
-    {
-      "epoch": 18.777574967405474,
-      "grad_norm": 1.5886493921279907,
-      "learning_rate": 6.432763819095478e-05,
-      "loss": 5.4969,
-      "step": 36006
-    },
-    {
-      "epoch": 18.778096479791394,
-      "grad_norm": 1.5960646867752075,
-      "learning_rate": 6.432663316582915e-05,
-      "loss": 4.8358,
-      "step": 36007
-    },
-    {
-      "epoch": 18.778617992177313,
-      "grad_norm": 1.4446533918380737,
-      "learning_rate": 6.432562814070352e-05,
-      "loss": 5.2533,
-      "step": 36008
-    },
-    {
-      "epoch": 18.779139504563233,
-      "grad_norm": 1.350625991821289,
-      "learning_rate": 6.43246231155779e-05,
-      "loss": 5.6095,
-      "step": 36009
-    },
-    {
-      "epoch": 18.779661016949152,
-      "grad_norm": 1.4180002212524414,
-      "learning_rate": 6.432361809045226e-05,
-      "loss": 5.4002,
-      "step": 36010
-    },
-    {
-      "epoch": 18.78018252933507,
-      "grad_norm": 1.3934980630874634,
-      "learning_rate": 6.432261306532663e-05,
-      "loss": 5.5321,
-      "step": 36011
-    },
-    {
-      "epoch": 18.78070404172099,
-      "grad_norm": 1.456923007965088,
-      "learning_rate": 6.4321608040201e-05,
-      "loss": 5.8476,
-      "step": 36012
-    },
-    {
-      "epoch": 18.78122555410691,
-      "grad_norm": 1.4901783466339111,
-      "learning_rate": 6.432060301507538e-05,
-      "loss": 4.9783,
-      "step": 36013
-    },
-    {
-      "epoch": 18.78174706649283,
-      "grad_norm": 1.5256454944610596,
-      "learning_rate": 6.431959798994976e-05,
-      "loss": 4.8575,
-      "step": 36014
-    },
-    {
-      "epoch": 18.78226857887875,
-      "grad_norm": 1.4159520864486694,
-      "learning_rate": 6.431859296482412e-05,
-      "loss": 5.7019,
-      "step": 36015
-    },
-    {
-      "epoch": 18.78279009126467,
-      "grad_norm": 1.4676216840744019,
-      "learning_rate": 6.43175879396985e-05,
-      "loss": 4.9824,
-      "step": 36016
-    },
-    {
-      "epoch": 18.78331160365059,
-      "grad_norm": 1.5069823265075684,
-      "learning_rate": 6.431658291457287e-05,
-      "loss": 5.3904,
-      "step": 36017
-    },
-    {
-      "epoch": 18.783833116036504,
-      "grad_norm": 1.574047327041626,
-      "learning_rate": 6.431557788944724e-05,
-      "loss": 5.4438,
-      "step": 36018
-    },
-    {
-      "epoch": 18.784354628422424,
-      "grad_norm": 1.4844332933425903,
-      "learning_rate": 6.431457286432161e-05,
-      "loss": 5.1596,
-      "step": 36019
-    },
-    {
-      "epoch": 18.784876140808343,
-      "grad_norm": 1.6502926349639893,
-      "learning_rate": 6.431356783919599e-05,
-      "loss": 5.2086,
-      "step": 36020
-    },
-    {
-      "epoch": 18.785397653194263,
-      "grad_norm": 1.4404512643814087,
-      "learning_rate": 6.431256281407035e-05,
-      "loss": 5.0398,
-      "step": 36021
-    },
-    {
-      "epoch": 18.785919165580182,
-      "grad_norm": 1.5548290014266968,
-      "learning_rate": 6.431155778894473e-05,
-      "loss": 5.4709,
-      "step": 36022
-    },
-    {
-      "epoch": 18.7864406779661,
-      "grad_norm": 1.6028720140457153,
-      "learning_rate": 6.43105527638191e-05,
-      "loss": 5.17,
-      "step": 36023
-    },
-    {
-      "epoch": 18.78696219035202,
-      "grad_norm": 1.516111969947815,
-      "learning_rate": 6.430954773869347e-05,
-      "loss": 4.9369,
-      "step": 36024
-    },
-    {
-      "epoch": 18.78748370273794,
-      "grad_norm": 1.4384026527404785,
-      "learning_rate": 6.430854271356785e-05,
-      "loss": 4.9671,
-      "step": 36025
-    },
-    {
-      "epoch": 18.78800521512386,
-      "grad_norm": 1.4424917697906494,
-      "learning_rate": 6.430753768844221e-05,
-      "loss": 5.4508,
-      "step": 36026
-    },
-    {
-      "epoch": 18.78852672750978,
-      "grad_norm": 1.4626296758651733,
-      "learning_rate": 6.430653266331659e-05,
-      "loss": 5.5546,
-      "step": 36027
-    },
-    {
-      "epoch": 18.7890482398957,
-      "grad_norm": 1.5384074449539185,
-      "learning_rate": 6.430552763819095e-05,
-      "loss": 5.2985,
-      "step": 36028
-    },
-    {
-      "epoch": 18.789569752281615,
-      "grad_norm": 1.4901114702224731,
-      "learning_rate": 6.430452261306533e-05,
-      "loss": 5.698,
-      "step": 36029
-    },
-    {
-      "epoch": 18.790091264667534,
-      "grad_norm": 1.6116293668746948,
-      "learning_rate": 6.43035175879397e-05,
-      "loss": 4.6158,
-      "step": 36030
-    },
-    {
-      "epoch": 18.790612777053454,
-      "grad_norm": 1.6068745851516724,
-      "learning_rate": 6.430251256281407e-05,
-      "loss": 5.5172,
-      "step": 36031
-    },
-    {
-      "epoch": 18.791134289439373,
-      "grad_norm": 1.544826626777649,
-      "learning_rate": 6.430150753768844e-05,
-      "loss": 5.3338,
-      "step": 36032
-    },
-    {
-      "epoch": 18.791655801825293,
-      "grad_norm": 1.5216360092163086,
-      "learning_rate": 6.430050251256282e-05,
-      "loss": 5.1879,
-      "step": 36033
-    },
-    {
-      "epoch": 18.792177314211212,
-      "grad_norm": 1.494059443473816,
-      "learning_rate": 6.42994974874372e-05,
-      "loss": 5.557,
-      "step": 36034
-    },
-    {
-      "epoch": 18.79269882659713,
-      "grad_norm": 1.464355707168579,
-      "learning_rate": 6.429849246231157e-05,
-      "loss": 5.1137,
-      "step": 36035
-    },
-    {
-      "epoch": 18.79322033898305,
-      "grad_norm": 1.4487351179122925,
-      "learning_rate": 6.429748743718594e-05,
-      "loss": 4.7595,
-      "step": 36036
-    },
-    {
-      "epoch": 18.79374185136897,
-      "grad_norm": 1.3909642696380615,
-      "learning_rate": 6.42964824120603e-05,
-      "loss": 5.5904,
-      "step": 36037
-    },
-    {
-      "epoch": 18.79426336375489,
-      "grad_norm": 1.5639530420303345,
-      "learning_rate": 6.429547738693468e-05,
-      "loss": 5.5759,
-      "step": 36038
-    },
-    {
-      "epoch": 18.79478487614081,
-      "grad_norm": 1.4806147813796997,
-      "learning_rate": 6.429447236180904e-05,
-      "loss": 5.4042,
-      "step": 36039
-    },
-    {
-      "epoch": 18.79530638852673,
-      "grad_norm": 1.5979598760604858,
-      "learning_rate": 6.429346733668342e-05,
-      "loss": 4.8099,
-      "step": 36040
-    },
-    {
-      "epoch": 18.795827900912645,
-      "grad_norm": 1.509364128112793,
-      "learning_rate": 6.429246231155778e-05,
-      "loss": 4.9721,
-      "step": 36041
-    },
-    {
-      "epoch": 18.796349413298564,
-      "grad_norm": 1.4226834774017334,
-      "learning_rate": 6.429145728643216e-05,
-      "loss": 4.9521,
-      "step": 36042
-    },
-    {
-      "epoch": 18.796870925684484,
-      "grad_norm": 1.4375622272491455,
-      "learning_rate": 6.429045226130654e-05,
-      "loss": 5.7821,
-      "step": 36043
-    },
-    {
-      "epoch": 18.797392438070403,
-      "grad_norm": 1.4903947114944458,
-      "learning_rate": 6.428944723618092e-05,
-      "loss": 5.1339,
-      "step": 36044
-    },
-    {
-      "epoch": 18.797913950456323,
-      "grad_norm": 1.6116632223129272,
-      "learning_rate": 6.428844221105528e-05,
-      "loss": 4.8203,
-      "step": 36045
-    },
-    {
-      "epoch": 18.798435462842242,
-      "grad_norm": 1.4433597326278687,
-      "learning_rate": 6.428743718592966e-05,
-      "loss": 5.5789,
-      "step": 36046
-    },
-    {
-      "epoch": 18.798956975228162,
-      "grad_norm": 1.5817699432373047,
-      "learning_rate": 6.428643216080402e-05,
-      "loss": 5.3466,
-      "step": 36047
-    },
-    {
-      "epoch": 18.79947848761408,
-      "grad_norm": 1.5329012870788574,
-      "learning_rate": 6.42854271356784e-05,
-      "loss": 5.064,
-      "step": 36048
-    },
-    {
-      "epoch": 18.8,
-      "grad_norm": 1.5491595268249512,
-      "learning_rate": 6.428442211055277e-05,
-      "loss": 4.861,
-      "step": 36049
-    },
-    {
-      "epoch": 18.80052151238592,
-      "grad_norm": 1.5053445100784302,
-      "learning_rate": 6.428341708542713e-05,
-      "loss": 5.2523,
-      "step": 36050
-    },
-    {
-      "epoch": 18.80104302477184,
-      "grad_norm": 1.5213111639022827,
-      "learning_rate": 6.428241206030151e-05,
-      "loss": 5.5073,
-      "step": 36051
-    },
-    {
-      "epoch": 18.80156453715776,
-      "grad_norm": 1.483224868774414,
-      "learning_rate": 6.428140703517587e-05,
-      "loss": 5.2766,
-      "step": 36052
-    },
-    {
-      "epoch": 18.802086049543675,
-      "grad_norm": 1.4543591737747192,
-      "learning_rate": 6.428040201005025e-05,
-      "loss": 5.4459,
-      "step": 36053
-    },
-    {
-      "epoch": 18.802607561929594,
-      "grad_norm": 1.4117722511291504,
-      "learning_rate": 6.427939698492463e-05,
-      "loss": 5.3106,
-      "step": 36054
-    },
-    {
-      "epoch": 18.803129074315514,
-      "grad_norm": 1.439598798751831,
-      "learning_rate": 6.4278391959799e-05,
-      "loss": 5.0341,
-      "step": 36055
-    },
-    {
-      "epoch": 18.803650586701433,
-      "grad_norm": 1.470990777015686,
-      "learning_rate": 6.427738693467337e-05,
-      "loss": 5.1052,
-      "step": 36056
-    },
-    {
-      "epoch": 18.804172099087353,
-      "grad_norm": 1.5455282926559448,
-      "learning_rate": 6.427638190954775e-05,
-      "loss": 4.9904,
-      "step": 36057
-    },
-    {
-      "epoch": 18.804693611473272,
-      "grad_norm": 1.4508600234985352,
-      "learning_rate": 6.427537688442211e-05,
-      "loss": 5.5365,
-      "step": 36058
-    },
-    {
-      "epoch": 18.805215123859192,
-      "grad_norm": 1.4391283988952637,
-      "learning_rate": 6.427437185929649e-05,
-      "loss": 5.6732,
-      "step": 36059
-    },
-    {
-      "epoch": 18.80573663624511,
-      "grad_norm": 1.4928991794586182,
-      "learning_rate": 6.427336683417085e-05,
-      "loss": 5.267,
-      "step": 36060
-    },
-    {
-      "epoch": 18.80625814863103,
-      "grad_norm": 1.504171371459961,
-      "learning_rate": 6.427236180904523e-05,
-      "loss": 4.806,
-      "step": 36061
-    },
-    {
-      "epoch": 18.80677966101695,
-      "grad_norm": 1.5151803493499756,
-      "learning_rate": 6.42713567839196e-05,
-      "loss": 5.2609,
-      "step": 36062
-    },
-    {
-      "epoch": 18.80730117340287,
-      "grad_norm": 1.5344252586364746,
-      "learning_rate": 6.427035175879397e-05,
-      "loss": 5.4522,
-      "step": 36063
-    },
-    {
-      "epoch": 18.80782268578879,
-      "grad_norm": 1.4511191844940186,
-      "learning_rate": 6.426934673366835e-05,
-      "loss": 5.552,
-      "step": 36064
-    },
-    {
-      "epoch": 18.808344198174705,
-      "grad_norm": 1.5820181369781494,
-      "learning_rate": 6.426834170854272e-05,
-      "loss": 5.2848,
-      "step": 36065
-    },
-    {
-      "epoch": 18.808865710560625,
-      "grad_norm": 1.4871257543563843,
-      "learning_rate": 6.42673366834171e-05,
-      "loss": 5.2409,
-      "step": 36066
-    },
-    {
-      "epoch": 18.809387222946544,
-      "grad_norm": 1.5056977272033691,
-      "learning_rate": 6.426633165829146e-05,
-      "loss": 4.9768,
-      "step": 36067
-    },
-    {
-      "epoch": 18.809908735332463,
-      "grad_norm": 1.5940700769424438,
-      "learning_rate": 6.426532663316584e-05,
-      "loss": 5.254,
-      "step": 36068
-    },
-    {
-      "epoch": 18.810430247718383,
-      "grad_norm": 1.492523193359375,
-      "learning_rate": 6.42643216080402e-05,
-      "loss": 4.9955,
-      "step": 36069
-    },
-    {
-      "epoch": 18.810951760104302,
-      "grad_norm": 1.476087212562561,
-      "learning_rate": 6.426331658291458e-05,
-      "loss": 5.2455,
-      "step": 36070
-    },
-    {
-      "epoch": 18.811473272490222,
-      "grad_norm": 1.4331799745559692,
-      "learning_rate": 6.426231155778894e-05,
-      "loss": 5.4012,
-      "step": 36071
-    },
-    {
-      "epoch": 18.81199478487614,
-      "grad_norm": 1.444035530090332,
-      "learning_rate": 6.426130653266332e-05,
-      "loss": 5.2477,
-      "step": 36072
-    },
-    {
-      "epoch": 18.81251629726206,
-      "grad_norm": 1.4727575778961182,
-      "learning_rate": 6.426030150753768e-05,
-      "loss": 5.4795,
-      "step": 36073
-    },
-    {
-      "epoch": 18.81303780964798,
-      "grad_norm": 1.4322048425674438,
-      "learning_rate": 6.425929648241206e-05,
-      "loss": 5.2227,
-      "step": 36074
-    },
-    {
-      "epoch": 18.8135593220339,
-      "grad_norm": 1.4474053382873535,
-      "learning_rate": 6.425829145728644e-05,
-      "loss": 5.2086,
-      "step": 36075
-    },
-    {
-      "epoch": 18.81408083441982,
-      "grad_norm": 1.4845854043960571,
-      "learning_rate": 6.425728643216082e-05,
-      "loss": 5.5118,
-      "step": 36076
-    },
-    {
-      "epoch": 18.814602346805735,
-      "grad_norm": 1.385987639427185,
-      "learning_rate": 6.425628140703518e-05,
-      "loss": 5.3209,
-      "step": 36077
-    },
-    {
-      "epoch": 18.815123859191655,
-      "grad_norm": 1.4590829610824585,
-      "learning_rate": 6.425527638190955e-05,
-      "loss": 5.5627,
-      "step": 36078
-    },
-    {
-      "epoch": 18.815645371577574,
-      "grad_norm": 1.51578688621521,
-      "learning_rate": 6.425427135678392e-05,
-      "loss": 5.4238,
-      "step": 36079
-    },
-    {
-      "epoch": 18.816166883963493,
-      "grad_norm": 1.431358814239502,
-      "learning_rate": 6.425326633165829e-05,
-      "loss": 5.2687,
-      "step": 36080
-    },
-    {
-      "epoch": 18.816688396349413,
-      "grad_norm": 1.5030032396316528,
-      "learning_rate": 6.425226130653267e-05,
-      "loss": 5.2143,
-      "step": 36081
-    },
-    {
-      "epoch": 18.817209908735332,
-      "grad_norm": 1.5048463344573975,
-      "learning_rate": 6.425125628140703e-05,
-      "loss": 5.2971,
-      "step": 36082
-    },
-    {
-      "epoch": 18.817731421121252,
-      "grad_norm": 1.5764142274856567,
-      "learning_rate": 6.425025125628141e-05,
-      "loss": 4.8038,
-      "step": 36083
-    },
-    {
-      "epoch": 18.81825293350717,
-      "grad_norm": 1.581382155418396,
-      "learning_rate": 6.424924623115577e-05,
-      "loss": 5.2385,
-      "step": 36084
-    },
-    {
-      "epoch": 18.81877444589309,
-      "grad_norm": 1.636052131652832,
-      "learning_rate": 6.424824120603015e-05,
-      "loss": 5.0311,
-      "step": 36085
-    },
-    {
-      "epoch": 18.81929595827901,
-      "grad_norm": 1.6845051050186157,
-      "learning_rate": 6.424723618090453e-05,
-      "loss": 4.889,
-      "step": 36086
-    },
-    {
-      "epoch": 18.81981747066493,
-      "grad_norm": 1.5255126953125,
-      "learning_rate": 6.42462311557789e-05,
-      "loss": 5.1711,
-      "step": 36087
-    },
-    {
-      "epoch": 18.820338983050846,
-      "grad_norm": 1.5924474000930786,
-      "learning_rate": 6.424522613065327e-05,
-      "loss": 5.2888,
-      "step": 36088
-    },
-    {
-      "epoch": 18.820860495436765,
-      "grad_norm": 1.471899390220642,
-      "learning_rate": 6.424422110552765e-05,
-      "loss": 5.2278,
-      "step": 36089
-    },
-    {
-      "epoch": 18.821382007822685,
-      "grad_norm": 1.5269067287445068,
-      "learning_rate": 6.424321608040201e-05,
-      "loss": 5.2542,
-      "step": 36090
-    },
-    {
-      "epoch": 18.821903520208604,
-      "grad_norm": 1.5013633966445923,
-      "learning_rate": 6.424221105527638e-05,
-      "loss": 4.661,
-      "step": 36091
-    },
-    {
-      "epoch": 18.822425032594523,
-      "grad_norm": 1.4967750310897827,
-      "learning_rate": 6.424120603015075e-05,
-      "loss": 4.9938,
-      "step": 36092
-    },
-    {
-      "epoch": 18.822946544980443,
-      "grad_norm": 1.6618824005126953,
-      "learning_rate": 6.424020100502512e-05,
-      "loss": 5.3534,
-      "step": 36093
-    },
-    {
-      "epoch": 18.823468057366362,
-      "grad_norm": 1.5356123447418213,
-      "learning_rate": 6.42391959798995e-05,
-      "loss": 5.2957,
-      "step": 36094
-    },
-    {
-      "epoch": 18.823989569752282,
-      "grad_norm": 1.6126197576522827,
-      "learning_rate": 6.423819095477387e-05,
-      "loss": 5.3641,
-      "step": 36095
-    },
-    {
-      "epoch": 18.8245110821382,
-      "grad_norm": 1.4369739294052124,
-      "learning_rate": 6.423718592964825e-05,
-      "loss": 5.4883,
-      "step": 36096
-    },
-    {
-      "epoch": 18.82503259452412,
-      "grad_norm": 1.4441215991973877,
-      "learning_rate": 6.423618090452262e-05,
-      "loss": 4.685,
-      "step": 36097
-    },
-    {
-      "epoch": 18.82555410691004,
-      "grad_norm": 1.465701937675476,
-      "learning_rate": 6.4235175879397e-05,
-      "loss": 5.3209,
-      "step": 36098
-    },
-    {
-      "epoch": 18.82607561929596,
-      "grad_norm": 1.5968279838562012,
-      "learning_rate": 6.423417085427136e-05,
-      "loss": 4.8652,
-      "step": 36099
-    },
-    {
-      "epoch": 18.82659713168188,
-      "grad_norm": 1.4692115783691406,
-      "learning_rate": 6.423316582914574e-05,
-      "loss": 5.377,
-      "step": 36100
-    },
-    {
-      "epoch": 18.827118644067795,
-      "grad_norm": 1.4457939863204956,
-      "learning_rate": 6.42321608040201e-05,
-      "loss": 5.2752,
-      "step": 36101
-    },
-    {
-      "epoch": 18.827640156453715,
-      "grad_norm": 1.5108660459518433,
-      "learning_rate": 6.423115577889448e-05,
-      "loss": 5.1735,
-      "step": 36102
-    },
-    {
-      "epoch": 18.828161668839634,
-      "grad_norm": 1.5377298593521118,
-      "learning_rate": 6.423015075376884e-05,
-      "loss": 5.2679,
-      "step": 36103
-    },
-    {
-      "epoch": 18.828683181225554,
-      "grad_norm": 1.3418734073638916,
-      "learning_rate": 6.422914572864321e-05,
-      "loss": 5.0887,
-      "step": 36104
-    },
-    {
-      "epoch": 18.829204693611473,
-      "grad_norm": 1.4896987676620483,
-      "learning_rate": 6.422814070351759e-05,
-      "loss": 5.2941,
-      "step": 36105
-    },
-    {
-      "epoch": 18.829726205997392,
-      "grad_norm": 1.5635124444961548,
-      "learning_rate": 6.422713567839196e-05,
-      "loss": 5.5966,
-      "step": 36106
-    },
-    {
-      "epoch": 18.830247718383312,
-      "grad_norm": Infinity,
-      "learning_rate": 6.422713567839196e-05,
-      "loss": 4.8514,
-      "step": 36107
-    },
-    {
-      "epoch": 18.83076923076923,
-      "grad_norm": 1.451920986175537,
-      "learning_rate": 6.422613065326634e-05,
-      "loss": 5.3455,
-      "step": 36108
-    },
-    {
-      "epoch": 18.83129074315515,
-      "grad_norm": 1.5483399629592896,
-      "learning_rate": 6.42251256281407e-05,
-      "loss": 5.2777,
-      "step": 36109
-    },
-    {
-      "epoch": 18.83181225554107,
-      "grad_norm": 1.5404943227767944,
-      "learning_rate": 6.422412060301508e-05,
-      "loss": 5.5924,
-      "step": 36110
-    },
-    {
-      "epoch": 18.83233376792699,
-      "grad_norm": 1.4740252494812012,
-      "learning_rate": 6.422311557788945e-05,
-      "loss": 5.6482,
-      "step": 36111
-    },
-    {
-      "epoch": 18.832855280312906,
-      "grad_norm": 1.510061264038086,
-      "learning_rate": 6.422211055276383e-05,
-      "loss": 5.3455,
-      "step": 36112
-    },
-    {
-      "epoch": 18.833376792698825,
-      "grad_norm": 1.5111699104309082,
-      "learning_rate": 6.422110552763819e-05,
-      "loss": 5.4569,
-      "step": 36113
-    },
-    {
-      "epoch": 18.833898305084745,
-      "grad_norm": 1.4755043983459473,
-      "learning_rate": 6.422010050251257e-05,
-      "loss": 5.2555,
-      "step": 36114
-    },
-    {
-      "epoch": 18.834419817470664,
-      "grad_norm": 1.4287505149841309,
-      "learning_rate": 6.421909547738693e-05,
-      "loss": 5.45,
-      "step": 36115
-    },
-    {
-      "epoch": 18.834941329856584,
-      "grad_norm": 1.5825756788253784,
-      "learning_rate": 6.421809045226131e-05,
-      "loss": 5.3062,
-      "step": 36116
-    },
-    {
-      "epoch": 18.835462842242503,
-      "grad_norm": 1.4397374391555786,
-      "learning_rate": 6.421708542713569e-05,
-      "loss": 5.4148,
-      "step": 36117
-    },
-    {
-      "epoch": 18.835984354628422,
-      "grad_norm": 1.4076414108276367,
-      "learning_rate": 6.421608040201005e-05,
-      "loss": 5.4608,
-      "step": 36118
-    },
-    {
-      "epoch": 18.836505867014342,
-      "grad_norm": 1.4894334077835083,
-      "learning_rate": 6.421507537688443e-05,
-      "loss": 5.6938,
-      "step": 36119
-    },
-    {
-      "epoch": 18.83702737940026,
-      "grad_norm": 1.4657782316207886,
-      "learning_rate": 6.42140703517588e-05,
-      "loss": 5.497,
-      "step": 36120
-    },
-    {
-      "epoch": 18.83754889178618,
-      "grad_norm": 1.4420604705810547,
-      "learning_rate": 6.421306532663317e-05,
-      "loss": 5.3888,
-      "step": 36121
-    },
-    {
-      "epoch": 18.8380704041721,
-      "grad_norm": 1.3516170978546143,
-      "learning_rate": 6.421206030150754e-05,
-      "loss": 4.6801,
-      "step": 36122
-    },
-    {
-      "epoch": 18.83859191655802,
-      "grad_norm": 1.532957673072815,
-      "learning_rate": 6.421105527638191e-05,
-      "loss": 5.2078,
-      "step": 36123
-    },
-    {
-      "epoch": 18.839113428943936,
-      "grad_norm": 1.4825836420059204,
-      "learning_rate": 6.421005025125628e-05,
-      "loss": 5.3477,
-      "step": 36124
-    },
-    {
-      "epoch": 18.839634941329855,
-      "grad_norm": 1.5326720476150513,
-      "learning_rate": 6.420904522613066e-05,
-      "loss": 4.6331,
-      "step": 36125
-    },
-    {
-      "epoch": 18.840156453715775,
-      "grad_norm": 1.4762625694274902,
-      "learning_rate": 6.420804020100502e-05,
-      "loss": 5.4156,
-      "step": 36126
-    },
-    {
-      "epoch": 18.840677966101694,
-      "grad_norm": 1.5373018980026245,
-      "learning_rate": 6.42070351758794e-05,
-      "loss": 5.1059,
-      "step": 36127
-    },
-    {
-      "epoch": 18.841199478487614,
-      "grad_norm": 1.528296709060669,
-      "learning_rate": 6.420603015075378e-05,
-      "loss": 5.1478,
-      "step": 36128
-    },
-    {
-      "epoch": 18.841720990873533,
-      "grad_norm": 1.5118441581726074,
-      "learning_rate": 6.420502512562815e-05,
-      "loss": 5.1437,
-      "step": 36129
-    },
-    {
-      "epoch": 18.842242503259452,
-      "grad_norm": 1.5167953968048096,
-      "learning_rate": 6.420402010050252e-05,
-      "loss": 5.3557,
-      "step": 36130
-    },
-    {
-      "epoch": 18.842764015645372,
-      "grad_norm": 1.506500244140625,
-      "learning_rate": 6.420301507537688e-05,
-      "loss": 5.4865,
-      "step": 36131
-    },
-    {
-      "epoch": 18.84328552803129,
-      "grad_norm": 1.5104295015335083,
-      "learning_rate": 6.420201005025126e-05,
-      "loss": 5.3155,
-      "step": 36132
-    },
-    {
-      "epoch": 18.84380704041721,
-      "grad_norm": 1.5090841054916382,
-      "learning_rate": 6.420100502512562e-05,
-      "loss": 5.2028,
-      "step": 36133
-    },
-    {
-      "epoch": 18.84432855280313,
-      "grad_norm": 1.527465581893921,
-      "learning_rate": 6.42e-05,
-      "loss": 5.3877,
-      "step": 36134
-    },
-    {
-      "epoch": 18.84485006518905,
-      "grad_norm": 1.4443304538726807,
-      "learning_rate": 6.419899497487437e-05,
-      "loss": 5.7053,
-      "step": 36135
-    },
-    {
-      "epoch": 18.845371577574966,
-      "grad_norm": 1.480575680732727,
-      "learning_rate": 6.419798994974874e-05,
-      "loss": 5.336,
-      "step": 36136
-    },
-    {
-      "epoch": 18.845893089960885,
-      "grad_norm": 1.5768316984176636,
-      "learning_rate": 6.419698492462312e-05,
-      "loss": 4.5486,
-      "step": 36137
-    },
-    {
-      "epoch": 18.846414602346805,
-      "grad_norm": 1.4727017879486084,
-      "learning_rate": 6.41959798994975e-05,
-      "loss": 5.2689,
-      "step": 36138
-    },
-    {
-      "epoch": 18.846936114732724,
-      "grad_norm": 1.5554426908493042,
-      "learning_rate": 6.419497487437186e-05,
-      "loss": 4.9502,
-      "step": 36139
-    },
-    {
-      "epoch": 18.847457627118644,
-      "grad_norm": 1.6397885084152222,
-      "learning_rate": 6.419396984924624e-05,
-      "loss": 5.5564,
-      "step": 36140
-    },
-    {
-      "epoch": 18.847979139504563,
-      "grad_norm": 1.5764665603637695,
-      "learning_rate": 6.41929648241206e-05,
-      "loss": 5.7077,
-      "step": 36141
-    },
-    {
-      "epoch": 18.848500651890483,
-      "grad_norm": 1.5532081127166748,
-      "learning_rate": 6.419195979899498e-05,
-      "loss": 5.6569,
-      "step": 36142
-    },
-    {
-      "epoch": 18.849022164276402,
-      "grad_norm": 1.4353117942810059,
-      "learning_rate": 6.419095477386935e-05,
-      "loss": 5.3956,
-      "step": 36143
-    },
-    {
-      "epoch": 18.84954367666232,
-      "grad_norm": 1.6036063432693481,
-      "learning_rate": 6.418994974874371e-05,
-      "loss": 5.0505,
-      "step": 36144
-    },
-    {
-      "epoch": 18.85006518904824,
-      "grad_norm": 1.6070334911346436,
-      "learning_rate": 6.418894472361809e-05,
-      "loss": 5.1452,
-      "step": 36145
-    },
-    {
-      "epoch": 18.85058670143416,
-      "grad_norm": 1.4581546783447266,
-      "learning_rate": 6.418793969849245e-05,
-      "loss": 5.3766,
-      "step": 36146
-    },
-    {
-      "epoch": 18.85110821382008,
-      "grad_norm": 1.5928704738616943,
-      "learning_rate": 6.418693467336683e-05,
-      "loss": 5.3131,
-      "step": 36147
-    },
-    {
-      "epoch": 18.851629726205996,
-      "grad_norm": 1.4039537906646729,
-      "learning_rate": 6.418592964824121e-05,
-      "loss": 5.1189,
-      "step": 36148
-    },
-    {
-      "epoch": 18.852151238591915,
-      "grad_norm": 1.4385722875595093,
-      "learning_rate": 6.418492462311559e-05,
-      "loss": 5.3115,
-      "step": 36149
-    },
-    {
-      "epoch": 18.852672750977835,
-      "grad_norm": 1.549341082572937,
-      "learning_rate": 6.418391959798995e-05,
-      "loss": 5.1702,
-      "step": 36150
-    },
-    {
-      "epoch": 18.853194263363754,
-      "grad_norm": 1.5574125051498413,
-      "learning_rate": 6.418291457286433e-05,
-      "loss": 5.4016,
-      "step": 36151
-    },
-    {
-      "epoch": 18.853715775749674,
-      "grad_norm": 1.4301645755767822,
-      "learning_rate": 6.41819095477387e-05,
-      "loss": 5.3148,
-      "step": 36152
-    },
-    {
-      "epoch": 18.854237288135593,
-      "grad_norm": 1.4813175201416016,
-      "learning_rate": 6.418090452261307e-05,
-      "loss": 5.4461,
-      "step": 36153
-    },
-    {
-      "epoch": 18.854758800521513,
-      "grad_norm": 1.4868955612182617,
-      "learning_rate": 6.417989949748744e-05,
-      "loss": 5.5258,
-      "step": 36154
-    },
-    {
-      "epoch": 18.855280312907432,
-      "grad_norm": 1.41548490524292,
-      "learning_rate": 6.417889447236181e-05,
-      "loss": 5.7512,
-      "step": 36155
-    },
-    {
-      "epoch": 18.85580182529335,
-      "grad_norm": 1.651811122894287,
-      "learning_rate": 6.417788944723618e-05,
-      "loss": 4.6088,
-      "step": 36156
-    },
-    {
-      "epoch": 18.85632333767927,
-      "grad_norm": 1.407164216041565,
-      "learning_rate": 6.417688442211056e-05,
-      "loss": 5.4805,
-      "step": 36157
-    },
-    {
-      "epoch": 18.85684485006519,
-      "grad_norm": 1.5518611669540405,
-      "learning_rate": 6.417587939698493e-05,
-      "loss": 5.1078,
-      "step": 36158
-    },
-    {
-      "epoch": 18.85736636245111,
-      "grad_norm": 1.5467054843902588,
-      "learning_rate": 6.41748743718593e-05,
-      "loss": 5.4464,
-      "step": 36159
-    },
-    {
-      "epoch": 18.857887874837026,
-      "grad_norm": 1.4400272369384766,
-      "learning_rate": 6.417386934673368e-05,
-      "loss": 5.5942,
-      "step": 36160
-    },
-    {
-      "epoch": 18.858409387222945,
-      "grad_norm": 1.6352183818817139,
-      "learning_rate": 6.417286432160804e-05,
-      "loss": 5.4098,
-      "step": 36161
-    },
-    {
-      "epoch": 18.858930899608865,
-      "grad_norm": 1.4745895862579346,
-      "learning_rate": 6.417185929648242e-05,
-      "loss": 5.2622,
-      "step": 36162
-    },
-    {
-      "epoch": 18.859452411994784,
-      "grad_norm": 1.5266833305358887,
-      "learning_rate": 6.417085427135678e-05,
-      "loss": 5.373,
-      "step": 36163
-    },
-    {
-      "epoch": 18.859973924380704,
-      "grad_norm": 1.5257980823516846,
-      "learning_rate": 6.416984924623116e-05,
-      "loss": 5.6442,
-      "step": 36164
-    },
-    {
-      "epoch": 18.860495436766623,
-      "grad_norm": 1.5030465126037598,
-      "learning_rate": 6.416884422110552e-05,
-      "loss": 4.8075,
-      "step": 36165
-    },
-    {
-      "epoch": 18.861016949152543,
-      "grad_norm": 1.5006550550460815,
-      "learning_rate": 6.41678391959799e-05,
-      "loss": 5.7362,
-      "step": 36166
-    },
-    {
-      "epoch": 18.861538461538462,
-      "grad_norm": 1.541820764541626,
-      "learning_rate": 6.416683417085427e-05,
-      "loss": 5.1161,
-      "step": 36167
-    },
-    {
-      "epoch": 18.86205997392438,
-      "grad_norm": 1.418144941329956,
-      "learning_rate": 6.416582914572864e-05,
-      "loss": 5.4494,
-      "step": 36168
-    },
-    {
-      "epoch": 18.8625814863103,
-      "grad_norm": 1.4381521940231323,
-      "learning_rate": 6.416482412060302e-05,
-      "loss": 5.2734,
-      "step": 36169
-    },
-    {
-      "epoch": 18.86310299869622,
-      "grad_norm": 1.519903540611267,
-      "learning_rate": 6.41638190954774e-05,
-      "loss": 5.4159,
-      "step": 36170
-    },
-    {
-      "epoch": 18.86362451108214,
-      "grad_norm": 1.5013196468353271,
-      "learning_rate": 6.416281407035176e-05,
-      "loss": 5.121,
-      "step": 36171
-    },
-    {
-      "epoch": 18.864146023468056,
-      "grad_norm": 1.4469938278198242,
-      "learning_rate": 6.416180904522613e-05,
-      "loss": 5.3576,
-      "step": 36172
-    },
-    {
-      "epoch": 18.864667535853975,
-      "grad_norm": 1.504755973815918,
-      "learning_rate": 6.416080402010051e-05,
-      "loss": 5.0637,
-      "step": 36173
-    },
-    {
-      "epoch": 18.865189048239895,
-      "grad_norm": 1.5609660148620605,
-      "learning_rate": 6.415979899497487e-05,
-      "loss": 5.1176,
-      "step": 36174
-    },
-    {
-      "epoch": 18.865710560625814,
-      "grad_norm": 1.4792709350585938,
-      "learning_rate": 6.415879396984925e-05,
-      "loss": 5.3452,
-      "step": 36175
-    },
-    {
-      "epoch": 18.866232073011734,
-      "grad_norm": 1.6429461240768433,
-      "learning_rate": 6.415778894472361e-05,
-      "loss": 4.4522,
-      "step": 36176
-    },
-    {
-      "epoch": 18.866753585397653,
-      "grad_norm": 1.5945658683776855,
-      "learning_rate": 6.415678391959799e-05,
-      "loss": 4.6088,
-      "step": 36177
-    },
-    {
-      "epoch": 18.867275097783573,
-      "grad_norm": 1.464270830154419,
-      "learning_rate": 6.415577889447237e-05,
-      "loss": 5.6644,
-      "step": 36178
-    },
-    {
-      "epoch": 18.867796610169492,
-      "grad_norm": 1.5146056413650513,
-      "learning_rate": 6.415477386934675e-05,
-      "loss": 5.625,
-      "step": 36179
-    },
-    {
-      "epoch": 18.86831812255541,
-      "grad_norm": 1.5702866315841675,
-      "learning_rate": 6.415376884422111e-05,
-      "loss": 4.7964,
-      "step": 36180
-    },
-    {
-      "epoch": 18.86883963494133,
-      "grad_norm": 1.4289476871490479,
-      "learning_rate": 6.415276381909549e-05,
-      "loss": 5.6681,
-      "step": 36181
-    },
-    {
-      "epoch": 18.86936114732725,
-      "grad_norm": 1.6191656589508057,
-      "learning_rate": 6.415175879396985e-05,
-      "loss": 4.7698,
-      "step": 36182
-    },
-    {
-      "epoch": 18.86988265971317,
-      "grad_norm": 1.5618919134140015,
-      "learning_rate": 6.415075376884423e-05,
-      "loss": 5.1574,
-      "step": 36183
-    },
-    {
-      "epoch": 18.870404172099086,
-      "grad_norm": 1.555629014968872,
-      "learning_rate": 6.41497487437186e-05,
-      "loss": 5.3518,
-      "step": 36184
-    },
-    {
-      "epoch": 18.870925684485005,
-      "grad_norm": 1.428525686264038,
-      "learning_rate": 6.414874371859296e-05,
-      "loss": 5.4144,
-      "step": 36185
-    },
-    {
-      "epoch": 18.871447196870925,
-      "grad_norm": 1.4993728399276733,
-      "learning_rate": 6.414773869346734e-05,
-      "loss": 5.3761,
-      "step": 36186
-    },
-    {
-      "epoch": 18.871968709256844,
-      "grad_norm": 1.5075191259384155,
-      "learning_rate": 6.41467336683417e-05,
-      "loss": 5.0582,
-      "step": 36187
-    },
-    {
-      "epoch": 18.872490221642764,
-      "grad_norm": 1.4090582132339478,
-      "learning_rate": 6.414572864321608e-05,
-      "loss": 5.2219,
-      "step": 36188
-    },
-    {
-      "epoch": 18.873011734028683,
-      "grad_norm": 1.5127609968185425,
-      "learning_rate": 6.414472361809046e-05,
-      "loss": 4.7753,
-      "step": 36189
-    },
-    {
-      "epoch": 18.873533246414603,
-      "grad_norm": 1.4577314853668213,
-      "learning_rate": 6.414371859296484e-05,
-      "loss": 5.3963,
-      "step": 36190
-    },
-    {
-      "epoch": 18.874054758800522,
-      "grad_norm": 1.683051586151123,
-      "learning_rate": 6.41427135678392e-05,
-      "loss": 4.5575,
-      "step": 36191
-    },
-    {
-      "epoch": 18.87457627118644,
-      "grad_norm": 1.4989440441131592,
-      "learning_rate": 6.414170854271358e-05,
-      "loss": 5.295,
-      "step": 36192
-    },
-    {
-      "epoch": 18.87509778357236,
-      "grad_norm": 1.4751557111740112,
-      "learning_rate": 6.414070351758794e-05,
-      "loss": 5.5287,
-      "step": 36193
-    },
-    {
-      "epoch": 18.87561929595828,
-      "grad_norm": 1.4696446657180786,
-      "learning_rate": 6.413969849246232e-05,
-      "loss": 5.0422,
-      "step": 36194
-    },
-    {
-      "epoch": 18.876140808344196,
-      "grad_norm": 1.6558433771133423,
-      "learning_rate": 6.413869346733668e-05,
-      "loss": 4.8848,
-      "step": 36195
-    },
-    {
-      "epoch": 18.876662320730116,
-      "grad_norm": 1.3356479406356812,
-      "learning_rate": 6.413768844221106e-05,
-      "loss": 5.3599,
-      "step": 36196
-    },
-    {
-      "epoch": 18.877183833116035,
-      "grad_norm": 1.4148725271224976,
-      "learning_rate": 6.413668341708543e-05,
-      "loss": 5.1478,
-      "step": 36197
-    },
-    {
-      "epoch": 18.877705345501955,
-      "grad_norm": 1.5477231740951538,
-      "learning_rate": 6.41356783919598e-05,
-      "loss": 5.4894,
-      "step": 36198
-    },
-    {
-      "epoch": 18.878226857887874,
-      "grad_norm": 1.4769586324691772,
-      "learning_rate": 6.413467336683418e-05,
-      "loss": 5.5393,
-      "step": 36199
-    },
-    {
-      "epoch": 18.878748370273794,
-      "grad_norm": 1.5485038757324219,
-      "learning_rate": 6.413366834170855e-05,
-      "loss": 5.4837,
-      "step": 36200
-    },
-    {
-      "epoch": 18.879269882659713,
-      "grad_norm": 1.4943690299987793,
-      "learning_rate": 6.413266331658292e-05,
-      "loss": 5.6402,
-      "step": 36201
-    },
-    {
-      "epoch": 18.879791395045633,
-      "grad_norm": 1.426095724105835,
-      "learning_rate": 6.413165829145729e-05,
-      "loss": 5.2068,
-      "step": 36202
-    },
-    {
-      "epoch": 18.880312907431552,
-      "grad_norm": 1.410951852798462,
-      "learning_rate": 6.413065326633167e-05,
-      "loss": 5.4858,
-      "step": 36203
-    },
-    {
-      "epoch": 18.88083441981747,
-      "grad_norm": 1.4840108156204224,
-      "learning_rate": 6.412964824120603e-05,
-      "loss": 5.1904,
-      "step": 36204
-    },
-    {
-      "epoch": 18.88135593220339,
-      "grad_norm": 1.5329428911209106,
-      "learning_rate": 6.412864321608041e-05,
-      "loss": 4.9698,
-      "step": 36205
-    },
-    {
-      "epoch": 18.88187744458931,
-      "grad_norm": 1.5627620220184326,
-      "learning_rate": 6.412763819095477e-05,
-      "loss": 4.9359,
-      "step": 36206
-    },
-    {
-      "epoch": 18.88239895697523,
-      "grad_norm": 1.5360904932022095,
-      "learning_rate": 6.412663316582915e-05,
-      "loss": 4.9533,
-      "step": 36207
-    },
-    {
-      "epoch": 18.882920469361146,
-      "grad_norm": 1.4812527894973755,
-      "learning_rate": 6.412562814070351e-05,
-      "loss": 5.069,
-      "step": 36208
-    },
-    {
-      "epoch": 18.883441981747065,
-      "grad_norm": 1.6092174053192139,
-      "learning_rate": 6.412462311557789e-05,
-      "loss": 4.9909,
-      "step": 36209
-    },
-    {
-      "epoch": 18.883963494132985,
-      "grad_norm": 1.415519118309021,
-      "learning_rate": 6.412361809045227e-05,
-      "loss": 5.6325,
-      "step": 36210
-    },
-    {
-      "epoch": 18.884485006518904,
-      "grad_norm": 1.4367152452468872,
-      "learning_rate": 6.412261306532663e-05,
-      "loss": 5.0477,
-      "step": 36211
-    },
-    {
-      "epoch": 18.885006518904824,
-      "grad_norm": 1.4186406135559082,
-      "learning_rate": 6.412160804020101e-05,
-      "loss": 5.6373,
-      "step": 36212
-    },
-    {
-      "epoch": 18.885528031290743,
-      "grad_norm": 1.3370003700256348,
-      "learning_rate": 6.412060301507538e-05,
-      "loss": 5.0403,
-      "step": 36213
-    },
-    {
-      "epoch": 18.886049543676663,
-      "grad_norm": 1.3751076459884644,
-      "learning_rate": 6.411959798994975e-05,
-      "loss": 5.5715,
-      "step": 36214
-    },
-    {
-      "epoch": 18.886571056062582,
-      "grad_norm": 1.5535491704940796,
-      "learning_rate": 6.411859296482412e-05,
-      "loss": 5.1477,
-      "step": 36215
-    },
-    {
-      "epoch": 18.8870925684485,
-      "grad_norm": 1.5548862218856812,
-      "learning_rate": 6.41175879396985e-05,
-      "loss": 5.4689,
-      "step": 36216
-    },
-    {
-      "epoch": 18.88761408083442,
-      "grad_norm": 1.6026164293289185,
-      "learning_rate": 6.411658291457286e-05,
-      "loss": 4.7455,
-      "step": 36217
-    },
-    {
-      "epoch": 18.88813559322034,
-      "grad_norm": 1.4965481758117676,
-      "learning_rate": 6.411557788944724e-05,
-      "loss": 5.4677,
-      "step": 36218
-    },
-    {
-      "epoch": 18.888657105606256,
-      "grad_norm": 1.517593502998352,
-      "learning_rate": 6.411457286432162e-05,
-      "loss": 5.47,
-      "step": 36219
-    },
-    {
-      "epoch": 18.889178617992176,
-      "grad_norm": 1.476999044418335,
-      "learning_rate": 6.4113567839196e-05,
-      "loss": 5.4333,
-      "step": 36220
-    },
-    {
-      "epoch": 18.889700130378095,
-      "grad_norm": 1.4837355613708496,
-      "learning_rate": 6.411256281407036e-05,
-      "loss": 5.6582,
-      "step": 36221
-    },
-    {
-      "epoch": 18.890221642764015,
-      "grad_norm": 1.4419198036193848,
-      "learning_rate": 6.411155778894474e-05,
-      "loss": 5.5284,
-      "step": 36222
-    },
-    {
-      "epoch": 18.890743155149934,
-      "grad_norm": 1.5367094278335571,
-      "learning_rate": 6.41105527638191e-05,
-      "loss": 4.9759,
-      "step": 36223
-    },
-    {
-      "epoch": 18.891264667535854,
-      "grad_norm": 1.3293428421020508,
-      "learning_rate": 6.410954773869346e-05,
-      "loss": 4.8429,
-      "step": 36224
-    },
-    {
-      "epoch": 18.891786179921773,
-      "grad_norm": 1.4388452768325806,
-      "learning_rate": 6.410854271356784e-05,
-      "loss": 5.3419,
-      "step": 36225
-    },
-    {
-      "epoch": 18.892307692307693,
-      "grad_norm": 1.537064552307129,
-      "learning_rate": 6.41075376884422e-05,
-      "loss": 5.3593,
-      "step": 36226
-    },
-    {
-      "epoch": 18.892829204693612,
-      "grad_norm": 1.5382177829742432,
-      "learning_rate": 6.410653266331658e-05,
-      "loss": 5.3,
-      "step": 36227
-    },
-    {
-      "epoch": 18.89335071707953,
-      "grad_norm": 1.4849672317504883,
-      "learning_rate": 6.410552763819095e-05,
-      "loss": 5.032,
-      "step": 36228
-    },
-    {
-      "epoch": 18.89387222946545,
-      "grad_norm": 1.4590810537338257,
-      "learning_rate": 6.410452261306533e-05,
-      "loss": 5.278,
-      "step": 36229
-    },
-    {
-      "epoch": 18.89439374185137,
-      "grad_norm": 1.5812088251113892,
-      "learning_rate": 6.41035175879397e-05,
-      "loss": 5.8198,
-      "step": 36230
-    },
-    {
-      "epoch": 18.894915254237286,
-      "grad_norm": 1.4040464162826538,
-      "learning_rate": 6.410251256281408e-05,
-      "loss": 4.9985,
-      "step": 36231
-    },
-    {
-      "epoch": 18.895436766623206,
-      "grad_norm": 1.4660000801086426,
-      "learning_rate": 6.410150753768845e-05,
-      "loss": 5.6376,
-      "step": 36232
-    },
-    {
-      "epoch": 18.895958279009125,
-      "grad_norm": 1.5645114183425903,
-      "learning_rate": 6.410050251256282e-05,
-      "loss": 5.267,
-      "step": 36233
-    },
-    {
-      "epoch": 18.896479791395045,
-      "grad_norm": 1.4926010370254517,
-      "learning_rate": 6.409949748743719e-05,
-      "loss": 5.2456,
-      "step": 36234
-    },
-    {
-      "epoch": 18.897001303780964,
-      "grad_norm": 1.6270456314086914,
-      "learning_rate": 6.409849246231157e-05,
-      "loss": 4.8949,
-      "step": 36235
-    },
-    {
-      "epoch": 18.897522816166884,
-      "grad_norm": 1.4939253330230713,
-      "learning_rate": 6.409748743718593e-05,
-      "loss": 4.7594,
-      "step": 36236
-    },
-    {
-      "epoch": 18.898044328552803,
-      "grad_norm": 1.540641188621521,
-      "learning_rate": 6.40964824120603e-05,
-      "loss": 4.6584,
-      "step": 36237
-    },
-    {
-      "epoch": 18.898565840938723,
-      "grad_norm": 1.4230870008468628,
-      "learning_rate": 6.409547738693467e-05,
-      "loss": 5.6161,
-      "step": 36238
-    },
-    {
-      "epoch": 18.899087353324642,
-      "grad_norm": 1.5657404661178589,
-      "learning_rate": 6.409447236180904e-05,
-      "loss": 4.9062,
-      "step": 36239
-    },
-    {
-      "epoch": 18.89960886571056,
-      "grad_norm": 1.5106126070022583,
-      "learning_rate": 6.409346733668341e-05,
-      "loss": 5.5243,
-      "step": 36240
-    },
-    {
-      "epoch": 18.90013037809648,
-      "grad_norm": 1.3716099262237549,
-      "learning_rate": 6.409246231155779e-05,
-      "loss": 5.5547,
-      "step": 36241
-    },
-    {
-      "epoch": 18.9006518904824,
-      "grad_norm": 1.449914574623108,
-      "learning_rate": 6.409145728643217e-05,
-      "loss": 5.3415,
-      "step": 36242
-    },
-    {
-      "epoch": 18.901173402868316,
-      "grad_norm": 1.4893356561660767,
-      "learning_rate": 6.409045226130653e-05,
-      "loss": 5.4049,
-      "step": 36243
-    },
-    {
-      "epoch": 18.901694915254236,
-      "grad_norm": 1.4949487447738647,
-      "learning_rate": 6.408944723618091e-05,
-      "loss": 5.4026,
-      "step": 36244
-    },
-    {
-      "epoch": 18.902216427640155,
-      "grad_norm": 1.4770277738571167,
-      "learning_rate": 6.408844221105528e-05,
-      "loss": 5.0551,
-      "step": 36245
-    },
-    {
-      "epoch": 18.902737940026075,
-      "grad_norm": 1.5279767513275146,
-      "learning_rate": 6.408743718592965e-05,
-      "loss": 5.3989,
-      "step": 36246
-    },
-    {
-      "epoch": 18.903259452411994,
-      "grad_norm": 1.39360773563385,
-      "learning_rate": 6.408643216080402e-05,
-      "loss": 4.9144,
-      "step": 36247
-    },
-    {
-      "epoch": 18.903780964797914,
-      "grad_norm": 1.6055200099945068,
-      "learning_rate": 6.40854271356784e-05,
-      "loss": 4.7782,
-      "step": 36248
-    },
-    {
-      "epoch": 18.904302477183833,
-      "grad_norm": 1.4155393838882446,
-      "learning_rate": 6.408442211055276e-05,
-      "loss": 5.2885,
-      "step": 36249
-    },
-    {
-      "epoch": 18.904823989569753,
-      "grad_norm": 1.534401774406433,
-      "learning_rate": 6.408341708542714e-05,
-      "loss": 5.2177,
-      "step": 36250
-    },
-    {
-      "epoch": 18.905345501955672,
-      "grad_norm": 1.4786614179611206,
-      "learning_rate": 6.408241206030152e-05,
-      "loss": 5.0644,
-      "step": 36251
-    },
-    {
-      "epoch": 18.90586701434159,
-      "grad_norm": 1.5742942094802856,
-      "learning_rate": 6.408140703517588e-05,
-      "loss": 5.2134,
-      "step": 36252
-    },
-    {
-      "epoch": 18.90638852672751,
-      "grad_norm": 1.5724858045578003,
-      "learning_rate": 6.408040201005026e-05,
-      "loss": 5.0606,
-      "step": 36253
-    },
-    {
-      "epoch": 18.90691003911343,
-      "grad_norm": 1.4771459102630615,
-      "learning_rate": 6.407939698492462e-05,
-      "loss": 5.5542,
-      "step": 36254
-    },
-    {
-      "epoch": 18.907431551499347,
-      "grad_norm": 1.5088170766830444,
-      "learning_rate": 6.4078391959799e-05,
-      "loss": 5.0269,
-      "step": 36255
-    },
-    {
-      "epoch": 18.907953063885266,
-      "grad_norm": 1.5115199089050293,
-      "learning_rate": 6.407738693467337e-05,
-      "loss": 5.3521,
-      "step": 36256
-    },
-    {
-      "epoch": 18.908474576271185,
-      "grad_norm": 1.5129362344741821,
-      "learning_rate": 6.407638190954774e-05,
-      "loss": 4.9043,
-      "step": 36257
-    },
-    {
-      "epoch": 18.908996088657105,
-      "grad_norm": 1.5595279932022095,
-      "learning_rate": 6.407537688442211e-05,
-      "loss": 5.4909,
-      "step": 36258
-    },
-    {
-      "epoch": 18.909517601043024,
-      "grad_norm": 1.4718765020370483,
-      "learning_rate": 6.407437185929649e-05,
-      "loss": 5.0456,
-      "step": 36259
-    },
-    {
-      "epoch": 18.910039113428944,
-      "grad_norm": 1.6125346422195435,
-      "learning_rate": 6.407336683417085e-05,
-      "loss": 5.1327,
-      "step": 36260
-    },
-    {
-      "epoch": 18.910560625814863,
-      "grad_norm": 1.5639050006866455,
-      "learning_rate": 6.407236180904523e-05,
-      "loss": 5.3801,
-      "step": 36261
-    },
-    {
-      "epoch": 18.911082138200783,
-      "grad_norm": 1.6114286184310913,
-      "learning_rate": 6.40713567839196e-05,
-      "loss": 4.9951,
-      "step": 36262
-    },
-    {
-      "epoch": 18.911603650586702,
-      "grad_norm": 1.4626898765563965,
-      "learning_rate": 6.407035175879398e-05,
-      "loss": 5.3983,
-      "step": 36263
-    },
-    {
-      "epoch": 18.91212516297262,
-      "grad_norm": 1.4688057899475098,
-      "learning_rate": 6.406934673366835e-05,
-      "loss": 5.3434,
-      "step": 36264
-    },
-    {
-      "epoch": 18.91264667535854,
-      "grad_norm": 1.415566325187683,
-      "learning_rate": 6.406834170854271e-05,
-      "loss": 5.7499,
-      "step": 36265
-    },
-    {
-      "epoch": 18.91316818774446,
-      "grad_norm": 1.4452229738235474,
-      "learning_rate": 6.406733668341709e-05,
-      "loss": 5.4272,
-      "step": 36266
-    },
-    {
-      "epoch": 18.913689700130377,
-      "grad_norm": 1.5191586017608643,
-      "learning_rate": 6.406633165829145e-05,
-      "loss": 5.1583,
-      "step": 36267
-    },
-    {
-      "epoch": 18.914211212516296,
-      "grad_norm": 1.381970763206482,
-      "learning_rate": 6.406532663316583e-05,
-      "loss": 5.4643,
-      "step": 36268
-    },
-    {
-      "epoch": 18.914732724902215,
-      "grad_norm": 1.5015459060668945,
-      "learning_rate": 6.40643216080402e-05,
-      "loss": 5.3138,
-      "step": 36269
-    },
-    {
-      "epoch": 18.915254237288135,
-      "grad_norm": 1.5229072570800781,
-      "learning_rate": 6.406331658291457e-05,
-      "loss": 4.8448,
-      "step": 36270
-    },
-    {
-      "epoch": 18.915775749674054,
-      "grad_norm": 1.4963020086288452,
-      "learning_rate": 6.406231155778895e-05,
-      "loss": 5.2093,
-      "step": 36271
-    },
-    {
-      "epoch": 18.916297262059974,
-      "grad_norm": 1.4080429077148438,
-      "learning_rate": 6.406130653266333e-05,
-      "loss": 5.5149,
-      "step": 36272
-    },
-    {
-      "epoch": 18.916818774445893,
-      "grad_norm": 1.5292623043060303,
-      "learning_rate": 6.40603015075377e-05,
-      "loss": 5.2252,
-      "step": 36273
-    },
-    {
-      "epoch": 18.917340286831813,
-      "grad_norm": 1.3280973434448242,
-      "learning_rate": 6.405929648241207e-05,
-      "loss": 5.8189,
-      "step": 36274
-    },
-    {
-      "epoch": 18.917861799217732,
-      "grad_norm": 1.5941418409347534,
-      "learning_rate": 6.405829145728644e-05,
-      "loss": 4.955,
-      "step": 36275
-    },
-    {
-      "epoch": 18.91838331160365,
-      "grad_norm": 1.5420382022857666,
-      "learning_rate": 6.405728643216081e-05,
-      "loss": 5.2073,
-      "step": 36276
-    },
-    {
-      "epoch": 18.91890482398957,
-      "grad_norm": 1.4519355297088623,
-      "learning_rate": 6.405628140703518e-05,
-      "loss": 5.6307,
-      "step": 36277
-    },
-    {
-      "epoch": 18.919426336375487,
-      "grad_norm": 1.5104440450668335,
-      "learning_rate": 6.405527638190954e-05,
-      "loss": 5.6372,
-      "step": 36278
-    },
-    {
-      "epoch": 18.919947848761407,
-      "grad_norm": 1.470607042312622,
-      "learning_rate": 6.405427135678392e-05,
-      "loss": 5.0959,
-      "step": 36279
-    },
-    {
-      "epoch": 18.920469361147326,
-      "grad_norm": 1.5599162578582764,
-      "learning_rate": 6.405326633165828e-05,
-      "loss": 4.8659,
-      "step": 36280
-    },
-    {
-      "epoch": 18.920990873533245,
-      "grad_norm": 1.583740234375,
-      "learning_rate": 6.405226130653266e-05,
-      "loss": 5.3151,
-      "step": 36281
-    },
-    {
-      "epoch": 18.921512385919165,
-      "grad_norm": 1.5363425016403198,
-      "learning_rate": 6.405125628140704e-05,
-      "loss": 5.0949,
-      "step": 36282
-    },
-    {
-      "epoch": 18.922033898305084,
-      "grad_norm": 1.3608624935150146,
-      "learning_rate": 6.405025125628142e-05,
-      "loss": 5.348,
-      "step": 36283
-    },
-    {
-      "epoch": 18.922555410691004,
-      "grad_norm": 1.427709937095642,
-      "learning_rate": 6.404924623115578e-05,
-      "loss": 5.6384,
-      "step": 36284
-    },
-    {
-      "epoch": 18.923076923076923,
-      "grad_norm": 1.5290457010269165,
-      "learning_rate": 6.404824120603016e-05,
-      "loss": 5.4431,
-      "step": 36285
-    },
-    {
-      "epoch": 18.923598435462843,
-      "grad_norm": 1.4528067111968994,
-      "learning_rate": 6.404723618090452e-05,
-      "loss": 5.2626,
-      "step": 36286
-    },
-    {
-      "epoch": 18.924119947848762,
-      "grad_norm": 1.4299062490463257,
-      "learning_rate": 6.40462311557789e-05,
-      "loss": 5.53,
-      "step": 36287
-    },
-    {
-      "epoch": 18.92464146023468,
-      "grad_norm": 1.513229489326477,
-      "learning_rate": 6.404522613065327e-05,
-      "loss": 5.1313,
-      "step": 36288
-    },
-    {
-      "epoch": 18.9251629726206,
-      "grad_norm": 1.4293532371520996,
-      "learning_rate": 6.404422110552764e-05,
-      "loss": 5.2942,
-      "step": 36289
-    },
-    {
-      "epoch": 18.92568448500652,
-      "grad_norm": 1.5070669651031494,
-      "learning_rate": 6.404321608040201e-05,
-      "loss": 5.2009,
-      "step": 36290
-    },
-    {
-      "epoch": 18.926205997392437,
-      "grad_norm": 1.3984112739562988,
-      "learning_rate": 6.404221105527639e-05,
-      "loss": 5.0371,
-      "step": 36291
-    },
-    {
-      "epoch": 18.926727509778356,
-      "grad_norm": 1.4432040452957153,
-      "learning_rate": 6.404120603015076e-05,
-      "loss": 5.5534,
-      "step": 36292
-    },
-    {
-      "epoch": 18.927249022164276,
-      "grad_norm": 1.5959495306015015,
-      "learning_rate": 6.404020100502513e-05,
-      "loss": 5.3301,
-      "step": 36293
-    },
-    {
-      "epoch": 18.927770534550195,
-      "grad_norm": 1.4927648305892944,
-      "learning_rate": 6.40391959798995e-05,
-      "loss": 4.9522,
-      "step": 36294
-    },
-    {
-      "epoch": 18.928292046936114,
-      "grad_norm": 1.4653586149215698,
-      "learning_rate": 6.403819095477387e-05,
-      "loss": 5.0528,
-      "step": 36295
-    },
-    {
-      "epoch": 18.928813559322034,
-      "grad_norm": 1.5588667392730713,
-      "learning_rate": 6.403718592964825e-05,
-      "loss": 4.6189,
-      "step": 36296
-    },
-    {
-      "epoch": 18.929335071707953,
-      "grad_norm": 1.532081127166748,
-      "learning_rate": 6.403618090452261e-05,
-      "loss": 5.3178,
-      "step": 36297
-    },
-    {
-      "epoch": 18.929856584093873,
-      "grad_norm": 1.5147581100463867,
-      "learning_rate": 6.403517587939699e-05,
-      "loss": 5.1475,
-      "step": 36298
-    },
-    {
-      "epoch": 18.930378096479792,
-      "grad_norm": 1.6128572225570679,
-      "learning_rate": 6.403417085427135e-05,
-      "loss": 5.1585,
-      "step": 36299
-    },
-    {
-      "epoch": 18.93089960886571,
-      "grad_norm": 1.5722092390060425,
-      "learning_rate": 6.403316582914573e-05,
-      "loss": 5.4534,
-      "step": 36300
-    },
-    {
-      "epoch": 18.93142112125163,
-      "grad_norm": 1.4924544095993042,
-      "learning_rate": 6.40321608040201e-05,
-      "loss": 5.381,
-      "step": 36301
-    },
-    {
-      "epoch": 18.931942633637547,
-      "grad_norm": 1.6154961585998535,
-      "learning_rate": 6.403115577889447e-05,
-      "loss": 5.2774,
-      "step": 36302
-    },
-    {
-      "epoch": 18.932464146023467,
-      "grad_norm": 1.497407078742981,
-      "learning_rate": 6.403015075376885e-05,
-      "loss": 5.1194,
-      "step": 36303
-    },
-    {
-      "epoch": 18.932985658409386,
-      "grad_norm": 1.711442470550537,
-      "learning_rate": 6.402914572864322e-05,
-      "loss": 4.4724,
-      "step": 36304
-    },
-    {
-      "epoch": 18.933507170795306,
-      "grad_norm": 1.5402292013168335,
-      "learning_rate": 6.40281407035176e-05,
-      "loss": 5.4043,
-      "step": 36305
-    },
-    {
-      "epoch": 18.934028683181225,
-      "grad_norm": 1.5562809705734253,
-      "learning_rate": 6.402713567839196e-05,
-      "loss": 5.442,
-      "step": 36306
-    },
-    {
-      "epoch": 18.934550195567144,
-      "grad_norm": 1.5067366361618042,
-      "learning_rate": 6.402613065326634e-05,
-      "loss": 5.784,
-      "step": 36307
-    },
-    {
-      "epoch": 18.935071707953064,
-      "grad_norm": 1.4032869338989258,
-      "learning_rate": 6.40251256281407e-05,
-      "loss": 5.679,
-      "step": 36308
-    },
-    {
-      "epoch": 18.935593220338983,
-      "grad_norm": 1.4953317642211914,
-      "learning_rate": 6.402412060301508e-05,
-      "loss": 5.0396,
-      "step": 36309
-    },
-    {
-      "epoch": 18.936114732724903,
-      "grad_norm": 1.5478482246398926,
-      "learning_rate": 6.402311557788944e-05,
-      "loss": 5.0592,
-      "step": 36310
-    },
-    {
-      "epoch": 18.936636245110822,
-      "grad_norm": 1.5210950374603271,
-      "learning_rate": 6.402211055276382e-05,
-      "loss": 4.7872,
-      "step": 36311
-    },
-    {
-      "epoch": 18.937157757496742,
-      "grad_norm": 1.5244574546813965,
-      "learning_rate": 6.40211055276382e-05,
-      "loss": 5.439,
-      "step": 36312
-    },
-    {
-      "epoch": 18.93767926988266,
-      "grad_norm": 1.5688457489013672,
-      "learning_rate": 6.402010050251258e-05,
-      "loss": 5.3549,
-      "step": 36313
-    },
-    {
-      "epoch": 18.938200782268577,
-      "grad_norm": 1.5115646123886108,
-      "learning_rate": 6.401909547738694e-05,
-      "loss": 4.9958,
-      "step": 36314
-    },
-    {
-      "epoch": 18.938722294654497,
-      "grad_norm": 1.4146430492401123,
-      "learning_rate": 6.401809045226132e-05,
-      "loss": 5.4082,
-      "step": 36315
-    },
-    {
-      "epoch": 18.939243807040416,
-      "grad_norm": 1.4981220960617065,
-      "learning_rate": 6.401708542713568e-05,
-      "loss": 5.0569,
-      "step": 36316
-    },
-    {
-      "epoch": 18.939765319426336,
-      "grad_norm": 1.5089508295059204,
-      "learning_rate": 6.401608040201005e-05,
-      "loss": 5.2073,
-      "step": 36317
-    },
-    {
-      "epoch": 18.940286831812255,
-      "grad_norm": 1.6205538511276245,
-      "learning_rate": 6.401507537688442e-05,
-      "loss": 4.5975,
-      "step": 36318
-    },
-    {
-      "epoch": 18.940808344198174,
-      "grad_norm": 1.595171570777893,
-      "learning_rate": 6.401407035175879e-05,
-      "loss": 5.6086,
-      "step": 36319
-    },
-    {
-      "epoch": 18.941329856584094,
-      "grad_norm": 1.5850378274917603,
-      "learning_rate": 6.401306532663317e-05,
-      "loss": 5.3111,
-      "step": 36320
-    },
-    {
-      "epoch": 18.941851368970013,
-      "grad_norm": 1.5371571779251099,
-      "learning_rate": 6.401206030150753e-05,
-      "loss": 5.109,
-      "step": 36321
-    },
-    {
-      "epoch": 18.942372881355933,
-      "grad_norm": 1.5438822507858276,
-      "learning_rate": 6.401105527638191e-05,
-      "loss": 5.2879,
-      "step": 36322
-    },
-    {
-      "epoch": 18.942894393741852,
-      "grad_norm": 1.548506498336792,
-      "learning_rate": 6.401005025125629e-05,
-      "loss": 4.7352,
-      "step": 36323
-    },
-    {
-      "epoch": 18.943415906127772,
-      "grad_norm": 1.563644289970398,
-      "learning_rate": 6.400904522613066e-05,
-      "loss": 4.8368,
-      "step": 36324
-    },
-    {
-      "epoch": 18.94393741851369,
-      "grad_norm": 1.5550519227981567,
-      "learning_rate": 6.400804020100503e-05,
-      "loss": 4.7403,
-      "step": 36325
-    },
-    {
-      "epoch": 18.944458930899607,
-      "grad_norm": 1.5488699674606323,
-      "learning_rate": 6.40070351758794e-05,
-      "loss": 5.0924,
-      "step": 36326
-    },
-    {
-      "epoch": 18.944980443285527,
-      "grad_norm": 1.5228408575057983,
-      "learning_rate": 6.400603015075377e-05,
-      "loss": 4.8859,
-      "step": 36327
-    },
-    {
-      "epoch": 18.945501955671446,
-      "grad_norm": 1.5239155292510986,
-      "learning_rate": 6.400502512562815e-05,
-      "loss": 5.4988,
-      "step": 36328
-    },
-    {
-      "epoch": 18.946023468057366,
-      "grad_norm": 1.5264171361923218,
-      "learning_rate": 6.400402010050251e-05,
-      "loss": 5.5263,
-      "step": 36329
-    },
-    {
-      "epoch": 18.946544980443285,
-      "grad_norm": 1.4120473861694336,
-      "learning_rate": 6.400301507537688e-05,
-      "loss": 5.0121,
-      "step": 36330
-    },
-    {
-      "epoch": 18.947066492829205,
-      "grad_norm": 1.5149757862091064,
-      "learning_rate": 6.400201005025126e-05,
-      "loss": 5.4649,
-      "step": 36331
-    },
-    {
-      "epoch": 18.947588005215124,
-      "grad_norm": 1.6116588115692139,
-      "learning_rate": 6.400100502512563e-05,
-      "loss": 5.4928,
-      "step": 36332
-    },
-    {
-      "epoch": 18.948109517601043,
-      "grad_norm": 1.4784910678863525,
-      "learning_rate": 6.400000000000001e-05,
-      "loss": 5.099,
-      "step": 36333
-    },
-    {
-      "epoch": 18.948631029986963,
-      "grad_norm": 1.5227853059768677,
-      "learning_rate": 6.399899497487437e-05,
-      "loss": 5.1638,
-      "step": 36334
-    },
-    {
-      "epoch": 18.949152542372882,
-      "grad_norm": 1.5314266681671143,
-      "learning_rate": 6.399798994974875e-05,
-      "loss": 4.7931,
-      "step": 36335
-    },
-    {
-      "epoch": 18.949674054758802,
-      "grad_norm": 1.7029438018798828,
-      "learning_rate": 6.399698492462312e-05,
-      "loss": 5.1359,
-      "step": 36336
-    },
-    {
-      "epoch": 18.95019556714472,
-      "grad_norm": 1.5243538618087769,
-      "learning_rate": 6.39959798994975e-05,
-      "loss": 5.3436,
-      "step": 36337
-    },
-    {
-      "epoch": 18.950717079530637,
-      "grad_norm": 1.4239857196807861,
-      "learning_rate": 6.399497487437186e-05,
-      "loss": 5.6126,
-      "step": 36338
-    },
-    {
-      "epoch": 18.951238591916557,
-      "grad_norm": 1.542472004890442,
-      "learning_rate": 6.399396984924624e-05,
-      "loss": 5.3236,
-      "step": 36339
-    },
-    {
-      "epoch": 18.951760104302476,
-      "grad_norm": 1.4876352548599243,
-      "learning_rate": 6.39929648241206e-05,
-      "loss": 5.3279,
-      "step": 36340
-    },
-    {
-      "epoch": 18.952281616688396,
-      "grad_norm": 1.4652842283248901,
-      "learning_rate": 6.399195979899498e-05,
-      "loss": 4.48,
-      "step": 36341
-    },
-    {
-      "epoch": 18.952803129074315,
-      "grad_norm": 1.3962911367416382,
-      "learning_rate": 6.399095477386934e-05,
-      "loss": 5.5819,
-      "step": 36342
-    },
-    {
-      "epoch": 18.953324641460235,
-      "grad_norm": 1.4073245525360107,
-      "learning_rate": 6.398994974874372e-05,
-      "loss": 5.2623,
-      "step": 36343
-    },
-    {
-      "epoch": 18.953846153846154,
-      "grad_norm": 1.5652891397476196,
-      "learning_rate": 6.39889447236181e-05,
-      "loss": 4.485,
-      "step": 36344
-    },
-    {
-      "epoch": 18.954367666232073,
-      "grad_norm": 1.494625210762024,
-      "learning_rate": 6.398793969849246e-05,
-      "loss": 4.5897,
-      "step": 36345
-    },
-    {
-      "epoch": 18.954889178617993,
-      "grad_norm": 1.5235084295272827,
-      "learning_rate": 6.398693467336684e-05,
-      "loss": 5.4312,
-      "step": 36346
-    },
-    {
-      "epoch": 18.955410691003912,
-      "grad_norm": 1.5086491107940674,
-      "learning_rate": 6.39859296482412e-05,
-      "loss": 5.3733,
-      "step": 36347
-    },
-    {
-      "epoch": 18.955932203389832,
-      "grad_norm": 1.399393081665039,
-      "learning_rate": 6.398492462311558e-05,
-      "loss": 5.6341,
-      "step": 36348
-    },
-    {
-      "epoch": 18.95645371577575,
-      "grad_norm": 1.6006807088851929,
-      "learning_rate": 6.398391959798995e-05,
-      "loss": 5.485,
-      "step": 36349
-    },
-    {
-      "epoch": 18.956975228161667,
-      "grad_norm": 1.5168216228485107,
-      "learning_rate": 6.398291457286433e-05,
-      "loss": 5.2761,
-      "step": 36350
-    },
-    {
-      "epoch": 18.957496740547587,
-      "grad_norm": 1.352529764175415,
-      "learning_rate": 6.398190954773869e-05,
-      "loss": 5.6392,
-      "step": 36351
-    },
-    {
-      "epoch": 18.958018252933506,
-      "grad_norm": 1.7657673358917236,
-      "learning_rate": 6.398090452261307e-05,
-      "loss": 4.7742,
-      "step": 36352
-    },
-    {
-      "epoch": 18.958539765319426,
-      "grad_norm": 1.475597858428955,
-      "learning_rate": 6.397989949748745e-05,
-      "loss": 5.2816,
-      "step": 36353
-    },
-    {
-      "epoch": 18.959061277705345,
-      "grad_norm": 1.504118800163269,
-      "learning_rate": 6.397889447236182e-05,
-      "loss": 5.5358,
-      "step": 36354
-    },
-    {
-      "epoch": 18.959582790091265,
-      "grad_norm": 1.5021814107894897,
-      "learning_rate": 6.397788944723619e-05,
-      "loss": 5.0309,
-      "step": 36355
-    },
-    {
-      "epoch": 18.960104302477184,
-      "grad_norm": 1.5989235639572144,
-      "learning_rate": 6.397688442211057e-05,
-      "loss": 5.4913,
-      "step": 36356
-    },
-    {
-      "epoch": 18.960625814863103,
-      "grad_norm": 1.4271684885025024,
-      "learning_rate": 6.397587939698493e-05,
-      "loss": 5.6721,
-      "step": 36357
-    },
-    {
-      "epoch": 18.961147327249023,
-      "grad_norm": 1.6072583198547363,
-      "learning_rate": 6.39748743718593e-05,
-      "loss": 5.0589,
-      "step": 36358
-    },
-    {
-      "epoch": 18.961668839634942,
-      "grad_norm": 1.5302656888961792,
-      "learning_rate": 6.397386934673367e-05,
-      "loss": 5.143,
-      "step": 36359
-    },
-    {
-      "epoch": 18.962190352020862,
-      "grad_norm": 1.420785665512085,
-      "learning_rate": 6.397286432160804e-05,
-      "loss": 5.3878,
-      "step": 36360
-    },
-    {
-      "epoch": 18.96271186440678,
-      "grad_norm": 1.4502924680709839,
-      "learning_rate": 6.397185929648241e-05,
-      "loss": 5.2445,
-      "step": 36361
-    },
-    {
-      "epoch": 18.963233376792697,
-      "grad_norm": 1.5360506772994995,
-      "learning_rate": 6.397085427135678e-05,
-      "loss": 5.0256,
-      "step": 36362
-    },
-    {
-      "epoch": 18.963754889178617,
-      "grad_norm": 1.5216636657714844,
-      "learning_rate": 6.396984924623116e-05,
-      "loss": 5.621,
-      "step": 36363
-    },
-    {
-      "epoch": 18.964276401564536,
-      "grad_norm": 1.5981419086456299,
-      "learning_rate": 6.396884422110553e-05,
-      "loss": 5.2899,
-      "step": 36364
-    },
-    {
-      "epoch": 18.964797913950456,
-      "grad_norm": 1.4571863412857056,
-      "learning_rate": 6.396783919597991e-05,
-      "loss": 5.0993,
-      "step": 36365
-    },
-    {
-      "epoch": 18.965319426336375,
-      "grad_norm": 1.5443283319473267,
-      "learning_rate": 6.396683417085428e-05,
-      "loss": 5.4587,
-      "step": 36366
-    },
-    {
-      "epoch": 18.965840938722295,
-      "grad_norm": 1.5857517719268799,
-      "learning_rate": 6.396582914572865e-05,
-      "loss": 4.9785,
-      "step": 36367
-    },
-    {
-      "epoch": 18.966362451108214,
-      "grad_norm": 1.5439568758010864,
-      "learning_rate": 6.396482412060302e-05,
-      "loss": 5.2066,
-      "step": 36368
-    },
-    {
-      "epoch": 18.966883963494134,
-      "grad_norm": 1.5963279008865356,
-      "learning_rate": 6.39638190954774e-05,
-      "loss": 5.2824,
-      "step": 36369
-    },
-    {
-      "epoch": 18.967405475880053,
-      "grad_norm": 1.5480585098266602,
-      "learning_rate": 6.396281407035176e-05,
-      "loss": 5.5399,
-      "step": 36370
-    },
-    {
-      "epoch": 18.967926988265972,
-      "grad_norm": 1.6044108867645264,
-      "learning_rate": 6.396180904522612e-05,
-      "loss": 4.8455,
-      "step": 36371
-    },
-    {
-      "epoch": 18.968448500651892,
-      "grad_norm": 1.6003870964050293,
-      "learning_rate": 6.39608040201005e-05,
-      "loss": 4.9205,
-      "step": 36372
-    },
-    {
-      "epoch": 18.96897001303781,
-      "grad_norm": 1.481105923652649,
-      "learning_rate": 6.395979899497488e-05,
-      "loss": 5.0724,
-      "step": 36373
-    },
-    {
-      "epoch": 18.969491525423727,
-      "grad_norm": 1.4993730783462524,
-      "learning_rate": 6.395879396984926e-05,
-      "loss": 5.4487,
-      "step": 36374
-    },
-    {
-      "epoch": 18.970013037809647,
-      "grad_norm": 1.443029761314392,
-      "learning_rate": 6.395778894472362e-05,
-      "loss": 5.1632,
-      "step": 36375
-    },
-    {
-      "epoch": 18.970534550195566,
-      "grad_norm": 1.5736000537872314,
-      "learning_rate": 6.3956783919598e-05,
-      "loss": 5.199,
-      "step": 36376
-    },
-    {
-      "epoch": 18.971056062581486,
-      "grad_norm": 1.5692921876907349,
-      "learning_rate": 6.395577889447236e-05,
-      "loss": 4.8155,
-      "step": 36377
-    },
-    {
-      "epoch": 18.971577574967405,
-      "grad_norm": 1.5377954244613647,
-      "learning_rate": 6.395477386934674e-05,
-      "loss": 5.1676,
-      "step": 36378
-    },
-    {
-      "epoch": 18.972099087353325,
-      "grad_norm": 1.523167371749878,
-      "learning_rate": 6.39537688442211e-05,
-      "loss": 5.5383,
-      "step": 36379
-    },
-    {
-      "epoch": 18.972620599739244,
-      "grad_norm": 1.5092257261276245,
-      "learning_rate": 6.395276381909548e-05,
-      "loss": 5.3072,
-      "step": 36380
-    },
-    {
-      "epoch": 18.973142112125164,
-      "grad_norm": 1.4799182415008545,
-      "learning_rate": 6.395175879396985e-05,
-      "loss": 5.1202,
-      "step": 36381
-    },
-    {
-      "epoch": 18.973663624511083,
-      "grad_norm": 1.4440547227859497,
-      "learning_rate": 6.395075376884423e-05,
-      "loss": 5.2105,
-      "step": 36382
-    },
-    {
-      "epoch": 18.974185136897002,
-      "grad_norm": 1.565718173980713,
-      "learning_rate": 6.394974874371859e-05,
-      "loss": 5.1607,
-      "step": 36383
-    },
-    {
-      "epoch": 18.974706649282922,
-      "grad_norm": 1.5154064893722534,
-      "learning_rate": 6.394874371859297e-05,
-      "loss": 5.1281,
-      "step": 36384
-    },
-    {
-      "epoch": 18.975228161668838,
-      "grad_norm": 1.5831751823425293,
-      "learning_rate": 6.394773869346735e-05,
-      "loss": 5.3439,
-      "step": 36385
-    },
-    {
-      "epoch": 18.975749674054757,
-      "grad_norm": 1.510176658630371,
-      "learning_rate": 6.394673366834171e-05,
-      "loss": 5.4103,
-      "step": 36386
-    },
-    {
-      "epoch": 18.976271186440677,
-      "grad_norm": 1.4515647888183594,
-      "learning_rate": 6.394572864321609e-05,
-      "loss": 5.6268,
-      "step": 36387
-    },
-    {
-      "epoch": 18.976792698826596,
-      "grad_norm": 1.393825888633728,
-      "learning_rate": 6.394472361809045e-05,
-      "loss": 5.5522,
-      "step": 36388
-    },
-    {
-      "epoch": 18.977314211212516,
-      "grad_norm": 1.4746861457824707,
-      "learning_rate": 6.394371859296483e-05,
-      "loss": 5.3951,
-      "step": 36389
-    },
-    {
-      "epoch": 18.977835723598435,
-      "grad_norm": 1.5197993516921997,
-      "learning_rate": 6.39427135678392e-05,
-      "loss": 5.0111,
-      "step": 36390
-    },
-    {
-      "epoch": 18.978357235984355,
-      "grad_norm": 1.5499286651611328,
-      "learning_rate": 6.394170854271357e-05,
-      "loss": 5.2341,
-      "step": 36391
-    },
-    {
-      "epoch": 18.978878748370274,
-      "grad_norm": 1.4168745279312134,
-      "learning_rate": 6.394070351758794e-05,
-      "loss": 5.3837,
-      "step": 36392
-    },
-    {
-      "epoch": 18.979400260756194,
-      "grad_norm": 1.4090020656585693,
-      "learning_rate": 6.393969849246231e-05,
-      "loss": 5.3674,
-      "step": 36393
-    },
-    {
-      "epoch": 18.979921773142113,
-      "grad_norm": 1.5054775476455688,
-      "learning_rate": 6.393869346733669e-05,
-      "loss": 5.3589,
-      "step": 36394
-    },
-    {
-      "epoch": 18.980443285528033,
-      "grad_norm": 1.536129117012024,
-      "learning_rate": 6.393768844221107e-05,
-      "loss": 5.1643,
-      "step": 36395
-    },
-    {
-      "epoch": 18.980964797913952,
-      "grad_norm": 1.5385679006576538,
-      "learning_rate": 6.393668341708543e-05,
-      "loss": 4.9146,
-      "step": 36396
-    },
-    {
-      "epoch": 18.98148631029987,
-      "grad_norm": 1.4788508415222168,
-      "learning_rate": 6.39356783919598e-05,
-      "loss": 5.2362,
-      "step": 36397
-    },
-    {
-      "epoch": 18.982007822685787,
-      "grad_norm": 1.4977564811706543,
-      "learning_rate": 6.393467336683418e-05,
-      "loss": 5.3996,
-      "step": 36398
-    },
-    {
-      "epoch": 18.982529335071707,
-      "grad_norm": 1.4907888174057007,
-      "learning_rate": 6.393366834170854e-05,
-      "loss": 5.5442,
-      "step": 36399
-    },
-    {
-      "epoch": 18.983050847457626,
-      "grad_norm": 1.4022825956344604,
-      "learning_rate": 6.393266331658292e-05,
-      "loss": 5.51,
-      "step": 36400
-    },
-    {
-      "epoch": 18.983572359843546,
-      "grad_norm": 1.5182963609695435,
-      "learning_rate": 6.393165829145728e-05,
-      "loss": 5.1229,
-      "step": 36401
-    },
-    {
-      "epoch": 18.984093872229465,
-      "grad_norm": 1.5249119997024536,
-      "learning_rate": 6.393065326633166e-05,
-      "loss": 5.1968,
-      "step": 36402
-    },
-    {
-      "epoch": 18.984615384615385,
-      "grad_norm": 1.4615013599395752,
-      "learning_rate": 6.392964824120602e-05,
-      "loss": 5.2629,
-      "step": 36403
-    },
-    {
-      "epoch": 18.985136897001304,
-      "grad_norm": 1.473999261856079,
-      "learning_rate": 6.39286432160804e-05,
-      "loss": 5.4992,
-      "step": 36404
-    },
-    {
-      "epoch": 18.985658409387224,
-      "grad_norm": 1.4354653358459473,
-      "learning_rate": 6.392763819095478e-05,
-      "loss": 4.9285,
-      "step": 36405
-    },
-    {
-      "epoch": 18.986179921773143,
-      "grad_norm": 1.6344817876815796,
-      "learning_rate": 6.392663316582916e-05,
-      "loss": 5.1617,
-      "step": 36406
-    },
-    {
-      "epoch": 18.986701434159063,
-      "grad_norm": 1.5544605255126953,
-      "learning_rate": 6.392562814070352e-05,
-      "loss": 5.482,
-      "step": 36407
-    },
-    {
-      "epoch": 18.987222946544982,
-      "grad_norm": 1.4644179344177246,
-      "learning_rate": 6.39246231155779e-05,
-      "loss": 5.5054,
-      "step": 36408
-    },
-    {
-      "epoch": 18.987744458930898,
-      "grad_norm": 1.363000512123108,
-      "learning_rate": 6.392361809045226e-05,
-      "loss": 5.1858,
-      "step": 36409
-    },
-    {
-      "epoch": 18.988265971316817,
-      "grad_norm": 1.4165173768997192,
-      "learning_rate": 6.392261306532663e-05,
-      "loss": 5.2271,
-      "step": 36410
-    },
-    {
-      "epoch": 18.988787483702737,
-      "grad_norm": 1.5173660516738892,
-      "learning_rate": 6.392160804020101e-05,
-      "loss": 5.3691,
-      "step": 36411
-    },
-    {
-      "epoch": 18.989308996088656,
-      "grad_norm": 1.3722710609436035,
-      "learning_rate": 6.392060301507537e-05,
-      "loss": 5.5191,
-      "step": 36412
-    },
-    {
-      "epoch": 18.989830508474576,
-      "grad_norm": 1.490736484527588,
-      "learning_rate": 6.391959798994975e-05,
-      "loss": 5.502,
-      "step": 36413
-    },
-    {
-      "epoch": 18.990352020860495,
-      "grad_norm": 1.508996605873108,
-      "learning_rate": 6.391859296482411e-05,
-      "loss": 5.1428,
-      "step": 36414
-    },
-    {
-      "epoch": 18.990873533246415,
-      "grad_norm": 1.5832735300064087,
-      "learning_rate": 6.391758793969849e-05,
-      "loss": 5.3106,
-      "step": 36415
-    },
-    {
-      "epoch": 18.991395045632334,
-      "grad_norm": 1.6087205410003662,
-      "learning_rate": 6.391658291457287e-05,
-      "loss": 5.3761,
-      "step": 36416
-    },
-    {
-      "epoch": 18.991916558018254,
-      "grad_norm": 1.5368000268936157,
-      "learning_rate": 6.391557788944725e-05,
-      "loss": 5.506,
-      "step": 36417
-    },
-    {
-      "epoch": 18.992438070404173,
-      "grad_norm": 1.5611159801483154,
-      "learning_rate": 6.391457286432161e-05,
-      "loss": 5.2054,
-      "step": 36418
-    },
-    {
-      "epoch": 18.992959582790093,
-      "grad_norm": 1.4925205707550049,
-      "learning_rate": 6.391356783919599e-05,
-      "loss": 5.2638,
-      "step": 36419
-    },
-    {
-      "epoch": 18.993481095176012,
-      "grad_norm": 1.4701879024505615,
-      "learning_rate": 6.391256281407035e-05,
-      "loss": 5.7408,
-      "step": 36420
-    },
-    {
-      "epoch": 18.994002607561928,
-      "grad_norm": 1.449507474899292,
-      "learning_rate": 6.391155778894473e-05,
-      "loss": 5.7025,
-      "step": 36421
-    },
-    {
-      "epoch": 18.994524119947847,
-      "grad_norm": 1.5352505445480347,
-      "learning_rate": 6.39105527638191e-05,
-      "loss": 5.4507,
-      "step": 36422
-    },
-    {
-      "epoch": 18.995045632333767,
-      "grad_norm": 1.6228381395339966,
-      "learning_rate": 6.390954773869346e-05,
-      "loss": 5.0959,
-      "step": 36423
-    },
-    {
-      "epoch": 18.995567144719686,
-      "grad_norm": 1.5831514596939087,
-      "learning_rate": 6.390854271356784e-05,
-      "loss": 5.3233,
-      "step": 36424
-    },
-    {
-      "epoch": 18.996088657105606,
-      "grad_norm": 1.456286907196045,
-      "learning_rate": 6.390753768844222e-05,
-      "loss": 5.5739,
-      "step": 36425
-    },
-    {
-      "epoch": 18.996610169491525,
-      "grad_norm": 1.5418107509613037,
-      "learning_rate": 6.390653266331659e-05,
-      "loss": 5.4024,
-      "step": 36426
-    },
-    {
-      "epoch": 18.997131681877445,
-      "grad_norm": 1.3206504583358765,
-      "learning_rate": 6.390552763819096e-05,
-      "loss": 5.5905,
-      "step": 36427
-    },
-    {
-      "epoch": 18.997653194263364,
-      "grad_norm": 1.5055292844772339,
-      "learning_rate": 6.390452261306534e-05,
-      "loss": 5.0207,
-      "step": 36428
-    },
-    {
-      "epoch": 18.998174706649284,
-      "grad_norm": 1.3307043313980103,
-      "learning_rate": 6.39035175879397e-05,
-      "loss": 4.848,
-      "step": 36429
-    },
-    {
-      "epoch": 18.998696219035203,
-      "grad_norm": 1.4130092859268188,
-      "learning_rate": 6.390251256281408e-05,
-      "loss": 5.1887,
-      "step": 36430
-    },
-    {
-      "epoch": 18.999217731421123,
-      "grad_norm": 1.5378360748291016,
-      "learning_rate": 6.390150753768844e-05,
-      "loss": 5.5752,
-      "step": 36431
-    },
-    {
-      "epoch": 18.999739243807042,
-      "grad_norm": 1.5153274536132812,
-      "learning_rate": 6.390050251256282e-05,
-      "loss": 5.5561,
-      "step": 36432
-    },
-    {
-      "epoch": 19.000260756192958,
-      "grad_norm": 1.4498170614242554,
-      "learning_rate": 6.389949748743718e-05,
-      "loss": 5.812,
-      "step": 36433
-    },
-    {
-      "epoch": 19.000782268578877,
-      "grad_norm": 1.4657938480377197,
-      "learning_rate": 6.389849246231156e-05,
-      "loss": 5.3594,
-      "step": 36434
-    },
-    {
-      "epoch": 19.001303780964797,
-      "grad_norm": 1.5366445779800415,
-      "learning_rate": 6.389748743718593e-05,
-      "loss": 5.3402,
-      "step": 36435
-    },
-    {
-      "epoch": 19.001825293350716,
-      "grad_norm": 1.6665617227554321,
-      "learning_rate": 6.38964824120603e-05,
-      "loss": 5.2992,
-      "step": 36436
-    },
-    {
-      "epoch": 19.002346805736636,
-      "grad_norm": 1.477283000946045,
-      "learning_rate": 6.389547738693468e-05,
-      "loss": 5.2712,
-      "step": 36437
-    },
-    {
-      "epoch": 19.002868318122555,
-      "grad_norm": 1.4257358312606812,
-      "learning_rate": 6.389447236180905e-05,
-      "loss": 5.6476,
-      "step": 36438
-    },
-    {
-      "epoch": 19.003389830508475,
-      "grad_norm": 1.5418530702590942,
-      "learning_rate": 6.389346733668342e-05,
-      "loss": 5.3388,
-      "step": 36439
-    },
-    {
-      "epoch": 19.003911342894394,
-      "grad_norm": 1.3973771333694458,
-      "learning_rate": 6.389246231155779e-05,
-      "loss": 4.8605,
-      "step": 36440
-    },
-    {
-      "epoch": 19.004432855280314,
-      "grad_norm": 1.510692834854126,
-      "learning_rate": 6.389145728643217e-05,
-      "loss": 5.2427,
-      "step": 36441
-    },
-    {
-      "epoch": 19.004954367666233,
-      "grad_norm": 1.4329012632369995,
-      "learning_rate": 6.389045226130653e-05,
-      "loss": 5.4912,
-      "step": 36442
-    },
-    {
-      "epoch": 19.005475880052153,
-      "grad_norm": 1.3920092582702637,
-      "learning_rate": 6.388944723618091e-05,
-      "loss": 5.2158,
-      "step": 36443
-    },
-    {
-      "epoch": 19.005997392438072,
-      "grad_norm": 1.531011700630188,
-      "learning_rate": 6.388844221105527e-05,
-      "loss": 5.5185,
-      "step": 36444
-    },
-    {
-      "epoch": 19.006518904823988,
-      "grad_norm": 1.7794307470321655,
-      "learning_rate": 6.388743718592965e-05,
-      "loss": 4.9228,
-      "step": 36445
-    },
-    {
-      "epoch": 19.007040417209907,
-      "grad_norm": 1.6493555307388306,
-      "learning_rate": 6.388643216080403e-05,
-      "loss": 4.8674,
-      "step": 36446
-    },
-    {
-      "epoch": 19.007561929595827,
-      "grad_norm": 1.436018943786621,
-      "learning_rate": 6.38854271356784e-05,
-      "loss": 5.1412,
-      "step": 36447
-    },
-    {
-      "epoch": 19.008083441981746,
-      "grad_norm": 1.4513887166976929,
-      "learning_rate": 6.388442211055277e-05,
-      "loss": 5.5302,
-      "step": 36448
-    },
-    {
-      "epoch": 19.008604954367666,
-      "grad_norm": 1.4046200513839722,
-      "learning_rate": 6.388341708542715e-05,
-      "loss": 5.5264,
-      "step": 36449
-    },
-    {
-      "epoch": 19.009126466753585,
-      "grad_norm": 1.4527695178985596,
-      "learning_rate": 6.388241206030151e-05,
-      "loss": 4.807,
-      "step": 36450
-    },
-    {
-      "epoch": 19.009647979139505,
-      "grad_norm": 1.4804778099060059,
-      "learning_rate": 6.388140703517588e-05,
-      "loss": 5.2941,
-      "step": 36451
-    },
-    {
-      "epoch": 19.010169491525424,
-      "grad_norm": 1.4075803756713867,
-      "learning_rate": 6.388040201005025e-05,
-      "loss": 5.3264,
-      "step": 36452
-    },
-    {
-      "epoch": 19.010691003911344,
-      "grad_norm": 1.472478985786438,
-      "learning_rate": 6.387939698492462e-05,
-      "loss": 5.326,
-      "step": 36453
-    },
-    {
-      "epoch": 19.011212516297263,
-      "grad_norm": 1.424439787864685,
-      "learning_rate": 6.3878391959799e-05,
-      "loss": 5.3651,
-      "step": 36454
-    },
-    {
-      "epoch": 19.011734028683183,
-      "grad_norm": 1.4651509523391724,
-      "learning_rate": 6.387738693467336e-05,
-      "loss": 4.9113,
-      "step": 36455
-    },
-    {
-      "epoch": 19.012255541069102,
-      "grad_norm": 1.5615006685256958,
-      "learning_rate": 6.387638190954774e-05,
-      "loss": 5.1468,
-      "step": 36456
-    },
-    {
-      "epoch": 19.012777053455018,
-      "grad_norm": 1.4683409929275513,
-      "learning_rate": 6.387537688442212e-05,
-      "loss": 5.4286,
-      "step": 36457
-    },
-    {
-      "epoch": 19.013298565840937,
-      "grad_norm": 1.4806933403015137,
-      "learning_rate": 6.38743718592965e-05,
-      "loss": 5.2872,
-      "step": 36458
-    },
-    {
-      "epoch": 19.013820078226857,
-      "grad_norm": 1.400978922843933,
-      "learning_rate": 6.387336683417086e-05,
-      "loss": 5.6631,
-      "step": 36459
-    },
-    {
-      "epoch": 19.014341590612776,
-      "grad_norm": 1.4781547784805298,
-      "learning_rate": 6.387236180904524e-05,
-      "loss": 5.3451,
-      "step": 36460
-    },
-    {
-      "epoch": 19.014863102998696,
-      "grad_norm": 1.4590935707092285,
-      "learning_rate": 6.38713567839196e-05,
-      "loss": 5.3484,
-      "step": 36461
-    },
-    {
-      "epoch": 19.015384615384615,
-      "grad_norm": 1.5076823234558105,
-      "learning_rate": 6.387035175879398e-05,
-      "loss": 5.279,
-      "step": 36462
-    },
-    {
-      "epoch": 19.015906127770535,
-      "grad_norm": 1.5022331476211548,
-      "learning_rate": 6.386934673366834e-05,
-      "loss": 4.1949,
-      "step": 36463
-    },
-    {
-      "epoch": 19.016427640156454,
-      "grad_norm": 1.5384430885314941,
-      "learning_rate": 6.38683417085427e-05,
-      "loss": 5.0968,
-      "step": 36464
-    },
-    {
-      "epoch": 19.016949152542374,
-      "grad_norm": 1.5601495504379272,
-      "learning_rate": 6.386733668341708e-05,
-      "loss": 5.4517,
-      "step": 36465
-    },
-    {
-      "epoch": 19.017470664928293,
-      "grad_norm": 1.5081920623779297,
-      "learning_rate": 6.386633165829146e-05,
-      "loss": 5.4368,
-      "step": 36466
-    },
-    {
-      "epoch": 19.017992177314213,
-      "grad_norm": 1.5114338397979736,
-      "learning_rate": 6.386532663316584e-05,
-      "loss": 5.4837,
-      "step": 36467
-    },
-    {
-      "epoch": 19.018513689700132,
-      "grad_norm": 1.4646339416503906,
-      "learning_rate": 6.38643216080402e-05,
-      "loss": 5.4553,
-      "step": 36468
-    },
-    {
-      "epoch": 19.019035202086048,
-      "grad_norm": 1.6463786363601685,
-      "learning_rate": 6.386331658291458e-05,
-      "loss": 5.0925,
-      "step": 36469
-    },
-    {
-      "epoch": 19.019556714471967,
-      "grad_norm": 1.493266224861145,
-      "learning_rate": 6.386231155778895e-05,
-      "loss": 5.1596,
-      "step": 36470
-    },
-    {
-      "epoch": 19.020078226857887,
-      "grad_norm": 1.5242294073104858,
-      "learning_rate": 6.386130653266332e-05,
-      "loss": 5.4097,
-      "step": 36471
-    },
-    {
-      "epoch": 19.020599739243806,
-      "grad_norm": 1.5348807573318481,
-      "learning_rate": 6.386030150753769e-05,
-      "loss": 5.0576,
-      "step": 36472
-    },
-    {
-      "epoch": 19.021121251629726,
-      "grad_norm": 1.5065150260925293,
-      "learning_rate": 6.385929648241207e-05,
-      "loss": 5.7928,
-      "step": 36473
-    },
-    {
-      "epoch": 19.021642764015645,
-      "grad_norm": 1.395912766456604,
-      "learning_rate": 6.385829145728643e-05,
-      "loss": 5.5437,
-      "step": 36474
-    },
-    {
-      "epoch": 19.022164276401565,
-      "grad_norm": 1.4397414922714233,
-      "learning_rate": 6.385728643216081e-05,
-      "loss": 5.5168,
-      "step": 36475
-    },
-    {
-      "epoch": 19.022685788787484,
-      "grad_norm": 1.4051600694656372,
-      "learning_rate": 6.385628140703517e-05,
-      "loss": 4.7824,
-      "step": 36476
-    },
-    {
-      "epoch": 19.023207301173404,
-      "grad_norm": 1.4835326671600342,
-      "learning_rate": 6.385527638190955e-05,
-      "loss": 5.6432,
-      "step": 36477
-    },
-    {
-      "epoch": 19.023728813559323,
-      "grad_norm": 1.5000804662704468,
-      "learning_rate": 6.385427135678393e-05,
-      "loss": 5.496,
-      "step": 36478
-    },
-    {
-      "epoch": 19.024250325945243,
-      "grad_norm": 1.5501950979232788,
-      "learning_rate": 6.385326633165829e-05,
-      "loss": 5.4527,
-      "step": 36479
-    },
-    {
-      "epoch": 19.02477183833116,
-      "grad_norm": 1.4072762727737427,
-      "learning_rate": 6.385226130653267e-05,
-      "loss": 5.6153,
-      "step": 36480
-    },
-    {
-      "epoch": 19.025293350717078,
-      "grad_norm": 1.399797797203064,
-      "learning_rate": 6.385125628140703e-05,
-      "loss": 5.4149,
-      "step": 36481
-    },
-    {
-      "epoch": 19.025814863102998,
-      "grad_norm": 1.6464295387268066,
-      "learning_rate": 6.385025125628141e-05,
-      "loss": 4.8524,
-      "step": 36482
-    },
-    {
-      "epoch": 19.026336375488917,
-      "grad_norm": 1.603816032409668,
-      "learning_rate": 6.384924623115578e-05,
-      "loss": 4.9213,
-      "step": 36483
-    },
-    {
-      "epoch": 19.026857887874836,
-      "grad_norm": 1.501155138015747,
-      "learning_rate": 6.384824120603015e-05,
-      "loss": 5.3296,
-      "step": 36484
-    },
-    {
-      "epoch": 19.027379400260756,
-      "grad_norm": 1.4615312814712524,
-      "learning_rate": 6.384723618090452e-05,
-      "loss": 4.9667,
-      "step": 36485
-    },
-    {
-      "epoch": 19.027900912646675,
-      "grad_norm": 1.5007359981536865,
-      "learning_rate": 6.38462311557789e-05,
-      "loss": 5.3927,
-      "step": 36486
-    },
-    {
-      "epoch": 19.028422425032595,
-      "grad_norm": 1.4750616550445557,
-      "learning_rate": 6.384522613065327e-05,
-      "loss": 5.2445,
-      "step": 36487
-    },
-    {
-      "epoch": 19.028943937418514,
-      "grad_norm": 1.6725836992263794,
-      "learning_rate": 6.384422110552765e-05,
-      "loss": 4.7844,
-      "step": 36488
-    },
-    {
-      "epoch": 19.029465449804434,
-      "grad_norm": 1.4435590505599976,
-      "learning_rate": 6.384321608040202e-05,
-      "loss": 5.5217,
-      "step": 36489
-    },
-    {
-      "epoch": 19.029986962190353,
-      "grad_norm": 1.4350696802139282,
-      "learning_rate": 6.384221105527638e-05,
-      "loss": 5.4831,
-      "step": 36490
-    },
-    {
-      "epoch": 19.030508474576273,
-      "grad_norm": 1.4217041730880737,
-      "learning_rate": 6.384120603015076e-05,
-      "loss": 5.4418,
-      "step": 36491
-    },
-    {
-      "epoch": 19.03102998696219,
-      "grad_norm": 1.4142299890518188,
-      "learning_rate": 6.384020100502512e-05,
-      "loss": 5.796,
-      "step": 36492
-    },
-    {
-      "epoch": 19.031551499348108,
-      "grad_norm": 1.5102044343948364,
-      "learning_rate": 6.38391959798995e-05,
-      "loss": 5.1056,
-      "step": 36493
-    },
-    {
-      "epoch": 19.032073011734028,
-      "grad_norm": 1.5957304239273071,
-      "learning_rate": 6.383819095477387e-05,
-      "loss": 5.441,
-      "step": 36494
-    },
-    {
-      "epoch": 19.032594524119947,
-      "grad_norm": 1.422868013381958,
-      "learning_rate": 6.383718592964824e-05,
-      "loss": 4.9152,
-      "step": 36495
-    },
-    {
-      "epoch": 19.033116036505866,
-      "grad_norm": 1.4956989288330078,
-      "learning_rate": 6.383618090452261e-05,
-      "loss": 5.2306,
-      "step": 36496
-    },
-    {
-      "epoch": 19.033637548891786,
-      "grad_norm": 1.5107190608978271,
-      "learning_rate": 6.383517587939699e-05,
-      "loss": 5.4643,
-      "step": 36497
-    },
-    {
-      "epoch": 19.034159061277705,
-      "grad_norm": 1.5417370796203613,
-      "learning_rate": 6.383417085427136e-05,
-      "loss": 5.1927,
-      "step": 36498
-    },
-    {
-      "epoch": 19.034680573663625,
-      "grad_norm": 1.391204833984375,
-      "learning_rate": 6.383316582914574e-05,
-      "loss": 5.282,
-      "step": 36499
-    },
-    {
-      "epoch": 19.035202086049544,
-      "grad_norm": 1.5229167938232422,
-      "learning_rate": 6.38321608040201e-05,
-      "loss": 5.4241,
-      "step": 36500
-    },
-    {
-      "epoch": 19.035723598435464,
-      "grad_norm": 1.5110135078430176,
-      "learning_rate": 6.383115577889448e-05,
-      "loss": 4.9811,
-      "step": 36501
-    },
-    {
-      "epoch": 19.036245110821383,
-      "grad_norm": 1.5072760581970215,
-      "learning_rate": 6.383015075376885e-05,
-      "loss": 5.1421,
-      "step": 36502
-    },
-    {
-      "epoch": 19.036766623207303,
-      "grad_norm": 1.499022364616394,
-      "learning_rate": 6.382914572864321e-05,
-      "loss": 5.4395,
-      "step": 36503
-    },
-    {
-      "epoch": 19.03728813559322,
-      "grad_norm": 1.4008692502975464,
-      "learning_rate": 6.382814070351759e-05,
-      "loss": 5.4162,
-      "step": 36504
-    },
-    {
-      "epoch": 19.037809647979138,
-      "grad_norm": 1.6091115474700928,
-      "learning_rate": 6.382713567839195e-05,
-      "loss": 5.0299,
-      "step": 36505
-    },
-    {
-      "epoch": 19.038331160365058,
-      "grad_norm": 1.591385841369629,
-      "learning_rate": 6.382613065326633e-05,
-      "loss": 5.0674,
-      "step": 36506
-    },
-    {
-      "epoch": 19.038852672750977,
-      "grad_norm": 1.4665042161941528,
-      "learning_rate": 6.382512562814071e-05,
-      "loss": 5.3333,
-      "step": 36507
-    },
-    {
-      "epoch": 19.039374185136897,
-      "grad_norm": 1.5577585697174072,
-      "learning_rate": 6.382412060301509e-05,
-      "loss": 5.5395,
-      "step": 36508
-    },
-    {
-      "epoch": 19.039895697522816,
-      "grad_norm": 1.5146288871765137,
-      "learning_rate": 6.382311557788945e-05,
-      "loss": 5.5293,
-      "step": 36509
-    },
-    {
-      "epoch": 19.040417209908735,
-      "grad_norm": 1.5342308282852173,
-      "learning_rate": 6.382211055276383e-05,
-      "loss": 5.0126,
-      "step": 36510
-    },
-    {
-      "epoch": 19.040938722294655,
-      "grad_norm": 1.4932032823562622,
-      "learning_rate": 6.38211055276382e-05,
-      "loss": 4.8517,
-      "step": 36511
-    },
-    {
-      "epoch": 19.041460234680574,
-      "grad_norm": 1.514173150062561,
-      "learning_rate": 6.382010050251257e-05,
-      "loss": 5.2624,
-      "step": 36512
-    },
-    {
-      "epoch": 19.041981747066494,
-      "grad_norm": 1.4733203649520874,
-      "learning_rate": 6.381909547738694e-05,
-      "loss": 5.6135,
-      "step": 36513
-    },
-    {
-      "epoch": 19.042503259452413,
-      "grad_norm": 1.5975241661071777,
-      "learning_rate": 6.381809045226131e-05,
-      "loss": 5.2755,
-      "step": 36514
-    },
-    {
-      "epoch": 19.043024771838333,
-      "grad_norm": 1.4733407497406006,
-      "learning_rate": 6.381708542713568e-05,
-      "loss": 5.4039,
-      "step": 36515
-    },
-    {
-      "epoch": 19.04354628422425,
-      "grad_norm": 1.4654357433319092,
-      "learning_rate": 6.381608040201004e-05,
-      "loss": 5.4956,
-      "step": 36516
-    },
-    {
-      "epoch": 19.044067796610168,
-      "grad_norm": 1.487318515777588,
-      "learning_rate": 6.381507537688442e-05,
-      "loss": 5.6068,
-      "step": 36517
-    },
-    {
-      "epoch": 19.044589308996088,
-      "grad_norm": 1.5200185775756836,
-      "learning_rate": 6.38140703517588e-05,
-      "loss": 5.5286,
-      "step": 36518
-    },
-    {
-      "epoch": 19.045110821382007,
-      "grad_norm": 1.5736103057861328,
-      "learning_rate": 6.381306532663318e-05,
-      "loss": 5.3334,
-      "step": 36519
-    },
-    {
-      "epoch": 19.045632333767927,
-      "grad_norm": 1.4616059064865112,
-      "learning_rate": 6.381206030150754e-05,
-      "loss": 5.4751,
-      "step": 36520
-    },
-    {
-      "epoch": 19.046153846153846,
-      "grad_norm": 1.6072853803634644,
-      "learning_rate": 6.381105527638192e-05,
-      "loss": 5.0529,
-      "step": 36521
-    },
-    {
-      "epoch": 19.046675358539765,
-      "grad_norm": 1.6348141431808472,
-      "learning_rate": 6.381005025125628e-05,
-      "loss": 5.2557,
-      "step": 36522
-    },
-    {
-      "epoch": 19.047196870925685,
-      "grad_norm": 1.4991405010223389,
-      "learning_rate": 6.380904522613066e-05,
-      "loss": 5.6913,
-      "step": 36523
-    },
-    {
-      "epoch": 19.047718383311604,
-      "grad_norm": 1.5230505466461182,
-      "learning_rate": 6.380804020100502e-05,
-      "loss": 5.4666,
-      "step": 36524
-    },
-    {
-      "epoch": 19.048239895697524,
-      "grad_norm": 1.5461788177490234,
-      "learning_rate": 6.38070351758794e-05,
-      "loss": 5.4028,
-      "step": 36525
-    },
-    {
-      "epoch": 19.048761408083443,
-      "grad_norm": 1.4686331748962402,
-      "learning_rate": 6.380603015075377e-05,
-      "loss": 5.2241,
-      "step": 36526
-    },
-    {
-      "epoch": 19.049282920469363,
-      "grad_norm": 1.5382760763168335,
-      "learning_rate": 6.380502512562814e-05,
-      "loss": 5.0061,
-      "step": 36527
-    },
-    {
-      "epoch": 19.04980443285528,
-      "grad_norm": 1.5505443811416626,
-      "learning_rate": 6.380402010050252e-05,
-      "loss": 5.4528,
-      "step": 36528
-    },
-    {
-      "epoch": 19.050325945241198,
-      "grad_norm": 1.4329123497009277,
-      "learning_rate": 6.38030150753769e-05,
-      "loss": 5.6293,
-      "step": 36529
-    },
-    {
-      "epoch": 19.050847457627118,
-      "grad_norm": 1.5236488580703735,
-      "learning_rate": 6.380201005025126e-05,
-      "loss": 5.4299,
-      "step": 36530
-    },
-    {
-      "epoch": 19.051368970013037,
-      "grad_norm": 1.5810667276382446,
-      "learning_rate": 6.380100502512563e-05,
-      "loss": 5.294,
-      "step": 36531
-    },
-    {
-      "epoch": 19.051890482398957,
-      "grad_norm": 1.6183044910430908,
-      "learning_rate": 6.38e-05,
-      "loss": 5.161,
-      "step": 36532
-    },
-    {
-      "epoch": 19.052411994784876,
-      "grad_norm": 1.475118637084961,
-      "learning_rate": 6.379899497487437e-05,
-      "loss": 5.2044,
-      "step": 36533
-    },
-    {
-      "epoch": 19.052933507170795,
-      "grad_norm": 1.473341703414917,
-      "learning_rate": 6.379798994974875e-05,
-      "loss": 5.1578,
-      "step": 36534
-    },
-    {
-      "epoch": 19.053455019556715,
-      "grad_norm": 1.4159265756607056,
-      "learning_rate": 6.379698492462311e-05,
-      "loss": 5.5254,
-      "step": 36535
-    },
-    {
-      "epoch": 19.053976531942634,
-      "grad_norm": 1.757319450378418,
-      "learning_rate": 6.379597989949749e-05,
-      "loss": 4.918,
-      "step": 36536
-    },
-    {
-      "epoch": 19.054498044328554,
-      "grad_norm": 1.3613264560699463,
-      "learning_rate": 6.379497487437185e-05,
-      "loss": 5.5014,
-      "step": 36537
-    },
-    {
-      "epoch": 19.055019556714473,
-      "grad_norm": 1.5040158033370972,
-      "learning_rate": 6.379396984924623e-05,
-      "loss": 4.9637,
-      "step": 36538
-    },
-    {
-      "epoch": 19.055541069100393,
-      "grad_norm": 1.5155035257339478,
-      "learning_rate": 6.379296482412061e-05,
-      "loss": 5.3774,
-      "step": 36539
-    },
-    {
-      "epoch": 19.05606258148631,
-      "grad_norm": 1.5342602729797363,
-      "learning_rate": 6.379195979899499e-05,
-      "loss": 5.2978,
-      "step": 36540
-    },
-    {
-      "epoch": 19.056584093872228,
-      "grad_norm": 1.4987059831619263,
-      "learning_rate": 6.379095477386935e-05,
-      "loss": 5.2034,
-      "step": 36541
-    },
-    {
-      "epoch": 19.057105606258148,
-      "grad_norm": 1.4527872800827026,
-      "learning_rate": 6.378994974874373e-05,
-      "loss": 5.3865,
-      "step": 36542
-    },
-    {
-      "epoch": 19.057627118644067,
-      "grad_norm": 1.4571586847305298,
-      "learning_rate": 6.37889447236181e-05,
-      "loss": 5.5738,
-      "step": 36543
-    },
-    {
-      "epoch": 19.058148631029987,
-      "grad_norm": 1.4864294528961182,
-      "learning_rate": 6.378793969849246e-05,
-      "loss": 5.1064,
-      "step": 36544
-    },
-    {
-      "epoch": 19.058670143415906,
-      "grad_norm": 1.54171621799469,
-      "learning_rate": 6.378693467336684e-05,
-      "loss": 5.1263,
-      "step": 36545
-    },
-    {
-      "epoch": 19.059191655801826,
-      "grad_norm": 1.5183409452438354,
-      "learning_rate": 6.37859296482412e-05,
-      "loss": 4.8823,
-      "step": 36546
-    },
-    {
-      "epoch": 19.059713168187745,
-      "grad_norm": 1.5672645568847656,
-      "learning_rate": 6.378492462311558e-05,
-      "loss": 4.8046,
-      "step": 36547
-    },
-    {
-      "epoch": 19.060234680573664,
-      "grad_norm": 1.4592701196670532,
-      "learning_rate": 6.378391959798996e-05,
-      "loss": 5.5306,
-      "step": 36548
-    },
-    {
-      "epoch": 19.060756192959584,
-      "grad_norm": 1.4693222045898438,
-      "learning_rate": 6.378291457286433e-05,
-      "loss": 5.4183,
-      "step": 36549
-    },
-    {
-      "epoch": 19.061277705345503,
-      "grad_norm": 1.4959423542022705,
-      "learning_rate": 6.37819095477387e-05,
-      "loss": 5.2775,
-      "step": 36550
-    },
-    {
-      "epoch": 19.061799217731423,
-      "grad_norm": 1.5169011354446411,
-      "learning_rate": 6.378090452261308e-05,
-      "loss": 5.5043,
-      "step": 36551
-    },
-    {
-      "epoch": 19.06232073011734,
-      "grad_norm": 1.4433250427246094,
-      "learning_rate": 6.377989949748744e-05,
-      "loss": 5.4013,
-      "step": 36552
-    },
-    {
-      "epoch": 19.062842242503258,
-      "grad_norm": 1.5331544876098633,
-      "learning_rate": 6.377889447236182e-05,
-      "loss": 4.9131,
-      "step": 36553
-    },
-    {
-      "epoch": 19.063363754889178,
-      "grad_norm": 1.6172598600387573,
-      "learning_rate": 6.377788944723618e-05,
-      "loss": 5.0992,
-      "step": 36554
-    },
-    {
-      "epoch": 19.063885267275097,
-      "grad_norm": 1.6446874141693115,
-      "learning_rate": 6.377688442211056e-05,
-      "loss": 4.5598,
-      "step": 36555
-    },
-    {
-      "epoch": 19.064406779661017,
-      "grad_norm": 1.6799814701080322,
-      "learning_rate": 6.377587939698492e-05,
-      "loss": 4.8566,
-      "step": 36556
-    },
-    {
-      "epoch": 19.064928292046936,
-      "grad_norm": 1.6671537160873413,
-      "learning_rate": 6.377487437185929e-05,
-      "loss": 5.4178,
-      "step": 36557
-    },
-    {
-      "epoch": 19.065449804432856,
-      "grad_norm": 1.6838674545288086,
-      "learning_rate": 6.377386934673367e-05,
-      "loss": 5.0808,
-      "step": 36558
-    },
-    {
-      "epoch": 19.065971316818775,
-      "grad_norm": 1.6032413244247437,
-      "learning_rate": 6.377286432160804e-05,
-      "loss": 5.0605,
-      "step": 36559
-    },
-    {
-      "epoch": 19.066492829204694,
-      "grad_norm": 1.5606235265731812,
-      "learning_rate": 6.377185929648242e-05,
-      "loss": 5.3278,
-      "step": 36560
-    },
-    {
-      "epoch": 19.067014341590614,
-      "grad_norm": 1.4803038835525513,
-      "learning_rate": 6.377085427135679e-05,
-      "loss": 5.6547,
-      "step": 36561
-    },
-    {
-      "epoch": 19.067535853976533,
-      "grad_norm": 1.5093634128570557,
-      "learning_rate": 6.376984924623116e-05,
-      "loss": 5.2453,
-      "step": 36562
-    },
-    {
-      "epoch": 19.068057366362453,
-      "grad_norm": 1.5818591117858887,
-      "learning_rate": 6.376884422110553e-05,
-      "loss": 5.366,
-      "step": 36563
-    },
-    {
-      "epoch": 19.06857887874837,
-      "grad_norm": 1.543556571006775,
-      "learning_rate": 6.37678391959799e-05,
-      "loss": 5.2485,
-      "step": 36564
-    },
-    {
-      "epoch": 19.06910039113429,
-      "grad_norm": 1.5193630456924438,
-      "learning_rate": 6.376683417085427e-05,
-      "loss": 5.2204,
-      "step": 36565
-    },
-    {
-      "epoch": 19.069621903520208,
-      "grad_norm": 1.562666654586792,
-      "learning_rate": 6.376582914572865e-05,
-      "loss": 5.0643,
-      "step": 36566
-    },
-    {
-      "epoch": 19.070143415906127,
-      "grad_norm": 1.5432220697402954,
-      "learning_rate": 6.376482412060301e-05,
-      "loss": 5.401,
-      "step": 36567
-    },
-    {
-      "epoch": 19.070664928292047,
-      "grad_norm": 1.6121044158935547,
-      "learning_rate": 6.376381909547739e-05,
-      "loss": 5.0907,
-      "step": 36568
-    },
-    {
-      "epoch": 19.071186440677966,
-      "grad_norm": 1.477071762084961,
-      "learning_rate": 6.376281407035176e-05,
-      "loss": 5.5063,
-      "step": 36569
-    },
-    {
-      "epoch": 19.071707953063886,
-      "grad_norm": 1.5291582345962524,
-      "learning_rate": 6.376180904522613e-05,
-      "loss": 5.4276,
-      "step": 36570
-    },
-    {
-      "epoch": 19.072229465449805,
-      "grad_norm": 1.6010515689849854,
-      "learning_rate": 6.376080402010051e-05,
-      "loss": 4.6766,
-      "step": 36571
-    },
-    {
-      "epoch": 19.072750977835724,
-      "grad_norm": 1.5369950532913208,
-      "learning_rate": 6.375979899497487e-05,
-      "loss": 5.354,
-      "step": 36572
-    },
-    {
-      "epoch": 19.073272490221644,
-      "grad_norm": 1.6257680654525757,
-      "learning_rate": 6.375879396984925e-05,
-      "loss": 4.9673,
-      "step": 36573
-    },
-    {
-      "epoch": 19.073794002607563,
-      "grad_norm": 1.3857817649841309,
-      "learning_rate": 6.375778894472362e-05,
-      "loss": 4.3686,
-      "step": 36574
-    },
-    {
-      "epoch": 19.07431551499348,
-      "grad_norm": 1.5214663743972778,
-      "learning_rate": 6.3756783919598e-05,
-      "loss": 5.4902,
-      "step": 36575
-    },
-    {
-      "epoch": 19.0748370273794,
-      "grad_norm": 1.4161297082901,
-      "learning_rate": 6.375577889447236e-05,
-      "loss": 5.6936,
-      "step": 36576
-    },
-    {
-      "epoch": 19.07535853976532,
-      "grad_norm": 1.5125356912612915,
-      "learning_rate": 6.375477386934674e-05,
-      "loss": 5.6921,
-      "step": 36577
-    },
-    {
-      "epoch": 19.075880052151238,
-      "grad_norm": 1.4842946529388428,
-      "learning_rate": 6.37537688442211e-05,
-      "loss": 5.6686,
-      "step": 36578
-    },
-    {
-      "epoch": 19.076401564537157,
-      "grad_norm": 1.4877196550369263,
-      "learning_rate": 6.375276381909548e-05,
-      "loss": 5.6936,
-      "step": 36579
-    },
-    {
-      "epoch": 19.076923076923077,
-      "grad_norm": 1.462684154510498,
-      "learning_rate": 6.375175879396986e-05,
-      "loss": 5.1331,
-      "step": 36580
-    },
-    {
-      "epoch": 19.077444589308996,
-      "grad_norm": 1.510770559310913,
-      "learning_rate": 6.375075376884423e-05,
-      "loss": 5.0803,
-      "step": 36581
-    },
-    {
-      "epoch": 19.077966101694916,
-      "grad_norm": 1.5933682918548584,
-      "learning_rate": 6.37497487437186e-05,
-      "loss": 5.0981,
-      "step": 36582
-    },
-    {
-      "epoch": 19.078487614080835,
-      "grad_norm": 1.4598448276519775,
-      "learning_rate": 6.374874371859296e-05,
-      "loss": 5.0569,
-      "step": 36583
-    },
-    {
-      "epoch": 19.079009126466755,
-      "grad_norm": 1.6159898042678833,
-      "learning_rate": 6.374773869346734e-05,
-      "loss": 4.8164,
-      "step": 36584
-    },
-    {
-      "epoch": 19.079530638852674,
-      "grad_norm": 1.4630815982818604,
-      "learning_rate": 6.37467336683417e-05,
-      "loss": 5.3505,
-      "step": 36585
-    },
-    {
-      "epoch": 19.080052151238593,
-      "grad_norm": 1.4564765691757202,
-      "learning_rate": 6.374572864321608e-05,
-      "loss": 5.5377,
-      "step": 36586
-    },
-    {
-      "epoch": 19.08057366362451,
-      "grad_norm": 1.5831578969955444,
-      "learning_rate": 6.374472361809045e-05,
-      "loss": 5.0242,
-      "step": 36587
-    },
-    {
-      "epoch": 19.08109517601043,
-      "grad_norm": 1.4714878797531128,
-      "learning_rate": 6.374371859296483e-05,
-      "loss": 5.4547,
-      "step": 36588
-    },
-    {
-      "epoch": 19.08161668839635,
-      "grad_norm": 1.492266058921814,
-      "learning_rate": 6.374271356783919e-05,
-      "loss": 5.3775,
-      "step": 36589
-    },
-    {
-      "epoch": 19.082138200782268,
-      "grad_norm": 1.4242206811904907,
-      "learning_rate": 6.374170854271357e-05,
-      "loss": 5.4771,
-      "step": 36590
-    },
-    {
-      "epoch": 19.082659713168187,
-      "grad_norm": 1.4412202835083008,
-      "learning_rate": 6.374070351758795e-05,
-      "loss": 5.4024,
-      "step": 36591
-    },
-    {
-      "epoch": 19.083181225554107,
-      "grad_norm": 1.4410947561264038,
-      "learning_rate": 6.373969849246232e-05,
-      "loss": 5.2565,
-      "step": 36592
-    },
-    {
-      "epoch": 19.083702737940026,
-      "grad_norm": 1.4974161386489868,
-      "learning_rate": 6.373869346733669e-05,
-      "loss": 5.1217,
-      "step": 36593
-    },
-    {
-      "epoch": 19.084224250325946,
-      "grad_norm": 1.516315221786499,
-      "learning_rate": 6.373768844221107e-05,
-      "loss": 5.3457,
-      "step": 36594
-    },
-    {
-      "epoch": 19.084745762711865,
-      "grad_norm": 1.5183523893356323,
-      "learning_rate": 6.373668341708543e-05,
-      "loss": 4.8831,
-      "step": 36595
-    },
-    {
-      "epoch": 19.085267275097785,
-      "grad_norm": 1.5317456722259521,
-      "learning_rate": 6.37356783919598e-05,
-      "loss": 5.4046,
-      "step": 36596
-    },
-    {
-      "epoch": 19.085788787483704,
-      "grad_norm": 1.4952510595321655,
-      "learning_rate": 6.373467336683417e-05,
-      "loss": 5.7256,
-      "step": 36597
-    },
-    {
-      "epoch": 19.086310299869623,
-      "grad_norm": 1.515519380569458,
-      "learning_rate": 6.373366834170854e-05,
-      "loss": 5.4626,
-      "step": 36598
-    },
-    {
-      "epoch": 19.08683181225554,
-      "grad_norm": 1.6007788181304932,
-      "learning_rate": 6.373266331658291e-05,
-      "loss": 5.3058,
-      "step": 36599
-    },
-    {
-      "epoch": 19.08735332464146,
-      "grad_norm": 1.6001778841018677,
-      "learning_rate": 6.373165829145729e-05,
-      "loss": 5.0834,
-      "step": 36600
-    },
-    {
-      "epoch": 19.08787483702738,
-      "grad_norm": 1.5111722946166992,
-      "learning_rate": 6.373065326633167e-05,
-      "loss": 5.3711,
-      "step": 36601
-    },
-    {
-      "epoch": 19.088396349413298,
-      "grad_norm": 1.53868567943573,
-      "learning_rate": 6.372964824120603e-05,
-      "loss": 4.7364,
-      "step": 36602
-    },
-    {
-      "epoch": 19.088917861799217,
-      "grad_norm": 1.5261507034301758,
-      "learning_rate": 6.372864321608041e-05,
-      "loss": 5.1144,
-      "step": 36603
-    },
-    {
-      "epoch": 19.089439374185137,
-      "grad_norm": 1.563050627708435,
-      "learning_rate": 6.372763819095478e-05,
-      "loss": 5.2185,
-      "step": 36604
-    },
-    {
-      "epoch": 19.089960886571056,
-      "grad_norm": 1.6243329048156738,
-      "learning_rate": 6.372663316582915e-05,
-      "loss": 5.4385,
-      "step": 36605
-    },
-    {
-      "epoch": 19.090482398956976,
-      "grad_norm": 1.4927983283996582,
-      "learning_rate": 6.372562814070352e-05,
-      "loss": 4.7481,
-      "step": 36606
-    },
-    {
-      "epoch": 19.091003911342895,
-      "grad_norm": 1.6091526746749878,
-      "learning_rate": 6.37246231155779e-05,
-      "loss": 5.0116,
-      "step": 36607
-    },
-    {
-      "epoch": 19.091525423728815,
-      "grad_norm": 1.5347274541854858,
-      "learning_rate": 6.372361809045226e-05,
-      "loss": 5.2605,
-      "step": 36608
-    },
-    {
-      "epoch": 19.092046936114734,
-      "grad_norm": 1.5131807327270508,
-      "learning_rate": 6.372261306532664e-05,
-      "loss": 5.2358,
-      "step": 36609
-    },
-    {
-      "epoch": 19.092568448500653,
-      "grad_norm": 1.5860117673873901,
-      "learning_rate": 6.3721608040201e-05,
-      "loss": 4.2869,
-      "step": 36610
-    },
-    {
-      "epoch": 19.09308996088657,
-      "grad_norm": 1.569191575050354,
-      "learning_rate": 6.372060301507538e-05,
-      "loss": 4.7987,
-      "step": 36611
-    },
-    {
-      "epoch": 19.09361147327249,
-      "grad_norm": 1.5398077964782715,
-      "learning_rate": 6.371959798994976e-05,
-      "loss": 5.1365,
-      "step": 36612
-    },
-    {
-      "epoch": 19.09413298565841,
-      "grad_norm": 1.4519495964050293,
-      "learning_rate": 6.371859296482412e-05,
-      "loss": 5.5062,
-      "step": 36613
-    },
-    {
-      "epoch": 19.094654498044328,
-      "grad_norm": 1.5475866794586182,
-      "learning_rate": 6.37175879396985e-05,
-      "loss": 5.4709,
-      "step": 36614
-    },
-    {
-      "epoch": 19.095176010430247,
-      "grad_norm": 1.482215166091919,
-      "learning_rate": 6.371658291457286e-05,
-      "loss": 5.4117,
-      "step": 36615
-    },
-    {
-      "epoch": 19.095697522816167,
-      "grad_norm": 1.5125893354415894,
-      "learning_rate": 6.371557788944724e-05,
-      "loss": 5.1712,
-      "step": 36616
-    },
-    {
-      "epoch": 19.096219035202086,
-      "grad_norm": 1.5505627393722534,
-      "learning_rate": 6.37145728643216e-05,
-      "loss": 4.6246,
-      "step": 36617
-    },
-    {
-      "epoch": 19.096740547588006,
-      "grad_norm": 1.5194066762924194,
-      "learning_rate": 6.371356783919598e-05,
-      "loss": 5.5317,
-      "step": 36618
-    },
-    {
-      "epoch": 19.097262059973925,
-      "grad_norm": 1.6286145448684692,
-      "learning_rate": 6.371256281407035e-05,
-      "loss": 5.2625,
-      "step": 36619
-    },
-    {
-      "epoch": 19.097783572359845,
-      "grad_norm": 1.5543354749679565,
-      "learning_rate": 6.371155778894473e-05,
-      "loss": 5.3475,
-      "step": 36620
-    },
-    {
-      "epoch": 19.098305084745764,
-      "grad_norm": 1.5435441732406616,
-      "learning_rate": 6.37105527638191e-05,
-      "loss": 5.7836,
-      "step": 36621
-    },
-    {
-      "epoch": 19.098826597131684,
-      "grad_norm": 1.5697647333145142,
-      "learning_rate": 6.370954773869348e-05,
-      "loss": 5.5183,
-      "step": 36622
-    },
-    {
-      "epoch": 19.0993481095176,
-      "grad_norm": 1.414245367050171,
-      "learning_rate": 6.370854271356785e-05,
-      "loss": 5.1591,
-      "step": 36623
-    },
-    {
-      "epoch": 19.09986962190352,
-      "grad_norm": 1.4757938385009766,
-      "learning_rate": 6.370753768844221e-05,
-      "loss": 5.1909,
-      "step": 36624
-    },
-    {
-      "epoch": 19.10039113428944,
-      "grad_norm": 1.4704240560531616,
-      "learning_rate": 6.370653266331659e-05,
-      "loss": 4.8111,
-      "step": 36625
-    },
-    {
-      "epoch": 19.100912646675358,
-      "grad_norm": 1.5882112979888916,
-      "learning_rate": 6.370552763819095e-05,
-      "loss": 5.3636,
-      "step": 36626
-    },
-    {
-      "epoch": 19.101434159061277,
-      "grad_norm": 1.4745354652404785,
-      "learning_rate": 6.370452261306533e-05,
-      "loss": 5.0016,
-      "step": 36627
-    },
-    {
-      "epoch": 19.101955671447197,
-      "grad_norm": 1.4902204275131226,
-      "learning_rate": 6.37035175879397e-05,
-      "loss": 5.0485,
-      "step": 36628
-    },
-    {
-      "epoch": 19.102477183833116,
-      "grad_norm": 1.547548532485962,
-      "learning_rate": 6.370251256281407e-05,
-      "loss": 4.6967,
-      "step": 36629
-    },
-    {
-      "epoch": 19.102998696219036,
-      "grad_norm": 1.4887185096740723,
-      "learning_rate": 6.370150753768844e-05,
-      "loss": 5.3562,
-      "step": 36630
-    },
-    {
-      "epoch": 19.103520208604955,
-      "grad_norm": 1.453380823135376,
-      "learning_rate": 6.370050251256281e-05,
-      "loss": 5.0912,
-      "step": 36631
-    },
-    {
-      "epoch": 19.104041720990875,
-      "grad_norm": 1.5910871028900146,
-      "learning_rate": 6.369949748743719e-05,
-      "loss": 5.1222,
-      "step": 36632
-    },
-    {
-      "epoch": 19.104563233376794,
-      "grad_norm": 1.508381962776184,
-      "learning_rate": 6.369849246231157e-05,
-      "loss": 5.1156,
-      "step": 36633
-    },
-    {
-      "epoch": 19.105084745762714,
-      "grad_norm": 1.5169093608856201,
-      "learning_rate": 6.369748743718593e-05,
-      "loss": 5.444,
-      "step": 36634
-    },
-    {
-      "epoch": 19.10560625814863,
-      "grad_norm": 1.4904595613479614,
-      "learning_rate": 6.369648241206031e-05,
-      "loss": 5.091,
-      "step": 36635
-    },
-    {
-      "epoch": 19.10612777053455,
-      "grad_norm": 1.5362987518310547,
-      "learning_rate": 6.369547738693468e-05,
-      "loss": 5.0145,
-      "step": 36636
-    },
-    {
-      "epoch": 19.10664928292047,
-      "grad_norm": 1.5943119525909424,
-      "learning_rate": 6.369447236180904e-05,
-      "loss": 5.1033,
-      "step": 36637
-    },
-    {
-      "epoch": 19.107170795306388,
-      "grad_norm": 1.8877424001693726,
-      "learning_rate": 6.369346733668342e-05,
-      "loss": 4.3844,
-      "step": 36638
-    },
-    {
-      "epoch": 19.107692307692307,
-      "grad_norm": 1.522752046585083,
-      "learning_rate": 6.369246231155778e-05,
-      "loss": 5.6336,
-      "step": 36639
-    },
-    {
-      "epoch": 19.108213820078227,
-      "grad_norm": 1.5923227071762085,
-      "learning_rate": 6.369145728643216e-05,
-      "loss": 5.1493,
-      "step": 36640
-    },
-    {
-      "epoch": 19.108735332464146,
-      "grad_norm": 1.5385719537734985,
-      "learning_rate": 6.369045226130654e-05,
-      "loss": 5.2029,
-      "step": 36641
-    },
-    {
-      "epoch": 19.109256844850066,
-      "grad_norm": 1.4773063659667969,
-      "learning_rate": 6.368944723618092e-05,
-      "loss": 5.3868,
-      "step": 36642
-    },
-    {
-      "epoch": 19.109778357235985,
-      "grad_norm": 1.523688793182373,
-      "learning_rate": 6.368844221105528e-05,
-      "loss": 5.4164,
-      "step": 36643
-    },
-    {
-      "epoch": 19.110299869621905,
-      "grad_norm": 1.4781272411346436,
-      "learning_rate": 6.368743718592966e-05,
-      "loss": 5.3316,
-      "step": 36644
-    },
-    {
-      "epoch": 19.110821382007824,
-      "grad_norm": 1.5225461721420288,
-      "learning_rate": 6.368643216080402e-05,
-      "loss": 5.4767,
-      "step": 36645
-    },
-    {
-      "epoch": 19.111342894393744,
-      "grad_norm": 1.5292868614196777,
-      "learning_rate": 6.36854271356784e-05,
-      "loss": 5.3126,
-      "step": 36646
-    },
-    {
-      "epoch": 19.11186440677966,
-      "grad_norm": 1.4513019323349,
-      "learning_rate": 6.368442211055276e-05,
-      "loss": 5.4346,
-      "step": 36647
-    },
-    {
-      "epoch": 19.11238591916558,
-      "grad_norm": 1.5161265134811401,
-      "learning_rate": 6.368341708542714e-05,
-      "loss": 5.283,
-      "step": 36648
-    },
-    {
-      "epoch": 19.1129074315515,
-      "grad_norm": 1.483806848526001,
-      "learning_rate": 6.368241206030151e-05,
-      "loss": 5.59,
-      "step": 36649
-    },
-    {
-      "epoch": 19.113428943937418,
-      "grad_norm": 1.5715782642364502,
-      "learning_rate": 6.368140703517587e-05,
-      "loss": 5.1454,
-      "step": 36650
-    },
-    {
-      "epoch": 19.113950456323337,
-      "grad_norm": 1.9295519590377808,
-      "learning_rate": 6.368040201005025e-05,
-      "loss": 4.9613,
-      "step": 36651
-    },
-    {
-      "epoch": 19.114471968709257,
-      "grad_norm": 1.5225756168365479,
-      "learning_rate": 6.367939698492463e-05,
-      "loss": 5.497,
-      "step": 36652
-    },
-    {
-      "epoch": 19.114993481095176,
-      "grad_norm": 1.585646629333496,
-      "learning_rate": 6.3678391959799e-05,
-      "loss": 4.8232,
-      "step": 36653
-    },
-    {
-      "epoch": 19.115514993481096,
-      "grad_norm": 1.4593799114227295,
-      "learning_rate": 6.367738693467337e-05,
-      "loss": 5.4117,
-      "step": 36654
-    },
-    {
-      "epoch": 19.116036505867015,
-      "grad_norm": 1.4406676292419434,
-      "learning_rate": 6.367638190954775e-05,
-      "loss": 5.5029,
-      "step": 36655
-    },
-    {
-      "epoch": 19.116558018252935,
-      "grad_norm": 1.45802640914917,
-      "learning_rate": 6.367537688442211e-05,
-      "loss": 5.4294,
-      "step": 36656
-    },
-    {
-      "epoch": 19.117079530638854,
-      "grad_norm": 1.4387221336364746,
-      "learning_rate": 6.367437185929649e-05,
-      "loss": 5.5685,
-      "step": 36657
-    },
-    {
-      "epoch": 19.117601043024774,
-      "grad_norm": 1.4047938585281372,
-      "learning_rate": 6.367336683417085e-05,
-      "loss": 5.6091,
-      "step": 36658
-    },
-    {
-      "epoch": 19.11812255541069,
-      "grad_norm": 1.4875246286392212,
-      "learning_rate": 6.367236180904523e-05,
-      "loss": 5.2661,
-      "step": 36659
-    },
-    {
-      "epoch": 19.11864406779661,
-      "grad_norm": 1.5932927131652832,
-      "learning_rate": 6.36713567839196e-05,
-      "loss": 5.1668,
-      "step": 36660
-    },
-    {
-      "epoch": 19.11916558018253,
-      "grad_norm": 1.4666935205459595,
-      "learning_rate": 6.367035175879397e-05,
-      "loss": 5.2257,
-      "step": 36661
-    },
-    {
-      "epoch": 19.119687092568448,
-      "grad_norm": 1.6545078754425049,
-      "learning_rate": 6.366934673366835e-05,
-      "loss": 4.8783,
-      "step": 36662
-    },
-    {
-      "epoch": 19.120208604954367,
-      "grad_norm": 1.4923994541168213,
-      "learning_rate": 6.366834170854272e-05,
-      "loss": 5.4607,
-      "step": 36663
-    },
-    {
-      "epoch": 19.120730117340287,
-      "grad_norm": 1.4454237222671509,
-      "learning_rate": 6.366733668341709e-05,
-      "loss": 5.6011,
-      "step": 36664
-    },
-    {
-      "epoch": 19.121251629726206,
-      "grad_norm": 1.544828176498413,
-      "learning_rate": 6.366633165829146e-05,
-      "loss": 5.2578,
-      "step": 36665
-    },
-    {
-      "epoch": 19.121773142112126,
-      "grad_norm": 1.5267301797866821,
-      "learning_rate": 6.366532663316584e-05,
-      "loss": 5.2091,
-      "step": 36666
-    },
-    {
-      "epoch": 19.122294654498045,
-      "grad_norm": 1.4454506635665894,
-      "learning_rate": 6.36643216080402e-05,
-      "loss": 5.6723,
-      "step": 36667
-    },
-    {
-      "epoch": 19.122816166883965,
-      "grad_norm": 1.4652971029281616,
-      "learning_rate": 6.366331658291458e-05,
-      "loss": 5.1305,
-      "step": 36668
-    },
-    {
-      "epoch": 19.123337679269884,
-      "grad_norm": 1.6296154260635376,
-      "learning_rate": 6.366231155778894e-05,
-      "loss": 4.9305,
-      "step": 36669
-    },
-    {
-      "epoch": 19.1238591916558,
-      "grad_norm": 1.498972773551941,
-      "learning_rate": 6.366130653266332e-05,
-      "loss": 5.5859,
-      "step": 36670
-    },
-    {
-      "epoch": 19.12438070404172,
-      "grad_norm": 1.5503907203674316,
-      "learning_rate": 6.366030150753768e-05,
-      "loss": 5.2768,
-      "step": 36671
-    },
-    {
-      "epoch": 19.12490221642764,
-      "grad_norm": 1.580798625946045,
-      "learning_rate": 6.365929648241206e-05,
-      "loss": 4.6612,
-      "step": 36672
-    },
-    {
-      "epoch": 19.12542372881356,
-      "grad_norm": 1.5076416730880737,
-      "learning_rate": 6.365829145728644e-05,
-      "loss": 5.6805,
-      "step": 36673
-    },
-    {
-      "epoch": 19.125945241199478,
-      "grad_norm": 1.4653459787368774,
-      "learning_rate": 6.365728643216082e-05,
-      "loss": 5.4984,
-      "step": 36674
-    },
-    {
-      "epoch": 19.126466753585397,
-      "grad_norm": 1.5098403692245483,
-      "learning_rate": 6.365628140703518e-05,
-      "loss": 5.5689,
-      "step": 36675
-    },
-    {
-      "epoch": 19.126988265971317,
-      "grad_norm": 1.5670430660247803,
-      "learning_rate": 6.365527638190955e-05,
-      "loss": 5.4045,
-      "step": 36676
-    },
-    {
-      "epoch": 19.127509778357236,
-      "grad_norm": 1.6365662813186646,
-      "learning_rate": 6.365427135678392e-05,
-      "loss": 4.9307,
-      "step": 36677
-    },
-    {
-      "epoch": 19.128031290743156,
-      "grad_norm": 1.390795111656189,
-      "learning_rate": 6.365326633165829e-05,
-      "loss": 5.4529,
-      "step": 36678
-    },
-    {
-      "epoch": 19.128552803129075,
-      "grad_norm": 1.5535682439804077,
-      "learning_rate": 6.365226130653267e-05,
-      "loss": 4.885,
-      "step": 36679
-    },
-    {
-      "epoch": 19.129074315514995,
-      "grad_norm": 1.6099435091018677,
-      "learning_rate": 6.365125628140703e-05,
-      "loss": 5.5162,
-      "step": 36680
-    },
-    {
-      "epoch": 19.129595827900914,
-      "grad_norm": 1.4569209814071655,
-      "learning_rate": 6.365025125628141e-05,
-      "loss": 5.2667,
-      "step": 36681
-    },
-    {
-      "epoch": 19.13011734028683,
-      "grad_norm": 1.5156009197235107,
-      "learning_rate": 6.364924623115579e-05,
-      "loss": 5.4432,
-      "step": 36682
-    },
-    {
-      "epoch": 19.13063885267275,
-      "grad_norm": 1.4014796018600464,
-      "learning_rate": 6.364824120603016e-05,
-      "loss": 5.4455,
-      "step": 36683
-    },
-    {
-      "epoch": 19.13116036505867,
-      "grad_norm": 1.5175632238388062,
-      "learning_rate": 6.364723618090453e-05,
-      "loss": 5.5212,
-      "step": 36684
-    },
-    {
-      "epoch": 19.13168187744459,
-      "grad_norm": 1.5460649728775024,
-      "learning_rate": 6.36462311557789e-05,
-      "loss": 5.2031,
-      "step": 36685
-    },
-    {
-      "epoch": 19.132203389830508,
-      "grad_norm": 1.5894851684570312,
-      "learning_rate": 6.364522613065327e-05,
-      "loss": 5.0905,
-      "step": 36686
-    },
-    {
-      "epoch": 19.132724902216427,
-      "grad_norm": 1.450091004371643,
-      "learning_rate": 6.364422110552765e-05,
-      "loss": 5.7567,
-      "step": 36687
-    },
-    {
-      "epoch": 19.133246414602347,
-      "grad_norm": 1.463516116142273,
-      "learning_rate": 6.364321608040201e-05,
-      "loss": 5.249,
-      "step": 36688
-    },
-    {
-      "epoch": 19.133767926988266,
-      "grad_norm": 1.557680606842041,
-      "learning_rate": 6.364221105527638e-05,
-      "loss": 5.2718,
-      "step": 36689
-    },
-    {
-      "epoch": 19.134289439374186,
-      "grad_norm": 1.5359597206115723,
-      "learning_rate": 6.364120603015075e-05,
-      "loss": 5.1885,
-      "step": 36690
-    },
-    {
-      "epoch": 19.134810951760105,
-      "grad_norm": 1.513721227645874,
-      "learning_rate": 6.364020100502512e-05,
-      "loss": 5.2285,
-      "step": 36691
-    },
-    {
-      "epoch": 19.135332464146025,
-      "grad_norm": 1.5070686340332031,
-      "learning_rate": 6.36391959798995e-05,
-      "loss": 5.2909,
-      "step": 36692
-    },
-    {
-      "epoch": 19.135853976531944,
-      "grad_norm": 1.6413143873214722,
-      "learning_rate": 6.363819095477387e-05,
-      "loss": 5.2341,
-      "step": 36693
-    },
-    {
-      "epoch": 19.13637548891786,
-      "grad_norm": 1.4899499416351318,
-      "learning_rate": 6.363718592964825e-05,
-      "loss": 4.8422,
-      "step": 36694
-    },
-    {
-      "epoch": 19.13689700130378,
-      "grad_norm": 1.553184151649475,
-      "learning_rate": 6.363618090452262e-05,
-      "loss": 5.456,
-      "step": 36695
-    },
-    {
-      "epoch": 19.1374185136897,
-      "grad_norm": 1.5641809701919556,
-      "learning_rate": 6.3635175879397e-05,
-      "loss": 4.6049,
-      "step": 36696
-    },
-    {
-      "epoch": 19.13794002607562,
-      "grad_norm": 1.4524518251419067,
-      "learning_rate": 6.363417085427136e-05,
-      "loss": 5.2979,
-      "step": 36697
-    },
-    {
-      "epoch": 19.138461538461538,
-      "grad_norm": 1.5345206260681152,
-      "learning_rate": 6.363316582914574e-05,
-      "loss": 4.7639,
-      "step": 36698
-    },
-    {
-      "epoch": 19.138983050847457,
-      "grad_norm": 1.3980988264083862,
-      "learning_rate": 6.36321608040201e-05,
-      "loss": 5.2046,
-      "step": 36699
-    },
-    {
-      "epoch": 19.139504563233377,
-      "grad_norm": 1.519384503364563,
-      "learning_rate": 6.363115577889448e-05,
-      "loss": 5.0816,
-      "step": 36700
-    },
-    {
-      "epoch": 19.140026075619296,
-      "grad_norm": 1.4587405920028687,
-      "learning_rate": 6.363015075376884e-05,
-      "loss": 5.2695,
-      "step": 36701
-    },
-    {
-      "epoch": 19.140547588005216,
-      "grad_norm": 1.5970838069915771,
-      "learning_rate": 6.362914572864322e-05,
-      "loss": 5.1794,
-      "step": 36702
-    },
-    {
-      "epoch": 19.141069100391135,
-      "grad_norm": 1.4958934783935547,
-      "learning_rate": 6.36281407035176e-05,
-      "loss": 5.2617,
-      "step": 36703
-    },
-    {
-      "epoch": 19.141590612777055,
-      "grad_norm": 1.515602707862854,
-      "learning_rate": 6.362713567839196e-05,
-      "loss": 5.5411,
-      "step": 36704
-    },
-    {
-      "epoch": 19.142112125162974,
-      "grad_norm": 1.4052804708480835,
-      "learning_rate": 6.362613065326634e-05,
-      "loss": 5.8909,
-      "step": 36705
-    },
-    {
-      "epoch": 19.14263363754889,
-      "grad_norm": 1.4969438314437866,
-      "learning_rate": 6.36251256281407e-05,
-      "loss": 5.366,
-      "step": 36706
-    },
-    {
-      "epoch": 19.14315514993481,
-      "grad_norm": 1.5021653175354004,
-      "learning_rate": 6.362412060301508e-05,
-      "loss": 5.0925,
-      "step": 36707
-    },
-    {
-      "epoch": 19.14367666232073,
-      "grad_norm": 1.423660159111023,
-      "learning_rate": 6.362311557788945e-05,
-      "loss": 5.4912,
-      "step": 36708
-    },
-    {
-      "epoch": 19.14419817470665,
-      "grad_norm": 1.5736644268035889,
-      "learning_rate": 6.362211055276382e-05,
-      "loss": 5.1471,
-      "step": 36709
-    },
-    {
-      "epoch": 19.144719687092568,
-      "grad_norm": 1.5139347314834595,
-      "learning_rate": 6.362110552763819e-05,
-      "loss": 5.2065,
-      "step": 36710
-    },
-    {
-      "epoch": 19.145241199478487,
-      "grad_norm": 1.5472419261932373,
-      "learning_rate": 6.362010050251257e-05,
-      "loss": 5.2008,
-      "step": 36711
-    },
-    {
-      "epoch": 19.145762711864407,
-      "grad_norm": 1.3990893363952637,
-      "learning_rate": 6.361909547738693e-05,
-      "loss": 5.6672,
-      "step": 36712
-    },
-    {
-      "epoch": 19.146284224250326,
-      "grad_norm": 1.4895142316818237,
-      "learning_rate": 6.361809045226131e-05,
-      "loss": 5.4265,
-      "step": 36713
-    },
-    {
-      "epoch": 19.146805736636246,
-      "grad_norm": 1.5120099782943726,
-      "learning_rate": 6.361708542713569e-05,
-      "loss": 5.7039,
-      "step": 36714
-    },
-    {
-      "epoch": 19.147327249022165,
-      "grad_norm": 1.5210435390472412,
-      "learning_rate": 6.361608040201006e-05,
-      "loss": 5.5757,
-      "step": 36715
-    },
-    {
-      "epoch": 19.147848761408085,
-      "grad_norm": 1.439422845840454,
-      "learning_rate": 6.361507537688443e-05,
-      "loss": 5.5971,
-      "step": 36716
-    },
-    {
-      "epoch": 19.148370273794004,
-      "grad_norm": 1.4505170583724976,
-      "learning_rate": 6.361407035175879e-05,
-      "loss": 5.5323,
-      "step": 36717
-    },
-    {
-      "epoch": 19.14889178617992,
-      "grad_norm": 1.6118375062942505,
-      "learning_rate": 6.361306532663317e-05,
-      "loss": 5.0422,
-      "step": 36718
-    },
-    {
-      "epoch": 19.14941329856584,
-      "grad_norm": 1.4072320461273193,
-      "learning_rate": 6.361206030150753e-05,
-      "loss": 5.846,
-      "step": 36719
-    },
-    {
-      "epoch": 19.14993481095176,
-      "grad_norm": 1.4924556016921997,
-      "learning_rate": 6.361105527638191e-05,
-      "loss": 5.2857,
-      "step": 36720
-    },
-    {
-      "epoch": 19.15045632333768,
-      "grad_norm": 1.4349915981292725,
-      "learning_rate": 6.361005025125628e-05,
-      "loss": 5.372,
-      "step": 36721
-    },
-    {
-      "epoch": 19.150977835723598,
-      "grad_norm": 1.5508767366409302,
-      "learning_rate": 6.360904522613065e-05,
-      "loss": 5.0448,
-      "step": 36722
-    },
-    {
-      "epoch": 19.151499348109517,
-      "grad_norm": 1.4822667837142944,
-      "learning_rate": 6.360804020100503e-05,
-      "loss": 5.1718,
-      "step": 36723
-    },
-    {
-      "epoch": 19.152020860495437,
-      "grad_norm": 1.5021687746047974,
-      "learning_rate": 6.360703517587941e-05,
-      "loss": 5.3285,
-      "step": 36724
-    },
-    {
-      "epoch": 19.152542372881356,
-      "grad_norm": 1.5797396898269653,
-      "learning_rate": 6.360603015075377e-05,
-      "loss": 4.7987,
-      "step": 36725
-    },
-    {
-      "epoch": 19.153063885267276,
-      "grad_norm": 1.550919771194458,
-      "learning_rate": 6.360502512562815e-05,
-      "loss": 5.483,
-      "step": 36726
-    },
-    {
-      "epoch": 19.153585397653195,
-      "grad_norm": 1.5435576438903809,
-      "learning_rate": 6.360402010050252e-05,
-      "loss": 5.1494,
-      "step": 36727
-    },
-    {
-      "epoch": 19.154106910039115,
-      "grad_norm": 1.5107754468917847,
-      "learning_rate": 6.36030150753769e-05,
-      "loss": 5.3998,
-      "step": 36728
-    },
-    {
-      "epoch": 19.154628422425034,
-      "grad_norm": 1.5041286945343018,
-      "learning_rate": 6.360201005025126e-05,
-      "loss": 5.3143,
-      "step": 36729
-    },
-    {
-      "epoch": 19.15514993481095,
-      "grad_norm": 1.4255475997924805,
-      "learning_rate": 6.360100502512562e-05,
-      "loss": 5.5105,
-      "step": 36730
-    },
-    {
-      "epoch": 19.15567144719687,
-      "grad_norm": 1.6097819805145264,
-      "learning_rate": 6.36e-05,
-      "loss": 4.6624,
-      "step": 36731
-    },
-    {
-      "epoch": 19.15619295958279,
-      "grad_norm": 1.448428750038147,
-      "learning_rate": 6.359899497487437e-05,
-      "loss": 5.5891,
-      "step": 36732
-    },
-    {
-      "epoch": 19.15671447196871,
-      "grad_norm": 1.6476534605026245,
-      "learning_rate": 6.359798994974874e-05,
-      "loss": 5.0132,
-      "step": 36733
-    },
-    {
-      "epoch": 19.157235984354628,
-      "grad_norm": 1.4491063356399536,
-      "learning_rate": 6.359698492462312e-05,
-      "loss": 5.2638,
-      "step": 36734
-    },
-    {
-      "epoch": 19.157757496740548,
-      "grad_norm": 1.4501032829284668,
-      "learning_rate": 6.35959798994975e-05,
-      "loss": 5.4452,
-      "step": 36735
-    },
-    {
-      "epoch": 19.158279009126467,
-      "grad_norm": 1.5640023946762085,
-      "learning_rate": 6.359497487437186e-05,
-      "loss": 5.1814,
-      "step": 36736
-    },
-    {
-      "epoch": 19.158800521512386,
-      "grad_norm": 1.462880253791809,
-      "learning_rate": 6.359396984924624e-05,
-      "loss": 5.1419,
-      "step": 36737
-    },
-    {
-      "epoch": 19.159322033898306,
-      "grad_norm": 1.3963650465011597,
-      "learning_rate": 6.35929648241206e-05,
-      "loss": 5.7111,
-      "step": 36738
-    },
-    {
-      "epoch": 19.159843546284225,
-      "grad_norm": 1.5613547563552856,
-      "learning_rate": 6.359195979899498e-05,
-      "loss": 5.1805,
-      "step": 36739
-    },
-    {
-      "epoch": 19.160365058670145,
-      "grad_norm": 1.4732403755187988,
-      "learning_rate": 6.359095477386935e-05,
-      "loss": 5.4086,
-      "step": 36740
-    },
-    {
-      "epoch": 19.160886571056064,
-      "grad_norm": 1.4188930988311768,
-      "learning_rate": 6.358994974874373e-05,
-      "loss": 5.4696,
-      "step": 36741
-    },
-    {
-      "epoch": 19.16140808344198,
-      "grad_norm": 1.5713647603988647,
-      "learning_rate": 6.358894472361809e-05,
-      "loss": 4.8627,
-      "step": 36742
-    },
-    {
-      "epoch": 19.1619295958279,
-      "grad_norm": 1.5737862586975098,
-      "learning_rate": 6.358793969849245e-05,
-      "loss": 5.2368,
-      "step": 36743
-    },
-    {
-      "epoch": 19.16245110821382,
-      "grad_norm": 1.4565887451171875,
-      "learning_rate": 6.358693467336683e-05,
-      "loss": 5.0875,
-      "step": 36744
-    },
-    {
-      "epoch": 19.16297262059974,
-      "grad_norm": 1.4393203258514404,
-      "learning_rate": 6.358592964824121e-05,
-      "loss": 5.2719,
-      "step": 36745
-    },
-    {
-      "epoch": 19.163494132985658,
-      "grad_norm": 1.711273431777954,
-      "learning_rate": 6.358492462311559e-05,
-      "loss": 4.6576,
-      "step": 36746
-    },
-    {
-      "epoch": 19.164015645371578,
-      "grad_norm": 1.5344659090042114,
-      "learning_rate": 6.358391959798995e-05,
-      "loss": 5.1877,
-      "step": 36747
-    },
-    {
-      "epoch": 19.164537157757497,
-      "grad_norm": 1.4457507133483887,
-      "learning_rate": 6.358291457286433e-05,
-      "loss": 5.3724,
-      "step": 36748
-    },
-    {
-      "epoch": 19.165058670143416,
-      "grad_norm": 1.5495229959487915,
-      "learning_rate": 6.35819095477387e-05,
-      "loss": 5.1971,
-      "step": 36749
-    },
-    {
-      "epoch": 19.165580182529336,
-      "grad_norm": 1.5006396770477295,
-      "learning_rate": 6.358090452261307e-05,
-      "loss": 5.5803,
-      "step": 36750
-    },
-    {
-      "epoch": 19.166101694915255,
-      "grad_norm": 1.5812116861343384,
-      "learning_rate": 6.357989949748744e-05,
-      "loss": 5.0205,
-      "step": 36751
-    },
-    {
-      "epoch": 19.166623207301175,
-      "grad_norm": 1.5254671573638916,
-      "learning_rate": 6.357889447236181e-05,
-      "loss": 5.1561,
-      "step": 36752
-    },
-    {
-      "epoch": 19.167144719687094,
-      "grad_norm": 1.4609397649765015,
-      "learning_rate": 6.357788944723618e-05,
-      "loss": 5.0739,
-      "step": 36753
-    },
-    {
-      "epoch": 19.16766623207301,
-      "grad_norm": 1.593964695930481,
-      "learning_rate": 6.357688442211056e-05,
-      "loss": 5.4341,
-      "step": 36754
-    },
-    {
-      "epoch": 19.16818774445893,
-      "grad_norm": 1.5672165155410767,
-      "learning_rate": 6.357587939698493e-05,
-      "loss": 5.3494,
-      "step": 36755
-    },
-    {
-      "epoch": 19.16870925684485,
-      "grad_norm": 1.514840006828308,
-      "learning_rate": 6.35748743718593e-05,
-      "loss": 5.3843,
-      "step": 36756
-    },
-    {
-      "epoch": 19.16923076923077,
-      "grad_norm": 1.5333455801010132,
-      "learning_rate": 6.357386934673368e-05,
-      "loss": 5.0843,
-      "step": 36757
-    },
-    {
-      "epoch": 19.169752281616688,
-      "grad_norm": 1.435300588607788,
-      "learning_rate": 6.357286432160804e-05,
-      "loss": 5.2276,
-      "step": 36758
-    },
-    {
-      "epoch": 19.170273794002608,
-      "grad_norm": 1.4682116508483887,
-      "learning_rate": 6.357185929648242e-05,
-      "loss": 5.5223,
-      "step": 36759
-    },
-    {
-      "epoch": 19.170795306388527,
-      "grad_norm": 1.5379986763000488,
-      "learning_rate": 6.357085427135678e-05,
-      "loss": 5.249,
-      "step": 36760
-    },
-    {
-      "epoch": 19.171316818774446,
-      "grad_norm": 1.6812586784362793,
-      "learning_rate": 6.356984924623116e-05,
-      "loss": 5.2766,
-      "step": 36761
-    },
-    {
-      "epoch": 19.171838331160366,
-      "grad_norm": 1.508056402206421,
-      "learning_rate": 6.356884422110552e-05,
-      "loss": 5.0748,
-      "step": 36762
-    },
-    {
-      "epoch": 19.172359843546285,
-      "grad_norm": 1.5261307954788208,
-      "learning_rate": 6.35678391959799e-05,
-      "loss": 5.2117,
-      "step": 36763
-    },
-    {
-      "epoch": 19.172881355932205,
-      "grad_norm": 1.5969741344451904,
-      "learning_rate": 6.356683417085427e-05,
-      "loss": 4.8812,
-      "step": 36764
-    },
-    {
-      "epoch": 19.17340286831812,
-      "grad_norm": 1.4700530767440796,
-      "learning_rate": 6.356582914572864e-05,
-      "loss": 5.6947,
-      "step": 36765
-    },
-    {
-      "epoch": 19.17392438070404,
-      "grad_norm": 1.667560338973999,
-      "learning_rate": 6.356482412060302e-05,
-      "loss": 5.18,
-      "step": 36766
-    },
-    {
-      "epoch": 19.17444589308996,
-      "grad_norm": 1.4506348371505737,
-      "learning_rate": 6.35638190954774e-05,
-      "loss": 5.4567,
-      "step": 36767
-    },
-    {
-      "epoch": 19.17496740547588,
-      "grad_norm": 1.5589851140975952,
-      "learning_rate": 6.356281407035176e-05,
-      "loss": 5.1669,
-      "step": 36768
-    },
-    {
-      "epoch": 19.1754889178618,
-      "grad_norm": 1.488966941833496,
-      "learning_rate": 6.356180904522613e-05,
-      "loss": 5.6474,
-      "step": 36769
-    },
-    {
-      "epoch": 19.176010430247718,
-      "grad_norm": 1.4959592819213867,
-      "learning_rate": 6.35608040201005e-05,
-      "loss": 5.3277,
-      "step": 36770
-    },
-    {
-      "epoch": 19.176531942633638,
-      "grad_norm": 1.4872239828109741,
-      "learning_rate": 6.355979899497487e-05,
-      "loss": 4.7649,
-      "step": 36771
-    },
-    {
-      "epoch": 19.177053455019557,
-      "grad_norm": 1.3071706295013428,
-      "learning_rate": 6.355879396984925e-05,
-      "loss": 5.7023,
-      "step": 36772
-    },
-    {
-      "epoch": 19.177574967405477,
-      "grad_norm": 1.4639469385147095,
-      "learning_rate": 6.355778894472361e-05,
-      "loss": 5.094,
-      "step": 36773
-    },
-    {
-      "epoch": 19.178096479791396,
-      "grad_norm": 1.461814045906067,
-      "learning_rate": 6.355678391959799e-05,
-      "loss": 5.6071,
-      "step": 36774
-    },
-    {
-      "epoch": 19.178617992177315,
-      "grad_norm": 1.4291685819625854,
-      "learning_rate": 6.355577889447237e-05,
-      "loss": 5.619,
-      "step": 36775
-    },
-    {
-      "epoch": 19.179139504563235,
-      "grad_norm": 1.5129731893539429,
-      "learning_rate": 6.355477386934675e-05,
-      "loss": 4.7896,
-      "step": 36776
-    },
-    {
-      "epoch": 19.17966101694915,
-      "grad_norm": 1.447580337524414,
-      "learning_rate": 6.355376884422111e-05,
-      "loss": 5.4114,
-      "step": 36777
-    },
-    {
-      "epoch": 19.18018252933507,
-      "grad_norm": 1.4864143133163452,
-      "learning_rate": 6.355276381909549e-05,
-      "loss": 4.917,
-      "step": 36778
-    },
-    {
-      "epoch": 19.18070404172099,
-      "grad_norm": 1.4383151531219482,
-      "learning_rate": 6.355175879396985e-05,
-      "loss": 5.608,
-      "step": 36779
-    },
-    {
-      "epoch": 19.18122555410691,
-      "grad_norm": 1.5035390853881836,
-      "learning_rate": 6.355075376884423e-05,
-      "loss": 5.5742,
-      "step": 36780
-    },
-    {
-      "epoch": 19.18174706649283,
-      "grad_norm": 1.4522112607955933,
-      "learning_rate": 6.35497487437186e-05,
-      "loss": 4.9576,
-      "step": 36781
-    },
-    {
-      "epoch": 19.182268578878748,
-      "grad_norm": 1.4563677310943604,
-      "learning_rate": 6.354874371859296e-05,
-      "loss": 5.539,
-      "step": 36782
-    },
-    {
-      "epoch": 19.182790091264668,
-      "grad_norm": 1.439776062965393,
-      "learning_rate": 6.354773869346734e-05,
-      "loss": 5.5646,
-      "step": 36783
-    },
-    {
-      "epoch": 19.183311603650587,
-      "grad_norm": 1.4637523889541626,
-      "learning_rate": 6.35467336683417e-05,
-      "loss": 5.616,
-      "step": 36784
-    },
-    {
-      "epoch": 19.183833116036507,
-      "grad_norm": 1.6338952779769897,
-      "learning_rate": 6.354572864321608e-05,
-      "loss": 4.8155,
-      "step": 36785
-    },
-    {
-      "epoch": 19.184354628422426,
-      "grad_norm": 1.721680760383606,
-      "learning_rate": 6.354472361809046e-05,
-      "loss": 5.2407,
-      "step": 36786
-    },
-    {
-      "epoch": 19.184876140808345,
-      "grad_norm": 1.3643306493759155,
-      "learning_rate": 6.354371859296483e-05,
-      "loss": 5.6971,
-      "step": 36787
-    },
-    {
-      "epoch": 19.185397653194265,
-      "grad_norm": 1.6221256256103516,
-      "learning_rate": 6.35427135678392e-05,
-      "loss": 5.1571,
-      "step": 36788
-    },
-    {
-      "epoch": 19.18591916558018,
-      "grad_norm": 1.4043117761611938,
-      "learning_rate": 6.354170854271358e-05,
-      "loss": 5.5485,
-      "step": 36789
-    },
-    {
-      "epoch": 19.1864406779661,
-      "grad_norm": 1.4833513498306274,
-      "learning_rate": 6.354070351758794e-05,
-      "loss": 5.2985,
-      "step": 36790
-    },
-    {
-      "epoch": 19.18696219035202,
-      "grad_norm": 1.6393437385559082,
-      "learning_rate": 6.353969849246232e-05,
-      "loss": 5.322,
-      "step": 36791
-    },
-    {
-      "epoch": 19.18748370273794,
-      "grad_norm": 1.4731892347335815,
-      "learning_rate": 6.353869346733668e-05,
-      "loss": 5.4432,
-      "step": 36792
-    },
-    {
-      "epoch": 19.18800521512386,
-      "grad_norm": 1.5758880376815796,
-      "learning_rate": 6.353768844221106e-05,
-      "loss": 5.0793,
-      "step": 36793
-    },
-    {
-      "epoch": 19.188526727509778,
-      "grad_norm": 1.5631556510925293,
-      "learning_rate": 6.353668341708542e-05,
-      "loss": 5.0857,
-      "step": 36794
-    },
-    {
-      "epoch": 19.189048239895698,
-      "grad_norm": 1.5493124723434448,
-      "learning_rate": 6.35356783919598e-05,
-      "loss": 5.4534,
-      "step": 36795
-    },
-    {
-      "epoch": 19.189569752281617,
-      "grad_norm": 1.452446699142456,
-      "learning_rate": 6.353467336683418e-05,
-      "loss": 5.5016,
-      "step": 36796
-    },
-    {
-      "epoch": 19.190091264667537,
-      "grad_norm": 1.5019510984420776,
-      "learning_rate": 6.353366834170854e-05,
-      "loss": 5.0226,
-      "step": 36797
-    },
-    {
-      "epoch": 19.190612777053456,
-      "grad_norm": 1.5245356559753418,
-      "learning_rate": 6.353266331658292e-05,
-      "loss": 5.1922,
-      "step": 36798
-    },
-    {
-      "epoch": 19.191134289439375,
-      "grad_norm": 1.5438936948776245,
-      "learning_rate": 6.353165829145729e-05,
-      "loss": 5.3812,
-      "step": 36799
-    },
-    {
-      "epoch": 19.191655801825295,
-      "grad_norm": 1.6431673765182495,
-      "learning_rate": 6.353065326633166e-05,
-      "loss": 4.8042,
-      "step": 36800
-    },
-    {
-      "epoch": 19.19217731421121,
-      "grad_norm": 1.5544520616531372,
-      "learning_rate": 6.352964824120603e-05,
-      "loss": 5.3633,
-      "step": 36801
-    },
-    {
-      "epoch": 19.19269882659713,
-      "grad_norm": 1.439686894416809,
-      "learning_rate": 6.35286432160804e-05,
-      "loss": 5.8256,
-      "step": 36802
-    },
-    {
-      "epoch": 19.19322033898305,
-      "grad_norm": 1.4189653396606445,
-      "learning_rate": 6.352763819095477e-05,
-      "loss": 5.6226,
-      "step": 36803
-    },
-    {
-      "epoch": 19.19374185136897,
-      "grad_norm": 1.5858079195022583,
-      "learning_rate": 6.352663316582915e-05,
-      "loss": 4.9998,
-      "step": 36804
-    },
-    {
-      "epoch": 19.19426336375489,
-      "grad_norm": 1.5106598138809204,
-      "learning_rate": 6.352562814070351e-05,
-      "loss": 5.2401,
-      "step": 36805
-    },
-    {
-      "epoch": 19.194784876140808,
-      "grad_norm": 1.5000866651535034,
-      "learning_rate": 6.352462311557789e-05,
-      "loss": 4.7224,
-      "step": 36806
-    },
-    {
-      "epoch": 19.195306388526728,
-      "grad_norm": 1.5284020900726318,
-      "learning_rate": 6.352361809045227e-05,
-      "loss": 5.0151,
-      "step": 36807
-    },
-    {
-      "epoch": 19.195827900912647,
-      "grad_norm": 1.5355902910232544,
-      "learning_rate": 6.352261306532665e-05,
-      "loss": 5.1299,
-      "step": 36808
-    },
-    {
-      "epoch": 19.196349413298567,
-      "grad_norm": 1.601561427116394,
-      "learning_rate": 6.352160804020101e-05,
-      "loss": 5.4361,
-      "step": 36809
-    },
-    {
-      "epoch": 19.196870925684486,
-      "grad_norm": 1.605168342590332,
-      "learning_rate": 6.352060301507538e-05,
-      "loss": 4.852,
-      "step": 36810
-    },
-    {
-      "epoch": 19.197392438070406,
-      "grad_norm": 1.6390225887298584,
-      "learning_rate": 6.351959798994975e-05,
-      "loss": 5.3375,
-      "step": 36811
-    },
-    {
-      "epoch": 19.197913950456325,
-      "grad_norm": 1.5888879299163818,
-      "learning_rate": 6.351859296482412e-05,
-      "loss": 5.013,
-      "step": 36812
-    },
-    {
-      "epoch": 19.19843546284224,
-      "grad_norm": 1.4969247579574585,
-      "learning_rate": 6.35175879396985e-05,
-      "loss": 5.4347,
-      "step": 36813
-    },
-    {
-      "epoch": 19.19895697522816,
-      "grad_norm": 1.6226699352264404,
-      "learning_rate": 6.351658291457286e-05,
-      "loss": 5.0224,
-      "step": 36814
-    },
-    {
-      "epoch": 19.19947848761408,
-      "grad_norm": 1.5531498193740845,
-      "learning_rate": 6.351557788944724e-05,
-      "loss": 5.3062,
-      "step": 36815
-    },
-    {
-      "epoch": 19.2,
-      "grad_norm": 1.434910774230957,
-      "learning_rate": 6.351457286432161e-05,
-      "loss": 5.5387,
-      "step": 36816
-    },
-    {
-      "epoch": 19.20052151238592,
-      "grad_norm": 1.4743329286575317,
-      "learning_rate": 6.351356783919599e-05,
-      "loss": 4.9693,
-      "step": 36817
-    },
-    {
-      "epoch": 19.201043024771838,
-      "grad_norm": 1.6672056913375854,
-      "learning_rate": 6.351256281407036e-05,
-      "loss": 4.8509,
-      "step": 36818
-    },
-    {
-      "epoch": 19.201564537157758,
-      "grad_norm": 1.6185163259506226,
-      "learning_rate": 6.351155778894473e-05,
-      "loss": 5.3013,
-      "step": 36819
-    },
-    {
-      "epoch": 19.202086049543677,
-      "grad_norm": 1.499768853187561,
-      "learning_rate": 6.35105527638191e-05,
-      "loss": 5.2753,
-      "step": 36820
-    },
-    {
-      "epoch": 19.202607561929597,
-      "grad_norm": 1.6774084568023682,
-      "learning_rate": 6.350954773869348e-05,
-      "loss": 5.1828,
-      "step": 36821
-    },
-    {
-      "epoch": 19.203129074315516,
-      "grad_norm": 1.4903481006622314,
-      "learning_rate": 6.350854271356784e-05,
-      "loss": 4.9863,
-      "step": 36822
-    },
-    {
-      "epoch": 19.203650586701436,
-      "grad_norm": 1.634554147720337,
-      "learning_rate": 6.35075376884422e-05,
-      "loss": 5.3333,
-      "step": 36823
-    },
-    {
-      "epoch": 19.204172099087355,
-      "grad_norm": 1.5463955402374268,
-      "learning_rate": 6.350653266331658e-05,
-      "loss": 5.1927,
-      "step": 36824
-    },
-    {
-      "epoch": 19.20469361147327,
-      "grad_norm": 1.5335463285446167,
-      "learning_rate": 6.350552763819095e-05,
-      "loss": 5.6385,
-      "step": 36825
-    },
-    {
-      "epoch": 19.20521512385919,
-      "grad_norm": 1.4487677812576294,
-      "learning_rate": 6.350452261306533e-05,
-      "loss": 5.4294,
-      "step": 36826
-    },
-    {
-      "epoch": 19.20573663624511,
-      "grad_norm": 1.4891185760498047,
-      "learning_rate": 6.35035175879397e-05,
-      "loss": 5.3142,
-      "step": 36827
-    },
-    {
-      "epoch": 19.20625814863103,
-      "grad_norm": 1.6271955966949463,
-      "learning_rate": 6.350251256281408e-05,
-      "loss": 5.4027,
-      "step": 36828
-    },
-    {
-      "epoch": 19.20677966101695,
-      "grad_norm": 1.6103308200836182,
-      "learning_rate": 6.350150753768845e-05,
-      "loss": 5.3095,
-      "step": 36829
-    },
-    {
-      "epoch": 19.20730117340287,
-      "grad_norm": 1.5681896209716797,
-      "learning_rate": 6.350050251256282e-05,
-      "loss": 5.2383,
-      "step": 36830
-    },
-    {
-      "epoch": 19.207822685788788,
-      "grad_norm": 1.5440258979797363,
-      "learning_rate": 6.349949748743719e-05,
-      "loss": 5.376,
-      "step": 36831
-    },
-    {
-      "epoch": 19.208344198174707,
-      "grad_norm": 1.476962685585022,
-      "learning_rate": 6.349849246231157e-05,
-      "loss": 5.45,
-      "step": 36832
-    },
-    {
-      "epoch": 19.208865710560627,
-      "grad_norm": 1.3936923742294312,
-      "learning_rate": 6.349748743718593e-05,
-      "loss": 5.0786,
-      "step": 36833
-    },
-    {
-      "epoch": 19.209387222946546,
-      "grad_norm": 1.4395536184310913,
-      "learning_rate": 6.349648241206031e-05,
-      "loss": 5.5912,
-      "step": 36834
-    },
-    {
-      "epoch": 19.209908735332466,
-      "grad_norm": 1.5874015092849731,
-      "learning_rate": 6.349547738693467e-05,
-      "loss": 5.1842,
-      "step": 36835
-    },
-    {
-      "epoch": 19.210430247718385,
-      "grad_norm": 1.4127824306488037,
-      "learning_rate": 6.349447236180905e-05,
-      "loss": 4.7363,
-      "step": 36836
-    },
-    {
-      "epoch": 19.2109517601043,
-      "grad_norm": 1.4570986032485962,
-      "learning_rate": 6.349346733668343e-05,
-      "loss": 5.4811,
-      "step": 36837
-    },
-    {
-      "epoch": 19.21147327249022,
-      "grad_norm": 1.4810131788253784,
-      "learning_rate": 6.349246231155779e-05,
-      "loss": 5.4135,
-      "step": 36838
-    },
-    {
-      "epoch": 19.21199478487614,
-      "grad_norm": 1.6047756671905518,
-      "learning_rate": 6.349145728643217e-05,
-      "loss": 4.9279,
-      "step": 36839
-    },
-    {
-      "epoch": 19.21251629726206,
-      "grad_norm": 1.500387191772461,
-      "learning_rate": 6.349045226130653e-05,
-      "loss": 4.7995,
-      "step": 36840
-    },
-    {
-      "epoch": 19.21303780964798,
-      "grad_norm": 1.5432854890823364,
-      "learning_rate": 6.348944723618091e-05,
-      "loss": 5.3838,
-      "step": 36841
-    },
-    {
-      "epoch": 19.2135593220339,
-      "grad_norm": 1.4466543197631836,
-      "learning_rate": 6.348844221105528e-05,
-      "loss": 5.3336,
-      "step": 36842
-    },
-    {
-      "epoch": 19.214080834419818,
-      "grad_norm": 1.5626252889633179,
-      "learning_rate": 6.348743718592965e-05,
-      "loss": 5.5827,
-      "step": 36843
-    },
-    {
-      "epoch": 19.214602346805737,
-      "grad_norm": 1.5846391916275024,
-      "learning_rate": 6.348643216080402e-05,
-      "loss": 5.4965,
-      "step": 36844
-    },
-    {
-      "epoch": 19.215123859191657,
-      "grad_norm": 1.5805906057357788,
-      "learning_rate": 6.34854271356784e-05,
-      "loss": 5.6206,
-      "step": 36845
-    },
-    {
-      "epoch": 19.215645371577576,
-      "grad_norm": 1.5605173110961914,
-      "learning_rate": 6.348442211055276e-05,
-      "loss": 5.4108,
-      "step": 36846
-    },
-    {
-      "epoch": 19.216166883963496,
-      "grad_norm": 1.5346235036849976,
-      "learning_rate": 6.348341708542714e-05,
-      "loss": 5.419,
-      "step": 36847
-    },
-    {
-      "epoch": 19.216688396349415,
-      "grad_norm": 1.381447672843933,
-      "learning_rate": 6.348241206030152e-05,
-      "loss": 4.5753,
-      "step": 36848
-    },
-    {
-      "epoch": 19.21720990873533,
-      "grad_norm": 1.6419659852981567,
-      "learning_rate": 6.348140703517588e-05,
-      "loss": 5.0773,
-      "step": 36849
-    },
-    {
-      "epoch": 19.21773142112125,
-      "grad_norm": 1.4510810375213623,
-      "learning_rate": 6.348040201005026e-05,
-      "loss": 5.709,
-      "step": 36850
-    },
-    {
-      "epoch": 19.21825293350717,
-      "grad_norm": 1.534165859222412,
-      "learning_rate": 6.347939698492462e-05,
-      "loss": 5.59,
-      "step": 36851
-    },
-    {
-      "epoch": 19.21877444589309,
-      "grad_norm": 1.5674177408218384,
-      "learning_rate": 6.3478391959799e-05,
-      "loss": 5.2574,
-      "step": 36852
-    },
-    {
-      "epoch": 19.21929595827901,
-      "grad_norm": 1.443166971206665,
-      "learning_rate": 6.347738693467336e-05,
-      "loss": 5.6958,
-      "step": 36853
-    },
-    {
-      "epoch": 19.21981747066493,
-      "grad_norm": 1.4661056995391846,
-      "learning_rate": 6.347638190954774e-05,
-      "loss": 5.1274,
-      "step": 36854
-    },
-    {
-      "epoch": 19.220338983050848,
-      "grad_norm": 1.628048300743103,
-      "learning_rate": 6.34753768844221e-05,
-      "loss": 5.0349,
-      "step": 36855
-    },
-    {
-      "epoch": 19.220860495436767,
-      "grad_norm": 1.49128258228302,
-      "learning_rate": 6.347437185929648e-05,
-      "loss": 5.4518,
-      "step": 36856
-    },
-    {
-      "epoch": 19.221382007822687,
-      "grad_norm": 1.4849789142608643,
-      "learning_rate": 6.347336683417086e-05,
-      "loss": 5.6713,
-      "step": 36857
-    },
-    {
-      "epoch": 19.221903520208606,
-      "grad_norm": 1.6297568082809448,
-      "learning_rate": 6.347236180904524e-05,
-      "loss": 4.7927,
-      "step": 36858
-    },
-    {
-      "epoch": 19.222425032594526,
-      "grad_norm": 1.4678562879562378,
-      "learning_rate": 6.34713567839196e-05,
-      "loss": 5.3589,
-      "step": 36859
-    },
-    {
-      "epoch": 19.22294654498044,
-      "grad_norm": 1.4813907146453857,
-      "learning_rate": 6.347035175879398e-05,
-      "loss": 5.4533,
-      "step": 36860
-    },
-    {
-      "epoch": 19.22346805736636,
-      "grad_norm": 1.5493189096450806,
-      "learning_rate": 6.346934673366835e-05,
-      "loss": 5.3319,
-      "step": 36861
-    },
-    {
-      "epoch": 19.22398956975228,
-      "grad_norm": 1.4202569723129272,
-      "learning_rate": 6.346834170854271e-05,
-      "loss": 4.9688,
-      "step": 36862
-    },
-    {
-      "epoch": 19.2245110821382,
-      "grad_norm": 1.45529305934906,
-      "learning_rate": 6.346733668341709e-05,
-      "loss": 5.271,
-      "step": 36863
-    },
-    {
-      "epoch": 19.22503259452412,
-      "grad_norm": 1.5787216424942017,
-      "learning_rate": 6.346633165829145e-05,
-      "loss": 4.9847,
-      "step": 36864
-    },
-    {
-      "epoch": 19.22555410691004,
-      "grad_norm": 1.7754377126693726,
-      "learning_rate": 6.346532663316583e-05,
-      "loss": 4.4854,
-      "step": 36865
-    },
-    {
-      "epoch": 19.22607561929596,
-      "grad_norm": 1.476708173751831,
-      "learning_rate": 6.34643216080402e-05,
-      "loss": 5.4723,
-      "step": 36866
-    },
-    {
-      "epoch": 19.226597131681878,
-      "grad_norm": 1.489437460899353,
-      "learning_rate": 6.346331658291457e-05,
-      "loss": 5.0789,
-      "step": 36867
-    },
-    {
-      "epoch": 19.227118644067797,
-      "grad_norm": 1.492141604423523,
-      "learning_rate": 6.346231155778895e-05,
-      "loss": 5.3091,
-      "step": 36868
-    },
-    {
-      "epoch": 19.227640156453717,
-      "grad_norm": 1.4575738906860352,
-      "learning_rate": 6.346130653266333e-05,
-      "loss": 5.7123,
-      "step": 36869
-    },
-    {
-      "epoch": 19.228161668839636,
-      "grad_norm": 1.484814167022705,
-      "learning_rate": 6.346030150753769e-05,
-      "loss": 5.5521,
-      "step": 36870
-    },
-    {
-      "epoch": 19.228683181225556,
-      "grad_norm": 1.4843875169754028,
-      "learning_rate": 6.345929648241207e-05,
-      "loss": 5.4517,
-      "step": 36871
-    },
-    {
-      "epoch": 19.22920469361147,
-      "grad_norm": 1.4497064352035522,
-      "learning_rate": 6.345829145728643e-05,
-      "loss": 5.0019,
-      "step": 36872
-    },
-    {
-      "epoch": 19.22972620599739,
-      "grad_norm": 1.5412789583206177,
-      "learning_rate": 6.345728643216081e-05,
-      "loss": 4.7164,
-      "step": 36873
-    },
-    {
-      "epoch": 19.23024771838331,
-      "grad_norm": 1.4401829242706299,
-      "learning_rate": 6.345628140703518e-05,
-      "loss": 5.4663,
-      "step": 36874
-    },
-    {
-      "epoch": 19.23076923076923,
-      "grad_norm": 1.6556521654129028,
-      "learning_rate": 6.345527638190954e-05,
-      "loss": 5.1637,
-      "step": 36875
-    },
-    {
-      "epoch": 19.23129074315515,
-      "grad_norm": 1.4327027797698975,
-      "learning_rate": 6.345427135678392e-05,
-      "loss": 5.6371,
-      "step": 36876
-    },
-    {
-      "epoch": 19.23181225554107,
-      "grad_norm": 1.6042107343673706,
-      "learning_rate": 6.34532663316583e-05,
-      "loss": 4.8231,
-      "step": 36877
-    },
-    {
-      "epoch": 19.23233376792699,
-      "grad_norm": 1.4737534523010254,
-      "learning_rate": 6.345226130653267e-05,
-      "loss": 5.5331,
-      "step": 36878
-    },
-    {
-      "epoch": 19.232855280312908,
-      "grad_norm": 1.5860285758972168,
-      "learning_rate": 6.345125628140704e-05,
-      "loss": 4.9038,
-      "step": 36879
-    },
-    {
-      "epoch": 19.233376792698827,
-      "grad_norm": 1.3962498903274536,
-      "learning_rate": 6.345025125628142e-05,
-      "loss": 5.6514,
-      "step": 36880
-    },
-    {
-      "epoch": 19.233898305084747,
-      "grad_norm": 1.6860625743865967,
-      "learning_rate": 6.344924623115578e-05,
-      "loss": 5.0277,
-      "step": 36881
-    },
-    {
-      "epoch": 19.234419817470666,
-      "grad_norm": 1.4662601947784424,
-      "learning_rate": 6.344824120603016e-05,
-      "loss": 5.2007,
-      "step": 36882
-    },
-    {
-      "epoch": 19.234941329856586,
-      "grad_norm": 1.4557886123657227,
-      "learning_rate": 6.344723618090452e-05,
-      "loss": 5.5913,
-      "step": 36883
-    },
-    {
-      "epoch": 19.2354628422425,
-      "grad_norm": 1.5355714559555054,
-      "learning_rate": 6.34462311557789e-05,
-      "loss": 5.1394,
-      "step": 36884
-    },
-    {
-      "epoch": 19.23598435462842,
-      "grad_norm": 1.529517412185669,
-      "learning_rate": 6.344522613065326e-05,
-      "loss": 5.7525,
-      "step": 36885
-    },
-    {
-      "epoch": 19.23650586701434,
-      "grad_norm": 1.4590818881988525,
-      "learning_rate": 6.344422110552764e-05,
-      "loss": 5.1185,
-      "step": 36886
-    },
-    {
-      "epoch": 19.23702737940026,
-      "grad_norm": 1.6308363676071167,
-      "learning_rate": 6.344321608040201e-05,
-      "loss": 4.5298,
-      "step": 36887
-    },
-    {
-      "epoch": 19.23754889178618,
-      "grad_norm": 1.6101256608963013,
-      "learning_rate": 6.344221105527638e-05,
-      "loss": 5.475,
-      "step": 36888
-    },
-    {
-      "epoch": 19.2380704041721,
-      "grad_norm": 1.4836009740829468,
-      "learning_rate": 6.344120603015076e-05,
-      "loss": 5.3024,
-      "step": 36889
-    },
-    {
-      "epoch": 19.23859191655802,
-      "grad_norm": 1.5156887769699097,
-      "learning_rate": 6.344020100502513e-05,
-      "loss": 5.0775,
-      "step": 36890
-    },
-    {
-      "epoch": 19.239113428943938,
-      "grad_norm": 1.5670709609985352,
-      "learning_rate": 6.34391959798995e-05,
-      "loss": 5.6457,
-      "step": 36891
-    },
-    {
-      "epoch": 19.239634941329857,
-      "grad_norm": 1.5336781740188599,
-      "learning_rate": 6.343819095477387e-05,
-      "loss": 5.1949,
-      "step": 36892
-    },
-    {
-      "epoch": 19.240156453715777,
-      "grad_norm": 1.4362921714782715,
-      "learning_rate": 6.343718592964825e-05,
-      "loss": 5.5624,
-      "step": 36893
-    },
-    {
-      "epoch": 19.240677966101696,
-      "grad_norm": 1.687852144241333,
-      "learning_rate": 6.343618090452261e-05,
-      "loss": 5.0137,
-      "step": 36894
-    },
-    {
-      "epoch": 19.241199478487616,
-      "grad_norm": 1.4673957824707031,
-      "learning_rate": 6.343517587939699e-05,
-      "loss": 5.4657,
-      "step": 36895
-    },
-    {
-      "epoch": 19.24172099087353,
-      "grad_norm": 1.449641466140747,
-      "learning_rate": 6.343417085427135e-05,
-      "loss": 5.3665,
-      "step": 36896
-    },
-    {
-      "epoch": 19.24224250325945,
-      "grad_norm": 1.4761532545089722,
-      "learning_rate": 6.343316582914573e-05,
-      "loss": 4.6071,
-      "step": 36897
-    },
-    {
-      "epoch": 19.24276401564537,
-      "grad_norm": 1.4906343221664429,
-      "learning_rate": 6.34321608040201e-05,
-      "loss": 5.523,
-      "step": 36898
-    },
-    {
-      "epoch": 19.24328552803129,
-      "grad_norm": 1.6711719036102295,
-      "learning_rate": 6.343115577889447e-05,
-      "loss": 5.0888,
-      "step": 36899
-    },
-    {
-      "epoch": 19.24380704041721,
-      "grad_norm": 1.5183913707733154,
-      "learning_rate": 6.343015075376885e-05,
-      "loss": 5.4914,
-      "step": 36900
-    },
-    {
-      "epoch": 19.24432855280313,
-      "grad_norm": 1.5291370153427124,
-      "learning_rate": 6.342914572864323e-05,
-      "loss": 5.1587,
-      "step": 36901
-    },
-    {
-      "epoch": 19.24485006518905,
-      "grad_norm": 1.5439311265945435,
-      "learning_rate": 6.342814070351759e-05,
-      "loss": 4.7495,
-      "step": 36902
-    },
-    {
-      "epoch": 19.245371577574968,
-      "grad_norm": 1.484724760055542,
-      "learning_rate": 6.342713567839196e-05,
-      "loss": 5.1885,
-      "step": 36903
-    },
-    {
-      "epoch": 19.245893089960887,
-      "grad_norm": 1.4891382455825806,
-      "learning_rate": 6.342613065326634e-05,
-      "loss": 5.5848,
-      "step": 36904
-    },
-    {
-      "epoch": 19.246414602346807,
-      "grad_norm": 1.5138885974884033,
-      "learning_rate": 6.34251256281407e-05,
-      "loss": 5.3052,
-      "step": 36905
-    },
-    {
-      "epoch": 19.246936114732726,
-      "grad_norm": 1.6237576007843018,
-      "learning_rate": 6.342412060301508e-05,
-      "loss": 4.905,
-      "step": 36906
-    },
-    {
-      "epoch": 19.247457627118646,
-      "grad_norm": 1.4857239723205566,
-      "learning_rate": 6.342311557788944e-05,
-      "loss": 5.2908,
-      "step": 36907
-    },
-    {
-      "epoch": 19.24797913950456,
-      "grad_norm": 1.6112921237945557,
-      "learning_rate": 6.342211055276382e-05,
-      "loss": 4.8324,
-      "step": 36908
-    },
-    {
-      "epoch": 19.24850065189048,
-      "grad_norm": 1.4779003858566284,
-      "learning_rate": 6.34211055276382e-05,
-      "loss": 5.6801,
-      "step": 36909
-    },
-    {
-      "epoch": 19.2490221642764,
-      "grad_norm": 1.391730546951294,
-      "learning_rate": 6.342010050251258e-05,
-      "loss": 5.4727,
-      "step": 36910
-    },
-    {
-      "epoch": 19.24954367666232,
-      "grad_norm": 1.5292145013809204,
-      "learning_rate": 6.341909547738694e-05,
-      "loss": 4.8833,
-      "step": 36911
-    },
-    {
-      "epoch": 19.25006518904824,
-      "grad_norm": 1.4813306331634521,
-      "learning_rate": 6.341809045226132e-05,
-      "loss": 5.2467,
-      "step": 36912
-    },
-    {
-      "epoch": 19.25058670143416,
-      "grad_norm": 1.4558688402175903,
-      "learning_rate": 6.341708542713568e-05,
-      "loss": 5.5094,
-      "step": 36913
-    },
-    {
-      "epoch": 19.25110821382008,
-      "grad_norm": 1.5205247402191162,
-      "learning_rate": 6.341608040201006e-05,
-      "loss": 5.0898,
-      "step": 36914
-    },
-    {
-      "epoch": 19.251629726205998,
-      "grad_norm": 1.5173863172531128,
-      "learning_rate": 6.341507537688442e-05,
-      "loss": 5.4024,
-      "step": 36915
-    },
-    {
-      "epoch": 19.252151238591917,
-      "grad_norm": 1.5855995416641235,
-      "learning_rate": 6.341407035175879e-05,
-      "loss": 5.2493,
-      "step": 36916
-    },
-    {
-      "epoch": 19.252672750977837,
-      "grad_norm": 1.4934707880020142,
-      "learning_rate": 6.341306532663317e-05,
-      "loss": 4.9347,
-      "step": 36917
-    },
-    {
-      "epoch": 19.253194263363756,
-      "grad_norm": 1.5738959312438965,
-      "learning_rate": 6.341206030150753e-05,
-      "loss": 5.0986,
-      "step": 36918
-    },
-    {
-      "epoch": 19.253715775749676,
-      "grad_norm": 1.435224175453186,
-      "learning_rate": 6.341105527638191e-05,
-      "loss": 5.2367,
-      "step": 36919
-    },
-    {
-      "epoch": 19.25423728813559,
-      "grad_norm": 1.5308747291564941,
-      "learning_rate": 6.341005025125629e-05,
-      "loss": 4.9427,
-      "step": 36920
-    },
-    {
-      "epoch": 19.25475880052151,
-      "grad_norm": 1.5276764631271362,
-      "learning_rate": 6.340904522613066e-05,
-      "loss": 5.0881,
-      "step": 36921
-    },
-    {
-      "epoch": 19.25528031290743,
-      "grad_norm": 1.3677276372909546,
-      "learning_rate": 6.340804020100503e-05,
-      "loss": 5.5581,
-      "step": 36922
-    },
-    {
-      "epoch": 19.25580182529335,
-      "grad_norm": 1.5276122093200684,
-      "learning_rate": 6.34070351758794e-05,
-      "loss": 5.3721,
-      "step": 36923
-    },
-    {
-      "epoch": 19.25632333767927,
-      "grad_norm": 1.3930071592330933,
-      "learning_rate": 6.340603015075377e-05,
-      "loss": 5.5698,
-      "step": 36924
-    },
-    {
-      "epoch": 19.25684485006519,
-      "grad_norm": 1.5334687232971191,
-      "learning_rate": 6.340502512562815e-05,
-      "loss": 5.4448,
-      "step": 36925
-    },
-    {
-      "epoch": 19.25736636245111,
-      "grad_norm": 1.5532457828521729,
-      "learning_rate": 6.340402010050251e-05,
-      "loss": 5.3501,
-      "step": 36926
-    },
-    {
-      "epoch": 19.257887874837028,
-      "grad_norm": 1.536877989768982,
-      "learning_rate": 6.340301507537689e-05,
-      "loss": 5.756,
-      "step": 36927
-    },
-    {
-      "epoch": 19.258409387222947,
-      "grad_norm": 1.5355427265167236,
-      "learning_rate": 6.340201005025125e-05,
-      "loss": 5.2108,
-      "step": 36928
-    },
-    {
-      "epoch": 19.258930899608867,
-      "grad_norm": 1.4861279726028442,
-      "learning_rate": 6.340100502512563e-05,
-      "loss": 5.3329,
-      "step": 36929
-    },
-    {
-      "epoch": 19.259452411994786,
-      "grad_norm": 1.4631205797195435,
-      "learning_rate": 6.340000000000001e-05,
-      "loss": 5.1522,
-      "step": 36930
-    },
-    {
-      "epoch": 19.259973924380706,
-      "grad_norm": 1.4840400218963623,
-      "learning_rate": 6.339899497487437e-05,
-      "loss": 5.1723,
-      "step": 36931
-    },
-    {
-      "epoch": 19.26049543676662,
-      "grad_norm": 1.5493239164352417,
-      "learning_rate": 6.339798994974875e-05,
-      "loss": 5.145,
-      "step": 36932
-    },
-    {
-      "epoch": 19.26101694915254,
-      "grad_norm": 1.5541789531707764,
-      "learning_rate": 6.339698492462312e-05,
-      "loss": 5.6941,
-      "step": 36933
-    },
-    {
-      "epoch": 19.26153846153846,
-      "grad_norm": 1.593537449836731,
-      "learning_rate": 6.33959798994975e-05,
-      "loss": 5.5049,
-      "step": 36934
-    },
-    {
-      "epoch": 19.26205997392438,
-      "grad_norm": 1.5609166622161865,
-      "learning_rate": 6.339497487437186e-05,
-      "loss": 5.4556,
-      "step": 36935
-    },
-    {
-      "epoch": 19.2625814863103,
-      "grad_norm": 1.4707036018371582,
-      "learning_rate": 6.339396984924624e-05,
-      "loss": 5.2974,
-      "step": 36936
-    },
-    {
-      "epoch": 19.26310299869622,
-      "grad_norm": 1.4938044548034668,
-      "learning_rate": 6.33929648241206e-05,
-      "loss": 5.3953,
-      "step": 36937
-    },
-    {
-      "epoch": 19.26362451108214,
-      "grad_norm": 1.614590048789978,
-      "learning_rate": 6.339195979899498e-05,
-      "loss": 5.3994,
-      "step": 36938
-    },
-    {
-      "epoch": 19.264146023468058,
-      "grad_norm": 1.5082640647888184,
-      "learning_rate": 6.339095477386934e-05,
-      "loss": 5.241,
-      "step": 36939
-    },
-    {
-      "epoch": 19.264667535853977,
-      "grad_norm": 1.6660542488098145,
-      "learning_rate": 6.338994974874372e-05,
-      "loss": 4.5689,
-      "step": 36940
-    },
-    {
-      "epoch": 19.265189048239897,
-      "grad_norm": 1.620902419090271,
-      "learning_rate": 6.33889447236181e-05,
-      "loss": 5.2737,
-      "step": 36941
-    },
-    {
-      "epoch": 19.265710560625816,
-      "grad_norm": 1.464135766029358,
-      "learning_rate": 6.338793969849246e-05,
-      "loss": 5.2803,
-      "step": 36942
-    },
-    {
-      "epoch": 19.266232073011736,
-      "grad_norm": 1.5170650482177734,
-      "learning_rate": 6.338693467336684e-05,
-      "loss": 5.4217,
-      "step": 36943
-    },
-    {
-      "epoch": 19.26675358539765,
-      "grad_norm": 1.6289030313491821,
-      "learning_rate": 6.33859296482412e-05,
-      "loss": 4.9642,
-      "step": 36944
-    },
-    {
-      "epoch": 19.26727509778357,
-      "grad_norm": 1.6857249736785889,
-      "learning_rate": 6.338492462311558e-05,
-      "loss": 5.2609,
-      "step": 36945
-    },
-    {
-      "epoch": 19.26779661016949,
-      "grad_norm": 1.5560426712036133,
-      "learning_rate": 6.338391959798995e-05,
-      "loss": 5.1615,
-      "step": 36946
-    },
-    {
-      "epoch": 19.26831812255541,
-      "grad_norm": 1.5423920154571533,
-      "learning_rate": 6.338291457286432e-05,
-      "loss": 5.5504,
-      "step": 36947
-    },
-    {
-      "epoch": 19.26883963494133,
-      "grad_norm": 1.5581380128860474,
-      "learning_rate": 6.338190954773869e-05,
-      "loss": 5.2994,
-      "step": 36948
-    },
-    {
-      "epoch": 19.26936114732725,
-      "grad_norm": 1.473510980606079,
-      "learning_rate": 6.338090452261307e-05,
-      "loss": 5.3017,
-      "step": 36949
-    },
-    {
-      "epoch": 19.26988265971317,
-      "grad_norm": 1.5097030401229858,
-      "learning_rate": 6.337989949748744e-05,
-      "loss": 5.4822,
-      "step": 36950
-    },
-    {
-      "epoch": 19.270404172099088,
-      "grad_norm": 1.6531695127487183,
-      "learning_rate": 6.337889447236182e-05,
-      "loss": 5.11,
-      "step": 36951
-    },
-    {
-      "epoch": 19.270925684485007,
-      "grad_norm": 1.4781410694122314,
-      "learning_rate": 6.337788944723619e-05,
-      "loss": 5.2756,
-      "step": 36952
-    },
-    {
-      "epoch": 19.271447196870927,
-      "grad_norm": 1.440588355064392,
-      "learning_rate": 6.337688442211056e-05,
-      "loss": 5.1442,
-      "step": 36953
-    },
-    {
-      "epoch": 19.271968709256846,
-      "grad_norm": 1.429126262664795,
-      "learning_rate": 6.337587939698493e-05,
-      "loss": 5.4935,
-      "step": 36954
-    },
-    {
-      "epoch": 19.272490221642762,
-      "grad_norm": 1.5640674829483032,
-      "learning_rate": 6.337487437185929e-05,
-      "loss": 5.1456,
-      "step": 36955
-    },
-    {
-      "epoch": 19.27301173402868,
-      "grad_norm": 1.5524988174438477,
-      "learning_rate": 6.337386934673367e-05,
-      "loss": 4.6682,
-      "step": 36956
-    },
-    {
-      "epoch": 19.2735332464146,
-      "grad_norm": 1.6039880514144897,
-      "learning_rate": 6.337286432160803e-05,
-      "loss": 5.557,
-      "step": 36957
-    },
-    {
-      "epoch": 19.27405475880052,
-      "grad_norm": 1.5884076356887817,
-      "learning_rate": 6.337185929648241e-05,
-      "loss": 5.2362,
-      "step": 36958
-    },
-    {
-      "epoch": 19.27457627118644,
-      "grad_norm": 1.5353890657424927,
-      "learning_rate": 6.337085427135678e-05,
-      "loss": 4.8334,
-      "step": 36959
-    },
-    {
-      "epoch": 19.27509778357236,
-      "grad_norm": 1.6452640295028687,
-      "learning_rate": 6.336984924623115e-05,
-      "loss": 5.3837,
-      "step": 36960
-    },
-    {
-      "epoch": 19.27561929595828,
-      "grad_norm": 1.5327543020248413,
-      "learning_rate": 6.336884422110553e-05,
-      "loss": 5.5065,
-      "step": 36961
-    },
-    {
-      "epoch": 19.2761408083442,
-      "grad_norm": 1.554150938987732,
-      "learning_rate": 6.336783919597991e-05,
-      "loss": 5.0363,
-      "step": 36962
-    },
-    {
-      "epoch": 19.276662320730118,
-      "grad_norm": 1.4831501245498657,
-      "learning_rate": 6.336683417085427e-05,
-      "loss": 4.9323,
-      "step": 36963
-    },
-    {
-      "epoch": 19.277183833116037,
-      "grad_norm": 1.4818572998046875,
-      "learning_rate": 6.336582914572865e-05,
-      "loss": 5.5608,
-      "step": 36964
-    },
-    {
-      "epoch": 19.277705345501957,
-      "grad_norm": 1.4605873823165894,
-      "learning_rate": 6.336482412060302e-05,
-      "loss": 5.2915,
-      "step": 36965
-    },
-    {
-      "epoch": 19.278226857887876,
-      "grad_norm": 1.5389604568481445,
-      "learning_rate": 6.33638190954774e-05,
-      "loss": 4.7589,
-      "step": 36966
-    },
-    {
-      "epoch": 19.278748370273792,
-      "grad_norm": 1.5020772218704224,
-      "learning_rate": 6.336281407035176e-05,
-      "loss": 5.5426,
-      "step": 36967
-    },
-    {
-      "epoch": 19.27926988265971,
-      "grad_norm": 1.5246011018753052,
-      "learning_rate": 6.336180904522612e-05,
-      "loss": 5.2249,
-      "step": 36968
-    },
-    {
-      "epoch": 19.27979139504563,
-      "grad_norm": 1.4716694355010986,
-      "learning_rate": 6.33608040201005e-05,
-      "loss": 5.5469,
-      "step": 36969
-    },
-    {
-      "epoch": 19.28031290743155,
-      "grad_norm": 1.5546784400939941,
-      "learning_rate": 6.335979899497488e-05,
-      "loss": 5.2756,
-      "step": 36970
-    },
-    {
-      "epoch": 19.28083441981747,
-      "grad_norm": 1.5236930847167969,
-      "learning_rate": 6.335879396984926e-05,
-      "loss": 5.084,
-      "step": 36971
-    },
-    {
-      "epoch": 19.28135593220339,
-      "grad_norm": 1.636731743812561,
-      "learning_rate": 6.335778894472362e-05,
-      "loss": 5.2955,
-      "step": 36972
-    },
-    {
-      "epoch": 19.28187744458931,
-      "grad_norm": 1.525723934173584,
-      "learning_rate": 6.3356783919598e-05,
-      "loss": 5.5118,
-      "step": 36973
-    },
-    {
-      "epoch": 19.28239895697523,
-      "grad_norm": 1.5043638944625854,
-      "learning_rate": 6.335577889447236e-05,
-      "loss": 5.458,
-      "step": 36974
-    },
-    {
-      "epoch": 19.282920469361148,
-      "grad_norm": 1.5032025575637817,
-      "learning_rate": 6.335477386934674e-05,
-      "loss": 5.5807,
-      "step": 36975
-    },
-    {
-      "epoch": 19.283441981747067,
-      "grad_norm": 1.4374566078186035,
-      "learning_rate": 6.33537688442211e-05,
-      "loss": 5.7435,
-      "step": 36976
-    },
-    {
-      "epoch": 19.283963494132987,
-      "grad_norm": 1.5648101568222046,
-      "learning_rate": 6.335276381909548e-05,
-      "loss": 5.6351,
-      "step": 36977
-    },
-    {
-      "epoch": 19.284485006518906,
-      "grad_norm": 1.6216416358947754,
-      "learning_rate": 6.335175879396985e-05,
-      "loss": 5.6382,
-      "step": 36978
-    },
-    {
-      "epoch": 19.285006518904822,
-      "grad_norm": 1.4489424228668213,
-      "learning_rate": 6.335075376884423e-05,
-      "loss": 5.7505,
-      "step": 36979
-    },
-    {
-      "epoch": 19.285528031290742,
-      "grad_norm": 1.5426013469696045,
-      "learning_rate": 6.334974874371859e-05,
-      "loss": 5.5321,
-      "step": 36980
-    },
-    {
-      "epoch": 19.28604954367666,
-      "grad_norm": 1.529561161994934,
-      "learning_rate": 6.334874371859297e-05,
-      "loss": 5.1118,
-      "step": 36981
-    },
-    {
-      "epoch": 19.28657105606258,
-      "grad_norm": 1.445841908454895,
-      "learning_rate": 6.334773869346734e-05,
-      "loss": 5.66,
-      "step": 36982
-    },
-    {
-      "epoch": 19.2870925684485,
-      "grad_norm": 1.5005908012390137,
-      "learning_rate": 6.334673366834171e-05,
-      "loss": 5.524,
-      "step": 36983
-    },
-    {
-      "epoch": 19.28761408083442,
-      "grad_norm": 1.5497288703918457,
-      "learning_rate": 6.334572864321609e-05,
-      "loss": 5.4262,
-      "step": 36984
-    },
-    {
-      "epoch": 19.28813559322034,
-      "grad_norm": 1.54937744140625,
-      "learning_rate": 6.334472361809045e-05,
-      "loss": 5.3626,
-      "step": 36985
-    },
-    {
-      "epoch": 19.28865710560626,
-      "grad_norm": 1.5920655727386475,
-      "learning_rate": 6.334371859296483e-05,
-      "loss": 5.1219,
-      "step": 36986
-    },
-    {
-      "epoch": 19.289178617992178,
-      "grad_norm": 1.4081624746322632,
-      "learning_rate": 6.33427135678392e-05,
-      "loss": 5.7339,
-      "step": 36987
-    },
-    {
-      "epoch": 19.289700130378097,
-      "grad_norm": 1.5500937700271606,
-      "learning_rate": 6.334170854271357e-05,
-      "loss": 5.3096,
-      "step": 36988
-    },
-    {
-      "epoch": 19.290221642764017,
-      "grad_norm": 1.564881443977356,
-      "learning_rate": 6.334070351758794e-05,
-      "loss": 5.0718,
-      "step": 36989
-    },
-    {
-      "epoch": 19.290743155149936,
-      "grad_norm": 1.5192279815673828,
-      "learning_rate": 6.333969849246231e-05,
-      "loss": 5.0125,
-      "step": 36990
-    },
-    {
-      "epoch": 19.291264667535852,
-      "grad_norm": 1.4879258871078491,
-      "learning_rate": 6.333869346733669e-05,
-      "loss": 5.0246,
-      "step": 36991
-    },
-    {
-      "epoch": 19.291786179921772,
-      "grad_norm": 1.4760690927505493,
-      "learning_rate": 6.333768844221107e-05,
-      "loss": 5.3884,
-      "step": 36992
-    },
-    {
-      "epoch": 19.29230769230769,
-      "grad_norm": 1.4323617219924927,
-      "learning_rate": 6.333668341708543e-05,
-      "loss": 5.6063,
-      "step": 36993
-    },
-    {
-      "epoch": 19.29282920469361,
-      "grad_norm": 1.4789483547210693,
-      "learning_rate": 6.333567839195981e-05,
-      "loss": 5.4262,
-      "step": 36994
-    },
-    {
-      "epoch": 19.29335071707953,
-      "grad_norm": 1.4340264797210693,
-      "learning_rate": 6.333467336683418e-05,
-      "loss": 5.5647,
-      "step": 36995
-    },
-    {
-      "epoch": 19.29387222946545,
-      "grad_norm": 1.4710607528686523,
-      "learning_rate": 6.333366834170854e-05,
-      "loss": 5.6171,
-      "step": 36996
-    },
-    {
-      "epoch": 19.29439374185137,
-      "grad_norm": 1.5966708660125732,
-      "learning_rate": 6.333266331658292e-05,
-      "loss": 5.3733,
-      "step": 36997
-    },
-    {
-      "epoch": 19.29491525423729,
-      "grad_norm": 1.5931636095046997,
-      "learning_rate": 6.333165829145728e-05,
-      "loss": 5.1957,
-      "step": 36998
-    },
-    {
-      "epoch": 19.295436766623208,
-      "grad_norm": 1.5114933252334595,
-      "learning_rate": 6.333065326633166e-05,
-      "loss": 5.2603,
-      "step": 36999
-    },
-    {
-      "epoch": 19.295958279009128,
-      "grad_norm": 1.4077621698379517,
-      "learning_rate": 6.332964824120602e-05,
-      "loss": 5.1852,
-      "step": 37000
-    },
-    {
-      "epoch": 19.296479791395047,
-      "grad_norm": 1.5346025228500366,
-      "learning_rate": 6.33286432160804e-05,
-      "loss": 5.1795,
-      "step": 37001
-    },
-    {
-      "epoch": 19.297001303780966,
-      "grad_norm": 1.7000153064727783,
-      "learning_rate": 6.332763819095478e-05,
-      "loss": 4.8945,
-      "step": 37002
-    },
-    {
-      "epoch": 19.297522816166882,
-      "grad_norm": 1.5090970993041992,
-      "learning_rate": 6.332663316582916e-05,
-      "loss": 5.0531,
-      "step": 37003
-    },
-    {
-      "epoch": 19.298044328552802,
-      "grad_norm": 1.6013832092285156,
-      "learning_rate": 6.332562814070352e-05,
-      "loss": 5.0197,
-      "step": 37004
-    },
-    {
-      "epoch": 19.29856584093872,
-      "grad_norm": 1.710984468460083,
-      "learning_rate": 6.33246231155779e-05,
-      "loss": 5.0483,
-      "step": 37005
-    },
-    {
-      "epoch": 19.29908735332464,
-      "grad_norm": 1.511130928993225,
-      "learning_rate": 6.332361809045226e-05,
-      "loss": 4.8332,
-      "step": 37006
-    },
-    {
-      "epoch": 19.29960886571056,
-      "grad_norm": 1.4409373998641968,
-      "learning_rate": 6.332261306532664e-05,
-      "loss": 5.5192,
-      "step": 37007
-    },
-    {
-      "epoch": 19.30013037809648,
-      "grad_norm": 1.410717248916626,
-      "learning_rate": 6.3321608040201e-05,
-      "loss": 5.8575,
-      "step": 37008
-    },
-    {
-      "epoch": 19.3006518904824,
-      "grad_norm": 1.4723435640335083,
-      "learning_rate": 6.332060301507537e-05,
-      "loss": 5.5105,
-      "step": 37009
-    },
-    {
-      "epoch": 19.30117340286832,
-      "grad_norm": 1.4109721183776855,
-      "learning_rate": 6.331959798994975e-05,
-      "loss": 5.1421,
-      "step": 37010
-    },
-    {
-      "epoch": 19.301694915254238,
-      "grad_norm": 1.450457215309143,
-      "learning_rate": 6.331859296482413e-05,
-      "loss": 5.5569,
-      "step": 37011
-    },
-    {
-      "epoch": 19.302216427640158,
-      "grad_norm": 1.3753290176391602,
-      "learning_rate": 6.33175879396985e-05,
-      "loss": 5.4319,
-      "step": 37012
-    },
-    {
-      "epoch": 19.302737940026077,
-      "grad_norm": 1.5892975330352783,
-      "learning_rate": 6.331658291457287e-05,
-      "loss": 5.2195,
-      "step": 37013
-    },
-    {
-      "epoch": 19.303259452411996,
-      "grad_norm": 1.611024022102356,
-      "learning_rate": 6.331557788944725e-05,
-      "loss": 5.1911,
-      "step": 37014
-    },
-    {
-      "epoch": 19.303780964797912,
-      "grad_norm": 1.5561045408248901,
-      "learning_rate": 6.331457286432161e-05,
-      "loss": 5.0182,
-      "step": 37015
-    },
-    {
-      "epoch": 19.304302477183832,
-      "grad_norm": 1.3958892822265625,
-      "learning_rate": 6.331356783919599e-05,
-      "loss": 5.5558,
-      "step": 37016
-    },
-    {
-      "epoch": 19.30482398956975,
-      "grad_norm": 1.5448585748672485,
-      "learning_rate": 6.331256281407035e-05,
-      "loss": 4.0007,
-      "step": 37017
-    },
-    {
-      "epoch": 19.30534550195567,
-      "grad_norm": 1.4566551446914673,
-      "learning_rate": 6.331155778894473e-05,
-      "loss": 5.4945,
-      "step": 37018
-    },
-    {
-      "epoch": 19.30586701434159,
-      "grad_norm": 1.5056254863739014,
-      "learning_rate": 6.33105527638191e-05,
-      "loss": 4.992,
-      "step": 37019
-    },
-    {
-      "epoch": 19.30638852672751,
-      "grad_norm": 1.4802403450012207,
-      "learning_rate": 6.330954773869347e-05,
-      "loss": 5.1516,
-      "step": 37020
-    },
-    {
-      "epoch": 19.30691003911343,
-      "grad_norm": 1.5480082035064697,
-      "learning_rate": 6.330854271356784e-05,
-      "loss": 5.5624,
-      "step": 37021
-    },
-    {
-      "epoch": 19.30743155149935,
-      "grad_norm": 1.5364937782287598,
-      "learning_rate": 6.330753768844221e-05,
-      "loss": 5.049,
-      "step": 37022
-    },
-    {
-      "epoch": 19.307953063885268,
-      "grad_norm": 1.6175222396850586,
-      "learning_rate": 6.330653266331659e-05,
-      "loss": 4.6898,
-      "step": 37023
-    },
-    {
-      "epoch": 19.308474576271188,
-      "grad_norm": 1.3721835613250732,
-      "learning_rate": 6.330552763819096e-05,
-      "loss": 5.1701,
-      "step": 37024
-    },
-    {
-      "epoch": 19.308996088657107,
-      "grad_norm": 1.5426119565963745,
-      "learning_rate": 6.330452261306533e-05,
-      "loss": 5.1736,
-      "step": 37025
-    },
-    {
-      "epoch": 19.309517601043027,
-      "grad_norm": 1.5243970155715942,
-      "learning_rate": 6.33035175879397e-05,
-      "loss": 5.1837,
-      "step": 37026
-    },
-    {
-      "epoch": 19.310039113428942,
-      "grad_norm": 1.519598126411438,
-      "learning_rate": 6.330251256281408e-05,
-      "loss": 5.479,
-      "step": 37027
-    },
-    {
-      "epoch": 19.310560625814862,
-      "grad_norm": 1.491750717163086,
-      "learning_rate": 6.330150753768844e-05,
-      "loss": 4.9179,
-      "step": 37028
-    },
-    {
-      "epoch": 19.31108213820078,
-      "grad_norm": 1.5575309991836548,
-      "learning_rate": 6.330050251256282e-05,
-      "loss": 4.5672,
-      "step": 37029
-    },
-    {
-      "epoch": 19.3116036505867,
-      "grad_norm": 1.5337893962860107,
-      "learning_rate": 6.329949748743718e-05,
-      "loss": 5.4104,
-      "step": 37030
-    },
-    {
-      "epoch": 19.31212516297262,
-      "grad_norm": 1.444305181503296,
-      "learning_rate": 6.329849246231156e-05,
-      "loss": 5.0335,
-      "step": 37031
-    },
-    {
-      "epoch": 19.31264667535854,
-      "grad_norm": 1.3032722473144531,
-      "learning_rate": 6.329748743718594e-05,
-      "loss": 5.132,
-      "step": 37032
-    },
-    {
-      "epoch": 19.31316818774446,
-      "grad_norm": 1.4571813344955444,
-      "learning_rate": 6.329648241206032e-05,
-      "loss": 5.6325,
-      "step": 37033
-    },
-    {
-      "epoch": 19.31368970013038,
-      "grad_norm": 1.3492635488510132,
-      "learning_rate": 6.329547738693468e-05,
-      "loss": 5.0656,
-      "step": 37034
-    },
-    {
-      "epoch": 19.314211212516298,
-      "grad_norm": 1.4143400192260742,
-      "learning_rate": 6.329447236180904e-05,
-      "loss": 5.2471,
-      "step": 37035
-    },
-    {
-      "epoch": 19.314732724902218,
-      "grad_norm": 1.5203739404678345,
-      "learning_rate": 6.329346733668342e-05,
-      "loss": 4.7781,
-      "step": 37036
-    },
-    {
-      "epoch": 19.315254237288137,
-      "grad_norm": 1.5239968299865723,
-      "learning_rate": 6.329246231155779e-05,
-      "loss": 5.3128,
-      "step": 37037
-    },
-    {
-      "epoch": 19.315775749674053,
-      "grad_norm": 1.4645259380340576,
-      "learning_rate": 6.329145728643216e-05,
-      "loss": 5.3126,
-      "step": 37038
-    },
-    {
-      "epoch": 19.316297262059972,
-      "grad_norm": 1.5467439889907837,
-      "learning_rate": 6.329045226130653e-05,
-      "loss": 5.4619,
-      "step": 37039
-    },
-    {
-      "epoch": 19.316818774445892,
-      "grad_norm": 1.5318442583084106,
-      "learning_rate": 6.32894472361809e-05,
-      "loss": 5.3047,
-      "step": 37040
-    },
-    {
-      "epoch": 19.31734028683181,
-      "grad_norm": 1.5713196992874146,
-      "learning_rate": 6.328844221105527e-05,
-      "loss": 5.1694,
-      "step": 37041
-    },
-    {
-      "epoch": 19.31786179921773,
-      "grad_norm": 1.5806161165237427,
-      "learning_rate": 6.328743718592965e-05,
-      "loss": 5.6307,
-      "step": 37042
-    },
-    {
-      "epoch": 19.31838331160365,
-      "grad_norm": 1.5741537809371948,
-      "learning_rate": 6.328643216080403e-05,
-      "loss": 5.3628,
-      "step": 37043
-    },
-    {
-      "epoch": 19.31890482398957,
-      "grad_norm": 1.5723493099212646,
-      "learning_rate": 6.32854271356784e-05,
-      "loss": 5.5148,
-      "step": 37044
-    },
-    {
-      "epoch": 19.31942633637549,
-      "grad_norm": 1.5280975103378296,
-      "learning_rate": 6.328442211055277e-05,
-      "loss": 5.2795,
-      "step": 37045
-    },
-    {
-      "epoch": 19.31994784876141,
-      "grad_norm": 1.6120126247406006,
-      "learning_rate": 6.328341708542715e-05,
-      "loss": 4.9336,
-      "step": 37046
-    },
-    {
-      "epoch": 19.320469361147328,
-      "grad_norm": 1.616157054901123,
-      "learning_rate": 6.328241206030151e-05,
-      "loss": 5.4347,
-      "step": 37047
-    },
-    {
-      "epoch": 19.320990873533248,
-      "grad_norm": 1.4936193227767944,
-      "learning_rate": 6.328140703517588e-05,
-      "loss": 5.3858,
-      "step": 37048
-    },
-    {
-      "epoch": 19.321512385919167,
-      "grad_norm": 1.510345697402954,
-      "learning_rate": 6.328040201005025e-05,
-      "loss": 4.8475,
-      "step": 37049
-    },
-    {
-      "epoch": 19.322033898305083,
-      "grad_norm": 1.5535348653793335,
-      "learning_rate": 6.327939698492462e-05,
-      "loss": 5.1352,
-      "step": 37050
-    },
-    {
-      "epoch": 19.322555410691002,
-      "grad_norm": 1.4663493633270264,
-      "learning_rate": 6.3278391959799e-05,
-      "loss": 5.141,
-      "step": 37051
-    },
-    {
-      "epoch": 19.323076923076922,
-      "grad_norm": 1.5262188911437988,
-      "learning_rate": 6.327738693467337e-05,
-      "loss": 5.1044,
-      "step": 37052
-    },
-    {
-      "epoch": 19.32359843546284,
-      "grad_norm": 1.391670823097229,
-      "learning_rate": 6.327638190954775e-05,
-      "loss": 5.6062,
-      "step": 37053
-    },
-    {
-      "epoch": 19.32411994784876,
-      "grad_norm": 1.5654526948928833,
-      "learning_rate": 6.327537688442211e-05,
-      "loss": 5.4623,
-      "step": 37054
-    },
-    {
-      "epoch": 19.32464146023468,
-      "grad_norm": 1.4188475608825684,
-      "learning_rate": 6.327437185929649e-05,
-      "loss": 5.0733,
-      "step": 37055
-    },
-    {
-      "epoch": 19.3251629726206,
-      "grad_norm": 1.451438546180725,
-      "learning_rate": 6.327336683417086e-05,
-      "loss": 5.2397,
-      "step": 37056
-    },
-    {
-      "epoch": 19.32568448500652,
-      "grad_norm": 1.5199898481369019,
-      "learning_rate": 6.327236180904523e-05,
-      "loss": 5.1675,
-      "step": 37057
-    },
-    {
-      "epoch": 19.32620599739244,
-      "grad_norm": 1.4219231605529785,
-      "learning_rate": 6.32713567839196e-05,
-      "loss": 4.5553,
-      "step": 37058
-    },
-    {
-      "epoch": 19.326727509778358,
-      "grad_norm": 1.477130651473999,
-      "learning_rate": 6.327035175879398e-05,
-      "loss": 5.2435,
-      "step": 37059
-    },
-    {
-      "epoch": 19.327249022164278,
-      "grad_norm": 1.5212773084640503,
-      "learning_rate": 6.326934673366834e-05,
-      "loss": 5.2433,
-      "step": 37060
-    },
-    {
-      "epoch": 19.327770534550197,
-      "grad_norm": 1.4492864608764648,
-      "learning_rate": 6.326834170854272e-05,
-      "loss": 4.9764,
-      "step": 37061
-    },
-    {
-      "epoch": 19.328292046936113,
-      "grad_norm": 1.5227199792861938,
-      "learning_rate": 6.326733668341708e-05,
-      "loss": 5.5993,
-      "step": 37062
-    },
-    {
-      "epoch": 19.328813559322032,
-      "grad_norm": 1.3341858386993408,
-      "learning_rate": 6.326633165829146e-05,
-      "loss": 5.3899,
-      "step": 37063
-    },
-    {
-      "epoch": 19.329335071707952,
-      "grad_norm": 1.4651274681091309,
-      "learning_rate": 6.326532663316584e-05,
-      "loss": 5.3836,
-      "step": 37064
-    },
-    {
-      "epoch": 19.32985658409387,
-      "grad_norm": 1.557844877243042,
-      "learning_rate": 6.32643216080402e-05,
-      "loss": 5.8565,
-      "step": 37065
-    },
-    {
-      "epoch": 19.33037809647979,
-      "grad_norm": 1.4713866710662842,
-      "learning_rate": 6.326331658291458e-05,
-      "loss": 5.0659,
-      "step": 37066
-    },
-    {
-      "epoch": 19.33089960886571,
-      "grad_norm": 1.4952170848846436,
-      "learning_rate": 6.326231155778895e-05,
-      "loss": 5.5119,
-      "step": 37067
-    },
-    {
-      "epoch": 19.33142112125163,
-      "grad_norm": 1.4446187019348145,
-      "learning_rate": 6.326130653266332e-05,
-      "loss": 5.23,
-      "step": 37068
-    },
-    {
-      "epoch": 19.33194263363755,
-      "grad_norm": 1.3665767908096313,
-      "learning_rate": 6.326030150753769e-05,
-      "loss": 5.5613,
-      "step": 37069
-    },
-    {
-      "epoch": 19.33246414602347,
-      "grad_norm": 1.4433425664901733,
-      "learning_rate": 6.325929648241207e-05,
-      "loss": 5.3037,
-      "step": 37070
-    },
-    {
-      "epoch": 19.332985658409388,
-      "grad_norm": 1.4764549732208252,
-      "learning_rate": 6.325829145728643e-05,
-      "loss": 5.4244,
-      "step": 37071
-    },
-    {
-      "epoch": 19.333507170795308,
-      "grad_norm": 1.4227054119110107,
-      "learning_rate": 6.325728643216081e-05,
-      "loss": 5.5109,
-      "step": 37072
-    },
-    {
-      "epoch": 19.334028683181227,
-      "grad_norm": 1.590537667274475,
-      "learning_rate": 6.325628140703517e-05,
-      "loss": 4.9628,
-      "step": 37073
-    },
-    {
-      "epoch": 19.334550195567143,
-      "grad_norm": 1.5459904670715332,
-      "learning_rate": 6.325527638190955e-05,
-      "loss": 5.4186,
-      "step": 37074
-    },
-    {
-      "epoch": 19.335071707953063,
-      "grad_norm": 1.4941155910491943,
-      "learning_rate": 6.325427135678393e-05,
-      "loss": 5.5436,
-      "step": 37075
-    },
-    {
-      "epoch": 19.335593220338982,
-      "grad_norm": 1.5174111127853394,
-      "learning_rate": 6.325326633165829e-05,
-      "loss": 5.5859,
-      "step": 37076
-    },
-    {
-      "epoch": 19.3361147327249,
-      "grad_norm": 1.3993632793426514,
-      "learning_rate": 6.325226130653267e-05,
-      "loss": 5.5369,
-      "step": 37077
-    },
-    {
-      "epoch": 19.33663624511082,
-      "grad_norm": 1.520076036453247,
-      "learning_rate": 6.325125628140703e-05,
-      "loss": 4.9423,
-      "step": 37078
-    },
-    {
-      "epoch": 19.33715775749674,
-      "grad_norm": 1.476530909538269,
-      "learning_rate": 6.325025125628141e-05,
-      "loss": 5.491,
-      "step": 37079
-    },
-    {
-      "epoch": 19.33767926988266,
-      "grad_norm": 1.497899055480957,
-      "learning_rate": 6.324924623115578e-05,
-      "loss": 5.3347,
-      "step": 37080
-    },
-    {
-      "epoch": 19.33820078226858,
-      "grad_norm": 1.6110990047454834,
-      "learning_rate": 6.324824120603015e-05,
-      "loss": 5.0702,
-      "step": 37081
-    },
-    {
-      "epoch": 19.3387222946545,
-      "grad_norm": 1.511828064918518,
-      "learning_rate": 6.324723618090452e-05,
-      "loss": 4.9993,
-      "step": 37082
-    },
-    {
-      "epoch": 19.33924380704042,
-      "grad_norm": 1.3735371828079224,
-      "learning_rate": 6.32462311557789e-05,
-      "loss": 5.8972,
-      "step": 37083
-    },
-    {
-      "epoch": 19.339765319426338,
-      "grad_norm": 1.4447212219238281,
-      "learning_rate": 6.324522613065327e-05,
-      "loss": 5.4688,
-      "step": 37084
-    },
-    {
-      "epoch": 19.340286831812257,
-      "grad_norm": 1.4063836336135864,
-      "learning_rate": 6.324422110552765e-05,
-      "loss": 5.2246,
-      "step": 37085
-    },
-    {
-      "epoch": 19.340808344198173,
-      "grad_norm": 1.4194180965423584,
-      "learning_rate": 6.324321608040202e-05,
-      "loss": 5.5086,
-      "step": 37086
-    },
-    {
-      "epoch": 19.341329856584093,
-      "grad_norm": 1.440015196800232,
-      "learning_rate": 6.32422110552764e-05,
-      "loss": 5.0298,
-      "step": 37087
-    },
-    {
-      "epoch": 19.341851368970012,
-      "grad_norm": 1.549672245979309,
-      "learning_rate": 6.324120603015076e-05,
-      "loss": 5.1629,
-      "step": 37088
-    },
-    {
-      "epoch": 19.34237288135593,
-      "grad_norm": 1.545289158821106,
-      "learning_rate": 6.324020100502512e-05,
-      "loss": 5.3528,
-      "step": 37089
-    },
-    {
-      "epoch": 19.34289439374185,
-      "grad_norm": 1.5546116828918457,
-      "learning_rate": 6.32391959798995e-05,
-      "loss": 5.4503,
-      "step": 37090
-    },
-    {
-      "epoch": 19.34341590612777,
-      "grad_norm": 1.4881292581558228,
-      "learning_rate": 6.323819095477386e-05,
-      "loss": 5.4775,
-      "step": 37091
-    },
-    {
-      "epoch": 19.34393741851369,
-      "grad_norm": 1.4726693630218506,
-      "learning_rate": 6.323718592964824e-05,
-      "loss": 5.452,
-      "step": 37092
-    },
-    {
-      "epoch": 19.34445893089961,
-      "grad_norm": 1.5209449529647827,
-      "learning_rate": 6.32361809045226e-05,
-      "loss": 5.2722,
-      "step": 37093
-    },
-    {
-      "epoch": 19.34498044328553,
-      "grad_norm": 1.5259023904800415,
-      "learning_rate": 6.323517587939698e-05,
-      "loss": 5.3292,
-      "step": 37094
-    },
-    {
-      "epoch": 19.34550195567145,
-      "grad_norm": 1.4960318803787231,
-      "learning_rate": 6.323417085427136e-05,
-      "loss": 5.4169,
-      "step": 37095
-    },
-    {
-      "epoch": 19.346023468057368,
-      "grad_norm": 1.6557385921478271,
-      "learning_rate": 6.323316582914574e-05,
-      "loss": 5.1099,
-      "step": 37096
-    },
-    {
-      "epoch": 19.346544980443287,
-      "grad_norm": 1.5973485708236694,
-      "learning_rate": 6.32321608040201e-05,
-      "loss": 4.8045,
-      "step": 37097
-    },
-    {
-      "epoch": 19.347066492829203,
-      "grad_norm": 1.703015923500061,
-      "learning_rate": 6.323115577889448e-05,
-      "loss": 4.6729,
-      "step": 37098
-    },
-    {
-      "epoch": 19.347588005215123,
-      "grad_norm": 1.5253891944885254,
-      "learning_rate": 6.323015075376885e-05,
-      "loss": 5.37,
-      "step": 37099
-    },
-    {
-      "epoch": 19.348109517601042,
-      "grad_norm": 1.587241768836975,
-      "learning_rate": 6.322914572864322e-05,
-      "loss": 5.337,
-      "step": 37100
-    },
-    {
-      "epoch": 19.34863102998696,
-      "grad_norm": 1.5907340049743652,
-      "learning_rate": 6.322814070351759e-05,
-      "loss": 5.0856,
-      "step": 37101
-    },
-    {
-      "epoch": 19.34915254237288,
-      "grad_norm": 1.68035888671875,
-      "learning_rate": 6.322713567839195e-05,
-      "loss": 5.2875,
-      "step": 37102
-    },
-    {
-      "epoch": 19.3496740547588,
-      "grad_norm": 1.4841622114181519,
-      "learning_rate": 6.322613065326633e-05,
-      "loss": 4.9518,
-      "step": 37103
-    },
-    {
-      "epoch": 19.35019556714472,
-      "grad_norm": 1.5364917516708374,
-      "learning_rate": 6.322512562814071e-05,
-      "loss": 5.4003,
-      "step": 37104
-    },
-    {
-      "epoch": 19.35071707953064,
-      "grad_norm": 1.4059512615203857,
-      "learning_rate": 6.322412060301509e-05,
-      "loss": 5.2571,
-      "step": 37105
-    },
-    {
-      "epoch": 19.35123859191656,
-      "grad_norm": 1.6129995584487915,
-      "learning_rate": 6.322311557788945e-05,
-      "loss": 4.8882,
-      "step": 37106
-    },
-    {
-      "epoch": 19.35176010430248,
-      "grad_norm": 1.5219221115112305,
-      "learning_rate": 6.322211055276383e-05,
-      "loss": 5.4591,
-      "step": 37107
-    },
-    {
-      "epoch": 19.352281616688398,
-      "grad_norm": 1.5046744346618652,
-      "learning_rate": 6.322110552763819e-05,
-      "loss": 5.1932,
-      "step": 37108
-    },
-    {
-      "epoch": 19.352803129074317,
-      "grad_norm": 1.640858769416809,
-      "learning_rate": 6.322010050251257e-05,
-      "loss": 4.9884,
-      "step": 37109
-    },
-    {
-      "epoch": 19.353324641460233,
-      "grad_norm": 1.5399210453033447,
-      "learning_rate": 6.321909547738693e-05,
-      "loss": 5.4409,
-      "step": 37110
-    },
-    {
-      "epoch": 19.353846153846153,
-      "grad_norm": 1.4734032154083252,
-      "learning_rate": 6.321809045226131e-05,
-      "loss": 5.5055,
-      "step": 37111
-    },
-    {
-      "epoch": 19.354367666232072,
-      "grad_norm": 1.6703662872314453,
-      "learning_rate": 6.321708542713568e-05,
-      "loss": 5.5863,
-      "step": 37112
-    },
-    {
-      "epoch": 19.35488917861799,
-      "grad_norm": 1.5822257995605469,
-      "learning_rate": 6.321608040201005e-05,
-      "loss": 5.2479,
-      "step": 37113
-    },
-    {
-      "epoch": 19.35541069100391,
-      "grad_norm": 1.5929591655731201,
-      "learning_rate": 6.321507537688442e-05,
-      "loss": 5.2665,
-      "step": 37114
-    },
-    {
-      "epoch": 19.35593220338983,
-      "grad_norm": 1.5502634048461914,
-      "learning_rate": 6.32140703517588e-05,
-      "loss": 5.4112,
-      "step": 37115
-    },
-    {
-      "epoch": 19.35645371577575,
-      "grad_norm": 1.504163146018982,
-      "learning_rate": 6.321306532663317e-05,
-      "loss": 5.7628,
-      "step": 37116
-    },
-    {
-      "epoch": 19.35697522816167,
-      "grad_norm": 1.5411347150802612,
-      "learning_rate": 6.321206030150754e-05,
-      "loss": 5.441,
-      "step": 37117
-    },
-    {
-      "epoch": 19.35749674054759,
-      "grad_norm": 1.5393457412719727,
-      "learning_rate": 6.321105527638192e-05,
-      "loss": 5.4607,
-      "step": 37118
-    },
-    {
-      "epoch": 19.35801825293351,
-      "grad_norm": 1.5102952718734741,
-      "learning_rate": 6.321005025125628e-05,
-      "loss": 5.0934,
-      "step": 37119
-    },
-    {
-      "epoch": 19.358539765319428,
-      "grad_norm": 1.6449638605117798,
-      "learning_rate": 6.320904522613066e-05,
-      "loss": 5.2486,
-      "step": 37120
-    },
-    {
-      "epoch": 19.359061277705347,
-      "grad_norm": 1.4409879446029663,
-      "learning_rate": 6.320804020100502e-05,
-      "loss": 5.6407,
-      "step": 37121
-    },
-    {
-      "epoch": 19.359582790091263,
-      "grad_norm": 1.5536202192306519,
-      "learning_rate": 6.32070351758794e-05,
-      "loss": 5.259,
-      "step": 37122
-    },
-    {
-      "epoch": 19.360104302477183,
-      "grad_norm": 1.4153131246566772,
-      "learning_rate": 6.320603015075376e-05,
-      "loss": 5.1754,
-      "step": 37123
-    },
-    {
-      "epoch": 19.360625814863102,
-      "grad_norm": 1.4174671173095703,
-      "learning_rate": 6.320502512562814e-05,
-      "loss": 5.3541,
-      "step": 37124
-    },
-    {
-      "epoch": 19.36114732724902,
-      "grad_norm": 1.5260249376296997,
-      "learning_rate": 6.320402010050252e-05,
-      "loss": 4.8405,
-      "step": 37125
-    },
-    {
-      "epoch": 19.36166883963494,
-      "grad_norm": 1.4917182922363281,
-      "learning_rate": 6.32030150753769e-05,
-      "loss": 5.4925,
-      "step": 37126
-    },
-    {
-      "epoch": 19.36219035202086,
-      "grad_norm": 1.599554181098938,
-      "learning_rate": 6.320201005025126e-05,
-      "loss": 5.2647,
-      "step": 37127
-    },
-    {
-      "epoch": 19.36271186440678,
-      "grad_norm": 1.5066841840744019,
-      "learning_rate": 6.320100502512563e-05,
-      "loss": 5.1892,
-      "step": 37128
-    },
-    {
-      "epoch": 19.3632333767927,
-      "grad_norm": 1.503145694732666,
-      "learning_rate": 6.32e-05,
-      "loss": 4.9431,
-      "step": 37129
-    },
-    {
-      "epoch": 19.36375488917862,
-      "grad_norm": 1.4827722311019897,
-      "learning_rate": 6.319899497487437e-05,
-      "loss": 4.7366,
-      "step": 37130
-    },
-    {
-      "epoch": 19.36427640156454,
-      "grad_norm": 1.4827415943145752,
-      "learning_rate": 6.319798994974875e-05,
-      "loss": 5.32,
-      "step": 37131
-    },
-    {
-      "epoch": 19.364797913950458,
-      "grad_norm": 1.5232783555984497,
-      "learning_rate": 6.319698492462311e-05,
-      "loss": 5.3045,
-      "step": 37132
-    },
-    {
-      "epoch": 19.365319426336377,
-      "grad_norm": 1.5192553997039795,
-      "learning_rate": 6.319597989949749e-05,
-      "loss": 4.8675,
-      "step": 37133
-    },
-    {
-      "epoch": 19.365840938722293,
-      "grad_norm": 1.5345357656478882,
-      "learning_rate": 6.319497487437185e-05,
-      "loss": 5.319,
-      "step": 37134
-    },
-    {
-      "epoch": 19.366362451108213,
-      "grad_norm": 1.5717198848724365,
-      "learning_rate": 6.319396984924623e-05,
-      "loss": 5.2626,
-      "step": 37135
-    },
-    {
-      "epoch": 19.366883963494132,
-      "grad_norm": 1.5151216983795166,
-      "learning_rate": 6.319296482412061e-05,
-      "loss": 5.2011,
-      "step": 37136
-    },
-    {
-      "epoch": 19.36740547588005,
-      "grad_norm": 1.580197811126709,
-      "learning_rate": 6.319195979899499e-05,
-      "loss": 5.0186,
-      "step": 37137
-    },
-    {
-      "epoch": 19.36792698826597,
-      "grad_norm": 1.4931432008743286,
-      "learning_rate": 6.319095477386935e-05,
-      "loss": 5.0168,
-      "step": 37138
-    },
-    {
-      "epoch": 19.36844850065189,
-      "grad_norm": 1.5474936962127686,
-      "learning_rate": 6.318994974874373e-05,
-      "loss": 4.8129,
-      "step": 37139
-    },
-    {
-      "epoch": 19.36897001303781,
-      "grad_norm": 1.4531172513961792,
-      "learning_rate": 6.318894472361809e-05,
-      "loss": 5.4508,
-      "step": 37140
-    },
-    {
-      "epoch": 19.36949152542373,
-      "grad_norm": 1.5608268976211548,
-      "learning_rate": 6.318793969849246e-05,
-      "loss": 5.1669,
-      "step": 37141
-    },
-    {
-      "epoch": 19.37001303780965,
-      "grad_norm": 1.523646593093872,
-      "learning_rate": 6.318693467336684e-05,
-      "loss": 5.4248,
-      "step": 37142
-    },
-    {
-      "epoch": 19.37053455019557,
-      "grad_norm": 1.5354210138320923,
-      "learning_rate": 6.31859296482412e-05,
-      "loss": 5.3409,
-      "step": 37143
-    },
-    {
-      "epoch": 19.371056062581488,
-      "grad_norm": 1.5741533041000366,
-      "learning_rate": 6.318492462311558e-05,
-      "loss": 5.0989,
-      "step": 37144
-    },
-    {
-      "epoch": 19.371577574967404,
-      "grad_norm": 1.6116901636123657,
-      "learning_rate": 6.318391959798996e-05,
-      "loss": 5.3771,
-      "step": 37145
-    },
-    {
-      "epoch": 19.372099087353323,
-      "grad_norm": 1.4636410474777222,
-      "learning_rate": 6.318291457286433e-05,
-      "loss": 5.7453,
-      "step": 37146
-    },
-    {
-      "epoch": 19.372620599739243,
-      "grad_norm": 1.532753348350525,
-      "learning_rate": 6.31819095477387e-05,
-      "loss": 5.1635,
-      "step": 37147
-    },
-    {
-      "epoch": 19.373142112125162,
-      "grad_norm": 1.4354628324508667,
-      "learning_rate": 6.318090452261308e-05,
-      "loss": 5.078,
-      "step": 37148
-    },
-    {
-      "epoch": 19.37366362451108,
-      "grad_norm": 1.4830875396728516,
-      "learning_rate": 6.317989949748744e-05,
-      "loss": 5.2865,
-      "step": 37149
-    },
-    {
-      "epoch": 19.374185136897,
-      "grad_norm": 1.390494465827942,
-      "learning_rate": 6.317889447236182e-05,
-      "loss": 4.8135,
-      "step": 37150
-    },
-    {
-      "epoch": 19.37470664928292,
-      "grad_norm": 1.4069291353225708,
-      "learning_rate": 6.317788944723618e-05,
-      "loss": 5.5019,
-      "step": 37151
-    },
-    {
-      "epoch": 19.37522816166884,
-      "grad_norm": 1.520014762878418,
-      "learning_rate": 6.317688442211056e-05,
-      "loss": 5.469,
-      "step": 37152
-    },
-    {
-      "epoch": 19.37574967405476,
-      "grad_norm": 1.543834924697876,
-      "learning_rate": 6.317587939698492e-05,
-      "loss": 4.9698,
-      "step": 37153
-    },
-    {
-      "epoch": 19.37627118644068,
-      "grad_norm": 1.4796831607818604,
-      "learning_rate": 6.31748743718593e-05,
-      "loss": 5.5475,
-      "step": 37154
-    },
-    {
-      "epoch": 19.3767926988266,
-      "grad_norm": 1.471279263496399,
-      "learning_rate": 6.317386934673367e-05,
-      "loss": 4.8241,
-      "step": 37155
-    },
-    {
-      "epoch": 19.377314211212518,
-      "grad_norm": 1.569157361984253,
-      "learning_rate": 6.317286432160804e-05,
-      "loss": 5.7546,
-      "step": 37156
-    },
-    {
-      "epoch": 19.377835723598434,
-      "grad_norm": 1.5846130847930908,
-      "learning_rate": 6.317185929648242e-05,
-      "loss": 5.2128,
-      "step": 37157
-    },
-    {
-      "epoch": 19.378357235984353,
-      "grad_norm": 1.6638171672821045,
-      "learning_rate": 6.317085427135679e-05,
-      "loss": 5.3114,
-      "step": 37158
-    },
-    {
-      "epoch": 19.378878748370273,
-      "grad_norm": 1.4874213933944702,
-      "learning_rate": 6.316984924623116e-05,
-      "loss": 4.9927,
-      "step": 37159
-    },
-    {
-      "epoch": 19.379400260756192,
-      "grad_norm": 1.6096141338348389,
-      "learning_rate": 6.316884422110553e-05,
-      "loss": 4.9932,
-      "step": 37160
-    },
-    {
-      "epoch": 19.37992177314211,
-      "grad_norm": 1.4527568817138672,
-      "learning_rate": 6.31678391959799e-05,
-      "loss": 5.1799,
-      "step": 37161
-    },
-    {
-      "epoch": 19.38044328552803,
-      "grad_norm": 1.3724361658096313,
-      "learning_rate": 6.316683417085427e-05,
-      "loss": 5.3762,
-      "step": 37162
-    },
-    {
-      "epoch": 19.38096479791395,
-      "grad_norm": 1.448685884475708,
-      "learning_rate": 6.316582914572865e-05,
-      "loss": 5.766,
-      "step": 37163
-    },
-    {
-      "epoch": 19.38148631029987,
-      "grad_norm": 1.5163205862045288,
-      "learning_rate": 6.316482412060301e-05,
-      "loss": 5.1211,
-      "step": 37164
-    },
-    {
-      "epoch": 19.38200782268579,
-      "grad_norm": 1.5153307914733887,
-      "learning_rate": 6.316381909547739e-05,
-      "loss": 5.5643,
-      "step": 37165
-    },
-    {
-      "epoch": 19.38252933507171,
-      "grad_norm": 1.5981110334396362,
-      "learning_rate": 6.316281407035177e-05,
-      "loss": 4.7278,
-      "step": 37166
-    },
-    {
-      "epoch": 19.38305084745763,
-      "grad_norm": 1.5576859712600708,
-      "learning_rate": 6.316180904522615e-05,
-      "loss": 5.7755,
-      "step": 37167
-    },
-    {
-      "epoch": 19.383572359843548,
-      "grad_norm": 1.4697257280349731,
-      "learning_rate": 6.316080402010051e-05,
-      "loss": 5.3523,
-      "step": 37168
-    },
-    {
-      "epoch": 19.384093872229464,
-      "grad_norm": 1.5471277236938477,
-      "learning_rate": 6.315979899497487e-05,
-      "loss": 4.5826,
-      "step": 37169
-    },
-    {
-      "epoch": 19.384615384615383,
-      "grad_norm": 1.4236682653427124,
-      "learning_rate": 6.315879396984925e-05,
-      "loss": 5.5613,
-      "step": 37170
-    },
-    {
-      "epoch": 19.385136897001303,
-      "grad_norm": 1.5070534944534302,
-      "learning_rate": 6.315778894472362e-05,
-      "loss": 5.3282,
-      "step": 37171
-    },
-    {
-      "epoch": 19.385658409387222,
-      "grad_norm": 1.4455013275146484,
-      "learning_rate": 6.3156783919598e-05,
-      "loss": 5.2175,
-      "step": 37172
-    },
-    {
-      "epoch": 19.38617992177314,
-      "grad_norm": 1.4438649415969849,
-      "learning_rate": 6.315577889447236e-05,
-      "loss": 5.5264,
-      "step": 37173
-    },
-    {
-      "epoch": 19.38670143415906,
-      "grad_norm": 1.523566722869873,
-      "learning_rate": 6.315477386934674e-05,
-      "loss": 5.3955,
-      "step": 37174
-    },
-    {
-      "epoch": 19.38722294654498,
-      "grad_norm": 1.515112280845642,
-      "learning_rate": 6.31537688442211e-05,
-      "loss": 5.424,
-      "step": 37175
-    },
-    {
-      "epoch": 19.3877444589309,
-      "grad_norm": 1.6169953346252441,
-      "learning_rate": 6.315276381909548e-05,
-      "loss": 5.0866,
-      "step": 37176
-    },
-    {
-      "epoch": 19.38826597131682,
-      "grad_norm": 1.5474475622177124,
-      "learning_rate": 6.315175879396986e-05,
-      "loss": 5.1086,
-      "step": 37177
-    },
-    {
-      "epoch": 19.38878748370274,
-      "grad_norm": 1.5691883563995361,
-      "learning_rate": 6.315075376884423e-05,
-      "loss": 5.3893,
-      "step": 37178
-    },
-    {
-      "epoch": 19.38930899608866,
-      "grad_norm": 1.4614287614822388,
-      "learning_rate": 6.31497487437186e-05,
-      "loss": 4.8538,
-      "step": 37179
-    },
-    {
-      "epoch": 19.389830508474578,
-      "grad_norm": 1.499411940574646,
-      "learning_rate": 6.314874371859298e-05,
-      "loss": 5.592,
-      "step": 37180
-    },
-    {
-      "epoch": 19.390352020860494,
-      "grad_norm": 1.5720584392547607,
-      "learning_rate": 6.314773869346734e-05,
-      "loss": 5.2985,
-      "step": 37181
-    },
-    {
-      "epoch": 19.390873533246413,
-      "grad_norm": 1.6698482036590576,
-      "learning_rate": 6.31467336683417e-05,
-      "loss": 5.2644,
-      "step": 37182
-    },
-    {
-      "epoch": 19.391395045632333,
-      "grad_norm": 1.524125337600708,
-      "learning_rate": 6.314572864321608e-05,
-      "loss": 5.1625,
-      "step": 37183
-    },
-    {
-      "epoch": 19.391916558018252,
-      "grad_norm": 1.497400164604187,
-      "learning_rate": 6.314472361809045e-05,
-      "loss": 5.0787,
-      "step": 37184
-    },
-    {
-      "epoch": 19.39243807040417,
-      "grad_norm": 1.393250823020935,
-      "learning_rate": 6.314371859296482e-05,
-      "loss": 4.6639,
-      "step": 37185
-    },
-    {
-      "epoch": 19.39295958279009,
-      "grad_norm": 1.509372353553772,
-      "learning_rate": 6.31427135678392e-05,
-      "loss": 5.2969,
-      "step": 37186
-    },
-    {
-      "epoch": 19.39348109517601,
-      "grad_norm": 1.5587784051895142,
-      "learning_rate": 6.314170854271358e-05,
-      "loss": 5.5173,
-      "step": 37187
-    },
-    {
-      "epoch": 19.39400260756193,
-      "grad_norm": 1.4776206016540527,
-      "learning_rate": 6.314070351758794e-05,
-      "loss": 5.4593,
-      "step": 37188
-    },
-    {
-      "epoch": 19.39452411994785,
-      "grad_norm": 1.5518769025802612,
-      "learning_rate": 6.313969849246232e-05,
-      "loss": 4.6403,
-      "step": 37189
-    },
-    {
-      "epoch": 19.39504563233377,
-      "grad_norm": 1.533639669418335,
-      "learning_rate": 6.313869346733669e-05,
-      "loss": 5.1828,
-      "step": 37190
-    },
-    {
-      "epoch": 19.39556714471969,
-      "grad_norm": 1.6061030626296997,
-      "learning_rate": 6.313768844221106e-05,
-      "loss": 4.6874,
-      "step": 37191
-    },
-    {
-      "epoch": 19.396088657105608,
-      "grad_norm": 1.4598623514175415,
-      "learning_rate": 6.313668341708543e-05,
-      "loss": 5.1044,
-      "step": 37192
-    },
-    {
-      "epoch": 19.396610169491524,
-      "grad_norm": 1.4822458028793335,
-      "learning_rate": 6.31356783919598e-05,
-      "loss": 5.5344,
-      "step": 37193
-    },
-    {
-      "epoch": 19.397131681877443,
-      "grad_norm": 1.5150407552719116,
-      "learning_rate": 6.313467336683417e-05,
-      "loss": 5.1483,
-      "step": 37194
-    },
-    {
-      "epoch": 19.397653194263363,
-      "grad_norm": 1.5336297750473022,
-      "learning_rate": 6.313366834170853e-05,
-      "loss": 5.3455,
-      "step": 37195
-    },
-    {
-      "epoch": 19.398174706649282,
-      "grad_norm": 1.4362671375274658,
-      "learning_rate": 6.313266331658291e-05,
-      "loss": 5.544,
-      "step": 37196
-    },
-    {
-      "epoch": 19.3986962190352,
-      "grad_norm": 1.4620648622512817,
-      "learning_rate": 6.313165829145729e-05,
-      "loss": 5.1571,
-      "step": 37197
-    },
-    {
-      "epoch": 19.39921773142112,
-      "grad_norm": 1.4666802883148193,
-      "learning_rate": 6.313065326633167e-05,
-      "loss": 5.0899,
-      "step": 37198
-    },
-    {
-      "epoch": 19.39973924380704,
-      "grad_norm": 1.4427659511566162,
-      "learning_rate": 6.312964824120603e-05,
-      "loss": 5.1333,
-      "step": 37199
-    },
-    {
-      "epoch": 19.40026075619296,
-      "grad_norm": 1.452598214149475,
-      "learning_rate": 6.312864321608041e-05,
-      "loss": 4.9409,
-      "step": 37200
-    },
-    {
-      "epoch": 19.40078226857888,
-      "grad_norm": 1.655092716217041,
-      "learning_rate": 6.312763819095477e-05,
-      "loss": 5.014,
-      "step": 37201
-    },
-    {
-      "epoch": 19.4013037809648,
-      "grad_norm": 1.5137276649475098,
-      "learning_rate": 6.312663316582915e-05,
-      "loss": 5.1864,
-      "step": 37202
-    },
-    {
-      "epoch": 19.40182529335072,
-      "grad_norm": 1.560375690460205,
-      "learning_rate": 6.312562814070352e-05,
-      "loss": 5.4316,
-      "step": 37203
-    },
-    {
-      "epoch": 19.402346805736638,
-      "grad_norm": 1.5381048917770386,
-      "learning_rate": 6.31246231155779e-05,
-      "loss": 4.7365,
-      "step": 37204
-    },
-    {
-      "epoch": 19.402868318122554,
-      "grad_norm": 1.4114108085632324,
-      "learning_rate": 6.312361809045226e-05,
-      "loss": 5.9943,
-      "step": 37205
-    },
-    {
-      "epoch": 19.403389830508473,
-      "grad_norm": 1.756009578704834,
-      "learning_rate": 6.312261306532664e-05,
-      "loss": 5.3833,
-      "step": 37206
-    },
-    {
-      "epoch": 19.403911342894393,
-      "grad_norm": 1.4191328287124634,
-      "learning_rate": 6.312160804020101e-05,
-      "loss": 4.8109,
-      "step": 37207
-    },
-    {
-      "epoch": 19.404432855280312,
-      "grad_norm": 1.4172736406326294,
-      "learning_rate": 6.312060301507538e-05,
-      "loss": 5.4628,
-      "step": 37208
-    },
-    {
-      "epoch": 19.40495436766623,
-      "grad_norm": 1.6597981452941895,
-      "learning_rate": 6.311959798994976e-05,
-      "loss": 4.8706,
-      "step": 37209
-    },
-    {
-      "epoch": 19.40547588005215,
-      "grad_norm": 1.5348907709121704,
-      "learning_rate": 6.311859296482412e-05,
-      "loss": 5.3727,
-      "step": 37210
-    },
-    {
-      "epoch": 19.40599739243807,
-      "grad_norm": 1.5799050331115723,
-      "learning_rate": 6.31175879396985e-05,
-      "loss": 5.4354,
-      "step": 37211
-    },
-    {
-      "epoch": 19.40651890482399,
-      "grad_norm": 1.5251089334487915,
-      "learning_rate": 6.311658291457286e-05,
-      "loss": 5.3834,
-      "step": 37212
-    },
-    {
-      "epoch": 19.40704041720991,
-      "grad_norm": 1.4434407949447632,
-      "learning_rate": 6.311557788944724e-05,
-      "loss": 5.3558,
-      "step": 37213
-    },
-    {
-      "epoch": 19.40756192959583,
-      "grad_norm": 1.5814133882522583,
-      "learning_rate": 6.31145728643216e-05,
-      "loss": 5.2056,
-      "step": 37214
-    },
-    {
-      "epoch": 19.40808344198175,
-      "grad_norm": 1.609458088874817,
-      "learning_rate": 6.311356783919598e-05,
-      "loss": 5.6505,
-      "step": 37215
-    },
-    {
-      "epoch": 19.408604954367668,
-      "grad_norm": 1.6148344278335571,
-      "learning_rate": 6.311256281407035e-05,
-      "loss": 5.252,
-      "step": 37216
-    },
-    {
-      "epoch": 19.409126466753584,
-      "grad_norm": 1.559738278388977,
-      "learning_rate": 6.311155778894473e-05,
-      "loss": 5.2684,
-      "step": 37217
-    },
-    {
-      "epoch": 19.409647979139503,
-      "grad_norm": 1.470294713973999,
-      "learning_rate": 6.31105527638191e-05,
-      "loss": 5.3866,
-      "step": 37218
-    },
-    {
-      "epoch": 19.410169491525423,
-      "grad_norm": 1.3613698482513428,
-      "learning_rate": 6.310954773869348e-05,
-      "loss": 5.3841,
-      "step": 37219
-    },
-    {
-      "epoch": 19.410691003911342,
-      "grad_norm": 1.487229824066162,
-      "learning_rate": 6.310854271356785e-05,
-      "loss": 5.0261,
-      "step": 37220
-    },
-    {
-      "epoch": 19.41121251629726,
-      "grad_norm": 1.6053881645202637,
-      "learning_rate": 6.310753768844221e-05,
-      "loss": 5.3562,
-      "step": 37221
-    },
-    {
-      "epoch": 19.41173402868318,
-      "grad_norm": 1.5299992561340332,
-      "learning_rate": 6.310653266331659e-05,
-      "loss": 5.3561,
-      "step": 37222
-    },
-    {
-      "epoch": 19.4122555410691,
-      "grad_norm": 1.5384016036987305,
-      "learning_rate": 6.310552763819095e-05,
-      "loss": 4.6831,
-      "step": 37223
-    },
-    {
-      "epoch": 19.41277705345502,
-      "grad_norm": 1.5678869485855103,
-      "learning_rate": 6.310452261306533e-05,
-      "loss": 5.1611,
-      "step": 37224
-    },
-    {
-      "epoch": 19.41329856584094,
-      "grad_norm": 1.4890213012695312,
-      "learning_rate": 6.31035175879397e-05,
-      "loss": 5.0934,
-      "step": 37225
-    },
-    {
-      "epoch": 19.41382007822686,
-      "grad_norm": 1.4977561235427856,
-      "learning_rate": 6.310251256281407e-05,
-      "loss": 5.1157,
-      "step": 37226
-    },
-    {
-      "epoch": 19.41434159061278,
-      "grad_norm": 1.5159685611724854,
-      "learning_rate": 6.310150753768844e-05,
-      "loss": 5.5567,
-      "step": 37227
-    },
-    {
-      "epoch": 19.414863102998694,
-      "grad_norm": 1.5723031759262085,
-      "learning_rate": 6.310050251256281e-05,
-      "loss": 5.4754,
-      "step": 37228
-    },
-    {
-      "epoch": 19.415384615384614,
-      "grad_norm": 1.5406476259231567,
-      "learning_rate": 6.309949748743719e-05,
-      "loss": 5.2093,
-      "step": 37229
-    },
-    {
-      "epoch": 19.415906127770533,
-      "grad_norm": 1.4024009704589844,
-      "learning_rate": 6.309849246231157e-05,
-      "loss": 5.265,
-      "step": 37230
-    },
-    {
-      "epoch": 19.416427640156453,
-      "grad_norm": 1.4275270700454712,
-      "learning_rate": 6.309748743718593e-05,
-      "loss": 4.9464,
-      "step": 37231
-    },
-    {
-      "epoch": 19.416949152542372,
-      "grad_norm": 1.5062472820281982,
-      "learning_rate": 6.309648241206031e-05,
-      "loss": 5.0283,
-      "step": 37232
-    },
-    {
-      "epoch": 19.41747066492829,
-      "grad_norm": 1.4368422031402588,
-      "learning_rate": 6.309547738693468e-05,
-      "loss": 5.4871,
-      "step": 37233
-    },
-    {
-      "epoch": 19.41799217731421,
-      "grad_norm": 1.6018155813217163,
-      "learning_rate": 6.309447236180904e-05,
-      "loss": 5.2045,
-      "step": 37234
-    },
-    {
-      "epoch": 19.41851368970013,
-      "grad_norm": 1.5483193397521973,
-      "learning_rate": 6.309346733668342e-05,
-      "loss": 5.1693,
-      "step": 37235
-    },
-    {
-      "epoch": 19.41903520208605,
-      "grad_norm": 1.5866891145706177,
-      "learning_rate": 6.309246231155778e-05,
-      "loss": 5.3571,
-      "step": 37236
-    },
-    {
-      "epoch": 19.41955671447197,
-      "grad_norm": 1.5753493309020996,
-      "learning_rate": 6.309145728643216e-05,
-      "loss": 5.2074,
-      "step": 37237
-    },
-    {
-      "epoch": 19.42007822685789,
-      "grad_norm": 1.5064932107925415,
-      "learning_rate": 6.309045226130654e-05,
-      "loss": 5.521,
-      "step": 37238
-    },
-    {
-      "epoch": 19.42059973924381,
-      "grad_norm": 1.583794355392456,
-      "learning_rate": 6.308944723618092e-05,
-      "loss": 5.6401,
-      "step": 37239
-    },
-    {
-      "epoch": 19.421121251629724,
-      "grad_norm": 1.7183427810668945,
-      "learning_rate": 6.308844221105528e-05,
-      "loss": 5.0604,
-      "step": 37240
-    },
-    {
-      "epoch": 19.421642764015644,
-      "grad_norm": 1.4739699363708496,
-      "learning_rate": 6.308743718592966e-05,
-      "loss": 5.6901,
-      "step": 37241
-    },
-    {
-      "epoch": 19.422164276401563,
-      "grad_norm": 1.602225661277771,
-      "learning_rate": 6.308643216080402e-05,
-      "loss": 5.2367,
-      "step": 37242
-    },
-    {
-      "epoch": 19.422685788787483,
-      "grad_norm": 1.4460948705673218,
-      "learning_rate": 6.30854271356784e-05,
-      "loss": 5.6569,
-      "step": 37243
-    },
-    {
-      "epoch": 19.423207301173402,
-      "grad_norm": 1.569582223892212,
-      "learning_rate": 6.308442211055276e-05,
-      "loss": 5.5332,
-      "step": 37244
-    },
-    {
-      "epoch": 19.423728813559322,
-      "grad_norm": 1.575559139251709,
-      "learning_rate": 6.308341708542714e-05,
-      "loss": 5.565,
-      "step": 37245
-    },
-    {
-      "epoch": 19.42425032594524,
-      "grad_norm": 1.5398699045181274,
-      "learning_rate": 6.30824120603015e-05,
-      "loss": 5.384,
-      "step": 37246
-    },
-    {
-      "epoch": 19.42477183833116,
-      "grad_norm": 1.5141342878341675,
-      "learning_rate": 6.308140703517588e-05,
-      "loss": 5.3436,
-      "step": 37247
-    },
-    {
-      "epoch": 19.42529335071708,
-      "grad_norm": 1.5421720743179321,
-      "learning_rate": 6.308040201005025e-05,
-      "loss": 5.4526,
-      "step": 37248
-    },
-    {
-      "epoch": 19.425814863103,
-      "grad_norm": 1.475429654121399,
-      "learning_rate": 6.307939698492463e-05,
-      "loss": 5.3894,
-      "step": 37249
-    },
-    {
-      "epoch": 19.42633637548892,
-      "grad_norm": 1.575569987297058,
-      "learning_rate": 6.3078391959799e-05,
-      "loss": 5.478,
-      "step": 37250
-    },
-    {
-      "epoch": 19.42685788787484,
-      "grad_norm": 1.4934653043746948,
-      "learning_rate": 6.307738693467337e-05,
-      "loss": 5.7038,
-      "step": 37251
-    },
-    {
-      "epoch": 19.427379400260754,
-      "grad_norm": 1.4902020692825317,
-      "learning_rate": 6.307638190954775e-05,
-      "loss": 5.5386,
-      "step": 37252
-    },
-    {
-      "epoch": 19.427900912646674,
-      "grad_norm": 1.519992470741272,
-      "learning_rate": 6.307537688442211e-05,
-      "loss": 5.0999,
-      "step": 37253
-    },
-    {
-      "epoch": 19.428422425032593,
-      "grad_norm": 1.9740692377090454,
-      "learning_rate": 6.307437185929649e-05,
-      "loss": 5.151,
-      "step": 37254
-    },
-    {
-      "epoch": 19.428943937418513,
-      "grad_norm": 1.5554062128067017,
-      "learning_rate": 6.307336683417085e-05,
-      "loss": 5.1663,
-      "step": 37255
-    },
-    {
-      "epoch": 19.429465449804432,
-      "grad_norm": 1.4907764196395874,
-      "learning_rate": 6.307236180904523e-05,
-      "loss": 5.4956,
-      "step": 37256
-    },
-    {
-      "epoch": 19.429986962190352,
-      "grad_norm": 1.504427433013916,
-      "learning_rate": 6.30713567839196e-05,
-      "loss": 5.2189,
-      "step": 37257
-    },
-    {
-      "epoch": 19.43050847457627,
-      "grad_norm": 1.4602121114730835,
-      "learning_rate": 6.307035175879397e-05,
-      "loss": 5.4462,
-      "step": 37258
-    },
-    {
-      "epoch": 19.43102998696219,
-      "grad_norm": 1.4749996662139893,
-      "learning_rate": 6.306934673366835e-05,
-      "loss": 5.4368,
-      "step": 37259
-    },
-    {
-      "epoch": 19.43155149934811,
-      "grad_norm": 1.6066274642944336,
-      "learning_rate": 6.306834170854273e-05,
-      "loss": 4.8398,
-      "step": 37260
-    },
-    {
-      "epoch": 19.43207301173403,
-      "grad_norm": 1.488830804824829,
-      "learning_rate": 6.306733668341709e-05,
-      "loss": 5.5997,
-      "step": 37261
-    },
-    {
-      "epoch": 19.43259452411995,
-      "grad_norm": 1.4294817447662354,
-      "learning_rate": 6.306633165829146e-05,
-      "loss": 5.2905,
-      "step": 37262
-    },
-    {
-      "epoch": 19.43311603650587,
-      "grad_norm": 1.55461585521698,
-      "learning_rate": 6.306532663316583e-05,
-      "loss": 5.4724,
-      "step": 37263
-    },
-    {
-      "epoch": 19.433637548891785,
-      "grad_norm": 1.5447301864624023,
-      "learning_rate": 6.30643216080402e-05,
-      "loss": 5.2932,
-      "step": 37264
-    },
-    {
-      "epoch": 19.434159061277704,
-      "grad_norm": 1.4225887060165405,
-      "learning_rate": 6.306331658291458e-05,
-      "loss": 5.5343,
-      "step": 37265
-    },
-    {
-      "epoch": 19.434680573663623,
-      "grad_norm": 1.4907426834106445,
-      "learning_rate": 6.306231155778894e-05,
-      "loss": 4.9991,
-      "step": 37266
-    },
-    {
-      "epoch": 19.435202086049543,
-      "grad_norm": 1.5347858667373657,
-      "learning_rate": 6.306130653266332e-05,
-      "loss": 5.1725,
-      "step": 37267
-    },
-    {
-      "epoch": 19.435723598435462,
-      "grad_norm": 1.6455618143081665,
-      "learning_rate": 6.306030150753768e-05,
-      "loss": 5.0643,
-      "step": 37268
-    },
-    {
-      "epoch": 19.436245110821382,
-      "grad_norm": 1.580331802368164,
-      "learning_rate": 6.305929648241206e-05,
-      "loss": 4.9387,
-      "step": 37269
-    },
-    {
-      "epoch": 19.4367666232073,
-      "grad_norm": 1.5663504600524902,
-      "learning_rate": 6.305829145728644e-05,
-      "loss": 5.1564,
-      "step": 37270
-    },
-    {
-      "epoch": 19.43728813559322,
-      "grad_norm": 1.4984995126724243,
-      "learning_rate": 6.305728643216082e-05,
-      "loss": 5.0121,
-      "step": 37271
-    },
-    {
-      "epoch": 19.43780964797914,
-      "grad_norm": 1.522286057472229,
-      "learning_rate": 6.305628140703518e-05,
-      "loss": 5.4763,
-      "step": 37272
-    },
-    {
-      "epoch": 19.43833116036506,
-      "grad_norm": 1.5093772411346436,
-      "learning_rate": 6.305527638190956e-05,
-      "loss": 5.4766,
-      "step": 37273
-    },
-    {
-      "epoch": 19.43885267275098,
-      "grad_norm": 1.4995659589767456,
-      "learning_rate": 6.305427135678392e-05,
-      "loss": 5.2377,
-      "step": 37274
-    },
-    {
-      "epoch": 19.4393741851369,
-      "grad_norm": 1.5183773040771484,
-      "learning_rate": 6.305326633165829e-05,
-      "loss": 4.9542,
-      "step": 37275
-    },
-    {
-      "epoch": 19.439895697522815,
-      "grad_norm": 1.486081838607788,
-      "learning_rate": 6.305226130653266e-05,
-      "loss": 5.4226,
-      "step": 37276
-    },
-    {
-      "epoch": 19.440417209908734,
-      "grad_norm": 1.5112732648849487,
-      "learning_rate": 6.305125628140703e-05,
-      "loss": 5.2461,
-      "step": 37277
-    },
-    {
-      "epoch": 19.440938722294653,
-      "grad_norm": 1.5091496706008911,
-      "learning_rate": 6.30502512562814e-05,
-      "loss": 5.4919,
-      "step": 37278
-    },
-    {
-      "epoch": 19.441460234680573,
-      "grad_norm": 1.4098429679870605,
-      "learning_rate": 6.304924623115578e-05,
-      "loss": 5.8317,
-      "step": 37279
-    },
-    {
-      "epoch": 19.441981747066492,
-      "grad_norm": 1.521650791168213,
-      "learning_rate": 6.304824120603016e-05,
-      "loss": 5.3419,
-      "step": 37280
-    },
-    {
-      "epoch": 19.442503259452412,
-      "grad_norm": 1.4135823249816895,
-      "learning_rate": 6.304723618090453e-05,
-      "loss": 5.5825,
-      "step": 37281
-    },
-    {
-      "epoch": 19.44302477183833,
-      "grad_norm": 1.4460318088531494,
-      "learning_rate": 6.30462311557789e-05,
-      "loss": 5.4736,
-      "step": 37282
-    },
-    {
-      "epoch": 19.44354628422425,
-      "grad_norm": 1.6256747245788574,
-      "learning_rate": 6.304522613065327e-05,
-      "loss": 4.9557,
-      "step": 37283
-    },
-    {
-      "epoch": 19.44406779661017,
-      "grad_norm": 1.418568730354309,
-      "learning_rate": 6.304422110552765e-05,
-      "loss": 5.2874,
-      "step": 37284
-    },
-    {
-      "epoch": 19.44458930899609,
-      "grad_norm": 1.5083210468292236,
-      "learning_rate": 6.304321608040201e-05,
-      "loss": 5.4561,
-      "step": 37285
-    },
-    {
-      "epoch": 19.44511082138201,
-      "grad_norm": 1.5574920177459717,
-      "learning_rate": 6.304221105527639e-05,
-      "loss": 5.1684,
-      "step": 37286
-    },
-    {
-      "epoch": 19.44563233376793,
-      "grad_norm": 1.6772947311401367,
-      "learning_rate": 6.304120603015075e-05,
-      "loss": 5.0604,
-      "step": 37287
-    },
-    {
-      "epoch": 19.446153846153845,
-      "grad_norm": 1.4920275211334229,
-      "learning_rate": 6.304020100502512e-05,
-      "loss": 5.3984,
-      "step": 37288
-    },
-    {
-      "epoch": 19.446675358539764,
-      "grad_norm": 1.5521111488342285,
-      "learning_rate": 6.30391959798995e-05,
-      "loss": 5.4675,
-      "step": 37289
-    },
-    {
-      "epoch": 19.447196870925683,
-      "grad_norm": 1.5375505685806274,
-      "learning_rate": 6.303819095477387e-05,
-      "loss": 5.1897,
-      "step": 37290
-    },
-    {
-      "epoch": 19.447718383311603,
-      "grad_norm": 1.5636916160583496,
-      "learning_rate": 6.303718592964825e-05,
-      "loss": 5.2452,
-      "step": 37291
-    },
-    {
-      "epoch": 19.448239895697522,
-      "grad_norm": 1.5020076036453247,
-      "learning_rate": 6.303618090452261e-05,
-      "loss": 5.5928,
-      "step": 37292
-    },
-    {
-      "epoch": 19.448761408083442,
-      "grad_norm": 1.5464260578155518,
-      "learning_rate": 6.303517587939699e-05,
-      "loss": 5.4023,
-      "step": 37293
-    },
-    {
-      "epoch": 19.44928292046936,
-      "grad_norm": 1.420599341392517,
-      "learning_rate": 6.303417085427136e-05,
-      "loss": 5.6967,
-      "step": 37294
-    },
-    {
-      "epoch": 19.44980443285528,
-      "grad_norm": 1.493533968925476,
-      "learning_rate": 6.303316582914573e-05,
-      "loss": 4.9638,
-      "step": 37295
-    },
-    {
-      "epoch": 19.4503259452412,
-      "grad_norm": 1.5219279527664185,
-      "learning_rate": 6.30321608040201e-05,
-      "loss": 5.168,
-      "step": 37296
-    },
-    {
-      "epoch": 19.45084745762712,
-      "grad_norm": 1.4469178915023804,
-      "learning_rate": 6.303115577889448e-05,
-      "loss": 5.5201,
-      "step": 37297
-    },
-    {
-      "epoch": 19.45136897001304,
-      "grad_norm": 1.6388332843780518,
-      "learning_rate": 6.303015075376884e-05,
-      "loss": 5.0313,
-      "step": 37298
-    },
-    {
-      "epoch": 19.45189048239896,
-      "grad_norm": 1.402552604675293,
-      "learning_rate": 6.302914572864322e-05,
-      "loss": 5.049,
-      "step": 37299
-    },
-    {
-      "epoch": 19.452411994784875,
-      "grad_norm": 1.4198784828186035,
-      "learning_rate": 6.30281407035176e-05,
-      "loss": 5.7073,
-      "step": 37300
-    },
-    {
-      "epoch": 19.452933507170794,
-      "grad_norm": 1.4833933115005493,
-      "learning_rate": 6.302713567839196e-05,
-      "loss": 5.2907,
-      "step": 37301
-    },
-    {
-      "epoch": 19.453455019556714,
-      "grad_norm": 1.4306979179382324,
-      "learning_rate": 6.302613065326634e-05,
-      "loss": 5.4608,
-      "step": 37302
-    },
-    {
-      "epoch": 19.453976531942633,
-      "grad_norm": 1.4787994623184204,
-      "learning_rate": 6.30251256281407e-05,
-      "loss": 5.3125,
-      "step": 37303
-    },
-    {
-      "epoch": 19.454498044328552,
-      "grad_norm": 1.4288735389709473,
-      "learning_rate": 6.302412060301508e-05,
-      "loss": 5.2381,
-      "step": 37304
-    },
-    {
-      "epoch": 19.455019556714472,
-      "grad_norm": 1.341896414756775,
-      "learning_rate": 6.302311557788945e-05,
-      "loss": 5.5058,
-      "step": 37305
-    },
-    {
-      "epoch": 19.45554106910039,
-      "grad_norm": 1.5564379692077637,
-      "learning_rate": 6.302211055276382e-05,
-      "loss": 5.4784,
-      "step": 37306
-    },
-    {
-      "epoch": 19.45606258148631,
-      "grad_norm": 1.6468271017074585,
-      "learning_rate": 6.302110552763819e-05,
-      "loss": 4.9523,
-      "step": 37307
-    },
-    {
-      "epoch": 19.45658409387223,
-      "grad_norm": 1.4131602048873901,
-      "learning_rate": 6.302010050251257e-05,
-      "loss": 5.4534,
-      "step": 37308
-    },
-    {
-      "epoch": 19.45710560625815,
-      "grad_norm": 1.5736078023910522,
-      "learning_rate": 6.301909547738693e-05,
-      "loss": 5.0176,
-      "step": 37309
-    },
-    {
-      "epoch": 19.45762711864407,
-      "grad_norm": 1.4280222654342651,
-      "learning_rate": 6.301809045226131e-05,
-      "loss": 5.2648,
-      "step": 37310
-    },
-    {
-      "epoch": 19.45814863102999,
-      "grad_norm": 1.4703959226608276,
-      "learning_rate": 6.301708542713569e-05,
-      "loss": 5.5572,
-      "step": 37311
-    },
-    {
-      "epoch": 19.458670143415905,
-      "grad_norm": 1.501389980316162,
-      "learning_rate": 6.301608040201006e-05,
-      "loss": 4.7714,
-      "step": 37312
-    },
-    {
-      "epoch": 19.459191655801824,
-      "grad_norm": 1.5663864612579346,
-      "learning_rate": 6.301507537688443e-05,
-      "loss": 4.9158,
-      "step": 37313
-    },
-    {
-      "epoch": 19.459713168187744,
-      "grad_norm": 1.5946868658065796,
-      "learning_rate": 6.301407035175879e-05,
-      "loss": 4.8477,
-      "step": 37314
-    },
-    {
-      "epoch": 19.460234680573663,
-      "grad_norm": 1.4271913766860962,
-      "learning_rate": 6.301306532663317e-05,
-      "loss": 5.4187,
-      "step": 37315
-    },
-    {
-      "epoch": 19.460756192959582,
-      "grad_norm": 1.4007647037506104,
-      "learning_rate": 6.301206030150753e-05,
-      "loss": 5.1085,
-      "step": 37316
-    },
-    {
-      "epoch": 19.461277705345502,
-      "grad_norm": 1.4066468477249146,
-      "learning_rate": 6.301105527638191e-05,
-      "loss": 5.3755,
-      "step": 37317
-    },
-    {
-      "epoch": 19.46179921773142,
-      "grad_norm": 1.612689733505249,
-      "learning_rate": 6.301005025125628e-05,
-      "loss": 4.5325,
-      "step": 37318
-    },
-    {
-      "epoch": 19.46232073011734,
-      "grad_norm": 1.5783993005752563,
-      "learning_rate": 6.300904522613065e-05,
-      "loss": 5.2231,
-      "step": 37319
-    },
-    {
-      "epoch": 19.46284224250326,
-      "grad_norm": 1.4529337882995605,
-      "learning_rate": 6.300804020100503e-05,
-      "loss": 5.774,
-      "step": 37320
-    },
-    {
-      "epoch": 19.46336375488918,
-      "grad_norm": 1.5188593864440918,
-      "learning_rate": 6.300703517587941e-05,
-      "loss": 5.7188,
-      "step": 37321
-    },
-    {
-      "epoch": 19.4638852672751,
-      "grad_norm": 1.5579113960266113,
-      "learning_rate": 6.300603015075377e-05,
-      "loss": 5.2305,
-      "step": 37322
-    },
-    {
-      "epoch": 19.46440677966102,
-      "grad_norm": 1.4882932901382446,
-      "learning_rate": 6.300502512562815e-05,
-      "loss": 5.1786,
-      "step": 37323
-    },
-    {
-      "epoch": 19.464928292046935,
-      "grad_norm": 1.5697908401489258,
-      "learning_rate": 6.300402010050252e-05,
-      "loss": 4.9005,
-      "step": 37324
-    },
-    {
-      "epoch": 19.465449804432854,
-      "grad_norm": 1.3653483390808105,
-      "learning_rate": 6.30030150753769e-05,
-      "loss": 5.6436,
-      "step": 37325
-    },
-    {
-      "epoch": 19.465971316818774,
-      "grad_norm": 1.416479229927063,
-      "learning_rate": 6.300201005025126e-05,
-      "loss": 5.2184,
-      "step": 37326
-    },
-    {
-      "epoch": 19.466492829204693,
-      "grad_norm": 1.54704749584198,
-      "learning_rate": 6.300100502512562e-05,
-      "loss": 5.0929,
-      "step": 37327
-    },
-    {
-      "epoch": 19.467014341590613,
-      "grad_norm": 1.5011281967163086,
-      "learning_rate": 6.3e-05,
-      "loss": 5.5939,
-      "step": 37328
-    },
-    {
-      "epoch": 19.467535853976532,
-      "grad_norm": 1.5293469429016113,
-      "learning_rate": 6.299899497487436e-05,
-      "loss": 5.0441,
-      "step": 37329
-    },
-    {
-      "epoch": 19.46805736636245,
-      "grad_norm": 1.578230857849121,
-      "learning_rate": 6.299798994974874e-05,
-      "loss": 4.9815,
-      "step": 37330
-    },
-    {
-      "epoch": 19.46857887874837,
-      "grad_norm": 1.4376766681671143,
-      "learning_rate": 6.299698492462312e-05,
-      "loss": 5.383,
-      "step": 37331
-    },
-    {
-      "epoch": 19.46910039113429,
-      "grad_norm": 1.5833238363265991,
-      "learning_rate": 6.29959798994975e-05,
-      "loss": 5.3092,
-      "step": 37332
-    },
-    {
-      "epoch": 19.46962190352021,
-      "grad_norm": 1.5258651971817017,
-      "learning_rate": 6.299497487437186e-05,
-      "loss": 4.6548,
-      "step": 37333
-    },
-    {
-      "epoch": 19.47014341590613,
-      "grad_norm": 1.5256139039993286,
-      "learning_rate": 6.299396984924624e-05,
-      "loss": 5.6907,
-      "step": 37334
-    },
-    {
-      "epoch": 19.470664928292045,
-      "grad_norm": 1.6399085521697998,
-      "learning_rate": 6.29929648241206e-05,
-      "loss": 4.6621,
-      "step": 37335
-    },
-    {
-      "epoch": 19.471186440677965,
-      "grad_norm": 1.5760735273361206,
-      "learning_rate": 6.299195979899498e-05,
-      "loss": 5.3937,
-      "step": 37336
-    },
-    {
-      "epoch": 19.471707953063884,
-      "grad_norm": 1.657070517539978,
-      "learning_rate": 6.299095477386935e-05,
-      "loss": 5.4718,
-      "step": 37337
-    },
-    {
-      "epoch": 19.472229465449804,
-      "grad_norm": 1.5149132013320923,
-      "learning_rate": 6.298994974874372e-05,
-      "loss": 5.2985,
-      "step": 37338
-    },
-    {
-      "epoch": 19.472750977835723,
-      "grad_norm": 1.502623438835144,
-      "learning_rate": 6.298894472361809e-05,
-      "loss": 5.4277,
-      "step": 37339
-    },
-    {
-      "epoch": 19.473272490221643,
-      "grad_norm": 1.5746971368789673,
-      "learning_rate": 6.298793969849247e-05,
-      "loss": 5.1482,
-      "step": 37340
-    },
-    {
-      "epoch": 19.473794002607562,
-      "grad_norm": 1.5487323999404907,
-      "learning_rate": 6.298693467336684e-05,
-      "loss": 5.3819,
-      "step": 37341
-    },
-    {
-      "epoch": 19.47431551499348,
-      "grad_norm": 1.6664633750915527,
-      "learning_rate": 6.298592964824121e-05,
-      "loss": 5.1589,
-      "step": 37342
-    },
-    {
-      "epoch": 19.4748370273794,
-      "grad_norm": 1.679703712463379,
-      "learning_rate": 6.298492462311559e-05,
-      "loss": 5.2432,
-      "step": 37343
-    },
-    {
-      "epoch": 19.47535853976532,
-      "grad_norm": 1.556282877922058,
-      "learning_rate": 6.298391959798995e-05,
-      "loss": 5.226,
-      "step": 37344
-    },
-    {
-      "epoch": 19.47588005215124,
-      "grad_norm": 1.5468372106552124,
-      "learning_rate": 6.298291457286433e-05,
-      "loss": 5.1671,
-      "step": 37345
-    },
-    {
-      "epoch": 19.47640156453716,
-      "grad_norm": 1.5020066499710083,
-      "learning_rate": 6.298190954773869e-05,
-      "loss": 5.4424,
-      "step": 37346
-    },
-    {
-      "epoch": 19.476923076923075,
-      "grad_norm": 1.4907149076461792,
-      "learning_rate": 6.298090452261307e-05,
-      "loss": 5.4206,
-      "step": 37347
-    },
-    {
-      "epoch": 19.477444589308995,
-      "grad_norm": 1.5200107097625732,
-      "learning_rate": 6.297989949748743e-05,
-      "loss": 5.817,
-      "step": 37348
-    },
-    {
-      "epoch": 19.477966101694914,
-      "grad_norm": 1.4848884344100952,
-      "learning_rate": 6.297889447236181e-05,
-      "loss": 5.0979,
-      "step": 37349
-    },
-    {
-      "epoch": 19.478487614080834,
-      "grad_norm": 1.5240153074264526,
-      "learning_rate": 6.297788944723618e-05,
-      "loss": 5.428,
-      "step": 37350
-    },
-    {
-      "epoch": 19.479009126466753,
-      "grad_norm": 1.6533585786819458,
-      "learning_rate": 6.297688442211055e-05,
-      "loss": 5.5099,
-      "step": 37351
-    },
-    {
-      "epoch": 19.479530638852673,
-      "grad_norm": 1.7335293292999268,
-      "learning_rate": 6.297587939698493e-05,
-      "loss": 5.0175,
-      "step": 37352
-    },
-    {
-      "epoch": 19.480052151238592,
-      "grad_norm": 1.518573522567749,
-      "learning_rate": 6.297487437185931e-05,
-      "loss": 5.2858,
-      "step": 37353
-    },
-    {
-      "epoch": 19.48057366362451,
-      "grad_norm": 1.4349814653396606,
-      "learning_rate": 6.297386934673367e-05,
-      "loss": 5.8855,
-      "step": 37354
-    },
-    {
-      "epoch": 19.48109517601043,
-      "grad_norm": 1.6711530685424805,
-      "learning_rate": 6.297286432160804e-05,
-      "loss": 5.1467,
-      "step": 37355
-    },
-    {
-      "epoch": 19.48161668839635,
-      "grad_norm": 1.5619884729385376,
-      "learning_rate": 6.297185929648242e-05,
-      "loss": 5.4166,
-      "step": 37356
-    },
-    {
-      "epoch": 19.48213820078227,
-      "grad_norm": 1.4860761165618896,
-      "learning_rate": 6.297085427135678e-05,
-      "loss": 5.2916,
-      "step": 37357
-    },
-    {
-      "epoch": 19.48265971316819,
-      "grad_norm": 1.5150444507598877,
-      "learning_rate": 6.296984924623116e-05,
-      "loss": 5.1904,
-      "step": 37358
-    },
-    {
-      "epoch": 19.483181225554105,
-      "grad_norm": 1.5208262205123901,
-      "learning_rate": 6.296884422110552e-05,
-      "loss": 4.955,
-      "step": 37359
-    },
-    {
-      "epoch": 19.483702737940025,
-      "grad_norm": 1.535759687423706,
-      "learning_rate": 6.29678391959799e-05,
-      "loss": 5.3011,
-      "step": 37360
-    },
-    {
-      "epoch": 19.484224250325944,
-      "grad_norm": 1.5087066888809204,
-      "learning_rate": 6.296683417085428e-05,
-      "loss": 5.1726,
-      "step": 37361
-    },
-    {
-      "epoch": 19.484745762711864,
-      "grad_norm": 1.5211093425750732,
-      "learning_rate": 6.296582914572866e-05,
-      "loss": 5.1205,
-      "step": 37362
-    },
-    {
-      "epoch": 19.485267275097783,
-      "grad_norm": 1.5269780158996582,
-      "learning_rate": 6.296482412060302e-05,
-      "loss": 5.2067,
-      "step": 37363
-    },
-    {
-      "epoch": 19.485788787483703,
-      "grad_norm": 1.4537190198898315,
-      "learning_rate": 6.29638190954774e-05,
-      "loss": 5.5389,
-      "step": 37364
-    },
-    {
-      "epoch": 19.486310299869622,
-      "grad_norm": 1.4940224885940552,
-      "learning_rate": 6.296281407035176e-05,
-      "loss": 5.0526,
-      "step": 37365
-    },
-    {
-      "epoch": 19.48683181225554,
-      "grad_norm": 1.5141774415969849,
-      "learning_rate": 6.296180904522614e-05,
-      "loss": 5.6359,
-      "step": 37366
-    },
-    {
-      "epoch": 19.48735332464146,
-      "grad_norm": 1.5057686567306519,
-      "learning_rate": 6.29608040201005e-05,
-      "loss": 5.8364,
-      "step": 37367
-    },
-    {
-      "epoch": 19.48787483702738,
-      "grad_norm": 1.545931100845337,
-      "learning_rate": 6.295979899497487e-05,
-      "loss": 5.3301,
-      "step": 37368
-    },
-    {
-      "epoch": 19.4883963494133,
-      "grad_norm": 1.4198249578475952,
-      "learning_rate": 6.295879396984925e-05,
-      "loss": 5.3101,
-      "step": 37369
-    },
-    {
-      "epoch": 19.48891786179922,
-      "grad_norm": 1.5603911876678467,
-      "learning_rate": 6.295778894472361e-05,
-      "loss": 5.0234,
-      "step": 37370
-    },
-    {
-      "epoch": 19.489439374185135,
-      "grad_norm": 1.4139888286590576,
-      "learning_rate": 6.295678391959799e-05,
-      "loss": 4.992,
-      "step": 37371
-    },
-    {
-      "epoch": 19.489960886571055,
-      "grad_norm": 1.4619789123535156,
-      "learning_rate": 6.295577889447237e-05,
-      "loss": 5.3645,
-      "step": 37372
-    },
-    {
-      "epoch": 19.490482398956974,
-      "grad_norm": 1.4725536108016968,
-      "learning_rate": 6.295477386934674e-05,
-      "loss": 5.1141,
-      "step": 37373
-    },
-    {
-      "epoch": 19.491003911342894,
-      "grad_norm": 1.5416631698608398,
-      "learning_rate": 6.295376884422111e-05,
-      "loss": 5.143,
-      "step": 37374
-    },
-    {
-      "epoch": 19.491525423728813,
-      "grad_norm": 1.5572096109390259,
-      "learning_rate": 6.295276381909549e-05,
-      "loss": 5.4227,
-      "step": 37375
-    },
-    {
-      "epoch": 19.492046936114733,
-      "grad_norm": 1.5277761220932007,
-      "learning_rate": 6.295175879396985e-05,
-      "loss": 5.6088,
-      "step": 37376
-    },
-    {
-      "epoch": 19.492568448500652,
-      "grad_norm": 1.4732385873794556,
-      "learning_rate": 6.295075376884423e-05,
-      "loss": 5.6025,
-      "step": 37377
-    },
-    {
-      "epoch": 19.49308996088657,
-      "grad_norm": 1.5243701934814453,
-      "learning_rate": 6.294974874371859e-05,
-      "loss": 5.3534,
-      "step": 37378
-    },
-    {
-      "epoch": 19.49361147327249,
-      "grad_norm": 1.4268461465835571,
-      "learning_rate": 6.294874371859297e-05,
-      "loss": 5.4602,
-      "step": 37379
-    },
-    {
-      "epoch": 19.49413298565841,
-      "grad_norm": 1.6526100635528564,
-      "learning_rate": 6.294773869346734e-05,
-      "loss": 4.9007,
-      "step": 37380
-    },
-    {
-      "epoch": 19.49465449804433,
-      "grad_norm": 1.5461496114730835,
-      "learning_rate": 6.294673366834171e-05,
-      "loss": 5.207,
-      "step": 37381
-    },
-    {
-      "epoch": 19.49517601043025,
-      "grad_norm": 1.4336929321289062,
-      "learning_rate": 6.294572864321609e-05,
-      "loss": 5.4362,
-      "step": 37382
-    },
-    {
-      "epoch": 19.495697522816165,
-      "grad_norm": 1.496029257774353,
-      "learning_rate": 6.294472361809046e-05,
-      "loss": 5.4709,
-      "step": 37383
-    },
-    {
-      "epoch": 19.496219035202085,
-      "grad_norm": 1.4630082845687866,
-      "learning_rate": 6.294371859296483e-05,
-      "loss": 5.5147,
-      "step": 37384
-    },
-    {
-      "epoch": 19.496740547588004,
-      "grad_norm": 1.467675805091858,
-      "learning_rate": 6.29427135678392e-05,
-      "loss": 5.1248,
-      "step": 37385
-    },
-    {
-      "epoch": 19.497262059973924,
-      "grad_norm": 1.6054035425186157,
-      "learning_rate": 6.294170854271358e-05,
-      "loss": 5.4171,
-      "step": 37386
-    },
-    {
-      "epoch": 19.497783572359843,
-      "grad_norm": 1.576171875,
-      "learning_rate": 6.294070351758794e-05,
-      "loss": 5.35,
-      "step": 37387
-    },
-    {
-      "epoch": 19.498305084745763,
-      "grad_norm": 1.5490212440490723,
-      "learning_rate": 6.293969849246232e-05,
-      "loss": 5.2228,
-      "step": 37388
-    },
-    {
-      "epoch": 19.498826597131682,
-      "grad_norm": 1.468489646911621,
-      "learning_rate": 6.293869346733668e-05,
-      "loss": 5.5472,
-      "step": 37389
-    },
-    {
-      "epoch": 19.4993481095176,
-      "grad_norm": 1.8731147050857544,
-      "learning_rate": 6.293768844221106e-05,
-      "loss": 4.5855,
-      "step": 37390
-    },
-    {
-      "epoch": 19.49986962190352,
-      "grad_norm": 1.525352120399475,
-      "learning_rate": 6.293668341708542e-05,
-      "loss": 5.3013,
-      "step": 37391
-    },
-    {
-      "epoch": 19.50039113428944,
-      "grad_norm": 1.5804531574249268,
-      "learning_rate": 6.29356783919598e-05,
-      "loss": 5.2738,
-      "step": 37392
-    },
-    {
-      "epoch": 19.50091264667536,
-      "grad_norm": 1.5275425910949707,
-      "learning_rate": 6.293467336683418e-05,
-      "loss": 5.4325,
-      "step": 37393
-    },
-    {
-      "epoch": 19.50143415906128,
-      "grad_norm": 1.3847652673721313,
-      "learning_rate": 6.293366834170854e-05,
-      "loss": 4.6321,
-      "step": 37394
-    },
-    {
-      "epoch": 19.501955671447195,
-      "grad_norm": 1.6265003681182861,
-      "learning_rate": 6.293266331658292e-05,
-      "loss": 5.2354,
-      "step": 37395
-    },
-    {
-      "epoch": 19.502477183833115,
-      "grad_norm": 1.4969078302383423,
-      "learning_rate": 6.293165829145729e-05,
-      "loss": 5.5002,
-      "step": 37396
-    },
-    {
-      "epoch": 19.502998696219034,
-      "grad_norm": 1.4565858840942383,
-      "learning_rate": 6.293065326633166e-05,
-      "loss": 5.5101,
-      "step": 37397
-    },
-    {
-      "epoch": 19.503520208604954,
-      "grad_norm": 1.4968385696411133,
-      "learning_rate": 6.292964824120603e-05,
-      "loss": 5.2815,
-      "step": 37398
-    },
-    {
-      "epoch": 19.504041720990873,
-      "grad_norm": 1.4382553100585938,
-      "learning_rate": 6.29286432160804e-05,
-      "loss": 5.6591,
-      "step": 37399
-    },
-    {
-      "epoch": 19.504563233376793,
-      "grad_norm": 1.502134919166565,
-      "learning_rate": 6.292763819095477e-05,
-      "loss": 5.4695,
-      "step": 37400
-    },
-    {
-      "epoch": 19.505084745762712,
-      "grad_norm": 1.5889514684677124,
-      "learning_rate": 6.292663316582915e-05,
-      "loss": 5.2816,
-      "step": 37401
-    },
-    {
-      "epoch": 19.50560625814863,
-      "grad_norm": 1.7039299011230469,
-      "learning_rate": 6.292562814070351e-05,
-      "loss": 5.0244,
-      "step": 37402
-    },
-    {
-      "epoch": 19.50612777053455,
-      "grad_norm": 1.4913221597671509,
-      "learning_rate": 6.292462311557789e-05,
-      "loss": 5.0428,
-      "step": 37403
-    },
-    {
-      "epoch": 19.50664928292047,
-      "grad_norm": 1.5622193813323975,
-      "learning_rate": 6.292361809045227e-05,
-      "loss": 5.0815,
-      "step": 37404
-    },
-    {
-      "epoch": 19.50717079530639,
-      "grad_norm": 1.5759025812149048,
-      "learning_rate": 6.292261306532665e-05,
-      "loss": 4.9676,
-      "step": 37405
-    },
-    {
-      "epoch": 19.50769230769231,
-      "grad_norm": 1.4436661005020142,
-      "learning_rate": 6.292160804020101e-05,
-      "loss": 5.4342,
-      "step": 37406
-    },
-    {
-      "epoch": 19.508213820078225,
-      "grad_norm": 1.501172423362732,
-      "learning_rate": 6.292060301507537e-05,
-      "loss": 5.6958,
-      "step": 37407
-    },
-    {
-      "epoch": 19.508735332464145,
-      "grad_norm": 1.4558210372924805,
-      "learning_rate": 6.291959798994975e-05,
-      "loss": 4.8576,
-      "step": 37408
-    },
-    {
-      "epoch": 19.509256844850064,
-      "grad_norm": 1.4975327253341675,
-      "learning_rate": 6.291859296482412e-05,
-      "loss": 5.134,
-      "step": 37409
-    },
-    {
-      "epoch": 19.509778357235984,
-      "grad_norm": 1.4757581949234009,
-      "learning_rate": 6.29175879396985e-05,
-      "loss": 5.7075,
-      "step": 37410
-    },
-    {
-      "epoch": 19.510299869621903,
-      "grad_norm": 1.5648298263549805,
-      "learning_rate": 6.291658291457286e-05,
-      "loss": 5.2389,
-      "step": 37411
-    },
-    {
-      "epoch": 19.510821382007823,
-      "grad_norm": 1.6016122102737427,
-      "learning_rate": 6.291557788944724e-05,
-      "loss": 4.7801,
-      "step": 37412
-    },
-    {
-      "epoch": 19.511342894393742,
-      "grad_norm": 1.4645133018493652,
-      "learning_rate": 6.291457286432161e-05,
-      "loss": 5.4104,
-      "step": 37413
-    },
-    {
-      "epoch": 19.51186440677966,
-      "grad_norm": 1.4620314836502075,
-      "learning_rate": 6.291356783919599e-05,
-      "loss": 5.0597,
-      "step": 37414
-    },
-    {
-      "epoch": 19.51238591916558,
-      "grad_norm": 1.4295207262039185,
-      "learning_rate": 6.291256281407036e-05,
-      "loss": 5.3201,
-      "step": 37415
-    },
-    {
-      "epoch": 19.5129074315515,
-      "grad_norm": 1.4604666233062744,
-      "learning_rate": 6.291155778894473e-05,
-      "loss": 5.4127,
-      "step": 37416
-    },
-    {
-      "epoch": 19.51342894393742,
-      "grad_norm": 1.5334651470184326,
-      "learning_rate": 6.29105527638191e-05,
-      "loss": 5.3196,
-      "step": 37417
-    },
-    {
-      "epoch": 19.513950456323336,
-      "grad_norm": 1.5397158861160278,
-      "learning_rate": 6.290954773869348e-05,
-      "loss": 5.1471,
-      "step": 37418
-    },
-    {
-      "epoch": 19.514471968709255,
-      "grad_norm": 1.4055157899856567,
-      "learning_rate": 6.290854271356784e-05,
-      "loss": 5.2655,
-      "step": 37419
-    },
-    {
-      "epoch": 19.514993481095175,
-      "grad_norm": 1.4520074129104614,
-      "learning_rate": 6.290753768844222e-05,
-      "loss": 5.5835,
-      "step": 37420
-    },
-    {
-      "epoch": 19.515514993481094,
-      "grad_norm": 1.5064903497695923,
-      "learning_rate": 6.290653266331658e-05,
-      "loss": 5.1775,
-      "step": 37421
-    },
-    {
-      "epoch": 19.516036505867014,
-      "grad_norm": 1.4846075773239136,
-      "learning_rate": 6.290552763819095e-05,
-      "loss": 5.2161,
-      "step": 37422
-    },
-    {
-      "epoch": 19.516558018252933,
-      "grad_norm": 1.4353463649749756,
-      "learning_rate": 6.290452261306532e-05,
-      "loss": 5.1005,
-      "step": 37423
-    },
-    {
-      "epoch": 19.517079530638853,
-      "grad_norm": 1.4706088304519653,
-      "learning_rate": 6.29035175879397e-05,
-      "loss": 5.4241,
-      "step": 37424
-    },
-    {
-      "epoch": 19.517601043024772,
-      "grad_norm": 1.5441031455993652,
-      "learning_rate": 6.290251256281408e-05,
-      "loss": 5.0299,
-      "step": 37425
-    },
-    {
-      "epoch": 19.51812255541069,
-      "grad_norm": 1.5890207290649414,
-      "learning_rate": 6.290150753768844e-05,
-      "loss": 4.8899,
-      "step": 37426
-    },
-    {
-      "epoch": 19.51864406779661,
-      "grad_norm": 1.5170304775238037,
-      "learning_rate": 6.290050251256282e-05,
-      "loss": 5.4802,
-      "step": 37427
-    },
-    {
-      "epoch": 19.51916558018253,
-      "grad_norm": 1.4741225242614746,
-      "learning_rate": 6.289949748743719e-05,
-      "loss": 5.1109,
-      "step": 37428
-    },
-    {
-      "epoch": 19.51968709256845,
-      "grad_norm": 1.4384995698928833,
-      "learning_rate": 6.289849246231156e-05,
-      "loss": 5.4311,
-      "step": 37429
-    },
-    {
-      "epoch": 19.52020860495437,
-      "grad_norm": 1.5288879871368408,
-      "learning_rate": 6.289748743718593e-05,
-      "loss": 5.3896,
-      "step": 37430
-    },
-    {
-      "epoch": 19.520730117340285,
-      "grad_norm": 1.590402603149414,
-      "learning_rate": 6.28964824120603e-05,
-      "loss": 4.6944,
-      "step": 37431
-    },
-    {
-      "epoch": 19.521251629726205,
-      "grad_norm": 1.6978366374969482,
-      "learning_rate": 6.289547738693467e-05,
-      "loss": 4.8139,
-      "step": 37432
-    },
-    {
-      "epoch": 19.521773142112124,
-      "grad_norm": 1.6142961978912354,
-      "learning_rate": 6.289447236180905e-05,
-      "loss": 4.7552,
-      "step": 37433
-    },
-    {
-      "epoch": 19.522294654498044,
-      "grad_norm": 1.5438636541366577,
-      "learning_rate": 6.289346733668343e-05,
-      "loss": 5.3293,
-      "step": 37434
-    },
-    {
-      "epoch": 19.522816166883963,
-      "grad_norm": 1.4304120540618896,
-      "learning_rate": 6.289246231155779e-05,
-      "loss": 5.0658,
-      "step": 37435
-    },
-    {
-      "epoch": 19.523337679269883,
-      "grad_norm": 1.544596552848816,
-      "learning_rate": 6.289145728643217e-05,
-      "loss": 5.0356,
-      "step": 37436
-    },
-    {
-      "epoch": 19.523859191655802,
-      "grad_norm": 1.3785384893417358,
-      "learning_rate": 6.289045226130653e-05,
-      "loss": 5.6539,
-      "step": 37437
-    },
-    {
-      "epoch": 19.52438070404172,
-      "grad_norm": 1.5315215587615967,
-      "learning_rate": 6.288944723618091e-05,
-      "loss": 5.1332,
-      "step": 37438
-    },
-    {
-      "epoch": 19.52490221642764,
-      "grad_norm": 1.5645599365234375,
-      "learning_rate": 6.288844221105527e-05,
-      "loss": 4.8235,
-      "step": 37439
-    },
-    {
-      "epoch": 19.52542372881356,
-      "grad_norm": 1.51400625705719,
-      "learning_rate": 6.288743718592965e-05,
-      "loss": 5.4424,
-      "step": 37440
-    },
-    {
-      "epoch": 19.52594524119948,
-      "grad_norm": 1.4766571521759033,
-      "learning_rate": 6.288643216080402e-05,
-      "loss": 5.5457,
-      "step": 37441
-    },
-    {
-      "epoch": 19.526466753585396,
-      "grad_norm": 1.4231263399124146,
-      "learning_rate": 6.28854271356784e-05,
-      "loss": 5.2135,
-      "step": 37442
-    },
-    {
-      "epoch": 19.526988265971315,
-      "grad_norm": 1.6407511234283447,
-      "learning_rate": 6.288442211055276e-05,
-      "loss": 5.0354,
-      "step": 37443
-    },
-    {
-      "epoch": 19.527509778357235,
-      "grad_norm": 1.515535831451416,
-      "learning_rate": 6.288341708542714e-05,
-      "loss": 5.4358,
-      "step": 37444
-    },
-    {
-      "epoch": 19.528031290743154,
-      "grad_norm": 1.5242292881011963,
-      "learning_rate": 6.288241206030151e-05,
-      "loss": 5.5564,
-      "step": 37445
-    },
-    {
-      "epoch": 19.528552803129074,
-      "grad_norm": 1.5468130111694336,
-      "learning_rate": 6.288140703517589e-05,
-      "loss": 5.5873,
-      "step": 37446
-    },
-    {
-      "epoch": 19.529074315514993,
-      "grad_norm": 1.4609287977218628,
-      "learning_rate": 6.288040201005026e-05,
-      "loss": 5.7852,
-      "step": 37447
-    },
-    {
-      "epoch": 19.529595827900913,
-      "grad_norm": 1.4723258018493652,
-      "learning_rate": 6.287939698492462e-05,
-      "loss": 5.633,
-      "step": 37448
-    },
-    {
-      "epoch": 19.530117340286832,
-      "grad_norm": 1.5329313278198242,
-      "learning_rate": 6.2878391959799e-05,
-      "loss": 5.4327,
-      "step": 37449
-    },
-    {
-      "epoch": 19.53063885267275,
-      "grad_norm": 1.5618882179260254,
-      "learning_rate": 6.287738693467336e-05,
-      "loss": 5.363,
-      "step": 37450
-    },
-    {
-      "epoch": 19.53116036505867,
-      "grad_norm": 1.494633436203003,
-      "learning_rate": 6.287638190954774e-05,
-      "loss": 5.3407,
-      "step": 37451
-    },
-    {
-      "epoch": 19.53168187744459,
-      "grad_norm": 1.554058313369751,
-      "learning_rate": 6.28753768844221e-05,
-      "loss": 5.2192,
-      "step": 37452
-    },
-    {
-      "epoch": 19.53220338983051,
-      "grad_norm": 1.5984495878219604,
-      "learning_rate": 6.287437185929648e-05,
-      "loss": 5.1761,
-      "step": 37453
-    },
-    {
-      "epoch": 19.532724902216426,
-      "grad_norm": 1.6217752695083618,
-      "learning_rate": 6.287336683417086e-05,
-      "loss": 5.0292,
-      "step": 37454
-    },
-    {
-      "epoch": 19.533246414602345,
-      "grad_norm": 1.528728723526001,
-      "learning_rate": 6.287236180904524e-05,
-      "loss": 5.4789,
-      "step": 37455
-    },
-    {
-      "epoch": 19.533767926988265,
-      "grad_norm": 1.5629384517669678,
-      "learning_rate": 6.28713567839196e-05,
-      "loss": 5.2033,
-      "step": 37456
-    },
-    {
-      "epoch": 19.534289439374184,
-      "grad_norm": 1.4917786121368408,
-      "learning_rate": 6.287035175879398e-05,
-      "loss": 5.3138,
-      "step": 37457
-    },
-    {
-      "epoch": 19.534810951760104,
-      "grad_norm": 1.6691511869430542,
-      "learning_rate": 6.286934673366835e-05,
-      "loss": 5.307,
-      "step": 37458
-    },
-    {
-      "epoch": 19.535332464146023,
-      "grad_norm": 1.5341060161590576,
-      "learning_rate": 6.286834170854272e-05,
-      "loss": 5.4368,
-      "step": 37459
-    },
-    {
-      "epoch": 19.535853976531943,
-      "grad_norm": 1.5439406633377075,
-      "learning_rate": 6.286733668341709e-05,
-      "loss": 5.2648,
-      "step": 37460
-    },
-    {
-      "epoch": 19.536375488917862,
-      "grad_norm": 1.5557283163070679,
-      "learning_rate": 6.286633165829145e-05,
-      "loss": 5.4651,
-      "step": 37461
-    },
-    {
-      "epoch": 19.53689700130378,
-      "grad_norm": 1.4684910774230957,
-      "learning_rate": 6.286532663316583e-05,
-      "loss": 5.2367,
-      "step": 37462
-    },
-    {
-      "epoch": 19.5374185136897,
-      "grad_norm": 1.430099368095398,
-      "learning_rate": 6.28643216080402e-05,
-      "loss": 5.5626,
-      "step": 37463
-    },
-    {
-      "epoch": 19.53794002607562,
-      "grad_norm": 1.4476091861724854,
-      "learning_rate": 6.286331658291457e-05,
-      "loss": 5.1099,
-      "step": 37464
-    },
-    {
-      "epoch": 19.53846153846154,
-      "grad_norm": 1.5505305528640747,
-      "learning_rate": 6.286231155778895e-05,
-      "loss": 5.2486,
-      "step": 37465
-    },
-    {
-      "epoch": 19.538983050847456,
-      "grad_norm": 1.4357610940933228,
-      "learning_rate": 6.286130653266333e-05,
-      "loss": 4.7092,
-      "step": 37466
-    },
-    {
-      "epoch": 19.539504563233375,
-      "grad_norm": 1.442327618598938,
-      "learning_rate": 6.286030150753769e-05,
-      "loss": 5.4076,
-      "step": 37467
-    },
-    {
-      "epoch": 19.540026075619295,
-      "grad_norm": 1.4965447187423706,
-      "learning_rate": 6.285929648241207e-05,
-      "loss": 5.7182,
-      "step": 37468
-    },
-    {
-      "epoch": 19.540547588005214,
-      "grad_norm": 1.4274893999099731,
-      "learning_rate": 6.285829145728643e-05,
-      "loss": 5.6063,
-      "step": 37469
-    },
-    {
-      "epoch": 19.541069100391134,
-      "grad_norm": 1.4446918964385986,
-      "learning_rate": 6.285728643216081e-05,
-      "loss": 5.4581,
-      "step": 37470
-    },
-    {
-      "epoch": 19.541590612777053,
-      "grad_norm": 1.4935729503631592,
-      "learning_rate": 6.285628140703518e-05,
-      "loss": 5.1614,
-      "step": 37471
-    },
-    {
-      "epoch": 19.542112125162973,
-      "grad_norm": 1.6776635646820068,
-      "learning_rate": 6.285527638190955e-05,
-      "loss": 4.6668,
-      "step": 37472
-    },
-    {
-      "epoch": 19.542633637548892,
-      "grad_norm": 1.4967087507247925,
-      "learning_rate": 6.285427135678392e-05,
-      "loss": 5.6446,
-      "step": 37473
-    },
-    {
-      "epoch": 19.54315514993481,
-      "grad_norm": 1.6585583686828613,
-      "learning_rate": 6.28532663316583e-05,
-      "loss": 5.1118,
-      "step": 37474
-    },
-    {
-      "epoch": 19.54367666232073,
-      "grad_norm": 1.5823849439620972,
-      "learning_rate": 6.285226130653267e-05,
-      "loss": 5.2226,
-      "step": 37475
-    },
-    {
-      "epoch": 19.54419817470665,
-      "grad_norm": 1.548598289489746,
-      "learning_rate": 6.285125628140704e-05,
-      "loss": 5.4156,
-      "step": 37476
-    },
-    {
-      "epoch": 19.54471968709257,
-      "grad_norm": 1.4502068758010864,
-      "learning_rate": 6.285025125628142e-05,
-      "loss": 4.6628,
-      "step": 37477
-    },
-    {
-      "epoch": 19.545241199478486,
-      "grad_norm": 1.391873836517334,
-      "learning_rate": 6.284924623115578e-05,
-      "loss": 5.6292,
-      "step": 37478
-    },
-    {
-      "epoch": 19.545762711864406,
-      "grad_norm": 1.5155748128890991,
-      "learning_rate": 6.284824120603016e-05,
-      "loss": 5.4312,
-      "step": 37479
-    },
-    {
-      "epoch": 19.546284224250325,
-      "grad_norm": 1.5796457529067993,
-      "learning_rate": 6.284723618090452e-05,
-      "loss": 5.0863,
-      "step": 37480
-    },
-    {
-      "epoch": 19.546805736636244,
-      "grad_norm": 1.536846399307251,
-      "learning_rate": 6.28462311557789e-05,
-      "loss": 5.4377,
-      "step": 37481
-    },
-    {
-      "epoch": 19.547327249022164,
-      "grad_norm": 1.585491418838501,
-      "learning_rate": 6.284522613065326e-05,
-      "loss": 5.0156,
-      "step": 37482
-    },
-    {
-      "epoch": 19.547848761408083,
-      "grad_norm": 1.6833882331848145,
-      "learning_rate": 6.284422110552764e-05,
-      "loss": 4.8335,
-      "step": 37483
-    },
-    {
-      "epoch": 19.548370273794003,
-      "grad_norm": 1.5156689882278442,
-      "learning_rate": 6.2843216080402e-05,
-      "loss": 4.9986,
-      "step": 37484
-    },
-    {
-      "epoch": 19.548891786179922,
-      "grad_norm": 1.426401138305664,
-      "learning_rate": 6.284221105527638e-05,
-      "loss": 5.8179,
-      "step": 37485
-    },
-    {
-      "epoch": 19.54941329856584,
-      "grad_norm": 1.5560848712921143,
-      "learning_rate": 6.284120603015076e-05,
-      "loss": 5.3171,
-      "step": 37486
-    },
-    {
-      "epoch": 19.54993481095176,
-      "grad_norm": 1.5668470859527588,
-      "learning_rate": 6.284020100502513e-05,
-      "loss": 5.1523,
-      "step": 37487
-    },
-    {
-      "epoch": 19.55045632333768,
-      "grad_norm": 1.4982200860977173,
-      "learning_rate": 6.28391959798995e-05,
-      "loss": 5.3335,
-      "step": 37488
-    },
-    {
-      "epoch": 19.5509778357236,
-      "grad_norm": 1.5658247470855713,
-      "learning_rate": 6.283819095477387e-05,
-      "loss": 5.3295,
-      "step": 37489
-    },
-    {
-      "epoch": 19.551499348109516,
-      "grad_norm": 1.405168056488037,
-      "learning_rate": 6.283718592964825e-05,
-      "loss": 5.5674,
-      "step": 37490
-    },
-    {
-      "epoch": 19.552020860495436,
-      "grad_norm": 1.4721736907958984,
-      "learning_rate": 6.283618090452261e-05,
-      "loss": 5.6618,
-      "step": 37491
-    },
-    {
-      "epoch": 19.552542372881355,
-      "grad_norm": 1.4577295780181885,
-      "learning_rate": 6.283517587939699e-05,
-      "loss": 5.1682,
-      "step": 37492
-    },
-    {
-      "epoch": 19.553063885267274,
-      "grad_norm": 1.537173867225647,
-      "learning_rate": 6.283417085427135e-05,
-      "loss": 5.4976,
-      "step": 37493
-    },
-    {
-      "epoch": 19.553585397653194,
-      "grad_norm": 1.5518070459365845,
-      "learning_rate": 6.283316582914573e-05,
-      "loss": 5.5164,
-      "step": 37494
-    },
-    {
-      "epoch": 19.554106910039113,
-      "grad_norm": 1.455331802368164,
-      "learning_rate": 6.283216080402011e-05,
-      "loss": 5.6294,
-      "step": 37495
-    },
-    {
-      "epoch": 19.554628422425033,
-      "grad_norm": 1.4764269590377808,
-      "learning_rate": 6.283115577889449e-05,
-      "loss": 5.4742,
-      "step": 37496
-    },
-    {
-      "epoch": 19.555149934810952,
-      "grad_norm": 1.4882757663726807,
-      "learning_rate": 6.283015075376885e-05,
-      "loss": 5.5503,
-      "step": 37497
-    },
-    {
-      "epoch": 19.555671447196872,
-      "grad_norm": 1.4425289630889893,
-      "learning_rate": 6.282914572864323e-05,
-      "loss": 5.2949,
-      "step": 37498
-    },
-    {
-      "epoch": 19.55619295958279,
-      "grad_norm": 1.537475824356079,
-      "learning_rate": 6.282814070351759e-05,
-      "loss": 5.5836,
-      "step": 37499
-    },
-    {
-      "epoch": 19.55671447196871,
-      "grad_norm": 1.4583909511566162,
-      "learning_rate": 6.282713567839196e-05,
-      "loss": 5.2354,
-      "step": 37500
-    },
-    {
-      "epoch": 19.557235984354627,
-      "grad_norm": 1.4233314990997314,
-      "learning_rate": 6.282613065326633e-05,
-      "loss": 5.6107,
-      "step": 37501
-    },
-    {
-      "epoch": 19.557757496740546,
-      "grad_norm": 1.4939262866973877,
-      "learning_rate": 6.28251256281407e-05,
-      "loss": 5.591,
-      "step": 37502
-    },
-    {
-      "epoch": 19.558279009126466,
-      "grad_norm": 1.4904749393463135,
-      "learning_rate": 6.282412060301508e-05,
-      "loss": 5.1102,
-      "step": 37503
-    },
-    {
-      "epoch": 19.558800521512385,
-      "grad_norm": 1.5590099096298218,
-      "learning_rate": 6.282311557788944e-05,
-      "loss": 4.6874,
-      "step": 37504
-    },
-    {
-      "epoch": 19.559322033898304,
-      "grad_norm": 1.55458402633667,
-      "learning_rate": 6.282211055276382e-05,
-      "loss": 4.9892,
-      "step": 37505
-    },
-    {
-      "epoch": 19.559843546284224,
-      "grad_norm": 1.6992926597595215,
-      "learning_rate": 6.28211055276382e-05,
-      "loss": 4.987,
-      "step": 37506
-    },
-    {
-      "epoch": 19.560365058670143,
-      "grad_norm": 1.3965368270874023,
-      "learning_rate": 6.282010050251257e-05,
-      "loss": 5.3039,
-      "step": 37507
-    },
-    {
-      "epoch": 19.560886571056063,
-      "grad_norm": 1.4321143627166748,
-      "learning_rate": 6.281909547738694e-05,
-      "loss": 5.4974,
-      "step": 37508
-    },
-    {
-      "epoch": 19.561408083441982,
-      "grad_norm": 1.50222647190094,
-      "learning_rate": 6.281809045226132e-05,
-      "loss": 5.1136,
-      "step": 37509
-    },
-    {
-      "epoch": 19.561929595827902,
-      "grad_norm": 1.5559730529785156,
-      "learning_rate": 6.281708542713568e-05,
-      "loss": 5.5758,
-      "step": 37510
-    },
-    {
-      "epoch": 19.56245110821382,
-      "grad_norm": 1.4700188636779785,
-      "learning_rate": 6.281608040201006e-05,
-      "loss": 5.6807,
-      "step": 37511
-    },
-    {
-      "epoch": 19.56297262059974,
-      "grad_norm": 1.4860584735870361,
-      "learning_rate": 6.281507537688442e-05,
-      "loss": 5.3676,
-      "step": 37512
-    },
-    {
-      "epoch": 19.56349413298566,
-      "grad_norm": 1.4578760862350464,
-      "learning_rate": 6.28140703517588e-05,
-      "loss": 5.3806,
-      "step": 37513
-    },
-    {
-      "epoch": 19.564015645371576,
-      "grad_norm": 1.3947306871414185,
-      "learning_rate": 6.281306532663316e-05,
-      "loss": 5.4997,
-      "step": 37514
-    },
-    {
-      "epoch": 19.564537157757496,
-      "grad_norm": 1.4823371171951294,
-      "learning_rate": 6.281206030150754e-05,
-      "loss": 5.272,
-      "step": 37515
-    },
-    {
-      "epoch": 19.565058670143415,
-      "grad_norm": 1.4940375089645386,
-      "learning_rate": 6.281105527638192e-05,
-      "loss": 5.3478,
-      "step": 37516
-    },
-    {
-      "epoch": 19.565580182529335,
-      "grad_norm": 1.4621760845184326,
-      "learning_rate": 6.281005025125628e-05,
-      "loss": 5.3436,
-      "step": 37517
-    },
-    {
-      "epoch": 19.566101694915254,
-      "grad_norm": 1.6056277751922607,
-      "learning_rate": 6.280904522613066e-05,
-      "loss": 5.2662,
-      "step": 37518
-    },
-    {
-      "epoch": 19.566623207301173,
-      "grad_norm": 1.461944818496704,
-      "learning_rate": 6.280804020100503e-05,
-      "loss": 5.169,
-      "step": 37519
-    },
-    {
-      "epoch": 19.567144719687093,
-      "grad_norm": 1.5362483263015747,
-      "learning_rate": 6.28070351758794e-05,
-      "loss": 4.8722,
-      "step": 37520
-    },
-    {
-      "epoch": 19.567666232073012,
-      "grad_norm": 1.6124696731567383,
-      "learning_rate": 6.280603015075377e-05,
-      "loss": 4.8766,
-      "step": 37521
-    },
-    {
-      "epoch": 19.568187744458932,
-      "grad_norm": 1.556427240371704,
-      "learning_rate": 6.280502512562815e-05,
-      "loss": 4.9939,
-      "step": 37522
-    },
-    {
-      "epoch": 19.56870925684485,
-      "grad_norm": 1.5003563165664673,
-      "learning_rate": 6.280402010050251e-05,
-      "loss": 5.2814,
-      "step": 37523
-    },
-    {
-      "epoch": 19.56923076923077,
-      "grad_norm": 1.5423680543899536,
-      "learning_rate": 6.280301507537689e-05,
-      "loss": 5.3686,
-      "step": 37524
-    },
-    {
-      "epoch": 19.569752281616687,
-      "grad_norm": 1.6424102783203125,
-      "learning_rate": 6.280201005025125e-05,
-      "loss": 5.544,
-      "step": 37525
-    },
-    {
-      "epoch": 19.570273794002606,
-      "grad_norm": 1.7642232179641724,
-      "learning_rate": 6.280100502512563e-05,
-      "loss": 4.9591,
-      "step": 37526
-    },
-    {
-      "epoch": 19.570795306388526,
-      "grad_norm": 1.4571833610534668,
-      "learning_rate": 6.280000000000001e-05,
-      "loss": 5.2986,
-      "step": 37527
-    },
-    {
-      "epoch": 19.571316818774445,
-      "grad_norm": 1.4873907566070557,
-      "learning_rate": 6.279899497487437e-05,
-      "loss": 5.4254,
-      "step": 37528
-    },
-    {
-      "epoch": 19.571838331160365,
-      "grad_norm": 1.4587247371673584,
-      "learning_rate": 6.279798994974875e-05,
-      "loss": 5.5735,
-      "step": 37529
-    },
-    {
-      "epoch": 19.572359843546284,
-      "grad_norm": 1.6283143758773804,
-      "learning_rate": 6.279698492462311e-05,
-      "loss": 5.089,
-      "step": 37530
-    },
-    {
-      "epoch": 19.572881355932203,
-      "grad_norm": 1.646662712097168,
-      "learning_rate": 6.279597989949749e-05,
-      "loss": 4.7903,
-      "step": 37531
-    },
-    {
-      "epoch": 19.573402868318123,
-      "grad_norm": 1.4829233884811401,
-      "learning_rate": 6.279497487437186e-05,
-      "loss": 5.5188,
-      "step": 37532
-    },
-    {
-      "epoch": 19.573924380704042,
-      "grad_norm": 1.62949538230896,
-      "learning_rate": 6.279396984924623e-05,
-      "loss": 4.7079,
-      "step": 37533
-    },
-    {
-      "epoch": 19.574445893089962,
-      "grad_norm": 1.532666802406311,
-      "learning_rate": 6.27929648241206e-05,
-      "loss": 5.2303,
-      "step": 37534
-    },
-    {
-      "epoch": 19.57496740547588,
-      "grad_norm": 1.6331340074539185,
-      "learning_rate": 6.279195979899498e-05,
-      "loss": 5.6569,
-      "step": 37535
-    },
-    {
-      "epoch": 19.5754889178618,
-      "grad_norm": 1.6694132089614868,
-      "learning_rate": 6.279095477386935e-05,
-      "loss": 5.2909,
-      "step": 37536
-    },
-    {
-      "epoch": 19.576010430247717,
-      "grad_norm": 1.6019362211227417,
-      "learning_rate": 6.278994974874373e-05,
-      "loss": 5.4318,
-      "step": 37537
-    },
-    {
-      "epoch": 19.576531942633636,
-      "grad_norm": 1.5223824977874756,
-      "learning_rate": 6.27889447236181e-05,
-      "loss": 4.7419,
-      "step": 37538
-    },
-    {
-      "epoch": 19.577053455019556,
-      "grad_norm": 1.4767334461212158,
-      "learning_rate": 6.278793969849247e-05,
-      "loss": 5.3535,
-      "step": 37539
-    },
-    {
-      "epoch": 19.577574967405475,
-      "grad_norm": 1.5782471895217896,
-      "learning_rate": 6.278693467336684e-05,
-      "loss": 5.1381,
-      "step": 37540
-    },
-    {
-      "epoch": 19.578096479791395,
-      "grad_norm": 1.5867197513580322,
-      "learning_rate": 6.27859296482412e-05,
-      "loss": 5.2569,
-      "step": 37541
-    },
-    {
-      "epoch": 19.578617992177314,
-      "grad_norm": 1.5299867391586304,
-      "learning_rate": 6.278492462311558e-05,
-      "loss": 5.0196,
-      "step": 37542
-    },
-    {
-      "epoch": 19.579139504563233,
-      "grad_norm": 1.504335880279541,
-      "learning_rate": 6.278391959798995e-05,
-      "loss": 5.0948,
-      "step": 37543
-    },
-    {
-      "epoch": 19.579661016949153,
-      "grad_norm": 1.4975916147232056,
-      "learning_rate": 6.278291457286432e-05,
-      "loss": 5.6012,
-      "step": 37544
-    },
-    {
-      "epoch": 19.580182529335072,
-      "grad_norm": 1.6199015378952026,
-      "learning_rate": 6.278190954773869e-05,
-      "loss": 5.6067,
-      "step": 37545
-    },
-    {
-      "epoch": 19.580704041720992,
-      "grad_norm": 1.5995440483093262,
-      "learning_rate": 6.278090452261307e-05,
-      "loss": 5.0517,
-      "step": 37546
-    },
-    {
-      "epoch": 19.58122555410691,
-      "grad_norm": 1.5336226224899292,
-      "learning_rate": 6.277989949748744e-05,
-      "loss": 5.2133,
-      "step": 37547
-    },
-    {
-      "epoch": 19.58174706649283,
-      "grad_norm": 1.4558686017990112,
-      "learning_rate": 6.277889447236182e-05,
-      "loss": 5.4197,
-      "step": 37548
-    },
-    {
-      "epoch": 19.582268578878747,
-      "grad_norm": 1.5445451736450195,
-      "learning_rate": 6.277788944723619e-05,
-      "loss": 5.2175,
-      "step": 37549
-    },
-    {
-      "epoch": 19.582790091264666,
-      "grad_norm": 1.478776454925537,
-      "learning_rate": 6.277688442211056e-05,
-      "loss": 5.3742,
-      "step": 37550
-    },
-    {
-      "epoch": 19.583311603650586,
-      "grad_norm": 1.583756446838379,
-      "learning_rate": 6.277587939698493e-05,
-      "loss": 5.229,
-      "step": 37551
-    },
-    {
-      "epoch": 19.583833116036505,
-      "grad_norm": 1.6518765687942505,
-      "learning_rate": 6.27748743718593e-05,
-      "loss": 4.6818,
-      "step": 37552
-    },
-    {
-      "epoch": 19.584354628422425,
-      "grad_norm": 1.4976038932800293,
-      "learning_rate": 6.277386934673367e-05,
-      "loss": 5.4114,
-      "step": 37553
-    },
-    {
-      "epoch": 19.584876140808344,
-      "grad_norm": 1.3866770267486572,
-      "learning_rate": 6.277286432160803e-05,
-      "loss": 5.405,
-      "step": 37554
-    },
-    {
-      "epoch": 19.585397653194264,
-      "grad_norm": 1.5570836067199707,
-      "learning_rate": 6.277185929648241e-05,
-      "loss": 5.3777,
-      "step": 37555
-    },
-    {
-      "epoch": 19.585919165580183,
-      "grad_norm": 1.51217782497406,
-      "learning_rate": 6.277085427135679e-05,
-      "loss": 5.2728,
-      "step": 37556
-    },
-    {
-      "epoch": 19.586440677966102,
-      "grad_norm": 1.4295389652252197,
-      "learning_rate": 6.276984924623117e-05,
-      "loss": 5.6566,
-      "step": 37557
-    },
-    {
-      "epoch": 19.586962190352022,
-      "grad_norm": 1.582895040512085,
-      "learning_rate": 6.276884422110553e-05,
-      "loss": 5.2283,
-      "step": 37558
-    },
-    {
-      "epoch": 19.58748370273794,
-      "grad_norm": 1.5975871086120605,
-      "learning_rate": 6.276783919597991e-05,
-      "loss": 4.8399,
-      "step": 37559
-    },
-    {
-      "epoch": 19.58800521512386,
-      "grad_norm": 1.5466372966766357,
-      "learning_rate": 6.276683417085427e-05,
-      "loss": 5.2187,
-      "step": 37560
-    },
-    {
-      "epoch": 19.588526727509777,
-      "grad_norm": 1.6074717044830322,
-      "learning_rate": 6.276582914572865e-05,
-      "loss": 5.2391,
-      "step": 37561
-    },
-    {
-      "epoch": 19.589048239895696,
-      "grad_norm": 1.4472650289535522,
-      "learning_rate": 6.276482412060302e-05,
-      "loss": 5.3369,
-      "step": 37562
-    },
-    {
-      "epoch": 19.589569752281616,
-      "grad_norm": 1.5407295227050781,
-      "learning_rate": 6.27638190954774e-05,
-      "loss": 5.4798,
-      "step": 37563
-    },
-    {
-      "epoch": 19.590091264667535,
-      "grad_norm": 1.4994728565216064,
-      "learning_rate": 6.276281407035176e-05,
-      "loss": 5.3987,
-      "step": 37564
-    },
-    {
-      "epoch": 19.590612777053455,
-      "grad_norm": 1.391235589981079,
-      "learning_rate": 6.276180904522614e-05,
-      "loss": 4.9206,
-      "step": 37565
-    },
-    {
-      "epoch": 19.591134289439374,
-      "grad_norm": 1.470260500907898,
-      "learning_rate": 6.27608040201005e-05,
-      "loss": 5.3633,
-      "step": 37566
-    },
-    {
-      "epoch": 19.591655801825294,
-      "grad_norm": 1.4692065715789795,
-      "learning_rate": 6.275979899497488e-05,
-      "loss": 5.3129,
-      "step": 37567
-    },
-    {
-      "epoch": 19.592177314211213,
-      "grad_norm": 1.5469679832458496,
-      "learning_rate": 6.275879396984926e-05,
-      "loss": 4.7844,
-      "step": 37568
-    },
-    {
-      "epoch": 19.592698826597132,
-      "grad_norm": 1.6025574207305908,
-      "learning_rate": 6.275778894472362e-05,
-      "loss": 5.0649,
-      "step": 37569
-    },
-    {
-      "epoch": 19.593220338983052,
-      "grad_norm": 1.45127272605896,
-      "learning_rate": 6.2756783919598e-05,
-      "loss": 5.2027,
-      "step": 37570
-    },
-    {
-      "epoch": 19.59374185136897,
-      "grad_norm": 1.496595025062561,
-      "learning_rate": 6.275577889447236e-05,
-      "loss": 5.1162,
-      "step": 37571
-    },
-    {
-      "epoch": 19.59426336375489,
-      "grad_norm": 1.502832293510437,
-      "learning_rate": 6.275477386934674e-05,
-      "loss": 5.6622,
-      "step": 37572
-    },
-    {
-      "epoch": 19.594784876140807,
-      "grad_norm": 1.4173272848129272,
-      "learning_rate": 6.27537688442211e-05,
-      "loss": 5.5247,
-      "step": 37573
-    },
-    {
-      "epoch": 19.595306388526726,
-      "grad_norm": 1.5187772512435913,
-      "learning_rate": 6.275276381909548e-05,
-      "loss": 5.591,
-      "step": 37574
-    },
-    {
-      "epoch": 19.595827900912646,
-      "grad_norm": 1.5332928895950317,
-      "learning_rate": 6.275175879396985e-05,
-      "loss": 5.3176,
-      "step": 37575
-    },
-    {
-      "epoch": 19.596349413298565,
-      "grad_norm": 1.6054776906967163,
-      "learning_rate": 6.275075376884422e-05,
-      "loss": 5.5292,
-      "step": 37576
-    },
-    {
-      "epoch": 19.596870925684485,
-      "grad_norm": 1.5825316905975342,
-      "learning_rate": 6.274974874371859e-05,
-      "loss": 4.8125,
-      "step": 37577
-    },
-    {
-      "epoch": 19.597392438070404,
-      "grad_norm": 1.494697093963623,
-      "learning_rate": 6.274874371859297e-05,
-      "loss": 5.2449,
-      "step": 37578
-    },
-    {
-      "epoch": 19.597913950456324,
-      "grad_norm": 1.5960270166397095,
-      "learning_rate": 6.274773869346734e-05,
-      "loss": 5.0087,
-      "step": 37579
-    },
-    {
-      "epoch": 19.598435462842243,
-      "grad_norm": 1.541136622428894,
-      "learning_rate": 6.274673366834171e-05,
-      "loss": 5.2362,
-      "step": 37580
-    },
-    {
-      "epoch": 19.598956975228162,
-      "grad_norm": 1.3700608015060425,
-      "learning_rate": 6.274572864321609e-05,
-      "loss": 5.8244,
-      "step": 37581
-    },
-    {
-      "epoch": 19.599478487614082,
-      "grad_norm": 1.4307323694229126,
-      "learning_rate": 6.274472361809045e-05,
-      "loss": 5.0418,
-      "step": 37582
-    },
-    {
-      "epoch": 19.6,
-      "grad_norm": 1.4940412044525146,
-      "learning_rate": 6.274371859296483e-05,
-      "loss": 5.4635,
-      "step": 37583
-    },
-    {
-      "epoch": 19.60052151238592,
-      "grad_norm": 1.5038869380950928,
-      "learning_rate": 6.274271356783919e-05,
-      "loss": 5.2173,
-      "step": 37584
-    },
-    {
-      "epoch": 19.601043024771837,
-      "grad_norm": 1.6027987003326416,
-      "learning_rate": 6.274170854271357e-05,
-      "loss": 4.8352,
-      "step": 37585
-    },
-    {
-      "epoch": 19.601564537157756,
-      "grad_norm": 1.4765115976333618,
-      "learning_rate": 6.274070351758793e-05,
-      "loss": 5.3093,
-      "step": 37586
-    },
-    {
-      "epoch": 19.602086049543676,
-      "grad_norm": 1.5005013942718506,
-      "learning_rate": 6.273969849246231e-05,
-      "loss": 5.242,
-      "step": 37587
-    },
-    {
-      "epoch": 19.602607561929595,
-      "grad_norm": 1.4754964113235474,
-      "learning_rate": 6.273869346733669e-05,
-      "loss": 5.0594,
-      "step": 37588
-    },
-    {
-      "epoch": 19.603129074315515,
-      "grad_norm": 1.446650743484497,
-      "learning_rate": 6.273768844221107e-05,
-      "loss": 5.1257,
-      "step": 37589
-    },
-    {
-      "epoch": 19.603650586701434,
-      "grad_norm": 1.4760850667953491,
-      "learning_rate": 6.273668341708543e-05,
-      "loss": 5.1751,
-      "step": 37590
-    },
-    {
-      "epoch": 19.604172099087354,
-      "grad_norm": 1.550814151763916,
-      "learning_rate": 6.273567839195981e-05,
-      "loss": 5.6976,
-      "step": 37591
-    },
-    {
-      "epoch": 19.604693611473273,
-      "grad_norm": 1.4828503131866455,
-      "learning_rate": 6.273467336683417e-05,
-      "loss": 5.7513,
-      "step": 37592
-    },
-    {
-      "epoch": 19.605215123859193,
-      "grad_norm": 1.5224705934524536,
-      "learning_rate": 6.273366834170854e-05,
-      "loss": 5.3214,
-      "step": 37593
-    },
-    {
-      "epoch": 19.605736636245112,
-      "grad_norm": 1.4208425283432007,
-      "learning_rate": 6.273266331658292e-05,
-      "loss": 5.3941,
-      "step": 37594
-    },
-    {
-      "epoch": 19.60625814863103,
-      "grad_norm": 1.626675009727478,
-      "learning_rate": 6.273165829145728e-05,
-      "loss": 5.0443,
-      "step": 37595
-    },
-    {
-      "epoch": 19.60677966101695,
-      "grad_norm": 1.5927811861038208,
-      "learning_rate": 6.273065326633166e-05,
-      "loss": 4.8796,
-      "step": 37596
-    },
-    {
-      "epoch": 19.607301173402867,
-      "grad_norm": 1.4182718992233276,
-      "learning_rate": 6.272964824120602e-05,
-      "loss": 5.5953,
-      "step": 37597
-    },
-    {
-      "epoch": 19.607822685788786,
-      "grad_norm": 1.3938947916030884,
-      "learning_rate": 6.27286432160804e-05,
-      "loss": 5.7721,
-      "step": 37598
-    },
-    {
-      "epoch": 19.608344198174706,
-      "grad_norm": 1.5096721649169922,
-      "learning_rate": 6.272763819095478e-05,
-      "loss": 5.3634,
-      "step": 37599
-    },
-    {
-      "epoch": 19.608865710560625,
-      "grad_norm": 1.4636753797531128,
-      "learning_rate": 6.272663316582916e-05,
-      "loss": 5.6679,
-      "step": 37600
-    },
-    {
-      "epoch": 19.609387222946545,
-      "grad_norm": 1.5411666631698608,
-      "learning_rate": 6.272562814070352e-05,
-      "loss": 5.3685,
-      "step": 37601
-    },
-    {
-      "epoch": 19.609908735332464,
-      "grad_norm": 1.5273106098175049,
-      "learning_rate": 6.27246231155779e-05,
-      "loss": 5.4342,
-      "step": 37602
-    },
-    {
-      "epoch": 19.610430247718384,
-      "grad_norm": 1.51449716091156,
-      "learning_rate": 6.272361809045226e-05,
-      "loss": 5.3833,
-      "step": 37603
-    },
-    {
-      "epoch": 19.610951760104303,
-      "grad_norm": 1.3736467361450195,
-      "learning_rate": 6.272261306532664e-05,
-      "loss": 5.3503,
-      "step": 37604
-    },
-    {
-      "epoch": 19.611473272490223,
-      "grad_norm": 1.4307125806808472,
-      "learning_rate": 6.2721608040201e-05,
-      "loss": 5.684,
-      "step": 37605
-    },
-    {
-      "epoch": 19.611994784876142,
-      "grad_norm": 1.5913656949996948,
-      "learning_rate": 6.272060301507538e-05,
-      "loss": 5.2938,
-      "step": 37606
-    },
-    {
-      "epoch": 19.61251629726206,
-      "grad_norm": 1.4736003875732422,
-      "learning_rate": 6.271959798994975e-05,
-      "loss": 5.2257,
-      "step": 37607
-    },
-    {
-      "epoch": 19.613037809647977,
-      "grad_norm": 1.5495914220809937,
-      "learning_rate": 6.271859296482412e-05,
-      "loss": 5.067,
-      "step": 37608
-    },
-    {
-      "epoch": 19.613559322033897,
-      "grad_norm": 1.458991289138794,
-      "learning_rate": 6.27175879396985e-05,
-      "loss": 5.6035,
-      "step": 37609
-    },
-    {
-      "epoch": 19.614080834419816,
-      "grad_norm": 1.5523844957351685,
-      "learning_rate": 6.271658291457287e-05,
-      "loss": 5.2848,
-      "step": 37610
-    },
-    {
-      "epoch": 19.614602346805736,
-      "grad_norm": 1.5820930004119873,
-      "learning_rate": 6.271557788944724e-05,
-      "loss": 5.1735,
-      "step": 37611
-    },
-    {
-      "epoch": 19.615123859191655,
-      "grad_norm": 1.5134700536727905,
-      "learning_rate": 6.271457286432161e-05,
-      "loss": 4.6003,
-      "step": 37612
-    },
-    {
-      "epoch": 19.615645371577575,
-      "grad_norm": 1.4905681610107422,
-      "learning_rate": 6.271356783919599e-05,
-      "loss": 5.4521,
-      "step": 37613
-    },
-    {
-      "epoch": 19.616166883963494,
-      "grad_norm": 1.4900822639465332,
-      "learning_rate": 6.271256281407035e-05,
-      "loss": 5.4371,
-      "step": 37614
-    },
-    {
-      "epoch": 19.616688396349414,
-      "grad_norm": 1.4078669548034668,
-      "learning_rate": 6.271155778894473e-05,
-      "loss": 4.8921,
-      "step": 37615
-    },
-    {
-      "epoch": 19.617209908735333,
-      "grad_norm": 1.565551519393921,
-      "learning_rate": 6.271055276381909e-05,
-      "loss": 5.3998,
-      "step": 37616
-    },
-    {
-      "epoch": 19.617731421121253,
-      "grad_norm": 1.4072154760360718,
-      "learning_rate": 6.270954773869347e-05,
-      "loss": 5.1363,
-      "step": 37617
-    },
-    {
-      "epoch": 19.618252933507172,
-      "grad_norm": 1.4375736713409424,
-      "learning_rate": 6.270854271356784e-05,
-      "loss": 4.5125,
-      "step": 37618
-    },
-    {
-      "epoch": 19.61877444589309,
-      "grad_norm": 1.4776878356933594,
-      "learning_rate": 6.270753768844221e-05,
-      "loss": 5.2903,
-      "step": 37619
-    },
-    {
-      "epoch": 19.619295958279007,
-      "grad_norm": 1.435483694076538,
-      "learning_rate": 6.270653266331659e-05,
-      "loss": 5.2808,
-      "step": 37620
-    },
-    {
-      "epoch": 19.619817470664927,
-      "grad_norm": 1.6224850416183472,
-      "learning_rate": 6.270552763819096e-05,
-      "loss": 5.5047,
-      "step": 37621
-    },
-    {
-      "epoch": 19.620338983050846,
-      "grad_norm": 1.7488044500350952,
-      "learning_rate": 6.270452261306533e-05,
-      "loss": 4.3046,
-      "step": 37622
-    },
-    {
-      "epoch": 19.620860495436766,
-      "grad_norm": 1.4290629625320435,
-      "learning_rate": 6.27035175879397e-05,
-      "loss": 5.5056,
-      "step": 37623
-    },
-    {
-      "epoch": 19.621382007822685,
-      "grad_norm": 1.4691249132156372,
-      "learning_rate": 6.270251256281408e-05,
-      "loss": 5.3162,
-      "step": 37624
-    },
-    {
-      "epoch": 19.621903520208605,
-      "grad_norm": 1.4943100214004517,
-      "learning_rate": 6.270150753768844e-05,
-      "loss": 5.4307,
-      "step": 37625
-    },
-    {
-      "epoch": 19.622425032594524,
-      "grad_norm": 1.5160104036331177,
-      "learning_rate": 6.270050251256282e-05,
-      "loss": 5.2294,
-      "step": 37626
-    },
-    {
-      "epoch": 19.622946544980444,
-      "grad_norm": 1.5675759315490723,
-      "learning_rate": 6.269949748743718e-05,
-      "loss": 5.1675,
-      "step": 37627
-    },
-    {
-      "epoch": 19.623468057366363,
-      "grad_norm": 1.4964457750320435,
-      "learning_rate": 6.269849246231156e-05,
-      "loss": 5.3027,
-      "step": 37628
-    },
-    {
-      "epoch": 19.623989569752283,
-      "grad_norm": 1.741913080215454,
-      "learning_rate": 6.269748743718594e-05,
-      "loss": 4.5475,
-      "step": 37629
-    },
-    {
-      "epoch": 19.624511082138202,
-      "grad_norm": 1.4432686567306519,
-      "learning_rate": 6.269648241206032e-05,
-      "loss": 5.6251,
-      "step": 37630
-    },
-    {
-      "epoch": 19.62503259452412,
-      "grad_norm": 1.5002470016479492,
-      "learning_rate": 6.269547738693468e-05,
-      "loss": 5.543,
-      "step": 37631
-    },
-    {
-      "epoch": 19.625554106910037,
-      "grad_norm": 1.4617347717285156,
-      "learning_rate": 6.269447236180906e-05,
-      "loss": 4.9584,
-      "step": 37632
-    },
-    {
-      "epoch": 19.626075619295957,
-      "grad_norm": 1.554542899131775,
-      "learning_rate": 6.269346733668342e-05,
-      "loss": 5.8027,
-      "step": 37633
-    },
-    {
-      "epoch": 19.626597131681876,
-      "grad_norm": 1.450566291809082,
-      "learning_rate": 6.269246231155779e-05,
-      "loss": 5.5116,
-      "step": 37634
-    },
-    {
-      "epoch": 19.627118644067796,
-      "grad_norm": 1.5099658966064453,
-      "learning_rate": 6.269145728643216e-05,
-      "loss": 5.4224,
-      "step": 37635
-    },
-    {
-      "epoch": 19.627640156453715,
-      "grad_norm": 1.5328614711761475,
-      "learning_rate": 6.269045226130653e-05,
-      "loss": 5.506,
-      "step": 37636
-    },
-    {
-      "epoch": 19.628161668839635,
-      "grad_norm": 1.6219265460968018,
-      "learning_rate": 6.26894472361809e-05,
-      "loss": 4.9139,
-      "step": 37637
-    },
-    {
-      "epoch": 19.628683181225554,
-      "grad_norm": 1.6054140329360962,
-      "learning_rate": 6.268844221105527e-05,
-      "loss": 5.379,
-      "step": 37638
-    },
-    {
-      "epoch": 19.629204693611474,
-      "grad_norm": 1.5029445886611938,
-      "learning_rate": 6.268743718592965e-05,
-      "loss": 5.3809,
-      "step": 37639
-    },
-    {
-      "epoch": 19.629726205997393,
-      "grad_norm": 1.4691535234451294,
-      "learning_rate": 6.268643216080403e-05,
-      "loss": 5.4861,
-      "step": 37640
-    },
-    {
-      "epoch": 19.630247718383313,
-      "grad_norm": 1.445295810699463,
-      "learning_rate": 6.26854271356784e-05,
-      "loss": 5.5806,
-      "step": 37641
-    },
-    {
-      "epoch": 19.630769230769232,
-      "grad_norm": 1.4059147834777832,
-      "learning_rate": 6.268442211055277e-05,
-      "loss": 5.6724,
-      "step": 37642
-    },
-    {
-      "epoch": 19.63129074315515,
-      "grad_norm": 1.4308180809020996,
-      "learning_rate": 6.268341708542715e-05,
-      "loss": 5.4149,
-      "step": 37643
-    },
-    {
-      "epoch": 19.631812255541067,
-      "grad_norm": 1.5484178066253662,
-      "learning_rate": 6.268241206030151e-05,
-      "loss": 5.064,
-      "step": 37644
-    },
-    {
-      "epoch": 19.632333767926987,
-      "grad_norm": 1.4323620796203613,
-      "learning_rate": 6.268140703517589e-05,
-      "loss": 5.5499,
-      "step": 37645
-    },
-    {
-      "epoch": 19.632855280312906,
-      "grad_norm": 1.6358072757720947,
-      "learning_rate": 6.268040201005025e-05,
-      "loss": 5.0233,
-      "step": 37646
-    },
-    {
-      "epoch": 19.633376792698826,
-      "grad_norm": 1.4526996612548828,
-      "learning_rate": 6.267939698492462e-05,
-      "loss": 5.3598,
-      "step": 37647
-    },
-    {
-      "epoch": 19.633898305084745,
-      "grad_norm": 1.4445244073867798,
-      "learning_rate": 6.2678391959799e-05,
-      "loss": 5.3885,
-      "step": 37648
-    },
-    {
-      "epoch": 19.634419817470665,
-      "grad_norm": 1.5908727645874023,
-      "learning_rate": 6.267738693467337e-05,
-      "loss": 5.194,
-      "step": 37649
-    },
-    {
-      "epoch": 19.634941329856584,
-      "grad_norm": 1.6128159761428833,
-      "learning_rate": 6.267638190954775e-05,
-      "loss": 5.2101,
-      "step": 37650
-    },
-    {
-      "epoch": 19.635462842242504,
-      "grad_norm": 1.5362493991851807,
-      "learning_rate": 6.267537688442211e-05,
-      "loss": 5.4873,
-      "step": 37651
-    },
-    {
-      "epoch": 19.635984354628423,
-      "grad_norm": 1.5178464651107788,
-      "learning_rate": 6.267437185929649e-05,
-      "loss": 5.3452,
-      "step": 37652
-    },
-    {
-      "epoch": 19.636505867014343,
-      "grad_norm": 1.576580286026001,
-      "learning_rate": 6.267336683417086e-05,
-      "loss": 5.2976,
-      "step": 37653
-    },
-    {
-      "epoch": 19.637027379400262,
-      "grad_norm": 1.43178129196167,
-      "learning_rate": 6.267236180904523e-05,
-      "loss": 5.7311,
-      "step": 37654
-    },
-    {
-      "epoch": 19.63754889178618,
-      "grad_norm": 1.5006719827651978,
-      "learning_rate": 6.26713567839196e-05,
-      "loss": 5.6302,
-      "step": 37655
-    },
-    {
-      "epoch": 19.638070404172097,
-      "grad_norm": 1.5514339208602905,
-      "learning_rate": 6.267035175879398e-05,
-      "loss": 5.3075,
-      "step": 37656
-    },
-    {
-      "epoch": 19.638591916558017,
-      "grad_norm": 1.5391095876693726,
-      "learning_rate": 6.266934673366834e-05,
-      "loss": 4.9979,
-      "step": 37657
-    },
-    {
-      "epoch": 19.639113428943936,
-      "grad_norm": 1.56154465675354,
-      "learning_rate": 6.266834170854272e-05,
-      "loss": 4.8533,
-      "step": 37658
-    },
-    {
-      "epoch": 19.639634941329856,
-      "grad_norm": 1.4972847700119019,
-      "learning_rate": 6.266733668341708e-05,
-      "loss": 5.3583,
-      "step": 37659
-    },
-    {
-      "epoch": 19.640156453715775,
-      "grad_norm": 1.49470055103302,
-      "learning_rate": 6.266633165829146e-05,
-      "loss": 5.2732,
-      "step": 37660
-    },
-    {
-      "epoch": 19.640677966101695,
-      "grad_norm": 1.4360369443893433,
-      "learning_rate": 6.266532663316584e-05,
-      "loss": 5.5246,
-      "step": 37661
-    },
-    {
-      "epoch": 19.641199478487614,
-      "grad_norm": 1.4864399433135986,
-      "learning_rate": 6.26643216080402e-05,
-      "loss": 5.6585,
-      "step": 37662
-    },
-    {
-      "epoch": 19.641720990873534,
-      "grad_norm": 1.4260280132293701,
-      "learning_rate": 6.266331658291458e-05,
-      "loss": 5.4164,
-      "step": 37663
-    },
-    {
-      "epoch": 19.642242503259453,
-      "grad_norm": 1.7840913534164429,
-      "learning_rate": 6.266231155778894e-05,
-      "loss": 4.2047,
-      "step": 37664
-    },
-    {
-      "epoch": 19.642764015645373,
-      "grad_norm": 1.465456247329712,
-      "learning_rate": 6.266130653266332e-05,
-      "loss": 5.6235,
-      "step": 37665
-    },
-    {
-      "epoch": 19.643285528031292,
-      "grad_norm": 1.6156920194625854,
-      "learning_rate": 6.266030150753769e-05,
-      "loss": 5.1372,
-      "step": 37666
-    },
-    {
-      "epoch": 19.64380704041721,
-      "grad_norm": 1.5833464860916138,
-      "learning_rate": 6.265929648241206e-05,
-      "loss": 5.6132,
-      "step": 37667
-    },
-    {
-      "epoch": 19.644328552803128,
-      "grad_norm": 1.4955601692199707,
-      "learning_rate": 6.265829145728643e-05,
-      "loss": 5.5649,
-      "step": 37668
-    },
-    {
-      "epoch": 19.644850065189047,
-      "grad_norm": 1.4541289806365967,
-      "learning_rate": 6.26572864321608e-05,
-      "loss": 5.4881,
-      "step": 37669
-    },
-    {
-      "epoch": 19.645371577574966,
-      "grad_norm": 1.608767032623291,
-      "learning_rate": 6.265628140703518e-05,
-      "loss": 5.198,
-      "step": 37670
-    },
-    {
-      "epoch": 19.645893089960886,
-      "grad_norm": 1.4522544145584106,
-      "learning_rate": 6.265527638190956e-05,
-      "loss": 5.331,
-      "step": 37671
-    },
-    {
-      "epoch": 19.646414602346805,
-      "grad_norm": 1.4974626302719116,
-      "learning_rate": 6.265427135678393e-05,
-      "loss": 5.3959,
-      "step": 37672
-    },
-    {
-      "epoch": 19.646936114732725,
-      "grad_norm": 1.5265084505081177,
-      "learning_rate": 6.265326633165829e-05,
-      "loss": 5.1391,
-      "step": 37673
-    },
-    {
-      "epoch": 19.647457627118644,
-      "grad_norm": 1.4977551698684692,
-      "learning_rate": 6.265226130653267e-05,
-      "loss": 5.144,
-      "step": 37674
-    },
-    {
-      "epoch": 19.647979139504564,
-      "grad_norm": 1.4018847942352295,
-      "learning_rate": 6.265125628140703e-05,
-      "loss": 5.8881,
-      "step": 37675
-    },
-    {
-      "epoch": 19.648500651890483,
-      "grad_norm": 1.5085691213607788,
-      "learning_rate": 6.265025125628141e-05,
-      "loss": 5.6515,
-      "step": 37676
-    },
-    {
-      "epoch": 19.649022164276403,
-      "grad_norm": 1.5129985809326172,
-      "learning_rate": 6.264924623115577e-05,
-      "loss": 4.6483,
-      "step": 37677
-    },
-    {
-      "epoch": 19.649543676662322,
-      "grad_norm": 1.505188226699829,
-      "learning_rate": 6.264824120603015e-05,
-      "loss": 5.5479,
-      "step": 37678
-    },
-    {
-      "epoch": 19.65006518904824,
-      "grad_norm": 1.464761734008789,
-      "learning_rate": 6.264723618090452e-05,
-      "loss": 5.7337,
-      "step": 37679
-    },
-    {
-      "epoch": 19.650586701434158,
-      "grad_norm": 1.4081918001174927,
-      "learning_rate": 6.26462311557789e-05,
-      "loss": 5.3074,
-      "step": 37680
-    },
-    {
-      "epoch": 19.651108213820077,
-      "grad_norm": 1.5023560523986816,
-      "learning_rate": 6.264522613065327e-05,
-      "loss": 5.535,
-      "step": 37681
-    },
-    {
-      "epoch": 19.651629726205996,
-      "grad_norm": 1.3931162357330322,
-      "learning_rate": 6.264422110552765e-05,
-      "loss": 5.3269,
-      "step": 37682
-    },
-    {
-      "epoch": 19.652151238591916,
-      "grad_norm": 1.5462297201156616,
-      "learning_rate": 6.264321608040201e-05,
-      "loss": 4.4096,
-      "step": 37683
-    },
-    {
-      "epoch": 19.652672750977835,
-      "grad_norm": 1.5756338834762573,
-      "learning_rate": 6.264221105527639e-05,
-      "loss": 5.2541,
-      "step": 37684
-    },
-    {
-      "epoch": 19.653194263363755,
-      "grad_norm": 1.5468881130218506,
-      "learning_rate": 6.264120603015076e-05,
-      "loss": 5.296,
-      "step": 37685
-    },
-    {
-      "epoch": 19.653715775749674,
-      "grad_norm": 1.64215886592865,
-      "learning_rate": 6.264020100502512e-05,
-      "loss": 5.0474,
-      "step": 37686
-    },
-    {
-      "epoch": 19.654237288135594,
-      "grad_norm": 1.5330426692962646,
-      "learning_rate": 6.26391959798995e-05,
-      "loss": 4.6743,
-      "step": 37687
-    },
-    {
-      "epoch": 19.654758800521513,
-      "grad_norm": 1.566285490989685,
-      "learning_rate": 6.263819095477386e-05,
-      "loss": 5.219,
-      "step": 37688
-    },
-    {
-      "epoch": 19.655280312907433,
-      "grad_norm": 1.596718668937683,
-      "learning_rate": 6.263718592964824e-05,
-      "loss": 5.4018,
-      "step": 37689
-    },
-    {
-      "epoch": 19.655801825293352,
-      "grad_norm": 1.639778733253479,
-      "learning_rate": 6.263618090452262e-05,
-      "loss": 5.1801,
-      "step": 37690
-    },
-    {
-      "epoch": 19.656323337679268,
-      "grad_norm": 1.445666790008545,
-      "learning_rate": 6.2635175879397e-05,
-      "loss": 4.8818,
-      "step": 37691
-    },
-    {
-      "epoch": 19.656844850065188,
-      "grad_norm": 1.5076650381088257,
-      "learning_rate": 6.263417085427136e-05,
-      "loss": 5.271,
-      "step": 37692
-    },
-    {
-      "epoch": 19.657366362451107,
-      "grad_norm": 1.4407343864440918,
-      "learning_rate": 6.263316582914574e-05,
-      "loss": 5.4046,
-      "step": 37693
-    },
-    {
-      "epoch": 19.657887874837026,
-      "grad_norm": 1.5431572198867798,
-      "learning_rate": 6.26321608040201e-05,
-      "loss": 5.5447,
-      "step": 37694
-    },
-    {
-      "epoch": 19.658409387222946,
-      "grad_norm": 1.6018019914627075,
-      "learning_rate": 6.263115577889448e-05,
-      "loss": 4.9305,
-      "step": 37695
-    },
-    {
-      "epoch": 19.658930899608865,
-      "grad_norm": 1.6334898471832275,
-      "learning_rate": 6.263015075376885e-05,
-      "loss": 5.2877,
-      "step": 37696
-    },
-    {
-      "epoch": 19.659452411994785,
-      "grad_norm": 1.4812756776809692,
-      "learning_rate": 6.262914572864322e-05,
-      "loss": 5.5454,
-      "step": 37697
-    },
-    {
-      "epoch": 19.659973924380704,
-      "grad_norm": 1.405326247215271,
-      "learning_rate": 6.262814070351759e-05,
-      "loss": 5.3841,
-      "step": 37698
-    },
-    {
-      "epoch": 19.660495436766624,
-      "grad_norm": 1.4963828325271606,
-      "learning_rate": 6.262713567839197e-05,
-      "loss": 5.1248,
-      "step": 37699
-    },
-    {
-      "epoch": 19.661016949152543,
-      "grad_norm": 1.4346935749053955,
-      "learning_rate": 6.262613065326633e-05,
-      "loss": 4.7042,
-      "step": 37700
-    },
-    {
-      "epoch": 19.661538461538463,
-      "grad_norm": 1.4753185510635376,
-      "learning_rate": 6.262512562814071e-05,
-      "loss": 5.2905,
-      "step": 37701
-    },
-    {
-      "epoch": 19.662059973924382,
-      "grad_norm": 1.4415271282196045,
-      "learning_rate": 6.262412060301508e-05,
-      "loss": 5.4079,
-      "step": 37702
-    },
-    {
-      "epoch": 19.6625814863103,
-      "grad_norm": 1.4726511240005493,
-      "learning_rate": 6.262311557788945e-05,
-      "loss": 4.812,
-      "step": 37703
-    },
-    {
-      "epoch": 19.663102998696218,
-      "grad_norm": 1.477829933166504,
-      "learning_rate": 6.262211055276383e-05,
-      "loss": 5.453,
-      "step": 37704
-    },
-    {
-      "epoch": 19.663624511082137,
-      "grad_norm": 1.5461270809173584,
-      "learning_rate": 6.262110552763819e-05,
-      "loss": 4.8501,
-      "step": 37705
-    },
-    {
-      "epoch": 19.664146023468057,
-      "grad_norm": 1.5563873052597046,
-      "learning_rate": 6.262010050251257e-05,
-      "loss": 5.3168,
-      "step": 37706
-    },
-    {
-      "epoch": 19.664667535853976,
-      "grad_norm": 1.491679310798645,
-      "learning_rate": 6.261909547738693e-05,
-      "loss": 5.3774,
-      "step": 37707
-    },
-    {
-      "epoch": 19.665189048239895,
-      "grad_norm": 1.5337982177734375,
-      "learning_rate": 6.261809045226131e-05,
-      "loss": 5.4782,
-      "step": 37708
-    },
-    {
-      "epoch": 19.665710560625815,
-      "grad_norm": 1.9858604669570923,
-      "learning_rate": 6.261708542713568e-05,
-      "loss": 4.8895,
-      "step": 37709
-    },
-    {
-      "epoch": 19.666232073011734,
-      "grad_norm": 1.5323824882507324,
-      "learning_rate": 6.261608040201005e-05,
-      "loss": 5.0373,
-      "step": 37710
-    },
-    {
-      "epoch": 19.666753585397654,
-      "grad_norm": 1.5937687158584595,
-      "learning_rate": 6.261507537688443e-05,
-      "loss": 5.2399,
-      "step": 37711
-    },
-    {
-      "epoch": 19.667275097783573,
-      "grad_norm": 1.4722965955734253,
-      "learning_rate": 6.261407035175881e-05,
-      "loss": 5.4287,
-      "step": 37712
-    },
-    {
-      "epoch": 19.667796610169493,
-      "grad_norm": 1.477921962738037,
-      "learning_rate": 6.261306532663317e-05,
-      "loss": 5.3096,
-      "step": 37713
-    },
-    {
-      "epoch": 19.668318122555412,
-      "grad_norm": 1.3599882125854492,
-      "learning_rate": 6.261206030150754e-05,
-      "loss": 4.4335,
-      "step": 37714
-    },
-    {
-      "epoch": 19.668839634941328,
-      "grad_norm": 1.5106711387634277,
-      "learning_rate": 6.261105527638192e-05,
-      "loss": 4.9404,
-      "step": 37715
-    },
-    {
-      "epoch": 19.669361147327248,
-      "grad_norm": 1.4957044124603271,
-      "learning_rate": 6.261005025125628e-05,
-      "loss": 5.1446,
-      "step": 37716
-    },
-    {
-      "epoch": 19.669882659713167,
-      "grad_norm": 1.4679063558578491,
-      "learning_rate": 6.260904522613066e-05,
-      "loss": 5.3381,
-      "step": 37717
-    },
-    {
-      "epoch": 19.670404172099087,
-      "grad_norm": 1.5154857635498047,
-      "learning_rate": 6.260804020100502e-05,
-      "loss": 5.5359,
-      "step": 37718
-    },
-    {
-      "epoch": 19.670925684485006,
-      "grad_norm": 1.5127296447753906,
-      "learning_rate": 6.26070351758794e-05,
-      "loss": 5.4362,
-      "step": 37719
-    },
-    {
-      "epoch": 19.671447196870925,
-      "grad_norm": 1.628928542137146,
-      "learning_rate": 6.260603015075376e-05,
-      "loss": 5.218,
-      "step": 37720
-    },
-    {
-      "epoch": 19.671968709256845,
-      "grad_norm": 1.502188801765442,
-      "learning_rate": 6.260502512562814e-05,
-      "loss": 5.5269,
-      "step": 37721
-    },
-    {
-      "epoch": 19.672490221642764,
-      "grad_norm": 1.488709807395935,
-      "learning_rate": 6.260402010050252e-05,
-      "loss": 5.0923,
-      "step": 37722
-    },
-    {
-      "epoch": 19.673011734028684,
-      "grad_norm": 1.4561975002288818,
-      "learning_rate": 6.26030150753769e-05,
-      "loss": 5.7409,
-      "step": 37723
-    },
-    {
-      "epoch": 19.673533246414603,
-      "grad_norm": 1.5619852542877197,
-      "learning_rate": 6.260201005025126e-05,
-      "loss": 5.19,
-      "step": 37724
-    },
-    {
-      "epoch": 19.674054758800523,
-      "grad_norm": 1.457887887954712,
-      "learning_rate": 6.260100502512564e-05,
-      "loss": 5.4114,
-      "step": 37725
-    },
-    {
-      "epoch": 19.674576271186442,
-      "grad_norm": 1.5466421842575073,
-      "learning_rate": 6.26e-05,
-      "loss": 5.1683,
-      "step": 37726
-    },
-    {
-      "epoch": 19.675097783572358,
-      "grad_norm": 1.5486266613006592,
-      "learning_rate": 6.259899497487437e-05,
-      "loss": 5.1546,
-      "step": 37727
-    },
-    {
-      "epoch": 19.675619295958278,
-      "grad_norm": 1.5473213195800781,
-      "learning_rate": 6.259798994974875e-05,
-      "loss": 5.3938,
-      "step": 37728
-    },
-    {
-      "epoch": 19.676140808344197,
-      "grad_norm": 1.4843559265136719,
-      "learning_rate": 6.259698492462311e-05,
-      "loss": 5.6119,
-      "step": 37729
-    },
-    {
-      "epoch": 19.676662320730117,
-      "grad_norm": 1.477612018585205,
-      "learning_rate": 6.259597989949749e-05,
-      "loss": 5.3071,
-      "step": 37730
-    },
-    {
-      "epoch": 19.677183833116036,
-      "grad_norm": 1.435394525527954,
-      "learning_rate": 6.259497487437185e-05,
-      "loss": 5.4872,
-      "step": 37731
-    },
-    {
-      "epoch": 19.677705345501955,
-      "grad_norm": 1.3838953971862793,
-      "learning_rate": 6.259396984924623e-05,
-      "loss": 5.7507,
-      "step": 37732
-    },
-    {
-      "epoch": 19.678226857887875,
-      "grad_norm": 1.4059388637542725,
-      "learning_rate": 6.259296482412061e-05,
-      "loss": 5.7324,
-      "step": 37733
-    },
-    {
-      "epoch": 19.678748370273794,
-      "grad_norm": 1.5324269533157349,
-      "learning_rate": 6.259195979899499e-05,
-      "loss": 5.0334,
-      "step": 37734
-    },
-    {
-      "epoch": 19.679269882659714,
-      "grad_norm": 1.525691270828247,
-      "learning_rate": 6.259095477386935e-05,
-      "loss": 5.0279,
-      "step": 37735
-    },
-    {
-      "epoch": 19.679791395045633,
-      "grad_norm": 1.5493007898330688,
-      "learning_rate": 6.258994974874373e-05,
-      "loss": 5.2455,
-      "step": 37736
-    },
-    {
-      "epoch": 19.680312907431553,
-      "grad_norm": 1.5372636318206787,
-      "learning_rate": 6.258894472361809e-05,
-      "loss": 5.0935,
-      "step": 37737
-    },
-    {
-      "epoch": 19.680834419817472,
-      "grad_norm": 1.441118597984314,
-      "learning_rate": 6.258793969849247e-05,
-      "loss": 5.3325,
-      "step": 37738
-    },
-    {
-      "epoch": 19.681355932203388,
-      "grad_norm": 1.5197871923446655,
-      "learning_rate": 6.258693467336683e-05,
-      "loss": 5.4174,
-      "step": 37739
-    },
-    {
-      "epoch": 19.681877444589308,
-      "grad_norm": 1.4089950323104858,
-      "learning_rate": 6.25859296482412e-05,
-      "loss": 5.5369,
-      "step": 37740
-    },
-    {
-      "epoch": 19.682398956975227,
-      "grad_norm": 1.5526202917099,
-      "learning_rate": 6.258492462311558e-05,
-      "loss": 5.123,
-      "step": 37741
-    },
-    {
-      "epoch": 19.682920469361147,
-      "grad_norm": 1.5623321533203125,
-      "learning_rate": 6.258391959798995e-05,
-      "loss": 4.9804,
-      "step": 37742
-    },
-    {
-      "epoch": 19.683441981747066,
-      "grad_norm": 1.5126683712005615,
-      "learning_rate": 6.258291457286433e-05,
-      "loss": 5.4399,
-      "step": 37743
-    },
-    {
-      "epoch": 19.683963494132986,
-      "grad_norm": 1.4831838607788086,
-      "learning_rate": 6.25819095477387e-05,
-      "loss": 4.9608,
-      "step": 37744
-    },
-    {
-      "epoch": 19.684485006518905,
-      "grad_norm": 1.5250674486160278,
-      "learning_rate": 6.258090452261307e-05,
-      "loss": 5.172,
-      "step": 37745
-    },
-    {
-      "epoch": 19.685006518904824,
-      "grad_norm": 1.4581689834594727,
-      "learning_rate": 6.257989949748744e-05,
-      "loss": 5.4039,
-      "step": 37746
-    },
-    {
-      "epoch": 19.685528031290744,
-      "grad_norm": 1.5732287168502808,
-      "learning_rate": 6.257889447236182e-05,
-      "loss": 4.6906,
-      "step": 37747
-    },
-    {
-      "epoch": 19.686049543676663,
-      "grad_norm": 1.5067365169525146,
-      "learning_rate": 6.257788944723618e-05,
-      "loss": 5.4114,
-      "step": 37748
-    },
-    {
-      "epoch": 19.686571056062583,
-      "grad_norm": 1.5502276420593262,
-      "learning_rate": 6.257688442211056e-05,
-      "loss": 5.0355,
-      "step": 37749
-    },
-    {
-      "epoch": 19.687092568448502,
-      "grad_norm": 1.5083999633789062,
-      "learning_rate": 6.257587939698492e-05,
-      "loss": 5.2252,
-      "step": 37750
-    },
-    {
-      "epoch": 19.687614080834418,
-      "grad_norm": 1.569830298423767,
-      "learning_rate": 6.25748743718593e-05,
-      "loss": 5.0373,
-      "step": 37751
-    },
-    {
-      "epoch": 19.688135593220338,
-      "grad_norm": 1.610499620437622,
-      "learning_rate": 6.257386934673366e-05,
-      "loss": 5.2319,
-      "step": 37752
-    },
-    {
-      "epoch": 19.688657105606257,
-      "grad_norm": 1.4957754611968994,
-      "learning_rate": 6.257286432160804e-05,
-      "loss": 5.494,
-      "step": 37753
-    },
-    {
-      "epoch": 19.689178617992177,
-      "grad_norm": 1.6002564430236816,
-      "learning_rate": 6.257185929648242e-05,
-      "loss": 5.113,
-      "step": 37754
-    },
-    {
-      "epoch": 19.689700130378096,
-      "grad_norm": 1.5098798274993896,
-      "learning_rate": 6.257085427135678e-05,
-      "loss": 5.4296,
-      "step": 37755
-    },
-    {
-      "epoch": 19.690221642764016,
-      "grad_norm": 1.6264398097991943,
-      "learning_rate": 6.256984924623116e-05,
-      "loss": 4.9672,
-      "step": 37756
-    },
-    {
-      "epoch": 19.690743155149935,
-      "grad_norm": 1.3986549377441406,
-      "learning_rate": 6.256884422110553e-05,
-      "loss": 5.2593,
-      "step": 37757
-    },
-    {
-      "epoch": 19.691264667535854,
-      "grad_norm": 1.440482497215271,
-      "learning_rate": 6.25678391959799e-05,
-      "loss": 5.5761,
-      "step": 37758
-    },
-    {
-      "epoch": 19.691786179921774,
-      "grad_norm": 1.5494471788406372,
-      "learning_rate": 6.256683417085427e-05,
-      "loss": 5.503,
-      "step": 37759
-    },
-    {
-      "epoch": 19.692307692307693,
-      "grad_norm": 1.50187349319458,
-      "learning_rate": 6.256582914572865e-05,
-      "loss": 4.5023,
-      "step": 37760
-    },
-    {
-      "epoch": 19.692829204693613,
-      "grad_norm": 1.5744003057479858,
-      "learning_rate": 6.256482412060301e-05,
-      "loss": 4.9213,
-      "step": 37761
-    },
-    {
-      "epoch": 19.693350717079532,
-      "grad_norm": 1.455190896987915,
-      "learning_rate": 6.256381909547739e-05,
-      "loss": 5.4741,
-      "step": 37762
-    },
-    {
-      "epoch": 19.69387222946545,
-      "grad_norm": 1.5965027809143066,
-      "learning_rate": 6.256281407035177e-05,
-      "loss": 5.3953,
-      "step": 37763
-    },
-    {
-      "epoch": 19.694393741851368,
-      "grad_norm": 1.544043779373169,
-      "learning_rate": 6.256180904522614e-05,
-      "loss": 4.8939,
-      "step": 37764
-    },
-    {
-      "epoch": 19.694915254237287,
-      "grad_norm": 1.5488483905792236,
-      "learning_rate": 6.256080402010051e-05,
-      "loss": 5.3882,
-      "step": 37765
-    },
-    {
-      "epoch": 19.695436766623207,
-      "grad_norm": 1.462303876876831,
-      "learning_rate": 6.255979899497487e-05,
-      "loss": 5.0861,
-      "step": 37766
-    },
-    {
-      "epoch": 19.695958279009126,
-      "grad_norm": 1.6059184074401855,
-      "learning_rate": 6.255879396984925e-05,
-      "loss": 5.2646,
-      "step": 37767
-    },
-    {
-      "epoch": 19.696479791395046,
-      "grad_norm": 1.4851332902908325,
-      "learning_rate": 6.255778894472362e-05,
-      "loss": 5.5366,
-      "step": 37768
-    },
-    {
-      "epoch": 19.697001303780965,
-      "grad_norm": 1.4932317733764648,
-      "learning_rate": 6.255678391959799e-05,
-      "loss": 5.1809,
-      "step": 37769
-    },
-    {
-      "epoch": 19.697522816166884,
-      "grad_norm": 1.507222294807434,
-      "learning_rate": 6.255577889447236e-05,
-      "loss": 5.0557,
-      "step": 37770
-    },
-    {
-      "epoch": 19.698044328552804,
-      "grad_norm": 1.4107666015625,
-      "learning_rate": 6.255477386934673e-05,
-      "loss": 5.1493,
-      "step": 37771
-    },
-    {
-      "epoch": 19.698565840938723,
-      "grad_norm": 1.613178014755249,
-      "learning_rate": 6.25537688442211e-05,
-      "loss": 5.396,
-      "step": 37772
-    },
-    {
-      "epoch": 19.699087353324643,
-      "grad_norm": 1.4749068021774292,
-      "learning_rate": 6.255276381909548e-05,
-      "loss": 5.6813,
-      "step": 37773
-    },
-    {
-      "epoch": 19.69960886571056,
-      "grad_norm": 1.693174123764038,
-      "learning_rate": 6.255175879396985e-05,
-      "loss": 5.1729,
-      "step": 37774
-    },
-    {
-      "epoch": 19.70013037809648,
-      "grad_norm": 1.438726782798767,
-      "learning_rate": 6.255075376884423e-05,
-      "loss": 5.7595,
-      "step": 37775
-    },
-    {
-      "epoch": 19.700651890482398,
-      "grad_norm": 1.439436435699463,
-      "learning_rate": 6.25497487437186e-05,
-      "loss": 4.9995,
-      "step": 37776
-    },
-    {
-      "epoch": 19.701173402868317,
-      "grad_norm": 1.388101577758789,
-      "learning_rate": 6.254874371859297e-05,
-      "loss": 5.3256,
-      "step": 37777
-    },
-    {
-      "epoch": 19.701694915254237,
-      "grad_norm": 1.556438684463501,
-      "learning_rate": 6.254773869346734e-05,
-      "loss": 5.2898,
-      "step": 37778
-    },
-    {
-      "epoch": 19.702216427640156,
-      "grad_norm": 1.6184853315353394,
-      "learning_rate": 6.25467336683417e-05,
-      "loss": 5.334,
-      "step": 37779
-    },
-    {
-      "epoch": 19.702737940026076,
-      "grad_norm": 1.504356026649475,
-      "learning_rate": 6.254572864321608e-05,
-      "loss": 5.2109,
-      "step": 37780
-    },
-    {
-      "epoch": 19.703259452411995,
-      "grad_norm": 1.4765268564224243,
-      "learning_rate": 6.254472361809045e-05,
-      "loss": 5.548,
-      "step": 37781
-    },
-    {
-      "epoch": 19.703780964797915,
-      "grad_norm": 1.479588508605957,
-      "learning_rate": 6.254371859296482e-05,
-      "loss": 5.3464,
-      "step": 37782
-    },
-    {
-      "epoch": 19.704302477183834,
-      "grad_norm": 1.4930155277252197,
-      "learning_rate": 6.25427135678392e-05,
-      "loss": 5.114,
-      "step": 37783
-    },
-    {
-      "epoch": 19.704823989569753,
-      "grad_norm": 1.5175000429153442,
-      "learning_rate": 6.254170854271358e-05,
-      "loss": 5.0785,
-      "step": 37784
-    },
-    {
-      "epoch": 19.705345501955673,
-      "grad_norm": 1.5469098091125488,
-      "learning_rate": 6.254070351758794e-05,
-      "loss": 5.4572,
-      "step": 37785
-    },
-    {
-      "epoch": 19.705867014341592,
-      "grad_norm": 1.6336795091629028,
-      "learning_rate": 6.253969849246232e-05,
-      "loss": 5.1819,
-      "step": 37786
-    },
-    {
-      "epoch": 19.70638852672751,
-      "grad_norm": 1.6442217826843262,
-      "learning_rate": 6.253869346733669e-05,
-      "loss": 4.8745,
-      "step": 37787
-    },
-    {
-      "epoch": 19.706910039113428,
-      "grad_norm": 1.5792070627212524,
-      "learning_rate": 6.253768844221106e-05,
-      "loss": 5.2774,
-      "step": 37788
-    },
-    {
-      "epoch": 19.707431551499347,
-      "grad_norm": 1.4850196838378906,
-      "learning_rate": 6.253668341708543e-05,
-      "loss": 5.1423,
-      "step": 37789
-    },
-    {
-      "epoch": 19.707953063885267,
-      "grad_norm": 1.451205849647522,
-      "learning_rate": 6.25356783919598e-05,
-      "loss": 5.513,
-      "step": 37790
-    },
-    {
-      "epoch": 19.708474576271186,
-      "grad_norm": 1.4933828115463257,
-      "learning_rate": 6.253467336683417e-05,
-      "loss": 5.6319,
-      "step": 37791
-    },
-    {
-      "epoch": 19.708996088657106,
-      "grad_norm": 1.5349767208099365,
-      "learning_rate": 6.253366834170855e-05,
-      "loss": 5.112,
-      "step": 37792
-    },
-    {
-      "epoch": 19.709517601043025,
-      "grad_norm": 1.5057306289672852,
-      "learning_rate": 6.253266331658291e-05,
-      "loss": 4.8375,
-      "step": 37793
-    },
-    {
-      "epoch": 19.710039113428945,
-      "grad_norm": 1.5328397750854492,
-      "learning_rate": 6.253165829145729e-05,
-      "loss": 5.3934,
-      "step": 37794
-    },
-    {
-      "epoch": 19.710560625814864,
-      "grad_norm": 1.5326961278915405,
-      "learning_rate": 6.253065326633167e-05,
-      "loss": 5.3965,
-      "step": 37795
-    },
-    {
-      "epoch": 19.711082138200783,
-      "grad_norm": 1.4378913640975952,
-      "learning_rate": 6.252964824120603e-05,
-      "loss": 5.7705,
-      "step": 37796
-    },
-    {
-      "epoch": 19.711603650586703,
-      "grad_norm": 1.5641452074050903,
-      "learning_rate": 6.252864321608041e-05,
-      "loss": 5.2552,
-      "step": 37797
-    },
-    {
-      "epoch": 19.71212516297262,
-      "grad_norm": 1.4623820781707764,
-      "learning_rate": 6.252763819095477e-05,
-      "loss": 5.1806,
-      "step": 37798
-    },
-    {
-      "epoch": 19.71264667535854,
-      "grad_norm": 1.4451181888580322,
-      "learning_rate": 6.252663316582915e-05,
-      "loss": 5.5482,
-      "step": 37799
-    },
-    {
-      "epoch": 19.713168187744458,
-      "grad_norm": 1.4379671812057495,
-      "learning_rate": 6.252562814070352e-05,
-      "loss": 5.0405,
-      "step": 37800
-    },
-    {
-      "epoch": 19.713689700130377,
-      "grad_norm": 1.6108121871948242,
-      "learning_rate": 6.25246231155779e-05,
-      "loss": 5.1643,
-      "step": 37801
-    },
-    {
-      "epoch": 19.714211212516297,
-      "grad_norm": 1.434021234512329,
-      "learning_rate": 6.252361809045226e-05,
-      "loss": 5.7419,
-      "step": 37802
-    },
-    {
-      "epoch": 19.714732724902216,
-      "grad_norm": 1.6004538536071777,
-      "learning_rate": 6.252261306532664e-05,
-      "loss": 5.4776,
-      "step": 37803
-    },
-    {
-      "epoch": 19.715254237288136,
-      "grad_norm": 1.6579724550247192,
-      "learning_rate": 6.252160804020101e-05,
-      "loss": 5.0847,
-      "step": 37804
-    },
-    {
-      "epoch": 19.715775749674055,
-      "grad_norm": 1.5093960762023926,
-      "learning_rate": 6.252060301507539e-05,
-      "loss": 5.2128,
-      "step": 37805
-    },
-    {
-      "epoch": 19.716297262059975,
-      "grad_norm": 1.4800500869750977,
-      "learning_rate": 6.251959798994976e-05,
-      "loss": 5.2295,
-      "step": 37806
-    },
-    {
-      "epoch": 19.716818774445894,
-      "grad_norm": 1.459172010421753,
-      "learning_rate": 6.251859296482412e-05,
-      "loss": 5.5372,
-      "step": 37807
-    },
-    {
-      "epoch": 19.717340286831814,
-      "grad_norm": 1.4869908094406128,
-      "learning_rate": 6.25175879396985e-05,
-      "loss": 5.4684,
-      "step": 37808
-    },
-    {
-      "epoch": 19.717861799217733,
-      "grad_norm": 1.4305486679077148,
-      "learning_rate": 6.251658291457286e-05,
-      "loss": 5.5493,
-      "step": 37809
-    },
-    {
-      "epoch": 19.71838331160365,
-      "grad_norm": 1.462227463722229,
-      "learning_rate": 6.251557788944724e-05,
-      "loss": 4.9258,
-      "step": 37810
-    },
-    {
-      "epoch": 19.71890482398957,
-      "grad_norm": 1.5091060400009155,
-      "learning_rate": 6.25145728643216e-05,
-      "loss": 5.4094,
-      "step": 37811
-    },
-    {
-      "epoch": 19.719426336375488,
-      "grad_norm": 1.5153913497924805,
-      "learning_rate": 6.251356783919598e-05,
-      "loss": 5.24,
-      "step": 37812
-    },
-    {
-      "epoch": 19.719947848761407,
-      "grad_norm": 1.5013216733932495,
-      "learning_rate": 6.251256281407035e-05,
-      "loss": 4.5806,
-      "step": 37813
-    },
-    {
-      "epoch": 19.720469361147327,
-      "grad_norm": 1.5234694480895996,
-      "learning_rate": 6.251155778894472e-05,
-      "loss": 5.2576,
-      "step": 37814
-    },
-    {
-      "epoch": 19.720990873533246,
-      "grad_norm": 1.411712408065796,
-      "learning_rate": 6.25105527638191e-05,
-      "loss": 5.6703,
-      "step": 37815
-    },
-    {
-      "epoch": 19.721512385919166,
-      "grad_norm": 1.4780032634735107,
-      "learning_rate": 6.250954773869348e-05,
-      "loss": 5.5083,
-      "step": 37816
-    },
-    {
-      "epoch": 19.722033898305085,
-      "grad_norm": 1.5177258253097534,
-      "learning_rate": 6.250854271356784e-05,
-      "loss": 5.7371,
-      "step": 37817
-    },
-    {
-      "epoch": 19.722555410691005,
-      "grad_norm": 1.442824363708496,
-      "learning_rate": 6.250753768844222e-05,
-      "loss": 4.8508,
-      "step": 37818
-    },
-    {
-      "epoch": 19.723076923076924,
-      "grad_norm": 1.5782191753387451,
-      "learning_rate": 6.250653266331659e-05,
-      "loss": 5.2344,
-      "step": 37819
-    },
-    {
-      "epoch": 19.723598435462844,
-      "grad_norm": 1.513321042060852,
-      "learning_rate": 6.250552763819095e-05,
-      "loss": 5.2039,
-      "step": 37820
-    },
-    {
-      "epoch": 19.724119947848763,
-      "grad_norm": 1.5104763507843018,
-      "learning_rate": 6.250452261306533e-05,
-      "loss": 5.3411,
-      "step": 37821
-    },
-    {
-      "epoch": 19.72464146023468,
-      "grad_norm": 1.5687726736068726,
-      "learning_rate": 6.250351758793969e-05,
-      "loss": 5.3951,
-      "step": 37822
-    },
-    {
-      "epoch": 19.7251629726206,
-      "grad_norm": 1.4483989477157593,
-      "learning_rate": 6.250251256281407e-05,
-      "loss": 5.7638,
-      "step": 37823
-    },
-    {
-      "epoch": 19.725684485006518,
-      "grad_norm": 1.5446287393569946,
-      "learning_rate": 6.250150753768845e-05,
-      "loss": 5.5115,
-      "step": 37824
-    },
-    {
-      "epoch": 19.726205997392437,
-      "grad_norm": 1.5480376482009888,
-      "learning_rate": 6.250050251256283e-05,
-      "loss": 4.7803,
-      "step": 37825
-    },
-    {
-      "epoch": 19.726727509778357,
-      "grad_norm": 1.5428684949874878,
-      "learning_rate": 6.249949748743719e-05,
-      "loss": 4.9716,
-      "step": 37826
-    },
-    {
-      "epoch": 19.727249022164276,
-      "grad_norm": 1.5290207862854004,
-      "learning_rate": 6.249849246231157e-05,
-      "loss": 5.2947,
-      "step": 37827
-    },
-    {
-      "epoch": 19.727770534550196,
-      "grad_norm": 1.4951180219650269,
-      "learning_rate": 6.249748743718593e-05,
-      "loss": 4.4958,
-      "step": 37828
-    },
-    {
-      "epoch": 19.728292046936115,
-      "grad_norm": 1.4740008115768433,
-      "learning_rate": 6.249648241206031e-05,
-      "loss": 5.2654,
-      "step": 37829
-    },
-    {
-      "epoch": 19.728813559322035,
-      "grad_norm": 1.5051628351211548,
-      "learning_rate": 6.249547738693467e-05,
-      "loss": 5.1327,
-      "step": 37830
-    },
-    {
-      "epoch": 19.729335071707954,
-      "grad_norm": 1.6264816522598267,
-      "learning_rate": 6.249447236180905e-05,
-      "loss": 4.9842,
-      "step": 37831
-    },
-    {
-      "epoch": 19.729856584093874,
-      "grad_norm": 1.5140687227249146,
-      "learning_rate": 6.249346733668342e-05,
-      "loss": 4.8895,
-      "step": 37832
-    },
-    {
-      "epoch": 19.730378096479793,
-      "grad_norm": 1.4988443851470947,
-      "learning_rate": 6.249246231155778e-05,
-      "loss": 5.5743,
-      "step": 37833
-    },
-    {
-      "epoch": 19.73089960886571,
-      "grad_norm": 1.433921456336975,
-      "learning_rate": 6.249145728643216e-05,
-      "loss": 5.3213,
-      "step": 37834
-    },
-    {
-      "epoch": 19.73142112125163,
-      "grad_norm": 1.4616093635559082,
-      "learning_rate": 6.249045226130654e-05,
-      "loss": 5.2684,
-      "step": 37835
-    },
-    {
-      "epoch": 19.731942633637548,
-      "grad_norm": 1.4831050634384155,
-      "learning_rate": 6.248944723618091e-05,
-      "loss": 5.7318,
-      "step": 37836
-    },
-    {
-      "epoch": 19.732464146023467,
-      "grad_norm": 1.5151301622390747,
-      "learning_rate": 6.248844221105528e-05,
-      "loss": 5.4818,
-      "step": 37837
-    },
-    {
-      "epoch": 19.732985658409387,
-      "grad_norm": 1.5885009765625,
-      "learning_rate": 6.248743718592966e-05,
-      "loss": 5.328,
-      "step": 37838
-    },
-    {
-      "epoch": 19.733507170795306,
-      "grad_norm": 1.5005912780761719,
-      "learning_rate": 6.248643216080402e-05,
-      "loss": 5.2076,
-      "step": 37839
-    },
-    {
-      "epoch": 19.734028683181226,
-      "grad_norm": 1.4275038242340088,
-      "learning_rate": 6.24854271356784e-05,
-      "loss": 5.1536,
-      "step": 37840
-    },
-    {
-      "epoch": 19.734550195567145,
-      "grad_norm": 1.6004914045333862,
-      "learning_rate": 6.248442211055276e-05,
-      "loss": 5.3333,
-      "step": 37841
-    },
-    {
-      "epoch": 19.735071707953065,
-      "grad_norm": 1.5905088186264038,
-      "learning_rate": 6.248341708542714e-05,
-      "loss": 4.7954,
-      "step": 37842
-    },
-    {
-      "epoch": 19.735593220338984,
-      "grad_norm": 1.3899710178375244,
-      "learning_rate": 6.24824120603015e-05,
-      "loss": 4.2084,
-      "step": 37843
-    },
-    {
-      "epoch": 19.736114732724904,
-      "grad_norm": 1.581222653388977,
-      "learning_rate": 6.248140703517588e-05,
-      "loss": 5.0418,
-      "step": 37844
-    },
-    {
-      "epoch": 19.736636245110823,
-      "grad_norm": 1.5096070766448975,
-      "learning_rate": 6.248040201005026e-05,
-      "loss": 5.6808,
-      "step": 37845
-    },
-    {
-      "epoch": 19.73715775749674,
-      "grad_norm": 1.4473145008087158,
-      "learning_rate": 6.247939698492462e-05,
-      "loss": 5.0456,
-      "step": 37846
-    },
-    {
-      "epoch": 19.73767926988266,
-      "grad_norm": 1.4710744619369507,
-      "learning_rate": 6.2478391959799e-05,
-      "loss": 5.5726,
-      "step": 37847
-    },
-    {
-      "epoch": 19.738200782268578,
-      "grad_norm": 1.4862555265426636,
-      "learning_rate": 6.247738693467337e-05,
-      "loss": 5.1843,
-      "step": 37848
-    },
-    {
-      "epoch": 19.738722294654497,
-      "grad_norm": 1.5971487760543823,
-      "learning_rate": 6.247638190954774e-05,
-      "loss": 5.1056,
-      "step": 37849
-    },
-    {
-      "epoch": 19.739243807040417,
-      "grad_norm": 1.450016975402832,
-      "learning_rate": 6.247537688442211e-05,
-      "loss": 5.4541,
-      "step": 37850
-    },
-    {
-      "epoch": 19.739765319426336,
-      "grad_norm": 1.494406819343567,
-      "learning_rate": 6.247437185929649e-05,
-      "loss": 4.9449,
-      "step": 37851
-    },
-    {
-      "epoch": 19.740286831812256,
-      "grad_norm": 1.4493924379348755,
-      "learning_rate": 6.247336683417085e-05,
-      "loss": 4.8702,
-      "step": 37852
-    },
-    {
-      "epoch": 19.740808344198175,
-      "grad_norm": 1.5237534046173096,
-      "learning_rate": 6.247236180904523e-05,
-      "loss": 5.021,
-      "step": 37853
-    },
-    {
-      "epoch": 19.741329856584095,
-      "grad_norm": 1.3231899738311768,
-      "learning_rate": 6.247135678391959e-05,
-      "loss": 5.3605,
-      "step": 37854
-    },
-    {
-      "epoch": 19.741851368970014,
-      "grad_norm": 1.510671854019165,
-      "learning_rate": 6.247035175879397e-05,
-      "loss": 5.0252,
-      "step": 37855
-    },
-    {
-      "epoch": 19.742372881355934,
-      "grad_norm": 1.491106390953064,
-      "learning_rate": 6.246934673366835e-05,
-      "loss": 5.2942,
-      "step": 37856
-    },
-    {
-      "epoch": 19.742894393741853,
-      "grad_norm": 1.5492260456085205,
-      "learning_rate": 6.246834170854273e-05,
-      "loss": 5.5271,
-      "step": 37857
-    },
-    {
-      "epoch": 19.74341590612777,
-      "grad_norm": 1.4395604133605957,
-      "learning_rate": 6.246733668341709e-05,
-      "loss": 5.3536,
-      "step": 37858
-    },
-    {
-      "epoch": 19.74393741851369,
-      "grad_norm": 1.4375420808792114,
-      "learning_rate": 6.246633165829146e-05,
-      "loss": 5.6136,
-      "step": 37859
-    },
-    {
-      "epoch": 19.744458930899608,
-      "grad_norm": 1.3927416801452637,
-      "learning_rate": 6.246532663316583e-05,
-      "loss": 5.4892,
-      "step": 37860
-    },
-    {
-      "epoch": 19.744980443285527,
-      "grad_norm": 1.4871512651443481,
-      "learning_rate": 6.24643216080402e-05,
-      "loss": 5.3566,
-      "step": 37861
-    },
-    {
-      "epoch": 19.745501955671447,
-      "grad_norm": 1.4942916631698608,
-      "learning_rate": 6.246331658291458e-05,
-      "loss": 5.4643,
-      "step": 37862
-    },
-    {
-      "epoch": 19.746023468057366,
-      "grad_norm": 1.5165129899978638,
-      "learning_rate": 6.246231155778894e-05,
-      "loss": 4.7036,
-      "step": 37863
-    },
-    {
-      "epoch": 19.746544980443286,
-      "grad_norm": 1.4744783639907837,
-      "learning_rate": 6.246130653266332e-05,
-      "loss": 5.2952,
-      "step": 37864
-    },
-    {
-      "epoch": 19.747066492829205,
-      "grad_norm": 1.5818086862564087,
-      "learning_rate": 6.24603015075377e-05,
-      "loss": 4.7931,
-      "step": 37865
-    },
-    {
-      "epoch": 19.747588005215125,
-      "grad_norm": 1.559005618095398,
-      "learning_rate": 6.245929648241207e-05,
-      "loss": 5.2935,
-      "step": 37866
-    },
-    {
-      "epoch": 19.748109517601044,
-      "grad_norm": 1.5769602060317993,
-      "learning_rate": 6.245829145728644e-05,
-      "loss": 5.5368,
-      "step": 37867
-    },
-    {
-      "epoch": 19.748631029986964,
-      "grad_norm": 1.4772735834121704,
-      "learning_rate": 6.245728643216082e-05,
-      "loss": 5.065,
-      "step": 37868
-    },
-    {
-      "epoch": 19.749152542372883,
-      "grad_norm": 1.6992079019546509,
-      "learning_rate": 6.245628140703518e-05,
-      "loss": 4.9169,
-      "step": 37869
-    },
-    {
-      "epoch": 19.7496740547588,
-      "grad_norm": 1.5731006860733032,
-      "learning_rate": 6.245527638190956e-05,
-      "loss": 5.2755,
-      "step": 37870
-    },
-    {
-      "epoch": 19.75019556714472,
-      "grad_norm": 1.5770598649978638,
-      "learning_rate": 6.245427135678392e-05,
-      "loss": 5.3413,
-      "step": 37871
-    },
-    {
-      "epoch": 19.750717079530638,
-      "grad_norm": 1.4807945489883423,
-      "learning_rate": 6.24532663316583e-05,
-      "loss": 5.4304,
-      "step": 37872
-    },
-    {
-      "epoch": 19.751238591916557,
-      "grad_norm": 1.4881576299667358,
-      "learning_rate": 6.245226130653266e-05,
-      "loss": 5.3335,
-      "step": 37873
-    },
-    {
-      "epoch": 19.751760104302477,
-      "grad_norm": 1.4085116386413574,
-      "learning_rate": 6.245125628140703e-05,
-      "loss": 5.1879,
-      "step": 37874
-    },
-    {
-      "epoch": 19.752281616688396,
-      "grad_norm": 1.3979103565216064,
-      "learning_rate": 6.24502512562814e-05,
-      "loss": 5.1988,
-      "step": 37875
-    },
-    {
-      "epoch": 19.752803129074316,
-      "grad_norm": 1.5044398307800293,
-      "learning_rate": 6.244924623115578e-05,
-      "loss": 5.0744,
-      "step": 37876
-    },
-    {
-      "epoch": 19.753324641460235,
-      "grad_norm": 1.3856213092803955,
-      "learning_rate": 6.244824120603016e-05,
-      "loss": 5.45,
-      "step": 37877
-    },
-    {
-      "epoch": 19.753846153846155,
-      "grad_norm": 1.638953685760498,
-      "learning_rate": 6.244723618090453e-05,
-      "loss": 5.3878,
-      "step": 37878
-    },
-    {
-      "epoch": 19.754367666232074,
-      "grad_norm": 1.4830732345581055,
-      "learning_rate": 6.24462311557789e-05,
-      "loss": 5.073,
-      "step": 37879
-    },
-    {
-      "epoch": 19.754889178617994,
-      "grad_norm": 1.6036432981491089,
-      "learning_rate": 6.244522613065327e-05,
-      "loss": 5.1853,
-      "step": 37880
-    },
-    {
-      "epoch": 19.75541069100391,
-      "grad_norm": 1.4311625957489014,
-      "learning_rate": 6.244422110552765e-05,
-      "loss": 5.6717,
-      "step": 37881
-    },
-    {
-      "epoch": 19.75593220338983,
-      "grad_norm": 1.5456891059875488,
-      "learning_rate": 6.244321608040201e-05,
-      "loss": 5.2779,
-      "step": 37882
-    },
-    {
-      "epoch": 19.75645371577575,
-      "grad_norm": 1.4808745384216309,
-      "learning_rate": 6.244221105527639e-05,
-      "loss": 5.4308,
-      "step": 37883
-    },
-    {
-      "epoch": 19.756975228161668,
-      "grad_norm": 1.6404788494110107,
-      "learning_rate": 6.244120603015075e-05,
-      "loss": 4.6887,
-      "step": 37884
-    },
-    {
-      "epoch": 19.757496740547587,
-      "grad_norm": 1.6766293048858643,
-      "learning_rate": 6.244020100502513e-05,
-      "loss": 5.2705,
-      "step": 37885
-    },
-    {
-      "epoch": 19.758018252933507,
-      "grad_norm": 1.459170937538147,
-      "learning_rate": 6.243919597989951e-05,
-      "loss": 5.1837,
-      "step": 37886
-    },
-    {
-      "epoch": 19.758539765319426,
-      "grad_norm": 1.4335148334503174,
-      "learning_rate": 6.243819095477387e-05,
-      "loss": 5.6594,
-      "step": 37887
-    },
-    {
-      "epoch": 19.759061277705346,
-      "grad_norm": 1.5123573541641235,
-      "learning_rate": 6.243718592964825e-05,
-      "loss": 5.3992,
-      "step": 37888
-    },
-    {
-      "epoch": 19.759582790091265,
-      "grad_norm": 1.4564015865325928,
-      "learning_rate": 6.243618090452261e-05,
-      "loss": 5.4688,
-      "step": 37889
-    },
-    {
-      "epoch": 19.760104302477185,
-      "grad_norm": 1.5538996458053589,
-      "learning_rate": 6.243517587939699e-05,
-      "loss": 5.2644,
-      "step": 37890
-    },
-    {
-      "epoch": 19.760625814863104,
-      "grad_norm": 1.4885549545288086,
-      "learning_rate": 6.243417085427136e-05,
-      "loss": 4.8914,
-      "step": 37891
-    },
-    {
-      "epoch": 19.761147327249024,
-      "grad_norm": 1.5047187805175781,
-      "learning_rate": 6.243316582914573e-05,
-      "loss": 5.3489,
-      "step": 37892
-    },
-    {
-      "epoch": 19.761668839634943,
-      "grad_norm": 1.5376787185668945,
-      "learning_rate": 6.24321608040201e-05,
-      "loss": 4.8675,
-      "step": 37893
-    },
-    {
-      "epoch": 19.76219035202086,
-      "grad_norm": 1.5502614974975586,
-      "learning_rate": 6.243115577889448e-05,
-      "loss": 5.3057,
-      "step": 37894
-    },
-    {
-      "epoch": 19.76271186440678,
-      "grad_norm": 1.5024405717849731,
-      "learning_rate": 6.243015075376884e-05,
-      "loss": 5.4517,
-      "step": 37895
-    },
-    {
-      "epoch": 19.763233376792698,
-      "grad_norm": 1.478054404258728,
-      "learning_rate": 6.242914572864322e-05,
-      "loss": 5.5504,
-      "step": 37896
-    },
-    {
-      "epoch": 19.763754889178617,
-      "grad_norm": 1.4159501791000366,
-      "learning_rate": 6.24281407035176e-05,
-      "loss": 4.9256,
-      "step": 37897
-    },
-    {
-      "epoch": 19.764276401564537,
-      "grad_norm": 1.5996776819229126,
-      "learning_rate": 6.242713567839197e-05,
-      "loss": 5.2488,
-      "step": 37898
-    },
-    {
-      "epoch": 19.764797913950456,
-      "grad_norm": 1.3936280012130737,
-      "learning_rate": 6.242613065326634e-05,
-      "loss": 5.4103,
-      "step": 37899
-    },
-    {
-      "epoch": 19.765319426336376,
-      "grad_norm": 1.4405661821365356,
-      "learning_rate": 6.24251256281407e-05,
-      "loss": 5.5773,
-      "step": 37900
-    },
-    {
-      "epoch": 19.765840938722295,
-      "grad_norm": 1.6067155599594116,
-      "learning_rate": 6.242412060301508e-05,
-      "loss": 5.2757,
-      "step": 37901
-    },
-    {
-      "epoch": 19.766362451108215,
-      "grad_norm": 1.5616137981414795,
-      "learning_rate": 6.242311557788944e-05,
-      "loss": 5.0349,
-      "step": 37902
-    },
-    {
-      "epoch": 19.766883963494134,
-      "grad_norm": 1.45431387424469,
-      "learning_rate": 6.242211055276382e-05,
-      "loss": 5.581,
-      "step": 37903
-    },
-    {
-      "epoch": 19.767405475880054,
-      "grad_norm": 1.5646575689315796,
-      "learning_rate": 6.242110552763819e-05,
-      "loss": 5.4041,
-      "step": 37904
-    },
-    {
-      "epoch": 19.76792698826597,
-      "grad_norm": 1.534603238105774,
-      "learning_rate": 6.242010050251256e-05,
-      "loss": 5.2977,
-      "step": 37905
-    },
-    {
-      "epoch": 19.76844850065189,
-      "grad_norm": 1.3910027742385864,
-      "learning_rate": 6.241909547738693e-05,
-      "loss": 5.7659,
-      "step": 37906
-    },
-    {
-      "epoch": 19.76897001303781,
-      "grad_norm": 1.561954140663147,
-      "learning_rate": 6.24180904522613e-05,
-      "loss": 5.3435,
-      "step": 37907
-    },
-    {
-      "epoch": 19.769491525423728,
-      "grad_norm": 1.3996227979660034,
-      "learning_rate": 6.241708542713568e-05,
-      "loss": 5.7221,
-      "step": 37908
-    },
-    {
-      "epoch": 19.770013037809647,
-      "grad_norm": 1.5145820379257202,
-      "learning_rate": 6.241608040201006e-05,
-      "loss": 4.8949,
-      "step": 37909
-    },
-    {
-      "epoch": 19.770534550195567,
-      "grad_norm": 1.5196846723556519,
-      "learning_rate": 6.241507537688443e-05,
-      "loss": 5.5613,
-      "step": 37910
-    },
-    {
-      "epoch": 19.771056062581486,
-      "grad_norm": 1.6124944686889648,
-      "learning_rate": 6.24140703517588e-05,
-      "loss": 5.4493,
-      "step": 37911
-    },
-    {
-      "epoch": 19.771577574967406,
-      "grad_norm": 1.518170952796936,
-      "learning_rate": 6.241306532663317e-05,
-      "loss": 5.1691,
-      "step": 37912
-    },
-    {
-      "epoch": 19.772099087353325,
-      "grad_norm": 1.507743239402771,
-      "learning_rate": 6.241206030150753e-05,
-      "loss": 4.3393,
-      "step": 37913
-    },
-    {
-      "epoch": 19.772620599739245,
-      "grad_norm": 1.4252804517745972,
-      "learning_rate": 6.241105527638191e-05,
-      "loss": 5.4384,
-      "step": 37914
-    },
-    {
-      "epoch": 19.773142112125164,
-      "grad_norm": 1.4261409044265747,
-      "learning_rate": 6.241005025125627e-05,
-      "loss": 5.182,
-      "step": 37915
-    },
-    {
-      "epoch": 19.773663624511084,
-      "grad_norm": 1.6514545679092407,
-      "learning_rate": 6.240904522613065e-05,
-      "loss": 5.1475,
-      "step": 37916
-    },
-    {
-      "epoch": 19.774185136897,
-      "grad_norm": 1.5993057489395142,
-      "learning_rate": 6.240804020100503e-05,
-      "loss": 5.1386,
-      "step": 37917
-    },
-    {
-      "epoch": 19.77470664928292,
-      "grad_norm": 1.5274536609649658,
-      "learning_rate": 6.240703517587941e-05,
-      "loss": 4.7128,
-      "step": 37918
-    },
-    {
-      "epoch": 19.77522816166884,
-      "grad_norm": 1.4728682041168213,
-      "learning_rate": 6.240603015075377e-05,
-      "loss": 5.7014,
-      "step": 37919
-    },
-    {
-      "epoch": 19.775749674054758,
-      "grad_norm": 1.5567195415496826,
-      "learning_rate": 6.240502512562815e-05,
-      "loss": 5.2897,
-      "step": 37920
-    },
-    {
-      "epoch": 19.776271186440677,
-      "grad_norm": 1.5591979026794434,
-      "learning_rate": 6.240402010050251e-05,
-      "loss": 4.785,
-      "step": 37921
-    },
-    {
-      "epoch": 19.776792698826597,
-      "grad_norm": 1.4901695251464844,
-      "learning_rate": 6.240301507537689e-05,
-      "loss": 5.5088,
-      "step": 37922
-    },
-    {
-      "epoch": 19.777314211212516,
-      "grad_norm": 1.5758423805236816,
-      "learning_rate": 6.240201005025126e-05,
-      "loss": 5.4174,
-      "step": 37923
-    },
-    {
-      "epoch": 19.777835723598436,
-      "grad_norm": 1.6006519794464111,
-      "learning_rate": 6.240100502512563e-05,
-      "loss": 4.9604,
-      "step": 37924
-    },
-    {
-      "epoch": 19.778357235984355,
-      "grad_norm": 1.5476160049438477,
-      "learning_rate": 6.24e-05,
-      "loss": 4.912,
-      "step": 37925
-    },
-    {
-      "epoch": 19.778878748370275,
-      "grad_norm": 1.553688645362854,
-      "learning_rate": 6.239899497487436e-05,
-      "loss": 5.3405,
-      "step": 37926
-    },
-    {
-      "epoch": 19.779400260756194,
-      "grad_norm": 1.5041866302490234,
-      "learning_rate": 6.239798994974874e-05,
-      "loss": 5.2451,
-      "step": 37927
-    },
-    {
-      "epoch": 19.779921773142114,
-      "grad_norm": 1.5947567224502563,
-      "learning_rate": 6.239698492462312e-05,
-      "loss": 5.357,
-      "step": 37928
-    },
-    {
-      "epoch": 19.78044328552803,
-      "grad_norm": 1.445551872253418,
-      "learning_rate": 6.23959798994975e-05,
-      "loss": 5.5081,
-      "step": 37929
-    },
-    {
-      "epoch": 19.78096479791395,
-      "grad_norm": 1.4842232465744019,
-      "learning_rate": 6.239497487437186e-05,
-      "loss": 5.2026,
-      "step": 37930
-    },
-    {
-      "epoch": 19.78148631029987,
-      "grad_norm": 1.618261456489563,
-      "learning_rate": 6.239396984924624e-05,
-      "loss": 5.254,
-      "step": 37931
-    },
-    {
-      "epoch": 19.782007822685788,
-      "grad_norm": 1.6319878101348877,
-      "learning_rate": 6.23929648241206e-05,
-      "loss": 5.2088,
-      "step": 37932
-    },
-    {
-      "epoch": 19.782529335071708,
-      "grad_norm": 1.631272554397583,
-      "learning_rate": 6.239195979899498e-05,
-      "loss": 4.4034,
-      "step": 37933
-    },
-    {
-      "epoch": 19.783050847457627,
-      "grad_norm": 1.582678198814392,
-      "learning_rate": 6.239095477386935e-05,
-      "loss": 4.6646,
-      "step": 37934
-    },
-    {
-      "epoch": 19.783572359843546,
-      "grad_norm": 1.624849557876587,
-      "learning_rate": 6.238994974874372e-05,
-      "loss": 5.4356,
-      "step": 37935
-    },
-    {
-      "epoch": 19.784093872229466,
-      "grad_norm": 1.5852569341659546,
-      "learning_rate": 6.238894472361809e-05,
-      "loss": 5.3472,
-      "step": 37936
-    },
-    {
-      "epoch": 19.784615384615385,
-      "grad_norm": 1.5713002681732178,
-      "learning_rate": 6.238793969849247e-05,
-      "loss": 5.8222,
-      "step": 37937
-    },
-    {
-      "epoch": 19.785136897001305,
-      "grad_norm": 1.6058754920959473,
-      "learning_rate": 6.238693467336684e-05,
-      "loss": 5.301,
-      "step": 37938
-    },
-    {
-      "epoch": 19.785658409387224,
-      "grad_norm": 1.4954670667648315,
-      "learning_rate": 6.238592964824121e-05,
-      "loss": 5.6014,
-      "step": 37939
-    },
-    {
-      "epoch": 19.786179921773144,
-      "grad_norm": 1.6555826663970947,
-      "learning_rate": 6.238492462311558e-05,
-      "loss": 5.096,
-      "step": 37940
-    },
-    {
-      "epoch": 19.78670143415906,
-      "grad_norm": 1.4372092485427856,
-      "learning_rate": 6.238391959798995e-05,
-      "loss": 5.05,
-      "step": 37941
-    },
-    {
-      "epoch": 19.78722294654498,
-      "grad_norm": 1.5183593034744263,
-      "learning_rate": 6.238291457286433e-05,
-      "loss": 5.0198,
-      "step": 37942
-    },
-    {
-      "epoch": 19.7877444589309,
-      "grad_norm": 1.54574716091156,
-      "learning_rate": 6.238190954773869e-05,
-      "loss": 5.0073,
-      "step": 37943
-    },
-    {
-      "epoch": 19.788265971316818,
-      "grad_norm": 1.5377893447875977,
-      "learning_rate": 6.238090452261307e-05,
-      "loss": 5.4108,
-      "step": 37944
-    },
-    {
-      "epoch": 19.788787483702738,
-      "grad_norm": 1.5202785730361938,
-      "learning_rate": 6.237989949748743e-05,
-      "loss": 5.4633,
-      "step": 37945
-    },
-    {
-      "epoch": 19.789308996088657,
-      "grad_norm": 1.457612156867981,
-      "learning_rate": 6.237889447236181e-05,
-      "loss": 5.0786,
-      "step": 37946
-    },
-    {
-      "epoch": 19.789830508474576,
-      "grad_norm": 1.5222173929214478,
-      "learning_rate": 6.237788944723618e-05,
-      "loss": 5.0004,
-      "step": 37947
-    },
-    {
-      "epoch": 19.790352020860496,
-      "grad_norm": 1.410109281539917,
-      "learning_rate": 6.237688442211055e-05,
-      "loss": 5.2428,
-      "step": 37948
-    },
-    {
-      "epoch": 19.790873533246415,
-      "grad_norm": 1.531026840209961,
-      "learning_rate": 6.237587939698493e-05,
-      "loss": 5.4401,
-      "step": 37949
-    },
-    {
-      "epoch": 19.791395045632335,
-      "grad_norm": 1.6082311868667603,
-      "learning_rate": 6.237487437185931e-05,
-      "loss": 5.1511,
-      "step": 37950
-    },
-    {
-      "epoch": 19.791916558018254,
-      "grad_norm": 1.5064042806625366,
-      "learning_rate": 6.237386934673367e-05,
-      "loss": 4.7823,
-      "step": 37951
-    },
-    {
-      "epoch": 19.792438070404174,
-      "grad_norm": 1.5897302627563477,
-      "learning_rate": 6.237286432160804e-05,
-      "loss": 5.196,
-      "step": 37952
-    },
-    {
-      "epoch": 19.79295958279009,
-      "grad_norm": 1.5613031387329102,
-      "learning_rate": 6.237185929648242e-05,
-      "loss": 5.3008,
-      "step": 37953
-    },
-    {
-      "epoch": 19.79348109517601,
-      "grad_norm": 1.513170599937439,
-      "learning_rate": 6.237085427135678e-05,
-      "loss": 4.8138,
-      "step": 37954
-    },
-    {
-      "epoch": 19.79400260756193,
-      "grad_norm": 1.534480333328247,
-      "learning_rate": 6.236984924623116e-05,
-      "loss": 5.5589,
-      "step": 37955
-    },
-    {
-      "epoch": 19.794524119947848,
-      "grad_norm": 1.5180028676986694,
-      "learning_rate": 6.236884422110552e-05,
-      "loss": 5.6102,
-      "step": 37956
-    },
-    {
-      "epoch": 19.795045632333768,
-      "grad_norm": 1.3917313814163208,
-      "learning_rate": 6.23678391959799e-05,
-      "loss": 4.9517,
-      "step": 37957
-    },
-    {
-      "epoch": 19.795567144719687,
-      "grad_norm": 1.543663740158081,
-      "learning_rate": 6.236683417085428e-05,
-      "loss": 5.3366,
-      "step": 37958
-    },
-    {
-      "epoch": 19.796088657105607,
-      "grad_norm": 1.4171018600463867,
-      "learning_rate": 6.236582914572866e-05,
-      "loss": 5.616,
-      "step": 37959
-    },
-    {
-      "epoch": 19.796610169491526,
-      "grad_norm": 1.5353766679763794,
-      "learning_rate": 6.236482412060302e-05,
-      "loss": 5.531,
-      "step": 37960
-    },
-    {
-      "epoch": 19.797131681877445,
-      "grad_norm": 1.4977765083312988,
-      "learning_rate": 6.23638190954774e-05,
-      "loss": 5.1348,
-      "step": 37961
-    },
-    {
-      "epoch": 19.797653194263365,
-      "grad_norm": 1.5105645656585693,
-      "learning_rate": 6.236281407035176e-05,
-      "loss": 5.7478,
-      "step": 37962
-    },
-    {
-      "epoch": 19.798174706649284,
-      "grad_norm": 1.4885071516036987,
-      "learning_rate": 6.236180904522614e-05,
-      "loss": 5.0597,
-      "step": 37963
-    },
-    {
-      "epoch": 19.7986962190352,
-      "grad_norm": 1.6074190139770508,
-      "learning_rate": 6.23608040201005e-05,
-      "loss": 4.1745,
-      "step": 37964
-    },
-    {
-      "epoch": 19.79921773142112,
-      "grad_norm": 1.4167407751083374,
-      "learning_rate": 6.235979899497488e-05,
-      "loss": 5.9348,
-      "step": 37965
-    },
-    {
-      "epoch": 19.79973924380704,
-      "grad_norm": 1.4617085456848145,
-      "learning_rate": 6.235879396984925e-05,
-      "loss": 5.5878,
-      "step": 37966
-    },
-    {
-      "epoch": 19.80026075619296,
-      "grad_norm": 1.5409557819366455,
-      "learning_rate": 6.235778894472361e-05,
-      "loss": 5.1928,
-      "step": 37967
-    },
-    {
-      "epoch": 19.800782268578878,
-      "grad_norm": 1.5670197010040283,
-      "learning_rate": 6.235678391959799e-05,
-      "loss": 5.0331,
-      "step": 37968
-    },
-    {
-      "epoch": 19.801303780964798,
-      "grad_norm": 1.4615172147750854,
-      "learning_rate": 6.235577889447237e-05,
-      "loss": 5.2866,
-      "step": 37969
-    },
-    {
-      "epoch": 19.801825293350717,
-      "grad_norm": 1.4920437335968018,
-      "learning_rate": 6.235477386934674e-05,
-      "loss": 5.3059,
-      "step": 37970
-    },
-    {
-      "epoch": 19.802346805736637,
-      "grad_norm": 1.4995673894882202,
-      "learning_rate": 6.235376884422111e-05,
-      "loss": 5.1503,
-      "step": 37971
-    },
-    {
-      "epoch": 19.802868318122556,
-      "grad_norm": 1.530464768409729,
-      "learning_rate": 6.235276381909549e-05,
-      "loss": 4.8525,
-      "step": 37972
-    },
-    {
-      "epoch": 19.803389830508475,
-      "grad_norm": 1.4550292491912842,
-      "learning_rate": 6.235175879396985e-05,
-      "loss": 5.317,
-      "step": 37973
-    },
-    {
-      "epoch": 19.803911342894395,
-      "grad_norm": 1.5828214883804321,
-      "learning_rate": 6.235075376884423e-05,
-      "loss": 5.4188,
-      "step": 37974
-    },
-    {
-      "epoch": 19.804432855280314,
-      "grad_norm": 1.4898735284805298,
-      "learning_rate": 6.234974874371859e-05,
-      "loss": 5.151,
-      "step": 37975
-    },
-    {
-      "epoch": 19.804954367666234,
-      "grad_norm": 1.5183597803115845,
-      "learning_rate": 6.234874371859297e-05,
-      "loss": 4.9042,
-      "step": 37976
-    },
-    {
-      "epoch": 19.80547588005215,
-      "grad_norm": 1.5182961225509644,
-      "learning_rate": 6.234773869346733e-05,
-      "loss": 5.1099,
-      "step": 37977
-    },
-    {
-      "epoch": 19.80599739243807,
-      "grad_norm": 1.554877758026123,
-      "learning_rate": 6.234673366834171e-05,
-      "loss": 4.5279,
-      "step": 37978
-    },
-    {
-      "epoch": 19.80651890482399,
-      "grad_norm": 1.5954318046569824,
-      "learning_rate": 6.234572864321609e-05,
-      "loss": 5.2507,
-      "step": 37979
-    },
-    {
-      "epoch": 19.807040417209908,
-      "grad_norm": 1.5655957460403442,
-      "learning_rate": 6.234472361809045e-05,
-      "loss": 5.2511,
-      "step": 37980
-    },
-    {
-      "epoch": 19.807561929595828,
-      "grad_norm": 1.600876808166504,
-      "learning_rate": 6.234371859296483e-05,
-      "loss": 5.3997,
-      "step": 37981
-    },
-    {
-      "epoch": 19.808083441981747,
-      "grad_norm": 1.4760515689849854,
-      "learning_rate": 6.23427135678392e-05,
-      "loss": 5.6209,
-      "step": 37982
-    },
-    {
-      "epoch": 19.808604954367667,
-      "grad_norm": 1.5411690473556519,
-      "learning_rate": 6.234170854271357e-05,
-      "loss": 5.2906,
-      "step": 37983
-    },
-    {
-      "epoch": 19.809126466753586,
-      "grad_norm": 1.3796164989471436,
-      "learning_rate": 6.234070351758794e-05,
-      "loss": 5.6678,
-      "step": 37984
-    },
-    {
-      "epoch": 19.809647979139505,
-      "grad_norm": 1.5040525197982788,
-      "learning_rate": 6.233969849246232e-05,
-      "loss": 5.1564,
-      "step": 37985
-    },
-    {
-      "epoch": 19.810169491525425,
-      "grad_norm": 1.504699468612671,
-      "learning_rate": 6.233869346733668e-05,
-      "loss": 5.3602,
-      "step": 37986
-    },
-    {
-      "epoch": 19.810691003911344,
-      "grad_norm": 1.3707400560379028,
-      "learning_rate": 6.233768844221106e-05,
-      "loss": 5.6488,
-      "step": 37987
-    },
-    {
-      "epoch": 19.81121251629726,
-      "grad_norm": 1.589952826499939,
-      "learning_rate": 6.233668341708542e-05,
-      "loss": 5.508,
-      "step": 37988
-    },
-    {
-      "epoch": 19.81173402868318,
-      "grad_norm": 1.4846524000167847,
-      "learning_rate": 6.23356783919598e-05,
-      "loss": 4.7587,
-      "step": 37989
-    },
-    {
-      "epoch": 19.8122555410691,
-      "grad_norm": 1.4885581731796265,
-      "learning_rate": 6.233467336683418e-05,
-      "loss": 5.7483,
-      "step": 37990
-    },
-    {
-      "epoch": 19.81277705345502,
-      "grad_norm": 1.443688988685608,
-      "learning_rate": 6.233366834170856e-05,
-      "loss": 5.5744,
-      "step": 37991
-    },
-    {
-      "epoch": 19.813298565840938,
-      "grad_norm": 1.4822362661361694,
-      "learning_rate": 6.233266331658292e-05,
-      "loss": 5.6398,
-      "step": 37992
-    },
-    {
-      "epoch": 19.813820078226858,
-      "grad_norm": 1.5231585502624512,
-      "learning_rate": 6.233165829145728e-05,
-      "loss": 4.8695,
-      "step": 37993
-    },
-    {
-      "epoch": 19.814341590612777,
-      "grad_norm": 1.5086250305175781,
-      "learning_rate": 6.233065326633166e-05,
-      "loss": 5.2059,
-      "step": 37994
-    },
-    {
-      "epoch": 19.814863102998697,
-      "grad_norm": 1.513839840888977,
-      "learning_rate": 6.232964824120603e-05,
-      "loss": 5.6504,
-      "step": 37995
-    },
-    {
-      "epoch": 19.815384615384616,
-      "grad_norm": 1.5438265800476074,
-      "learning_rate": 6.23286432160804e-05,
-      "loss": 5.3498,
-      "step": 37996
-    },
-    {
-      "epoch": 19.815906127770536,
-      "grad_norm": 1.6211137771606445,
-      "learning_rate": 6.232763819095477e-05,
-      "loss": 5.095,
-      "step": 37997
-    },
-    {
-      "epoch": 19.816427640156455,
-      "grad_norm": 1.534665584564209,
-      "learning_rate": 6.232663316582915e-05,
-      "loss": 5.5688,
-      "step": 37998
-    },
-    {
-      "epoch": 19.816949152542374,
-      "grad_norm": 1.517006278038025,
-      "learning_rate": 6.232562814070352e-05,
-      "loss": 5.4629,
-      "step": 37999
-    },
-    {
-      "epoch": 19.81747066492829,
-      "grad_norm": 1.6067497730255127,
-      "learning_rate": 6.23246231155779e-05,
-      "loss": 5.231,
-      "step": 38000
-    },
-    {
-      "epoch": 19.81799217731421,
-      "grad_norm": 1.624896764755249,
-      "learning_rate": 6.232361809045227e-05,
-      "loss": 5.4177,
-      "step": 38001
-    },
-    {
-      "epoch": 19.81851368970013,
-      "grad_norm": 1.704095721244812,
-      "learning_rate": 6.232261306532664e-05,
-      "loss": 5.3702,
-      "step": 38002
-    },
-    {
-      "epoch": 19.81903520208605,
-      "grad_norm": 1.5762667655944824,
-      "learning_rate": 6.232160804020101e-05,
-      "loss": 5.0502,
-      "step": 38003
-    },
-    {
-      "epoch": 19.819556714471968,
-      "grad_norm": 1.501335859298706,
-      "learning_rate": 6.232060301507539e-05,
-      "loss": 5.061,
-      "step": 38004
-    },
-    {
-      "epoch": 19.820078226857888,
-      "grad_norm": 2.148737668991089,
-      "learning_rate": 6.231959798994975e-05,
-      "loss": 4.6081,
-      "step": 38005
-    },
-    {
-      "epoch": 19.820599739243807,
-      "grad_norm": 1.45090651512146,
-      "learning_rate": 6.231859296482412e-05,
-      "loss": 4.9161,
-      "step": 38006
-    },
-    {
-      "epoch": 19.821121251629727,
-      "grad_norm": 1.464372992515564,
-      "learning_rate": 6.231758793969849e-05,
-      "loss": 5.4176,
-      "step": 38007
-    },
-    {
-      "epoch": 19.821642764015646,
-      "grad_norm": 1.4733357429504395,
-      "learning_rate": 6.231658291457286e-05,
-      "loss": 5.3914,
-      "step": 38008
-    },
-    {
-      "epoch": 19.822164276401566,
-      "grad_norm": 1.3993964195251465,
-      "learning_rate": 6.231557788944723e-05,
-      "loss": 5.5884,
-      "step": 38009
-    },
-    {
-      "epoch": 19.822685788787485,
-      "grad_norm": 1.5895321369171143,
-      "learning_rate": 6.231457286432161e-05,
-      "loss": 5.2185,
-      "step": 38010
-    },
-    {
-      "epoch": 19.823207301173404,
-      "grad_norm": 1.6116583347320557,
-      "learning_rate": 6.231356783919599e-05,
-      "loss": 4.7286,
-      "step": 38011
-    },
-    {
-      "epoch": 19.82372881355932,
-      "grad_norm": 1.4666268825531006,
-      "learning_rate": 6.231256281407035e-05,
-      "loss": 5.1135,
-      "step": 38012
-    },
-    {
-      "epoch": 19.82425032594524,
-      "grad_norm": 1.5283137559890747,
-      "learning_rate": 6.231155778894473e-05,
-      "loss": 4.8095,
-      "step": 38013
-    },
-    {
-      "epoch": 19.82477183833116,
-      "grad_norm": 1.5556869506835938,
-      "learning_rate": 6.23105527638191e-05,
-      "loss": 5.3667,
-      "step": 38014
-    },
-    {
-      "epoch": 19.82529335071708,
-      "grad_norm": 1.4338719844818115,
-      "learning_rate": 6.230954773869347e-05,
-      "loss": 5.4184,
-      "step": 38015
-    },
-    {
-      "epoch": 19.825814863103,
-      "grad_norm": 1.48334801197052,
-      "learning_rate": 6.230854271356784e-05,
-      "loss": 5.2668,
-      "step": 38016
-    },
-    {
-      "epoch": 19.826336375488918,
-      "grad_norm": 1.4707789421081543,
-      "learning_rate": 6.230753768844222e-05,
-      "loss": 5.5442,
-      "step": 38017
-    },
-    {
-      "epoch": 19.826857887874837,
-      "grad_norm": 1.5436851978302002,
-      "learning_rate": 6.230653266331658e-05,
-      "loss": 4.8246,
-      "step": 38018
-    },
-    {
-      "epoch": 19.827379400260757,
-      "grad_norm": 1.40238356590271,
-      "learning_rate": 6.230552763819096e-05,
-      "loss": 5.2318,
-      "step": 38019
-    },
-    {
-      "epoch": 19.827900912646676,
-      "grad_norm": 1.562132477760315,
-      "learning_rate": 6.230452261306534e-05,
-      "loss": 5.4156,
-      "step": 38020
-    },
-    {
-      "epoch": 19.828422425032596,
-      "grad_norm": 1.5392040014266968,
-      "learning_rate": 6.23035175879397e-05,
-      "loss": 5.4697,
-      "step": 38021
-    },
-    {
-      "epoch": 19.828943937418515,
-      "grad_norm": 1.4969412088394165,
-      "learning_rate": 6.230251256281408e-05,
-      "loss": 5.2666,
-      "step": 38022
-    },
-    {
-      "epoch": 19.829465449804434,
-      "grad_norm": 1.5227526426315308,
-      "learning_rate": 6.230150753768844e-05,
-      "loss": 5.1344,
-      "step": 38023
-    },
-    {
-      "epoch": 19.82998696219035,
-      "grad_norm": 1.471665382385254,
-      "learning_rate": 6.230050251256282e-05,
-      "loss": 5.1429,
-      "step": 38024
-    },
-    {
-      "epoch": 19.83050847457627,
-      "grad_norm": 1.4799587726593018,
-      "learning_rate": 6.229949748743719e-05,
-      "loss": 4.9117,
-      "step": 38025
-    },
-    {
-      "epoch": 19.83102998696219,
-      "grad_norm": 1.625701904296875,
-      "learning_rate": 6.229849246231156e-05,
-      "loss": 5.2241,
-      "step": 38026
-    },
-    {
-      "epoch": 19.83155149934811,
-      "grad_norm": 1.5110909938812256,
-      "learning_rate": 6.229748743718593e-05,
-      "loss": 4.9161,
-      "step": 38027
-    },
-    {
-      "epoch": 19.83207301173403,
-      "grad_norm": 1.4727725982666016,
-      "learning_rate": 6.22964824120603e-05,
-      "loss": 5.3655,
-      "step": 38028
-    },
-    {
-      "epoch": 19.832594524119948,
-      "grad_norm": 1.5114096403121948,
-      "learning_rate": 6.229547738693467e-05,
-      "loss": 5.381,
-      "step": 38029
-    },
-    {
-      "epoch": 19.833116036505867,
-      "grad_norm": 1.6261932849884033,
-      "learning_rate": 6.229447236180905e-05,
-      "loss": 5.3231,
-      "step": 38030
-    },
-    {
-      "epoch": 19.833637548891787,
-      "grad_norm": 1.558958649635315,
-      "learning_rate": 6.229346733668343e-05,
-      "loss": 4.9138,
-      "step": 38031
-    },
-    {
-      "epoch": 19.834159061277706,
-      "grad_norm": 1.6377509832382202,
-      "learning_rate": 6.229246231155779e-05,
-      "loss": 4.5949,
-      "step": 38032
-    },
-    {
-      "epoch": 19.834680573663626,
-      "grad_norm": 1.4197070598602295,
-      "learning_rate": 6.229145728643217e-05,
-      "loss": 5.3803,
-      "step": 38033
-    },
-    {
-      "epoch": 19.835202086049545,
-      "grad_norm": 1.4567188024520874,
-      "learning_rate": 6.229045226130653e-05,
-      "loss": 5.5583,
-      "step": 38034
-    },
-    {
-      "epoch": 19.835723598435465,
-      "grad_norm": 1.525467038154602,
-      "learning_rate": 6.228944723618091e-05,
-      "loss": 5.1433,
-      "step": 38035
-    },
-    {
-      "epoch": 19.83624511082138,
-      "grad_norm": 1.5763332843780518,
-      "learning_rate": 6.228844221105527e-05,
-      "loss": 5.2427,
-      "step": 38036
-    },
-    {
-      "epoch": 19.8367666232073,
-      "grad_norm": 1.5999610424041748,
-      "learning_rate": 6.228743718592965e-05,
-      "loss": 5.4941,
-      "step": 38037
-    },
-    {
-      "epoch": 19.83728813559322,
-      "grad_norm": 1.5742014646530151,
-      "learning_rate": 6.228643216080402e-05,
-      "loss": 5.4147,
-      "step": 38038
-    },
-    {
-      "epoch": 19.83780964797914,
-      "grad_norm": 1.6569479703903198,
-      "learning_rate": 6.22854271356784e-05,
-      "loss": 5.1514,
-      "step": 38039
-    },
-    {
-      "epoch": 19.83833116036506,
-      "grad_norm": 1.541446328163147,
-      "learning_rate": 6.228442211055277e-05,
-      "loss": 5.3103,
-      "step": 38040
-    },
-    {
-      "epoch": 19.838852672750978,
-      "grad_norm": 1.6543647050857544,
-      "learning_rate": 6.228341708542715e-05,
-      "loss": 4.8745,
-      "step": 38041
-    },
-    {
-      "epoch": 19.839374185136897,
-      "grad_norm": 1.4653807878494263,
-      "learning_rate": 6.228241206030151e-05,
-      "loss": 5.3947,
-      "step": 38042
-    },
-    {
-      "epoch": 19.839895697522817,
-      "grad_norm": 1.5723772048950195,
-      "learning_rate": 6.228140703517589e-05,
-      "loss": 5.4279,
-      "step": 38043
-    },
-    {
-      "epoch": 19.840417209908736,
-      "grad_norm": 1.4371367692947388,
-      "learning_rate": 6.228040201005026e-05,
-      "loss": 5.4613,
-      "step": 38044
-    },
-    {
-      "epoch": 19.840938722294656,
-      "grad_norm": 1.3919565677642822,
-      "learning_rate": 6.227939698492462e-05,
-      "loss": 5.686,
-      "step": 38045
-    },
-    {
-      "epoch": 19.841460234680575,
-      "grad_norm": 1.613714575767517,
-      "learning_rate": 6.2278391959799e-05,
-      "loss": 5.4214,
-      "step": 38046
-    },
-    {
-      "epoch": 19.841981747066495,
-      "grad_norm": 1.465040683746338,
-      "learning_rate": 6.227738693467336e-05,
-      "loss": 5.1775,
-      "step": 38047
-    },
-    {
-      "epoch": 19.84250325945241,
-      "grad_norm": 1.5393273830413818,
-      "learning_rate": 6.227638190954774e-05,
-      "loss": 5.2407,
-      "step": 38048
-    },
-    {
-      "epoch": 19.84302477183833,
-      "grad_norm": 1.5612901449203491,
-      "learning_rate": 6.22753768844221e-05,
-      "loss": 4.8177,
-      "step": 38049
-    },
-    {
-      "epoch": 19.84354628422425,
-      "grad_norm": 1.5520093441009521,
-      "learning_rate": 6.227437185929648e-05,
-      "loss": 4.6307,
-      "step": 38050
-    },
-    {
-      "epoch": 19.84406779661017,
-      "grad_norm": 1.4969916343688965,
-      "learning_rate": 6.227336683417086e-05,
-      "loss": 5.4872,
-      "step": 38051
-    },
-    {
-      "epoch": 19.84458930899609,
-      "grad_norm": 1.460471510887146,
-      "learning_rate": 6.227236180904524e-05,
-      "loss": 5.5307,
-      "step": 38052
-    },
-    {
-      "epoch": 19.845110821382008,
-      "grad_norm": 1.4979948997497559,
-      "learning_rate": 6.22713567839196e-05,
-      "loss": 5.2736,
-      "step": 38053
-    },
-    {
-      "epoch": 19.845632333767927,
-      "grad_norm": 1.5570520162582397,
-      "learning_rate": 6.227035175879398e-05,
-      "loss": 5.0863,
-      "step": 38054
-    },
-    {
-      "epoch": 19.846153846153847,
-      "grad_norm": 1.5851945877075195,
-      "learning_rate": 6.226934673366834e-05,
-      "loss": 5.3066,
-      "step": 38055
-    },
-    {
-      "epoch": 19.846675358539766,
-      "grad_norm": 1.4957586526870728,
-      "learning_rate": 6.226834170854272e-05,
-      "loss": 5.5075,
-      "step": 38056
-    },
-    {
-      "epoch": 19.847196870925686,
-      "grad_norm": 1.723888635635376,
-      "learning_rate": 6.226733668341709e-05,
-      "loss": 4.7469,
-      "step": 38057
-    },
-    {
-      "epoch": 19.847718383311605,
-      "grad_norm": 1.4808363914489746,
-      "learning_rate": 6.226633165829146e-05,
-      "loss": 5.3174,
-      "step": 38058
-    },
-    {
-      "epoch": 19.848239895697525,
-      "grad_norm": 1.4626599550247192,
-      "learning_rate": 6.226532663316583e-05,
-      "loss": 5.4731,
-      "step": 38059
-    },
-    {
-      "epoch": 19.84876140808344,
-      "grad_norm": 1.6283854246139526,
-      "learning_rate": 6.226432160804019e-05,
-      "loss": 5.6096,
-      "step": 38060
-    },
-    {
-      "epoch": 19.84928292046936,
-      "grad_norm": 1.5155583620071411,
-      "learning_rate": 6.226331658291457e-05,
-      "loss": 5.3367,
-      "step": 38061
-    },
-    {
-      "epoch": 19.84980443285528,
-      "grad_norm": 1.631174087524414,
-      "learning_rate": 6.226231155778895e-05,
-      "loss": 5.4392,
-      "step": 38062
-    },
-    {
-      "epoch": 19.8503259452412,
-      "grad_norm": 1.6434369087219238,
-      "learning_rate": 6.226130653266333e-05,
-      "loss": 5.1768,
-      "step": 38063
-    },
-    {
-      "epoch": 19.85084745762712,
-      "grad_norm": 1.601196527481079,
-      "learning_rate": 6.226030150753769e-05,
-      "loss": 5.4631,
-      "step": 38064
-    },
-    {
-      "epoch": 19.851368970013038,
-      "grad_norm": 1.4975336790084839,
-      "learning_rate": 6.225929648241207e-05,
-      "loss": 5.5221,
-      "step": 38065
-    },
-    {
-      "epoch": 19.851890482398957,
-      "grad_norm": 1.5268256664276123,
-      "learning_rate": 6.225829145728643e-05,
-      "loss": 5.2,
-      "step": 38066
-    },
-    {
-      "epoch": 19.852411994784877,
-      "grad_norm": 1.6482908725738525,
-      "learning_rate": 6.225728643216081e-05,
-      "loss": 5.017,
-      "step": 38067
-    },
-    {
-      "epoch": 19.852933507170796,
-      "grad_norm": 1.5571174621582031,
-      "learning_rate": 6.225628140703517e-05,
-      "loss": 5.2844,
-      "step": 38068
-    },
-    {
-      "epoch": 19.853455019556716,
-      "grad_norm": 1.6609224081039429,
-      "learning_rate": 6.225527638190955e-05,
-      "loss": 5.5004,
-      "step": 38069
-    },
-    {
-      "epoch": 19.853976531942635,
-      "grad_norm": 1.583567500114441,
-      "learning_rate": 6.225427135678392e-05,
-      "loss": 5.0465,
-      "step": 38070
-    },
-    {
-      "epoch": 19.85449804432855,
-      "grad_norm": 1.367842674255371,
-      "learning_rate": 6.22532663316583e-05,
-      "loss": 5.7591,
-      "step": 38071
-    },
-    {
-      "epoch": 19.85501955671447,
-      "grad_norm": 1.5644207000732422,
-      "learning_rate": 6.225226130653267e-05,
-      "loss": 5.0872,
-      "step": 38072
-    },
-    {
-      "epoch": 19.85554106910039,
-      "grad_norm": 1.4315071105957031,
-      "learning_rate": 6.225125628140704e-05,
-      "loss": 4.8854,
-      "step": 38073
-    },
-    {
-      "epoch": 19.85606258148631,
-      "grad_norm": 1.5598132610321045,
-      "learning_rate": 6.225025125628141e-05,
-      "loss": 5.5906,
-      "step": 38074
-    },
-    {
-      "epoch": 19.85658409387223,
-      "grad_norm": 1.4406980276107788,
-      "learning_rate": 6.224924623115578e-05,
-      "loss": 5.3627,
-      "step": 38075
-    },
-    {
-      "epoch": 19.85710560625815,
-      "grad_norm": 1.5026601552963257,
-      "learning_rate": 6.224824120603016e-05,
-      "loss": 4.8173,
-      "step": 38076
-    },
-    {
-      "epoch": 19.857627118644068,
-      "grad_norm": 1.4293718338012695,
-      "learning_rate": 6.224723618090452e-05,
-      "loss": 5.5717,
-      "step": 38077
-    },
-    {
-      "epoch": 19.858148631029987,
-      "grad_norm": 1.4541141986846924,
-      "learning_rate": 6.22462311557789e-05,
-      "loss": 5.1425,
-      "step": 38078
-    },
-    {
-      "epoch": 19.858670143415907,
-      "grad_norm": 1.5465751886367798,
-      "learning_rate": 6.224522613065326e-05,
-      "loss": 4.8516,
-      "step": 38079
-    },
-    {
-      "epoch": 19.859191655801826,
-      "grad_norm": 1.5505495071411133,
-      "learning_rate": 6.224422110552764e-05,
-      "loss": 5.4299,
-      "step": 38080
-    },
-    {
-      "epoch": 19.859713168187746,
-      "grad_norm": 1.5347495079040527,
-      "learning_rate": 6.2243216080402e-05,
-      "loss": 5.1183,
-      "step": 38081
-    },
-    {
-      "epoch": 19.860234680573665,
-      "grad_norm": 1.5249630212783813,
-      "learning_rate": 6.224221105527638e-05,
-      "loss": 5.0767,
-      "step": 38082
-    },
-    {
-      "epoch": 19.860756192959585,
-      "grad_norm": 1.6139954328536987,
-      "learning_rate": 6.224120603015076e-05,
-      "loss": 4.7193,
-      "step": 38083
-    },
-    {
-      "epoch": 19.8612777053455,
-      "grad_norm": 1.3960182666778564,
-      "learning_rate": 6.224020100502514e-05,
-      "loss": 5.7539,
-      "step": 38084
-    },
-    {
-      "epoch": 19.86179921773142,
-      "grad_norm": 1.5544743537902832,
-      "learning_rate": 6.22391959798995e-05,
-      "loss": 5.4848,
-      "step": 38085
-    },
-    {
-      "epoch": 19.86232073011734,
-      "grad_norm": 1.4651744365692139,
-      "learning_rate": 6.223819095477387e-05,
-      "loss": 4.9023,
-      "step": 38086
-    },
-    {
-      "epoch": 19.86284224250326,
-      "grad_norm": 1.378217339515686,
-      "learning_rate": 6.223718592964824e-05,
-      "loss": 5.5016,
-      "step": 38087
-    },
-    {
-      "epoch": 19.86336375488918,
-      "grad_norm": 1.5137689113616943,
-      "learning_rate": 6.223618090452261e-05,
-      "loss": 5.2097,
-      "step": 38088
-    },
-    {
-      "epoch": 19.863885267275098,
-      "grad_norm": 1.4685672521591187,
-      "learning_rate": 6.223517587939699e-05,
-      "loss": 5.2524,
-      "step": 38089
-    },
-    {
-      "epoch": 19.864406779661017,
-      "grad_norm": 1.4706100225448608,
-      "learning_rate": 6.223417085427135e-05,
-      "loss": 5.2208,
-      "step": 38090
-    },
-    {
-      "epoch": 19.864928292046937,
-      "grad_norm": 1.4041749238967896,
-      "learning_rate": 6.223316582914573e-05,
-      "loss": 5.3538,
-      "step": 38091
-    },
-    {
-      "epoch": 19.865449804432856,
-      "grad_norm": 1.5164860486984253,
-      "learning_rate": 6.223216080402011e-05,
-      "loss": 5.2101,
-      "step": 38092
-    },
-    {
-      "epoch": 19.865971316818776,
-      "grad_norm": 1.4187732934951782,
-      "learning_rate": 6.223115577889448e-05,
-      "loss": 5.0221,
-      "step": 38093
-    },
-    {
-      "epoch": 19.866492829204695,
-      "grad_norm": 1.5141184329986572,
-      "learning_rate": 6.223015075376885e-05,
-      "loss": 5.4959,
-      "step": 38094
-    },
-    {
-      "epoch": 19.86701434159061,
-      "grad_norm": 1.5288240909576416,
-      "learning_rate": 6.222914572864323e-05,
-      "loss": 5.3952,
-      "step": 38095
-    },
-    {
-      "epoch": 19.86753585397653,
-      "grad_norm": 1.4411494731903076,
-      "learning_rate": 6.222814070351759e-05,
-      "loss": 5.4695,
-      "step": 38096
-    },
-    {
-      "epoch": 19.86805736636245,
-      "grad_norm": 1.428125023841858,
-      "learning_rate": 6.222713567839197e-05,
-      "loss": 5.3916,
-      "step": 38097
-    },
-    {
-      "epoch": 19.86857887874837,
-      "grad_norm": 1.389845609664917,
-      "learning_rate": 6.222613065326633e-05,
-      "loss": 5.7761,
-      "step": 38098
-    },
-    {
-      "epoch": 19.86910039113429,
-      "grad_norm": 1.4753390550613403,
-      "learning_rate": 6.22251256281407e-05,
-      "loss": 5.7875,
-      "step": 38099
-    },
-    {
-      "epoch": 19.86962190352021,
-      "grad_norm": 1.570302128791809,
-      "learning_rate": 6.222412060301508e-05,
-      "loss": 5.0258,
-      "step": 38100
-    },
-    {
-      "epoch": 19.870143415906128,
-      "grad_norm": 1.5758273601531982,
-      "learning_rate": 6.222311557788944e-05,
-      "loss": 5.1677,
-      "step": 38101
-    },
-    {
-      "epoch": 19.870664928292047,
-      "grad_norm": 1.4620864391326904,
-      "learning_rate": 6.222211055276382e-05,
-      "loss": 5.4278,
-      "step": 38102
-    },
-    {
-      "epoch": 19.871186440677967,
-      "grad_norm": 1.450624942779541,
-      "learning_rate": 6.22211055276382e-05,
-      "loss": 5.7174,
-      "step": 38103
-    },
-    {
-      "epoch": 19.871707953063886,
-      "grad_norm": 1.5064587593078613,
-      "learning_rate": 6.222010050251257e-05,
-      "loss": 5.6004,
-      "step": 38104
-    },
-    {
-      "epoch": 19.872229465449806,
-      "grad_norm": 1.4028217792510986,
-      "learning_rate": 6.221909547738694e-05,
-      "loss": 5.5525,
-      "step": 38105
-    },
-    {
-      "epoch": 19.872750977835725,
-      "grad_norm": 1.4189088344573975,
-      "learning_rate": 6.221809045226132e-05,
-      "loss": 5.6656,
-      "step": 38106
-    },
-    {
-      "epoch": 19.87327249022164,
-      "grad_norm": 1.433026671409607,
-      "learning_rate": 6.221708542713568e-05,
-      "loss": 5.0296,
-      "step": 38107
-    },
-    {
-      "epoch": 19.87379400260756,
-      "grad_norm": 1.619960904121399,
-      "learning_rate": 6.221608040201006e-05,
-      "loss": 5.2135,
-      "step": 38108
-    },
-    {
-      "epoch": 19.87431551499348,
-      "grad_norm": 1.4980055093765259,
-      "learning_rate": 6.221507537688442e-05,
-      "loss": 5.4753,
-      "step": 38109
-    },
-    {
-      "epoch": 19.8748370273794,
-      "grad_norm": 1.436659812927246,
-      "learning_rate": 6.22140703517588e-05,
-      "loss": 5.0363,
-      "step": 38110
-    },
-    {
-      "epoch": 19.87535853976532,
-      "grad_norm": 1.4961206912994385,
-      "learning_rate": 6.221306532663316e-05,
-      "loss": 5.4807,
-      "step": 38111
-    },
-    {
-      "epoch": 19.87588005215124,
-      "grad_norm": 1.480179786682129,
-      "learning_rate": 6.221206030150754e-05,
-      "loss": 5.5051,
-      "step": 38112
-    },
-    {
-      "epoch": 19.876401564537158,
-      "grad_norm": 1.4929604530334473,
-      "learning_rate": 6.221105527638192e-05,
-      "loss": 5.3254,
-      "step": 38113
-    },
-    {
-      "epoch": 19.876923076923077,
-      "grad_norm": 1.5296822786331177,
-      "learning_rate": 6.221005025125628e-05,
-      "loss": 5.2485,
-      "step": 38114
-    },
-    {
-      "epoch": 19.877444589308997,
-      "grad_norm": 1.557716727256775,
-      "learning_rate": 6.220904522613066e-05,
-      "loss": 4.9773,
-      "step": 38115
-    },
-    {
-      "epoch": 19.877966101694916,
-      "grad_norm": 1.4799778461456299,
-      "learning_rate": 6.220804020100503e-05,
-      "loss": 5.4878,
-      "step": 38116
-    },
-    {
-      "epoch": 19.878487614080836,
-      "grad_norm": 1.4203803539276123,
-      "learning_rate": 6.22070351758794e-05,
-      "loss": 5.7354,
-      "step": 38117
-    },
-    {
-      "epoch": 19.879009126466755,
-      "grad_norm": 1.4516133069992065,
-      "learning_rate": 6.220603015075377e-05,
-      "loss": 5.558,
-      "step": 38118
-    },
-    {
-      "epoch": 19.87953063885267,
-      "grad_norm": 1.5652246475219727,
-      "learning_rate": 6.220502512562815e-05,
-      "loss": 4.752,
-      "step": 38119
-    },
-    {
-      "epoch": 19.88005215123859,
-      "grad_norm": 1.4018625020980835,
-      "learning_rate": 6.220402010050251e-05,
-      "loss": 5.4666,
-      "step": 38120
-    },
-    {
-      "epoch": 19.88057366362451,
-      "grad_norm": 1.5595145225524902,
-      "learning_rate": 6.220301507537689e-05,
-      "loss": 5.5208,
-      "step": 38121
-    },
-    {
-      "epoch": 19.88109517601043,
-      "grad_norm": 1.4398109912872314,
-      "learning_rate": 6.220201005025125e-05,
-      "loss": 5.6426,
-      "step": 38122
-    },
-    {
-      "epoch": 19.88161668839635,
-      "grad_norm": 1.6427068710327148,
-      "learning_rate": 6.220100502512563e-05,
-      "loss": 4.7887,
-      "step": 38123
-    },
-    {
-      "epoch": 19.88213820078227,
-      "grad_norm": 1.6221426725387573,
-      "learning_rate": 6.220000000000001e-05,
-      "loss": 4.5102,
-      "step": 38124
-    },
-    {
-      "epoch": 19.882659713168188,
-      "grad_norm": 1.498834490776062,
-      "learning_rate": 6.219899497487437e-05,
-      "loss": 5.4057,
-      "step": 38125
-    },
-    {
-      "epoch": 19.883181225554107,
-      "grad_norm": 1.4918246269226074,
-      "learning_rate": 6.219798994974875e-05,
-      "loss": 5.2287,
-      "step": 38126
-    },
-    {
-      "epoch": 19.883702737940027,
-      "grad_norm": 1.6179622411727905,
-      "learning_rate": 6.219698492462311e-05,
-      "loss": 4.8194,
-      "step": 38127
-    },
-    {
-      "epoch": 19.884224250325946,
-      "grad_norm": 1.6153532266616821,
-      "learning_rate": 6.219597989949749e-05,
-      "loss": 4.994,
-      "step": 38128
-    },
-    {
-      "epoch": 19.884745762711866,
-      "grad_norm": 1.5056079626083374,
-      "learning_rate": 6.219497487437186e-05,
-      "loss": 5.2541,
-      "step": 38129
-    },
-    {
-      "epoch": 19.885267275097785,
-      "grad_norm": 1.5173094272613525,
-      "learning_rate": 6.219396984924623e-05,
-      "loss": 4.8681,
-      "step": 38130
-    },
-    {
-      "epoch": 19.8857887874837,
-      "grad_norm": 1.4664044380187988,
-      "learning_rate": 6.21929648241206e-05,
-      "loss": 5.5091,
-      "step": 38131
-    },
-    {
-      "epoch": 19.88631029986962,
-      "grad_norm": 1.5254746675491333,
-      "learning_rate": 6.219195979899498e-05,
-      "loss": 5.1538,
-      "step": 38132
-    },
-    {
-      "epoch": 19.88683181225554,
-      "grad_norm": 1.5804431438446045,
-      "learning_rate": 6.219095477386935e-05,
-      "loss": 5.2793,
-      "step": 38133
-    },
-    {
-      "epoch": 19.88735332464146,
-      "grad_norm": 1.4151618480682373,
-      "learning_rate": 6.218994974874373e-05,
-      "loss": 5.1958,
-      "step": 38134
-    },
-    {
-      "epoch": 19.88787483702738,
-      "grad_norm": 1.5552607774734497,
-      "learning_rate": 6.21889447236181e-05,
-      "loss": 5.1933,
-      "step": 38135
-    },
-    {
-      "epoch": 19.8883963494133,
-      "grad_norm": 1.4916871786117554,
-      "learning_rate": 6.218793969849247e-05,
-      "loss": 5.5832,
-      "step": 38136
-    },
-    {
-      "epoch": 19.888917861799218,
-      "grad_norm": 1.6262755393981934,
-      "learning_rate": 6.218693467336684e-05,
-      "loss": 4.8665,
-      "step": 38137
-    },
-    {
-      "epoch": 19.889439374185137,
-      "grad_norm": 1.5290266275405884,
-      "learning_rate": 6.21859296482412e-05,
-      "loss": 5.2546,
-      "step": 38138
-    },
-    {
-      "epoch": 19.889960886571057,
-      "grad_norm": 1.4643021821975708,
-      "learning_rate": 6.218492462311558e-05,
-      "loss": 5.6464,
-      "step": 38139
-    },
-    {
-      "epoch": 19.890482398956976,
-      "grad_norm": 1.4137722253799438,
-      "learning_rate": 6.218391959798994e-05,
-      "loss": 4.9865,
-      "step": 38140
-    },
-    {
-      "epoch": 19.891003911342896,
-      "grad_norm": 1.4223109483718872,
-      "learning_rate": 6.218291457286432e-05,
-      "loss": 5.629,
-      "step": 38141
-    },
-    {
-      "epoch": 19.891525423728815,
-      "grad_norm": 1.4793031215667725,
-      "learning_rate": 6.218190954773869e-05,
-      "loss": 5.0585,
-      "step": 38142
-    },
-    {
-      "epoch": 19.89204693611473,
-      "grad_norm": 1.5550274848937988,
-      "learning_rate": 6.218090452261306e-05,
-      "loss": 4.5658,
-      "step": 38143
-    },
-    {
-      "epoch": 19.89256844850065,
-      "grad_norm": 1.5505367517471313,
-      "learning_rate": 6.217989949748744e-05,
-      "loss": 4.9584,
-      "step": 38144
-    },
-    {
-      "epoch": 19.89308996088657,
-      "grad_norm": 1.4890332221984863,
-      "learning_rate": 6.217889447236182e-05,
-      "loss": 5.3106,
-      "step": 38145
-    },
-    {
-      "epoch": 19.89361147327249,
-      "grad_norm": 1.5142959356307983,
-      "learning_rate": 6.217788944723618e-05,
-      "loss": 5.1914,
-      "step": 38146
-    },
-    {
-      "epoch": 19.89413298565841,
-      "grad_norm": 1.4256622791290283,
-      "learning_rate": 6.217688442211056e-05,
-      "loss": 5.5909,
-      "step": 38147
-    },
-    {
-      "epoch": 19.89465449804433,
-      "grad_norm": 1.4868316650390625,
-      "learning_rate": 6.217587939698493e-05,
-      "loss": 5.3283,
-      "step": 38148
-    },
-    {
-      "epoch": 19.895176010430248,
-      "grad_norm": 1.531471848487854,
-      "learning_rate": 6.21748743718593e-05,
-      "loss": 4.2213,
-      "step": 38149
-    },
-    {
-      "epoch": 19.895697522816167,
-      "grad_norm": 1.5173839330673218,
-      "learning_rate": 6.217386934673367e-05,
-      "loss": 5.342,
-      "step": 38150
-    },
-    {
-      "epoch": 19.896219035202087,
-      "grad_norm": 1.498688817024231,
-      "learning_rate": 6.217286432160805e-05,
-      "loss": 5.4315,
-      "step": 38151
-    },
-    {
-      "epoch": 19.896740547588006,
-      "grad_norm": 1.3598445653915405,
-      "learning_rate": 6.217185929648241e-05,
-      "loss": 4.921,
-      "step": 38152
-    },
-    {
-      "epoch": 19.897262059973926,
-      "grad_norm": 1.461531162261963,
-      "learning_rate": 6.217085427135679e-05,
-      "loss": 5.71,
-      "step": 38153
-    },
-    {
-      "epoch": 19.89778357235984,
-      "grad_norm": 1.5162572860717773,
-      "learning_rate": 6.216984924623117e-05,
-      "loss": 5.2894,
-      "step": 38154
-    },
-    {
-      "epoch": 19.89830508474576,
-      "grad_norm": 1.477660059928894,
-      "learning_rate": 6.216884422110553e-05,
-      "loss": 5.3669,
-      "step": 38155
-    },
-    {
-      "epoch": 19.89882659713168,
-      "grad_norm": 1.6160160303115845,
-      "learning_rate": 6.216783919597991e-05,
-      "loss": 4.8417,
-      "step": 38156
-    },
-    {
-      "epoch": 19.8993481095176,
-      "grad_norm": 1.5090303421020508,
-      "learning_rate": 6.216683417085427e-05,
-      "loss": 5.3364,
-      "step": 38157
-    },
-    {
-      "epoch": 19.89986962190352,
-      "grad_norm": 1.4430267810821533,
-      "learning_rate": 6.216582914572865e-05,
-      "loss": 5.3982,
-      "step": 38158
-    },
-    {
-      "epoch": 19.90039113428944,
-      "grad_norm": 1.49335515499115,
-      "learning_rate": 6.216482412060301e-05,
-      "loss": 5.0079,
-      "step": 38159
-    },
-    {
-      "epoch": 19.90091264667536,
-      "grad_norm": 1.4905319213867188,
-      "learning_rate": 6.216381909547739e-05,
-      "loss": 5.2631,
-      "step": 38160
-    },
-    {
-      "epoch": 19.901434159061278,
-      "grad_norm": 1.4842778444290161,
-      "learning_rate": 6.216281407035176e-05,
-      "loss": 5.4243,
-      "step": 38161
-    },
-    {
-      "epoch": 19.901955671447197,
-      "grad_norm": 1.4998399019241333,
-      "learning_rate": 6.216180904522613e-05,
-      "loss": 4.7933,
-      "step": 38162
-    },
-    {
-      "epoch": 19.902477183833117,
-      "grad_norm": 1.4168721437454224,
-      "learning_rate": 6.21608040201005e-05,
-      "loss": 5.4006,
-      "step": 38163
-    },
-    {
-      "epoch": 19.902998696219036,
-      "grad_norm": 1.7023727893829346,
-      "learning_rate": 6.215979899497488e-05,
-      "loss": 5.024,
-      "step": 38164
-    },
-    {
-      "epoch": 19.903520208604956,
-      "grad_norm": 1.5093584060668945,
-      "learning_rate": 6.215879396984925e-05,
-      "loss": 5.4139,
-      "step": 38165
-    },
-    {
-      "epoch": 19.904041720990875,
-      "grad_norm": 1.5375622510910034,
-      "learning_rate": 6.215778894472362e-05,
-      "loss": 5.381,
-      "step": 38166
-    },
-    {
-      "epoch": 19.90456323337679,
-      "grad_norm": 1.5324050188064575,
-      "learning_rate": 6.2156783919598e-05,
-      "loss": 5.2118,
-      "step": 38167
-    },
-    {
-      "epoch": 19.90508474576271,
-      "grad_norm": 1.4609447717666626,
-      "learning_rate": 6.215577889447236e-05,
-      "loss": 4.709,
-      "step": 38168
-    },
-    {
-      "epoch": 19.90560625814863,
-      "grad_norm": 1.4736759662628174,
-      "learning_rate": 6.215477386934674e-05,
-      "loss": 5.4876,
-      "step": 38169
-    },
-    {
-      "epoch": 19.90612777053455,
-      "grad_norm": 1.4363259077072144,
-      "learning_rate": 6.21537688442211e-05,
-      "loss": 4.7938,
-      "step": 38170
-    },
-    {
-      "epoch": 19.90664928292047,
-      "grad_norm": 1.446358323097229,
-      "learning_rate": 6.215276381909548e-05,
-      "loss": 5.587,
-      "step": 38171
-    },
-    {
-      "epoch": 19.90717079530639,
-      "grad_norm": 1.515824794769287,
-      "learning_rate": 6.215175879396985e-05,
-      "loss": 5.3575,
-      "step": 38172
-    },
-    {
-      "epoch": 19.907692307692308,
-      "grad_norm": 1.4200032949447632,
-      "learning_rate": 6.215075376884422e-05,
-      "loss": 4.5062,
-      "step": 38173
-    },
-    {
-      "epoch": 19.908213820078227,
-      "grad_norm": 1.3990845680236816,
-      "learning_rate": 6.21497487437186e-05,
-      "loss": 5.1961,
-      "step": 38174
-    },
-    {
-      "epoch": 19.908735332464147,
-      "grad_norm": 1.483668565750122,
-      "learning_rate": 6.214874371859298e-05,
-      "loss": 4.9355,
-      "step": 38175
-    },
-    {
-      "epoch": 19.909256844850066,
-      "grad_norm": 1.5391247272491455,
-      "learning_rate": 6.214773869346734e-05,
-      "loss": 4.6317,
-      "step": 38176
-    },
-    {
-      "epoch": 19.909778357235986,
-      "grad_norm": 1.472774624824524,
-      "learning_rate": 6.214673366834172e-05,
-      "loss": 5.5288,
-      "step": 38177
-    },
-    {
-      "epoch": 19.910299869621902,
-      "grad_norm": 1.4478139877319336,
-      "learning_rate": 6.214572864321609e-05,
-      "loss": 5.4348,
-      "step": 38178
-    },
-    {
-      "epoch": 19.91082138200782,
-      "grad_norm": 1.4396167993545532,
-      "learning_rate": 6.214472361809045e-05,
-      "loss": 5.2163,
-      "step": 38179
-    },
-    {
-      "epoch": 19.91134289439374,
-      "grad_norm": 1.4811393022537231,
-      "learning_rate": 6.214371859296483e-05,
-      "loss": 5.3467,
-      "step": 38180
-    },
-    {
-      "epoch": 19.91186440677966,
-      "grad_norm": 1.5732979774475098,
-      "learning_rate": 6.214271356783919e-05,
-      "loss": 5.3665,
-      "step": 38181
-    },
-    {
-      "epoch": 19.91238591916558,
-      "grad_norm": 1.4391684532165527,
-      "learning_rate": 6.214170854271357e-05,
-      "loss": 5.9374,
-      "step": 38182
-    },
-    {
-      "epoch": 19.9129074315515,
-      "grad_norm": 1.5206786394119263,
-      "learning_rate": 6.214070351758793e-05,
-      "loss": 5.4421,
-      "step": 38183
-    },
-    {
-      "epoch": 19.91342894393742,
-      "grad_norm": 1.4679421186447144,
-      "learning_rate": 6.213969849246231e-05,
-      "loss": 5.2188,
-      "step": 38184
-    },
-    {
-      "epoch": 19.913950456323338,
-      "grad_norm": 1.562019944190979,
-      "learning_rate": 6.213869346733669e-05,
-      "loss": 5.58,
-      "step": 38185
-    },
-    {
-      "epoch": 19.914471968709258,
-      "grad_norm": 1.4214998483657837,
-      "learning_rate": 6.213768844221107e-05,
-      "loss": 5.4855,
-      "step": 38186
-    },
-    {
-      "epoch": 19.914993481095177,
-      "grad_norm": 1.5634026527404785,
-      "learning_rate": 6.213668341708543e-05,
-      "loss": 5.0633,
-      "step": 38187
-    },
-    {
-      "epoch": 19.915514993481096,
-      "grad_norm": 1.5691609382629395,
-      "learning_rate": 6.213567839195981e-05,
-      "loss": 5.3099,
-      "step": 38188
-    },
-    {
-      "epoch": 19.916036505867016,
-      "grad_norm": 1.5828901529312134,
-      "learning_rate": 6.213467336683417e-05,
-      "loss": 5.3425,
-      "step": 38189
-    },
-    {
-      "epoch": 19.916558018252932,
-      "grad_norm": 1.4627023935317993,
-      "learning_rate": 6.213366834170855e-05,
-      "loss": 5.525,
-      "step": 38190
-    },
-    {
-      "epoch": 19.91707953063885,
-      "grad_norm": 1.436439037322998,
-      "learning_rate": 6.213266331658292e-05,
-      "loss": 5.7044,
-      "step": 38191
-    },
-    {
-      "epoch": 19.91760104302477,
-      "grad_norm": 1.528156042098999,
-      "learning_rate": 6.213165829145728e-05,
-      "loss": 5.4609,
-      "step": 38192
-    },
-    {
-      "epoch": 19.91812255541069,
-      "grad_norm": 1.3592088222503662,
-      "learning_rate": 6.213065326633166e-05,
-      "loss": 5.565,
-      "step": 38193
-    },
-    {
-      "epoch": 19.91864406779661,
-      "grad_norm": 1.4342150688171387,
-      "learning_rate": 6.212964824120604e-05,
-      "loss": 5.4675,
-      "step": 38194
-    },
-    {
-      "epoch": 19.91916558018253,
-      "grad_norm": 1.3914097547531128,
-      "learning_rate": 6.212864321608041e-05,
-      "loss": 5.8754,
-      "step": 38195
-    },
-    {
-      "epoch": 19.91968709256845,
-      "grad_norm": 1.3995081186294556,
-      "learning_rate": 6.212763819095478e-05,
-      "loss": 5.0407,
-      "step": 38196
-    },
-    {
-      "epoch": 19.920208604954368,
-      "grad_norm": 1.4648427963256836,
-      "learning_rate": 6.212663316582916e-05,
-      "loss": 5.3271,
-      "step": 38197
-    },
-    {
-      "epoch": 19.920730117340288,
-      "grad_norm": 1.386121392250061,
-      "learning_rate": 6.212562814070352e-05,
-      "loss": 5.6145,
-      "step": 38198
-    },
-    {
-      "epoch": 19.921251629726207,
-      "grad_norm": 1.4974291324615479,
-      "learning_rate": 6.21246231155779e-05,
-      "loss": 4.7265,
-      "step": 38199
-    },
-    {
-      "epoch": 19.921773142112126,
-      "grad_norm": 1.4993605613708496,
-      "learning_rate": 6.212361809045226e-05,
-      "loss": 5.1517,
-      "step": 38200
-    },
-    {
-      "epoch": 19.922294654498046,
-      "grad_norm": 1.7060974836349487,
-      "learning_rate": 6.212261306532664e-05,
-      "loss": 5.1871,
-      "step": 38201
-    },
-    {
-      "epoch": 19.922816166883962,
-      "grad_norm": 1.5000334978103638,
-      "learning_rate": 6.2121608040201e-05,
-      "loss": 5.4437,
-      "step": 38202
-    },
-    {
-      "epoch": 19.92333767926988,
-      "grad_norm": 1.5411897897720337,
-      "learning_rate": 6.212060301507538e-05,
-      "loss": 5.4723,
-      "step": 38203
-    },
-    {
-      "epoch": 19.9238591916558,
-      "grad_norm": 1.485679030418396,
-      "learning_rate": 6.211959798994975e-05,
-      "loss": 5.0698,
-      "step": 38204
-    },
-    {
-      "epoch": 19.92438070404172,
-      "grad_norm": 1.4133687019348145,
-      "learning_rate": 6.211859296482412e-05,
-      "loss": 4.3637,
-      "step": 38205
-    },
-    {
-      "epoch": 19.92490221642764,
-      "grad_norm": 1.444465160369873,
-      "learning_rate": 6.21175879396985e-05,
-      "loss": 4.9207,
-      "step": 38206
-    },
-    {
-      "epoch": 19.92542372881356,
-      "grad_norm": 1.4572407007217407,
-      "learning_rate": 6.211658291457287e-05,
-      "loss": 5.3377,
-      "step": 38207
-    },
-    {
-      "epoch": 19.92594524119948,
-      "grad_norm": 1.569268822669983,
-      "learning_rate": 6.211557788944724e-05,
-      "loss": 5.0791,
-      "step": 38208
-    },
-    {
-      "epoch": 19.926466753585398,
-      "grad_norm": 1.512372612953186,
-      "learning_rate": 6.211457286432161e-05,
-      "loss": 5.4726,
-      "step": 38209
-    },
-    {
-      "epoch": 19.926988265971318,
-      "grad_norm": 1.5261820554733276,
-      "learning_rate": 6.211356783919599e-05,
-      "loss": 5.1882,
-      "step": 38210
-    },
-    {
-      "epoch": 19.927509778357237,
-      "grad_norm": 1.5607030391693115,
-      "learning_rate": 6.211256281407035e-05,
-      "loss": 5.3144,
-      "step": 38211
-    },
-    {
-      "epoch": 19.928031290743156,
-      "grad_norm": 1.6497431993484497,
-      "learning_rate": 6.211155778894473e-05,
-      "loss": 5.2583,
-      "step": 38212
-    },
-    {
-      "epoch": 19.928552803129076,
-      "grad_norm": 1.500420331954956,
-      "learning_rate": 6.211055276381909e-05,
-      "loss": 5.3685,
-      "step": 38213
-    },
-    {
-      "epoch": 19.929074315514992,
-      "grad_norm": 1.4268224239349365,
-      "learning_rate": 6.210954773869347e-05,
-      "loss": 5.5477,
-      "step": 38214
-    },
-    {
-      "epoch": 19.92959582790091,
-      "grad_norm": 1.5568667650222778,
-      "learning_rate": 6.210854271356785e-05,
-      "loss": 4.8751,
-      "step": 38215
-    },
-    {
-      "epoch": 19.93011734028683,
-      "grad_norm": 1.579789400100708,
-      "learning_rate": 6.210753768844223e-05,
-      "loss": 5.1579,
-      "step": 38216
-    },
-    {
-      "epoch": 19.93063885267275,
-      "grad_norm": 1.480023741722107,
-      "learning_rate": 6.210653266331659e-05,
-      "loss": 5.3738,
-      "step": 38217
-    },
-    {
-      "epoch": 19.93116036505867,
-      "grad_norm": 1.5211929082870483,
-      "learning_rate": 6.210552763819095e-05,
-      "loss": 5.1022,
-      "step": 38218
-    },
-    {
-      "epoch": 19.93168187744459,
-      "grad_norm": 1.4928573369979858,
-      "learning_rate": 6.210452261306533e-05,
-      "loss": 5.267,
-      "step": 38219
-    },
-    {
-      "epoch": 19.93220338983051,
-      "grad_norm": 1.6059174537658691,
-      "learning_rate": 6.21035175879397e-05,
-      "loss": 4.565,
-      "step": 38220
-    },
-    {
-      "epoch": 19.932724902216428,
-      "grad_norm": 1.387036919593811,
-      "learning_rate": 6.210251256281407e-05,
-      "loss": 5.0912,
-      "step": 38221
-    },
-    {
-      "epoch": 19.933246414602348,
-      "grad_norm": 1.4522606134414673,
-      "learning_rate": 6.210150753768844e-05,
-      "loss": 5.5052,
-      "step": 38222
-    },
-    {
-      "epoch": 19.933767926988267,
-      "grad_norm": 1.3652220964431763,
-      "learning_rate": 6.210050251256282e-05,
-      "loss": 5.4326,
-      "step": 38223
-    },
-    {
-      "epoch": 19.934289439374187,
-      "grad_norm": 1.460880994796753,
-      "learning_rate": 6.209949748743718e-05,
-      "loss": 5.6017,
-      "step": 38224
-    },
-    {
-      "epoch": 19.934810951760106,
-      "grad_norm": 1.4801867008209229,
-      "learning_rate": 6.209849246231156e-05,
-      "loss": 5.3201,
-      "step": 38225
-    },
-    {
-      "epoch": 19.935332464146022,
-      "grad_norm": 1.4583388566970825,
-      "learning_rate": 6.209748743718594e-05,
-      "loss": 5.4775,
-      "step": 38226
-    },
-    {
-      "epoch": 19.93585397653194,
-      "grad_norm": 1.499648094177246,
-      "learning_rate": 6.209648241206031e-05,
-      "loss": 5.2127,
-      "step": 38227
-    },
-    {
-      "epoch": 19.93637548891786,
-      "grad_norm": 1.5414307117462158,
-      "learning_rate": 6.209547738693468e-05,
-      "loss": 5.5175,
-      "step": 38228
-    },
-    {
-      "epoch": 19.93689700130378,
-      "grad_norm": 1.5581543445587158,
-      "learning_rate": 6.209447236180906e-05,
-      "loss": 5.4307,
-      "step": 38229
-    },
-    {
-      "epoch": 19.9374185136897,
-      "grad_norm": 1.5137557983398438,
-      "learning_rate": 6.209346733668342e-05,
-      "loss": 5.1771,
-      "step": 38230
-    },
-    {
-      "epoch": 19.93794002607562,
-      "grad_norm": 1.5382746458053589,
-      "learning_rate": 6.209246231155778e-05,
-      "loss": 4.8634,
-      "step": 38231
-    },
-    {
-      "epoch": 19.93846153846154,
-      "grad_norm": 1.453658103942871,
-      "learning_rate": 6.209145728643216e-05,
-      "loss": 5.041,
-      "step": 38232
-    },
-    {
-      "epoch": 19.938983050847458,
-      "grad_norm": 1.5428329706192017,
-      "learning_rate": 6.209045226130653e-05,
-      "loss": 5.0689,
-      "step": 38233
-    },
-    {
-      "epoch": 19.939504563233378,
-      "grad_norm": 1.6713454723358154,
-      "learning_rate": 6.20894472361809e-05,
-      "loss": 4.6827,
-      "step": 38234
-    },
-    {
-      "epoch": 19.940026075619297,
-      "grad_norm": 1.572962760925293,
-      "learning_rate": 6.208844221105527e-05,
-      "loss": 5.4348,
-      "step": 38235
-    },
-    {
-      "epoch": 19.940547588005217,
-      "grad_norm": 1.4381239414215088,
-      "learning_rate": 6.208743718592965e-05,
-      "loss": 5.3034,
-      "step": 38236
-    },
-    {
-      "epoch": 19.941069100391136,
-      "grad_norm": 1.5577903985977173,
-      "learning_rate": 6.208643216080402e-05,
-      "loss": 5.0238,
-      "step": 38237
-    },
-    {
-      "epoch": 19.941590612777052,
-      "grad_norm": 1.5006067752838135,
-      "learning_rate": 6.20854271356784e-05,
-      "loss": 5.3092,
-      "step": 38238
-    },
-    {
-      "epoch": 19.94211212516297,
-      "grad_norm": 1.4910619258880615,
-      "learning_rate": 6.208442211055277e-05,
-      "loss": 5.5342,
-      "step": 38239
-    },
-    {
-      "epoch": 19.94263363754889,
-      "grad_norm": 1.4897197484970093,
-      "learning_rate": 6.208341708542714e-05,
-      "loss": 5.4778,
-      "step": 38240
-    },
-    {
-      "epoch": 19.94315514993481,
-      "grad_norm": 1.578852891921997,
-      "learning_rate": 6.208241206030151e-05,
-      "loss": 5.2282,
-      "step": 38241
-    },
-    {
-      "epoch": 19.94367666232073,
-      "grad_norm": 1.5442230701446533,
-      "learning_rate": 6.208140703517589e-05,
-      "loss": 5.0632,
-      "step": 38242
-    },
-    {
-      "epoch": 19.94419817470665,
-      "grad_norm": 1.5175142288208008,
-      "learning_rate": 6.208040201005025e-05,
-      "loss": 5.895,
-      "step": 38243
-    },
-    {
-      "epoch": 19.94471968709257,
-      "grad_norm": 1.5281156301498413,
-      "learning_rate": 6.207939698492463e-05,
-      "loss": 5.3891,
-      "step": 38244
-    },
-    {
-      "epoch": 19.945241199478488,
-      "grad_norm": 1.616683840751648,
-      "learning_rate": 6.207839195979899e-05,
-      "loss": 5.2726,
-      "step": 38245
-    },
-    {
-      "epoch": 19.945762711864408,
-      "grad_norm": 1.6550337076187134,
-      "learning_rate": 6.207738693467337e-05,
-      "loss": 4.6522,
-      "step": 38246
-    },
-    {
-      "epoch": 19.946284224250327,
-      "grad_norm": 1.5008409023284912,
-      "learning_rate": 6.207638190954775e-05,
-      "loss": 5.3102,
-      "step": 38247
-    },
-    {
-      "epoch": 19.946805736636247,
-      "grad_norm": 1.6091493368148804,
-      "learning_rate": 6.207537688442211e-05,
-      "loss": 5.0733,
-      "step": 38248
-    },
-    {
-      "epoch": 19.947327249022166,
-      "grad_norm": 1.4614239931106567,
-      "learning_rate": 6.207437185929649e-05,
-      "loss": 5.5875,
-      "step": 38249
-    },
-    {
-      "epoch": 19.947848761408082,
-      "grad_norm": 1.460827350616455,
-      "learning_rate": 6.207336683417085e-05,
-      "loss": 5.675,
-      "step": 38250
-    },
-    {
-      "epoch": 19.948370273794,
-      "grad_norm": 1.5288445949554443,
-      "learning_rate": 6.207236180904523e-05,
-      "loss": 5.2832,
-      "step": 38251
-    },
-    {
-      "epoch": 19.94889178617992,
-      "grad_norm": 1.5294309854507446,
-      "learning_rate": 6.20713567839196e-05,
-      "loss": 4.9536,
-      "step": 38252
-    },
-    {
-      "epoch": 19.94941329856584,
-      "grad_norm": 1.5054280757904053,
-      "learning_rate": 6.207035175879397e-05,
-      "loss": 5.4047,
-      "step": 38253
-    },
-    {
-      "epoch": 19.94993481095176,
-      "grad_norm": 1.6656136512756348,
-      "learning_rate": 6.206934673366834e-05,
-      "loss": 5.0735,
-      "step": 38254
-    },
-    {
-      "epoch": 19.95045632333768,
-      "grad_norm": 1.4616235494613647,
-      "learning_rate": 6.206834170854272e-05,
-      "loss": 5.4085,
-      "step": 38255
-    },
-    {
-      "epoch": 19.9509778357236,
-      "grad_norm": 1.5389493703842163,
-      "learning_rate": 6.206733668341708e-05,
-      "loss": 4.961,
-      "step": 38256
-    },
-    {
-      "epoch": 19.951499348109518,
-      "grad_norm": 1.4426450729370117,
-      "learning_rate": 6.206633165829146e-05,
-      "loss": 5.5164,
-      "step": 38257
-    },
-    {
-      "epoch": 19.952020860495438,
-      "grad_norm": 1.469177007675171,
-      "learning_rate": 6.206532663316584e-05,
-      "loss": 5.3368,
-      "step": 38258
-    },
-    {
-      "epoch": 19.952542372881357,
-      "grad_norm": 1.5553861856460571,
-      "learning_rate": 6.20643216080402e-05,
-      "loss": 5.289,
-      "step": 38259
-    },
-    {
-      "epoch": 19.953063885267277,
-      "grad_norm": 1.5059492588043213,
-      "learning_rate": 6.206331658291458e-05,
-      "loss": 5.4239,
-      "step": 38260
-    },
-    {
-      "epoch": 19.953585397653193,
-      "grad_norm": 1.5416524410247803,
-      "learning_rate": 6.206231155778894e-05,
-      "loss": 5.112,
-      "step": 38261
-    },
-    {
-      "epoch": 19.954106910039112,
-      "grad_norm": 1.53389573097229,
-      "learning_rate": 6.206130653266332e-05,
-      "loss": 5.1772,
-      "step": 38262
-    },
-    {
-      "epoch": 19.95462842242503,
-      "grad_norm": 1.5212841033935547,
-      "learning_rate": 6.206030150753769e-05,
-      "loss": 5.3659,
-      "step": 38263
-    },
-    {
-      "epoch": 19.95514993481095,
-      "grad_norm": 1.7363357543945312,
-      "learning_rate": 6.205929648241206e-05,
-      "loss": 4.8177,
-      "step": 38264
-    },
-    {
-      "epoch": 19.95567144719687,
-      "grad_norm": 1.4889999628067017,
-      "learning_rate": 6.205829145728643e-05,
-      "loss": 5.366,
-      "step": 38265
-    },
-    {
-      "epoch": 19.95619295958279,
-      "grad_norm": 1.517215371131897,
-      "learning_rate": 6.20572864321608e-05,
-      "loss": 5.418,
-      "step": 38266
-    },
-    {
-      "epoch": 19.95671447196871,
-      "grad_norm": 1.475551724433899,
-      "learning_rate": 6.205628140703518e-05,
-      "loss": 5.3996,
-      "step": 38267
-    },
-    {
-      "epoch": 19.95723598435463,
-      "grad_norm": 1.5664246082305908,
-      "learning_rate": 6.205527638190956e-05,
-      "loss": 5.4523,
-      "step": 38268
-    },
-    {
-      "epoch": 19.957757496740548,
-      "grad_norm": 1.5209778547286987,
-      "learning_rate": 6.205427135678393e-05,
-      "loss": 5.5006,
-      "step": 38269
-    },
-    {
-      "epoch": 19.958279009126468,
-      "grad_norm": 1.5092730522155762,
-      "learning_rate": 6.20532663316583e-05,
-      "loss": 4.9878,
-      "step": 38270
-    },
-    {
-      "epoch": 19.958800521512387,
-      "grad_norm": 1.461768388748169,
-      "learning_rate": 6.205226130653267e-05,
-      "loss": 5.5595,
-      "step": 38271
-    },
-    {
-      "epoch": 19.959322033898307,
-      "grad_norm": 1.7307533025741577,
-      "learning_rate": 6.205125628140703e-05,
-      "loss": 4.8672,
-      "step": 38272
-    },
-    {
-      "epoch": 19.959843546284226,
-      "grad_norm": 1.5503100156784058,
-      "learning_rate": 6.205025125628141e-05,
-      "loss": 5.2294,
-      "step": 38273
-    },
-    {
-      "epoch": 19.960365058670142,
-      "grad_norm": 1.490294337272644,
-      "learning_rate": 6.204924623115577e-05,
-      "loss": 4.8679,
-      "step": 38274
-    },
-    {
-      "epoch": 19.96088657105606,
-      "grad_norm": 1.4753285646438599,
-      "learning_rate": 6.204824120603015e-05,
-      "loss": 5.3733,
-      "step": 38275
-    },
-    {
-      "epoch": 19.96140808344198,
-      "grad_norm": 1.6775805950164795,
-      "learning_rate": 6.204723618090452e-05,
-      "loss": 5.3552,
-      "step": 38276
-    },
-    {
-      "epoch": 19.9619295958279,
-      "grad_norm": 1.5332751274108887,
-      "learning_rate": 6.20462311557789e-05,
-      "loss": 5.5506,
-      "step": 38277
-    },
-    {
-      "epoch": 19.96245110821382,
-      "grad_norm": 1.5473403930664062,
-      "learning_rate": 6.204522613065327e-05,
-      "loss": 5.1156,
-      "step": 38278
-    },
-    {
-      "epoch": 19.96297262059974,
-      "grad_norm": 1.5635771751403809,
-      "learning_rate": 6.204422110552765e-05,
-      "loss": 4.7657,
-      "step": 38279
-    },
-    {
-      "epoch": 19.96349413298566,
-      "grad_norm": 1.455479383468628,
-      "learning_rate": 6.204321608040201e-05,
-      "loss": 5.2005,
-      "step": 38280
-    },
-    {
-      "epoch": 19.96401564537158,
-      "grad_norm": 1.551373839378357,
-      "learning_rate": 6.204221105527639e-05,
-      "loss": 5.3,
-      "step": 38281
-    },
-    {
-      "epoch": 19.964537157757498,
-      "grad_norm": 1.428611159324646,
-      "learning_rate": 6.204120603015076e-05,
-      "loss": 5.1217,
-      "step": 38282
-    },
-    {
-      "epoch": 19.965058670143417,
-      "grad_norm": 1.4961553812026978,
-      "learning_rate": 6.204020100502513e-05,
-      "loss": 5.3162,
-      "step": 38283
-    },
-    {
-      "epoch": 19.965580182529337,
-      "grad_norm": 1.5838223695755005,
-      "learning_rate": 6.20391959798995e-05,
-      "loss": 5.0899,
-      "step": 38284
-    },
-    {
-      "epoch": 19.966101694915253,
-      "grad_norm": 1.5647261142730713,
-      "learning_rate": 6.203819095477386e-05,
-      "loss": 5.204,
-      "step": 38285
-    },
-    {
-      "epoch": 19.966623207301172,
-      "grad_norm": 1.4564136266708374,
-      "learning_rate": 6.203718592964824e-05,
-      "loss": 5.5877,
-      "step": 38286
-    },
-    {
-      "epoch": 19.96714471968709,
-      "grad_norm": 1.4654467105865479,
-      "learning_rate": 6.203618090452262e-05,
-      "loss": 5.4059,
-      "step": 38287
-    },
-    {
-      "epoch": 19.96766623207301,
-      "grad_norm": 1.4822232723236084,
-      "learning_rate": 6.2035175879397e-05,
-      "loss": 5.4883,
-      "step": 38288
-    },
-    {
-      "epoch": 19.96818774445893,
-      "grad_norm": 1.4983853101730347,
-      "learning_rate": 6.203417085427136e-05,
-      "loss": 5.0726,
-      "step": 38289
-    },
-    {
-      "epoch": 19.96870925684485,
-      "grad_norm": 1.5626475811004639,
-      "learning_rate": 6.203316582914574e-05,
-      "loss": 5.3834,
-      "step": 38290
-    },
-    {
-      "epoch": 19.96923076923077,
-      "grad_norm": 1.4274855852127075,
-      "learning_rate": 6.20321608040201e-05,
-      "loss": 5.6162,
-      "step": 38291
-    },
-    {
-      "epoch": 19.96975228161669,
-      "grad_norm": 1.5643380880355835,
-      "learning_rate": 6.203115577889448e-05,
-      "loss": 5.6857,
-      "step": 38292
-    },
-    {
-      "epoch": 19.97027379400261,
-      "grad_norm": 1.492038607597351,
-      "learning_rate": 6.203015075376884e-05,
-      "loss": 5.3568,
-      "step": 38293
-    },
-    {
-      "epoch": 19.970795306388528,
-      "grad_norm": 1.524474859237671,
-      "learning_rate": 6.202914572864322e-05,
-      "loss": 5.5221,
-      "step": 38294
-    },
-    {
-      "epoch": 19.971316818774447,
-      "grad_norm": 1.5520282983779907,
-      "learning_rate": 6.202814070351759e-05,
-      "loss": 4.8685,
-      "step": 38295
-    },
-    {
-      "epoch": 19.971838331160367,
-      "grad_norm": 1.4594491720199585,
-      "learning_rate": 6.202713567839196e-05,
-      "loss": 5.5657,
-      "step": 38296
-    },
-    {
-      "epoch": 19.972359843546283,
-      "grad_norm": 1.428917646408081,
-      "learning_rate": 6.202613065326633e-05,
-      "loss": 5.7834,
-      "step": 38297
-    },
-    {
-      "epoch": 19.972881355932202,
-      "grad_norm": 1.534371018409729,
-      "learning_rate": 6.20251256281407e-05,
-      "loss": 4.8774,
-      "step": 38298
-    },
-    {
-      "epoch": 19.97340286831812,
-      "grad_norm": 1.5672705173492432,
-      "learning_rate": 6.202412060301508e-05,
-      "loss": 5.3706,
-      "step": 38299
-    },
-    {
-      "epoch": 19.97392438070404,
-      "grad_norm": 1.5536917448043823,
-      "learning_rate": 6.202311557788945e-05,
-      "loss": 4.9969,
-      "step": 38300
-    },
-    {
-      "epoch": 19.97444589308996,
-      "grad_norm": 1.5125210285186768,
-      "learning_rate": 6.202211055276383e-05,
-      "loss": 5.4372,
-      "step": 38301
-    },
-    {
-      "epoch": 19.97496740547588,
-      "grad_norm": 1.4361066818237305,
-      "learning_rate": 6.202110552763819e-05,
-      "loss": 5.0774,
-      "step": 38302
-    },
-    {
-      "epoch": 19.9754889178618,
-      "grad_norm": 1.5917662382125854,
-      "learning_rate": 6.202010050251257e-05,
-      "loss": 4.5265,
-      "step": 38303
-    },
-    {
-      "epoch": 19.97601043024772,
-      "grad_norm": 1.5558242797851562,
-      "learning_rate": 6.201909547738693e-05,
-      "loss": 5.1149,
-      "step": 38304
-    },
-    {
-      "epoch": 19.97653194263364,
-      "grad_norm": 1.4747296571731567,
-      "learning_rate": 6.201809045226131e-05,
-      "loss": 5.5078,
-      "step": 38305
-    },
-    {
-      "epoch": 19.977053455019558,
-      "grad_norm": 1.590537667274475,
-      "learning_rate": 6.201708542713567e-05,
-      "loss": 4.7747,
-      "step": 38306
-    },
-    {
-      "epoch": 19.977574967405477,
-      "grad_norm": 1.449483871459961,
-      "learning_rate": 6.201608040201005e-05,
-      "loss": 5.2204,
-      "step": 38307
-    },
-    {
-      "epoch": 19.978096479791397,
-      "grad_norm": 1.3806482553482056,
-      "learning_rate": 6.201507537688443e-05,
-      "loss": 5.6059,
-      "step": 38308
-    },
-    {
-      "epoch": 19.978617992177313,
-      "grad_norm": 1.481704592704773,
-      "learning_rate": 6.201407035175881e-05,
-      "loss": 5.6264,
-      "step": 38309
-    },
-    {
-      "epoch": 19.979139504563232,
-      "grad_norm": Infinity,
-      "learning_rate": 6.201407035175881e-05,
-      "loss": 5.3032,
-      "step": 38310
-    },
-    {
-      "epoch": 19.97966101694915,
-      "grad_norm": 1.51116943359375,
-      "learning_rate": 6.201306532663317e-05,
-      "loss": 5.2179,
-      "step": 38311
-    },
-    {
-      "epoch": 19.98018252933507,
-      "grad_norm": 1.5021331310272217,
-      "learning_rate": 6.201206030150754e-05,
-      "loss": 5.6067,
-      "step": 38312
-    },
-    {
-      "epoch": 19.98070404172099,
-      "grad_norm": 1.4031672477722168,
-      "learning_rate": 6.201105527638191e-05,
-      "loss": 5.4398,
-      "step": 38313
-    },
-    {
-      "epoch": 19.98122555410691,
-      "grad_norm": 1.4915101528167725,
-      "learning_rate": 6.201005025125628e-05,
-      "loss": 5.519,
-      "step": 38314
-    },
-    {
-      "epoch": 19.98174706649283,
-      "grad_norm": 1.5065876245498657,
-      "learning_rate": 6.200904522613066e-05,
-      "loss": 5.0572,
-      "step": 38315
-    },
-    {
-      "epoch": 19.98226857887875,
-      "grad_norm": 1.4821653366088867,
-      "learning_rate": 6.200804020100502e-05,
-      "loss": 5.5387,
-      "step": 38316
-    },
-    {
-      "epoch": 19.98279009126467,
-      "grad_norm": 1.4739785194396973,
-      "learning_rate": 6.20070351758794e-05,
-      "loss": 5.4162,
-      "step": 38317
-    },
-    {
-      "epoch": 19.983311603650588,
-      "grad_norm": 1.4773660898208618,
-      "learning_rate": 6.200603015075376e-05,
-      "loss": 5.5508,
-      "step": 38318
-    },
-    {
-      "epoch": 19.983833116036507,
-      "grad_norm": 1.6143766641616821,
-      "learning_rate": 6.200502512562814e-05,
-      "loss": 5.2878,
-      "step": 38319
-    },
-    {
-      "epoch": 19.984354628422427,
-      "grad_norm": 1.536381483078003,
-      "learning_rate": 6.200402010050252e-05,
-      "loss": 4.9506,
-      "step": 38320
-    },
-    {
-      "epoch": 19.984876140808343,
-      "grad_norm": 1.4987151622772217,
-      "learning_rate": 6.20030150753769e-05,
-      "loss": 5.298,
-      "step": 38321
-    },
-    {
-      "epoch": 19.985397653194262,
-      "grad_norm": 1.5543229579925537,
-      "learning_rate": 6.200201005025126e-05,
-      "loss": 5.2184,
-      "step": 38322
-    },
-    {
-      "epoch": 19.98591916558018,
-      "grad_norm": 1.497510313987732,
-      "learning_rate": 6.200100502512564e-05,
-      "loss": 5.3506,
-      "step": 38323
-    },
-    {
-      "epoch": 19.9864406779661,
-      "grad_norm": 1.4817461967468262,
-      "learning_rate": 6.2e-05,
-      "loss": 5.1208,
-      "step": 38324
-    },
-    {
-      "epoch": 19.98696219035202,
-      "grad_norm": 1.4718077182769775,
-      "learning_rate": 6.199899497487438e-05,
-      "loss": 5.3013,
-      "step": 38325
-    },
-    {
-      "epoch": 19.98748370273794,
-      "grad_norm": 1.6227890253067017,
-      "learning_rate": 6.199798994974874e-05,
-      "loss": 4.8002,
-      "step": 38326
-    },
-    {
-      "epoch": 19.98800521512386,
-      "grad_norm": 1.4617931842803955,
-      "learning_rate": 6.199698492462311e-05,
-      "loss": 5.5222,
-      "step": 38327
-    },
-    {
-      "epoch": 19.98852672750978,
-      "grad_norm": 1.4979294538497925,
-      "learning_rate": 6.199597989949749e-05,
-      "loss": 5.3152,
-      "step": 38328
-    },
-    {
-      "epoch": 19.9890482398957,
-      "grad_norm": 1.4967265129089355,
-      "learning_rate": 6.199497487437186e-05,
-      "loss": 5.0194,
-      "step": 38329
-    },
-    {
-      "epoch": 19.989569752281618,
-      "grad_norm": 1.577396035194397,
-      "learning_rate": 6.199396984924624e-05,
-      "loss": 5.285,
-      "step": 38330
-    },
-    {
-      "epoch": 19.990091264667537,
-      "grad_norm": 1.4141192436218262,
-      "learning_rate": 6.199296482412061e-05,
-      "loss": 5.4598,
-      "step": 38331
-    },
-    {
-      "epoch": 19.990612777053457,
-      "grad_norm": 1.7098125219345093,
-      "learning_rate": 6.199195979899498e-05,
-      "loss": 4.5923,
-      "step": 38332
-    },
-    {
-      "epoch": 19.991134289439373,
-      "grad_norm": 1.480564832687378,
-      "learning_rate": 6.199095477386935e-05,
-      "loss": 5.5478,
-      "step": 38333
-    },
-    {
-      "epoch": 19.991655801825292,
-      "grad_norm": 1.4652355909347534,
-      "learning_rate": 6.198994974874373e-05,
-      "loss": 5.4401,
-      "step": 38334
-    },
-    {
-      "epoch": 19.99217731421121,
-      "grad_norm": 1.4865045547485352,
-      "learning_rate": 6.198894472361809e-05,
-      "loss": 5.1785,
-      "step": 38335
-    },
-    {
-      "epoch": 19.99269882659713,
-      "grad_norm": 1.6490668058395386,
-      "learning_rate": 6.198793969849247e-05,
-      "loss": 5.1903,
-      "step": 38336
-    },
-    {
-      "epoch": 19.99322033898305,
-      "grad_norm": 1.4895744323730469,
-      "learning_rate": 6.198693467336683e-05,
-      "loss": 5.154,
-      "step": 38337
-    },
-    {
-      "epoch": 19.99374185136897,
-      "grad_norm": 1.494354248046875,
-      "learning_rate": 6.198592964824121e-05,
-      "loss": 5.6342,
-      "step": 38338
-    },
-    {
-      "epoch": 19.99426336375489,
-      "grad_norm": 1.5652570724487305,
-      "learning_rate": 6.198492462311558e-05,
-      "loss": 4.8167,
-      "step": 38339
-    },
-    {
-      "epoch": 19.99478487614081,
-      "grad_norm": 1.5241749286651611,
-      "learning_rate": 6.198391959798995e-05,
-      "loss": 5.0666,
-      "step": 38340
-    },
-    {
-      "epoch": 19.99530638852673,
-      "grad_norm": 1.5899394750595093,
-      "learning_rate": 6.198291457286433e-05,
-      "loss": 5.7087,
-      "step": 38341
-    },
-    {
-      "epoch": 19.995827900912648,
-      "grad_norm": 1.6270229816436768,
-      "learning_rate": 6.19819095477387e-05,
-      "loss": 5.4033,
-      "step": 38342
-    },
-    {
-      "epoch": 19.996349413298567,
-      "grad_norm": 1.4796744585037231,
-      "learning_rate": 6.198090452261307e-05,
-      "loss": 4.8208,
-      "step": 38343
-    },
-    {
-      "epoch": 19.996870925684483,
-      "grad_norm": 1.5884077548980713,
-      "learning_rate": 6.197989949748744e-05,
-      "loss": 4.8114,
-      "step": 38344
-    },
-    {
-      "epoch": 19.997392438070403,
-      "grad_norm": 1.5920491218566895,
-      "learning_rate": 6.197889447236182e-05,
-      "loss": 5.3038,
-      "step": 38345
-    },
-    {
-      "epoch": 19.997913950456322,
-      "grad_norm": 1.4736557006835938,
-      "learning_rate": 6.197788944723618e-05,
-      "loss": 5.5012,
-      "step": 38346
-    },
-    {
-      "epoch": 19.99843546284224,
-      "grad_norm": 1.6492938995361328,
-      "learning_rate": 6.197688442211056e-05,
-      "loss": 4.8642,
-      "step": 38347
-    },
-    {
-      "epoch": 19.99895697522816,
-      "grad_norm": 1.4494704008102417,
-      "learning_rate": 6.197587939698492e-05,
-      "loss": 5.2915,
-      "step": 38348
-    },
-    {
-      "epoch": 19.99947848761408,
-      "grad_norm": 1.5519440174102783,
-      "learning_rate": 6.19748743718593e-05,
-      "loss": 4.7947,
-      "step": 38349
-    },
-    {
-      "epoch": 20.0,
-      "grad_norm": 1.7670592069625854,
-      "learning_rate": 6.197386934673368e-05,
-      "loss": 5.0612,
-      "step": 38350
-    },
-    {
-      "epoch": 20.00052151238592,
-      "grad_norm": 1.4991551637649536,
-      "learning_rate": 6.197286432160805e-05,
-      "loss": 5.0481,
-      "step": 38351
-    },
-    {
-      "epoch": 20.00104302477184,
-      "grad_norm": 1.5075438022613525,
-      "learning_rate": 6.197185929648242e-05,
-      "loss": 5.2745,
-      "step": 38352
-    },
-    {
-      "epoch": 20.00156453715776,
-      "grad_norm": 1.5979329347610474,
-      "learning_rate": 6.197085427135678e-05,
-      "loss": 5.4155,
-      "step": 38353
-    },
-    {
-      "epoch": 20.002086049543678,
-      "grad_norm": 1.5887905359268188,
-      "learning_rate": 6.196984924623116e-05,
-      "loss": 4.9441,
-      "step": 38354
-    },
-    {
-      "epoch": 20.002607561929597,
-      "grad_norm": 1.5453920364379883,
-      "learning_rate": 6.196884422110553e-05,
-      "loss": 5.5281,
-      "step": 38355
-    },
-    {
-      "epoch": 20.003129074315513,
-      "grad_norm": 1.5092674493789673,
-      "learning_rate": 6.19678391959799e-05,
-      "loss": 5.3524,
-      "step": 38356
-    },
-    {
-      "epoch": 20.003650586701433,
-      "grad_norm": 1.5544883012771606,
-      "learning_rate": 6.196683417085427e-05,
-      "loss": 5.4691,
-      "step": 38357
-    },
-    {
-      "epoch": 20.004172099087352,
-      "grad_norm": 1.4410196542739868,
-      "learning_rate": 6.196582914572865e-05,
-      "loss": 4.9703,
-      "step": 38358
-    },
-    {
-      "epoch": 20.00469361147327,
-      "grad_norm": 1.503227949142456,
-      "learning_rate": 6.196482412060301e-05,
-      "loss": 5.4416,
-      "step": 38359
-    },
-    {
-      "epoch": 20.00521512385919,
-      "grad_norm": 1.4966946840286255,
-      "learning_rate": 6.196381909547739e-05,
-      "loss": 4.7147,
-      "step": 38360
-    },
-    {
-      "epoch": 20.00573663624511,
-      "grad_norm": 1.542250156402588,
-      "learning_rate": 6.196281407035177e-05,
-      "loss": 5.1907,
-      "step": 38361
-    },
-    {
-      "epoch": 20.00625814863103,
-      "grad_norm": 1.478912591934204,
-      "learning_rate": 6.196180904522614e-05,
-      "loss": 5.469,
-      "step": 38362
-    },
-    {
-      "epoch": 20.00677966101695,
-      "grad_norm": 1.4546810388565063,
-      "learning_rate": 6.196080402010051e-05,
-      "loss": 4.7283,
-      "step": 38363
-    },
-    {
-      "epoch": 20.00730117340287,
-      "grad_norm": 1.4002141952514648,
-      "learning_rate": 6.195979899497489e-05,
-      "loss": 4.9661,
-      "step": 38364
-    },
-    {
-      "epoch": 20.00782268578879,
-      "grad_norm": 1.4379291534423828,
-      "learning_rate": 6.195879396984925e-05,
-      "loss": 5.4524,
-      "step": 38365
-    },
-    {
-      "epoch": 20.008344198174708,
-      "grad_norm": 1.453396201133728,
-      "learning_rate": 6.195778894472361e-05,
-      "loss": 5.2206,
-      "step": 38366
-    },
-    {
-      "epoch": 20.008865710560627,
-      "grad_norm": 1.5911533832550049,
-      "learning_rate": 6.195678391959799e-05,
-      "loss": 5.0024,
-      "step": 38367
-    },
-    {
-      "epoch": 20.009387222946543,
-      "grad_norm": 1.5506181716918945,
-      "learning_rate": 6.195577889447236e-05,
-      "loss": 4.8984,
-      "step": 38368
-    },
-    {
-      "epoch": 20.009908735332463,
-      "grad_norm": 1.531511902809143,
-      "learning_rate": 6.195477386934673e-05,
-      "loss": 5.112,
-      "step": 38369
-    },
-    {
-      "epoch": 20.010430247718382,
-      "grad_norm": 1.4704431295394897,
-      "learning_rate": 6.195376884422111e-05,
-      "loss": 5.2469,
-      "step": 38370
-    },
-    {
-      "epoch": 20.0109517601043,
-      "grad_norm": 1.482861876487732,
-      "learning_rate": 6.195276381909549e-05,
-      "loss": 5.2703,
-      "step": 38371
-    },
-    {
-      "epoch": 20.01147327249022,
-      "grad_norm": 1.3952692747116089,
-      "learning_rate": 6.195175879396985e-05,
-      "loss": 5.6045,
-      "step": 38372
-    },
-    {
-      "epoch": 20.01199478487614,
-      "grad_norm": 1.463992714881897,
-      "learning_rate": 6.195075376884423e-05,
-      "loss": 5.2716,
-      "step": 38373
-    },
-    {
-      "epoch": 20.01251629726206,
-      "grad_norm": 1.5541954040527344,
-      "learning_rate": 6.19497487437186e-05,
-      "loss": 5.1242,
-      "step": 38374
-    },
-    {
-      "epoch": 20.01303780964798,
-      "grad_norm": 1.5170527696609497,
-      "learning_rate": 6.194874371859297e-05,
-      "loss": 5.2831,
-      "step": 38375
-    },
-    {
-      "epoch": 20.0135593220339,
-      "grad_norm": 1.5074068307876587,
-      "learning_rate": 6.194773869346734e-05,
-      "loss": 4.9258,
-      "step": 38376
-    },
-    {
-      "epoch": 20.01408083441982,
-      "grad_norm": 1.4942498207092285,
-      "learning_rate": 6.194673366834172e-05,
-      "loss": 5.442,
-      "step": 38377
-    },
-    {
-      "epoch": 20.014602346805738,
-      "grad_norm": 1.554732084274292,
-      "learning_rate": 6.194572864321608e-05,
-      "loss": 4.9194,
-      "step": 38378
-    },
-    {
-      "epoch": 20.015123859191657,
-      "grad_norm": 1.6314871311187744,
-      "learning_rate": 6.194472361809044e-05,
-      "loss": 5.437,
-      "step": 38379
-    },
-    {
-      "epoch": 20.015645371577573,
-      "grad_norm": 1.5272865295410156,
-      "learning_rate": 6.194371859296482e-05,
-      "loss": 4.9432,
-      "step": 38380
-    },
-    {
-      "epoch": 20.016166883963493,
-      "grad_norm": 1.544514536857605,
-      "learning_rate": 6.19427135678392e-05,
-      "loss": 5.093,
-      "step": 38381
-    },
-    {
-      "epoch": 20.016688396349412,
-      "grad_norm": 1.5155658721923828,
-      "learning_rate": 6.194170854271358e-05,
-      "loss": 5.0238,
-      "step": 38382
-    },
-    {
-      "epoch": 20.01720990873533,
-      "grad_norm": 1.590888261795044,
-      "learning_rate": 6.194070351758794e-05,
-      "loss": 5.2328,
-      "step": 38383
-    },
-    {
-      "epoch": 20.01773142112125,
-      "grad_norm": 1.5151185989379883,
-      "learning_rate": 6.193969849246232e-05,
-      "loss": 5.07,
-      "step": 38384
-    },
-    {
-      "epoch": 20.01825293350717,
-      "grad_norm": 1.49502694606781,
-      "learning_rate": 6.193869346733668e-05,
-      "loss": 5.1957,
-      "step": 38385
-    },
-    {
-      "epoch": 20.01877444589309,
-      "grad_norm": 1.5676205158233643,
-      "learning_rate": 6.193768844221106e-05,
-      "loss": 5.5284,
-      "step": 38386
-    },
-    {
-      "epoch": 20.01929595827901,
-      "grad_norm": 1.3869047164916992,
-      "learning_rate": 6.193668341708543e-05,
-      "loss": 5.3116,
-      "step": 38387
-    },
-    {
-      "epoch": 20.01981747066493,
-      "grad_norm": 1.587816834449768,
-      "learning_rate": 6.19356783919598e-05,
-      "loss": 5.137,
-      "step": 38388
-    },
-    {
-      "epoch": 20.02033898305085,
-      "grad_norm": 1.4600805044174194,
-      "learning_rate": 6.193467336683417e-05,
-      "loss": 4.8914,
-      "step": 38389
-    },
-    {
-      "epoch": 20.020860495436768,
-      "grad_norm": 1.598531723022461,
-      "learning_rate": 6.193366834170855e-05,
-      "loss": 5.5073,
-      "step": 38390
-    },
-    {
-      "epoch": 20.021382007822687,
-      "grad_norm": 1.4647703170776367,
-      "learning_rate": 6.193266331658291e-05,
-      "loss": 5.0429,
-      "step": 38391
-    },
-    {
-      "epoch": 20.021903520208603,
-      "grad_norm": 1.545299768447876,
-      "learning_rate": 6.193165829145729e-05,
-      "loss": 5.4232,
-      "step": 38392
-    },
-    {
-      "epoch": 20.022425032594523,
-      "grad_norm": 1.6224243640899658,
-      "learning_rate": 6.193065326633167e-05,
-      "loss": 5.0732,
-      "step": 38393
-    },
-    {
-      "epoch": 20.022946544980442,
-      "grad_norm": 1.4504443407058716,
-      "learning_rate": 6.192964824120603e-05,
-      "loss": 5.6731,
-      "step": 38394
-    },
-    {
-      "epoch": 20.02346805736636,
-      "grad_norm": 1.5926960706710815,
-      "learning_rate": 6.192864321608041e-05,
-      "loss": 5.6407,
-      "step": 38395
-    },
-    {
-      "epoch": 20.02398956975228,
-      "grad_norm": 1.5381401777267456,
-      "learning_rate": 6.192763819095477e-05,
-      "loss": 5.4208,
-      "step": 38396
-    },
-    {
-      "epoch": 20.0245110821382,
-      "grad_norm": 1.6528329849243164,
-      "learning_rate": 6.192663316582915e-05,
-      "loss": 4.7361,
-      "step": 38397
-    },
-    {
-      "epoch": 20.02503259452412,
-      "grad_norm": 1.6026922464370728,
-      "learning_rate": 6.192562814070351e-05,
-      "loss": 5.5348,
-      "step": 38398
-    },
-    {
-      "epoch": 20.02555410691004,
-      "grad_norm": 1.551564335823059,
-      "learning_rate": 6.192462311557789e-05,
-      "loss": 5.4295,
-      "step": 38399
-    },
-    {
-      "epoch": 20.02607561929596,
-      "grad_norm": 1.5610331296920776,
-      "learning_rate": 6.192361809045226e-05,
-      "loss": 5.3704,
-      "step": 38400
-    },
-    {
-      "epoch": 20.02659713168188,
-      "grad_norm": 1.5625617504119873,
-      "learning_rate": 6.192261306532663e-05,
-      "loss": 5.058,
-      "step": 38401
-    },
-    {
-      "epoch": 20.027118644067798,
-      "grad_norm": 1.3963245153427124,
-      "learning_rate": 6.192160804020101e-05,
-      "loss": 5.7464,
-      "step": 38402
-    },
-    {
-      "epoch": 20.027640156453717,
-      "grad_norm": 1.5664513111114502,
-      "learning_rate": 6.192060301507539e-05,
-      "loss": 4.778,
-      "step": 38403
-    },
-    {
-      "epoch": 20.028161668839633,
-      "grad_norm": 1.4568419456481934,
-      "learning_rate": 6.191959798994975e-05,
-      "loss": 5.6187,
-      "step": 38404
-    },
-    {
-      "epoch": 20.028683181225553,
-      "grad_norm": 1.5702580213546753,
-      "learning_rate": 6.191859296482412e-05,
-      "loss": 5.4227,
-      "step": 38405
-    },
-    {
-      "epoch": 20.029204693611472,
-      "grad_norm": 1.5674772262573242,
-      "learning_rate": 6.19175879396985e-05,
-      "loss": 5.3551,
-      "step": 38406
-    },
-    {
-      "epoch": 20.02972620599739,
-      "grad_norm": 1.5585013628005981,
-      "learning_rate": 6.191658291457286e-05,
-      "loss": 5.1697,
-      "step": 38407
-    },
-    {
-      "epoch": 20.03024771838331,
-      "grad_norm": 1.642568826675415,
-      "learning_rate": 6.191557788944724e-05,
-      "loss": 5.1243,
-      "step": 38408
-    },
-    {
-      "epoch": 20.03076923076923,
-      "grad_norm": 1.5893622636795044,
-      "learning_rate": 6.19145728643216e-05,
-      "loss": 4.8495,
-      "step": 38409
-    },
-    {
-      "epoch": 20.03129074315515,
-      "grad_norm": 1.5313950777053833,
-      "learning_rate": 6.191356783919598e-05,
-      "loss": 5.0677,
-      "step": 38410
-    },
-    {
-      "epoch": 20.03181225554107,
-      "grad_norm": 1.4648582935333252,
-      "learning_rate": 6.191256281407035e-05,
-      "loss": 5.5975,
-      "step": 38411
-    },
-    {
-      "epoch": 20.03233376792699,
-      "grad_norm": 1.5536527633666992,
-      "learning_rate": 6.191155778894472e-05,
-      "loss": 5.0323,
-      "step": 38412
-    },
-    {
-      "epoch": 20.03285528031291,
-      "grad_norm": 1.557538390159607,
-      "learning_rate": 6.19105527638191e-05,
-      "loss": 5.0763,
-      "step": 38413
-    },
-    {
-      "epoch": 20.033376792698828,
-      "grad_norm": 1.4583901166915894,
-      "learning_rate": 6.190954773869348e-05,
-      "loss": 5.4506,
-      "step": 38414
-    },
-    {
-      "epoch": 20.033898305084747,
-      "grad_norm": 1.5870263576507568,
-      "learning_rate": 6.190854271356784e-05,
-      "loss": 5.0072,
-      "step": 38415
-    },
-    {
-      "epoch": 20.034419817470663,
-      "grad_norm": 1.5255483388900757,
-      "learning_rate": 6.190753768844222e-05,
-      "loss": 5.3761,
-      "step": 38416
-    },
-    {
-      "epoch": 20.034941329856583,
-      "grad_norm": 1.5742522478103638,
-      "learning_rate": 6.190653266331659e-05,
-      "loss": 5.2756,
-      "step": 38417
-    },
-    {
-      "epoch": 20.035462842242502,
-      "grad_norm": 1.5301873683929443,
-      "learning_rate": 6.190552763819096e-05,
-      "loss": 4.8402,
-      "step": 38418
-    },
-    {
-      "epoch": 20.03598435462842,
-      "grad_norm": 1.5005854368209839,
-      "learning_rate": 6.190452261306533e-05,
-      "loss": 5.2236,
-      "step": 38419
-    },
-    {
-      "epoch": 20.03650586701434,
-      "grad_norm": 1.4867439270019531,
-      "learning_rate": 6.190351758793969e-05,
-      "loss": 5.0934,
-      "step": 38420
-    },
-    {
-      "epoch": 20.03702737940026,
-      "grad_norm": 1.5350110530853271,
-      "learning_rate": 6.190251256281407e-05,
-      "loss": 5.5612,
-      "step": 38421
-    },
-    {
-      "epoch": 20.03754889178618,
-      "grad_norm": 1.4691708087921143,
-      "learning_rate": 6.190150753768845e-05,
-      "loss": 5.5288,
-      "step": 38422
-    },
-    {
-      "epoch": 20.0380704041721,
-      "grad_norm": 1.512378215789795,
-      "learning_rate": 6.190050251256282e-05,
-      "loss": 4.7789,
-      "step": 38423
-    },
-    {
-      "epoch": 20.03859191655802,
-      "grad_norm": 1.4904849529266357,
-      "learning_rate": 6.189949748743719e-05,
-      "loss": 5.4314,
-      "step": 38424
-    },
-    {
-      "epoch": 20.03911342894394,
-      "grad_norm": 1.4355167150497437,
-      "learning_rate": 6.189849246231157e-05,
-      "loss": 5.5798,
-      "step": 38425
-    },
-    {
-      "epoch": 20.039634941329858,
-      "grad_norm": 1.4825509786605835,
-      "learning_rate": 6.189748743718593e-05,
-      "loss": 5.4187,
-      "step": 38426
-    },
-    {
-      "epoch": 20.040156453715777,
-      "grad_norm": 1.5370471477508545,
-      "learning_rate": 6.189648241206031e-05,
-      "loss": 5.4004,
-      "step": 38427
-    },
-    {
-      "epoch": 20.040677966101693,
-      "grad_norm": 1.4789812564849854,
-      "learning_rate": 6.189547738693467e-05,
-      "loss": 5.6624,
-      "step": 38428
-    },
-    {
-      "epoch": 20.041199478487613,
-      "grad_norm": 1.4799489974975586,
-      "learning_rate": 6.189447236180905e-05,
-      "loss": 5.2217,
-      "step": 38429
-    },
-    {
-      "epoch": 20.041720990873532,
-      "grad_norm": 1.5122636556625366,
-      "learning_rate": 6.189346733668342e-05,
-      "loss": 4.8582,
-      "step": 38430
-    },
-    {
-      "epoch": 20.042242503259452,
-      "grad_norm": 1.543449878692627,
-      "learning_rate": 6.18924623115578e-05,
-      "loss": 5.2453,
-      "step": 38431
-    },
-    {
-      "epoch": 20.04276401564537,
-      "grad_norm": 1.5097383260726929,
-      "learning_rate": 6.189145728643216e-05,
-      "loss": 5.0603,
-      "step": 38432
-    },
-    {
-      "epoch": 20.04328552803129,
-      "grad_norm": 1.4585131406784058,
-      "learning_rate": 6.189045226130654e-05,
-      "loss": 5.3117,
-      "step": 38433
-    },
-    {
-      "epoch": 20.04380704041721,
-      "grad_norm": 1.5462977886199951,
-      "learning_rate": 6.188944723618091e-05,
-      "loss": 5.0589,
-      "step": 38434
-    },
-    {
-      "epoch": 20.04432855280313,
-      "grad_norm": 1.436532735824585,
-      "learning_rate": 6.188844221105528e-05,
-      "loss": 5.6479,
-      "step": 38435
-    },
-    {
-      "epoch": 20.04485006518905,
-      "grad_norm": 1.4964061975479126,
-      "learning_rate": 6.188743718592966e-05,
-      "loss": 5.0756,
-      "step": 38436
-    },
-    {
-      "epoch": 20.04537157757497,
-      "grad_norm": 1.4857820272445679,
-      "learning_rate": 6.188643216080402e-05,
-      "loss": 5.39,
-      "step": 38437
-    },
-    {
-      "epoch": 20.045893089960888,
-      "grad_norm": 1.4438871145248413,
-      "learning_rate": 6.18854271356784e-05,
-      "loss": 4.9049,
-      "step": 38438
-    },
-    {
-      "epoch": 20.046414602346807,
-      "grad_norm": 1.551041841506958,
-      "learning_rate": 6.188442211055276e-05,
-      "loss": 4.8662,
-      "step": 38439
-    },
-    {
-      "epoch": 20.046936114732723,
-      "grad_norm": 1.5907386541366577,
-      "learning_rate": 6.188341708542714e-05,
-      "loss": 4.9587,
-      "step": 38440
-    },
-    {
-      "epoch": 20.047457627118643,
-      "grad_norm": 1.4488600492477417,
-      "learning_rate": 6.18824120603015e-05,
-      "loss": 5.4533,
-      "step": 38441
-    },
-    {
-      "epoch": 20.047979139504562,
-      "grad_norm": 1.4760756492614746,
-      "learning_rate": 6.188140703517588e-05,
-      "loss": 5.5056,
-      "step": 38442
-    },
-    {
-      "epoch": 20.048500651890482,
-      "grad_norm": 1.4011493921279907,
-      "learning_rate": 6.188040201005026e-05,
-      "loss": 5.743,
-      "step": 38443
-    },
-    {
-      "epoch": 20.0490221642764,
-      "grad_norm": 1.5420926809310913,
-      "learning_rate": 6.187939698492464e-05,
-      "loss": 5.4779,
-      "step": 38444
-    },
-    {
-      "epoch": 20.04954367666232,
-      "grad_norm": 1.4932574033737183,
-      "learning_rate": 6.1878391959799e-05,
-      "loss": 5.3426,
-      "step": 38445
-    },
-    {
-      "epoch": 20.05006518904824,
-      "grad_norm": 1.466773509979248,
-      "learning_rate": 6.187738693467337e-05,
-      "loss": 5.6908,
-      "step": 38446
-    },
-    {
-      "epoch": 20.05058670143416,
-      "grad_norm": 1.4316977262496948,
-      "learning_rate": 6.187638190954774e-05,
-      "loss": 5.0673,
-      "step": 38447
-    },
-    {
-      "epoch": 20.05110821382008,
-      "grad_norm": 1.5069615840911865,
-      "learning_rate": 6.187537688442211e-05,
-      "loss": 5.3042,
-      "step": 38448
-    },
-    {
-      "epoch": 20.051629726206,
-      "grad_norm": 1.4266856908798218,
-      "learning_rate": 6.187437185929649e-05,
-      "loss": 5.8059,
-      "step": 38449
-    },
-    {
-      "epoch": 20.052151238591918,
-      "grad_norm": 1.508659839630127,
-      "learning_rate": 6.187336683417085e-05,
-      "loss": 5.6164,
-      "step": 38450
-    },
-    {
-      "epoch": 20.052672750977834,
-      "grad_norm": 1.5979679822921753,
-      "learning_rate": 6.187236180904523e-05,
-      "loss": 4.9606,
-      "step": 38451
-    },
-    {
-      "epoch": 20.053194263363753,
-      "grad_norm": 1.755431056022644,
-      "learning_rate": 6.187135678391959e-05,
-      "loss": 4.9024,
-      "step": 38452
-    },
-    {
-      "epoch": 20.053715775749673,
-      "grad_norm": 1.6115193367004395,
-      "learning_rate": 6.187035175879397e-05,
-      "loss": 4.673,
-      "step": 38453
-    },
-    {
-      "epoch": 20.054237288135592,
-      "grad_norm": 1.4572842121124268,
-      "learning_rate": 6.186934673366835e-05,
-      "loss": 5.317,
-      "step": 38454
-    },
-    {
-      "epoch": 20.054758800521512,
-      "grad_norm": 1.66477370262146,
-      "learning_rate": 6.186834170854273e-05,
-      "loss": 5.1272,
-      "step": 38455
-    },
-    {
-      "epoch": 20.05528031290743,
-      "grad_norm": 1.5466547012329102,
-      "learning_rate": 6.186733668341709e-05,
-      "loss": 4.9858,
-      "step": 38456
-    },
-    {
-      "epoch": 20.05580182529335,
-      "grad_norm": 1.4053394794464111,
-      "learning_rate": 6.186633165829147e-05,
-      "loss": 5.6473,
-      "step": 38457
-    },
-    {
-      "epoch": 20.05632333767927,
-      "grad_norm": 1.480117917060852,
-      "learning_rate": 6.186532663316583e-05,
-      "loss": 5.2639,
-      "step": 38458
-    },
-    {
-      "epoch": 20.05684485006519,
-      "grad_norm": 1.6367876529693604,
-      "learning_rate": 6.18643216080402e-05,
-      "loss": 4.9262,
-      "step": 38459
-    },
-    {
-      "epoch": 20.05736636245111,
-      "grad_norm": 1.5097612142562866,
-      "learning_rate": 6.186331658291457e-05,
-      "loss": 5.1067,
-      "step": 38460
-    },
-    {
-      "epoch": 20.05788787483703,
-      "grad_norm": 1.4982255697250366,
-      "learning_rate": 6.186231155778894e-05,
-      "loss": 5.7607,
-      "step": 38461
-    },
-    {
-      "epoch": 20.058409387222948,
-      "grad_norm": 1.635248064994812,
-      "learning_rate": 6.186130653266332e-05,
-      "loss": 5.104,
-      "step": 38462
-    },
-    {
-      "epoch": 20.058930899608864,
-      "grad_norm": 1.446808099746704,
-      "learning_rate": 6.18603015075377e-05,
-      "loss": 5.5374,
-      "step": 38463
-    },
-    {
-      "epoch": 20.059452411994783,
-      "grad_norm": 1.4944933652877808,
-      "learning_rate": 6.185929648241207e-05,
-      "loss": 5.4496,
-      "step": 38464
-    },
-    {
-      "epoch": 20.059973924380703,
-      "grad_norm": 1.5801339149475098,
-      "learning_rate": 6.185829145728644e-05,
-      "loss": 5.5124,
-      "step": 38465
-    },
-    {
-      "epoch": 20.060495436766622,
-      "grad_norm": 1.5195292234420776,
-      "learning_rate": 6.185728643216081e-05,
-      "loss": 5.101,
-      "step": 38466
-    },
-    {
-      "epoch": 20.061016949152542,
-      "grad_norm": 1.4888685941696167,
-      "learning_rate": 6.185628140703518e-05,
-      "loss": 5.0348,
-      "step": 38467
-    },
-    {
-      "epoch": 20.06153846153846,
-      "grad_norm": 1.5030338764190674,
-      "learning_rate": 6.185527638190956e-05,
-      "loss": 4.9932,
-      "step": 38468
-    },
-    {
-      "epoch": 20.06205997392438,
-      "grad_norm": 1.5674996376037598,
-      "learning_rate": 6.185427135678392e-05,
-      "loss": 5.2346,
-      "step": 38469
-    },
-    {
-      "epoch": 20.0625814863103,
-      "grad_norm": 1.5799942016601562,
-      "learning_rate": 6.18532663316583e-05,
-      "loss": 5.3575,
-      "step": 38470
-    },
-    {
-      "epoch": 20.06310299869622,
-      "grad_norm": 1.4671993255615234,
-      "learning_rate": 6.185226130653266e-05,
-      "loss": 5.436,
-      "step": 38471
-    },
-    {
-      "epoch": 20.06362451108214,
-      "grad_norm": 1.5939100980758667,
-      "learning_rate": 6.185125628140703e-05,
-      "loss": 5.2234,
-      "step": 38472
-    },
-    {
-      "epoch": 20.06414602346806,
-      "grad_norm": 1.534515380859375,
-      "learning_rate": 6.18502512562814e-05,
-      "loss": 4.6969,
-      "step": 38473
-    },
-    {
-      "epoch": 20.064667535853978,
-      "grad_norm": 1.5435783863067627,
-      "learning_rate": 6.184924623115578e-05,
-      "loss": 5.4023,
-      "step": 38474
-    },
-    {
-      "epoch": 20.065189048239894,
-      "grad_norm": 1.5855234861373901,
-      "learning_rate": 6.184824120603016e-05,
-      "loss": 4.8683,
-      "step": 38475
-    },
-    {
-      "epoch": 20.065710560625813,
-      "grad_norm": 1.4912241697311401,
-      "learning_rate": 6.184723618090452e-05,
-      "loss": 5.3877,
-      "step": 38476
-    },
-    {
-      "epoch": 20.066232073011733,
-      "grad_norm": 1.502462387084961,
-      "learning_rate": 6.18462311557789e-05,
-      "loss": 5.6019,
-      "step": 38477
-    },
-    {
-      "epoch": 20.066753585397652,
-      "grad_norm": 1.551421880722046,
-      "learning_rate": 6.184522613065327e-05,
-      "loss": 5.3343,
-      "step": 38478
-    },
-    {
-      "epoch": 20.067275097783572,
-      "grad_norm": 1.5208722352981567,
-      "learning_rate": 6.184422110552764e-05,
-      "loss": 5.0349,
-      "step": 38479
-    },
-    {
-      "epoch": 20.06779661016949,
-      "grad_norm": 1.4309135675430298,
-      "learning_rate": 6.184321608040201e-05,
-      "loss": 5.3525,
-      "step": 38480
-    },
-    {
-      "epoch": 20.06831812255541,
-      "grad_norm": 1.5060803890228271,
-      "learning_rate": 6.184221105527639e-05,
-      "loss": 5.2212,
-      "step": 38481
-    },
-    {
-      "epoch": 20.06883963494133,
-      "grad_norm": 1.456944465637207,
-      "learning_rate": 6.184120603015075e-05,
-      "loss": 5.3446,
-      "step": 38482
-    },
-    {
-      "epoch": 20.06936114732725,
-      "grad_norm": 1.5476816892623901,
-      "learning_rate": 6.184020100502513e-05,
-      "loss": 5.2039,
-      "step": 38483
-    },
-    {
-      "epoch": 20.06988265971317,
-      "grad_norm": 1.5427982807159424,
-      "learning_rate": 6.18391959798995e-05,
-      "loss": 4.9848,
-      "step": 38484
-    },
-    {
-      "epoch": 20.07040417209909,
-      "grad_norm": 1.5014585256576538,
-      "learning_rate": 6.183819095477387e-05,
-      "loss": 5.3976,
-      "step": 38485
-    },
-    {
-      "epoch": 20.070925684485008,
-      "grad_norm": 1.53799569606781,
-      "learning_rate": 6.183718592964825e-05,
-      "loss": 4.7688,
-      "step": 38486
-    },
-    {
-      "epoch": 20.071447196870924,
-      "grad_norm": 1.5701849460601807,
-      "learning_rate": 6.183618090452261e-05,
-      "loss": 4.6645,
-      "step": 38487
-    },
-    {
-      "epoch": 20.071968709256844,
-      "grad_norm": 1.5447826385498047,
-      "learning_rate": 6.183517587939699e-05,
-      "loss": 5.5599,
-      "step": 38488
-    },
-    {
-      "epoch": 20.072490221642763,
-      "grad_norm": 1.5328484773635864,
-      "learning_rate": 6.183417085427135e-05,
-      "loss": 5.6055,
-      "step": 38489
-    },
-    {
-      "epoch": 20.073011734028682,
-      "grad_norm": 1.597350001335144,
-      "learning_rate": 6.183316582914573e-05,
-      "loss": 5.2684,
-      "step": 38490
-    },
-    {
-      "epoch": 20.073533246414602,
-      "grad_norm": 1.5024501085281372,
-      "learning_rate": 6.18321608040201e-05,
-      "loss": 5.5539,
-      "step": 38491
-    },
-    {
-      "epoch": 20.07405475880052,
-      "grad_norm": 1.5314534902572632,
-      "learning_rate": 6.183115577889447e-05,
-      "loss": 5.2335,
-      "step": 38492
-    },
-    {
-      "epoch": 20.07457627118644,
-      "grad_norm": 1.6881284713745117,
-      "learning_rate": 6.183015075376884e-05,
-      "loss": 4.8615,
-      "step": 38493
-    },
-    {
-      "epoch": 20.07509778357236,
-      "grad_norm": 1.5235024690628052,
-      "learning_rate": 6.182914572864322e-05,
-      "loss": 5.0037,
-      "step": 38494
-    },
-    {
-      "epoch": 20.07561929595828,
-      "grad_norm": 1.5761784315109253,
-      "learning_rate": 6.18281407035176e-05,
-      "loss": 5.1075,
-      "step": 38495
-    },
-    {
-      "epoch": 20.0761408083442,
-      "grad_norm": 1.5596257448196411,
-      "learning_rate": 6.182713567839197e-05,
-      "loss": 4.8236,
-      "step": 38496
-    },
-    {
-      "epoch": 20.07666232073012,
-      "grad_norm": 1.5764896869659424,
-      "learning_rate": 6.182613065326634e-05,
-      "loss": 5.2057,
-      "step": 38497
-    },
-    {
-      "epoch": 20.077183833116038,
-      "grad_norm": 1.5630866289138794,
-      "learning_rate": 6.18251256281407e-05,
-      "loss": 5.222,
-      "step": 38498
-    },
-    {
-      "epoch": 20.077705345501954,
-      "grad_norm": 1.4481397867202759,
-      "learning_rate": 6.182412060301508e-05,
-      "loss": 5.4137,
-      "step": 38499
-    },
-    {
-      "epoch": 20.078226857887874,
-      "grad_norm": 1.4301581382751465,
-      "learning_rate": 6.182311557788944e-05,
-      "loss": 5.1653,
-      "step": 38500
-    },
-    {
-      "epoch": 20.078748370273793,
-      "grad_norm": 1.6191366910934448,
-      "learning_rate": 6.182211055276382e-05,
-      "loss": 4.8714,
-      "step": 38501
-    },
-    {
-      "epoch": 20.079269882659712,
-      "grad_norm": 1.448967695236206,
-      "learning_rate": 6.182110552763819e-05,
-      "loss": 5.3517,
-      "step": 38502
-    },
-    {
-      "epoch": 20.079791395045632,
-      "grad_norm": 1.455005407333374,
-      "learning_rate": 6.182010050251256e-05,
-      "loss": 4.832,
-      "step": 38503
-    },
-    {
-      "epoch": 20.08031290743155,
-      "grad_norm": 1.4309290647506714,
-      "learning_rate": 6.181909547738694e-05,
-      "loss": 5.4649,
-      "step": 38504
-    },
-    {
-      "epoch": 20.08083441981747,
-      "grad_norm": 1.4761850833892822,
-      "learning_rate": 6.181809045226132e-05,
-      "loss": 5.4937,
-      "step": 38505
-    },
-    {
-      "epoch": 20.08135593220339,
-      "grad_norm": 1.533424973487854,
-      "learning_rate": 6.181708542713568e-05,
-      "loss": 5.5969,
-      "step": 38506
-    },
-    {
-      "epoch": 20.08187744458931,
-      "grad_norm": 1.4834061861038208,
-      "learning_rate": 6.181608040201006e-05,
-      "loss": 4.6386,
-      "step": 38507
-    },
-    {
-      "epoch": 20.08239895697523,
-      "grad_norm": 1.5381070375442505,
-      "learning_rate": 6.181507537688443e-05,
-      "loss": 5.0967,
-      "step": 38508
-    },
-    {
-      "epoch": 20.08292046936115,
-      "grad_norm": 1.4019349813461304,
-      "learning_rate": 6.18140703517588e-05,
-      "loss": 4.6498,
-      "step": 38509
-    },
-    {
-      "epoch": 20.083441981747068,
-      "grad_norm": 1.573056697845459,
-      "learning_rate": 6.181306532663317e-05,
-      "loss": 5.3443,
-      "step": 38510
-    },
-    {
-      "epoch": 20.083963494132984,
-      "grad_norm": 1.5414665937423706,
-      "learning_rate": 6.181206030150755e-05,
-      "loss": 5.4254,
-      "step": 38511
-    },
-    {
-      "epoch": 20.084485006518904,
-      "grad_norm": 1.452172875404358,
-      "learning_rate": 6.181105527638191e-05,
-      "loss": 5.659,
-      "step": 38512
-    },
-    {
-      "epoch": 20.085006518904823,
-      "grad_norm": 1.4776780605316162,
-      "learning_rate": 6.181005025125627e-05,
-      "loss": 5.3069,
-      "step": 38513
-    },
-    {
-      "epoch": 20.085528031290742,
-      "grad_norm": 1.4714559316635132,
-      "learning_rate": 6.180904522613065e-05,
-      "loss": 5.4353,
-      "step": 38514
-    },
-    {
-      "epoch": 20.086049543676662,
-      "grad_norm": 1.5292280912399292,
-      "learning_rate": 6.180804020100503e-05,
-      "loss": 5.4732,
-      "step": 38515
-    },
-    {
-      "epoch": 20.08657105606258,
-      "grad_norm": 1.5390563011169434,
-      "learning_rate": 6.180703517587941e-05,
-      "loss": 5.1489,
-      "step": 38516
-    },
-    {
-      "epoch": 20.0870925684485,
-      "grad_norm": 1.5135494470596313,
-      "learning_rate": 6.180603015075377e-05,
-      "loss": 5.3783,
-      "step": 38517
-    },
-    {
-      "epoch": 20.08761408083442,
-      "grad_norm": 1.503393530845642,
-      "learning_rate": 6.180502512562815e-05,
-      "loss": 5.5059,
-      "step": 38518
-    },
-    {
-      "epoch": 20.08813559322034,
-      "grad_norm": 1.566335678100586,
-      "learning_rate": 6.180402010050251e-05,
-      "loss": 5.031,
-      "step": 38519
-    },
-    {
-      "epoch": 20.08865710560626,
-      "grad_norm": 1.5201599597930908,
-      "learning_rate": 6.180301507537689e-05,
-      "loss": 4.9813,
-      "step": 38520
-    },
-    {
-      "epoch": 20.08917861799218,
-      "grad_norm": 1.494736671447754,
-      "learning_rate": 6.180201005025126e-05,
-      "loss": 5.3736,
-      "step": 38521
-    },
-    {
-      "epoch": 20.089700130378098,
-      "grad_norm": 1.4916061162948608,
-      "learning_rate": 6.180100502512563e-05,
-      "loss": 5.2267,
-      "step": 38522
-    },
-    {
-      "epoch": 20.090221642764014,
-      "grad_norm": 1.4387646913528442,
-      "learning_rate": 6.18e-05,
-      "loss": 5.5116,
-      "step": 38523
-    },
-    {
-      "epoch": 20.090743155149934,
-      "grad_norm": 1.4683613777160645,
-      "learning_rate": 6.179899497487438e-05,
-      "loss": 5.0749,
-      "step": 38524
-    },
-    {
-      "epoch": 20.091264667535853,
-      "grad_norm": 1.5496107339859009,
-      "learning_rate": 6.179798994974875e-05,
-      "loss": 4.8635,
-      "step": 38525
-    },
-    {
-      "epoch": 20.091786179921773,
-      "grad_norm": 1.5670212507247925,
-      "learning_rate": 6.179698492462312e-05,
-      "loss": 4.7193,
-      "step": 38526
-    },
-    {
-      "epoch": 20.092307692307692,
-      "grad_norm": 1.6760023832321167,
-      "learning_rate": 6.17959798994975e-05,
-      "loss": 4.7146,
-      "step": 38527
-    },
-    {
-      "epoch": 20.09282920469361,
-      "grad_norm": 1.6453315019607544,
-      "learning_rate": 6.179497487437186e-05,
-      "loss": 5.575,
-      "step": 38528
-    },
-    {
-      "epoch": 20.09335071707953,
-      "grad_norm": 1.6068875789642334,
-      "learning_rate": 6.179396984924624e-05,
-      "loss": 5.6045,
-      "step": 38529
-    },
-    {
-      "epoch": 20.09387222946545,
-      "grad_norm": 1.4400311708450317,
-      "learning_rate": 6.17929648241206e-05,
-      "loss": 5.3338,
-      "step": 38530
-    },
-    {
-      "epoch": 20.09439374185137,
-      "grad_norm": 1.463665246963501,
-      "learning_rate": 6.179195979899498e-05,
-      "loss": 5.4333,
-      "step": 38531
-    },
-    {
-      "epoch": 20.09491525423729,
-      "grad_norm": 1.4831563234329224,
-      "learning_rate": 6.179095477386934e-05,
-      "loss": 5.5331,
-      "step": 38532
-    },
-    {
-      "epoch": 20.09543676662321,
-      "grad_norm": 1.5855357646942139,
-      "learning_rate": 6.178994974874372e-05,
-      "loss": 5.0125,
-      "step": 38533
-    },
-    {
-      "epoch": 20.09595827900913,
-      "grad_norm": 1.495572805404663,
-      "learning_rate": 6.178894472361809e-05,
-      "loss": 5.7021,
-      "step": 38534
-    },
-    {
-      "epoch": 20.096479791395044,
-      "grad_norm": 1.4683012962341309,
-      "learning_rate": 6.178793969849246e-05,
-      "loss": 5.3324,
-      "step": 38535
-    },
-    {
-      "epoch": 20.097001303780964,
-      "grad_norm": 1.7425317764282227,
-      "learning_rate": 6.178693467336684e-05,
-      "loss": 5.3896,
-      "step": 38536
-    },
-    {
-      "epoch": 20.097522816166883,
-      "grad_norm": 1.4906350374221802,
-      "learning_rate": 6.178592964824122e-05,
-      "loss": 5.5473,
-      "step": 38537
-    },
-    {
-      "epoch": 20.098044328552803,
-      "grad_norm": 1.5587809085845947,
-      "learning_rate": 6.178492462311558e-05,
-      "loss": 5.0221,
-      "step": 38538
-    },
-    {
-      "epoch": 20.098565840938722,
-      "grad_norm": 1.655808925628662,
-      "learning_rate": 6.178391959798995e-05,
-      "loss": 5.0678,
-      "step": 38539
-    },
-    {
-      "epoch": 20.09908735332464,
-      "grad_norm": 1.4382820129394531,
-      "learning_rate": 6.178291457286433e-05,
-      "loss": 5.5419,
-      "step": 38540
-    },
-    {
-      "epoch": 20.09960886571056,
-      "grad_norm": 1.5981334447860718,
-      "learning_rate": 6.178190954773869e-05,
-      "loss": 5.009,
-      "step": 38541
-    },
-    {
-      "epoch": 20.10013037809648,
-      "grad_norm": 1.475565791130066,
-      "learning_rate": 6.178090452261307e-05,
-      "loss": 4.9232,
-      "step": 38542
-    },
-    {
-      "epoch": 20.1006518904824,
-      "grad_norm": 1.655060052871704,
-      "learning_rate": 6.177989949748743e-05,
-      "loss": 5.2709,
-      "step": 38543
-    },
-    {
-      "epoch": 20.10117340286832,
-      "grad_norm": 1.405463457107544,
-      "learning_rate": 6.177889447236181e-05,
-      "loss": 5.574,
-      "step": 38544
-    },
-    {
-      "epoch": 20.10169491525424,
-      "grad_norm": 1.4235963821411133,
-      "learning_rate": 6.177788944723619e-05,
-      "loss": 5.5684,
-      "step": 38545
-    },
-    {
-      "epoch": 20.102216427640155,
-      "grad_norm": 1.578945279121399,
-      "learning_rate": 6.177688442211057e-05,
-      "loss": 5.2131,
-      "step": 38546
-    },
-    {
-      "epoch": 20.102737940026074,
-      "grad_norm": 1.4678783416748047,
-      "learning_rate": 6.177587939698493e-05,
-      "loss": 5.6409,
-      "step": 38547
-    },
-    {
-      "epoch": 20.103259452411994,
-      "grad_norm": 1.5249897241592407,
-      "learning_rate": 6.177487437185931e-05,
-      "loss": 5.2485,
-      "step": 38548
-    },
-    {
-      "epoch": 20.103780964797913,
-      "grad_norm": 1.3929344415664673,
-      "learning_rate": 6.177386934673367e-05,
-      "loss": 5.7874,
-      "step": 38549
-    },
-    {
-      "epoch": 20.104302477183833,
-      "grad_norm": 1.5698636770248413,
-      "learning_rate": 6.177286432160805e-05,
-      "loss": 5.1236,
-      "step": 38550
-    },
-    {
-      "epoch": 20.104823989569752,
-      "grad_norm": 1.5356385707855225,
-      "learning_rate": 6.177185929648241e-05,
-      "loss": 5.1866,
-      "step": 38551
-    },
-    {
-      "epoch": 20.10534550195567,
-      "grad_norm": 1.5756529569625854,
-      "learning_rate": 6.177085427135678e-05,
-      "loss": 5.0823,
-      "step": 38552
-    },
-    {
-      "epoch": 20.10586701434159,
-      "grad_norm": 1.631987452507019,
-      "learning_rate": 6.176984924623116e-05,
-      "loss": 5.015,
-      "step": 38553
-    },
-    {
-      "epoch": 20.10638852672751,
-      "grad_norm": 1.4420053958892822,
-      "learning_rate": 6.176884422110552e-05,
-      "loss": 5.1761,
-      "step": 38554
-    },
-    {
-      "epoch": 20.10691003911343,
-      "grad_norm": 1.5529810190200806,
-      "learning_rate": 6.17678391959799e-05,
-      "loss": 5.1748,
-      "step": 38555
-    },
-    {
-      "epoch": 20.10743155149935,
-      "grad_norm": 1.5226433277130127,
-      "learning_rate": 6.176683417085428e-05,
-      "loss": 5.5658,
-      "step": 38556
-    },
-    {
-      "epoch": 20.10795306388527,
-      "grad_norm": 1.5063096284866333,
-      "learning_rate": 6.176582914572865e-05,
-      "loss": 5.3938,
-      "step": 38557
-    },
-    {
-      "epoch": 20.108474576271185,
-      "grad_norm": 1.465323805809021,
-      "learning_rate": 6.176482412060302e-05,
-      "loss": 5.4899,
-      "step": 38558
-    },
-    {
-      "epoch": 20.108996088657104,
-      "grad_norm": 1.571457862854004,
-      "learning_rate": 6.17638190954774e-05,
-      "loss": 5.0036,
-      "step": 38559
-    },
-    {
-      "epoch": 20.109517601043024,
-      "grad_norm": 1.561355710029602,
-      "learning_rate": 6.176281407035176e-05,
-      "loss": 5.291,
-      "step": 38560
-    },
-    {
-      "epoch": 20.110039113428943,
-      "grad_norm": 1.5571049451828003,
-      "learning_rate": 6.176180904522614e-05,
-      "loss": 5.8926,
-      "step": 38561
-    },
-    {
-      "epoch": 20.110560625814863,
-      "grad_norm": 1.5960458517074585,
-      "learning_rate": 6.17608040201005e-05,
-      "loss": 5.2228,
-      "step": 38562
-    },
-    {
-      "epoch": 20.111082138200782,
-      "grad_norm": 1.4739036560058594,
-      "learning_rate": 6.175979899497488e-05,
-      "loss": 5.2315,
-      "step": 38563
-    },
-    {
-      "epoch": 20.1116036505867,
-      "grad_norm": 1.560827612876892,
-      "learning_rate": 6.175879396984924e-05,
-      "loss": 4.9656,
-      "step": 38564
-    },
-    {
-      "epoch": 20.11212516297262,
-      "grad_norm": 1.4980367422103882,
-      "learning_rate": 6.175778894472361e-05,
-      "loss": 5.0656,
-      "step": 38565
-    },
-    {
-      "epoch": 20.11264667535854,
-      "grad_norm": 1.438679575920105,
-      "learning_rate": 6.175678391959799e-05,
-      "loss": 4.7451,
-      "step": 38566
-    },
-    {
-      "epoch": 20.11316818774446,
-      "grad_norm": 1.404511570930481,
-      "learning_rate": 6.175577889447236e-05,
-      "loss": 5.7304,
-      "step": 38567
-    },
-    {
-      "epoch": 20.11368970013038,
-      "grad_norm": 1.4491422176361084,
-      "learning_rate": 6.175477386934674e-05,
-      "loss": 5.3432,
-      "step": 38568
-    },
-    {
-      "epoch": 20.1142112125163,
-      "grad_norm": 1.558864951133728,
-      "learning_rate": 6.175376884422111e-05,
-      "loss": 5.3514,
-      "step": 38569
-    },
-    {
-      "epoch": 20.114732724902215,
-      "grad_norm": 1.4954278469085693,
-      "learning_rate": 6.175276381909548e-05,
-      "loss": 5.2829,
-      "step": 38570
-    },
-    {
-      "epoch": 20.115254237288134,
-      "grad_norm": 1.5221354961395264,
-      "learning_rate": 6.175175879396985e-05,
-      "loss": 5.2184,
-      "step": 38571
-    },
-    {
-      "epoch": 20.115775749674054,
-      "grad_norm": 1.4322227239608765,
-      "learning_rate": 6.175075376884423e-05,
-      "loss": 5.1887,
-      "step": 38572
-    },
-    {
-      "epoch": 20.116297262059973,
-      "grad_norm": 1.552506923675537,
-      "learning_rate": 6.174974874371859e-05,
-      "loss": 5.3747,
-      "step": 38573
-    },
-    {
-      "epoch": 20.116818774445893,
-      "grad_norm": 1.6186401844024658,
-      "learning_rate": 6.174874371859297e-05,
-      "loss": 4.3894,
-      "step": 38574
-    },
-    {
-      "epoch": 20.117340286831812,
-      "grad_norm": 1.479161024093628,
-      "learning_rate": 6.174773869346733e-05,
-      "loss": 5.4412,
-      "step": 38575
-    },
-    {
-      "epoch": 20.11786179921773,
-      "grad_norm": 1.5059820413589478,
-      "learning_rate": 6.174673366834171e-05,
-      "loss": 5.3221,
-      "step": 38576
-    },
-    {
-      "epoch": 20.11838331160365,
-      "grad_norm": 1.537973165512085,
-      "learning_rate": 6.174572864321609e-05,
-      "loss": 5.3658,
-      "step": 38577
-    },
-    {
-      "epoch": 20.11890482398957,
-      "grad_norm": 1.4992201328277588,
-      "learning_rate": 6.174472361809045e-05,
-      "loss": 5.2155,
-      "step": 38578
-    },
-    {
-      "epoch": 20.11942633637549,
-      "grad_norm": 1.398603916168213,
-      "learning_rate": 6.174371859296483e-05,
-      "loss": 5.5064,
-      "step": 38579
-    },
-    {
-      "epoch": 20.11994784876141,
-      "grad_norm": 1.5721955299377441,
-      "learning_rate": 6.17427135678392e-05,
-      "loss": 4.9478,
-      "step": 38580
-    },
-    {
-      "epoch": 20.12046936114733,
-      "grad_norm": 1.50698983669281,
-      "learning_rate": 6.174170854271357e-05,
-      "loss": 4.8125,
-      "step": 38581
-    },
-    {
-      "epoch": 20.120990873533245,
-      "grad_norm": 1.4552392959594727,
-      "learning_rate": 6.174070351758794e-05,
-      "loss": 5.0577,
-      "step": 38582
-    },
-    {
-      "epoch": 20.121512385919164,
-      "grad_norm": 1.3634613752365112,
-      "learning_rate": 6.173969849246232e-05,
-      "loss": 5.8709,
-      "step": 38583
-    },
-    {
-      "epoch": 20.122033898305084,
-      "grad_norm": 1.615193247795105,
-      "learning_rate": 6.173869346733668e-05,
-      "loss": 5.0255,
-      "step": 38584
-    },
-    {
-      "epoch": 20.122555410691003,
-      "grad_norm": 1.461206316947937,
-      "learning_rate": 6.173768844221106e-05,
-      "loss": 5.0813,
-      "step": 38585
-    },
-    {
-      "epoch": 20.123076923076923,
-      "grad_norm": 1.9166373014450073,
-      "learning_rate": 6.173668341708542e-05,
-      "loss": 4.8397,
-      "step": 38586
-    },
-    {
-      "epoch": 20.123598435462842,
-      "grad_norm": 1.6170282363891602,
-      "learning_rate": 6.17356783919598e-05,
-      "loss": 5.0112,
-      "step": 38587
-    },
-    {
-      "epoch": 20.12411994784876,
-      "grad_norm": 1.5713094472885132,
-      "learning_rate": 6.173467336683418e-05,
-      "loss": 4.8932,
-      "step": 38588
-    },
-    {
-      "epoch": 20.12464146023468,
-      "grad_norm": 1.4417084455490112,
-      "learning_rate": 6.173366834170856e-05,
-      "loss": 5.453,
-      "step": 38589
-    },
-    {
-      "epoch": 20.1251629726206,
-      "grad_norm": 1.4695556163787842,
-      "learning_rate": 6.173266331658292e-05,
-      "loss": 5.3211,
-      "step": 38590
-    },
-    {
-      "epoch": 20.12568448500652,
-      "grad_norm": 1.5307310819625854,
-      "learning_rate": 6.173165829145728e-05,
-      "loss": 5.5081,
-      "step": 38591
-    },
-    {
-      "epoch": 20.12620599739244,
-      "grad_norm": 1.5614421367645264,
-      "learning_rate": 6.173065326633166e-05,
-      "loss": 4.6494,
-      "step": 38592
-    },
-    {
-      "epoch": 20.12672750977836,
-      "grad_norm": 1.4611682891845703,
-      "learning_rate": 6.172964824120603e-05,
-      "loss": 5.3175,
-      "step": 38593
-    },
-    {
-      "epoch": 20.127249022164275,
-      "grad_norm": 1.6017662286758423,
-      "learning_rate": 6.17286432160804e-05,
-      "loss": 4.9029,
-      "step": 38594
-    },
-    {
-      "epoch": 20.127770534550194,
-      "grad_norm": 1.4874608516693115,
-      "learning_rate": 6.172763819095477e-05,
-      "loss": 5.4427,
-      "step": 38595
-    },
-    {
-      "epoch": 20.128292046936114,
-      "grad_norm": 1.5627154111862183,
-      "learning_rate": 6.172663316582915e-05,
-      "loss": 5.4239,
-      "step": 38596
-    },
-    {
-      "epoch": 20.128813559322033,
-      "grad_norm": 1.5337722301483154,
-      "learning_rate": 6.172562814070352e-05,
-      "loss": 5.0623,
-      "step": 38597
-    },
-    {
-      "epoch": 20.129335071707953,
-      "grad_norm": 1.557196855545044,
-      "learning_rate": 6.17246231155779e-05,
-      "loss": 5.3106,
-      "step": 38598
-    },
-    {
-      "epoch": 20.129856584093872,
-      "grad_norm": 1.57356858253479,
-      "learning_rate": 6.172361809045227e-05,
-      "loss": 5.0455,
-      "step": 38599
-    },
-    {
-      "epoch": 20.13037809647979,
-      "grad_norm": 1.4545104503631592,
-      "learning_rate": 6.172261306532664e-05,
-      "loss": 5.4049,
-      "step": 38600
-    },
-    {
-      "epoch": 20.13089960886571,
-      "grad_norm": 1.4383140802383423,
-      "learning_rate": 6.172160804020101e-05,
-      "loss": 5.6317,
-      "step": 38601
-    },
-    {
-      "epoch": 20.13142112125163,
-      "grad_norm": 1.4411762952804565,
-      "learning_rate": 6.172060301507539e-05,
-      "loss": 5.6866,
-      "step": 38602
-    },
-    {
-      "epoch": 20.13194263363755,
-      "grad_norm": 1.4621316194534302,
-      "learning_rate": 6.171959798994975e-05,
-      "loss": 5.7408,
-      "step": 38603
-    },
-    {
-      "epoch": 20.13246414602347,
-      "grad_norm": 1.5645672082901,
-      "learning_rate": 6.171859296482413e-05,
-      "loss": 5.1359,
-      "step": 38604
-    },
-    {
-      "epoch": 20.13298565840939,
-      "grad_norm": 1.5689210891723633,
-      "learning_rate": 6.171758793969849e-05,
-      "loss": 5.2809,
-      "step": 38605
-    },
-    {
-      "epoch": 20.133507170795305,
-      "grad_norm": 1.6742188930511475,
-      "learning_rate": 6.171658291457286e-05,
-      "loss": 5.2945,
-      "step": 38606
-    },
-    {
-      "epoch": 20.134028683181224,
-      "grad_norm": 1.4837597608566284,
-      "learning_rate": 6.171557788944723e-05,
-      "loss": 5.6012,
-      "step": 38607
-    },
-    {
-      "epoch": 20.134550195567144,
-      "grad_norm": 1.5046340227127075,
-      "learning_rate": 6.171457286432161e-05,
-      "loss": 5.6144,
-      "step": 38608
-    },
-    {
-      "epoch": 20.135071707953063,
-      "grad_norm": 1.644165277481079,
-      "learning_rate": 6.171356783919599e-05,
-      "loss": 5.2163,
-      "step": 38609
-    },
-    {
-      "epoch": 20.135593220338983,
-      "grad_norm": 1.4691990613937378,
-      "learning_rate": 6.171256281407035e-05,
-      "loss": 5.1466,
-      "step": 38610
-    },
-    {
-      "epoch": 20.136114732724902,
-      "grad_norm": 1.4932986497879028,
-      "learning_rate": 6.171155778894473e-05,
-      "loss": 5.7095,
-      "step": 38611
-    },
-    {
-      "epoch": 20.13663624511082,
-      "grad_norm": 1.5249931812286377,
-      "learning_rate": 6.17105527638191e-05,
-      "loss": 5.201,
-      "step": 38612
-    },
-    {
-      "epoch": 20.13715775749674,
-      "grad_norm": 1.4290790557861328,
-      "learning_rate": 6.170954773869347e-05,
-      "loss": 5.3403,
-      "step": 38613
-    },
-    {
-      "epoch": 20.13767926988266,
-      "grad_norm": 1.5265834331512451,
-      "learning_rate": 6.170854271356784e-05,
-      "loss": 5.1564,
-      "step": 38614
-    },
-    {
-      "epoch": 20.13820078226858,
-      "grad_norm": 1.5553442239761353,
-      "learning_rate": 6.170753768844222e-05,
-      "loss": 5.1308,
-      "step": 38615
-    },
-    {
-      "epoch": 20.1387222946545,
-      "grad_norm": 1.556208848953247,
-      "learning_rate": 6.170653266331658e-05,
-      "loss": 5.2464,
-      "step": 38616
-    },
-    {
-      "epoch": 20.13924380704042,
-      "grad_norm": 1.4675995111465454,
-      "learning_rate": 6.170552763819096e-05,
-      "loss": 5.3501,
-      "step": 38617
-    },
-    {
-      "epoch": 20.139765319426335,
-      "grad_norm": 1.556364893913269,
-      "learning_rate": 6.170452261306534e-05,
-      "loss": 5.6397,
-      "step": 38618
-    },
-    {
-      "epoch": 20.140286831812254,
-      "grad_norm": 1.7445341348648071,
-      "learning_rate": 6.17035175879397e-05,
-      "loss": 4.4371,
-      "step": 38619
-    },
-    {
-      "epoch": 20.140808344198174,
-      "grad_norm": 1.6007767915725708,
-      "learning_rate": 6.170251256281408e-05,
-      "loss": 5.3573,
-      "step": 38620
-    },
-    {
-      "epoch": 20.141329856584093,
-      "grad_norm": 1.6099179983139038,
-      "learning_rate": 6.170150753768844e-05,
-      "loss": 5.2296,
-      "step": 38621
-    },
-    {
-      "epoch": 20.141851368970013,
-      "grad_norm": 1.5737632513046265,
-      "learning_rate": 6.170050251256282e-05,
-      "loss": 5.4101,
-      "step": 38622
-    },
-    {
-      "epoch": 20.142372881355932,
-      "grad_norm": 1.5527278184890747,
-      "learning_rate": 6.169949748743718e-05,
-      "loss": 5.1139,
-      "step": 38623
-    },
-    {
-      "epoch": 20.14289439374185,
-      "grad_norm": 1.5857445001602173,
-      "learning_rate": 6.169849246231156e-05,
-      "loss": 5.3968,
-      "step": 38624
-    },
-    {
-      "epoch": 20.14341590612777,
-      "grad_norm": 1.484298825263977,
-      "learning_rate": 6.169748743718593e-05,
-      "loss": 5.2113,
-      "step": 38625
-    },
-    {
-      "epoch": 20.14393741851369,
-      "grad_norm": 1.5412592887878418,
-      "learning_rate": 6.16964824120603e-05,
-      "loss": 5.2804,
-      "step": 38626
-    },
-    {
-      "epoch": 20.14445893089961,
-      "grad_norm": 1.4974403381347656,
-      "learning_rate": 6.169547738693467e-05,
-      "loss": 5.375,
-      "step": 38627
-    },
-    {
-      "epoch": 20.14498044328553,
-      "grad_norm": 1.6157993078231812,
-      "learning_rate": 6.169447236180905e-05,
-      "loss": 5.2016,
-      "step": 38628
-    },
-    {
-      "epoch": 20.14550195567145,
-      "grad_norm": 1.571169137954712,
-      "learning_rate": 6.169346733668342e-05,
-      "loss": 4.4208,
-      "step": 38629
-    },
-    {
-      "epoch": 20.146023468057365,
-      "grad_norm": 1.428978443145752,
-      "learning_rate": 6.16924623115578e-05,
-      "loss": 5.568,
-      "step": 38630
-    },
-    {
-      "epoch": 20.146544980443284,
-      "grad_norm": 1.478332281112671,
-      "learning_rate": 6.169145728643217e-05,
-      "loss": 5.4636,
-      "step": 38631
-    },
-    {
-      "epoch": 20.147066492829204,
-      "grad_norm": 1.4465315341949463,
-      "learning_rate": 6.169045226130653e-05,
-      "loss": 5.2345,
-      "step": 38632
-    },
-    {
-      "epoch": 20.147588005215123,
-      "grad_norm": 1.4252408742904663,
-      "learning_rate": 6.168944723618091e-05,
-      "loss": 5.5172,
-      "step": 38633
-    },
-    {
-      "epoch": 20.148109517601043,
-      "grad_norm": 1.4164742231369019,
-      "learning_rate": 6.168844221105527e-05,
-      "loss": 5.2634,
-      "step": 38634
-    },
-    {
-      "epoch": 20.148631029986962,
-      "grad_norm": 1.615415096282959,
-      "learning_rate": 6.168743718592965e-05,
-      "loss": 5.5901,
-      "step": 38635
-    },
-    {
-      "epoch": 20.14915254237288,
-      "grad_norm": 1.532148003578186,
-      "learning_rate": 6.168643216080401e-05,
-      "loss": 5.1214,
-      "step": 38636
-    },
-    {
-      "epoch": 20.1496740547588,
-      "grad_norm": 1.5574300289154053,
-      "learning_rate": 6.168542713567839e-05,
-      "loss": 5.5176,
-      "step": 38637
-    },
-    {
-      "epoch": 20.15019556714472,
-      "grad_norm": 1.6399857997894287,
-      "learning_rate": 6.168442211055277e-05,
-      "loss": 5.4529,
-      "step": 38638
-    },
-    {
-      "epoch": 20.15071707953064,
-      "grad_norm": 1.5439329147338867,
-      "learning_rate": 6.168341708542715e-05,
-      "loss": 5.3374,
-      "step": 38639
-    },
-    {
-      "epoch": 20.15123859191656,
-      "grad_norm": 1.5264943838119507,
-      "learning_rate": 6.168241206030151e-05,
-      "loss": 5.5258,
-      "step": 38640
-    },
-    {
-      "epoch": 20.151760104302475,
-      "grad_norm": 1.5109431743621826,
-      "learning_rate": 6.168140703517589e-05,
-      "loss": 5.3282,
-      "step": 38641
-    },
-    {
-      "epoch": 20.152281616688395,
-      "grad_norm": 1.5024092197418213,
-      "learning_rate": 6.168040201005025e-05,
-      "loss": 5.4645,
-      "step": 38642
-    },
-    {
-      "epoch": 20.152803129074314,
-      "grad_norm": 1.5650906562805176,
-      "learning_rate": 6.167939698492463e-05,
-      "loss": 5.3073,
-      "step": 38643
-    },
-    {
-      "epoch": 20.153324641460234,
-      "grad_norm": 1.5399954319000244,
-      "learning_rate": 6.1678391959799e-05,
-      "loss": 4.9895,
-      "step": 38644
-    },
-    {
-      "epoch": 20.153846153846153,
-      "grad_norm": 1.5510436296463013,
-      "learning_rate": 6.167738693467336e-05,
-      "loss": 5.3852,
-      "step": 38645
-    },
-    {
-      "epoch": 20.154367666232073,
-      "grad_norm": 1.566744327545166,
-      "learning_rate": 6.167638190954774e-05,
-      "loss": 5.175,
-      "step": 38646
-    },
-    {
-      "epoch": 20.154889178617992,
-      "grad_norm": 1.5004661083221436,
-      "learning_rate": 6.16753768844221e-05,
-      "loss": 5.266,
-      "step": 38647
-    },
-    {
-      "epoch": 20.15541069100391,
-      "grad_norm": 1.5350818634033203,
-      "learning_rate": 6.167437185929648e-05,
-      "loss": 5.1929,
-      "step": 38648
-    },
-    {
-      "epoch": 20.15593220338983,
-      "grad_norm": 1.539420247077942,
-      "learning_rate": 6.167336683417086e-05,
-      "loss": 5.2232,
-      "step": 38649
-    },
-    {
-      "epoch": 20.15645371577575,
-      "grad_norm": 1.6019084453582764,
-      "learning_rate": 6.167236180904524e-05,
-      "loss": 5.2918,
-      "step": 38650
-    },
-    {
-      "epoch": 20.15697522816167,
-      "grad_norm": 1.476466417312622,
-      "learning_rate": 6.16713567839196e-05,
-      "loss": 5.2299,
-      "step": 38651
-    },
-    {
-      "epoch": 20.15749674054759,
-      "grad_norm": 1.5431426763534546,
-      "learning_rate": 6.167035175879398e-05,
-      "loss": 5.1314,
-      "step": 38652
-    },
-    {
-      "epoch": 20.158018252933505,
-      "grad_norm": 1.5486465692520142,
-      "learning_rate": 6.166934673366834e-05,
-      "loss": 4.7199,
-      "step": 38653
-    },
-    {
-      "epoch": 20.158539765319425,
-      "grad_norm": 1.5984957218170166,
-      "learning_rate": 6.166834170854272e-05,
-      "loss": 5.3981,
-      "step": 38654
-    },
-    {
-      "epoch": 20.159061277705344,
-      "grad_norm": 1.5200992822647095,
-      "learning_rate": 6.166733668341709e-05,
-      "loss": 5.3666,
-      "step": 38655
-    },
-    {
-      "epoch": 20.159582790091264,
-      "grad_norm": 1.6320445537567139,
-      "learning_rate": 6.166633165829146e-05,
-      "loss": 5.0336,
-      "step": 38656
-    },
-    {
-      "epoch": 20.160104302477183,
-      "grad_norm": 1.6178951263427734,
-      "learning_rate": 6.166532663316583e-05,
-      "loss": 5.2347,
-      "step": 38657
-    },
-    {
-      "epoch": 20.160625814863103,
-      "grad_norm": 1.6060850620269775,
-      "learning_rate": 6.16643216080402e-05,
-      "loss": 5.4943,
-      "step": 38658
-    },
-    {
-      "epoch": 20.161147327249022,
-      "grad_norm": 1.4348559379577637,
-      "learning_rate": 6.166331658291458e-05,
-      "loss": 5.0866,
-      "step": 38659
-    },
-    {
-      "epoch": 20.16166883963494,
-      "grad_norm": 1.5869314670562744,
-      "learning_rate": 6.166231155778895e-05,
-      "loss": 5.4427,
-      "step": 38660
-    },
-    {
-      "epoch": 20.16219035202086,
-      "grad_norm": 1.5707459449768066,
-      "learning_rate": 6.166130653266332e-05,
-      "loss": 4.9008,
-      "step": 38661
-    },
-    {
-      "epoch": 20.16271186440678,
-      "grad_norm": 1.59604811668396,
-      "learning_rate": 6.166030150753769e-05,
-      "loss": 5.0528,
-      "step": 38662
-    },
-    {
-      "epoch": 20.1632333767927,
-      "grad_norm": 1.3991843461990356,
-      "learning_rate": 6.165929648241207e-05,
-      "loss": 5.2545,
-      "step": 38663
-    },
-    {
-      "epoch": 20.16375488917862,
-      "grad_norm": 1.445783019065857,
-      "learning_rate": 6.165829145728643e-05,
-      "loss": 5.7448,
-      "step": 38664
-    },
-    {
-      "epoch": 20.164276401564535,
-      "grad_norm": 1.605556607246399,
-      "learning_rate": 6.165728643216081e-05,
-      "loss": 4.818,
-      "step": 38665
-    },
-    {
-      "epoch": 20.164797913950455,
-      "grad_norm": 1.5854179859161377,
-      "learning_rate": 6.165628140703517e-05,
-      "loss": 5.0649,
-      "step": 38666
-    },
-    {
-      "epoch": 20.165319426336374,
-      "grad_norm": 1.4289045333862305,
-      "learning_rate": 6.165527638190955e-05,
-      "loss": 5.0823,
-      "step": 38667
-    },
-    {
-      "epoch": 20.165840938722294,
-      "grad_norm": 1.534348726272583,
-      "learning_rate": 6.165427135678392e-05,
-      "loss": 5.5518,
-      "step": 38668
-    },
-    {
-      "epoch": 20.166362451108213,
-      "grad_norm": 1.5558463335037231,
-      "learning_rate": 6.16532663316583e-05,
-      "loss": 5.6122,
-      "step": 38669
-    },
-    {
-      "epoch": 20.166883963494133,
-      "grad_norm": 1.5152127742767334,
-      "learning_rate": 6.165226130653267e-05,
-      "loss": 5.6267,
-      "step": 38670
-    },
-    {
-      "epoch": 20.167405475880052,
-      "grad_norm": 1.5328717231750488,
-      "learning_rate": 6.165125628140704e-05,
-      "loss": 5.1223,
-      "step": 38671
-    },
-    {
-      "epoch": 20.16792698826597,
-      "grad_norm": 1.5127960443496704,
-      "learning_rate": 6.165025125628141e-05,
-      "loss": 5.0523,
-      "step": 38672
-    },
-    {
-      "epoch": 20.16844850065189,
-      "grad_norm": 1.5602655410766602,
-      "learning_rate": 6.164924623115578e-05,
-      "loss": 5.4612,
-      "step": 38673
-    },
-    {
-      "epoch": 20.16897001303781,
-      "grad_norm": 1.5324584245681763,
-      "learning_rate": 6.164824120603016e-05,
-      "loss": 5.0763,
-      "step": 38674
-    },
-    {
-      "epoch": 20.16949152542373,
-      "grad_norm": 1.4503953456878662,
-      "learning_rate": 6.164723618090452e-05,
-      "loss": 5.7364,
-      "step": 38675
-    },
-    {
-      "epoch": 20.17001303780965,
-      "grad_norm": 1.5634844303131104,
-      "learning_rate": 6.16462311557789e-05,
-      "loss": 5.068,
-      "step": 38676
-    },
-    {
-      "epoch": 20.170534550195566,
-      "grad_norm": 1.5550934076309204,
-      "learning_rate": 6.164522613065326e-05,
-      "loss": 5.1348,
-      "step": 38677
-    },
-    {
-      "epoch": 20.171056062581485,
-      "grad_norm": 1.5124202966690063,
-      "learning_rate": 6.164422110552764e-05,
-      "loss": 5.2685,
-      "step": 38678
-    },
-    {
-      "epoch": 20.171577574967404,
-      "grad_norm": 1.5481520891189575,
-      "learning_rate": 6.164321608040202e-05,
-      "loss": 5.3227,
-      "step": 38679
-    },
-    {
-      "epoch": 20.172099087353324,
-      "grad_norm": 1.480562686920166,
-      "learning_rate": 6.16422110552764e-05,
-      "loss": 5.244,
-      "step": 38680
-    },
-    {
-      "epoch": 20.172620599739243,
-      "grad_norm": 1.5333843231201172,
-      "learning_rate": 6.164120603015076e-05,
-      "loss": 4.7925,
-      "step": 38681
-    },
-    {
-      "epoch": 20.173142112125163,
-      "grad_norm": 1.5219974517822266,
-      "learning_rate": 6.164020100502514e-05,
-      "loss": 5.5282,
-      "step": 38682
-    },
-    {
-      "epoch": 20.173663624511082,
-      "grad_norm": 1.5302159786224365,
-      "learning_rate": 6.16391959798995e-05,
-      "loss": 5.599,
-      "step": 38683
-    },
-    {
-      "epoch": 20.174185136897,
-      "grad_norm": 1.4423072338104248,
-      "learning_rate": 6.163819095477388e-05,
-      "loss": 5.5866,
-      "step": 38684
-    },
-    {
-      "epoch": 20.17470664928292,
-      "grad_norm": 1.5089632272720337,
-      "learning_rate": 6.163718592964824e-05,
-      "loss": 5.5561,
-      "step": 38685
-    },
-    {
-      "epoch": 20.17522816166884,
-      "grad_norm": 1.494275450706482,
-      "learning_rate": 6.163618090452261e-05,
-      "loss": 5.4933,
-      "step": 38686
-    },
-    {
-      "epoch": 20.17574967405476,
-      "grad_norm": 1.5518304109573364,
-      "learning_rate": 6.163517587939699e-05,
-      "loss": 5.1324,
-      "step": 38687
-    },
-    {
-      "epoch": 20.17627118644068,
-      "grad_norm": 1.6110292673110962,
-      "learning_rate": 6.163417085427135e-05,
-      "loss": 5.096,
-      "step": 38688
-    },
-    {
-      "epoch": 20.176792698826596,
-      "grad_norm": 1.4092140197753906,
-      "learning_rate": 6.163316582914573e-05,
-      "loss": 4.3003,
-      "step": 38689
-    },
-    {
-      "epoch": 20.177314211212515,
-      "grad_norm": 1.4345428943634033,
-      "learning_rate": 6.16321608040201e-05,
-      "loss": 5.3422,
-      "step": 38690
-    },
-    {
-      "epoch": 20.177835723598434,
-      "grad_norm": 1.5705370903015137,
-      "learning_rate": 6.163115577889448e-05,
-      "loss": 5.4919,
-      "step": 38691
-    },
-    {
-      "epoch": 20.178357235984354,
-      "grad_norm": 1.5396379232406616,
-      "learning_rate": 6.163015075376885e-05,
-      "loss": 5.2908,
-      "step": 38692
-    },
-    {
-      "epoch": 20.178878748370273,
-      "grad_norm": 2.0110301971435547,
-      "learning_rate": 6.162914572864323e-05,
-      "loss": 5.1717,
-      "step": 38693
-    },
-    {
-      "epoch": 20.179400260756193,
-      "grad_norm": 1.6019877195358276,
-      "learning_rate": 6.162814070351759e-05,
-      "loss": 5.5089,
-      "step": 38694
-    },
-    {
-      "epoch": 20.179921773142112,
-      "grad_norm": 1.5300891399383545,
-      "learning_rate": 6.162713567839197e-05,
-      "loss": 5.4047,
-      "step": 38695
-    },
-    {
-      "epoch": 20.180443285528032,
-      "grad_norm": 1.590362548828125,
-      "learning_rate": 6.162613065326633e-05,
-      "loss": 5.2489,
-      "step": 38696
-    },
-    {
-      "epoch": 20.18096479791395,
-      "grad_norm": 1.5862038135528564,
-      "learning_rate": 6.162512562814071e-05,
-      "loss": 4.9228,
-      "step": 38697
-    },
-    {
-      "epoch": 20.18148631029987,
-      "grad_norm": 1.4764759540557861,
-      "learning_rate": 6.162412060301507e-05,
-      "loss": 5.6622,
-      "step": 38698
-    },
-    {
-      "epoch": 20.18200782268579,
-      "grad_norm": 1.4284685850143433,
-      "learning_rate": 6.162311557788945e-05,
-      "loss": 5.2658,
-      "step": 38699
-    },
-    {
-      "epoch": 20.18252933507171,
-      "grad_norm": 1.4263982772827148,
-      "learning_rate": 6.162211055276383e-05,
-      "loss": 5.7437,
-      "step": 38700
-    },
-    {
-      "epoch": 20.183050847457626,
-      "grad_norm": 1.5178149938583374,
-      "learning_rate": 6.16211055276382e-05,
-      "loss": 5.7092,
-      "step": 38701
-    },
-    {
-      "epoch": 20.183572359843545,
-      "grad_norm": 1.5071231126785278,
-      "learning_rate": 6.162010050251257e-05,
-      "loss": 5.3013,
-      "step": 38702
-    },
-    {
-      "epoch": 20.184093872229464,
-      "grad_norm": 1.575380563735962,
-      "learning_rate": 6.161909547738694e-05,
-      "loss": 5.2241,
-      "step": 38703
-    },
-    {
-      "epoch": 20.184615384615384,
-      "grad_norm": 1.644905924797058,
-      "learning_rate": 6.161809045226131e-05,
-      "loss": 5.1562,
-      "step": 38704
-    },
-    {
-      "epoch": 20.185136897001303,
-      "grad_norm": 1.4710954427719116,
-      "learning_rate": 6.161708542713568e-05,
-      "loss": 5.5142,
-      "step": 38705
-    },
-    {
-      "epoch": 20.185658409387223,
-      "grad_norm": 1.4019136428833008,
-      "learning_rate": 6.161608040201006e-05,
-      "loss": 5.4239,
-      "step": 38706
-    },
-    {
-      "epoch": 20.186179921773142,
-      "grad_norm": 1.4951791763305664,
-      "learning_rate": 6.161507537688442e-05,
-      "loss": 4.6579,
-      "step": 38707
-    },
-    {
-      "epoch": 20.186701434159062,
-      "grad_norm": 1.5546544790267944,
-      "learning_rate": 6.16140703517588e-05,
-      "loss": 5.2469,
-      "step": 38708
-    },
-    {
-      "epoch": 20.18722294654498,
-      "grad_norm": 1.5502313375473022,
-      "learning_rate": 6.161306532663316e-05,
-      "loss": 4.629,
-      "step": 38709
-    },
-    {
-      "epoch": 20.1877444589309,
-      "grad_norm": 1.53248131275177,
-      "learning_rate": 6.161206030150754e-05,
-      "loss": 5.5422,
-      "step": 38710
-    },
-    {
-      "epoch": 20.18826597131682,
-      "grad_norm": 1.5384491682052612,
-      "learning_rate": 6.161105527638192e-05,
-      "loss": 5.5666,
-      "step": 38711
-    },
-    {
-      "epoch": 20.18878748370274,
-      "grad_norm": 1.4549921751022339,
-      "learning_rate": 6.161005025125628e-05,
-      "loss": 5.6778,
-      "step": 38712
-    },
-    {
-      "epoch": 20.189308996088656,
-      "grad_norm": 1.3993197679519653,
-      "learning_rate": 6.160904522613066e-05,
-      "loss": 5.5648,
-      "step": 38713
-    },
-    {
-      "epoch": 20.189830508474575,
-      "grad_norm": 1.4174277782440186,
-      "learning_rate": 6.160804020100502e-05,
-      "loss": 5.5546,
-      "step": 38714
-    },
-    {
-      "epoch": 20.190352020860495,
-      "grad_norm": 1.6495996713638306,
-      "learning_rate": 6.16070351758794e-05,
-      "loss": 5.1813,
-      "step": 38715
-    },
-    {
-      "epoch": 20.190873533246414,
-      "grad_norm": 1.5890578031539917,
-      "learning_rate": 6.160603015075377e-05,
-      "loss": 5.1441,
-      "step": 38716
-    },
-    {
-      "epoch": 20.191395045632333,
-      "grad_norm": 1.5440881252288818,
-      "learning_rate": 6.160502512562814e-05,
-      "loss": 5.3273,
-      "step": 38717
-    },
-    {
-      "epoch": 20.191916558018253,
-      "grad_norm": 1.4479783773422241,
-      "learning_rate": 6.160402010050251e-05,
-      "loss": 5.331,
-      "step": 38718
-    },
-    {
-      "epoch": 20.192438070404172,
-      "grad_norm": 1.4906413555145264,
-      "learning_rate": 6.160301507537689e-05,
-      "loss": 5.2766,
-      "step": 38719
-    },
-    {
-      "epoch": 20.192959582790092,
-      "grad_norm": 1.6548335552215576,
-      "learning_rate": 6.160201005025125e-05,
-      "loss": 5.1345,
-      "step": 38720
-    },
-    {
-      "epoch": 20.19348109517601,
-      "grad_norm": 1.5615806579589844,
-      "learning_rate": 6.160100502512563e-05,
-      "loss": 5.1456,
-      "step": 38721
-    },
-    {
-      "epoch": 20.19400260756193,
-      "grad_norm": 1.6723201274871826,
-      "learning_rate": 6.16e-05,
-      "loss": 4.8182,
-      "step": 38722
-    },
-    {
-      "epoch": 20.19452411994785,
-      "grad_norm": 1.6104260683059692,
-      "learning_rate": 6.159899497487438e-05,
-      "loss": 5.149,
-      "step": 38723
-    },
-    {
-      "epoch": 20.195045632333766,
-      "grad_norm": 1.4527119398117065,
-      "learning_rate": 6.159798994974875e-05,
-      "loss": 5.4252,
-      "step": 38724
-    },
-    {
-      "epoch": 20.195567144719686,
-      "grad_norm": 1.4763766527175903,
-      "learning_rate": 6.159698492462311e-05,
-      "loss": 5.8091,
-      "step": 38725
-    },
-    {
-      "epoch": 20.196088657105605,
-      "grad_norm": 1.4666398763656616,
-      "learning_rate": 6.159597989949749e-05,
-      "loss": 5.3024,
-      "step": 38726
-    },
-    {
-      "epoch": 20.196610169491525,
-      "grad_norm": 1.557051181793213,
-      "learning_rate": 6.159497487437186e-05,
-      "loss": 4.9849,
-      "step": 38727
-    },
-    {
-      "epoch": 20.197131681877444,
-      "grad_norm": 1.580950140953064,
-      "learning_rate": 6.159396984924623e-05,
-      "loss": 5.052,
-      "step": 38728
-    },
-    {
-      "epoch": 20.197653194263363,
-      "grad_norm": 1.5560944080352783,
-      "learning_rate": 6.15929648241206e-05,
-      "loss": 5.379,
-      "step": 38729
-    },
-    {
-      "epoch": 20.198174706649283,
-      "grad_norm": 1.5086809396743774,
-      "learning_rate": 6.159195979899497e-05,
-      "loss": 5.3295,
-      "step": 38730
-    },
-    {
-      "epoch": 20.198696219035202,
-      "grad_norm": 1.4507672786712646,
-      "learning_rate": 6.159095477386935e-05,
-      "loss": 5.4818,
-      "step": 38731
-    },
-    {
-      "epoch": 20.199217731421122,
-      "grad_norm": 1.5740967988967896,
-      "learning_rate": 6.158994974874373e-05,
-      "loss": 4.9317,
-      "step": 38732
-    },
-    {
-      "epoch": 20.19973924380704,
-      "grad_norm": 1.6050904989242554,
-      "learning_rate": 6.15889447236181e-05,
-      "loss": 5.3358,
-      "step": 38733
-    },
-    {
-      "epoch": 20.20026075619296,
-      "grad_norm": 1.5623772144317627,
-      "learning_rate": 6.158793969849247e-05,
-      "loss": 5.2654,
-      "step": 38734
-    },
-    {
-      "epoch": 20.20078226857888,
-      "grad_norm": 1.5484157800674438,
-      "learning_rate": 6.158693467336684e-05,
-      "loss": 4.8295,
-      "step": 38735
-    },
-    {
-      "epoch": 20.201303780964796,
-      "grad_norm": 1.5139784812927246,
-      "learning_rate": 6.158592964824121e-05,
-      "loss": 4.6044,
-      "step": 38736
-    },
-    {
-      "epoch": 20.201825293350716,
-      "grad_norm": 1.591526985168457,
-      "learning_rate": 6.158492462311558e-05,
-      "loss": 5.2619,
-      "step": 38737
-    },
-    {
-      "epoch": 20.202346805736635,
-      "grad_norm": 1.5008195638656616,
-      "learning_rate": 6.158391959798994e-05,
-      "loss": 5.1938,
-      "step": 38738
-    },
-    {
-      "epoch": 20.202868318122555,
-      "grad_norm": 1.5061639547348022,
-      "learning_rate": 6.158291457286432e-05,
-      "loss": 5.3082,
-      "step": 38739
-    },
-    {
-      "epoch": 20.203389830508474,
-      "grad_norm": 1.4229801893234253,
-      "learning_rate": 6.158190954773869e-05,
-      "loss": 5.8232,
-      "step": 38740
-    },
-    {
-      "epoch": 20.203911342894393,
-      "grad_norm": 1.5004254579544067,
-      "learning_rate": 6.158090452261306e-05,
-      "loss": 5.2783,
-      "step": 38741
-    },
-    {
-      "epoch": 20.204432855280313,
-      "grad_norm": 1.5466594696044922,
-      "learning_rate": 6.157989949748744e-05,
-      "loss": 5.0645,
-      "step": 38742
-    },
-    {
-      "epoch": 20.204954367666232,
-      "grad_norm": 1.518619418144226,
-      "learning_rate": 6.157889447236182e-05,
-      "loss": 5.5035,
-      "step": 38743
-    },
-    {
-      "epoch": 20.205475880052152,
-      "grad_norm": 1.4989635944366455,
-      "learning_rate": 6.157788944723618e-05,
-      "loss": 5.4791,
-      "step": 38744
-    },
-    {
-      "epoch": 20.20599739243807,
-      "grad_norm": 1.459814190864563,
-      "learning_rate": 6.157688442211056e-05,
-      "loss": 5.6589,
-      "step": 38745
-    },
-    {
-      "epoch": 20.20651890482399,
-      "grad_norm": 1.5033451318740845,
-      "learning_rate": 6.157587939698493e-05,
-      "loss": 5.0698,
-      "step": 38746
-    },
-    {
-      "epoch": 20.20704041720991,
-      "grad_norm": 1.4514124393463135,
-      "learning_rate": 6.15748743718593e-05,
-      "loss": 5.1882,
-      "step": 38747
-    },
-    {
-      "epoch": 20.207561929595826,
-      "grad_norm": 1.44204843044281,
-      "learning_rate": 6.157386934673367e-05,
-      "loss": 4.6966,
-      "step": 38748
-    },
-    {
-      "epoch": 20.208083441981746,
-      "grad_norm": 1.6254905462265015,
-      "learning_rate": 6.157286432160805e-05,
-      "loss": 5.5012,
-      "step": 38749
-    },
-    {
-      "epoch": 20.208604954367665,
-      "grad_norm": 1.5037719011306763,
-      "learning_rate": 6.157185929648241e-05,
-      "loss": 4.9922,
-      "step": 38750
-    },
-    {
-      "epoch": 20.209126466753585,
-      "grad_norm": 1.5800042152404785,
-      "learning_rate": 6.157085427135679e-05,
-      "loss": 5.2329,
-      "step": 38751
-    },
-    {
-      "epoch": 20.209647979139504,
-      "grad_norm": 1.4816044569015503,
-      "learning_rate": 6.156984924623117e-05,
-      "loss": 5.5745,
-      "step": 38752
-    },
-    {
-      "epoch": 20.210169491525424,
-      "grad_norm": 1.603934407234192,
-      "learning_rate": 6.156884422110553e-05,
-      "loss": 5.2091,
-      "step": 38753
-    },
-    {
-      "epoch": 20.210691003911343,
-      "grad_norm": 1.3865203857421875,
-      "learning_rate": 6.156783919597991e-05,
-      "loss": 5.5024,
-      "step": 38754
-    },
-    {
-      "epoch": 20.211212516297262,
-      "grad_norm": 1.6053229570388794,
-      "learning_rate": 6.156683417085427e-05,
-      "loss": 5.4094,
-      "step": 38755
-    },
-    {
-      "epoch": 20.211734028683182,
-      "grad_norm": 1.4980101585388184,
-      "learning_rate": 6.156582914572865e-05,
-      "loss": 5.5568,
-      "step": 38756
-    },
-    {
-      "epoch": 20.2122555410691,
-      "grad_norm": 1.5558741092681885,
-      "learning_rate": 6.156482412060301e-05,
-      "loss": 5.1382,
-      "step": 38757
-    },
-    {
-      "epoch": 20.21277705345502,
-      "grad_norm": 1.4673783779144287,
-      "learning_rate": 6.156381909547739e-05,
-      "loss": 5.6164,
-      "step": 38758
-    },
-    {
-      "epoch": 20.21329856584094,
-      "grad_norm": 1.5643507242202759,
-      "learning_rate": 6.156281407035176e-05,
-      "loss": 5.5524,
-      "step": 38759
-    },
-    {
-      "epoch": 20.213820078226856,
-      "grad_norm": 1.5511448383331299,
-      "learning_rate": 6.156180904522613e-05,
-      "loss": 5.1981,
-      "step": 38760
-    },
-    {
-      "epoch": 20.214341590612776,
-      "grad_norm": 1.6431188583374023,
-      "learning_rate": 6.15608040201005e-05,
-      "loss": 4.9717,
-      "step": 38761
-    },
-    {
-      "epoch": 20.214863102998695,
-      "grad_norm": 1.4964299201965332,
-      "learning_rate": 6.155979899497488e-05,
-      "loss": 5.6104,
-      "step": 38762
-    },
-    {
-      "epoch": 20.215384615384615,
-      "grad_norm": 1.451062560081482,
-      "learning_rate": 6.155879396984925e-05,
-      "loss": 4.3644,
-      "step": 38763
-    },
-    {
-      "epoch": 20.215906127770534,
-      "grad_norm": 1.4916681051254272,
-      "learning_rate": 6.155778894472362e-05,
-      "loss": 5.0697,
-      "step": 38764
-    },
-    {
-      "epoch": 20.216427640156454,
-      "grad_norm": 1.4641718864440918,
-      "learning_rate": 6.1556783919598e-05,
-      "loss": 5.4302,
-      "step": 38765
-    },
-    {
-      "epoch": 20.216949152542373,
-      "grad_norm": 1.4921820163726807,
-      "learning_rate": 6.155577889447236e-05,
-      "loss": 5.3383,
-      "step": 38766
-    },
-    {
-      "epoch": 20.217470664928292,
-      "grad_norm": 1.583154559135437,
-      "learning_rate": 6.155477386934674e-05,
-      "loss": 5.113,
-      "step": 38767
-    },
-    {
-      "epoch": 20.217992177314212,
-      "grad_norm": 1.5793778896331787,
-      "learning_rate": 6.15537688442211e-05,
-      "loss": 5.3046,
-      "step": 38768
-    },
-    {
-      "epoch": 20.21851368970013,
-      "grad_norm": 1.526437759399414,
-      "learning_rate": 6.155276381909548e-05,
-      "loss": 5.0363,
-      "step": 38769
-    },
-    {
-      "epoch": 20.21903520208605,
-      "grad_norm": 1.5280686616897583,
-      "learning_rate": 6.155175879396984e-05,
-      "loss": 5.8405,
-      "step": 38770
-    },
-    {
-      "epoch": 20.21955671447197,
-      "grad_norm": 1.626832365989685,
-      "learning_rate": 6.155075376884422e-05,
-      "loss": 5.034,
-      "step": 38771
-    },
-    {
-      "epoch": 20.220078226857886,
-      "grad_norm": 1.589781403541565,
-      "learning_rate": 6.15497487437186e-05,
-      "loss": 4.8952,
-      "step": 38772
-    },
-    {
-      "epoch": 20.220599739243806,
-      "grad_norm": 1.562495470046997,
-      "learning_rate": 6.154874371859298e-05,
-      "loss": 5.4704,
-      "step": 38773
-    },
-    {
-      "epoch": 20.221121251629725,
-      "grad_norm": 1.5884995460510254,
-      "learning_rate": 6.154773869346734e-05,
-      "loss": 5.3986,
-      "step": 38774
-    },
-    {
-      "epoch": 20.221642764015645,
-      "grad_norm": 1.5738818645477295,
-      "learning_rate": 6.154673366834172e-05,
-      "loss": 5.0101,
-      "step": 38775
-    },
-    {
-      "epoch": 20.222164276401564,
-      "grad_norm": 1.5095783472061157,
-      "learning_rate": 6.154572864321608e-05,
-      "loss": 5.1955,
-      "step": 38776
-    },
-    {
-      "epoch": 20.222685788787484,
-      "grad_norm": 1.4916489124298096,
-      "learning_rate": 6.154472361809046e-05,
-      "loss": 5.2976,
-      "step": 38777
-    },
-    {
-      "epoch": 20.223207301173403,
-      "grad_norm": 1.4893434047698975,
-      "learning_rate": 6.154371859296483e-05,
-      "loss": 5.4731,
-      "step": 38778
-    },
-    {
-      "epoch": 20.223728813559323,
-      "grad_norm": 1.39323091506958,
-      "learning_rate": 6.154271356783919e-05,
-      "loss": 5.3792,
-      "step": 38779
-    },
-    {
-      "epoch": 20.224250325945242,
-      "grad_norm": 1.4979183673858643,
-      "learning_rate": 6.154170854271357e-05,
-      "loss": 5.5766,
-      "step": 38780
-    },
-    {
-      "epoch": 20.22477183833116,
-      "grad_norm": 1.5326833724975586,
-      "learning_rate": 6.154070351758793e-05,
-      "loss": 5.0445,
-      "step": 38781
-    },
-    {
-      "epoch": 20.22529335071708,
-      "grad_norm": 1.5402565002441406,
-      "learning_rate": 6.153969849246231e-05,
-      "loss": 4.9427,
-      "step": 38782
-    },
-    {
-      "epoch": 20.225814863103,
-      "grad_norm": 1.6214228868484497,
-      "learning_rate": 6.153869346733669e-05,
-      "loss": 5.4567,
-      "step": 38783
-    },
-    {
-      "epoch": 20.226336375488916,
-      "grad_norm": 1.5909479856491089,
-      "learning_rate": 6.153768844221107e-05,
-      "loss": 4.8861,
-      "step": 38784
-    },
-    {
-      "epoch": 20.226857887874836,
-      "grad_norm": 1.5378336906433105,
-      "learning_rate": 6.153668341708543e-05,
-      "loss": 5.1053,
-      "step": 38785
-    },
-    {
-      "epoch": 20.227379400260755,
-      "grad_norm": 1.5735191106796265,
-      "learning_rate": 6.153567839195981e-05,
-      "loss": 5.3115,
-      "step": 38786
-    },
-    {
-      "epoch": 20.227900912646675,
-      "grad_norm": 1.6072754859924316,
-      "learning_rate": 6.153467336683417e-05,
-      "loss": 4.6644,
-      "step": 38787
-    },
-    {
-      "epoch": 20.228422425032594,
-      "grad_norm": 1.3864837884902954,
-      "learning_rate": 6.153366834170855e-05,
-      "loss": 5.5825,
-      "step": 38788
-    },
-    {
-      "epoch": 20.228943937418514,
-      "grad_norm": 1.6598174571990967,
-      "learning_rate": 6.153266331658291e-05,
-      "loss": 4.4815,
-      "step": 38789
-    },
-    {
-      "epoch": 20.229465449804433,
-      "grad_norm": 1.593738317489624,
-      "learning_rate": 6.153165829145729e-05,
-      "loss": 5.2755,
-      "step": 38790
-    },
-    {
-      "epoch": 20.229986962190353,
-      "grad_norm": 1.5579906702041626,
-      "learning_rate": 6.153065326633166e-05,
-      "loss": 5.2043,
-      "step": 38791
-    },
-    {
-      "epoch": 20.230508474576272,
-      "grad_norm": 1.4602326154708862,
-      "learning_rate": 6.152964824120603e-05,
-      "loss": 4.8043,
-      "step": 38792
-    },
-    {
-      "epoch": 20.23102998696219,
-      "grad_norm": 1.4637948274612427,
-      "learning_rate": 6.152864321608041e-05,
-      "loss": 5.6723,
-      "step": 38793
-    },
-    {
-      "epoch": 20.23155149934811,
-      "grad_norm": 1.4770451784133911,
-      "learning_rate": 6.152763819095478e-05,
-      "loss": 5.3415,
-      "step": 38794
-    },
-    {
-      "epoch": 20.23207301173403,
-      "grad_norm": 1.3597707748413086,
-      "learning_rate": 6.152663316582915e-05,
-      "loss": 5.7734,
-      "step": 38795
-    },
-    {
-      "epoch": 20.232594524119946,
-      "grad_norm": 1.532223105430603,
-      "learning_rate": 6.152562814070352e-05,
-      "loss": 5.4036,
-      "step": 38796
-    },
-    {
-      "epoch": 20.233116036505866,
-      "grad_norm": 1.3785998821258545,
-      "learning_rate": 6.15246231155779e-05,
-      "loss": 5.4101,
-      "step": 38797
-    },
-    {
-      "epoch": 20.233637548891785,
-      "grad_norm": 1.540358066558838,
-      "learning_rate": 6.152361809045226e-05,
-      "loss": 5.4152,
-      "step": 38798
-    },
-    {
-      "epoch": 20.234159061277705,
-      "grad_norm": 1.451528787612915,
-      "learning_rate": 6.152261306532664e-05,
-      "loss": 5.0288,
-      "step": 38799
-    },
-    {
-      "epoch": 20.234680573663624,
-      "grad_norm": 1.5864378213882446,
-      "learning_rate": 6.1521608040201e-05,
-      "loss": 5.1697,
-      "step": 38800
-    },
-    {
-      "epoch": 20.235202086049544,
-      "grad_norm": 1.5600072145462036,
-      "learning_rate": 6.152060301507538e-05,
-      "loss": 5.0437,
-      "step": 38801
-    },
-    {
-      "epoch": 20.235723598435463,
-      "grad_norm": 1.5615878105163574,
-      "learning_rate": 6.151959798994974e-05,
-      "loss": 4.9346,
-      "step": 38802
-    },
-    {
-      "epoch": 20.236245110821383,
-      "grad_norm": 1.4146225452423096,
-      "learning_rate": 6.151859296482412e-05,
-      "loss": 5.5031,
-      "step": 38803
-    },
-    {
-      "epoch": 20.236766623207302,
-      "grad_norm": 1.5940543413162231,
-      "learning_rate": 6.15175879396985e-05,
-      "loss": 5.3446,
-      "step": 38804
-    },
-    {
-      "epoch": 20.23728813559322,
-      "grad_norm": 1.5338404178619385,
-      "learning_rate": 6.151658291457286e-05,
-      "loss": 5.2861,
-      "step": 38805
-    },
-    {
-      "epoch": 20.23780964797914,
-      "grad_norm": 1.6093162298202515,
-      "learning_rate": 6.151557788944724e-05,
-      "loss": 5.5733,
-      "step": 38806
-    },
-    {
-      "epoch": 20.23833116036506,
-      "grad_norm": 1.5166798830032349,
-      "learning_rate": 6.151457286432161e-05,
-      "loss": 4.9553,
-      "step": 38807
-    },
-    {
-      "epoch": 20.238852672750976,
-      "grad_norm": 1.4622304439544678,
-      "learning_rate": 6.151356783919598e-05,
-      "loss": 5.664,
-      "step": 38808
-    },
-    {
-      "epoch": 20.239374185136896,
-      "grad_norm": 1.4933046102523804,
-      "learning_rate": 6.151256281407035e-05,
-      "loss": 4.556,
-      "step": 38809
-    },
-    {
-      "epoch": 20.239895697522815,
-      "grad_norm": 1.5863615274429321,
-      "learning_rate": 6.151155778894473e-05,
-      "loss": 5.2111,
-      "step": 38810
-    },
-    {
-      "epoch": 20.240417209908735,
-      "grad_norm": 1.5060656070709229,
-      "learning_rate": 6.151055276381909e-05,
-      "loss": 5.4308,
-      "step": 38811
-    },
-    {
-      "epoch": 20.240938722294654,
-      "grad_norm": 1.4615967273712158,
-      "learning_rate": 6.150954773869347e-05,
-      "loss": 5.883,
-      "step": 38812
-    },
-    {
-      "epoch": 20.241460234680574,
-      "grad_norm": 1.4310134649276733,
-      "learning_rate": 6.150854271356785e-05,
-      "loss": 5.5608,
-      "step": 38813
-    },
-    {
-      "epoch": 20.241981747066493,
-      "grad_norm": 1.444650650024414,
-      "learning_rate": 6.150753768844222e-05,
-      "loss": 5.1797,
-      "step": 38814
-    },
-    {
-      "epoch": 20.242503259452413,
-      "grad_norm": 1.6298139095306396,
-      "learning_rate": 6.150653266331659e-05,
-      "loss": 5.0461,
-      "step": 38815
-    },
-    {
-      "epoch": 20.243024771838332,
-      "grad_norm": 1.5772768259048462,
-      "learning_rate": 6.150552763819097e-05,
-      "loss": 4.8134,
-      "step": 38816
-    },
-    {
-      "epoch": 20.24354628422425,
-      "grad_norm": 1.4875357151031494,
-      "learning_rate": 6.150452261306533e-05,
-      "loss": 5.5939,
-      "step": 38817
-    },
-    {
-      "epoch": 20.24406779661017,
-      "grad_norm": 1.443407654762268,
-      "learning_rate": 6.15035175879397e-05,
-      "loss": 5.6761,
-      "step": 38818
-    },
-    {
-      "epoch": 20.24458930899609,
-      "grad_norm": 1.488156795501709,
-      "learning_rate": 6.150251256281407e-05,
-      "loss": 5.3069,
-      "step": 38819
-    },
-    {
-      "epoch": 20.245110821382006,
-      "grad_norm": 1.4697115421295166,
-      "learning_rate": 6.150150753768844e-05,
-      "loss": 5.7286,
-      "step": 38820
-    },
-    {
-      "epoch": 20.245632333767926,
-      "grad_norm": 1.5686125755310059,
-      "learning_rate": 6.150050251256282e-05,
-      "loss": 4.8816,
-      "step": 38821
-    },
-    {
-      "epoch": 20.246153846153845,
-      "grad_norm": 1.6092056035995483,
-      "learning_rate": 6.149949748743718e-05,
-      "loss": 5.0332,
-      "step": 38822
-    },
-    {
-      "epoch": 20.246675358539765,
-      "grad_norm": 1.4796721935272217,
-      "learning_rate": 6.149849246231156e-05,
-      "loss": 5.1921,
-      "step": 38823
-    },
-    {
-      "epoch": 20.247196870925684,
-      "grad_norm": 1.6173590421676636,
-      "learning_rate": 6.149748743718594e-05,
-      "loss": 4.9955,
-      "step": 38824
-    },
-    {
-      "epoch": 20.247718383311604,
-      "grad_norm": 1.621656060218811,
-      "learning_rate": 6.149648241206031e-05,
-      "loss": 5.2917,
-      "step": 38825
-    },
-    {
-      "epoch": 20.248239895697523,
-      "grad_norm": 1.3533858060836792,
-      "learning_rate": 6.149547738693468e-05,
-      "loss": 5.6795,
-      "step": 38826
-    },
-    {
-      "epoch": 20.248761408083443,
-      "grad_norm": 1.5897855758666992,
-      "learning_rate": 6.149447236180906e-05,
-      "loss": 5.2868,
-      "step": 38827
-    },
-    {
-      "epoch": 20.249282920469362,
-      "grad_norm": 1.4255290031433105,
-      "learning_rate": 6.149346733668342e-05,
-      "loss": 5.6681,
-      "step": 38828
-    },
-    {
-      "epoch": 20.24980443285528,
-      "grad_norm": 1.4905554056167603,
-      "learning_rate": 6.14924623115578e-05,
-      "loss": 4.9006,
-      "step": 38829
-    },
-    {
-      "epoch": 20.2503259452412,
-      "grad_norm": 1.6326013803482056,
-      "learning_rate": 6.149145728643216e-05,
-      "loss": 5.2261,
-      "step": 38830
-    },
-    {
-      "epoch": 20.250847457627117,
-      "grad_norm": 1.55451500415802,
-      "learning_rate": 6.149045226130653e-05,
-      "loss": 5.4792,
-      "step": 38831
-    },
-    {
-      "epoch": 20.251368970013036,
-      "grad_norm": 1.4840644598007202,
-      "learning_rate": 6.14894472361809e-05,
-      "loss": 5.4717,
-      "step": 38832
-    },
-    {
-      "epoch": 20.251890482398956,
-      "grad_norm": 1.5045934915542603,
-      "learning_rate": 6.148844221105528e-05,
-      "loss": 5.4938,
-      "step": 38833
-    },
-    {
-      "epoch": 20.252411994784875,
-      "grad_norm": 1.5231295824050903,
-      "learning_rate": 6.148743718592966e-05,
-      "loss": 4.9925,
-      "step": 38834
-    },
-    {
-      "epoch": 20.252933507170795,
-      "grad_norm": 1.4792925119400024,
-      "learning_rate": 6.148643216080402e-05,
-      "loss": 5.6739,
-      "step": 38835
-    },
-    {
-      "epoch": 20.253455019556714,
-      "grad_norm": 1.5991653203964233,
-      "learning_rate": 6.14854271356784e-05,
-      "loss": 5.5947,
-      "step": 38836
-    },
-    {
-      "epoch": 20.253976531942634,
-      "grad_norm": 1.6063240766525269,
-      "learning_rate": 6.148442211055277e-05,
-      "loss": 5.6252,
-      "step": 38837
-    },
-    {
-      "epoch": 20.254498044328553,
-      "grad_norm": 1.4987483024597168,
-      "learning_rate": 6.148341708542714e-05,
-      "loss": 5.459,
-      "step": 38838
-    },
-    {
-      "epoch": 20.255019556714473,
-      "grad_norm": 1.4964790344238281,
-      "learning_rate": 6.148241206030151e-05,
-      "loss": 5.5283,
-      "step": 38839
-    },
-    {
-      "epoch": 20.255541069100392,
-      "grad_norm": 1.439732551574707,
-      "learning_rate": 6.148140703517589e-05,
-      "loss": 5.2748,
-      "step": 38840
-    },
-    {
-      "epoch": 20.25606258148631,
-      "grad_norm": 1.4735980033874512,
-      "learning_rate": 6.148040201005025e-05,
-      "loss": 5.0775,
-      "step": 38841
-    },
-    {
-      "epoch": 20.25658409387223,
-      "grad_norm": 1.637336015701294,
-      "learning_rate": 6.147939698492463e-05,
-      "loss": 5.0317,
-      "step": 38842
-    },
-    {
-      "epoch": 20.257105606258147,
-      "grad_norm": 1.7421656847000122,
-      "learning_rate": 6.147839195979899e-05,
-      "loss": 4.839,
-      "step": 38843
-    },
-    {
-      "epoch": 20.257627118644066,
-      "grad_norm": 1.575321078300476,
-      "learning_rate": 6.147738693467337e-05,
-      "loss": 5.4285,
-      "step": 38844
-    },
-    {
-      "epoch": 20.258148631029986,
-      "grad_norm": 1.659794569015503,
-      "learning_rate": 6.147638190954775e-05,
-      "loss": 5.581,
-      "step": 38845
-    },
-    {
-      "epoch": 20.258670143415905,
-      "grad_norm": 1.4580484628677368,
-      "learning_rate": 6.147537688442211e-05,
-      "loss": 5.3262,
-      "step": 38846
-    },
-    {
-      "epoch": 20.259191655801825,
-      "grad_norm": 1.5640933513641357,
-      "learning_rate": 6.147437185929649e-05,
-      "loss": 4.7674,
-      "step": 38847
-    },
-    {
-      "epoch": 20.259713168187744,
-      "grad_norm": 1.4847848415374756,
-      "learning_rate": 6.147336683417085e-05,
-      "loss": 5.7128,
-      "step": 38848
-    },
-    {
-      "epoch": 20.260234680573664,
-      "grad_norm": 1.547215461730957,
-      "learning_rate": 6.147236180904523e-05,
-      "loss": 5.3642,
-      "step": 38849
-    },
-    {
-      "epoch": 20.260756192959583,
-      "grad_norm": 1.4898394346237183,
-      "learning_rate": 6.14713567839196e-05,
-      "loss": 5.5543,
-      "step": 38850
-    },
-    {
-      "epoch": 20.261277705345503,
-      "grad_norm": 1.487136960029602,
-      "learning_rate": 6.147035175879397e-05,
-      "loss": 5.2852,
-      "step": 38851
-    },
-    {
-      "epoch": 20.261799217731422,
-      "grad_norm": 1.5314915180206299,
-      "learning_rate": 6.146934673366834e-05,
-      "loss": 5.5285,
-      "step": 38852
-    },
-    {
-      "epoch": 20.26232073011734,
-      "grad_norm": 1.6019150018692017,
-      "learning_rate": 6.146834170854272e-05,
-      "loss": 5.0234,
-      "step": 38853
-    },
-    {
-      "epoch": 20.26284224250326,
-      "grad_norm": 1.465004563331604,
-      "learning_rate": 6.14673366834171e-05,
-      "loss": 5.4149,
-      "step": 38854
-    },
-    {
-      "epoch": 20.263363754889177,
-      "grad_norm": 1.4701566696166992,
-      "learning_rate": 6.146633165829147e-05,
-      "loss": 5.2787,
-      "step": 38855
-    },
-    {
-      "epoch": 20.263885267275096,
-      "grad_norm": 1.4532235860824585,
-      "learning_rate": 6.146532663316584e-05,
-      "loss": 5.728,
-      "step": 38856
-    },
-    {
-      "epoch": 20.264406779661016,
-      "grad_norm": 1.5708858966827393,
-      "learning_rate": 6.14643216080402e-05,
-      "loss": 5.075,
-      "step": 38857
-    },
-    {
-      "epoch": 20.264928292046935,
-      "grad_norm": 1.4573662281036377,
-      "learning_rate": 6.146331658291458e-05,
-      "loss": 5.4086,
-      "step": 38858
-    },
-    {
-      "epoch": 20.265449804432855,
-      "grad_norm": 1.404242992401123,
-      "learning_rate": 6.146231155778894e-05,
-      "loss": 4.7339,
-      "step": 38859
-    },
-    {
-      "epoch": 20.265971316818774,
-      "grad_norm": 1.4999991655349731,
-      "learning_rate": 6.146130653266332e-05,
-      "loss": 5.6051,
-      "step": 38860
-    },
-    {
-      "epoch": 20.266492829204694,
-      "grad_norm": 1.5094375610351562,
-      "learning_rate": 6.146030150753768e-05,
-      "loss": 5.4867,
-      "step": 38861
-    },
-    {
-      "epoch": 20.267014341590613,
-      "grad_norm": 1.4439409971237183,
-      "learning_rate": 6.145929648241206e-05,
-      "loss": 5.2829,
-      "step": 38862
-    },
-    {
-      "epoch": 20.267535853976533,
-      "grad_norm": 1.5473002195358276,
-      "learning_rate": 6.145829145728643e-05,
-      "loss": 4.9703,
-      "step": 38863
-    },
-    {
-      "epoch": 20.268057366362452,
-      "grad_norm": 1.5631005764007568,
-      "learning_rate": 6.14572864321608e-05,
-      "loss": 5.2262,
-      "step": 38864
-    },
-    {
-      "epoch": 20.26857887874837,
-      "grad_norm": 1.5108368396759033,
-      "learning_rate": 6.145628140703518e-05,
-      "loss": 5.1104,
-      "step": 38865
-    },
-    {
-      "epoch": 20.26910039113429,
-      "grad_norm": 1.588607668876648,
-      "learning_rate": 6.145527638190956e-05,
-      "loss": 5.4489,
-      "step": 38866
-    },
-    {
-      "epoch": 20.269621903520207,
-      "grad_norm": 1.5584031343460083,
-      "learning_rate": 6.145427135678392e-05,
-      "loss": 4.8257,
-      "step": 38867
-    },
-    {
-      "epoch": 20.270143415906126,
-      "grad_norm": 1.5939509868621826,
-      "learning_rate": 6.14532663316583e-05,
-      "loss": 5.2251,
-      "step": 38868
-    },
-    {
-      "epoch": 20.270664928292046,
-      "grad_norm": 1.534656286239624,
-      "learning_rate": 6.145226130653267e-05,
-      "loss": 5.141,
-      "step": 38869
-    },
-    {
-      "epoch": 20.271186440677965,
-      "grad_norm": 1.550120234489441,
-      "learning_rate": 6.145125628140704e-05,
-      "loss": 5.3279,
-      "step": 38870
-    },
-    {
-      "epoch": 20.271707953063885,
-      "grad_norm": 1.5358978509902954,
-      "learning_rate": 6.145025125628141e-05,
-      "loss": 5.0588,
-      "step": 38871
-    },
-    {
-      "epoch": 20.272229465449804,
-      "grad_norm": 1.5765703916549683,
-      "learning_rate": 6.144924623115577e-05,
-      "loss": 4.7941,
-      "step": 38872
-    },
-    {
-      "epoch": 20.272750977835724,
-      "grad_norm": 1.4579061269760132,
-      "learning_rate": 6.144824120603015e-05,
-      "loss": 5.201,
-      "step": 38873
-    },
-    {
-      "epoch": 20.273272490221643,
-      "grad_norm": 1.525787115097046,
-      "learning_rate": 6.144723618090453e-05,
-      "loss": 4.7932,
-      "step": 38874
-    },
-    {
-      "epoch": 20.273794002607563,
-      "grad_norm": 1.549570918083191,
-      "learning_rate": 6.14462311557789e-05,
-      "loss": 5.5149,
-      "step": 38875
-    },
-    {
-      "epoch": 20.274315514993482,
-      "grad_norm": 1.5403556823730469,
-      "learning_rate": 6.144522613065327e-05,
-      "loss": 5.1153,
-      "step": 38876
-    },
-    {
-      "epoch": 20.2748370273794,
-      "grad_norm": 1.57559335231781,
-      "learning_rate": 6.144422110552765e-05,
-      "loss": 5.0205,
-      "step": 38877
-    },
-    {
-      "epoch": 20.27535853976532,
-      "grad_norm": 1.4571866989135742,
-      "learning_rate": 6.144321608040201e-05,
-      "loss": 5.7639,
-      "step": 38878
-    },
-    {
-      "epoch": 20.275880052151237,
-      "grad_norm": 1.5154949426651,
-      "learning_rate": 6.144221105527639e-05,
-      "loss": 4.924,
-      "step": 38879
-    },
-    {
-      "epoch": 20.276401564537156,
-      "grad_norm": 1.5160510540008545,
-      "learning_rate": 6.144120603015075e-05,
-      "loss": 5.4196,
-      "step": 38880
-    },
-    {
-      "epoch": 20.276923076923076,
-      "grad_norm": 1.4504001140594482,
-      "learning_rate": 6.144020100502513e-05,
-      "loss": 5.4527,
-      "step": 38881
-    },
-    {
-      "epoch": 20.277444589308995,
-      "grad_norm": 1.6148287057876587,
-      "learning_rate": 6.14391959798995e-05,
-      "loss": 5.5924,
-      "step": 38882
-    },
-    {
-      "epoch": 20.277966101694915,
-      "grad_norm": 1.531987190246582,
-      "learning_rate": 6.143819095477387e-05,
-      "loss": 4.5716,
-      "step": 38883
-    },
-    {
-      "epoch": 20.278487614080834,
-      "grad_norm": 1.5701878070831299,
-      "learning_rate": 6.143718592964824e-05,
-      "loss": 4.9742,
-      "step": 38884
-    },
-    {
-      "epoch": 20.279009126466754,
-      "grad_norm": 1.496734619140625,
-      "learning_rate": 6.143618090452262e-05,
-      "loss": 5.5565,
-      "step": 38885
-    },
-    {
-      "epoch": 20.279530638852673,
-      "grad_norm": 1.4867477416992188,
-      "learning_rate": 6.1435175879397e-05,
-      "loss": 5.2839,
-      "step": 38886
-    },
-    {
-      "epoch": 20.280052151238593,
-      "grad_norm": 1.5807726383209229,
-      "learning_rate": 6.143417085427136e-05,
-      "loss": 4.8485,
-      "step": 38887
-    },
-    {
-      "epoch": 20.280573663624512,
-      "grad_norm": 1.6672688722610474,
-      "learning_rate": 6.143316582914574e-05,
-      "loss": 4.6759,
-      "step": 38888
-    },
-    {
-      "epoch": 20.28109517601043,
-      "grad_norm": 1.534371018409729,
-      "learning_rate": 6.14321608040201e-05,
-      "loss": 5.3765,
-      "step": 38889
-    },
-    {
-      "epoch": 20.28161668839635,
-      "grad_norm": 1.4836466312408447,
-      "learning_rate": 6.143115577889448e-05,
-      "loss": 5.4724,
-      "step": 38890
-    },
-    {
-      "epoch": 20.282138200782267,
-      "grad_norm": 1.4868474006652832,
-      "learning_rate": 6.143015075376884e-05,
-      "loss": 5.1598,
-      "step": 38891
-    },
-    {
-      "epoch": 20.282659713168186,
-      "grad_norm": 1.51041579246521,
-      "learning_rate": 6.142914572864322e-05,
-      "loss": 4.9031,
-      "step": 38892
-    },
-    {
-      "epoch": 20.283181225554106,
-      "grad_norm": 1.4888370037078857,
-      "learning_rate": 6.142814070351759e-05,
-      "loss": 5.2599,
-      "step": 38893
-    },
-    {
-      "epoch": 20.283702737940025,
-      "grad_norm": 1.4658361673355103,
-      "learning_rate": 6.142713567839196e-05,
-      "loss": 5.6359,
-      "step": 38894
-    },
-    {
-      "epoch": 20.284224250325945,
-      "grad_norm": 1.598999261856079,
-      "learning_rate": 6.142613065326633e-05,
-      "loss": 4.8818,
-      "step": 38895
-    },
-    {
-      "epoch": 20.284745762711864,
-      "grad_norm": 1.4351089000701904,
-      "learning_rate": 6.14251256281407e-05,
-      "loss": 5.5946,
-      "step": 38896
-    },
-    {
-      "epoch": 20.285267275097784,
-      "grad_norm": 1.378549337387085,
-      "learning_rate": 6.142412060301508e-05,
-      "loss": 4.8948,
-      "step": 38897
-    },
-    {
-      "epoch": 20.285788787483703,
-      "grad_norm": 1.5197657346725464,
-      "learning_rate": 6.142311557788945e-05,
-      "loss": 5.2293,
-      "step": 38898
-    },
-    {
-      "epoch": 20.286310299869623,
-      "grad_norm": 1.5250890254974365,
-      "learning_rate": 6.142211055276382e-05,
-      "loss": 5.128,
-      "step": 38899
-    },
-    {
-      "epoch": 20.286831812255542,
-      "grad_norm": 1.4345823526382446,
-      "learning_rate": 6.142110552763819e-05,
-      "loss": 5.3729,
-      "step": 38900
-    },
-    {
-      "epoch": 20.28735332464146,
-      "grad_norm": 1.4713832139968872,
-      "learning_rate": 6.142010050251257e-05,
-      "loss": 5.0619,
-      "step": 38901
-    },
-    {
-      "epoch": 20.28787483702738,
-      "grad_norm": 1.5416258573532104,
-      "learning_rate": 6.141909547738693e-05,
-      "loss": 5.327,
-      "step": 38902
-    },
-    {
-      "epoch": 20.288396349413297,
-      "grad_norm": 1.6048376560211182,
-      "learning_rate": 6.141809045226131e-05,
-      "loss": 5.318,
-      "step": 38903
-    },
-    {
-      "epoch": 20.288917861799217,
-      "grad_norm": 1.8454334735870361,
-      "learning_rate": 6.141708542713567e-05,
-      "loss": 4.76,
-      "step": 38904
-    },
-    {
-      "epoch": 20.289439374185136,
-      "grad_norm": 1.4412399530410767,
-      "learning_rate": 6.141608040201005e-05,
-      "loss": 5.2224,
-      "step": 38905
-    },
-    {
-      "epoch": 20.289960886571055,
-      "grad_norm": 1.4729448556900024,
-      "learning_rate": 6.141507537688443e-05,
-      "loss": 5.6735,
-      "step": 38906
-    },
-    {
-      "epoch": 20.290482398956975,
-      "grad_norm": 1.4507662057876587,
-      "learning_rate": 6.141407035175881e-05,
-      "loss": 5.0682,
-      "step": 38907
-    },
-    {
-      "epoch": 20.291003911342894,
-      "grad_norm": 1.5077345371246338,
-      "learning_rate": 6.141306532663317e-05,
-      "loss": 4.9456,
-      "step": 38908
-    },
-    {
-      "epoch": 20.291525423728814,
-      "grad_norm": 1.5296891927719116,
-      "learning_rate": 6.141206030150755e-05,
-      "loss": 5.0344,
-      "step": 38909
-    },
-    {
-      "epoch": 20.292046936114733,
-      "grad_norm": 1.4503198862075806,
-      "learning_rate": 6.141105527638191e-05,
-      "loss": 5.7422,
-      "step": 38910
-    },
-    {
-      "epoch": 20.292568448500653,
-      "grad_norm": 1.5403175354003906,
-      "learning_rate": 6.141005025125628e-05,
-      "loss": 5.46,
-      "step": 38911
-    },
-    {
-      "epoch": 20.293089960886572,
-      "grad_norm": 1.6044268608093262,
-      "learning_rate": 6.140904522613066e-05,
-      "loss": 5.2829,
-      "step": 38912
-    },
-    {
-      "epoch": 20.29361147327249,
-      "grad_norm": 1.5751328468322754,
-      "learning_rate": 6.140804020100502e-05,
-      "loss": 5.2846,
-      "step": 38913
-    },
-    {
-      "epoch": 20.294132985658408,
-      "grad_norm": 1.4373857975006104,
-      "learning_rate": 6.14070351758794e-05,
-      "loss": 5.6541,
-      "step": 38914
-    },
-    {
-      "epoch": 20.294654498044327,
-      "grad_norm": 1.4545822143554688,
-      "learning_rate": 6.140603015075376e-05,
-      "loss": 4.9846,
-      "step": 38915
-    },
-    {
-      "epoch": 20.295176010430247,
-      "grad_norm": 1.640956163406372,
-      "learning_rate": 6.140502512562814e-05,
-      "loss": 4.9976,
-      "step": 38916
-    },
-    {
-      "epoch": 20.295697522816166,
-      "grad_norm": 1.5287598371505737,
-      "learning_rate": 6.140402010050252e-05,
-      "loss": 5.272,
-      "step": 38917
-    },
-    {
-      "epoch": 20.296219035202085,
-      "grad_norm": 1.6217031478881836,
-      "learning_rate": 6.14030150753769e-05,
-      "loss": 5.0824,
-      "step": 38918
-    },
-    {
-      "epoch": 20.296740547588005,
-      "grad_norm": 1.4491448402404785,
-      "learning_rate": 6.140201005025126e-05,
-      "loss": 5.2191,
-      "step": 38919
-    },
-    {
-      "epoch": 20.297262059973924,
-      "grad_norm": 1.5324313640594482,
-      "learning_rate": 6.140100502512564e-05,
-      "loss": 5.2206,
-      "step": 38920
-    },
-    {
-      "epoch": 20.297783572359844,
-      "grad_norm": 1.455832839012146,
-      "learning_rate": 6.14e-05,
-      "loss": 5.5269,
-      "step": 38921
-    },
-    {
-      "epoch": 20.298305084745763,
-      "grad_norm": 1.5848917961120605,
-      "learning_rate": 6.139899497487438e-05,
-      "loss": 5.6879,
-      "step": 38922
-    },
-    {
-      "epoch": 20.298826597131683,
-      "grad_norm": 1.508481502532959,
-      "learning_rate": 6.139798994974874e-05,
-      "loss": 5.1503,
-      "step": 38923
-    },
-    {
-      "epoch": 20.299348109517602,
-      "grad_norm": 1.6080601215362549,
-      "learning_rate": 6.139698492462311e-05,
-      "loss": 5.1727,
-      "step": 38924
-    },
-    {
-      "epoch": 20.29986962190352,
-      "grad_norm": 1.5506482124328613,
-      "learning_rate": 6.139597989949749e-05,
-      "loss": 5.2633,
-      "step": 38925
-    },
-    {
-      "epoch": 20.300391134289438,
-      "grad_norm": 1.4910155534744263,
-      "learning_rate": 6.139497487437186e-05,
-      "loss": 5.1452,
-      "step": 38926
-    },
-    {
-      "epoch": 20.300912646675357,
-      "grad_norm": 1.5613034963607788,
-      "learning_rate": 6.139396984924624e-05,
-      "loss": 5.0506,
-      "step": 38927
-    },
-    {
-      "epoch": 20.301434159061277,
-      "grad_norm": 1.5240859985351562,
-      "learning_rate": 6.13929648241206e-05,
-      "loss": 5.3936,
-      "step": 38928
-    },
-    {
-      "epoch": 20.301955671447196,
-      "grad_norm": 1.4174308776855469,
-      "learning_rate": 6.139195979899498e-05,
-      "loss": 5.5042,
-      "step": 38929
-    },
-    {
-      "epoch": 20.302477183833116,
-      "grad_norm": 1.3952689170837402,
-      "learning_rate": 6.139095477386935e-05,
-      "loss": 5.0879,
-      "step": 38930
-    },
-    {
-      "epoch": 20.302998696219035,
-      "grad_norm": 1.5198860168457031,
-      "learning_rate": 6.138994974874373e-05,
-      "loss": 5.1708,
-      "step": 38931
-    },
-    {
-      "epoch": 20.303520208604954,
-      "grad_norm": 1.3821953535079956,
-      "learning_rate": 6.138894472361809e-05,
-      "loss": 5.562,
-      "step": 38932
-    },
-    {
-      "epoch": 20.304041720990874,
-      "grad_norm": 1.582653522491455,
-      "learning_rate": 6.138793969849247e-05,
-      "loss": 5.0181,
-      "step": 38933
-    },
-    {
-      "epoch": 20.304563233376793,
-      "grad_norm": 1.4439626932144165,
-      "learning_rate": 6.138693467336683e-05,
-      "loss": 5.1071,
-      "step": 38934
-    },
-    {
-      "epoch": 20.305084745762713,
-      "grad_norm": 1.4683512449264526,
-      "learning_rate": 6.138592964824121e-05,
-      "loss": 5.4204,
-      "step": 38935
-    },
-    {
-      "epoch": 20.305606258148632,
-      "grad_norm": 1.5318756103515625,
-      "learning_rate": 6.138492462311557e-05,
-      "loss": 5.3111,
-      "step": 38936
-    },
-    {
-      "epoch": 20.30612777053455,
-      "grad_norm": 1.5385550260543823,
-      "learning_rate": 6.138391959798995e-05,
-      "loss": 5.1534,
-      "step": 38937
-    },
-    {
-      "epoch": 20.306649282920468,
-      "grad_norm": 1.6668226718902588,
-      "learning_rate": 6.138291457286433e-05,
-      "loss": 4.3553,
-      "step": 38938
-    },
-    {
-      "epoch": 20.307170795306387,
-      "grad_norm": 1.4483027458190918,
-      "learning_rate": 6.13819095477387e-05,
-      "loss": 5.2084,
-      "step": 38939
-    },
-    {
-      "epoch": 20.307692307692307,
-      "grad_norm": 1.515108585357666,
-      "learning_rate": 6.138090452261307e-05,
-      "loss": 5.097,
-      "step": 38940
-    },
-    {
-      "epoch": 20.308213820078226,
-      "grad_norm": 1.4388914108276367,
-      "learning_rate": 6.137989949748744e-05,
-      "loss": 5.4656,
-      "step": 38941
-    },
-    {
-      "epoch": 20.308735332464146,
-      "grad_norm": 1.5518642663955688,
-      "learning_rate": 6.137889447236181e-05,
-      "loss": 5.2271,
-      "step": 38942
-    },
-    {
-      "epoch": 20.309256844850065,
-      "grad_norm": 1.513202428817749,
-      "learning_rate": 6.137788944723618e-05,
-      "loss": 5.3544,
-      "step": 38943
-    },
-    {
-      "epoch": 20.309778357235984,
-      "grad_norm": 1.5441851615905762,
-      "learning_rate": 6.137688442211056e-05,
-      "loss": 5.1668,
-      "step": 38944
-    },
-    {
-      "epoch": 20.310299869621904,
-      "grad_norm": 1.4754576683044434,
-      "learning_rate": 6.137587939698492e-05,
-      "loss": 5.1113,
-      "step": 38945
-    },
-    {
-      "epoch": 20.310821382007823,
-      "grad_norm": 1.6175622940063477,
-      "learning_rate": 6.13748743718593e-05,
-      "loss": 5.3332,
-      "step": 38946
-    },
-    {
-      "epoch": 20.311342894393743,
-      "grad_norm": 1.5251471996307373,
-      "learning_rate": 6.137386934673368e-05,
-      "loss": 5.1744,
-      "step": 38947
-    },
-    {
-      "epoch": 20.311864406779662,
-      "grad_norm": 1.5198582410812378,
-      "learning_rate": 6.137286432160805e-05,
-      "loss": 5.5547,
-      "step": 38948
-    },
-    {
-      "epoch": 20.312385919165582,
-      "grad_norm": 1.625289797782898,
-      "learning_rate": 6.137185929648242e-05,
-      "loss": 5.0455,
-      "step": 38949
-    },
-    {
-      "epoch": 20.312907431551498,
-      "grad_norm": 1.5139297246932983,
-      "learning_rate": 6.137085427135678e-05,
-      "loss": 5.0525,
-      "step": 38950
-    },
-    {
-      "epoch": 20.313428943937417,
-      "grad_norm": 1.5706355571746826,
-      "learning_rate": 6.136984924623116e-05,
-      "loss": 5.4745,
-      "step": 38951
-    },
-    {
-      "epoch": 20.313950456323337,
-      "grad_norm": 1.5285108089447021,
-      "learning_rate": 6.136884422110552e-05,
-      "loss": 5.5993,
-      "step": 38952
-    },
-    {
-      "epoch": 20.314471968709256,
-      "grad_norm": 1.6407617330551147,
-      "learning_rate": 6.13678391959799e-05,
-      "loss": 4.7721,
-      "step": 38953
-    },
-    {
-      "epoch": 20.314993481095176,
-      "grad_norm": 1.5688549280166626,
-      "learning_rate": 6.136683417085427e-05,
-      "loss": 5.4783,
-      "step": 38954
-    },
-    {
-      "epoch": 20.315514993481095,
-      "grad_norm": 1.4947811365127563,
-      "learning_rate": 6.136582914572864e-05,
-      "loss": 5.6479,
-      "step": 38955
-    },
-    {
-      "epoch": 20.316036505867014,
-      "grad_norm": 1.586327075958252,
-      "learning_rate": 6.136482412060301e-05,
-      "loss": 4.9663,
-      "step": 38956
-    },
-    {
-      "epoch": 20.316558018252934,
-      "grad_norm": 1.5007834434509277,
-      "learning_rate": 6.136381909547739e-05,
-      "loss": 4.9901,
-      "step": 38957
-    },
-    {
-      "epoch": 20.317079530638853,
-      "grad_norm": 1.7143219709396362,
-      "learning_rate": 6.136281407035176e-05,
-      "loss": 5.022,
-      "step": 38958
-    },
-    {
-      "epoch": 20.317601043024773,
-      "grad_norm": 1.592516303062439,
-      "learning_rate": 6.136180904522614e-05,
-      "loss": 5.1819,
-      "step": 38959
-    },
-    {
-      "epoch": 20.318122555410692,
-      "grad_norm": 1.5869141817092896,
-      "learning_rate": 6.13608040201005e-05,
-      "loss": 5.4718,
-      "step": 38960
-    },
-    {
-      "epoch": 20.318644067796612,
-      "grad_norm": 1.4803764820098877,
-      "learning_rate": 6.135979899497488e-05,
-      "loss": 5.2283,
-      "step": 38961
-    },
-    {
-      "epoch": 20.319165580182528,
-      "grad_norm": 1.473483920097351,
-      "learning_rate": 6.135879396984925e-05,
-      "loss": 5.6238,
-      "step": 38962
-    },
-    {
-      "epoch": 20.319687092568447,
-      "grad_norm": 1.499606728553772,
-      "learning_rate": 6.135778894472363e-05,
-      "loss": 5.5183,
-      "step": 38963
-    },
-    {
-      "epoch": 20.320208604954367,
-      "grad_norm": 1.530145287513733,
-      "learning_rate": 6.135678391959799e-05,
-      "loss": 5.3769,
-      "step": 38964
-    },
-    {
-      "epoch": 20.320730117340286,
-      "grad_norm": 1.500069499015808,
-      "learning_rate": 6.135577889447236e-05,
-      "loss": 4.775,
-      "step": 38965
-    },
-    {
-      "epoch": 20.321251629726206,
-      "grad_norm": 1.3895155191421509,
-      "learning_rate": 6.135477386934673e-05,
-      "loss": 5.7833,
-      "step": 38966
-    },
-    {
-      "epoch": 20.321773142112125,
-      "grad_norm": 1.4355623722076416,
-      "learning_rate": 6.135376884422111e-05,
-      "loss": 5.4409,
-      "step": 38967
-    },
-    {
-      "epoch": 20.322294654498045,
-      "grad_norm": 1.5078169107437134,
-      "learning_rate": 6.135276381909549e-05,
-      "loss": 4.9519,
-      "step": 38968
-    },
-    {
-      "epoch": 20.322816166883964,
-      "grad_norm": 1.5116007328033447,
-      "learning_rate": 6.135175879396985e-05,
-      "loss": 5.4823,
-      "step": 38969
-    },
-    {
-      "epoch": 20.323337679269883,
-      "grad_norm": 1.439882516860962,
-      "learning_rate": 6.135075376884423e-05,
-      "loss": 5.598,
-      "step": 38970
-    },
-    {
-      "epoch": 20.323859191655803,
-      "grad_norm": 1.6008241176605225,
-      "learning_rate": 6.13497487437186e-05,
-      "loss": 5.4548,
-      "step": 38971
-    },
-    {
-      "epoch": 20.324380704041722,
-      "grad_norm": 1.4671337604522705,
-      "learning_rate": 6.134874371859297e-05,
-      "loss": 5.586,
-      "step": 38972
-    },
-    {
-      "epoch": 20.324902216427642,
-      "grad_norm": 1.577132225036621,
-      "learning_rate": 6.134773869346734e-05,
-      "loss": 4.9266,
-      "step": 38973
-    },
-    {
-      "epoch": 20.325423728813558,
-      "grad_norm": 1.432394027709961,
-      "learning_rate": 6.134673366834171e-05,
-      "loss": 5.1645,
-      "step": 38974
-    },
-    {
-      "epoch": 20.325945241199477,
-      "grad_norm": 1.6237038373947144,
-      "learning_rate": 6.134572864321608e-05,
-      "loss": 5.1524,
-      "step": 38975
-    },
-    {
-      "epoch": 20.326466753585397,
-      "grad_norm": 1.5865139961242676,
-      "learning_rate": 6.134472361809046e-05,
-      "loss": 5.1184,
-      "step": 38976
-    },
-    {
-      "epoch": 20.326988265971316,
-      "grad_norm": 1.4645291566848755,
-      "learning_rate": 6.134371859296482e-05,
-      "loss": 5.42,
-      "step": 38977
-    },
-    {
-      "epoch": 20.327509778357236,
-      "grad_norm": 1.4829438924789429,
-      "learning_rate": 6.13427135678392e-05,
-      "loss": 5.2008,
-      "step": 38978
-    },
-    {
-      "epoch": 20.328031290743155,
-      "grad_norm": 1.441131591796875,
-      "learning_rate": 6.134170854271358e-05,
-      "loss": 5.635,
-      "step": 38979
-    },
-    {
-      "epoch": 20.328552803129075,
-      "grad_norm": 1.4779366254806519,
-      "learning_rate": 6.134070351758794e-05,
-      "loss": 5.3405,
-      "step": 38980
-    },
-    {
-      "epoch": 20.329074315514994,
-      "grad_norm": 1.4987767934799194,
-      "learning_rate": 6.133969849246232e-05,
-      "loss": 5.0317,
-      "step": 38981
-    },
-    {
-      "epoch": 20.329595827900913,
-      "grad_norm": 1.5270270109176636,
-      "learning_rate": 6.133869346733668e-05,
-      "loss": 5.2158,
-      "step": 38982
-    },
-    {
-      "epoch": 20.330117340286833,
-      "grad_norm": 1.648970365524292,
-      "learning_rate": 6.133768844221106e-05,
-      "loss": 4.6947,
-      "step": 38983
-    },
-    {
-      "epoch": 20.330638852672752,
-      "grad_norm": 1.6118874549865723,
-      "learning_rate": 6.133668341708543e-05,
-      "loss": 5.3233,
-      "step": 38984
-    },
-    {
-      "epoch": 20.331160365058672,
-      "grad_norm": 1.5482470989227295,
-      "learning_rate": 6.13356783919598e-05,
-      "loss": 5.2165,
-      "step": 38985
-    },
-    {
-      "epoch": 20.331681877444588,
-      "grad_norm": 1.5853029489517212,
-      "learning_rate": 6.133467336683417e-05,
-      "loss": 4.8628,
-      "step": 38986
-    },
-    {
-      "epoch": 20.332203389830507,
-      "grad_norm": 1.5033226013183594,
-      "learning_rate": 6.133366834170855e-05,
-      "loss": 5.3854,
-      "step": 38987
-    },
-    {
-      "epoch": 20.332724902216427,
-      "grad_norm": 1.6188560724258423,
-      "learning_rate": 6.133266331658292e-05,
-      "loss": 5.2042,
-      "step": 38988
-    },
-    {
-      "epoch": 20.333246414602346,
-      "grad_norm": 1.6049602031707764,
-      "learning_rate": 6.13316582914573e-05,
-      "loss": 4.5603,
-      "step": 38989
-    },
-    {
-      "epoch": 20.333767926988266,
-      "grad_norm": 1.6667308807373047,
-      "learning_rate": 6.133065326633167e-05,
-      "loss": 5.0909,
-      "step": 38990
-    },
-    {
-      "epoch": 20.334289439374185,
-      "grad_norm": 1.532781958580017,
-      "learning_rate": 6.132964824120603e-05,
-      "loss": 5.115,
-      "step": 38991
-    },
-    {
-      "epoch": 20.334810951760105,
-      "grad_norm": 1.5039350986480713,
-      "learning_rate": 6.132864321608041e-05,
-      "loss": 5.416,
-      "step": 38992
-    },
-    {
-      "epoch": 20.335332464146024,
-      "grad_norm": 1.4996626377105713,
-      "learning_rate": 6.132763819095477e-05,
-      "loss": 5.6543,
-      "step": 38993
-    },
-    {
-      "epoch": 20.335853976531943,
-      "grad_norm": 1.4587947130203247,
-      "learning_rate": 6.132663316582915e-05,
-      "loss": 5.528,
-      "step": 38994
-    },
-    {
-      "epoch": 20.336375488917863,
-      "grad_norm": 1.6142001152038574,
-      "learning_rate": 6.132562814070351e-05,
-      "loss": 5.4709,
-      "step": 38995
-    },
-    {
-      "epoch": 20.336897001303782,
-      "grad_norm": 1.4943121671676636,
-      "learning_rate": 6.132462311557789e-05,
-      "loss": 5.0228,
-      "step": 38996
-    },
-    {
-      "epoch": 20.3374185136897,
-      "grad_norm": 1.60222589969635,
-      "learning_rate": 6.132361809045226e-05,
-      "loss": 5.3925,
-      "step": 38997
-    },
-    {
-      "epoch": 20.337940026075618,
-      "grad_norm": 1.4420368671417236,
-      "learning_rate": 6.132261306532663e-05,
-      "loss": 5.776,
-      "step": 38998
-    },
-    {
-      "epoch": 20.338461538461537,
-      "grad_norm": 1.5439374446868896,
-      "learning_rate": 6.132160804020101e-05,
-      "loss": 4.8956,
-      "step": 38999
-    },
-    {
-      "epoch": 20.338983050847457,
-      "grad_norm": 1.4922178983688354,
-      "learning_rate": 6.132060301507539e-05,
-      "loss": 5.2825,
-      "step": 39000
-    },
-    {
-      "epoch": 20.339504563233376,
-      "grad_norm": 1.4057302474975586,
-      "learning_rate": 6.131959798994975e-05,
-      "loss": 5.6373,
-      "step": 39001
-    },
-    {
-      "epoch": 20.340026075619296,
-      "grad_norm": 1.4534698724746704,
-      "learning_rate": 6.131859296482413e-05,
-      "loss": 5.5259,
-      "step": 39002
-    },
-    {
-      "epoch": 20.340547588005215,
-      "grad_norm": 1.5071537494659424,
-      "learning_rate": 6.13175879396985e-05,
-      "loss": 4.3066,
-      "step": 39003
-    },
-    {
-      "epoch": 20.341069100391135,
-      "grad_norm": 1.4628397226333618,
-      "learning_rate": 6.131658291457286e-05,
-      "loss": 5.1649,
-      "step": 39004
-    },
-    {
-      "epoch": 20.341590612777054,
-      "grad_norm": 1.5064796209335327,
-      "learning_rate": 6.131557788944724e-05,
-      "loss": 5.5567,
-      "step": 39005
-    },
-    {
-      "epoch": 20.342112125162974,
-      "grad_norm": 1.5753031969070435,
-      "learning_rate": 6.13145728643216e-05,
-      "loss": 5.135,
-      "step": 39006
-    },
-    {
-      "epoch": 20.342633637548893,
-      "grad_norm": 1.6645632982254028,
-      "learning_rate": 6.131356783919598e-05,
-      "loss": 4.3897,
-      "step": 39007
-    },
-    {
-      "epoch": 20.343155149934812,
-      "grad_norm": 1.5725023746490479,
-      "learning_rate": 6.131256281407036e-05,
-      "loss": 4.7589,
-      "step": 39008
-    },
-    {
-      "epoch": 20.343676662320732,
-      "grad_norm": 1.5553046464920044,
-      "learning_rate": 6.131155778894474e-05,
-      "loss": 4.8187,
-      "step": 39009
-    },
-    {
-      "epoch": 20.344198174706648,
-      "grad_norm": 1.551418662071228,
-      "learning_rate": 6.13105527638191e-05,
-      "loss": 5.4027,
-      "step": 39010
-    },
-    {
-      "epoch": 20.344719687092567,
-      "grad_norm": 1.5608811378479004,
-      "learning_rate": 6.130954773869348e-05,
-      "loss": 5.6389,
-      "step": 39011
-    },
-    {
-      "epoch": 20.345241199478487,
-      "grad_norm": 1.6240736246109009,
-      "learning_rate": 6.130854271356784e-05,
-      "loss": 5.0011,
-      "step": 39012
-    },
-    {
-      "epoch": 20.345762711864406,
-      "grad_norm": 1.516531229019165,
-      "learning_rate": 6.130753768844222e-05,
-      "loss": 5.3784,
-      "step": 39013
-    },
-    {
-      "epoch": 20.346284224250326,
-      "grad_norm": 1.5389937162399292,
-      "learning_rate": 6.130653266331658e-05,
-      "loss": 5.3206,
-      "step": 39014
-    },
-    {
-      "epoch": 20.346805736636245,
-      "grad_norm": 1.5237147808074951,
-      "learning_rate": 6.130552763819096e-05,
-      "loss": 5.5844,
-      "step": 39015
-    },
-    {
-      "epoch": 20.347327249022165,
-      "grad_norm": 1.4138357639312744,
-      "learning_rate": 6.130452261306533e-05,
-      "loss": 5.4417,
-      "step": 39016
-    },
-    {
-      "epoch": 20.347848761408084,
-      "grad_norm": 1.6032546758651733,
-      "learning_rate": 6.130351758793969e-05,
-      "loss": 5.1158,
-      "step": 39017
-    },
-    {
-      "epoch": 20.348370273794004,
-      "grad_norm": 1.590702772140503,
-      "learning_rate": 6.130251256281407e-05,
-      "loss": 4.1346,
-      "step": 39018
-    },
-    {
-      "epoch": 20.348891786179923,
-      "grad_norm": 1.5176955461502075,
-      "learning_rate": 6.130150753768845e-05,
-      "loss": 5.7603,
-      "step": 39019
-    },
-    {
-      "epoch": 20.349413298565842,
-      "grad_norm": 1.535244345664978,
-      "learning_rate": 6.130050251256282e-05,
-      "loss": 4.7776,
-      "step": 39020
-    },
-    {
-      "epoch": 20.34993481095176,
-      "grad_norm": 1.6550427675247192,
-      "learning_rate": 6.129949748743719e-05,
-      "loss": 5.3546,
-      "step": 39021
-    },
-    {
-      "epoch": 20.350456323337678,
-      "grad_norm": 1.5235251188278198,
-      "learning_rate": 6.129849246231157e-05,
-      "loss": 5.1901,
-      "step": 39022
-    },
-    {
-      "epoch": 20.350977835723597,
-      "grad_norm": 1.5626846551895142,
-      "learning_rate": 6.129748743718593e-05,
-      "loss": 5.0505,
-      "step": 39023
-    },
-    {
-      "epoch": 20.351499348109517,
-      "grad_norm": 1.5114418268203735,
-      "learning_rate": 6.129648241206031e-05,
-      "loss": 4.9701,
-      "step": 39024
-    },
-    {
-      "epoch": 20.352020860495436,
-      "grad_norm": 1.4964227676391602,
-      "learning_rate": 6.129547738693467e-05,
-      "loss": 5.4275,
-      "step": 39025
-    },
-    {
-      "epoch": 20.352542372881356,
-      "grad_norm": 1.5470404624938965,
-      "learning_rate": 6.129447236180905e-05,
-      "loss": 4.594,
-      "step": 39026
-    },
-    {
-      "epoch": 20.353063885267275,
-      "grad_norm": 1.5921827554702759,
-      "learning_rate": 6.129346733668341e-05,
-      "loss": 5.1422,
-      "step": 39027
-    },
-    {
-      "epoch": 20.353585397653195,
-      "grad_norm": 1.5512826442718506,
-      "learning_rate": 6.129246231155779e-05,
-      "loss": 5.1524,
-      "step": 39028
-    },
-    {
-      "epoch": 20.354106910039114,
-      "grad_norm": 1.3832217454910278,
-      "learning_rate": 6.129145728643217e-05,
-      "loss": 5.5142,
-      "step": 39029
-    },
-    {
-      "epoch": 20.354628422425034,
-      "grad_norm": 1.5491729974746704,
-      "learning_rate": 6.129045226130653e-05,
-      "loss": 5.1783,
-      "step": 39030
-    },
-    {
-      "epoch": 20.355149934810953,
-      "grad_norm": 1.5985569953918457,
-      "learning_rate": 6.128944723618091e-05,
-      "loss": 4.7055,
-      "step": 39031
-    },
-    {
-      "epoch": 20.355671447196872,
-      "grad_norm": 1.5520789623260498,
-      "learning_rate": 6.128844221105528e-05,
-      "loss": 5.1855,
-      "step": 39032
-    },
-    {
-      "epoch": 20.35619295958279,
-      "grad_norm": 1.5621240139007568,
-      "learning_rate": 6.128743718592965e-05,
-      "loss": 5.2747,
-      "step": 39033
-    },
-    {
-      "epoch": 20.356714471968708,
-      "grad_norm": 1.4997938871383667,
-      "learning_rate": 6.128643216080402e-05,
-      "loss": 5.4516,
-      "step": 39034
-    },
-    {
-      "epoch": 20.357235984354627,
-      "grad_norm": 1.6321650743484497,
-      "learning_rate": 6.12854271356784e-05,
-      "loss": 5.2169,
-      "step": 39035
-    },
-    {
-      "epoch": 20.357757496740547,
-      "grad_norm": 1.5056384801864624,
-      "learning_rate": 6.128442211055276e-05,
-      "loss": 5.3889,
-      "step": 39036
-    },
-    {
-      "epoch": 20.358279009126466,
-      "grad_norm": 1.5455660820007324,
-      "learning_rate": 6.128341708542714e-05,
-      "loss": 5.4873,
-      "step": 39037
-    },
-    {
-      "epoch": 20.358800521512386,
-      "grad_norm": 1.4223275184631348,
-      "learning_rate": 6.12824120603015e-05,
-      "loss": 5.2598,
-      "step": 39038
-    },
-    {
-      "epoch": 20.359322033898305,
-      "grad_norm": 1.7271310091018677,
-      "learning_rate": 6.128140703517588e-05,
-      "loss": 4.7592,
-      "step": 39039
-    },
-    {
-      "epoch": 20.359843546284225,
-      "grad_norm": 1.3792104721069336,
-      "learning_rate": 6.128040201005026e-05,
-      "loss": 5.232,
-      "step": 39040
-    },
-    {
-      "epoch": 20.360365058670144,
-      "grad_norm": 1.6100587844848633,
-      "learning_rate": 6.127939698492464e-05,
-      "loss": 5.3511,
-      "step": 39041
-    },
-    {
-      "epoch": 20.360886571056064,
-      "grad_norm": 1.6421093940734863,
-      "learning_rate": 6.1278391959799e-05,
-      "loss": 5.2991,
-      "step": 39042
-    },
-    {
-      "epoch": 20.361408083441983,
-      "grad_norm": 1.5531854629516602,
-      "learning_rate": 6.127738693467336e-05,
-      "loss": 4.9846,
-      "step": 39043
-    },
-    {
-      "epoch": 20.361929595827903,
-      "grad_norm": 1.509024739265442,
-      "learning_rate": 6.127638190954774e-05,
-      "loss": 5.3293,
-      "step": 39044
-    },
-    {
-      "epoch": 20.36245110821382,
-      "grad_norm": 1.599167823791504,
-      "learning_rate": 6.127537688442211e-05,
-      "loss": 5.2123,
-      "step": 39045
-    },
-    {
-      "epoch": 20.362972620599738,
-      "grad_norm": 1.4890978336334229,
-      "learning_rate": 6.127437185929648e-05,
-      "loss": 5.4733,
-      "step": 39046
-    },
-    {
-      "epoch": 20.363494132985657,
-      "grad_norm": 1.5485782623291016,
-      "learning_rate": 6.127336683417085e-05,
-      "loss": 5.1353,
-      "step": 39047
-    },
-    {
-      "epoch": 20.364015645371577,
-      "grad_norm": 1.521567940711975,
-      "learning_rate": 6.127236180904523e-05,
-      "loss": 5.0983,
-      "step": 39048
-    },
-    {
-      "epoch": 20.364537157757496,
-      "grad_norm": 1.465916633605957,
-      "learning_rate": 6.127135678391959e-05,
-      "loss": 5.4984,
-      "step": 39049
-    },
-    {
-      "epoch": 20.365058670143416,
-      "grad_norm": 1.5336171388626099,
-      "learning_rate": 6.127035175879397e-05,
-      "loss": 5.3191,
-      "step": 39050
-    },
-    {
-      "epoch": 20.365580182529335,
-      "grad_norm": 1.538618564605713,
-      "learning_rate": 6.126934673366835e-05,
-      "loss": 5.3631,
-      "step": 39051
-    },
-    {
-      "epoch": 20.366101694915255,
-      "grad_norm": 1.5858047008514404,
-      "learning_rate": 6.126834170854272e-05,
-      "loss": 4.9975,
-      "step": 39052
-    },
-    {
-      "epoch": 20.366623207301174,
-      "grad_norm": 1.5527147054672241,
-      "learning_rate": 6.126733668341709e-05,
-      "loss": 5.5239,
-      "step": 39053
-    },
-    {
-      "epoch": 20.367144719687094,
-      "grad_norm": 1.5079368352890015,
-      "learning_rate": 6.126633165829147e-05,
-      "loss": 5.5225,
-      "step": 39054
-    },
-    {
-      "epoch": 20.367666232073013,
-      "grad_norm": 1.5120962858200073,
-      "learning_rate": 6.126532663316583e-05,
-      "loss": 5.3256,
-      "step": 39055
-    },
-    {
-      "epoch": 20.368187744458933,
-      "grad_norm": 1.5087090730667114,
-      "learning_rate": 6.126432160804021e-05,
-      "loss": 5.3695,
-      "step": 39056
-    },
-    {
-      "epoch": 20.36870925684485,
-      "grad_norm": 1.5351253747940063,
-      "learning_rate": 6.126331658291457e-05,
-      "loss": 5.2667,
-      "step": 39057
-    },
-    {
-      "epoch": 20.369230769230768,
-      "grad_norm": 1.5603647232055664,
-      "learning_rate": 6.126231155778894e-05,
-      "loss": 5.12,
-      "step": 39058
-    },
-    {
-      "epoch": 20.369752281616687,
-      "grad_norm": 1.6095304489135742,
-      "learning_rate": 6.126130653266332e-05,
-      "loss": 4.5243,
-      "step": 39059
-    },
-    {
-      "epoch": 20.370273794002607,
-      "grad_norm": 1.4848781824111938,
-      "learning_rate": 6.126030150753769e-05,
-      "loss": 5.1777,
-      "step": 39060
-    },
-    {
-      "epoch": 20.370795306388526,
-      "grad_norm": 1.540753960609436,
-      "learning_rate": 6.125929648241207e-05,
-      "loss": 4.9288,
-      "step": 39061
-    },
-    {
-      "epoch": 20.371316818774446,
-      "grad_norm": 1.5512019395828247,
-      "learning_rate": 6.125829145728644e-05,
-      "loss": 4.8188,
-      "step": 39062
-    },
-    {
-      "epoch": 20.371838331160365,
-      "grad_norm": 1.5257257223129272,
-      "learning_rate": 6.125728643216081e-05,
-      "loss": 5.4349,
-      "step": 39063
-    },
-    {
-      "epoch": 20.372359843546285,
-      "grad_norm": 1.6056439876556396,
-      "learning_rate": 6.125628140703518e-05,
-      "loss": 5.3893,
-      "step": 39064
-    },
-    {
-      "epoch": 20.372881355932204,
-      "grad_norm": 1.5176252126693726,
-      "learning_rate": 6.125527638190956e-05,
-      "loss": 4.8913,
-      "step": 39065
-    },
-    {
-      "epoch": 20.373402868318124,
-      "grad_norm": 1.5902800559997559,
-      "learning_rate": 6.125427135678392e-05,
-      "loss": 5.3132,
-      "step": 39066
-    },
-    {
-      "epoch": 20.373924380704043,
-      "grad_norm": 1.456115484237671,
-      "learning_rate": 6.12532663316583e-05,
-      "loss": 5.5958,
-      "step": 39067
-    },
-    {
-      "epoch": 20.374445893089963,
-      "grad_norm": 1.4676082134246826,
-      "learning_rate": 6.125226130653266e-05,
-      "loss": 5.7208,
-      "step": 39068
-    },
-    {
-      "epoch": 20.37496740547588,
-      "grad_norm": 1.3680689334869385,
-      "learning_rate": 6.125125628140704e-05,
-      "loss": 4.8922,
-      "step": 39069
-    },
-    {
-      "epoch": 20.375488917861798,
-      "grad_norm": 1.550582766532898,
-      "learning_rate": 6.12502512562814e-05,
-      "loss": 4.938,
-      "step": 39070
-    },
-    {
-      "epoch": 20.376010430247717,
-      "grad_norm": 1.6184269189834595,
-      "learning_rate": 6.124924623115578e-05,
-      "loss": 5.2703,
-      "step": 39071
-    },
-    {
-      "epoch": 20.376531942633637,
-      "grad_norm": 1.5899096727371216,
-      "learning_rate": 6.124824120603016e-05,
-      "loss": 5.0748,
-      "step": 39072
-    },
-    {
-      "epoch": 20.377053455019556,
-      "grad_norm": 1.558464765548706,
-      "learning_rate": 6.124723618090452e-05,
-      "loss": 5.3616,
-      "step": 39073
-    },
-    {
-      "epoch": 20.377574967405476,
-      "grad_norm": 1.5973069667816162,
-      "learning_rate": 6.12462311557789e-05,
-      "loss": 4.4712,
-      "step": 39074
-    },
-    {
-      "epoch": 20.378096479791395,
-      "grad_norm": 1.5174379348754883,
-      "learning_rate": 6.124522613065327e-05,
-      "loss": 5.6013,
-      "step": 39075
-    },
-    {
-      "epoch": 20.378617992177315,
-      "grad_norm": 1.587668538093567,
-      "learning_rate": 6.124422110552764e-05,
-      "loss": 5.0132,
-      "step": 39076
-    },
-    {
-      "epoch": 20.379139504563234,
-      "grad_norm": 1.4951015710830688,
-      "learning_rate": 6.124321608040201e-05,
-      "loss": 5.1294,
-      "step": 39077
-    },
-    {
-      "epoch": 20.379661016949154,
-      "grad_norm": 1.501212239265442,
-      "learning_rate": 6.124221105527639e-05,
-      "loss": 5.2672,
-      "step": 39078
-    },
-    {
-      "epoch": 20.380182529335073,
-      "grad_norm": 1.4350157976150513,
-      "learning_rate": 6.124120603015075e-05,
-      "loss": 5.1364,
-      "step": 39079
-    },
-    {
-      "epoch": 20.380704041720993,
-      "grad_norm": 1.4404677152633667,
-      "learning_rate": 6.124020100502513e-05,
-      "loss": 5.6768,
-      "step": 39080
-    },
-    {
-      "epoch": 20.38122555410691,
-      "grad_norm": 1.4691946506500244,
-      "learning_rate": 6.12391959798995e-05,
-      "loss": 5.258,
-      "step": 39081
-    },
-    {
-      "epoch": 20.381747066492828,
-      "grad_norm": 1.592268943786621,
-      "learning_rate": 6.123819095477388e-05,
-      "loss": 5.3727,
-      "step": 39082
-    },
-    {
-      "epoch": 20.382268578878747,
-      "grad_norm": 1.5507586002349854,
-      "learning_rate": 6.123718592964825e-05,
-      "loss": 4.9015,
-      "step": 39083
-    },
-    {
-      "epoch": 20.382790091264667,
-      "grad_norm": 1.4289288520812988,
-      "learning_rate": 6.123618090452261e-05,
-      "loss": 5.2646,
-      "step": 39084
-    },
-    {
-      "epoch": 20.383311603650586,
-      "grad_norm": 1.566942811012268,
-      "learning_rate": 6.123517587939699e-05,
-      "loss": 5.1315,
-      "step": 39085
-    },
-    {
-      "epoch": 20.383833116036506,
-      "grad_norm": 1.4899202585220337,
-      "learning_rate": 6.123417085427135e-05,
-      "loss": 5.5673,
-      "step": 39086
-    },
-    {
-      "epoch": 20.384354628422425,
-      "grad_norm": 1.497159481048584,
-      "learning_rate": 6.123316582914573e-05,
-      "loss": 5.3705,
-      "step": 39087
-    },
-    {
-      "epoch": 20.384876140808345,
-      "grad_norm": 1.5891838073730469,
-      "learning_rate": 6.12321608040201e-05,
-      "loss": 5.0816,
-      "step": 39088
-    },
-    {
-      "epoch": 20.385397653194264,
-      "grad_norm": 1.6166093349456787,
-      "learning_rate": 6.123115577889447e-05,
-      "loss": 4.8056,
-      "step": 39089
-    },
-    {
-      "epoch": 20.385919165580184,
-      "grad_norm": 1.5258103609085083,
-      "learning_rate": 6.123015075376884e-05,
-      "loss": 4.975,
-      "step": 39090
-    },
-    {
-      "epoch": 20.386440677966103,
-      "grad_norm": 1.5827770233154297,
-      "learning_rate": 6.122914572864322e-05,
-      "loss": 5.3785,
-      "step": 39091
-    },
-    {
-      "epoch": 20.386962190352023,
-      "grad_norm": 1.5706799030303955,
-      "learning_rate": 6.12281407035176e-05,
-      "loss": 5.2484,
-      "step": 39092
-    },
-    {
-      "epoch": 20.38748370273794,
-      "grad_norm": 1.4847716093063354,
-      "learning_rate": 6.122713567839197e-05,
-      "loss": 5.3095,
-      "step": 39093
-    },
-    {
-      "epoch": 20.388005215123858,
-      "grad_norm": 1.5119279623031616,
-      "learning_rate": 6.122613065326634e-05,
-      "loss": 5.6387,
-      "step": 39094
-    },
-    {
-      "epoch": 20.388526727509777,
-      "grad_norm": 1.545487642288208,
-      "learning_rate": 6.122512562814071e-05,
-      "loss": 5.599,
-      "step": 39095
-    },
-    {
-      "epoch": 20.389048239895697,
-      "grad_norm": 1.476743221282959,
-      "learning_rate": 6.122412060301508e-05,
-      "loss": 5.6704,
-      "step": 39096
-    },
-    {
-      "epoch": 20.389569752281616,
-      "grad_norm": 1.4909508228302002,
-      "learning_rate": 6.122311557788944e-05,
-      "loss": 5.2748,
-      "step": 39097
-    },
-    {
-      "epoch": 20.390091264667536,
-      "grad_norm": 1.4329723119735718,
-      "learning_rate": 6.122211055276382e-05,
-      "loss": 5.4032,
-      "step": 39098
-    },
-    {
-      "epoch": 20.390612777053455,
-      "grad_norm": 1.5163997411727905,
-      "learning_rate": 6.122110552763818e-05,
-      "loss": 5.1672,
-      "step": 39099
-    },
-    {
-      "epoch": 20.391134289439375,
-      "grad_norm": 1.5240025520324707,
-      "learning_rate": 6.122010050251256e-05,
-      "loss": 5.441,
-      "step": 39100
-    },
-    {
-      "epoch": 20.391655801825294,
-      "grad_norm": 1.5419052839279175,
-      "learning_rate": 6.121909547738694e-05,
-      "loss": 5.3711,
-      "step": 39101
-    },
-    {
-      "epoch": 20.392177314211214,
-      "grad_norm": 1.4940335750579834,
-      "learning_rate": 6.121809045226132e-05,
-      "loss": 5.3362,
-      "step": 39102
-    },
-    {
-      "epoch": 20.392698826597133,
-      "grad_norm": 1.5756571292877197,
-      "learning_rate": 6.121708542713568e-05,
-      "loss": 4.5446,
-      "step": 39103
-    },
-    {
-      "epoch": 20.39322033898305,
-      "grad_norm": 1.5270276069641113,
-      "learning_rate": 6.121608040201006e-05,
-      "loss": 5.0847,
-      "step": 39104
-    },
-    {
-      "epoch": 20.39374185136897,
-      "grad_norm": 1.548668384552002,
-      "learning_rate": 6.121507537688442e-05,
-      "loss": 5.3695,
-      "step": 39105
-    },
-    {
-      "epoch": 20.394263363754888,
-      "grad_norm": 1.4212546348571777,
-      "learning_rate": 6.12140703517588e-05,
-      "loss": 5.3153,
-      "step": 39106
-    },
-    {
-      "epoch": 20.394784876140807,
-      "grad_norm": 1.5117441415786743,
-      "learning_rate": 6.121306532663317e-05,
-      "loss": 5.4751,
-      "step": 39107
-    },
-    {
-      "epoch": 20.395306388526727,
-      "grad_norm": 1.365805983543396,
-      "learning_rate": 6.121206030150754e-05,
-      "loss": 5.8326,
-      "step": 39108
-    },
-    {
-      "epoch": 20.395827900912646,
-      "grad_norm": 1.5644761323928833,
-      "learning_rate": 6.121105527638191e-05,
-      "loss": 4.9445,
-      "step": 39109
-    },
-    {
-      "epoch": 20.396349413298566,
-      "grad_norm": 1.5726583003997803,
-      "learning_rate": 6.121005025125627e-05,
-      "loss": 5.2832,
-      "step": 39110
-    },
-    {
-      "epoch": 20.396870925684485,
-      "grad_norm": 1.4119763374328613,
-      "learning_rate": 6.120904522613065e-05,
-      "loss": 5.6537,
-      "step": 39111
-    },
-    {
-      "epoch": 20.397392438070405,
-      "grad_norm": 1.4441804885864258,
-      "learning_rate": 6.120804020100503e-05,
-      "loss": 5.4232,
-      "step": 39112
-    },
-    {
-      "epoch": 20.397913950456324,
-      "grad_norm": 1.5257585048675537,
-      "learning_rate": 6.12070351758794e-05,
-      "loss": 4.9246,
-      "step": 39113
-    },
-    {
-      "epoch": 20.398435462842244,
-      "grad_norm": 1.6661722660064697,
-      "learning_rate": 6.120603015075377e-05,
-      "loss": 5.1367,
-      "step": 39114
-    },
-    {
-      "epoch": 20.398956975228163,
-      "grad_norm": 1.5135685205459595,
-      "learning_rate": 6.120502512562815e-05,
-      "loss": 5.0041,
-      "step": 39115
-    },
-    {
-      "epoch": 20.39947848761408,
-      "grad_norm": 1.5521724224090576,
-      "learning_rate": 6.120402010050251e-05,
-      "loss": 5.6358,
-      "step": 39116
-    },
-    {
-      "epoch": 20.4,
-      "grad_norm": 1.4465432167053223,
-      "learning_rate": 6.120301507537689e-05,
-      "loss": 5.8501,
-      "step": 39117
-    },
-    {
-      "epoch": 20.400521512385918,
-      "grad_norm": 1.5845431089401245,
-      "learning_rate": 6.120201005025125e-05,
-      "loss": 5.306,
-      "step": 39118
-    },
-    {
-      "epoch": 20.401043024771838,
-      "grad_norm": 1.5388000011444092,
-      "learning_rate": 6.120100502512563e-05,
-      "loss": 5.2636,
-      "step": 39119
-    },
-    {
-      "epoch": 20.401564537157757,
-      "grad_norm": 1.568698763847351,
-      "learning_rate": 6.12e-05,
-      "loss": 5.0672,
-      "step": 39120
-    },
-    {
-      "epoch": 20.402086049543676,
-      "grad_norm": 1.642250418663025,
-      "learning_rate": 6.119899497487437e-05,
-      "loss": 5.0477,
-      "step": 39121
-    },
-    {
-      "epoch": 20.402607561929596,
-      "grad_norm": 1.4970066547393799,
-      "learning_rate": 6.119798994974875e-05,
-      "loss": 5.7093,
-      "step": 39122
-    },
-    {
-      "epoch": 20.403129074315515,
-      "grad_norm": 1.4126724004745483,
-      "learning_rate": 6.119698492462312e-05,
-      "loss": 5.501,
-      "step": 39123
-    },
-    {
-      "epoch": 20.403650586701435,
-      "grad_norm": 1.59050714969635,
-      "learning_rate": 6.11959798994975e-05,
-      "loss": 4.9626,
-      "step": 39124
-    },
-    {
-      "epoch": 20.404172099087354,
-      "grad_norm": 1.5347026586532593,
-      "learning_rate": 6.119497487437186e-05,
-      "loss": 5.2787,
-      "step": 39125
-    },
-    {
-      "epoch": 20.404693611473274,
-      "grad_norm": 1.6729317903518677,
-      "learning_rate": 6.119396984924624e-05,
-      "loss": 5.2993,
-      "step": 39126
-    },
-    {
-      "epoch": 20.405215123859193,
-      "grad_norm": 1.5412582159042358,
-      "learning_rate": 6.11929648241206e-05,
-      "loss": 5.4915,
-      "step": 39127
-    },
-    {
-      "epoch": 20.40573663624511,
-      "grad_norm": 1.4775314331054688,
-      "learning_rate": 6.119195979899498e-05,
-      "loss": 5.1497,
-      "step": 39128
-    },
-    {
-      "epoch": 20.40625814863103,
-      "grad_norm": 1.5748220682144165,
-      "learning_rate": 6.119095477386934e-05,
-      "loss": 5.094,
-      "step": 39129
-    },
-    {
-      "epoch": 20.406779661016948,
-      "grad_norm": 1.4944684505462646,
-      "learning_rate": 6.118994974874372e-05,
-      "loss": 5.6307,
-      "step": 39130
-    },
-    {
-      "epoch": 20.407301173402868,
-      "grad_norm": 1.4201107025146484,
-      "learning_rate": 6.118894472361809e-05,
-      "loss": 5.7429,
-      "step": 39131
-    },
-    {
-      "epoch": 20.407822685788787,
-      "grad_norm": 1.566638469696045,
-      "learning_rate": 6.118793969849246e-05,
-      "loss": 5.1446,
-      "step": 39132
-    },
-    {
-      "epoch": 20.408344198174706,
-      "grad_norm": 1.4989084005355835,
-      "learning_rate": 6.118693467336684e-05,
-      "loss": 5.0668,
-      "step": 39133
-    },
-    {
-      "epoch": 20.408865710560626,
-      "grad_norm": 1.566967248916626,
-      "learning_rate": 6.118592964824122e-05,
-      "loss": 5.4076,
-      "step": 39134
-    },
-    {
-      "epoch": 20.409387222946545,
-      "grad_norm": 1.6560512781143188,
-      "learning_rate": 6.118492462311558e-05,
-      "loss": 4.9024,
-      "step": 39135
-    },
-    {
-      "epoch": 20.409908735332465,
-      "grad_norm": 1.5748809576034546,
-      "learning_rate": 6.118391959798996e-05,
-      "loss": 5.2771,
-      "step": 39136
-    },
-    {
-      "epoch": 20.410430247718384,
-      "grad_norm": 1.5949653387069702,
-      "learning_rate": 6.118291457286433e-05,
-      "loss": 5.4099,
-      "step": 39137
-    },
-    {
-      "epoch": 20.410951760104304,
-      "grad_norm": 1.6597089767456055,
-      "learning_rate": 6.118190954773869e-05,
-      "loss": 5.3435,
-      "step": 39138
-    },
-    {
-      "epoch": 20.411473272490223,
-      "grad_norm": 1.569689154624939,
-      "learning_rate": 6.118090452261307e-05,
-      "loss": 5.3263,
-      "step": 39139
-    },
-    {
-      "epoch": 20.41199478487614,
-      "grad_norm": 1.5590767860412598,
-      "learning_rate": 6.117989949748743e-05,
-      "loss": 5.3222,
-      "step": 39140
-    },
-    {
-      "epoch": 20.41251629726206,
-      "grad_norm": 1.573208212852478,
-      "learning_rate": 6.117889447236181e-05,
-      "loss": 5.2463,
-      "step": 39141
-    },
-    {
-      "epoch": 20.413037809647978,
-      "grad_norm": 1.6863559484481812,
-      "learning_rate": 6.117788944723619e-05,
-      "loss": 5.4989,
-      "step": 39142
-    },
-    {
-      "epoch": 20.413559322033898,
-      "grad_norm": 1.6088703870773315,
-      "learning_rate": 6.117688442211056e-05,
-      "loss": 5.3232,
-      "step": 39143
-    },
-    {
-      "epoch": 20.414080834419817,
-      "grad_norm": 1.528210997581482,
-      "learning_rate": 6.117587939698493e-05,
-      "loss": 5.6033,
-      "step": 39144
-    },
-    {
-      "epoch": 20.414602346805736,
-      "grad_norm": 1.6275591850280762,
-      "learning_rate": 6.117487437185931e-05,
-      "loss": 5.844,
-      "step": 39145
-    },
-    {
-      "epoch": 20.415123859191656,
-      "grad_norm": 1.556462287902832,
-      "learning_rate": 6.117386934673367e-05,
-      "loss": 4.6781,
-      "step": 39146
-    },
-    {
-      "epoch": 20.415645371577575,
-      "grad_norm": 1.5109871625900269,
-      "learning_rate": 6.117286432160805e-05,
-      "loss": 5.4118,
-      "step": 39147
-    },
-    {
-      "epoch": 20.416166883963495,
-      "grad_norm": 1.5809930562973022,
-      "learning_rate": 6.117185929648241e-05,
-      "loss": 4.8862,
-      "step": 39148
-    },
-    {
-      "epoch": 20.416688396349414,
-      "grad_norm": 1.5480096340179443,
-      "learning_rate": 6.117085427135679e-05,
-      "loss": 5.2902,
-      "step": 39149
-    },
-    {
-      "epoch": 20.417209908735334,
-      "grad_norm": 1.5376540422439575,
-      "learning_rate": 6.116984924623116e-05,
-      "loss": 5.4216,
-      "step": 39150
-    },
-    {
-      "epoch": 20.417731421121253,
-      "grad_norm": 1.6855875253677368,
-      "learning_rate": 6.116884422110552e-05,
-      "loss": 4.5425,
-      "step": 39151
-    },
-    {
-      "epoch": 20.41825293350717,
-      "grad_norm": 1.6210318803787231,
-      "learning_rate": 6.11678391959799e-05,
-      "loss": 5.0848,
-      "step": 39152
-    },
-    {
-      "epoch": 20.41877444589309,
-      "grad_norm": 1.6392858028411865,
-      "learning_rate": 6.116683417085428e-05,
-      "loss": 4.9807,
-      "step": 39153
-    },
-    {
-      "epoch": 20.419295958279008,
-      "grad_norm": 1.653232455253601,
-      "learning_rate": 6.116582914572865e-05,
-      "loss": 5.2879,
-      "step": 39154
-    },
-    {
-      "epoch": 20.419817470664928,
-      "grad_norm": 1.643033742904663,
-      "learning_rate": 6.116482412060302e-05,
-      "loss": 5.1741,
-      "step": 39155
-    },
-    {
-      "epoch": 20.420338983050847,
-      "grad_norm": 1.5167585611343384,
-      "learning_rate": 6.11638190954774e-05,
-      "loss": 5.6661,
-      "step": 39156
-    },
-    {
-      "epoch": 20.420860495436767,
-      "grad_norm": 1.5611923933029175,
-      "learning_rate": 6.116281407035176e-05,
-      "loss": 5.4721,
-      "step": 39157
-    },
-    {
-      "epoch": 20.421382007822686,
-      "grad_norm": 1.5772531032562256,
-      "learning_rate": 6.116180904522614e-05,
-      "loss": 5.3757,
-      "step": 39158
-    },
-    {
-      "epoch": 20.421903520208605,
-      "grad_norm": 1.5584921836853027,
-      "learning_rate": 6.11608040201005e-05,
-      "loss": 4.846,
-      "step": 39159
-    },
-    {
-      "epoch": 20.422425032594525,
-      "grad_norm": 1.5448660850524902,
-      "learning_rate": 6.115979899497488e-05,
-      "loss": 5.0396,
-      "step": 39160
-    },
-    {
-      "epoch": 20.422946544980444,
-      "grad_norm": 1.5479423999786377,
-      "learning_rate": 6.115879396984924e-05,
-      "loss": 5.1938,
-      "step": 39161
-    },
-    {
-      "epoch": 20.423468057366364,
-      "grad_norm": 1.4870119094848633,
-      "learning_rate": 6.115778894472362e-05,
-      "loss": 5.3273,
-      "step": 39162
-    },
-    {
-      "epoch": 20.423989569752283,
-      "grad_norm": 1.557992935180664,
-      "learning_rate": 6.1156783919598e-05,
-      "loss": 5.3715,
-      "step": 39163
-    },
-    {
-      "epoch": 20.4245110821382,
-      "grad_norm": 1.4583230018615723,
-      "learning_rate": 6.115577889447236e-05,
-      "loss": 5.34,
-      "step": 39164
-    },
-    {
-      "epoch": 20.42503259452412,
-      "grad_norm": 1.5566149950027466,
-      "learning_rate": 6.115477386934674e-05,
-      "loss": 5.1713,
-      "step": 39165
-    },
-    {
-      "epoch": 20.425554106910038,
-      "grad_norm": 1.4666873216629028,
-      "learning_rate": 6.11537688442211e-05,
-      "loss": 5.3737,
-      "step": 39166
-    },
-    {
-      "epoch": 20.426075619295958,
-      "grad_norm": 1.5978419780731201,
-      "learning_rate": 6.115276381909548e-05,
-      "loss": 4.9386,
-      "step": 39167
-    },
-    {
-      "epoch": 20.426597131681877,
-      "grad_norm": 1.451133370399475,
-      "learning_rate": 6.115175879396985e-05,
-      "loss": 5.7342,
-      "step": 39168
-    },
-    {
-      "epoch": 20.427118644067797,
-      "grad_norm": 1.6309210062026978,
-      "learning_rate": 6.115075376884423e-05,
-      "loss": 4.6965,
-      "step": 39169
-    },
-    {
-      "epoch": 20.427640156453716,
-      "grad_norm": 1.4942840337753296,
-      "learning_rate": 6.114974874371859e-05,
-      "loss": 5.468,
-      "step": 39170
-    },
-    {
-      "epoch": 20.428161668839635,
-      "grad_norm": 1.4919672012329102,
-      "learning_rate": 6.114874371859297e-05,
-      "loss": 5.5374,
-      "step": 39171
-    },
-    {
-      "epoch": 20.428683181225555,
-      "grad_norm": 1.4606890678405762,
-      "learning_rate": 6.114773869346733e-05,
-      "loss": 5.3868,
-      "step": 39172
-    },
-    {
-      "epoch": 20.429204693611474,
-      "grad_norm": 1.445459246635437,
-      "learning_rate": 6.114673366834171e-05,
-      "loss": 5.749,
-      "step": 39173
-    },
-    {
-      "epoch": 20.429726205997394,
-      "grad_norm": 1.6059123277664185,
-      "learning_rate": 6.114572864321609e-05,
-      "loss": 5.2537,
-      "step": 39174
-    },
-    {
-      "epoch": 20.430247718383313,
-      "grad_norm": 1.61715567111969,
-      "learning_rate": 6.114472361809047e-05,
-      "loss": 5.348,
-      "step": 39175
-    },
-    {
-      "epoch": 20.43076923076923,
-      "grad_norm": 1.5249617099761963,
-      "learning_rate": 6.114371859296483e-05,
-      "loss": 5.587,
-      "step": 39176
-    },
-    {
-      "epoch": 20.43129074315515,
-      "grad_norm": 1.499993085861206,
-      "learning_rate": 6.11427135678392e-05,
-      "loss": 5.5058,
-      "step": 39177
-    },
-    {
-      "epoch": 20.431812255541068,
-      "grad_norm": 1.5570179224014282,
-      "learning_rate": 6.114170854271357e-05,
-      "loss": 5.6777,
-      "step": 39178
-    },
-    {
-      "epoch": 20.432333767926988,
-      "grad_norm": 1.6335922479629517,
-      "learning_rate": 6.114070351758794e-05,
-      "loss": 5.1646,
-      "step": 39179
-    },
-    {
-      "epoch": 20.432855280312907,
-      "grad_norm": 1.5311355590820312,
-      "learning_rate": 6.113969849246231e-05,
-      "loss": 5.4258,
-      "step": 39180
-    },
-    {
-      "epoch": 20.433376792698827,
-      "grad_norm": 1.510166049003601,
-      "learning_rate": 6.113869346733668e-05,
-      "loss": 5.1941,
-      "step": 39181
-    },
-    {
-      "epoch": 20.433898305084746,
-      "grad_norm": 1.691862940788269,
-      "learning_rate": 6.113768844221106e-05,
-      "loss": 4.5655,
-      "step": 39182
-    },
-    {
-      "epoch": 20.434419817470665,
-      "grad_norm": 1.5858371257781982,
-      "learning_rate": 6.113668341708543e-05,
-      "loss": 5.4818,
-      "step": 39183
-    },
-    {
-      "epoch": 20.434941329856585,
-      "grad_norm": 1.6371864080429077,
-      "learning_rate": 6.113567839195981e-05,
-      "loss": 4.891,
-      "step": 39184
-    },
-    {
-      "epoch": 20.435462842242504,
-      "grad_norm": 1.647875189781189,
-      "learning_rate": 6.113467336683418e-05,
-      "loss": 5.0171,
-      "step": 39185
-    },
-    {
-      "epoch": 20.435984354628424,
-      "grad_norm": 1.6351635456085205,
-      "learning_rate": 6.113366834170855e-05,
-      "loss": 5.2249,
-      "step": 39186
-    },
-    {
-      "epoch": 20.43650586701434,
-      "grad_norm": 1.546930193901062,
-      "learning_rate": 6.113266331658292e-05,
-      "loss": 5.0325,
-      "step": 39187
-    },
-    {
-      "epoch": 20.43702737940026,
-      "grad_norm": 1.6765334606170654,
-      "learning_rate": 6.11316582914573e-05,
-      "loss": 5.2269,
-      "step": 39188
-    },
-    {
-      "epoch": 20.43754889178618,
-      "grad_norm": 1.4339665174484253,
-      "learning_rate": 6.113065326633166e-05,
-      "loss": 5.4486,
-      "step": 39189
-    },
-    {
-      "epoch": 20.438070404172098,
-      "grad_norm": 1.5157150030136108,
-      "learning_rate": 6.112964824120602e-05,
-      "loss": 5.4449,
-      "step": 39190
-    },
-    {
-      "epoch": 20.438591916558018,
-      "grad_norm": 1.4742637872695923,
-      "learning_rate": 6.11286432160804e-05,
-      "loss": 5.3775,
-      "step": 39191
-    },
-    {
-      "epoch": 20.439113428943937,
-      "grad_norm": 1.5768489837646484,
-      "learning_rate": 6.112763819095477e-05,
-      "loss": 5.3375,
-      "step": 39192
-    },
-    {
-      "epoch": 20.439634941329857,
-      "grad_norm": 1.5502500534057617,
-      "learning_rate": 6.112663316582914e-05,
-      "loss": 5.3344,
-      "step": 39193
-    },
-    {
-      "epoch": 20.440156453715776,
-      "grad_norm": 1.462510585784912,
-      "learning_rate": 6.112562814070352e-05,
-      "loss": 5.2455,
-      "step": 39194
-    },
-    {
-      "epoch": 20.440677966101696,
-      "grad_norm": 1.4972797632217407,
-      "learning_rate": 6.11246231155779e-05,
-      "loss": 5.4439,
-      "step": 39195
-    },
-    {
-      "epoch": 20.441199478487615,
-      "grad_norm": 1.4721014499664307,
-      "learning_rate": 6.112361809045226e-05,
-      "loss": 5.4188,
-      "step": 39196
-    },
-    {
-      "epoch": 20.441720990873534,
-      "grad_norm": 1.483281135559082,
-      "learning_rate": 6.112261306532664e-05,
-      "loss": 5.4233,
-      "step": 39197
-    },
-    {
-      "epoch": 20.442242503259454,
-      "grad_norm": 1.6059657335281372,
-      "learning_rate": 6.1121608040201e-05,
-      "loss": 5.0426,
-      "step": 39198
-    },
-    {
-      "epoch": 20.442764015645373,
-      "grad_norm": 1.4513877630233765,
-      "learning_rate": 6.112060301507538e-05,
-      "loss": 5.7841,
-      "step": 39199
-    },
-    {
-      "epoch": 20.44328552803129,
-      "grad_norm": 1.502186894416809,
-      "learning_rate": 6.111959798994975e-05,
-      "loss": 5.2735,
-      "step": 39200
-    },
-    {
-      "epoch": 20.44380704041721,
-      "grad_norm": 1.637000560760498,
-      "learning_rate": 6.111859296482413e-05,
-      "loss": 4.9812,
-      "step": 39201
-    },
-    {
-      "epoch": 20.444328552803128,
-      "grad_norm": 1.4670974016189575,
-      "learning_rate": 6.111758793969849e-05,
-      "loss": 5.4211,
-      "step": 39202
-    },
-    {
-      "epoch": 20.444850065189048,
-      "grad_norm": 1.5129468441009521,
-      "learning_rate": 6.111658291457287e-05,
-      "loss": 5.2644,
-      "step": 39203
-    },
-    {
-      "epoch": 20.445371577574967,
-      "grad_norm": 1.4795082807540894,
-      "learning_rate": 6.111557788944725e-05,
-      "loss": 5.5089,
-      "step": 39204
-    },
-    {
-      "epoch": 20.445893089960887,
-      "grad_norm": 1.4679921865463257,
-      "learning_rate": 6.111457286432161e-05,
-      "loss": 5.3141,
-      "step": 39205
-    },
-    {
-      "epoch": 20.446414602346806,
-      "grad_norm": 1.5355230569839478,
-      "learning_rate": 6.111356783919599e-05,
-      "loss": 5.6228,
-      "step": 39206
-    },
-    {
-      "epoch": 20.446936114732726,
-      "grad_norm": 1.4816838502883911,
-      "learning_rate": 6.111256281407035e-05,
-      "loss": 5.1623,
-      "step": 39207
-    },
-    {
-      "epoch": 20.447457627118645,
-      "grad_norm": 1.514768362045288,
-      "learning_rate": 6.111155778894473e-05,
-      "loss": 5.2152,
-      "step": 39208
-    },
-    {
-      "epoch": 20.447979139504564,
-      "grad_norm": 1.4594944715499878,
-      "learning_rate": 6.11105527638191e-05,
-      "loss": 5.3124,
-      "step": 39209
-    },
-    {
-      "epoch": 20.448500651890484,
-      "grad_norm": 1.6265099048614502,
-      "learning_rate": 6.110954773869347e-05,
-      "loss": 5.1866,
-      "step": 39210
-    },
-    {
-      "epoch": 20.4490221642764,
-      "grad_norm": 1.4618339538574219,
-      "learning_rate": 6.110854271356784e-05,
-      "loss": 4.6618,
-      "step": 39211
-    },
-    {
-      "epoch": 20.44954367666232,
-      "grad_norm": 1.4374983310699463,
-      "learning_rate": 6.110753768844221e-05,
-      "loss": 5.4524,
-      "step": 39212
-    },
-    {
-      "epoch": 20.45006518904824,
-      "grad_norm": 1.4458436965942383,
-      "learning_rate": 6.110653266331658e-05,
-      "loss": 5.4549,
-      "step": 39213
-    },
-    {
-      "epoch": 20.45058670143416,
-      "grad_norm": 1.5121983289718628,
-      "learning_rate": 6.110552763819096e-05,
-      "loss": 5.4661,
-      "step": 39214
-    },
-    {
-      "epoch": 20.451108213820078,
-      "grad_norm": 1.4858014583587646,
-      "learning_rate": 6.110452261306533e-05,
-      "loss": 5.1677,
-      "step": 39215
-    },
-    {
-      "epoch": 20.451629726205997,
-      "grad_norm": 1.5144952535629272,
-      "learning_rate": 6.11035175879397e-05,
-      "loss": 5.1272,
-      "step": 39216
-    },
-    {
-      "epoch": 20.452151238591917,
-      "grad_norm": 1.5353615283966064,
-      "learning_rate": 6.110251256281408e-05,
-      "loss": 5.3849,
-      "step": 39217
-    },
-    {
-      "epoch": 20.452672750977836,
-      "grad_norm": 1.5405497550964355,
-      "learning_rate": 6.110150753768844e-05,
-      "loss": 5.0416,
-      "step": 39218
-    },
-    {
-      "epoch": 20.453194263363756,
-      "grad_norm": 1.659932017326355,
-      "learning_rate": 6.110050251256282e-05,
-      "loss": 5.146,
-      "step": 39219
-    },
-    {
-      "epoch": 20.453715775749675,
-      "grad_norm": 1.574473261833191,
-      "learning_rate": 6.109949748743718e-05,
-      "loss": 5.3654,
-      "step": 39220
-    },
-    {
-      "epoch": 20.454237288135594,
-      "grad_norm": 1.5099859237670898,
-      "learning_rate": 6.109849246231156e-05,
-      "loss": 5.415,
-      "step": 39221
-    },
-    {
-      "epoch": 20.454758800521514,
-      "grad_norm": 1.4849106073379517,
-      "learning_rate": 6.109748743718593e-05,
-      "loss": 5.4117,
-      "step": 39222
-    },
-    {
-      "epoch": 20.45528031290743,
-      "grad_norm": 1.4712542295455933,
-      "learning_rate": 6.10964824120603e-05,
-      "loss": 5.3523,
-      "step": 39223
-    },
-    {
-      "epoch": 20.45580182529335,
-      "grad_norm": 1.6596806049346924,
-      "learning_rate": 6.109547738693467e-05,
-      "loss": 4.9182,
-      "step": 39224
-    },
-    {
-      "epoch": 20.45632333767927,
-      "grad_norm": 1.6348894834518433,
-      "learning_rate": 6.109447236180905e-05,
-      "loss": 4.8502,
-      "step": 39225
-    },
-    {
-      "epoch": 20.45684485006519,
-      "grad_norm": 1.6819812059402466,
-      "learning_rate": 6.109346733668342e-05,
-      "loss": 4.9856,
-      "step": 39226
-    },
-    {
-      "epoch": 20.457366362451108,
-      "grad_norm": 1.464590072631836,
-      "learning_rate": 6.10924623115578e-05,
-      "loss": 5.7342,
-      "step": 39227
-    },
-    {
-      "epoch": 20.457887874837027,
-      "grad_norm": 1.4976483583450317,
-      "learning_rate": 6.109145728643217e-05,
-      "loss": 5.1473,
-      "step": 39228
-    },
-    {
-      "epoch": 20.458409387222947,
-      "grad_norm": 1.4871894121170044,
-      "learning_rate": 6.109045226130654e-05,
-      "loss": 5.7313,
-      "step": 39229
-    },
-    {
-      "epoch": 20.458930899608866,
-      "grad_norm": 1.6520514488220215,
-      "learning_rate": 6.108944723618091e-05,
-      "loss": 5.283,
-      "step": 39230
-    },
-    {
-      "epoch": 20.459452411994786,
-      "grad_norm": 1.5655161142349243,
-      "learning_rate": 6.108844221105527e-05,
-      "loss": 5.2381,
-      "step": 39231
-    },
-    {
-      "epoch": 20.459973924380705,
-      "grad_norm": 1.5517462491989136,
-      "learning_rate": 6.108743718592965e-05,
-      "loss": 4.9867,
-      "step": 39232
-    },
-    {
-      "epoch": 20.460495436766625,
-      "grad_norm": 1.5415490865707397,
-      "learning_rate": 6.108643216080401e-05,
-      "loss": 5.505,
-      "step": 39233
-    },
-    {
-      "epoch": 20.461016949152544,
-      "grad_norm": 1.611360788345337,
-      "learning_rate": 6.108542713567839e-05,
-      "loss": 5.1393,
-      "step": 39234
-    },
-    {
-      "epoch": 20.46153846153846,
-      "grad_norm": 1.5025660991668701,
-      "learning_rate": 6.108442211055277e-05,
-      "loss": 5.4201,
-      "step": 39235
-    },
-    {
-      "epoch": 20.46205997392438,
-      "grad_norm": 1.4526742696762085,
-      "learning_rate": 6.108341708542715e-05,
-      "loss": 5.4098,
-      "step": 39236
-    },
-    {
-      "epoch": 20.4625814863103,
-      "grad_norm": 1.4513765573501587,
-      "learning_rate": 6.108241206030151e-05,
-      "loss": 5.1847,
-      "step": 39237
-    },
-    {
-      "epoch": 20.46310299869622,
-      "grad_norm": 1.5924623012542725,
-      "learning_rate": 6.108140703517589e-05,
-      "loss": 5.2675,
-      "step": 39238
-    },
-    {
-      "epoch": 20.463624511082138,
-      "grad_norm": 1.610077977180481,
-      "learning_rate": 6.108040201005025e-05,
-      "loss": 5.3199,
-      "step": 39239
-    },
-    {
-      "epoch": 20.464146023468057,
-      "grad_norm": 1.4391592741012573,
-      "learning_rate": 6.107939698492463e-05,
-      "loss": 5.203,
-      "step": 39240
-    },
-    {
-      "epoch": 20.464667535853977,
-      "grad_norm": 1.4315147399902344,
-      "learning_rate": 6.1078391959799e-05,
-      "loss": 5.5225,
-      "step": 39241
-    },
-    {
-      "epoch": 20.465189048239896,
-      "grad_norm": 1.4747183322906494,
-      "learning_rate": 6.107738693467337e-05,
-      "loss": 5.5624,
-      "step": 39242
-    },
-    {
-      "epoch": 20.465710560625816,
-      "grad_norm": 1.4516838788986206,
-      "learning_rate": 6.107638190954774e-05,
-      "loss": 5.31,
-      "step": 39243
-    },
-    {
-      "epoch": 20.466232073011735,
-      "grad_norm": 1.6226801872253418,
-      "learning_rate": 6.10753768844221e-05,
-      "loss": 4.8929,
-      "step": 39244
-    },
-    {
-      "epoch": 20.466753585397655,
-      "grad_norm": 1.4757024049758911,
-      "learning_rate": 6.107437185929648e-05,
-      "loss": 5.2256,
-      "step": 39245
-    },
-    {
-      "epoch": 20.467275097783574,
-      "grad_norm": 1.8639414310455322,
-      "learning_rate": 6.107336683417086e-05,
-      "loss": 4.8316,
-      "step": 39246
-    },
-    {
-      "epoch": 20.46779661016949,
-      "grad_norm": 1.4654583930969238,
-      "learning_rate": 6.107236180904524e-05,
-      "loss": 5.6576,
-      "step": 39247
-    },
-    {
-      "epoch": 20.46831812255541,
-      "grad_norm": 1.5810139179229736,
-      "learning_rate": 6.10713567839196e-05,
-      "loss": 5.4652,
-      "step": 39248
-    },
-    {
-      "epoch": 20.46883963494133,
-      "grad_norm": 1.6509428024291992,
-      "learning_rate": 6.107035175879398e-05,
-      "loss": 5.2123,
-      "step": 39249
-    },
-    {
-      "epoch": 20.46936114732725,
-      "grad_norm": 1.4936994314193726,
-      "learning_rate": 6.106934673366834e-05,
-      "loss": 5.0404,
-      "step": 39250
-    },
-    {
-      "epoch": 20.469882659713168,
-      "grad_norm": 1.5470963716506958,
-      "learning_rate": 6.106834170854272e-05,
-      "loss": 4.9574,
-      "step": 39251
-    },
-    {
-      "epoch": 20.470404172099087,
-      "grad_norm": 1.5847070217132568,
-      "learning_rate": 6.106733668341708e-05,
-      "loss": 5.3926,
-      "step": 39252
-    },
-    {
-      "epoch": 20.470925684485007,
-      "grad_norm": 1.471047043800354,
-      "learning_rate": 6.106633165829146e-05,
-      "loss": 5.2604,
-      "step": 39253
-    },
-    {
-      "epoch": 20.471447196870926,
-      "grad_norm": 1.8438405990600586,
-      "learning_rate": 6.106532663316583e-05,
-      "loss": 5.0121,
-      "step": 39254
-    },
-    {
-      "epoch": 20.471968709256846,
-      "grad_norm": 1.5046148300170898,
-      "learning_rate": 6.10643216080402e-05,
-      "loss": 5.4523,
-      "step": 39255
-    },
-    {
-      "epoch": 20.472490221642765,
-      "grad_norm": 1.4635696411132812,
-      "learning_rate": 6.106331658291458e-05,
-      "loss": 5.5091,
-      "step": 39256
-    },
-    {
-      "epoch": 20.473011734028685,
-      "grad_norm": 1.5024489164352417,
-      "learning_rate": 6.106231155778895e-05,
-      "loss": 5.2482,
-      "step": 39257
-    },
-    {
-      "epoch": 20.473533246414604,
-      "grad_norm": 1.581940770149231,
-      "learning_rate": 6.106130653266332e-05,
-      "loss": 5.3334,
-      "step": 39258
-    },
-    {
-      "epoch": 20.47405475880052,
-      "grad_norm": 1.5555427074432373,
-      "learning_rate": 6.106030150753769e-05,
-      "loss": 5.3749,
-      "step": 39259
-    },
-    {
-      "epoch": 20.47457627118644,
-      "grad_norm": 1.5814120769500732,
-      "learning_rate": 6.105929648241207e-05,
-      "loss": 5.0587,
-      "step": 39260
-    },
-    {
-      "epoch": 20.47509778357236,
-      "grad_norm": 1.5108728408813477,
-      "learning_rate": 6.105829145728643e-05,
-      "loss": 5.3813,
-      "step": 39261
-    },
-    {
-      "epoch": 20.47561929595828,
-      "grad_norm": 1.575006365776062,
-      "learning_rate": 6.105728643216081e-05,
-      "loss": 5.1302,
-      "step": 39262
-    },
-    {
-      "epoch": 20.476140808344198,
-      "grad_norm": 1.5255606174468994,
-      "learning_rate": 6.105628140703517e-05,
-      "loss": 4.9115,
-      "step": 39263
-    },
-    {
-      "epoch": 20.476662320730117,
-      "grad_norm": 1.5296711921691895,
-      "learning_rate": 6.105527638190955e-05,
-      "loss": 5.3278,
-      "step": 39264
-    },
-    {
-      "epoch": 20.477183833116037,
-      "grad_norm": 1.5366332530975342,
-      "learning_rate": 6.105427135678391e-05,
-      "loss": 5.5106,
-      "step": 39265
-    },
-    {
-      "epoch": 20.477705345501956,
-      "grad_norm": 1.585924744606018,
-      "learning_rate": 6.105326633165829e-05,
-      "loss": 4.8841,
-      "step": 39266
-    },
-    {
-      "epoch": 20.478226857887876,
-      "grad_norm": 1.4706027507781982,
-      "learning_rate": 6.105226130653267e-05,
-      "loss": 5.686,
-      "step": 39267
-    },
-    {
-      "epoch": 20.478748370273795,
-      "grad_norm": 1.5690721273422241,
-      "learning_rate": 6.105125628140705e-05,
-      "loss": 5.265,
-      "step": 39268
-    },
-    {
-      "epoch": 20.479269882659715,
-      "grad_norm": 1.4599096775054932,
-      "learning_rate": 6.105025125628141e-05,
-      "loss": 5.3254,
-      "step": 39269
-    },
-    {
-      "epoch": 20.479791395045634,
-      "grad_norm": 1.5083411931991577,
-      "learning_rate": 6.104924623115578e-05,
-      "loss": 5.6519,
-      "step": 39270
-    },
-    {
-      "epoch": 20.48031290743155,
-      "grad_norm": 1.470821499824524,
-      "learning_rate": 6.104824120603015e-05,
-      "loss": 5.3031,
-      "step": 39271
-    },
-    {
-      "epoch": 20.48083441981747,
-      "grad_norm": 1.5599735975265503,
-      "learning_rate": 6.104723618090452e-05,
-      "loss": 5.4226,
-      "step": 39272
-    },
-    {
-      "epoch": 20.48135593220339,
-      "grad_norm": 1.4817174673080444,
-      "learning_rate": 6.10462311557789e-05,
-      "loss": 5.5017,
-      "step": 39273
-    },
-    {
-      "epoch": 20.48187744458931,
-      "grad_norm": 1.5488996505737305,
-      "learning_rate": 6.104522613065326e-05,
-      "loss": 5.0908,
-      "step": 39274
-    },
-    {
-      "epoch": 20.482398956975228,
-      "grad_norm": 1.5446836948394775,
-      "learning_rate": 6.104422110552764e-05,
-      "loss": 5.2754,
-      "step": 39275
-    },
-    {
-      "epoch": 20.482920469361147,
-      "grad_norm": 1.7011641263961792,
-      "learning_rate": 6.104321608040202e-05,
-      "loss": 5.5051,
-      "step": 39276
-    },
-    {
-      "epoch": 20.483441981747067,
-      "grad_norm": 1.506603717803955,
-      "learning_rate": 6.10422110552764e-05,
-      "loss": 5.0818,
-      "step": 39277
-    },
-    {
-      "epoch": 20.483963494132986,
-      "grad_norm": 1.61571204662323,
-      "learning_rate": 6.104120603015076e-05,
-      "loss": 5.4402,
-      "step": 39278
-    },
-    {
-      "epoch": 20.484485006518906,
-      "grad_norm": 1.5746686458587646,
-      "learning_rate": 6.104020100502514e-05,
-      "loss": 5.1099,
-      "step": 39279
-    },
-    {
-      "epoch": 20.485006518904825,
-      "grad_norm": 1.7260318994522095,
-      "learning_rate": 6.10391959798995e-05,
-      "loss": 5.1174,
-      "step": 39280
-    },
-    {
-      "epoch": 20.485528031290745,
-      "grad_norm": 1.5812554359436035,
-      "learning_rate": 6.103819095477388e-05,
-      "loss": 5.0754,
-      "step": 39281
-    },
-    {
-      "epoch": 20.486049543676664,
-      "grad_norm": 1.6353305578231812,
-      "learning_rate": 6.103718592964824e-05,
-      "loss": 5.0108,
-      "step": 39282
-    },
-    {
-      "epoch": 20.48657105606258,
-      "grad_norm": 1.4977896213531494,
-      "learning_rate": 6.103618090452261e-05,
-      "loss": 5.3568,
-      "step": 39283
-    },
-    {
-      "epoch": 20.4870925684485,
-      "grad_norm": 1.5921846628189087,
-      "learning_rate": 6.103517587939698e-05,
-      "loss": 5.6261,
-      "step": 39284
-    },
-    {
-      "epoch": 20.48761408083442,
-      "grad_norm": 1.479851484298706,
-      "learning_rate": 6.1034170854271356e-05,
-      "loss": 5.4782,
-      "step": 39285
-    },
-    {
-      "epoch": 20.48813559322034,
-      "grad_norm": 1.5566447973251343,
-      "learning_rate": 6.1033165829145734e-05,
-      "loss": 5.3916,
-      "step": 39286
-    },
-    {
-      "epoch": 20.488657105606258,
-      "grad_norm": 1.6301968097686768,
-      "learning_rate": 6.10321608040201e-05,
-      "loss": 5.179,
-      "step": 39287
-    },
-    {
-      "epoch": 20.489178617992177,
-      "grad_norm": 1.6149908304214478,
-      "learning_rate": 6.1031155778894476e-05,
-      "loss": 4.7348,
-      "step": 39288
-    },
-    {
-      "epoch": 20.489700130378097,
-      "grad_norm": 1.4717965126037598,
-      "learning_rate": 6.103015075376885e-05,
-      "loss": 5.4496,
-      "step": 39289
-    },
-    {
-      "epoch": 20.490221642764016,
-      "grad_norm": 1.4751356840133667,
-      "learning_rate": 6.1029145728643225e-05,
-      "loss": 5.5731,
-      "step": 39290
-    },
-    {
-      "epoch": 20.490743155149936,
-      "grad_norm": 1.7154840230941772,
-      "learning_rate": 6.102814070351759e-05,
-      "loss": 5.1433,
-      "step": 39291
-    },
-    {
-      "epoch": 20.491264667535855,
-      "grad_norm": 1.461373209953308,
-      "learning_rate": 6.102713567839197e-05,
-      "loss": 5.455,
-      "step": 39292
-    },
-    {
-      "epoch": 20.491786179921775,
-      "grad_norm": 1.5330816507339478,
-      "learning_rate": 6.102613065326633e-05,
-      "loss": 5.479,
-      "step": 39293
-    },
-    {
-      "epoch": 20.49230769230769,
-      "grad_norm": 1.511727213859558,
-      "learning_rate": 6.102512562814071e-05,
-      "loss": 5.2021,
-      "step": 39294
-    },
-    {
-      "epoch": 20.49282920469361,
-      "grad_norm": 1.5649257898330688,
-      "learning_rate": 6.102412060301508e-05,
-      "loss": 5.4161,
-      "step": 39295
-    },
-    {
-      "epoch": 20.49335071707953,
-      "grad_norm": 1.5506330728530884,
-      "learning_rate": 6.1023115577889444e-05,
-      "loss": 5.6021,
-      "step": 39296
-    },
-    {
-      "epoch": 20.49387222946545,
-      "grad_norm": 1.5315881967544556,
-      "learning_rate": 6.102211055276382e-05,
-      "loss": 4.9586,
-      "step": 39297
-    },
-    {
-      "epoch": 20.49439374185137,
-      "grad_norm": 1.7004175186157227,
-      "learning_rate": 6.1021105527638186e-05,
-      "loss": 4.7243,
-      "step": 39298
-    },
-    {
-      "epoch": 20.494915254237288,
-      "grad_norm": 1.442698359489441,
-      "learning_rate": 6.1020100502512564e-05,
-      "loss": 5.3075,
-      "step": 39299
-    },
-    {
-      "epoch": 20.495436766623207,
-      "grad_norm": 1.4816226959228516,
-      "learning_rate": 6.1019095477386935e-05,
-      "loss": 5.5148,
-      "step": 39300
-    },
-    {
-      "epoch": 20.495958279009127,
-      "grad_norm": 1.3376781940460205,
-      "learning_rate": 6.101809045226131e-05,
-      "loss": 5.7856,
-      "step": 39301
-    },
-    {
-      "epoch": 20.496479791395046,
-      "grad_norm": 1.428850769996643,
-      "learning_rate": 6.101708542713568e-05,
-      "loss": 5.4598,
-      "step": 39302
-    },
-    {
-      "epoch": 20.497001303780966,
-      "grad_norm": 1.4255536794662476,
-      "learning_rate": 6.1016080402010055e-05,
-      "loss": 5.3882,
-      "step": 39303
-    },
-    {
-      "epoch": 20.497522816166885,
-      "grad_norm": 1.6311066150665283,
-      "learning_rate": 6.1015075376884426e-05,
-      "loss": 5.1186,
-      "step": 39304
-    },
-    {
-      "epoch": 20.498044328552805,
-      "grad_norm": 1.402056097984314,
-      "learning_rate": 6.1014070351758804e-05,
-      "loss": 5.6825,
-      "step": 39305
-    },
-    {
-      "epoch": 20.49856584093872,
-      "grad_norm": 1.555078148841858,
-      "learning_rate": 6.101306532663317e-05,
-      "loss": 5.4639,
-      "step": 39306
-    },
-    {
-      "epoch": 20.49908735332464,
-      "grad_norm": 1.4640473127365112,
-      "learning_rate": 6.1012060301507546e-05,
-      "loss": 5.5691,
-      "step": 39307
-    },
-    {
-      "epoch": 20.49960886571056,
-      "grad_norm": 1.532235860824585,
-      "learning_rate": 6.101105527638191e-05,
-      "loss": 5.4043,
-      "step": 39308
-    },
-    {
-      "epoch": 20.50013037809648,
-      "grad_norm": 1.9495059251785278,
-      "learning_rate": 6.101005025125628e-05,
-      "loss": 4.4664,
-      "step": 39309
-    },
-    {
-      "epoch": 20.5006518904824,
-      "grad_norm": 1.5615993738174438,
-      "learning_rate": 6.100904522613066e-05,
-      "loss": 5.2994,
-      "step": 39310
-    },
-    {
-      "epoch": 20.501173402868318,
-      "grad_norm": 1.519203782081604,
-      "learning_rate": 6.1008040201005024e-05,
-      "loss": 4.7192,
-      "step": 39311
-    },
-    {
-      "epoch": 20.501694915254237,
-      "grad_norm": 1.5965657234191895,
-      "learning_rate": 6.10070351758794e-05,
-      "loss": 5.5527,
-      "step": 39312
-    },
-    {
-      "epoch": 20.502216427640157,
-      "grad_norm": 1.5153594017028809,
-      "learning_rate": 6.1006030150753766e-05,
-      "loss": 5.504,
-      "step": 39313
-    },
-    {
-      "epoch": 20.502737940026076,
-      "grad_norm": 1.4216541051864624,
-      "learning_rate": 6.1005025125628143e-05,
-      "loss": 5.5458,
-      "step": 39314
-    },
-    {
-      "epoch": 20.503259452411996,
-      "grad_norm": 1.528558373451233,
-      "learning_rate": 6.1004020100502515e-05,
-      "loss": 5.194,
-      "step": 39315
-    },
-    {
-      "epoch": 20.503780964797915,
-      "grad_norm": 1.5273253917694092,
-      "learning_rate": 6.100301507537689e-05,
-      "loss": 4.8383,
-      "step": 39316
-    },
-    {
-      "epoch": 20.504302477183835,
-      "grad_norm": 1.35048508644104,
-      "learning_rate": 6.100201005025126e-05,
-      "loss": 4.7407,
-      "step": 39317
-    },
-    {
-      "epoch": 20.50482398956975,
-      "grad_norm": 1.5074056386947632,
-      "learning_rate": 6.1001005025125634e-05,
-      "loss": 5.3911,
-      "step": 39318
-    },
-    {
-      "epoch": 20.50534550195567,
-      "grad_norm": 1.495975375175476,
-      "learning_rate": 6.1e-05,
-      "loss": 5.265,
-      "step": 39319
-    },
-    {
-      "epoch": 20.50586701434159,
-      "grad_norm": 1.5368255376815796,
-      "learning_rate": 6.0998994974874377e-05,
-      "loss": 5.0668,
-      "step": 39320
-    },
-    {
-      "epoch": 20.50638852672751,
-      "grad_norm": 1.4520970582962036,
-      "learning_rate": 6.099798994974875e-05,
-      "loss": 5.4531,
-      "step": 39321
-    },
-    {
-      "epoch": 20.50691003911343,
-      "grad_norm": 1.4855247735977173,
-      "learning_rate": 6.0996984924623125e-05,
-      "loss": 4.6894,
-      "step": 39322
-    },
-    {
-      "epoch": 20.507431551499348,
-      "grad_norm": 1.6418043375015259,
-      "learning_rate": 6.099597989949749e-05,
-      "loss": 5.0289,
-      "step": 39323
-    },
-    {
-      "epoch": 20.507953063885267,
-      "grad_norm": 1.5531940460205078,
-      "learning_rate": 6.099497487437186e-05,
-      "loss": 5.5306,
-      "step": 39324
-    },
-    {
-      "epoch": 20.508474576271187,
-      "grad_norm": 1.5128228664398193,
-      "learning_rate": 6.099396984924624e-05,
-      "loss": 4.9188,
-      "step": 39325
-    },
-    {
-      "epoch": 20.508996088657106,
-      "grad_norm": 1.4974497556686401,
-      "learning_rate": 6.09929648241206e-05,
-      "loss": 5.313,
-      "step": 39326
-    },
-    {
-      "epoch": 20.509517601043026,
-      "grad_norm": 1.6299883127212524,
-      "learning_rate": 6.099195979899498e-05,
-      "loss": 4.8403,
-      "step": 39327
-    },
-    {
-      "epoch": 20.510039113428945,
-      "grad_norm": 1.551372766494751,
-      "learning_rate": 6.0990954773869345e-05,
-      "loss": 5.4634,
-      "step": 39328
-    },
-    {
-      "epoch": 20.510560625814865,
-      "grad_norm": 1.5218173265457153,
-      "learning_rate": 6.098994974874372e-05,
-      "loss": 5.1974,
-      "step": 39329
-    },
-    {
-      "epoch": 20.51108213820078,
-      "grad_norm": 1.4747446775436401,
-      "learning_rate": 6.0988944723618094e-05,
-      "loss": 5.1762,
-      "step": 39330
-    },
-    {
-      "epoch": 20.5116036505867,
-      "grad_norm": 1.6974270343780518,
-      "learning_rate": 6.098793969849247e-05,
-      "loss": 4.8721,
-      "step": 39331
-    },
-    {
-      "epoch": 20.51212516297262,
-      "grad_norm": 1.5009088516235352,
-      "learning_rate": 6.0986934673366836e-05,
-      "loss": 5.0687,
-      "step": 39332
-    },
-    {
-      "epoch": 20.51264667535854,
-      "grad_norm": 1.543951392173767,
-      "learning_rate": 6.0985929648241214e-05,
-      "loss": 5.0862,
-      "step": 39333
-    },
-    {
-      "epoch": 20.51316818774446,
-      "grad_norm": 1.546045184135437,
-      "learning_rate": 6.098492462311558e-05,
-      "loss": 5.3447,
-      "step": 39334
-    },
-    {
-      "epoch": 20.513689700130378,
-      "grad_norm": 1.4431819915771484,
-      "learning_rate": 6.0983919597989956e-05,
-      "loss": 5.3641,
-      "step": 39335
-    },
-    {
-      "epoch": 20.514211212516297,
-      "grad_norm": 1.4401087760925293,
-      "learning_rate": 6.098291457286433e-05,
-      "loss": 5.4214,
-      "step": 39336
-    },
-    {
-      "epoch": 20.514732724902217,
-      "grad_norm": 1.5025286674499512,
-      "learning_rate": 6.098190954773869e-05,
-      "loss": 4.9465,
-      "step": 39337
-    },
-    {
-      "epoch": 20.515254237288136,
-      "grad_norm": 1.4988389015197754,
-      "learning_rate": 6.098090452261307e-05,
-      "loss": 5.4828,
-      "step": 39338
-    },
-    {
-      "epoch": 20.515775749674056,
-      "grad_norm": 1.4675507545471191,
-      "learning_rate": 6.097989949748743e-05,
-      "loss": 5.2669,
-      "step": 39339
-    },
-    {
-      "epoch": 20.516297262059975,
-      "grad_norm": 1.472285509109497,
-      "learning_rate": 6.097889447236181e-05,
-      "loss": 5.3688,
-      "step": 39340
-    },
-    {
-      "epoch": 20.516818774445895,
-      "grad_norm": 1.4791316986083984,
-      "learning_rate": 6.097788944723618e-05,
-      "loss": 5.1714,
-      "step": 39341
-    },
-    {
-      "epoch": 20.51734028683181,
-      "grad_norm": 1.4187883138656616,
-      "learning_rate": 6.097688442211056e-05,
-      "loss": 5.5328,
-      "step": 39342
-    },
-    {
-      "epoch": 20.51786179921773,
-      "grad_norm": 1.5268903970718384,
-      "learning_rate": 6.0975879396984924e-05,
-      "loss": 5.3544,
-      "step": 39343
-    },
-    {
-      "epoch": 20.51838331160365,
-      "grad_norm": 1.591433048248291,
-      "learning_rate": 6.09748743718593e-05,
-      "loss": 5.5056,
-      "step": 39344
-    },
-    {
-      "epoch": 20.51890482398957,
-      "grad_norm": 1.6669795513153076,
-      "learning_rate": 6.0973869346733666e-05,
-      "loss": 5.4731,
-      "step": 39345
-    },
-    {
-      "epoch": 20.51942633637549,
-      "grad_norm": 1.4789931774139404,
-      "learning_rate": 6.0972864321608044e-05,
-      "loss": 5.4483,
-      "step": 39346
-    },
-    {
-      "epoch": 20.519947848761408,
-      "grad_norm": 1.6159300804138184,
-      "learning_rate": 6.0971859296482415e-05,
-      "loss": 4.892,
-      "step": 39347
-    },
-    {
-      "epoch": 20.520469361147327,
-      "grad_norm": 1.3779780864715576,
-      "learning_rate": 6.097085427135679e-05,
-      "loss": 5.4075,
-      "step": 39348
-    },
-    {
-      "epoch": 20.520990873533247,
-      "grad_norm": 1.543480634689331,
-      "learning_rate": 6.096984924623116e-05,
-      "loss": 5.3206,
-      "step": 39349
-    },
-    {
-      "epoch": 20.521512385919166,
-      "grad_norm": 1.5860562324523926,
-      "learning_rate": 6.096884422110553e-05,
-      "loss": 5.3015,
-      "step": 39350
-    },
-    {
-      "epoch": 20.522033898305086,
-      "grad_norm": 1.4648233652114868,
-      "learning_rate": 6.0967839195979906e-05,
-      "loss": 5.3243,
-      "step": 39351
-    },
-    {
-      "epoch": 20.522555410691005,
-      "grad_norm": 1.6231242418289185,
-      "learning_rate": 6.096683417085427e-05,
-      "loss": 4.8403,
-      "step": 39352
-    },
-    {
-      "epoch": 20.523076923076925,
-      "grad_norm": 1.4124469757080078,
-      "learning_rate": 6.096582914572865e-05,
-      "loss": 5.3699,
-      "step": 39353
-    },
-    {
-      "epoch": 20.52359843546284,
-      "grad_norm": 1.4449304342269897,
-      "learning_rate": 6.096482412060301e-05,
-      "loss": 5.4157,
-      "step": 39354
-    },
-    {
-      "epoch": 20.52411994784876,
-      "grad_norm": 1.4580754041671753,
-      "learning_rate": 6.096381909547739e-05,
-      "loss": 5.3524,
-      "step": 39355
-    },
-    {
-      "epoch": 20.52464146023468,
-      "grad_norm": 1.4876171350479126,
-      "learning_rate": 6.096281407035176e-05,
-      "loss": 4.6418,
-      "step": 39356
-    },
-    {
-      "epoch": 20.5251629726206,
-      "grad_norm": 1.4255387783050537,
-      "learning_rate": 6.096180904522614e-05,
-      "loss": 5.443,
-      "step": 39357
-    },
-    {
-      "epoch": 20.52568448500652,
-      "grad_norm": 1.560710072517395,
-      "learning_rate": 6.0960804020100504e-05,
-      "loss": 5.5355,
-      "step": 39358
-    },
-    {
-      "epoch": 20.526205997392438,
-      "grad_norm": 1.402662754058838,
-      "learning_rate": 6.095979899497488e-05,
-      "loss": 4.5727,
-      "step": 39359
-    },
-    {
-      "epoch": 20.526727509778357,
-      "grad_norm": 1.4244173765182495,
-      "learning_rate": 6.0958793969849246e-05,
-      "loss": 5.6894,
-      "step": 39360
-    },
-    {
-      "epoch": 20.527249022164277,
-      "grad_norm": 1.4352562427520752,
-      "learning_rate": 6.0957788944723624e-05,
-      "loss": 4.8222,
-      "step": 39361
-    },
-    {
-      "epoch": 20.527770534550196,
-      "grad_norm": 1.5219932794570923,
-      "learning_rate": 6.0956783919597995e-05,
-      "loss": 5.2166,
-      "step": 39362
-    },
-    {
-      "epoch": 20.528292046936116,
-      "grad_norm": 1.5687726736068726,
-      "learning_rate": 6.095577889447236e-05,
-      "loss": 5.521,
-      "step": 39363
-    },
-    {
-      "epoch": 20.528813559322035,
-      "grad_norm": 1.6225850582122803,
-      "learning_rate": 6.095477386934674e-05,
-      "loss": 5.0688,
-      "step": 39364
-    },
-    {
-      "epoch": 20.529335071707955,
-      "grad_norm": 1.5433387756347656,
-      "learning_rate": 6.09537688442211e-05,
-      "loss": 5.1837,
-      "step": 39365
-    },
-    {
-      "epoch": 20.52985658409387,
-      "grad_norm": 1.4791748523712158,
-      "learning_rate": 6.095276381909548e-05,
-      "loss": 4.2965,
-      "step": 39366
-    },
-    {
-      "epoch": 20.53037809647979,
-      "grad_norm": 1.5635392665863037,
-      "learning_rate": 6.095175879396985e-05,
-      "loss": 4.865,
-      "step": 39367
-    },
-    {
-      "epoch": 20.53089960886571,
-      "grad_norm": 1.5515981912612915,
-      "learning_rate": 6.095075376884423e-05,
-      "loss": 5.3706,
-      "step": 39368
-    },
-    {
-      "epoch": 20.53142112125163,
-      "grad_norm": 1.529679775238037,
-      "learning_rate": 6.094974874371859e-05,
-      "loss": 5.3264,
-      "step": 39369
-    },
-    {
-      "epoch": 20.53194263363755,
-      "grad_norm": 1.4900251626968384,
-      "learning_rate": 6.094874371859297e-05,
-      "loss": 5.5384,
-      "step": 39370
-    },
-    {
-      "epoch": 20.532464146023468,
-      "grad_norm": 1.4516819715499878,
-      "learning_rate": 6.094773869346734e-05,
-      "loss": 5.6852,
-      "step": 39371
-    },
-    {
-      "epoch": 20.532985658409387,
-      "grad_norm": 1.6036107540130615,
-      "learning_rate": 6.094673366834172e-05,
-      "loss": 5.3745,
-      "step": 39372
-    },
-    {
-      "epoch": 20.533507170795307,
-      "grad_norm": 1.6219121217727661,
-      "learning_rate": 6.094572864321608e-05,
-      "loss": 5.3281,
-      "step": 39373
-    },
-    {
-      "epoch": 20.534028683181226,
-      "grad_norm": 1.4624727964401245,
-      "learning_rate": 6.094472361809046e-05,
-      "loss": 5.5453,
-      "step": 39374
-    },
-    {
-      "epoch": 20.534550195567146,
-      "grad_norm": 1.6031413078308105,
-      "learning_rate": 6.0943718592964825e-05,
-      "loss": 5.0036,
-      "step": 39375
-    },
-    {
-      "epoch": 20.535071707953065,
-      "grad_norm": 1.54746413230896,
-      "learning_rate": 6.0942713567839196e-05,
-      "loss": 5.3341,
-      "step": 39376
-    },
-    {
-      "epoch": 20.53559322033898,
-      "grad_norm": 1.595733404159546,
-      "learning_rate": 6.0941708542713574e-05,
-      "loss": 5.3268,
-      "step": 39377
-    },
-    {
-      "epoch": 20.5361147327249,
-      "grad_norm": 1.5625718832015991,
-      "learning_rate": 6.094070351758794e-05,
-      "loss": 5.4521,
-      "step": 39378
-    },
-    {
-      "epoch": 20.53663624511082,
-      "grad_norm": 1.6058812141418457,
-      "learning_rate": 6.0939698492462316e-05,
-      "loss": 5.2401,
-      "step": 39379
-    },
-    {
-      "epoch": 20.53715775749674,
-      "grad_norm": 1.5571467876434326,
-      "learning_rate": 6.093869346733668e-05,
-      "loss": 5.3479,
-      "step": 39380
-    },
-    {
-      "epoch": 20.53767926988266,
-      "grad_norm": 1.5247060060501099,
-      "learning_rate": 6.093768844221106e-05,
-      "loss": 5.5184,
-      "step": 39381
-    },
-    {
-      "epoch": 20.53820078226858,
-      "grad_norm": 1.5689324140548706,
-      "learning_rate": 6.093668341708543e-05,
-      "loss": 5.7521,
-      "step": 39382
-    },
-    {
-      "epoch": 20.538722294654498,
-      "grad_norm": 1.6441315412521362,
-      "learning_rate": 6.093567839195981e-05,
-      "loss": 5.0187,
-      "step": 39383
-    },
-    {
-      "epoch": 20.539243807040418,
-      "grad_norm": 1.5081523656845093,
-      "learning_rate": 6.093467336683417e-05,
-      "loss": 5.2386,
-      "step": 39384
-    },
-    {
-      "epoch": 20.539765319426337,
-      "grad_norm": 1.588078498840332,
-      "learning_rate": 6.093366834170855e-05,
-      "loss": 5.1381,
-      "step": 39385
-    },
-    {
-      "epoch": 20.540286831812256,
-      "grad_norm": 1.6446964740753174,
-      "learning_rate": 6.0932663316582913e-05,
-      "loss": 5.1853,
-      "step": 39386
-    },
-    {
-      "epoch": 20.540808344198176,
-      "grad_norm": 1.5577090978622437,
-      "learning_rate": 6.093165829145729e-05,
-      "loss": 5.232,
-      "step": 39387
-    },
-    {
-      "epoch": 20.541329856584095,
-      "grad_norm": 1.6448345184326172,
-      "learning_rate": 6.093065326633166e-05,
-      "loss": 5.0021,
-      "step": 39388
-    },
-    {
-      "epoch": 20.541851368970015,
-      "grad_norm": 1.476454734802246,
-      "learning_rate": 6.0929648241206027e-05,
-      "loss": 5.6116,
-      "step": 39389
-    },
-    {
-      "epoch": 20.54237288135593,
-      "grad_norm": 1.5331435203552246,
-      "learning_rate": 6.0928643216080404e-05,
-      "loss": 5.2872,
-      "step": 39390
-    },
-    {
-      "epoch": 20.54289439374185,
-      "grad_norm": 1.5109621286392212,
-      "learning_rate": 6.0927638190954775e-05,
-      "loss": 5.3416,
-      "step": 39391
-    },
-    {
-      "epoch": 20.54341590612777,
-      "grad_norm": 1.4367281198501587,
-      "learning_rate": 6.092663316582915e-05,
-      "loss": 5.3387,
-      "step": 39392
-    },
-    {
-      "epoch": 20.54393741851369,
-      "grad_norm": 1.6654082536697388,
-      "learning_rate": 6.092562814070352e-05,
-      "loss": 4.5465,
-      "step": 39393
-    },
-    {
-      "epoch": 20.54445893089961,
-      "grad_norm": 1.5357283353805542,
-      "learning_rate": 6.0924623115577895e-05,
-      "loss": 5.6508,
-      "step": 39394
-    },
-    {
-      "epoch": 20.544980443285528,
-      "grad_norm": 1.4247822761535645,
-      "learning_rate": 6.092361809045226e-05,
-      "loss": 5.4437,
-      "step": 39395
-    },
-    {
-      "epoch": 20.545501955671448,
-      "grad_norm": 1.5824918746948242,
-      "learning_rate": 6.092261306532664e-05,
-      "loss": 5.2144,
-      "step": 39396
-    },
-    {
-      "epoch": 20.546023468057367,
-      "grad_norm": 1.4608968496322632,
-      "learning_rate": 6.092160804020101e-05,
-      "loss": 5.2426,
-      "step": 39397
-    },
-    {
-      "epoch": 20.546544980443286,
-      "grad_norm": 1.4815287590026855,
-      "learning_rate": 6.0920603015075386e-05,
-      "loss": 5.4486,
-      "step": 39398
-    },
-    {
-      "epoch": 20.547066492829206,
-      "grad_norm": 1.546678066253662,
-      "learning_rate": 6.091959798994975e-05,
-      "loss": 5.3167,
-      "step": 39399
-    },
-    {
-      "epoch": 20.547588005215125,
-      "grad_norm": 1.4661179780960083,
-      "learning_rate": 6.091859296482413e-05,
-      "loss": 5.4071,
-      "step": 39400
-    },
-    {
-      "epoch": 20.54810951760104,
-      "grad_norm": 1.5644499063491821,
-      "learning_rate": 6.091758793969849e-05,
-      "loss": 4.7755,
-      "step": 39401
-    },
-    {
-      "epoch": 20.54863102998696,
-      "grad_norm": 1.493137001991272,
-      "learning_rate": 6.0916582914572864e-05,
-      "loss": 5.5109,
-      "step": 39402
-    },
-    {
-      "epoch": 20.54915254237288,
-      "grad_norm": 1.5213409662246704,
-      "learning_rate": 6.091557788944724e-05,
-      "loss": 5.0593,
-      "step": 39403
-    },
-    {
-      "epoch": 20.5496740547588,
-      "grad_norm": 1.5352951288223267,
-      "learning_rate": 6.0914572864321606e-05,
-      "loss": 5.6213,
-      "step": 39404
-    },
-    {
-      "epoch": 20.55019556714472,
-      "grad_norm": 1.4461774826049805,
-      "learning_rate": 6.0913567839195984e-05,
-      "loss": 5.4989,
-      "step": 39405
-    },
-    {
-      "epoch": 20.55071707953064,
-      "grad_norm": 1.6170477867126465,
-      "learning_rate": 6.091256281407035e-05,
-      "loss": 5.505,
-      "step": 39406
-    },
-    {
-      "epoch": 20.551238591916558,
-      "grad_norm": 1.4707235097885132,
-      "learning_rate": 6.0911557788944726e-05,
-      "loss": 5.133,
-      "step": 39407
-    },
-    {
-      "epoch": 20.551760104302478,
-      "grad_norm": 1.526599645614624,
-      "learning_rate": 6.09105527638191e-05,
-      "loss": 4.9323,
-      "step": 39408
-    },
-    {
-      "epoch": 20.552281616688397,
-      "grad_norm": 1.5554530620574951,
-      "learning_rate": 6.0909547738693475e-05,
-      "loss": 5.1551,
-      "step": 39409
-    },
-    {
-      "epoch": 20.552803129074317,
-      "grad_norm": 1.5319855213165283,
-      "learning_rate": 6.090854271356784e-05,
-      "loss": 4.7772,
-      "step": 39410
-    },
-    {
-      "epoch": 20.553324641460236,
-      "grad_norm": 1.4569919109344482,
-      "learning_rate": 6.090753768844222e-05,
-      "loss": 5.5713,
-      "step": 39411
-    },
-    {
-      "epoch": 20.553846153846155,
-      "grad_norm": 1.6351990699768066,
-      "learning_rate": 6.090653266331659e-05,
-      "loss": 5.202,
-      "step": 39412
-    },
-    {
-      "epoch": 20.55436766623207,
-      "grad_norm": 1.5905256271362305,
-      "learning_rate": 6.0905527638190966e-05,
-      "loss": 5.1375,
-      "step": 39413
-    },
-    {
-      "epoch": 20.55488917861799,
-      "grad_norm": 1.7902603149414062,
-      "learning_rate": 6.090452261306533e-05,
-      "loss": 5.7295,
-      "step": 39414
-    },
-    {
-      "epoch": 20.55541069100391,
-      "grad_norm": 1.6377737522125244,
-      "learning_rate": 6.090351758793971e-05,
-      "loss": 4.7697,
-      "step": 39415
-    },
-    {
-      "epoch": 20.55593220338983,
-      "grad_norm": 1.5110399723052979,
-      "learning_rate": 6.090251256281407e-05,
-      "loss": 5.1417,
-      "step": 39416
-    },
-    {
-      "epoch": 20.55645371577575,
-      "grad_norm": 1.5144548416137695,
-      "learning_rate": 6.090150753768844e-05,
-      "loss": 5.1747,
-      "step": 39417
-    },
-    {
-      "epoch": 20.55697522816167,
-      "grad_norm": 1.5214625597000122,
-      "learning_rate": 6.090050251256282e-05,
-      "loss": 5.3781,
-      "step": 39418
-    },
-    {
-      "epoch": 20.557496740547588,
-      "grad_norm": 1.3808985948562622,
-      "learning_rate": 6.0899497487437185e-05,
-      "loss": 5.5016,
-      "step": 39419
-    },
-    {
-      "epoch": 20.558018252933508,
-      "grad_norm": 1.5280511379241943,
-      "learning_rate": 6.089849246231156e-05,
-      "loss": 5.503,
-      "step": 39420
-    },
-    {
-      "epoch": 20.558539765319427,
-      "grad_norm": 1.4521267414093018,
-      "learning_rate": 6.089748743718593e-05,
-      "loss": 5.4234,
-      "step": 39421
-    },
-    {
-      "epoch": 20.559061277705347,
-      "grad_norm": 1.4117993116378784,
-      "learning_rate": 6.0896482412060305e-05,
-      "loss": 5.8622,
-      "step": 39422
-    },
-    {
-      "epoch": 20.559582790091266,
-      "grad_norm": 1.4404985904693604,
-      "learning_rate": 6.0895477386934676e-05,
-      "loss": 5.44,
-      "step": 39423
-    },
-    {
-      "epoch": 20.560104302477185,
-      "grad_norm": 1.5323840379714966,
-      "learning_rate": 6.0894472361809054e-05,
-      "loss": 5.4201,
-      "step": 39424
-    },
-    {
-      "epoch": 20.5606258148631,
-      "grad_norm": 1.5557546615600586,
-      "learning_rate": 6.089346733668342e-05,
-      "loss": 5.2495,
-      "step": 39425
-    },
-    {
-      "epoch": 20.56114732724902,
-      "grad_norm": 1.4497714042663574,
-      "learning_rate": 6.0892462311557796e-05,
-      "loss": 5.2772,
-      "step": 39426
-    },
-    {
-      "epoch": 20.56166883963494,
-      "grad_norm": 1.4826403856277466,
-      "learning_rate": 6.089145728643216e-05,
-      "loss": 5.1972,
-      "step": 39427
-    },
-    {
-      "epoch": 20.56219035202086,
-      "grad_norm": 1.4309693574905396,
-      "learning_rate": 6.089045226130654e-05,
-      "loss": 5.2927,
-      "step": 39428
-    },
-    {
-      "epoch": 20.56271186440678,
-      "grad_norm": 1.6100399494171143,
-      "learning_rate": 6.088944723618091e-05,
-      "loss": 4.951,
-      "step": 39429
-    },
-    {
-      "epoch": 20.5632333767927,
-      "grad_norm": 1.5190948247909546,
-      "learning_rate": 6.0888442211055274e-05,
-      "loss": 5.207,
-      "step": 39430
-    },
-    {
-      "epoch": 20.563754889178618,
-      "grad_norm": 1.4372402429580688,
-      "learning_rate": 6.088743718592965e-05,
-      "loss": 5.4809,
-      "step": 39431
-    },
-    {
-      "epoch": 20.564276401564538,
-      "grad_norm": 1.450183629989624,
-      "learning_rate": 6.0886432160804016e-05,
-      "loss": 5.4361,
-      "step": 39432
-    },
-    {
-      "epoch": 20.564797913950457,
-      "grad_norm": 1.6359535455703735,
-      "learning_rate": 6.0885427135678393e-05,
-      "loss": 4.7277,
-      "step": 39433
-    },
-    {
-      "epoch": 20.565319426336377,
-      "grad_norm": 1.4612196683883667,
-      "learning_rate": 6.0884422110552765e-05,
-      "loss": 5.3584,
-      "step": 39434
-    },
-    {
-      "epoch": 20.565840938722296,
-      "grad_norm": 1.5562682151794434,
-      "learning_rate": 6.088341708542714e-05,
-      "loss": 5.044,
-      "step": 39435
-    },
-    {
-      "epoch": 20.566362451108215,
-      "grad_norm": 1.5783534049987793,
-      "learning_rate": 6.088241206030151e-05,
-      "loss": 5.2621,
-      "step": 39436
-    },
-    {
-      "epoch": 20.56688396349413,
-      "grad_norm": 1.5121049880981445,
-      "learning_rate": 6.0881407035175884e-05,
-      "loss": 4.9292,
-      "step": 39437
-    },
-    {
-      "epoch": 20.56740547588005,
-      "grad_norm": 1.6122666597366333,
-      "learning_rate": 6.0880402010050256e-05,
-      "loss": 5.0181,
-      "step": 39438
-    },
-    {
-      "epoch": 20.56792698826597,
-      "grad_norm": 1.5026912689208984,
-      "learning_rate": 6.087939698492463e-05,
-      "loss": 5.2683,
-      "step": 39439
-    },
-    {
-      "epoch": 20.56844850065189,
-      "grad_norm": 1.4204106330871582,
-      "learning_rate": 6.0878391959799e-05,
-      "loss": 5.5137,
-      "step": 39440
-    },
-    {
-      "epoch": 20.56897001303781,
-      "grad_norm": 1.641945481300354,
-      "learning_rate": 6.0877386934673375e-05,
-      "loss": 4.9206,
-      "step": 39441
-    },
-    {
-      "epoch": 20.56949152542373,
-      "grad_norm": 1.5947831869125366,
-      "learning_rate": 6.087638190954774e-05,
-      "loss": 5.4065,
-      "step": 39442
-    },
-    {
-      "epoch": 20.570013037809648,
-      "grad_norm": 1.611958384513855,
-      "learning_rate": 6.087537688442211e-05,
-      "loss": 5.0679,
-      "step": 39443
-    },
-    {
-      "epoch": 20.570534550195568,
-      "grad_norm": 1.5477532148361206,
-      "learning_rate": 6.087437185929649e-05,
-      "loss": 5.5253,
-      "step": 39444
-    },
-    {
-      "epoch": 20.571056062581487,
-      "grad_norm": 1.5274550914764404,
-      "learning_rate": 6.087336683417085e-05,
-      "loss": 5.42,
-      "step": 39445
-    },
-    {
-      "epoch": 20.571577574967407,
-      "grad_norm": 1.5009684562683105,
-      "learning_rate": 6.087236180904523e-05,
-      "loss": 5.4681,
-      "step": 39446
-    },
-    {
-      "epoch": 20.572099087353326,
-      "grad_norm": 1.5220609903335571,
-      "learning_rate": 6.0871356783919595e-05,
-      "loss": 5.2783,
-      "step": 39447
-    },
-    {
-      "epoch": 20.572620599739246,
-      "grad_norm": 1.6397931575775146,
-      "learning_rate": 6.087035175879397e-05,
-      "loss": 5.5902,
-      "step": 39448
-    },
-    {
-      "epoch": 20.57314211212516,
-      "grad_norm": 1.4499843120574951,
-      "learning_rate": 6.0869346733668344e-05,
-      "loss": 5.3103,
-      "step": 39449
-    },
-    {
-      "epoch": 20.57366362451108,
-      "grad_norm": 1.3678580522537231,
-      "learning_rate": 6.086834170854272e-05,
-      "loss": 4.9661,
-      "step": 39450
-    },
-    {
-      "epoch": 20.574185136897,
-      "grad_norm": 1.504172682762146,
-      "learning_rate": 6.0867336683417086e-05,
-      "loss": 5.3044,
-      "step": 39451
-    },
-    {
-      "epoch": 20.57470664928292,
-      "grad_norm": 1.5688613653182983,
-      "learning_rate": 6.0866331658291464e-05,
-      "loss": 5.1527,
-      "step": 39452
-    },
-    {
-      "epoch": 20.57522816166884,
-      "grad_norm": 1.4701305627822876,
-      "learning_rate": 6.086532663316583e-05,
-      "loss": 5.6125,
-      "step": 39453
-    },
-    {
-      "epoch": 20.57574967405476,
-      "grad_norm": 1.5266461372375488,
-      "learning_rate": 6.0864321608040206e-05,
-      "loss": 5.3919,
-      "step": 39454
-    },
-    {
-      "epoch": 20.576271186440678,
-      "grad_norm": 1.5410836935043335,
-      "learning_rate": 6.086331658291458e-05,
-      "loss": 4.9354,
-      "step": 39455
-    },
-    {
-      "epoch": 20.576792698826598,
-      "grad_norm": 1.47757089138031,
-      "learning_rate": 6.086231155778894e-05,
-      "loss": 5.4498,
-      "step": 39456
-    },
-    {
-      "epoch": 20.577314211212517,
-      "grad_norm": 1.5593279600143433,
-      "learning_rate": 6.086130653266332e-05,
-      "loss": 4.706,
-      "step": 39457
-    },
-    {
-      "epoch": 20.577835723598437,
-      "grad_norm": 1.5631204843521118,
-      "learning_rate": 6.086030150753769e-05,
-      "loss": 5.2485,
-      "step": 39458
-    },
-    {
-      "epoch": 20.578357235984356,
-      "grad_norm": 1.5886244773864746,
-      "learning_rate": 6.085929648241207e-05,
-      "loss": 4.7512,
-      "step": 39459
-    },
-    {
-      "epoch": 20.578878748370272,
-      "grad_norm": 1.4447076320648193,
-      "learning_rate": 6.085829145728643e-05,
-      "loss": 5.5566,
-      "step": 39460
-    },
-    {
-      "epoch": 20.57940026075619,
-      "grad_norm": 1.3683221340179443,
-      "learning_rate": 6.085728643216081e-05,
-      "loss": 5.3489,
-      "step": 39461
-    },
-    {
-      "epoch": 20.57992177314211,
-      "grad_norm": 1.506969690322876,
-      "learning_rate": 6.0856281407035174e-05,
-      "loss": 5.3612,
-      "step": 39462
-    },
-    {
-      "epoch": 20.58044328552803,
-      "grad_norm": 1.473534107208252,
-      "learning_rate": 6.085527638190955e-05,
-      "loss": 5.748,
-      "step": 39463
-    },
-    {
-      "epoch": 20.58096479791395,
-      "grad_norm": 1.3983136415481567,
-      "learning_rate": 6.085427135678392e-05,
-      "loss": 5.3461,
-      "step": 39464
-    },
-    {
-      "epoch": 20.58148631029987,
-      "grad_norm": 1.5360524654388428,
-      "learning_rate": 6.08532663316583e-05,
-      "loss": 5.2095,
-      "step": 39465
-    },
-    {
-      "epoch": 20.58200782268579,
-      "grad_norm": 1.469625473022461,
-      "learning_rate": 6.0852261306532665e-05,
-      "loss": 5.0779,
-      "step": 39466
-    },
-    {
-      "epoch": 20.58252933507171,
-      "grad_norm": 1.5269228219985962,
-      "learning_rate": 6.085125628140704e-05,
-      "loss": 4.9791,
-      "step": 39467
-    },
-    {
-      "epoch": 20.583050847457628,
-      "grad_norm": 1.5214533805847168,
-      "learning_rate": 6.085025125628141e-05,
-      "loss": 5.511,
-      "step": 39468
-    },
-    {
-      "epoch": 20.583572359843547,
-      "grad_norm": 1.4907368421554565,
-      "learning_rate": 6.084924623115578e-05,
-      "loss": 5.5028,
-      "step": 39469
-    },
-    {
-      "epoch": 20.584093872229467,
-      "grad_norm": 1.6120752096176147,
-      "learning_rate": 6.0848241206030156e-05,
-      "loss": 5.0576,
-      "step": 39470
-    },
-    {
-      "epoch": 20.584615384615386,
-      "grad_norm": 1.574238657951355,
-      "learning_rate": 6.084723618090452e-05,
-      "loss": 4.9604,
-      "step": 39471
-    },
-    {
-      "epoch": 20.585136897001306,
-      "grad_norm": 1.4375874996185303,
-      "learning_rate": 6.08462311557789e-05,
-      "loss": 5.6272,
-      "step": 39472
-    },
-    {
-      "epoch": 20.58565840938722,
-      "grad_norm": 1.5566617250442505,
-      "learning_rate": 6.084522613065326e-05,
-      "loss": 5.2702,
-      "step": 39473
-    },
-    {
-      "epoch": 20.58617992177314,
-      "grad_norm": 1.4980430603027344,
-      "learning_rate": 6.084422110552764e-05,
-      "loss": 5.2359,
-      "step": 39474
-    },
-    {
-      "epoch": 20.58670143415906,
-      "grad_norm": 1.4835655689239502,
-      "learning_rate": 6.084321608040201e-05,
-      "loss": 5.0265,
-      "step": 39475
-    },
-    {
-      "epoch": 20.58722294654498,
-      "grad_norm": 1.55263090133667,
-      "learning_rate": 6.084221105527639e-05,
-      "loss": 5.579,
-      "step": 39476
-    },
-    {
-      "epoch": 20.5877444589309,
-      "grad_norm": 1.648334264755249,
-      "learning_rate": 6.0841206030150754e-05,
-      "loss": 5.1807,
-      "step": 39477
-    },
-    {
-      "epoch": 20.58826597131682,
-      "grad_norm": 1.586987018585205,
-      "learning_rate": 6.084020100502513e-05,
-      "loss": 5.2615,
-      "step": 39478
-    },
-    {
-      "epoch": 20.58878748370274,
-      "grad_norm": 1.4790875911712646,
-      "learning_rate": 6.08391959798995e-05,
-      "loss": 5.4848,
-      "step": 39479
-    },
-    {
-      "epoch": 20.589308996088658,
-      "grad_norm": 1.5620747804641724,
-      "learning_rate": 6.083819095477388e-05,
-      "loss": 5.1716,
-      "step": 39480
-    },
-    {
-      "epoch": 20.589830508474577,
-      "grad_norm": 1.5078541040420532,
-      "learning_rate": 6.0837185929648245e-05,
-      "loss": 5.0079,
-      "step": 39481
-    },
-    {
-      "epoch": 20.590352020860497,
-      "grad_norm": 1.4756380319595337,
-      "learning_rate": 6.083618090452261e-05,
-      "loss": 5.3985,
-      "step": 39482
-    },
-    {
-      "epoch": 20.590873533246416,
-      "grad_norm": 1.601255178451538,
-      "learning_rate": 6.083517587939699e-05,
-      "loss": 5.1856,
-      "step": 39483
-    },
-    {
-      "epoch": 20.591395045632332,
-      "grad_norm": 1.4116466045379639,
-      "learning_rate": 6.083417085427136e-05,
-      "loss": 5.4748,
-      "step": 39484
-    },
-    {
-      "epoch": 20.59191655801825,
-      "grad_norm": 1.4749634265899658,
-      "learning_rate": 6.0833165829145736e-05,
-      "loss": 5.4071,
-      "step": 39485
-    },
-    {
-      "epoch": 20.59243807040417,
-      "grad_norm": 1.4961082935333252,
-      "learning_rate": 6.08321608040201e-05,
-      "loss": 5.5108,
-      "step": 39486
-    },
-    {
-      "epoch": 20.59295958279009,
-      "grad_norm": 1.7391705513000488,
-      "learning_rate": 6.083115577889448e-05,
-      "loss": 5.0293,
-      "step": 39487
-    },
-    {
-      "epoch": 20.59348109517601,
-      "grad_norm": 1.4929144382476807,
-      "learning_rate": 6.083015075376884e-05,
-      "loss": 5.5841,
-      "step": 39488
-    },
-    {
-      "epoch": 20.59400260756193,
-      "grad_norm": 1.5907635688781738,
-      "learning_rate": 6.082914572864322e-05,
-      "loss": 4.9332,
-      "step": 39489
-    },
-    {
-      "epoch": 20.59452411994785,
-      "grad_norm": 1.5337276458740234,
-      "learning_rate": 6.082814070351759e-05,
-      "loss": 5.4847,
-      "step": 39490
-    },
-    {
-      "epoch": 20.59504563233377,
-      "grad_norm": 1.6205893754959106,
-      "learning_rate": 6.082713567839197e-05,
-      "loss": 5.0998,
-      "step": 39491
-    },
-    {
-      "epoch": 20.595567144719688,
-      "grad_norm": 1.4510581493377686,
-      "learning_rate": 6.082613065326633e-05,
-      "loss": 5.4899,
-      "step": 39492
-    },
-    {
-      "epoch": 20.596088657105607,
-      "grad_norm": 1.5349059104919434,
-      "learning_rate": 6.082512562814071e-05,
-      "loss": 5.448,
-      "step": 39493
-    },
-    {
-      "epoch": 20.596610169491527,
-      "grad_norm": 1.655225157737732,
-      "learning_rate": 6.0824120603015075e-05,
-      "loss": 5.0008,
-      "step": 39494
-    },
-    {
-      "epoch": 20.597131681877446,
-      "grad_norm": 1.622947096824646,
-      "learning_rate": 6.082311557788945e-05,
-      "loss": 5.115,
-      "step": 39495
-    },
-    {
-      "epoch": 20.597653194263362,
-      "grad_norm": 1.4599788188934326,
-      "learning_rate": 6.0822110552763824e-05,
-      "loss": 5.6301,
-      "step": 39496
-    },
-    {
-      "epoch": 20.59817470664928,
-      "grad_norm": 1.6917649507522583,
-      "learning_rate": 6.082110552763819e-05,
-      "loss": 5.0848,
-      "step": 39497
-    },
-    {
-      "epoch": 20.5986962190352,
-      "grad_norm": 1.4262168407440186,
-      "learning_rate": 6.0820100502512566e-05,
-      "loss": 5.3798,
-      "step": 39498
-    },
-    {
-      "epoch": 20.59921773142112,
-      "grad_norm": 1.5220191478729248,
-      "learning_rate": 6.081909547738693e-05,
-      "loss": 5.1316,
-      "step": 39499
-    },
-    {
-      "epoch": 20.59973924380704,
-      "grad_norm": 1.4505013227462769,
-      "learning_rate": 6.081809045226131e-05,
-      "loss": 5.5714,
-      "step": 39500
-    },
-    {
-      "epoch": 20.60026075619296,
-      "grad_norm": 1.4815289974212646,
-      "learning_rate": 6.081708542713568e-05,
-      "loss": 5.0286,
-      "step": 39501
-    },
-    {
-      "epoch": 20.60078226857888,
-      "grad_norm": 1.54777991771698,
-      "learning_rate": 6.081608040201006e-05,
-      "loss": 4.8765,
-      "step": 39502
-    },
-    {
-      "epoch": 20.6013037809648,
-      "grad_norm": 1.4883688688278198,
-      "learning_rate": 6.081507537688442e-05,
-      "loss": 4.9519,
-      "step": 39503
-    },
-    {
-      "epoch": 20.601825293350718,
-      "grad_norm": 1.4861541986465454,
-      "learning_rate": 6.08140703517588e-05,
-      "loss": 5.4027,
-      "step": 39504
-    },
-    {
-      "epoch": 20.602346805736637,
-      "grad_norm": 1.6370915174484253,
-      "learning_rate": 6.081306532663317e-05,
-      "loss": 4.831,
-      "step": 39505
-    },
-    {
-      "epoch": 20.602868318122557,
-      "grad_norm": 1.5818172693252563,
-      "learning_rate": 6.081206030150755e-05,
-      "loss": 4.8733,
-      "step": 39506
-    },
-    {
-      "epoch": 20.603389830508476,
-      "grad_norm": 1.4497076272964478,
-      "learning_rate": 6.081105527638191e-05,
-      "loss": 5.6802,
-      "step": 39507
-    },
-    {
-      "epoch": 20.603911342894392,
-      "grad_norm": 1.494806170463562,
-      "learning_rate": 6.081005025125629e-05,
-      "loss": 5.1823,
-      "step": 39508
-    },
-    {
-      "epoch": 20.60443285528031,
-      "grad_norm": 1.5388914346694946,
-      "learning_rate": 6.0809045226130654e-05,
-      "loss": 4.8847,
-      "step": 39509
-    },
-    {
-      "epoch": 20.60495436766623,
-      "grad_norm": 1.6000969409942627,
-      "learning_rate": 6.0808040201005025e-05,
-      "loss": 5.1719,
-      "step": 39510
-    },
-    {
-      "epoch": 20.60547588005215,
-      "grad_norm": 1.5799751281738281,
-      "learning_rate": 6.08070351758794e-05,
-      "loss": 5.3492,
-      "step": 39511
-    },
-    {
-      "epoch": 20.60599739243807,
-      "grad_norm": 1.5413357019424438,
-      "learning_rate": 6.080603015075377e-05,
-      "loss": 4.9651,
-      "step": 39512
-    },
-    {
-      "epoch": 20.60651890482399,
-      "grad_norm": 1.460109829902649,
-      "learning_rate": 6.0805025125628145e-05,
-      "loss": 5.1348,
-      "step": 39513
-    },
-    {
-      "epoch": 20.60704041720991,
-      "grad_norm": 1.5084465742111206,
-      "learning_rate": 6.080402010050251e-05,
-      "loss": 4.9882,
-      "step": 39514
-    },
-    {
-      "epoch": 20.60756192959583,
-      "grad_norm": 1.46568763256073,
-      "learning_rate": 6.080301507537689e-05,
-      "loss": 4.6816,
-      "step": 39515
-    },
-    {
-      "epoch": 20.608083441981748,
-      "grad_norm": 1.523721694946289,
-      "learning_rate": 6.080201005025126e-05,
-      "loss": 5.0635,
-      "step": 39516
-    },
-    {
-      "epoch": 20.608604954367667,
-      "grad_norm": 1.568245768547058,
-      "learning_rate": 6.0801005025125636e-05,
-      "loss": 5.5582,
-      "step": 39517
-    },
-    {
-      "epoch": 20.609126466753587,
-      "grad_norm": 1.4532641172409058,
-      "learning_rate": 6.08e-05,
-      "loss": 5.6076,
-      "step": 39518
-    },
-    {
-      "epoch": 20.609647979139506,
-      "grad_norm": 1.527274250984192,
-      "learning_rate": 6.079899497487438e-05,
-      "loss": 5.5414,
-      "step": 39519
-    },
-    {
-      "epoch": 20.610169491525422,
-      "grad_norm": 1.5536152124404907,
-      "learning_rate": 6.079798994974874e-05,
-      "loss": 5.2574,
-      "step": 39520
-    },
-    {
-      "epoch": 20.61069100391134,
-      "grad_norm": 1.600590705871582,
-      "learning_rate": 6.079698492462312e-05,
-      "loss": 4.9268,
-      "step": 39521
-    },
-    {
-      "epoch": 20.61121251629726,
-      "grad_norm": 1.5556484460830688,
-      "learning_rate": 6.079597989949749e-05,
-      "loss": 4.7436,
-      "step": 39522
-    },
-    {
-      "epoch": 20.61173402868318,
-      "grad_norm": 1.600376844406128,
-      "learning_rate": 6.0794974874371856e-05,
-      "loss": 4.9797,
-      "step": 39523
-    },
-    {
-      "epoch": 20.6122555410691,
-      "grad_norm": 1.6710917949676514,
-      "learning_rate": 6.0793969849246234e-05,
-      "loss": 5.0929,
-      "step": 39524
-    },
-    {
-      "epoch": 20.61277705345502,
-      "grad_norm": 1.6319414377212524,
-      "learning_rate": 6.0792964824120605e-05,
-      "loss": 4.7066,
-      "step": 39525
-    },
-    {
-      "epoch": 20.61329856584094,
-      "grad_norm": 1.6205097436904907,
-      "learning_rate": 6.079195979899498e-05,
-      "loss": 5.2977,
-      "step": 39526
-    },
-    {
-      "epoch": 20.61382007822686,
-      "grad_norm": 1.4910601377487183,
-      "learning_rate": 6.079095477386935e-05,
-      "loss": 4.7023,
-      "step": 39527
-    },
-    {
-      "epoch": 20.614341590612778,
-      "grad_norm": 1.448333501815796,
-      "learning_rate": 6.0789949748743725e-05,
-      "loss": 5.8077,
-      "step": 39528
-    },
-    {
-      "epoch": 20.614863102998697,
-      "grad_norm": 1.457947850227356,
-      "learning_rate": 6.078894472361809e-05,
-      "loss": 5.5462,
-      "step": 39529
-    },
-    {
-      "epoch": 20.615384615384617,
-      "grad_norm": 1.4835091829299927,
-      "learning_rate": 6.078793969849247e-05,
-      "loss": 5.3286,
-      "step": 39530
-    },
-    {
-      "epoch": 20.615906127770536,
-      "grad_norm": 1.4405900239944458,
-      "learning_rate": 6.078693467336684e-05,
-      "loss": 5.3172,
-      "step": 39531
-    },
-    {
-      "epoch": 20.616427640156452,
-      "grad_norm": 1.4012694358825684,
-      "learning_rate": 6.0785929648241216e-05,
-      "loss": 5.3669,
-      "step": 39532
-    },
-    {
-      "epoch": 20.61694915254237,
-      "grad_norm": 1.3665969371795654,
-      "learning_rate": 6.078492462311558e-05,
-      "loss": 5.5389,
-      "step": 39533
-    },
-    {
-      "epoch": 20.61747066492829,
-      "grad_norm": 1.4401227235794067,
-      "learning_rate": 6.078391959798996e-05,
-      "loss": 5.5834,
-      "step": 39534
-    },
-    {
-      "epoch": 20.61799217731421,
-      "grad_norm": 1.44844388961792,
-      "learning_rate": 6.078291457286432e-05,
-      "loss": 5.2541,
-      "step": 39535
-    },
-    {
-      "epoch": 20.61851368970013,
-      "grad_norm": 1.4645276069641113,
-      "learning_rate": 6.078190954773869e-05,
-      "loss": 5.6591,
-      "step": 39536
-    },
-    {
-      "epoch": 20.61903520208605,
-      "grad_norm": 1.5074151754379272,
-      "learning_rate": 6.078090452261307e-05,
-      "loss": 5.1913,
-      "step": 39537
-    },
-    {
-      "epoch": 20.61955671447197,
-      "grad_norm": 1.5155527591705322,
-      "learning_rate": 6.0779899497487435e-05,
-      "loss": 5.3865,
-      "step": 39538
-    },
-    {
-      "epoch": 20.62007822685789,
-      "grad_norm": 1.4852958917617798,
-      "learning_rate": 6.077889447236181e-05,
-      "loss": 4.9842,
-      "step": 39539
-    },
-    {
-      "epoch": 20.620599739243808,
-      "grad_norm": 1.511531949043274,
-      "learning_rate": 6.077788944723618e-05,
-      "loss": 5.5712,
-      "step": 39540
-    },
-    {
-      "epoch": 20.621121251629727,
-      "grad_norm": 1.6284053325653076,
-      "learning_rate": 6.0776884422110555e-05,
-      "loss": 5.133,
-      "step": 39541
-    },
-    {
-      "epoch": 20.621642764015647,
-      "grad_norm": 1.4609575271606445,
-      "learning_rate": 6.0775879396984926e-05,
-      "loss": 5.5109,
-      "step": 39542
-    },
-    {
-      "epoch": 20.622164276401566,
-      "grad_norm": 1.5815978050231934,
-      "learning_rate": 6.0774874371859304e-05,
-      "loss": 5.4523,
-      "step": 39543
-    },
-    {
-      "epoch": 20.622685788787482,
-      "grad_norm": 1.497287631034851,
-      "learning_rate": 6.077386934673367e-05,
-      "loss": 5.4339,
-      "step": 39544
-    },
-    {
-      "epoch": 20.6232073011734,
-      "grad_norm": 1.506528615951538,
-      "learning_rate": 6.0772864321608046e-05,
-      "loss": 5.285,
-      "step": 39545
-    },
-    {
-      "epoch": 20.62372881355932,
-      "grad_norm": 1.5729997158050537,
-      "learning_rate": 6.077185929648242e-05,
-      "loss": 5.3299,
-      "step": 39546
-    },
-    {
-      "epoch": 20.62425032594524,
-      "grad_norm": 1.4790287017822266,
-      "learning_rate": 6.0770854271356795e-05,
-      "loss": 5.045,
-      "step": 39547
-    },
-    {
-      "epoch": 20.62477183833116,
-      "grad_norm": 1.4249814748764038,
-      "learning_rate": 6.076984924623116e-05,
-      "loss": 5.4296,
-      "step": 39548
-    },
-    {
-      "epoch": 20.62529335071708,
-      "grad_norm": 1.5219523906707764,
-      "learning_rate": 6.0768844221105524e-05,
-      "loss": 5.4245,
-      "step": 39549
-    },
-    {
-      "epoch": 20.625814863103,
-      "grad_norm": 1.627550482749939,
-      "learning_rate": 6.07678391959799e-05,
-      "loss": 5.2756,
-      "step": 39550
-    },
-    {
-      "epoch": 20.62633637548892,
-      "grad_norm": 1.5636066198349,
-      "learning_rate": 6.076683417085427e-05,
-      "loss": 5.5839,
-      "step": 39551
-    },
-    {
-      "epoch": 20.626857887874838,
-      "grad_norm": 1.5191186666488647,
-      "learning_rate": 6.076582914572865e-05,
-      "loss": 5.322,
-      "step": 39552
-    },
-    {
-      "epoch": 20.627379400260757,
-      "grad_norm": 1.4723035097122192,
-      "learning_rate": 6.0764824120603015e-05,
-      "loss": 5.5293,
-      "step": 39553
-    },
-    {
-      "epoch": 20.627900912646677,
-      "grad_norm": 1.5393054485321045,
-      "learning_rate": 6.076381909547739e-05,
-      "loss": 4.9926,
-      "step": 39554
-    },
-    {
-      "epoch": 20.628422425032596,
-      "grad_norm": 1.442798137664795,
-      "learning_rate": 6.076281407035176e-05,
-      "loss": 5.4889,
-      "step": 39555
-    },
-    {
-      "epoch": 20.628943937418512,
-      "grad_norm": 1.5653629302978516,
-      "learning_rate": 6.0761809045226134e-05,
-      "loss": 5.5575,
-      "step": 39556
-    },
-    {
-      "epoch": 20.62946544980443,
-      "grad_norm": 1.4968180656433105,
-      "learning_rate": 6.0760804020100506e-05,
-      "loss": 5.8928,
-      "step": 39557
-    },
-    {
-      "epoch": 20.62998696219035,
-      "grad_norm": 1.4395779371261597,
-      "learning_rate": 6.075979899497488e-05,
-      "loss": 5.3611,
-      "step": 39558
-    },
-    {
-      "epoch": 20.63050847457627,
-      "grad_norm": 1.4254486560821533,
-      "learning_rate": 6.075879396984925e-05,
-      "loss": 5.5934,
-      "step": 39559
-    },
-    {
-      "epoch": 20.63102998696219,
-      "grad_norm": 1.4557905197143555,
-      "learning_rate": 6.0757788944723625e-05,
-      "loss": 5.5971,
-      "step": 39560
-    },
-    {
-      "epoch": 20.63155149934811,
-      "grad_norm": 1.5105934143066406,
-      "learning_rate": 6.075678391959799e-05,
-      "loss": 4.9387,
-      "step": 39561
-    },
-    {
-      "epoch": 20.63207301173403,
-      "grad_norm": 1.5561039447784424,
-      "learning_rate": 6.075577889447236e-05,
-      "loss": 5.2759,
-      "step": 39562
-    },
-    {
-      "epoch": 20.63259452411995,
-      "grad_norm": 1.3952305316925049,
-      "learning_rate": 6.075477386934674e-05,
-      "loss": 4.9302,
-      "step": 39563
-    },
-    {
-      "epoch": 20.633116036505868,
-      "grad_norm": 1.5250401496887207,
-      "learning_rate": 6.07537688442211e-05,
-      "loss": 5.224,
-      "step": 39564
-    },
-    {
-      "epoch": 20.633637548891787,
-      "grad_norm": 1.6308919191360474,
-      "learning_rate": 6.075276381909548e-05,
-      "loss": 5.0653,
-      "step": 39565
-    },
-    {
-      "epoch": 20.634159061277707,
-      "grad_norm": 1.4848275184631348,
-      "learning_rate": 6.075175879396985e-05,
-      "loss": 4.702,
-      "step": 39566
-    },
-    {
-      "epoch": 20.634680573663623,
-      "grad_norm": 1.5058636665344238,
-      "learning_rate": 6.075075376884423e-05,
-      "loss": 5.1686,
-      "step": 39567
-    },
-    {
-      "epoch": 20.635202086049542,
-      "grad_norm": 1.5807693004608154,
-      "learning_rate": 6.0749748743718594e-05,
-      "loss": 5.3642,
-      "step": 39568
-    },
-    {
-      "epoch": 20.63572359843546,
-      "grad_norm": 1.400421380996704,
-      "learning_rate": 6.074874371859297e-05,
-      "loss": 5.8684,
-      "step": 39569
-    },
-    {
-      "epoch": 20.63624511082138,
-      "grad_norm": 1.5733561515808105,
-      "learning_rate": 6.0747738693467336e-05,
-      "loss": 5.7101,
-      "step": 39570
-    },
-    {
-      "epoch": 20.6367666232073,
-      "grad_norm": 1.549376130104065,
-      "learning_rate": 6.0746733668341714e-05,
-      "loss": 4.9195,
-      "step": 39571
-    },
-    {
-      "epoch": 20.63728813559322,
-      "grad_norm": 1.4177049398422241,
-      "learning_rate": 6.0745728643216085e-05,
-      "loss": 5.1644,
-      "step": 39572
-    },
-    {
-      "epoch": 20.63780964797914,
-      "grad_norm": 1.3962563276290894,
-      "learning_rate": 6.074472361809046e-05,
-      "loss": 5.0966,
-      "step": 39573
-    },
-    {
-      "epoch": 20.63833116036506,
-      "grad_norm": 1.5535407066345215,
-      "learning_rate": 6.074371859296483e-05,
-      "loss": 5.1353,
-      "step": 39574
-    },
-    {
-      "epoch": 20.63885267275098,
-      "grad_norm": 1.412893295288086,
-      "learning_rate": 6.074271356783919e-05,
-      "loss": 5.629,
-      "step": 39575
-    },
-    {
-      "epoch": 20.639374185136898,
-      "grad_norm": 1.6047130823135376,
-      "learning_rate": 6.074170854271357e-05,
-      "loss": 5.0091,
-      "step": 39576
-    },
-    {
-      "epoch": 20.639895697522817,
-      "grad_norm": 1.509798526763916,
-      "learning_rate": 6.074070351758794e-05,
-      "loss": 5.4926,
-      "step": 39577
-    },
-    {
-      "epoch": 20.640417209908737,
-      "grad_norm": 1.5203871726989746,
-      "learning_rate": 6.073969849246232e-05,
-      "loss": 5.5318,
-      "step": 39578
-    },
-    {
-      "epoch": 20.640938722294656,
-      "grad_norm": 1.4767358303070068,
-      "learning_rate": 6.073869346733668e-05,
-      "loss": 5.3676,
-      "step": 39579
-    },
-    {
-      "epoch": 20.641460234680572,
-      "grad_norm": 1.5503684282302856,
-      "learning_rate": 6.073768844221106e-05,
-      "loss": 5.1495,
-      "step": 39580
-    },
-    {
-      "epoch": 20.64198174706649,
-      "grad_norm": 1.5477228164672852,
-      "learning_rate": 6.0736683417085424e-05,
-      "loss": 5.2735,
-      "step": 39581
-    },
-    {
-      "epoch": 20.64250325945241,
-      "grad_norm": 1.5348738431930542,
-      "learning_rate": 6.07356783919598e-05,
-      "loss": 5.2717,
-      "step": 39582
-    },
-    {
-      "epoch": 20.64302477183833,
-      "grad_norm": 1.5274512767791748,
-      "learning_rate": 6.073467336683417e-05,
-      "loss": 5.3435,
-      "step": 39583
-    },
-    {
-      "epoch": 20.64354628422425,
-      "grad_norm": 1.4755600690841675,
-      "learning_rate": 6.073366834170855e-05,
-      "loss": 5.5555,
-      "step": 39584
-    },
-    {
-      "epoch": 20.64406779661017,
-      "grad_norm": 1.5469361543655396,
-      "learning_rate": 6.0732663316582915e-05,
-      "loss": 5.205,
-      "step": 39585
-    },
-    {
-      "epoch": 20.64458930899609,
-      "grad_norm": 1.611266851425171,
-      "learning_rate": 6.073165829145729e-05,
-      "loss": 5.1897,
-      "step": 39586
-    },
-    {
-      "epoch": 20.64511082138201,
-      "grad_norm": 1.4687286615371704,
-      "learning_rate": 6.073065326633166e-05,
-      "loss": 4.7972,
-      "step": 39587
-    },
-    {
-      "epoch": 20.645632333767928,
-      "grad_norm": 2.843839168548584,
-      "learning_rate": 6.0729648241206035e-05,
-      "loss": 5.1662,
-      "step": 39588
-    },
-    {
-      "epoch": 20.646153846153847,
-      "grad_norm": 1.5008503198623657,
-      "learning_rate": 6.0728643216080406e-05,
-      "loss": 5.3612,
-      "step": 39589
-    },
-    {
-      "epoch": 20.646675358539767,
-      "grad_norm": 1.557961106300354,
-      "learning_rate": 6.072763819095477e-05,
-      "loss": 5.3283,
-      "step": 39590
-    },
-    {
-      "epoch": 20.647196870925683,
-      "grad_norm": 1.5532467365264893,
-      "learning_rate": 6.072663316582915e-05,
-      "loss": 4.8042,
-      "step": 39591
-    },
-    {
-      "epoch": 20.647718383311602,
-      "grad_norm": 1.5034074783325195,
-      "learning_rate": 6.072562814070352e-05,
-      "loss": 4.9573,
-      "step": 39592
-    },
-    {
-      "epoch": 20.64823989569752,
-      "grad_norm": 1.5052039623260498,
-      "learning_rate": 6.07246231155779e-05,
-      "loss": 5.2592,
-      "step": 39593
-    },
-    {
-      "epoch": 20.64876140808344,
-      "grad_norm": 1.5911632776260376,
-      "learning_rate": 6.072361809045226e-05,
-      "loss": 5.1202,
-      "step": 39594
-    },
-    {
-      "epoch": 20.64928292046936,
-      "grad_norm": 1.5099008083343506,
-      "learning_rate": 6.072261306532664e-05,
-      "loss": 5.2463,
-      "step": 39595
-    },
-    {
-      "epoch": 20.64980443285528,
-      "grad_norm": 1.5546927452087402,
-      "learning_rate": 6.0721608040201004e-05,
-      "loss": 5.2759,
-      "step": 39596
-    },
-    {
-      "epoch": 20.6503259452412,
-      "grad_norm": 1.4716174602508545,
-      "learning_rate": 6.072060301507538e-05,
-      "loss": 5.5306,
-      "step": 39597
-    },
-    {
-      "epoch": 20.65084745762712,
-      "grad_norm": 1.476719856262207,
-      "learning_rate": 6.071959798994975e-05,
-      "loss": 5.0525,
-      "step": 39598
-    },
-    {
-      "epoch": 20.65136897001304,
-      "grad_norm": 1.4898656606674194,
-      "learning_rate": 6.071859296482413e-05,
-      "loss": 5.0215,
-      "step": 39599
-    },
-    {
-      "epoch": 20.651890482398958,
-      "grad_norm": 1.3997082710266113,
-      "learning_rate": 6.0717587939698495e-05,
-      "loss": 5.7421,
-      "step": 39600
-    },
-    {
-      "epoch": 20.652411994784877,
-      "grad_norm": 1.5321660041809082,
-      "learning_rate": 6.071658291457287e-05,
-      "loss": 5.2857,
-      "step": 39601
-    },
-    {
-      "epoch": 20.652933507170797,
-      "grad_norm": 1.5518457889556885,
-      "learning_rate": 6.071557788944724e-05,
-      "loss": 4.5992,
-      "step": 39602
-    },
-    {
-      "epoch": 20.653455019556713,
-      "grad_norm": 1.588607668876648,
-      "learning_rate": 6.071457286432161e-05,
-      "loss": 5.1962,
-      "step": 39603
-    },
-    {
-      "epoch": 20.653976531942632,
-      "grad_norm": 1.6560239791870117,
-      "learning_rate": 6.0713567839195986e-05,
-      "loss": 5.0278,
-      "step": 39604
-    },
-    {
-      "epoch": 20.65449804432855,
-      "grad_norm": 1.4610310792922974,
-      "learning_rate": 6.071256281407035e-05,
-      "loss": 5.6813,
-      "step": 39605
-    },
-    {
-      "epoch": 20.65501955671447,
-      "grad_norm": 1.6536736488342285,
-      "learning_rate": 6.071155778894473e-05,
-      "loss": 5.2484,
-      "step": 39606
-    },
-    {
-      "epoch": 20.65554106910039,
-      "grad_norm": 1.5251692533493042,
-      "learning_rate": 6.071055276381909e-05,
-      "loss": 5.5003,
-      "step": 39607
-    },
-    {
-      "epoch": 20.65606258148631,
-      "grad_norm": 1.461708426475525,
-      "learning_rate": 6.070954773869347e-05,
-      "loss": 5.2168,
-      "step": 39608
-    },
-    {
-      "epoch": 20.65658409387223,
-      "grad_norm": 1.4689382314682007,
-      "learning_rate": 6.070854271356784e-05,
-      "loss": 5.2112,
-      "step": 39609
-    },
-    {
-      "epoch": 20.65710560625815,
-      "grad_norm": 1.552442193031311,
-      "learning_rate": 6.070753768844222e-05,
-      "loss": 4.8881,
-      "step": 39610
-    },
-    {
-      "epoch": 20.65762711864407,
-      "grad_norm": 1.8615492582321167,
-      "learning_rate": 6.070653266331658e-05,
-      "loss": 5.1729,
-      "step": 39611
-    },
-    {
-      "epoch": 20.658148631029988,
-      "grad_norm": 1.5445365905761719,
-      "learning_rate": 6.070552763819096e-05,
-      "loss": 5.2546,
-      "step": 39612
-    },
-    {
-      "epoch": 20.658670143415907,
-      "grad_norm": 1.5995570421218872,
-      "learning_rate": 6.070452261306533e-05,
-      "loss": 5.6215,
-      "step": 39613
-    },
-    {
-      "epoch": 20.659191655801827,
-      "grad_norm": 1.5695749521255493,
-      "learning_rate": 6.070351758793971e-05,
-      "loss": 4.6172,
-      "step": 39614
-    },
-    {
-      "epoch": 20.659713168187743,
-      "grad_norm": 1.6100183725357056,
-      "learning_rate": 6.0702512562814074e-05,
-      "loss": 4.8409,
-      "step": 39615
-    },
-    {
-      "epoch": 20.660234680573662,
-      "grad_norm": 1.5412354469299316,
-      "learning_rate": 6.070150753768844e-05,
-      "loss": 5.4417,
-      "step": 39616
-    },
-    {
-      "epoch": 20.66075619295958,
-      "grad_norm": 1.5960217714309692,
-      "learning_rate": 6.0700502512562816e-05,
-      "loss": 5.077,
-      "step": 39617
-    },
-    {
-      "epoch": 20.6612777053455,
-      "grad_norm": 1.4516631364822388,
-      "learning_rate": 6.069949748743719e-05,
-      "loss": 5.6287,
-      "step": 39618
-    },
-    {
-      "epoch": 20.66179921773142,
-      "grad_norm": 1.5934832096099854,
-      "learning_rate": 6.0698492462311565e-05,
-      "loss": 5.3119,
-      "step": 39619
-    },
-    {
-      "epoch": 20.66232073011734,
-      "grad_norm": 1.5315513610839844,
-      "learning_rate": 6.069748743718593e-05,
-      "loss": 5.1281,
-      "step": 39620
-    },
-    {
-      "epoch": 20.66284224250326,
-      "grad_norm": 1.5744798183441162,
-      "learning_rate": 6.069648241206031e-05,
-      "loss": 5.7684,
-      "step": 39621
-    },
-    {
-      "epoch": 20.66336375488918,
-      "grad_norm": 1.48493230342865,
-      "learning_rate": 6.069547738693467e-05,
-      "loss": 5.0384,
-      "step": 39622
-    },
-    {
-      "epoch": 20.6638852672751,
-      "grad_norm": 1.6417945623397827,
-      "learning_rate": 6.069447236180905e-05,
-      "loss": 4.8651,
-      "step": 39623
-    },
-    {
-      "epoch": 20.664406779661018,
-      "grad_norm": 1.575063943862915,
-      "learning_rate": 6.069346733668342e-05,
-      "loss": 5.1701,
-      "step": 39624
-    },
-    {
-      "epoch": 20.664928292046937,
-      "grad_norm": 1.5158315896987915,
-      "learning_rate": 6.06924623115578e-05,
-      "loss": 5.2761,
-      "step": 39625
-    },
-    {
-      "epoch": 20.665449804432857,
-      "grad_norm": 1.4806312322616577,
-      "learning_rate": 6.069145728643216e-05,
-      "loss": 5.2551,
-      "step": 39626
-    },
-    {
-      "epoch": 20.665971316818773,
-      "grad_norm": 1.5863522291183472,
-      "learning_rate": 6.069045226130654e-05,
-      "loss": 4.9347,
-      "step": 39627
-    },
-    {
-      "epoch": 20.666492829204692,
-      "grad_norm": 1.6204432249069214,
-      "learning_rate": 6.0689447236180904e-05,
-      "loss": 5.3022,
-      "step": 39628
-    },
-    {
-      "epoch": 20.667014341590612,
-      "grad_norm": 1.4518715143203735,
-      "learning_rate": 6.0688442211055275e-05,
-      "loss": 5.591,
-      "step": 39629
-    },
-    {
-      "epoch": 20.66753585397653,
-      "grad_norm": 1.4831781387329102,
-      "learning_rate": 6.068743718592965e-05,
-      "loss": 4.9552,
-      "step": 39630
-    },
-    {
-      "epoch": 20.66805736636245,
-      "grad_norm": 1.4503012895584106,
-      "learning_rate": 6.068643216080402e-05,
-      "loss": 4.9653,
-      "step": 39631
-    },
-    {
-      "epoch": 20.66857887874837,
-      "grad_norm": 1.5947444438934326,
-      "learning_rate": 6.0685427135678395e-05,
-      "loss": 5.4755,
-      "step": 39632
-    },
-    {
-      "epoch": 20.66910039113429,
-      "grad_norm": 1.6464022397994995,
-      "learning_rate": 6.0684422110552766e-05,
-      "loss": 4.9038,
-      "step": 39633
-    },
-    {
-      "epoch": 20.66962190352021,
-      "grad_norm": 1.7467135190963745,
-      "learning_rate": 6.0683417085427144e-05,
-      "loss": 5.139,
-      "step": 39634
-    },
-    {
-      "epoch": 20.67014341590613,
-      "grad_norm": 1.4447052478790283,
-      "learning_rate": 6.068241206030151e-05,
-      "loss": 5.5703,
-      "step": 39635
-    },
-    {
-      "epoch": 20.670664928292048,
-      "grad_norm": 1.4708151817321777,
-      "learning_rate": 6.0681407035175886e-05,
-      "loss": 5.3737,
-      "step": 39636
-    },
-    {
-      "epoch": 20.671186440677968,
-      "grad_norm": 1.4892035722732544,
-      "learning_rate": 6.068040201005025e-05,
-      "loss": 4.665,
-      "step": 39637
-    },
-    {
-      "epoch": 20.671707953063887,
-      "grad_norm": 1.614014744758606,
-      "learning_rate": 6.067939698492463e-05,
-      "loss": 5.1746,
-      "step": 39638
-    },
-    {
-      "epoch": 20.672229465449803,
-      "grad_norm": 1.3913809061050415,
-      "learning_rate": 6.0678391959799e-05,
-      "loss": 5.2747,
-      "step": 39639
-    },
-    {
-      "epoch": 20.672750977835722,
-      "grad_norm": 1.4664220809936523,
-      "learning_rate": 6.067738693467338e-05,
-      "loss": 5.5587,
-      "step": 39640
-    },
-    {
-      "epoch": 20.673272490221642,
-      "grad_norm": 1.5246155261993408,
-      "learning_rate": 6.067638190954774e-05,
-      "loss": 5.223,
-      "step": 39641
-    },
-    {
-      "epoch": 20.67379400260756,
-      "grad_norm": 1.5164976119995117,
-      "learning_rate": 6.0675376884422106e-05,
-      "loss": 5.3623,
-      "step": 39642
-    },
-    {
-      "epoch": 20.67431551499348,
-      "grad_norm": 1.5193370580673218,
-      "learning_rate": 6.0674371859296484e-05,
-      "loss": 5.215,
-      "step": 39643
-    },
-    {
-      "epoch": 20.6748370273794,
-      "grad_norm": 1.384103536605835,
-      "learning_rate": 6.0673366834170855e-05,
-      "loss": 4.7754,
-      "step": 39644
-    },
-    {
-      "epoch": 20.67535853976532,
-      "grad_norm": 1.607710599899292,
-      "learning_rate": 6.067236180904523e-05,
-      "loss": 5.1621,
-      "step": 39645
-    },
-    {
-      "epoch": 20.67588005215124,
-      "grad_norm": 1.5798412561416626,
-      "learning_rate": 6.06713567839196e-05,
-      "loss": 4.9091,
-      "step": 39646
-    },
-    {
-      "epoch": 20.67640156453716,
-      "grad_norm": 1.4984991550445557,
-      "learning_rate": 6.0670351758793975e-05,
-      "loss": 5.3283,
-      "step": 39647
-    },
-    {
-      "epoch": 20.676923076923078,
-      "grad_norm": 1.456598162651062,
-      "learning_rate": 6.066934673366834e-05,
-      "loss": 5.2855,
-      "step": 39648
-    },
-    {
-      "epoch": 20.677444589308998,
-      "grad_norm": 1.545089602470398,
-      "learning_rate": 6.066834170854272e-05,
-      "loss": 5.2577,
-      "step": 39649
-    },
-    {
-      "epoch": 20.677966101694913,
-      "grad_norm": 1.6115305423736572,
-      "learning_rate": 6.066733668341709e-05,
-      "loss": 4.7761,
-      "step": 39650
-    },
-    {
-      "epoch": 20.678487614080833,
-      "grad_norm": 1.640951156616211,
-      "learning_rate": 6.0666331658291466e-05,
-      "loss": 5.0087,
-      "step": 39651
-    },
-    {
-      "epoch": 20.679009126466752,
-      "grad_norm": 1.422574520111084,
-      "learning_rate": 6.066532663316583e-05,
-      "loss": 5.6314,
-      "step": 39652
-    },
-    {
-      "epoch": 20.679530638852672,
-      "grad_norm": 1.5351743698120117,
-      "learning_rate": 6.066432160804021e-05,
-      "loss": 5.6608,
-      "step": 39653
-    },
-    {
-      "epoch": 20.68005215123859,
-      "grad_norm": 1.5877008438110352,
-      "learning_rate": 6.066331658291458e-05,
-      "loss": 5.2251,
-      "step": 39654
-    },
-    {
-      "epoch": 20.68057366362451,
-      "grad_norm": 1.6308757066726685,
-      "learning_rate": 6.066231155778894e-05,
-      "loss": 5.3004,
-      "step": 39655
-    },
-    {
-      "epoch": 20.68109517601043,
-      "grad_norm": 1.4891036748886108,
-      "learning_rate": 6.066130653266332e-05,
-      "loss": 5.2709,
-      "step": 39656
-    },
-    {
-      "epoch": 20.68161668839635,
-      "grad_norm": 1.5929555892944336,
-      "learning_rate": 6.0660301507537685e-05,
-      "loss": 5.1301,
-      "step": 39657
-    },
-    {
-      "epoch": 20.68213820078227,
-      "grad_norm": 1.4326093196868896,
-      "learning_rate": 6.065929648241206e-05,
-      "loss": 5.3955,
-      "step": 39658
-    },
-    {
-      "epoch": 20.68265971316819,
-      "grad_norm": 1.4973465204238892,
-      "learning_rate": 6.0658291457286434e-05,
-      "loss": 5.5967,
-      "step": 39659
-    },
-    {
-      "epoch": 20.683181225554108,
-      "grad_norm": 1.519844889640808,
-      "learning_rate": 6.065728643216081e-05,
-      "loss": 5.5867,
-      "step": 39660
-    },
-    {
-      "epoch": 20.683702737940028,
-      "grad_norm": 1.4781899452209473,
-      "learning_rate": 6.0656281407035176e-05,
-      "loss": 5.2477,
-      "step": 39661
-    },
-    {
-      "epoch": 20.684224250325947,
-      "grad_norm": 1.4774765968322754,
-      "learning_rate": 6.0655276381909554e-05,
-      "loss": 5.0375,
-      "step": 39662
-    },
-    {
-      "epoch": 20.684745762711863,
-      "grad_norm": 1.664771556854248,
-      "learning_rate": 6.065427135678392e-05,
-      "loss": 5.1633,
-      "step": 39663
-    },
-    {
-      "epoch": 20.685267275097782,
-      "grad_norm": 1.452271580696106,
-      "learning_rate": 6.0653266331658296e-05,
-      "loss": 5.2912,
-      "step": 39664
-    },
-    {
-      "epoch": 20.685788787483702,
-      "grad_norm": 1.5154359340667725,
-      "learning_rate": 6.065226130653267e-05,
-      "loss": 5.2144,
-      "step": 39665
-    },
-    {
-      "epoch": 20.68631029986962,
-      "grad_norm": 1.5261255502700806,
-      "learning_rate": 6.0651256281407045e-05,
-      "loss": 5.1983,
-      "step": 39666
-    },
-    {
-      "epoch": 20.68683181225554,
-      "grad_norm": 1.4827284812927246,
-      "learning_rate": 6.065025125628141e-05,
-      "loss": 5.6337,
-      "step": 39667
-    },
-    {
-      "epoch": 20.68735332464146,
-      "grad_norm": 1.419244647026062,
-      "learning_rate": 6.0649246231155774e-05,
-      "loss": 5.6312,
-      "step": 39668
-    },
-    {
-      "epoch": 20.68787483702738,
-      "grad_norm": 1.500741958618164,
-      "learning_rate": 6.064824120603015e-05,
-      "loss": 5.3013,
-      "step": 39669
-    },
-    {
-      "epoch": 20.6883963494133,
-      "grad_norm": 1.5659035444259644,
-      "learning_rate": 6.064723618090452e-05,
-      "loss": 5.4496,
-      "step": 39670
-    },
-    {
-      "epoch": 20.68891786179922,
-      "grad_norm": 1.4246485233306885,
-      "learning_rate": 6.06462311557789e-05,
-      "loss": 5.8158,
-      "step": 39671
-    },
-    {
-      "epoch": 20.689439374185138,
-      "grad_norm": 1.5327054262161255,
-      "learning_rate": 6.0645226130653265e-05,
-      "loss": 5.5889,
-      "step": 39672
-    },
-    {
-      "epoch": 20.689960886571058,
-      "grad_norm": 1.4768836498260498,
-      "learning_rate": 6.064422110552764e-05,
-      "loss": 5.3547,
-      "step": 39673
-    },
-    {
-      "epoch": 20.690482398956973,
-      "grad_norm": 1.4829355478286743,
-      "learning_rate": 6.064321608040201e-05,
-      "loss": 5.0484,
-      "step": 39674
-    },
-    {
-      "epoch": 20.691003911342893,
-      "grad_norm": 1.4919958114624023,
-      "learning_rate": 6.0642211055276384e-05,
-      "loss": 4.9077,
-      "step": 39675
-    },
-    {
-      "epoch": 20.691525423728812,
-      "grad_norm": 1.4404346942901611,
-      "learning_rate": 6.0641206030150756e-05,
-      "loss": 5.7267,
-      "step": 39676
-    },
-    {
-      "epoch": 20.692046936114732,
-      "grad_norm": 1.586681604385376,
-      "learning_rate": 6.064020100502513e-05,
-      "loss": 5.1418,
-      "step": 39677
-    },
-    {
-      "epoch": 20.69256844850065,
-      "grad_norm": 1.4936484098434448,
-      "learning_rate": 6.06391959798995e-05,
-      "loss": 5.6661,
-      "step": 39678
-    },
-    {
-      "epoch": 20.69308996088657,
-      "grad_norm": 1.5253725051879883,
-      "learning_rate": 6.0638190954773875e-05,
-      "loss": 5.4229,
-      "step": 39679
-    },
-    {
-      "epoch": 20.69361147327249,
-      "grad_norm": 1.5829837322235107,
-      "learning_rate": 6.0637185929648247e-05,
-      "loss": 5.115,
-      "step": 39680
-    },
-    {
-      "epoch": 20.69413298565841,
-      "grad_norm": 1.5386154651641846,
-      "learning_rate": 6.0636180904522624e-05,
-      "loss": 5.4345,
-      "step": 39681
-    },
-    {
-      "epoch": 20.69465449804433,
-      "grad_norm": 1.4490097761154175,
-      "learning_rate": 6.063517587939699e-05,
-      "loss": 5.1775,
-      "step": 39682
-    },
-    {
-      "epoch": 20.69517601043025,
-      "grad_norm": 1.5062490701675415,
-      "learning_rate": 6.063417085427135e-05,
-      "loss": 5.4783,
-      "step": 39683
-    },
-    {
-      "epoch": 20.695697522816168,
-      "grad_norm": 1.6646472215652466,
-      "learning_rate": 6.063316582914573e-05,
-      "loss": 5.4656,
-      "step": 39684
-    },
-    {
-      "epoch": 20.696219035202088,
-      "grad_norm": 1.5669032335281372,
-      "learning_rate": 6.06321608040201e-05,
-      "loss": 5.1794,
-      "step": 39685
-    },
-    {
-      "epoch": 20.696740547588004,
-      "grad_norm": 1.6104190349578857,
-      "learning_rate": 6.063115577889448e-05,
-      "loss": 4.7927,
-      "step": 39686
-    },
-    {
-      "epoch": 20.697262059973923,
-      "grad_norm": 1.4725980758666992,
-      "learning_rate": 6.0630150753768844e-05,
-      "loss": 5.1545,
-      "step": 39687
-    },
-    {
-      "epoch": 20.697783572359842,
-      "grad_norm": 1.4490851163864136,
-      "learning_rate": 6.062914572864322e-05,
-      "loss": 5.4966,
-      "step": 39688
-    },
-    {
-      "epoch": 20.698305084745762,
-      "grad_norm": 1.5122294425964355,
-      "learning_rate": 6.0628140703517586e-05,
-      "loss": 5.4112,
-      "step": 39689
-    },
-    {
-      "epoch": 20.69882659713168,
-      "grad_norm": 1.3854050636291504,
-      "learning_rate": 6.0627135678391964e-05,
-      "loss": 5.6632,
-      "step": 39690
-    },
-    {
-      "epoch": 20.6993481095176,
-      "grad_norm": 1.464016079902649,
-      "learning_rate": 6.0626130653266335e-05,
-      "loss": 5.1103,
-      "step": 39691
-    },
-    {
-      "epoch": 20.69986962190352,
-      "grad_norm": 1.5136443376541138,
-      "learning_rate": 6.062512562814071e-05,
-      "loss": 5.4228,
-      "step": 39692
-    },
-    {
-      "epoch": 20.70039113428944,
-      "grad_norm": 1.6263755559921265,
-      "learning_rate": 6.062412060301508e-05,
-      "loss": 5.2283,
-      "step": 39693
-    },
-    {
-      "epoch": 20.70091264667536,
-      "grad_norm": 1.5163757801055908,
-      "learning_rate": 6.0623115577889455e-05,
-      "loss": 5.1964,
-      "step": 39694
-    },
-    {
-      "epoch": 20.70143415906128,
-      "grad_norm": 1.481034755706787,
-      "learning_rate": 6.062211055276382e-05,
-      "loss": 5.2653,
-      "step": 39695
-    },
-    {
-      "epoch": 20.701955671447198,
-      "grad_norm": 1.5828278064727783,
-      "learning_rate": 6.062110552763819e-05,
-      "loss": 5.4427,
-      "step": 39696
-    },
-    {
-      "epoch": 20.702477183833118,
-      "grad_norm": 1.5008360147476196,
-      "learning_rate": 6.062010050251257e-05,
-      "loss": 4.5259,
-      "step": 39697
-    },
-    {
-      "epoch": 20.702998696219034,
-      "grad_norm": 1.4895949363708496,
-      "learning_rate": 6.061909547738693e-05,
-      "loss": 5.6325,
-      "step": 39698
-    },
-    {
-      "epoch": 20.703520208604953,
-      "grad_norm": 1.436837911605835,
-      "learning_rate": 6.061809045226131e-05,
-      "loss": 5.627,
-      "step": 39699
-    },
-    {
-      "epoch": 20.704041720990872,
-      "grad_norm": 1.5880860090255737,
-      "learning_rate": 6.061708542713568e-05,
-      "loss": 4.9563,
-      "step": 39700
-    },
-    {
-      "epoch": 20.704563233376792,
-      "grad_norm": 1.4658045768737793,
-      "learning_rate": 6.061608040201006e-05,
-      "loss": 5.315,
-      "step": 39701
-    },
-    {
-      "epoch": 20.70508474576271,
-      "grad_norm": 1.5733054876327515,
-      "learning_rate": 6.061507537688442e-05,
-      "loss": 5.1037,
-      "step": 39702
-    },
-    {
-      "epoch": 20.70560625814863,
-      "grad_norm": 1.5945769548416138,
-      "learning_rate": 6.06140703517588e-05,
-      "loss": 4.8891,
-      "step": 39703
-    },
-    {
-      "epoch": 20.70612777053455,
-      "grad_norm": 1.5164551734924316,
-      "learning_rate": 6.0613065326633165e-05,
-      "loss": 4.8761,
-      "step": 39704
-    },
-    {
-      "epoch": 20.70664928292047,
-      "grad_norm": 1.5165902376174927,
-      "learning_rate": 6.061206030150754e-05,
-      "loss": 5.2305,
-      "step": 39705
-    },
-    {
-      "epoch": 20.70717079530639,
-      "grad_norm": 1.4905699491500854,
-      "learning_rate": 6.0611055276381914e-05,
-      "loss": 5.3704,
-      "step": 39706
-    },
-    {
-      "epoch": 20.70769230769231,
-      "grad_norm": 1.4542635679244995,
-      "learning_rate": 6.061005025125629e-05,
-      "loss": 5.2978,
-      "step": 39707
-    },
-    {
-      "epoch": 20.708213820078228,
-      "grad_norm": 1.5128779411315918,
-      "learning_rate": 6.0609045226130656e-05,
-      "loss": 5.4578,
-      "step": 39708
-    },
-    {
-      "epoch": 20.708735332464148,
-      "grad_norm": 1.486722469329834,
-      "learning_rate": 6.060804020100502e-05,
-      "loss": 4.7348,
-      "step": 39709
-    },
-    {
-      "epoch": 20.709256844850064,
-      "grad_norm": 1.461548089981079,
-      "learning_rate": 6.06070351758794e-05,
-      "loss": 5.5692,
-      "step": 39710
-    },
-    {
-      "epoch": 20.709778357235983,
-      "grad_norm": 1.4298595190048218,
-      "learning_rate": 6.060603015075377e-05,
-      "loss": 5.455,
-      "step": 39711
-    },
-    {
-      "epoch": 20.710299869621903,
-      "grad_norm": 1.5788531303405762,
-      "learning_rate": 6.060502512562815e-05,
-      "loss": 4.9828,
-      "step": 39712
-    },
-    {
-      "epoch": 20.710821382007822,
-      "grad_norm": 1.4364399909973145,
-      "learning_rate": 6.060402010050251e-05,
-      "loss": 5.6243,
-      "step": 39713
-    },
-    {
-      "epoch": 20.71134289439374,
-      "grad_norm": 1.448679804801941,
-      "learning_rate": 6.060301507537689e-05,
-      "loss": 5.2822,
-      "step": 39714
-    },
-    {
-      "epoch": 20.71186440677966,
-      "grad_norm": 1.632962703704834,
-      "learning_rate": 6.0602010050251254e-05,
-      "loss": 5.2341,
-      "step": 39715
-    },
-    {
-      "epoch": 20.71238591916558,
-      "grad_norm": 1.5599056482315063,
-      "learning_rate": 6.060100502512563e-05,
-      "loss": 5.524,
-      "step": 39716
-    },
-    {
-      "epoch": 20.7129074315515,
-      "grad_norm": 1.5329740047454834,
-      "learning_rate": 6.06e-05,
-      "loss": 5.3756,
-      "step": 39717
-    },
-    {
-      "epoch": 20.71342894393742,
-      "grad_norm": 1.4626308679580688,
-      "learning_rate": 6.059899497487438e-05,
-      "loss": 5.1181,
-      "step": 39718
-    },
-    {
-      "epoch": 20.71395045632334,
-      "grad_norm": 1.4925694465637207,
-      "learning_rate": 6.0597989949748745e-05,
-      "loss": 5.2811,
-      "step": 39719
-    },
-    {
-      "epoch": 20.714471968709258,
-      "grad_norm": 1.5021424293518066,
-      "learning_rate": 6.059698492462312e-05,
-      "loss": 5.3948,
-      "step": 39720
-    },
-    {
-      "epoch": 20.714993481095178,
-      "grad_norm": 1.6545783281326294,
-      "learning_rate": 6.0595979899497494e-05,
-      "loss": 4.6826,
-      "step": 39721
-    },
-    {
-      "epoch": 20.715514993481094,
-      "grad_norm": 1.5768743753433228,
-      "learning_rate": 6.059497487437186e-05,
-      "loss": 5.5131,
-      "step": 39722
-    },
-    {
-      "epoch": 20.716036505867013,
-      "grad_norm": 1.390627145767212,
-      "learning_rate": 6.0593969849246236e-05,
-      "loss": 5.6471,
-      "step": 39723
-    },
-    {
-      "epoch": 20.716558018252933,
-      "grad_norm": 1.5367443561553955,
-      "learning_rate": 6.05929648241206e-05,
-      "loss": 4.9448,
-      "step": 39724
-    },
-    {
-      "epoch": 20.717079530638852,
-      "grad_norm": 1.5401952266693115,
-      "learning_rate": 6.059195979899498e-05,
-      "loss": 5.3741,
-      "step": 39725
-    },
-    {
-      "epoch": 20.71760104302477,
-      "grad_norm": 1.6512949466705322,
-      "learning_rate": 6.059095477386935e-05,
-      "loss": 5.0313,
-      "step": 39726
-    },
-    {
-      "epoch": 20.71812255541069,
-      "grad_norm": 1.5453531742095947,
-      "learning_rate": 6.058994974874373e-05,
-      "loss": 4.8017,
-      "step": 39727
-    },
-    {
-      "epoch": 20.71864406779661,
-      "grad_norm": 1.5704362392425537,
-      "learning_rate": 6.058894472361809e-05,
-      "loss": 5.5133,
-      "step": 39728
-    },
-    {
-      "epoch": 20.71916558018253,
-      "grad_norm": 1.4298028945922852,
-      "learning_rate": 6.058793969849247e-05,
-      "loss": 5.2659,
-      "step": 39729
-    },
-    {
-      "epoch": 20.71968709256845,
-      "grad_norm": 1.5677509307861328,
-      "learning_rate": 6.058693467336683e-05,
-      "loss": 5.5212,
-      "step": 39730
-    },
-    {
-      "epoch": 20.72020860495437,
-      "grad_norm": 1.5907855033874512,
-      "learning_rate": 6.058592964824121e-05,
-      "loss": 5.4314,
-      "step": 39731
-    },
-    {
-      "epoch": 20.72073011734029,
-      "grad_norm": 1.4736322164535522,
-      "learning_rate": 6.058492462311558e-05,
-      "loss": 4.6506,
-      "step": 39732
-    },
-    {
-      "epoch": 20.721251629726208,
-      "grad_norm": 1.6135705709457397,
-      "learning_rate": 6.058391959798996e-05,
-      "loss": 5.0766,
-      "step": 39733
-    },
-    {
-      "epoch": 20.721773142112124,
-      "grad_norm": 1.609515905380249,
-      "learning_rate": 6.0582914572864324e-05,
-      "loss": 5.2169,
-      "step": 39734
-    },
-    {
-      "epoch": 20.722294654498043,
-      "grad_norm": 1.5136406421661377,
-      "learning_rate": 6.058190954773869e-05,
-      "loss": 5.4938,
-      "step": 39735
-    },
-    {
-      "epoch": 20.722816166883963,
-      "grad_norm": 1.3771679401397705,
-      "learning_rate": 6.0580904522613066e-05,
-      "loss": 5.7415,
-      "step": 39736
-    },
-    {
-      "epoch": 20.723337679269882,
-      "grad_norm": 1.4625364542007446,
-      "learning_rate": 6.057989949748744e-05,
-      "loss": 5.3351,
-      "step": 39737
-    },
-    {
-      "epoch": 20.7238591916558,
-      "grad_norm": 1.4680311679840088,
-      "learning_rate": 6.0578894472361815e-05,
-      "loss": 5.7621,
-      "step": 39738
-    },
-    {
-      "epoch": 20.72438070404172,
-      "grad_norm": 1.4878965616226196,
-      "learning_rate": 6.057788944723618e-05,
-      "loss": 5.4067,
-      "step": 39739
-    },
-    {
-      "epoch": 20.72490221642764,
-      "grad_norm": 1.3818632364273071,
-      "learning_rate": 6.057688442211056e-05,
-      "loss": 5.5887,
-      "step": 39740
-    },
-    {
-      "epoch": 20.72542372881356,
-      "grad_norm": 1.4431909322738647,
-      "learning_rate": 6.057587939698493e-05,
-      "loss": 5.2527,
-      "step": 39741
-    },
-    {
-      "epoch": 20.72594524119948,
-      "grad_norm": 1.433445930480957,
-      "learning_rate": 6.0574874371859306e-05,
-      "loss": 5.1012,
-      "step": 39742
-    },
-    {
-      "epoch": 20.7264667535854,
-      "grad_norm": 1.4869837760925293,
-      "learning_rate": 6.057386934673367e-05,
-      "loss": 5.2846,
-      "step": 39743
-    },
-    {
-      "epoch": 20.72698826597132,
-      "grad_norm": 1.55743408203125,
-      "learning_rate": 6.057286432160805e-05,
-      "loss": 5.0319,
-      "step": 39744
-    },
-    {
-      "epoch": 20.727509778357238,
-      "grad_norm": 1.4954663515090942,
-      "learning_rate": 6.057185929648241e-05,
-      "loss": 5.3212,
-      "step": 39745
-    },
-    {
-      "epoch": 20.728031290743154,
-      "grad_norm": 1.5554485321044922,
-      "learning_rate": 6.057085427135679e-05,
-      "loss": 4.932,
-      "step": 39746
-    },
-    {
-      "epoch": 20.728552803129073,
-      "grad_norm": 1.6133321523666382,
-      "learning_rate": 6.056984924623116e-05,
-      "loss": 5.2647,
-      "step": 39747
-    },
-    {
-      "epoch": 20.729074315514993,
-      "grad_norm": 1.5690889358520508,
-      "learning_rate": 6.0568844221105525e-05,
-      "loss": 4.9635,
-      "step": 39748
-    },
-    {
-      "epoch": 20.729595827900912,
-      "grad_norm": 1.571123719215393,
-      "learning_rate": 6.05678391959799e-05,
-      "loss": 4.5981,
-      "step": 39749
-    },
-    {
-      "epoch": 20.73011734028683,
-      "grad_norm": 1.5257474184036255,
-      "learning_rate": 6.056683417085427e-05,
-      "loss": 5.3633,
-      "step": 39750
-    },
-    {
-      "epoch": 20.73063885267275,
-      "grad_norm": 1.4855064153671265,
-      "learning_rate": 6.0565829145728645e-05,
-      "loss": 5.6341,
-      "step": 39751
-    },
-    {
-      "epoch": 20.73116036505867,
-      "grad_norm": 1.6079655885696411,
-      "learning_rate": 6.0564824120603016e-05,
-      "loss": 5.0567,
-      "step": 39752
-    },
-    {
-      "epoch": 20.73168187744459,
-      "grad_norm": 1.5049333572387695,
-      "learning_rate": 6.0563819095477394e-05,
-      "loss": 5.5275,
-      "step": 39753
-    },
-    {
-      "epoch": 20.73220338983051,
-      "grad_norm": 1.526338815689087,
-      "learning_rate": 6.056281407035176e-05,
-      "loss": 5.2811,
-      "step": 39754
-    },
-    {
-      "epoch": 20.73272490221643,
-      "grad_norm": 1.5201102495193481,
-      "learning_rate": 6.0561809045226136e-05,
-      "loss": 5.5283,
-      "step": 39755
-    },
-    {
-      "epoch": 20.73324641460235,
-      "grad_norm": 1.6012699604034424,
-      "learning_rate": 6.05608040201005e-05,
-      "loss": 5.2551,
-      "step": 39756
-    },
-    {
-      "epoch": 20.733767926988264,
-      "grad_norm": 1.4903942346572876,
-      "learning_rate": 6.055979899497488e-05,
-      "loss": 5.5,
-      "step": 39757
-    },
-    {
-      "epoch": 20.734289439374184,
-      "grad_norm": 1.6086437702178955,
-      "learning_rate": 6.055879396984925e-05,
-      "loss": 5.5873,
-      "step": 39758
-    },
-    {
-      "epoch": 20.734810951760103,
-      "grad_norm": 1.5619721412658691,
-      "learning_rate": 6.055778894472363e-05,
-      "loss": 4.9673,
-      "step": 39759
-    },
-    {
-      "epoch": 20.735332464146023,
-      "grad_norm": 1.5650744438171387,
-      "learning_rate": 6.055678391959799e-05,
-      "loss": 4.3292,
-      "step": 39760
-    },
-    {
-      "epoch": 20.735853976531942,
-      "grad_norm": 1.517235279083252,
-      "learning_rate": 6.0555778894472356e-05,
-      "loss": 5.2056,
-      "step": 39761
-    },
-    {
-      "epoch": 20.73637548891786,
-      "grad_norm": 1.7537144422531128,
-      "learning_rate": 6.0554773869346734e-05,
-      "loss": 4.4085,
-      "step": 39762
-    },
-    {
-      "epoch": 20.73689700130378,
-      "grad_norm": 1.5310475826263428,
-      "learning_rate": 6.0553768844221105e-05,
-      "loss": 5.6475,
-      "step": 39763
-    },
-    {
-      "epoch": 20.7374185136897,
-      "grad_norm": 1.5442466735839844,
-      "learning_rate": 6.055276381909548e-05,
-      "loss": 5.3191,
-      "step": 39764
-    },
-    {
-      "epoch": 20.73794002607562,
-      "grad_norm": 1.5191656351089478,
-      "learning_rate": 6.055175879396985e-05,
-      "loss": 5.4069,
-      "step": 39765
-    },
-    {
-      "epoch": 20.73846153846154,
-      "grad_norm": 1.4222437143325806,
-      "learning_rate": 6.0550753768844225e-05,
-      "loss": 5.7249,
-      "step": 39766
-    },
-    {
-      "epoch": 20.73898305084746,
-      "grad_norm": 1.4908653497695923,
-      "learning_rate": 6.0549748743718596e-05,
-      "loss": 5.613,
-      "step": 39767
-    },
-    {
-      "epoch": 20.73950456323338,
-      "grad_norm": 1.5584337711334229,
-      "learning_rate": 6.0548743718592974e-05,
-      "loss": 5.3289,
-      "step": 39768
-    },
-    {
-      "epoch": 20.740026075619298,
-      "grad_norm": 1.5817900896072388,
-      "learning_rate": 6.054773869346734e-05,
-      "loss": 4.9118,
-      "step": 39769
-    },
-    {
-      "epoch": 20.740547588005214,
-      "grad_norm": 1.4674632549285889,
-      "learning_rate": 6.0546733668341716e-05,
-      "loss": 5.5338,
-      "step": 39770
-    },
-    {
-      "epoch": 20.741069100391133,
-      "grad_norm": 1.5344324111938477,
-      "learning_rate": 6.054572864321608e-05,
-      "loss": 5.5126,
-      "step": 39771
-    },
-    {
-      "epoch": 20.741590612777053,
-      "grad_norm": 1.515645146369934,
-      "learning_rate": 6.054472361809046e-05,
-      "loss": 5.2481,
-      "step": 39772
-    },
-    {
-      "epoch": 20.742112125162972,
-      "grad_norm": 1.47941255569458,
-      "learning_rate": 6.054371859296483e-05,
-      "loss": 5.6244,
-      "step": 39773
-    },
-    {
-      "epoch": 20.74263363754889,
-      "grad_norm": 1.5873017311096191,
-      "learning_rate": 6.054271356783921e-05,
-      "loss": 5.1201,
-      "step": 39774
-    },
-    {
-      "epoch": 20.74315514993481,
-      "grad_norm": 1.5855456590652466,
-      "learning_rate": 6.054170854271357e-05,
-      "loss": 5.1248,
-      "step": 39775
-    },
-    {
-      "epoch": 20.74367666232073,
-      "grad_norm": 1.5980288982391357,
-      "learning_rate": 6.0540703517587935e-05,
-      "loss": 5.0643,
-      "step": 39776
-    },
-    {
-      "epoch": 20.74419817470665,
-      "grad_norm": 1.5541105270385742,
-      "learning_rate": 6.053969849246231e-05,
-      "loss": 5.4156,
-      "step": 39777
-    },
-    {
-      "epoch": 20.74471968709257,
-      "grad_norm": 1.4421323537826538,
-      "learning_rate": 6.0538693467336684e-05,
-      "loss": 5.5487,
-      "step": 39778
-    },
-    {
-      "epoch": 20.74524119947849,
-      "grad_norm": 1.4712700843811035,
-      "learning_rate": 6.053768844221106e-05,
-      "loss": 5.399,
-      "step": 39779
-    },
-    {
-      "epoch": 20.74576271186441,
-      "grad_norm": 1.466018557548523,
-      "learning_rate": 6.0536683417085426e-05,
-      "loss": 4.8383,
-      "step": 39780
-    },
-    {
-      "epoch": 20.746284224250324,
-      "grad_norm": 1.5672438144683838,
-      "learning_rate": 6.0535678391959804e-05,
-      "loss": 5.0133,
-      "step": 39781
-    },
-    {
-      "epoch": 20.746805736636244,
-      "grad_norm": 1.5790257453918457,
-      "learning_rate": 6.053467336683417e-05,
-      "loss": 5.1544,
-      "step": 39782
-    },
-    {
-      "epoch": 20.747327249022163,
-      "grad_norm": 1.5260897874832153,
-      "learning_rate": 6.0533668341708546e-05,
-      "loss": 5.2601,
-      "step": 39783
-    },
-    {
-      "epoch": 20.747848761408083,
-      "grad_norm": 1.5079429149627686,
-      "learning_rate": 6.053266331658292e-05,
-      "loss": 5.343,
-      "step": 39784
-    },
-    {
-      "epoch": 20.748370273794002,
-      "grad_norm": 1.4808851480484009,
-      "learning_rate": 6.0531658291457295e-05,
-      "loss": 5.5798,
-      "step": 39785
-    },
-    {
-      "epoch": 20.74889178617992,
-      "grad_norm": 1.553613305091858,
-      "learning_rate": 6.053065326633166e-05,
-      "loss": 5.0639,
-      "step": 39786
-    },
-    {
-      "epoch": 20.74941329856584,
-      "grad_norm": 1.5047022104263306,
-      "learning_rate": 6.052964824120604e-05,
-      "loss": 5.0546,
-      "step": 39787
-    },
-    {
-      "epoch": 20.74993481095176,
-      "grad_norm": 1.512042760848999,
-      "learning_rate": 6.052864321608041e-05,
-      "loss": 5.3988,
-      "step": 39788
-    },
-    {
-      "epoch": 20.75045632333768,
-      "grad_norm": 1.5827561616897583,
-      "learning_rate": 6.052763819095477e-05,
-      "loss": 5.4751,
-      "step": 39789
-    },
-    {
-      "epoch": 20.7509778357236,
-      "grad_norm": 1.5344197750091553,
-      "learning_rate": 6.052663316582915e-05,
-      "loss": 5.6466,
-      "step": 39790
-    },
-    {
-      "epoch": 20.75149934810952,
-      "grad_norm": 1.4560250043869019,
-      "learning_rate": 6.0525628140703515e-05,
-      "loss": 5.7269,
-      "step": 39791
-    },
-    {
-      "epoch": 20.75202086049544,
-      "grad_norm": 1.6703704595565796,
-      "learning_rate": 6.052462311557789e-05,
-      "loss": 5.0548,
-      "step": 39792
-    },
-    {
-      "epoch": 20.752542372881354,
-      "grad_norm": 1.6121684312820435,
-      "learning_rate": 6.0523618090452263e-05,
-      "loss": 5.3991,
-      "step": 39793
-    },
-    {
-      "epoch": 20.753063885267274,
-      "grad_norm": 1.5770491361618042,
-      "learning_rate": 6.052261306532664e-05,
-      "loss": 4.9547,
-      "step": 39794
-    },
-    {
-      "epoch": 20.753585397653193,
-      "grad_norm": 1.5843219757080078,
-      "learning_rate": 6.0521608040201006e-05,
-      "loss": 5.4891,
-      "step": 39795
-    },
-    {
-      "epoch": 20.754106910039113,
-      "grad_norm": 1.429232120513916,
-      "learning_rate": 6.0520603015075383e-05,
-      "loss": 4.9084,
-      "step": 39796
-    },
-    {
-      "epoch": 20.754628422425032,
-      "grad_norm": 1.50742769241333,
-      "learning_rate": 6.051959798994975e-05,
-      "loss": 5.1303,
-      "step": 39797
-    },
-    {
-      "epoch": 20.75514993481095,
-      "grad_norm": 1.4886915683746338,
-      "learning_rate": 6.0518592964824126e-05,
-      "loss": 5.1417,
-      "step": 39798
-    },
-    {
-      "epoch": 20.75567144719687,
-      "grad_norm": 1.5768355131149292,
-      "learning_rate": 6.0517587939698497e-05,
-      "loss": 5.4628,
-      "step": 39799
-    },
-    {
-      "epoch": 20.75619295958279,
-      "grad_norm": 1.484576940536499,
-      "learning_rate": 6.0516582914572874e-05,
-      "loss": 5.7427,
-      "step": 39800
-    },
-    {
-      "epoch": 20.75671447196871,
-      "grad_norm": 1.456667423248291,
-      "learning_rate": 6.051557788944724e-05,
-      "loss": 5.041,
-      "step": 39801
-    },
-    {
-      "epoch": 20.75723598435463,
-      "grad_norm": 1.5110745429992676,
-      "learning_rate": 6.05145728643216e-05,
-      "loss": 5.2433,
-      "step": 39802
-    },
-    {
-      "epoch": 20.75775749674055,
-      "grad_norm": 1.4115207195281982,
-      "learning_rate": 6.051356783919598e-05,
-      "loss": 5.4306,
-      "step": 39803
-    },
-    {
-      "epoch": 20.75827900912647,
-      "grad_norm": 1.579367995262146,
-      "learning_rate": 6.051256281407035e-05,
-      "loss": 5.2368,
-      "step": 39804
-    },
-    {
-      "epoch": 20.758800521512384,
-      "grad_norm": 1.4721304178237915,
-      "learning_rate": 6.051155778894473e-05,
-      "loss": 5.2844,
-      "step": 39805
-    },
-    {
-      "epoch": 20.759322033898304,
-      "grad_norm": 1.4999877214431763,
-      "learning_rate": 6.0510552763819094e-05,
-      "loss": 5.6235,
-      "step": 39806
-    },
-    {
-      "epoch": 20.759843546284223,
-      "grad_norm": 1.6428139209747314,
-      "learning_rate": 6.050954773869347e-05,
-      "loss": 5.3177,
-      "step": 39807
-    },
-    {
-      "epoch": 20.760365058670143,
-      "grad_norm": 1.550929307937622,
-      "learning_rate": 6.050854271356784e-05,
-      "loss": 5.2686,
-      "step": 39808
-    },
-    {
-      "epoch": 20.760886571056062,
-      "grad_norm": 1.4819782972335815,
-      "learning_rate": 6.050753768844222e-05,
-      "loss": 5.5498,
-      "step": 39809
-    },
-    {
-      "epoch": 20.76140808344198,
-      "grad_norm": 1.4415926933288574,
-      "learning_rate": 6.0506532663316585e-05,
-      "loss": 5.045,
-      "step": 39810
-    },
-    {
-      "epoch": 20.7619295958279,
-      "grad_norm": 1.4308089017868042,
-      "learning_rate": 6.050552763819096e-05,
-      "loss": 5.5597,
-      "step": 39811
-    },
-    {
-      "epoch": 20.76245110821382,
-      "grad_norm": 1.5281836986541748,
-      "learning_rate": 6.050452261306533e-05,
-      "loss": 5.6835,
-      "step": 39812
-    },
-    {
-      "epoch": 20.76297262059974,
-      "grad_norm": 1.6392236948013306,
-      "learning_rate": 6.0503517587939705e-05,
-      "loss": 4.9248,
-      "step": 39813
-    },
-    {
-      "epoch": 20.76349413298566,
-      "grad_norm": 1.5648926496505737,
-      "learning_rate": 6.0502512562814076e-05,
-      "loss": 5.3039,
-      "step": 39814
-    },
-    {
-      "epoch": 20.76401564537158,
-      "grad_norm": 1.4695162773132324,
-      "learning_rate": 6.050150753768844e-05,
-      "loss": 5.2771,
-      "step": 39815
-    },
-    {
-      "epoch": 20.7645371577575,
-      "grad_norm": 1.455304741859436,
-      "learning_rate": 6.050050251256282e-05,
-      "loss": 5.4062,
-      "step": 39816
-    },
-    {
-      "epoch": 20.765058670143414,
-      "grad_norm": 1.500754952430725,
-      "learning_rate": 6.049949748743718e-05,
-      "loss": 5.4293,
-      "step": 39817
-    },
-    {
-      "epoch": 20.765580182529334,
-      "grad_norm": 1.6058802604675293,
-      "learning_rate": 6.049849246231156e-05,
-      "loss": 5.4247,
-      "step": 39818
-    },
-    {
-      "epoch": 20.766101694915253,
-      "grad_norm": 1.5973701477050781,
-      "learning_rate": 6.049748743718593e-05,
-      "loss": 5.1767,
-      "step": 39819
-    },
-    {
-      "epoch": 20.766623207301173,
-      "grad_norm": 1.5378121137619019,
-      "learning_rate": 6.049648241206031e-05,
-      "loss": 5.4508,
-      "step": 39820
-    },
-    {
-      "epoch": 20.767144719687092,
-      "grad_norm": 1.4487895965576172,
-      "learning_rate": 6.049547738693467e-05,
-      "loss": 5.385,
-      "step": 39821
-    },
-    {
-      "epoch": 20.76766623207301,
-      "grad_norm": 1.4878191947937012,
-      "learning_rate": 6.049447236180905e-05,
-      "loss": 5.0518,
-      "step": 39822
-    },
-    {
-      "epoch": 20.76818774445893,
-      "grad_norm": 1.5537110567092896,
-      "learning_rate": 6.0493467336683415e-05,
-      "loss": 5.254,
-      "step": 39823
-    },
-    {
-      "epoch": 20.76870925684485,
-      "grad_norm": 1.5708539485931396,
-      "learning_rate": 6.049246231155779e-05,
-      "loss": 5.5589,
-      "step": 39824
-    },
-    {
-      "epoch": 20.76923076923077,
-      "grad_norm": 1.5808264017105103,
-      "learning_rate": 6.0491457286432164e-05,
-      "loss": 4.7519,
-      "step": 39825
-    },
-    {
-      "epoch": 20.76975228161669,
-      "grad_norm": 1.555550217628479,
-      "learning_rate": 6.049045226130654e-05,
-      "loss": 5.5497,
-      "step": 39826
-    },
-    {
-      "epoch": 20.77027379400261,
-      "grad_norm": 1.6048319339752197,
-      "learning_rate": 6.0489447236180906e-05,
-      "loss": 5.4876,
-      "step": 39827
-    },
-    {
-      "epoch": 20.77079530638853,
-      "grad_norm": 1.5714980363845825,
-      "learning_rate": 6.048844221105527e-05,
-      "loss": 5.2624,
-      "step": 39828
-    },
-    {
-      "epoch": 20.771316818774444,
-      "grad_norm": 1.6500221490859985,
-      "learning_rate": 6.048743718592965e-05,
-      "loss": 5.3867,
-      "step": 39829
-    },
-    {
-      "epoch": 20.771838331160364,
-      "grad_norm": 1.4877474308013916,
-      "learning_rate": 6.048643216080402e-05,
-      "loss": 5.7078,
-      "step": 39830
-    },
-    {
-      "epoch": 20.772359843546283,
-      "grad_norm": 1.5882434844970703,
-      "learning_rate": 6.04854271356784e-05,
-      "loss": 5.1284,
-      "step": 39831
-    },
-    {
-      "epoch": 20.772881355932203,
-      "grad_norm": 1.5127390623092651,
-      "learning_rate": 6.048442211055276e-05,
-      "loss": 5.0011,
-      "step": 39832
-    },
-    {
-      "epoch": 20.773402868318122,
-      "grad_norm": 1.5394182205200195,
-      "learning_rate": 6.048341708542714e-05,
-      "loss": 5.4263,
-      "step": 39833
-    },
-    {
-      "epoch": 20.77392438070404,
-      "grad_norm": 1.508617639541626,
-      "learning_rate": 6.048241206030151e-05,
-      "loss": 5.3419,
-      "step": 39834
-    },
-    {
-      "epoch": 20.77444589308996,
-      "grad_norm": 1.4134243726730347,
-      "learning_rate": 6.048140703517589e-05,
-      "loss": 5.7115,
-      "step": 39835
-    },
-    {
-      "epoch": 20.77496740547588,
-      "grad_norm": 1.4994351863861084,
-      "learning_rate": 6.048040201005025e-05,
-      "loss": 5.4973,
-      "step": 39836
-    },
-    {
-      "epoch": 20.7754889178618,
-      "grad_norm": 1.4198002815246582,
-      "learning_rate": 6.047939698492463e-05,
-      "loss": 5.2973,
-      "step": 39837
-    },
-    {
-      "epoch": 20.77601043024772,
-      "grad_norm": 1.5055568218231201,
-      "learning_rate": 6.0478391959798995e-05,
-      "loss": 5.1774,
-      "step": 39838
-    },
-    {
-      "epoch": 20.77653194263364,
-      "grad_norm": 1.4391368627548218,
-      "learning_rate": 6.047738693467337e-05,
-      "loss": 5.7397,
-      "step": 39839
-    },
-    {
-      "epoch": 20.777053455019555,
-      "grad_norm": 1.5909496545791626,
-      "learning_rate": 6.0476381909547744e-05,
-      "loss": 5.1295,
-      "step": 39840
-    },
-    {
-      "epoch": 20.777574967405474,
-      "grad_norm": 1.5903970003128052,
-      "learning_rate": 6.047537688442211e-05,
-      "loss": 5.135,
-      "step": 39841
-    },
-    {
-      "epoch": 20.778096479791394,
-      "grad_norm": 1.4708921909332275,
-      "learning_rate": 6.0474371859296486e-05,
-      "loss": 5.5788,
-      "step": 39842
-    },
-    {
-      "epoch": 20.778617992177313,
-      "grad_norm": 1.3997968435287476,
-      "learning_rate": 6.047336683417085e-05,
-      "loss": 5.6846,
-      "step": 39843
-    },
-    {
-      "epoch": 20.779139504563233,
-      "grad_norm": 1.6252117156982422,
-      "learning_rate": 6.047236180904523e-05,
-      "loss": 5.4533,
-      "step": 39844
-    },
-    {
-      "epoch": 20.779661016949152,
-      "grad_norm": 1.515307068824768,
-      "learning_rate": 6.04713567839196e-05,
-      "loss": 5.4832,
-      "step": 39845
-    },
-    {
-      "epoch": 20.78018252933507,
-      "grad_norm": 1.7766960859298706,
-      "learning_rate": 6.047035175879398e-05,
-      "loss": 5.4787,
-      "step": 39846
-    },
-    {
-      "epoch": 20.78070404172099,
-      "grad_norm": 1.410578727722168,
-      "learning_rate": 6.046934673366834e-05,
-      "loss": 5.664,
-      "step": 39847
-    },
-    {
-      "epoch": 20.78122555410691,
-      "grad_norm": 1.4207721948623657,
-      "learning_rate": 6.046834170854272e-05,
-      "loss": 5.5757,
-      "step": 39848
-    },
-    {
-      "epoch": 20.78174706649283,
-      "grad_norm": 1.5073217153549194,
-      "learning_rate": 6.046733668341708e-05,
-      "loss": 5.3726,
-      "step": 39849
-    },
-    {
-      "epoch": 20.78226857887875,
-      "grad_norm": 1.5117231607437134,
-      "learning_rate": 6.046633165829146e-05,
-      "loss": 5.3304,
-      "step": 39850
-    },
-    {
-      "epoch": 20.78279009126467,
-      "grad_norm": 1.631959080696106,
-      "learning_rate": 6.046532663316583e-05,
-      "loss": 4.6168,
-      "step": 39851
-    },
-    {
-      "epoch": 20.78331160365059,
-      "grad_norm": 1.6062592267990112,
-      "learning_rate": 6.046432160804021e-05,
-      "loss": 5.378,
-      "step": 39852
-    },
-    {
-      "epoch": 20.783833116036504,
-      "grad_norm": 1.61146879196167,
-      "learning_rate": 6.0463316582914574e-05,
-      "loss": 5.3774,
-      "step": 39853
-    },
-    {
-      "epoch": 20.784354628422424,
-      "grad_norm": 1.574057698249817,
-      "learning_rate": 6.0462311557788945e-05,
-      "loss": 5.5193,
-      "step": 39854
-    },
-    {
-      "epoch": 20.784876140808343,
-      "grad_norm": 1.5598788261413574,
-      "learning_rate": 6.046130653266332e-05,
-      "loss": 5.188,
-      "step": 39855
-    },
-    {
-      "epoch": 20.785397653194263,
-      "grad_norm": 1.4512594938278198,
-      "learning_rate": 6.046030150753769e-05,
-      "loss": 5.465,
-      "step": 39856
-    },
-    {
-      "epoch": 20.785919165580182,
-      "grad_norm": 1.619172215461731,
-      "learning_rate": 6.0459296482412065e-05,
-      "loss": 4.7827,
-      "step": 39857
-    },
-    {
-      "epoch": 20.7864406779661,
-      "grad_norm": 1.5253984928131104,
-      "learning_rate": 6.045829145728643e-05,
-      "loss": 5.6884,
-      "step": 39858
-    },
-    {
-      "epoch": 20.78696219035202,
-      "grad_norm": 1.5285894870758057,
-      "learning_rate": 6.045728643216081e-05,
-      "loss": 5.343,
-      "step": 39859
-    },
-    {
-      "epoch": 20.78748370273794,
-      "grad_norm": 1.4380850791931152,
-      "learning_rate": 6.045628140703518e-05,
-      "loss": 5.4463,
-      "step": 39860
-    },
-    {
-      "epoch": 20.78800521512386,
-      "grad_norm": 1.7334266901016235,
-      "learning_rate": 6.0455276381909556e-05,
-      "loss": 4.3763,
-      "step": 39861
-    },
-    {
-      "epoch": 20.78852672750978,
-      "grad_norm": 1.4341411590576172,
-      "learning_rate": 6.045427135678392e-05,
-      "loss": 5.4011,
-      "step": 39862
-    },
-    {
-      "epoch": 20.7890482398957,
-      "grad_norm": 1.5777955055236816,
-      "learning_rate": 6.04532663316583e-05,
-      "loss": 5.4974,
-      "step": 39863
-    },
-    {
-      "epoch": 20.789569752281615,
-      "grad_norm": 1.4106073379516602,
-      "learning_rate": 6.045226130653266e-05,
-      "loss": 5.5509,
-      "step": 39864
-    },
-    {
-      "epoch": 20.790091264667534,
-      "grad_norm": 1.4069663286209106,
-      "learning_rate": 6.045125628140704e-05,
-      "loss": 5.1899,
-      "step": 39865
-    },
-    {
-      "epoch": 20.790612777053454,
-      "grad_norm": 1.4910531044006348,
-      "learning_rate": 6.045025125628141e-05,
-      "loss": 5.0952,
-      "step": 39866
-    },
-    {
-      "epoch": 20.791134289439373,
-      "grad_norm": 1.4396672248840332,
-      "learning_rate": 6.044924623115579e-05,
-      "loss": 5.1368,
-      "step": 39867
-    },
-    {
-      "epoch": 20.791655801825293,
-      "grad_norm": 1.5694156885147095,
-      "learning_rate": 6.044824120603015e-05,
-      "loss": 5.2465,
-      "step": 39868
-    },
-    {
-      "epoch": 20.792177314211212,
-      "grad_norm": 1.3601713180541992,
-      "learning_rate": 6.044723618090452e-05,
-      "loss": 4.721,
-      "step": 39869
-    },
-    {
-      "epoch": 20.79269882659713,
-      "grad_norm": 1.3907192945480347,
-      "learning_rate": 6.0446231155778895e-05,
-      "loss": 5.4633,
-      "step": 39870
-    },
-    {
-      "epoch": 20.79322033898305,
-      "grad_norm": 1.4873064756393433,
-      "learning_rate": 6.0445226130653266e-05,
-      "loss": 5.6899,
-      "step": 39871
-    },
-    {
-      "epoch": 20.79374185136897,
-      "grad_norm": 1.4381142854690552,
-      "learning_rate": 6.0444221105527644e-05,
-      "loss": 5.5967,
-      "step": 39872
-    },
-    {
-      "epoch": 20.79426336375489,
-      "grad_norm": 1.4789106845855713,
-      "learning_rate": 6.044321608040201e-05,
-      "loss": 4.2494,
-      "step": 39873
-    },
-    {
-      "epoch": 20.79478487614081,
-      "grad_norm": 1.672726035118103,
-      "learning_rate": 6.0442211055276386e-05,
-      "loss": 4.83,
-      "step": 39874
-    },
-    {
-      "epoch": 20.79530638852673,
-      "grad_norm": 1.6276755332946777,
-      "learning_rate": 6.044120603015076e-05,
-      "loss": 5.1632,
-      "step": 39875
-    },
-    {
-      "epoch": 20.795827900912645,
-      "grad_norm": 1.5594706535339355,
-      "learning_rate": 6.0440201005025135e-05,
-      "loss": 5.4431,
-      "step": 39876
-    },
-    {
-      "epoch": 20.796349413298564,
-      "grad_norm": 1.6361676454544067,
-      "learning_rate": 6.04391959798995e-05,
-      "loss": 5.1623,
-      "step": 39877
-    },
-    {
-      "epoch": 20.796870925684484,
-      "grad_norm": 1.5639970302581787,
-      "learning_rate": 6.043819095477388e-05,
-      "loss": 4.7074,
-      "step": 39878
-    },
-    {
-      "epoch": 20.797392438070403,
-      "grad_norm": 1.642386794090271,
-      "learning_rate": 6.043718592964824e-05,
-      "loss": 4.812,
-      "step": 39879
-    },
-    {
-      "epoch": 20.797913950456323,
-      "grad_norm": 1.4882620573043823,
-      "learning_rate": 6.043618090452262e-05,
-      "loss": 5.4587,
-      "step": 39880
-    },
-    {
-      "epoch": 20.798435462842242,
-      "grad_norm": 1.4787524938583374,
-      "learning_rate": 6.043517587939699e-05,
-      "loss": 5.3744,
-      "step": 39881
-    },
-    {
-      "epoch": 20.798956975228162,
-      "grad_norm": 1.483515977859497,
-      "learning_rate": 6.0434170854271355e-05,
-      "loss": 5.019,
-      "step": 39882
-    },
-    {
-      "epoch": 20.79947848761408,
-      "grad_norm": 1.5972687005996704,
-      "learning_rate": 6.043316582914573e-05,
-      "loss": 5.2904,
-      "step": 39883
-    },
-    {
-      "epoch": 20.8,
-      "grad_norm": 1.4910686016082764,
-      "learning_rate": 6.04321608040201e-05,
-      "loss": 5.6277,
-      "step": 39884
-    },
-    {
-      "epoch": 20.80052151238592,
-      "grad_norm": 1.4789589643478394,
-      "learning_rate": 6.0431155778894475e-05,
-      "loss": 5.5252,
-      "step": 39885
-    },
-    {
-      "epoch": 20.80104302477184,
-      "grad_norm": 1.5255082845687866,
-      "learning_rate": 6.0430150753768846e-05,
-      "loss": 4.5761,
-      "step": 39886
-    },
-    {
-      "epoch": 20.80156453715776,
-      "grad_norm": 1.440598487854004,
-      "learning_rate": 6.0429145728643224e-05,
-      "loss": 5.2248,
-      "step": 39887
-    },
-    {
-      "epoch": 20.802086049543675,
-      "grad_norm": 1.4021072387695312,
-      "learning_rate": 6.042814070351759e-05,
-      "loss": 5.3205,
-      "step": 39888
-    },
-    {
-      "epoch": 20.802607561929594,
-      "grad_norm": 1.4987972974777222,
-      "learning_rate": 6.0427135678391966e-05,
-      "loss": 5.4646,
-      "step": 39889
-    },
-    {
-      "epoch": 20.803129074315514,
-      "grad_norm": 1.49226713180542,
-      "learning_rate": 6.042613065326633e-05,
-      "loss": 5.0423,
-      "step": 39890
-    },
-    {
-      "epoch": 20.803650586701433,
-      "grad_norm": 1.5054028034210205,
-      "learning_rate": 6.042512562814071e-05,
-      "loss": 5.5228,
-      "step": 39891
-    },
-    {
-      "epoch": 20.804172099087353,
-      "grad_norm": 1.6311123371124268,
-      "learning_rate": 6.042412060301508e-05,
-      "loss": 5.1491,
-      "step": 39892
-    },
-    {
-      "epoch": 20.804693611473272,
-      "grad_norm": 1.6004329919815063,
-      "learning_rate": 6.042311557788946e-05,
-      "loss": 5.3356,
-      "step": 39893
-    },
-    {
-      "epoch": 20.805215123859192,
-      "grad_norm": 1.5624935626983643,
-      "learning_rate": 6.042211055276382e-05,
-      "loss": 4.4504,
-      "step": 39894
-    },
-    {
-      "epoch": 20.80573663624511,
-      "grad_norm": 1.4448143243789673,
-      "learning_rate": 6.042110552763819e-05,
-      "loss": 5.4986,
-      "step": 39895
-    },
-    {
-      "epoch": 20.80625814863103,
-      "grad_norm": 1.511906623840332,
-      "learning_rate": 6.042010050251257e-05,
-      "loss": 4.5871,
-      "step": 39896
-    },
-    {
-      "epoch": 20.80677966101695,
-      "grad_norm": 1.5463826656341553,
-      "learning_rate": 6.0419095477386934e-05,
-      "loss": 5.2836,
-      "step": 39897
-    },
-    {
-      "epoch": 20.80730117340287,
-      "grad_norm": 1.5231024026870728,
-      "learning_rate": 6.041809045226131e-05,
-      "loss": 5.3128,
-      "step": 39898
-    },
-    {
-      "epoch": 20.80782268578879,
-      "grad_norm": 1.4506170749664307,
-      "learning_rate": 6.0417085427135676e-05,
-      "loss": 5.4482,
-      "step": 39899
-    },
-    {
-      "epoch": 20.808344198174705,
-      "grad_norm": 1.4772603511810303,
-      "learning_rate": 6.0416080402010054e-05,
-      "loss": 5.052,
-      "step": 39900
-    },
-    {
-      "epoch": 20.808865710560625,
-      "grad_norm": 1.4169104099273682,
-      "learning_rate": 6.0415075376884425e-05,
-      "loss": 5.3644,
-      "step": 39901
-    },
-    {
-      "epoch": 20.809387222946544,
-      "grad_norm": 1.5253602266311646,
-      "learning_rate": 6.04140703517588e-05,
-      "loss": 4.8209,
-      "step": 39902
-    },
-    {
-      "epoch": 20.809908735332463,
-      "grad_norm": 1.4498074054718018,
-      "learning_rate": 6.041306532663317e-05,
-      "loss": 5.8008,
-      "step": 39903
-    },
-    {
-      "epoch": 20.810430247718383,
-      "grad_norm": 1.45956289768219,
-      "learning_rate": 6.0412060301507545e-05,
-      "loss": 5.6856,
-      "step": 39904
-    },
-    {
-      "epoch": 20.810951760104302,
-      "grad_norm": 1.5131890773773193,
-      "learning_rate": 6.041105527638191e-05,
-      "loss": 5.4971,
-      "step": 39905
-    },
-    {
-      "epoch": 20.811473272490222,
-      "grad_norm": 1.450623631477356,
-      "learning_rate": 6.041005025125629e-05,
-      "loss": 5.3875,
-      "step": 39906
-    },
-    {
-      "epoch": 20.81199478487614,
-      "grad_norm": 1.482239007949829,
-      "learning_rate": 6.040904522613066e-05,
-      "loss": 5.0981,
-      "step": 39907
-    },
-    {
-      "epoch": 20.81251629726206,
-      "grad_norm": 1.5412629842758179,
-      "learning_rate": 6.040804020100502e-05,
-      "loss": 5.45,
-      "step": 39908
-    },
-    {
-      "epoch": 20.81303780964798,
-      "grad_norm": 1.5341947078704834,
-      "learning_rate": 6.04070351758794e-05,
-      "loss": 5.4021,
-      "step": 39909
-    },
-    {
-      "epoch": 20.8135593220339,
-      "grad_norm": 1.4398082494735718,
-      "learning_rate": 6.0406030150753765e-05,
-      "loss": 4.7041,
-      "step": 39910
-    },
-    {
-      "epoch": 20.81408083441982,
-      "grad_norm": 1.5265710353851318,
-      "learning_rate": 6.040502512562814e-05,
-      "loss": 5.397,
-      "step": 39911
-    },
-    {
-      "epoch": 20.814602346805735,
-      "grad_norm": 1.4480386972427368,
-      "learning_rate": 6.0404020100502513e-05,
-      "loss": 5.1887,
-      "step": 39912
-    },
-    {
-      "epoch": 20.815123859191655,
-      "grad_norm": 1.6141515970230103,
-      "learning_rate": 6.040301507537689e-05,
-      "loss": 4.79,
-      "step": 39913
-    },
-    {
-      "epoch": 20.815645371577574,
-      "grad_norm": 1.4284404516220093,
-      "learning_rate": 6.0402010050251256e-05,
-      "loss": 5.6329,
-      "step": 39914
-    },
-    {
-      "epoch": 20.816166883963493,
-      "grad_norm": 1.6170051097869873,
-      "learning_rate": 6.0401005025125633e-05,
-      "loss": 4.7098,
-      "step": 39915
-    },
-    {
-      "epoch": 20.816688396349413,
-      "grad_norm": 1.4958391189575195,
-      "learning_rate": 6.04e-05,
-      "loss": 5.4194,
-      "step": 39916
-    },
-    {
-      "epoch": 20.817209908735332,
-      "grad_norm": 1.4733611345291138,
-      "learning_rate": 6.0398994974874376e-05,
-      "loss": 4.8584,
-      "step": 39917
-    },
-    {
-      "epoch": 20.817731421121252,
-      "grad_norm": 1.4669216871261597,
-      "learning_rate": 6.0397989949748747e-05,
-      "loss": 5.0529,
-      "step": 39918
-    },
-    {
-      "epoch": 20.81825293350717,
-      "grad_norm": 1.3859708309173584,
-      "learning_rate": 6.0396984924623124e-05,
-      "loss": 5.6833,
-      "step": 39919
-    },
-    {
-      "epoch": 20.81877444589309,
-      "grad_norm": 1.476605772972107,
-      "learning_rate": 6.039597989949749e-05,
-      "loss": 5.5159,
-      "step": 39920
-    },
-    {
-      "epoch": 20.81929595827901,
-      "grad_norm": 1.3268121480941772,
-      "learning_rate": 6.039497487437186e-05,
-      "loss": 5.0258,
-      "step": 39921
-    },
-    {
-      "epoch": 20.81981747066493,
-      "grad_norm": 1.4842336177825928,
-      "learning_rate": 6.039396984924624e-05,
-      "loss": 5.4764,
-      "step": 39922
-    },
-    {
-      "epoch": 20.820338983050846,
-      "grad_norm": 1.4540526866912842,
-      "learning_rate": 6.03929648241206e-05,
-      "loss": 5.2722,
-      "step": 39923
-    },
-    {
-      "epoch": 20.820860495436765,
-      "grad_norm": 1.599642276763916,
-      "learning_rate": 6.039195979899498e-05,
-      "loss": 5.1843,
-      "step": 39924
-    },
-    {
-      "epoch": 20.821382007822685,
-      "grad_norm": 1.591845154762268,
-      "learning_rate": 6.0390954773869344e-05,
-      "loss": 5.0244,
-      "step": 39925
-    },
-    {
-      "epoch": 20.821903520208604,
-      "grad_norm": 1.504915714263916,
-      "learning_rate": 6.038994974874372e-05,
-      "loss": 5.058,
-      "step": 39926
-    },
-    {
-      "epoch": 20.822425032594523,
-      "grad_norm": 1.4430899620056152,
-      "learning_rate": 6.038894472361809e-05,
-      "loss": 5.4913,
-      "step": 39927
-    },
-    {
-      "epoch": 20.822946544980443,
-      "grad_norm": 1.511671543121338,
-      "learning_rate": 6.038793969849247e-05,
-      "loss": 5.2106,
-      "step": 39928
-    },
-    {
-      "epoch": 20.823468057366362,
-      "grad_norm": 1.545095443725586,
-      "learning_rate": 6.0386934673366835e-05,
-      "loss": 4.6872,
-      "step": 39929
-    },
-    {
-      "epoch": 20.823989569752282,
-      "grad_norm": 1.5059067010879517,
-      "learning_rate": 6.038592964824121e-05,
-      "loss": 5.1071,
-      "step": 39930
-    },
-    {
-      "epoch": 20.8245110821382,
-      "grad_norm": 1.4671326875686646,
-      "learning_rate": 6.038492462311558e-05,
-      "loss": 4.4065,
-      "step": 39931
-    },
-    {
-      "epoch": 20.82503259452412,
-      "grad_norm": 1.5128850936889648,
-      "learning_rate": 6.0383919597989955e-05,
-      "loss": 5.0276,
-      "step": 39932
-    },
-    {
-      "epoch": 20.82555410691004,
-      "grad_norm": 1.4708514213562012,
-      "learning_rate": 6.0382914572864326e-05,
-      "loss": 5.5917,
-      "step": 39933
-    },
-    {
-      "epoch": 20.82607561929596,
-      "grad_norm": 1.5712562799453735,
-      "learning_rate": 6.038190954773869e-05,
-      "loss": 5.2616,
-      "step": 39934
-    },
-    {
-      "epoch": 20.82659713168188,
-      "grad_norm": 1.564740777015686,
-      "learning_rate": 6.038090452261307e-05,
-      "loss": 4.8635,
-      "step": 39935
-    },
-    {
-      "epoch": 20.827118644067795,
-      "grad_norm": 1.4187517166137695,
-      "learning_rate": 6.037989949748743e-05,
-      "loss": 5.3705,
-      "step": 39936
-    },
-    {
-      "epoch": 20.827640156453715,
-      "grad_norm": 1.506408929824829,
-      "learning_rate": 6.037889447236181e-05,
-      "loss": 5.403,
-      "step": 39937
-    },
-    {
-      "epoch": 20.828161668839634,
-      "grad_norm": 1.4963879585266113,
-      "learning_rate": 6.037788944723618e-05,
-      "loss": 5.3444,
-      "step": 39938
-    },
-    {
-      "epoch": 20.828683181225554,
-      "grad_norm": 1.688828945159912,
-      "learning_rate": 6.037688442211056e-05,
-      "loss": 4.7799,
-      "step": 39939
-    },
-    {
-      "epoch": 20.829204693611473,
-      "grad_norm": 1.4276642799377441,
-      "learning_rate": 6.037587939698492e-05,
-      "loss": 5.6716,
-      "step": 39940
-    },
-    {
-      "epoch": 20.829726205997392,
-      "grad_norm": 1.593644142150879,
-      "learning_rate": 6.03748743718593e-05,
-      "loss": 5.4766,
-      "step": 39941
-    },
-    {
-      "epoch": 20.830247718383312,
-      "grad_norm": 1.44269859790802,
-      "learning_rate": 6.037386934673367e-05,
-      "loss": 5.5356,
-      "step": 39942
-    },
-    {
-      "epoch": 20.83076923076923,
-      "grad_norm": 1.70134699344635,
-      "learning_rate": 6.037286432160805e-05,
-      "loss": 4.5554,
-      "step": 39943
-    },
-    {
-      "epoch": 20.83129074315515,
-      "grad_norm": 1.62839937210083,
-      "learning_rate": 6.0371859296482414e-05,
-      "loss": 5.496,
-      "step": 39944
-    },
-    {
-      "epoch": 20.83181225554107,
-      "grad_norm": 1.4503368139266968,
-      "learning_rate": 6.037085427135679e-05,
-      "loss": 5.4414,
-      "step": 39945
-    },
-    {
-      "epoch": 20.83233376792699,
-      "grad_norm": 1.4764113426208496,
-      "learning_rate": 6.0369849246231156e-05,
-      "loss": 5.5379,
-      "step": 39946
-    },
-    {
-      "epoch": 20.832855280312906,
-      "grad_norm": 1.439281702041626,
-      "learning_rate": 6.0368844221105534e-05,
-      "loss": 5.0078,
-      "step": 39947
-    },
-    {
-      "epoch": 20.833376792698825,
-      "grad_norm": 1.5833196640014648,
-      "learning_rate": 6.0367839195979905e-05,
-      "loss": 5.0277,
-      "step": 39948
-    },
-    {
-      "epoch": 20.833898305084745,
-      "grad_norm": 1.518999457359314,
-      "learning_rate": 6.036683417085427e-05,
-      "loss": 5.6431,
-      "step": 39949
-    },
-    {
-      "epoch": 20.834419817470664,
-      "grad_norm": 1.6284846067428589,
-      "learning_rate": 6.036582914572865e-05,
-      "loss": 4.7729,
-      "step": 39950
-    },
-    {
-      "epoch": 20.834941329856584,
-      "grad_norm": 1.5679675340652466,
-      "learning_rate": 6.036482412060301e-05,
-      "loss": 4.313,
-      "step": 39951
-    },
-    {
-      "epoch": 20.835462842242503,
-      "grad_norm": 1.4857149124145508,
-      "learning_rate": 6.036381909547739e-05,
-      "loss": 5.4421,
-      "step": 39952
-    },
-    {
-      "epoch": 20.835984354628422,
-      "grad_norm": 1.5602260828018188,
-      "learning_rate": 6.036281407035176e-05,
-      "loss": 5.374,
-      "step": 39953
-    },
-    {
-      "epoch": 20.836505867014342,
-      "grad_norm": 1.5773767232894897,
-      "learning_rate": 6.036180904522614e-05,
-      "loss": 5.0935,
-      "step": 39954
-    },
-    {
-      "epoch": 20.83702737940026,
-      "grad_norm": 1.648660659790039,
-      "learning_rate": 6.03608040201005e-05,
-      "loss": 5.1309,
-      "step": 39955
-    },
-    {
-      "epoch": 20.83754889178618,
-      "grad_norm": 1.6436294317245483,
-      "learning_rate": 6.035979899497488e-05,
-      "loss": 4.7443,
-      "step": 39956
-    },
-    {
-      "epoch": 20.8380704041721,
-      "grad_norm": 1.491809606552124,
-      "learning_rate": 6.0358793969849245e-05,
-      "loss": 5.4018,
-      "step": 39957
-    },
-    {
-      "epoch": 20.83859191655802,
-      "grad_norm": 1.5448087453842163,
-      "learning_rate": 6.035778894472362e-05,
-      "loss": 4.9539,
-      "step": 39958
-    },
-    {
-      "epoch": 20.839113428943936,
-      "grad_norm": 1.5586057901382446,
-      "learning_rate": 6.0356783919597994e-05,
-      "loss": 5.0509,
-      "step": 39959
-    },
-    {
-      "epoch": 20.839634941329855,
-      "grad_norm": 1.5103428363800049,
-      "learning_rate": 6.035577889447237e-05,
-      "loss": 4.9019,
-      "step": 39960
-    },
-    {
-      "epoch": 20.840156453715775,
-      "grad_norm": 1.4897074699401855,
-      "learning_rate": 6.0354773869346736e-05,
-      "loss": 4.7142,
-      "step": 39961
-    },
-    {
-      "epoch": 20.840677966101694,
-      "grad_norm": 1.5955137014389038,
-      "learning_rate": 6.035376884422111e-05,
-      "loss": 5.1867,
-      "step": 39962
-    },
-    {
-      "epoch": 20.841199478487614,
-      "grad_norm": 1.4793031215667725,
-      "learning_rate": 6.0352763819095485e-05,
-      "loss": 5.6091,
-      "step": 39963
-    },
-    {
-      "epoch": 20.841720990873533,
-      "grad_norm": 1.517891526222229,
-      "learning_rate": 6.035175879396985e-05,
-      "loss": 4.7439,
-      "step": 39964
-    },
-    {
-      "epoch": 20.842242503259452,
-      "grad_norm": 1.4355560541152954,
-      "learning_rate": 6.035075376884423e-05,
-      "loss": 5.569,
-      "step": 39965
-    },
-    {
-      "epoch": 20.842764015645372,
-      "grad_norm": 1.3943548202514648,
-      "learning_rate": 6.034974874371859e-05,
-      "loss": 5.3957,
-      "step": 39966
-    },
-    {
-      "epoch": 20.84328552803129,
-      "grad_norm": 1.5681556463241577,
-      "learning_rate": 6.034874371859297e-05,
-      "loss": 4.3901,
-      "step": 39967
-    },
-    {
-      "epoch": 20.84380704041721,
-      "grad_norm": 1.5573011636734009,
-      "learning_rate": 6.034773869346734e-05,
-      "loss": 5.3652,
-      "step": 39968
-    },
-    {
-      "epoch": 20.84432855280313,
-      "grad_norm": 1.5746870040893555,
-      "learning_rate": 6.034673366834172e-05,
-      "loss": 5.527,
-      "step": 39969
-    },
-    {
-      "epoch": 20.84485006518905,
-      "grad_norm": 1.4059549570083618,
-      "learning_rate": 6.034572864321608e-05,
-      "loss": 5.6607,
-      "step": 39970
-    },
-    {
-      "epoch": 20.845371577574966,
-      "grad_norm": 1.5632845163345337,
-      "learning_rate": 6.034472361809046e-05,
-      "loss": 5.184,
-      "step": 39971
-    },
-    {
-      "epoch": 20.845893089960885,
-      "grad_norm": 1.5040395259857178,
-      "learning_rate": 6.0343718592964824e-05,
-      "loss": 5.4614,
-      "step": 39972
-    },
-    {
-      "epoch": 20.846414602346805,
-      "grad_norm": 1.5227994918823242,
-      "learning_rate": 6.03427135678392e-05,
-      "loss": 5.2957,
-      "step": 39973
-    },
-    {
-      "epoch": 20.846936114732724,
-      "grad_norm": 1.4963380098342896,
-      "learning_rate": 6.034170854271357e-05,
-      "loss": 5.071,
-      "step": 39974
-    },
-    {
-      "epoch": 20.847457627118644,
-      "grad_norm": 1.522970199584961,
-      "learning_rate": 6.034070351758794e-05,
-      "loss": 5.3396,
-      "step": 39975
-    },
-    {
-      "epoch": 20.847979139504563,
-      "grad_norm": 1.5952558517456055,
-      "learning_rate": 6.0339698492462315e-05,
-      "loss": 5.5151,
-      "step": 39976
-    },
-    {
-      "epoch": 20.848500651890483,
-      "grad_norm": 1.6090325117111206,
-      "learning_rate": 6.033869346733668e-05,
-      "loss": 5.1836,
-      "step": 39977
-    },
-    {
-      "epoch": 20.849022164276402,
-      "grad_norm": 1.5110201835632324,
-      "learning_rate": 6.033768844221106e-05,
-      "loss": 5.2427,
-      "step": 39978
-    },
-    {
-      "epoch": 20.84954367666232,
-      "grad_norm": 1.4837230443954468,
-      "learning_rate": 6.033668341708543e-05,
-      "loss": 5.5362,
-      "step": 39979
-    },
-    {
-      "epoch": 20.85006518904824,
-      "grad_norm": 1.6052162647247314,
-      "learning_rate": 6.0335678391959806e-05,
-      "loss": 5.2376,
-      "step": 39980
-    },
-    {
-      "epoch": 20.85058670143416,
-      "grad_norm": 1.503879189491272,
-      "learning_rate": 6.033467336683417e-05,
-      "loss": 4.8504,
-      "step": 39981
-    },
-    {
-      "epoch": 20.85110821382008,
-      "grad_norm": 1.6010560989379883,
-      "learning_rate": 6.033366834170855e-05,
-      "loss": 5.2417,
-      "step": 39982
-    },
-    {
-      "epoch": 20.851629726205996,
-      "grad_norm": 1.5949069261550903,
-      "learning_rate": 6.033266331658292e-05,
-      "loss": 5.1719,
-      "step": 39983
-    },
-    {
-      "epoch": 20.852151238591915,
-      "grad_norm": 1.499493956565857,
-      "learning_rate": 6.03316582914573e-05,
-      "loss": 5.3564,
-      "step": 39984
-    },
-    {
-      "epoch": 20.852672750977835,
-      "grad_norm": 1.5797473192214966,
-      "learning_rate": 6.033065326633166e-05,
-      "loss": 5.3165,
-      "step": 39985
-    },
-    {
-      "epoch": 20.853194263363754,
-      "grad_norm": 1.549668550491333,
-      "learning_rate": 6.032964824120604e-05,
-      "loss": 5.0041,
-      "step": 39986
-    },
-    {
-      "epoch": 20.853715775749674,
-      "grad_norm": 1.4657344818115234,
-      "learning_rate": 6.03286432160804e-05,
-      "loss": 5.2258,
-      "step": 39987
-    },
-    {
-      "epoch": 20.854237288135593,
-      "grad_norm": 1.5446922779083252,
-      "learning_rate": 6.0327638190954774e-05,
-      "loss": 5.4646,
-      "step": 39988
-    },
-    {
-      "epoch": 20.854758800521513,
-      "grad_norm": 1.5638127326965332,
-      "learning_rate": 6.032663316582915e-05,
-      "loss": 4.5802,
-      "step": 39989
-    },
-    {
-      "epoch": 20.855280312907432,
-      "grad_norm": 1.4872932434082031,
-      "learning_rate": 6.0325628140703517e-05,
-      "loss": 5.5491,
-      "step": 39990
-    },
-    {
-      "epoch": 20.85580182529335,
-      "grad_norm": 1.495995044708252,
-      "learning_rate": 6.0324623115577894e-05,
-      "loss": 5.0393,
-      "step": 39991
-    },
-    {
-      "epoch": 20.85632333767927,
-      "grad_norm": 1.5403873920440674,
-      "learning_rate": 6.032361809045226e-05,
-      "loss": 4.5354,
-      "step": 39992
-    },
-    {
-      "epoch": 20.85684485006519,
-      "grad_norm": 1.4766384363174438,
-      "learning_rate": 6.0322613065326636e-05,
-      "loss": 5.6803,
-      "step": 39993
-    },
-    {
-      "epoch": 20.85736636245111,
-      "grad_norm": 1.4905648231506348,
-      "learning_rate": 6.032160804020101e-05,
-      "loss": 5.097,
-      "step": 39994
-    },
-    {
-      "epoch": 20.857887874837026,
-      "grad_norm": 1.433417558670044,
-      "learning_rate": 6.0320603015075385e-05,
-      "loss": 5.726,
-      "step": 39995
-    },
-    {
-      "epoch": 20.858409387222945,
-      "grad_norm": 2.1533966064453125,
-      "learning_rate": 6.031959798994975e-05,
-      "loss": 5.0923,
-      "step": 39996
-    },
-    {
-      "epoch": 20.858930899608865,
-      "grad_norm": 1.5557854175567627,
-      "learning_rate": 6.031859296482413e-05,
-      "loss": 4.7194,
-      "step": 39997
-    },
-    {
-      "epoch": 20.859452411994784,
-      "grad_norm": 1.5013822317123413,
-      "learning_rate": 6.031758793969849e-05,
-      "loss": 4.8325,
-      "step": 39998
-    },
-    {
-      "epoch": 20.859973924380704,
-      "grad_norm": 1.6386536359786987,
-      "learning_rate": 6.031658291457287e-05,
-      "loss": 5.0144,
-      "step": 39999
-    },
-    {
-      "epoch": 20.860495436766623,
-      "grad_norm": 1.6077555418014526,
-      "learning_rate": 6.031557788944724e-05,
-      "loss": 4.9266,
-      "step": 40000
-    },
-    {
-      "epoch": 20.860495436766623,
-      "eval_loss": 5.391127109527588,
-      "eval_runtime": 42.6724,
-      "eval_samples_per_second": 28.73,
-      "eval_steps_per_second": 3.609,
-      "step": 40000
-    },
-    {
-      "epoch": 20.861016949152543,
-      "grad_norm": 1.6149286031723022,
-      "learning_rate": 6.0314572864321605e-05,
-      "loss": 5.2929,
-      "step": 40001
-    },
-    {
-      "epoch": 20.861538461538462,
-      "grad_norm": 1.5909291505813599,
-      "learning_rate": 6.031356783919598e-05,
-      "loss": 4.9987,
-      "step": 40002
-    },
-    {
-      "epoch": 20.86205997392438,
-      "grad_norm": 1.5366368293762207,
-      "learning_rate": 6.031256281407035e-05,
-      "loss": 5.5527,
-      "step": 40003
-    },
-    {
-      "epoch": 20.8625814863103,
-      "grad_norm": 1.485438585281372,
-      "learning_rate": 6.0311557788944725e-05,
-      "loss": 5.507,
-      "step": 40004
-    },
-    {
-      "epoch": 20.86310299869622,
-      "grad_norm": 1.5034315586090088,
-      "learning_rate": 6.0310552763819096e-05,
-      "loss": 5.1777,
-      "step": 40005
-    },
-    {
-      "epoch": 20.86362451108214,
-      "grad_norm": 1.5249171257019043,
-      "learning_rate": 6.0309547738693474e-05,
-      "loss": 5.3676,
-      "step": 40006
-    },
-    {
-      "epoch": 20.864146023468056,
-      "grad_norm": 1.542660117149353,
-      "learning_rate": 6.030854271356784e-05,
-      "loss": 4.9075,
-      "step": 40007
-    },
-    {
-      "epoch": 20.864667535853975,
-      "grad_norm": 1.4002628326416016,
-      "learning_rate": 6.0307537688442216e-05,
-      "loss": 5.8253,
-      "step": 40008
-    },
-    {
-      "epoch": 20.865189048239895,
-      "grad_norm": 1.4901403188705444,
-      "learning_rate": 6.030653266331659e-05,
-      "loss": 5.288,
-      "step": 40009
-    },
-    {
-      "epoch": 20.865710560625814,
-      "grad_norm": 1.4464964866638184,
-      "learning_rate": 6.0305527638190965e-05,
-      "loss": 5.5196,
-      "step": 40010
-    },
-    {
-      "epoch": 20.866232073011734,
-      "grad_norm": 1.4654141664505005,
-      "learning_rate": 6.030452261306533e-05,
-      "loss": 5.2431,
-      "step": 40011
-    },
-    {
-      "epoch": 20.866753585397653,
-      "grad_norm": 1.608750343322754,
-      "learning_rate": 6.030351758793971e-05,
-      "loss": 4.9492,
-      "step": 40012
-    },
-    {
-      "epoch": 20.867275097783573,
-      "grad_norm": 1.5605833530426025,
-      "learning_rate": 6.030251256281407e-05,
-      "loss": 5.288,
-      "step": 40013
-    },
-    {
-      "epoch": 20.867796610169492,
-      "grad_norm": 1.5330029726028442,
-      "learning_rate": 6.030150753768844e-05,
-      "loss": 5.093,
-      "step": 40014
-    },
-    {
-      "epoch": 20.86831812255541,
-      "grad_norm": 1.636875867843628,
-      "learning_rate": 6.030050251256282e-05,
-      "loss": 5.5587,
-      "step": 40015
-    },
-    {
-      "epoch": 20.86883963494133,
-      "grad_norm": 1.5427441596984863,
-      "learning_rate": 6.0299497487437184e-05,
-      "loss": 5.5173,
-      "step": 40016
-    },
-    {
-      "epoch": 20.86936114732725,
-      "grad_norm": 1.5639699697494507,
-      "learning_rate": 6.029849246231156e-05,
-      "loss": 5.026,
-      "step": 40017
-    },
-    {
-      "epoch": 20.86988265971317,
-      "grad_norm": 1.4935379028320312,
-      "learning_rate": 6.0297487437185926e-05,
-      "loss": 5.3788,
-      "step": 40018
-    },
-    {
-      "epoch": 20.870404172099086,
-      "grad_norm": 1.5325037240982056,
-      "learning_rate": 6.0296482412060304e-05,
-      "loss": 5.2197,
-      "step": 40019
-    },
-    {
-      "epoch": 20.870925684485005,
-      "grad_norm": 1.5644073486328125,
-      "learning_rate": 6.0295477386934675e-05,
-      "loss": 5.2995,
-      "step": 40020
-    },
-    {
-      "epoch": 20.871447196870925,
-      "grad_norm": 1.5578187704086304,
-      "learning_rate": 6.029447236180905e-05,
-      "loss": 5.3543,
-      "step": 40021
-    },
-    {
-      "epoch": 20.871968709256844,
-      "grad_norm": 1.5423818826675415,
-      "learning_rate": 6.029346733668342e-05,
-      "loss": 5.1625,
-      "step": 40022
-    },
-    {
-      "epoch": 20.872490221642764,
-      "grad_norm": 1.5208367109298706,
-      "learning_rate": 6.0292462311557795e-05,
-      "loss": 5.2998,
-      "step": 40023
-    },
-    {
-      "epoch": 20.873011734028683,
-      "grad_norm": 1.4979146718978882,
-      "learning_rate": 6.029145728643216e-05,
-      "loss": 5.5241,
-      "step": 40024
-    },
-    {
-      "epoch": 20.873533246414603,
-      "grad_norm": 1.5274099111557007,
-      "learning_rate": 6.029045226130654e-05,
-      "loss": 5.2596,
-      "step": 40025
-    },
-    {
-      "epoch": 20.874054758800522,
-      "grad_norm": 1.5418825149536133,
-      "learning_rate": 6.028944723618091e-05,
-      "loss": 5.3175,
-      "step": 40026
-    },
-    {
-      "epoch": 20.87457627118644,
-      "grad_norm": 1.4363479614257812,
-      "learning_rate": 6.028844221105527e-05,
-      "loss": 5.4367,
-      "step": 40027
-    },
-    {
-      "epoch": 20.87509778357236,
-      "grad_norm": 1.5427542924880981,
-      "learning_rate": 6.028743718592965e-05,
-      "loss": 5.3513,
-      "step": 40028
-    },
-    {
-      "epoch": 20.87561929595828,
-      "grad_norm": 1.4335319995880127,
-      "learning_rate": 6.028643216080402e-05,
-      "loss": 5.2387,
-      "step": 40029
-    },
-    {
-      "epoch": 20.876140808344196,
-      "grad_norm": 1.3532804250717163,
-      "learning_rate": 6.02854271356784e-05,
-      "loss": 5.1876,
-      "step": 40030
-    },
-    {
-      "epoch": 20.876662320730116,
-      "grad_norm": 1.5246355533599854,
-      "learning_rate": 6.0284422110552764e-05,
-      "loss": 4.9018,
-      "step": 40031
-    },
-    {
-      "epoch": 20.877183833116035,
-      "grad_norm": 1.6411628723144531,
-      "learning_rate": 6.028341708542714e-05,
-      "loss": 4.7051,
-      "step": 40032
-    },
-    {
-      "epoch": 20.877705345501955,
-      "grad_norm": 1.4828377962112427,
-      "learning_rate": 6.0282412060301506e-05,
-      "loss": 5.5515,
-      "step": 40033
-    },
-    {
-      "epoch": 20.878226857887874,
-      "grad_norm": 1.5776827335357666,
-      "learning_rate": 6.0281407035175883e-05,
-      "loss": 5.3735,
-      "step": 40034
-    },
-    {
-      "epoch": 20.878748370273794,
-      "grad_norm": 1.5174503326416016,
-      "learning_rate": 6.0280402010050254e-05,
-      "loss": 5.1007,
-      "step": 40035
-    },
-    {
-      "epoch": 20.879269882659713,
-      "grad_norm": 1.5100367069244385,
-      "learning_rate": 6.027939698492463e-05,
-      "loss": 5.0767,
-      "step": 40036
-    },
-    {
-      "epoch": 20.879791395045633,
-      "grad_norm": 1.4190988540649414,
-      "learning_rate": 6.0278391959798997e-05,
-      "loss": 5.7459,
-      "step": 40037
-    },
-    {
-      "epoch": 20.880312907431552,
-      "grad_norm": 1.489125370979309,
-      "learning_rate": 6.0277386934673374e-05,
-      "loss": 5.5291,
-      "step": 40038
-    },
-    {
-      "epoch": 20.88083441981747,
-      "grad_norm": 1.5731769800186157,
-      "learning_rate": 6.027638190954774e-05,
-      "loss": 4.7828,
-      "step": 40039
-    },
-    {
-      "epoch": 20.88135593220339,
-      "grad_norm": 1.5434705018997192,
-      "learning_rate": 6.0275376884422117e-05,
-      "loss": 5.3044,
-      "step": 40040
-    },
-    {
-      "epoch": 20.88187744458931,
-      "grad_norm": 1.4419794082641602,
-      "learning_rate": 6.027437185929649e-05,
-      "loss": 5.3339,
-      "step": 40041
-    },
-    {
-      "epoch": 20.88239895697523,
-      "grad_norm": 1.5613213777542114,
-      "learning_rate": 6.027336683417085e-05,
-      "loss": 5.1324,
-      "step": 40042
-    },
-    {
-      "epoch": 20.882920469361146,
-      "grad_norm": 1.5402007102966309,
-      "learning_rate": 6.027236180904523e-05,
-      "loss": 4.8199,
-      "step": 40043
-    },
-    {
-      "epoch": 20.883441981747065,
-      "grad_norm": 1.4976166486740112,
-      "learning_rate": 6.0271356783919594e-05,
-      "loss": 5.3158,
-      "step": 40044
-    },
-    {
-      "epoch": 20.883963494132985,
-      "grad_norm": 1.4589430093765259,
-      "learning_rate": 6.027035175879397e-05,
-      "loss": 5.0514,
-      "step": 40045
-    },
-    {
-      "epoch": 20.884485006518904,
-      "grad_norm": 1.5237559080123901,
-      "learning_rate": 6.026934673366834e-05,
-      "loss": 5.452,
-      "step": 40046
-    },
-    {
-      "epoch": 20.885006518904824,
-      "grad_norm": 1.544539451599121,
-      "learning_rate": 6.026834170854272e-05,
-      "loss": 5.4719,
-      "step": 40047
-    },
-    {
-      "epoch": 20.885528031290743,
-      "grad_norm": 1.4824291467666626,
-      "learning_rate": 6.0267336683417085e-05,
-      "loss": 5.2715,
-      "step": 40048
-    },
-    {
-      "epoch": 20.886049543676663,
-      "grad_norm": 1.5072659254074097,
-      "learning_rate": 6.026633165829146e-05,
-      "loss": 5.2842,
-      "step": 40049
-    },
-    {
-      "epoch": 20.886571056062582,
-      "grad_norm": 1.4685763120651245,
-      "learning_rate": 6.0265326633165834e-05,
-      "loss": 5.6424,
-      "step": 40050
-    },
-    {
-      "epoch": 20.8870925684485,
-      "grad_norm": 1.4253073930740356,
-      "learning_rate": 6.026432160804021e-05,
-      "loss": 5.1098,
-      "step": 40051
-    },
-    {
-      "epoch": 20.88761408083442,
-      "grad_norm": 1.632025122642517,
-      "learning_rate": 6.0263316582914576e-05,
-      "loss": 5.1759,
-      "step": 40052
-    },
-    {
-      "epoch": 20.88813559322034,
-      "grad_norm": 1.6079111099243164,
-      "learning_rate": 6.0262311557788954e-05,
-      "loss": 5.2461,
-      "step": 40053
-    },
-    {
-      "epoch": 20.888657105606256,
-      "grad_norm": 1.416792631149292,
-      "learning_rate": 6.026130653266332e-05,
-      "loss": 5.1232,
-      "step": 40054
-    },
-    {
-      "epoch": 20.889178617992176,
-      "grad_norm": 1.4809178113937378,
-      "learning_rate": 6.026030150753769e-05,
-      "loss": 4.9761,
-      "step": 40055
-    },
-    {
-      "epoch": 20.889700130378095,
-      "grad_norm": 1.568742275238037,
-      "learning_rate": 6.025929648241207e-05,
-      "loss": 5.2317,
-      "step": 40056
-    },
-    {
-      "epoch": 20.890221642764015,
-      "grad_norm": 1.4731265306472778,
-      "learning_rate": 6.025829145728643e-05,
-      "loss": 5.6155,
-      "step": 40057
-    },
-    {
-      "epoch": 20.890743155149934,
-      "grad_norm": 1.5951759815216064,
-      "learning_rate": 6.025728643216081e-05,
-      "loss": 5.4796,
-      "step": 40058
-    },
-    {
-      "epoch": 20.891264667535854,
-      "grad_norm": 1.6033629179000854,
-      "learning_rate": 6.025628140703517e-05,
-      "loss": 5.3336,
-      "step": 40059
-    },
-    {
-      "epoch": 20.891786179921773,
-      "grad_norm": 1.514505386352539,
-      "learning_rate": 6.025527638190955e-05,
-      "loss": 5.4626,
-      "step": 40060
-    },
-    {
-      "epoch": 20.892307692307693,
-      "grad_norm": 1.7006981372833252,
-      "learning_rate": 6.025427135678392e-05,
-      "loss": 5.0514,
-      "step": 40061
-    },
-    {
-      "epoch": 20.892829204693612,
-      "grad_norm": 1.4950098991394043,
-      "learning_rate": 6.02532663316583e-05,
-      "loss": 5.4885,
-      "step": 40062
-    },
-    {
-      "epoch": 20.89335071707953,
-      "grad_norm": 1.5678675174713135,
-      "learning_rate": 6.0252261306532664e-05,
-      "loss": 5.4155,
-      "step": 40063
-    },
-    {
-      "epoch": 20.89387222946545,
-      "grad_norm": 1.4668123722076416,
-      "learning_rate": 6.025125628140704e-05,
-      "loss": 5.6335,
-      "step": 40064
-    },
-    {
-      "epoch": 20.89439374185137,
-      "grad_norm": 1.480548620223999,
-      "learning_rate": 6.0250251256281406e-05,
-      "loss": 5.3845,
-      "step": 40065
-    },
-    {
-      "epoch": 20.894915254237286,
-      "grad_norm": 1.6238813400268555,
-      "learning_rate": 6.0249246231155784e-05,
-      "loss": 4.848,
-      "step": 40066
-    },
-    {
-      "epoch": 20.895436766623206,
-      "grad_norm": 1.395859956741333,
-      "learning_rate": 6.0248241206030155e-05,
-      "loss": 5.6509,
-      "step": 40067
-    },
-    {
-      "epoch": 20.895958279009125,
-      "grad_norm": 1.474166989326477,
-      "learning_rate": 6.024723618090452e-05,
-      "loss": 5.745,
-      "step": 40068
-    },
-    {
-      "epoch": 20.896479791395045,
-      "grad_norm": 1.6041585206985474,
-      "learning_rate": 6.02462311557789e-05,
-      "loss": 4.9707,
-      "step": 40069
-    },
-    {
-      "epoch": 20.897001303780964,
-      "grad_norm": 1.6987378597259521,
-      "learning_rate": 6.024522613065327e-05,
-      "loss": 4.9491,
-      "step": 40070
-    },
-    {
-      "epoch": 20.897522816166884,
-      "grad_norm": 1.4874483346939087,
-      "learning_rate": 6.0244221105527646e-05,
-      "loss": 5.4322,
-      "step": 40071
-    },
-    {
-      "epoch": 20.898044328552803,
-      "grad_norm": 1.586398959159851,
-      "learning_rate": 6.024321608040201e-05,
-      "loss": 5.3099,
-      "step": 40072
-    },
-    {
-      "epoch": 20.898565840938723,
-      "grad_norm": 1.4801256656646729,
-      "learning_rate": 6.024221105527639e-05,
-      "loss": 5.2541,
-      "step": 40073
-    },
-    {
-      "epoch": 20.899087353324642,
-      "grad_norm": 1.5793423652648926,
-      "learning_rate": 6.024120603015075e-05,
-      "loss": 5.5438,
-      "step": 40074
-    },
-    {
-      "epoch": 20.89960886571056,
-      "grad_norm": 1.5230382680892944,
-      "learning_rate": 6.024020100502513e-05,
-      "loss": 5.3552,
-      "step": 40075
-    },
-    {
-      "epoch": 20.90013037809648,
-      "grad_norm": 1.529515266418457,
-      "learning_rate": 6.02391959798995e-05,
-      "loss": 5.3351,
-      "step": 40076
-    },
-    {
-      "epoch": 20.9006518904824,
-      "grad_norm": 1.3799694776535034,
-      "learning_rate": 6.023819095477388e-05,
-      "loss": 4.9661,
-      "step": 40077
-    },
-    {
-      "epoch": 20.901173402868316,
-      "grad_norm": 1.545832633972168,
-      "learning_rate": 6.0237185929648244e-05,
-      "loss": 5.1994,
-      "step": 40078
-    },
-    {
-      "epoch": 20.901694915254236,
-      "grad_norm": 1.4758573770523071,
-      "learning_rate": 6.023618090452262e-05,
-      "loss": 5.3581,
-      "step": 40079
-    },
-    {
-      "epoch": 20.902216427640155,
-      "grad_norm": 1.453084111213684,
-      "learning_rate": 6.0235175879396986e-05,
-      "loss": 4.898,
-      "step": 40080
-    },
-    {
-      "epoch": 20.902737940026075,
-      "grad_norm": 1.666993260383606,
-      "learning_rate": 6.023417085427136e-05,
-      "loss": 5.0067,
-      "step": 40081
-    },
-    {
-      "epoch": 20.903259452411994,
-      "grad_norm": 1.5796171426773071,
-      "learning_rate": 6.0233165829145735e-05,
-      "loss": 4.2646,
-      "step": 40082
-    },
-    {
-      "epoch": 20.903780964797914,
-      "grad_norm": 1.6471984386444092,
-      "learning_rate": 6.02321608040201e-05,
-      "loss": 5.2119,
-      "step": 40083
-    },
-    {
-      "epoch": 20.904302477183833,
-      "grad_norm": 1.579105019569397,
-      "learning_rate": 6.023115577889448e-05,
-      "loss": 5.2941,
-      "step": 40084
-    },
-    {
-      "epoch": 20.904823989569753,
-      "grad_norm": 1.777175784111023,
-      "learning_rate": 6.023015075376884e-05,
-      "loss": 5.3527,
-      "step": 40085
-    },
-    {
-      "epoch": 20.905345501955672,
-      "grad_norm": 1.5237518548965454,
-      "learning_rate": 6.022914572864322e-05,
-      "loss": 5.469,
-      "step": 40086
-    },
-    {
-      "epoch": 20.90586701434159,
-      "grad_norm": 1.5888643264770508,
-      "learning_rate": 6.022814070351759e-05,
-      "loss": 5.3022,
-      "step": 40087
-    },
-    {
-      "epoch": 20.90638852672751,
-      "grad_norm": 1.5788813829421997,
-      "learning_rate": 6.022713567839197e-05,
-      "loss": 5.6265,
-      "step": 40088
-    },
-    {
-      "epoch": 20.90691003911343,
-      "grad_norm": 1.5781879425048828,
-      "learning_rate": 6.022613065326633e-05,
-      "loss": 5.3788,
-      "step": 40089
-    },
-    {
-      "epoch": 20.907431551499347,
-      "grad_norm": 1.6271270513534546,
-      "learning_rate": 6.022512562814071e-05,
-      "loss": 4.7775,
-      "step": 40090
-    },
-    {
-      "epoch": 20.907953063885266,
-      "grad_norm": 1.5963928699493408,
-      "learning_rate": 6.0224120603015074e-05,
-      "loss": 5.0235,
-      "step": 40091
-    },
-    {
-      "epoch": 20.908474576271185,
-      "grad_norm": 1.5063369274139404,
-      "learning_rate": 6.022311557788945e-05,
-      "loss": 4.9472,
-      "step": 40092
-    },
-    {
-      "epoch": 20.908996088657105,
-      "grad_norm": 1.6844631433486938,
-      "learning_rate": 6.022211055276382e-05,
-      "loss": 4.9336,
-      "step": 40093
-    },
-    {
-      "epoch": 20.909517601043024,
-      "grad_norm": 1.438459873199463,
-      "learning_rate": 6.022110552763819e-05,
-      "loss": 5.439,
-      "step": 40094
-    },
-    {
-      "epoch": 20.910039113428944,
-      "grad_norm": 1.4677313566207886,
-      "learning_rate": 6.0220100502512565e-05,
-      "loss": 5.3671,
-      "step": 40095
-    },
-    {
-      "epoch": 20.910560625814863,
-      "grad_norm": 1.5824062824249268,
-      "learning_rate": 6.0219095477386936e-05,
-      "loss": 5.3408,
-      "step": 40096
-    },
-    {
-      "epoch": 20.911082138200783,
-      "grad_norm": 1.68326997756958,
-      "learning_rate": 6.0218090452261314e-05,
-      "loss": 5.1346,
-      "step": 40097
-    },
-    {
-      "epoch": 20.911603650586702,
-      "grad_norm": 1.5087056159973145,
-      "learning_rate": 6.021708542713568e-05,
-      "loss": 4.9556,
-      "step": 40098
-    },
-    {
-      "epoch": 20.91212516297262,
-      "grad_norm": 1.4743117094039917,
-      "learning_rate": 6.0216080402010056e-05,
-      "loss": 5.5545,
-      "step": 40099
-    },
-    {
-      "epoch": 20.91264667535854,
-      "grad_norm": 1.4857059717178345,
-      "learning_rate": 6.021507537688442e-05,
-      "loss": 5.0298,
-      "step": 40100
-    },
-    {
-      "epoch": 20.91316818774446,
-      "grad_norm": 1.5393195152282715,
-      "learning_rate": 6.02140703517588e-05,
-      "loss": 5.0775,
-      "step": 40101
-    },
-    {
-      "epoch": 20.913689700130377,
-      "grad_norm": 1.5671625137329102,
-      "learning_rate": 6.021306532663317e-05,
-      "loss": 4.6823,
-      "step": 40102
-    },
-    {
-      "epoch": 20.914211212516296,
-      "grad_norm": 1.5019588470458984,
-      "learning_rate": 6.021206030150755e-05,
-      "loss": 5.2135,
-      "step": 40103
-    },
-    {
-      "epoch": 20.914732724902215,
-      "grad_norm": 1.4813984632492065,
-      "learning_rate": 6.021105527638191e-05,
-      "loss": 5.5481,
-      "step": 40104
-    },
-    {
-      "epoch": 20.915254237288135,
-      "grad_norm": 1.5667246580123901,
-      "learning_rate": 6.021005025125629e-05,
-      "loss": 5.4134,
-      "step": 40105
-    },
-    {
-      "epoch": 20.915775749674054,
-      "grad_norm": 1.424500823020935,
-      "learning_rate": 6.020904522613065e-05,
-      "loss": 5.5741,
-      "step": 40106
-    },
-    {
-      "epoch": 20.916297262059974,
-      "grad_norm": 1.5366369485855103,
-      "learning_rate": 6.0208040201005024e-05,
-      "loss": 5.1487,
-      "step": 40107
-    },
-    {
-      "epoch": 20.916818774445893,
-      "grad_norm": 1.5908422470092773,
-      "learning_rate": 6.02070351758794e-05,
-      "loss": 4.9716,
-      "step": 40108
-    },
-    {
-      "epoch": 20.917340286831813,
-      "grad_norm": 1.495580792427063,
-      "learning_rate": 6.0206030150753767e-05,
-      "loss": 5.1251,
-      "step": 40109
-    },
-    {
-      "epoch": 20.917861799217732,
-      "grad_norm": 1.63461172580719,
-      "learning_rate": 6.0205025125628144e-05,
-      "loss": 4.99,
-      "step": 40110
-    },
-    {
-      "epoch": 20.91838331160365,
-      "grad_norm": 1.5443072319030762,
-      "learning_rate": 6.020402010050251e-05,
-      "loss": 5.5153,
-      "step": 40111
-    },
-    {
-      "epoch": 20.91890482398957,
-      "grad_norm": 1.5649410486221313,
-      "learning_rate": 6.0203015075376886e-05,
-      "loss": 5.2708,
-      "step": 40112
-    },
-    {
-      "epoch": 20.919426336375487,
-      "grad_norm": 1.5051201581954956,
-      "learning_rate": 6.020201005025126e-05,
-      "loss": 4.9989,
-      "step": 40113
-    },
-    {
-      "epoch": 20.919947848761407,
-      "grad_norm": 1.5114772319793701,
-      "learning_rate": 6.0201005025125635e-05,
-      "loss": 5.2138,
-      "step": 40114
-    },
-    {
-      "epoch": 20.920469361147326,
-      "grad_norm": 1.5812599658966064,
-      "learning_rate": 6.02e-05,
-      "loss": 5.666,
-      "step": 40115
-    },
-    {
-      "epoch": 20.920990873533245,
-      "grad_norm": 1.653073787689209,
-      "learning_rate": 6.019899497487438e-05,
-      "loss": 5.2132,
-      "step": 40116
-    },
-    {
-      "epoch": 20.921512385919165,
-      "grad_norm": 1.5771384239196777,
-      "learning_rate": 6.019798994974875e-05,
-      "loss": 5.0369,
-      "step": 40117
-    },
-    {
-      "epoch": 20.922033898305084,
-      "grad_norm": 1.4368616342544556,
-      "learning_rate": 6.0196984924623126e-05,
-      "loss": 5.1551,
-      "step": 40118
-    },
-    {
-      "epoch": 20.922555410691004,
-      "grad_norm": 1.5372655391693115,
-      "learning_rate": 6.019597989949749e-05,
-      "loss": 4.8848,
-      "step": 40119
-    },
-    {
-      "epoch": 20.923076923076923,
-      "grad_norm": 1.7004553079605103,
-      "learning_rate": 6.0194974874371855e-05,
-      "loss": 5.3764,
-      "step": 40120
-    },
-    {
-      "epoch": 20.923598435462843,
-      "grad_norm": 1.6447502374649048,
-      "learning_rate": 6.019396984924623e-05,
-      "loss": 5.3838,
-      "step": 40121
-    },
-    {
-      "epoch": 20.924119947848762,
-      "grad_norm": 1.6129873991012573,
-      "learning_rate": 6.0192964824120604e-05,
-      "loss": 5.2763,
-      "step": 40122
-    },
-    {
-      "epoch": 20.92464146023468,
-      "grad_norm": 1.7042375802993774,
-      "learning_rate": 6.019195979899498e-05,
-      "loss": 4.6105,
-      "step": 40123
-    },
-    {
-      "epoch": 20.9251629726206,
-      "grad_norm": 1.54409658908844,
-      "learning_rate": 6.0190954773869346e-05,
-      "loss": 5.0121,
-      "step": 40124
-    },
-    {
-      "epoch": 20.92568448500652,
-      "grad_norm": 1.4965101480484009,
-      "learning_rate": 6.0189949748743724e-05,
-      "loss": 5.1868,
-      "step": 40125
-    },
-    {
-      "epoch": 20.926205997392437,
-      "grad_norm": 1.6124176979064941,
-      "learning_rate": 6.018894472361809e-05,
-      "loss": 5.3203,
-      "step": 40126
-    },
-    {
-      "epoch": 20.926727509778356,
-      "grad_norm": 1.590895414352417,
-      "learning_rate": 6.0187939698492466e-05,
-      "loss": 5.5363,
-      "step": 40127
-    },
-    {
-      "epoch": 20.927249022164276,
-      "grad_norm": 1.6438887119293213,
-      "learning_rate": 6.018693467336684e-05,
-      "loss": 4.9561,
-      "step": 40128
-    },
-    {
-      "epoch": 20.927770534550195,
-      "grad_norm": 1.4365217685699463,
-      "learning_rate": 6.0185929648241215e-05,
-      "loss": 5.3499,
-      "step": 40129
-    },
-    {
-      "epoch": 20.928292046936114,
-      "grad_norm": 1.5837492942810059,
-      "learning_rate": 6.018492462311558e-05,
-      "loss": 5.1114,
-      "step": 40130
-    },
-    {
-      "epoch": 20.928813559322034,
-      "grad_norm": 1.5927616357803345,
-      "learning_rate": 6.018391959798996e-05,
-      "loss": 5.5207,
-      "step": 40131
-    },
-    {
-      "epoch": 20.929335071707953,
-      "grad_norm": 1.4584507942199707,
-      "learning_rate": 6.018291457286432e-05,
-      "loss": 5.6504,
-      "step": 40132
-    },
-    {
-      "epoch": 20.929856584093873,
-      "grad_norm": 1.5015032291412354,
-      "learning_rate": 6.01819095477387e-05,
-      "loss": 5.5236,
-      "step": 40133
-    },
-    {
-      "epoch": 20.930378096479792,
-      "grad_norm": 1.580650806427002,
-      "learning_rate": 6.018090452261307e-05,
-      "loss": 5.0829,
-      "step": 40134
-    },
-    {
-      "epoch": 20.93089960886571,
-      "grad_norm": 1.4846616983413696,
-      "learning_rate": 6.0179899497487434e-05,
-      "loss": 5.2264,
-      "step": 40135
-    },
-    {
-      "epoch": 20.93142112125163,
-      "grad_norm": 1.5967588424682617,
-      "learning_rate": 6.017889447236181e-05,
-      "loss": 5.0715,
-      "step": 40136
-    },
-    {
-      "epoch": 20.931942633637547,
-      "grad_norm": 1.533526062965393,
-      "learning_rate": 6.017788944723618e-05,
-      "loss": 5.0155,
-      "step": 40137
-    },
-    {
-      "epoch": 20.932464146023467,
-      "grad_norm": 1.5463529825210571,
-      "learning_rate": 6.017688442211056e-05,
-      "loss": 5.3019,
-      "step": 40138
-    },
-    {
-      "epoch": 20.932985658409386,
-      "grad_norm": 1.6065436601638794,
-      "learning_rate": 6.0175879396984925e-05,
-      "loss": 5.3666,
-      "step": 40139
-    },
-    {
-      "epoch": 20.933507170795306,
-      "grad_norm": 1.5459471940994263,
-      "learning_rate": 6.01748743718593e-05,
-      "loss": 5.6053,
-      "step": 40140
-    },
-    {
-      "epoch": 20.934028683181225,
-      "grad_norm": 1.467740774154663,
-      "learning_rate": 6.017386934673367e-05,
-      "loss": 5.6041,
-      "step": 40141
-    },
-    {
-      "epoch": 20.934550195567144,
-      "grad_norm": 1.5223225355148315,
-      "learning_rate": 6.0172864321608045e-05,
-      "loss": 5.4651,
-      "step": 40142
-    },
-    {
-      "epoch": 20.935071707953064,
-      "grad_norm": 1.5120021104812622,
-      "learning_rate": 6.0171859296482416e-05,
-      "loss": 5.4131,
-      "step": 40143
-    },
-    {
-      "epoch": 20.935593220338983,
-      "grad_norm": 1.5661296844482422,
-      "learning_rate": 6.0170854271356794e-05,
-      "loss": 5.1209,
-      "step": 40144
-    },
-    {
-      "epoch": 20.936114732724903,
-      "grad_norm": 1.5404590368270874,
-      "learning_rate": 6.016984924623116e-05,
-      "loss": 4.9498,
-      "step": 40145
-    },
-    {
-      "epoch": 20.936636245110822,
-      "grad_norm": 1.4349257946014404,
-      "learning_rate": 6.0168844221105536e-05,
-      "loss": 5.6255,
-      "step": 40146
-    },
-    {
-      "epoch": 20.937157757496742,
-      "grad_norm": 1.528143048286438,
-      "learning_rate": 6.01678391959799e-05,
-      "loss": 5.5342,
-      "step": 40147
-    },
-    {
-      "epoch": 20.93767926988266,
-      "grad_norm": 1.3296117782592773,
-      "learning_rate": 6.016683417085427e-05,
-      "loss": 5.5537,
-      "step": 40148
-    },
-    {
-      "epoch": 20.938200782268577,
-      "grad_norm": 1.451151967048645,
-      "learning_rate": 6.016582914572865e-05,
-      "loss": 4.9141,
-      "step": 40149
-    },
-    {
-      "epoch": 20.938722294654497,
-      "grad_norm": 1.4518226385116577,
-      "learning_rate": 6.0164824120603014e-05,
-      "loss": 5.6612,
-      "step": 40150
-    },
-    {
-      "epoch": 20.939243807040416,
-      "grad_norm": 1.462399959564209,
-      "learning_rate": 6.016381909547739e-05,
-      "loss": 5.1106,
-      "step": 40151
-    },
-    {
-      "epoch": 20.939765319426336,
-      "grad_norm": 1.4577128887176514,
-      "learning_rate": 6.0162814070351756e-05,
-      "loss": 5.1552,
-      "step": 40152
-    },
-    {
-      "epoch": 20.940286831812255,
-      "grad_norm": 1.551796555519104,
-      "learning_rate": 6.0161809045226133e-05,
-      "loss": 4.8739,
-      "step": 40153
-    },
-    {
-      "epoch": 20.940808344198174,
-      "grad_norm": 1.4434036016464233,
-      "learning_rate": 6.0160804020100505e-05,
-      "loss": 5.5439,
-      "step": 40154
-    },
-    {
-      "epoch": 20.941329856584094,
-      "grad_norm": 1.5747034549713135,
-      "learning_rate": 6.015979899497488e-05,
-      "loss": 5.4643,
-      "step": 40155
-    },
-    {
-      "epoch": 20.941851368970013,
-      "grad_norm": 1.5744515657424927,
-      "learning_rate": 6.0158793969849247e-05,
-      "loss": 4.9625,
-      "step": 40156
-    },
-    {
-      "epoch": 20.942372881355933,
-      "grad_norm": 1.5187890529632568,
-      "learning_rate": 6.0157788944723624e-05,
-      "loss": 5.1006,
-      "step": 40157
-    },
-    {
-      "epoch": 20.942894393741852,
-      "grad_norm": 1.6071891784667969,
-      "learning_rate": 6.015678391959799e-05,
-      "loss": 5.3865,
-      "step": 40158
-    },
-    {
-      "epoch": 20.943415906127772,
-      "grad_norm": 1.5806126594543457,
-      "learning_rate": 6.0155778894472367e-05,
-      "loss": 4.5366,
-      "step": 40159
-    },
-    {
-      "epoch": 20.94393741851369,
-      "grad_norm": 1.518614649772644,
-      "learning_rate": 6.015477386934674e-05,
-      "loss": 5.5511,
-      "step": 40160
-    },
-    {
-      "epoch": 20.944458930899607,
-      "grad_norm": 1.5267651081085205,
-      "learning_rate": 6.01537688442211e-05,
-      "loss": 5.4554,
-      "step": 40161
-    },
-    {
-      "epoch": 20.944980443285527,
-      "grad_norm": 1.4261300563812256,
-      "learning_rate": 6.015276381909548e-05,
-      "loss": 5.1089,
-      "step": 40162
-    },
-    {
-      "epoch": 20.945501955671446,
-      "grad_norm": 1.571205735206604,
-      "learning_rate": 6.015175879396985e-05,
-      "loss": 5.0308,
-      "step": 40163
-    },
-    {
-      "epoch": 20.946023468057366,
-      "grad_norm": 1.557940125465393,
-      "learning_rate": 6.015075376884423e-05,
-      "loss": 5.1122,
-      "step": 40164
-    },
-    {
-      "epoch": 20.946544980443285,
-      "grad_norm": 1.5680867433547974,
-      "learning_rate": 6.014974874371859e-05,
-      "loss": 5.3282,
-      "step": 40165
-    },
-    {
-      "epoch": 20.947066492829205,
-      "grad_norm": 1.5002192258834839,
-      "learning_rate": 6.014874371859297e-05,
-      "loss": 5.4396,
-      "step": 40166
-    },
-    {
-      "epoch": 20.947588005215124,
-      "grad_norm": 1.5569390058517456,
-      "learning_rate": 6.0147738693467335e-05,
-      "loss": 5.0725,
-      "step": 40167
-    },
-    {
-      "epoch": 20.948109517601043,
-      "grad_norm": 1.3556677103042603,
-      "learning_rate": 6.014673366834171e-05,
-      "loss": 5.8337,
-      "step": 40168
-    },
-    {
-      "epoch": 20.948631029986963,
-      "grad_norm": 1.4943475723266602,
-      "learning_rate": 6.0145728643216084e-05,
-      "loss": 5.2591,
-      "step": 40169
-    },
-    {
-      "epoch": 20.949152542372882,
-      "grad_norm": 1.454598307609558,
-      "learning_rate": 6.014472361809046e-05,
-      "loss": 4.9122,
-      "step": 40170
-    },
-    {
-      "epoch": 20.949674054758802,
-      "grad_norm": 1.5308911800384521,
-      "learning_rate": 6.0143718592964826e-05,
-      "loss": 5.2948,
-      "step": 40171
-    },
-    {
-      "epoch": 20.95019556714472,
-      "grad_norm": 1.5228846073150635,
-      "learning_rate": 6.0142713567839204e-05,
-      "loss": 4.532,
-      "step": 40172
-    },
-    {
-      "epoch": 20.950717079530637,
-      "grad_norm": 1.5819575786590576,
-      "learning_rate": 6.014170854271357e-05,
-      "loss": 5.5765,
-      "step": 40173
-    },
-    {
-      "epoch": 20.951238591916557,
-      "grad_norm": 1.5619804859161377,
-      "learning_rate": 6.014070351758794e-05,
-      "loss": 5.2392,
-      "step": 40174
-    },
-    {
-      "epoch": 20.951760104302476,
-      "grad_norm": 1.698674201965332,
-      "learning_rate": 6.013969849246232e-05,
-      "loss": 5.2774,
-      "step": 40175
-    },
-    {
-      "epoch": 20.952281616688396,
-      "grad_norm": 1.613785743713379,
-      "learning_rate": 6.013869346733668e-05,
-      "loss": 5.2368,
-      "step": 40176
-    },
-    {
-      "epoch": 20.952803129074315,
-      "grad_norm": 1.4727815389633179,
-      "learning_rate": 6.013768844221106e-05,
-      "loss": 5.43,
-      "step": 40177
-    },
-    {
-      "epoch": 20.953324641460235,
-      "grad_norm": 1.5783919095993042,
-      "learning_rate": 6.013668341708542e-05,
-      "loss": 4.9756,
-      "step": 40178
-    },
-    {
-      "epoch": 20.953846153846154,
-      "grad_norm": 1.5336049795150757,
-      "learning_rate": 6.01356783919598e-05,
-      "loss": 5.6235,
-      "step": 40179
-    },
-    {
-      "epoch": 20.954367666232073,
-      "grad_norm": 1.4777024984359741,
-      "learning_rate": 6.013467336683417e-05,
-      "loss": 5.5426,
-      "step": 40180
-    },
-    {
-      "epoch": 20.954889178617993,
-      "grad_norm": 1.597744107246399,
-      "learning_rate": 6.013366834170855e-05,
-      "loss": 4.9515,
-      "step": 40181
-    },
-    {
-      "epoch": 20.955410691003912,
-      "grad_norm": 1.4273394346237183,
-      "learning_rate": 6.0132663316582914e-05,
-      "loss": 5.4339,
-      "step": 40182
-    },
-    {
-      "epoch": 20.955932203389832,
-      "grad_norm": 1.486538052558899,
-      "learning_rate": 6.013165829145729e-05,
-      "loss": 5.5648,
-      "step": 40183
-    },
-    {
-      "epoch": 20.95645371577575,
-      "grad_norm": 1.4847687482833862,
-      "learning_rate": 6.013065326633166e-05,
-      "loss": 5.1431,
-      "step": 40184
-    },
-    {
-      "epoch": 20.956975228161667,
-      "grad_norm": 1.5765632390975952,
-      "learning_rate": 6.012964824120604e-05,
-      "loss": 5.2142,
-      "step": 40185
-    },
-    {
-      "epoch": 20.957496740547587,
-      "grad_norm": 1.5786784887313843,
-      "learning_rate": 6.0128643216080405e-05,
-      "loss": 4.8707,
-      "step": 40186
-    },
-    {
-      "epoch": 20.958018252933506,
-      "grad_norm": 1.5200437307357788,
-      "learning_rate": 6.012763819095477e-05,
-      "loss": 5.3054,
-      "step": 40187
-    },
-    {
-      "epoch": 20.958539765319426,
-      "grad_norm": 1.5504025220870972,
-      "learning_rate": 6.012663316582915e-05,
-      "loss": 5.5163,
-      "step": 40188
-    },
-    {
-      "epoch": 20.959061277705345,
-      "grad_norm": 1.545135498046875,
-      "learning_rate": 6.012562814070352e-05,
-      "loss": 5.623,
-      "step": 40189
-    },
-    {
-      "epoch": 20.959582790091265,
-      "grad_norm": 1.5906264781951904,
-      "learning_rate": 6.0124623115577896e-05,
-      "loss": 5.4274,
-      "step": 40190
-    },
-    {
-      "epoch": 20.960104302477184,
-      "grad_norm": 1.453804850578308,
-      "learning_rate": 6.012361809045226e-05,
-      "loss": 5.2976,
-      "step": 40191
-    },
-    {
-      "epoch": 20.960625814863103,
-      "grad_norm": 1.5727537870407104,
-      "learning_rate": 6.012261306532664e-05,
-      "loss": 5.2347,
-      "step": 40192
-    },
-    {
-      "epoch": 20.961147327249023,
-      "grad_norm": 1.548128604888916,
-      "learning_rate": 6.0121608040201e-05,
-      "loss": 5.3548,
-      "step": 40193
-    },
-    {
-      "epoch": 20.961668839634942,
-      "grad_norm": 1.5605583190917969,
-      "learning_rate": 6.012060301507538e-05,
-      "loss": 5.481,
-      "step": 40194
-    },
-    {
-      "epoch": 20.962190352020862,
-      "grad_norm": 1.5379910469055176,
-      "learning_rate": 6.011959798994975e-05,
-      "loss": 5.6524,
-      "step": 40195
-    },
-    {
-      "epoch": 20.96271186440678,
-      "grad_norm": 1.6494733095169067,
-      "learning_rate": 6.011859296482413e-05,
-      "loss": 5.2571,
-      "step": 40196
-    },
-    {
-      "epoch": 20.963233376792697,
-      "grad_norm": 1.5354130268096924,
-      "learning_rate": 6.0117587939698494e-05,
-      "loss": 5.0124,
-      "step": 40197
-    },
-    {
-      "epoch": 20.963754889178617,
-      "grad_norm": 1.546015977859497,
-      "learning_rate": 6.011658291457287e-05,
-      "loss": 5.2774,
-      "step": 40198
-    },
-    {
-      "epoch": 20.964276401564536,
-      "grad_norm": 1.5150935649871826,
-      "learning_rate": 6.0115577889447236e-05,
-      "loss": 5.215,
-      "step": 40199
-    },
-    {
-      "epoch": 20.964797913950456,
-      "grad_norm": 1.6140588521957397,
-      "learning_rate": 6.011457286432161e-05,
-      "loss": 5.0133,
-      "step": 40200
-    },
-    {
-      "epoch": 20.965319426336375,
-      "grad_norm": 1.518580436706543,
-      "learning_rate": 6.0113567839195985e-05,
-      "loss": 5.2849,
-      "step": 40201
-    },
-    {
-      "epoch": 20.965840938722295,
-      "grad_norm": 1.620582938194275,
-      "learning_rate": 6.011256281407035e-05,
-      "loss": 4.7791,
-      "step": 40202
-    },
-    {
-      "epoch": 20.966362451108214,
-      "grad_norm": 1.541845440864563,
-      "learning_rate": 6.011155778894473e-05,
-      "loss": 5.3425,
-      "step": 40203
-    },
-    {
-      "epoch": 20.966883963494134,
-      "grad_norm": 1.657497525215149,
-      "learning_rate": 6.01105527638191e-05,
-      "loss": 5.6106,
-      "step": 40204
-    },
-    {
-      "epoch": 20.967405475880053,
-      "grad_norm": 1.4770286083221436,
-      "learning_rate": 6.0109547738693476e-05,
-      "loss": 5.6313,
-      "step": 40205
-    },
-    {
-      "epoch": 20.967926988265972,
-      "grad_norm": 1.4619706869125366,
-      "learning_rate": 6.010854271356784e-05,
-      "loss": 5.228,
-      "step": 40206
-    },
-    {
-      "epoch": 20.968448500651892,
-      "grad_norm": 1.4926683902740479,
-      "learning_rate": 6.010753768844222e-05,
-      "loss": 5.41,
-      "step": 40207
-    },
-    {
-      "epoch": 20.96897001303781,
-      "grad_norm": 1.5441648960113525,
-      "learning_rate": 6.010653266331658e-05,
-      "loss": 5.5308,
-      "step": 40208
-    },
-    {
-      "epoch": 20.969491525423727,
-      "grad_norm": 1.4317371845245361,
-      "learning_rate": 6.010552763819096e-05,
-      "loss": 5.3666,
-      "step": 40209
-    },
-    {
-      "epoch": 20.970013037809647,
-      "grad_norm": 1.6250205039978027,
-      "learning_rate": 6.010452261306533e-05,
-      "loss": 5.25,
-      "step": 40210
-    },
-    {
-      "epoch": 20.970534550195566,
-      "grad_norm": 1.6558005809783936,
-      "learning_rate": 6.010351758793971e-05,
-      "loss": 4.9811,
-      "step": 40211
-    },
-    {
-      "epoch": 20.971056062581486,
-      "grad_norm": 1.4917068481445312,
-      "learning_rate": 6.010251256281407e-05,
-      "loss": 5.4748,
-      "step": 40212
-    },
-    {
-      "epoch": 20.971577574967405,
-      "grad_norm": 1.6469484567642212,
-      "learning_rate": 6.010150753768844e-05,
-      "loss": 5.1952,
-      "step": 40213
-    },
-    {
-      "epoch": 20.972099087353325,
-      "grad_norm": 1.5276374816894531,
-      "learning_rate": 6.0100502512562815e-05,
-      "loss": 5.3595,
-      "step": 40214
-    },
-    {
-      "epoch": 20.972620599739244,
-      "grad_norm": 1.4749454259872437,
-      "learning_rate": 6.0099497487437186e-05,
-      "loss": 4.772,
-      "step": 40215
-    },
-    {
-      "epoch": 20.973142112125164,
-      "grad_norm": 1.6171891689300537,
-      "learning_rate": 6.0098492462311564e-05,
-      "loss": 5.2312,
-      "step": 40216
-    },
-    {
-      "epoch": 20.973663624511083,
-      "grad_norm": 1.4953112602233887,
-      "learning_rate": 6.009748743718593e-05,
-      "loss": 5.4778,
-      "step": 40217
-    },
-    {
-      "epoch": 20.974185136897002,
-      "grad_norm": 1.4792813062667847,
-      "learning_rate": 6.0096482412060306e-05,
-      "loss": 5.6235,
-      "step": 40218
-    },
-    {
-      "epoch": 20.974706649282922,
-      "grad_norm": 1.4887871742248535,
-      "learning_rate": 6.009547738693467e-05,
-      "loss": 5.4614,
-      "step": 40219
-    },
-    {
-      "epoch": 20.975228161668838,
-      "grad_norm": 1.5162588357925415,
-      "learning_rate": 6.009447236180905e-05,
-      "loss": 5.1471,
-      "step": 40220
-    },
-    {
-      "epoch": 20.975749674054757,
-      "grad_norm": 1.6672980785369873,
-      "learning_rate": 6.009346733668342e-05,
-      "loss": 5.1384,
-      "step": 40221
-    },
-    {
-      "epoch": 20.976271186440677,
-      "grad_norm": 1.4811428785324097,
-      "learning_rate": 6.00924623115578e-05,
-      "loss": 5.3882,
-      "step": 40222
-    },
-    {
-      "epoch": 20.976792698826596,
-      "grad_norm": 1.4637203216552734,
-      "learning_rate": 6.009145728643216e-05,
-      "loss": 4.81,
-      "step": 40223
-    },
-    {
-      "epoch": 20.977314211212516,
-      "grad_norm": 1.538106083869934,
-      "learning_rate": 6.009045226130654e-05,
-      "loss": 5.2851,
-      "step": 40224
-    },
-    {
-      "epoch": 20.977835723598435,
-      "grad_norm": 1.6831083297729492,
-      "learning_rate": 6.008944723618091e-05,
-      "loss": 5.1094,
-      "step": 40225
-    },
-    {
-      "epoch": 20.978357235984355,
-      "grad_norm": 1.6933670043945312,
-      "learning_rate": 6.008844221105529e-05,
-      "loss": 4.8109,
-      "step": 40226
-    },
-    {
-      "epoch": 20.978878748370274,
-      "grad_norm": 1.7358567714691162,
-      "learning_rate": 6.008743718592965e-05,
-      "loss": 5.0023,
-      "step": 40227
-    },
-    {
-      "epoch": 20.979400260756194,
-      "grad_norm": 1.5748038291931152,
-      "learning_rate": 6.0086432160804017e-05,
-      "loss": 5.2449,
-      "step": 40228
-    },
-    {
-      "epoch": 20.979921773142113,
-      "grad_norm": 1.4807603359222412,
-      "learning_rate": 6.0085427135678394e-05,
-      "loss": 5.4888,
-      "step": 40229
-    },
-    {
-      "epoch": 20.980443285528033,
-      "grad_norm": 1.5149019956588745,
-      "learning_rate": 6.0084422110552765e-05,
-      "loss": 5.4159,
-      "step": 40230
-    },
-    {
-      "epoch": 20.980964797913952,
-      "grad_norm": 1.6893212795257568,
-      "learning_rate": 6.008341708542714e-05,
-      "loss": 5.2402,
-      "step": 40231
-    },
-    {
-      "epoch": 20.98148631029987,
-      "grad_norm": 1.4182393550872803,
-      "learning_rate": 6.008241206030151e-05,
-      "loss": 5.3413,
-      "step": 40232
-    },
-    {
-      "epoch": 20.982007822685787,
-      "grad_norm": 1.4657871723175049,
-      "learning_rate": 6.0081407035175885e-05,
-      "loss": 5.4494,
-      "step": 40233
-    },
-    {
-      "epoch": 20.982529335071707,
-      "grad_norm": 1.7487273216247559,
-      "learning_rate": 6.008040201005025e-05,
-      "loss": 4.4625,
-      "step": 40234
-    },
-    {
-      "epoch": 20.983050847457626,
-      "grad_norm": 1.4355273246765137,
-      "learning_rate": 6.007939698492463e-05,
-      "loss": 5.463,
-      "step": 40235
-    },
-    {
-      "epoch": 20.983572359843546,
-      "grad_norm": 1.5604431629180908,
-      "learning_rate": 6.0078391959799e-05,
-      "loss": 5.0324,
-      "step": 40236
-    },
-    {
-      "epoch": 20.984093872229465,
-      "grad_norm": 1.652076005935669,
-      "learning_rate": 6.0077386934673376e-05,
-      "loss": 4.6932,
-      "step": 40237
-    },
-    {
-      "epoch": 20.984615384615385,
-      "grad_norm": 1.5230931043624878,
-      "learning_rate": 6.007638190954774e-05,
-      "loss": 5.2558,
-      "step": 40238
-    },
-    {
-      "epoch": 20.985136897001304,
-      "grad_norm": 1.5501974821090698,
-      "learning_rate": 6.007537688442212e-05,
-      "loss": 4.6476,
-      "step": 40239
-    },
-    {
-      "epoch": 20.985658409387224,
-      "grad_norm": 1.4960957765579224,
-      "learning_rate": 6.007437185929648e-05,
-      "loss": 5.3601,
-      "step": 40240
-    },
-    {
-      "epoch": 20.986179921773143,
-      "grad_norm": 1.4244786500930786,
-      "learning_rate": 6.0073366834170854e-05,
-      "loss": 5.2403,
-      "step": 40241
-    },
-    {
-      "epoch": 20.986701434159063,
-      "grad_norm": 1.503225326538086,
-      "learning_rate": 6.007236180904523e-05,
-      "loss": 5.4623,
-      "step": 40242
-    },
-    {
-      "epoch": 20.987222946544982,
-      "grad_norm": 1.591659426689148,
-      "learning_rate": 6.0071356783919596e-05,
-      "loss": 5.2805,
-      "step": 40243
-    },
-    {
-      "epoch": 20.987744458930898,
-      "grad_norm": 1.465066909790039,
-      "learning_rate": 6.0070351758793974e-05,
-      "loss": 5.2604,
-      "step": 40244
-    },
-    {
-      "epoch": 20.988265971316817,
-      "grad_norm": 1.5989128351211548,
-      "learning_rate": 6.006934673366834e-05,
-      "loss": 5.1662,
-      "step": 40245
-    },
-    {
-      "epoch": 20.988787483702737,
-      "grad_norm": 1.4269365072250366,
-      "learning_rate": 6.0068341708542716e-05,
-      "loss": 5.4351,
-      "step": 40246
-    },
-    {
-      "epoch": 20.989308996088656,
-      "grad_norm": 1.6657384634017944,
-      "learning_rate": 6.006733668341709e-05,
-      "loss": 5.5203,
-      "step": 40247
-    },
-    {
-      "epoch": 20.989830508474576,
-      "grad_norm": 1.4499242305755615,
-      "learning_rate": 6.0066331658291465e-05,
-      "loss": 5.2816,
-      "step": 40248
-    },
-    {
-      "epoch": 20.990352020860495,
-      "grad_norm": 1.4440535306930542,
-      "learning_rate": 6.006532663316583e-05,
-      "loss": 5.416,
-      "step": 40249
-    },
-    {
-      "epoch": 20.990873533246415,
-      "grad_norm": 1.5198662281036377,
-      "learning_rate": 6.006432160804021e-05,
-      "loss": 5.6567,
-      "step": 40250
-    },
-    {
-      "epoch": 20.991395045632334,
-      "grad_norm": 1.4505279064178467,
-      "learning_rate": 6.006331658291458e-05,
-      "loss": 5.2816,
-      "step": 40251
-    },
-    {
-      "epoch": 20.991916558018254,
-      "grad_norm": 1.5252782106399536,
-      "learning_rate": 6.0062311557788956e-05,
-      "loss": 5.2111,
-      "step": 40252
-    },
-    {
-      "epoch": 20.992438070404173,
-      "grad_norm": 1.4900864362716675,
-      "learning_rate": 6.006130653266332e-05,
-      "loss": 5.5018,
-      "step": 40253
-    },
-    {
-      "epoch": 20.992959582790093,
-      "grad_norm": 1.50275719165802,
-      "learning_rate": 6.0060301507537684e-05,
-      "loss": 5.7173,
-      "step": 40254
-    },
-    {
-      "epoch": 20.993481095176012,
-      "grad_norm": 1.359324336051941,
-      "learning_rate": 6.005929648241206e-05,
-      "loss": 5.7325,
-      "step": 40255
-    },
-    {
-      "epoch": 20.994002607561928,
-      "grad_norm": 1.518571376800537,
-      "learning_rate": 6.005829145728643e-05,
-      "loss": 5.1995,
-      "step": 40256
-    },
-    {
-      "epoch": 20.994524119947847,
-      "grad_norm": 1.4319403171539307,
-      "learning_rate": 6.005728643216081e-05,
-      "loss": 5.2674,
-      "step": 40257
-    },
-    {
-      "epoch": 20.995045632333767,
-      "grad_norm": 1.4090073108673096,
-      "learning_rate": 6.0056281407035175e-05,
-      "loss": 5.3663,
-      "step": 40258
-    },
-    {
-      "epoch": 20.995567144719686,
-      "grad_norm": 1.5255705118179321,
-      "learning_rate": 6.005527638190955e-05,
-      "loss": 5.1647,
-      "step": 40259
-    },
-    {
-      "epoch": 20.996088657105606,
-      "grad_norm": 1.588913083076477,
-      "learning_rate": 6.005427135678392e-05,
-      "loss": 5.0062,
-      "step": 40260
-    },
-    {
-      "epoch": 20.996610169491525,
-      "grad_norm": 1.45968496799469,
-      "learning_rate": 6.0053266331658295e-05,
-      "loss": 4.8246,
-      "step": 40261
-    },
-    {
-      "epoch": 20.997131681877445,
-      "grad_norm": 1.59174644947052,
-      "learning_rate": 6.0052261306532666e-05,
-      "loss": 5.0051,
-      "step": 40262
-    },
-    {
-      "epoch": 20.997653194263364,
-      "grad_norm": 1.6332316398620605,
-      "learning_rate": 6.0051256281407044e-05,
-      "loss": 5.4038,
-      "step": 40263
-    },
-    {
-      "epoch": 20.998174706649284,
-      "grad_norm": 1.4970576763153076,
-      "learning_rate": 6.005025125628141e-05,
-      "loss": 5.2375,
-      "step": 40264
-    },
-    {
-      "epoch": 20.998696219035203,
-      "grad_norm": 1.5869061946868896,
-      "learning_rate": 6.0049246231155786e-05,
-      "loss": 5.453,
-      "step": 40265
-    },
-    {
-      "epoch": 20.999217731421123,
-      "grad_norm": 1.5531926155090332,
-      "learning_rate": 6.004824120603015e-05,
-      "loss": 4.8579,
-      "step": 40266
-    },
-    {
-      "epoch": 20.999739243807042,
-      "grad_norm": 1.4642184972763062,
-      "learning_rate": 6.004723618090452e-05,
-      "loss": 5.4726,
-      "step": 40267
-    },
-    {
-      "epoch": 21.000260756192958,
-      "grad_norm": 1.6982650756835938,
-      "learning_rate": 6.00462311557789e-05,
-      "loss": 5.3608,
-      "step": 40268
-    },
-    {
-      "epoch": 21.000782268578877,
-      "grad_norm": 1.4933518171310425,
-      "learning_rate": 6.0045226130653264e-05,
-      "loss": 5.3215,
-      "step": 40269
-    },
-    {
-      "epoch": 21.001303780964797,
-      "grad_norm": 1.5109684467315674,
-      "learning_rate": 6.004422110552764e-05,
-      "loss": 5.4334,
-      "step": 40270
-    },
-    {
-      "epoch": 21.001825293350716,
-      "grad_norm": 1.6125102043151855,
-      "learning_rate": 6.004321608040201e-05,
-      "loss": 5.0635,
-      "step": 40271
-    },
-    {
-      "epoch": 21.002346805736636,
-      "grad_norm": 1.5816333293914795,
-      "learning_rate": 6.004221105527639e-05,
-      "loss": 5.1659,
-      "step": 40272
-    },
-    {
-      "epoch": 21.002868318122555,
-      "grad_norm": 1.4574201107025146,
-      "learning_rate": 6.0041206030150755e-05,
-      "loss": 5.4722,
-      "step": 40273
-    },
-    {
-      "epoch": 21.003389830508475,
-      "grad_norm": 1.635692834854126,
-      "learning_rate": 6.004020100502513e-05,
-      "loss": 5.3357,
-      "step": 40274
-    },
-    {
-      "epoch": 21.003911342894394,
-      "grad_norm": 1.554926872253418,
-      "learning_rate": 6.00391959798995e-05,
-      "loss": 4.4909,
-      "step": 40275
-    },
-    {
-      "epoch": 21.004432855280314,
-      "grad_norm": 1.556113839149475,
-      "learning_rate": 6.0038190954773874e-05,
-      "loss": 5.5364,
-      "step": 40276
-    },
-    {
-      "epoch": 21.004954367666233,
-      "grad_norm": 1.5638800859451294,
-      "learning_rate": 6.0037185929648246e-05,
-      "loss": 5.4427,
-      "step": 40277
-    },
-    {
-      "epoch": 21.005475880052153,
-      "grad_norm": 1.4955930709838867,
-      "learning_rate": 6.003618090452262e-05,
-      "loss": 5.5536,
-      "step": 40278
-    },
-    {
-      "epoch": 21.005997392438072,
-      "grad_norm": 1.482633352279663,
-      "learning_rate": 6.003517587939699e-05,
-      "loss": 5.4127,
-      "step": 40279
-    },
-    {
-      "epoch": 21.006518904823988,
-      "grad_norm": 1.4531344175338745,
-      "learning_rate": 6.003417085427135e-05,
-      "loss": 5.7888,
-      "step": 40280
-    },
-    {
-      "epoch": 21.007040417209907,
-      "grad_norm": 1.6312295198440552,
-      "learning_rate": 6.003316582914573e-05,
-      "loss": 4.6033,
-      "step": 40281
-    },
-    {
-      "epoch": 21.007561929595827,
-      "grad_norm": 1.5481932163238525,
-      "learning_rate": 6.00321608040201e-05,
-      "loss": 5.3524,
-      "step": 40282
-    },
-    {
-      "epoch": 21.008083441981746,
-      "grad_norm": 1.5337562561035156,
-      "learning_rate": 6.003115577889448e-05,
-      "loss": 5.1111,
-      "step": 40283
-    },
-    {
-      "epoch": 21.008604954367666,
-      "grad_norm": 1.5550724267959595,
-      "learning_rate": 6.003015075376884e-05,
-      "loss": 5.1641,
-      "step": 40284
-    },
-    {
-      "epoch": 21.009126466753585,
-      "grad_norm": 1.4227324724197388,
-      "learning_rate": 6.002914572864322e-05,
-      "loss": 5.7504,
-      "step": 40285
-    },
-    {
-      "epoch": 21.009647979139505,
-      "grad_norm": 1.5139414072036743,
-      "learning_rate": 6.0028140703517585e-05,
-      "loss": 5.7246,
-      "step": 40286
-    },
-    {
-      "epoch": 21.010169491525424,
-      "grad_norm": 1.5168286561965942,
-      "learning_rate": 6.002713567839196e-05,
-      "loss": 5.5099,
-      "step": 40287
-    },
-    {
-      "epoch": 21.010691003911344,
-      "grad_norm": 1.5149667263031006,
-      "learning_rate": 6.0026130653266334e-05,
-      "loss": 5.1914,
-      "step": 40288
-    },
-    {
-      "epoch": 21.011212516297263,
-      "grad_norm": 1.5573232173919678,
-      "learning_rate": 6.002512562814071e-05,
-      "loss": 4.9893,
-      "step": 40289
-    },
-    {
-      "epoch": 21.011734028683183,
-      "grad_norm": 1.5841418504714966,
-      "learning_rate": 6.0024120603015076e-05,
-      "loss": 5.3715,
-      "step": 40290
-    },
-    {
-      "epoch": 21.012255541069102,
-      "grad_norm": 1.4941086769104004,
-      "learning_rate": 6.0023115577889454e-05,
-      "loss": 5.5093,
-      "step": 40291
-    },
-    {
-      "epoch": 21.012777053455018,
-      "grad_norm": 1.5442582368850708,
-      "learning_rate": 6.0022110552763825e-05,
-      "loss": 5.2348,
-      "step": 40292
-    },
-    {
-      "epoch": 21.013298565840937,
-      "grad_norm": 1.6026268005371094,
-      "learning_rate": 6.002110552763819e-05,
-      "loss": 5.1517,
-      "step": 40293
-    },
-    {
-      "epoch": 21.013820078226857,
-      "grad_norm": 1.547715663909912,
-      "learning_rate": 6.002010050251257e-05,
-      "loss": 5.3268,
-      "step": 40294
-    },
-    {
-      "epoch": 21.014341590612776,
-      "grad_norm": 1.587480068206787,
-      "learning_rate": 6.001909547738693e-05,
-      "loss": 5.2388,
-      "step": 40295
-    },
-    {
-      "epoch": 21.014863102998696,
-      "grad_norm": 1.482025146484375,
-      "learning_rate": 6.001809045226131e-05,
-      "loss": 5.2134,
-      "step": 40296
-    },
-    {
-      "epoch": 21.015384615384615,
-      "grad_norm": 1.5048632621765137,
-      "learning_rate": 6.001708542713568e-05,
-      "loss": 5.4017,
-      "step": 40297
-    },
-    {
-      "epoch": 21.015906127770535,
-      "grad_norm": 1.4276080131530762,
-      "learning_rate": 6.001608040201006e-05,
-      "loss": 5.334,
-      "step": 40298
-    },
-    {
-      "epoch": 21.016427640156454,
-      "grad_norm": 1.504310131072998,
-      "learning_rate": 6.001507537688442e-05,
-      "loss": 5.4576,
-      "step": 40299
-    },
-    {
-      "epoch": 21.016949152542374,
-      "grad_norm": 1.5477471351623535,
-      "learning_rate": 6.00140703517588e-05,
-      "loss": 5.3124,
-      "step": 40300
-    },
-    {
-      "epoch": 21.017470664928293,
-      "grad_norm": 1.5822176933288574,
-      "learning_rate": 6.0013065326633164e-05,
-      "loss": 5.3247,
-      "step": 40301
-    },
-    {
-      "epoch": 21.017992177314213,
-      "grad_norm": 1.561895728111267,
-      "learning_rate": 6.001206030150754e-05,
-      "loss": 5.1954,
-      "step": 40302
-    },
-    {
-      "epoch": 21.018513689700132,
-      "grad_norm": 1.594109296798706,
-      "learning_rate": 6.001105527638191e-05,
-      "loss": 5.2637,
-      "step": 40303
-    },
-    {
-      "epoch": 21.019035202086048,
-      "grad_norm": 1.5589921474456787,
-      "learning_rate": 6.001005025125629e-05,
-      "loss": 5.0376,
-      "step": 40304
-    },
-    {
-      "epoch": 21.019556714471967,
-      "grad_norm": 1.4877195358276367,
-      "learning_rate": 6.0009045226130655e-05,
-      "loss": 5.2987,
-      "step": 40305
-    },
-    {
-      "epoch": 21.020078226857887,
-      "grad_norm": 1.6553051471710205,
-      "learning_rate": 6.000804020100503e-05,
-      "loss": 4.8251,
-      "step": 40306
-    },
-    {
-      "epoch": 21.020599739243806,
-      "grad_norm": 1.6016215085983276,
-      "learning_rate": 6.00070351758794e-05,
-      "loss": 5.4391,
-      "step": 40307
-    },
-    {
-      "epoch": 21.021121251629726,
-      "grad_norm": 1.4515763521194458,
-      "learning_rate": 6.000603015075377e-05,
-      "loss": 5.2177,
-      "step": 40308
-    },
-    {
-      "epoch": 21.021642764015645,
-      "grad_norm": 1.7168132066726685,
-      "learning_rate": 6.0005025125628146e-05,
-      "loss": 4.6944,
-      "step": 40309
-    },
-    {
-      "epoch": 21.022164276401565,
-      "grad_norm": 1.453839659690857,
-      "learning_rate": 6.000402010050251e-05,
-      "loss": 4.7506,
-      "step": 40310
-    },
-    {
-      "epoch": 21.022685788787484,
-      "grad_norm": 1.4824360609054565,
-      "learning_rate": 6.000301507537689e-05,
-      "loss": 4.8908,
-      "step": 40311
-    },
-    {
-      "epoch": 21.023207301173404,
-      "grad_norm": 1.4710594415664673,
-      "learning_rate": 6.000201005025126e-05,
-      "loss": 5.3352,
-      "step": 40312
-    },
-    {
-      "epoch": 21.023728813559323,
-      "grad_norm": 1.490584135055542,
-      "learning_rate": 6.000100502512564e-05,
-      "loss": 4.9869,
-      "step": 40313
-    },
-    {
-      "epoch": 21.024250325945243,
-      "grad_norm": 1.477036714553833,
-      "learning_rate": 6e-05,
-      "loss": 5.0558,
-      "step": 40314
-    },
-    {
-      "epoch": 21.02477183833116,
-      "grad_norm": 1.438130259513855,
-      "learning_rate": 5.999899497487438e-05,
-      "loss": 5.3332,
-      "step": 40315
-    },
-    {
-      "epoch": 21.025293350717078,
-      "grad_norm": 1.4845020771026611,
-      "learning_rate": 5.9997989949748744e-05,
-      "loss": 5.0556,
-      "step": 40316
-    },
-    {
-      "epoch": 21.025814863102998,
-      "grad_norm": 1.4898744821548462,
-      "learning_rate": 5.999698492462312e-05,
-      "loss": 4.9486,
-      "step": 40317
-    },
-    {
-      "epoch": 21.026336375488917,
-      "grad_norm": 1.5747170448303223,
-      "learning_rate": 5.999597989949749e-05,
-      "loss": 4.8809,
-      "step": 40318
-    },
-    {
-      "epoch": 21.026857887874836,
-      "grad_norm": 1.4557090997695923,
-      "learning_rate": 5.999497487437187e-05,
-      "loss": 5.5055,
-      "step": 40319
-    },
-    {
-      "epoch": 21.027379400260756,
-      "grad_norm": 1.5053032636642456,
-      "learning_rate": 5.9993969849246235e-05,
-      "loss": 5.508,
-      "step": 40320
-    },
-    {
-      "epoch": 21.027900912646675,
-      "grad_norm": 1.5862587690353394,
-      "learning_rate": 5.99929648241206e-05,
-      "loss": 5.3276,
-      "step": 40321
-    },
-    {
-      "epoch": 21.028422425032595,
-      "grad_norm": 1.4656000137329102,
-      "learning_rate": 5.999195979899498e-05,
-      "loss": 5.8587,
-      "step": 40322
-    },
-    {
-      "epoch": 21.028943937418514,
-      "grad_norm": 1.5410146713256836,
-      "learning_rate": 5.999095477386935e-05,
-      "loss": 5.2779,
-      "step": 40323
-    },
-    {
-      "epoch": 21.029465449804434,
-      "grad_norm": 1.547837257385254,
-      "learning_rate": 5.9989949748743726e-05,
-      "loss": 5.1354,
-      "step": 40324
-    },
-    {
-      "epoch": 21.029986962190353,
-      "grad_norm": 1.4998126029968262,
-      "learning_rate": 5.998894472361809e-05,
-      "loss": 5.5004,
-      "step": 40325
-    },
-    {
-      "epoch": 21.030508474576273,
-      "grad_norm": 1.572510838508606,
-      "learning_rate": 5.998793969849247e-05,
-      "loss": 5.2196,
-      "step": 40326
-    },
-    {
-      "epoch": 21.03102998696219,
-      "grad_norm": 1.5417941808700562,
-      "learning_rate": 5.998693467336683e-05,
-      "loss": 5.5647,
-      "step": 40327
-    },
-    {
-      "epoch": 21.031551499348108,
-      "grad_norm": 1.535940170288086,
-      "learning_rate": 5.998592964824121e-05,
-      "loss": 5.064,
-      "step": 40328
-    },
-    {
-      "epoch": 21.032073011734028,
-      "grad_norm": 1.4961856603622437,
-      "learning_rate": 5.998492462311558e-05,
-      "loss": 5.323,
-      "step": 40329
-    },
-    {
-      "epoch": 21.032594524119947,
-      "grad_norm": 1.4591492414474487,
-      "learning_rate": 5.998391959798996e-05,
-      "loss": 5.6348,
-      "step": 40330
-    },
-    {
-      "epoch": 21.033116036505866,
-      "grad_norm": 1.413661241531372,
-      "learning_rate": 5.998291457286432e-05,
-      "loss": 5.5023,
-      "step": 40331
-    },
-    {
-      "epoch": 21.033637548891786,
-      "grad_norm": 1.5521583557128906,
-      "learning_rate": 5.99819095477387e-05,
-      "loss": 5.208,
-      "step": 40332
-    },
-    {
-      "epoch": 21.034159061277705,
-      "grad_norm": 1.5993579626083374,
-      "learning_rate": 5.9980904522613065e-05,
-      "loss": 5.2177,
-      "step": 40333
-    },
-    {
-      "epoch": 21.034680573663625,
-      "grad_norm": 1.5065300464630127,
-      "learning_rate": 5.9979899497487436e-05,
-      "loss": 5.1744,
-      "step": 40334
-    },
-    {
-      "epoch": 21.035202086049544,
-      "grad_norm": 1.5574893951416016,
-      "learning_rate": 5.9978894472361814e-05,
-      "loss": 4.7886,
-      "step": 40335
-    },
-    {
-      "epoch": 21.035723598435464,
-      "grad_norm": 1.5106351375579834,
-      "learning_rate": 5.997788944723618e-05,
-      "loss": 4.9583,
-      "step": 40336
-    },
-    {
-      "epoch": 21.036245110821383,
-      "grad_norm": 1.5090610980987549,
-      "learning_rate": 5.9976884422110556e-05,
-      "loss": 5.4181,
-      "step": 40337
-    },
-    {
-      "epoch": 21.036766623207303,
-      "grad_norm": 1.668508529663086,
-      "learning_rate": 5.997587939698493e-05,
-      "loss": 5.013,
-      "step": 40338
-    },
-    {
-      "epoch": 21.03728813559322,
-      "grad_norm": 1.5508103370666504,
-      "learning_rate": 5.9974874371859305e-05,
-      "loss": 4.9643,
-      "step": 40339
-    },
-    {
-      "epoch": 21.037809647979138,
-      "grad_norm": 1.6395087242126465,
-      "learning_rate": 5.997386934673367e-05,
-      "loss": 4.7375,
-      "step": 40340
-    },
-    {
-      "epoch": 21.038331160365058,
-      "grad_norm": 1.4180349111557007,
-      "learning_rate": 5.997286432160805e-05,
-      "loss": 4.5332,
-      "step": 40341
-    },
-    {
-      "epoch": 21.038852672750977,
-      "grad_norm": 1.4339677095413208,
-      "learning_rate": 5.997185929648241e-05,
-      "loss": 5.5386,
-      "step": 40342
-    },
-    {
-      "epoch": 21.039374185136897,
-      "grad_norm": 1.3930926322937012,
-      "learning_rate": 5.997085427135679e-05,
-      "loss": 5.6717,
-      "step": 40343
-    },
-    {
-      "epoch": 21.039895697522816,
-      "grad_norm": 1.4621506929397583,
-      "learning_rate": 5.996984924623116e-05,
-      "loss": 5.3347,
-      "step": 40344
-    },
-    {
-      "epoch": 21.040417209908735,
-      "grad_norm": 1.5797785520553589,
-      "learning_rate": 5.996884422110554e-05,
-      "loss": 5.4533,
-      "step": 40345
-    },
-    {
-      "epoch": 21.040938722294655,
-      "grad_norm": 1.4994438886642456,
-      "learning_rate": 5.99678391959799e-05,
-      "loss": 5.3379,
-      "step": 40346
-    },
-    {
-      "epoch": 21.041460234680574,
-      "grad_norm": 1.4864157438278198,
-      "learning_rate": 5.9966834170854267e-05,
-      "loss": 5.2781,
-      "step": 40347
-    },
-    {
-      "epoch": 21.041981747066494,
-      "grad_norm": 1.4713951349258423,
-      "learning_rate": 5.9965829145728644e-05,
-      "loss": 5.1767,
-      "step": 40348
-    },
-    {
-      "epoch": 21.042503259452413,
-      "grad_norm": 1.4549949169158936,
-      "learning_rate": 5.9964824120603015e-05,
-      "loss": 5.1135,
-      "step": 40349
-    },
-    {
-      "epoch": 21.043024771838333,
-      "grad_norm": 1.5170758962631226,
-      "learning_rate": 5.996381909547739e-05,
-      "loss": 5.5312,
-      "step": 40350
-    },
-    {
-      "epoch": 21.04354628422425,
-      "grad_norm": 1.5293550491333008,
-      "learning_rate": 5.996281407035176e-05,
-      "loss": 4.8149,
-      "step": 40351
-    },
-    {
-      "epoch": 21.044067796610168,
-      "grad_norm": 1.6020863056182861,
-      "learning_rate": 5.9961809045226135e-05,
-      "loss": 5.4051,
-      "step": 40352
-    },
-    {
-      "epoch": 21.044589308996088,
-      "grad_norm": 1.5399811267852783,
-      "learning_rate": 5.99608040201005e-05,
-      "loss": 5.0463,
-      "step": 40353
-    },
-    {
-      "epoch": 21.045110821382007,
-      "grad_norm": 1.4660260677337646,
-      "learning_rate": 5.995979899497488e-05,
-      "loss": 5.3122,
-      "step": 40354
-    },
-    {
-      "epoch": 21.045632333767927,
-      "grad_norm": 1.5565156936645508,
-      "learning_rate": 5.995879396984925e-05,
-      "loss": 4.9425,
-      "step": 40355
-    },
-    {
-      "epoch": 21.046153846153846,
-      "grad_norm": 1.5840297937393188,
-      "learning_rate": 5.9957788944723626e-05,
-      "loss": 5.2442,
-      "step": 40356
-    },
-    {
-      "epoch": 21.046675358539765,
-      "grad_norm": 1.4347575902938843,
-      "learning_rate": 5.995678391959799e-05,
-      "loss": 5.764,
-      "step": 40357
-    },
-    {
-      "epoch": 21.047196870925685,
-      "grad_norm": 1.4967448711395264,
-      "learning_rate": 5.995577889447237e-05,
-      "loss": 5.4049,
-      "step": 40358
-    },
-    {
-      "epoch": 21.047718383311604,
-      "grad_norm": 1.7114477157592773,
-      "learning_rate": 5.995477386934674e-05,
-      "loss": 5.2523,
-      "step": 40359
-    },
-    {
-      "epoch": 21.048239895697524,
-      "grad_norm": 1.5145765542984009,
-      "learning_rate": 5.9953768844221104e-05,
-      "loss": 5.3668,
-      "step": 40360
-    },
-    {
-      "epoch": 21.048761408083443,
-      "grad_norm": 1.6865406036376953,
-      "learning_rate": 5.995276381909548e-05,
-      "loss": 4.7117,
-      "step": 40361
-    },
-    {
-      "epoch": 21.049282920469363,
-      "grad_norm": 1.5630327463150024,
-      "learning_rate": 5.9951758793969846e-05,
-      "loss": 5.0823,
-      "step": 40362
-    },
-    {
-      "epoch": 21.04980443285528,
-      "grad_norm": 1.5027399063110352,
-      "learning_rate": 5.9950753768844224e-05,
-      "loss": 5.0845,
-      "step": 40363
-    },
-    {
-      "epoch": 21.050325945241198,
-      "grad_norm": 1.4975191354751587,
-      "learning_rate": 5.9949748743718595e-05,
-      "loss": 5.6971,
-      "step": 40364
-    },
-    {
-      "epoch": 21.050847457627118,
-      "grad_norm": 1.5459964275360107,
-      "learning_rate": 5.994874371859297e-05,
-      "loss": 5.1648,
-      "step": 40365
-    },
-    {
-      "epoch": 21.051368970013037,
-      "grad_norm": 1.6261237859725952,
-      "learning_rate": 5.994773869346734e-05,
-      "loss": 5.1223,
-      "step": 40366
-    },
-    {
-      "epoch": 21.051890482398957,
-      "grad_norm": 1.5345220565795898,
-      "learning_rate": 5.9946733668341715e-05,
-      "loss": 5.3707,
-      "step": 40367
-    },
-    {
-      "epoch": 21.052411994784876,
-      "grad_norm": 1.645136833190918,
-      "learning_rate": 5.994572864321608e-05,
-      "loss": 4.6298,
-      "step": 40368
-    },
-    {
-      "epoch": 21.052933507170795,
-      "grad_norm": 1.6052868366241455,
-      "learning_rate": 5.994472361809046e-05,
-      "loss": 4.5733,
-      "step": 40369
-    },
-    {
-      "epoch": 21.053455019556715,
-      "grad_norm": 1.4566707611083984,
-      "learning_rate": 5.994371859296483e-05,
-      "loss": 5.5273,
-      "step": 40370
-    },
-    {
-      "epoch": 21.053976531942634,
-      "grad_norm": 1.6116071939468384,
-      "learning_rate": 5.9942713567839206e-05,
-      "loss": 4.9594,
-      "step": 40371
-    },
-    {
-      "epoch": 21.054498044328554,
-      "grad_norm": 1.4607092142105103,
-      "learning_rate": 5.994170854271357e-05,
-      "loss": 5.3639,
-      "step": 40372
-    },
-    {
-      "epoch": 21.055019556714473,
-      "grad_norm": 1.4738078117370605,
-      "learning_rate": 5.9940703517587934e-05,
-      "loss": 5.3527,
-      "step": 40373
-    },
-    {
-      "epoch": 21.055541069100393,
-      "grad_norm": 1.5473918914794922,
-      "learning_rate": 5.993969849246231e-05,
-      "loss": 5.5514,
-      "step": 40374
-    },
-    {
-      "epoch": 21.05606258148631,
-      "grad_norm": 1.5012526512145996,
-      "learning_rate": 5.993869346733668e-05,
-      "loss": 5.1012,
-      "step": 40375
-    },
-    {
-      "epoch": 21.056584093872228,
-      "grad_norm": 1.4310847520828247,
-      "learning_rate": 5.993768844221106e-05,
-      "loss": 5.6591,
-      "step": 40376
-    },
-    {
-      "epoch": 21.057105606258148,
-      "grad_norm": 1.6541026830673218,
-      "learning_rate": 5.9936683417085425e-05,
-      "loss": 5.0254,
-      "step": 40377
-    },
-    {
-      "epoch": 21.057627118644067,
-      "grad_norm": 1.4835400581359863,
-      "learning_rate": 5.99356783919598e-05,
-      "loss": 5.4378,
-      "step": 40378
-    },
-    {
-      "epoch": 21.058148631029987,
-      "grad_norm": 1.5869765281677246,
-      "learning_rate": 5.9934673366834174e-05,
-      "loss": 5.2637,
-      "step": 40379
-    },
-    {
-      "epoch": 21.058670143415906,
-      "grad_norm": 1.5224345922470093,
-      "learning_rate": 5.993366834170855e-05,
-      "loss": 5.3463,
-      "step": 40380
-    },
-    {
-      "epoch": 21.059191655801826,
-      "grad_norm": 1.5414509773254395,
-      "learning_rate": 5.9932663316582916e-05,
-      "loss": 4.9937,
-      "step": 40381
-    },
-    {
-      "epoch": 21.059713168187745,
-      "grad_norm": 1.5173695087432861,
-      "learning_rate": 5.9931658291457294e-05,
-      "loss": 5.2013,
-      "step": 40382
-    },
-    {
-      "epoch": 21.060234680573664,
-      "grad_norm": 1.549431324005127,
-      "learning_rate": 5.993065326633166e-05,
-      "loss": 4.7318,
-      "step": 40383
-    },
-    {
-      "epoch": 21.060756192959584,
-      "grad_norm": 1.4791259765625,
-      "learning_rate": 5.9929648241206036e-05,
-      "loss": 5.2644,
-      "step": 40384
-    },
-    {
-      "epoch": 21.061277705345503,
-      "grad_norm": 1.5860217809677124,
-      "learning_rate": 5.992864321608041e-05,
-      "loss": 5.5841,
-      "step": 40385
-    },
-    {
-      "epoch": 21.061799217731423,
-      "grad_norm": 1.5154565572738647,
-      "learning_rate": 5.992763819095477e-05,
-      "loss": 5.2084,
-      "step": 40386
-    },
-    {
-      "epoch": 21.06232073011734,
-      "grad_norm": 1.5342824459075928,
-      "learning_rate": 5.992663316582915e-05,
-      "loss": 4.9991,
-      "step": 40387
-    },
-    {
-      "epoch": 21.062842242503258,
-      "grad_norm": 1.4464399814605713,
-      "learning_rate": 5.9925628140703514e-05,
-      "loss": 5.5001,
-      "step": 40388
-    },
-    {
-      "epoch": 21.063363754889178,
-      "grad_norm": 1.519986867904663,
-      "learning_rate": 5.992462311557789e-05,
-      "loss": 4.7058,
-      "step": 40389
-    },
-    {
-      "epoch": 21.063885267275097,
-      "grad_norm": 1.4727811813354492,
-      "learning_rate": 5.992361809045226e-05,
-      "loss": 5.2539,
-      "step": 40390
-    },
-    {
-      "epoch": 21.064406779661017,
-      "grad_norm": 1.5164682865142822,
-      "learning_rate": 5.992261306532664e-05,
-      "loss": 5.1668,
-      "step": 40391
-    },
-    {
-      "epoch": 21.064928292046936,
-      "grad_norm": 1.5629678964614868,
-      "learning_rate": 5.9921608040201005e-05,
-      "loss": 5.4169,
-      "step": 40392
-    },
-    {
-      "epoch": 21.065449804432856,
-      "grad_norm": 1.4999115467071533,
-      "learning_rate": 5.992060301507538e-05,
-      "loss": 5.0804,
-      "step": 40393
-    },
-    {
-      "epoch": 21.065971316818775,
-      "grad_norm": 1.4797821044921875,
-      "learning_rate": 5.991959798994975e-05,
-      "loss": 5.5736,
-      "step": 40394
-    },
-    {
-      "epoch": 21.066492829204694,
-      "grad_norm": 1.550120234489441,
-      "learning_rate": 5.9918592964824124e-05,
-      "loss": 5.4076,
-      "step": 40395
-    },
-    {
-      "epoch": 21.067014341590614,
-      "grad_norm": 1.390349268913269,
-      "learning_rate": 5.9917587939698496e-05,
-      "loss": 5.3583,
-      "step": 40396
-    },
-    {
-      "epoch": 21.067535853976533,
-      "grad_norm": 1.5997527837753296,
-      "learning_rate": 5.991658291457287e-05,
-      "loss": 5.1002,
-      "step": 40397
-    },
-    {
-      "epoch": 21.068057366362453,
-      "grad_norm": 1.5309362411499023,
-      "learning_rate": 5.991557788944724e-05,
-      "loss": 5.4891,
-      "step": 40398
-    },
-    {
-      "epoch": 21.06857887874837,
-      "grad_norm": 1.5490130186080933,
-      "learning_rate": 5.9914572864321615e-05,
-      "loss": 5.1167,
-      "step": 40399
-    },
-    {
-      "epoch": 21.06910039113429,
-      "grad_norm": 1.6302458047866821,
-      "learning_rate": 5.9913567839195987e-05,
-      "loss": 5.3459,
-      "step": 40400
-    },
-    {
-      "epoch": 21.069621903520208,
-      "grad_norm": 1.5515536069869995,
-      "learning_rate": 5.991256281407035e-05,
-      "loss": 5.4387,
-      "step": 40401
-    },
-    {
-      "epoch": 21.070143415906127,
-      "grad_norm": 1.499293565750122,
-      "learning_rate": 5.991155778894473e-05,
-      "loss": 5.2539,
-      "step": 40402
-    },
-    {
-      "epoch": 21.070664928292047,
-      "grad_norm": 1.5099488496780396,
-      "learning_rate": 5.991055276381909e-05,
-      "loss": 5.2149,
-      "step": 40403
-    },
-    {
-      "epoch": 21.071186440677966,
-      "grad_norm": 1.4873926639556885,
-      "learning_rate": 5.990954773869347e-05,
-      "loss": 5.1351,
-      "step": 40404
-    },
-    {
-      "epoch": 21.071707953063886,
-      "grad_norm": 1.5114779472351074,
-      "learning_rate": 5.990854271356784e-05,
-      "loss": 5.2743,
-      "step": 40405
-    },
-    {
-      "epoch": 21.072229465449805,
-      "grad_norm": 1.443228840827942,
-      "learning_rate": 5.990753768844222e-05,
-      "loss": 5.7163,
-      "step": 40406
-    },
-    {
-      "epoch": 21.072750977835724,
-      "grad_norm": 1.435282826423645,
-      "learning_rate": 5.9906532663316584e-05,
-      "loss": 5.6784,
-      "step": 40407
-    },
-    {
-      "epoch": 21.073272490221644,
-      "grad_norm": 1.4752062559127808,
-      "learning_rate": 5.990552763819096e-05,
-      "loss": 5.2758,
-      "step": 40408
-    },
-    {
-      "epoch": 21.073794002607563,
-      "grad_norm": 1.5468318462371826,
-      "learning_rate": 5.9904522613065326e-05,
-      "loss": 5.4254,
-      "step": 40409
-    },
-    {
-      "epoch": 21.07431551499348,
-      "grad_norm": 1.4777799844741821,
-      "learning_rate": 5.9903517587939704e-05,
-      "loss": 5.0548,
-      "step": 40410
-    },
-    {
-      "epoch": 21.0748370273794,
-      "grad_norm": 1.5729069709777832,
-      "learning_rate": 5.9902512562814075e-05,
-      "loss": 5.4661,
-      "step": 40411
-    },
-    {
-      "epoch": 21.07535853976532,
-      "grad_norm": 1.5045760869979858,
-      "learning_rate": 5.990150753768845e-05,
-      "loss": 5.2886,
-      "step": 40412
-    },
-    {
-      "epoch": 21.075880052151238,
-      "grad_norm": 1.5258417129516602,
-      "learning_rate": 5.990050251256282e-05,
-      "loss": 5.3664,
-      "step": 40413
-    },
-    {
-      "epoch": 21.076401564537157,
-      "grad_norm": 1.539175271987915,
-      "learning_rate": 5.989949748743718e-05,
-      "loss": 5.2761,
-      "step": 40414
-    },
-    {
-      "epoch": 21.076923076923077,
-      "grad_norm": 1.436463475227356,
-      "learning_rate": 5.989849246231156e-05,
-      "loss": 5.2673,
-      "step": 40415
-    },
-    {
-      "epoch": 21.077444589308996,
-      "grad_norm": 1.5854564905166626,
-      "learning_rate": 5.989748743718593e-05,
-      "loss": 5.0315,
-      "step": 40416
-    },
-    {
-      "epoch": 21.077966101694916,
-      "grad_norm": 1.5790973901748657,
-      "learning_rate": 5.989648241206031e-05,
-      "loss": 5.1532,
-      "step": 40417
-    },
-    {
-      "epoch": 21.078487614080835,
-      "grad_norm": 1.5370365381240845,
-      "learning_rate": 5.989547738693467e-05,
-      "loss": 5.1437,
-      "step": 40418
-    },
-    {
-      "epoch": 21.079009126466755,
-      "grad_norm": 1.5858910083770752,
-      "learning_rate": 5.989447236180905e-05,
-      "loss": 5.1078,
-      "step": 40419
-    },
-    {
-      "epoch": 21.079530638852674,
-      "grad_norm": 1.5030497312545776,
-      "learning_rate": 5.9893467336683414e-05,
-      "loss": 5.1256,
-      "step": 40420
-    },
-    {
-      "epoch": 21.080052151238593,
-      "grad_norm": 1.722938060760498,
-      "learning_rate": 5.989246231155779e-05,
-      "loss": 5.1341,
-      "step": 40421
-    },
-    {
-      "epoch": 21.08057366362451,
-      "grad_norm": 1.5431077480316162,
-      "learning_rate": 5.989145728643216e-05,
-      "loss": 5.3995,
-      "step": 40422
-    },
-    {
-      "epoch": 21.08109517601043,
-      "grad_norm": 1.5121657848358154,
-      "learning_rate": 5.989045226130654e-05,
-      "loss": 5.4958,
-      "step": 40423
-    },
-    {
-      "epoch": 21.08161668839635,
-      "grad_norm": 1.6048474311828613,
-      "learning_rate": 5.9889447236180905e-05,
-      "loss": 5.5061,
-      "step": 40424
-    },
-    {
-      "epoch": 21.082138200782268,
-      "grad_norm": 1.537729024887085,
-      "learning_rate": 5.988844221105528e-05,
-      "loss": 4.9886,
-      "step": 40425
-    },
-    {
-      "epoch": 21.082659713168187,
-      "grad_norm": 1.6699798107147217,
-      "learning_rate": 5.9887437185929654e-05,
-      "loss": 5.0797,
-      "step": 40426
-    },
-    {
-      "epoch": 21.083181225554107,
-      "grad_norm": 1.454174280166626,
-      "learning_rate": 5.988643216080402e-05,
-      "loss": 5.0666,
-      "step": 40427
-    },
-    {
-      "epoch": 21.083702737940026,
-      "grad_norm": 1.600732684135437,
-      "learning_rate": 5.9885427135678396e-05,
-      "loss": 5.1711,
-      "step": 40428
-    },
-    {
-      "epoch": 21.084224250325946,
-      "grad_norm": 1.4605021476745605,
-      "learning_rate": 5.988442211055276e-05,
-      "loss": 5.4381,
-      "step": 40429
-    },
-    {
-      "epoch": 21.084745762711865,
-      "grad_norm": 1.3942335844039917,
-      "learning_rate": 5.988341708542714e-05,
-      "loss": 5.7274,
-      "step": 40430
-    },
-    {
-      "epoch": 21.085267275097785,
-      "grad_norm": 1.6235052347183228,
-      "learning_rate": 5.988241206030151e-05,
-      "loss": 4.9994,
-      "step": 40431
-    },
-    {
-      "epoch": 21.085788787483704,
-      "grad_norm": 1.50924551486969,
-      "learning_rate": 5.988140703517589e-05,
-      "loss": 5.5725,
-      "step": 40432
-    },
-    {
-      "epoch": 21.086310299869623,
-      "grad_norm": 1.4373935461044312,
-      "learning_rate": 5.988040201005025e-05,
-      "loss": 5.3389,
-      "step": 40433
-    },
-    {
-      "epoch": 21.08683181225554,
-      "grad_norm": 1.537576675415039,
-      "learning_rate": 5.987939698492463e-05,
-      "loss": 4.7413,
-      "step": 40434
-    },
-    {
-      "epoch": 21.08735332464146,
-      "grad_norm": 1.4725618362426758,
-      "learning_rate": 5.9878391959798994e-05,
-      "loss": 5.555,
-      "step": 40435
-    },
-    {
-      "epoch": 21.08787483702738,
-      "grad_norm": 1.5339632034301758,
-      "learning_rate": 5.987738693467337e-05,
-      "loss": 5.0229,
-      "step": 40436
-    },
-    {
-      "epoch": 21.088396349413298,
-      "grad_norm": 1.5626211166381836,
-      "learning_rate": 5.987638190954774e-05,
-      "loss": 5.4258,
-      "step": 40437
-    },
-    {
-      "epoch": 21.088917861799217,
-      "grad_norm": 1.5049450397491455,
-      "learning_rate": 5.987537688442212e-05,
-      "loss": 5.2395,
-      "step": 40438
-    },
-    {
-      "epoch": 21.089439374185137,
-      "grad_norm": 1.7836427688598633,
-      "learning_rate": 5.9874371859296485e-05,
-      "loss": 4.8023,
-      "step": 40439
-    },
-    {
-      "epoch": 21.089960886571056,
-      "grad_norm": 1.5529954433441162,
-      "learning_rate": 5.987336683417085e-05,
-      "loss": 5.3701,
-      "step": 40440
-    },
-    {
-      "epoch": 21.090482398956976,
-      "grad_norm": 1.4089865684509277,
-      "learning_rate": 5.987236180904523e-05,
-      "loss": 5.3612,
-      "step": 40441
-    },
-    {
-      "epoch": 21.091003911342895,
-      "grad_norm": 1.4377624988555908,
-      "learning_rate": 5.98713567839196e-05,
-      "loss": 5.4074,
-      "step": 40442
-    },
-    {
-      "epoch": 21.091525423728815,
-      "grad_norm": 1.4648789167404175,
-      "learning_rate": 5.9870351758793976e-05,
-      "loss": 5.3629,
-      "step": 40443
-    },
-    {
-      "epoch": 21.092046936114734,
-      "grad_norm": 1.5598760843276978,
-      "learning_rate": 5.986934673366834e-05,
-      "loss": 5.3171,
-      "step": 40444
-    },
-    {
-      "epoch": 21.092568448500653,
-      "grad_norm": 1.4527603387832642,
-      "learning_rate": 5.986834170854272e-05,
-      "loss": 5.2719,
-      "step": 40445
-    },
-    {
-      "epoch": 21.09308996088657,
-      "grad_norm": 1.5831148624420166,
-      "learning_rate": 5.986733668341709e-05,
-      "loss": 5.2809,
-      "step": 40446
-    },
-    {
-      "epoch": 21.09361147327249,
-      "grad_norm": 1.6403896808624268,
-      "learning_rate": 5.9866331658291467e-05,
-      "loss": 4.6864,
-      "step": 40447
-    },
-    {
-      "epoch": 21.09413298565841,
-      "grad_norm": 1.5236560106277466,
-      "learning_rate": 5.986532663316583e-05,
-      "loss": 5.6539,
-      "step": 40448
-    },
-    {
-      "epoch": 21.094654498044328,
-      "grad_norm": 1.5617449283599854,
-      "learning_rate": 5.986432160804021e-05,
-      "loss": 5.5145,
-      "step": 40449
-    },
-    {
-      "epoch": 21.095176010430247,
-      "grad_norm": 1.4581414461135864,
-      "learning_rate": 5.986331658291457e-05,
-      "loss": 5.2671,
-      "step": 40450
-    },
-    {
-      "epoch": 21.095697522816167,
-      "grad_norm": 1.586154580116272,
-      "learning_rate": 5.986231155778895e-05,
-      "loss": 5.3387,
-      "step": 40451
-    },
-    {
-      "epoch": 21.096219035202086,
-      "grad_norm": 1.5706770420074463,
-      "learning_rate": 5.986130653266332e-05,
-      "loss": 5.0477,
-      "step": 40452
-    },
-    {
-      "epoch": 21.096740547588006,
-      "grad_norm": 1.4694738388061523,
-      "learning_rate": 5.9860301507537686e-05,
-      "loss": 5.4925,
-      "step": 40453
-    },
-    {
-      "epoch": 21.097262059973925,
-      "grad_norm": 1.4194875955581665,
-      "learning_rate": 5.9859296482412064e-05,
-      "loss": 4.8194,
-      "step": 40454
-    },
-    {
-      "epoch": 21.097783572359845,
-      "grad_norm": 1.522809386253357,
-      "learning_rate": 5.985829145728643e-05,
-      "loss": 5.121,
-      "step": 40455
-    },
-    {
-      "epoch": 21.098305084745764,
-      "grad_norm": 1.4670060873031616,
-      "learning_rate": 5.9857286432160806e-05,
-      "loss": 5.4332,
-      "step": 40456
-    },
-    {
-      "epoch": 21.098826597131684,
-      "grad_norm": 1.4306670427322388,
-      "learning_rate": 5.985628140703518e-05,
-      "loss": 5.2564,
-      "step": 40457
-    },
-    {
-      "epoch": 21.0993481095176,
-      "grad_norm": 1.5054138898849487,
-      "learning_rate": 5.9855276381909555e-05,
-      "loss": 5.3078,
-      "step": 40458
-    },
-    {
-      "epoch": 21.09986962190352,
-      "grad_norm": 1.5119010210037231,
-      "learning_rate": 5.985427135678392e-05,
-      "loss": 5.1843,
-      "step": 40459
-    },
-    {
-      "epoch": 21.10039113428944,
-      "grad_norm": 1.596413016319275,
-      "learning_rate": 5.98532663316583e-05,
-      "loss": 4.879,
-      "step": 40460
-    },
-    {
-      "epoch": 21.100912646675358,
-      "grad_norm": 1.6137243509292603,
-      "learning_rate": 5.985226130653266e-05,
-      "loss": 5.172,
-      "step": 40461
-    },
-    {
-      "epoch": 21.101434159061277,
-      "grad_norm": 1.4975477457046509,
-      "learning_rate": 5.985125628140704e-05,
-      "loss": 5.2918,
-      "step": 40462
-    },
-    {
-      "epoch": 21.101955671447197,
-      "grad_norm": 1.5288376808166504,
-      "learning_rate": 5.985025125628141e-05,
-      "loss": 5.2361,
-      "step": 40463
-    },
-    {
-      "epoch": 21.102477183833116,
-      "grad_norm": 1.6010289192199707,
-      "learning_rate": 5.984924623115579e-05,
-      "loss": 5.0082,
-      "step": 40464
-    },
-    {
-      "epoch": 21.102998696219036,
-      "grad_norm": 1.6673272848129272,
-      "learning_rate": 5.984824120603015e-05,
-      "loss": 5.4964,
-      "step": 40465
-    },
-    {
-      "epoch": 21.103520208604955,
-      "grad_norm": 1.6450858116149902,
-      "learning_rate": 5.984723618090452e-05,
-      "loss": 5.022,
-      "step": 40466
-    },
-    {
-      "epoch": 21.104041720990875,
-      "grad_norm": 1.566383719444275,
-      "learning_rate": 5.98462311557789e-05,
-      "loss": 5.6457,
-      "step": 40467
-    },
-    {
-      "epoch": 21.104563233376794,
-      "grad_norm": 1.6621265411376953,
-      "learning_rate": 5.9845226130653265e-05,
-      "loss": 5.1344,
-      "step": 40468
-    },
-    {
-      "epoch": 21.105084745762714,
-      "grad_norm": 1.4125103950500488,
-      "learning_rate": 5.984422110552764e-05,
-      "loss": 5.1382,
-      "step": 40469
-    },
-    {
-      "epoch": 21.10560625814863,
-      "grad_norm": 1.476352572441101,
-      "learning_rate": 5.984321608040201e-05,
-      "loss": 5.0588,
-      "step": 40470
-    },
-    {
-      "epoch": 21.10612777053455,
-      "grad_norm": 1.5042837858200073,
-      "learning_rate": 5.9842211055276385e-05,
-      "loss": 5.4934,
-      "step": 40471
-    },
-    {
-      "epoch": 21.10664928292047,
-      "grad_norm": 1.451910376548767,
-      "learning_rate": 5.9841206030150756e-05,
-      "loss": 5.1945,
-      "step": 40472
-    },
-    {
-      "epoch": 21.107170795306388,
-      "grad_norm": 1.6424407958984375,
-      "learning_rate": 5.9840201005025134e-05,
-      "loss": 5.3011,
-      "step": 40473
-    },
-    {
-      "epoch": 21.107692307692307,
-      "grad_norm": 1.587241530418396,
-      "learning_rate": 5.98391959798995e-05,
-      "loss": 5.1846,
-      "step": 40474
-    },
-    {
-      "epoch": 21.108213820078227,
-      "grad_norm": 1.4831129312515259,
-      "learning_rate": 5.9838190954773876e-05,
-      "loss": 5.4498,
-      "step": 40475
-    },
-    {
-      "epoch": 21.108735332464146,
-      "grad_norm": 1.5142009258270264,
-      "learning_rate": 5.983718592964824e-05,
-      "loss": 5.1398,
-      "step": 40476
-    },
-    {
-      "epoch": 21.109256844850066,
-      "grad_norm": 1.4450006484985352,
-      "learning_rate": 5.983618090452262e-05,
-      "loss": 5.6326,
-      "step": 40477
-    },
-    {
-      "epoch": 21.109778357235985,
-      "grad_norm": 1.4935756921768188,
-      "learning_rate": 5.983517587939699e-05,
-      "loss": 5.6283,
-      "step": 40478
-    },
-    {
-      "epoch": 21.110299869621905,
-      "grad_norm": 1.6060832738876343,
-      "learning_rate": 5.9834170854271354e-05,
-      "loss": 5.2143,
-      "step": 40479
-    },
-    {
-      "epoch": 21.110821382007824,
-      "grad_norm": 1.5077277421951294,
-      "learning_rate": 5.983316582914573e-05,
-      "loss": 5.3397,
-      "step": 40480
-    },
-    {
-      "epoch": 21.111342894393744,
-      "grad_norm": 1.5558044910430908,
-      "learning_rate": 5.9832160804020096e-05,
-      "loss": 5.5193,
-      "step": 40481
-    },
-    {
-      "epoch": 21.11186440677966,
-      "grad_norm": 1.5218029022216797,
-      "learning_rate": 5.9831155778894474e-05,
-      "loss": 4.7618,
-      "step": 40482
-    },
-    {
-      "epoch": 21.11238591916558,
-      "grad_norm": 1.5532203912734985,
-      "learning_rate": 5.9830150753768845e-05,
-      "loss": 5.1237,
-      "step": 40483
-    },
-    {
-      "epoch": 21.1129074315515,
-      "grad_norm": 1.5586994886398315,
-      "learning_rate": 5.982914572864322e-05,
-      "loss": 5.4746,
-      "step": 40484
-    },
-    {
-      "epoch": 21.113428943937418,
-      "grad_norm": 1.4544874429702759,
-      "learning_rate": 5.982814070351759e-05,
-      "loss": 4.9114,
-      "step": 40485
-    },
-    {
-      "epoch": 21.113950456323337,
-      "grad_norm": 1.5348527431488037,
-      "learning_rate": 5.9827135678391965e-05,
-      "loss": 5.3531,
-      "step": 40486
-    },
-    {
-      "epoch": 21.114471968709257,
-      "grad_norm": 1.4761649370193481,
-      "learning_rate": 5.982613065326633e-05,
-      "loss": 5.8766,
-      "step": 40487
-    },
-    {
-      "epoch": 21.114993481095176,
-      "grad_norm": 1.4431233406066895,
-      "learning_rate": 5.982512562814071e-05,
-      "loss": 5.8608,
-      "step": 40488
-    },
-    {
-      "epoch": 21.115514993481096,
-      "grad_norm": 1.4149786233901978,
-      "learning_rate": 5.982412060301508e-05,
-      "loss": 4.7492,
-      "step": 40489
-    },
-    {
-      "epoch": 21.116036505867015,
-      "grad_norm": 1.5748789310455322,
-      "learning_rate": 5.9823115577889456e-05,
-      "loss": 4.7643,
-      "step": 40490
-    },
-    {
-      "epoch": 21.116558018252935,
-      "grad_norm": 1.5708955526351929,
-      "learning_rate": 5.982211055276382e-05,
-      "loss": 5.1629,
-      "step": 40491
-    },
-    {
-      "epoch": 21.117079530638854,
-      "grad_norm": 1.4050750732421875,
-      "learning_rate": 5.98211055276382e-05,
-      "loss": 5.7844,
-      "step": 40492
-    },
-    {
-      "epoch": 21.117601043024774,
-      "grad_norm": 1.5531952381134033,
-      "learning_rate": 5.982010050251257e-05,
-      "loss": 4.6696,
-      "step": 40493
-    },
-    {
-      "epoch": 21.11812255541069,
-      "grad_norm": 1.5518747568130493,
-      "learning_rate": 5.981909547738693e-05,
-      "loss": 5.2633,
-      "step": 40494
-    },
-    {
-      "epoch": 21.11864406779661,
-      "grad_norm": 1.50668466091156,
-      "learning_rate": 5.981809045226131e-05,
-      "loss": 5.5256,
-      "step": 40495
-    },
-    {
-      "epoch": 21.11916558018253,
-      "grad_norm": 1.5196616649627686,
-      "learning_rate": 5.9817085427135675e-05,
-      "loss": 5.666,
-      "step": 40496
-    },
-    {
-      "epoch": 21.119687092568448,
-      "grad_norm": 1.5376895666122437,
-      "learning_rate": 5.981608040201005e-05,
-      "loss": 5.1744,
-      "step": 40497
-    },
-    {
-      "epoch": 21.120208604954367,
-      "grad_norm": 1.5256839990615845,
-      "learning_rate": 5.9815075376884424e-05,
-      "loss": 5.6403,
-      "step": 40498
-    },
-    {
-      "epoch": 21.120730117340287,
-      "grad_norm": 1.4935588836669922,
-      "learning_rate": 5.98140703517588e-05,
-      "loss": 5.5343,
-      "step": 40499
-    },
-    {
-      "epoch": 21.121251629726206,
-      "grad_norm": 1.5131702423095703,
-      "learning_rate": 5.9813065326633166e-05,
-      "loss": 5.5956,
-      "step": 40500
-    },
-    {
-      "epoch": 21.121773142112126,
-      "grad_norm": 1.5082676410675049,
-      "learning_rate": 5.9812060301507544e-05,
-      "loss": 4.5166,
-      "step": 40501
-    },
-    {
-      "epoch": 21.122294654498045,
-      "grad_norm": 1.4995970726013184,
-      "learning_rate": 5.981105527638191e-05,
-      "loss": 5.2458,
-      "step": 40502
-    },
-    {
-      "epoch": 21.122816166883965,
-      "grad_norm": 1.5533170700073242,
-      "learning_rate": 5.9810050251256286e-05,
-      "loss": 5.2451,
-      "step": 40503
-    },
-    {
-      "epoch": 21.123337679269884,
-      "grad_norm": 1.5492286682128906,
-      "learning_rate": 5.980904522613066e-05,
-      "loss": 5.0129,
-      "step": 40504
-    },
-    {
-      "epoch": 21.1238591916558,
-      "grad_norm": 1.49004065990448,
-      "learning_rate": 5.9808040201005035e-05,
-      "loss": 5.2184,
-      "step": 40505
-    },
-    {
-      "epoch": 21.12438070404172,
-      "grad_norm": 1.5308761596679688,
-      "learning_rate": 5.98070351758794e-05,
-      "loss": 5.4461,
-      "step": 40506
-    },
-    {
-      "epoch": 21.12490221642764,
-      "grad_norm": 1.4886794090270996,
-      "learning_rate": 5.9806030150753764e-05,
-      "loss": 4.7275,
-      "step": 40507
-    },
-    {
-      "epoch": 21.12542372881356,
-      "grad_norm": 1.5160338878631592,
-      "learning_rate": 5.980502512562814e-05,
-      "loss": 5.4562,
-      "step": 40508
-    },
-    {
-      "epoch": 21.125945241199478,
-      "grad_norm": 1.5570695400238037,
-      "learning_rate": 5.980402010050251e-05,
-      "loss": 5.5712,
-      "step": 40509
-    },
-    {
-      "epoch": 21.126466753585397,
-      "grad_norm": 1.5468798875808716,
-      "learning_rate": 5.980301507537689e-05,
-      "loss": 5.1783,
-      "step": 40510
-    },
-    {
-      "epoch": 21.126988265971317,
-      "grad_norm": 1.6814110279083252,
-      "learning_rate": 5.9802010050251255e-05,
-      "loss": 5.2683,
-      "step": 40511
-    },
-    {
-      "epoch": 21.127509778357236,
-      "grad_norm": 1.6233142614364624,
-      "learning_rate": 5.980100502512563e-05,
-      "loss": 4.8768,
-      "step": 40512
-    },
-    {
-      "epoch": 21.128031290743156,
-      "grad_norm": 1.4384679794311523,
-      "learning_rate": 5.9800000000000003e-05,
-      "loss": 5.3773,
-      "step": 40513
-    },
-    {
-      "epoch": 21.128552803129075,
-      "grad_norm": 1.5343406200408936,
-      "learning_rate": 5.979899497487438e-05,
-      "loss": 5.1732,
-      "step": 40514
-    },
-    {
-      "epoch": 21.129074315514995,
-      "grad_norm": 1.6052615642547607,
-      "learning_rate": 5.9797989949748746e-05,
-      "loss": 4.64,
-      "step": 40515
-    },
-    {
-      "epoch": 21.129595827900914,
-      "grad_norm": 1.455440878868103,
-      "learning_rate": 5.979698492462312e-05,
-      "loss": 5.7678,
-      "step": 40516
-    },
-    {
-      "epoch": 21.13011734028683,
-      "grad_norm": 1.512150764465332,
-      "learning_rate": 5.979597989949749e-05,
-      "loss": 5.1306,
-      "step": 40517
-    },
-    {
-      "epoch": 21.13063885267275,
-      "grad_norm": 1.604032039642334,
-      "learning_rate": 5.9794974874371865e-05,
-      "loss": 4.9549,
-      "step": 40518
-    },
-    {
-      "epoch": 21.13116036505867,
-      "grad_norm": 1.4354671239852905,
-      "learning_rate": 5.9793969849246237e-05,
-      "loss": 5.6652,
-      "step": 40519
-    },
-    {
-      "epoch": 21.13168187744459,
-      "grad_norm": 1.5312098264694214,
-      "learning_rate": 5.97929648241206e-05,
-      "loss": 4.7124,
-      "step": 40520
-    },
-    {
-      "epoch": 21.132203389830508,
-      "grad_norm": 1.5357425212860107,
-      "learning_rate": 5.979195979899498e-05,
-      "loss": 4.8568,
-      "step": 40521
-    },
-    {
-      "epoch": 21.132724902216427,
-      "grad_norm": 1.5877001285552979,
-      "learning_rate": 5.979095477386934e-05,
-      "loss": 5.4052,
-      "step": 40522
-    },
-    {
-      "epoch": 21.133246414602347,
-      "grad_norm": 1.447835087776184,
-      "learning_rate": 5.978994974874372e-05,
-      "loss": 5.7247,
-      "step": 40523
-    },
-    {
-      "epoch": 21.133767926988266,
-      "grad_norm": 1.463950753211975,
-      "learning_rate": 5.978894472361809e-05,
-      "loss": 5.0692,
-      "step": 40524
-    },
-    {
-      "epoch": 21.134289439374186,
-      "grad_norm": 1.4584983587265015,
-      "learning_rate": 5.978793969849247e-05,
-      "loss": 5.164,
-      "step": 40525
-    },
-    {
-      "epoch": 21.134810951760105,
-      "grad_norm": 1.4880266189575195,
-      "learning_rate": 5.9786934673366834e-05,
-      "loss": 5.4359,
-      "step": 40526
-    },
-    {
-      "epoch": 21.135332464146025,
-      "grad_norm": 1.5281604528427124,
-      "learning_rate": 5.978592964824121e-05,
-      "loss": 5.3851,
-      "step": 40527
-    },
-    {
-      "epoch": 21.135853976531944,
-      "grad_norm": 1.4921480417251587,
-      "learning_rate": 5.9784924623115576e-05,
-      "loss": 5.1255,
-      "step": 40528
-    },
-    {
-      "epoch": 21.13637548891786,
-      "grad_norm": 1.4789621829986572,
-      "learning_rate": 5.9783919597989954e-05,
-      "loss": 5.2672,
-      "step": 40529
-    },
-    {
-      "epoch": 21.13689700130378,
-      "grad_norm": 1.551680088043213,
-      "learning_rate": 5.9782914572864325e-05,
-      "loss": 5.704,
-      "step": 40530
-    },
-    {
-      "epoch": 21.1374185136897,
-      "grad_norm": 1.4833424091339111,
-      "learning_rate": 5.97819095477387e-05,
-      "loss": 4.8105,
-      "step": 40531
-    },
-    {
-      "epoch": 21.13794002607562,
-      "grad_norm": 1.5106178522109985,
-      "learning_rate": 5.978090452261307e-05,
-      "loss": 5.2242,
-      "step": 40532
-    },
-    {
-      "epoch": 21.138461538461538,
-      "grad_norm": 1.4758743047714233,
-      "learning_rate": 5.977989949748744e-05,
-      "loss": 5.1175,
-      "step": 40533
-    },
-    {
-      "epoch": 21.138983050847457,
-      "grad_norm": 1.6112853288650513,
-      "learning_rate": 5.9778894472361816e-05,
-      "loss": 5.3677,
-      "step": 40534
-    },
-    {
-      "epoch": 21.139504563233377,
-      "grad_norm": 1.6271826028823853,
-      "learning_rate": 5.977788944723618e-05,
-      "loss": 5.2227,
-      "step": 40535
-    },
-    {
-      "epoch": 21.140026075619296,
-      "grad_norm": 1.5717853307724,
-      "learning_rate": 5.977688442211056e-05,
-      "loss": 5.2404,
-      "step": 40536
-    },
-    {
-      "epoch": 21.140547588005216,
-      "grad_norm": 1.5730926990509033,
-      "learning_rate": 5.977587939698492e-05,
-      "loss": 5.2007,
-      "step": 40537
-    },
-    {
-      "epoch": 21.141069100391135,
-      "grad_norm": 1.4699945449829102,
-      "learning_rate": 5.97748743718593e-05,
-      "loss": 5.3718,
-      "step": 40538
-    },
-    {
-      "epoch": 21.141590612777055,
-      "grad_norm": 1.454193353652954,
-      "learning_rate": 5.977386934673367e-05,
-      "loss": 5.6283,
-      "step": 40539
-    },
-    {
-      "epoch": 21.142112125162974,
-      "grad_norm": 1.621497392654419,
-      "learning_rate": 5.977286432160805e-05,
-      "loss": 4.8496,
-      "step": 40540
-    },
-    {
-      "epoch": 21.14263363754889,
-      "grad_norm": 1.419335126876831,
-      "learning_rate": 5.977185929648241e-05,
-      "loss": 5.5454,
-      "step": 40541
-    },
-    {
-      "epoch": 21.14315514993481,
-      "grad_norm": 1.5860153436660767,
-      "learning_rate": 5.977085427135679e-05,
-      "loss": 5.375,
-      "step": 40542
-    },
-    {
-      "epoch": 21.14367666232073,
-      "grad_norm": 1.5255674123764038,
-      "learning_rate": 5.9769849246231155e-05,
-      "loss": 5.2427,
-      "step": 40543
-    },
-    {
-      "epoch": 21.14419817470665,
-      "grad_norm": 1.5108267068862915,
-      "learning_rate": 5.976884422110553e-05,
-      "loss": 4.6561,
-      "step": 40544
-    },
-    {
-      "epoch": 21.144719687092568,
-      "grad_norm": 1.5271291732788086,
-      "learning_rate": 5.9767839195979904e-05,
-      "loss": 5.6177,
-      "step": 40545
-    },
-    {
-      "epoch": 21.145241199478487,
-      "grad_norm": 1.4991004467010498,
-      "learning_rate": 5.976683417085427e-05,
-      "loss": 5.7441,
-      "step": 40546
-    },
-    {
-      "epoch": 21.145762711864407,
-      "grad_norm": 1.5990164279937744,
-      "learning_rate": 5.9765829145728646e-05,
-      "loss": 5.1927,
-      "step": 40547
-    },
-    {
-      "epoch": 21.146284224250326,
-      "grad_norm": 1.6530146598815918,
-      "learning_rate": 5.976482412060301e-05,
-      "loss": 4.975,
-      "step": 40548
-    },
-    {
-      "epoch": 21.146805736636246,
-      "grad_norm": 1.5724226236343384,
-      "learning_rate": 5.976381909547739e-05,
-      "loss": 5.4426,
-      "step": 40549
-    },
-    {
-      "epoch": 21.147327249022165,
-      "grad_norm": 1.5909117460250854,
-      "learning_rate": 5.976281407035176e-05,
-      "loss": 5.5623,
-      "step": 40550
-    },
-    {
-      "epoch": 21.147848761408085,
-      "grad_norm": 1.4992179870605469,
-      "learning_rate": 5.976180904522614e-05,
-      "loss": 5.6115,
-      "step": 40551
-    },
-    {
-      "epoch": 21.148370273794004,
-      "grad_norm": 1.6057852506637573,
-      "learning_rate": 5.97608040201005e-05,
-      "loss": 4.8385,
-      "step": 40552
-    },
-    {
-      "epoch": 21.14889178617992,
-      "grad_norm": 1.5973511934280396,
-      "learning_rate": 5.975979899497488e-05,
-      "loss": 5.0595,
-      "step": 40553
-    },
-    {
-      "epoch": 21.14941329856584,
-      "grad_norm": 1.4936127662658691,
-      "learning_rate": 5.975879396984925e-05,
-      "loss": 5.4076,
-      "step": 40554
-    },
-    {
-      "epoch": 21.14993481095176,
-      "grad_norm": 1.4421006441116333,
-      "learning_rate": 5.975778894472363e-05,
-      "loss": 5.6904,
-      "step": 40555
-    },
-    {
-      "epoch": 21.15045632333768,
-      "grad_norm": 1.5135644674301147,
-      "learning_rate": 5.975678391959799e-05,
-      "loss": 5.0872,
-      "step": 40556
-    },
-    {
-      "epoch": 21.150977835723598,
-      "grad_norm": 1.6418724060058594,
-      "learning_rate": 5.975577889447237e-05,
-      "loss": 4.8918,
-      "step": 40557
-    },
-    {
-      "epoch": 21.151499348109517,
-      "grad_norm": 1.429557204246521,
-      "learning_rate": 5.9754773869346735e-05,
-      "loss": 5.7187,
-      "step": 40558
-    },
-    {
-      "epoch": 21.152020860495437,
-      "grad_norm": 1.525814175605774,
-      "learning_rate": 5.9753768844221106e-05,
-      "loss": 5.2657,
-      "step": 40559
-    },
-    {
-      "epoch": 21.152542372881356,
-      "grad_norm": 1.494579553604126,
-      "learning_rate": 5.9752763819095484e-05,
-      "loss": 5.3456,
-      "step": 40560
-    },
-    {
-      "epoch": 21.153063885267276,
-      "grad_norm": 1.578065037727356,
-      "learning_rate": 5.975175879396985e-05,
-      "loss": 5.3397,
-      "step": 40561
-    },
-    {
-      "epoch": 21.153585397653195,
-      "grad_norm": 1.457777976989746,
-      "learning_rate": 5.9750753768844226e-05,
-      "loss": 5.3574,
-      "step": 40562
-    },
-    {
-      "epoch": 21.154106910039115,
-      "grad_norm": 1.5035675764083862,
-      "learning_rate": 5.974974874371859e-05,
-      "loss": 5.3374,
-      "step": 40563
-    },
-    {
-      "epoch": 21.154628422425034,
-      "grad_norm": 1.5819815397262573,
-      "learning_rate": 5.974874371859297e-05,
-      "loss": 5.4262,
-      "step": 40564
-    },
-    {
-      "epoch": 21.15514993481095,
-      "grad_norm": 1.4558929204940796,
-      "learning_rate": 5.974773869346734e-05,
-      "loss": 5.7661,
-      "step": 40565
-    },
-    {
-      "epoch": 21.15567144719687,
-      "grad_norm": 1.6401617527008057,
-      "learning_rate": 5.9746733668341717e-05,
-      "loss": 5.178,
-      "step": 40566
-    },
-    {
-      "epoch": 21.15619295958279,
-      "grad_norm": 1.592689871788025,
-      "learning_rate": 5.974572864321608e-05,
-      "loss": 5.3349,
-      "step": 40567
-    },
-    {
-      "epoch": 21.15671447196871,
-      "grad_norm": 1.5307620763778687,
-      "learning_rate": 5.974472361809046e-05,
-      "loss": 5.1594,
-      "step": 40568
-    },
-    {
-      "epoch": 21.157235984354628,
-      "grad_norm": 1.5390926599502563,
-      "learning_rate": 5.974371859296482e-05,
-      "loss": 5.2129,
-      "step": 40569
-    },
-    {
-      "epoch": 21.157757496740548,
-      "grad_norm": 1.5479118824005127,
-      "learning_rate": 5.97427135678392e-05,
-      "loss": 5.3018,
-      "step": 40570
-    },
-    {
-      "epoch": 21.158279009126467,
-      "grad_norm": 1.535560131072998,
-      "learning_rate": 5.974170854271357e-05,
-      "loss": 5.3338,
-      "step": 40571
-    },
-    {
-      "epoch": 21.158800521512386,
-      "grad_norm": 1.65035879611969,
-      "learning_rate": 5.9740703517587936e-05,
-      "loss": 5.0294,
-      "step": 40572
-    },
-    {
-      "epoch": 21.159322033898306,
-      "grad_norm": 1.452447533607483,
-      "learning_rate": 5.9739698492462314e-05,
-      "loss": 5.6015,
-      "step": 40573
-    },
-    {
-      "epoch": 21.159843546284225,
-      "grad_norm": 1.6139521598815918,
-      "learning_rate": 5.973869346733668e-05,
-      "loss": 5.1903,
-      "step": 40574
-    },
-    {
-      "epoch": 21.160365058670145,
-      "grad_norm": 1.4652280807495117,
-      "learning_rate": 5.9737688442211056e-05,
-      "loss": 5.1274,
-      "step": 40575
-    },
-    {
-      "epoch": 21.160886571056064,
-      "grad_norm": 1.5154614448547363,
-      "learning_rate": 5.973668341708543e-05,
-      "loss": 4.8997,
-      "step": 40576
-    },
-    {
-      "epoch": 21.16140808344198,
-      "grad_norm": 1.41560697555542,
-      "learning_rate": 5.9735678391959805e-05,
-      "loss": 5.4255,
-      "step": 40577
-    },
-    {
-      "epoch": 21.1619295958279,
-      "grad_norm": 1.496713638305664,
-      "learning_rate": 5.973467336683417e-05,
-      "loss": 5.6983,
-      "step": 40578
-    },
-    {
-      "epoch": 21.16245110821382,
-      "grad_norm": 1.449817180633545,
-      "learning_rate": 5.973366834170855e-05,
-      "loss": 5.304,
-      "step": 40579
-    },
-    {
-      "epoch": 21.16297262059974,
-      "grad_norm": 1.5906455516815186,
-      "learning_rate": 5.973266331658292e-05,
-      "loss": 5.1318,
-      "step": 40580
-    },
-    {
-      "epoch": 21.163494132985658,
-      "grad_norm": 1.5979536771774292,
-      "learning_rate": 5.9731658291457296e-05,
-      "loss": 5.091,
-      "step": 40581
-    },
-    {
-      "epoch": 21.164015645371578,
-      "grad_norm": 1.5023746490478516,
-      "learning_rate": 5.973065326633166e-05,
-      "loss": 4.8425,
-      "step": 40582
-    },
-    {
-      "epoch": 21.164537157757497,
-      "grad_norm": 1.5801231861114502,
-      "learning_rate": 5.972964824120604e-05,
-      "loss": 5.3994,
-      "step": 40583
-    },
-    {
-      "epoch": 21.165058670143416,
-      "grad_norm": 1.469386339187622,
-      "learning_rate": 5.97286432160804e-05,
-      "loss": 5.5612,
-      "step": 40584
-    },
-    {
-      "epoch": 21.165580182529336,
-      "grad_norm": 1.4849580526351929,
-      "learning_rate": 5.972763819095478e-05,
-      "loss": 5.3121,
-      "step": 40585
-    },
-    {
-      "epoch": 21.166101694915255,
-      "grad_norm": 1.4124133586883545,
-      "learning_rate": 5.972663316582915e-05,
-      "loss": 5.6649,
-      "step": 40586
-    },
-    {
-      "epoch": 21.166623207301175,
-      "grad_norm": 1.5104923248291016,
-      "learning_rate": 5.9725628140703515e-05,
-      "loss": 5.2213,
-      "step": 40587
-    },
-    {
-      "epoch": 21.167144719687094,
-      "grad_norm": 1.5204596519470215,
-      "learning_rate": 5.972462311557789e-05,
-      "loss": 5.4067,
-      "step": 40588
-    },
-    {
-      "epoch": 21.16766623207301,
-      "grad_norm": 1.5049691200256348,
-      "learning_rate": 5.972361809045226e-05,
-      "loss": 5.2132,
-      "step": 40589
-    },
-    {
-      "epoch": 21.16818774445893,
-      "grad_norm": 1.4714080095291138,
-      "learning_rate": 5.9722613065326635e-05,
-      "loss": 5.4587,
-      "step": 40590
-    },
-    {
-      "epoch": 21.16870925684485,
-      "grad_norm": 1.5509520769119263,
-      "learning_rate": 5.9721608040201006e-05,
-      "loss": 5.5615,
-      "step": 40591
-    },
-    {
-      "epoch": 21.16923076923077,
-      "grad_norm": 1.6020852327346802,
-      "learning_rate": 5.9720603015075384e-05,
-      "loss": 5.4443,
-      "step": 40592
-    },
-    {
-      "epoch": 21.169752281616688,
-      "grad_norm": 1.6457250118255615,
-      "learning_rate": 5.971959798994975e-05,
-      "loss": 4.787,
-      "step": 40593
-    },
-    {
-      "epoch": 21.170273794002608,
-      "grad_norm": 1.6122372150421143,
-      "learning_rate": 5.9718592964824126e-05,
-      "loss": 4.9118,
-      "step": 40594
-    },
-    {
-      "epoch": 21.170795306388527,
-      "grad_norm": 1.5317331552505493,
-      "learning_rate": 5.971758793969849e-05,
-      "loss": 5.3112,
-      "step": 40595
-    },
-    {
-      "epoch": 21.171316818774446,
-      "grad_norm": 1.6187509298324585,
-      "learning_rate": 5.971658291457287e-05,
-      "loss": 4.6214,
-      "step": 40596
-    },
-    {
-      "epoch": 21.171838331160366,
-      "grad_norm": 1.607183575630188,
-      "learning_rate": 5.971557788944724e-05,
-      "loss": 4.893,
-      "step": 40597
-    },
-    {
-      "epoch": 21.172359843546285,
-      "grad_norm": 1.5166726112365723,
-      "learning_rate": 5.971457286432162e-05,
-      "loss": 5.3135,
-      "step": 40598
-    },
-    {
-      "epoch": 21.172881355932205,
-      "grad_norm": 1.5715097188949585,
-      "learning_rate": 5.971356783919598e-05,
-      "loss": 5.2823,
-      "step": 40599
-    },
-    {
-      "epoch": 21.17340286831812,
-      "grad_norm": 1.5899711847305298,
-      "learning_rate": 5.971256281407035e-05,
-      "loss": 5.0503,
-      "step": 40600
-    },
-    {
-      "epoch": 21.17392438070404,
-      "grad_norm": 1.6250227689743042,
-      "learning_rate": 5.971155778894473e-05,
-      "loss": 4.9744,
-      "step": 40601
-    },
-    {
-      "epoch": 21.17444589308996,
-      "grad_norm": 1.425146460533142,
-      "learning_rate": 5.9710552763819095e-05,
-      "loss": 5.1954,
-      "step": 40602
-    },
-    {
-      "epoch": 21.17496740547588,
-      "grad_norm": 1.4517695903778076,
-      "learning_rate": 5.970954773869347e-05,
-      "loss": 5.1685,
-      "step": 40603
-    },
-    {
-      "epoch": 21.1754889178618,
-      "grad_norm": 1.5541102886199951,
-      "learning_rate": 5.970854271356784e-05,
-      "loss": 5.2239,
-      "step": 40604
-    },
-    {
-      "epoch": 21.176010430247718,
-      "grad_norm": 1.6134741306304932,
-      "learning_rate": 5.9707537688442215e-05,
-      "loss": 5.0682,
-      "step": 40605
-    },
-    {
-      "epoch": 21.176531942633638,
-      "grad_norm": 1.4248137474060059,
-      "learning_rate": 5.9706532663316586e-05,
-      "loss": 5.6403,
-      "step": 40606
-    },
-    {
-      "epoch": 21.177053455019557,
-      "grad_norm": 1.6588658094406128,
-      "learning_rate": 5.9705527638190964e-05,
-      "loss": 4.6825,
-      "step": 40607
-    },
-    {
-      "epoch": 21.177574967405477,
-      "grad_norm": 1.5338305234909058,
-      "learning_rate": 5.970452261306533e-05,
-      "loss": 5.3615,
-      "step": 40608
-    },
-    {
-      "epoch": 21.178096479791396,
-      "grad_norm": 1.6117602586746216,
-      "learning_rate": 5.9703517587939706e-05,
-      "loss": 5.0227,
-      "step": 40609
-    },
-    {
-      "epoch": 21.178617992177315,
-      "grad_norm": 1.490416407585144,
-      "learning_rate": 5.970251256281407e-05,
-      "loss": 4.9704,
-      "step": 40610
-    },
-    {
-      "epoch": 21.179139504563235,
-      "grad_norm": 1.5879480838775635,
-      "learning_rate": 5.970150753768845e-05,
-      "loss": 5.0938,
-      "step": 40611
-    },
-    {
-      "epoch": 21.17966101694915,
-      "grad_norm": 1.4804872274398804,
-      "learning_rate": 5.970050251256282e-05,
-      "loss": 5.3505,
-      "step": 40612
-    },
-    {
-      "epoch": 21.18018252933507,
-      "grad_norm": 1.462067723274231,
-      "learning_rate": 5.969949748743718e-05,
-      "loss": 5.4705,
-      "step": 40613
-    },
-    {
-      "epoch": 21.18070404172099,
-      "grad_norm": 1.5462214946746826,
-      "learning_rate": 5.969849246231156e-05,
-      "loss": 5.2449,
-      "step": 40614
-    },
-    {
-      "epoch": 21.18122555410691,
-      "grad_norm": 1.567888855934143,
-      "learning_rate": 5.9697487437185925e-05,
-      "loss": 5.6053,
-      "step": 40615
-    },
-    {
-      "epoch": 21.18174706649283,
-      "grad_norm": 1.5803216695785522,
-      "learning_rate": 5.96964824120603e-05,
-      "loss": 5.4036,
-      "step": 40616
-    },
-    {
-      "epoch": 21.182268578878748,
-      "grad_norm": 1.4339674711227417,
-      "learning_rate": 5.9695477386934674e-05,
-      "loss": 5.7128,
-      "step": 40617
-    },
-    {
-      "epoch": 21.182790091264668,
-      "grad_norm": 1.3847391605377197,
-      "learning_rate": 5.969447236180905e-05,
-      "loss": 5.4878,
-      "step": 40618
-    },
-    {
-      "epoch": 21.183311603650587,
-      "grad_norm": 1.6508862972259521,
-      "learning_rate": 5.9693467336683416e-05,
-      "loss": 4.5501,
-      "step": 40619
-    },
-    {
-      "epoch": 21.183833116036507,
-      "grad_norm": 1.540549874305725,
-      "learning_rate": 5.9692462311557794e-05,
-      "loss": 5.4801,
-      "step": 40620
-    },
-    {
-      "epoch": 21.184354628422426,
-      "grad_norm": 1.4584678411483765,
-      "learning_rate": 5.9691457286432165e-05,
-      "loss": 5.3343,
-      "step": 40621
-    },
-    {
-      "epoch": 21.184876140808345,
-      "grad_norm": 1.6501909494400024,
-      "learning_rate": 5.969045226130654e-05,
-      "loss": 4.759,
-      "step": 40622
-    },
-    {
-      "epoch": 21.185397653194265,
-      "grad_norm": 1.5307040214538574,
-      "learning_rate": 5.968944723618091e-05,
-      "loss": 5.3919,
-      "step": 40623
-    },
-    {
-      "epoch": 21.18591916558018,
-      "grad_norm": 1.576462984085083,
-      "learning_rate": 5.9688442211055285e-05,
-      "loss": 5.4487,
-      "step": 40624
-    },
-    {
-      "epoch": 21.1864406779661,
-      "grad_norm": 1.4717215299606323,
-      "learning_rate": 5.968743718592965e-05,
-      "loss": 5.4381,
-      "step": 40625
-    },
-    {
-      "epoch": 21.18696219035202,
-      "grad_norm": 1.44826078414917,
-      "learning_rate": 5.968643216080402e-05,
-      "loss": 5.453,
-      "step": 40626
-    },
-    {
-      "epoch": 21.18748370273794,
-      "grad_norm": 1.5031006336212158,
-      "learning_rate": 5.96854271356784e-05,
-      "loss": 5.5376,
-      "step": 40627
-    },
-    {
-      "epoch": 21.18800521512386,
-      "grad_norm": 1.656994104385376,
-      "learning_rate": 5.968442211055276e-05,
-      "loss": 5.1949,
-      "step": 40628
-    },
-    {
-      "epoch": 21.188526727509778,
-      "grad_norm": 1.4495015144348145,
-      "learning_rate": 5.968341708542714e-05,
-      "loss": 4.555,
-      "step": 40629
-    },
-    {
-      "epoch": 21.189048239895698,
-      "grad_norm": 1.5009620189666748,
-      "learning_rate": 5.9682412060301505e-05,
-      "loss": 5.3268,
-      "step": 40630
-    },
-    {
-      "epoch": 21.189569752281617,
-      "grad_norm": 1.581083059310913,
-      "learning_rate": 5.968140703517588e-05,
-      "loss": 5.1681,
-      "step": 40631
-    },
-    {
-      "epoch": 21.190091264667537,
-      "grad_norm": 1.4545484781265259,
-      "learning_rate": 5.9680402010050253e-05,
-      "loss": 5.9612,
-      "step": 40632
-    },
-    {
-      "epoch": 21.190612777053456,
-      "grad_norm": 1.434973955154419,
-      "learning_rate": 5.967939698492463e-05,
-      "loss": 5.623,
-      "step": 40633
-    },
-    {
-      "epoch": 21.191134289439375,
-      "grad_norm": 1.6205551624298096,
-      "learning_rate": 5.9678391959798996e-05,
-      "loss": 5.0967,
-      "step": 40634
-    },
-    {
-      "epoch": 21.191655801825295,
-      "grad_norm": 1.5244802236557007,
-      "learning_rate": 5.967738693467337e-05,
-      "loss": 4.9634,
-      "step": 40635
-    },
-    {
-      "epoch": 21.19217731421121,
-      "grad_norm": 1.5559725761413574,
-      "learning_rate": 5.967638190954774e-05,
-      "loss": 5.0815,
-      "step": 40636
-    },
-    {
-      "epoch": 21.19269882659713,
-      "grad_norm": 1.438287377357483,
-      "learning_rate": 5.9675376884422115e-05,
-      "loss": 5.0869,
-      "step": 40637
-    },
-    {
-      "epoch": 21.19322033898305,
-      "grad_norm": 1.5143187046051025,
-      "learning_rate": 5.9674371859296487e-05,
-      "loss": 4.9993,
-      "step": 40638
-    },
-    {
-      "epoch": 21.19374185136897,
-      "grad_norm": 1.6917345523834229,
-      "learning_rate": 5.967336683417085e-05,
-      "loss": 4.8606,
-      "step": 40639
-    },
-    {
-      "epoch": 21.19426336375489,
-      "grad_norm": 1.5854580402374268,
-      "learning_rate": 5.967236180904523e-05,
-      "loss": 5.0629,
-      "step": 40640
-    },
-    {
-      "epoch": 21.194784876140808,
-      "grad_norm": 1.5831925868988037,
-      "learning_rate": 5.96713567839196e-05,
-      "loss": 5.2907,
-      "step": 40641
-    },
-    {
-      "epoch": 21.195306388526728,
-      "grad_norm": 1.62052583694458,
-      "learning_rate": 5.967035175879398e-05,
-      "loss": 4.5758,
-      "step": 40642
-    },
-    {
-      "epoch": 21.195827900912647,
-      "grad_norm": 1.4569064378738403,
-      "learning_rate": 5.966934673366834e-05,
-      "loss": 5.7979,
-      "step": 40643
-    },
-    {
-      "epoch": 21.196349413298567,
-      "grad_norm": 1.736686110496521,
-      "learning_rate": 5.966834170854272e-05,
-      "loss": 4.8302,
-      "step": 40644
-    },
-    {
-      "epoch": 21.196870925684486,
-      "grad_norm": 1.4086229801177979,
-      "learning_rate": 5.9667336683417084e-05,
-      "loss": 5.8459,
-      "step": 40645
-    },
-    {
-      "epoch": 21.197392438070406,
-      "grad_norm": 1.435834288597107,
-      "learning_rate": 5.966633165829146e-05,
-      "loss": 5.5059,
-      "step": 40646
-    },
-    {
-      "epoch": 21.197913950456325,
-      "grad_norm": 1.5013762712478638,
-      "learning_rate": 5.966532663316583e-05,
-      "loss": 5.4594,
-      "step": 40647
-    },
-    {
-      "epoch": 21.19843546284224,
-      "grad_norm": 1.6662782430648804,
-      "learning_rate": 5.966432160804021e-05,
-      "loss": 5.1555,
-      "step": 40648
-    },
-    {
-      "epoch": 21.19895697522816,
-      "grad_norm": 1.5300575494766235,
-      "learning_rate": 5.9663316582914575e-05,
-      "loss": 4.824,
-      "step": 40649
-    },
-    {
-      "epoch": 21.19947848761408,
-      "grad_norm": 1.4639308452606201,
-      "learning_rate": 5.966231155778895e-05,
-      "loss": 5.4297,
-      "step": 40650
-    },
-    {
-      "epoch": 21.2,
-      "grad_norm": 1.5549782514572144,
-      "learning_rate": 5.966130653266332e-05,
-      "loss": 5.2158,
-      "step": 40651
-    },
-    {
-      "epoch": 21.20052151238592,
-      "grad_norm": 1.417754054069519,
-      "learning_rate": 5.966030150753769e-05,
-      "loss": 5.7853,
-      "step": 40652
-    },
-    {
-      "epoch": 21.201043024771838,
-      "grad_norm": 1.5551081895828247,
-      "learning_rate": 5.9659296482412066e-05,
-      "loss": 5.0077,
-      "step": 40653
-    },
-    {
-      "epoch": 21.201564537157758,
-      "grad_norm": 1.4925522804260254,
-      "learning_rate": 5.965829145728643e-05,
-      "loss": 4.8515,
-      "step": 40654
-    },
-    {
-      "epoch": 21.202086049543677,
-      "grad_norm": 1.4862408638000488,
-      "learning_rate": 5.965728643216081e-05,
-      "loss": 5.2133,
-      "step": 40655
-    },
-    {
-      "epoch": 21.202607561929597,
-      "grad_norm": 1.3617701530456543,
-      "learning_rate": 5.965628140703517e-05,
-      "loss": 5.7975,
-      "step": 40656
-    },
-    {
-      "epoch": 21.203129074315516,
-      "grad_norm": 1.461024284362793,
-      "learning_rate": 5.965527638190955e-05,
-      "loss": 5.6379,
-      "step": 40657
-    },
-    {
-      "epoch": 21.203650586701436,
-      "grad_norm": 1.647446870803833,
-      "learning_rate": 5.965427135678392e-05,
-      "loss": 4.6717,
-      "step": 40658
-    },
-    {
-      "epoch": 21.204172099087355,
-      "grad_norm": 1.418860673904419,
-      "learning_rate": 5.96532663316583e-05,
-      "loss": 4.6865,
-      "step": 40659
-    },
-    {
-      "epoch": 21.20469361147327,
-      "grad_norm": 1.5359752178192139,
-      "learning_rate": 5.965226130653266e-05,
-      "loss": 5.2944,
-      "step": 40660
-    },
-    {
-      "epoch": 21.20521512385919,
-      "grad_norm": 1.468827247619629,
-      "learning_rate": 5.965125628140704e-05,
-      "loss": 5.2504,
-      "step": 40661
-    },
-    {
-      "epoch": 21.20573663624511,
-      "grad_norm": 1.582008957862854,
-      "learning_rate": 5.9650251256281405e-05,
-      "loss": 5.3548,
-      "step": 40662
-    },
-    {
-      "epoch": 21.20625814863103,
-      "grad_norm": 1.504395604133606,
-      "learning_rate": 5.964924623115578e-05,
-      "loss": 5.1717,
-      "step": 40663
-    },
-    {
-      "epoch": 21.20677966101695,
-      "grad_norm": 1.5606064796447754,
-      "learning_rate": 5.9648241206030154e-05,
-      "loss": 4.8401,
-      "step": 40664
-    },
-    {
-      "epoch": 21.20730117340287,
-      "grad_norm": 1.5171400308609009,
-      "learning_rate": 5.964723618090452e-05,
-      "loss": 5.0575,
-      "step": 40665
-    },
-    {
-      "epoch": 21.207822685788788,
-      "grad_norm": 1.61776864528656,
-      "learning_rate": 5.9646231155778896e-05,
-      "loss": 4.9886,
-      "step": 40666
-    },
-    {
-      "epoch": 21.208344198174707,
-      "grad_norm": 1.5356152057647705,
-      "learning_rate": 5.964522613065327e-05,
-      "loss": 5.4236,
-      "step": 40667
-    },
-    {
-      "epoch": 21.208865710560627,
-      "grad_norm": 1.5440709590911865,
-      "learning_rate": 5.9644221105527645e-05,
-      "loss": 5.1622,
-      "step": 40668
-    },
-    {
-      "epoch": 21.209387222946546,
-      "grad_norm": 1.65593421459198,
-      "learning_rate": 5.964321608040201e-05,
-      "loss": 4.5791,
-      "step": 40669
-    },
-    {
-      "epoch": 21.209908735332466,
-      "grad_norm": 1.571873426437378,
-      "learning_rate": 5.964221105527639e-05,
-      "loss": 4.9211,
-      "step": 40670
-    },
-    {
-      "epoch": 21.210430247718385,
-      "grad_norm": 1.574280023574829,
-      "learning_rate": 5.964120603015075e-05,
-      "loss": 5.2429,
-      "step": 40671
-    },
-    {
-      "epoch": 21.2109517601043,
-      "grad_norm": 1.3971047401428223,
-      "learning_rate": 5.964020100502513e-05,
-      "loss": 5.6198,
-      "step": 40672
-    },
-    {
-      "epoch": 21.21147327249022,
-      "grad_norm": 1.462854027748108,
-      "learning_rate": 5.96391959798995e-05,
-      "loss": 5.2688,
-      "step": 40673
-    },
-    {
-      "epoch": 21.21199478487614,
-      "grad_norm": 1.5383580923080444,
-      "learning_rate": 5.963819095477388e-05,
-      "loss": 4.2754,
-      "step": 40674
-    },
-    {
-      "epoch": 21.21251629726206,
-      "grad_norm": 1.6117019653320312,
-      "learning_rate": 5.963718592964824e-05,
-      "loss": 5.115,
-      "step": 40675
-    },
-    {
-      "epoch": 21.21303780964798,
-      "grad_norm": 1.5221608877182007,
-      "learning_rate": 5.963618090452262e-05,
-      "loss": 5.2615,
-      "step": 40676
-    },
-    {
-      "epoch": 21.2135593220339,
-      "grad_norm": 1.4906824827194214,
-      "learning_rate": 5.9635175879396985e-05,
-      "loss": 5.3335,
-      "step": 40677
-    },
-    {
-      "epoch": 21.214080834419818,
-      "grad_norm": 1.5931057929992676,
-      "learning_rate": 5.963417085427136e-05,
-      "loss": 5.1498,
-      "step": 40678
-    },
-    {
-      "epoch": 21.214602346805737,
-      "grad_norm": 1.52841317653656,
-      "learning_rate": 5.9633165829145734e-05,
-      "loss": 5.2862,
-      "step": 40679
-    },
-    {
-      "epoch": 21.215123859191657,
-      "grad_norm": 1.6055434942245483,
-      "learning_rate": 5.96321608040201e-05,
-      "loss": 5.5136,
-      "step": 40680
-    },
-    {
-      "epoch": 21.215645371577576,
-      "grad_norm": 1.7348253726959229,
-      "learning_rate": 5.9631155778894476e-05,
-      "loss": 5.2745,
-      "step": 40681
-    },
-    {
-      "epoch": 21.216166883963496,
-      "grad_norm": 1.6879568099975586,
-      "learning_rate": 5.963015075376884e-05,
-      "loss": 5.0558,
-      "step": 40682
-    },
-    {
-      "epoch": 21.216688396349415,
-      "grad_norm": 1.5645301342010498,
-      "learning_rate": 5.962914572864322e-05,
-      "loss": 4.4842,
-      "step": 40683
-    },
-    {
-      "epoch": 21.21720990873533,
-      "grad_norm": 1.5281503200531006,
-      "learning_rate": 5.962814070351759e-05,
-      "loss": 5.3088,
-      "step": 40684
-    },
-    {
-      "epoch": 21.21773142112125,
-      "grad_norm": 1.4888880252838135,
-      "learning_rate": 5.962713567839197e-05,
-      "loss": 5.4383,
-      "step": 40685
-    },
-    {
-      "epoch": 21.21825293350717,
-      "grad_norm": 1.561076283454895,
-      "learning_rate": 5.962613065326633e-05,
-      "loss": 5.1468,
-      "step": 40686
-    },
-    {
-      "epoch": 21.21877444589309,
-      "grad_norm": 1.4732894897460938,
-      "learning_rate": 5.962512562814071e-05,
-      "loss": 5.5201,
-      "step": 40687
-    },
-    {
-      "epoch": 21.21929595827901,
-      "grad_norm": 1.6400636434555054,
-      "learning_rate": 5.962412060301508e-05,
-      "loss": 4.476,
-      "step": 40688
-    },
-    {
-      "epoch": 21.21981747066493,
-      "grad_norm": 1.4940588474273682,
-      "learning_rate": 5.962311557788946e-05,
-      "loss": 4.7539,
-      "step": 40689
-    },
-    {
-      "epoch": 21.220338983050848,
-      "grad_norm": 1.4909626245498657,
-      "learning_rate": 5.962211055276382e-05,
-      "loss": 5.4266,
-      "step": 40690
-    },
-    {
-      "epoch": 21.220860495436767,
-      "grad_norm": 1.6058056354522705,
-      "learning_rate": 5.96211055276382e-05,
-      "loss": 4.961,
-      "step": 40691
-    },
-    {
-      "epoch": 21.221382007822687,
-      "grad_norm": 1.4971870183944702,
-      "learning_rate": 5.9620100502512564e-05,
-      "loss": 5.6193,
-      "step": 40692
-    },
-    {
-      "epoch": 21.221903520208606,
-      "grad_norm": 1.44918692111969,
-      "learning_rate": 5.9619095477386935e-05,
-      "loss": 5.4614,
-      "step": 40693
-    },
-    {
-      "epoch": 21.222425032594526,
-      "grad_norm": 1.5328431129455566,
-      "learning_rate": 5.961809045226131e-05,
-      "loss": 5.2643,
-      "step": 40694
-    },
-    {
-      "epoch": 21.22294654498044,
-      "grad_norm": 1.4669524431228638,
-      "learning_rate": 5.961708542713568e-05,
-      "loss": 5.0707,
-      "step": 40695
-    },
-    {
-      "epoch": 21.22346805736636,
-      "grad_norm": 1.4946417808532715,
-      "learning_rate": 5.9616080402010055e-05,
-      "loss": 5.5828,
-      "step": 40696
-    },
-    {
-      "epoch": 21.22398956975228,
-      "grad_norm": 1.4900370836257935,
-      "learning_rate": 5.961507537688442e-05,
-      "loss": 5.2675,
-      "step": 40697
-    },
-    {
-      "epoch": 21.2245110821382,
-      "grad_norm": 1.5576834678649902,
-      "learning_rate": 5.96140703517588e-05,
-      "loss": 4.7376,
-      "step": 40698
-    },
-    {
-      "epoch": 21.22503259452412,
-      "grad_norm": 1.4590167999267578,
-      "learning_rate": 5.961306532663317e-05,
-      "loss": 5.3745,
-      "step": 40699
-    },
-    {
-      "epoch": 21.22555410691004,
-      "grad_norm": 1.6524975299835205,
-      "learning_rate": 5.9612060301507546e-05,
-      "loss": 4.8418,
-      "step": 40700
-    },
-    {
-      "epoch": 21.22607561929596,
-      "grad_norm": 1.517913818359375,
-      "learning_rate": 5.961105527638191e-05,
-      "loss": 5.3512,
-      "step": 40701
-    },
-    {
-      "epoch": 21.226597131681878,
-      "grad_norm": 1.5073001384735107,
-      "learning_rate": 5.961005025125629e-05,
-      "loss": 4.8733,
-      "step": 40702
-    },
-    {
-      "epoch": 21.227118644067797,
-      "grad_norm": 1.6129887104034424,
-      "learning_rate": 5.960904522613065e-05,
-      "loss": 5.4667,
-      "step": 40703
-    },
-    {
-      "epoch": 21.227640156453717,
-      "grad_norm": 1.4693940877914429,
-      "learning_rate": 5.960804020100503e-05,
-      "loss": 5.4461,
-      "step": 40704
-    },
-    {
-      "epoch": 21.228161668839636,
-      "grad_norm": 1.5384207963943481,
-      "learning_rate": 5.96070351758794e-05,
-      "loss": 5.2487,
-      "step": 40705
-    },
-    {
-      "epoch": 21.228683181225556,
-      "grad_norm": 1.626287579536438,
-      "learning_rate": 5.9606030150753765e-05,
-      "loss": 4.7199,
-      "step": 40706
-    },
-    {
-      "epoch": 21.22920469361147,
-      "grad_norm": 1.6779847145080566,
-      "learning_rate": 5.960502512562814e-05,
-      "loss": 5.2069,
-      "step": 40707
-    },
-    {
-      "epoch": 21.22972620599739,
-      "grad_norm": 1.5597922801971436,
-      "learning_rate": 5.9604020100502514e-05,
-      "loss": 5.4751,
-      "step": 40708
-    },
-    {
-      "epoch": 21.23024771838331,
-      "grad_norm": 1.5380358695983887,
-      "learning_rate": 5.960301507537689e-05,
-      "loss": 5.369,
-      "step": 40709
-    },
-    {
-      "epoch": 21.23076923076923,
-      "grad_norm": 1.4512758255004883,
-      "learning_rate": 5.9602010050251256e-05,
-      "loss": 5.2023,
-      "step": 40710
-    },
-    {
-      "epoch": 21.23129074315515,
-      "grad_norm": 1.5846806764602661,
-      "learning_rate": 5.9601005025125634e-05,
-      "loss": 5.0814,
-      "step": 40711
-    },
-    {
-      "epoch": 21.23181225554107,
-      "grad_norm": 1.4970588684082031,
-      "learning_rate": 5.96e-05,
-      "loss": 5.3202,
-      "step": 40712
-    },
-    {
-      "epoch": 21.23233376792699,
-      "grad_norm": 1.4731444120407104,
-      "learning_rate": 5.9598994974874376e-05,
-      "loss": 5.5008,
-      "step": 40713
-    },
-    {
-      "epoch": 21.232855280312908,
-      "grad_norm": 1.372088074684143,
-      "learning_rate": 5.959798994974875e-05,
-      "loss": 5.5865,
-      "step": 40714
-    },
-    {
-      "epoch": 21.233376792698827,
-      "grad_norm": 1.712204098701477,
-      "learning_rate": 5.9596984924623125e-05,
-      "loss": 5.1757,
-      "step": 40715
-    },
-    {
-      "epoch": 21.233898305084747,
-      "grad_norm": 1.4219629764556885,
-      "learning_rate": 5.959597989949749e-05,
-      "loss": 5.6423,
-      "step": 40716
-    },
-    {
-      "epoch": 21.234419817470666,
-      "grad_norm": 1.51162588596344,
-      "learning_rate": 5.959497487437187e-05,
-      "loss": 5.495,
-      "step": 40717
-    },
-    {
-      "epoch": 21.234941329856586,
-      "grad_norm": 1.4844706058502197,
-      "learning_rate": 5.959396984924623e-05,
-      "loss": 5.5553,
-      "step": 40718
-    },
-    {
-      "epoch": 21.2354628422425,
-      "grad_norm": 1.5182784795761108,
-      "learning_rate": 5.95929648241206e-05,
-      "loss": 5.2003,
-      "step": 40719
-    },
-    {
-      "epoch": 21.23598435462842,
-      "grad_norm": 1.5879640579223633,
-      "learning_rate": 5.959195979899498e-05,
-      "loss": 5.3074,
-      "step": 40720
-    },
-    {
-      "epoch": 21.23650586701434,
-      "grad_norm": 1.5592247247695923,
-      "learning_rate": 5.9590954773869345e-05,
-      "loss": 5.2428,
-      "step": 40721
-    },
-    {
-      "epoch": 21.23702737940026,
-      "grad_norm": 1.5055763721466064,
-      "learning_rate": 5.958994974874372e-05,
-      "loss": 5.3936,
-      "step": 40722
-    },
-    {
-      "epoch": 21.23754889178618,
-      "grad_norm": 1.5267512798309326,
-      "learning_rate": 5.958894472361809e-05,
-      "loss": 5.0857,
-      "step": 40723
-    },
-    {
-      "epoch": 21.2380704041721,
-      "grad_norm": 1.5841096639633179,
-      "learning_rate": 5.9587939698492465e-05,
-      "loss": 4.9565,
-      "step": 40724
-    },
-    {
-      "epoch": 21.23859191655802,
-      "grad_norm": 1.5224006175994873,
-      "learning_rate": 5.9586934673366836e-05,
-      "loss": 5.169,
-      "step": 40725
-    },
-    {
-      "epoch": 21.239113428943938,
-      "grad_norm": 1.5191453695297241,
-      "learning_rate": 5.9585929648241214e-05,
-      "loss": 5.613,
-      "step": 40726
-    },
-    {
-      "epoch": 21.239634941329857,
-      "grad_norm": 1.5786255598068237,
-      "learning_rate": 5.958492462311558e-05,
-      "loss": 5.3386,
-      "step": 40727
-    },
-    {
-      "epoch": 21.240156453715777,
-      "grad_norm": 1.4854495525360107,
-      "learning_rate": 5.9583919597989956e-05,
-      "loss": 5.1638,
-      "step": 40728
-    },
-    {
-      "epoch": 21.240677966101696,
-      "grad_norm": 1.646871566772461,
-      "learning_rate": 5.958291457286433e-05,
-      "loss": 4.7628,
-      "step": 40729
-    },
-    {
-      "epoch": 21.241199478487616,
-      "grad_norm": 1.3875795602798462,
-      "learning_rate": 5.9581909547738705e-05,
-      "loss": 5.5777,
-      "step": 40730
-    },
-    {
-      "epoch": 21.24172099087353,
-      "grad_norm": 1.3808659315109253,
-      "learning_rate": 5.958090452261307e-05,
-      "loss": 5.5746,
-      "step": 40731
-    },
-    {
-      "epoch": 21.24224250325945,
-      "grad_norm": 1.5761308670043945,
-      "learning_rate": 5.957989949748743e-05,
-      "loss": 4.9748,
-      "step": 40732
-    },
-    {
-      "epoch": 21.24276401564537,
-      "grad_norm": 1.5349422693252563,
-      "learning_rate": 5.957889447236181e-05,
-      "loss": 5.3548,
-      "step": 40733
-    },
-    {
-      "epoch": 21.24328552803129,
-      "grad_norm": 1.5793297290802002,
-      "learning_rate": 5.957788944723618e-05,
-      "loss": 5.2812,
-      "step": 40734
-    },
-    {
-      "epoch": 21.24380704041721,
-      "grad_norm": 1.4175671339035034,
-      "learning_rate": 5.957688442211056e-05,
-      "loss": 5.2258,
-      "step": 40735
-    },
-    {
-      "epoch": 21.24432855280313,
-      "grad_norm": 1.5572582483291626,
-      "learning_rate": 5.9575879396984924e-05,
-      "loss": 5.0401,
-      "step": 40736
-    },
-    {
-      "epoch": 21.24485006518905,
-      "grad_norm": 1.4599263668060303,
-      "learning_rate": 5.95748743718593e-05,
-      "loss": 5.3435,
-      "step": 40737
-    },
-    {
-      "epoch": 21.245371577574968,
-      "grad_norm": 1.5179498195648193,
-      "learning_rate": 5.9573869346733666e-05,
-      "loss": 5.486,
-      "step": 40738
-    },
-    {
-      "epoch": 21.245893089960887,
-      "grad_norm": 1.5965720415115356,
-      "learning_rate": 5.9572864321608044e-05,
-      "loss": 4.8161,
-      "step": 40739
-    },
-    {
-      "epoch": 21.246414602346807,
-      "grad_norm": 1.508539080619812,
-      "learning_rate": 5.9571859296482415e-05,
-      "loss": 5.2906,
-      "step": 40740
-    },
-    {
-      "epoch": 21.246936114732726,
-      "grad_norm": 1.4128646850585938,
-      "learning_rate": 5.957085427135679e-05,
-      "loss": 5.0294,
-      "step": 40741
-    },
-    {
-      "epoch": 21.247457627118646,
-      "grad_norm": 1.4843413829803467,
-      "learning_rate": 5.956984924623116e-05,
-      "loss": 4.9894,
-      "step": 40742
-    },
-    {
-      "epoch": 21.24797913950456,
-      "grad_norm": 1.5948389768600464,
-      "learning_rate": 5.9568844221105535e-05,
-      "loss": 5.2611,
-      "step": 40743
-    },
-    {
-      "epoch": 21.24850065189048,
-      "grad_norm": 1.5166685581207275,
-      "learning_rate": 5.95678391959799e-05,
-      "loss": 5.1153,
-      "step": 40744
-    },
-    {
-      "epoch": 21.2490221642764,
-      "grad_norm": 1.7201374769210815,
-      "learning_rate": 5.956683417085427e-05,
-      "loss": 5.2824,
-      "step": 40745
-    },
-    {
-      "epoch": 21.24954367666232,
-      "grad_norm": 1.5264537334442139,
-      "learning_rate": 5.956582914572865e-05,
-      "loss": 5.0989,
-      "step": 40746
-    },
-    {
-      "epoch": 21.25006518904824,
-      "grad_norm": 1.4438689947128296,
-      "learning_rate": 5.956482412060301e-05,
-      "loss": 5.3879,
-      "step": 40747
-    },
-    {
-      "epoch": 21.25058670143416,
-      "grad_norm": 1.4175055027008057,
-      "learning_rate": 5.956381909547739e-05,
-      "loss": 4.7027,
-      "step": 40748
-    },
-    {
-      "epoch": 21.25110821382008,
-      "grad_norm": 1.570082187652588,
-      "learning_rate": 5.9562814070351755e-05,
-      "loss": 5.1995,
-      "step": 40749
-    },
-    {
-      "epoch": 21.251629726205998,
-      "grad_norm": 1.6001746654510498,
-      "learning_rate": 5.956180904522613e-05,
-      "loss": 5.2271,
-      "step": 40750
-    },
-    {
-      "epoch": 21.252151238591917,
-      "grad_norm": 1.5747109651565552,
-      "learning_rate": 5.9560804020100503e-05,
-      "loss": 5.2623,
-      "step": 40751
-    },
-    {
-      "epoch": 21.252672750977837,
-      "grad_norm": 1.5383111238479614,
-      "learning_rate": 5.955979899497488e-05,
-      "loss": 5.575,
-      "step": 40752
-    },
-    {
-      "epoch": 21.253194263363756,
-      "grad_norm": 1.471304178237915,
-      "learning_rate": 5.9558793969849246e-05,
-      "loss": 5.3038,
-      "step": 40753
-    },
-    {
-      "epoch": 21.253715775749676,
-      "grad_norm": 1.5575268268585205,
-      "learning_rate": 5.9557788944723623e-05,
-      "loss": 4.7148,
-      "step": 40754
-    },
-    {
-      "epoch": 21.25423728813559,
-      "grad_norm": 1.4256460666656494,
-      "learning_rate": 5.9556783919597994e-05,
-      "loss": 5.6964,
-      "step": 40755
-    },
-    {
-      "epoch": 21.25475880052151,
-      "grad_norm": 1.5142194032669067,
-      "learning_rate": 5.955577889447237e-05,
-      "loss": 5.5685,
-      "step": 40756
-    },
-    {
-      "epoch": 21.25528031290743,
-      "grad_norm": 1.657931923866272,
-      "learning_rate": 5.9554773869346737e-05,
-      "loss": 5.0898,
-      "step": 40757
-    },
-    {
-      "epoch": 21.25580182529335,
-      "grad_norm": 1.5456980466842651,
-      "learning_rate": 5.9553768844221114e-05,
-      "loss": 5.1292,
-      "step": 40758
-    },
-    {
-      "epoch": 21.25632333767927,
-      "grad_norm": 1.5730788707733154,
-      "learning_rate": 5.955276381909548e-05,
-      "loss": 5.281,
-      "step": 40759
-    },
-    {
-      "epoch": 21.25684485006519,
-      "grad_norm": 1.5308345556259155,
-      "learning_rate": 5.955175879396985e-05,
-      "loss": 5.2778,
-      "step": 40760
-    },
-    {
-      "epoch": 21.25736636245111,
-      "grad_norm": 1.5675803422927856,
-      "learning_rate": 5.955075376884423e-05,
-      "loss": 4.955,
-      "step": 40761
-    },
-    {
-      "epoch": 21.257887874837028,
-      "grad_norm": 1.5283371210098267,
-      "learning_rate": 5.954974874371859e-05,
-      "loss": 4.9722,
-      "step": 40762
-    },
-    {
-      "epoch": 21.258409387222947,
-      "grad_norm": 1.522217035293579,
-      "learning_rate": 5.954874371859297e-05,
-      "loss": 5.3346,
-      "step": 40763
-    },
-    {
-      "epoch": 21.258930899608867,
-      "grad_norm": 1.6428052186965942,
-      "learning_rate": 5.9547738693467334e-05,
-      "loss": 5.3357,
-      "step": 40764
-    },
-    {
-      "epoch": 21.259452411994786,
-      "grad_norm": 1.5825231075286865,
-      "learning_rate": 5.954673366834171e-05,
-      "loss": 4.8384,
-      "step": 40765
-    },
-    {
-      "epoch": 21.259973924380706,
-      "grad_norm": 1.5588725805282593,
-      "learning_rate": 5.954572864321608e-05,
-      "loss": 5.3867,
-      "step": 40766
-    },
-    {
-      "epoch": 21.26049543676662,
-      "grad_norm": 1.54362952709198,
-      "learning_rate": 5.954472361809046e-05,
-      "loss": 5.1471,
-      "step": 40767
-    },
-    {
-      "epoch": 21.26101694915254,
-      "grad_norm": 1.5852717161178589,
-      "learning_rate": 5.9543718592964825e-05,
-      "loss": 5.2636,
-      "step": 40768
-    },
-    {
-      "epoch": 21.26153846153846,
-      "grad_norm": 1.5657789707183838,
-      "learning_rate": 5.95427135678392e-05,
-      "loss": 4.725,
-      "step": 40769
-    },
-    {
-      "epoch": 21.26205997392438,
-      "grad_norm": 1.561194658279419,
-      "learning_rate": 5.954170854271357e-05,
-      "loss": 5.0246,
-      "step": 40770
-    },
-    {
-      "epoch": 21.2625814863103,
-      "grad_norm": 1.5646576881408691,
-      "learning_rate": 5.9540703517587945e-05,
-      "loss": 4.9327,
-      "step": 40771
-    },
-    {
-      "epoch": 21.26310299869622,
-      "grad_norm": 1.6228318214416504,
-      "learning_rate": 5.9539698492462316e-05,
-      "loss": 5.09,
-      "step": 40772
-    },
-    {
-      "epoch": 21.26362451108214,
-      "grad_norm": 1.5183427333831787,
-      "learning_rate": 5.953869346733668e-05,
-      "loss": 5.1868,
-      "step": 40773
-    },
-    {
-      "epoch": 21.264146023468058,
-      "grad_norm": 1.6428048610687256,
-      "learning_rate": 5.953768844221106e-05,
-      "loss": 5.0283,
-      "step": 40774
-    },
-    {
-      "epoch": 21.264667535853977,
-      "grad_norm": 1.564725637435913,
-      "learning_rate": 5.953668341708543e-05,
-      "loss": 4.6594,
-      "step": 40775
-    },
-    {
-      "epoch": 21.265189048239897,
-      "grad_norm": 1.5947381258010864,
-      "learning_rate": 5.953567839195981e-05,
-      "loss": 5.4537,
-      "step": 40776
-    },
-    {
-      "epoch": 21.265710560625816,
-      "grad_norm": 1.53286874294281,
-      "learning_rate": 5.953467336683417e-05,
-      "loss": 4.7313,
-      "step": 40777
-    },
-    {
-      "epoch": 21.266232073011736,
-      "grad_norm": 1.7175650596618652,
-      "learning_rate": 5.953366834170855e-05,
-      "loss": 5.0876,
-      "step": 40778
-    },
-    {
-      "epoch": 21.26675358539765,
-      "grad_norm": 1.606264591217041,
-      "learning_rate": 5.953266331658291e-05,
-      "loss": 5.0314,
-      "step": 40779
-    },
-    {
-      "epoch": 21.26727509778357,
-      "grad_norm": 1.5979820489883423,
-      "learning_rate": 5.953165829145729e-05,
-      "loss": 5.2411,
-      "step": 40780
-    },
-    {
-      "epoch": 21.26779661016949,
-      "grad_norm": 1.6378285884857178,
-      "learning_rate": 5.953065326633166e-05,
-      "loss": 5.0771,
-      "step": 40781
-    },
-    {
-      "epoch": 21.26831812255541,
-      "grad_norm": 1.4904508590698242,
-      "learning_rate": 5.952964824120604e-05,
-      "loss": 5.5173,
-      "step": 40782
-    },
-    {
-      "epoch": 21.26883963494133,
-      "grad_norm": 1.6446096897125244,
-      "learning_rate": 5.9528643216080404e-05,
-      "loss": 4.5363,
-      "step": 40783
-    },
-    {
-      "epoch": 21.26936114732725,
-      "grad_norm": 1.5241581201553345,
-      "learning_rate": 5.952763819095478e-05,
-      "loss": 5.1654,
-      "step": 40784
-    },
-    {
-      "epoch": 21.26988265971317,
-      "grad_norm": 1.547930121421814,
-      "learning_rate": 5.9526633165829146e-05,
-      "loss": 5.085,
-      "step": 40785
-    },
-    {
-      "epoch": 21.270404172099088,
-      "grad_norm": 1.5604771375656128,
-      "learning_rate": 5.952562814070352e-05,
-      "loss": 5.3727,
-      "step": 40786
-    },
-    {
-      "epoch": 21.270925684485007,
-      "grad_norm": 1.662916660308838,
-      "learning_rate": 5.9524623115577895e-05,
-      "loss": 5.2492,
-      "step": 40787
-    },
-    {
-      "epoch": 21.271447196870927,
-      "grad_norm": 1.7025134563446045,
-      "learning_rate": 5.952361809045226e-05,
-      "loss": 5.3127,
-      "step": 40788
-    },
-    {
-      "epoch": 21.271968709256846,
-      "grad_norm": 1.5748717784881592,
-      "learning_rate": 5.952261306532664e-05,
-      "loss": 4.988,
-      "step": 40789
-    },
-    {
-      "epoch": 21.272490221642762,
-      "grad_norm": 1.5459505319595337,
-      "learning_rate": 5.9521608040201e-05,
-      "loss": 5.5061,
-      "step": 40790
-    },
-    {
-      "epoch": 21.27301173402868,
-      "grad_norm": 1.4504777193069458,
-      "learning_rate": 5.952060301507538e-05,
-      "loss": 5.3292,
-      "step": 40791
-    },
-    {
-      "epoch": 21.2735332464146,
-      "grad_norm": 1.5916770696640015,
-      "learning_rate": 5.951959798994975e-05,
-      "loss": 4.8519,
-      "step": 40792
-    },
-    {
-      "epoch": 21.27405475880052,
-      "grad_norm": 1.5238193273544312,
-      "learning_rate": 5.951859296482413e-05,
-      "loss": 5.6013,
-      "step": 40793
-    },
-    {
-      "epoch": 21.27457627118644,
-      "grad_norm": 1.4957349300384521,
-      "learning_rate": 5.951758793969849e-05,
-      "loss": 5.3082,
-      "step": 40794
-    },
-    {
-      "epoch": 21.27509778357236,
-      "grad_norm": 1.524200201034546,
-      "learning_rate": 5.951658291457287e-05,
-      "loss": 5.6481,
-      "step": 40795
-    },
-    {
-      "epoch": 21.27561929595828,
-      "grad_norm": 1.4160727262496948,
-      "learning_rate": 5.951557788944724e-05,
-      "loss": 5.6878,
-      "step": 40796
-    },
-    {
-      "epoch": 21.2761408083442,
-      "grad_norm": 1.5115456581115723,
-      "learning_rate": 5.951457286432162e-05,
-      "loss": 5.4487,
-      "step": 40797
-    },
-    {
-      "epoch": 21.276662320730118,
-      "grad_norm": 1.7243092060089111,
-      "learning_rate": 5.9513567839195984e-05,
-      "loss": 4.4869,
-      "step": 40798
-    },
-    {
-      "epoch": 21.277183833116037,
-      "grad_norm": 1.5576086044311523,
-      "learning_rate": 5.951256281407035e-05,
-      "loss": 5.4606,
-      "step": 40799
-    },
-    {
-      "epoch": 21.277705345501957,
-      "grad_norm": 1.5223690271377563,
-      "learning_rate": 5.9511557788944726e-05,
-      "loss": 5.3064,
-      "step": 40800
-    },
-    {
-      "epoch": 21.278226857887876,
-      "grad_norm": 1.7975701093673706,
-      "learning_rate": 5.95105527638191e-05,
-      "loss": 5.0122,
-      "step": 40801
-    },
-    {
-      "epoch": 21.278748370273792,
-      "grad_norm": Infinity,
-      "learning_rate": 5.95105527638191e-05,
-      "loss": 4.0356,
-      "step": 40802
-    },
-    {
-      "epoch": 21.27926988265971,
-      "grad_norm": 1.5575697422027588,
-      "learning_rate": 5.9509547738693475e-05,
-      "loss": 5.5336,
-      "step": 40803
-    },
-    {
-      "epoch": 21.27979139504563,
-      "grad_norm": 1.4135459661483765,
-      "learning_rate": 5.950854271356784e-05,
-      "loss": 5.486,
-      "step": 40804
-    },
-    {
-      "epoch": 21.28031290743155,
-      "grad_norm": 1.5902893543243408,
-      "learning_rate": 5.950753768844222e-05,
-      "loss": 5.3169,
-      "step": 40805
-    },
-    {
-      "epoch": 21.28083441981747,
-      "grad_norm": 1.6225979328155518,
-      "learning_rate": 5.950653266331658e-05,
-      "loss": 5.1091,
-      "step": 40806
-    },
-    {
-      "epoch": 21.28135593220339,
-      "grad_norm": 1.4464935064315796,
-      "learning_rate": 5.950552763819096e-05,
-      "loss": 5.198,
-      "step": 40807
-    },
-    {
-      "epoch": 21.28187744458931,
-      "grad_norm": 1.5664358139038086,
-      "learning_rate": 5.950452261306533e-05,
-      "loss": 5.302,
-      "step": 40808
-    },
-    {
-      "epoch": 21.28239895697523,
-      "grad_norm": 1.519441843032837,
-      "learning_rate": 5.950351758793971e-05,
-      "loss": 5.065,
-      "step": 40809
-    },
-    {
-      "epoch": 21.282920469361148,
-      "grad_norm": 1.554803729057312,
-      "learning_rate": 5.950251256281407e-05,
-      "loss": 5.3305,
-      "step": 40810
-    },
-    {
-      "epoch": 21.283441981747067,
-      "grad_norm": 1.404586911201477,
-      "learning_rate": 5.950150753768845e-05,
-      "loss": 5.2685,
-      "step": 40811
-    },
-    {
-      "epoch": 21.283963494132987,
-      "grad_norm": 1.5561023950576782,
-      "learning_rate": 5.9500502512562814e-05,
-      "loss": 5.1981,
-      "step": 40812
-    },
-    {
-      "epoch": 21.284485006518906,
-      "grad_norm": 1.5188193321228027,
-      "learning_rate": 5.9499497487437185e-05,
-      "loss": 4.8598,
-      "step": 40813
-    },
-    {
-      "epoch": 21.285006518904822,
-      "grad_norm": 1.563275933265686,
-      "learning_rate": 5.949849246231156e-05,
-      "loss": 5.1793,
-      "step": 40814
-    },
-    {
-      "epoch": 21.285528031290742,
-      "grad_norm": 1.5542573928833008,
-      "learning_rate": 5.949748743718593e-05,
-      "loss": 5.2132,
-      "step": 40815
-    },
-    {
-      "epoch": 21.28604954367666,
-      "grad_norm": 1.497840166091919,
-      "learning_rate": 5.9496482412060305e-05,
-      "loss": 5.6925,
-      "step": 40816
-    },
-    {
-      "epoch": 21.28657105606258,
-      "grad_norm": 1.5876822471618652,
-      "learning_rate": 5.9495477386934676e-05,
-      "loss": 5.0219,
-      "step": 40817
-    },
-    {
-      "epoch": 21.2870925684485,
-      "grad_norm": 1.4898223876953125,
-      "learning_rate": 5.9494472361809054e-05,
-      "loss": 5.3229,
-      "step": 40818
-    },
-    {
-      "epoch": 21.28761408083442,
-      "grad_norm": 1.5026066303253174,
-      "learning_rate": 5.949346733668342e-05,
-      "loss": 5.6066,
-      "step": 40819
-    },
-    {
-      "epoch": 21.28813559322034,
-      "grad_norm": 1.5082387924194336,
-      "learning_rate": 5.9492462311557796e-05,
-      "loss": 5.3428,
-      "step": 40820
-    },
-    {
-      "epoch": 21.28865710560626,
-      "grad_norm": 1.5716331005096436,
-      "learning_rate": 5.949145728643216e-05,
-      "loss": 5.5041,
-      "step": 40821
-    },
-    {
-      "epoch": 21.289178617992178,
-      "grad_norm": 1.5413674116134644,
-      "learning_rate": 5.949045226130654e-05,
-      "loss": 5.262,
-      "step": 40822
-    },
-    {
-      "epoch": 21.289700130378097,
-      "grad_norm": 1.451149582862854,
-      "learning_rate": 5.948944723618091e-05,
-      "loss": 5.272,
-      "step": 40823
-    },
-    {
-      "epoch": 21.290221642764017,
-      "grad_norm": 1.5008817911148071,
-      "learning_rate": 5.948844221105529e-05,
-      "loss": 5.4996,
-      "step": 40824
-    },
-    {
-      "epoch": 21.290743155149936,
-      "grad_norm": 1.447507619857788,
-      "learning_rate": 5.948743718592965e-05,
-      "loss": 5.5659,
-      "step": 40825
-    },
-    {
-      "epoch": 21.291264667535852,
-      "grad_norm": 1.4228202104568481,
-      "learning_rate": 5.9486432160804016e-05,
-      "loss": 5.6322,
-      "step": 40826
-    },
-    {
-      "epoch": 21.291786179921772,
-      "grad_norm": 1.4734055995941162,
-      "learning_rate": 5.948542713567839e-05,
-      "loss": 5.6801,
-      "step": 40827
-    },
-    {
-      "epoch": 21.29230769230769,
-      "grad_norm": 1.5387847423553467,
-      "learning_rate": 5.9484422110552764e-05,
-      "loss": 5.5591,
-      "step": 40828
-    },
-    {
-      "epoch": 21.29282920469361,
-      "grad_norm": 1.406144380569458,
-      "learning_rate": 5.948341708542714e-05,
-      "loss": 5.366,
-      "step": 40829
-    },
-    {
-      "epoch": 21.29335071707953,
-      "grad_norm": 1.406234860420227,
-      "learning_rate": 5.9482412060301506e-05,
-      "loss": 5.1303,
-      "step": 40830
-    },
-    {
-      "epoch": 21.29387222946545,
-      "grad_norm": 1.4289017915725708,
-      "learning_rate": 5.9481407035175884e-05,
-      "loss": 5.1968,
-      "step": 40831
-    },
-    {
-      "epoch": 21.29439374185137,
-      "grad_norm": 1.585181474685669,
-      "learning_rate": 5.948040201005025e-05,
-      "loss": 5.4074,
-      "step": 40832
-    },
-    {
-      "epoch": 21.29491525423729,
-      "grad_norm": 1.6468982696533203,
-      "learning_rate": 5.9479396984924626e-05,
-      "loss": 5.1316,
-      "step": 40833
-    },
-    {
-      "epoch": 21.295436766623208,
-      "grad_norm": 1.5228610038757324,
-      "learning_rate": 5.9478391959799e-05,
-      "loss": 5.5807,
-      "step": 40834
-    },
-    {
-      "epoch": 21.295958279009128,
-      "grad_norm": 1.491750955581665,
-      "learning_rate": 5.9477386934673375e-05,
-      "loss": 5.3337,
-      "step": 40835
-    },
-    {
-      "epoch": 21.296479791395047,
-      "grad_norm": 1.4969621896743774,
-      "learning_rate": 5.947638190954774e-05,
-      "loss": 5.1844,
-      "step": 40836
-    },
-    {
-      "epoch": 21.297001303780966,
-      "grad_norm": 1.5636826753616333,
-      "learning_rate": 5.947537688442212e-05,
-      "loss": 5.4384,
-      "step": 40837
-    },
-    {
-      "epoch": 21.297522816166882,
-      "grad_norm": 1.674810528755188,
-      "learning_rate": 5.947437185929648e-05,
-      "loss": 5.2384,
-      "step": 40838
-    },
-    {
-      "epoch": 21.298044328552802,
-      "grad_norm": 1.5583528280258179,
-      "learning_rate": 5.947336683417085e-05,
-      "loss": 5.404,
-      "step": 40839
-    },
-    {
-      "epoch": 21.29856584093872,
-      "grad_norm": 1.571091890335083,
-      "learning_rate": 5.947236180904523e-05,
-      "loss": 5.2897,
-      "step": 40840
-    },
-    {
-      "epoch": 21.29908735332464,
-      "grad_norm": 1.5122441053390503,
-      "learning_rate": 5.9471356783919595e-05,
-      "loss": 4.8536,
-      "step": 40841
-    },
-    {
-      "epoch": 21.29960886571056,
-      "grad_norm": 1.4769506454467773,
-      "learning_rate": 5.947035175879397e-05,
-      "loss": 5.6909,
-      "step": 40842
-    },
-    {
-      "epoch": 21.30013037809648,
-      "grad_norm": 1.4571459293365479,
-      "learning_rate": 5.9469346733668344e-05,
-      "loss": 5.4572,
-      "step": 40843
-    },
-    {
-      "epoch": 21.3006518904824,
-      "grad_norm": 1.543034315109253,
-      "learning_rate": 5.946834170854272e-05,
-      "loss": 5.3452,
-      "step": 40844
-    },
-    {
-      "epoch": 21.30117340286832,
-      "grad_norm": 1.587611436843872,
-      "learning_rate": 5.9467336683417086e-05,
-      "loss": 5.0673,
-      "step": 40845
-    },
-    {
-      "epoch": 21.301694915254238,
-      "grad_norm": 1.6007832288742065,
-      "learning_rate": 5.9466331658291464e-05,
-      "loss": 5.0559,
-      "step": 40846
-    },
-    {
-      "epoch": 21.302216427640158,
-      "grad_norm": 1.5647269487380981,
-      "learning_rate": 5.946532663316583e-05,
-      "loss": 5.4144,
-      "step": 40847
-    },
-    {
-      "epoch": 21.302737940026077,
-      "grad_norm": 1.5859144926071167,
-      "learning_rate": 5.9464321608040206e-05,
-      "loss": 5.229,
-      "step": 40848
-    },
-    {
-      "epoch": 21.303259452411996,
-      "grad_norm": 1.5060439109802246,
-      "learning_rate": 5.946331658291458e-05,
-      "loss": 5.4049,
-      "step": 40849
-    },
-    {
-      "epoch": 21.303780964797912,
-      "grad_norm": 1.6064549684524536,
-      "learning_rate": 5.9462311557788955e-05,
-      "loss": 5.3078,
-      "step": 40850
-    },
-    {
-      "epoch": 21.304302477183832,
-      "grad_norm": 1.5979022979736328,
-      "learning_rate": 5.946130653266332e-05,
-      "loss": 5.0798,
-      "step": 40851
-    },
-    {
-      "epoch": 21.30482398956975,
-      "grad_norm": 1.4101814031600952,
-      "learning_rate": 5.94603015075377e-05,
-      "loss": 5.1673,
-      "step": 40852
-    },
-    {
-      "epoch": 21.30534550195567,
-      "grad_norm": 1.4803450107574463,
-      "learning_rate": 5.945929648241206e-05,
-      "loss": 5.685,
-      "step": 40853
-    },
-    {
-      "epoch": 21.30586701434159,
-      "grad_norm": 1.4151355028152466,
-      "learning_rate": 5.945829145728643e-05,
-      "loss": 5.6212,
-      "step": 40854
-    },
-    {
-      "epoch": 21.30638852672751,
-      "grad_norm": 1.4481624364852905,
-      "learning_rate": 5.945728643216081e-05,
-      "loss": 5.6481,
-      "step": 40855
-    },
-    {
-      "epoch": 21.30691003911343,
-      "grad_norm": 1.4457807540893555,
-      "learning_rate": 5.9456281407035174e-05,
-      "loss": 5.6049,
-      "step": 40856
-    },
-    {
-      "epoch": 21.30743155149935,
-      "grad_norm": 1.7257094383239746,
-      "learning_rate": 5.945527638190955e-05,
-      "loss": 4.7543,
-      "step": 40857
-    },
-    {
-      "epoch": 21.307953063885268,
-      "grad_norm": 1.5033897161483765,
-      "learning_rate": 5.9454271356783916e-05,
-      "loss": 5.2218,
-      "step": 40858
-    },
-    {
-      "epoch": 21.308474576271188,
-      "grad_norm": 1.4415249824523926,
-      "learning_rate": 5.9453266331658294e-05,
-      "loss": 5.4551,
-      "step": 40859
-    },
-    {
-      "epoch": 21.308996088657107,
-      "grad_norm": 1.6185505390167236,
-      "learning_rate": 5.9452261306532665e-05,
-      "loss": 5.0613,
-      "step": 40860
-    },
-    {
-      "epoch": 21.309517601043027,
-      "grad_norm": 1.5475696325302124,
-      "learning_rate": 5.945125628140704e-05,
-      "loss": 5.5256,
-      "step": 40861
-    },
-    {
-      "epoch": 21.310039113428942,
-      "grad_norm": 1.5712594985961914,
-      "learning_rate": 5.945025125628141e-05,
-      "loss": 4.9915,
-      "step": 40862
-    },
-    {
-      "epoch": 21.310560625814862,
-      "grad_norm": 1.6783169507980347,
-      "learning_rate": 5.9449246231155785e-05,
-      "loss": 5.3569,
-      "step": 40863
-    },
-    {
-      "epoch": 21.31108213820078,
-      "grad_norm": 1.5205625295639038,
-      "learning_rate": 5.9448241206030156e-05,
-      "loss": 4.2032,
-      "step": 40864
-    },
-    {
-      "epoch": 21.3116036505867,
-      "grad_norm": 1.5024465322494507,
-      "learning_rate": 5.9447236180904534e-05,
-      "loss": 5.2259,
-      "step": 40865
-    },
-    {
-      "epoch": 21.31212516297262,
-      "grad_norm": 1.5990606546401978,
-      "learning_rate": 5.94462311557789e-05,
-      "loss": 5.2908,
-      "step": 40866
-    },
-    {
-      "epoch": 21.31264667535854,
-      "grad_norm": 1.7005352973937988,
-      "learning_rate": 5.944522613065326e-05,
-      "loss": 4.8477,
-      "step": 40867
-    },
-    {
-      "epoch": 21.31316818774446,
-      "grad_norm": 1.4687269926071167,
-      "learning_rate": 5.944422110552764e-05,
-      "loss": 4.8357,
-      "step": 40868
-    },
-    {
-      "epoch": 21.31368970013038,
-      "grad_norm": 1.5487514734268188,
-      "learning_rate": 5.944321608040201e-05,
-      "loss": 5.3452,
-      "step": 40869
-    },
-    {
-      "epoch": 21.314211212516298,
-      "grad_norm": 1.5928713083267212,
-      "learning_rate": 5.944221105527639e-05,
-      "loss": 4.9163,
-      "step": 40870
-    },
-    {
-      "epoch": 21.314732724902218,
-      "grad_norm": 1.4816498756408691,
-      "learning_rate": 5.9441206030150753e-05,
-      "loss": 5.5664,
-      "step": 40871
-    },
-    {
-      "epoch": 21.315254237288137,
-      "grad_norm": 1.4892343282699585,
-      "learning_rate": 5.944020100502513e-05,
-      "loss": 5.2443,
-      "step": 40872
-    },
-    {
-      "epoch": 21.315775749674053,
-      "grad_norm": 1.5128345489501953,
-      "learning_rate": 5.9439195979899496e-05,
-      "loss": 5.6021,
-      "step": 40873
-    },
-    {
-      "epoch": 21.316297262059972,
-      "grad_norm": 1.555304765701294,
-      "learning_rate": 5.9438190954773873e-05,
-      "loss": 5.4058,
-      "step": 40874
-    },
-    {
-      "epoch": 21.316818774445892,
-      "grad_norm": 1.5632262229919434,
-      "learning_rate": 5.9437185929648244e-05,
-      "loss": 4.8621,
-      "step": 40875
-    },
-    {
-      "epoch": 21.31734028683181,
-      "grad_norm": 1.4406673908233643,
-      "learning_rate": 5.943618090452262e-05,
-      "loss": 5.6848,
-      "step": 40876
-    },
-    {
-      "epoch": 21.31786179921773,
-      "grad_norm": 1.511650562286377,
-      "learning_rate": 5.9435175879396987e-05,
-      "loss": 5.5256,
-      "step": 40877
-    },
-    {
-      "epoch": 21.31838331160365,
-      "grad_norm": 1.4770852327346802,
-      "learning_rate": 5.9434170854271364e-05,
-      "loss": 5.0751,
-      "step": 40878
-    },
-    {
-      "epoch": 21.31890482398957,
-      "grad_norm": 1.649117350578308,
-      "learning_rate": 5.943316582914573e-05,
-      "loss": 5.2123,
-      "step": 40879
-    },
-    {
-      "epoch": 21.31942633637549,
-      "grad_norm": 1.532670021057129,
-      "learning_rate": 5.94321608040201e-05,
-      "loss": 5.3957,
-      "step": 40880
-    },
-    {
-      "epoch": 21.31994784876141,
-      "grad_norm": 1.4848772287368774,
-      "learning_rate": 5.943115577889448e-05,
-      "loss": 5.0417,
-      "step": 40881
-    },
-    {
-      "epoch": 21.320469361147328,
-      "grad_norm": 1.581044316291809,
-      "learning_rate": 5.943015075376884e-05,
-      "loss": 5.2071,
-      "step": 40882
-    },
-    {
-      "epoch": 21.320990873533248,
-      "grad_norm": 1.6283496618270874,
-      "learning_rate": 5.942914572864322e-05,
-      "loss": 4.9686,
-      "step": 40883
-    },
-    {
-      "epoch": 21.321512385919167,
-      "grad_norm": 1.5617624521255493,
-      "learning_rate": 5.942814070351759e-05,
-      "loss": 4.9491,
-      "step": 40884
-    },
-    {
-      "epoch": 21.322033898305083,
-      "grad_norm": 1.5271189212799072,
-      "learning_rate": 5.942713567839197e-05,
-      "loss": 5.5983,
-      "step": 40885
-    },
-    {
-      "epoch": 21.322555410691002,
-      "grad_norm": 1.740585207939148,
-      "learning_rate": 5.942613065326633e-05,
-      "loss": 4.5544,
-      "step": 40886
-    },
-    {
-      "epoch": 21.323076923076922,
-      "grad_norm": 1.5167793035507202,
-      "learning_rate": 5.942512562814071e-05,
-      "loss": 5.5948,
-      "step": 40887
-    },
-    {
-      "epoch": 21.32359843546284,
-      "grad_norm": 1.4684667587280273,
-      "learning_rate": 5.9424120603015075e-05,
-      "loss": 5.2968,
-      "step": 40888
-    },
-    {
-      "epoch": 21.32411994784876,
-      "grad_norm": 1.5451933145523071,
-      "learning_rate": 5.942311557788945e-05,
-      "loss": 5.3635,
-      "step": 40889
-    },
-    {
-      "epoch": 21.32464146023468,
-      "grad_norm": 1.4969037771224976,
-      "learning_rate": 5.9422110552763824e-05,
-      "loss": 5.3454,
-      "step": 40890
-    },
-    {
-      "epoch": 21.3251629726206,
-      "grad_norm": 1.4507194757461548,
-      "learning_rate": 5.94211055276382e-05,
-      "loss": 4.9199,
-      "step": 40891
-    },
-    {
-      "epoch": 21.32568448500652,
-      "grad_norm": 1.4298386573791504,
-      "learning_rate": 5.9420100502512566e-05,
-      "loss": 5.5899,
-      "step": 40892
-    },
-    {
-      "epoch": 21.32620599739244,
-      "grad_norm": 1.4388476610183716,
-      "learning_rate": 5.941909547738693e-05,
-      "loss": 5.3227,
-      "step": 40893
-    },
-    {
-      "epoch": 21.326727509778358,
-      "grad_norm": 1.6280312538146973,
-      "learning_rate": 5.941809045226131e-05,
-      "loss": 5.4076,
-      "step": 40894
-    },
-    {
-      "epoch": 21.327249022164278,
-      "grad_norm": 1.4186466932296753,
-      "learning_rate": 5.941708542713568e-05,
-      "loss": 5.6743,
-      "step": 40895
-    },
-    {
-      "epoch": 21.327770534550197,
-      "grad_norm": 1.4409352540969849,
-      "learning_rate": 5.941608040201006e-05,
-      "loss": 4.8889,
-      "step": 40896
-    },
-    {
-      "epoch": 21.328292046936113,
-      "grad_norm": 1.5117113590240479,
-      "learning_rate": 5.941507537688442e-05,
-      "loss": 5.5685,
-      "step": 40897
-    },
-    {
-      "epoch": 21.328813559322032,
-      "grad_norm": 1.615553617477417,
-      "learning_rate": 5.94140703517588e-05,
-      "loss": 5.2513,
-      "step": 40898
-    },
-    {
-      "epoch": 21.329335071707952,
-      "grad_norm": 1.4416154623031616,
-      "learning_rate": 5.941306532663316e-05,
-      "loss": 5.7036,
-      "step": 40899
-    },
-    {
-      "epoch": 21.32985658409387,
-      "grad_norm": 1.512376070022583,
-      "learning_rate": 5.941206030150754e-05,
-      "loss": 5.4488,
-      "step": 40900
-    },
-    {
-      "epoch": 21.33037809647979,
-      "grad_norm": 1.5365761518478394,
-      "learning_rate": 5.941105527638191e-05,
-      "loss": 5.0196,
-      "step": 40901
-    },
-    {
-      "epoch": 21.33089960886571,
-      "grad_norm": 1.58588445186615,
-      "learning_rate": 5.941005025125629e-05,
-      "loss": 5.0691,
-      "step": 40902
-    },
-    {
-      "epoch": 21.33142112125163,
-      "grad_norm": 1.6771491765975952,
-      "learning_rate": 5.9409045226130654e-05,
-      "loss": 5.1472,
-      "step": 40903
-    },
-    {
-      "epoch": 21.33194263363755,
-      "grad_norm": 1.4331022500991821,
-      "learning_rate": 5.940804020100503e-05,
-      "loss": 5.3881,
-      "step": 40904
-    },
-    {
-      "epoch": 21.33246414602347,
-      "grad_norm": 1.6036385297775269,
-      "learning_rate": 5.9407035175879396e-05,
-      "loss": 5.1325,
-      "step": 40905
-    },
-    {
-      "epoch": 21.332985658409388,
-      "grad_norm": 1.7574537992477417,
-      "learning_rate": 5.940603015075377e-05,
-      "loss": 5.3321,
-      "step": 40906
-    },
-    {
-      "epoch": 21.333507170795308,
-      "grad_norm": 1.5995945930480957,
-      "learning_rate": 5.9405025125628145e-05,
-      "loss": 4.899,
-      "step": 40907
-    },
-    {
-      "epoch": 21.334028683181227,
-      "grad_norm": 1.6289172172546387,
-      "learning_rate": 5.940402010050251e-05,
-      "loss": 4.5806,
-      "step": 40908
-    },
-    {
-      "epoch": 21.334550195567143,
-      "grad_norm": 1.5385991334915161,
-      "learning_rate": 5.940301507537689e-05,
-      "loss": 5.793,
-      "step": 40909
-    },
-    {
-      "epoch": 21.335071707953063,
-      "grad_norm": 1.5260945558547974,
-      "learning_rate": 5.940201005025126e-05,
-      "loss": 5.7182,
-      "step": 40910
-    },
-    {
-      "epoch": 21.335593220338982,
-      "grad_norm": 1.5184907913208008,
-      "learning_rate": 5.9401005025125636e-05,
-      "loss": 5.3771,
-      "step": 40911
-    },
-    {
-      "epoch": 21.3361147327249,
-      "grad_norm": 1.4863297939300537,
-      "learning_rate": 5.94e-05,
-      "loss": 5.5571,
-      "step": 40912
-    },
-    {
-      "epoch": 21.33663624511082,
-      "grad_norm": 1.484132170677185,
-      "learning_rate": 5.939899497487438e-05,
-      "loss": 5.0135,
-      "step": 40913
-    },
-    {
-      "epoch": 21.33715775749674,
-      "grad_norm": 1.5695160627365112,
-      "learning_rate": 5.939798994974874e-05,
-      "loss": 4.8249,
-      "step": 40914
-    },
-    {
-      "epoch": 21.33767926988266,
-      "grad_norm": 1.5659550428390503,
-      "learning_rate": 5.939698492462312e-05,
-      "loss": 5.3158,
-      "step": 40915
-    },
-    {
-      "epoch": 21.33820078226858,
-      "grad_norm": 1.445351243019104,
-      "learning_rate": 5.939597989949749e-05,
-      "loss": 5.4378,
-      "step": 40916
-    },
-    {
-      "epoch": 21.3387222946545,
-      "grad_norm": 1.3961231708526611,
-      "learning_rate": 5.939497487437187e-05,
-      "loss": 5.5526,
-      "step": 40917
-    },
-    {
-      "epoch": 21.33924380704042,
-      "grad_norm": 1.6317304372787476,
-      "learning_rate": 5.9393969849246234e-05,
-      "loss": 4.9291,
-      "step": 40918
-    },
-    {
-      "epoch": 21.339765319426338,
-      "grad_norm": 1.5887010097503662,
-      "learning_rate": 5.93929648241206e-05,
-      "loss": 5.3342,
-      "step": 40919
-    },
-    {
-      "epoch": 21.340286831812257,
-      "grad_norm": 1.512373447418213,
-      "learning_rate": 5.9391959798994976e-05,
-      "loss": 4.9451,
-      "step": 40920
-    },
-    {
-      "epoch": 21.340808344198173,
-      "grad_norm": 1.599632740020752,
-      "learning_rate": 5.939095477386935e-05,
-      "loss": 4.7479,
-      "step": 40921
-    },
-    {
-      "epoch": 21.341329856584093,
-      "grad_norm": 1.5299012660980225,
-      "learning_rate": 5.9389949748743725e-05,
-      "loss": 4.8165,
-      "step": 40922
-    },
-    {
-      "epoch": 21.341851368970012,
-      "grad_norm": 1.5663541555404663,
-      "learning_rate": 5.938894472361809e-05,
-      "loss": 5.1421,
-      "step": 40923
-    },
-    {
-      "epoch": 21.34237288135593,
-      "grad_norm": 1.5739582777023315,
-      "learning_rate": 5.938793969849247e-05,
-      "loss": 4.9402,
-      "step": 40924
-    },
-    {
-      "epoch": 21.34289439374185,
-      "grad_norm": 1.4704631567001343,
-      "learning_rate": 5.938693467336683e-05,
-      "loss": 5.1615,
-      "step": 40925
-    },
-    {
-      "epoch": 21.34341590612777,
-      "grad_norm": 1.584706425666809,
-      "learning_rate": 5.938592964824121e-05,
-      "loss": 5.3041,
-      "step": 40926
-    },
-    {
-      "epoch": 21.34393741851369,
-      "grad_norm": 1.586729884147644,
-      "learning_rate": 5.938492462311558e-05,
-      "loss": 5.0961,
-      "step": 40927
-    },
-    {
-      "epoch": 21.34445893089961,
-      "grad_norm": 1.4849932193756104,
-      "learning_rate": 5.938391959798996e-05,
-      "loss": 5.3816,
-      "step": 40928
-    },
-    {
-      "epoch": 21.34498044328553,
-      "grad_norm": 1.4750339984893799,
-      "learning_rate": 5.938291457286432e-05,
-      "loss": 4.7413,
-      "step": 40929
-    },
-    {
-      "epoch": 21.34550195567145,
-      "grad_norm": 1.4762884378433228,
-      "learning_rate": 5.93819095477387e-05,
-      "loss": 5.4931,
-      "step": 40930
-    },
-    {
-      "epoch": 21.346023468057368,
-      "grad_norm": 1.5032840967178345,
-      "learning_rate": 5.938090452261307e-05,
-      "loss": 5.1127,
-      "step": 40931
-    },
-    {
-      "epoch": 21.346544980443287,
-      "grad_norm": 1.4938700199127197,
-      "learning_rate": 5.9379899497487435e-05,
-      "loss": 5.6202,
-      "step": 40932
-    },
-    {
-      "epoch": 21.347066492829203,
-      "grad_norm": 1.4965375661849976,
-      "learning_rate": 5.937889447236181e-05,
-      "loss": 5.1272,
-      "step": 40933
-    },
-    {
-      "epoch": 21.347588005215123,
-      "grad_norm": 1.555314540863037,
-      "learning_rate": 5.937788944723618e-05,
-      "loss": 5.4571,
-      "step": 40934
-    },
-    {
-      "epoch": 21.348109517601042,
-      "grad_norm": 1.571028470993042,
-      "learning_rate": 5.9376884422110555e-05,
-      "loss": 5.4981,
-      "step": 40935
-    },
-    {
-      "epoch": 21.34863102998696,
-      "grad_norm": 1.4949557781219482,
-      "learning_rate": 5.9375879396984926e-05,
-      "loss": 5.3821,
-      "step": 40936
-    },
-    {
-      "epoch": 21.34915254237288,
-      "grad_norm": 1.583286166191101,
-      "learning_rate": 5.9374874371859304e-05,
-      "loss": 5.1934,
-      "step": 40937
-    },
-    {
-      "epoch": 21.3496740547588,
-      "grad_norm": 1.495627999305725,
-      "learning_rate": 5.937386934673367e-05,
-      "loss": 5.0903,
-      "step": 40938
-    },
-    {
-      "epoch": 21.35019556714472,
-      "grad_norm": 1.6588165760040283,
-      "learning_rate": 5.9372864321608046e-05,
-      "loss": 4.8647,
-      "step": 40939
-    },
-    {
-      "epoch": 21.35071707953064,
-      "grad_norm": 1.5746252536773682,
-      "learning_rate": 5.937185929648241e-05,
-      "loss": 5.019,
-      "step": 40940
-    },
-    {
-      "epoch": 21.35123859191656,
-      "grad_norm": 1.5444270372390747,
-      "learning_rate": 5.937085427135679e-05,
-      "loss": 5.4598,
-      "step": 40941
-    },
-    {
-      "epoch": 21.35176010430248,
-      "grad_norm": 1.5286359786987305,
-      "learning_rate": 5.936984924623116e-05,
-      "loss": 5.3243,
-      "step": 40942
-    },
-    {
-      "epoch": 21.352281616688398,
-      "grad_norm": 1.552099347114563,
-      "learning_rate": 5.936884422110554e-05,
-      "loss": 5.4579,
-      "step": 40943
-    },
-    {
-      "epoch": 21.352803129074317,
-      "grad_norm": 1.4719291925430298,
-      "learning_rate": 5.93678391959799e-05,
-      "loss": 5.0688,
-      "step": 40944
-    },
-    {
-      "epoch": 21.353324641460233,
-      "grad_norm": 1.4922274351119995,
-      "learning_rate": 5.936683417085428e-05,
-      "loss": 4.3699,
-      "step": 40945
-    },
-    {
-      "epoch": 21.353846153846153,
-      "grad_norm": 1.474350929260254,
-      "learning_rate": 5.936582914572864e-05,
-      "loss": 5.5728,
-      "step": 40946
-    },
-    {
-      "epoch": 21.354367666232072,
-      "grad_norm": 1.575473427772522,
-      "learning_rate": 5.9364824120603014e-05,
-      "loss": 5.5566,
-      "step": 40947
-    },
-    {
-      "epoch": 21.35488917861799,
-      "grad_norm": 1.4146181344985962,
-      "learning_rate": 5.936381909547739e-05,
-      "loss": 5.8061,
-      "step": 40948
-    },
-    {
-      "epoch": 21.35541069100391,
-      "grad_norm": 1.403114914894104,
-      "learning_rate": 5.9362814070351757e-05,
-      "loss": 5.0211,
-      "step": 40949
-    },
-    {
-      "epoch": 21.35593220338983,
-      "grad_norm": 1.4499248266220093,
-      "learning_rate": 5.9361809045226134e-05,
-      "loss": 5.5439,
-      "step": 40950
-    },
-    {
-      "epoch": 21.35645371577575,
-      "grad_norm": 1.723794937133789,
-      "learning_rate": 5.9360804020100505e-05,
-      "loss": 4.7494,
-      "step": 40951
-    },
-    {
-      "epoch": 21.35697522816167,
-      "grad_norm": 1.460012435913086,
-      "learning_rate": 5.935979899497488e-05,
-      "loss": 5.63,
-      "step": 40952
-    },
-    {
-      "epoch": 21.35749674054759,
-      "grad_norm": 1.6020454168319702,
-      "learning_rate": 5.935879396984925e-05,
-      "loss": 5.2717,
-      "step": 40953
-    },
-    {
-      "epoch": 21.35801825293351,
-      "grad_norm": 1.6452161073684692,
-      "learning_rate": 5.9357788944723625e-05,
-      "loss": 5.3466,
-      "step": 40954
-    },
-    {
-      "epoch": 21.358539765319428,
-      "grad_norm": 1.7217261791229248,
-      "learning_rate": 5.935678391959799e-05,
-      "loss": 5.0425,
-      "step": 40955
-    },
-    {
-      "epoch": 21.359061277705347,
-      "grad_norm": 1.4887254238128662,
-      "learning_rate": 5.935577889447237e-05,
-      "loss": 5.2533,
-      "step": 40956
-    },
-    {
-      "epoch": 21.359582790091263,
-      "grad_norm": 1.5125799179077148,
-      "learning_rate": 5.935477386934674e-05,
-      "loss": 5.7371,
-      "step": 40957
-    },
-    {
-      "epoch": 21.360104302477183,
-      "grad_norm": 1.4952157735824585,
-      "learning_rate": 5.9353768844221116e-05,
-      "loss": 5.3257,
-      "step": 40958
-    },
-    {
-      "epoch": 21.360625814863102,
-      "grad_norm": 1.5983624458312988,
-      "learning_rate": 5.935276381909548e-05,
-      "loss": 5.343,
-      "step": 40959
-    },
-    {
-      "epoch": 21.36114732724902,
-      "grad_norm": 1.6501446962356567,
-      "learning_rate": 5.9351758793969845e-05,
-      "loss": 4.8979,
-      "step": 40960
-    },
-    {
-      "epoch": 21.36166883963494,
-      "grad_norm": 1.5375128984451294,
-      "learning_rate": 5.935075376884422e-05,
-      "loss": 5.6032,
-      "step": 40961
-    },
-    {
-      "epoch": 21.36219035202086,
-      "grad_norm": 1.4957034587860107,
-      "learning_rate": 5.9349748743718594e-05,
-      "loss": 5.6672,
-      "step": 40962
-    },
-    {
-      "epoch": 21.36271186440678,
-      "grad_norm": 1.6168785095214844,
-      "learning_rate": 5.934874371859297e-05,
-      "loss": 5.0255,
-      "step": 40963
-    },
-    {
-      "epoch": 21.3632333767927,
-      "grad_norm": 1.4650394916534424,
-      "learning_rate": 5.9347738693467336e-05,
-      "loss": 5.611,
-      "step": 40964
-    },
-    {
-      "epoch": 21.36375488917862,
-      "grad_norm": 1.4465699195861816,
-      "learning_rate": 5.9346733668341714e-05,
-      "loss": 5.5756,
-      "step": 40965
-    },
-    {
-      "epoch": 21.36427640156454,
-      "grad_norm": 1.4711436033248901,
-      "learning_rate": 5.934572864321608e-05,
-      "loss": 5.3214,
-      "step": 40966
-    },
-    {
-      "epoch": 21.364797913950458,
-      "grad_norm": 1.452285647392273,
-      "learning_rate": 5.9344723618090456e-05,
-      "loss": 5.3879,
-      "step": 40967
-    },
-    {
-      "epoch": 21.365319426336377,
-      "grad_norm": 1.4728838205337524,
-      "learning_rate": 5.934371859296483e-05,
-      "loss": 5.107,
-      "step": 40968
-    },
-    {
-      "epoch": 21.365840938722293,
-      "grad_norm": 1.64299738407135,
-      "learning_rate": 5.9342713567839205e-05,
-      "loss": 5.0141,
-      "step": 40969
-    },
-    {
-      "epoch": 21.366362451108213,
-      "grad_norm": 1.570068597793579,
-      "learning_rate": 5.934170854271357e-05,
-      "loss": 5.2134,
-      "step": 40970
-    },
-    {
-      "epoch": 21.366883963494132,
-      "grad_norm": 1.563784122467041,
-      "learning_rate": 5.934070351758795e-05,
-      "loss": 5.4986,
-      "step": 40971
-    },
-    {
-      "epoch": 21.36740547588005,
-      "grad_norm": 1.534156084060669,
-      "learning_rate": 5.933969849246232e-05,
-      "loss": 5.0891,
-      "step": 40972
-    },
-    {
-      "epoch": 21.36792698826597,
-      "grad_norm": 1.6588236093521118,
-      "learning_rate": 5.933869346733668e-05,
-      "loss": 4.5432,
-      "step": 40973
-    },
-    {
-      "epoch": 21.36844850065189,
-      "grad_norm": 1.6133077144622803,
-      "learning_rate": 5.933768844221106e-05,
-      "loss": 4.9522,
-      "step": 40974
-    },
-    {
-      "epoch": 21.36897001303781,
-      "grad_norm": 1.6876251697540283,
-      "learning_rate": 5.9336683417085424e-05,
-      "loss": 4.6777,
-      "step": 40975
-    },
-    {
-      "epoch": 21.36949152542373,
-      "grad_norm": 1.5592303276062012,
-      "learning_rate": 5.93356783919598e-05,
-      "loss": 5.25,
-      "step": 40976
-    },
-    {
-      "epoch": 21.37001303780965,
-      "grad_norm": 1.4913634061813354,
-      "learning_rate": 5.933467336683417e-05,
-      "loss": 5.4922,
-      "step": 40977
-    },
-    {
-      "epoch": 21.37053455019557,
-      "grad_norm": 1.6158907413482666,
-      "learning_rate": 5.933366834170855e-05,
-      "loss": 5.0464,
-      "step": 40978
-    },
-    {
-      "epoch": 21.371056062581488,
-      "grad_norm": 1.4802371263504028,
-      "learning_rate": 5.9332663316582915e-05,
-      "loss": 5.655,
-      "step": 40979
-    },
-    {
-      "epoch": 21.371577574967404,
-      "grad_norm": 1.5686805248260498,
-      "learning_rate": 5.933165829145729e-05,
-      "loss": 5.0281,
-      "step": 40980
-    },
-    {
-      "epoch": 21.372099087353323,
-      "grad_norm": 1.5785152912139893,
-      "learning_rate": 5.933065326633166e-05,
-      "loss": 5.3534,
-      "step": 40981
-    },
-    {
-      "epoch": 21.372620599739243,
-      "grad_norm": 1.6537035703659058,
-      "learning_rate": 5.9329648241206035e-05,
-      "loss": 4.4954,
-      "step": 40982
-    },
-    {
-      "epoch": 21.373142112125162,
-      "grad_norm": 1.52735435962677,
-      "learning_rate": 5.9328643216080406e-05,
-      "loss": 5.5233,
-      "step": 40983
-    },
-    {
-      "epoch": 21.37366362451108,
-      "grad_norm": 1.357649564743042,
-      "learning_rate": 5.9327638190954784e-05,
-      "loss": 5.7687,
-      "step": 40984
-    },
-    {
-      "epoch": 21.374185136897,
-      "grad_norm": 1.472169041633606,
-      "learning_rate": 5.932663316582915e-05,
-      "loss": 5.3882,
-      "step": 40985
-    },
-    {
-      "epoch": 21.37470664928292,
-      "grad_norm": 1.5623241662979126,
-      "learning_rate": 5.932562814070351e-05,
-      "loss": 5.0453,
-      "step": 40986
-    },
-    {
-      "epoch": 21.37522816166884,
-      "grad_norm": 1.563472867012024,
-      "learning_rate": 5.932462311557789e-05,
-      "loss": 5.1863,
-      "step": 40987
-    },
-    {
-      "epoch": 21.37574967405476,
-      "grad_norm": 1.6141726970672607,
-      "learning_rate": 5.932361809045226e-05,
-      "loss": 5.3087,
-      "step": 40988
-    },
-    {
-      "epoch": 21.37627118644068,
-      "grad_norm": 1.5981614589691162,
-      "learning_rate": 5.932261306532664e-05,
-      "loss": 5.1248,
-      "step": 40989
-    },
-    {
-      "epoch": 21.3767926988266,
-      "grad_norm": 1.6629509925842285,
-      "learning_rate": 5.9321608040201004e-05,
-      "loss": 5.3344,
-      "step": 40990
-    },
-    {
-      "epoch": 21.377314211212518,
-      "grad_norm": 1.5006574392318726,
-      "learning_rate": 5.932060301507538e-05,
-      "loss": 4.8448,
-      "step": 40991
-    },
-    {
-      "epoch": 21.377835723598434,
-      "grad_norm": 1.757929801940918,
-      "learning_rate": 5.9319597989949746e-05,
-      "loss": 5.2148,
-      "step": 40992
-    },
-    {
-      "epoch": 21.378357235984353,
-      "grad_norm": 1.7399604320526123,
-      "learning_rate": 5.9318592964824123e-05,
-      "loss": 5.3898,
-      "step": 40993
-    },
-    {
-      "epoch": 21.378878748370273,
-      "grad_norm": 1.5058708190917969,
-      "learning_rate": 5.9317587939698494e-05,
-      "loss": 5.2788,
-      "step": 40994
-    },
-    {
-      "epoch": 21.379400260756192,
-      "grad_norm": 1.5097706317901611,
-      "learning_rate": 5.931658291457287e-05,
-      "loss": 4.9801,
-      "step": 40995
-    },
-    {
-      "epoch": 21.37992177314211,
-      "grad_norm": 1.58722722530365,
-      "learning_rate": 5.9315577889447237e-05,
-      "loss": 5.5879,
-      "step": 40996
-    },
-    {
-      "epoch": 21.38044328552803,
-      "grad_norm": 1.5102382898330688,
-      "learning_rate": 5.9314572864321614e-05,
-      "loss": 4.9375,
-      "step": 40997
-    },
-    {
-      "epoch": 21.38096479791395,
-      "grad_norm": 1.5648524761199951,
-      "learning_rate": 5.9313567839195985e-05,
-      "loss": 5.1801,
-      "step": 40998
-    },
-    {
-      "epoch": 21.38148631029987,
-      "grad_norm": 1.5839422941207886,
-      "learning_rate": 5.931256281407035e-05,
-      "loss": 4.4988,
-      "step": 40999
-    },
-    {
-      "epoch": 21.38200782268579,
-      "grad_norm": 1.633887529373169,
-      "learning_rate": 5.931155778894473e-05,
-      "loss": 5.37,
-      "step": 41000
-    },
-    {
-      "epoch": 21.38252933507171,
-      "grad_norm": 1.614801049232483,
-      "learning_rate": 5.931055276381909e-05,
-      "loss": 5.5887,
-      "step": 41001
-    },
-    {
-      "epoch": 21.38305084745763,
-      "grad_norm": 1.4197735786437988,
-      "learning_rate": 5.930954773869347e-05,
-      "loss": 5.7268,
-      "step": 41002
-    },
-    {
-      "epoch": 21.383572359843548,
-      "grad_norm": 1.737823486328125,
-      "learning_rate": 5.930854271356784e-05,
-      "loss": 4.3521,
-      "step": 41003
-    },
-    {
-      "epoch": 21.384093872229464,
-      "grad_norm": 1.5161893367767334,
-      "learning_rate": 5.930753768844222e-05,
-      "loss": 5.1994,
-      "step": 41004
-    },
-    {
-      "epoch": 21.384615384615383,
-      "grad_norm": 1.593271255493164,
-      "learning_rate": 5.930653266331658e-05,
-      "loss": 5.1088,
-      "step": 41005
-    },
-    {
-      "epoch": 21.385136897001303,
-      "grad_norm": 1.5263408422470093,
-      "learning_rate": 5.930552763819096e-05,
-      "loss": 5.4207,
-      "step": 41006
-    },
-    {
-      "epoch": 21.385658409387222,
-      "grad_norm": 1.4992153644561768,
-      "learning_rate": 5.9304522613065325e-05,
-      "loss": 5.2577,
-      "step": 41007
-    },
-    {
-      "epoch": 21.38617992177314,
-      "grad_norm": 1.6267746686935425,
-      "learning_rate": 5.93035175879397e-05,
-      "loss": 5.3733,
-      "step": 41008
-    },
-    {
-      "epoch": 21.38670143415906,
-      "grad_norm": 1.5455204248428345,
-      "learning_rate": 5.9302512562814074e-05,
-      "loss": 5.4886,
-      "step": 41009
-    },
-    {
-      "epoch": 21.38722294654498,
-      "grad_norm": 1.4201644659042358,
-      "learning_rate": 5.930150753768845e-05,
-      "loss": 5.5605,
-      "step": 41010
-    },
-    {
-      "epoch": 21.3877444589309,
-      "grad_norm": 1.6348481178283691,
-      "learning_rate": 5.9300502512562816e-05,
-      "loss": 4.6502,
-      "step": 41011
-    },
-    {
-      "epoch": 21.38826597131682,
-      "grad_norm": 1.5664865970611572,
-      "learning_rate": 5.929949748743718e-05,
-      "loss": 5.1563,
-      "step": 41012
-    },
-    {
-      "epoch": 21.38878748370274,
-      "grad_norm": 1.602364182472229,
-      "learning_rate": 5.929849246231156e-05,
-      "loss": 4.9187,
-      "step": 41013
-    },
-    {
-      "epoch": 21.38930899608866,
-      "grad_norm": 1.5807092189788818,
-      "learning_rate": 5.929748743718593e-05,
-      "loss": 5.0818,
-      "step": 41014
-    },
-    {
-      "epoch": 21.389830508474578,
-      "grad_norm": 1.5047962665557861,
-      "learning_rate": 5.929648241206031e-05,
-      "loss": 4.9442,
-      "step": 41015
-    },
-    {
-      "epoch": 21.390352020860494,
-      "grad_norm": 1.5893853902816772,
-      "learning_rate": 5.929547738693467e-05,
-      "loss": 5.1936,
-      "step": 41016
-    },
-    {
-      "epoch": 21.390873533246413,
-      "grad_norm": 1.511821985244751,
-      "learning_rate": 5.929447236180905e-05,
-      "loss": 5.561,
-      "step": 41017
-    },
-    {
-      "epoch": 21.391395045632333,
-      "grad_norm": 1.5792101621627808,
-      "learning_rate": 5.929346733668342e-05,
-      "loss": 5.1699,
-      "step": 41018
-    },
-    {
-      "epoch": 21.391916558018252,
-      "grad_norm": 1.629119873046875,
-      "learning_rate": 5.92924623115578e-05,
-      "loss": 4.8822,
-      "step": 41019
-    },
-    {
-      "epoch": 21.39243807040417,
-      "grad_norm": 1.5236436128616333,
-      "learning_rate": 5.929145728643216e-05,
-      "loss": 5.2727,
-      "step": 41020
-    },
-    {
-      "epoch": 21.39295958279009,
-      "grad_norm": 1.5177249908447266,
-      "learning_rate": 5.929045226130654e-05,
-      "loss": 5.7773,
-      "step": 41021
-    },
-    {
-      "epoch": 21.39348109517601,
-      "grad_norm": 1.5594468116760254,
-      "learning_rate": 5.9289447236180904e-05,
-      "loss": 5.3569,
-      "step": 41022
-    },
-    {
-      "epoch": 21.39400260756193,
-      "grad_norm": 1.6227952241897583,
-      "learning_rate": 5.928844221105528e-05,
-      "loss": 5.2662,
-      "step": 41023
-    },
-    {
-      "epoch": 21.39452411994785,
-      "grad_norm": 1.5346007347106934,
-      "learning_rate": 5.928743718592965e-05,
-      "loss": 5.0805,
-      "step": 41024
-    },
-    {
-      "epoch": 21.39504563233377,
-      "grad_norm": 1.4589375257492065,
-      "learning_rate": 5.928643216080402e-05,
-      "loss": 5.4739,
-      "step": 41025
-    },
-    {
-      "epoch": 21.39556714471969,
-      "grad_norm": 1.450156569480896,
-      "learning_rate": 5.9285427135678395e-05,
-      "loss": 5.5263,
-      "step": 41026
-    },
-    {
-      "epoch": 21.396088657105608,
-      "grad_norm": 1.507494330406189,
-      "learning_rate": 5.928442211055276e-05,
-      "loss": 5.6013,
-      "step": 41027
-    },
-    {
-      "epoch": 21.396610169491524,
-      "grad_norm": 1.4497063159942627,
-      "learning_rate": 5.928341708542714e-05,
-      "loss": 5.4827,
-      "step": 41028
-    },
-    {
-      "epoch": 21.397131681877443,
-      "grad_norm": 1.56150221824646,
-      "learning_rate": 5.928241206030151e-05,
-      "loss": 5.373,
-      "step": 41029
-    },
-    {
-      "epoch": 21.397653194263363,
-      "grad_norm": 1.4473379850387573,
-      "learning_rate": 5.9281407035175886e-05,
-      "loss": 5.4087,
-      "step": 41030
-    },
-    {
-      "epoch": 21.398174706649282,
-      "grad_norm": 1.510840654373169,
-      "learning_rate": 5.928040201005025e-05,
-      "loss": 5.1225,
-      "step": 41031
-    },
-    {
-      "epoch": 21.3986962190352,
-      "grad_norm": 1.6419967412948608,
-      "learning_rate": 5.927939698492463e-05,
-      "loss": 5.3574,
-      "step": 41032
-    },
-    {
-      "epoch": 21.39921773142112,
-      "grad_norm": 1.5846710205078125,
-      "learning_rate": 5.927839195979899e-05,
-      "loss": 5.1445,
-      "step": 41033
-    },
-    {
-      "epoch": 21.39973924380704,
-      "grad_norm": 1.5163947343826294,
-      "learning_rate": 5.927738693467337e-05,
-      "loss": 5.1643,
-      "step": 41034
-    },
-    {
-      "epoch": 21.40026075619296,
-      "grad_norm": 1.4968881607055664,
-      "learning_rate": 5.927638190954774e-05,
-      "loss": 5.4778,
-      "step": 41035
-    },
-    {
-      "epoch": 21.40078226857888,
-      "grad_norm": 1.5700329542160034,
-      "learning_rate": 5.927537688442212e-05,
-      "loss": 5.0708,
-      "step": 41036
-    },
-    {
-      "epoch": 21.4013037809648,
-      "grad_norm": 1.6675209999084473,
-      "learning_rate": 5.9274371859296484e-05,
-      "loss": 4.9417,
-      "step": 41037
-    },
-    {
-      "epoch": 21.40182529335072,
-      "grad_norm": 1.5523279905319214,
-      "learning_rate": 5.927336683417086e-05,
-      "loss": 5.3075,
-      "step": 41038
-    },
-    {
-      "epoch": 21.402346805736638,
-      "grad_norm": 1.484291434288025,
-      "learning_rate": 5.927236180904523e-05,
-      "loss": 4.4498,
-      "step": 41039
-    },
-    {
-      "epoch": 21.402868318122554,
-      "grad_norm": 1.5777686834335327,
-      "learning_rate": 5.92713567839196e-05,
-      "loss": 4.9492,
-      "step": 41040
-    },
-    {
-      "epoch": 21.403389830508473,
-      "grad_norm": 1.4950445890426636,
-      "learning_rate": 5.9270351758793975e-05,
-      "loss": 5.5504,
-      "step": 41041
-    },
-    {
-      "epoch": 21.403911342894393,
-      "grad_norm": 1.4731183052062988,
-      "learning_rate": 5.926934673366834e-05,
-      "loss": 5.6877,
-      "step": 41042
-    },
-    {
-      "epoch": 21.404432855280312,
-      "grad_norm": 1.510014533996582,
-      "learning_rate": 5.926834170854272e-05,
-      "loss": 5.124,
-      "step": 41043
-    },
-    {
-      "epoch": 21.40495436766623,
-      "grad_norm": 1.5817887783050537,
-      "learning_rate": 5.926733668341709e-05,
-      "loss": 5.2589,
-      "step": 41044
-    },
-    {
-      "epoch": 21.40547588005215,
-      "grad_norm": 1.5144833326339722,
-      "learning_rate": 5.9266331658291466e-05,
-      "loss": 5.3821,
-      "step": 41045
-    },
-    {
-      "epoch": 21.40599739243807,
-      "grad_norm": 1.4214595556259155,
-      "learning_rate": 5.926532663316583e-05,
-      "loss": 5.3311,
-      "step": 41046
-    },
-    {
-      "epoch": 21.40651890482399,
-      "grad_norm": 1.6286941766738892,
-      "learning_rate": 5.926432160804021e-05,
-      "loss": 5.1227,
-      "step": 41047
-    },
-    {
-      "epoch": 21.40704041720991,
-      "grad_norm": 1.5133235454559326,
-      "learning_rate": 5.926331658291457e-05,
-      "loss": 5.1979,
-      "step": 41048
-    },
-    {
-      "epoch": 21.40756192959583,
-      "grad_norm": 2.3715803623199463,
-      "learning_rate": 5.926231155778895e-05,
-      "loss": 4.8061,
-      "step": 41049
-    },
-    {
-      "epoch": 21.40808344198175,
-      "grad_norm": 1.6216412782669067,
-      "learning_rate": 5.926130653266332e-05,
-      "loss": 5.5482,
-      "step": 41050
-    },
-    {
-      "epoch": 21.408604954367668,
-      "grad_norm": 1.4913182258605957,
-      "learning_rate": 5.92603015075377e-05,
-      "loss": 5.3412,
-      "step": 41051
-    },
-    {
-      "epoch": 21.409126466753584,
-      "grad_norm": 1.6990469694137573,
-      "learning_rate": 5.925929648241206e-05,
-      "loss": 4.9918,
-      "step": 41052
-    },
-    {
-      "epoch": 21.409647979139503,
-      "grad_norm": 1.4792122840881348,
-      "learning_rate": 5.925829145728643e-05,
-      "loss": 5.4069,
-      "step": 41053
-    },
-    {
-      "epoch": 21.410169491525423,
-      "grad_norm": 1.4729372262954712,
-      "learning_rate": 5.9257286432160805e-05,
-      "loss": 5.4093,
-      "step": 41054
-    },
-    {
-      "epoch": 21.410691003911342,
-      "grad_norm": 1.496396780014038,
-      "learning_rate": 5.9256281407035176e-05,
-      "loss": 5.3636,
-      "step": 41055
-    },
-    {
-      "epoch": 21.41121251629726,
-      "grad_norm": 1.4755038022994995,
-      "learning_rate": 5.9255276381909554e-05,
-      "loss": 4.7169,
-      "step": 41056
-    },
-    {
-      "epoch": 21.41173402868318,
-      "grad_norm": 1.4520140886306763,
-      "learning_rate": 5.925427135678392e-05,
-      "loss": 5.4506,
-      "step": 41057
-    },
-    {
-      "epoch": 21.4122555410691,
-      "grad_norm": 1.5533554553985596,
-      "learning_rate": 5.9253266331658296e-05,
-      "loss": 5.0492,
-      "step": 41058
-    },
-    {
-      "epoch": 21.41277705345502,
-      "grad_norm": 1.3886674642562866,
-      "learning_rate": 5.925226130653267e-05,
-      "loss": 5.724,
-      "step": 41059
-    },
-    {
-      "epoch": 21.41329856584094,
-      "grad_norm": 1.717076301574707,
-      "learning_rate": 5.9251256281407045e-05,
-      "loss": 5.2265,
-      "step": 41060
-    },
-    {
-      "epoch": 21.41382007822686,
-      "grad_norm": 1.3786736726760864,
-      "learning_rate": 5.925025125628141e-05,
-      "loss": 4.8147,
-      "step": 41061
-    },
-    {
-      "epoch": 21.41434159061278,
-      "grad_norm": 1.5504847764968872,
-      "learning_rate": 5.924924623115579e-05,
-      "loss": 5.3173,
-      "step": 41062
-    },
-    {
-      "epoch": 21.414863102998694,
-      "grad_norm": 1.4121050834655762,
-      "learning_rate": 5.924824120603015e-05,
-      "loss": 5.3676,
-      "step": 41063
-    },
-    {
-      "epoch": 21.415384615384614,
-      "grad_norm": 1.6170237064361572,
-      "learning_rate": 5.924723618090453e-05,
-      "loss": 4.8045,
-      "step": 41064
-    },
-    {
-      "epoch": 21.415906127770533,
-      "grad_norm": 1.4522677659988403,
-      "learning_rate": 5.92462311557789e-05,
-      "loss": 5.5759,
-      "step": 41065
-    },
-    {
-      "epoch": 21.416427640156453,
-      "grad_norm": 1.5410972833633423,
-      "learning_rate": 5.9245226130653264e-05,
-      "loss": 5.2174,
-      "step": 41066
-    },
-    {
-      "epoch": 21.416949152542372,
-      "grad_norm": 1.5329581499099731,
-      "learning_rate": 5.924422110552764e-05,
-      "loss": 3.7775,
-      "step": 41067
-    },
-    {
-      "epoch": 21.41747066492829,
-      "grad_norm": 1.652663230895996,
-      "learning_rate": 5.9243216080402007e-05,
-      "loss": 4.9956,
-      "step": 41068
-    },
-    {
-      "epoch": 21.41799217731421,
-      "grad_norm": 1.514403223991394,
-      "learning_rate": 5.9242211055276384e-05,
-      "loss": 5.4549,
-      "step": 41069
-    },
-    {
-      "epoch": 21.41851368970013,
-      "grad_norm": 1.7414703369140625,
-      "learning_rate": 5.9241206030150755e-05,
-      "loss": 4.9867,
-      "step": 41070
-    },
-    {
-      "epoch": 21.41903520208605,
-      "grad_norm": 1.6054807901382446,
-      "learning_rate": 5.924020100502513e-05,
-      "loss": 5.0059,
-      "step": 41071
-    },
-    {
-      "epoch": 21.41955671447197,
-      "grad_norm": 1.5161943435668945,
-      "learning_rate": 5.92391959798995e-05,
-      "loss": 5.6421,
-      "step": 41072
-    },
-    {
-      "epoch": 21.42007822685789,
-      "grad_norm": 1.4522037506103516,
-      "learning_rate": 5.9238190954773875e-05,
-      "loss": 5.8762,
-      "step": 41073
-    },
-    {
-      "epoch": 21.42059973924381,
-      "grad_norm": 2.422438144683838,
-      "learning_rate": 5.923718592964824e-05,
-      "loss": 4.9055,
-      "step": 41074
-    },
-    {
-      "epoch": 21.421121251629724,
-      "grad_norm": 1.5230520963668823,
-      "learning_rate": 5.923618090452262e-05,
-      "loss": 4.7994,
-      "step": 41075
-    },
-    {
-      "epoch": 21.421642764015644,
-      "grad_norm": 1.567209005355835,
-      "learning_rate": 5.923517587939699e-05,
-      "loss": 5.1228,
-      "step": 41076
-    },
-    {
-      "epoch": 21.422164276401563,
-      "grad_norm": 1.48659348487854,
-      "learning_rate": 5.9234170854271366e-05,
-      "loss": 5.208,
-      "step": 41077
-    },
-    {
-      "epoch": 21.422685788787483,
-      "grad_norm": 1.6859955787658691,
-      "learning_rate": 5.923316582914573e-05,
-      "loss": 4.7495,
-      "step": 41078
-    },
-    {
-      "epoch": 21.423207301173402,
-      "grad_norm": 1.4938021898269653,
-      "learning_rate": 5.9232160804020095e-05,
-      "loss": 5.488,
-      "step": 41079
-    },
-    {
-      "epoch": 21.423728813559322,
-      "grad_norm": 1.5396283864974976,
-      "learning_rate": 5.923115577889447e-05,
-      "loss": 5.1227,
-      "step": 41080
-    },
-    {
-      "epoch": 21.42425032594524,
-      "grad_norm": 1.5156424045562744,
-      "learning_rate": 5.9230150753768844e-05,
-      "loss": 5.3561,
-      "step": 41081
-    },
-    {
-      "epoch": 21.42477183833116,
-      "grad_norm": 1.6391777992248535,
-      "learning_rate": 5.922914572864322e-05,
-      "loss": 4.9924,
-      "step": 41082
-    },
-    {
-      "epoch": 21.42529335071708,
-      "grad_norm": 1.520200490951538,
-      "learning_rate": 5.9228140703517586e-05,
-      "loss": 5.0437,
-      "step": 41083
-    },
-    {
-      "epoch": 21.425814863103,
-      "grad_norm": 1.6187902688980103,
-      "learning_rate": 5.9227135678391964e-05,
-      "loss": 5.0699,
-      "step": 41084
-    },
-    {
-      "epoch": 21.42633637548892,
-      "grad_norm": 1.7063201665878296,
-      "learning_rate": 5.9226130653266335e-05,
-      "loss": 5.1236,
-      "step": 41085
-    },
-    {
-      "epoch": 21.42685788787484,
-      "grad_norm": 1.551129937171936,
-      "learning_rate": 5.922512562814071e-05,
-      "loss": 5.3837,
-      "step": 41086
-    },
-    {
-      "epoch": 21.427379400260754,
-      "grad_norm": 1.7293967008590698,
-      "learning_rate": 5.922412060301508e-05,
-      "loss": 4.9043,
-      "step": 41087
-    },
-    {
-      "epoch": 21.427900912646674,
-      "grad_norm": 1.5616123676300049,
-      "learning_rate": 5.9223115577889455e-05,
-      "loss": 4.9117,
-      "step": 41088
-    },
-    {
-      "epoch": 21.428422425032593,
-      "grad_norm": 1.9327290058135986,
-      "learning_rate": 5.922211055276382e-05,
-      "loss": 5.4518,
-      "step": 41089
-    },
-    {
-      "epoch": 21.428943937418513,
-      "grad_norm": 1.5790907144546509,
-      "learning_rate": 5.92211055276382e-05,
-      "loss": 4.5076,
-      "step": 41090
-    },
-    {
-      "epoch": 21.429465449804432,
-      "grad_norm": 1.6041349172592163,
-      "learning_rate": 5.922010050251257e-05,
-      "loss": 4.966,
-      "step": 41091
-    },
-    {
-      "epoch": 21.429986962190352,
-      "grad_norm": 1.5374449491500854,
-      "learning_rate": 5.921909547738693e-05,
-      "loss": 4.8887,
-      "step": 41092
-    },
-    {
-      "epoch": 21.43050847457627,
-      "grad_norm": 1.623756766319275,
-      "learning_rate": 5.921809045226131e-05,
-      "loss": 5.4154,
-      "step": 41093
-    },
-    {
-      "epoch": 21.43102998696219,
-      "grad_norm": 1.5386654138565063,
-      "learning_rate": 5.9217085427135674e-05,
-      "loss": 4.9311,
-      "step": 41094
-    },
-    {
-      "epoch": 21.43155149934811,
-      "grad_norm": 1.716391682624817,
-      "learning_rate": 5.921608040201005e-05,
-      "loss": 5.122,
-      "step": 41095
-    },
-    {
-      "epoch": 21.43207301173403,
-      "grad_norm": 1.5140138864517212,
-      "learning_rate": 5.921507537688442e-05,
-      "loss": 4.8581,
-      "step": 41096
-    },
-    {
-      "epoch": 21.43259452411995,
-      "grad_norm": 1.5096015930175781,
-      "learning_rate": 5.92140703517588e-05,
-      "loss": 5.2939,
-      "step": 41097
-    },
-    {
-      "epoch": 21.43311603650587,
-      "grad_norm": 1.6007031202316284,
-      "learning_rate": 5.9213065326633165e-05,
-      "loss": 5.1037,
-      "step": 41098
-    },
-    {
-      "epoch": 21.433637548891785,
-      "grad_norm": 1.56480872631073,
-      "learning_rate": 5.921206030150754e-05,
-      "loss": 5.0266,
-      "step": 41099
-    },
-    {
-      "epoch": 21.434159061277704,
-      "grad_norm": 1.545104742050171,
-      "learning_rate": 5.921105527638191e-05,
-      "loss": 5.3377,
-      "step": 41100
-    },
-    {
-      "epoch": 21.434680573663623,
-      "grad_norm": 1.5894389152526855,
-      "learning_rate": 5.9210050251256285e-05,
-      "loss": 4.8444,
-      "step": 41101
-    },
-    {
-      "epoch": 21.435202086049543,
-      "grad_norm": 1.5818474292755127,
-      "learning_rate": 5.9209045226130656e-05,
-      "loss": 5.2247,
-      "step": 41102
-    },
-    {
-      "epoch": 21.435723598435462,
-      "grad_norm": 1.6359614133834839,
-      "learning_rate": 5.9208040201005034e-05,
-      "loss": 5.1891,
-      "step": 41103
-    },
-    {
-      "epoch": 21.436245110821382,
-      "grad_norm": 1.4823541641235352,
-      "learning_rate": 5.92070351758794e-05,
-      "loss": 4.9077,
-      "step": 41104
-    },
-    {
-      "epoch": 21.4367666232073,
-      "grad_norm": 1.4793648719787598,
-      "learning_rate": 5.920603015075377e-05,
-      "loss": 5.2415,
-      "step": 41105
-    },
-    {
-      "epoch": 21.43728813559322,
-      "grad_norm": 1.5936557054519653,
-      "learning_rate": 5.920502512562815e-05,
-      "loss": 5.1797,
-      "step": 41106
-    },
-    {
-      "epoch": 21.43780964797914,
-      "grad_norm": 1.4779069423675537,
-      "learning_rate": 5.920402010050251e-05,
-      "loss": 5.1808,
-      "step": 41107
-    },
-    {
-      "epoch": 21.43833116036506,
-      "grad_norm": 1.5176036357879639,
-      "learning_rate": 5.920301507537689e-05,
-      "loss": 5.0645,
-      "step": 41108
-    },
-    {
-      "epoch": 21.43885267275098,
-      "grad_norm": 1.4553302526474,
-      "learning_rate": 5.9202010050251254e-05,
-      "loss": 4.9093,
-      "step": 41109
-    },
-    {
-      "epoch": 21.4393741851369,
-      "grad_norm": 1.455739140510559,
-      "learning_rate": 5.920100502512563e-05,
-      "loss": 5.5781,
-      "step": 41110
-    },
-    {
-      "epoch": 21.439895697522815,
-      "grad_norm": 1.4968539476394653,
-      "learning_rate": 5.92e-05,
-      "loss": 5.5818,
-      "step": 41111
-    },
-    {
-      "epoch": 21.440417209908734,
-      "grad_norm": 1.5650750398635864,
-      "learning_rate": 5.919899497487438e-05,
-      "loss": 5.0588,
-      "step": 41112
-    },
-    {
-      "epoch": 21.440938722294653,
-      "grad_norm": 1.4428352117538452,
-      "learning_rate": 5.9197989949748745e-05,
-      "loss": 5.3827,
-      "step": 41113
-    },
-    {
-      "epoch": 21.441460234680573,
-      "grad_norm": 1.5384511947631836,
-      "learning_rate": 5.919698492462312e-05,
-      "loss": 5.256,
-      "step": 41114
-    },
-    {
-      "epoch": 21.441981747066492,
-      "grad_norm": 1.6068804264068604,
-      "learning_rate": 5.9195979899497487e-05,
-      "loss": 5.3996,
-      "step": 41115
-    },
-    {
-      "epoch": 21.442503259452412,
-      "grad_norm": 1.5617308616638184,
-      "learning_rate": 5.9194974874371864e-05,
-      "loss": 4.982,
-      "step": 41116
-    },
-    {
-      "epoch": 21.44302477183833,
-      "grad_norm": 1.4507203102111816,
-      "learning_rate": 5.9193969849246235e-05,
-      "loss": 5.2052,
-      "step": 41117
-    },
-    {
-      "epoch": 21.44354628422425,
-      "grad_norm": 1.5799821615219116,
-      "learning_rate": 5.919296482412061e-05,
-      "loss": 4.9293,
-      "step": 41118
-    },
-    {
-      "epoch": 21.44406779661017,
-      "grad_norm": 1.4922505617141724,
-      "learning_rate": 5.919195979899498e-05,
-      "loss": 5.1822,
-      "step": 41119
-    },
-    {
-      "epoch": 21.44458930899609,
-      "grad_norm": 1.4315698146820068,
-      "learning_rate": 5.919095477386934e-05,
-      "loss": 5.3661,
-      "step": 41120
-    },
-    {
-      "epoch": 21.44511082138201,
-      "grad_norm": 1.5785952806472778,
-      "learning_rate": 5.918994974874372e-05,
-      "loss": 5.0197,
-      "step": 41121
-    },
-    {
-      "epoch": 21.44563233376793,
-      "grad_norm": 1.4807422161102295,
-      "learning_rate": 5.918894472361809e-05,
-      "loss": 5.5114,
-      "step": 41122
-    },
-    {
-      "epoch": 21.446153846153845,
-      "grad_norm": 1.4588969945907593,
-      "learning_rate": 5.918793969849247e-05,
-      "loss": 5.4579,
-      "step": 41123
-    },
-    {
-      "epoch": 21.446675358539764,
-      "grad_norm": 1.5385782718658447,
-      "learning_rate": 5.918693467336683e-05,
-      "loss": 5.071,
-      "step": 41124
-    },
-    {
-      "epoch": 21.447196870925683,
-      "grad_norm": 1.5653952360153198,
-      "learning_rate": 5.918592964824121e-05,
-      "loss": 4.7828,
-      "step": 41125
-    },
-    {
-      "epoch": 21.447718383311603,
-      "grad_norm": 1.605215072631836,
-      "learning_rate": 5.918492462311558e-05,
-      "loss": 5.1829,
-      "step": 41126
-    },
-    {
-      "epoch": 21.448239895697522,
-      "grad_norm": 1.592134714126587,
-      "learning_rate": 5.918391959798996e-05,
-      "loss": 4.7724,
-      "step": 41127
-    },
-    {
-      "epoch": 21.448761408083442,
-      "grad_norm": 1.565037727355957,
-      "learning_rate": 5.9182914572864324e-05,
-      "loss": 5.0003,
-      "step": 41128
-    },
-    {
-      "epoch": 21.44928292046936,
-      "grad_norm": 1.5050938129425049,
-      "learning_rate": 5.91819095477387e-05,
-      "loss": 5.1437,
-      "step": 41129
-    },
-    {
-      "epoch": 21.44980443285528,
-      "grad_norm": 1.5295896530151367,
-      "learning_rate": 5.9180904522613066e-05,
-      "loss": 4.923,
-      "step": 41130
-    },
-    {
-      "epoch": 21.4503259452412,
-      "grad_norm": 1.5047335624694824,
-      "learning_rate": 5.9179899497487444e-05,
-      "loss": 5.7705,
-      "step": 41131
-    },
-    {
-      "epoch": 21.45084745762712,
-      "grad_norm": 1.5995969772338867,
-      "learning_rate": 5.9178894472361815e-05,
-      "loss": 5.0854,
-      "step": 41132
-    },
-    {
-      "epoch": 21.45136897001304,
-      "grad_norm": 1.4608933925628662,
-      "learning_rate": 5.917788944723618e-05,
-      "loss": 4.8073,
-      "step": 41133
-    },
-    {
-      "epoch": 21.45189048239896,
-      "grad_norm": 1.5143768787384033,
-      "learning_rate": 5.917688442211056e-05,
-      "loss": 5.3879,
-      "step": 41134
-    },
-    {
-      "epoch": 21.452411994784875,
-      "grad_norm": 1.5135085582733154,
-      "learning_rate": 5.917587939698492e-05,
-      "loss": 5.0996,
-      "step": 41135
-    },
-    {
-      "epoch": 21.452933507170794,
-      "grad_norm": 1.4530612230300903,
-      "learning_rate": 5.91748743718593e-05,
-      "loss": 4.8794,
-      "step": 41136
-    },
-    {
-      "epoch": 21.453455019556714,
-      "grad_norm": 1.4465845823287964,
-      "learning_rate": 5.917386934673367e-05,
-      "loss": 5.2073,
-      "step": 41137
-    },
-    {
-      "epoch": 21.453976531942633,
-      "grad_norm": 1.470965027809143,
-      "learning_rate": 5.917286432160805e-05,
-      "loss": 4.5187,
-      "step": 41138
-    },
-    {
-      "epoch": 21.454498044328552,
-      "grad_norm": 1.5436632633209229,
-      "learning_rate": 5.917185929648241e-05,
-      "loss": 5.2408,
-      "step": 41139
-    },
-    {
-      "epoch": 21.455019556714472,
-      "grad_norm": 1.5198332071304321,
-      "learning_rate": 5.917085427135679e-05,
-      "loss": 5.4214,
-      "step": 41140
-    },
-    {
-      "epoch": 21.45554106910039,
-      "grad_norm": 1.4951589107513428,
-      "learning_rate": 5.9169849246231154e-05,
-      "loss": 5.3258,
-      "step": 41141
-    },
-    {
-      "epoch": 21.45606258148631,
-      "grad_norm": 1.559004783630371,
-      "learning_rate": 5.916884422110553e-05,
-      "loss": 4.5089,
-      "step": 41142
-    },
-    {
-      "epoch": 21.45658409387223,
-      "grad_norm": 1.5241467952728271,
-      "learning_rate": 5.91678391959799e-05,
-      "loss": 5.4665,
-      "step": 41143
-    },
-    {
-      "epoch": 21.45710560625815,
-      "grad_norm": 1.4609414339065552,
-      "learning_rate": 5.916683417085428e-05,
-      "loss": 5.4369,
-      "step": 41144
-    },
-    {
-      "epoch": 21.45762711864407,
-      "grad_norm": 1.4938504695892334,
-      "learning_rate": 5.9165829145728645e-05,
-      "loss": 5.3748,
-      "step": 41145
-    },
-    {
-      "epoch": 21.45814863102999,
-      "grad_norm": 1.5268255472183228,
-      "learning_rate": 5.9164824120603016e-05,
-      "loss": 5.405,
-      "step": 41146
-    },
-    {
-      "epoch": 21.458670143415905,
-      "grad_norm": 1.5430594682693481,
-      "learning_rate": 5.9163819095477394e-05,
-      "loss": 5.4704,
-      "step": 41147
-    },
-    {
-      "epoch": 21.459191655801824,
-      "grad_norm": 1.4992681741714478,
-      "learning_rate": 5.916281407035176e-05,
-      "loss": 5.1042,
-      "step": 41148
-    },
-    {
-      "epoch": 21.459713168187744,
-      "grad_norm": 1.674988865852356,
-      "learning_rate": 5.9161809045226136e-05,
-      "loss": 5.0209,
-      "step": 41149
-    },
-    {
-      "epoch": 21.460234680573663,
-      "grad_norm": 1.5126831531524658,
-      "learning_rate": 5.91608040201005e-05,
-      "loss": 5.5444,
-      "step": 41150
-    },
-    {
-      "epoch": 21.460756192959582,
-      "grad_norm": 1.5453957319259644,
-      "learning_rate": 5.915979899497488e-05,
-      "loss": 4.7466,
-      "step": 41151
-    },
-    {
-      "epoch": 21.461277705345502,
-      "grad_norm": 1.556870460510254,
-      "learning_rate": 5.915879396984925e-05,
-      "loss": 5.3538,
-      "step": 41152
-    },
-    {
-      "epoch": 21.46179921773142,
-      "grad_norm": 1.5816092491149902,
-      "learning_rate": 5.915778894472363e-05,
-      "loss": 5.5292,
-      "step": 41153
-    },
-    {
-      "epoch": 21.46232073011734,
-      "grad_norm": 1.503212809562683,
-      "learning_rate": 5.915678391959799e-05,
-      "loss": 5.5807,
-      "step": 41154
-    },
-    {
-      "epoch": 21.46284224250326,
-      "grad_norm": 1.5021730661392212,
-      "learning_rate": 5.915577889447237e-05,
-      "loss": 5.4327,
-      "step": 41155
-    },
-    {
-      "epoch": 21.46336375488918,
-      "grad_norm": 1.5607396364212036,
-      "learning_rate": 5.9154773869346734e-05,
-      "loss": 5.505,
-      "step": 41156
-    },
-    {
-      "epoch": 21.4638852672751,
-      "grad_norm": 1.5450488328933716,
-      "learning_rate": 5.915376884422111e-05,
-      "loss": 5.0243,
-      "step": 41157
-    },
-    {
-      "epoch": 21.46440677966102,
-      "grad_norm": 1.5298491716384888,
-      "learning_rate": 5.915276381909548e-05,
-      "loss": 5.6584,
-      "step": 41158
-    },
-    {
-      "epoch": 21.464928292046935,
-      "grad_norm": 1.5874018669128418,
-      "learning_rate": 5.915175879396985e-05,
-      "loss": 5.5467,
-      "step": 41159
-    },
-    {
-      "epoch": 21.465449804432854,
-      "grad_norm": 1.4667235612869263,
-      "learning_rate": 5.9150753768844225e-05,
-      "loss": 5.333,
-      "step": 41160
-    },
-    {
-      "epoch": 21.465971316818774,
-      "grad_norm": 1.4391800165176392,
-      "learning_rate": 5.914974874371859e-05,
-      "loss": 4.4899,
-      "step": 41161
-    },
-    {
-      "epoch": 21.466492829204693,
-      "grad_norm": 1.5391337871551514,
-      "learning_rate": 5.914874371859297e-05,
-      "loss": 4.8772,
-      "step": 41162
-    },
-    {
-      "epoch": 21.467014341590613,
-      "grad_norm": 1.576309323310852,
-      "learning_rate": 5.914773869346734e-05,
-      "loss": 5.2822,
-      "step": 41163
-    },
-    {
-      "epoch": 21.467535853976532,
-      "grad_norm": 1.5789377689361572,
-      "learning_rate": 5.9146733668341716e-05,
-      "loss": 5.1393,
-      "step": 41164
-    },
-    {
-      "epoch": 21.46805736636245,
-      "grad_norm": 1.4880632162094116,
-      "learning_rate": 5.914572864321608e-05,
-      "loss": 5.1259,
-      "step": 41165
-    },
-    {
-      "epoch": 21.46857887874837,
-      "grad_norm": 1.5775562524795532,
-      "learning_rate": 5.914472361809046e-05,
-      "loss": 5.0851,
-      "step": 41166
-    },
-    {
-      "epoch": 21.46910039113429,
-      "grad_norm": 1.5226553678512573,
-      "learning_rate": 5.914371859296482e-05,
-      "loss": 5.3097,
-      "step": 41167
-    },
-    {
-      "epoch": 21.46962190352021,
-      "grad_norm": 1.601026177406311,
-      "learning_rate": 5.91427135678392e-05,
-      "loss": 4.9864,
-      "step": 41168
-    },
-    {
-      "epoch": 21.47014341590613,
-      "grad_norm": 1.5087001323699951,
-      "learning_rate": 5.914170854271357e-05,
-      "loss": 5.3757,
-      "step": 41169
-    },
-    {
-      "epoch": 21.470664928292045,
-      "grad_norm": 1.5159602165222168,
-      "learning_rate": 5.914070351758795e-05,
-      "loss": 5.4397,
-      "step": 41170
-    },
-    {
-      "epoch": 21.471186440677965,
-      "grad_norm": 1.506822109222412,
-      "learning_rate": 5.913969849246231e-05,
-      "loss": 5.7466,
-      "step": 41171
-    },
-    {
-      "epoch": 21.471707953063884,
-      "grad_norm": 1.566779375076294,
-      "learning_rate": 5.9138693467336684e-05,
-      "loss": 5.6096,
-      "step": 41172
-    },
-    {
-      "epoch": 21.472229465449804,
-      "grad_norm": 1.5106900930404663,
-      "learning_rate": 5.913768844221106e-05,
-      "loss": 5.332,
-      "step": 41173
-    },
-    {
-      "epoch": 21.472750977835723,
-      "grad_norm": 1.556606411933899,
-      "learning_rate": 5.9136683417085426e-05,
-      "loss": 5.6151,
-      "step": 41174
-    },
-    {
-      "epoch": 21.473272490221643,
-      "grad_norm": 1.449682354927063,
-      "learning_rate": 5.9135678391959804e-05,
-      "loss": 4.7145,
-      "step": 41175
-    },
-    {
-      "epoch": 21.473794002607562,
-      "grad_norm": 1.4287097454071045,
-      "learning_rate": 5.913467336683417e-05,
-      "loss": 5.5583,
-      "step": 41176
-    },
-    {
-      "epoch": 21.47431551499348,
-      "grad_norm": 1.6095407009124756,
-      "learning_rate": 5.9133668341708546e-05,
-      "loss": 5.0712,
-      "step": 41177
-    },
-    {
-      "epoch": 21.4748370273794,
-      "grad_norm": 1.4746060371398926,
-      "learning_rate": 5.913266331658292e-05,
-      "loss": 5.2952,
-      "step": 41178
-    },
-    {
-      "epoch": 21.47535853976532,
-      "grad_norm": 1.5001964569091797,
-      "learning_rate": 5.9131658291457295e-05,
-      "loss": 5.2824,
-      "step": 41179
-    },
-    {
-      "epoch": 21.47588005215124,
-      "grad_norm": 1.5815166234970093,
-      "learning_rate": 5.913065326633166e-05,
-      "loss": 4.8536,
-      "step": 41180
-    },
-    {
-      "epoch": 21.47640156453716,
-      "grad_norm": 1.5653260946273804,
-      "learning_rate": 5.912964824120604e-05,
-      "loss": 4.7497,
-      "step": 41181
-    },
-    {
-      "epoch": 21.476923076923075,
-      "grad_norm": 1.4771090745925903,
-      "learning_rate": 5.91286432160804e-05,
-      "loss": 5.5982,
-      "step": 41182
-    },
-    {
-      "epoch": 21.477444589308995,
-      "grad_norm": 1.4325979948043823,
-      "learning_rate": 5.912763819095478e-05,
-      "loss": 5.4012,
-      "step": 41183
-    },
-    {
-      "epoch": 21.477966101694914,
-      "grad_norm": 1.557810664176941,
-      "learning_rate": 5.912663316582915e-05,
-      "loss": 5.3068,
-      "step": 41184
-    },
-    {
-      "epoch": 21.478487614080834,
-      "grad_norm": 1.3967316150665283,
-      "learning_rate": 5.9125628140703514e-05,
-      "loss": 5.626,
-      "step": 41185
-    },
-    {
-      "epoch": 21.479009126466753,
-      "grad_norm": 1.5346946716308594,
-      "learning_rate": 5.912462311557789e-05,
-      "loss": 5.5987,
-      "step": 41186
-    },
-    {
-      "epoch": 21.479530638852673,
-      "grad_norm": 1.6300305128097534,
-      "learning_rate": 5.9123618090452257e-05,
-      "loss": 4.9651,
-      "step": 41187
-    },
-    {
-      "epoch": 21.480052151238592,
-      "grad_norm": 1.4661213159561157,
-      "learning_rate": 5.9122613065326634e-05,
-      "loss": 5.3466,
-      "step": 41188
-    },
-    {
-      "epoch": 21.48057366362451,
-      "grad_norm": 1.6083879470825195,
-      "learning_rate": 5.9121608040201005e-05,
-      "loss": 4.7983,
-      "step": 41189
-    },
-    {
-      "epoch": 21.48109517601043,
-      "grad_norm": 1.4596834182739258,
-      "learning_rate": 5.912060301507538e-05,
-      "loss": 5.6572,
-      "step": 41190
-    },
-    {
-      "epoch": 21.48161668839635,
-      "grad_norm": 1.563409686088562,
-      "learning_rate": 5.911959798994975e-05,
-      "loss": 5.0891,
-      "step": 41191
-    },
-    {
-      "epoch": 21.48213820078227,
-      "grad_norm": 1.56592857837677,
-      "learning_rate": 5.9118592964824125e-05,
-      "loss": 5.3416,
-      "step": 41192
-    },
-    {
-      "epoch": 21.48265971316819,
-      "grad_norm": 1.5389330387115479,
-      "learning_rate": 5.9117587939698496e-05,
-      "loss": 4.9226,
-      "step": 41193
-    },
-    {
-      "epoch": 21.483181225554105,
-      "grad_norm": 1.5529509782791138,
-      "learning_rate": 5.9116582914572874e-05,
-      "loss": 4.3574,
-      "step": 41194
-    },
-    {
-      "epoch": 21.483702737940025,
-      "grad_norm": 1.624025583267212,
-      "learning_rate": 5.911557788944724e-05,
-      "loss": 5.0792,
-      "step": 41195
-    },
-    {
-      "epoch": 21.484224250325944,
-      "grad_norm": 1.4370590448379517,
-      "learning_rate": 5.9114572864321616e-05,
-      "loss": 5.7706,
-      "step": 41196
-    },
-    {
-      "epoch": 21.484745762711864,
-      "grad_norm": 1.5869472026824951,
-      "learning_rate": 5.911356783919598e-05,
-      "loss": 4.9356,
-      "step": 41197
-    },
-    {
-      "epoch": 21.485267275097783,
-      "grad_norm": 1.5522187948226929,
-      "learning_rate": 5.911256281407035e-05,
-      "loss": 5.1373,
-      "step": 41198
-    },
-    {
-      "epoch": 21.485788787483703,
-      "grad_norm": 1.5030452013015747,
-      "learning_rate": 5.911155778894473e-05,
-      "loss": 5.3523,
-      "step": 41199
-    },
-    {
-      "epoch": 21.486310299869622,
-      "grad_norm": 1.514507532119751,
-      "learning_rate": 5.9110552763819094e-05,
-      "loss": 5.1713,
-      "step": 41200
-    },
-    {
-      "epoch": 21.48683181225554,
-      "grad_norm": 1.3508753776550293,
-      "learning_rate": 5.910954773869347e-05,
-      "loss": 5.8561,
-      "step": 41201
-    },
-    {
-      "epoch": 21.48735332464146,
-      "grad_norm": 1.5350942611694336,
-      "learning_rate": 5.9108542713567836e-05,
-      "loss": 5.1713,
-      "step": 41202
-    },
-    {
-      "epoch": 21.48787483702738,
-      "grad_norm": 1.5217972993850708,
-      "learning_rate": 5.9107537688442214e-05,
-      "loss": 5.3972,
-      "step": 41203
-    },
-    {
-      "epoch": 21.4883963494133,
-      "grad_norm": 1.617126703262329,
-      "learning_rate": 5.9106532663316585e-05,
-      "loss": 4.8247,
-      "step": 41204
-    },
-    {
-      "epoch": 21.48891786179922,
-      "grad_norm": 1.5497868061065674,
-      "learning_rate": 5.910552763819096e-05,
-      "loss": 5.3665,
-      "step": 41205
-    },
-    {
-      "epoch": 21.489439374185135,
-      "grad_norm": 1.6398006677627563,
-      "learning_rate": 5.910452261306533e-05,
-      "loss": 4.9331,
-      "step": 41206
-    },
-    {
-      "epoch": 21.489960886571055,
-      "grad_norm": 1.4751466512680054,
-      "learning_rate": 5.9103517587939705e-05,
-      "loss": 5.6548,
-      "step": 41207
-    },
-    {
-      "epoch": 21.490482398956974,
-      "grad_norm": 1.5717668533325195,
-      "learning_rate": 5.910251256281407e-05,
-      "loss": 4.7295,
-      "step": 41208
-    },
-    {
-      "epoch": 21.491003911342894,
-      "grad_norm": 1.566636562347412,
-      "learning_rate": 5.910150753768845e-05,
-      "loss": 4.6615,
-      "step": 41209
-    },
-    {
-      "epoch": 21.491525423728813,
-      "grad_norm": 1.6807433366775513,
-      "learning_rate": 5.910050251256282e-05,
-      "loss": 5.0344,
-      "step": 41210
-    },
-    {
-      "epoch": 21.492046936114733,
-      "grad_norm": 1.6894841194152832,
-      "learning_rate": 5.9099497487437196e-05,
-      "loss": 5.3511,
-      "step": 41211
-    },
-    {
-      "epoch": 21.492568448500652,
-      "grad_norm": 1.6384341716766357,
-      "learning_rate": 5.909849246231156e-05,
-      "loss": 5.4068,
-      "step": 41212
-    },
-    {
-      "epoch": 21.49308996088657,
-      "grad_norm": 1.5998798608779907,
-      "learning_rate": 5.909748743718593e-05,
-      "loss": 5.6401,
-      "step": 41213
-    },
-    {
-      "epoch": 21.49361147327249,
-      "grad_norm": 1.5968462228775024,
-      "learning_rate": 5.909648241206031e-05,
-      "loss": 5.5511,
-      "step": 41214
-    },
-    {
-      "epoch": 21.49413298565841,
-      "grad_norm": 1.611171007156372,
-      "learning_rate": 5.909547738693467e-05,
-      "loss": 5.2913,
-      "step": 41215
-    },
-    {
-      "epoch": 21.49465449804433,
-      "grad_norm": 1.4959501028060913,
-      "learning_rate": 5.909447236180905e-05,
-      "loss": 4.6228,
-      "step": 41216
-    },
-    {
-      "epoch": 21.49517601043025,
-      "grad_norm": 1.6977726221084595,
-      "learning_rate": 5.9093467336683415e-05,
-      "loss": 4.667,
-      "step": 41217
-    },
-    {
-      "epoch": 21.495697522816165,
-      "grad_norm": 1.6247860193252563,
-      "learning_rate": 5.909246231155779e-05,
-      "loss": 5.254,
-      "step": 41218
-    },
-    {
-      "epoch": 21.496219035202085,
-      "grad_norm": 1.4898165464401245,
-      "learning_rate": 5.9091457286432164e-05,
-      "loss": 5.5982,
-      "step": 41219
-    },
-    {
-      "epoch": 21.496740547588004,
-      "grad_norm": 1.4868561029434204,
-      "learning_rate": 5.909045226130654e-05,
-      "loss": 5.6851,
-      "step": 41220
-    },
-    {
-      "epoch": 21.497262059973924,
-      "grad_norm": 1.432295322418213,
-      "learning_rate": 5.9089447236180906e-05,
-      "loss": 5.3737,
-      "step": 41221
-    },
-    {
-      "epoch": 21.497783572359843,
-      "grad_norm": 1.6012927293777466,
-      "learning_rate": 5.9088442211055284e-05,
-      "loss": 5.3639,
-      "step": 41222
-    },
-    {
-      "epoch": 21.498305084745763,
-      "grad_norm": 1.5758346319198608,
-      "learning_rate": 5.908743718592965e-05,
-      "loss": 5.6594,
-      "step": 41223
-    },
-    {
-      "epoch": 21.498826597131682,
-      "grad_norm": 1.4600738286972046,
-      "learning_rate": 5.9086432160804026e-05,
-      "loss": 5.4422,
-      "step": 41224
-    },
-    {
-      "epoch": 21.4993481095176,
-      "grad_norm": 1.4165698289871216,
-      "learning_rate": 5.90854271356784e-05,
-      "loss": 5.2058,
-      "step": 41225
-    },
-    {
-      "epoch": 21.49986962190352,
-      "grad_norm": 1.6297355890274048,
-      "learning_rate": 5.908442211055276e-05,
-      "loss": 4.8911,
-      "step": 41226
-    },
-    {
-      "epoch": 21.50039113428944,
-      "grad_norm": 1.505671501159668,
-      "learning_rate": 5.908341708542714e-05,
-      "loss": 5.4934,
-      "step": 41227
-    },
-    {
-      "epoch": 21.50091264667536,
-      "grad_norm": 1.6158794164657593,
-      "learning_rate": 5.9082412060301504e-05,
-      "loss": 5.009,
-      "step": 41228
-    },
-    {
-      "epoch": 21.50143415906128,
-      "grad_norm": 1.6683810949325562,
-      "learning_rate": 5.908140703517588e-05,
-      "loss": 5.3823,
-      "step": 41229
-    },
-    {
-      "epoch": 21.501955671447195,
-      "grad_norm": 1.4975764751434326,
-      "learning_rate": 5.908040201005025e-05,
-      "loss": 5.5457,
-      "step": 41230
-    },
-    {
-      "epoch": 21.502477183833115,
-      "grad_norm": 1.5996835231781006,
-      "learning_rate": 5.907939698492463e-05,
-      "loss": 5.4359,
-      "step": 41231
-    },
-    {
-      "epoch": 21.502998696219034,
-      "grad_norm": 1.4670857191085815,
-      "learning_rate": 5.9078391959798995e-05,
-      "loss": 5.7115,
-      "step": 41232
-    },
-    {
-      "epoch": 21.503520208604954,
-      "grad_norm": 1.6114745140075684,
-      "learning_rate": 5.907738693467337e-05,
-      "loss": 4.9153,
-      "step": 41233
-    },
-    {
-      "epoch": 21.504041720990873,
-      "grad_norm": 1.4757940769195557,
-      "learning_rate": 5.907638190954774e-05,
-      "loss": 4.8792,
-      "step": 41234
-    },
-    {
-      "epoch": 21.504563233376793,
-      "grad_norm": 1.5152256488800049,
-      "learning_rate": 5.9075376884422114e-05,
-      "loss": 4.943,
-      "step": 41235
-    },
-    {
-      "epoch": 21.505084745762712,
-      "grad_norm": 1.5824360847473145,
-      "learning_rate": 5.9074371859296486e-05,
-      "loss": 4.2691,
-      "step": 41236
-    },
-    {
-      "epoch": 21.50560625814863,
-      "grad_norm": 1.4625403881072998,
-      "learning_rate": 5.907336683417086e-05,
-      "loss": 5.1327,
-      "step": 41237
-    },
-    {
-      "epoch": 21.50612777053455,
-      "grad_norm": 1.5414425134658813,
-      "learning_rate": 5.907236180904523e-05,
-      "loss": 5.2404,
-      "step": 41238
-    },
-    {
-      "epoch": 21.50664928292047,
-      "grad_norm": 1.5251132249832153,
-      "learning_rate": 5.90713567839196e-05,
-      "loss": 5.0439,
-      "step": 41239
-    },
-    {
-      "epoch": 21.50717079530639,
-      "grad_norm": 1.4669935703277588,
-      "learning_rate": 5.9070351758793976e-05,
-      "loss": 5.3542,
-      "step": 41240
-    },
-    {
-      "epoch": 21.50769230769231,
-      "grad_norm": 1.4769057035446167,
-      "learning_rate": 5.906934673366834e-05,
-      "loss": 5.6058,
-      "step": 41241
-    },
-    {
-      "epoch": 21.508213820078225,
-      "grad_norm": 1.5450319051742554,
-      "learning_rate": 5.906834170854272e-05,
-      "loss": 4.9746,
-      "step": 41242
-    },
-    {
-      "epoch": 21.508735332464145,
-      "grad_norm": 1.526537537574768,
-      "learning_rate": 5.906733668341708e-05,
-      "loss": 5.1116,
-      "step": 41243
-    },
-    {
-      "epoch": 21.509256844850064,
-      "grad_norm": 1.5868839025497437,
-      "learning_rate": 5.906633165829146e-05,
-      "loss": 5.3606,
-      "step": 41244
-    },
-    {
-      "epoch": 21.509778357235984,
-      "grad_norm": 1.5160589218139648,
-      "learning_rate": 5.906532663316583e-05,
-      "loss": 5.3751,
-      "step": 41245
-    },
-    {
-      "epoch": 21.510299869621903,
-      "grad_norm": 1.5750348567962646,
-      "learning_rate": 5.906432160804021e-05,
-      "loss": 5.2339,
-      "step": 41246
-    },
-    {
-      "epoch": 21.510821382007823,
-      "grad_norm": 1.6170661449432373,
-      "learning_rate": 5.9063316582914574e-05,
-      "loss": 5.3353,
-      "step": 41247
-    },
-    {
-      "epoch": 21.511342894393742,
-      "grad_norm": 1.6072708368301392,
-      "learning_rate": 5.906231155778895e-05,
-      "loss": 4.995,
-      "step": 41248
-    },
-    {
-      "epoch": 21.51186440677966,
-      "grad_norm": 1.4909777641296387,
-      "learning_rate": 5.9061306532663316e-05,
-      "loss": 5.1838,
-      "step": 41249
-    },
-    {
-      "epoch": 21.51238591916558,
-      "grad_norm": 1.5419069528579712,
-      "learning_rate": 5.9060301507537694e-05,
-      "loss": 4.9955,
-      "step": 41250
-    },
-    {
-      "epoch": 21.5129074315515,
-      "grad_norm": 1.5469324588775635,
-      "learning_rate": 5.9059296482412065e-05,
-      "loss": 5.3478,
-      "step": 41251
-    },
-    {
-      "epoch": 21.51342894393742,
-      "grad_norm": 1.4407187700271606,
-      "learning_rate": 5.905829145728643e-05,
-      "loss": 5.4094,
-      "step": 41252
-    },
-    {
-      "epoch": 21.513950456323336,
-      "grad_norm": 1.500580072402954,
-      "learning_rate": 5.905728643216081e-05,
-      "loss": 5.5507,
-      "step": 41253
-    },
-    {
-      "epoch": 21.514471968709255,
-      "grad_norm": 1.5069688558578491,
-      "learning_rate": 5.905628140703517e-05,
-      "loss": 5.2306,
-      "step": 41254
-    },
-    {
-      "epoch": 21.514993481095175,
-      "grad_norm": 1.4819751977920532,
-      "learning_rate": 5.905527638190955e-05,
-      "loss": 5.1309,
-      "step": 41255
-    },
-    {
-      "epoch": 21.515514993481094,
-      "grad_norm": 1.5541964769363403,
-      "learning_rate": 5.905427135678392e-05,
-      "loss": 5.2847,
-      "step": 41256
-    },
-    {
-      "epoch": 21.516036505867014,
-      "grad_norm": 1.4017810821533203,
-      "learning_rate": 5.90532663316583e-05,
-      "loss": 5.343,
-      "step": 41257
-    },
-    {
-      "epoch": 21.516558018252933,
-      "grad_norm": 1.4409304857254028,
-      "learning_rate": 5.905226130653266e-05,
-      "loss": 5.1683,
-      "step": 41258
-    },
-    {
-      "epoch": 21.517079530638853,
-      "grad_norm": 1.569568395614624,
-      "learning_rate": 5.905125628140704e-05,
-      "loss": 5.0821,
-      "step": 41259
-    },
-    {
-      "epoch": 21.517601043024772,
-      "grad_norm": 1.5077970027923584,
-      "learning_rate": 5.905025125628141e-05,
-      "loss": 5.3681,
-      "step": 41260
-    },
-    {
-      "epoch": 21.51812255541069,
-      "grad_norm": 1.605093240737915,
-      "learning_rate": 5.904924623115579e-05,
-      "loss": 5.0028,
-      "step": 41261
-    },
-    {
-      "epoch": 21.51864406779661,
-      "grad_norm": 1.5409423112869263,
-      "learning_rate": 5.904824120603015e-05,
-      "loss": 5.3555,
-      "step": 41262
-    },
-    {
-      "epoch": 21.51916558018253,
-      "grad_norm": 1.5781704187393188,
-      "learning_rate": 5.904723618090453e-05,
-      "loss": 5.2161,
-      "step": 41263
-    },
-    {
-      "epoch": 21.51968709256845,
-      "grad_norm": 1.4119279384613037,
-      "learning_rate": 5.9046231155778895e-05,
-      "loss": 5.4529,
-      "step": 41264
-    },
-    {
-      "epoch": 21.52020860495437,
-      "grad_norm": 1.5030242204666138,
-      "learning_rate": 5.9045226130653266e-05,
-      "loss": 5.2937,
-      "step": 41265
-    },
-    {
-      "epoch": 21.520730117340285,
-      "grad_norm": 1.556138277053833,
-      "learning_rate": 5.9044221105527644e-05,
-      "loss": 5.0308,
-      "step": 41266
-    },
-    {
-      "epoch": 21.521251629726205,
-      "grad_norm": 1.6110644340515137,
-      "learning_rate": 5.904321608040201e-05,
-      "loss": 5.2451,
-      "step": 41267
-    },
-    {
-      "epoch": 21.521773142112124,
-      "grad_norm": 1.553777813911438,
-      "learning_rate": 5.9042211055276386e-05,
-      "loss": 5.1459,
-      "step": 41268
-    },
-    {
-      "epoch": 21.522294654498044,
-      "grad_norm": 1.5336902141571045,
-      "learning_rate": 5.904120603015075e-05,
-      "loss": 5.3676,
-      "step": 41269
-    },
-    {
-      "epoch": 21.522816166883963,
-      "grad_norm": 1.6905298233032227,
-      "learning_rate": 5.904020100502513e-05,
-      "loss": 5.0414,
-      "step": 41270
-    },
-    {
-      "epoch": 21.523337679269883,
-      "grad_norm": 1.4861394166946411,
-      "learning_rate": 5.90391959798995e-05,
-      "loss": 5.3022,
-      "step": 41271
-    },
-    {
-      "epoch": 21.523859191655802,
-      "grad_norm": 1.5998461246490479,
-      "learning_rate": 5.903819095477388e-05,
-      "loss": 4.3752,
-      "step": 41272
-    },
-    {
-      "epoch": 21.52438070404172,
-      "grad_norm": 1.4904248714447021,
-      "learning_rate": 5.903718592964824e-05,
-      "loss": 5.1484,
-      "step": 41273
-    },
-    {
-      "epoch": 21.52490221642764,
-      "grad_norm": 1.5549038648605347,
-      "learning_rate": 5.903618090452262e-05,
-      "loss": 5.7047,
-      "step": 41274
-    },
-    {
-      "epoch": 21.52542372881356,
-      "grad_norm": 1.4862509965896606,
-      "learning_rate": 5.9035175879396984e-05,
-      "loss": 5.0035,
-      "step": 41275
-    },
-    {
-      "epoch": 21.52594524119948,
-      "grad_norm": 1.575300931930542,
-      "learning_rate": 5.903417085427136e-05,
-      "loss": 5.1487,
-      "step": 41276
-    },
-    {
-      "epoch": 21.526466753585396,
-      "grad_norm": 1.49616277217865,
-      "learning_rate": 5.903316582914573e-05,
-      "loss": 5.6388,
-      "step": 41277
-    },
-    {
-      "epoch": 21.526988265971315,
-      "grad_norm": 1.5445679426193237,
-      "learning_rate": 5.90321608040201e-05,
-      "loss": 5.0637,
-      "step": 41278
-    },
-    {
-      "epoch": 21.527509778357235,
-      "grad_norm": 1.5683672428131104,
-      "learning_rate": 5.9031155778894475e-05,
-      "loss": 5.0367,
-      "step": 41279
-    },
-    {
-      "epoch": 21.528031290743154,
-      "grad_norm": 1.6211491823196411,
-      "learning_rate": 5.9030150753768846e-05,
-      "loss": 4.2479,
-      "step": 41280
-    },
-    {
-      "epoch": 21.528552803129074,
-      "grad_norm": 1.5930051803588867,
-      "learning_rate": 5.9029145728643223e-05,
-      "loss": 5.1962,
-      "step": 41281
-    },
-    {
-      "epoch": 21.529074315514993,
-      "grad_norm": 1.5900702476501465,
-      "learning_rate": 5.902814070351759e-05,
-      "loss": 5.1131,
-      "step": 41282
-    },
-    {
-      "epoch": 21.529595827900913,
-      "grad_norm": 1.7908767461776733,
-      "learning_rate": 5.9027135678391966e-05,
-      "loss": 5.0271,
-      "step": 41283
-    },
-    {
-      "epoch": 21.530117340286832,
-      "grad_norm": 1.503699779510498,
-      "learning_rate": 5.902613065326633e-05,
-      "loss": 5.4777,
-      "step": 41284
-    },
-    {
-      "epoch": 21.53063885267275,
-      "grad_norm": 1.3899527788162231,
-      "learning_rate": 5.902512562814071e-05,
-      "loss": 5.297,
-      "step": 41285
-    },
-    {
-      "epoch": 21.53116036505867,
-      "grad_norm": 1.7177962064743042,
-      "learning_rate": 5.902412060301508e-05,
-      "loss": 5.026,
-      "step": 41286
-    },
-    {
-      "epoch": 21.53168187744459,
-      "grad_norm": 1.455560564994812,
-      "learning_rate": 5.9023115577889457e-05,
-      "loss": 5.5653,
-      "step": 41287
-    },
-    {
-      "epoch": 21.53220338983051,
-      "grad_norm": 1.5129448175430298,
-      "learning_rate": 5.902211055276382e-05,
-      "loss": 5.5221,
-      "step": 41288
-    },
-    {
-      "epoch": 21.532724902216426,
-      "grad_norm": 1.4738863706588745,
-      "learning_rate": 5.90211055276382e-05,
-      "loss": 5.4537,
-      "step": 41289
-    },
-    {
-      "epoch": 21.533246414602345,
-      "grad_norm": 1.5757029056549072,
-      "learning_rate": 5.902010050251256e-05,
-      "loss": 5.483,
-      "step": 41290
-    },
-    {
-      "epoch": 21.533767926988265,
-      "grad_norm": 1.5015990734100342,
-      "learning_rate": 5.9019095477386934e-05,
-      "loss": 5.2347,
-      "step": 41291
-    },
-    {
-      "epoch": 21.534289439374184,
-      "grad_norm": 1.533408522605896,
-      "learning_rate": 5.901809045226131e-05,
-      "loss": 5.534,
-      "step": 41292
-    },
-    {
-      "epoch": 21.534810951760104,
-      "grad_norm": 1.5703332424163818,
-      "learning_rate": 5.9017085427135676e-05,
-      "loss": 5.3702,
-      "step": 41293
-    },
-    {
-      "epoch": 21.535332464146023,
-      "grad_norm": 1.5454661846160889,
-      "learning_rate": 5.9016080402010054e-05,
-      "loss": 5.2517,
-      "step": 41294
-    },
-    {
-      "epoch": 21.535853976531943,
-      "grad_norm": 1.4862533807754517,
-      "learning_rate": 5.901507537688442e-05,
-      "loss": 5.6052,
-      "step": 41295
-    },
-    {
-      "epoch": 21.536375488917862,
-      "grad_norm": 1.6355597972869873,
-      "learning_rate": 5.9014070351758796e-05,
-      "loss": 5.1414,
-      "step": 41296
-    },
-    {
-      "epoch": 21.53689700130378,
-      "grad_norm": 1.6251059770584106,
-      "learning_rate": 5.901306532663317e-05,
-      "loss": 5.0651,
-      "step": 41297
-    },
-    {
-      "epoch": 21.5374185136897,
-      "grad_norm": 1.4672293663024902,
-      "learning_rate": 5.9012060301507545e-05,
-      "loss": 4.8981,
-      "step": 41298
-    },
-    {
-      "epoch": 21.53794002607562,
-      "grad_norm": 1.5937309265136719,
-      "learning_rate": 5.901105527638191e-05,
-      "loss": 5.0438,
-      "step": 41299
-    },
-    {
-      "epoch": 21.53846153846154,
-      "grad_norm": 1.5574020147323608,
-      "learning_rate": 5.901005025125629e-05,
-      "loss": 5.3621,
-      "step": 41300
-    },
-    {
-      "epoch": 21.538983050847456,
-      "grad_norm": 1.6874536275863647,
-      "learning_rate": 5.900904522613066e-05,
-      "loss": 5.1124,
-      "step": 41301
-    },
-    {
-      "epoch": 21.539504563233375,
-      "grad_norm": 1.4630059003829956,
-      "learning_rate": 5.9008040201005036e-05,
-      "loss": 4.9967,
-      "step": 41302
-    },
-    {
-      "epoch": 21.540026075619295,
-      "grad_norm": 1.4255181550979614,
-      "learning_rate": 5.90070351758794e-05,
-      "loss": 5.541,
-      "step": 41303
-    },
-    {
-      "epoch": 21.540547588005214,
-      "grad_norm": 1.4596010446548462,
-      "learning_rate": 5.900603015075378e-05,
-      "loss": 5.226,
-      "step": 41304
-    },
-    {
-      "epoch": 21.541069100391134,
-      "grad_norm": 1.4685242176055908,
-      "learning_rate": 5.900502512562814e-05,
-      "loss": 5.2777,
-      "step": 41305
-    },
-    {
-      "epoch": 21.541590612777053,
-      "grad_norm": 1.469725251197815,
-      "learning_rate": 5.900402010050251e-05,
-      "loss": 5.4477,
-      "step": 41306
-    },
-    {
-      "epoch": 21.542112125162973,
-      "grad_norm": 1.4878336191177368,
-      "learning_rate": 5.900301507537689e-05,
-      "loss": 5.4303,
-      "step": 41307
-    },
-    {
-      "epoch": 21.542633637548892,
-      "grad_norm": 1.538132905960083,
-      "learning_rate": 5.9002010050251255e-05,
-      "loss": 5.1373,
-      "step": 41308
-    },
-    {
-      "epoch": 21.54315514993481,
-      "grad_norm": 1.6088814735412598,
-      "learning_rate": 5.900100502512563e-05,
-      "loss": 5.1346,
-      "step": 41309
-    },
-    {
-      "epoch": 21.54367666232073,
-      "grad_norm": 1.4650967121124268,
-      "learning_rate": 5.9e-05,
-      "loss": 5.4505,
-      "step": 41310
-    },
-    {
-      "epoch": 21.54419817470665,
-      "grad_norm": 1.6112236976623535,
-      "learning_rate": 5.8998994974874375e-05,
-      "loss": 5.1138,
-      "step": 41311
-    },
-    {
-      "epoch": 21.54471968709257,
-      "grad_norm": 1.6032872200012207,
-      "learning_rate": 5.8997989949748746e-05,
-      "loss": 5.239,
-      "step": 41312
-    },
-    {
-      "epoch": 21.545241199478486,
-      "grad_norm": 1.429436445236206,
-      "learning_rate": 5.8996984924623124e-05,
-      "loss": 5.3193,
-      "step": 41313
-    },
-    {
-      "epoch": 21.545762711864406,
-      "grad_norm": 1.4618483781814575,
-      "learning_rate": 5.899597989949749e-05,
-      "loss": 5.4398,
-      "step": 41314
-    },
-    {
-      "epoch": 21.546284224250325,
-      "grad_norm": 1.5813241004943848,
-      "learning_rate": 5.8994974874371866e-05,
-      "loss": 5.142,
-      "step": 41315
-    },
-    {
-      "epoch": 21.546805736636244,
-      "grad_norm": 1.4722589254379272,
-      "learning_rate": 5.899396984924623e-05,
-      "loss": 5.3079,
-      "step": 41316
-    },
-    {
-      "epoch": 21.547327249022164,
-      "grad_norm": 1.4642257690429688,
-      "learning_rate": 5.899296482412061e-05,
-      "loss": 5.3665,
-      "step": 41317
-    },
-    {
-      "epoch": 21.547848761408083,
-      "grad_norm": 1.525261640548706,
-      "learning_rate": 5.899195979899498e-05,
-      "loss": 5.2653,
-      "step": 41318
-    },
-    {
-      "epoch": 21.548370273794003,
-      "grad_norm": 1.459911823272705,
-      "learning_rate": 5.8990954773869344e-05,
-      "loss": 5.3607,
-      "step": 41319
-    },
-    {
-      "epoch": 21.548891786179922,
-      "grad_norm": 1.5273574590682983,
-      "learning_rate": 5.898994974874372e-05,
-      "loss": 5.5016,
-      "step": 41320
-    },
-    {
-      "epoch": 21.54941329856584,
-      "grad_norm": 1.5402766466140747,
-      "learning_rate": 5.8988944723618086e-05,
-      "loss": 5.3229,
-      "step": 41321
-    },
-    {
-      "epoch": 21.54993481095176,
-      "grad_norm": 1.4661905765533447,
-      "learning_rate": 5.8987939698492464e-05,
-      "loss": 5.4789,
-      "step": 41322
-    },
-    {
-      "epoch": 21.55045632333768,
-      "grad_norm": 1.509519100189209,
-      "learning_rate": 5.8986934673366835e-05,
-      "loss": 5.215,
-      "step": 41323
-    },
-    {
-      "epoch": 21.5509778357236,
-      "grad_norm": 1.557570219039917,
-      "learning_rate": 5.898592964824121e-05,
-      "loss": 5.5647,
-      "step": 41324
-    },
-    {
-      "epoch": 21.551499348109516,
-      "grad_norm": 1.5032492876052856,
-      "learning_rate": 5.898492462311558e-05,
-      "loss": 5.5752,
-      "step": 41325
-    },
-    {
-      "epoch": 21.552020860495436,
-      "grad_norm": 1.4643694162368774,
-      "learning_rate": 5.8983919597989955e-05,
-      "loss": 5.6583,
-      "step": 41326
-    },
-    {
-      "epoch": 21.552542372881355,
-      "grad_norm": 1.4774478673934937,
-      "learning_rate": 5.8982914572864326e-05,
-      "loss": 5.3805,
-      "step": 41327
-    },
-    {
-      "epoch": 21.553063885267274,
-      "grad_norm": 1.580359697341919,
-      "learning_rate": 5.8981909547738704e-05,
-      "loss": 4.2801,
-      "step": 41328
-    },
-    {
-      "epoch": 21.553585397653194,
-      "grad_norm": 1.5588352680206299,
-      "learning_rate": 5.898090452261307e-05,
-      "loss": 5.2692,
-      "step": 41329
-    },
-    {
-      "epoch": 21.554106910039113,
-      "grad_norm": 1.5803859233856201,
-      "learning_rate": 5.8979899497487446e-05,
-      "loss": 5.2465,
-      "step": 41330
-    },
-    {
-      "epoch": 21.554628422425033,
-      "grad_norm": 1.4927420616149902,
-      "learning_rate": 5.897889447236181e-05,
-      "loss": 5.4267,
-      "step": 41331
-    },
-    {
-      "epoch": 21.555149934810952,
-      "grad_norm": 1.5084747076034546,
-      "learning_rate": 5.897788944723618e-05,
-      "loss": 5.1386,
-      "step": 41332
-    },
-    {
-      "epoch": 21.555671447196872,
-      "grad_norm": 1.533982276916504,
-      "learning_rate": 5.897688442211056e-05,
-      "loss": 5.1234,
-      "step": 41333
-    },
-    {
-      "epoch": 21.55619295958279,
-      "grad_norm": 1.5401102304458618,
-      "learning_rate": 5.897587939698492e-05,
-      "loss": 5.1858,
-      "step": 41334
-    },
-    {
-      "epoch": 21.55671447196871,
-      "grad_norm": 1.6077932119369507,
-      "learning_rate": 5.89748743718593e-05,
-      "loss": 5.2746,
-      "step": 41335
-    },
-    {
-      "epoch": 21.557235984354627,
-      "grad_norm": 1.6266292333602905,
-      "learning_rate": 5.8973869346733665e-05,
-      "loss": 4.8293,
-      "step": 41336
-    },
-    {
-      "epoch": 21.557757496740546,
-      "grad_norm": 1.6232482194900513,
-      "learning_rate": 5.897286432160804e-05,
-      "loss": 4.933,
-      "step": 41337
-    },
-    {
-      "epoch": 21.558279009126466,
-      "grad_norm": 1.4842878580093384,
-      "learning_rate": 5.8971859296482414e-05,
-      "loss": 5.3509,
-      "step": 41338
-    },
-    {
-      "epoch": 21.558800521512385,
-      "grad_norm": 1.4475979804992676,
-      "learning_rate": 5.897085427135679e-05,
-      "loss": 5.7354,
-      "step": 41339
-    },
-    {
-      "epoch": 21.559322033898304,
-      "grad_norm": 1.5124280452728271,
-      "learning_rate": 5.8969849246231156e-05,
-      "loss": 5.2072,
-      "step": 41340
-    },
-    {
-      "epoch": 21.559843546284224,
-      "grad_norm": 1.508368730545044,
-      "learning_rate": 5.8968844221105534e-05,
-      "loss": 5.2261,
-      "step": 41341
-    },
-    {
-      "epoch": 21.560365058670143,
-      "grad_norm": 1.5105043649673462,
-      "learning_rate": 5.89678391959799e-05,
-      "loss": 5.5376,
-      "step": 41342
-    },
-    {
-      "epoch": 21.560886571056063,
-      "grad_norm": 1.4866056442260742,
-      "learning_rate": 5.8966834170854276e-05,
-      "loss": 5.5118,
-      "step": 41343
-    },
-    {
-      "epoch": 21.561408083441982,
-      "grad_norm": 1.5670864582061768,
-      "learning_rate": 5.896582914572865e-05,
-      "loss": 5.3386,
-      "step": 41344
-    },
-    {
-      "epoch": 21.561929595827902,
-      "grad_norm": 1.5481849908828735,
-      "learning_rate": 5.896482412060301e-05,
-      "loss": 4.9978,
-      "step": 41345
-    },
-    {
-      "epoch": 21.56245110821382,
-      "grad_norm": 1.557443618774414,
-      "learning_rate": 5.896381909547739e-05,
-      "loss": 5.3842,
-      "step": 41346
-    },
-    {
-      "epoch": 21.56297262059974,
-      "grad_norm": 1.6475753784179688,
-      "learning_rate": 5.896281407035176e-05,
-      "loss": 5.0819,
-      "step": 41347
-    },
-    {
-      "epoch": 21.56349413298566,
-      "grad_norm": 1.558729887008667,
-      "learning_rate": 5.896180904522614e-05,
-      "loss": 5.0014,
-      "step": 41348
-    },
-    {
-      "epoch": 21.564015645371576,
-      "grad_norm": 1.4555788040161133,
-      "learning_rate": 5.89608040201005e-05,
-      "loss": 5.4457,
-      "step": 41349
-    },
-    {
-      "epoch": 21.564537157757496,
-      "grad_norm": 1.5509744882583618,
-      "learning_rate": 5.895979899497488e-05,
-      "loss": 5.2223,
-      "step": 41350
-    },
-    {
-      "epoch": 21.565058670143415,
-      "grad_norm": 1.5769461393356323,
-      "learning_rate": 5.8958793969849245e-05,
-      "loss": 4.6862,
-      "step": 41351
-    },
-    {
-      "epoch": 21.565580182529335,
-      "grad_norm": 1.5209707021713257,
-      "learning_rate": 5.895778894472362e-05,
-      "loss": 5.4294,
-      "step": 41352
-    },
-    {
-      "epoch": 21.566101694915254,
-      "grad_norm": 1.5281062126159668,
-      "learning_rate": 5.8956783919597993e-05,
-      "loss": 5.2457,
-      "step": 41353
-    },
-    {
-      "epoch": 21.566623207301173,
-      "grad_norm": 1.636772632598877,
-      "learning_rate": 5.895577889447237e-05,
-      "loss": 5.0229,
-      "step": 41354
-    },
-    {
-      "epoch": 21.567144719687093,
-      "grad_norm": 1.528497338294983,
-      "learning_rate": 5.8954773869346736e-05,
-      "loss": 5.5249,
-      "step": 41355
-    },
-    {
-      "epoch": 21.567666232073012,
-      "grad_norm": 1.5780446529388428,
-      "learning_rate": 5.895376884422111e-05,
-      "loss": 5.4808,
-      "step": 41356
-    },
-    {
-      "epoch": 21.568187744458932,
-      "grad_norm": 1.536303162574768,
-      "learning_rate": 5.895276381909548e-05,
-      "loss": 5.3068,
-      "step": 41357
-    },
-    {
-      "epoch": 21.56870925684485,
-      "grad_norm": 1.5733270645141602,
-      "learning_rate": 5.895175879396985e-05,
-      "loss": 5.2632,
-      "step": 41358
-    },
-    {
-      "epoch": 21.56923076923077,
-      "grad_norm": 1.6079787015914917,
-      "learning_rate": 5.8950753768844227e-05,
-      "loss": 5.4695,
-      "step": 41359
-    },
-    {
-      "epoch": 21.569752281616687,
-      "grad_norm": 1.4912632703781128,
-      "learning_rate": 5.894974874371859e-05,
-      "loss": 5.6474,
-      "step": 41360
-    },
-    {
-      "epoch": 21.570273794002606,
-      "grad_norm": 1.4878822565078735,
-      "learning_rate": 5.894874371859297e-05,
-      "loss": 5.4958,
-      "step": 41361
-    },
-    {
-      "epoch": 21.570795306388526,
-      "grad_norm": 1.5711787939071655,
-      "learning_rate": 5.894773869346733e-05,
-      "loss": 4.7963,
-      "step": 41362
-    },
-    {
-      "epoch": 21.571316818774445,
-      "grad_norm": 1.536095380783081,
-      "learning_rate": 5.894673366834171e-05,
-      "loss": 5.3576,
-      "step": 41363
-    },
-    {
-      "epoch": 21.571838331160365,
-      "grad_norm": 1.5319575071334839,
-      "learning_rate": 5.894572864321608e-05,
-      "loss": 5.2181,
-      "step": 41364
-    },
-    {
-      "epoch": 21.572359843546284,
-      "grad_norm": 1.606568694114685,
-      "learning_rate": 5.894472361809046e-05,
-      "loss": 5.1574,
-      "step": 41365
-    },
-    {
-      "epoch": 21.572881355932203,
-      "grad_norm": 1.537639856338501,
-      "learning_rate": 5.8943718592964824e-05,
-      "loss": 4.8327,
-      "step": 41366
-    },
-    {
-      "epoch": 21.573402868318123,
-      "grad_norm": 1.6431291103363037,
-      "learning_rate": 5.89427135678392e-05,
-      "loss": 5.5306,
-      "step": 41367
-    },
-    {
-      "epoch": 21.573924380704042,
-      "grad_norm": 1.5570900440216064,
-      "learning_rate": 5.894170854271357e-05,
-      "loss": 4.7533,
-      "step": 41368
-    },
-    {
-      "epoch": 21.574445893089962,
-      "grad_norm": 1.518526315689087,
-      "learning_rate": 5.894070351758795e-05,
-      "loss": 5.4788,
-      "step": 41369
-    },
-    {
-      "epoch": 21.57496740547588,
-      "grad_norm": 1.5892174243927002,
-      "learning_rate": 5.8939698492462315e-05,
-      "loss": 5.2001,
-      "step": 41370
-    },
-    {
-      "epoch": 21.5754889178618,
-      "grad_norm": 1.501483678817749,
-      "learning_rate": 5.893869346733668e-05,
-      "loss": 5.3999,
-      "step": 41371
-    },
-    {
-      "epoch": 21.576010430247717,
-      "grad_norm": 1.5541528463363647,
-      "learning_rate": 5.893768844221106e-05,
-      "loss": 4.5509,
-      "step": 41372
-    },
-    {
-      "epoch": 21.576531942633636,
-      "grad_norm": 1.4519033432006836,
-      "learning_rate": 5.893668341708543e-05,
-      "loss": 5.6925,
-      "step": 41373
-    },
-    {
-      "epoch": 21.577053455019556,
-      "grad_norm": 1.6237998008728027,
-      "learning_rate": 5.8935678391959806e-05,
-      "loss": 4.3589,
-      "step": 41374
-    },
-    {
-      "epoch": 21.577574967405475,
-      "grad_norm": 1.4941470623016357,
-      "learning_rate": 5.893467336683417e-05,
-      "loss": 5.3427,
-      "step": 41375
-    },
-    {
-      "epoch": 21.578096479791395,
-      "grad_norm": 1.5373578071594238,
-      "learning_rate": 5.893366834170855e-05,
-      "loss": 5.1249,
-      "step": 41376
-    },
-    {
-      "epoch": 21.578617992177314,
-      "grad_norm": 1.5348938703536987,
-      "learning_rate": 5.893266331658291e-05,
-      "loss": 5.2206,
-      "step": 41377
-    },
-    {
-      "epoch": 21.579139504563233,
-      "grad_norm": 1.5524977445602417,
-      "learning_rate": 5.893165829145729e-05,
-      "loss": 5.5673,
-      "step": 41378
-    },
-    {
-      "epoch": 21.579661016949153,
-      "grad_norm": 1.5716183185577393,
-      "learning_rate": 5.893065326633166e-05,
-      "loss": 5.4244,
-      "step": 41379
-    },
-    {
-      "epoch": 21.580182529335072,
-      "grad_norm": 1.5389695167541504,
-      "learning_rate": 5.892964824120604e-05,
-      "loss": 5.5832,
-      "step": 41380
-    },
-    {
-      "epoch": 21.580704041720992,
-      "grad_norm": 1.5567045211791992,
-      "learning_rate": 5.89286432160804e-05,
-      "loss": 5.4572,
-      "step": 41381
-    },
-    {
-      "epoch": 21.58122555410691,
-      "grad_norm": 1.499077320098877,
-      "learning_rate": 5.892763819095478e-05,
-      "loss": 5.3614,
-      "step": 41382
-    },
-    {
-      "epoch": 21.58174706649283,
-      "grad_norm": 1.4536120891571045,
-      "learning_rate": 5.8926633165829145e-05,
-      "loss": 5.4685,
-      "step": 41383
-    },
-    {
-      "epoch": 21.582268578878747,
-      "grad_norm": 1.468278169631958,
-      "learning_rate": 5.8925628140703516e-05,
-      "loss": 5.4499,
-      "step": 41384
-    },
-    {
-      "epoch": 21.582790091264666,
-      "grad_norm": 1.6011974811553955,
-      "learning_rate": 5.8924623115577894e-05,
-      "loss": 5.2173,
-      "step": 41385
-    },
-    {
-      "epoch": 21.583311603650586,
-      "grad_norm": 1.5428211688995361,
-      "learning_rate": 5.892361809045226e-05,
-      "loss": 4.6602,
-      "step": 41386
-    },
-    {
-      "epoch": 21.583833116036505,
-      "grad_norm": 1.6843336820602417,
-      "learning_rate": 5.8922613065326636e-05,
-      "loss": 5.0611,
-      "step": 41387
-    },
-    {
-      "epoch": 21.584354628422425,
-      "grad_norm": 1.5177103281021118,
-      "learning_rate": 5.892160804020101e-05,
-      "loss": 5.2427,
-      "step": 41388
-    },
-    {
-      "epoch": 21.584876140808344,
-      "grad_norm": 1.7024086713790894,
-      "learning_rate": 5.8920603015075385e-05,
-      "loss": 4.8055,
-      "step": 41389
-    },
-    {
-      "epoch": 21.585397653194264,
-      "grad_norm": 1.453627347946167,
-      "learning_rate": 5.891959798994975e-05,
-      "loss": 5.4637,
-      "step": 41390
-    },
-    {
-      "epoch": 21.585919165580183,
-      "grad_norm": 1.5850744247436523,
-      "learning_rate": 5.891859296482413e-05,
-      "loss": 5.5119,
-      "step": 41391
-    },
-    {
-      "epoch": 21.586440677966102,
-      "grad_norm": 1.5077663660049438,
-      "learning_rate": 5.891758793969849e-05,
-      "loss": 5.1874,
-      "step": 41392
-    },
-    {
-      "epoch": 21.586962190352022,
-      "grad_norm": 1.5064185857772827,
-      "learning_rate": 5.891658291457287e-05,
-      "loss": 5.3789,
-      "step": 41393
-    },
-    {
-      "epoch": 21.58748370273794,
-      "grad_norm": 1.5786540508270264,
-      "learning_rate": 5.891557788944724e-05,
-      "loss": 5.0895,
-      "step": 41394
-    },
-    {
-      "epoch": 21.58800521512386,
-      "grad_norm": 1.5935633182525635,
-      "learning_rate": 5.891457286432162e-05,
-      "loss": 5.5072,
-      "step": 41395
-    },
-    {
-      "epoch": 21.588526727509777,
-      "grad_norm": 1.4772313833236694,
-      "learning_rate": 5.891356783919598e-05,
-      "loss": 5.3167,
-      "step": 41396
-    },
-    {
-      "epoch": 21.589048239895696,
-      "grad_norm": 1.5155328512191772,
-      "learning_rate": 5.891256281407036e-05,
-      "loss": 5.3329,
-      "step": 41397
-    },
-    {
-      "epoch": 21.589569752281616,
-      "grad_norm": 1.4700196981430054,
-      "learning_rate": 5.8911557788944725e-05,
-      "loss": 5.0588,
-      "step": 41398
-    },
-    {
-      "epoch": 21.590091264667535,
-      "grad_norm": 1.6144379377365112,
-      "learning_rate": 5.8910552763819096e-05,
-      "loss": 5.1268,
-      "step": 41399
-    },
-    {
-      "epoch": 21.590612777053455,
-      "grad_norm": 1.4258018732070923,
-      "learning_rate": 5.8909547738693474e-05,
-      "loss": 5.3086,
-      "step": 41400
-    },
-    {
-      "epoch": 21.591134289439374,
-      "grad_norm": 1.5602774620056152,
-      "learning_rate": 5.890854271356784e-05,
-      "loss": 5.1959,
-      "step": 41401
-    },
-    {
-      "epoch": 21.591655801825294,
-      "grad_norm": 1.5990005731582642,
-      "learning_rate": 5.8907537688442216e-05,
-      "loss": 4.9382,
-      "step": 41402
-    },
-    {
-      "epoch": 21.592177314211213,
-      "grad_norm": 1.5364447832107544,
-      "learning_rate": 5.890653266331658e-05,
-      "loss": 5.4361,
-      "step": 41403
-    },
-    {
-      "epoch": 21.592698826597132,
-      "grad_norm": 1.6290539503097534,
-      "learning_rate": 5.890552763819096e-05,
-      "loss": 5.3129,
-      "step": 41404
-    },
-    {
-      "epoch": 21.593220338983052,
-      "grad_norm": 1.6921759843826294,
-      "learning_rate": 5.890452261306533e-05,
-      "loss": 5.1706,
-      "step": 41405
-    },
-    {
-      "epoch": 21.59374185136897,
-      "grad_norm": 1.4879244565963745,
-      "learning_rate": 5.8903517587939707e-05,
-      "loss": 5.1502,
-      "step": 41406
-    },
-    {
-      "epoch": 21.59426336375489,
-      "grad_norm": 1.53236722946167,
-      "learning_rate": 5.890251256281407e-05,
-      "loss": 5.3556,
-      "step": 41407
-    },
-    {
-      "epoch": 21.594784876140807,
-      "grad_norm": 1.4980614185333252,
-      "learning_rate": 5.890150753768845e-05,
-      "loss": 5.6075,
-      "step": 41408
-    },
-    {
-      "epoch": 21.595306388526726,
-      "grad_norm": 1.5190949440002441,
-      "learning_rate": 5.890050251256281e-05,
-      "loss": 5.1779,
-      "step": 41409
-    },
-    {
-      "epoch": 21.595827900912646,
-      "grad_norm": 1.4911565780639648,
-      "learning_rate": 5.889949748743719e-05,
-      "loss": 5.2891,
-      "step": 41410
-    },
-    {
-      "epoch": 21.596349413298565,
-      "grad_norm": 1.500127911567688,
-      "learning_rate": 5.889849246231156e-05,
-      "loss": 5.3406,
-      "step": 41411
-    },
-    {
-      "epoch": 21.596870925684485,
-      "grad_norm": 1.4777846336364746,
-      "learning_rate": 5.8897487437185926e-05,
-      "loss": 5.5058,
-      "step": 41412
-    },
-    {
-      "epoch": 21.597392438070404,
-      "grad_norm": 1.6216222047805786,
-      "learning_rate": 5.8896482412060304e-05,
-      "loss": 5.24,
-      "step": 41413
-    },
-    {
-      "epoch": 21.597913950456324,
-      "grad_norm": 1.4594931602478027,
-      "learning_rate": 5.8895477386934675e-05,
-      "loss": 5.4462,
-      "step": 41414
-    },
-    {
-      "epoch": 21.598435462842243,
-      "grad_norm": 1.5225138664245605,
-      "learning_rate": 5.889447236180905e-05,
-      "loss": 4.7676,
-      "step": 41415
-    },
-    {
-      "epoch": 21.598956975228162,
-      "grad_norm": 1.5399303436279297,
-      "learning_rate": 5.889346733668342e-05,
-      "loss": 4.5742,
-      "step": 41416
-    },
-    {
-      "epoch": 21.599478487614082,
-      "grad_norm": 1.567804217338562,
-      "learning_rate": 5.8892462311557795e-05,
-      "loss": 5.1795,
-      "step": 41417
-    },
-    {
-      "epoch": 21.6,
-      "grad_norm": 1.497216820716858,
-      "learning_rate": 5.889145728643216e-05,
-      "loss": 5.6878,
-      "step": 41418
-    },
-    {
-      "epoch": 21.60052151238592,
-      "grad_norm": 1.544082760810852,
-      "learning_rate": 5.889045226130654e-05,
-      "loss": 5.3631,
-      "step": 41419
-    },
-    {
-      "epoch": 21.601043024771837,
-      "grad_norm": 1.9596598148345947,
-      "learning_rate": 5.888944723618091e-05,
-      "loss": 4.3475,
-      "step": 41420
-    },
-    {
-      "epoch": 21.601564537157756,
-      "grad_norm": 1.464604139328003,
-      "learning_rate": 5.8888442211055286e-05,
-      "loss": 5.4788,
-      "step": 41421
-    },
-    {
-      "epoch": 21.602086049543676,
-      "grad_norm": 1.601670742034912,
-      "learning_rate": 5.888743718592965e-05,
-      "loss": 5.1724,
-      "step": 41422
-    },
-    {
-      "epoch": 21.602607561929595,
-      "grad_norm": 1.5380523204803467,
-      "learning_rate": 5.888643216080403e-05,
-      "loss": 5.2525,
-      "step": 41423
-    },
-    {
-      "epoch": 21.603129074315515,
-      "grad_norm": 1.5500893592834473,
-      "learning_rate": 5.888542713567839e-05,
-      "loss": 5.4176,
-      "step": 41424
-    },
-    {
-      "epoch": 21.603650586701434,
-      "grad_norm": 1.5062679052352905,
-      "learning_rate": 5.888442211055276e-05,
-      "loss": 5.262,
-      "step": 41425
-    },
-    {
-      "epoch": 21.604172099087354,
-      "grad_norm": 1.4823168516159058,
-      "learning_rate": 5.888341708542714e-05,
-      "loss": 4.8063,
-      "step": 41426
-    },
-    {
-      "epoch": 21.604693611473273,
-      "grad_norm": 1.592042326927185,
-      "learning_rate": 5.8882412060301505e-05,
-      "loss": 5.2058,
-      "step": 41427
-    },
-    {
-      "epoch": 21.605215123859193,
-      "grad_norm": 1.8300832509994507,
-      "learning_rate": 5.888140703517588e-05,
-      "loss": 4.9953,
-      "step": 41428
-    },
-    {
-      "epoch": 21.605736636245112,
-      "grad_norm": 1.5570449829101562,
-      "learning_rate": 5.888040201005025e-05,
-      "loss": 4.8435,
-      "step": 41429
-    },
-    {
-      "epoch": 21.60625814863103,
-      "grad_norm": 1.4779289960861206,
-      "learning_rate": 5.8879396984924625e-05,
-      "loss": 5.4483,
-      "step": 41430
-    },
-    {
-      "epoch": 21.60677966101695,
-      "grad_norm": 1.5822534561157227,
-      "learning_rate": 5.8878391959798996e-05,
-      "loss": 5.1704,
-      "step": 41431
-    },
-    {
-      "epoch": 21.607301173402867,
-      "grad_norm": 1.5070769786834717,
-      "learning_rate": 5.8877386934673374e-05,
-      "loss": 5.4858,
-      "step": 41432
-    },
-    {
-      "epoch": 21.607822685788786,
-      "grad_norm": 1.536942958831787,
-      "learning_rate": 5.887638190954774e-05,
-      "loss": 5.4964,
-      "step": 41433
-    },
-    {
-      "epoch": 21.608344198174706,
-      "grad_norm": 1.6103627681732178,
-      "learning_rate": 5.8875376884422116e-05,
-      "loss": 5.0719,
-      "step": 41434
-    },
-    {
-      "epoch": 21.608865710560625,
-      "grad_norm": 1.5022697448730469,
-      "learning_rate": 5.887437185929649e-05,
-      "loss": 5.3846,
-      "step": 41435
-    },
-    {
-      "epoch": 21.609387222946545,
-      "grad_norm": 1.4528095722198486,
-      "learning_rate": 5.8873366834170865e-05,
-      "loss": 5.0784,
-      "step": 41436
-    },
-    {
-      "epoch": 21.609908735332464,
-      "grad_norm": 1.4652621746063232,
-      "learning_rate": 5.887236180904523e-05,
-      "loss": 5.4985,
-      "step": 41437
-    },
-    {
-      "epoch": 21.610430247718384,
-      "grad_norm": 1.5103555917739868,
-      "learning_rate": 5.8871356783919594e-05,
-      "loss": 5.4056,
-      "step": 41438
-    },
-    {
-      "epoch": 21.610951760104303,
-      "grad_norm": 1.4960732460021973,
-      "learning_rate": 5.887035175879397e-05,
-      "loss": 4.8726,
-      "step": 41439
-    },
-    {
-      "epoch": 21.611473272490223,
-      "grad_norm": 1.5556795597076416,
-      "learning_rate": 5.886934673366834e-05,
-      "loss": 5.5837,
-      "step": 41440
-    },
-    {
-      "epoch": 21.611994784876142,
-      "grad_norm": 1.7180513143539429,
-      "learning_rate": 5.886834170854272e-05,
-      "loss": 5.3557,
-      "step": 41441
-    },
-    {
-      "epoch": 21.61251629726206,
-      "grad_norm": 1.6356369256973267,
-      "learning_rate": 5.8867336683417085e-05,
-      "loss": 5.3709,
-      "step": 41442
-    },
-    {
-      "epoch": 21.613037809647977,
-      "grad_norm": 1.7187151908874512,
-      "learning_rate": 5.886633165829146e-05,
-      "loss": 4.6471,
-      "step": 41443
-    },
-    {
-      "epoch": 21.613559322033897,
-      "grad_norm": 1.5300993919372559,
-      "learning_rate": 5.886532663316583e-05,
-      "loss": 5.6507,
-      "step": 41444
-    },
-    {
-      "epoch": 21.614080834419816,
-      "grad_norm": 1.6551614999771118,
-      "learning_rate": 5.8864321608040205e-05,
-      "loss": 5.4529,
-      "step": 41445
-    },
-    {
-      "epoch": 21.614602346805736,
-      "grad_norm": 1.5432894229888916,
-      "learning_rate": 5.8863316582914576e-05,
-      "loss": 5.5323,
-      "step": 41446
-    },
-    {
-      "epoch": 21.615123859191655,
-      "grad_norm": 1.4930585622787476,
-      "learning_rate": 5.8862311557788954e-05,
-      "loss": 5.5351,
-      "step": 41447
-    },
-    {
-      "epoch": 21.615645371577575,
-      "grad_norm": 1.5499262809753418,
-      "learning_rate": 5.886130653266332e-05,
-      "loss": 5.2547,
-      "step": 41448
-    },
-    {
-      "epoch": 21.616166883963494,
-      "grad_norm": 1.5132684707641602,
-      "learning_rate": 5.8860301507537696e-05,
-      "loss": 5.5468,
-      "step": 41449
-    },
-    {
-      "epoch": 21.616688396349414,
-      "grad_norm": 1.5655758380889893,
-      "learning_rate": 5.885929648241206e-05,
-      "loss": 4.9281,
-      "step": 41450
-    },
-    {
-      "epoch": 21.617209908735333,
-      "grad_norm": 1.4573997259140015,
-      "learning_rate": 5.885829145728643e-05,
-      "loss": 5.5674,
-      "step": 41451
-    },
-    {
-      "epoch": 21.617731421121253,
-      "grad_norm": 1.5619337558746338,
-      "learning_rate": 5.885728643216081e-05,
-      "loss": 5.366,
-      "step": 41452
-    },
-    {
-      "epoch": 21.618252933507172,
-      "grad_norm": 1.6491507291793823,
-      "learning_rate": 5.885628140703517e-05,
-      "loss": 5.1505,
-      "step": 41453
-    },
-    {
-      "epoch": 21.61877444589309,
-      "grad_norm": 1.6573599576950073,
-      "learning_rate": 5.885527638190955e-05,
-      "loss": 5.1203,
-      "step": 41454
-    },
-    {
-      "epoch": 21.619295958279007,
-      "grad_norm": 1.5246413946151733,
-      "learning_rate": 5.885427135678392e-05,
-      "loss": 5.501,
-      "step": 41455
-    },
-    {
-      "epoch": 21.619817470664927,
-      "grad_norm": 1.5017472505569458,
-      "learning_rate": 5.88532663316583e-05,
-      "loss": 5.2805,
-      "step": 41456
-    },
-    {
-      "epoch": 21.620338983050846,
-      "grad_norm": 1.4843274354934692,
-      "learning_rate": 5.8852261306532664e-05,
-      "loss": 5.6728,
-      "step": 41457
-    },
-    {
-      "epoch": 21.620860495436766,
-      "grad_norm": 1.503167748451233,
-      "learning_rate": 5.885125628140704e-05,
-      "loss": 5.271,
-      "step": 41458
-    },
-    {
-      "epoch": 21.621382007822685,
-      "grad_norm": 1.516230583190918,
-      "learning_rate": 5.8850251256281406e-05,
-      "loss": 5.2359,
-      "step": 41459
-    },
-    {
-      "epoch": 21.621903520208605,
-      "grad_norm": 1.5255297422409058,
-      "learning_rate": 5.8849246231155784e-05,
-      "loss": 5.5463,
-      "step": 41460
-    },
-    {
-      "epoch": 21.622425032594524,
-      "grad_norm": 1.4131778478622437,
-      "learning_rate": 5.8848241206030155e-05,
-      "loss": 5.7039,
-      "step": 41461
-    },
-    {
-      "epoch": 21.622946544980444,
-      "grad_norm": 1.4790581464767456,
-      "learning_rate": 5.884723618090453e-05,
-      "loss": 5.4384,
-      "step": 41462
-    },
-    {
-      "epoch": 21.623468057366363,
-      "grad_norm": 1.5140877962112427,
-      "learning_rate": 5.88462311557789e-05,
-      "loss": 5.2616,
-      "step": 41463
-    },
-    {
-      "epoch": 21.623989569752283,
-      "grad_norm": 1.440026879310608,
-      "learning_rate": 5.884522613065326e-05,
-      "loss": 5.756,
-      "step": 41464
-    },
-    {
-      "epoch": 21.624511082138202,
-      "grad_norm": 1.4127751588821411,
-      "learning_rate": 5.884422110552764e-05,
-      "loss": 5.8806,
-      "step": 41465
-    },
-    {
-      "epoch": 21.62503259452412,
-      "grad_norm": 1.499885082244873,
-      "learning_rate": 5.884321608040201e-05,
-      "loss": 5.6502,
-      "step": 41466
-    },
-    {
-      "epoch": 21.625554106910037,
-      "grad_norm": 1.549926996231079,
-      "learning_rate": 5.884221105527639e-05,
-      "loss": 5.5028,
-      "step": 41467
-    },
-    {
-      "epoch": 21.626075619295957,
-      "grad_norm": 1.4875452518463135,
-      "learning_rate": 5.884120603015075e-05,
-      "loss": 5.2259,
-      "step": 41468
-    },
-    {
-      "epoch": 21.626597131681876,
-      "grad_norm": 1.4460562467575073,
-      "learning_rate": 5.884020100502513e-05,
-      "loss": 5.5204,
-      "step": 41469
-    },
-    {
-      "epoch": 21.627118644067796,
-      "grad_norm": 1.6393229961395264,
-      "learning_rate": 5.8839195979899495e-05,
-      "loss": 5.0958,
-      "step": 41470
-    },
-    {
-      "epoch": 21.627640156453715,
-      "grad_norm": 1.4342387914657593,
-      "learning_rate": 5.883819095477387e-05,
-      "loss": 5.5004,
-      "step": 41471
-    },
-    {
-      "epoch": 21.628161668839635,
-      "grad_norm": 1.622225284576416,
-      "learning_rate": 5.8837185929648243e-05,
-      "loss": 4.9858,
-      "step": 41472
-    },
-    {
-      "epoch": 21.628683181225554,
-      "grad_norm": 1.5272003412246704,
-      "learning_rate": 5.883618090452262e-05,
-      "loss": 5.5978,
-      "step": 41473
-    },
-    {
-      "epoch": 21.629204693611474,
-      "grad_norm": 1.5245200395584106,
-      "learning_rate": 5.8835175879396986e-05,
-      "loss": 4.8239,
-      "step": 41474
-    },
-    {
-      "epoch": 21.629726205997393,
-      "grad_norm": 1.5967583656311035,
-      "learning_rate": 5.883417085427136e-05,
-      "loss": 4.9959,
-      "step": 41475
-    },
-    {
-      "epoch": 21.630247718383313,
-      "grad_norm": 1.5394117832183838,
-      "learning_rate": 5.8833165829145734e-05,
-      "loss": 5.088,
-      "step": 41476
-    },
-    {
-      "epoch": 21.630769230769232,
-      "grad_norm": 1.521960973739624,
-      "learning_rate": 5.88321608040201e-05,
-      "loss": 5.3275,
-      "step": 41477
-    },
-    {
-      "epoch": 21.63129074315515,
-      "grad_norm": 1.5794556140899658,
-      "learning_rate": 5.8831155778894477e-05,
-      "loss": 5.0139,
-      "step": 41478
-    },
-    {
-      "epoch": 21.631812255541067,
-      "grad_norm": 1.5904173851013184,
-      "learning_rate": 5.883015075376884e-05,
-      "loss": 4.9991,
-      "step": 41479
-    },
-    {
-      "epoch": 21.632333767926987,
-      "grad_norm": 1.6401469707489014,
-      "learning_rate": 5.882914572864322e-05,
-      "loss": 4.8051,
-      "step": 41480
-    },
-    {
-      "epoch": 21.632855280312906,
-      "grad_norm": 1.5257781744003296,
-      "learning_rate": 5.882814070351759e-05,
-      "loss": 5.6436,
-      "step": 41481
-    },
-    {
-      "epoch": 21.633376792698826,
-      "grad_norm": 1.5611255168914795,
-      "learning_rate": 5.882713567839197e-05,
-      "loss": 5.3355,
-      "step": 41482
-    },
-    {
-      "epoch": 21.633898305084745,
-      "grad_norm": 1.5504740476608276,
-      "learning_rate": 5.882613065326633e-05,
-      "loss": 5.2409,
-      "step": 41483
-    },
-    {
-      "epoch": 21.634419817470665,
-      "grad_norm": 1.5472609996795654,
-      "learning_rate": 5.882512562814071e-05,
-      "loss": 5.0568,
-      "step": 41484
-    },
-    {
-      "epoch": 21.634941329856584,
-      "grad_norm": 1.612432599067688,
-      "learning_rate": 5.8824120603015074e-05,
-      "loss": 4.7758,
-      "step": 41485
-    },
-    {
-      "epoch": 21.635462842242504,
-      "grad_norm": 1.4828989505767822,
-      "learning_rate": 5.882311557788945e-05,
-      "loss": 5.2986,
-      "step": 41486
-    },
-    {
-      "epoch": 21.635984354628423,
-      "grad_norm": 1.5615484714508057,
-      "learning_rate": 5.882211055276382e-05,
-      "loss": 5.2926,
-      "step": 41487
-    },
-    {
-      "epoch": 21.636505867014343,
-      "grad_norm": 1.6252198219299316,
-      "learning_rate": 5.88211055276382e-05,
-      "loss": 5.1225,
-      "step": 41488
-    },
-    {
-      "epoch": 21.637027379400262,
-      "grad_norm": 1.58849036693573,
-      "learning_rate": 5.8820100502512565e-05,
-      "loss": 5.5416,
-      "step": 41489
-    },
-    {
-      "epoch": 21.63754889178618,
-      "grad_norm": 1.4664409160614014,
-      "learning_rate": 5.881909547738694e-05,
-      "loss": 5.5028,
-      "step": 41490
-    },
-    {
-      "epoch": 21.638070404172097,
-      "grad_norm": 1.5577236413955688,
-      "learning_rate": 5.881809045226131e-05,
-      "loss": 5.353,
-      "step": 41491
-    },
-    {
-      "epoch": 21.638591916558017,
-      "grad_norm": 1.4853779077529907,
-      "learning_rate": 5.881708542713568e-05,
-      "loss": 5.2698,
-      "step": 41492
-    },
-    {
-      "epoch": 21.639113428943936,
-      "grad_norm": 1.490096926689148,
-      "learning_rate": 5.8816080402010056e-05,
-      "loss": 5.2485,
-      "step": 41493
-    },
-    {
-      "epoch": 21.639634941329856,
-      "grad_norm": 1.630578637123108,
-      "learning_rate": 5.881507537688442e-05,
-      "loss": 4.6888,
-      "step": 41494
-    },
-    {
-      "epoch": 21.640156453715775,
-      "grad_norm": 1.5800950527191162,
-      "learning_rate": 5.88140703517588e-05,
-      "loss": 4.942,
-      "step": 41495
-    },
-    {
-      "epoch": 21.640677966101695,
-      "grad_norm": 1.5310500860214233,
-      "learning_rate": 5.881306532663316e-05,
-      "loss": 5.3328,
-      "step": 41496
-    },
-    {
-      "epoch": 21.641199478487614,
-      "grad_norm": 1.4441509246826172,
-      "learning_rate": 5.881206030150754e-05,
-      "loss": 5.2632,
-      "step": 41497
-    },
-    {
-      "epoch": 21.641720990873534,
-      "grad_norm": 1.4731454849243164,
-      "learning_rate": 5.881105527638191e-05,
-      "loss": 5.4027,
-      "step": 41498
-    },
-    {
-      "epoch": 21.642242503259453,
-      "grad_norm": 1.4906498193740845,
-      "learning_rate": 5.881005025125629e-05,
-      "loss": 5.4936,
-      "step": 41499
-    },
-    {
-      "epoch": 21.642764015645373,
-      "grad_norm": 1.5505999326705933,
-      "learning_rate": 5.880904522613065e-05,
-      "loss": 4.9277,
-      "step": 41500
-    },
-    {
-      "epoch": 21.643285528031292,
-      "grad_norm": 1.5863078832626343,
-      "learning_rate": 5.880804020100503e-05,
-      "loss": 5.4625,
-      "step": 41501
-    },
-    {
-      "epoch": 21.64380704041721,
-      "grad_norm": 1.4908655881881714,
-      "learning_rate": 5.88070351758794e-05,
-      "loss": 5.541,
-      "step": 41502
-    },
-    {
-      "epoch": 21.644328552803128,
-      "grad_norm": 1.560986042022705,
-      "learning_rate": 5.880603015075378e-05,
-      "loss": 5.3541,
-      "step": 41503
-    },
-    {
-      "epoch": 21.644850065189047,
-      "grad_norm": 1.5521982908248901,
-      "learning_rate": 5.8805025125628144e-05,
-      "loss": 5.1449,
-      "step": 41504
-    },
-    {
-      "epoch": 21.645371577574966,
-      "grad_norm": 1.5278199911117554,
-      "learning_rate": 5.880402010050251e-05,
-      "loss": 5.0264,
-      "step": 41505
-    },
-    {
-      "epoch": 21.645893089960886,
-      "grad_norm": 1.6114189624786377,
-      "learning_rate": 5.8803015075376886e-05,
-      "loss": 4.7129,
-      "step": 41506
-    },
-    {
-      "epoch": 21.646414602346805,
-      "grad_norm": 1.5774400234222412,
-      "learning_rate": 5.880201005025126e-05,
-      "loss": 5.4961,
-      "step": 41507
-    },
-    {
-      "epoch": 21.646936114732725,
-      "grad_norm": 1.488282561302185,
-      "learning_rate": 5.8801005025125635e-05,
-      "loss": 5.2712,
-      "step": 41508
-    },
-    {
-      "epoch": 21.647457627118644,
-      "grad_norm": 1.5817421674728394,
-      "learning_rate": 5.88e-05,
-      "loss": 4.9787,
-      "step": 41509
-    },
-    {
-      "epoch": 21.647979139504564,
-      "grad_norm": 1.458275318145752,
-      "learning_rate": 5.879899497487438e-05,
-      "loss": 4.8619,
-      "step": 41510
-    },
-    {
-      "epoch": 21.648500651890483,
-      "grad_norm": 1.4415987730026245,
-      "learning_rate": 5.879798994974874e-05,
-      "loss": 5.6537,
-      "step": 41511
-    },
-    {
-      "epoch": 21.649022164276403,
-      "grad_norm": 1.614202618598938,
-      "learning_rate": 5.879698492462312e-05,
-      "loss": 4.9395,
-      "step": 41512
-    },
-    {
-      "epoch": 21.649543676662322,
-      "grad_norm": 1.5437053442001343,
-      "learning_rate": 5.879597989949749e-05,
-      "loss": 5.349,
-      "step": 41513
-    },
-    {
-      "epoch": 21.65006518904824,
-      "grad_norm": 1.5099420547485352,
-      "learning_rate": 5.879497487437187e-05,
-      "loss": 5.148,
-      "step": 41514
-    },
-    {
-      "epoch": 21.650586701434158,
-      "grad_norm": 1.5536085367202759,
-      "learning_rate": 5.879396984924623e-05,
-      "loss": 5.5164,
-      "step": 41515
-    },
-    {
-      "epoch": 21.651108213820077,
-      "grad_norm": 1.562759518623352,
-      "learning_rate": 5.879296482412061e-05,
-      "loss": 4.7856,
-      "step": 41516
-    },
-    {
-      "epoch": 21.651629726205996,
-      "grad_norm": 1.6061336994171143,
-      "learning_rate": 5.8791959798994975e-05,
-      "loss": 5.3023,
-      "step": 41517
-    },
-    {
-      "epoch": 21.652151238591916,
-      "grad_norm": 1.6152944564819336,
-      "learning_rate": 5.8790954773869346e-05,
-      "loss": 5.3574,
-      "step": 41518
-    },
-    {
-      "epoch": 21.652672750977835,
-      "grad_norm": 1.4823570251464844,
-      "learning_rate": 5.8789949748743724e-05,
-      "loss": 5.5355,
-      "step": 41519
-    },
-    {
-      "epoch": 21.653194263363755,
-      "grad_norm": 1.7027454376220703,
-      "learning_rate": 5.878894472361809e-05,
-      "loss": 5.1259,
-      "step": 41520
-    },
-    {
-      "epoch": 21.653715775749674,
-      "grad_norm": 1.5516915321350098,
-      "learning_rate": 5.8787939698492466e-05,
-      "loss": 5.3549,
-      "step": 41521
-    },
-    {
-      "epoch": 21.654237288135594,
-      "grad_norm": 1.4874484539031982,
-      "learning_rate": 5.878693467336684e-05,
-      "loss": 5.3714,
-      "step": 41522
-    },
-    {
-      "epoch": 21.654758800521513,
-      "grad_norm": 1.5007872581481934,
-      "learning_rate": 5.8785929648241215e-05,
-      "loss": 5.3796,
-      "step": 41523
-    },
-    {
-      "epoch": 21.655280312907433,
-      "grad_norm": 1.6526179313659668,
-      "learning_rate": 5.878492462311558e-05,
-      "loss": 5.0339,
-      "step": 41524
-    },
-    {
-      "epoch": 21.655801825293352,
-      "grad_norm": 1.4845941066741943,
-      "learning_rate": 5.878391959798996e-05,
-      "loss": 5.3272,
-      "step": 41525
-    },
-    {
-      "epoch": 21.656323337679268,
-      "grad_norm": 1.5386086702346802,
-      "learning_rate": 5.878291457286432e-05,
-      "loss": 5.3645,
-      "step": 41526
-    },
-    {
-      "epoch": 21.656844850065188,
-      "grad_norm": 1.5495390892028809,
-      "learning_rate": 5.87819095477387e-05,
-      "loss": 5.5022,
-      "step": 41527
-    },
-    {
-      "epoch": 21.657366362451107,
-      "grad_norm": 1.5641697645187378,
-      "learning_rate": 5.878090452261307e-05,
-      "loss": 5.1136,
-      "step": 41528
-    },
-    {
-      "epoch": 21.657887874837026,
-      "grad_norm": 1.6152046918869019,
-      "learning_rate": 5.877989949748745e-05,
-      "loss": 5.3907,
-      "step": 41529
-    },
-    {
-      "epoch": 21.658409387222946,
-      "grad_norm": 1.566256046295166,
-      "learning_rate": 5.877889447236181e-05,
-      "loss": 4.9571,
-      "step": 41530
-    },
-    {
-      "epoch": 21.658930899608865,
-      "grad_norm": 1.5805747509002686,
-      "learning_rate": 5.8777889447236176e-05,
-      "loss": 5.4289,
-      "step": 41531
-    },
-    {
-      "epoch": 21.659452411994785,
-      "grad_norm": 1.4535826444625854,
-      "learning_rate": 5.8776884422110554e-05,
-      "loss": 5.2906,
-      "step": 41532
-    },
-    {
-      "epoch": 21.659973924380704,
-      "grad_norm": 1.5007268190383911,
-      "learning_rate": 5.8775879396984925e-05,
-      "loss": 5.5228,
-      "step": 41533
-    },
-    {
-      "epoch": 21.660495436766624,
-      "grad_norm": 1.4440490007400513,
-      "learning_rate": 5.87748743718593e-05,
-      "loss": 5.5366,
-      "step": 41534
-    },
-    {
-      "epoch": 21.661016949152543,
-      "grad_norm": 1.4966466426849365,
-      "learning_rate": 5.877386934673367e-05,
-      "loss": 5.1826,
-      "step": 41535
-    },
-    {
-      "epoch": 21.661538461538463,
-      "grad_norm": 1.5930553674697876,
-      "learning_rate": 5.8772864321608045e-05,
-      "loss": 5.0775,
-      "step": 41536
-    },
-    {
-      "epoch": 21.662059973924382,
-      "grad_norm": 1.4895614385604858,
-      "learning_rate": 5.877185929648241e-05,
-      "loss": 5.1094,
-      "step": 41537
-    },
-    {
-      "epoch": 21.6625814863103,
-      "grad_norm": 1.3979547023773193,
-      "learning_rate": 5.877085427135679e-05,
-      "loss": 5.7944,
-      "step": 41538
-    },
-    {
-      "epoch": 21.663102998696218,
-      "grad_norm": 1.588274359703064,
-      "learning_rate": 5.876984924623116e-05,
-      "loss": 4.9553,
-      "step": 41539
-    },
-    {
-      "epoch": 21.663624511082137,
-      "grad_norm": 1.3533085584640503,
-      "learning_rate": 5.8768844221105536e-05,
-      "loss": 4.8717,
-      "step": 41540
-    },
-    {
-      "epoch": 21.664146023468057,
-      "grad_norm": 1.5132156610488892,
-      "learning_rate": 5.87678391959799e-05,
-      "loss": 5.631,
-      "step": 41541
-    },
-    {
-      "epoch": 21.664667535853976,
-      "grad_norm": 1.5336025953292847,
-      "learning_rate": 5.876683417085428e-05,
-      "loss": 5.3086,
-      "step": 41542
-    },
-    {
-      "epoch": 21.665189048239895,
-      "grad_norm": 1.6147290468215942,
-      "learning_rate": 5.876582914572865e-05,
-      "loss": 5.1867,
-      "step": 41543
-    },
-    {
-      "epoch": 21.665710560625815,
-      "grad_norm": 1.463538408279419,
-      "learning_rate": 5.876482412060301e-05,
-      "loss": 5.3344,
-      "step": 41544
-    },
-    {
-      "epoch": 21.666232073011734,
-      "grad_norm": 1.5623269081115723,
-      "learning_rate": 5.876381909547739e-05,
-      "loss": 5.3158,
-      "step": 41545
-    },
-    {
-      "epoch": 21.666753585397654,
-      "grad_norm": 1.6242740154266357,
-      "learning_rate": 5.8762814070351755e-05,
-      "loss": 5.3291,
-      "step": 41546
-    },
-    {
-      "epoch": 21.667275097783573,
-      "grad_norm": 1.5234125852584839,
-      "learning_rate": 5.876180904522613e-05,
-      "loss": 4.8262,
-      "step": 41547
-    },
-    {
-      "epoch": 21.667796610169493,
-      "grad_norm": 1.5326701402664185,
-      "learning_rate": 5.8760804020100504e-05,
-      "loss": 4.5862,
-      "step": 41548
-    },
-    {
-      "epoch": 21.668318122555412,
-      "grad_norm": 1.6132596731185913,
-      "learning_rate": 5.875979899497488e-05,
-      "loss": 5.0954,
-      "step": 41549
-    },
-    {
-      "epoch": 21.668839634941328,
-      "grad_norm": 1.5785390138626099,
-      "learning_rate": 5.8758793969849246e-05,
-      "loss": 5.1461,
-      "step": 41550
-    },
-    {
-      "epoch": 21.669361147327248,
-      "grad_norm": 1.5764044523239136,
-      "learning_rate": 5.8757788944723624e-05,
-      "loss": 5.2634,
-      "step": 41551
-    },
-    {
-      "epoch": 21.669882659713167,
-      "grad_norm": 1.587061882019043,
-      "learning_rate": 5.875678391959799e-05,
-      "loss": 4.8855,
-      "step": 41552
-    },
-    {
-      "epoch": 21.670404172099087,
-      "grad_norm": 1.5338783264160156,
-      "learning_rate": 5.8755778894472366e-05,
-      "loss": 4.9377,
-      "step": 41553
-    },
-    {
-      "epoch": 21.670925684485006,
-      "grad_norm": 1.5670100450515747,
-      "learning_rate": 5.875477386934674e-05,
-      "loss": 5.1425,
-      "step": 41554
-    },
-    {
-      "epoch": 21.671447196870925,
-      "grad_norm": 1.5008275508880615,
-      "learning_rate": 5.8753768844221115e-05,
-      "loss": 5.2058,
-      "step": 41555
-    },
-    {
-      "epoch": 21.671968709256845,
-      "grad_norm": 1.5660088062286377,
-      "learning_rate": 5.875276381909548e-05,
-      "loss": 4.8861,
-      "step": 41556
-    },
-    {
-      "epoch": 21.672490221642764,
-      "grad_norm": 1.652713656425476,
-      "learning_rate": 5.8751758793969844e-05,
-      "loss": 4.2282,
-      "step": 41557
-    },
-    {
-      "epoch": 21.673011734028684,
-      "grad_norm": 1.5557433366775513,
-      "learning_rate": 5.875075376884422e-05,
-      "loss": 5.0507,
-      "step": 41558
-    },
-    {
-      "epoch": 21.673533246414603,
-      "grad_norm": 1.505203366279602,
-      "learning_rate": 5.874974874371859e-05,
-      "loss": 5.2477,
-      "step": 41559
-    },
-    {
-      "epoch": 21.674054758800523,
-      "grad_norm": 1.6316397190093994,
-      "learning_rate": 5.874874371859297e-05,
-      "loss": 5.0936,
-      "step": 41560
-    },
-    {
-      "epoch": 21.674576271186442,
-      "grad_norm": 1.431463360786438,
-      "learning_rate": 5.8747738693467335e-05,
-      "loss": 5.566,
-      "step": 41561
-    },
-    {
-      "epoch": 21.675097783572358,
-      "grad_norm": 1.481339454650879,
-      "learning_rate": 5.874673366834171e-05,
-      "loss": 4.68,
-      "step": 41562
-    },
-    {
-      "epoch": 21.675619295958278,
-      "grad_norm": 1.456303358078003,
-      "learning_rate": 5.874572864321608e-05,
-      "loss": 5.2258,
-      "step": 41563
-    },
-    {
-      "epoch": 21.676140808344197,
-      "grad_norm": 1.4916270971298218,
-      "learning_rate": 5.8744723618090455e-05,
-      "loss": 5.338,
-      "step": 41564
-    },
-    {
-      "epoch": 21.676662320730117,
-      "grad_norm": 1.5957539081573486,
-      "learning_rate": 5.8743718592964826e-05,
-      "loss": 4.7849,
-      "step": 41565
-    },
-    {
-      "epoch": 21.677183833116036,
-      "grad_norm": 1.6977516412734985,
-      "learning_rate": 5.8742713567839204e-05,
-      "loss": 5.4997,
-      "step": 41566
-    },
-    {
-      "epoch": 21.677705345501955,
-      "grad_norm": 1.4743386507034302,
-      "learning_rate": 5.874170854271357e-05,
-      "loss": 5.1304,
-      "step": 41567
-    },
-    {
-      "epoch": 21.678226857887875,
-      "grad_norm": 1.4511635303497314,
-      "learning_rate": 5.8740703517587946e-05,
-      "loss": 5.2283,
-      "step": 41568
-    },
-    {
-      "epoch": 21.678748370273794,
-      "grad_norm": 1.4672003984451294,
-      "learning_rate": 5.873969849246232e-05,
-      "loss": 5.5206,
-      "step": 41569
-    },
-    {
-      "epoch": 21.679269882659714,
-      "grad_norm": 1.6245176792144775,
-      "learning_rate": 5.8738693467336695e-05,
-      "loss": 4.8676,
-      "step": 41570
-    },
-    {
-      "epoch": 21.679791395045633,
-      "grad_norm": 1.6278373003005981,
-      "learning_rate": 5.873768844221106e-05,
-      "loss": 5.0047,
-      "step": 41571
-    },
-    {
-      "epoch": 21.680312907431553,
-      "grad_norm": 1.533719778060913,
-      "learning_rate": 5.873668341708542e-05,
-      "loss": 5.4471,
-      "step": 41572
-    },
-    {
-      "epoch": 21.680834419817472,
-      "grad_norm": 1.5350322723388672,
-      "learning_rate": 5.87356783919598e-05,
-      "loss": 5.7294,
-      "step": 41573
-    },
-    {
-      "epoch": 21.681355932203388,
-      "grad_norm": 1.6318467855453491,
-      "learning_rate": 5.873467336683417e-05,
-      "loss": 5.2185,
-      "step": 41574
-    },
-    {
-      "epoch": 21.681877444589308,
-      "grad_norm": 1.5409212112426758,
-      "learning_rate": 5.873366834170855e-05,
-      "loss": 5.0666,
-      "step": 41575
-    },
-    {
-      "epoch": 21.682398956975227,
-      "grad_norm": 1.5142617225646973,
-      "learning_rate": 5.8732663316582914e-05,
-      "loss": 5.1279,
-      "step": 41576
-    },
-    {
-      "epoch": 21.682920469361147,
-      "grad_norm": 1.5297150611877441,
-      "learning_rate": 5.873165829145729e-05,
-      "loss": 5.4343,
-      "step": 41577
-    },
-    {
-      "epoch": 21.683441981747066,
-      "grad_norm": 1.5091995000839233,
-      "learning_rate": 5.8730653266331656e-05,
-      "loss": 5.4554,
-      "step": 41578
-    },
-    {
-      "epoch": 21.683963494132986,
-      "grad_norm": 1.5110188722610474,
-      "learning_rate": 5.8729648241206034e-05,
-      "loss": 5.3325,
-      "step": 41579
-    },
-    {
-      "epoch": 21.684485006518905,
-      "grad_norm": 1.6137596368789673,
-      "learning_rate": 5.8728643216080405e-05,
-      "loss": 4.6743,
-      "step": 41580
-    },
-    {
-      "epoch": 21.685006518904824,
-      "grad_norm": 1.5791412591934204,
-      "learning_rate": 5.872763819095478e-05,
-      "loss": 5.3418,
-      "step": 41581
-    },
-    {
-      "epoch": 21.685528031290744,
-      "grad_norm": 1.5611743927001953,
-      "learning_rate": 5.872663316582915e-05,
-      "loss": 5.1687,
-      "step": 41582
-    },
-    {
-      "epoch": 21.686049543676663,
-      "grad_norm": 1.37284517288208,
-      "learning_rate": 5.8725628140703525e-05,
-      "loss": 5.5268,
-      "step": 41583
-    },
-    {
-      "epoch": 21.686571056062583,
-      "grad_norm": 1.5076850652694702,
-      "learning_rate": 5.872462311557789e-05,
-      "loss": 5.4081,
-      "step": 41584
-    },
-    {
-      "epoch": 21.687092568448502,
-      "grad_norm": 1.4455592632293701,
-      "learning_rate": 5.872361809045226e-05,
-      "loss": 5.8342,
-      "step": 41585
-    },
-    {
-      "epoch": 21.687614080834418,
-      "grad_norm": 1.5059542655944824,
-      "learning_rate": 5.872261306532664e-05,
-      "loss": 5.2419,
-      "step": 41586
-    },
-    {
-      "epoch": 21.688135593220338,
-      "grad_norm": 1.5242027044296265,
-      "learning_rate": 5.8721608040201e-05,
-      "loss": 5.2147,
-      "step": 41587
-    },
-    {
-      "epoch": 21.688657105606257,
-      "grad_norm": 1.5656870603561401,
-      "learning_rate": 5.872060301507538e-05,
-      "loss": 4.9617,
-      "step": 41588
-    },
-    {
-      "epoch": 21.689178617992177,
-      "grad_norm": 1.5875540971755981,
-      "learning_rate": 5.871959798994975e-05,
-      "loss": 5.3399,
-      "step": 41589
-    },
-    {
-      "epoch": 21.689700130378096,
-      "grad_norm": 1.5428742170333862,
-      "learning_rate": 5.871859296482413e-05,
-      "loss": 5.0035,
-      "step": 41590
-    },
-    {
-      "epoch": 21.690221642764016,
-      "grad_norm": 1.5891075134277344,
-      "learning_rate": 5.8717587939698493e-05,
-      "loss": 5.0829,
-      "step": 41591
-    },
-    {
-      "epoch": 21.690743155149935,
-      "grad_norm": 1.4274256229400635,
-      "learning_rate": 5.871658291457287e-05,
-      "loss": 5.3249,
-      "step": 41592
-    },
-    {
-      "epoch": 21.691264667535854,
-      "grad_norm": 1.5292534828186035,
-      "learning_rate": 5.8715577889447236e-05,
-      "loss": 5.4152,
-      "step": 41593
-    },
-    {
-      "epoch": 21.691786179921774,
-      "grad_norm": 1.5141563415527344,
-      "learning_rate": 5.8714572864321613e-05,
-      "loss": 5.2902,
-      "step": 41594
-    },
-    {
-      "epoch": 21.692307692307693,
-      "grad_norm": 1.4516974687576294,
-      "learning_rate": 5.8713567839195984e-05,
-      "loss": 5.1744,
-      "step": 41595
-    },
-    {
-      "epoch": 21.692829204693613,
-      "grad_norm": 1.49988853931427,
-      "learning_rate": 5.871256281407036e-05,
-      "loss": 5.1353,
-      "step": 41596
-    },
-    {
-      "epoch": 21.693350717079532,
-      "grad_norm": 1.5363038778305054,
-      "learning_rate": 5.8711557788944727e-05,
-      "loss": 5.1725,
-      "step": 41597
-    },
-    {
-      "epoch": 21.69387222946545,
-      "grad_norm": 1.5775610208511353,
-      "learning_rate": 5.871055276381909e-05,
-      "loss": 5.1296,
-      "step": 41598
-    },
-    {
-      "epoch": 21.694393741851368,
-      "grad_norm": 1.363779902458191,
-      "learning_rate": 5.870954773869347e-05,
-      "loss": 4.7818,
-      "step": 41599
-    },
-    {
-      "epoch": 21.694915254237287,
-      "grad_norm": 1.5950227975845337,
-      "learning_rate": 5.870854271356784e-05,
-      "loss": 5.2163,
-      "step": 41600
-    },
-    {
-      "epoch": 21.695436766623207,
-      "grad_norm": 1.561263918876648,
-      "learning_rate": 5.870753768844222e-05,
-      "loss": 5.6703,
-      "step": 41601
-    },
-    {
-      "epoch": 21.695958279009126,
-      "grad_norm": 1.5532898902893066,
-      "learning_rate": 5.870653266331658e-05,
-      "loss": 5.2946,
-      "step": 41602
-    },
-    {
-      "epoch": 21.696479791395046,
-      "grad_norm": 1.4461994171142578,
-      "learning_rate": 5.870552763819096e-05,
-      "loss": 4.5124,
-      "step": 41603
-    },
-    {
-      "epoch": 21.697001303780965,
-      "grad_norm": 1.554764986038208,
-      "learning_rate": 5.8704522613065324e-05,
-      "loss": 5.2938,
-      "step": 41604
-    },
-    {
-      "epoch": 21.697522816166884,
-      "grad_norm": 1.4569101333618164,
-      "learning_rate": 5.87035175879397e-05,
-      "loss": 5.3154,
-      "step": 41605
-    },
-    {
-      "epoch": 21.698044328552804,
-      "grad_norm": 1.532118320465088,
-      "learning_rate": 5.870251256281407e-05,
-      "loss": 5.3158,
-      "step": 41606
-    },
-    {
-      "epoch": 21.698565840938723,
-      "grad_norm": 1.5417871475219727,
-      "learning_rate": 5.870150753768845e-05,
-      "loss": 5.3223,
-      "step": 41607
-    },
-    {
-      "epoch": 21.699087353324643,
-      "grad_norm": 1.534792423248291,
-      "learning_rate": 5.8700502512562815e-05,
-      "loss": 5.2436,
-      "step": 41608
-    },
-    {
-      "epoch": 21.69960886571056,
-      "grad_norm": 1.5698204040527344,
-      "learning_rate": 5.869949748743719e-05,
-      "loss": 5.2673,
-      "step": 41609
-    },
-    {
-      "epoch": 21.70013037809648,
-      "grad_norm": 1.601309061050415,
-      "learning_rate": 5.8698492462311564e-05,
-      "loss": 5.0187,
-      "step": 41610
-    },
-    {
-      "epoch": 21.700651890482398,
-      "grad_norm": 1.5326530933380127,
-      "learning_rate": 5.869748743718593e-05,
-      "loss": 5.0773,
-      "step": 41611
-    },
-    {
-      "epoch": 21.701173402868317,
-      "grad_norm": 1.4165189266204834,
-      "learning_rate": 5.8696482412060306e-05,
-      "loss": 5.7365,
-      "step": 41612
-    },
-    {
-      "epoch": 21.701694915254237,
-      "grad_norm": 1.4711717367172241,
-      "learning_rate": 5.869547738693467e-05,
-      "loss": 5.0257,
-      "step": 41613
-    },
-    {
-      "epoch": 21.702216427640156,
-      "grad_norm": 1.440752625465393,
-      "learning_rate": 5.869447236180905e-05,
-      "loss": 5.2014,
-      "step": 41614
-    },
-    {
-      "epoch": 21.702737940026076,
-      "grad_norm": 1.5573344230651855,
-      "learning_rate": 5.869346733668342e-05,
-      "loss": 4.7664,
-      "step": 41615
-    },
-    {
-      "epoch": 21.703259452411995,
-      "grad_norm": 1.523794412612915,
-      "learning_rate": 5.86924623115578e-05,
-      "loss": 5.0491,
-      "step": 41616
-    },
-    {
-      "epoch": 21.703780964797915,
-      "grad_norm": 1.498833417892456,
-      "learning_rate": 5.869145728643216e-05,
-      "loss": 5.6219,
-      "step": 41617
-    },
-    {
-      "epoch": 21.704302477183834,
-      "grad_norm": 1.5770301818847656,
-      "learning_rate": 5.869045226130654e-05,
-      "loss": 5.2613,
-      "step": 41618
-    },
-    {
-      "epoch": 21.704823989569753,
-      "grad_norm": 1.5487691164016724,
-      "learning_rate": 5.86894472361809e-05,
-      "loss": 5.4751,
-      "step": 41619
-    },
-    {
-      "epoch": 21.705345501955673,
-      "grad_norm": 1.5106180906295776,
-      "learning_rate": 5.868844221105528e-05,
-      "loss": 5.3268,
-      "step": 41620
-    },
-    {
-      "epoch": 21.705867014341592,
-      "grad_norm": 1.4114280939102173,
-      "learning_rate": 5.868743718592965e-05,
-      "loss": 5.1584,
-      "step": 41621
-    },
-    {
-      "epoch": 21.70638852672751,
-      "grad_norm": 1.5481749773025513,
-      "learning_rate": 5.868643216080403e-05,
-      "loss": 5.5984,
-      "step": 41622
-    },
-    {
-      "epoch": 21.706910039113428,
-      "grad_norm": 1.5671601295471191,
-      "learning_rate": 5.8685427135678394e-05,
-      "loss": 5.1909,
-      "step": 41623
-    },
-    {
-      "epoch": 21.707431551499347,
-      "grad_norm": 1.5201764106750488,
-      "learning_rate": 5.868442211055276e-05,
-      "loss": 5.2519,
-      "step": 41624
-    },
-    {
-      "epoch": 21.707953063885267,
-      "grad_norm": 1.474036693572998,
-      "learning_rate": 5.8683417085427136e-05,
-      "loss": 5.602,
-      "step": 41625
-    },
-    {
-      "epoch": 21.708474576271186,
-      "grad_norm": 1.5074152946472168,
-      "learning_rate": 5.868241206030151e-05,
-      "loss": 5.5343,
-      "step": 41626
-    },
-    {
-      "epoch": 21.708996088657106,
-      "grad_norm": 1.5228127241134644,
-      "learning_rate": 5.8681407035175885e-05,
-      "loss": 5.1297,
-      "step": 41627
-    },
-    {
-      "epoch": 21.709517601043025,
-      "grad_norm": 1.5907186269760132,
-      "learning_rate": 5.868040201005025e-05,
-      "loss": 5.2322,
-      "step": 41628
-    },
-    {
-      "epoch": 21.710039113428945,
-      "grad_norm": 1.4514117240905762,
-      "learning_rate": 5.867939698492463e-05,
-      "loss": 5.5359,
-      "step": 41629
-    },
-    {
-      "epoch": 21.710560625814864,
-      "grad_norm": 1.6561555862426758,
-      "learning_rate": 5.8678391959799e-05,
-      "loss": 4.887,
-      "step": 41630
-    },
-    {
-      "epoch": 21.711082138200783,
-      "grad_norm": 1.5670137405395508,
-      "learning_rate": 5.8677386934673376e-05,
-      "loss": 5.1571,
-      "step": 41631
-    },
-    {
-      "epoch": 21.711603650586703,
-      "grad_norm": 1.5550730228424072,
-      "learning_rate": 5.867638190954774e-05,
-      "loss": 4.9928,
-      "step": 41632
-    },
-    {
-      "epoch": 21.71212516297262,
-      "grad_norm": 1.4351286888122559,
-      "learning_rate": 5.867537688442212e-05,
-      "loss": 5.2484,
-      "step": 41633
-    },
-    {
-      "epoch": 21.71264667535854,
-      "grad_norm": 2.0400471687316895,
-      "learning_rate": 5.867437185929648e-05,
-      "loss": 5.1955,
-      "step": 41634
-    },
-    {
-      "epoch": 21.713168187744458,
-      "grad_norm": 1.6497867107391357,
-      "learning_rate": 5.867336683417086e-05,
-      "loss": 5.0971,
-      "step": 41635
-    },
-    {
-      "epoch": 21.713689700130377,
-      "grad_norm": 1.5154831409454346,
-      "learning_rate": 5.867236180904523e-05,
-      "loss": 5.1647,
-      "step": 41636
-    },
-    {
-      "epoch": 21.714211212516297,
-      "grad_norm": 1.4534724950790405,
-      "learning_rate": 5.8671356783919596e-05,
-      "loss": 5.7118,
-      "step": 41637
-    },
-    {
-      "epoch": 21.714732724902216,
-      "grad_norm": 1.5982604026794434,
-      "learning_rate": 5.8670351758793974e-05,
-      "loss": 4.9627,
-      "step": 41638
-    },
-    {
-      "epoch": 21.715254237288136,
-      "grad_norm": 1.5213985443115234,
-      "learning_rate": 5.866934673366834e-05,
-      "loss": 5.1192,
-      "step": 41639
-    },
-    {
-      "epoch": 21.715775749674055,
-      "grad_norm": 1.5826442241668701,
-      "learning_rate": 5.8668341708542716e-05,
-      "loss": 5.801,
-      "step": 41640
-    },
-    {
-      "epoch": 21.716297262059975,
-      "grad_norm": 1.467716097831726,
-      "learning_rate": 5.866733668341709e-05,
-      "loss": 5.3388,
-      "step": 41641
-    },
-    {
-      "epoch": 21.716818774445894,
-      "grad_norm": 1.5774949789047241,
-      "learning_rate": 5.8666331658291465e-05,
-      "loss": 5.0351,
-      "step": 41642
-    },
-    {
-      "epoch": 21.717340286831814,
-      "grad_norm": 1.4917237758636475,
-      "learning_rate": 5.866532663316583e-05,
-      "loss": 5.4311,
-      "step": 41643
-    },
-    {
-      "epoch": 21.717861799217733,
-      "grad_norm": 1.6009315252304077,
-      "learning_rate": 5.866432160804021e-05,
-      "loss": 4.9347,
-      "step": 41644
-    },
-    {
-      "epoch": 21.71838331160365,
-      "grad_norm": 1.579338550567627,
-      "learning_rate": 5.866331658291457e-05,
-      "loss": 4.8105,
-      "step": 41645
-    },
-    {
-      "epoch": 21.71890482398957,
-      "grad_norm": 1.5788494348526,
-      "learning_rate": 5.866231155778895e-05,
-      "loss": 5.0322,
-      "step": 41646
-    },
-    {
-      "epoch": 21.719426336375488,
-      "grad_norm": 1.458644151687622,
-      "learning_rate": 5.866130653266332e-05,
-      "loss": 4.8138,
-      "step": 41647
-    },
-    {
-      "epoch": 21.719947848761407,
-      "grad_norm": 1.5361034870147705,
-      "learning_rate": 5.86603015075377e-05,
-      "loss": 5.7437,
-      "step": 41648
-    },
-    {
-      "epoch": 21.720469361147327,
-      "grad_norm": 1.492159366607666,
-      "learning_rate": 5.865929648241206e-05,
-      "loss": 5.298,
-      "step": 41649
-    },
-    {
-      "epoch": 21.720990873533246,
-      "grad_norm": 1.527204990386963,
-      "learning_rate": 5.8658291457286426e-05,
-      "loss": 5.2036,
-      "step": 41650
-    },
-    {
-      "epoch": 21.721512385919166,
-      "grad_norm": 1.6302096843719482,
-      "learning_rate": 5.8657286432160804e-05,
-      "loss": 4.7824,
-      "step": 41651
-    },
-    {
-      "epoch": 21.722033898305085,
-      "grad_norm": 1.4915475845336914,
-      "learning_rate": 5.8656281407035175e-05,
-      "loss": 5.1224,
-      "step": 41652
-    },
-    {
-      "epoch": 21.722555410691005,
-      "grad_norm": 1.4843460321426392,
-      "learning_rate": 5.865527638190955e-05,
-      "loss": 4.8146,
-      "step": 41653
-    },
-    {
-      "epoch": 21.723076923076924,
-      "grad_norm": 1.5194110870361328,
-      "learning_rate": 5.865427135678392e-05,
-      "loss": 5.3773,
-      "step": 41654
-    },
-    {
-      "epoch": 21.723598435462844,
-      "grad_norm": 1.4112035036087036,
-      "learning_rate": 5.8653266331658295e-05,
-      "loss": 5.3597,
-      "step": 41655
-    },
-    {
-      "epoch": 21.724119947848763,
-      "grad_norm": 1.5047193765640259,
-      "learning_rate": 5.8652261306532666e-05,
-      "loss": 4.9178,
-      "step": 41656
-    },
-    {
-      "epoch": 21.72464146023468,
-      "grad_norm": 1.525019645690918,
-      "learning_rate": 5.8651256281407044e-05,
-      "loss": 5.6815,
-      "step": 41657
-    },
-    {
-      "epoch": 21.7251629726206,
-      "grad_norm": 1.4826734066009521,
-      "learning_rate": 5.865025125628141e-05,
-      "loss": 5.7038,
-      "step": 41658
-    },
-    {
-      "epoch": 21.725684485006518,
-      "grad_norm": 1.6101466417312622,
-      "learning_rate": 5.8649246231155786e-05,
-      "loss": 4.978,
-      "step": 41659
-    },
-    {
-      "epoch": 21.726205997392437,
-      "grad_norm": 1.5250028371810913,
-      "learning_rate": 5.864824120603015e-05,
-      "loss": 4.9208,
-      "step": 41660
-    },
-    {
-      "epoch": 21.726727509778357,
-      "grad_norm": 1.823823094367981,
-      "learning_rate": 5.864723618090453e-05,
-      "loss": 5.2502,
-      "step": 41661
-    },
-    {
-      "epoch": 21.727249022164276,
-      "grad_norm": 1.5998808145523071,
-      "learning_rate": 5.86462311557789e-05,
-      "loss": 5.2485,
-      "step": 41662
-    },
-    {
-      "epoch": 21.727770534550196,
-      "grad_norm": 1.6679456233978271,
-      "learning_rate": 5.864522613065328e-05,
-      "loss": 4.9546,
-      "step": 41663
-    },
-    {
-      "epoch": 21.728292046936115,
-      "grad_norm": 1.6930371522903442,
-      "learning_rate": 5.864422110552764e-05,
-      "loss": 5.2218,
-      "step": 41664
-    },
-    {
-      "epoch": 21.728813559322035,
-      "grad_norm": 1.6375067234039307,
-      "learning_rate": 5.8643216080402005e-05,
-      "loss": 5.2093,
-      "step": 41665
-    },
-    {
-      "epoch": 21.729335071707954,
-      "grad_norm": 1.603919267654419,
-      "learning_rate": 5.864221105527638e-05,
-      "loss": 4.7537,
-      "step": 41666
-    },
-    {
-      "epoch": 21.729856584093874,
-      "grad_norm": 1.5303230285644531,
-      "learning_rate": 5.8641206030150754e-05,
-      "loss": 5.3871,
-      "step": 41667
-    },
-    {
-      "epoch": 21.730378096479793,
-      "grad_norm": 1.5369701385498047,
-      "learning_rate": 5.864020100502513e-05,
-      "loss": 5.4282,
-      "step": 41668
-    },
-    {
-      "epoch": 21.73089960886571,
-      "grad_norm": 1.7466238737106323,
-      "learning_rate": 5.8639195979899496e-05,
-      "loss": 5.4601,
-      "step": 41669
-    },
-    {
-      "epoch": 21.73142112125163,
-      "grad_norm": 1.6249661445617676,
-      "learning_rate": 5.8638190954773874e-05,
-      "loss": 5.259,
-      "step": 41670
-    },
-    {
-      "epoch": 21.731942633637548,
-      "grad_norm": 1.6299608945846558,
-      "learning_rate": 5.863718592964824e-05,
-      "loss": 5.1786,
-      "step": 41671
-    },
-    {
-      "epoch": 21.732464146023467,
-      "grad_norm": 1.4250259399414062,
-      "learning_rate": 5.8636180904522616e-05,
-      "loss": 5.5702,
-      "step": 41672
-    },
-    {
-      "epoch": 21.732985658409387,
-      "grad_norm": 1.6248235702514648,
-      "learning_rate": 5.863517587939699e-05,
-      "loss": 5.1458,
-      "step": 41673
-    },
-    {
-      "epoch": 21.733507170795306,
-      "grad_norm": 1.634889006614685,
-      "learning_rate": 5.8634170854271365e-05,
-      "loss": 5.1358,
-      "step": 41674
-    },
-    {
-      "epoch": 21.734028683181226,
-      "grad_norm": 1.544450283050537,
-      "learning_rate": 5.863316582914573e-05,
-      "loss": 5.3923,
-      "step": 41675
-    },
-    {
-      "epoch": 21.734550195567145,
-      "grad_norm": 1.513262152671814,
-      "learning_rate": 5.863216080402011e-05,
-      "loss": 5.5082,
-      "step": 41676
-    },
-    {
-      "epoch": 21.735071707953065,
-      "grad_norm": 1.4429874420166016,
-      "learning_rate": 5.863115577889448e-05,
-      "loss": 4.7651,
-      "step": 41677
-    },
-    {
-      "epoch": 21.735593220338984,
-      "grad_norm": 1.4627509117126465,
-      "learning_rate": 5.863015075376884e-05,
-      "loss": 5.2569,
-      "step": 41678
-    },
-    {
-      "epoch": 21.736114732724904,
-      "grad_norm": 1.4974406957626343,
-      "learning_rate": 5.862914572864322e-05,
-      "loss": 5.3774,
-      "step": 41679
-    },
-    {
-      "epoch": 21.736636245110823,
-      "grad_norm": 1.583622932434082,
-      "learning_rate": 5.8628140703517585e-05,
-      "loss": 5.1374,
-      "step": 41680
-    },
-    {
-      "epoch": 21.73715775749674,
-      "grad_norm": 1.51862633228302,
-      "learning_rate": 5.862713567839196e-05,
-      "loss": 5.6368,
-      "step": 41681
-    },
-    {
-      "epoch": 21.73767926988266,
-      "grad_norm": 1.5752471685409546,
-      "learning_rate": 5.8626130653266334e-05,
-      "loss": 5.3291,
-      "step": 41682
-    },
-    {
-      "epoch": 21.738200782268578,
-      "grad_norm": 1.560415506362915,
-      "learning_rate": 5.862512562814071e-05,
-      "loss": 5.5238,
-      "step": 41683
-    },
-    {
-      "epoch": 21.738722294654497,
-      "grad_norm": 1.452353596687317,
-      "learning_rate": 5.8624120603015076e-05,
-      "loss": 5.1288,
-      "step": 41684
-    },
-    {
-      "epoch": 21.739243807040417,
-      "grad_norm": 1.4789149761199951,
-      "learning_rate": 5.8623115577889454e-05,
-      "loss": 5.412,
-      "step": 41685
-    },
-    {
-      "epoch": 21.739765319426336,
-      "grad_norm": 1.5746650695800781,
-      "learning_rate": 5.862211055276382e-05,
-      "loss": 5.1628,
-      "step": 41686
-    },
-    {
-      "epoch": 21.740286831812256,
-      "grad_norm": 1.5910027027130127,
-      "learning_rate": 5.8621105527638196e-05,
-      "loss": 5.2831,
-      "step": 41687
-    },
-    {
-      "epoch": 21.740808344198175,
-      "grad_norm": 1.7910445928573608,
-      "learning_rate": 5.862010050251257e-05,
-      "loss": 5.447,
-      "step": 41688
-    },
-    {
-      "epoch": 21.741329856584095,
-      "grad_norm": 1.575778603553772,
-      "learning_rate": 5.8619095477386945e-05,
-      "loss": 5.6634,
-      "step": 41689
-    },
-    {
-      "epoch": 21.741851368970014,
-      "grad_norm": 1.575588583946228,
-      "learning_rate": 5.861809045226131e-05,
-      "loss": 5.1386,
-      "step": 41690
-    },
-    {
-      "epoch": 21.742372881355934,
-      "grad_norm": 1.5298939943313599,
-      "learning_rate": 5.861708542713567e-05,
-      "loss": 5.3302,
-      "step": 41691
-    },
-    {
-      "epoch": 21.742894393741853,
-      "grad_norm": 1.5256588459014893,
-      "learning_rate": 5.861608040201005e-05,
-      "loss": 5.2342,
-      "step": 41692
-    },
-    {
-      "epoch": 21.74341590612777,
-      "grad_norm": 1.5763205289840698,
-      "learning_rate": 5.861507537688442e-05,
-      "loss": 5.6516,
-      "step": 41693
-    },
-    {
-      "epoch": 21.74393741851369,
-      "grad_norm": 1.7419191598892212,
-      "learning_rate": 5.86140703517588e-05,
-      "loss": 5.0366,
-      "step": 41694
-    },
-    {
-      "epoch": 21.744458930899608,
-      "grad_norm": 1.5829273462295532,
-      "learning_rate": 5.8613065326633164e-05,
-      "loss": 5.3145,
-      "step": 41695
-    },
-    {
-      "epoch": 21.744980443285527,
-      "grad_norm": 1.4798948764801025,
-      "learning_rate": 5.861206030150754e-05,
-      "loss": 5.1799,
-      "step": 41696
-    },
-    {
-      "epoch": 21.745501955671447,
-      "grad_norm": 1.4848198890686035,
-      "learning_rate": 5.861105527638191e-05,
-      "loss": 5.2231,
-      "step": 41697
-    },
-    {
-      "epoch": 21.746023468057366,
-      "grad_norm": 1.4181134700775146,
-      "learning_rate": 5.861005025125629e-05,
-      "loss": 5.4051,
-      "step": 41698
-    },
-    {
-      "epoch": 21.746544980443286,
-      "grad_norm": 1.524461269378662,
-      "learning_rate": 5.8609045226130655e-05,
-      "loss": 5.2468,
-      "step": 41699
-    },
-    {
-      "epoch": 21.747066492829205,
-      "grad_norm": 1.5203258991241455,
-      "learning_rate": 5.860804020100503e-05,
-      "loss": 5.2497,
-      "step": 41700
-    },
-    {
-      "epoch": 21.747588005215125,
-      "grad_norm": 1.4700255393981934,
-      "learning_rate": 5.86070351758794e-05,
-      "loss": 5.3482,
-      "step": 41701
-    },
-    {
-      "epoch": 21.748109517601044,
-      "grad_norm": 1.5455230474472046,
-      "learning_rate": 5.8606030150753775e-05,
-      "loss": 5.3132,
-      "step": 41702
-    },
-    {
-      "epoch": 21.748631029986964,
-      "grad_norm": 1.4175063371658325,
-      "learning_rate": 5.8605025125628146e-05,
-      "loss": 5.5718,
-      "step": 41703
-    },
-    {
-      "epoch": 21.749152542372883,
-      "grad_norm": 1.5279626846313477,
-      "learning_rate": 5.860402010050251e-05,
-      "loss": 5.5703,
-      "step": 41704
-    },
-    {
-      "epoch": 21.7496740547588,
-      "grad_norm": 1.7537732124328613,
-      "learning_rate": 5.860301507537689e-05,
-      "loss": 4.9383,
-      "step": 41705
-    },
-    {
-      "epoch": 21.75019556714472,
-      "grad_norm": 1.39093017578125,
-      "learning_rate": 5.860201005025125e-05,
-      "loss": 5.4712,
-      "step": 41706
-    },
-    {
-      "epoch": 21.750717079530638,
-      "grad_norm": 1.5120582580566406,
-      "learning_rate": 5.860100502512563e-05,
-      "loss": 5.1623,
-      "step": 41707
-    },
-    {
-      "epoch": 21.751238591916557,
-      "grad_norm": 1.5481622219085693,
-      "learning_rate": 5.86e-05,
-      "loss": 5.3943,
-      "step": 41708
-    },
-    {
-      "epoch": 21.751760104302477,
-      "grad_norm": 1.6215606927871704,
-      "learning_rate": 5.859899497487438e-05,
-      "loss": 4.9913,
-      "step": 41709
-    },
-    {
-      "epoch": 21.752281616688396,
-      "grad_norm": 1.5443366765975952,
-      "learning_rate": 5.8597989949748743e-05,
-      "loss": 5.0037,
-      "step": 41710
-    },
-    {
-      "epoch": 21.752803129074316,
-      "grad_norm": 1.4347741603851318,
-      "learning_rate": 5.859698492462312e-05,
-      "loss": 5.7168,
-      "step": 41711
-    },
-    {
-      "epoch": 21.753324641460235,
-      "grad_norm": 1.6700955629348755,
-      "learning_rate": 5.8595979899497486e-05,
-      "loss": 4.9891,
-      "step": 41712
-    },
-    {
-      "epoch": 21.753846153846155,
-      "grad_norm": 1.4156585931777954,
-      "learning_rate": 5.8594974874371863e-05,
-      "loss": 5.5044,
-      "step": 41713
-    },
-    {
-      "epoch": 21.754367666232074,
-      "grad_norm": 1.514775276184082,
-      "learning_rate": 5.8593969849246234e-05,
-      "loss": 5.3508,
-      "step": 41714
-    },
-    {
-      "epoch": 21.754889178617994,
-      "grad_norm": 1.5144778490066528,
-      "learning_rate": 5.859296482412061e-05,
-      "loss": 5.0875,
-      "step": 41715
-    },
-    {
-      "epoch": 21.75541069100391,
-      "grad_norm": 1.5499054193496704,
-      "learning_rate": 5.8591959798994977e-05,
-      "loss": 5.0937,
-      "step": 41716
-    },
-    {
-      "epoch": 21.75593220338983,
-      "grad_norm": 1.5318212509155273,
-      "learning_rate": 5.859095477386935e-05,
-      "loss": 5.3987,
-      "step": 41717
-    },
-    {
-      "epoch": 21.75645371577575,
-      "grad_norm": Infinity,
-      "learning_rate": 5.859095477386935e-05,
-      "loss": 5.3748,
-      "step": 41718
-    },
-    {
-      "epoch": 21.756975228161668,
-      "grad_norm": 1.549851417541504,
-      "learning_rate": 5.8589949748743725e-05,
-      "loss": 5.5775,
-      "step": 41719
-    },
-    {
-      "epoch": 21.757496740547587,
-      "grad_norm": 1.5160008668899536,
-      "learning_rate": 5.858894472361809e-05,
-      "loss": 5.1359,
-      "step": 41720
-    },
-    {
-      "epoch": 21.758018252933507,
-      "grad_norm": 1.4741359949111938,
-      "learning_rate": 5.858793969849247e-05,
-      "loss": 5.502,
-      "step": 41721
-    },
-    {
-      "epoch": 21.758539765319426,
-      "grad_norm": 1.5032916069030762,
-      "learning_rate": 5.858693467336683e-05,
-      "loss": 5.5692,
-      "step": 41722
-    },
-    {
-      "epoch": 21.759061277705346,
-      "grad_norm": 1.6086719036102295,
-      "learning_rate": 5.858592964824121e-05,
-      "loss": 5.2595,
-      "step": 41723
-    },
-    {
-      "epoch": 21.759582790091265,
-      "grad_norm": 1.532251000404358,
-      "learning_rate": 5.858492462311558e-05,
-      "loss": 5.2012,
-      "step": 41724
-    },
-    {
-      "epoch": 21.760104302477185,
-      "grad_norm": 1.4939624071121216,
-      "learning_rate": 5.858391959798996e-05,
-      "loss": 5.3989,
-      "step": 41725
-    },
-    {
-      "epoch": 21.760625814863104,
-      "grad_norm": 1.3975390195846558,
-      "learning_rate": 5.858291457286432e-05,
-      "loss": 5.0917,
-      "step": 41726
-    },
-    {
-      "epoch": 21.761147327249024,
-      "grad_norm": 1.6127427816390991,
-      "learning_rate": 5.85819095477387e-05,
-      "loss": 4.5582,
-      "step": 41727
-    },
-    {
-      "epoch": 21.761668839634943,
-      "grad_norm": 1.4647825956344604,
-      "learning_rate": 5.8580904522613065e-05,
-      "loss": 5.7997,
-      "step": 41728
-    },
-    {
-      "epoch": 21.76219035202086,
-      "grad_norm": 1.5685114860534668,
-      "learning_rate": 5.857989949748744e-05,
-      "loss": 5.3508,
-      "step": 41729
-    },
-    {
-      "epoch": 21.76271186440678,
-      "grad_norm": 1.6191972494125366,
-      "learning_rate": 5.8578894472361814e-05,
-      "loss": 5.1166,
-      "step": 41730
-    },
-    {
-      "epoch": 21.763233376792698,
-      "grad_norm": 1.5419520139694214,
-      "learning_rate": 5.857788944723618e-05,
-      "loss": 5.7086,
-      "step": 41731
-    },
-    {
-      "epoch": 21.763754889178617,
-      "grad_norm": 1.4646605253219604,
-      "learning_rate": 5.8576884422110556e-05,
-      "loss": 5.236,
-      "step": 41732
-    },
-    {
-      "epoch": 21.764276401564537,
-      "grad_norm": 1.4417520761489868,
-      "learning_rate": 5.857587939698492e-05,
-      "loss": 5.6661,
-      "step": 41733
-    },
-    {
-      "epoch": 21.764797913950456,
-      "grad_norm": 1.8630716800689697,
-      "learning_rate": 5.85748743718593e-05,
-      "loss": 4.6237,
-      "step": 41734
-    },
-    {
-      "epoch": 21.765319426336376,
-      "grad_norm": 1.4803338050842285,
-      "learning_rate": 5.857386934673367e-05,
-      "loss": 5.6184,
-      "step": 41735
-    },
-    {
-      "epoch": 21.765840938722295,
-      "grad_norm": 1.5654864311218262,
-      "learning_rate": 5.857286432160805e-05,
-      "loss": 5.0281,
-      "step": 41736
-    },
-    {
-      "epoch": 21.766362451108215,
-      "grad_norm": 1.6804496049880981,
-      "learning_rate": 5.857185929648241e-05,
-      "loss": 4.8331,
-      "step": 41737
-    },
-    {
-      "epoch": 21.766883963494134,
-      "grad_norm": 1.4900615215301514,
-      "learning_rate": 5.857085427135679e-05,
-      "loss": 5.4467,
-      "step": 41738
-    },
-    {
-      "epoch": 21.767405475880054,
-      "grad_norm": 1.652434229850769,
-      "learning_rate": 5.856984924623115e-05,
-      "loss": 4.9778,
-      "step": 41739
-    },
-    {
-      "epoch": 21.76792698826597,
-      "grad_norm": 1.5790510177612305,
-      "learning_rate": 5.856884422110553e-05,
-      "loss": 5.1748,
-      "step": 41740
-    },
-    {
-      "epoch": 21.76844850065189,
-      "grad_norm": 1.6165881156921387,
-      "learning_rate": 5.85678391959799e-05,
-      "loss": 5.1735,
-      "step": 41741
-    },
-    {
-      "epoch": 21.76897001303781,
-      "grad_norm": 1.461808681488037,
-      "learning_rate": 5.856683417085428e-05,
-      "loss": 5.5187,
-      "step": 41742
-    },
-    {
-      "epoch": 21.769491525423728,
-      "grad_norm": 1.5976723432540894,
-      "learning_rate": 5.8565829145728644e-05,
-      "loss": 5.355,
-      "step": 41743
-    },
-    {
-      "epoch": 21.770013037809647,
-      "grad_norm": 1.521122694015503,
-      "learning_rate": 5.8564824120603015e-05,
-      "loss": 5.38,
-      "step": 41744
-    },
-    {
-      "epoch": 21.770534550195567,
-      "grad_norm": 1.4872664213180542,
-      "learning_rate": 5.856381909547739e-05,
-      "loss": 5.4462,
-      "step": 41745
-    },
-    {
-      "epoch": 21.771056062581486,
-      "grad_norm": 1.6282308101654053,
-      "learning_rate": 5.856281407035176e-05,
-      "loss": 4.6676,
-      "step": 41746
-    },
-    {
-      "epoch": 21.771577574967406,
-      "grad_norm": 1.6099826097488403,
-      "learning_rate": 5.8561809045226135e-05,
-      "loss": 5.2836,
-      "step": 41747
-    },
-    {
-      "epoch": 21.772099087353325,
-      "grad_norm": 1.5733752250671387,
-      "learning_rate": 5.85608040201005e-05,
-      "loss": 5.0979,
-      "step": 41748
-    },
-    {
-      "epoch": 21.772620599739245,
-      "grad_norm": 1.5208629369735718,
-      "learning_rate": 5.855979899497488e-05,
-      "loss": 5.2863,
-      "step": 41749
-    },
-    {
-      "epoch": 21.773142112125164,
-      "grad_norm": 1.5577448606491089,
-      "learning_rate": 5.855879396984925e-05,
-      "loss": 5.0803,
-      "step": 41750
-    },
-    {
-      "epoch": 21.773663624511084,
-      "grad_norm": 1.6274619102478027,
-      "learning_rate": 5.8557788944723626e-05,
-      "loss": 5.3288,
-      "step": 41751
-    },
-    {
-      "epoch": 21.774185136897,
-      "grad_norm": 1.6342273950576782,
-      "learning_rate": 5.855678391959799e-05,
-      "loss": 4.7541,
-      "step": 41752
-    },
-    {
-      "epoch": 21.77470664928292,
-      "grad_norm": 1.561427116394043,
-      "learning_rate": 5.855577889447237e-05,
-      "loss": 5.3606,
-      "step": 41753
-    },
-    {
-      "epoch": 21.77522816166884,
-      "grad_norm": 1.557403802871704,
-      "learning_rate": 5.855477386934673e-05,
-      "loss": 5.6472,
-      "step": 41754
-    },
-    {
-      "epoch": 21.775749674054758,
-      "grad_norm": 1.5467591285705566,
-      "learning_rate": 5.855376884422111e-05,
-      "loss": 5.3031,
-      "step": 41755
-    },
-    {
-      "epoch": 21.776271186440677,
-      "grad_norm": 1.5626713037490845,
-      "learning_rate": 5.855276381909548e-05,
-      "loss": 5.5909,
-      "step": 41756
-    },
-    {
-      "epoch": 21.776792698826597,
-      "grad_norm": 1.5318406820297241,
-      "learning_rate": 5.855175879396986e-05,
-      "loss": 5.1521,
-      "step": 41757
-    },
-    {
-      "epoch": 21.777314211212516,
-      "grad_norm": 1.5830122232437134,
-      "learning_rate": 5.8550753768844224e-05,
-      "loss": 4.9899,
-      "step": 41758
-    },
-    {
-      "epoch": 21.777835723598436,
-      "grad_norm": 1.5559252500534058,
-      "learning_rate": 5.854974874371859e-05,
-      "loss": 5.3508,
-      "step": 41759
-    },
-    {
-      "epoch": 21.778357235984355,
-      "grad_norm": 1.5764858722686768,
-      "learning_rate": 5.8548743718592966e-05,
-      "loss": 5.2253,
-      "step": 41760
-    },
-    {
-      "epoch": 21.778878748370275,
-      "grad_norm": 1.6277918815612793,
-      "learning_rate": 5.854773869346734e-05,
-      "loss": 5.2671,
-      "step": 41761
-    },
-    {
-      "epoch": 21.779400260756194,
-      "grad_norm": 1.4668973684310913,
-      "learning_rate": 5.8546733668341715e-05,
-      "loss": 5.3443,
-      "step": 41762
-    },
-    {
-      "epoch": 21.779921773142114,
-      "grad_norm": 1.5415254831314087,
-      "learning_rate": 5.854572864321608e-05,
-      "loss": 5.6908,
-      "step": 41763
-    },
-    {
-      "epoch": 21.78044328552803,
-      "grad_norm": 1.3588570356369019,
-      "learning_rate": 5.854472361809046e-05,
-      "loss": 4.9589,
-      "step": 41764
-    },
-    {
-      "epoch": 21.78096479791395,
-      "grad_norm": 1.6194422245025635,
-      "learning_rate": 5.854371859296483e-05,
-      "loss": 5.2047,
-      "step": 41765
-    },
-    {
-      "epoch": 21.78148631029987,
-      "grad_norm": 1.505468726158142,
-      "learning_rate": 5.8542713567839206e-05,
-      "loss": 5.5038,
-      "step": 41766
-    },
-    {
-      "epoch": 21.782007822685788,
-      "grad_norm": 1.5702511072158813,
-      "learning_rate": 5.854170854271357e-05,
-      "loss": 5.4089,
-      "step": 41767
-    },
-    {
-      "epoch": 21.782529335071708,
-      "grad_norm": 1.5578513145446777,
-      "learning_rate": 5.854070351758795e-05,
-      "loss": 4.8612,
-      "step": 41768
-    },
-    {
-      "epoch": 21.783050847457627,
-      "grad_norm": 1.5202736854553223,
-      "learning_rate": 5.853969849246231e-05,
-      "loss": 5.6449,
-      "step": 41769
-    },
-    {
-      "epoch": 21.783572359843546,
-      "grad_norm": 1.527382493019104,
-      "learning_rate": 5.853869346733669e-05,
-      "loss": 5.5107,
-      "step": 41770
-    },
-    {
-      "epoch": 21.784093872229466,
-      "grad_norm": 1.585921049118042,
-      "learning_rate": 5.853768844221106e-05,
-      "loss": 5.1988,
-      "step": 41771
-    },
-    {
-      "epoch": 21.784615384615385,
-      "grad_norm": 1.4409370422363281,
-      "learning_rate": 5.8536683417085425e-05,
-      "loss": 5.2969,
-      "step": 41772
-    },
-    {
-      "epoch": 21.785136897001305,
-      "grad_norm": 1.6492187976837158,
-      "learning_rate": 5.85356783919598e-05,
-      "loss": 5.5865,
-      "step": 41773
-    },
-    {
-      "epoch": 21.785658409387224,
-      "grad_norm": 1.593255877494812,
-      "learning_rate": 5.853467336683417e-05,
-      "loss": 4.968,
-      "step": 41774
-    },
-    {
-      "epoch": 21.786179921773144,
-      "grad_norm": 1.4729552268981934,
-      "learning_rate": 5.8533668341708545e-05,
-      "loss": 5.3996,
-      "step": 41775
-    },
-    {
-      "epoch": 21.78670143415906,
-      "grad_norm": 1.4222331047058105,
-      "learning_rate": 5.8532663316582916e-05,
-      "loss": 5.4531,
-      "step": 41776
-    },
-    {
-      "epoch": 21.78722294654498,
-      "grad_norm": 1.3882203102111816,
-      "learning_rate": 5.8531658291457294e-05,
-      "loss": 5.3101,
-      "step": 41777
-    },
-    {
-      "epoch": 21.7877444589309,
-      "grad_norm": 1.4726414680480957,
-      "learning_rate": 5.853065326633166e-05,
-      "loss": 5.57,
-      "step": 41778
-    },
-    {
-      "epoch": 21.788265971316818,
-      "grad_norm": 1.4402600526809692,
-      "learning_rate": 5.8529648241206036e-05,
-      "loss": 5.2553,
-      "step": 41779
-    },
-    {
-      "epoch": 21.788787483702738,
-      "grad_norm": 1.489721655845642,
-      "learning_rate": 5.85286432160804e-05,
-      "loss": 5.7216,
-      "step": 41780
-    },
-    {
-      "epoch": 21.789308996088657,
-      "grad_norm": 1.547945499420166,
-      "learning_rate": 5.852763819095478e-05,
-      "loss": 5.2273,
-      "step": 41781
-    },
-    {
-      "epoch": 21.789830508474576,
-      "grad_norm": 1.4512337446212769,
-      "learning_rate": 5.852663316582915e-05,
-      "loss": 5.5932,
-      "step": 41782
-    },
-    {
-      "epoch": 21.790352020860496,
-      "grad_norm": 1.5749917030334473,
-      "learning_rate": 5.852562814070353e-05,
-      "loss": 5.2333,
-      "step": 41783
-    },
-    {
-      "epoch": 21.790873533246415,
-      "grad_norm": 1.603283405303955,
-      "learning_rate": 5.852462311557789e-05,
-      "loss": 4.8951,
-      "step": 41784
-    },
-    {
-      "epoch": 21.791395045632335,
-      "grad_norm": 1.5073927640914917,
-      "learning_rate": 5.852361809045226e-05,
-      "loss": 5.0157,
-      "step": 41785
-    },
-    {
-      "epoch": 21.791916558018254,
-      "grad_norm": 1.5076488256454468,
-      "learning_rate": 5.852261306532664e-05,
-      "loss": 5.5838,
-      "step": 41786
-    },
-    {
-      "epoch": 21.792438070404174,
-      "grad_norm": 1.5056546926498413,
-      "learning_rate": 5.8521608040201004e-05,
-      "loss": 5.1899,
-      "step": 41787
-    },
-    {
-      "epoch": 21.79295958279009,
-      "grad_norm": 1.6219654083251953,
-      "learning_rate": 5.852060301507538e-05,
-      "loss": 4.816,
-      "step": 41788
-    },
-    {
-      "epoch": 21.79348109517601,
-      "grad_norm": 1.5256848335266113,
-      "learning_rate": 5.8519597989949746e-05,
-      "loss": 5.3018,
-      "step": 41789
-    },
-    {
-      "epoch": 21.79400260756193,
-      "grad_norm": 1.5361566543579102,
-      "learning_rate": 5.8518592964824124e-05,
-      "loss": 5.1922,
-      "step": 41790
-    },
-    {
-      "epoch": 21.794524119947848,
-      "grad_norm": 1.4824978113174438,
-      "learning_rate": 5.8517587939698495e-05,
-      "loss": 5.4743,
-      "step": 41791
-    },
-    {
-      "epoch": 21.795045632333768,
-      "grad_norm": 1.5215915441513062,
-      "learning_rate": 5.851658291457287e-05,
-      "loss": 5.2588,
-      "step": 41792
-    },
-    {
-      "epoch": 21.795567144719687,
-      "grad_norm": 1.543323040008545,
-      "learning_rate": 5.851557788944724e-05,
-      "loss": 5.4174,
-      "step": 41793
-    },
-    {
-      "epoch": 21.796088657105607,
-      "grad_norm": 1.539798617362976,
-      "learning_rate": 5.8514572864321615e-05,
-      "loss": 4.834,
-      "step": 41794
-    },
-    {
-      "epoch": 21.796610169491526,
-      "grad_norm": 1.5915992259979248,
-      "learning_rate": 5.851356783919598e-05,
-      "loss": 4.8217,
-      "step": 41795
-    },
-    {
-      "epoch": 21.797131681877445,
-      "grad_norm": 1.5297062397003174,
-      "learning_rate": 5.851256281407036e-05,
-      "loss": 5.4361,
-      "step": 41796
-    },
-    {
-      "epoch": 21.797653194263365,
-      "grad_norm": 1.612471580505371,
-      "learning_rate": 5.851155778894473e-05,
-      "loss": 5.1828,
-      "step": 41797
-    },
-    {
-      "epoch": 21.798174706649284,
-      "grad_norm": 1.528226613998413,
-      "learning_rate": 5.851055276381909e-05,
-      "loss": 5.2382,
-      "step": 41798
-    },
-    {
-      "epoch": 21.7986962190352,
-      "grad_norm": 1.4575625658035278,
-      "learning_rate": 5.850954773869347e-05,
-      "loss": 4.9815,
-      "step": 41799
-    },
-    {
-      "epoch": 21.79921773142112,
-      "grad_norm": 1.635818600654602,
-      "learning_rate": 5.8508542713567835e-05,
-      "loss": 4.4711,
-      "step": 41800
-    },
-    {
-      "epoch": 21.79973924380704,
-      "grad_norm": 1.6312175989151,
-      "learning_rate": 5.850753768844221e-05,
-      "loss": 5.1643,
-      "step": 41801
-    },
-    {
-      "epoch": 21.80026075619296,
-      "grad_norm": 1.6046547889709473,
-      "learning_rate": 5.8506532663316584e-05,
-      "loss": 5.0013,
-      "step": 41802
-    },
-    {
-      "epoch": 21.800782268578878,
-      "grad_norm": 1.6288723945617676,
-      "learning_rate": 5.850552763819096e-05,
-      "loss": 5.5005,
-      "step": 41803
-    },
-    {
-      "epoch": 21.801303780964798,
-      "grad_norm": 1.47405207157135,
-      "learning_rate": 5.8504522613065326e-05,
-      "loss": 5.5191,
-      "step": 41804
-    },
-    {
-      "epoch": 21.801825293350717,
-      "grad_norm": 1.4922550916671753,
-      "learning_rate": 5.8503517587939704e-05,
-      "loss": 5.2235,
-      "step": 41805
-    },
-    {
-      "epoch": 21.802346805736637,
-      "grad_norm": 1.5171229839324951,
-      "learning_rate": 5.8502512562814075e-05,
-      "loss": 5.0464,
-      "step": 41806
-    },
-    {
-      "epoch": 21.802868318122556,
-      "grad_norm": 1.509813666343689,
-      "learning_rate": 5.850150753768845e-05,
-      "loss": 4.7567,
-      "step": 41807
-    },
-    {
-      "epoch": 21.803389830508475,
-      "grad_norm": 1.5275622606277466,
-      "learning_rate": 5.850050251256282e-05,
-      "loss": 5.3084,
-      "step": 41808
-    },
-    {
-      "epoch": 21.803911342894395,
-      "grad_norm": 1.542248010635376,
-      "learning_rate": 5.8499497487437195e-05,
-      "loss": 4.9625,
-      "step": 41809
-    },
-    {
-      "epoch": 21.804432855280314,
-      "grad_norm": 1.425857424736023,
-      "learning_rate": 5.849849246231156e-05,
-      "loss": 5.0354,
-      "step": 41810
-    },
-    {
-      "epoch": 21.804954367666234,
-      "grad_norm": 1.5100351572036743,
-      "learning_rate": 5.849748743718593e-05,
-      "loss": 5.6914,
-      "step": 41811
-    },
-    {
-      "epoch": 21.80547588005215,
-      "grad_norm": 1.5530920028686523,
-      "learning_rate": 5.849648241206031e-05,
-      "loss": 5.3472,
-      "step": 41812
-    },
-    {
-      "epoch": 21.80599739243807,
-      "grad_norm": 1.6062617301940918,
-      "learning_rate": 5.849547738693467e-05,
-      "loss": 5.0467,
-      "step": 41813
-    },
-    {
-      "epoch": 21.80651890482399,
-      "grad_norm": 1.8850963115692139,
-      "learning_rate": 5.849447236180905e-05,
-      "loss": 4.9317,
-      "step": 41814
-    },
-    {
-      "epoch": 21.807040417209908,
-      "grad_norm": 1.5544040203094482,
-      "learning_rate": 5.8493467336683414e-05,
-      "loss": 5.3197,
-      "step": 41815
-    },
-    {
-      "epoch": 21.807561929595828,
-      "grad_norm": 1.486678123474121,
-      "learning_rate": 5.849246231155779e-05,
-      "loss": 5.5676,
-      "step": 41816
-    },
-    {
-      "epoch": 21.808083441981747,
-      "grad_norm": 1.5463550090789795,
-      "learning_rate": 5.849145728643216e-05,
-      "loss": 4.6434,
-      "step": 41817
-    },
-    {
-      "epoch": 21.808604954367667,
-      "grad_norm": 1.6071466207504272,
-      "learning_rate": 5.849045226130654e-05,
-      "loss": 5.2877,
-      "step": 41818
-    },
-    {
-      "epoch": 21.809126466753586,
-      "grad_norm": 1.6218366622924805,
-      "learning_rate": 5.8489447236180905e-05,
-      "loss": 4.7233,
-      "step": 41819
-    },
-    {
-      "epoch": 21.809647979139505,
-      "grad_norm": 1.5549222230911255,
-      "learning_rate": 5.848844221105528e-05,
-      "loss": 5.3702,
-      "step": 41820
-    },
-    {
-      "epoch": 21.810169491525425,
-      "grad_norm": 1.5836418867111206,
-      "learning_rate": 5.848743718592965e-05,
-      "loss": 5.2021,
-      "step": 41821
-    },
-    {
-      "epoch": 21.810691003911344,
-      "grad_norm": 1.544958472251892,
-      "learning_rate": 5.8486432160804025e-05,
-      "loss": 5.4492,
-      "step": 41822
-    },
-    {
-      "epoch": 21.81121251629726,
-      "grad_norm": 1.5403763055801392,
-      "learning_rate": 5.8485427135678396e-05,
-      "loss": 5.3777,
-      "step": 41823
-    },
-    {
-      "epoch": 21.81173402868318,
-      "grad_norm": 1.48047935962677,
-      "learning_rate": 5.848442211055276e-05,
-      "loss": 4.9786,
-      "step": 41824
-    },
-    {
-      "epoch": 21.8122555410691,
-      "grad_norm": 1.5669288635253906,
-      "learning_rate": 5.848341708542714e-05,
-      "loss": 4.8041,
-      "step": 41825
-    },
-    {
-      "epoch": 21.81277705345502,
-      "grad_norm": 1.5604220628738403,
-      "learning_rate": 5.84824120603015e-05,
-      "loss": 5.4154,
-      "step": 41826
-    },
-    {
-      "epoch": 21.813298565840938,
-      "grad_norm": 1.5233486890792847,
-      "learning_rate": 5.848140703517588e-05,
-      "loss": 5.4174,
-      "step": 41827
-    },
-    {
-      "epoch": 21.813820078226858,
-      "grad_norm": 1.5509045124053955,
-      "learning_rate": 5.848040201005025e-05,
-      "loss": 4.9884,
-      "step": 41828
-    },
-    {
-      "epoch": 21.814341590612777,
-      "grad_norm": 1.485085129737854,
-      "learning_rate": 5.847939698492463e-05,
-      "loss": 5.3725,
-      "step": 41829
-    },
-    {
-      "epoch": 21.814863102998697,
-      "grad_norm": 1.5500649213790894,
-      "learning_rate": 5.8478391959798993e-05,
-      "loss": 5.2316,
-      "step": 41830
-    },
-    {
-      "epoch": 21.815384615384616,
-      "grad_norm": 1.6023967266082764,
-      "learning_rate": 5.847738693467337e-05,
-      "loss": 4.9724,
-      "step": 41831
-    },
-    {
-      "epoch": 21.815906127770536,
-      "grad_norm": 1.6323894262313843,
-      "learning_rate": 5.847638190954774e-05,
-      "loss": 5.0212,
-      "step": 41832
-    },
-    {
-      "epoch": 21.816427640156455,
-      "grad_norm": 1.4577659368515015,
-      "learning_rate": 5.847537688442212e-05,
-      "loss": 5.4725,
-      "step": 41833
-    },
-    {
-      "epoch": 21.816949152542374,
-      "grad_norm": 1.5011835098266602,
-      "learning_rate": 5.8474371859296484e-05,
-      "loss": 5.0388,
-      "step": 41834
-    },
-    {
-      "epoch": 21.81747066492829,
-      "grad_norm": 1.6251546144485474,
-      "learning_rate": 5.847336683417086e-05,
-      "loss": 5.2904,
-      "step": 41835
-    },
-    {
-      "epoch": 21.81799217731421,
-      "grad_norm": 1.5147134065628052,
-      "learning_rate": 5.8472361809045227e-05,
-      "loss": 4.9802,
-      "step": 41836
-    },
-    {
-      "epoch": 21.81851368970013,
-      "grad_norm": 1.5521677732467651,
-      "learning_rate": 5.84713567839196e-05,
-      "loss": 4.7293,
-      "step": 41837
-    },
-    {
-      "epoch": 21.81903520208605,
-      "grad_norm": 1.462471604347229,
-      "learning_rate": 5.8470351758793975e-05,
-      "loss": 5.4529,
-      "step": 41838
-    },
-    {
-      "epoch": 21.819556714471968,
-      "grad_norm": 1.6528427600860596,
-      "learning_rate": 5.846934673366834e-05,
-      "loss": 5.0635,
-      "step": 41839
-    },
-    {
-      "epoch": 21.820078226857888,
-      "grad_norm": 1.6585314273834229,
-      "learning_rate": 5.846834170854272e-05,
-      "loss": 4.7191,
-      "step": 41840
-    },
-    {
-      "epoch": 21.820599739243807,
-      "grad_norm": 1.5294302701950073,
-      "learning_rate": 5.846733668341708e-05,
-      "loss": 4.8398,
-      "step": 41841
-    },
-    {
-      "epoch": 21.821121251629727,
-      "grad_norm": 1.616568684577942,
-      "learning_rate": 5.846633165829146e-05,
-      "loss": 5.236,
-      "step": 41842
-    },
-    {
-      "epoch": 21.821642764015646,
-      "grad_norm": 1.4857842922210693,
-      "learning_rate": 5.846532663316583e-05,
-      "loss": 5.4738,
-      "step": 41843
-    },
-    {
-      "epoch": 21.822164276401566,
-      "grad_norm": 1.6349114179611206,
-      "learning_rate": 5.846432160804021e-05,
-      "loss": 4.6649,
-      "step": 41844
-    },
-    {
-      "epoch": 21.822685788787485,
-      "grad_norm": 1.538591980934143,
-      "learning_rate": 5.846331658291457e-05,
-      "loss": 5.3922,
-      "step": 41845
-    },
-    {
-      "epoch": 21.823207301173404,
-      "grad_norm": 1.5729013681411743,
-      "learning_rate": 5.846231155778895e-05,
-      "loss": 5.5455,
-      "step": 41846
-    },
-    {
-      "epoch": 21.82372881355932,
-      "grad_norm": 1.5337170362472534,
-      "learning_rate": 5.8461306532663315e-05,
-      "loss": 5.6652,
-      "step": 41847
-    },
-    {
-      "epoch": 21.82425032594524,
-      "grad_norm": 1.5598108768463135,
-      "learning_rate": 5.846030150753769e-05,
-      "loss": 5.4816,
-      "step": 41848
-    },
-    {
-      "epoch": 21.82477183833116,
-      "grad_norm": 1.6860698461532593,
-      "learning_rate": 5.8459296482412064e-05,
-      "loss": 4.9557,
-      "step": 41849
-    },
-    {
-      "epoch": 21.82529335071708,
-      "grad_norm": 1.5731563568115234,
-      "learning_rate": 5.845829145728644e-05,
-      "loss": 5.12,
-      "step": 41850
-    },
-    {
-      "epoch": 21.825814863103,
-      "grad_norm": 1.5078023672103882,
-      "learning_rate": 5.8457286432160806e-05,
-      "loss": 5.2813,
-      "step": 41851
-    },
-    {
-      "epoch": 21.826336375488918,
-      "grad_norm": 1.6467094421386719,
-      "learning_rate": 5.845628140703518e-05,
-      "loss": 5.095,
-      "step": 41852
-    },
-    {
-      "epoch": 21.826857887874837,
-      "grad_norm": 1.5297105312347412,
-      "learning_rate": 5.8455276381909555e-05,
-      "loss": 5.0524,
-      "step": 41853
-    },
-    {
-      "epoch": 21.827379400260757,
-      "grad_norm": 1.4620307683944702,
-      "learning_rate": 5.845427135678392e-05,
-      "loss": 5.3246,
-      "step": 41854
-    },
-    {
-      "epoch": 21.827900912646676,
-      "grad_norm": 1.5144662857055664,
-      "learning_rate": 5.84532663316583e-05,
-      "loss": 5.2477,
-      "step": 41855
-    },
-    {
-      "epoch": 21.828422425032596,
-      "grad_norm": 1.4711079597473145,
-      "learning_rate": 5.845226130653266e-05,
-      "loss": 5.6519,
-      "step": 41856
-    },
-    {
-      "epoch": 21.828943937418515,
-      "grad_norm": 1.588218331336975,
-      "learning_rate": 5.845125628140704e-05,
-      "loss": 5.0603,
-      "step": 41857
-    },
-    {
-      "epoch": 21.829465449804434,
-      "grad_norm": 1.49828040599823,
-      "learning_rate": 5.845025125628141e-05,
-      "loss": 5.4423,
-      "step": 41858
-    },
-    {
-      "epoch": 21.82998696219035,
-      "grad_norm": 1.5939992666244507,
-      "learning_rate": 5.844924623115579e-05,
-      "loss": 5.0538,
-      "step": 41859
-    },
-    {
-      "epoch": 21.83050847457627,
-      "grad_norm": 1.5635906457901,
-      "learning_rate": 5.844824120603015e-05,
-      "loss": 4.9217,
-      "step": 41860
-    },
-    {
-      "epoch": 21.83102998696219,
-      "grad_norm": 1.6150535345077515,
-      "learning_rate": 5.844723618090453e-05,
-      "loss": 5.4034,
-      "step": 41861
-    },
-    {
-      "epoch": 21.83155149934811,
-      "grad_norm": 1.5982779264450073,
-      "learning_rate": 5.8446231155778894e-05,
-      "loss": 5.0582,
-      "step": 41862
-    },
-    {
-      "epoch": 21.83207301173403,
-      "grad_norm": 1.5209678411483765,
-      "learning_rate": 5.844522613065327e-05,
-      "loss": 5.0012,
-      "step": 41863
-    },
-    {
-      "epoch": 21.832594524119948,
-      "grad_norm": 1.7833702564239502,
-      "learning_rate": 5.844422110552764e-05,
-      "loss": 4.9613,
-      "step": 41864
-    },
-    {
-      "epoch": 21.833116036505867,
-      "grad_norm": 1.5641496181488037,
-      "learning_rate": 5.844321608040201e-05,
-      "loss": 5.5788,
-      "step": 41865
-    },
-    {
-      "epoch": 21.833637548891787,
-      "grad_norm": 1.5508756637573242,
-      "learning_rate": 5.8442211055276385e-05,
-      "loss": 5.5326,
-      "step": 41866
-    },
-    {
-      "epoch": 21.834159061277706,
-      "grad_norm": 1.498386025428772,
-      "learning_rate": 5.844120603015075e-05,
-      "loss": 5.4402,
-      "step": 41867
-    },
-    {
-      "epoch": 21.834680573663626,
-      "grad_norm": 1.5959192514419556,
-      "learning_rate": 5.844020100502513e-05,
-      "loss": 5.0828,
-      "step": 41868
-    },
-    {
-      "epoch": 21.835202086049545,
-      "grad_norm": 1.4415475130081177,
-      "learning_rate": 5.84391959798995e-05,
-      "loss": 5.1377,
-      "step": 41869
-    },
-    {
-      "epoch": 21.835723598435465,
-      "grad_norm": 1.5473214387893677,
-      "learning_rate": 5.8438190954773876e-05,
-      "loss": 5.2678,
-      "step": 41870
-    },
-    {
-      "epoch": 21.83624511082138,
-      "grad_norm": 1.5427993535995483,
-      "learning_rate": 5.843718592964824e-05,
-      "loss": 5.2679,
-      "step": 41871
-    },
-    {
-      "epoch": 21.8367666232073,
-      "grad_norm": 1.4379045963287354,
-      "learning_rate": 5.843618090452262e-05,
-      "loss": 5.2339,
-      "step": 41872
-    },
-    {
-      "epoch": 21.83728813559322,
-      "grad_norm": 1.5980710983276367,
-      "learning_rate": 5.843517587939699e-05,
-      "loss": 5.5609,
-      "step": 41873
-    },
-    {
-      "epoch": 21.83780964797914,
-      "grad_norm": 1.4317413568496704,
-      "learning_rate": 5.843417085427137e-05,
-      "loss": 5.3896,
-      "step": 41874
-    },
-    {
-      "epoch": 21.83833116036506,
-      "grad_norm": 1.6057835817337036,
-      "learning_rate": 5.843316582914573e-05,
-      "loss": 5.4141,
-      "step": 41875
-    },
-    {
-      "epoch": 21.838852672750978,
-      "grad_norm": 1.4632275104522705,
-      "learning_rate": 5.843216080402011e-05,
-      "loss": 5.4878,
-      "step": 41876
-    },
-    {
-      "epoch": 21.839374185136897,
-      "grad_norm": 1.455889105796814,
-      "learning_rate": 5.8431155778894474e-05,
-      "loss": 5.5863,
-      "step": 41877
-    },
-    {
-      "epoch": 21.839895697522817,
-      "grad_norm": 1.5341012477874756,
-      "learning_rate": 5.8430150753768845e-05,
-      "loss": 5.1537,
-      "step": 41878
-    },
-    {
-      "epoch": 21.840417209908736,
-      "grad_norm": 1.4913326501846313,
-      "learning_rate": 5.842914572864322e-05,
-      "loss": 5.1791,
-      "step": 41879
-    },
-    {
-      "epoch": 21.840938722294656,
-      "grad_norm": 1.5357085466384888,
-      "learning_rate": 5.842814070351759e-05,
-      "loss": 5.0442,
-      "step": 41880
-    },
-    {
-      "epoch": 21.841460234680575,
-      "grad_norm": 1.6094989776611328,
-      "learning_rate": 5.8427135678391965e-05,
-      "loss": 5.2962,
-      "step": 41881
-    },
-    {
-      "epoch": 21.841981747066495,
-      "grad_norm": 1.5590009689331055,
-      "learning_rate": 5.842613065326633e-05,
-      "loss": 5.3653,
-      "step": 41882
-    },
-    {
-      "epoch": 21.84250325945241,
-      "grad_norm": 1.5385627746582031,
-      "learning_rate": 5.842512562814071e-05,
-      "loss": 4.9555,
-      "step": 41883
-    },
-    {
-      "epoch": 21.84302477183833,
-      "grad_norm": 1.57291841506958,
-      "learning_rate": 5.842412060301508e-05,
-      "loss": 5.1792,
-      "step": 41884
-    },
-    {
-      "epoch": 21.84354628422425,
-      "grad_norm": 1.5229089260101318,
-      "learning_rate": 5.8423115577889456e-05,
-      "loss": 5.4457,
-      "step": 41885
-    },
-    {
-      "epoch": 21.84406779661017,
-      "grad_norm": 1.3921809196472168,
-      "learning_rate": 5.842211055276382e-05,
-      "loss": 5.1839,
-      "step": 41886
-    },
-    {
-      "epoch": 21.84458930899609,
-      "grad_norm": 1.4522595405578613,
-      "learning_rate": 5.84211055276382e-05,
-      "loss": 5.5781,
-      "step": 41887
-    },
-    {
-      "epoch": 21.845110821382008,
-      "grad_norm": 1.5894471406936646,
-      "learning_rate": 5.842010050251256e-05,
-      "loss": 4.8582,
-      "step": 41888
-    },
-    {
-      "epoch": 21.845632333767927,
-      "grad_norm": 1.4380908012390137,
-      "learning_rate": 5.841909547738694e-05,
-      "loss": 5.6892,
-      "step": 41889
-    },
-    {
-      "epoch": 21.846153846153847,
-      "grad_norm": 1.4884122610092163,
-      "learning_rate": 5.841809045226131e-05,
-      "loss": 5.4317,
-      "step": 41890
-    },
-    {
-      "epoch": 21.846675358539766,
-      "grad_norm": 1.496809959411621,
-      "learning_rate": 5.8417085427135675e-05,
-      "loss": 5.4385,
-      "step": 41891
-    },
-    {
-      "epoch": 21.847196870925686,
-      "grad_norm": 1.6301608085632324,
-      "learning_rate": 5.841608040201005e-05,
-      "loss": 5.1881,
-      "step": 41892
-    },
-    {
-      "epoch": 21.847718383311605,
-      "grad_norm": 1.483633041381836,
-      "learning_rate": 5.841507537688442e-05,
-      "loss": 5.3472,
-      "step": 41893
-    },
-    {
-      "epoch": 21.848239895697525,
-      "grad_norm": 1.599779486656189,
-      "learning_rate": 5.8414070351758795e-05,
-      "loss": 4.5516,
-      "step": 41894
-    },
-    {
-      "epoch": 21.84876140808344,
-      "grad_norm": 1.4741852283477783,
-      "learning_rate": 5.8413065326633166e-05,
-      "loss": 5.4519,
-      "step": 41895
-    },
-    {
-      "epoch": 21.84928292046936,
-      "grad_norm": 1.520350694656372,
-      "learning_rate": 5.8412060301507544e-05,
-      "loss": 5.2032,
-      "step": 41896
-    },
-    {
-      "epoch": 21.84980443285528,
-      "grad_norm": 1.6136078834533691,
-      "learning_rate": 5.841105527638191e-05,
-      "loss": 4.9613,
-      "step": 41897
-    },
-    {
-      "epoch": 21.8503259452412,
-      "grad_norm": 1.4915363788604736,
-      "learning_rate": 5.8410050251256286e-05,
-      "loss": 5.0082,
-      "step": 41898
-    },
-    {
-      "epoch": 21.85084745762712,
-      "grad_norm": 1.452694058418274,
-      "learning_rate": 5.840904522613066e-05,
-      "loss": 5.5215,
-      "step": 41899
-    },
-    {
-      "epoch": 21.851368970013038,
-      "grad_norm": 1.4307173490524292,
-      "learning_rate": 5.8408040201005035e-05,
-      "loss": 5.1033,
-      "step": 41900
-    },
-    {
-      "epoch": 21.851890482398957,
-      "grad_norm": 1.4931042194366455,
-      "learning_rate": 5.84070351758794e-05,
-      "loss": 5.4082,
-      "step": 41901
-    },
-    {
-      "epoch": 21.852411994784877,
-      "grad_norm": 1.4555625915527344,
-      "learning_rate": 5.840603015075378e-05,
-      "loss": 5.0541,
-      "step": 41902
-    },
-    {
-      "epoch": 21.852933507170796,
-      "grad_norm": 1.7087438106536865,
-      "learning_rate": 5.840502512562814e-05,
-      "loss": 4.6929,
-      "step": 41903
-    },
-    {
-      "epoch": 21.853455019556716,
-      "grad_norm": 1.40498685836792,
-      "learning_rate": 5.840402010050251e-05,
-      "loss": 5.364,
-      "step": 41904
-    },
-    {
-      "epoch": 21.853976531942635,
-      "grad_norm": 1.659467339515686,
-      "learning_rate": 5.840301507537689e-05,
-      "loss": 5.1238,
-      "step": 41905
-    },
-    {
-      "epoch": 21.85449804432855,
-      "grad_norm": 1.6358180046081543,
-      "learning_rate": 5.8402010050251254e-05,
-      "loss": 4.7675,
-      "step": 41906
-    },
-    {
-      "epoch": 21.85501955671447,
-      "grad_norm": 1.4843775033950806,
-      "learning_rate": 5.840100502512563e-05,
-      "loss": 5.4466,
-      "step": 41907
-    },
-    {
-      "epoch": 21.85554106910039,
-      "grad_norm": 1.5180821418762207,
-      "learning_rate": 5.8399999999999997e-05,
-      "loss": 5.36,
-      "step": 41908
-    },
-    {
-      "epoch": 21.85606258148631,
-      "grad_norm": 1.4790209531784058,
-      "learning_rate": 5.8398994974874374e-05,
-      "loss": 5.6131,
-      "step": 41909
-    },
-    {
-      "epoch": 21.85658409387223,
-      "grad_norm": 1.6452873945236206,
-      "learning_rate": 5.8397989949748745e-05,
-      "loss": 4.8184,
-      "step": 41910
-    },
-    {
-      "epoch": 21.85710560625815,
-      "grad_norm": 1.5504568815231323,
-      "learning_rate": 5.839698492462312e-05,
-      "loss": 4.8807,
-      "step": 41911
-    },
-    {
-      "epoch": 21.857627118644068,
-      "grad_norm": 1.5455732345581055,
-      "learning_rate": 5.839597989949749e-05,
-      "loss": 5.6304,
-      "step": 41912
-    },
-    {
-      "epoch": 21.858148631029987,
-      "grad_norm": 1.582024335861206,
-      "learning_rate": 5.8394974874371865e-05,
-      "loss": 5.0792,
-      "step": 41913
-    },
-    {
-      "epoch": 21.858670143415907,
-      "grad_norm": 1.5457736253738403,
-      "learning_rate": 5.839396984924623e-05,
-      "loss": 5.1021,
-      "step": 41914
-    },
-    {
-      "epoch": 21.859191655801826,
-      "grad_norm": 1.5175983905792236,
-      "learning_rate": 5.839296482412061e-05,
-      "loss": 5.4904,
-      "step": 41915
-    },
-    {
-      "epoch": 21.859713168187746,
-      "grad_norm": 1.5347245931625366,
-      "learning_rate": 5.839195979899498e-05,
-      "loss": 4.8764,
-      "step": 41916
-    },
-    {
-      "epoch": 21.860234680573665,
-      "grad_norm": 1.5013045072555542,
-      "learning_rate": 5.839095477386934e-05,
-      "loss": 5.6277,
-      "step": 41917
-    },
-    {
-      "epoch": 21.860756192959585,
-      "grad_norm": 1.5162074565887451,
-      "learning_rate": 5.838994974874372e-05,
-      "loss": 5.4623,
-      "step": 41918
-    },
-    {
-      "epoch": 21.8612777053455,
-      "grad_norm": 1.5441786050796509,
-      "learning_rate": 5.838894472361809e-05,
-      "loss": 5.4825,
-      "step": 41919
-    },
-    {
-      "epoch": 21.86179921773142,
-      "grad_norm": 1.509523630142212,
-      "learning_rate": 5.838793969849247e-05,
-      "loss": 5.2309,
-      "step": 41920
-    },
-    {
-      "epoch": 21.86232073011734,
-      "grad_norm": 1.5717930793762207,
-      "learning_rate": 5.8386934673366834e-05,
-      "loss": 5.3303,
-      "step": 41921
-    },
-    {
-      "epoch": 21.86284224250326,
-      "grad_norm": 1.4411300420761108,
-      "learning_rate": 5.838592964824121e-05,
-      "loss": 5.3709,
-      "step": 41922
-    },
-    {
-      "epoch": 21.86336375488918,
-      "grad_norm": 1.5654906034469604,
-      "learning_rate": 5.8384924623115576e-05,
-      "loss": 5.6031,
-      "step": 41923
-    },
-    {
-      "epoch": 21.863885267275098,
-      "grad_norm": 1.5137501955032349,
-      "learning_rate": 5.8383919597989954e-05,
-      "loss": 5.4535,
-      "step": 41924
-    },
-    {
-      "epoch": 21.864406779661017,
-      "grad_norm": 1.5513395071029663,
-      "learning_rate": 5.8382914572864325e-05,
-      "loss": 5.543,
-      "step": 41925
-    },
-    {
-      "epoch": 21.864928292046937,
-      "grad_norm": 1.488561749458313,
-      "learning_rate": 5.83819095477387e-05,
-      "loss": 5.7962,
-      "step": 41926
-    },
-    {
-      "epoch": 21.865449804432856,
-      "grad_norm": 1.5669893026351929,
-      "learning_rate": 5.838090452261307e-05,
-      "loss": 5.074,
-      "step": 41927
-    },
-    {
-      "epoch": 21.865971316818776,
-      "grad_norm": 1.4754858016967773,
-      "learning_rate": 5.8379899497487445e-05,
-      "loss": 5.0765,
-      "step": 41928
-    },
-    {
-      "epoch": 21.866492829204695,
-      "grad_norm": 1.5180875062942505,
-      "learning_rate": 5.837889447236181e-05,
-      "loss": 5.2106,
-      "step": 41929
-    },
-    {
-      "epoch": 21.86701434159061,
-      "grad_norm": 1.5075641870498657,
-      "learning_rate": 5.837788944723619e-05,
-      "loss": 5.3933,
-      "step": 41930
-    },
-    {
-      "epoch": 21.86753585397653,
-      "grad_norm": 1.6527711153030396,
-      "learning_rate": 5.837688442211056e-05,
-      "loss": 5.3742,
-      "step": 41931
-    },
-    {
-      "epoch": 21.86805736636245,
-      "grad_norm": 1.514183521270752,
-      "learning_rate": 5.837587939698492e-05,
-      "loss": 5.3399,
-      "step": 41932
-    },
-    {
-      "epoch": 21.86857887874837,
-      "grad_norm": 1.5060375928878784,
-      "learning_rate": 5.83748743718593e-05,
-      "loss": 5.5315,
-      "step": 41933
-    },
-    {
-      "epoch": 21.86910039113429,
-      "grad_norm": 1.5075675249099731,
-      "learning_rate": 5.8373869346733664e-05,
-      "loss": 5.5403,
-      "step": 41934
-    },
-    {
-      "epoch": 21.86962190352021,
-      "grad_norm": 1.5204569101333618,
-      "learning_rate": 5.837286432160804e-05,
-      "loss": 5.2979,
-      "step": 41935
-    },
-    {
-      "epoch": 21.870143415906128,
-      "grad_norm": 1.596657633781433,
-      "learning_rate": 5.837185929648241e-05,
-      "loss": 5.1186,
-      "step": 41936
-    },
-    {
-      "epoch": 21.870664928292047,
-      "grad_norm": 1.5224225521087646,
-      "learning_rate": 5.837085427135679e-05,
-      "loss": 5.4516,
-      "step": 41937
-    },
-    {
-      "epoch": 21.871186440677967,
-      "grad_norm": 1.5894834995269775,
-      "learning_rate": 5.8369849246231155e-05,
-      "loss": 5.1063,
-      "step": 41938
-    },
-    {
-      "epoch": 21.871707953063886,
-      "grad_norm": 1.538542628288269,
-      "learning_rate": 5.836884422110553e-05,
-      "loss": 5.3148,
-      "step": 41939
-    },
-    {
-      "epoch": 21.872229465449806,
-      "grad_norm": 1.4963231086730957,
-      "learning_rate": 5.8367839195979904e-05,
-      "loss": 5.4986,
-      "step": 41940
-    },
-    {
-      "epoch": 21.872750977835725,
-      "grad_norm": 1.485551357269287,
-      "learning_rate": 5.836683417085428e-05,
-      "loss": 5.5251,
-      "step": 41941
-    },
-    {
-      "epoch": 21.87327249022164,
-      "grad_norm": 1.6024152040481567,
-      "learning_rate": 5.8365829145728646e-05,
-      "loss": 4.663,
-      "step": 41942
-    },
-    {
-      "epoch": 21.87379400260756,
-      "grad_norm": 1.5055773258209229,
-      "learning_rate": 5.8364824120603024e-05,
-      "loss": 5.2487,
-      "step": 41943
-    },
-    {
-      "epoch": 21.87431551499348,
-      "grad_norm": 1.6701358556747437,
-      "learning_rate": 5.836381909547739e-05,
-      "loss": 4.6191,
-      "step": 41944
-    },
-    {
-      "epoch": 21.8748370273794,
-      "grad_norm": 1.6039685010910034,
-      "learning_rate": 5.836281407035176e-05,
-      "loss": 5.546,
-      "step": 41945
-    },
-    {
-      "epoch": 21.87535853976532,
-      "grad_norm": 1.561792254447937,
-      "learning_rate": 5.836180904522614e-05,
-      "loss": 5.427,
-      "step": 41946
-    },
-    {
-      "epoch": 21.87588005215124,
-      "grad_norm": 1.5535634756088257,
-      "learning_rate": 5.83608040201005e-05,
-      "loss": 5.4515,
-      "step": 41947
-    },
-    {
-      "epoch": 21.876401564537158,
-      "grad_norm": 1.5093504190444946,
-      "learning_rate": 5.835979899497488e-05,
-      "loss": 5.5186,
-      "step": 41948
-    },
-    {
-      "epoch": 21.876923076923077,
-      "grad_norm": 1.546369194984436,
-      "learning_rate": 5.8358793969849244e-05,
-      "loss": 5.3234,
-      "step": 41949
-    },
-    {
-      "epoch": 21.877444589308997,
-      "grad_norm": 1.66780686378479,
-      "learning_rate": 5.835778894472362e-05,
-      "loss": 4.4648,
-      "step": 41950
-    },
-    {
-      "epoch": 21.877966101694916,
-      "grad_norm": 1.4737446308135986,
-      "learning_rate": 5.835678391959799e-05,
-      "loss": 5.5348,
-      "step": 41951
-    },
-    {
-      "epoch": 21.878487614080836,
-      "grad_norm": 1.50537109375,
-      "learning_rate": 5.835577889447237e-05,
-      "loss": 5.7937,
-      "step": 41952
-    },
-    {
-      "epoch": 21.879009126466755,
-      "grad_norm": 1.4941002130508423,
-      "learning_rate": 5.8354773869346734e-05,
-      "loss": 5.0339,
-      "step": 41953
-    },
-    {
-      "epoch": 21.87953063885267,
-      "grad_norm": 1.4653023481369019,
-      "learning_rate": 5.835376884422111e-05,
-      "loss": 5.3883,
-      "step": 41954
-    },
-    {
-      "epoch": 21.88005215123859,
-      "grad_norm": 1.513156533241272,
-      "learning_rate": 5.8352763819095477e-05,
-      "loss": 5.4429,
-      "step": 41955
-    },
-    {
-      "epoch": 21.88057366362451,
-      "grad_norm": 1.4504368305206299,
-      "learning_rate": 5.8351758793969854e-05,
-      "loss": 5.4428,
-      "step": 41956
-    },
-    {
-      "epoch": 21.88109517601043,
-      "grad_norm": 1.4611902236938477,
-      "learning_rate": 5.8350753768844225e-05,
-      "loss": 5.6856,
-      "step": 41957
-    },
-    {
-      "epoch": 21.88161668839635,
-      "grad_norm": 1.5420540571212769,
-      "learning_rate": 5.834974874371859e-05,
-      "loss": 5.0313,
-      "step": 41958
-    },
-    {
-      "epoch": 21.88213820078227,
-      "grad_norm": 1.3872066736221313,
-      "learning_rate": 5.834874371859297e-05,
-      "loss": 4.8624,
-      "step": 41959
-    },
-    {
-      "epoch": 21.882659713168188,
-      "grad_norm": 1.564996600151062,
-      "learning_rate": 5.834773869346734e-05,
-      "loss": 4.9693,
-      "step": 41960
-    },
-    {
-      "epoch": 21.883181225554107,
-      "grad_norm": 1.514563798904419,
-      "learning_rate": 5.8346733668341716e-05,
-      "loss": 5.3859,
-      "step": 41961
-    },
-    {
-      "epoch": 21.883702737940027,
-      "grad_norm": 1.595097303390503,
-      "learning_rate": 5.834572864321608e-05,
-      "loss": 5.466,
-      "step": 41962
-    },
-    {
-      "epoch": 21.884224250325946,
-      "grad_norm": 1.6685523986816406,
-      "learning_rate": 5.834472361809046e-05,
-      "loss": 4.9853,
-      "step": 41963
-    },
-    {
-      "epoch": 21.884745762711866,
-      "grad_norm": 1.5359693765640259,
-      "learning_rate": 5.834371859296482e-05,
-      "loss": 5.0661,
-      "step": 41964
-    },
-    {
-      "epoch": 21.885267275097785,
-      "grad_norm": 1.4945323467254639,
-      "learning_rate": 5.83427135678392e-05,
-      "loss": 5.0705,
-      "step": 41965
-    },
-    {
-      "epoch": 21.8857887874837,
-      "grad_norm": 1.497876763343811,
-      "learning_rate": 5.834170854271357e-05,
-      "loss": 5.5896,
-      "step": 41966
-    },
-    {
-      "epoch": 21.88631029986962,
-      "grad_norm": 1.4886724948883057,
-      "learning_rate": 5.834070351758795e-05,
-      "loss": 5.5292,
-      "step": 41967
-    },
-    {
-      "epoch": 21.88683181225554,
-      "grad_norm": 1.4681150913238525,
-      "learning_rate": 5.8339698492462314e-05,
-      "loss": 5.4734,
-      "step": 41968
-    },
-    {
-      "epoch": 21.88735332464146,
-      "grad_norm": 1.5266600847244263,
-      "learning_rate": 5.833869346733669e-05,
-      "loss": 5.1259,
-      "step": 41969
-    },
-    {
-      "epoch": 21.88787483702738,
-      "grad_norm": 1.4793592691421509,
-      "learning_rate": 5.8337688442211056e-05,
-      "loss": 5.1573,
-      "step": 41970
-    },
-    {
-      "epoch": 21.8883963494133,
-      "grad_norm": 1.5315974950790405,
-      "learning_rate": 5.833668341708543e-05,
-      "loss": 5.2563,
-      "step": 41971
-    },
-    {
-      "epoch": 21.888917861799218,
-      "grad_norm": 1.5838834047317505,
-      "learning_rate": 5.8335678391959805e-05,
-      "loss": 4.9376,
-      "step": 41972
-    },
-    {
-      "epoch": 21.889439374185137,
-      "grad_norm": 1.538482666015625,
-      "learning_rate": 5.833467336683417e-05,
-      "loss": 5.3885,
-      "step": 41973
-    },
-    {
-      "epoch": 21.889960886571057,
-      "grad_norm": 1.5373525619506836,
-      "learning_rate": 5.833366834170855e-05,
-      "loss": 5.2494,
-      "step": 41974
-    },
-    {
-      "epoch": 21.890482398956976,
-      "grad_norm": 1.649280309677124,
-      "learning_rate": 5.833266331658291e-05,
-      "loss": 4.7732,
-      "step": 41975
-    },
-    {
-      "epoch": 21.891003911342896,
-      "grad_norm": 1.5088695287704468,
-      "learning_rate": 5.833165829145729e-05,
-      "loss": 5.2788,
-      "step": 41976
-    },
-    {
-      "epoch": 21.891525423728815,
-      "grad_norm": 1.5766627788543701,
-      "learning_rate": 5.833065326633166e-05,
-      "loss": 5.3025,
-      "step": 41977
-    },
-    {
-      "epoch": 21.89204693611473,
-      "grad_norm": 1.5534403324127197,
-      "learning_rate": 5.832964824120604e-05,
-      "loss": 4.9831,
-      "step": 41978
-    },
-    {
-      "epoch": 21.89256844850065,
-      "grad_norm": 1.5899075269699097,
-      "learning_rate": 5.83286432160804e-05,
-      "loss": 4.8409,
-      "step": 41979
-    },
-    {
-      "epoch": 21.89308996088657,
-      "grad_norm": 1.5571039915084839,
-      "learning_rate": 5.832763819095478e-05,
-      "loss": 5.4544,
-      "step": 41980
-    },
-    {
-      "epoch": 21.89361147327249,
-      "grad_norm": 1.5878766775131226,
-      "learning_rate": 5.8326633165829144e-05,
-      "loss": 5.1214,
-      "step": 41981
-    },
-    {
-      "epoch": 21.89413298565841,
-      "grad_norm": 1.527906060218811,
-      "learning_rate": 5.832562814070352e-05,
-      "loss": 5.5225,
-      "step": 41982
-    },
-    {
-      "epoch": 21.89465449804433,
-      "grad_norm": 1.4888696670532227,
-      "learning_rate": 5.832462311557789e-05,
-      "loss": 5.7317,
-      "step": 41983
-    },
-    {
-      "epoch": 21.895176010430248,
-      "grad_norm": 1.4556025266647339,
-      "learning_rate": 5.832361809045226e-05,
-      "loss": 5.2562,
-      "step": 41984
-    },
-    {
-      "epoch": 21.895697522816167,
-      "grad_norm": 1.4777439832687378,
-      "learning_rate": 5.8322613065326635e-05,
-      "loss": 5.6589,
-      "step": 41985
-    },
-    {
-      "epoch": 21.896219035202087,
-      "grad_norm": 1.5699248313903809,
-      "learning_rate": 5.8321608040201006e-05,
-      "loss": 4.9895,
-      "step": 41986
-    },
-    {
-      "epoch": 21.896740547588006,
-      "grad_norm": 1.5966968536376953,
-      "learning_rate": 5.8320603015075384e-05,
-      "loss": 5.2519,
-      "step": 41987
-    },
-    {
-      "epoch": 21.897262059973926,
-      "grad_norm": 1.5805176496505737,
-      "learning_rate": 5.831959798994975e-05,
-      "loss": 4.756,
-      "step": 41988
-    },
-    {
-      "epoch": 21.89778357235984,
-      "grad_norm": 1.4405996799468994,
-      "learning_rate": 5.8318592964824126e-05,
-      "loss": 5.5167,
-      "step": 41989
-    },
-    {
-      "epoch": 21.89830508474576,
-      "grad_norm": 1.5649290084838867,
-      "learning_rate": 5.831758793969849e-05,
-      "loss": 5.4918,
-      "step": 41990
-    },
-    {
-      "epoch": 21.89882659713168,
-      "grad_norm": 1.5600794553756714,
-      "learning_rate": 5.831658291457287e-05,
-      "loss": 5.5562,
-      "step": 41991
-    },
-    {
-      "epoch": 21.8993481095176,
-      "grad_norm": 1.4807024002075195,
-      "learning_rate": 5.831557788944724e-05,
-      "loss": 5.2046,
-      "step": 41992
-    },
-    {
-      "epoch": 21.89986962190352,
-      "grad_norm": 1.6025186777114868,
-      "learning_rate": 5.831457286432162e-05,
-      "loss": 5.3162,
-      "step": 41993
-    },
-    {
-      "epoch": 21.90039113428944,
-      "grad_norm": 1.5062063932418823,
-      "learning_rate": 5.831356783919598e-05,
-      "loss": 5.7683,
-      "step": 41994
-    },
-    {
-      "epoch": 21.90091264667536,
-      "grad_norm": 1.5127612352371216,
-      "learning_rate": 5.831256281407036e-05,
-      "loss": 5.3821,
-      "step": 41995
-    },
-    {
-      "epoch": 21.901434159061278,
-      "grad_norm": 1.4498271942138672,
-      "learning_rate": 5.8311557788944724e-05,
-      "loss": 5.2588,
-      "step": 41996
-    },
-    {
-      "epoch": 21.901955671447197,
-      "grad_norm": 1.6701335906982422,
-      "learning_rate": 5.8310552763819095e-05,
-      "loss": 5.0367,
-      "step": 41997
-    },
-    {
-      "epoch": 21.902477183833117,
-      "grad_norm": 1.5532876253128052,
-      "learning_rate": 5.830954773869347e-05,
-      "loss": 5.3337,
-      "step": 41998
-    },
-    {
-      "epoch": 21.902998696219036,
-      "grad_norm": 1.5312162637710571,
-      "learning_rate": 5.830854271356784e-05,
-      "loss": 5.3679,
-      "step": 41999
-    },
-    {
-      "epoch": 21.903520208604956,
-      "grad_norm": 1.5049115419387817,
-      "learning_rate": 5.8307537688442215e-05,
-      "loss": 5.4632,
-      "step": 42000
-    },
-    {
-      "epoch": 21.904041720990875,
-      "grad_norm": 1.5623483657836914,
-      "learning_rate": 5.830653266331658e-05,
-      "loss": 5.4609,
-      "step": 42001
-    },
-    {
-      "epoch": 21.90456323337679,
-      "grad_norm": 1.5191283226013184,
-      "learning_rate": 5.830552763819096e-05,
-      "loss": 5.6238,
-      "step": 42002
-    },
-    {
-      "epoch": 21.90508474576271,
-      "grad_norm": 1.4985300302505493,
-      "learning_rate": 5.830452261306533e-05,
-      "loss": 5.3175,
-      "step": 42003
-    },
-    {
-      "epoch": 21.90560625814863,
-      "grad_norm": 1.3812922239303589,
-      "learning_rate": 5.8303517587939706e-05,
-      "loss": 5.7412,
-      "step": 42004
-    },
-    {
-      "epoch": 21.90612777053455,
-      "grad_norm": 1.5603723526000977,
-      "learning_rate": 5.830251256281407e-05,
-      "loss": 5.0244,
-      "step": 42005
-    },
-    {
-      "epoch": 21.90664928292047,
-      "grad_norm": 1.5485970973968506,
-      "learning_rate": 5.830150753768845e-05,
-      "loss": 5.1907,
-      "step": 42006
-    },
-    {
-      "epoch": 21.90717079530639,
-      "grad_norm": 1.5613195896148682,
-      "learning_rate": 5.830050251256282e-05,
-      "loss": 5.0657,
-      "step": 42007
-    },
-    {
-      "epoch": 21.907692307692308,
-      "grad_norm": 1.5981497764587402,
-      "learning_rate": 5.8299497487437197e-05,
-      "loss": 5.0002,
-      "step": 42008
-    },
-    {
-      "epoch": 21.908213820078227,
-      "grad_norm": 1.5079877376556396,
-      "learning_rate": 5.829849246231156e-05,
-      "loss": 5.6621,
-      "step": 42009
-    },
-    {
-      "epoch": 21.908735332464147,
-      "grad_norm": 1.6818382740020752,
-      "learning_rate": 5.8297487437185925e-05,
-      "loss": 4.1537,
-      "step": 42010
-    },
-    {
-      "epoch": 21.909256844850066,
-      "grad_norm": 1.4978188276290894,
-      "learning_rate": 5.82964824120603e-05,
-      "loss": 5.5116,
-      "step": 42011
-    },
-    {
-      "epoch": 21.909778357235986,
-      "grad_norm": 1.5603281259536743,
-      "learning_rate": 5.8295477386934674e-05,
-      "loss": 5.4766,
-      "step": 42012
-    },
-    {
-      "epoch": 21.910299869621902,
-      "grad_norm": 1.6450597047805786,
-      "learning_rate": 5.829447236180905e-05,
-      "loss": 4.7945,
-      "step": 42013
-    },
-    {
-      "epoch": 21.91082138200782,
-      "grad_norm": 1.583295464515686,
-      "learning_rate": 5.8293467336683416e-05,
-      "loss": 5.1988,
-      "step": 42014
-    },
-    {
-      "epoch": 21.91134289439374,
-      "grad_norm": 1.3789119720458984,
-      "learning_rate": 5.8292462311557794e-05,
-      "loss": 5.756,
-      "step": 42015
-    },
-    {
-      "epoch": 21.91186440677966,
-      "grad_norm": 1.475592851638794,
-      "learning_rate": 5.829145728643216e-05,
-      "loss": 5.5703,
-      "step": 42016
-    },
-    {
-      "epoch": 21.91238591916558,
-      "grad_norm": 1.6182458400726318,
-      "learning_rate": 5.8290452261306536e-05,
-      "loss": 5.2822,
-      "step": 42017
-    },
-    {
-      "epoch": 21.9129074315515,
-      "grad_norm": 1.365044116973877,
-      "learning_rate": 5.828944723618091e-05,
-      "loss": 4.9048,
-      "step": 42018
-    },
-    {
-      "epoch": 21.91342894393742,
-      "grad_norm": 1.7405434846878052,
-      "learning_rate": 5.8288442211055285e-05,
-      "loss": 4.6075,
-      "step": 42019
-    },
-    {
-      "epoch": 21.913950456323338,
-      "grad_norm": 1.5701165199279785,
-      "learning_rate": 5.828743718592965e-05,
-      "loss": 5.5114,
-      "step": 42020
-    },
-    {
-      "epoch": 21.914471968709258,
-      "grad_norm": 1.6888223886489868,
-      "learning_rate": 5.828643216080403e-05,
-      "loss": 4.8161,
-      "step": 42021
-    },
-    {
-      "epoch": 21.914993481095177,
-      "grad_norm": 1.5084335803985596,
-      "learning_rate": 5.828542713567839e-05,
-      "loss": 5.3072,
-      "step": 42022
-    },
-    {
-      "epoch": 21.915514993481096,
-      "grad_norm": 1.5703974962234497,
-      "learning_rate": 5.828442211055277e-05,
-      "loss": 5.0759,
-      "step": 42023
-    },
-    {
-      "epoch": 21.916036505867016,
-      "grad_norm": 1.4951527118682861,
-      "learning_rate": 5.828341708542714e-05,
-      "loss": 5.2994,
-      "step": 42024
-    },
-    {
-      "epoch": 21.916558018252932,
-      "grad_norm": 1.608962059020996,
-      "learning_rate": 5.8282412060301504e-05,
-      "loss": 5.3088,
-      "step": 42025
-    },
-    {
-      "epoch": 21.91707953063885,
-      "grad_norm": 1.5105503797531128,
-      "learning_rate": 5.828140703517588e-05,
-      "loss": 5.4178,
-      "step": 42026
-    },
-    {
-      "epoch": 21.91760104302477,
-      "grad_norm": 1.4586607217788696,
-      "learning_rate": 5.828040201005025e-05,
-      "loss": 5.5626,
-      "step": 42027
-    },
-    {
-      "epoch": 21.91812255541069,
-      "grad_norm": 1.542940616607666,
-      "learning_rate": 5.827939698492463e-05,
-      "loss": 5.2173,
-      "step": 42028
-    },
-    {
-      "epoch": 21.91864406779661,
-      "grad_norm": 1.457555890083313,
-      "learning_rate": 5.8278391959798995e-05,
-      "loss": 5.4643,
-      "step": 42029
-    },
-    {
-      "epoch": 21.91916558018253,
-      "grad_norm": 1.5745370388031006,
-      "learning_rate": 5.827738693467337e-05,
-      "loss": 5.224,
-      "step": 42030
-    },
-    {
-      "epoch": 21.91968709256845,
-      "grad_norm": 1.5205106735229492,
-      "learning_rate": 5.827638190954774e-05,
-      "loss": 5.596,
-      "step": 42031
-    },
-    {
-      "epoch": 21.920208604954368,
-      "grad_norm": 1.482153058052063,
-      "learning_rate": 5.8275376884422115e-05,
-      "loss": 5.3695,
-      "step": 42032
-    },
-    {
-      "epoch": 21.920730117340288,
-      "grad_norm": 1.5774900913238525,
-      "learning_rate": 5.8274371859296486e-05,
-      "loss": 5.3925,
-      "step": 42033
-    },
-    {
-      "epoch": 21.921251629726207,
-      "grad_norm": 1.6781002283096313,
-      "learning_rate": 5.8273366834170864e-05,
-      "loss": 4.7475,
-      "step": 42034
-    },
-    {
-      "epoch": 21.921773142112126,
-      "grad_norm": 1.580794095993042,
-      "learning_rate": 5.827236180904523e-05,
-      "loss": 4.872,
-      "step": 42035
-    },
-    {
-      "epoch": 21.922294654498046,
-      "grad_norm": 1.5266492366790771,
-      "learning_rate": 5.8271356783919606e-05,
-      "loss": 5.3525,
-      "step": 42036
-    },
-    {
-      "epoch": 21.922816166883962,
-      "grad_norm": 1.5178310871124268,
-      "learning_rate": 5.827035175879397e-05,
-      "loss": 5.4739,
-      "step": 42037
-    },
-    {
-      "epoch": 21.92333767926988,
-      "grad_norm": 1.4860700368881226,
-      "learning_rate": 5.826934673366834e-05,
-      "loss": 5.1177,
-      "step": 42038
-    },
-    {
-      "epoch": 21.9238591916558,
-      "grad_norm": 1.5002357959747314,
-      "learning_rate": 5.826834170854272e-05,
-      "loss": 5.6885,
-      "step": 42039
-    },
-    {
-      "epoch": 21.92438070404172,
-      "grad_norm": 1.6520437002182007,
-      "learning_rate": 5.8267336683417084e-05,
-      "loss": 5.4964,
-      "step": 42040
-    },
-    {
-      "epoch": 21.92490221642764,
-      "grad_norm": 1.4138195514678955,
-      "learning_rate": 5.826633165829146e-05,
-      "loss": 5.1667,
-      "step": 42041
-    },
-    {
-      "epoch": 21.92542372881356,
-      "grad_norm": 1.5630048513412476,
-      "learning_rate": 5.8265326633165826e-05,
-      "loss": 5.5006,
-      "step": 42042
-    },
-    {
-      "epoch": 21.92594524119948,
-      "grad_norm": 1.5247585773468018,
-      "learning_rate": 5.8264321608040204e-05,
-      "loss": 5.3403,
-      "step": 42043
-    },
-    {
-      "epoch": 21.926466753585398,
-      "grad_norm": 1.5392122268676758,
-      "learning_rate": 5.8263316582914575e-05,
-      "loss": 5.3917,
-      "step": 42044
-    },
-    {
-      "epoch": 21.926988265971318,
-      "grad_norm": 1.6335647106170654,
-      "learning_rate": 5.826231155778895e-05,
-      "loss": 5.1803,
-      "step": 42045
-    },
-    {
-      "epoch": 21.927509778357237,
-      "grad_norm": 1.5000687837600708,
-      "learning_rate": 5.826130653266332e-05,
-      "loss": 5.1532,
-      "step": 42046
-    },
-    {
-      "epoch": 21.928031290743156,
-      "grad_norm": 1.5084764957427979,
-      "learning_rate": 5.8260301507537695e-05,
-      "loss": 5.581,
-      "step": 42047
-    },
-    {
-      "epoch": 21.928552803129076,
-      "grad_norm": 1.50753653049469,
-      "learning_rate": 5.8259296482412066e-05,
-      "loss": 4.9312,
-      "step": 42048
-    },
-    {
-      "epoch": 21.929074315514992,
-      "grad_norm": 1.632819414138794,
-      "learning_rate": 5.8258291457286444e-05,
-      "loss": 4.6885,
-      "step": 42049
-    },
-    {
-      "epoch": 21.92959582790091,
-      "grad_norm": 1.549473524093628,
-      "learning_rate": 5.825728643216081e-05,
-      "loss": 4.8301,
-      "step": 42050
-    },
-    {
-      "epoch": 21.93011734028683,
-      "grad_norm": 1.3727803230285645,
-      "learning_rate": 5.825628140703517e-05,
-      "loss": 5.4922,
-      "step": 42051
-    },
-    {
-      "epoch": 21.93063885267275,
-      "grad_norm": 1.435546875,
-      "learning_rate": 5.825527638190955e-05,
-      "loss": 4.8336,
-      "step": 42052
-    },
-    {
-      "epoch": 21.93116036505867,
-      "grad_norm": 1.588252305984497,
-      "learning_rate": 5.825427135678392e-05,
-      "loss": 5.195,
-      "step": 42053
-    },
-    {
-      "epoch": 21.93168187744459,
-      "grad_norm": 1.6049964427947998,
-      "learning_rate": 5.82532663316583e-05,
-      "loss": 5.123,
-      "step": 42054
-    },
-    {
-      "epoch": 21.93220338983051,
-      "grad_norm": 1.5056451559066772,
-      "learning_rate": 5.825226130653266e-05,
-      "loss": 5.436,
-      "step": 42055
-    },
-    {
-      "epoch": 21.932724902216428,
-      "grad_norm": 1.5290497541427612,
-      "learning_rate": 5.825125628140704e-05,
-      "loss": 4.9612,
-      "step": 42056
-    },
-    {
-      "epoch": 21.933246414602348,
-      "grad_norm": 1.5625461339950562,
-      "learning_rate": 5.8250251256281405e-05,
-      "loss": 5.5316,
-      "step": 42057
-    },
-    {
-      "epoch": 21.933767926988267,
-      "grad_norm": 1.5629171133041382,
-      "learning_rate": 5.824924623115578e-05,
-      "loss": 5.2947,
-      "step": 42058
-    },
-    {
-      "epoch": 21.934289439374187,
-      "grad_norm": 1.436341404914856,
-      "learning_rate": 5.8248241206030154e-05,
-      "loss": 5.5308,
-      "step": 42059
-    },
-    {
-      "epoch": 21.934810951760106,
-      "grad_norm": 1.532678484916687,
-      "learning_rate": 5.824723618090453e-05,
-      "loss": 5.4107,
-      "step": 42060
-    },
-    {
-      "epoch": 21.935332464146022,
-      "grad_norm": 1.4010720252990723,
-      "learning_rate": 5.8246231155778896e-05,
-      "loss": 5.826,
-      "step": 42061
-    },
-    {
-      "epoch": 21.93585397653194,
-      "grad_norm": 1.6664141416549683,
-      "learning_rate": 5.8245226130653274e-05,
-      "loss": 4.8218,
-      "step": 42062
-    },
-    {
-      "epoch": 21.93637548891786,
-      "grad_norm": 1.524713397026062,
-      "learning_rate": 5.824422110552764e-05,
-      "loss": 5.5256,
-      "step": 42063
-    },
-    {
-      "epoch": 21.93689700130378,
-      "grad_norm": 1.5958356857299805,
-      "learning_rate": 5.824321608040201e-05,
-      "loss": 5.3986,
-      "step": 42064
-    },
-    {
-      "epoch": 21.9374185136897,
-      "grad_norm": 1.4752930402755737,
-      "learning_rate": 5.824221105527639e-05,
-      "loss": 4.4377,
-      "step": 42065
-    },
-    {
-      "epoch": 21.93794002607562,
-      "grad_norm": 1.6368935108184814,
-      "learning_rate": 5.824120603015075e-05,
-      "loss": 5.0611,
-      "step": 42066
-    },
-    {
-      "epoch": 21.93846153846154,
-      "grad_norm": 1.5998173952102661,
-      "learning_rate": 5.824020100502513e-05,
-      "loss": 5.4908,
-      "step": 42067
-    },
-    {
-      "epoch": 21.938983050847458,
-      "grad_norm": 1.5348546504974365,
-      "learning_rate": 5.8239195979899494e-05,
-      "loss": 5.052,
-      "step": 42068
-    },
-    {
-      "epoch": 21.939504563233378,
-      "grad_norm": 1.6235400438308716,
-      "learning_rate": 5.823819095477387e-05,
-      "loss": 5.3374,
-      "step": 42069
-    },
-    {
-      "epoch": 21.940026075619297,
-      "grad_norm": 1.466383934020996,
-      "learning_rate": 5.823718592964824e-05,
-      "loss": 5.4586,
-      "step": 42070
-    },
-    {
-      "epoch": 21.940547588005217,
-      "grad_norm": 1.57613205909729,
-      "learning_rate": 5.823618090452262e-05,
-      "loss": 5.1459,
-      "step": 42071
-    },
-    {
-      "epoch": 21.941069100391136,
-      "grad_norm": 1.4336999654769897,
-      "learning_rate": 5.8235175879396985e-05,
-      "loss": 5.2142,
-      "step": 42072
-    },
-    {
-      "epoch": 21.941590612777052,
-      "grad_norm": 1.4654468297958374,
-      "learning_rate": 5.823417085427136e-05,
-      "loss": 5.4257,
-      "step": 42073
-    },
-    {
-      "epoch": 21.94211212516297,
-      "grad_norm": 1.5660783052444458,
-      "learning_rate": 5.8233165829145733e-05,
-      "loss": 5.0364,
-      "step": 42074
-    },
-    {
-      "epoch": 21.94263363754889,
-      "grad_norm": 1.640815019607544,
-      "learning_rate": 5.823216080402011e-05,
-      "loss": 5.5139,
-      "step": 42075
-    },
-    {
-      "epoch": 21.94315514993481,
-      "grad_norm": 1.4631438255310059,
-      "learning_rate": 5.8231155778894475e-05,
-      "loss": 5.2812,
-      "step": 42076
-    },
-    {
-      "epoch": 21.94367666232073,
-      "grad_norm": 1.5351910591125488,
-      "learning_rate": 5.823015075376884e-05,
-      "loss": 5.5789,
-      "step": 42077
-    },
-    {
-      "epoch": 21.94419817470665,
-      "grad_norm": 1.5925915241241455,
-      "learning_rate": 5.822914572864322e-05,
-      "loss": 4.9249,
-      "step": 42078
-    },
-    {
-      "epoch": 21.94471968709257,
-      "grad_norm": 1.5787886381149292,
-      "learning_rate": 5.822814070351759e-05,
-      "loss": 4.9654,
-      "step": 42079
-    },
-    {
-      "epoch": 21.945241199478488,
-      "grad_norm": 1.529506802558899,
-      "learning_rate": 5.8227135678391966e-05,
-      "loss": 4.9347,
-      "step": 42080
-    },
-    {
-      "epoch": 21.945762711864408,
-      "grad_norm": 1.4680368900299072,
-      "learning_rate": 5.822613065326633e-05,
-      "loss": 5.5374,
-      "step": 42081
-    },
-    {
-      "epoch": 21.946284224250327,
-      "grad_norm": 1.707362174987793,
-      "learning_rate": 5.822512562814071e-05,
-      "loss": 4.9373,
-      "step": 42082
-    },
-    {
-      "epoch": 21.946805736636247,
-      "grad_norm": 1.4404871463775635,
-      "learning_rate": 5.822412060301507e-05,
-      "loss": 5.635,
-      "step": 42083
-    },
-    {
-      "epoch": 21.947327249022166,
-      "grad_norm": 1.4904073476791382,
-      "learning_rate": 5.822311557788945e-05,
-      "loss": 5.3745,
-      "step": 42084
-    },
-    {
-      "epoch": 21.947848761408082,
-      "grad_norm": 1.474326491355896,
-      "learning_rate": 5.822211055276382e-05,
-      "loss": 5.5313,
-      "step": 42085
-    },
-    {
-      "epoch": 21.948370273794,
-      "grad_norm": 1.4067562818527222,
-      "learning_rate": 5.82211055276382e-05,
-      "loss": 5.4885,
-      "step": 42086
-    },
-    {
-      "epoch": 21.94889178617992,
-      "grad_norm": 1.5355733633041382,
-      "learning_rate": 5.8220100502512564e-05,
-      "loss": 5.1048,
-      "step": 42087
-    },
-    {
-      "epoch": 21.94941329856584,
-      "grad_norm": 1.4649490118026733,
-      "learning_rate": 5.821909547738694e-05,
-      "loss": 5.4616,
-      "step": 42088
-    },
-    {
-      "epoch": 21.94993481095176,
-      "grad_norm": 1.4604824781417847,
-      "learning_rate": 5.8218090452261306e-05,
-      "loss": 4.8086,
-      "step": 42089
-    },
-    {
-      "epoch": 21.95045632333768,
-      "grad_norm": 1.4685108661651611,
-      "learning_rate": 5.821708542713568e-05,
-      "loss": 4.9491,
-      "step": 42090
-    },
-    {
-      "epoch": 21.9509778357236,
-      "grad_norm": 1.5075209140777588,
-      "learning_rate": 5.8216080402010055e-05,
-      "loss": 5.3008,
-      "step": 42091
-    },
-    {
-      "epoch": 21.951499348109518,
-      "grad_norm": 1.4777847528457642,
-      "learning_rate": 5.821507537688442e-05,
-      "loss": 5.4415,
-      "step": 42092
-    },
-    {
-      "epoch": 21.952020860495438,
-      "grad_norm": 1.5505176782608032,
-      "learning_rate": 5.82140703517588e-05,
-      "loss": 5.1612,
-      "step": 42093
-    },
-    {
-      "epoch": 21.952542372881357,
-      "grad_norm": 1.5828884840011597,
-      "learning_rate": 5.821306532663317e-05,
-      "loss": 5.2586,
-      "step": 42094
-    },
-    {
-      "epoch": 21.953063885267277,
-      "grad_norm": 1.4432294368743896,
-      "learning_rate": 5.8212060301507546e-05,
-      "loss": 5.6862,
-      "step": 42095
-    },
-    {
-      "epoch": 21.953585397653193,
-      "grad_norm": 1.4948519468307495,
-      "learning_rate": 5.821105527638191e-05,
-      "loss": 4.6054,
-      "step": 42096
-    },
-    {
-      "epoch": 21.954106910039112,
-      "grad_norm": 1.5245412588119507,
-      "learning_rate": 5.821005025125629e-05,
-      "loss": 5.5406,
-      "step": 42097
-    },
-    {
-      "epoch": 21.95462842242503,
-      "grad_norm": 1.5176786184310913,
-      "learning_rate": 5.820904522613065e-05,
-      "loss": 5.3078,
-      "step": 42098
-    },
-    {
-      "epoch": 21.95514993481095,
-      "grad_norm": 1.5956100225448608,
-      "learning_rate": 5.820804020100503e-05,
-      "loss": 4.9958,
-      "step": 42099
-    },
-    {
-      "epoch": 21.95567144719687,
-      "grad_norm": 1.531893014907837,
-      "learning_rate": 5.82070351758794e-05,
-      "loss": 5.36,
-      "step": 42100
-    },
-    {
-      "epoch": 21.95619295958279,
-      "grad_norm": 1.5048855543136597,
-      "learning_rate": 5.820603015075378e-05,
-      "loss": 5.5331,
-      "step": 42101
-    },
-    {
-      "epoch": 21.95671447196871,
-      "grad_norm": 1.5997244119644165,
-      "learning_rate": 5.820502512562814e-05,
-      "loss": 4.6256,
-      "step": 42102
-    },
-    {
-      "epoch": 21.95723598435463,
-      "grad_norm": 1.5563286542892456,
-      "learning_rate": 5.820402010050251e-05,
-      "loss": 5.346,
-      "step": 42103
-    },
-    {
-      "epoch": 21.957757496740548,
-      "grad_norm": 1.5403648614883423,
-      "learning_rate": 5.8203015075376885e-05,
-      "loss": 5.3513,
-      "step": 42104
-    },
-    {
-      "epoch": 21.958279009126468,
-      "grad_norm": 1.4803470373153687,
-      "learning_rate": 5.8202010050251256e-05,
-      "loss": 5.3674,
-      "step": 42105
-    },
-    {
-      "epoch": 21.958800521512387,
-      "grad_norm": 1.6556289196014404,
-      "learning_rate": 5.8201005025125634e-05,
-      "loss": 4.6369,
-      "step": 42106
-    },
-    {
-      "epoch": 21.959322033898307,
-      "grad_norm": 1.5434647798538208,
-      "learning_rate": 5.82e-05,
-      "loss": 4.8886,
-      "step": 42107
-    },
-    {
-      "epoch": 21.959843546284226,
-      "grad_norm": 1.7631635665893555,
-      "learning_rate": 5.8198994974874376e-05,
-      "loss": 4.9237,
-      "step": 42108
-    },
-    {
-      "epoch": 21.960365058670142,
-      "grad_norm": 1.4906246662139893,
-      "learning_rate": 5.819798994974874e-05,
-      "loss": 5.5423,
-      "step": 42109
-    },
-    {
-      "epoch": 21.96088657105606,
-      "grad_norm": 1.560929536819458,
-      "learning_rate": 5.819698492462312e-05,
-      "loss": 5.2586,
-      "step": 42110
-    },
-    {
-      "epoch": 21.96140808344198,
-      "grad_norm": 1.5039037466049194,
-      "learning_rate": 5.819597989949749e-05,
-      "loss": 5.6529,
-      "step": 42111
-    },
-    {
-      "epoch": 21.9619295958279,
-      "grad_norm": 1.5942096710205078,
-      "learning_rate": 5.819497487437187e-05,
-      "loss": 5.1574,
-      "step": 42112
-    },
-    {
-      "epoch": 21.96245110821382,
-      "grad_norm": 1.391998529434204,
-      "learning_rate": 5.819396984924623e-05,
-      "loss": 4.5104,
-      "step": 42113
-    },
-    {
-      "epoch": 21.96297262059974,
-      "grad_norm": 1.4537588357925415,
-      "learning_rate": 5.819296482412061e-05,
-      "loss": 5.476,
-      "step": 42114
-    },
-    {
-      "epoch": 21.96349413298566,
-      "grad_norm": 1.5042388439178467,
-      "learning_rate": 5.819195979899498e-05,
-      "loss": 5.2225,
-      "step": 42115
-    },
-    {
-      "epoch": 21.96401564537158,
-      "grad_norm": 1.5018008947372437,
-      "learning_rate": 5.819095477386936e-05,
-      "loss": 4.6879,
-      "step": 42116
-    },
-    {
-      "epoch": 21.964537157757498,
-      "grad_norm": 1.5898692607879639,
-      "learning_rate": 5.818994974874372e-05,
-      "loss": 5.5802,
-      "step": 42117
-    },
-    {
-      "epoch": 21.965058670143417,
-      "grad_norm": 1.5287984609603882,
-      "learning_rate": 5.818894472361809e-05,
-      "loss": 5.3689,
-      "step": 42118
-    },
-    {
-      "epoch": 21.965580182529337,
-      "grad_norm": 1.5139858722686768,
-      "learning_rate": 5.8187939698492465e-05,
-      "loss": 5.1533,
-      "step": 42119
-    },
-    {
-      "epoch": 21.966101694915253,
-      "grad_norm": 1.5237902402877808,
-      "learning_rate": 5.8186934673366836e-05,
-      "loss": 5.5651,
-      "step": 42120
-    },
-    {
-      "epoch": 21.966623207301172,
-      "grad_norm": 1.4550294876098633,
-      "learning_rate": 5.8185929648241213e-05,
-      "loss": 5.0271,
-      "step": 42121
-    },
-    {
-      "epoch": 21.96714471968709,
-      "grad_norm": 1.4642835855484009,
-      "learning_rate": 5.818492462311558e-05,
-      "loss": 5.4201,
-      "step": 42122
-    },
-    {
-      "epoch": 21.96766623207301,
-      "grad_norm": 1.4589670896530151,
-      "learning_rate": 5.8183919597989956e-05,
-      "loss": 5.6589,
-      "step": 42123
-    },
-    {
-      "epoch": 21.96818774445893,
-      "grad_norm": 1.5269198417663574,
-      "learning_rate": 5.818291457286432e-05,
-      "loss": 5.346,
-      "step": 42124
-    },
-    {
-      "epoch": 21.96870925684485,
-      "grad_norm": 1.5818287134170532,
-      "learning_rate": 5.81819095477387e-05,
-      "loss": 5.2311,
-      "step": 42125
-    },
-    {
-      "epoch": 21.96923076923077,
-      "grad_norm": 1.3386163711547852,
-      "learning_rate": 5.818090452261307e-05,
-      "loss": 5.6011,
-      "step": 42126
-    },
-    {
-      "epoch": 21.96975228161669,
-      "grad_norm": 1.5470058917999268,
-      "learning_rate": 5.8179899497487447e-05,
-      "loss": 5.1538,
-      "step": 42127
-    },
-    {
-      "epoch": 21.97027379400261,
-      "grad_norm": 1.731455683708191,
-      "learning_rate": 5.817889447236181e-05,
-      "loss": 4.7328,
-      "step": 42128
-    },
-    {
-      "epoch": 21.970795306388528,
-      "grad_norm": 1.4202128648757935,
-      "learning_rate": 5.817788944723619e-05,
-      "loss": 5.0677,
-      "step": 42129
-    },
-    {
-      "epoch": 21.971316818774447,
-      "grad_norm": 1.843341588973999,
-      "learning_rate": 5.817688442211055e-05,
-      "loss": 4.9524,
-      "step": 42130
-    },
-    {
-      "epoch": 21.971838331160367,
-      "grad_norm": 1.4760949611663818,
-      "learning_rate": 5.8175879396984924e-05,
-      "loss": 5.499,
-      "step": 42131
-    },
-    {
-      "epoch": 21.972359843546283,
-      "grad_norm": 1.572011947631836,
-      "learning_rate": 5.81748743718593e-05,
-      "loss": 4.9848,
-      "step": 42132
-    },
-    {
-      "epoch": 21.972881355932202,
-      "grad_norm": 1.5579276084899902,
-      "learning_rate": 5.8173869346733666e-05,
-      "loss": 5.1625,
-      "step": 42133
-    },
-    {
-      "epoch": 21.97340286831812,
-      "grad_norm": 1.6083675622940063,
-      "learning_rate": 5.8172864321608044e-05,
-      "loss": 5.3133,
-      "step": 42134
-    },
-    {
-      "epoch": 21.97392438070404,
-      "grad_norm": 1.4813382625579834,
-      "learning_rate": 5.8171859296482415e-05,
-      "loss": 4.9943,
-      "step": 42135
-    },
-    {
-      "epoch": 21.97444589308996,
-      "grad_norm": 1.5064232349395752,
-      "learning_rate": 5.817085427135679e-05,
-      "loss": 4.9173,
-      "step": 42136
-    },
-    {
-      "epoch": 21.97496740547588,
-      "grad_norm": 1.5259687900543213,
-      "learning_rate": 5.816984924623116e-05,
-      "loss": 5.3921,
-      "step": 42137
-    },
-    {
-      "epoch": 21.9754889178618,
-      "grad_norm": 1.6212314367294312,
-      "learning_rate": 5.8168844221105535e-05,
-      "loss": 4.6806,
-      "step": 42138
-    },
-    {
-      "epoch": 21.97601043024772,
-      "grad_norm": 1.5363695621490479,
-      "learning_rate": 5.81678391959799e-05,
-      "loss": 5.3511,
-      "step": 42139
-    },
-    {
-      "epoch": 21.97653194263364,
-      "grad_norm": 1.5135668516159058,
-      "learning_rate": 5.816683417085428e-05,
-      "loss": 5.4472,
-      "step": 42140
-    },
-    {
-      "epoch": 21.977053455019558,
-      "grad_norm": 1.5170632600784302,
-      "learning_rate": 5.816582914572865e-05,
-      "loss": 5.3477,
-      "step": 42141
-    },
-    {
-      "epoch": 21.977574967405477,
-      "grad_norm": 1.4819536209106445,
-      "learning_rate": 5.8164824120603026e-05,
-      "loss": 4.5059,
-      "step": 42142
-    },
-    {
-      "epoch": 21.978096479791397,
-      "grad_norm": 1.5759046077728271,
-      "learning_rate": 5.816381909547739e-05,
-      "loss": 4.8937,
-      "step": 42143
-    },
-    {
-      "epoch": 21.978617992177313,
-      "grad_norm": 1.5767240524291992,
-      "learning_rate": 5.8162814070351754e-05,
-      "loss": 5.2315,
-      "step": 42144
-    },
-    {
-      "epoch": 21.979139504563232,
-      "grad_norm": 1.4092458486557007,
-      "learning_rate": 5.816180904522613e-05,
-      "loss": 5.6344,
-      "step": 42145
-    },
-    {
-      "epoch": 21.97966101694915,
-      "grad_norm": 1.664224624633789,
-      "learning_rate": 5.81608040201005e-05,
-      "loss": 4.7398,
-      "step": 42146
-    },
-    {
-      "epoch": 21.98018252933507,
-      "grad_norm": 1.536131501197815,
-      "learning_rate": 5.815979899497488e-05,
-      "loss": 5.0595,
-      "step": 42147
-    },
-    {
-      "epoch": 21.98070404172099,
-      "grad_norm": 1.4872055053710938,
-      "learning_rate": 5.8158793969849245e-05,
-      "loss": 5.4025,
-      "step": 42148
-    },
-    {
-      "epoch": 21.98122555410691,
-      "grad_norm": 1.5688204765319824,
-      "learning_rate": 5.815778894472362e-05,
-      "loss": 4.8293,
-      "step": 42149
-    },
-    {
-      "epoch": 21.98174706649283,
-      "grad_norm": 1.4652063846588135,
-      "learning_rate": 5.815678391959799e-05,
-      "loss": 5.4027,
-      "step": 42150
-    },
-    {
-      "epoch": 21.98226857887875,
-      "grad_norm": 1.6321640014648438,
-      "learning_rate": 5.8155778894472365e-05,
-      "loss": 4.9994,
-      "step": 42151
-    },
-    {
-      "epoch": 21.98279009126467,
-      "grad_norm": 1.4840806722640991,
-      "learning_rate": 5.8154773869346736e-05,
-      "loss": 5.3424,
-      "step": 42152
-    },
-    {
-      "epoch": 21.983311603650588,
-      "grad_norm": 1.4152530431747437,
-      "learning_rate": 5.8153768844221114e-05,
-      "loss": 5.5168,
-      "step": 42153
-    },
-    {
-      "epoch": 21.983833116036507,
-      "grad_norm": 1.4785317182540894,
-      "learning_rate": 5.815276381909548e-05,
-      "loss": 5.4903,
-      "step": 42154
-    },
-    {
-      "epoch": 21.984354628422427,
-      "grad_norm": 1.5743228197097778,
-      "learning_rate": 5.8151758793969856e-05,
-      "loss": 5.2291,
-      "step": 42155
-    },
-    {
-      "epoch": 21.984876140808343,
-      "grad_norm": 1.4972797632217407,
-      "learning_rate": 5.815075376884422e-05,
-      "loss": 5.7381,
-      "step": 42156
-    },
-    {
-      "epoch": 21.985397653194262,
-      "grad_norm": 1.5571222305297852,
-      "learning_rate": 5.814974874371859e-05,
-      "loss": 5.0386,
-      "step": 42157
-    },
-    {
-      "epoch": 21.98591916558018,
-      "grad_norm": 1.4645054340362549,
-      "learning_rate": 5.814874371859297e-05,
-      "loss": 5.4694,
-      "step": 42158
-    },
-    {
-      "epoch": 21.9864406779661,
-      "grad_norm": 1.5832356214523315,
-      "learning_rate": 5.8147738693467334e-05,
-      "loss": 5.1461,
-      "step": 42159
-    },
-    {
-      "epoch": 21.98696219035202,
-      "grad_norm": 1.4825469255447388,
-      "learning_rate": 5.814673366834171e-05,
-      "loss": 4.9455,
-      "step": 42160
-    },
-    {
-      "epoch": 21.98748370273794,
-      "grad_norm": 1.5109059810638428,
-      "learning_rate": 5.814572864321608e-05,
-      "loss": 5.174,
-      "step": 42161
-    },
-    {
-      "epoch": 21.98800521512386,
-      "grad_norm": 1.4460844993591309,
-      "learning_rate": 5.814472361809046e-05,
-      "loss": 5.5148,
-      "step": 42162
-    },
-    {
-      "epoch": 21.98852672750978,
-      "grad_norm": 1.9474440813064575,
-      "learning_rate": 5.8143718592964825e-05,
-      "loss": 5.2287,
-      "step": 42163
-    },
-    {
-      "epoch": 21.9890482398957,
-      "grad_norm": 1.4937024116516113,
-      "learning_rate": 5.81427135678392e-05,
-      "loss": 5.2602,
-      "step": 42164
-    },
-    {
-      "epoch": 21.989569752281618,
-      "grad_norm": 1.4989162683486938,
-      "learning_rate": 5.814170854271357e-05,
-      "loss": 5.3758,
-      "step": 42165
-    },
-    {
-      "epoch": 21.990091264667537,
-      "grad_norm": 1.4958611726760864,
-      "learning_rate": 5.8140703517587945e-05,
-      "loss": 5.5151,
-      "step": 42166
-    },
-    {
-      "epoch": 21.990612777053457,
-      "grad_norm": 1.544307827949524,
-      "learning_rate": 5.8139698492462316e-05,
-      "loss": 5.607,
-      "step": 42167
-    },
-    {
-      "epoch": 21.991134289439373,
-      "grad_norm": 1.561173915863037,
-      "learning_rate": 5.8138693467336694e-05,
-      "loss": 5.0536,
-      "step": 42168
-    },
-    {
-      "epoch": 21.991655801825292,
-      "grad_norm": 1.5484232902526855,
-      "learning_rate": 5.813768844221106e-05,
-      "loss": 5.3116,
-      "step": 42169
-    },
-    {
-      "epoch": 21.99217731421121,
-      "grad_norm": 1.6643389463424683,
-      "learning_rate": 5.813668341708542e-05,
-      "loss": 4.7715,
-      "step": 42170
-    },
-    {
-      "epoch": 21.99269882659713,
-      "grad_norm": 1.4491355419158936,
-      "learning_rate": 5.81356783919598e-05,
-      "loss": 5.2348,
-      "step": 42171
-    },
-    {
-      "epoch": 21.99322033898305,
-      "grad_norm": 1.5861461162567139,
-      "learning_rate": 5.813467336683417e-05,
-      "loss": 5.0803,
-      "step": 42172
-    },
-    {
-      "epoch": 21.99374185136897,
-      "grad_norm": 1.5061039924621582,
-      "learning_rate": 5.813366834170855e-05,
-      "loss": 4.7567,
-      "step": 42173
-    },
-    {
-      "epoch": 21.99426336375489,
-      "grad_norm": 1.4513672590255737,
-      "learning_rate": 5.813266331658291e-05,
-      "loss": 5.5294,
-      "step": 42174
-    },
-    {
-      "epoch": 21.99478487614081,
-      "grad_norm": 1.5401742458343506,
-      "learning_rate": 5.813165829145729e-05,
-      "loss": 4.9539,
-      "step": 42175
-    },
-    {
-      "epoch": 21.99530638852673,
-      "grad_norm": 1.579932689666748,
-      "learning_rate": 5.8130653266331655e-05,
-      "loss": 4.8499,
-      "step": 42176
-    },
-    {
-      "epoch": 21.995827900912648,
-      "grad_norm": 1.570725917816162,
-      "learning_rate": 5.812964824120603e-05,
-      "loss": 5.3096,
-      "step": 42177
-    },
-    {
-      "epoch": 21.996349413298567,
-      "grad_norm": 1.5353684425354004,
-      "learning_rate": 5.8128643216080404e-05,
-      "loss": 5.2757,
-      "step": 42178
-    },
-    {
-      "epoch": 21.996870925684483,
-      "grad_norm": 1.4778517484664917,
-      "learning_rate": 5.812763819095478e-05,
-      "loss": 4.4032,
-      "step": 42179
-    },
-    {
-      "epoch": 21.997392438070403,
-      "grad_norm": 1.5564367771148682,
-      "learning_rate": 5.8126633165829146e-05,
-      "loss": 5.4146,
-      "step": 42180
-    },
-    {
-      "epoch": 21.997913950456322,
-      "grad_norm": 1.5175681114196777,
-      "learning_rate": 5.8125628140703524e-05,
-      "loss": 5.2948,
-      "step": 42181
-    },
-    {
-      "epoch": 21.99843546284224,
-      "grad_norm": 1.4250667095184326,
-      "learning_rate": 5.8124623115577895e-05,
-      "loss": 5.5665,
-      "step": 42182
-    },
-    {
-      "epoch": 21.99895697522816,
-      "grad_norm": 1.6407155990600586,
-      "learning_rate": 5.812361809045226e-05,
-      "loss": 5.2181,
-      "step": 42183
-    },
-    {
-      "epoch": 21.99947848761408,
-      "grad_norm": 1.6010212898254395,
-      "learning_rate": 5.812261306532664e-05,
-      "loss": 4.9045,
-      "step": 42184
-    },
-    {
-      "epoch": 22.0,
-      "grad_norm": 1.7898919582366943,
-      "learning_rate": 5.8121608040201e-05,
-      "loss": 5.1907,
-      "step": 42185
-    },
-    {
-      "epoch": 22.00052151238592,
-      "grad_norm": 1.5473474264144897,
-      "learning_rate": 5.812060301507538e-05,
-      "loss": 5.0991,
-      "step": 42186
-    },
-    {
-      "epoch": 22.00104302477184,
-      "grad_norm": 1.5148394107818604,
-      "learning_rate": 5.811959798994975e-05,
-      "loss": 5.2961,
-      "step": 42187
-    },
-    {
-      "epoch": 22.00156453715776,
-      "grad_norm": 1.4671710729599,
-      "learning_rate": 5.811859296482413e-05,
-      "loss": 5.5163,
-      "step": 42188
-    },
-    {
-      "epoch": 22.002086049543678,
-      "grad_norm": 1.3937492370605469,
-      "learning_rate": 5.811758793969849e-05,
-      "loss": 5.8544,
-      "step": 42189
-    },
-    {
-      "epoch": 22.002607561929597,
-      "grad_norm": 1.4880857467651367,
-      "learning_rate": 5.811658291457287e-05,
-      "loss": 5.5973,
-      "step": 42190
-    },
-    {
-      "epoch": 22.003129074315513,
-      "grad_norm": 1.5444488525390625,
-      "learning_rate": 5.8115577889447235e-05,
-      "loss": 4.7218,
-      "step": 42191
-    },
-    {
-      "epoch": 22.003650586701433,
-      "grad_norm": 1.5536354780197144,
-      "learning_rate": 5.811457286432161e-05,
-      "loss": 5.0955,
-      "step": 42192
-    },
-    {
-      "epoch": 22.004172099087352,
-      "grad_norm": 1.5018465518951416,
-      "learning_rate": 5.8113567839195983e-05,
-      "loss": 4.915,
-      "step": 42193
-    },
-    {
-      "epoch": 22.00469361147327,
-      "grad_norm": 1.4450932741165161,
-      "learning_rate": 5.811256281407036e-05,
-      "loss": 5.2797,
-      "step": 42194
-    },
-    {
-      "epoch": 22.00521512385919,
-      "grad_norm": 1.5485340356826782,
-      "learning_rate": 5.8111557788944726e-05,
-      "loss": 5.3059,
-      "step": 42195
-    },
-    {
-      "epoch": 22.00573663624511,
-      "grad_norm": 1.5409448146820068,
-      "learning_rate": 5.811055276381909e-05,
-      "loss": 5.0839,
-      "step": 42196
-    },
-    {
-      "epoch": 22.00625814863103,
-      "grad_norm": 1.5119385719299316,
-      "learning_rate": 5.810954773869347e-05,
-      "loss": 5.6625,
-      "step": 42197
-    },
-    {
-      "epoch": 22.00677966101695,
-      "grad_norm": 1.4794385433197021,
-      "learning_rate": 5.810854271356784e-05,
-      "loss": 5.6239,
-      "step": 42198
-    },
-    {
-      "epoch": 22.00730117340287,
-      "grad_norm": 1.6015331745147705,
-      "learning_rate": 5.8107537688442216e-05,
-      "loss": 5.4303,
-      "step": 42199
-    },
-    {
-      "epoch": 22.00782268578879,
-      "grad_norm": 1.5086472034454346,
-      "learning_rate": 5.810653266331658e-05,
-      "loss": 4.7496,
-      "step": 42200
-    },
-    {
-      "epoch": 22.008344198174708,
-      "grad_norm": 1.500246524810791,
-      "learning_rate": 5.810552763819096e-05,
-      "loss": 5.4023,
-      "step": 42201
-    },
-    {
-      "epoch": 22.008865710560627,
-      "grad_norm": 1.6118499040603638,
-      "learning_rate": 5.810452261306533e-05,
-      "loss": 4.6176,
-      "step": 42202
-    },
-    {
-      "epoch": 22.009387222946543,
-      "grad_norm": 1.5325058698654175,
-      "learning_rate": 5.810351758793971e-05,
-      "loss": 5.085,
-      "step": 42203
-    },
-    {
-      "epoch": 22.009908735332463,
-      "grad_norm": 1.4773192405700684,
-      "learning_rate": 5.810251256281407e-05,
-      "loss": 5.2834,
-      "step": 42204
-    },
-    {
-      "epoch": 22.010430247718382,
-      "grad_norm": 1.5281156301498413,
-      "learning_rate": 5.810150753768845e-05,
-      "loss": 4.7989,
-      "step": 42205
-    },
-    {
-      "epoch": 22.0109517601043,
-      "grad_norm": 1.4840501546859741,
-      "learning_rate": 5.8100502512562814e-05,
-      "loss": 5.3223,
-      "step": 42206
-    },
-    {
-      "epoch": 22.01147327249022,
-      "grad_norm": 1.6088289022445679,
-      "learning_rate": 5.809949748743719e-05,
-      "loss": 5.3307,
-      "step": 42207
-    },
-    {
-      "epoch": 22.01199478487614,
-      "grad_norm": 1.4916146993637085,
-      "learning_rate": 5.809849246231156e-05,
-      "loss": 5.5598,
-      "step": 42208
-    },
-    {
-      "epoch": 22.01251629726206,
-      "grad_norm": 1.659359097480774,
-      "learning_rate": 5.809748743718594e-05,
-      "loss": 5.3009,
-      "step": 42209
-    },
-    {
-      "epoch": 22.01303780964798,
-      "grad_norm": 1.6196494102478027,
-      "learning_rate": 5.8096482412060305e-05,
-      "loss": 5.0247,
-      "step": 42210
-    },
-    {
-      "epoch": 22.0135593220339,
-      "grad_norm": 1.5432448387145996,
-      "learning_rate": 5.809547738693467e-05,
-      "loss": 5.2458,
-      "step": 42211
-    },
-    {
-      "epoch": 22.01408083441982,
-      "grad_norm": 1.5713915824890137,
-      "learning_rate": 5.809447236180905e-05,
-      "loss": 5.0024,
-      "step": 42212
-    },
-    {
-      "epoch": 22.014602346805738,
-      "grad_norm": 1.5571273565292358,
-      "learning_rate": 5.809346733668342e-05,
-      "loss": 5.2121,
-      "step": 42213
-    },
-    {
-      "epoch": 22.015123859191657,
-      "grad_norm": 1.5246895551681519,
-      "learning_rate": 5.8092462311557796e-05,
-      "loss": 5.6447,
-      "step": 42214
-    },
-    {
-      "epoch": 22.015645371577573,
-      "grad_norm": 1.535440444946289,
-      "learning_rate": 5.809145728643216e-05,
-      "loss": 4.9861,
-      "step": 42215
-    },
-    {
-      "epoch": 22.016166883963493,
-      "grad_norm": 1.5461304187774658,
-      "learning_rate": 5.809045226130654e-05,
-      "loss": 5.3143,
-      "step": 42216
-    },
-    {
-      "epoch": 22.016688396349412,
-      "grad_norm": 1.6400445699691772,
-      "learning_rate": 5.80894472361809e-05,
-      "loss": 5.1612,
-      "step": 42217
-    },
-    {
-      "epoch": 22.01720990873533,
-      "grad_norm": 1.642970323562622,
-      "learning_rate": 5.808844221105528e-05,
-      "loss": 4.7964,
-      "step": 42218
-    },
-    {
-      "epoch": 22.01773142112125,
-      "grad_norm": 1.5452964305877686,
-      "learning_rate": 5.808743718592965e-05,
-      "loss": 5.1517,
-      "step": 42219
-    },
-    {
-      "epoch": 22.01825293350717,
-      "grad_norm": 1.5313557386398315,
-      "learning_rate": 5.808643216080403e-05,
-      "loss": 5.5918,
-      "step": 42220
-    },
-    {
-      "epoch": 22.01877444589309,
-      "grad_norm": 1.5130937099456787,
-      "learning_rate": 5.808542713567839e-05,
-      "loss": 5.1508,
-      "step": 42221
-    },
-    {
-      "epoch": 22.01929595827901,
-      "grad_norm": 1.5912034511566162,
-      "learning_rate": 5.808442211055277e-05,
-      "loss": 5.1843,
-      "step": 42222
-    },
-    {
-      "epoch": 22.01981747066493,
-      "grad_norm": 1.4688340425491333,
-      "learning_rate": 5.8083417085427135e-05,
-      "loss": 5.1524,
-      "step": 42223
-    },
-    {
-      "epoch": 22.02033898305085,
-      "grad_norm": 1.627386212348938,
-      "learning_rate": 5.8082412060301506e-05,
-      "loss": 5.142,
-      "step": 42224
-    },
-    {
-      "epoch": 22.020860495436768,
-      "grad_norm": 1.5849157571792603,
-      "learning_rate": 5.8081407035175884e-05,
-      "loss": 4.9926,
-      "step": 42225
-    },
-    {
-      "epoch": 22.021382007822687,
-      "grad_norm": 1.4365702867507935,
-      "learning_rate": 5.808040201005025e-05,
-      "loss": 5.5992,
-      "step": 42226
-    },
-    {
-      "epoch": 22.021903520208603,
-      "grad_norm": 1.5197229385375977,
-      "learning_rate": 5.8079396984924626e-05,
-      "loss": 5.2509,
-      "step": 42227
-    },
-    {
-      "epoch": 22.022425032594523,
-      "grad_norm": 1.437827229499817,
-      "learning_rate": 5.8078391959799e-05,
-      "loss": 5.3122,
-      "step": 42228
-    },
-    {
-      "epoch": 22.022946544980442,
-      "grad_norm": 1.5956943035125732,
-      "learning_rate": 5.8077386934673375e-05,
-      "loss": 5.0064,
-      "step": 42229
-    },
-    {
-      "epoch": 22.02346805736636,
-      "grad_norm": 1.535682201385498,
-      "learning_rate": 5.807638190954774e-05,
-      "loss": 5.527,
-      "step": 42230
-    },
-    {
-      "epoch": 22.02398956975228,
-      "grad_norm": 1.4262722730636597,
-      "learning_rate": 5.807537688442212e-05,
-      "loss": 5.5058,
-      "step": 42231
-    },
-    {
-      "epoch": 22.0245110821382,
-      "grad_norm": 1.5825973749160767,
-      "learning_rate": 5.807437185929648e-05,
-      "loss": 5.3274,
-      "step": 42232
-    },
-    {
-      "epoch": 22.02503259452412,
-      "grad_norm": 1.5673924684524536,
-      "learning_rate": 5.807336683417086e-05,
-      "loss": 5.4373,
-      "step": 42233
-    },
-    {
-      "epoch": 22.02555410691004,
-      "grad_norm": 1.5093003511428833,
-      "learning_rate": 5.807236180904523e-05,
-      "loss": 4.931,
-      "step": 42234
-    },
-    {
-      "epoch": 22.02607561929596,
-      "grad_norm": 1.5039019584655762,
-      "learning_rate": 5.807135678391961e-05,
-      "loss": 5.2337,
-      "step": 42235
-    },
-    {
-      "epoch": 22.02659713168188,
-      "grad_norm": 1.6816078424453735,
-      "learning_rate": 5.807035175879397e-05,
-      "loss": 5.3646,
-      "step": 42236
-    },
-    {
-      "epoch": 22.027118644067798,
-      "grad_norm": 1.4858564138412476,
-      "learning_rate": 5.806934673366834e-05,
-      "loss": 5.2436,
-      "step": 42237
-    },
-    {
-      "epoch": 22.027640156453717,
-      "grad_norm": 1.5412591695785522,
-      "learning_rate": 5.8068341708542715e-05,
-      "loss": 5.0854,
-      "step": 42238
-    },
-    {
-      "epoch": 22.028161668839633,
-      "grad_norm": 1.5451936721801758,
-      "learning_rate": 5.8067336683417086e-05,
-      "loss": 5.2861,
-      "step": 42239
-    },
-    {
-      "epoch": 22.028683181225553,
-      "grad_norm": 1.5802855491638184,
-      "learning_rate": 5.8066331658291463e-05,
-      "loss": 5.1191,
-      "step": 42240
-    },
-    {
-      "epoch": 22.029204693611472,
-      "grad_norm": 1.5186986923217773,
-      "learning_rate": 5.806532663316583e-05,
-      "loss": 5.4269,
-      "step": 42241
-    },
-    {
-      "epoch": 22.02972620599739,
-      "grad_norm": 1.4904438257217407,
-      "learning_rate": 5.8064321608040206e-05,
-      "loss": 5.2085,
-      "step": 42242
-    },
-    {
-      "epoch": 22.03024771838331,
-      "grad_norm": 1.6654895544052124,
-      "learning_rate": 5.806331658291457e-05,
-      "loss": 5.2702,
-      "step": 42243
-    },
-    {
-      "epoch": 22.03076923076923,
-      "grad_norm": 1.5423372983932495,
-      "learning_rate": 5.806231155778895e-05,
-      "loss": 5.1416,
-      "step": 42244
-    },
-    {
-      "epoch": 22.03129074315515,
-      "grad_norm": 1.5702077150344849,
-      "learning_rate": 5.806130653266332e-05,
-      "loss": 5.2851,
-      "step": 42245
-    },
-    {
-      "epoch": 22.03181225554107,
-      "grad_norm": 1.480294942855835,
-      "learning_rate": 5.8060301507537697e-05,
-      "loss": 5.287,
-      "step": 42246
-    },
-    {
-      "epoch": 22.03233376792699,
-      "grad_norm": 1.5242141485214233,
-      "learning_rate": 5.805929648241206e-05,
-      "loss": 5.3468,
-      "step": 42247
-    },
-    {
-      "epoch": 22.03285528031291,
-      "grad_norm": 1.4922040700912476,
-      "learning_rate": 5.805829145728644e-05,
-      "loss": 5.282,
-      "step": 42248
-    },
-    {
-      "epoch": 22.033376792698828,
-      "grad_norm": 1.5157889127731323,
-      "learning_rate": 5.805728643216081e-05,
-      "loss": 5.1929,
-      "step": 42249
-    },
-    {
-      "epoch": 22.033898305084747,
-      "grad_norm": 1.5766901969909668,
-      "learning_rate": 5.8056281407035174e-05,
-      "loss": 5.112,
-      "step": 42250
-    },
-    {
-      "epoch": 22.034419817470663,
-      "grad_norm": 1.5137531757354736,
-      "learning_rate": 5.805527638190955e-05,
-      "loss": 5.2829,
-      "step": 42251
-    },
-    {
-      "epoch": 22.034941329856583,
-      "grad_norm": 1.5672322511672974,
-      "learning_rate": 5.8054271356783916e-05,
-      "loss": 4.947,
-      "step": 42252
-    },
-    {
-      "epoch": 22.035462842242502,
-      "grad_norm": 1.6018668413162231,
-      "learning_rate": 5.8053266331658294e-05,
-      "loss": 5.4892,
-      "step": 42253
-    },
-    {
-      "epoch": 22.03598435462842,
-      "grad_norm": 1.5134884119033813,
-      "learning_rate": 5.8052261306532665e-05,
-      "loss": 5.3792,
-      "step": 42254
-    },
-    {
-      "epoch": 22.03650586701434,
-      "grad_norm": 1.6152496337890625,
-      "learning_rate": 5.805125628140704e-05,
-      "loss": 5.0169,
-      "step": 42255
-    },
-    {
-      "epoch": 22.03702737940026,
-      "grad_norm": 1.490946650505066,
-      "learning_rate": 5.805025125628141e-05,
-      "loss": 5.5264,
-      "step": 42256
-    },
-    {
-      "epoch": 22.03754889178618,
-      "grad_norm": 1.5038586854934692,
-      "learning_rate": 5.8049246231155785e-05,
-      "loss": 5.1481,
-      "step": 42257
-    },
-    {
-      "epoch": 22.0380704041721,
-      "grad_norm": 1.5374336242675781,
-      "learning_rate": 5.804824120603015e-05,
-      "loss": 5.2445,
-      "step": 42258
-    },
-    {
-      "epoch": 22.03859191655802,
-      "grad_norm": 1.5626789331436157,
-      "learning_rate": 5.804723618090453e-05,
-      "loss": 5.3559,
-      "step": 42259
-    },
-    {
-      "epoch": 22.03911342894394,
-      "grad_norm": 1.6444525718688965,
-      "learning_rate": 5.80462311557789e-05,
-      "loss": 5.5492,
-      "step": 42260
-    },
-    {
-      "epoch": 22.039634941329858,
-      "grad_norm": 1.637704610824585,
-      "learning_rate": 5.8045226130653276e-05,
-      "loss": 5.387,
-      "step": 42261
-    },
-    {
-      "epoch": 22.040156453715777,
-      "grad_norm": 1.4715656042099,
-      "learning_rate": 5.804422110552764e-05,
-      "loss": 5.633,
-      "step": 42262
-    },
-    {
-      "epoch": 22.040677966101693,
-      "grad_norm": 1.4672647714614868,
-      "learning_rate": 5.8043216080402004e-05,
-      "loss": 5.3676,
-      "step": 42263
-    },
-    {
-      "epoch": 22.041199478487613,
-      "grad_norm": 1.4865983724594116,
-      "learning_rate": 5.804221105527638e-05,
-      "loss": 4.928,
-      "step": 42264
-    },
-    {
-      "epoch": 22.041720990873532,
-      "grad_norm": 1.5633963346481323,
-      "learning_rate": 5.804120603015075e-05,
-      "loss": 5.281,
-      "step": 42265
-    },
-    {
-      "epoch": 22.042242503259452,
-      "grad_norm": 1.6120134592056274,
-      "learning_rate": 5.804020100502513e-05,
-      "loss": 5.5161,
-      "step": 42266
-    },
-    {
-      "epoch": 22.04276401564537,
-      "grad_norm": 1.4965249300003052,
-      "learning_rate": 5.8039195979899495e-05,
-      "loss": 5.0576,
-      "step": 42267
-    },
-    {
-      "epoch": 22.04328552803129,
-      "grad_norm": 1.5903080701828003,
-      "learning_rate": 5.803819095477387e-05,
-      "loss": 5.3705,
-      "step": 42268
-    },
-    {
-      "epoch": 22.04380704041721,
-      "grad_norm": 1.5209394693374634,
-      "learning_rate": 5.8037185929648244e-05,
-      "loss": 5.2117,
-      "step": 42269
-    },
-    {
-      "epoch": 22.04432855280313,
-      "grad_norm": 1.5541586875915527,
-      "learning_rate": 5.803618090452262e-05,
-      "loss": 5.4584,
-      "step": 42270
-    },
-    {
-      "epoch": 22.04485006518905,
-      "grad_norm": 1.6272467374801636,
-      "learning_rate": 5.8035175879396986e-05,
-      "loss": 4.9546,
-      "step": 42271
-    },
-    {
-      "epoch": 22.04537157757497,
-      "grad_norm": 1.5151827335357666,
-      "learning_rate": 5.8034170854271364e-05,
-      "loss": 4.986,
-      "step": 42272
-    },
-    {
-      "epoch": 22.045893089960888,
-      "grad_norm": 1.4931079149246216,
-      "learning_rate": 5.803316582914573e-05,
-      "loss": 5.475,
-      "step": 42273
-    },
-    {
-      "epoch": 22.046414602346807,
-      "grad_norm": 1.543350100517273,
-      "learning_rate": 5.8032160804020106e-05,
-      "loss": 5.2002,
-      "step": 42274
-    },
-    {
-      "epoch": 22.046936114732723,
-      "grad_norm": 1.544380784034729,
-      "learning_rate": 5.803115577889448e-05,
-      "loss": 5.4718,
-      "step": 42275
-    },
-    {
-      "epoch": 22.047457627118643,
-      "grad_norm": 1.4293986558914185,
-      "learning_rate": 5.803015075376884e-05,
-      "loss": 5.5374,
-      "step": 42276
-    },
-    {
-      "epoch": 22.047979139504562,
-      "grad_norm": 1.3892316818237305,
-      "learning_rate": 5.802914572864322e-05,
-      "loss": 5.156,
-      "step": 42277
-    },
-    {
-      "epoch": 22.048500651890482,
-      "grad_norm": 1.507064700126648,
-      "learning_rate": 5.8028140703517584e-05,
-      "loss": 5.5612,
-      "step": 42278
-    },
-    {
-      "epoch": 22.0490221642764,
-      "grad_norm": 1.5701760053634644,
-      "learning_rate": 5.802713567839196e-05,
-      "loss": 5.5168,
-      "step": 42279
-    },
-    {
-      "epoch": 22.04954367666232,
-      "grad_norm": 1.6472729444503784,
-      "learning_rate": 5.802613065326633e-05,
-      "loss": 4.9352,
-      "step": 42280
-    },
-    {
-      "epoch": 22.05006518904824,
-      "grad_norm": 1.5006089210510254,
-      "learning_rate": 5.802512562814071e-05,
-      "loss": 5.5991,
-      "step": 42281
-    },
-    {
-      "epoch": 22.05058670143416,
-      "grad_norm": 1.6083143949508667,
-      "learning_rate": 5.8024120603015075e-05,
-      "loss": 5.6905,
-      "step": 42282
-    },
-    {
-      "epoch": 22.05110821382008,
-      "grad_norm": 1.5426721572875977,
-      "learning_rate": 5.802311557788945e-05,
-      "loss": 5.4596,
-      "step": 42283
-    },
-    {
-      "epoch": 22.051629726206,
-      "grad_norm": 1.5273357629776,
-      "learning_rate": 5.802211055276382e-05,
-      "loss": 5.4297,
-      "step": 42284
-    },
-    {
-      "epoch": 22.052151238591918,
-      "grad_norm": 1.4488849639892578,
-      "learning_rate": 5.8021105527638195e-05,
-      "loss": 5.8062,
-      "step": 42285
-    },
-    {
-      "epoch": 22.052672750977834,
-      "grad_norm": 1.6520737409591675,
-      "learning_rate": 5.8020100502512566e-05,
-      "loss": 5.0421,
-      "step": 42286
-    },
-    {
-      "epoch": 22.053194263363753,
-      "grad_norm": 1.4769448041915894,
-      "learning_rate": 5.8019095477386944e-05,
-      "loss": 5.0619,
-      "step": 42287
-    },
-    {
-      "epoch": 22.053715775749673,
-      "grad_norm": 1.5856926441192627,
-      "learning_rate": 5.801809045226131e-05,
-      "loss": 5.3888,
-      "step": 42288
-    },
-    {
-      "epoch": 22.054237288135592,
-      "grad_norm": 1.8263765573501587,
-      "learning_rate": 5.801708542713568e-05,
-      "loss": 5.1289,
-      "step": 42289
-    },
-    {
-      "epoch": 22.054758800521512,
-      "grad_norm": 1.525852918624878,
-      "learning_rate": 5.801608040201006e-05,
-      "loss": 5.2605,
-      "step": 42290
-    },
-    {
-      "epoch": 22.05528031290743,
-      "grad_norm": 1.5276423692703247,
-      "learning_rate": 5.801507537688442e-05,
-      "loss": 4.9702,
-      "step": 42291
-    },
-    {
-      "epoch": 22.05580182529335,
-      "grad_norm": 1.549009084701538,
-      "learning_rate": 5.80140703517588e-05,
-      "loss": 5.2373,
-      "step": 42292
-    },
-    {
-      "epoch": 22.05632333767927,
-      "grad_norm": 1.575695514678955,
-      "learning_rate": 5.801306532663316e-05,
-      "loss": 5.1719,
-      "step": 42293
-    },
-    {
-      "epoch": 22.05684485006519,
-      "grad_norm": 1.4412477016448975,
-      "learning_rate": 5.801206030150754e-05,
-      "loss": 5.718,
-      "step": 42294
-    },
-    {
-      "epoch": 22.05736636245111,
-      "grad_norm": 1.4487104415893555,
-      "learning_rate": 5.801105527638191e-05,
-      "loss": 5.2456,
-      "step": 42295
-    },
-    {
-      "epoch": 22.05788787483703,
-      "grad_norm": 1.4973044395446777,
-      "learning_rate": 5.801005025125629e-05,
-      "loss": 4.6319,
-      "step": 42296
-    },
-    {
-      "epoch": 22.058409387222948,
-      "grad_norm": 1.4930790662765503,
-      "learning_rate": 5.8009045226130654e-05,
-      "loss": 5.3642,
-      "step": 42297
-    },
-    {
-      "epoch": 22.058930899608864,
-      "grad_norm": 1.5872652530670166,
-      "learning_rate": 5.800804020100503e-05,
-      "loss": 5.2263,
-      "step": 42298
-    },
-    {
-      "epoch": 22.059452411994783,
-      "grad_norm": 1.6044197082519531,
-      "learning_rate": 5.8007035175879396e-05,
-      "loss": 5.4276,
-      "step": 42299
-    },
-    {
-      "epoch": 22.059973924380703,
-      "grad_norm": 1.6059513092041016,
-      "learning_rate": 5.8006030150753774e-05,
-      "loss": 5.1901,
-      "step": 42300
-    },
-    {
-      "epoch": 22.060495436766622,
-      "grad_norm": 1.5164756774902344,
-      "learning_rate": 5.8005025125628145e-05,
-      "loss": 5.5163,
-      "step": 42301
-    },
-    {
-      "epoch": 22.061016949152542,
-      "grad_norm": 1.5234034061431885,
-      "learning_rate": 5.800402010050252e-05,
-      "loss": 5.4376,
-      "step": 42302
-    },
-    {
-      "epoch": 22.06153846153846,
-      "grad_norm": 1.716391921043396,
-      "learning_rate": 5.800301507537689e-05,
-      "loss": 5.3651,
-      "step": 42303
-    },
-    {
-      "epoch": 22.06205997392438,
-      "grad_norm": 1.5376743078231812,
-      "learning_rate": 5.800201005025125e-05,
-      "loss": 5.5506,
-      "step": 42304
-    },
-    {
-      "epoch": 22.0625814863103,
-      "grad_norm": 1.5157427787780762,
-      "learning_rate": 5.800100502512563e-05,
-      "loss": 5.1742,
-      "step": 42305
-    },
-    {
-      "epoch": 22.06310299869622,
-      "grad_norm": 1.6346800327301025,
-      "learning_rate": 5.8e-05,
-      "loss": 4.662,
-      "step": 42306
-    },
-    {
-      "epoch": 22.06362451108214,
-      "grad_norm": 1.6226364374160767,
-      "learning_rate": 5.799899497487438e-05,
-      "loss": 5.0782,
-      "step": 42307
-    },
-    {
-      "epoch": 22.06414602346806,
-      "grad_norm": 1.505082368850708,
-      "learning_rate": 5.799798994974874e-05,
-      "loss": 5.4002,
-      "step": 42308
-    },
-    {
-      "epoch": 22.064667535853978,
-      "grad_norm": 1.552036166191101,
-      "learning_rate": 5.799698492462312e-05,
-      "loss": 5.013,
-      "step": 42309
-    },
-    {
-      "epoch": 22.065189048239894,
-      "grad_norm": 1.4929243326187134,
-      "learning_rate": 5.7995979899497485e-05,
-      "loss": 4.7207,
-      "step": 42310
-    },
-    {
-      "epoch": 22.065710560625813,
-      "grad_norm": 1.5077801942825317,
-      "learning_rate": 5.799497487437186e-05,
-      "loss": 4.8949,
-      "step": 42311
-    },
-    {
-      "epoch": 22.066232073011733,
-      "grad_norm": 1.5349164009094238,
-      "learning_rate": 5.7993969849246233e-05,
-      "loss": 5.5173,
-      "step": 42312
-    },
-    {
-      "epoch": 22.066753585397652,
-      "grad_norm": 1.5867209434509277,
-      "learning_rate": 5.799296482412061e-05,
-      "loss": 4.9735,
-      "step": 42313
-    },
-    {
-      "epoch": 22.067275097783572,
-      "grad_norm": 1.4702298641204834,
-      "learning_rate": 5.7991959798994976e-05,
-      "loss": 5.6164,
-      "step": 42314
-    },
-    {
-      "epoch": 22.06779661016949,
-      "grad_norm": 1.6182284355163574,
-      "learning_rate": 5.799095477386935e-05,
-      "loss": 4.7971,
-      "step": 42315
-    },
-    {
-      "epoch": 22.06831812255541,
-      "grad_norm": 1.5292675495147705,
-      "learning_rate": 5.7989949748743724e-05,
-      "loss": 5.3853,
-      "step": 42316
-    },
-    {
-      "epoch": 22.06883963494133,
-      "grad_norm": 1.476227879524231,
-      "learning_rate": 5.798894472361809e-05,
-      "loss": 4.7682,
-      "step": 42317
-    },
-    {
-      "epoch": 22.06936114732725,
-      "grad_norm": 1.5176769495010376,
-      "learning_rate": 5.7987939698492467e-05,
-      "loss": 5.385,
-      "step": 42318
-    },
-    {
-      "epoch": 22.06988265971317,
-      "grad_norm": 1.5595006942749023,
-      "learning_rate": 5.798693467336683e-05,
-      "loss": 4.7278,
-      "step": 42319
-    },
-    {
-      "epoch": 22.07040417209909,
-      "grad_norm": 1.5096850395202637,
-      "learning_rate": 5.798592964824121e-05,
-      "loss": 4.5756,
-      "step": 42320
-    },
-    {
-      "epoch": 22.070925684485008,
-      "grad_norm": 1.6761773824691772,
-      "learning_rate": 5.798492462311558e-05,
-      "loss": 5.5695,
-      "step": 42321
-    },
-    {
-      "epoch": 22.071447196870924,
-      "grad_norm": 1.6459242105484009,
-      "learning_rate": 5.798391959798996e-05,
-      "loss": 4.791,
-      "step": 42322
-    },
-    {
-      "epoch": 22.071968709256844,
-      "grad_norm": 1.6878302097320557,
-      "learning_rate": 5.798291457286432e-05,
-      "loss": 5.3555,
-      "step": 42323
-    },
-    {
-      "epoch": 22.072490221642763,
-      "grad_norm": 1.4874836206436157,
-      "learning_rate": 5.79819095477387e-05,
-      "loss": 5.4683,
-      "step": 42324
-    },
-    {
-      "epoch": 22.073011734028682,
-      "grad_norm": 1.672272801399231,
-      "learning_rate": 5.7980904522613064e-05,
-      "loss": 5.577,
-      "step": 42325
-    },
-    {
-      "epoch": 22.073533246414602,
-      "grad_norm": 1.6163147687911987,
-      "learning_rate": 5.797989949748744e-05,
-      "loss": 5.3418,
-      "step": 42326
-    },
-    {
-      "epoch": 22.07405475880052,
-      "grad_norm": 1.4997661113739014,
-      "learning_rate": 5.797889447236181e-05,
-      "loss": 5.092,
-      "step": 42327
-    },
-    {
-      "epoch": 22.07457627118644,
-      "grad_norm": 1.5598405599594116,
-      "learning_rate": 5.797788944723619e-05,
-      "loss": 5.1498,
-      "step": 42328
-    },
-    {
-      "epoch": 22.07509778357236,
-      "grad_norm": 1.5226545333862305,
-      "learning_rate": 5.7976884422110555e-05,
-      "loss": 5.3728,
-      "step": 42329
-    },
-    {
-      "epoch": 22.07561929595828,
-      "grad_norm": 1.5446579456329346,
-      "learning_rate": 5.797587939698492e-05,
-      "loss": 5.27,
-      "step": 42330
-    },
-    {
-      "epoch": 22.0761408083442,
-      "grad_norm": 1.5285747051239014,
-      "learning_rate": 5.79748743718593e-05,
-      "loss": 5.4291,
-      "step": 42331
-    },
-    {
-      "epoch": 22.07666232073012,
-      "grad_norm": 1.5519615411758423,
-      "learning_rate": 5.797386934673367e-05,
-      "loss": 5.5362,
-      "step": 42332
-    },
-    {
-      "epoch": 22.077183833116038,
-      "grad_norm": 1.5340614318847656,
-      "learning_rate": 5.7972864321608046e-05,
-      "loss": 5.1594,
-      "step": 42333
-    },
-    {
-      "epoch": 22.077705345501954,
-      "grad_norm": 1.4453030824661255,
-      "learning_rate": 5.797185929648241e-05,
-      "loss": 5.1403,
-      "step": 42334
-    },
-    {
-      "epoch": 22.078226857887874,
-      "grad_norm": 1.57541024684906,
-      "learning_rate": 5.797085427135679e-05,
-      "loss": 4.6054,
-      "step": 42335
-    },
-    {
-      "epoch": 22.078748370273793,
-      "grad_norm": 1.6429253816604614,
-      "learning_rate": 5.796984924623116e-05,
-      "loss": 5.4739,
-      "step": 42336
-    },
-    {
-      "epoch": 22.079269882659712,
-      "grad_norm": 1.4531382322311401,
-      "learning_rate": 5.796884422110554e-05,
-      "loss": 5.2818,
-      "step": 42337
-    },
-    {
-      "epoch": 22.079791395045632,
-      "grad_norm": 1.5491150617599487,
-      "learning_rate": 5.79678391959799e-05,
-      "loss": 5.4018,
-      "step": 42338
-    },
-    {
-      "epoch": 22.08031290743155,
-      "grad_norm": 1.4766401052474976,
-      "learning_rate": 5.796683417085428e-05,
-      "loss": 5.4083,
-      "step": 42339
-    },
-    {
-      "epoch": 22.08083441981747,
-      "grad_norm": 1.4444125890731812,
-      "learning_rate": 5.796582914572864e-05,
-      "loss": 5.4281,
-      "step": 42340
-    },
-    {
-      "epoch": 22.08135593220339,
-      "grad_norm": 1.5357054471969604,
-      "learning_rate": 5.796482412060302e-05,
-      "loss": 5.4772,
-      "step": 42341
-    },
-    {
-      "epoch": 22.08187744458931,
-      "grad_norm": 1.655051350593567,
-      "learning_rate": 5.796381909547739e-05,
-      "loss": 5.0499,
-      "step": 42342
-    },
-    {
-      "epoch": 22.08239895697523,
-      "grad_norm": 1.5851609706878662,
-      "learning_rate": 5.7962814070351756e-05,
-      "loss": 5.1557,
-      "step": 42343
-    },
-    {
-      "epoch": 22.08292046936115,
-      "grad_norm": 1.523876428604126,
-      "learning_rate": 5.7961809045226134e-05,
-      "loss": 5.3931,
-      "step": 42344
-    },
-    {
-      "epoch": 22.083441981747068,
-      "grad_norm": 1.5812326669692993,
-      "learning_rate": 5.79608040201005e-05,
-      "loss": 4.8676,
-      "step": 42345
-    },
-    {
-      "epoch": 22.083963494132984,
-      "grad_norm": 1.4729214906692505,
-      "learning_rate": 5.7959798994974876e-05,
-      "loss": 5.6163,
-      "step": 42346
-    },
-    {
-      "epoch": 22.084485006518904,
-      "grad_norm": 1.4948015213012695,
-      "learning_rate": 5.795879396984925e-05,
-      "loss": 4.8288,
-      "step": 42347
-    },
-    {
-      "epoch": 22.085006518904823,
-      "grad_norm": 1.4991923570632935,
-      "learning_rate": 5.7957788944723625e-05,
-      "loss": 5.0738,
-      "step": 42348
-    },
-    {
-      "epoch": 22.085528031290742,
-      "grad_norm": 1.5286527872085571,
-      "learning_rate": 5.795678391959799e-05,
-      "loss": 5.3468,
-      "step": 42349
-    },
-    {
-      "epoch": 22.086049543676662,
-      "grad_norm": 1.5830589532852173,
-      "learning_rate": 5.795577889447237e-05,
-      "loss": 4.9178,
-      "step": 42350
-    },
-    {
-      "epoch": 22.08657105606258,
-      "grad_norm": 1.6079641580581665,
-      "learning_rate": 5.795477386934673e-05,
-      "loss": 5.4443,
-      "step": 42351
-    },
-    {
-      "epoch": 22.0870925684485,
-      "grad_norm": 1.6402677297592163,
-      "learning_rate": 5.795376884422111e-05,
-      "loss": 5.287,
-      "step": 42352
-    },
-    {
-      "epoch": 22.08761408083442,
-      "grad_norm": 1.485399603843689,
-      "learning_rate": 5.795276381909548e-05,
-      "loss": 5.5949,
-      "step": 42353
-    },
-    {
-      "epoch": 22.08813559322034,
-      "grad_norm": 1.5588369369506836,
-      "learning_rate": 5.795175879396986e-05,
-      "loss": 4.6217,
-      "step": 42354
-    },
-    {
-      "epoch": 22.08865710560626,
-      "grad_norm": 1.5767165422439575,
-      "learning_rate": 5.795075376884422e-05,
-      "loss": 4.9396,
-      "step": 42355
-    },
-    {
-      "epoch": 22.08917861799218,
-      "grad_norm": 1.529052734375,
-      "learning_rate": 5.7949748743718594e-05,
-      "loss": 5.5324,
-      "step": 42356
-    },
-    {
-      "epoch": 22.089700130378098,
-      "grad_norm": 1.5543524026870728,
-      "learning_rate": 5.794874371859297e-05,
-      "loss": 5.13,
-      "step": 42357
-    },
-    {
-      "epoch": 22.090221642764014,
-      "grad_norm": 1.684630036354065,
-      "learning_rate": 5.7947738693467336e-05,
-      "loss": 4.9293,
-      "step": 42358
-    },
-    {
-      "epoch": 22.090743155149934,
-      "grad_norm": 1.4738095998764038,
-      "learning_rate": 5.7946733668341714e-05,
-      "loss": 5.0208,
-      "step": 42359
-    },
-    {
-      "epoch": 22.091264667535853,
-      "grad_norm": 1.5739991664886475,
-      "learning_rate": 5.794572864321608e-05,
-      "loss": 5.2874,
-      "step": 42360
-    },
-    {
-      "epoch": 22.091786179921773,
-      "grad_norm": 1.4648410081863403,
-      "learning_rate": 5.7944723618090456e-05,
-      "loss": 5.3737,
-      "step": 42361
-    },
-    {
-      "epoch": 22.092307692307692,
-      "grad_norm": 1.4977666139602661,
-      "learning_rate": 5.794371859296483e-05,
-      "loss": 5.5731,
-      "step": 42362
-    },
-    {
-      "epoch": 22.09282920469361,
-      "grad_norm": 1.5609028339385986,
-      "learning_rate": 5.7942713567839204e-05,
-      "loss": 5.3642,
-      "step": 42363
-    },
-    {
-      "epoch": 22.09335071707953,
-      "grad_norm": 1.626520037651062,
-      "learning_rate": 5.794170854271357e-05,
-      "loss": 4.8408,
-      "step": 42364
-    },
-    {
-      "epoch": 22.09387222946545,
-      "grad_norm": 1.6557269096374512,
-      "learning_rate": 5.7940703517587947e-05,
-      "loss": 5.3467,
-      "step": 42365
-    },
-    {
-      "epoch": 22.09439374185137,
-      "grad_norm": 1.462580680847168,
-      "learning_rate": 5.793969849246231e-05,
-      "loss": 5.4928,
-      "step": 42366
-    },
-    {
-      "epoch": 22.09491525423729,
-      "grad_norm": 1.534935474395752,
-      "learning_rate": 5.793869346733669e-05,
-      "loss": 5.4451,
-      "step": 42367
-    },
-    {
-      "epoch": 22.09543676662321,
-      "grad_norm": 1.5024702548980713,
-      "learning_rate": 5.793768844221106e-05,
-      "loss": 5.3948,
-      "step": 42368
-    },
-    {
-      "epoch": 22.09595827900913,
-      "grad_norm": 1.5085101127624512,
-      "learning_rate": 5.7936683417085424e-05,
-      "loss": 5.4895,
-      "step": 42369
-    },
-    {
-      "epoch": 22.096479791395044,
-      "grad_norm": 1.580207109451294,
-      "learning_rate": 5.79356783919598e-05,
-      "loss": 5.1627,
-      "step": 42370
-    },
-    {
-      "epoch": 22.097001303780964,
-      "grad_norm": 1.4655929803848267,
-      "learning_rate": 5.7934673366834166e-05,
-      "loss": 5.6526,
-      "step": 42371
-    },
-    {
-      "epoch": 22.097522816166883,
-      "grad_norm": 1.5333974361419678,
-      "learning_rate": 5.7933668341708544e-05,
-      "loss": 5.0227,
-      "step": 42372
-    },
-    {
-      "epoch": 22.098044328552803,
-      "grad_norm": 1.520455002784729,
-      "learning_rate": 5.7932663316582915e-05,
-      "loss": 5.4336,
-      "step": 42373
-    },
-    {
-      "epoch": 22.098565840938722,
-      "grad_norm": 1.5777113437652588,
-      "learning_rate": 5.793165829145729e-05,
-      "loss": 5.3191,
-      "step": 42374
-    },
-    {
-      "epoch": 22.09908735332464,
-      "grad_norm": 1.527712106704712,
-      "learning_rate": 5.793065326633166e-05,
-      "loss": 5.528,
-      "step": 42375
-    },
-    {
-      "epoch": 22.09960886571056,
-      "grad_norm": 1.4661518335342407,
-      "learning_rate": 5.7929648241206035e-05,
-      "loss": 5.3659,
-      "step": 42376
-    },
-    {
-      "epoch": 22.10013037809648,
-      "grad_norm": 1.5879225730895996,
-      "learning_rate": 5.7928643216080406e-05,
-      "loss": 5.1932,
-      "step": 42377
-    },
-    {
-      "epoch": 22.1006518904824,
-      "grad_norm": 1.7491592168807983,
-      "learning_rate": 5.7927638190954784e-05,
-      "loss": 4.8636,
-      "step": 42378
-    },
-    {
-      "epoch": 22.10117340286832,
-      "grad_norm": 1.456304907798767,
-      "learning_rate": 5.792663316582915e-05,
-      "loss": 5.1481,
-      "step": 42379
-    },
-    {
-      "epoch": 22.10169491525424,
-      "grad_norm": 1.5472043752670288,
-      "learning_rate": 5.7925628140703526e-05,
-      "loss": 5.1285,
-      "step": 42380
-    },
-    {
-      "epoch": 22.102216427640155,
-      "grad_norm": 1.5230553150177002,
-      "learning_rate": 5.792462311557789e-05,
-      "loss": 5.2753,
-      "step": 42381
-    },
-    {
-      "epoch": 22.102737940026074,
-      "grad_norm": 1.587855339050293,
-      "learning_rate": 5.792361809045227e-05,
-      "loss": 5.0918,
-      "step": 42382
-    },
-    {
-      "epoch": 22.103259452411994,
-      "grad_norm": 1.5250290632247925,
-      "learning_rate": 5.792261306532664e-05,
-      "loss": 5.3373,
-      "step": 42383
-    },
-    {
-      "epoch": 22.103780964797913,
-      "grad_norm": 1.7338262796401978,
-      "learning_rate": 5.7921608040201e-05,
-      "loss": 4.9569,
-      "step": 42384
-    },
-    {
-      "epoch": 22.104302477183833,
-      "grad_norm": 1.4724820852279663,
-      "learning_rate": 5.792060301507538e-05,
-      "loss": 5.3365,
-      "step": 42385
-    },
-    {
-      "epoch": 22.104823989569752,
-      "grad_norm": 1.5225813388824463,
-      "learning_rate": 5.7919597989949745e-05,
-      "loss": 5.0477,
-      "step": 42386
-    },
-    {
-      "epoch": 22.10534550195567,
-      "grad_norm": 1.4993162155151367,
-      "learning_rate": 5.791859296482412e-05,
-      "loss": 5.3746,
-      "step": 42387
-    },
-    {
-      "epoch": 22.10586701434159,
-      "grad_norm": 1.568951964378357,
-      "learning_rate": 5.7917587939698494e-05,
-      "loss": 5.5184,
-      "step": 42388
-    },
-    {
-      "epoch": 22.10638852672751,
-      "grad_norm": 1.534873366355896,
-      "learning_rate": 5.791658291457287e-05,
-      "loss": 5.3923,
-      "step": 42389
-    },
-    {
-      "epoch": 22.10691003911343,
-      "grad_norm": 1.561029076576233,
-      "learning_rate": 5.7915577889447236e-05,
-      "loss": 4.8038,
-      "step": 42390
-    },
-    {
-      "epoch": 22.10743155149935,
-      "grad_norm": 1.5880491733551025,
-      "learning_rate": 5.7914572864321614e-05,
-      "loss": 5.0686,
-      "step": 42391
-    },
-    {
-      "epoch": 22.10795306388527,
-      "grad_norm": 1.5562140941619873,
-      "learning_rate": 5.791356783919598e-05,
-      "loss": 5.1571,
-      "step": 42392
-    },
-    {
-      "epoch": 22.108474576271185,
-      "grad_norm": 1.558252215385437,
-      "learning_rate": 5.7912562814070356e-05,
-      "loss": 5.3734,
-      "step": 42393
-    },
-    {
-      "epoch": 22.108996088657104,
-      "grad_norm": 1.6999043226242065,
-      "learning_rate": 5.791155778894473e-05,
-      "loss": 5.3371,
-      "step": 42394
-    },
-    {
-      "epoch": 22.109517601043024,
-      "grad_norm": 1.5452244281768799,
-      "learning_rate": 5.7910552763819105e-05,
-      "loss": 5.3142,
-      "step": 42395
-    },
-    {
-      "epoch": 22.110039113428943,
-      "grad_norm": 1.4969416856765747,
-      "learning_rate": 5.790954773869347e-05,
-      "loss": 5.5275,
-      "step": 42396
-    },
-    {
-      "epoch": 22.110560625814863,
-      "grad_norm": 1.7184135913848877,
-      "learning_rate": 5.7908542713567834e-05,
-      "loss": 5.2162,
-      "step": 42397
-    },
-    {
-      "epoch": 22.111082138200782,
-      "grad_norm": 1.5630230903625488,
-      "learning_rate": 5.790753768844221e-05,
-      "loss": 5.4283,
-      "step": 42398
-    },
-    {
-      "epoch": 22.1116036505867,
-      "grad_norm": 1.5013071298599243,
-      "learning_rate": 5.790653266331658e-05,
-      "loss": 5.5491,
-      "step": 42399
-    },
-    {
-      "epoch": 22.11212516297262,
-      "grad_norm": 1.5905548334121704,
-      "learning_rate": 5.790552763819096e-05,
-      "loss": 5.1892,
-      "step": 42400
-    },
-    {
-      "epoch": 22.11264667535854,
-      "grad_norm": 1.5060958862304688,
-      "learning_rate": 5.7904522613065325e-05,
-      "loss": 5.6362,
-      "step": 42401
-    },
-    {
-      "epoch": 22.11316818774446,
-      "grad_norm": 1.455564022064209,
-      "learning_rate": 5.79035175879397e-05,
-      "loss": 5.1936,
-      "step": 42402
-    },
-    {
-      "epoch": 22.11368970013038,
-      "grad_norm": 1.4664674997329712,
-      "learning_rate": 5.7902512562814074e-05,
-      "loss": 5.5896,
-      "step": 42403
-    },
-    {
-      "epoch": 22.1142112125163,
-      "grad_norm": 1.4814037084579468,
-      "learning_rate": 5.790150753768845e-05,
-      "loss": 5.2531,
-      "step": 42404
-    },
-    {
-      "epoch": 22.114732724902215,
-      "grad_norm": 1.4196617603302002,
-      "learning_rate": 5.7900502512562816e-05,
-      "loss": 5.4499,
-      "step": 42405
-    },
-    {
-      "epoch": 22.115254237288134,
-      "grad_norm": 1.360715389251709,
-      "learning_rate": 5.7899497487437194e-05,
-      "loss": 4.9149,
-      "step": 42406
-    },
-    {
-      "epoch": 22.115775749674054,
-      "grad_norm": 1.5204967260360718,
-      "learning_rate": 5.789849246231156e-05,
-      "loss": 5.1063,
-      "step": 42407
-    },
-    {
-      "epoch": 22.116297262059973,
-      "grad_norm": 1.4832125902175903,
-      "learning_rate": 5.7897487437185936e-05,
-      "loss": 5.2975,
-      "step": 42408
-    },
-    {
-      "epoch": 22.116818774445893,
-      "grad_norm": 1.7650363445281982,
-      "learning_rate": 5.789648241206031e-05,
-      "loss": 4.9583,
-      "step": 42409
-    },
-    {
-      "epoch": 22.117340286831812,
-      "grad_norm": 1.5806444883346558,
-      "learning_rate": 5.789547738693467e-05,
-      "loss": 5.29,
-      "step": 42410
-    },
-    {
-      "epoch": 22.11786179921773,
-      "grad_norm": 1.5122441053390503,
-      "learning_rate": 5.789447236180905e-05,
-      "loss": 5.4857,
-      "step": 42411
-    },
-    {
-      "epoch": 22.11838331160365,
-      "grad_norm": 1.6182904243469238,
-      "learning_rate": 5.789346733668341e-05,
-      "loss": 4.6349,
-      "step": 42412
-    },
-    {
-      "epoch": 22.11890482398957,
-      "grad_norm": 1.5898643732070923,
-      "learning_rate": 5.789246231155779e-05,
-      "loss": 4.9678,
-      "step": 42413
-    },
-    {
-      "epoch": 22.11942633637549,
-      "grad_norm": 1.505020022392273,
-      "learning_rate": 5.789145728643216e-05,
-      "loss": 5.2337,
-      "step": 42414
-    },
-    {
-      "epoch": 22.11994784876141,
-      "grad_norm": 1.5221680402755737,
-      "learning_rate": 5.789045226130654e-05,
-      "loss": 5.1032,
-      "step": 42415
-    },
-    {
-      "epoch": 22.12046936114733,
-      "grad_norm": 1.5927999019622803,
-      "learning_rate": 5.7889447236180904e-05,
-      "loss": 4.9481,
-      "step": 42416
-    },
-    {
-      "epoch": 22.120990873533245,
-      "grad_norm": 1.4744617938995361,
-      "learning_rate": 5.788844221105528e-05,
-      "loss": 5.5805,
-      "step": 42417
-    },
-    {
-      "epoch": 22.121512385919164,
-      "grad_norm": 1.4752416610717773,
-      "learning_rate": 5.7887437185929646e-05,
-      "loss": 5.1436,
-      "step": 42418
-    },
-    {
-      "epoch": 22.122033898305084,
-      "grad_norm": 1.4079784154891968,
-      "learning_rate": 5.7886432160804024e-05,
-      "loss": 5.5562,
-      "step": 42419
-    },
-    {
-      "epoch": 22.122555410691003,
-      "grad_norm": 1.5233315229415894,
-      "learning_rate": 5.7885427135678395e-05,
-      "loss": 5.5424,
-      "step": 42420
-    },
-    {
-      "epoch": 22.123076923076923,
-      "grad_norm": 1.612548828125,
-      "learning_rate": 5.788442211055277e-05,
-      "loss": 4.9846,
-      "step": 42421
-    },
-    {
-      "epoch": 22.123598435462842,
-      "grad_norm": 1.4921735525131226,
-      "learning_rate": 5.788341708542714e-05,
-      "loss": 5.2717,
-      "step": 42422
-    },
-    {
-      "epoch": 22.12411994784876,
-      "grad_norm": 1.5684130191802979,
-      "learning_rate": 5.788241206030151e-05,
-      "loss": 5.1035,
-      "step": 42423
-    },
-    {
-      "epoch": 22.12464146023468,
-      "grad_norm": 1.5260653495788574,
-      "learning_rate": 5.7881407035175886e-05,
-      "loss": 5.1471,
-      "step": 42424
-    },
-    {
-      "epoch": 22.1251629726206,
-      "grad_norm": 1.4931695461273193,
-      "learning_rate": 5.788040201005025e-05,
-      "loss": 5.3753,
-      "step": 42425
-    },
-    {
-      "epoch": 22.12568448500652,
-      "grad_norm": 1.5880813598632812,
-      "learning_rate": 5.787939698492463e-05,
-      "loss": 5.3603,
-      "step": 42426
-    },
-    {
-      "epoch": 22.12620599739244,
-      "grad_norm": 1.4524049758911133,
-      "learning_rate": 5.787839195979899e-05,
-      "loss": 5.2705,
-      "step": 42427
-    },
-    {
-      "epoch": 22.12672750977836,
-      "grad_norm": 1.5040369033813477,
-      "learning_rate": 5.787738693467337e-05,
-      "loss": 5.0453,
-      "step": 42428
-    },
-    {
-      "epoch": 22.127249022164275,
-      "grad_norm": 1.5458515882492065,
-      "learning_rate": 5.787638190954774e-05,
-      "loss": 5.2505,
-      "step": 42429
-    },
-    {
-      "epoch": 22.127770534550194,
-      "grad_norm": 1.498842477798462,
-      "learning_rate": 5.787537688442212e-05,
-      "loss": 5.5789,
-      "step": 42430
-    },
-    {
-      "epoch": 22.128292046936114,
-      "grad_norm": 1.5059565305709839,
-      "learning_rate": 5.7874371859296483e-05,
-      "loss": 5.3404,
-      "step": 42431
-    },
-    {
-      "epoch": 22.128813559322033,
-      "grad_norm": 1.5272843837738037,
-      "learning_rate": 5.787336683417086e-05,
-      "loss": 5.4054,
-      "step": 42432
-    },
-    {
-      "epoch": 22.129335071707953,
-      "grad_norm": 1.566743016242981,
-      "learning_rate": 5.7872361809045226e-05,
-      "loss": 5.2445,
-      "step": 42433
-    },
-    {
-      "epoch": 22.129856584093872,
-      "grad_norm": 1.5728050470352173,
-      "learning_rate": 5.78713567839196e-05,
-      "loss": 5.0965,
-      "step": 42434
-    },
-    {
-      "epoch": 22.13037809647979,
-      "grad_norm": 1.5137262344360352,
-      "learning_rate": 5.7870351758793974e-05,
-      "loss": 5.4532,
-      "step": 42435
-    },
-    {
-      "epoch": 22.13089960886571,
-      "grad_norm": 1.5174181461334229,
-      "learning_rate": 5.786934673366834e-05,
-      "loss": 5.4427,
-      "step": 42436
-    },
-    {
-      "epoch": 22.13142112125163,
-      "grad_norm": 1.4334295988082886,
-      "learning_rate": 5.7868341708542717e-05,
-      "loss": 5.9383,
-      "step": 42437
-    },
-    {
-      "epoch": 22.13194263363755,
-      "grad_norm": 1.5448453426361084,
-      "learning_rate": 5.786733668341708e-05,
-      "loss": 5.3783,
-      "step": 42438
-    },
-    {
-      "epoch": 22.13246414602347,
-      "grad_norm": 1.5559437274932861,
-      "learning_rate": 5.786633165829146e-05,
-      "loss": 5.0697,
-      "step": 42439
-    },
-    {
-      "epoch": 22.13298565840939,
-      "grad_norm": 1.5171972513198853,
-      "learning_rate": 5.786532663316583e-05,
-      "loss": 5.3929,
-      "step": 42440
-    },
-    {
-      "epoch": 22.133507170795305,
-      "grad_norm": 1.522678256034851,
-      "learning_rate": 5.786432160804021e-05,
-      "loss": 5.4585,
-      "step": 42441
-    },
-    {
-      "epoch": 22.134028683181224,
-      "grad_norm": 1.4943335056304932,
-      "learning_rate": 5.786331658291457e-05,
-      "loss": 5.5293,
-      "step": 42442
-    },
-    {
-      "epoch": 22.134550195567144,
-      "grad_norm": 1.5469905138015747,
-      "learning_rate": 5.786231155778895e-05,
-      "loss": 5.137,
-      "step": 42443
-    },
-    {
-      "epoch": 22.135071707953063,
-      "grad_norm": 1.6245616674423218,
-      "learning_rate": 5.786130653266332e-05,
-      "loss": 5.1763,
-      "step": 42444
-    },
-    {
-      "epoch": 22.135593220338983,
-      "grad_norm": 1.5286318063735962,
-      "learning_rate": 5.78603015075377e-05,
-      "loss": 4.6862,
-      "step": 42445
-    },
-    {
-      "epoch": 22.136114732724902,
-      "grad_norm": 1.614904761314392,
-      "learning_rate": 5.785929648241206e-05,
-      "loss": 5.5257,
-      "step": 42446
-    },
-    {
-      "epoch": 22.13663624511082,
-      "grad_norm": 1.5107609033584595,
-      "learning_rate": 5.785829145728644e-05,
-      "loss": 5.3874,
-      "step": 42447
-    },
-    {
-      "epoch": 22.13715775749674,
-      "grad_norm": 1.6361048221588135,
-      "learning_rate": 5.7857286432160805e-05,
-      "loss": 5.2898,
-      "step": 42448
-    },
-    {
-      "epoch": 22.13767926988266,
-      "grad_norm": 1.6439523696899414,
-      "learning_rate": 5.7856281407035176e-05,
-      "loss": 4.8607,
-      "step": 42449
-    },
-    {
-      "epoch": 22.13820078226858,
-      "grad_norm": 1.4825921058654785,
-      "learning_rate": 5.7855276381909554e-05,
-      "loss": 5.1329,
-      "step": 42450
-    },
-    {
-      "epoch": 22.1387222946545,
-      "grad_norm": 1.5446991920471191,
-      "learning_rate": 5.785427135678392e-05,
-      "loss": 5.2465,
-      "step": 42451
-    },
-    {
-      "epoch": 22.13924380704042,
-      "grad_norm": 1.4345476627349854,
-      "learning_rate": 5.7853266331658296e-05,
-      "loss": 4.8215,
-      "step": 42452
-    },
-    {
-      "epoch": 22.139765319426335,
-      "grad_norm": 1.669278860092163,
-      "learning_rate": 5.785226130653266e-05,
-      "loss": 4.9634,
-      "step": 42453
-    },
-    {
-      "epoch": 22.140286831812254,
-      "grad_norm": 1.4580302238464355,
-      "learning_rate": 5.785125628140704e-05,
-      "loss": 5.333,
-      "step": 42454
-    },
-    {
-      "epoch": 22.140808344198174,
-      "grad_norm": 1.5962154865264893,
-      "learning_rate": 5.785025125628141e-05,
-      "loss": 5.0291,
-      "step": 42455
-    },
-    {
-      "epoch": 22.141329856584093,
-      "grad_norm": 1.5726583003997803,
-      "learning_rate": 5.784924623115579e-05,
-      "loss": 5.2439,
-      "step": 42456
-    },
-    {
-      "epoch": 22.141851368970013,
-      "grad_norm": 1.5028910636901855,
-      "learning_rate": 5.784824120603015e-05,
-      "loss": 5.4363,
-      "step": 42457
-    },
-    {
-      "epoch": 22.142372881355932,
-      "grad_norm": 1.581496238708496,
-      "learning_rate": 5.784723618090453e-05,
-      "loss": 5.0561,
-      "step": 42458
-    },
-    {
-      "epoch": 22.14289439374185,
-      "grad_norm": 1.6305019855499268,
-      "learning_rate": 5.784623115577889e-05,
-      "loss": 5.0364,
-      "step": 42459
-    },
-    {
-      "epoch": 22.14341590612777,
-      "grad_norm": 1.952868103981018,
-      "learning_rate": 5.784522613065327e-05,
-      "loss": 4.9313,
-      "step": 42460
-    },
-    {
-      "epoch": 22.14393741851369,
-      "grad_norm": 1.5616897344589233,
-      "learning_rate": 5.784422110552764e-05,
-      "loss": 5.5213,
-      "step": 42461
-    },
-    {
-      "epoch": 22.14445893089961,
-      "grad_norm": 1.5195074081420898,
-      "learning_rate": 5.7843216080402006e-05,
-      "loss": 5.3328,
-      "step": 42462
-    },
-    {
-      "epoch": 22.14498044328553,
-      "grad_norm": 1.4279369115829468,
-      "learning_rate": 5.7842211055276384e-05,
-      "loss": 5.736,
-      "step": 42463
-    },
-    {
-      "epoch": 22.14550195567145,
-      "grad_norm": 1.4495587348937988,
-      "learning_rate": 5.7841206030150755e-05,
-      "loss": 4.7346,
-      "step": 42464
-    },
-    {
-      "epoch": 22.146023468057365,
-      "grad_norm": 1.4699784517288208,
-      "learning_rate": 5.784020100502513e-05,
-      "loss": 5.0945,
-      "step": 42465
-    },
-    {
-      "epoch": 22.146544980443284,
-      "grad_norm": 1.706677794456482,
-      "learning_rate": 5.78391959798995e-05,
-      "loss": 4.8237,
-      "step": 42466
-    },
-    {
-      "epoch": 22.147066492829204,
-      "grad_norm": 1.6133407354354858,
-      "learning_rate": 5.7838190954773875e-05,
-      "loss": 4.9414,
-      "step": 42467
-    },
-    {
-      "epoch": 22.147588005215123,
-      "grad_norm": 1.4920588731765747,
-      "learning_rate": 5.783718592964824e-05,
-      "loss": 5.4182,
-      "step": 42468
-    },
-    {
-      "epoch": 22.148109517601043,
-      "grad_norm": 1.6906228065490723,
-      "learning_rate": 5.783618090452262e-05,
-      "loss": 5.1222,
-      "step": 42469
-    },
-    {
-      "epoch": 22.148631029986962,
-      "grad_norm": 1.5719300508499146,
-      "learning_rate": 5.783517587939699e-05,
-      "loss": 5.4841,
-      "step": 42470
-    },
-    {
-      "epoch": 22.14915254237288,
-      "grad_norm": 1.4884006977081299,
-      "learning_rate": 5.7834170854271366e-05,
-      "loss": 5.3088,
-      "step": 42471
-    },
-    {
-      "epoch": 22.1496740547588,
-      "grad_norm": 1.5142745971679688,
-      "learning_rate": 5.783316582914573e-05,
-      "loss": 4.8832,
-      "step": 42472
-    },
-    {
-      "epoch": 22.15019556714472,
-      "grad_norm": 1.527806043624878,
-      "learning_rate": 5.783216080402011e-05,
-      "loss": 5.2662,
-      "step": 42473
-    },
-    {
-      "epoch": 22.15071707953064,
-      "grad_norm": 1.5068389177322388,
-      "learning_rate": 5.783115577889447e-05,
-      "loss": 5.0321,
-      "step": 42474
-    },
-    {
-      "epoch": 22.15123859191656,
-      "grad_norm": 1.7204368114471436,
-      "learning_rate": 5.783015075376885e-05,
-      "loss": 4.2287,
-      "step": 42475
-    },
-    {
-      "epoch": 22.151760104302475,
-      "grad_norm": 1.623386025428772,
-      "learning_rate": 5.782914572864322e-05,
-      "loss": 5.3817,
-      "step": 42476
-    },
-    {
-      "epoch": 22.152281616688395,
-      "grad_norm": 1.5028835535049438,
-      "learning_rate": 5.7828140703517586e-05,
-      "loss": 5.2626,
-      "step": 42477
-    },
-    {
-      "epoch": 22.152803129074314,
-      "grad_norm": 1.5511829853057861,
-      "learning_rate": 5.7827135678391964e-05,
-      "loss": 5.147,
-      "step": 42478
-    },
-    {
-      "epoch": 22.153324641460234,
-      "grad_norm": 1.5281428098678589,
-      "learning_rate": 5.782613065326633e-05,
-      "loss": 5.6311,
-      "step": 42479
-    },
-    {
-      "epoch": 22.153846153846153,
-      "grad_norm": 1.6268402338027954,
-      "learning_rate": 5.7825125628140706e-05,
-      "loss": 5.0836,
-      "step": 42480
-    },
-    {
-      "epoch": 22.154367666232073,
-      "grad_norm": 1.614466667175293,
-      "learning_rate": 5.782412060301508e-05,
-      "loss": 5.2388,
-      "step": 42481
-    },
-    {
-      "epoch": 22.154889178617992,
-      "grad_norm": 1.6218923330307007,
-      "learning_rate": 5.7823115577889455e-05,
-      "loss": 5.4546,
-      "step": 42482
-    },
-    {
-      "epoch": 22.15541069100391,
-      "grad_norm": 1.5170239210128784,
-      "learning_rate": 5.782211055276382e-05,
-      "loss": 5.5063,
-      "step": 42483
-    },
-    {
-      "epoch": 22.15593220338983,
-      "grad_norm": 1.614268183708191,
-      "learning_rate": 5.78211055276382e-05,
-      "loss": 4.7181,
-      "step": 42484
-    },
-    {
-      "epoch": 22.15645371577575,
-      "grad_norm": 1.5421935319900513,
-      "learning_rate": 5.782010050251256e-05,
-      "loss": 4.39,
-      "step": 42485
-    },
-    {
-      "epoch": 22.15697522816167,
-      "grad_norm": 1.5646377801895142,
-      "learning_rate": 5.781909547738694e-05,
-      "loss": 4.9358,
-      "step": 42486
-    },
-    {
-      "epoch": 22.15749674054759,
-      "grad_norm": 1.472159504890442,
-      "learning_rate": 5.781809045226131e-05,
-      "loss": 5.5702,
-      "step": 42487
-    },
-    {
-      "epoch": 22.158018252933505,
-      "grad_norm": 1.6557201147079468,
-      "learning_rate": 5.781708542713569e-05,
-      "loss": 5.433,
-      "step": 42488
-    },
-    {
-      "epoch": 22.158539765319425,
-      "grad_norm": 1.5124895572662354,
-      "learning_rate": 5.781608040201005e-05,
-      "loss": 5.2726,
-      "step": 42489
-    },
-    {
-      "epoch": 22.159061277705344,
-      "grad_norm": 1.5945518016815186,
-      "learning_rate": 5.781507537688442e-05,
-      "loss": 5.4863,
-      "step": 42490
-    },
-    {
-      "epoch": 22.159582790091264,
-      "grad_norm": 1.6163686513900757,
-      "learning_rate": 5.78140703517588e-05,
-      "loss": 4.9146,
-      "step": 42491
-    },
-    {
-      "epoch": 22.160104302477183,
-      "grad_norm": 1.4189941883087158,
-      "learning_rate": 5.7813065326633165e-05,
-      "loss": 5.5887,
-      "step": 42492
-    },
-    {
-      "epoch": 22.160625814863103,
-      "grad_norm": 1.5492609739303589,
-      "learning_rate": 5.781206030150754e-05,
-      "loss": 5.2678,
-      "step": 42493
-    },
-    {
-      "epoch": 22.161147327249022,
-      "grad_norm": 1.5212125778198242,
-      "learning_rate": 5.781105527638191e-05,
-      "loss": 5.5067,
-      "step": 42494
-    },
-    {
-      "epoch": 22.16166883963494,
-      "grad_norm": 1.6353012323379517,
-      "learning_rate": 5.7810050251256285e-05,
-      "loss": 5.1052,
-      "step": 42495
-    },
-    {
-      "epoch": 22.16219035202086,
-      "grad_norm": 1.5542573928833008,
-      "learning_rate": 5.7809045226130656e-05,
-      "loss": 4.6794,
-      "step": 42496
-    },
-    {
-      "epoch": 22.16271186440678,
-      "grad_norm": 1.530464768409729,
-      "learning_rate": 5.7808040201005034e-05,
-      "loss": 5.2556,
-      "step": 42497
-    },
-    {
-      "epoch": 22.1632333767927,
-      "grad_norm": 1.4862868785858154,
-      "learning_rate": 5.78070351758794e-05,
-      "loss": 5.3835,
-      "step": 42498
-    },
-    {
-      "epoch": 22.16375488917862,
-      "grad_norm": 1.6068871021270752,
-      "learning_rate": 5.7806030150753776e-05,
-      "loss": 5.1598,
-      "step": 42499
-    },
-    {
-      "epoch": 22.164276401564535,
-      "grad_norm": 1.404327154159546,
-      "learning_rate": 5.780502512562814e-05,
-      "loss": 5.7232,
-      "step": 42500
-    },
-    {
-      "epoch": 22.164797913950455,
-      "grad_norm": 1.6383136510849,
-      "learning_rate": 5.780402010050252e-05,
-      "loss": 4.8621,
-      "step": 42501
-    },
-    {
-      "epoch": 22.165319426336374,
-      "grad_norm": 1.65328049659729,
-      "learning_rate": 5.780301507537689e-05,
-      "loss": 4.988,
-      "step": 42502
-    },
-    {
-      "epoch": 22.165840938722294,
-      "grad_norm": 1.4483566284179688,
-      "learning_rate": 5.780201005025125e-05,
-      "loss": 5.5363,
-      "step": 42503
-    },
-    {
-      "epoch": 22.166362451108213,
-      "grad_norm": 1.7229372262954712,
-      "learning_rate": 5.780100502512563e-05,
-      "loss": 4.9346,
-      "step": 42504
-    },
-    {
-      "epoch": 22.166883963494133,
-      "grad_norm": 1.5034925937652588,
-      "learning_rate": 5.7799999999999995e-05,
-      "loss": 4.5823,
-      "step": 42505
-    },
-    {
-      "epoch": 22.167405475880052,
-      "grad_norm": 1.5120712518692017,
-      "learning_rate": 5.779899497487437e-05,
-      "loss": 5.554,
-      "step": 42506
-    },
-    {
-      "epoch": 22.16792698826597,
-      "grad_norm": 1.5640324354171753,
-      "learning_rate": 5.7797989949748744e-05,
-      "loss": 5.0052,
-      "step": 42507
-    },
-    {
-      "epoch": 22.16844850065189,
-      "grad_norm": 1.5640054941177368,
-      "learning_rate": 5.779698492462312e-05,
-      "loss": 5.2134,
-      "step": 42508
-    },
-    {
-      "epoch": 22.16897001303781,
-      "grad_norm": 1.4938898086547852,
-      "learning_rate": 5.7795979899497486e-05,
-      "loss": 5.2013,
-      "step": 42509
-    },
-    {
-      "epoch": 22.16949152542373,
-      "grad_norm": 1.4905893802642822,
-      "learning_rate": 5.7794974874371864e-05,
-      "loss": 5.0494,
-      "step": 42510
-    },
-    {
-      "epoch": 22.17001303780965,
-      "grad_norm": 1.5026718378067017,
-      "learning_rate": 5.7793969849246235e-05,
-      "loss": 5.2788,
-      "step": 42511
-    },
-    {
-      "epoch": 22.170534550195566,
-      "grad_norm": 1.4981063604354858,
-      "learning_rate": 5.779296482412061e-05,
-      "loss": 4.6325,
-      "step": 42512
-    },
-    {
-      "epoch": 22.171056062581485,
-      "grad_norm": 1.6805537939071655,
-      "learning_rate": 5.779195979899498e-05,
-      "loss": 5.4506,
-      "step": 42513
-    },
-    {
-      "epoch": 22.171577574967404,
-      "grad_norm": 1.5829294919967651,
-      "learning_rate": 5.7790954773869355e-05,
-      "loss": 5.3093,
-      "step": 42514
-    },
-    {
-      "epoch": 22.172099087353324,
-      "grad_norm": 1.5502252578735352,
-      "learning_rate": 5.778994974874372e-05,
-      "loss": 5.1075,
-      "step": 42515
-    },
-    {
-      "epoch": 22.172620599739243,
-      "grad_norm": 1.6039518117904663,
-      "learning_rate": 5.778894472361809e-05,
-      "loss": 4.8972,
-      "step": 42516
-    },
-    {
-      "epoch": 22.173142112125163,
-      "grad_norm": 1.592633605003357,
-      "learning_rate": 5.778793969849247e-05,
-      "loss": 5.474,
-      "step": 42517
-    },
-    {
-      "epoch": 22.173663624511082,
-      "grad_norm": 1.5572813749313354,
-      "learning_rate": 5.778693467336683e-05,
-      "loss": 5.1167,
-      "step": 42518
-    },
-    {
-      "epoch": 22.174185136897,
-      "grad_norm": 1.5340052843093872,
-      "learning_rate": 5.778592964824121e-05,
-      "loss": 5.2024,
-      "step": 42519
-    },
-    {
-      "epoch": 22.17470664928292,
-      "grad_norm": 1.4748904705047607,
-      "learning_rate": 5.7784924623115575e-05,
-      "loss": 5.2721,
-      "step": 42520
-    },
-    {
-      "epoch": 22.17522816166884,
-      "grad_norm": 1.5882571935653687,
-      "learning_rate": 5.778391959798995e-05,
-      "loss": 5.1149,
-      "step": 42521
-    },
-    {
-      "epoch": 22.17574967405476,
-      "grad_norm": 1.6267693042755127,
-      "learning_rate": 5.7782914572864324e-05,
-      "loss": 5.2073,
-      "step": 42522
-    },
-    {
-      "epoch": 22.17627118644068,
-      "grad_norm": 1.5906232595443726,
-      "learning_rate": 5.77819095477387e-05,
-      "loss": 5.3144,
-      "step": 42523
-    },
-    {
-      "epoch": 22.176792698826596,
-      "grad_norm": 1.5212279558181763,
-      "learning_rate": 5.7780904522613066e-05,
-      "loss": 5.3769,
-      "step": 42524
-    },
-    {
-      "epoch": 22.177314211212515,
-      "grad_norm": 1.5507404804229736,
-      "learning_rate": 5.7779899497487444e-05,
-      "loss": 5.3783,
-      "step": 42525
-    },
-    {
-      "epoch": 22.177835723598434,
-      "grad_norm": 1.4369686841964722,
-      "learning_rate": 5.777889447236181e-05,
-      "loss": 5.501,
-      "step": 42526
-    },
-    {
-      "epoch": 22.178357235984354,
-      "grad_norm": 1.5280250310897827,
-      "learning_rate": 5.7777889447236186e-05,
-      "loss": 5.0346,
-      "step": 42527
-    },
-    {
-      "epoch": 22.178878748370273,
-      "grad_norm": 1.5499906539916992,
-      "learning_rate": 5.777688442211056e-05,
-      "loss": 5.246,
-      "step": 42528
-    },
-    {
-      "epoch": 22.179400260756193,
-      "grad_norm": 1.5179470777511597,
-      "learning_rate": 5.777587939698492e-05,
-      "loss": 5.1357,
-      "step": 42529
-    },
-    {
-      "epoch": 22.179921773142112,
-      "grad_norm": 1.5376241207122803,
-      "learning_rate": 5.77748743718593e-05,
-      "loss": 5.0403,
-      "step": 42530
-    },
-    {
-      "epoch": 22.180443285528032,
-      "grad_norm": 1.707153081893921,
-      "learning_rate": 5.777386934673367e-05,
-      "loss": 5.0533,
-      "step": 42531
-    },
-    {
-      "epoch": 22.18096479791395,
-      "grad_norm": 1.514082908630371,
-      "learning_rate": 5.777286432160805e-05,
-      "loss": 5.1297,
-      "step": 42532
-    },
-    {
-      "epoch": 22.18148631029987,
-      "grad_norm": 1.6012250185012817,
-      "learning_rate": 5.777185929648241e-05,
-      "loss": 5.0699,
-      "step": 42533
-    },
-    {
-      "epoch": 22.18200782268579,
-      "grad_norm": 1.5024631023406982,
-      "learning_rate": 5.777085427135679e-05,
-      "loss": 4.3468,
-      "step": 42534
-    },
-    {
-      "epoch": 22.18252933507171,
-      "grad_norm": 1.494377613067627,
-      "learning_rate": 5.7769849246231154e-05,
-      "loss": 5.3719,
-      "step": 42535
-    },
-    {
-      "epoch": 22.183050847457626,
-      "grad_norm": 1.5651891231536865,
-      "learning_rate": 5.776884422110553e-05,
-      "loss": 5.5365,
-      "step": 42536
-    },
-    {
-      "epoch": 22.183572359843545,
-      "grad_norm": 1.612901210784912,
-      "learning_rate": 5.77678391959799e-05,
-      "loss": 4.9181,
-      "step": 42537
-    },
-    {
-      "epoch": 22.184093872229464,
-      "grad_norm": 1.442598819732666,
-      "learning_rate": 5.776683417085428e-05,
-      "loss": 5.2512,
-      "step": 42538
-    },
-    {
-      "epoch": 22.184615384615384,
-      "grad_norm": 1.598738431930542,
-      "learning_rate": 5.7765829145728645e-05,
-      "loss": 4.9234,
-      "step": 42539
-    },
-    {
-      "epoch": 22.185136897001303,
-      "grad_norm": 2.189188003540039,
-      "learning_rate": 5.776482412060302e-05,
-      "loss": 4.7128,
-      "step": 42540
-    },
-    {
-      "epoch": 22.185658409387223,
-      "grad_norm": 1.5383086204528809,
-      "learning_rate": 5.776381909547739e-05,
-      "loss": 4.9109,
-      "step": 42541
-    },
-    {
-      "epoch": 22.186179921773142,
-      "grad_norm": 1.6215639114379883,
-      "learning_rate": 5.776281407035176e-05,
-      "loss": 5.2811,
-      "step": 42542
-    },
-    {
-      "epoch": 22.186701434159062,
-      "grad_norm": 1.4468625783920288,
-      "learning_rate": 5.7761809045226136e-05,
-      "loss": 5.3306,
-      "step": 42543
-    },
-    {
-      "epoch": 22.18722294654498,
-      "grad_norm": 1.5875988006591797,
-      "learning_rate": 5.77608040201005e-05,
-      "loss": 5.2358,
-      "step": 42544
-    },
-    {
-      "epoch": 22.1877444589309,
-      "grad_norm": 1.4534822702407837,
-      "learning_rate": 5.775979899497488e-05,
-      "loss": 5.481,
-      "step": 42545
-    },
-    {
-      "epoch": 22.18826597131682,
-      "grad_norm": 1.495900273323059,
-      "learning_rate": 5.775879396984924e-05,
-      "loss": 5.312,
-      "step": 42546
-    },
-    {
-      "epoch": 22.18878748370274,
-      "grad_norm": 1.4406330585479736,
-      "learning_rate": 5.775778894472362e-05,
-      "loss": 5.6595,
-      "step": 42547
-    },
-    {
-      "epoch": 22.189308996088656,
-      "grad_norm": 1.5420492887496948,
-      "learning_rate": 5.775678391959799e-05,
-      "loss": 4.8101,
-      "step": 42548
-    },
-    {
-      "epoch": 22.189830508474575,
-      "grad_norm": 1.567585825920105,
-      "learning_rate": 5.775577889447237e-05,
-      "loss": 4.9244,
-      "step": 42549
-    },
-    {
-      "epoch": 22.190352020860495,
-      "grad_norm": 1.5246533155441284,
-      "learning_rate": 5.7754773869346733e-05,
-      "loss": 5.5939,
-      "step": 42550
-    },
-    {
-      "epoch": 22.190873533246414,
-      "grad_norm": 1.62144935131073,
-      "learning_rate": 5.775376884422111e-05,
-      "loss": 5.1667,
-      "step": 42551
-    },
-    {
-      "epoch": 22.191395045632333,
-      "grad_norm": 1.5452485084533691,
-      "learning_rate": 5.7752763819095476e-05,
-      "loss": 4.8752,
-      "step": 42552
-    },
-    {
-      "epoch": 22.191916558018253,
-      "grad_norm": 1.4873158931732178,
-      "learning_rate": 5.7751758793969853e-05,
-      "loss": 5.4687,
-      "step": 42553
-    },
-    {
-      "epoch": 22.192438070404172,
-      "grad_norm": 1.5990431308746338,
-      "learning_rate": 5.7750753768844224e-05,
-      "loss": 5.3453,
-      "step": 42554
-    },
-    {
-      "epoch": 22.192959582790092,
-      "grad_norm": 1.5737054347991943,
-      "learning_rate": 5.774974874371859e-05,
-      "loss": 5.4789,
-      "step": 42555
-    },
-    {
-      "epoch": 22.19348109517601,
-      "grad_norm": 1.5673563480377197,
-      "learning_rate": 5.7748743718592967e-05,
-      "loss": 5.342,
-      "step": 42556
-    },
-    {
-      "epoch": 22.19400260756193,
-      "grad_norm": 1.882723093032837,
-      "learning_rate": 5.774773869346734e-05,
-      "loss": 5.2589,
-      "step": 42557
-    },
-    {
-      "epoch": 22.19452411994785,
-      "grad_norm": 1.5392907857894897,
-      "learning_rate": 5.7746733668341715e-05,
-      "loss": 4.9383,
-      "step": 42558
-    },
-    {
-      "epoch": 22.195045632333766,
-      "grad_norm": 1.4543898105621338,
-      "learning_rate": 5.774572864321608e-05,
-      "loss": 5.2059,
-      "step": 42559
-    },
-    {
-      "epoch": 22.195567144719686,
-      "grad_norm": 1.4797441959381104,
-      "learning_rate": 5.774472361809046e-05,
-      "loss": 5.4585,
-      "step": 42560
-    },
-    {
-      "epoch": 22.196088657105605,
-      "grad_norm": 1.5450032949447632,
-      "learning_rate": 5.774371859296482e-05,
-      "loss": 5.3872,
-      "step": 42561
-    },
-    {
-      "epoch": 22.196610169491525,
-      "grad_norm": 1.5037400722503662,
-      "learning_rate": 5.77427135678392e-05,
-      "loss": 5.4688,
-      "step": 42562
-    },
-    {
-      "epoch": 22.197131681877444,
-      "grad_norm": 1.5387507677078247,
-      "learning_rate": 5.774170854271357e-05,
-      "loss": 5.0315,
-      "step": 42563
-    },
-    {
-      "epoch": 22.197653194263363,
-      "grad_norm": 1.5901215076446533,
-      "learning_rate": 5.774070351758795e-05,
-      "loss": 5.0382,
-      "step": 42564
-    },
-    {
-      "epoch": 22.198174706649283,
-      "grad_norm": 1.4452202320098877,
-      "learning_rate": 5.773969849246231e-05,
-      "loss": 5.4021,
-      "step": 42565
-    },
-    {
-      "epoch": 22.198696219035202,
-      "grad_norm": 1.4574350118637085,
-      "learning_rate": 5.773869346733669e-05,
-      "loss": 5.4893,
-      "step": 42566
-    },
-    {
-      "epoch": 22.199217731421122,
-      "grad_norm": 1.4673196077346802,
-      "learning_rate": 5.7737688442211055e-05,
-      "loss": 5.0766,
-      "step": 42567
-    },
-    {
-      "epoch": 22.19973924380704,
-      "grad_norm": 1.6169793605804443,
-      "learning_rate": 5.773668341708543e-05,
-      "loss": 5.4718,
-      "step": 42568
-    },
-    {
-      "epoch": 22.20026075619296,
-      "grad_norm": 1.5590616464614868,
-      "learning_rate": 5.7735678391959804e-05,
-      "loss": 5.3002,
-      "step": 42569
-    },
-    {
-      "epoch": 22.20078226857888,
-      "grad_norm": 1.5284812450408936,
-      "learning_rate": 5.773467336683417e-05,
-      "loss": 5.4893,
-      "step": 42570
-    },
-    {
-      "epoch": 22.201303780964796,
-      "grad_norm": 1.5278363227844238,
-      "learning_rate": 5.7733668341708546e-05,
-      "loss": 5.0048,
-      "step": 42571
-    },
-    {
-      "epoch": 22.201825293350716,
-      "grad_norm": 1.552660346031189,
-      "learning_rate": 5.773266331658291e-05,
-      "loss": 5.176,
-      "step": 42572
-    },
-    {
-      "epoch": 22.202346805736635,
-      "grad_norm": 1.583893060684204,
-      "learning_rate": 5.773165829145729e-05,
-      "loss": 5.2547,
-      "step": 42573
-    },
-    {
-      "epoch": 22.202868318122555,
-      "grad_norm": 1.5119929313659668,
-      "learning_rate": 5.773065326633166e-05,
-      "loss": 5.3093,
-      "step": 42574
-    },
-    {
-      "epoch": 22.203389830508474,
-      "grad_norm": 1.5216015577316284,
-      "learning_rate": 5.772964824120604e-05,
-      "loss": 5.152,
-      "step": 42575
-    },
-    {
-      "epoch": 22.203911342894393,
-      "grad_norm": 1.648738145828247,
-      "learning_rate": 5.77286432160804e-05,
-      "loss": 5.1119,
-      "step": 42576
-    },
-    {
-      "epoch": 22.204432855280313,
-      "grad_norm": 1.555527925491333,
-      "learning_rate": 5.772763819095478e-05,
-      "loss": 5.3321,
-      "step": 42577
-    },
-    {
-      "epoch": 22.204954367666232,
-      "grad_norm": 1.6664870977401733,
-      "learning_rate": 5.772663316582915e-05,
-      "loss": 5.1898,
-      "step": 42578
-    },
-    {
-      "epoch": 22.205475880052152,
-      "grad_norm": 1.571927785873413,
-      "learning_rate": 5.772562814070353e-05,
-      "loss": 5.311,
-      "step": 42579
-    },
-    {
-      "epoch": 22.20599739243807,
-      "grad_norm": 1.555804967880249,
-      "learning_rate": 5.772462311557789e-05,
-      "loss": 4.9896,
-      "step": 42580
-    },
-    {
-      "epoch": 22.20651890482399,
-      "grad_norm": 1.4492287635803223,
-      "learning_rate": 5.772361809045227e-05,
-      "loss": 4.875,
-      "step": 42581
-    },
-    {
-      "epoch": 22.20704041720991,
-      "grad_norm": 1.5096968412399292,
-      "learning_rate": 5.7722613065326634e-05,
-      "loss": 5.338,
-      "step": 42582
-    },
-    {
-      "epoch": 22.207561929595826,
-      "grad_norm": 1.5618921518325806,
-      "learning_rate": 5.7721608040201005e-05,
-      "loss": 5.4177,
-      "step": 42583
-    },
-    {
-      "epoch": 22.208083441981746,
-      "grad_norm": 1.5084766149520874,
-      "learning_rate": 5.772060301507538e-05,
-      "loss": 5.2551,
-      "step": 42584
-    },
-    {
-      "epoch": 22.208604954367665,
-      "grad_norm": 1.511855125427246,
-      "learning_rate": 5.771959798994975e-05,
-      "loss": 5.4964,
-      "step": 42585
-    },
-    {
-      "epoch": 22.209126466753585,
-      "grad_norm": 1.546643614768982,
-      "learning_rate": 5.7718592964824125e-05,
-      "loss": 5.4038,
-      "step": 42586
-    },
-    {
-      "epoch": 22.209647979139504,
-      "grad_norm": 1.3998074531555176,
-      "learning_rate": 5.771758793969849e-05,
-      "loss": 4.7129,
-      "step": 42587
-    },
-    {
-      "epoch": 22.210169491525424,
-      "grad_norm": 1.4719446897506714,
-      "learning_rate": 5.771658291457287e-05,
-      "loss": 5.4134,
-      "step": 42588
-    },
-    {
-      "epoch": 22.210691003911343,
-      "grad_norm": 1.4947519302368164,
-      "learning_rate": 5.771557788944724e-05,
-      "loss": 5.0808,
-      "step": 42589
-    },
-    {
-      "epoch": 22.211212516297262,
-      "grad_norm": 1.574833869934082,
-      "learning_rate": 5.7714572864321616e-05,
-      "loss": 5.4438,
-      "step": 42590
-    },
-    {
-      "epoch": 22.211734028683182,
-      "grad_norm": 1.5651153326034546,
-      "learning_rate": 5.771356783919598e-05,
-      "loss": 5.1063,
-      "step": 42591
-    },
-    {
-      "epoch": 22.2122555410691,
-      "grad_norm": 1.568068504333496,
-      "learning_rate": 5.771256281407036e-05,
-      "loss": 4.8297,
-      "step": 42592
-    },
-    {
-      "epoch": 22.21277705345502,
-      "grad_norm": 1.5472921133041382,
-      "learning_rate": 5.771155778894472e-05,
-      "loss": 5.2556,
-      "step": 42593
-    },
-    {
-      "epoch": 22.21329856584094,
-      "grad_norm": 1.4602279663085938,
-      "learning_rate": 5.77105527638191e-05,
-      "loss": 5.339,
-      "step": 42594
-    },
-    {
-      "epoch": 22.213820078226856,
-      "grad_norm": 1.5650819540023804,
-      "learning_rate": 5.770954773869347e-05,
-      "loss": 5.4447,
-      "step": 42595
-    },
-    {
-      "epoch": 22.214341590612776,
-      "grad_norm": 1.503077507019043,
-      "learning_rate": 5.7708542713567836e-05,
-      "loss": 5.2636,
-      "step": 42596
-    },
-    {
-      "epoch": 22.214863102998695,
-      "grad_norm": 1.4853368997573853,
-      "learning_rate": 5.7707537688442214e-05,
-      "loss": 5.2138,
-      "step": 42597
-    },
-    {
-      "epoch": 22.215384615384615,
-      "grad_norm": 1.4757829904556274,
-      "learning_rate": 5.7706532663316585e-05,
-      "loss": 5.1226,
-      "step": 42598
-    },
-    {
-      "epoch": 22.215906127770534,
-      "grad_norm": 1.415135383605957,
-      "learning_rate": 5.770552763819096e-05,
-      "loss": 5.7026,
-      "step": 42599
-    },
-    {
-      "epoch": 22.216427640156454,
-      "grad_norm": 1.472921371459961,
-      "learning_rate": 5.770452261306533e-05,
-      "loss": 5.4842,
-      "step": 42600
-    },
-    {
-      "epoch": 22.216949152542373,
-      "grad_norm": 1.436500072479248,
-      "learning_rate": 5.7703517587939705e-05,
-      "loss": 5.5856,
-      "step": 42601
-    },
-    {
-      "epoch": 22.217470664928292,
-      "grad_norm": 1.537264108657837,
-      "learning_rate": 5.770251256281407e-05,
-      "loss": 5.0244,
-      "step": 42602
-    },
-    {
-      "epoch": 22.217992177314212,
-      "grad_norm": 1.4670379161834717,
-      "learning_rate": 5.770150753768845e-05,
-      "loss": 5.5084,
-      "step": 42603
-    },
-    {
-      "epoch": 22.21851368970013,
-      "grad_norm": 1.5485451221466064,
-      "learning_rate": 5.770050251256282e-05,
-      "loss": 5.2092,
-      "step": 42604
-    },
-    {
-      "epoch": 22.21903520208605,
-      "grad_norm": 1.590876579284668,
-      "learning_rate": 5.7699497487437196e-05,
-      "loss": 5.1619,
-      "step": 42605
-    },
-    {
-      "epoch": 22.21955671447197,
-      "grad_norm": 1.6168969869613647,
-      "learning_rate": 5.769849246231156e-05,
-      "loss": 5.2726,
-      "step": 42606
-    },
-    {
-      "epoch": 22.220078226857886,
-      "grad_norm": 1.643803358078003,
-      "learning_rate": 5.769748743718594e-05,
-      "loss": 5.1866,
-      "step": 42607
-    },
-    {
-      "epoch": 22.220599739243806,
-      "grad_norm": 1.4727168083190918,
-      "learning_rate": 5.76964824120603e-05,
-      "loss": 5.0001,
-      "step": 42608
-    },
-    {
-      "epoch": 22.221121251629725,
-      "grad_norm": 1.6031196117401123,
-      "learning_rate": 5.769547738693467e-05,
-      "loss": 5.1832,
-      "step": 42609
-    },
-    {
-      "epoch": 22.221642764015645,
-      "grad_norm": 1.6368236541748047,
-      "learning_rate": 5.769447236180905e-05,
-      "loss": 5.5941,
-      "step": 42610
-    },
-    {
-      "epoch": 22.222164276401564,
-      "grad_norm": 1.5174440145492554,
-      "learning_rate": 5.7693467336683415e-05,
-      "loss": 5.2497,
-      "step": 42611
-    },
-    {
-      "epoch": 22.222685788787484,
-      "grad_norm": 1.575860857963562,
-      "learning_rate": 5.769246231155779e-05,
-      "loss": 5.3528,
-      "step": 42612
-    },
-    {
-      "epoch": 22.223207301173403,
-      "grad_norm": 1.6122685670852661,
-      "learning_rate": 5.769145728643216e-05,
-      "loss": 4.8929,
-      "step": 42613
-    },
-    {
-      "epoch": 22.223728813559323,
-      "grad_norm": 1.5797749757766724,
-      "learning_rate": 5.7690452261306535e-05,
-      "loss": 5.2694,
-      "step": 42614
-    },
-    {
-      "epoch": 22.224250325945242,
-      "grad_norm": 1.57999587059021,
-      "learning_rate": 5.7689447236180906e-05,
-      "loss": 5.6052,
-      "step": 42615
-    },
-    {
-      "epoch": 22.22477183833116,
-      "grad_norm": 1.6453219652175903,
-      "learning_rate": 5.7688442211055284e-05,
-      "loss": 5.2249,
-      "step": 42616
-    },
-    {
-      "epoch": 22.22529335071708,
-      "grad_norm": 1.5031291246414185,
-      "learning_rate": 5.768743718592965e-05,
-      "loss": 5.2297,
-      "step": 42617
-    },
-    {
-      "epoch": 22.225814863103,
-      "grad_norm": 1.5103729963302612,
-      "learning_rate": 5.7686432160804026e-05,
-      "loss": 5.1614,
-      "step": 42618
-    },
-    {
-      "epoch": 22.226336375488916,
-      "grad_norm": 1.5939828157424927,
-      "learning_rate": 5.76854271356784e-05,
-      "loss": 5.4588,
-      "step": 42619
-    },
-    {
-      "epoch": 22.226857887874836,
-      "grad_norm": 1.5063965320587158,
-      "learning_rate": 5.7684422110552775e-05,
-      "loss": 5.1008,
-      "step": 42620
-    },
-    {
-      "epoch": 22.227379400260755,
-      "grad_norm": 1.5316174030303955,
-      "learning_rate": 5.768341708542714e-05,
-      "loss": 5.2719,
-      "step": 42621
-    },
-    {
-      "epoch": 22.227900912646675,
-      "grad_norm": 1.5397471189498901,
-      "learning_rate": 5.7682412060301503e-05,
-      "loss": 4.8511,
-      "step": 42622
-    },
-    {
-      "epoch": 22.228422425032594,
-      "grad_norm": 1.4139792919158936,
-      "learning_rate": 5.768140703517588e-05,
-      "loss": 5.6913,
-      "step": 42623
-    },
-    {
-      "epoch": 22.228943937418514,
-      "grad_norm": 1.5370490550994873,
-      "learning_rate": 5.768040201005025e-05,
-      "loss": 5.4794,
-      "step": 42624
-    },
-    {
-      "epoch": 22.229465449804433,
-      "grad_norm": 1.4744714498519897,
-      "learning_rate": 5.767939698492463e-05,
-      "loss": 5.7551,
-      "step": 42625
-    },
-    {
-      "epoch": 22.229986962190353,
-      "grad_norm": 1.560343623161316,
-      "learning_rate": 5.7678391959798994e-05,
-      "loss": 5.1483,
-      "step": 42626
-    },
-    {
-      "epoch": 22.230508474576272,
-      "grad_norm": 1.6559511423110962,
-      "learning_rate": 5.767738693467337e-05,
-      "loss": 4.7718,
-      "step": 42627
-    },
-    {
-      "epoch": 22.23102998696219,
-      "grad_norm": 1.5244826078414917,
-      "learning_rate": 5.7676381909547736e-05,
-      "loss": 5.1453,
-      "step": 42628
-    },
-    {
-      "epoch": 22.23155149934811,
-      "grad_norm": 1.5855116844177246,
-      "learning_rate": 5.7675376884422114e-05,
-      "loss": 5.5052,
-      "step": 42629
-    },
-    {
-      "epoch": 22.23207301173403,
-      "grad_norm": 1.6794129610061646,
-      "learning_rate": 5.7674371859296485e-05,
-      "loss": 5.1415,
-      "step": 42630
-    },
-    {
-      "epoch": 22.232594524119946,
-      "grad_norm": 1.4752428531646729,
-      "learning_rate": 5.767336683417086e-05,
-      "loss": 5.4253,
-      "step": 42631
-    },
-    {
-      "epoch": 22.233116036505866,
-      "grad_norm": 1.6020498275756836,
-      "learning_rate": 5.767236180904523e-05,
-      "loss": 4.6551,
-      "step": 42632
-    },
-    {
-      "epoch": 22.233637548891785,
-      "grad_norm": 1.5850929021835327,
-      "learning_rate": 5.7671356783919605e-05,
-      "loss": 5.1329,
-      "step": 42633
-    },
-    {
-      "epoch": 22.234159061277705,
-      "grad_norm": 1.4890735149383545,
-      "learning_rate": 5.767035175879397e-05,
-      "loss": 5.2005,
-      "step": 42634
-    },
-    {
-      "epoch": 22.234680573663624,
-      "grad_norm": 1.5829156637191772,
-      "learning_rate": 5.766934673366834e-05,
-      "loss": 5.3421,
-      "step": 42635
-    },
-    {
-      "epoch": 22.235202086049544,
-      "grad_norm": 1.5378198623657227,
-      "learning_rate": 5.766834170854272e-05,
-      "loss": 5.2257,
-      "step": 42636
-    },
-    {
-      "epoch": 22.235723598435463,
-      "grad_norm": 1.5660229921340942,
-      "learning_rate": 5.766733668341708e-05,
-      "loss": 4.8735,
-      "step": 42637
-    },
-    {
-      "epoch": 22.236245110821383,
-      "grad_norm": 1.567365050315857,
-      "learning_rate": 5.766633165829146e-05,
-      "loss": 5.6089,
-      "step": 42638
-    },
-    {
-      "epoch": 22.236766623207302,
-      "grad_norm": 1.5342856645584106,
-      "learning_rate": 5.7665326633165825e-05,
-      "loss": 5.2709,
-      "step": 42639
-    },
-    {
-      "epoch": 22.23728813559322,
-      "grad_norm": 1.562403917312622,
-      "learning_rate": 5.76643216080402e-05,
-      "loss": 5.3583,
-      "step": 42640
-    },
-    {
-      "epoch": 22.23780964797914,
-      "grad_norm": 1.5080257654190063,
-      "learning_rate": 5.7663316582914574e-05,
-      "loss": 5.4751,
-      "step": 42641
-    },
-    {
-      "epoch": 22.23833116036506,
-      "grad_norm": 1.5445575714111328,
-      "learning_rate": 5.766231155778895e-05,
-      "loss": 5.1796,
-      "step": 42642
-    },
-    {
-      "epoch": 22.238852672750976,
-      "grad_norm": 1.5741121768951416,
-      "learning_rate": 5.7661306532663316e-05,
-      "loss": 5.0369,
-      "step": 42643
-    },
-    {
-      "epoch": 22.239374185136896,
-      "grad_norm": 1.522300362586975,
-      "learning_rate": 5.7660301507537694e-05,
-      "loss": 4.8851,
-      "step": 42644
-    },
-    {
-      "epoch": 22.239895697522815,
-      "grad_norm": 1.6120563745498657,
-      "learning_rate": 5.7659296482412065e-05,
-      "loss": 5.1793,
-      "step": 42645
-    },
-    {
-      "epoch": 22.240417209908735,
-      "grad_norm": 1.6236398220062256,
-      "learning_rate": 5.765829145728644e-05,
-      "loss": 5.3586,
-      "step": 42646
-    },
-    {
-      "epoch": 22.240938722294654,
-      "grad_norm": 1.5386542081832886,
-      "learning_rate": 5.765728643216081e-05,
-      "loss": 5.2244,
-      "step": 42647
-    },
-    {
-      "epoch": 22.241460234680574,
-      "grad_norm": 1.5857760906219482,
-      "learning_rate": 5.765628140703517e-05,
-      "loss": 4.3847,
-      "step": 42648
-    },
-    {
-      "epoch": 22.241981747066493,
-      "grad_norm": 1.465994119644165,
-      "learning_rate": 5.765527638190955e-05,
-      "loss": 5.6288,
-      "step": 42649
-    },
-    {
-      "epoch": 22.242503259452413,
-      "grad_norm": 1.6312357187271118,
-      "learning_rate": 5.765427135678392e-05,
-      "loss": 5.0444,
-      "step": 42650
-    },
-    {
-      "epoch": 22.243024771838332,
-      "grad_norm": 1.6125133037567139,
-      "learning_rate": 5.76532663316583e-05,
-      "loss": 5.4235,
-      "step": 42651
-    },
-    {
-      "epoch": 22.24354628422425,
-      "grad_norm": 1.491666316986084,
-      "learning_rate": 5.765226130653266e-05,
-      "loss": 5.6472,
-      "step": 42652
-    },
-    {
-      "epoch": 22.24406779661017,
-      "grad_norm": 1.4415994882583618,
-      "learning_rate": 5.765125628140704e-05,
-      "loss": 5.4673,
-      "step": 42653
-    },
-    {
-      "epoch": 22.24458930899609,
-      "grad_norm": 1.5354139804840088,
-      "learning_rate": 5.7650251256281404e-05,
-      "loss": 5.0521,
-      "step": 42654
-    },
-    {
-      "epoch": 22.245110821382006,
-      "grad_norm": 1.6082944869995117,
-      "learning_rate": 5.764924623115578e-05,
-      "loss": 5.0043,
-      "step": 42655
-    },
-    {
-      "epoch": 22.245632333767926,
-      "grad_norm": 1.551696538925171,
-      "learning_rate": 5.764824120603015e-05,
-      "loss": 5.1367,
-      "step": 42656
-    },
-    {
-      "epoch": 22.246153846153845,
-      "grad_norm": 1.6853907108306885,
-      "learning_rate": 5.764723618090453e-05,
-      "loss": 4.8054,
-      "step": 42657
-    },
-    {
-      "epoch": 22.246675358539765,
-      "grad_norm": 1.5112788677215576,
-      "learning_rate": 5.7646231155778895e-05,
-      "loss": 5.517,
-      "step": 42658
-    },
-    {
-      "epoch": 22.247196870925684,
-      "grad_norm": 1.5705713033676147,
-      "learning_rate": 5.764522613065327e-05,
-      "loss": 5.3934,
-      "step": 42659
-    },
-    {
-      "epoch": 22.247718383311604,
-      "grad_norm": 1.591688632965088,
-      "learning_rate": 5.764422110552764e-05,
-      "loss": 5.2686,
-      "step": 42660
-    },
-    {
-      "epoch": 22.248239895697523,
-      "grad_norm": 1.5254555940628052,
-      "learning_rate": 5.7643216080402015e-05,
-      "loss": 5.2984,
-      "step": 42661
-    },
-    {
-      "epoch": 22.248761408083443,
-      "grad_norm": 1.5278242826461792,
-      "learning_rate": 5.7642211055276386e-05,
-      "loss": 5.0187,
-      "step": 42662
-    },
-    {
-      "epoch": 22.249282920469362,
-      "grad_norm": 1.6001027822494507,
-      "learning_rate": 5.764120603015075e-05,
-      "loss": 5.2682,
-      "step": 42663
-    },
-    {
-      "epoch": 22.24980443285528,
-      "grad_norm": 1.5511822700500488,
-      "learning_rate": 5.764020100502513e-05,
-      "loss": 5.4129,
-      "step": 42664
-    },
-    {
-      "epoch": 22.2503259452412,
-      "grad_norm": 1.6065582036972046,
-      "learning_rate": 5.76391959798995e-05,
-      "loss": 5.0766,
-      "step": 42665
-    },
-    {
-      "epoch": 22.250847457627117,
-      "grad_norm": 1.474036693572998,
-      "learning_rate": 5.763819095477388e-05,
-      "loss": 5.4186,
-      "step": 42666
-    },
-    {
-      "epoch": 22.251368970013036,
-      "grad_norm": 1.562574863433838,
-      "learning_rate": 5.763718592964824e-05,
-      "loss": 5.2641,
-      "step": 42667
-    },
-    {
-      "epoch": 22.251890482398956,
-      "grad_norm": 1.5029574632644653,
-      "learning_rate": 5.763618090452262e-05,
-      "loss": 5.4474,
-      "step": 42668
-    },
-    {
-      "epoch": 22.252411994784875,
-      "grad_norm": 1.4710605144500732,
-      "learning_rate": 5.7635175879396983e-05,
-      "loss": 5.5298,
-      "step": 42669
-    },
-    {
-      "epoch": 22.252933507170795,
-      "grad_norm": 1.6180781126022339,
-      "learning_rate": 5.763417085427136e-05,
-      "loss": 5.1952,
-      "step": 42670
-    },
-    {
-      "epoch": 22.253455019556714,
-      "grad_norm": 1.5945063829421997,
-      "learning_rate": 5.763316582914573e-05,
-      "loss": 4.74,
-      "step": 42671
-    },
-    {
-      "epoch": 22.253976531942634,
-      "grad_norm": 1.4723384380340576,
-      "learning_rate": 5.763216080402011e-05,
-      "loss": 5.239,
-      "step": 42672
-    },
-    {
-      "epoch": 22.254498044328553,
-      "grad_norm": 1.5497794151306152,
-      "learning_rate": 5.7631155778894474e-05,
-      "loss": 4.9599,
-      "step": 42673
-    },
-    {
-      "epoch": 22.255019556714473,
-      "grad_norm": 1.659677505493164,
-      "learning_rate": 5.763015075376885e-05,
-      "loss": 4.9014,
-      "step": 42674
-    },
-    {
-      "epoch": 22.255541069100392,
-      "grad_norm": 1.5408847332000732,
-      "learning_rate": 5.7629145728643217e-05,
-      "loss": 5.1412,
-      "step": 42675
-    },
-    {
-      "epoch": 22.25606258148631,
-      "grad_norm": 1.638092041015625,
-      "learning_rate": 5.762814070351759e-05,
-      "loss": 4.9378,
-      "step": 42676
-    },
-    {
-      "epoch": 22.25658409387223,
-      "grad_norm": 1.5986945629119873,
-      "learning_rate": 5.7627135678391965e-05,
-      "loss": 5.2219,
-      "step": 42677
-    },
-    {
-      "epoch": 22.257105606258147,
-      "grad_norm": 1.5849286317825317,
-      "learning_rate": 5.762613065326633e-05,
-      "loss": 4.862,
-      "step": 42678
-    },
-    {
-      "epoch": 22.257627118644066,
-      "grad_norm": 1.5108551979064941,
-      "learning_rate": 5.762512562814071e-05,
-      "loss": 4.6383,
-      "step": 42679
-    },
-    {
-      "epoch": 22.258148631029986,
-      "grad_norm": 1.5310474634170532,
-      "learning_rate": 5.762412060301507e-05,
-      "loss": 5.1257,
-      "step": 42680
-    },
-    {
-      "epoch": 22.258670143415905,
-      "grad_norm": 1.5359526872634888,
-      "learning_rate": 5.762311557788945e-05,
-      "loss": 5.2523,
-      "step": 42681
-    },
-    {
-      "epoch": 22.259191655801825,
-      "grad_norm": 1.6780747175216675,
-      "learning_rate": 5.762211055276382e-05,
-      "loss": 4.8469,
-      "step": 42682
-    },
-    {
-      "epoch": 22.259713168187744,
-      "grad_norm": 1.5906333923339844,
-      "learning_rate": 5.76211055276382e-05,
-      "loss": 5.0673,
-      "step": 42683
-    },
-    {
-      "epoch": 22.260234680573664,
-      "grad_norm": 1.6442816257476807,
-      "learning_rate": 5.762010050251256e-05,
-      "loss": 4.4878,
-      "step": 42684
-    },
-    {
-      "epoch": 22.260756192959583,
-      "grad_norm": 1.498566746711731,
-      "learning_rate": 5.761909547738694e-05,
-      "loss": 5.1336,
-      "step": 42685
-    },
-    {
-      "epoch": 22.261277705345503,
-      "grad_norm": 1.5446151494979858,
-      "learning_rate": 5.761809045226131e-05,
-      "loss": 4.9948,
-      "step": 42686
-    },
-    {
-      "epoch": 22.261799217731422,
-      "grad_norm": 1.5894014835357666,
-      "learning_rate": 5.761708542713569e-05,
-      "loss": 4.9815,
-      "step": 42687
-    },
-    {
-      "epoch": 22.26232073011734,
-      "grad_norm": 1.5603742599487305,
-      "learning_rate": 5.7616080402010054e-05,
-      "loss": 5.0045,
-      "step": 42688
-    },
-    {
-      "epoch": 22.26284224250326,
-      "grad_norm": 1.611146092414856,
-      "learning_rate": 5.761507537688442e-05,
-      "loss": 4.7783,
-      "step": 42689
-    },
-    {
-      "epoch": 22.263363754889177,
-      "grad_norm": 1.7263976335525513,
-      "learning_rate": 5.7614070351758796e-05,
-      "loss": 4.9833,
-      "step": 42690
-    },
-    {
-      "epoch": 22.263885267275096,
-      "grad_norm": 1.6046302318572998,
-      "learning_rate": 5.761306532663317e-05,
-      "loss": 4.9028,
-      "step": 42691
-    },
-    {
-      "epoch": 22.264406779661016,
-      "grad_norm": 1.5123279094696045,
-      "learning_rate": 5.7612060301507545e-05,
-      "loss": 5.3621,
-      "step": 42692
-    },
-    {
-      "epoch": 22.264928292046935,
-      "grad_norm": 1.5057975053787231,
-      "learning_rate": 5.761105527638191e-05,
-      "loss": 5.5031,
-      "step": 42693
-    },
-    {
-      "epoch": 22.265449804432855,
-      "grad_norm": 1.5535962581634521,
-      "learning_rate": 5.761005025125629e-05,
-      "loss": 5.0217,
-      "step": 42694
-    },
-    {
-      "epoch": 22.265971316818774,
-      "grad_norm": 1.497111439704895,
-      "learning_rate": 5.760904522613065e-05,
-      "loss": 5.2261,
-      "step": 42695
-    },
-    {
-      "epoch": 22.266492829204694,
-      "grad_norm": 1.4009497165679932,
-      "learning_rate": 5.760804020100503e-05,
-      "loss": 5.7061,
-      "step": 42696
-    },
-    {
-      "epoch": 22.267014341590613,
-      "grad_norm": 1.5643246173858643,
-      "learning_rate": 5.76070351758794e-05,
-      "loss": 5.2161,
-      "step": 42697
-    },
-    {
-      "epoch": 22.267535853976533,
-      "grad_norm": 1.478999137878418,
-      "learning_rate": 5.760603015075378e-05,
-      "loss": 5.6338,
-      "step": 42698
-    },
-    {
-      "epoch": 22.268057366362452,
-      "grad_norm": 1.4726502895355225,
-      "learning_rate": 5.760502512562814e-05,
-      "loss": 5.2562,
-      "step": 42699
-    },
-    {
-      "epoch": 22.26857887874837,
-      "grad_norm": 1.5999311208724976,
-      "learning_rate": 5.760402010050252e-05,
-      "loss": 5.2911,
-      "step": 42700
-    },
-    {
-      "epoch": 22.26910039113429,
-      "grad_norm": 1.487545132637024,
-      "learning_rate": 5.7603015075376884e-05,
-      "loss": 5.6426,
-      "step": 42701
-    },
-    {
-      "epoch": 22.269621903520207,
-      "grad_norm": 1.4669175148010254,
-      "learning_rate": 5.7602010050251255e-05,
-      "loss": 5.429,
-      "step": 42702
-    },
-    {
-      "epoch": 22.270143415906126,
-      "grad_norm": 1.6137542724609375,
-      "learning_rate": 5.760100502512563e-05,
-      "loss": 4.8515,
-      "step": 42703
-    },
-    {
-      "epoch": 22.270664928292046,
-      "grad_norm": 1.616709589958191,
-      "learning_rate": 5.76e-05,
-      "loss": 4.9934,
-      "step": 42704
-    },
-    {
-      "epoch": 22.271186440677965,
-      "grad_norm": 1.5681886672973633,
-      "learning_rate": 5.7598994974874375e-05,
-      "loss": 5.2474,
-      "step": 42705
-    },
-    {
-      "epoch": 22.271707953063885,
-      "grad_norm": 1.469547986984253,
-      "learning_rate": 5.7597989949748746e-05,
-      "loss": 5.6365,
-      "step": 42706
-    },
-    {
-      "epoch": 22.272229465449804,
-      "grad_norm": 1.513516902923584,
-      "learning_rate": 5.7596984924623124e-05,
-      "loss": 5.1288,
-      "step": 42707
-    },
-    {
-      "epoch": 22.272750977835724,
-      "grad_norm": 1.6259602308273315,
-      "learning_rate": 5.759597989949749e-05,
-      "loss": 5.0155,
-      "step": 42708
-    },
-    {
-      "epoch": 22.273272490221643,
-      "grad_norm": 1.5893603563308716,
-      "learning_rate": 5.7594974874371866e-05,
-      "loss": 5.6015,
-      "step": 42709
-    },
-    {
-      "epoch": 22.273794002607563,
-      "grad_norm": 1.6439648866653442,
-      "learning_rate": 5.759396984924623e-05,
-      "loss": 5.1084,
-      "step": 42710
-    },
-    {
-      "epoch": 22.274315514993482,
-      "grad_norm": 1.5470153093338013,
-      "learning_rate": 5.759296482412061e-05,
-      "loss": 5.6569,
-      "step": 42711
-    },
-    {
-      "epoch": 22.2748370273794,
-      "grad_norm": 1.5094358921051025,
-      "learning_rate": 5.759195979899498e-05,
-      "loss": 5.4451,
-      "step": 42712
-    },
-    {
-      "epoch": 22.27535853976532,
-      "grad_norm": 1.486604928970337,
-      "learning_rate": 5.759095477386936e-05,
-      "loss": 5.7262,
-      "step": 42713
-    },
-    {
-      "epoch": 22.275880052151237,
-      "grad_norm": 1.4622840881347656,
-      "learning_rate": 5.758994974874372e-05,
-      "loss": 5.562,
-      "step": 42714
-    },
-    {
-      "epoch": 22.276401564537156,
-      "grad_norm": 1.4377809762954712,
-      "learning_rate": 5.7588944723618086e-05,
-      "loss": 5.4962,
-      "step": 42715
-    },
-    {
-      "epoch": 22.276923076923076,
-      "grad_norm": 1.5447652339935303,
-      "learning_rate": 5.7587939698492464e-05,
-      "loss": 5.1716,
-      "step": 42716
-    },
-    {
-      "epoch": 22.277444589308995,
-      "grad_norm": 1.5258245468139648,
-      "learning_rate": 5.7586934673366835e-05,
-      "loss": 5.4097,
-      "step": 42717
-    },
-    {
-      "epoch": 22.277966101694915,
-      "grad_norm": 1.539490818977356,
-      "learning_rate": 5.758592964824121e-05,
-      "loss": 5.4347,
-      "step": 42718
-    },
-    {
-      "epoch": 22.278487614080834,
-      "grad_norm": 1.4843858480453491,
-      "learning_rate": 5.758492462311558e-05,
-      "loss": 5.5729,
-      "step": 42719
-    },
-    {
-      "epoch": 22.279009126466754,
-      "grad_norm": 1.6736210584640503,
-      "learning_rate": 5.7583919597989955e-05,
-      "loss": 5.3521,
-      "step": 42720
-    },
-    {
-      "epoch": 22.279530638852673,
-      "grad_norm": 1.5727450847625732,
-      "learning_rate": 5.758291457286432e-05,
-      "loss": 5.2599,
-      "step": 42721
-    },
-    {
-      "epoch": 22.280052151238593,
-      "grad_norm": 1.6118577718734741,
-      "learning_rate": 5.75819095477387e-05,
-      "loss": 5.2005,
-      "step": 42722
-    },
-    {
-      "epoch": 22.280573663624512,
-      "grad_norm": 1.5826573371887207,
-      "learning_rate": 5.758090452261307e-05,
-      "loss": 5.1751,
-      "step": 42723
-    },
-    {
-      "epoch": 22.28109517601043,
-      "grad_norm": 1.535997748374939,
-      "learning_rate": 5.7579899497487446e-05,
-      "loss": 5.227,
-      "step": 42724
-    },
-    {
-      "epoch": 22.28161668839635,
-      "grad_norm": 1.5772911310195923,
-      "learning_rate": 5.757889447236181e-05,
-      "loss": 5.4234,
-      "step": 42725
-    },
-    {
-      "epoch": 22.282138200782267,
-      "grad_norm": 1.594077467918396,
-      "learning_rate": 5.757788944723619e-05,
-      "loss": 5.0254,
-      "step": 42726
-    },
-    {
-      "epoch": 22.282659713168186,
-      "grad_norm": 1.4820584058761597,
-      "learning_rate": 5.757688442211055e-05,
-      "loss": 5.6656,
-      "step": 42727
-    },
-    {
-      "epoch": 22.283181225554106,
-      "grad_norm": 1.6719586849212646,
-      "learning_rate": 5.757587939698492e-05,
-      "loss": 5.105,
-      "step": 42728
-    },
-    {
-      "epoch": 22.283702737940025,
-      "grad_norm": 1.520719289779663,
-      "learning_rate": 5.75748743718593e-05,
-      "loss": 5.7191,
-      "step": 42729
-    },
-    {
-      "epoch": 22.284224250325945,
-      "grad_norm": 1.4987751245498657,
-      "learning_rate": 5.7573869346733665e-05,
-      "loss": 5.6759,
-      "step": 42730
-    },
-    {
-      "epoch": 22.284745762711864,
-      "grad_norm": 1.5564543008804321,
-      "learning_rate": 5.757286432160804e-05,
-      "loss": 5.1286,
-      "step": 42731
-    },
-    {
-      "epoch": 22.285267275097784,
-      "grad_norm": 1.5780175924301147,
-      "learning_rate": 5.7571859296482414e-05,
-      "loss": 5.3732,
-      "step": 42732
-    },
-    {
-      "epoch": 22.285788787483703,
-      "grad_norm": 1.6035034656524658,
-      "learning_rate": 5.757085427135679e-05,
-      "loss": 5.1144,
-      "step": 42733
-    },
-    {
-      "epoch": 22.286310299869623,
-      "grad_norm": 1.526979923248291,
-      "learning_rate": 5.7569849246231156e-05,
-      "loss": 5.4491,
-      "step": 42734
-    },
-    {
-      "epoch": 22.286831812255542,
-      "grad_norm": 1.6305540800094604,
-      "learning_rate": 5.7568844221105534e-05,
-      "loss": 4.5524,
-      "step": 42735
-    },
-    {
-      "epoch": 22.28735332464146,
-      "grad_norm": 1.720597743988037,
-      "learning_rate": 5.75678391959799e-05,
-      "loss": 4.7765,
-      "step": 42736
-    },
-    {
-      "epoch": 22.28787483702738,
-      "grad_norm": 1.5495737791061401,
-      "learning_rate": 5.7566834170854276e-05,
-      "loss": 5.6303,
-      "step": 42737
-    },
-    {
-      "epoch": 22.288396349413297,
-      "grad_norm": 1.6587685346603394,
-      "learning_rate": 5.756582914572865e-05,
-      "loss": 4.6155,
-      "step": 42738
-    },
-    {
-      "epoch": 22.288917861799217,
-      "grad_norm": 1.630443811416626,
-      "learning_rate": 5.7564824120603025e-05,
-      "loss": 4.7045,
-      "step": 42739
-    },
-    {
-      "epoch": 22.289439374185136,
-      "grad_norm": 1.5973398685455322,
-      "learning_rate": 5.756381909547739e-05,
-      "loss": 5.0058,
-      "step": 42740
-    },
-    {
-      "epoch": 22.289960886571055,
-      "grad_norm": 1.4852455854415894,
-      "learning_rate": 5.756281407035177e-05,
-      "loss": 5.102,
-      "step": 42741
-    },
-    {
-      "epoch": 22.290482398956975,
-      "grad_norm": 1.6017577648162842,
-      "learning_rate": 5.756180904522613e-05,
-      "loss": 5.1057,
-      "step": 42742
-    },
-    {
-      "epoch": 22.291003911342894,
-      "grad_norm": 1.5645556449890137,
-      "learning_rate": 5.75608040201005e-05,
-      "loss": 5.118,
-      "step": 42743
-    },
-    {
-      "epoch": 22.291525423728814,
-      "grad_norm": 1.639432668685913,
-      "learning_rate": 5.755979899497488e-05,
-      "loss": 5.255,
-      "step": 42744
-    },
-    {
-      "epoch": 22.292046936114733,
-      "grad_norm": 1.4921025037765503,
-      "learning_rate": 5.7558793969849244e-05,
-      "loss": 4.6801,
-      "step": 42745
-    },
-    {
-      "epoch": 22.292568448500653,
-      "grad_norm": 1.5907570123672485,
-      "learning_rate": 5.755778894472362e-05,
-      "loss": 5.1928,
-      "step": 42746
-    },
-    {
-      "epoch": 22.293089960886572,
-      "grad_norm": 1.5050814151763916,
-      "learning_rate": 5.7556783919597986e-05,
-      "loss": 5.2712,
-      "step": 42747
-    },
-    {
-      "epoch": 22.29361147327249,
-      "grad_norm": 1.6353517770767212,
-      "learning_rate": 5.7555778894472364e-05,
-      "loss": 4.9787,
-      "step": 42748
-    },
-    {
-      "epoch": 22.294132985658408,
-      "grad_norm": 1.5008080005645752,
-      "learning_rate": 5.7554773869346735e-05,
-      "loss": 4.942,
-      "step": 42749
-    },
-    {
-      "epoch": 22.294654498044327,
-      "grad_norm": 1.474360466003418,
-      "learning_rate": 5.755376884422111e-05,
-      "loss": 5.0111,
-      "step": 42750
-    },
-    {
-      "epoch": 22.295176010430247,
-      "grad_norm": 1.6336030960083008,
-      "learning_rate": 5.755276381909548e-05,
-      "loss": 4.8312,
-      "step": 42751
-    },
-    {
-      "epoch": 22.295697522816166,
-      "grad_norm": 1.5718523263931274,
-      "learning_rate": 5.7551758793969855e-05,
-      "loss": 5.3217,
-      "step": 42752
-    },
-    {
-      "epoch": 22.296219035202085,
-      "grad_norm": 1.4818423986434937,
-      "learning_rate": 5.7550753768844226e-05,
-      "loss": 4.8477,
-      "step": 42753
-    },
-    {
-      "epoch": 22.296740547588005,
-      "grad_norm": 1.5473694801330566,
-      "learning_rate": 5.7549748743718604e-05,
-      "loss": 5.1857,
-      "step": 42754
-    },
-    {
-      "epoch": 22.297262059973924,
-      "grad_norm": 1.4837837219238281,
-      "learning_rate": 5.754874371859297e-05,
-      "loss": 5.1828,
-      "step": 42755
-    },
-    {
-      "epoch": 22.297783572359844,
-      "grad_norm": 1.4898433685302734,
-      "learning_rate": 5.754773869346733e-05,
-      "loss": 4.9015,
-      "step": 42756
-    },
-    {
-      "epoch": 22.298305084745763,
-      "grad_norm": 1.5580596923828125,
-      "learning_rate": 5.754673366834171e-05,
-      "loss": 5.2694,
-      "step": 42757
-    },
-    {
-      "epoch": 22.298826597131683,
-      "grad_norm": 1.4999011754989624,
-      "learning_rate": 5.754572864321608e-05,
-      "loss": 5.658,
-      "step": 42758
-    },
-    {
-      "epoch": 22.299348109517602,
-      "grad_norm": 1.586396336555481,
-      "learning_rate": 5.754472361809046e-05,
-      "loss": 5.2743,
-      "step": 42759
-    },
-    {
-      "epoch": 22.29986962190352,
-      "grad_norm": 1.470942735671997,
-      "learning_rate": 5.7543718592964824e-05,
-      "loss": 5.637,
-      "step": 42760
-    },
-    {
-      "epoch": 22.300391134289438,
-      "grad_norm": 1.5429136753082275,
-      "learning_rate": 5.75427135678392e-05,
-      "loss": 5.474,
-      "step": 42761
-    },
-    {
-      "epoch": 22.300912646675357,
-      "grad_norm": 1.4912092685699463,
-      "learning_rate": 5.7541708542713566e-05,
-      "loss": 5.4072,
-      "step": 42762
-    },
-    {
-      "epoch": 22.301434159061277,
-      "grad_norm": 1.5132808685302734,
-      "learning_rate": 5.7540703517587944e-05,
-      "loss": 5.3368,
-      "step": 42763
-    },
-    {
-      "epoch": 22.301955671447196,
-      "grad_norm": 1.5622069835662842,
-      "learning_rate": 5.7539698492462315e-05,
-      "loss": 5.487,
-      "step": 42764
-    },
-    {
-      "epoch": 22.302477183833116,
-      "grad_norm": 1.437434434890747,
-      "learning_rate": 5.753869346733669e-05,
-      "loss": 5.7006,
-      "step": 42765
-    },
-    {
-      "epoch": 22.302998696219035,
-      "grad_norm": 1.803671956062317,
-      "learning_rate": 5.753768844221106e-05,
-      "loss": 4.1344,
-      "step": 42766
-    },
-    {
-      "epoch": 22.303520208604954,
-      "grad_norm": 1.652134895324707,
-      "learning_rate": 5.7536683417085435e-05,
-      "loss": 5.193,
-      "step": 42767
-    },
-    {
-      "epoch": 22.304041720990874,
-      "grad_norm": 1.446913242340088,
-      "learning_rate": 5.75356783919598e-05,
-      "loss": 5.5676,
-      "step": 42768
-    },
-    {
-      "epoch": 22.304563233376793,
-      "grad_norm": 1.8638867139816284,
-      "learning_rate": 5.753467336683417e-05,
-      "loss": 5.3321,
-      "step": 42769
-    },
-    {
-      "epoch": 22.305084745762713,
-      "grad_norm": 1.418784499168396,
-      "learning_rate": 5.753366834170855e-05,
-      "loss": 5.5671,
-      "step": 42770
-    },
-    {
-      "epoch": 22.305606258148632,
-      "grad_norm": 1.6472268104553223,
-      "learning_rate": 5.753266331658291e-05,
-      "loss": 5.1672,
-      "step": 42771
-    },
-    {
-      "epoch": 22.30612777053455,
-      "grad_norm": 1.4837148189544678,
-      "learning_rate": 5.753165829145729e-05,
-      "loss": 5.6534,
-      "step": 42772
-    },
-    {
-      "epoch": 22.306649282920468,
-      "grad_norm": 1.597639560699463,
-      "learning_rate": 5.753065326633166e-05,
-      "loss": 4.6664,
-      "step": 42773
-    },
-    {
-      "epoch": 22.307170795306387,
-      "grad_norm": 1.5298717021942139,
-      "learning_rate": 5.752964824120604e-05,
-      "loss": 5.1607,
-      "step": 42774
-    },
-    {
-      "epoch": 22.307692307692307,
-      "grad_norm": 1.553389310836792,
-      "learning_rate": 5.75286432160804e-05,
-      "loss": 4.8918,
-      "step": 42775
-    },
-    {
-      "epoch": 22.308213820078226,
-      "grad_norm": 1.7909811735153198,
-      "learning_rate": 5.752763819095478e-05,
-      "loss": 4.8344,
-      "step": 42776
-    },
-    {
-      "epoch": 22.308735332464146,
-      "grad_norm": 1.8073606491088867,
-      "learning_rate": 5.7526633165829145e-05,
-      "loss": 4.7238,
-      "step": 42777
-    },
-    {
-      "epoch": 22.309256844850065,
-      "grad_norm": 1.4976232051849365,
-      "learning_rate": 5.752562814070352e-05,
-      "loss": 5.3658,
-      "step": 42778
-    },
-    {
-      "epoch": 22.309778357235984,
-      "grad_norm": 1.501645803451538,
-      "learning_rate": 5.7524623115577894e-05,
-      "loss": 5.6423,
-      "step": 42779
-    },
-    {
-      "epoch": 22.310299869621904,
-      "grad_norm": 1.4912315607070923,
-      "learning_rate": 5.752361809045227e-05,
-      "loss": 5.471,
-      "step": 42780
-    },
-    {
-      "epoch": 22.310821382007823,
-      "grad_norm": 1.384901762008667,
-      "learning_rate": 5.7522613065326636e-05,
-      "loss": 5.7815,
-      "step": 42781
-    },
-    {
-      "epoch": 22.311342894393743,
-      "grad_norm": 1.4566527605056763,
-      "learning_rate": 5.7521608040201e-05,
-      "loss": 5.3961,
-      "step": 42782
-    },
-    {
-      "epoch": 22.311864406779662,
-      "grad_norm": 1.528397798538208,
-      "learning_rate": 5.752060301507538e-05,
-      "loss": 5.2963,
-      "step": 42783
-    },
-    {
-      "epoch": 22.312385919165582,
-      "grad_norm": 1.603379726409912,
-      "learning_rate": 5.751959798994975e-05,
-      "loss": 5.0302,
-      "step": 42784
-    },
-    {
-      "epoch": 22.312907431551498,
-      "grad_norm": 1.5466679334640503,
-      "learning_rate": 5.751859296482413e-05,
-      "loss": 5.2818,
-      "step": 42785
-    },
-    {
-      "epoch": 22.313428943937417,
-      "grad_norm": 1.4719234704971313,
-      "learning_rate": 5.751758793969849e-05,
-      "loss": 5.57,
-      "step": 42786
-    },
-    {
-      "epoch": 22.313950456323337,
-      "grad_norm": 1.5690820217132568,
-      "learning_rate": 5.751658291457287e-05,
-      "loss": 4.9966,
-      "step": 42787
-    },
-    {
-      "epoch": 22.314471968709256,
-      "grad_norm": 1.5730342864990234,
-      "learning_rate": 5.7515577889447233e-05,
-      "loss": 5.2985,
-      "step": 42788
-    },
-    {
-      "epoch": 22.314993481095176,
-      "grad_norm": 1.6106051206588745,
-      "learning_rate": 5.751457286432161e-05,
-      "loss": 5.0798,
-      "step": 42789
-    },
-    {
-      "epoch": 22.315514993481095,
-      "grad_norm": 1.557536005973816,
-      "learning_rate": 5.751356783919598e-05,
-      "loss": 4.9497,
-      "step": 42790
-    },
-    {
-      "epoch": 22.316036505867014,
-      "grad_norm": 1.4732604026794434,
-      "learning_rate": 5.751256281407036e-05,
-      "loss": 5.569,
-      "step": 42791
-    },
-    {
-      "epoch": 22.316558018252934,
-      "grad_norm": 1.564284086227417,
-      "learning_rate": 5.7511557788944724e-05,
-      "loss": 5.6254,
-      "step": 42792
-    },
-    {
-      "epoch": 22.317079530638853,
-      "grad_norm": 1.4350491762161255,
-      "learning_rate": 5.75105527638191e-05,
-      "loss": 5.6616,
-      "step": 42793
-    },
-    {
-      "epoch": 22.317601043024773,
-      "grad_norm": 1.528190016746521,
-      "learning_rate": 5.750954773869347e-05,
-      "loss": 5.2763,
-      "step": 42794
-    },
-    {
-      "epoch": 22.318122555410692,
-      "grad_norm": 1.4811400175094604,
-      "learning_rate": 5.750854271356784e-05,
-      "loss": 5.3562,
-      "step": 42795
-    },
-    {
-      "epoch": 22.318644067796612,
-      "grad_norm": 1.5091955661773682,
-      "learning_rate": 5.7507537688442215e-05,
-      "loss": 5.4477,
-      "step": 42796
-    },
-    {
-      "epoch": 22.319165580182528,
-      "grad_norm": 1.545815110206604,
-      "learning_rate": 5.750653266331658e-05,
-      "loss": 5.258,
-      "step": 42797
-    },
-    {
-      "epoch": 22.319687092568447,
-      "grad_norm": 1.5791497230529785,
-      "learning_rate": 5.750552763819096e-05,
-      "loss": 5.1381,
-      "step": 42798
-    },
-    {
-      "epoch": 22.320208604954367,
-      "grad_norm": 1.5134024620056152,
-      "learning_rate": 5.750452261306533e-05,
-      "loss": 5.3597,
-      "step": 42799
-    },
-    {
-      "epoch": 22.320730117340286,
-      "grad_norm": 1.4704854488372803,
-      "learning_rate": 5.7503517587939706e-05,
-      "loss": 5.4095,
-      "step": 42800
-    },
-    {
-      "epoch": 22.321251629726206,
-      "grad_norm": 1.592573881149292,
-      "learning_rate": 5.750251256281407e-05,
-      "loss": 4.9296,
-      "step": 42801
-    },
-    {
-      "epoch": 22.321773142112125,
-      "grad_norm": 1.4801677465438843,
-      "learning_rate": 5.750150753768845e-05,
-      "loss": 5.5423,
-      "step": 42802
-    },
-    {
-      "epoch": 22.322294654498045,
-      "grad_norm": 1.5843042135238647,
-      "learning_rate": 5.750050251256281e-05,
-      "loss": 5.2979,
-      "step": 42803
-    },
-    {
-      "epoch": 22.322816166883964,
-      "grad_norm": 1.4889599084854126,
-      "learning_rate": 5.749949748743719e-05,
-      "loss": 4.9618,
-      "step": 42804
-    },
-    {
-      "epoch": 22.323337679269883,
-      "grad_norm": 1.6002346277236938,
-      "learning_rate": 5.749849246231156e-05,
-      "loss": 5.064,
-      "step": 42805
-    },
-    {
-      "epoch": 22.323859191655803,
-      "grad_norm": 1.5704405307769775,
-      "learning_rate": 5.749748743718594e-05,
-      "loss": 5.2321,
-      "step": 42806
-    },
-    {
-      "epoch": 22.324380704041722,
-      "grad_norm": 1.6929042339324951,
-      "learning_rate": 5.7496482412060304e-05,
-      "loss": 5.0314,
-      "step": 42807
-    },
-    {
-      "epoch": 22.324902216427642,
-      "grad_norm": 1.650964617729187,
-      "learning_rate": 5.749547738693467e-05,
-      "loss": 4.8591,
-      "step": 42808
-    },
-    {
-      "epoch": 22.325423728813558,
-      "grad_norm": 1.540054202079773,
-      "learning_rate": 5.7494472361809046e-05,
-      "loss": 4.9632,
-      "step": 42809
-    },
-    {
-      "epoch": 22.325945241199477,
-      "grad_norm": 1.5971612930297852,
-      "learning_rate": 5.749346733668342e-05,
-      "loss": 5.1937,
-      "step": 42810
-    },
-    {
-      "epoch": 22.326466753585397,
-      "grad_norm": 1.6140475273132324,
-      "learning_rate": 5.7492462311557795e-05,
-      "loss": 5.2626,
-      "step": 42811
-    },
-    {
-      "epoch": 22.326988265971316,
-      "grad_norm": 1.4521375894546509,
-      "learning_rate": 5.749145728643216e-05,
-      "loss": 5.4056,
-      "step": 42812
-    },
-    {
-      "epoch": 22.327509778357236,
-      "grad_norm": 1.5422048568725586,
-      "learning_rate": 5.749045226130654e-05,
-      "loss": 5.6258,
-      "step": 42813
-    },
-    {
-      "epoch": 22.328031290743155,
-      "grad_norm": 1.5375878810882568,
-      "learning_rate": 5.74894472361809e-05,
-      "loss": 5.2817,
-      "step": 42814
-    },
-    {
-      "epoch": 22.328552803129075,
-      "grad_norm": 1.488816261291504,
-      "learning_rate": 5.748844221105528e-05,
-      "loss": 5.0139,
-      "step": 42815
-    },
-    {
-      "epoch": 22.329074315514994,
-      "grad_norm": 1.6023046970367432,
-      "learning_rate": 5.748743718592965e-05,
-      "loss": 4.9498,
-      "step": 42816
-    },
-    {
-      "epoch": 22.329595827900913,
-      "grad_norm": 1.5314183235168457,
-      "learning_rate": 5.748643216080403e-05,
-      "loss": 5.1151,
-      "step": 42817
-    },
-    {
-      "epoch": 22.330117340286833,
-      "grad_norm": 1.526774287223816,
-      "learning_rate": 5.748542713567839e-05,
-      "loss": 5.2484,
-      "step": 42818
-    },
-    {
-      "epoch": 22.330638852672752,
-      "grad_norm": 1.4831304550170898,
-      "learning_rate": 5.748442211055277e-05,
-      "loss": 5.437,
-      "step": 42819
-    },
-    {
-      "epoch": 22.331160365058672,
-      "grad_norm": 1.6507471799850464,
-      "learning_rate": 5.748341708542714e-05,
-      "loss": 5.0606,
-      "step": 42820
-    },
-    {
-      "epoch": 22.331681877444588,
-      "grad_norm": 1.602585792541504,
-      "learning_rate": 5.7482412060301505e-05,
-      "loss": 5.513,
-      "step": 42821
-    },
-    {
-      "epoch": 22.332203389830507,
-      "grad_norm": 1.504675030708313,
-      "learning_rate": 5.748140703517588e-05,
-      "loss": 5.3852,
-      "step": 42822
-    },
-    {
-      "epoch": 22.332724902216427,
-      "grad_norm": 1.4387164115905762,
-      "learning_rate": 5.748040201005025e-05,
-      "loss": 5.8431,
-      "step": 42823
-    },
-    {
-      "epoch": 22.333246414602346,
-      "grad_norm": 1.5318255424499512,
-      "learning_rate": 5.7479396984924625e-05,
-      "loss": 5.3473,
-      "step": 42824
-    },
-    {
-      "epoch": 22.333767926988266,
-      "grad_norm": 1.6261584758758545,
-      "learning_rate": 5.7478391959798996e-05,
-      "loss": 5.0366,
-      "step": 42825
-    },
-    {
-      "epoch": 22.334289439374185,
-      "grad_norm": 1.4903706312179565,
-      "learning_rate": 5.7477386934673374e-05,
-      "loss": 5.3913,
-      "step": 42826
-    },
-    {
-      "epoch": 22.334810951760105,
-      "grad_norm": 1.565327763557434,
-      "learning_rate": 5.747638190954774e-05,
-      "loss": 4.9777,
-      "step": 42827
-    },
-    {
-      "epoch": 22.335332464146024,
-      "grad_norm": 1.6317617893218994,
-      "learning_rate": 5.7475376884422116e-05,
-      "loss": 4.3646,
-      "step": 42828
-    },
-    {
-      "epoch": 22.335853976531943,
-      "grad_norm": 1.4455677270889282,
-      "learning_rate": 5.747437185929648e-05,
-      "loss": 5.6147,
-      "step": 42829
-    },
-    {
-      "epoch": 22.336375488917863,
-      "grad_norm": 1.4496139287948608,
-      "learning_rate": 5.747336683417086e-05,
-      "loss": 5.3208,
-      "step": 42830
-    },
-    {
-      "epoch": 22.336897001303782,
-      "grad_norm": 1.5573583841323853,
-      "learning_rate": 5.747236180904523e-05,
-      "loss": 5.3901,
-      "step": 42831
-    },
-    {
-      "epoch": 22.3374185136897,
-      "grad_norm": 1.6755626201629639,
-      "learning_rate": 5.747135678391961e-05,
-      "loss": 5.0111,
-      "step": 42832
-    },
-    {
-      "epoch": 22.337940026075618,
-      "grad_norm": 1.5682693719863892,
-      "learning_rate": 5.747035175879397e-05,
-      "loss": 4.9562,
-      "step": 42833
-    },
-    {
-      "epoch": 22.338461538461537,
-      "grad_norm": 1.5217403173446655,
-      "learning_rate": 5.746934673366835e-05,
-      "loss": 5.3804,
-      "step": 42834
-    },
-    {
-      "epoch": 22.338983050847457,
-      "grad_norm": 1.5286898612976074,
-      "learning_rate": 5.7468341708542714e-05,
-      "loss": 5.297,
-      "step": 42835
-    },
-    {
-      "epoch": 22.339504563233376,
-      "grad_norm": 1.5463917255401611,
-      "learning_rate": 5.7467336683417085e-05,
-      "loss": 5.2,
-      "step": 42836
-    },
-    {
-      "epoch": 22.340026075619296,
-      "grad_norm": 1.484580636024475,
-      "learning_rate": 5.746633165829146e-05,
-      "loss": 4.3636,
-      "step": 42837
-    },
-    {
-      "epoch": 22.340547588005215,
-      "grad_norm": 1.5628561973571777,
-      "learning_rate": 5.746532663316583e-05,
-      "loss": 5.5169,
-      "step": 42838
-    },
-    {
-      "epoch": 22.341069100391135,
-      "grad_norm": 1.4254413843154907,
-      "learning_rate": 5.7464321608040205e-05,
-      "loss": 5.0598,
-      "step": 42839
-    },
-    {
-      "epoch": 22.341590612777054,
-      "grad_norm": 1.5039458274841309,
-      "learning_rate": 5.7463316582914576e-05,
-      "loss": 5.3165,
-      "step": 42840
-    },
-    {
-      "epoch": 22.342112125162974,
-      "grad_norm": 1.5416696071624756,
-      "learning_rate": 5.7462311557788953e-05,
-      "loss": 5.2756,
-      "step": 42841
-    },
-    {
-      "epoch": 22.342633637548893,
-      "grad_norm": 1.6005851030349731,
-      "learning_rate": 5.746130653266332e-05,
-      "loss": 5.2146,
-      "step": 42842
-    },
-    {
-      "epoch": 22.343155149934812,
-      "grad_norm": 1.5308830738067627,
-      "learning_rate": 5.7460301507537696e-05,
-      "loss": 5.7603,
-      "step": 42843
-    },
-    {
-      "epoch": 22.343676662320732,
-      "grad_norm": 1.510621190071106,
-      "learning_rate": 5.745929648241206e-05,
-      "loss": 5.3567,
-      "step": 42844
-    },
-    {
-      "epoch": 22.344198174706648,
-      "grad_norm": 1.4573752880096436,
-      "learning_rate": 5.745829145728644e-05,
-      "loss": 5.563,
-      "step": 42845
-    },
-    {
-      "epoch": 22.344719687092567,
-      "grad_norm": 1.5491293668746948,
-      "learning_rate": 5.745728643216081e-05,
-      "loss": 5.2459,
-      "step": 42846
-    },
-    {
-      "epoch": 22.345241199478487,
-      "grad_norm": 1.5386322736740112,
-      "learning_rate": 5.7456281407035187e-05,
-      "loss": 4.7565,
-      "step": 42847
-    },
-    {
-      "epoch": 22.345762711864406,
-      "grad_norm": 1.685471773147583,
-      "learning_rate": 5.745527638190955e-05,
-      "loss": 5.2997,
-      "step": 42848
-    },
-    {
-      "epoch": 22.346284224250326,
-      "grad_norm": 1.5531773567199707,
-      "learning_rate": 5.7454271356783915e-05,
-      "loss": 5.0679,
-      "step": 42849
-    },
-    {
-      "epoch": 22.346805736636245,
-      "grad_norm": 1.622462511062622,
-      "learning_rate": 5.745326633165829e-05,
-      "loss": 5.4753,
-      "step": 42850
-    },
-    {
-      "epoch": 22.347327249022165,
-      "grad_norm": 1.5963082313537598,
-      "learning_rate": 5.7452261306532664e-05,
-      "loss": 5.0217,
-      "step": 42851
-    },
-    {
-      "epoch": 22.347848761408084,
-      "grad_norm": 1.4886051416397095,
-      "learning_rate": 5.745125628140704e-05,
-      "loss": 5.4631,
-      "step": 42852
-    },
-    {
-      "epoch": 22.348370273794004,
-      "grad_norm": 1.5990064144134521,
-      "learning_rate": 5.7450251256281406e-05,
-      "loss": 5.1362,
-      "step": 42853
-    },
-    {
-      "epoch": 22.348891786179923,
-      "grad_norm": 1.575565218925476,
-      "learning_rate": 5.7449246231155784e-05,
-      "loss": 5.2738,
-      "step": 42854
-    },
-    {
-      "epoch": 22.349413298565842,
-      "grad_norm": 1.4949923753738403,
-      "learning_rate": 5.744824120603015e-05,
-      "loss": 5.3391,
-      "step": 42855
-    },
-    {
-      "epoch": 22.34993481095176,
-      "grad_norm": 1.6351250410079956,
-      "learning_rate": 5.7447236180904526e-05,
-      "loss": 4.8036,
-      "step": 42856
-    },
-    {
-      "epoch": 22.350456323337678,
-      "grad_norm": 1.4721981287002563,
-      "learning_rate": 5.74462311557789e-05,
-      "loss": 5.2592,
-      "step": 42857
-    },
-    {
-      "epoch": 22.350977835723597,
-      "grad_norm": 1.525246024131775,
-      "learning_rate": 5.7445226130653275e-05,
-      "loss": 4.9373,
-      "step": 42858
-    },
-    {
-      "epoch": 22.351499348109517,
-      "grad_norm": 1.607305645942688,
-      "learning_rate": 5.744422110552764e-05,
-      "loss": 5.5217,
-      "step": 42859
-    },
-    {
-      "epoch": 22.352020860495436,
-      "grad_norm": 1.5342620611190796,
-      "learning_rate": 5.744321608040202e-05,
-      "loss": 5.37,
-      "step": 42860
-    },
-    {
-      "epoch": 22.352542372881356,
-      "grad_norm": 1.6928009986877441,
-      "learning_rate": 5.744221105527639e-05,
-      "loss": 5.5065,
-      "step": 42861
-    },
-    {
-      "epoch": 22.353063885267275,
-      "grad_norm": 1.533974289894104,
-      "learning_rate": 5.744120603015075e-05,
-      "loss": 5.3807,
-      "step": 42862
-    },
-    {
-      "epoch": 22.353585397653195,
-      "grad_norm": 1.5266495943069458,
-      "learning_rate": 5.744020100502513e-05,
-      "loss": 5.4767,
-      "step": 42863
-    },
-    {
-      "epoch": 22.354106910039114,
-      "grad_norm": 1.6541231870651245,
-      "learning_rate": 5.7439195979899494e-05,
-      "loss": 4.9945,
-      "step": 42864
-    },
-    {
-      "epoch": 22.354628422425034,
-      "grad_norm": 1.5179107189178467,
-      "learning_rate": 5.743819095477387e-05,
-      "loss": 5.4464,
-      "step": 42865
-    },
-    {
-      "epoch": 22.355149934810953,
-      "grad_norm": 1.6335253715515137,
-      "learning_rate": 5.743718592964824e-05,
-      "loss": 4.8241,
-      "step": 42866
-    },
-    {
-      "epoch": 22.355671447196872,
-      "grad_norm": 1.492908239364624,
-      "learning_rate": 5.743618090452262e-05,
-      "loss": 5.2071,
-      "step": 42867
-    },
-    {
-      "epoch": 22.35619295958279,
-      "grad_norm": 1.5183640718460083,
-      "learning_rate": 5.7435175879396985e-05,
-      "loss": 5.0362,
-      "step": 42868
-    },
-    {
-      "epoch": 22.356714471968708,
-      "grad_norm": 1.4831368923187256,
-      "learning_rate": 5.743417085427136e-05,
-      "loss": 5.014,
-      "step": 42869
-    },
-    {
-      "epoch": 22.357235984354627,
-      "grad_norm": 1.5525263547897339,
-      "learning_rate": 5.743316582914573e-05,
-      "loss": 5.4015,
-      "step": 42870
-    },
-    {
-      "epoch": 22.357757496740547,
-      "grad_norm": 1.5319117307662964,
-      "learning_rate": 5.7432160804020105e-05,
-      "loss": 5.3297,
-      "step": 42871
-    },
-    {
-      "epoch": 22.358279009126466,
-      "grad_norm": 1.4968235492706299,
-      "learning_rate": 5.7431155778894476e-05,
-      "loss": 5.3102,
-      "step": 42872
-    },
-    {
-      "epoch": 22.358800521512386,
-      "grad_norm": 1.566992163658142,
-      "learning_rate": 5.7430150753768854e-05,
-      "loss": 4.7126,
-      "step": 42873
-    },
-    {
-      "epoch": 22.359322033898305,
-      "grad_norm": 1.46524977684021,
-      "learning_rate": 5.742914572864322e-05,
-      "loss": 5.1122,
-      "step": 42874
-    },
-    {
-      "epoch": 22.359843546284225,
-      "grad_norm": 1.51374351978302,
-      "learning_rate": 5.742814070351758e-05,
-      "loss": 5.1658,
-      "step": 42875
-    },
-    {
-      "epoch": 22.360365058670144,
-      "grad_norm": 1.5765234231948853,
-      "learning_rate": 5.742713567839196e-05,
-      "loss": 5.3228,
-      "step": 42876
-    },
-    {
-      "epoch": 22.360886571056064,
-      "grad_norm": 1.4952105283737183,
-      "learning_rate": 5.742613065326633e-05,
-      "loss": 5.4397,
-      "step": 42877
-    },
-    {
-      "epoch": 22.361408083441983,
-      "grad_norm": 1.5856044292449951,
-      "learning_rate": 5.742512562814071e-05,
-      "loss": 5.305,
-      "step": 42878
-    },
-    {
-      "epoch": 22.361929595827903,
-      "grad_norm": 1.5251072645187378,
-      "learning_rate": 5.7424120603015074e-05,
-      "loss": 4.8119,
-      "step": 42879
-    },
-    {
-      "epoch": 22.36245110821382,
-      "grad_norm": 1.482258677482605,
-      "learning_rate": 5.742311557788945e-05,
-      "loss": 5.3897,
-      "step": 42880
-    },
-    {
-      "epoch": 22.362972620599738,
-      "grad_norm": 1.4763643741607666,
-      "learning_rate": 5.742211055276382e-05,
-      "loss": 5.4899,
-      "step": 42881
-    },
-    {
-      "epoch": 22.363494132985657,
-      "grad_norm": 1.5505344867706299,
-      "learning_rate": 5.74211055276382e-05,
-      "loss": 4.772,
-      "step": 42882
-    },
-    {
-      "epoch": 22.364015645371577,
-      "grad_norm": 1.6290355920791626,
-      "learning_rate": 5.7420100502512565e-05,
-      "loss": 5.4754,
-      "step": 42883
-    },
-    {
-      "epoch": 22.364537157757496,
-      "grad_norm": 1.5098071098327637,
-      "learning_rate": 5.741909547738694e-05,
-      "loss": 5.3604,
-      "step": 42884
-    },
-    {
-      "epoch": 22.365058670143416,
-      "grad_norm": 1.5537015199661255,
-      "learning_rate": 5.741809045226131e-05,
-      "loss": 4.7007,
-      "step": 42885
-    },
-    {
-      "epoch": 22.365580182529335,
-      "grad_norm": 1.5832258462905884,
-      "learning_rate": 5.7417085427135685e-05,
-      "loss": 4.6154,
-      "step": 42886
-    },
-    {
-      "epoch": 22.366101694915255,
-      "grad_norm": 1.550389289855957,
-      "learning_rate": 5.7416080402010056e-05,
-      "loss": 5.1796,
-      "step": 42887
-    },
-    {
-      "epoch": 22.366623207301174,
-      "grad_norm": 1.5595495700836182,
-      "learning_rate": 5.741507537688442e-05,
-      "loss": 5.2152,
-      "step": 42888
-    },
-    {
-      "epoch": 22.367144719687094,
-      "grad_norm": 1.5427892208099365,
-      "learning_rate": 5.74140703517588e-05,
-      "loss": 5.2097,
-      "step": 42889
-    },
-    {
-      "epoch": 22.367666232073013,
-      "grad_norm": 1.5164943933486938,
-      "learning_rate": 5.741306532663316e-05,
-      "loss": 5.34,
-      "step": 42890
-    },
-    {
-      "epoch": 22.368187744458933,
-      "grad_norm": 1.6050875186920166,
-      "learning_rate": 5.741206030150754e-05,
-      "loss": 5.0682,
-      "step": 42891
-    },
-    {
-      "epoch": 22.36870925684485,
-      "grad_norm": 1.515204906463623,
-      "learning_rate": 5.741105527638191e-05,
-      "loss": 5.0972,
-      "step": 42892
-    },
-    {
-      "epoch": 22.369230769230768,
-      "grad_norm": 1.5969642400741577,
-      "learning_rate": 5.741005025125629e-05,
-      "loss": 4.9931,
-      "step": 42893
-    },
-    {
-      "epoch": 22.369752281616687,
-      "grad_norm": 1.5290542840957642,
-      "learning_rate": 5.740904522613065e-05,
-      "loss": 5.3126,
-      "step": 42894
-    },
-    {
-      "epoch": 22.370273794002607,
-      "grad_norm": 1.4441026449203491,
-      "learning_rate": 5.740804020100503e-05,
-      "loss": 5.3445,
-      "step": 42895
-    },
-    {
-      "epoch": 22.370795306388526,
-      "grad_norm": 1.6258403062820435,
-      "learning_rate": 5.7407035175879395e-05,
-      "loss": 5.2518,
-      "step": 42896
-    },
-    {
-      "epoch": 22.371316818774446,
-      "grad_norm": 1.612577199935913,
-      "learning_rate": 5.740603015075377e-05,
-      "loss": 5.3864,
-      "step": 42897
-    },
-    {
-      "epoch": 22.371838331160365,
-      "grad_norm": 1.6264102458953857,
-      "learning_rate": 5.7405025125628144e-05,
-      "loss": 5.149,
-      "step": 42898
-    },
-    {
-      "epoch": 22.372359843546285,
-      "grad_norm": 1.521488904953003,
-      "learning_rate": 5.740402010050252e-05,
-      "loss": 5.5002,
-      "step": 42899
-    },
-    {
-      "epoch": 22.372881355932204,
-      "grad_norm": 1.628991961479187,
-      "learning_rate": 5.7403015075376886e-05,
-      "loss": 5.3411,
-      "step": 42900
-    },
-    {
-      "epoch": 22.373402868318124,
-      "grad_norm": 1.5440915822982788,
-      "learning_rate": 5.740201005025125e-05,
-      "loss": 5.3417,
-      "step": 42901
-    },
-    {
-      "epoch": 22.373924380704043,
-      "grad_norm": 1.5288612842559814,
-      "learning_rate": 5.740100502512563e-05,
-      "loss": 4.8967,
-      "step": 42902
-    },
-    {
-      "epoch": 22.374445893089963,
-      "grad_norm": 1.5228627920150757,
-      "learning_rate": 5.74e-05,
-      "loss": 5.184,
-      "step": 42903
-    },
-    {
-      "epoch": 22.37496740547588,
-      "grad_norm": 1.5160425901412964,
-      "learning_rate": 5.739899497487438e-05,
-      "loss": 5.5055,
-      "step": 42904
-    },
-    {
-      "epoch": 22.375488917861798,
-      "grad_norm": 1.5593863725662231,
-      "learning_rate": 5.739798994974874e-05,
-      "loss": 4.9926,
-      "step": 42905
-    },
-    {
-      "epoch": 22.376010430247717,
-      "grad_norm": 1.5880602598190308,
-      "learning_rate": 5.739698492462312e-05,
-      "loss": 5.5848,
-      "step": 42906
-    },
-    {
-      "epoch": 22.376531942633637,
-      "grad_norm": 1.6088818311691284,
-      "learning_rate": 5.739597989949749e-05,
-      "loss": 5.259,
-      "step": 42907
-    },
-    {
-      "epoch": 22.377053455019556,
-      "grad_norm": 1.5144789218902588,
-      "learning_rate": 5.739497487437187e-05,
-      "loss": 5.4824,
-      "step": 42908
-    },
-    {
-      "epoch": 22.377574967405476,
-      "grad_norm": 1.4692318439483643,
-      "learning_rate": 5.739396984924623e-05,
-      "loss": 5.177,
-      "step": 42909
-    },
-    {
-      "epoch": 22.378096479791395,
-      "grad_norm": 1.5088121891021729,
-      "learning_rate": 5.739296482412061e-05,
-      "loss": 4.8333,
-      "step": 42910
-    },
-    {
-      "epoch": 22.378617992177315,
-      "grad_norm": 1.4833545684814453,
-      "learning_rate": 5.7391959798994974e-05,
-      "loss": 5.33,
-      "step": 42911
-    },
-    {
-      "epoch": 22.379139504563234,
-      "grad_norm": 1.6378998756408691,
-      "learning_rate": 5.739095477386935e-05,
-      "loss": 5.0495,
-      "step": 42912
-    },
-    {
-      "epoch": 22.379661016949154,
-      "grad_norm": 1.4974472522735596,
-      "learning_rate": 5.738994974874372e-05,
-      "loss": 5.2579,
-      "step": 42913
-    },
-    {
-      "epoch": 22.380182529335073,
-      "grad_norm": 1.5015650987625122,
-      "learning_rate": 5.738894472361809e-05,
-      "loss": 5.3076,
-      "step": 42914
-    },
-    {
-      "epoch": 22.380704041720993,
-      "grad_norm": 1.544075608253479,
-      "learning_rate": 5.7387939698492465e-05,
-      "loss": 5.154,
-      "step": 42915
-    },
-    {
-      "epoch": 22.38122555410691,
-      "grad_norm": 1.5891059637069702,
-      "learning_rate": 5.738693467336683e-05,
-      "loss": 5.5685,
-      "step": 42916
-    },
-    {
-      "epoch": 22.381747066492828,
-      "grad_norm": 1.5550227165222168,
-      "learning_rate": 5.738592964824121e-05,
-      "loss": 4.8232,
-      "step": 42917
-    },
-    {
-      "epoch": 22.382268578878747,
-      "grad_norm": 1.5777685642242432,
-      "learning_rate": 5.738492462311558e-05,
-      "loss": 4.8894,
-      "step": 42918
-    },
-    {
-      "epoch": 22.382790091264667,
-      "grad_norm": 1.4557856321334839,
-      "learning_rate": 5.7383919597989956e-05,
-      "loss": 5.3715,
-      "step": 42919
-    },
-    {
-      "epoch": 22.383311603650586,
-      "grad_norm": 1.5824731588363647,
-      "learning_rate": 5.738291457286432e-05,
-      "loss": 5.6409,
-      "step": 42920
-    },
-    {
-      "epoch": 22.383833116036506,
-      "grad_norm": 1.6751708984375,
-      "learning_rate": 5.73819095477387e-05,
-      "loss": 4.6927,
-      "step": 42921
-    },
-    {
-      "epoch": 22.384354628422425,
-      "grad_norm": 1.4853863716125488,
-      "learning_rate": 5.738090452261306e-05,
-      "loss": 5.4917,
-      "step": 42922
-    },
-    {
-      "epoch": 22.384876140808345,
-      "grad_norm": 1.5500450134277344,
-      "learning_rate": 5.737989949748744e-05,
-      "loss": 5.5968,
-      "step": 42923
-    },
-    {
-      "epoch": 22.385397653194264,
-      "grad_norm": 1.614044427871704,
-      "learning_rate": 5.737889447236181e-05,
-      "loss": 5.1832,
-      "step": 42924
-    },
-    {
-      "epoch": 22.385919165580184,
-      "grad_norm": 1.578064203262329,
-      "learning_rate": 5.737788944723619e-05,
-      "loss": 4.5597,
-      "step": 42925
-    },
-    {
-      "epoch": 22.386440677966103,
-      "grad_norm": 1.491530179977417,
-      "learning_rate": 5.7376884422110554e-05,
-      "loss": 5.5475,
-      "step": 42926
-    },
-    {
-      "epoch": 22.386962190352023,
-      "grad_norm": 1.5222409963607788,
-      "learning_rate": 5.737587939698493e-05,
-      "loss": 5.2512,
-      "step": 42927
-    },
-    {
-      "epoch": 22.38748370273794,
-      "grad_norm": 1.5096936225891113,
-      "learning_rate": 5.73748743718593e-05,
-      "loss": 5.493,
-      "step": 42928
-    },
-    {
-      "epoch": 22.388005215123858,
-      "grad_norm": 1.3845131397247314,
-      "learning_rate": 5.737386934673367e-05,
-      "loss": 4.8905,
-      "step": 42929
-    },
-    {
-      "epoch": 22.388526727509777,
-      "grad_norm": 1.4581518173217773,
-      "learning_rate": 5.7372864321608045e-05,
-      "loss": 4.7885,
-      "step": 42930
-    },
-    {
-      "epoch": 22.389048239895697,
-      "grad_norm": 1.5288833379745483,
-      "learning_rate": 5.737185929648241e-05,
-      "loss": 5.0267,
-      "step": 42931
-    },
-    {
-      "epoch": 22.389569752281616,
-      "grad_norm": 1.5153896808624268,
-      "learning_rate": 5.737085427135679e-05,
-      "loss": 5.0125,
-      "step": 42932
-    },
-    {
-      "epoch": 22.390091264667536,
-      "grad_norm": 1.4864826202392578,
-      "learning_rate": 5.736984924623116e-05,
-      "loss": 5.3303,
-      "step": 42933
-    },
-    {
-      "epoch": 22.390612777053455,
-      "grad_norm": 1.44159734249115,
-      "learning_rate": 5.7368844221105536e-05,
-      "loss": 5.3555,
-      "step": 42934
-    },
-    {
-      "epoch": 22.391134289439375,
-      "grad_norm": 1.5137139558792114,
-      "learning_rate": 5.73678391959799e-05,
-      "loss": 4.8672,
-      "step": 42935
-    },
-    {
-      "epoch": 22.391655801825294,
-      "grad_norm": 1.5595335960388184,
-      "learning_rate": 5.736683417085428e-05,
-      "loss": 5.2208,
-      "step": 42936
-    },
-    {
-      "epoch": 22.392177314211214,
-      "grad_norm": 1.5753421783447266,
-      "learning_rate": 5.736582914572864e-05,
-      "loss": 4.6185,
-      "step": 42937
-    },
-    {
-      "epoch": 22.392698826597133,
-      "grad_norm": 1.4661716222763062,
-      "learning_rate": 5.736482412060302e-05,
-      "loss": 5.1462,
-      "step": 42938
-    },
-    {
-      "epoch": 22.39322033898305,
-      "grad_norm": 1.6038013696670532,
-      "learning_rate": 5.736381909547739e-05,
-      "loss": 5.2655,
-      "step": 42939
-    },
-    {
-      "epoch": 22.39374185136897,
-      "grad_norm": 1.5865533351898193,
-      "learning_rate": 5.736281407035177e-05,
-      "loss": 5.2245,
-      "step": 42940
-    },
-    {
-      "epoch": 22.394263363754888,
-      "grad_norm": 1.623478889465332,
-      "learning_rate": 5.736180904522613e-05,
-      "loss": 5.2023,
-      "step": 42941
-    },
-    {
-      "epoch": 22.394784876140807,
-      "grad_norm": 1.6189968585968018,
-      "learning_rate": 5.73608040201005e-05,
-      "loss": 4.8899,
-      "step": 42942
-    },
-    {
-      "epoch": 22.395306388526727,
-      "grad_norm": 1.5673750638961792,
-      "learning_rate": 5.7359798994974875e-05,
-      "loss": 5.3843,
-      "step": 42943
-    },
-    {
-      "epoch": 22.395827900912646,
-      "grad_norm": 1.4984862804412842,
-      "learning_rate": 5.7358793969849246e-05,
-      "loss": 5.6647,
-      "step": 42944
-    },
-    {
-      "epoch": 22.396349413298566,
-      "grad_norm": 1.6237870454788208,
-      "learning_rate": 5.7357788944723624e-05,
-      "loss": 4.4427,
-      "step": 42945
-    },
-    {
-      "epoch": 22.396870925684485,
-      "grad_norm": 1.6366217136383057,
-      "learning_rate": 5.735678391959799e-05,
-      "loss": 4.952,
-      "step": 42946
-    },
-    {
-      "epoch": 22.397392438070405,
-      "grad_norm": 1.5805312395095825,
-      "learning_rate": 5.7355778894472366e-05,
-      "loss": 5.1587,
-      "step": 42947
-    },
-    {
-      "epoch": 22.397913950456324,
-      "grad_norm": 1.5899157524108887,
-      "learning_rate": 5.735477386934674e-05,
-      "loss": 4.9688,
-      "step": 42948
-    },
-    {
-      "epoch": 22.398435462842244,
-      "grad_norm": 1.6820751428604126,
-      "learning_rate": 5.7353768844221115e-05,
-      "loss": 5.0211,
-      "step": 42949
-    },
-    {
-      "epoch": 22.398956975228163,
-      "grad_norm": 1.5500439405441284,
-      "learning_rate": 5.735276381909548e-05,
-      "loss": 5.35,
-      "step": 42950
-    },
-    {
-      "epoch": 22.39947848761408,
-      "grad_norm": 1.4282385110855103,
-      "learning_rate": 5.735175879396986e-05,
-      "loss": 5.0786,
-      "step": 42951
-    },
-    {
-      "epoch": 22.4,
-      "grad_norm": 1.5523983240127563,
-      "learning_rate": 5.735075376884422e-05,
-      "loss": 5.4911,
-      "step": 42952
-    },
-    {
-      "epoch": 22.400521512385918,
-      "grad_norm": 1.5324007272720337,
-      "learning_rate": 5.73497487437186e-05,
-      "loss": 5.3575,
-      "step": 42953
-    },
-    {
-      "epoch": 22.401043024771838,
-      "grad_norm": 1.5103315114974976,
-      "learning_rate": 5.734874371859297e-05,
-      "loss": 5.5573,
-      "step": 42954
-    },
-    {
-      "epoch": 22.401564537157757,
-      "grad_norm": 1.4810384511947632,
-      "learning_rate": 5.7347738693467335e-05,
-      "loss": 5.2345,
-      "step": 42955
-    },
-    {
-      "epoch": 22.402086049543676,
-      "grad_norm": 1.564318060874939,
-      "learning_rate": 5.734673366834171e-05,
-      "loss": 4.9503,
-      "step": 42956
-    },
-    {
-      "epoch": 22.402607561929596,
-      "grad_norm": 1.509112000465393,
-      "learning_rate": 5.734572864321608e-05,
-      "loss": 5.2697,
-      "step": 42957
-    },
-    {
-      "epoch": 22.403129074315515,
-      "grad_norm": 1.582565426826477,
-      "learning_rate": 5.7344723618090455e-05,
-      "loss": 5.5892,
-      "step": 42958
-    },
-    {
-      "epoch": 22.403650586701435,
-      "grad_norm": 1.498915672302246,
-      "learning_rate": 5.7343718592964826e-05,
-      "loss": 5.299,
-      "step": 42959
-    },
-    {
-      "epoch": 22.404172099087354,
-      "grad_norm": 1.4977563619613647,
-      "learning_rate": 5.7342713567839203e-05,
-      "loss": 5.4458,
-      "step": 42960
-    },
-    {
-      "epoch": 22.404693611473274,
-      "grad_norm": 1.560768961906433,
-      "learning_rate": 5.734170854271357e-05,
-      "loss": 5.3105,
-      "step": 42961
-    },
-    {
-      "epoch": 22.405215123859193,
-      "grad_norm": 1.5477086305618286,
-      "learning_rate": 5.7340703517587946e-05,
-      "loss": 5.4601,
-      "step": 42962
-    },
-    {
-      "epoch": 22.40573663624511,
-      "grad_norm": 1.5421011447906494,
-      "learning_rate": 5.733969849246231e-05,
-      "loss": 5.4691,
-      "step": 42963
-    },
-    {
-      "epoch": 22.40625814863103,
-      "grad_norm": 1.4226195812225342,
-      "learning_rate": 5.733869346733669e-05,
-      "loss": 5.7444,
-      "step": 42964
-    },
-    {
-      "epoch": 22.406779661016948,
-      "grad_norm": 1.4875434637069702,
-      "learning_rate": 5.733768844221106e-05,
-      "loss": 5.3218,
-      "step": 42965
-    },
-    {
-      "epoch": 22.407301173402868,
-      "grad_norm": 1.6513762474060059,
-      "learning_rate": 5.7336683417085437e-05,
-      "loss": 5.2812,
-      "step": 42966
-    },
-    {
-      "epoch": 22.407822685788787,
-      "grad_norm": 1.5039398670196533,
-      "learning_rate": 5.73356783919598e-05,
-      "loss": 5.3846,
-      "step": 42967
-    },
-    {
-      "epoch": 22.408344198174706,
-      "grad_norm": 1.6373463869094849,
-      "learning_rate": 5.7334673366834165e-05,
-      "loss": 5.2182,
-      "step": 42968
-    },
-    {
-      "epoch": 22.408865710560626,
-      "grad_norm": 1.5168533325195312,
-      "learning_rate": 5.733366834170854e-05,
-      "loss": 5.1719,
-      "step": 42969
-    },
-    {
-      "epoch": 22.409387222946545,
-      "grad_norm": 1.510921597480774,
-      "learning_rate": 5.7332663316582914e-05,
-      "loss": 5.241,
-      "step": 42970
-    },
-    {
-      "epoch": 22.409908735332465,
-      "grad_norm": 1.5986089706420898,
-      "learning_rate": 5.733165829145729e-05,
-      "loss": 5.2137,
-      "step": 42971
-    },
-    {
-      "epoch": 22.410430247718384,
-      "grad_norm": 1.586984634399414,
-      "learning_rate": 5.7330653266331656e-05,
-      "loss": 4.6787,
-      "step": 42972
-    },
-    {
-      "epoch": 22.410951760104304,
-      "grad_norm": 1.56332266330719,
-      "learning_rate": 5.7329648241206034e-05,
-      "loss": 4.9515,
-      "step": 42973
-    },
-    {
-      "epoch": 22.411473272490223,
-      "grad_norm": 1.6697046756744385,
-      "learning_rate": 5.7328643216080405e-05,
-      "loss": 4.6514,
-      "step": 42974
-    },
-    {
-      "epoch": 22.41199478487614,
-      "grad_norm": 1.5866037607192993,
-      "learning_rate": 5.732763819095478e-05,
-      "loss": 5.0813,
-      "step": 42975
-    },
-    {
-      "epoch": 22.41251629726206,
-      "grad_norm": 1.5401126146316528,
-      "learning_rate": 5.732663316582915e-05,
-      "loss": 5.526,
-      "step": 42976
-    },
-    {
-      "epoch": 22.413037809647978,
-      "grad_norm": 1.5919543504714966,
-      "learning_rate": 5.7325628140703525e-05,
-      "loss": 5.5045,
-      "step": 42977
-    },
-    {
-      "epoch": 22.413559322033898,
-      "grad_norm": 1.5107676982879639,
-      "learning_rate": 5.732462311557789e-05,
-      "loss": 5.4299,
-      "step": 42978
-    },
-    {
-      "epoch": 22.414080834419817,
-      "grad_norm": 1.5541752576828003,
-      "learning_rate": 5.732361809045227e-05,
-      "loss": 5.1719,
-      "step": 42979
-    },
-    {
-      "epoch": 22.414602346805736,
-      "grad_norm": 1.5348137617111206,
-      "learning_rate": 5.732261306532664e-05,
-      "loss": 5.1652,
-      "step": 42980
-    },
-    {
-      "epoch": 22.415123859191656,
-      "grad_norm": 1.5209633111953735,
-      "learning_rate": 5.7321608040201e-05,
-      "loss": 5.2305,
-      "step": 42981
-    },
-    {
-      "epoch": 22.415645371577575,
-      "grad_norm": 1.4900665283203125,
-      "learning_rate": 5.732060301507538e-05,
-      "loss": 5.0474,
-      "step": 42982
-    },
-    {
-      "epoch": 22.416166883963495,
-      "grad_norm": 1.4426689147949219,
-      "learning_rate": 5.7319597989949744e-05,
-      "loss": 5.84,
-      "step": 42983
-    },
-    {
-      "epoch": 22.416688396349414,
-      "grad_norm": 1.5433100461959839,
-      "learning_rate": 5.731859296482412e-05,
-      "loss": 5.419,
-      "step": 42984
-    },
-    {
-      "epoch": 22.417209908735334,
-      "grad_norm": 1.5430413484573364,
-      "learning_rate": 5.731758793969849e-05,
-      "loss": 5.3476,
-      "step": 42985
-    },
-    {
-      "epoch": 22.417731421121253,
-      "grad_norm": 1.5598622560501099,
-      "learning_rate": 5.731658291457287e-05,
-      "loss": 5.0451,
-      "step": 42986
-    },
-    {
-      "epoch": 22.41825293350717,
-      "grad_norm": 1.5985435247421265,
-      "learning_rate": 5.7315577889447235e-05,
-      "loss": 5.3728,
-      "step": 42987
-    },
-    {
-      "epoch": 22.41877444589309,
-      "grad_norm": 1.5666905641555786,
-      "learning_rate": 5.731457286432161e-05,
-      "loss": 5.452,
-      "step": 42988
-    },
-    {
-      "epoch": 22.419295958279008,
-      "grad_norm": 1.497247338294983,
-      "learning_rate": 5.731356783919598e-05,
-      "loss": 5.391,
-      "step": 42989
-    },
-    {
-      "epoch": 22.419817470664928,
-      "grad_norm": 1.470530390739441,
-      "learning_rate": 5.7312562814070355e-05,
-      "loss": 5.3001,
-      "step": 42990
-    },
-    {
-      "epoch": 22.420338983050847,
-      "grad_norm": 1.6626722812652588,
-      "learning_rate": 5.7311557788944726e-05,
-      "loss": 5.0425,
-      "step": 42991
-    },
-    {
-      "epoch": 22.420860495436767,
-      "grad_norm": 1.5917035341262817,
-      "learning_rate": 5.7310552763819104e-05,
-      "loss": 5.3316,
-      "step": 42992
-    },
-    {
-      "epoch": 22.421382007822686,
-      "grad_norm": 1.630299687385559,
-      "learning_rate": 5.730954773869347e-05,
-      "loss": 5.0101,
-      "step": 42993
-    },
-    {
-      "epoch": 22.421903520208605,
-      "grad_norm": 1.5384160280227661,
-      "learning_rate": 5.730854271356784e-05,
-      "loss": 5.2237,
-      "step": 42994
-    },
-    {
-      "epoch": 22.422425032594525,
-      "grad_norm": 1.5404452085494995,
-      "learning_rate": 5.730753768844222e-05,
-      "loss": 5.5181,
-      "step": 42995
-    },
-    {
-      "epoch": 22.422946544980444,
-      "grad_norm": 1.5374116897583008,
-      "learning_rate": 5.730653266331658e-05,
-      "loss": 5.3683,
-      "step": 42996
-    },
-    {
-      "epoch": 22.423468057366364,
-      "grad_norm": 1.521562099456787,
-      "learning_rate": 5.730552763819096e-05,
-      "loss": 4.9144,
-      "step": 42997
-    },
-    {
-      "epoch": 22.423989569752283,
-      "grad_norm": 1.6409480571746826,
-      "learning_rate": 5.7304522613065324e-05,
-      "loss": 5.4403,
-      "step": 42998
-    },
-    {
-      "epoch": 22.4245110821382,
-      "grad_norm": 1.4521045684814453,
-      "learning_rate": 5.73035175879397e-05,
-      "loss": 5.3833,
-      "step": 42999
-    },
-    {
-      "epoch": 22.42503259452412,
-      "grad_norm": 1.6786061525344849,
-      "learning_rate": 5.730251256281407e-05,
-      "loss": 4.5115,
-      "step": 43000
-    },
-    {
-      "epoch": 22.425554106910038,
-      "grad_norm": 1.4974408149719238,
-      "learning_rate": 5.730150753768845e-05,
-      "loss": 5.3547,
-      "step": 43001
-    },
-    {
-      "epoch": 22.426075619295958,
-      "grad_norm": 1.3582251071929932,
-      "learning_rate": 5.7300502512562815e-05,
-      "loss": 5.0592,
-      "step": 43002
-    },
-    {
-      "epoch": 22.426597131681877,
-      "grad_norm": 1.5153381824493408,
-      "learning_rate": 5.729949748743719e-05,
-      "loss": 5.1461,
-      "step": 43003
-    },
-    {
-      "epoch": 22.427118644067797,
-      "grad_norm": 1.5432884693145752,
-      "learning_rate": 5.729849246231156e-05,
-      "loss": 5.5511,
-      "step": 43004
-    },
-    {
-      "epoch": 22.427640156453716,
-      "grad_norm": 1.4955365657806396,
-      "learning_rate": 5.7297487437185935e-05,
-      "loss": 5.46,
-      "step": 43005
-    },
-    {
-      "epoch": 22.428161668839635,
-      "grad_norm": 1.6270861625671387,
-      "learning_rate": 5.7296482412060306e-05,
-      "loss": 5.0261,
-      "step": 43006
-    },
-    {
-      "epoch": 22.428683181225555,
-      "grad_norm": 1.5490330457687378,
-      "learning_rate": 5.729547738693467e-05,
-      "loss": 5.3966,
-      "step": 43007
-    },
-    {
-      "epoch": 22.429204693611474,
-      "grad_norm": 1.567928433418274,
-      "learning_rate": 5.729447236180905e-05,
-      "loss": 5.3603,
-      "step": 43008
-    },
-    {
-      "epoch": 22.429726205997394,
-      "grad_norm": 1.6198670864105225,
-      "learning_rate": 5.729346733668341e-05,
-      "loss": 4.7273,
-      "step": 43009
-    },
-    {
-      "epoch": 22.430247718383313,
-      "grad_norm": 1.5863292217254639,
-      "learning_rate": 5.729246231155779e-05,
-      "loss": 5.4659,
-      "step": 43010
-    },
-    {
-      "epoch": 22.43076923076923,
-      "grad_norm": 1.5651425123214722,
-      "learning_rate": 5.729145728643216e-05,
-      "loss": 5.0377,
-      "step": 43011
-    },
-    {
-      "epoch": 22.43129074315515,
-      "grad_norm": 1.5528302192687988,
-      "learning_rate": 5.729045226130654e-05,
-      "loss": 5.4088,
-      "step": 43012
-    },
-    {
-      "epoch": 22.431812255541068,
-      "grad_norm": 1.4889565706253052,
-      "learning_rate": 5.72894472361809e-05,
-      "loss": 5.467,
-      "step": 43013
-    },
-    {
-      "epoch": 22.432333767926988,
-      "grad_norm": 1.5110816955566406,
-      "learning_rate": 5.728844221105528e-05,
-      "loss": 5.2857,
-      "step": 43014
-    },
-    {
-      "epoch": 22.432855280312907,
-      "grad_norm": 1.5241477489471436,
-      "learning_rate": 5.728743718592965e-05,
-      "loss": 5.3819,
-      "step": 43015
-    },
-    {
-      "epoch": 22.433376792698827,
-      "grad_norm": 1.5583972930908203,
-      "learning_rate": 5.728643216080403e-05,
-      "loss": 5.3085,
-      "step": 43016
-    },
-    {
-      "epoch": 22.433898305084746,
-      "grad_norm": 1.5643582344055176,
-      "learning_rate": 5.7285427135678394e-05,
-      "loss": 5.4384,
-      "step": 43017
-    },
-    {
-      "epoch": 22.434419817470665,
-      "grad_norm": 1.5266704559326172,
-      "learning_rate": 5.728442211055277e-05,
-      "loss": 5.442,
-      "step": 43018
-    },
-    {
-      "epoch": 22.434941329856585,
-      "grad_norm": 1.509541392326355,
-      "learning_rate": 5.7283417085427136e-05,
-      "loss": 5.2518,
-      "step": 43019
-    },
-    {
-      "epoch": 22.435462842242504,
-      "grad_norm": 1.5320398807525635,
-      "learning_rate": 5.7282412060301514e-05,
-      "loss": 5.1012,
-      "step": 43020
-    },
-    {
-      "epoch": 22.435984354628424,
-      "grad_norm": 1.4879629611968994,
-      "learning_rate": 5.7281407035175885e-05,
-      "loss": 5.4898,
-      "step": 43021
-    },
-    {
-      "epoch": 22.43650586701434,
-      "grad_norm": 1.6283085346221924,
-      "learning_rate": 5.728040201005025e-05,
-      "loss": 4.9876,
-      "step": 43022
-    },
-    {
-      "epoch": 22.43702737940026,
-      "grad_norm": 1.5218249559402466,
-      "learning_rate": 5.727939698492463e-05,
-      "loss": 5.3664,
-      "step": 43023
-    },
-    {
-      "epoch": 22.43754889178618,
-      "grad_norm": 1.5896350145339966,
-      "learning_rate": 5.727839195979899e-05,
-      "loss": 5.276,
-      "step": 43024
-    },
-    {
-      "epoch": 22.438070404172098,
-      "grad_norm": 1.5657119750976562,
-      "learning_rate": 5.727738693467337e-05,
-      "loss": 5.306,
-      "step": 43025
-    },
-    {
-      "epoch": 22.438591916558018,
-      "grad_norm": 1.5224825143814087,
-      "learning_rate": 5.727638190954774e-05,
-      "loss": 5.342,
-      "step": 43026
-    },
-    {
-      "epoch": 22.439113428943937,
-      "grad_norm": 1.5480314493179321,
-      "learning_rate": 5.727537688442212e-05,
-      "loss": 3.8716,
-      "step": 43027
-    },
-    {
-      "epoch": 22.439634941329857,
-      "grad_norm": 1.5204529762268066,
-      "learning_rate": 5.727437185929648e-05,
-      "loss": 5.4225,
-      "step": 43028
-    },
-    {
-      "epoch": 22.440156453715776,
-      "grad_norm": 1.6223480701446533,
-      "learning_rate": 5.727336683417086e-05,
-      "loss": 5.0503,
-      "step": 43029
-    },
-    {
-      "epoch": 22.440677966101696,
-      "grad_norm": 1.5132330656051636,
-      "learning_rate": 5.7272361809045225e-05,
-      "loss": 5.0694,
-      "step": 43030
-    },
-    {
-      "epoch": 22.441199478487615,
-      "grad_norm": 1.5760234594345093,
-      "learning_rate": 5.72713567839196e-05,
-      "loss": 4.8916,
-      "step": 43031
-    },
-    {
-      "epoch": 22.441720990873534,
-      "grad_norm": 1.4590411186218262,
-      "learning_rate": 5.7270351758793973e-05,
-      "loss": 4.6364,
-      "step": 43032
-    },
-    {
-      "epoch": 22.442242503259454,
-      "grad_norm": 1.6879725456237793,
-      "learning_rate": 5.726934673366835e-05,
-      "loss": 5.4472,
-      "step": 43033
-    },
-    {
-      "epoch": 22.442764015645373,
-      "grad_norm": 1.487238883972168,
-      "learning_rate": 5.7268341708542715e-05,
-      "loss": 4.9121,
-      "step": 43034
-    },
-    {
-      "epoch": 22.44328552803129,
-      "grad_norm": 1.806837797164917,
-      "learning_rate": 5.7267336683417087e-05,
-      "loss": 5.123,
-      "step": 43035
-    },
-    {
-      "epoch": 22.44380704041721,
-      "grad_norm": 1.6022913455963135,
-      "learning_rate": 5.7266331658291464e-05,
-      "loss": 4.8379,
-      "step": 43036
-    },
-    {
-      "epoch": 22.444328552803128,
-      "grad_norm": 1.5299046039581299,
-      "learning_rate": 5.726532663316583e-05,
-      "loss": 5.0465,
-      "step": 43037
-    },
-    {
-      "epoch": 22.444850065189048,
-      "grad_norm": 1.4908334016799927,
-      "learning_rate": 5.7264321608040206e-05,
-      "loss": 5.672,
-      "step": 43038
-    },
-    {
-      "epoch": 22.445371577574967,
-      "grad_norm": 1.4566712379455566,
-      "learning_rate": 5.726331658291457e-05,
-      "loss": 5.5562,
-      "step": 43039
-    },
-    {
-      "epoch": 22.445893089960887,
-      "grad_norm": 1.5142180919647217,
-      "learning_rate": 5.726231155778895e-05,
-      "loss": 5.3831,
-      "step": 43040
-    },
-    {
-      "epoch": 22.446414602346806,
-      "grad_norm": 1.5296202898025513,
-      "learning_rate": 5.726130653266332e-05,
-      "loss": 5.6536,
-      "step": 43041
-    },
-    {
-      "epoch": 22.446936114732726,
-      "grad_norm": 1.5888808965682983,
-      "learning_rate": 5.72603015075377e-05,
-      "loss": 4.6286,
-      "step": 43042
-    },
-    {
-      "epoch": 22.447457627118645,
-      "grad_norm": 1.6591140031814575,
-      "learning_rate": 5.725929648241206e-05,
-      "loss": 4.5574,
-      "step": 43043
-    },
-    {
-      "epoch": 22.447979139504564,
-      "grad_norm": 1.4406663179397583,
-      "learning_rate": 5.725829145728644e-05,
-      "loss": 5.1034,
-      "step": 43044
-    },
-    {
-      "epoch": 22.448500651890484,
-      "grad_norm": 1.486299753189087,
-      "learning_rate": 5.7257286432160804e-05,
-      "loss": 5.3618,
-      "step": 43045
-    },
-    {
-      "epoch": 22.4490221642764,
-      "grad_norm": 1.5371942520141602,
-      "learning_rate": 5.725628140703518e-05,
-      "loss": 5.0789,
-      "step": 43046
-    },
-    {
-      "epoch": 22.44954367666232,
-      "grad_norm": 1.413948655128479,
-      "learning_rate": 5.725527638190955e-05,
-      "loss": 5.255,
-      "step": 43047
-    },
-    {
-      "epoch": 22.45006518904824,
-      "grad_norm": 1.5162546634674072,
-      "learning_rate": 5.725427135678392e-05,
-      "loss": 4.6927,
-      "step": 43048
-    },
-    {
-      "epoch": 22.45058670143416,
-      "grad_norm": 1.53749680519104,
-      "learning_rate": 5.7253266331658295e-05,
-      "loss": 5.3948,
-      "step": 43049
-    },
-    {
-      "epoch": 22.451108213820078,
-      "grad_norm": 1.3882874250411987,
-      "learning_rate": 5.725226130653266e-05,
-      "loss": 5.6269,
-      "step": 43050
-    },
-    {
-      "epoch": 22.451629726205997,
-      "grad_norm": 1.4944310188293457,
-      "learning_rate": 5.725125628140704e-05,
-      "loss": 5.1606,
-      "step": 43051
-    },
-    {
-      "epoch": 22.452151238591917,
-      "grad_norm": 1.5099520683288574,
-      "learning_rate": 5.725025125628141e-05,
-      "loss": 5.3241,
-      "step": 43052
-    },
-    {
-      "epoch": 22.452672750977836,
-      "grad_norm": 1.5625684261322021,
-      "learning_rate": 5.7249246231155786e-05,
-      "loss": 5.1926,
-      "step": 43053
-    },
-    {
-      "epoch": 22.453194263363756,
-      "grad_norm": 1.5217818021774292,
-      "learning_rate": 5.724824120603015e-05,
-      "loss": 5.3976,
-      "step": 43054
-    },
-    {
-      "epoch": 22.453715775749675,
-      "grad_norm": 1.533125877380371,
-      "learning_rate": 5.724723618090453e-05,
-      "loss": 5.1846,
-      "step": 43055
-    },
-    {
-      "epoch": 22.454237288135594,
-      "grad_norm": 1.3770337104797363,
-      "learning_rate": 5.724623115577889e-05,
-      "loss": 5.3023,
-      "step": 43056
-    },
-    {
-      "epoch": 22.454758800521514,
-      "grad_norm": 1.5092167854309082,
-      "learning_rate": 5.724522613065327e-05,
-      "loss": 5.5155,
-      "step": 43057
-    },
-    {
-      "epoch": 22.45528031290743,
-      "grad_norm": 1.472259759902954,
-      "learning_rate": 5.724422110552764e-05,
-      "loss": 5.2524,
-      "step": 43058
-    },
-    {
-      "epoch": 22.45580182529335,
-      "grad_norm": 1.5265469551086426,
-      "learning_rate": 5.724321608040202e-05,
-      "loss": 5.2307,
-      "step": 43059
-    },
-    {
-      "epoch": 22.45632333767927,
-      "grad_norm": 1.4518202543258667,
-      "learning_rate": 5.724221105527638e-05,
-      "loss": 4.85,
-      "step": 43060
-    },
-    {
-      "epoch": 22.45684485006519,
-      "grad_norm": 1.7922356128692627,
-      "learning_rate": 5.7241206030150754e-05,
-      "loss": 5.3117,
-      "step": 43061
-    },
-    {
-      "epoch": 22.457366362451108,
-      "grad_norm": 1.5231406688690186,
-      "learning_rate": 5.724020100502513e-05,
-      "loss": 5.3571,
-      "step": 43062
-    },
-    {
-      "epoch": 22.457887874837027,
-      "grad_norm": 1.5115879774093628,
-      "learning_rate": 5.7239195979899496e-05,
-      "loss": 4.6838,
-      "step": 43063
-    },
-    {
-      "epoch": 22.458409387222947,
-      "grad_norm": 1.5500391721725464,
-      "learning_rate": 5.7238190954773874e-05,
-      "loss": 5.6006,
-      "step": 43064
-    },
-    {
-      "epoch": 22.458930899608866,
-      "grad_norm": 1.4544562101364136,
-      "learning_rate": 5.723718592964824e-05,
-      "loss": 5.5626,
-      "step": 43065
-    },
-    {
-      "epoch": 22.459452411994786,
-      "grad_norm": 1.4498063325881958,
-      "learning_rate": 5.7236180904522616e-05,
-      "loss": 5.2381,
-      "step": 43066
-    },
-    {
-      "epoch": 22.459973924380705,
-      "grad_norm": 1.4926817417144775,
-      "learning_rate": 5.723517587939699e-05,
-      "loss": 5.476,
-      "step": 43067
-    },
-    {
-      "epoch": 22.460495436766625,
-      "grad_norm": 1.4490236043930054,
-      "learning_rate": 5.7234170854271365e-05,
-      "loss": 5.1975,
-      "step": 43068
-    },
-    {
-      "epoch": 22.461016949152544,
-      "grad_norm": 1.6239135265350342,
-      "learning_rate": 5.723316582914573e-05,
-      "loss": 5.3905,
-      "step": 43069
-    },
-    {
-      "epoch": 22.46153846153846,
-      "grad_norm": 1.4835656881332397,
-      "learning_rate": 5.723216080402011e-05,
-      "loss": 5.5078,
-      "step": 43070
-    },
-    {
-      "epoch": 22.46205997392438,
-      "grad_norm": 1.5136959552764893,
-      "learning_rate": 5.723115577889447e-05,
-      "loss": 4.8098,
-      "step": 43071
-    },
-    {
-      "epoch": 22.4625814863103,
-      "grad_norm": 1.4116190671920776,
-      "learning_rate": 5.723015075376885e-05,
-      "loss": 5.7298,
-      "step": 43072
-    },
-    {
-      "epoch": 22.46310299869622,
-      "grad_norm": 1.6326230764389038,
-      "learning_rate": 5.722914572864322e-05,
-      "loss": 4.9614,
-      "step": 43073
-    },
-    {
-      "epoch": 22.463624511082138,
-      "grad_norm": 1.6463686227798462,
-      "learning_rate": 5.7228140703517585e-05,
-      "loss": 4.9892,
-      "step": 43074
-    },
-    {
-      "epoch": 22.464146023468057,
-      "grad_norm": 1.4874809980392456,
-      "learning_rate": 5.722713567839196e-05,
-      "loss": 5.3407,
-      "step": 43075
-    },
-    {
-      "epoch": 22.464667535853977,
-      "grad_norm": 1.5444296598434448,
-      "learning_rate": 5.722613065326633e-05,
-      "loss": 5.3499,
-      "step": 43076
-    },
-    {
-      "epoch": 22.465189048239896,
-      "grad_norm": 1.4579188823699951,
-      "learning_rate": 5.7225125628140705e-05,
-      "loss": 5.1586,
-      "step": 43077
-    },
-    {
-      "epoch": 22.465710560625816,
-      "grad_norm": 1.607480764389038,
-      "learning_rate": 5.7224120603015076e-05,
-      "loss": 4.5511,
-      "step": 43078
-    },
-    {
-      "epoch": 22.466232073011735,
-      "grad_norm": 1.5057421922683716,
-      "learning_rate": 5.7223115577889453e-05,
-      "loss": 5.3737,
-      "step": 43079
-    },
-    {
-      "epoch": 22.466753585397655,
-      "grad_norm": 1.5903584957122803,
-      "learning_rate": 5.722211055276382e-05,
-      "loss": 5.0403,
-      "step": 43080
-    },
-    {
-      "epoch": 22.467275097783574,
-      "grad_norm": 1.4893826246261597,
-      "learning_rate": 5.7221105527638196e-05,
-      "loss": 5.484,
-      "step": 43081
-    },
-    {
-      "epoch": 22.46779661016949,
-      "grad_norm": 1.4427158832550049,
-      "learning_rate": 5.722010050251257e-05,
-      "loss": 5.5129,
-      "step": 43082
-    },
-    {
-      "epoch": 22.46831812255541,
-      "grad_norm": 1.5021981000900269,
-      "learning_rate": 5.7219095477386944e-05,
-      "loss": 4.8702,
-      "step": 43083
-    },
-    {
-      "epoch": 22.46883963494133,
-      "grad_norm": 1.5542888641357422,
-      "learning_rate": 5.721809045226131e-05,
-      "loss": 5.2173,
-      "step": 43084
-    },
-    {
-      "epoch": 22.46936114732725,
-      "grad_norm": 1.600624918937683,
-      "learning_rate": 5.7217085427135687e-05,
-      "loss": 4.5454,
-      "step": 43085
-    },
-    {
-      "epoch": 22.469882659713168,
-      "grad_norm": 1.5310766696929932,
-      "learning_rate": 5.721608040201005e-05,
-      "loss": 4.563,
-      "step": 43086
-    },
-    {
-      "epoch": 22.470404172099087,
-      "grad_norm": 1.5951251983642578,
-      "learning_rate": 5.721507537688442e-05,
-      "loss": 5.1721,
-      "step": 43087
-    },
-    {
-      "epoch": 22.470925684485007,
-      "grad_norm": 1.4577196836471558,
-      "learning_rate": 5.72140703517588e-05,
-      "loss": 5.6319,
-      "step": 43088
-    },
-    {
-      "epoch": 22.471447196870926,
-      "grad_norm": 1.5762759447097778,
-      "learning_rate": 5.7213065326633164e-05,
-      "loss": 4.8252,
-      "step": 43089
-    },
-    {
-      "epoch": 22.471968709256846,
-      "grad_norm": 1.5954216718673706,
-      "learning_rate": 5.721206030150754e-05,
-      "loss": 4.9435,
-      "step": 43090
-    },
-    {
-      "epoch": 22.472490221642765,
-      "grad_norm": 1.5123989582061768,
-      "learning_rate": 5.7211055276381906e-05,
-      "loss": 4.5009,
-      "step": 43091
-    },
-    {
-      "epoch": 22.473011734028685,
-      "grad_norm": 1.5634748935699463,
-      "learning_rate": 5.7210050251256284e-05,
-      "loss": 4.8732,
-      "step": 43092
-    },
-    {
-      "epoch": 22.473533246414604,
-      "grad_norm": 1.5364056825637817,
-      "learning_rate": 5.7209045226130655e-05,
-      "loss": 5.4201,
-      "step": 43093
-    },
-    {
-      "epoch": 22.47405475880052,
-      "grad_norm": 1.5165671110153198,
-      "learning_rate": 5.720804020100503e-05,
-      "loss": 5.5555,
-      "step": 43094
-    },
-    {
-      "epoch": 22.47457627118644,
-      "grad_norm": 1.589097023010254,
-      "learning_rate": 5.72070351758794e-05,
-      "loss": 5.1722,
-      "step": 43095
-    },
-    {
-      "epoch": 22.47509778357236,
-      "grad_norm": 1.7934550046920776,
-      "learning_rate": 5.7206030150753775e-05,
-      "loss": 4.9193,
-      "step": 43096
-    },
-    {
-      "epoch": 22.47561929595828,
-      "grad_norm": 1.5561072826385498,
-      "learning_rate": 5.720502512562814e-05,
-      "loss": 5.3533,
-      "step": 43097
-    },
-    {
-      "epoch": 22.476140808344198,
-      "grad_norm": 1.523532748222351,
-      "learning_rate": 5.720402010050252e-05,
-      "loss": 5.2988,
-      "step": 43098
-    },
-    {
-      "epoch": 22.476662320730117,
-      "grad_norm": 1.6269596815109253,
-      "learning_rate": 5.720301507537689e-05,
-      "loss": 5.1066,
-      "step": 43099
-    },
-    {
-      "epoch": 22.477183833116037,
-      "grad_norm": 1.5015316009521484,
-      "learning_rate": 5.720201005025125e-05,
-      "loss": 5.4576,
-      "step": 43100
-    },
-    {
-      "epoch": 22.477705345501956,
-      "grad_norm": 1.5973457098007202,
-      "learning_rate": 5.720100502512563e-05,
-      "loss": 5.3419,
-      "step": 43101
-    },
-    {
-      "epoch": 22.478226857887876,
-      "grad_norm": 1.6131362915039062,
-      "learning_rate": 5.72e-05,
-      "loss": 4.7468,
-      "step": 43102
-    },
-    {
-      "epoch": 22.478748370273795,
-      "grad_norm": 1.5770792961120605,
-      "learning_rate": 5.719899497487438e-05,
-      "loss": 5.3632,
-      "step": 43103
-    },
-    {
-      "epoch": 22.479269882659715,
-      "grad_norm": 1.5619451999664307,
-      "learning_rate": 5.719798994974874e-05,
-      "loss": 5.6035,
-      "step": 43104
-    },
-    {
-      "epoch": 22.479791395045634,
-      "grad_norm": 1.5660187005996704,
-      "learning_rate": 5.719698492462312e-05,
-      "loss": 5.2898,
-      "step": 43105
-    },
-    {
-      "epoch": 22.48031290743155,
-      "grad_norm": 1.5590558052062988,
-      "learning_rate": 5.7195979899497485e-05,
-      "loss": 5.0141,
-      "step": 43106
-    },
-    {
-      "epoch": 22.48083441981747,
-      "grad_norm": 1.6828688383102417,
-      "learning_rate": 5.719497487437186e-05,
-      "loss": 5.0394,
-      "step": 43107
-    },
-    {
-      "epoch": 22.48135593220339,
-      "grad_norm": 1.4853003025054932,
-      "learning_rate": 5.7193969849246234e-05,
-      "loss": 5.1917,
-      "step": 43108
-    },
-    {
-      "epoch": 22.48187744458931,
-      "grad_norm": 1.5906245708465576,
-      "learning_rate": 5.719296482412061e-05,
-      "loss": 4.9636,
-      "step": 43109
-    },
-    {
-      "epoch": 22.482398956975228,
-      "grad_norm": 1.4996970891952515,
-      "learning_rate": 5.7191959798994976e-05,
-      "loss": 5.4102,
-      "step": 43110
-    },
-    {
-      "epoch": 22.482920469361147,
-      "grad_norm": 1.5439682006835938,
-      "learning_rate": 5.7190954773869354e-05,
-      "loss": 5.3798,
-      "step": 43111
-    },
-    {
-      "epoch": 22.483441981747067,
-      "grad_norm": 1.5278795957565308,
-      "learning_rate": 5.718994974874372e-05,
-      "loss": 5.3671,
-      "step": 43112
-    },
-    {
-      "epoch": 22.483963494132986,
-      "grad_norm": 1.5425803661346436,
-      "learning_rate": 5.7188944723618096e-05,
-      "loss": 5.0125,
-      "step": 43113
-    },
-    {
-      "epoch": 22.484485006518906,
-      "grad_norm": 1.456473469734192,
-      "learning_rate": 5.718793969849247e-05,
-      "loss": 4.7902,
-      "step": 43114
-    },
-    {
-      "epoch": 22.485006518904825,
-      "grad_norm": 1.4851598739624023,
-      "learning_rate": 5.718693467336683e-05,
-      "loss": 5.6178,
-      "step": 43115
-    },
-    {
-      "epoch": 22.485528031290745,
-      "grad_norm": 1.537054419517517,
-      "learning_rate": 5.718592964824121e-05,
-      "loss": 4.928,
-      "step": 43116
-    },
-    {
-      "epoch": 22.486049543676664,
-      "grad_norm": 1.537340760231018,
-      "learning_rate": 5.7184924623115574e-05,
-      "loss": 4.874,
-      "step": 43117
-    },
-    {
-      "epoch": 22.48657105606258,
-      "grad_norm": 1.6599732637405396,
-      "learning_rate": 5.718391959798995e-05,
-      "loss": 5.2564,
-      "step": 43118
-    },
-    {
-      "epoch": 22.4870925684485,
-      "grad_norm": 1.506908893585205,
-      "learning_rate": 5.718291457286432e-05,
-      "loss": 5.1215,
-      "step": 43119
-    },
-    {
-      "epoch": 22.48761408083442,
-      "grad_norm": 1.5170080661773682,
-      "learning_rate": 5.71819095477387e-05,
-      "loss": 5.3622,
-      "step": 43120
-    },
-    {
-      "epoch": 22.48813559322034,
-      "grad_norm": 1.5598951578140259,
-      "learning_rate": 5.7180904522613065e-05,
-      "loss": 5.3573,
-      "step": 43121
-    },
-    {
-      "epoch": 22.488657105606258,
-      "grad_norm": 1.4889683723449707,
-      "learning_rate": 5.717989949748744e-05,
-      "loss": 5.1189,
-      "step": 43122
-    },
-    {
-      "epoch": 22.489178617992177,
-      "grad_norm": 1.6903903484344482,
-      "learning_rate": 5.7178894472361814e-05,
-      "loss": 4.7277,
-      "step": 43123
-    },
-    {
-      "epoch": 22.489700130378097,
-      "grad_norm": 1.5174801349639893,
-      "learning_rate": 5.717788944723619e-05,
-      "loss": 5.0754,
-      "step": 43124
-    },
-    {
-      "epoch": 22.490221642764016,
-      "grad_norm": 1.4573112726211548,
-      "learning_rate": 5.7176884422110556e-05,
-      "loss": 5.1322,
-      "step": 43125
-    },
-    {
-      "epoch": 22.490743155149936,
-      "grad_norm": 1.7390567064285278,
-      "learning_rate": 5.7175879396984934e-05,
-      "loss": 4.3391,
-      "step": 43126
-    },
-    {
-      "epoch": 22.491264667535855,
-      "grad_norm": 1.5083625316619873,
-      "learning_rate": 5.71748743718593e-05,
-      "loss": 5.3055,
-      "step": 43127
-    },
-    {
-      "epoch": 22.491786179921775,
-      "grad_norm": 1.455742597579956,
-      "learning_rate": 5.717386934673367e-05,
-      "loss": 4.8702,
-      "step": 43128
-    },
-    {
-      "epoch": 22.49230769230769,
-      "grad_norm": 1.4063392877578735,
-      "learning_rate": 5.717286432160805e-05,
-      "loss": 5.7432,
-      "step": 43129
-    },
-    {
-      "epoch": 22.49282920469361,
-      "grad_norm": 1.5482995510101318,
-      "learning_rate": 5.717185929648241e-05,
-      "loss": 5.4364,
-      "step": 43130
-    },
-    {
-      "epoch": 22.49335071707953,
-      "grad_norm": 1.5649389028549194,
-      "learning_rate": 5.717085427135679e-05,
-      "loss": 5.2203,
-      "step": 43131
-    },
-    {
-      "epoch": 22.49387222946545,
-      "grad_norm": 1.4924832582473755,
-      "learning_rate": 5.716984924623115e-05,
-      "loss": 5.2333,
-      "step": 43132
-    },
-    {
-      "epoch": 22.49439374185137,
-      "grad_norm": 1.5364832878112793,
-      "learning_rate": 5.716884422110553e-05,
-      "loss": 5.7237,
-      "step": 43133
-    },
-    {
-      "epoch": 22.494915254237288,
-      "grad_norm": 1.5215322971343994,
-      "learning_rate": 5.71678391959799e-05,
-      "loss": 5.149,
-      "step": 43134
-    },
-    {
-      "epoch": 22.495436766623207,
-      "grad_norm": 1.5609171390533447,
-      "learning_rate": 5.716683417085428e-05,
-      "loss": 5.1574,
-      "step": 43135
-    },
-    {
-      "epoch": 22.495958279009127,
-      "grad_norm": 1.614754319190979,
-      "learning_rate": 5.7165829145728644e-05,
-      "loss": 4.8837,
-      "step": 43136
-    },
-    {
-      "epoch": 22.496479791395046,
-      "grad_norm": 1.4762547016143799,
-      "learning_rate": 5.716482412060302e-05,
-      "loss": 5.4924,
-      "step": 43137
-    },
-    {
-      "epoch": 22.497001303780966,
-      "grad_norm": 1.4839476346969604,
-      "learning_rate": 5.7163819095477386e-05,
-      "loss": 4.8394,
-      "step": 43138
-    },
-    {
-      "epoch": 22.497522816166885,
-      "grad_norm": 1.523445725440979,
-      "learning_rate": 5.7162814070351764e-05,
-      "loss": 5.5008,
-      "step": 43139
-    },
-    {
-      "epoch": 22.498044328552805,
-      "grad_norm": 1.4341411590576172,
-      "learning_rate": 5.7161809045226135e-05,
-      "loss": 5.5009,
-      "step": 43140
-    },
-    {
-      "epoch": 22.49856584093872,
-      "grad_norm": 1.42341947555542,
-      "learning_rate": 5.71608040201005e-05,
-      "loss": 5.2516,
-      "step": 43141
-    },
-    {
-      "epoch": 22.49908735332464,
-      "grad_norm": 1.481767177581787,
-      "learning_rate": 5.715979899497488e-05,
-      "loss": 5.57,
-      "step": 43142
-    },
-    {
-      "epoch": 22.49960886571056,
-      "grad_norm": 1.5396596193313599,
-      "learning_rate": 5.715879396984924e-05,
-      "loss": 4.9598,
-      "step": 43143
-    },
-    {
-      "epoch": 22.50013037809648,
-      "grad_norm": 1.5281959772109985,
-      "learning_rate": 5.715778894472362e-05,
-      "loss": 4.8319,
-      "step": 43144
-    },
-    {
-      "epoch": 22.5006518904824,
-      "grad_norm": 1.525223970413208,
-      "learning_rate": 5.715678391959799e-05,
-      "loss": 5.4895,
-      "step": 43145
-    },
-    {
-      "epoch": 22.501173402868318,
-      "grad_norm": 1.5709556341171265,
-      "learning_rate": 5.715577889447237e-05,
-      "loss": 4.3263,
-      "step": 43146
-    },
-    {
-      "epoch": 22.501694915254237,
-      "grad_norm": 1.6439722776412964,
-      "learning_rate": 5.715477386934673e-05,
-      "loss": 5.2131,
-      "step": 43147
-    },
-    {
-      "epoch": 22.502216427640157,
-      "grad_norm": 1.4806957244873047,
-      "learning_rate": 5.715376884422111e-05,
-      "loss": 5.4224,
-      "step": 43148
-    },
-    {
-      "epoch": 22.502737940026076,
-      "grad_norm": 1.5501161813735962,
-      "learning_rate": 5.715276381909548e-05,
-      "loss": 5.5685,
-      "step": 43149
-    },
-    {
-      "epoch": 22.503259452411996,
-      "grad_norm": 1.6109122037887573,
-      "learning_rate": 5.715175879396986e-05,
-      "loss": 5.4455,
-      "step": 43150
-    },
-    {
-      "epoch": 22.503780964797915,
-      "grad_norm": 1.4691832065582275,
-      "learning_rate": 5.7150753768844223e-05,
-      "loss": 5.0062,
-      "step": 43151
-    },
-    {
-      "epoch": 22.504302477183835,
-      "grad_norm": 1.500876545906067,
-      "learning_rate": 5.71497487437186e-05,
-      "loss": 5.4606,
-      "step": 43152
-    },
-    {
-      "epoch": 22.50482398956975,
-      "grad_norm": 1.5278596878051758,
-      "learning_rate": 5.7148743718592966e-05,
-      "loss": 4.8661,
-      "step": 43153
-    },
-    {
-      "epoch": 22.50534550195567,
-      "grad_norm": 1.6564992666244507,
-      "learning_rate": 5.7147738693467337e-05,
-      "loss": 5.6592,
-      "step": 43154
-    },
-    {
-      "epoch": 22.50586701434159,
-      "grad_norm": 1.6532273292541504,
-      "learning_rate": 5.7146733668341714e-05,
-      "loss": 5.337,
-      "step": 43155
-    },
-    {
-      "epoch": 22.50638852672751,
-      "grad_norm": 1.554994821548462,
-      "learning_rate": 5.714572864321608e-05,
-      "loss": 4.941,
-      "step": 43156
-    },
-    {
-      "epoch": 22.50691003911343,
-      "grad_norm": 1.61863374710083,
-      "learning_rate": 5.7144723618090456e-05,
-      "loss": 4.9318,
-      "step": 43157
-    },
-    {
-      "epoch": 22.507431551499348,
-      "grad_norm": 1.517767071723938,
-      "learning_rate": 5.714371859296482e-05,
-      "loss": 5.3932,
-      "step": 43158
-    },
-    {
-      "epoch": 22.507953063885267,
-      "grad_norm": 1.6253814697265625,
-      "learning_rate": 5.71427135678392e-05,
-      "loss": 5.2172,
-      "step": 43159
-    },
-    {
-      "epoch": 22.508474576271187,
-      "grad_norm": 1.5596435070037842,
-      "learning_rate": 5.714170854271357e-05,
-      "loss": 5.3367,
-      "step": 43160
-    },
-    {
-      "epoch": 22.508996088657106,
-      "grad_norm": 1.6528273820877075,
-      "learning_rate": 5.714070351758795e-05,
-      "loss": 4.3917,
-      "step": 43161
-    },
-    {
-      "epoch": 22.509517601043026,
-      "grad_norm": 1.5229084491729736,
-      "learning_rate": 5.713969849246231e-05,
-      "loss": 5.387,
-      "step": 43162
-    },
-    {
-      "epoch": 22.510039113428945,
-      "grad_norm": 1.491413950920105,
-      "learning_rate": 5.713869346733669e-05,
-      "loss": 5.2887,
-      "step": 43163
-    },
-    {
-      "epoch": 22.510560625814865,
-      "grad_norm": 1.642224669456482,
-      "learning_rate": 5.7137688442211054e-05,
-      "loss": 5.1975,
-      "step": 43164
-    },
-    {
-      "epoch": 22.51108213820078,
-      "grad_norm": 1.4375232458114624,
-      "learning_rate": 5.713668341708543e-05,
-      "loss": 5.5921,
-      "step": 43165
-    },
-    {
-      "epoch": 22.5116036505867,
-      "grad_norm": 1.6763157844543457,
-      "learning_rate": 5.71356783919598e-05,
-      "loss": 4.9597,
-      "step": 43166
-    },
-    {
-      "epoch": 22.51212516297262,
-      "grad_norm": 1.5744147300720215,
-      "learning_rate": 5.713467336683417e-05,
-      "loss": 5.1265,
-      "step": 43167
-    },
-    {
-      "epoch": 22.51264667535854,
-      "grad_norm": 1.6696281433105469,
-      "learning_rate": 5.7133668341708545e-05,
-      "loss": 5.2229,
-      "step": 43168
-    },
-    {
-      "epoch": 22.51316818774446,
-      "grad_norm": 1.6615289449691772,
-      "learning_rate": 5.7132663316582916e-05,
-      "loss": 5.0917,
-      "step": 43169
-    },
-    {
-      "epoch": 22.513689700130378,
-      "grad_norm": 1.5789470672607422,
-      "learning_rate": 5.7131658291457294e-05,
-      "loss": 5.4439,
-      "step": 43170
-    },
-    {
-      "epoch": 22.514211212516297,
-      "grad_norm": 1.71641206741333,
-      "learning_rate": 5.713065326633166e-05,
-      "loss": 4.9477,
-      "step": 43171
-    },
-    {
-      "epoch": 22.514732724902217,
-      "grad_norm": 1.5849716663360596,
-      "learning_rate": 5.7129648241206036e-05,
-      "loss": 5.3681,
-      "step": 43172
-    },
-    {
-      "epoch": 22.515254237288136,
-      "grad_norm": 1.6634455919265747,
-      "learning_rate": 5.71286432160804e-05,
-      "loss": 4.9376,
-      "step": 43173
-    },
-    {
-      "epoch": 22.515775749674056,
-      "grad_norm": 1.6493333578109741,
-      "learning_rate": 5.712763819095478e-05,
-      "loss": 4.6737,
-      "step": 43174
-    },
-    {
-      "epoch": 22.516297262059975,
-      "grad_norm": 1.5191906690597534,
-      "learning_rate": 5.712663316582915e-05,
-      "loss": 5.2476,
-      "step": 43175
-    },
-    {
-      "epoch": 22.516818774445895,
-      "grad_norm": 1.4210448265075684,
-      "learning_rate": 5.712562814070353e-05,
-      "loss": 5.3542,
-      "step": 43176
-    },
-    {
-      "epoch": 22.51734028683181,
-      "grad_norm": 1.5321118831634521,
-      "learning_rate": 5.712462311557789e-05,
-      "loss": 5.2545,
-      "step": 43177
-    },
-    {
-      "epoch": 22.51786179921773,
-      "grad_norm": 1.5318512916564941,
-      "learning_rate": 5.712361809045227e-05,
-      "loss": 5.1407,
-      "step": 43178
-    },
-    {
-      "epoch": 22.51838331160365,
-      "grad_norm": 1.5893195867538452,
-      "learning_rate": 5.712261306532663e-05,
-      "loss": 4.7959,
-      "step": 43179
-    },
-    {
-      "epoch": 22.51890482398957,
-      "grad_norm": 1.54738187789917,
-      "learning_rate": 5.7121608040201004e-05,
-      "loss": 5.4886,
-      "step": 43180
-    },
-    {
-      "epoch": 22.51942633637549,
-      "grad_norm": 1.490206241607666,
-      "learning_rate": 5.712060301507538e-05,
-      "loss": 5.1809,
-      "step": 43181
-    },
-    {
-      "epoch": 22.519947848761408,
-      "grad_norm": 1.5779560804367065,
-      "learning_rate": 5.7119597989949746e-05,
-      "loss": 5.2665,
-      "step": 43182
-    },
-    {
-      "epoch": 22.520469361147327,
-      "grad_norm": 1.5259690284729004,
-      "learning_rate": 5.7118592964824124e-05,
-      "loss": 5.4031,
-      "step": 43183
-    },
-    {
-      "epoch": 22.520990873533247,
-      "grad_norm": 1.6987825632095337,
-      "learning_rate": 5.711758793969849e-05,
-      "loss": 5.1216,
-      "step": 43184
-    },
-    {
-      "epoch": 22.521512385919166,
-      "grad_norm": 1.5258303880691528,
-      "learning_rate": 5.7116582914572866e-05,
-      "loss": 5.3339,
-      "step": 43185
-    },
-    {
-      "epoch": 22.522033898305086,
-      "grad_norm": 1.5030094385147095,
-      "learning_rate": 5.711557788944724e-05,
-      "loss": 5.479,
-      "step": 43186
-    },
-    {
-      "epoch": 22.522555410691005,
-      "grad_norm": 1.5304136276245117,
-      "learning_rate": 5.7114572864321615e-05,
-      "loss": 5.7803,
-      "step": 43187
-    },
-    {
-      "epoch": 22.523076923076925,
-      "grad_norm": 1.5807774066925049,
-      "learning_rate": 5.711356783919598e-05,
-      "loss": 5.3349,
-      "step": 43188
-    },
-    {
-      "epoch": 22.52359843546284,
-      "grad_norm": 1.551010251045227,
-      "learning_rate": 5.711256281407036e-05,
-      "loss": 5.6706,
-      "step": 43189
-    },
-    {
-      "epoch": 22.52411994784876,
-      "grad_norm": 1.4996862411499023,
-      "learning_rate": 5.711155778894473e-05,
-      "loss": 5.402,
-      "step": 43190
-    },
-    {
-      "epoch": 22.52464146023468,
-      "grad_norm": 1.4900217056274414,
-      "learning_rate": 5.7110552763819106e-05,
-      "loss": 5.647,
-      "step": 43191
-    },
-    {
-      "epoch": 22.5251629726206,
-      "grad_norm": 1.6283787488937378,
-      "learning_rate": 5.710954773869347e-05,
-      "loss": 5.2732,
-      "step": 43192
-    },
-    {
-      "epoch": 22.52568448500652,
-      "grad_norm": 1.5593897104263306,
-      "learning_rate": 5.710854271356785e-05,
-      "loss": 5.4187,
-      "step": 43193
-    },
-    {
-      "epoch": 22.526205997392438,
-      "grad_norm": 1.5734307765960693,
-      "learning_rate": 5.710753768844221e-05,
-      "loss": 5.4,
-      "step": 43194
-    },
-    {
-      "epoch": 22.526727509778357,
-      "grad_norm": 1.4677157402038574,
-      "learning_rate": 5.7106532663316584e-05,
-      "loss": 4.5543,
-      "step": 43195
-    },
-    {
-      "epoch": 22.527249022164277,
-      "grad_norm": 1.5963696241378784,
-      "learning_rate": 5.710552763819096e-05,
-      "loss": 5.2349,
-      "step": 43196
-    },
-    {
-      "epoch": 22.527770534550196,
-      "grad_norm": 1.5544975996017456,
-      "learning_rate": 5.7104522613065326e-05,
-      "loss": 5.1337,
-      "step": 43197
-    },
-    {
-      "epoch": 22.528292046936116,
-      "grad_norm": 1.642124891281128,
-      "learning_rate": 5.7103517587939703e-05,
-      "loss": 5.4498,
-      "step": 43198
-    },
-    {
-      "epoch": 22.528813559322035,
-      "grad_norm": 1.6113452911376953,
-      "learning_rate": 5.710251256281407e-05,
-      "loss": 5.2065,
-      "step": 43199
-    },
-    {
-      "epoch": 22.529335071707955,
-      "grad_norm": 1.5183161497116089,
-      "learning_rate": 5.7101507537688446e-05,
-      "loss": 5.1414,
-      "step": 43200
-    },
-    {
-      "epoch": 22.52985658409387,
-      "grad_norm": 1.6247782707214355,
-      "learning_rate": 5.710050251256282e-05,
-      "loss": 5.1617,
-      "step": 43201
-    },
-    {
-      "epoch": 22.53037809647979,
-      "grad_norm": 1.5843122005462646,
-      "learning_rate": 5.7099497487437194e-05,
-      "loss": 5.3948,
-      "step": 43202
-    },
-    {
-      "epoch": 22.53089960886571,
-      "grad_norm": 1.4967448711395264,
-      "learning_rate": 5.709849246231156e-05,
-      "loss": 5.5053,
-      "step": 43203
-    },
-    {
-      "epoch": 22.53142112125163,
-      "grad_norm": 1.552748680114746,
-      "learning_rate": 5.7097487437185937e-05,
-      "loss": 5.6544,
-      "step": 43204
-    },
-    {
-      "epoch": 22.53194263363755,
-      "grad_norm": 1.5219686031341553,
-      "learning_rate": 5.70964824120603e-05,
-      "loss": 4.9539,
-      "step": 43205
-    },
-    {
-      "epoch": 22.532464146023468,
-      "grad_norm": 1.48931086063385,
-      "learning_rate": 5.709547738693468e-05,
-      "loss": 5.2648,
-      "step": 43206
-    },
-    {
-      "epoch": 22.532985658409387,
-      "grad_norm": 1.6440725326538086,
-      "learning_rate": 5.709447236180905e-05,
-      "loss": 5.2661,
-      "step": 43207
-    },
-    {
-      "epoch": 22.533507170795307,
-      "grad_norm": 1.608794927597046,
-      "learning_rate": 5.7093467336683414e-05,
-      "loss": 4.9503,
-      "step": 43208
-    },
-    {
-      "epoch": 22.534028683181226,
-      "grad_norm": 1.5316861867904663,
-      "learning_rate": 5.709246231155779e-05,
-      "loss": 5.509,
-      "step": 43209
-    },
-    {
-      "epoch": 22.534550195567146,
-      "grad_norm": 1.587078332901001,
-      "learning_rate": 5.709145728643216e-05,
-      "loss": 5.3329,
-      "step": 43210
-    },
-    {
-      "epoch": 22.535071707953065,
-      "grad_norm": 1.5029144287109375,
-      "learning_rate": 5.709045226130654e-05,
-      "loss": 5.4065,
-      "step": 43211
-    },
-    {
-      "epoch": 22.53559322033898,
-      "grad_norm": 1.5604274272918701,
-      "learning_rate": 5.7089447236180905e-05,
-      "loss": 4.9827,
-      "step": 43212
-    },
-    {
-      "epoch": 22.5361147327249,
-      "grad_norm": 1.6063064336776733,
-      "learning_rate": 5.708844221105528e-05,
-      "loss": 5.5832,
-      "step": 43213
-    },
-    {
-      "epoch": 22.53663624511082,
-      "grad_norm": 1.5680820941925049,
-      "learning_rate": 5.708743718592965e-05,
-      "loss": 5.0059,
-      "step": 43214
-    },
-    {
-      "epoch": 22.53715775749674,
-      "grad_norm": 1.5184288024902344,
-      "learning_rate": 5.7086432160804025e-05,
-      "loss": 4.5837,
-      "step": 43215
-    },
-    {
-      "epoch": 22.53767926988266,
-      "grad_norm": 1.5021497011184692,
-      "learning_rate": 5.7085427135678396e-05,
-      "loss": 4.8913,
-      "step": 43216
-    },
-    {
-      "epoch": 22.53820078226858,
-      "grad_norm": 1.8004798889160156,
-      "learning_rate": 5.7084422110552774e-05,
-      "loss": 4.8703,
-      "step": 43217
-    },
-    {
-      "epoch": 22.538722294654498,
-      "grad_norm": 1.599009394645691,
-      "learning_rate": 5.708341708542714e-05,
-      "loss": 5.3061,
-      "step": 43218
-    },
-    {
-      "epoch": 22.539243807040418,
-      "grad_norm": 1.5108356475830078,
-      "learning_rate": 5.7082412060301516e-05,
-      "loss": 5.5498,
-      "step": 43219
-    },
-    {
-      "epoch": 22.539765319426337,
-      "grad_norm": 1.545542597770691,
-      "learning_rate": 5.708140703517588e-05,
-      "loss": 4.9423,
-      "step": 43220
-    },
-    {
-      "epoch": 22.540286831812256,
-      "grad_norm": 1.5357881784439087,
-      "learning_rate": 5.708040201005025e-05,
-      "loss": 5.3218,
-      "step": 43221
-    },
-    {
-      "epoch": 22.540808344198176,
-      "grad_norm": 1.4995262622833252,
-      "learning_rate": 5.707939698492463e-05,
-      "loss": 5.5552,
-      "step": 43222
-    },
-    {
-      "epoch": 22.541329856584095,
-      "grad_norm": 1.522287130355835,
-      "learning_rate": 5.707839195979899e-05,
-      "loss": 5.4942,
-      "step": 43223
-    },
-    {
-      "epoch": 22.541851368970015,
-      "grad_norm": 1.590205192565918,
-      "learning_rate": 5.707738693467337e-05,
-      "loss": 4.6225,
-      "step": 43224
-    },
-    {
-      "epoch": 22.54237288135593,
-      "grad_norm": 1.5339124202728271,
-      "learning_rate": 5.7076381909547735e-05,
-      "loss": 4.895,
-      "step": 43225
-    },
-    {
-      "epoch": 22.54289439374185,
-      "grad_norm": 1.4556055068969727,
-      "learning_rate": 5.707537688442211e-05,
-      "loss": 5.2244,
-      "step": 43226
-    },
-    {
-      "epoch": 22.54341590612777,
-      "grad_norm": 1.5607589483261108,
-      "learning_rate": 5.7074371859296484e-05,
-      "loss": 5.1324,
-      "step": 43227
-    },
-    {
-      "epoch": 22.54393741851369,
-      "grad_norm": 1.5456149578094482,
-      "learning_rate": 5.707336683417086e-05,
-      "loss": 5.144,
-      "step": 43228
-    },
-    {
-      "epoch": 22.54445893089961,
-      "grad_norm": 1.6929330825805664,
-      "learning_rate": 5.7072361809045226e-05,
-      "loss": 5.1076,
-      "step": 43229
-    },
-    {
-      "epoch": 22.544980443285528,
-      "grad_norm": 1.5696947574615479,
-      "learning_rate": 5.7071356783919604e-05,
-      "loss": 5.37,
-      "step": 43230
-    },
-    {
-      "epoch": 22.545501955671448,
-      "grad_norm": 1.5648627281188965,
-      "learning_rate": 5.707035175879397e-05,
-      "loss": 5.3787,
-      "step": 43231
-    },
-    {
-      "epoch": 22.546023468057367,
-      "grad_norm": 1.5468299388885498,
-      "learning_rate": 5.7069346733668346e-05,
-      "loss": 4.8519,
-      "step": 43232
-    },
-    {
-      "epoch": 22.546544980443286,
-      "grad_norm": 1.586899995803833,
-      "learning_rate": 5.706834170854272e-05,
-      "loss": 5.3725,
-      "step": 43233
-    },
-    {
-      "epoch": 22.547066492829206,
-      "grad_norm": 1.4647307395935059,
-      "learning_rate": 5.706733668341708e-05,
-      "loss": 5.1812,
-      "step": 43234
-    },
-    {
-      "epoch": 22.547588005215125,
-      "grad_norm": 1.4945837259292603,
-      "learning_rate": 5.706633165829146e-05,
-      "loss": 5.4369,
-      "step": 43235
-    },
-    {
-      "epoch": 22.54810951760104,
-      "grad_norm": 1.4614646434783936,
-      "learning_rate": 5.706532663316583e-05,
-      "loss": 5.7362,
-      "step": 43236
-    },
-    {
-      "epoch": 22.54863102998696,
-      "grad_norm": 1.6803538799285889,
-      "learning_rate": 5.706432160804021e-05,
-      "loss": 5.319,
-      "step": 43237
-    },
-    {
-      "epoch": 22.54915254237288,
-      "grad_norm": 1.5274139642715454,
-      "learning_rate": 5.706331658291457e-05,
-      "loss": 5.2676,
-      "step": 43238
-    },
-    {
-      "epoch": 22.5496740547588,
-      "grad_norm": 1.5153086185455322,
-      "learning_rate": 5.706231155778895e-05,
-      "loss": 4.9862,
-      "step": 43239
-    },
-    {
-      "epoch": 22.55019556714472,
-      "grad_norm": 1.463563323020935,
-      "learning_rate": 5.7061306532663315e-05,
-      "loss": 5.6523,
-      "step": 43240
-    },
-    {
-      "epoch": 22.55071707953064,
-      "grad_norm": 1.5825313329696655,
-      "learning_rate": 5.706030150753769e-05,
-      "loss": 5.0187,
-      "step": 43241
-    },
-    {
-      "epoch": 22.551238591916558,
-      "grad_norm": 1.6053706407546997,
-      "learning_rate": 5.7059296482412064e-05,
-      "loss": 5.4993,
-      "step": 43242
-    },
-    {
-      "epoch": 22.551760104302478,
-      "grad_norm": 1.5472259521484375,
-      "learning_rate": 5.705829145728644e-05,
-      "loss": 5.1998,
-      "step": 43243
-    },
-    {
-      "epoch": 22.552281616688397,
-      "grad_norm": 1.5163027048110962,
-      "learning_rate": 5.7057286432160806e-05,
-      "loss": 4.5642,
-      "step": 43244
-    },
-    {
-      "epoch": 22.552803129074317,
-      "grad_norm": 1.5083364248275757,
-      "learning_rate": 5.7056281407035184e-05,
-      "loss": 5.2114,
-      "step": 43245
-    },
-    {
-      "epoch": 22.553324641460236,
-      "grad_norm": 1.5288965702056885,
-      "learning_rate": 5.705527638190955e-05,
-      "loss": 5.0589,
-      "step": 43246
-    },
-    {
-      "epoch": 22.553846153846155,
-      "grad_norm": 1.5227463245391846,
-      "learning_rate": 5.705427135678392e-05,
-      "loss": 5.4135,
-      "step": 43247
-    },
-    {
-      "epoch": 22.55436766623207,
-      "grad_norm": 1.4686754941940308,
-      "learning_rate": 5.70532663316583e-05,
-      "loss": 5.6706,
-      "step": 43248
-    },
-    {
-      "epoch": 22.55488917861799,
-      "grad_norm": 1.6276915073394775,
-      "learning_rate": 5.705226130653266e-05,
-      "loss": 5.0946,
-      "step": 43249
-    },
-    {
-      "epoch": 22.55541069100391,
-      "grad_norm": 1.4741880893707275,
-      "learning_rate": 5.705125628140704e-05,
-      "loss": 5.0679,
-      "step": 43250
-    },
-    {
-      "epoch": 22.55593220338983,
-      "grad_norm": 1.4882673025131226,
-      "learning_rate": 5.70502512562814e-05,
-      "loss": 5.0204,
-      "step": 43251
-    },
-    {
-      "epoch": 22.55645371577575,
-      "grad_norm": 1.5444152355194092,
-      "learning_rate": 5.704924623115578e-05,
-      "loss": 4.809,
-      "step": 43252
-    },
-    {
-      "epoch": 22.55697522816167,
-      "grad_norm": 1.5663658380508423,
-      "learning_rate": 5.704824120603015e-05,
-      "loss": 5.0445,
-      "step": 43253
-    },
-    {
-      "epoch": 22.557496740547588,
-      "grad_norm": 1.4032074213027954,
-      "learning_rate": 5.704723618090453e-05,
-      "loss": 5.7648,
-      "step": 43254
-    },
-    {
-      "epoch": 22.558018252933508,
-      "grad_norm": 1.657432198524475,
-      "learning_rate": 5.7046231155778894e-05,
-      "loss": 5.3118,
-      "step": 43255
-    },
-    {
-      "epoch": 22.558539765319427,
-      "grad_norm": 1.3873841762542725,
-      "learning_rate": 5.704522613065327e-05,
-      "loss": 5.4019,
-      "step": 43256
-    },
-    {
-      "epoch": 22.559061277705347,
-      "grad_norm": 1.4490346908569336,
-      "learning_rate": 5.704422110552764e-05,
-      "loss": 4.9548,
-      "step": 43257
-    },
-    {
-      "epoch": 22.559582790091266,
-      "grad_norm": 1.5590282678604126,
-      "learning_rate": 5.704321608040202e-05,
-      "loss": 5.5496,
-      "step": 43258
-    },
-    {
-      "epoch": 22.560104302477185,
-      "grad_norm": 1.5294862985610962,
-      "learning_rate": 5.7042211055276385e-05,
-      "loss": 4.7322,
-      "step": 43259
-    },
-    {
-      "epoch": 22.5606258148631,
-      "grad_norm": 1.4792033433914185,
-      "learning_rate": 5.704120603015075e-05,
-      "loss": 5.7083,
-      "step": 43260
-    },
-    {
-      "epoch": 22.56114732724902,
-      "grad_norm": 1.58607017993927,
-      "learning_rate": 5.704020100502513e-05,
-      "loss": 5.1641,
-      "step": 43261
-    },
-    {
-      "epoch": 22.56166883963494,
-      "grad_norm": 1.5614137649536133,
-      "learning_rate": 5.70391959798995e-05,
-      "loss": 4.8442,
-      "step": 43262
-    },
-    {
-      "epoch": 22.56219035202086,
-      "grad_norm": 1.601027488708496,
-      "learning_rate": 5.7038190954773876e-05,
-      "loss": 5.186,
-      "step": 43263
-    },
-    {
-      "epoch": 22.56271186440678,
-      "grad_norm": 1.5703307390213013,
-      "learning_rate": 5.703718592964824e-05,
-      "loss": 5.3353,
-      "step": 43264
-    },
-    {
-      "epoch": 22.5632333767927,
-      "grad_norm": 1.5101289749145508,
-      "learning_rate": 5.703618090452262e-05,
-      "loss": 4.6302,
-      "step": 43265
-    },
-    {
-      "epoch": 22.563754889178618,
-      "grad_norm": 1.6383060216903687,
-      "learning_rate": 5.703517587939698e-05,
-      "loss": 5.0555,
-      "step": 43266
-    },
-    {
-      "epoch": 22.564276401564538,
-      "grad_norm": 1.725376844406128,
-      "learning_rate": 5.703417085427136e-05,
-      "loss": 5.076,
-      "step": 43267
-    },
-    {
-      "epoch": 22.564797913950457,
-      "grad_norm": 1.5537540912628174,
-      "learning_rate": 5.703316582914573e-05,
-      "loss": 5.5138,
-      "step": 43268
-    },
-    {
-      "epoch": 22.565319426336377,
-      "grad_norm": 1.530961036682129,
-      "learning_rate": 5.703216080402011e-05,
-      "loss": 5.296,
-      "step": 43269
-    },
-    {
-      "epoch": 22.565840938722296,
-      "grad_norm": 1.5897762775421143,
-      "learning_rate": 5.7031155778894473e-05,
-      "loss": 5.2919,
-      "step": 43270
-    },
-    {
-      "epoch": 22.566362451108215,
-      "grad_norm": 1.5545247793197632,
-      "learning_rate": 5.703015075376885e-05,
-      "loss": 5.5083,
-      "step": 43271
-    },
-    {
-      "epoch": 22.56688396349413,
-      "grad_norm": 1.4740970134735107,
-      "learning_rate": 5.7029145728643216e-05,
-      "loss": 4.9061,
-      "step": 43272
-    },
-    {
-      "epoch": 22.56740547588005,
-      "grad_norm": 1.5766983032226562,
-      "learning_rate": 5.7028140703517587e-05,
-      "loss": 5.5919,
-      "step": 43273
-    },
-    {
-      "epoch": 22.56792698826597,
-      "grad_norm": 1.6207754611968994,
-      "learning_rate": 5.7027135678391964e-05,
-      "loss": 4.7018,
-      "step": 43274
-    },
-    {
-      "epoch": 22.56844850065189,
-      "grad_norm": 1.4908854961395264,
-      "learning_rate": 5.702613065326633e-05,
-      "loss": 5.1436,
-      "step": 43275
-    },
-    {
-      "epoch": 22.56897001303781,
-      "grad_norm": 1.617387056350708,
-      "learning_rate": 5.7025125628140707e-05,
-      "loss": 5.4443,
-      "step": 43276
-    },
-    {
-      "epoch": 22.56949152542373,
-      "grad_norm": 1.6835154294967651,
-      "learning_rate": 5.702412060301508e-05,
-      "loss": 5.3108,
-      "step": 43277
-    },
-    {
-      "epoch": 22.570013037809648,
-      "grad_norm": 1.5083848237991333,
-      "learning_rate": 5.7023115577889455e-05,
-      "loss": 5.4245,
-      "step": 43278
-    },
-    {
-      "epoch": 22.570534550195568,
-      "grad_norm": 1.4063540697097778,
-      "learning_rate": 5.702211055276382e-05,
-      "loss": 5.6357,
-      "step": 43279
-    },
-    {
-      "epoch": 22.571056062581487,
-      "grad_norm": 1.603441834449768,
-      "learning_rate": 5.70211055276382e-05,
-      "loss": 5.1902,
-      "step": 43280
-    },
-    {
-      "epoch": 22.571577574967407,
-      "grad_norm": 1.493664264678955,
-      "learning_rate": 5.702010050251256e-05,
-      "loss": 5.3887,
-      "step": 43281
-    },
-    {
-      "epoch": 22.572099087353326,
-      "grad_norm": 1.6283291578292847,
-      "learning_rate": 5.701909547738694e-05,
-      "loss": 4.7791,
-      "step": 43282
-    },
-    {
-      "epoch": 22.572620599739246,
-      "grad_norm": 1.5124313831329346,
-      "learning_rate": 5.701809045226131e-05,
-      "loss": 5.3175,
-      "step": 43283
-    },
-    {
-      "epoch": 22.57314211212516,
-      "grad_norm": 1.4233287572860718,
-      "learning_rate": 5.701708542713569e-05,
-      "loss": 5.6461,
-      "step": 43284
-    },
-    {
-      "epoch": 22.57366362451108,
-      "grad_norm": 1.551692008972168,
-      "learning_rate": 5.701608040201005e-05,
-      "loss": 5.4017,
-      "step": 43285
-    },
-    {
-      "epoch": 22.574185136897,
-      "grad_norm": 1.4999221563339233,
-      "learning_rate": 5.701507537688443e-05,
-      "loss": 5.7175,
-      "step": 43286
-    },
-    {
-      "epoch": 22.57470664928292,
-      "grad_norm": 1.5960990190505981,
-      "learning_rate": 5.7014070351758795e-05,
-      "loss": 4.5643,
-      "step": 43287
-    },
-    {
-      "epoch": 22.57522816166884,
-      "grad_norm": 1.4346483945846558,
-      "learning_rate": 5.7013065326633166e-05,
-      "loss": 5.6836,
-      "step": 43288
-    },
-    {
-      "epoch": 22.57574967405476,
-      "grad_norm": 1.56142258644104,
-      "learning_rate": 5.7012060301507544e-05,
-      "loss": 5.1943,
-      "step": 43289
-    },
-    {
-      "epoch": 22.576271186440678,
-      "grad_norm": 1.4480229616165161,
-      "learning_rate": 5.701105527638191e-05,
-      "loss": 5.0704,
-      "step": 43290
-    },
-    {
-      "epoch": 22.576792698826598,
-      "grad_norm": 1.407570242881775,
-      "learning_rate": 5.7010050251256286e-05,
-      "loss": 5.6288,
-      "step": 43291
-    },
-    {
-      "epoch": 22.577314211212517,
-      "grad_norm": 1.5344690084457397,
-      "learning_rate": 5.700904522613065e-05,
-      "loss": 5.5939,
-      "step": 43292
-    },
-    {
-      "epoch": 22.577835723598437,
-      "grad_norm": 1.677284598350525,
-      "learning_rate": 5.700804020100503e-05,
-      "loss": 4.953,
-      "step": 43293
-    },
-    {
-      "epoch": 22.578357235984356,
-      "grad_norm": 1.5601017475128174,
-      "learning_rate": 5.70070351758794e-05,
-      "loss": 5.1063,
-      "step": 43294
-    },
-    {
-      "epoch": 22.578878748370272,
-      "grad_norm": 1.4369345903396606,
-      "learning_rate": 5.700603015075378e-05,
-      "loss": 5.7115,
-      "step": 43295
-    },
-    {
-      "epoch": 22.57940026075619,
-      "grad_norm": 1.4899661540985107,
-      "learning_rate": 5.700502512562814e-05,
-      "loss": 5.4828,
-      "step": 43296
-    },
-    {
-      "epoch": 22.57992177314211,
-      "grad_norm": 1.593403697013855,
-      "learning_rate": 5.700402010050252e-05,
-      "loss": 5.3735,
-      "step": 43297
-    },
-    {
-      "epoch": 22.58044328552803,
-      "grad_norm": 1.502347707748413,
-      "learning_rate": 5.700301507537688e-05,
-      "loss": 5.5141,
-      "step": 43298
-    },
-    {
-      "epoch": 22.58096479791395,
-      "grad_norm": 1.551188588142395,
-      "learning_rate": 5.700201005025126e-05,
-      "loss": 5.4645,
-      "step": 43299
-    },
-    {
-      "epoch": 22.58148631029987,
-      "grad_norm": 1.475858211517334,
-      "learning_rate": 5.700100502512563e-05,
-      "loss": 5.4393,
-      "step": 43300
-    },
-    {
-      "epoch": 22.58200782268579,
-      "grad_norm": 1.5221503973007202,
-      "learning_rate": 5.6999999999999996e-05,
-      "loss": 5.3587,
-      "step": 43301
-    },
-    {
-      "epoch": 22.58252933507171,
-      "grad_norm": 1.4866483211517334,
-      "learning_rate": 5.6998994974874374e-05,
-      "loss": 5.3347,
-      "step": 43302
-    },
-    {
-      "epoch": 22.583050847457628,
-      "grad_norm": 1.4972871541976929,
-      "learning_rate": 5.6997989949748745e-05,
-      "loss": 5.1276,
-      "step": 43303
-    },
-    {
-      "epoch": 22.583572359843547,
-      "grad_norm": 1.4681146144866943,
-      "learning_rate": 5.699698492462312e-05,
-      "loss": 5.1675,
-      "step": 43304
-    },
-    {
-      "epoch": 22.584093872229467,
-      "grad_norm": 1.4607173204421997,
-      "learning_rate": 5.699597989949749e-05,
-      "loss": 5.4636,
-      "step": 43305
-    },
-    {
-      "epoch": 22.584615384615386,
-      "grad_norm": 1.6550503969192505,
-      "learning_rate": 5.6994974874371865e-05,
-      "loss": 4.2732,
-      "step": 43306
-    },
-    {
-      "epoch": 22.585136897001306,
-      "grad_norm": 1.6234793663024902,
-      "learning_rate": 5.699396984924623e-05,
-      "loss": 5.1332,
-      "step": 43307
-    },
-    {
-      "epoch": 22.58565840938722,
-      "grad_norm": 1.5409679412841797,
-      "learning_rate": 5.699296482412061e-05,
-      "loss": 5.4075,
-      "step": 43308
-    },
-    {
-      "epoch": 22.58617992177314,
-      "grad_norm": 1.5418065786361694,
-      "learning_rate": 5.699195979899498e-05,
-      "loss": 5.3218,
-      "step": 43309
-    },
-    {
-      "epoch": 22.58670143415906,
-      "grad_norm": 1.5750799179077148,
-      "learning_rate": 5.6990954773869356e-05,
-      "loss": 5.3465,
-      "step": 43310
-    },
-    {
-      "epoch": 22.58722294654498,
-      "grad_norm": 2.0511574745178223,
-      "learning_rate": 5.698994974874372e-05,
-      "loss": 4.5436,
-      "step": 43311
-    },
-    {
-      "epoch": 22.5877444589309,
-      "grad_norm": 1.7447589635849,
-      "learning_rate": 5.69889447236181e-05,
-      "loss": 5.212,
-      "step": 43312
-    },
-    {
-      "epoch": 22.58826597131682,
-      "grad_norm": 1.5185309648513794,
-      "learning_rate": 5.698793969849246e-05,
-      "loss": 5.4494,
-      "step": 43313
-    },
-    {
-      "epoch": 22.58878748370274,
-      "grad_norm": 1.567466139793396,
-      "learning_rate": 5.6986934673366834e-05,
-      "loss": 5.2042,
-      "step": 43314
-    },
-    {
-      "epoch": 22.589308996088658,
-      "grad_norm": 1.5263041257858276,
-      "learning_rate": 5.698592964824121e-05,
-      "loss": 5.116,
-      "step": 43315
-    },
-    {
-      "epoch": 22.589830508474577,
-      "grad_norm": 1.5116162300109863,
-      "learning_rate": 5.6984924623115576e-05,
-      "loss": 5.2255,
-      "step": 43316
-    },
-    {
-      "epoch": 22.590352020860497,
-      "grad_norm": 1.4477955102920532,
-      "learning_rate": 5.6983919597989954e-05,
-      "loss": 5.507,
-      "step": 43317
-    },
-    {
-      "epoch": 22.590873533246416,
-      "grad_norm": 1.481142520904541,
-      "learning_rate": 5.698291457286432e-05,
-      "loss": 5.4643,
-      "step": 43318
-    },
-    {
-      "epoch": 22.591395045632332,
-      "grad_norm": 1.499234914779663,
-      "learning_rate": 5.6981909547738696e-05,
-      "loss": 5.2349,
-      "step": 43319
-    },
-    {
-      "epoch": 22.59191655801825,
-      "grad_norm": 1.4883562326431274,
-      "learning_rate": 5.698090452261307e-05,
-      "loss": 5.5232,
-      "step": 43320
-    },
-    {
-      "epoch": 22.59243807040417,
-      "grad_norm": 1.7106218338012695,
-      "learning_rate": 5.6979899497487445e-05,
-      "loss": 5.412,
-      "step": 43321
-    },
-    {
-      "epoch": 22.59295958279009,
-      "grad_norm": 1.4643250703811646,
-      "learning_rate": 5.697889447236181e-05,
-      "loss": 5.6347,
-      "step": 43322
-    },
-    {
-      "epoch": 22.59348109517601,
-      "grad_norm": 1.4930530786514282,
-      "learning_rate": 5.6977889447236187e-05,
-      "loss": 4.5618,
-      "step": 43323
-    },
-    {
-      "epoch": 22.59400260756193,
-      "grad_norm": 1.5150835514068604,
-      "learning_rate": 5.697688442211056e-05,
-      "loss": 5.4472,
-      "step": 43324
-    },
-    {
-      "epoch": 22.59452411994785,
-      "grad_norm": 1.5299272537231445,
-      "learning_rate": 5.6975879396984935e-05,
-      "loss": 5.2204,
-      "step": 43325
-    },
-    {
-      "epoch": 22.59504563233377,
-      "grad_norm": 1.5457578897476196,
-      "learning_rate": 5.69748743718593e-05,
-      "loss": 4.6898,
-      "step": 43326
-    },
-    {
-      "epoch": 22.595567144719688,
-      "grad_norm": 1.4869896173477173,
-      "learning_rate": 5.6973869346733664e-05,
-      "loss": 5.5308,
-      "step": 43327
-    },
-    {
-      "epoch": 22.596088657105607,
-      "grad_norm": 1.5084805488586426,
-      "learning_rate": 5.697286432160804e-05,
-      "loss": 5.5239,
-      "step": 43328
-    },
-    {
-      "epoch": 22.596610169491527,
-      "grad_norm": 1.5118868350982666,
-      "learning_rate": 5.697185929648241e-05,
-      "loss": 5.3679,
-      "step": 43329
-    },
-    {
-      "epoch": 22.597131681877446,
-      "grad_norm": 1.5370091199874878,
-      "learning_rate": 5.697085427135679e-05,
-      "loss": 5.1151,
-      "step": 43330
-    },
-    {
-      "epoch": 22.597653194263362,
-      "grad_norm": 1.5238187313079834,
-      "learning_rate": 5.6969849246231155e-05,
-      "loss": 5.3439,
-      "step": 43331
-    },
-    {
-      "epoch": 22.59817470664928,
-      "grad_norm": 1.5585284233093262,
-      "learning_rate": 5.696884422110553e-05,
-      "loss": 4.9867,
-      "step": 43332
-    },
-    {
-      "epoch": 22.5986962190352,
-      "grad_norm": 1.588065266609192,
-      "learning_rate": 5.69678391959799e-05,
-      "loss": 5.1078,
-      "step": 43333
-    },
-    {
-      "epoch": 22.59921773142112,
-      "grad_norm": 1.561334490776062,
-      "learning_rate": 5.6966834170854275e-05,
-      "loss": 5.4776,
-      "step": 43334
-    },
-    {
-      "epoch": 22.59973924380704,
-      "grad_norm": 1.4545437097549438,
-      "learning_rate": 5.6965829145728646e-05,
-      "loss": 5.6543,
-      "step": 43335
-    },
-    {
-      "epoch": 22.60026075619296,
-      "grad_norm": 1.490424633026123,
-      "learning_rate": 5.6964824120603024e-05,
-      "loss": 5.4052,
-      "step": 43336
-    },
-    {
-      "epoch": 22.60078226857888,
-      "grad_norm": 1.4967079162597656,
-      "learning_rate": 5.696381909547739e-05,
-      "loss": 5.2738,
-      "step": 43337
-    },
-    {
-      "epoch": 22.6013037809648,
-      "grad_norm": 1.5988097190856934,
-      "learning_rate": 5.6962814070351766e-05,
-      "loss": 5.1009,
-      "step": 43338
-    },
-    {
-      "epoch": 22.601825293350718,
-      "grad_norm": 1.5766143798828125,
-      "learning_rate": 5.696180904522613e-05,
-      "loss": 5.6338,
-      "step": 43339
-    },
-    {
-      "epoch": 22.602346805736637,
-      "grad_norm": 1.5173710584640503,
-      "learning_rate": 5.69608040201005e-05,
-      "loss": 5.3059,
-      "step": 43340
-    },
-    {
-      "epoch": 22.602868318122557,
-      "grad_norm": 1.589680552482605,
-      "learning_rate": 5.695979899497488e-05,
-      "loss": 4.6069,
-      "step": 43341
-    },
-    {
-      "epoch": 22.603389830508476,
-      "grad_norm": 1.4725210666656494,
-      "learning_rate": 5.695879396984924e-05,
-      "loss": 5.4619,
-      "step": 43342
-    },
-    {
-      "epoch": 22.603911342894392,
-      "grad_norm": 1.5072097778320312,
-      "learning_rate": 5.695778894472362e-05,
-      "loss": 5.273,
-      "step": 43343
-    },
-    {
-      "epoch": 22.60443285528031,
-      "grad_norm": 1.6802442073822021,
-      "learning_rate": 5.695678391959799e-05,
-      "loss": 5.3098,
-      "step": 43344
-    },
-    {
-      "epoch": 22.60495436766623,
-      "grad_norm": 1.5687181949615479,
-      "learning_rate": 5.695577889447237e-05,
-      "loss": 5.3782,
-      "step": 43345
-    },
-    {
-      "epoch": 22.60547588005215,
-      "grad_norm": 1.557298183441162,
-      "learning_rate": 5.6954773869346734e-05,
-      "loss": 4.2172,
-      "step": 43346
-    },
-    {
-      "epoch": 22.60599739243807,
-      "grad_norm": 1.4804941415786743,
-      "learning_rate": 5.695376884422111e-05,
-      "loss": 5.5081,
-      "step": 43347
-    },
-    {
-      "epoch": 22.60651890482399,
-      "grad_norm": 1.5630719661712646,
-      "learning_rate": 5.6952763819095476e-05,
-      "loss": 5.2766,
-      "step": 43348
-    },
-    {
-      "epoch": 22.60704041720991,
-      "grad_norm": 1.5567946434020996,
-      "learning_rate": 5.6951758793969854e-05,
-      "loss": 5.383,
-      "step": 43349
-    },
-    {
-      "epoch": 22.60756192959583,
-      "grad_norm": 1.6519356966018677,
-      "learning_rate": 5.6950753768844225e-05,
-      "loss": 4.7608,
-      "step": 43350
-    },
-    {
-      "epoch": 22.608083441981748,
-      "grad_norm": 1.5048415660858154,
-      "learning_rate": 5.69497487437186e-05,
-      "loss": 5.087,
-      "step": 43351
-    },
-    {
-      "epoch": 22.608604954367667,
-      "grad_norm": 1.5025955438613892,
-      "learning_rate": 5.694874371859297e-05,
-      "loss": 5.6412,
-      "step": 43352
-    },
-    {
-      "epoch": 22.609126466753587,
-      "grad_norm": 1.4400184154510498,
-      "learning_rate": 5.694773869346733e-05,
-      "loss": 5.3645,
-      "step": 43353
-    },
-    {
-      "epoch": 22.609647979139506,
-      "grad_norm": 1.5571807622909546,
-      "learning_rate": 5.694673366834171e-05,
-      "loss": 4.4673,
-      "step": 43354
-    },
-    {
-      "epoch": 22.610169491525422,
-      "grad_norm": 1.501349925994873,
-      "learning_rate": 5.694572864321608e-05,
-      "loss": 5.3425,
-      "step": 43355
-    },
-    {
-      "epoch": 22.61069100391134,
-      "grad_norm": 1.6236090660095215,
-      "learning_rate": 5.694472361809046e-05,
-      "loss": 4.658,
-      "step": 43356
-    },
-    {
-      "epoch": 22.61121251629726,
-      "grad_norm": 1.526179313659668,
-      "learning_rate": 5.694371859296482e-05,
-      "loss": 4.9504,
-      "step": 43357
-    },
-    {
-      "epoch": 22.61173402868318,
-      "grad_norm": 1.5293267965316772,
-      "learning_rate": 5.69427135678392e-05,
-      "loss": 5.1592,
-      "step": 43358
-    },
-    {
-      "epoch": 22.6122555410691,
-      "grad_norm": 1.6158748865127563,
-      "learning_rate": 5.6941708542713565e-05,
-      "loss": 5.191,
-      "step": 43359
-    },
-    {
-      "epoch": 22.61277705345502,
-      "grad_norm": 1.4765465259552002,
-      "learning_rate": 5.694070351758794e-05,
-      "loss": 5.3882,
-      "step": 43360
-    },
-    {
-      "epoch": 22.61329856584094,
-      "grad_norm": 1.545562505722046,
-      "learning_rate": 5.6939698492462314e-05,
-      "loss": 4.7515,
-      "step": 43361
-    },
-    {
-      "epoch": 22.61382007822686,
-      "grad_norm": 1.5682399272918701,
-      "learning_rate": 5.693869346733669e-05,
-      "loss": 5.7196,
-      "step": 43362
-    },
-    {
-      "epoch": 22.614341590612778,
-      "grad_norm": 1.6733752489089966,
-      "learning_rate": 5.6937688442211056e-05,
-      "loss": 4.8292,
-      "step": 43363
-    },
-    {
-      "epoch": 22.614863102998697,
-      "grad_norm": 1.580146074295044,
-      "learning_rate": 5.6936683417085434e-05,
-      "loss": 5.0742,
-      "step": 43364
-    },
-    {
-      "epoch": 22.615384615384617,
-      "grad_norm": 1.4860153198242188,
-      "learning_rate": 5.6935678391959805e-05,
-      "loss": 5.4281,
-      "step": 43365
-    },
-    {
-      "epoch": 22.615906127770536,
-      "grad_norm": 1.531622290611267,
-      "learning_rate": 5.693467336683417e-05,
-      "loss": 5.33,
-      "step": 43366
-    },
-    {
-      "epoch": 22.616427640156452,
-      "grad_norm": 1.4644767045974731,
-      "learning_rate": 5.693366834170855e-05,
-      "loss": 5.4932,
-      "step": 43367
-    },
-    {
-      "epoch": 22.61694915254237,
-      "grad_norm": 1.5520808696746826,
-      "learning_rate": 5.693266331658291e-05,
-      "loss": 4.9282,
-      "step": 43368
-    },
-    {
-      "epoch": 22.61747066492829,
-      "grad_norm": 1.4997872114181519,
-      "learning_rate": 5.693165829145729e-05,
-      "loss": 5.1316,
-      "step": 43369
-    },
-    {
-      "epoch": 22.61799217731421,
-      "grad_norm": 1.5395225286483765,
-      "learning_rate": 5.693065326633166e-05,
-      "loss": 5.3594,
-      "step": 43370
-    },
-    {
-      "epoch": 22.61851368970013,
-      "grad_norm": 1.6091073751449585,
-      "learning_rate": 5.692964824120604e-05,
-      "loss": 4.8329,
-      "step": 43371
-    },
-    {
-      "epoch": 22.61903520208605,
-      "grad_norm": 1.6259477138519287,
-      "learning_rate": 5.69286432160804e-05,
-      "loss": 4.9108,
-      "step": 43372
-    },
-    {
-      "epoch": 22.61955671447197,
-      "grad_norm": 1.5601993799209595,
-      "learning_rate": 5.692763819095478e-05,
-      "loss": 5.1551,
-      "step": 43373
-    },
-    {
-      "epoch": 22.62007822685789,
-      "grad_norm": 1.6736907958984375,
-      "learning_rate": 5.6926633165829144e-05,
-      "loss": 4.7729,
-      "step": 43374
-    },
-    {
-      "epoch": 22.620599739243808,
-      "grad_norm": 1.617360234260559,
-      "learning_rate": 5.692562814070352e-05,
-      "loss": 5.4219,
-      "step": 43375
-    },
-    {
-      "epoch": 22.621121251629727,
-      "grad_norm": 1.5513502359390259,
-      "learning_rate": 5.692462311557789e-05,
-      "loss": 5.0617,
-      "step": 43376
-    },
-    {
-      "epoch": 22.621642764015647,
-      "grad_norm": 1.5137771368026733,
-      "learning_rate": 5.692361809045227e-05,
-      "loss": 5.2299,
-      "step": 43377
-    },
-    {
-      "epoch": 22.622164276401566,
-      "grad_norm": 1.4652163982391357,
-      "learning_rate": 5.6922613065326635e-05,
-      "loss": 5.5083,
-      "step": 43378
-    },
-    {
-      "epoch": 22.622685788787482,
-      "grad_norm": 1.447622537612915,
-      "learning_rate": 5.692160804020101e-05,
-      "loss": 5.6127,
-      "step": 43379
-    },
-    {
-      "epoch": 22.6232073011734,
-      "grad_norm": 1.6096493005752563,
-      "learning_rate": 5.692060301507538e-05,
-      "loss": 4.89,
-      "step": 43380
-    },
-    {
-      "epoch": 22.62372881355932,
-      "grad_norm": 1.4738688468933105,
-      "learning_rate": 5.691959798994975e-05,
-      "loss": 5.2871,
-      "step": 43381
-    },
-    {
-      "epoch": 22.62425032594524,
-      "grad_norm": 1.5074107646942139,
-      "learning_rate": 5.6918592964824126e-05,
-      "loss": 5.5065,
-      "step": 43382
-    },
-    {
-      "epoch": 22.62477183833116,
-      "grad_norm": 1.4599087238311768,
-      "learning_rate": 5.691758793969849e-05,
-      "loss": 5.5128,
-      "step": 43383
-    },
-    {
-      "epoch": 22.62529335071708,
-      "grad_norm": 1.5737236738204956,
-      "learning_rate": 5.691658291457287e-05,
-      "loss": 5.1804,
-      "step": 43384
-    },
-    {
-      "epoch": 22.625814863103,
-      "grad_norm": 1.5290586948394775,
-      "learning_rate": 5.691557788944723e-05,
-      "loss": 5.443,
-      "step": 43385
-    },
-    {
-      "epoch": 22.62633637548892,
-      "grad_norm": 1.628075361251831,
-      "learning_rate": 5.691457286432161e-05,
-      "loss": 5.3524,
-      "step": 43386
-    },
-    {
-      "epoch": 22.626857887874838,
-      "grad_norm": 1.5058979988098145,
-      "learning_rate": 5.691356783919598e-05,
-      "loss": 5.5274,
-      "step": 43387
-    },
-    {
-      "epoch": 22.627379400260757,
-      "grad_norm": 1.4948521852493286,
-      "learning_rate": 5.691256281407036e-05,
-      "loss": 5.2274,
-      "step": 43388
-    },
-    {
-      "epoch": 22.627900912646677,
-      "grad_norm": 1.455642580986023,
-      "learning_rate": 5.6911557788944723e-05,
-      "loss": 4.8948,
-      "step": 43389
-    },
-    {
-      "epoch": 22.628422425032596,
-      "grad_norm": 1.4949259757995605,
-      "learning_rate": 5.69105527638191e-05,
-      "loss": 5.6025,
-      "step": 43390
-    },
-    {
-      "epoch": 22.628943937418512,
-      "grad_norm": 1.5313342809677124,
-      "learning_rate": 5.690954773869347e-05,
-      "loss": 5.6891,
-      "step": 43391
-    },
-    {
-      "epoch": 22.62946544980443,
-      "grad_norm": 1.5686628818511963,
-      "learning_rate": 5.690854271356785e-05,
-      "loss": 5.3456,
-      "step": 43392
-    },
-    {
-      "epoch": 22.62998696219035,
-      "grad_norm": 1.6155208349227905,
-      "learning_rate": 5.6907537688442214e-05,
-      "loss": 4.7524,
-      "step": 43393
-    },
-    {
-      "epoch": 22.63050847457627,
-      "grad_norm": 1.396476149559021,
-      "learning_rate": 5.690653266331658e-05,
-      "loss": 5.3261,
-      "step": 43394
-    },
-    {
-      "epoch": 22.63102998696219,
-      "grad_norm": 1.5512588024139404,
-      "learning_rate": 5.6905527638190957e-05,
-      "loss": 4.4608,
-      "step": 43395
-    },
-    {
-      "epoch": 22.63155149934811,
-      "grad_norm": 1.4881954193115234,
-      "learning_rate": 5.690452261306533e-05,
-      "loss": 5.3872,
-      "step": 43396
-    },
-    {
-      "epoch": 22.63207301173403,
-      "grad_norm": 1.5852798223495483,
-      "learning_rate": 5.6903517587939705e-05,
-      "loss": 5.2358,
-      "step": 43397
-    },
-    {
-      "epoch": 22.63259452411995,
-      "grad_norm": 1.5538225173950195,
-      "learning_rate": 5.690251256281407e-05,
-      "loss": 5.3084,
-      "step": 43398
-    },
-    {
-      "epoch": 22.633116036505868,
-      "grad_norm": 1.6098660230636597,
-      "learning_rate": 5.690150753768845e-05,
-      "loss": 5.207,
-      "step": 43399
-    },
-    {
-      "epoch": 22.633637548891787,
-      "grad_norm": 1.544649600982666,
-      "learning_rate": 5.690050251256281e-05,
-      "loss": 5.0862,
-      "step": 43400
-    },
-    {
-      "epoch": 22.634159061277707,
-      "grad_norm": 1.5251119136810303,
-      "learning_rate": 5.689949748743719e-05,
-      "loss": 4.37,
-      "step": 43401
-    },
-    {
-      "epoch": 22.634680573663623,
-      "grad_norm": 1.5308527946472168,
-      "learning_rate": 5.689849246231156e-05,
-      "loss": 5.4535,
-      "step": 43402
-    },
-    {
-      "epoch": 22.635202086049542,
-      "grad_norm": 1.5559757947921753,
-      "learning_rate": 5.689748743718594e-05,
-      "loss": 5.4594,
-      "step": 43403
-    },
-    {
-      "epoch": 22.63572359843546,
-      "grad_norm": 1.6279411315917969,
-      "learning_rate": 5.68964824120603e-05,
-      "loss": 5.1286,
-      "step": 43404
-    },
-    {
-      "epoch": 22.63624511082138,
-      "grad_norm": 1.6169337034225464,
-      "learning_rate": 5.689547738693468e-05,
-      "loss": 5.5433,
-      "step": 43405
-    },
-    {
-      "epoch": 22.6367666232073,
-      "grad_norm": 1.470681071281433,
-      "learning_rate": 5.6894472361809045e-05,
-      "loss": 5.3321,
-      "step": 43406
-    },
-    {
-      "epoch": 22.63728813559322,
-      "grad_norm": 1.7545480728149414,
-      "learning_rate": 5.6893467336683416e-05,
-      "loss": 4.5339,
-      "step": 43407
-    },
-    {
-      "epoch": 22.63780964797914,
-      "grad_norm": 1.5468413829803467,
-      "learning_rate": 5.6892462311557794e-05,
-      "loss": 5.5344,
-      "step": 43408
-    },
-    {
-      "epoch": 22.63833116036506,
-      "grad_norm": 1.5201261043548584,
-      "learning_rate": 5.689145728643216e-05,
-      "loss": 5.2349,
-      "step": 43409
-    },
-    {
-      "epoch": 22.63885267275098,
-      "grad_norm": 1.591415286064148,
-      "learning_rate": 5.6890452261306536e-05,
-      "loss": 5.4789,
-      "step": 43410
-    },
-    {
-      "epoch": 22.639374185136898,
-      "grad_norm": 1.5852774381637573,
-      "learning_rate": 5.688944723618091e-05,
-      "loss": 4.9207,
-      "step": 43411
-    },
-    {
-      "epoch": 22.639895697522817,
-      "grad_norm": 1.5791733264923096,
-      "learning_rate": 5.6888442211055285e-05,
-      "loss": 5.2969,
-      "step": 43412
-    },
-    {
-      "epoch": 22.640417209908737,
-      "grad_norm": 1.498121976852417,
-      "learning_rate": 5.688743718592965e-05,
-      "loss": 5.5087,
-      "step": 43413
-    },
-    {
-      "epoch": 22.640938722294656,
-      "grad_norm": 1.600346565246582,
-      "learning_rate": 5.688643216080403e-05,
-      "loss": 5.0717,
-      "step": 43414
-    },
-    {
-      "epoch": 22.641460234680572,
-      "grad_norm": 1.5682787895202637,
-      "learning_rate": 5.688542713567839e-05,
-      "loss": 5.4815,
-      "step": 43415
-    },
-    {
-      "epoch": 22.64198174706649,
-      "grad_norm": 1.5074329376220703,
-      "learning_rate": 5.688442211055277e-05,
-      "loss": 5.5268,
-      "step": 43416
-    },
-    {
-      "epoch": 22.64250325945241,
-      "grad_norm": 1.5524739027023315,
-      "learning_rate": 5.688341708542714e-05,
-      "loss": 5.5597,
-      "step": 43417
-    },
-    {
-      "epoch": 22.64302477183833,
-      "grad_norm": 1.5351697206497192,
-      "learning_rate": 5.688241206030152e-05,
-      "loss": 5.0179,
-      "step": 43418
-    },
-    {
-      "epoch": 22.64354628422425,
-      "grad_norm": 1.496140480041504,
-      "learning_rate": 5.688140703517588e-05,
-      "loss": 5.3386,
-      "step": 43419
-    },
-    {
-      "epoch": 22.64406779661017,
-      "grad_norm": 1.448280692100525,
-      "learning_rate": 5.6880402010050246e-05,
-      "loss": 5.4146,
-      "step": 43420
-    },
-    {
-      "epoch": 22.64458930899609,
-      "grad_norm": 1.6069144010543823,
-      "learning_rate": 5.6879396984924624e-05,
-      "loss": 5.3766,
-      "step": 43421
-    },
-    {
-      "epoch": 22.64511082138201,
-      "grad_norm": 1.459661841392517,
-      "learning_rate": 5.6878391959798995e-05,
-      "loss": 4.4722,
-      "step": 43422
-    },
-    {
-      "epoch": 22.645632333767928,
-      "grad_norm": 1.5729111433029175,
-      "learning_rate": 5.687738693467337e-05,
-      "loss": 5.4468,
-      "step": 43423
-    },
-    {
-      "epoch": 22.646153846153847,
-      "grad_norm": 1.602390170097351,
-      "learning_rate": 5.687638190954774e-05,
-      "loss": 5.2264,
-      "step": 43424
-    },
-    {
-      "epoch": 22.646675358539767,
-      "grad_norm": 1.680341124534607,
-      "learning_rate": 5.6875376884422115e-05,
-      "loss": 5.0955,
-      "step": 43425
-    },
-    {
-      "epoch": 22.647196870925683,
-      "grad_norm": 1.6376413106918335,
-      "learning_rate": 5.687437185929648e-05,
-      "loss": 4.767,
-      "step": 43426
-    },
-    {
-      "epoch": 22.647718383311602,
-      "grad_norm": 1.5398024320602417,
-      "learning_rate": 5.687336683417086e-05,
-      "loss": 5.6833,
-      "step": 43427
-    },
-    {
-      "epoch": 22.64823989569752,
-      "grad_norm": 1.5230238437652588,
-      "learning_rate": 5.687236180904523e-05,
-      "loss": 5.3886,
-      "step": 43428
-    },
-    {
-      "epoch": 22.64876140808344,
-      "grad_norm": 1.5326732397079468,
-      "learning_rate": 5.6871356783919606e-05,
-      "loss": 5.5722,
-      "step": 43429
-    },
-    {
-      "epoch": 22.64928292046936,
-      "grad_norm": 1.5411186218261719,
-      "learning_rate": 5.687035175879397e-05,
-      "loss": 5.2122,
-      "step": 43430
-    },
-    {
-      "epoch": 22.64980443285528,
-      "grad_norm": 1.5744762420654297,
-      "learning_rate": 5.686934673366835e-05,
-      "loss": 4.9478,
-      "step": 43431
-    },
-    {
-      "epoch": 22.6503259452412,
-      "grad_norm": 1.6029685735702515,
-      "learning_rate": 5.686834170854272e-05,
-      "loss": 4.4127,
-      "step": 43432
-    },
-    {
-      "epoch": 22.65084745762712,
-      "grad_norm": 1.4619553089141846,
-      "learning_rate": 5.6867336683417084e-05,
-      "loss": 5.5804,
-      "step": 43433
-    },
-    {
-      "epoch": 22.65136897001304,
-      "grad_norm": 1.495229959487915,
-      "learning_rate": 5.686633165829146e-05,
-      "loss": 5.3774,
-      "step": 43434
-    },
-    {
-      "epoch": 22.651890482398958,
-      "grad_norm": 1.512961983680725,
-      "learning_rate": 5.6865326633165826e-05,
-      "loss": 5.2747,
-      "step": 43435
-    },
-    {
-      "epoch": 22.652411994784877,
-      "grad_norm": 1.6186894178390503,
-      "learning_rate": 5.6864321608040204e-05,
-      "loss": 4.6514,
-      "step": 43436
-    },
-    {
-      "epoch": 22.652933507170797,
-      "grad_norm": 1.6435412168502808,
-      "learning_rate": 5.6863316582914575e-05,
-      "loss": 5.2517,
-      "step": 43437
-    },
-    {
-      "epoch": 22.653455019556713,
-      "grad_norm": 1.561131238937378,
-      "learning_rate": 5.686231155778895e-05,
-      "loss": 5.1358,
-      "step": 43438
-    },
-    {
-      "epoch": 22.653976531942632,
-      "grad_norm": 1.530440092086792,
-      "learning_rate": 5.686130653266332e-05,
-      "loss": 5.0956,
-      "step": 43439
-    },
-    {
-      "epoch": 22.65449804432855,
-      "grad_norm": 1.5329136848449707,
-      "learning_rate": 5.6860301507537695e-05,
-      "loss": 5.2689,
-      "step": 43440
-    },
-    {
-      "epoch": 22.65501955671447,
-      "grad_norm": 1.511812448501587,
-      "learning_rate": 5.685929648241206e-05,
-      "loss": 5.2321,
-      "step": 43441
-    },
-    {
-      "epoch": 22.65554106910039,
-      "grad_norm": 1.5152184963226318,
-      "learning_rate": 5.685829145728644e-05,
-      "loss": 4.9315,
-      "step": 43442
-    },
-    {
-      "epoch": 22.65606258148631,
-      "grad_norm": 1.6392571926116943,
-      "learning_rate": 5.685728643216081e-05,
-      "loss": 5.094,
-      "step": 43443
-    },
-    {
-      "epoch": 22.65658409387223,
-      "grad_norm": 1.3999741077423096,
-      "learning_rate": 5.6856281407035186e-05,
-      "loss": 5.3789,
-      "step": 43444
-    },
-    {
-      "epoch": 22.65710560625815,
-      "grad_norm": 1.5350711345672607,
-      "learning_rate": 5.685527638190955e-05,
-      "loss": 5.225,
-      "step": 43445
-    },
-    {
-      "epoch": 22.65762711864407,
-      "grad_norm": 1.5857648849487305,
-      "learning_rate": 5.6854271356783914e-05,
-      "loss": 5.2556,
-      "step": 43446
-    },
-    {
-      "epoch": 22.658148631029988,
-      "grad_norm": 1.46377694606781,
-      "learning_rate": 5.685326633165829e-05,
-      "loss": 5.6548,
-      "step": 43447
-    },
-    {
-      "epoch": 22.658670143415907,
-      "grad_norm": 1.5281000137329102,
-      "learning_rate": 5.685226130653266e-05,
-      "loss": 5.6763,
-      "step": 43448
-    },
-    {
-      "epoch": 22.659191655801827,
-      "grad_norm": 1.4469417333602905,
-      "learning_rate": 5.685125628140704e-05,
-      "loss": 5.4109,
-      "step": 43449
-    },
-    {
-      "epoch": 22.659713168187743,
-      "grad_norm": 1.5894666910171509,
-      "learning_rate": 5.6850251256281405e-05,
-      "loss": 5.0664,
-      "step": 43450
-    },
-    {
-      "epoch": 22.660234680573662,
-      "grad_norm": 1.5560327768325806,
-      "learning_rate": 5.684924623115578e-05,
-      "loss": 5.498,
-      "step": 43451
-    },
-    {
-      "epoch": 22.66075619295958,
-      "grad_norm": 1.506399393081665,
-      "learning_rate": 5.6848241206030154e-05,
-      "loss": 5.3683,
-      "step": 43452
-    },
-    {
-      "epoch": 22.6612777053455,
-      "grad_norm": 1.5730129480361938,
-      "learning_rate": 5.684723618090453e-05,
-      "loss": 5.2027,
-      "step": 43453
-    },
-    {
-      "epoch": 22.66179921773142,
-      "grad_norm": 1.5869864225387573,
-      "learning_rate": 5.6846231155778896e-05,
-      "loss": 4.9663,
-      "step": 43454
-    },
-    {
-      "epoch": 22.66232073011734,
-      "grad_norm": 1.370853066444397,
-      "learning_rate": 5.6845226130653274e-05,
-      "loss": 5.8782,
-      "step": 43455
-    },
-    {
-      "epoch": 22.66284224250326,
-      "grad_norm": 1.5110903978347778,
-      "learning_rate": 5.684422110552764e-05,
-      "loss": 5.4926,
-      "step": 43456
-    },
-    {
-      "epoch": 22.66336375488918,
-      "grad_norm": 1.5697135925292969,
-      "learning_rate": 5.6843216080402016e-05,
-      "loss": 5.2669,
-      "step": 43457
-    },
-    {
-      "epoch": 22.6638852672751,
-      "grad_norm": 1.5262030363082886,
-      "learning_rate": 5.684221105527639e-05,
-      "loss": 5.1738,
-      "step": 43458
-    },
-    {
-      "epoch": 22.664406779661018,
-      "grad_norm": 1.6138776540756226,
-      "learning_rate": 5.684120603015075e-05,
-      "loss": 4.9495,
-      "step": 43459
-    },
-    {
-      "epoch": 22.664928292046937,
-      "grad_norm": 1.5255464315414429,
-      "learning_rate": 5.684020100502513e-05,
-      "loss": 5.8651,
-      "step": 43460
-    },
-    {
-      "epoch": 22.665449804432857,
-      "grad_norm": 1.4810463190078735,
-      "learning_rate": 5.683919597989949e-05,
-      "loss": 5.5218,
-      "step": 43461
-    },
-    {
-      "epoch": 22.665971316818773,
-      "grad_norm": 1.510490894317627,
-      "learning_rate": 5.683819095477387e-05,
-      "loss": 5.3512,
-      "step": 43462
-    },
-    {
-      "epoch": 22.666492829204692,
-      "grad_norm": 1.6599408388137817,
-      "learning_rate": 5.683718592964824e-05,
-      "loss": 4.6698,
-      "step": 43463
-    },
-    {
-      "epoch": 22.667014341590612,
-      "grad_norm": 1.5557852983474731,
-      "learning_rate": 5.683618090452262e-05,
-      "loss": 5.1996,
-      "step": 43464
-    },
-    {
-      "epoch": 22.66753585397653,
-      "grad_norm": 1.4641211032867432,
-      "learning_rate": 5.6835175879396984e-05,
-      "loss": 5.4641,
-      "step": 43465
-    },
-    {
-      "epoch": 22.66805736636245,
-      "grad_norm": 1.5515233278274536,
-      "learning_rate": 5.683417085427136e-05,
-      "loss": 4.9672,
-      "step": 43466
-    },
-    {
-      "epoch": 22.66857887874837,
-      "grad_norm": 1.4964193105697632,
-      "learning_rate": 5.6833165829145726e-05,
-      "loss": 5.4476,
-      "step": 43467
-    },
-    {
-      "epoch": 22.66910039113429,
-      "grad_norm": 1.560524582862854,
-      "learning_rate": 5.6832160804020104e-05,
-      "loss": 5.0456,
-      "step": 43468
-    },
-    {
-      "epoch": 22.66962190352021,
-      "grad_norm": 1.5053558349609375,
-      "learning_rate": 5.6831155778894475e-05,
-      "loss": 5.7724,
-      "step": 43469
-    },
-    {
-      "epoch": 22.67014341590613,
-      "grad_norm": 1.4117577075958252,
-      "learning_rate": 5.683015075376885e-05,
-      "loss": 5.0519,
-      "step": 43470
-    },
-    {
-      "epoch": 22.670664928292048,
-      "grad_norm": 1.4025828838348389,
-      "learning_rate": 5.682914572864322e-05,
-      "loss": 4.9893,
-      "step": 43471
-    },
-    {
-      "epoch": 22.671186440677968,
-      "grad_norm": 1.6143975257873535,
-      "learning_rate": 5.6828140703517595e-05,
-      "loss": 5.1728,
-      "step": 43472
-    },
-    {
-      "epoch": 22.671707953063887,
-      "grad_norm": 1.624085783958435,
-      "learning_rate": 5.682713567839196e-05,
-      "loss": 4.8093,
-      "step": 43473
-    },
-    {
-      "epoch": 22.672229465449803,
-      "grad_norm": 1.5144000053405762,
-      "learning_rate": 5.682613065326633e-05,
-      "loss": 5.3705,
-      "step": 43474
-    },
-    {
-      "epoch": 22.672750977835722,
-      "grad_norm": 1.5274183750152588,
-      "learning_rate": 5.682512562814071e-05,
-      "loss": 5.3755,
-      "step": 43475
-    },
-    {
-      "epoch": 22.673272490221642,
-      "grad_norm": 1.468226432800293,
-      "learning_rate": 5.682412060301507e-05,
-      "loss": 5.4724,
-      "step": 43476
-    },
-    {
-      "epoch": 22.67379400260756,
-      "grad_norm": 1.4956412315368652,
-      "learning_rate": 5.682311557788945e-05,
-      "loss": 5.2382,
-      "step": 43477
-    },
-    {
-      "epoch": 22.67431551499348,
-      "grad_norm": 1.5511806011199951,
-      "learning_rate": 5.682211055276382e-05,
-      "loss": 5.3287,
-      "step": 43478
-    },
-    {
-      "epoch": 22.6748370273794,
-      "grad_norm": 1.6077880859375,
-      "learning_rate": 5.68211055276382e-05,
-      "loss": 4.9851,
-      "step": 43479
-    },
-    {
-      "epoch": 22.67535853976532,
-      "grad_norm": 1.5828624963760376,
-      "learning_rate": 5.6820100502512564e-05,
-      "loss": 5.4475,
-      "step": 43480
-    },
-    {
-      "epoch": 22.67588005215124,
-      "grad_norm": 1.4829890727996826,
-      "learning_rate": 5.681909547738694e-05,
-      "loss": 5.6844,
-      "step": 43481
-    },
-    {
-      "epoch": 22.67640156453716,
-      "grad_norm": 1.503535270690918,
-      "learning_rate": 5.6818090452261306e-05,
-      "loss": 4.952,
-      "step": 43482
-    },
-    {
-      "epoch": 22.676923076923078,
-      "grad_norm": 1.7117077112197876,
-      "learning_rate": 5.6817085427135684e-05,
-      "loss": 5.4892,
-      "step": 43483
-    },
-    {
-      "epoch": 22.677444589308998,
-      "grad_norm": 1.5806976556777954,
-      "learning_rate": 5.6816080402010055e-05,
-      "loss": 5.5147,
-      "step": 43484
-    },
-    {
-      "epoch": 22.677966101694913,
-      "grad_norm": 1.714493989944458,
-      "learning_rate": 5.681507537688443e-05,
-      "loss": 5.1543,
-      "step": 43485
-    },
-    {
-      "epoch": 22.678487614080833,
-      "grad_norm": 1.5772976875305176,
-      "learning_rate": 5.68140703517588e-05,
-      "loss": 5.0268,
-      "step": 43486
-    },
-    {
-      "epoch": 22.679009126466752,
-      "grad_norm": 1.585015892982483,
-      "learning_rate": 5.681306532663316e-05,
-      "loss": 5.7949,
-      "step": 43487
-    },
-    {
-      "epoch": 22.679530638852672,
-      "grad_norm": 1.4531642198562622,
-      "learning_rate": 5.681206030150754e-05,
-      "loss": 5.3754,
-      "step": 43488
-    },
-    {
-      "epoch": 22.68005215123859,
-      "grad_norm": 1.4830397367477417,
-      "learning_rate": 5.681105527638191e-05,
-      "loss": 4.8818,
-      "step": 43489
-    },
-    {
-      "epoch": 22.68057366362451,
-      "grad_norm": 1.594510793685913,
-      "learning_rate": 5.681005025125629e-05,
-      "loss": 4.9427,
-      "step": 43490
-    },
-    {
-      "epoch": 22.68109517601043,
-      "grad_norm": 1.6483129262924194,
-      "learning_rate": 5.680904522613065e-05,
-      "loss": 5.4484,
-      "step": 43491
-    },
-    {
-      "epoch": 22.68161668839635,
-      "grad_norm": 1.7799484729766846,
-      "learning_rate": 5.680804020100503e-05,
-      "loss": 4.49,
-      "step": 43492
-    },
-    {
-      "epoch": 22.68213820078227,
-      "grad_norm": 1.4736367464065552,
-      "learning_rate": 5.6807035175879394e-05,
-      "loss": 5.2339,
-      "step": 43493
-    },
-    {
-      "epoch": 22.68265971316819,
-      "grad_norm": 1.5133734941482544,
-      "learning_rate": 5.680603015075377e-05,
-      "loss": 5.3411,
-      "step": 43494
-    },
-    {
-      "epoch": 22.683181225554108,
-      "grad_norm": 1.626164436340332,
-      "learning_rate": 5.680502512562814e-05,
-      "loss": 4.5177,
-      "step": 43495
-    },
-    {
-      "epoch": 22.683702737940028,
-      "grad_norm": 1.632803201675415,
-      "learning_rate": 5.680402010050252e-05,
-      "loss": 4.6958,
-      "step": 43496
-    },
-    {
-      "epoch": 22.684224250325947,
-      "grad_norm": 1.4988389015197754,
-      "learning_rate": 5.6803015075376885e-05,
-      "loss": 5.4149,
-      "step": 43497
-    },
-    {
-      "epoch": 22.684745762711863,
-      "grad_norm": 1.5293842554092407,
-      "learning_rate": 5.680201005025126e-05,
-      "loss": 5.531,
-      "step": 43498
-    },
-    {
-      "epoch": 22.685267275097782,
-      "grad_norm": 1.604077696800232,
-      "learning_rate": 5.6801005025125634e-05,
-      "loss": 5.2417,
-      "step": 43499
-    },
-    {
-      "epoch": 22.685788787483702,
-      "grad_norm": 1.4824504852294922,
-      "learning_rate": 5.68e-05,
-      "loss": 5.3253,
-      "step": 43500
-    },
-    {
-      "epoch": 22.68631029986962,
-      "grad_norm": 1.5647428035736084,
-      "learning_rate": 5.6798994974874376e-05,
-      "loss": 5.5283,
-      "step": 43501
-    },
-    {
-      "epoch": 22.68683181225554,
-      "grad_norm": 1.624413251876831,
-      "learning_rate": 5.679798994974874e-05,
-      "loss": 4.9395,
-      "step": 43502
-    },
-    {
-      "epoch": 22.68735332464146,
-      "grad_norm": 1.4406791925430298,
-      "learning_rate": 5.679698492462312e-05,
-      "loss": 5.2668,
-      "step": 43503
-    },
-    {
-      "epoch": 22.68787483702738,
-      "grad_norm": 1.4966073036193848,
-      "learning_rate": 5.679597989949749e-05,
-      "loss": 5.357,
-      "step": 43504
-    },
-    {
-      "epoch": 22.6883963494133,
-      "grad_norm": 1.5166990756988525,
-      "learning_rate": 5.679497487437187e-05,
-      "loss": 5.5096,
-      "step": 43505
-    },
-    {
-      "epoch": 22.68891786179922,
-      "grad_norm": 1.568264126777649,
-      "learning_rate": 5.679396984924623e-05,
-      "loss": 5.3761,
-      "step": 43506
-    },
-    {
-      "epoch": 22.689439374185138,
-      "grad_norm": 1.5365759134292603,
-      "learning_rate": 5.679296482412061e-05,
-      "loss": 5.2755,
-      "step": 43507
-    },
-    {
-      "epoch": 22.689960886571058,
-      "grad_norm": 1.5792359113693237,
-      "learning_rate": 5.6791959798994973e-05,
-      "loss": 5.0718,
-      "step": 43508
-    },
-    {
-      "epoch": 22.690482398956973,
-      "grad_norm": 1.488196849822998,
-      "learning_rate": 5.679095477386935e-05,
-      "loss": 5.6517,
-      "step": 43509
-    },
-    {
-      "epoch": 22.691003911342893,
-      "grad_norm": 1.5041993856430054,
-      "learning_rate": 5.678994974874372e-05,
-      "loss": 5.0485,
-      "step": 43510
-    },
-    {
-      "epoch": 22.691525423728812,
-      "grad_norm": 1.5619808435440063,
-      "learning_rate": 5.67889447236181e-05,
-      "loss": 4.5185,
-      "step": 43511
-    },
-    {
-      "epoch": 22.692046936114732,
-      "grad_norm": 1.5983145236968994,
-      "learning_rate": 5.6787939698492464e-05,
-      "loss": 5.2646,
-      "step": 43512
-    },
-    {
-      "epoch": 22.69256844850065,
-      "grad_norm": 1.5716911554336548,
-      "learning_rate": 5.678693467336683e-05,
-      "loss": 5.2386,
-      "step": 43513
-    },
-    {
-      "epoch": 22.69308996088657,
-      "grad_norm": 1.590446949005127,
-      "learning_rate": 5.6785929648241207e-05,
-      "loss": 5.2132,
-      "step": 43514
-    },
-    {
-      "epoch": 22.69361147327249,
-      "grad_norm": 1.6289048194885254,
-      "learning_rate": 5.678492462311558e-05,
-      "loss": 5.3967,
-      "step": 43515
-    },
-    {
-      "epoch": 22.69413298565841,
-      "grad_norm": 1.5387171506881714,
-      "learning_rate": 5.6783919597989955e-05,
-      "loss": 5.2391,
-      "step": 43516
-    },
-    {
-      "epoch": 22.69465449804433,
-      "grad_norm": 1.7043304443359375,
-      "learning_rate": 5.678291457286432e-05,
-      "loss": 5.0614,
-      "step": 43517
-    },
-    {
-      "epoch": 22.69517601043025,
-      "grad_norm": 1.5253762006759644,
-      "learning_rate": 5.67819095477387e-05,
-      "loss": 5.6708,
-      "step": 43518
-    },
-    {
-      "epoch": 22.695697522816168,
-      "grad_norm": 1.5298278331756592,
-      "learning_rate": 5.678090452261307e-05,
-      "loss": 5.2562,
-      "step": 43519
-    },
-    {
-      "epoch": 22.696219035202088,
-      "grad_norm": 1.509393334388733,
-      "learning_rate": 5.6779899497487446e-05,
-      "loss": 5.2055,
-      "step": 43520
-    },
-    {
-      "epoch": 22.696740547588004,
-      "grad_norm": 1.4686716794967651,
-      "learning_rate": 5.677889447236181e-05,
-      "loss": 5.4163,
-      "step": 43521
-    },
-    {
-      "epoch": 22.697262059973923,
-      "grad_norm": 1.708779215812683,
-      "learning_rate": 5.677788944723619e-05,
-      "loss": 5.1745,
-      "step": 43522
-    },
-    {
-      "epoch": 22.697783572359842,
-      "grad_norm": 1.5766597986221313,
-      "learning_rate": 5.677688442211055e-05,
-      "loss": 5.1826,
-      "step": 43523
-    },
-    {
-      "epoch": 22.698305084745762,
-      "grad_norm": 1.630409598350525,
-      "learning_rate": 5.677587939698493e-05,
-      "loss": 5.0573,
-      "step": 43524
-    },
-    {
-      "epoch": 22.69882659713168,
-      "grad_norm": 1.6618056297302246,
-      "learning_rate": 5.67748743718593e-05,
-      "loss": 4.916,
-      "step": 43525
-    },
-    {
-      "epoch": 22.6993481095176,
-      "grad_norm": 1.4817008972167969,
-      "learning_rate": 5.6773869346733666e-05,
-      "loss": 5.1963,
-      "step": 43526
-    },
-    {
-      "epoch": 22.69986962190352,
-      "grad_norm": 1.5734859704971313,
-      "learning_rate": 5.6772864321608044e-05,
-      "loss": 5.0859,
-      "step": 43527
-    },
-    {
-      "epoch": 22.70039113428944,
-      "grad_norm": 1.5953967571258545,
-      "learning_rate": 5.677185929648241e-05,
-      "loss": 5.3767,
-      "step": 43528
-    },
-    {
-      "epoch": 22.70091264667536,
-      "grad_norm": 1.5312923192977905,
-      "learning_rate": 5.6770854271356786e-05,
-      "loss": 5.2412,
-      "step": 43529
-    },
-    {
-      "epoch": 22.70143415906128,
-      "grad_norm": 1.4814808368682861,
-      "learning_rate": 5.676984924623116e-05,
-      "loss": 5.6486,
-      "step": 43530
-    },
-    {
-      "epoch": 22.701955671447198,
-      "grad_norm": 1.40824294090271,
-      "learning_rate": 5.6768844221105535e-05,
-      "loss": 5.5594,
-      "step": 43531
-    },
-    {
-      "epoch": 22.702477183833118,
-      "grad_norm": 1.492762804031372,
-      "learning_rate": 5.67678391959799e-05,
-      "loss": 4.9764,
-      "step": 43532
-    },
-    {
-      "epoch": 22.702998696219034,
-      "grad_norm": 1.5337538719177246,
-      "learning_rate": 5.676683417085428e-05,
-      "loss": 5.7506,
-      "step": 43533
-    },
-    {
-      "epoch": 22.703520208604953,
-      "grad_norm": 1.5257457494735718,
-      "learning_rate": 5.676582914572864e-05,
-      "loss": 5.2241,
-      "step": 43534
-    },
-    {
-      "epoch": 22.704041720990872,
-      "grad_norm": 1.679463267326355,
-      "learning_rate": 5.676482412060302e-05,
-      "loss": 4.7901,
-      "step": 43535
-    },
-    {
-      "epoch": 22.704563233376792,
-      "grad_norm": 1.5392781496047974,
-      "learning_rate": 5.676381909547739e-05,
-      "loss": 5.0232,
-      "step": 43536
-    },
-    {
-      "epoch": 22.70508474576271,
-      "grad_norm": 1.5237269401550293,
-      "learning_rate": 5.676281407035177e-05,
-      "loss": 5.5379,
-      "step": 43537
-    },
-    {
-      "epoch": 22.70560625814863,
-      "grad_norm": 1.4810911417007446,
-      "learning_rate": 5.676180904522613e-05,
-      "loss": 5.6565,
-      "step": 43538
-    },
-    {
-      "epoch": 22.70612777053455,
-      "grad_norm": 1.4005436897277832,
-      "learning_rate": 5.67608040201005e-05,
-      "loss": 5.393,
-      "step": 43539
-    },
-    {
-      "epoch": 22.70664928292047,
-      "grad_norm": 1.6453535556793213,
-      "learning_rate": 5.675979899497488e-05,
-      "loss": 5.1728,
-      "step": 43540
-    },
-    {
-      "epoch": 22.70717079530639,
-      "grad_norm": 1.5412641763687134,
-      "learning_rate": 5.6758793969849245e-05,
-      "loss": 5.409,
-      "step": 43541
-    },
-    {
-      "epoch": 22.70769230769231,
-      "grad_norm": 1.6259503364562988,
-      "learning_rate": 5.675778894472362e-05,
-      "loss": 4.4565,
-      "step": 43542
-    },
-    {
-      "epoch": 22.708213820078228,
-      "grad_norm": 1.551547884941101,
-      "learning_rate": 5.675678391959799e-05,
-      "loss": 5.1626,
-      "step": 43543
-    },
-    {
-      "epoch": 22.708735332464148,
-      "grad_norm": 1.6154043674468994,
-      "learning_rate": 5.6755778894472365e-05,
-      "loss": 4.972,
-      "step": 43544
-    },
-    {
-      "epoch": 22.709256844850064,
-      "grad_norm": 1.4810090065002441,
-      "learning_rate": 5.6754773869346736e-05,
-      "loss": 4.8992,
-      "step": 43545
-    },
-    {
-      "epoch": 22.709778357235983,
-      "grad_norm": 1.4425731897354126,
-      "learning_rate": 5.6753768844221114e-05,
-      "loss": 5.3868,
-      "step": 43546
-    },
-    {
-      "epoch": 22.710299869621903,
-      "grad_norm": 1.5424957275390625,
-      "learning_rate": 5.675276381909548e-05,
-      "loss": 5.0856,
-      "step": 43547
-    },
-    {
-      "epoch": 22.710821382007822,
-      "grad_norm": 1.5184839963912964,
-      "learning_rate": 5.6751758793969856e-05,
-      "loss": 5.228,
-      "step": 43548
-    },
-    {
-      "epoch": 22.71134289439374,
-      "grad_norm": 1.5284631252288818,
-      "learning_rate": 5.675075376884422e-05,
-      "loss": 5.183,
-      "step": 43549
-    },
-    {
-      "epoch": 22.71186440677966,
-      "grad_norm": 1.5117998123168945,
-      "learning_rate": 5.67497487437186e-05,
-      "loss": 5.2011,
-      "step": 43550
-    },
-    {
-      "epoch": 22.71238591916558,
-      "grad_norm": 1.5185693502426147,
-      "learning_rate": 5.674874371859297e-05,
-      "loss": 5.0064,
-      "step": 43551
-    },
-    {
-      "epoch": 22.7129074315515,
-      "grad_norm": 1.5872184038162231,
-      "learning_rate": 5.674773869346735e-05,
-      "loss": 4.9059,
-      "step": 43552
-    },
-    {
-      "epoch": 22.71342894393742,
-      "grad_norm": 1.5759347677230835,
-      "learning_rate": 5.674673366834171e-05,
-      "loss": 5.5491,
-      "step": 43553
-    },
-    {
-      "epoch": 22.71395045632334,
-      "grad_norm": 1.7255204916000366,
-      "learning_rate": 5.6745728643216076e-05,
-      "loss": 4.7667,
-      "step": 43554
-    },
-    {
-      "epoch": 22.714471968709258,
-      "grad_norm": 1.613344669342041,
-      "learning_rate": 5.6744723618090454e-05,
-      "loss": 4.965,
-      "step": 43555
-    },
-    {
-      "epoch": 22.714993481095178,
-      "grad_norm": 1.6315727233886719,
-      "learning_rate": 5.6743718592964825e-05,
-      "loss": 5.1913,
-      "step": 43556
-    },
-    {
-      "epoch": 22.715514993481094,
-      "grad_norm": 1.446747899055481,
-      "learning_rate": 5.67427135678392e-05,
-      "loss": 5.3358,
-      "step": 43557
-    },
-    {
-      "epoch": 22.716036505867013,
-      "grad_norm": 1.5313606262207031,
-      "learning_rate": 5.674170854271357e-05,
-      "loss": 4.5914,
-      "step": 43558
-    },
-    {
-      "epoch": 22.716558018252933,
-      "grad_norm": 1.5648066997528076,
-      "learning_rate": 5.6740703517587945e-05,
-      "loss": 5.1248,
-      "step": 43559
-    },
-    {
-      "epoch": 22.717079530638852,
-      "grad_norm": 1.560665249824524,
-      "learning_rate": 5.673969849246231e-05,
-      "loss": 5.186,
-      "step": 43560
-    },
-    {
-      "epoch": 22.71760104302477,
-      "grad_norm": 1.577545404434204,
-      "learning_rate": 5.673869346733669e-05,
-      "loss": 4.8958,
-      "step": 43561
-    },
-    {
-      "epoch": 22.71812255541069,
-      "grad_norm": 1.577321171760559,
-      "learning_rate": 5.673768844221106e-05,
-      "loss": 5.4797,
-      "step": 43562
-    },
-    {
-      "epoch": 22.71864406779661,
-      "grad_norm": 1.5836830139160156,
-      "learning_rate": 5.6736683417085436e-05,
-      "loss": 5.1977,
-      "step": 43563
-    },
-    {
-      "epoch": 22.71916558018253,
-      "grad_norm": 1.5391831398010254,
-      "learning_rate": 5.67356783919598e-05,
-      "loss": 5.2764,
-      "step": 43564
-    },
-    {
-      "epoch": 22.71968709256845,
-      "grad_norm": 1.477423071861267,
-      "learning_rate": 5.673467336683418e-05,
-      "loss": 5.5323,
-      "step": 43565
-    },
-    {
-      "epoch": 22.72020860495437,
-      "grad_norm": 1.504712700843811,
-      "learning_rate": 5.673366834170855e-05,
-      "loss": 5.3355,
-      "step": 43566
-    },
-    {
-      "epoch": 22.72073011734029,
-      "grad_norm": 1.5606032609939575,
-      "learning_rate": 5.673266331658291e-05,
-      "loss": 5.0343,
-      "step": 43567
-    },
-    {
-      "epoch": 22.721251629726208,
-      "grad_norm": 1.583422303199768,
-      "learning_rate": 5.673165829145729e-05,
-      "loss": 4.6195,
-      "step": 43568
-    },
-    {
-      "epoch": 22.721773142112124,
-      "grad_norm": 1.5231785774230957,
-      "learning_rate": 5.6730653266331655e-05,
-      "loss": 5.2844,
-      "step": 43569
-    },
-    {
-      "epoch": 22.722294654498043,
-      "grad_norm": 1.5133657455444336,
-      "learning_rate": 5.672964824120603e-05,
-      "loss": 4.6609,
-      "step": 43570
-    },
-    {
-      "epoch": 22.722816166883963,
-      "grad_norm": 1.4149539470672607,
-      "learning_rate": 5.6728643216080404e-05,
-      "loss": 5.6956,
-      "step": 43571
-    },
-    {
-      "epoch": 22.723337679269882,
-      "grad_norm": 1.5304936170578003,
-      "learning_rate": 5.672763819095478e-05,
-      "loss": 4.742,
-      "step": 43572
-    },
-    {
-      "epoch": 22.7238591916558,
-      "grad_norm": 1.5497275590896606,
-      "learning_rate": 5.6726633165829146e-05,
-      "loss": 5.1553,
-      "step": 43573
-    },
-    {
-      "epoch": 22.72438070404172,
-      "grad_norm": 1.5889712572097778,
-      "learning_rate": 5.6725628140703524e-05,
-      "loss": 4.9864,
-      "step": 43574
-    },
-    {
-      "epoch": 22.72490221642764,
-      "grad_norm": 1.5712617635726929,
-      "learning_rate": 5.672462311557789e-05,
-      "loss": 5.3212,
-      "step": 43575
-    },
-    {
-      "epoch": 22.72542372881356,
-      "grad_norm": 1.6376333236694336,
-      "learning_rate": 5.6723618090452266e-05,
-      "loss": 5.1262,
-      "step": 43576
-    },
-    {
-      "epoch": 22.72594524119948,
-      "grad_norm": 1.70259690284729,
-      "learning_rate": 5.672261306532664e-05,
-      "loss": 4.6448,
-      "step": 43577
-    },
-    {
-      "epoch": 22.7264667535854,
-      "grad_norm": 1.5344536304473877,
-      "learning_rate": 5.6721608040201015e-05,
-      "loss": 5.5073,
-      "step": 43578
-    },
-    {
-      "epoch": 22.72698826597132,
-      "grad_norm": 1.509818196296692,
-      "learning_rate": 5.672060301507538e-05,
-      "loss": 5.6096,
-      "step": 43579
-    },
-    {
-      "epoch": 22.727509778357238,
-      "grad_norm": 1.4955521821975708,
-      "learning_rate": 5.6719597989949743e-05,
-      "loss": 5.2584,
-      "step": 43580
-    },
-    {
-      "epoch": 22.728031290743154,
-      "grad_norm": 1.5107946395874023,
-      "learning_rate": 5.671859296482412e-05,
-      "loss": 5.6334,
-      "step": 43581
-    },
-    {
-      "epoch": 22.728552803129073,
-      "grad_norm": 1.4461408853530884,
-      "learning_rate": 5.671758793969849e-05,
-      "loss": 5.7721,
-      "step": 43582
-    },
-    {
-      "epoch": 22.729074315514993,
-      "grad_norm": 1.4458807706832886,
-      "learning_rate": 5.671658291457287e-05,
-      "loss": 5.1345,
-      "step": 43583
-    },
-    {
-      "epoch": 22.729595827900912,
-      "grad_norm": 1.6711066961288452,
-      "learning_rate": 5.6715577889447234e-05,
-      "loss": 5.0777,
-      "step": 43584
-    },
-    {
-      "epoch": 22.73011734028683,
-      "grad_norm": 1.4790852069854736,
-      "learning_rate": 5.671457286432161e-05,
-      "loss": 4.7271,
-      "step": 43585
-    },
-    {
-      "epoch": 22.73063885267275,
-      "grad_norm": 1.621751308441162,
-      "learning_rate": 5.671356783919598e-05,
-      "loss": 5.2924,
-      "step": 43586
-    },
-    {
-      "epoch": 22.73116036505867,
-      "grad_norm": 1.5565738677978516,
-      "learning_rate": 5.671256281407036e-05,
-      "loss": 5.1618,
-      "step": 43587
-    },
-    {
-      "epoch": 22.73168187744459,
-      "grad_norm": 1.584689974784851,
-      "learning_rate": 5.6711557788944725e-05,
-      "loss": 5.5078,
-      "step": 43588
-    },
-    {
-      "epoch": 22.73220338983051,
-      "grad_norm": 1.4949774742126465,
-      "learning_rate": 5.67105527638191e-05,
-      "loss": 5.4743,
-      "step": 43589
-    },
-    {
-      "epoch": 22.73272490221643,
-      "grad_norm": 1.613179326057434,
-      "learning_rate": 5.670954773869347e-05,
-      "loss": 5.5513,
-      "step": 43590
-    },
-    {
-      "epoch": 22.73324641460235,
-      "grad_norm": 1.5137758255004883,
-      "learning_rate": 5.6708542713567845e-05,
-      "loss": 5.2433,
-      "step": 43591
-    },
-    {
-      "epoch": 22.733767926988264,
-      "grad_norm": 1.6054010391235352,
-      "learning_rate": 5.6707537688442216e-05,
-      "loss": 4.5095,
-      "step": 43592
-    },
-    {
-      "epoch": 22.734289439374184,
-      "grad_norm": 1.4707938432693481,
-      "learning_rate": 5.670653266331658e-05,
-      "loss": 5.2525,
-      "step": 43593
-    },
-    {
-      "epoch": 22.734810951760103,
-      "grad_norm": 1.4152129888534546,
-      "learning_rate": 5.670552763819096e-05,
-      "loss": 5.447,
-      "step": 43594
-    },
-    {
-      "epoch": 22.735332464146023,
-      "grad_norm": 1.513128399848938,
-      "learning_rate": 5.670452261306532e-05,
-      "loss": 5.4371,
-      "step": 43595
-    },
-    {
-      "epoch": 22.735853976531942,
-      "grad_norm": 1.4943568706512451,
-      "learning_rate": 5.67035175879397e-05,
-      "loss": 5.5873,
-      "step": 43596
-    },
-    {
-      "epoch": 22.73637548891786,
-      "grad_norm": 1.648838996887207,
-      "learning_rate": 5.670251256281407e-05,
-      "loss": 4.5718,
-      "step": 43597
-    },
-    {
-      "epoch": 22.73689700130378,
-      "grad_norm": 1.4185198545455933,
-      "learning_rate": 5.670150753768845e-05,
-      "loss": 4.454,
-      "step": 43598
-    },
-    {
-      "epoch": 22.7374185136897,
-      "grad_norm": 1.6395576000213623,
-      "learning_rate": 5.6700502512562814e-05,
-      "loss": 5.5014,
-      "step": 43599
-    },
-    {
-      "epoch": 22.73794002607562,
-      "grad_norm": 1.5954861640930176,
-      "learning_rate": 5.669949748743719e-05,
-      "loss": 5.1225,
-      "step": 43600
-    },
-    {
-      "epoch": 22.73846153846154,
-      "grad_norm": 1.5396878719329834,
-      "learning_rate": 5.6698492462311556e-05,
-      "loss": 4.7217,
-      "step": 43601
-    },
-    {
-      "epoch": 22.73898305084746,
-      "grad_norm": 1.638129711151123,
-      "learning_rate": 5.6697487437185934e-05,
-      "loss": 4.7643,
-      "step": 43602
-    },
-    {
-      "epoch": 22.73950456323338,
-      "grad_norm": 1.5255956649780273,
-      "learning_rate": 5.6696482412060305e-05,
-      "loss": 5.3576,
-      "step": 43603
-    },
-    {
-      "epoch": 22.740026075619298,
-      "grad_norm": 1.4064580202102661,
-      "learning_rate": 5.669547738693468e-05,
-      "loss": 5.1621,
-      "step": 43604
-    },
-    {
-      "epoch": 22.740547588005214,
-      "grad_norm": 1.5537289381027222,
-      "learning_rate": 5.669447236180905e-05,
-      "loss": 5.1802,
-      "step": 43605
-    },
-    {
-      "epoch": 22.741069100391133,
-      "grad_norm": 1.5225088596343994,
-      "learning_rate": 5.669346733668342e-05,
-      "loss": 5.1985,
-      "step": 43606
-    },
-    {
-      "epoch": 22.741590612777053,
-      "grad_norm": 1.5258302688598633,
-      "learning_rate": 5.6692462311557796e-05,
-      "loss": 5.0387,
-      "step": 43607
-    },
-    {
-      "epoch": 22.742112125162972,
-      "grad_norm": 1.5152956247329712,
-      "learning_rate": 5.669145728643216e-05,
-      "loss": 5.3807,
-      "step": 43608
-    },
-    {
-      "epoch": 22.74263363754889,
-      "grad_norm": 1.544751763343811,
-      "learning_rate": 5.669045226130654e-05,
-      "loss": 4.9786,
-      "step": 43609
-    },
-    {
-      "epoch": 22.74315514993481,
-      "grad_norm": 1.5272228717803955,
-      "learning_rate": 5.66894472361809e-05,
-      "loss": 5.3058,
-      "step": 43610
-    },
-    {
-      "epoch": 22.74367666232073,
-      "grad_norm": 1.5379263162612915,
-      "learning_rate": 5.668844221105528e-05,
-      "loss": 5.6403,
-      "step": 43611
-    },
-    {
-      "epoch": 22.74419817470665,
-      "grad_norm": 1.498295545578003,
-      "learning_rate": 5.668743718592965e-05,
-      "loss": 5.4049,
-      "step": 43612
-    },
-    {
-      "epoch": 22.74471968709257,
-      "grad_norm": 1.626025676727295,
-      "learning_rate": 5.668643216080403e-05,
-      "loss": 4.6974,
-      "step": 43613
-    },
-    {
-      "epoch": 22.74524119947849,
-      "grad_norm": 1.5138968229293823,
-      "learning_rate": 5.668542713567839e-05,
-      "loss": 4.8628,
-      "step": 43614
-    },
-    {
-      "epoch": 22.74576271186441,
-      "grad_norm": 1.4963243007659912,
-      "learning_rate": 5.668442211055277e-05,
-      "loss": 5.403,
-      "step": 43615
-    },
-    {
-      "epoch": 22.746284224250324,
-      "grad_norm": 1.6218912601470947,
-      "learning_rate": 5.6683417085427135e-05,
-      "loss": 4.9604,
-      "step": 43616
-    },
-    {
-      "epoch": 22.746805736636244,
-      "grad_norm": 1.5605095624923706,
-      "learning_rate": 5.668241206030151e-05,
-      "loss": 5.2476,
-      "step": 43617
-    },
-    {
-      "epoch": 22.747327249022163,
-      "grad_norm": 1.641647458076477,
-      "learning_rate": 5.6681407035175884e-05,
-      "loss": 5.1801,
-      "step": 43618
-    },
-    {
-      "epoch": 22.747848761408083,
-      "grad_norm": 1.6971056461334229,
-      "learning_rate": 5.668040201005025e-05,
-      "loss": 4.7762,
-      "step": 43619
-    },
-    {
-      "epoch": 22.748370273794002,
-      "grad_norm": 1.4995695352554321,
-      "learning_rate": 5.6679396984924626e-05,
-      "loss": 5.3086,
-      "step": 43620
-    },
-    {
-      "epoch": 22.74889178617992,
-      "grad_norm": 1.5065323114395142,
-      "learning_rate": 5.667839195979899e-05,
-      "loss": 5.2079,
-      "step": 43621
-    },
-    {
-      "epoch": 22.74941329856584,
-      "grad_norm": 1.5797990560531616,
-      "learning_rate": 5.667738693467337e-05,
-      "loss": 4.6228,
-      "step": 43622
-    },
-    {
-      "epoch": 22.74993481095176,
-      "grad_norm": 1.6195157766342163,
-      "learning_rate": 5.667638190954774e-05,
-      "loss": 4.7835,
-      "step": 43623
-    },
-    {
-      "epoch": 22.75045632333768,
-      "grad_norm": 1.5439677238464355,
-      "learning_rate": 5.667537688442212e-05,
-      "loss": 5.42,
-      "step": 43624
-    },
-    {
-      "epoch": 22.7509778357236,
-      "grad_norm": 1.4853371381759644,
-      "learning_rate": 5.667437185929648e-05,
-      "loss": 5.4518,
-      "step": 43625
-    },
-    {
-      "epoch": 22.75149934810952,
-      "grad_norm": 1.675188422203064,
-      "learning_rate": 5.667336683417086e-05,
-      "loss": 4.5801,
-      "step": 43626
-    },
-    {
-      "epoch": 22.75202086049544,
-      "grad_norm": 1.5656832456588745,
-      "learning_rate": 5.6672361809045223e-05,
-      "loss": 4.7563,
-      "step": 43627
-    },
-    {
-      "epoch": 22.752542372881354,
-      "grad_norm": 1.5228904485702515,
-      "learning_rate": 5.66713567839196e-05,
-      "loss": 5.4426,
-      "step": 43628
-    },
-    {
-      "epoch": 22.753063885267274,
-      "grad_norm": 1.4836459159851074,
-      "learning_rate": 5.667035175879397e-05,
-      "loss": 5.2775,
-      "step": 43629
-    },
-    {
-      "epoch": 22.753585397653193,
-      "grad_norm": 1.4909400939941406,
-      "learning_rate": 5.666934673366835e-05,
-      "loss": 5.5038,
-      "step": 43630
-    },
-    {
-      "epoch": 22.754106910039113,
-      "grad_norm": 1.6502809524536133,
-      "learning_rate": 5.6668341708542714e-05,
-      "loss": 4.5246,
-      "step": 43631
-    },
-    {
-      "epoch": 22.754628422425032,
-      "grad_norm": 1.5173877477645874,
-      "learning_rate": 5.6667336683417086e-05,
-      "loss": 5.4248,
-      "step": 43632
-    },
-    {
-      "epoch": 22.75514993481095,
-      "grad_norm": 1.4898436069488525,
-      "learning_rate": 5.666633165829146e-05,
-      "loss": 5.3518,
-      "step": 43633
-    },
-    {
-      "epoch": 22.75567144719687,
-      "grad_norm": 1.4200117588043213,
-      "learning_rate": 5.666532663316583e-05,
-      "loss": 5.3838,
-      "step": 43634
-    },
-    {
-      "epoch": 22.75619295958279,
-      "grad_norm": 1.4129799604415894,
-      "learning_rate": 5.6664321608040205e-05,
-      "loss": 5.588,
-      "step": 43635
-    },
-    {
-      "epoch": 22.75671447196871,
-      "grad_norm": 1.5679535865783691,
-      "learning_rate": 5.666331658291457e-05,
-      "loss": 5.3686,
-      "step": 43636
-    },
-    {
-      "epoch": 22.75723598435463,
-      "grad_norm": 1.5806092023849487,
-      "learning_rate": 5.666231155778895e-05,
-      "loss": 5.3805,
-      "step": 43637
-    },
-    {
-      "epoch": 22.75775749674055,
-      "grad_norm": 1.495985507965088,
-      "learning_rate": 5.666130653266332e-05,
-      "loss": 5.2669,
-      "step": 43638
-    },
-    {
-      "epoch": 22.75827900912647,
-      "grad_norm": 1.5262519121170044,
-      "learning_rate": 5.6660301507537696e-05,
-      "loss": 5.2356,
-      "step": 43639
-    },
-    {
-      "epoch": 22.758800521512384,
-      "grad_norm": 1.499624252319336,
-      "learning_rate": 5.665929648241206e-05,
-      "loss": 5.5129,
-      "step": 43640
-    },
-    {
-      "epoch": 22.759322033898304,
-      "grad_norm": 1.5143710374832153,
-      "learning_rate": 5.665829145728644e-05,
-      "loss": 5.2874,
-      "step": 43641
-    },
-    {
-      "epoch": 22.759843546284223,
-      "grad_norm": 1.472136378288269,
-      "learning_rate": 5.66572864321608e-05,
-      "loss": 5.5729,
-      "step": 43642
-    },
-    {
-      "epoch": 22.760365058670143,
-      "grad_norm": 1.537054181098938,
-      "learning_rate": 5.665628140703518e-05,
-      "loss": 5.208,
-      "step": 43643
-    },
-    {
-      "epoch": 22.760886571056062,
-      "grad_norm": 1.6284784078598022,
-      "learning_rate": 5.665527638190955e-05,
-      "loss": 5.0598,
-      "step": 43644
-    },
-    {
-      "epoch": 22.76140808344198,
-      "grad_norm": 1.519278645515442,
-      "learning_rate": 5.665427135678393e-05,
-      "loss": 5.0009,
-      "step": 43645
-    },
-    {
-      "epoch": 22.7619295958279,
-      "grad_norm": 1.576280117034912,
-      "learning_rate": 5.6653266331658294e-05,
-      "loss": 5.037,
-      "step": 43646
-    },
-    {
-      "epoch": 22.76245110821382,
-      "grad_norm": 1.5863081216812134,
-      "learning_rate": 5.665226130653266e-05,
-      "loss": 5.021,
-      "step": 43647
-    },
-    {
-      "epoch": 22.76297262059974,
-      "grad_norm": 1.430732250213623,
-      "learning_rate": 5.6651256281407036e-05,
-      "loss": 5.5501,
-      "step": 43648
-    },
-    {
-      "epoch": 22.76349413298566,
-      "grad_norm": 1.6200342178344727,
-      "learning_rate": 5.665025125628141e-05,
-      "loss": 5.0242,
-      "step": 43649
-    },
-    {
-      "epoch": 22.76401564537158,
-      "grad_norm": 1.5858516693115234,
-      "learning_rate": 5.6649246231155785e-05,
-      "loss": 4.6684,
-      "step": 43650
-    },
-    {
-      "epoch": 22.7645371577575,
-      "grad_norm": 1.5747095346450806,
-      "learning_rate": 5.664824120603015e-05,
-      "loss": 5.3528,
-      "step": 43651
-    },
-    {
-      "epoch": 22.765058670143414,
-      "grad_norm": 1.570696234703064,
-      "learning_rate": 5.664723618090453e-05,
-      "loss": 5.0764,
-      "step": 43652
-    },
-    {
-      "epoch": 22.765580182529334,
-      "grad_norm": 1.5419296026229858,
-      "learning_rate": 5.66462311557789e-05,
-      "loss": 5.2731,
-      "step": 43653
-    },
-    {
-      "epoch": 22.766101694915253,
-      "grad_norm": 1.4512871503829956,
-      "learning_rate": 5.6645226130653276e-05,
-      "loss": 5.3827,
-      "step": 43654
-    },
-    {
-      "epoch": 22.766623207301173,
-      "grad_norm": 1.6451237201690674,
-      "learning_rate": 5.664422110552764e-05,
-      "loss": 5.1082,
-      "step": 43655
-    },
-    {
-      "epoch": 22.767144719687092,
-      "grad_norm": 1.506790280342102,
-      "learning_rate": 5.664321608040202e-05,
-      "loss": 4.8938,
-      "step": 43656
-    },
-    {
-      "epoch": 22.76766623207301,
-      "grad_norm": 1.4181106090545654,
-      "learning_rate": 5.664221105527638e-05,
-      "loss": 5.2883,
-      "step": 43657
-    },
-    {
-      "epoch": 22.76818774445893,
-      "grad_norm": 1.5718648433685303,
-      "learning_rate": 5.664120603015076e-05,
-      "loss": 5.6015,
-      "step": 43658
-    },
-    {
-      "epoch": 22.76870925684485,
-      "grad_norm": 1.5471014976501465,
-      "learning_rate": 5.664020100502513e-05,
-      "loss": 5.6134,
-      "step": 43659
-    },
-    {
-      "epoch": 22.76923076923077,
-      "grad_norm": 1.61711847782135,
-      "learning_rate": 5.6639195979899495e-05,
-      "loss": 5.2856,
-      "step": 43660
-    },
-    {
-      "epoch": 22.76975228161669,
-      "grad_norm": 1.5757273435592651,
-      "learning_rate": 5.663819095477387e-05,
-      "loss": 5.446,
-      "step": 43661
-    },
-    {
-      "epoch": 22.77027379400261,
-      "grad_norm": 1.520330786705017,
-      "learning_rate": 5.663718592964824e-05,
-      "loss": 5.016,
-      "step": 43662
-    },
-    {
-      "epoch": 22.77079530638853,
-      "grad_norm": 1.4801751375198364,
-      "learning_rate": 5.6636180904522615e-05,
-      "loss": 5.6165,
-      "step": 43663
-    },
-    {
-      "epoch": 22.771316818774444,
-      "grad_norm": 1.459517240524292,
-      "learning_rate": 5.6635175879396986e-05,
-      "loss": 5.5898,
-      "step": 43664
-    },
-    {
-      "epoch": 22.771838331160364,
-      "grad_norm": 1.4061470031738281,
-      "learning_rate": 5.6634170854271364e-05,
-      "loss": 4.627,
-      "step": 43665
-    },
-    {
-      "epoch": 22.772359843546283,
-      "grad_norm": 1.5996545553207397,
-      "learning_rate": 5.663316582914573e-05,
-      "loss": 5.1153,
-      "step": 43666
-    },
-    {
-      "epoch": 22.772881355932203,
-      "grad_norm": 1.5378884077072144,
-      "learning_rate": 5.6632160804020106e-05,
-      "loss": 5.0596,
-      "step": 43667
-    },
-    {
-      "epoch": 22.773402868318122,
-      "grad_norm": 1.655340552330017,
-      "learning_rate": 5.663115577889447e-05,
-      "loss": 5.1982,
-      "step": 43668
-    },
-    {
-      "epoch": 22.77392438070404,
-      "grad_norm": 1.4504578113555908,
-      "learning_rate": 5.663015075376885e-05,
-      "loss": 5.1665,
-      "step": 43669
-    },
-    {
-      "epoch": 22.77444589308996,
-      "grad_norm": 1.5471187829971313,
-      "learning_rate": 5.662914572864322e-05,
-      "loss": 5.3317,
-      "step": 43670
-    },
-    {
-      "epoch": 22.77496740547588,
-      "grad_norm": 1.6404417753219604,
-      "learning_rate": 5.66281407035176e-05,
-      "loss": 5.0919,
-      "step": 43671
-    },
-    {
-      "epoch": 22.7754889178618,
-      "grad_norm": 1.596103549003601,
-      "learning_rate": 5.662713567839196e-05,
-      "loss": 5.4668,
-      "step": 43672
-    },
-    {
-      "epoch": 22.77601043024772,
-      "grad_norm": 1.7144513130187988,
-      "learning_rate": 5.662613065326633e-05,
-      "loss": 4.876,
-      "step": 43673
-    },
-    {
-      "epoch": 22.77653194263364,
-      "grad_norm": 1.5253721475601196,
-      "learning_rate": 5.662512562814071e-05,
-      "loss": 5.3484,
-      "step": 43674
-    },
-    {
-      "epoch": 22.777053455019555,
-      "grad_norm": 1.4473891258239746,
-      "learning_rate": 5.6624120603015075e-05,
-      "loss": 5.5389,
-      "step": 43675
-    },
-    {
-      "epoch": 22.777574967405474,
-      "grad_norm": 1.6750437021255493,
-      "learning_rate": 5.662311557788945e-05,
-      "loss": 4.8897,
-      "step": 43676
-    },
-    {
-      "epoch": 22.778096479791394,
-      "grad_norm": 1.4849916696548462,
-      "learning_rate": 5.662211055276382e-05,
-      "loss": 4.7024,
-      "step": 43677
-    },
-    {
-      "epoch": 22.778617992177313,
-      "grad_norm": 1.4258695840835571,
-      "learning_rate": 5.6621105527638195e-05,
-      "loss": 5.734,
-      "step": 43678
-    },
-    {
-      "epoch": 22.779139504563233,
-      "grad_norm": 1.505090355873108,
-      "learning_rate": 5.6620100502512566e-05,
-      "loss": 5.2473,
-      "step": 43679
-    },
-    {
-      "epoch": 22.779661016949152,
-      "grad_norm": 1.5054527521133423,
-      "learning_rate": 5.6619095477386943e-05,
-      "loss": 5.366,
-      "step": 43680
-    },
-    {
-      "epoch": 22.78018252933507,
-      "grad_norm": 1.4825023412704468,
-      "learning_rate": 5.661809045226131e-05,
-      "loss": 5.4887,
-      "step": 43681
-    },
-    {
-      "epoch": 22.78070404172099,
-      "grad_norm": 1.4821866750717163,
-      "learning_rate": 5.6617085427135686e-05,
-      "loss": 5.0848,
-      "step": 43682
-    },
-    {
-      "epoch": 22.78122555410691,
-      "grad_norm": 1.5326170921325684,
-      "learning_rate": 5.661608040201005e-05,
-      "loss": 5.0806,
-      "step": 43683
-    },
-    {
-      "epoch": 22.78174706649283,
-      "grad_norm": 1.5626336336135864,
-      "learning_rate": 5.661507537688443e-05,
-      "loss": 5.3584,
-      "step": 43684
-    },
-    {
-      "epoch": 22.78226857887875,
-      "grad_norm": 1.6134955883026123,
-      "learning_rate": 5.66140703517588e-05,
-      "loss": 5.2807,
-      "step": 43685
-    },
-    {
-      "epoch": 22.78279009126467,
-      "grad_norm": 1.5278500318527222,
-      "learning_rate": 5.661306532663316e-05,
-      "loss": 5.6251,
-      "step": 43686
-    },
-    {
-      "epoch": 22.78331160365059,
-      "grad_norm": 1.418513298034668,
-      "learning_rate": 5.661206030150754e-05,
-      "loss": 5.3588,
-      "step": 43687
-    },
-    {
-      "epoch": 22.783833116036504,
-      "grad_norm": 1.490968108177185,
-      "learning_rate": 5.6611055276381905e-05,
-      "loss": 5.4776,
-      "step": 43688
-    },
-    {
-      "epoch": 22.784354628422424,
-      "grad_norm": 1.5030300617218018,
-      "learning_rate": 5.661005025125628e-05,
-      "loss": 5.7577,
-      "step": 43689
-    },
-    {
-      "epoch": 22.784876140808343,
-      "grad_norm": 1.8143651485443115,
-      "learning_rate": 5.6609045226130654e-05,
-      "loss": 4.2511,
-      "step": 43690
-    },
-    {
-      "epoch": 22.785397653194263,
-      "grad_norm": 1.4392682313919067,
-      "learning_rate": 5.660804020100503e-05,
-      "loss": 5.5684,
-      "step": 43691
-    },
-    {
-      "epoch": 22.785919165580182,
-      "grad_norm": 1.6126751899719238,
-      "learning_rate": 5.6607035175879396e-05,
-      "loss": 5.2665,
-      "step": 43692
-    },
-    {
-      "epoch": 22.7864406779661,
-      "grad_norm": 1.6236506700515747,
-      "learning_rate": 5.6606030150753774e-05,
-      "loss": 4.7849,
-      "step": 43693
-    },
-    {
-      "epoch": 22.78696219035202,
-      "grad_norm": 1.595902442932129,
-      "learning_rate": 5.6605025125628145e-05,
-      "loss": 5.1276,
-      "step": 43694
-    },
-    {
-      "epoch": 22.78748370273794,
-      "grad_norm": 1.487149953842163,
-      "learning_rate": 5.660402010050252e-05,
-      "loss": 5.5149,
-      "step": 43695
-    },
-    {
-      "epoch": 22.78800521512386,
-      "grad_norm": 1.501965880393982,
-      "learning_rate": 5.660301507537689e-05,
-      "loss": 5.1254,
-      "step": 43696
-    },
-    {
-      "epoch": 22.78852672750978,
-      "grad_norm": 1.4730650186538696,
-      "learning_rate": 5.6602010050251265e-05,
-      "loss": 5.3045,
-      "step": 43697
-    },
-    {
-      "epoch": 22.7890482398957,
-      "grad_norm": 1.5798925161361694,
-      "learning_rate": 5.660100502512563e-05,
-      "loss": 5.0667,
-      "step": 43698
-    },
-    {
-      "epoch": 22.789569752281615,
-      "grad_norm": 1.4892210960388184,
-      "learning_rate": 5.66e-05,
-      "loss": 5.5592,
-      "step": 43699
-    },
-    {
-      "epoch": 22.790091264667534,
-      "grad_norm": 1.6364331245422363,
-      "learning_rate": 5.659899497487438e-05,
-      "loss": 4.9781,
-      "step": 43700
-    },
-    {
-      "epoch": 22.790612777053454,
-      "grad_norm": 1.6389548778533936,
-      "learning_rate": 5.659798994974874e-05,
-      "loss": 5.5653,
-      "step": 43701
-    },
-    {
-      "epoch": 22.791134289439373,
-      "grad_norm": 1.5353292226791382,
-      "learning_rate": 5.659698492462312e-05,
-      "loss": 5.4065,
-      "step": 43702
-    },
-    {
-      "epoch": 22.791655801825293,
-      "grad_norm": 1.5009474754333496,
-      "learning_rate": 5.6595979899497484e-05,
-      "loss": 5.3319,
-      "step": 43703
-    },
-    {
-      "epoch": 22.792177314211212,
-      "grad_norm": 1.518190860748291,
-      "learning_rate": 5.659497487437186e-05,
-      "loss": 5.2438,
-      "step": 43704
-    },
-    {
-      "epoch": 22.79269882659713,
-      "grad_norm": 1.6329461336135864,
-      "learning_rate": 5.659396984924623e-05,
-      "loss": 4.9717,
-      "step": 43705
-    },
-    {
-      "epoch": 22.79322033898305,
-      "grad_norm": 1.4716770648956299,
-      "learning_rate": 5.659296482412061e-05,
-      "loss": 5.648,
-      "step": 43706
-    },
-    {
-      "epoch": 22.79374185136897,
-      "grad_norm": 1.5126428604125977,
-      "learning_rate": 5.6591959798994975e-05,
-      "loss": 5.3561,
-      "step": 43707
-    },
-    {
-      "epoch": 22.79426336375489,
-      "grad_norm": 1.528153657913208,
-      "learning_rate": 5.659095477386935e-05,
-      "loss": 4.6661,
-      "step": 43708
-    },
-    {
-      "epoch": 22.79478487614081,
-      "grad_norm": 1.6828837394714355,
-      "learning_rate": 5.658994974874372e-05,
-      "loss": 4.9548,
-      "step": 43709
-    },
-    {
-      "epoch": 22.79530638852673,
-      "grad_norm": 1.4357631206512451,
-      "learning_rate": 5.6588944723618095e-05,
-      "loss": 4.8507,
-      "step": 43710
-    },
-    {
-      "epoch": 22.795827900912645,
-      "grad_norm": 1.5777868032455444,
-      "learning_rate": 5.6587939698492466e-05,
-      "loss": 4.9381,
-      "step": 43711
-    },
-    {
-      "epoch": 22.796349413298564,
-      "grad_norm": 1.623633861541748,
-      "learning_rate": 5.658693467336683e-05,
-      "loss": 5.3188,
-      "step": 43712
-    },
-    {
-      "epoch": 22.796870925684484,
-      "grad_norm": 1.5988550186157227,
-      "learning_rate": 5.658592964824121e-05,
-      "loss": 5.5118,
-      "step": 43713
-    },
-    {
-      "epoch": 22.797392438070403,
-      "grad_norm": 1.6397204399108887,
-      "learning_rate": 5.658492462311557e-05,
-      "loss": 4.7462,
-      "step": 43714
-    },
-    {
-      "epoch": 22.797913950456323,
-      "grad_norm": 1.6496988534927368,
-      "learning_rate": 5.658391959798995e-05,
-      "loss": 5.0277,
-      "step": 43715
-    },
-    {
-      "epoch": 22.798435462842242,
-      "grad_norm": 1.6283150911331177,
-      "learning_rate": 5.658291457286432e-05,
-      "loss": 5.1489,
-      "step": 43716
-    },
-    {
-      "epoch": 22.798956975228162,
-      "grad_norm": 1.6909691095352173,
-      "learning_rate": 5.65819095477387e-05,
-      "loss": 5.1794,
-      "step": 43717
-    },
-    {
-      "epoch": 22.79947848761408,
-      "grad_norm": 1.5947465896606445,
-      "learning_rate": 5.6580904522613064e-05,
-      "loss": 4.8501,
-      "step": 43718
-    },
-    {
-      "epoch": 22.8,
-      "grad_norm": 1.5666594505310059,
-      "learning_rate": 5.657989949748744e-05,
-      "loss": 5.3232,
-      "step": 43719
-    },
-    {
-      "epoch": 22.80052151238592,
-      "grad_norm": 1.61076819896698,
-      "learning_rate": 5.657889447236181e-05,
-      "loss": 5.2836,
-      "step": 43720
-    },
-    {
-      "epoch": 22.80104302477184,
-      "grad_norm": 1.5377427339553833,
-      "learning_rate": 5.657788944723619e-05,
-      "loss": 5.4385,
-      "step": 43721
-    },
-    {
-      "epoch": 22.80156453715776,
-      "grad_norm": 1.6580294370651245,
-      "learning_rate": 5.6576884422110555e-05,
-      "loss": 5.3174,
-      "step": 43722
-    },
-    {
-      "epoch": 22.802086049543675,
-      "grad_norm": 1.5692241191864014,
-      "learning_rate": 5.657587939698493e-05,
-      "loss": 5.3678,
-      "step": 43723
-    },
-    {
-      "epoch": 22.802607561929594,
-      "grad_norm": 1.5294373035430908,
-      "learning_rate": 5.65748743718593e-05,
-      "loss": 5.4117,
-      "step": 43724
-    },
-    {
-      "epoch": 22.803129074315514,
-      "grad_norm": 1.6298224925994873,
-      "learning_rate": 5.657386934673367e-05,
-      "loss": 4.9828,
-      "step": 43725
-    },
-    {
-      "epoch": 22.803650586701433,
-      "grad_norm": 1.4418774843215942,
-      "learning_rate": 5.6572864321608046e-05,
-      "loss": 5.5381,
-      "step": 43726
-    },
-    {
-      "epoch": 22.804172099087353,
-      "grad_norm": 1.491703748703003,
-      "learning_rate": 5.657185929648241e-05,
-      "loss": 5.377,
-      "step": 43727
-    },
-    {
-      "epoch": 22.804693611473272,
-      "grad_norm": 1.4799867868423462,
-      "learning_rate": 5.657085427135679e-05,
-      "loss": 5.2699,
-      "step": 43728
-    },
-    {
-      "epoch": 22.805215123859192,
-      "grad_norm": 1.6676127910614014,
-      "learning_rate": 5.656984924623115e-05,
-      "loss": 4.9483,
-      "step": 43729
-    },
-    {
-      "epoch": 22.80573663624511,
-      "grad_norm": 1.5966219902038574,
-      "learning_rate": 5.656884422110553e-05,
-      "loss": 4.9256,
-      "step": 43730
-    },
-    {
-      "epoch": 22.80625814863103,
-      "grad_norm": 1.5287697315216064,
-      "learning_rate": 5.65678391959799e-05,
-      "loss": 4.6827,
-      "step": 43731
-    },
-    {
-      "epoch": 22.80677966101695,
-      "grad_norm": 1.5581183433532715,
-      "learning_rate": 5.656683417085428e-05,
-      "loss": 4.9505,
-      "step": 43732
-    },
-    {
-      "epoch": 22.80730117340287,
-      "grad_norm": 1.5147632360458374,
-      "learning_rate": 5.656582914572864e-05,
-      "loss": 5.4739,
-      "step": 43733
-    },
-    {
-      "epoch": 22.80782268578879,
-      "grad_norm": 1.501495361328125,
-      "learning_rate": 5.656482412060302e-05,
-      "loss": 5.2612,
-      "step": 43734
-    },
-    {
-      "epoch": 22.808344198174705,
-      "grad_norm": 1.5001682043075562,
-      "learning_rate": 5.6563819095477385e-05,
-      "loss": 5.5923,
-      "step": 43735
-    },
-    {
-      "epoch": 22.808865710560625,
-      "grad_norm": 1.4991084337234497,
-      "learning_rate": 5.656281407035176e-05,
-      "loss": 5.6115,
-      "step": 43736
-    },
-    {
-      "epoch": 22.809387222946544,
-      "grad_norm": 1.5743961334228516,
-      "learning_rate": 5.6561809045226134e-05,
-      "loss": 5.5638,
-      "step": 43737
-    },
-    {
-      "epoch": 22.809908735332463,
-      "grad_norm": 1.4118157625198364,
-      "learning_rate": 5.656080402010051e-05,
-      "loss": 5.5267,
-      "step": 43738
-    },
-    {
-      "epoch": 22.810430247718383,
-      "grad_norm": 1.4575284719467163,
-      "learning_rate": 5.6559798994974876e-05,
-      "loss": 5.3721,
-      "step": 43739
-    },
-    {
-      "epoch": 22.810951760104302,
-      "grad_norm": 1.5031465291976929,
-      "learning_rate": 5.655879396984925e-05,
-      "loss": 5.5379,
-      "step": 43740
-    },
-    {
-      "epoch": 22.811473272490222,
-      "grad_norm": 1.505236268043518,
-      "learning_rate": 5.6557788944723625e-05,
-      "loss": 5.6081,
-      "step": 43741
-    },
-    {
-      "epoch": 22.81199478487614,
-      "grad_norm": 1.6393318176269531,
-      "learning_rate": 5.655678391959799e-05,
-      "loss": 4.4834,
-      "step": 43742
-    },
-    {
-      "epoch": 22.81251629726206,
-      "grad_norm": 1.5184906721115112,
-      "learning_rate": 5.655577889447237e-05,
-      "loss": 5.1623,
-      "step": 43743
-    },
-    {
-      "epoch": 22.81303780964798,
-      "grad_norm": 1.4365428686141968,
-      "learning_rate": 5.655477386934673e-05,
-      "loss": 4.9359,
-      "step": 43744
-    },
-    {
-      "epoch": 22.8135593220339,
-      "grad_norm": 1.4715070724487305,
-      "learning_rate": 5.655376884422111e-05,
-      "loss": 5.5884,
-      "step": 43745
-    },
-    {
-      "epoch": 22.81408083441982,
-      "grad_norm": 1.5084565877914429,
-      "learning_rate": 5.655276381909548e-05,
-      "loss": 5.5499,
-      "step": 43746
-    },
-    {
-      "epoch": 22.814602346805735,
-      "grad_norm": 1.6595224142074585,
-      "learning_rate": 5.655175879396986e-05,
-      "loss": 4.6722,
-      "step": 43747
-    },
-    {
-      "epoch": 22.815123859191655,
-      "grad_norm": 1.5206230878829956,
-      "learning_rate": 5.655075376884422e-05,
-      "loss": 5.0612,
-      "step": 43748
-    },
-    {
-      "epoch": 22.815645371577574,
-      "grad_norm": 1.5778412818908691,
-      "learning_rate": 5.65497487437186e-05,
-      "loss": 5.3522,
-      "step": 43749
-    },
-    {
-      "epoch": 22.816166883963493,
-      "grad_norm": 1.5016103982925415,
-      "learning_rate": 5.6548743718592964e-05,
-      "loss": 4.654,
-      "step": 43750
-    },
-    {
-      "epoch": 22.816688396349413,
-      "grad_norm": 1.651724934577942,
-      "learning_rate": 5.654773869346734e-05,
-      "loss": 5.2623,
-      "step": 43751
-    },
-    {
-      "epoch": 22.817209908735332,
-      "grad_norm": 1.5032416582107544,
-      "learning_rate": 5.654673366834171e-05,
-      "loss": 4.9326,
-      "step": 43752
-    },
-    {
-      "epoch": 22.817731421121252,
-      "grad_norm": 1.735308051109314,
-      "learning_rate": 5.654572864321608e-05,
-      "loss": 4.4942,
-      "step": 43753
-    },
-    {
-      "epoch": 22.81825293350717,
-      "grad_norm": 1.5778172016143799,
-      "learning_rate": 5.6544723618090455e-05,
-      "loss": 5.4034,
-      "step": 43754
-    },
-    {
-      "epoch": 22.81877444589309,
-      "grad_norm": 1.4365259408950806,
-      "learning_rate": 5.654371859296482e-05,
-      "loss": 5.8424,
-      "step": 43755
-    },
-    {
-      "epoch": 22.81929595827901,
-      "grad_norm": 1.5548961162567139,
-      "learning_rate": 5.65427135678392e-05,
-      "loss": 5.5016,
-      "step": 43756
-    },
-    {
-      "epoch": 22.81981747066493,
-      "grad_norm": 1.593336582183838,
-      "learning_rate": 5.654170854271357e-05,
-      "loss": 5.1074,
-      "step": 43757
-    },
-    {
-      "epoch": 22.820338983050846,
-      "grad_norm": 1.566320538520813,
-      "learning_rate": 5.6540703517587946e-05,
-      "loss": 5.4542,
-      "step": 43758
-    },
-    {
-      "epoch": 22.820860495436765,
-      "grad_norm": 1.5194801092147827,
-      "learning_rate": 5.653969849246231e-05,
-      "loss": 5.8924,
-      "step": 43759
-    },
-    {
-      "epoch": 22.821382007822685,
-      "grad_norm": 1.6163641214370728,
-      "learning_rate": 5.653869346733669e-05,
-      "loss": 5.0887,
-      "step": 43760
-    },
-    {
-      "epoch": 22.821903520208604,
-      "grad_norm": 1.512014389038086,
-      "learning_rate": 5.653768844221106e-05,
-      "loss": 5.5713,
-      "step": 43761
-    },
-    {
-      "epoch": 22.822425032594523,
-      "grad_norm": 1.5809484720230103,
-      "learning_rate": 5.653668341708544e-05,
-      "loss": 5.3977,
-      "step": 43762
-    },
-    {
-      "epoch": 22.822946544980443,
-      "grad_norm": 1.5775283575057983,
-      "learning_rate": 5.65356783919598e-05,
-      "loss": 5.2115,
-      "step": 43763
-    },
-    {
-      "epoch": 22.823468057366362,
-      "grad_norm": 1.5577479600906372,
-      "learning_rate": 5.653467336683418e-05,
-      "loss": 5.2705,
-      "step": 43764
-    },
-    {
-      "epoch": 22.823989569752282,
-      "grad_norm": 1.497727870941162,
-      "learning_rate": 5.6533668341708544e-05,
-      "loss": 5.6414,
-      "step": 43765
-    },
-    {
-      "epoch": 22.8245110821382,
-      "grad_norm": 1.556006908416748,
-      "learning_rate": 5.6532663316582915e-05,
-      "loss": 5.4465,
-      "step": 43766
-    },
-    {
-      "epoch": 22.82503259452412,
-      "grad_norm": 1.4713867902755737,
-      "learning_rate": 5.653165829145729e-05,
-      "loss": 5.4795,
-      "step": 43767
-    },
-    {
-      "epoch": 22.82555410691004,
-      "grad_norm": 1.454753041267395,
-      "learning_rate": 5.653065326633166e-05,
-      "loss": 5.5017,
-      "step": 43768
-    },
-    {
-      "epoch": 22.82607561929596,
-      "grad_norm": 1.5429636240005493,
-      "learning_rate": 5.6529648241206035e-05,
-      "loss": 5.4768,
-      "step": 43769
-    },
-    {
-      "epoch": 22.82659713168188,
-      "grad_norm": 1.5849156379699707,
-      "learning_rate": 5.65286432160804e-05,
-      "loss": 5.3157,
-      "step": 43770
-    },
-    {
-      "epoch": 22.827118644067795,
-      "grad_norm": 1.5864230394363403,
-      "learning_rate": 5.652763819095478e-05,
-      "loss": 5.4186,
-      "step": 43771
-    },
-    {
-      "epoch": 22.827640156453715,
-      "grad_norm": 1.614202857017517,
-      "learning_rate": 5.652663316582915e-05,
-      "loss": 5.3282,
-      "step": 43772
-    },
-    {
-      "epoch": 22.828161668839634,
-      "grad_norm": 1.6227835416793823,
-      "learning_rate": 5.6525628140703526e-05,
-      "loss": 5.0294,
-      "step": 43773
-    },
-    {
-      "epoch": 22.828683181225554,
-      "grad_norm": 1.5998501777648926,
-      "learning_rate": 5.652462311557789e-05,
-      "loss": 5.3034,
-      "step": 43774
-    },
-    {
-      "epoch": 22.829204693611473,
-      "grad_norm": 1.5297520160675049,
-      "learning_rate": 5.652361809045227e-05,
-      "loss": 5.4418,
-      "step": 43775
-    },
-    {
-      "epoch": 22.829726205997392,
-      "grad_norm": 1.5936472415924072,
-      "learning_rate": 5.652261306532663e-05,
-      "loss": 4.991,
-      "step": 43776
-    },
-    {
-      "epoch": 22.830247718383312,
-      "grad_norm": 1.5370672941207886,
-      "learning_rate": 5.652160804020101e-05,
-      "loss": 4.7693,
-      "step": 43777
-    },
-    {
-      "epoch": 22.83076923076923,
-      "grad_norm": 1.5695016384124756,
-      "learning_rate": 5.652060301507538e-05,
-      "loss": 4.9633,
-      "step": 43778
-    },
-    {
-      "epoch": 22.83129074315515,
-      "grad_norm": 1.5994375944137573,
-      "learning_rate": 5.6519597989949745e-05,
-      "loss": 4.5993,
-      "step": 43779
-    },
-    {
-      "epoch": 22.83181225554107,
-      "grad_norm": 1.573333501815796,
-      "learning_rate": 5.651859296482412e-05,
-      "loss": 5.1097,
-      "step": 43780
-    },
-    {
-      "epoch": 22.83233376792699,
-      "grad_norm": 1.539019227027893,
-      "learning_rate": 5.6517587939698494e-05,
-      "loss": 4.6058,
-      "step": 43781
-    },
-    {
-      "epoch": 22.832855280312906,
-      "grad_norm": 1.5624301433563232,
-      "learning_rate": 5.651658291457287e-05,
-      "loss": 5.1848,
-      "step": 43782
-    },
-    {
-      "epoch": 22.833376792698825,
-      "grad_norm": 1.649697184562683,
-      "learning_rate": 5.6515577889447236e-05,
-      "loss": 5.0209,
-      "step": 43783
-    },
-    {
-      "epoch": 22.833898305084745,
-      "grad_norm": 1.5021121501922607,
-      "learning_rate": 5.6514572864321614e-05,
-      "loss": 5.3816,
-      "step": 43784
-    },
-    {
-      "epoch": 22.834419817470664,
-      "grad_norm": 1.494035005569458,
-      "learning_rate": 5.651356783919598e-05,
-      "loss": 5.485,
-      "step": 43785
-    },
-    {
-      "epoch": 22.834941329856584,
-      "grad_norm": 1.4962831735610962,
-      "learning_rate": 5.6512562814070356e-05,
-      "loss": 5.7114,
-      "step": 43786
-    },
-    {
-      "epoch": 22.835462842242503,
-      "grad_norm": 1.634596347808838,
-      "learning_rate": 5.651155778894473e-05,
-      "loss": 4.941,
-      "step": 43787
-    },
-    {
-      "epoch": 22.835984354628422,
-      "grad_norm": 1.51726496219635,
-      "learning_rate": 5.6510552763819105e-05,
-      "loss": 5.5643,
-      "step": 43788
-    },
-    {
-      "epoch": 22.836505867014342,
-      "grad_norm": 1.649924397468567,
-      "learning_rate": 5.650954773869347e-05,
-      "loss": 4.7783,
-      "step": 43789
-    },
-    {
-      "epoch": 22.83702737940026,
-      "grad_norm": 1.5116080045700073,
-      "learning_rate": 5.650854271356785e-05,
-      "loss": 5.5925,
-      "step": 43790
-    },
-    {
-      "epoch": 22.83754889178618,
-      "grad_norm": 1.5655758380889893,
-      "learning_rate": 5.650753768844221e-05,
-      "loss": 5.1,
-      "step": 43791
-    },
-    {
-      "epoch": 22.8380704041721,
-      "grad_norm": 1.6500061750411987,
-      "learning_rate": 5.650653266331658e-05,
-      "loss": 5.2737,
-      "step": 43792
-    },
-    {
-      "epoch": 22.83859191655802,
-      "grad_norm": 1.5976057052612305,
-      "learning_rate": 5.650552763819096e-05,
-      "loss": 4.888,
-      "step": 43793
-    },
-    {
-      "epoch": 22.839113428943936,
-      "grad_norm": 1.5810602903366089,
-      "learning_rate": 5.6504522613065325e-05,
-      "loss": 5.3982,
-      "step": 43794
-    },
-    {
-      "epoch": 22.839634941329855,
-      "grad_norm": 1.641796350479126,
-      "learning_rate": 5.65035175879397e-05,
-      "loss": 4.8946,
-      "step": 43795
-    },
-    {
-      "epoch": 22.840156453715775,
-      "grad_norm": 1.6535438299179077,
-      "learning_rate": 5.650251256281407e-05,
-      "loss": 5.1788,
-      "step": 43796
-    },
-    {
-      "epoch": 22.840677966101694,
-      "grad_norm": 1.4792399406433105,
-      "learning_rate": 5.6501507537688445e-05,
-      "loss": 5.6275,
-      "step": 43797
-    },
-    {
-      "epoch": 22.841199478487614,
-      "grad_norm": 1.5986177921295166,
-      "learning_rate": 5.6500502512562816e-05,
-      "loss": 5.2438,
-      "step": 43798
-    },
-    {
-      "epoch": 22.841720990873533,
-      "grad_norm": 1.6343178749084473,
-      "learning_rate": 5.6499497487437193e-05,
-      "loss": 5.2472,
-      "step": 43799
-    },
-    {
-      "epoch": 22.842242503259452,
-      "grad_norm": 1.5091255903244019,
-      "learning_rate": 5.649849246231156e-05,
-      "loss": 5.3155,
-      "step": 43800
-    },
-    {
-      "epoch": 22.842764015645372,
-      "grad_norm": 1.4162400960922241,
-      "learning_rate": 5.6497487437185936e-05,
-      "loss": 4.9852,
-      "step": 43801
-    },
-    {
-      "epoch": 22.84328552803129,
-      "grad_norm": 1.543320655822754,
-      "learning_rate": 5.64964824120603e-05,
-      "loss": 5.1768,
-      "step": 43802
-    },
-    {
-      "epoch": 22.84380704041721,
-      "grad_norm": 1.5972241163253784,
-      "learning_rate": 5.649547738693468e-05,
-      "loss": 4.4758,
-      "step": 43803
-    },
-    {
-      "epoch": 22.84432855280313,
-      "grad_norm": 1.4737979173660278,
-      "learning_rate": 5.649447236180905e-05,
-      "loss": 5.127,
-      "step": 43804
-    },
-    {
-      "epoch": 22.84485006518905,
-      "grad_norm": 1.430006742477417,
-      "learning_rate": 5.649346733668341e-05,
-      "loss": 5.5857,
-      "step": 43805
-    },
-    {
-      "epoch": 22.845371577574966,
-      "grad_norm": 1.587385654449463,
-      "learning_rate": 5.649246231155779e-05,
-      "loss": 5.192,
-      "step": 43806
-    },
-    {
-      "epoch": 22.845893089960885,
-      "grad_norm": 1.4399969577789307,
-      "learning_rate": 5.649145728643216e-05,
-      "loss": 5.5704,
-      "step": 43807
-    },
-    {
-      "epoch": 22.846414602346805,
-      "grad_norm": 1.637709140777588,
-      "learning_rate": 5.649045226130654e-05,
-      "loss": 5.2747,
-      "step": 43808
-    },
-    {
-      "epoch": 22.846936114732724,
-      "grad_norm": 1.5029748678207397,
-      "learning_rate": 5.6489447236180904e-05,
-      "loss": 5.0861,
-      "step": 43809
-    },
-    {
-      "epoch": 22.847457627118644,
-      "grad_norm": 1.5541677474975586,
-      "learning_rate": 5.648844221105528e-05,
-      "loss": 5.6417,
-      "step": 43810
-    },
-    {
-      "epoch": 22.847979139504563,
-      "grad_norm": 1.4661240577697754,
-      "learning_rate": 5.6487437185929646e-05,
-      "loss": 5.4102,
-      "step": 43811
-    },
-    {
-      "epoch": 22.848500651890483,
-      "grad_norm": 1.5125446319580078,
-      "learning_rate": 5.6486432160804024e-05,
-      "loss": 5.3428,
-      "step": 43812
-    },
-    {
-      "epoch": 22.849022164276402,
-      "grad_norm": 1.5155744552612305,
-      "learning_rate": 5.6485427135678395e-05,
-      "loss": 5.3729,
-      "step": 43813
-    },
-    {
-      "epoch": 22.84954367666232,
-      "grad_norm": 1.488637924194336,
-      "learning_rate": 5.648442211055277e-05,
-      "loss": 4.7946,
-      "step": 43814
-    },
-    {
-      "epoch": 22.85006518904824,
-      "grad_norm": 1.4626681804656982,
-      "learning_rate": 5.648341708542714e-05,
-      "loss": 5.5681,
-      "step": 43815
-    },
-    {
-      "epoch": 22.85058670143416,
-      "grad_norm": 1.6661564111709595,
-      "learning_rate": 5.6482412060301515e-05,
-      "loss": 5.1619,
-      "step": 43816
-    },
-    {
-      "epoch": 22.85110821382008,
-      "grad_norm": 1.4850658178329468,
-      "learning_rate": 5.648140703517588e-05,
-      "loss": 4.9396,
-      "step": 43817
-    },
-    {
-      "epoch": 22.851629726205996,
-      "grad_norm": 1.4683902263641357,
-      "learning_rate": 5.648040201005025e-05,
-      "loss": 5.4609,
-      "step": 43818
-    },
-    {
-      "epoch": 22.852151238591915,
-      "grad_norm": 1.5296398401260376,
-      "learning_rate": 5.647939698492463e-05,
-      "loss": 5.1031,
-      "step": 43819
-    },
-    {
-      "epoch": 22.852672750977835,
-      "grad_norm": 1.5345362424850464,
-      "learning_rate": 5.647839195979899e-05,
-      "loss": 5.4742,
-      "step": 43820
-    },
-    {
-      "epoch": 22.853194263363754,
-      "grad_norm": 1.5610917806625366,
-      "learning_rate": 5.647738693467337e-05,
-      "loss": 4.8429,
-      "step": 43821
-    },
-    {
-      "epoch": 22.853715775749674,
-      "grad_norm": 1.4760565757751465,
-      "learning_rate": 5.6476381909547734e-05,
-      "loss": 5.7243,
-      "step": 43822
-    },
-    {
-      "epoch": 22.854237288135593,
-      "grad_norm": 1.6025846004486084,
-      "learning_rate": 5.647537688442211e-05,
-      "loss": 5.3551,
-      "step": 43823
-    },
-    {
-      "epoch": 22.854758800521513,
-      "grad_norm": 1.5385838747024536,
-      "learning_rate": 5.647437185929648e-05,
-      "loss": 5.5729,
-      "step": 43824
-    },
-    {
-      "epoch": 22.855280312907432,
-      "grad_norm": 1.5963371992111206,
-      "learning_rate": 5.647336683417086e-05,
-      "loss": 5.3661,
-      "step": 43825
-    },
-    {
-      "epoch": 22.85580182529335,
-      "grad_norm": 1.447227120399475,
-      "learning_rate": 5.6472361809045225e-05,
-      "loss": 5.2161,
-      "step": 43826
-    },
-    {
-      "epoch": 22.85632333767927,
-      "grad_norm": 1.5842950344085693,
-      "learning_rate": 5.64713567839196e-05,
-      "loss": 5.4134,
-      "step": 43827
-    },
-    {
-      "epoch": 22.85684485006519,
-      "grad_norm": 1.5333112478256226,
-      "learning_rate": 5.6470351758793974e-05,
-      "loss": 4.8267,
-      "step": 43828
-    },
-    {
-      "epoch": 22.85736636245111,
-      "grad_norm": 1.6053402423858643,
-      "learning_rate": 5.646934673366835e-05,
-      "loss": 5.1661,
-      "step": 43829
-    },
-    {
-      "epoch": 22.857887874837026,
-      "grad_norm": 1.5621650218963623,
-      "learning_rate": 5.6468341708542716e-05,
-      "loss": 5.1015,
-      "step": 43830
-    },
-    {
-      "epoch": 22.858409387222945,
-      "grad_norm": 1.551958441734314,
-      "learning_rate": 5.6467336683417094e-05,
-      "loss": 5.4141,
-      "step": 43831
-    },
-    {
-      "epoch": 22.858930899608865,
-      "grad_norm": 1.679864525794983,
-      "learning_rate": 5.646633165829146e-05,
-      "loss": 4.7649,
-      "step": 43832
-    },
-    {
-      "epoch": 22.859452411994784,
-      "grad_norm": 1.5776410102844238,
-      "learning_rate": 5.646532663316583e-05,
-      "loss": 5.2124,
-      "step": 43833
-    },
-    {
-      "epoch": 22.859973924380704,
-      "grad_norm": 1.4428001642227173,
-      "learning_rate": 5.646432160804021e-05,
-      "loss": 5.602,
-      "step": 43834
-    },
-    {
-      "epoch": 22.860495436766623,
-      "grad_norm": 1.5516902208328247,
-      "learning_rate": 5.646331658291457e-05,
-      "loss": 5.3725,
-      "step": 43835
-    },
-    {
-      "epoch": 22.861016949152543,
-      "grad_norm": 1.505524754524231,
-      "learning_rate": 5.646231155778895e-05,
-      "loss": 5.2809,
-      "step": 43836
-    },
-    {
-      "epoch": 22.861538461538462,
-      "grad_norm": 1.484533429145813,
-      "learning_rate": 5.6461306532663314e-05,
-      "loss": 5.2707,
-      "step": 43837
-    },
-    {
-      "epoch": 22.86205997392438,
-      "grad_norm": 1.579887866973877,
-      "learning_rate": 5.646030150753769e-05,
-      "loss": 5.341,
-      "step": 43838
-    },
-    {
-      "epoch": 22.8625814863103,
-      "grad_norm": 1.4718706607818604,
-      "learning_rate": 5.645929648241206e-05,
-      "loss": 5.0232,
-      "step": 43839
-    },
-    {
-      "epoch": 22.86310299869622,
-      "grad_norm": 1.5966250896453857,
-      "learning_rate": 5.645829145728644e-05,
-      "loss": 5.0821,
-      "step": 43840
-    },
-    {
-      "epoch": 22.86362451108214,
-      "grad_norm": 1.5993702411651611,
-      "learning_rate": 5.6457286432160805e-05,
-      "loss": 5.2378,
-      "step": 43841
-    },
-    {
-      "epoch": 22.864146023468056,
-      "grad_norm": 1.5641024112701416,
-      "learning_rate": 5.645628140703518e-05,
-      "loss": 5.5057,
-      "step": 43842
-    },
-    {
-      "epoch": 22.864667535853975,
-      "grad_norm": 1.4936962127685547,
-      "learning_rate": 5.645527638190955e-05,
-      "loss": 5.0434,
-      "step": 43843
-    },
-    {
-      "epoch": 22.865189048239895,
-      "grad_norm": 1.6132298707962036,
-      "learning_rate": 5.6454271356783925e-05,
-      "loss": 4.9867,
-      "step": 43844
-    },
-    {
-      "epoch": 22.865710560625814,
-      "grad_norm": 1.5181260108947754,
-      "learning_rate": 5.6453266331658296e-05,
-      "loss": 4.7652,
-      "step": 43845
-    },
-    {
-      "epoch": 22.866232073011734,
-      "grad_norm": 1.5254075527191162,
-      "learning_rate": 5.645226130653266e-05,
-      "loss": 5.4906,
-      "step": 43846
-    },
-    {
-      "epoch": 22.866753585397653,
-      "grad_norm": 1.46632981300354,
-      "learning_rate": 5.645125628140704e-05,
-      "loss": 5.3874,
-      "step": 43847
-    },
-    {
-      "epoch": 22.867275097783573,
-      "grad_norm": 1.5208100080490112,
-      "learning_rate": 5.645025125628141e-05,
-      "loss": 5.1825,
-      "step": 43848
-    },
-    {
-      "epoch": 22.867796610169492,
-      "grad_norm": 1.5796253681182861,
-      "learning_rate": 5.644924623115579e-05,
-      "loss": 5.1691,
-      "step": 43849
-    },
-    {
-      "epoch": 22.86831812255541,
-      "grad_norm": 1.5706462860107422,
-      "learning_rate": 5.644824120603015e-05,
-      "loss": 5.3786,
-      "step": 43850
-    },
-    {
-      "epoch": 22.86883963494133,
-      "grad_norm": 1.5800679922103882,
-      "learning_rate": 5.644723618090453e-05,
-      "loss": 4.4353,
-      "step": 43851
-    },
-    {
-      "epoch": 22.86936114732725,
-      "grad_norm": 1.5463582277297974,
-      "learning_rate": 5.644623115577889e-05,
-      "loss": 5.4958,
-      "step": 43852
-    },
-    {
-      "epoch": 22.86988265971317,
-      "grad_norm": 1.491467833518982,
-      "learning_rate": 5.644522613065327e-05,
-      "loss": 5.4322,
-      "step": 43853
-    },
-    {
-      "epoch": 22.870404172099086,
-      "grad_norm": 1.4935115575790405,
-      "learning_rate": 5.644422110552764e-05,
-      "loss": 5.5245,
-      "step": 43854
-    },
-    {
-      "epoch": 22.870925684485005,
-      "grad_norm": 1.5543060302734375,
-      "learning_rate": 5.644321608040202e-05,
-      "loss": 5.0588,
-      "step": 43855
-    },
-    {
-      "epoch": 22.871447196870925,
-      "grad_norm": 1.5403547286987305,
-      "learning_rate": 5.6442211055276384e-05,
-      "loss": 4.8794,
-      "step": 43856
-    },
-    {
-      "epoch": 22.871968709256844,
-      "grad_norm": 1.495591640472412,
-      "learning_rate": 5.644120603015076e-05,
-      "loss": 4.8834,
-      "step": 43857
-    },
-    {
-      "epoch": 22.872490221642764,
-      "grad_norm": 1.5178899765014648,
-      "learning_rate": 5.6440201005025126e-05,
-      "loss": 5.5146,
-      "step": 43858
-    },
-    {
-      "epoch": 22.873011734028683,
-      "grad_norm": 1.4939416646957397,
-      "learning_rate": 5.64391959798995e-05,
-      "loss": 5.2657,
-      "step": 43859
-    },
-    {
-      "epoch": 22.873533246414603,
-      "grad_norm": 1.6456714868545532,
-      "learning_rate": 5.6438190954773875e-05,
-      "loss": 4.96,
-      "step": 43860
-    },
-    {
-      "epoch": 22.874054758800522,
-      "grad_norm": 1.5073128938674927,
-      "learning_rate": 5.643718592964824e-05,
-      "loss": 5.442,
-      "step": 43861
-    },
-    {
-      "epoch": 22.87457627118644,
-      "grad_norm": 1.436726450920105,
-      "learning_rate": 5.643618090452262e-05,
-      "loss": 5.3891,
-      "step": 43862
-    },
-    {
-      "epoch": 22.87509778357236,
-      "grad_norm": 1.4376827478408813,
-      "learning_rate": 5.643517587939698e-05,
-      "loss": 5.5528,
-      "step": 43863
-    },
-    {
-      "epoch": 22.87561929595828,
-      "grad_norm": 1.5143874883651733,
-      "learning_rate": 5.643417085427136e-05,
-      "loss": 5.5992,
-      "step": 43864
-    },
-    {
-      "epoch": 22.876140808344196,
-      "grad_norm": 1.540245532989502,
-      "learning_rate": 5.643316582914573e-05,
-      "loss": 5.3623,
-      "step": 43865
-    },
-    {
-      "epoch": 22.876662320730116,
-      "grad_norm": 1.5702929496765137,
-      "learning_rate": 5.643216080402011e-05,
-      "loss": 5.0962,
-      "step": 43866
-    },
-    {
-      "epoch": 22.877183833116035,
-      "grad_norm": 1.481192946434021,
-      "learning_rate": 5.643115577889447e-05,
-      "loss": 5.0792,
-      "step": 43867
-    },
-    {
-      "epoch": 22.877705345501955,
-      "grad_norm": 1.585952639579773,
-      "learning_rate": 5.643015075376885e-05,
-      "loss": 5.0503,
-      "step": 43868
-    },
-    {
-      "epoch": 22.878226857887874,
-      "grad_norm": 1.4856120347976685,
-      "learning_rate": 5.642914572864322e-05,
-      "loss": 5.5093,
-      "step": 43869
-    },
-    {
-      "epoch": 22.878748370273794,
-      "grad_norm": 1.5142191648483276,
-      "learning_rate": 5.64281407035176e-05,
-      "loss": 4.9326,
-      "step": 43870
-    },
-    {
-      "epoch": 22.879269882659713,
-      "grad_norm": 1.5204355716705322,
-      "learning_rate": 5.642713567839196e-05,
-      "loss": 5.3328,
-      "step": 43871
-    },
-    {
-      "epoch": 22.879791395045633,
-      "grad_norm": 1.5076828002929688,
-      "learning_rate": 5.642613065326633e-05,
-      "loss": 5.5636,
-      "step": 43872
-    },
-    {
-      "epoch": 22.880312907431552,
-      "grad_norm": 1.4129658937454224,
-      "learning_rate": 5.6425125628140705e-05,
-      "loss": 5.5727,
-      "step": 43873
-    },
-    {
-      "epoch": 22.88083441981747,
-      "grad_norm": 1.5949684381484985,
-      "learning_rate": 5.6424120603015077e-05,
-      "loss": 5.3183,
-      "step": 43874
-    },
-    {
-      "epoch": 22.88135593220339,
-      "grad_norm": 1.3961563110351562,
-      "learning_rate": 5.6423115577889454e-05,
-      "loss": 5.6692,
-      "step": 43875
-    },
-    {
-      "epoch": 22.88187744458931,
-      "grad_norm": 1.5556163787841797,
-      "learning_rate": 5.642211055276382e-05,
-      "loss": 4.9951,
-      "step": 43876
-    },
-    {
-      "epoch": 22.88239895697523,
-      "grad_norm": 1.5672740936279297,
-      "learning_rate": 5.6421105527638196e-05,
-      "loss": 4.9473,
-      "step": 43877
-    },
-    {
-      "epoch": 22.882920469361146,
-      "grad_norm": 1.604331135749817,
-      "learning_rate": 5.642010050251256e-05,
-      "loss": 5.255,
-      "step": 43878
-    },
-    {
-      "epoch": 22.883441981747065,
-      "grad_norm": 1.6034636497497559,
-      "learning_rate": 5.641909547738694e-05,
-      "loss": 4.9468,
-      "step": 43879
-    },
-    {
-      "epoch": 22.883963494132985,
-      "grad_norm": 1.5386732816696167,
-      "learning_rate": 5.641809045226131e-05,
-      "loss": 5.6785,
-      "step": 43880
-    },
-    {
-      "epoch": 22.884485006518904,
-      "grad_norm": 1.488684892654419,
-      "learning_rate": 5.641708542713569e-05,
-      "loss": 5.2526,
-      "step": 43881
-    },
-    {
-      "epoch": 22.885006518904824,
-      "grad_norm": 1.5124902725219727,
-      "learning_rate": 5.641608040201005e-05,
-      "loss": 5.2503,
-      "step": 43882
-    },
-    {
-      "epoch": 22.885528031290743,
-      "grad_norm": 1.6074758768081665,
-      "learning_rate": 5.641507537688443e-05,
-      "loss": 5.2478,
-      "step": 43883
-    },
-    {
-      "epoch": 22.886049543676663,
-      "grad_norm": 1.5183216333389282,
-      "learning_rate": 5.6414070351758794e-05,
-      "loss": 5.7435,
-      "step": 43884
-    },
-    {
-      "epoch": 22.886571056062582,
-      "grad_norm": 1.5744880437850952,
-      "learning_rate": 5.6413065326633165e-05,
-      "loss": 5.2411,
-      "step": 43885
-    },
-    {
-      "epoch": 22.8870925684485,
-      "grad_norm": 1.7396072149276733,
-      "learning_rate": 5.641206030150754e-05,
-      "loss": 5.1458,
-      "step": 43886
-    },
-    {
-      "epoch": 22.88761408083442,
-      "grad_norm": 1.6306849718093872,
-      "learning_rate": 5.641105527638191e-05,
-      "loss": 5.313,
-      "step": 43887
-    },
-    {
-      "epoch": 22.88813559322034,
-      "grad_norm": 1.620241403579712,
-      "learning_rate": 5.6410050251256285e-05,
-      "loss": 5.2111,
-      "step": 43888
-    },
-    {
-      "epoch": 22.888657105606256,
-      "grad_norm": 1.5316517353057861,
-      "learning_rate": 5.640904522613065e-05,
-      "loss": 5.7275,
-      "step": 43889
-    },
-    {
-      "epoch": 22.889178617992176,
-      "grad_norm": 1.5592721700668335,
-      "learning_rate": 5.640804020100503e-05,
-      "loss": 5.3795,
-      "step": 43890
-    },
-    {
-      "epoch": 22.889700130378095,
-      "grad_norm": 1.5572409629821777,
-      "learning_rate": 5.64070351758794e-05,
-      "loss": 5.2895,
-      "step": 43891
-    },
-    {
-      "epoch": 22.890221642764015,
-      "grad_norm": 1.583119511604309,
-      "learning_rate": 5.6406030150753776e-05,
-      "loss": 5.2647,
-      "step": 43892
-    },
-    {
-      "epoch": 22.890743155149934,
-      "grad_norm": 1.5672944784164429,
-      "learning_rate": 5.640502512562814e-05,
-      "loss": 5.0594,
-      "step": 43893
-    },
-    {
-      "epoch": 22.891264667535854,
-      "grad_norm": 1.6076546907424927,
-      "learning_rate": 5.640402010050252e-05,
-      "loss": 5.2662,
-      "step": 43894
-    },
-    {
-      "epoch": 22.891786179921773,
-      "grad_norm": 1.592577576637268,
-      "learning_rate": 5.640301507537689e-05,
-      "loss": 5.4113,
-      "step": 43895
-    },
-    {
-      "epoch": 22.892307692307693,
-      "grad_norm": 1.5166155099868774,
-      "learning_rate": 5.640201005025127e-05,
-      "loss": 5.7303,
-      "step": 43896
-    },
-    {
-      "epoch": 22.892829204693612,
-      "grad_norm": 1.6370842456817627,
-      "learning_rate": 5.640100502512563e-05,
-      "loss": 5.0385,
-      "step": 43897
-    },
-    {
-      "epoch": 22.89335071707953,
-      "grad_norm": 1.5847370624542236,
-      "learning_rate": 5.6399999999999995e-05,
-      "loss": 5.3656,
-      "step": 43898
-    },
-    {
-      "epoch": 22.89387222946545,
-      "grad_norm": 1.5062470436096191,
-      "learning_rate": 5.639899497487437e-05,
-      "loss": 5.2647,
-      "step": 43899
-    },
-    {
-      "epoch": 22.89439374185137,
-      "grad_norm": 1.5968453884124756,
-      "learning_rate": 5.6397989949748744e-05,
-      "loss": 5.1529,
-      "step": 43900
-    },
-    {
-      "epoch": 22.894915254237286,
-      "grad_norm": 1.5363004207611084,
-      "learning_rate": 5.639698492462312e-05,
-      "loss": 5.5286,
-      "step": 43901
-    },
-    {
-      "epoch": 22.895436766623206,
-      "grad_norm": 1.585430383682251,
-      "learning_rate": 5.6395979899497486e-05,
-      "loss": 5.0544,
-      "step": 43902
-    },
-    {
-      "epoch": 22.895958279009125,
-      "grad_norm": 1.5483131408691406,
-      "learning_rate": 5.6394974874371864e-05,
-      "loss": 5.1722,
-      "step": 43903
-    },
-    {
-      "epoch": 22.896479791395045,
-      "grad_norm": 1.5975139141082764,
-      "learning_rate": 5.639396984924623e-05,
-      "loss": 4.9969,
-      "step": 43904
-    },
-    {
-      "epoch": 22.897001303780964,
-      "grad_norm": 1.6602686643600464,
-      "learning_rate": 5.6392964824120606e-05,
-      "loss": 4.9658,
-      "step": 43905
-    },
-    {
-      "epoch": 22.897522816166884,
-      "grad_norm": 1.5871829986572266,
-      "learning_rate": 5.639195979899498e-05,
-      "loss": 4.8261,
-      "step": 43906
-    },
-    {
-      "epoch": 22.898044328552803,
-      "grad_norm": 1.5655676126480103,
-      "learning_rate": 5.6390954773869355e-05,
-      "loss": 4.9143,
-      "step": 43907
-    },
-    {
-      "epoch": 22.898565840938723,
-      "grad_norm": 1.5854569673538208,
-      "learning_rate": 5.638994974874372e-05,
-      "loss": 5.7372,
-      "step": 43908
-    },
-    {
-      "epoch": 22.899087353324642,
-      "grad_norm": 1.5272400379180908,
-      "learning_rate": 5.63889447236181e-05,
-      "loss": 5.1663,
-      "step": 43909
-    },
-    {
-      "epoch": 22.89960886571056,
-      "grad_norm": 1.4531322717666626,
-      "learning_rate": 5.638793969849246e-05,
-      "loss": 5.0429,
-      "step": 43910
-    },
-    {
-      "epoch": 22.90013037809648,
-      "grad_norm": 1.4522771835327148,
-      "learning_rate": 5.638693467336683e-05,
-      "loss": 5.7825,
-      "step": 43911
-    },
-    {
-      "epoch": 22.9006518904824,
-      "grad_norm": 1.5248713493347168,
-      "learning_rate": 5.638592964824121e-05,
-      "loss": 5.5188,
-      "step": 43912
-    },
-    {
-      "epoch": 22.901173402868316,
-      "grad_norm": 1.6154943704605103,
-      "learning_rate": 5.6384924623115575e-05,
-      "loss": 5.5475,
-      "step": 43913
-    },
-    {
-      "epoch": 22.901694915254236,
-      "grad_norm": 1.6890860795974731,
-      "learning_rate": 5.638391959798995e-05,
-      "loss": 4.4913,
-      "step": 43914
-    },
-    {
-      "epoch": 22.902216427640155,
-      "grad_norm": 1.8384783267974854,
-      "learning_rate": 5.6382914572864324e-05,
-      "loss": 4.8023,
-      "step": 43915
-    },
-    {
-      "epoch": 22.902737940026075,
-      "grad_norm": 1.6328279972076416,
-      "learning_rate": 5.63819095477387e-05,
-      "loss": 4.7488,
-      "step": 43916
-    },
-    {
-      "epoch": 22.903259452411994,
-      "grad_norm": 1.5609350204467773,
-      "learning_rate": 5.6380904522613066e-05,
-      "loss": 5.459,
-      "step": 43917
-    },
-    {
-      "epoch": 22.903780964797914,
-      "grad_norm": 1.6764976978302002,
-      "learning_rate": 5.6379899497487443e-05,
-      "loss": 5.4223,
-      "step": 43918
-    },
-    {
-      "epoch": 22.904302477183833,
-      "grad_norm": 1.5950367450714111,
-      "learning_rate": 5.637889447236181e-05,
-      "loss": 5.0014,
-      "step": 43919
-    },
-    {
-      "epoch": 22.904823989569753,
-      "grad_norm": 1.7364130020141602,
-      "learning_rate": 5.6377889447236186e-05,
-      "loss": 4.8181,
-      "step": 43920
-    },
-    {
-      "epoch": 22.905345501955672,
-      "grad_norm": 1.6436392068862915,
-      "learning_rate": 5.637688442211056e-05,
-      "loss": 4.9525,
-      "step": 43921
-    },
-    {
-      "epoch": 22.90586701434159,
-      "grad_norm": 1.6760200262069702,
-      "learning_rate": 5.6375879396984934e-05,
-      "loss": 5.0236,
-      "step": 43922
-    },
-    {
-      "epoch": 22.90638852672751,
-      "grad_norm": 1.6810145378112793,
-      "learning_rate": 5.63748743718593e-05,
-      "loss": 5.2108,
-      "step": 43923
-    },
-    {
-      "epoch": 22.90691003911343,
-      "grad_norm": 1.5254544019699097,
-      "learning_rate": 5.6373869346733677e-05,
-      "loss": 5.1979,
-      "step": 43924
-    },
-    {
-      "epoch": 22.907431551499347,
-      "grad_norm": 1.5837843418121338,
-      "learning_rate": 5.637286432160804e-05,
-      "loss": 5.0538,
-      "step": 43925
-    },
-    {
-      "epoch": 22.907953063885266,
-      "grad_norm": 1.6617910861968994,
-      "learning_rate": 5.637185929648241e-05,
-      "loss": 5.2948,
-      "step": 43926
-    },
-    {
-      "epoch": 22.908474576271185,
-      "grad_norm": 1.5291924476623535,
-      "learning_rate": 5.637085427135679e-05,
-      "loss": 5.3043,
-      "step": 43927
-    },
-    {
-      "epoch": 22.908996088657105,
-      "grad_norm": 1.5804210901260376,
-      "learning_rate": 5.6369849246231154e-05,
-      "loss": 5.1193,
-      "step": 43928
-    },
-    {
-      "epoch": 22.909517601043024,
-      "grad_norm": 1.6330556869506836,
-      "learning_rate": 5.636884422110553e-05,
-      "loss": 4.9202,
-      "step": 43929
-    },
-    {
-      "epoch": 22.910039113428944,
-      "grad_norm": 1.4870970249176025,
-      "learning_rate": 5.6367839195979896e-05,
-      "loss": 5.3474,
-      "step": 43930
-    },
-    {
-      "epoch": 22.910560625814863,
-      "grad_norm": 1.579213261604309,
-      "learning_rate": 5.6366834170854274e-05,
-      "loss": 5.1351,
-      "step": 43931
-    },
-    {
-      "epoch": 22.911082138200783,
-      "grad_norm": 1.581527829170227,
-      "learning_rate": 5.6365829145728645e-05,
-      "loss": 4.9337,
-      "step": 43932
-    },
-    {
-      "epoch": 22.911603650586702,
-      "grad_norm": 1.6781792640686035,
-      "learning_rate": 5.636482412060302e-05,
-      "loss": 5.3971,
-      "step": 43933
-    },
-    {
-      "epoch": 22.91212516297262,
-      "grad_norm": 1.5742840766906738,
-      "learning_rate": 5.636381909547739e-05,
-      "loss": 4.7718,
-      "step": 43934
-    },
-    {
-      "epoch": 22.91264667535854,
-      "grad_norm": 1.512369155883789,
-      "learning_rate": 5.6362814070351765e-05,
-      "loss": 4.5918,
-      "step": 43935
-    },
-    {
-      "epoch": 22.91316818774446,
-      "grad_norm": 1.4804518222808838,
-      "learning_rate": 5.6361809045226136e-05,
-      "loss": 4.9479,
-      "step": 43936
-    },
-    {
-      "epoch": 22.913689700130377,
-      "grad_norm": 1.6695044040679932,
-      "learning_rate": 5.6360804020100514e-05,
-      "loss": 4.3998,
-      "step": 43937
-    },
-    {
-      "epoch": 22.914211212516296,
-      "grad_norm": 1.6352910995483398,
-      "learning_rate": 5.635979899497488e-05,
-      "loss": 4.7448,
-      "step": 43938
-    },
-    {
-      "epoch": 22.914732724902215,
-      "grad_norm": 1.617153286933899,
-      "learning_rate": 5.635879396984924e-05,
-      "loss": 5.0969,
-      "step": 43939
-    },
-    {
-      "epoch": 22.915254237288135,
-      "grad_norm": 1.6170361042022705,
-      "learning_rate": 5.635778894472362e-05,
-      "loss": 4.6339,
-      "step": 43940
-    },
-    {
-      "epoch": 22.915775749674054,
-      "grad_norm": 1.722010850906372,
-      "learning_rate": 5.635678391959799e-05,
-      "loss": 4.8016,
-      "step": 43941
-    },
-    {
-      "epoch": 22.916297262059974,
-      "grad_norm": 1.5365633964538574,
-      "learning_rate": 5.635577889447237e-05,
-      "loss": 5.1739,
-      "step": 43942
-    },
-    {
-      "epoch": 22.916818774445893,
-      "grad_norm": 1.5213146209716797,
-      "learning_rate": 5.635477386934673e-05,
-      "loss": 5.5579,
-      "step": 43943
-    },
-    {
-      "epoch": 22.917340286831813,
-      "grad_norm": 1.475122094154358,
-      "learning_rate": 5.635376884422111e-05,
-      "loss": 5.6066,
-      "step": 43944
-    },
-    {
-      "epoch": 22.917861799217732,
-      "grad_norm": 1.663979411125183,
-      "learning_rate": 5.6352763819095475e-05,
-      "loss": 4.6066,
-      "step": 43945
-    },
-    {
-      "epoch": 22.91838331160365,
-      "grad_norm": 1.420501708984375,
-      "learning_rate": 5.635175879396985e-05,
-      "loss": 5.5343,
-      "step": 43946
-    },
-    {
-      "epoch": 22.91890482398957,
-      "grad_norm": 1.6265175342559814,
-      "learning_rate": 5.6350753768844224e-05,
-      "loss": 5.3059,
-      "step": 43947
-    },
-    {
-      "epoch": 22.919426336375487,
-      "grad_norm": 1.5129711627960205,
-      "learning_rate": 5.63497487437186e-05,
-      "loss": 4.9467,
-      "step": 43948
-    },
-    {
-      "epoch": 22.919947848761407,
-      "grad_norm": 1.6144676208496094,
-      "learning_rate": 5.6348743718592966e-05,
-      "loss": 4.7979,
-      "step": 43949
-    },
-    {
-      "epoch": 22.920469361147326,
-      "grad_norm": 1.5220717191696167,
-      "learning_rate": 5.6347738693467344e-05,
-      "loss": 5.3558,
-      "step": 43950
-    },
-    {
-      "epoch": 22.920990873533245,
-      "grad_norm": 1.6656690835952759,
-      "learning_rate": 5.634673366834171e-05,
-      "loss": 5.1728,
-      "step": 43951
-    },
-    {
-      "epoch": 22.921512385919165,
-      "grad_norm": 1.6289803981781006,
-      "learning_rate": 5.634572864321608e-05,
-      "loss": 5.3415,
-      "step": 43952
-    },
-    {
-      "epoch": 22.922033898305084,
-      "grad_norm": 1.6315346956253052,
-      "learning_rate": 5.634472361809046e-05,
-      "loss": 4.6387,
-      "step": 43953
-    },
-    {
-      "epoch": 22.922555410691004,
-      "grad_norm": 1.6612775325775146,
-      "learning_rate": 5.634371859296482e-05,
-      "loss": 5.735,
-      "step": 43954
-    },
-    {
-      "epoch": 22.923076923076923,
-      "grad_norm": 1.5979325771331787,
-      "learning_rate": 5.63427135678392e-05,
-      "loss": 5.1228,
-      "step": 43955
-    },
-    {
-      "epoch": 22.923598435462843,
-      "grad_norm": 1.5397437810897827,
-      "learning_rate": 5.6341708542713564e-05,
-      "loss": 5.4637,
-      "step": 43956
-    },
-    {
-      "epoch": 22.924119947848762,
-      "grad_norm": 1.5566164255142212,
-      "learning_rate": 5.634070351758794e-05,
-      "loss": 5.0734,
-      "step": 43957
-    },
-    {
-      "epoch": 22.92464146023468,
-      "grad_norm": 1.628625512123108,
-      "learning_rate": 5.633969849246231e-05,
-      "loss": 4.4375,
-      "step": 43958
-    },
-    {
-      "epoch": 22.9251629726206,
-      "grad_norm": 1.5245404243469238,
-      "learning_rate": 5.633869346733669e-05,
-      "loss": 5.4129,
-      "step": 43959
-    },
-    {
-      "epoch": 22.92568448500652,
-      "grad_norm": 1.636980652809143,
-      "learning_rate": 5.6337688442211055e-05,
-      "loss": 5.0094,
-      "step": 43960
-    },
-    {
-      "epoch": 22.926205997392437,
-      "grad_norm": 1.50666081905365,
-      "learning_rate": 5.633668341708543e-05,
-      "loss": 5.3941,
-      "step": 43961
-    },
-    {
-      "epoch": 22.926727509778356,
-      "grad_norm": 1.6201801300048828,
-      "learning_rate": 5.6335678391959804e-05,
-      "loss": 4.8278,
-      "step": 43962
-    },
-    {
-      "epoch": 22.927249022164276,
-      "grad_norm": 1.4927394390106201,
-      "learning_rate": 5.633467336683418e-05,
-      "loss": 5.0763,
-      "step": 43963
-    },
-    {
-      "epoch": 22.927770534550195,
-      "grad_norm": 1.4961588382720947,
-      "learning_rate": 5.6333668341708546e-05,
-      "loss": 5.1818,
-      "step": 43964
-    },
-    {
-      "epoch": 22.928292046936114,
-      "grad_norm": 1.429048776626587,
-      "learning_rate": 5.633266331658291e-05,
-      "loss": 5.0859,
-      "step": 43965
-    },
-    {
-      "epoch": 22.928813559322034,
-      "grad_norm": 1.5219444036483765,
-      "learning_rate": 5.633165829145729e-05,
-      "loss": 5.1233,
-      "step": 43966
-    },
-    {
-      "epoch": 22.929335071707953,
-      "grad_norm": 1.5636223554611206,
-      "learning_rate": 5.633065326633166e-05,
-      "loss": 5.2895,
-      "step": 43967
-    },
-    {
-      "epoch": 22.929856584093873,
-      "grad_norm": 1.5056648254394531,
-      "learning_rate": 5.632964824120604e-05,
-      "loss": 5.181,
-      "step": 43968
-    },
-    {
-      "epoch": 22.930378096479792,
-      "grad_norm": 1.5227999687194824,
-      "learning_rate": 5.63286432160804e-05,
-      "loss": 5.3637,
-      "step": 43969
-    },
-    {
-      "epoch": 22.93089960886571,
-      "grad_norm": 1.4446873664855957,
-      "learning_rate": 5.632763819095478e-05,
-      "loss": 5.3276,
-      "step": 43970
-    },
-    {
-      "epoch": 22.93142112125163,
-      "grad_norm": 1.3778117895126343,
-      "learning_rate": 5.632663316582914e-05,
-      "loss": 5.653,
-      "step": 43971
-    },
-    {
-      "epoch": 22.931942633637547,
-      "grad_norm": 1.5520516633987427,
-      "learning_rate": 5.632562814070352e-05,
-      "loss": 5.1753,
-      "step": 43972
-    },
-    {
-      "epoch": 22.932464146023467,
-      "grad_norm": 1.7097194194793701,
-      "learning_rate": 5.632462311557789e-05,
-      "loss": 4.4798,
-      "step": 43973
-    },
-    {
-      "epoch": 22.932985658409386,
-      "grad_norm": 1.4768521785736084,
-      "learning_rate": 5.632361809045227e-05,
-      "loss": 4.8343,
-      "step": 43974
-    },
-    {
-      "epoch": 22.933507170795306,
-      "grad_norm": 1.5331894159317017,
-      "learning_rate": 5.6322613065326634e-05,
-      "loss": 5.5059,
-      "step": 43975
-    },
-    {
-      "epoch": 22.934028683181225,
-      "grad_norm": 1.5519486665725708,
-      "learning_rate": 5.632160804020101e-05,
-      "loss": 4.948,
-      "step": 43976
-    },
-    {
-      "epoch": 22.934550195567144,
-      "grad_norm": 1.53237783908844,
-      "learning_rate": 5.6320603015075376e-05,
-      "loss": 5.1662,
-      "step": 43977
-    },
-    {
-      "epoch": 22.935071707953064,
-      "grad_norm": 1.5506590604782104,
-      "learning_rate": 5.631959798994975e-05,
-      "loss": 5.3458,
-      "step": 43978
-    },
-    {
-      "epoch": 22.935593220338983,
-      "grad_norm": 1.5153002738952637,
-      "learning_rate": 5.6318592964824125e-05,
-      "loss": 5.4112,
-      "step": 43979
-    },
-    {
-      "epoch": 22.936114732724903,
-      "grad_norm": 1.4981226921081543,
-      "learning_rate": 5.631758793969849e-05,
-      "loss": 5.6101,
-      "step": 43980
-    },
-    {
-      "epoch": 22.936636245110822,
-      "grad_norm": 1.5827105045318604,
-      "learning_rate": 5.631658291457287e-05,
-      "loss": 5.3776,
-      "step": 43981
-    },
-    {
-      "epoch": 22.937157757496742,
-      "grad_norm": 1.5604897737503052,
-      "learning_rate": 5.631557788944724e-05,
-      "loss": 5.4678,
-      "step": 43982
-    },
-    {
-      "epoch": 22.93767926988266,
-      "grad_norm": 1.5260343551635742,
-      "learning_rate": 5.6314572864321616e-05,
-      "loss": 5.2747,
-      "step": 43983
-    },
-    {
-      "epoch": 22.938200782268577,
-      "grad_norm": 1.5876272916793823,
-      "learning_rate": 5.631356783919598e-05,
-      "loss": 4.8567,
-      "step": 43984
-    },
-    {
-      "epoch": 22.938722294654497,
-      "grad_norm": 1.5781205892562866,
-      "learning_rate": 5.631256281407036e-05,
-      "loss": 5.289,
-      "step": 43985
-    },
-    {
-      "epoch": 22.939243807040416,
-      "grad_norm": 1.4727349281311035,
-      "learning_rate": 5.631155778894472e-05,
-      "loss": 5.378,
-      "step": 43986
-    },
-    {
-      "epoch": 22.939765319426336,
-      "grad_norm": 1.6080297231674194,
-      "learning_rate": 5.63105527638191e-05,
-      "loss": 5.2689,
-      "step": 43987
-    },
-    {
-      "epoch": 22.940286831812255,
-      "grad_norm": 1.4950214624404907,
-      "learning_rate": 5.630954773869347e-05,
-      "loss": 5.6667,
-      "step": 43988
-    },
-    {
-      "epoch": 22.940808344198174,
-      "grad_norm": 1.5363363027572632,
-      "learning_rate": 5.630854271356785e-05,
-      "loss": 5.3952,
-      "step": 43989
-    },
-    {
-      "epoch": 22.941329856584094,
-      "grad_norm": 1.5760457515716553,
-      "learning_rate": 5.6307537688442213e-05,
-      "loss": 5.0277,
-      "step": 43990
-    },
-    {
-      "epoch": 22.941851368970013,
-      "grad_norm": 1.6175216436386108,
-      "learning_rate": 5.630653266331658e-05,
-      "loss": 5.2345,
-      "step": 43991
-    },
-    {
-      "epoch": 22.942372881355933,
-      "grad_norm": 1.4629230499267578,
-      "learning_rate": 5.6305527638190956e-05,
-      "loss": 5.3791,
-      "step": 43992
-    },
-    {
-      "epoch": 22.942894393741852,
-      "grad_norm": 1.660480260848999,
-      "learning_rate": 5.6304522613065327e-05,
-      "loss": 5.3354,
-      "step": 43993
-    },
-    {
-      "epoch": 22.943415906127772,
-      "grad_norm": 1.625388264656067,
-      "learning_rate": 5.6303517587939704e-05,
-      "loss": 5.3221,
-      "step": 43994
-    },
-    {
-      "epoch": 22.94393741851369,
-      "grad_norm": 1.4898459911346436,
-      "learning_rate": 5.630251256281407e-05,
-      "loss": 5.582,
-      "step": 43995
-    },
-    {
-      "epoch": 22.944458930899607,
-      "grad_norm": 1.5073834657669067,
-      "learning_rate": 5.6301507537688446e-05,
-      "loss": 5.1166,
-      "step": 43996
-    },
-    {
-      "epoch": 22.944980443285527,
-      "grad_norm": 1.6113460063934326,
-      "learning_rate": 5.630050251256281e-05,
-      "loss": 5.3444,
-      "step": 43997
-    },
-    {
-      "epoch": 22.945501955671446,
-      "grad_norm": 1.4706408977508545,
-      "learning_rate": 5.629949748743719e-05,
-      "loss": 5.4544,
-      "step": 43998
-    },
-    {
-      "epoch": 22.946023468057366,
-      "grad_norm": 1.4544531106948853,
-      "learning_rate": 5.629849246231156e-05,
-      "loss": 5.5088,
-      "step": 43999
-    },
-    {
-      "epoch": 22.946544980443285,
-      "grad_norm": 1.4736666679382324,
-      "learning_rate": 5.629748743718594e-05,
-      "loss": 5.1489,
-      "step": 44000
-    },
-    {
-      "epoch": 22.947066492829205,
-      "grad_norm": 1.663905143737793,
-      "learning_rate": 5.62964824120603e-05,
-      "loss": 5.3924,
-      "step": 44001
-    },
-    {
-      "epoch": 22.947588005215124,
-      "grad_norm": 1.5415337085723877,
-      "learning_rate": 5.629547738693468e-05,
-      "loss": 4.9444,
-      "step": 44002
-    },
-    {
-      "epoch": 22.948109517601043,
-      "grad_norm": 1.5413964986801147,
-      "learning_rate": 5.629447236180905e-05,
-      "loss": 4.9241,
-      "step": 44003
-    },
-    {
-      "epoch": 22.948631029986963,
-      "grad_norm": 1.537776231765747,
-      "learning_rate": 5.629346733668343e-05,
-      "loss": 5.6537,
-      "step": 44004
-    },
-    {
-      "epoch": 22.949152542372882,
-      "grad_norm": 1.559747576713562,
-      "learning_rate": 5.629246231155779e-05,
-      "loss": 5.4506,
-      "step": 44005
-    },
-    {
-      "epoch": 22.949674054758802,
-      "grad_norm": 1.6596547365188599,
-      "learning_rate": 5.629145728643216e-05,
-      "loss": 4.7549,
-      "step": 44006
-    },
-    {
-      "epoch": 22.95019556714472,
-      "grad_norm": 1.6191202402114868,
-      "learning_rate": 5.6290452261306535e-05,
-      "loss": 4.9306,
-      "step": 44007
-    },
-    {
-      "epoch": 22.950717079530637,
-      "grad_norm": 1.6380170583724976,
-      "learning_rate": 5.6289447236180906e-05,
-      "loss": 5.3569,
-      "step": 44008
-    },
-    {
-      "epoch": 22.951238591916557,
-      "grad_norm": 1.566148042678833,
-      "learning_rate": 5.6288442211055284e-05,
-      "loss": 5.3534,
-      "step": 44009
-    },
-    {
-      "epoch": 22.951760104302476,
-      "grad_norm": 1.5120725631713867,
-      "learning_rate": 5.628743718592965e-05,
-      "loss": 5.1883,
-      "step": 44010
-    },
-    {
-      "epoch": 22.952281616688396,
-      "grad_norm": 1.590471625328064,
-      "learning_rate": 5.6286432160804026e-05,
-      "loss": 4.9957,
-      "step": 44011
-    },
-    {
-      "epoch": 22.952803129074315,
-      "grad_norm": 1.6497929096221924,
-      "learning_rate": 5.628542713567839e-05,
-      "loss": 5.25,
-      "step": 44012
-    },
-    {
-      "epoch": 22.953324641460235,
-      "grad_norm": 1.5450183153152466,
-      "learning_rate": 5.628442211055277e-05,
-      "loss": 5.407,
-      "step": 44013
-    },
-    {
-      "epoch": 22.953846153846154,
-      "grad_norm": 1.41344153881073,
-      "learning_rate": 5.628341708542714e-05,
-      "loss": 5.5918,
-      "step": 44014
-    },
-    {
-      "epoch": 22.954367666232073,
-      "grad_norm": 1.5517079830169678,
-      "learning_rate": 5.628241206030152e-05,
-      "loss": 5.2776,
-      "step": 44015
-    },
-    {
-      "epoch": 22.954889178617993,
-      "grad_norm": 1.4785363674163818,
-      "learning_rate": 5.628140703517588e-05,
-      "loss": 4.9469,
-      "step": 44016
-    },
-    {
-      "epoch": 22.955410691003912,
-      "grad_norm": 1.6513373851776123,
-      "learning_rate": 5.628040201005026e-05,
-      "loss": 5.0082,
-      "step": 44017
-    },
-    {
-      "epoch": 22.955932203389832,
-      "grad_norm": 1.4752473831176758,
-      "learning_rate": 5.627939698492462e-05,
-      "loss": 5.1943,
-      "step": 44018
-    },
-    {
-      "epoch": 22.95645371577575,
-      "grad_norm": 1.589176893234253,
-      "learning_rate": 5.6278391959798994e-05,
-      "loss": 5.4167,
-      "step": 44019
-    },
-    {
-      "epoch": 22.956975228161667,
-      "grad_norm": 1.8575412034988403,
-      "learning_rate": 5.627738693467337e-05,
-      "loss": 4.9391,
-      "step": 44020
-    },
-    {
-      "epoch": 22.957496740547587,
-      "grad_norm": 1.5082495212554932,
-      "learning_rate": 5.6276381909547736e-05,
-      "loss": 5.5104,
-      "step": 44021
-    },
-    {
-      "epoch": 22.958018252933506,
-      "grad_norm": 1.4279236793518066,
-      "learning_rate": 5.6275376884422114e-05,
-      "loss": 5.4063,
-      "step": 44022
-    },
-    {
-      "epoch": 22.958539765319426,
-      "grad_norm": 1.474502682685852,
-      "learning_rate": 5.6274371859296485e-05,
-      "loss": 5.5541,
-      "step": 44023
-    },
-    {
-      "epoch": 22.959061277705345,
-      "grad_norm": 1.583530306816101,
-      "learning_rate": 5.627336683417086e-05,
-      "loss": 4.8739,
-      "step": 44024
-    },
-    {
-      "epoch": 22.959582790091265,
-      "grad_norm": 1.576289176940918,
-      "learning_rate": 5.627236180904523e-05,
-      "loss": 5.1191,
-      "step": 44025
-    },
-    {
-      "epoch": 22.960104302477184,
-      "grad_norm": 1.4606496095657349,
-      "learning_rate": 5.6271356783919605e-05,
-      "loss": 5.3671,
-      "step": 44026
-    },
-    {
-      "epoch": 22.960625814863103,
-      "grad_norm": 1.463059425354004,
-      "learning_rate": 5.627035175879397e-05,
-      "loss": 5.4798,
-      "step": 44027
-    },
-    {
-      "epoch": 22.961147327249023,
-      "grad_norm": 1.6371541023254395,
-      "learning_rate": 5.626934673366835e-05,
-      "loss": 5.1206,
-      "step": 44028
-    },
-    {
-      "epoch": 22.961668839634942,
-      "grad_norm": 1.532850742340088,
-      "learning_rate": 5.626834170854272e-05,
-      "loss": 4.8538,
-      "step": 44029
-    },
-    {
-      "epoch": 22.962190352020862,
-      "grad_norm": 1.5197837352752686,
-      "learning_rate": 5.6267336683417096e-05,
-      "loss": 4.9883,
-      "step": 44030
-    },
-    {
-      "epoch": 22.96271186440678,
-      "grad_norm": 1.6664855480194092,
-      "learning_rate": 5.626633165829146e-05,
-      "loss": 5.0164,
-      "step": 44031
-    },
-    {
-      "epoch": 22.963233376792697,
-      "grad_norm": 1.4934005737304688,
-      "learning_rate": 5.6265326633165825e-05,
-      "loss": 4.6812,
-      "step": 44032
-    },
-    {
-      "epoch": 22.963754889178617,
-      "grad_norm": 1.5279492139816284,
-      "learning_rate": 5.62643216080402e-05,
-      "loss": 5.1764,
-      "step": 44033
-    },
-    {
-      "epoch": 22.964276401564536,
-      "grad_norm": 1.513526439666748,
-      "learning_rate": 5.6263316582914574e-05,
-      "loss": 5.1018,
-      "step": 44034
-    },
-    {
-      "epoch": 22.964797913950456,
-      "grad_norm": 1.7110165357589722,
-      "learning_rate": 5.626231155778895e-05,
-      "loss": 5.056,
-      "step": 44035
-    },
-    {
-      "epoch": 22.965319426336375,
-      "grad_norm": 1.5855284929275513,
-      "learning_rate": 5.6261306532663316e-05,
-      "loss": 5.5769,
-      "step": 44036
-    },
-    {
-      "epoch": 22.965840938722295,
-      "grad_norm": 1.6046127080917358,
-      "learning_rate": 5.6260301507537693e-05,
-      "loss": 4.9911,
-      "step": 44037
-    },
-    {
-      "epoch": 22.966362451108214,
-      "grad_norm": 1.5545600652694702,
-      "learning_rate": 5.625929648241206e-05,
-      "loss": 4.9362,
-      "step": 44038
-    },
-    {
-      "epoch": 22.966883963494134,
-      "grad_norm": 1.5197006464004517,
-      "learning_rate": 5.6258291457286436e-05,
-      "loss": 5.2668,
-      "step": 44039
-    },
-    {
-      "epoch": 22.967405475880053,
-      "grad_norm": 1.4803029298782349,
-      "learning_rate": 5.625728643216081e-05,
-      "loss": 5.4607,
-      "step": 44040
-    },
-    {
-      "epoch": 22.967926988265972,
-      "grad_norm": 1.5844285488128662,
-      "learning_rate": 5.6256281407035184e-05,
-      "loss": 5.3269,
-      "step": 44041
-    },
-    {
-      "epoch": 22.968448500651892,
-      "grad_norm": 1.5247690677642822,
-      "learning_rate": 5.625527638190955e-05,
-      "loss": 5.4722,
-      "step": 44042
-    },
-    {
-      "epoch": 22.96897001303781,
-      "grad_norm": 1.5167207717895508,
-      "learning_rate": 5.6254271356783927e-05,
-      "loss": 5.169,
-      "step": 44043
-    },
-    {
-      "epoch": 22.969491525423727,
-      "grad_norm": 1.4671735763549805,
-      "learning_rate": 5.625326633165829e-05,
-      "loss": 5.0933,
-      "step": 44044
-    },
-    {
-      "epoch": 22.970013037809647,
-      "grad_norm": 1.6494500637054443,
-      "learning_rate": 5.625226130653266e-05,
-      "loss": 5.6151,
-      "step": 44045
-    },
-    {
-      "epoch": 22.970534550195566,
-      "grad_norm": 1.4672765731811523,
-      "learning_rate": 5.625125628140704e-05,
-      "loss": 5.037,
-      "step": 44046
-    },
-    {
-      "epoch": 22.971056062581486,
-      "grad_norm": 1.50511634349823,
-      "learning_rate": 5.6250251256281404e-05,
-      "loss": 5.5448,
-      "step": 44047
-    },
-    {
-      "epoch": 22.971577574967405,
-      "grad_norm": 1.507594108581543,
-      "learning_rate": 5.624924623115578e-05,
-      "loss": 4.9921,
-      "step": 44048
-    },
-    {
-      "epoch": 22.972099087353325,
-      "grad_norm": 1.5252268314361572,
-      "learning_rate": 5.624824120603015e-05,
-      "loss": 5.2267,
-      "step": 44049
-    },
-    {
-      "epoch": 22.972620599739244,
-      "grad_norm": 1.4307701587677002,
-      "learning_rate": 5.624723618090453e-05,
-      "loss": 4.8045,
-      "step": 44050
-    },
-    {
-      "epoch": 22.973142112125164,
-      "grad_norm": 1.5661019086837769,
-      "learning_rate": 5.6246231155778895e-05,
-      "loss": 4.5491,
-      "step": 44051
-    },
-    {
-      "epoch": 22.973663624511083,
-      "grad_norm": 1.4553507566452026,
-      "learning_rate": 5.624522613065327e-05,
-      "loss": 5.6224,
-      "step": 44052
-    },
-    {
-      "epoch": 22.974185136897002,
-      "grad_norm": 1.5095007419586182,
-      "learning_rate": 5.624422110552764e-05,
-      "loss": 5.3017,
-      "step": 44053
-    },
-    {
-      "epoch": 22.974706649282922,
-      "grad_norm": 1.5380406379699707,
-      "learning_rate": 5.6243216080402015e-05,
-      "loss": 5.3244,
-      "step": 44054
-    },
-    {
-      "epoch": 22.975228161668838,
-      "grad_norm": 1.5967761278152466,
-      "learning_rate": 5.6242211055276386e-05,
-      "loss": 4.918,
-      "step": 44055
-    },
-    {
-      "epoch": 22.975749674054757,
-      "grad_norm": 1.4848142862319946,
-      "learning_rate": 5.6241206030150764e-05,
-      "loss": 5.5117,
-      "step": 44056
-    },
-    {
-      "epoch": 22.976271186440677,
-      "grad_norm": 1.523545503616333,
-      "learning_rate": 5.624020100502513e-05,
-      "loss": 5.4724,
-      "step": 44057
-    },
-    {
-      "epoch": 22.976792698826596,
-      "grad_norm": 1.503098487854004,
-      "learning_rate": 5.623919597989949e-05,
-      "loss": 5.0057,
-      "step": 44058
-    },
-    {
-      "epoch": 22.977314211212516,
-      "grad_norm": 1.584805965423584,
-      "learning_rate": 5.623819095477387e-05,
-      "loss": 5.3204,
-      "step": 44059
-    },
-    {
-      "epoch": 22.977835723598435,
-      "grad_norm": 1.4231570959091187,
-      "learning_rate": 5.623718592964824e-05,
-      "loss": 5.4983,
-      "step": 44060
-    },
-    {
-      "epoch": 22.978357235984355,
-      "grad_norm": 1.4428253173828125,
-      "learning_rate": 5.623618090452262e-05,
-      "loss": 5.5572,
-      "step": 44061
-    },
-    {
-      "epoch": 22.978878748370274,
-      "grad_norm": 1.4895906448364258,
-      "learning_rate": 5.623517587939698e-05,
-      "loss": 5.6394,
-      "step": 44062
-    },
-    {
-      "epoch": 22.979400260756194,
-      "grad_norm": 1.5029640197753906,
-      "learning_rate": 5.623417085427136e-05,
-      "loss": 5.6583,
-      "step": 44063
-    },
-    {
-      "epoch": 22.979921773142113,
-      "grad_norm": 1.5950064659118652,
-      "learning_rate": 5.6233165829145725e-05,
-      "loss": 5.5935,
-      "step": 44064
-    },
-    {
-      "epoch": 22.980443285528033,
-      "grad_norm": 1.480576992034912,
-      "learning_rate": 5.62321608040201e-05,
-      "loss": 5.1865,
-      "step": 44065
-    },
-    {
-      "epoch": 22.980964797913952,
-      "grad_norm": 1.4549455642700195,
-      "learning_rate": 5.6231155778894474e-05,
-      "loss": 5.5733,
-      "step": 44066
-    },
-    {
-      "epoch": 22.98148631029987,
-      "grad_norm": 1.456506371498108,
-      "learning_rate": 5.623015075376885e-05,
-      "loss": 5.3811,
-      "step": 44067
-    },
-    {
-      "epoch": 22.982007822685787,
-      "grad_norm": 1.4077600240707397,
-      "learning_rate": 5.6229145728643216e-05,
-      "loss": 5.7994,
-      "step": 44068
-    },
-    {
-      "epoch": 22.982529335071707,
-      "grad_norm": 1.627037763595581,
-      "learning_rate": 5.6228140703517594e-05,
-      "loss": 5.3829,
-      "step": 44069
-    },
-    {
-      "epoch": 22.983050847457626,
-      "grad_norm": 1.5012351274490356,
-      "learning_rate": 5.6227135678391965e-05,
-      "loss": 5.0029,
-      "step": 44070
-    },
-    {
-      "epoch": 22.983572359843546,
-      "grad_norm": 1.4944490194320679,
-      "learning_rate": 5.622613065326633e-05,
-      "loss": 5.1808,
-      "step": 44071
-    },
-    {
-      "epoch": 22.984093872229465,
-      "grad_norm": 1.4852514266967773,
-      "learning_rate": 5.622512562814071e-05,
-      "loss": 5.3792,
-      "step": 44072
-    },
-    {
-      "epoch": 22.984615384615385,
-      "grad_norm": 1.4140602350234985,
-      "learning_rate": 5.622412060301507e-05,
-      "loss": 5.3983,
-      "step": 44073
-    },
-    {
-      "epoch": 22.985136897001304,
-      "grad_norm": 1.6279915571212769,
-      "learning_rate": 5.622311557788945e-05,
-      "loss": 5.0152,
-      "step": 44074
-    },
-    {
-      "epoch": 22.985658409387224,
-      "grad_norm": 1.5198726654052734,
-      "learning_rate": 5.622211055276382e-05,
-      "loss": 5.588,
-      "step": 44075
-    },
-    {
-      "epoch": 22.986179921773143,
-      "grad_norm": 1.6081055402755737,
-      "learning_rate": 5.62211055276382e-05,
-      "loss": 4.7823,
-      "step": 44076
-    },
-    {
-      "epoch": 22.986701434159063,
-      "grad_norm": 1.6730215549468994,
-      "learning_rate": 5.622010050251256e-05,
-      "loss": 4.7682,
-      "step": 44077
-    },
-    {
-      "epoch": 22.987222946544982,
-      "grad_norm": 1.5244570970535278,
-      "learning_rate": 5.621909547738694e-05,
-      "loss": 5.2134,
-      "step": 44078
-    },
-    {
-      "epoch": 22.987744458930898,
-      "grad_norm": 1.636431336402893,
-      "learning_rate": 5.6218090452261305e-05,
-      "loss": 5.1304,
-      "step": 44079
-    },
-    {
-      "epoch": 22.988265971316817,
-      "grad_norm": 1.498121976852417,
-      "learning_rate": 5.621708542713568e-05,
-      "loss": 5.4388,
-      "step": 44080
-    },
-    {
-      "epoch": 22.988787483702737,
-      "grad_norm": 1.5559110641479492,
-      "learning_rate": 5.6216080402010054e-05,
-      "loss": 5.2462,
-      "step": 44081
-    },
-    {
-      "epoch": 22.989308996088656,
-      "grad_norm": 1.574946403503418,
-      "learning_rate": 5.621507537688443e-05,
-      "loss": 5.5067,
-      "step": 44082
-    },
-    {
-      "epoch": 22.989830508474576,
-      "grad_norm": 1.4182963371276855,
-      "learning_rate": 5.6214070351758796e-05,
-      "loss": 5.4216,
-      "step": 44083
-    },
-    {
-      "epoch": 22.990352020860495,
-      "grad_norm": 1.535606861114502,
-      "learning_rate": 5.621306532663316e-05,
-      "loss": 5.4884,
-      "step": 44084
-    },
-    {
-      "epoch": 22.990873533246415,
-      "grad_norm": 1.5597097873687744,
-      "learning_rate": 5.621206030150754e-05,
-      "loss": 5.729,
-      "step": 44085
-    },
-    {
-      "epoch": 22.991395045632334,
-      "grad_norm": 1.6042708158493042,
-      "learning_rate": 5.621105527638191e-05,
-      "loss": 5.0583,
-      "step": 44086
-    },
-    {
-      "epoch": 22.991916558018254,
-      "grad_norm": 1.4815045595169067,
-      "learning_rate": 5.621005025125629e-05,
-      "loss": 5.5103,
-      "step": 44087
-    },
-    {
-      "epoch": 22.992438070404173,
-      "grad_norm": 1.559187412261963,
-      "learning_rate": 5.620904522613065e-05,
-      "loss": 5.3178,
-      "step": 44088
-    },
-    {
-      "epoch": 22.992959582790093,
-      "grad_norm": 1.653410792350769,
-      "learning_rate": 5.620804020100503e-05,
-      "loss": 4.3553,
-      "step": 44089
-    },
-    {
-      "epoch": 22.993481095176012,
-      "grad_norm": 1.5962815284729004,
-      "learning_rate": 5.62070351758794e-05,
-      "loss": 4.8565,
-      "step": 44090
-    },
-    {
-      "epoch": 22.994002607561928,
-      "grad_norm": 1.5313242673873901,
-      "learning_rate": 5.620603015075378e-05,
-      "loss": 5.213,
-      "step": 44091
-    },
-    {
-      "epoch": 22.994524119947847,
-      "grad_norm": 1.4439717531204224,
-      "learning_rate": 5.620502512562814e-05,
-      "loss": 4.7648,
-      "step": 44092
-    },
-    {
-      "epoch": 22.995045632333767,
-      "grad_norm": 1.6210914850234985,
-      "learning_rate": 5.620402010050252e-05,
-      "loss": 5.4878,
-      "step": 44093
-    },
-    {
-      "epoch": 22.995567144719686,
-      "grad_norm": 1.5509672164916992,
-      "learning_rate": 5.6203015075376884e-05,
-      "loss": 5.1722,
-      "step": 44094
-    },
-    {
-      "epoch": 22.996088657105606,
-      "grad_norm": 1.5761781930923462,
-      "learning_rate": 5.620201005025126e-05,
-      "loss": 5.049,
-      "step": 44095
-    },
-    {
-      "epoch": 22.996610169491525,
-      "grad_norm": 1.4706286191940308,
-      "learning_rate": 5.620100502512563e-05,
-      "loss": 5.0305,
-      "step": 44096
-    },
-    {
-      "epoch": 22.997131681877445,
-      "grad_norm": 1.5995993614196777,
-      "learning_rate": 5.620000000000001e-05,
-      "loss": 5.2825,
-      "step": 44097
-    },
-    {
-      "epoch": 22.997653194263364,
-      "grad_norm": 1.561200499534607,
-      "learning_rate": 5.6198994974874375e-05,
-      "loss": 5.629,
-      "step": 44098
-    },
-    {
-      "epoch": 22.998174706649284,
-      "grad_norm": 1.61214280128479,
-      "learning_rate": 5.619798994974874e-05,
-      "loss": 5.1613,
-      "step": 44099
-    },
-    {
-      "epoch": 22.998696219035203,
-      "grad_norm": 1.5085150003433228,
-      "learning_rate": 5.619698492462312e-05,
-      "loss": 5.3224,
-      "step": 44100
-    },
-    {
-      "epoch": 22.999217731421123,
-      "grad_norm": 1.4961739778518677,
-      "learning_rate": 5.619597989949749e-05,
-      "loss": 5.1694,
-      "step": 44101
-    },
-    {
-      "epoch": 22.999739243807042,
-      "grad_norm": 1.4383876323699951,
-      "learning_rate": 5.6194974874371866e-05,
-      "loss": 5.6287,
-      "step": 44102
-    },
-    {
-      "epoch": 23.000260756192958,
-      "grad_norm": 1.7588353157043457,
-      "learning_rate": 5.619396984924623e-05,
-      "loss": 5.1551,
-      "step": 44103
-    },
-    {
-      "epoch": 23.000782268578877,
-      "grad_norm": 1.5699337720870972,
-      "learning_rate": 5.619296482412061e-05,
-      "loss": 5.1483,
-      "step": 44104
-    },
-    {
-      "epoch": 23.001303780964797,
-      "grad_norm": 1.5176074504852295,
-      "learning_rate": 5.619195979899497e-05,
-      "loss": 5.2755,
-      "step": 44105
-    },
-    {
-      "epoch": 23.001825293350716,
-      "grad_norm": 1.4847779273986816,
-      "learning_rate": 5.619095477386935e-05,
-      "loss": 5.2375,
-      "step": 44106
-    },
-    {
-      "epoch": 23.002346805736636,
-      "grad_norm": 1.5303823947906494,
-      "learning_rate": 5.618994974874372e-05,
-      "loss": 4.9822,
-      "step": 44107
-    },
-    {
-      "epoch": 23.002868318122555,
-      "grad_norm": 1.5105905532836914,
-      "learning_rate": 5.61889447236181e-05,
-      "loss": 5.5286,
-      "step": 44108
-    },
-    {
-      "epoch": 23.003389830508475,
-      "grad_norm": 1.4539484977722168,
-      "learning_rate": 5.6187939698492463e-05,
-      "loss": 5.0467,
-      "step": 44109
-    },
-    {
-      "epoch": 23.003911342894394,
-      "grad_norm": 1.5218104124069214,
-      "learning_rate": 5.618693467336684e-05,
-      "loss": 4.9173,
-      "step": 44110
-    },
-    {
-      "epoch": 23.004432855280314,
-      "grad_norm": 1.5706593990325928,
-      "learning_rate": 5.618592964824121e-05,
-      "loss": 5.3231,
-      "step": 44111
-    },
-    {
-      "epoch": 23.004954367666233,
-      "grad_norm": 1.58208429813385,
-      "learning_rate": 5.6184924623115577e-05,
-      "loss": 5.1382,
-      "step": 44112
-    },
-    {
-      "epoch": 23.005475880052153,
-      "grad_norm": 1.5139538049697876,
-      "learning_rate": 5.6183919597989954e-05,
-      "loss": 4.7019,
-      "step": 44113
-    },
-    {
-      "epoch": 23.005997392438072,
-      "grad_norm": 1.5079801082611084,
-      "learning_rate": 5.618291457286432e-05,
-      "loss": 5.1059,
-      "step": 44114
-    },
-    {
-      "epoch": 23.006518904823988,
-      "grad_norm": 1.4938979148864746,
-      "learning_rate": 5.6181909547738697e-05,
-      "loss": 5.3422,
-      "step": 44115
-    },
-    {
-      "epoch": 23.007040417209907,
-      "grad_norm": 1.5331518650054932,
-      "learning_rate": 5.618090452261307e-05,
-      "loss": 5.3532,
-      "step": 44116
-    },
-    {
-      "epoch": 23.007561929595827,
-      "grad_norm": 1.5456804037094116,
-      "learning_rate": 5.6179899497487445e-05,
-      "loss": 5.2898,
-      "step": 44117
-    },
-    {
-      "epoch": 23.008083441981746,
-      "grad_norm": 1.5163803100585938,
-      "learning_rate": 5.617889447236181e-05,
-      "loss": 5.0099,
-      "step": 44118
-    },
-    {
-      "epoch": 23.008604954367666,
-      "grad_norm": 1.524611473083496,
-      "learning_rate": 5.617788944723619e-05,
-      "loss": 5.3774,
-      "step": 44119
-    },
-    {
-      "epoch": 23.009126466753585,
-      "grad_norm": 1.5253468751907349,
-      "learning_rate": 5.617688442211055e-05,
-      "loss": 5.2014,
-      "step": 44120
-    },
-    {
-      "epoch": 23.009647979139505,
-      "grad_norm": 1.6026999950408936,
-      "learning_rate": 5.617587939698493e-05,
-      "loss": 4.7867,
-      "step": 44121
-    },
-    {
-      "epoch": 23.010169491525424,
-      "grad_norm": 1.4861558675765991,
-      "learning_rate": 5.61748743718593e-05,
-      "loss": 5.4878,
-      "step": 44122
-    },
-    {
-      "epoch": 23.010691003911344,
-      "grad_norm": 1.6254584789276123,
-      "learning_rate": 5.617386934673368e-05,
-      "loss": 5.2507,
-      "step": 44123
-    },
-    {
-      "epoch": 23.011212516297263,
-      "grad_norm": 1.6275973320007324,
-      "learning_rate": 5.617286432160804e-05,
-      "loss": 4.9022,
-      "step": 44124
-    },
-    {
-      "epoch": 23.011734028683183,
-      "grad_norm": 1.5329769849777222,
-      "learning_rate": 5.617185929648241e-05,
-      "loss": 5.0549,
-      "step": 44125
-    },
-    {
-      "epoch": 23.012255541069102,
-      "grad_norm": 1.4493244886398315,
-      "learning_rate": 5.6170854271356785e-05,
-      "loss": 5.6359,
-      "step": 44126
-    },
-    {
-      "epoch": 23.012777053455018,
-      "grad_norm": 1.471423625946045,
-      "learning_rate": 5.6169849246231156e-05,
-      "loss": 5.1138,
-      "step": 44127
-    },
-    {
-      "epoch": 23.013298565840937,
-      "grad_norm": 1.5144530534744263,
-      "learning_rate": 5.6168844221105534e-05,
-      "loss": 4.8477,
-      "step": 44128
-    },
-    {
-      "epoch": 23.013820078226857,
-      "grad_norm": 1.5249199867248535,
-      "learning_rate": 5.61678391959799e-05,
-      "loss": 5.191,
-      "step": 44129
-    },
-    {
-      "epoch": 23.014341590612776,
-      "grad_norm": 1.5001307725906372,
-      "learning_rate": 5.6166834170854276e-05,
-      "loss": 4.9884,
-      "step": 44130
-    },
-    {
-      "epoch": 23.014863102998696,
-      "grad_norm": 1.423403024673462,
-      "learning_rate": 5.616582914572864e-05,
-      "loss": 5.8198,
-      "step": 44131
-    },
-    {
-      "epoch": 23.015384615384615,
-      "grad_norm": 1.530036211013794,
-      "learning_rate": 5.616482412060302e-05,
-      "loss": 5.6179,
-      "step": 44132
-    },
-    {
-      "epoch": 23.015906127770535,
-      "grad_norm": 1.5245168209075928,
-      "learning_rate": 5.616381909547739e-05,
-      "loss": 4.6743,
-      "step": 44133
-    },
-    {
-      "epoch": 23.016427640156454,
-      "grad_norm": 1.4966111183166504,
-      "learning_rate": 5.616281407035177e-05,
-      "loss": 4.8368,
-      "step": 44134
-    },
-    {
-      "epoch": 23.016949152542374,
-      "grad_norm": 1.5477303266525269,
-      "learning_rate": 5.616180904522613e-05,
-      "loss": 5.7182,
-      "step": 44135
-    },
-    {
-      "epoch": 23.017470664928293,
-      "grad_norm": 1.4636286497116089,
-      "learning_rate": 5.616080402010051e-05,
-      "loss": 5.3811,
-      "step": 44136
-    },
-    {
-      "epoch": 23.017992177314213,
-      "grad_norm": 1.540224313735962,
-      "learning_rate": 5.615979899497488e-05,
-      "loss": 4.7569,
-      "step": 44137
-    },
-    {
-      "epoch": 23.018513689700132,
-      "grad_norm": 1.5466463565826416,
-      "learning_rate": 5.6158793969849244e-05,
-      "loss": 5.5355,
-      "step": 44138
-    },
-    {
-      "epoch": 23.019035202086048,
-      "grad_norm": 1.4627565145492554,
-      "learning_rate": 5.615778894472362e-05,
-      "loss": 5.3994,
-      "step": 44139
-    },
-    {
-      "epoch": 23.019556714471967,
-      "grad_norm": 1.5818090438842773,
-      "learning_rate": 5.6156783919597986e-05,
-      "loss": 5.1401,
-      "step": 44140
-    },
-    {
-      "epoch": 23.020078226857887,
-      "grad_norm": 1.4703141450881958,
-      "learning_rate": 5.6155778894472364e-05,
-      "loss": 5.6937,
-      "step": 44141
-    },
-    {
-      "epoch": 23.020599739243806,
-      "grad_norm": 1.574371576309204,
-      "learning_rate": 5.6154773869346735e-05,
-      "loss": 4.7369,
-      "step": 44142
-    },
-    {
-      "epoch": 23.021121251629726,
-      "grad_norm": 1.6514657735824585,
-      "learning_rate": 5.615376884422111e-05,
-      "loss": 4.8413,
-      "step": 44143
-    },
-    {
-      "epoch": 23.021642764015645,
-      "grad_norm": 1.6309975385665894,
-      "learning_rate": 5.615276381909548e-05,
-      "loss": 4.756,
-      "step": 44144
-    },
-    {
-      "epoch": 23.022164276401565,
-      "grad_norm": 1.5601089000701904,
-      "learning_rate": 5.6151758793969855e-05,
-      "loss": 5.2113,
-      "step": 44145
-    },
-    {
-      "epoch": 23.022685788787484,
-      "grad_norm": 1.6040171384811401,
-      "learning_rate": 5.615075376884422e-05,
-      "loss": 4.9641,
-      "step": 44146
-    },
-    {
-      "epoch": 23.023207301173404,
-      "grad_norm": 1.5096713304519653,
-      "learning_rate": 5.61497487437186e-05,
-      "loss": 5.2857,
-      "step": 44147
-    },
-    {
-      "epoch": 23.023728813559323,
-      "grad_norm": 1.5144857168197632,
-      "learning_rate": 5.614874371859297e-05,
-      "loss": 5.4664,
-      "step": 44148
-    },
-    {
-      "epoch": 23.024250325945243,
-      "grad_norm": 1.6723887920379639,
-      "learning_rate": 5.6147738693467346e-05,
-      "loss": 5.3033,
-      "step": 44149
-    },
-    {
-      "epoch": 23.02477183833116,
-      "grad_norm": 1.5519890785217285,
-      "learning_rate": 5.614673366834171e-05,
-      "loss": 5.5244,
-      "step": 44150
-    },
-    {
-      "epoch": 23.025293350717078,
-      "grad_norm": 1.5825716257095337,
-      "learning_rate": 5.6145728643216075e-05,
-      "loss": 5.413,
-      "step": 44151
-    },
-    {
-      "epoch": 23.025814863102998,
-      "grad_norm": 1.5205601453781128,
-      "learning_rate": 5.614472361809045e-05,
-      "loss": 5.36,
-      "step": 44152
-    },
-    {
-      "epoch": 23.026336375488917,
-      "grad_norm": 1.6226478815078735,
-      "learning_rate": 5.6143718592964824e-05,
-      "loss": 4.679,
-      "step": 44153
-    },
-    {
-      "epoch": 23.026857887874836,
-      "grad_norm": 1.5495645999908447,
-      "learning_rate": 5.61427135678392e-05,
-      "loss": 5.2031,
-      "step": 44154
-    },
-    {
-      "epoch": 23.027379400260756,
-      "grad_norm": 1.6152679920196533,
-      "learning_rate": 5.6141708542713566e-05,
-      "loss": 5.5204,
-      "step": 44155
-    },
-    {
-      "epoch": 23.027900912646675,
-      "grad_norm": 1.5449550151824951,
-      "learning_rate": 5.6140703517587944e-05,
-      "loss": 5.2636,
-      "step": 44156
-    },
-    {
-      "epoch": 23.028422425032595,
-      "grad_norm": 1.5193078517913818,
-      "learning_rate": 5.6139698492462315e-05,
-      "loss": 5.3413,
-      "step": 44157
-    },
-    {
-      "epoch": 23.028943937418514,
-      "grad_norm": 1.4681317806243896,
-      "learning_rate": 5.613869346733669e-05,
-      "loss": 5.2517,
-      "step": 44158
-    },
-    {
-      "epoch": 23.029465449804434,
-      "grad_norm": 1.672638177871704,
-      "learning_rate": 5.613768844221106e-05,
-      "loss": 5.1265,
-      "step": 44159
-    },
-    {
-      "epoch": 23.029986962190353,
-      "grad_norm": 1.6082420349121094,
-      "learning_rate": 5.6136683417085434e-05,
-      "loss": 5.2193,
-      "step": 44160
-    },
-    {
-      "epoch": 23.030508474576273,
-      "grad_norm": 1.5401055812835693,
-      "learning_rate": 5.61356783919598e-05,
-      "loss": 5.4234,
-      "step": 44161
-    },
-    {
-      "epoch": 23.03102998696219,
-      "grad_norm": 1.536392092704773,
-      "learning_rate": 5.6134673366834177e-05,
-      "loss": 4.9333,
-      "step": 44162
-    },
-    {
-      "epoch": 23.031551499348108,
-      "grad_norm": 1.5998786687850952,
-      "learning_rate": 5.613366834170855e-05,
-      "loss": 4.7861,
-      "step": 44163
-    },
-    {
-      "epoch": 23.032073011734028,
-      "grad_norm": 1.6624573469161987,
-      "learning_rate": 5.613266331658291e-05,
-      "loss": 5.2664,
-      "step": 44164
-    },
-    {
-      "epoch": 23.032594524119947,
-      "grad_norm": 1.5876541137695312,
-      "learning_rate": 5.613165829145729e-05,
-      "loss": 5.0848,
-      "step": 44165
-    },
-    {
-      "epoch": 23.033116036505866,
-      "grad_norm": 1.6187584400177002,
-      "learning_rate": 5.6130653266331654e-05,
-      "loss": 5.4199,
-      "step": 44166
-    },
-    {
-      "epoch": 23.033637548891786,
-      "grad_norm": 1.64701509475708,
-      "learning_rate": 5.612964824120603e-05,
-      "loss": 4.9288,
-      "step": 44167
-    },
-    {
-      "epoch": 23.034159061277705,
-      "grad_norm": 1.4942572116851807,
-      "learning_rate": 5.61286432160804e-05,
-      "loss": 4.9982,
-      "step": 44168
-    },
-    {
-      "epoch": 23.034680573663625,
-      "grad_norm": 1.4706209897994995,
-      "learning_rate": 5.612763819095478e-05,
-      "loss": 5.4747,
-      "step": 44169
-    },
-    {
-      "epoch": 23.035202086049544,
-      "grad_norm": 1.558253526687622,
-      "learning_rate": 5.6126633165829145e-05,
-      "loss": 5.3631,
-      "step": 44170
-    },
-    {
-      "epoch": 23.035723598435464,
-      "grad_norm": 1.670030951499939,
-      "learning_rate": 5.612562814070352e-05,
-      "loss": 5.1758,
-      "step": 44171
-    },
-    {
-      "epoch": 23.036245110821383,
-      "grad_norm": 1.5509953498840332,
-      "learning_rate": 5.612462311557789e-05,
-      "loss": 5.2983,
-      "step": 44172
-    },
-    {
-      "epoch": 23.036766623207303,
-      "grad_norm": 1.565963625907898,
-      "learning_rate": 5.6123618090452265e-05,
-      "loss": 5.4427,
-      "step": 44173
-    },
-    {
-      "epoch": 23.03728813559322,
-      "grad_norm": 1.510735273361206,
-      "learning_rate": 5.6122613065326636e-05,
-      "loss": 5.3611,
-      "step": 44174
-    },
-    {
-      "epoch": 23.037809647979138,
-      "grad_norm": 1.6084152460098267,
-      "learning_rate": 5.6121608040201014e-05,
-      "loss": 5.1011,
-      "step": 44175
-    },
-    {
-      "epoch": 23.038331160365058,
-      "grad_norm": 1.625915288925171,
-      "learning_rate": 5.612060301507538e-05,
-      "loss": 5.1559,
-      "step": 44176
-    },
-    {
-      "epoch": 23.038852672750977,
-      "grad_norm": 1.5887471437454224,
-      "learning_rate": 5.611959798994975e-05,
-      "loss": 5.2114,
-      "step": 44177
-    },
-    {
-      "epoch": 23.039374185136897,
-      "grad_norm": 1.6323591470718384,
-      "learning_rate": 5.611859296482413e-05,
-      "loss": 5.0946,
-      "step": 44178
-    },
-    {
-      "epoch": 23.039895697522816,
-      "grad_norm": 1.5416592359542847,
-      "learning_rate": 5.611758793969849e-05,
-      "loss": 5.5736,
-      "step": 44179
-    },
-    {
-      "epoch": 23.040417209908735,
-      "grad_norm": 1.55796480178833,
-      "learning_rate": 5.611658291457287e-05,
-      "loss": 5.2377,
-      "step": 44180
-    },
-    {
-      "epoch": 23.040938722294655,
-      "grad_norm": 1.5152515172958374,
-      "learning_rate": 5.611557788944723e-05,
-      "loss": 4.8527,
-      "step": 44181
-    },
-    {
-      "epoch": 23.041460234680574,
-      "grad_norm": 1.5090440511703491,
-      "learning_rate": 5.611457286432161e-05,
-      "loss": 5.3446,
-      "step": 44182
-    },
-    {
-      "epoch": 23.041981747066494,
-      "grad_norm": 1.5127770900726318,
-      "learning_rate": 5.611356783919598e-05,
-      "loss": 5.0866,
-      "step": 44183
-    },
-    {
-      "epoch": 23.042503259452413,
-      "grad_norm": 1.6319607496261597,
-      "learning_rate": 5.611256281407036e-05,
-      "loss": 5.1327,
-      "step": 44184
-    },
-    {
-      "epoch": 23.043024771838333,
-      "grad_norm": 1.5124109983444214,
-      "learning_rate": 5.6111557788944724e-05,
-      "loss": 5.4293,
-      "step": 44185
-    },
-    {
-      "epoch": 23.04354628422425,
-      "grad_norm": 1.5612196922302246,
-      "learning_rate": 5.61105527638191e-05,
-      "loss": 5.3042,
-      "step": 44186
-    },
-    {
-      "epoch": 23.044067796610168,
-      "grad_norm": 1.5254265069961548,
-      "learning_rate": 5.6109547738693466e-05,
-      "loss": 5.3866,
-      "step": 44187
-    },
-    {
-      "epoch": 23.044589308996088,
-      "grad_norm": 1.4587345123291016,
-      "learning_rate": 5.6108542713567844e-05,
-      "loss": 5.2136,
-      "step": 44188
-    },
-    {
-      "epoch": 23.045110821382007,
-      "grad_norm": 1.529735803604126,
-      "learning_rate": 5.6107537688442215e-05,
-      "loss": 5.1422,
-      "step": 44189
-    },
-    {
-      "epoch": 23.045632333767927,
-      "grad_norm": 1.5934793949127197,
-      "learning_rate": 5.610653266331659e-05,
-      "loss": 5.3534,
-      "step": 44190
-    },
-    {
-      "epoch": 23.046153846153846,
-      "grad_norm": 1.532606840133667,
-      "learning_rate": 5.610552763819096e-05,
-      "loss": 5.3825,
-      "step": 44191
-    },
-    {
-      "epoch": 23.046675358539765,
-      "grad_norm": 1.5712941884994507,
-      "learning_rate": 5.610452261306532e-05,
-      "loss": 4.9236,
-      "step": 44192
-    },
-    {
-      "epoch": 23.047196870925685,
-      "grad_norm": 1.6492092609405518,
-      "learning_rate": 5.61035175879397e-05,
-      "loss": 5.4112,
-      "step": 44193
-    },
-    {
-      "epoch": 23.047718383311604,
-      "grad_norm": 1.476919412612915,
-      "learning_rate": 5.610251256281407e-05,
-      "loss": 5.5795,
-      "step": 44194
-    },
-    {
-      "epoch": 23.048239895697524,
-      "grad_norm": 1.5565747022628784,
-      "learning_rate": 5.610150753768845e-05,
-      "loss": 4.9406,
-      "step": 44195
-    },
-    {
-      "epoch": 23.048761408083443,
-      "grad_norm": 1.516579031944275,
-      "learning_rate": 5.610050251256281e-05,
-      "loss": 5.4549,
-      "step": 44196
-    },
-    {
-      "epoch": 23.049282920469363,
-      "grad_norm": 1.5050814151763916,
-      "learning_rate": 5.609949748743719e-05,
-      "loss": 5.3274,
-      "step": 44197
-    },
-    {
-      "epoch": 23.04980443285528,
-      "grad_norm": 1.6160473823547363,
-      "learning_rate": 5.609849246231156e-05,
-      "loss": 5.3057,
-      "step": 44198
-    },
-    {
-      "epoch": 23.050325945241198,
-      "grad_norm": 1.6080831289291382,
-      "learning_rate": 5.609748743718594e-05,
-      "loss": 5.1603,
-      "step": 44199
-    },
-    {
-      "epoch": 23.050847457627118,
-      "grad_norm": 1.5089322328567505,
-      "learning_rate": 5.6096482412060304e-05,
-      "loss": 5.4709,
-      "step": 44200
-    },
-    {
-      "epoch": 23.051368970013037,
-      "grad_norm": 1.5144751071929932,
-      "learning_rate": 5.609547738693468e-05,
-      "loss": 5.4352,
-      "step": 44201
-    },
-    {
-      "epoch": 23.051890482398957,
-      "grad_norm": 1.647049069404602,
-      "learning_rate": 5.6094472361809046e-05,
-      "loss": 5.0064,
-      "step": 44202
-    },
-    {
-      "epoch": 23.052411994784876,
-      "grad_norm": 1.4825514554977417,
-      "learning_rate": 5.6093467336683424e-05,
-      "loss": 5.3918,
-      "step": 44203
-    },
-    {
-      "epoch": 23.052933507170795,
-      "grad_norm": 1.544574499130249,
-      "learning_rate": 5.6092462311557795e-05,
-      "loss": 4.9544,
-      "step": 44204
-    },
-    {
-      "epoch": 23.053455019556715,
-      "grad_norm": 1.49387526512146,
-      "learning_rate": 5.609145728643216e-05,
-      "loss": 5.1768,
-      "step": 44205
-    },
-    {
-      "epoch": 23.053976531942634,
-      "grad_norm": 1.5668026208877563,
-      "learning_rate": 5.609045226130654e-05,
-      "loss": 5.1151,
-      "step": 44206
-    },
-    {
-      "epoch": 23.054498044328554,
-      "grad_norm": 1.5981688499450684,
-      "learning_rate": 5.60894472361809e-05,
-      "loss": 4.9203,
-      "step": 44207
-    },
-    {
-      "epoch": 23.055019556714473,
-      "grad_norm": 1.5040651559829712,
-      "learning_rate": 5.608844221105528e-05,
-      "loss": 5.7794,
-      "step": 44208
-    },
-    {
-      "epoch": 23.055541069100393,
-      "grad_norm": 1.5354117155075073,
-      "learning_rate": 5.608743718592965e-05,
-      "loss": 4.9094,
-      "step": 44209
-    },
-    {
-      "epoch": 23.05606258148631,
-      "grad_norm": 1.5024762153625488,
-      "learning_rate": 5.608643216080403e-05,
-      "loss": 5.2287,
-      "step": 44210
-    },
-    {
-      "epoch": 23.056584093872228,
-      "grad_norm": 1.5759599208831787,
-      "learning_rate": 5.608542713567839e-05,
-      "loss": 5.2285,
-      "step": 44211
-    },
-    {
-      "epoch": 23.057105606258148,
-      "grad_norm": 1.6399372816085815,
-      "learning_rate": 5.608442211055277e-05,
-      "loss": 5.271,
-      "step": 44212
-    },
-    {
-      "epoch": 23.057627118644067,
-      "grad_norm": 1.7120574712753296,
-      "learning_rate": 5.6083417085427134e-05,
-      "loss": 4.9534,
-      "step": 44213
-    },
-    {
-      "epoch": 23.058148631029987,
-      "grad_norm": 1.4270570278167725,
-      "learning_rate": 5.608241206030151e-05,
-      "loss": 5.1419,
-      "step": 44214
-    },
-    {
-      "epoch": 23.058670143415906,
-      "grad_norm": 1.4865541458129883,
-      "learning_rate": 5.608140703517588e-05,
-      "loss": 5.0917,
-      "step": 44215
-    },
-    {
-      "epoch": 23.059191655801826,
-      "grad_norm": 1.5296885967254639,
-      "learning_rate": 5.608040201005026e-05,
-      "loss": 5.4378,
-      "step": 44216
-    },
-    {
-      "epoch": 23.059713168187745,
-      "grad_norm": 1.5675642490386963,
-      "learning_rate": 5.6079396984924625e-05,
-      "loss": 5.3041,
-      "step": 44217
-    },
-    {
-      "epoch": 23.060234680573664,
-      "grad_norm": 1.6579577922821045,
-      "learning_rate": 5.607839195979899e-05,
-      "loss": 5.0469,
-      "step": 44218
-    },
-    {
-      "epoch": 23.060756192959584,
-      "grad_norm": 1.663954734802246,
-      "learning_rate": 5.607738693467337e-05,
-      "loss": 4.8878,
-      "step": 44219
-    },
-    {
-      "epoch": 23.061277705345503,
-      "grad_norm": 1.6589109897613525,
-      "learning_rate": 5.607638190954774e-05,
-      "loss": 5.3474,
-      "step": 44220
-    },
-    {
-      "epoch": 23.061799217731423,
-      "grad_norm": 1.5335230827331543,
-      "learning_rate": 5.6075376884422116e-05,
-      "loss": 5.5685,
-      "step": 44221
-    },
-    {
-      "epoch": 23.06232073011734,
-      "grad_norm": 1.5106971263885498,
-      "learning_rate": 5.607437185929648e-05,
-      "loss": 5.7113,
-      "step": 44222
-    },
-    {
-      "epoch": 23.062842242503258,
-      "grad_norm": 1.7632415294647217,
-      "learning_rate": 5.607336683417086e-05,
-      "loss": 4.7541,
-      "step": 44223
-    },
-    {
-      "epoch": 23.063363754889178,
-      "grad_norm": 1.6437299251556396,
-      "learning_rate": 5.607236180904523e-05,
-      "loss": 5.0254,
-      "step": 44224
-    },
-    {
-      "epoch": 23.063885267275097,
-      "grad_norm": 1.583186388015747,
-      "learning_rate": 5.607135678391961e-05,
-      "loss": 5.4276,
-      "step": 44225
-    },
-    {
-      "epoch": 23.064406779661017,
-      "grad_norm": 1.5530083179473877,
-      "learning_rate": 5.607035175879397e-05,
-      "loss": 5.0412,
-      "step": 44226
-    },
-    {
-      "epoch": 23.064928292046936,
-      "grad_norm": 1.4764927625656128,
-      "learning_rate": 5.606934673366835e-05,
-      "loss": 5.7177,
-      "step": 44227
-    },
-    {
-      "epoch": 23.065449804432856,
-      "grad_norm": 1.50925612449646,
-      "learning_rate": 5.6068341708542713e-05,
-      "loss": 5.4734,
-      "step": 44228
-    },
-    {
-      "epoch": 23.065971316818775,
-      "grad_norm": 1.6366883516311646,
-      "learning_rate": 5.606733668341709e-05,
-      "loss": 5.053,
-      "step": 44229
-    },
-    {
-      "epoch": 23.066492829204694,
-      "grad_norm": 1.5717936754226685,
-      "learning_rate": 5.606633165829146e-05,
-      "loss": 5.4273,
-      "step": 44230
-    },
-    {
-      "epoch": 23.067014341590614,
-      "grad_norm": 1.6765187978744507,
-      "learning_rate": 5.6065326633165827e-05,
-      "loss": 4.8675,
-      "step": 44231
-    },
-    {
-      "epoch": 23.067535853976533,
-      "grad_norm": 1.4949336051940918,
-      "learning_rate": 5.6064321608040204e-05,
-      "loss": 5.5861,
-      "step": 44232
-    },
-    {
-      "epoch": 23.068057366362453,
-      "grad_norm": 1.5550392866134644,
-      "learning_rate": 5.606331658291457e-05,
-      "loss": 5.509,
-      "step": 44233
-    },
-    {
-      "epoch": 23.06857887874837,
-      "grad_norm": 1.5104155540466309,
-      "learning_rate": 5.6062311557788947e-05,
-      "loss": 5.6239,
-      "step": 44234
-    },
-    {
-      "epoch": 23.06910039113429,
-      "grad_norm": 1.46573007106781,
-      "learning_rate": 5.606130653266332e-05,
-      "loss": 5.6563,
-      "step": 44235
-    },
-    {
-      "epoch": 23.069621903520208,
-      "grad_norm": 1.550082802772522,
-      "learning_rate": 5.6060301507537695e-05,
-      "loss": 5.3968,
-      "step": 44236
-    },
-    {
-      "epoch": 23.070143415906127,
-      "grad_norm": 1.489371418952942,
-      "learning_rate": 5.605929648241206e-05,
-      "loss": 4.8999,
-      "step": 44237
-    },
-    {
-      "epoch": 23.070664928292047,
-      "grad_norm": 1.4827178716659546,
-      "learning_rate": 5.605829145728644e-05,
-      "loss": 5.733,
-      "step": 44238
-    },
-    {
-      "epoch": 23.071186440677966,
-      "grad_norm": 1.6949259042739868,
-      "learning_rate": 5.60572864321608e-05,
-      "loss": 5.2909,
-      "step": 44239
-    },
-    {
-      "epoch": 23.071707953063886,
-      "grad_norm": 1.5821691751480103,
-      "learning_rate": 5.605628140703518e-05,
-      "loss": 4.9813,
-      "step": 44240
-    },
-    {
-      "epoch": 23.072229465449805,
-      "grad_norm": 1.587161898612976,
-      "learning_rate": 5.605527638190955e-05,
-      "loss": 5.0957,
-      "step": 44241
-    },
-    {
-      "epoch": 23.072750977835724,
-      "grad_norm": 1.546392798423767,
-      "learning_rate": 5.605427135678393e-05,
-      "loss": 5.3123,
-      "step": 44242
-    },
-    {
-      "epoch": 23.073272490221644,
-      "grad_norm": 1.6802620887756348,
-      "learning_rate": 5.605326633165829e-05,
-      "loss": 5.2688,
-      "step": 44243
-    },
-    {
-      "epoch": 23.073794002607563,
-      "grad_norm": 1.464930534362793,
-      "learning_rate": 5.6052261306532664e-05,
-      "loss": 4.9244,
-      "step": 44244
-    },
-    {
-      "epoch": 23.07431551499348,
-      "grad_norm": 1.5040459632873535,
-      "learning_rate": 5.605125628140704e-05,
-      "loss": 5.1904,
-      "step": 44245
-    },
-    {
-      "epoch": 23.0748370273794,
-      "grad_norm": 1.5453670024871826,
-      "learning_rate": 5.6050251256281406e-05,
-      "loss": 5.2639,
-      "step": 44246
-    },
-    {
-      "epoch": 23.07535853976532,
-      "grad_norm": 1.4370310306549072,
-      "learning_rate": 5.6049246231155784e-05,
-      "loss": 5.4263,
-      "step": 44247
-    },
-    {
-      "epoch": 23.075880052151238,
-      "grad_norm": 1.5279301404953003,
-      "learning_rate": 5.604824120603015e-05,
-      "loss": 5.1857,
-      "step": 44248
-    },
-    {
-      "epoch": 23.076401564537157,
-      "grad_norm": 1.5973496437072754,
-      "learning_rate": 5.6047236180904526e-05,
-      "loss": 5.5631,
-      "step": 44249
-    },
-    {
-      "epoch": 23.076923076923077,
-      "grad_norm": 1.5487520694732666,
-      "learning_rate": 5.60462311557789e-05,
-      "loss": 5.6208,
-      "step": 44250
-    },
-    {
-      "epoch": 23.077444589308996,
-      "grad_norm": 1.5417556762695312,
-      "learning_rate": 5.6045226130653275e-05,
-      "loss": 5.2185,
-      "step": 44251
-    },
-    {
-      "epoch": 23.077966101694916,
-      "grad_norm": 1.5724568367004395,
-      "learning_rate": 5.604422110552764e-05,
-      "loss": 5.3911,
-      "step": 44252
-    },
-    {
-      "epoch": 23.078487614080835,
-      "grad_norm": 1.4482685327529907,
-      "learning_rate": 5.604321608040202e-05,
-      "loss": 5.4551,
-      "step": 44253
-    },
-    {
-      "epoch": 23.079009126466755,
-      "grad_norm": 1.545122504234314,
-      "learning_rate": 5.604221105527638e-05,
-      "loss": 5.6441,
-      "step": 44254
-    },
-    {
-      "epoch": 23.079530638852674,
-      "grad_norm": 1.5829100608825684,
-      "learning_rate": 5.604120603015076e-05,
-      "loss": 5.2844,
-      "step": 44255
-    },
-    {
-      "epoch": 23.080052151238593,
-      "grad_norm": 1.4594438076019287,
-      "learning_rate": 5.604020100502513e-05,
-      "loss": 5.5087,
-      "step": 44256
-    },
-    {
-      "epoch": 23.08057366362451,
-      "grad_norm": 1.5534204244613647,
-      "learning_rate": 5.6039195979899494e-05,
-      "loss": 4.8756,
-      "step": 44257
-    },
-    {
-      "epoch": 23.08109517601043,
-      "grad_norm": 1.5856492519378662,
-      "learning_rate": 5.603819095477387e-05,
-      "loss": 4.669,
-      "step": 44258
-    },
-    {
-      "epoch": 23.08161668839635,
-      "grad_norm": 1.5068422555923462,
-      "learning_rate": 5.6037185929648236e-05,
-      "loss": 5.4423,
-      "step": 44259
-    },
-    {
-      "epoch": 23.082138200782268,
-      "grad_norm": 1.5549417734146118,
-      "learning_rate": 5.6036180904522614e-05,
-      "loss": 4.6242,
-      "step": 44260
-    },
-    {
-      "epoch": 23.082659713168187,
-      "grad_norm": 1.528660774230957,
-      "learning_rate": 5.6035175879396985e-05,
-      "loss": 4.8343,
-      "step": 44261
-    },
-    {
-      "epoch": 23.083181225554107,
-      "grad_norm": 1.5528937578201294,
-      "learning_rate": 5.603417085427136e-05,
-      "loss": 5.4004,
-      "step": 44262
-    },
-    {
-      "epoch": 23.083702737940026,
-      "grad_norm": 1.5217008590698242,
-      "learning_rate": 5.603316582914573e-05,
-      "loss": 5.17,
-      "step": 44263
-    },
-    {
-      "epoch": 23.084224250325946,
-      "grad_norm": 1.6222697496414185,
-      "learning_rate": 5.6032160804020105e-05,
-      "loss": 4.9701,
-      "step": 44264
-    },
-    {
-      "epoch": 23.084745762711865,
-      "grad_norm": 1.460544466972351,
-      "learning_rate": 5.6031155778894476e-05,
-      "loss": 5.5013,
-      "step": 44265
-    },
-    {
-      "epoch": 23.085267275097785,
-      "grad_norm": 1.5549968481063843,
-      "learning_rate": 5.6030150753768854e-05,
-      "loss": 4.955,
-      "step": 44266
-    },
-    {
-      "epoch": 23.085788787483704,
-      "grad_norm": 1.558011531829834,
-      "learning_rate": 5.602914572864322e-05,
-      "loss": 4.9515,
-      "step": 44267
-    },
-    {
-      "epoch": 23.086310299869623,
-      "grad_norm": 1.5533313751220703,
-      "learning_rate": 5.6028140703517596e-05,
-      "loss": 4.9929,
-      "step": 44268
-    },
-    {
-      "epoch": 23.08683181225554,
-      "grad_norm": 1.4526426792144775,
-      "learning_rate": 5.602713567839196e-05,
-      "loss": 5.3377,
-      "step": 44269
-    },
-    {
-      "epoch": 23.08735332464146,
-      "grad_norm": 1.5141342878341675,
-      "learning_rate": 5.602613065326633e-05,
-      "loss": 5.3245,
-      "step": 44270
-    },
-    {
-      "epoch": 23.08787483702738,
-      "grad_norm": 1.5991771221160889,
-      "learning_rate": 5.602512562814071e-05,
-      "loss": 5.368,
-      "step": 44271
-    },
-    {
-      "epoch": 23.088396349413298,
-      "grad_norm": 1.588348388671875,
-      "learning_rate": 5.6024120603015074e-05,
-      "loss": 4.8774,
-      "step": 44272
-    },
-    {
-      "epoch": 23.088917861799217,
-      "grad_norm": 1.6172094345092773,
-      "learning_rate": 5.602311557788945e-05,
-      "loss": 5.0733,
-      "step": 44273
-    },
-    {
-      "epoch": 23.089439374185137,
-      "grad_norm": 1.5130037069320679,
-      "learning_rate": 5.6022110552763816e-05,
-      "loss": 5.5688,
-      "step": 44274
-    },
-    {
-      "epoch": 23.089960886571056,
-      "grad_norm": 1.4715625047683716,
-      "learning_rate": 5.6021105527638194e-05,
-      "loss": 5.3636,
-      "step": 44275
-    },
-    {
-      "epoch": 23.090482398956976,
-      "grad_norm": 1.522261381149292,
-      "learning_rate": 5.6020100502512565e-05,
-      "loss": 5.5567,
-      "step": 44276
-    },
-    {
-      "epoch": 23.091003911342895,
-      "grad_norm": 1.6823046207427979,
-      "learning_rate": 5.601909547738694e-05,
-      "loss": 5.0217,
-      "step": 44277
-    },
-    {
-      "epoch": 23.091525423728815,
-      "grad_norm": 1.5491161346435547,
-      "learning_rate": 5.601809045226131e-05,
-      "loss": 5.3432,
-      "step": 44278
-    },
-    {
-      "epoch": 23.092046936114734,
-      "grad_norm": 1.6071226596832275,
-      "learning_rate": 5.6017085427135685e-05,
-      "loss": 5.5206,
-      "step": 44279
-    },
-    {
-      "epoch": 23.092568448500653,
-      "grad_norm": 1.6054600477218628,
-      "learning_rate": 5.601608040201005e-05,
-      "loss": 4.6726,
-      "step": 44280
-    },
-    {
-      "epoch": 23.09308996088657,
-      "grad_norm": 1.7696481943130493,
-      "learning_rate": 5.6015075376884427e-05,
-      "loss": 5.1434,
-      "step": 44281
-    },
-    {
-      "epoch": 23.09361147327249,
-      "grad_norm": 1.606623649597168,
-      "learning_rate": 5.60140703517588e-05,
-      "loss": 5.4651,
-      "step": 44282
-    },
-    {
-      "epoch": 23.09413298565841,
-      "grad_norm": 1.6091371774673462,
-      "learning_rate": 5.6013065326633175e-05,
-      "loss": 5.0726,
-      "step": 44283
-    },
-    {
-      "epoch": 23.094654498044328,
-      "grad_norm": 1.4765387773513794,
-      "learning_rate": 5.601206030150754e-05,
-      "loss": 5.6324,
-      "step": 44284
-    },
-    {
-      "epoch": 23.095176010430247,
-      "grad_norm": 1.5956339836120605,
-      "learning_rate": 5.6011055276381904e-05,
-      "loss": 5.245,
-      "step": 44285
-    },
-    {
-      "epoch": 23.095697522816167,
-      "grad_norm": 1.5123881101608276,
-      "learning_rate": 5.601005025125628e-05,
-      "loss": 5.4311,
-      "step": 44286
-    },
-    {
-      "epoch": 23.096219035202086,
-      "grad_norm": 1.603606104850769,
-      "learning_rate": 5.600904522613065e-05,
-      "loss": 5.2715,
-      "step": 44287
-    },
-    {
-      "epoch": 23.096740547588006,
-      "grad_norm": 1.5447793006896973,
-      "learning_rate": 5.600804020100503e-05,
-      "loss": 4.6607,
-      "step": 44288
-    },
-    {
-      "epoch": 23.097262059973925,
-      "grad_norm": 1.4804331064224243,
-      "learning_rate": 5.6007035175879395e-05,
-      "loss": 5.4718,
-      "step": 44289
-    },
-    {
-      "epoch": 23.097783572359845,
-      "grad_norm": 1.5432261228561401,
-      "learning_rate": 5.600603015075377e-05,
-      "loss": 5.5115,
-      "step": 44290
-    },
-    {
-      "epoch": 23.098305084745764,
-      "grad_norm": 1.531092643737793,
-      "learning_rate": 5.6005025125628144e-05,
-      "loss": 5.2517,
-      "step": 44291
-    },
-    {
-      "epoch": 23.098826597131684,
-      "grad_norm": 1.518325924873352,
-      "learning_rate": 5.600402010050252e-05,
-      "loss": 5.6896,
-      "step": 44292
-    },
-    {
-      "epoch": 23.0993481095176,
-      "grad_norm": 1.5546977519989014,
-      "learning_rate": 5.6003015075376886e-05,
-      "loss": 5.3904,
-      "step": 44293
-    },
-    {
-      "epoch": 23.09986962190352,
-      "grad_norm": 1.5527348518371582,
-      "learning_rate": 5.6002010050251264e-05,
-      "loss": 4.9812,
-      "step": 44294
-    },
-    {
-      "epoch": 23.10039113428944,
-      "grad_norm": 1.6087510585784912,
-      "learning_rate": 5.600100502512563e-05,
-      "loss": 5.2385,
-      "step": 44295
-    },
-    {
-      "epoch": 23.100912646675358,
-      "grad_norm": 1.5110647678375244,
-      "learning_rate": 5.6000000000000006e-05,
-      "loss": 5.4561,
-      "step": 44296
-    },
-    {
-      "epoch": 23.101434159061277,
-      "grad_norm": 1.5034502744674683,
-      "learning_rate": 5.599899497487438e-05,
-      "loss": 5.268,
-      "step": 44297
-    },
-    {
-      "epoch": 23.101955671447197,
-      "grad_norm": 1.7065238952636719,
-      "learning_rate": 5.599798994974874e-05,
-      "loss": 4.1829,
-      "step": 44298
-    },
-    {
-      "epoch": 23.102477183833116,
-      "grad_norm": 1.6263313293457031,
-      "learning_rate": 5.599698492462312e-05,
-      "loss": 5.1696,
-      "step": 44299
-    },
-    {
-      "epoch": 23.102998696219036,
-      "grad_norm": 1.4984503984451294,
-      "learning_rate": 5.599597989949748e-05,
-      "loss": 5.3902,
-      "step": 44300
-    },
-    {
-      "epoch": 23.103520208604955,
-      "grad_norm": 1.7102168798446655,
-      "learning_rate": 5.599497487437186e-05,
-      "loss": 4.8354,
-      "step": 44301
-    },
-    {
-      "epoch": 23.104041720990875,
-      "grad_norm": 1.6121397018432617,
-      "learning_rate": 5.599396984924623e-05,
-      "loss": 5.2388,
-      "step": 44302
-    },
-    {
-      "epoch": 23.104563233376794,
-      "grad_norm": 1.5432770252227783,
-      "learning_rate": 5.599296482412061e-05,
-      "loss": 4.2624,
-      "step": 44303
-    },
-    {
-      "epoch": 23.105084745762714,
-      "grad_norm": 1.5199192762374878,
-      "learning_rate": 5.5991959798994974e-05,
-      "loss": 4.9791,
-      "step": 44304
-    },
-    {
-      "epoch": 23.10560625814863,
-      "grad_norm": 1.6821264028549194,
-      "learning_rate": 5.599095477386935e-05,
-      "loss": 5.0596,
-      "step": 44305
-    },
-    {
-      "epoch": 23.10612777053455,
-      "grad_norm": 1.3521559238433838,
-      "learning_rate": 5.5989949748743716e-05,
-      "loss": 4.6266,
-      "step": 44306
-    },
-    {
-      "epoch": 23.10664928292047,
-      "grad_norm": 1.5250223875045776,
-      "learning_rate": 5.5988944723618094e-05,
-      "loss": 5.4424,
-      "step": 44307
-    },
-    {
-      "epoch": 23.107170795306388,
-      "grad_norm": 1.5268884897232056,
-      "learning_rate": 5.5987939698492465e-05,
-      "loss": 5.427,
-      "step": 44308
-    },
-    {
-      "epoch": 23.107692307692307,
-      "grad_norm": 1.6352870464324951,
-      "learning_rate": 5.598693467336684e-05,
-      "loss": 5.1343,
-      "step": 44309
-    },
-    {
-      "epoch": 23.108213820078227,
-      "grad_norm": 1.48247230052948,
-      "learning_rate": 5.598592964824121e-05,
-      "loss": 5.5253,
-      "step": 44310
-    },
-    {
-      "epoch": 23.108735332464146,
-      "grad_norm": 1.6788885593414307,
-      "learning_rate": 5.598492462311558e-05,
-      "loss": 5.1573,
-      "step": 44311
-    },
-    {
-      "epoch": 23.109256844850066,
-      "grad_norm": 1.485263705253601,
-      "learning_rate": 5.5983919597989956e-05,
-      "loss": 5.0941,
-      "step": 44312
-    },
-    {
-      "epoch": 23.109778357235985,
-      "grad_norm": 1.5086501836776733,
-      "learning_rate": 5.598291457286432e-05,
-      "loss": 4.968,
-      "step": 44313
-    },
-    {
-      "epoch": 23.110299869621905,
-      "grad_norm": 1.558056116104126,
-      "learning_rate": 5.59819095477387e-05,
-      "loss": 5.2162,
-      "step": 44314
-    },
-    {
-      "epoch": 23.110821382007824,
-      "grad_norm": 1.5805771350860596,
-      "learning_rate": 5.598090452261306e-05,
-      "loss": 5.5491,
-      "step": 44315
-    },
-    {
-      "epoch": 23.111342894393744,
-      "grad_norm": 1.5781910419464111,
-      "learning_rate": 5.597989949748744e-05,
-      "loss": 5.2424,
-      "step": 44316
-    },
-    {
-      "epoch": 23.11186440677966,
-      "grad_norm": 1.5647194385528564,
-      "learning_rate": 5.597889447236181e-05,
-      "loss": 5.5491,
-      "step": 44317
-    },
-    {
-      "epoch": 23.11238591916558,
-      "grad_norm": 1.6656441688537598,
-      "learning_rate": 5.597788944723619e-05,
-      "loss": 4.8497,
-      "step": 44318
-    },
-    {
-      "epoch": 23.1129074315515,
-      "grad_norm": 1.639399528503418,
-      "learning_rate": 5.5976884422110554e-05,
-      "loss": 4.903,
-      "step": 44319
-    },
-    {
-      "epoch": 23.113428943937418,
-      "grad_norm": 1.6096041202545166,
-      "learning_rate": 5.597587939698493e-05,
-      "loss": 4.9045,
-      "step": 44320
-    },
-    {
-      "epoch": 23.113950456323337,
-      "grad_norm": 1.515893578529358,
-      "learning_rate": 5.5974874371859296e-05,
-      "loss": 5.266,
-      "step": 44321
-    },
-    {
-      "epoch": 23.114471968709257,
-      "grad_norm": 1.492468237876892,
-      "learning_rate": 5.5973869346733674e-05,
-      "loss": 5.3952,
-      "step": 44322
-    },
-    {
-      "epoch": 23.114993481095176,
-      "grad_norm": 1.536942720413208,
-      "learning_rate": 5.5972864321608045e-05,
-      "loss": 4.7623,
-      "step": 44323
-    },
-    {
-      "epoch": 23.115514993481096,
-      "grad_norm": 1.6108567714691162,
-      "learning_rate": 5.597185929648241e-05,
-      "loss": 5.3394,
-      "step": 44324
-    },
-    {
-      "epoch": 23.116036505867015,
-      "grad_norm": 1.5863429307937622,
-      "learning_rate": 5.597085427135679e-05,
-      "loss": 5.2572,
-      "step": 44325
-    },
-    {
-      "epoch": 23.116558018252935,
-      "grad_norm": 1.5053595304489136,
-      "learning_rate": 5.596984924623115e-05,
-      "loss": 5.3792,
-      "step": 44326
-    },
-    {
-      "epoch": 23.117079530638854,
-      "grad_norm": 1.5470268726348877,
-      "learning_rate": 5.596884422110553e-05,
-      "loss": 5.3229,
-      "step": 44327
-    },
-    {
-      "epoch": 23.117601043024774,
-      "grad_norm": 1.5217698812484741,
-      "learning_rate": 5.59678391959799e-05,
-      "loss": 5.2699,
-      "step": 44328
-    },
-    {
-      "epoch": 23.11812255541069,
-      "grad_norm": 1.4709482192993164,
-      "learning_rate": 5.596683417085428e-05,
-      "loss": 5.4574,
-      "step": 44329
-    },
-    {
-      "epoch": 23.11864406779661,
-      "grad_norm": 1.493037223815918,
-      "learning_rate": 5.596582914572864e-05,
-      "loss": 5.3327,
-      "step": 44330
-    },
-    {
-      "epoch": 23.11916558018253,
-      "grad_norm": 1.6582324504852295,
-      "learning_rate": 5.596482412060302e-05,
-      "loss": 5.0363,
-      "step": 44331
-    },
-    {
-      "epoch": 23.119687092568448,
-      "grad_norm": 1.5751782655715942,
-      "learning_rate": 5.596381909547739e-05,
-      "loss": 5.4138,
-      "step": 44332
-    },
-    {
-      "epoch": 23.120208604954367,
-      "grad_norm": 1.4760597944259644,
-      "learning_rate": 5.596281407035177e-05,
-      "loss": 5.0749,
-      "step": 44333
-    },
-    {
-      "epoch": 23.120730117340287,
-      "grad_norm": 1.7289230823516846,
-      "learning_rate": 5.596180904522613e-05,
-      "loss": 5.1463,
-      "step": 44334
-    },
-    {
-      "epoch": 23.121251629726206,
-      "grad_norm": 1.5487769842147827,
-      "learning_rate": 5.596080402010051e-05,
-      "loss": 5.2902,
-      "step": 44335
-    },
-    {
-      "epoch": 23.121773142112126,
-      "grad_norm": 1.5186022520065308,
-      "learning_rate": 5.5959798994974875e-05,
-      "loss": 4.9507,
-      "step": 44336
-    },
-    {
-      "epoch": 23.122294654498045,
-      "grad_norm": 1.5960909128189087,
-      "learning_rate": 5.5958793969849246e-05,
-      "loss": 5.5566,
-      "step": 44337
-    },
-    {
-      "epoch": 23.122816166883965,
-      "grad_norm": 1.5690075159072876,
-      "learning_rate": 5.5957788944723624e-05,
-      "loss": 4.8766,
-      "step": 44338
-    },
-    {
-      "epoch": 23.123337679269884,
-      "grad_norm": 1.534613847732544,
-      "learning_rate": 5.595678391959799e-05,
-      "loss": 5.4237,
-      "step": 44339
-    },
-    {
-      "epoch": 23.1238591916558,
-      "grad_norm": 1.5109068155288696,
-      "learning_rate": 5.5955778894472366e-05,
-      "loss": 5.0691,
-      "step": 44340
-    },
-    {
-      "epoch": 23.12438070404172,
-      "grad_norm": 1.4922438859939575,
-      "learning_rate": 5.595477386934673e-05,
-      "loss": 5.1997,
-      "step": 44341
-    },
-    {
-      "epoch": 23.12490221642764,
-      "grad_norm": 1.500375747680664,
-      "learning_rate": 5.595376884422111e-05,
-      "loss": 5.5567,
-      "step": 44342
-    },
-    {
-      "epoch": 23.12542372881356,
-      "grad_norm": 1.5292985439300537,
-      "learning_rate": 5.595276381909548e-05,
-      "loss": 4.749,
-      "step": 44343
-    },
-    {
-      "epoch": 23.125945241199478,
-      "grad_norm": 1.5737888813018799,
-      "learning_rate": 5.595175879396986e-05,
-      "loss": 5.1188,
-      "step": 44344
-    },
-    {
-      "epoch": 23.126466753585397,
-      "grad_norm": 1.8988252878189087,
-      "learning_rate": 5.595075376884422e-05,
-      "loss": 4.609,
-      "step": 44345
-    },
-    {
-      "epoch": 23.126988265971317,
-      "grad_norm": 1.5698645114898682,
-      "learning_rate": 5.59497487437186e-05,
-      "loss": 5.2377,
-      "step": 44346
-    },
-    {
-      "epoch": 23.127509778357236,
-      "grad_norm": 1.6264585256576538,
-      "learning_rate": 5.5948743718592963e-05,
-      "loss": 5.4621,
-      "step": 44347
-    },
-    {
-      "epoch": 23.128031290743156,
-      "grad_norm": 1.5163516998291016,
-      "learning_rate": 5.594773869346734e-05,
-      "loss": 5.6817,
-      "step": 44348
-    },
-    {
-      "epoch": 23.128552803129075,
-      "grad_norm": 1.5212162733078003,
-      "learning_rate": 5.594673366834171e-05,
-      "loss": 5.4772,
-      "step": 44349
-    },
-    {
-      "epoch": 23.129074315514995,
-      "grad_norm": 1.754381775856018,
-      "learning_rate": 5.5945728643216077e-05,
-      "loss": 4.4816,
-      "step": 44350
-    },
-    {
-      "epoch": 23.129595827900914,
-      "grad_norm": 1.5987505912780762,
-      "learning_rate": 5.5944723618090454e-05,
-      "loss": 5.1579,
-      "step": 44351
-    },
-    {
-      "epoch": 23.13011734028683,
-      "grad_norm": 1.5706615447998047,
-      "learning_rate": 5.5943718592964825e-05,
-      "loss": 5.1844,
-      "step": 44352
-    },
-    {
-      "epoch": 23.13063885267275,
-      "grad_norm": 1.5856157541275024,
-      "learning_rate": 5.59427135678392e-05,
-      "loss": 4.2906,
-      "step": 44353
-    },
-    {
-      "epoch": 23.13116036505867,
-      "grad_norm": 1.4615163803100586,
-      "learning_rate": 5.594170854271357e-05,
-      "loss": 4.9227,
-      "step": 44354
-    },
-    {
-      "epoch": 23.13168187744459,
-      "grad_norm": 1.497867465019226,
-      "learning_rate": 5.5940703517587945e-05,
-      "loss": 5.5688,
-      "step": 44355
-    },
-    {
-      "epoch": 23.132203389830508,
-      "grad_norm": 1.5673774480819702,
-      "learning_rate": 5.593969849246231e-05,
-      "loss": 5.1855,
-      "step": 44356
-    },
-    {
-      "epoch": 23.132724902216427,
-      "grad_norm": 1.4880610704421997,
-      "learning_rate": 5.593869346733669e-05,
-      "loss": 5.284,
-      "step": 44357
-    },
-    {
-      "epoch": 23.133246414602347,
-      "grad_norm": 1.5867726802825928,
-      "learning_rate": 5.593768844221106e-05,
-      "loss": 4.5313,
-      "step": 44358
-    },
-    {
-      "epoch": 23.133767926988266,
-      "grad_norm": 1.5662530660629272,
-      "learning_rate": 5.5936683417085436e-05,
-      "loss": 5.3711,
-      "step": 44359
-    },
-    {
-      "epoch": 23.134289439374186,
-      "grad_norm": 1.525652289390564,
-      "learning_rate": 5.59356783919598e-05,
-      "loss": 4.8821,
-      "step": 44360
-    },
-    {
-      "epoch": 23.134810951760105,
-      "grad_norm": 1.5443443059921265,
-      "learning_rate": 5.593467336683418e-05,
-      "loss": 4.9643,
-      "step": 44361
-    },
-    {
-      "epoch": 23.135332464146025,
-      "grad_norm": 1.5732898712158203,
-      "learning_rate": 5.593366834170854e-05,
-      "loss": 5.506,
-      "step": 44362
-    },
-    {
-      "epoch": 23.135853976531944,
-      "grad_norm": 1.6634172201156616,
-      "learning_rate": 5.593266331658292e-05,
-      "loss": 5.2097,
-      "step": 44363
-    },
-    {
-      "epoch": 23.13637548891786,
-      "grad_norm": 1.4624149799346924,
-      "learning_rate": 5.593165829145729e-05,
-      "loss": 5.2691,
-      "step": 44364
-    },
-    {
-      "epoch": 23.13689700130378,
-      "grad_norm": 1.4564409255981445,
-      "learning_rate": 5.5930653266331656e-05,
-      "loss": 5.4651,
-      "step": 44365
-    },
-    {
-      "epoch": 23.1374185136897,
-      "grad_norm": 1.6802165508270264,
-      "learning_rate": 5.5929648241206034e-05,
-      "loss": 5.1465,
-      "step": 44366
-    },
-    {
-      "epoch": 23.13794002607562,
-      "grad_norm": 1.6761503219604492,
-      "learning_rate": 5.59286432160804e-05,
-      "loss": 4.6668,
-      "step": 44367
-    },
-    {
-      "epoch": 23.138461538461538,
-      "grad_norm": 1.4565293788909912,
-      "learning_rate": 5.5927638190954776e-05,
-      "loss": 5.5904,
-      "step": 44368
-    },
-    {
-      "epoch": 23.138983050847457,
-      "grad_norm": 1.6015913486480713,
-      "learning_rate": 5.592663316582915e-05,
-      "loss": 5.3259,
-      "step": 44369
-    },
-    {
-      "epoch": 23.139504563233377,
-      "grad_norm": 1.6245182752609253,
-      "learning_rate": 5.5925628140703525e-05,
-      "loss": 4.7964,
-      "step": 44370
-    },
-    {
-      "epoch": 23.140026075619296,
-      "grad_norm": 1.6014807224273682,
-      "learning_rate": 5.592462311557789e-05,
-      "loss": 5.4318,
-      "step": 44371
-    },
-    {
-      "epoch": 23.140547588005216,
-      "grad_norm": 1.4273537397384644,
-      "learning_rate": 5.592361809045227e-05,
-      "loss": 5.0628,
-      "step": 44372
-    },
-    {
-      "epoch": 23.141069100391135,
-      "grad_norm": 1.4863674640655518,
-      "learning_rate": 5.592261306532663e-05,
-      "loss": 5.4974,
-      "step": 44373
-    },
-    {
-      "epoch": 23.141590612777055,
-      "grad_norm": 1.6562001705169678,
-      "learning_rate": 5.592160804020101e-05,
-      "loss": 5.4295,
-      "step": 44374
-    },
-    {
-      "epoch": 23.142112125162974,
-      "grad_norm": 1.5794895887374878,
-      "learning_rate": 5.592060301507538e-05,
-      "loss": 5.3571,
-      "step": 44375
-    },
-    {
-      "epoch": 23.14263363754889,
-      "grad_norm": 1.5250059366226196,
-      "learning_rate": 5.591959798994976e-05,
-      "loss": 5.3526,
-      "step": 44376
-    },
-    {
-      "epoch": 23.14315514993481,
-      "grad_norm": 1.545509934425354,
-      "learning_rate": 5.591859296482412e-05,
-      "loss": 5.358,
-      "step": 44377
-    },
-    {
-      "epoch": 23.14367666232073,
-      "grad_norm": 1.5436042547225952,
-      "learning_rate": 5.591758793969849e-05,
-      "loss": 5.2048,
-      "step": 44378
-    },
-    {
-      "epoch": 23.14419817470665,
-      "grad_norm": 1.5172964334487915,
-      "learning_rate": 5.591658291457287e-05,
-      "loss": 5.4697,
-      "step": 44379
-    },
-    {
-      "epoch": 23.144719687092568,
-      "grad_norm": 1.674127221107483,
-      "learning_rate": 5.5915577889447235e-05,
-      "loss": 5.1954,
-      "step": 44380
-    },
-    {
-      "epoch": 23.145241199478487,
-      "grad_norm": 1.5467638969421387,
-      "learning_rate": 5.591457286432161e-05,
-      "loss": 5.3445,
-      "step": 44381
-    },
-    {
-      "epoch": 23.145762711864407,
-      "grad_norm": 1.5212962627410889,
-      "learning_rate": 5.591356783919598e-05,
-      "loss": 5.3549,
-      "step": 44382
-    },
-    {
-      "epoch": 23.146284224250326,
-      "grad_norm": 1.5430058240890503,
-      "learning_rate": 5.5912562814070355e-05,
-      "loss": 5.4836,
-      "step": 44383
-    },
-    {
-      "epoch": 23.146805736636246,
-      "grad_norm": 1.554516077041626,
-      "learning_rate": 5.5911557788944726e-05,
-      "loss": 4.7715,
-      "step": 44384
-    },
-    {
-      "epoch": 23.147327249022165,
-      "grad_norm": 1.5024774074554443,
-      "learning_rate": 5.5910552763819104e-05,
-      "loss": 5.3976,
-      "step": 44385
-    },
-    {
-      "epoch": 23.147848761408085,
-      "grad_norm": 1.5544353723526,
-      "learning_rate": 5.590954773869347e-05,
-      "loss": 4.6706,
-      "step": 44386
-    },
-    {
-      "epoch": 23.148370273794004,
-      "grad_norm": 1.6513954401016235,
-      "learning_rate": 5.5908542713567846e-05,
-      "loss": 5.4086,
-      "step": 44387
-    },
-    {
-      "epoch": 23.14889178617992,
-      "grad_norm": 1.8536309003829956,
-      "learning_rate": 5.590753768844221e-05,
-      "loss": 5.1364,
-      "step": 44388
-    },
-    {
-      "epoch": 23.14941329856584,
-      "grad_norm": 1.5676305294036865,
-      "learning_rate": 5.590653266331659e-05,
-      "loss": 5.2804,
-      "step": 44389
-    },
-    {
-      "epoch": 23.14993481095176,
-      "grad_norm": 1.4657251834869385,
-      "learning_rate": 5.590552763819096e-05,
-      "loss": 4.6826,
-      "step": 44390
-    },
-    {
-      "epoch": 23.15045632333768,
-      "grad_norm": 1.5766276121139526,
-      "learning_rate": 5.5904522613065324e-05,
-      "loss": 5.4326,
-      "step": 44391
-    },
-    {
-      "epoch": 23.150977835723598,
-      "grad_norm": 1.4632548093795776,
-      "learning_rate": 5.59035175879397e-05,
-      "loss": 4.9899,
-      "step": 44392
-    },
-    {
-      "epoch": 23.151499348109517,
-      "grad_norm": 1.5896002054214478,
-      "learning_rate": 5.5902512562814066e-05,
-      "loss": 5.1503,
-      "step": 44393
-    },
-    {
-      "epoch": 23.152020860495437,
-      "grad_norm": 1.4700225591659546,
-      "learning_rate": 5.5901507537688444e-05,
-      "loss": 5.0012,
-      "step": 44394
-    },
-    {
-      "epoch": 23.152542372881356,
-      "grad_norm": 1.5365785360336304,
-      "learning_rate": 5.5900502512562815e-05,
-      "loss": 5.3897,
-      "step": 44395
-    },
-    {
-      "epoch": 23.153063885267276,
-      "grad_norm": 1.547487497329712,
-      "learning_rate": 5.589949748743719e-05,
-      "loss": 5.2639,
-      "step": 44396
-    },
-    {
-      "epoch": 23.153585397653195,
-      "grad_norm": 1.551567792892456,
-      "learning_rate": 5.589849246231156e-05,
-      "loss": 5.1781,
-      "step": 44397
-    },
-    {
-      "epoch": 23.154106910039115,
-      "grad_norm": 1.5515793561935425,
-      "learning_rate": 5.5897487437185935e-05,
-      "loss": 5.0279,
-      "step": 44398
-    },
-    {
-      "epoch": 23.154628422425034,
-      "grad_norm": 1.5107226371765137,
-      "learning_rate": 5.5896482412060306e-05,
-      "loss": 5.2351,
-      "step": 44399
-    },
-    {
-      "epoch": 23.15514993481095,
-      "grad_norm": 1.6052024364471436,
-      "learning_rate": 5.5895477386934683e-05,
-      "loss": 4.8875,
-      "step": 44400
-    },
-    {
-      "epoch": 23.15567144719687,
-      "grad_norm": 1.7481673955917358,
-      "learning_rate": 5.589447236180905e-05,
-      "loss": 4.5725,
-      "step": 44401
-    },
-    {
-      "epoch": 23.15619295958279,
-      "grad_norm": 1.544325351715088,
-      "learning_rate": 5.5893467336683426e-05,
-      "loss": 5.1971,
-      "step": 44402
-    },
-    {
-      "epoch": 23.15671447196871,
-      "grad_norm": 1.5476394891738892,
-      "learning_rate": 5.589246231155779e-05,
-      "loss": 4.8267,
-      "step": 44403
-    },
-    {
-      "epoch": 23.157235984354628,
-      "grad_norm": 1.4956297874450684,
-      "learning_rate": 5.589145728643216e-05,
-      "loss": 5.4084,
-      "step": 44404
-    },
-    {
-      "epoch": 23.157757496740548,
-      "grad_norm": 1.6533668041229248,
-      "learning_rate": 5.589045226130654e-05,
-      "loss": 5.2848,
-      "step": 44405
-    },
-    {
-      "epoch": 23.158279009126467,
-      "grad_norm": 1.4833753108978271,
-      "learning_rate": 5.58894472361809e-05,
-      "loss": 5.2977,
-      "step": 44406
-    },
-    {
-      "epoch": 23.158800521512386,
-      "grad_norm": 1.5922671556472778,
-      "learning_rate": 5.588844221105528e-05,
-      "loss": 5.2135,
-      "step": 44407
-    },
-    {
-      "epoch": 23.159322033898306,
-      "grad_norm": 1.5757970809936523,
-      "learning_rate": 5.5887437185929645e-05,
-      "loss": 5.3662,
-      "step": 44408
-    },
-    {
-      "epoch": 23.159843546284225,
-      "grad_norm": 1.6806377172470093,
-      "learning_rate": 5.588643216080402e-05,
-      "loss": 4.703,
-      "step": 44409
-    },
-    {
-      "epoch": 23.160365058670145,
-      "grad_norm": 1.5572504997253418,
-      "learning_rate": 5.5885427135678394e-05,
-      "loss": 4.8505,
-      "step": 44410
-    },
-    {
-      "epoch": 23.160886571056064,
-      "grad_norm": 1.5185115337371826,
-      "learning_rate": 5.588442211055277e-05,
-      "loss": 5.1827,
-      "step": 44411
-    },
-    {
-      "epoch": 23.16140808344198,
-      "grad_norm": 1.4492722749710083,
-      "learning_rate": 5.5883417085427136e-05,
-      "loss": 5.7801,
-      "step": 44412
-    },
-    {
-      "epoch": 23.1619295958279,
-      "grad_norm": 1.540126085281372,
-      "learning_rate": 5.5882412060301514e-05,
-      "loss": 4.996,
-      "step": 44413
-    },
-    {
-      "epoch": 23.16245110821382,
-      "grad_norm": 1.493816614151001,
-      "learning_rate": 5.588140703517588e-05,
-      "loss": 5.26,
-      "step": 44414
-    },
-    {
-      "epoch": 23.16297262059974,
-      "grad_norm": 1.6985938549041748,
-      "learning_rate": 5.5880402010050256e-05,
-      "loss": 4.8891,
-      "step": 44415
-    },
-    {
-      "epoch": 23.163494132985658,
-      "grad_norm": 1.5733305215835571,
-      "learning_rate": 5.587939698492463e-05,
-      "loss": 5.2268,
-      "step": 44416
-    },
-    {
-      "epoch": 23.164015645371578,
-      "grad_norm": 1.4618334770202637,
-      "learning_rate": 5.587839195979899e-05,
-      "loss": 5.803,
-      "step": 44417
-    },
-    {
-      "epoch": 23.164537157757497,
-      "grad_norm": 1.5179309844970703,
-      "learning_rate": 5.587738693467337e-05,
-      "loss": 5.3328,
-      "step": 44418
-    },
-    {
-      "epoch": 23.165058670143416,
-      "grad_norm": 1.5752873420715332,
-      "learning_rate": 5.587638190954774e-05,
-      "loss": 5.2079,
-      "step": 44419
-    },
-    {
-      "epoch": 23.165580182529336,
-      "grad_norm": 1.6247870922088623,
-      "learning_rate": 5.587537688442212e-05,
-      "loss": 5.001,
-      "step": 44420
-    },
-    {
-      "epoch": 23.166101694915255,
-      "grad_norm": 1.529240608215332,
-      "learning_rate": 5.587437185929648e-05,
-      "loss": 5.4824,
-      "step": 44421
-    },
-    {
-      "epoch": 23.166623207301175,
-      "grad_norm": 1.5531929731369019,
-      "learning_rate": 5.587336683417086e-05,
-      "loss": 5.0454,
-      "step": 44422
-    },
-    {
-      "epoch": 23.167144719687094,
-      "grad_norm": 1.4715605974197388,
-      "learning_rate": 5.5872361809045224e-05,
-      "loss": 5.4019,
-      "step": 44423
-    },
-    {
-      "epoch": 23.16766623207301,
-      "grad_norm": 1.606982946395874,
-      "learning_rate": 5.58713567839196e-05,
-      "loss": 4.6081,
-      "step": 44424
-    },
-    {
-      "epoch": 23.16818774445893,
-      "grad_norm": 1.568732500076294,
-      "learning_rate": 5.587035175879397e-05,
-      "loss": 5.0155,
-      "step": 44425
-    },
-    {
-      "epoch": 23.16870925684485,
-      "grad_norm": 1.5307775735855103,
-      "learning_rate": 5.586934673366835e-05,
-      "loss": 5.669,
-      "step": 44426
-    },
-    {
-      "epoch": 23.16923076923077,
-      "grad_norm": 1.5937563180923462,
-      "learning_rate": 5.5868341708542715e-05,
-      "loss": 5.6476,
-      "step": 44427
-    },
-    {
-      "epoch": 23.169752281616688,
-      "grad_norm": 1.5149341821670532,
-      "learning_rate": 5.586733668341709e-05,
-      "loss": 4.7297,
-      "step": 44428
-    },
-    {
-      "epoch": 23.170273794002608,
-      "grad_norm": 1.595287561416626,
-      "learning_rate": 5.586633165829146e-05,
-      "loss": 5.3566,
-      "step": 44429
-    },
-    {
-      "epoch": 23.170795306388527,
-      "grad_norm": 1.510013222694397,
-      "learning_rate": 5.586532663316583e-05,
-      "loss": 5.151,
-      "step": 44430
-    },
-    {
-      "epoch": 23.171316818774446,
-      "grad_norm": 1.5315250158309937,
-      "learning_rate": 5.5864321608040206e-05,
-      "loss": 5.2568,
-      "step": 44431
-    },
-    {
-      "epoch": 23.171838331160366,
-      "grad_norm": 1.5053879022598267,
-      "learning_rate": 5.586331658291457e-05,
-      "loss": 5.1245,
-      "step": 44432
-    },
-    {
-      "epoch": 23.172359843546285,
-      "grad_norm": 1.530985713005066,
-      "learning_rate": 5.586231155778895e-05,
-      "loss": 5.2794,
-      "step": 44433
-    },
-    {
-      "epoch": 23.172881355932205,
-      "grad_norm": 1.506066918373108,
-      "learning_rate": 5.586130653266331e-05,
-      "loss": 5.4207,
-      "step": 44434
-    },
-    {
-      "epoch": 23.17340286831812,
-      "grad_norm": 1.517920970916748,
-      "learning_rate": 5.586030150753769e-05,
-      "loss": 5.116,
-      "step": 44435
-    },
-    {
-      "epoch": 23.17392438070404,
-      "grad_norm": 1.6365300416946411,
-      "learning_rate": 5.585929648241206e-05,
-      "loss": 5.2444,
-      "step": 44436
-    },
-    {
-      "epoch": 23.17444589308996,
-      "grad_norm": 1.533740758895874,
-      "learning_rate": 5.585829145728644e-05,
-      "loss": 5.4651,
-      "step": 44437
-    },
-    {
-      "epoch": 23.17496740547588,
-      "grad_norm": 1.5531818866729736,
-      "learning_rate": 5.5857286432160804e-05,
-      "loss": 5.0552,
-      "step": 44438
-    },
-    {
-      "epoch": 23.1754889178618,
-      "grad_norm": 1.5670596361160278,
-      "learning_rate": 5.585628140703518e-05,
-      "loss": 5.2571,
-      "step": 44439
-    },
-    {
-      "epoch": 23.176010430247718,
-      "grad_norm": 1.482662320137024,
-      "learning_rate": 5.585527638190955e-05,
-      "loss": 5.3953,
-      "step": 44440
-    },
-    {
-      "epoch": 23.176531942633638,
-      "grad_norm": 1.5225999355316162,
-      "learning_rate": 5.585427135678393e-05,
-      "loss": 5.5203,
-      "step": 44441
-    },
-    {
-      "epoch": 23.177053455019557,
-      "grad_norm": 1.5714598894119263,
-      "learning_rate": 5.5853266331658295e-05,
-      "loss": 5.2719,
-      "step": 44442
-    },
-    {
-      "epoch": 23.177574967405477,
-      "grad_norm": 1.7030725479125977,
-      "learning_rate": 5.585226130653266e-05,
-      "loss": 5.0575,
-      "step": 44443
-    },
-    {
-      "epoch": 23.178096479791396,
-      "grad_norm": 1.5519673824310303,
-      "learning_rate": 5.585125628140704e-05,
-      "loss": 4.963,
-      "step": 44444
-    },
-    {
-      "epoch": 23.178617992177315,
-      "grad_norm": 1.4521359205245972,
-      "learning_rate": 5.585025125628141e-05,
-      "loss": 5.2794,
-      "step": 44445
-    },
-    {
-      "epoch": 23.179139504563235,
-      "grad_norm": 1.5054372549057007,
-      "learning_rate": 5.5849246231155786e-05,
-      "loss": 5.0013,
-      "step": 44446
-    },
-    {
-      "epoch": 23.17966101694915,
-      "grad_norm": 1.4701169729232788,
-      "learning_rate": 5.584824120603015e-05,
-      "loss": 4.7161,
-      "step": 44447
-    },
-    {
-      "epoch": 23.18018252933507,
-      "grad_norm": 1.5844577550888062,
-      "learning_rate": 5.584723618090453e-05,
-      "loss": 5.6168,
-      "step": 44448
-    },
-    {
-      "epoch": 23.18070404172099,
-      "grad_norm": 1.6560114622116089,
-      "learning_rate": 5.584623115577889e-05,
-      "loss": 5.2664,
-      "step": 44449
-    },
-    {
-      "epoch": 23.18122555410691,
-      "grad_norm": 1.6769298315048218,
-      "learning_rate": 5.584522613065327e-05,
-      "loss": 5.5789,
-      "step": 44450
-    },
-    {
-      "epoch": 23.18174706649283,
-      "grad_norm": 1.581703782081604,
-      "learning_rate": 5.584422110552764e-05,
-      "loss": 5.6421,
-      "step": 44451
-    },
-    {
-      "epoch": 23.182268578878748,
-      "grad_norm": 1.4715690612792969,
-      "learning_rate": 5.584321608040202e-05,
-      "loss": 5.4105,
-      "step": 44452
-    },
-    {
-      "epoch": 23.182790091264668,
-      "grad_norm": 1.5141658782958984,
-      "learning_rate": 5.584221105527638e-05,
-      "loss": 4.9959,
-      "step": 44453
-    },
-    {
-      "epoch": 23.183311603650587,
-      "grad_norm": 1.4889776706695557,
-      "learning_rate": 5.584120603015076e-05,
-      "loss": 4.7765,
-      "step": 44454
-    },
-    {
-      "epoch": 23.183833116036507,
-      "grad_norm": 1.7164026498794556,
-      "learning_rate": 5.5840201005025125e-05,
-      "loss": 4.8498,
-      "step": 44455
-    },
-    {
-      "epoch": 23.184354628422426,
-      "grad_norm": 1.5603432655334473,
-      "learning_rate": 5.58391959798995e-05,
-      "loss": 4.6686,
-      "step": 44456
-    },
-    {
-      "epoch": 23.184876140808345,
-      "grad_norm": 1.697520136833191,
-      "learning_rate": 5.5838190954773874e-05,
-      "loss": 4.1408,
-      "step": 44457
-    },
-    {
-      "epoch": 23.185397653194265,
-      "grad_norm": 1.5019550323486328,
-      "learning_rate": 5.583718592964824e-05,
-      "loss": 5.6475,
-      "step": 44458
-    },
-    {
-      "epoch": 23.18591916558018,
-      "grad_norm": 1.5064529180526733,
-      "learning_rate": 5.5836180904522616e-05,
-      "loss": 5.7946,
-      "step": 44459
-    },
-    {
-      "epoch": 23.1864406779661,
-      "grad_norm": 1.7036986351013184,
-      "learning_rate": 5.583517587939698e-05,
-      "loss": 5.3093,
-      "step": 44460
-    },
-    {
-      "epoch": 23.18696219035202,
-      "grad_norm": 1.6460987329483032,
-      "learning_rate": 5.583417085427136e-05,
-      "loss": 5.4136,
-      "step": 44461
-    },
-    {
-      "epoch": 23.18748370273794,
-      "grad_norm": 1.5598304271697998,
-      "learning_rate": 5.583316582914573e-05,
-      "loss": 5.3218,
-      "step": 44462
-    },
-    {
-      "epoch": 23.18800521512386,
-      "grad_norm": 1.5708069801330566,
-      "learning_rate": 5.583216080402011e-05,
-      "loss": 4.8947,
-      "step": 44463
-    },
-    {
-      "epoch": 23.188526727509778,
-      "grad_norm": 1.5182583332061768,
-      "learning_rate": 5.583115577889447e-05,
-      "loss": 5.3932,
-      "step": 44464
-    },
-    {
-      "epoch": 23.189048239895698,
-      "grad_norm": 1.5008070468902588,
-      "learning_rate": 5.583015075376885e-05,
-      "loss": 5.3346,
-      "step": 44465
-    },
-    {
-      "epoch": 23.189569752281617,
-      "grad_norm": 1.5545766353607178,
-      "learning_rate": 5.582914572864322e-05,
-      "loss": 4.873,
-      "step": 44466
-    },
-    {
-      "epoch": 23.190091264667537,
-      "grad_norm": 1.611024022102356,
-      "learning_rate": 5.58281407035176e-05,
-      "loss": 5.224,
-      "step": 44467
-    },
-    {
-      "epoch": 23.190612777053456,
-      "grad_norm": 1.59691321849823,
-      "learning_rate": 5.582713567839196e-05,
-      "loss": 5.3982,
-      "step": 44468
-    },
-    {
-      "epoch": 23.191134289439375,
-      "grad_norm": 1.4951989650726318,
-      "learning_rate": 5.582613065326634e-05,
-      "loss": 5.0871,
-      "step": 44469
-    },
-    {
-      "epoch": 23.191655801825295,
-      "grad_norm": 1.46759033203125,
-      "learning_rate": 5.5825125628140704e-05,
-      "loss": 4.6307,
-      "step": 44470
-    },
-    {
-      "epoch": 23.19217731421121,
-      "grad_norm": 1.5732370615005493,
-      "learning_rate": 5.5824120603015076e-05,
-      "loss": 4.9385,
-      "step": 44471
-    },
-    {
-      "epoch": 23.19269882659713,
-      "grad_norm": 1.5035287141799927,
-      "learning_rate": 5.582311557788945e-05,
-      "loss": 5.1477,
-      "step": 44472
-    },
-    {
-      "epoch": 23.19322033898305,
-      "grad_norm": 1.554451584815979,
-      "learning_rate": 5.582211055276382e-05,
-      "loss": 5.5065,
-      "step": 44473
-    },
-    {
-      "epoch": 23.19374185136897,
-      "grad_norm": 1.5065127611160278,
-      "learning_rate": 5.5821105527638195e-05,
-      "loss": 5.5938,
-      "step": 44474
-    },
-    {
-      "epoch": 23.19426336375489,
-      "grad_norm": 1.661399483680725,
-      "learning_rate": 5.582010050251256e-05,
-      "loss": 5.2313,
-      "step": 44475
-    },
-    {
-      "epoch": 23.194784876140808,
-      "grad_norm": 1.7264972925186157,
-      "learning_rate": 5.581909547738694e-05,
-      "loss": 4.737,
-      "step": 44476
-    },
-    {
-      "epoch": 23.195306388526728,
-      "grad_norm": 1.5699737071990967,
-      "learning_rate": 5.581809045226131e-05,
-      "loss": 4.8192,
-      "step": 44477
-    },
-    {
-      "epoch": 23.195827900912647,
-      "grad_norm": 1.6524354219436646,
-      "learning_rate": 5.5817085427135686e-05,
-      "loss": 5.2304,
-      "step": 44478
-    },
-    {
-      "epoch": 23.196349413298567,
-      "grad_norm": 1.584173321723938,
-      "learning_rate": 5.581608040201005e-05,
-      "loss": 4.9043,
-      "step": 44479
-    },
-    {
-      "epoch": 23.196870925684486,
-      "grad_norm": 1.4571678638458252,
-      "learning_rate": 5.581507537688443e-05,
-      "loss": 4.0244,
-      "step": 44480
-    },
-    {
-      "epoch": 23.197392438070406,
-      "grad_norm": 1.5093358755111694,
-      "learning_rate": 5.581407035175879e-05,
-      "loss": 4.8153,
-      "step": 44481
-    },
-    {
-      "epoch": 23.197913950456325,
-      "grad_norm": 1.523886799812317,
-      "learning_rate": 5.581306532663317e-05,
-      "loss": 5.4279,
-      "step": 44482
-    },
-    {
-      "epoch": 23.19843546284224,
-      "grad_norm": 1.587496042251587,
-      "learning_rate": 5.581206030150754e-05,
-      "loss": 4.7321,
-      "step": 44483
-    },
-    {
-      "epoch": 23.19895697522816,
-      "grad_norm": 1.7043652534484863,
-      "learning_rate": 5.5811055276381906e-05,
-      "loss": 4.759,
-      "step": 44484
-    },
-    {
-      "epoch": 23.19947848761408,
-      "grad_norm": 1.701104760169983,
-      "learning_rate": 5.5810050251256284e-05,
-      "loss": 4.3673,
-      "step": 44485
-    },
-    {
-      "epoch": 23.2,
-      "grad_norm": 1.7615993022918701,
-      "learning_rate": 5.5809045226130655e-05,
-      "loss": 5.1632,
-      "step": 44486
-    },
-    {
-      "epoch": 23.20052151238592,
-      "grad_norm": 1.6056160926818848,
-      "learning_rate": 5.580804020100503e-05,
-      "loss": 4.8911,
-      "step": 44487
-    },
-    {
-      "epoch": 23.201043024771838,
-      "grad_norm": 1.561873197555542,
-      "learning_rate": 5.58070351758794e-05,
-      "loss": 5.2888,
-      "step": 44488
-    },
-    {
-      "epoch": 23.201564537157758,
-      "grad_norm": 1.5198965072631836,
-      "learning_rate": 5.5806030150753775e-05,
-      "loss": 4.9964,
-      "step": 44489
-    },
-    {
-      "epoch": 23.202086049543677,
-      "grad_norm": 1.4538987874984741,
-      "learning_rate": 5.580502512562814e-05,
-      "loss": 5.2676,
-      "step": 44490
-    },
-    {
-      "epoch": 23.202607561929597,
-      "grad_norm": 1.5355035066604614,
-      "learning_rate": 5.580402010050252e-05,
-      "loss": 5.211,
-      "step": 44491
-    },
-    {
-      "epoch": 23.203129074315516,
-      "grad_norm": 1.4796754121780396,
-      "learning_rate": 5.580301507537689e-05,
-      "loss": 5.4725,
-      "step": 44492
-    },
-    {
-      "epoch": 23.203650586701436,
-      "grad_norm": 1.5519839525222778,
-      "learning_rate": 5.5802010050251266e-05,
-      "loss": 5.4003,
-      "step": 44493
-    },
-    {
-      "epoch": 23.204172099087355,
-      "grad_norm": 1.517124056816101,
-      "learning_rate": 5.580100502512563e-05,
-      "loss": 5.5417,
-      "step": 44494
-    },
-    {
-      "epoch": 23.20469361147327,
-      "grad_norm": 1.506738305091858,
-      "learning_rate": 5.580000000000001e-05,
-      "loss": 5.5064,
-      "step": 44495
-    },
-    {
-      "epoch": 23.20521512385919,
-      "grad_norm": 1.531636357307434,
-      "learning_rate": 5.579899497487437e-05,
-      "loss": 5.1309,
-      "step": 44496
-    },
-    {
-      "epoch": 23.20573663624511,
-      "grad_norm": 1.5367642641067505,
-      "learning_rate": 5.579798994974874e-05,
-      "loss": 5.1897,
-      "step": 44497
-    },
-    {
-      "epoch": 23.20625814863103,
-      "grad_norm": 1.6637266874313354,
-      "learning_rate": 5.579698492462312e-05,
-      "loss": 4.9355,
-      "step": 44498
-    },
-    {
-      "epoch": 23.20677966101695,
-      "grad_norm": 1.6399122476577759,
-      "learning_rate": 5.5795979899497485e-05,
-      "loss": 5.1819,
-      "step": 44499
-    },
-    {
-      "epoch": 23.20730117340287,
-      "grad_norm": 1.6781072616577148,
-      "learning_rate": 5.579497487437186e-05,
-      "loss": 4.8043,
-      "step": 44500
-    },
-    {
-      "epoch": 23.207822685788788,
-      "grad_norm": 1.6894584894180298,
-      "learning_rate": 5.579396984924623e-05,
-      "loss": 4.621,
-      "step": 44501
-    },
-    {
-      "epoch": 23.208344198174707,
-      "grad_norm": 1.5262281894683838,
-      "learning_rate": 5.5792964824120605e-05,
-      "loss": 4.9033,
-      "step": 44502
-    },
-    {
-      "epoch": 23.208865710560627,
-      "grad_norm": 1.6599876880645752,
-      "learning_rate": 5.5791959798994976e-05,
-      "loss": 5.0329,
-      "step": 44503
-    },
-    {
-      "epoch": 23.209387222946546,
-      "grad_norm": 1.5819079875946045,
-      "learning_rate": 5.5790954773869354e-05,
-      "loss": 5.4085,
-      "step": 44504
-    },
-    {
-      "epoch": 23.209908735332466,
-      "grad_norm": 1.7141226530075073,
-      "learning_rate": 5.578994974874372e-05,
-      "loss": 5.1057,
-      "step": 44505
-    },
-    {
-      "epoch": 23.210430247718385,
-      "grad_norm": 1.5411040782928467,
-      "learning_rate": 5.5788944723618096e-05,
-      "loss": 5.3083,
-      "step": 44506
-    },
-    {
-      "epoch": 23.2109517601043,
-      "grad_norm": 1.556691288948059,
-      "learning_rate": 5.578793969849247e-05,
-      "loss": 5.7389,
-      "step": 44507
-    },
-    {
-      "epoch": 23.21147327249022,
-      "grad_norm": 1.6186749935150146,
-      "learning_rate": 5.5786934673366845e-05,
-      "loss": 4.9765,
-      "step": 44508
-    },
-    {
-      "epoch": 23.21199478487614,
-      "grad_norm": 1.490837574005127,
-      "learning_rate": 5.578592964824121e-05,
-      "loss": 5.3557,
-      "step": 44509
-    },
-    {
-      "epoch": 23.21251629726206,
-      "grad_norm": 1.619564175605774,
-      "learning_rate": 5.5784924623115574e-05,
-      "loss": 5.4258,
-      "step": 44510
-    },
-    {
-      "epoch": 23.21303780964798,
-      "grad_norm": 1.5919963121414185,
-      "learning_rate": 5.578391959798995e-05,
-      "loss": 5.5385,
-      "step": 44511
-    },
-    {
-      "epoch": 23.2135593220339,
-      "grad_norm": 1.6380077600479126,
-      "learning_rate": 5.578291457286432e-05,
-      "loss": 5.3102,
-      "step": 44512
-    },
-    {
-      "epoch": 23.214080834419818,
-      "grad_norm": 1.5933871269226074,
-      "learning_rate": 5.57819095477387e-05,
-      "loss": 4.7929,
-      "step": 44513
-    },
-    {
-      "epoch": 23.214602346805737,
-      "grad_norm": 1.5525047779083252,
-      "learning_rate": 5.5780904522613065e-05,
-      "loss": 5.0254,
-      "step": 44514
-    },
-    {
-      "epoch": 23.215123859191657,
-      "grad_norm": 1.5207644701004028,
-      "learning_rate": 5.577989949748744e-05,
-      "loss": 5.4492,
-      "step": 44515
-    },
-    {
-      "epoch": 23.215645371577576,
-      "grad_norm": 1.5767146348953247,
-      "learning_rate": 5.577889447236181e-05,
-      "loss": 5.7119,
-      "step": 44516
-    },
-    {
-      "epoch": 23.216166883963496,
-      "grad_norm": 1.5263234376907349,
-      "learning_rate": 5.5777889447236185e-05,
-      "loss": 5.3596,
-      "step": 44517
-    },
-    {
-      "epoch": 23.216688396349415,
-      "grad_norm": 1.5918166637420654,
-      "learning_rate": 5.5776884422110556e-05,
-      "loss": 5.184,
-      "step": 44518
-    },
-    {
-      "epoch": 23.21720990873533,
-      "grad_norm": 1.643988013267517,
-      "learning_rate": 5.5775879396984933e-05,
-      "loss": 4.4758,
-      "step": 44519
-    },
-    {
-      "epoch": 23.21773142112125,
-      "grad_norm": 1.5149116516113281,
-      "learning_rate": 5.57748743718593e-05,
-      "loss": 5.7059,
-      "step": 44520
-    },
-    {
-      "epoch": 23.21825293350717,
-      "grad_norm": 1.594844937324524,
-      "learning_rate": 5.5773869346733676e-05,
-      "loss": 5.2617,
-      "step": 44521
-    },
-    {
-      "epoch": 23.21877444589309,
-      "grad_norm": 1.43764066696167,
-      "learning_rate": 5.577286432160804e-05,
-      "loss": 5.2321,
-      "step": 44522
-    },
-    {
-      "epoch": 23.21929595827901,
-      "grad_norm": 1.481082797050476,
-      "learning_rate": 5.577185929648241e-05,
-      "loss": 5.6576,
-      "step": 44523
-    },
-    {
-      "epoch": 23.21981747066493,
-      "grad_norm": 1.6023286581039429,
-      "learning_rate": 5.577085427135679e-05,
-      "loss": 4.7092,
-      "step": 44524
-    },
-    {
-      "epoch": 23.220338983050848,
-      "grad_norm": 1.5849754810333252,
-      "learning_rate": 5.576984924623115e-05,
-      "loss": 5.287,
-      "step": 44525
-    },
-    {
-      "epoch": 23.220860495436767,
-      "grad_norm": 1.6785475015640259,
-      "learning_rate": 5.576884422110553e-05,
-      "loss": 5.2086,
-      "step": 44526
-    },
-    {
-      "epoch": 23.221382007822687,
-      "grad_norm": 1.5969887971878052,
-      "learning_rate": 5.57678391959799e-05,
-      "loss": 5.2268,
-      "step": 44527
-    },
-    {
-      "epoch": 23.221903520208606,
-      "grad_norm": 1.5361576080322266,
-      "learning_rate": 5.576683417085428e-05,
-      "loss": 5.1579,
-      "step": 44528
-    },
-    {
-      "epoch": 23.222425032594526,
-      "grad_norm": 1.6871178150177002,
-      "learning_rate": 5.5765829145728644e-05,
-      "loss": 4.6787,
-      "step": 44529
-    },
-    {
-      "epoch": 23.22294654498044,
-      "grad_norm": 1.7713544368743896,
-      "learning_rate": 5.576482412060302e-05,
-      "loss": 4.4551,
-      "step": 44530
-    },
-    {
-      "epoch": 23.22346805736636,
-      "grad_norm": 1.6521567106246948,
-      "learning_rate": 5.5763819095477386e-05,
-      "loss": 5.434,
-      "step": 44531
-    },
-    {
-      "epoch": 23.22398956975228,
-      "grad_norm": 1.587735652923584,
-      "learning_rate": 5.5762814070351764e-05,
-      "loss": 4.6837,
-      "step": 44532
-    },
-    {
-      "epoch": 23.2245110821382,
-      "grad_norm": 1.5835282802581787,
-      "learning_rate": 5.5761809045226135e-05,
-      "loss": 5.4976,
-      "step": 44533
-    },
-    {
-      "epoch": 23.22503259452412,
-      "grad_norm": 1.5470564365386963,
-      "learning_rate": 5.576080402010051e-05,
-      "loss": 4.704,
-      "step": 44534
-    },
-    {
-      "epoch": 23.22555410691004,
-      "grad_norm": 1.5559953451156616,
-      "learning_rate": 5.575979899497488e-05,
-      "loss": 5.1195,
-      "step": 44535
-    },
-    {
-      "epoch": 23.22607561929596,
-      "grad_norm": 1.6401594877243042,
-      "learning_rate": 5.575879396984924e-05,
-      "loss": 5.3782,
-      "step": 44536
-    },
-    {
-      "epoch": 23.226597131681878,
-      "grad_norm": 1.5583429336547852,
-      "learning_rate": 5.575778894472362e-05,
-      "loss": 5.1234,
-      "step": 44537
-    },
-    {
-      "epoch": 23.227118644067797,
-      "grad_norm": 1.5059813261032104,
-      "learning_rate": 5.575678391959799e-05,
-      "loss": 4.9452,
-      "step": 44538
-    },
-    {
-      "epoch": 23.227640156453717,
-      "grad_norm": 1.5531635284423828,
-      "learning_rate": 5.575577889447237e-05,
-      "loss": 5.2235,
-      "step": 44539
-    },
-    {
-      "epoch": 23.228161668839636,
-      "grad_norm": 1.48867666721344,
-      "learning_rate": 5.575477386934673e-05,
-      "loss": 5.2885,
-      "step": 44540
-    },
-    {
-      "epoch": 23.228683181225556,
-      "grad_norm": 1.5812493562698364,
-      "learning_rate": 5.575376884422111e-05,
-      "loss": 5.0846,
-      "step": 44541
-    },
-    {
-      "epoch": 23.22920469361147,
-      "grad_norm": 1.6124461889266968,
-      "learning_rate": 5.5752763819095474e-05,
-      "loss": 5.5369,
-      "step": 44542
-    },
-    {
-      "epoch": 23.22972620599739,
-      "grad_norm": 1.521266222000122,
-      "learning_rate": 5.575175879396985e-05,
-      "loss": 5.0622,
-      "step": 44543
-    },
-    {
-      "epoch": 23.23024771838331,
-      "grad_norm": 1.5308573246002197,
-      "learning_rate": 5.575075376884422e-05,
-      "loss": 5.4341,
-      "step": 44544
-    },
-    {
-      "epoch": 23.23076923076923,
-      "grad_norm": 1.633890151977539,
-      "learning_rate": 5.57497487437186e-05,
-      "loss": 5.1463,
-      "step": 44545
-    },
-    {
-      "epoch": 23.23129074315515,
-      "grad_norm": 1.4661734104156494,
-      "learning_rate": 5.5748743718592965e-05,
-      "loss": 5.1726,
-      "step": 44546
-    },
-    {
-      "epoch": 23.23181225554107,
-      "grad_norm": 1.6232037544250488,
-      "learning_rate": 5.574773869346734e-05,
-      "loss": 5.1109,
-      "step": 44547
-    },
-    {
-      "epoch": 23.23233376792699,
-      "grad_norm": 1.5435011386871338,
-      "learning_rate": 5.574673366834171e-05,
-      "loss": 5.6018,
-      "step": 44548
-    },
-    {
-      "epoch": 23.232855280312908,
-      "grad_norm": 1.5316187143325806,
-      "learning_rate": 5.5745728643216085e-05,
-      "loss": 4.92,
-      "step": 44549
-    },
-    {
-      "epoch": 23.233376792698827,
-      "grad_norm": 1.7112205028533936,
-      "learning_rate": 5.5744723618090456e-05,
-      "loss": 5.3565,
-      "step": 44550
-    },
-    {
-      "epoch": 23.233898305084747,
-      "grad_norm": 1.6411887407302856,
-      "learning_rate": 5.574371859296482e-05,
-      "loss": 4.9356,
-      "step": 44551
-    },
-    {
-      "epoch": 23.234419817470666,
-      "grad_norm": 1.5089813470840454,
-      "learning_rate": 5.57427135678392e-05,
-      "loss": 4.9833,
-      "step": 44552
-    },
-    {
-      "epoch": 23.234941329856586,
-      "grad_norm": 1.590040922164917,
-      "learning_rate": 5.574170854271357e-05,
-      "loss": 4.9826,
-      "step": 44553
-    },
-    {
-      "epoch": 23.2354628422425,
-      "grad_norm": 1.4724701642990112,
-      "learning_rate": 5.574070351758795e-05,
-      "loss": 5.2234,
-      "step": 44554
-    },
-    {
-      "epoch": 23.23598435462842,
-      "grad_norm": 1.4800435304641724,
-      "learning_rate": 5.573969849246231e-05,
-      "loss": 5.4086,
-      "step": 44555
-    },
-    {
-      "epoch": 23.23650586701434,
-      "grad_norm": 1.6484026908874512,
-      "learning_rate": 5.573869346733669e-05,
-      "loss": 5.2167,
-      "step": 44556
-    },
-    {
-      "epoch": 23.23702737940026,
-      "grad_norm": 1.492709755897522,
-      "learning_rate": 5.5737688442211054e-05,
-      "loss": 5.3841,
-      "step": 44557
-    },
-    {
-      "epoch": 23.23754889178618,
-      "grad_norm": 1.666269063949585,
-      "learning_rate": 5.573668341708543e-05,
-      "loss": 4.9408,
-      "step": 44558
-    },
-    {
-      "epoch": 23.2380704041721,
-      "grad_norm": 1.4974778890609741,
-      "learning_rate": 5.57356783919598e-05,
-      "loss": 5.5913,
-      "step": 44559
-    },
-    {
-      "epoch": 23.23859191655802,
-      "grad_norm": 1.5328329801559448,
-      "learning_rate": 5.573467336683418e-05,
-      "loss": 5.6629,
-      "step": 44560
-    },
-    {
-      "epoch": 23.239113428943938,
-      "grad_norm": 1.5449854135513306,
-      "learning_rate": 5.5733668341708545e-05,
-      "loss": 5.3747,
-      "step": 44561
-    },
-    {
-      "epoch": 23.239634941329857,
-      "grad_norm": 1.439741611480713,
-      "learning_rate": 5.573266331658292e-05,
-      "loss": 5.5506,
-      "step": 44562
-    },
-    {
-      "epoch": 23.240156453715777,
-      "grad_norm": 1.5875928401947021,
-      "learning_rate": 5.573165829145729e-05,
-      "loss": 5.5345,
-      "step": 44563
-    },
-    {
-      "epoch": 23.240677966101696,
-      "grad_norm": 1.6554032564163208,
-      "learning_rate": 5.573065326633166e-05,
-      "loss": 5.0586,
-      "step": 44564
-    },
-    {
-      "epoch": 23.241199478487616,
-      "grad_norm": 1.599313497543335,
-      "learning_rate": 5.5729648241206036e-05,
-      "loss": 5.2891,
-      "step": 44565
-    },
-    {
-      "epoch": 23.24172099087353,
-      "grad_norm": 1.510029911994934,
-      "learning_rate": 5.57286432160804e-05,
-      "loss": 5.3181,
-      "step": 44566
-    },
-    {
-      "epoch": 23.24224250325945,
-      "grad_norm": 1.562924861907959,
-      "learning_rate": 5.572763819095478e-05,
-      "loss": 5.3408,
-      "step": 44567
-    },
-    {
-      "epoch": 23.24276401564537,
-      "grad_norm": 1.5319468975067139,
-      "learning_rate": 5.572663316582914e-05,
-      "loss": 5.6065,
-      "step": 44568
-    },
-    {
-      "epoch": 23.24328552803129,
-      "grad_norm": 1.5445629358291626,
-      "learning_rate": 5.572562814070352e-05,
-      "loss": 4.9126,
-      "step": 44569
-    },
-    {
-      "epoch": 23.24380704041721,
-      "grad_norm": 1.4859198331832886,
-      "learning_rate": 5.572462311557789e-05,
-      "loss": 5.2278,
-      "step": 44570
-    },
-    {
-      "epoch": 23.24432855280313,
-      "grad_norm": 1.520655870437622,
-      "learning_rate": 5.572361809045227e-05,
-      "loss": 5.4659,
-      "step": 44571
-    },
-    {
-      "epoch": 23.24485006518905,
-      "grad_norm": 1.532092809677124,
-      "learning_rate": 5.572261306532663e-05,
-      "loss": 5.0158,
-      "step": 44572
-    },
-    {
-      "epoch": 23.245371577574968,
-      "grad_norm": 1.5513187646865845,
-      "learning_rate": 5.572160804020101e-05,
-      "loss": 5.4661,
-      "step": 44573
-    },
-    {
-      "epoch": 23.245893089960887,
-      "grad_norm": 1.5081251859664917,
-      "learning_rate": 5.572060301507538e-05,
-      "loss": 5.2861,
-      "step": 44574
-    },
-    {
-      "epoch": 23.246414602346807,
-      "grad_norm": 1.651164174079895,
-      "learning_rate": 5.571959798994976e-05,
-      "loss": 5.1333,
-      "step": 44575
-    },
-    {
-      "epoch": 23.246936114732726,
-      "grad_norm": 1.5426654815673828,
-      "learning_rate": 5.5718592964824124e-05,
-      "loss": 5.3585,
-      "step": 44576
-    },
-    {
-      "epoch": 23.247457627118646,
-      "grad_norm": 1.5087411403656006,
-      "learning_rate": 5.571758793969849e-05,
-      "loss": 5.3314,
-      "step": 44577
-    },
-    {
-      "epoch": 23.24797913950456,
-      "grad_norm": 1.5931810140609741,
-      "learning_rate": 5.5716582914572866e-05,
-      "loss": 5.3147,
-      "step": 44578
-    },
-    {
-      "epoch": 23.24850065189048,
-      "grad_norm": 1.5942541360855103,
-      "learning_rate": 5.571557788944724e-05,
-      "loss": 4.936,
-      "step": 44579
-    },
-    {
-      "epoch": 23.2490221642764,
-      "grad_norm": 1.5566232204437256,
-      "learning_rate": 5.5714572864321615e-05,
-      "loss": 5.4375,
-      "step": 44580
-    },
-    {
-      "epoch": 23.24954367666232,
-      "grad_norm": 1.5772749185562134,
-      "learning_rate": 5.571356783919598e-05,
-      "loss": 5.3146,
-      "step": 44581
-    },
-    {
-      "epoch": 23.25006518904824,
-      "grad_norm": 1.6166068315505981,
-      "learning_rate": 5.571256281407036e-05,
-      "loss": 4.1991,
-      "step": 44582
-    },
-    {
-      "epoch": 23.25058670143416,
-      "grad_norm": 1.603832483291626,
-      "learning_rate": 5.571155778894472e-05,
-      "loss": 5.1745,
-      "step": 44583
-    },
-    {
-      "epoch": 23.25110821382008,
-      "grad_norm": 1.4461263418197632,
-      "learning_rate": 5.57105527638191e-05,
-      "loss": 5.3029,
-      "step": 44584
-    },
-    {
-      "epoch": 23.251629726205998,
-      "grad_norm": 1.4724007844924927,
-      "learning_rate": 5.570954773869347e-05,
-      "loss": 5.3787,
-      "step": 44585
-    },
-    {
-      "epoch": 23.252151238591917,
-      "grad_norm": 1.7117221355438232,
-      "learning_rate": 5.570854271356785e-05,
-      "loss": 4.7482,
-      "step": 44586
-    },
-    {
-      "epoch": 23.252672750977837,
-      "grad_norm": 1.4404044151306152,
-      "learning_rate": 5.570753768844221e-05,
-      "loss": 5.7402,
-      "step": 44587
-    },
-    {
-      "epoch": 23.253194263363756,
-      "grad_norm": 1.5791313648223877,
-      "learning_rate": 5.570653266331659e-05,
-      "loss": 5.2975,
-      "step": 44588
-    },
-    {
-      "epoch": 23.253715775749676,
-      "grad_norm": 1.4962552785873413,
-      "learning_rate": 5.5705527638190954e-05,
-      "loss": 4.9357,
-      "step": 44589
-    },
-    {
-      "epoch": 23.25423728813559,
-      "grad_norm": 1.474942684173584,
-      "learning_rate": 5.5704522613065326e-05,
-      "loss": 5.0679,
-      "step": 44590
-    },
-    {
-      "epoch": 23.25475880052151,
-      "grad_norm": 1.4585527181625366,
-      "learning_rate": 5.57035175879397e-05,
-      "loss": 5.331,
-      "step": 44591
-    },
-    {
-      "epoch": 23.25528031290743,
-      "grad_norm": 1.5089999437332153,
-      "learning_rate": 5.570251256281407e-05,
-      "loss": 5.6188,
-      "step": 44592
-    },
-    {
-      "epoch": 23.25580182529335,
-      "grad_norm": 1.4410741329193115,
-      "learning_rate": 5.5701507537688445e-05,
-      "loss": 5.8025,
-      "step": 44593
-    },
-    {
-      "epoch": 23.25632333767927,
-      "grad_norm": 1.4977535009384155,
-      "learning_rate": 5.5700502512562817e-05,
-      "loss": 5.4563,
-      "step": 44594
-    },
-    {
-      "epoch": 23.25684485006519,
-      "grad_norm": 1.5469807386398315,
-      "learning_rate": 5.5699497487437194e-05,
-      "loss": 5.3378,
-      "step": 44595
-    },
-    {
-      "epoch": 23.25736636245111,
-      "grad_norm": 1.6360353231430054,
-      "learning_rate": 5.569849246231156e-05,
-      "loss": 5.3596,
-      "step": 44596
-    },
-    {
-      "epoch": 23.257887874837028,
-      "grad_norm": 1.6265379190444946,
-      "learning_rate": 5.5697487437185936e-05,
-      "loss": 5.3563,
-      "step": 44597
-    },
-    {
-      "epoch": 23.258409387222947,
-      "grad_norm": 1.667091965675354,
-      "learning_rate": 5.56964824120603e-05,
-      "loss": 4.49,
-      "step": 44598
-    },
-    {
-      "epoch": 23.258930899608867,
-      "grad_norm": 1.6011992692947388,
-      "learning_rate": 5.569547738693468e-05,
-      "loss": 4.8326,
-      "step": 44599
-    },
-    {
-      "epoch": 23.259452411994786,
-      "grad_norm": 1.5255705118179321,
-      "learning_rate": 5.569447236180905e-05,
-      "loss": 5.4961,
-      "step": 44600
-    },
-    {
-      "epoch": 23.259973924380706,
-      "grad_norm": 1.6424628496170044,
-      "learning_rate": 5.569346733668343e-05,
-      "loss": 4.845,
-      "step": 44601
-    },
-    {
-      "epoch": 23.26049543676662,
-      "grad_norm": 1.6299266815185547,
-      "learning_rate": 5.569246231155779e-05,
-      "loss": 4.9694,
-      "step": 44602
-    },
-    {
-      "epoch": 23.26101694915254,
-      "grad_norm": 1.5316828489303589,
-      "learning_rate": 5.5691457286432156e-05,
-      "loss": 5.757,
-      "step": 44603
-    },
-    {
-      "epoch": 23.26153846153846,
-      "grad_norm": 1.5627888441085815,
-      "learning_rate": 5.5690452261306534e-05,
-      "loss": 4.6872,
-      "step": 44604
-    },
-    {
-      "epoch": 23.26205997392438,
-      "grad_norm": 1.5855426788330078,
-      "learning_rate": 5.5689447236180905e-05,
-      "loss": 5.2008,
-      "step": 44605
-    },
-    {
-      "epoch": 23.2625814863103,
-      "grad_norm": 1.5990697145462036,
-      "learning_rate": 5.568844221105528e-05,
-      "loss": 4.9874,
-      "step": 44606
-    },
-    {
-      "epoch": 23.26310299869622,
-      "grad_norm": 1.4784404039382935,
-      "learning_rate": 5.568743718592965e-05,
-      "loss": 5.4411,
-      "step": 44607
-    },
-    {
-      "epoch": 23.26362451108214,
-      "grad_norm": 1.7317475080490112,
-      "learning_rate": 5.5686432160804025e-05,
-      "loss": 4.9679,
-      "step": 44608
-    },
-    {
-      "epoch": 23.264146023468058,
-      "grad_norm": 1.6227879524230957,
-      "learning_rate": 5.568542713567839e-05,
-      "loss": 4.8079,
-      "step": 44609
-    },
-    {
-      "epoch": 23.264667535853977,
-      "grad_norm": 1.6126012802124023,
-      "learning_rate": 5.568442211055277e-05,
-      "loss": 4.8223,
-      "step": 44610
-    },
-    {
-      "epoch": 23.265189048239897,
-      "grad_norm": 1.4931504726409912,
-      "learning_rate": 5.568341708542714e-05,
-      "loss": 5.1269,
-      "step": 44611
-    },
-    {
-      "epoch": 23.265710560625816,
-      "grad_norm": 1.596244215965271,
-      "learning_rate": 5.5682412060301516e-05,
-      "loss": 5.1975,
-      "step": 44612
-    },
-    {
-      "epoch": 23.266232073011736,
-      "grad_norm": 1.5338290929794312,
-      "learning_rate": 5.568140703517588e-05,
-      "loss": 4.8642,
-      "step": 44613
-    },
-    {
-      "epoch": 23.26675358539765,
-      "grad_norm": 1.5319815874099731,
-      "learning_rate": 5.568040201005026e-05,
-      "loss": 5.4331,
-      "step": 44614
-    },
-    {
-      "epoch": 23.26727509778357,
-      "grad_norm": 1.5527443885803223,
-      "learning_rate": 5.567939698492462e-05,
-      "loss": 4.3405,
-      "step": 44615
-    },
-    {
-      "epoch": 23.26779661016949,
-      "grad_norm": 1.6191926002502441,
-      "learning_rate": 5.567839195979899e-05,
-      "loss": 5.5103,
-      "step": 44616
-    },
-    {
-      "epoch": 23.26831812255541,
-      "grad_norm": 1.5476422309875488,
-      "learning_rate": 5.567738693467337e-05,
-      "loss": 5.1155,
-      "step": 44617
-    },
-    {
-      "epoch": 23.26883963494133,
-      "grad_norm": 1.5693473815917969,
-      "learning_rate": 5.5676381909547735e-05,
-      "loss": 5.1467,
-      "step": 44618
-    },
-    {
-      "epoch": 23.26936114732725,
-      "grad_norm": 1.503140926361084,
-      "learning_rate": 5.567537688442211e-05,
-      "loss": 5.2797,
-      "step": 44619
-    },
-    {
-      "epoch": 23.26988265971317,
-      "grad_norm": 1.5679562091827393,
-      "learning_rate": 5.5674371859296484e-05,
-      "loss": 5.2887,
-      "step": 44620
-    },
-    {
-      "epoch": 23.270404172099088,
-      "grad_norm": 1.5854188203811646,
-      "learning_rate": 5.567336683417086e-05,
-      "loss": 5.3159,
-      "step": 44621
-    },
-    {
-      "epoch": 23.270925684485007,
-      "grad_norm": 1.772110104560852,
-      "learning_rate": 5.5672361809045226e-05,
-      "loss": 5.1824,
-      "step": 44622
-    },
-    {
-      "epoch": 23.271447196870927,
-      "grad_norm": 1.6210100650787354,
-      "learning_rate": 5.5671356783919604e-05,
-      "loss": 5.0699,
-      "step": 44623
-    },
-    {
-      "epoch": 23.271968709256846,
-      "grad_norm": 1.5166345834732056,
-      "learning_rate": 5.567035175879397e-05,
-      "loss": 5.3716,
-      "step": 44624
-    },
-    {
-      "epoch": 23.272490221642762,
-      "grad_norm": 1.5097739696502686,
-      "learning_rate": 5.5669346733668346e-05,
-      "loss": 5.1122,
-      "step": 44625
-    },
-    {
-      "epoch": 23.27301173402868,
-      "grad_norm": 1.5546506643295288,
-      "learning_rate": 5.566834170854272e-05,
-      "loss": 5.1947,
-      "step": 44626
-    },
-    {
-      "epoch": 23.2735332464146,
-      "grad_norm": 1.5289373397827148,
-      "learning_rate": 5.5667336683417095e-05,
-      "loss": 4.4029,
-      "step": 44627
-    },
-    {
-      "epoch": 23.27405475880052,
-      "grad_norm": 1.5476421117782593,
-      "learning_rate": 5.566633165829146e-05,
-      "loss": 5.4249,
-      "step": 44628
-    },
-    {
-      "epoch": 23.27457627118644,
-      "grad_norm": 1.5300356149673462,
-      "learning_rate": 5.5665326633165824e-05,
-      "loss": 5.2788,
-      "step": 44629
-    },
-    {
-      "epoch": 23.27509778357236,
-      "grad_norm": 1.485335350036621,
-      "learning_rate": 5.56643216080402e-05,
-      "loss": 5.0492,
-      "step": 44630
-    },
-    {
-      "epoch": 23.27561929595828,
-      "grad_norm": 1.605517029762268,
-      "learning_rate": 5.566331658291457e-05,
-      "loss": 4.7811,
-      "step": 44631
-    },
-    {
-      "epoch": 23.2761408083442,
-      "grad_norm": 1.5763812065124512,
-      "learning_rate": 5.566231155778895e-05,
-      "loss": 5.2471,
-      "step": 44632
-    },
-    {
-      "epoch": 23.276662320730118,
-      "grad_norm": 1.4646650552749634,
-      "learning_rate": 5.5661306532663315e-05,
-      "loss": 4.9819,
-      "step": 44633
-    },
-    {
-      "epoch": 23.277183833116037,
-      "grad_norm": 1.566563367843628,
-      "learning_rate": 5.566030150753769e-05,
-      "loss": 5.1293,
-      "step": 44634
-    },
-    {
-      "epoch": 23.277705345501957,
-      "grad_norm": 1.5506607294082642,
-      "learning_rate": 5.565929648241206e-05,
-      "loss": 5.1751,
-      "step": 44635
-    },
-    {
-      "epoch": 23.278226857887876,
-      "grad_norm": 1.5204654932022095,
-      "learning_rate": 5.5658291457286435e-05,
-      "loss": 5.2847,
-      "step": 44636
-    },
-    {
-      "epoch": 23.278748370273792,
-      "grad_norm": 1.6279391050338745,
-      "learning_rate": 5.5657286432160806e-05,
-      "loss": 5.4812,
-      "step": 44637
-    },
-    {
-      "epoch": 23.27926988265971,
-      "grad_norm": 1.4877694845199585,
-      "learning_rate": 5.5656281407035183e-05,
-      "loss": 5.377,
-      "step": 44638
-    },
-    {
-      "epoch": 23.27979139504563,
-      "grad_norm": 1.5620474815368652,
-      "learning_rate": 5.565527638190955e-05,
-      "loss": 5.6881,
-      "step": 44639
-    },
-    {
-      "epoch": 23.28031290743155,
-      "grad_norm": 1.641066551208496,
-      "learning_rate": 5.5654271356783926e-05,
-      "loss": 5.1048,
-      "step": 44640
-    },
-    {
-      "epoch": 23.28083441981747,
-      "grad_norm": 1.6184463500976562,
-      "learning_rate": 5.5653266331658297e-05,
-      "loss": 5.1682,
-      "step": 44641
-    },
-    {
-      "epoch": 23.28135593220339,
-      "grad_norm": 1.4685239791870117,
-      "learning_rate": 5.5652261306532674e-05,
-      "loss": 5.3044,
-      "step": 44642
-    },
-    {
-      "epoch": 23.28187744458931,
-      "grad_norm": 1.4906600713729858,
-      "learning_rate": 5.565125628140704e-05,
-      "loss": 5.6824,
-      "step": 44643
-    },
-    {
-      "epoch": 23.28239895697523,
-      "grad_norm": 1.5627871751785278,
-      "learning_rate": 5.56502512562814e-05,
-      "loss": 5.3077,
-      "step": 44644
-    },
-    {
-      "epoch": 23.282920469361148,
-      "grad_norm": 1.6152735948562622,
-      "learning_rate": 5.564924623115578e-05,
-      "loss": 4.8679,
-      "step": 44645
-    },
-    {
-      "epoch": 23.283441981747067,
-      "grad_norm": 1.6646863222122192,
-      "learning_rate": 5.564824120603015e-05,
-      "loss": 4.7536,
-      "step": 44646
-    },
-    {
-      "epoch": 23.283963494132987,
-      "grad_norm": 1.6920599937438965,
-      "learning_rate": 5.564723618090453e-05,
-      "loss": 4.7927,
-      "step": 44647
-    },
-    {
-      "epoch": 23.284485006518906,
-      "grad_norm": 1.538078784942627,
-      "learning_rate": 5.5646231155778894e-05,
-      "loss": 5.5078,
-      "step": 44648
-    },
-    {
-      "epoch": 23.285006518904822,
-      "grad_norm": 1.4506512880325317,
-      "learning_rate": 5.564522613065327e-05,
-      "loss": 5.5572,
-      "step": 44649
-    },
-    {
-      "epoch": 23.285528031290742,
-      "grad_norm": 1.5637023448944092,
-      "learning_rate": 5.5644221105527636e-05,
-      "loss": 4.9838,
-      "step": 44650
-    },
-    {
-      "epoch": 23.28604954367666,
-      "grad_norm": 1.6314842700958252,
-      "learning_rate": 5.5643216080402014e-05,
-      "loss": 5.2783,
-      "step": 44651
-    },
-    {
-      "epoch": 23.28657105606258,
-      "grad_norm": 1.5503811836242676,
-      "learning_rate": 5.5642211055276385e-05,
-      "loss": 5.4445,
-      "step": 44652
-    },
-    {
-      "epoch": 23.2870925684485,
-      "grad_norm": 1.6422772407531738,
-      "learning_rate": 5.564120603015076e-05,
-      "loss": 5.585,
-      "step": 44653
-    },
-    {
-      "epoch": 23.28761408083442,
-      "grad_norm": 1.581883430480957,
-      "learning_rate": 5.564020100502513e-05,
-      "loss": 5.385,
-      "step": 44654
-    },
-    {
-      "epoch": 23.28813559322034,
-      "grad_norm": 1.5608153343200684,
-      "learning_rate": 5.5639195979899505e-05,
-      "loss": 4.7424,
-      "step": 44655
-    },
-    {
-      "epoch": 23.28865710560626,
-      "grad_norm": 1.4820151329040527,
-      "learning_rate": 5.563819095477387e-05,
-      "loss": 5.5058,
-      "step": 44656
-    },
-    {
-      "epoch": 23.289178617992178,
-      "grad_norm": 1.5099780559539795,
-      "learning_rate": 5.563718592964824e-05,
-      "loss": 5.4671,
-      "step": 44657
-    },
-    {
-      "epoch": 23.289700130378097,
-      "grad_norm": 1.8262561559677124,
-      "learning_rate": 5.563618090452262e-05,
-      "loss": 4.6794,
-      "step": 44658
-    },
-    {
-      "epoch": 23.290221642764017,
-      "grad_norm": 1.477083444595337,
-      "learning_rate": 5.563517587939698e-05,
-      "loss": 5.5808,
-      "step": 44659
-    },
-    {
-      "epoch": 23.290743155149936,
-      "grad_norm": 1.5138734579086304,
-      "learning_rate": 5.563417085427136e-05,
-      "loss": 5.489,
-      "step": 44660
-    },
-    {
-      "epoch": 23.291264667535852,
-      "grad_norm": 1.5622485876083374,
-      "learning_rate": 5.563316582914573e-05,
-      "loss": 5.0328,
-      "step": 44661
-    },
-    {
-      "epoch": 23.291786179921772,
-      "grad_norm": 1.637935996055603,
-      "learning_rate": 5.563216080402011e-05,
-      "loss": 5.0845,
-      "step": 44662
-    },
-    {
-      "epoch": 23.29230769230769,
-      "grad_norm": 1.5235352516174316,
-      "learning_rate": 5.563115577889447e-05,
-      "loss": 5.4494,
-      "step": 44663
-    },
-    {
-      "epoch": 23.29282920469361,
-      "grad_norm": 1.6391088962554932,
-      "learning_rate": 5.563015075376885e-05,
-      "loss": 5.3723,
-      "step": 44664
-    },
-    {
-      "epoch": 23.29335071707953,
-      "grad_norm": 1.5789251327514648,
-      "learning_rate": 5.5629145728643215e-05,
-      "loss": 5.5773,
-      "step": 44665
-    },
-    {
-      "epoch": 23.29387222946545,
-      "grad_norm": 1.515793800354004,
-      "learning_rate": 5.562814070351759e-05,
-      "loss": 5.2731,
-      "step": 44666
-    },
-    {
-      "epoch": 23.29439374185137,
-      "grad_norm": 1.5366852283477783,
-      "learning_rate": 5.5627135678391964e-05,
-      "loss": 4.8907,
-      "step": 44667
-    },
-    {
-      "epoch": 23.29491525423729,
-      "grad_norm": 1.5618301630020142,
-      "learning_rate": 5.562613065326634e-05,
-      "loss": 5.3766,
-      "step": 44668
-    },
-    {
-      "epoch": 23.295436766623208,
-      "grad_norm": 1.4979588985443115,
-      "learning_rate": 5.5625125628140706e-05,
-      "loss": 5.4347,
-      "step": 44669
-    },
-    {
-      "epoch": 23.295958279009128,
-      "grad_norm": 1.5338482856750488,
-      "learning_rate": 5.562412060301507e-05,
-      "loss": 5.4531,
-      "step": 44670
-    },
-    {
-      "epoch": 23.296479791395047,
-      "grad_norm": 1.5686256885528564,
-      "learning_rate": 5.562311557788945e-05,
-      "loss": 5.1434,
-      "step": 44671
-    },
-    {
-      "epoch": 23.297001303780966,
-      "grad_norm": 1.5425318479537964,
-      "learning_rate": 5.562211055276382e-05,
-      "loss": 5.2741,
-      "step": 44672
-    },
-    {
-      "epoch": 23.297522816166882,
-      "grad_norm": 1.6043047904968262,
-      "learning_rate": 5.56211055276382e-05,
-      "loss": 5.4325,
-      "step": 44673
-    },
-    {
-      "epoch": 23.298044328552802,
-      "grad_norm": 1.3888049125671387,
-      "learning_rate": 5.562010050251256e-05,
-      "loss": 5.5591,
-      "step": 44674
-    },
-    {
-      "epoch": 23.29856584093872,
-      "grad_norm": 1.5566939115524292,
-      "learning_rate": 5.561909547738694e-05,
-      "loss": 5.1854,
-      "step": 44675
-    },
-    {
-      "epoch": 23.29908735332464,
-      "grad_norm": 1.5312343835830688,
-      "learning_rate": 5.5618090452261304e-05,
-      "loss": 5.6062,
-      "step": 44676
-    },
-    {
-      "epoch": 23.29960886571056,
-      "grad_norm": 1.7183352708816528,
-      "learning_rate": 5.561708542713568e-05,
-      "loss": 4.8673,
-      "step": 44677
-    },
-    {
-      "epoch": 23.30013037809648,
-      "grad_norm": 1.5134196281433105,
-      "learning_rate": 5.561608040201005e-05,
-      "loss": 5.7735,
-      "step": 44678
-    },
-    {
-      "epoch": 23.3006518904824,
-      "grad_norm": 1.5826233625411987,
-      "learning_rate": 5.561507537688443e-05,
-      "loss": 5.0083,
-      "step": 44679
-    },
-    {
-      "epoch": 23.30117340286832,
-      "grad_norm": 1.5253071784973145,
-      "learning_rate": 5.5614070351758795e-05,
-      "loss": 4.9844,
-      "step": 44680
-    },
-    {
-      "epoch": 23.301694915254238,
-      "grad_norm": 1.4610686302185059,
-      "learning_rate": 5.561306532663317e-05,
-      "loss": 5.1329,
-      "step": 44681
-    },
-    {
-      "epoch": 23.302216427640158,
-      "grad_norm": 1.5844380855560303,
-      "learning_rate": 5.5612060301507544e-05,
-      "loss": 5.3277,
-      "step": 44682
-    },
-    {
-      "epoch": 23.302737940026077,
-      "grad_norm": 1.5018259286880493,
-      "learning_rate": 5.561105527638191e-05,
-      "loss": 5.5396,
-      "step": 44683
-    },
-    {
-      "epoch": 23.303259452411996,
-      "grad_norm": 1.47543466091156,
-      "learning_rate": 5.5610050251256286e-05,
-      "loss": 5.3904,
-      "step": 44684
-    },
-    {
-      "epoch": 23.303780964797912,
-      "grad_norm": 1.67115318775177,
-      "learning_rate": 5.560904522613065e-05,
-      "loss": 4.7273,
-      "step": 44685
-    },
-    {
-      "epoch": 23.304302477183832,
-      "grad_norm": 1.5946059226989746,
-      "learning_rate": 5.560804020100503e-05,
-      "loss": 5.6025,
-      "step": 44686
-    },
-    {
-      "epoch": 23.30482398956975,
-      "grad_norm": 1.4989362955093384,
-      "learning_rate": 5.56070351758794e-05,
-      "loss": 5.676,
-      "step": 44687
-    },
-    {
-      "epoch": 23.30534550195567,
-      "grad_norm": 1.6304326057434082,
-      "learning_rate": 5.560603015075378e-05,
-      "loss": 5.4481,
-      "step": 44688
-    },
-    {
-      "epoch": 23.30586701434159,
-      "grad_norm": 1.501067876815796,
-      "learning_rate": 5.560502512562814e-05,
-      "loss": 5.7041,
-      "step": 44689
-    },
-    {
-      "epoch": 23.30638852672751,
-      "grad_norm": 1.65953528881073,
-      "learning_rate": 5.560402010050252e-05,
-      "loss": 5.273,
-      "step": 44690
-    },
-    {
-      "epoch": 23.30691003911343,
-      "grad_norm": 1.5365334749221802,
-      "learning_rate": 5.560301507537688e-05,
-      "loss": 5.1931,
-      "step": 44691
-    },
-    {
-      "epoch": 23.30743155149935,
-      "grad_norm": 1.5538578033447266,
-      "learning_rate": 5.560201005025126e-05,
-      "loss": 5.3223,
-      "step": 44692
-    },
-    {
-      "epoch": 23.307953063885268,
-      "grad_norm": 1.6225897073745728,
-      "learning_rate": 5.560100502512563e-05,
-      "loss": 5.1897,
-      "step": 44693
-    },
-    {
-      "epoch": 23.308474576271188,
-      "grad_norm": 1.5602236986160278,
-      "learning_rate": 5.560000000000001e-05,
-      "loss": 5.1927,
-      "step": 44694
-    },
-    {
-      "epoch": 23.308996088657107,
-      "grad_norm": 1.4370789527893066,
-      "learning_rate": 5.5598994974874374e-05,
-      "loss": 5.9174,
-      "step": 44695
-    },
-    {
-      "epoch": 23.309517601043027,
-      "grad_norm": 1.5780892372131348,
-      "learning_rate": 5.559798994974874e-05,
-      "loss": 5.2164,
-      "step": 44696
-    },
-    {
-      "epoch": 23.310039113428942,
-      "grad_norm": 1.5019656419754028,
-      "learning_rate": 5.5596984924623116e-05,
-      "loss": 5.6995,
-      "step": 44697
-    },
-    {
-      "epoch": 23.310560625814862,
-      "grad_norm": 1.533696174621582,
-      "learning_rate": 5.559597989949749e-05,
-      "loss": 5.4244,
-      "step": 44698
-    },
-    {
-      "epoch": 23.31108213820078,
-      "grad_norm": 1.6366585493087769,
-      "learning_rate": 5.5594974874371865e-05,
-      "loss": 5.4108,
-      "step": 44699
-    },
-    {
-      "epoch": 23.3116036505867,
-      "grad_norm": 1.603880524635315,
-      "learning_rate": 5.559396984924623e-05,
-      "loss": 5.401,
-      "step": 44700
-    },
-    {
-      "epoch": 23.31212516297262,
-      "grad_norm": 1.5485318899154663,
-      "learning_rate": 5.559296482412061e-05,
-      "loss": 5.1255,
-      "step": 44701
-    },
-    {
-      "epoch": 23.31264667535854,
-      "grad_norm": 1.4589794874191284,
-      "learning_rate": 5.559195979899497e-05,
-      "loss": 5.0422,
-      "step": 44702
-    },
-    {
-      "epoch": 23.31316818774446,
-      "grad_norm": 1.582326889038086,
-      "learning_rate": 5.559095477386935e-05,
-      "loss": 5.1916,
-      "step": 44703
-    },
-    {
-      "epoch": 23.31368970013038,
-      "grad_norm": 1.6801213026046753,
-      "learning_rate": 5.558994974874372e-05,
-      "loss": 4.6877,
-      "step": 44704
-    },
-    {
-      "epoch": 23.314211212516298,
-      "grad_norm": 1.6560171842575073,
-      "learning_rate": 5.55889447236181e-05,
-      "loss": 5.0668,
-      "step": 44705
-    },
-    {
-      "epoch": 23.314732724902218,
-      "grad_norm": 1.5631471872329712,
-      "learning_rate": 5.558793969849246e-05,
-      "loss": 4.9834,
-      "step": 44706
-    },
-    {
-      "epoch": 23.315254237288137,
-      "grad_norm": 1.5147922039031982,
-      "learning_rate": 5.558693467336684e-05,
-      "loss": 5.2403,
-      "step": 44707
-    },
-    {
-      "epoch": 23.315775749674053,
-      "grad_norm": 1.5578421354293823,
-      "learning_rate": 5.558592964824121e-05,
-      "loss": 5.0552,
-      "step": 44708
-    },
-    {
-      "epoch": 23.316297262059972,
-      "grad_norm": 1.614723563194275,
-      "learning_rate": 5.5584924623115576e-05,
-      "loss": 5.0896,
-      "step": 44709
-    },
-    {
-      "epoch": 23.316818774445892,
-      "grad_norm": 1.5920286178588867,
-      "learning_rate": 5.558391959798995e-05,
-      "loss": 5.5287,
-      "step": 44710
-    },
-    {
-      "epoch": 23.31734028683181,
-      "grad_norm": 1.552654504776001,
-      "learning_rate": 5.558291457286432e-05,
-      "loss": 4.8938,
-      "step": 44711
-    },
-    {
-      "epoch": 23.31786179921773,
-      "grad_norm": 1.6547778844833374,
-      "learning_rate": 5.5581909547738695e-05,
-      "loss": 5.0849,
-      "step": 44712
-    },
-    {
-      "epoch": 23.31838331160365,
-      "grad_norm": 1.6074649095535278,
-      "learning_rate": 5.5580904522613067e-05,
-      "loss": 4.9143,
-      "step": 44713
-    },
-    {
-      "epoch": 23.31890482398957,
-      "grad_norm": 1.5663902759552002,
-      "learning_rate": 5.5579899497487444e-05,
-      "loss": 5.0095,
-      "step": 44714
-    },
-    {
-      "epoch": 23.31942633637549,
-      "grad_norm": 1.5822302103042603,
-      "learning_rate": 5.557889447236181e-05,
-      "loss": 4.7694,
-      "step": 44715
-    },
-    {
-      "epoch": 23.31994784876141,
-      "grad_norm": 1.638737678527832,
-      "learning_rate": 5.5577889447236186e-05,
-      "loss": 4.7813,
-      "step": 44716
-    },
-    {
-      "epoch": 23.320469361147328,
-      "grad_norm": 1.5123647451400757,
-      "learning_rate": 5.557688442211055e-05,
-      "loss": 5.1303,
-      "step": 44717
-    },
-    {
-      "epoch": 23.320990873533248,
-      "grad_norm": 1.5520777702331543,
-      "learning_rate": 5.557587939698493e-05,
-      "loss": 5.4999,
-      "step": 44718
-    },
-    {
-      "epoch": 23.321512385919167,
-      "grad_norm": 1.6192748546600342,
-      "learning_rate": 5.55748743718593e-05,
-      "loss": 4.9425,
-      "step": 44719
-    },
-    {
-      "epoch": 23.322033898305083,
-      "grad_norm": 1.6600195169448853,
-      "learning_rate": 5.557386934673368e-05,
-      "loss": 4.7403,
-      "step": 44720
-    },
-    {
-      "epoch": 23.322555410691002,
-      "grad_norm": 1.6267646551132202,
-      "learning_rate": 5.557286432160804e-05,
-      "loss": 5.0642,
-      "step": 44721
-    },
-    {
-      "epoch": 23.323076923076922,
-      "grad_norm": 1.507372498512268,
-      "learning_rate": 5.5571859296482406e-05,
-      "loss": 5.4397,
-      "step": 44722
-    },
-    {
-      "epoch": 23.32359843546284,
-      "grad_norm": 1.5454317331314087,
-      "learning_rate": 5.5570854271356784e-05,
-      "loss": 5.1344,
-      "step": 44723
-    },
-    {
-      "epoch": 23.32411994784876,
-      "grad_norm": 1.5941543579101562,
-      "learning_rate": 5.5569849246231155e-05,
-      "loss": 5.2755,
-      "step": 44724
-    },
-    {
-      "epoch": 23.32464146023468,
-      "grad_norm": 1.4954109191894531,
-      "learning_rate": 5.556884422110553e-05,
-      "loss": 5.3311,
-      "step": 44725
-    },
-    {
-      "epoch": 23.3251629726206,
-      "grad_norm": 1.505150556564331,
-      "learning_rate": 5.55678391959799e-05,
-      "loss": 5.3077,
-      "step": 44726
-    },
-    {
-      "epoch": 23.32568448500652,
-      "grad_norm": 1.5327328443527222,
-      "learning_rate": 5.5566834170854275e-05,
-      "loss": 5.3264,
-      "step": 44727
-    },
-    {
-      "epoch": 23.32620599739244,
-      "grad_norm": 1.5768507719039917,
-      "learning_rate": 5.5565829145728646e-05,
-      "loss": 5.5773,
-      "step": 44728
-    },
-    {
-      "epoch": 23.326727509778358,
-      "grad_norm": 1.6567356586456299,
-      "learning_rate": 5.5564824120603024e-05,
-      "loss": 5.0229,
-      "step": 44729
-    },
-    {
-      "epoch": 23.327249022164278,
-      "grad_norm": 1.5159434080123901,
-      "learning_rate": 5.556381909547739e-05,
-      "loss": 5.1724,
-      "step": 44730
-    },
-    {
-      "epoch": 23.327770534550197,
-      "grad_norm": 1.5891367197036743,
-      "learning_rate": 5.5562814070351766e-05,
-      "loss": 5.1333,
-      "step": 44731
-    },
-    {
-      "epoch": 23.328292046936113,
-      "grad_norm": 1.633366584777832,
-      "learning_rate": 5.556180904522613e-05,
-      "loss": 5.0104,
-      "step": 44732
-    },
-    {
-      "epoch": 23.328813559322032,
-      "grad_norm": 1.709047794342041,
-      "learning_rate": 5.556080402010051e-05,
-      "loss": 5.0638,
-      "step": 44733
-    },
-    {
-      "epoch": 23.329335071707952,
-      "grad_norm": 1.5495436191558838,
-      "learning_rate": 5.555979899497488e-05,
-      "loss": 5.3463,
-      "step": 44734
-    },
-    {
-      "epoch": 23.32985658409387,
-      "grad_norm": 1.558257818222046,
-      "learning_rate": 5.555879396984926e-05,
-      "loss": 5.55,
-      "step": 44735
-    },
-    {
-      "epoch": 23.33037809647979,
-      "grad_norm": 1.5250074863433838,
-      "learning_rate": 5.555778894472362e-05,
-      "loss": 5.6959,
-      "step": 44736
-    },
-    {
-      "epoch": 23.33089960886571,
-      "grad_norm": 1.485108733177185,
-      "learning_rate": 5.5556783919597985e-05,
-      "loss": 5.5015,
-      "step": 44737
-    },
-    {
-      "epoch": 23.33142112125163,
-      "grad_norm": 1.6164953708648682,
-      "learning_rate": 5.555577889447236e-05,
-      "loss": 4.6788,
-      "step": 44738
-    },
-    {
-      "epoch": 23.33194263363755,
-      "grad_norm": 1.5041263103485107,
-      "learning_rate": 5.5554773869346734e-05,
-      "loss": 5.1798,
-      "step": 44739
-    },
-    {
-      "epoch": 23.33246414602347,
-      "grad_norm": 1.634890079498291,
-      "learning_rate": 5.555376884422111e-05,
-      "loss": 4.4108,
-      "step": 44740
-    },
-    {
-      "epoch": 23.332985658409388,
-      "grad_norm": 1.5192242860794067,
-      "learning_rate": 5.5552763819095476e-05,
-      "loss": 4.9664,
-      "step": 44741
-    },
-    {
-      "epoch": 23.333507170795308,
-      "grad_norm": 1.644005298614502,
-      "learning_rate": 5.5551758793969854e-05,
-      "loss": 4.7573,
-      "step": 44742
-    },
-    {
-      "epoch": 23.334028683181227,
-      "grad_norm": 1.6740736961364746,
-      "learning_rate": 5.555075376884422e-05,
-      "loss": 4.898,
-      "step": 44743
-    },
-    {
-      "epoch": 23.334550195567143,
-      "grad_norm": 1.5707627534866333,
-      "learning_rate": 5.5549748743718596e-05,
-      "loss": 5.1425,
-      "step": 44744
-    },
-    {
-      "epoch": 23.335071707953063,
-      "grad_norm": 1.692326545715332,
-      "learning_rate": 5.554874371859297e-05,
-      "loss": 5.1639,
-      "step": 44745
-    },
-    {
-      "epoch": 23.335593220338982,
-      "grad_norm": 1.54416024684906,
-      "learning_rate": 5.5547738693467345e-05,
-      "loss": 5.3127,
-      "step": 44746
-    },
-    {
-      "epoch": 23.3361147327249,
-      "grad_norm": 1.5275280475616455,
-      "learning_rate": 5.554673366834171e-05,
-      "loss": 5.4431,
-      "step": 44747
-    },
-    {
-      "epoch": 23.33663624511082,
-      "grad_norm": 1.5547983646392822,
-      "learning_rate": 5.554572864321609e-05,
-      "loss": 5.414,
-      "step": 44748
-    },
-    {
-      "epoch": 23.33715775749674,
-      "grad_norm": 1.6379680633544922,
-      "learning_rate": 5.554472361809046e-05,
-      "loss": 5.1368,
-      "step": 44749
-    },
-    {
-      "epoch": 23.33767926988266,
-      "grad_norm": 1.561330795288086,
-      "learning_rate": 5.554371859296482e-05,
-      "loss": 5.2506,
-      "step": 44750
-    },
-    {
-      "epoch": 23.33820078226858,
-      "grad_norm": 1.6030808687210083,
-      "learning_rate": 5.55427135678392e-05,
-      "loss": 5.1105,
-      "step": 44751
-    },
-    {
-      "epoch": 23.3387222946545,
-      "grad_norm": 1.5966078042984009,
-      "learning_rate": 5.5541708542713565e-05,
-      "loss": 5.507,
-      "step": 44752
-    },
-    {
-      "epoch": 23.33924380704042,
-      "grad_norm": 1.5918004512786865,
-      "learning_rate": 5.554070351758794e-05,
-      "loss": 4.7794,
-      "step": 44753
-    },
-    {
-      "epoch": 23.339765319426338,
-      "grad_norm": 1.7015801668167114,
-      "learning_rate": 5.5539698492462314e-05,
-      "loss": 4.5018,
-      "step": 44754
-    },
-    {
-      "epoch": 23.340286831812257,
-      "grad_norm": 1.5586298704147339,
-      "learning_rate": 5.553869346733669e-05,
-      "loss": 5.1593,
-      "step": 44755
-    },
-    {
-      "epoch": 23.340808344198173,
-      "grad_norm": 1.5368242263793945,
-      "learning_rate": 5.5537688442211056e-05,
-      "loss": 5.4901,
-      "step": 44756
-    },
-    {
-      "epoch": 23.341329856584093,
-      "grad_norm": 1.598795771598816,
-      "learning_rate": 5.5536683417085433e-05,
-      "loss": 5.4654,
-      "step": 44757
-    },
-    {
-      "epoch": 23.341851368970012,
-      "grad_norm": 1.5153985023498535,
-      "learning_rate": 5.55356783919598e-05,
-      "loss": 5.317,
-      "step": 44758
-    },
-    {
-      "epoch": 23.34237288135593,
-      "grad_norm": 1.5620619058609009,
-      "learning_rate": 5.5534673366834176e-05,
-      "loss": 5.0553,
-      "step": 44759
-    },
-    {
-      "epoch": 23.34289439374185,
-      "grad_norm": 1.44981849193573,
-      "learning_rate": 5.5533668341708547e-05,
-      "loss": 5.4125,
-      "step": 44760
-    },
-    {
-      "epoch": 23.34341590612777,
-      "grad_norm": 1.5409274101257324,
-      "learning_rate": 5.5532663316582924e-05,
-      "loss": 5.447,
-      "step": 44761
-    },
-    {
-      "epoch": 23.34393741851369,
-      "grad_norm": 1.5615460872650146,
-      "learning_rate": 5.553165829145729e-05,
-      "loss": 4.8259,
-      "step": 44762
-    },
-    {
-      "epoch": 23.34445893089961,
-      "grad_norm": 1.555101990699768,
-      "learning_rate": 5.553065326633165e-05,
-      "loss": 5.1489,
-      "step": 44763
-    },
-    {
-      "epoch": 23.34498044328553,
-      "grad_norm": 1.532172679901123,
-      "learning_rate": 5.552964824120603e-05,
-      "loss": 5.1653,
-      "step": 44764
-    },
-    {
-      "epoch": 23.34550195567145,
-      "grad_norm": 1.624721884727478,
-      "learning_rate": 5.55286432160804e-05,
-      "loss": 5.1131,
-      "step": 44765
-    },
-    {
-      "epoch": 23.346023468057368,
-      "grad_norm": 1.4859825372695923,
-      "learning_rate": 5.552763819095478e-05,
-      "loss": 5.6562,
-      "step": 44766
-    },
-    {
-      "epoch": 23.346544980443287,
-      "grad_norm": 1.4772056341171265,
-      "learning_rate": 5.5526633165829144e-05,
-      "loss": 5.4459,
-      "step": 44767
-    },
-    {
-      "epoch": 23.347066492829203,
-      "grad_norm": 1.6053836345672607,
-      "learning_rate": 5.552562814070352e-05,
-      "loss": 4.8979,
-      "step": 44768
-    },
-    {
-      "epoch": 23.347588005215123,
-      "grad_norm": 1.481070637702942,
-      "learning_rate": 5.552462311557789e-05,
-      "loss": 5.5769,
-      "step": 44769
-    },
-    {
-      "epoch": 23.348109517601042,
-      "grad_norm": 1.5127042531967163,
-      "learning_rate": 5.552361809045227e-05,
-      "loss": 5.5403,
-      "step": 44770
-    },
-    {
-      "epoch": 23.34863102998696,
-      "grad_norm": 1.5336209535598755,
-      "learning_rate": 5.5522613065326635e-05,
-      "loss": 5.3403,
-      "step": 44771
-    },
-    {
-      "epoch": 23.34915254237288,
-      "grad_norm": 1.5843722820281982,
-      "learning_rate": 5.552160804020101e-05,
-      "loss": 5.3198,
-      "step": 44772
-    },
-    {
-      "epoch": 23.3496740547588,
-      "grad_norm": 1.5614718198776245,
-      "learning_rate": 5.552060301507538e-05,
-      "loss": 5.2318,
-      "step": 44773
-    },
-    {
-      "epoch": 23.35019556714472,
-      "grad_norm": 1.413976788520813,
-      "learning_rate": 5.5519597989949755e-05,
-      "loss": 4.9059,
-      "step": 44774
-    },
-    {
-      "epoch": 23.35071707953064,
-      "grad_norm": 1.6387206315994263,
-      "learning_rate": 5.5518592964824126e-05,
-      "loss": 5.1424,
-      "step": 44775
-    },
-    {
-      "epoch": 23.35123859191656,
-      "grad_norm": 1.5690563917160034,
-      "learning_rate": 5.551758793969849e-05,
-      "loss": 5.3351,
-      "step": 44776
-    },
-    {
-      "epoch": 23.35176010430248,
-      "grad_norm": 1.5309547185897827,
-      "learning_rate": 5.551658291457287e-05,
-      "loss": 5.595,
-      "step": 44777
-    },
-    {
-      "epoch": 23.352281616688398,
-      "grad_norm": 1.5046021938323975,
-      "learning_rate": 5.551557788944723e-05,
-      "loss": 5.2392,
-      "step": 44778
-    },
-    {
-      "epoch": 23.352803129074317,
-      "grad_norm": 1.491847276687622,
-      "learning_rate": 5.551457286432161e-05,
-      "loss": 5.5747,
-      "step": 44779
-    },
-    {
-      "epoch": 23.353324641460233,
-      "grad_norm": 1.5293476581573486,
-      "learning_rate": 5.551356783919598e-05,
-      "loss": 5.036,
-      "step": 44780
-    },
-    {
-      "epoch": 23.353846153846153,
-      "grad_norm": 1.5138404369354248,
-      "learning_rate": 5.551256281407036e-05,
-      "loss": 4.9775,
-      "step": 44781
-    },
-    {
-      "epoch": 23.354367666232072,
-      "grad_norm": 1.4660429954528809,
-      "learning_rate": 5.551155778894472e-05,
-      "loss": 5.4673,
-      "step": 44782
-    },
-    {
-      "epoch": 23.35488917861799,
-      "grad_norm": 1.5658761262893677,
-      "learning_rate": 5.55105527638191e-05,
-      "loss": 5.4376,
-      "step": 44783
-    },
-    {
-      "epoch": 23.35541069100391,
-      "grad_norm": 1.5566550493240356,
-      "learning_rate": 5.5509547738693465e-05,
-      "loss": 4.3535,
-      "step": 44784
-    },
-    {
-      "epoch": 23.35593220338983,
-      "grad_norm": 1.5932620763778687,
-      "learning_rate": 5.550854271356784e-05,
-      "loss": 5.4128,
-      "step": 44785
-    },
-    {
-      "epoch": 23.35645371577575,
-      "grad_norm": 1.4946129322052002,
-      "learning_rate": 5.5507537688442214e-05,
-      "loss": 5.4542,
-      "step": 44786
-    },
-    {
-      "epoch": 23.35697522816167,
-      "grad_norm": 1.4808534383773804,
-      "learning_rate": 5.550653266331659e-05,
-      "loss": 5.4266,
-      "step": 44787
-    },
-    {
-      "epoch": 23.35749674054759,
-      "grad_norm": 1.5076165199279785,
-      "learning_rate": 5.5505527638190956e-05,
-      "loss": 5.3677,
-      "step": 44788
-    },
-    {
-      "epoch": 23.35801825293351,
-      "grad_norm": 1.6001607179641724,
-      "learning_rate": 5.550452261306532e-05,
-      "loss": 5.1336,
-      "step": 44789
-    },
-    {
-      "epoch": 23.358539765319428,
-      "grad_norm": 1.5280483961105347,
-      "learning_rate": 5.55035175879397e-05,
-      "loss": 5.3155,
-      "step": 44790
-    },
-    {
-      "epoch": 23.359061277705347,
-      "grad_norm": 1.580587387084961,
-      "learning_rate": 5.550251256281407e-05,
-      "loss": 5.2879,
-      "step": 44791
-    },
-    {
-      "epoch": 23.359582790091263,
-      "grad_norm": 1.5784662961959839,
-      "learning_rate": 5.550150753768845e-05,
-      "loss": 5.6049,
-      "step": 44792
-    },
-    {
-      "epoch": 23.360104302477183,
-      "grad_norm": 1.668789029121399,
-      "learning_rate": 5.550050251256281e-05,
-      "loss": 4.3553,
-      "step": 44793
-    },
-    {
-      "epoch": 23.360625814863102,
-      "grad_norm": 1.6764118671417236,
-      "learning_rate": 5.549949748743719e-05,
-      "loss": 4.9711,
-      "step": 44794
-    },
-    {
-      "epoch": 23.36114732724902,
-      "grad_norm": 1.6770001649856567,
-      "learning_rate": 5.549849246231156e-05,
-      "loss": 5.4995,
-      "step": 44795
-    },
-    {
-      "epoch": 23.36166883963494,
-      "grad_norm": 1.498279333114624,
-      "learning_rate": 5.549748743718594e-05,
-      "loss": 5.1829,
-      "step": 44796
-    },
-    {
-      "epoch": 23.36219035202086,
-      "grad_norm": 1.5488457679748535,
-      "learning_rate": 5.54964824120603e-05,
-      "loss": 4.6627,
-      "step": 44797
-    },
-    {
-      "epoch": 23.36271186440678,
-      "grad_norm": 1.5784153938293457,
-      "learning_rate": 5.549547738693468e-05,
-      "loss": 5.2863,
-      "step": 44798
-    },
-    {
-      "epoch": 23.3632333767927,
-      "grad_norm": 1.44713294506073,
-      "learning_rate": 5.5494472361809045e-05,
-      "loss": 5.6199,
-      "step": 44799
-    },
-    {
-      "epoch": 23.36375488917862,
-      "grad_norm": 1.566400170326233,
-      "learning_rate": 5.549346733668342e-05,
-      "loss": 4.9808,
-      "step": 44800
-    },
-    {
-      "epoch": 23.36427640156454,
-      "grad_norm": 1.4777164459228516,
-      "learning_rate": 5.5492462311557794e-05,
-      "loss": 5.6646,
-      "step": 44801
-    },
-    {
-      "epoch": 23.364797913950458,
-      "grad_norm": 1.4942063093185425,
-      "learning_rate": 5.549145728643216e-05,
-      "loss": 5.1504,
-      "step": 44802
-    },
-    {
-      "epoch": 23.365319426336377,
-      "grad_norm": 1.614269495010376,
-      "learning_rate": 5.5490452261306536e-05,
-      "loss": 5.4345,
-      "step": 44803
-    },
-    {
-      "epoch": 23.365840938722293,
-      "grad_norm": 1.5554661750793457,
-      "learning_rate": 5.54894472361809e-05,
-      "loss": 4.8948,
-      "step": 44804
-    },
-    {
-      "epoch": 23.366362451108213,
-      "grad_norm": 1.6437816619873047,
-      "learning_rate": 5.548844221105528e-05,
-      "loss": 5.1617,
-      "step": 44805
-    },
-    {
-      "epoch": 23.366883963494132,
-      "grad_norm": 1.6021653413772583,
-      "learning_rate": 5.548743718592965e-05,
-      "loss": 4.676,
-      "step": 44806
-    },
-    {
-      "epoch": 23.36740547588005,
-      "grad_norm": 1.5827009677886963,
-      "learning_rate": 5.548643216080403e-05,
-      "loss": 4.9689,
-      "step": 44807
-    },
-    {
-      "epoch": 23.36792698826597,
-      "grad_norm": 1.528337001800537,
-      "learning_rate": 5.548542713567839e-05,
-      "loss": 5.04,
-      "step": 44808
-    },
-    {
-      "epoch": 23.36844850065189,
-      "grad_norm": 1.5188783407211304,
-      "learning_rate": 5.548442211055277e-05,
-      "loss": 5.1398,
-      "step": 44809
-    },
-    {
-      "epoch": 23.36897001303781,
-      "grad_norm": 1.4692344665527344,
-      "learning_rate": 5.548341708542713e-05,
-      "loss": 5.4519,
-      "step": 44810
-    },
-    {
-      "epoch": 23.36949152542373,
-      "grad_norm": 1.5242841243743896,
-      "learning_rate": 5.548241206030151e-05,
-      "loss": 5.2651,
-      "step": 44811
-    },
-    {
-      "epoch": 23.37001303780965,
-      "grad_norm": 1.4630486965179443,
-      "learning_rate": 5.548140703517588e-05,
-      "loss": 4.091,
-      "step": 44812
-    },
-    {
-      "epoch": 23.37053455019557,
-      "grad_norm": 1.5034470558166504,
-      "learning_rate": 5.548040201005026e-05,
-      "loss": 4.1589,
-      "step": 44813
-    },
-    {
-      "epoch": 23.371056062581488,
-      "grad_norm": 1.5538959503173828,
-      "learning_rate": 5.5479396984924624e-05,
-      "loss": 5.5367,
-      "step": 44814
-    },
-    {
-      "epoch": 23.371577574967404,
-      "grad_norm": 1.5861825942993164,
-      "learning_rate": 5.5478391959799e-05,
-      "loss": 5.4655,
-      "step": 44815
-    },
-    {
-      "epoch": 23.372099087353323,
-      "grad_norm": 1.6194576025009155,
-      "learning_rate": 5.547738693467337e-05,
-      "loss": 5.1032,
-      "step": 44816
-    },
-    {
-      "epoch": 23.372620599739243,
-      "grad_norm": 1.559667944908142,
-      "learning_rate": 5.547638190954774e-05,
-      "loss": 5.166,
-      "step": 44817
-    },
-    {
-      "epoch": 23.373142112125162,
-      "grad_norm": 1.4487583637237549,
-      "learning_rate": 5.5475376884422115e-05,
-      "loss": 5.6305,
-      "step": 44818
-    },
-    {
-      "epoch": 23.37366362451108,
-      "grad_norm": 1.5856473445892334,
-      "learning_rate": 5.547437185929648e-05,
-      "loss": 5.189,
-      "step": 44819
-    },
-    {
-      "epoch": 23.374185136897,
-      "grad_norm": 1.4817148447036743,
-      "learning_rate": 5.547336683417086e-05,
-      "loss": 5.6641,
-      "step": 44820
-    },
-    {
-      "epoch": 23.37470664928292,
-      "grad_norm": 1.5517505407333374,
-      "learning_rate": 5.547236180904523e-05,
-      "loss": 5.3899,
-      "step": 44821
-    },
-    {
-      "epoch": 23.37522816166884,
-      "grad_norm": 1.56109619140625,
-      "learning_rate": 5.5471356783919606e-05,
-      "loss": 5.2616,
-      "step": 44822
-    },
-    {
-      "epoch": 23.37574967405476,
-      "grad_norm": 1.5524286031723022,
-      "learning_rate": 5.547035175879397e-05,
-      "loss": 5.4948,
-      "step": 44823
-    },
-    {
-      "epoch": 23.37627118644068,
-      "grad_norm": 1.6954971551895142,
-      "learning_rate": 5.546934673366835e-05,
-      "loss": 5.1275,
-      "step": 44824
-    },
-    {
-      "epoch": 23.3767926988266,
-      "grad_norm": 1.6407297849655151,
-      "learning_rate": 5.546834170854271e-05,
-      "loss": 4.9815,
-      "step": 44825
-    },
-    {
-      "epoch": 23.377314211212518,
-      "grad_norm": 1.6697989702224731,
-      "learning_rate": 5.546733668341709e-05,
-      "loss": 5.6146,
-      "step": 44826
-    },
-    {
-      "epoch": 23.377835723598434,
-      "grad_norm": 1.619157075881958,
-      "learning_rate": 5.546633165829146e-05,
-      "loss": 4.8785,
-      "step": 44827
-    },
-    {
-      "epoch": 23.378357235984353,
-      "grad_norm": 1.7644245624542236,
-      "learning_rate": 5.546532663316584e-05,
-      "loss": 4.9313,
-      "step": 44828
-    },
-    {
-      "epoch": 23.378878748370273,
-      "grad_norm": 1.6359869241714478,
-      "learning_rate": 5.54643216080402e-05,
-      "loss": 5.0221,
-      "step": 44829
-    },
-    {
-      "epoch": 23.379400260756192,
-      "grad_norm": 1.5231208801269531,
-      "learning_rate": 5.546331658291457e-05,
-      "loss": 5.6047,
-      "step": 44830
-    },
-    {
-      "epoch": 23.37992177314211,
-      "grad_norm": 1.4292277097702026,
-      "learning_rate": 5.5462311557788945e-05,
-      "loss": 5.4968,
-      "step": 44831
-    },
-    {
-      "epoch": 23.38044328552803,
-      "grad_norm": 1.549662470817566,
-      "learning_rate": 5.5461306532663317e-05,
-      "loss": 5.24,
-      "step": 44832
-    },
-    {
-      "epoch": 23.38096479791395,
-      "grad_norm": 1.6458088159561157,
-      "learning_rate": 5.5460301507537694e-05,
-      "loss": 5.3427,
-      "step": 44833
-    },
-    {
-      "epoch": 23.38148631029987,
-      "grad_norm": 1.5357705354690552,
-      "learning_rate": 5.545929648241206e-05,
-      "loss": 5.49,
-      "step": 44834
-    },
-    {
-      "epoch": 23.38200782268579,
-      "grad_norm": 1.533947229385376,
-      "learning_rate": 5.5458291457286436e-05,
-      "loss": 5.5063,
-      "step": 44835
-    },
-    {
-      "epoch": 23.38252933507171,
-      "grad_norm": 1.6017277240753174,
-      "learning_rate": 5.545728643216081e-05,
-      "loss": 5.2308,
-      "step": 44836
-    },
-    {
-      "epoch": 23.38305084745763,
-      "grad_norm": 1.5580233335494995,
-      "learning_rate": 5.5456281407035185e-05,
-      "loss": 5.2376,
-      "step": 44837
-    },
-    {
-      "epoch": 23.383572359843548,
-      "grad_norm": 1.5221823453903198,
-      "learning_rate": 5.545527638190955e-05,
-      "loss": 4.8828,
-      "step": 44838
-    },
-    {
-      "epoch": 23.384093872229464,
-      "grad_norm": 1.576883316040039,
-      "learning_rate": 5.545427135678393e-05,
-      "loss": 4.9445,
-      "step": 44839
-    },
-    {
-      "epoch": 23.384615384615383,
-      "grad_norm": 1.6334574222564697,
-      "learning_rate": 5.545326633165829e-05,
-      "loss": 5.1537,
-      "step": 44840
-    },
-    {
-      "epoch": 23.385136897001303,
-      "grad_norm": 1.5723347663879395,
-      "learning_rate": 5.545226130653267e-05,
-      "loss": 5.4723,
-      "step": 44841
-    },
-    {
-      "epoch": 23.385658409387222,
-      "grad_norm": 1.570426106452942,
-      "learning_rate": 5.545125628140704e-05,
-      "loss": 5.3406,
-      "step": 44842
-    },
-    {
-      "epoch": 23.38617992177314,
-      "grad_norm": 1.5493618249893188,
-      "learning_rate": 5.5450251256281405e-05,
-      "loss": 5.657,
-      "step": 44843
-    },
-    {
-      "epoch": 23.38670143415906,
-      "grad_norm": 1.5081101655960083,
-      "learning_rate": 5.544924623115578e-05,
-      "loss": 5.3488,
-      "step": 44844
-    },
-    {
-      "epoch": 23.38722294654498,
-      "grad_norm": 1.4700024127960205,
-      "learning_rate": 5.544824120603015e-05,
-      "loss": 5.6245,
-      "step": 44845
-    },
-    {
-      "epoch": 23.3877444589309,
-      "grad_norm": 1.5843256711959839,
-      "learning_rate": 5.5447236180904525e-05,
-      "loss": 5.1503,
-      "step": 44846
-    },
-    {
-      "epoch": 23.38826597131682,
-      "grad_norm": 1.538255214691162,
-      "learning_rate": 5.5446231155778896e-05,
-      "loss": 5.1317,
-      "step": 44847
-    },
-    {
-      "epoch": 23.38878748370274,
-      "grad_norm": 1.5703628063201904,
-      "learning_rate": 5.5445226130653274e-05,
-      "loss": 5.4458,
-      "step": 44848
-    },
-    {
-      "epoch": 23.38930899608866,
-      "grad_norm": 1.4976208209991455,
-      "learning_rate": 5.544422110552764e-05,
-      "loss": 5.5471,
-      "step": 44849
-    },
-    {
-      "epoch": 23.389830508474578,
-      "grad_norm": 1.5688999891281128,
-      "learning_rate": 5.5443216080402016e-05,
-      "loss": 5.437,
-      "step": 44850
-    },
-    {
-      "epoch": 23.390352020860494,
-      "grad_norm": 1.506462812423706,
-      "learning_rate": 5.544221105527638e-05,
-      "loss": 4.8639,
-      "step": 44851
-    },
-    {
-      "epoch": 23.390873533246413,
-      "grad_norm": 1.6201246976852417,
-      "learning_rate": 5.544120603015076e-05,
-      "loss": 4.8506,
-      "step": 44852
-    },
-    {
-      "epoch": 23.391395045632333,
-      "grad_norm": 1.6703269481658936,
-      "learning_rate": 5.544020100502513e-05,
-      "loss": 5.0466,
-      "step": 44853
-    },
-    {
-      "epoch": 23.391916558018252,
-      "grad_norm": 1.523182988166809,
-      "learning_rate": 5.543919597989951e-05,
-      "loss": 5.3471,
-      "step": 44854
-    },
-    {
-      "epoch": 23.39243807040417,
-      "grad_norm": 1.5678662061691284,
-      "learning_rate": 5.543819095477387e-05,
-      "loss": 5.1039,
-      "step": 44855
-    },
-    {
-      "epoch": 23.39295958279009,
-      "grad_norm": 1.5313336849212646,
-      "learning_rate": 5.543718592964824e-05,
-      "loss": 5.2902,
-      "step": 44856
-    },
-    {
-      "epoch": 23.39348109517601,
-      "grad_norm": 1.5380008220672607,
-      "learning_rate": 5.543618090452262e-05,
-      "loss": 5.0938,
-      "step": 44857
-    },
-    {
-      "epoch": 23.39400260756193,
-      "grad_norm": 1.4714242219924927,
-      "learning_rate": 5.5435175879396984e-05,
-      "loss": 4.9877,
-      "step": 44858
-    },
-    {
-      "epoch": 23.39452411994785,
-      "grad_norm": 1.5945611000061035,
-      "learning_rate": 5.543417085427136e-05,
-      "loss": 5.496,
-      "step": 44859
-    },
-    {
-      "epoch": 23.39504563233377,
-      "grad_norm": 1.4515812397003174,
-      "learning_rate": 5.5433165829145726e-05,
-      "loss": 5.3086,
-      "step": 44860
-    },
-    {
-      "epoch": 23.39556714471969,
-      "grad_norm": 1.5526423454284668,
-      "learning_rate": 5.5432160804020104e-05,
-      "loss": 5.4301,
-      "step": 44861
-    },
-    {
-      "epoch": 23.396088657105608,
-      "grad_norm": 1.5716462135314941,
-      "learning_rate": 5.5431155778894475e-05,
-      "loss": 4.1672,
-      "step": 44862
-    },
-    {
-      "epoch": 23.396610169491524,
-      "grad_norm": 1.457054853439331,
-      "learning_rate": 5.543015075376885e-05,
-      "loss": 5.6789,
-      "step": 44863
-    },
-    {
-      "epoch": 23.397131681877443,
-      "grad_norm": 1.7080209255218506,
-      "learning_rate": 5.542914572864322e-05,
-      "loss": 4.794,
-      "step": 44864
-    },
-    {
-      "epoch": 23.397653194263363,
-      "grad_norm": 1.5621246099472046,
-      "learning_rate": 5.5428140703517595e-05,
-      "loss": 5.1312,
-      "step": 44865
-    },
-    {
-      "epoch": 23.398174706649282,
-      "grad_norm": 1.541440725326538,
-      "learning_rate": 5.542713567839196e-05,
-      "loss": 5.048,
-      "step": 44866
-    },
-    {
-      "epoch": 23.3986962190352,
-      "grad_norm": 1.558325171470642,
-      "learning_rate": 5.542613065326634e-05,
-      "loss": 5.2787,
-      "step": 44867
-    },
-    {
-      "epoch": 23.39921773142112,
-      "grad_norm": 1.4823945760726929,
-      "learning_rate": 5.542512562814071e-05,
-      "loss": 4.3879,
-      "step": 44868
-    },
-    {
-      "epoch": 23.39973924380704,
-      "grad_norm": 1.4838858842849731,
-      "learning_rate": 5.542412060301507e-05,
-      "loss": 5.6677,
-      "step": 44869
-    },
-    {
-      "epoch": 23.40026075619296,
-      "grad_norm": 1.4814224243164062,
-      "learning_rate": 5.542311557788945e-05,
-      "loss": 5.4327,
-      "step": 44870
-    },
-    {
-      "epoch": 23.40078226857888,
-      "grad_norm": 1.5469328165054321,
-      "learning_rate": 5.5422110552763815e-05,
-      "loss": 5.2989,
-      "step": 44871
-    },
-    {
-      "epoch": 23.4013037809648,
-      "grad_norm": 1.5777119398117065,
-      "learning_rate": 5.542110552763819e-05,
-      "loss": 5.4819,
-      "step": 44872
-    },
-    {
-      "epoch": 23.40182529335072,
-      "grad_norm": 1.6021257638931274,
-      "learning_rate": 5.5420100502512564e-05,
-      "loss": 5.397,
-      "step": 44873
-    },
-    {
-      "epoch": 23.402346805736638,
-      "grad_norm": 1.5289041996002197,
-      "learning_rate": 5.541909547738694e-05,
-      "loss": 5.1859,
-      "step": 44874
-    },
-    {
-      "epoch": 23.402868318122554,
-      "grad_norm": 1.5088396072387695,
-      "learning_rate": 5.5418090452261306e-05,
-      "loss": 5.2348,
-      "step": 44875
-    },
-    {
-      "epoch": 23.403389830508473,
-      "grad_norm": 1.5135935544967651,
-      "learning_rate": 5.5417085427135683e-05,
-      "loss": 5.3577,
-      "step": 44876
-    },
-    {
-      "epoch": 23.403911342894393,
-      "grad_norm": 1.5188326835632324,
-      "learning_rate": 5.541608040201005e-05,
-      "loss": 5.3038,
-      "step": 44877
-    },
-    {
-      "epoch": 23.404432855280312,
-      "grad_norm": 1.6530377864837646,
-      "learning_rate": 5.5415075376884426e-05,
-      "loss": 4.2765,
-      "step": 44878
-    },
-    {
-      "epoch": 23.40495436766623,
-      "grad_norm": 1.6188114881515503,
-      "learning_rate": 5.54140703517588e-05,
-      "loss": 5.3767,
-      "step": 44879
-    },
-    {
-      "epoch": 23.40547588005215,
-      "grad_norm": 1.609874963760376,
-      "learning_rate": 5.5413065326633174e-05,
-      "loss": 5.063,
-      "step": 44880
-    },
-    {
-      "epoch": 23.40599739243807,
-      "grad_norm": 1.7933354377746582,
-      "learning_rate": 5.541206030150754e-05,
-      "loss": 4.5534,
-      "step": 44881
-    },
-    {
-      "epoch": 23.40651890482399,
-      "grad_norm": 1.5888760089874268,
-      "learning_rate": 5.541105527638191e-05,
-      "loss": 5.1161,
-      "step": 44882
-    },
-    {
-      "epoch": 23.40704041720991,
-      "grad_norm": 1.5926038026809692,
-      "learning_rate": 5.541005025125629e-05,
-      "loss": 5.0075,
-      "step": 44883
-    },
-    {
-      "epoch": 23.40756192959583,
-      "grad_norm": 1.3859573602676392,
-      "learning_rate": 5.540904522613065e-05,
-      "loss": 5.6684,
-      "step": 44884
-    },
-    {
-      "epoch": 23.40808344198175,
-      "grad_norm": 1.602033257484436,
-      "learning_rate": 5.540804020100503e-05,
-      "loss": 5.3442,
-      "step": 44885
-    },
-    {
-      "epoch": 23.408604954367668,
-      "grad_norm": 1.546629548072815,
-      "learning_rate": 5.5407035175879394e-05,
-      "loss": 5.4811,
-      "step": 44886
-    },
-    {
-      "epoch": 23.409126466753584,
-      "grad_norm": 1.4871773719787598,
-      "learning_rate": 5.540603015075377e-05,
-      "loss": 5.7262,
-      "step": 44887
-    },
-    {
-      "epoch": 23.409647979139503,
-      "grad_norm": 1.4014136791229248,
-      "learning_rate": 5.540502512562814e-05,
-      "loss": 5.4074,
-      "step": 44888
-    },
-    {
-      "epoch": 23.410169491525423,
-      "grad_norm": 1.6141948699951172,
-      "learning_rate": 5.540402010050252e-05,
-      "loss": 5.3655,
-      "step": 44889
-    },
-    {
-      "epoch": 23.410691003911342,
-      "grad_norm": 1.480941653251648,
-      "learning_rate": 5.5403015075376885e-05,
-      "loss": 5.3766,
-      "step": 44890
-    },
-    {
-      "epoch": 23.41121251629726,
-      "grad_norm": 1.6376103162765503,
-      "learning_rate": 5.540201005025126e-05,
-      "loss": 5.0988,
-      "step": 44891
-    },
-    {
-      "epoch": 23.41173402868318,
-      "grad_norm": 1.6026805639266968,
-      "learning_rate": 5.540100502512563e-05,
-      "loss": 5.0585,
-      "step": 44892
-    },
-    {
-      "epoch": 23.4122555410691,
-      "grad_norm": 1.6583120822906494,
-      "learning_rate": 5.5400000000000005e-05,
-      "loss": 5.2839,
-      "step": 44893
-    },
-    {
-      "epoch": 23.41277705345502,
-      "grad_norm": 1.5726120471954346,
-      "learning_rate": 5.5398994974874376e-05,
-      "loss": 4.922,
-      "step": 44894
-    },
-    {
-      "epoch": 23.41329856584094,
-      "grad_norm": 1.5792877674102783,
-      "learning_rate": 5.539798994974874e-05,
-      "loss": 4.9949,
-      "step": 44895
-    },
-    {
-      "epoch": 23.41382007822686,
-      "grad_norm": 1.4660223722457886,
-      "learning_rate": 5.539698492462312e-05,
-      "loss": 5.633,
-      "step": 44896
-    },
-    {
-      "epoch": 23.41434159061278,
-      "grad_norm": 1.4152697324752808,
-      "learning_rate": 5.539597989949748e-05,
-      "loss": 5.31,
-      "step": 44897
-    },
-    {
-      "epoch": 23.414863102998694,
-      "grad_norm": 1.4570657014846802,
-      "learning_rate": 5.539497487437186e-05,
-      "loss": 5.4195,
-      "step": 44898
-    },
-    {
-      "epoch": 23.415384615384614,
-      "grad_norm": 1.5786123275756836,
-      "learning_rate": 5.539396984924623e-05,
-      "loss": 5.2348,
-      "step": 44899
-    },
-    {
-      "epoch": 23.415906127770533,
-      "grad_norm": 1.586459994316101,
-      "learning_rate": 5.539296482412061e-05,
-      "loss": 5.3063,
-      "step": 44900
-    },
-    {
-      "epoch": 23.416427640156453,
-      "grad_norm": 1.5395164489746094,
-      "learning_rate": 5.539195979899497e-05,
-      "loss": 5.0393,
-      "step": 44901
-    },
-    {
-      "epoch": 23.416949152542372,
-      "grad_norm": 1.4962353706359863,
-      "learning_rate": 5.539095477386935e-05,
-      "loss": 5.3445,
-      "step": 44902
-    },
-    {
-      "epoch": 23.41747066492829,
-      "grad_norm": 1.5453811883926392,
-      "learning_rate": 5.538994974874372e-05,
-      "loss": 5.334,
-      "step": 44903
-    },
-    {
-      "epoch": 23.41799217731421,
-      "grad_norm": 1.6115304231643677,
-      "learning_rate": 5.53889447236181e-05,
-      "loss": 5.1814,
-      "step": 44904
-    },
-    {
-      "epoch": 23.41851368970013,
-      "grad_norm": 1.5551193952560425,
-      "learning_rate": 5.5387939698492464e-05,
-      "loss": 5.1218,
-      "step": 44905
-    },
-    {
-      "epoch": 23.41903520208605,
-      "grad_norm": 1.5897161960601807,
-      "learning_rate": 5.538693467336684e-05,
-      "loss": 4.8642,
-      "step": 44906
-    },
-    {
-      "epoch": 23.41955671447197,
-      "grad_norm": 1.6551594734191895,
-      "learning_rate": 5.5385929648241206e-05,
-      "loss": 4.8146,
-      "step": 44907
-    },
-    {
-      "epoch": 23.42007822685789,
-      "grad_norm": 2.199145793914795,
-      "learning_rate": 5.5384924623115584e-05,
-      "loss": 4.8774,
-      "step": 44908
-    },
-    {
-      "epoch": 23.42059973924381,
-      "grad_norm": 1.5411579608917236,
-      "learning_rate": 5.5383919597989955e-05,
-      "loss": 5.3269,
-      "step": 44909
-    },
-    {
-      "epoch": 23.421121251629724,
-      "grad_norm": 1.4980368614196777,
-      "learning_rate": 5.538291457286432e-05,
-      "loss": 5.4747,
-      "step": 44910
-    },
-    {
-      "epoch": 23.421642764015644,
-      "grad_norm": 1.6050505638122559,
-      "learning_rate": 5.53819095477387e-05,
-      "loss": 5.1229,
-      "step": 44911
-    },
-    {
-      "epoch": 23.422164276401563,
-      "grad_norm": 1.4569580554962158,
-      "learning_rate": 5.538090452261306e-05,
-      "loss": 5.8128,
-      "step": 44912
-    },
-    {
-      "epoch": 23.422685788787483,
-      "grad_norm": 1.4727782011032104,
-      "learning_rate": 5.537989949748744e-05,
-      "loss": 5.0906,
-      "step": 44913
-    },
-    {
-      "epoch": 23.423207301173402,
-      "grad_norm": 1.7206674814224243,
-      "learning_rate": 5.537889447236181e-05,
-      "loss": 4.9645,
-      "step": 44914
-    },
-    {
-      "epoch": 23.423728813559322,
-      "grad_norm": 1.4313329458236694,
-      "learning_rate": 5.537788944723619e-05,
-      "loss": 5.3439,
-      "step": 44915
-    },
-    {
-      "epoch": 23.42425032594524,
-      "grad_norm": 1.613979458808899,
-      "learning_rate": 5.537688442211055e-05,
-      "loss": 5.2015,
-      "step": 44916
-    },
-    {
-      "epoch": 23.42477183833116,
-      "grad_norm": 1.6038111448287964,
-      "learning_rate": 5.537587939698493e-05,
-      "loss": 5.3326,
-      "step": 44917
-    },
-    {
-      "epoch": 23.42529335071708,
-      "grad_norm": 1.5882784128189087,
-      "learning_rate": 5.5374874371859295e-05,
-      "loss": 5.2954,
-      "step": 44918
-    },
-    {
-      "epoch": 23.425814863103,
-      "grad_norm": 1.6646360158920288,
-      "learning_rate": 5.537386934673367e-05,
-      "loss": 5.2949,
-      "step": 44919
-    },
-    {
-      "epoch": 23.42633637548892,
-      "grad_norm": 1.6462442874908447,
-      "learning_rate": 5.5372864321608044e-05,
-      "loss": 5.1533,
-      "step": 44920
-    },
-    {
-      "epoch": 23.42685788787484,
-      "grad_norm": 1.4788990020751953,
-      "learning_rate": 5.537185929648242e-05,
-      "loss": 4.9927,
-      "step": 44921
-    },
-    {
-      "epoch": 23.427379400260754,
-      "grad_norm": 1.533339500427246,
-      "learning_rate": 5.5370854271356786e-05,
-      "loss": 5.2571,
-      "step": 44922
-    },
-    {
-      "epoch": 23.427900912646674,
-      "grad_norm": 1.6141910552978516,
-      "learning_rate": 5.536984924623116e-05,
-      "loss": 5.3332,
-      "step": 44923
-    },
-    {
-      "epoch": 23.428422425032593,
-      "grad_norm": 1.5784156322479248,
-      "learning_rate": 5.5368844221105535e-05,
-      "loss": 5.0632,
-      "step": 44924
-    },
-    {
-      "epoch": 23.428943937418513,
-      "grad_norm": 1.5659290552139282,
-      "learning_rate": 5.53678391959799e-05,
-      "loss": 5.513,
-      "step": 44925
-    },
-    {
-      "epoch": 23.429465449804432,
-      "grad_norm": 1.7315040826797485,
-      "learning_rate": 5.536683417085428e-05,
-      "loss": 4.5922,
-      "step": 44926
-    },
-    {
-      "epoch": 23.429986962190352,
-      "grad_norm": 1.5244708061218262,
-      "learning_rate": 5.536582914572864e-05,
-      "loss": 5.3048,
-      "step": 44927
-    },
-    {
-      "epoch": 23.43050847457627,
-      "grad_norm": 1.5128982067108154,
-      "learning_rate": 5.536482412060302e-05,
-      "loss": 5.6634,
-      "step": 44928
-    },
-    {
-      "epoch": 23.43102998696219,
-      "grad_norm": 1.599971055984497,
-      "learning_rate": 5.536381909547739e-05,
-      "loss": 5.2492,
-      "step": 44929
-    },
-    {
-      "epoch": 23.43155149934811,
-      "grad_norm": 1.4794501066207886,
-      "learning_rate": 5.536281407035177e-05,
-      "loss": 5.0591,
-      "step": 44930
-    },
-    {
-      "epoch": 23.43207301173403,
-      "grad_norm": 1.4518438577651978,
-      "learning_rate": 5.536180904522613e-05,
-      "loss": 5.6221,
-      "step": 44931
-    },
-    {
-      "epoch": 23.43259452411995,
-      "grad_norm": 1.4881998300552368,
-      "learning_rate": 5.536080402010051e-05,
-      "loss": 5.3812,
-      "step": 44932
-    },
-    {
-      "epoch": 23.43311603650587,
-      "grad_norm": 1.5835877656936646,
-      "learning_rate": 5.5359798994974874e-05,
-      "loss": 5.0213,
-      "step": 44933
-    },
-    {
-      "epoch": 23.433637548891785,
-      "grad_norm": 1.540120005607605,
-      "learning_rate": 5.535879396984925e-05,
-      "loss": 5.3484,
-      "step": 44934
-    },
-    {
-      "epoch": 23.434159061277704,
-      "grad_norm": 1.5184276103973389,
-      "learning_rate": 5.535778894472362e-05,
-      "loss": 5.7223,
-      "step": 44935
-    },
-    {
-      "epoch": 23.434680573663623,
-      "grad_norm": 1.4848181009292603,
-      "learning_rate": 5.535678391959799e-05,
-      "loss": 5.2341,
-      "step": 44936
-    },
-    {
-      "epoch": 23.435202086049543,
-      "grad_norm": 1.6670907735824585,
-      "learning_rate": 5.5355778894472365e-05,
-      "loss": 4.8004,
-      "step": 44937
-    },
-    {
-      "epoch": 23.435723598435462,
-      "grad_norm": 1.458491325378418,
-      "learning_rate": 5.535477386934673e-05,
-      "loss": 5.253,
-      "step": 44938
-    },
-    {
-      "epoch": 23.436245110821382,
-      "grad_norm": 1.4633561372756958,
-      "learning_rate": 5.535376884422111e-05,
-      "loss": 4.8008,
-      "step": 44939
-    },
-    {
-      "epoch": 23.4367666232073,
-      "grad_norm": 1.5619767904281616,
-      "learning_rate": 5.535276381909548e-05,
-      "loss": 5.4754,
-      "step": 44940
-    },
-    {
-      "epoch": 23.43728813559322,
-      "grad_norm": 1.618003487586975,
-      "learning_rate": 5.5351758793969856e-05,
-      "loss": 5.3342,
-      "step": 44941
-    },
-    {
-      "epoch": 23.43780964797914,
-      "grad_norm": 1.5485447645187378,
-      "learning_rate": 5.535075376884422e-05,
-      "loss": 5.4219,
-      "step": 44942
-    },
-    {
-      "epoch": 23.43833116036506,
-      "grad_norm": 1.600870132446289,
-      "learning_rate": 5.53497487437186e-05,
-      "loss": 5.4811,
-      "step": 44943
-    },
-    {
-      "epoch": 23.43885267275098,
-      "grad_norm": 1.4615960121154785,
-      "learning_rate": 5.534874371859297e-05,
-      "loss": 5.1631,
-      "step": 44944
-    },
-    {
-      "epoch": 23.4393741851369,
-      "grad_norm": 1.6294506788253784,
-      "learning_rate": 5.534773869346735e-05,
-      "loss": 5.3379,
-      "step": 44945
-    },
-    {
-      "epoch": 23.439895697522815,
-      "grad_norm": 1.5913339853286743,
-      "learning_rate": 5.534673366834171e-05,
-      "loss": 5.2392,
-      "step": 44946
-    },
-    {
-      "epoch": 23.440417209908734,
-      "grad_norm": 1.5544298887252808,
-      "learning_rate": 5.534572864321609e-05,
-      "loss": 5.2093,
-      "step": 44947
-    },
-    {
-      "epoch": 23.440938722294653,
-      "grad_norm": 1.5079365968704224,
-      "learning_rate": 5.5344723618090453e-05,
-      "loss": 5.2516,
-      "step": 44948
-    },
-    {
-      "epoch": 23.441460234680573,
-      "grad_norm": 1.5498720407485962,
-      "learning_rate": 5.5343718592964824e-05,
-      "loss": 5.1318,
-      "step": 44949
-    },
-    {
-      "epoch": 23.441981747066492,
-      "grad_norm": 1.5529274940490723,
-      "learning_rate": 5.53427135678392e-05,
-      "loss": 5.0902,
-      "step": 44950
-    },
-    {
-      "epoch": 23.442503259452412,
-      "grad_norm": 1.494611144065857,
-      "learning_rate": 5.5341708542713567e-05,
-      "loss": 5.0036,
-      "step": 44951
-    },
-    {
-      "epoch": 23.44302477183833,
-      "grad_norm": 1.6769306659698486,
-      "learning_rate": 5.5340703517587944e-05,
-      "loss": 4.7166,
-      "step": 44952
-    },
-    {
-      "epoch": 23.44354628422425,
-      "grad_norm": 1.5726009607315063,
-      "learning_rate": 5.533969849246231e-05,
-      "loss": 5.4001,
-      "step": 44953
-    },
-    {
-      "epoch": 23.44406779661017,
-      "grad_norm": 1.5431960821151733,
-      "learning_rate": 5.5338693467336686e-05,
-      "loss": 4.9114,
-      "step": 44954
-    },
-    {
-      "epoch": 23.44458930899609,
-      "grad_norm": 1.4167211055755615,
-      "learning_rate": 5.533768844221106e-05,
-      "loss": 5.3506,
-      "step": 44955
-    },
-    {
-      "epoch": 23.44511082138201,
-      "grad_norm": 1.4785369634628296,
-      "learning_rate": 5.5336683417085435e-05,
-      "loss": 5.5445,
-      "step": 44956
-    },
-    {
-      "epoch": 23.44563233376793,
-      "grad_norm": 1.662868618965149,
-      "learning_rate": 5.53356783919598e-05,
-      "loss": 5.1304,
-      "step": 44957
-    },
-    {
-      "epoch": 23.446153846153845,
-      "grad_norm": 1.7496687173843384,
-      "learning_rate": 5.533467336683418e-05,
-      "loss": 4.8252,
-      "step": 44958
-    },
-    {
-      "epoch": 23.446675358539764,
-      "grad_norm": 1.65021550655365,
-      "learning_rate": 5.533366834170854e-05,
-      "loss": 5.2008,
-      "step": 44959
-    },
-    {
-      "epoch": 23.447196870925683,
-      "grad_norm": 1.4982399940490723,
-      "learning_rate": 5.533266331658292e-05,
-      "loss": 5.496,
-      "step": 44960
-    },
-    {
-      "epoch": 23.447718383311603,
-      "grad_norm": 1.6222468614578247,
-      "learning_rate": 5.533165829145729e-05,
-      "loss": 4.6989,
-      "step": 44961
-    },
-    {
-      "epoch": 23.448239895697522,
-      "grad_norm": 1.6078394651412964,
-      "learning_rate": 5.5330653266331655e-05,
-      "loss": 5.3305,
-      "step": 44962
-    },
-    {
-      "epoch": 23.448761408083442,
-      "grad_norm": 1.5726734399795532,
-      "learning_rate": 5.532964824120603e-05,
-      "loss": 5.0791,
-      "step": 44963
-    },
-    {
-      "epoch": 23.44928292046936,
-      "grad_norm": 1.512732982635498,
-      "learning_rate": 5.53286432160804e-05,
-      "loss": 5.3907,
-      "step": 44964
-    },
-    {
-      "epoch": 23.44980443285528,
-      "grad_norm": 1.6080949306488037,
-      "learning_rate": 5.5327638190954775e-05,
-      "loss": 5.2398,
-      "step": 44965
-    },
-    {
-      "epoch": 23.4503259452412,
-      "grad_norm": 1.5002926588058472,
-      "learning_rate": 5.5326633165829146e-05,
-      "loss": 5.185,
-      "step": 44966
-    },
-    {
-      "epoch": 23.45084745762712,
-      "grad_norm": 1.5301555395126343,
-      "learning_rate": 5.5325628140703524e-05,
-      "loss": 5.0533,
-      "step": 44967
-    },
-    {
-      "epoch": 23.45136897001304,
-      "grad_norm": 1.6135509014129639,
-      "learning_rate": 5.532462311557789e-05,
-      "loss": 5.3513,
-      "step": 44968
-    },
-    {
-      "epoch": 23.45189048239896,
-      "grad_norm": 1.5998921394348145,
-      "learning_rate": 5.5323618090452266e-05,
-      "loss": 5.1526,
-      "step": 44969
-    },
-    {
-      "epoch": 23.452411994784875,
-      "grad_norm": 1.5436983108520508,
-      "learning_rate": 5.532261306532664e-05,
-      "loss": 5.2006,
-      "step": 44970
-    },
-    {
-      "epoch": 23.452933507170794,
-      "grad_norm": 1.5689977407455444,
-      "learning_rate": 5.5321608040201015e-05,
-      "loss": 5.2237,
-      "step": 44971
-    },
-    {
-      "epoch": 23.453455019556714,
-      "grad_norm": 1.4503297805786133,
-      "learning_rate": 5.532060301507538e-05,
-      "loss": 5.4209,
-      "step": 44972
-    },
-    {
-      "epoch": 23.453976531942633,
-      "grad_norm": 1.5644619464874268,
-      "learning_rate": 5.531959798994976e-05,
-      "loss": 5.4503,
-      "step": 44973
-    },
-    {
-      "epoch": 23.454498044328552,
-      "grad_norm": 1.5319979190826416,
-      "learning_rate": 5.531859296482412e-05,
-      "loss": 5.168,
-      "step": 44974
-    },
-    {
-      "epoch": 23.455019556714472,
-      "grad_norm": 1.4716757535934448,
-      "learning_rate": 5.531758793969849e-05,
-      "loss": 5.3626,
-      "step": 44975
-    },
-    {
-      "epoch": 23.45554106910039,
-      "grad_norm": 1.5932561159133911,
-      "learning_rate": 5.531658291457287e-05,
-      "loss": 5.0097,
-      "step": 44976
-    },
-    {
-      "epoch": 23.45606258148631,
-      "grad_norm": 1.7147753238677979,
-      "learning_rate": 5.5315577889447234e-05,
-      "loss": 4.615,
-      "step": 44977
-    },
-    {
-      "epoch": 23.45658409387223,
-      "grad_norm": 1.7185699939727783,
-      "learning_rate": 5.531457286432161e-05,
-      "loss": 5.266,
-      "step": 44978
-    },
-    {
-      "epoch": 23.45710560625815,
-      "grad_norm": 1.6241304874420166,
-      "learning_rate": 5.5313567839195976e-05,
-      "loss": 5.5261,
-      "step": 44979
-    },
-    {
-      "epoch": 23.45762711864407,
-      "grad_norm": 1.5530451536178589,
-      "learning_rate": 5.5312562814070354e-05,
-      "loss": 5.4027,
-      "step": 44980
-    },
-    {
-      "epoch": 23.45814863102999,
-      "grad_norm": 1.5205875635147095,
-      "learning_rate": 5.5311557788944725e-05,
-      "loss": 5.5387,
-      "step": 44981
-    },
-    {
-      "epoch": 23.458670143415905,
-      "grad_norm": 1.4860084056854248,
-      "learning_rate": 5.53105527638191e-05,
-      "loss": 5.5759,
-      "step": 44982
-    },
-    {
-      "epoch": 23.459191655801824,
-      "grad_norm": 1.6300199031829834,
-      "learning_rate": 5.530954773869347e-05,
-      "loss": 5.1469,
-      "step": 44983
-    },
-    {
-      "epoch": 23.459713168187744,
-      "grad_norm": 1.537933349609375,
-      "learning_rate": 5.5308542713567845e-05,
-      "loss": 5.3099,
-      "step": 44984
-    },
-    {
-      "epoch": 23.460234680573663,
-      "grad_norm": 1.581310749053955,
-      "learning_rate": 5.530753768844221e-05,
-      "loss": 5.4324,
-      "step": 44985
-    },
-    {
-      "epoch": 23.460756192959582,
-      "grad_norm": 1.4879015684127808,
-      "learning_rate": 5.530653266331659e-05,
-      "loss": 5.1618,
-      "step": 44986
-    },
-    {
-      "epoch": 23.461277705345502,
-      "grad_norm": 1.5952553749084473,
-      "learning_rate": 5.530552763819096e-05,
-      "loss": 4.6894,
-      "step": 44987
-    },
-    {
-      "epoch": 23.46179921773142,
-      "grad_norm": 1.5920072793960571,
-      "learning_rate": 5.530452261306532e-05,
-      "loss": 5.4025,
-      "step": 44988
-    },
-    {
-      "epoch": 23.46232073011734,
-      "grad_norm": 1.7405831813812256,
-      "learning_rate": 5.53035175879397e-05,
-      "loss": 4.6019,
-      "step": 44989
-    },
-    {
-      "epoch": 23.46284224250326,
-      "grad_norm": 1.5043755769729614,
-      "learning_rate": 5.530251256281407e-05,
-      "loss": 5.6957,
-      "step": 44990
-    },
-    {
-      "epoch": 23.46336375488918,
-      "grad_norm": 1.6279869079589844,
-      "learning_rate": 5.530150753768845e-05,
-      "loss": 4.9453,
-      "step": 44991
-    },
-    {
-      "epoch": 23.4638852672751,
-      "grad_norm": 1.621447205543518,
-      "learning_rate": 5.5300502512562814e-05,
-      "loss": 5.2583,
-      "step": 44992
-    },
-    {
-      "epoch": 23.46440677966102,
-      "grad_norm": 1.4669862985610962,
-      "learning_rate": 5.529949748743719e-05,
-      "loss": 5.5507,
-      "step": 44993
-    },
-    {
-      "epoch": 23.464928292046935,
-      "grad_norm": 1.5908455848693848,
-      "learning_rate": 5.5298492462311556e-05,
-      "loss": 5.604,
-      "step": 44994
-    },
-    {
-      "epoch": 23.465449804432854,
-      "grad_norm": 1.5916928052902222,
-      "learning_rate": 5.5297487437185933e-05,
-      "loss": 5.2786,
-      "step": 44995
-    },
-    {
-      "epoch": 23.465971316818774,
-      "grad_norm": 1.626563310623169,
-      "learning_rate": 5.5296482412060305e-05,
-      "loss": 5.1868,
-      "step": 44996
-    },
-    {
-      "epoch": 23.466492829204693,
-      "grad_norm": 1.466921329498291,
-      "learning_rate": 5.529547738693468e-05,
-      "loss": 5.3598,
-      "step": 44997
-    },
-    {
-      "epoch": 23.467014341590613,
-      "grad_norm": 1.5370981693267822,
-      "learning_rate": 5.529447236180905e-05,
-      "loss": 5.4535,
-      "step": 44998
-    },
-    {
-      "epoch": 23.467535853976532,
-      "grad_norm": 1.565901279449463,
-      "learning_rate": 5.5293467336683424e-05,
-      "loss": 5.3905,
-      "step": 44999
-    },
-    {
-      "epoch": 23.46805736636245,
-      "grad_norm": 1.5909582376480103,
-      "learning_rate": 5.529246231155779e-05,
-      "loss": 5.1209,
-      "step": 45000
-    },
-    {
-      "epoch": 23.46805736636245,
-      "eval_loss": 5.347715377807617,
-      "eval_runtime": 42.6716,
-      "eval_samples_per_second": 28.731,
-      "eval_steps_per_second": 3.609,
-      "step": 45000
-    },
-    {
-      "epoch": 23.46857887874837,
-      "grad_norm": 1.4612352848052979,
-      "learning_rate": 5.5291457286432167e-05,
-      "loss": 5.1561,
-      "step": 45001
-    },
-    {
-      "epoch": 23.46910039113429,
-      "grad_norm": 1.4771522283554077,
-      "learning_rate": 5.529045226130654e-05,
-      "loss": 5.3964,
-      "step": 45002
-    },
-    {
-      "epoch": 23.46962190352021,
-      "grad_norm": 1.610545039176941,
-      "learning_rate": 5.52894472361809e-05,
-      "loss": 5.1105,
-      "step": 45003
-    },
-    {
-      "epoch": 23.47014341590613,
-      "grad_norm": 1.508809208869934,
-      "learning_rate": 5.528844221105528e-05,
-      "loss": 5.336,
-      "step": 45004
-    },
-    {
-      "epoch": 23.470664928292045,
-      "grad_norm": 1.5830525159835815,
-      "learning_rate": 5.5287437185929644e-05,
-      "loss": 5.4074,
-      "step": 45005
-    },
-    {
-      "epoch": 23.471186440677965,
-      "grad_norm": 1.5725648403167725,
-      "learning_rate": 5.528643216080402e-05,
-      "loss": 5.3925,
-      "step": 45006
-    },
-    {
-      "epoch": 23.471707953063884,
-      "grad_norm": 1.500412940979004,
-      "learning_rate": 5.528542713567839e-05,
-      "loss": 5.5655,
-      "step": 45007
-    },
-    {
-      "epoch": 23.472229465449804,
-      "grad_norm": 1.565407156944275,
-      "learning_rate": 5.528442211055277e-05,
-      "loss": 5.4518,
-      "step": 45008
-    },
-    {
-      "epoch": 23.472750977835723,
-      "grad_norm": 1.591524600982666,
-      "learning_rate": 5.5283417085427135e-05,
-      "loss": 5.4515,
-      "step": 45009
-    },
-    {
-      "epoch": 23.473272490221643,
-      "grad_norm": 1.6261792182922363,
-      "learning_rate": 5.528241206030151e-05,
-      "loss": 5.3386,
-      "step": 45010
-    },
-    {
-      "epoch": 23.473794002607562,
-      "grad_norm": 1.4288294315338135,
-      "learning_rate": 5.5281407035175884e-05,
-      "loss": 5.7177,
-      "step": 45011
-    },
-    {
-      "epoch": 23.47431551499348,
-      "grad_norm": 1.6563650369644165,
-      "learning_rate": 5.528040201005026e-05,
-      "loss": 5.2406,
-      "step": 45012
-    },
-    {
-      "epoch": 23.4748370273794,
-      "grad_norm": 1.48397696018219,
-      "learning_rate": 5.5279396984924626e-05,
-      "loss": 5.659,
-      "step": 45013
-    },
-    {
-      "epoch": 23.47535853976532,
-      "grad_norm": 1.5655341148376465,
-      "learning_rate": 5.5278391959799004e-05,
-      "loss": 5.0201,
-      "step": 45014
-    },
-    {
-      "epoch": 23.47588005215124,
-      "grad_norm": 1.4763957262039185,
-      "learning_rate": 5.527738693467337e-05,
-      "loss": 5.6983,
-      "step": 45015
-    },
-    {
-      "epoch": 23.47640156453716,
-      "grad_norm": 1.5843526124954224,
-      "learning_rate": 5.527638190954774e-05,
-      "loss": 4.9945,
-      "step": 45016
-    },
-    {
-      "epoch": 23.476923076923075,
-      "grad_norm": 1.4974619150161743,
-      "learning_rate": 5.527537688442212e-05,
-      "loss": 5.3289,
-      "step": 45017
-    },
-    {
-      "epoch": 23.477444589308995,
-      "grad_norm": 1.598821997642517,
-      "learning_rate": 5.527437185929648e-05,
-      "loss": 5.2562,
-      "step": 45018
-    },
-    {
-      "epoch": 23.477966101694914,
-      "grad_norm": 1.6421034336090088,
-      "learning_rate": 5.527336683417086e-05,
-      "loss": 5.0628,
-      "step": 45019
-    },
-    {
-      "epoch": 23.478487614080834,
-      "grad_norm": 1.3977463245391846,
-      "learning_rate": 5.527236180904522e-05,
-      "loss": 4.9353,
-      "step": 45020
-    },
-    {
-      "epoch": 23.479009126466753,
-      "grad_norm": 1.5751514434814453,
-      "learning_rate": 5.52713567839196e-05,
-      "loss": 4.7121,
-      "step": 45021
-    },
-    {
-      "epoch": 23.479530638852673,
-      "grad_norm": 1.5127168893814087,
-      "learning_rate": 5.527035175879397e-05,
-      "loss": 5.3457,
-      "step": 45022
-    },
-    {
-      "epoch": 23.480052151238592,
-      "grad_norm": 1.47398841381073,
-      "learning_rate": 5.526934673366835e-05,
-      "loss": 5.1318,
-      "step": 45023
-    },
-    {
-      "epoch": 23.48057366362451,
-      "grad_norm": 1.4608360528945923,
-      "learning_rate": 5.5268341708542714e-05,
-      "loss": 5.4417,
-      "step": 45024
-    },
-    {
-      "epoch": 23.48109517601043,
-      "grad_norm": 1.483828067779541,
-      "learning_rate": 5.526733668341709e-05,
-      "loss": 5.4418,
-      "step": 45025
-    },
-    {
-      "epoch": 23.48161668839635,
-      "grad_norm": 1.7154289484024048,
-      "learning_rate": 5.5266331658291456e-05,
-      "loss": 5.4141,
-      "step": 45026
-    },
-    {
-      "epoch": 23.48213820078227,
-      "grad_norm": 1.5350192785263062,
-      "learning_rate": 5.5265326633165834e-05,
-      "loss": 4.5566,
-      "step": 45027
-    },
-    {
-      "epoch": 23.48265971316819,
-      "grad_norm": 1.5952281951904297,
-      "learning_rate": 5.5264321608040205e-05,
-      "loss": 5.2085,
-      "step": 45028
-    },
-    {
-      "epoch": 23.483181225554105,
-      "grad_norm": 1.4713040590286255,
-      "learning_rate": 5.526331658291457e-05,
-      "loss": 5.3471,
-      "step": 45029
-    },
-    {
-      "epoch": 23.483702737940025,
-      "grad_norm": 1.6902929544448853,
-      "learning_rate": 5.526231155778895e-05,
-      "loss": 5.2995,
-      "step": 45030
-    },
-    {
-      "epoch": 23.484224250325944,
-      "grad_norm": 1.4889923334121704,
-      "learning_rate": 5.526130653266331e-05,
-      "loss": 5.0452,
-      "step": 45031
-    },
-    {
-      "epoch": 23.484745762711864,
-      "grad_norm": 1.5931788682937622,
-      "learning_rate": 5.526030150753769e-05,
-      "loss": 5.1785,
-      "step": 45032
-    },
-    {
-      "epoch": 23.485267275097783,
-      "grad_norm": 1.524707555770874,
-      "learning_rate": 5.525929648241206e-05,
-      "loss": 5.0066,
-      "step": 45033
-    },
-    {
-      "epoch": 23.485788787483703,
-      "grad_norm": 1.4234238862991333,
-      "learning_rate": 5.525829145728644e-05,
-      "loss": 5.0855,
-      "step": 45034
-    },
-    {
-      "epoch": 23.486310299869622,
-      "grad_norm": 1.5294870138168335,
-      "learning_rate": 5.52572864321608e-05,
-      "loss": 5.1467,
-      "step": 45035
-    },
-    {
-      "epoch": 23.48683181225554,
-      "grad_norm": 1.5130001306533813,
-      "learning_rate": 5.525628140703518e-05,
-      "loss": 4.8101,
-      "step": 45036
-    },
-    {
-      "epoch": 23.48735332464146,
-      "grad_norm": 1.5002493858337402,
-      "learning_rate": 5.525527638190955e-05,
-      "loss": 5.3782,
-      "step": 45037
-    },
-    {
-      "epoch": 23.48787483702738,
-      "grad_norm": 1.7037707567214966,
-      "learning_rate": 5.525427135678393e-05,
-      "loss": 4.7603,
-      "step": 45038
-    },
-    {
-      "epoch": 23.4883963494133,
-      "grad_norm": 1.521496057510376,
-      "learning_rate": 5.5253266331658294e-05,
-      "loss": 5.4269,
-      "step": 45039
-    },
-    {
-      "epoch": 23.48891786179922,
-      "grad_norm": 1.6229301691055298,
-      "learning_rate": 5.525226130653267e-05,
-      "loss": 4.833,
-      "step": 45040
-    },
-    {
-      "epoch": 23.489439374185135,
-      "grad_norm": 1.658880591392517,
-      "learning_rate": 5.5251256281407036e-05,
-      "loss": 5.1882,
-      "step": 45041
-    },
-    {
-      "epoch": 23.489960886571055,
-      "grad_norm": 1.529630184173584,
-      "learning_rate": 5.525025125628141e-05,
-      "loss": 5.1141,
-      "step": 45042
-    },
-    {
-      "epoch": 23.490482398956974,
-      "grad_norm": 1.714160442352295,
-      "learning_rate": 5.5249246231155785e-05,
-      "loss": 4.7222,
-      "step": 45043
-    },
-    {
-      "epoch": 23.491003911342894,
-      "grad_norm": 1.4787331819534302,
-      "learning_rate": 5.524824120603015e-05,
-      "loss": 4.9673,
-      "step": 45044
-    },
-    {
-      "epoch": 23.491525423728813,
-      "grad_norm": 1.556222915649414,
-      "learning_rate": 5.524723618090453e-05,
-      "loss": 5.4708,
-      "step": 45045
-    },
-    {
-      "epoch": 23.492046936114733,
-      "grad_norm": 1.5868175029754639,
-      "learning_rate": 5.524623115577889e-05,
-      "loss": 5.1471,
-      "step": 45046
-    },
-    {
-      "epoch": 23.492568448500652,
-      "grad_norm": 1.5689142942428589,
-      "learning_rate": 5.524522613065327e-05,
-      "loss": 5.2815,
-      "step": 45047
-    },
-    {
-      "epoch": 23.49308996088657,
-      "grad_norm": 1.530940294265747,
-      "learning_rate": 5.524422110552764e-05,
-      "loss": 5.8015,
-      "step": 45048
-    },
-    {
-      "epoch": 23.49361147327249,
-      "grad_norm": 1.564393162727356,
-      "learning_rate": 5.524321608040202e-05,
-      "loss": 5.6875,
-      "step": 45049
-    },
-    {
-      "epoch": 23.49413298565841,
-      "grad_norm": 1.6032979488372803,
-      "learning_rate": 5.524221105527638e-05,
-      "loss": 4.9147,
-      "step": 45050
-    },
-    {
-      "epoch": 23.49465449804433,
-      "grad_norm": 1.3761777877807617,
-      "learning_rate": 5.524120603015076e-05,
-      "loss": 4.7668,
-      "step": 45051
-    },
-    {
-      "epoch": 23.49517601043025,
-      "grad_norm": 1.7062867879867554,
-      "learning_rate": 5.5240201005025124e-05,
-      "loss": 5.1074,
-      "step": 45052
-    },
-    {
-      "epoch": 23.495697522816165,
-      "grad_norm": 1.5078603029251099,
-      "learning_rate": 5.52391959798995e-05,
-      "loss": 5.0677,
-      "step": 45053
-    },
-    {
-      "epoch": 23.496219035202085,
-      "grad_norm": 1.5129741430282593,
-      "learning_rate": 5.523819095477387e-05,
-      "loss": 5.3398,
-      "step": 45054
-    },
-    {
-      "epoch": 23.496740547588004,
-      "grad_norm": 1.5182323455810547,
-      "learning_rate": 5.523718592964824e-05,
-      "loss": 5.4624,
-      "step": 45055
-    },
-    {
-      "epoch": 23.497262059973924,
-      "grad_norm": 1.4875986576080322,
-      "learning_rate": 5.5236180904522615e-05,
-      "loss": 5.1147,
-      "step": 45056
-    },
-    {
-      "epoch": 23.497783572359843,
-      "grad_norm": 1.7591707706451416,
-      "learning_rate": 5.5235175879396986e-05,
-      "loss": 4.9003,
-      "step": 45057
-    },
-    {
-      "epoch": 23.498305084745763,
-      "grad_norm": 1.5873141288757324,
-      "learning_rate": 5.5234170854271364e-05,
-      "loss": 4.5909,
-      "step": 45058
-    },
-    {
-      "epoch": 23.498826597131682,
-      "grad_norm": 1.6183271408081055,
-      "learning_rate": 5.523316582914573e-05,
-      "loss": 5.0348,
-      "step": 45059
-    },
-    {
-      "epoch": 23.4993481095176,
-      "grad_norm": 1.5475101470947266,
-      "learning_rate": 5.5232160804020106e-05,
-      "loss": 5.3695,
-      "step": 45060
-    },
-    {
-      "epoch": 23.49986962190352,
-      "grad_norm": 1.5247535705566406,
-      "learning_rate": 5.523115577889447e-05,
-      "loss": 5.2003,
-      "step": 45061
-    },
-    {
-      "epoch": 23.50039113428944,
-      "grad_norm": 1.6140753030776978,
-      "learning_rate": 5.523015075376885e-05,
-      "loss": 5.2165,
-      "step": 45062
-    },
-    {
-      "epoch": 23.50091264667536,
-      "grad_norm": 1.5212832689285278,
-      "learning_rate": 5.522914572864322e-05,
-      "loss": 5.3545,
-      "step": 45063
-    },
-    {
-      "epoch": 23.50143415906128,
-      "grad_norm": 1.5814288854599,
-      "learning_rate": 5.52281407035176e-05,
-      "loss": 5.6908,
-      "step": 45064
-    },
-    {
-      "epoch": 23.501955671447195,
-      "grad_norm": 1.5545016527175903,
-      "learning_rate": 5.522713567839196e-05,
-      "loss": 5.0501,
-      "step": 45065
-    },
-    {
-      "epoch": 23.502477183833115,
-      "grad_norm": 1.5836246013641357,
-      "learning_rate": 5.522613065326634e-05,
-      "loss": 5.3706,
-      "step": 45066
-    },
-    {
-      "epoch": 23.502998696219034,
-      "grad_norm": 1.5750166177749634,
-      "learning_rate": 5.5225125628140703e-05,
-      "loss": 5.3314,
-      "step": 45067
-    },
-    {
-      "epoch": 23.503520208604954,
-      "grad_norm": 1.5846303701400757,
-      "learning_rate": 5.5224120603015074e-05,
-      "loss": 4.9436,
-      "step": 45068
-    },
-    {
-      "epoch": 23.504041720990873,
-      "grad_norm": 1.5784651041030884,
-      "learning_rate": 5.522311557788945e-05,
-      "loss": 5.355,
-      "step": 45069
-    },
-    {
-      "epoch": 23.504563233376793,
-      "grad_norm": 1.619275689125061,
-      "learning_rate": 5.5222110552763817e-05,
-      "loss": 5.189,
-      "step": 45070
-    },
-    {
-      "epoch": 23.505084745762712,
-      "grad_norm": 1.5252935886383057,
-      "learning_rate": 5.5221105527638194e-05,
-      "loss": 5.2501,
-      "step": 45071
-    },
-    {
-      "epoch": 23.50560625814863,
-      "grad_norm": 1.4414507150650024,
-      "learning_rate": 5.522010050251256e-05,
-      "loss": 5.4781,
-      "step": 45072
-    },
-    {
-      "epoch": 23.50612777053455,
-      "grad_norm": 1.5977143049240112,
-      "learning_rate": 5.5219095477386937e-05,
-      "loss": 5.4054,
-      "step": 45073
-    },
-    {
-      "epoch": 23.50664928292047,
-      "grad_norm": 1.5444453954696655,
-      "learning_rate": 5.521809045226131e-05,
-      "loss": 5.1981,
-      "step": 45074
-    },
-    {
-      "epoch": 23.50717079530639,
-      "grad_norm": 1.5270720720291138,
-      "learning_rate": 5.5217085427135685e-05,
-      "loss": 5.0579,
-      "step": 45075
-    },
-    {
-      "epoch": 23.50769230769231,
-      "grad_norm": 1.548493504524231,
-      "learning_rate": 5.521608040201005e-05,
-      "loss": 5.0191,
-      "step": 45076
-    },
-    {
-      "epoch": 23.508213820078225,
-      "grad_norm": 1.5735608339309692,
-      "learning_rate": 5.521507537688443e-05,
-      "loss": 4.6551,
-      "step": 45077
-    },
-    {
-      "epoch": 23.508735332464145,
-      "grad_norm": 1.5660817623138428,
-      "learning_rate": 5.52140703517588e-05,
-      "loss": 4.7702,
-      "step": 45078
-    },
-    {
-      "epoch": 23.509256844850064,
-      "grad_norm": 1.5818474292755127,
-      "learning_rate": 5.5213065326633176e-05,
-      "loss": 5.3364,
-      "step": 45079
-    },
-    {
-      "epoch": 23.509778357235984,
-      "grad_norm": 1.510961651802063,
-      "learning_rate": 5.521206030150754e-05,
-      "loss": 5.3769,
-      "step": 45080
-    },
-    {
-      "epoch": 23.510299869621903,
-      "grad_norm": 1.5668984651565552,
-      "learning_rate": 5.5211055276381905e-05,
-      "loss": 5.5767,
-      "step": 45081
-    },
-    {
-      "epoch": 23.510821382007823,
-      "grad_norm": 1.5317301750183105,
-      "learning_rate": 5.521005025125628e-05,
-      "loss": 5.0237,
-      "step": 45082
-    },
-    {
-      "epoch": 23.511342894393742,
-      "grad_norm": 1.5393564701080322,
-      "learning_rate": 5.5209045226130654e-05,
-      "loss": 5.335,
-      "step": 45083
-    },
-    {
-      "epoch": 23.51186440677966,
-      "grad_norm": 1.5804204940795898,
-      "learning_rate": 5.520804020100503e-05,
-      "loss": 5.4783,
-      "step": 45084
-    },
-    {
-      "epoch": 23.51238591916558,
-      "grad_norm": 1.6279453039169312,
-      "learning_rate": 5.5207035175879396e-05,
-      "loss": 5.363,
-      "step": 45085
-    },
-    {
-      "epoch": 23.5129074315515,
-      "grad_norm": 1.6443127393722534,
-      "learning_rate": 5.5206030150753774e-05,
-      "loss": 5.3165,
-      "step": 45086
-    },
-    {
-      "epoch": 23.51342894393742,
-      "grad_norm": 1.684666633605957,
-      "learning_rate": 5.520502512562814e-05,
-      "loss": 5.1392,
-      "step": 45087
-    },
-    {
-      "epoch": 23.513950456323336,
-      "grad_norm": 1.5392242670059204,
-      "learning_rate": 5.5204020100502516e-05,
-      "loss": 5.2445,
-      "step": 45088
-    },
-    {
-      "epoch": 23.514471968709255,
-      "grad_norm": 1.5514860153198242,
-      "learning_rate": 5.520301507537689e-05,
-      "loss": 5.5353,
-      "step": 45089
-    },
-    {
-      "epoch": 23.514993481095175,
-      "grad_norm": 1.5972399711608887,
-      "learning_rate": 5.5202010050251265e-05,
-      "loss": 5.3071,
-      "step": 45090
-    },
-    {
-      "epoch": 23.515514993481094,
-      "grad_norm": 1.5658745765686035,
-      "learning_rate": 5.520100502512563e-05,
-      "loss": 5.1803,
-      "step": 45091
-    },
-    {
-      "epoch": 23.516036505867014,
-      "grad_norm": 1.5933376550674438,
-      "learning_rate": 5.520000000000001e-05,
-      "loss": 5.4945,
-      "step": 45092
-    },
-    {
-      "epoch": 23.516558018252933,
-      "grad_norm": 1.4929301738739014,
-      "learning_rate": 5.519899497487437e-05,
-      "loss": 5.3845,
-      "step": 45093
-    },
-    {
-      "epoch": 23.517079530638853,
-      "grad_norm": 1.5012513399124146,
-      "learning_rate": 5.519798994974875e-05,
-      "loss": 5.2765,
-      "step": 45094
-    },
-    {
-      "epoch": 23.517601043024772,
-      "grad_norm": 1.498672604560852,
-      "learning_rate": 5.519698492462312e-05,
-      "loss": 4.9528,
-      "step": 45095
-    },
-    {
-      "epoch": 23.51812255541069,
-      "grad_norm": 1.7148864269256592,
-      "learning_rate": 5.5195979899497484e-05,
-      "loss": 5.6454,
-      "step": 45096
-    },
-    {
-      "epoch": 23.51864406779661,
-      "grad_norm": 1.685937762260437,
-      "learning_rate": 5.519497487437186e-05,
-      "loss": 4.9297,
-      "step": 45097
-    },
-    {
-      "epoch": 23.51916558018253,
-      "grad_norm": 1.542743444442749,
-      "learning_rate": 5.519396984924623e-05,
-      "loss": 5.0165,
-      "step": 45098
-    },
-    {
-      "epoch": 23.51968709256845,
-      "grad_norm": 1.6755121946334839,
-      "learning_rate": 5.519296482412061e-05,
-      "loss": 5.0977,
-      "step": 45099
-    },
-    {
-      "epoch": 23.52020860495437,
-      "grad_norm": 1.4921413660049438,
-      "learning_rate": 5.5191959798994975e-05,
-      "loss": 5.3896,
-      "step": 45100
-    },
-    {
-      "epoch": 23.520730117340285,
-      "grad_norm": 1.6777423620224,
-      "learning_rate": 5.519095477386935e-05,
-      "loss": 5.4064,
-      "step": 45101
-    },
-    {
-      "epoch": 23.521251629726205,
-      "grad_norm": 1.5083500146865845,
-      "learning_rate": 5.518994974874372e-05,
-      "loss": 5.7307,
-      "step": 45102
-    },
-    {
-      "epoch": 23.521773142112124,
-      "grad_norm": 1.5942606925964355,
-      "learning_rate": 5.5188944723618095e-05,
-      "loss": 5.262,
-      "step": 45103
-    },
-    {
-      "epoch": 23.522294654498044,
-      "grad_norm": 1.5755467414855957,
-      "learning_rate": 5.5187939698492466e-05,
-      "loss": 5.2819,
-      "step": 45104
-    },
-    {
-      "epoch": 23.522816166883963,
-      "grad_norm": 1.5891109704971313,
-      "learning_rate": 5.5186934673366844e-05,
-      "loss": 4.9078,
-      "step": 45105
-    },
-    {
-      "epoch": 23.523337679269883,
-      "grad_norm": 1.529007911682129,
-      "learning_rate": 5.518592964824121e-05,
-      "loss": 5.5188,
-      "step": 45106
-    },
-    {
-      "epoch": 23.523859191655802,
-      "grad_norm": 1.4872251749038696,
-      "learning_rate": 5.5184924623115586e-05,
-      "loss": 5.2645,
-      "step": 45107
-    },
-    {
-      "epoch": 23.52438070404172,
-      "grad_norm": 1.6716161966323853,
-      "learning_rate": 5.518391959798995e-05,
-      "loss": 5.3739,
-      "step": 45108
-    },
-    {
-      "epoch": 23.52490221642764,
-      "grad_norm": 1.503151774406433,
-      "learning_rate": 5.518291457286432e-05,
-      "loss": 5.3105,
-      "step": 45109
-    },
-    {
-      "epoch": 23.52542372881356,
-      "grad_norm": 1.6200631856918335,
-      "learning_rate": 5.51819095477387e-05,
-      "loss": 4.7666,
-      "step": 45110
-    },
-    {
-      "epoch": 23.52594524119948,
-      "grad_norm": 1.6448620557785034,
-      "learning_rate": 5.5180904522613064e-05,
-      "loss": 5.4608,
-      "step": 45111
-    },
-    {
-      "epoch": 23.526466753585396,
-      "grad_norm": 1.600652813911438,
-      "learning_rate": 5.517989949748744e-05,
-      "loss": 4.8478,
-      "step": 45112
-    },
-    {
-      "epoch": 23.526988265971315,
-      "grad_norm": 1.551236867904663,
-      "learning_rate": 5.5178894472361806e-05,
-      "loss": 5.4102,
-      "step": 45113
-    },
-    {
-      "epoch": 23.527509778357235,
-      "grad_norm": 1.4383474588394165,
-      "learning_rate": 5.5177889447236184e-05,
-      "loss": 4.9946,
-      "step": 45114
-    },
-    {
-      "epoch": 23.528031290743154,
-      "grad_norm": 1.4939708709716797,
-      "learning_rate": 5.5176884422110555e-05,
-      "loss": 5.5127,
-      "step": 45115
-    },
-    {
-      "epoch": 23.528552803129074,
-      "grad_norm": 1.6286665201187134,
-      "learning_rate": 5.517587939698493e-05,
-      "loss": 4.9197,
-      "step": 45116
-    },
-    {
-      "epoch": 23.529074315514993,
-      "grad_norm": 1.4824366569519043,
-      "learning_rate": 5.51748743718593e-05,
-      "loss": 4.561,
-      "step": 45117
-    },
-    {
-      "epoch": 23.529595827900913,
-      "grad_norm": 1.524134635925293,
-      "learning_rate": 5.5173869346733674e-05,
-      "loss": 5.4218,
-      "step": 45118
-    },
-    {
-      "epoch": 23.530117340286832,
-      "grad_norm": 1.425180435180664,
-      "learning_rate": 5.517286432160804e-05,
-      "loss": 5.5388,
-      "step": 45119
-    },
-    {
-      "epoch": 23.53063885267275,
-      "grad_norm": 1.4767742156982422,
-      "learning_rate": 5.5171859296482417e-05,
-      "loss": 5.2861,
-      "step": 45120
-    },
-    {
-      "epoch": 23.53116036505867,
-      "grad_norm": 1.5365070104599,
-      "learning_rate": 5.517085427135679e-05,
-      "loss": 5.191,
-      "step": 45121
-    },
-    {
-      "epoch": 23.53168187744459,
-      "grad_norm": 1.491297721862793,
-      "learning_rate": 5.516984924623115e-05,
-      "loss": 5.0401,
-      "step": 45122
-    },
-    {
-      "epoch": 23.53220338983051,
-      "grad_norm": 1.701664686203003,
-      "learning_rate": 5.516884422110553e-05,
-      "loss": 4.8309,
-      "step": 45123
-    },
-    {
-      "epoch": 23.532724902216426,
-      "grad_norm": 1.5275886058807373,
-      "learning_rate": 5.51678391959799e-05,
-      "loss": 4.7821,
-      "step": 45124
-    },
-    {
-      "epoch": 23.533246414602345,
-      "grad_norm": 1.6087068319320679,
-      "learning_rate": 5.516683417085428e-05,
-      "loss": 4.9402,
-      "step": 45125
-    },
-    {
-      "epoch": 23.533767926988265,
-      "grad_norm": 1.6394647359848022,
-      "learning_rate": 5.516582914572864e-05,
-      "loss": 5.5365,
-      "step": 45126
-    },
-    {
-      "epoch": 23.534289439374184,
-      "grad_norm": 1.5398077964782715,
-      "learning_rate": 5.516482412060302e-05,
-      "loss": 5.1148,
-      "step": 45127
-    },
-    {
-      "epoch": 23.534810951760104,
-      "grad_norm": 1.6146302223205566,
-      "learning_rate": 5.5163819095477385e-05,
-      "loss": 4.9711,
-      "step": 45128
-    },
-    {
-      "epoch": 23.535332464146023,
-      "grad_norm": 1.5116279125213623,
-      "learning_rate": 5.516281407035176e-05,
-      "loss": 5.5458,
-      "step": 45129
-    },
-    {
-      "epoch": 23.535853976531943,
-      "grad_norm": 1.5122876167297363,
-      "learning_rate": 5.5161809045226134e-05,
-      "loss": 5.4928,
-      "step": 45130
-    },
-    {
-      "epoch": 23.536375488917862,
-      "grad_norm": 1.5672675371170044,
-      "learning_rate": 5.516080402010051e-05,
-      "loss": 5.2158,
-      "step": 45131
-    },
-    {
-      "epoch": 23.53689700130378,
-      "grad_norm": 1.6747959852218628,
-      "learning_rate": 5.5159798994974876e-05,
-      "loss": 4.941,
-      "step": 45132
-    },
-    {
-      "epoch": 23.5374185136897,
-      "grad_norm": 1.7433459758758545,
-      "learning_rate": 5.5158793969849254e-05,
-      "loss": 5.0159,
-      "step": 45133
-    },
-    {
-      "epoch": 23.53794002607562,
-      "grad_norm": 1.521287202835083,
-      "learning_rate": 5.515778894472362e-05,
-      "loss": 4.9901,
-      "step": 45134
-    },
-    {
-      "epoch": 23.53846153846154,
-      "grad_norm": 1.5506750345230103,
-      "learning_rate": 5.515678391959799e-05,
-      "loss": 5.0103,
-      "step": 45135
-    },
-    {
-      "epoch": 23.538983050847456,
-      "grad_norm": 1.6222010850906372,
-      "learning_rate": 5.515577889447237e-05,
-      "loss": 5.2147,
-      "step": 45136
-    },
-    {
-      "epoch": 23.539504563233375,
-      "grad_norm": 1.597408413887024,
-      "learning_rate": 5.515477386934673e-05,
-      "loss": 5.2873,
-      "step": 45137
-    },
-    {
-      "epoch": 23.540026075619295,
-      "grad_norm": 1.4901565313339233,
-      "learning_rate": 5.515376884422111e-05,
-      "loss": 5.3233,
-      "step": 45138
-    },
-    {
-      "epoch": 23.540547588005214,
-      "grad_norm": 1.472495675086975,
-      "learning_rate": 5.515276381909547e-05,
-      "loss": 5.1347,
-      "step": 45139
-    },
-    {
-      "epoch": 23.541069100391134,
-      "grad_norm": 1.5472133159637451,
-      "learning_rate": 5.515175879396985e-05,
-      "loss": 5.4053,
-      "step": 45140
-    },
-    {
-      "epoch": 23.541590612777053,
-      "grad_norm": 1.3495067358016968,
-      "learning_rate": 5.515075376884422e-05,
-      "loss": 5.862,
-      "step": 45141
-    },
-    {
-      "epoch": 23.542112125162973,
-      "grad_norm": 1.5349332094192505,
-      "learning_rate": 5.51497487437186e-05,
-      "loss": 4.9914,
-      "step": 45142
-    },
-    {
-      "epoch": 23.542633637548892,
-      "grad_norm": 1.5606087446212769,
-      "learning_rate": 5.5148743718592964e-05,
-      "loss": 4.4232,
-      "step": 45143
-    },
-    {
-      "epoch": 23.54315514993481,
-      "grad_norm": 1.6113998889923096,
-      "learning_rate": 5.514773869346734e-05,
-      "loss": 4.8711,
-      "step": 45144
-    },
-    {
-      "epoch": 23.54367666232073,
-      "grad_norm": 1.5672622919082642,
-      "learning_rate": 5.514673366834171e-05,
-      "loss": 5.2802,
-      "step": 45145
-    },
-    {
-      "epoch": 23.54419817470665,
-      "grad_norm": 1.5892664194107056,
-      "learning_rate": 5.514572864321609e-05,
-      "loss": 4.989,
-      "step": 45146
-    },
-    {
-      "epoch": 23.54471968709257,
-      "grad_norm": 1.6486588716506958,
-      "learning_rate": 5.5144723618090455e-05,
-      "loss": 5.1183,
-      "step": 45147
-    },
-    {
-      "epoch": 23.545241199478486,
-      "grad_norm": 1.488864541053772,
-      "learning_rate": 5.514371859296482e-05,
-      "loss": 5.4434,
-      "step": 45148
-    },
-    {
-      "epoch": 23.545762711864406,
-      "grad_norm": 1.4675776958465576,
-      "learning_rate": 5.51427135678392e-05,
-      "loss": 5.5661,
-      "step": 45149
-    },
-    {
-      "epoch": 23.546284224250325,
-      "grad_norm": 1.5645157098770142,
-      "learning_rate": 5.514170854271357e-05,
-      "loss": 5.0815,
-      "step": 45150
-    },
-    {
-      "epoch": 23.546805736636244,
-      "grad_norm": 1.6107739210128784,
-      "learning_rate": 5.5140703517587946e-05,
-      "loss": 5.3177,
-      "step": 45151
-    },
-    {
-      "epoch": 23.547327249022164,
-      "grad_norm": 1.6017814874649048,
-      "learning_rate": 5.513969849246231e-05,
-      "loss": 5.4182,
-      "step": 45152
-    },
-    {
-      "epoch": 23.547848761408083,
-      "grad_norm": 1.5589812994003296,
-      "learning_rate": 5.513869346733669e-05,
-      "loss": 5.1569,
-      "step": 45153
-    },
-    {
-      "epoch": 23.548370273794003,
-      "grad_norm": 1.6451574563980103,
-      "learning_rate": 5.513768844221105e-05,
-      "loss": 4.7715,
-      "step": 45154
-    },
-    {
-      "epoch": 23.548891786179922,
-      "grad_norm": 1.6286566257476807,
-      "learning_rate": 5.513668341708543e-05,
-      "loss": 5.1163,
-      "step": 45155
-    },
-    {
-      "epoch": 23.54941329856584,
-      "grad_norm": 1.5601731538772583,
-      "learning_rate": 5.51356783919598e-05,
-      "loss": 5.158,
-      "step": 45156
-    },
-    {
-      "epoch": 23.54993481095176,
-      "grad_norm": 1.6185295581817627,
-      "learning_rate": 5.513467336683418e-05,
-      "loss": 4.8593,
-      "step": 45157
-    },
-    {
-      "epoch": 23.55045632333768,
-      "grad_norm": 1.5209765434265137,
-      "learning_rate": 5.5133668341708544e-05,
-      "loss": 5.1588,
-      "step": 45158
-    },
-    {
-      "epoch": 23.5509778357236,
-      "grad_norm": 1.5844625234603882,
-      "learning_rate": 5.513266331658292e-05,
-      "loss": 4.7454,
-      "step": 45159
-    },
-    {
-      "epoch": 23.551499348109516,
-      "grad_norm": 1.56330144405365,
-      "learning_rate": 5.5131658291457286e-05,
-      "loss": 5.3782,
-      "step": 45160
-    },
-    {
-      "epoch": 23.552020860495436,
-      "grad_norm": 1.5634639263153076,
-      "learning_rate": 5.513065326633166e-05,
-      "loss": 5.2309,
-      "step": 45161
-    },
-    {
-      "epoch": 23.552542372881355,
-      "grad_norm": 1.5007258653640747,
-      "learning_rate": 5.5129648241206035e-05,
-      "loss": 4.598,
-      "step": 45162
-    },
-    {
-      "epoch": 23.553063885267274,
-      "grad_norm": 1.4781451225280762,
-      "learning_rate": 5.51286432160804e-05,
-      "loss": 5.658,
-      "step": 45163
-    },
-    {
-      "epoch": 23.553585397653194,
-      "grad_norm": 1.5620771646499634,
-      "learning_rate": 5.512763819095478e-05,
-      "loss": 5.4203,
-      "step": 45164
-    },
-    {
-      "epoch": 23.554106910039113,
-      "grad_norm": 1.6020227670669556,
-      "learning_rate": 5.512663316582915e-05,
-      "loss": 5.4699,
-      "step": 45165
-    },
-    {
-      "epoch": 23.554628422425033,
-      "grad_norm": 1.7678658962249756,
-      "learning_rate": 5.5125628140703526e-05,
-      "loss": 4.9287,
-      "step": 45166
-    },
-    {
-      "epoch": 23.555149934810952,
-      "grad_norm": 1.5473146438598633,
-      "learning_rate": 5.512462311557789e-05,
-      "loss": 5.2885,
-      "step": 45167
-    },
-    {
-      "epoch": 23.555671447196872,
-      "grad_norm": 1.6008638143539429,
-      "learning_rate": 5.512361809045227e-05,
-      "loss": 5.3295,
-      "step": 45168
-    },
-    {
-      "epoch": 23.55619295958279,
-      "grad_norm": 1.7466163635253906,
-      "learning_rate": 5.512261306532663e-05,
-      "loss": 4.7897,
-      "step": 45169
-    },
-    {
-      "epoch": 23.55671447196871,
-      "grad_norm": 1.585847020149231,
-      "learning_rate": 5.512160804020101e-05,
-      "loss": 5.5803,
-      "step": 45170
-    },
-    {
-      "epoch": 23.557235984354627,
-      "grad_norm": 1.5720826387405396,
-      "learning_rate": 5.512060301507538e-05,
-      "loss": 5.1498,
-      "step": 45171
-    },
-    {
-      "epoch": 23.557757496740546,
-      "grad_norm": 1.5221236944198608,
-      "learning_rate": 5.511959798994976e-05,
-      "loss": 5.3903,
-      "step": 45172
-    },
-    {
-      "epoch": 23.558279009126466,
-      "grad_norm": 1.53538978099823,
-      "learning_rate": 5.511859296482412e-05,
-      "loss": 5.5028,
-      "step": 45173
-    },
-    {
-      "epoch": 23.558800521512385,
-      "grad_norm": 1.60988187789917,
-      "learning_rate": 5.51175879396985e-05,
-      "loss": 4.1883,
-      "step": 45174
-    },
-    {
-      "epoch": 23.559322033898304,
-      "grad_norm": 1.6525676250457764,
-      "learning_rate": 5.5116582914572865e-05,
-      "loss": 4.9079,
-      "step": 45175
-    },
-    {
-      "epoch": 23.559843546284224,
-      "grad_norm": 1.5581003427505493,
-      "learning_rate": 5.5115577889447236e-05,
-      "loss": 5.1303,
-      "step": 45176
-    },
-    {
-      "epoch": 23.560365058670143,
-      "grad_norm": 1.4955047369003296,
-      "learning_rate": 5.5114572864321614e-05,
-      "loss": 5.5612,
-      "step": 45177
-    },
-    {
-      "epoch": 23.560886571056063,
-      "grad_norm": 1.4893156290054321,
-      "learning_rate": 5.511356783919598e-05,
-      "loss": 4.9962,
-      "step": 45178
-    },
-    {
-      "epoch": 23.561408083441982,
-      "grad_norm": 1.5251617431640625,
-      "learning_rate": 5.5112562814070356e-05,
-      "loss": 4.8926,
-      "step": 45179
-    },
-    {
-      "epoch": 23.561929595827902,
-      "grad_norm": 1.574378252029419,
-      "learning_rate": 5.511155778894472e-05,
-      "loss": 5.4265,
-      "step": 45180
-    },
-    {
-      "epoch": 23.56245110821382,
-      "grad_norm": 1.6475176811218262,
-      "learning_rate": 5.51105527638191e-05,
-      "loss": 4.9328,
-      "step": 45181
-    },
-    {
-      "epoch": 23.56297262059974,
-      "grad_norm": 1.612644076347351,
-      "learning_rate": 5.510954773869347e-05,
-      "loss": 5.1772,
-      "step": 45182
-    },
-    {
-      "epoch": 23.56349413298566,
-      "grad_norm": 1.5410295724868774,
-      "learning_rate": 5.510854271356785e-05,
-      "loss": 5.4076,
-      "step": 45183
-    },
-    {
-      "epoch": 23.564015645371576,
-      "grad_norm": 1.5774060487747192,
-      "learning_rate": 5.510753768844221e-05,
-      "loss": 5.2228,
-      "step": 45184
-    },
-    {
-      "epoch": 23.564537157757496,
-      "grad_norm": 1.6848657131195068,
-      "learning_rate": 5.510653266331659e-05,
-      "loss": 4.6848,
-      "step": 45185
-    },
-    {
-      "epoch": 23.565058670143415,
-      "grad_norm": 1.668009638786316,
-      "learning_rate": 5.510552763819096e-05,
-      "loss": 4.8957,
-      "step": 45186
-    },
-    {
-      "epoch": 23.565580182529335,
-      "grad_norm": 1.5051920413970947,
-      "learning_rate": 5.510452261306534e-05,
-      "loss": 5.4218,
-      "step": 45187
-    },
-    {
-      "epoch": 23.566101694915254,
-      "grad_norm": 1.5622782707214355,
-      "learning_rate": 5.51035175879397e-05,
-      "loss": 5.2783,
-      "step": 45188
-    },
-    {
-      "epoch": 23.566623207301173,
-      "grad_norm": 1.5067251920700073,
-      "learning_rate": 5.5102512562814067e-05,
-      "loss": 5.3043,
-      "step": 45189
-    },
-    {
-      "epoch": 23.567144719687093,
-      "grad_norm": 1.4599759578704834,
-      "learning_rate": 5.5101507537688444e-05,
-      "loss": 5.302,
-      "step": 45190
-    },
-    {
-      "epoch": 23.567666232073012,
-      "grad_norm": 1.6299688816070557,
-      "learning_rate": 5.5100502512562815e-05,
-      "loss": 5.2553,
-      "step": 45191
-    },
-    {
-      "epoch": 23.568187744458932,
-      "grad_norm": 1.6544501781463623,
-      "learning_rate": 5.509949748743719e-05,
-      "loss": 4.8659,
-      "step": 45192
-    },
-    {
-      "epoch": 23.56870925684485,
-      "grad_norm": 1.585411548614502,
-      "learning_rate": 5.509849246231156e-05,
-      "loss": 5.0501,
-      "step": 45193
-    },
-    {
-      "epoch": 23.56923076923077,
-      "grad_norm": 1.687155842781067,
-      "learning_rate": 5.5097487437185935e-05,
-      "loss": 5.2617,
-      "step": 45194
-    },
-    {
-      "epoch": 23.569752281616687,
-      "grad_norm": 1.5199084281921387,
-      "learning_rate": 5.50964824120603e-05,
-      "loss": 5.4283,
-      "step": 45195
-    },
-    {
-      "epoch": 23.570273794002606,
-      "grad_norm": 1.6087290048599243,
-      "learning_rate": 5.509547738693468e-05,
-      "loss": 5.3532,
-      "step": 45196
-    },
-    {
-      "epoch": 23.570795306388526,
-      "grad_norm": 1.5891603231430054,
-      "learning_rate": 5.509447236180905e-05,
-      "loss": 4.9642,
-      "step": 45197
-    },
-    {
-      "epoch": 23.571316818774445,
-      "grad_norm": 1.5311102867126465,
-      "learning_rate": 5.5093467336683426e-05,
-      "loss": 5.4067,
-      "step": 45198
-    },
-    {
-      "epoch": 23.571838331160365,
-      "grad_norm": 1.4934756755828857,
-      "learning_rate": 5.509246231155779e-05,
-      "loss": 5.3914,
-      "step": 45199
-    },
-    {
-      "epoch": 23.572359843546284,
-      "grad_norm": 1.5320544242858887,
-      "learning_rate": 5.509145728643217e-05,
-      "loss": 5.5151,
-      "step": 45200
-    },
-    {
-      "epoch": 23.572881355932203,
-      "grad_norm": 1.7711284160614014,
-      "learning_rate": 5.509045226130653e-05,
-      "loss": 4.9506,
-      "step": 45201
-    },
-    {
-      "epoch": 23.573402868318123,
-      "grad_norm": 1.6130386590957642,
-      "learning_rate": 5.5089447236180904e-05,
-      "loss": 4.8548,
-      "step": 45202
-    },
-    {
-      "epoch": 23.573924380704042,
-      "grad_norm": 1.5604774951934814,
-      "learning_rate": 5.508844221105528e-05,
-      "loss": 5.5504,
-      "step": 45203
-    },
-    {
-      "epoch": 23.574445893089962,
-      "grad_norm": 1.4515950679779053,
-      "learning_rate": 5.5087437185929646e-05,
-      "loss": 5.43,
-      "step": 45204
-    },
-    {
-      "epoch": 23.57496740547588,
-      "grad_norm": 1.5292539596557617,
-      "learning_rate": 5.5086432160804024e-05,
-      "loss": 5.2935,
-      "step": 45205
-    },
-    {
-      "epoch": 23.5754889178618,
-      "grad_norm": 1.5900356769561768,
-      "learning_rate": 5.508542713567839e-05,
-      "loss": 5.4402,
-      "step": 45206
-    },
-    {
-      "epoch": 23.576010430247717,
-      "grad_norm": 1.5139657258987427,
-      "learning_rate": 5.5084422110552766e-05,
-      "loss": 4.8596,
-      "step": 45207
-    },
-    {
-      "epoch": 23.576531942633636,
-      "grad_norm": 1.4952867031097412,
-      "learning_rate": 5.508341708542714e-05,
-      "loss": 5.3075,
-      "step": 45208
-    },
-    {
-      "epoch": 23.577053455019556,
-      "grad_norm": 1.5425843000411987,
-      "learning_rate": 5.5082412060301515e-05,
-      "loss": 5.5149,
-      "step": 45209
-    },
-    {
-      "epoch": 23.577574967405475,
-      "grad_norm": 1.5261116027832031,
-      "learning_rate": 5.508140703517588e-05,
-      "loss": 5.3478,
-      "step": 45210
-    },
-    {
-      "epoch": 23.578096479791395,
-      "grad_norm": 1.403017520904541,
-      "learning_rate": 5.508040201005026e-05,
-      "loss": 4.1932,
-      "step": 45211
-    },
-    {
-      "epoch": 23.578617992177314,
-      "grad_norm": 1.5294585227966309,
-      "learning_rate": 5.507939698492463e-05,
-      "loss": 5.5274,
-      "step": 45212
-    },
-    {
-      "epoch": 23.579139504563233,
-      "grad_norm": 1.707960844039917,
-      "learning_rate": 5.5078391959799006e-05,
-      "loss": 5.0986,
-      "step": 45213
-    },
-    {
-      "epoch": 23.579661016949153,
-      "grad_norm": 1.4405442476272583,
-      "learning_rate": 5.507738693467337e-05,
-      "loss": 5.4452,
-      "step": 45214
-    },
-    {
-      "epoch": 23.580182529335072,
-      "grad_norm": 1.6236084699630737,
-      "learning_rate": 5.5076381909547734e-05,
-      "loss": 5.2728,
-      "step": 45215
-    },
-    {
-      "epoch": 23.580704041720992,
-      "grad_norm": 1.5963010787963867,
-      "learning_rate": 5.507537688442211e-05,
-      "loss": 5.5195,
-      "step": 45216
-    },
-    {
-      "epoch": 23.58122555410691,
-      "grad_norm": 1.5697970390319824,
-      "learning_rate": 5.507437185929648e-05,
-      "loss": 5.6127,
-      "step": 45217
-    },
-    {
-      "epoch": 23.58174706649283,
-      "grad_norm": 1.5806407928466797,
-      "learning_rate": 5.507336683417086e-05,
-      "loss": 5.1282,
-      "step": 45218
-    },
-    {
-      "epoch": 23.582268578878747,
-      "grad_norm": 1.4499099254608154,
-      "learning_rate": 5.5072361809045225e-05,
-      "loss": 5.722,
-      "step": 45219
-    },
-    {
-      "epoch": 23.582790091264666,
-      "grad_norm": 1.5740374326705933,
-      "learning_rate": 5.50713567839196e-05,
-      "loss": 5.1852,
-      "step": 45220
-    },
-    {
-      "epoch": 23.583311603650586,
-      "grad_norm": 1.7124232053756714,
-      "learning_rate": 5.507035175879397e-05,
-      "loss": 5.2668,
-      "step": 45221
-    },
-    {
-      "epoch": 23.583833116036505,
-      "grad_norm": 1.5908503532409668,
-      "learning_rate": 5.5069346733668345e-05,
-      "loss": 5.0748,
-      "step": 45222
-    },
-    {
-      "epoch": 23.584354628422425,
-      "grad_norm": 1.5473947525024414,
-      "learning_rate": 5.5068341708542716e-05,
-      "loss": 5.4818,
-      "step": 45223
-    },
-    {
-      "epoch": 23.584876140808344,
-      "grad_norm": 1.558558702468872,
-      "learning_rate": 5.5067336683417094e-05,
-      "loss": 5.2866,
-      "step": 45224
-    },
-    {
-      "epoch": 23.585397653194264,
-      "grad_norm": 1.521040678024292,
-      "learning_rate": 5.506633165829146e-05,
-      "loss": 5.6871,
-      "step": 45225
-    },
-    {
-      "epoch": 23.585919165580183,
-      "grad_norm": 1.5595734119415283,
-      "learning_rate": 5.5065326633165836e-05,
-      "loss": 5.4927,
-      "step": 45226
-    },
-    {
-      "epoch": 23.586440677966102,
-      "grad_norm": 1.522033929824829,
-      "learning_rate": 5.50643216080402e-05,
-      "loss": 4.9662,
-      "step": 45227
-    },
-    {
-      "epoch": 23.586962190352022,
-      "grad_norm": 1.548729419708252,
-      "learning_rate": 5.506331658291457e-05,
-      "loss": 5.2696,
-      "step": 45228
-    },
-    {
-      "epoch": 23.58748370273794,
-      "grad_norm": 1.5914177894592285,
-      "learning_rate": 5.506231155778895e-05,
-      "loss": 4.8992,
-      "step": 45229
-    },
-    {
-      "epoch": 23.58800521512386,
-      "grad_norm": 1.6128554344177246,
-      "learning_rate": 5.5061306532663314e-05,
-      "loss": 5.3303,
-      "step": 45230
-    },
-    {
-      "epoch": 23.588526727509777,
-      "grad_norm": 1.4604445695877075,
-      "learning_rate": 5.506030150753769e-05,
-      "loss": 5.7304,
-      "step": 45231
-    },
-    {
-      "epoch": 23.589048239895696,
-      "grad_norm": 1.4736802577972412,
-      "learning_rate": 5.505929648241206e-05,
-      "loss": 5.51,
-      "step": 45232
-    },
-    {
-      "epoch": 23.589569752281616,
-      "grad_norm": 1.5698902606964111,
-      "learning_rate": 5.505829145728644e-05,
-      "loss": 5.2221,
-      "step": 45233
-    },
-    {
-      "epoch": 23.590091264667535,
-      "grad_norm": 1.5999317169189453,
-      "learning_rate": 5.5057286432160805e-05,
-      "loss": 5.3865,
-      "step": 45234
-    },
-    {
-      "epoch": 23.590612777053455,
-      "grad_norm": 1.528164029121399,
-      "learning_rate": 5.505628140703518e-05,
-      "loss": 5.2398,
-      "step": 45235
-    },
-    {
-      "epoch": 23.591134289439374,
-      "grad_norm": 1.5013551712036133,
-      "learning_rate": 5.505527638190955e-05,
-      "loss": 5.7581,
-      "step": 45236
-    },
-    {
-      "epoch": 23.591655801825294,
-      "grad_norm": 1.5219188928604126,
-      "learning_rate": 5.5054271356783925e-05,
-      "loss": 4.9211,
-      "step": 45237
-    },
-    {
-      "epoch": 23.592177314211213,
-      "grad_norm": 1.4756724834442139,
-      "learning_rate": 5.5053266331658296e-05,
-      "loss": 5.5095,
-      "step": 45238
-    },
-    {
-      "epoch": 23.592698826597132,
-      "grad_norm": 1.670495867729187,
-      "learning_rate": 5.5052261306532673e-05,
-      "loss": 5.1948,
-      "step": 45239
-    },
-    {
-      "epoch": 23.593220338983052,
-      "grad_norm": 1.6014080047607422,
-      "learning_rate": 5.505125628140704e-05,
-      "loss": 4.8507,
-      "step": 45240
-    },
-    {
-      "epoch": 23.59374185136897,
-      "grad_norm": 1.5810050964355469,
-      "learning_rate": 5.50502512562814e-05,
-      "loss": 5.1633,
-      "step": 45241
-    },
-    {
-      "epoch": 23.59426336375489,
-      "grad_norm": 1.573981523513794,
-      "learning_rate": 5.504924623115578e-05,
-      "loss": 5.4471,
-      "step": 45242
-    },
-    {
-      "epoch": 23.594784876140807,
-      "grad_norm": 1.6828585863113403,
-      "learning_rate": 5.504824120603015e-05,
-      "loss": 5.1566,
-      "step": 45243
-    },
-    {
-      "epoch": 23.595306388526726,
-      "grad_norm": 1.6379529237747192,
-      "learning_rate": 5.504723618090453e-05,
-      "loss": 4.7593,
-      "step": 45244
-    },
-    {
-      "epoch": 23.595827900912646,
-      "grad_norm": 1.6080615520477295,
-      "learning_rate": 5.504623115577889e-05,
-      "loss": 5.4472,
-      "step": 45245
-    },
-    {
-      "epoch": 23.596349413298565,
-      "grad_norm": 1.5610448122024536,
-      "learning_rate": 5.504522613065327e-05,
-      "loss": 5.5119,
-      "step": 45246
-    },
-    {
-      "epoch": 23.596870925684485,
-      "grad_norm": 1.6764482259750366,
-      "learning_rate": 5.5044221105527635e-05,
-      "loss": 4.5919,
-      "step": 45247
-    },
-    {
-      "epoch": 23.597392438070404,
-      "grad_norm": 1.6057460308074951,
-      "learning_rate": 5.504321608040201e-05,
-      "loss": 5.1253,
-      "step": 45248
-    },
-    {
-      "epoch": 23.597913950456324,
-      "grad_norm": 1.4662604331970215,
-      "learning_rate": 5.5042211055276384e-05,
-      "loss": 4.9188,
-      "step": 45249
-    },
-    {
-      "epoch": 23.598435462842243,
-      "grad_norm": 1.4419028759002686,
-      "learning_rate": 5.504120603015076e-05,
-      "loss": 5.2653,
-      "step": 45250
-    },
-    {
-      "epoch": 23.598956975228162,
-      "grad_norm": 1.4595078229904175,
-      "learning_rate": 5.5040201005025126e-05,
-      "loss": 5.5243,
-      "step": 45251
-    },
-    {
-      "epoch": 23.599478487614082,
-      "grad_norm": 1.6463961601257324,
-      "learning_rate": 5.5039195979899504e-05,
-      "loss": 4.919,
-      "step": 45252
-    },
-    {
-      "epoch": 23.6,
-      "grad_norm": 1.5563658475875854,
-      "learning_rate": 5.5038190954773875e-05,
-      "loss": 5.0708,
-      "step": 45253
-    },
-    {
-      "epoch": 23.60052151238592,
-      "grad_norm": 1.5095690488815308,
-      "learning_rate": 5.503718592964824e-05,
-      "loss": 4.7389,
-      "step": 45254
-    },
-    {
-      "epoch": 23.601043024771837,
-      "grad_norm": 1.6362158060073853,
-      "learning_rate": 5.503618090452262e-05,
-      "loss": 4.465,
-      "step": 45255
-    },
-    {
-      "epoch": 23.601564537157756,
-      "grad_norm": 1.6024861335754395,
-      "learning_rate": 5.503517587939698e-05,
-      "loss": 4.7564,
-      "step": 45256
-    },
-    {
-      "epoch": 23.602086049543676,
-      "grad_norm": 1.6377919912338257,
-      "learning_rate": 5.503417085427136e-05,
-      "loss": 4.8566,
-      "step": 45257
-    },
-    {
-      "epoch": 23.602607561929595,
-      "grad_norm": 1.4983742237091064,
-      "learning_rate": 5.503316582914573e-05,
-      "loss": 5.1027,
-      "step": 45258
-    },
-    {
-      "epoch": 23.603129074315515,
-      "grad_norm": 1.578943133354187,
-      "learning_rate": 5.503216080402011e-05,
-      "loss": 5.241,
-      "step": 45259
-    },
-    {
-      "epoch": 23.603650586701434,
-      "grad_norm": 1.545652151107788,
-      "learning_rate": 5.503115577889447e-05,
-      "loss": 5.4391,
-      "step": 45260
-    },
-    {
-      "epoch": 23.604172099087354,
-      "grad_norm": 1.4662399291992188,
-      "learning_rate": 5.503015075376885e-05,
-      "loss": 5.0974,
-      "step": 45261
-    },
-    {
-      "epoch": 23.604693611473273,
-      "grad_norm": 1.5108139514923096,
-      "learning_rate": 5.5029145728643214e-05,
-      "loss": 5.7098,
-      "step": 45262
-    },
-    {
-      "epoch": 23.605215123859193,
-      "grad_norm": 1.4686954021453857,
-      "learning_rate": 5.502814070351759e-05,
-      "loss": 5.3942,
-      "step": 45263
-    },
-    {
-      "epoch": 23.605736636245112,
-      "grad_norm": 1.624500036239624,
-      "learning_rate": 5.502713567839196e-05,
-      "loss": 4.8201,
-      "step": 45264
-    },
-    {
-      "epoch": 23.60625814863103,
-      "grad_norm": 1.5154461860656738,
-      "learning_rate": 5.502613065326634e-05,
-      "loss": 5.428,
-      "step": 45265
-    },
-    {
-      "epoch": 23.60677966101695,
-      "grad_norm": 1.572263479232788,
-      "learning_rate": 5.5025125628140705e-05,
-      "loss": 5.1764,
-      "step": 45266
-    },
-    {
-      "epoch": 23.607301173402867,
-      "grad_norm": 1.5162533521652222,
-      "learning_rate": 5.502412060301508e-05,
-      "loss": 4.6596,
-      "step": 45267
-    },
-    {
-      "epoch": 23.607822685788786,
-      "grad_norm": 1.5757859945297241,
-      "learning_rate": 5.502311557788945e-05,
-      "loss": 5.2003,
-      "step": 45268
-    },
-    {
-      "epoch": 23.608344198174706,
-      "grad_norm": 1.7127867937088013,
-      "learning_rate": 5.502211055276382e-05,
-      "loss": 5.3312,
-      "step": 45269
-    },
-    {
-      "epoch": 23.608865710560625,
-      "grad_norm": 1.5678132772445679,
-      "learning_rate": 5.5021105527638196e-05,
-      "loss": 5.4483,
-      "step": 45270
-    },
-    {
-      "epoch": 23.609387222946545,
-      "grad_norm": 1.6114519834518433,
-      "learning_rate": 5.502010050251256e-05,
-      "loss": 5.6719,
-      "step": 45271
-    },
-    {
-      "epoch": 23.609908735332464,
-      "grad_norm": 1.5737437009811401,
-      "learning_rate": 5.501909547738694e-05,
-      "loss": 5.0614,
-      "step": 45272
-    },
-    {
-      "epoch": 23.610430247718384,
-      "grad_norm": 1.5928925275802612,
-      "learning_rate": 5.501809045226131e-05,
-      "loss": 5.3108,
-      "step": 45273
-    },
-    {
-      "epoch": 23.610951760104303,
-      "grad_norm": 1.560725212097168,
-      "learning_rate": 5.501708542713569e-05,
-      "loss": 5.5557,
-      "step": 45274
-    },
-    {
-      "epoch": 23.611473272490223,
-      "grad_norm": 1.6528434753417969,
-      "learning_rate": 5.501608040201005e-05,
-      "loss": 5.1965,
-      "step": 45275
-    },
-    {
-      "epoch": 23.611994784876142,
-      "grad_norm": 1.6121461391448975,
-      "learning_rate": 5.501507537688443e-05,
-      "loss": 5.1921,
-      "step": 45276
-    },
-    {
-      "epoch": 23.61251629726206,
-      "grad_norm": 1.55232572555542,
-      "learning_rate": 5.5014070351758794e-05,
-      "loss": 5.2445,
-      "step": 45277
-    },
-    {
-      "epoch": 23.613037809647977,
-      "grad_norm": 1.5917562246322632,
-      "learning_rate": 5.501306532663317e-05,
-      "loss": 5.3772,
-      "step": 45278
-    },
-    {
-      "epoch": 23.613559322033897,
-      "grad_norm": 1.5021917819976807,
-      "learning_rate": 5.501206030150754e-05,
-      "loss": 4.7516,
-      "step": 45279
-    },
-    {
-      "epoch": 23.614080834419816,
-      "grad_norm": 1.5773768424987793,
-      "learning_rate": 5.501105527638192e-05,
-      "loss": 5.1406,
-      "step": 45280
-    },
-    {
-      "epoch": 23.614602346805736,
-      "grad_norm": 1.5924891233444214,
-      "learning_rate": 5.5010050251256285e-05,
-      "loss": 5.1268,
-      "step": 45281
-    },
-    {
-      "epoch": 23.615123859191655,
-      "grad_norm": 1.5497431755065918,
-      "learning_rate": 5.500904522613065e-05,
-      "loss": 5.1177,
-      "step": 45282
-    },
-    {
-      "epoch": 23.615645371577575,
-      "grad_norm": 1.475248098373413,
-      "learning_rate": 5.500804020100503e-05,
-      "loss": 4.9263,
-      "step": 45283
-    },
-    {
-      "epoch": 23.616166883963494,
-      "grad_norm": 1.4931033849716187,
-      "learning_rate": 5.50070351758794e-05,
-      "loss": 5.0478,
-      "step": 45284
-    },
-    {
-      "epoch": 23.616688396349414,
-      "grad_norm": 1.5530096292495728,
-      "learning_rate": 5.5006030150753776e-05,
-      "loss": 4.8942,
-      "step": 45285
-    },
-    {
-      "epoch": 23.617209908735333,
-      "grad_norm": 1.5607755184173584,
-      "learning_rate": 5.500502512562814e-05,
-      "loss": 5.4289,
-      "step": 45286
-    },
-    {
-      "epoch": 23.617731421121253,
-      "grad_norm": 1.6028773784637451,
-      "learning_rate": 5.500402010050252e-05,
-      "loss": 5.3057,
-      "step": 45287
-    },
-    {
-      "epoch": 23.618252933507172,
-      "grad_norm": 1.6737390756607056,
-      "learning_rate": 5.500301507537688e-05,
-      "loss": 5.149,
-      "step": 45288
-    },
-    {
-      "epoch": 23.61877444589309,
-      "grad_norm": 1.4922083616256714,
-      "learning_rate": 5.500201005025126e-05,
-      "loss": 5.5795,
-      "step": 45289
-    },
-    {
-      "epoch": 23.619295958279007,
-      "grad_norm": 1.7001910209655762,
-      "learning_rate": 5.500100502512563e-05,
-      "loss": 5.0932,
-      "step": 45290
-    },
-    {
-      "epoch": 23.619817470664927,
-      "grad_norm": 1.5723458528518677,
-      "learning_rate": 5.500000000000001e-05,
-      "loss": 5.5213,
-      "step": 45291
-    },
-    {
-      "epoch": 23.620338983050846,
-      "grad_norm": 1.5070350170135498,
-      "learning_rate": 5.499899497487437e-05,
-      "loss": 4.6884,
-      "step": 45292
-    },
-    {
-      "epoch": 23.620860495436766,
-      "grad_norm": 1.519818663597107,
-      "learning_rate": 5.499798994974875e-05,
-      "loss": 5.2898,
-      "step": 45293
-    },
-    {
-      "epoch": 23.621382007822685,
-      "grad_norm": 1.549261212348938,
-      "learning_rate": 5.4996984924623115e-05,
-      "loss": 5.2018,
-      "step": 45294
-    },
-    {
-      "epoch": 23.621903520208605,
-      "grad_norm": 1.4388186931610107,
-      "learning_rate": 5.4995979899497486e-05,
-      "loss": 5.5087,
-      "step": 45295
-    },
-    {
-      "epoch": 23.622425032594524,
-      "grad_norm": 1.5441579818725586,
-      "learning_rate": 5.4994974874371864e-05,
-      "loss": 5.6334,
-      "step": 45296
-    },
-    {
-      "epoch": 23.622946544980444,
-      "grad_norm": 1.3835536241531372,
-      "learning_rate": 5.499396984924623e-05,
-      "loss": 5.3155,
-      "step": 45297
-    },
-    {
-      "epoch": 23.623468057366363,
-      "grad_norm": 1.5017846822738647,
-      "learning_rate": 5.4992964824120606e-05,
-      "loss": 5.3502,
-      "step": 45298
-    },
-    {
-      "epoch": 23.623989569752283,
-      "grad_norm": 1.60231614112854,
-      "learning_rate": 5.499195979899498e-05,
-      "loss": 4.7698,
-      "step": 45299
-    },
-    {
-      "epoch": 23.624511082138202,
-      "grad_norm": 1.6030080318450928,
-      "learning_rate": 5.4990954773869355e-05,
-      "loss": 4.6274,
-      "step": 45300
-    },
-    {
-      "epoch": 23.62503259452412,
-      "grad_norm": 1.4322714805603027,
-      "learning_rate": 5.498994974874372e-05,
-      "loss": 5.4673,
-      "step": 45301
-    },
-    {
-      "epoch": 23.625554106910037,
-      "grad_norm": 1.6278003454208374,
-      "learning_rate": 5.49889447236181e-05,
-      "loss": 5.3381,
-      "step": 45302
-    },
-    {
-      "epoch": 23.626075619295957,
-      "grad_norm": 1.6379565000534058,
-      "learning_rate": 5.498793969849246e-05,
-      "loss": 5.3837,
-      "step": 45303
-    },
-    {
-      "epoch": 23.626597131681876,
-      "grad_norm": 1.5679290294647217,
-      "learning_rate": 5.498693467336684e-05,
-      "loss": 5.177,
-      "step": 45304
-    },
-    {
-      "epoch": 23.627118644067796,
-      "grad_norm": 1.6396254301071167,
-      "learning_rate": 5.498592964824121e-05,
-      "loss": 4.9183,
-      "step": 45305
-    },
-    {
-      "epoch": 23.627640156453715,
-      "grad_norm": 1.5953056812286377,
-      "learning_rate": 5.498492462311559e-05,
-      "loss": 4.9769,
-      "step": 45306
-    },
-    {
-      "epoch": 23.628161668839635,
-      "grad_norm": 1.5951595306396484,
-      "learning_rate": 5.498391959798995e-05,
-      "loss": 5.0724,
-      "step": 45307
-    },
-    {
-      "epoch": 23.628683181225554,
-      "grad_norm": 1.5496015548706055,
-      "learning_rate": 5.498291457286432e-05,
-      "loss": 5.2458,
-      "step": 45308
-    },
-    {
-      "epoch": 23.629204693611474,
-      "grad_norm": 1.6966696977615356,
-      "learning_rate": 5.4981909547738694e-05,
-      "loss": 4.9441,
-      "step": 45309
-    },
-    {
-      "epoch": 23.629726205997393,
-      "grad_norm": 1.5768994092941284,
-      "learning_rate": 5.4980904522613065e-05,
-      "loss": 5.39,
-      "step": 45310
-    },
-    {
-      "epoch": 23.630247718383313,
-      "grad_norm": 1.5399878025054932,
-      "learning_rate": 5.497989949748744e-05,
-      "loss": 5.3583,
-      "step": 45311
-    },
-    {
-      "epoch": 23.630769230769232,
-      "grad_norm": 1.5750014781951904,
-      "learning_rate": 5.497889447236181e-05,
-      "loss": 5.2349,
-      "step": 45312
-    },
-    {
-      "epoch": 23.63129074315515,
-      "grad_norm": 1.5129845142364502,
-      "learning_rate": 5.4977889447236185e-05,
-      "loss": 5.4352,
-      "step": 45313
-    },
-    {
-      "epoch": 23.631812255541067,
-      "grad_norm": 1.4325238466262817,
-      "learning_rate": 5.497688442211055e-05,
-      "loss": 5.6206,
-      "step": 45314
-    },
-    {
-      "epoch": 23.632333767926987,
-      "grad_norm": 1.631861686706543,
-      "learning_rate": 5.497587939698493e-05,
-      "loss": 4.8691,
-      "step": 45315
-    },
-    {
-      "epoch": 23.632855280312906,
-      "grad_norm": 1.5593675374984741,
-      "learning_rate": 5.49748743718593e-05,
-      "loss": 4.968,
-      "step": 45316
-    },
-    {
-      "epoch": 23.633376792698826,
-      "grad_norm": 1.6038849353790283,
-      "learning_rate": 5.4973869346733676e-05,
-      "loss": 4.9043,
-      "step": 45317
-    },
-    {
-      "epoch": 23.633898305084745,
-      "grad_norm": 1.5373280048370361,
-      "learning_rate": 5.497286432160804e-05,
-      "loss": 4.8561,
-      "step": 45318
-    },
-    {
-      "epoch": 23.634419817470665,
-      "grad_norm": 1.5237919092178345,
-      "learning_rate": 5.497185929648242e-05,
-      "loss": 5.4643,
-      "step": 45319
-    },
-    {
-      "epoch": 23.634941329856584,
-      "grad_norm": 1.6127339601516724,
-      "learning_rate": 5.497085427135679e-05,
-      "loss": 5.1082,
-      "step": 45320
-    },
-    {
-      "epoch": 23.635462842242504,
-      "grad_norm": 1.5417441129684448,
-      "learning_rate": 5.4969849246231154e-05,
-      "loss": 5.3815,
-      "step": 45321
-    },
-    {
-      "epoch": 23.635984354628423,
-      "grad_norm": 1.8096040487289429,
-      "learning_rate": 5.496884422110553e-05,
-      "loss": 5.1914,
-      "step": 45322
-    },
-    {
-      "epoch": 23.636505867014343,
-      "grad_norm": 1.4839853048324585,
-      "learning_rate": 5.4967839195979896e-05,
-      "loss": 5.5535,
-      "step": 45323
-    },
-    {
-      "epoch": 23.637027379400262,
-      "grad_norm": 1.5061789751052856,
-      "learning_rate": 5.4966834170854274e-05,
-      "loss": 5.2739,
-      "step": 45324
-    },
-    {
-      "epoch": 23.63754889178618,
-      "grad_norm": 1.6278361082077026,
-      "learning_rate": 5.4965829145728645e-05,
-      "loss": 5.0997,
-      "step": 45325
-    },
-    {
-      "epoch": 23.638070404172097,
-      "grad_norm": 1.5376230478286743,
-      "learning_rate": 5.496482412060302e-05,
-      "loss": 5.1841,
-      "step": 45326
-    },
-    {
-      "epoch": 23.638591916558017,
-      "grad_norm": 1.5682841539382935,
-      "learning_rate": 5.496381909547739e-05,
-      "loss": 5.4658,
-      "step": 45327
-    },
-    {
-      "epoch": 23.639113428943936,
-      "grad_norm": 1.4551482200622559,
-      "learning_rate": 5.4962814070351765e-05,
-      "loss": 5.3131,
-      "step": 45328
-    },
-    {
-      "epoch": 23.639634941329856,
-      "grad_norm": 1.5070785284042358,
-      "learning_rate": 5.496180904522613e-05,
-      "loss": 4.858,
-      "step": 45329
-    },
-    {
-      "epoch": 23.640156453715775,
-      "grad_norm": 1.4977874755859375,
-      "learning_rate": 5.496080402010051e-05,
-      "loss": 4.9099,
-      "step": 45330
-    },
-    {
-      "epoch": 23.640677966101695,
-      "grad_norm": 1.4867523908615112,
-      "learning_rate": 5.495979899497488e-05,
-      "loss": 5.5148,
-      "step": 45331
-    },
-    {
-      "epoch": 23.641199478487614,
-      "grad_norm": 1.4235230684280396,
-      "learning_rate": 5.4958793969849256e-05,
-      "loss": 5.4484,
-      "step": 45332
-    },
-    {
-      "epoch": 23.641720990873534,
-      "grad_norm": 1.65019953250885,
-      "learning_rate": 5.495778894472362e-05,
-      "loss": 5.0744,
-      "step": 45333
-    },
-    {
-      "epoch": 23.642242503259453,
-      "grad_norm": 1.4573196172714233,
-      "learning_rate": 5.4956783919597984e-05,
-      "loss": 5.8144,
-      "step": 45334
-    },
-    {
-      "epoch": 23.642764015645373,
-      "grad_norm": 1.5646789073944092,
-      "learning_rate": 5.495577889447236e-05,
-      "loss": 5.2421,
-      "step": 45335
-    },
-    {
-      "epoch": 23.643285528031292,
-      "grad_norm": 1.6756519079208374,
-      "learning_rate": 5.495477386934673e-05,
-      "loss": 5.4637,
-      "step": 45336
-    },
-    {
-      "epoch": 23.64380704041721,
-      "grad_norm": 1.5998197793960571,
-      "learning_rate": 5.495376884422111e-05,
-      "loss": 5.3628,
-      "step": 45337
-    },
-    {
-      "epoch": 23.644328552803128,
-      "grad_norm": 1.7166519165039062,
-      "learning_rate": 5.4952763819095475e-05,
-      "loss": 4.4805,
-      "step": 45338
-    },
-    {
-      "epoch": 23.644850065189047,
-      "grad_norm": 1.544224500656128,
-      "learning_rate": 5.495175879396985e-05,
-      "loss": 4.9427,
-      "step": 45339
-    },
-    {
-      "epoch": 23.645371577574966,
-      "grad_norm": 1.5469862222671509,
-      "learning_rate": 5.4950753768844224e-05,
-      "loss": 5.0593,
-      "step": 45340
-    },
-    {
-      "epoch": 23.645893089960886,
-      "grad_norm": 1.5188933610916138,
-      "learning_rate": 5.49497487437186e-05,
-      "loss": 5.5464,
-      "step": 45341
-    },
-    {
-      "epoch": 23.646414602346805,
-      "grad_norm": 1.5509384870529175,
-      "learning_rate": 5.4948743718592966e-05,
-      "loss": 4.7517,
-      "step": 45342
-    },
-    {
-      "epoch": 23.646936114732725,
-      "grad_norm": 1.5761544704437256,
-      "learning_rate": 5.4947738693467344e-05,
-      "loss": 5.3398,
-      "step": 45343
-    },
-    {
-      "epoch": 23.647457627118644,
-      "grad_norm": 1.5717899799346924,
-      "learning_rate": 5.494673366834171e-05,
-      "loss": 5.5484,
-      "step": 45344
-    },
-    {
-      "epoch": 23.647979139504564,
-      "grad_norm": 1.6187279224395752,
-      "learning_rate": 5.4945728643216086e-05,
-      "loss": 5.2785,
-      "step": 45345
-    },
-    {
-      "epoch": 23.648500651890483,
-      "grad_norm": 1.5319993495941162,
-      "learning_rate": 5.494472361809046e-05,
-      "loss": 4.7357,
-      "step": 45346
-    },
-    {
-      "epoch": 23.649022164276403,
-      "grad_norm": 1.7056225538253784,
-      "learning_rate": 5.494371859296482e-05,
-      "loss": 5.0886,
-      "step": 45347
-    },
-    {
-      "epoch": 23.649543676662322,
-      "grad_norm": 1.5748392343521118,
-      "learning_rate": 5.49427135678392e-05,
-      "loss": 5.1046,
-      "step": 45348
-    },
-    {
-      "epoch": 23.65006518904824,
-      "grad_norm": 1.5663747787475586,
-      "learning_rate": 5.4941708542713564e-05,
-      "loss": 4.9925,
-      "step": 45349
-    },
-    {
-      "epoch": 23.650586701434158,
-      "grad_norm": 1.730271577835083,
-      "learning_rate": 5.494070351758794e-05,
-      "loss": 4.9597,
-      "step": 45350
-    },
-    {
-      "epoch": 23.651108213820077,
-      "grad_norm": 1.636501669883728,
-      "learning_rate": 5.493969849246231e-05,
-      "loss": 5.3739,
-      "step": 45351
-    },
-    {
-      "epoch": 23.651629726205996,
-      "grad_norm": 1.6183212995529175,
-      "learning_rate": 5.493869346733669e-05,
-      "loss": 5.1531,
-      "step": 45352
-    },
-    {
-      "epoch": 23.652151238591916,
-      "grad_norm": 1.5543594360351562,
-      "learning_rate": 5.4937688442211055e-05,
-      "loss": 5.5124,
-      "step": 45353
-    },
-    {
-      "epoch": 23.652672750977835,
-      "grad_norm": 1.4713082313537598,
-      "learning_rate": 5.493668341708543e-05,
-      "loss": 5.6816,
-      "step": 45354
-    },
-    {
-      "epoch": 23.653194263363755,
-      "grad_norm": 1.432661771774292,
-      "learning_rate": 5.49356783919598e-05,
-      "loss": 5.6227,
-      "step": 45355
-    },
-    {
-      "epoch": 23.653715775749674,
-      "grad_norm": 1.4964239597320557,
-      "learning_rate": 5.4934673366834175e-05,
-      "loss": 5.4864,
-      "step": 45356
-    },
-    {
-      "epoch": 23.654237288135594,
-      "grad_norm": 1.6548165082931519,
-      "learning_rate": 5.4933668341708546e-05,
-      "loss": 4.9732,
-      "step": 45357
-    },
-    {
-      "epoch": 23.654758800521513,
-      "grad_norm": 1.5675036907196045,
-      "learning_rate": 5.4932663316582923e-05,
-      "loss": 5.2138,
-      "step": 45358
-    },
-    {
-      "epoch": 23.655280312907433,
-      "grad_norm": 1.6943219900131226,
-      "learning_rate": 5.493165829145729e-05,
-      "loss": 4.9591,
-      "step": 45359
-    },
-    {
-      "epoch": 23.655801825293352,
-      "grad_norm": 1.5690582990646362,
-      "learning_rate": 5.4930653266331666e-05,
-      "loss": 5.6462,
-      "step": 45360
-    },
-    {
-      "epoch": 23.656323337679268,
-      "grad_norm": 1.67183256149292,
-      "learning_rate": 5.492964824120603e-05,
-      "loss": 5.0037,
-      "step": 45361
-    },
-    {
-      "epoch": 23.656844850065188,
-      "grad_norm": 1.4693117141723633,
-      "learning_rate": 5.49286432160804e-05,
-      "loss": 5.654,
-      "step": 45362
-    },
-    {
-      "epoch": 23.657366362451107,
-      "grad_norm": 1.5728647708892822,
-      "learning_rate": 5.492763819095478e-05,
-      "loss": 4.91,
-      "step": 45363
-    },
-    {
-      "epoch": 23.657887874837026,
-      "grad_norm": 1.5370826721191406,
-      "learning_rate": 5.492663316582914e-05,
-      "loss": 5.1949,
-      "step": 45364
-    },
-    {
-      "epoch": 23.658409387222946,
-      "grad_norm": 1.6270078420639038,
-      "learning_rate": 5.492562814070352e-05,
-      "loss": 4.7613,
-      "step": 45365
-    },
-    {
-      "epoch": 23.658930899608865,
-      "grad_norm": 1.4937224388122559,
-      "learning_rate": 5.492462311557789e-05,
-      "loss": 4.8513,
-      "step": 45366
-    },
-    {
-      "epoch": 23.659452411994785,
-      "grad_norm": 1.5443580150604248,
-      "learning_rate": 5.492361809045227e-05,
-      "loss": 4.3389,
-      "step": 45367
-    },
-    {
-      "epoch": 23.659973924380704,
-      "grad_norm": 1.4476935863494873,
-      "learning_rate": 5.4922613065326634e-05,
-      "loss": 5.473,
-      "step": 45368
-    },
-    {
-      "epoch": 23.660495436766624,
-      "grad_norm": 1.6652225255966187,
-      "learning_rate": 5.492160804020101e-05,
-      "loss": 5.1673,
-      "step": 45369
-    },
-    {
-      "epoch": 23.661016949152543,
-      "grad_norm": 1.5816748142242432,
-      "learning_rate": 5.4920603015075376e-05,
-      "loss": 4.8805,
-      "step": 45370
-    },
-    {
-      "epoch": 23.661538461538463,
-      "grad_norm": 1.3939203023910522,
-      "learning_rate": 5.4919597989949754e-05,
-      "loss": 4.869,
-      "step": 45371
-    },
-    {
-      "epoch": 23.662059973924382,
-      "grad_norm": 1.4879183769226074,
-      "learning_rate": 5.4918592964824125e-05,
-      "loss": 5.3605,
-      "step": 45372
-    },
-    {
-      "epoch": 23.6625814863103,
-      "grad_norm": 1.5105429887771606,
-      "learning_rate": 5.49175879396985e-05,
-      "loss": 5.5926,
-      "step": 45373
-    },
-    {
-      "epoch": 23.663102998696218,
-      "grad_norm": 1.5205044746398926,
-      "learning_rate": 5.491658291457287e-05,
-      "loss": 5.1191,
-      "step": 45374
-    },
-    {
-      "epoch": 23.663624511082137,
-      "grad_norm": 1.5895963907241821,
-      "learning_rate": 5.491557788944723e-05,
-      "loss": 5.2034,
-      "step": 45375
-    },
-    {
-      "epoch": 23.664146023468057,
-      "grad_norm": 1.7297574281692505,
-      "learning_rate": 5.491457286432161e-05,
-      "loss": 4.3387,
-      "step": 45376
-    },
-    {
-      "epoch": 23.664667535853976,
-      "grad_norm": 1.525249719619751,
-      "learning_rate": 5.491356783919598e-05,
-      "loss": 5.5687,
-      "step": 45377
-    },
-    {
-      "epoch": 23.665189048239895,
-      "grad_norm": 1.5461193323135376,
-      "learning_rate": 5.491256281407036e-05,
-      "loss": 5.5893,
-      "step": 45378
-    },
-    {
-      "epoch": 23.665710560625815,
-      "grad_norm": 1.559857726097107,
-      "learning_rate": 5.491155778894472e-05,
-      "loss": 5.55,
-      "step": 45379
-    },
-    {
-      "epoch": 23.666232073011734,
-      "grad_norm": 1.5784025192260742,
-      "learning_rate": 5.49105527638191e-05,
-      "loss": 5.3672,
-      "step": 45380
-    },
-    {
-      "epoch": 23.666753585397654,
-      "grad_norm": 1.5045815706253052,
-      "learning_rate": 5.4909547738693464e-05,
-      "loss": 5.3037,
-      "step": 45381
-    },
-    {
-      "epoch": 23.667275097783573,
-      "grad_norm": 1.5105022192001343,
-      "learning_rate": 5.490854271356784e-05,
-      "loss": 5.4813,
-      "step": 45382
-    },
-    {
-      "epoch": 23.667796610169493,
-      "grad_norm": 1.4519191980361938,
-      "learning_rate": 5.490753768844221e-05,
-      "loss": 5.3916,
-      "step": 45383
-    },
-    {
-      "epoch": 23.668318122555412,
-      "grad_norm": 1.5912586450576782,
-      "learning_rate": 5.490653266331659e-05,
-      "loss": 5.3732,
-      "step": 45384
-    },
-    {
-      "epoch": 23.668839634941328,
-      "grad_norm": 1.4389643669128418,
-      "learning_rate": 5.4905527638190955e-05,
-      "loss": 5.4426,
-      "step": 45385
-    },
-    {
-      "epoch": 23.669361147327248,
-      "grad_norm": 1.6408048868179321,
-      "learning_rate": 5.490452261306533e-05,
-      "loss": 5.0375,
-      "step": 45386
-    },
-    {
-      "epoch": 23.669882659713167,
-      "grad_norm": 1.5693790912628174,
-      "learning_rate": 5.4903517587939704e-05,
-      "loss": 5.0791,
-      "step": 45387
-    },
-    {
-      "epoch": 23.670404172099087,
-      "grad_norm": 1.5060327053070068,
-      "learning_rate": 5.490251256281407e-05,
-      "loss": 5.101,
-      "step": 45388
-    },
-    {
-      "epoch": 23.670925684485006,
-      "grad_norm": 1.512671947479248,
-      "learning_rate": 5.4901507537688446e-05,
-      "loss": 4.958,
-      "step": 45389
-    },
-    {
-      "epoch": 23.671447196870925,
-      "grad_norm": 1.6752002239227295,
-      "learning_rate": 5.490050251256281e-05,
-      "loss": 4.775,
-      "step": 45390
-    },
-    {
-      "epoch": 23.671968709256845,
-      "grad_norm": 1.4634467363357544,
-      "learning_rate": 5.489949748743719e-05,
-      "loss": 5.2737,
-      "step": 45391
-    },
-    {
-      "epoch": 23.672490221642764,
-      "grad_norm": 1.5844084024429321,
-      "learning_rate": 5.489849246231156e-05,
-      "loss": 4.9158,
-      "step": 45392
-    },
-    {
-      "epoch": 23.673011734028684,
-      "grad_norm": 1.5587491989135742,
-      "learning_rate": 5.489748743718594e-05,
-      "loss": 5.4425,
-      "step": 45393
-    },
-    {
-      "epoch": 23.673533246414603,
-      "grad_norm": 1.5607906579971313,
-      "learning_rate": 5.48964824120603e-05,
-      "loss": 5.3201,
-      "step": 45394
-    },
-    {
-      "epoch": 23.674054758800523,
-      "grad_norm": 1.5762008428573608,
-      "learning_rate": 5.489547738693468e-05,
-      "loss": 5.5178,
-      "step": 45395
-    },
-    {
-      "epoch": 23.674576271186442,
-      "grad_norm": 1.585372805595398,
-      "learning_rate": 5.4894472361809044e-05,
-      "loss": 5.1511,
-      "step": 45396
-    },
-    {
-      "epoch": 23.675097783572358,
-      "grad_norm": 1.546237587928772,
-      "learning_rate": 5.489346733668342e-05,
-      "loss": 5.2264,
-      "step": 45397
-    },
-    {
-      "epoch": 23.675619295958278,
-      "grad_norm": 1.5854939222335815,
-      "learning_rate": 5.489246231155779e-05,
-      "loss": 4.8491,
-      "step": 45398
-    },
-    {
-      "epoch": 23.676140808344197,
-      "grad_norm": 1.575784683227539,
-      "learning_rate": 5.489145728643217e-05,
-      "loss": 5.2168,
-      "step": 45399
-    },
-    {
-      "epoch": 23.676662320730117,
-      "grad_norm": 1.493667721748352,
-      "learning_rate": 5.4890452261306535e-05,
-      "loss": 5.4677,
-      "step": 45400
-    },
-    {
-      "epoch": 23.677183833116036,
-      "grad_norm": 1.5314258337020874,
-      "learning_rate": 5.48894472361809e-05,
-      "loss": 5.3516,
-      "step": 45401
-    },
-    {
-      "epoch": 23.677705345501955,
-      "grad_norm": 1.573110580444336,
-      "learning_rate": 5.488844221105528e-05,
-      "loss": 5.2845,
-      "step": 45402
-    },
-    {
-      "epoch": 23.678226857887875,
-      "grad_norm": 1.5564453601837158,
-      "learning_rate": 5.488743718592965e-05,
-      "loss": 5.282,
-      "step": 45403
-    },
-    {
-      "epoch": 23.678748370273794,
-      "grad_norm": 1.6481062173843384,
-      "learning_rate": 5.4886432160804026e-05,
-      "loss": 5.2711,
-      "step": 45404
-    },
-    {
-      "epoch": 23.679269882659714,
-      "grad_norm": 1.561964750289917,
-      "learning_rate": 5.488542713567839e-05,
-      "loss": 5.1987,
-      "step": 45405
-    },
-    {
-      "epoch": 23.679791395045633,
-      "grad_norm": 1.559525489807129,
-      "learning_rate": 5.488442211055277e-05,
-      "loss": 5.3966,
-      "step": 45406
-    },
-    {
-      "epoch": 23.680312907431553,
-      "grad_norm": 1.5913983583450317,
-      "learning_rate": 5.488341708542714e-05,
-      "loss": 5.4745,
-      "step": 45407
-    },
-    {
-      "epoch": 23.680834419817472,
-      "grad_norm": 1.5240859985351562,
-      "learning_rate": 5.488241206030152e-05,
-      "loss": 5.4063,
-      "step": 45408
-    },
-    {
-      "epoch": 23.681355932203388,
-      "grad_norm": 1.517305612564087,
-      "learning_rate": 5.488140703517588e-05,
-      "loss": 5.3938,
-      "step": 45409
-    },
-    {
-      "epoch": 23.681877444589308,
-      "grad_norm": 1.6201436519622803,
-      "learning_rate": 5.488040201005026e-05,
-      "loss": 5.1827,
-      "step": 45410
-    },
-    {
-      "epoch": 23.682398956975227,
-      "grad_norm": 1.5816996097564697,
-      "learning_rate": 5.487939698492462e-05,
-      "loss": 4.7747,
-      "step": 45411
-    },
-    {
-      "epoch": 23.682920469361147,
-      "grad_norm": 1.4409162998199463,
-      "learning_rate": 5.4878391959799e-05,
-      "loss": 4.351,
-      "step": 45412
-    },
-    {
-      "epoch": 23.683441981747066,
-      "grad_norm": 1.5661442279815674,
-      "learning_rate": 5.487738693467337e-05,
-      "loss": 4.9628,
-      "step": 45413
-    },
-    {
-      "epoch": 23.683963494132986,
-      "grad_norm": 1.5423375368118286,
-      "learning_rate": 5.4876381909547736e-05,
-      "loss": 5.4432,
-      "step": 45414
-    },
-    {
-      "epoch": 23.684485006518905,
-      "grad_norm": 1.5334928035736084,
-      "learning_rate": 5.4875376884422114e-05,
-      "loss": 5.2809,
-      "step": 45415
-    },
-    {
-      "epoch": 23.685006518904824,
-      "grad_norm": 1.5452947616577148,
-      "learning_rate": 5.487437185929648e-05,
-      "loss": 5.4066,
-      "step": 45416
-    },
-    {
-      "epoch": 23.685528031290744,
-      "grad_norm": 1.4427522420883179,
-      "learning_rate": 5.4873366834170856e-05,
-      "loss": 5.6173,
-      "step": 45417
-    },
-    {
-      "epoch": 23.686049543676663,
-      "grad_norm": 1.6459593772888184,
-      "learning_rate": 5.487236180904523e-05,
-      "loss": 5.114,
-      "step": 45418
-    },
-    {
-      "epoch": 23.686571056062583,
-      "grad_norm": 1.6027591228485107,
-      "learning_rate": 5.4871356783919605e-05,
-      "loss": 5.3219,
-      "step": 45419
-    },
-    {
-      "epoch": 23.687092568448502,
-      "grad_norm": 1.5153411626815796,
-      "learning_rate": 5.487035175879397e-05,
-      "loss": 4.7753,
-      "step": 45420
-    },
-    {
-      "epoch": 23.687614080834418,
-      "grad_norm": 1.5862008333206177,
-      "learning_rate": 5.486934673366835e-05,
-      "loss": 4.8958,
-      "step": 45421
-    },
-    {
-      "epoch": 23.688135593220338,
-      "grad_norm": 1.5831340551376343,
-      "learning_rate": 5.486834170854271e-05,
-      "loss": 5.1558,
-      "step": 45422
-    },
-    {
-      "epoch": 23.688657105606257,
-      "grad_norm": 1.5397521257400513,
-      "learning_rate": 5.486733668341709e-05,
-      "loss": 5.1662,
-      "step": 45423
-    },
-    {
-      "epoch": 23.689178617992177,
-      "grad_norm": 1.6853231191635132,
-      "learning_rate": 5.486633165829146e-05,
-      "loss": 4.8772,
-      "step": 45424
-    },
-    {
-      "epoch": 23.689700130378096,
-      "grad_norm": 1.5655403137207031,
-      "learning_rate": 5.486532663316584e-05,
-      "loss": 5.305,
-      "step": 45425
-    },
-    {
-      "epoch": 23.690221642764016,
-      "grad_norm": 1.588850498199463,
-      "learning_rate": 5.48643216080402e-05,
-      "loss": 5.2129,
-      "step": 45426
-    },
-    {
-      "epoch": 23.690743155149935,
-      "grad_norm": 1.681321620941162,
-      "learning_rate": 5.4863316582914573e-05,
-      "loss": 4.908,
-      "step": 45427
-    },
-    {
-      "epoch": 23.691264667535854,
-      "grad_norm": 1.4943418502807617,
-      "learning_rate": 5.486231155778895e-05,
-      "loss": 5.6367,
-      "step": 45428
-    },
-    {
-      "epoch": 23.691786179921774,
-      "grad_norm": 1.5288238525390625,
-      "learning_rate": 5.4861306532663316e-05,
-      "loss": 4.7913,
-      "step": 45429
-    },
-    {
-      "epoch": 23.692307692307693,
-      "grad_norm": 1.54961097240448,
-      "learning_rate": 5.486030150753769e-05,
-      "loss": 5.1195,
-      "step": 45430
-    },
-    {
-      "epoch": 23.692829204693613,
-      "grad_norm": 1.470288634300232,
-      "learning_rate": 5.485929648241206e-05,
-      "loss": 5.7391,
-      "step": 45431
-    },
-    {
-      "epoch": 23.693350717079532,
-      "grad_norm": 1.9362585544586182,
-      "learning_rate": 5.4858291457286435e-05,
-      "loss": 5.0007,
-      "step": 45432
-    },
-    {
-      "epoch": 23.69387222946545,
-      "grad_norm": 1.7284375429153442,
-      "learning_rate": 5.4857286432160806e-05,
-      "loss": 5.1278,
-      "step": 45433
-    },
-    {
-      "epoch": 23.694393741851368,
-      "grad_norm": 1.6771650314331055,
-      "learning_rate": 5.4856281407035184e-05,
-      "loss": 4.8117,
-      "step": 45434
-    },
-    {
-      "epoch": 23.694915254237287,
-      "grad_norm": 1.537010908126831,
-      "learning_rate": 5.485527638190955e-05,
-      "loss": 5.3271,
-      "step": 45435
-    },
-    {
-      "epoch": 23.695436766623207,
-      "grad_norm": 1.5736531019210815,
-      "learning_rate": 5.4854271356783926e-05,
-      "loss": 4.7159,
-      "step": 45436
-    },
-    {
-      "epoch": 23.695958279009126,
-      "grad_norm": 1.5919716358184814,
-      "learning_rate": 5.485326633165829e-05,
-      "loss": 5.2741,
-      "step": 45437
-    },
-    {
-      "epoch": 23.696479791395046,
-      "grad_norm": 1.5706818103790283,
-      "learning_rate": 5.485226130653267e-05,
-      "loss": 5.3578,
-      "step": 45438
-    },
-    {
-      "epoch": 23.697001303780965,
-      "grad_norm": 1.6277132034301758,
-      "learning_rate": 5.485125628140704e-05,
-      "loss": 5.1102,
-      "step": 45439
-    },
-    {
-      "epoch": 23.697522816166884,
-      "grad_norm": 1.588730812072754,
-      "learning_rate": 5.4850251256281404e-05,
-      "loss": 5.1082,
-      "step": 45440
-    },
-    {
-      "epoch": 23.698044328552804,
-      "grad_norm": 1.6546802520751953,
-      "learning_rate": 5.484924623115578e-05,
-      "loss": 5.1389,
-      "step": 45441
-    },
-    {
-      "epoch": 23.698565840938723,
-      "grad_norm": 1.4681358337402344,
-      "learning_rate": 5.4848241206030146e-05,
-      "loss": 4.9354,
-      "step": 45442
-    },
-    {
-      "epoch": 23.699087353324643,
-      "grad_norm": 1.7487012147903442,
-      "learning_rate": 5.4847236180904524e-05,
-      "loss": 5.1995,
-      "step": 45443
-    },
-    {
-      "epoch": 23.69960886571056,
-      "grad_norm": 1.5226659774780273,
-      "learning_rate": 5.4846231155778895e-05,
-      "loss": 4.7945,
-      "step": 45444
-    },
-    {
-      "epoch": 23.70013037809648,
-      "grad_norm": 1.52105712890625,
-      "learning_rate": 5.484522613065327e-05,
-      "loss": 5.3399,
-      "step": 45445
-    },
-    {
-      "epoch": 23.700651890482398,
-      "grad_norm": 1.5224363803863525,
-      "learning_rate": 5.484422110552764e-05,
-      "loss": 4.7415,
-      "step": 45446
-    },
-    {
-      "epoch": 23.701173402868317,
-      "grad_norm": 1.4834359884262085,
-      "learning_rate": 5.4843216080402015e-05,
-      "loss": 5.3466,
-      "step": 45447
-    },
-    {
-      "epoch": 23.701694915254237,
-      "grad_norm": 1.655357837677002,
-      "learning_rate": 5.484221105527638e-05,
-      "loss": 4.6303,
-      "step": 45448
-    },
-    {
-      "epoch": 23.702216427640156,
-      "grad_norm": 1.557248830795288,
-      "learning_rate": 5.484120603015076e-05,
-      "loss": 5.2264,
-      "step": 45449
-    },
-    {
-      "epoch": 23.702737940026076,
-      "grad_norm": 1.5627228021621704,
-      "learning_rate": 5.484020100502513e-05,
-      "loss": 5.4834,
-      "step": 45450
-    },
-    {
-      "epoch": 23.703259452411995,
-      "grad_norm": 1.855590581893921,
-      "learning_rate": 5.4839195979899506e-05,
-      "loss": 4.911,
-      "step": 45451
-    },
-    {
-      "epoch": 23.703780964797915,
-      "grad_norm": 1.6151762008666992,
-      "learning_rate": 5.483819095477387e-05,
-      "loss": 5.2705,
-      "step": 45452
-    },
-    {
-      "epoch": 23.704302477183834,
-      "grad_norm": 1.5826822519302368,
-      "learning_rate": 5.483718592964825e-05,
-      "loss": 4.9398,
-      "step": 45453
-    },
-    {
-      "epoch": 23.704823989569753,
-      "grad_norm": 1.6045047044754028,
-      "learning_rate": 5.483618090452262e-05,
-      "loss": 4.8894,
-      "step": 45454
-    },
-    {
-      "epoch": 23.705345501955673,
-      "grad_norm": 1.5597285032272339,
-      "learning_rate": 5.483517587939698e-05,
-      "loss": 5.552,
-      "step": 45455
-    },
-    {
-      "epoch": 23.705867014341592,
-      "grad_norm": 1.6307579278945923,
-      "learning_rate": 5.483417085427136e-05,
-      "loss": 5.2149,
-      "step": 45456
-    },
-    {
-      "epoch": 23.70638852672751,
-      "grad_norm": 1.6405891180038452,
-      "learning_rate": 5.4833165829145725e-05,
-      "loss": 5.451,
-      "step": 45457
-    },
-    {
-      "epoch": 23.706910039113428,
-      "grad_norm": 1.5551711320877075,
-      "learning_rate": 5.48321608040201e-05,
-      "loss": 5.2975,
-      "step": 45458
-    },
-    {
-      "epoch": 23.707431551499347,
-      "grad_norm": 1.5141947269439697,
-      "learning_rate": 5.4831155778894474e-05,
-      "loss": 5.1663,
-      "step": 45459
-    },
-    {
-      "epoch": 23.707953063885267,
-      "grad_norm": 1.542397379875183,
-      "learning_rate": 5.483015075376885e-05,
-      "loss": 4.888,
-      "step": 45460
-    },
-    {
-      "epoch": 23.708474576271186,
-      "grad_norm": 1.5886623859405518,
-      "learning_rate": 5.4829145728643216e-05,
-      "loss": 5.5378,
-      "step": 45461
-    },
-    {
-      "epoch": 23.708996088657106,
-      "grad_norm": 1.6089768409729004,
-      "learning_rate": 5.4828140703517594e-05,
-      "loss": 4.5744,
-      "step": 45462
-    },
-    {
-      "epoch": 23.709517601043025,
-      "grad_norm": 1.6160433292388916,
-      "learning_rate": 5.482713567839196e-05,
-      "loss": 5.2513,
-      "step": 45463
-    },
-    {
-      "epoch": 23.710039113428945,
-      "grad_norm": 1.475787878036499,
-      "learning_rate": 5.4826130653266336e-05,
-      "loss": 5.4489,
-      "step": 45464
-    },
-    {
-      "epoch": 23.710560625814864,
-      "grad_norm": 1.5249608755111694,
-      "learning_rate": 5.482512562814071e-05,
-      "loss": 5.453,
-      "step": 45465
-    },
-    {
-      "epoch": 23.711082138200783,
-      "grad_norm": 1.67307710647583,
-      "learning_rate": 5.4824120603015085e-05,
-      "loss": 5.2043,
-      "step": 45466
-    },
-    {
-      "epoch": 23.711603650586703,
-      "grad_norm": 1.6462039947509766,
-      "learning_rate": 5.482311557788945e-05,
-      "loss": 5.3003,
-      "step": 45467
-    },
-    {
-      "epoch": 23.71212516297262,
-      "grad_norm": 1.5233888626098633,
-      "learning_rate": 5.4822110552763814e-05,
-      "loss": 5.4387,
-      "step": 45468
-    },
-    {
-      "epoch": 23.71264667535854,
-      "grad_norm": 1.471303105354309,
-      "learning_rate": 5.482110552763819e-05,
-      "loss": 4.8486,
-      "step": 45469
-    },
-    {
-      "epoch": 23.713168187744458,
-      "grad_norm": 1.535362720489502,
-      "learning_rate": 5.482010050251256e-05,
-      "loss": 5.4574,
-      "step": 45470
-    },
-    {
-      "epoch": 23.713689700130377,
-      "grad_norm": 1.5225898027420044,
-      "learning_rate": 5.481909547738694e-05,
-      "loss": 5.0967,
-      "step": 45471
-    },
-    {
-      "epoch": 23.714211212516297,
-      "grad_norm": 1.5924068689346313,
-      "learning_rate": 5.4818090452261305e-05,
-      "loss": 4.9581,
-      "step": 45472
-    },
-    {
-      "epoch": 23.714732724902216,
-      "grad_norm": 1.5639668703079224,
-      "learning_rate": 5.481708542713568e-05,
-      "loss": 5.152,
-      "step": 45473
-    },
-    {
-      "epoch": 23.715254237288136,
-      "grad_norm": 1.5610477924346924,
-      "learning_rate": 5.4816080402010053e-05,
-      "loss": 5.6806,
-      "step": 45474
-    },
-    {
-      "epoch": 23.715775749674055,
-      "grad_norm": 1.5582925081253052,
-      "learning_rate": 5.481507537688443e-05,
-      "loss": 5.4608,
-      "step": 45475
-    },
-    {
-      "epoch": 23.716297262059975,
-      "grad_norm": 1.5266163349151611,
-      "learning_rate": 5.4814070351758796e-05,
-      "loss": 5.4823,
-      "step": 45476
-    },
-    {
-      "epoch": 23.716818774445894,
-      "grad_norm": 1.6936920881271362,
-      "learning_rate": 5.4813065326633173e-05,
-      "loss": 5.1694,
-      "step": 45477
-    },
-    {
-      "epoch": 23.717340286831814,
-      "grad_norm": 1.6649174690246582,
-      "learning_rate": 5.481206030150754e-05,
-      "loss": 5.0661,
-      "step": 45478
-    },
-    {
-      "epoch": 23.717861799217733,
-      "grad_norm": 1.5385949611663818,
-      "learning_rate": 5.4811055276381916e-05,
-      "loss": 5.3011,
-      "step": 45479
-    },
-    {
-      "epoch": 23.71838331160365,
-      "grad_norm": 1.5461658239364624,
-      "learning_rate": 5.4810050251256287e-05,
-      "loss": 5.0898,
-      "step": 45480
-    },
-    {
-      "epoch": 23.71890482398957,
-      "grad_norm": 1.5695945024490356,
-      "learning_rate": 5.480904522613065e-05,
-      "loss": 4.8006,
-      "step": 45481
-    },
-    {
-      "epoch": 23.719426336375488,
-      "grad_norm": 1.5563256740570068,
-      "learning_rate": 5.480804020100503e-05,
-      "loss": 5.7115,
-      "step": 45482
-    },
-    {
-      "epoch": 23.719947848761407,
-      "grad_norm": 1.721711277961731,
-      "learning_rate": 5.480703517587939e-05,
-      "loss": 5.0098,
-      "step": 45483
-    },
-    {
-      "epoch": 23.720469361147327,
-      "grad_norm": 1.4925111532211304,
-      "learning_rate": 5.480603015075377e-05,
-      "loss": 5.543,
-      "step": 45484
-    },
-    {
-      "epoch": 23.720990873533246,
-      "grad_norm": 1.6744385957717896,
-      "learning_rate": 5.480502512562814e-05,
-      "loss": 4.4586,
-      "step": 45485
-    },
-    {
-      "epoch": 23.721512385919166,
-      "grad_norm": 1.5619972944259644,
-      "learning_rate": 5.480402010050252e-05,
-      "loss": 5.3269,
-      "step": 45486
-    },
-    {
-      "epoch": 23.722033898305085,
-      "grad_norm": 1.5236601829528809,
-      "learning_rate": 5.4803015075376884e-05,
-      "loss": 5.5697,
-      "step": 45487
-    },
-    {
-      "epoch": 23.722555410691005,
-      "grad_norm": 1.4542951583862305,
-      "learning_rate": 5.480201005025126e-05,
-      "loss": 4.5981,
-      "step": 45488
-    },
-    {
-      "epoch": 23.723076923076924,
-      "grad_norm": 1.4952211380004883,
-      "learning_rate": 5.4801005025125626e-05,
-      "loss": 5.5162,
-      "step": 45489
-    },
-    {
-      "epoch": 23.723598435462844,
-      "grad_norm": 1.6091101169586182,
-      "learning_rate": 5.4800000000000004e-05,
-      "loss": 5.4687,
-      "step": 45490
-    },
-    {
-      "epoch": 23.724119947848763,
-      "grad_norm": 1.7095035314559937,
-      "learning_rate": 5.4798994974874375e-05,
-      "loss": 5.1773,
-      "step": 45491
-    },
-    {
-      "epoch": 23.72464146023468,
-      "grad_norm": 1.5495740175247192,
-      "learning_rate": 5.479798994974875e-05,
-      "loss": 5.3442,
-      "step": 45492
-    },
-    {
-      "epoch": 23.7251629726206,
-      "grad_norm": 1.51512610912323,
-      "learning_rate": 5.479698492462312e-05,
-      "loss": 5.3729,
-      "step": 45493
-    },
-    {
-      "epoch": 23.725684485006518,
-      "grad_norm": 1.5282114744186401,
-      "learning_rate": 5.479597989949749e-05,
-      "loss": 5.5463,
-      "step": 45494
-    },
-    {
-      "epoch": 23.726205997392437,
-      "grad_norm": 1.5310368537902832,
-      "learning_rate": 5.4794974874371866e-05,
-      "loss": 5.4015,
-      "step": 45495
-    },
-    {
-      "epoch": 23.726727509778357,
-      "grad_norm": 1.4125651121139526,
-      "learning_rate": 5.479396984924623e-05,
-      "loss": 5.5265,
-      "step": 45496
-    },
-    {
-      "epoch": 23.727249022164276,
-      "grad_norm": 1.5056511163711548,
-      "learning_rate": 5.479296482412061e-05,
-      "loss": 5.1968,
-      "step": 45497
-    },
-    {
-      "epoch": 23.727770534550196,
-      "grad_norm": 1.4995546340942383,
-      "learning_rate": 5.479195979899497e-05,
-      "loss": 5.0752,
-      "step": 45498
-    },
-    {
-      "epoch": 23.728292046936115,
-      "grad_norm": 1.5345330238342285,
-      "learning_rate": 5.479095477386935e-05,
-      "loss": 5.1321,
-      "step": 45499
-    },
-    {
-      "epoch": 23.728813559322035,
-      "grad_norm": 1.6317113637924194,
-      "learning_rate": 5.478994974874372e-05,
-      "loss": 4.4865,
-      "step": 45500
-    },
-    {
-      "epoch": 23.729335071707954,
-      "grad_norm": 1.471787691116333,
-      "learning_rate": 5.47889447236181e-05,
-      "loss": 4.4581,
-      "step": 45501
-    },
-    {
-      "epoch": 23.729856584093874,
-      "grad_norm": 1.5323214530944824,
-      "learning_rate": 5.478793969849246e-05,
-      "loss": 5.2919,
-      "step": 45502
-    },
-    {
-      "epoch": 23.730378096479793,
-      "grad_norm": 1.5690193176269531,
-      "learning_rate": 5.478693467336684e-05,
-      "loss": 5.0818,
-      "step": 45503
-    },
-    {
-      "epoch": 23.73089960886571,
-      "grad_norm": 1.6435353755950928,
-      "learning_rate": 5.4785929648241205e-05,
-      "loss": 4.9514,
-      "step": 45504
-    },
-    {
-      "epoch": 23.73142112125163,
-      "grad_norm": 1.5845906734466553,
-      "learning_rate": 5.478492462311558e-05,
-      "loss": 5.1475,
-      "step": 45505
-    },
-    {
-      "epoch": 23.731942633637548,
-      "grad_norm": 1.4370909929275513,
-      "learning_rate": 5.4783919597989954e-05,
-      "loss": 5.6316,
-      "step": 45506
-    },
-    {
-      "epoch": 23.732464146023467,
-      "grad_norm": 1.515535593032837,
-      "learning_rate": 5.478291457286432e-05,
-      "loss": 5.3143,
-      "step": 45507
-    },
-    {
-      "epoch": 23.732985658409387,
-      "grad_norm": 1.5976325273513794,
-      "learning_rate": 5.4781909547738696e-05,
-      "loss": 4.8167,
-      "step": 45508
-    },
-    {
-      "epoch": 23.733507170795306,
-      "grad_norm": 1.4917724132537842,
-      "learning_rate": 5.478090452261306e-05,
-      "loss": 5.2286,
-      "step": 45509
-    },
-    {
-      "epoch": 23.734028683181226,
-      "grad_norm": 1.5505422353744507,
-      "learning_rate": 5.477989949748744e-05,
-      "loss": 4.9157,
-      "step": 45510
-    },
-    {
-      "epoch": 23.734550195567145,
-      "grad_norm": 1.5433779954910278,
-      "learning_rate": 5.477889447236181e-05,
-      "loss": 5.599,
-      "step": 45511
-    },
-    {
-      "epoch": 23.735071707953065,
-      "grad_norm": 1.5939396619796753,
-      "learning_rate": 5.477788944723619e-05,
-      "loss": 5.3695,
-      "step": 45512
-    },
-    {
-      "epoch": 23.735593220338984,
-      "grad_norm": 1.619171142578125,
-      "learning_rate": 5.477688442211055e-05,
-      "loss": 4.9266,
-      "step": 45513
-    },
-    {
-      "epoch": 23.736114732724904,
-      "grad_norm": 1.67328679561615,
-      "learning_rate": 5.477587939698493e-05,
-      "loss": 4.8544,
-      "step": 45514
-    },
-    {
-      "epoch": 23.736636245110823,
-      "grad_norm": 1.4480946063995361,
-      "learning_rate": 5.47748743718593e-05,
-      "loss": 5.7374,
-      "step": 45515
-    },
-    {
-      "epoch": 23.73715775749674,
-      "grad_norm": 1.5192031860351562,
-      "learning_rate": 5.477386934673368e-05,
-      "loss": 5.6697,
-      "step": 45516
-    },
-    {
-      "epoch": 23.73767926988266,
-      "grad_norm": 1.531114101409912,
-      "learning_rate": 5.477286432160804e-05,
-      "loss": 4.9454,
-      "step": 45517
-    },
-    {
-      "epoch": 23.738200782268578,
-      "grad_norm": 1.5330220460891724,
-      "learning_rate": 5.477185929648242e-05,
-      "loss": 5.4402,
-      "step": 45518
-    },
-    {
-      "epoch": 23.738722294654497,
-      "grad_norm": 1.4450418949127197,
-      "learning_rate": 5.4770854271356785e-05,
-      "loss": 5.4586,
-      "step": 45519
-    },
-    {
-      "epoch": 23.739243807040417,
-      "grad_norm": 1.5722280740737915,
-      "learning_rate": 5.4769849246231156e-05,
-      "loss": 5.0428,
-      "step": 45520
-    },
-    {
-      "epoch": 23.739765319426336,
-      "grad_norm": 1.5784227848052979,
-      "learning_rate": 5.4768844221105534e-05,
-      "loss": 4.9772,
-      "step": 45521
-    },
-    {
-      "epoch": 23.740286831812256,
-      "grad_norm": 1.465787649154663,
-      "learning_rate": 5.47678391959799e-05,
-      "loss": 5.3156,
-      "step": 45522
-    },
-    {
-      "epoch": 23.740808344198175,
-      "grad_norm": 1.4775776863098145,
-      "learning_rate": 5.4766834170854276e-05,
-      "loss": 5.7627,
-      "step": 45523
-    },
-    {
-      "epoch": 23.741329856584095,
-      "grad_norm": 1.5245797634124756,
-      "learning_rate": 5.476582914572864e-05,
-      "loss": 5.0894,
-      "step": 45524
-    },
-    {
-      "epoch": 23.741851368970014,
-      "grad_norm": 1.5634026527404785,
-      "learning_rate": 5.476482412060302e-05,
-      "loss": 5.0069,
-      "step": 45525
-    },
-    {
-      "epoch": 23.742372881355934,
-      "grad_norm": 1.5170888900756836,
-      "learning_rate": 5.476381909547739e-05,
-      "loss": 5.4412,
-      "step": 45526
-    },
-    {
-      "epoch": 23.742894393741853,
-      "grad_norm": 1.6286404132843018,
-      "learning_rate": 5.476281407035177e-05,
-      "loss": 5.2049,
-      "step": 45527
-    },
-    {
-      "epoch": 23.74341590612777,
-      "grad_norm": 1.634924292564392,
-      "learning_rate": 5.476180904522613e-05,
-      "loss": 5.1179,
-      "step": 45528
-    },
-    {
-      "epoch": 23.74393741851369,
-      "grad_norm": 1.6193797588348389,
-      "learning_rate": 5.476080402010051e-05,
-      "loss": 5.3293,
-      "step": 45529
-    },
-    {
-      "epoch": 23.744458930899608,
-      "grad_norm": 1.631040096282959,
-      "learning_rate": 5.475979899497487e-05,
-      "loss": 4.5779,
-      "step": 45530
-    },
-    {
-      "epoch": 23.744980443285527,
-      "grad_norm": 1.5833927392959595,
-      "learning_rate": 5.475879396984925e-05,
-      "loss": 4.867,
-      "step": 45531
-    },
-    {
-      "epoch": 23.745501955671447,
-      "grad_norm": 1.459072470664978,
-      "learning_rate": 5.475778894472362e-05,
-      "loss": 5.2162,
-      "step": 45532
-    },
-    {
-      "epoch": 23.746023468057366,
-      "grad_norm": 1.5365707874298096,
-      "learning_rate": 5.4756783919597986e-05,
-      "loss": 5.2657,
-      "step": 45533
-    },
-    {
-      "epoch": 23.746544980443286,
-      "grad_norm": 1.5581051111221313,
-      "learning_rate": 5.4755778894472364e-05,
-      "loss": 5.3734,
-      "step": 45534
-    },
-    {
-      "epoch": 23.747066492829205,
-      "grad_norm": 1.4761043787002563,
-      "learning_rate": 5.475477386934673e-05,
-      "loss": 5.4168,
-      "step": 45535
-    },
-    {
-      "epoch": 23.747588005215125,
-      "grad_norm": 1.5213805437088013,
-      "learning_rate": 5.4753768844221106e-05,
-      "loss": 5.2026,
-      "step": 45536
-    },
-    {
-      "epoch": 23.748109517601044,
-      "grad_norm": 1.5675749778747559,
-      "learning_rate": 5.475276381909548e-05,
-      "loss": 5.2758,
-      "step": 45537
-    },
-    {
-      "epoch": 23.748631029986964,
-      "grad_norm": 1.5312714576721191,
-      "learning_rate": 5.4751758793969855e-05,
-      "loss": 4.8544,
-      "step": 45538
-    },
-    {
-      "epoch": 23.749152542372883,
-      "grad_norm": 1.5608341693878174,
-      "learning_rate": 5.475075376884422e-05,
-      "loss": 5.5526,
-      "step": 45539
-    },
-    {
-      "epoch": 23.7496740547588,
-      "grad_norm": 1.5707173347473145,
-      "learning_rate": 5.47497487437186e-05,
-      "loss": 5.0062,
-      "step": 45540
-    },
-    {
-      "epoch": 23.75019556714472,
-      "grad_norm": 1.5798749923706055,
-      "learning_rate": 5.474874371859297e-05,
-      "loss": 5.3482,
-      "step": 45541
-    },
-    {
-      "epoch": 23.750717079530638,
-      "grad_norm": 1.6384940147399902,
-      "learning_rate": 5.4747738693467346e-05,
-      "loss": 4.9713,
-      "step": 45542
-    },
-    {
-      "epoch": 23.751238591916557,
-      "grad_norm": 1.6084184646606445,
-      "learning_rate": 5.474673366834171e-05,
-      "loss": 5.5291,
-      "step": 45543
-    },
-    {
-      "epoch": 23.751760104302477,
-      "grad_norm": 1.5509833097457886,
-      "learning_rate": 5.474572864321609e-05,
-      "loss": 5.0936,
-      "step": 45544
-    },
-    {
-      "epoch": 23.752281616688396,
-      "grad_norm": 1.5234421491622925,
-      "learning_rate": 5.474472361809045e-05,
-      "loss": 4.8785,
-      "step": 45545
-    },
-    {
-      "epoch": 23.752803129074316,
-      "grad_norm": 1.5684385299682617,
-      "learning_rate": 5.474371859296483e-05,
-      "loss": 5.442,
-      "step": 45546
-    },
-    {
-      "epoch": 23.753324641460235,
-      "grad_norm": 1.5926090478897095,
-      "learning_rate": 5.47427135678392e-05,
-      "loss": 4.8603,
-      "step": 45547
-    },
-    {
-      "epoch": 23.753846153846155,
-      "grad_norm": 1.6869227886199951,
-      "learning_rate": 5.4741708542713566e-05,
-      "loss": 4.8772,
-      "step": 45548
-    },
-    {
-      "epoch": 23.754367666232074,
-      "grad_norm": 1.5697087049484253,
-      "learning_rate": 5.474070351758794e-05,
-      "loss": 4.9237,
-      "step": 45549
-    },
-    {
-      "epoch": 23.754889178617994,
-      "grad_norm": 1.5976415872573853,
-      "learning_rate": 5.473969849246231e-05,
-      "loss": 5.4077,
-      "step": 45550
-    },
-    {
-      "epoch": 23.75541069100391,
-      "grad_norm": 1.5596749782562256,
-      "learning_rate": 5.4738693467336685e-05,
-      "loss": 5.501,
-      "step": 45551
-    },
-    {
-      "epoch": 23.75593220338983,
-      "grad_norm": 1.537361741065979,
-      "learning_rate": 5.4737688442211057e-05,
-      "loss": 5.2404,
-      "step": 45552
-    },
-    {
-      "epoch": 23.75645371577575,
-      "grad_norm": 1.53136146068573,
-      "learning_rate": 5.4736683417085434e-05,
-      "loss": 5.2777,
-      "step": 45553
-    },
-    {
-      "epoch": 23.756975228161668,
-      "grad_norm": 1.6369484663009644,
-      "learning_rate": 5.47356783919598e-05,
-      "loss": 5.2479,
-      "step": 45554
-    },
-    {
-      "epoch": 23.757496740547587,
-      "grad_norm": 1.6579148769378662,
-      "learning_rate": 5.4734673366834176e-05,
-      "loss": 5.2407,
-      "step": 45555
-    },
-    {
-      "epoch": 23.758018252933507,
-      "grad_norm": 1.6201000213623047,
-      "learning_rate": 5.473366834170854e-05,
-      "loss": 5.4951,
-      "step": 45556
-    },
-    {
-      "epoch": 23.758539765319426,
-      "grad_norm": 1.5092310905456543,
-      "learning_rate": 5.473266331658292e-05,
-      "loss": 5.5003,
-      "step": 45557
-    },
-    {
-      "epoch": 23.759061277705346,
-      "grad_norm": 1.520350694656372,
-      "learning_rate": 5.473165829145729e-05,
-      "loss": 5.1771,
-      "step": 45558
-    },
-    {
-      "epoch": 23.759582790091265,
-      "grad_norm": 1.580060362815857,
-      "learning_rate": 5.473065326633167e-05,
-      "loss": 5.0116,
-      "step": 45559
-    },
-    {
-      "epoch": 23.760104302477185,
-      "grad_norm": 1.5320247411727905,
-      "learning_rate": 5.472964824120603e-05,
-      "loss": 5.1429,
-      "step": 45560
-    },
-    {
-      "epoch": 23.760625814863104,
-      "grad_norm": 1.60117506980896,
-      "learning_rate": 5.47286432160804e-05,
-      "loss": 5.0238,
-      "step": 45561
-    },
-    {
-      "epoch": 23.761147327249024,
-      "grad_norm": 1.4952518939971924,
-      "learning_rate": 5.472763819095478e-05,
-      "loss": 5.4559,
-      "step": 45562
-    },
-    {
-      "epoch": 23.761668839634943,
-      "grad_norm": 1.5346426963806152,
-      "learning_rate": 5.4726633165829145e-05,
-      "loss": 5.5317,
-      "step": 45563
-    },
-    {
-      "epoch": 23.76219035202086,
-      "grad_norm": 1.591434359550476,
-      "learning_rate": 5.472562814070352e-05,
-      "loss": 5.2117,
-      "step": 45564
-    },
-    {
-      "epoch": 23.76271186440678,
-      "grad_norm": 1.5169031620025635,
-      "learning_rate": 5.472462311557789e-05,
-      "loss": 5.1653,
-      "step": 45565
-    },
-    {
-      "epoch": 23.763233376792698,
-      "grad_norm": 1.6451448202133179,
-      "learning_rate": 5.4723618090452265e-05,
-      "loss": 5.3259,
-      "step": 45566
-    },
-    {
-      "epoch": 23.763754889178617,
-      "grad_norm": 1.5538430213928223,
-      "learning_rate": 5.4722613065326636e-05,
-      "loss": 5.5687,
-      "step": 45567
-    },
-    {
-      "epoch": 23.764276401564537,
-      "grad_norm": 1.5899289846420288,
-      "learning_rate": 5.4721608040201014e-05,
-      "loss": 5.3915,
-      "step": 45568
-    },
-    {
-      "epoch": 23.764797913950456,
-      "grad_norm": 1.5633107423782349,
-      "learning_rate": 5.472060301507538e-05,
-      "loss": 5.1921,
-      "step": 45569
-    },
-    {
-      "epoch": 23.765319426336376,
-      "grad_norm": 1.5991624593734741,
-      "learning_rate": 5.4719597989949756e-05,
-      "loss": 5.5313,
-      "step": 45570
-    },
-    {
-      "epoch": 23.765840938722295,
-      "grad_norm": 1.6889333724975586,
-      "learning_rate": 5.471859296482412e-05,
-      "loss": 5.1803,
-      "step": 45571
-    },
-    {
-      "epoch": 23.766362451108215,
-      "grad_norm": 1.661473035812378,
-      "learning_rate": 5.47175879396985e-05,
-      "loss": 4.9393,
-      "step": 45572
-    },
-    {
-      "epoch": 23.766883963494134,
-      "grad_norm": 1.5111528635025024,
-      "learning_rate": 5.471658291457287e-05,
-      "loss": 5.6494,
-      "step": 45573
-    },
-    {
-      "epoch": 23.767405475880054,
-      "grad_norm": 1.5687949657440186,
-      "learning_rate": 5.471557788944723e-05,
-      "loss": 5.0417,
-      "step": 45574
-    },
-    {
-      "epoch": 23.76792698826597,
-      "grad_norm": 1.5879193544387817,
-      "learning_rate": 5.471457286432161e-05,
-      "loss": 4.5277,
-      "step": 45575
-    },
-    {
-      "epoch": 23.76844850065189,
-      "grad_norm": 1.6603033542633057,
-      "learning_rate": 5.4713567839195975e-05,
-      "loss": 5.1101,
-      "step": 45576
-    },
-    {
-      "epoch": 23.76897001303781,
-      "grad_norm": 1.4515095949172974,
-      "learning_rate": 5.471256281407035e-05,
-      "loss": 5.3845,
-      "step": 45577
-    },
-    {
-      "epoch": 23.769491525423728,
-      "grad_norm": 1.6455930471420288,
-      "learning_rate": 5.4711557788944724e-05,
-      "loss": 5.2771,
-      "step": 45578
-    },
-    {
-      "epoch": 23.770013037809647,
-      "grad_norm": 1.5000885725021362,
-      "learning_rate": 5.47105527638191e-05,
-      "loss": 5.6286,
-      "step": 45579
-    },
-    {
-      "epoch": 23.770534550195567,
-      "grad_norm": 1.471977710723877,
-      "learning_rate": 5.4709547738693466e-05,
-      "loss": 5.4358,
-      "step": 45580
-    },
-    {
-      "epoch": 23.771056062581486,
-      "grad_norm": 1.4891124963760376,
-      "learning_rate": 5.4708542713567844e-05,
-      "loss": 5.4437,
-      "step": 45581
-    },
-    {
-      "epoch": 23.771577574967406,
-      "grad_norm": 1.5085984468460083,
-      "learning_rate": 5.4707537688442215e-05,
-      "loss": 5.5172,
-      "step": 45582
-    },
-    {
-      "epoch": 23.772099087353325,
-      "grad_norm": 1.5924235582351685,
-      "learning_rate": 5.470653266331659e-05,
-      "loss": 5.3704,
-      "step": 45583
-    },
-    {
-      "epoch": 23.772620599739245,
-      "grad_norm": 1.5920051336288452,
-      "learning_rate": 5.470552763819096e-05,
-      "loss": 5.1687,
-      "step": 45584
-    },
-    {
-      "epoch": 23.773142112125164,
-      "grad_norm": 1.5796808004379272,
-      "learning_rate": 5.4704522613065335e-05,
-      "loss": 5.5073,
-      "step": 45585
-    },
-    {
-      "epoch": 23.773663624511084,
-      "grad_norm": 1.583308458328247,
-      "learning_rate": 5.47035175879397e-05,
-      "loss": 5.4615,
-      "step": 45586
-    },
-    {
-      "epoch": 23.774185136897,
-      "grad_norm": 1.4598954916000366,
-      "learning_rate": 5.470251256281407e-05,
-      "loss": 5.2191,
-      "step": 45587
-    },
-    {
-      "epoch": 23.77470664928292,
-      "grad_norm": 1.5341423749923706,
-      "learning_rate": 5.470150753768845e-05,
-      "loss": 5.1383,
-      "step": 45588
-    },
-    {
-      "epoch": 23.77522816166884,
-      "grad_norm": 1.5312010049819946,
-      "learning_rate": 5.470050251256281e-05,
-      "loss": 5.5275,
-      "step": 45589
-    },
-    {
-      "epoch": 23.775749674054758,
-      "grad_norm": 1.5399258136749268,
-      "learning_rate": 5.469949748743719e-05,
-      "loss": 5.0674,
-      "step": 45590
-    },
-    {
-      "epoch": 23.776271186440677,
-      "grad_norm": 1.561847448348999,
-      "learning_rate": 5.4698492462311555e-05,
-      "loss": 5.0787,
-      "step": 45591
-    },
-    {
-      "epoch": 23.776792698826597,
-      "grad_norm": 1.6160967350006104,
-      "learning_rate": 5.469748743718593e-05,
-      "loss": 5.1464,
-      "step": 45592
-    },
-    {
-      "epoch": 23.777314211212516,
-      "grad_norm": 1.6874266862869263,
-      "learning_rate": 5.4696482412060304e-05,
-      "loss": 4.8269,
-      "step": 45593
-    },
-    {
-      "epoch": 23.777835723598436,
-      "grad_norm": 1.64320707321167,
-      "learning_rate": 5.469547738693468e-05,
-      "loss": 5.3372,
-      "step": 45594
-    },
-    {
-      "epoch": 23.778357235984355,
-      "grad_norm": 1.5179800987243652,
-      "learning_rate": 5.4694472361809046e-05,
-      "loss": 5.0057,
-      "step": 45595
-    },
-    {
-      "epoch": 23.778878748370275,
-      "grad_norm": 1.5386121273040771,
-      "learning_rate": 5.4693467336683423e-05,
-      "loss": 5.4595,
-      "step": 45596
-    },
-    {
-      "epoch": 23.779400260756194,
-      "grad_norm": 1.5363490581512451,
-      "learning_rate": 5.469246231155779e-05,
-      "loss": 5.23,
-      "step": 45597
-    },
-    {
-      "epoch": 23.779921773142114,
-      "grad_norm": 1.564741849899292,
-      "learning_rate": 5.4691457286432166e-05,
-      "loss": 5.337,
-      "step": 45598
-    },
-    {
-      "epoch": 23.78044328552803,
-      "grad_norm": 1.522445797920227,
-      "learning_rate": 5.4690452261306537e-05,
-      "loss": 4.8393,
-      "step": 45599
-    },
-    {
-      "epoch": 23.78096479791395,
-      "grad_norm": 1.7139244079589844,
-      "learning_rate": 5.46894472361809e-05,
-      "loss": 5.0846,
-      "step": 45600
-    },
-    {
-      "epoch": 23.78148631029987,
-      "grad_norm": 1.5847437381744385,
-      "learning_rate": 5.468844221105528e-05,
-      "loss": 5.1546,
-      "step": 45601
-    },
-    {
-      "epoch": 23.782007822685788,
-      "grad_norm": 1.4729623794555664,
-      "learning_rate": 5.468743718592965e-05,
-      "loss": 5.3434,
-      "step": 45602
-    },
-    {
-      "epoch": 23.782529335071708,
-      "grad_norm": 1.5097603797912598,
-      "learning_rate": 5.468643216080403e-05,
-      "loss": 5.3511,
-      "step": 45603
-    },
-    {
-      "epoch": 23.783050847457627,
-      "grad_norm": 1.4371808767318726,
-      "learning_rate": 5.468542713567839e-05,
-      "loss": 5.2706,
-      "step": 45604
-    },
-    {
-      "epoch": 23.783572359843546,
-      "grad_norm": 1.5548120737075806,
-      "learning_rate": 5.468442211055277e-05,
-      "loss": 5.1314,
-      "step": 45605
-    },
-    {
-      "epoch": 23.784093872229466,
-      "grad_norm": 1.6140737533569336,
-      "learning_rate": 5.4683417085427134e-05,
-      "loss": 4.4126,
-      "step": 45606
-    },
-    {
-      "epoch": 23.784615384615385,
-      "grad_norm": 1.5003042221069336,
-      "learning_rate": 5.468241206030151e-05,
-      "loss": 5.1528,
-      "step": 45607
-    },
-    {
-      "epoch": 23.785136897001305,
-      "grad_norm": 1.5096466541290283,
-      "learning_rate": 5.468140703517588e-05,
-      "loss": 5.626,
-      "step": 45608
-    },
-    {
-      "epoch": 23.785658409387224,
-      "grad_norm": 1.5922995805740356,
-      "learning_rate": 5.468040201005026e-05,
-      "loss": 5.0398,
-      "step": 45609
-    },
-    {
-      "epoch": 23.786179921773144,
-      "grad_norm": 1.5243654251098633,
-      "learning_rate": 5.4679396984924625e-05,
-      "loss": 5.2806,
-      "step": 45610
-    },
-    {
-      "epoch": 23.78670143415906,
-      "grad_norm": 1.5414623022079468,
-      "learning_rate": 5.4678391959799e-05,
-      "loss": 5.2246,
-      "step": 45611
-    },
-    {
-      "epoch": 23.78722294654498,
-      "grad_norm": 1.5485609769821167,
-      "learning_rate": 5.467738693467337e-05,
-      "loss": 5.0763,
-      "step": 45612
-    },
-    {
-      "epoch": 23.7877444589309,
-      "grad_norm": 1.4935978651046753,
-      "learning_rate": 5.467638190954774e-05,
-      "loss": 5.3177,
-      "step": 45613
-    },
-    {
-      "epoch": 23.788265971316818,
-      "grad_norm": 1.5225372314453125,
-      "learning_rate": 5.4675376884422116e-05,
-      "loss": 5.1687,
-      "step": 45614
-    },
-    {
-      "epoch": 23.788787483702738,
-      "grad_norm": 1.5105504989624023,
-      "learning_rate": 5.467437185929648e-05,
-      "loss": 5.3346,
-      "step": 45615
-    },
-    {
-      "epoch": 23.789308996088657,
-      "grad_norm": 1.578044056892395,
-      "learning_rate": 5.467336683417086e-05,
-      "loss": 5.3649,
-      "step": 45616
-    },
-    {
-      "epoch": 23.789830508474576,
-      "grad_norm": 1.566383719444275,
-      "learning_rate": 5.467236180904522e-05,
-      "loss": 5.363,
-      "step": 45617
-    },
-    {
-      "epoch": 23.790352020860496,
-      "grad_norm": 1.5391912460327148,
-      "learning_rate": 5.46713567839196e-05,
-      "loss": 5.4415,
-      "step": 45618
-    },
-    {
-      "epoch": 23.790873533246415,
-      "grad_norm": 1.4659645557403564,
-      "learning_rate": 5.467035175879397e-05,
-      "loss": 5.426,
-      "step": 45619
-    },
-    {
-      "epoch": 23.791395045632335,
-      "grad_norm": 1.5128706693649292,
-      "learning_rate": 5.466934673366835e-05,
-      "loss": 4.6104,
-      "step": 45620
-    },
-    {
-      "epoch": 23.791916558018254,
-      "grad_norm": 1.4967721700668335,
-      "learning_rate": 5.466834170854271e-05,
-      "loss": 5.2518,
-      "step": 45621
-    },
-    {
-      "epoch": 23.792438070404174,
-      "grad_norm": 1.5066938400268555,
-      "learning_rate": 5.466733668341709e-05,
-      "loss": 5.5306,
-      "step": 45622
-    },
-    {
-      "epoch": 23.79295958279009,
-      "grad_norm": 1.6101067066192627,
-      "learning_rate": 5.4666331658291455e-05,
-      "loss": 4.4855,
-      "step": 45623
-    },
-    {
-      "epoch": 23.79348109517601,
-      "grad_norm": 1.513290524482727,
-      "learning_rate": 5.466532663316583e-05,
-      "loss": 5.6409,
-      "step": 45624
-    },
-    {
-      "epoch": 23.79400260756193,
-      "grad_norm": 1.5240391492843628,
-      "learning_rate": 5.4664321608040204e-05,
-      "loss": 5.3027,
-      "step": 45625
-    },
-    {
-      "epoch": 23.794524119947848,
-      "grad_norm": 1.6967500448226929,
-      "learning_rate": 5.466331658291458e-05,
-      "loss": 5.3196,
-      "step": 45626
-    },
-    {
-      "epoch": 23.795045632333768,
-      "grad_norm": 1.4742120504379272,
-      "learning_rate": 5.4662311557788946e-05,
-      "loss": 5.4384,
-      "step": 45627
-    },
-    {
-      "epoch": 23.795567144719687,
-      "grad_norm": 1.6050699949264526,
-      "learning_rate": 5.466130653266332e-05,
-      "loss": 5.3646,
-      "step": 45628
-    },
-    {
-      "epoch": 23.796088657105607,
-      "grad_norm": 1.526998519897461,
-      "learning_rate": 5.4660301507537695e-05,
-      "loss": 5.3179,
-      "step": 45629
-    },
-    {
-      "epoch": 23.796610169491526,
-      "grad_norm": 1.6623457670211792,
-      "learning_rate": 5.465929648241206e-05,
-      "loss": 5.2424,
-      "step": 45630
-    },
-    {
-      "epoch": 23.797131681877445,
-      "grad_norm": 1.6069891452789307,
-      "learning_rate": 5.465829145728644e-05,
-      "loss": 5.4848,
-      "step": 45631
-    },
-    {
-      "epoch": 23.797653194263365,
-      "grad_norm": 1.5887254476547241,
-      "learning_rate": 5.46572864321608e-05,
-      "loss": 5.6769,
-      "step": 45632
-    },
-    {
-      "epoch": 23.798174706649284,
-      "grad_norm": 1.463175892829895,
-      "learning_rate": 5.465628140703518e-05,
-      "loss": 5.4114,
-      "step": 45633
-    },
-    {
-      "epoch": 23.7986962190352,
-      "grad_norm": 1.4851170778274536,
-      "learning_rate": 5.465527638190955e-05,
-      "loss": 5.2941,
-      "step": 45634
-    },
-    {
-      "epoch": 23.79921773142112,
-      "grad_norm": 1.5653049945831299,
-      "learning_rate": 5.465427135678393e-05,
-      "loss": 5.2004,
-      "step": 45635
-    },
-    {
-      "epoch": 23.79973924380704,
-      "grad_norm": 1.502803921699524,
-      "learning_rate": 5.465326633165829e-05,
-      "loss": 5.1226,
-      "step": 45636
-    },
-    {
-      "epoch": 23.80026075619296,
-      "grad_norm": 1.555456280708313,
-      "learning_rate": 5.465226130653267e-05,
-      "loss": 5.6344,
-      "step": 45637
-    },
-    {
-      "epoch": 23.800782268578878,
-      "grad_norm": 1.564927577972412,
-      "learning_rate": 5.4651256281407035e-05,
-      "loss": 4.7929,
-      "step": 45638
-    },
-    {
-      "epoch": 23.801303780964798,
-      "grad_norm": 1.5598373413085938,
-      "learning_rate": 5.465025125628141e-05,
-      "loss": 4.9837,
-      "step": 45639
-    },
-    {
-      "epoch": 23.801825293350717,
-      "grad_norm": 1.5646661520004272,
-      "learning_rate": 5.4649246231155784e-05,
-      "loss": 5.3377,
-      "step": 45640
-    },
-    {
-      "epoch": 23.802346805736637,
-      "grad_norm": 1.4906476736068726,
-      "learning_rate": 5.464824120603015e-05,
-      "loss": 5.4458,
-      "step": 45641
-    },
-    {
-      "epoch": 23.802868318122556,
-      "grad_norm": 1.566502571105957,
-      "learning_rate": 5.4647236180904526e-05,
-      "loss": 5.2145,
-      "step": 45642
-    },
-    {
-      "epoch": 23.803389830508475,
-      "grad_norm": 1.5425204038619995,
-      "learning_rate": 5.464623115577889e-05,
-      "loss": 5.169,
-      "step": 45643
-    },
-    {
-      "epoch": 23.803911342894395,
-      "grad_norm": 1.5663312673568726,
-      "learning_rate": 5.464522613065327e-05,
-      "loss": 5.2089,
-      "step": 45644
-    },
-    {
-      "epoch": 23.804432855280314,
-      "grad_norm": 1.6487261056900024,
-      "learning_rate": 5.464422110552764e-05,
-      "loss": 5.1783,
-      "step": 45645
-    },
-    {
-      "epoch": 23.804954367666234,
-      "grad_norm": 1.7251083850860596,
-      "learning_rate": 5.464321608040202e-05,
-      "loss": 5.4514,
-      "step": 45646
-    },
-    {
-      "epoch": 23.80547588005215,
-      "grad_norm": 1.53765070438385,
-      "learning_rate": 5.464221105527638e-05,
-      "loss": 5.0057,
-      "step": 45647
-    },
-    {
-      "epoch": 23.80599739243807,
-      "grad_norm": 1.654558777809143,
-      "learning_rate": 5.464120603015076e-05,
-      "loss": 5.5086,
-      "step": 45648
-    },
-    {
-      "epoch": 23.80651890482399,
-      "grad_norm": 1.5266557931900024,
-      "learning_rate": 5.464020100502513e-05,
-      "loss": 5.03,
-      "step": 45649
-    },
-    {
-      "epoch": 23.807040417209908,
-      "grad_norm": 1.5755503177642822,
-      "learning_rate": 5.463919597989951e-05,
-      "loss": 5.2452,
-      "step": 45650
-    },
-    {
-      "epoch": 23.807561929595828,
-      "grad_norm": 1.5978163480758667,
-      "learning_rate": 5.463819095477387e-05,
-      "loss": 5.3655,
-      "step": 45651
-    },
-    {
-      "epoch": 23.808083441981747,
-      "grad_norm": 1.6048599481582642,
-      "learning_rate": 5.463718592964825e-05,
-      "loss": 4.978,
-      "step": 45652
-    },
-    {
-      "epoch": 23.808604954367667,
-      "grad_norm": 1.467402696609497,
-      "learning_rate": 5.4636180904522614e-05,
-      "loss": 5.325,
-      "step": 45653
-    },
-    {
-      "epoch": 23.809126466753586,
-      "grad_norm": 1.495947241783142,
-      "learning_rate": 5.4635175879396985e-05,
-      "loss": 5.4659,
-      "step": 45654
-    },
-    {
-      "epoch": 23.809647979139505,
-      "grad_norm": 1.57086181640625,
-      "learning_rate": 5.463417085427136e-05,
-      "loss": 5.1865,
-      "step": 45655
-    },
-    {
-      "epoch": 23.810169491525425,
-      "grad_norm": 1.5319677591323853,
-      "learning_rate": 5.463316582914573e-05,
-      "loss": 5.5961,
-      "step": 45656
-    },
-    {
-      "epoch": 23.810691003911344,
-      "grad_norm": 1.6074886322021484,
-      "learning_rate": 5.4632160804020105e-05,
-      "loss": 5.2843,
-      "step": 45657
-    },
-    {
-      "epoch": 23.81121251629726,
-      "grad_norm": 1.5399596691131592,
-      "learning_rate": 5.463115577889447e-05,
-      "loss": 4.7537,
-      "step": 45658
-    },
-    {
-      "epoch": 23.81173402868318,
-      "grad_norm": 1.5703710317611694,
-      "learning_rate": 5.463015075376885e-05,
-      "loss": 5.2689,
-      "step": 45659
-    },
-    {
-      "epoch": 23.8122555410691,
-      "grad_norm": 1.573734164237976,
-      "learning_rate": 5.462914572864322e-05,
-      "loss": 5.2427,
-      "step": 45660
-    },
-    {
-      "epoch": 23.81277705345502,
-      "grad_norm": 1.5512616634368896,
-      "learning_rate": 5.4628140703517596e-05,
-      "loss": 5.505,
-      "step": 45661
-    },
-    {
-      "epoch": 23.813298565840938,
-      "grad_norm": 1.673316240310669,
-      "learning_rate": 5.462713567839196e-05,
-      "loss": 5.3981,
-      "step": 45662
-    },
-    {
-      "epoch": 23.813820078226858,
-      "grad_norm": 1.6535769701004028,
-      "learning_rate": 5.462613065326634e-05,
-      "loss": 5.2198,
-      "step": 45663
-    },
-    {
-      "epoch": 23.814341590612777,
-      "grad_norm": 1.7144695520401,
-      "learning_rate": 5.46251256281407e-05,
-      "loss": 5.4355,
-      "step": 45664
-    },
-    {
-      "epoch": 23.814863102998697,
-      "grad_norm": 1.5820033550262451,
-      "learning_rate": 5.462412060301508e-05,
-      "loss": 5.0149,
-      "step": 45665
-    },
-    {
-      "epoch": 23.815384615384616,
-      "grad_norm": 1.4936332702636719,
-      "learning_rate": 5.462311557788945e-05,
-      "loss": 5.1354,
-      "step": 45666
-    },
-    {
-      "epoch": 23.815906127770536,
-      "grad_norm": 1.5168439149856567,
-      "learning_rate": 5.4622110552763816e-05,
-      "loss": 5.3773,
-      "step": 45667
-    },
-    {
-      "epoch": 23.816427640156455,
-      "grad_norm": 1.6429424285888672,
-      "learning_rate": 5.462110552763819e-05,
-      "loss": 5.0809,
-      "step": 45668
-    },
-    {
-      "epoch": 23.816949152542374,
-      "grad_norm": 2.0559117794036865,
-      "learning_rate": 5.4620100502512564e-05,
-      "loss": 4.6248,
-      "step": 45669
-    },
-    {
-      "epoch": 23.81747066492829,
-      "grad_norm": 1.6072907447814941,
-      "learning_rate": 5.461909547738694e-05,
-      "loss": 4.6996,
-      "step": 45670
-    },
-    {
-      "epoch": 23.81799217731421,
-      "grad_norm": 1.5074412822723389,
-      "learning_rate": 5.4618090452261307e-05,
-      "loss": 5.1692,
-      "step": 45671
-    },
-    {
-      "epoch": 23.81851368970013,
-      "grad_norm": 1.5723017454147339,
-      "learning_rate": 5.4617085427135684e-05,
-      "loss": 5.4292,
-      "step": 45672
-    },
-    {
-      "epoch": 23.81903520208605,
-      "grad_norm": 1.530968427658081,
-      "learning_rate": 5.461608040201005e-05,
-      "loss": 5.3773,
-      "step": 45673
-    },
-    {
-      "epoch": 23.819556714471968,
-      "grad_norm": 1.6167012453079224,
-      "learning_rate": 5.4615075376884426e-05,
-      "loss": 5.3367,
-      "step": 45674
-    },
-    {
-      "epoch": 23.820078226857888,
-      "grad_norm": 1.6550828218460083,
-      "learning_rate": 5.46140703517588e-05,
-      "loss": 5.2219,
-      "step": 45675
-    },
-    {
-      "epoch": 23.820599739243807,
-      "grad_norm": 1.5960091352462769,
-      "learning_rate": 5.4613065326633175e-05,
-      "loss": 4.8715,
-      "step": 45676
-    },
-    {
-      "epoch": 23.821121251629727,
-      "grad_norm": 1.5195292234420776,
-      "learning_rate": 5.461206030150754e-05,
-      "loss": 5.294,
-      "step": 45677
-    },
-    {
-      "epoch": 23.821642764015646,
-      "grad_norm": 1.543068289756775,
-      "learning_rate": 5.461105527638192e-05,
-      "loss": 5.0974,
-      "step": 45678
-    },
-    {
-      "epoch": 23.822164276401566,
-      "grad_norm": 1.5892126560211182,
-      "learning_rate": 5.461005025125628e-05,
-      "loss": 4.8178,
-      "step": 45679
-    },
-    {
-      "epoch": 23.822685788787485,
-      "grad_norm": 1.594329595565796,
-      "learning_rate": 5.460904522613065e-05,
-      "loss": 5.1244,
-      "step": 45680
-    },
-    {
-      "epoch": 23.823207301173404,
-      "grad_norm": 1.6278249025344849,
-      "learning_rate": 5.460804020100503e-05,
-      "loss": 5.0369,
-      "step": 45681
-    },
-    {
-      "epoch": 23.82372881355932,
-      "grad_norm": 1.6407253742218018,
-      "learning_rate": 5.4607035175879395e-05,
-      "loss": 5.4542,
-      "step": 45682
-    },
-    {
-      "epoch": 23.82425032594524,
-      "grad_norm": 1.4998034238815308,
-      "learning_rate": 5.460603015075377e-05,
-      "loss": 4.4948,
-      "step": 45683
-    },
-    {
-      "epoch": 23.82477183833116,
-      "grad_norm": 1.5681369304656982,
-      "learning_rate": 5.460502512562814e-05,
-      "loss": 5.03,
-      "step": 45684
-    },
-    {
-      "epoch": 23.82529335071708,
-      "grad_norm": 1.566685676574707,
-      "learning_rate": 5.4604020100502515e-05,
-      "loss": 5.4447,
-      "step": 45685
-    },
-    {
-      "epoch": 23.825814863103,
-      "grad_norm": 1.536383867263794,
-      "learning_rate": 5.4603015075376886e-05,
-      "loss": 5.0774,
-      "step": 45686
-    },
-    {
-      "epoch": 23.826336375488918,
-      "grad_norm": 1.5938314199447632,
-      "learning_rate": 5.4602010050251264e-05,
-      "loss": 5.4869,
-      "step": 45687
-    },
-    {
-      "epoch": 23.826857887874837,
-      "grad_norm": 1.5806763172149658,
-      "learning_rate": 5.460100502512563e-05,
-      "loss": 5.3257,
-      "step": 45688
-    },
-    {
-      "epoch": 23.827379400260757,
-      "grad_norm": 1.4978464841842651,
-      "learning_rate": 5.4600000000000006e-05,
-      "loss": 5.0467,
-      "step": 45689
-    },
-    {
-      "epoch": 23.827900912646676,
-      "grad_norm": 1.521727442741394,
-      "learning_rate": 5.459899497487437e-05,
-      "loss": 5.4827,
-      "step": 45690
-    },
-    {
-      "epoch": 23.828422425032596,
-      "grad_norm": 1.478752851486206,
-      "learning_rate": 5.459798994974875e-05,
-      "loss": 4.9442,
-      "step": 45691
-    },
-    {
-      "epoch": 23.828943937418515,
-      "grad_norm": 1.5004795789718628,
-      "learning_rate": 5.459698492462312e-05,
-      "loss": 5.3536,
-      "step": 45692
-    },
-    {
-      "epoch": 23.829465449804434,
-      "grad_norm": 1.4914175271987915,
-      "learning_rate": 5.459597989949748e-05,
-      "loss": 5.4205,
-      "step": 45693
-    },
-    {
-      "epoch": 23.82998696219035,
-      "grad_norm": 1.5124778747558594,
-      "learning_rate": 5.459497487437186e-05,
-      "loss": 5.4302,
-      "step": 45694
-    },
-    {
-      "epoch": 23.83050847457627,
-      "grad_norm": 1.5605061054229736,
-      "learning_rate": 5.459396984924623e-05,
-      "loss": 5.3654,
-      "step": 45695
-    },
-    {
-      "epoch": 23.83102998696219,
-      "grad_norm": 1.5202621221542358,
-      "learning_rate": 5.459296482412061e-05,
-      "loss": 5.2558,
-      "step": 45696
-    },
-    {
-      "epoch": 23.83155149934811,
-      "grad_norm": 1.5348337888717651,
-      "learning_rate": 5.4591959798994974e-05,
-      "loss": 5.3585,
-      "step": 45697
-    },
-    {
-      "epoch": 23.83207301173403,
-      "grad_norm": 1.4834576845169067,
-      "learning_rate": 5.459095477386935e-05,
-      "loss": 5.7349,
-      "step": 45698
-    },
-    {
-      "epoch": 23.832594524119948,
-      "grad_norm": 1.663902759552002,
-      "learning_rate": 5.4589949748743716e-05,
-      "loss": 4.9678,
-      "step": 45699
-    },
-    {
-      "epoch": 23.833116036505867,
-      "grad_norm": 1.4950532913208008,
-      "learning_rate": 5.4588944723618094e-05,
-      "loss": 5.215,
-      "step": 45700
-    },
-    {
-      "epoch": 23.833637548891787,
-      "grad_norm": 1.5009773969650269,
-      "learning_rate": 5.4587939698492465e-05,
-      "loss": 5.5561,
-      "step": 45701
-    },
-    {
-      "epoch": 23.834159061277706,
-      "grad_norm": 1.5879104137420654,
-      "learning_rate": 5.458693467336684e-05,
-      "loss": 5.0261,
-      "step": 45702
-    },
-    {
-      "epoch": 23.834680573663626,
-      "grad_norm": 1.5260707139968872,
-      "learning_rate": 5.458592964824121e-05,
-      "loss": 5.1565,
-      "step": 45703
-    },
-    {
-      "epoch": 23.835202086049545,
-      "grad_norm": 1.5117911100387573,
-      "learning_rate": 5.4584924623115585e-05,
-      "loss": 5.14,
-      "step": 45704
-    },
-    {
-      "epoch": 23.835723598435465,
-      "grad_norm": 1.5996063947677612,
-      "learning_rate": 5.458391959798995e-05,
-      "loss": 5.0497,
-      "step": 45705
-    },
-    {
-      "epoch": 23.83624511082138,
-      "grad_norm": 1.6874775886535645,
-      "learning_rate": 5.458291457286432e-05,
-      "loss": 4.8702,
-      "step": 45706
-    },
-    {
-      "epoch": 23.8367666232073,
-      "grad_norm": 1.5974851846694946,
-      "learning_rate": 5.45819095477387e-05,
-      "loss": 5.5712,
-      "step": 45707
-    },
-    {
-      "epoch": 23.83728813559322,
-      "grad_norm": 1.5320372581481934,
-      "learning_rate": 5.458090452261306e-05,
-      "loss": 5.0822,
-      "step": 45708
-    },
-    {
-      "epoch": 23.83780964797914,
-      "grad_norm": 1.510243535041809,
-      "learning_rate": 5.457989949748744e-05,
-      "loss": 5.4158,
-      "step": 45709
-    },
-    {
-      "epoch": 23.83833116036506,
-      "grad_norm": 1.4690251350402832,
-      "learning_rate": 5.4578894472361805e-05,
-      "loss": 5.1816,
-      "step": 45710
-    },
-    {
-      "epoch": 23.838852672750978,
-      "grad_norm": 1.6103930473327637,
-      "learning_rate": 5.457788944723618e-05,
-      "loss": 5.0808,
-      "step": 45711
-    },
-    {
-      "epoch": 23.839374185136897,
-      "grad_norm": 1.5905827283859253,
-      "learning_rate": 5.4576884422110554e-05,
-      "loss": 5.3192,
-      "step": 45712
-    },
-    {
-      "epoch": 23.839895697522817,
-      "grad_norm": 1.5647701025009155,
-      "learning_rate": 5.457587939698493e-05,
-      "loss": 5.3459,
-      "step": 45713
-    },
-    {
-      "epoch": 23.840417209908736,
-      "grad_norm": 1.488312005996704,
-      "learning_rate": 5.4574874371859296e-05,
-      "loss": 5.1826,
-      "step": 45714
-    },
-    {
-      "epoch": 23.840938722294656,
-      "grad_norm": 1.6420440673828125,
-      "learning_rate": 5.4573869346733673e-05,
-      "loss": 5.2485,
-      "step": 45715
-    },
-    {
-      "epoch": 23.841460234680575,
-      "grad_norm": 1.7368983030319214,
-      "learning_rate": 5.4572864321608045e-05,
-      "loss": 4.6659,
-      "step": 45716
-    },
-    {
-      "epoch": 23.841981747066495,
-      "grad_norm": 1.5079989433288574,
-      "learning_rate": 5.457185929648242e-05,
-      "loss": 5.4024,
-      "step": 45717
-    },
-    {
-      "epoch": 23.84250325945241,
-      "grad_norm": 1.5000303983688354,
-      "learning_rate": 5.457085427135679e-05,
-      "loss": 5.3031,
-      "step": 45718
-    },
-    {
-      "epoch": 23.84302477183833,
-      "grad_norm": 1.5231084823608398,
-      "learning_rate": 5.4569849246231164e-05,
-      "loss": 5.1191,
-      "step": 45719
-    },
-    {
-      "epoch": 23.84354628422425,
-      "grad_norm": 1.5415656566619873,
-      "learning_rate": 5.456884422110553e-05,
-      "loss": 5.1228,
-      "step": 45720
-    },
-    {
-      "epoch": 23.84406779661017,
-      "grad_norm": 1.5478147268295288,
-      "learning_rate": 5.45678391959799e-05,
-      "loss": 5.3524,
-      "step": 45721
-    },
-    {
-      "epoch": 23.84458930899609,
-      "grad_norm": 1.5636478662490845,
-      "learning_rate": 5.456683417085428e-05,
-      "loss": 5.2231,
-      "step": 45722
-    },
-    {
-      "epoch": 23.845110821382008,
-      "grad_norm": 1.559726595878601,
-      "learning_rate": 5.456582914572864e-05,
-      "loss": 5.0369,
-      "step": 45723
-    },
-    {
-      "epoch": 23.845632333767927,
-      "grad_norm": 1.5596660375595093,
-      "learning_rate": 5.456482412060302e-05,
-      "loss": 5.2007,
-      "step": 45724
-    },
-    {
-      "epoch": 23.846153846153847,
-      "grad_norm": 1.5686088800430298,
-      "learning_rate": 5.4563819095477384e-05,
-      "loss": 5.1237,
-      "step": 45725
-    },
-    {
-      "epoch": 23.846675358539766,
-      "grad_norm": 1.5370724201202393,
-      "learning_rate": 5.456281407035176e-05,
-      "loss": 5.3505,
-      "step": 45726
-    },
-    {
-      "epoch": 23.847196870925686,
-      "grad_norm": 1.808107614517212,
-      "learning_rate": 5.456180904522613e-05,
-      "loss": 5.0098,
-      "step": 45727
-    },
-    {
-      "epoch": 23.847718383311605,
-      "grad_norm": 1.6387404203414917,
-      "learning_rate": 5.456080402010051e-05,
-      "loss": 5.1394,
-      "step": 45728
-    },
-    {
-      "epoch": 23.848239895697525,
-      "grad_norm": 1.5786752700805664,
-      "learning_rate": 5.4559798994974875e-05,
-      "loss": 4.8787,
-      "step": 45729
-    },
-    {
-      "epoch": 23.84876140808344,
-      "grad_norm": 1.5562694072723389,
-      "learning_rate": 5.455879396984925e-05,
-      "loss": 5.3608,
-      "step": 45730
-    },
-    {
-      "epoch": 23.84928292046936,
-      "grad_norm": 1.571564793586731,
-      "learning_rate": 5.455778894472362e-05,
-      "loss": 5.3569,
-      "step": 45731
-    },
-    {
-      "epoch": 23.84980443285528,
-      "grad_norm": 1.517909288406372,
-      "learning_rate": 5.4556783919597995e-05,
-      "loss": 5.233,
-      "step": 45732
-    },
-    {
-      "epoch": 23.8503259452412,
-      "grad_norm": 1.5479347705841064,
-      "learning_rate": 5.4555778894472366e-05,
-      "loss": 5.1617,
-      "step": 45733
-    },
-    {
-      "epoch": 23.85084745762712,
-      "grad_norm": 1.5677481889724731,
-      "learning_rate": 5.455477386934673e-05,
-      "loss": 4.386,
-      "step": 45734
-    },
-    {
-      "epoch": 23.851368970013038,
-      "grad_norm": 1.4351811408996582,
-      "learning_rate": 5.455376884422111e-05,
-      "loss": 4.9599,
-      "step": 45735
-    },
-    {
-      "epoch": 23.851890482398957,
-      "grad_norm": 1.6870310306549072,
-      "learning_rate": 5.455276381909548e-05,
-      "loss": 5.3797,
-      "step": 45736
-    },
-    {
-      "epoch": 23.852411994784877,
-      "grad_norm": 1.5158500671386719,
-      "learning_rate": 5.455175879396986e-05,
-      "loss": 4.9093,
-      "step": 45737
-    },
-    {
-      "epoch": 23.852933507170796,
-      "grad_norm": 1.5210378170013428,
-      "learning_rate": 5.455075376884422e-05,
-      "loss": 5.2741,
-      "step": 45738
-    },
-    {
-      "epoch": 23.853455019556716,
-      "grad_norm": 1.5630955696105957,
-      "learning_rate": 5.45497487437186e-05,
-      "loss": 4.6626,
-      "step": 45739
-    },
-    {
-      "epoch": 23.853976531942635,
-      "grad_norm": 1.6099202632904053,
-      "learning_rate": 5.454874371859296e-05,
-      "loss": 5.2313,
-      "step": 45740
-    },
-    {
-      "epoch": 23.85449804432855,
-      "grad_norm": 1.541640043258667,
-      "learning_rate": 5.454773869346734e-05,
-      "loss": 5.5699,
-      "step": 45741
-    },
-    {
-      "epoch": 23.85501955671447,
-      "grad_norm": 1.5464141368865967,
-      "learning_rate": 5.454673366834171e-05,
-      "loss": 4.3741,
-      "step": 45742
-    },
-    {
-      "epoch": 23.85554106910039,
-      "grad_norm": 1.5519157648086548,
-      "learning_rate": 5.454572864321609e-05,
-      "loss": 4.9094,
-      "step": 45743
-    },
-    {
-      "epoch": 23.85606258148631,
-      "grad_norm": 1.5708643198013306,
-      "learning_rate": 5.4544723618090454e-05,
-      "loss": 5.0217,
-      "step": 45744
-    },
-    {
-      "epoch": 23.85658409387223,
-      "grad_norm": 1.517678141593933,
-      "learning_rate": 5.454371859296483e-05,
-      "loss": 5.6667,
-      "step": 45745
-    },
-    {
-      "epoch": 23.85710560625815,
-      "grad_norm": 1.5696351528167725,
-      "learning_rate": 5.4542713567839196e-05,
-      "loss": 4.7892,
-      "step": 45746
-    },
-    {
-      "epoch": 23.857627118644068,
-      "grad_norm": 1.5545626878738403,
-      "learning_rate": 5.454170854271357e-05,
-      "loss": 5.1106,
-      "step": 45747
-    },
-    {
-      "epoch": 23.858148631029987,
-      "grad_norm": 1.3935390710830688,
-      "learning_rate": 5.4540703517587945e-05,
-      "loss": 5.6974,
-      "step": 45748
-    },
-    {
-      "epoch": 23.858670143415907,
-      "grad_norm": 1.4122824668884277,
-      "learning_rate": 5.453969849246231e-05,
-      "loss": 5.413,
-      "step": 45749
-    },
-    {
-      "epoch": 23.859191655801826,
-      "grad_norm": 1.5244327783584595,
-      "learning_rate": 5.453869346733669e-05,
-      "loss": 5.6904,
-      "step": 45750
-    },
-    {
-      "epoch": 23.859713168187746,
-      "grad_norm": 1.5274535417556763,
-      "learning_rate": 5.453768844221105e-05,
-      "loss": 5.0018,
-      "step": 45751
-    },
-    {
-      "epoch": 23.860234680573665,
-      "grad_norm": 1.535589575767517,
-      "learning_rate": 5.453668341708543e-05,
-      "loss": 4.7691,
-      "step": 45752
-    },
-    {
-      "epoch": 23.860756192959585,
-      "grad_norm": 1.4804866313934326,
-      "learning_rate": 5.45356783919598e-05,
-      "loss": 5.453,
-      "step": 45753
-    },
-    {
-      "epoch": 23.8612777053455,
-      "grad_norm": 1.4854577779769897,
-      "learning_rate": 5.453467336683418e-05,
-      "loss": 5.4004,
-      "step": 45754
-    },
-    {
-      "epoch": 23.86179921773142,
-      "grad_norm": 1.5883245468139648,
-      "learning_rate": 5.453366834170854e-05,
-      "loss": 5.6003,
-      "step": 45755
-    },
-    {
-      "epoch": 23.86232073011734,
-      "grad_norm": 1.4717642068862915,
-      "learning_rate": 5.453266331658292e-05,
-      "loss": 5.5304,
-      "step": 45756
-    },
-    {
-      "epoch": 23.86284224250326,
-      "grad_norm": 1.5408867597579956,
-      "learning_rate": 5.453165829145729e-05,
-      "loss": 5.6797,
-      "step": 45757
-    },
-    {
-      "epoch": 23.86336375488918,
-      "grad_norm": 1.5372098684310913,
-      "learning_rate": 5.453065326633167e-05,
-      "loss": 5.5028,
-      "step": 45758
-    },
-    {
-      "epoch": 23.863885267275098,
-      "grad_norm": 1.5802035331726074,
-      "learning_rate": 5.4529648241206034e-05,
-      "loss": 5.249,
-      "step": 45759
-    },
-    {
-      "epoch": 23.864406779661017,
-      "grad_norm": 1.6118299961090088,
-      "learning_rate": 5.45286432160804e-05,
-      "loss": 5.283,
-      "step": 45760
-    },
-    {
-      "epoch": 23.864928292046937,
-      "grad_norm": 1.4433091878890991,
-      "learning_rate": 5.4527638190954776e-05,
-      "loss": 5.4359,
-      "step": 45761
-    },
-    {
-      "epoch": 23.865449804432856,
-      "grad_norm": 1.5282405614852905,
-      "learning_rate": 5.452663316582915e-05,
-      "loss": 4.7196,
-      "step": 45762
-    },
-    {
-      "epoch": 23.865971316818776,
-      "grad_norm": 1.6439448595046997,
-      "learning_rate": 5.4525628140703525e-05,
-      "loss": 5.1954,
-      "step": 45763
-    },
-    {
-      "epoch": 23.866492829204695,
-      "grad_norm": 1.5624064207077026,
-      "learning_rate": 5.452462311557789e-05,
-      "loss": 5.1915,
-      "step": 45764
-    },
-    {
-      "epoch": 23.86701434159061,
-      "grad_norm": 1.5416712760925293,
-      "learning_rate": 5.452361809045227e-05,
-      "loss": 4.7749,
-      "step": 45765
-    },
-    {
-      "epoch": 23.86753585397653,
-      "grad_norm": 1.4940491914749146,
-      "learning_rate": 5.452261306532663e-05,
-      "loss": 5.2536,
-      "step": 45766
-    },
-    {
-      "epoch": 23.86805736636245,
-      "grad_norm": 1.5270013809204102,
-      "learning_rate": 5.452160804020101e-05,
-      "loss": 5.591,
-      "step": 45767
-    },
-    {
-      "epoch": 23.86857887874837,
-      "grad_norm": 1.53765070438385,
-      "learning_rate": 5.452060301507538e-05,
-      "loss": 5.3056,
-      "step": 45768
-    },
-    {
-      "epoch": 23.86910039113429,
-      "grad_norm": 1.4766361713409424,
-      "learning_rate": 5.451959798994976e-05,
-      "loss": 4.7719,
-      "step": 45769
-    },
-    {
-      "epoch": 23.86962190352021,
-      "grad_norm": 1.6917356252670288,
-      "learning_rate": 5.451859296482412e-05,
-      "loss": 4.4614,
-      "step": 45770
-    },
-    {
-      "epoch": 23.870143415906128,
-      "grad_norm": 1.5858254432678223,
-      "learning_rate": 5.45175879396985e-05,
-      "loss": 5.4034,
-      "step": 45771
-    },
-    {
-      "epoch": 23.870664928292047,
-      "grad_norm": 1.6354995965957642,
-      "learning_rate": 5.4516582914572864e-05,
-      "loss": 5.2476,
-      "step": 45772
-    },
-    {
-      "epoch": 23.871186440677967,
-      "grad_norm": 1.6223936080932617,
-      "learning_rate": 5.4515577889447235e-05,
-      "loss": 5.1587,
-      "step": 45773
-    },
-    {
-      "epoch": 23.871707953063886,
-      "grad_norm": 1.551016092300415,
-      "learning_rate": 5.451457286432161e-05,
-      "loss": 5.4821,
-      "step": 45774
-    },
-    {
-      "epoch": 23.872229465449806,
-      "grad_norm": 1.6104533672332764,
-      "learning_rate": 5.451356783919598e-05,
-      "loss": 4.8969,
-      "step": 45775
-    },
-    {
-      "epoch": 23.872750977835725,
-      "grad_norm": 1.5861245393753052,
-      "learning_rate": 5.4512562814070355e-05,
-      "loss": 4.5855,
-      "step": 45776
-    },
-    {
-      "epoch": 23.87327249022164,
-      "grad_norm": 1.6187744140625,
-      "learning_rate": 5.451155778894472e-05,
-      "loss": 4.8712,
-      "step": 45777
-    },
-    {
-      "epoch": 23.87379400260756,
-      "grad_norm": 1.448129415512085,
-      "learning_rate": 5.45105527638191e-05,
-      "loss": 5.8031,
-      "step": 45778
-    },
-    {
-      "epoch": 23.87431551499348,
-      "grad_norm": 1.6637630462646484,
-      "learning_rate": 5.450954773869347e-05,
-      "loss": 5.1826,
-      "step": 45779
-    },
-    {
-      "epoch": 23.8748370273794,
-      "grad_norm": 1.6522270441055298,
-      "learning_rate": 5.4508542713567846e-05,
-      "loss": 4.9361,
-      "step": 45780
-    },
-    {
-      "epoch": 23.87535853976532,
-      "grad_norm": 1.5531680583953857,
-      "learning_rate": 5.450753768844221e-05,
-      "loss": 5.3039,
-      "step": 45781
-    },
-    {
-      "epoch": 23.87588005215124,
-      "grad_norm": 1.481773853302002,
-      "learning_rate": 5.450653266331659e-05,
-      "loss": 5.2235,
-      "step": 45782
-    },
-    {
-      "epoch": 23.876401564537158,
-      "grad_norm": 1.5604844093322754,
-      "learning_rate": 5.450552763819096e-05,
-      "loss": 4.8617,
-      "step": 45783
-    },
-    {
-      "epoch": 23.876923076923077,
-      "grad_norm": 1.5179202556610107,
-      "learning_rate": 5.450452261306534e-05,
-      "loss": 5.0294,
-      "step": 45784
-    },
-    {
-      "epoch": 23.877444589308997,
-      "grad_norm": 1.4755017757415771,
-      "learning_rate": 5.45035175879397e-05,
-      "loss": 5.2671,
-      "step": 45785
-    },
-    {
-      "epoch": 23.877966101694916,
-      "grad_norm": 1.504150629043579,
-      "learning_rate": 5.4502512562814066e-05,
-      "loss": 5.3906,
-      "step": 45786
-    },
-    {
-      "epoch": 23.878487614080836,
-      "grad_norm": 1.4744025468826294,
-      "learning_rate": 5.4501507537688443e-05,
-      "loss": 5.1405,
-      "step": 45787
-    },
-    {
-      "epoch": 23.879009126466755,
-      "grad_norm": 1.6154359579086304,
-      "learning_rate": 5.4500502512562814e-05,
-      "loss": 4.7258,
-      "step": 45788
-    },
-    {
-      "epoch": 23.87953063885267,
-      "grad_norm": 1.512521743774414,
-      "learning_rate": 5.449949748743719e-05,
-      "loss": 5.4261,
-      "step": 45789
-    },
-    {
-      "epoch": 23.88005215123859,
-      "grad_norm": 1.576574683189392,
-      "learning_rate": 5.4498492462311557e-05,
-      "loss": 5.1451,
-      "step": 45790
-    },
-    {
-      "epoch": 23.88057366362451,
-      "grad_norm": 1.3990269899368286,
-      "learning_rate": 5.4497487437185934e-05,
-      "loss": 5.0301,
-      "step": 45791
-    },
-    {
-      "epoch": 23.88109517601043,
-      "grad_norm": 1.4903937578201294,
-      "learning_rate": 5.44964824120603e-05,
-      "loss": 4.9783,
-      "step": 45792
-    },
-    {
-      "epoch": 23.88161668839635,
-      "grad_norm": 1.4282925128936768,
-      "learning_rate": 5.4495477386934676e-05,
-      "loss": 5.7536,
-      "step": 45793
-    },
-    {
-      "epoch": 23.88213820078227,
-      "grad_norm": 1.5289151668548584,
-      "learning_rate": 5.449447236180905e-05,
-      "loss": 5.3803,
-      "step": 45794
-    },
-    {
-      "epoch": 23.882659713168188,
-      "grad_norm": 1.4598572254180908,
-      "learning_rate": 5.4493467336683425e-05,
-      "loss": 5.6005,
-      "step": 45795
-    },
-    {
-      "epoch": 23.883181225554107,
-      "grad_norm": 1.5788817405700684,
-      "learning_rate": 5.449246231155779e-05,
-      "loss": 5.1587,
-      "step": 45796
-    },
-    {
-      "epoch": 23.883702737940027,
-      "grad_norm": 1.5729764699935913,
-      "learning_rate": 5.449145728643217e-05,
-      "loss": 5.1848,
-      "step": 45797
-    },
-    {
-      "epoch": 23.884224250325946,
-      "grad_norm": 1.5400643348693848,
-      "learning_rate": 5.449045226130653e-05,
-      "loss": 5.5227,
-      "step": 45798
-    },
-    {
-      "epoch": 23.884745762711866,
-      "grad_norm": 1.490918517112732,
-      "learning_rate": 5.44894472361809e-05,
-      "loss": 5.4188,
-      "step": 45799
-    },
-    {
-      "epoch": 23.885267275097785,
-      "grad_norm": 1.5947996377944946,
-      "learning_rate": 5.448844221105528e-05,
-      "loss": 5.3103,
-      "step": 45800
-    },
-    {
-      "epoch": 23.8857887874837,
-      "grad_norm": 1.5015416145324707,
-      "learning_rate": 5.4487437185929645e-05,
-      "loss": 5.5447,
-      "step": 45801
-    },
-    {
-      "epoch": 23.88631029986962,
-      "grad_norm": 1.51035475730896,
-      "learning_rate": 5.448643216080402e-05,
-      "loss": 5.4498,
-      "step": 45802
-    },
-    {
-      "epoch": 23.88683181225554,
-      "grad_norm": 1.5182982683181763,
-      "learning_rate": 5.4485427135678394e-05,
-      "loss": 5.1971,
-      "step": 45803
-    },
-    {
-      "epoch": 23.88735332464146,
-      "grad_norm": 1.5741297006607056,
-      "learning_rate": 5.448442211055277e-05,
-      "loss": 5.2922,
-      "step": 45804
-    },
-    {
-      "epoch": 23.88787483702738,
-      "grad_norm": 1.5525071620941162,
-      "learning_rate": 5.4483417085427136e-05,
-      "loss": 5.4244,
-      "step": 45805
-    },
-    {
-      "epoch": 23.8883963494133,
-      "grad_norm": 1.5326189994812012,
-      "learning_rate": 5.4482412060301514e-05,
-      "loss": 4.7822,
-      "step": 45806
-    },
-    {
-      "epoch": 23.888917861799218,
-      "grad_norm": 1.6522706747055054,
-      "learning_rate": 5.448140703517588e-05,
-      "loss": 5.2093,
-      "step": 45807
-    },
-    {
-      "epoch": 23.889439374185137,
-      "grad_norm": 1.5380135774612427,
-      "learning_rate": 5.4480402010050256e-05,
-      "loss": 5.3221,
-      "step": 45808
-    },
-    {
-      "epoch": 23.889960886571057,
-      "grad_norm": 1.541497826576233,
-      "learning_rate": 5.447939698492463e-05,
-      "loss": 4.6299,
-      "step": 45809
-    },
-    {
-      "epoch": 23.890482398956976,
-      "grad_norm": 1.5378525257110596,
-      "learning_rate": 5.4478391959799005e-05,
-      "loss": 5.1103,
-      "step": 45810
-    },
-    {
-      "epoch": 23.891003911342896,
-      "grad_norm": 1.5975524187088013,
-      "learning_rate": 5.447738693467337e-05,
-      "loss": 5.1851,
-      "step": 45811
-    },
-    {
-      "epoch": 23.891525423728815,
-      "grad_norm": 1.6927083730697632,
-      "learning_rate": 5.447638190954775e-05,
-      "loss": 4.4792,
-      "step": 45812
-    },
-    {
-      "epoch": 23.89204693611473,
-      "grad_norm": 1.562325358390808,
-      "learning_rate": 5.447537688442211e-05,
-      "loss": 5.3374,
-      "step": 45813
-    },
-    {
-      "epoch": 23.89256844850065,
-      "grad_norm": 1.5862953662872314,
-      "learning_rate": 5.447437185929648e-05,
-      "loss": 5.5624,
-      "step": 45814
-    },
-    {
-      "epoch": 23.89308996088657,
-      "grad_norm": 1.595808744430542,
-      "learning_rate": 5.447336683417086e-05,
-      "loss": 5.2084,
-      "step": 45815
-    },
-    {
-      "epoch": 23.89361147327249,
-      "grad_norm": 1.6932799816131592,
-      "learning_rate": 5.4472361809045224e-05,
-      "loss": 5.1516,
-      "step": 45816
-    },
-    {
-      "epoch": 23.89413298565841,
-      "grad_norm": 1.4883548021316528,
-      "learning_rate": 5.44713567839196e-05,
-      "loss": 5.1434,
-      "step": 45817
-    },
-    {
-      "epoch": 23.89465449804433,
-      "grad_norm": 1.5288604497909546,
-      "learning_rate": 5.4470351758793966e-05,
-      "loss": 5.2264,
-      "step": 45818
-    },
-    {
-      "epoch": 23.895176010430248,
-      "grad_norm": 1.6474590301513672,
-      "learning_rate": 5.4469346733668344e-05,
-      "loss": 5.1152,
-      "step": 45819
-    },
-    {
-      "epoch": 23.895697522816167,
-      "grad_norm": 1.639183521270752,
-      "learning_rate": 5.4468341708542715e-05,
-      "loss": 5.4581,
-      "step": 45820
-    },
-    {
-      "epoch": 23.896219035202087,
-      "grad_norm": 1.6723036766052246,
-      "learning_rate": 5.446733668341709e-05,
-      "loss": 5.4748,
-      "step": 45821
-    },
-    {
-      "epoch": 23.896740547588006,
-      "grad_norm": 1.7622416019439697,
-      "learning_rate": 5.446633165829146e-05,
-      "loss": 5.0331,
-      "step": 45822
-    },
-    {
-      "epoch": 23.897262059973926,
-      "grad_norm": 1.594859004020691,
-      "learning_rate": 5.4465326633165835e-05,
-      "loss": 5.2682,
-      "step": 45823
-    },
-    {
-      "epoch": 23.89778357235984,
-      "grad_norm": 1.5091334581375122,
-      "learning_rate": 5.4464321608040206e-05,
-      "loss": 5.3948,
-      "step": 45824
-    },
-    {
-      "epoch": 23.89830508474576,
-      "grad_norm": 1.4702675342559814,
-      "learning_rate": 5.4463316582914584e-05,
-      "loss": 5.348,
-      "step": 45825
-    },
-    {
-      "epoch": 23.89882659713168,
-      "grad_norm": 1.536076307296753,
-      "learning_rate": 5.446231155778895e-05,
-      "loss": 5.3346,
-      "step": 45826
-    },
-    {
-      "epoch": 23.8993481095176,
-      "grad_norm": 1.4801017045974731,
-      "learning_rate": 5.446130653266331e-05,
-      "loss": 5.5954,
-      "step": 45827
-    },
-    {
-      "epoch": 23.89986962190352,
-      "grad_norm": 1.4148688316345215,
-      "learning_rate": 5.446030150753769e-05,
-      "loss": 5.4358,
-      "step": 45828
-    },
-    {
-      "epoch": 23.90039113428944,
-      "grad_norm": 1.5147449970245361,
-      "learning_rate": 5.445929648241206e-05,
-      "loss": 5.5276,
-      "step": 45829
-    },
-    {
-      "epoch": 23.90091264667536,
-      "grad_norm": 1.5393526554107666,
-      "learning_rate": 5.445829145728644e-05,
-      "loss": 5.3064,
-      "step": 45830
-    },
-    {
-      "epoch": 23.901434159061278,
-      "grad_norm": 1.4648455381393433,
-      "learning_rate": 5.4457286432160804e-05,
-      "loss": 5.0436,
-      "step": 45831
-    },
-    {
-      "epoch": 23.901955671447197,
-      "grad_norm": 1.5096162557601929,
-      "learning_rate": 5.445628140703518e-05,
-      "loss": 4.9541,
-      "step": 45832
-    },
-    {
-      "epoch": 23.902477183833117,
-      "grad_norm": 1.5777872800827026,
-      "learning_rate": 5.4455276381909546e-05,
-      "loss": 5.452,
-      "step": 45833
-    },
-    {
-      "epoch": 23.902998696219036,
-      "grad_norm": 1.6403920650482178,
-      "learning_rate": 5.4454271356783923e-05,
-      "loss": 4.8935,
-      "step": 45834
-    },
-    {
-      "epoch": 23.903520208604956,
-      "grad_norm": 1.5506229400634766,
-      "learning_rate": 5.4453266331658295e-05,
-      "loss": 5.1948,
-      "step": 45835
-    },
-    {
-      "epoch": 23.904041720990875,
-      "grad_norm": 1.551715612411499,
-      "learning_rate": 5.445226130653267e-05,
-      "loss": 4.6453,
-      "step": 45836
-    },
-    {
-      "epoch": 23.90456323337679,
-      "grad_norm": 1.545270562171936,
-      "learning_rate": 5.445125628140704e-05,
-      "loss": 5.0619,
-      "step": 45837
-    },
-    {
-      "epoch": 23.90508474576271,
-      "grad_norm": 1.6402338743209839,
-      "learning_rate": 5.4450251256281414e-05,
-      "loss": 5.1712,
-      "step": 45838
-    },
-    {
-      "epoch": 23.90560625814863,
-      "grad_norm": 1.5181512832641602,
-      "learning_rate": 5.444924623115578e-05,
-      "loss": 5.1897,
-      "step": 45839
-    },
-    {
-      "epoch": 23.90612777053455,
-      "grad_norm": 1.585888147354126,
-      "learning_rate": 5.444824120603015e-05,
-      "loss": 4.7403,
-      "step": 45840
-    },
-    {
-      "epoch": 23.90664928292047,
-      "grad_norm": 1.6686553955078125,
-      "learning_rate": 5.444723618090453e-05,
-      "loss": 5.1283,
-      "step": 45841
-    },
-    {
-      "epoch": 23.90717079530639,
-      "grad_norm": 1.4981985092163086,
-      "learning_rate": 5.444623115577889e-05,
-      "loss": 5.5001,
-      "step": 45842
-    },
-    {
-      "epoch": 23.907692307692308,
-      "grad_norm": 1.551850438117981,
-      "learning_rate": 5.444522613065327e-05,
-      "loss": 5.2693,
-      "step": 45843
-    },
-    {
-      "epoch": 23.908213820078227,
-      "grad_norm": 1.5057674646377563,
-      "learning_rate": 5.444422110552764e-05,
-      "loss": 5.1421,
-      "step": 45844
-    },
-    {
-      "epoch": 23.908735332464147,
-      "grad_norm": 1.686669111251831,
-      "learning_rate": 5.444321608040202e-05,
-      "loss": 4.8524,
-      "step": 45845
-    },
-    {
-      "epoch": 23.909256844850066,
-      "grad_norm": 1.644380807876587,
-      "learning_rate": 5.444221105527638e-05,
-      "loss": 4.4684,
-      "step": 45846
-    },
-    {
-      "epoch": 23.909778357235986,
-      "grad_norm": 1.5846244096755981,
-      "learning_rate": 5.444120603015076e-05,
-      "loss": 5.4292,
-      "step": 45847
-    },
-    {
-      "epoch": 23.910299869621902,
-      "grad_norm": 1.5941330194473267,
-      "learning_rate": 5.4440201005025125e-05,
-      "loss": 5.2438,
-      "step": 45848
-    },
-    {
-      "epoch": 23.91082138200782,
-      "grad_norm": 1.5942860841751099,
-      "learning_rate": 5.44391959798995e-05,
-      "loss": 5.2057,
-      "step": 45849
-    },
-    {
-      "epoch": 23.91134289439374,
-      "grad_norm": 1.5306389331817627,
-      "learning_rate": 5.4438190954773874e-05,
-      "loss": 5.6217,
-      "step": 45850
-    },
-    {
-      "epoch": 23.91186440677966,
-      "grad_norm": 1.7005889415740967,
-      "learning_rate": 5.443718592964825e-05,
-      "loss": 4.7101,
-      "step": 45851
-    },
-    {
-      "epoch": 23.91238591916558,
-      "grad_norm": 1.5270657539367676,
-      "learning_rate": 5.4436180904522616e-05,
-      "loss": 4.7201,
-      "step": 45852
-    },
-    {
-      "epoch": 23.9129074315515,
-      "grad_norm": 1.5456143617630005,
-      "learning_rate": 5.443517587939698e-05,
-      "loss": 5.1935,
-      "step": 45853
-    },
-    {
-      "epoch": 23.91342894393742,
-      "grad_norm": 1.4218398332595825,
-      "learning_rate": 5.443417085427136e-05,
-      "loss": 5.7742,
-      "step": 45854
-    },
-    {
-      "epoch": 23.913950456323338,
-      "grad_norm": 1.5944900512695312,
-      "learning_rate": 5.443316582914573e-05,
-      "loss": 5.5022,
-      "step": 45855
-    },
-    {
-      "epoch": 23.914471968709258,
-      "grad_norm": 1.5528451204299927,
-      "learning_rate": 5.443216080402011e-05,
-      "loss": 4.9247,
-      "step": 45856
-    },
-    {
-      "epoch": 23.914993481095177,
-      "grad_norm": 1.5426291227340698,
-      "learning_rate": 5.443115577889447e-05,
-      "loss": 5.251,
-      "step": 45857
-    },
-    {
-      "epoch": 23.915514993481096,
-      "grad_norm": 1.589393973350525,
-      "learning_rate": 5.443015075376885e-05,
-      "loss": 5.4151,
-      "step": 45858
-    },
-    {
-      "epoch": 23.916036505867016,
-      "grad_norm": 1.58730947971344,
-      "learning_rate": 5.442914572864321e-05,
-      "loss": 5.6786,
-      "step": 45859
-    },
-    {
-      "epoch": 23.916558018252932,
-      "grad_norm": 1.4604380130767822,
-      "learning_rate": 5.442814070351759e-05,
-      "loss": 5.7695,
-      "step": 45860
-    },
-    {
-      "epoch": 23.91707953063885,
-      "grad_norm": 1.6044037342071533,
-      "learning_rate": 5.442713567839196e-05,
-      "loss": 5.1913,
-      "step": 45861
-    },
-    {
-      "epoch": 23.91760104302477,
-      "grad_norm": 1.4892586469650269,
-      "learning_rate": 5.442613065326634e-05,
-      "loss": 4.5983,
-      "step": 45862
-    },
-    {
-      "epoch": 23.91812255541069,
-      "grad_norm": 1.6101680994033813,
-      "learning_rate": 5.4425125628140704e-05,
-      "loss": 4.62,
-      "step": 45863
-    },
-    {
-      "epoch": 23.91864406779661,
-      "grad_norm": 1.5110687017440796,
-      "learning_rate": 5.442412060301508e-05,
-      "loss": 5.2401,
-      "step": 45864
-    },
-    {
-      "epoch": 23.91916558018253,
-      "grad_norm": 1.5721936225891113,
-      "learning_rate": 5.4423115577889446e-05,
-      "loss": 5.088,
-      "step": 45865
-    },
-    {
-      "epoch": 23.91968709256845,
-      "grad_norm": 1.5956507921218872,
-      "learning_rate": 5.442211055276382e-05,
-      "loss": 5.1531,
-      "step": 45866
-    },
-    {
-      "epoch": 23.920208604954368,
-      "grad_norm": 1.5419212579727173,
-      "learning_rate": 5.4421105527638195e-05,
-      "loss": 5.0128,
-      "step": 45867
-    },
-    {
-      "epoch": 23.920730117340288,
-      "grad_norm": 1.4806262254714966,
-      "learning_rate": 5.442010050251256e-05,
-      "loss": 4.7288,
-      "step": 45868
-    },
-    {
-      "epoch": 23.921251629726207,
-      "grad_norm": 1.5445619821548462,
-      "learning_rate": 5.441909547738694e-05,
-      "loss": 5.5026,
-      "step": 45869
-    },
-    {
-      "epoch": 23.921773142112126,
-      "grad_norm": 1.4414212703704834,
-      "learning_rate": 5.441809045226131e-05,
-      "loss": 5.7686,
-      "step": 45870
-    },
-    {
-      "epoch": 23.922294654498046,
-      "grad_norm": 1.581486463546753,
-      "learning_rate": 5.4417085427135686e-05,
-      "loss": 4.7736,
-      "step": 45871
-    },
-    {
-      "epoch": 23.922816166883962,
-      "grad_norm": 1.5704773664474487,
-      "learning_rate": 5.441608040201005e-05,
-      "loss": 5.4808,
-      "step": 45872
-    },
-    {
-      "epoch": 23.92333767926988,
-      "grad_norm": 1.507904291152954,
-      "learning_rate": 5.441507537688443e-05,
-      "loss": 5.1613,
-      "step": 45873
-    },
-    {
-      "epoch": 23.9238591916558,
-      "grad_norm": 1.6657450199127197,
-      "learning_rate": 5.441407035175879e-05,
-      "loss": 4.7872,
-      "step": 45874
-    },
-    {
-      "epoch": 23.92438070404172,
-      "grad_norm": 1.533423662185669,
-      "learning_rate": 5.441306532663317e-05,
-      "loss": 5.3441,
-      "step": 45875
-    },
-    {
-      "epoch": 23.92490221642764,
-      "grad_norm": 1.4304890632629395,
-      "learning_rate": 5.441206030150754e-05,
-      "loss": 5.471,
-      "step": 45876
-    },
-    {
-      "epoch": 23.92542372881356,
-      "grad_norm": 1.5482558012008667,
-      "learning_rate": 5.441105527638192e-05,
-      "loss": 5.0198,
-      "step": 45877
-    },
-    {
-      "epoch": 23.92594524119948,
-      "grad_norm": 1.5618070363998413,
-      "learning_rate": 5.4410050251256284e-05,
-      "loss": 5.3508,
-      "step": 45878
-    },
-    {
-      "epoch": 23.926466753585398,
-      "grad_norm": 1.5809727907180786,
-      "learning_rate": 5.440904522613065e-05,
-      "loss": 5.2821,
-      "step": 45879
-    },
-    {
-      "epoch": 23.926988265971318,
-      "grad_norm": 1.7704592943191528,
-      "learning_rate": 5.4408040201005026e-05,
-      "loss": 4.9802,
-      "step": 45880
-    },
-    {
-      "epoch": 23.927509778357237,
-      "grad_norm": 1.6758694648742676,
-      "learning_rate": 5.44070351758794e-05,
-      "loss": 5.5108,
-      "step": 45881
-    },
-    {
-      "epoch": 23.928031290743156,
-      "grad_norm": 1.5804835557937622,
-      "learning_rate": 5.4406030150753775e-05,
-      "loss": 4.8507,
-      "step": 45882
-    },
-    {
-      "epoch": 23.928552803129076,
-      "grad_norm": 1.5148566961288452,
-      "learning_rate": 5.440502512562814e-05,
-      "loss": 5.3239,
-      "step": 45883
-    },
-    {
-      "epoch": 23.929074315514992,
-      "grad_norm": 1.539462924003601,
-      "learning_rate": 5.440402010050252e-05,
-      "loss": 5.3225,
-      "step": 45884
-    },
-    {
-      "epoch": 23.92959582790091,
-      "grad_norm": 1.6392443180084229,
-      "learning_rate": 5.440301507537688e-05,
-      "loss": 5.3838,
-      "step": 45885
-    },
-    {
-      "epoch": 23.93011734028683,
-      "grad_norm": 1.5067055225372314,
-      "learning_rate": 5.440201005025126e-05,
-      "loss": 5.6791,
-      "step": 45886
-    },
-    {
-      "epoch": 23.93063885267275,
-      "grad_norm": 1.731990933418274,
-      "learning_rate": 5.440100502512563e-05,
-      "loss": 5.2506,
-      "step": 45887
-    },
-    {
-      "epoch": 23.93116036505867,
-      "grad_norm": 1.6715755462646484,
-      "learning_rate": 5.440000000000001e-05,
-      "loss": 5.1057,
-      "step": 45888
-    },
-    {
-      "epoch": 23.93168187744459,
-      "grad_norm": 1.5228095054626465,
-      "learning_rate": 5.439899497487437e-05,
-      "loss": 5.3782,
-      "step": 45889
-    },
-    {
-      "epoch": 23.93220338983051,
-      "grad_norm": 1.5447797775268555,
-      "learning_rate": 5.439798994974875e-05,
-      "loss": 5.081,
-      "step": 45890
-    },
-    {
-      "epoch": 23.932724902216428,
-      "grad_norm": 1.530352234840393,
-      "learning_rate": 5.439698492462312e-05,
-      "loss": 5.4885,
-      "step": 45891
-    },
-    {
-      "epoch": 23.933246414602348,
-      "grad_norm": 1.4631054401397705,
-      "learning_rate": 5.4395979899497485e-05,
-      "loss": 5.2858,
-      "step": 45892
-    },
-    {
-      "epoch": 23.933767926988267,
-      "grad_norm": 1.5115141868591309,
-      "learning_rate": 5.439497487437186e-05,
-      "loss": 5.4553,
-      "step": 45893
-    },
-    {
-      "epoch": 23.934289439374187,
-      "grad_norm": 1.616053819656372,
-      "learning_rate": 5.439396984924623e-05,
-      "loss": 5.0099,
-      "step": 45894
-    },
-    {
-      "epoch": 23.934810951760106,
-      "grad_norm": 1.5843645334243774,
-      "learning_rate": 5.4392964824120605e-05,
-      "loss": 5.3652,
-      "step": 45895
-    },
-    {
-      "epoch": 23.935332464146022,
-      "grad_norm": 1.5625044107437134,
-      "learning_rate": 5.4391959798994976e-05,
-      "loss": 5.8486,
-      "step": 45896
-    },
-    {
-      "epoch": 23.93585397653194,
-      "grad_norm": 1.5380901098251343,
-      "learning_rate": 5.4390954773869354e-05,
-      "loss": 5.1604,
-      "step": 45897
-    },
-    {
-      "epoch": 23.93637548891786,
-      "grad_norm": 1.5540616512298584,
-      "learning_rate": 5.438994974874372e-05,
-      "loss": 5.2653,
-      "step": 45898
-    },
-    {
-      "epoch": 23.93689700130378,
-      "grad_norm": 1.6385149955749512,
-      "learning_rate": 5.4388944723618096e-05,
-      "loss": 5.0696,
-      "step": 45899
-    },
-    {
-      "epoch": 23.9374185136897,
-      "grad_norm": 1.659623384475708,
-      "learning_rate": 5.438793969849246e-05,
-      "loss": 5.3515,
-      "step": 45900
-    },
-    {
-      "epoch": 23.93794002607562,
-      "grad_norm": 1.5500977039337158,
-      "learning_rate": 5.438693467336684e-05,
-      "loss": 5.2196,
-      "step": 45901
-    },
-    {
-      "epoch": 23.93846153846154,
-      "grad_norm": 1.5647515058517456,
-      "learning_rate": 5.438592964824121e-05,
-      "loss": 4.5635,
-      "step": 45902
-    },
-    {
-      "epoch": 23.938983050847458,
-      "grad_norm": 1.6098263263702393,
-      "learning_rate": 5.438492462311559e-05,
-      "loss": 5.4984,
-      "step": 45903
-    },
-    {
-      "epoch": 23.939504563233378,
-      "grad_norm": 1.624824047088623,
-      "learning_rate": 5.438391959798995e-05,
-      "loss": 5.6292,
-      "step": 45904
-    },
-    {
-      "epoch": 23.940026075619297,
-      "grad_norm": 1.4059901237487793,
-      "learning_rate": 5.438291457286433e-05,
-      "loss": 5.4482,
-      "step": 45905
-    },
-    {
-      "epoch": 23.940547588005217,
-      "grad_norm": 1.4479320049285889,
-      "learning_rate": 5.4381909547738693e-05,
-      "loss": 5.2356,
-      "step": 45906
-    },
-    {
-      "epoch": 23.941069100391136,
-      "grad_norm": 1.651581048965454,
-      "learning_rate": 5.4380904522613064e-05,
-      "loss": 5.3036,
-      "step": 45907
-    },
-    {
-      "epoch": 23.941590612777052,
-      "grad_norm": 1.38026762008667,
-      "learning_rate": 5.437989949748744e-05,
-      "loss": 4.7287,
-      "step": 45908
-    },
-    {
-      "epoch": 23.94211212516297,
-      "grad_norm": 1.5506731271743774,
-      "learning_rate": 5.4378894472361807e-05,
-      "loss": 4.3538,
-      "step": 45909
-    },
-    {
-      "epoch": 23.94263363754889,
-      "grad_norm": 1.7149958610534668,
-      "learning_rate": 5.4377889447236184e-05,
-      "loss": 4.998,
-      "step": 45910
-    },
-    {
-      "epoch": 23.94315514993481,
-      "grad_norm": 1.5911314487457275,
-      "learning_rate": 5.4376884422110555e-05,
-      "loss": 5.5602,
-      "step": 45911
-    },
-    {
-      "epoch": 23.94367666232073,
-      "grad_norm": 1.52681303024292,
-      "learning_rate": 5.437587939698493e-05,
-      "loss": 4.8873,
-      "step": 45912
-    },
-    {
-      "epoch": 23.94419817470665,
-      "grad_norm": 1.5438066720962524,
-      "learning_rate": 5.43748743718593e-05,
-      "loss": 5.5888,
-      "step": 45913
-    },
-    {
-      "epoch": 23.94471968709257,
-      "grad_norm": 1.5009058713912964,
-      "learning_rate": 5.4373869346733675e-05,
-      "loss": 5.2002,
-      "step": 45914
-    },
-    {
-      "epoch": 23.945241199478488,
-      "grad_norm": 1.5067570209503174,
-      "learning_rate": 5.437286432160804e-05,
-      "loss": 5.2242,
-      "step": 45915
-    },
-    {
-      "epoch": 23.945762711864408,
-      "grad_norm": 1.6327003240585327,
-      "learning_rate": 5.437185929648242e-05,
-      "loss": 4.9546,
-      "step": 45916
-    },
-    {
-      "epoch": 23.946284224250327,
-      "grad_norm": 1.5817492008209229,
-      "learning_rate": 5.437085427135679e-05,
-      "loss": 5.449,
-      "step": 45917
-    },
-    {
-      "epoch": 23.946805736636247,
-      "grad_norm": 1.6199629306793213,
-      "learning_rate": 5.4369849246231166e-05,
-      "loss": 5.3517,
-      "step": 45918
-    },
-    {
-      "epoch": 23.947327249022166,
-      "grad_norm": 1.4944099187850952,
-      "learning_rate": 5.436884422110553e-05,
-      "loss": 4.7674,
-      "step": 45919
-    },
-    {
-      "epoch": 23.947848761408082,
-      "grad_norm": 1.4839332103729248,
-      "learning_rate": 5.4367839195979895e-05,
-      "loss": 5.1236,
-      "step": 45920
-    },
-    {
-      "epoch": 23.948370273794,
-      "grad_norm": 1.6466569900512695,
-      "learning_rate": 5.436683417085427e-05,
-      "loss": 4.8823,
-      "step": 45921
-    },
-    {
-      "epoch": 23.94889178617992,
-      "grad_norm": 1.5306611061096191,
-      "learning_rate": 5.4365829145728644e-05,
-      "loss": 5.1877,
-      "step": 45922
-    },
-    {
-      "epoch": 23.94941329856584,
-      "grad_norm": 1.474365472793579,
-      "learning_rate": 5.436482412060302e-05,
-      "loss": 5.4896,
-      "step": 45923
-    },
-    {
-      "epoch": 23.94993481095176,
-      "grad_norm": 1.5926676988601685,
-      "learning_rate": 5.4363819095477386e-05,
-      "loss": 4.9817,
-      "step": 45924
-    },
-    {
-      "epoch": 23.95045632333768,
-      "grad_norm": 1.6341049671173096,
-      "learning_rate": 5.4362814070351764e-05,
-      "loss": 4.8117,
-      "step": 45925
-    },
-    {
-      "epoch": 23.9509778357236,
-      "grad_norm": 1.6057682037353516,
-      "learning_rate": 5.436180904522613e-05,
-      "loss": 5.1572,
-      "step": 45926
-    },
-    {
-      "epoch": 23.951499348109518,
-      "grad_norm": 1.5786569118499756,
-      "learning_rate": 5.4360804020100506e-05,
-      "loss": 5.1309,
-      "step": 45927
-    },
-    {
-      "epoch": 23.952020860495438,
-      "grad_norm": 1.499778151512146,
-      "learning_rate": 5.435979899497488e-05,
-      "loss": 5.5189,
-      "step": 45928
-    },
-    {
-      "epoch": 23.952542372881357,
-      "grad_norm": 1.5492783784866333,
-      "learning_rate": 5.4358793969849255e-05,
-      "loss": 5.2531,
-      "step": 45929
-    },
-    {
-      "epoch": 23.953063885267277,
-      "grad_norm": 1.6237252950668335,
-      "learning_rate": 5.435778894472362e-05,
-      "loss": 5.0436,
-      "step": 45930
-    },
-    {
-      "epoch": 23.953585397653193,
-      "grad_norm": 1.3664621114730835,
-      "learning_rate": 5.4356783919598e-05,
-      "loss": 5.6392,
-      "step": 45931
-    },
-    {
-      "epoch": 23.954106910039112,
-      "grad_norm": 1.502312183380127,
-      "learning_rate": 5.435577889447237e-05,
-      "loss": 5.2975,
-      "step": 45932
-    },
-    {
-      "epoch": 23.95462842242503,
-      "grad_norm": 1.543453574180603,
-      "learning_rate": 5.435477386934673e-05,
-      "loss": 5.3388,
-      "step": 45933
-    },
-    {
-      "epoch": 23.95514993481095,
-      "grad_norm": 1.5563774108886719,
-      "learning_rate": 5.435376884422111e-05,
-      "loss": 5.3539,
-      "step": 45934
-    },
-    {
-      "epoch": 23.95567144719687,
-      "grad_norm": 1.46745765209198,
-      "learning_rate": 5.4352763819095474e-05,
-      "loss": 4.8749,
-      "step": 45935
-    },
-    {
-      "epoch": 23.95619295958279,
-      "grad_norm": 1.5346966981887817,
-      "learning_rate": 5.435175879396985e-05,
-      "loss": 5.385,
-      "step": 45936
-    },
-    {
-      "epoch": 23.95671447196871,
-      "grad_norm": 1.5657808780670166,
-      "learning_rate": 5.435075376884422e-05,
-      "loss": 5.0163,
-      "step": 45937
-    },
-    {
-      "epoch": 23.95723598435463,
-      "grad_norm": 1.5875946283340454,
-      "learning_rate": 5.43497487437186e-05,
-      "loss": 5.1982,
-      "step": 45938
-    },
-    {
-      "epoch": 23.957757496740548,
-      "grad_norm": 1.5437651872634888,
-      "learning_rate": 5.4348743718592965e-05,
-      "loss": 5.0069,
-      "step": 45939
-    },
-    {
-      "epoch": 23.958279009126468,
-      "grad_norm": 1.4974595308303833,
-      "learning_rate": 5.434773869346734e-05,
-      "loss": 5.324,
-      "step": 45940
-    },
-    {
-      "epoch": 23.958800521512387,
-      "grad_norm": 1.5313531160354614,
-      "learning_rate": 5.434673366834171e-05,
-      "loss": 5.6401,
-      "step": 45941
-    },
-    {
-      "epoch": 23.959322033898307,
-      "grad_norm": 1.545339584350586,
-      "learning_rate": 5.4345728643216085e-05,
-      "loss": 5.5891,
-      "step": 45942
-    },
-    {
-      "epoch": 23.959843546284226,
-      "grad_norm": 1.5711708068847656,
-      "learning_rate": 5.4344723618090456e-05,
-      "loss": 5.4524,
-      "step": 45943
-    },
-    {
-      "epoch": 23.960365058670142,
-      "grad_norm": 1.5513815879821777,
-      "learning_rate": 5.4343718592964834e-05,
-      "loss": 5.2216,
-      "step": 45944
-    },
-    {
-      "epoch": 23.96088657105606,
-      "grad_norm": 1.4729284048080444,
-      "learning_rate": 5.43427135678392e-05,
-      "loss": 5.4566,
-      "step": 45945
-    },
-    {
-      "epoch": 23.96140808344198,
-      "grad_norm": 1.5640850067138672,
-      "learning_rate": 5.434170854271356e-05,
-      "loss": 5.3657,
-      "step": 45946
-    },
-    {
-      "epoch": 23.9619295958279,
-      "grad_norm": 1.5616397857666016,
-      "learning_rate": 5.434070351758794e-05,
-      "loss": 4.732,
-      "step": 45947
-    },
-    {
-      "epoch": 23.96245110821382,
-      "grad_norm": 1.4964392185211182,
-      "learning_rate": 5.433969849246231e-05,
-      "loss": 5.3881,
-      "step": 45948
-    },
-    {
-      "epoch": 23.96297262059974,
-      "grad_norm": 1.5051445960998535,
-      "learning_rate": 5.433869346733669e-05,
-      "loss": 5.6716,
-      "step": 45949
-    },
-    {
-      "epoch": 23.96349413298566,
-      "grad_norm": 1.5547480583190918,
-      "learning_rate": 5.4337688442211054e-05,
-      "loss": 4.6809,
-      "step": 45950
-    },
-    {
-      "epoch": 23.96401564537158,
-      "grad_norm": 1.636846661567688,
-      "learning_rate": 5.433668341708543e-05,
-      "loss": 4.8921,
-      "step": 45951
-    },
-    {
-      "epoch": 23.964537157757498,
-      "grad_norm": 1.6491202116012573,
-      "learning_rate": 5.4335678391959796e-05,
-      "loss": 4.9124,
-      "step": 45952
-    },
-    {
-      "epoch": 23.965058670143417,
-      "grad_norm": 1.6907470226287842,
-      "learning_rate": 5.4334673366834173e-05,
-      "loss": 5.2623,
-      "step": 45953
-    },
-    {
-      "epoch": 23.965580182529337,
-      "grad_norm": 1.6396926641464233,
-      "learning_rate": 5.4333668341708545e-05,
-      "loss": 5.6804,
-      "step": 45954
-    },
-    {
-      "epoch": 23.966101694915253,
-      "grad_norm": 1.702321171760559,
-      "learning_rate": 5.433266331658292e-05,
-      "loss": 5.0645,
-      "step": 45955
-    },
-    {
-      "epoch": 23.966623207301172,
-      "grad_norm": 1.6079583168029785,
-      "learning_rate": 5.433165829145729e-05,
-      "loss": 5.0371,
-      "step": 45956
-    },
-    {
-      "epoch": 23.96714471968709,
-      "grad_norm": 1.6777218580245972,
-      "learning_rate": 5.4330653266331664e-05,
-      "loss": 5.171,
-      "step": 45957
-    },
-    {
-      "epoch": 23.96766623207301,
-      "grad_norm": 1.6809300184249878,
-      "learning_rate": 5.4329648241206036e-05,
-      "loss": 4.8227,
-      "step": 45958
-    },
-    {
-      "epoch": 23.96818774445893,
-      "grad_norm": 1.5920047760009766,
-      "learning_rate": 5.43286432160804e-05,
-      "loss": 5.1429,
-      "step": 45959
-    },
-    {
-      "epoch": 23.96870925684485,
-      "grad_norm": 1.508156418800354,
-      "learning_rate": 5.432763819095478e-05,
-      "loss": 5.3045,
-      "step": 45960
-    },
-    {
-      "epoch": 23.96923076923077,
-      "grad_norm": 1.5615688562393188,
-      "learning_rate": 5.432663316582914e-05,
-      "loss": 5.2084,
-      "step": 45961
-    },
-    {
-      "epoch": 23.96975228161669,
-      "grad_norm": 1.5932196378707886,
-      "learning_rate": 5.432562814070352e-05,
-      "loss": 5.135,
-      "step": 45962
-    },
-    {
-      "epoch": 23.97027379400261,
-      "grad_norm": 1.6053498983383179,
-      "learning_rate": 5.432462311557789e-05,
-      "loss": 4.8316,
-      "step": 45963
-    },
-    {
-      "epoch": 23.970795306388528,
-      "grad_norm": 1.4911683797836304,
-      "learning_rate": 5.432361809045227e-05,
-      "loss": 5.6475,
-      "step": 45964
-    },
-    {
-      "epoch": 23.971316818774447,
-      "grad_norm": 1.5564721822738647,
-      "learning_rate": 5.432261306532663e-05,
-      "loss": 5.4389,
-      "step": 45965
-    },
-    {
-      "epoch": 23.971838331160367,
-      "grad_norm": 1.5763745307922363,
-      "learning_rate": 5.432160804020101e-05,
-      "loss": 5.0623,
-      "step": 45966
-    },
-    {
-      "epoch": 23.972359843546283,
-      "grad_norm": 1.5209274291992188,
-      "learning_rate": 5.4320603015075375e-05,
-      "loss": 5.4824,
-      "step": 45967
-    },
-    {
-      "epoch": 23.972881355932202,
-      "grad_norm": 1.5195144414901733,
-      "learning_rate": 5.431959798994975e-05,
-      "loss": 4.6157,
-      "step": 45968
-    },
-    {
-      "epoch": 23.97340286831812,
-      "grad_norm": 1.5060853958129883,
-      "learning_rate": 5.4318592964824124e-05,
-      "loss": 5.596,
-      "step": 45969
-    },
-    {
-      "epoch": 23.97392438070404,
-      "grad_norm": 1.469939947128296,
-      "learning_rate": 5.43175879396985e-05,
-      "loss": 5.6327,
-      "step": 45970
-    },
-    {
-      "epoch": 23.97444589308996,
-      "grad_norm": 1.5959091186523438,
-      "learning_rate": 5.4316582914572866e-05,
-      "loss": 5.4911,
-      "step": 45971
-    },
-    {
-      "epoch": 23.97496740547588,
-      "grad_norm": 1.6032130718231201,
-      "learning_rate": 5.431557788944723e-05,
-      "loss": 5.2095,
-      "step": 45972
-    },
-    {
-      "epoch": 23.9754889178618,
-      "grad_norm": 1.549489140510559,
-      "learning_rate": 5.431457286432161e-05,
-      "loss": 5.634,
-      "step": 45973
-    },
-    {
-      "epoch": 23.97601043024772,
-      "grad_norm": 1.576242446899414,
-      "learning_rate": 5.431356783919598e-05,
-      "loss": 4.6322,
-      "step": 45974
-    },
-    {
-      "epoch": 23.97653194263364,
-      "grad_norm": 1.7056081295013428,
-      "learning_rate": 5.431256281407036e-05,
-      "loss": 4.8858,
-      "step": 45975
-    },
-    {
-      "epoch": 23.977053455019558,
-      "grad_norm": 1.6119083166122437,
-      "learning_rate": 5.431155778894472e-05,
-      "loss": 4.8088,
-      "step": 45976
-    },
-    {
-      "epoch": 23.977574967405477,
-      "grad_norm": 1.5760382413864136,
-      "learning_rate": 5.43105527638191e-05,
-      "loss": 5.564,
-      "step": 45977
-    },
-    {
-      "epoch": 23.978096479791397,
-      "grad_norm": 1.648408055305481,
-      "learning_rate": 5.430954773869347e-05,
-      "loss": 5.2894,
-      "step": 45978
-    },
-    {
-      "epoch": 23.978617992177313,
-      "grad_norm": 1.579911231994629,
-      "learning_rate": 5.430854271356785e-05,
-      "loss": 5.1021,
-      "step": 45979
-    },
-    {
-      "epoch": 23.979139504563232,
-      "grad_norm": 1.5487580299377441,
-      "learning_rate": 5.430753768844221e-05,
-      "loss": 5.6105,
-      "step": 45980
-    },
-    {
-      "epoch": 23.97966101694915,
-      "grad_norm": 1.5298149585723877,
-      "learning_rate": 5.430653266331659e-05,
-      "loss": 5.5857,
-      "step": 45981
-    },
-    {
-      "epoch": 23.98018252933507,
-      "grad_norm": 1.604411244392395,
-      "learning_rate": 5.4305527638190954e-05,
-      "loss": 5.2439,
-      "step": 45982
-    },
-    {
-      "epoch": 23.98070404172099,
-      "grad_norm": 1.6002827882766724,
-      "learning_rate": 5.430452261306533e-05,
-      "loss": 5.3309,
-      "step": 45983
-    },
-    {
-      "epoch": 23.98122555410691,
-      "grad_norm": 1.6413325071334839,
-      "learning_rate": 5.43035175879397e-05,
-      "loss": 5.5603,
-      "step": 45984
-    },
-    {
-      "epoch": 23.98174706649283,
-      "grad_norm": 1.4506750106811523,
-      "learning_rate": 5.430251256281408e-05,
-      "loss": 5.654,
-      "step": 45985
-    },
-    {
-      "epoch": 23.98226857887875,
-      "grad_norm": 1.4530068635940552,
-      "learning_rate": 5.4301507537688445e-05,
-      "loss": 4.8243,
-      "step": 45986
-    },
-    {
-      "epoch": 23.98279009126467,
-      "grad_norm": 1.5964878797531128,
-      "learning_rate": 5.430050251256281e-05,
-      "loss": 4.9618,
-      "step": 45987
-    },
-    {
-      "epoch": 23.983311603650588,
-      "grad_norm": 1.5209447145462036,
-      "learning_rate": 5.429949748743719e-05,
-      "loss": 5.3976,
-      "step": 45988
-    },
-    {
-      "epoch": 23.983833116036507,
-      "grad_norm": Infinity,
-      "learning_rate": 5.429949748743719e-05,
-      "loss": 4.8725,
-      "step": 45989
-    },
-    {
-      "epoch": 23.984354628422427,
-      "grad_norm": 1.5815855264663696,
-      "learning_rate": 5.429849246231156e-05,
-      "loss": 5.3473,
-      "step": 45990
-    },
-    {
-      "epoch": 23.984876140808343,
-      "grad_norm": 1.5126343965530396,
-      "learning_rate": 5.4297487437185936e-05,
-      "loss": 5.3729,
-      "step": 45991
-    },
-    {
-      "epoch": 23.985397653194262,
-      "grad_norm": 1.6722396612167358,
-      "learning_rate": 5.42964824120603e-05,
-      "loss": 5.3587,
-      "step": 45992
-    },
-    {
-      "epoch": 23.98591916558018,
-      "grad_norm": 1.5411133766174316,
-      "learning_rate": 5.429547738693468e-05,
-      "loss": 5.5628,
-      "step": 45993
-    },
-    {
-      "epoch": 23.9864406779661,
-      "grad_norm": 1.5605889558792114,
-      "learning_rate": 5.429447236180904e-05,
-      "loss": 4.8569,
-      "step": 45994
-    },
-    {
-      "epoch": 23.98696219035202,
-      "grad_norm": 1.5758171081542969,
-      "learning_rate": 5.429346733668342e-05,
-      "loss": 4.7693,
-      "step": 45995
-    },
-    {
-      "epoch": 23.98748370273794,
-      "grad_norm": 1.5309958457946777,
-      "learning_rate": 5.429246231155779e-05,
-      "loss": 4.937,
-      "step": 45996
-    },
-    {
-      "epoch": 23.98800521512386,
-      "grad_norm": 1.4814704656600952,
-      "learning_rate": 5.429145728643217e-05,
-      "loss": 5.3306,
-      "step": 45997
-    },
-    {
-      "epoch": 23.98852672750978,
-      "grad_norm": 1.511664867401123,
-      "learning_rate": 5.4290452261306534e-05,
-      "loss": 5.3207,
-      "step": 45998
-    },
-    {
-      "epoch": 23.9890482398957,
-      "grad_norm": 1.4380537271499634,
-      "learning_rate": 5.428944723618091e-05,
-      "loss": 5.5652,
-      "step": 45999
-    },
-    {
-      "epoch": 23.989569752281618,
-      "grad_norm": 1.4643508195877075,
-      "learning_rate": 5.428844221105528e-05,
-      "loss": 5.5917,
-      "step": 46000
-    },
-    {
-      "epoch": 23.990091264667537,
-      "grad_norm": 1.6133898496627808,
-      "learning_rate": 5.428743718592965e-05,
-      "loss": 5.2042,
-      "step": 46001
-    },
-    {
-      "epoch": 23.990612777053457,
-      "grad_norm": 1.6193649768829346,
-      "learning_rate": 5.4286432160804025e-05,
-      "loss": 5.175,
-      "step": 46002
-    },
-    {
-      "epoch": 23.991134289439373,
-      "grad_norm": 1.4771482944488525,
-      "learning_rate": 5.428542713567839e-05,
-      "loss": 5.3236,
-      "step": 46003
-    },
-    {
-      "epoch": 23.991655801825292,
-      "grad_norm": 1.6050174236297607,
-      "learning_rate": 5.428442211055277e-05,
-      "loss": 4.4981,
-      "step": 46004
-    },
-    {
-      "epoch": 23.99217731421121,
-      "grad_norm": 1.7049267292022705,
-      "learning_rate": 5.428341708542714e-05,
-      "loss": 5.0193,
-      "step": 46005
-    },
-    {
-      "epoch": 23.99269882659713,
-      "grad_norm": 1.4995810985565186,
-      "learning_rate": 5.4282412060301516e-05,
-      "loss": 5.6079,
-      "step": 46006
-    },
-    {
-      "epoch": 23.99322033898305,
-      "grad_norm": 1.4873191118240356,
-      "learning_rate": 5.428140703517588e-05,
-      "loss": 5.1858,
-      "step": 46007
-    },
-    {
-      "epoch": 23.99374185136897,
-      "grad_norm": 1.5495026111602783,
-      "learning_rate": 5.428040201005026e-05,
-      "loss": 5.0814,
-      "step": 46008
-    },
-    {
-      "epoch": 23.99426336375489,
-      "grad_norm": 1.4731147289276123,
-      "learning_rate": 5.427939698492462e-05,
-      "loss": 5.3038,
-      "step": 46009
-    },
-    {
-      "epoch": 23.99478487614081,
-      "grad_norm": 1.5070608854293823,
-      "learning_rate": 5.4278391959799e-05,
-      "loss": 4.8748,
-      "step": 46010
-    },
-    {
-      "epoch": 23.99530638852673,
-      "grad_norm": 1.5443145036697388,
-      "learning_rate": 5.427738693467337e-05,
-      "loss": 5.2913,
-      "step": 46011
-    },
-    {
-      "epoch": 23.995827900912648,
-      "grad_norm": 1.5552198886871338,
-      "learning_rate": 5.427638190954775e-05,
-      "loss": 4.9417,
-      "step": 46012
-    },
-    {
-      "epoch": 23.996349413298567,
-      "grad_norm": 1.669577717781067,
-      "learning_rate": 5.427537688442211e-05,
-      "loss": 5.5687,
-      "step": 46013
-    },
-    {
-      "epoch": 23.996870925684483,
-      "grad_norm": 1.8017690181732178,
-      "learning_rate": 5.427437185929648e-05,
-      "loss": 5.1779,
-      "step": 46014
-    },
-    {
-      "epoch": 23.997392438070403,
-      "grad_norm": 1.5378084182739258,
-      "learning_rate": 5.4273366834170855e-05,
-      "loss": 5.3497,
-      "step": 46015
-    },
-    {
-      "epoch": 23.997913950456322,
-      "grad_norm": 1.5156443119049072,
-      "learning_rate": 5.4272361809045226e-05,
-      "loss": 5.1203,
-      "step": 46016
-    },
-    {
-      "epoch": 23.99843546284224,
-      "grad_norm": 1.5367891788482666,
-      "learning_rate": 5.4271356783919604e-05,
-      "loss": 5.5455,
-      "step": 46017
-    },
-    {
-      "epoch": 23.99895697522816,
-      "grad_norm": 1.4522937536239624,
-      "learning_rate": 5.427035175879397e-05,
-      "loss": 5.6112,
-      "step": 46018
-    },
-    {
-      "epoch": 23.99947848761408,
-      "grad_norm": 1.613660216331482,
-      "learning_rate": 5.4269346733668346e-05,
-      "loss": 4.9531,
-      "step": 46019
-    },
-    {
-      "epoch": 24.0,
-      "grad_norm": 1.6710158586502075,
-      "learning_rate": 5.426834170854271e-05,
-      "loss": 5.321,
-      "step": 46020
-    },
-    {
-      "epoch": 24.00052151238592,
-      "grad_norm": 1.4684683084487915,
-      "learning_rate": 5.426733668341709e-05,
-      "loss": 5.3245,
-      "step": 46021
-    },
-    {
-      "epoch": 24.00104302477184,
-      "grad_norm": 1.606246829032898,
-      "learning_rate": 5.426633165829146e-05,
-      "loss": 5.2038,
-      "step": 46022
-    },
-    {
-      "epoch": 24.00156453715776,
-      "grad_norm": 1.5719938278198242,
-      "learning_rate": 5.426532663316584e-05,
-      "loss": 5.137,
-      "step": 46023
-    },
-    {
-      "epoch": 24.002086049543678,
-      "grad_norm": 1.5940433740615845,
-      "learning_rate": 5.42643216080402e-05,
-      "loss": 5.0891,
-      "step": 46024
-    },
-    {
-      "epoch": 24.002607561929597,
-      "grad_norm": 1.5087535381317139,
-      "learning_rate": 5.426331658291458e-05,
-      "loss": 5.1106,
-      "step": 46025
-    },
-    {
-      "epoch": 24.003129074315513,
-      "grad_norm": 1.52946937084198,
-      "learning_rate": 5.426231155778895e-05,
-      "loss": 4.906,
-      "step": 46026
-    },
-    {
-      "epoch": 24.003650586701433,
-      "grad_norm": 1.617351770401001,
-      "learning_rate": 5.4261306532663314e-05,
-      "loss": 4.8505,
-      "step": 46027
-    },
-    {
-      "epoch": 24.004172099087352,
-      "grad_norm": 1.5695254802703857,
-      "learning_rate": 5.426030150753769e-05,
-      "loss": 4.6777,
-      "step": 46028
-    },
-    {
-      "epoch": 24.00469361147327,
-      "grad_norm": 1.6444867849349976,
-      "learning_rate": 5.4259296482412057e-05,
-      "loss": 5.2649,
-      "step": 46029
-    },
-    {
-      "epoch": 24.00521512385919,
-      "grad_norm": 1.4997265338897705,
-      "learning_rate": 5.4258291457286434e-05,
-      "loss": 5.5159,
-      "step": 46030
-    },
-    {
-      "epoch": 24.00573663624511,
-      "grad_norm": 1.588230013847351,
-      "learning_rate": 5.4257286432160805e-05,
-      "loss": 4.848,
-      "step": 46031
-    },
-    {
-      "epoch": 24.00625814863103,
-      "grad_norm": 1.5247836112976074,
-      "learning_rate": 5.425628140703518e-05,
-      "loss": 5.714,
-      "step": 46032
-    },
-    {
-      "epoch": 24.00677966101695,
-      "grad_norm": 1.7232493162155151,
-      "learning_rate": 5.425527638190955e-05,
-      "loss": 4.978,
-      "step": 46033
-    },
-    {
-      "epoch": 24.00730117340287,
-      "grad_norm": 1.5294561386108398,
-      "learning_rate": 5.4254271356783925e-05,
-      "loss": 5.3366,
-      "step": 46034
-    },
-    {
-      "epoch": 24.00782268578879,
-      "grad_norm": 1.461788296699524,
-      "learning_rate": 5.425326633165829e-05,
-      "loss": 4.8345,
-      "step": 46035
-    },
-    {
-      "epoch": 24.008344198174708,
-      "grad_norm": 1.569153070449829,
-      "learning_rate": 5.425226130653267e-05,
-      "loss": 4.7699,
-      "step": 46036
-    },
-    {
-      "epoch": 24.008865710560627,
-      "grad_norm": 1.569054126739502,
-      "learning_rate": 5.425125628140704e-05,
-      "loss": 5.3337,
-      "step": 46037
-    },
-    {
-      "epoch": 24.009387222946543,
-      "grad_norm": 1.6185903549194336,
-      "learning_rate": 5.4250251256281416e-05,
-      "loss": 5.5041,
-      "step": 46038
-    },
-    {
-      "epoch": 24.009908735332463,
-      "grad_norm": 1.5332311391830444,
-      "learning_rate": 5.424924623115578e-05,
-      "loss": 5.4248,
-      "step": 46039
-    },
-    {
-      "epoch": 24.010430247718382,
-      "grad_norm": 1.5594563484191895,
-      "learning_rate": 5.4248241206030145e-05,
-      "loss": 5.3164,
-      "step": 46040
-    },
-    {
-      "epoch": 24.0109517601043,
-      "grad_norm": 1.5309391021728516,
-      "learning_rate": 5.424723618090452e-05,
-      "loss": 5.5143,
-      "step": 46041
-    },
-    {
-      "epoch": 24.01147327249022,
-      "grad_norm": 1.5855588912963867,
-      "learning_rate": 5.4246231155778894e-05,
-      "loss": 5.2378,
-      "step": 46042
-    },
-    {
-      "epoch": 24.01199478487614,
-      "grad_norm": 1.496659517288208,
-      "learning_rate": 5.424522613065327e-05,
-      "loss": 5.6601,
-      "step": 46043
-    },
-    {
-      "epoch": 24.01251629726206,
-      "grad_norm": 1.6735243797302246,
-      "learning_rate": 5.4244221105527636e-05,
-      "loss": 4.9908,
-      "step": 46044
-    },
-    {
-      "epoch": 24.01303780964798,
-      "grad_norm": 1.645653247833252,
-      "learning_rate": 5.4243216080402014e-05,
-      "loss": 5.0656,
-      "step": 46045
-    },
-    {
-      "epoch": 24.0135593220339,
-      "grad_norm": 1.6388884782791138,
-      "learning_rate": 5.4242211055276385e-05,
-      "loss": 5.1619,
-      "step": 46046
-    },
-    {
-      "epoch": 24.01408083441982,
-      "grad_norm": 1.6696414947509766,
-      "learning_rate": 5.424120603015076e-05,
-      "loss": 4.9574,
-      "step": 46047
-    },
-    {
-      "epoch": 24.014602346805738,
-      "grad_norm": 1.5975017547607422,
-      "learning_rate": 5.424020100502513e-05,
-      "loss": 5.12,
-      "step": 46048
-    },
-    {
-      "epoch": 24.015123859191657,
-      "grad_norm": 1.532721757888794,
-      "learning_rate": 5.4239195979899505e-05,
-      "loss": 5.4742,
-      "step": 46049
-    },
-    {
-      "epoch": 24.015645371577573,
-      "grad_norm": 1.5162363052368164,
-      "learning_rate": 5.423819095477387e-05,
-      "loss": 4.8136,
-      "step": 46050
-    },
-    {
-      "epoch": 24.016166883963493,
-      "grad_norm": 2.0477590560913086,
-      "learning_rate": 5.423718592964825e-05,
-      "loss": 4.7733,
-      "step": 46051
-    },
-    {
-      "epoch": 24.016688396349412,
-      "grad_norm": 1.5923370122909546,
-      "learning_rate": 5.423618090452262e-05,
-      "loss": 5.1871,
-      "step": 46052
-    },
-    {
-      "epoch": 24.01720990873533,
-      "grad_norm": 1.4261493682861328,
-      "learning_rate": 5.423517587939698e-05,
-      "loss": 5.7367,
-      "step": 46053
-    },
-    {
-      "epoch": 24.01773142112125,
-      "grad_norm": 1.600920557975769,
-      "learning_rate": 5.423417085427136e-05,
-      "loss": 4.4217,
-      "step": 46054
-    },
-    {
-      "epoch": 24.01825293350717,
-      "grad_norm": 1.651667594909668,
-      "learning_rate": 5.4233165829145724e-05,
-      "loss": 5.0547,
-      "step": 46055
-    },
-    {
-      "epoch": 24.01877444589309,
-      "grad_norm": 1.4112615585327148,
-      "learning_rate": 5.42321608040201e-05,
-      "loss": 5.3478,
-      "step": 46056
-    },
-    {
-      "epoch": 24.01929595827901,
-      "grad_norm": 1.5659563541412354,
-      "learning_rate": 5.423115577889447e-05,
-      "loss": 5.4329,
-      "step": 46057
-    },
-    {
-      "epoch": 24.01981747066493,
-      "grad_norm": 1.51419997215271,
-      "learning_rate": 5.423015075376885e-05,
-      "loss": 5.4193,
-      "step": 46058
-    },
-    {
-      "epoch": 24.02033898305085,
-      "grad_norm": 1.546014666557312,
-      "learning_rate": 5.4229145728643215e-05,
-      "loss": 5.276,
-      "step": 46059
-    },
-    {
-      "epoch": 24.020860495436768,
-      "grad_norm": 1.5950777530670166,
-      "learning_rate": 5.422814070351759e-05,
-      "loss": 5.1339,
-      "step": 46060
-    },
-    {
-      "epoch": 24.021382007822687,
-      "grad_norm": 1.4555310010910034,
-      "learning_rate": 5.422713567839196e-05,
-      "loss": 5.6,
-      "step": 46061
-    },
-    {
-      "epoch": 24.021903520208603,
-      "grad_norm": 1.5049126148223877,
-      "learning_rate": 5.4226130653266335e-05,
-      "loss": 5.5152,
-      "step": 46062
-    },
-    {
-      "epoch": 24.022425032594523,
-      "grad_norm": 1.4997739791870117,
-      "learning_rate": 5.4225125628140706e-05,
-      "loss": 5.4713,
-      "step": 46063
-    },
-    {
-      "epoch": 24.022946544980442,
-      "grad_norm": 1.556372880935669,
-      "learning_rate": 5.4224120603015084e-05,
-      "loss": 5.0858,
-      "step": 46064
-    },
-    {
-      "epoch": 24.02346805736636,
-      "grad_norm": 1.4993529319763184,
-      "learning_rate": 5.422311557788945e-05,
-      "loss": 5.0031,
-      "step": 46065
-    },
-    {
-      "epoch": 24.02398956975228,
-      "grad_norm": 1.5561082363128662,
-      "learning_rate": 5.422211055276382e-05,
-      "loss": 5.1331,
-      "step": 46066
-    },
-    {
-      "epoch": 24.0245110821382,
-      "grad_norm": 1.4982510805130005,
-      "learning_rate": 5.42211055276382e-05,
-      "loss": 5.4972,
-      "step": 46067
-    },
-    {
-      "epoch": 24.02503259452412,
-      "grad_norm": 1.620223879814148,
-      "learning_rate": 5.422010050251256e-05,
-      "loss": 5.1852,
-      "step": 46068
-    },
-    {
-      "epoch": 24.02555410691004,
-      "grad_norm": 1.6083669662475586,
-      "learning_rate": 5.421909547738694e-05,
-      "loss": 5.1749,
-      "step": 46069
-    },
-    {
-      "epoch": 24.02607561929596,
-      "grad_norm": 1.60512375831604,
-      "learning_rate": 5.4218090452261304e-05,
-      "loss": 5.1214,
-      "step": 46070
-    },
-    {
-      "epoch": 24.02659713168188,
-      "grad_norm": 1.50897216796875,
-      "learning_rate": 5.421708542713568e-05,
-      "loss": 5.4374,
-      "step": 46071
-    },
-    {
-      "epoch": 24.027118644067798,
-      "grad_norm": 1.5528723001480103,
-      "learning_rate": 5.421608040201005e-05,
-      "loss": 5.4749,
-      "step": 46072
-    },
-    {
-      "epoch": 24.027640156453717,
-      "grad_norm": 1.4743549823760986,
-      "learning_rate": 5.421507537688443e-05,
-      "loss": 5.5223,
-      "step": 46073
-    },
-    {
-      "epoch": 24.028161668839633,
-      "grad_norm": 1.5961159467697144,
-      "learning_rate": 5.4214070351758795e-05,
-      "loss": 5.4454,
-      "step": 46074
-    },
-    {
-      "epoch": 24.028683181225553,
-      "grad_norm": 1.793000340461731,
-      "learning_rate": 5.421306532663317e-05,
-      "loss": 4.731,
-      "step": 46075
-    },
-    {
-      "epoch": 24.029204693611472,
-      "grad_norm": 1.6856452226638794,
-      "learning_rate": 5.421206030150754e-05,
-      "loss": 5.3187,
-      "step": 46076
-    },
-    {
-      "epoch": 24.02972620599739,
-      "grad_norm": 1.454793930053711,
-      "learning_rate": 5.4211055276381914e-05,
-      "loss": 5.7701,
-      "step": 46077
-    },
-    {
-      "epoch": 24.03024771838331,
-      "grad_norm": 1.481123924255371,
-      "learning_rate": 5.4210050251256286e-05,
-      "loss": 5.3467,
-      "step": 46078
-    },
-    {
-      "epoch": 24.03076923076923,
-      "grad_norm": 1.4650630950927734,
-      "learning_rate": 5.420904522613066e-05,
-      "loss": 4.8783,
-      "step": 46079
-    },
-    {
-      "epoch": 24.03129074315515,
-      "grad_norm": 1.5515016317367554,
-      "learning_rate": 5.420804020100503e-05,
-      "loss": 5.5613,
-      "step": 46080
-    },
-    {
-      "epoch": 24.03181225554107,
-      "grad_norm": 1.511457085609436,
-      "learning_rate": 5.420703517587939e-05,
-      "loss": 5.2557,
-      "step": 46081
-    },
-    {
-      "epoch": 24.03233376792699,
-      "grad_norm": 1.6389484405517578,
-      "learning_rate": 5.420603015075377e-05,
-      "loss": 5.1918,
-      "step": 46082
-    },
-    {
-      "epoch": 24.03285528031291,
-      "grad_norm": 1.5714011192321777,
-      "learning_rate": 5.420502512562814e-05,
-      "loss": 5.4691,
-      "step": 46083
-    },
-    {
-      "epoch": 24.033376792698828,
-      "grad_norm": 1.618318796157837,
-      "learning_rate": 5.420402010050252e-05,
-      "loss": 4.9458,
-      "step": 46084
-    },
-    {
-      "epoch": 24.033898305084747,
-      "grad_norm": 1.6347686052322388,
-      "learning_rate": 5.420301507537688e-05,
-      "loss": 5.3703,
-      "step": 46085
-    },
-    {
-      "epoch": 24.034419817470663,
-      "grad_norm": 1.5274754762649536,
-      "learning_rate": 5.420201005025126e-05,
-      "loss": 5.6552,
-      "step": 46086
-    },
-    {
-      "epoch": 24.034941329856583,
-      "grad_norm": 1.522687315940857,
-      "learning_rate": 5.420100502512563e-05,
-      "loss": 5.4275,
-      "step": 46087
-    },
-    {
-      "epoch": 24.035462842242502,
-      "grad_norm": 1.539440631866455,
-      "learning_rate": 5.420000000000001e-05,
-      "loss": 4.6405,
-      "step": 46088
-    },
-    {
-      "epoch": 24.03598435462842,
-      "grad_norm": 1.5528409481048584,
-      "learning_rate": 5.4198994974874374e-05,
-      "loss": 5.4346,
-      "step": 46089
-    },
-    {
-      "epoch": 24.03650586701434,
-      "grad_norm": 1.5965327024459839,
-      "learning_rate": 5.419798994974875e-05,
-      "loss": 4.638,
-      "step": 46090
-    },
-    {
-      "epoch": 24.03702737940026,
-      "grad_norm": 1.5039515495300293,
-      "learning_rate": 5.4196984924623116e-05,
-      "loss": 5.6826,
-      "step": 46091
-    },
-    {
-      "epoch": 24.03754889178618,
-      "grad_norm": 1.6296741962432861,
-      "learning_rate": 5.4195979899497494e-05,
-      "loss": 4.9458,
-      "step": 46092
-    },
-    {
-      "epoch": 24.0380704041721,
-      "grad_norm": 1.5898256301879883,
-      "learning_rate": 5.4194974874371865e-05,
-      "loss": 5.0443,
-      "step": 46093
-    },
-    {
-      "epoch": 24.03859191655802,
-      "grad_norm": 1.5694644451141357,
-      "learning_rate": 5.419396984924623e-05,
-      "loss": 5.3816,
-      "step": 46094
-    },
-    {
-      "epoch": 24.03911342894394,
-      "grad_norm": 1.4784035682678223,
-      "learning_rate": 5.419296482412061e-05,
-      "loss": 4.9497,
-      "step": 46095
-    },
-    {
-      "epoch": 24.039634941329858,
-      "grad_norm": 1.5374153852462769,
-      "learning_rate": 5.419195979899497e-05,
-      "loss": 5.3113,
-      "step": 46096
-    },
-    {
-      "epoch": 24.040156453715777,
-      "grad_norm": 1.671799659729004,
-      "learning_rate": 5.419095477386935e-05,
-      "loss": 5.1469,
-      "step": 46097
-    },
-    {
-      "epoch": 24.040677966101693,
-      "grad_norm": 1.7802436351776123,
-      "learning_rate": 5.418994974874372e-05,
-      "loss": 5.039,
-      "step": 46098
-    },
-    {
-      "epoch": 24.041199478487613,
-      "grad_norm": 1.7207837104797363,
-      "learning_rate": 5.41889447236181e-05,
-      "loss": 4.6694,
-      "step": 46099
-    },
-    {
-      "epoch": 24.041720990873532,
-      "grad_norm": 1.5498486757278442,
-      "learning_rate": 5.418793969849246e-05,
-      "loss": 4.9114,
-      "step": 46100
-    },
-    {
-      "epoch": 24.042242503259452,
-      "grad_norm": 1.6191529035568237,
-      "learning_rate": 5.418693467336684e-05,
-      "loss": 5.2036,
-      "step": 46101
-    },
-    {
-      "epoch": 24.04276401564537,
-      "grad_norm": 1.580148696899414,
-      "learning_rate": 5.4185929648241204e-05,
-      "loss": 5.053,
-      "step": 46102
-    },
-    {
-      "epoch": 24.04328552803129,
-      "grad_norm": 1.4993054866790771,
-      "learning_rate": 5.418492462311558e-05,
-      "loss": 5.4512,
-      "step": 46103
-    },
-    {
-      "epoch": 24.04380704041721,
-      "grad_norm": 1.642574667930603,
-      "learning_rate": 5.418391959798995e-05,
-      "loss": 5.1678,
-      "step": 46104
-    },
-    {
-      "epoch": 24.04432855280313,
-      "grad_norm": 1.6004410982131958,
-      "learning_rate": 5.418291457286433e-05,
-      "loss": 4.4682,
-      "step": 46105
-    },
-    {
-      "epoch": 24.04485006518905,
-      "grad_norm": 1.6212694644927979,
-      "learning_rate": 5.4181909547738695e-05,
-      "loss": 5.0306,
-      "step": 46106
-    },
-    {
-      "epoch": 24.04537157757497,
-      "grad_norm": 1.5393000841140747,
-      "learning_rate": 5.418090452261306e-05,
-      "loss": 5.5827,
-      "step": 46107
-    },
-    {
-      "epoch": 24.045893089960888,
-      "grad_norm": 1.6398674249649048,
-      "learning_rate": 5.417989949748744e-05,
-      "loss": 4.8379,
-      "step": 46108
-    },
-    {
-      "epoch": 24.046414602346807,
-      "grad_norm": 1.5472393035888672,
-      "learning_rate": 5.417889447236181e-05,
-      "loss": 5.0969,
-      "step": 46109
-    },
-    {
-      "epoch": 24.046936114732723,
-      "grad_norm": 1.607438564300537,
-      "learning_rate": 5.4177889447236186e-05,
-      "loss": 5.1359,
-      "step": 46110
-    },
-    {
-      "epoch": 24.047457627118643,
-      "grad_norm": 1.4798245429992676,
-      "learning_rate": 5.417688442211055e-05,
-      "loss": 5.4496,
-      "step": 46111
-    },
-    {
-      "epoch": 24.047979139504562,
-      "grad_norm": 1.615877389907837,
-      "learning_rate": 5.417587939698493e-05,
-      "loss": 5.3755,
-      "step": 46112
-    },
-    {
-      "epoch": 24.048500651890482,
-      "grad_norm": 1.4993796348571777,
-      "learning_rate": 5.41748743718593e-05,
-      "loss": 5.2106,
-      "step": 46113
-    },
-    {
-      "epoch": 24.0490221642764,
-      "grad_norm": 1.5948331356048584,
-      "learning_rate": 5.417386934673368e-05,
-      "loss": 4.8748,
-      "step": 46114
-    },
-    {
-      "epoch": 24.04954367666232,
-      "grad_norm": 1.5158321857452393,
-      "learning_rate": 5.417286432160804e-05,
-      "loss": 5.172,
-      "step": 46115
-    },
-    {
-      "epoch": 24.05006518904824,
-      "grad_norm": 1.513336181640625,
-      "learning_rate": 5.417185929648242e-05,
-      "loss": 5.1024,
-      "step": 46116
-    },
-    {
-      "epoch": 24.05058670143416,
-      "grad_norm": 1.5061228275299072,
-      "learning_rate": 5.4170854271356784e-05,
-      "loss": 5.2831,
-      "step": 46117
-    },
-    {
-      "epoch": 24.05110821382008,
-      "grad_norm": 1.5956271886825562,
-      "learning_rate": 5.416984924623116e-05,
-      "loss": 4.9538,
-      "step": 46118
-    },
-    {
-      "epoch": 24.051629726206,
-      "grad_norm": 1.5196985006332397,
-      "learning_rate": 5.416884422110553e-05,
-      "loss": 5.1045,
-      "step": 46119
-    },
-    {
-      "epoch": 24.052151238591918,
-      "grad_norm": 1.6203173398971558,
-      "learning_rate": 5.41678391959799e-05,
-      "loss": 5.3808,
-      "step": 46120
-    },
-    {
-      "epoch": 24.052672750977834,
-      "grad_norm": 1.612718105316162,
-      "learning_rate": 5.4166834170854275e-05,
-      "loss": 5.2778,
-      "step": 46121
-    },
-    {
-      "epoch": 24.053194263363753,
-      "grad_norm": 1.5709731578826904,
-      "learning_rate": 5.416582914572864e-05,
-      "loss": 5.3889,
-      "step": 46122
-    },
-    {
-      "epoch": 24.053715775749673,
-      "grad_norm": 1.4780852794647217,
-      "learning_rate": 5.416482412060302e-05,
-      "loss": 5.0213,
-      "step": 46123
-    },
-    {
-      "epoch": 24.054237288135592,
-      "grad_norm": 1.5251973867416382,
-      "learning_rate": 5.416381909547739e-05,
-      "loss": 4.9747,
-      "step": 46124
-    },
-    {
-      "epoch": 24.054758800521512,
-      "grad_norm": 1.4882943630218506,
-      "learning_rate": 5.4162814070351766e-05,
-      "loss": 5.4213,
-      "step": 46125
-    },
-    {
-      "epoch": 24.05528031290743,
-      "grad_norm": 1.6323121786117554,
-      "learning_rate": 5.416180904522613e-05,
-      "loss": 4.8621,
-      "step": 46126
-    },
-    {
-      "epoch": 24.05580182529335,
-      "grad_norm": 1.6088634729385376,
-      "learning_rate": 5.416080402010051e-05,
-      "loss": 5.0845,
-      "step": 46127
-    },
-    {
-      "epoch": 24.05632333767927,
-      "grad_norm": 1.7182953357696533,
-      "learning_rate": 5.415979899497487e-05,
-      "loss": 4.815,
-      "step": 46128
-    },
-    {
-      "epoch": 24.05684485006519,
-      "grad_norm": 1.5520579814910889,
-      "learning_rate": 5.415879396984925e-05,
-      "loss": 5.4736,
-      "step": 46129
-    },
-    {
-      "epoch": 24.05736636245111,
-      "grad_norm": 1.624547004699707,
-      "learning_rate": 5.415778894472362e-05,
-      "loss": 5.0882,
-      "step": 46130
-    },
-    {
-      "epoch": 24.05788787483703,
-      "grad_norm": 1.6157820224761963,
-      "learning_rate": 5.4156783919598e-05,
-      "loss": 5.4171,
-      "step": 46131
-    },
-    {
-      "epoch": 24.058409387222948,
-      "grad_norm": 1.5797287225723267,
-      "learning_rate": 5.415577889447236e-05,
-      "loss": 5.019,
-      "step": 46132
-    },
-    {
-      "epoch": 24.058930899608864,
-      "grad_norm": 1.559063196182251,
-      "learning_rate": 5.4154773869346734e-05,
-      "loss": 4.7679,
-      "step": 46133
-    },
-    {
-      "epoch": 24.059452411994783,
-      "grad_norm": 1.62010657787323,
-      "learning_rate": 5.415376884422111e-05,
-      "loss": 5.3454,
-      "step": 46134
-    },
-    {
-      "epoch": 24.059973924380703,
-      "grad_norm": 1.639583706855774,
-      "learning_rate": 5.4152763819095476e-05,
-      "loss": 5.1264,
-      "step": 46135
-    },
-    {
-      "epoch": 24.060495436766622,
-      "grad_norm": 1.5523625612258911,
-      "learning_rate": 5.4151758793969854e-05,
-      "loss": 5.1476,
-      "step": 46136
-    },
-    {
-      "epoch": 24.061016949152542,
-      "grad_norm": 1.6235785484313965,
-      "learning_rate": 5.415075376884422e-05,
-      "loss": 4.9372,
-      "step": 46137
-    },
-    {
-      "epoch": 24.06153846153846,
-      "grad_norm": 1.5566412210464478,
-      "learning_rate": 5.4149748743718596e-05,
-      "loss": 5.5051,
-      "step": 46138
-    },
-    {
-      "epoch": 24.06205997392438,
-      "grad_norm": 1.6164720058441162,
-      "learning_rate": 5.414874371859297e-05,
-      "loss": 4.5743,
-      "step": 46139
-    },
-    {
-      "epoch": 24.0625814863103,
-      "grad_norm": 1.5199687480926514,
-      "learning_rate": 5.4147738693467345e-05,
-      "loss": 5.4282,
-      "step": 46140
-    },
-    {
-      "epoch": 24.06310299869622,
-      "grad_norm": 1.6363195180892944,
-      "learning_rate": 5.414673366834171e-05,
-      "loss": 4.3532,
-      "step": 46141
-    },
-    {
-      "epoch": 24.06362451108214,
-      "grad_norm": 1.5296297073364258,
-      "learning_rate": 5.414572864321609e-05,
-      "loss": 5.3846,
-      "step": 46142
-    },
-    {
-      "epoch": 24.06414602346806,
-      "grad_norm": 1.6112593412399292,
-      "learning_rate": 5.414472361809045e-05,
-      "loss": 5.0458,
-      "step": 46143
-    },
-    {
-      "epoch": 24.064667535853978,
-      "grad_norm": 1.5586239099502563,
-      "learning_rate": 5.414371859296483e-05,
-      "loss": 5.0657,
-      "step": 46144
-    },
-    {
-      "epoch": 24.065189048239894,
-      "grad_norm": 1.5466039180755615,
-      "learning_rate": 5.41427135678392e-05,
-      "loss": 5.5197,
-      "step": 46145
-    },
-    {
-      "epoch": 24.065710560625813,
-      "grad_norm": 1.5270720720291138,
-      "learning_rate": 5.4141708542713564e-05,
-      "loss": 5.4398,
-      "step": 46146
-    },
-    {
-      "epoch": 24.066232073011733,
-      "grad_norm": 1.4853025674819946,
-      "learning_rate": 5.414070351758794e-05,
-      "loss": 5.2081,
-      "step": 46147
-    },
-    {
-      "epoch": 24.066753585397652,
-      "grad_norm": 1.4869545698165894,
-      "learning_rate": 5.4139698492462307e-05,
-      "loss": 5.0266,
-      "step": 46148
-    },
-    {
-      "epoch": 24.067275097783572,
-      "grad_norm": 1.5332838296890259,
-      "learning_rate": 5.4138693467336684e-05,
-      "loss": 5.5649,
-      "step": 46149
-    },
-    {
-      "epoch": 24.06779661016949,
-      "grad_norm": 1.5586276054382324,
-      "learning_rate": 5.4137688442211055e-05,
-      "loss": 4.9503,
-      "step": 46150
-    },
-    {
-      "epoch": 24.06831812255541,
-      "grad_norm": 1.4693835973739624,
-      "learning_rate": 5.413668341708543e-05,
-      "loss": 5.2637,
-      "step": 46151
-    },
-    {
-      "epoch": 24.06883963494133,
-      "grad_norm": 1.6159244775772095,
-      "learning_rate": 5.41356783919598e-05,
-      "loss": 4.9665,
-      "step": 46152
-    },
-    {
-      "epoch": 24.06936114732725,
-      "grad_norm": 1.6912510395050049,
-      "learning_rate": 5.4134673366834175e-05,
-      "loss": 4.9367,
-      "step": 46153
-    },
-    {
-      "epoch": 24.06988265971317,
-      "grad_norm": 1.5656617879867554,
-      "learning_rate": 5.4133668341708546e-05,
-      "loss": 4.9491,
-      "step": 46154
-    },
-    {
-      "epoch": 24.07040417209909,
-      "grad_norm": 1.546600580215454,
-      "learning_rate": 5.4132663316582924e-05,
-      "loss": 5.5664,
-      "step": 46155
-    },
-    {
-      "epoch": 24.070925684485008,
-      "grad_norm": 1.6194109916687012,
-      "learning_rate": 5.413165829145729e-05,
-      "loss": 4.3685,
-      "step": 46156
-    },
-    {
-      "epoch": 24.071447196870924,
-      "grad_norm": 1.514235019683838,
-      "learning_rate": 5.4130653266331666e-05,
-      "loss": 5.4297,
-      "step": 46157
-    },
-    {
-      "epoch": 24.071968709256844,
-      "grad_norm": 1.4718817472457886,
-      "learning_rate": 5.412964824120603e-05,
-      "loss": 5.6158,
-      "step": 46158
-    },
-    {
-      "epoch": 24.072490221642763,
-      "grad_norm": 1.4517574310302734,
-      "learning_rate": 5.41286432160804e-05,
-      "loss": 5.5131,
-      "step": 46159
-    },
-    {
-      "epoch": 24.073011734028682,
-      "grad_norm": 1.551672101020813,
-      "learning_rate": 5.412763819095478e-05,
-      "loss": 5.3943,
-      "step": 46160
-    },
-    {
-      "epoch": 24.073533246414602,
-      "grad_norm": 1.5712032318115234,
-      "learning_rate": 5.4126633165829144e-05,
-      "loss": 4.7555,
-      "step": 46161
-    },
-    {
-      "epoch": 24.07405475880052,
-      "grad_norm": 1.4879182577133179,
-      "learning_rate": 5.412562814070352e-05,
-      "loss": 5.353,
-      "step": 46162
-    },
-    {
-      "epoch": 24.07457627118644,
-      "grad_norm": 1.5015766620635986,
-      "learning_rate": 5.4124623115577886e-05,
-      "loss": 5.0462,
-      "step": 46163
-    },
-    {
-      "epoch": 24.07509778357236,
-      "grad_norm": 1.5358426570892334,
-      "learning_rate": 5.4123618090452264e-05,
-      "loss": 4.8209,
-      "step": 46164
-    },
-    {
-      "epoch": 24.07561929595828,
-      "grad_norm": 1.445813536643982,
-      "learning_rate": 5.4122613065326635e-05,
-      "loss": 5.7537,
-      "step": 46165
-    },
-    {
-      "epoch": 24.0761408083442,
-      "grad_norm": 1.541701316833496,
-      "learning_rate": 5.412160804020101e-05,
-      "loss": 5.4983,
-      "step": 46166
-    },
-    {
-      "epoch": 24.07666232073012,
-      "grad_norm": 1.4989402294158936,
-      "learning_rate": 5.412060301507538e-05,
-      "loss": 5.1591,
-      "step": 46167
-    },
-    {
-      "epoch": 24.077183833116038,
-      "grad_norm": 1.4399510622024536,
-      "learning_rate": 5.4119597989949755e-05,
-      "loss": 5.603,
-      "step": 46168
-    },
-    {
-      "epoch": 24.077705345501954,
-      "grad_norm": 1.5259425640106201,
-      "learning_rate": 5.411859296482412e-05,
-      "loss": 5.2099,
-      "step": 46169
-    },
-    {
-      "epoch": 24.078226857887874,
-      "grad_norm": 1.7379735708236694,
-      "learning_rate": 5.41175879396985e-05,
-      "loss": 5.1335,
-      "step": 46170
-    },
-    {
-      "epoch": 24.078748370273793,
-      "grad_norm": 1.5350654125213623,
-      "learning_rate": 5.411658291457287e-05,
-      "loss": 5.1806,
-      "step": 46171
-    },
-    {
-      "epoch": 24.079269882659712,
-      "grad_norm": 1.5073529481887817,
-      "learning_rate": 5.4115577889447246e-05,
-      "loss": 5.2096,
-      "step": 46172
-    },
-    {
-      "epoch": 24.079791395045632,
-      "grad_norm": 1.676622748374939,
-      "learning_rate": 5.411457286432161e-05,
-      "loss": 4.8476,
-      "step": 46173
-    },
-    {
-      "epoch": 24.08031290743155,
-      "grad_norm": 1.5904616117477417,
-      "learning_rate": 5.411356783919598e-05,
-      "loss": 4.9586,
-      "step": 46174
-    },
-    {
-      "epoch": 24.08083441981747,
-      "grad_norm": 1.6008751392364502,
-      "learning_rate": 5.411256281407036e-05,
-      "loss": 4.9652,
-      "step": 46175
-    },
-    {
-      "epoch": 24.08135593220339,
-      "grad_norm": 1.6593527793884277,
-      "learning_rate": 5.411155778894472e-05,
-      "loss": 5.2798,
-      "step": 46176
-    },
-    {
-      "epoch": 24.08187744458931,
-      "grad_norm": 1.4933072328567505,
-      "learning_rate": 5.41105527638191e-05,
-      "loss": 5.5527,
-      "step": 46177
-    },
-    {
-      "epoch": 24.08239895697523,
-      "grad_norm": 1.5341237783432007,
-      "learning_rate": 5.4109547738693465e-05,
-      "loss": 4.9895,
-      "step": 46178
-    },
-    {
-      "epoch": 24.08292046936115,
-      "grad_norm": 1.5785399675369263,
-      "learning_rate": 5.410854271356784e-05,
-      "loss": 5.2123,
-      "step": 46179
-    },
-    {
-      "epoch": 24.083441981747068,
-      "grad_norm": 1.5738776922225952,
-      "learning_rate": 5.4107537688442214e-05,
-      "loss": 4.0493,
-      "step": 46180
-    },
-    {
-      "epoch": 24.083963494132984,
-      "grad_norm": 1.4765321016311646,
-      "learning_rate": 5.410653266331659e-05,
-      "loss": 5.4741,
-      "step": 46181
-    },
-    {
-      "epoch": 24.084485006518904,
-      "grad_norm": 1.6573847532272339,
-      "learning_rate": 5.4105527638190956e-05,
-      "loss": 5.1741,
-      "step": 46182
-    },
-    {
-      "epoch": 24.085006518904823,
-      "grad_norm": 1.5710257291793823,
-      "learning_rate": 5.4104522613065334e-05,
-      "loss": 5.0465,
-      "step": 46183
-    },
-    {
-      "epoch": 24.085528031290742,
-      "grad_norm": 1.5279206037521362,
-      "learning_rate": 5.41035175879397e-05,
-      "loss": 5.1994,
-      "step": 46184
-    },
-    {
-      "epoch": 24.086049543676662,
-      "grad_norm": 1.4894112348556519,
-      "learning_rate": 5.4102512562814076e-05,
-      "loss": 5.095,
-      "step": 46185
-    },
-    {
-      "epoch": 24.08657105606258,
-      "grad_norm": 1.6499632596969604,
-      "learning_rate": 5.410150753768845e-05,
-      "loss": 4.7079,
-      "step": 46186
-    },
-    {
-      "epoch": 24.0870925684485,
-      "grad_norm": 1.558045506477356,
-      "learning_rate": 5.410050251256281e-05,
-      "loss": 5.2951,
-      "step": 46187
-    },
-    {
-      "epoch": 24.08761408083442,
-      "grad_norm": 1.6158416271209717,
-      "learning_rate": 5.409949748743719e-05,
-      "loss": 5.0308,
-      "step": 46188
-    },
-    {
-      "epoch": 24.08813559322034,
-      "grad_norm": 1.5349894762039185,
-      "learning_rate": 5.4098492462311554e-05,
-      "loss": 4.6311,
-      "step": 46189
-    },
-    {
-      "epoch": 24.08865710560626,
-      "grad_norm": 1.6273276805877686,
-      "learning_rate": 5.409748743718593e-05,
-      "loss": 5.1533,
-      "step": 46190
-    },
-    {
-      "epoch": 24.08917861799218,
-      "grad_norm": 1.6878405809402466,
-      "learning_rate": 5.40964824120603e-05,
-      "loss": 5.1714,
-      "step": 46191
-    },
-    {
-      "epoch": 24.089700130378098,
-      "grad_norm": 1.6074060201644897,
-      "learning_rate": 5.409547738693468e-05,
-      "loss": 4.9237,
-      "step": 46192
-    },
-    {
-      "epoch": 24.090221642764014,
-      "grad_norm": 1.5225616693496704,
-      "learning_rate": 5.4094472361809045e-05,
-      "loss": 4.9143,
-      "step": 46193
-    },
-    {
-      "epoch": 24.090743155149934,
-      "grad_norm": 1.6235140562057495,
-      "learning_rate": 5.409346733668342e-05,
-      "loss": 5.3798,
-      "step": 46194
-    },
-    {
-      "epoch": 24.091264667535853,
-      "grad_norm": 1.5328065156936646,
-      "learning_rate": 5.409246231155779e-05,
-      "loss": 5.1268,
-      "step": 46195
-    },
-    {
-      "epoch": 24.091786179921773,
-      "grad_norm": 1.5106827020645142,
-      "learning_rate": 5.4091457286432165e-05,
-      "loss": 5.3365,
-      "step": 46196
-    },
-    {
-      "epoch": 24.092307692307692,
-      "grad_norm": 1.5720000267028809,
-      "learning_rate": 5.4090452261306536e-05,
-      "loss": 5.1118,
-      "step": 46197
-    },
-    {
-      "epoch": 24.09282920469361,
-      "grad_norm": 1.6293867826461792,
-      "learning_rate": 5.4089447236180913e-05,
-      "loss": 4.8491,
-      "step": 46198
-    },
-    {
-      "epoch": 24.09335071707953,
-      "grad_norm": 1.4419485330581665,
-      "learning_rate": 5.408844221105528e-05,
-      "loss": 5.4614,
-      "step": 46199
-    },
-    {
-      "epoch": 24.09387222946545,
-      "grad_norm": 1.5313202142715454,
-      "learning_rate": 5.408743718592965e-05,
-      "loss": 5.6681,
-      "step": 46200
-    },
-    {
-      "epoch": 24.09439374185137,
-      "grad_norm": 1.5183393955230713,
-      "learning_rate": 5.4086432160804027e-05,
-      "loss": 5.4697,
-      "step": 46201
-    },
-    {
-      "epoch": 24.09491525423729,
-      "grad_norm": 1.4516030550003052,
-      "learning_rate": 5.408542713567839e-05,
-      "loss": 5.5997,
-      "step": 46202
-    },
-    {
-      "epoch": 24.09543676662321,
-      "grad_norm": 1.6628334522247314,
-      "learning_rate": 5.408442211055277e-05,
-      "loss": 5.2474,
-      "step": 46203
-    },
-    {
-      "epoch": 24.09595827900913,
-      "grad_norm": 1.6739684343338013,
-      "learning_rate": 5.408341708542713e-05,
-      "loss": 4.8565,
-      "step": 46204
-    },
-    {
-      "epoch": 24.096479791395044,
-      "grad_norm": 1.6147239208221436,
-      "learning_rate": 5.408241206030151e-05,
-      "loss": 4.6339,
-      "step": 46205
-    },
-    {
-      "epoch": 24.097001303780964,
-      "grad_norm": 1.5889710187911987,
-      "learning_rate": 5.408140703517588e-05,
-      "loss": 4.8023,
-      "step": 46206
-    },
-    {
-      "epoch": 24.097522816166883,
-      "grad_norm": 1.5924924612045288,
-      "learning_rate": 5.408040201005026e-05,
-      "loss": 5.337,
-      "step": 46207
-    },
-    {
-      "epoch": 24.098044328552803,
-      "grad_norm": 1.4940680265426636,
-      "learning_rate": 5.4079396984924624e-05,
-      "loss": 5.545,
-      "step": 46208
-    },
-    {
-      "epoch": 24.098565840938722,
-      "grad_norm": 1.614648461341858,
-      "learning_rate": 5.4078391959799e-05,
-      "loss": 5.076,
-      "step": 46209
-    },
-    {
-      "epoch": 24.09908735332464,
-      "grad_norm": 1.5647549629211426,
-      "learning_rate": 5.4077386934673366e-05,
-      "loss": 5.2867,
-      "step": 46210
-    },
-    {
-      "epoch": 24.09960886571056,
-      "grad_norm": 1.614320993423462,
-      "learning_rate": 5.4076381909547744e-05,
-      "loss": 4.7765,
-      "step": 46211
-    },
-    {
-      "epoch": 24.10013037809648,
-      "grad_norm": 1.981935739517212,
-      "learning_rate": 5.4075376884422115e-05,
-      "loss": 5.3463,
-      "step": 46212
-    },
-    {
-      "epoch": 24.1006518904824,
-      "grad_norm": 1.599616527557373,
-      "learning_rate": 5.407437185929648e-05,
-      "loss": 5.1578,
-      "step": 46213
-    },
-    {
-      "epoch": 24.10117340286832,
-      "grad_norm": 1.4636040925979614,
-      "learning_rate": 5.407336683417086e-05,
-      "loss": 5.2045,
-      "step": 46214
-    },
-    {
-      "epoch": 24.10169491525424,
-      "grad_norm": 1.4532266855239868,
-      "learning_rate": 5.407236180904522e-05,
-      "loss": 5.7606,
-      "step": 46215
-    },
-    {
-      "epoch": 24.102216427640155,
-      "grad_norm": 1.483828067779541,
-      "learning_rate": 5.40713567839196e-05,
-      "loss": 5.6761,
-      "step": 46216
-    },
-    {
-      "epoch": 24.102737940026074,
-      "grad_norm": 1.6140031814575195,
-      "learning_rate": 5.407035175879397e-05,
-      "loss": 4.9897,
-      "step": 46217
-    },
-    {
-      "epoch": 24.103259452411994,
-      "grad_norm": 1.6210917234420776,
-      "learning_rate": 5.406934673366835e-05,
-      "loss": 4.5014,
-      "step": 46218
-    },
-    {
-      "epoch": 24.103780964797913,
-      "grad_norm": 1.536994218826294,
-      "learning_rate": 5.406834170854271e-05,
-      "loss": 5.0039,
-      "step": 46219
-    },
-    {
-      "epoch": 24.104302477183833,
-      "grad_norm": 1.5318061113357544,
-      "learning_rate": 5.406733668341709e-05,
-      "loss": 5.2248,
-      "step": 46220
-    },
-    {
-      "epoch": 24.104823989569752,
-      "grad_norm": 1.5586416721343994,
-      "learning_rate": 5.406633165829146e-05,
-      "loss": 5.1045,
-      "step": 46221
-    },
-    {
-      "epoch": 24.10534550195567,
-      "grad_norm": 1.5919859409332275,
-      "learning_rate": 5.406532663316584e-05,
-      "loss": 5.5821,
-      "step": 46222
-    },
-    {
-      "epoch": 24.10586701434159,
-      "grad_norm": 1.4573848247528076,
-      "learning_rate": 5.40643216080402e-05,
-      "loss": 4.99,
-      "step": 46223
-    },
-    {
-      "epoch": 24.10638852672751,
-      "grad_norm": 1.5790798664093018,
-      "learning_rate": 5.406331658291458e-05,
-      "loss": 5.1041,
-      "step": 46224
-    },
-    {
-      "epoch": 24.10691003911343,
-      "grad_norm": 1.5711389780044556,
-      "learning_rate": 5.4062311557788945e-05,
-      "loss": 5.1393,
-      "step": 46225
-    },
-    {
-      "epoch": 24.10743155149935,
-      "grad_norm": 1.6000858545303345,
-      "learning_rate": 5.4061306532663316e-05,
-      "loss": 4.9326,
-      "step": 46226
-    },
-    {
-      "epoch": 24.10795306388527,
-      "grad_norm": 1.4644230604171753,
-      "learning_rate": 5.4060301507537694e-05,
-      "loss": 5.8591,
-      "step": 46227
-    },
-    {
-      "epoch": 24.108474576271185,
-      "grad_norm": 1.475772500038147,
-      "learning_rate": 5.405929648241206e-05,
-      "loss": 5.3003,
-      "step": 46228
-    },
-    {
-      "epoch": 24.108996088657104,
-      "grad_norm": 1.5789530277252197,
-      "learning_rate": 5.4058291457286436e-05,
-      "loss": 4.184,
-      "step": 46229
-    },
-    {
-      "epoch": 24.109517601043024,
-      "grad_norm": 1.614521861076355,
-      "learning_rate": 5.40572864321608e-05,
-      "loss": 4.7985,
-      "step": 46230
-    },
-    {
-      "epoch": 24.110039113428943,
-      "grad_norm": 1.458031177520752,
-      "learning_rate": 5.405628140703518e-05,
-      "loss": 5.5268,
-      "step": 46231
-    },
-    {
-      "epoch": 24.110560625814863,
-      "grad_norm": 1.5837531089782715,
-      "learning_rate": 5.405527638190955e-05,
-      "loss": 4.8697,
-      "step": 46232
-    },
-    {
-      "epoch": 24.111082138200782,
-      "grad_norm": 1.5353851318359375,
-      "learning_rate": 5.405427135678393e-05,
-      "loss": 4.8964,
-      "step": 46233
-    },
-    {
-      "epoch": 24.1116036505867,
-      "grad_norm": 1.5475854873657227,
-      "learning_rate": 5.405326633165829e-05,
-      "loss": 5.51,
-      "step": 46234
-    },
-    {
-      "epoch": 24.11212516297262,
-      "grad_norm": 1.4811782836914062,
-      "learning_rate": 5.405226130653267e-05,
-      "loss": 5.159,
-      "step": 46235
-    },
-    {
-      "epoch": 24.11264667535854,
-      "grad_norm": 1.460315227508545,
-      "learning_rate": 5.4051256281407034e-05,
-      "loss": 5.527,
-      "step": 46236
-    },
-    {
-      "epoch": 24.11316818774446,
-      "grad_norm": 1.568555474281311,
-      "learning_rate": 5.405025125628141e-05,
-      "loss": 5.1229,
-      "step": 46237
-    },
-    {
-      "epoch": 24.11368970013038,
-      "grad_norm": 1.5880444049835205,
-      "learning_rate": 5.404924623115578e-05,
-      "loss": 5.2115,
-      "step": 46238
-    },
-    {
-      "epoch": 24.1142112125163,
-      "grad_norm": 1.5320460796356201,
-      "learning_rate": 5.404824120603015e-05,
-      "loss": 4.9427,
-      "step": 46239
-    },
-    {
-      "epoch": 24.114732724902215,
-      "grad_norm": 1.5335010290145874,
-      "learning_rate": 5.4047236180904525e-05,
-      "loss": 5.3671,
-      "step": 46240
-    },
-    {
-      "epoch": 24.115254237288134,
-      "grad_norm": 1.556148886680603,
-      "learning_rate": 5.4046231155778896e-05,
-      "loss": 5.3116,
-      "step": 46241
-    },
-    {
-      "epoch": 24.115775749674054,
-      "grad_norm": 1.4792262315750122,
-      "learning_rate": 5.4045226130653274e-05,
-      "loss": 4.9028,
-      "step": 46242
-    },
-    {
-      "epoch": 24.116297262059973,
-      "grad_norm": 1.5185506343841553,
-      "learning_rate": 5.404422110552764e-05,
-      "loss": 5.2637,
-      "step": 46243
-    },
-    {
-      "epoch": 24.116818774445893,
-      "grad_norm": 1.428697109222412,
-      "learning_rate": 5.4043216080402016e-05,
-      "loss": 5.7535,
-      "step": 46244
-    },
-    {
-      "epoch": 24.117340286831812,
-      "grad_norm": 1.5568270683288574,
-      "learning_rate": 5.404221105527638e-05,
-      "loss": 5.3336,
-      "step": 46245
-    },
-    {
-      "epoch": 24.11786179921773,
-      "grad_norm": 1.5282292366027832,
-      "learning_rate": 5.404120603015076e-05,
-      "loss": 5.4707,
-      "step": 46246
-    },
-    {
-      "epoch": 24.11838331160365,
-      "grad_norm": 1.4758955240249634,
-      "learning_rate": 5.404020100502513e-05,
-      "loss": 5.6032,
-      "step": 46247
-    },
-    {
-      "epoch": 24.11890482398957,
-      "grad_norm": 1.6480367183685303,
-      "learning_rate": 5.403919597989951e-05,
-      "loss": 5.0456,
-      "step": 46248
-    },
-    {
-      "epoch": 24.11942633637549,
-      "grad_norm": 1.4414865970611572,
-      "learning_rate": 5.403819095477387e-05,
-      "loss": 5.4307,
-      "step": 46249
-    },
-    {
-      "epoch": 24.11994784876141,
-      "grad_norm": 1.6009303331375122,
-      "learning_rate": 5.403718592964825e-05,
-      "loss": 4.8828,
-      "step": 46250
-    },
-    {
-      "epoch": 24.12046936114733,
-      "grad_norm": 1.6142356395721436,
-      "learning_rate": 5.403618090452261e-05,
-      "loss": 5.1711,
-      "step": 46251
-    },
-    {
-      "epoch": 24.120990873533245,
-      "grad_norm": 1.5713139772415161,
-      "learning_rate": 5.4035175879396984e-05,
-      "loss": 4.7361,
-      "step": 46252
-    },
-    {
-      "epoch": 24.121512385919164,
-      "grad_norm": 1.5003362894058228,
-      "learning_rate": 5.403417085427136e-05,
-      "loss": 4.7122,
-      "step": 46253
-    },
-    {
-      "epoch": 24.122033898305084,
-      "grad_norm": 1.5010409355163574,
-      "learning_rate": 5.4033165829145726e-05,
-      "loss": 5.7927,
-      "step": 46254
-    },
-    {
-      "epoch": 24.122555410691003,
-      "grad_norm": 1.4410560131072998,
-      "learning_rate": 5.4032160804020104e-05,
-      "loss": 5.218,
-      "step": 46255
-    },
-    {
-      "epoch": 24.123076923076923,
-      "grad_norm": 1.5949164628982544,
-      "learning_rate": 5.403115577889447e-05,
-      "loss": 4.7966,
-      "step": 46256
-    },
-    {
-      "epoch": 24.123598435462842,
-      "grad_norm": 1.7690244913101196,
-      "learning_rate": 5.4030150753768846e-05,
-      "loss": 4.9925,
-      "step": 46257
-    },
-    {
-      "epoch": 24.12411994784876,
-      "grad_norm": 1.5081486701965332,
-      "learning_rate": 5.402914572864322e-05,
-      "loss": 5.5065,
-      "step": 46258
-    },
-    {
-      "epoch": 24.12464146023468,
-      "grad_norm": 1.5617952346801758,
-      "learning_rate": 5.4028140703517595e-05,
-      "loss": 5.4693,
-      "step": 46259
-    },
-    {
-      "epoch": 24.1251629726206,
-      "grad_norm": 1.5824449062347412,
-      "learning_rate": 5.402713567839196e-05,
-      "loss": 5.2092,
-      "step": 46260
-    },
-    {
-      "epoch": 24.12568448500652,
-      "grad_norm": 1.5250182151794434,
-      "learning_rate": 5.402613065326634e-05,
-      "loss": 5.5775,
-      "step": 46261
-    },
-    {
-      "epoch": 24.12620599739244,
-      "grad_norm": 1.5588918924331665,
-      "learning_rate": 5.402512562814071e-05,
-      "loss": 4.7993,
-      "step": 46262
-    },
-    {
-      "epoch": 24.12672750977836,
-      "grad_norm": 1.4875248670578003,
-      "learning_rate": 5.4024120603015086e-05,
-      "loss": 5.0728,
-      "step": 46263
-    },
-    {
-      "epoch": 24.127249022164275,
-      "grad_norm": 1.5333912372589111,
-      "learning_rate": 5.402311557788945e-05,
-      "loss": 5.4437,
-      "step": 46264
-    },
-    {
-      "epoch": 24.127770534550194,
-      "grad_norm": 1.6242471933364868,
-      "learning_rate": 5.402211055276383e-05,
-      "loss": 4.8073,
-      "step": 46265
-    },
-    {
-      "epoch": 24.128292046936114,
-      "grad_norm": 1.5310648679733276,
-      "learning_rate": 5.402110552763819e-05,
-      "loss": 5.73,
-      "step": 46266
-    },
-    {
-      "epoch": 24.128813559322033,
-      "grad_norm": 1.5524113178253174,
-      "learning_rate": 5.4020100502512563e-05,
-      "loss": 5.2066,
-      "step": 46267
-    },
-    {
-      "epoch": 24.129335071707953,
-      "grad_norm": 1.5742655992507935,
-      "learning_rate": 5.401909547738694e-05,
-      "loss": 5.1954,
-      "step": 46268
-    },
-    {
-      "epoch": 24.129856584093872,
-      "grad_norm": 1.5146701335906982,
-      "learning_rate": 5.4018090452261305e-05,
-      "loss": 5.2763,
-      "step": 46269
-    },
-    {
-      "epoch": 24.13037809647979,
-      "grad_norm": 1.447706699371338,
-      "learning_rate": 5.401708542713568e-05,
-      "loss": 5.3817,
-      "step": 46270
-    },
-    {
-      "epoch": 24.13089960886571,
-      "grad_norm": 1.6465013027191162,
-      "learning_rate": 5.401608040201005e-05,
-      "loss": 4.3353,
-      "step": 46271
-    },
-    {
-      "epoch": 24.13142112125163,
-      "grad_norm": 1.6284083127975464,
-      "learning_rate": 5.4015075376884425e-05,
-      "loss": 4.835,
-      "step": 46272
-    },
-    {
-      "epoch": 24.13194263363755,
-      "grad_norm": 1.5540298223495483,
-      "learning_rate": 5.4014070351758796e-05,
-      "loss": 4.9539,
-      "step": 46273
-    },
-    {
-      "epoch": 24.13246414602347,
-      "grad_norm": 1.6565953493118286,
-      "learning_rate": 5.4013065326633174e-05,
-      "loss": 4.8887,
-      "step": 46274
-    },
-    {
-      "epoch": 24.13298565840939,
-      "grad_norm": 1.5295827388763428,
-      "learning_rate": 5.401206030150754e-05,
-      "loss": 5.5523,
-      "step": 46275
-    },
-    {
-      "epoch": 24.133507170795305,
-      "grad_norm": 1.649246096611023,
-      "learning_rate": 5.4011055276381916e-05,
-      "loss": 4.9998,
-      "step": 46276
-    },
-    {
-      "epoch": 24.134028683181224,
-      "grad_norm": 1.4382654428482056,
-      "learning_rate": 5.401005025125628e-05,
-      "loss": 5.6922,
-      "step": 46277
-    },
-    {
-      "epoch": 24.134550195567144,
-      "grad_norm": 1.6374479532241821,
-      "learning_rate": 5.400904522613066e-05,
-      "loss": 5.0115,
-      "step": 46278
-    },
-    {
-      "epoch": 24.135071707953063,
-      "grad_norm": 1.6577038764953613,
-      "learning_rate": 5.400804020100503e-05,
-      "loss": 4.8425,
-      "step": 46279
-    },
-    {
-      "epoch": 24.135593220338983,
-      "grad_norm": 1.469879150390625,
-      "learning_rate": 5.4007035175879394e-05,
-      "loss": 5.7147,
-      "step": 46280
-    },
-    {
-      "epoch": 24.136114732724902,
-      "grad_norm": 1.6344963312149048,
-      "learning_rate": 5.400603015075377e-05,
-      "loss": 4.824,
-      "step": 46281
-    },
-    {
-      "epoch": 24.13663624511082,
-      "grad_norm": 1.4712461233139038,
-      "learning_rate": 5.4005025125628136e-05,
-      "loss": 5.3909,
-      "step": 46282
-    },
-    {
-      "epoch": 24.13715775749674,
-      "grad_norm": 1.5455083847045898,
-      "learning_rate": 5.4004020100502514e-05,
-      "loss": 5.1313,
-      "step": 46283
-    },
-    {
-      "epoch": 24.13767926988266,
-      "grad_norm": 1.5361002683639526,
-      "learning_rate": 5.4003015075376885e-05,
-      "loss": 4.8464,
-      "step": 46284
-    },
-    {
-      "epoch": 24.13820078226858,
-      "grad_norm": 1.6290690898895264,
-      "learning_rate": 5.400201005025126e-05,
-      "loss": 4.7786,
-      "step": 46285
-    },
-    {
-      "epoch": 24.1387222946545,
-      "grad_norm": 1.540895938873291,
-      "learning_rate": 5.400100502512563e-05,
-      "loss": 5.3719,
-      "step": 46286
-    },
-    {
-      "epoch": 24.13924380704042,
-      "grad_norm": 1.4879103899002075,
-      "learning_rate": 5.4000000000000005e-05,
-      "loss": 5.0299,
-      "step": 46287
-    },
-    {
-      "epoch": 24.139765319426335,
-      "grad_norm": 1.6355894804000854,
-      "learning_rate": 5.3998994974874376e-05,
-      "loss": 5.6305,
-      "step": 46288
-    },
-    {
-      "epoch": 24.140286831812254,
-      "grad_norm": 1.5325566530227661,
-      "learning_rate": 5.3997989949748754e-05,
-      "loss": 5.5452,
-      "step": 46289
-    },
-    {
-      "epoch": 24.140808344198174,
-      "grad_norm": 1.5134621858596802,
-      "learning_rate": 5.399698492462312e-05,
-      "loss": 5.4308,
-      "step": 46290
-    },
-    {
-      "epoch": 24.141329856584093,
-      "grad_norm": 1.4936336278915405,
-      "learning_rate": 5.3995979899497496e-05,
-      "loss": 5.3204,
-      "step": 46291
-    },
-    {
-      "epoch": 24.141851368970013,
-      "grad_norm": 1.5561835765838623,
-      "learning_rate": 5.399497487437186e-05,
-      "loss": 4.6903,
-      "step": 46292
-    },
-    {
-      "epoch": 24.142372881355932,
-      "grad_norm": 1.465512990951538,
-      "learning_rate": 5.399396984924623e-05,
-      "loss": 5.803,
-      "step": 46293
-    },
-    {
-      "epoch": 24.14289439374185,
-      "grad_norm": 1.5997511148452759,
-      "learning_rate": 5.399296482412061e-05,
-      "loss": 5.2638,
-      "step": 46294
-    },
-    {
-      "epoch": 24.14341590612777,
-      "grad_norm": 1.7532638311386108,
-      "learning_rate": 5.399195979899497e-05,
-      "loss": 4.8933,
-      "step": 46295
-    },
-    {
-      "epoch": 24.14393741851369,
-      "grad_norm": 1.5372366905212402,
-      "learning_rate": 5.399095477386935e-05,
-      "loss": 5.2893,
-      "step": 46296
-    },
-    {
-      "epoch": 24.14445893089961,
-      "grad_norm": 1.59455144405365,
-      "learning_rate": 5.3989949748743715e-05,
-      "loss": 5.3657,
-      "step": 46297
-    },
-    {
-      "epoch": 24.14498044328553,
-      "grad_norm": 1.5500688552856445,
-      "learning_rate": 5.398894472361809e-05,
-      "loss": 4.9478,
-      "step": 46298
-    },
-    {
-      "epoch": 24.14550195567145,
-      "grad_norm": 1.5390684604644775,
-      "learning_rate": 5.3987939698492464e-05,
-      "loss": 5.3699,
-      "step": 46299
-    },
-    {
-      "epoch": 24.146023468057365,
-      "grad_norm": 1.537137746810913,
-      "learning_rate": 5.398693467336684e-05,
-      "loss": 4.9374,
-      "step": 46300
-    },
-    {
-      "epoch": 24.146544980443284,
-      "grad_norm": 1.5696587562561035,
-      "learning_rate": 5.3985929648241206e-05,
-      "loss": 5.4639,
-      "step": 46301
-    },
-    {
-      "epoch": 24.147066492829204,
-      "grad_norm": 1.6008490324020386,
-      "learning_rate": 5.3984924623115584e-05,
-      "loss": 4.2608,
-      "step": 46302
-    },
-    {
-      "epoch": 24.147588005215123,
-      "grad_norm": 1.6336522102355957,
-      "learning_rate": 5.398391959798995e-05,
-      "loss": 4.8231,
-      "step": 46303
-    },
-    {
-      "epoch": 24.148109517601043,
-      "grad_norm": 1.5787789821624756,
-      "learning_rate": 5.3982914572864326e-05,
-      "loss": 5.1621,
-      "step": 46304
-    },
-    {
-      "epoch": 24.148631029986962,
-      "grad_norm": 1.6193997859954834,
-      "learning_rate": 5.39819095477387e-05,
-      "loss": 4.949,
-      "step": 46305
-    },
-    {
-      "epoch": 24.14915254237288,
-      "grad_norm": 1.558630108833313,
-      "learning_rate": 5.398090452261306e-05,
-      "loss": 5.5907,
-      "step": 46306
-    },
-    {
-      "epoch": 24.1496740547588,
-      "grad_norm": 1.5566539764404297,
-      "learning_rate": 5.397989949748744e-05,
-      "loss": 4.7883,
-      "step": 46307
-    },
-    {
-      "epoch": 24.15019556714472,
-      "grad_norm": 1.4818211793899536,
-      "learning_rate": 5.397889447236181e-05,
-      "loss": 5.2953,
-      "step": 46308
-    },
-    {
-      "epoch": 24.15071707953064,
-      "grad_norm": 1.5886698961257935,
-      "learning_rate": 5.397788944723619e-05,
-      "loss": 5.3193,
-      "step": 46309
-    },
-    {
-      "epoch": 24.15123859191656,
-      "grad_norm": 1.498441219329834,
-      "learning_rate": 5.397688442211055e-05,
-      "loss": 4.8786,
-      "step": 46310
-    },
-    {
-      "epoch": 24.151760104302475,
-      "grad_norm": 1.6078786849975586,
-      "learning_rate": 5.397587939698493e-05,
-      "loss": 5.5112,
-      "step": 46311
-    },
-    {
-      "epoch": 24.152281616688395,
-      "grad_norm": 1.5618846416473389,
-      "learning_rate": 5.3974874371859295e-05,
-      "loss": 5.2562,
-      "step": 46312
-    },
-    {
-      "epoch": 24.152803129074314,
-      "grad_norm": 1.5169811248779297,
-      "learning_rate": 5.397386934673367e-05,
-      "loss": 5.2605,
-      "step": 46313
-    },
-    {
-      "epoch": 24.153324641460234,
-      "grad_norm": 1.534814715385437,
-      "learning_rate": 5.3972864321608043e-05,
-      "loss": 5.3092,
-      "step": 46314
-    },
-    {
-      "epoch": 24.153846153846153,
-      "grad_norm": 1.4739606380462646,
-      "learning_rate": 5.397185929648242e-05,
-      "loss": 4.8877,
-      "step": 46315
-    },
-    {
-      "epoch": 24.154367666232073,
-      "grad_norm": 1.7033811807632446,
-      "learning_rate": 5.3970854271356786e-05,
-      "loss": 4.6204,
-      "step": 46316
-    },
-    {
-      "epoch": 24.154889178617992,
-      "grad_norm": 1.3983705043792725,
-      "learning_rate": 5.3969849246231163e-05,
-      "loss": 5.5582,
-      "step": 46317
-    },
-    {
-      "epoch": 24.15541069100391,
-      "grad_norm": 1.5880297422409058,
-      "learning_rate": 5.396884422110553e-05,
-      "loss": 4.8663,
-      "step": 46318
-    },
-    {
-      "epoch": 24.15593220338983,
-      "grad_norm": 1.6449909210205078,
-      "learning_rate": 5.39678391959799e-05,
-      "loss": 5.05,
-      "step": 46319
-    },
-    {
-      "epoch": 24.15645371577575,
-      "grad_norm": 1.5571874380111694,
-      "learning_rate": 5.3966834170854277e-05,
-      "loss": 5.3514,
-      "step": 46320
-    },
-    {
-      "epoch": 24.15697522816167,
-      "grad_norm": 1.6356078386306763,
-      "learning_rate": 5.396582914572864e-05,
-      "loss": 4.5578,
-      "step": 46321
-    },
-    {
-      "epoch": 24.15749674054759,
-      "grad_norm": 1.519140362739563,
-      "learning_rate": 5.396482412060302e-05,
-      "loss": 4.9288,
-      "step": 46322
-    },
-    {
-      "epoch": 24.158018252933505,
-      "grad_norm": 1.564781904220581,
-      "learning_rate": 5.396381909547738e-05,
-      "loss": 4.9108,
-      "step": 46323
-    },
-    {
-      "epoch": 24.158539765319425,
-      "grad_norm": 1.5291354656219482,
-      "learning_rate": 5.396281407035176e-05,
-      "loss": 5.234,
-      "step": 46324
-    },
-    {
-      "epoch": 24.159061277705344,
-      "grad_norm": 1.5344220399856567,
-      "learning_rate": 5.396180904522613e-05,
-      "loss": 5.3909,
-      "step": 46325
-    },
-    {
-      "epoch": 24.159582790091264,
-      "grad_norm": 1.4855977296829224,
-      "learning_rate": 5.396080402010051e-05,
-      "loss": 5.3035,
-      "step": 46326
-    },
-    {
-      "epoch": 24.160104302477183,
-      "grad_norm": 1.4897464513778687,
-      "learning_rate": 5.3959798994974874e-05,
-      "loss": 5.3991,
-      "step": 46327
-    },
-    {
-      "epoch": 24.160625814863103,
-      "grad_norm": 1.5425324440002441,
-      "learning_rate": 5.395879396984925e-05,
-      "loss": 5.5859,
-      "step": 46328
-    },
-    {
-      "epoch": 24.161147327249022,
-      "grad_norm": 1.7401981353759766,
-      "learning_rate": 5.395778894472362e-05,
-      "loss": 5.0742,
-      "step": 46329
-    },
-    {
-      "epoch": 24.16166883963494,
-      "grad_norm": 1.5870991945266724,
-      "learning_rate": 5.3956783919598e-05,
-      "loss": 5.1538,
-      "step": 46330
-    },
-    {
-      "epoch": 24.16219035202086,
-      "grad_norm": 1.567669153213501,
-      "learning_rate": 5.3955778894472365e-05,
-      "loss": 5.1611,
-      "step": 46331
-    },
-    {
-      "epoch": 24.16271186440678,
-      "grad_norm": 1.5696760416030884,
-      "learning_rate": 5.395477386934673e-05,
-      "loss": 5.2454,
-      "step": 46332
-    },
-    {
-      "epoch": 24.1632333767927,
-      "grad_norm": 1.686755657196045,
-      "learning_rate": 5.395376884422111e-05,
-      "loss": 5.1962,
-      "step": 46333
-    },
-    {
-      "epoch": 24.16375488917862,
-      "grad_norm": 1.598188877105713,
-      "learning_rate": 5.395276381909548e-05,
-      "loss": 5.2327,
-      "step": 46334
-    },
-    {
-      "epoch": 24.164276401564535,
-      "grad_norm": 1.5255699157714844,
-      "learning_rate": 5.3951758793969856e-05,
-      "loss": 5.5893,
-      "step": 46335
-    },
-    {
-      "epoch": 24.164797913950455,
-      "grad_norm": 1.7398474216461182,
-      "learning_rate": 5.395075376884422e-05,
-      "loss": 5.2375,
-      "step": 46336
-    },
-    {
-      "epoch": 24.165319426336374,
-      "grad_norm": 1.4708869457244873,
-      "learning_rate": 5.39497487437186e-05,
-      "loss": 5.3966,
-      "step": 46337
-    },
-    {
-      "epoch": 24.165840938722294,
-      "grad_norm": 1.7286466360092163,
-      "learning_rate": 5.394874371859296e-05,
-      "loss": 5.0607,
-      "step": 46338
-    },
-    {
-      "epoch": 24.166362451108213,
-      "grad_norm": 1.5741547346115112,
-      "learning_rate": 5.394773869346734e-05,
-      "loss": 5.0348,
-      "step": 46339
-    },
-    {
-      "epoch": 24.166883963494133,
-      "grad_norm": 1.5429822206497192,
-      "learning_rate": 5.394673366834171e-05,
-      "loss": 5.6848,
-      "step": 46340
-    },
-    {
-      "epoch": 24.167405475880052,
-      "grad_norm": 1.378768801689148,
-      "learning_rate": 5.394572864321609e-05,
-      "loss": 4.9461,
-      "step": 46341
-    },
-    {
-      "epoch": 24.16792698826597,
-      "grad_norm": 1.6272180080413818,
-      "learning_rate": 5.394472361809045e-05,
-      "loss": 5.0955,
-      "step": 46342
-    },
-    {
-      "epoch": 24.16844850065189,
-      "grad_norm": 1.5709151029586792,
-      "learning_rate": 5.394371859296483e-05,
-      "loss": 4.9627,
-      "step": 46343
-    },
-    {
-      "epoch": 24.16897001303781,
-      "grad_norm": 1.6599886417388916,
-      "learning_rate": 5.3942713567839195e-05,
-      "loss": 4.9125,
-      "step": 46344
-    },
-    {
-      "epoch": 24.16949152542373,
-      "grad_norm": 1.4783902168273926,
-      "learning_rate": 5.3941708542713566e-05,
-      "loss": 5.5028,
-      "step": 46345
-    },
-    {
-      "epoch": 24.17001303780965,
-      "grad_norm": 1.7009825706481934,
-      "learning_rate": 5.3940703517587944e-05,
-      "loss": 5.2658,
-      "step": 46346
-    },
-    {
-      "epoch": 24.170534550195566,
-      "grad_norm": 1.6247358322143555,
-      "learning_rate": 5.393969849246231e-05,
-      "loss": 4.6903,
-      "step": 46347
-    },
-    {
-      "epoch": 24.171056062581485,
-      "grad_norm": 1.4544612169265747,
-      "learning_rate": 5.3938693467336686e-05,
-      "loss": 4.3902,
-      "step": 46348
-    },
-    {
-      "epoch": 24.171577574967404,
-      "grad_norm": 1.6079128980636597,
-      "learning_rate": 5.393768844221105e-05,
-      "loss": 4.8316,
-      "step": 46349
-    },
-    {
-      "epoch": 24.172099087353324,
-      "grad_norm": 1.5350109338760376,
-      "learning_rate": 5.393668341708543e-05,
-      "loss": 5.4547,
-      "step": 46350
-    },
-    {
-      "epoch": 24.172620599739243,
-      "grad_norm": 1.517871618270874,
-      "learning_rate": 5.39356783919598e-05,
-      "loss": 5.4808,
-      "step": 46351
-    },
-    {
-      "epoch": 24.173142112125163,
-      "grad_norm": 1.6508053541183472,
-      "learning_rate": 5.393467336683418e-05,
-      "loss": 4.9221,
-      "step": 46352
-    },
-    {
-      "epoch": 24.173663624511082,
-      "grad_norm": 1.4982821941375732,
-      "learning_rate": 5.393366834170854e-05,
-      "loss": 5.5118,
-      "step": 46353
-    },
-    {
-      "epoch": 24.174185136897,
-      "grad_norm": 1.518595576286316,
-      "learning_rate": 5.393266331658292e-05,
-      "loss": 5.7856,
-      "step": 46354
-    },
-    {
-      "epoch": 24.17470664928292,
-      "grad_norm": 1.6157419681549072,
-      "learning_rate": 5.393165829145729e-05,
-      "loss": 5.2395,
-      "step": 46355
-    },
-    {
-      "epoch": 24.17522816166884,
-      "grad_norm": 1.5971299409866333,
-      "learning_rate": 5.393065326633167e-05,
-      "loss": 4.818,
-      "step": 46356
-    },
-    {
-      "epoch": 24.17574967405476,
-      "grad_norm": 1.585545539855957,
-      "learning_rate": 5.392964824120603e-05,
-      "loss": 5.2571,
-      "step": 46357
-    },
-    {
-      "epoch": 24.17627118644068,
-      "grad_norm": 1.50228750705719,
-      "learning_rate": 5.392864321608041e-05,
-      "loss": 5.3955,
-      "step": 46358
-    },
-    {
-      "epoch": 24.176792698826596,
-      "grad_norm": 1.5857266187667847,
-      "learning_rate": 5.3927638190954775e-05,
-      "loss": 4.9964,
-      "step": 46359
-    },
-    {
-      "epoch": 24.177314211212515,
-      "grad_norm": 1.5820868015289307,
-      "learning_rate": 5.3926633165829146e-05,
-      "loss": 4.7745,
-      "step": 46360
-    },
-    {
-      "epoch": 24.177835723598434,
-      "grad_norm": 1.5530641078948975,
-      "learning_rate": 5.3925628140703524e-05,
-      "loss": 5.4098,
-      "step": 46361
-    },
-    {
-      "epoch": 24.178357235984354,
-      "grad_norm": 1.6045013666152954,
-      "learning_rate": 5.392462311557789e-05,
-      "loss": 5.2229,
-      "step": 46362
-    },
-    {
-      "epoch": 24.178878748370273,
-      "grad_norm": 1.6189947128295898,
-      "learning_rate": 5.3923618090452266e-05,
-      "loss": 4.9864,
-      "step": 46363
-    },
-    {
-      "epoch": 24.179400260756193,
-      "grad_norm": 1.69171941280365,
-      "learning_rate": 5.392261306532663e-05,
-      "loss": 4.4985,
-      "step": 46364
-    },
-    {
-      "epoch": 24.179921773142112,
-      "grad_norm": 1.6774797439575195,
-      "learning_rate": 5.392160804020101e-05,
-      "loss": 5.2277,
-      "step": 46365
-    },
-    {
-      "epoch": 24.180443285528032,
-      "grad_norm": 1.603043556213379,
-      "learning_rate": 5.392060301507538e-05,
-      "loss": 5.6007,
-      "step": 46366
-    },
-    {
-      "epoch": 24.18096479791395,
-      "grad_norm": 1.5529862642288208,
-      "learning_rate": 5.391959798994976e-05,
-      "loss": 5.1274,
-      "step": 46367
-    },
-    {
-      "epoch": 24.18148631029987,
-      "grad_norm": 1.6332619190216064,
-      "learning_rate": 5.391859296482412e-05,
-      "loss": 5.043,
-      "step": 46368
-    },
-    {
-      "epoch": 24.18200782268579,
-      "grad_norm": 1.625966191291809,
-      "learning_rate": 5.39175879396985e-05,
-      "loss": 5.7316,
-      "step": 46369
-    },
-    {
-      "epoch": 24.18252933507171,
-      "grad_norm": 1.6669849157333374,
-      "learning_rate": 5.391658291457286e-05,
-      "loss": 5.3203,
-      "step": 46370
-    },
-    {
-      "epoch": 24.183050847457626,
-      "grad_norm": 1.6942963600158691,
-      "learning_rate": 5.391557788944724e-05,
-      "loss": 4.5723,
-      "step": 46371
-    },
-    {
-      "epoch": 24.183572359843545,
-      "grad_norm": 1.5928059816360474,
-      "learning_rate": 5.391457286432161e-05,
-      "loss": 4.968,
-      "step": 46372
-    },
-    {
-      "epoch": 24.184093872229464,
-      "grad_norm": 1.6965670585632324,
-      "learning_rate": 5.3913567839195976e-05,
-      "loss": 4.8172,
-      "step": 46373
-    },
-    {
-      "epoch": 24.184615384615384,
-      "grad_norm": 1.5861245393753052,
-      "learning_rate": 5.3912562814070354e-05,
-      "loss": 5.1711,
-      "step": 46374
-    },
-    {
-      "epoch": 24.185136897001303,
-      "grad_norm": 1.5411503314971924,
-      "learning_rate": 5.3911557788944725e-05,
-      "loss": 5.3123,
-      "step": 46375
-    },
-    {
-      "epoch": 24.185658409387223,
-      "grad_norm": 1.5282597541809082,
-      "learning_rate": 5.39105527638191e-05,
-      "loss": 5.5071,
-      "step": 46376
-    },
-    {
-      "epoch": 24.186179921773142,
-      "grad_norm": 1.6094508171081543,
-      "learning_rate": 5.390954773869347e-05,
-      "loss": 4.9529,
-      "step": 46377
-    },
-    {
-      "epoch": 24.186701434159062,
-      "grad_norm": 1.552862524986267,
-      "learning_rate": 5.3908542713567845e-05,
-      "loss": 5.11,
-      "step": 46378
-    },
-    {
-      "epoch": 24.18722294654498,
-      "grad_norm": 1.5963221788406372,
-      "learning_rate": 5.390753768844221e-05,
-      "loss": 5.5186,
-      "step": 46379
-    },
-    {
-      "epoch": 24.1877444589309,
-      "grad_norm": 1.5504820346832275,
-      "learning_rate": 5.390653266331659e-05,
-      "loss": 5.1927,
-      "step": 46380
-    },
-    {
-      "epoch": 24.18826597131682,
-      "grad_norm": 1.5412381887435913,
-      "learning_rate": 5.390552763819096e-05,
-      "loss": 5.5401,
-      "step": 46381
-    },
-    {
-      "epoch": 24.18878748370274,
-      "grad_norm": 1.6276432275772095,
-      "learning_rate": 5.3904522613065336e-05,
-      "loss": 5.2388,
-      "step": 46382
-    },
-    {
-      "epoch": 24.189308996088656,
-      "grad_norm": 1.8161855936050415,
-      "learning_rate": 5.39035175879397e-05,
-      "loss": 4.5698,
-      "step": 46383
-    },
-    {
-      "epoch": 24.189830508474575,
-      "grad_norm": 1.5865027904510498,
-      "learning_rate": 5.390251256281408e-05,
-      "loss": 5.5825,
-      "step": 46384
-    },
-    {
-      "epoch": 24.190352020860495,
-      "grad_norm": 1.48198664188385,
-      "learning_rate": 5.390150753768844e-05,
-      "loss": 5.2726,
-      "step": 46385
-    },
-    {
-      "epoch": 24.190873533246414,
-      "grad_norm": 1.586185097694397,
-      "learning_rate": 5.3900502512562813e-05,
-      "loss": 5.1222,
-      "step": 46386
-    },
-    {
-      "epoch": 24.191395045632333,
-      "grad_norm": 1.6536496877670288,
-      "learning_rate": 5.389949748743719e-05,
-      "loss": 5.2579,
-      "step": 46387
-    },
-    {
-      "epoch": 24.191916558018253,
-      "grad_norm": 1.5486581325531006,
-      "learning_rate": 5.3898492462311556e-05,
-      "loss": 5.2792,
-      "step": 46388
-    },
-    {
-      "epoch": 24.192438070404172,
-      "grad_norm": 1.5157922506332397,
-      "learning_rate": 5.389748743718593e-05,
-      "loss": 5.3727,
-      "step": 46389
-    },
-    {
-      "epoch": 24.192959582790092,
-      "grad_norm": 1.6127725839614868,
-      "learning_rate": 5.38964824120603e-05,
-      "loss": 5.0878,
-      "step": 46390
-    },
-    {
-      "epoch": 24.19348109517601,
-      "grad_norm": 1.6107511520385742,
-      "learning_rate": 5.3895477386934675e-05,
-      "loss": 5.1785,
-      "step": 46391
-    },
-    {
-      "epoch": 24.19400260756193,
-      "grad_norm": 1.5164779424667358,
-      "learning_rate": 5.3894472361809046e-05,
-      "loss": 5.3219,
-      "step": 46392
-    },
-    {
-      "epoch": 24.19452411994785,
-      "grad_norm": 1.4946221113204956,
-      "learning_rate": 5.3893467336683424e-05,
-      "loss": 5.1246,
-      "step": 46393
-    },
-    {
-      "epoch": 24.195045632333766,
-      "grad_norm": 1.4637478590011597,
-      "learning_rate": 5.389246231155779e-05,
-      "loss": 5.2091,
-      "step": 46394
-    },
-    {
-      "epoch": 24.195567144719686,
-      "grad_norm": 1.597446084022522,
-      "learning_rate": 5.3891457286432166e-05,
-      "loss": 4.578,
-      "step": 46395
-    },
-    {
-      "epoch": 24.196088657105605,
-      "grad_norm": 1.6217025518417358,
-      "learning_rate": 5.389045226130654e-05,
-      "loss": 5.4475,
-      "step": 46396
-    },
-    {
-      "epoch": 24.196610169491525,
-      "grad_norm": 1.4581788778305054,
-      "learning_rate": 5.3889447236180915e-05,
-      "loss": 5.1782,
-      "step": 46397
-    },
-    {
-      "epoch": 24.197131681877444,
-      "grad_norm": 1.4902821779251099,
-      "learning_rate": 5.388844221105528e-05,
-      "loss": 5.6112,
-      "step": 46398
-    },
-    {
-      "epoch": 24.197653194263363,
-      "grad_norm": 1.575924277305603,
-      "learning_rate": 5.3887437185929644e-05,
-      "loss": 5.1034,
-      "step": 46399
-    },
-    {
-      "epoch": 24.198174706649283,
-      "grad_norm": 1.569297194480896,
-      "learning_rate": 5.388643216080402e-05,
-      "loss": 5.4409,
-      "step": 46400
-    },
-    {
-      "epoch": 24.198696219035202,
-      "grad_norm": 1.7097095251083374,
-      "learning_rate": 5.388542713567839e-05,
-      "loss": 5.2224,
-      "step": 46401
-    },
-    {
-      "epoch": 24.199217731421122,
-      "grad_norm": 1.5452251434326172,
-      "learning_rate": 5.388442211055277e-05,
-      "loss": 5.4659,
-      "step": 46402
-    },
-    {
-      "epoch": 24.19973924380704,
-      "grad_norm": 1.5139878988265991,
-      "learning_rate": 5.3883417085427135e-05,
-      "loss": 5.6455,
-      "step": 46403
-    },
-    {
-      "epoch": 24.20026075619296,
-      "grad_norm": 1.5644828081130981,
-      "learning_rate": 5.388241206030151e-05,
-      "loss": 5.2003,
-      "step": 46404
-    },
-    {
-      "epoch": 24.20078226857888,
-      "grad_norm": 1.5589828491210938,
-      "learning_rate": 5.388140703517588e-05,
-      "loss": 5.0927,
-      "step": 46405
-    },
-    {
-      "epoch": 24.201303780964796,
-      "grad_norm": 1.557769536972046,
-      "learning_rate": 5.3880402010050255e-05,
-      "loss": 4.8181,
-      "step": 46406
-    },
-    {
-      "epoch": 24.201825293350716,
-      "grad_norm": 1.5085036754608154,
-      "learning_rate": 5.3879396984924626e-05,
-      "loss": 5.2977,
-      "step": 46407
-    },
-    {
-      "epoch": 24.202346805736635,
-      "grad_norm": 1.5682294368743896,
-      "learning_rate": 5.3878391959799004e-05,
-      "loss": 4.6436,
-      "step": 46408
-    },
-    {
-      "epoch": 24.202868318122555,
-      "grad_norm": 1.6205408573150635,
-      "learning_rate": 5.387738693467337e-05,
-      "loss": 5.397,
-      "step": 46409
-    },
-    {
-      "epoch": 24.203389830508474,
-      "grad_norm": 1.5650115013122559,
-      "learning_rate": 5.3876381909547746e-05,
-      "loss": 5.3839,
-      "step": 46410
-    },
-    {
-      "epoch": 24.203911342894393,
-      "grad_norm": 1.536582589149475,
-      "learning_rate": 5.387537688442211e-05,
-      "loss": 5.0777,
-      "step": 46411
-    },
-    {
-      "epoch": 24.204432855280313,
-      "grad_norm": 1.6016658544540405,
-      "learning_rate": 5.387437185929648e-05,
-      "loss": 5.1283,
-      "step": 46412
-    },
-    {
-      "epoch": 24.204954367666232,
-      "grad_norm": 1.5407428741455078,
-      "learning_rate": 5.387336683417086e-05,
-      "loss": 5.3833,
-      "step": 46413
-    },
-    {
-      "epoch": 24.205475880052152,
-      "grad_norm": 1.4931832551956177,
-      "learning_rate": 5.387236180904522e-05,
-      "loss": 5.2158,
-      "step": 46414
-    },
-    {
-      "epoch": 24.20599739243807,
-      "grad_norm": 1.5440089702606201,
-      "learning_rate": 5.38713567839196e-05,
-      "loss": 5.5148,
-      "step": 46415
-    },
-    {
-      "epoch": 24.20651890482399,
-      "grad_norm": 1.483192801475525,
-      "learning_rate": 5.387035175879397e-05,
-      "loss": 5.3218,
-      "step": 46416
-    },
-    {
-      "epoch": 24.20704041720991,
-      "grad_norm": 1.5212008953094482,
-      "learning_rate": 5.386934673366835e-05,
-      "loss": 5.4369,
-      "step": 46417
-    },
-    {
-      "epoch": 24.207561929595826,
-      "grad_norm": 1.668747901916504,
-      "learning_rate": 5.3868341708542714e-05,
-      "loss": 5.1273,
-      "step": 46418
-    },
-    {
-      "epoch": 24.208083441981746,
-      "grad_norm": 1.6282336711883545,
-      "learning_rate": 5.386733668341709e-05,
-      "loss": 5.1495,
-      "step": 46419
-    },
-    {
-      "epoch": 24.208604954367665,
-      "grad_norm": 1.5587127208709717,
-      "learning_rate": 5.3866331658291456e-05,
-      "loss": 5.0995,
-      "step": 46420
-    },
-    {
-      "epoch": 24.209126466753585,
-      "grad_norm": 1.5617177486419678,
-      "learning_rate": 5.3865326633165834e-05,
-      "loss": 4.9212,
-      "step": 46421
-    },
-    {
-      "epoch": 24.209647979139504,
-      "grad_norm": 1.6100897789001465,
-      "learning_rate": 5.3864321608040205e-05,
-      "loss": 4.9351,
-      "step": 46422
-    },
-    {
-      "epoch": 24.210169491525424,
-      "grad_norm": 1.535861611366272,
-      "learning_rate": 5.386331658291458e-05,
-      "loss": 5.3917,
-      "step": 46423
-    },
-    {
-      "epoch": 24.210691003911343,
-      "grad_norm": 1.662685751914978,
-      "learning_rate": 5.386231155778895e-05,
-      "loss": 5.2119,
-      "step": 46424
-    },
-    {
-      "epoch": 24.211212516297262,
-      "grad_norm": 1.5320243835449219,
-      "learning_rate": 5.386130653266331e-05,
-      "loss": 5.2787,
-      "step": 46425
-    },
-    {
-      "epoch": 24.211734028683182,
-      "grad_norm": 1.4730128049850464,
-      "learning_rate": 5.386030150753769e-05,
-      "loss": 5.4908,
-      "step": 46426
-    },
-    {
-      "epoch": 24.2122555410691,
-      "grad_norm": 1.5467630624771118,
-      "learning_rate": 5.385929648241206e-05,
-      "loss": 5.4555,
-      "step": 46427
-    },
-    {
-      "epoch": 24.21277705345502,
-      "grad_norm": 1.5134776830673218,
-      "learning_rate": 5.385829145728644e-05,
-      "loss": 5.5516,
-      "step": 46428
-    },
-    {
-      "epoch": 24.21329856584094,
-      "grad_norm": 1.5399893522262573,
-      "learning_rate": 5.38572864321608e-05,
-      "loss": 4.8632,
-      "step": 46429
-    },
-    {
-      "epoch": 24.213820078226856,
-      "grad_norm": 1.6998852491378784,
-      "learning_rate": 5.385628140703518e-05,
-      "loss": 5.0927,
-      "step": 46430
-    },
-    {
-      "epoch": 24.214341590612776,
-      "grad_norm": 1.5616800785064697,
-      "learning_rate": 5.3855276381909545e-05,
-      "loss": 5.4029,
-      "step": 46431
-    },
-    {
-      "epoch": 24.214863102998695,
-      "grad_norm": 1.5536319017410278,
-      "learning_rate": 5.385427135678392e-05,
-      "loss": 4.8738,
-      "step": 46432
-    },
-    {
-      "epoch": 24.215384615384615,
-      "grad_norm": 1.5367240905761719,
-      "learning_rate": 5.3853266331658293e-05,
-      "loss": 5.1782,
-      "step": 46433
-    },
-    {
-      "epoch": 24.215906127770534,
-      "grad_norm": 1.7184617519378662,
-      "learning_rate": 5.385226130653267e-05,
-      "loss": 4.8078,
-      "step": 46434
-    },
-    {
-      "epoch": 24.216427640156454,
-      "grad_norm": 1.5036720037460327,
-      "learning_rate": 5.3851256281407036e-05,
-      "loss": 5.4638,
-      "step": 46435
-    },
-    {
-      "epoch": 24.216949152542373,
-      "grad_norm": 1.6218678951263428,
-      "learning_rate": 5.3850251256281413e-05,
-      "loss": 4.955,
-      "step": 46436
-    },
-    {
-      "epoch": 24.217470664928292,
-      "grad_norm": 1.5789510011672974,
-      "learning_rate": 5.384924623115578e-05,
-      "loss": 5.3132,
-      "step": 46437
-    },
-    {
-      "epoch": 24.217992177314212,
-      "grad_norm": 1.6274701356887817,
-      "learning_rate": 5.3848241206030156e-05,
-      "loss": 5.077,
-      "step": 46438
-    },
-    {
-      "epoch": 24.21851368970013,
-      "grad_norm": 1.6282942295074463,
-      "learning_rate": 5.3847236180904527e-05,
-      "loss": 5.1151,
-      "step": 46439
-    },
-    {
-      "epoch": 24.21903520208605,
-      "grad_norm": 1.5768746137619019,
-      "learning_rate": 5.384623115577889e-05,
-      "loss": 4.2336,
-      "step": 46440
-    },
-    {
-      "epoch": 24.21955671447197,
-      "grad_norm": 1.5219004154205322,
-      "learning_rate": 5.384522613065327e-05,
-      "loss": 5.3608,
-      "step": 46441
-    },
-    {
-      "epoch": 24.220078226857886,
-      "grad_norm": 1.5060242414474487,
-      "learning_rate": 5.384422110552764e-05,
-      "loss": 5.3703,
-      "step": 46442
-    },
-    {
-      "epoch": 24.220599739243806,
-      "grad_norm": 1.5818642377853394,
-      "learning_rate": 5.384321608040202e-05,
-      "loss": 5.349,
-      "step": 46443
-    },
-    {
-      "epoch": 24.221121251629725,
-      "grad_norm": 1.4774409532546997,
-      "learning_rate": 5.384221105527638e-05,
-      "loss": 5.2047,
-      "step": 46444
-    },
-    {
-      "epoch": 24.221642764015645,
-      "grad_norm": 1.5908737182617188,
-      "learning_rate": 5.384120603015076e-05,
-      "loss": 4.992,
-      "step": 46445
-    },
-    {
-      "epoch": 24.222164276401564,
-      "grad_norm": 1.5936996936798096,
-      "learning_rate": 5.3840201005025124e-05,
-      "loss": 4.9241,
-      "step": 46446
-    },
-    {
-      "epoch": 24.222685788787484,
-      "grad_norm": 1.633881688117981,
-      "learning_rate": 5.38391959798995e-05,
-      "loss": 4.6818,
-      "step": 46447
-    },
-    {
-      "epoch": 24.223207301173403,
-      "grad_norm": 1.5220927000045776,
-      "learning_rate": 5.383819095477387e-05,
-      "loss": 5.5926,
-      "step": 46448
-    },
-    {
-      "epoch": 24.223728813559323,
-      "grad_norm": 1.9805582761764526,
-      "learning_rate": 5.383718592964825e-05,
-      "loss": 4.2995,
-      "step": 46449
-    },
-    {
-      "epoch": 24.224250325945242,
-      "grad_norm": 1.5875651836395264,
-      "learning_rate": 5.3836180904522615e-05,
-      "loss": 5.7699,
-      "step": 46450
-    },
-    {
-      "epoch": 24.22477183833116,
-      "grad_norm": 1.5394421815872192,
-      "learning_rate": 5.383517587939699e-05,
-      "loss": 4.8898,
-      "step": 46451
-    },
-    {
-      "epoch": 24.22529335071708,
-      "grad_norm": 1.648266077041626,
-      "learning_rate": 5.383417085427136e-05,
-      "loss": 4.8772,
-      "step": 46452
-    },
-    {
-      "epoch": 24.225814863103,
-      "grad_norm": 1.443962812423706,
-      "learning_rate": 5.383316582914573e-05,
-      "loss": 5.7601,
-      "step": 46453
-    },
-    {
-      "epoch": 24.226336375488916,
-      "grad_norm": 1.5913971662521362,
-      "learning_rate": 5.3832160804020106e-05,
-      "loss": 4.7301,
-      "step": 46454
-    },
-    {
-      "epoch": 24.226857887874836,
-      "grad_norm": 1.5389988422393799,
-      "learning_rate": 5.383115577889447e-05,
-      "loss": 4.9215,
-      "step": 46455
-    },
-    {
-      "epoch": 24.227379400260755,
-      "grad_norm": 1.6058571338653564,
-      "learning_rate": 5.383015075376885e-05,
-      "loss": 4.9166,
-      "step": 46456
-    },
-    {
-      "epoch": 24.227900912646675,
-      "grad_norm": 1.5236693620681763,
-      "learning_rate": 5.382914572864321e-05,
-      "loss": 5.2888,
-      "step": 46457
-    },
-    {
-      "epoch": 24.228422425032594,
-      "grad_norm": 1.5186322927474976,
-      "learning_rate": 5.382814070351759e-05,
-      "loss": 5.3366,
-      "step": 46458
-    },
-    {
-      "epoch": 24.228943937418514,
-      "grad_norm": 1.619195818901062,
-      "learning_rate": 5.382713567839196e-05,
-      "loss": 4.7162,
-      "step": 46459
-    },
-    {
-      "epoch": 24.229465449804433,
-      "grad_norm": 1.6159449815750122,
-      "learning_rate": 5.382613065326634e-05,
-      "loss": 5.1724,
-      "step": 46460
-    },
-    {
-      "epoch": 24.229986962190353,
-      "grad_norm": 1.5176600217819214,
-      "learning_rate": 5.38251256281407e-05,
-      "loss": 5.188,
-      "step": 46461
-    },
-    {
-      "epoch": 24.230508474576272,
-      "grad_norm": 1.5957595109939575,
-      "learning_rate": 5.382412060301508e-05,
-      "loss": 5.3252,
-      "step": 46462
-    },
-    {
-      "epoch": 24.23102998696219,
-      "grad_norm": 1.6791824102401733,
-      "learning_rate": 5.382311557788945e-05,
-      "loss": 5.0806,
-      "step": 46463
-    },
-    {
-      "epoch": 24.23155149934811,
-      "grad_norm": 1.5334595441818237,
-      "learning_rate": 5.382211055276383e-05,
-      "loss": 5.2478,
-      "step": 46464
-    },
-    {
-      "epoch": 24.23207301173403,
-      "grad_norm": 1.6755985021591187,
-      "learning_rate": 5.3821105527638194e-05,
-      "loss": 5.2174,
-      "step": 46465
-    },
-    {
-      "epoch": 24.232594524119946,
-      "grad_norm": 1.5232495069503784,
-      "learning_rate": 5.382010050251256e-05,
-      "loss": 4.8776,
-      "step": 46466
-    },
-    {
-      "epoch": 24.233116036505866,
-      "grad_norm": 1.4794540405273438,
-      "learning_rate": 5.3819095477386936e-05,
-      "loss": 5.3278,
-      "step": 46467
-    },
-    {
-      "epoch": 24.233637548891785,
-      "grad_norm": 1.651078462600708,
-      "learning_rate": 5.381809045226131e-05,
-      "loss": 4.7419,
-      "step": 46468
-    },
-    {
-      "epoch": 24.234159061277705,
-      "grad_norm": 1.504529356956482,
-      "learning_rate": 5.3817085427135685e-05,
-      "loss": 5.0214,
-      "step": 46469
-    },
-    {
-      "epoch": 24.234680573663624,
-      "grad_norm": 1.5812921524047852,
-      "learning_rate": 5.381608040201005e-05,
-      "loss": 5.558,
-      "step": 46470
-    },
-    {
-      "epoch": 24.235202086049544,
-      "grad_norm": 1.766411304473877,
-      "learning_rate": 5.381507537688443e-05,
-      "loss": 4.633,
-      "step": 46471
-    },
-    {
-      "epoch": 24.235723598435463,
-      "grad_norm": 1.5372164249420166,
-      "learning_rate": 5.381407035175879e-05,
-      "loss": 5.0309,
-      "step": 46472
-    },
-    {
-      "epoch": 24.236245110821383,
-      "grad_norm": 1.4685935974121094,
-      "learning_rate": 5.381306532663317e-05,
-      "loss": 4.728,
-      "step": 46473
-    },
-    {
-      "epoch": 24.236766623207302,
-      "grad_norm": 1.6498103141784668,
-      "learning_rate": 5.381206030150754e-05,
-      "loss": 5.2901,
-      "step": 46474
-    },
-    {
-      "epoch": 24.23728813559322,
-      "grad_norm": 1.551617980003357,
-      "learning_rate": 5.381105527638192e-05,
-      "loss": 5.4879,
-      "step": 46475
-    },
-    {
-      "epoch": 24.23780964797914,
-      "grad_norm": 1.6688379049301147,
-      "learning_rate": 5.381005025125628e-05,
-      "loss": 4.9153,
-      "step": 46476
-    },
-    {
-      "epoch": 24.23833116036506,
-      "grad_norm": 1.5804507732391357,
-      "learning_rate": 5.380904522613066e-05,
-      "loss": 5.2222,
-      "step": 46477
-    },
-    {
-      "epoch": 24.238852672750976,
-      "grad_norm": 1.4998183250427246,
-      "learning_rate": 5.3808040201005025e-05,
-      "loss": 5.1247,
-      "step": 46478
-    },
-    {
-      "epoch": 24.239374185136896,
-      "grad_norm": 1.4556199312210083,
-      "learning_rate": 5.3807035175879396e-05,
-      "loss": 5.5402,
-      "step": 46479
-    },
-    {
-      "epoch": 24.239895697522815,
-      "grad_norm": 1.4645341634750366,
-      "learning_rate": 5.3806030150753774e-05,
-      "loss": 5.5644,
-      "step": 46480
-    },
-    {
-      "epoch": 24.240417209908735,
-      "grad_norm": 1.6064203977584839,
-      "learning_rate": 5.380502512562814e-05,
-      "loss": 4.8966,
-      "step": 46481
-    },
-    {
-      "epoch": 24.240938722294654,
-      "grad_norm": 1.659401297569275,
-      "learning_rate": 5.3804020100502516e-05,
-      "loss": 4.7466,
-      "step": 46482
-    },
-    {
-      "epoch": 24.241460234680574,
-      "grad_norm": 1.527489185333252,
-      "learning_rate": 5.380301507537689e-05,
-      "loss": 5.5247,
-      "step": 46483
-    },
-    {
-      "epoch": 24.241981747066493,
-      "grad_norm": 1.4719699621200562,
-      "learning_rate": 5.3802010050251265e-05,
-      "loss": 5.4873,
-      "step": 46484
-    },
-    {
-      "epoch": 24.242503259452413,
-      "grad_norm": 1.5057063102722168,
-      "learning_rate": 5.380100502512563e-05,
-      "loss": 5.1073,
-      "step": 46485
-    },
-    {
-      "epoch": 24.243024771838332,
-      "grad_norm": 1.6750850677490234,
-      "learning_rate": 5.380000000000001e-05,
-      "loss": 5.1175,
-      "step": 46486
-    },
-    {
-      "epoch": 24.24354628422425,
-      "grad_norm": 1.6372199058532715,
-      "learning_rate": 5.379899497487437e-05,
-      "loss": 4.9759,
-      "step": 46487
-    },
-    {
-      "epoch": 24.24406779661017,
-      "grad_norm": 1.6231390237808228,
-      "learning_rate": 5.379798994974875e-05,
-      "loss": 5.3047,
-      "step": 46488
-    },
-    {
-      "epoch": 24.24458930899609,
-      "grad_norm": 1.5272561311721802,
-      "learning_rate": 5.379698492462312e-05,
-      "loss": 5.3407,
-      "step": 46489
-    },
-    {
-      "epoch": 24.245110821382006,
-      "grad_norm": 1.548416018486023,
-      "learning_rate": 5.37959798994975e-05,
-      "loss": 5.126,
-      "step": 46490
-    },
-    {
-      "epoch": 24.245632333767926,
-      "grad_norm": 1.6990758180618286,
-      "learning_rate": 5.379497487437186e-05,
-      "loss": 4.5233,
-      "step": 46491
-    },
-    {
-      "epoch": 24.246153846153845,
-      "grad_norm": 1.5591294765472412,
-      "learning_rate": 5.3793969849246226e-05,
-      "loss": 5.5398,
-      "step": 46492
-    },
-    {
-      "epoch": 24.246675358539765,
-      "grad_norm": 1.5645341873168945,
-      "learning_rate": 5.3792964824120604e-05,
-      "loss": 5.5628,
-      "step": 46493
-    },
-    {
-      "epoch": 24.247196870925684,
-      "grad_norm": 1.5691587924957275,
-      "learning_rate": 5.3791959798994975e-05,
-      "loss": 5.3879,
-      "step": 46494
-    },
-    {
-      "epoch": 24.247718383311604,
-      "grad_norm": 1.6278003454208374,
-      "learning_rate": 5.379095477386935e-05,
-      "loss": 4.6923,
-      "step": 46495
-    },
-    {
-      "epoch": 24.248239895697523,
-      "grad_norm": 1.6052556037902832,
-      "learning_rate": 5.378994974874372e-05,
-      "loss": 5.4894,
-      "step": 46496
-    },
-    {
-      "epoch": 24.248761408083443,
-      "grad_norm": 1.5084013938903809,
-      "learning_rate": 5.3788944723618095e-05,
-      "loss": 5.5154,
-      "step": 46497
-    },
-    {
-      "epoch": 24.249282920469362,
-      "grad_norm": 1.6078141927719116,
-      "learning_rate": 5.378793969849246e-05,
-      "loss": 5.2446,
-      "step": 46498
-    },
-    {
-      "epoch": 24.24980443285528,
-      "grad_norm": 1.5285695791244507,
-      "learning_rate": 5.378693467336684e-05,
-      "loss": 5.4987,
-      "step": 46499
-    },
-    {
-      "epoch": 24.2503259452412,
-      "grad_norm": 1.524895429611206,
-      "learning_rate": 5.378592964824121e-05,
-      "loss": 5.1031,
-      "step": 46500
-    },
-    {
-      "epoch": 24.250847457627117,
-      "grad_norm": 1.5889954566955566,
-      "learning_rate": 5.3784924623115586e-05,
-      "loss": 4.9945,
-      "step": 46501
-    },
-    {
-      "epoch": 24.251368970013036,
-      "grad_norm": 1.5890040397644043,
-      "learning_rate": 5.378391959798995e-05,
-      "loss": 5.0514,
-      "step": 46502
-    },
-    {
-      "epoch": 24.251890482398956,
-      "grad_norm": 1.4910718202590942,
-      "learning_rate": 5.378291457286433e-05,
-      "loss": 5.0523,
-      "step": 46503
-    },
-    {
-      "epoch": 24.252411994784875,
-      "grad_norm": 1.5374865531921387,
-      "learning_rate": 5.37819095477387e-05,
-      "loss": 5.0909,
-      "step": 46504
-    },
-    {
-      "epoch": 24.252933507170795,
-      "grad_norm": 1.4775243997573853,
-      "learning_rate": 5.3780904522613063e-05,
-      "loss": 5.1644,
-      "step": 46505
-    },
-    {
-      "epoch": 24.253455019556714,
-      "grad_norm": 1.6279429197311401,
-      "learning_rate": 5.377989949748744e-05,
-      "loss": 5.4946,
-      "step": 46506
-    },
-    {
-      "epoch": 24.253976531942634,
-      "grad_norm": 1.4730029106140137,
-      "learning_rate": 5.3778894472361806e-05,
-      "loss": 5.4121,
-      "step": 46507
-    },
-    {
-      "epoch": 24.254498044328553,
-      "grad_norm": 1.51401948928833,
-      "learning_rate": 5.377788944723618e-05,
-      "loss": 5.1287,
-      "step": 46508
-    },
-    {
-      "epoch": 24.255019556714473,
-      "grad_norm": 1.630916953086853,
-      "learning_rate": 5.3776884422110554e-05,
-      "loss": 4.7815,
-      "step": 46509
-    },
-    {
-      "epoch": 24.255541069100392,
-      "grad_norm": 1.7061710357666016,
-      "learning_rate": 5.377587939698493e-05,
-      "loss": 4.9695,
-      "step": 46510
-    },
-    {
-      "epoch": 24.25606258148631,
-      "grad_norm": 1.7191863059997559,
-      "learning_rate": 5.3774874371859297e-05,
-      "loss": 4.8089,
-      "step": 46511
-    },
-    {
-      "epoch": 24.25658409387223,
-      "grad_norm": 1.546297311782837,
-      "learning_rate": 5.3773869346733674e-05,
-      "loss": 5.2331,
-      "step": 46512
-    },
-    {
-      "epoch": 24.257105606258147,
-      "grad_norm": 1.5563076734542847,
-      "learning_rate": 5.377286432160804e-05,
-      "loss": 5.3063,
-      "step": 46513
-    },
-    {
-      "epoch": 24.257627118644066,
-      "grad_norm": 1.601966142654419,
-      "learning_rate": 5.3771859296482416e-05,
-      "loss": 4.7513,
-      "step": 46514
-    },
-    {
-      "epoch": 24.258148631029986,
-      "grad_norm": 1.479156494140625,
-      "learning_rate": 5.377085427135679e-05,
-      "loss": 5.3641,
-      "step": 46515
-    },
-    {
-      "epoch": 24.258670143415905,
-      "grad_norm": 1.8991185426712036,
-      "learning_rate": 5.3769849246231165e-05,
-      "loss": 5.3831,
-      "step": 46516
-    },
-    {
-      "epoch": 24.259191655801825,
-      "grad_norm": 1.662251353263855,
-      "learning_rate": 5.376884422110553e-05,
-      "loss": 4.8394,
-      "step": 46517
-    },
-    {
-      "epoch": 24.259713168187744,
-      "grad_norm": 1.5951800346374512,
-      "learning_rate": 5.3767839195979894e-05,
-      "loss": 5.3197,
-      "step": 46518
-    },
-    {
-      "epoch": 24.260234680573664,
-      "grad_norm": 1.5974843502044678,
-      "learning_rate": 5.376683417085427e-05,
-      "loss": 5.0499,
-      "step": 46519
-    },
-    {
-      "epoch": 24.260756192959583,
-      "grad_norm": 1.5448158979415894,
-      "learning_rate": 5.376582914572864e-05,
-      "loss": 5.2154,
-      "step": 46520
-    },
-    {
-      "epoch": 24.261277705345503,
-      "grad_norm": 1.5328088998794556,
-      "learning_rate": 5.376482412060302e-05,
-      "loss": 5.4032,
-      "step": 46521
-    },
-    {
-      "epoch": 24.261799217731422,
-      "grad_norm": 1.557132601737976,
-      "learning_rate": 5.3763819095477385e-05,
-      "loss": 5.2291,
-      "step": 46522
-    },
-    {
-      "epoch": 24.26232073011734,
-      "grad_norm": 1.569765329360962,
-      "learning_rate": 5.376281407035176e-05,
-      "loss": 5.0498,
-      "step": 46523
-    },
-    {
-      "epoch": 24.26284224250326,
-      "grad_norm": 1.536246418952942,
-      "learning_rate": 5.376180904522613e-05,
-      "loss": 4.6585,
-      "step": 46524
-    },
-    {
-      "epoch": 24.263363754889177,
-      "grad_norm": 1.5787760019302368,
-      "learning_rate": 5.3760804020100505e-05,
-      "loss": 5.4537,
-      "step": 46525
-    },
-    {
-      "epoch": 24.263885267275096,
-      "grad_norm": 1.798959493637085,
-      "learning_rate": 5.3759798994974876e-05,
-      "loss": 5.227,
-      "step": 46526
-    },
-    {
-      "epoch": 24.264406779661016,
-      "grad_norm": 1.5277239084243774,
-      "learning_rate": 5.3758793969849254e-05,
-      "loss": 5.5021,
-      "step": 46527
-    },
-    {
-      "epoch": 24.264928292046935,
-      "grad_norm": 1.596158742904663,
-      "learning_rate": 5.375778894472362e-05,
-      "loss": 4.7899,
-      "step": 46528
-    },
-    {
-      "epoch": 24.265449804432855,
-      "grad_norm": 1.546547770500183,
-      "learning_rate": 5.3756783919597996e-05,
-      "loss": 4.9555,
-      "step": 46529
-    },
-    {
-      "epoch": 24.265971316818774,
-      "grad_norm": 1.5198386907577515,
-      "learning_rate": 5.375577889447237e-05,
-      "loss": 5.18,
-      "step": 46530
-    },
-    {
-      "epoch": 24.266492829204694,
-      "grad_norm": 1.5776407718658447,
-      "learning_rate": 5.3754773869346745e-05,
-      "loss": 5.1132,
-      "step": 46531
-    },
-    {
-      "epoch": 24.267014341590613,
-      "grad_norm": 1.5499372482299805,
-      "learning_rate": 5.375376884422111e-05,
-      "loss": 5.4776,
-      "step": 46532
-    },
-    {
-      "epoch": 24.267535853976533,
-      "grad_norm": 1.4727933406829834,
-      "learning_rate": 5.375276381909547e-05,
-      "loss": 5.1737,
-      "step": 46533
-    },
-    {
-      "epoch": 24.268057366362452,
-      "grad_norm": 1.5558862686157227,
-      "learning_rate": 5.375175879396985e-05,
-      "loss": 4.9744,
-      "step": 46534
-    },
-    {
-      "epoch": 24.26857887874837,
-      "grad_norm": 1.5459965467453003,
-      "learning_rate": 5.375075376884422e-05,
-      "loss": 5.5401,
-      "step": 46535
-    },
-    {
-      "epoch": 24.26910039113429,
-      "grad_norm": 1.41295325756073,
-      "learning_rate": 5.37497487437186e-05,
-      "loss": 5.1173,
-      "step": 46536
-    },
-    {
-      "epoch": 24.269621903520207,
-      "grad_norm": 1.6529011726379395,
-      "learning_rate": 5.3748743718592964e-05,
-      "loss": 5.5341,
-      "step": 46537
-    },
-    {
-      "epoch": 24.270143415906126,
-      "grad_norm": 1.530126929283142,
-      "learning_rate": 5.374773869346734e-05,
-      "loss": 4.4887,
-      "step": 46538
-    },
-    {
-      "epoch": 24.270664928292046,
-      "grad_norm": 1.585639238357544,
-      "learning_rate": 5.3746733668341706e-05,
-      "loss": 4.7209,
-      "step": 46539
-    },
-    {
-      "epoch": 24.271186440677965,
-      "grad_norm": 1.591572880744934,
-      "learning_rate": 5.3745728643216084e-05,
-      "loss": 4.658,
-      "step": 46540
-    },
-    {
-      "epoch": 24.271707953063885,
-      "grad_norm": 1.478265643119812,
-      "learning_rate": 5.3744723618090455e-05,
-      "loss": 4.9451,
-      "step": 46541
-    },
-    {
-      "epoch": 24.272229465449804,
-      "grad_norm": 1.565230369567871,
-      "learning_rate": 5.374371859296483e-05,
-      "loss": 5.2774,
-      "step": 46542
-    },
-    {
-      "epoch": 24.272750977835724,
-      "grad_norm": 1.4534052610397339,
-      "learning_rate": 5.37427135678392e-05,
-      "loss": 5.164,
-      "step": 46543
-    },
-    {
-      "epoch": 24.273272490221643,
-      "grad_norm": 1.60517418384552,
-      "learning_rate": 5.3741708542713575e-05,
-      "loss": 5.0051,
-      "step": 46544
-    },
-    {
-      "epoch": 24.273794002607563,
-      "grad_norm": 1.5081274509429932,
-      "learning_rate": 5.374070351758794e-05,
-      "loss": 4.9418,
-      "step": 46545
-    },
-    {
-      "epoch": 24.274315514993482,
-      "grad_norm": 1.5396013259887695,
-      "learning_rate": 5.373969849246231e-05,
-      "loss": 5.3833,
-      "step": 46546
-    },
-    {
-      "epoch": 24.2748370273794,
-      "grad_norm": 1.5851737260818481,
-      "learning_rate": 5.373869346733669e-05,
-      "loss": 5.0194,
-      "step": 46547
-    },
-    {
-      "epoch": 24.27535853976532,
-      "grad_norm": 1.6919569969177246,
-      "learning_rate": 5.373768844221105e-05,
-      "loss": 5.1377,
-      "step": 46548
-    },
-    {
-      "epoch": 24.275880052151237,
-      "grad_norm": 1.5245801210403442,
-      "learning_rate": 5.373668341708543e-05,
-      "loss": 5.4671,
-      "step": 46549
-    },
-    {
-      "epoch": 24.276401564537156,
-      "grad_norm": 1.650509238243103,
-      "learning_rate": 5.37356783919598e-05,
-      "loss": 5.125,
-      "step": 46550
-    },
-    {
-      "epoch": 24.276923076923076,
-      "grad_norm": 1.6186389923095703,
-      "learning_rate": 5.373467336683418e-05,
-      "loss": 5.0391,
-      "step": 46551
-    },
-    {
-      "epoch": 24.277444589308995,
-      "grad_norm": 1.5488924980163574,
-      "learning_rate": 5.3733668341708544e-05,
-      "loss": 5.4188,
-      "step": 46552
-    },
-    {
-      "epoch": 24.277966101694915,
-      "grad_norm": 1.623411774635315,
-      "learning_rate": 5.373266331658292e-05,
-      "loss": 5.4039,
-      "step": 46553
-    },
-    {
-      "epoch": 24.278487614080834,
-      "grad_norm": 1.573259949684143,
-      "learning_rate": 5.3731658291457286e-05,
-      "loss": 5.5853,
-      "step": 46554
-    },
-    {
-      "epoch": 24.279009126466754,
-      "grad_norm": 1.6182838678359985,
-      "learning_rate": 5.3730653266331663e-05,
-      "loss": 5.2829,
-      "step": 46555
-    },
-    {
-      "epoch": 24.279530638852673,
-      "grad_norm": 1.6129504442214966,
-      "learning_rate": 5.3729648241206034e-05,
-      "loss": 5.3202,
-      "step": 46556
-    },
-    {
-      "epoch": 24.280052151238593,
-      "grad_norm": 1.6232938766479492,
-      "learning_rate": 5.372864321608041e-05,
-      "loss": 5.0764,
-      "step": 46557
-    },
-    {
-      "epoch": 24.280573663624512,
-      "grad_norm": 1.7150017023086548,
-      "learning_rate": 5.3727638190954777e-05,
-      "loss": 4.3729,
-      "step": 46558
-    },
-    {
-      "epoch": 24.28109517601043,
-      "grad_norm": 1.663986325263977,
-      "learning_rate": 5.372663316582914e-05,
-      "loss": 5.3287,
-      "step": 46559
-    },
-    {
-      "epoch": 24.28161668839635,
-      "grad_norm": 1.5687485933303833,
-      "learning_rate": 5.372562814070352e-05,
-      "loss": 5.5443,
-      "step": 46560
-    },
-    {
-      "epoch": 24.282138200782267,
-      "grad_norm": 1.5422264337539673,
-      "learning_rate": 5.372462311557789e-05,
-      "loss": 4.3648,
-      "step": 46561
-    },
-    {
-      "epoch": 24.282659713168186,
-      "grad_norm": 1.5882489681243896,
-      "learning_rate": 5.372361809045227e-05,
-      "loss": 4.4674,
-      "step": 46562
-    },
-    {
-      "epoch": 24.283181225554106,
-      "grad_norm": 1.628889799118042,
-      "learning_rate": 5.372261306532663e-05,
-      "loss": 4.7572,
-      "step": 46563
-    },
-    {
-      "epoch": 24.283702737940025,
-      "grad_norm": 1.552642583847046,
-      "learning_rate": 5.372160804020101e-05,
-      "loss": 5.3804,
-      "step": 46564
-    },
-    {
-      "epoch": 24.284224250325945,
-      "grad_norm": 1.535701036453247,
-      "learning_rate": 5.3720603015075374e-05,
-      "loss": 5.547,
-      "step": 46565
-    },
-    {
-      "epoch": 24.284745762711864,
-      "grad_norm": 1.5851973295211792,
-      "learning_rate": 5.371959798994975e-05,
-      "loss": 5.3646,
-      "step": 46566
-    },
-    {
-      "epoch": 24.285267275097784,
-      "grad_norm": 1.5524189472198486,
-      "learning_rate": 5.371859296482412e-05,
-      "loss": 5.4781,
-      "step": 46567
-    },
-    {
-      "epoch": 24.285788787483703,
-      "grad_norm": 1.6161832809448242,
-      "learning_rate": 5.37175879396985e-05,
-      "loss": 5.0334,
-      "step": 46568
-    },
-    {
-      "epoch": 24.286310299869623,
-      "grad_norm": 1.5426084995269775,
-      "learning_rate": 5.3716582914572865e-05,
-      "loss": 5.3827,
-      "step": 46569
-    },
-    {
-      "epoch": 24.286831812255542,
-      "grad_norm": 1.6541011333465576,
-      "learning_rate": 5.371557788944724e-05,
-      "loss": 4.8641,
-      "step": 46570
-    },
-    {
-      "epoch": 24.28735332464146,
-      "grad_norm": 1.6597360372543335,
-      "learning_rate": 5.3714572864321614e-05,
-      "loss": 5.084,
-      "step": 46571
-    },
-    {
-      "epoch": 24.28787483702738,
-      "grad_norm": 1.5408251285552979,
-      "learning_rate": 5.371356783919598e-05,
-      "loss": 5.1376,
-      "step": 46572
-    },
-    {
-      "epoch": 24.288396349413297,
-      "grad_norm": 1.583795428276062,
-      "learning_rate": 5.3712562814070356e-05,
-      "loss": 5.1934,
-      "step": 46573
-    },
-    {
-      "epoch": 24.288917861799217,
-      "grad_norm": 1.534959316253662,
-      "learning_rate": 5.371155778894472e-05,
-      "loss": 5.6932,
-      "step": 46574
-    },
-    {
-      "epoch": 24.289439374185136,
-      "grad_norm": 1.569667100906372,
-      "learning_rate": 5.37105527638191e-05,
-      "loss": 5.257,
-      "step": 46575
-    },
-    {
-      "epoch": 24.289960886571055,
-      "grad_norm": 1.5968492031097412,
-      "learning_rate": 5.370954773869347e-05,
-      "loss": 4.7974,
-      "step": 46576
-    },
-    {
-      "epoch": 24.290482398956975,
-      "grad_norm": 1.7246915102005005,
-      "learning_rate": 5.370854271356785e-05,
-      "loss": 4.9212,
-      "step": 46577
-    },
-    {
-      "epoch": 24.291003911342894,
-      "grad_norm": 1.6612392663955688,
-      "learning_rate": 5.370753768844221e-05,
-      "loss": 5.1836,
-      "step": 46578
-    },
-    {
-      "epoch": 24.291525423728814,
-      "grad_norm": 1.5125454664230347,
-      "learning_rate": 5.370653266331659e-05,
-      "loss": 5.4666,
-      "step": 46579
-    },
-    {
-      "epoch": 24.292046936114733,
-      "grad_norm": 1.5160938501358032,
-      "learning_rate": 5.370552763819095e-05,
-      "loss": 5.2386,
-      "step": 46580
-    },
-    {
-      "epoch": 24.292568448500653,
-      "grad_norm": 1.6713393926620483,
-      "learning_rate": 5.370452261306533e-05,
-      "loss": 5.1998,
-      "step": 46581
-    },
-    {
-      "epoch": 24.293089960886572,
-      "grad_norm": 1.5227710008621216,
-      "learning_rate": 5.37035175879397e-05,
-      "loss": 4.9512,
-      "step": 46582
-    },
-    {
-      "epoch": 24.29361147327249,
-      "grad_norm": 1.5369834899902344,
-      "learning_rate": 5.370251256281408e-05,
-      "loss": 5.2184,
-      "step": 46583
-    },
-    {
-      "epoch": 24.294132985658408,
-      "grad_norm": 1.5856119394302368,
-      "learning_rate": 5.3701507537688444e-05,
-      "loss": 5.275,
-      "step": 46584
-    },
-    {
-      "epoch": 24.294654498044327,
-      "grad_norm": 1.6110239028930664,
-      "learning_rate": 5.370050251256281e-05,
-      "loss": 5.4654,
-      "step": 46585
-    },
-    {
-      "epoch": 24.295176010430247,
-      "grad_norm": 1.589174509048462,
-      "learning_rate": 5.3699497487437186e-05,
-      "loss": 5.0054,
-      "step": 46586
-    },
-    {
-      "epoch": 24.295697522816166,
-      "grad_norm": 1.4716191291809082,
-      "learning_rate": 5.369849246231156e-05,
-      "loss": 5.4205,
-      "step": 46587
-    },
-    {
-      "epoch": 24.296219035202085,
-      "grad_norm": 1.5405246019363403,
-      "learning_rate": 5.3697487437185935e-05,
-      "loss": 5.5069,
-      "step": 46588
-    },
-    {
-      "epoch": 24.296740547588005,
-      "grad_norm": 1.6221007108688354,
-      "learning_rate": 5.36964824120603e-05,
-      "loss": 4.8461,
-      "step": 46589
-    },
-    {
-      "epoch": 24.297262059973924,
-      "grad_norm": 1.5237624645233154,
-      "learning_rate": 5.369547738693468e-05,
-      "loss": 4.2474,
-      "step": 46590
-    },
-    {
-      "epoch": 24.297783572359844,
-      "grad_norm": 1.6060062646865845,
-      "learning_rate": 5.369447236180905e-05,
-      "loss": 5.299,
-      "step": 46591
-    },
-    {
-      "epoch": 24.298305084745763,
-      "grad_norm": 1.5928664207458496,
-      "learning_rate": 5.3693467336683426e-05,
-      "loss": 5.1822,
-      "step": 46592
-    },
-    {
-      "epoch": 24.298826597131683,
-      "grad_norm": 1.6345442533493042,
-      "learning_rate": 5.369246231155779e-05,
-      "loss": 4.9443,
-      "step": 46593
-    },
-    {
-      "epoch": 24.299348109517602,
-      "grad_norm": 1.4771085977554321,
-      "learning_rate": 5.369145728643217e-05,
-      "loss": 5.2035,
-      "step": 46594
-    },
-    {
-      "epoch": 24.29986962190352,
-      "grad_norm": 1.6538231372833252,
-      "learning_rate": 5.369045226130653e-05,
-      "loss": 4.7779,
-      "step": 46595
-    },
-    {
-      "epoch": 24.300391134289438,
-      "grad_norm": 1.5676822662353516,
-      "learning_rate": 5.368944723618091e-05,
-      "loss": 5.1694,
-      "step": 46596
-    },
-    {
-      "epoch": 24.300912646675357,
-      "grad_norm": 1.4690879583358765,
-      "learning_rate": 5.368844221105528e-05,
-      "loss": 5.0034,
-      "step": 46597
-    },
-    {
-      "epoch": 24.301434159061277,
-      "grad_norm": 1.5651763677597046,
-      "learning_rate": 5.3687437185929646e-05,
-      "loss": 5.2618,
-      "step": 46598
-    },
-    {
-      "epoch": 24.301955671447196,
-      "grad_norm": 1.5007787942886353,
-      "learning_rate": 5.3686432160804024e-05,
-      "loss": 5.2824,
-      "step": 46599
-    },
-    {
-      "epoch": 24.302477183833116,
-      "grad_norm": 1.588273525238037,
-      "learning_rate": 5.368542713567839e-05,
-      "loss": 5.2891,
-      "step": 46600
-    },
-    {
-      "epoch": 24.302998696219035,
-      "grad_norm": 1.5493247509002686,
-      "learning_rate": 5.3684422110552766e-05,
-      "loss": 4.4648,
-      "step": 46601
-    },
-    {
-      "epoch": 24.303520208604954,
-      "grad_norm": 1.62213933467865,
-      "learning_rate": 5.368341708542714e-05,
-      "loss": 4.064,
-      "step": 46602
-    },
-    {
-      "epoch": 24.304041720990874,
-      "grad_norm": 1.5776082277297974,
-      "learning_rate": 5.3682412060301515e-05,
-      "loss": 5.3707,
-      "step": 46603
-    },
-    {
-      "epoch": 24.304563233376793,
-      "grad_norm": 1.6238863468170166,
-      "learning_rate": 5.368140703517588e-05,
-      "loss": 5.1448,
-      "step": 46604
-    },
-    {
-      "epoch": 24.305084745762713,
-      "grad_norm": 1.517470121383667,
-      "learning_rate": 5.368040201005026e-05,
-      "loss": 4.9888,
-      "step": 46605
-    },
-    {
-      "epoch": 24.305606258148632,
-      "grad_norm": 1.5910130739212036,
-      "learning_rate": 5.367939698492462e-05,
-      "loss": 5.2658,
-      "step": 46606
-    },
-    {
-      "epoch": 24.30612777053455,
-      "grad_norm": 1.6356123685836792,
-      "learning_rate": 5.3678391959799e-05,
-      "loss": 5.0292,
-      "step": 46607
-    },
-    {
-      "epoch": 24.306649282920468,
-      "grad_norm": 1.5970298051834106,
-      "learning_rate": 5.367738693467337e-05,
-      "loss": 5.6171,
-      "step": 46608
-    },
-    {
-      "epoch": 24.307170795306387,
-      "grad_norm": 1.726518154144287,
-      "learning_rate": 5.367638190954775e-05,
-      "loss": 5.0374,
-      "step": 46609
-    },
-    {
-      "epoch": 24.307692307692307,
-      "grad_norm": 1.4900022745132446,
-      "learning_rate": 5.367537688442211e-05,
-      "loss": 5.4899,
-      "step": 46610
-    },
-    {
-      "epoch": 24.308213820078226,
-      "grad_norm": 1.4991257190704346,
-      "learning_rate": 5.3674371859296476e-05,
-      "loss": 5.125,
-      "step": 46611
-    },
-    {
-      "epoch": 24.308735332464146,
-      "grad_norm": 1.5483914613723755,
-      "learning_rate": 5.3673366834170854e-05,
-      "loss": 4.646,
-      "step": 46612
-    },
-    {
-      "epoch": 24.309256844850065,
-      "grad_norm": 1.6165622472763062,
-      "learning_rate": 5.3672361809045225e-05,
-      "loss": 5.1141,
-      "step": 46613
-    },
-    {
-      "epoch": 24.309778357235984,
-      "grad_norm": 1.48163902759552,
-      "learning_rate": 5.36713567839196e-05,
-      "loss": 5.1583,
-      "step": 46614
-    },
-    {
-      "epoch": 24.310299869621904,
-      "grad_norm": 1.6403732299804688,
-      "learning_rate": 5.367035175879397e-05,
-      "loss": 5.4645,
-      "step": 46615
-    },
-    {
-      "epoch": 24.310821382007823,
-      "grad_norm": 1.5265711545944214,
-      "learning_rate": 5.3669346733668345e-05,
-      "loss": 5.3366,
-      "step": 46616
-    },
-    {
-      "epoch": 24.311342894393743,
-      "grad_norm": 1.5642144680023193,
-      "learning_rate": 5.3668341708542716e-05,
-      "loss": 5.106,
-      "step": 46617
-    },
-    {
-      "epoch": 24.311864406779662,
-      "grad_norm": 1.5955440998077393,
-      "learning_rate": 5.3667336683417094e-05,
-      "loss": 5.2983,
-      "step": 46618
-    },
-    {
-      "epoch": 24.312385919165582,
-      "grad_norm": 1.602898120880127,
-      "learning_rate": 5.366633165829146e-05,
-      "loss": 5.3281,
-      "step": 46619
-    },
-    {
-      "epoch": 24.312907431551498,
-      "grad_norm": 1.6177289485931396,
-      "learning_rate": 5.3665326633165836e-05,
-      "loss": 5.1293,
-      "step": 46620
-    },
-    {
-      "epoch": 24.313428943937417,
-      "grad_norm": 2.3731181621551514,
-      "learning_rate": 5.36643216080402e-05,
-      "loss": 4.4469,
-      "step": 46621
-    },
-    {
-      "epoch": 24.313950456323337,
-      "grad_norm": 1.5811498165130615,
-      "learning_rate": 5.366331658291458e-05,
-      "loss": 5.0165,
-      "step": 46622
-    },
-    {
-      "epoch": 24.314471968709256,
-      "grad_norm": 1.5065430402755737,
-      "learning_rate": 5.366231155778895e-05,
-      "loss": 5.3461,
-      "step": 46623
-    },
-    {
-      "epoch": 24.314993481095176,
-      "grad_norm": 1.5695322751998901,
-      "learning_rate": 5.366130653266333e-05,
-      "loss": 5.3073,
-      "step": 46624
-    },
-    {
-      "epoch": 24.315514993481095,
-      "grad_norm": 1.6277281045913696,
-      "learning_rate": 5.366030150753769e-05,
-      "loss": 5.0665,
-      "step": 46625
-    },
-    {
-      "epoch": 24.316036505867014,
-      "grad_norm": 1.5736382007598877,
-      "learning_rate": 5.3659296482412056e-05,
-      "loss": 5.0317,
-      "step": 46626
-    },
-    {
-      "epoch": 24.316558018252934,
-      "grad_norm": 1.6100175380706787,
-      "learning_rate": 5.365829145728643e-05,
-      "loss": 5.1464,
-      "step": 46627
-    },
-    {
-      "epoch": 24.317079530638853,
-      "grad_norm": 1.522641897201538,
-      "learning_rate": 5.3657286432160804e-05,
-      "loss": 5.4654,
-      "step": 46628
-    },
-    {
-      "epoch": 24.317601043024773,
-      "grad_norm": 1.547913908958435,
-      "learning_rate": 5.365628140703518e-05,
-      "loss": 5.3363,
-      "step": 46629
-    },
-    {
-      "epoch": 24.318122555410692,
-      "grad_norm": 1.5869567394256592,
-      "learning_rate": 5.3655276381909547e-05,
-      "loss": 5.2518,
-      "step": 46630
-    },
-    {
-      "epoch": 24.318644067796612,
-      "grad_norm": 1.5204222202301025,
-      "learning_rate": 5.3654271356783924e-05,
-      "loss": 5.4275,
-      "step": 46631
-    },
-    {
-      "epoch": 24.319165580182528,
-      "grad_norm": 1.5300172567367554,
-      "learning_rate": 5.365326633165829e-05,
-      "loss": 5.4153,
-      "step": 46632
-    },
-    {
-      "epoch": 24.319687092568447,
-      "grad_norm": 1.5143780708312988,
-      "learning_rate": 5.3652261306532666e-05,
-      "loss": 4.8942,
-      "step": 46633
-    },
-    {
-      "epoch": 24.320208604954367,
-      "grad_norm": 1.5416815280914307,
-      "learning_rate": 5.365125628140704e-05,
-      "loss": 5.5277,
-      "step": 46634
-    },
-    {
-      "epoch": 24.320730117340286,
-      "grad_norm": 1.4537303447723389,
-      "learning_rate": 5.3650251256281415e-05,
-      "loss": 5.6494,
-      "step": 46635
-    },
-    {
-      "epoch": 24.321251629726206,
-      "grad_norm": 1.4255781173706055,
-      "learning_rate": 5.364924623115578e-05,
-      "loss": 5.7643,
-      "step": 46636
-    },
-    {
-      "epoch": 24.321773142112125,
-      "grad_norm": 1.5455092191696167,
-      "learning_rate": 5.364824120603016e-05,
-      "loss": 4.8143,
-      "step": 46637
-    },
-    {
-      "epoch": 24.322294654498045,
-      "grad_norm": 1.591746211051941,
-      "learning_rate": 5.364723618090453e-05,
-      "loss": 5.2743,
-      "step": 46638
-    },
-    {
-      "epoch": 24.322816166883964,
-      "grad_norm": 1.55540132522583,
-      "learning_rate": 5.364623115577889e-05,
-      "loss": 4.9435,
-      "step": 46639
-    },
-    {
-      "epoch": 24.323337679269883,
-      "grad_norm": 1.6163493394851685,
-      "learning_rate": 5.364522613065327e-05,
-      "loss": 5.2061,
-      "step": 46640
-    },
-    {
-      "epoch": 24.323859191655803,
-      "grad_norm": 1.462853193283081,
-      "learning_rate": 5.3644221105527635e-05,
-      "loss": 5.3553,
-      "step": 46641
-    },
-    {
-      "epoch": 24.324380704041722,
-      "grad_norm": 1.5892717838287354,
-      "learning_rate": 5.364321608040201e-05,
-      "loss": 4.9348,
-      "step": 46642
-    },
-    {
-      "epoch": 24.324902216427642,
-      "grad_norm": 1.5374174118041992,
-      "learning_rate": 5.3642211055276384e-05,
-      "loss": 5.2828,
-      "step": 46643
-    },
-    {
-      "epoch": 24.325423728813558,
-      "grad_norm": 1.5996121168136597,
-      "learning_rate": 5.364120603015076e-05,
-      "loss": 5.1828,
-      "step": 46644
-    },
-    {
-      "epoch": 24.325945241199477,
-      "grad_norm": 1.515793800354004,
-      "learning_rate": 5.3640201005025126e-05,
-      "loss": 5.3633,
-      "step": 46645
-    },
-    {
-      "epoch": 24.326466753585397,
-      "grad_norm": 1.637360692024231,
-      "learning_rate": 5.3639195979899504e-05,
-      "loss": 5.3676,
-      "step": 46646
-    },
-    {
-      "epoch": 24.326988265971316,
-      "grad_norm": 1.577966570854187,
-      "learning_rate": 5.363819095477387e-05,
-      "loss": 5.0286,
-      "step": 46647
-    },
-    {
-      "epoch": 24.327509778357236,
-      "grad_norm": 1.678457260131836,
-      "learning_rate": 5.3637185929648246e-05,
-      "loss": 4.3423,
-      "step": 46648
-    },
-    {
-      "epoch": 24.328031290743155,
-      "grad_norm": 1.600327968597412,
-      "learning_rate": 5.363618090452262e-05,
-      "loss": 5.3261,
-      "step": 46649
-    },
-    {
-      "epoch": 24.328552803129075,
-      "grad_norm": 1.6013729572296143,
-      "learning_rate": 5.3635175879396995e-05,
-      "loss": 5.2041,
-      "step": 46650
-    },
-    {
-      "epoch": 24.329074315514994,
-      "grad_norm": 1.5849155187606812,
-      "learning_rate": 5.363417085427136e-05,
-      "loss": 5.3744,
-      "step": 46651
-    },
-    {
-      "epoch": 24.329595827900913,
-      "grad_norm": 1.6084797382354736,
-      "learning_rate": 5.363316582914572e-05,
-      "loss": 5.318,
-      "step": 46652
-    },
-    {
-      "epoch": 24.330117340286833,
-      "grad_norm": 1.536217451095581,
-      "learning_rate": 5.36321608040201e-05,
-      "loss": 5.5239,
-      "step": 46653
-    },
-    {
-      "epoch": 24.330638852672752,
-      "grad_norm": 1.59565269947052,
-      "learning_rate": 5.363115577889447e-05,
-      "loss": 5.2355,
-      "step": 46654
-    },
-    {
-      "epoch": 24.331160365058672,
-      "grad_norm": 1.504576325416565,
-      "learning_rate": 5.363015075376885e-05,
-      "loss": 5.3442,
-      "step": 46655
-    },
-    {
-      "epoch": 24.331681877444588,
-      "grad_norm": 1.6171420812606812,
-      "learning_rate": 5.3629145728643214e-05,
-      "loss": 4.9218,
-      "step": 46656
-    },
-    {
-      "epoch": 24.332203389830507,
-      "grad_norm": 1.454347014427185,
-      "learning_rate": 5.362814070351759e-05,
-      "loss": 5.6167,
-      "step": 46657
-    },
-    {
-      "epoch": 24.332724902216427,
-      "grad_norm": 1.7542415857315063,
-      "learning_rate": 5.362713567839196e-05,
-      "loss": 5.0358,
-      "step": 46658
-    },
-    {
-      "epoch": 24.333246414602346,
-      "grad_norm": 1.6092952489852905,
-      "learning_rate": 5.362613065326634e-05,
-      "loss": 5.2362,
-      "step": 46659
-    },
-    {
-      "epoch": 24.333767926988266,
-      "grad_norm": 1.4792824983596802,
-      "learning_rate": 5.3625125628140705e-05,
-      "loss": 5.07,
-      "step": 46660
-    },
-    {
-      "epoch": 24.334289439374185,
-      "grad_norm": 1.492871880531311,
-      "learning_rate": 5.362412060301508e-05,
-      "loss": 5.0638,
-      "step": 46661
-    },
-    {
-      "epoch": 24.334810951760105,
-      "grad_norm": 1.620306372642517,
-      "learning_rate": 5.362311557788945e-05,
-      "loss": 4.8931,
-      "step": 46662
-    },
-    {
-      "epoch": 24.335332464146024,
-      "grad_norm": 1.4903837442398071,
-      "learning_rate": 5.3622110552763825e-05,
-      "loss": 5.5703,
-      "step": 46663
-    },
-    {
-      "epoch": 24.335853976531943,
-      "grad_norm": 1.5272018909454346,
-      "learning_rate": 5.3621105527638196e-05,
-      "loss": 5.2872,
-      "step": 46664
-    },
-    {
-      "epoch": 24.336375488917863,
-      "grad_norm": 1.5871680974960327,
-      "learning_rate": 5.362010050251256e-05,
-      "loss": 5.4374,
-      "step": 46665
-    },
-    {
-      "epoch": 24.336897001303782,
-      "grad_norm": 1.5554306507110596,
-      "learning_rate": 5.361909547738694e-05,
-      "loss": 5.5505,
-      "step": 46666
-    },
-    {
-      "epoch": 24.3374185136897,
-      "grad_norm": 1.4673306941986084,
-      "learning_rate": 5.36180904522613e-05,
-      "loss": 5.7145,
-      "step": 46667
-    },
-    {
-      "epoch": 24.337940026075618,
-      "grad_norm": 1.4878606796264648,
-      "learning_rate": 5.361708542713568e-05,
-      "loss": 5.4413,
-      "step": 46668
-    },
-    {
-      "epoch": 24.338461538461537,
-      "grad_norm": 1.541118860244751,
-      "learning_rate": 5.361608040201005e-05,
-      "loss": 5.0377,
-      "step": 46669
-    },
-    {
-      "epoch": 24.338983050847457,
-      "grad_norm": 1.6206892728805542,
-      "learning_rate": 5.361507537688443e-05,
-      "loss": 5.2933,
-      "step": 46670
-    },
-    {
-      "epoch": 24.339504563233376,
-      "grad_norm": 1.446699857711792,
-      "learning_rate": 5.3614070351758794e-05,
-      "loss": 5.3192,
-      "step": 46671
-    },
-    {
-      "epoch": 24.340026075619296,
-      "grad_norm": 1.4445785284042358,
-      "learning_rate": 5.361306532663317e-05,
-      "loss": 5.5648,
-      "step": 46672
-    },
-    {
-      "epoch": 24.340547588005215,
-      "grad_norm": 1.6103436946868896,
-      "learning_rate": 5.3612060301507536e-05,
-      "loss": 4.9791,
-      "step": 46673
-    },
-    {
-      "epoch": 24.341069100391135,
-      "grad_norm": 1.6132456064224243,
-      "learning_rate": 5.3611055276381913e-05,
-      "loss": 5.0591,
-      "step": 46674
-    },
-    {
-      "epoch": 24.341590612777054,
-      "grad_norm": 1.5596740245819092,
-      "learning_rate": 5.3610050251256285e-05,
-      "loss": 5.0358,
-      "step": 46675
-    },
-    {
-      "epoch": 24.342112125162974,
-      "grad_norm": 1.5720112323760986,
-      "learning_rate": 5.360904522613066e-05,
-      "loss": 4.9,
-      "step": 46676
-    },
-    {
-      "epoch": 24.342633637548893,
-      "grad_norm": 1.4962886571884155,
-      "learning_rate": 5.360804020100503e-05,
-      "loss": 5.2838,
-      "step": 46677
-    },
-    {
-      "epoch": 24.343155149934812,
-      "grad_norm": 1.487217903137207,
-      "learning_rate": 5.360703517587939e-05,
-      "loss": 5.8337,
-      "step": 46678
-    },
-    {
-      "epoch": 24.343676662320732,
-      "grad_norm": 1.4660305976867676,
-      "learning_rate": 5.360603015075377e-05,
-      "loss": 5.3055,
-      "step": 46679
-    },
-    {
-      "epoch": 24.344198174706648,
-      "grad_norm": 1.5468034744262695,
-      "learning_rate": 5.360502512562814e-05,
-      "loss": 5.5191,
-      "step": 46680
-    },
-    {
-      "epoch": 24.344719687092567,
-      "grad_norm": 1.5067448616027832,
-      "learning_rate": 5.360402010050252e-05,
-      "loss": 5.4391,
-      "step": 46681
-    },
-    {
-      "epoch": 24.345241199478487,
-      "grad_norm": 1.5891979932785034,
-      "learning_rate": 5.360301507537688e-05,
-      "loss": 4.8512,
-      "step": 46682
-    },
-    {
-      "epoch": 24.345762711864406,
-      "grad_norm": 1.4967737197875977,
-      "learning_rate": 5.360201005025126e-05,
-      "loss": 5.8709,
-      "step": 46683
-    },
-    {
-      "epoch": 24.346284224250326,
-      "grad_norm": 1.5120757818222046,
-      "learning_rate": 5.360100502512563e-05,
-      "loss": 5.2026,
-      "step": 46684
-    },
-    {
-      "epoch": 24.346805736636245,
-      "grad_norm": 1.6322177648544312,
-      "learning_rate": 5.360000000000001e-05,
-      "loss": 4.5448,
-      "step": 46685
-    },
-    {
-      "epoch": 24.347327249022165,
-      "grad_norm": 1.665804147720337,
-      "learning_rate": 5.359899497487437e-05,
-      "loss": 5.0943,
-      "step": 46686
-    },
-    {
-      "epoch": 24.347848761408084,
-      "grad_norm": 1.5306041240692139,
-      "learning_rate": 5.359798994974875e-05,
-      "loss": 5.3209,
-      "step": 46687
-    },
-    {
-      "epoch": 24.348370273794004,
-      "grad_norm": 1.536742091178894,
-      "learning_rate": 5.3596984924623115e-05,
-      "loss": 5.2572,
-      "step": 46688
-    },
-    {
-      "epoch": 24.348891786179923,
-      "grad_norm": 1.5175718069076538,
-      "learning_rate": 5.359597989949749e-05,
-      "loss": 5.0965,
-      "step": 46689
-    },
-    {
-      "epoch": 24.349413298565842,
-      "grad_norm": 1.5491081476211548,
-      "learning_rate": 5.3594974874371864e-05,
-      "loss": 5.6037,
-      "step": 46690
-    },
-    {
-      "epoch": 24.34993481095176,
-      "grad_norm": 1.6356558799743652,
-      "learning_rate": 5.359396984924623e-05,
-      "loss": 4.9422,
-      "step": 46691
-    },
-    {
-      "epoch": 24.350456323337678,
-      "grad_norm": 1.551336407661438,
-      "learning_rate": 5.3592964824120606e-05,
-      "loss": 4.9636,
-      "step": 46692
-    },
-    {
-      "epoch": 24.350977835723597,
-      "grad_norm": 1.4739060401916504,
-      "learning_rate": 5.359195979899497e-05,
-      "loss": 4.7283,
-      "step": 46693
-    },
-    {
-      "epoch": 24.351499348109517,
-      "grad_norm": 1.6668118238449097,
-      "learning_rate": 5.359095477386935e-05,
-      "loss": 5.2958,
-      "step": 46694
-    },
-    {
-      "epoch": 24.352020860495436,
-      "grad_norm": 1.5188933610916138,
-      "learning_rate": 5.358994974874372e-05,
-      "loss": 5.5809,
-      "step": 46695
-    },
-    {
-      "epoch": 24.352542372881356,
-      "grad_norm": 1.5982838869094849,
-      "learning_rate": 5.35889447236181e-05,
-      "loss": 5.3575,
-      "step": 46696
-    },
-    {
-      "epoch": 24.353063885267275,
-      "grad_norm": 1.6466450691223145,
-      "learning_rate": 5.358793969849246e-05,
-      "loss": 5.3003,
-      "step": 46697
-    },
-    {
-      "epoch": 24.353585397653195,
-      "grad_norm": 1.6770554780960083,
-      "learning_rate": 5.358693467336684e-05,
-      "loss": 5.5338,
-      "step": 46698
-    },
-    {
-      "epoch": 24.354106910039114,
-      "grad_norm": 1.539605736732483,
-      "learning_rate": 5.35859296482412e-05,
-      "loss": 5.3957,
-      "step": 46699
-    },
-    {
-      "epoch": 24.354628422425034,
-      "grad_norm": 1.5029267072677612,
-      "learning_rate": 5.358492462311558e-05,
-      "loss": 4.8365,
-      "step": 46700
-    },
-    {
-      "epoch": 24.355149934810953,
-      "grad_norm": 1.5314891338348389,
-      "learning_rate": 5.358391959798995e-05,
-      "loss": 5.1098,
-      "step": 46701
-    },
-    {
-      "epoch": 24.355671447196872,
-      "grad_norm": 1.4592232704162598,
-      "learning_rate": 5.358291457286433e-05,
-      "loss": 5.0325,
-      "step": 46702
-    },
-    {
-      "epoch": 24.35619295958279,
-      "grad_norm": 1.5268797874450684,
-      "learning_rate": 5.3581909547738694e-05,
-      "loss": 5.1898,
-      "step": 46703
-    },
-    {
-      "epoch": 24.356714471968708,
-      "grad_norm": 1.5811846256256104,
-      "learning_rate": 5.3580904522613065e-05,
-      "loss": 4.6398,
-      "step": 46704
-    },
-    {
-      "epoch": 24.357235984354627,
-      "grad_norm": 1.4974379539489746,
-      "learning_rate": 5.357989949748744e-05,
-      "loss": 5.4097,
-      "step": 46705
-    },
-    {
-      "epoch": 24.357757496740547,
-      "grad_norm": 1.5028765201568604,
-      "learning_rate": 5.357889447236181e-05,
-      "loss": 4.9995,
-      "step": 46706
-    },
-    {
-      "epoch": 24.358279009126466,
-      "grad_norm": 1.5281312465667725,
-      "learning_rate": 5.3577889447236185e-05,
-      "loss": 5.5609,
-      "step": 46707
-    },
-    {
-      "epoch": 24.358800521512386,
-      "grad_norm": 1.6143425703048706,
-      "learning_rate": 5.357688442211055e-05,
-      "loss": 5.4462,
-      "step": 46708
-    },
-    {
-      "epoch": 24.359322033898305,
-      "grad_norm": 1.517938494682312,
-      "learning_rate": 5.357587939698493e-05,
-      "loss": 5.5515,
-      "step": 46709
-    },
-    {
-      "epoch": 24.359843546284225,
-      "grad_norm": 1.488358974456787,
-      "learning_rate": 5.35748743718593e-05,
-      "loss": 5.7042,
-      "step": 46710
-    },
-    {
-      "epoch": 24.360365058670144,
-      "grad_norm": 1.5919100046157837,
-      "learning_rate": 5.3573869346733676e-05,
-      "loss": 5.364,
-      "step": 46711
-    },
-    {
-      "epoch": 24.360886571056064,
-      "grad_norm": 1.5082749128341675,
-      "learning_rate": 5.357286432160804e-05,
-      "loss": 5.7376,
-      "step": 46712
-    },
-    {
-      "epoch": 24.361408083441983,
-      "grad_norm": 1.468997836112976,
-      "learning_rate": 5.357185929648242e-05,
-      "loss": 5.2231,
-      "step": 46713
-    },
-    {
-      "epoch": 24.361929595827903,
-      "grad_norm": 1.6588627099990845,
-      "learning_rate": 5.357085427135678e-05,
-      "loss": 4.8492,
-      "step": 46714
-    },
-    {
-      "epoch": 24.36245110821382,
-      "grad_norm": 1.5386675596237183,
-      "learning_rate": 5.356984924623116e-05,
-      "loss": 5.6983,
-      "step": 46715
-    },
-    {
-      "epoch": 24.362972620599738,
-      "grad_norm": 1.9397474527359009,
-      "learning_rate": 5.356884422110553e-05,
-      "loss": 5.226,
-      "step": 46716
-    },
-    {
-      "epoch": 24.363494132985657,
-      "grad_norm": 1.6140294075012207,
-      "learning_rate": 5.356783919597991e-05,
-      "loss": 5.1547,
-      "step": 46717
-    },
-    {
-      "epoch": 24.364015645371577,
-      "grad_norm": 1.6452679634094238,
-      "learning_rate": 5.3566834170854274e-05,
-      "loss": 4.864,
-      "step": 46718
-    },
-    {
-      "epoch": 24.364537157757496,
-      "grad_norm": 1.611659288406372,
-      "learning_rate": 5.356582914572864e-05,
-      "loss": 4.9852,
-      "step": 46719
-    },
-    {
-      "epoch": 24.365058670143416,
-      "grad_norm": 1.5865225791931152,
-      "learning_rate": 5.3564824120603016e-05,
-      "loss": 4.9751,
-      "step": 46720
-    },
-    {
-      "epoch": 24.365580182529335,
-      "grad_norm": 1.5311130285263062,
-      "learning_rate": 5.356381909547739e-05,
-      "loss": 5.5341,
-      "step": 46721
-    },
-    {
-      "epoch": 24.366101694915255,
-      "grad_norm": 1.6892858743667603,
-      "learning_rate": 5.3562814070351765e-05,
-      "loss": 4.8011,
-      "step": 46722
-    },
-    {
-      "epoch": 24.366623207301174,
-      "grad_norm": 1.5774376392364502,
-      "learning_rate": 5.356180904522613e-05,
-      "loss": 4.9398,
-      "step": 46723
-    },
-    {
-      "epoch": 24.367144719687094,
-      "grad_norm": 1.680722951889038,
-      "learning_rate": 5.356080402010051e-05,
-      "loss": 5.0496,
-      "step": 46724
-    },
-    {
-      "epoch": 24.367666232073013,
-      "grad_norm": 1.5784188508987427,
-      "learning_rate": 5.355979899497488e-05,
-      "loss": 5.0779,
-      "step": 46725
-    },
-    {
-      "epoch": 24.368187744458933,
-      "grad_norm": 1.5683897733688354,
-      "learning_rate": 5.3558793969849256e-05,
-      "loss": 5.136,
-      "step": 46726
-    },
-    {
-      "epoch": 24.36870925684485,
-      "grad_norm": 1.6975905895233154,
-      "learning_rate": 5.355778894472362e-05,
-      "loss": 4.8463,
-      "step": 46727
-    },
-    {
-      "epoch": 24.369230769230768,
-      "grad_norm": 1.5479676723480225,
-      "learning_rate": 5.3556783919598e-05,
-      "loss": 5.259,
-      "step": 46728
-    },
-    {
-      "epoch": 24.369752281616687,
-      "grad_norm": 1.5467514991760254,
-      "learning_rate": 5.355577889447236e-05,
-      "loss": 5.3925,
-      "step": 46729
-    },
-    {
-      "epoch": 24.370273794002607,
-      "grad_norm": 1.4710688591003418,
-      "learning_rate": 5.355477386934674e-05,
-      "loss": 5.5907,
-      "step": 46730
-    },
-    {
-      "epoch": 24.370795306388526,
-      "grad_norm": 1.5912089347839355,
-      "learning_rate": 5.355376884422111e-05,
-      "loss": 5.2704,
-      "step": 46731
-    },
-    {
-      "epoch": 24.371316818774446,
-      "grad_norm": 1.6285303831100464,
-      "learning_rate": 5.3552763819095475e-05,
-      "loss": 5.2124,
-      "step": 46732
-    },
-    {
-      "epoch": 24.371838331160365,
-      "grad_norm": 1.572697639465332,
-      "learning_rate": 5.355175879396985e-05,
-      "loss": 5.1687,
-      "step": 46733
-    },
-    {
-      "epoch": 24.372359843546285,
-      "grad_norm": 1.6056311130523682,
-      "learning_rate": 5.355075376884422e-05,
-      "loss": 5.2608,
-      "step": 46734
-    },
-    {
-      "epoch": 24.372881355932204,
-      "grad_norm": 1.653411626815796,
-      "learning_rate": 5.3549748743718595e-05,
-      "loss": 4.6762,
-      "step": 46735
-    },
-    {
-      "epoch": 24.373402868318124,
-      "grad_norm": 1.5580683946609497,
-      "learning_rate": 5.3548743718592966e-05,
-      "loss": 5.5959,
-      "step": 46736
-    },
-    {
-      "epoch": 24.373924380704043,
-      "grad_norm": 1.6278222799301147,
-      "learning_rate": 5.3547738693467344e-05,
-      "loss": 5.4653,
-      "step": 46737
-    },
-    {
-      "epoch": 24.374445893089963,
-      "grad_norm": 1.4319900274276733,
-      "learning_rate": 5.354673366834171e-05,
-      "loss": 5.5572,
-      "step": 46738
-    },
-    {
-      "epoch": 24.37496740547588,
-      "grad_norm": 1.564563512802124,
-      "learning_rate": 5.3545728643216086e-05,
-      "loss": 5.212,
-      "step": 46739
-    },
-    {
-      "epoch": 24.375488917861798,
-      "grad_norm": 1.5445950031280518,
-      "learning_rate": 5.354472361809045e-05,
-      "loss": 5.4503,
-      "step": 46740
-    },
-    {
-      "epoch": 24.376010430247717,
-      "grad_norm": 1.5167509317398071,
-      "learning_rate": 5.354371859296483e-05,
-      "loss": 5.4314,
-      "step": 46741
-    },
-    {
-      "epoch": 24.376531942633637,
-      "grad_norm": 1.472383737564087,
-      "learning_rate": 5.35427135678392e-05,
-      "loss": 5.4351,
-      "step": 46742
-    },
-    {
-      "epoch": 24.377053455019556,
-      "grad_norm": 1.536608338356018,
-      "learning_rate": 5.354170854271358e-05,
-      "loss": 5.0218,
-      "step": 46743
-    },
-    {
-      "epoch": 24.377574967405476,
-      "grad_norm": 1.5551252365112305,
-      "learning_rate": 5.354070351758794e-05,
-      "loss": 5.4413,
-      "step": 46744
-    },
-    {
-      "epoch": 24.378096479791395,
-      "grad_norm": 1.5637702941894531,
-      "learning_rate": 5.353969849246231e-05,
-      "loss": 5.1231,
-      "step": 46745
-    },
-    {
-      "epoch": 24.378617992177315,
-      "grad_norm": 1.5864009857177734,
-      "learning_rate": 5.353869346733669e-05,
-      "loss": 5.0547,
-      "step": 46746
-    },
-    {
-      "epoch": 24.379139504563234,
-      "grad_norm": 1.7575881481170654,
-      "learning_rate": 5.3537688442211054e-05,
-      "loss": 5.3276,
-      "step": 46747
-    },
-    {
-      "epoch": 24.379661016949154,
-      "grad_norm": 1.548041582107544,
-      "learning_rate": 5.353668341708543e-05,
-      "loss": 4.6297,
-      "step": 46748
-    },
-    {
-      "epoch": 24.380182529335073,
-      "grad_norm": 1.608337640762329,
-      "learning_rate": 5.3535678391959797e-05,
-      "loss": 5.34,
-      "step": 46749
-    },
-    {
-      "epoch": 24.380704041720993,
-      "grad_norm": 1.4815932512283325,
-      "learning_rate": 5.3534673366834174e-05,
-      "loss": 5.3316,
-      "step": 46750
-    },
-    {
-      "epoch": 24.38122555410691,
-      "grad_norm": 1.62247633934021,
-      "learning_rate": 5.3533668341708545e-05,
-      "loss": 4.6363,
-      "step": 46751
-    },
-    {
-      "epoch": 24.381747066492828,
-      "grad_norm": 1.492600679397583,
-      "learning_rate": 5.353266331658292e-05,
-      "loss": 5.578,
-      "step": 46752
-    },
-    {
-      "epoch": 24.382268578878747,
-      "grad_norm": 1.534112572669983,
-      "learning_rate": 5.353165829145729e-05,
-      "loss": 5.3305,
-      "step": 46753
-    },
-    {
-      "epoch": 24.382790091264667,
-      "grad_norm": 1.5693928003311157,
-      "learning_rate": 5.3530653266331665e-05,
-      "loss": 5.1411,
-      "step": 46754
-    },
-    {
-      "epoch": 24.383311603650586,
-      "grad_norm": 1.5709365606307983,
-      "learning_rate": 5.352964824120603e-05,
-      "loss": 5.1967,
-      "step": 46755
-    },
-    {
-      "epoch": 24.383833116036506,
-      "grad_norm": 1.5470030307769775,
-      "learning_rate": 5.352864321608041e-05,
-      "loss": 5.3642,
-      "step": 46756
-    },
-    {
-      "epoch": 24.384354628422425,
-      "grad_norm": NaN,
-      "learning_rate": 5.352864321608041e-05,
-      "loss": 5.4593,
-      "step": 46757
-    },
-    {
-      "epoch": 24.384876140808345,
-      "grad_norm": 1.6188031435012817,
-      "learning_rate": 5.352763819095478e-05,
-      "loss": 5.2543,
-      "step": 46758
-    },
-    {
-      "epoch": 24.385397653194264,
-      "grad_norm": 1.4927293062210083,
-      "learning_rate": 5.352663316582914e-05,
-      "loss": 4.9877,
-      "step": 46759
-    },
-    {
-      "epoch": 24.385919165580184,
-      "grad_norm": 1.559382677078247,
-      "learning_rate": 5.352562814070352e-05,
-      "loss": 5.026,
-      "step": 46760
-    },
-    {
-      "epoch": 24.386440677966103,
-      "grad_norm": 1.5425792932510376,
-      "learning_rate": 5.3524623115577885e-05,
-      "loss": 5.2035,
-      "step": 46761
-    },
-    {
-      "epoch": 24.386962190352023,
-      "grad_norm": 1.5445119142532349,
-      "learning_rate": 5.352361809045226e-05,
-      "loss": 5.2641,
-      "step": 46762
-    },
-    {
-      "epoch": 24.38748370273794,
-      "grad_norm": 1.7350221872329712,
-      "learning_rate": 5.3522613065326634e-05,
-      "loss": 4.8422,
-      "step": 46763
-    },
-    {
-      "epoch": 24.388005215123858,
-      "grad_norm": 1.4480572938919067,
-      "learning_rate": 5.352160804020101e-05,
-      "loss": 4.9186,
-      "step": 46764
-    },
-    {
-      "epoch": 24.388526727509777,
-      "grad_norm": 1.5679616928100586,
-      "learning_rate": 5.3520603015075376e-05,
-      "loss": 5.4042,
-      "step": 46765
-    },
-    {
-      "epoch": 24.389048239895697,
-      "grad_norm": 1.5019046068191528,
-      "learning_rate": 5.3519597989949754e-05,
-      "loss": 4.8916,
-      "step": 46766
-    },
-    {
-      "epoch": 24.389569752281616,
-      "grad_norm": 1.5769399404525757,
-      "learning_rate": 5.351859296482412e-05,
-      "loss": 4.7467,
-      "step": 46767
-    },
-    {
-      "epoch": 24.390091264667536,
-      "grad_norm": 1.6964269876480103,
-      "learning_rate": 5.3517587939698496e-05,
-      "loss": 4.539,
-      "step": 46768
-    },
-    {
-      "epoch": 24.390612777053455,
-      "grad_norm": 1.490473985671997,
-      "learning_rate": 5.351658291457287e-05,
-      "loss": 5.5297,
-      "step": 46769
-    },
-    {
-      "epoch": 24.391134289439375,
-      "grad_norm": 1.5852210521697998,
-      "learning_rate": 5.3515577889447245e-05,
-      "loss": 5.0051,
-      "step": 46770
-    },
-    {
-      "epoch": 24.391655801825294,
-      "grad_norm": 1.552804946899414,
-      "learning_rate": 5.351457286432161e-05,
-      "loss": 5.042,
-      "step": 46771
-    },
-    {
-      "epoch": 24.392177314211214,
-      "grad_norm": 1.603550910949707,
-      "learning_rate": 5.351356783919598e-05,
-      "loss": 4.9104,
-      "step": 46772
-    },
-    {
-      "epoch": 24.392698826597133,
-      "grad_norm": 1.5778529644012451,
-      "learning_rate": 5.351256281407036e-05,
-      "loss": 5.4691,
-      "step": 46773
-    },
-    {
-      "epoch": 24.39322033898305,
-      "grad_norm": 1.6180814504623413,
-      "learning_rate": 5.351155778894472e-05,
-      "loss": 5.0149,
-      "step": 46774
-    },
-    {
-      "epoch": 24.39374185136897,
-      "grad_norm": 1.4708882570266724,
-      "learning_rate": 5.35105527638191e-05,
-      "loss": 5.7194,
-      "step": 46775
-    },
-    {
-      "epoch": 24.394263363754888,
-      "grad_norm": 1.4812366962432861,
-      "learning_rate": 5.3509547738693464e-05,
-      "loss": 5.3904,
-      "step": 46776
-    },
-    {
-      "epoch": 24.394784876140807,
-      "grad_norm": 1.4794608354568481,
-      "learning_rate": 5.350854271356784e-05,
-      "loss": 5.6902,
-      "step": 46777
-    },
-    {
-      "epoch": 24.395306388526727,
-      "grad_norm": 1.5230708122253418,
-      "learning_rate": 5.350753768844221e-05,
-      "loss": 5.4786,
-      "step": 46778
-    },
-    {
-      "epoch": 24.395827900912646,
-      "grad_norm": 1.5241135358810425,
-      "learning_rate": 5.350653266331659e-05,
-      "loss": 4.6597,
-      "step": 46779
-    },
-    {
-      "epoch": 24.396349413298566,
-      "grad_norm": 1.5163906812667847,
-      "learning_rate": 5.3505527638190955e-05,
-      "loss": 5.1006,
-      "step": 46780
-    },
-    {
-      "epoch": 24.396870925684485,
-      "grad_norm": 1.4672882556915283,
-      "learning_rate": 5.350452261306533e-05,
-      "loss": 5.3952,
-      "step": 46781
-    },
-    {
-      "epoch": 24.397392438070405,
-      "grad_norm": 1.5129591226577759,
-      "learning_rate": 5.35035175879397e-05,
-      "loss": 5.6294,
-      "step": 46782
-    },
-    {
-      "epoch": 24.397913950456324,
-      "grad_norm": 1.5973355770111084,
-      "learning_rate": 5.3502512562814075e-05,
-      "loss": 5.1503,
-      "step": 46783
-    },
-    {
-      "epoch": 24.398435462842244,
-      "grad_norm": 1.5719927549362183,
-      "learning_rate": 5.3501507537688446e-05,
-      "loss": 5.143,
-      "step": 46784
-    },
-    {
-      "epoch": 24.398956975228163,
-      "grad_norm": 1.5816539525985718,
-      "learning_rate": 5.350050251256281e-05,
-      "loss": 5.2509,
-      "step": 46785
-    },
-    {
-      "epoch": 24.39947848761408,
-      "grad_norm": 1.4487128257751465,
-      "learning_rate": 5.349949748743719e-05,
-      "loss": 5.4624,
-      "step": 46786
-    },
-    {
-      "epoch": 24.4,
-      "grad_norm": 1.5997023582458496,
-      "learning_rate": 5.349849246231155e-05,
-      "loss": 5.0839,
-      "step": 46787
-    },
-    {
-      "epoch": 24.400521512385918,
-      "grad_norm": 1.605520248413086,
-      "learning_rate": 5.349748743718593e-05,
-      "loss": 4.9188,
-      "step": 46788
-    },
-    {
-      "epoch": 24.401043024771838,
-      "grad_norm": 1.49250066280365,
-      "learning_rate": 5.34964824120603e-05,
-      "loss": 5.2519,
-      "step": 46789
-    },
-    {
-      "epoch": 24.401564537157757,
-      "grad_norm": 1.5662202835083008,
-      "learning_rate": 5.349547738693468e-05,
-      "loss": 4.9463,
-      "step": 46790
-    },
-    {
-      "epoch": 24.402086049543676,
-      "grad_norm": 1.58832848072052,
-      "learning_rate": 5.3494472361809044e-05,
-      "loss": 4.8115,
-      "step": 46791
-    },
-    {
-      "epoch": 24.402607561929596,
-      "grad_norm": 1.5292195081710815,
-      "learning_rate": 5.349346733668342e-05,
-      "loss": 5.5499,
-      "step": 46792
-    },
-    {
-      "epoch": 24.403129074315515,
-      "grad_norm": 1.6421623229980469,
-      "learning_rate": 5.349246231155779e-05,
-      "loss": 5.3064,
-      "step": 46793
-    },
-    {
-      "epoch": 24.403650586701435,
-      "grad_norm": 1.4181722402572632,
-      "learning_rate": 5.349145728643217e-05,
-      "loss": 5.5014,
-      "step": 46794
-    },
-    {
-      "epoch": 24.404172099087354,
-      "grad_norm": 1.5616059303283691,
-      "learning_rate": 5.3490452261306535e-05,
-      "loss": 4.7792,
-      "step": 46795
-    },
-    {
-      "epoch": 24.404693611473274,
-      "grad_norm": 1.5391825437545776,
-      "learning_rate": 5.348944723618091e-05,
-      "loss": 5.0982,
-      "step": 46796
-    },
-    {
-      "epoch": 24.405215123859193,
-      "grad_norm": 1.5276191234588623,
-      "learning_rate": 5.348844221105528e-05,
-      "loss": 5.2672,
-      "step": 46797
-    },
-    {
-      "epoch": 24.40573663624511,
-      "grad_norm": 1.4793869256973267,
-      "learning_rate": 5.348743718592965e-05,
-      "loss": 5.2786,
-      "step": 46798
-    },
-    {
-      "epoch": 24.40625814863103,
-      "grad_norm": 1.496817708015442,
-      "learning_rate": 5.3486432160804026e-05,
-      "loss": 4.3457,
-      "step": 46799
-    },
-    {
-      "epoch": 24.406779661016948,
-      "grad_norm": 1.4432053565979004,
-      "learning_rate": 5.348542713567839e-05,
-      "loss": 5.3139,
-      "step": 46800
-    },
-    {
-      "epoch": 24.407301173402868,
-      "grad_norm": 1.6009502410888672,
-      "learning_rate": 5.348442211055277e-05,
-      "loss": 5.1683,
-      "step": 46801
-    },
-    {
-      "epoch": 24.407822685788787,
-      "grad_norm": 1.520005464553833,
-      "learning_rate": 5.348341708542713e-05,
-      "loss": 5.5065,
-      "step": 46802
-    },
-    {
-      "epoch": 24.408344198174706,
-      "grad_norm": 1.528971552848816,
-      "learning_rate": 5.348241206030151e-05,
-      "loss": 5.2842,
-      "step": 46803
-    },
-    {
-      "epoch": 24.408865710560626,
-      "grad_norm": 1.4042320251464844,
-      "learning_rate": 5.348140703517588e-05,
-      "loss": 4.8447,
-      "step": 46804
-    },
-    {
-      "epoch": 24.409387222946545,
-      "grad_norm": 1.5521961450576782,
-      "learning_rate": 5.348040201005026e-05,
-      "loss": 5.4643,
-      "step": 46805
-    },
-    {
-      "epoch": 24.409908735332465,
-      "grad_norm": 1.5985453128814697,
-      "learning_rate": 5.347939698492462e-05,
-      "loss": 5.1848,
-      "step": 46806
-    },
-    {
-      "epoch": 24.410430247718384,
-      "grad_norm": 1.6031867265701294,
-      "learning_rate": 5.3478391959799e-05,
-      "loss": 5.4451,
-      "step": 46807
-    },
-    {
-      "epoch": 24.410951760104304,
-      "grad_norm": 1.4890942573547363,
-      "learning_rate": 5.3477386934673365e-05,
-      "loss": 5.2139,
-      "step": 46808
-    },
-    {
-      "epoch": 24.411473272490223,
-      "grad_norm": 1.5691941976547241,
-      "learning_rate": 5.347638190954774e-05,
-      "loss": 5.3449,
-      "step": 46809
-    },
-    {
-      "epoch": 24.41199478487614,
-      "grad_norm": 1.612863540649414,
-      "learning_rate": 5.3475376884422114e-05,
-      "loss": 5.031,
-      "step": 46810
-    },
-    {
-      "epoch": 24.41251629726206,
-      "grad_norm": 1.4622020721435547,
-      "learning_rate": 5.347437185929649e-05,
-      "loss": 5.2702,
-      "step": 46811
-    },
-    {
-      "epoch": 24.413037809647978,
-      "grad_norm": 1.6028493642807007,
-      "learning_rate": 5.3473366834170856e-05,
-      "loss": 5.0679,
-      "step": 46812
-    },
-    {
-      "epoch": 24.413559322033898,
-      "grad_norm": 1.6639914512634277,
-      "learning_rate": 5.347236180904523e-05,
-      "loss": 5.3367,
-      "step": 46813
-    },
-    {
-      "epoch": 24.414080834419817,
-      "grad_norm": 1.5464037656784058,
-      "learning_rate": 5.3471356783919605e-05,
-      "loss": 5.5064,
-      "step": 46814
-    },
-    {
-      "epoch": 24.414602346805736,
-      "grad_norm": 1.5480848550796509,
-      "learning_rate": 5.347035175879397e-05,
-      "loss": 5.196,
-      "step": 46815
-    },
-    {
-      "epoch": 24.415123859191656,
-      "grad_norm": 1.489405870437622,
-      "learning_rate": 5.346934673366835e-05,
-      "loss": 5.8414,
-      "step": 46816
-    },
-    {
-      "epoch": 24.415645371577575,
-      "grad_norm": 1.534005880355835,
-      "learning_rate": 5.346834170854271e-05,
-      "loss": 5.4964,
-      "step": 46817
-    },
-    {
-      "epoch": 24.416166883963495,
-      "grad_norm": 1.5526580810546875,
-      "learning_rate": 5.346733668341709e-05,
-      "loss": 5.5629,
-      "step": 46818
-    },
-    {
-      "epoch": 24.416688396349414,
-      "grad_norm": 1.5449583530426025,
-      "learning_rate": 5.346633165829146e-05,
-      "loss": 5.1884,
-      "step": 46819
-    },
-    {
-      "epoch": 24.417209908735334,
-      "grad_norm": 1.6475977897644043,
-      "learning_rate": 5.346532663316584e-05,
-      "loss": 5.2614,
-      "step": 46820
-    },
-    {
-      "epoch": 24.417731421121253,
-      "grad_norm": 1.705411672592163,
-      "learning_rate": 5.34643216080402e-05,
-      "loss": 5.2128,
-      "step": 46821
-    },
-    {
-      "epoch": 24.41825293350717,
-      "grad_norm": 1.534668207168579,
-      "learning_rate": 5.346331658291458e-05,
-      "loss": 5.0137,
-      "step": 46822
-    },
-    {
-      "epoch": 24.41877444589309,
-      "grad_norm": 1.5202434062957764,
-      "learning_rate": 5.3462311557788944e-05,
-      "loss": 5.5161,
-      "step": 46823
-    },
-    {
-      "epoch": 24.419295958279008,
-      "grad_norm": 1.5040980577468872,
-      "learning_rate": 5.346130653266332e-05,
-      "loss": 5.4303,
-      "step": 46824
-    },
-    {
-      "epoch": 24.419817470664928,
-      "grad_norm": 1.5516712665557861,
-      "learning_rate": 5.346030150753769e-05,
-      "loss": 5.5705,
-      "step": 46825
-    },
-    {
-      "epoch": 24.420338983050847,
-      "grad_norm": 1.6989881992340088,
-      "learning_rate": 5.345929648241206e-05,
-      "loss": 5.2943,
-      "step": 46826
-    },
-    {
-      "epoch": 24.420860495436767,
-      "grad_norm": 1.4926403760910034,
-      "learning_rate": 5.3458291457286435e-05,
-      "loss": 5.4415,
-      "step": 46827
-    },
-    {
-      "epoch": 24.421382007822686,
-      "grad_norm": 1.5819379091262817,
-      "learning_rate": 5.34572864321608e-05,
-      "loss": 4.8879,
-      "step": 46828
-    },
-    {
-      "epoch": 24.421903520208605,
-      "grad_norm": 1.5946037769317627,
-      "learning_rate": 5.345628140703518e-05,
-      "loss": 5.3622,
-      "step": 46829
-    },
-    {
-      "epoch": 24.422425032594525,
-      "grad_norm": 1.4274275302886963,
-      "learning_rate": 5.345527638190955e-05,
-      "loss": 5.4451,
-      "step": 46830
-    },
-    {
-      "epoch": 24.422946544980444,
-      "grad_norm": 1.4762616157531738,
-      "learning_rate": 5.3454271356783926e-05,
-      "loss": 5.5169,
-      "step": 46831
-    },
-    {
-      "epoch": 24.423468057366364,
-      "grad_norm": 1.5574125051498413,
-      "learning_rate": 5.345326633165829e-05,
-      "loss": 5.2595,
-      "step": 46832
-    },
-    {
-      "epoch": 24.423989569752283,
-      "grad_norm": 1.628781795501709,
-      "learning_rate": 5.345226130653267e-05,
-      "loss": 5.3586,
-      "step": 46833
-    },
-    {
-      "epoch": 24.4245110821382,
-      "grad_norm": 1.5415176153182983,
-      "learning_rate": 5.345125628140704e-05,
-      "loss": 5.0242,
-      "step": 46834
-    },
-    {
-      "epoch": 24.42503259452412,
-      "grad_norm": 1.6040716171264648,
-      "learning_rate": 5.345025125628142e-05,
-      "loss": 5.0274,
-      "step": 46835
-    },
-    {
-      "epoch": 24.425554106910038,
-      "grad_norm": 1.6169217824935913,
-      "learning_rate": 5.344924623115578e-05,
-      "loss": 5.2575,
-      "step": 46836
-    },
-    {
-      "epoch": 24.426075619295958,
-      "grad_norm": 1.4125640392303467,
-      "learning_rate": 5.344824120603016e-05,
-      "loss": 5.4518,
-      "step": 46837
-    },
-    {
-      "epoch": 24.426597131681877,
-      "grad_norm": 1.5681970119476318,
-      "learning_rate": 5.3447236180904524e-05,
-      "loss": 5.2223,
-      "step": 46838
-    },
-    {
-      "epoch": 24.427118644067797,
-      "grad_norm": 1.7159295082092285,
-      "learning_rate": 5.3446231155778895e-05,
-      "loss": 5.001,
-      "step": 46839
-    },
-    {
-      "epoch": 24.427640156453716,
-      "grad_norm": 1.4049465656280518,
-      "learning_rate": 5.344522613065327e-05,
-      "loss": 5.6541,
-      "step": 46840
-    },
-    {
-      "epoch": 24.428161668839635,
-      "grad_norm": 1.5568429231643677,
-      "learning_rate": 5.344422110552764e-05,
-      "loss": 5.3861,
-      "step": 46841
-    },
-    {
-      "epoch": 24.428683181225555,
-      "grad_norm": 1.460735559463501,
-      "learning_rate": 5.3443216080402015e-05,
-      "loss": 5.464,
-      "step": 46842
-    },
-    {
-      "epoch": 24.429204693611474,
-      "grad_norm": 1.6145219802856445,
-      "learning_rate": 5.344221105527638e-05,
-      "loss": 5.4572,
-      "step": 46843
-    },
-    {
-      "epoch": 24.429726205997394,
-      "grad_norm": 1.638950228691101,
-      "learning_rate": 5.344120603015076e-05,
-      "loss": 4.7213,
-      "step": 46844
-    },
-    {
-      "epoch": 24.430247718383313,
-      "grad_norm": 1.4625078439712524,
-      "learning_rate": 5.344020100502513e-05,
-      "loss": 5.6535,
-      "step": 46845
-    },
-    {
-      "epoch": 24.43076923076923,
-      "grad_norm": 1.4761309623718262,
-      "learning_rate": 5.3439195979899506e-05,
-      "loss": 5.3702,
-      "step": 46846
-    },
-    {
-      "epoch": 24.43129074315515,
-      "grad_norm": 1.4490272998809814,
-      "learning_rate": 5.343819095477387e-05,
-      "loss": 5.4501,
-      "step": 46847
-    },
-    {
-      "epoch": 24.431812255541068,
-      "grad_norm": 1.5501612424850464,
-      "learning_rate": 5.343718592964825e-05,
-      "loss": 5.2548,
-      "step": 46848
-    },
-    {
-      "epoch": 24.432333767926988,
-      "grad_norm": 1.5925376415252686,
-      "learning_rate": 5.343618090452261e-05,
-      "loss": 4.67,
-      "step": 46849
-    },
-    {
-      "epoch": 24.432855280312907,
-      "grad_norm": 1.6174564361572266,
-      "learning_rate": 5.343517587939699e-05,
-      "loss": 5.0246,
-      "step": 46850
-    },
-    {
-      "epoch": 24.433376792698827,
-      "grad_norm": 1.4380359649658203,
-      "learning_rate": 5.343417085427136e-05,
-      "loss": 5.0052,
-      "step": 46851
-    },
-    {
-      "epoch": 24.433898305084746,
-      "grad_norm": 1.535888671875,
-      "learning_rate": 5.3433165829145725e-05,
-      "loss": 5.1546,
-      "step": 46852
-    },
-    {
-      "epoch": 24.434419817470665,
-      "grad_norm": 1.5637187957763672,
-      "learning_rate": 5.34321608040201e-05,
-      "loss": 5.2682,
-      "step": 46853
-    },
-    {
-      "epoch": 24.434941329856585,
-      "grad_norm": 1.5596562623977661,
-      "learning_rate": 5.343115577889447e-05,
-      "loss": 5.0149,
-      "step": 46854
-    },
-    {
-      "epoch": 24.435462842242504,
-      "grad_norm": 1.5106682777404785,
-      "learning_rate": 5.3430150753768845e-05,
-      "loss": 5.5416,
-      "step": 46855
-    },
-    {
-      "epoch": 24.435984354628424,
-      "grad_norm": 1.4926294088363647,
-      "learning_rate": 5.3429145728643216e-05,
-      "loss": 5.4728,
-      "step": 46856
-    },
-    {
-      "epoch": 24.43650586701434,
-      "grad_norm": 1.5117021799087524,
-      "learning_rate": 5.3428140703517594e-05,
-      "loss": 5.2663,
-      "step": 46857
-    },
-    {
-      "epoch": 24.43702737940026,
-      "grad_norm": 1.6741130352020264,
-      "learning_rate": 5.342713567839196e-05,
-      "loss": 5.19,
-      "step": 46858
-    },
-    {
-      "epoch": 24.43754889178618,
-      "grad_norm": 1.6343400478363037,
-      "learning_rate": 5.3426130653266336e-05,
-      "loss": 5.0055,
-      "step": 46859
-    },
-    {
-      "epoch": 24.438070404172098,
-      "grad_norm": 1.450929045677185,
-      "learning_rate": 5.342512562814071e-05,
-      "loss": 5.5479,
-      "step": 46860
-    },
-    {
-      "epoch": 24.438591916558018,
-      "grad_norm": 1.5731208324432373,
-      "learning_rate": 5.3424120603015085e-05,
-      "loss": 5.2433,
-      "step": 46861
-    },
-    {
-      "epoch": 24.439113428943937,
-      "grad_norm": 1.558265209197998,
-      "learning_rate": 5.342311557788945e-05,
-      "loss": 5.6118,
-      "step": 46862
-    },
-    {
-      "epoch": 24.439634941329857,
-      "grad_norm": 1.4933276176452637,
-      "learning_rate": 5.342211055276383e-05,
-      "loss": 5.5118,
-      "step": 46863
-    },
-    {
-      "epoch": 24.440156453715776,
-      "grad_norm": 1.555484414100647,
-      "learning_rate": 5.342110552763819e-05,
-      "loss": 5.1801,
-      "step": 46864
-    },
-    {
-      "epoch": 24.440677966101696,
-      "grad_norm": 1.7526689767837524,
-      "learning_rate": 5.342010050251256e-05,
-      "loss": 4.8368,
-      "step": 46865
-    },
-    {
-      "epoch": 24.441199478487615,
-      "grad_norm": 1.6551331281661987,
-      "learning_rate": 5.341909547738694e-05,
-      "loss": 4.9357,
-      "step": 46866
-    },
-    {
-      "epoch": 24.441720990873534,
-      "grad_norm": 1.5546497106552124,
-      "learning_rate": 5.3418090452261304e-05,
-      "loss": 5.3863,
-      "step": 46867
-    },
-    {
-      "epoch": 24.442242503259454,
-      "grad_norm": 1.5204728841781616,
-      "learning_rate": 5.341708542713568e-05,
-      "loss": 5.3732,
-      "step": 46868
-    },
-    {
-      "epoch": 24.442764015645373,
-      "grad_norm": 1.4931373596191406,
-      "learning_rate": 5.3416080402010047e-05,
-      "loss": 5.4782,
-      "step": 46869
-    },
-    {
-      "epoch": 24.44328552803129,
-      "grad_norm": 1.579802393913269,
-      "learning_rate": 5.3415075376884424e-05,
-      "loss": 4.8919,
-      "step": 46870
-    },
-    {
-      "epoch": 24.44380704041721,
-      "grad_norm": 1.5080891847610474,
-      "learning_rate": 5.3414070351758795e-05,
-      "loss": 5.0857,
-      "step": 46871
-    },
-    {
-      "epoch": 24.444328552803128,
-      "grad_norm": 1.5446221828460693,
-      "learning_rate": 5.341306532663317e-05,
-      "loss": 5.5737,
-      "step": 46872
-    },
-    {
-      "epoch": 24.444850065189048,
-      "grad_norm": 1.5653200149536133,
-      "learning_rate": 5.341206030150754e-05,
-      "loss": 5.3569,
-      "step": 46873
-    },
-    {
-      "epoch": 24.445371577574967,
-      "grad_norm": 1.5013320446014404,
-      "learning_rate": 5.3411055276381915e-05,
-      "loss": 5.2634,
-      "step": 46874
-    },
-    {
-      "epoch": 24.445893089960887,
-      "grad_norm": 1.560054898262024,
-      "learning_rate": 5.341005025125628e-05,
-      "loss": 5.2686,
-      "step": 46875
-    },
-    {
-      "epoch": 24.446414602346806,
-      "grad_norm": 1.5474975109100342,
-      "learning_rate": 5.340904522613066e-05,
-      "loss": 5.3209,
-      "step": 46876
-    },
-    {
-      "epoch": 24.446936114732726,
-      "grad_norm": 1.6157587766647339,
-      "learning_rate": 5.340804020100503e-05,
-      "loss": 4.9168,
-      "step": 46877
-    },
-    {
-      "epoch": 24.447457627118645,
-      "grad_norm": 1.5772098302841187,
-      "learning_rate": 5.340703517587939e-05,
-      "loss": 4.9507,
-      "step": 46878
-    },
-    {
-      "epoch": 24.447979139504564,
-      "grad_norm": 1.6372140645980835,
-      "learning_rate": 5.340603015075377e-05,
-      "loss": 4.8728,
-      "step": 46879
-    },
-    {
-      "epoch": 24.448500651890484,
-      "grad_norm": 1.5911861658096313,
-      "learning_rate": 5.340502512562814e-05,
-      "loss": 5.1891,
-      "step": 46880
-    },
-    {
-      "epoch": 24.4490221642764,
-      "grad_norm": 1.6129380464553833,
-      "learning_rate": 5.340402010050252e-05,
-      "loss": 5.1805,
-      "step": 46881
-    },
-    {
-      "epoch": 24.44954367666232,
-      "grad_norm": 1.608136773109436,
-      "learning_rate": 5.3403015075376884e-05,
-      "loss": 5.1125,
-      "step": 46882
-    },
-    {
-      "epoch": 24.45006518904824,
-      "grad_norm": 1.6620815992355347,
-      "learning_rate": 5.340201005025126e-05,
-      "loss": 4.864,
-      "step": 46883
-    },
-    {
-      "epoch": 24.45058670143416,
-      "grad_norm": 1.5843944549560547,
-      "learning_rate": 5.3401005025125626e-05,
-      "loss": 4.8988,
-      "step": 46884
-    },
-    {
-      "epoch": 24.451108213820078,
-      "grad_norm": 1.5106269121170044,
-      "learning_rate": 5.3400000000000004e-05,
-      "loss": 5.4836,
-      "step": 46885
-    },
-    {
-      "epoch": 24.451629726205997,
-      "grad_norm": 1.4851598739624023,
-      "learning_rate": 5.3398994974874375e-05,
-      "loss": 4.9493,
-      "step": 46886
-    },
-    {
-      "epoch": 24.452151238591917,
-      "grad_norm": 1.4578101634979248,
-      "learning_rate": 5.339798994974875e-05,
-      "loss": 5.5683,
-      "step": 46887
-    },
-    {
-      "epoch": 24.452672750977836,
-      "grad_norm": 1.584368109703064,
-      "learning_rate": 5.339698492462312e-05,
-      "loss": 5.3785,
-      "step": 46888
-    },
-    {
-      "epoch": 24.453194263363756,
-      "grad_norm": 1.4908372163772583,
-      "learning_rate": 5.3395979899497495e-05,
-      "loss": 5.4653,
-      "step": 46889
-    },
-    {
-      "epoch": 24.453715775749675,
-      "grad_norm": 1.5987317562103271,
-      "learning_rate": 5.339497487437186e-05,
-      "loss": 5.2023,
-      "step": 46890
-    },
-    {
-      "epoch": 24.454237288135594,
-      "grad_norm": 1.4295525550842285,
-      "learning_rate": 5.339396984924624e-05,
-      "loss": 5.553,
-      "step": 46891
-    },
-    {
-      "epoch": 24.454758800521514,
-      "grad_norm": 1.5404844284057617,
-      "learning_rate": 5.339296482412061e-05,
-      "loss": 5.0238,
-      "step": 46892
-    },
-    {
-      "epoch": 24.45528031290743,
-      "grad_norm": 1.549845576286316,
-      "learning_rate": 5.339195979899497e-05,
-      "loss": 4.8637,
-      "step": 46893
-    },
-    {
-      "epoch": 24.45580182529335,
-      "grad_norm": 1.541495442390442,
-      "learning_rate": 5.339095477386935e-05,
-      "loss": 5.3051,
-      "step": 46894
-    },
-    {
-      "epoch": 24.45632333767927,
-      "grad_norm": 1.4559043645858765,
-      "learning_rate": 5.3389949748743714e-05,
-      "loss": 4.6441,
-      "step": 46895
-    },
-    {
-      "epoch": 24.45684485006519,
-      "grad_norm": 1.6433387994766235,
-      "learning_rate": 5.338894472361809e-05,
-      "loss": 4.6646,
-      "step": 46896
-    },
-    {
-      "epoch": 24.457366362451108,
-      "grad_norm": 1.5289877653121948,
-      "learning_rate": 5.338793969849246e-05,
-      "loss": 5.2818,
-      "step": 46897
-    },
-    {
-      "epoch": 24.457887874837027,
-      "grad_norm": 1.5622875690460205,
-      "learning_rate": 5.338693467336684e-05,
-      "loss": 4.9651,
-      "step": 46898
-    },
-    {
-      "epoch": 24.458409387222947,
-      "grad_norm": 1.4806311130523682,
-      "learning_rate": 5.3385929648241205e-05,
-      "loss": 5.5735,
-      "step": 46899
-    },
-    {
-      "epoch": 24.458930899608866,
-      "grad_norm": 1.6699057817459106,
-      "learning_rate": 5.338492462311558e-05,
-      "loss": 5.2577,
-      "step": 46900
-    },
-    {
-      "epoch": 24.459452411994786,
-      "grad_norm": 1.5258805751800537,
-      "learning_rate": 5.3383919597989954e-05,
-      "loss": 4.7491,
-      "step": 46901
-    },
-    {
-      "epoch": 24.459973924380705,
-      "grad_norm": 1.5841914415359497,
-      "learning_rate": 5.338291457286433e-05,
-      "loss": 5.2938,
-      "step": 46902
-    },
-    {
-      "epoch": 24.460495436766625,
-      "grad_norm": 1.5252264738082886,
-      "learning_rate": 5.3381909547738696e-05,
-      "loss": 5.6679,
-      "step": 46903
-    },
-    {
-      "epoch": 24.461016949152544,
-      "grad_norm": 1.502155065536499,
-      "learning_rate": 5.3380904522613074e-05,
-      "loss": 5.5332,
-      "step": 46904
-    },
-    {
-      "epoch": 24.46153846153846,
-      "grad_norm": 1.5513098239898682,
-      "learning_rate": 5.337989949748744e-05,
-      "loss": 5.1876,
-      "step": 46905
-    },
-    {
-      "epoch": 24.46205997392438,
-      "grad_norm": 1.6216298341751099,
-      "learning_rate": 5.337889447236181e-05,
-      "loss": 4.5304,
-      "step": 46906
-    },
-    {
-      "epoch": 24.4625814863103,
-      "grad_norm": 1.6283618211746216,
-      "learning_rate": 5.337788944723619e-05,
-      "loss": 4.919,
-      "step": 46907
-    },
-    {
-      "epoch": 24.46310299869622,
-      "grad_norm": 1.5621066093444824,
-      "learning_rate": 5.337688442211055e-05,
-      "loss": 5.4725,
-      "step": 46908
-    },
-    {
-      "epoch": 24.463624511082138,
-      "grad_norm": 1.5952681303024292,
-      "learning_rate": 5.337587939698493e-05,
-      "loss": 5.4109,
-      "step": 46909
-    },
-    {
-      "epoch": 24.464146023468057,
-      "grad_norm": 1.5695478916168213,
-      "learning_rate": 5.3374874371859294e-05,
-      "loss": 5.464,
-      "step": 46910
-    },
-    {
-      "epoch": 24.464667535853977,
-      "grad_norm": 1.5776935815811157,
-      "learning_rate": 5.337386934673367e-05,
-      "loss": 5.5859,
-      "step": 46911
-    },
-    {
-      "epoch": 24.465189048239896,
-      "grad_norm": 1.7134361267089844,
-      "learning_rate": 5.337286432160804e-05,
-      "loss": 4.9267,
-      "step": 46912
-    },
-    {
-      "epoch": 24.465710560625816,
-      "grad_norm": 1.5193932056427002,
-      "learning_rate": 5.337185929648242e-05,
-      "loss": 5.3485,
-      "step": 46913
-    },
-    {
-      "epoch": 24.466232073011735,
-      "grad_norm": 1.6123600006103516,
-      "learning_rate": 5.3370854271356785e-05,
-      "loss": 5.5001,
-      "step": 46914
-    },
-    {
-      "epoch": 24.466753585397655,
-      "grad_norm": 1.5536575317382812,
-      "learning_rate": 5.336984924623116e-05,
-      "loss": 5.5449,
-      "step": 46915
-    },
-    {
-      "epoch": 24.467275097783574,
-      "grad_norm": 1.539025902748108,
-      "learning_rate": 5.336884422110553e-05,
-      "loss": 5.2784,
-      "step": 46916
-    },
-    {
-      "epoch": 24.46779661016949,
-      "grad_norm": 1.5732706785202026,
-      "learning_rate": 5.3367839195979904e-05,
-      "loss": 4.9103,
-      "step": 46917
-    },
-    {
-      "epoch": 24.46831812255541,
-      "grad_norm": 1.5040841102600098,
-      "learning_rate": 5.3366834170854276e-05,
-      "loss": 4.89,
-      "step": 46918
-    },
-    {
-      "epoch": 24.46883963494133,
-      "grad_norm": 1.6116715669631958,
-      "learning_rate": 5.336582914572864e-05,
-      "loss": 5.4721,
-      "step": 46919
-    },
-    {
-      "epoch": 24.46936114732725,
-      "grad_norm": 1.5664962530136108,
-      "learning_rate": 5.336482412060302e-05,
-      "loss": 5.4139,
-      "step": 46920
-    },
-    {
-      "epoch": 24.469882659713168,
-      "grad_norm": 1.6272259950637817,
-      "learning_rate": 5.336381909547739e-05,
-      "loss": 5.4449,
-      "step": 46921
-    },
-    {
-      "epoch": 24.470404172099087,
-      "grad_norm": 1.6525672674179077,
-      "learning_rate": 5.3362814070351767e-05,
-      "loss": 4.6873,
-      "step": 46922
-    },
-    {
-      "epoch": 24.470925684485007,
-      "grad_norm": 1.5854500532150269,
-      "learning_rate": 5.336180904522613e-05,
-      "loss": 5.1608,
-      "step": 46923
-    },
-    {
-      "epoch": 24.471447196870926,
-      "grad_norm": 1.4764102697372437,
-      "learning_rate": 5.336080402010051e-05,
-      "loss": 4.9864,
-      "step": 46924
-    },
-    {
-      "epoch": 24.471968709256846,
-      "grad_norm": 1.5040793418884277,
-      "learning_rate": 5.335979899497487e-05,
-      "loss": 5.4397,
-      "step": 46925
-    },
-    {
-      "epoch": 24.472490221642765,
-      "grad_norm": 1.5976850986480713,
-      "learning_rate": 5.335879396984925e-05,
-      "loss": 5.228,
-      "step": 46926
-    },
-    {
-      "epoch": 24.473011734028685,
-      "grad_norm": 1.5677300691604614,
-      "learning_rate": 5.335778894472362e-05,
-      "loss": 5.0082,
-      "step": 46927
-    },
-    {
-      "epoch": 24.473533246414604,
-      "grad_norm": 1.5198653936386108,
-      "learning_rate": 5.3356783919598e-05,
-      "loss": 5.0508,
-      "step": 46928
-    },
-    {
-      "epoch": 24.47405475880052,
-      "grad_norm": 1.603180170059204,
-      "learning_rate": 5.3355778894472364e-05,
-      "loss": 4.8876,
-      "step": 46929
-    },
-    {
-      "epoch": 24.47457627118644,
-      "grad_norm": 1.5647199153900146,
-      "learning_rate": 5.335477386934674e-05,
-      "loss": 5.1215,
-      "step": 46930
-    },
-    {
-      "epoch": 24.47509778357236,
-      "grad_norm": 1.5535557270050049,
-      "learning_rate": 5.3353768844221106e-05,
-      "loss": 4.9975,
-      "step": 46931
-    },
-    {
-      "epoch": 24.47561929595828,
-      "grad_norm": 1.6753103733062744,
-      "learning_rate": 5.335276381909548e-05,
-      "loss": 5.1001,
-      "step": 46932
-    },
-    {
-      "epoch": 24.476140808344198,
-      "grad_norm": 1.4759730100631714,
-      "learning_rate": 5.3351758793969855e-05,
-      "loss": 5.3686,
-      "step": 46933
-    },
-    {
-      "epoch": 24.476662320730117,
-      "grad_norm": 1.4556455612182617,
-      "learning_rate": 5.335075376884422e-05,
-      "loss": 5.6415,
-      "step": 46934
-    },
-    {
-      "epoch": 24.477183833116037,
-      "grad_norm": 1.5684616565704346,
-      "learning_rate": 5.33497487437186e-05,
-      "loss": 5.4931,
-      "step": 46935
-    },
-    {
-      "epoch": 24.477705345501956,
-      "grad_norm": 1.6388766765594482,
-      "learning_rate": 5.334874371859296e-05,
-      "loss": 5.2926,
-      "step": 46936
-    },
-    {
-      "epoch": 24.478226857887876,
-      "grad_norm": 1.498582124710083,
-      "learning_rate": 5.334773869346734e-05,
-      "loss": 4.901,
-      "step": 46937
-    },
-    {
-      "epoch": 24.478748370273795,
-      "grad_norm": 1.5016286373138428,
-      "learning_rate": 5.334673366834171e-05,
-      "loss": 5.4452,
-      "step": 46938
-    },
-    {
-      "epoch": 24.479269882659715,
-      "grad_norm": 1.6363579034805298,
-      "learning_rate": 5.334572864321609e-05,
-      "loss": 5.0962,
-      "step": 46939
-    },
-    {
-      "epoch": 24.479791395045634,
-      "grad_norm": 1.7017638683319092,
-      "learning_rate": 5.334472361809045e-05,
-      "loss": 5.2828,
-      "step": 46940
-    },
-    {
-      "epoch": 24.48031290743155,
-      "grad_norm": 1.5281842947006226,
-      "learning_rate": 5.334371859296483e-05,
-      "loss": 5.2497,
-      "step": 46941
-    },
-    {
-      "epoch": 24.48083441981747,
-      "grad_norm": 1.6567151546478271,
-      "learning_rate": 5.3342713567839194e-05,
-      "loss": 4.6082,
-      "step": 46942
-    },
-    {
-      "epoch": 24.48135593220339,
-      "grad_norm": 1.4884926080703735,
-      "learning_rate": 5.334170854271357e-05,
-      "loss": 5.4225,
-      "step": 46943
-    },
-    {
-      "epoch": 24.48187744458931,
-      "grad_norm": 1.4401594400405884,
-      "learning_rate": 5.334070351758794e-05,
-      "loss": 5.8562,
-      "step": 46944
-    },
-    {
-      "epoch": 24.482398956975228,
-      "grad_norm": 1.5681110620498657,
-      "learning_rate": 5.333969849246231e-05,
-      "loss": 5.2935,
-      "step": 46945
-    },
-    {
-      "epoch": 24.482920469361147,
-      "grad_norm": 1.611291766166687,
-      "learning_rate": 5.3338693467336685e-05,
-      "loss": 5.1855,
-      "step": 46946
-    },
-    {
-      "epoch": 24.483441981747067,
-      "grad_norm": 1.60612154006958,
-      "learning_rate": 5.3337688442211056e-05,
-      "loss": 5.4226,
-      "step": 46947
-    },
-    {
-      "epoch": 24.483963494132986,
-      "grad_norm": 1.5588703155517578,
-      "learning_rate": 5.3336683417085434e-05,
-      "loss": 5.514,
-      "step": 46948
-    },
-    {
-      "epoch": 24.484485006518906,
-      "grad_norm": 1.473008632659912,
-      "learning_rate": 5.33356783919598e-05,
-      "loss": 5.6299,
-      "step": 46949
-    },
-    {
-      "epoch": 24.485006518904825,
-      "grad_norm": 1.603568434715271,
-      "learning_rate": 5.3334673366834176e-05,
-      "loss": 5.1993,
-      "step": 46950
-    },
-    {
-      "epoch": 24.485528031290745,
-      "grad_norm": 1.6604666709899902,
-      "learning_rate": 5.333366834170854e-05,
-      "loss": 4.9037,
-      "step": 46951
-    },
-    {
-      "epoch": 24.486049543676664,
-      "grad_norm": 1.6940096616744995,
-      "learning_rate": 5.333266331658292e-05,
-      "loss": 5.1333,
-      "step": 46952
-    },
-    {
-      "epoch": 24.48657105606258,
-      "grad_norm": 1.576972484588623,
-      "learning_rate": 5.333165829145729e-05,
-      "loss": 5.0299,
-      "step": 46953
-    },
-    {
-      "epoch": 24.4870925684485,
-      "grad_norm": 1.527289628982544,
-      "learning_rate": 5.333065326633167e-05,
-      "loss": 5.2628,
-      "step": 46954
-    },
-    {
-      "epoch": 24.48761408083442,
-      "grad_norm": 1.542615532875061,
-      "learning_rate": 5.332964824120603e-05,
-      "loss": 5.3624,
-      "step": 46955
-    },
-    {
-      "epoch": 24.48813559322034,
-      "grad_norm": 1.6338961124420166,
-      "learning_rate": 5.332864321608041e-05,
-      "loss": 4.6834,
-      "step": 46956
-    },
-    {
-      "epoch": 24.488657105606258,
-      "grad_norm": 1.4894274473190308,
-      "learning_rate": 5.3327638190954774e-05,
-      "loss": 5.6548,
-      "step": 46957
-    },
-    {
-      "epoch": 24.489178617992177,
-      "grad_norm": 1.579152226448059,
-      "learning_rate": 5.3326633165829145e-05,
-      "loss": 5.3461,
-      "step": 46958
-    },
-    {
-      "epoch": 24.489700130378097,
-      "grad_norm": 1.503467321395874,
-      "learning_rate": 5.332562814070352e-05,
-      "loss": 5.6224,
-      "step": 46959
-    },
-    {
-      "epoch": 24.490221642764016,
-      "grad_norm": 1.669542908668518,
-      "learning_rate": 5.332462311557789e-05,
-      "loss": 5.3379,
-      "step": 46960
-    },
-    {
-      "epoch": 24.490743155149936,
-      "grad_norm": 1.5572022199630737,
-      "learning_rate": 5.3323618090452265e-05,
-      "loss": 5.4071,
-      "step": 46961
-    },
-    {
-      "epoch": 24.491264667535855,
-      "grad_norm": 1.5719079971313477,
-      "learning_rate": 5.332261306532663e-05,
-      "loss": 5.2748,
-      "step": 46962
-    },
-    {
-      "epoch": 24.491786179921775,
-      "grad_norm": 1.5602941513061523,
-      "learning_rate": 5.332160804020101e-05,
-      "loss": 4.9338,
-      "step": 46963
-    },
-    {
-      "epoch": 24.49230769230769,
-      "grad_norm": 1.5654107332229614,
-      "learning_rate": 5.332060301507538e-05,
-      "loss": 5.4114,
-      "step": 46964
-    },
-    {
-      "epoch": 24.49282920469361,
-      "grad_norm": 1.576171636581421,
-      "learning_rate": 5.3319597989949756e-05,
-      "loss": 4.6189,
-      "step": 46965
-    },
-    {
-      "epoch": 24.49335071707953,
-      "grad_norm": 1.5497735738754272,
-      "learning_rate": 5.331859296482412e-05,
-      "loss": 5.1549,
-      "step": 46966
-    },
-    {
-      "epoch": 24.49387222946545,
-      "grad_norm": 1.4493952989578247,
-      "learning_rate": 5.33175879396985e-05,
-      "loss": 5.6189,
-      "step": 46967
-    },
-    {
-      "epoch": 24.49439374185137,
-      "grad_norm": 1.5733602046966553,
-      "learning_rate": 5.331658291457287e-05,
-      "loss": 4.4805,
-      "step": 46968
-    },
-    {
-      "epoch": 24.494915254237288,
-      "grad_norm": 1.5480234622955322,
-      "learning_rate": 5.3315577889447247e-05,
-      "loss": 5.1361,
-      "step": 46969
-    },
-    {
-      "epoch": 24.495436766623207,
-      "grad_norm": 1.4671661853790283,
-      "learning_rate": 5.331457286432161e-05,
-      "loss": 5.3632,
-      "step": 46970
-    },
-    {
-      "epoch": 24.495958279009127,
-      "grad_norm": 2.0938878059387207,
-      "learning_rate": 5.3313567839195975e-05,
-      "loss": 4.5117,
-      "step": 46971
-    },
-    {
-      "epoch": 24.496479791395046,
-      "grad_norm": 1.5747873783111572,
-      "learning_rate": 5.331256281407035e-05,
-      "loss": 5.4546,
-      "step": 46972
-    },
-    {
-      "epoch": 24.497001303780966,
-      "grad_norm": 1.6440975666046143,
-      "learning_rate": 5.3311557788944724e-05,
-      "loss": 4.8502,
-      "step": 46973
-    },
-    {
-      "epoch": 24.497522816166885,
-      "grad_norm": 1.472338080406189,
-      "learning_rate": 5.33105527638191e-05,
-      "loss": 5.4147,
-      "step": 46974
-    },
-    {
-      "epoch": 24.498044328552805,
-      "grad_norm": 1.6443449258804321,
-      "learning_rate": 5.3309547738693466e-05,
-      "loss": 3.8802,
-      "step": 46975
-    },
-    {
-      "epoch": 24.49856584093872,
-      "grad_norm": 1.599511742591858,
-      "learning_rate": 5.3308542713567844e-05,
-      "loss": 5.0704,
-      "step": 46976
-    },
-    {
-      "epoch": 24.49908735332464,
-      "grad_norm": 1.6179391145706177,
-      "learning_rate": 5.330753768844221e-05,
-      "loss": 5.1508,
-      "step": 46977
-    },
-    {
-      "epoch": 24.49960886571056,
-      "grad_norm": 1.5943717956542969,
-      "learning_rate": 5.3306532663316586e-05,
-      "loss": 4.8495,
-      "step": 46978
-    },
-    {
-      "epoch": 24.50013037809648,
-      "grad_norm": 1.5914806127548218,
-      "learning_rate": 5.330552763819096e-05,
-      "loss": 4.9521,
-      "step": 46979
-    },
-    {
-      "epoch": 24.5006518904824,
-      "grad_norm": 1.4852319955825806,
-      "learning_rate": 5.3304522613065335e-05,
-      "loss": 5.5935,
-      "step": 46980
-    },
-    {
-      "epoch": 24.501173402868318,
-      "grad_norm": 1.5313832759857178,
-      "learning_rate": 5.33035175879397e-05,
-      "loss": 5.438,
-      "step": 46981
-    },
-    {
-      "epoch": 24.501694915254237,
-      "grad_norm": 1.5226150751113892,
-      "learning_rate": 5.330251256281408e-05,
-      "loss": 5.165,
-      "step": 46982
-    },
-    {
-      "epoch": 24.502216427640157,
-      "grad_norm": 1.6395223140716553,
-      "learning_rate": 5.330150753768844e-05,
-      "loss": 5.1003,
-      "step": 46983
-    },
-    {
-      "epoch": 24.502737940026076,
-      "grad_norm": 1.5655437707901,
-      "learning_rate": 5.330050251256282e-05,
-      "loss": 5.1513,
-      "step": 46984
-    },
-    {
-      "epoch": 24.503259452411996,
-      "grad_norm": 1.6374090909957886,
-      "learning_rate": 5.329949748743719e-05,
-      "loss": 5.2454,
-      "step": 46985
-    },
-    {
-      "epoch": 24.503780964797915,
-      "grad_norm": 1.6206214427947998,
-      "learning_rate": 5.3298492462311554e-05,
-      "loss": 5.2173,
-      "step": 46986
-    },
-    {
-      "epoch": 24.504302477183835,
-      "grad_norm": 1.5722882747650146,
-      "learning_rate": 5.329748743718593e-05,
-      "loss": 5.0491,
-      "step": 46987
-    },
-    {
-      "epoch": 24.50482398956975,
-      "grad_norm": 1.55980384349823,
-      "learning_rate": 5.32964824120603e-05,
-      "loss": 5.5175,
-      "step": 46988
-    },
-    {
-      "epoch": 24.50534550195567,
-      "grad_norm": 1.5735328197479248,
-      "learning_rate": 5.329547738693468e-05,
-      "loss": 4.7047,
-      "step": 46989
-    },
-    {
-      "epoch": 24.50586701434159,
-      "grad_norm": 1.4001905918121338,
-      "learning_rate": 5.3294472361809045e-05,
-      "loss": 5.4998,
-      "step": 46990
-    },
-    {
-      "epoch": 24.50638852672751,
-      "grad_norm": 1.5956040620803833,
-      "learning_rate": 5.329346733668342e-05,
-      "loss": 5.2431,
-      "step": 46991
-    },
-    {
-      "epoch": 24.50691003911343,
-      "grad_norm": 1.4089510440826416,
-      "learning_rate": 5.329246231155779e-05,
-      "loss": 4.8503,
-      "step": 46992
-    },
-    {
-      "epoch": 24.507431551499348,
-      "grad_norm": 1.5464413166046143,
-      "learning_rate": 5.3291457286432165e-05,
-      "loss": 5.271,
-      "step": 46993
-    },
-    {
-      "epoch": 24.507953063885267,
-      "grad_norm": 1.5495258569717407,
-      "learning_rate": 5.3290452261306536e-05,
-      "loss": 5.0857,
-      "step": 46994
-    },
-    {
-      "epoch": 24.508474576271187,
-      "grad_norm": 1.4467421770095825,
-      "learning_rate": 5.3289447236180914e-05,
-      "loss": 5.4756,
-      "step": 46995
-    },
-    {
-      "epoch": 24.508996088657106,
-      "grad_norm": 1.5310806035995483,
-      "learning_rate": 5.328844221105528e-05,
-      "loss": 5.2246,
-      "step": 46996
-    },
-    {
-      "epoch": 24.509517601043026,
-      "grad_norm": 1.5783644914627075,
-      "learning_rate": 5.3287437185929656e-05,
-      "loss": 5.3798,
-      "step": 46997
-    },
-    {
-      "epoch": 24.510039113428945,
-      "grad_norm": 1.5423732995986938,
-      "learning_rate": 5.328643216080402e-05,
-      "loss": 5.4275,
-      "step": 46998
-    },
-    {
-      "epoch": 24.510560625814865,
-      "grad_norm": 1.4872562885284424,
-      "learning_rate": 5.328542713567839e-05,
-      "loss": 5.5408,
-      "step": 46999
-    },
-    {
-      "epoch": 24.51108213820078,
-      "grad_norm": 1.6110937595367432,
-      "learning_rate": 5.328442211055277e-05,
-      "loss": 5.2964,
-      "step": 47000
-    },
-    {
-      "epoch": 24.5116036505867,
-      "grad_norm": 1.4966665506362915,
-      "learning_rate": 5.3283417085427134e-05,
-      "loss": 4.9393,
-      "step": 47001
-    },
-    {
-      "epoch": 24.51212516297262,
-      "grad_norm": 1.6254503726959229,
-      "learning_rate": 5.328241206030151e-05,
-      "loss": 4.9367,
-      "step": 47002
-    },
-    {
-      "epoch": 24.51264667535854,
-      "grad_norm": 1.672582745552063,
-      "learning_rate": 5.3281407035175876e-05,
-      "loss": 5.302,
-      "step": 47003
-    },
-    {
-      "epoch": 24.51316818774446,
-      "grad_norm": 1.5385371446609497,
-      "learning_rate": 5.3280402010050254e-05,
-      "loss": 5.0354,
-      "step": 47004
-    },
-    {
-      "epoch": 24.513689700130378,
-      "grad_norm": 1.5774885416030884,
-      "learning_rate": 5.3279396984924625e-05,
-      "loss": 5.3679,
-      "step": 47005
-    },
-    {
-      "epoch": 24.514211212516297,
-      "grad_norm": 1.4294776916503906,
-      "learning_rate": 5.3278391959799e-05,
-      "loss": 4.8923,
-      "step": 47006
-    },
-    {
-      "epoch": 24.514732724902217,
-      "grad_norm": 1.561663269996643,
-      "learning_rate": 5.327738693467337e-05,
-      "loss": 5.4225,
-      "step": 47007
-    },
-    {
-      "epoch": 24.515254237288136,
-      "grad_norm": 1.5611156225204468,
-      "learning_rate": 5.3276381909547745e-05,
-      "loss": 4.7216,
-      "step": 47008
-    },
-    {
-      "epoch": 24.515775749674056,
-      "grad_norm": 1.597525715827942,
-      "learning_rate": 5.327537688442211e-05,
-      "loss": 5.2043,
-      "step": 47009
-    },
-    {
-      "epoch": 24.516297262059975,
-      "grad_norm": 1.53656005859375,
-      "learning_rate": 5.327437185929649e-05,
-      "loss": 5.1759,
-      "step": 47010
-    },
-    {
-      "epoch": 24.516818774445895,
-      "grad_norm": 1.551531195640564,
-      "learning_rate": 5.327336683417086e-05,
-      "loss": 5.0734,
-      "step": 47011
-    },
-    {
-      "epoch": 24.51734028683181,
-      "grad_norm": 1.57008957862854,
-      "learning_rate": 5.327236180904522e-05,
-      "loss": 5.32,
-      "step": 47012
-    },
-    {
-      "epoch": 24.51786179921773,
-      "grad_norm": 1.5695199966430664,
-      "learning_rate": 5.32713567839196e-05,
-      "loss": 5.2148,
-      "step": 47013
-    },
-    {
-      "epoch": 24.51838331160365,
-      "grad_norm": 1.904607892036438,
-      "learning_rate": 5.327035175879397e-05,
-      "loss": 4.9105,
-      "step": 47014
-    },
-    {
-      "epoch": 24.51890482398957,
-      "grad_norm": 1.6498414278030396,
-      "learning_rate": 5.326934673366835e-05,
-      "loss": 4.965,
-      "step": 47015
-    },
-    {
-      "epoch": 24.51942633637549,
-      "grad_norm": 1.5195074081420898,
-      "learning_rate": 5.326834170854271e-05,
-      "loss": 5.2548,
-      "step": 47016
-    },
-    {
-      "epoch": 24.519947848761408,
-      "grad_norm": 1.530997395515442,
-      "learning_rate": 5.326733668341709e-05,
-      "loss": 5.5183,
-      "step": 47017
-    },
-    {
-      "epoch": 24.520469361147327,
-      "grad_norm": 1.640350580215454,
-      "learning_rate": 5.3266331658291455e-05,
-      "loss": 4.511,
-      "step": 47018
-    },
-    {
-      "epoch": 24.520990873533247,
-      "grad_norm": 1.6884610652923584,
-      "learning_rate": 5.326532663316583e-05,
-      "loss": 4.744,
-      "step": 47019
-    },
-    {
-      "epoch": 24.521512385919166,
-      "grad_norm": 1.608212947845459,
-      "learning_rate": 5.3264321608040204e-05,
-      "loss": 4.7739,
-      "step": 47020
-    },
-    {
-      "epoch": 24.522033898305086,
-      "grad_norm": 1.5452059507369995,
-      "learning_rate": 5.326331658291458e-05,
-      "loss": 5.4324,
-      "step": 47021
-    },
-    {
-      "epoch": 24.522555410691005,
-      "grad_norm": 1.6619318723678589,
-      "learning_rate": 5.3262311557788946e-05,
-      "loss": 5.1558,
-      "step": 47022
-    },
-    {
-      "epoch": 24.523076923076925,
-      "grad_norm": 1.5675123929977417,
-      "learning_rate": 5.3261306532663324e-05,
-      "loss": 5.4826,
-      "step": 47023
-    },
-    {
-      "epoch": 24.52359843546284,
-      "grad_norm": 1.5339269638061523,
-      "learning_rate": 5.326030150753769e-05,
-      "loss": 5.4646,
-      "step": 47024
-    },
-    {
-      "epoch": 24.52411994784876,
-      "grad_norm": 1.5070695877075195,
-      "learning_rate": 5.325929648241206e-05,
-      "loss": 5.389,
-      "step": 47025
-    },
-    {
-      "epoch": 24.52464146023468,
-      "grad_norm": 1.5428792238235474,
-      "learning_rate": 5.325829145728644e-05,
-      "loss": 5.4542,
-      "step": 47026
-    },
-    {
-      "epoch": 24.5251629726206,
-      "grad_norm": 1.47152841091156,
-      "learning_rate": 5.32572864321608e-05,
-      "loss": 5.2133,
-      "step": 47027
-    },
-    {
-      "epoch": 24.52568448500652,
-      "grad_norm": 1.6174442768096924,
-      "learning_rate": 5.325628140703518e-05,
-      "loss": 5.3646,
-      "step": 47028
-    },
-    {
-      "epoch": 24.526205997392438,
-      "grad_norm": 1.4819834232330322,
-      "learning_rate": 5.3255276381909544e-05,
-      "loss": 5.4802,
-      "step": 47029
-    },
-    {
-      "epoch": 24.526727509778357,
-      "grad_norm": 1.5882633924484253,
-      "learning_rate": 5.325427135678392e-05,
-      "loss": 5.0458,
-      "step": 47030
-    },
-    {
-      "epoch": 24.527249022164277,
-      "grad_norm": 1.5190577507019043,
-      "learning_rate": 5.325326633165829e-05,
-      "loss": 4.7818,
-      "step": 47031
-    },
-    {
-      "epoch": 24.527770534550196,
-      "grad_norm": 1.5614604949951172,
-      "learning_rate": 5.325226130653267e-05,
-      "loss": 4.9924,
-      "step": 47032
-    },
-    {
-      "epoch": 24.528292046936116,
-      "grad_norm": 1.5994915962219238,
-      "learning_rate": 5.3251256281407035e-05,
-      "loss": 5.3276,
-      "step": 47033
-    },
-    {
-      "epoch": 24.528813559322035,
-      "grad_norm": 1.6464288234710693,
-      "learning_rate": 5.325025125628141e-05,
-      "loss": 4.6116,
-      "step": 47034
-    },
-    {
-      "epoch": 24.529335071707955,
-      "grad_norm": 1.488739013671875,
-      "learning_rate": 5.3249246231155783e-05,
-      "loss": 5.2183,
-      "step": 47035
-    },
-    {
-      "epoch": 24.52985658409387,
-      "grad_norm": 1.572550892829895,
-      "learning_rate": 5.324824120603016e-05,
-      "loss": 5.3221,
-      "step": 47036
-    },
-    {
-      "epoch": 24.53037809647979,
-      "grad_norm": 1.6037341356277466,
-      "learning_rate": 5.3247236180904526e-05,
-      "loss": 5.4404,
-      "step": 47037
-    },
-    {
-      "epoch": 24.53089960886571,
-      "grad_norm": 1.5635998249053955,
-      "learning_rate": 5.324623115577889e-05,
-      "loss": 5.3136,
-      "step": 47038
-    },
-    {
-      "epoch": 24.53142112125163,
-      "grad_norm": 1.643830418586731,
-      "learning_rate": 5.324522613065327e-05,
-      "loss": 5.7225,
-      "step": 47039
-    },
-    {
-      "epoch": 24.53194263363755,
-      "grad_norm": 1.6241387128829956,
-      "learning_rate": 5.324422110552764e-05,
-      "loss": 4.8485,
-      "step": 47040
-    },
-    {
-      "epoch": 24.532464146023468,
-      "grad_norm": 1.5629453659057617,
-      "learning_rate": 5.3243216080402017e-05,
-      "loss": 5.4526,
-      "step": 47041
-    },
-    {
-      "epoch": 24.532985658409387,
-      "grad_norm": 1.5443285703659058,
-      "learning_rate": 5.324221105527638e-05,
-      "loss": 5.0476,
-      "step": 47042
-    },
-    {
-      "epoch": 24.533507170795307,
-      "grad_norm": 1.6248217821121216,
-      "learning_rate": 5.324120603015076e-05,
-      "loss": 5.2419,
-      "step": 47043
-    },
-    {
-      "epoch": 24.534028683181226,
-      "grad_norm": 1.6189706325531006,
-      "learning_rate": 5.324020100502512e-05,
-      "loss": 5.2651,
-      "step": 47044
-    },
-    {
-      "epoch": 24.534550195567146,
-      "grad_norm": 1.5307016372680664,
-      "learning_rate": 5.32391959798995e-05,
-      "loss": 5.1806,
-      "step": 47045
-    },
-    {
-      "epoch": 24.535071707953065,
-      "grad_norm": 1.5677441358566284,
-      "learning_rate": 5.323819095477387e-05,
-      "loss": 5.1639,
-      "step": 47046
-    },
-    {
-      "epoch": 24.53559322033898,
-      "grad_norm": 1.5195235013961792,
-      "learning_rate": 5.323718592964825e-05,
-      "loss": 5.376,
-      "step": 47047
-    },
-    {
-      "epoch": 24.5361147327249,
-      "grad_norm": 1.6116783618927002,
-      "learning_rate": 5.3236180904522614e-05,
-      "loss": 4.935,
-      "step": 47048
-    },
-    {
-      "epoch": 24.53663624511082,
-      "grad_norm": 1.5301162004470825,
-      "learning_rate": 5.323517587939699e-05,
-      "loss": 5.4768,
-      "step": 47049
-    },
-    {
-      "epoch": 24.53715775749674,
-      "grad_norm": 1.6664265394210815,
-      "learning_rate": 5.3234170854271356e-05,
-      "loss": 5.2695,
-      "step": 47050
-    },
-    {
-      "epoch": 24.53767926988266,
-      "grad_norm": 1.705824613571167,
-      "learning_rate": 5.323316582914573e-05,
-      "loss": 4.9167,
-      "step": 47051
-    },
-    {
-      "epoch": 24.53820078226858,
-      "grad_norm": 1.6121060848236084,
-      "learning_rate": 5.3232160804020105e-05,
-      "loss": 5.4096,
-      "step": 47052
-    },
-    {
-      "epoch": 24.538722294654498,
-      "grad_norm": 1.6011855602264404,
-      "learning_rate": 5.323115577889447e-05,
-      "loss": 5.1289,
-      "step": 47053
-    },
-    {
-      "epoch": 24.539243807040418,
-      "grad_norm": 1.5259665250778198,
-      "learning_rate": 5.323015075376885e-05,
-      "loss": 5.5757,
-      "step": 47054
-    },
-    {
-      "epoch": 24.539765319426337,
-      "grad_norm": 1.5274022817611694,
-      "learning_rate": 5.322914572864322e-05,
-      "loss": 5.1348,
-      "step": 47055
-    },
-    {
-      "epoch": 24.540286831812256,
-      "grad_norm": 1.5356996059417725,
-      "learning_rate": 5.3228140703517596e-05,
-      "loss": 5.4456,
-      "step": 47056
-    },
-    {
-      "epoch": 24.540808344198176,
-      "grad_norm": 1.562803030014038,
-      "learning_rate": 5.322713567839196e-05,
-      "loss": 5.1263,
-      "step": 47057
-    },
-    {
-      "epoch": 24.541329856584095,
-      "grad_norm": 1.5661804676055908,
-      "learning_rate": 5.322613065326634e-05,
-      "loss": 5.6693,
-      "step": 47058
-    },
-    {
-      "epoch": 24.541851368970015,
-      "grad_norm": 1.4674652814865112,
-      "learning_rate": 5.32251256281407e-05,
-      "loss": 5.1049,
-      "step": 47059
-    },
-    {
-      "epoch": 24.54237288135593,
-      "grad_norm": 1.8363059759140015,
-      "learning_rate": 5.322412060301508e-05,
-      "loss": 4.9358,
-      "step": 47060
-    },
-    {
-      "epoch": 24.54289439374185,
-      "grad_norm": 1.4799633026123047,
-      "learning_rate": 5.322311557788945e-05,
-      "loss": 5.1808,
-      "step": 47061
-    },
-    {
-      "epoch": 24.54341590612777,
-      "grad_norm": 1.532515287399292,
-      "learning_rate": 5.322211055276383e-05,
-      "loss": 5.0203,
-      "step": 47062
-    },
-    {
-      "epoch": 24.54393741851369,
-      "grad_norm": 1.574811577796936,
-      "learning_rate": 5.322110552763819e-05,
-      "loss": 5.1823,
-      "step": 47063
-    },
-    {
-      "epoch": 24.54445893089961,
-      "grad_norm": 1.5390830039978027,
-      "learning_rate": 5.322010050251256e-05,
-      "loss": 5.4646,
-      "step": 47064
-    },
-    {
-      "epoch": 24.544980443285528,
-      "grad_norm": 1.6037873029708862,
-      "learning_rate": 5.3219095477386935e-05,
-      "loss": 4.8655,
-      "step": 47065
-    },
-    {
-      "epoch": 24.545501955671448,
-      "grad_norm": 1.6616650819778442,
-      "learning_rate": 5.3218090452261306e-05,
-      "loss": 5.3843,
-      "step": 47066
-    },
-    {
-      "epoch": 24.546023468057367,
-      "grad_norm": 1.6359963417053223,
-      "learning_rate": 5.3217085427135684e-05,
-      "loss": 4.9845,
-      "step": 47067
-    },
-    {
-      "epoch": 24.546544980443286,
-      "grad_norm": 1.556873083114624,
-      "learning_rate": 5.321608040201005e-05,
-      "loss": 5.5559,
-      "step": 47068
-    },
-    {
-      "epoch": 24.547066492829206,
-      "grad_norm": 1.5560230016708374,
-      "learning_rate": 5.3215075376884426e-05,
-      "loss": 5.5444,
-      "step": 47069
-    },
-    {
-      "epoch": 24.547588005215125,
-      "grad_norm": 1.5889887809753418,
-      "learning_rate": 5.321407035175879e-05,
-      "loss": 5.1498,
-      "step": 47070
-    },
-    {
-      "epoch": 24.54810951760104,
-      "grad_norm": 1.5852190256118774,
-      "learning_rate": 5.321306532663317e-05,
-      "loss": 4.8279,
-      "step": 47071
-    },
-    {
-      "epoch": 24.54863102998696,
-      "grad_norm": 1.4976025819778442,
-      "learning_rate": 5.321206030150754e-05,
-      "loss": 5.5543,
-      "step": 47072
-    },
-    {
-      "epoch": 24.54915254237288,
-      "grad_norm": 1.601387619972229,
-      "learning_rate": 5.321105527638192e-05,
-      "loss": 5.1481,
-      "step": 47073
-    },
-    {
-      "epoch": 24.5496740547588,
-      "grad_norm": 1.5435199737548828,
-      "learning_rate": 5.321005025125628e-05,
-      "loss": 5.4969,
-      "step": 47074
-    },
-    {
-      "epoch": 24.55019556714472,
-      "grad_norm": 1.5487394332885742,
-      "learning_rate": 5.320904522613066e-05,
-      "loss": 4.6123,
-      "step": 47075
-    },
-    {
-      "epoch": 24.55071707953064,
-      "grad_norm": 1.6220908164978027,
-      "learning_rate": 5.320804020100503e-05,
-      "loss": 5.0785,
-      "step": 47076
-    },
-    {
-      "epoch": 24.551238591916558,
-      "grad_norm": 1.6250853538513184,
-      "learning_rate": 5.320703517587941e-05,
-      "loss": 5.0159,
-      "step": 47077
-    },
-    {
-      "epoch": 24.551760104302478,
-      "grad_norm": 1.5931196212768555,
-      "learning_rate": 5.320603015075377e-05,
-      "loss": 5.1938,
-      "step": 47078
-    },
-    {
-      "epoch": 24.552281616688397,
-      "grad_norm": 1.4731366634368896,
-      "learning_rate": 5.320502512562814e-05,
-      "loss": 5.3705,
-      "step": 47079
-    },
-    {
-      "epoch": 24.552803129074317,
-      "grad_norm": 1.5003907680511475,
-      "learning_rate": 5.3204020100502515e-05,
-      "loss": 5.0554,
-      "step": 47080
-    },
-    {
-      "epoch": 24.553324641460236,
-      "grad_norm": 1.6551129817962646,
-      "learning_rate": 5.3203015075376886e-05,
-      "loss": 5.0101,
-      "step": 47081
-    },
-    {
-      "epoch": 24.553846153846155,
-      "grad_norm": 1.5449728965759277,
-      "learning_rate": 5.3202010050251264e-05,
-      "loss": 5.5557,
-      "step": 47082
-    },
-    {
-      "epoch": 24.55436766623207,
-      "grad_norm": 1.7259401082992554,
-      "learning_rate": 5.320100502512563e-05,
-      "loss": 4.6966,
-      "step": 47083
-    },
-    {
-      "epoch": 24.55488917861799,
-      "grad_norm": 1.4797067642211914,
-      "learning_rate": 5.3200000000000006e-05,
-      "loss": 5.3778,
-      "step": 47084
-    },
-    {
-      "epoch": 24.55541069100391,
-      "grad_norm": 1.5345033407211304,
-      "learning_rate": 5.319899497487437e-05,
-      "loss": 5.3239,
-      "step": 47085
-    },
-    {
-      "epoch": 24.55593220338983,
-      "grad_norm": 1.6992316246032715,
-      "learning_rate": 5.319798994974875e-05,
-      "loss": 5.1966,
-      "step": 47086
-    },
-    {
-      "epoch": 24.55645371577575,
-      "grad_norm": 1.568390130996704,
-      "learning_rate": 5.319698492462312e-05,
-      "loss": 5.369,
-      "step": 47087
-    },
-    {
-      "epoch": 24.55697522816167,
-      "grad_norm": 1.5634956359863281,
-      "learning_rate": 5.31959798994975e-05,
-      "loss": 5.3814,
-      "step": 47088
-    },
-    {
-      "epoch": 24.557496740547588,
-      "grad_norm": 1.561781644821167,
-      "learning_rate": 5.319497487437186e-05,
-      "loss": 5.3529,
-      "step": 47089
-    },
-    {
-      "epoch": 24.558018252933508,
-      "grad_norm": 1.6695263385772705,
-      "learning_rate": 5.319396984924624e-05,
-      "loss": 5.0248,
-      "step": 47090
-    },
-    {
-      "epoch": 24.558539765319427,
-      "grad_norm": 1.7321109771728516,
-      "learning_rate": 5.31929648241206e-05,
-      "loss": 5.1224,
-      "step": 47091
-    },
-    {
-      "epoch": 24.559061277705347,
-      "grad_norm": 1.6410990953445435,
-      "learning_rate": 5.3191959798994974e-05,
-      "loss": 5.2466,
-      "step": 47092
-    },
-    {
-      "epoch": 24.559582790091266,
-      "grad_norm": 1.5850460529327393,
-      "learning_rate": 5.319095477386935e-05,
-      "loss": 5.0406,
-      "step": 47093
-    },
-    {
-      "epoch": 24.560104302477185,
-      "grad_norm": 1.6447933912277222,
-      "learning_rate": 5.3189949748743716e-05,
-      "loss": 4.9844,
-      "step": 47094
-    },
-    {
-      "epoch": 24.5606258148631,
-      "grad_norm": 1.5462090969085693,
-      "learning_rate": 5.3188944723618094e-05,
-      "loss": 5.3696,
-      "step": 47095
-    },
-    {
-      "epoch": 24.56114732724902,
-      "grad_norm": 1.6251468658447266,
-      "learning_rate": 5.318793969849246e-05,
-      "loss": 5.3425,
-      "step": 47096
-    },
-    {
-      "epoch": 24.56166883963494,
-      "grad_norm": 1.639073371887207,
-      "learning_rate": 5.3186934673366836e-05,
-      "loss": 5.0156,
-      "step": 47097
-    },
-    {
-      "epoch": 24.56219035202086,
-      "grad_norm": 1.6349979639053345,
-      "learning_rate": 5.318592964824121e-05,
-      "loss": 5.0279,
-      "step": 47098
-    },
-    {
-      "epoch": 24.56271186440678,
-      "grad_norm": 1.6422098875045776,
-      "learning_rate": 5.3184924623115585e-05,
-      "loss": 5.1782,
-      "step": 47099
-    },
-    {
-      "epoch": 24.5632333767927,
-      "grad_norm": 1.6538772583007812,
-      "learning_rate": 5.318391959798995e-05,
-      "loss": 5.0991,
-      "step": 47100
-    },
-    {
-      "epoch": 24.563754889178618,
-      "grad_norm": 1.6344274282455444,
-      "learning_rate": 5.318291457286433e-05,
-      "loss": 5.2534,
-      "step": 47101
-    },
-    {
-      "epoch": 24.564276401564538,
-      "grad_norm": 1.5293092727661133,
-      "learning_rate": 5.31819095477387e-05,
-      "loss": 4.9854,
-      "step": 47102
-    },
-    {
-      "epoch": 24.564797913950457,
-      "grad_norm": 1.5208077430725098,
-      "learning_rate": 5.3180904522613076e-05,
-      "loss": 5.4248,
-      "step": 47103
-    },
-    {
-      "epoch": 24.565319426336377,
-      "grad_norm": 1.5405877828598022,
-      "learning_rate": 5.317989949748744e-05,
-      "loss": 5.321,
-      "step": 47104
-    },
-    {
-      "epoch": 24.565840938722296,
-      "grad_norm": 1.4832463264465332,
-      "learning_rate": 5.3178894472361805e-05,
-      "loss": 5.2499,
-      "step": 47105
-    },
-    {
-      "epoch": 24.566362451108215,
-      "grad_norm": 1.541253924369812,
-      "learning_rate": 5.317788944723618e-05,
-      "loss": 5.3368,
-      "step": 47106
-    },
-    {
-      "epoch": 24.56688396349413,
-      "grad_norm": 1.5816291570663452,
-      "learning_rate": 5.317688442211055e-05,
-      "loss": 4.772,
-      "step": 47107
-    },
-    {
-      "epoch": 24.56740547588005,
-      "grad_norm": 1.488073706626892,
-      "learning_rate": 5.317587939698493e-05,
-      "loss": 4.7429,
-      "step": 47108
-    },
-    {
-      "epoch": 24.56792698826597,
-      "grad_norm": 1.603676438331604,
-      "learning_rate": 5.3174874371859295e-05,
-      "loss": 5.2272,
-      "step": 47109
-    },
-    {
-      "epoch": 24.56844850065189,
-      "grad_norm": 1.685495138168335,
-      "learning_rate": 5.317386934673367e-05,
-      "loss": 4.9442,
-      "step": 47110
-    },
-    {
-      "epoch": 24.56897001303781,
-      "grad_norm": 1.5859094858169556,
-      "learning_rate": 5.317286432160804e-05,
-      "loss": 5.2428,
-      "step": 47111
-    },
-    {
-      "epoch": 24.56949152542373,
-      "grad_norm": 1.5397531986236572,
-      "learning_rate": 5.3171859296482415e-05,
-      "loss": 5.2452,
-      "step": 47112
-    },
-    {
-      "epoch": 24.570013037809648,
-      "grad_norm": 1.5330406427383423,
-      "learning_rate": 5.3170854271356786e-05,
-      "loss": 5.435,
-      "step": 47113
-    },
-    {
-      "epoch": 24.570534550195568,
-      "grad_norm": 1.594334602355957,
-      "learning_rate": 5.3169849246231164e-05,
-      "loss": 5.2751,
-      "step": 47114
-    },
-    {
-      "epoch": 24.571056062581487,
-      "grad_norm": 1.499260663986206,
-      "learning_rate": 5.316884422110553e-05,
-      "loss": 5.4131,
-      "step": 47115
-    },
-    {
-      "epoch": 24.571577574967407,
-      "grad_norm": 1.531864881515503,
-      "learning_rate": 5.3167839195979906e-05,
-      "loss": 4.636,
-      "step": 47116
-    },
-    {
-      "epoch": 24.572099087353326,
-      "grad_norm": 1.588198184967041,
-      "learning_rate": 5.316683417085427e-05,
-      "loss": 4.982,
-      "step": 47117
-    },
-    {
-      "epoch": 24.572620599739246,
-      "grad_norm": 1.59525465965271,
-      "learning_rate": 5.316582914572864e-05,
-      "loss": 5.087,
-      "step": 47118
-    },
-    {
-      "epoch": 24.57314211212516,
-      "grad_norm": 1.6753292083740234,
-      "learning_rate": 5.316482412060302e-05,
-      "loss": 5.3368,
-      "step": 47119
-    },
-    {
-      "epoch": 24.57366362451108,
-      "grad_norm": 1.641148328781128,
-      "learning_rate": 5.3163819095477384e-05,
-      "loss": 4.9193,
-      "step": 47120
-    },
-    {
-      "epoch": 24.574185136897,
-      "grad_norm": 1.5460728406906128,
-      "learning_rate": 5.316281407035176e-05,
-      "loss": 5.2461,
-      "step": 47121
-    },
-    {
-      "epoch": 24.57470664928292,
-      "grad_norm": 1.5380750894546509,
-      "learning_rate": 5.316180904522613e-05,
-      "loss": 5.2758,
-      "step": 47122
-    },
-    {
-      "epoch": 24.57522816166884,
-      "grad_norm": 1.5853967666625977,
-      "learning_rate": 5.316080402010051e-05,
-      "loss": 4.9809,
-      "step": 47123
-    },
-    {
-      "epoch": 24.57574967405476,
-      "grad_norm": 1.4907805919647217,
-      "learning_rate": 5.3159798994974875e-05,
-      "loss": 4.5194,
-      "step": 47124
-    },
-    {
-      "epoch": 24.576271186440678,
-      "grad_norm": 1.5488293170928955,
-      "learning_rate": 5.315879396984925e-05,
-      "loss": 5.256,
-      "step": 47125
-    },
-    {
-      "epoch": 24.576792698826598,
-      "grad_norm": 1.6093873977661133,
-      "learning_rate": 5.315778894472362e-05,
-      "loss": 5.623,
-      "step": 47126
-    },
-    {
-      "epoch": 24.577314211212517,
-      "grad_norm": 1.5698741674423218,
-      "learning_rate": 5.3156783919597995e-05,
-      "loss": 5.1486,
-      "step": 47127
-    },
-    {
-      "epoch": 24.577835723598437,
-      "grad_norm": 1.5810225009918213,
-      "learning_rate": 5.3155778894472366e-05,
-      "loss": 5.387,
-      "step": 47128
-    },
-    {
-      "epoch": 24.578357235984356,
-      "grad_norm": 1.5035818815231323,
-      "learning_rate": 5.3154773869346744e-05,
-      "loss": 4.7217,
-      "step": 47129
-    },
-    {
-      "epoch": 24.578878748370272,
-      "grad_norm": 1.5102927684783936,
-      "learning_rate": 5.315376884422111e-05,
-      "loss": 5.1896,
-      "step": 47130
-    },
-    {
-      "epoch": 24.57940026075619,
-      "grad_norm": 1.4402732849121094,
-      "learning_rate": 5.315276381909547e-05,
-      "loss": 5.6598,
-      "step": 47131
-    },
-    {
-      "epoch": 24.57992177314211,
-      "grad_norm": 1.523047685623169,
-      "learning_rate": 5.315175879396985e-05,
-      "loss": 5.1284,
-      "step": 47132
-    },
-    {
-      "epoch": 24.58044328552803,
-      "grad_norm": 1.5802021026611328,
-      "learning_rate": 5.315075376884422e-05,
-      "loss": 5.1942,
-      "step": 47133
-    },
-    {
-      "epoch": 24.58096479791395,
-      "grad_norm": 1.6296401023864746,
-      "learning_rate": 5.31497487437186e-05,
-      "loss": 5.2271,
-      "step": 47134
-    },
-    {
-      "epoch": 24.58148631029987,
-      "grad_norm": 1.530971884727478,
-      "learning_rate": 5.314874371859296e-05,
-      "loss": 5.0256,
-      "step": 47135
-    },
-    {
-      "epoch": 24.58200782268579,
-      "grad_norm": 1.6363250017166138,
-      "learning_rate": 5.314773869346734e-05,
-      "loss": 5.3741,
-      "step": 47136
-    },
-    {
-      "epoch": 24.58252933507171,
-      "grad_norm": 1.5470699071884155,
-      "learning_rate": 5.3146733668341705e-05,
-      "loss": 5.9234,
-      "step": 47137
-    },
-    {
-      "epoch": 24.583050847457628,
-      "grad_norm": 1.463914394378662,
-      "learning_rate": 5.314572864321608e-05,
-      "loss": 5.5869,
-      "step": 47138
-    },
-    {
-      "epoch": 24.583572359843547,
-      "grad_norm": 1.4803458452224731,
-      "learning_rate": 5.3144723618090454e-05,
-      "loss": 5.4209,
-      "step": 47139
-    },
-    {
-      "epoch": 24.584093872229467,
-      "grad_norm": 1.5428608655929565,
-      "learning_rate": 5.314371859296483e-05,
-      "loss": 5.3396,
-      "step": 47140
-    },
-    {
-      "epoch": 24.584615384615386,
-      "grad_norm": 1.56388521194458,
-      "learning_rate": 5.3142713567839196e-05,
-      "loss": 5.0289,
-      "step": 47141
-    },
-    {
-      "epoch": 24.585136897001306,
-      "grad_norm": 1.6431182622909546,
-      "learning_rate": 5.3141708542713574e-05,
-      "loss": 5.4726,
-      "step": 47142
-    },
-    {
-      "epoch": 24.58565840938722,
-      "grad_norm": 1.6031723022460938,
-      "learning_rate": 5.3140703517587945e-05,
-      "loss": 5.2736,
-      "step": 47143
-    },
-    {
-      "epoch": 24.58617992177314,
-      "grad_norm": 1.5232685804367065,
-      "learning_rate": 5.313969849246231e-05,
-      "loss": 5.3398,
-      "step": 47144
-    },
-    {
-      "epoch": 24.58670143415906,
-      "grad_norm": 1.6355420351028442,
-      "learning_rate": 5.313869346733669e-05,
-      "loss": 5.0655,
-      "step": 47145
-    },
-    {
-      "epoch": 24.58722294654498,
-      "grad_norm": 1.623210072517395,
-      "learning_rate": 5.313768844221105e-05,
-      "loss": 5.1625,
-      "step": 47146
-    },
-    {
-      "epoch": 24.5877444589309,
-      "grad_norm": 1.5997458696365356,
-      "learning_rate": 5.313668341708543e-05,
-      "loss": 4.9867,
-      "step": 47147
-    },
-    {
-      "epoch": 24.58826597131682,
-      "grad_norm": 1.5831611156463623,
-      "learning_rate": 5.31356783919598e-05,
-      "loss": 5.1526,
-      "step": 47148
-    },
-    {
-      "epoch": 24.58878748370274,
-      "grad_norm": 1.4715105295181274,
-      "learning_rate": 5.313467336683418e-05,
-      "loss": 5.5879,
-      "step": 47149
-    },
-    {
-      "epoch": 24.589308996088658,
-      "grad_norm": 1.566053867340088,
-      "learning_rate": 5.313366834170854e-05,
-      "loss": 4.9439,
-      "step": 47150
-    },
-    {
-      "epoch": 24.589830508474577,
-      "grad_norm": 1.6109719276428223,
-      "learning_rate": 5.313266331658292e-05,
-      "loss": 5.1987,
-      "step": 47151
-    },
-    {
-      "epoch": 24.590352020860497,
-      "grad_norm": 1.4973137378692627,
-      "learning_rate": 5.3131658291457285e-05,
-      "loss": 5.5718,
-      "step": 47152
-    },
-    {
-      "epoch": 24.590873533246416,
-      "grad_norm": 1.6253700256347656,
-      "learning_rate": 5.313065326633166e-05,
-      "loss": 4.3589,
-      "step": 47153
-    },
-    {
-      "epoch": 24.591395045632332,
-      "grad_norm": 1.5874892473220825,
-      "learning_rate": 5.3129648241206033e-05,
-      "loss": 5.2369,
-      "step": 47154
-    },
-    {
-      "epoch": 24.59191655801825,
-      "grad_norm": 1.5949761867523193,
-      "learning_rate": 5.312864321608041e-05,
-      "loss": 5.3937,
-      "step": 47155
-    },
-    {
-      "epoch": 24.59243807040417,
-      "grad_norm": 1.5417932271957397,
-      "learning_rate": 5.3127638190954776e-05,
-      "loss": 5.4357,
-      "step": 47156
-    },
-    {
-      "epoch": 24.59295958279009,
-      "grad_norm": 1.5004431009292603,
-      "learning_rate": 5.312663316582914e-05,
-      "loss": 5.2604,
-      "step": 47157
-    },
-    {
-      "epoch": 24.59348109517601,
-      "grad_norm": 1.5038899183273315,
-      "learning_rate": 5.312562814070352e-05,
-      "loss": 5.3881,
-      "step": 47158
-    },
-    {
-      "epoch": 24.59400260756193,
-      "grad_norm": 1.653290033340454,
-      "learning_rate": 5.312462311557789e-05,
-      "loss": 4.9258,
-      "step": 47159
-    },
-    {
-      "epoch": 24.59452411994785,
-      "grad_norm": 1.5257371664047241,
-      "learning_rate": 5.3123618090452267e-05,
-      "loss": 5.2646,
-      "step": 47160
-    },
-    {
-      "epoch": 24.59504563233377,
-      "grad_norm": 1.5981290340423584,
-      "learning_rate": 5.312261306532663e-05,
-      "loss": 5.4452,
-      "step": 47161
-    },
-    {
-      "epoch": 24.595567144719688,
-      "grad_norm": 1.5279515981674194,
-      "learning_rate": 5.312160804020101e-05,
-      "loss": 5.4959,
-      "step": 47162
-    },
-    {
-      "epoch": 24.596088657105607,
-      "grad_norm": 1.5072613954544067,
-      "learning_rate": 5.312060301507538e-05,
-      "loss": 5.453,
-      "step": 47163
-    },
-    {
-      "epoch": 24.596610169491527,
-      "grad_norm": 1.6152540445327759,
-      "learning_rate": 5.311959798994976e-05,
-      "loss": 5.4195,
-      "step": 47164
-    },
-    {
-      "epoch": 24.597131681877446,
-      "grad_norm": 1.6595503091812134,
-      "learning_rate": 5.311859296482412e-05,
-      "loss": 5.0994,
-      "step": 47165
-    },
-    {
-      "epoch": 24.597653194263362,
-      "grad_norm": 1.5254706144332886,
-      "learning_rate": 5.31175879396985e-05,
-      "loss": 5.2007,
-      "step": 47166
-    },
-    {
-      "epoch": 24.59817470664928,
-      "grad_norm": 1.5437471866607666,
-      "learning_rate": 5.3116582914572864e-05,
-      "loss": 5.2794,
-      "step": 47167
-    },
-    {
-      "epoch": 24.5986962190352,
-      "grad_norm": 1.5850486755371094,
-      "learning_rate": 5.311557788944724e-05,
-      "loss": 5.0089,
-      "step": 47168
-    },
-    {
-      "epoch": 24.59921773142112,
-      "grad_norm": 1.5807647705078125,
-      "learning_rate": 5.311457286432161e-05,
-      "loss": 5.0487,
-      "step": 47169
-    },
-    {
-      "epoch": 24.59973924380704,
-      "grad_norm": 1.6075592041015625,
-      "learning_rate": 5.311356783919599e-05,
-      "loss": 5.2585,
-      "step": 47170
-    },
-    {
-      "epoch": 24.60026075619296,
-      "grad_norm": 1.6043047904968262,
-      "learning_rate": 5.3112562814070355e-05,
-      "loss": 5.1537,
-      "step": 47171
-    },
-    {
-      "epoch": 24.60078226857888,
-      "grad_norm": 1.6553950309753418,
-      "learning_rate": 5.311155778894472e-05,
-      "loss": 5.244,
-      "step": 47172
-    },
-    {
-      "epoch": 24.6013037809648,
-      "grad_norm": 1.560715913772583,
-      "learning_rate": 5.31105527638191e-05,
-      "loss": 4.8832,
-      "step": 47173
-    },
-    {
-      "epoch": 24.601825293350718,
-      "grad_norm": 1.5508004426956177,
-      "learning_rate": 5.310954773869347e-05,
-      "loss": 5.155,
-      "step": 47174
-    },
-    {
-      "epoch": 24.602346805736637,
-      "grad_norm": 1.567783236503601,
-      "learning_rate": 5.3108542713567846e-05,
-      "loss": 5.4905,
-      "step": 47175
-    },
-    {
-      "epoch": 24.602868318122557,
-      "grad_norm": 1.4671134948730469,
-      "learning_rate": 5.310753768844221e-05,
-      "loss": 4.9243,
-      "step": 47176
-    },
-    {
-      "epoch": 24.603389830508476,
-      "grad_norm": 1.534091830253601,
-      "learning_rate": 5.310653266331659e-05,
-      "loss": 5.3797,
-      "step": 47177
-    },
-    {
-      "epoch": 24.603911342894392,
-      "grad_norm": 1.4277266263961792,
-      "learning_rate": 5.310552763819095e-05,
-      "loss": 5.4755,
-      "step": 47178
-    },
-    {
-      "epoch": 24.60443285528031,
-      "grad_norm": 1.4792784452438354,
-      "learning_rate": 5.310452261306533e-05,
-      "loss": 4.5365,
-      "step": 47179
-    },
-    {
-      "epoch": 24.60495436766623,
-      "grad_norm": 1.489902377128601,
-      "learning_rate": 5.31035175879397e-05,
-      "loss": 5.4915,
-      "step": 47180
-    },
-    {
-      "epoch": 24.60547588005215,
-      "grad_norm": 1.5528783798217773,
-      "learning_rate": 5.310251256281408e-05,
-      "loss": 5.3994,
-      "step": 47181
-    },
-    {
-      "epoch": 24.60599739243807,
-      "grad_norm": 1.6299142837524414,
-      "learning_rate": 5.310150753768844e-05,
-      "loss": 5.4266,
-      "step": 47182
-    },
-    {
-      "epoch": 24.60651890482399,
-      "grad_norm": 1.475597620010376,
-      "learning_rate": 5.310050251256282e-05,
-      "loss": 5.3714,
-      "step": 47183
-    },
-    {
-      "epoch": 24.60704041720991,
-      "grad_norm": 1.4581830501556396,
-      "learning_rate": 5.3099497487437185e-05,
-      "loss": 5.4902,
-      "step": 47184
-    },
-    {
-      "epoch": 24.60756192959583,
-      "grad_norm": 1.5175524950027466,
-      "learning_rate": 5.3098492462311556e-05,
-      "loss": 5.2732,
-      "step": 47185
-    },
-    {
-      "epoch": 24.608083441981748,
-      "grad_norm": 1.531785249710083,
-      "learning_rate": 5.3097487437185934e-05,
-      "loss": 5.3204,
-      "step": 47186
-    },
-    {
-      "epoch": 24.608604954367667,
-      "grad_norm": 1.693729281425476,
-      "learning_rate": 5.30964824120603e-05,
-      "loss": 4.7665,
-      "step": 47187
-    },
-    {
-      "epoch": 24.609126466753587,
-      "grad_norm": 1.577835202217102,
-      "learning_rate": 5.3095477386934676e-05,
-      "loss": 5.3026,
-      "step": 47188
-    },
-    {
-      "epoch": 24.609647979139506,
-      "grad_norm": 1.558287262916565,
-      "learning_rate": 5.309447236180905e-05,
-      "loss": 5.197,
-      "step": 47189
-    },
-    {
-      "epoch": 24.610169491525422,
-      "grad_norm": 1.5376355648040771,
-      "learning_rate": 5.3093467336683425e-05,
-      "loss": 5.0287,
-      "step": 47190
-    },
-    {
-      "epoch": 24.61069100391134,
-      "grad_norm": 1.4702706336975098,
-      "learning_rate": 5.309246231155779e-05,
-      "loss": 5.0335,
-      "step": 47191
-    },
-    {
-      "epoch": 24.61121251629726,
-      "grad_norm": 1.7605355978012085,
-      "learning_rate": 5.309145728643217e-05,
-      "loss": 5.0939,
-      "step": 47192
-    },
-    {
-      "epoch": 24.61173402868318,
-      "grad_norm": 1.7096079587936401,
-      "learning_rate": 5.309045226130653e-05,
-      "loss": 4.9587,
-      "step": 47193
-    },
-    {
-      "epoch": 24.6122555410691,
-      "grad_norm": 1.6876126527786255,
-      "learning_rate": 5.308944723618091e-05,
-      "loss": 5.319,
-      "step": 47194
-    },
-    {
-      "epoch": 24.61277705345502,
-      "grad_norm": 1.523714303970337,
-      "learning_rate": 5.308844221105528e-05,
-      "loss": 5.2504,
-      "step": 47195
-    },
-    {
-      "epoch": 24.61329856584094,
-      "grad_norm": 1.5173941850662231,
-      "learning_rate": 5.308743718592966e-05,
-      "loss": 5.0652,
-      "step": 47196
-    },
-    {
-      "epoch": 24.61382007822686,
-      "grad_norm": 1.5298278331756592,
-      "learning_rate": 5.308643216080402e-05,
-      "loss": 5.2781,
-      "step": 47197
-    },
-    {
-      "epoch": 24.614341590612778,
-      "grad_norm": 1.6060129404067993,
-      "learning_rate": 5.308542713567839e-05,
-      "loss": 4.8044,
-      "step": 47198
-    },
-    {
-      "epoch": 24.614863102998697,
-      "grad_norm": 1.4899954795837402,
-      "learning_rate": 5.3084422110552765e-05,
-      "loss": 4.6245,
-      "step": 47199
-    },
-    {
-      "epoch": 24.615384615384617,
-      "grad_norm": 1.5288375616073608,
-      "learning_rate": 5.3083417085427136e-05,
-      "loss": 5.355,
-      "step": 47200
-    },
-    {
-      "epoch": 24.615906127770536,
-      "grad_norm": 1.6281635761260986,
-      "learning_rate": 5.3082412060301514e-05,
-      "loss": 4.9669,
-      "step": 47201
-    },
-    {
-      "epoch": 24.616427640156452,
-      "grad_norm": 1.6428015232086182,
-      "learning_rate": 5.308140703517588e-05,
-      "loss": 5.2924,
-      "step": 47202
-    },
-    {
-      "epoch": 24.61694915254237,
-      "grad_norm": 1.6521916389465332,
-      "learning_rate": 5.3080402010050256e-05,
-      "loss": 5.2068,
-      "step": 47203
-    },
-    {
-      "epoch": 24.61747066492829,
-      "grad_norm": 1.627158761024475,
-      "learning_rate": 5.307939698492462e-05,
-      "loss": 5.3274,
-      "step": 47204
-    },
-    {
-      "epoch": 24.61799217731421,
-      "grad_norm": 1.4687442779541016,
-      "learning_rate": 5.3078391959799e-05,
-      "loss": 5.6754,
-      "step": 47205
-    },
-    {
-      "epoch": 24.61851368970013,
-      "grad_norm": 1.5469396114349365,
-      "learning_rate": 5.307738693467337e-05,
-      "loss": 5.406,
-      "step": 47206
-    },
-    {
-      "epoch": 24.61903520208605,
-      "grad_norm": 1.6229835748672485,
-      "learning_rate": 5.307638190954775e-05,
-      "loss": 5.2833,
-      "step": 47207
-    },
-    {
-      "epoch": 24.61955671447197,
-      "grad_norm": 1.4443403482437134,
-      "learning_rate": 5.307537688442211e-05,
-      "loss": 5.4767,
-      "step": 47208
-    },
-    {
-      "epoch": 24.62007822685789,
-      "grad_norm": 1.721652626991272,
-      "learning_rate": 5.307437185929649e-05,
-      "loss": 5.2897,
-      "step": 47209
-    },
-    {
-      "epoch": 24.620599739243808,
-      "grad_norm": 1.523107647895813,
-      "learning_rate": 5.307336683417086e-05,
-      "loss": 5.4533,
-      "step": 47210
-    },
-    {
-      "epoch": 24.621121251629727,
-      "grad_norm": 1.5523746013641357,
-      "learning_rate": 5.3072361809045224e-05,
-      "loss": 5.2791,
-      "step": 47211
-    },
-    {
-      "epoch": 24.621642764015647,
-      "grad_norm": 1.5223809480667114,
-      "learning_rate": 5.30713567839196e-05,
-      "loss": 5.2408,
-      "step": 47212
-    },
-    {
-      "epoch": 24.622164276401566,
-      "grad_norm": 1.5237787961959839,
-      "learning_rate": 5.3070351758793966e-05,
-      "loss": 5.4868,
-      "step": 47213
-    },
-    {
-      "epoch": 24.622685788787482,
-      "grad_norm": 1.4946768283843994,
-      "learning_rate": 5.3069346733668344e-05,
-      "loss": 5.4956,
-      "step": 47214
-    },
-    {
-      "epoch": 24.6232073011734,
-      "grad_norm": 1.7775529623031616,
-      "learning_rate": 5.3068341708542715e-05,
-      "loss": 5.4592,
-      "step": 47215
-    },
-    {
-      "epoch": 24.62372881355932,
-      "grad_norm": 1.7927982807159424,
-      "learning_rate": 5.306733668341709e-05,
-      "loss": 5.2087,
-      "step": 47216
-    },
-    {
-      "epoch": 24.62425032594524,
-      "grad_norm": 1.5073119401931763,
-      "learning_rate": 5.306633165829146e-05,
-      "loss": 5.3644,
-      "step": 47217
-    },
-    {
-      "epoch": 24.62477183833116,
-      "grad_norm": 1.648723840713501,
-      "learning_rate": 5.3065326633165835e-05,
-      "loss": 5.2285,
-      "step": 47218
-    },
-    {
-      "epoch": 24.62529335071708,
-      "grad_norm": 1.5113118886947632,
-      "learning_rate": 5.30643216080402e-05,
-      "loss": 5.626,
-      "step": 47219
-    },
-    {
-      "epoch": 24.625814863103,
-      "grad_norm": 1.5916692018508911,
-      "learning_rate": 5.306331658291458e-05,
-      "loss": 4.2101,
-      "step": 47220
-    },
-    {
-      "epoch": 24.62633637548892,
-      "grad_norm": 1.5326213836669922,
-      "learning_rate": 5.306231155778895e-05,
-      "loss": 4.7042,
-      "step": 47221
-    },
-    {
-      "epoch": 24.626857887874838,
-      "grad_norm": 1.599189043045044,
-      "learning_rate": 5.3061306532663326e-05,
-      "loss": 5.1577,
-      "step": 47222
-    },
-    {
-      "epoch": 24.627379400260757,
-      "grad_norm": 1.583525538444519,
-      "learning_rate": 5.306030150753769e-05,
-      "loss": 5.2441,
-      "step": 47223
-    },
-    {
-      "epoch": 24.627900912646677,
-      "grad_norm": 1.5252430438995361,
-      "learning_rate": 5.3059296482412055e-05,
-      "loss": 5.708,
-      "step": 47224
-    },
-    {
-      "epoch": 24.628422425032596,
-      "grad_norm": 1.4827157258987427,
-      "learning_rate": 5.305829145728643e-05,
-      "loss": 5.1982,
-      "step": 47225
-    },
-    {
-      "epoch": 24.628943937418512,
-      "grad_norm": 1.5864847898483276,
-      "learning_rate": 5.3057286432160803e-05,
-      "loss": 5.2547,
-      "step": 47226
-    },
-    {
-      "epoch": 24.62946544980443,
-      "grad_norm": 1.5038081407546997,
-      "learning_rate": 5.305628140703518e-05,
-      "loss": 4.8134,
-      "step": 47227
-    },
-    {
-      "epoch": 24.62998696219035,
-      "grad_norm": 1.6411970853805542,
-      "learning_rate": 5.3055276381909546e-05,
-      "loss": 4.5142,
-      "step": 47228
-    },
-    {
-      "epoch": 24.63050847457627,
-      "grad_norm": 1.5785592794418335,
-      "learning_rate": 5.305427135678392e-05,
-      "loss": 4.997,
-      "step": 47229
-    },
-    {
-      "epoch": 24.63102998696219,
-      "grad_norm": 1.6743130683898926,
-      "learning_rate": 5.3053266331658294e-05,
-      "loss": 5.2298,
-      "step": 47230
-    },
-    {
-      "epoch": 24.63155149934811,
-      "grad_norm": 1.6261507272720337,
-      "learning_rate": 5.305226130653267e-05,
-      "loss": 5.495,
-      "step": 47231
-    },
-    {
-      "epoch": 24.63207301173403,
-      "grad_norm": 1.5224404335021973,
-      "learning_rate": 5.3051256281407036e-05,
-      "loss": 4.9609,
-      "step": 47232
-    },
-    {
-      "epoch": 24.63259452411995,
-      "grad_norm": 1.4874696731567383,
-      "learning_rate": 5.3050251256281414e-05,
-      "loss": 5.5073,
-      "step": 47233
-    },
-    {
-      "epoch": 24.633116036505868,
-      "grad_norm": 1.6006276607513428,
-      "learning_rate": 5.304924623115578e-05,
-      "loss": 4.5828,
-      "step": 47234
-    },
-    {
-      "epoch": 24.633637548891787,
-      "grad_norm": 1.6693974733352661,
-      "learning_rate": 5.3048241206030156e-05,
-      "loss": 4.8457,
-      "step": 47235
-    },
-    {
-      "epoch": 24.634159061277707,
-      "grad_norm": 1.506693720817566,
-      "learning_rate": 5.304723618090453e-05,
-      "loss": 5.1673,
-      "step": 47236
-    },
-    {
-      "epoch": 24.634680573663623,
-      "grad_norm": 1.5398550033569336,
-      "learning_rate": 5.304623115577889e-05,
-      "loss": 4.24,
-      "step": 47237
-    },
-    {
-      "epoch": 24.635202086049542,
-      "grad_norm": 1.5314178466796875,
-      "learning_rate": 5.304522613065327e-05,
-      "loss": 5.4885,
-      "step": 47238
-    },
-    {
-      "epoch": 24.63572359843546,
-      "grad_norm": 1.5919532775878906,
-      "learning_rate": 5.3044221105527634e-05,
-      "loss": 4.9412,
-      "step": 47239
-    },
-    {
-      "epoch": 24.63624511082138,
-      "grad_norm": 1.5714800357818604,
-      "learning_rate": 5.304321608040201e-05,
-      "loss": 5.1706,
-      "step": 47240
-    },
-    {
-      "epoch": 24.6367666232073,
-      "grad_norm": 1.5271435976028442,
-      "learning_rate": 5.304221105527638e-05,
-      "loss": 5.5402,
-      "step": 47241
-    },
-    {
-      "epoch": 24.63728813559322,
-      "grad_norm": 1.5567569732666016,
-      "learning_rate": 5.304120603015076e-05,
-      "loss": 5.1546,
-      "step": 47242
-    },
-    {
-      "epoch": 24.63780964797914,
-      "grad_norm": 1.656408429145813,
-      "learning_rate": 5.3040201005025125e-05,
-      "loss": 5.3986,
-      "step": 47243
-    },
-    {
-      "epoch": 24.63833116036506,
-      "grad_norm": 1.6177750825881958,
-      "learning_rate": 5.30391959798995e-05,
-      "loss": 5.27,
-      "step": 47244
-    },
-    {
-      "epoch": 24.63885267275098,
-      "grad_norm": 1.5970677137374878,
-      "learning_rate": 5.303819095477387e-05,
-      "loss": 5.2944,
-      "step": 47245
-    },
-    {
-      "epoch": 24.639374185136898,
-      "grad_norm": 1.4506051540374756,
-      "learning_rate": 5.3037185929648245e-05,
-      "loss": 5.5297,
-      "step": 47246
-    },
-    {
-      "epoch": 24.639895697522817,
-      "grad_norm": 1.6511797904968262,
-      "learning_rate": 5.3036180904522616e-05,
-      "loss": 4.8308,
-      "step": 47247
-    },
-    {
-      "epoch": 24.640417209908737,
-      "grad_norm": 1.5718356370925903,
-      "learning_rate": 5.3035175879396994e-05,
-      "loss": 5.1075,
-      "step": 47248
-    },
-    {
-      "epoch": 24.640938722294656,
-      "grad_norm": 1.768456220626831,
-      "learning_rate": 5.303417085427136e-05,
-      "loss": 5.164,
-      "step": 47249
-    },
-    {
-      "epoch": 24.641460234680572,
-      "grad_norm": 1.6422250270843506,
-      "learning_rate": 5.3033165829145736e-05,
-      "loss": 5.2793,
-      "step": 47250
-    },
-    {
-      "epoch": 24.64198174706649,
-      "grad_norm": 1.6708883047103882,
-      "learning_rate": 5.303216080402011e-05,
-      "loss": 5.0487,
-      "step": 47251
-    },
-    {
-      "epoch": 24.64250325945241,
-      "grad_norm": 1.6016300916671753,
-      "learning_rate": 5.303115577889447e-05,
-      "loss": 5.2212,
-      "step": 47252
-    },
-    {
-      "epoch": 24.64302477183833,
-      "grad_norm": 1.5365461111068726,
-      "learning_rate": 5.303015075376885e-05,
-      "loss": 5.5075,
-      "step": 47253
-    },
-    {
-      "epoch": 24.64354628422425,
-      "grad_norm": 1.605668306350708,
-      "learning_rate": 5.302914572864321e-05,
-      "loss": 5.0429,
-      "step": 47254
-    },
-    {
-      "epoch": 24.64406779661017,
-      "grad_norm": 1.4853752851486206,
-      "learning_rate": 5.302814070351759e-05,
-      "loss": 5.3686,
-      "step": 47255
-    },
-    {
-      "epoch": 24.64458930899609,
-      "grad_norm": 1.6953967809677124,
-      "learning_rate": 5.302713567839196e-05,
-      "loss": 5.1381,
-      "step": 47256
-    },
-    {
-      "epoch": 24.64511082138201,
-      "grad_norm": 1.5825589895248413,
-      "learning_rate": 5.302613065326634e-05,
-      "loss": 5.2158,
-      "step": 47257
-    },
-    {
-      "epoch": 24.645632333767928,
-      "grad_norm": 1.7015743255615234,
-      "learning_rate": 5.3025125628140704e-05,
-      "loss": 4.8415,
-      "step": 47258
-    },
-    {
-      "epoch": 24.646153846153847,
-      "grad_norm": 1.5502547025680542,
-      "learning_rate": 5.302412060301508e-05,
-      "loss": 5.2807,
-      "step": 47259
-    },
-    {
-      "epoch": 24.646675358539767,
-      "grad_norm": 2.1508536338806152,
-      "learning_rate": 5.3023115577889446e-05,
-      "loss": 5.3073,
-      "step": 47260
-    },
-    {
-      "epoch": 24.647196870925683,
-      "grad_norm": 1.6809701919555664,
-      "learning_rate": 5.3022110552763824e-05,
-      "loss": 5.1485,
-      "step": 47261
-    },
-    {
-      "epoch": 24.647718383311602,
-      "grad_norm": 1.5775620937347412,
-      "learning_rate": 5.3021105527638195e-05,
-      "loss": 5.0698,
-      "step": 47262
-    },
-    {
-      "epoch": 24.64823989569752,
-      "grad_norm": 1.5737042427062988,
-      "learning_rate": 5.302010050251257e-05,
-      "loss": 5.5859,
-      "step": 47263
-    },
-    {
-      "epoch": 24.64876140808344,
-      "grad_norm": 1.551084041595459,
-      "learning_rate": 5.301909547738694e-05,
-      "loss": 5.1445,
-      "step": 47264
-    },
-    {
-      "epoch": 24.64928292046936,
-      "grad_norm": 1.7412413358688354,
-      "learning_rate": 5.30180904522613e-05,
-      "loss": 4.7423,
-      "step": 47265
-    },
-    {
-      "epoch": 24.64980443285528,
-      "grad_norm": 1.522901177406311,
-      "learning_rate": 5.301708542713568e-05,
-      "loss": 5.4326,
-      "step": 47266
-    },
-    {
-      "epoch": 24.6503259452412,
-      "grad_norm": 1.7019038200378418,
-      "learning_rate": 5.301608040201005e-05,
-      "loss": 5.111,
-      "step": 47267
-    },
-    {
-      "epoch": 24.65084745762712,
-      "grad_norm": 1.5494191646575928,
-      "learning_rate": 5.301507537688443e-05,
-      "loss": 5.064,
-      "step": 47268
-    },
-    {
-      "epoch": 24.65136897001304,
-      "grad_norm": 1.7355725765228271,
-      "learning_rate": 5.301407035175879e-05,
-      "loss": 5.16,
-      "step": 47269
-    },
-    {
-      "epoch": 24.651890482398958,
-      "grad_norm": 1.4748327732086182,
-      "learning_rate": 5.301306532663317e-05,
-      "loss": 5.4889,
-      "step": 47270
-    },
-    {
-      "epoch": 24.652411994784877,
-      "grad_norm": 1.5693140029907227,
-      "learning_rate": 5.3012060301507535e-05,
-      "loss": 4.6313,
-      "step": 47271
-    },
-    {
-      "epoch": 24.652933507170797,
-      "grad_norm": 1.4056451320648193,
-      "learning_rate": 5.301105527638191e-05,
-      "loss": 5.9091,
-      "step": 47272
-    },
-    {
-      "epoch": 24.653455019556713,
-      "grad_norm": 1.6304889917373657,
-      "learning_rate": 5.3010050251256283e-05,
-      "loss": 4.8851,
-      "step": 47273
-    },
-    {
-      "epoch": 24.653976531942632,
-      "grad_norm": 1.7449266910552979,
-      "learning_rate": 5.300904522613066e-05,
-      "loss": 4.3523,
-      "step": 47274
-    },
-    {
-      "epoch": 24.65449804432855,
-      "grad_norm": 1.5028966665267944,
-      "learning_rate": 5.3008040201005026e-05,
-      "loss": 5.5223,
-      "step": 47275
-    },
-    {
-      "epoch": 24.65501955671447,
-      "grad_norm": 1.5807825326919556,
-      "learning_rate": 5.3007035175879403e-05,
-      "loss": 4.7024,
-      "step": 47276
-    },
-    {
-      "epoch": 24.65554106910039,
-      "grad_norm": 1.5285805463790894,
-      "learning_rate": 5.3006030150753774e-05,
-      "loss": 5.6849,
-      "step": 47277
-    },
-    {
-      "epoch": 24.65606258148631,
-      "grad_norm": 1.6249452829360962,
-      "learning_rate": 5.300502512562814e-05,
-      "loss": 4.5854,
-      "step": 47278
-    },
-    {
-      "epoch": 24.65658409387223,
-      "grad_norm": 1.557190179824829,
-      "learning_rate": 5.3004020100502517e-05,
-      "loss": 4.777,
-      "step": 47279
-    },
-    {
-      "epoch": 24.65710560625815,
-      "grad_norm": 1.5841377973556519,
-      "learning_rate": 5.300301507537688e-05,
-      "loss": 5.116,
-      "step": 47280
-    },
-    {
-      "epoch": 24.65762711864407,
-      "grad_norm": 1.4651930332183838,
-      "learning_rate": 5.300201005025126e-05,
-      "loss": 5.4987,
-      "step": 47281
-    },
-    {
-      "epoch": 24.658148631029988,
-      "grad_norm": 1.5852582454681396,
-      "learning_rate": 5.300100502512563e-05,
-      "loss": 5.2137,
-      "step": 47282
-    },
-    {
-      "epoch": 24.658670143415907,
-      "grad_norm": 1.6034446954727173,
-      "learning_rate": 5.300000000000001e-05,
-      "loss": 5.2335,
-      "step": 47283
-    },
-    {
-      "epoch": 24.659191655801827,
-      "grad_norm": 1.6141154766082764,
-      "learning_rate": 5.299899497487437e-05,
-      "loss": 5.2183,
-      "step": 47284
-    },
-    {
-      "epoch": 24.659713168187743,
-      "grad_norm": 1.5855337381362915,
-      "learning_rate": 5.299798994974875e-05,
-      "loss": 5.192,
-      "step": 47285
-    },
-    {
-      "epoch": 24.660234680573662,
-      "grad_norm": 1.5944453477859497,
-      "learning_rate": 5.2996984924623114e-05,
-      "loss": 5.5681,
-      "step": 47286
-    },
-    {
-      "epoch": 24.66075619295958,
-      "grad_norm": 1.455283284187317,
-      "learning_rate": 5.299597989949749e-05,
-      "loss": 5.0151,
-      "step": 47287
-    },
-    {
-      "epoch": 24.6612777053455,
-      "grad_norm": 1.498062252998352,
-      "learning_rate": 5.299497487437186e-05,
-      "loss": 5.4247,
-      "step": 47288
-    },
-    {
-      "epoch": 24.66179921773142,
-      "grad_norm": 1.5763134956359863,
-      "learning_rate": 5.299396984924624e-05,
-      "loss": 5.3298,
-      "step": 47289
-    },
-    {
-      "epoch": 24.66232073011734,
-      "grad_norm": 1.475059151649475,
-      "learning_rate": 5.2992964824120605e-05,
-      "loss": 5.477,
-      "step": 47290
-    },
-    {
-      "epoch": 24.66284224250326,
-      "grad_norm": 1.4729453325271606,
-      "learning_rate": 5.299195979899497e-05,
-      "loss": 5.5025,
-      "step": 47291
-    },
-    {
-      "epoch": 24.66336375488918,
-      "grad_norm": 1.4951444864273071,
-      "learning_rate": 5.299095477386935e-05,
-      "loss": 5.3979,
-      "step": 47292
-    },
-    {
-      "epoch": 24.6638852672751,
-      "grad_norm": 1.543735146522522,
-      "learning_rate": 5.298994974874372e-05,
-      "loss": 5.8062,
-      "step": 47293
-    },
-    {
-      "epoch": 24.664406779661018,
-      "grad_norm": 1.5200599431991577,
-      "learning_rate": 5.2988944723618096e-05,
-      "loss": 5.5263,
-      "step": 47294
-    },
-    {
-      "epoch": 24.664928292046937,
-      "grad_norm": 1.6580661535263062,
-      "learning_rate": 5.298793969849246e-05,
-      "loss": 5.0338,
-      "step": 47295
-    },
-    {
-      "epoch": 24.665449804432857,
-      "grad_norm": 1.5120186805725098,
-      "learning_rate": 5.298693467336684e-05,
-      "loss": 5.3792,
-      "step": 47296
-    },
-    {
-      "epoch": 24.665971316818773,
-      "grad_norm": 1.4971370697021484,
-      "learning_rate": 5.298592964824121e-05,
-      "loss": 5.6788,
-      "step": 47297
-    },
-    {
-      "epoch": 24.666492829204692,
-      "grad_norm": 1.5689584016799927,
-      "learning_rate": 5.298492462311559e-05,
-      "loss": 4.8034,
-      "step": 47298
-    },
-    {
-      "epoch": 24.667014341590612,
-      "grad_norm": 1.4943147897720337,
-      "learning_rate": 5.298391959798995e-05,
-      "loss": 5.5322,
-      "step": 47299
-    },
-    {
-      "epoch": 24.66753585397653,
-      "grad_norm": 1.5954899787902832,
-      "learning_rate": 5.298291457286433e-05,
-      "loss": 5.2418,
-      "step": 47300
-    },
-    {
-      "epoch": 24.66805736636245,
-      "grad_norm": 1.4898053407669067,
-      "learning_rate": 5.298190954773869e-05,
-      "loss": 5.6953,
-      "step": 47301
-    },
-    {
-      "epoch": 24.66857887874837,
-      "grad_norm": 1.5356955528259277,
-      "learning_rate": 5.298090452261307e-05,
-      "loss": 5.3747,
-      "step": 47302
-    },
-    {
-      "epoch": 24.66910039113429,
-      "grad_norm": 1.5907671451568604,
-      "learning_rate": 5.297989949748744e-05,
-      "loss": 5.3825,
-      "step": 47303
-    },
-    {
-      "epoch": 24.66962190352021,
-      "grad_norm": 1.6802881956100464,
-      "learning_rate": 5.2978894472361806e-05,
-      "loss": 5.0529,
-      "step": 47304
-    },
-    {
-      "epoch": 24.67014341590613,
-      "grad_norm": 1.4547600746154785,
-      "learning_rate": 5.2977889447236184e-05,
-      "loss": 5.3847,
-      "step": 47305
-    },
-    {
-      "epoch": 24.670664928292048,
-      "grad_norm": 1.559444785118103,
-      "learning_rate": 5.297688442211055e-05,
-      "loss": 5.1646,
-      "step": 47306
-    },
-    {
-      "epoch": 24.671186440677968,
-      "grad_norm": 1.5914020538330078,
-      "learning_rate": 5.2975879396984926e-05,
-      "loss": 5.335,
-      "step": 47307
-    },
-    {
-      "epoch": 24.671707953063887,
-      "grad_norm": 1.5467071533203125,
-      "learning_rate": 5.29748743718593e-05,
-      "loss": 5.14,
-      "step": 47308
-    },
-    {
-      "epoch": 24.672229465449803,
-      "grad_norm": 1.5258315801620483,
-      "learning_rate": 5.2973869346733675e-05,
-      "loss": 5.2126,
-      "step": 47309
-    },
-    {
-      "epoch": 24.672750977835722,
-      "grad_norm": 1.467491865158081,
-      "learning_rate": 5.297286432160804e-05,
-      "loss": 5.2486,
-      "step": 47310
-    },
-    {
-      "epoch": 24.673272490221642,
-      "grad_norm": 1.5919041633605957,
-      "learning_rate": 5.297185929648242e-05,
-      "loss": 5.4162,
-      "step": 47311
-    },
-    {
-      "epoch": 24.67379400260756,
-      "grad_norm": 1.4914984703063965,
-      "learning_rate": 5.297085427135678e-05,
-      "loss": 5.1323,
-      "step": 47312
-    },
-    {
-      "epoch": 24.67431551499348,
-      "grad_norm": 1.4902338981628418,
-      "learning_rate": 5.296984924623116e-05,
-      "loss": 5.2009,
-      "step": 47313
-    },
-    {
-      "epoch": 24.6748370273794,
-      "grad_norm": 1.5726829767227173,
-      "learning_rate": 5.296884422110553e-05,
-      "loss": 5.6046,
-      "step": 47314
-    },
-    {
-      "epoch": 24.67535853976532,
-      "grad_norm": 1.523964524269104,
-      "learning_rate": 5.296783919597991e-05,
-      "loss": 5.4268,
-      "step": 47315
-    },
-    {
-      "epoch": 24.67588005215124,
-      "grad_norm": 1.6206296682357788,
-      "learning_rate": 5.296683417085427e-05,
-      "loss": 4.8415,
-      "step": 47316
-    },
-    {
-      "epoch": 24.67640156453716,
-      "grad_norm": 1.713107705116272,
-      "learning_rate": 5.2965829145728644e-05,
-      "loss": 4.5193,
-      "step": 47317
-    },
-    {
-      "epoch": 24.676923076923078,
-      "grad_norm": 1.5280061960220337,
-      "learning_rate": 5.296482412060302e-05,
-      "loss": 5.6748,
-      "step": 47318
-    },
-    {
-      "epoch": 24.677444589308998,
-      "grad_norm": 1.6032605171203613,
-      "learning_rate": 5.2963819095477386e-05,
-      "loss": 4.3322,
-      "step": 47319
-    },
-    {
-      "epoch": 24.677966101694913,
-      "grad_norm": 1.5340440273284912,
-      "learning_rate": 5.2962814070351764e-05,
-      "loss": 5.2519,
-      "step": 47320
-    },
-    {
-      "epoch": 24.678487614080833,
-      "grad_norm": 1.4873908758163452,
-      "learning_rate": 5.296180904522613e-05,
-      "loss": 5.535,
-      "step": 47321
-    },
-    {
-      "epoch": 24.679009126466752,
-      "grad_norm": 1.5279886722564697,
-      "learning_rate": 5.2960804020100506e-05,
-      "loss": 5.643,
-      "step": 47322
-    },
-    {
-      "epoch": 24.679530638852672,
-      "grad_norm": 1.5305421352386475,
-      "learning_rate": 5.295979899497488e-05,
-      "loss": 5.2924,
-      "step": 47323
-    },
-    {
-      "epoch": 24.68005215123859,
-      "grad_norm": 1.594450831413269,
-      "learning_rate": 5.2958793969849255e-05,
-      "loss": 5.1636,
-      "step": 47324
-    },
-    {
-      "epoch": 24.68057366362451,
-      "grad_norm": 1.5831859111785889,
-      "learning_rate": 5.295778894472362e-05,
-      "loss": 4.9946,
-      "step": 47325
-    },
-    {
-      "epoch": 24.68109517601043,
-      "grad_norm": 1.548529863357544,
-      "learning_rate": 5.2956783919598e-05,
-      "loss": 5.3552,
-      "step": 47326
-    },
-    {
-      "epoch": 24.68161668839635,
-      "grad_norm": 1.7135651111602783,
-      "learning_rate": 5.295577889447236e-05,
-      "loss": 5.1989,
-      "step": 47327
-    },
-    {
-      "epoch": 24.68213820078227,
-      "grad_norm": 1.6396974325180054,
-      "learning_rate": 5.295477386934674e-05,
-      "loss": 5.0598,
-      "step": 47328
-    },
-    {
-      "epoch": 24.68265971316819,
-      "grad_norm": 1.5008044242858887,
-      "learning_rate": 5.295376884422111e-05,
-      "loss": 4.7358,
-      "step": 47329
-    },
-    {
-      "epoch": 24.683181225554108,
-      "grad_norm": 1.467037558555603,
-      "learning_rate": 5.2952763819095474e-05,
-      "loss": 5.6756,
-      "step": 47330
-    },
-    {
-      "epoch": 24.683702737940028,
-      "grad_norm": 1.5959055423736572,
-      "learning_rate": 5.295175879396985e-05,
-      "loss": 4.8113,
-      "step": 47331
-    },
-    {
-      "epoch": 24.684224250325947,
-      "grad_norm": 1.6172267198562622,
-      "learning_rate": 5.2950753768844216e-05,
-      "loss": 5.3932,
-      "step": 47332
-    },
-    {
-      "epoch": 24.684745762711863,
-      "grad_norm": 1.6276932954788208,
-      "learning_rate": 5.2949748743718594e-05,
-      "loss": 4.9819,
-      "step": 47333
-    },
-    {
-      "epoch": 24.685267275097782,
-      "grad_norm": 1.6367923021316528,
-      "learning_rate": 5.2948743718592965e-05,
-      "loss": 5.1476,
-      "step": 47334
-    },
-    {
-      "epoch": 24.685788787483702,
-      "grad_norm": 1.6635050773620605,
-      "learning_rate": 5.294773869346734e-05,
-      "loss": 4.9992,
-      "step": 47335
-    },
-    {
-      "epoch": 24.68631029986962,
-      "grad_norm": 1.5734236240386963,
-      "learning_rate": 5.294673366834171e-05,
-      "loss": 5.5752,
-      "step": 47336
-    },
-    {
-      "epoch": 24.68683181225554,
-      "grad_norm": 1.6350886821746826,
-      "learning_rate": 5.2945728643216085e-05,
-      "loss": 5.3038,
-      "step": 47337
-    },
-    {
-      "epoch": 24.68735332464146,
-      "grad_norm": 1.6359999179840088,
-      "learning_rate": 5.2944723618090456e-05,
-      "loss": 5.2646,
-      "step": 47338
-    },
-    {
-      "epoch": 24.68787483702738,
-      "grad_norm": 1.5551235675811768,
-      "learning_rate": 5.2943718592964834e-05,
-      "loss": 5.1918,
-      "step": 47339
-    },
-    {
-      "epoch": 24.6883963494133,
-      "grad_norm": 1.5145343542099,
-      "learning_rate": 5.29427135678392e-05,
-      "loss": 5.2244,
-      "step": 47340
-    },
-    {
-      "epoch": 24.68891786179922,
-      "grad_norm": 1.5511668920516968,
-      "learning_rate": 5.2941708542713576e-05,
-      "loss": 5.1972,
-      "step": 47341
-    },
-    {
-      "epoch": 24.689439374185138,
-      "grad_norm": 1.638988971710205,
-      "learning_rate": 5.294070351758794e-05,
-      "loss": 5.0287,
-      "step": 47342
-    },
-    {
-      "epoch": 24.689960886571058,
-      "grad_norm": 1.612149953842163,
-      "learning_rate": 5.293969849246232e-05,
-      "loss": 5.2798,
-      "step": 47343
-    },
-    {
-      "epoch": 24.690482398956973,
-      "grad_norm": 1.4347255229949951,
-      "learning_rate": 5.293869346733669e-05,
-      "loss": 5.2653,
-      "step": 47344
-    },
-    {
-      "epoch": 24.691003911342893,
-      "grad_norm": 1.5032137632369995,
-      "learning_rate": 5.2937688442211053e-05,
-      "loss": 5.1911,
-      "step": 47345
-    },
-    {
-      "epoch": 24.691525423728812,
-      "grad_norm": 1.573765516281128,
-      "learning_rate": 5.293668341708543e-05,
-      "loss": 5.5309,
-      "step": 47346
-    },
-    {
-      "epoch": 24.692046936114732,
-      "grad_norm": 1.4932886362075806,
-      "learning_rate": 5.2935678391959796e-05,
-      "loss": 4.6991,
-      "step": 47347
-    },
-    {
-      "epoch": 24.69256844850065,
-      "grad_norm": 1.6029568910598755,
-      "learning_rate": 5.293467336683417e-05,
-      "loss": 5.0425,
-      "step": 47348
-    },
-    {
-      "epoch": 24.69308996088657,
-      "grad_norm": 1.5358147621154785,
-      "learning_rate": 5.2933668341708544e-05,
-      "loss": 4.9356,
-      "step": 47349
-    },
-    {
-      "epoch": 24.69361147327249,
-      "grad_norm": 1.6393566131591797,
-      "learning_rate": 5.293266331658292e-05,
-      "loss": 5.4375,
-      "step": 47350
-    },
-    {
-      "epoch": 24.69413298565841,
-      "grad_norm": 1.4459666013717651,
-      "learning_rate": 5.2931658291457287e-05,
-      "loss": 5.2593,
-      "step": 47351
-    },
-    {
-      "epoch": 24.69465449804433,
-      "grad_norm": 1.4607338905334473,
-      "learning_rate": 5.2930653266331664e-05,
-      "loss": 5.2711,
-      "step": 47352
-    },
-    {
-      "epoch": 24.69517601043025,
-      "grad_norm": 1.5448263883590698,
-      "learning_rate": 5.292964824120603e-05,
-      "loss": 5.3287,
-      "step": 47353
-    },
-    {
-      "epoch": 24.695697522816168,
-      "grad_norm": 1.6037408113479614,
-      "learning_rate": 5.2928643216080406e-05,
-      "loss": 5.1141,
-      "step": 47354
-    },
-    {
-      "epoch": 24.696219035202088,
-      "grad_norm": 1.7563648223876953,
-      "learning_rate": 5.292763819095478e-05,
-      "loss": 5.1192,
-      "step": 47355
-    },
-    {
-      "epoch": 24.696740547588004,
-      "grad_norm": 1.6039814949035645,
-      "learning_rate": 5.2926633165829155e-05,
-      "loss": 5.2887,
-      "step": 47356
-    },
-    {
-      "epoch": 24.697262059973923,
-      "grad_norm": 1.928030014038086,
-      "learning_rate": 5.292562814070352e-05,
-      "loss": 4.884,
-      "step": 47357
-    },
-    {
-      "epoch": 24.697783572359842,
-      "grad_norm": 1.5670113563537598,
-      "learning_rate": 5.2924623115577884e-05,
-      "loss": 5.5185,
-      "step": 47358
-    },
-    {
-      "epoch": 24.698305084745762,
-      "grad_norm": 1.605115532875061,
-      "learning_rate": 5.292361809045226e-05,
-      "loss": 4.869,
-      "step": 47359
-    },
-    {
-      "epoch": 24.69882659713168,
-      "grad_norm": 1.5162445306777954,
-      "learning_rate": 5.292261306532663e-05,
-      "loss": 5.5156,
-      "step": 47360
-    },
-    {
-      "epoch": 24.6993481095176,
-      "grad_norm": 1.5317503213882446,
-      "learning_rate": 5.292160804020101e-05,
-      "loss": 5.4146,
-      "step": 47361
-    },
-    {
-      "epoch": 24.69986962190352,
-      "grad_norm": 1.465466022491455,
-      "learning_rate": 5.2920603015075375e-05,
-      "loss": 5.2047,
-      "step": 47362
-    },
-    {
-      "epoch": 24.70039113428944,
-      "grad_norm": 1.6061033010482788,
-      "learning_rate": 5.291959798994975e-05,
-      "loss": 5.0021,
-      "step": 47363
-    },
-    {
-      "epoch": 24.70091264667536,
-      "grad_norm": 1.519797682762146,
-      "learning_rate": 5.2918592964824124e-05,
-      "loss": 5.1277,
-      "step": 47364
-    },
-    {
-      "epoch": 24.70143415906128,
-      "grad_norm": 1.4747980833053589,
-      "learning_rate": 5.29175879396985e-05,
-      "loss": 5.4958,
-      "step": 47365
-    },
-    {
-      "epoch": 24.701955671447198,
-      "grad_norm": 1.5383646488189697,
-      "learning_rate": 5.2916582914572866e-05,
-      "loss": 5.0771,
-      "step": 47366
-    },
-    {
-      "epoch": 24.702477183833118,
-      "grad_norm": 1.6111302375793457,
-      "learning_rate": 5.2915577889447244e-05,
-      "loss": 5.3315,
-      "step": 47367
-    },
-    {
-      "epoch": 24.702998696219034,
-      "grad_norm": 1.63905930519104,
-      "learning_rate": 5.291457286432161e-05,
-      "loss": 5.225,
-      "step": 47368
-    },
-    {
-      "epoch": 24.703520208604953,
-      "grad_norm": 1.5803592205047607,
-      "learning_rate": 5.2913567839195986e-05,
-      "loss": 5.2988,
-      "step": 47369
-    },
-    {
-      "epoch": 24.704041720990872,
-      "grad_norm": 1.485611915588379,
-      "learning_rate": 5.291256281407036e-05,
-      "loss": 5.7067,
-      "step": 47370
-    },
-    {
-      "epoch": 24.704563233376792,
-      "grad_norm": 1.5183179378509521,
-      "learning_rate": 5.291155778894472e-05,
-      "loss": 4.9675,
-      "step": 47371
-    },
-    {
-      "epoch": 24.70508474576271,
-      "grad_norm": 1.4715124368667603,
-      "learning_rate": 5.29105527638191e-05,
-      "loss": 5.4153,
-      "step": 47372
-    },
-    {
-      "epoch": 24.70560625814863,
-      "grad_norm": 1.5149403810501099,
-      "learning_rate": 5.290954773869346e-05,
-      "loss": 5.6242,
-      "step": 47373
-    },
-    {
-      "epoch": 24.70612777053455,
-      "grad_norm": 1.5118502378463745,
-      "learning_rate": 5.290854271356784e-05,
-      "loss": 5.2936,
-      "step": 47374
-    },
-    {
-      "epoch": 24.70664928292047,
-      "grad_norm": 1.4715086221694946,
-      "learning_rate": 5.290753768844221e-05,
-      "loss": 5.7607,
-      "step": 47375
-    },
-    {
-      "epoch": 24.70717079530639,
-      "grad_norm": 1.602860927581787,
-      "learning_rate": 5.290653266331659e-05,
-      "loss": 5.1513,
-      "step": 47376
-    },
-    {
-      "epoch": 24.70769230769231,
-      "grad_norm": 1.5714776515960693,
-      "learning_rate": 5.2905527638190954e-05,
-      "loss": 4.7246,
-      "step": 47377
-    },
-    {
-      "epoch": 24.708213820078228,
-      "grad_norm": 1.4496737718582153,
-      "learning_rate": 5.290452261306533e-05,
-      "loss": 5.2328,
-      "step": 47378
-    },
-    {
-      "epoch": 24.708735332464148,
-      "grad_norm": 1.4641375541687012,
-      "learning_rate": 5.2903517587939696e-05,
-      "loss": 5.5792,
-      "step": 47379
-    },
-    {
-      "epoch": 24.709256844850064,
-      "grad_norm": 1.5336039066314697,
-      "learning_rate": 5.2902512562814074e-05,
-      "loss": 5.0133,
-      "step": 47380
-    },
-    {
-      "epoch": 24.709778357235983,
-      "grad_norm": 1.5651122331619263,
-      "learning_rate": 5.2901507537688445e-05,
-      "loss": 5.0655,
-      "step": 47381
-    },
-    {
-      "epoch": 24.710299869621903,
-      "grad_norm": 1.4556572437286377,
-      "learning_rate": 5.290050251256282e-05,
-      "loss": 5.4127,
-      "step": 47382
-    },
-    {
-      "epoch": 24.710821382007822,
-      "grad_norm": 1.6332452297210693,
-      "learning_rate": 5.289949748743719e-05,
-      "loss": 5.1906,
-      "step": 47383
-    },
-    {
-      "epoch": 24.71134289439374,
-      "grad_norm": 1.5959571599960327,
-      "learning_rate": 5.289849246231156e-05,
-      "loss": 4.6574,
-      "step": 47384
-    },
-    {
-      "epoch": 24.71186440677966,
-      "grad_norm": 1.610870361328125,
-      "learning_rate": 5.2897487437185936e-05,
-      "loss": 5.4142,
-      "step": 47385
-    },
-    {
-      "epoch": 24.71238591916558,
-      "grad_norm": 1.5060276985168457,
-      "learning_rate": 5.28964824120603e-05,
-      "loss": 4.9912,
-      "step": 47386
-    },
-    {
-      "epoch": 24.7129074315515,
-      "grad_norm": 1.5834510326385498,
-      "learning_rate": 5.289547738693468e-05,
-      "loss": 5.4094,
-      "step": 47387
-    },
-    {
-      "epoch": 24.71342894393742,
-      "grad_norm": 1.5250613689422607,
-      "learning_rate": 5.289447236180904e-05,
-      "loss": 5.1351,
-      "step": 47388
-    },
-    {
-      "epoch": 24.71395045632334,
-      "grad_norm": 1.5975128412246704,
-      "learning_rate": 5.289346733668342e-05,
-      "loss": 5.2784,
-      "step": 47389
-    },
-    {
-      "epoch": 24.714471968709258,
-      "grad_norm": 1.6757081747055054,
-      "learning_rate": 5.289246231155779e-05,
-      "loss": 4.8205,
-      "step": 47390
-    },
-    {
-      "epoch": 24.714993481095178,
-      "grad_norm": 1.6876505613327026,
-      "learning_rate": 5.289145728643217e-05,
-      "loss": 5.0773,
-      "step": 47391
-    },
-    {
-      "epoch": 24.715514993481094,
-      "grad_norm": 1.541818618774414,
-      "learning_rate": 5.2890452261306534e-05,
-      "loss": 5.0922,
-      "step": 47392
-    },
-    {
-      "epoch": 24.716036505867013,
-      "grad_norm": 1.5823266506195068,
-      "learning_rate": 5.288944723618091e-05,
-      "loss": 5.3117,
-      "step": 47393
-    },
-    {
-      "epoch": 24.716558018252933,
-      "grad_norm": 1.6786820888519287,
-      "learning_rate": 5.2888442211055276e-05,
-      "loss": 5.1552,
-      "step": 47394
-    },
-    {
-      "epoch": 24.717079530638852,
-      "grad_norm": 1.5583157539367676,
-      "learning_rate": 5.2887437185929653e-05,
-      "loss": 5.3845,
-      "step": 47395
-    },
-    {
-      "epoch": 24.71760104302477,
-      "grad_norm": 1.6496175527572632,
-      "learning_rate": 5.2886432160804024e-05,
-      "loss": 4.6893,
-      "step": 47396
-    },
-    {
-      "epoch": 24.71812255541069,
-      "grad_norm": 1.4989069700241089,
-      "learning_rate": 5.288542713567839e-05,
-      "loss": 5.4424,
-      "step": 47397
-    },
-    {
-      "epoch": 24.71864406779661,
-      "grad_norm": 1.5996272563934326,
-      "learning_rate": 5.2884422110552767e-05,
-      "loss": 5.203,
-      "step": 47398
-    },
-    {
-      "epoch": 24.71916558018253,
-      "grad_norm": 1.6613653898239136,
-      "learning_rate": 5.288341708542713e-05,
-      "loss": 5.5652,
-      "step": 47399
-    },
-    {
-      "epoch": 24.71968709256845,
-      "grad_norm": 1.5515891313552856,
-      "learning_rate": 5.288241206030151e-05,
-      "loss": 5.3419,
-      "step": 47400
-    },
-    {
-      "epoch": 24.72020860495437,
-      "grad_norm": 1.5039900541305542,
-      "learning_rate": 5.288140703517588e-05,
-      "loss": 5.0322,
-      "step": 47401
-    },
-    {
-      "epoch": 24.72073011734029,
-      "grad_norm": 1.5610084533691406,
-      "learning_rate": 5.288040201005026e-05,
-      "loss": 5.0695,
-      "step": 47402
-    },
-    {
-      "epoch": 24.721251629726208,
-      "grad_norm": 1.6099997758865356,
-      "learning_rate": 5.287939698492462e-05,
-      "loss": 5.2438,
-      "step": 47403
-    },
-    {
-      "epoch": 24.721773142112124,
-      "grad_norm": 1.5810344219207764,
-      "learning_rate": 5.2878391959799e-05,
-      "loss": 5.3501,
-      "step": 47404
-    },
-    {
-      "epoch": 24.722294654498043,
-      "grad_norm": 1.600861668586731,
-      "learning_rate": 5.287738693467337e-05,
-      "loss": 5.0323,
-      "step": 47405
-    },
-    {
-      "epoch": 24.722816166883963,
-      "grad_norm": 1.4808549880981445,
-      "learning_rate": 5.287638190954775e-05,
-      "loss": 4.6499,
-      "step": 47406
-    },
-    {
-      "epoch": 24.723337679269882,
-      "grad_norm": 1.7229527235031128,
-      "learning_rate": 5.287537688442211e-05,
-      "loss": 5.0125,
-      "step": 47407
-    },
-    {
-      "epoch": 24.7238591916558,
-      "grad_norm": 1.6062817573547363,
-      "learning_rate": 5.287437185929649e-05,
-      "loss": 5.3669,
-      "step": 47408
-    },
-    {
-      "epoch": 24.72438070404172,
-      "grad_norm": 1.6355072259902954,
-      "learning_rate": 5.2873366834170855e-05,
-      "loss": 5.568,
-      "step": 47409
-    },
-    {
-      "epoch": 24.72490221642764,
-      "grad_norm": 1.5378187894821167,
-      "learning_rate": 5.2872361809045226e-05,
-      "loss": 4.9895,
-      "step": 47410
-    },
-    {
-      "epoch": 24.72542372881356,
-      "grad_norm": 1.5877232551574707,
-      "learning_rate": 5.2871356783919604e-05,
-      "loss": 4.8618,
-      "step": 47411
-    },
-    {
-      "epoch": 24.72594524119948,
-      "grad_norm": 1.6423448324203491,
-      "learning_rate": 5.287035175879397e-05,
-      "loss": 4.8321,
-      "step": 47412
-    },
-    {
-      "epoch": 24.7264667535854,
-      "grad_norm": 1.5903444290161133,
-      "learning_rate": 5.2869346733668346e-05,
-      "loss": 5.0025,
-      "step": 47413
-    },
-    {
-      "epoch": 24.72698826597132,
-      "grad_norm": 1.5387601852416992,
-      "learning_rate": 5.286834170854271e-05,
-      "loss": 5.5535,
-      "step": 47414
-    },
-    {
-      "epoch": 24.727509778357238,
-      "grad_norm": 1.5133532285690308,
-      "learning_rate": 5.286733668341709e-05,
-      "loss": 5.4693,
-      "step": 47415
-    },
-    {
-      "epoch": 24.728031290743154,
-      "grad_norm": 1.6344048976898193,
-      "learning_rate": 5.286633165829146e-05,
-      "loss": 4.4188,
-      "step": 47416
-    },
-    {
-      "epoch": 24.728552803129073,
-      "grad_norm": 1.5636427402496338,
-      "learning_rate": 5.286532663316584e-05,
-      "loss": 3.7235,
-      "step": 47417
-    },
-    {
-      "epoch": 24.729074315514993,
-      "grad_norm": 1.5532876253128052,
-      "learning_rate": 5.28643216080402e-05,
-      "loss": 5.5252,
-      "step": 47418
-    },
-    {
-      "epoch": 24.729595827900912,
-      "grad_norm": 1.4795598983764648,
-      "learning_rate": 5.286331658291458e-05,
-      "loss": 5.1144,
-      "step": 47419
-    },
-    {
-      "epoch": 24.73011734028683,
-      "grad_norm": 1.466006875038147,
-      "learning_rate": 5.286231155778894e-05,
-      "loss": 5.118,
-      "step": 47420
-    },
-    {
-      "epoch": 24.73063885267275,
-      "grad_norm": 1.5018320083618164,
-      "learning_rate": 5.286130653266332e-05,
-      "loss": 5.4393,
-      "step": 47421
-    },
-    {
-      "epoch": 24.73116036505867,
-      "grad_norm": 1.6700994968414307,
-      "learning_rate": 5.286030150753769e-05,
-      "loss": 4.8299,
-      "step": 47422
-    },
-    {
-      "epoch": 24.73168187744459,
-      "grad_norm": 1.5098687410354614,
-      "learning_rate": 5.2859296482412056e-05,
-      "loss": 5.2341,
-      "step": 47423
-    },
-    {
-      "epoch": 24.73220338983051,
-      "grad_norm": 1.6486495733261108,
-      "learning_rate": 5.2858291457286434e-05,
-      "loss": 4.9364,
-      "step": 47424
-    },
-    {
-      "epoch": 24.73272490221643,
-      "grad_norm": 1.5243858098983765,
-      "learning_rate": 5.28572864321608e-05,
-      "loss": 5.3591,
-      "step": 47425
-    },
-    {
-      "epoch": 24.73324641460235,
-      "grad_norm": 1.586220145225525,
-      "learning_rate": 5.2856281407035176e-05,
-      "loss": 5.3367,
-      "step": 47426
-    },
-    {
-      "epoch": 24.733767926988264,
-      "grad_norm": 1.4399850368499756,
-      "learning_rate": 5.285527638190955e-05,
-      "loss": 5.6983,
-      "step": 47427
-    },
-    {
-      "epoch": 24.734289439374184,
-      "grad_norm": 1.555681586265564,
-      "learning_rate": 5.2854271356783925e-05,
-      "loss": 5.2923,
-      "step": 47428
-    },
-    {
-      "epoch": 24.734810951760103,
-      "grad_norm": 1.5340088605880737,
-      "learning_rate": 5.285326633165829e-05,
-      "loss": 5.5594,
-      "step": 47429
-    },
-    {
-      "epoch": 24.735332464146023,
-      "grad_norm": 1.5374293327331543,
-      "learning_rate": 5.285226130653267e-05,
-      "loss": 5.1602,
-      "step": 47430
-    },
-    {
-      "epoch": 24.735853976531942,
-      "grad_norm": 1.5442519187927246,
-      "learning_rate": 5.285125628140704e-05,
-      "loss": 5.2001,
-      "step": 47431
-    },
-    {
-      "epoch": 24.73637548891786,
-      "grad_norm": 1.5805193185806274,
-      "learning_rate": 5.2850251256281416e-05,
-      "loss": 5.2318,
-      "step": 47432
-    },
-    {
-      "epoch": 24.73689700130378,
-      "grad_norm": 1.5123268365859985,
-      "learning_rate": 5.284924623115578e-05,
-      "loss": 5.5972,
-      "step": 47433
-    },
-    {
-      "epoch": 24.7374185136897,
-      "grad_norm": 1.5266926288604736,
-      "learning_rate": 5.284824120603016e-05,
-      "loss": 5.4148,
-      "step": 47434
-    },
-    {
-      "epoch": 24.73794002607562,
-      "grad_norm": 1.5206873416900635,
-      "learning_rate": 5.284723618090452e-05,
-      "loss": 4.8806,
-      "step": 47435
-    },
-    {
-      "epoch": 24.73846153846154,
-      "grad_norm": 1.669305443763733,
-      "learning_rate": 5.28462311557789e-05,
-      "loss": 5.1472,
-      "step": 47436
-    },
-    {
-      "epoch": 24.73898305084746,
-      "grad_norm": 1.6166794300079346,
-      "learning_rate": 5.284522613065327e-05,
-      "loss": 5.4533,
-      "step": 47437
-    },
-    {
-      "epoch": 24.73950456323338,
-      "grad_norm": 1.558632493019104,
-      "learning_rate": 5.2844221105527636e-05,
-      "loss": 5.2843,
-      "step": 47438
-    },
-    {
-      "epoch": 24.740026075619298,
-      "grad_norm": 1.566023826599121,
-      "learning_rate": 5.2843216080402014e-05,
-      "loss": 5.0458,
-      "step": 47439
-    },
-    {
-      "epoch": 24.740547588005214,
-      "grad_norm": 1.5679587125778198,
-      "learning_rate": 5.284221105527638e-05,
-      "loss": 5.2118,
-      "step": 47440
-    },
-    {
-      "epoch": 24.741069100391133,
-      "grad_norm": 1.5457262992858887,
-      "learning_rate": 5.2841206030150756e-05,
-      "loss": 5.5065,
-      "step": 47441
-    },
-    {
-      "epoch": 24.741590612777053,
-      "grad_norm": 1.6512668132781982,
-      "learning_rate": 5.284020100502513e-05,
-      "loss": 4.8574,
-      "step": 47442
-    },
-    {
-      "epoch": 24.742112125162972,
-      "grad_norm": 1.5250366926193237,
-      "learning_rate": 5.2839195979899505e-05,
-      "loss": 5.2422,
-      "step": 47443
-    },
-    {
-      "epoch": 24.74263363754889,
-      "grad_norm": 1.5479857921600342,
-      "learning_rate": 5.283819095477387e-05,
-      "loss": 5.7012,
-      "step": 47444
-    },
-    {
-      "epoch": 24.74315514993481,
-      "grad_norm": 1.6642109155654907,
-      "learning_rate": 5.283718592964825e-05,
-      "loss": 4.9788,
-      "step": 47445
-    },
-    {
-      "epoch": 24.74367666232073,
-      "grad_norm": 1.4878603219985962,
-      "learning_rate": 5.283618090452261e-05,
-      "loss": 5.5579,
-      "step": 47446
-    },
-    {
-      "epoch": 24.74419817470665,
-      "grad_norm": 1.5626896619796753,
-      "learning_rate": 5.283517587939699e-05,
-      "loss": 5.2341,
-      "step": 47447
-    },
-    {
-      "epoch": 24.74471968709257,
-      "grad_norm": 1.5615829229354858,
-      "learning_rate": 5.283417085427136e-05,
-      "loss": 5.2907,
-      "step": 47448
-    },
-    {
-      "epoch": 24.74524119947849,
-      "grad_norm": 1.681559443473816,
-      "learning_rate": 5.283316582914574e-05,
-      "loss": 4.7711,
-      "step": 47449
-    },
-    {
-      "epoch": 24.74576271186441,
-      "grad_norm": 1.6382852792739868,
-      "learning_rate": 5.28321608040201e-05,
-      "loss": 5.1967,
-      "step": 47450
-    },
-    {
-      "epoch": 24.746284224250324,
-      "grad_norm": 1.4390983581542969,
-      "learning_rate": 5.283115577889447e-05,
-      "loss": 5.6046,
-      "step": 47451
-    },
-    {
-      "epoch": 24.746805736636244,
-      "grad_norm": 1.5192053318023682,
-      "learning_rate": 5.283015075376885e-05,
-      "loss": 5.3251,
-      "step": 47452
-    },
-    {
-      "epoch": 24.747327249022163,
-      "grad_norm": 1.6994962692260742,
-      "learning_rate": 5.2829145728643215e-05,
-      "loss": 5.1813,
-      "step": 47453
-    },
-    {
-      "epoch": 24.747848761408083,
-      "grad_norm": 1.5553594827651978,
-      "learning_rate": 5.282814070351759e-05,
-      "loss": 5.435,
-      "step": 47454
-    },
-    {
-      "epoch": 24.748370273794002,
-      "grad_norm": 1.5389695167541504,
-      "learning_rate": 5.282713567839196e-05,
-      "loss": 4.9239,
-      "step": 47455
-    },
-    {
-      "epoch": 24.74889178617992,
-      "grad_norm": 1.481982946395874,
-      "learning_rate": 5.2826130653266335e-05,
-      "loss": 5.416,
-      "step": 47456
-    },
-    {
-      "epoch": 24.74941329856584,
-      "grad_norm": 1.5305616855621338,
-      "learning_rate": 5.2825125628140706e-05,
-      "loss": 5.438,
-      "step": 47457
-    },
-    {
-      "epoch": 24.74993481095176,
-      "grad_norm": 1.6072165966033936,
-      "learning_rate": 5.2824120603015084e-05,
-      "loss": 5.1031,
-      "step": 47458
-    },
-    {
-      "epoch": 24.75045632333768,
-      "grad_norm": 1.5590033531188965,
-      "learning_rate": 5.282311557788945e-05,
-      "loss": 5.2212,
-      "step": 47459
-    },
-    {
-      "epoch": 24.7509778357236,
-      "grad_norm": 1.6058114767074585,
-      "learning_rate": 5.2822110552763826e-05,
-      "loss": 5.3505,
-      "step": 47460
-    },
-    {
-      "epoch": 24.75149934810952,
-      "grad_norm": 1.4668052196502686,
-      "learning_rate": 5.282110552763819e-05,
-      "loss": 5.2932,
-      "step": 47461
-    },
-    {
-      "epoch": 24.75202086049544,
-      "grad_norm": 1.5334019660949707,
-      "learning_rate": 5.282010050251257e-05,
-      "loss": 4.9809,
-      "step": 47462
-    },
-    {
-      "epoch": 24.752542372881354,
-      "grad_norm": 1.5113006830215454,
-      "learning_rate": 5.281909547738694e-05,
-      "loss": 5.6508,
-      "step": 47463
-    },
-    {
-      "epoch": 24.753063885267274,
-      "grad_norm": 1.5373015403747559,
-      "learning_rate": 5.2818090452261303e-05,
-      "loss": 5.5241,
-      "step": 47464
-    },
-    {
-      "epoch": 24.753585397653193,
-      "grad_norm": 1.5907377004623413,
-      "learning_rate": 5.281708542713568e-05,
-      "loss": 5.1051,
-      "step": 47465
-    },
-    {
-      "epoch": 24.754106910039113,
-      "grad_norm": 1.5339654684066772,
-      "learning_rate": 5.2816080402010046e-05,
-      "loss": 5.4189,
-      "step": 47466
-    },
-    {
-      "epoch": 24.754628422425032,
-      "grad_norm": 1.5293375253677368,
-      "learning_rate": 5.281507537688442e-05,
-      "loss": 5.2817,
-      "step": 47467
-    },
-    {
-      "epoch": 24.75514993481095,
-      "grad_norm": 1.4435278177261353,
-      "learning_rate": 5.2814070351758794e-05,
-      "loss": 5.5082,
-      "step": 47468
-    },
-    {
-      "epoch": 24.75567144719687,
-      "grad_norm": 1.6221555471420288,
-      "learning_rate": 5.281306532663317e-05,
-      "loss": 5.0163,
-      "step": 47469
-    },
-    {
-      "epoch": 24.75619295958279,
-      "grad_norm": 1.630165457725525,
-      "learning_rate": 5.2812060301507537e-05,
-      "loss": 4.9543,
-      "step": 47470
-    },
-    {
-      "epoch": 24.75671447196871,
-      "grad_norm": 1.4313089847564697,
-      "learning_rate": 5.2811055276381914e-05,
-      "loss": 4.9606,
-      "step": 47471
-    },
-    {
-      "epoch": 24.75723598435463,
-      "grad_norm": 1.469399333000183,
-      "learning_rate": 5.2810050251256285e-05,
-      "loss": 4.9652,
-      "step": 47472
-    },
-    {
-      "epoch": 24.75775749674055,
-      "grad_norm": 1.615159511566162,
-      "learning_rate": 5.280904522613066e-05,
-      "loss": 5.1887,
-      "step": 47473
-    },
-    {
-      "epoch": 24.75827900912647,
-      "grad_norm": 1.5276683568954468,
-      "learning_rate": 5.280804020100503e-05,
-      "loss": 5.4537,
-      "step": 47474
-    },
-    {
-      "epoch": 24.758800521512384,
-      "grad_norm": 1.577805519104004,
-      "learning_rate": 5.2807035175879405e-05,
-      "loss": 5.3674,
-      "step": 47475
-    },
-    {
-      "epoch": 24.759322033898304,
-      "grad_norm": 1.6209670305252075,
-      "learning_rate": 5.280603015075377e-05,
-      "loss": 5.3706,
-      "step": 47476
-    },
-    {
-      "epoch": 24.759843546284223,
-      "grad_norm": 1.5125975608825684,
-      "learning_rate": 5.280502512562814e-05,
-      "loss": 5.5455,
-      "step": 47477
-    },
-    {
-      "epoch": 24.760365058670143,
-      "grad_norm": 1.537079095840454,
-      "learning_rate": 5.280402010050252e-05,
-      "loss": 5.1048,
-      "step": 47478
-    },
-    {
-      "epoch": 24.760886571056062,
-      "grad_norm": 1.6048425436019897,
-      "learning_rate": 5.280301507537688e-05,
-      "loss": 5.1865,
-      "step": 47479
-    },
-    {
-      "epoch": 24.76140808344198,
-      "grad_norm": 1.6852270364761353,
-      "learning_rate": 5.280201005025126e-05,
-      "loss": 4.6009,
-      "step": 47480
-    },
-    {
-      "epoch": 24.7619295958279,
-      "grad_norm": 1.6097460985183716,
-      "learning_rate": 5.2801005025125625e-05,
-      "loss": 5.4573,
-      "step": 47481
-    },
-    {
-      "epoch": 24.76245110821382,
-      "grad_norm": 1.5865166187286377,
-      "learning_rate": 5.28e-05,
-      "loss": 5.4706,
-      "step": 47482
-    },
-    {
-      "epoch": 24.76297262059974,
-      "grad_norm": 1.6382147073745728,
-      "learning_rate": 5.2798994974874374e-05,
-      "loss": 4.2004,
-      "step": 47483
-    },
-    {
-      "epoch": 24.76349413298566,
-      "grad_norm": 1.54300057888031,
-      "learning_rate": 5.279798994974875e-05,
-      "loss": 4.8358,
-      "step": 47484
-    },
-    {
-      "epoch": 24.76401564537158,
-      "grad_norm": 1.6016281843185425,
-      "learning_rate": 5.2796984924623116e-05,
-      "loss": 5.0162,
-      "step": 47485
-    },
-    {
-      "epoch": 24.7645371577575,
-      "grad_norm": 1.6335874795913696,
-      "learning_rate": 5.2795979899497494e-05,
-      "loss": 5.2313,
-      "step": 47486
-    },
-    {
-      "epoch": 24.765058670143414,
-      "grad_norm": 1.5735015869140625,
-      "learning_rate": 5.279497487437186e-05,
-      "loss": 5.3067,
-      "step": 47487
-    },
-    {
-      "epoch": 24.765580182529334,
-      "grad_norm": 1.5651882886886597,
-      "learning_rate": 5.2793969849246236e-05,
-      "loss": 5.0935,
-      "step": 47488
-    },
-    {
-      "epoch": 24.766101694915253,
-      "grad_norm": 1.5960276126861572,
-      "learning_rate": 5.279296482412061e-05,
-      "loss": 5.1305,
-      "step": 47489
-    },
-    {
-      "epoch": 24.766623207301173,
-      "grad_norm": 1.575652003288269,
-      "learning_rate": 5.279195979899497e-05,
-      "loss": 5.1764,
-      "step": 47490
-    },
-    {
-      "epoch": 24.767144719687092,
-      "grad_norm": 1.5699725151062012,
-      "learning_rate": 5.279095477386935e-05,
-      "loss": 5.5052,
-      "step": 47491
-    },
-    {
-      "epoch": 24.76766623207301,
-      "grad_norm": 1.4981679916381836,
-      "learning_rate": 5.278994974874372e-05,
-      "loss": 4.8892,
-      "step": 47492
-    },
-    {
-      "epoch": 24.76818774445893,
-      "grad_norm": 1.5606601238250732,
-      "learning_rate": 5.27889447236181e-05,
-      "loss": 4.9481,
-      "step": 47493
-    },
-    {
-      "epoch": 24.76870925684485,
-      "grad_norm": 1.6086260080337524,
-      "learning_rate": 5.278793969849246e-05,
-      "loss": 4.9859,
-      "step": 47494
-    },
-    {
-      "epoch": 24.76923076923077,
-      "grad_norm": 1.5214101076126099,
-      "learning_rate": 5.278693467336684e-05,
-      "loss": 5.1028,
-      "step": 47495
-    },
-    {
-      "epoch": 24.76975228161669,
-      "grad_norm": 1.5095444917678833,
-      "learning_rate": 5.2785929648241204e-05,
-      "loss": 5.1418,
-      "step": 47496
-    },
-    {
-      "epoch": 24.77027379400261,
-      "grad_norm": 1.5300822257995605,
-      "learning_rate": 5.278492462311558e-05,
-      "loss": 5.2729,
-      "step": 47497
-    },
-    {
-      "epoch": 24.77079530638853,
-      "grad_norm": 1.5844762325286865,
-      "learning_rate": 5.278391959798995e-05,
-      "loss": 4.9532,
-      "step": 47498
-    },
-    {
-      "epoch": 24.771316818774444,
-      "grad_norm": 1.529147982597351,
-      "learning_rate": 5.278291457286433e-05,
-      "loss": 5.1491,
-      "step": 47499
-    },
-    {
-      "epoch": 24.771838331160364,
-      "grad_norm": 1.6927844285964966,
-      "learning_rate": 5.2781909547738695e-05,
-      "loss": 5.3014,
-      "step": 47500
-    },
-    {
-      "epoch": 24.772359843546283,
-      "grad_norm": 1.595672369003296,
-      "learning_rate": 5.278090452261307e-05,
-      "loss": 5.5442,
-      "step": 47501
-    },
-    {
-      "epoch": 24.772881355932203,
-      "grad_norm": 1.5376272201538086,
-      "learning_rate": 5.277989949748744e-05,
-      "loss": 5.1765,
-      "step": 47502
-    },
-    {
-      "epoch": 24.773402868318122,
-      "grad_norm": 1.589247703552246,
-      "learning_rate": 5.277889447236181e-05,
-      "loss": 5.2062,
-      "step": 47503
-    },
-    {
-      "epoch": 24.77392438070404,
-      "grad_norm": 1.5780093669891357,
-      "learning_rate": 5.2777889447236186e-05,
-      "loss": 5.51,
-      "step": 47504
-    },
-    {
-      "epoch": 24.77444589308996,
-      "grad_norm": 1.6503119468688965,
-      "learning_rate": 5.277688442211055e-05,
-      "loss": 5.1063,
-      "step": 47505
-    },
-    {
-      "epoch": 24.77496740547588,
-      "grad_norm": 1.4889123439788818,
-      "learning_rate": 5.277587939698493e-05,
-      "loss": 5.1427,
-      "step": 47506
-    },
-    {
-      "epoch": 24.7754889178618,
-      "grad_norm": 1.5205193758010864,
-      "learning_rate": 5.277487437185929e-05,
-      "loss": 5.0561,
-      "step": 47507
-    },
-    {
-      "epoch": 24.77601043024772,
-      "grad_norm": 1.6262480020523071,
-      "learning_rate": 5.277386934673367e-05,
-      "loss": 5.0302,
-      "step": 47508
-    },
-    {
-      "epoch": 24.77653194263364,
-      "grad_norm": 1.7596312761306763,
-      "learning_rate": 5.277286432160804e-05,
-      "loss": 5.0836,
-      "step": 47509
-    },
-    {
-      "epoch": 24.777053455019555,
-      "grad_norm": 1.4933420419692993,
-      "learning_rate": 5.277185929648242e-05,
-      "loss": 5.1002,
-      "step": 47510
-    },
-    {
-      "epoch": 24.777574967405474,
-      "grad_norm": 1.4821689128875732,
-      "learning_rate": 5.2770854271356784e-05,
-      "loss": 5.5749,
-      "step": 47511
-    },
-    {
-      "epoch": 24.778096479791394,
-      "grad_norm": 1.4744006395339966,
-      "learning_rate": 5.276984924623116e-05,
-      "loss": 5.5574,
-      "step": 47512
-    },
-    {
-      "epoch": 24.778617992177313,
-      "grad_norm": 1.543318271636963,
-      "learning_rate": 5.2768844221105526e-05,
-      "loss": 5.1245,
-      "step": 47513
-    },
-    {
-      "epoch": 24.779139504563233,
-      "grad_norm": 1.4699589014053345,
-      "learning_rate": 5.2767839195979903e-05,
-      "loss": 5.3484,
-      "step": 47514
-    },
-    {
-      "epoch": 24.779661016949152,
-      "grad_norm": 1.5591673851013184,
-      "learning_rate": 5.2766834170854275e-05,
-      "loss": 5.2485,
-      "step": 47515
-    },
-    {
-      "epoch": 24.78018252933507,
-      "grad_norm": 1.528746247291565,
-      "learning_rate": 5.276582914572864e-05,
-      "loss": 5.1657,
-      "step": 47516
-    },
-    {
-      "epoch": 24.78070404172099,
-      "grad_norm": 1.5867305994033813,
-      "learning_rate": 5.2764824120603017e-05,
-      "loss": 5.3466,
-      "step": 47517
-    },
-    {
-      "epoch": 24.78122555410691,
-      "grad_norm": 1.4916069507598877,
-      "learning_rate": 5.276381909547739e-05,
-      "loss": 5.0848,
-      "step": 47518
-    },
-    {
-      "epoch": 24.78174706649283,
-      "grad_norm": 1.4861212968826294,
-      "learning_rate": 5.2762814070351765e-05,
-      "loss": 5.4499,
-      "step": 47519
-    },
-    {
-      "epoch": 24.78226857887875,
-      "grad_norm": 1.6560192108154297,
-      "learning_rate": 5.276180904522613e-05,
-      "loss": 5.2978,
-      "step": 47520
-    },
-    {
-      "epoch": 24.78279009126467,
-      "grad_norm": 1.5494730472564697,
-      "learning_rate": 5.276080402010051e-05,
-      "loss": 5.4394,
-      "step": 47521
-    },
-    {
-      "epoch": 24.78331160365059,
-      "grad_norm": 1.5607541799545288,
-      "learning_rate": 5.275979899497487e-05,
-      "loss": 5.0194,
-      "step": 47522
-    },
-    {
-      "epoch": 24.783833116036504,
-      "grad_norm": 1.470194935798645,
-      "learning_rate": 5.275879396984925e-05,
-      "loss": 5.3755,
-      "step": 47523
-    },
-    {
-      "epoch": 24.784354628422424,
-      "grad_norm": 1.6068567037582397,
-      "learning_rate": 5.275778894472362e-05,
-      "loss": 5.2849,
-      "step": 47524
-    },
-    {
-      "epoch": 24.784876140808343,
-      "grad_norm": 1.6193372011184692,
-      "learning_rate": 5.2756783919598e-05,
-      "loss": 5.2435,
-      "step": 47525
-    },
-    {
-      "epoch": 24.785397653194263,
-      "grad_norm": 1.548972487449646,
-      "learning_rate": 5.275577889447236e-05,
-      "loss": 5.2963,
-      "step": 47526
-    },
-    {
-      "epoch": 24.785919165580182,
-      "grad_norm": 1.4897874593734741,
-      "learning_rate": 5.275477386934674e-05,
-      "loss": 5.4222,
-      "step": 47527
-    },
-    {
-      "epoch": 24.7864406779661,
-      "grad_norm": 1.5287671089172363,
-      "learning_rate": 5.2753768844221105e-05,
-      "loss": 5.038,
-      "step": 47528
-    },
-    {
-      "epoch": 24.78696219035202,
-      "grad_norm": 1.6856635808944702,
-      "learning_rate": 5.275276381909548e-05,
-      "loss": 5.2468,
-      "step": 47529
-    },
-    {
-      "epoch": 24.78748370273794,
-      "grad_norm": 1.577754020690918,
-      "learning_rate": 5.2751758793969854e-05,
-      "loss": 5.3736,
-      "step": 47530
-    },
-    {
-      "epoch": 24.78800521512386,
-      "grad_norm": 1.5250321626663208,
-      "learning_rate": 5.275075376884422e-05,
-      "loss": 5.3523,
-      "step": 47531
-    },
-    {
-      "epoch": 24.78852672750978,
-      "grad_norm": 1.534523367881775,
-      "learning_rate": 5.2749748743718596e-05,
-      "loss": 4.9717,
-      "step": 47532
-    },
-    {
-      "epoch": 24.7890482398957,
-      "grad_norm": 1.5770007371902466,
-      "learning_rate": 5.274874371859296e-05,
-      "loss": 5.2463,
-      "step": 47533
-    },
-    {
-      "epoch": 24.789569752281615,
-      "grad_norm": 1.489088535308838,
-      "learning_rate": 5.274773869346734e-05,
-      "loss": 5.6112,
-      "step": 47534
-    },
-    {
-      "epoch": 24.790091264667534,
-      "grad_norm": 1.5553216934204102,
-      "learning_rate": 5.274673366834171e-05,
-      "loss": 4.7451,
-      "step": 47535
-    },
-    {
-      "epoch": 24.790612777053454,
-      "grad_norm": 1.5538774728775024,
-      "learning_rate": 5.274572864321609e-05,
-      "loss": 5.2696,
-      "step": 47536
-    },
-    {
-      "epoch": 24.791134289439373,
-      "grad_norm": 1.5648025274276733,
-      "learning_rate": 5.274472361809045e-05,
-      "loss": 5.2637,
-      "step": 47537
-    },
-    {
-      "epoch": 24.791655801825293,
-      "grad_norm": 1.5613423585891724,
-      "learning_rate": 5.274371859296483e-05,
-      "loss": 5.1972,
-      "step": 47538
-    },
-    {
-      "epoch": 24.792177314211212,
-      "grad_norm": 1.4599372148513794,
-      "learning_rate": 5.27427135678392e-05,
-      "loss": 5.6729,
-      "step": 47539
-    },
-    {
-      "epoch": 24.79269882659713,
-      "grad_norm": 1.601422667503357,
-      "learning_rate": 5.274170854271358e-05,
-      "loss": 4.898,
-      "step": 47540
-    },
-    {
-      "epoch": 24.79322033898305,
-      "grad_norm": 1.4956389665603638,
-      "learning_rate": 5.274070351758794e-05,
-      "loss": 5.2339,
-      "step": 47541
-    },
-    {
-      "epoch": 24.79374185136897,
-      "grad_norm": 1.7296587228775024,
-      "learning_rate": 5.273969849246232e-05,
-      "loss": 4.6253,
-      "step": 47542
-    },
-    {
-      "epoch": 24.79426336375489,
-      "grad_norm": 1.6556662321090698,
-      "learning_rate": 5.2738693467336684e-05,
-      "loss": 5.2285,
-      "step": 47543
-    },
-    {
-      "epoch": 24.79478487614081,
-      "grad_norm": 1.6297032833099365,
-      "learning_rate": 5.2737688442211055e-05,
-      "loss": 5.0284,
-      "step": 47544
-    },
-    {
-      "epoch": 24.79530638852673,
-      "grad_norm": 1.6926029920578003,
-      "learning_rate": 5.273668341708543e-05,
-      "loss": 4.4239,
-      "step": 47545
-    },
-    {
-      "epoch": 24.795827900912645,
-      "grad_norm": 1.6048219203948975,
-      "learning_rate": 5.27356783919598e-05,
-      "loss": 5.3311,
-      "step": 47546
-    },
-    {
-      "epoch": 24.796349413298564,
-      "grad_norm": 1.5725462436676025,
-      "learning_rate": 5.2734673366834175e-05,
-      "loss": 5.465,
-      "step": 47547
-    },
-    {
-      "epoch": 24.796870925684484,
-      "grad_norm": 1.666869878768921,
-      "learning_rate": 5.273366834170854e-05,
-      "loss": 5.4491,
-      "step": 47548
-    },
-    {
-      "epoch": 24.797392438070403,
-      "grad_norm": 1.6140059232711792,
-      "learning_rate": 5.273266331658292e-05,
-      "loss": 5.3708,
-      "step": 47549
-    },
-    {
-      "epoch": 24.797913950456323,
-      "grad_norm": 1.5744291543960571,
-      "learning_rate": 5.273165829145729e-05,
-      "loss": 5.4412,
-      "step": 47550
-    },
-    {
-      "epoch": 24.798435462842242,
-      "grad_norm": 1.435795783996582,
-      "learning_rate": 5.2730653266331666e-05,
-      "loss": 5.4963,
-      "step": 47551
-    },
-    {
-      "epoch": 24.798956975228162,
-      "grad_norm": 1.4892674684524536,
-      "learning_rate": 5.272964824120603e-05,
-      "loss": 5.3157,
-      "step": 47552
-    },
-    {
-      "epoch": 24.79947848761408,
-      "grad_norm": 1.5376384258270264,
-      "learning_rate": 5.272864321608041e-05,
-      "loss": 4.2826,
-      "step": 47553
-    },
-    {
-      "epoch": 24.8,
-      "grad_norm": 1.6269644498825073,
-      "learning_rate": 5.272763819095477e-05,
-      "loss": 5.0349,
-      "step": 47554
-    },
-    {
-      "epoch": 24.80052151238592,
-      "grad_norm": 1.512412428855896,
-      "learning_rate": 5.272663316582915e-05,
-      "loss": 5.5586,
-      "step": 47555
-    },
-    {
-      "epoch": 24.80104302477184,
-      "grad_norm": 1.5930720567703247,
-      "learning_rate": 5.272562814070352e-05,
-      "loss": 5.0954,
-      "step": 47556
-    },
-    {
-      "epoch": 24.80156453715776,
-      "grad_norm": 1.6046547889709473,
-      "learning_rate": 5.2724623115577886e-05,
-      "loss": 4.8148,
-      "step": 47557
-    },
-    {
-      "epoch": 24.802086049543675,
-      "grad_norm": 1.5422947406768799,
-      "learning_rate": 5.2723618090452264e-05,
-      "loss": 5.4638,
-      "step": 47558
-    },
-    {
-      "epoch": 24.802607561929594,
-      "grad_norm": 1.632067084312439,
-      "learning_rate": 5.2722613065326635e-05,
-      "loss": 4.7168,
-      "step": 47559
-    },
-    {
-      "epoch": 24.803129074315514,
-      "grad_norm": 1.5230306386947632,
-      "learning_rate": 5.272160804020101e-05,
-      "loss": 5.4069,
-      "step": 47560
-    },
-    {
-      "epoch": 24.803650586701433,
-      "grad_norm": 1.592128038406372,
-      "learning_rate": 5.272060301507538e-05,
-      "loss": 5.4452,
-      "step": 47561
-    },
-    {
-      "epoch": 24.804172099087353,
-      "grad_norm": 1.5460057258605957,
-      "learning_rate": 5.2719597989949755e-05,
-      "loss": 5.4909,
-      "step": 47562
-    },
-    {
-      "epoch": 24.804693611473272,
-      "grad_norm": 1.5914688110351562,
-      "learning_rate": 5.271859296482412e-05,
-      "loss": 5.4366,
-      "step": 47563
-    },
-    {
-      "epoch": 24.805215123859192,
-      "grad_norm": 1.5336439609527588,
-      "learning_rate": 5.27175879396985e-05,
-      "loss": 5.3584,
-      "step": 47564
-    },
-    {
-      "epoch": 24.80573663624511,
-      "grad_norm": 1.4945257902145386,
-      "learning_rate": 5.271658291457287e-05,
-      "loss": 5.483,
-      "step": 47565
-    },
-    {
-      "epoch": 24.80625814863103,
-      "grad_norm": 1.542188048362732,
-      "learning_rate": 5.2715577889447246e-05,
-      "loss": 5.4905,
-      "step": 47566
-    },
-    {
-      "epoch": 24.80677966101695,
-      "grad_norm": 1.49358332157135,
-      "learning_rate": 5.271457286432161e-05,
-      "loss": 5.387,
-      "step": 47567
-    },
-    {
-      "epoch": 24.80730117340287,
-      "grad_norm": 1.5079996585845947,
-      "learning_rate": 5.271356783919599e-05,
-      "loss": 5.5255,
-      "step": 47568
-    },
-    {
-      "epoch": 24.80782268578879,
-      "grad_norm": 1.6331822872161865,
-      "learning_rate": 5.271256281407035e-05,
-      "loss": 5.0497,
-      "step": 47569
-    },
-    {
-      "epoch": 24.808344198174705,
-      "grad_norm": 1.538796305656433,
-      "learning_rate": 5.271155778894472e-05,
-      "loss": 5.3896,
-      "step": 47570
-    },
-    {
-      "epoch": 24.808865710560625,
-      "grad_norm": 1.538104772567749,
-      "learning_rate": 5.27105527638191e-05,
-      "loss": 5.333,
-      "step": 47571
-    },
-    {
-      "epoch": 24.809387222946544,
-      "grad_norm": 1.5917664766311646,
-      "learning_rate": 5.2709547738693465e-05,
-      "loss": 5.1779,
-      "step": 47572
-    },
-    {
-      "epoch": 24.809908735332463,
-      "grad_norm": 1.5296070575714111,
-      "learning_rate": 5.270854271356784e-05,
-      "loss": 5.2235,
-      "step": 47573
-    },
-    {
-      "epoch": 24.810430247718383,
-      "grad_norm": 1.6117219924926758,
-      "learning_rate": 5.270753768844221e-05,
-      "loss": 5.5542,
-      "step": 47574
-    },
-    {
-      "epoch": 24.810951760104302,
-      "grad_norm": 1.553123950958252,
-      "learning_rate": 5.2706532663316585e-05,
-      "loss": 5.2958,
-      "step": 47575
-    },
-    {
-      "epoch": 24.811473272490222,
-      "grad_norm": 1.5403311252593994,
-      "learning_rate": 5.2705527638190956e-05,
-      "loss": 5.4147,
-      "step": 47576
-    },
-    {
-      "epoch": 24.81199478487614,
-      "grad_norm": 1.5323834419250488,
-      "learning_rate": 5.2704522613065334e-05,
-      "loss": 5.2087,
-      "step": 47577
-    },
-    {
-      "epoch": 24.81251629726206,
-      "grad_norm": 1.5729739665985107,
-      "learning_rate": 5.27035175879397e-05,
-      "loss": 5.4178,
-      "step": 47578
-    },
-    {
-      "epoch": 24.81303780964798,
-      "grad_norm": 1.5241584777832031,
-      "learning_rate": 5.2702512562814076e-05,
-      "loss": 5.4662,
-      "step": 47579
-    },
-    {
-      "epoch": 24.8135593220339,
-      "grad_norm": 1.6713162660598755,
-      "learning_rate": 5.270150753768845e-05,
-      "loss": 4.9369,
-      "step": 47580
-    },
-    {
-      "epoch": 24.81408083441982,
-      "grad_norm": 1.700795292854309,
-      "learning_rate": 5.2700502512562825e-05,
-      "loss": 4.717,
-      "step": 47581
-    },
-    {
-      "epoch": 24.814602346805735,
-      "grad_norm": 1.6582183837890625,
-      "learning_rate": 5.269949748743719e-05,
-      "loss": 5.3538,
-      "step": 47582
-    },
-    {
-      "epoch": 24.815123859191655,
-      "grad_norm": 1.5852854251861572,
-      "learning_rate": 5.2698492462311553e-05,
-      "loss": 5.37,
-      "step": 47583
-    },
-    {
-      "epoch": 24.815645371577574,
-      "grad_norm": 1.545393705368042,
-      "learning_rate": 5.269748743718593e-05,
-      "loss": 5.7418,
-      "step": 47584
-    },
-    {
-      "epoch": 24.816166883963493,
-      "grad_norm": 1.5864192247390747,
-      "learning_rate": 5.26964824120603e-05,
-      "loss": 5.1175,
-      "step": 47585
-    },
-    {
-      "epoch": 24.816688396349413,
-      "grad_norm": 1.4018276929855347,
-      "learning_rate": 5.269547738693468e-05,
-      "loss": 5.059,
-      "step": 47586
-    },
-    {
-      "epoch": 24.817209908735332,
-      "grad_norm": 1.6075929403305054,
-      "learning_rate": 5.2694472361809044e-05,
-      "loss": 5.4799,
-      "step": 47587
-    },
-    {
-      "epoch": 24.817731421121252,
-      "grad_norm": 1.8427879810333252,
-      "learning_rate": 5.269346733668342e-05,
-      "loss": 4.8626,
-      "step": 47588
-    },
-    {
-      "epoch": 24.81825293350717,
-      "grad_norm": 1.5658609867095947,
-      "learning_rate": 5.2692462311557787e-05,
-      "loss": 5.422,
-      "step": 47589
-    },
-    {
-      "epoch": 24.81877444589309,
-      "grad_norm": 1.5325922966003418,
-      "learning_rate": 5.2691457286432164e-05,
-      "loss": 4.907,
-      "step": 47590
-    },
-    {
-      "epoch": 24.81929595827901,
-      "grad_norm": 1.5844037532806396,
-      "learning_rate": 5.2690452261306535e-05,
-      "loss": 5.2854,
-      "step": 47591
-    },
-    {
-      "epoch": 24.81981747066493,
-      "grad_norm": 1.5087043046951294,
-      "learning_rate": 5.268944723618091e-05,
-      "loss": 4.4148,
-      "step": 47592
-    },
-    {
-      "epoch": 24.820338983050846,
-      "grad_norm": 1.6013150215148926,
-      "learning_rate": 5.268844221105528e-05,
-      "loss": 4.9726,
-      "step": 47593
-    },
-    {
-      "epoch": 24.820860495436765,
-      "grad_norm": 1.706227421760559,
-      "learning_rate": 5.2687437185929655e-05,
-      "loss": 4.8196,
-      "step": 47594
-    },
-    {
-      "epoch": 24.821382007822685,
-      "grad_norm": 1.5933588743209839,
-      "learning_rate": 5.268643216080402e-05,
-      "loss": 5.3539,
-      "step": 47595
-    },
-    {
-      "epoch": 24.821903520208604,
-      "grad_norm": 1.5665749311447144,
-      "learning_rate": 5.268542713567839e-05,
-      "loss": 5.0776,
-      "step": 47596
-    },
-    {
-      "epoch": 24.822425032594523,
-      "grad_norm": 1.5875684022903442,
-      "learning_rate": 5.268442211055277e-05,
-      "loss": 4.8961,
-      "step": 47597
-    },
-    {
-      "epoch": 24.822946544980443,
-      "grad_norm": 1.6239757537841797,
-      "learning_rate": 5.268341708542713e-05,
-      "loss": 5.1489,
-      "step": 47598
-    },
-    {
-      "epoch": 24.823468057366362,
-      "grad_norm": 1.5420277118682861,
-      "learning_rate": 5.268241206030151e-05,
-      "loss": 5.5853,
-      "step": 47599
-    },
-    {
-      "epoch": 24.823989569752282,
-      "grad_norm": 1.571365237236023,
-      "learning_rate": 5.2681407035175875e-05,
-      "loss": 5.3365,
-      "step": 47600
-    },
-    {
-      "epoch": 24.8245110821382,
-      "grad_norm": 1.5884332656860352,
-      "learning_rate": 5.268040201005025e-05,
-      "loss": 5.5718,
-      "step": 47601
-    },
-    {
-      "epoch": 24.82503259452412,
-      "grad_norm": 1.6147994995117188,
-      "learning_rate": 5.2679396984924624e-05,
-      "loss": 4.8736,
-      "step": 47602
-    },
-    {
-      "epoch": 24.82555410691004,
-      "grad_norm": 1.5263615846633911,
-      "learning_rate": 5.2678391959799e-05,
-      "loss": 5.5705,
-      "step": 47603
-    },
-    {
-      "epoch": 24.82607561929596,
-      "grad_norm": 1.6066384315490723,
-      "learning_rate": 5.2677386934673366e-05,
-      "loss": 5.1235,
-      "step": 47604
-    },
-    {
-      "epoch": 24.82659713168188,
-      "grad_norm": 1.5310699939727783,
-      "learning_rate": 5.2676381909547744e-05,
-      "loss": 5.273,
-      "step": 47605
-    },
-    {
-      "epoch": 24.827118644067795,
-      "grad_norm": 1.6136231422424316,
-      "learning_rate": 5.2675376884422115e-05,
-      "loss": 5.1345,
-      "step": 47606
-    },
-    {
-      "epoch": 24.827640156453715,
-      "grad_norm": 1.5032501220703125,
-      "learning_rate": 5.267437185929649e-05,
-      "loss": 5.5692,
-      "step": 47607
-    },
-    {
-      "epoch": 24.828161668839634,
-      "grad_norm": 1.6017577648162842,
-      "learning_rate": 5.267336683417086e-05,
-      "loss": 4.8972,
-      "step": 47608
-    },
-    {
-      "epoch": 24.828683181225554,
-      "grad_norm": 1.604788899421692,
-      "learning_rate": 5.267236180904522e-05,
-      "loss": 5.3332,
-      "step": 47609
-    },
-    {
-      "epoch": 24.829204693611473,
-      "grad_norm": 1.4956939220428467,
-      "learning_rate": 5.26713567839196e-05,
-      "loss": 5.3561,
-      "step": 47610
-    },
-    {
-      "epoch": 24.829726205997392,
-      "grad_norm": 1.6135096549987793,
-      "learning_rate": 5.267035175879397e-05,
-      "loss": 4.7716,
-      "step": 47611
-    },
-    {
-      "epoch": 24.830247718383312,
-      "grad_norm": 1.5763779878616333,
-      "learning_rate": 5.266934673366835e-05,
-      "loss": 5.2531,
-      "step": 47612
-    },
-    {
-      "epoch": 24.83076923076923,
-      "grad_norm": 1.5152612924575806,
-      "learning_rate": 5.266834170854271e-05,
-      "loss": 5.0948,
-      "step": 47613
-    },
-    {
-      "epoch": 24.83129074315515,
-      "grad_norm": 1.4788919687271118,
-      "learning_rate": 5.266733668341709e-05,
-      "loss": 4.9009,
-      "step": 47614
-    },
-    {
-      "epoch": 24.83181225554107,
-      "grad_norm": 1.542481541633606,
-      "learning_rate": 5.2666331658291454e-05,
-      "loss": 5.4679,
-      "step": 47615
-    },
-    {
-      "epoch": 24.83233376792699,
-      "grad_norm": 1.6971243619918823,
-      "learning_rate": 5.266532663316583e-05,
-      "loss": 4.5892,
-      "step": 47616
-    },
-    {
-      "epoch": 24.832855280312906,
-      "grad_norm": 1.5795730352401733,
-      "learning_rate": 5.26643216080402e-05,
-      "loss": 5.5169,
-      "step": 47617
-    },
-    {
-      "epoch": 24.833376792698825,
-      "grad_norm": 1.6188507080078125,
-      "learning_rate": 5.266331658291458e-05,
-      "loss": 5.0766,
-      "step": 47618
-    },
-    {
-      "epoch": 24.833898305084745,
-      "grad_norm": 1.555129051208496,
-      "learning_rate": 5.2662311557788945e-05,
-      "loss": 4.7554,
-      "step": 47619
-    },
-    {
-      "epoch": 24.834419817470664,
-      "grad_norm": 1.565237283706665,
-      "learning_rate": 5.266130653266332e-05,
-      "loss": 4.9967,
-      "step": 47620
-    },
-    {
-      "epoch": 24.834941329856584,
-      "grad_norm": 1.4539389610290527,
-      "learning_rate": 5.266030150753769e-05,
-      "loss": 4.9517,
-      "step": 47621
-    },
-    {
-      "epoch": 24.835462842242503,
-      "grad_norm": 1.577468991279602,
-      "learning_rate": 5.2659296482412065e-05,
-      "loss": 5.1513,
-      "step": 47622
-    },
-    {
-      "epoch": 24.835984354628422,
-      "grad_norm": 1.58712899684906,
-      "learning_rate": 5.2658291457286436e-05,
-      "loss": 5.3448,
-      "step": 47623
-    },
-    {
-      "epoch": 24.836505867014342,
-      "grad_norm": 1.4740113019943237,
-      "learning_rate": 5.26572864321608e-05,
-      "loss": 5.3956,
-      "step": 47624
-    },
-    {
-      "epoch": 24.83702737940026,
-      "grad_norm": 1.5789072513580322,
-      "learning_rate": 5.265628140703518e-05,
-      "loss": 5.4184,
-      "step": 47625
-    },
-    {
-      "epoch": 24.83754889178618,
-      "grad_norm": 1.563567042350769,
-      "learning_rate": 5.265527638190955e-05,
-      "loss": 5.2468,
-      "step": 47626
-    },
-    {
-      "epoch": 24.8380704041721,
-      "grad_norm": 1.6770111322402954,
-      "learning_rate": 5.265427135678393e-05,
-      "loss": 4.9138,
-      "step": 47627
-    },
-    {
-      "epoch": 24.83859191655802,
-      "grad_norm": 1.5910348892211914,
-      "learning_rate": 5.265326633165829e-05,
-      "loss": 5.3359,
-      "step": 47628
-    },
-    {
-      "epoch": 24.839113428943936,
-      "grad_norm": 1.5392276048660278,
-      "learning_rate": 5.265226130653267e-05,
-      "loss": 5.3807,
-      "step": 47629
-    },
-    {
-      "epoch": 24.839634941329855,
-      "grad_norm": 1.5421252250671387,
-      "learning_rate": 5.2651256281407034e-05,
-      "loss": 5.2895,
-      "step": 47630
-    },
-    {
-      "epoch": 24.840156453715775,
-      "grad_norm": 1.5147168636322021,
-      "learning_rate": 5.265025125628141e-05,
-      "loss": 5.6014,
-      "step": 47631
-    },
-    {
-      "epoch": 24.840677966101694,
-      "grad_norm": 1.5462509393692017,
-      "learning_rate": 5.264924623115578e-05,
-      "loss": 5.1455,
-      "step": 47632
-    },
-    {
-      "epoch": 24.841199478487614,
-      "grad_norm": 1.4600012302398682,
-      "learning_rate": 5.264824120603016e-05,
-      "loss": 5.2715,
-      "step": 47633
-    },
-    {
-      "epoch": 24.841720990873533,
-      "grad_norm": 1.5341097116470337,
-      "learning_rate": 5.2647236180904525e-05,
-      "loss": 5.6873,
-      "step": 47634
-    },
-    {
-      "epoch": 24.842242503259452,
-      "grad_norm": 1.7246912717819214,
-      "learning_rate": 5.26462311557789e-05,
-      "loss": 4.8405,
-      "step": 47635
-    },
-    {
-      "epoch": 24.842764015645372,
-      "grad_norm": 1.5018664598464966,
-      "learning_rate": 5.264522613065327e-05,
-      "loss": 5.1872,
-      "step": 47636
-    },
-    {
-      "epoch": 24.84328552803129,
-      "grad_norm": 1.5575308799743652,
-      "learning_rate": 5.264422110552764e-05,
-      "loss": 5.1835,
-      "step": 47637
-    },
-    {
-      "epoch": 24.84380704041721,
-      "grad_norm": 1.5339595079421997,
-      "learning_rate": 5.2643216080402016e-05,
-      "loss": 5.3441,
-      "step": 47638
-    },
-    {
-      "epoch": 24.84432855280313,
-      "grad_norm": 1.4532644748687744,
-      "learning_rate": 5.264221105527638e-05,
-      "loss": 5.6701,
-      "step": 47639
-    },
-    {
-      "epoch": 24.84485006518905,
-      "grad_norm": 1.7795687913894653,
-      "learning_rate": 5.264120603015076e-05,
-      "loss": 4.6195,
-      "step": 47640
-    },
-    {
-      "epoch": 24.845371577574966,
-      "grad_norm": 1.5197502374649048,
-      "learning_rate": 5.264020100502512e-05,
-      "loss": 5.1521,
-      "step": 47641
-    },
-    {
-      "epoch": 24.845893089960885,
-      "grad_norm": 1.5420690774917603,
-      "learning_rate": 5.26391959798995e-05,
-      "loss": 5.3525,
-      "step": 47642
-    },
-    {
-      "epoch": 24.846414602346805,
-      "grad_norm": 1.5969605445861816,
-      "learning_rate": 5.263819095477387e-05,
-      "loss": 5.4385,
-      "step": 47643
-    },
-    {
-      "epoch": 24.846936114732724,
-      "grad_norm": 1.6672930717468262,
-      "learning_rate": 5.263718592964825e-05,
-      "loss": 5.2916,
-      "step": 47644
-    },
-    {
-      "epoch": 24.847457627118644,
-      "grad_norm": 1.4754692316055298,
-      "learning_rate": 5.263618090452261e-05,
-      "loss": 5.265,
-      "step": 47645
-    },
-    {
-      "epoch": 24.847979139504563,
-      "grad_norm": 1.5418092012405396,
-      "learning_rate": 5.263517587939699e-05,
-      "loss": 5.2349,
-      "step": 47646
-    },
-    {
-      "epoch": 24.848500651890483,
-      "grad_norm": 1.5092413425445557,
-      "learning_rate": 5.263417085427136e-05,
-      "loss": 5.4792,
-      "step": 47647
-    },
-    {
-      "epoch": 24.849022164276402,
-      "grad_norm": 1.5439201593399048,
-      "learning_rate": 5.263316582914574e-05,
-      "loss": 5.7572,
-      "step": 47648
-    },
-    {
-      "epoch": 24.84954367666232,
-      "grad_norm": 1.547318696975708,
-      "learning_rate": 5.2632160804020104e-05,
-      "loss": 5.2754,
-      "step": 47649
-    },
-    {
-      "epoch": 24.85006518904824,
-      "grad_norm": 1.562667727470398,
-      "learning_rate": 5.263115577889447e-05,
-      "loss": 5.3841,
-      "step": 47650
-    },
-    {
-      "epoch": 24.85058670143416,
-      "grad_norm": 1.667661428451538,
-      "learning_rate": 5.2630150753768846e-05,
-      "loss": 5.0221,
-      "step": 47651
-    },
-    {
-      "epoch": 24.85110821382008,
-      "grad_norm": 1.6626861095428467,
-      "learning_rate": 5.262914572864322e-05,
-      "loss": 5.0782,
-      "step": 47652
-    },
-    {
-      "epoch": 24.851629726205996,
-      "grad_norm": 1.5533127784729004,
-      "learning_rate": 5.2628140703517595e-05,
-      "loss": 5.2047,
-      "step": 47653
-    },
-    {
-      "epoch": 24.852151238591915,
-      "grad_norm": 1.532079815864563,
-      "learning_rate": 5.262713567839196e-05,
-      "loss": 5.6026,
-      "step": 47654
-    },
-    {
-      "epoch": 24.852672750977835,
-      "grad_norm": 1.5754894018173218,
-      "learning_rate": 5.262613065326634e-05,
-      "loss": 5.182,
-      "step": 47655
-    },
-    {
-      "epoch": 24.853194263363754,
-      "grad_norm": 1.5834249258041382,
-      "learning_rate": 5.26251256281407e-05,
-      "loss": 4.8145,
-      "step": 47656
-    },
-    {
-      "epoch": 24.853715775749674,
-      "grad_norm": 1.4557965993881226,
-      "learning_rate": 5.262412060301508e-05,
-      "loss": 5.4316,
-      "step": 47657
-    },
-    {
-      "epoch": 24.854237288135593,
-      "grad_norm": 1.5624897480010986,
-      "learning_rate": 5.262311557788945e-05,
-      "loss": 5.2327,
-      "step": 47658
-    },
-    {
-      "epoch": 24.854758800521513,
-      "grad_norm": 1.5320183038711548,
-      "learning_rate": 5.262211055276383e-05,
-      "loss": 5.1006,
-      "step": 47659
-    },
-    {
-      "epoch": 24.855280312907432,
-      "grad_norm": 1.5800933837890625,
-      "learning_rate": 5.262110552763819e-05,
-      "loss": 5.0254,
-      "step": 47660
-    },
-    {
-      "epoch": 24.85580182529335,
-      "grad_norm": 1.6376460790634155,
-      "learning_rate": 5.262010050251257e-05,
-      "loss": 5.4482,
-      "step": 47661
-    },
-    {
-      "epoch": 24.85632333767927,
-      "grad_norm": 1.5008492469787598,
-      "learning_rate": 5.2619095477386934e-05,
-      "loss": 5.2549,
-      "step": 47662
-    },
-    {
-      "epoch": 24.85684485006519,
-      "grad_norm": 1.4822336435317993,
-      "learning_rate": 5.2618090452261305e-05,
-      "loss": 5.4814,
-      "step": 47663
-    },
-    {
-      "epoch": 24.85736636245111,
-      "grad_norm": 1.501574993133545,
-      "learning_rate": 5.261708542713568e-05,
-      "loss": 4.6984,
-      "step": 47664
-    },
-    {
-      "epoch": 24.857887874837026,
-      "grad_norm": 1.503880262374878,
-      "learning_rate": 5.261608040201005e-05,
-      "loss": 5.1773,
-      "step": 47665
-    },
-    {
-      "epoch": 24.858409387222945,
-      "grad_norm": 1.5224677324295044,
-      "learning_rate": 5.2615075376884425e-05,
-      "loss": 5.6097,
-      "step": 47666
-    },
-    {
-      "epoch": 24.858930899608865,
-      "grad_norm": 1.4554351568222046,
-      "learning_rate": 5.2614070351758796e-05,
-      "loss": 5.2694,
-      "step": 47667
-    },
-    {
-      "epoch": 24.859452411994784,
-      "grad_norm": 1.5911177396774292,
-      "learning_rate": 5.2613065326633174e-05,
-      "loss": 5.5906,
-      "step": 47668
-    },
-    {
-      "epoch": 24.859973924380704,
-      "grad_norm": 1.5939325094223022,
-      "learning_rate": 5.261206030150754e-05,
-      "loss": 5.5919,
-      "step": 47669
-    },
-    {
-      "epoch": 24.860495436766623,
-      "grad_norm": 1.3965059518814087,
-      "learning_rate": 5.2611055276381916e-05,
-      "loss": 5.4885,
-      "step": 47670
-    },
-    {
-      "epoch": 24.861016949152543,
-      "grad_norm": 1.6884772777557373,
-      "learning_rate": 5.261005025125628e-05,
-      "loss": 5.425,
-      "step": 47671
-    },
-    {
-      "epoch": 24.861538461538462,
-      "grad_norm": 1.5756975412368774,
-      "learning_rate": 5.260904522613066e-05,
-      "loss": 4.6471,
-      "step": 47672
-    },
-    {
-      "epoch": 24.86205997392438,
-      "grad_norm": 1.567591905593872,
-      "learning_rate": 5.260804020100503e-05,
-      "loss": 5.3167,
-      "step": 47673
-    },
-    {
-      "epoch": 24.8625814863103,
-      "grad_norm": 1.510602355003357,
-      "learning_rate": 5.260703517587941e-05,
-      "loss": 5.2839,
-      "step": 47674
-    },
-    {
-      "epoch": 24.86310299869622,
-      "grad_norm": 1.507896900177002,
-      "learning_rate": 5.260603015075377e-05,
-      "loss": 5.4122,
-      "step": 47675
-    },
-    {
-      "epoch": 24.86362451108214,
-      "grad_norm": 1.570114016532898,
-      "learning_rate": 5.2605025125628136e-05,
-      "loss": 5.1531,
-      "step": 47676
-    },
-    {
-      "epoch": 24.864146023468056,
-      "grad_norm": 1.5529382228851318,
-      "learning_rate": 5.2604020100502514e-05,
-      "loss": 5.0554,
-      "step": 47677
-    },
-    {
-      "epoch": 24.864667535853975,
-      "grad_norm": 1.5655509233474731,
-      "learning_rate": 5.2603015075376885e-05,
-      "loss": 5.7001,
-      "step": 47678
-    },
-    {
-      "epoch": 24.865189048239895,
-      "grad_norm": 1.515552282333374,
-      "learning_rate": 5.260201005025126e-05,
-      "loss": 5.5979,
-      "step": 47679
-    },
-    {
-      "epoch": 24.865710560625814,
-      "grad_norm": 1.6701767444610596,
-      "learning_rate": 5.260100502512563e-05,
-      "loss": 4.4712,
-      "step": 47680
-    },
-    {
-      "epoch": 24.866232073011734,
-      "grad_norm": 1.6492589712142944,
-      "learning_rate": 5.2600000000000005e-05,
-      "loss": 5.0752,
-      "step": 47681
-    },
-    {
-      "epoch": 24.866753585397653,
-      "grad_norm": 1.5632976293563843,
-      "learning_rate": 5.259899497487437e-05,
-      "loss": 5.6909,
-      "step": 47682
-    },
-    {
-      "epoch": 24.867275097783573,
-      "grad_norm": 1.4932217597961426,
-      "learning_rate": 5.259798994974875e-05,
-      "loss": 4.872,
-      "step": 47683
-    },
-    {
-      "epoch": 24.867796610169492,
-      "grad_norm": 1.4718550443649292,
-      "learning_rate": 5.259698492462312e-05,
-      "loss": 5.4693,
-      "step": 47684
-    },
-    {
-      "epoch": 24.86831812255541,
-      "grad_norm": 1.6471961736679077,
-      "learning_rate": 5.2595979899497496e-05,
-      "loss": 5.541,
-      "step": 47685
-    },
-    {
-      "epoch": 24.86883963494133,
-      "grad_norm": 1.6098086833953857,
-      "learning_rate": 5.259497487437186e-05,
-      "loss": 5.155,
-      "step": 47686
-    },
-    {
-      "epoch": 24.86936114732725,
-      "grad_norm": 1.603498935699463,
-      "learning_rate": 5.259396984924624e-05,
-      "loss": 5.2024,
-      "step": 47687
-    },
-    {
-      "epoch": 24.86988265971317,
-      "grad_norm": 1.5907758474349976,
-      "learning_rate": 5.25929648241206e-05,
-      "loss": 5.1382,
-      "step": 47688
-    },
-    {
-      "epoch": 24.870404172099086,
-      "grad_norm": 1.70477294921875,
-      "learning_rate": 5.259195979899497e-05,
-      "loss": 4.7903,
-      "step": 47689
-    },
-    {
-      "epoch": 24.870925684485005,
-      "grad_norm": 1.5717171430587769,
-      "learning_rate": 5.259095477386935e-05,
-      "loss": 5.168,
-      "step": 47690
-    },
-    {
-      "epoch": 24.871447196870925,
-      "grad_norm": 1.671024203300476,
-      "learning_rate": 5.2589949748743715e-05,
-      "loss": 5.0703,
-      "step": 47691
-    },
-    {
-      "epoch": 24.871968709256844,
-      "grad_norm": 1.630667805671692,
-      "learning_rate": 5.258894472361809e-05,
-      "loss": 5.5467,
-      "step": 47692
-    },
-    {
-      "epoch": 24.872490221642764,
-      "grad_norm": 1.5529078245162964,
-      "learning_rate": 5.2587939698492464e-05,
-      "loss": 5.2816,
-      "step": 47693
-    },
-    {
-      "epoch": 24.873011734028683,
-      "grad_norm": 1.6567729711532593,
-      "learning_rate": 5.258693467336684e-05,
-      "loss": 4.4639,
-      "step": 47694
-    },
-    {
-      "epoch": 24.873533246414603,
-      "grad_norm": 1.6699813604354858,
-      "learning_rate": 5.2585929648241206e-05,
-      "loss": 4.678,
-      "step": 47695
-    },
-    {
-      "epoch": 24.874054758800522,
-      "grad_norm": 1.5209894180297852,
-      "learning_rate": 5.2584924623115584e-05,
-      "loss": 5.5606,
-      "step": 47696
-    },
-    {
-      "epoch": 24.87457627118644,
-      "grad_norm": 1.7648942470550537,
-      "learning_rate": 5.258391959798995e-05,
-      "loss": 4.9322,
-      "step": 47697
-    },
-    {
-      "epoch": 24.87509778357236,
-      "grad_norm": 1.5355969667434692,
-      "learning_rate": 5.2582914572864326e-05,
-      "loss": 5.3695,
-      "step": 47698
-    },
-    {
-      "epoch": 24.87561929595828,
-      "grad_norm": 1.6538645029067993,
-      "learning_rate": 5.25819095477387e-05,
-      "loss": 4.7162,
-      "step": 47699
-    },
-    {
-      "epoch": 24.876140808344196,
-      "grad_norm": 1.6203669309616089,
-      "learning_rate": 5.2580904522613075e-05,
-      "loss": 5.0451,
-      "step": 47700
-    },
-    {
-      "epoch": 24.876662320730116,
-      "grad_norm": 1.5546715259552002,
-      "learning_rate": 5.257989949748744e-05,
-      "loss": 5.3306,
-      "step": 47701
-    },
-    {
-      "epoch": 24.877183833116035,
-      "grad_norm": 1.4750310182571411,
-      "learning_rate": 5.257889447236182e-05,
-      "loss": 5.5758,
-      "step": 47702
-    },
-    {
-      "epoch": 24.877705345501955,
-      "grad_norm": 1.5209474563598633,
-      "learning_rate": 5.257788944723618e-05,
-      "loss": 5.0943,
-      "step": 47703
-    },
-    {
-      "epoch": 24.878226857887874,
-      "grad_norm": 1.5052242279052734,
-      "learning_rate": 5.257688442211055e-05,
-      "loss": 4.8533,
-      "step": 47704
-    },
-    {
-      "epoch": 24.878748370273794,
-      "grad_norm": 1.4948164224624634,
-      "learning_rate": 5.257587939698493e-05,
-      "loss": 5.4799,
-      "step": 47705
-    },
-    {
-      "epoch": 24.879269882659713,
-      "grad_norm": 1.5050519704818726,
-      "learning_rate": 5.2574874371859294e-05,
-      "loss": 5.3991,
-      "step": 47706
-    },
-    {
-      "epoch": 24.879791395045633,
-      "grad_norm": 1.632494568824768,
-      "learning_rate": 5.257386934673367e-05,
-      "loss": 4.6097,
-      "step": 47707
-    },
-    {
-      "epoch": 24.880312907431552,
-      "grad_norm": 1.5771527290344238,
-      "learning_rate": 5.2572864321608037e-05,
-      "loss": 4.8484,
-      "step": 47708
-    },
-    {
-      "epoch": 24.88083441981747,
-      "grad_norm": 1.603529691696167,
-      "learning_rate": 5.2571859296482414e-05,
-      "loss": 5.4611,
-      "step": 47709
-    },
-    {
-      "epoch": 24.88135593220339,
-      "grad_norm": 1.5545287132263184,
-      "learning_rate": 5.2570854271356785e-05,
-      "loss": 5.251,
-      "step": 47710
-    },
-    {
-      "epoch": 24.88187744458931,
-      "grad_norm": 1.5102593898773193,
-      "learning_rate": 5.256984924623116e-05,
-      "loss": 4.9732,
-      "step": 47711
-    },
-    {
-      "epoch": 24.88239895697523,
-      "grad_norm": 1.7157131433486938,
-      "learning_rate": 5.256884422110553e-05,
-      "loss": 4.9523,
-      "step": 47712
-    },
-    {
-      "epoch": 24.882920469361146,
-      "grad_norm": 1.5542281866073608,
-      "learning_rate": 5.2567839195979905e-05,
-      "loss": 5.516,
-      "step": 47713
-    },
-    {
-      "epoch": 24.883441981747065,
-      "grad_norm": 1.7762529850006104,
-      "learning_rate": 5.2566834170854276e-05,
-      "loss": 5.1571,
-      "step": 47714
-    },
-    {
-      "epoch": 24.883963494132985,
-      "grad_norm": 1.6818736791610718,
-      "learning_rate": 5.2565829145728654e-05,
-      "loss": 4.8064,
-      "step": 47715
-    },
-    {
-      "epoch": 24.884485006518904,
-      "grad_norm": 1.5859419107437134,
-      "learning_rate": 5.256482412060302e-05,
-      "loss": 4.8007,
-      "step": 47716
-    },
-    {
-      "epoch": 24.885006518904824,
-      "grad_norm": 1.5227797031402588,
-      "learning_rate": 5.256381909547738e-05,
-      "loss": 5.5076,
-      "step": 47717
-    },
-    {
-      "epoch": 24.885528031290743,
-      "grad_norm": 1.6248798370361328,
-      "learning_rate": 5.256281407035176e-05,
-      "loss": 5.464,
-      "step": 47718
-    },
-    {
-      "epoch": 24.886049543676663,
-      "grad_norm": 1.5068378448486328,
-      "learning_rate": 5.256180904522613e-05,
-      "loss": 4.8819,
-      "step": 47719
-    },
-    {
-      "epoch": 24.886571056062582,
-      "grad_norm": 1.5771691799163818,
-      "learning_rate": 5.256080402010051e-05,
-      "loss": 5.1556,
-      "step": 47720
-    },
-    {
-      "epoch": 24.8870925684485,
-      "grad_norm": 1.6090469360351562,
-      "learning_rate": 5.2559798994974874e-05,
-      "loss": 5.3581,
-      "step": 47721
-    },
-    {
-      "epoch": 24.88761408083442,
-      "grad_norm": 1.485993504524231,
-      "learning_rate": 5.255879396984925e-05,
-      "loss": 5.5114,
-      "step": 47722
-    },
-    {
-      "epoch": 24.88813559322034,
-      "grad_norm": 1.6352717876434326,
-      "learning_rate": 5.2557788944723616e-05,
-      "loss": 4.8617,
-      "step": 47723
-    },
-    {
-      "epoch": 24.888657105606256,
-      "grad_norm": 1.5232905149459839,
-      "learning_rate": 5.2556783919597994e-05,
-      "loss": 5.1275,
-      "step": 47724
-    },
-    {
-      "epoch": 24.889178617992176,
-      "grad_norm": 1.6271828413009644,
-      "learning_rate": 5.2555778894472365e-05,
-      "loss": 4.7737,
-      "step": 47725
-    },
-    {
-      "epoch": 24.889700130378095,
-      "grad_norm": 1.543155550956726,
-      "learning_rate": 5.255477386934674e-05,
-      "loss": 5.5462,
-      "step": 47726
-    },
-    {
-      "epoch": 24.890221642764015,
-      "grad_norm": 1.6370857954025269,
-      "learning_rate": 5.255376884422111e-05,
-      "loss": 5.2907,
-      "step": 47727
-    },
-    {
-      "epoch": 24.890743155149934,
-      "grad_norm": 1.6214169263839722,
-      "learning_rate": 5.2552763819095485e-05,
-      "loss": 4.706,
-      "step": 47728
-    },
-    {
-      "epoch": 24.891264667535854,
-      "grad_norm": 1.5547282695770264,
-      "learning_rate": 5.255175879396985e-05,
-      "loss": 5.1151,
-      "step": 47729
-    },
-    {
-      "epoch": 24.891786179921773,
-      "grad_norm": 1.5775498151779175,
-      "learning_rate": 5.255075376884422e-05,
-      "loss": 5.2578,
-      "step": 47730
-    },
-    {
-      "epoch": 24.892307692307693,
-      "grad_norm": 1.5966057777404785,
-      "learning_rate": 5.25497487437186e-05,
-      "loss": 5.2258,
-      "step": 47731
-    },
-    {
-      "epoch": 24.892829204693612,
-      "grad_norm": 1.6480427980422974,
-      "learning_rate": 5.254874371859296e-05,
-      "loss": 5.1445,
-      "step": 47732
-    },
-    {
-      "epoch": 24.89335071707953,
-      "grad_norm": 1.6053638458251953,
-      "learning_rate": 5.254773869346734e-05,
-      "loss": 4.6525,
-      "step": 47733
-    },
-    {
-      "epoch": 24.89387222946545,
-      "grad_norm": 1.5487043857574463,
-      "learning_rate": 5.254673366834171e-05,
-      "loss": 5.0401,
-      "step": 47734
-    },
-    {
-      "epoch": 24.89439374185137,
-      "grad_norm": 1.4264172315597534,
-      "learning_rate": 5.254572864321609e-05,
-      "loss": 5.3292,
-      "step": 47735
-    },
-    {
-      "epoch": 24.894915254237286,
-      "grad_norm": 1.5879815816879272,
-      "learning_rate": 5.254472361809045e-05,
-      "loss": 4.9973,
-      "step": 47736
-    },
-    {
-      "epoch": 24.895436766623206,
-      "grad_norm": 1.603525996208191,
-      "learning_rate": 5.254371859296483e-05,
-      "loss": 5.4693,
-      "step": 47737
-    },
-    {
-      "epoch": 24.895958279009125,
-      "grad_norm": 1.5971696376800537,
-      "learning_rate": 5.2542713567839195e-05,
-      "loss": 5.2515,
-      "step": 47738
-    },
-    {
-      "epoch": 24.896479791395045,
-      "grad_norm": 1.7853882312774658,
-      "learning_rate": 5.254170854271357e-05,
-      "loss": 5.0559,
-      "step": 47739
-    },
-    {
-      "epoch": 24.897001303780964,
-      "grad_norm": 1.5833179950714111,
-      "learning_rate": 5.2540703517587944e-05,
-      "loss": 5.64,
-      "step": 47740
-    },
-    {
-      "epoch": 24.897522816166884,
-      "grad_norm": 1.564195156097412,
-      "learning_rate": 5.253969849246232e-05,
-      "loss": 5.0239,
-      "step": 47741
-    },
-    {
-      "epoch": 24.898044328552803,
-      "grad_norm": 1.623983383178711,
-      "learning_rate": 5.2538693467336686e-05,
-      "loss": 4.668,
-      "step": 47742
-    },
-    {
-      "epoch": 24.898565840938723,
-      "grad_norm": 1.4910074472427368,
-      "learning_rate": 5.253768844221105e-05,
-      "loss": 5.4858,
-      "step": 47743
-    },
-    {
-      "epoch": 24.899087353324642,
-      "grad_norm": 1.6191582679748535,
-      "learning_rate": 5.253668341708543e-05,
-      "loss": 5.0819,
-      "step": 47744
-    },
-    {
-      "epoch": 24.89960886571056,
-      "grad_norm": 1.6309188604354858,
-      "learning_rate": 5.25356783919598e-05,
-      "loss": 4.4997,
-      "step": 47745
-    },
-    {
-      "epoch": 24.90013037809648,
-      "grad_norm": 1.7423293590545654,
-      "learning_rate": 5.253467336683418e-05,
-      "loss": 5.0749,
-      "step": 47746
-    },
-    {
-      "epoch": 24.9006518904824,
-      "grad_norm": 1.5687360763549805,
-      "learning_rate": 5.253366834170854e-05,
-      "loss": 5.1844,
-      "step": 47747
-    },
-    {
-      "epoch": 24.901173402868316,
-      "grad_norm": 1.5586190223693848,
-      "learning_rate": 5.253266331658292e-05,
-      "loss": 5.2014,
-      "step": 47748
-    },
-    {
-      "epoch": 24.901694915254236,
-      "grad_norm": 1.6881275177001953,
-      "learning_rate": 5.2531658291457284e-05,
-      "loss": 5.3192,
-      "step": 47749
-    },
-    {
-      "epoch": 24.902216427640155,
-      "grad_norm": 1.6396729946136475,
-      "learning_rate": 5.253065326633166e-05,
-      "loss": 5.1135,
-      "step": 47750
-    },
-    {
-      "epoch": 24.902737940026075,
-      "grad_norm": 1.5733470916748047,
-      "learning_rate": 5.252964824120603e-05,
-      "loss": 5.3723,
-      "step": 47751
-    },
-    {
-      "epoch": 24.903259452411994,
-      "grad_norm": 1.5795012712478638,
-      "learning_rate": 5.252864321608041e-05,
-      "loss": 4.9149,
-      "step": 47752
-    },
-    {
-      "epoch": 24.903780964797914,
-      "grad_norm": 1.8583464622497559,
-      "learning_rate": 5.2527638190954775e-05,
-      "loss": 4.7027,
-      "step": 47753
-    },
-    {
-      "epoch": 24.904302477183833,
-      "grad_norm": 1.6588373184204102,
-      "learning_rate": 5.252663316582915e-05,
-      "loss": 4.9412,
-      "step": 47754
-    },
-    {
-      "epoch": 24.904823989569753,
-      "grad_norm": 1.5213452577590942,
-      "learning_rate": 5.252562814070352e-05,
-      "loss": 4.8114,
-      "step": 47755
-    },
-    {
-      "epoch": 24.905345501955672,
-      "grad_norm": 1.6171296834945679,
-      "learning_rate": 5.252462311557789e-05,
-      "loss": 5.2567,
-      "step": 47756
-    },
-    {
-      "epoch": 24.90586701434159,
-      "grad_norm": 1.5120751857757568,
-      "learning_rate": 5.2523618090452266e-05,
-      "loss": 5.2787,
-      "step": 47757
-    },
-    {
-      "epoch": 24.90638852672751,
-      "grad_norm": 1.57416570186615,
-      "learning_rate": 5.252261306532663e-05,
-      "loss": 5.0991,
-      "step": 47758
-    },
-    {
-      "epoch": 24.90691003911343,
-      "grad_norm": 1.5638071298599243,
-      "learning_rate": 5.252160804020101e-05,
-      "loss": 5.2184,
-      "step": 47759
-    },
-    {
-      "epoch": 24.907431551499347,
-      "grad_norm": 1.546887755393982,
-      "learning_rate": 5.252060301507538e-05,
-      "loss": 5.0591,
-      "step": 47760
-    },
-    {
-      "epoch": 24.907953063885266,
-      "grad_norm": 1.6743097305297852,
-      "learning_rate": 5.2519597989949757e-05,
-      "loss": 4.2793,
-      "step": 47761
-    },
-    {
-      "epoch": 24.908474576271185,
-      "grad_norm": 1.5390578508377075,
-      "learning_rate": 5.251859296482412e-05,
-      "loss": 5.1922,
-      "step": 47762
-    },
-    {
-      "epoch": 24.908996088657105,
-      "grad_norm": 1.5575982332229614,
-      "learning_rate": 5.25175879396985e-05,
-      "loss": 5.2024,
-      "step": 47763
-    },
-    {
-      "epoch": 24.909517601043024,
-      "grad_norm": 1.5341124534606934,
-      "learning_rate": 5.251658291457286e-05,
-      "loss": 4.7451,
-      "step": 47764
-    },
-    {
-      "epoch": 24.910039113428944,
-      "grad_norm": 1.5733764171600342,
-      "learning_rate": 5.251557788944724e-05,
-      "loss": 5.5622,
-      "step": 47765
-    },
-    {
-      "epoch": 24.910560625814863,
-      "grad_norm": 1.5478358268737793,
-      "learning_rate": 5.251457286432161e-05,
-      "loss": 4.9181,
-      "step": 47766
-    },
-    {
-      "epoch": 24.911082138200783,
-      "grad_norm": 1.4321041107177734,
-      "learning_rate": 5.251356783919599e-05,
-      "loss": 5.6996,
-      "step": 47767
-    },
-    {
-      "epoch": 24.911603650586702,
-      "grad_norm": 1.5911256074905396,
-      "learning_rate": 5.2512562814070354e-05,
-      "loss": 5.3327,
-      "step": 47768
-    },
-    {
-      "epoch": 24.91212516297262,
-      "grad_norm": 1.4958876371383667,
-      "learning_rate": 5.251155778894472e-05,
-      "loss": 5.6893,
-      "step": 47769
-    },
-    {
-      "epoch": 24.91264667535854,
-      "grad_norm": 1.538758635520935,
-      "learning_rate": 5.2510552763819096e-05,
-      "loss": 5.6016,
-      "step": 47770
-    },
-    {
-      "epoch": 24.91316818774446,
-      "grad_norm": 1.5784504413604736,
-      "learning_rate": 5.250954773869347e-05,
-      "loss": 4.855,
-      "step": 47771
-    },
-    {
-      "epoch": 24.913689700130377,
-      "grad_norm": 1.6337946653366089,
-      "learning_rate": 5.2508542713567845e-05,
-      "loss": 5.1146,
-      "step": 47772
-    },
-    {
-      "epoch": 24.914211212516296,
-      "grad_norm": 1.522896647453308,
-      "learning_rate": 5.250753768844221e-05,
-      "loss": 5.3469,
-      "step": 47773
-    },
-    {
-      "epoch": 24.914732724902215,
-      "grad_norm": 1.5711334943771362,
-      "learning_rate": 5.250653266331659e-05,
-      "loss": 4.9305,
-      "step": 47774
-    },
-    {
-      "epoch": 24.915254237288135,
-      "grad_norm": 1.6397209167480469,
-      "learning_rate": 5.250552763819095e-05,
-      "loss": 5.0464,
-      "step": 47775
-    },
-    {
-      "epoch": 24.915775749674054,
-      "grad_norm": 1.6840505599975586,
-      "learning_rate": 5.250452261306533e-05,
-      "loss": 5.4156,
-      "step": 47776
-    },
-    {
-      "epoch": 24.916297262059974,
-      "grad_norm": 1.6195839643478394,
-      "learning_rate": 5.25035175879397e-05,
-      "loss": 5.41,
-      "step": 47777
-    },
-    {
-      "epoch": 24.916818774445893,
-      "grad_norm": 1.568685531616211,
-      "learning_rate": 5.250251256281408e-05,
-      "loss": 5.4221,
-      "step": 47778
-    },
-    {
-      "epoch": 24.917340286831813,
-      "grad_norm": 1.6174300909042358,
-      "learning_rate": 5.250150753768844e-05,
-      "loss": 4.7464,
-      "step": 47779
-    },
-    {
-      "epoch": 24.917861799217732,
-      "grad_norm": 1.5490334033966064,
-      "learning_rate": 5.250050251256282e-05,
-      "loss": 5.3887,
-      "step": 47780
-    },
-    {
-      "epoch": 24.91838331160365,
-      "grad_norm": 1.7119126319885254,
-      "learning_rate": 5.249949748743719e-05,
-      "loss": 4.7146,
-      "step": 47781
-    },
-    {
-      "epoch": 24.91890482398957,
-      "grad_norm": 1.6385048627853394,
-      "learning_rate": 5.2498492462311555e-05,
-      "loss": 4.9601,
-      "step": 47782
-    },
-    {
-      "epoch": 24.919426336375487,
-      "grad_norm": 1.5603028535842896,
-      "learning_rate": 5.249748743718593e-05,
-      "loss": 5.252,
-      "step": 47783
-    },
-    {
-      "epoch": 24.919947848761407,
-      "grad_norm": 1.5985231399536133,
-      "learning_rate": 5.24964824120603e-05,
-      "loss": 5.3915,
-      "step": 47784
-    },
-    {
-      "epoch": 24.920469361147326,
-      "grad_norm": 1.4744200706481934,
-      "learning_rate": 5.2495477386934675e-05,
-      "loss": 5.5178,
-      "step": 47785
-    },
-    {
-      "epoch": 24.920990873533245,
-      "grad_norm": 1.6509729623794556,
-      "learning_rate": 5.2494472361809046e-05,
-      "loss": 5.1761,
-      "step": 47786
-    },
-    {
-      "epoch": 24.921512385919165,
-      "grad_norm": 1.5562658309936523,
-      "learning_rate": 5.2493467336683424e-05,
-      "loss": 5.3523,
-      "step": 47787
-    },
-    {
-      "epoch": 24.922033898305084,
-      "grad_norm": 1.6325427293777466,
-      "learning_rate": 5.249246231155779e-05,
-      "loss": 4.9509,
-      "step": 47788
-    },
-    {
-      "epoch": 24.922555410691004,
-      "grad_norm": 1.4898924827575684,
-      "learning_rate": 5.2491457286432166e-05,
-      "loss": 5.384,
-      "step": 47789
-    },
-    {
-      "epoch": 24.923076923076923,
-      "grad_norm": 1.4454082250595093,
-      "learning_rate": 5.249045226130653e-05,
-      "loss": 5.1173,
-      "step": 47790
-    },
-    {
-      "epoch": 24.923598435462843,
-      "grad_norm": 1.522719144821167,
-      "learning_rate": 5.248944723618091e-05,
-      "loss": 5.2724,
-      "step": 47791
-    },
-    {
-      "epoch": 24.924119947848762,
-      "grad_norm": 1.6397840976715088,
-      "learning_rate": 5.248844221105528e-05,
-      "loss": 5.1661,
-      "step": 47792
-    },
-    {
-      "epoch": 24.92464146023468,
-      "grad_norm": 1.563125491142273,
-      "learning_rate": 5.248743718592966e-05,
-      "loss": 4.8694,
-      "step": 47793
-    },
-    {
-      "epoch": 24.9251629726206,
-      "grad_norm": 1.501402497291565,
-      "learning_rate": 5.248643216080402e-05,
-      "loss": 5.5893,
-      "step": 47794
-    },
-    {
-      "epoch": 24.92568448500652,
-      "grad_norm": 1.580520510673523,
-      "learning_rate": 5.24854271356784e-05,
-      "loss": 5.2237,
-      "step": 47795
-    },
-    {
-      "epoch": 24.926205997392437,
-      "grad_norm": 1.56383216381073,
-      "learning_rate": 5.2484422110552764e-05,
-      "loss": 4.8392,
-      "step": 47796
-    },
-    {
-      "epoch": 24.926727509778356,
-      "grad_norm": 1.5567612648010254,
-      "learning_rate": 5.2483417085427135e-05,
-      "loss": 5.2373,
-      "step": 47797
-    },
-    {
-      "epoch": 24.927249022164276,
-      "grad_norm": 1.5055768489837646,
-      "learning_rate": 5.248241206030151e-05,
-      "loss": 4.8063,
-      "step": 47798
-    },
-    {
-      "epoch": 24.927770534550195,
-      "grad_norm": 1.5126721858978271,
-      "learning_rate": 5.248140703517588e-05,
-      "loss": 5.4846,
-      "step": 47799
-    },
-    {
-      "epoch": 24.928292046936114,
-      "grad_norm": 1.479546070098877,
-      "learning_rate": 5.2480402010050255e-05,
-      "loss": 5.2124,
-      "step": 47800
-    },
-    {
-      "epoch": 24.928813559322034,
-      "grad_norm": 1.5349624156951904,
-      "learning_rate": 5.2479396984924626e-05,
-      "loss": 5.1559,
-      "step": 47801
-    },
-    {
-      "epoch": 24.929335071707953,
-      "grad_norm": 1.575971245765686,
-      "learning_rate": 5.2478391959799004e-05,
-      "loss": 5.4469,
-      "step": 47802
-    },
-    {
-      "epoch": 24.929856584093873,
-      "grad_norm": 1.5864543914794922,
-      "learning_rate": 5.247738693467337e-05,
-      "loss": 5.2331,
-      "step": 47803
-    },
-    {
-      "epoch": 24.930378096479792,
-      "grad_norm": 1.5096724033355713,
-      "learning_rate": 5.2476381909547746e-05,
-      "loss": 4.6205,
-      "step": 47804
-    },
-    {
-      "epoch": 24.93089960886571,
-      "grad_norm": 1.5665311813354492,
-      "learning_rate": 5.247537688442211e-05,
-      "loss": 4.6509,
-      "step": 47805
-    },
-    {
-      "epoch": 24.93142112125163,
-      "grad_norm": 1.529343605041504,
-      "learning_rate": 5.247437185929649e-05,
-      "loss": 5.2868,
-      "step": 47806
-    },
-    {
-      "epoch": 24.931942633637547,
-      "grad_norm": 1.5825896263122559,
-      "learning_rate": 5.247336683417086e-05,
-      "loss": 5.3469,
-      "step": 47807
-    },
-    {
-      "epoch": 24.932464146023467,
-      "grad_norm": 1.6180617809295654,
-      "learning_rate": 5.2472361809045237e-05,
-      "loss": 5.3011,
-      "step": 47808
-    },
-    {
-      "epoch": 24.932985658409386,
-      "grad_norm": 1.525445580482483,
-      "learning_rate": 5.24713567839196e-05,
-      "loss": 4.9811,
-      "step": 47809
-    },
-    {
-      "epoch": 24.933507170795306,
-      "grad_norm": 1.46649968624115,
-      "learning_rate": 5.2470351758793965e-05,
-      "loss": 4.7676,
-      "step": 47810
-    },
-    {
-      "epoch": 24.934028683181225,
-      "grad_norm": 1.604508638381958,
-      "learning_rate": 5.246934673366834e-05,
-      "loss": 5.3004,
-      "step": 47811
-    },
-    {
-      "epoch": 24.934550195567144,
-      "grad_norm": 1.587953805923462,
-      "learning_rate": 5.2468341708542714e-05,
-      "loss": 4.6854,
-      "step": 47812
-    },
-    {
-      "epoch": 24.935071707953064,
-      "grad_norm": 1.629517674446106,
-      "learning_rate": 5.246733668341709e-05,
-      "loss": 5.1924,
-      "step": 47813
-    },
-    {
-      "epoch": 24.935593220338983,
-      "grad_norm": 1.5708280801773071,
-      "learning_rate": 5.2466331658291456e-05,
-      "loss": 5.1146,
-      "step": 47814
-    },
-    {
-      "epoch": 24.936114732724903,
-      "grad_norm": 1.6082669496536255,
-      "learning_rate": 5.2465326633165834e-05,
-      "loss": 4.7022,
-      "step": 47815
-    },
-    {
-      "epoch": 24.936636245110822,
-      "grad_norm": 1.5975440740585327,
-      "learning_rate": 5.24643216080402e-05,
-      "loss": 5.44,
-      "step": 47816
-    },
-    {
-      "epoch": 24.937157757496742,
-      "grad_norm": 1.5923064947128296,
-      "learning_rate": 5.2463316582914576e-05,
-      "loss": 5.5848,
-      "step": 47817
-    },
-    {
-      "epoch": 24.93767926988266,
-      "grad_norm": 1.4620832204818726,
-      "learning_rate": 5.246231155778895e-05,
-      "loss": 5.2175,
-      "step": 47818
-    },
-    {
-      "epoch": 24.938200782268577,
-      "grad_norm": 1.6335848569869995,
-      "learning_rate": 5.2461306532663325e-05,
-      "loss": 5.4373,
-      "step": 47819
-    },
-    {
-      "epoch": 24.938722294654497,
-      "grad_norm": 1.5892499685287476,
-      "learning_rate": 5.246030150753769e-05,
-      "loss": 4.8808,
-      "step": 47820
-    },
-    {
-      "epoch": 24.939243807040416,
-      "grad_norm": 1.4610767364501953,
-      "learning_rate": 5.245929648241207e-05,
-      "loss": 5.3893,
-      "step": 47821
-    },
-    {
-      "epoch": 24.939765319426336,
-      "grad_norm": 1.6223840713500977,
-      "learning_rate": 5.245829145728644e-05,
-      "loss": 4.9274,
-      "step": 47822
-    },
-    {
-      "epoch": 24.940286831812255,
-      "grad_norm": 1.655417561531067,
-      "learning_rate": 5.24572864321608e-05,
-      "loss": 4.8352,
-      "step": 47823
-    },
-    {
-      "epoch": 24.940808344198174,
-      "grad_norm": 1.625246524810791,
-      "learning_rate": 5.245628140703518e-05,
-      "loss": 5.2717,
-      "step": 47824
-    },
-    {
-      "epoch": 24.941329856584094,
-      "grad_norm": 1.5081887245178223,
-      "learning_rate": 5.2455276381909544e-05,
-      "loss": 5.1119,
-      "step": 47825
-    },
-    {
-      "epoch": 24.941851368970013,
-      "grad_norm": 1.5487219095230103,
-      "learning_rate": 5.245427135678392e-05,
-      "loss": 4.9596,
-      "step": 47826
-    },
-    {
-      "epoch": 24.942372881355933,
-      "grad_norm": 1.5525285005569458,
-      "learning_rate": 5.245326633165829e-05,
-      "loss": 5.1782,
-      "step": 47827
-    },
-    {
-      "epoch": 24.942894393741852,
-      "grad_norm": 1.570200800895691,
-      "learning_rate": 5.245226130653267e-05,
-      "loss": 5.1889,
-      "step": 47828
-    },
-    {
-      "epoch": 24.943415906127772,
-      "grad_norm": 1.6159007549285889,
-      "learning_rate": 5.2451256281407035e-05,
-      "loss": 5.2351,
-      "step": 47829
-    },
-    {
-      "epoch": 24.94393741851369,
-      "grad_norm": 1.6352791786193848,
-      "learning_rate": 5.245025125628141e-05,
-      "loss": 5.5035,
-      "step": 47830
-    },
-    {
-      "epoch": 24.944458930899607,
-      "grad_norm": 1.5439342260360718,
-      "learning_rate": 5.244924623115578e-05,
-      "loss": 5.227,
-      "step": 47831
-    },
-    {
-      "epoch": 24.944980443285527,
-      "grad_norm": 1.6885194778442383,
-      "learning_rate": 5.2448241206030155e-05,
-      "loss": 4.7509,
-      "step": 47832
-    },
-    {
-      "epoch": 24.945501955671446,
-      "grad_norm": 1.5482327938079834,
-      "learning_rate": 5.2447236180904526e-05,
-      "loss": 5.0549,
-      "step": 47833
-    },
-    {
-      "epoch": 24.946023468057366,
-      "grad_norm": 1.6050736904144287,
-      "learning_rate": 5.2446231155778904e-05,
-      "loss": 5.2309,
-      "step": 47834
-    },
-    {
-      "epoch": 24.946544980443285,
-      "grad_norm": 1.650611162185669,
-      "learning_rate": 5.244522613065327e-05,
-      "loss": 5.2114,
-      "step": 47835
-    },
-    {
-      "epoch": 24.947066492829205,
-      "grad_norm": 1.5977762937545776,
-      "learning_rate": 5.244422110552763e-05,
-      "loss": 5.3857,
-      "step": 47836
-    },
-    {
-      "epoch": 24.947588005215124,
-      "grad_norm": 1.5972933769226074,
-      "learning_rate": 5.244321608040201e-05,
-      "loss": 4.8976,
-      "step": 47837
-    },
-    {
-      "epoch": 24.948109517601043,
-      "grad_norm": 1.5562890768051147,
-      "learning_rate": 5.244221105527638e-05,
-      "loss": 4.9802,
-      "step": 47838
-    },
-    {
-      "epoch": 24.948631029986963,
-      "grad_norm": 1.4937858581542969,
-      "learning_rate": 5.244120603015076e-05,
-      "loss": 5.3736,
-      "step": 47839
-    },
-    {
-      "epoch": 24.949152542372882,
-      "grad_norm": 1.6591356992721558,
-      "learning_rate": 5.2440201005025124e-05,
-      "loss": 5.1913,
-      "step": 47840
-    },
-    {
-      "epoch": 24.949674054758802,
-      "grad_norm": 1.650553584098816,
-      "learning_rate": 5.24391959798995e-05,
-      "loss": 5.2942,
-      "step": 47841
-    },
-    {
-      "epoch": 24.95019556714472,
-      "grad_norm": 1.531273365020752,
-      "learning_rate": 5.2438190954773866e-05,
-      "loss": 5.4793,
-      "step": 47842
-    },
-    {
-      "epoch": 24.950717079530637,
-      "grad_norm": 1.4849472045898438,
-      "learning_rate": 5.2437185929648244e-05,
-      "loss": 5.0029,
-      "step": 47843
-    },
-    {
-      "epoch": 24.951238591916557,
-      "grad_norm": 1.550865888595581,
-      "learning_rate": 5.2436180904522615e-05,
-      "loss": 5.1235,
-      "step": 47844
-    },
-    {
-      "epoch": 24.951760104302476,
-      "grad_norm": 1.67410409450531,
-      "learning_rate": 5.243517587939699e-05,
-      "loss": 5.028,
-      "step": 47845
-    },
-    {
-      "epoch": 24.952281616688396,
-      "grad_norm": 1.5550713539123535,
-      "learning_rate": 5.243417085427136e-05,
-      "loss": 5.3094,
-      "step": 47846
-    },
-    {
-      "epoch": 24.952803129074315,
-      "grad_norm": 1.4926759004592896,
-      "learning_rate": 5.2433165829145735e-05,
-      "loss": 4.7864,
-      "step": 47847
-    },
-    {
-      "epoch": 24.953324641460235,
-      "grad_norm": 1.5091379880905151,
-      "learning_rate": 5.2432160804020106e-05,
-      "loss": 5.639,
-      "step": 47848
-    },
-    {
-      "epoch": 24.953846153846154,
-      "grad_norm": 1.5771652460098267,
-      "learning_rate": 5.243115577889447e-05,
-      "loss": 5.3105,
-      "step": 47849
-    },
-    {
-      "epoch": 24.954367666232073,
-      "grad_norm": 1.619673490524292,
-      "learning_rate": 5.243015075376885e-05,
-      "loss": 4.6853,
-      "step": 47850
-    },
-    {
-      "epoch": 24.954889178617993,
-      "grad_norm": 1.643508791923523,
-      "learning_rate": 5.242914572864321e-05,
-      "loss": 5.1724,
-      "step": 47851
-    },
-    {
-      "epoch": 24.955410691003912,
-      "grad_norm": 1.526640772819519,
-      "learning_rate": 5.242814070351759e-05,
-      "loss": 5.337,
-      "step": 47852
-    },
-    {
-      "epoch": 24.955932203389832,
-      "grad_norm": 1.5601602792739868,
-      "learning_rate": 5.242713567839196e-05,
-      "loss": 5.4956,
-      "step": 47853
-    },
-    {
-      "epoch": 24.95645371577575,
-      "grad_norm": 1.5568257570266724,
-      "learning_rate": 5.242613065326634e-05,
-      "loss": 5.6838,
-      "step": 47854
-    },
-    {
-      "epoch": 24.956975228161667,
-      "grad_norm": 1.599039077758789,
-      "learning_rate": 5.24251256281407e-05,
-      "loss": 5.2723,
-      "step": 47855
-    },
-    {
-      "epoch": 24.957496740547587,
-      "grad_norm": 1.5869622230529785,
-      "learning_rate": 5.242412060301508e-05,
-      "loss": 5.1557,
-      "step": 47856
-    },
-    {
-      "epoch": 24.958018252933506,
-      "grad_norm": 1.547155499458313,
-      "learning_rate": 5.2423115577889445e-05,
-      "loss": 5.3211,
-      "step": 47857
-    },
-    {
-      "epoch": 24.958539765319426,
-      "grad_norm": 1.4963464736938477,
-      "learning_rate": 5.242211055276382e-05,
-      "loss": 5.4319,
-      "step": 47858
-    },
-    {
-      "epoch": 24.959061277705345,
-      "grad_norm": 1.5818794965744019,
-      "learning_rate": 5.2421105527638194e-05,
-      "loss": 4.9564,
-      "step": 47859
-    },
-    {
-      "epoch": 24.959582790091265,
-      "grad_norm": 1.60922372341156,
-      "learning_rate": 5.242010050251257e-05,
-      "loss": 5.6372,
-      "step": 47860
-    },
-    {
-      "epoch": 24.960104302477184,
-      "grad_norm": 1.5624961853027344,
-      "learning_rate": 5.2419095477386936e-05,
-      "loss": 5.4488,
-      "step": 47861
-    },
-    {
-      "epoch": 24.960625814863103,
-      "grad_norm": 1.547555685043335,
-      "learning_rate": 5.24180904522613e-05,
-      "loss": 5.3291,
-      "step": 47862
-    },
-    {
-      "epoch": 24.961147327249023,
-      "grad_norm": 1.5541396141052246,
-      "learning_rate": 5.241708542713568e-05,
-      "loss": 5.0084,
-      "step": 47863
-    },
-    {
-      "epoch": 24.961668839634942,
-      "grad_norm": 1.5431007146835327,
-      "learning_rate": 5.241608040201005e-05,
-      "loss": 5.3004,
-      "step": 47864
-    },
-    {
-      "epoch": 24.962190352020862,
-      "grad_norm": 1.6425081491470337,
-      "learning_rate": 5.241507537688443e-05,
-      "loss": 5.0095,
-      "step": 47865
-    },
-    {
-      "epoch": 24.96271186440678,
-      "grad_norm": 1.5993643999099731,
-      "learning_rate": 5.241407035175879e-05,
-      "loss": 5.2458,
-      "step": 47866
-    },
-    {
-      "epoch": 24.963233376792697,
-      "grad_norm": 1.5550777912139893,
-      "learning_rate": 5.241306532663317e-05,
-      "loss": 5.1539,
-      "step": 47867
-    },
-    {
-      "epoch": 24.963754889178617,
-      "grad_norm": 1.5212130546569824,
-      "learning_rate": 5.241206030150754e-05,
-      "loss": 4.9883,
-      "step": 47868
-    },
-    {
-      "epoch": 24.964276401564536,
-      "grad_norm": 1.6037548780441284,
-      "learning_rate": 5.241105527638192e-05,
-      "loss": 5.3218,
-      "step": 47869
-    },
-    {
-      "epoch": 24.964797913950456,
-      "grad_norm": 1.4925531148910522,
-      "learning_rate": 5.241005025125628e-05,
-      "loss": 5.5152,
-      "step": 47870
-    },
-    {
-      "epoch": 24.965319426336375,
-      "grad_norm": 1.5641018152236938,
-      "learning_rate": 5.240904522613066e-05,
-      "loss": 5.1541,
-      "step": 47871
-    },
-    {
-      "epoch": 24.965840938722295,
-      "grad_norm": 1.5720890760421753,
-      "learning_rate": 5.2408040201005025e-05,
-      "loss": 5.2101,
-      "step": 47872
-    },
-    {
-      "epoch": 24.966362451108214,
-      "grad_norm": 1.4760297536849976,
-      "learning_rate": 5.24070351758794e-05,
-      "loss": 5.6326,
-      "step": 47873
-    },
-    {
-      "epoch": 24.966883963494134,
-      "grad_norm": 1.6819473505020142,
-      "learning_rate": 5.2406030150753773e-05,
-      "loss": 5.0774,
-      "step": 47874
-    },
-    {
-      "epoch": 24.967405475880053,
-      "grad_norm": 1.536594033241272,
-      "learning_rate": 5.240502512562814e-05,
-      "loss": 5.234,
-      "step": 47875
-    },
-    {
-      "epoch": 24.967926988265972,
-      "grad_norm": 1.520174264907837,
-      "learning_rate": 5.2404020100502516e-05,
-      "loss": 5.253,
-      "step": 47876
-    },
-    {
-      "epoch": 24.968448500651892,
-      "grad_norm": 1.5528056621551514,
-      "learning_rate": 5.240301507537688e-05,
-      "loss": 5.6703,
-      "step": 47877
-    },
-    {
-      "epoch": 24.96897001303781,
-      "grad_norm": 1.610087513923645,
-      "learning_rate": 5.240201005025126e-05,
-      "loss": 5.4435,
-      "step": 47878
-    },
-    {
-      "epoch": 24.969491525423727,
-      "grad_norm": 1.4586291313171387,
-      "learning_rate": 5.240100502512563e-05,
-      "loss": 5.2981,
-      "step": 47879
-    },
-    {
-      "epoch": 24.970013037809647,
-      "grad_norm": 1.5283790826797485,
-      "learning_rate": 5.2400000000000007e-05,
-      "loss": 5.3911,
-      "step": 47880
-    },
-    {
-      "epoch": 24.970534550195566,
-      "grad_norm": 1.5742734670639038,
-      "learning_rate": 5.239899497487437e-05,
-      "loss": 5.6052,
-      "step": 47881
-    },
-    {
-      "epoch": 24.971056062581486,
-      "grad_norm": 1.5516984462738037,
-      "learning_rate": 5.239798994974875e-05,
-      "loss": 5.2629,
-      "step": 47882
-    },
-    {
-      "epoch": 24.971577574967405,
-      "grad_norm": 1.622444748878479,
-      "learning_rate": 5.239698492462311e-05,
-      "loss": 5.3749,
-      "step": 47883
-    },
-    {
-      "epoch": 24.972099087353325,
-      "grad_norm": 1.4948434829711914,
-      "learning_rate": 5.239597989949749e-05,
-      "loss": 5.6435,
-      "step": 47884
-    },
-    {
-      "epoch": 24.972620599739244,
-      "grad_norm": 1.5284301042556763,
-      "learning_rate": 5.239497487437186e-05,
-      "loss": 4.8262,
-      "step": 47885
-    },
-    {
-      "epoch": 24.973142112125164,
-      "grad_norm": 1.5983543395996094,
-      "learning_rate": 5.239396984924624e-05,
-      "loss": 4.7585,
-      "step": 47886
-    },
-    {
-      "epoch": 24.973663624511083,
-      "grad_norm": 1.4761754274368286,
-      "learning_rate": 5.2392964824120604e-05,
-      "loss": 5.2941,
-      "step": 47887
-    },
-    {
-      "epoch": 24.974185136897002,
-      "grad_norm": 1.4496870040893555,
-      "learning_rate": 5.239195979899498e-05,
-      "loss": 5.711,
-      "step": 47888
-    },
-    {
-      "epoch": 24.974706649282922,
-      "grad_norm": 1.530072569847107,
-      "learning_rate": 5.239095477386935e-05,
-      "loss": 5.3934,
-      "step": 47889
-    },
-    {
-      "epoch": 24.975228161668838,
-      "grad_norm": 1.5856667757034302,
-      "learning_rate": 5.238994974874372e-05,
-      "loss": 4.9854,
-      "step": 47890
-    },
-    {
-      "epoch": 24.975749674054757,
-      "grad_norm": 1.494483470916748,
-      "learning_rate": 5.2388944723618095e-05,
-      "loss": 5.0581,
-      "step": 47891
-    },
-    {
-      "epoch": 24.976271186440677,
-      "grad_norm": 1.594369649887085,
-      "learning_rate": 5.238793969849246e-05,
-      "loss": 5.1751,
-      "step": 47892
-    },
-    {
-      "epoch": 24.976792698826596,
-      "grad_norm": 1.4724606275558472,
-      "learning_rate": 5.238693467336684e-05,
-      "loss": 4.5484,
-      "step": 47893
-    },
-    {
-      "epoch": 24.977314211212516,
-      "grad_norm": 1.499943494796753,
-      "learning_rate": 5.238592964824121e-05,
-      "loss": 5.4197,
-      "step": 47894
-    },
-    {
-      "epoch": 24.977835723598435,
-      "grad_norm": 1.6119844913482666,
-      "learning_rate": 5.2384924623115586e-05,
-      "loss": 5.2106,
-      "step": 47895
-    },
-    {
-      "epoch": 24.978357235984355,
-      "grad_norm": 1.5109329223632812,
-      "learning_rate": 5.238391959798995e-05,
-      "loss": 5.4181,
-      "step": 47896
-    },
-    {
-      "epoch": 24.978878748370274,
-      "grad_norm": 1.5986223220825195,
-      "learning_rate": 5.238291457286433e-05,
-      "loss": 4.9645,
-      "step": 47897
-    },
-    {
-      "epoch": 24.979400260756194,
-      "grad_norm": 1.4698710441589355,
-      "learning_rate": 5.238190954773869e-05,
-      "loss": 5.4356,
-      "step": 47898
-    },
-    {
-      "epoch": 24.979921773142113,
-      "grad_norm": 1.4777920246124268,
-      "learning_rate": 5.238090452261307e-05,
-      "loss": 5.2425,
-      "step": 47899
-    },
-    {
-      "epoch": 24.980443285528033,
-      "grad_norm": 1.4898412227630615,
-      "learning_rate": 5.237989949748744e-05,
-      "loss": 5.556,
-      "step": 47900
-    },
-    {
-      "epoch": 24.980964797913952,
-      "grad_norm": 1.624388337135315,
-      "learning_rate": 5.237889447236182e-05,
-      "loss": 4.7454,
-      "step": 47901
-    },
-    {
-      "epoch": 24.98148631029987,
-      "grad_norm": 1.5074251890182495,
-      "learning_rate": 5.237788944723618e-05,
-      "loss": 5.4303,
-      "step": 47902
-    },
-    {
-      "epoch": 24.982007822685787,
-      "grad_norm": 1.5527795553207397,
-      "learning_rate": 5.237688442211055e-05,
-      "loss": 5.2367,
-      "step": 47903
-    },
-    {
-      "epoch": 24.982529335071707,
-      "grad_norm": 1.563732385635376,
-      "learning_rate": 5.2375879396984925e-05,
-      "loss": 5.2297,
-      "step": 47904
-    },
-    {
-      "epoch": 24.983050847457626,
-      "grad_norm": 1.6776899099349976,
-      "learning_rate": 5.2374874371859296e-05,
-      "loss": 4.5675,
-      "step": 47905
-    },
-    {
-      "epoch": 24.983572359843546,
-      "grad_norm": 1.5880287885665894,
-      "learning_rate": 5.2373869346733674e-05,
-      "loss": 5.2222,
-      "step": 47906
-    },
-    {
-      "epoch": 24.984093872229465,
-      "grad_norm": 1.6778610944747925,
-      "learning_rate": 5.237286432160804e-05,
-      "loss": 4.8532,
-      "step": 47907
-    },
-    {
-      "epoch": 24.984615384615385,
-      "grad_norm": 1.7292981147766113,
-      "learning_rate": 5.2371859296482416e-05,
-      "loss": 5.0938,
-      "step": 47908
-    },
-    {
-      "epoch": 24.985136897001304,
-      "grad_norm": 1.70377516746521,
-      "learning_rate": 5.237085427135679e-05,
-      "loss": 5.3753,
-      "step": 47909
-    },
-    {
-      "epoch": 24.985658409387224,
-      "grad_norm": 1.7160298824310303,
-      "learning_rate": 5.2369849246231165e-05,
-      "loss": 4.6394,
-      "step": 47910
-    },
-    {
-      "epoch": 24.986179921773143,
-      "grad_norm": 1.576291799545288,
-      "learning_rate": 5.236884422110553e-05,
-      "loss": 5.343,
-      "step": 47911
-    },
-    {
-      "epoch": 24.986701434159063,
-      "grad_norm": 1.6506407260894775,
-      "learning_rate": 5.236783919597991e-05,
-      "loss": 5.13,
-      "step": 47912
-    },
-    {
-      "epoch": 24.987222946544982,
-      "grad_norm": 1.6022146940231323,
-      "learning_rate": 5.236683417085427e-05,
-      "loss": 5.4724,
-      "step": 47913
-    },
-    {
-      "epoch": 24.987744458930898,
-      "grad_norm": 1.5101474523544312,
-      "learning_rate": 5.236582914572865e-05,
-      "loss": 5.2945,
-      "step": 47914
-    },
-    {
-      "epoch": 24.988265971316817,
-      "grad_norm": 1.5308173894882202,
-      "learning_rate": 5.236482412060302e-05,
-      "loss": 5.2781,
-      "step": 47915
-    },
-    {
-      "epoch": 24.988787483702737,
-      "grad_norm": 1.5754930973052979,
-      "learning_rate": 5.2363819095477385e-05,
-      "loss": 5.2611,
-      "step": 47916
-    },
-    {
-      "epoch": 24.989308996088656,
-      "grad_norm": 1.6063728332519531,
-      "learning_rate": 5.236281407035176e-05,
-      "loss": 5.3116,
-      "step": 47917
-    },
-    {
-      "epoch": 24.989830508474576,
-      "grad_norm": 1.6133534908294678,
-      "learning_rate": 5.236180904522613e-05,
-      "loss": 4.9285,
-      "step": 47918
-    },
-    {
-      "epoch": 24.990352020860495,
-      "grad_norm": 1.5047926902770996,
-      "learning_rate": 5.2360804020100505e-05,
-      "loss": 5.5971,
-      "step": 47919
-    },
-    {
-      "epoch": 24.990873533246415,
-      "grad_norm": 1.448617696762085,
-      "learning_rate": 5.2359798994974876e-05,
-      "loss": 5.3678,
-      "step": 47920
-    },
-    {
-      "epoch": 24.991395045632334,
-      "grad_norm": 1.6302272081375122,
-      "learning_rate": 5.2358793969849254e-05,
-      "loss": 5.0403,
-      "step": 47921
-    },
-    {
-      "epoch": 24.991916558018254,
-      "grad_norm": 1.6421219110488892,
-      "learning_rate": 5.235778894472362e-05,
-      "loss": 5.3084,
-      "step": 47922
-    },
-    {
-      "epoch": 24.992438070404173,
-      "grad_norm": 1.4586857557296753,
-      "learning_rate": 5.2356783919597996e-05,
-      "loss": 5.3428,
-      "step": 47923
-    },
-    {
-      "epoch": 24.992959582790093,
-      "grad_norm": 1.4806804656982422,
-      "learning_rate": 5.235577889447236e-05,
-      "loss": 5.5212,
-      "step": 47924
-    },
-    {
-      "epoch": 24.993481095176012,
-      "grad_norm": 1.6511995792388916,
-      "learning_rate": 5.235477386934674e-05,
-      "loss": 5.1964,
-      "step": 47925
-    },
-    {
-      "epoch": 24.994002607561928,
-      "grad_norm": 1.5205432176589966,
-      "learning_rate": 5.235376884422111e-05,
-      "loss": 5.3313,
-      "step": 47926
-    },
-    {
-      "epoch": 24.994524119947847,
-      "grad_norm": 1.5951286554336548,
-      "learning_rate": 5.2352763819095487e-05,
-      "loss": 5.2459,
-      "step": 47927
-    },
-    {
-      "epoch": 24.995045632333767,
-      "grad_norm": 1.6039103269577026,
-      "learning_rate": 5.235175879396985e-05,
-      "loss": 5.2013,
-      "step": 47928
-    },
-    {
-      "epoch": 24.995567144719686,
-      "grad_norm": 1.5081759691238403,
-      "learning_rate": 5.2350753768844215e-05,
-      "loss": 4.9872,
-      "step": 47929
-    },
-    {
-      "epoch": 24.996088657105606,
-      "grad_norm": 1.6342885494232178,
-      "learning_rate": 5.234974874371859e-05,
-      "loss": 4.9709,
-      "step": 47930
-    },
-    {
-      "epoch": 24.996610169491525,
-      "grad_norm": 1.6218138933181763,
-      "learning_rate": 5.2348743718592964e-05,
-      "loss": 5.3819,
-      "step": 47931
-    },
-    {
-      "epoch": 24.997131681877445,
-      "grad_norm": 1.5713235139846802,
-      "learning_rate": 5.234773869346734e-05,
-      "loss": 5.328,
-      "step": 47932
-    },
-    {
-      "epoch": 24.997653194263364,
-      "grad_norm": 1.6546761989593506,
-      "learning_rate": 5.2346733668341706e-05,
-      "loss": 5.2006,
-      "step": 47933
-    },
-    {
-      "epoch": 24.998174706649284,
-      "grad_norm": 1.5714281797409058,
-      "learning_rate": 5.2345728643216084e-05,
-      "loss": 5.452,
-      "step": 47934
-    },
-    {
-      "epoch": 24.998696219035203,
-      "grad_norm": 1.4970617294311523,
-      "learning_rate": 5.2344723618090455e-05,
-      "loss": 5.2018,
-      "step": 47935
-    },
-    {
-      "epoch": 24.999217731421123,
-      "grad_norm": 1.5740658044815063,
-      "learning_rate": 5.234371859296483e-05,
-      "loss": 5.5606,
-      "step": 47936
-    },
-    {
-      "epoch": 24.999739243807042,
-      "grad_norm": 1.5823925733566284,
-      "learning_rate": 5.23427135678392e-05,
-      "loss": 5.2419,
-      "step": 47937
-    },
-    {
-      "epoch": 25.000260756192958,
-      "grad_norm": 1.6449239253997803,
-      "learning_rate": 5.2341708542713575e-05,
-      "loss": 5.2363,
-      "step": 47938
-    },
-    {
-      "epoch": 25.000782268578877,
-      "grad_norm": 1.5797241926193237,
-      "learning_rate": 5.234070351758794e-05,
-      "loss": 5.0465,
-      "step": 47939
-    },
-    {
-      "epoch": 25.001303780964797,
-      "grad_norm": 1.5703134536743164,
-      "learning_rate": 5.233969849246232e-05,
-      "loss": 5.2689,
-      "step": 47940
-    },
-    {
-      "epoch": 25.001825293350716,
-      "grad_norm": 1.567196249961853,
-      "learning_rate": 5.233869346733669e-05,
-      "loss": 5.2132,
-      "step": 47941
-    },
-    {
-      "epoch": 25.002346805736636,
-      "grad_norm": 1.5825436115264893,
-      "learning_rate": 5.233768844221105e-05,
-      "loss": 5.0222,
-      "step": 47942
-    },
-    {
-      "epoch": 25.002868318122555,
-      "grad_norm": 1.5517722368240356,
-      "learning_rate": 5.233668341708543e-05,
-      "loss": 5.3922,
-      "step": 47943
-    },
-    {
-      "epoch": 25.003389830508475,
-      "grad_norm": 1.5709059238433838,
-      "learning_rate": 5.2335678391959794e-05,
-      "loss": 5.3029,
-      "step": 47944
-    },
-    {
-      "epoch": 25.003911342894394,
-      "grad_norm": 1.5589113235473633,
-      "learning_rate": 5.233467336683417e-05,
-      "loss": 5.2594,
-      "step": 47945
-    },
-    {
-      "epoch": 25.004432855280314,
-      "grad_norm": 1.5097631216049194,
-      "learning_rate": 5.233366834170854e-05,
-      "loss": 4.9391,
-      "step": 47946
-    },
-    {
-      "epoch": 25.004954367666233,
-      "grad_norm": 1.5377240180969238,
-      "learning_rate": 5.233266331658292e-05,
-      "loss": 5.5488,
-      "step": 47947
-    },
-    {
-      "epoch": 25.005475880052153,
-      "grad_norm": 1.5296876430511475,
-      "learning_rate": 5.2331658291457285e-05,
-      "loss": 5.0043,
-      "step": 47948
-    },
-    {
-      "epoch": 25.005997392438072,
-      "grad_norm": 1.6083464622497559,
-      "learning_rate": 5.233065326633166e-05,
-      "loss": 5.0694,
-      "step": 47949
-    },
-    {
-      "epoch": 25.006518904823988,
-      "grad_norm": 1.5442899465560913,
-      "learning_rate": 5.232964824120603e-05,
-      "loss": 5.4865,
-      "step": 47950
-    },
-    {
-      "epoch": 25.007040417209907,
-      "grad_norm": 1.6254340410232544,
-      "learning_rate": 5.2328643216080405e-05,
-      "loss": 5.4321,
-      "step": 47951
-    },
-    {
-      "epoch": 25.007561929595827,
-      "grad_norm": 1.6587506532669067,
-      "learning_rate": 5.2327638190954776e-05,
-      "loss": 5.0631,
-      "step": 47952
-    },
-    {
-      "epoch": 25.008083441981746,
-      "grad_norm": 1.5868816375732422,
-      "learning_rate": 5.2326633165829154e-05,
-      "loss": 5.3961,
-      "step": 47953
-    },
-    {
-      "epoch": 25.008604954367666,
-      "grad_norm": 1.4974640607833862,
-      "learning_rate": 5.232562814070352e-05,
-      "loss": 5.1597,
-      "step": 47954
-    },
-    {
-      "epoch": 25.009126466753585,
-      "grad_norm": 1.5757125616073608,
-      "learning_rate": 5.232462311557789e-05,
-      "loss": 5.1841,
-      "step": 47955
-    },
-    {
-      "epoch": 25.009647979139505,
-      "grad_norm": 1.4377646446228027,
-      "learning_rate": 5.232361809045227e-05,
-      "loss": 4.8552,
-      "step": 47956
-    },
-    {
-      "epoch": 25.010169491525424,
-      "grad_norm": 1.6160539388656616,
-      "learning_rate": 5.232261306532663e-05,
-      "loss": 5.178,
-      "step": 47957
-    },
-    {
-      "epoch": 25.010691003911344,
-      "grad_norm": 1.6632559299468994,
-      "learning_rate": 5.232160804020101e-05,
-      "loss": 5.0398,
-      "step": 47958
-    },
-    {
-      "epoch": 25.011212516297263,
-      "grad_norm": 1.5708813667297363,
-      "learning_rate": 5.2320603015075374e-05,
-      "loss": 5.4516,
-      "step": 47959
-    },
-    {
-      "epoch": 25.011734028683183,
-      "grad_norm": 1.4331462383270264,
-      "learning_rate": 5.231959798994975e-05,
-      "loss": 4.6523,
-      "step": 47960
-    },
-    {
-      "epoch": 25.012255541069102,
-      "grad_norm": 1.5590027570724487,
-      "learning_rate": 5.231859296482412e-05,
-      "loss": 4.9238,
-      "step": 47961
-    },
-    {
-      "epoch": 25.012777053455018,
-      "grad_norm": 1.491186261177063,
-      "learning_rate": 5.23175879396985e-05,
-      "loss": 5.4107,
-      "step": 47962
-    },
-    {
-      "epoch": 25.013298565840937,
-      "grad_norm": 1.4307039976119995,
-      "learning_rate": 5.2316582914572865e-05,
-      "loss": 5.2987,
-      "step": 47963
-    },
-    {
-      "epoch": 25.013820078226857,
-      "grad_norm": 1.5206658840179443,
-      "learning_rate": 5.231557788944724e-05,
-      "loss": 4.6962,
-      "step": 47964
-    },
-    {
-      "epoch": 25.014341590612776,
-      "grad_norm": 1.5465837717056274,
-      "learning_rate": 5.231457286432161e-05,
-      "loss": 5.3571,
-      "step": 47965
-    },
-    {
-      "epoch": 25.014863102998696,
-      "grad_norm": 1.5505754947662354,
-      "learning_rate": 5.2313567839195985e-05,
-      "loss": 5.3454,
-      "step": 47966
-    },
-    {
-      "epoch": 25.015384615384615,
-      "grad_norm": 1.50552237033844,
-      "learning_rate": 5.2312562814070356e-05,
-      "loss": 5.5509,
-      "step": 47967
-    },
-    {
-      "epoch": 25.015906127770535,
-      "grad_norm": 1.5563350915908813,
-      "learning_rate": 5.231155778894472e-05,
-      "loss": 5.2532,
-      "step": 47968
-    },
-    {
-      "epoch": 25.016427640156454,
-      "grad_norm": 1.5838203430175781,
-      "learning_rate": 5.23105527638191e-05,
-      "loss": 4.7039,
-      "step": 47969
-    },
-    {
-      "epoch": 25.016949152542374,
-      "grad_norm": 1.5300307273864746,
-      "learning_rate": 5.230954773869346e-05,
-      "loss": 5.2034,
-      "step": 47970
-    },
-    {
-      "epoch": 25.017470664928293,
-      "grad_norm": 1.5971177816390991,
-      "learning_rate": 5.230854271356784e-05,
-      "loss": 5.2748,
-      "step": 47971
-    },
-    {
-      "epoch": 25.017992177314213,
-      "grad_norm": 1.8589894771575928,
-      "learning_rate": 5.230753768844221e-05,
-      "loss": 4.8425,
-      "step": 47972
-    },
-    {
-      "epoch": 25.018513689700132,
-      "grad_norm": 1.5234215259552002,
-      "learning_rate": 5.230653266331659e-05,
-      "loss": 4.8835,
-      "step": 47973
-    },
-    {
-      "epoch": 25.019035202086048,
-      "grad_norm": 1.5195035934448242,
-      "learning_rate": 5.230552763819095e-05,
-      "loss": 5.4634,
-      "step": 47974
-    },
-    {
-      "epoch": 25.019556714471967,
-      "grad_norm": 1.7104145288467407,
-      "learning_rate": 5.230452261306533e-05,
-      "loss": 4.8629,
-      "step": 47975
-    },
-    {
-      "epoch": 25.020078226857887,
-      "grad_norm": 1.6063058376312256,
-      "learning_rate": 5.23035175879397e-05,
-      "loss": 5.003,
-      "step": 47976
-    },
-    {
-      "epoch": 25.020599739243806,
-      "grad_norm": 1.5604350566864014,
-      "learning_rate": 5.230251256281408e-05,
-      "loss": 4.911,
-      "step": 47977
-    },
-    {
-      "epoch": 25.021121251629726,
-      "grad_norm": 1.4927552938461304,
-      "learning_rate": 5.2301507537688444e-05,
-      "loss": 4.872,
-      "step": 47978
-    },
-    {
-      "epoch": 25.021642764015645,
-      "grad_norm": 1.5293821096420288,
-      "learning_rate": 5.230050251256282e-05,
-      "loss": 4.1729,
-      "step": 47979
-    },
-    {
-      "epoch": 25.022164276401565,
-      "grad_norm": 1.5910650491714478,
-      "learning_rate": 5.2299497487437186e-05,
-      "loss": 5.6085,
-      "step": 47980
-    },
-    {
-      "epoch": 25.022685788787484,
-      "grad_norm": 1.5396956205368042,
-      "learning_rate": 5.2298492462311564e-05,
-      "loss": 5.2807,
-      "step": 47981
-    },
-    {
-      "epoch": 25.023207301173404,
-      "grad_norm": 1.6444637775421143,
-      "learning_rate": 5.2297487437185935e-05,
-      "loss": 4.7798,
-      "step": 47982
-    },
-    {
-      "epoch": 25.023728813559323,
-      "grad_norm": 1.522187352180481,
-      "learning_rate": 5.22964824120603e-05,
-      "loss": 5.1165,
-      "step": 47983
-    },
-    {
-      "epoch": 25.024250325945243,
-      "grad_norm": 1.5728025436401367,
-      "learning_rate": 5.229547738693468e-05,
-      "loss": 4.8963,
-      "step": 47984
-    },
-    {
-      "epoch": 25.02477183833116,
-      "grad_norm": 1.5792328119277954,
-      "learning_rate": 5.229447236180904e-05,
-      "loss": 5.4274,
-      "step": 47985
-    },
-    {
-      "epoch": 25.025293350717078,
-      "grad_norm": 1.446913242340088,
-      "learning_rate": 5.229346733668342e-05,
-      "loss": 5.3593,
-      "step": 47986
-    },
-    {
-      "epoch": 25.025814863102998,
-      "grad_norm": 1.5294749736785889,
-      "learning_rate": 5.229246231155779e-05,
-      "loss": 5.4693,
-      "step": 47987
-    },
-    {
-      "epoch": 25.026336375488917,
-      "grad_norm": 1.6596821546554565,
-      "learning_rate": 5.229145728643217e-05,
-      "loss": 4.7274,
-      "step": 47988
-    },
-    {
-      "epoch": 25.026857887874836,
-      "grad_norm": 1.578134536743164,
-      "learning_rate": 5.229045226130653e-05,
-      "loss": 5.6329,
-      "step": 47989
-    },
-    {
-      "epoch": 25.027379400260756,
-      "grad_norm": 1.6660449504852295,
-      "learning_rate": 5.228944723618091e-05,
-      "loss": 4.8747,
-      "step": 47990
-    },
-    {
-      "epoch": 25.027900912646675,
-      "grad_norm": 1.5761691331863403,
-      "learning_rate": 5.2288442211055275e-05,
-      "loss": 5.011,
-      "step": 47991
-    },
-    {
-      "epoch": 25.028422425032595,
-      "grad_norm": 1.6081013679504395,
-      "learning_rate": 5.228743718592965e-05,
-      "loss": 5.2206,
-      "step": 47992
-    },
-    {
-      "epoch": 25.028943937418514,
-      "grad_norm": 1.6231704950332642,
-      "learning_rate": 5.2286432160804023e-05,
-      "loss": 5.2324,
-      "step": 47993
-    },
-    {
-      "epoch": 25.029465449804434,
-      "grad_norm": 1.5783947706222534,
-      "learning_rate": 5.22854271356784e-05,
-      "loss": 4.8951,
-      "step": 47994
-    },
-    {
-      "epoch": 25.029986962190353,
-      "grad_norm": 1.5610227584838867,
-      "learning_rate": 5.2284422110552766e-05,
-      "loss": 5.3224,
-      "step": 47995
-    },
-    {
-      "epoch": 25.030508474576273,
-      "grad_norm": 1.556490421295166,
-      "learning_rate": 5.2283417085427137e-05,
-      "loss": 5.1734,
-      "step": 47996
-    },
-    {
-      "epoch": 25.03102998696219,
-      "grad_norm": 1.512485384941101,
-      "learning_rate": 5.2282412060301514e-05,
-      "loss": 4.9708,
-      "step": 47997
-    },
-    {
-      "epoch": 25.031551499348108,
-      "grad_norm": 1.656604290008545,
-      "learning_rate": 5.228140703517588e-05,
-      "loss": 4.9955,
-      "step": 47998
-    },
-    {
-      "epoch": 25.032073011734028,
-      "grad_norm": 1.631657361984253,
-      "learning_rate": 5.2280402010050257e-05,
-      "loss": 5.3694,
-      "step": 47999
-    },
-    {
-      "epoch": 25.032594524119947,
-      "grad_norm": 1.5367200374603271,
-      "learning_rate": 5.227939698492462e-05,
-      "loss": 4.9226,
-      "step": 48000
-    },
-    {
-      "epoch": 25.033116036505866,
-      "grad_norm": 1.576295018196106,
-      "learning_rate": 5.2278391959799e-05,
-      "loss": 5.4135,
-      "step": 48001
-    },
-    {
-      "epoch": 25.033637548891786,
-      "grad_norm": 1.7286040782928467,
-      "learning_rate": 5.227738693467337e-05,
-      "loss": 4.6489,
-      "step": 48002
-    },
-    {
-      "epoch": 25.034159061277705,
-      "grad_norm": 1.515144944190979,
-      "learning_rate": 5.227638190954775e-05,
-      "loss": 5.5637,
-      "step": 48003
-    },
-    {
-      "epoch": 25.034680573663625,
-      "grad_norm": 1.460913062095642,
-      "learning_rate": 5.227537688442211e-05,
-      "loss": 5.2815,
-      "step": 48004
-    },
-    {
-      "epoch": 25.035202086049544,
-      "grad_norm": 1.5246466398239136,
-      "learning_rate": 5.227437185929649e-05,
-      "loss": 5.2387,
-      "step": 48005
-    },
-    {
-      "epoch": 25.035723598435464,
-      "grad_norm": 1.6306599378585815,
-      "learning_rate": 5.2273366834170854e-05,
-      "loss": 4.8912,
-      "step": 48006
-    },
-    {
-      "epoch": 25.036245110821383,
-      "grad_norm": 1.699123740196228,
-      "learning_rate": 5.227236180904523e-05,
-      "loss": 4.6782,
-      "step": 48007
-    },
-    {
-      "epoch": 25.036766623207303,
-      "grad_norm": 1.6518476009368896,
-      "learning_rate": 5.22713567839196e-05,
-      "loss": 4.9387,
-      "step": 48008
-    },
-    {
-      "epoch": 25.03728813559322,
-      "grad_norm": 1.6386560201644897,
-      "learning_rate": 5.227035175879397e-05,
-      "loss": 5.2649,
-      "step": 48009
-    },
-    {
-      "epoch": 25.037809647979138,
-      "grad_norm": 1.6140928268432617,
-      "learning_rate": 5.2269346733668345e-05,
-      "loss": 5.479,
-      "step": 48010
-    },
-    {
-      "epoch": 25.038331160365058,
-      "grad_norm": 1.5281693935394287,
-      "learning_rate": 5.226834170854271e-05,
-      "loss": 5.3758,
-      "step": 48011
-    },
-    {
-      "epoch": 25.038852672750977,
-      "grad_norm": 1.594447135925293,
-      "learning_rate": 5.226733668341709e-05,
-      "loss": 4.5486,
-      "step": 48012
-    },
-    {
-      "epoch": 25.039374185136897,
-      "grad_norm": 1.635145664215088,
-      "learning_rate": 5.226633165829146e-05,
-      "loss": 4.9519,
-      "step": 48013
-    },
-    {
-      "epoch": 25.039895697522816,
-      "grad_norm": 1.6227941513061523,
-      "learning_rate": 5.2265326633165836e-05,
-      "loss": 5.1203,
-      "step": 48014
-    },
-    {
-      "epoch": 25.040417209908735,
-      "grad_norm": 1.4908031225204468,
-      "learning_rate": 5.22643216080402e-05,
-      "loss": 5.2972,
-      "step": 48015
-    },
-    {
-      "epoch": 25.040938722294655,
-      "grad_norm": 1.6482861042022705,
-      "learning_rate": 5.226331658291458e-05,
-      "loss": 5.3533,
-      "step": 48016
-    },
-    {
-      "epoch": 25.041460234680574,
-      "grad_norm": 1.5024847984313965,
-      "learning_rate": 5.226231155778894e-05,
-      "loss": 5.3597,
-      "step": 48017
-    },
-    {
-      "epoch": 25.041981747066494,
-      "grad_norm": 1.4999306201934814,
-      "learning_rate": 5.226130653266332e-05,
-      "loss": 5.2932,
-      "step": 48018
-    },
-    {
-      "epoch": 25.042503259452413,
-      "grad_norm": 1.5252189636230469,
-      "learning_rate": 5.226030150753769e-05,
-      "loss": 5.3634,
-      "step": 48019
-    },
-    {
-      "epoch": 25.043024771838333,
-      "grad_norm": 1.650327205657959,
-      "learning_rate": 5.225929648241207e-05,
-      "loss": 5.2756,
-      "step": 48020
-    },
-    {
-      "epoch": 25.04354628422425,
-      "grad_norm": 1.5264599323272705,
-      "learning_rate": 5.225829145728643e-05,
-      "loss": 5.585,
-      "step": 48021
-    },
-    {
-      "epoch": 25.044067796610168,
-      "grad_norm": 1.5873053073883057,
-      "learning_rate": 5.2257286432160804e-05,
-      "loss": 5.0785,
-      "step": 48022
-    },
-    {
-      "epoch": 25.044589308996088,
-      "grad_norm": 1.6064082384109497,
-      "learning_rate": 5.225628140703518e-05,
-      "loss": 4.4133,
-      "step": 48023
-    },
-    {
-      "epoch": 25.045110821382007,
-      "grad_norm": 1.493614912033081,
-      "learning_rate": 5.2255276381909546e-05,
-      "loss": 5.232,
-      "step": 48024
-    },
-    {
-      "epoch": 25.045632333767927,
-      "grad_norm": 1.5414810180664062,
-      "learning_rate": 5.2254271356783924e-05,
-      "loss": 5.4797,
-      "step": 48025
-    },
-    {
-      "epoch": 25.046153846153846,
-      "grad_norm": 1.613867163658142,
-      "learning_rate": 5.225326633165829e-05,
-      "loss": 4.8193,
-      "step": 48026
-    },
-    {
-      "epoch": 25.046675358539765,
-      "grad_norm": 1.5164532661437988,
-      "learning_rate": 5.2252261306532666e-05,
-      "loss": 5.6607,
-      "step": 48027
-    },
-    {
-      "epoch": 25.047196870925685,
-      "grad_norm": 1.5713435411453247,
-      "learning_rate": 5.225125628140704e-05,
-      "loss": 4.9202,
-      "step": 48028
-    },
-    {
-      "epoch": 25.047718383311604,
-      "grad_norm": 1.6441740989685059,
-      "learning_rate": 5.2250251256281415e-05,
-      "loss": 5.5079,
-      "step": 48029
-    },
-    {
-      "epoch": 25.048239895697524,
-      "grad_norm": 1.7035290002822876,
-      "learning_rate": 5.224924623115578e-05,
-      "loss": 4.906,
-      "step": 48030
-    },
-    {
-      "epoch": 25.048761408083443,
-      "grad_norm": 2.085284471511841,
-      "learning_rate": 5.224824120603016e-05,
-      "loss": 5.0533,
-      "step": 48031
-    },
-    {
-      "epoch": 25.049282920469363,
-      "grad_norm": 1.619373083114624,
-      "learning_rate": 5.224723618090452e-05,
-      "loss": 5.522,
-      "step": 48032
-    },
-    {
-      "epoch": 25.04980443285528,
-      "grad_norm": 1.6298762559890747,
-      "learning_rate": 5.22462311557789e-05,
-      "loss": 5.5061,
-      "step": 48033
-    },
-    {
-      "epoch": 25.050325945241198,
-      "grad_norm": 1.6265641450881958,
-      "learning_rate": 5.224522613065327e-05,
-      "loss": 5.2259,
-      "step": 48034
-    },
-    {
-      "epoch": 25.050847457627118,
-      "grad_norm": 1.6626091003417969,
-      "learning_rate": 5.2244221105527635e-05,
-      "loss": 4.6286,
-      "step": 48035
-    },
-    {
-      "epoch": 25.051368970013037,
-      "grad_norm": 1.5121263265609741,
-      "learning_rate": 5.224321608040201e-05,
-      "loss": 4.7779,
-      "step": 48036
-    },
-    {
-      "epoch": 25.051890482398957,
-      "grad_norm": 1.5903284549713135,
-      "learning_rate": 5.224221105527638e-05,
-      "loss": 4.7092,
-      "step": 48037
-    },
-    {
-      "epoch": 25.052411994784876,
-      "grad_norm": 1.613397240638733,
-      "learning_rate": 5.2241206030150755e-05,
-      "loss": 5.0417,
-      "step": 48038
-    },
-    {
-      "epoch": 25.052933507170795,
-      "grad_norm": 1.5699125528335571,
-      "learning_rate": 5.2240201005025126e-05,
-      "loss": 5.5418,
-      "step": 48039
-    },
-    {
-      "epoch": 25.053455019556715,
-      "grad_norm": 1.6471492052078247,
-      "learning_rate": 5.2239195979899504e-05,
-      "loss": 5.1907,
-      "step": 48040
-    },
-    {
-      "epoch": 25.053976531942634,
-      "grad_norm": 1.6034982204437256,
-      "learning_rate": 5.223819095477387e-05,
-      "loss": 5.0717,
-      "step": 48041
-    },
-    {
-      "epoch": 25.054498044328554,
-      "grad_norm": 1.6490715742111206,
-      "learning_rate": 5.2237185929648246e-05,
-      "loss": 5.4298,
-      "step": 48042
-    },
-    {
-      "epoch": 25.055019556714473,
-      "grad_norm": 1.5352113246917725,
-      "learning_rate": 5.223618090452262e-05,
-      "loss": 5.1596,
-      "step": 48043
-    },
-    {
-      "epoch": 25.055541069100393,
-      "grad_norm": 1.549985647201538,
-      "learning_rate": 5.2235175879396995e-05,
-      "loss": 5.3552,
-      "step": 48044
-    },
-    {
-      "epoch": 25.05606258148631,
-      "grad_norm": 1.4620240926742554,
-      "learning_rate": 5.223417085427136e-05,
-      "loss": 5.5961,
-      "step": 48045
-    },
-    {
-      "epoch": 25.056584093872228,
-      "grad_norm": 1.5992226600646973,
-      "learning_rate": 5.223316582914574e-05,
-      "loss": 5.5577,
-      "step": 48046
-    },
-    {
-      "epoch": 25.057105606258148,
-      "grad_norm": 1.5935360193252563,
-      "learning_rate": 5.22321608040201e-05,
-      "loss": 5.3526,
-      "step": 48047
-    },
-    {
-      "epoch": 25.057627118644067,
-      "grad_norm": 1.5498789548873901,
-      "learning_rate": 5.223115577889447e-05,
-      "loss": 5.0973,
-      "step": 48048
-    },
-    {
-      "epoch": 25.058148631029987,
-      "grad_norm": 1.6259602308273315,
-      "learning_rate": 5.223015075376885e-05,
-      "loss": 5.0978,
-      "step": 48049
-    },
-    {
-      "epoch": 25.058670143415906,
-      "grad_norm": 1.574906349182129,
-      "learning_rate": 5.2229145728643214e-05,
-      "loss": 5.318,
-      "step": 48050
-    },
-    {
-      "epoch": 25.059191655801826,
-      "grad_norm": 1.5509297847747803,
-      "learning_rate": 5.222814070351759e-05,
-      "loss": 4.9509,
-      "step": 48051
-    },
-    {
-      "epoch": 25.059713168187745,
-      "grad_norm": 1.718526840209961,
-      "learning_rate": 5.2227135678391956e-05,
-      "loss": 5.4511,
-      "step": 48052
-    },
-    {
-      "epoch": 25.060234680573664,
-      "grad_norm": 1.6014487743377686,
-      "learning_rate": 5.2226130653266334e-05,
-      "loss": 5.4742,
-      "step": 48053
-    },
-    {
-      "epoch": 25.060756192959584,
-      "grad_norm": 1.7435474395751953,
-      "learning_rate": 5.2225125628140705e-05,
-      "loss": 5.337,
-      "step": 48054
-    },
-    {
-      "epoch": 25.061277705345503,
-      "grad_norm": 1.615799069404602,
-      "learning_rate": 5.222412060301508e-05,
-      "loss": 5.3563,
-      "step": 48055
-    },
-    {
-      "epoch": 25.061799217731423,
-      "grad_norm": 1.6639987230300903,
-      "learning_rate": 5.222311557788945e-05,
-      "loss": 4.7516,
-      "step": 48056
-    },
-    {
-      "epoch": 25.06232073011734,
-      "grad_norm": 1.520599365234375,
-      "learning_rate": 5.2222110552763825e-05,
-      "loss": 5.2604,
-      "step": 48057
-    },
-    {
-      "epoch": 25.062842242503258,
-      "grad_norm": 1.6133095026016235,
-      "learning_rate": 5.222110552763819e-05,
-      "loss": 4.711,
-      "step": 48058
-    },
-    {
-      "epoch": 25.063363754889178,
-      "grad_norm": 1.5816247463226318,
-      "learning_rate": 5.222010050251257e-05,
-      "loss": 5.128,
-      "step": 48059
-    },
-    {
-      "epoch": 25.063885267275097,
-      "grad_norm": 1.5816177129745483,
-      "learning_rate": 5.221909547738694e-05,
-      "loss": 5.1499,
-      "step": 48060
-    },
-    {
-      "epoch": 25.064406779661017,
-      "grad_norm": 1.5256105661392212,
-      "learning_rate": 5.2218090452261316e-05,
-      "loss": 5.1567,
-      "step": 48061
-    },
-    {
-      "epoch": 25.064928292046936,
-      "grad_norm": 1.6246858835220337,
-      "learning_rate": 5.221708542713568e-05,
-      "loss": 4.9064,
-      "step": 48062
-    },
-    {
-      "epoch": 25.065449804432856,
-      "grad_norm": 1.4724870920181274,
-      "learning_rate": 5.221608040201005e-05,
-      "loss": 5.6264,
-      "step": 48063
-    },
-    {
-      "epoch": 25.065971316818775,
-      "grad_norm": 1.6764525175094604,
-      "learning_rate": 5.221507537688443e-05,
-      "loss": 5.1681,
-      "step": 48064
-    },
-    {
-      "epoch": 25.066492829204694,
-      "grad_norm": 1.6687190532684326,
-      "learning_rate": 5.221407035175879e-05,
-      "loss": 4.8345,
-      "step": 48065
-    },
-    {
-      "epoch": 25.067014341590614,
-      "grad_norm": 1.5673115253448486,
-      "learning_rate": 5.221306532663317e-05,
-      "loss": 5.4683,
-      "step": 48066
-    },
-    {
-      "epoch": 25.067535853976533,
-      "grad_norm": 1.6957436800003052,
-      "learning_rate": 5.2212060301507535e-05,
-      "loss": 5.0998,
-      "step": 48067
-    },
-    {
-      "epoch": 25.068057366362453,
-      "grad_norm": 1.5277578830718994,
-      "learning_rate": 5.221105527638191e-05,
-      "loss": 5.4145,
-      "step": 48068
-    },
-    {
-      "epoch": 25.06857887874837,
-      "grad_norm": 1.6020982265472412,
-      "learning_rate": 5.2210050251256284e-05,
-      "loss": 5.3476,
-      "step": 48069
-    },
-    {
-      "epoch": 25.06910039113429,
-      "grad_norm": 1.6135144233703613,
-      "learning_rate": 5.220904522613066e-05,
-      "loss": 5.521,
-      "step": 48070
-    },
-    {
-      "epoch": 25.069621903520208,
-      "grad_norm": 1.6405580043792725,
-      "learning_rate": 5.2208040201005026e-05,
-      "loss": 5.0008,
-      "step": 48071
-    },
-    {
-      "epoch": 25.070143415906127,
-      "grad_norm": 1.5178943872451782,
-      "learning_rate": 5.2207035175879404e-05,
-      "loss": 5.1607,
-      "step": 48072
-    },
-    {
-      "epoch": 25.070664928292047,
-      "grad_norm": 1.576015591621399,
-      "learning_rate": 5.220603015075377e-05,
-      "loss": 4.7903,
-      "step": 48073
-    },
-    {
-      "epoch": 25.071186440677966,
-      "grad_norm": 1.5297528505325317,
-      "learning_rate": 5.2205025125628146e-05,
-      "loss": 5.6894,
-      "step": 48074
-    },
-    {
-      "epoch": 25.071707953063886,
-      "grad_norm": 1.5205433368682861,
-      "learning_rate": 5.220402010050252e-05,
-      "loss": 5.253,
-      "step": 48075
-    },
-    {
-      "epoch": 25.072229465449805,
-      "grad_norm": 1.5397348403930664,
-      "learning_rate": 5.220301507537688e-05,
-      "loss": 5.1561,
-      "step": 48076
-    },
-    {
-      "epoch": 25.072750977835724,
-      "grad_norm": 1.538046956062317,
-      "learning_rate": 5.220201005025126e-05,
-      "loss": 5.0638,
-      "step": 48077
-    },
-    {
-      "epoch": 25.073272490221644,
-      "grad_norm": 1.4839677810668945,
-      "learning_rate": 5.2201005025125624e-05,
-      "loss": 5.3218,
-      "step": 48078
-    },
-    {
-      "epoch": 25.073794002607563,
-      "grad_norm": 1.6622732877731323,
-      "learning_rate": 5.22e-05,
-      "loss": 5.5276,
-      "step": 48079
-    },
-    {
-      "epoch": 25.07431551499348,
-      "grad_norm": 1.6353181600570679,
-      "learning_rate": 5.219899497487437e-05,
-      "loss": 4.9584,
-      "step": 48080
-    },
-    {
-      "epoch": 25.0748370273794,
-      "grad_norm": 1.5373274087905884,
-      "learning_rate": 5.219798994974875e-05,
-      "loss": 5.3617,
-      "step": 48081
-    },
-    {
-      "epoch": 25.07535853976532,
-      "grad_norm": 1.6091219186782837,
-      "learning_rate": 5.2196984924623115e-05,
-      "loss": 5.3374,
-      "step": 48082
-    },
-    {
-      "epoch": 25.075880052151238,
-      "grad_norm": 1.6382367610931396,
-      "learning_rate": 5.219597989949749e-05,
-      "loss": 5.423,
-      "step": 48083
-    },
-    {
-      "epoch": 25.076401564537157,
-      "grad_norm": 1.4987819194793701,
-      "learning_rate": 5.219497487437186e-05,
-      "loss": 5.4526,
-      "step": 48084
-    },
-    {
-      "epoch": 25.076923076923077,
-      "grad_norm": 1.533179759979248,
-      "learning_rate": 5.2193969849246235e-05,
-      "loss": 5.0797,
-      "step": 48085
-    },
-    {
-      "epoch": 25.077444589308996,
-      "grad_norm": 1.604422688484192,
-      "learning_rate": 5.2192964824120606e-05,
-      "loss": 5.56,
-      "step": 48086
-    },
-    {
-      "epoch": 25.077966101694916,
-      "grad_norm": 1.6459507942199707,
-      "learning_rate": 5.2191959798994984e-05,
-      "loss": 5.1896,
-      "step": 48087
-    },
-    {
-      "epoch": 25.078487614080835,
-      "grad_norm": 1.5721250772476196,
-      "learning_rate": 5.219095477386935e-05,
-      "loss": 5.3563,
-      "step": 48088
-    },
-    {
-      "epoch": 25.079009126466755,
-      "grad_norm": 1.4425777196884155,
-      "learning_rate": 5.218994974874372e-05,
-      "loss": 5.6419,
-      "step": 48089
-    },
-    {
-      "epoch": 25.079530638852674,
-      "grad_norm": 1.5747184753417969,
-      "learning_rate": 5.21889447236181e-05,
-      "loss": 5.0476,
-      "step": 48090
-    },
-    {
-      "epoch": 25.080052151238593,
-      "grad_norm": 1.6852716207504272,
-      "learning_rate": 5.218793969849246e-05,
-      "loss": 4.1987,
-      "step": 48091
-    },
-    {
-      "epoch": 25.08057366362451,
-      "grad_norm": 1.590915322303772,
-      "learning_rate": 5.218693467336684e-05,
-      "loss": 5.1722,
-      "step": 48092
-    },
-    {
-      "epoch": 25.08109517601043,
-      "grad_norm": 1.5987131595611572,
-      "learning_rate": 5.21859296482412e-05,
-      "loss": 4.9057,
-      "step": 48093
-    },
-    {
-      "epoch": 25.08161668839635,
-      "grad_norm": 1.5028268098831177,
-      "learning_rate": 5.218492462311558e-05,
-      "loss": 5.0129,
-      "step": 48094
-    },
-    {
-      "epoch": 25.082138200782268,
-      "grad_norm": 1.4478603601455688,
-      "learning_rate": 5.218391959798995e-05,
-      "loss": 5.5567,
-      "step": 48095
-    },
-    {
-      "epoch": 25.082659713168187,
-      "grad_norm": 1.6364659070968628,
-      "learning_rate": 5.218291457286433e-05,
-      "loss": 3.9887,
-      "step": 48096
-    },
-    {
-      "epoch": 25.083181225554107,
-      "grad_norm": 1.6034468412399292,
-      "learning_rate": 5.2181909547738694e-05,
-      "loss": 5.2996,
-      "step": 48097
-    },
-    {
-      "epoch": 25.083702737940026,
-      "grad_norm": 1.4357606172561646,
-      "learning_rate": 5.218090452261307e-05,
-      "loss": 4.8466,
-      "step": 48098
-    },
-    {
-      "epoch": 25.084224250325946,
-      "grad_norm": 1.6267223358154297,
-      "learning_rate": 5.2179899497487436e-05,
-      "loss": 4.9795,
-      "step": 48099
-    },
-    {
-      "epoch": 25.084745762711865,
-      "grad_norm": 1.4889978170394897,
-      "learning_rate": 5.2178894472361814e-05,
-      "loss": 5.2792,
-      "step": 48100
-    },
-    {
-      "epoch": 25.085267275097785,
-      "grad_norm": 1.4590140581130981,
-      "learning_rate": 5.2177889447236185e-05,
-      "loss": 5.2074,
-      "step": 48101
-    },
-    {
-      "epoch": 25.085788787483704,
-      "grad_norm": 1.5663119554519653,
-      "learning_rate": 5.217688442211055e-05,
-      "loss": 5.3289,
-      "step": 48102
-    },
-    {
-      "epoch": 25.086310299869623,
-      "grad_norm": 1.6344408988952637,
-      "learning_rate": 5.217587939698493e-05,
-      "loss": 5.0389,
-      "step": 48103
-    },
-    {
-      "epoch": 25.08683181225554,
-      "grad_norm": 1.743761420249939,
-      "learning_rate": 5.217487437185929e-05,
-      "loss": 5.2697,
-      "step": 48104
-    },
-    {
-      "epoch": 25.08735332464146,
-      "grad_norm": 1.6246193647384644,
-      "learning_rate": 5.217386934673367e-05,
-      "loss": 4.9479,
-      "step": 48105
-    },
-    {
-      "epoch": 25.08787483702738,
-      "grad_norm": 1.5213711261749268,
-      "learning_rate": 5.217286432160804e-05,
-      "loss": 5.5987,
-      "step": 48106
-    },
-    {
-      "epoch": 25.088396349413298,
-      "grad_norm": 1.6578574180603027,
-      "learning_rate": 5.217185929648242e-05,
-      "loss": 5.1465,
-      "step": 48107
-    },
-    {
-      "epoch": 25.088917861799217,
-      "grad_norm": 1.609294056892395,
-      "learning_rate": 5.217085427135678e-05,
-      "loss": 5.3002,
-      "step": 48108
-    },
-    {
-      "epoch": 25.089439374185137,
-      "grad_norm": 1.5202021598815918,
-      "learning_rate": 5.216984924623116e-05,
-      "loss": 5.4344,
-      "step": 48109
-    },
-    {
-      "epoch": 25.089960886571056,
-      "grad_norm": 1.6063722372055054,
-      "learning_rate": 5.216884422110553e-05,
-      "loss": 5.2788,
-      "step": 48110
-    },
-    {
-      "epoch": 25.090482398956976,
-      "grad_norm": 1.7403017282485962,
-      "learning_rate": 5.216783919597991e-05,
-      "loss": 4.7151,
-      "step": 48111
-    },
-    {
-      "epoch": 25.091003911342895,
-      "grad_norm": 1.585898518562317,
-      "learning_rate": 5.2166834170854273e-05,
-      "loss": 5.3997,
-      "step": 48112
-    },
-    {
-      "epoch": 25.091525423728815,
-      "grad_norm": 1.6215565204620361,
-      "learning_rate": 5.216582914572865e-05,
-      "loss": 5.1765,
-      "step": 48113
-    },
-    {
-      "epoch": 25.092046936114734,
-      "grad_norm": 1.5886249542236328,
-      "learning_rate": 5.2164824120603016e-05,
-      "loss": 5.1362,
-      "step": 48114
-    },
-    {
-      "epoch": 25.092568448500653,
-      "grad_norm": 1.6035633087158203,
-      "learning_rate": 5.216381909547739e-05,
-      "loss": 5.3149,
-      "step": 48115
-    },
-    {
-      "epoch": 25.09308996088657,
-      "grad_norm": 1.7099478244781494,
-      "learning_rate": 5.2162814070351764e-05,
-      "loss": 5.0732,
-      "step": 48116
-    },
-    {
-      "epoch": 25.09361147327249,
-      "grad_norm": 1.5410817861557007,
-      "learning_rate": 5.216180904522613e-05,
-      "loss": 5.4971,
-      "step": 48117
-    },
-    {
-      "epoch": 25.09413298565841,
-      "grad_norm": 1.6006975173950195,
-      "learning_rate": 5.2160804020100507e-05,
-      "loss": 5.043,
-      "step": 48118
-    },
-    {
-      "epoch": 25.094654498044328,
-      "grad_norm": 1.6049522161483765,
-      "learning_rate": 5.215979899497487e-05,
-      "loss": 5.2449,
-      "step": 48119
-    },
-    {
-      "epoch": 25.095176010430247,
-      "grad_norm": 1.5726654529571533,
-      "learning_rate": 5.215879396984925e-05,
-      "loss": 5.4627,
-      "step": 48120
-    },
-    {
-      "epoch": 25.095697522816167,
-      "grad_norm": 1.6766661405563354,
-      "learning_rate": 5.215778894472362e-05,
-      "loss": 5.0462,
-      "step": 48121
-    },
-    {
-      "epoch": 25.096219035202086,
-      "grad_norm": 1.5718541145324707,
-      "learning_rate": 5.2156783919598e-05,
-      "loss": 5.0668,
-      "step": 48122
-    },
-    {
-      "epoch": 25.096740547588006,
-      "grad_norm": 1.5774974822998047,
-      "learning_rate": 5.215577889447236e-05,
-      "loss": 4.9374,
-      "step": 48123
-    },
-    {
-      "epoch": 25.097262059973925,
-      "grad_norm": 1.630448818206787,
-      "learning_rate": 5.215477386934674e-05,
-      "loss": 4.718,
-      "step": 48124
-    },
-    {
-      "epoch": 25.097783572359845,
-      "grad_norm": 1.6294201612472534,
-      "learning_rate": 5.2153768844221104e-05,
-      "loss": 4.81,
-      "step": 48125
-    },
-    {
-      "epoch": 25.098305084745764,
-      "grad_norm": 1.5562546253204346,
-      "learning_rate": 5.215276381909548e-05,
-      "loss": 4.9449,
-      "step": 48126
-    },
-    {
-      "epoch": 25.098826597131684,
-      "grad_norm": 1.608944058418274,
-      "learning_rate": 5.215175879396985e-05,
-      "loss": 5.3152,
-      "step": 48127
-    },
-    {
-      "epoch": 25.0993481095176,
-      "grad_norm": 1.5664875507354736,
-      "learning_rate": 5.215075376884422e-05,
-      "loss": 5.4298,
-      "step": 48128
-    },
-    {
-      "epoch": 25.09986962190352,
-      "grad_norm": 1.5402615070343018,
-      "learning_rate": 5.2149748743718595e-05,
-      "loss": 5.6319,
-      "step": 48129
-    },
-    {
-      "epoch": 25.10039113428944,
-      "grad_norm": 1.552189826965332,
-      "learning_rate": 5.2148743718592966e-05,
-      "loss": 5.4691,
-      "step": 48130
-    },
-    {
-      "epoch": 25.100912646675358,
-      "grad_norm": 1.6753851175308228,
-      "learning_rate": 5.2147738693467344e-05,
-      "loss": 5.0115,
-      "step": 48131
-    },
-    {
-      "epoch": 25.101434159061277,
-      "grad_norm": 1.532462239265442,
-      "learning_rate": 5.214673366834171e-05,
-      "loss": 5.4271,
-      "step": 48132
-    },
-    {
-      "epoch": 25.101955671447197,
-      "grad_norm": 1.635480284690857,
-      "learning_rate": 5.2145728643216086e-05,
-      "loss": 5.0256,
-      "step": 48133
-    },
-    {
-      "epoch": 25.102477183833116,
-      "grad_norm": 1.5448393821716309,
-      "learning_rate": 5.214472361809045e-05,
-      "loss": 5.1738,
-      "step": 48134
-    },
-    {
-      "epoch": 25.102998696219036,
-      "grad_norm": 1.5636242628097534,
-      "learning_rate": 5.214371859296483e-05,
-      "loss": 5.381,
-      "step": 48135
-    },
-    {
-      "epoch": 25.103520208604955,
-      "grad_norm": 1.501422643661499,
-      "learning_rate": 5.21427135678392e-05,
-      "loss": 3.8809,
-      "step": 48136
-    },
-    {
-      "epoch": 25.104041720990875,
-      "grad_norm": 1.5404072999954224,
-      "learning_rate": 5.214170854271358e-05,
-      "loss": 5.1133,
-      "step": 48137
-    },
-    {
-      "epoch": 25.104563233376794,
-      "grad_norm": 1.6260484457015991,
-      "learning_rate": 5.214070351758794e-05,
-      "loss": 5.1955,
-      "step": 48138
-    },
-    {
-      "epoch": 25.105084745762714,
-      "grad_norm": 1.5609612464904785,
-      "learning_rate": 5.213969849246232e-05,
-      "loss": 5.0974,
-      "step": 48139
-    },
-    {
-      "epoch": 25.10560625814863,
-      "grad_norm": 1.5654407739639282,
-      "learning_rate": 5.213869346733668e-05,
-      "loss": 5.0046,
-      "step": 48140
-    },
-    {
-      "epoch": 25.10612777053455,
-      "grad_norm": 1.618502140045166,
-      "learning_rate": 5.2137688442211054e-05,
-      "loss": 5.3164,
-      "step": 48141
-    },
-    {
-      "epoch": 25.10664928292047,
-      "grad_norm": 1.4652643203735352,
-      "learning_rate": 5.213668341708543e-05,
-      "loss": 4.6613,
-      "step": 48142
-    },
-    {
-      "epoch": 25.107170795306388,
-      "grad_norm": 1.6721347570419312,
-      "learning_rate": 5.2135678391959796e-05,
-      "loss": 5.2065,
-      "step": 48143
-    },
-    {
-      "epoch": 25.107692307692307,
-      "grad_norm": 1.497338056564331,
-      "learning_rate": 5.2134673366834174e-05,
-      "loss": 5.3773,
-      "step": 48144
-    },
-    {
-      "epoch": 25.108213820078227,
-      "grad_norm": 1.6226699352264404,
-      "learning_rate": 5.213366834170854e-05,
-      "loss": 5.4894,
-      "step": 48145
-    },
-    {
-      "epoch": 25.108735332464146,
-      "grad_norm": 1.6273462772369385,
-      "learning_rate": 5.2132663316582916e-05,
-      "loss": 5.1295,
-      "step": 48146
-    },
-    {
-      "epoch": 25.109256844850066,
-      "grad_norm": 1.575742483139038,
-      "learning_rate": 5.213165829145729e-05,
-      "loss": 5.021,
-      "step": 48147
-    },
-    {
-      "epoch": 25.109778357235985,
-      "grad_norm": 1.5621916055679321,
-      "learning_rate": 5.2130653266331665e-05,
-      "loss": 5.3857,
-      "step": 48148
-    },
-    {
-      "epoch": 25.110299869621905,
-      "grad_norm": 1.4941315650939941,
-      "learning_rate": 5.212964824120603e-05,
-      "loss": 5.2961,
-      "step": 48149
-    },
-    {
-      "epoch": 25.110821382007824,
-      "grad_norm": 1.5661746263504028,
-      "learning_rate": 5.212864321608041e-05,
-      "loss": 4.9118,
-      "step": 48150
-    },
-    {
-      "epoch": 25.111342894393744,
-      "grad_norm": 1.6174672842025757,
-      "learning_rate": 5.212763819095478e-05,
-      "loss": 4.8483,
-      "step": 48151
-    },
-    {
-      "epoch": 25.11186440677966,
-      "grad_norm": 1.5959138870239258,
-      "learning_rate": 5.2126633165829156e-05,
-      "loss": 5.1718,
-      "step": 48152
-    },
-    {
-      "epoch": 25.11238591916558,
-      "grad_norm": 1.4583885669708252,
-      "learning_rate": 5.212562814070352e-05,
-      "loss": 5.2302,
-      "step": 48153
-    },
-    {
-      "epoch": 25.1129074315515,
-      "grad_norm": 1.5151971578598022,
-      "learning_rate": 5.21246231155779e-05,
-      "loss": 5.2939,
-      "step": 48154
-    },
-    {
-      "epoch": 25.113428943937418,
-      "grad_norm": 1.5206096172332764,
-      "learning_rate": 5.212361809045226e-05,
-      "loss": 4.7663,
-      "step": 48155
-    },
-    {
-      "epoch": 25.113950456323337,
-      "grad_norm": 1.446768879890442,
-      "learning_rate": 5.2122613065326634e-05,
-      "loss": 5.5966,
-      "step": 48156
-    },
-    {
-      "epoch": 25.114471968709257,
-      "grad_norm": 1.590512990951538,
-      "learning_rate": 5.212160804020101e-05,
-      "loss": 5.0482,
-      "step": 48157
-    },
-    {
-      "epoch": 25.114993481095176,
-      "grad_norm": 1.5574973821640015,
-      "learning_rate": 5.2120603015075376e-05,
-      "loss": 5.0848,
-      "step": 48158
-    },
-    {
-      "epoch": 25.115514993481096,
-      "grad_norm": 1.5858995914459229,
-      "learning_rate": 5.2119597989949754e-05,
-      "loss": 5.0175,
-      "step": 48159
-    },
-    {
-      "epoch": 25.116036505867015,
-      "grad_norm": 1.6092939376831055,
-      "learning_rate": 5.211859296482412e-05,
-      "loss": 5.2783,
-      "step": 48160
-    },
-    {
-      "epoch": 25.116558018252935,
-      "grad_norm": 1.4780962467193604,
-      "learning_rate": 5.2117587939698496e-05,
-      "loss": 5.604,
-      "step": 48161
-    },
-    {
-      "epoch": 25.117079530638854,
-      "grad_norm": 1.6225965023040771,
-      "learning_rate": 5.211658291457287e-05,
-      "loss": 5.3382,
-      "step": 48162
-    },
-    {
-      "epoch": 25.117601043024774,
-      "grad_norm": 1.6012520790100098,
-      "learning_rate": 5.2115577889447245e-05,
-      "loss": 5.3415,
-      "step": 48163
-    },
-    {
-      "epoch": 25.11812255541069,
-      "grad_norm": 1.6274304389953613,
-      "learning_rate": 5.211457286432161e-05,
-      "loss": 4.8036,
-      "step": 48164
-    },
-    {
-      "epoch": 25.11864406779661,
-      "grad_norm": 1.5565248727798462,
-      "learning_rate": 5.211356783919599e-05,
-      "loss": 5.2834,
-      "step": 48165
-    },
-    {
-      "epoch": 25.11916558018253,
-      "grad_norm": 1.6478276252746582,
-      "learning_rate": 5.211256281407035e-05,
-      "loss": 4.9939,
-      "step": 48166
-    },
-    {
-      "epoch": 25.119687092568448,
-      "grad_norm": 1.5891567468643188,
-      "learning_rate": 5.211155778894473e-05,
-      "loss": 5.1703,
-      "step": 48167
-    },
-    {
-      "epoch": 25.120208604954367,
-      "grad_norm": 1.660111904144287,
-      "learning_rate": 5.21105527638191e-05,
-      "loss": 5.0909,
-      "step": 48168
-    },
-    {
-      "epoch": 25.120730117340287,
-      "grad_norm": 1.5375797748565674,
-      "learning_rate": 5.2109547738693464e-05,
-      "loss": 5.3287,
-      "step": 48169
-    },
-    {
-      "epoch": 25.121251629726206,
-      "grad_norm": 1.6365638971328735,
-      "learning_rate": 5.210854271356784e-05,
-      "loss": 5.0463,
-      "step": 48170
-    },
-    {
-      "epoch": 25.121773142112126,
-      "grad_norm": 1.5868256092071533,
-      "learning_rate": 5.2107537688442206e-05,
-      "loss": 5.3186,
-      "step": 48171
-    },
-    {
-      "epoch": 25.122294654498045,
-      "grad_norm": 1.5055153369903564,
-      "learning_rate": 5.2106532663316584e-05,
-      "loss": 5.7097,
-      "step": 48172
-    },
-    {
-      "epoch": 25.122816166883965,
-      "grad_norm": 1.635202407836914,
-      "learning_rate": 5.2105527638190955e-05,
-      "loss": 4.7157,
-      "step": 48173
-    },
-    {
-      "epoch": 25.123337679269884,
-      "grad_norm": 1.6698713302612305,
-      "learning_rate": 5.210452261306533e-05,
-      "loss": 5.2708,
-      "step": 48174
-    },
-    {
-      "epoch": 25.1238591916558,
-      "grad_norm": 1.646334171295166,
-      "learning_rate": 5.21035175879397e-05,
-      "loss": 5.097,
-      "step": 48175
-    },
-    {
-      "epoch": 25.12438070404172,
-      "grad_norm": 1.5443772077560425,
-      "learning_rate": 5.2102512562814075e-05,
-      "loss": 5.03,
-      "step": 48176
-    },
-    {
-      "epoch": 25.12490221642764,
-      "grad_norm": 1.578948736190796,
-      "learning_rate": 5.2101507537688446e-05,
-      "loss": 5.3146,
-      "step": 48177
-    },
-    {
-      "epoch": 25.12542372881356,
-      "grad_norm": 1.5742299556732178,
-      "learning_rate": 5.2100502512562824e-05,
-      "loss": 5.589,
-      "step": 48178
-    },
-    {
-      "epoch": 25.125945241199478,
-      "grad_norm": 1.5790657997131348,
-      "learning_rate": 5.209949748743719e-05,
-      "loss": 5.4131,
-      "step": 48179
-    },
-    {
-      "epoch": 25.126466753585397,
-      "grad_norm": 1.502204179763794,
-      "learning_rate": 5.2098492462311566e-05,
-      "loss": 5.5255,
-      "step": 48180
-    },
-    {
-      "epoch": 25.126988265971317,
-      "grad_norm": 1.6068446636199951,
-      "learning_rate": 5.209748743718593e-05,
-      "loss": 5.4598,
-      "step": 48181
-    },
-    {
-      "epoch": 25.127509778357236,
-      "grad_norm": 1.5747489929199219,
-      "learning_rate": 5.20964824120603e-05,
-      "loss": 4.9203,
-      "step": 48182
-    },
-    {
-      "epoch": 25.128031290743156,
-      "grad_norm": 1.4774532318115234,
-      "learning_rate": 5.209547738693468e-05,
-      "loss": 5.2928,
-      "step": 48183
-    },
-    {
-      "epoch": 25.128552803129075,
-      "grad_norm": 1.6008639335632324,
-      "learning_rate": 5.2094472361809043e-05,
-      "loss": 5.1,
-      "step": 48184
-    },
-    {
-      "epoch": 25.129074315514995,
-      "grad_norm": 1.4689936637878418,
-      "learning_rate": 5.209346733668342e-05,
-      "loss": 5.2165,
-      "step": 48185
-    },
-    {
-      "epoch": 25.129595827900914,
-      "grad_norm": 1.4817838668823242,
-      "learning_rate": 5.2092462311557786e-05,
-      "loss": 5.1851,
-      "step": 48186
-    },
-    {
-      "epoch": 25.13011734028683,
-      "grad_norm": 1.6935149431228638,
-      "learning_rate": 5.209145728643216e-05,
-      "loss": 4.9495,
-      "step": 48187
-    },
-    {
-      "epoch": 25.13063885267275,
-      "grad_norm": 1.4817986488342285,
-      "learning_rate": 5.2090452261306534e-05,
-      "loss": 5.4332,
-      "step": 48188
-    },
-    {
-      "epoch": 25.13116036505867,
-      "grad_norm": 1.4891773462295532,
-      "learning_rate": 5.208944723618091e-05,
-      "loss": 5.1221,
-      "step": 48189
-    },
-    {
-      "epoch": 25.13168187744459,
-      "grad_norm": 1.561461329460144,
-      "learning_rate": 5.2088442211055276e-05,
-      "loss": 5.2946,
-      "step": 48190
-    },
-    {
-      "epoch": 25.132203389830508,
-      "grad_norm": 1.5797206163406372,
-      "learning_rate": 5.2087437185929654e-05,
-      "loss": 5.0065,
-      "step": 48191
-    },
-    {
-      "epoch": 25.132724902216427,
-      "grad_norm": 1.6771870851516724,
-      "learning_rate": 5.208643216080402e-05,
-      "loss": 4.7235,
-      "step": 48192
-    },
-    {
-      "epoch": 25.133246414602347,
-      "grad_norm": 1.5815049409866333,
-      "learning_rate": 5.2085427135678396e-05,
-      "loss": 5.4588,
-      "step": 48193
-    },
-    {
-      "epoch": 25.133767926988266,
-      "grad_norm": 1.6302151679992676,
-      "learning_rate": 5.208442211055277e-05,
-      "loss": 4.9791,
-      "step": 48194
-    },
-    {
-      "epoch": 25.134289439374186,
-      "grad_norm": 1.4869463443756104,
-      "learning_rate": 5.208341708542713e-05,
-      "loss": 5.1828,
-      "step": 48195
-    },
-    {
-      "epoch": 25.134810951760105,
-      "grad_norm": 1.5574650764465332,
-      "learning_rate": 5.208241206030151e-05,
-      "loss": 5.0029,
-      "step": 48196
-    },
-    {
-      "epoch": 25.135332464146025,
-      "grad_norm": 1.5514665842056274,
-      "learning_rate": 5.208140703517588e-05,
-      "loss": 4.9966,
-      "step": 48197
-    },
-    {
-      "epoch": 25.135853976531944,
-      "grad_norm": 1.5175729990005493,
-      "learning_rate": 5.208040201005026e-05,
-      "loss": 5.2739,
-      "step": 48198
-    },
-    {
-      "epoch": 25.13637548891786,
-      "grad_norm": 1.5870496034622192,
-      "learning_rate": 5.207939698492462e-05,
-      "loss": 5.0509,
-      "step": 48199
-    },
-    {
-      "epoch": 25.13689700130378,
-      "grad_norm": 1.6089013814926147,
-      "learning_rate": 5.2078391959799e-05,
-      "loss": 4.8098,
-      "step": 48200
-    },
-    {
-      "epoch": 25.1374185136897,
-      "grad_norm": 1.5417670011520386,
-      "learning_rate": 5.2077386934673365e-05,
-      "loss": 5.2898,
-      "step": 48201
-    },
-    {
-      "epoch": 25.13794002607562,
-      "grad_norm": 1.742838978767395,
-      "learning_rate": 5.207638190954774e-05,
-      "loss": 5.0464,
-      "step": 48202
-    },
-    {
-      "epoch": 25.138461538461538,
-      "grad_norm": 1.5392214059829712,
-      "learning_rate": 5.2075376884422114e-05,
-      "loss": 4.871,
-      "step": 48203
-    },
-    {
-      "epoch": 25.138983050847457,
-      "grad_norm": 1.560907244682312,
-      "learning_rate": 5.207437185929649e-05,
-      "loss": 5.5693,
-      "step": 48204
-    },
-    {
-      "epoch": 25.139504563233377,
-      "grad_norm": 1.5184104442596436,
-      "learning_rate": 5.2073366834170856e-05,
-      "loss": 5.5042,
-      "step": 48205
-    },
-    {
-      "epoch": 25.140026075619296,
-      "grad_norm": 1.57496976852417,
-      "learning_rate": 5.2072361809045234e-05,
-      "loss": 5.4034,
-      "step": 48206
-    },
-    {
-      "epoch": 25.140547588005216,
-      "grad_norm": 1.626519799232483,
-      "learning_rate": 5.20713567839196e-05,
-      "loss": 5.2947,
-      "step": 48207
-    },
-    {
-      "epoch": 25.141069100391135,
-      "grad_norm": 1.5329514741897583,
-      "learning_rate": 5.207035175879397e-05,
-      "loss": 5.3089,
-      "step": 48208
-    },
-    {
-      "epoch": 25.141590612777055,
-      "grad_norm": 1.5630195140838623,
-      "learning_rate": 5.206934673366835e-05,
-      "loss": 5.2724,
-      "step": 48209
-    },
-    {
-      "epoch": 25.142112125162974,
-      "grad_norm": 1.5848133563995361,
-      "learning_rate": 5.206834170854271e-05,
-      "loss": 5.1734,
-      "step": 48210
-    },
-    {
-      "epoch": 25.14263363754889,
-      "grad_norm": 1.55434250831604,
-      "learning_rate": 5.206733668341709e-05,
-      "loss": 5.6921,
-      "step": 48211
-    },
-    {
-      "epoch": 25.14315514993481,
-      "grad_norm": 1.5273516178131104,
-      "learning_rate": 5.206633165829145e-05,
-      "loss": 5.3113,
-      "step": 48212
-    },
-    {
-      "epoch": 25.14367666232073,
-      "grad_norm": 1.5964916944503784,
-      "learning_rate": 5.206532663316583e-05,
-      "loss": 5.146,
-      "step": 48213
-    },
-    {
-      "epoch": 25.14419817470665,
-      "grad_norm": 1.6248948574066162,
-      "learning_rate": 5.20643216080402e-05,
-      "loss": 5.3516,
-      "step": 48214
-    },
-    {
-      "epoch": 25.144719687092568,
-      "grad_norm": 1.7536388635635376,
-      "learning_rate": 5.206331658291458e-05,
-      "loss": 5.2284,
-      "step": 48215
-    },
-    {
-      "epoch": 25.145241199478487,
-      "grad_norm": 1.6195341348648071,
-      "learning_rate": 5.2062311557788944e-05,
-      "loss": 5.1896,
-      "step": 48216
-    },
-    {
-      "epoch": 25.145762711864407,
-      "grad_norm": 1.5441358089447021,
-      "learning_rate": 5.206130653266332e-05,
-      "loss": 5.228,
-      "step": 48217
-    },
-    {
-      "epoch": 25.146284224250326,
-      "grad_norm": 1.6368287801742554,
-      "learning_rate": 5.206030150753769e-05,
-      "loss": 5.5487,
-      "step": 48218
-    },
-    {
-      "epoch": 25.146805736636246,
-      "grad_norm": 1.5946929454803467,
-      "learning_rate": 5.205929648241207e-05,
-      "loss": 4.5835,
-      "step": 48219
-    },
-    {
-      "epoch": 25.147327249022165,
-      "grad_norm": 1.424033522605896,
-      "learning_rate": 5.2058291457286435e-05,
-      "loss": 5.5432,
-      "step": 48220
-    },
-    {
-      "epoch": 25.147848761408085,
-      "grad_norm": 1.4372049570083618,
-      "learning_rate": 5.20572864321608e-05,
-      "loss": 5.0455,
-      "step": 48221
-    },
-    {
-      "epoch": 25.148370273794004,
-      "grad_norm": 1.5308092832565308,
-      "learning_rate": 5.205628140703518e-05,
-      "loss": 5.0259,
-      "step": 48222
-    },
-    {
-      "epoch": 25.14889178617992,
-      "grad_norm": 1.7394767999649048,
-      "learning_rate": 5.205527638190955e-05,
-      "loss": 4.9047,
-      "step": 48223
-    },
-    {
-      "epoch": 25.14941329856584,
-      "grad_norm": 1.9454396963119507,
-      "learning_rate": 5.2054271356783926e-05,
-      "loss": 4.8059,
-      "step": 48224
-    },
-    {
-      "epoch": 25.14993481095176,
-      "grad_norm": 1.652451992034912,
-      "learning_rate": 5.205326633165829e-05,
-      "loss": 5.2775,
-      "step": 48225
-    },
-    {
-      "epoch": 25.15045632333768,
-      "grad_norm": 1.5885117053985596,
-      "learning_rate": 5.205226130653267e-05,
-      "loss": 5.2052,
-      "step": 48226
-    },
-    {
-      "epoch": 25.150977835723598,
-      "grad_norm": 1.6662418842315674,
-      "learning_rate": 5.205125628140703e-05,
-      "loss": 4.8089,
-      "step": 48227
-    },
-    {
-      "epoch": 25.151499348109517,
-      "grad_norm": 1.6505861282348633,
-      "learning_rate": 5.205025125628141e-05,
-      "loss": 4.9137,
-      "step": 48228
-    },
-    {
-      "epoch": 25.152020860495437,
-      "grad_norm": 1.67071533203125,
-      "learning_rate": 5.204924623115578e-05,
-      "loss": 4.6297,
-      "step": 48229
-    },
-    {
-      "epoch": 25.152542372881356,
-      "grad_norm": 1.5707292556762695,
-      "learning_rate": 5.204824120603016e-05,
-      "loss": 5.2695,
-      "step": 48230
-    },
-    {
-      "epoch": 25.153063885267276,
-      "grad_norm": 1.7561732530593872,
-      "learning_rate": 5.2047236180904523e-05,
-      "loss": 4.7642,
-      "step": 48231
-    },
-    {
-      "epoch": 25.153585397653195,
-      "grad_norm": 1.5681991577148438,
-      "learning_rate": 5.20462311557789e-05,
-      "loss": 4.9484,
-      "step": 48232
-    },
-    {
-      "epoch": 25.154106910039115,
-      "grad_norm": 1.605025291442871,
-      "learning_rate": 5.2045226130653266e-05,
-      "loss": 5.0805,
-      "step": 48233
-    },
-    {
-      "epoch": 25.154628422425034,
-      "grad_norm": 1.622686743736267,
-      "learning_rate": 5.204422110552764e-05,
-      "loss": 4.8012,
-      "step": 48234
-    },
-    {
-      "epoch": 25.15514993481095,
-      "grad_norm": 1.5964213609695435,
-      "learning_rate": 5.2043216080402014e-05,
-      "loss": 5.1172,
-      "step": 48235
-    },
-    {
-      "epoch": 25.15567144719687,
-      "grad_norm": 1.6251388788223267,
-      "learning_rate": 5.204221105527638e-05,
-      "loss": 4.179,
-      "step": 48236
-    },
-    {
-      "epoch": 25.15619295958279,
-      "grad_norm": 1.4505014419555664,
-      "learning_rate": 5.2041206030150757e-05,
-      "loss": 5.2209,
-      "step": 48237
-    },
-    {
-      "epoch": 25.15671447196871,
-      "grad_norm": 1.5651741027832031,
-      "learning_rate": 5.204020100502513e-05,
-      "loss": 4.9761,
-      "step": 48238
-    },
-    {
-      "epoch": 25.157235984354628,
-      "grad_norm": 1.6765464544296265,
-      "learning_rate": 5.2039195979899505e-05,
-      "loss": 5.4433,
-      "step": 48239
-    },
-    {
-      "epoch": 25.157757496740548,
-      "grad_norm": 1.5783615112304688,
-      "learning_rate": 5.203819095477387e-05,
-      "loss": 4.7876,
-      "step": 48240
-    },
-    {
-      "epoch": 25.158279009126467,
-      "grad_norm": 1.5194097757339478,
-      "learning_rate": 5.203718592964825e-05,
-      "loss": 5.4855,
-      "step": 48241
-    },
-    {
-      "epoch": 25.158800521512386,
-      "grad_norm": 1.5367183685302734,
-      "learning_rate": 5.203618090452261e-05,
-      "loss": 4.9585,
-      "step": 48242
-    },
-    {
-      "epoch": 25.159322033898306,
-      "grad_norm": 1.6741039752960205,
-      "learning_rate": 5.203517587939699e-05,
-      "loss": 5.2096,
-      "step": 48243
-    },
-    {
-      "epoch": 25.159843546284225,
-      "grad_norm": 1.6310003995895386,
-      "learning_rate": 5.203417085427136e-05,
-      "loss": 5.1442,
-      "step": 48244
-    },
-    {
-      "epoch": 25.160365058670145,
-      "grad_norm": 1.4548531770706177,
-      "learning_rate": 5.203316582914574e-05,
-      "loss": 5.1668,
-      "step": 48245
-    },
-    {
-      "epoch": 25.160886571056064,
-      "grad_norm": 1.549142599105835,
-      "learning_rate": 5.20321608040201e-05,
-      "loss": 5.2885,
-      "step": 48246
-    },
-    {
-      "epoch": 25.16140808344198,
-      "grad_norm": 1.691920518875122,
-      "learning_rate": 5.203115577889448e-05,
-      "loss": 5.0742,
-      "step": 48247
-    },
-    {
-      "epoch": 25.1619295958279,
-      "grad_norm": 1.5953315496444702,
-      "learning_rate": 5.2030150753768845e-05,
-      "loss": 4.9535,
-      "step": 48248
-    },
-    {
-      "epoch": 25.16245110821382,
-      "grad_norm": 1.5538278818130493,
-      "learning_rate": 5.2029145728643216e-05,
-      "loss": 5.0225,
-      "step": 48249
-    },
-    {
-      "epoch": 25.16297262059974,
-      "grad_norm": 1.59413743019104,
-      "learning_rate": 5.2028140703517594e-05,
-      "loss": 5.1177,
-      "step": 48250
-    },
-    {
-      "epoch": 25.163494132985658,
-      "grad_norm": 1.7710264921188354,
-      "learning_rate": 5.202713567839196e-05,
-      "loss": 4.5543,
-      "step": 48251
-    },
-    {
-      "epoch": 25.164015645371578,
-      "grad_norm": 1.609657645225525,
-      "learning_rate": 5.2026130653266336e-05,
-      "loss": 5.361,
-      "step": 48252
-    },
-    {
-      "epoch": 25.164537157757497,
-      "grad_norm": 1.5920768976211548,
-      "learning_rate": 5.20251256281407e-05,
-      "loss": 5.3722,
-      "step": 48253
-    },
-    {
-      "epoch": 25.165058670143416,
-      "grad_norm": 1.5664012432098389,
-      "learning_rate": 5.202412060301508e-05,
-      "loss": 5.0528,
-      "step": 48254
-    },
-    {
-      "epoch": 25.165580182529336,
-      "grad_norm": 1.5830000638961792,
-      "learning_rate": 5.202311557788945e-05,
-      "loss": 5.2396,
-      "step": 48255
-    },
-    {
-      "epoch": 25.166101694915255,
-      "grad_norm": 1.5106652975082397,
-      "learning_rate": 5.202211055276383e-05,
-      "loss": 4.7429,
-      "step": 48256
-    },
-    {
-      "epoch": 25.166623207301175,
-      "grad_norm": 1.5451288223266602,
-      "learning_rate": 5.202110552763819e-05,
-      "loss": 5.763,
-      "step": 48257
-    },
-    {
-      "epoch": 25.167144719687094,
-      "grad_norm": 1.5612925291061401,
-      "learning_rate": 5.202010050251257e-05,
-      "loss": 4.5147,
-      "step": 48258
-    },
-    {
-      "epoch": 25.16766623207301,
-      "grad_norm": 1.46934175491333,
-      "learning_rate": 5.201909547738693e-05,
-      "loss": 5.4977,
-      "step": 48259
-    },
-    {
-      "epoch": 25.16818774445893,
-      "grad_norm": 1.5998698472976685,
-      "learning_rate": 5.201809045226131e-05,
-      "loss": 5.2459,
-      "step": 48260
-    },
-    {
-      "epoch": 25.16870925684485,
-      "grad_norm": 1.6068105697631836,
-      "learning_rate": 5.201708542713568e-05,
-      "loss": 5.2279,
-      "step": 48261
-    },
-    {
-      "epoch": 25.16923076923077,
-      "grad_norm": 1.5778433084487915,
-      "learning_rate": 5.2016080402010046e-05,
-      "loss": 5.5873,
-      "step": 48262
-    },
-    {
-      "epoch": 25.169752281616688,
-      "grad_norm": 1.5922578573226929,
-      "learning_rate": 5.2015075376884424e-05,
-      "loss": 5.4446,
-      "step": 48263
-    },
-    {
-      "epoch": 25.170273794002608,
-      "grad_norm": 1.5329680442810059,
-      "learning_rate": 5.2014070351758795e-05,
-      "loss": 4.7624,
-      "step": 48264
-    },
-    {
-      "epoch": 25.170795306388527,
-      "grad_norm": 1.5803509950637817,
-      "learning_rate": 5.201306532663317e-05,
-      "loss": 4.9062,
-      "step": 48265
-    },
-    {
-      "epoch": 25.171316818774446,
-      "grad_norm": 1.4776437282562256,
-      "learning_rate": 5.201206030150754e-05,
-      "loss": 5.4162,
-      "step": 48266
-    },
-    {
-      "epoch": 25.171838331160366,
-      "grad_norm": 1.5642379522323608,
-      "learning_rate": 5.2011055276381915e-05,
-      "loss": 5.4002,
-      "step": 48267
-    },
-    {
-      "epoch": 25.172359843546285,
-      "grad_norm": 1.5841513872146606,
-      "learning_rate": 5.201005025125628e-05,
-      "loss": 5.218,
-      "step": 48268
-    },
-    {
-      "epoch": 25.172881355932205,
-      "grad_norm": 1.5213850736618042,
-      "learning_rate": 5.200904522613066e-05,
-      "loss": 5.09,
-      "step": 48269
-    },
-    {
-      "epoch": 25.17340286831812,
-      "grad_norm": 1.6134103536605835,
-      "learning_rate": 5.200804020100503e-05,
-      "loss": 4.5599,
-      "step": 48270
-    },
-    {
-      "epoch": 25.17392438070404,
-      "grad_norm": 1.5928481817245483,
-      "learning_rate": 5.2007035175879406e-05,
-      "loss": 5.1158,
-      "step": 48271
-    },
-    {
-      "epoch": 25.17444589308996,
-      "grad_norm": 1.5425201654434204,
-      "learning_rate": 5.200603015075377e-05,
-      "loss": 5.5117,
-      "step": 48272
-    },
-    {
-      "epoch": 25.17496740547588,
-      "grad_norm": 1.50590181350708,
-      "learning_rate": 5.200502512562815e-05,
-      "loss": 4.9966,
-      "step": 48273
-    },
-    {
-      "epoch": 25.1754889178618,
-      "grad_norm": 1.586564064025879,
-      "learning_rate": 5.200402010050251e-05,
-      "loss": 5.1931,
-      "step": 48274
-    },
-    {
-      "epoch": 25.176010430247718,
-      "grad_norm": 1.6813656091690063,
-      "learning_rate": 5.2003015075376884e-05,
-      "loss": 5.3193,
-      "step": 48275
-    },
-    {
-      "epoch": 25.176531942633638,
-      "grad_norm": 1.583972692489624,
-      "learning_rate": 5.200201005025126e-05,
-      "loss": 5.2776,
-      "step": 48276
-    },
-    {
-      "epoch": 25.177053455019557,
-      "grad_norm": 1.694563627243042,
-      "learning_rate": 5.2001005025125626e-05,
-      "loss": 5.3477,
-      "step": 48277
-    },
-    {
-      "epoch": 25.177574967405477,
-      "grad_norm": 1.6316635608673096,
-      "learning_rate": 5.2000000000000004e-05,
-      "loss": 5.2086,
-      "step": 48278
-    },
-    {
-      "epoch": 25.178096479791396,
-      "grad_norm": 1.5485838651657104,
-      "learning_rate": 5.199899497487437e-05,
-      "loss": 5.5242,
-      "step": 48279
-    },
-    {
-      "epoch": 25.178617992177315,
-      "grad_norm": 1.649080514907837,
-      "learning_rate": 5.1997989949748746e-05,
-      "loss": 4.9274,
-      "step": 48280
-    },
-    {
-      "epoch": 25.179139504563235,
-      "grad_norm": 1.5844050645828247,
-      "learning_rate": 5.199698492462312e-05,
-      "loss": 5.1668,
-      "step": 48281
-    },
-    {
-      "epoch": 25.17966101694915,
-      "grad_norm": 1.5307279825210571,
-      "learning_rate": 5.1995979899497495e-05,
-      "loss": 5.2937,
-      "step": 48282
-    },
-    {
-      "epoch": 25.18018252933507,
-      "grad_norm": 1.5992871522903442,
-      "learning_rate": 5.199497487437186e-05,
-      "loss": 5.3289,
-      "step": 48283
-    },
-    {
-      "epoch": 25.18070404172099,
-      "grad_norm": 1.5622804164886475,
-      "learning_rate": 5.199396984924624e-05,
-      "loss": 5.1153,
-      "step": 48284
-    },
-    {
-      "epoch": 25.18122555410691,
-      "grad_norm": 1.6254280805587769,
-      "learning_rate": 5.199296482412061e-05,
-      "loss": 4.7735,
-      "step": 48285
-    },
-    {
-      "epoch": 25.18174706649283,
-      "grad_norm": 1.4742189645767212,
-      "learning_rate": 5.1991959798994986e-05,
-      "loss": 5.1951,
-      "step": 48286
-    },
-    {
-      "epoch": 25.182268578878748,
-      "grad_norm": 1.7222024202346802,
-      "learning_rate": 5.199095477386935e-05,
-      "loss": 5.3222,
-      "step": 48287
-    },
-    {
-      "epoch": 25.182790091264668,
-      "grad_norm": 1.5868488550186157,
-      "learning_rate": 5.1989949748743714e-05,
-      "loss": 5.3154,
-      "step": 48288
-    },
-    {
-      "epoch": 25.183311603650587,
-      "grad_norm": 1.6669461727142334,
-      "learning_rate": 5.198894472361809e-05,
-      "loss": 5.4858,
-      "step": 48289
-    },
-    {
-      "epoch": 25.183833116036507,
-      "grad_norm": 1.6729753017425537,
-      "learning_rate": 5.198793969849246e-05,
-      "loss": 5.2408,
-      "step": 48290
-    },
-    {
-      "epoch": 25.184354628422426,
-      "grad_norm": 1.4920904636383057,
-      "learning_rate": 5.198693467336684e-05,
-      "loss": 5.2494,
-      "step": 48291
-    },
-    {
-      "epoch": 25.184876140808345,
-      "grad_norm": 1.6071747541427612,
-      "learning_rate": 5.1985929648241205e-05,
-      "loss": 5.4232,
-      "step": 48292
-    },
-    {
-      "epoch": 25.185397653194265,
-      "grad_norm": 1.6151361465454102,
-      "learning_rate": 5.198492462311558e-05,
-      "loss": 5.3767,
-      "step": 48293
-    },
-    {
-      "epoch": 25.18591916558018,
-      "grad_norm": 1.5583058595657349,
-      "learning_rate": 5.198391959798995e-05,
-      "loss": 5.2987,
-      "step": 48294
-    },
-    {
-      "epoch": 25.1864406779661,
-      "grad_norm": 1.5590896606445312,
-      "learning_rate": 5.1982914572864325e-05,
-      "loss": 5.4547,
-      "step": 48295
-    },
-    {
-      "epoch": 25.18696219035202,
-      "grad_norm": 1.6528080701828003,
-      "learning_rate": 5.1981909547738696e-05,
-      "loss": 5.197,
-      "step": 48296
-    },
-    {
-      "epoch": 25.18748370273794,
-      "grad_norm": 1.5021320581436157,
-      "learning_rate": 5.1980904522613074e-05,
-      "loss": 4.9096,
-      "step": 48297
-    },
-    {
-      "epoch": 25.18800521512386,
-      "grad_norm": 1.5672109127044678,
-      "learning_rate": 5.197989949748744e-05,
-      "loss": 5.0116,
-      "step": 48298
-    },
-    {
-      "epoch": 25.188526727509778,
-      "grad_norm": 1.4901870489120483,
-      "learning_rate": 5.1978894472361816e-05,
-      "loss": 5.2673,
-      "step": 48299
-    },
-    {
-      "epoch": 25.189048239895698,
-      "grad_norm": 1.591065764427185,
-      "learning_rate": 5.197788944723618e-05,
-      "loss": 5.1592,
-      "step": 48300
-    },
-    {
-      "epoch": 25.189569752281617,
-      "grad_norm": 1.6320013999938965,
-      "learning_rate": 5.197688442211055e-05,
-      "loss": 5.55,
-      "step": 48301
-    },
-    {
-      "epoch": 25.190091264667537,
-      "grad_norm": 1.6306010484695435,
-      "learning_rate": 5.197587939698493e-05,
-      "loss": 5.1082,
-      "step": 48302
-    },
-    {
-      "epoch": 25.190612777053456,
-      "grad_norm": 1.5763851404190063,
-      "learning_rate": 5.1974874371859293e-05,
-      "loss": 5.0773,
-      "step": 48303
-    },
-    {
-      "epoch": 25.191134289439375,
-      "grad_norm": 1.5967589616775513,
-      "learning_rate": 5.197386934673367e-05,
-      "loss": 5.2321,
-      "step": 48304
-    },
-    {
-      "epoch": 25.191655801825295,
-      "grad_norm": 1.5414631366729736,
-      "learning_rate": 5.197286432160804e-05,
-      "loss": 5.1967,
-      "step": 48305
-    },
-    {
-      "epoch": 25.19217731421121,
-      "grad_norm": 1.4847177267074585,
-      "learning_rate": 5.197185929648242e-05,
-      "loss": 5.2536,
-      "step": 48306
-    },
-    {
-      "epoch": 25.19269882659713,
-      "grad_norm": 1.6727932691574097,
-      "learning_rate": 5.1970854271356784e-05,
-      "loss": 5.6573,
-      "step": 48307
-    },
-    {
-      "epoch": 25.19322033898305,
-      "grad_norm": 1.5710893869400024,
-      "learning_rate": 5.196984924623116e-05,
-      "loss": 5.1313,
-      "step": 48308
-    },
-    {
-      "epoch": 25.19374185136897,
-      "grad_norm": 1.6223132610321045,
-      "learning_rate": 5.1968844221105527e-05,
-      "loss": 5.1001,
-      "step": 48309
-    },
-    {
-      "epoch": 25.19426336375489,
-      "grad_norm": 1.5251421928405762,
-      "learning_rate": 5.1967839195979904e-05,
-      "loss": 5.0532,
-      "step": 48310
-    },
-    {
-      "epoch": 25.194784876140808,
-      "grad_norm": 1.5213109254837036,
-      "learning_rate": 5.1966834170854275e-05,
-      "loss": 5.0112,
-      "step": 48311
-    },
-    {
-      "epoch": 25.195306388526728,
-      "grad_norm": 1.608641505241394,
-      "learning_rate": 5.196582914572865e-05,
-      "loss": 5.616,
-      "step": 48312
-    },
-    {
-      "epoch": 25.195827900912647,
-      "grad_norm": 1.5241128206253052,
-      "learning_rate": 5.196482412060302e-05,
-      "loss": 5.3145,
-      "step": 48313
-    },
-    {
-      "epoch": 25.196349413298567,
-      "grad_norm": 1.6935138702392578,
-      "learning_rate": 5.196381909547738e-05,
-      "loss": 4.6836,
-      "step": 48314
-    },
-    {
-      "epoch": 25.196870925684486,
-      "grad_norm": 1.6507019996643066,
-      "learning_rate": 5.196281407035176e-05,
-      "loss": 5.0499,
-      "step": 48315
-    },
-    {
-      "epoch": 25.197392438070406,
-      "grad_norm": 1.596078872680664,
-      "learning_rate": 5.196180904522613e-05,
-      "loss": 4.6802,
-      "step": 48316
-    },
-    {
-      "epoch": 25.197913950456325,
-      "grad_norm": 1.546017050743103,
-      "learning_rate": 5.196080402010051e-05,
-      "loss": 4.7077,
-      "step": 48317
-    },
-    {
-      "epoch": 25.19843546284224,
-      "grad_norm": 1.629560112953186,
-      "learning_rate": 5.195979899497487e-05,
-      "loss": 5.0823,
-      "step": 48318
-    },
-    {
-      "epoch": 25.19895697522816,
-      "grad_norm": 1.6828365325927734,
-      "learning_rate": 5.195879396984925e-05,
-      "loss": 5.5038,
-      "step": 48319
-    },
-    {
-      "epoch": 25.19947848761408,
-      "grad_norm": 1.5470513105392456,
-      "learning_rate": 5.1957788944723615e-05,
-      "loss": 5.5871,
-      "step": 48320
-    },
-    {
-      "epoch": 25.2,
-      "grad_norm": 1.5760334730148315,
-      "learning_rate": 5.195678391959799e-05,
-      "loss": 5.1638,
-      "step": 48321
-    },
-    {
-      "epoch": 25.20052151238592,
-      "grad_norm": 1.4575612545013428,
-      "learning_rate": 5.1955778894472364e-05,
-      "loss": 5.5824,
-      "step": 48322
-    },
-    {
-      "epoch": 25.201043024771838,
-      "grad_norm": 1.734055519104004,
-      "learning_rate": 5.195477386934674e-05,
-      "loss": 4.7759,
-      "step": 48323
-    },
-    {
-      "epoch": 25.201564537157758,
-      "grad_norm": 1.6148805618286133,
-      "learning_rate": 5.1953768844221106e-05,
-      "loss": 5.0881,
-      "step": 48324
-    },
-    {
-      "epoch": 25.202086049543677,
-      "grad_norm": 1.574838399887085,
-      "learning_rate": 5.1952763819095484e-05,
-      "loss": 5.1639,
-      "step": 48325
-    },
-    {
-      "epoch": 25.202607561929597,
-      "grad_norm": 1.6231328248977661,
-      "learning_rate": 5.1951758793969855e-05,
-      "loss": 5.0672,
-      "step": 48326
-    },
-    {
-      "epoch": 25.203129074315516,
-      "grad_norm": 1.6548125743865967,
-      "learning_rate": 5.195075376884422e-05,
-      "loss": 5.4811,
-      "step": 48327
-    },
-    {
-      "epoch": 25.203650586701436,
-      "grad_norm": 1.6284483671188354,
-      "learning_rate": 5.19497487437186e-05,
-      "loss": 5.1359,
-      "step": 48328
-    },
-    {
-      "epoch": 25.204172099087355,
-      "grad_norm": 1.5638811588287354,
-      "learning_rate": 5.194874371859296e-05,
-      "loss": 5.5895,
-      "step": 48329
-    },
-    {
-      "epoch": 25.20469361147327,
-      "grad_norm": 1.6855058670043945,
-      "learning_rate": 5.194773869346734e-05,
-      "loss": 4.8557,
-      "step": 48330
-    },
-    {
-      "epoch": 25.20521512385919,
-      "grad_norm": 1.4649975299835205,
-      "learning_rate": 5.194673366834171e-05,
-      "loss": 5.2772,
-      "step": 48331
-    },
-    {
-      "epoch": 25.20573663624511,
-      "grad_norm": 1.6855249404907227,
-      "learning_rate": 5.194572864321609e-05,
-      "loss": 4.7377,
-      "step": 48332
-    },
-    {
-      "epoch": 25.20625814863103,
-      "grad_norm": 1.5740643739700317,
-      "learning_rate": 5.194472361809045e-05,
-      "loss": 5.3692,
-      "step": 48333
-    },
-    {
-      "epoch": 25.20677966101695,
-      "grad_norm": 1.50991690158844,
-      "learning_rate": 5.194371859296483e-05,
-      "loss": 4.9255,
-      "step": 48334
-    },
-    {
-      "epoch": 25.20730117340287,
-      "grad_norm": 1.4853595495224,
-      "learning_rate": 5.1942713567839194e-05,
-      "loss": 5.0932,
-      "step": 48335
-    },
-    {
-      "epoch": 25.207822685788788,
-      "grad_norm": 1.937727689743042,
-      "learning_rate": 5.194170854271357e-05,
-      "loss": 5.0133,
-      "step": 48336
-    },
-    {
-      "epoch": 25.208344198174707,
-      "grad_norm": 1.5737426280975342,
-      "learning_rate": 5.194070351758794e-05,
-      "loss": 5.6157,
-      "step": 48337
-    },
-    {
-      "epoch": 25.208865710560627,
-      "grad_norm": 1.6499024629592896,
-      "learning_rate": 5.193969849246232e-05,
-      "loss": 5.1248,
-      "step": 48338
-    },
-    {
-      "epoch": 25.209387222946546,
-      "grad_norm": 1.5909255743026733,
-      "learning_rate": 5.1938693467336685e-05,
-      "loss": 5.5741,
-      "step": 48339
-    },
-    {
-      "epoch": 25.209908735332466,
-      "grad_norm": 1.5819053649902344,
-      "learning_rate": 5.193768844221106e-05,
-      "loss": 4.8089,
-      "step": 48340
-    },
-    {
-      "epoch": 25.210430247718385,
-      "grad_norm": 1.6573951244354248,
-      "learning_rate": 5.193668341708543e-05,
-      "loss": 5.1161,
-      "step": 48341
-    },
-    {
-      "epoch": 25.2109517601043,
-      "grad_norm": 1.462242603302002,
-      "learning_rate": 5.19356783919598e-05,
-      "loss": 5.606,
-      "step": 48342
-    },
-    {
-      "epoch": 25.21147327249022,
-      "grad_norm": 1.5756440162658691,
-      "learning_rate": 5.1934673366834176e-05,
-      "loss": 5.0698,
-      "step": 48343
-    },
-    {
-      "epoch": 25.21199478487614,
-      "grad_norm": 1.5252902507781982,
-      "learning_rate": 5.193366834170854e-05,
-      "loss": 5.4856,
-      "step": 48344
-    },
-    {
-      "epoch": 25.21251629726206,
-      "grad_norm": 1.5936272144317627,
-      "learning_rate": 5.193266331658292e-05,
-      "loss": 5.1274,
-      "step": 48345
-    },
-    {
-      "epoch": 25.21303780964798,
-      "grad_norm": 1.6166918277740479,
-      "learning_rate": 5.193165829145728e-05,
-      "loss": 5.0241,
-      "step": 48346
-    },
-    {
-      "epoch": 25.2135593220339,
-      "grad_norm": 1.5729111433029175,
-      "learning_rate": 5.193065326633166e-05,
-      "loss": 5.1316,
-      "step": 48347
-    },
-    {
-      "epoch": 25.214080834419818,
-      "grad_norm": 1.5887471437454224,
-      "learning_rate": 5.192964824120603e-05,
-      "loss": 4.6782,
-      "step": 48348
-    },
-    {
-      "epoch": 25.214602346805737,
-      "grad_norm": 1.5337222814559937,
-      "learning_rate": 5.192864321608041e-05,
-      "loss": 5.0075,
-      "step": 48349
-    },
-    {
-      "epoch": 25.215123859191657,
-      "grad_norm": 1.489263653755188,
-      "learning_rate": 5.1927638190954774e-05,
-      "loss": 5.3088,
-      "step": 48350
-    },
-    {
-      "epoch": 25.215645371577576,
-      "grad_norm": 1.4759835004806519,
-      "learning_rate": 5.192663316582915e-05,
-      "loss": 5.4859,
-      "step": 48351
-    },
-    {
-      "epoch": 25.216166883963496,
-      "grad_norm": 1.6131857633590698,
-      "learning_rate": 5.192562814070352e-05,
-      "loss": 5.2151,
-      "step": 48352
-    },
-    {
-      "epoch": 25.216688396349415,
-      "grad_norm": 1.465651512145996,
-      "learning_rate": 5.19246231155779e-05,
-      "loss": 5.3619,
-      "step": 48353
-    },
-    {
-      "epoch": 25.21720990873533,
-      "grad_norm": 1.4825241565704346,
-      "learning_rate": 5.1923618090452264e-05,
-      "loss": 5.7105,
-      "step": 48354
-    },
-    {
-      "epoch": 25.21773142112125,
-      "grad_norm": 1.6847013235092163,
-      "learning_rate": 5.192261306532663e-05,
-      "loss": 4.9435,
-      "step": 48355
-    },
-    {
-      "epoch": 25.21825293350717,
-      "grad_norm": 1.5060125589370728,
-      "learning_rate": 5.1921608040201007e-05,
-      "loss": 5.6148,
-      "step": 48356
-    },
-    {
-      "epoch": 25.21877444589309,
-      "grad_norm": 1.5414056777954102,
-      "learning_rate": 5.192060301507538e-05,
-      "loss": 5.5711,
-      "step": 48357
-    },
-    {
-      "epoch": 25.21929595827901,
-      "grad_norm": 1.555281400680542,
-      "learning_rate": 5.1919597989949755e-05,
-      "loss": 5.485,
-      "step": 48358
-    },
-    {
-      "epoch": 25.21981747066493,
-      "grad_norm": 1.6709152460098267,
-      "learning_rate": 5.191859296482412e-05,
-      "loss": 5.1475,
-      "step": 48359
-    },
-    {
-      "epoch": 25.220338983050848,
-      "grad_norm": 1.4349151849746704,
-      "learning_rate": 5.19175879396985e-05,
-      "loss": 5.5852,
-      "step": 48360
-    },
-    {
-      "epoch": 25.220860495436767,
-      "grad_norm": 1.6444754600524902,
-      "learning_rate": 5.191658291457286e-05,
-      "loss": 5.2679,
-      "step": 48361
-    },
-    {
-      "epoch": 25.221382007822687,
-      "grad_norm": 1.5464152097702026,
-      "learning_rate": 5.191557788944724e-05,
-      "loss": 5.53,
-      "step": 48362
-    },
-    {
-      "epoch": 25.221903520208606,
-      "grad_norm": 1.6001051664352417,
-      "learning_rate": 5.191457286432161e-05,
-      "loss": 5.5129,
-      "step": 48363
-    },
-    {
-      "epoch": 25.222425032594526,
-      "grad_norm": 1.5279535055160522,
-      "learning_rate": 5.191356783919599e-05,
-      "loss": 5.6252,
-      "step": 48364
-    },
-    {
-      "epoch": 25.22294654498044,
-      "grad_norm": 1.611327052116394,
-      "learning_rate": 5.191256281407035e-05,
-      "loss": 5.5115,
-      "step": 48365
-    },
-    {
-      "epoch": 25.22346805736636,
-      "grad_norm": 1.472090721130371,
-      "learning_rate": 5.191155778894473e-05,
-      "loss": 5.1272,
-      "step": 48366
-    },
-    {
-      "epoch": 25.22398956975228,
-      "grad_norm": 1.6048035621643066,
-      "learning_rate": 5.1910552763819095e-05,
-      "loss": 5.2369,
-      "step": 48367
-    },
-    {
-      "epoch": 25.2245110821382,
-      "grad_norm": 1.4507238864898682,
-      "learning_rate": 5.1909547738693466e-05,
-      "loss": 4.7864,
-      "step": 48368
-    },
-    {
-      "epoch": 25.22503259452412,
-      "grad_norm": 1.5571866035461426,
-      "learning_rate": 5.1908542713567844e-05,
-      "loss": 5.579,
-      "step": 48369
-    },
-    {
-      "epoch": 25.22555410691004,
-      "grad_norm": 1.5456912517547607,
-      "learning_rate": 5.190753768844221e-05,
-      "loss": 5.1095,
-      "step": 48370
-    },
-    {
-      "epoch": 25.22607561929596,
-      "grad_norm": 1.6036648750305176,
-      "learning_rate": 5.1906532663316586e-05,
-      "loss": 4.9969,
-      "step": 48371
-    },
-    {
-      "epoch": 25.226597131681878,
-      "grad_norm": 1.5595334768295288,
-      "learning_rate": 5.190552763819096e-05,
-      "loss": 5.2629,
-      "step": 48372
-    },
-    {
-      "epoch": 25.227118644067797,
-      "grad_norm": 1.6687716245651245,
-      "learning_rate": 5.1904522613065335e-05,
-      "loss": 5.3684,
-      "step": 48373
-    },
-    {
-      "epoch": 25.227640156453717,
-      "grad_norm": 1.4662269353866577,
-      "learning_rate": 5.19035175879397e-05,
-      "loss": 5.8045,
-      "step": 48374
-    },
-    {
-      "epoch": 25.228161668839636,
-      "grad_norm": 1.5624916553497314,
-      "learning_rate": 5.190251256281408e-05,
-      "loss": 5.7236,
-      "step": 48375
-    },
-    {
-      "epoch": 25.228683181225556,
-      "grad_norm": 1.5957906246185303,
-      "learning_rate": 5.190150753768844e-05,
-      "loss": 5.6514,
-      "step": 48376
-    },
-    {
-      "epoch": 25.22920469361147,
-      "grad_norm": 1.6163588762283325,
-      "learning_rate": 5.190050251256282e-05,
-      "loss": 5.049,
-      "step": 48377
-    },
-    {
-      "epoch": 25.22972620599739,
-      "grad_norm": 1.4608922004699707,
-      "learning_rate": 5.189949748743719e-05,
-      "loss": 5.3884,
-      "step": 48378
-    },
-    {
-      "epoch": 25.23024771838331,
-      "grad_norm": 1.5689085721969604,
-      "learning_rate": 5.189849246231157e-05,
-      "loss": 5.5234,
-      "step": 48379
-    },
-    {
-      "epoch": 25.23076923076923,
-      "grad_norm": 1.5280333757400513,
-      "learning_rate": 5.189748743718593e-05,
-      "loss": 5.4396,
-      "step": 48380
-    },
-    {
-      "epoch": 25.23129074315515,
-      "grad_norm": 1.6102114915847778,
-      "learning_rate": 5.1896482412060296e-05,
-      "loss": 4.5859,
-      "step": 48381
-    },
-    {
-      "epoch": 25.23181225554107,
-      "grad_norm": 1.5372856855392456,
-      "learning_rate": 5.1895477386934674e-05,
-      "loss": 5.5929,
-      "step": 48382
-    },
-    {
-      "epoch": 25.23233376792699,
-      "grad_norm": 1.6287471055984497,
-      "learning_rate": 5.1894472361809045e-05,
-      "loss": 5.2827,
-      "step": 48383
-    },
-    {
-      "epoch": 25.232855280312908,
-      "grad_norm": 1.5909171104431152,
-      "learning_rate": 5.189346733668342e-05,
-      "loss": 5.3572,
-      "step": 48384
-    },
-    {
-      "epoch": 25.233376792698827,
-      "grad_norm": 1.5753047466278076,
-      "learning_rate": 5.189246231155779e-05,
-      "loss": 5.0252,
-      "step": 48385
-    },
-    {
-      "epoch": 25.233898305084747,
-      "grad_norm": 1.6671885251998901,
-      "learning_rate": 5.1891457286432165e-05,
-      "loss": 5.1712,
-      "step": 48386
-    },
-    {
-      "epoch": 25.234419817470666,
-      "grad_norm": 1.5014052391052246,
-      "learning_rate": 5.189045226130653e-05,
-      "loss": 4.9431,
-      "step": 48387
-    },
-    {
-      "epoch": 25.234941329856586,
-      "grad_norm": 1.730507493019104,
-      "learning_rate": 5.188944723618091e-05,
-      "loss": 5.2172,
-      "step": 48388
-    },
-    {
-      "epoch": 25.2354628422425,
-      "grad_norm": 1.5905430316925049,
-      "learning_rate": 5.188844221105528e-05,
-      "loss": 5.2787,
-      "step": 48389
-    },
-    {
-      "epoch": 25.23598435462842,
-      "grad_norm": 1.6713426113128662,
-      "learning_rate": 5.1887437185929656e-05,
-      "loss": 5.012,
-      "step": 48390
-    },
-    {
-      "epoch": 25.23650586701434,
-      "grad_norm": 1.545896291732788,
-      "learning_rate": 5.188643216080402e-05,
-      "loss": 4.9777,
-      "step": 48391
-    },
-    {
-      "epoch": 25.23702737940026,
-      "grad_norm": 1.4265631437301636,
-      "learning_rate": 5.18854271356784e-05,
-      "loss": 5.7121,
-      "step": 48392
-    },
-    {
-      "epoch": 25.23754889178618,
-      "grad_norm": 1.5716664791107178,
-      "learning_rate": 5.188442211055277e-05,
-      "loss": 5.1689,
-      "step": 48393
-    },
-    {
-      "epoch": 25.2380704041721,
-      "grad_norm": 1.7393176555633545,
-      "learning_rate": 5.1883417085427134e-05,
-      "loss": 4.2449,
-      "step": 48394
-    },
-    {
-      "epoch": 25.23859191655802,
-      "grad_norm": 1.6336759328842163,
-      "learning_rate": 5.188241206030151e-05,
-      "loss": 4.8914,
-      "step": 48395
-    },
-    {
-      "epoch": 25.239113428943938,
-      "grad_norm": 1.5827322006225586,
-      "learning_rate": 5.1881407035175876e-05,
-      "loss": 5.0529,
-      "step": 48396
-    },
-    {
-      "epoch": 25.239634941329857,
-      "grad_norm": 1.651581048965454,
-      "learning_rate": 5.1880402010050254e-05,
-      "loss": 5.2489,
-      "step": 48397
-    },
-    {
-      "epoch": 25.240156453715777,
-      "grad_norm": 1.474347710609436,
-      "learning_rate": 5.1879396984924625e-05,
-      "loss": 5.3414,
-      "step": 48398
-    },
-    {
-      "epoch": 25.240677966101696,
-      "grad_norm": 1.590214490890503,
-      "learning_rate": 5.1878391959799e-05,
-      "loss": 5.3885,
-      "step": 48399
-    },
-    {
-      "epoch": 25.241199478487616,
-      "grad_norm": 1.6012526750564575,
-      "learning_rate": 5.187738693467337e-05,
-      "loss": 5.0981,
-      "step": 48400
-    },
-    {
-      "epoch": 25.24172099087353,
-      "grad_norm": 1.6336185932159424,
-      "learning_rate": 5.1876381909547745e-05,
-      "loss": 5.2496,
-      "step": 48401
-    },
-    {
-      "epoch": 25.24224250325945,
-      "grad_norm": 1.4584846496582031,
-      "learning_rate": 5.187537688442211e-05,
-      "loss": 5.4841,
-      "step": 48402
-    },
-    {
-      "epoch": 25.24276401564537,
-      "grad_norm": 1.6554908752441406,
-      "learning_rate": 5.187437185929649e-05,
-      "loss": 4.8691,
-      "step": 48403
-    },
-    {
-      "epoch": 25.24328552803129,
-      "grad_norm": 1.5323480367660522,
-      "learning_rate": 5.187336683417086e-05,
-      "loss": 5.6671,
-      "step": 48404
-    },
-    {
-      "epoch": 25.24380704041721,
-      "grad_norm": 1.5569689273834229,
-      "learning_rate": 5.1872361809045236e-05,
-      "loss": 5.3156,
-      "step": 48405
-    },
-    {
-      "epoch": 25.24432855280313,
-      "grad_norm": 1.5573638677597046,
-      "learning_rate": 5.18713567839196e-05,
-      "loss": 5.7277,
-      "step": 48406
-    },
-    {
-      "epoch": 25.24485006518905,
-      "grad_norm": 1.5741711854934692,
-      "learning_rate": 5.1870351758793964e-05,
-      "loss": 5.4018,
-      "step": 48407
-    },
-    {
-      "epoch": 25.245371577574968,
-      "grad_norm": 1.660439372062683,
-      "learning_rate": 5.186934673366834e-05,
-      "loss": 4.9323,
-      "step": 48408
-    },
-    {
-      "epoch": 25.245893089960887,
-      "grad_norm": 1.547269344329834,
-      "learning_rate": 5.186834170854271e-05,
-      "loss": 5.0701,
-      "step": 48409
-    },
-    {
-      "epoch": 25.246414602346807,
-      "grad_norm": 1.5055551528930664,
-      "learning_rate": 5.186733668341709e-05,
-      "loss": 5.3178,
-      "step": 48410
-    },
-    {
-      "epoch": 25.246936114732726,
-      "grad_norm": 1.537407398223877,
-      "learning_rate": 5.1866331658291455e-05,
-      "loss": 4.9559,
-      "step": 48411
-    },
-    {
-      "epoch": 25.247457627118646,
-      "grad_norm": 1.5306118726730347,
-      "learning_rate": 5.186532663316583e-05,
-      "loss": 5.0451,
-      "step": 48412
-    },
-    {
-      "epoch": 25.24797913950456,
-      "grad_norm": 1.4959690570831299,
-      "learning_rate": 5.18643216080402e-05,
-      "loss": 4.9265,
-      "step": 48413
-    },
-    {
-      "epoch": 25.24850065189048,
-      "grad_norm": 1.5236923694610596,
-      "learning_rate": 5.1863316582914575e-05,
-      "loss": 4.8895,
-      "step": 48414
-    },
-    {
-      "epoch": 25.2490221642764,
-      "grad_norm": 1.6606241464614868,
-      "learning_rate": 5.1862311557788946e-05,
-      "loss": 4.8309,
-      "step": 48415
-    },
-    {
-      "epoch": 25.24954367666232,
-      "grad_norm": 1.5386115312576294,
-      "learning_rate": 5.1861306532663324e-05,
-      "loss": 4.7996,
-      "step": 48416
-    },
-    {
-      "epoch": 25.25006518904824,
-      "grad_norm": 1.6200599670410156,
-      "learning_rate": 5.186030150753769e-05,
-      "loss": 5.3818,
-      "step": 48417
-    },
-    {
-      "epoch": 25.25058670143416,
-      "grad_norm": 1.5998432636260986,
-      "learning_rate": 5.1859296482412066e-05,
-      "loss": 5.1543,
-      "step": 48418
-    },
-    {
-      "epoch": 25.25110821382008,
-      "grad_norm": 1.7899843454360962,
-      "learning_rate": 5.185829145728644e-05,
-      "loss": 4.9071,
-      "step": 48419
-    },
-    {
-      "epoch": 25.251629726205998,
-      "grad_norm": 1.6064870357513428,
-      "learning_rate": 5.18572864321608e-05,
-      "loss": 5.3772,
-      "step": 48420
-    },
-    {
-      "epoch": 25.252151238591917,
-      "grad_norm": 1.5558159351348877,
-      "learning_rate": 5.185628140703518e-05,
-      "loss": 4.9196,
-      "step": 48421
-    },
-    {
-      "epoch": 25.252672750977837,
-      "grad_norm": 1.6635066270828247,
-      "learning_rate": 5.1855276381909543e-05,
-      "loss": 5.0102,
-      "step": 48422
-    },
-    {
-      "epoch": 25.253194263363756,
-      "grad_norm": 1.5834736824035645,
-      "learning_rate": 5.185427135678392e-05,
-      "loss": 5.5086,
-      "step": 48423
-    },
-    {
-      "epoch": 25.253715775749676,
-      "grad_norm": 1.6167480945587158,
-      "learning_rate": 5.185326633165829e-05,
-      "loss": 5.1906,
-      "step": 48424
-    },
-    {
-      "epoch": 25.25423728813559,
-      "grad_norm": 1.5333375930786133,
-      "learning_rate": 5.185226130653267e-05,
-      "loss": 5.1787,
-      "step": 48425
-    },
-    {
-      "epoch": 25.25475880052151,
-      "grad_norm": 1.4776060581207275,
-      "learning_rate": 5.1851256281407034e-05,
-      "loss": 5.2085,
-      "step": 48426
-    },
-    {
-      "epoch": 25.25528031290743,
-      "grad_norm": 1.654433012008667,
-      "learning_rate": 5.185025125628141e-05,
-      "loss": 4.9291,
-      "step": 48427
-    },
-    {
-      "epoch": 25.25580182529335,
-      "grad_norm": 1.5435508489608765,
-      "learning_rate": 5.1849246231155777e-05,
-      "loss": 5.2624,
-      "step": 48428
-    },
-    {
-      "epoch": 25.25632333767927,
-      "grad_norm": 1.5785573720932007,
-      "learning_rate": 5.1848241206030154e-05,
-      "loss": 5.507,
-      "step": 48429
-    },
-    {
-      "epoch": 25.25684485006519,
-      "grad_norm": 1.5511677265167236,
-      "learning_rate": 5.1847236180904525e-05,
-      "loss": 5.3474,
-      "step": 48430
-    },
-    {
-      "epoch": 25.25736636245111,
-      "grad_norm": 1.5418386459350586,
-      "learning_rate": 5.18462311557789e-05,
-      "loss": 5.0291,
-      "step": 48431
-    },
-    {
-      "epoch": 25.257887874837028,
-      "grad_norm": 1.6104434728622437,
-      "learning_rate": 5.184522613065327e-05,
-      "loss": 4.8576,
-      "step": 48432
-    },
-    {
-      "epoch": 25.258409387222947,
-      "grad_norm": 1.6678651571273804,
-      "learning_rate": 5.1844221105527645e-05,
-      "loss": 5.2512,
-      "step": 48433
-    },
-    {
-      "epoch": 25.258930899608867,
-      "grad_norm": 1.6337233781814575,
-      "learning_rate": 5.184321608040201e-05,
-      "loss": 5.133,
-      "step": 48434
-    },
-    {
-      "epoch": 25.259452411994786,
-      "grad_norm": 1.6016641855239868,
-      "learning_rate": 5.184221105527638e-05,
-      "loss": 5.1501,
-      "step": 48435
-    },
-    {
-      "epoch": 25.259973924380706,
-      "grad_norm": 1.6046007871627808,
-      "learning_rate": 5.184120603015076e-05,
-      "loss": 5.2444,
-      "step": 48436
-    },
-    {
-      "epoch": 25.26049543676662,
-      "grad_norm": 1.5911872386932373,
-      "learning_rate": 5.184020100502512e-05,
-      "loss": 5.0116,
-      "step": 48437
-    },
-    {
-      "epoch": 25.26101694915254,
-      "grad_norm": 1.5764108896255493,
-      "learning_rate": 5.18391959798995e-05,
-      "loss": 5.4207,
-      "step": 48438
-    },
-    {
-      "epoch": 25.26153846153846,
-      "grad_norm": 1.498425841331482,
-      "learning_rate": 5.183819095477387e-05,
-      "loss": 5.3714,
-      "step": 48439
-    },
-    {
-      "epoch": 25.26205997392438,
-      "grad_norm": 1.8555607795715332,
-      "learning_rate": 5.183718592964825e-05,
-      "loss": 5.1469,
-      "step": 48440
-    },
-    {
-      "epoch": 25.2625814863103,
-      "grad_norm": 1.4443224668502808,
-      "learning_rate": 5.1836180904522614e-05,
-      "loss": 5.4355,
-      "step": 48441
-    },
-    {
-      "epoch": 25.26310299869622,
-      "grad_norm": 1.5884778499603271,
-      "learning_rate": 5.183517587939699e-05,
-      "loss": 5.3884,
-      "step": 48442
-    },
-    {
-      "epoch": 25.26362451108214,
-      "grad_norm": 1.5203505754470825,
-      "learning_rate": 5.1834170854271356e-05,
-      "loss": 5.256,
-      "step": 48443
-    },
-    {
-      "epoch": 25.264146023468058,
-      "grad_norm": 1.526780366897583,
-      "learning_rate": 5.1833165829145734e-05,
-      "loss": 5.5029,
-      "step": 48444
-    },
-    {
-      "epoch": 25.264667535853977,
-      "grad_norm": 1.5274556875228882,
-      "learning_rate": 5.1832160804020105e-05,
-      "loss": 5.3035,
-      "step": 48445
-    },
-    {
-      "epoch": 25.265189048239897,
-      "grad_norm": 1.662408471107483,
-      "learning_rate": 5.183115577889448e-05,
-      "loss": 5.3502,
-      "step": 48446
-    },
-    {
-      "epoch": 25.265710560625816,
-      "grad_norm": 1.4956525564193726,
-      "learning_rate": 5.183015075376885e-05,
-      "loss": 4.777,
-      "step": 48447
-    },
-    {
-      "epoch": 25.266232073011736,
-      "grad_norm": 1.5459717512130737,
-      "learning_rate": 5.182914572864321e-05,
-      "loss": 5.2627,
-      "step": 48448
-    },
-    {
-      "epoch": 25.26675358539765,
-      "grad_norm": 1.4863349199295044,
-      "learning_rate": 5.182814070351759e-05,
-      "loss": 5.1364,
-      "step": 48449
-    },
-    {
-      "epoch": 25.26727509778357,
-      "grad_norm": 1.6058123111724854,
-      "learning_rate": 5.182713567839196e-05,
-      "loss": 5.2122,
-      "step": 48450
-    },
-    {
-      "epoch": 25.26779661016949,
-      "grad_norm": 1.6187435388565063,
-      "learning_rate": 5.182613065326634e-05,
-      "loss": 5.5057,
-      "step": 48451
-    },
-    {
-      "epoch": 25.26831812255541,
-      "grad_norm": 1.6572860479354858,
-      "learning_rate": 5.18251256281407e-05,
-      "loss": 5.5885,
-      "step": 48452
-    },
-    {
-      "epoch": 25.26883963494133,
-      "grad_norm": 1.5660463571548462,
-      "learning_rate": 5.182412060301508e-05,
-      "loss": 5.0988,
-      "step": 48453
-    },
-    {
-      "epoch": 25.26936114732725,
-      "grad_norm": 1.6568342447280884,
-      "learning_rate": 5.1823115577889444e-05,
-      "loss": 5.0688,
-      "step": 48454
-    },
-    {
-      "epoch": 25.26988265971317,
-      "grad_norm": 1.5923150777816772,
-      "learning_rate": 5.182211055276382e-05,
-      "loss": 5.3963,
-      "step": 48455
-    },
-    {
-      "epoch": 25.270404172099088,
-      "grad_norm": 1.5179517269134521,
-      "learning_rate": 5.182110552763819e-05,
-      "loss": 5.4883,
-      "step": 48456
-    },
-    {
-      "epoch": 25.270925684485007,
-      "grad_norm": 1.649361491203308,
-      "learning_rate": 5.182010050251257e-05,
-      "loss": 5.3358,
-      "step": 48457
-    },
-    {
-      "epoch": 25.271447196870927,
-      "grad_norm": 1.7448351383209229,
-      "learning_rate": 5.1819095477386935e-05,
-      "loss": 5.2355,
-      "step": 48458
-    },
-    {
-      "epoch": 25.271968709256846,
-      "grad_norm": 1.5480570793151855,
-      "learning_rate": 5.181809045226131e-05,
-      "loss": 5.277,
-      "step": 48459
-    },
-    {
-      "epoch": 25.272490221642762,
-      "grad_norm": 1.5755810737609863,
-      "learning_rate": 5.1817085427135684e-05,
-      "loss": 5.1142,
-      "step": 48460
-    },
-    {
-      "epoch": 25.27301173402868,
-      "grad_norm": 1.6002317667007446,
-      "learning_rate": 5.181608040201005e-05,
-      "loss": 5.0888,
-      "step": 48461
-    },
-    {
-      "epoch": 25.2735332464146,
-      "grad_norm": 1.6315646171569824,
-      "learning_rate": 5.1815075376884426e-05,
-      "loss": 5.0488,
-      "step": 48462
-    },
-    {
-      "epoch": 25.27405475880052,
-      "grad_norm": 1.5270779132843018,
-      "learning_rate": 5.181407035175879e-05,
-      "loss": 4.6309,
-      "step": 48463
-    },
-    {
-      "epoch": 25.27457627118644,
-      "grad_norm": 1.5996840000152588,
-      "learning_rate": 5.181306532663317e-05,
-      "loss": 4.7386,
-      "step": 48464
-    },
-    {
-      "epoch": 25.27509778357236,
-      "grad_norm": 1.5075407028198242,
-      "learning_rate": 5.181206030150754e-05,
-      "loss": 5.0035,
-      "step": 48465
-    },
-    {
-      "epoch": 25.27561929595828,
-      "grad_norm": 1.5242680311203003,
-      "learning_rate": 5.181105527638192e-05,
-      "loss": 4.5605,
-      "step": 48466
-    },
-    {
-      "epoch": 25.2761408083442,
-      "grad_norm": 1.4950236082077026,
-      "learning_rate": 5.181005025125628e-05,
-      "loss": 5.1608,
-      "step": 48467
-    },
-    {
-      "epoch": 25.276662320730118,
-      "grad_norm": 1.6069222688674927,
-      "learning_rate": 5.180904522613066e-05,
-      "loss": 4.7771,
-      "step": 48468
-    },
-    {
-      "epoch": 25.277183833116037,
-      "grad_norm": 1.564070463180542,
-      "learning_rate": 5.1808040201005024e-05,
-      "loss": 5.6196,
-      "step": 48469
-    },
-    {
-      "epoch": 25.277705345501957,
-      "grad_norm": 1.691686987876892,
-      "learning_rate": 5.18070351758794e-05,
-      "loss": 5.0175,
-      "step": 48470
-    },
-    {
-      "epoch": 25.278226857887876,
-      "grad_norm": 1.7313019037246704,
-      "learning_rate": 5.180603015075377e-05,
-      "loss": 5.1403,
-      "step": 48471
-    },
-    {
-      "epoch": 25.278748370273792,
-      "grad_norm": 1.5979009866714478,
-      "learning_rate": 5.180502512562815e-05,
-      "loss": 5.2703,
-      "step": 48472
-    },
-    {
-      "epoch": 25.27926988265971,
-      "grad_norm": 1.7067283391952515,
-      "learning_rate": 5.1804020100502515e-05,
-      "loss": 4.9935,
-      "step": 48473
-    },
-    {
-      "epoch": 25.27979139504563,
-      "grad_norm": 1.604418158531189,
-      "learning_rate": 5.180301507537688e-05,
-      "loss": 4.9872,
-      "step": 48474
-    },
-    {
-      "epoch": 25.28031290743155,
-      "grad_norm": 1.525961995124817,
-      "learning_rate": 5.1802010050251257e-05,
-      "loss": 5.2086,
-      "step": 48475
-    },
-    {
-      "epoch": 25.28083441981747,
-      "grad_norm": 1.5795907974243164,
-      "learning_rate": 5.180100502512563e-05,
-      "loss": 4.8225,
-      "step": 48476
-    },
-    {
-      "epoch": 25.28135593220339,
-      "grad_norm": 1.6977088451385498,
-      "learning_rate": 5.1800000000000005e-05,
-      "loss": 5.0366,
-      "step": 48477
-    },
-    {
-      "epoch": 25.28187744458931,
-      "grad_norm": 1.6625896692276,
-      "learning_rate": 5.179899497487437e-05,
-      "loss": 5.3092,
-      "step": 48478
-    },
-    {
-      "epoch": 25.28239895697523,
-      "grad_norm": 1.5034955739974976,
-      "learning_rate": 5.179798994974875e-05,
-      "loss": 5.3432,
-      "step": 48479
-    },
-    {
-      "epoch": 25.282920469361148,
-      "grad_norm": 1.5685173273086548,
-      "learning_rate": 5.179698492462312e-05,
-      "loss": 5.4335,
-      "step": 48480
-    },
-    {
-      "epoch": 25.283441981747067,
-      "grad_norm": 1.5448129177093506,
-      "learning_rate": 5.1795979899497496e-05,
-      "loss": 5.5967,
-      "step": 48481
-    },
-    {
-      "epoch": 25.283963494132987,
-      "grad_norm": 1.5207968950271606,
-      "learning_rate": 5.179497487437186e-05,
-      "loss": 5.3052,
-      "step": 48482
-    },
-    {
-      "epoch": 25.284485006518906,
-      "grad_norm": 1.574852466583252,
-      "learning_rate": 5.179396984924624e-05,
-      "loss": 5.2387,
-      "step": 48483
-    },
-    {
-      "epoch": 25.285006518904822,
-      "grad_norm": 1.5602384805679321,
-      "learning_rate": 5.17929648241206e-05,
-      "loss": 4.7163,
-      "step": 48484
-    },
-    {
-      "epoch": 25.285528031290742,
-      "grad_norm": 1.8110796213150024,
-      "learning_rate": 5.179195979899498e-05,
-      "loss": 4.6229,
-      "step": 48485
-    },
-    {
-      "epoch": 25.28604954367666,
-      "grad_norm": 1.6231932640075684,
-      "learning_rate": 5.179095477386935e-05,
-      "loss": 4.975,
-      "step": 48486
-    },
-    {
-      "epoch": 25.28657105606258,
-      "grad_norm": 1.6373850107192993,
-      "learning_rate": 5.1789949748743716e-05,
-      "loss": 5.1001,
-      "step": 48487
-    },
-    {
-      "epoch": 25.2870925684485,
-      "grad_norm": 1.5988361835479736,
-      "learning_rate": 5.1788944723618094e-05,
-      "loss": 5.2723,
-      "step": 48488
-    },
-    {
-      "epoch": 25.28761408083442,
-      "grad_norm": 1.5776691436767578,
-      "learning_rate": 5.178793969849246e-05,
-      "loss": 5.5555,
-      "step": 48489
-    },
-    {
-      "epoch": 25.28813559322034,
-      "grad_norm": 1.6307975053787231,
-      "learning_rate": 5.1786934673366836e-05,
-      "loss": 5.2777,
-      "step": 48490
-    },
-    {
-      "epoch": 25.28865710560626,
-      "grad_norm": 1.6178057193756104,
-      "learning_rate": 5.178592964824121e-05,
-      "loss": 5.4523,
-      "step": 48491
-    },
-    {
-      "epoch": 25.289178617992178,
-      "grad_norm": 1.5960419178009033,
-      "learning_rate": 5.1784924623115585e-05,
-      "loss": 5.7774,
-      "step": 48492
-    },
-    {
-      "epoch": 25.289700130378097,
-      "grad_norm": 1.537475347518921,
-      "learning_rate": 5.178391959798995e-05,
-      "loss": 5.5407,
-      "step": 48493
-    },
-    {
-      "epoch": 25.290221642764017,
-      "grad_norm": 1.5164333581924438,
-      "learning_rate": 5.178291457286433e-05,
-      "loss": 5.6954,
-      "step": 48494
-    },
-    {
-      "epoch": 25.290743155149936,
-      "grad_norm": 1.5010545253753662,
-      "learning_rate": 5.178190954773869e-05,
-      "loss": 5.2889,
-      "step": 48495
-    },
-    {
-      "epoch": 25.291264667535852,
-      "grad_norm": 1.6553062200546265,
-      "learning_rate": 5.178090452261307e-05,
-      "loss": 4.7543,
-      "step": 48496
-    },
-    {
-      "epoch": 25.291786179921772,
-      "grad_norm": 1.5252145528793335,
-      "learning_rate": 5.177989949748744e-05,
-      "loss": 5.1668,
-      "step": 48497
-    },
-    {
-      "epoch": 25.29230769230769,
-      "grad_norm": 1.6622825860977173,
-      "learning_rate": 5.177889447236182e-05,
-      "loss": 4.4929,
-      "step": 48498
-    },
-    {
-      "epoch": 25.29282920469361,
-      "grad_norm": 1.5770264863967896,
-      "learning_rate": 5.177788944723618e-05,
-      "loss": 5.1255,
-      "step": 48499
-    },
-    {
-      "epoch": 25.29335071707953,
-      "grad_norm": 1.551990270614624,
-      "learning_rate": 5.1776884422110546e-05,
-      "loss": 5.7754,
-      "step": 48500
-    },
-    {
-      "epoch": 25.29387222946545,
-      "grad_norm": 1.4943437576293945,
-      "learning_rate": 5.1775879396984924e-05,
-      "loss": 5.2239,
-      "step": 48501
-    },
-    {
-      "epoch": 25.29439374185137,
-      "grad_norm": 1.563147783279419,
-      "learning_rate": 5.1774874371859295e-05,
-      "loss": 5.2623,
-      "step": 48502
-    },
-    {
-      "epoch": 25.29491525423729,
-      "grad_norm": 1.5493098497390747,
-      "learning_rate": 5.177386934673367e-05,
-      "loss": 4.8754,
-      "step": 48503
-    },
-    {
-      "epoch": 25.295436766623208,
-      "grad_norm": 1.5841995477676392,
-      "learning_rate": 5.177286432160804e-05,
-      "loss": 5.1965,
-      "step": 48504
-    },
-    {
-      "epoch": 25.295958279009128,
-      "grad_norm": 1.6192734241485596,
-      "learning_rate": 5.1771859296482415e-05,
-      "loss": 5.0682,
-      "step": 48505
-    },
-    {
-      "epoch": 25.296479791395047,
-      "grad_norm": 1.617760181427002,
-      "learning_rate": 5.1770854271356786e-05,
-      "loss": 5.0013,
-      "step": 48506
-    },
-    {
-      "epoch": 25.297001303780966,
-      "grad_norm": 1.559918999671936,
-      "learning_rate": 5.1769849246231164e-05,
-      "loss": 5.3393,
-      "step": 48507
-    },
-    {
-      "epoch": 25.297522816166882,
-      "grad_norm": 1.5646215677261353,
-      "learning_rate": 5.176884422110553e-05,
-      "loss": 5.1592,
-      "step": 48508
-    },
-    {
-      "epoch": 25.298044328552802,
-      "grad_norm": 1.565974473953247,
-      "learning_rate": 5.1767839195979906e-05,
-      "loss": 5.3576,
-      "step": 48509
-    },
-    {
-      "epoch": 25.29856584093872,
-      "grad_norm": 1.6343655586242676,
-      "learning_rate": 5.176683417085427e-05,
-      "loss": 5.0379,
-      "step": 48510
-    },
-    {
-      "epoch": 25.29908735332464,
-      "grad_norm": 1.516369104385376,
-      "learning_rate": 5.176582914572865e-05,
-      "loss": 5.1983,
-      "step": 48511
-    },
-    {
-      "epoch": 25.29960886571056,
-      "grad_norm": 1.6975189447402954,
-      "learning_rate": 5.176482412060302e-05,
-      "loss": 4.6323,
-      "step": 48512
-    },
-    {
-      "epoch": 25.30013037809648,
-      "grad_norm": 1.6835877895355225,
-      "learning_rate": 5.17638190954774e-05,
-      "loss": 4.9477,
-      "step": 48513
-    },
-    {
-      "epoch": 25.3006518904824,
-      "grad_norm": 1.602736473083496,
-      "learning_rate": 5.176281407035176e-05,
-      "loss": 5.3148,
-      "step": 48514
-    },
-    {
-      "epoch": 25.30117340286832,
-      "grad_norm": 1.5342170000076294,
-      "learning_rate": 5.1761809045226126e-05,
-      "loss": 4.976,
-      "step": 48515
-    },
-    {
-      "epoch": 25.301694915254238,
-      "grad_norm": 1.565859079360962,
-      "learning_rate": 5.1760804020100504e-05,
-      "loss": 4.8576,
-      "step": 48516
-    },
-    {
-      "epoch": 25.302216427640158,
-      "grad_norm": 1.6106871366500854,
-      "learning_rate": 5.1759798994974875e-05,
-      "loss": 4.9449,
-      "step": 48517
-    },
-    {
-      "epoch": 25.302737940026077,
-      "grad_norm": 1.6186937093734741,
-      "learning_rate": 5.175879396984925e-05,
-      "loss": 5.4573,
-      "step": 48518
-    },
-    {
-      "epoch": 25.303259452411996,
-      "grad_norm": 1.5835356712341309,
-      "learning_rate": 5.175778894472362e-05,
-      "loss": 5.0793,
-      "step": 48519
-    },
-    {
-      "epoch": 25.303780964797912,
-      "grad_norm": 1.6648069620132446,
-      "learning_rate": 5.1756783919597995e-05,
-      "loss": 4.7163,
-      "step": 48520
-    },
-    {
-      "epoch": 25.304302477183832,
-      "grad_norm": 1.529647946357727,
-      "learning_rate": 5.175577889447236e-05,
-      "loss": 5.2642,
-      "step": 48521
-    },
-    {
-      "epoch": 25.30482398956975,
-      "grad_norm": 1.667874813079834,
-      "learning_rate": 5.175477386934674e-05,
-      "loss": 5.09,
-      "step": 48522
-    },
-    {
-      "epoch": 25.30534550195567,
-      "grad_norm": 1.5039567947387695,
-      "learning_rate": 5.175376884422111e-05,
-      "loss": 4.975,
-      "step": 48523
-    },
-    {
-      "epoch": 25.30586701434159,
-      "grad_norm": 1.5900615453720093,
-      "learning_rate": 5.1752763819095486e-05,
-      "loss": 5.205,
-      "step": 48524
-    },
-    {
-      "epoch": 25.30638852672751,
-      "grad_norm": 1.6796729564666748,
-      "learning_rate": 5.175175879396985e-05,
-      "loss": 5.0242,
-      "step": 48525
-    },
-    {
-      "epoch": 25.30691003911343,
-      "grad_norm": 1.4709261655807495,
-      "learning_rate": 5.175075376884423e-05,
-      "loss": 5.6235,
-      "step": 48526
-    },
-    {
-      "epoch": 25.30743155149935,
-      "grad_norm": 1.5820796489715576,
-      "learning_rate": 5.17497487437186e-05,
-      "loss": 5.4441,
-      "step": 48527
-    },
-    {
-      "epoch": 25.307953063885268,
-      "grad_norm": 1.6090680360794067,
-      "learning_rate": 5.174874371859296e-05,
-      "loss": 5.1629,
-      "step": 48528
-    },
-    {
-      "epoch": 25.308474576271188,
-      "grad_norm": 1.615013837814331,
-      "learning_rate": 5.174773869346734e-05,
-      "loss": 5.6151,
-      "step": 48529
-    },
-    {
-      "epoch": 25.308996088657107,
-      "grad_norm": 1.6915628910064697,
-      "learning_rate": 5.1746733668341705e-05,
-      "loss": 5.1688,
-      "step": 48530
-    },
-    {
-      "epoch": 25.309517601043027,
-      "grad_norm": 1.5410512685775757,
-      "learning_rate": 5.174572864321608e-05,
-      "loss": 4.6584,
-      "step": 48531
-    },
-    {
-      "epoch": 25.310039113428942,
-      "grad_norm": 1.584467887878418,
-      "learning_rate": 5.1744723618090454e-05,
-      "loss": 5.0208,
-      "step": 48532
-    },
-    {
-      "epoch": 25.310560625814862,
-      "grad_norm": 1.6026755571365356,
-      "learning_rate": 5.174371859296483e-05,
-      "loss": 5.3615,
-      "step": 48533
-    },
-    {
-      "epoch": 25.31108213820078,
-      "grad_norm": 1.6326016187667847,
-      "learning_rate": 5.1742713567839196e-05,
-      "loss": 5.1819,
-      "step": 48534
-    },
-    {
-      "epoch": 25.3116036505867,
-      "grad_norm": 1.6311193704605103,
-      "learning_rate": 5.1741708542713574e-05,
-      "loss": 5.2053,
-      "step": 48535
-    },
-    {
-      "epoch": 25.31212516297262,
-      "grad_norm": 1.5434054136276245,
-      "learning_rate": 5.174070351758794e-05,
-      "loss": 4.8803,
-      "step": 48536
-    },
-    {
-      "epoch": 25.31264667535854,
-      "grad_norm": 1.5282293558120728,
-      "learning_rate": 5.1739698492462316e-05,
-      "loss": 5.2057,
-      "step": 48537
-    },
-    {
-      "epoch": 25.31316818774446,
-      "grad_norm": 1.5535529851913452,
-      "learning_rate": 5.173869346733669e-05,
-      "loss": 5.1293,
-      "step": 48538
-    },
-    {
-      "epoch": 25.31368970013038,
-      "grad_norm": 1.5664138793945312,
-      "learning_rate": 5.1737688442211065e-05,
-      "loss": 5.2861,
-      "step": 48539
-    },
-    {
-      "epoch": 25.314211212516298,
-      "grad_norm": 1.6479696035385132,
-      "learning_rate": 5.173668341708543e-05,
-      "loss": 5.303,
-      "step": 48540
-    },
-    {
-      "epoch": 25.314732724902218,
-      "grad_norm": 1.5793291330337524,
-      "learning_rate": 5.1735678391959793e-05,
-      "loss": 5.5506,
-      "step": 48541
-    },
-    {
-      "epoch": 25.315254237288137,
-      "grad_norm": 1.6412948369979858,
-      "learning_rate": 5.173467336683417e-05,
-      "loss": 5.4989,
-      "step": 48542
-    },
-    {
-      "epoch": 25.315775749674053,
-      "grad_norm": 1.6319665908813477,
-      "learning_rate": 5.173366834170854e-05,
-      "loss": 4.638,
-      "step": 48543
-    },
-    {
-      "epoch": 25.316297262059972,
-      "grad_norm": 1.5958175659179688,
-      "learning_rate": 5.173266331658292e-05,
-      "loss": 5.4106,
-      "step": 48544
-    },
-    {
-      "epoch": 25.316818774445892,
-      "grad_norm": 1.431693196296692,
-      "learning_rate": 5.1731658291457284e-05,
-      "loss": 5.5933,
-      "step": 48545
-    },
-    {
-      "epoch": 25.31734028683181,
-      "grad_norm": 1.4389233589172363,
-      "learning_rate": 5.173065326633166e-05,
-      "loss": 4.5911,
-      "step": 48546
-    },
-    {
-      "epoch": 25.31786179921773,
-      "grad_norm": 1.549519658088684,
-      "learning_rate": 5.172964824120603e-05,
-      "loss": 5.3768,
-      "step": 48547
-    },
-    {
-      "epoch": 25.31838331160365,
-      "grad_norm": 1.499122977256775,
-      "learning_rate": 5.172864321608041e-05,
-      "loss": 5.126,
-      "step": 48548
-    },
-    {
-      "epoch": 25.31890482398957,
-      "grad_norm": 1.6422274112701416,
-      "learning_rate": 5.1727638190954775e-05,
-      "loss": 4.5145,
-      "step": 48549
-    },
-    {
-      "epoch": 25.31942633637549,
-      "grad_norm": 1.648422122001648,
-      "learning_rate": 5.172663316582915e-05,
-      "loss": 5.189,
-      "step": 48550
-    },
-    {
-      "epoch": 25.31994784876141,
-      "grad_norm": 1.6189335584640503,
-      "learning_rate": 5.172562814070352e-05,
-      "loss": 5.2951,
-      "step": 48551
-    },
-    {
-      "epoch": 25.320469361147328,
-      "grad_norm": 1.6178325414657593,
-      "learning_rate": 5.1724623115577895e-05,
-      "loss": 4.9212,
-      "step": 48552
-    },
-    {
-      "epoch": 25.320990873533248,
-      "grad_norm": 1.6704673767089844,
-      "learning_rate": 5.1723618090452266e-05,
-      "loss": 4.8445,
-      "step": 48553
-    },
-    {
-      "epoch": 25.321512385919167,
-      "grad_norm": 1.6343128681182861,
-      "learning_rate": 5.172261306532663e-05,
-      "loss": 4.8825,
-      "step": 48554
-    },
-    {
-      "epoch": 25.322033898305083,
-      "grad_norm": 1.5836892127990723,
-      "learning_rate": 5.172160804020101e-05,
-      "loss": 5.4805,
-      "step": 48555
-    },
-    {
-      "epoch": 25.322555410691002,
-      "grad_norm": 1.576979398727417,
-      "learning_rate": 5.172060301507537e-05,
-      "loss": 4.9929,
-      "step": 48556
-    },
-    {
-      "epoch": 25.323076923076922,
-      "grad_norm": 1.5217466354370117,
-      "learning_rate": 5.171959798994975e-05,
-      "loss": 5.5212,
-      "step": 48557
-    },
-    {
-      "epoch": 25.32359843546284,
-      "grad_norm": 1.6653550863265991,
-      "learning_rate": 5.171859296482412e-05,
-      "loss": 5.0531,
-      "step": 48558
-    },
-    {
-      "epoch": 25.32411994784876,
-      "grad_norm": 1.4931222200393677,
-      "learning_rate": 5.17175879396985e-05,
-      "loss": 5.66,
-      "step": 48559
-    },
-    {
-      "epoch": 25.32464146023468,
-      "grad_norm": 1.631942629814148,
-      "learning_rate": 5.1716582914572864e-05,
-      "loss": 5.2773,
-      "step": 48560
-    },
-    {
-      "epoch": 25.3251629726206,
-      "grad_norm": 1.6180286407470703,
-      "learning_rate": 5.171557788944724e-05,
-      "loss": 5.1307,
-      "step": 48561
-    },
-    {
-      "epoch": 25.32568448500652,
-      "grad_norm": 1.460673213005066,
-      "learning_rate": 5.1714572864321606e-05,
-      "loss": 5.2143,
-      "step": 48562
-    },
-    {
-      "epoch": 25.32620599739244,
-      "grad_norm": 1.4587324857711792,
-      "learning_rate": 5.1713567839195984e-05,
-      "loss": 5.5697,
-      "step": 48563
-    },
-    {
-      "epoch": 25.326727509778358,
-      "grad_norm": 1.7429648637771606,
-      "learning_rate": 5.1712562814070355e-05,
-      "loss": 4.9627,
-      "step": 48564
-    },
-    {
-      "epoch": 25.327249022164278,
-      "grad_norm": 1.558127999305725,
-      "learning_rate": 5.171155778894473e-05,
-      "loss": 5.1761,
-      "step": 48565
-    },
-    {
-      "epoch": 25.327770534550197,
-      "grad_norm": 1.5692847967147827,
-      "learning_rate": 5.17105527638191e-05,
-      "loss": 5.736,
-      "step": 48566
-    },
-    {
-      "epoch": 25.328292046936113,
-      "grad_norm": 1.5903745889663696,
-      "learning_rate": 5.170954773869347e-05,
-      "loss": 5.5077,
-      "step": 48567
-    },
-    {
-      "epoch": 25.328813559322032,
-      "grad_norm": 1.6701359748840332,
-      "learning_rate": 5.1708542713567846e-05,
-      "loss": 5.1529,
-      "step": 48568
-    },
-    {
-      "epoch": 25.329335071707952,
-      "grad_norm": 1.5912038087844849,
-      "learning_rate": 5.170753768844221e-05,
-      "loss": 4.9469,
-      "step": 48569
-    },
-    {
-      "epoch": 25.32985658409387,
-      "grad_norm": 1.5388301610946655,
-      "learning_rate": 5.170653266331659e-05,
-      "loss": 5.1329,
-      "step": 48570
-    },
-    {
-      "epoch": 25.33037809647979,
-      "grad_norm": 1.7346123456954956,
-      "learning_rate": 5.170552763819095e-05,
-      "loss": 5.4142,
-      "step": 48571
-    },
-    {
-      "epoch": 25.33089960886571,
-      "grad_norm": 1.5514754056930542,
-      "learning_rate": 5.170452261306533e-05,
-      "loss": 5.4307,
-      "step": 48572
-    },
-    {
-      "epoch": 25.33142112125163,
-      "grad_norm": 1.4606399536132812,
-      "learning_rate": 5.17035175879397e-05,
-      "loss": 5.1033,
-      "step": 48573
-    },
-    {
-      "epoch": 25.33194263363755,
-      "grad_norm": 1.5193415880203247,
-      "learning_rate": 5.170251256281408e-05,
-      "loss": 5.5417,
-      "step": 48574
-    },
-    {
-      "epoch": 25.33246414602347,
-      "grad_norm": 1.5797133445739746,
-      "learning_rate": 5.170150753768844e-05,
-      "loss": 5.3006,
-      "step": 48575
-    },
-    {
-      "epoch": 25.332985658409388,
-      "grad_norm": 1.6541471481323242,
-      "learning_rate": 5.170050251256282e-05,
-      "loss": 5.3281,
-      "step": 48576
-    },
-    {
-      "epoch": 25.333507170795308,
-      "grad_norm": 1.4554307460784912,
-      "learning_rate": 5.1699497487437185e-05,
-      "loss": 5.784,
-      "step": 48577
-    },
-    {
-      "epoch": 25.334028683181227,
-      "grad_norm": 1.6292798519134521,
-      "learning_rate": 5.169849246231156e-05,
-      "loss": 5.016,
-      "step": 48578
-    },
-    {
-      "epoch": 25.334550195567143,
-      "grad_norm": 1.5957454442977905,
-      "learning_rate": 5.1697487437185934e-05,
-      "loss": 4.7679,
-      "step": 48579
-    },
-    {
-      "epoch": 25.335071707953063,
-      "grad_norm": 1.6116067171096802,
-      "learning_rate": 5.16964824120603e-05,
-      "loss": 5.1894,
-      "step": 48580
-    },
-    {
-      "epoch": 25.335593220338982,
-      "grad_norm": 1.5571861267089844,
-      "learning_rate": 5.1695477386934676e-05,
-      "loss": 4.4191,
-      "step": 48581
-    },
-    {
-      "epoch": 25.3361147327249,
-      "grad_norm": 1.5321491956710815,
-      "learning_rate": 5.169447236180904e-05,
-      "loss": 5.3026,
-      "step": 48582
-    },
-    {
-      "epoch": 25.33663624511082,
-      "grad_norm": 1.6711622476577759,
-      "learning_rate": 5.169346733668342e-05,
-      "loss": 5.139,
-      "step": 48583
-    },
-    {
-      "epoch": 25.33715775749674,
-      "grad_norm": 1.6178046464920044,
-      "learning_rate": 5.169246231155779e-05,
-      "loss": 5.519,
-      "step": 48584
-    },
-    {
-      "epoch": 25.33767926988266,
-      "grad_norm": 1.65463387966156,
-      "learning_rate": 5.169145728643217e-05,
-      "loss": 4.5465,
-      "step": 48585
-    },
-    {
-      "epoch": 25.33820078226858,
-      "grad_norm": 1.632018804550171,
-      "learning_rate": 5.169045226130653e-05,
-      "loss": 5.1444,
-      "step": 48586
-    },
-    {
-      "epoch": 25.3387222946545,
-      "grad_norm": 1.6084651947021484,
-      "learning_rate": 5.168944723618091e-05,
-      "loss": 5.1865,
-      "step": 48587
-    },
-    {
-      "epoch": 25.33924380704042,
-      "grad_norm": 1.5377250909805298,
-      "learning_rate": 5.1688442211055274e-05,
-      "loss": 5.5557,
-      "step": 48588
-    },
-    {
-      "epoch": 25.339765319426338,
-      "grad_norm": 1.5767958164215088,
-      "learning_rate": 5.168743718592965e-05,
-      "loss": 4.7571,
-      "step": 48589
-    },
-    {
-      "epoch": 25.340286831812257,
-      "grad_norm": 1.5704752206802368,
-      "learning_rate": 5.168643216080402e-05,
-      "loss": 5.4442,
-      "step": 48590
-    },
-    {
-      "epoch": 25.340808344198173,
-      "grad_norm": 1.6449116468429565,
-      "learning_rate": 5.16854271356784e-05,
-      "loss": 4.9448,
-      "step": 48591
-    },
-    {
-      "epoch": 25.341329856584093,
-      "grad_norm": 1.5385998487472534,
-      "learning_rate": 5.1684422110552765e-05,
-      "loss": 5.2356,
-      "step": 48592
-    },
-    {
-      "epoch": 25.341851368970012,
-      "grad_norm": 1.5015718936920166,
-      "learning_rate": 5.1683417085427136e-05,
-      "loss": 5.4391,
-      "step": 48593
-    },
-    {
-      "epoch": 25.34237288135593,
-      "grad_norm": 1.5409080982208252,
-      "learning_rate": 5.1682412060301513e-05,
-      "loss": 5.6314,
-      "step": 48594
-    },
-    {
-      "epoch": 25.34289439374185,
-      "grad_norm": 1.5694299936294556,
-      "learning_rate": 5.168140703517588e-05,
-      "loss": 5.443,
-      "step": 48595
-    },
-    {
-      "epoch": 25.34341590612777,
-      "grad_norm": 1.59146249294281,
-      "learning_rate": 5.1680402010050256e-05,
-      "loss": 4.8418,
-      "step": 48596
-    },
-    {
-      "epoch": 25.34393741851369,
-      "grad_norm": 1.5509833097457886,
-      "learning_rate": 5.167939698492462e-05,
-      "loss": 5.0286,
-      "step": 48597
-    },
-    {
-      "epoch": 25.34445893089961,
-      "grad_norm": 1.453006386756897,
-      "learning_rate": 5.1678391959799e-05,
-      "loss": 5.2511,
-      "step": 48598
-    },
-    {
-      "epoch": 25.34498044328553,
-      "grad_norm": 1.5223796367645264,
-      "learning_rate": 5.167738693467337e-05,
-      "loss": 5.1157,
-      "step": 48599
-    },
-    {
-      "epoch": 25.34550195567145,
-      "grad_norm": 1.590415120124817,
-      "learning_rate": 5.1676381909547746e-05,
-      "loss": 5.4028,
-      "step": 48600
-    },
-    {
-      "epoch": 25.346023468057368,
-      "grad_norm": 1.6210602521896362,
-      "learning_rate": 5.167537688442211e-05,
-      "loss": 5.7086,
-      "step": 48601
-    },
-    {
-      "epoch": 25.346544980443287,
-      "grad_norm": 1.5685685873031616,
-      "learning_rate": 5.167437185929649e-05,
-      "loss": 5.1974,
-      "step": 48602
-    },
-    {
-      "epoch": 25.347066492829203,
-      "grad_norm": 1.6845428943634033,
-      "learning_rate": 5.167336683417085e-05,
-      "loss": 4.6046,
-      "step": 48603
-    },
-    {
-      "epoch": 25.347588005215123,
-      "grad_norm": 1.512067437171936,
-      "learning_rate": 5.167236180904523e-05,
-      "loss": 4.8504,
-      "step": 48604
-    },
-    {
-      "epoch": 25.348109517601042,
-      "grad_norm": 1.674498438835144,
-      "learning_rate": 5.16713567839196e-05,
-      "loss": 5.2789,
-      "step": 48605
-    },
-    {
-      "epoch": 25.34863102998696,
-      "grad_norm": 1.5788142681121826,
-      "learning_rate": 5.167035175879398e-05,
-      "loss": 5.5211,
-      "step": 48606
-    },
-    {
-      "epoch": 25.34915254237288,
-      "grad_norm": 1.537176489830017,
-      "learning_rate": 5.1669346733668344e-05,
-      "loss": 5.3042,
-      "step": 48607
-    },
-    {
-      "epoch": 25.3496740547588,
-      "grad_norm": 1.532775640487671,
-      "learning_rate": 5.166834170854271e-05,
-      "loss": 5.422,
-      "step": 48608
-    },
-    {
-      "epoch": 25.35019556714472,
-      "grad_norm": 1.5452570915222168,
-      "learning_rate": 5.1667336683417086e-05,
-      "loss": 4.8775,
-      "step": 48609
-    },
-    {
-      "epoch": 25.35071707953064,
-      "grad_norm": 1.6748077869415283,
-      "learning_rate": 5.166633165829146e-05,
-      "loss": 5.4251,
-      "step": 48610
-    },
-    {
-      "epoch": 25.35123859191656,
-      "grad_norm": 1.656619906425476,
-      "learning_rate": 5.1665326633165835e-05,
-      "loss": 5.5135,
-      "step": 48611
-    },
-    {
-      "epoch": 25.35176010430248,
-      "grad_norm": 1.6418235301971436,
-      "learning_rate": 5.16643216080402e-05,
-      "loss": 5.195,
-      "step": 48612
-    },
-    {
-      "epoch": 25.352281616688398,
-      "grad_norm": 1.5956307649612427,
-      "learning_rate": 5.166331658291458e-05,
-      "loss": 5.6853,
-      "step": 48613
-    },
-    {
-      "epoch": 25.352803129074317,
-      "grad_norm": 1.624460220336914,
-      "learning_rate": 5.166231155778895e-05,
-      "loss": 5.4303,
-      "step": 48614
-    },
-    {
-      "epoch": 25.353324641460233,
-      "grad_norm": 1.624199390411377,
-      "learning_rate": 5.1661306532663326e-05,
-      "loss": 5.2319,
-      "step": 48615
-    },
-    {
-      "epoch": 25.353846153846153,
-      "grad_norm": 1.6969722509384155,
-      "learning_rate": 5.166030150753769e-05,
-      "loss": 5.3331,
-      "step": 48616
-    },
-    {
-      "epoch": 25.354367666232072,
-      "grad_norm": 1.5088402032852173,
-      "learning_rate": 5.165929648241207e-05,
-      "loss": 5.0024,
-      "step": 48617
-    },
-    {
-      "epoch": 25.35488917861799,
-      "grad_norm": 1.520193099975586,
-      "learning_rate": 5.165829145728643e-05,
-      "loss": 4.9225,
-      "step": 48618
-    },
-    {
-      "epoch": 25.35541069100391,
-      "grad_norm": 1.5763914585113525,
-      "learning_rate": 5.165728643216081e-05,
-      "loss": 5.2928,
-      "step": 48619
-    },
-    {
-      "epoch": 25.35593220338983,
-      "grad_norm": 1.597982406616211,
-      "learning_rate": 5.165628140703518e-05,
-      "loss": 4.9562,
-      "step": 48620
-    },
-    {
-      "epoch": 25.35645371577575,
-      "grad_norm": 1.5913941860198975,
-      "learning_rate": 5.1655276381909545e-05,
-      "loss": 4.9361,
-      "step": 48621
-    },
-    {
-      "epoch": 25.35697522816167,
-      "grad_norm": 1.876592993736267,
-      "learning_rate": 5.165427135678392e-05,
-      "loss": 4.678,
-      "step": 48622
-    },
-    {
-      "epoch": 25.35749674054759,
-      "grad_norm": 1.68411386013031,
-      "learning_rate": 5.165326633165829e-05,
-      "loss": 5.081,
-      "step": 48623
-    },
-    {
-      "epoch": 25.35801825293351,
-      "grad_norm": 1.5956672430038452,
-      "learning_rate": 5.1652261306532665e-05,
-      "loss": 5.5799,
-      "step": 48624
-    },
-    {
-      "epoch": 25.358539765319428,
-      "grad_norm": 1.5962228775024414,
-      "learning_rate": 5.1651256281407036e-05,
-      "loss": 5.0445,
-      "step": 48625
-    },
-    {
-      "epoch": 25.359061277705347,
-      "grad_norm": 1.6801810264587402,
-      "learning_rate": 5.1650251256281414e-05,
-      "loss": 5.3893,
-      "step": 48626
-    },
-    {
-      "epoch": 25.359582790091263,
-      "grad_norm": 1.5680267810821533,
-      "learning_rate": 5.164924623115578e-05,
-      "loss": 4.6678,
-      "step": 48627
-    },
-    {
-      "epoch": 25.360104302477183,
-      "grad_norm": 1.5068014860153198,
-      "learning_rate": 5.1648241206030156e-05,
-      "loss": 5.6945,
-      "step": 48628
-    },
-    {
-      "epoch": 25.360625814863102,
-      "grad_norm": 1.4337983131408691,
-      "learning_rate": 5.164723618090452e-05,
-      "loss": 4.8326,
-      "step": 48629
-    },
-    {
-      "epoch": 25.36114732724902,
-      "grad_norm": 1.585800290107727,
-      "learning_rate": 5.16462311557789e-05,
-      "loss": 5.0529,
-      "step": 48630
-    },
-    {
-      "epoch": 25.36166883963494,
-      "grad_norm": 1.5852980613708496,
-      "learning_rate": 5.164522613065327e-05,
-      "loss": 5.2876,
-      "step": 48631
-    },
-    {
-      "epoch": 25.36219035202086,
-      "grad_norm": 1.5891375541687012,
-      "learning_rate": 5.164422110552765e-05,
-      "loss": 4.7047,
-      "step": 48632
-    },
-    {
-      "epoch": 25.36271186440678,
-      "grad_norm": 1.5172505378723145,
-      "learning_rate": 5.164321608040201e-05,
-      "loss": 5.1684,
-      "step": 48633
-    },
-    {
-      "epoch": 25.3632333767927,
-      "grad_norm": 1.5760256052017212,
-      "learning_rate": 5.164221105527638e-05,
-      "loss": 5.2612,
-      "step": 48634
-    },
-    {
-      "epoch": 25.36375488917862,
-      "grad_norm": 1.607162594795227,
-      "learning_rate": 5.164120603015076e-05,
-      "loss": 5.3605,
-      "step": 48635
-    },
-    {
-      "epoch": 25.36427640156454,
-      "grad_norm": 1.5073364973068237,
-      "learning_rate": 5.1640201005025125e-05,
-      "loss": 5.6667,
-      "step": 48636
-    },
-    {
-      "epoch": 25.364797913950458,
-      "grad_norm": 1.5826934576034546,
-      "learning_rate": 5.16391959798995e-05,
-      "loss": 5.0118,
-      "step": 48637
-    },
-    {
-      "epoch": 25.365319426336377,
-      "grad_norm": 1.6883397102355957,
-      "learning_rate": 5.163819095477387e-05,
-      "loss": 4.5418,
-      "step": 48638
-    },
-    {
-      "epoch": 25.365840938722293,
-      "grad_norm": 1.5607984066009521,
-      "learning_rate": 5.1637185929648245e-05,
-      "loss": 4.9662,
-      "step": 48639
-    },
-    {
-      "epoch": 25.366362451108213,
-      "grad_norm": 1.5592013597488403,
-      "learning_rate": 5.1636180904522616e-05,
-      "loss": 5.1716,
-      "step": 48640
-    },
-    {
-      "epoch": 25.366883963494132,
-      "grad_norm": 1.6123462915420532,
-      "learning_rate": 5.1635175879396993e-05,
-      "loss": 5.177,
-      "step": 48641
-    },
-    {
-      "epoch": 25.36740547588005,
-      "grad_norm": 1.7034494876861572,
-      "learning_rate": 5.163417085427136e-05,
-      "loss": 4.3158,
-      "step": 48642
-    },
-    {
-      "epoch": 25.36792698826597,
-      "grad_norm": 1.6353645324707031,
-      "learning_rate": 5.1633165829145736e-05,
-      "loss": 4.8649,
-      "step": 48643
-    },
-    {
-      "epoch": 25.36844850065189,
-      "grad_norm": 1.640093207359314,
-      "learning_rate": 5.16321608040201e-05,
-      "loss": 5.1932,
-      "step": 48644
-    },
-    {
-      "epoch": 25.36897001303781,
-      "grad_norm": 1.5676757097244263,
-      "learning_rate": 5.163115577889448e-05,
-      "loss": 5.3293,
-      "step": 48645
-    },
-    {
-      "epoch": 25.36949152542373,
-      "grad_norm": 1.5381194353103638,
-      "learning_rate": 5.163015075376885e-05,
-      "loss": 5.2846,
-      "step": 48646
-    },
-    {
-      "epoch": 25.37001303780965,
-      "grad_norm": 1.5794904232025146,
-      "learning_rate": 5.162914572864321e-05,
-      "loss": 4.7776,
-      "step": 48647
-    },
-    {
-      "epoch": 25.37053455019557,
-      "grad_norm": 1.7643852233886719,
-      "learning_rate": 5.162814070351759e-05,
-      "loss": 4.8858,
-      "step": 48648
-    },
-    {
-      "epoch": 25.371056062581488,
-      "grad_norm": 1.4828307628631592,
-      "learning_rate": 5.1627135678391955e-05,
-      "loss": 5.6238,
-      "step": 48649
-    },
-    {
-      "epoch": 25.371577574967404,
-      "grad_norm": 1.583393931388855,
-      "learning_rate": 5.162613065326633e-05,
-      "loss": 4.7968,
-      "step": 48650
-    },
-    {
-      "epoch": 25.372099087353323,
-      "grad_norm": 1.5963013172149658,
-      "learning_rate": 5.1625125628140704e-05,
-      "loss": 5.3418,
-      "step": 48651
-    },
-    {
-      "epoch": 25.372620599739243,
-      "grad_norm": 1.6259045600891113,
-      "learning_rate": 5.162412060301508e-05,
-      "loss": 4.8592,
-      "step": 48652
-    },
-    {
-      "epoch": 25.373142112125162,
-      "grad_norm": 1.446448802947998,
-      "learning_rate": 5.1623115577889446e-05,
-      "loss": 5.5959,
-      "step": 48653
-    },
-    {
-      "epoch": 25.37366362451108,
-      "grad_norm": 1.5679539442062378,
-      "learning_rate": 5.1622110552763824e-05,
-      "loss": 5.1849,
-      "step": 48654
-    },
-    {
-      "epoch": 25.374185136897,
-      "grad_norm": 1.588621735572815,
-      "learning_rate": 5.1621105527638195e-05,
-      "loss": 4.7587,
-      "step": 48655
-    },
-    {
-      "epoch": 25.37470664928292,
-      "grad_norm": 1.5927817821502686,
-      "learning_rate": 5.162010050251257e-05,
-      "loss": 4.7108,
-      "step": 48656
-    },
-    {
-      "epoch": 25.37522816166884,
-      "grad_norm": 1.5864720344543457,
-      "learning_rate": 5.161909547738694e-05,
-      "loss": 5.1125,
-      "step": 48657
-    },
-    {
-      "epoch": 25.37574967405476,
-      "grad_norm": 1.5493046045303345,
-      "learning_rate": 5.1618090452261315e-05,
-      "loss": 5.3782,
-      "step": 48658
-    },
-    {
-      "epoch": 25.37627118644068,
-      "grad_norm": 1.5939159393310547,
-      "learning_rate": 5.161708542713568e-05,
-      "loss": 5.0131,
-      "step": 48659
-    },
-    {
-      "epoch": 25.3767926988266,
-      "grad_norm": 1.5327914953231812,
-      "learning_rate": 5.161608040201005e-05,
-      "loss": 5.4454,
-      "step": 48660
-    },
-    {
-      "epoch": 25.377314211212518,
-      "grad_norm": 1.4988529682159424,
-      "learning_rate": 5.161507537688443e-05,
-      "loss": 5.7728,
-      "step": 48661
-    },
-    {
-      "epoch": 25.377835723598434,
-      "grad_norm": 1.6155612468719482,
-      "learning_rate": 5.161407035175879e-05,
-      "loss": 4.9829,
-      "step": 48662
-    },
-    {
-      "epoch": 25.378357235984353,
-      "grad_norm": 1.5401945114135742,
-      "learning_rate": 5.161306532663317e-05,
-      "loss": 5.1764,
-      "step": 48663
-    },
-    {
-      "epoch": 25.378878748370273,
-      "grad_norm": 1.492969036102295,
-      "learning_rate": 5.1612060301507534e-05,
-      "loss": 5.5554,
-      "step": 48664
-    },
-    {
-      "epoch": 25.379400260756192,
-      "grad_norm": 1.6423238515853882,
-      "learning_rate": 5.161105527638191e-05,
-      "loss": 4.935,
-      "step": 48665
-    },
-    {
-      "epoch": 25.37992177314211,
-      "grad_norm": 1.6385939121246338,
-      "learning_rate": 5.161005025125628e-05,
-      "loss": 5.1219,
-      "step": 48666
-    },
-    {
-      "epoch": 25.38044328552803,
-      "grad_norm": 1.628400444984436,
-      "learning_rate": 5.160904522613066e-05,
-      "loss": 4.7691,
-      "step": 48667
-    },
-    {
-      "epoch": 25.38096479791395,
-      "grad_norm": 1.6573028564453125,
-      "learning_rate": 5.1608040201005025e-05,
-      "loss": 4.5814,
-      "step": 48668
-    },
-    {
-      "epoch": 25.38148631029987,
-      "grad_norm": 1.5827038288116455,
-      "learning_rate": 5.16070351758794e-05,
-      "loss": 5.2182,
-      "step": 48669
-    },
-    {
-      "epoch": 25.38200782268579,
-      "grad_norm": 1.596052885055542,
-      "learning_rate": 5.160603015075377e-05,
-      "loss": 5.36,
-      "step": 48670
-    },
-    {
-      "epoch": 25.38252933507171,
-      "grad_norm": 1.5833617448806763,
-      "learning_rate": 5.1605025125628145e-05,
-      "loss": 5.0968,
-      "step": 48671
-    },
-    {
-      "epoch": 25.38305084745763,
-      "grad_norm": 1.5769758224487305,
-      "learning_rate": 5.1604020100502516e-05,
-      "loss": 5.2713,
-      "step": 48672
-    },
-    {
-      "epoch": 25.383572359843548,
-      "grad_norm": 1.4942151308059692,
-      "learning_rate": 5.160301507537688e-05,
-      "loss": 5.605,
-      "step": 48673
-    },
-    {
-      "epoch": 25.384093872229464,
-      "grad_norm": 1.5421115159988403,
-      "learning_rate": 5.160201005025126e-05,
-      "loss": 5.2488,
-      "step": 48674
-    },
-    {
-      "epoch": 25.384615384615383,
-      "grad_norm": 1.4805301427841187,
-      "learning_rate": 5.160100502512562e-05,
-      "loss": 5.2521,
-      "step": 48675
-    },
-    {
-      "epoch": 25.385136897001303,
-      "grad_norm": 1.607869267463684,
-      "learning_rate": 5.16e-05,
-      "loss": 5.0921,
-      "step": 48676
-    },
-    {
-      "epoch": 25.385658409387222,
-      "grad_norm": 1.5150866508483887,
-      "learning_rate": 5.159899497487437e-05,
-      "loss": 5.1293,
-      "step": 48677
-    },
-    {
-      "epoch": 25.38617992177314,
-      "grad_norm": 1.5158594846725464,
-      "learning_rate": 5.159798994974875e-05,
-      "loss": 4.9655,
-      "step": 48678
-    },
-    {
-      "epoch": 25.38670143415906,
-      "grad_norm": 1.580691933631897,
-      "learning_rate": 5.1596984924623114e-05,
-      "loss": 5.4456,
-      "step": 48679
-    },
-    {
-      "epoch": 25.38722294654498,
-      "grad_norm": 1.4918469190597534,
-      "learning_rate": 5.159597989949749e-05,
-      "loss": 5.2154,
-      "step": 48680
-    },
-    {
-      "epoch": 25.3877444589309,
-      "grad_norm": 1.5084091424942017,
-      "learning_rate": 5.159497487437186e-05,
-      "loss": 5.5421,
-      "step": 48681
-    },
-    {
-      "epoch": 25.38826597131682,
-      "grad_norm": 1.574414849281311,
-      "learning_rate": 5.159396984924624e-05,
-      "loss": 5.4671,
-      "step": 48682
-    },
-    {
-      "epoch": 25.38878748370274,
-      "grad_norm": 1.5326447486877441,
-      "learning_rate": 5.1592964824120605e-05,
-      "loss": 5.3143,
-      "step": 48683
-    },
-    {
-      "epoch": 25.38930899608866,
-      "grad_norm": 1.6694937944412231,
-      "learning_rate": 5.159195979899498e-05,
-      "loss": 5.5988,
-      "step": 48684
-    },
-    {
-      "epoch": 25.389830508474578,
-      "grad_norm": 1.709242820739746,
-      "learning_rate": 5.159095477386935e-05,
-      "loss": 4.8631,
-      "step": 48685
-    },
-    {
-      "epoch": 25.390352020860494,
-      "grad_norm": 1.606035590171814,
-      "learning_rate": 5.158994974874372e-05,
-      "loss": 5.0814,
-      "step": 48686
-    },
-    {
-      "epoch": 25.390873533246413,
-      "grad_norm": 1.5427119731903076,
-      "learning_rate": 5.1588944723618096e-05,
-      "loss": 5.0456,
-      "step": 48687
-    },
-    {
-      "epoch": 25.391395045632333,
-      "grad_norm": 1.6057218313217163,
-      "learning_rate": 5.158793969849246e-05,
-      "loss": 4.7115,
-      "step": 48688
-    },
-    {
-      "epoch": 25.391916558018252,
-      "grad_norm": 1.5941840410232544,
-      "learning_rate": 5.158693467336684e-05,
-      "loss": 5.4703,
-      "step": 48689
-    },
-    {
-      "epoch": 25.39243807040417,
-      "grad_norm": 1.715334177017212,
-      "learning_rate": 5.15859296482412e-05,
-      "loss": 5.2313,
-      "step": 48690
-    },
-    {
-      "epoch": 25.39295958279009,
-      "grad_norm": 1.586108684539795,
-      "learning_rate": 5.158492462311558e-05,
-      "loss": 5.7314,
-      "step": 48691
-    },
-    {
-      "epoch": 25.39348109517601,
-      "grad_norm": 1.5748380422592163,
-      "learning_rate": 5.158391959798995e-05,
-      "loss": 5.3268,
-      "step": 48692
-    },
-    {
-      "epoch": 25.39400260756193,
-      "grad_norm": 1.4896392822265625,
-      "learning_rate": 5.158291457286433e-05,
-      "loss": 5.6592,
-      "step": 48693
-    },
-    {
-      "epoch": 25.39452411994785,
-      "grad_norm": 1.6265192031860352,
-      "learning_rate": 5.158190954773869e-05,
-      "loss": 4.9711,
-      "step": 48694
-    },
-    {
-      "epoch": 25.39504563233377,
-      "grad_norm": 1.4877095222473145,
-      "learning_rate": 5.158090452261307e-05,
-      "loss": 5.5476,
-      "step": 48695
-    },
-    {
-      "epoch": 25.39556714471969,
-      "grad_norm": 1.6540958881378174,
-      "learning_rate": 5.1579899497487435e-05,
-      "loss": 5.0944,
-      "step": 48696
-    },
-    {
-      "epoch": 25.396088657105608,
-      "grad_norm": 1.5424458980560303,
-      "learning_rate": 5.157889447236181e-05,
-      "loss": 5.1472,
-      "step": 48697
-    },
-    {
-      "epoch": 25.396610169491524,
-      "grad_norm": 1.6343233585357666,
-      "learning_rate": 5.1577889447236184e-05,
-      "loss": 5.2846,
-      "step": 48698
-    },
-    {
-      "epoch": 25.397131681877443,
-      "grad_norm": 1.503279447555542,
-      "learning_rate": 5.157688442211056e-05,
-      "loss": 5.6018,
-      "step": 48699
-    },
-    {
-      "epoch": 25.397653194263363,
-      "grad_norm": 1.5297571420669556,
-      "learning_rate": 5.1575879396984926e-05,
-      "loss": 5.7373,
-      "step": 48700
-    },
-    {
-      "epoch": 25.398174706649282,
-      "grad_norm": 1.5696423053741455,
-      "learning_rate": 5.15748743718593e-05,
-      "loss": 5.0616,
-      "step": 48701
-    },
-    {
-      "epoch": 25.3986962190352,
-      "grad_norm": 1.5223500728607178,
-      "learning_rate": 5.1573869346733675e-05,
-      "loss": 5.3451,
-      "step": 48702
-    },
-    {
-      "epoch": 25.39921773142112,
-      "grad_norm": 1.5900804996490479,
-      "learning_rate": 5.157286432160804e-05,
-      "loss": 5.0277,
-      "step": 48703
-    },
-    {
-      "epoch": 25.39973924380704,
-      "grad_norm": 1.7568047046661377,
-      "learning_rate": 5.157185929648242e-05,
-      "loss": 4.9208,
-      "step": 48704
-    },
-    {
-      "epoch": 25.40026075619296,
-      "grad_norm": 1.6192514896392822,
-      "learning_rate": 5.157085427135678e-05,
-      "loss": 5.2456,
-      "step": 48705
-    },
-    {
-      "epoch": 25.40078226857888,
-      "grad_norm": 1.6020219326019287,
-      "learning_rate": 5.156984924623116e-05,
-      "loss": 5.1556,
-      "step": 48706
-    },
-    {
-      "epoch": 25.4013037809648,
-      "grad_norm": 1.6122126579284668,
-      "learning_rate": 5.156884422110553e-05,
-      "loss": 5.1957,
-      "step": 48707
-    },
-    {
-      "epoch": 25.40182529335072,
-      "grad_norm": 1.5939961671829224,
-      "learning_rate": 5.156783919597991e-05,
-      "loss": 5.2139,
-      "step": 48708
-    },
-    {
-      "epoch": 25.402346805736638,
-      "grad_norm": 1.623036503791809,
-      "learning_rate": 5.156683417085427e-05,
-      "loss": 5.1474,
-      "step": 48709
-    },
-    {
-      "epoch": 25.402868318122554,
-      "grad_norm": 1.6067742109298706,
-      "learning_rate": 5.156582914572865e-05,
-      "loss": 5.1348,
-      "step": 48710
-    },
-    {
-      "epoch": 25.403389830508473,
-      "grad_norm": 1.5704131126403809,
-      "learning_rate": 5.1564824120603015e-05,
-      "loss": 4.8414,
-      "step": 48711
-    },
-    {
-      "epoch": 25.403911342894393,
-      "grad_norm": 1.625951886177063,
-      "learning_rate": 5.156381909547739e-05,
-      "loss": 5.1086,
-      "step": 48712
-    },
-    {
-      "epoch": 25.404432855280312,
-      "grad_norm": 1.5978426933288574,
-      "learning_rate": 5.1562814070351763e-05,
-      "loss": 5.4079,
-      "step": 48713
-    },
-    {
-      "epoch": 25.40495436766623,
-      "grad_norm": 1.6278952360153198,
-      "learning_rate": 5.156180904522613e-05,
-      "loss": 4.9326,
-      "step": 48714
-    },
-    {
-      "epoch": 25.40547588005215,
-      "grad_norm": 1.5088653564453125,
-      "learning_rate": 5.1560804020100506e-05,
-      "loss": 5.2265,
-      "step": 48715
-    },
-    {
-      "epoch": 25.40599739243807,
-      "grad_norm": 1.7388988733291626,
-      "learning_rate": 5.155979899497487e-05,
-      "loss": 4.9494,
-      "step": 48716
-    },
-    {
-      "epoch": 25.40651890482399,
-      "grad_norm": 1.5844447612762451,
-      "learning_rate": 5.155879396984925e-05,
-      "loss": 5.5313,
-      "step": 48717
-    },
-    {
-      "epoch": 25.40704041720991,
-      "grad_norm": 1.5349947214126587,
-      "learning_rate": 5.155778894472362e-05,
-      "loss": 5.369,
-      "step": 48718
-    },
-    {
-      "epoch": 25.40756192959583,
-      "grad_norm": 1.5794901847839355,
-      "learning_rate": 5.1556783919597997e-05,
-      "loss": 4.9558,
-      "step": 48719
-    },
-    {
-      "epoch": 25.40808344198175,
-      "grad_norm": 1.6343796253204346,
-      "learning_rate": 5.155577889447236e-05,
-      "loss": 5.2357,
-      "step": 48720
-    },
-    {
-      "epoch": 25.408604954367668,
-      "grad_norm": 1.5230633020401,
-      "learning_rate": 5.155477386934674e-05,
-      "loss": 5.2536,
-      "step": 48721
-    },
-    {
-      "epoch": 25.409126466753584,
-      "grad_norm": 1.7045761346817017,
-      "learning_rate": 5.155376884422111e-05,
-      "loss": 4.9698,
-      "step": 48722
-    },
-    {
-      "epoch": 25.409647979139503,
-      "grad_norm": 1.4593881368637085,
-      "learning_rate": 5.155276381909549e-05,
-      "loss": 5.7312,
-      "step": 48723
-    },
-    {
-      "epoch": 25.410169491525423,
-      "grad_norm": 1.5296324491500854,
-      "learning_rate": 5.155175879396985e-05,
-      "loss": 5.3096,
-      "step": 48724
-    },
-    {
-      "epoch": 25.410691003911342,
-      "grad_norm": 1.571215271949768,
-      "learning_rate": 5.155075376884423e-05,
-      "loss": 4.8248,
-      "step": 48725
-    },
-    {
-      "epoch": 25.41121251629726,
-      "grad_norm": 1.53542160987854,
-      "learning_rate": 5.1549748743718594e-05,
-      "loss": 5.5126,
-      "step": 48726
-    },
-    {
-      "epoch": 25.41173402868318,
-      "grad_norm": 1.5855152606964111,
-      "learning_rate": 5.1548743718592965e-05,
-      "loss": 5.0823,
-      "step": 48727
-    },
-    {
-      "epoch": 25.4122555410691,
-      "grad_norm": 1.618701457977295,
-      "learning_rate": 5.154773869346734e-05,
-      "loss": 5.4162,
-      "step": 48728
-    },
-    {
-      "epoch": 25.41277705345502,
-      "grad_norm": 1.708299160003662,
-      "learning_rate": 5.154673366834171e-05,
-      "loss": 5.2625,
-      "step": 48729
-    },
-    {
-      "epoch": 25.41329856584094,
-      "grad_norm": 1.601278305053711,
-      "learning_rate": 5.1545728643216085e-05,
-      "loss": 4.9366,
-      "step": 48730
-    },
-    {
-      "epoch": 25.41382007822686,
-      "grad_norm": 1.4437475204467773,
-      "learning_rate": 5.154472361809045e-05,
-      "loss": 5.4525,
-      "step": 48731
-    },
-    {
-      "epoch": 25.41434159061278,
-      "grad_norm": 1.5522147417068481,
-      "learning_rate": 5.154371859296483e-05,
-      "loss": 5.3465,
-      "step": 48732
-    },
-    {
-      "epoch": 25.414863102998694,
-      "grad_norm": 1.5948935747146606,
-      "learning_rate": 5.15427135678392e-05,
-      "loss": 4.9531,
-      "step": 48733
-    },
-    {
-      "epoch": 25.415384615384614,
-      "grad_norm": 1.5641202926635742,
-      "learning_rate": 5.1541708542713576e-05,
-      "loss": 5.5731,
-      "step": 48734
-    },
-    {
-      "epoch": 25.415906127770533,
-      "grad_norm": 1.5927212238311768,
-      "learning_rate": 5.154070351758794e-05,
-      "loss": 5.2419,
-      "step": 48735
-    },
-    {
-      "epoch": 25.416427640156453,
-      "grad_norm": 1.6920965909957886,
-      "learning_rate": 5.153969849246232e-05,
-      "loss": 5.2274,
-      "step": 48736
-    },
-    {
-      "epoch": 25.416949152542372,
-      "grad_norm": 1.56526780128479,
-      "learning_rate": 5.153869346733668e-05,
-      "loss": 4.893,
-      "step": 48737
-    },
-    {
-      "epoch": 25.41747066492829,
-      "grad_norm": 1.4940389394760132,
-      "learning_rate": 5.153768844221106e-05,
-      "loss": 5.2565,
-      "step": 48738
-    },
-    {
-      "epoch": 25.41799217731421,
-      "grad_norm": 1.5806639194488525,
-      "learning_rate": 5.153668341708543e-05,
-      "loss": 4.9654,
-      "step": 48739
-    },
-    {
-      "epoch": 25.41851368970013,
-      "grad_norm": 1.5217660665512085,
-      "learning_rate": 5.1535678391959795e-05,
-      "loss": 5.2316,
-      "step": 48740
-    },
-    {
-      "epoch": 25.41903520208605,
-      "grad_norm": 1.467848777770996,
-      "learning_rate": 5.153467336683417e-05,
-      "loss": 5.6382,
-      "step": 48741
-    },
-    {
-      "epoch": 25.41955671447197,
-      "grad_norm": 1.5235515832901,
-      "learning_rate": 5.153366834170854e-05,
-      "loss": 4.828,
-      "step": 48742
-    },
-    {
-      "epoch": 25.42007822685789,
-      "grad_norm": 1.5086933374404907,
-      "learning_rate": 5.1532663316582915e-05,
-      "loss": 5.2873,
-      "step": 48743
-    },
-    {
-      "epoch": 25.42059973924381,
-      "grad_norm": 1.7409090995788574,
-      "learning_rate": 5.1531658291457286e-05,
-      "loss": 5.2056,
-      "step": 48744
-    },
-    {
-      "epoch": 25.421121251629724,
-      "grad_norm": 1.6457839012145996,
-      "learning_rate": 5.1530653266331664e-05,
-      "loss": 5.4254,
-      "step": 48745
-    },
-    {
-      "epoch": 25.421642764015644,
-      "grad_norm": 1.5704584121704102,
-      "learning_rate": 5.152964824120603e-05,
-      "loss": 5.7799,
-      "step": 48746
-    },
-    {
-      "epoch": 25.422164276401563,
-      "grad_norm": 1.5595637559890747,
-      "learning_rate": 5.1528643216080406e-05,
-      "loss": 5.1484,
-      "step": 48747
-    },
-    {
-      "epoch": 25.422685788787483,
-      "grad_norm": 1.5638841390609741,
-      "learning_rate": 5.152763819095478e-05,
-      "loss": 5.4116,
-      "step": 48748
-    },
-    {
-      "epoch": 25.423207301173402,
-      "grad_norm": 1.4581124782562256,
-      "learning_rate": 5.1526633165829155e-05,
-      "loss": 5.3129,
-      "step": 48749
-    },
-    {
-      "epoch": 25.423728813559322,
-      "grad_norm": 1.6408509016036987,
-      "learning_rate": 5.152562814070352e-05,
-      "loss": 4.6183,
-      "step": 48750
-    },
-    {
-      "epoch": 25.42425032594524,
-      "grad_norm": 1.575932502746582,
-      "learning_rate": 5.15246231155779e-05,
-      "loss": 5.1306,
-      "step": 48751
-    },
-    {
-      "epoch": 25.42477183833116,
-      "grad_norm": 1.6156325340270996,
-      "learning_rate": 5.152361809045226e-05,
-      "loss": 5.7384,
-      "step": 48752
-    },
-    {
-      "epoch": 25.42529335071708,
-      "grad_norm": 1.6475096940994263,
-      "learning_rate": 5.152261306532663e-05,
-      "loss": 5.2152,
-      "step": 48753
-    },
-    {
-      "epoch": 25.425814863103,
-      "grad_norm": 1.6150480508804321,
-      "learning_rate": 5.152160804020101e-05,
-      "loss": 4.8916,
-      "step": 48754
-    },
-    {
-      "epoch": 25.42633637548892,
-      "grad_norm": 1.597909927368164,
-      "learning_rate": 5.1520603015075375e-05,
-      "loss": 5.1233,
-      "step": 48755
-    },
-    {
-      "epoch": 25.42685788787484,
-      "grad_norm": 1.6395503282546997,
-      "learning_rate": 5.151959798994975e-05,
-      "loss": 5.1619,
-      "step": 48756
-    },
-    {
-      "epoch": 25.427379400260754,
-      "grad_norm": 1.6429308652877808,
-      "learning_rate": 5.151859296482412e-05,
-      "loss": 4.9548,
-      "step": 48757
-    },
-    {
-      "epoch": 25.427900912646674,
-      "grad_norm": 1.670779824256897,
-      "learning_rate": 5.1517587939698495e-05,
-      "loss": 5.3706,
-      "step": 48758
-    },
-    {
-      "epoch": 25.428422425032593,
-      "grad_norm": 1.5284515619277954,
-      "learning_rate": 5.1516582914572866e-05,
-      "loss": 5.3276,
-      "step": 48759
-    },
-    {
-      "epoch": 25.428943937418513,
-      "grad_norm": 1.5489469766616821,
-      "learning_rate": 5.1515577889447244e-05,
-      "loss": 5.6399,
-      "step": 48760
-    },
-    {
-      "epoch": 25.429465449804432,
-      "grad_norm": 1.5555830001831055,
-      "learning_rate": 5.151457286432161e-05,
-      "loss": 5.4327,
-      "step": 48761
-    },
-    {
-      "epoch": 25.429986962190352,
-      "grad_norm": 1.5341514348983765,
-      "learning_rate": 5.1513567839195986e-05,
-      "loss": 5.1033,
-      "step": 48762
-    },
-    {
-      "epoch": 25.43050847457627,
-      "grad_norm": 1.5374161005020142,
-      "learning_rate": 5.151256281407035e-05,
-      "loss": 5.1961,
-      "step": 48763
-    },
-    {
-      "epoch": 25.43102998696219,
-      "grad_norm": 1.485835075378418,
-      "learning_rate": 5.151155778894473e-05,
-      "loss": 5.2652,
-      "step": 48764
-    },
-    {
-      "epoch": 25.43155149934811,
-      "grad_norm": 1.712064504623413,
-      "learning_rate": 5.15105527638191e-05,
-      "loss": 4.7791,
-      "step": 48765
-    },
-    {
-      "epoch": 25.43207301173403,
-      "grad_norm": 1.6681610345840454,
-      "learning_rate": 5.150954773869346e-05,
-      "loss": 5.7112,
-      "step": 48766
-    },
-    {
-      "epoch": 25.43259452411995,
-      "grad_norm": 1.5463600158691406,
-      "learning_rate": 5.150854271356784e-05,
-      "loss": 5.3073,
-      "step": 48767
-    },
-    {
-      "epoch": 25.43311603650587,
-      "grad_norm": 1.6510193347930908,
-      "learning_rate": 5.150753768844221e-05,
-      "loss": 5.0367,
-      "step": 48768
-    },
-    {
-      "epoch": 25.433637548891785,
-      "grad_norm": 1.556320309638977,
-      "learning_rate": 5.150653266331659e-05,
-      "loss": 5.2327,
-      "step": 48769
-    },
-    {
-      "epoch": 25.434159061277704,
-      "grad_norm": 1.535349726676941,
-      "learning_rate": 5.1505527638190954e-05,
-      "loss": 4.8515,
-      "step": 48770
-    },
-    {
-      "epoch": 25.434680573663623,
-      "grad_norm": 1.8054602146148682,
-      "learning_rate": 5.150452261306533e-05,
-      "loss": 5.0524,
-      "step": 48771
-    },
-    {
-      "epoch": 25.435202086049543,
-      "grad_norm": 1.5073297023773193,
-      "learning_rate": 5.1503517587939696e-05,
-      "loss": 5.1606,
-      "step": 48772
-    },
-    {
-      "epoch": 25.435723598435462,
-      "grad_norm": 1.5193510055541992,
-      "learning_rate": 5.1502512562814074e-05,
-      "loss": 5.6494,
-      "step": 48773
-    },
-    {
-      "epoch": 25.436245110821382,
-      "grad_norm": 1.5200436115264893,
-      "learning_rate": 5.1501507537688445e-05,
-      "loss": 5.6113,
-      "step": 48774
-    },
-    {
-      "epoch": 25.4367666232073,
-      "grad_norm": 1.5623139142990112,
-      "learning_rate": 5.150050251256282e-05,
-      "loss": 5.0449,
-      "step": 48775
-    },
-    {
-      "epoch": 25.43728813559322,
-      "grad_norm": 1.5080626010894775,
-      "learning_rate": 5.149949748743719e-05,
-      "loss": 5.5637,
-      "step": 48776
-    },
-    {
-      "epoch": 25.43780964797914,
-      "grad_norm": 1.57700514793396,
-      "learning_rate": 5.1498492462311565e-05,
-      "loss": 4.6756,
-      "step": 48777
-    },
-    {
-      "epoch": 25.43833116036506,
-      "grad_norm": 1.535646915435791,
-      "learning_rate": 5.149748743718593e-05,
-      "loss": 5.5462,
-      "step": 48778
-    },
-    {
-      "epoch": 25.43885267275098,
-      "grad_norm": 1.5468218326568604,
-      "learning_rate": 5.14964824120603e-05,
-      "loss": 5.6102,
-      "step": 48779
-    },
-    {
-      "epoch": 25.4393741851369,
-      "grad_norm": 1.6353203058242798,
-      "learning_rate": 5.149547738693468e-05,
-      "loss": 4.8514,
-      "step": 48780
-    },
-    {
-      "epoch": 25.439895697522815,
-      "grad_norm": 1.5476833581924438,
-      "learning_rate": 5.149447236180904e-05,
-      "loss": 5.2612,
-      "step": 48781
-    },
-    {
-      "epoch": 25.440417209908734,
-      "grad_norm": 1.535524606704712,
-      "learning_rate": 5.149346733668342e-05,
-      "loss": 5.2928,
-      "step": 48782
-    },
-    {
-      "epoch": 25.440938722294653,
-      "grad_norm": 1.5223387479782104,
-      "learning_rate": 5.1492462311557784e-05,
-      "loss": 5.4803,
-      "step": 48783
-    },
-    {
-      "epoch": 25.441460234680573,
-      "grad_norm": 1.6258329153060913,
-      "learning_rate": 5.149145728643216e-05,
-      "loss": 5.09,
-      "step": 48784
-    },
-    {
-      "epoch": 25.441981747066492,
-      "grad_norm": 1.6516965627670288,
-      "learning_rate": 5.149045226130653e-05,
-      "loss": 5.4866,
-      "step": 48785
-    },
-    {
-      "epoch": 25.442503259452412,
-      "grad_norm": 1.5564312934875488,
-      "learning_rate": 5.148944723618091e-05,
-      "loss": 5.2589,
-      "step": 48786
-    },
-    {
-      "epoch": 25.44302477183833,
-      "grad_norm": 1.6062028408050537,
-      "learning_rate": 5.1488442211055275e-05,
-      "loss": 5.1347,
-      "step": 48787
-    },
-    {
-      "epoch": 25.44354628422425,
-      "grad_norm": 1.5328680276870728,
-      "learning_rate": 5.148743718592965e-05,
-      "loss": 4.8408,
-      "step": 48788
-    },
-    {
-      "epoch": 25.44406779661017,
-      "grad_norm": 1.5672683715820312,
-      "learning_rate": 5.1486432160804024e-05,
-      "loss": 5.0368,
-      "step": 48789
-    },
-    {
-      "epoch": 25.44458930899609,
-      "grad_norm": 1.568207859992981,
-      "learning_rate": 5.14854271356784e-05,
-      "loss": 5.3468,
-      "step": 48790
-    },
-    {
-      "epoch": 25.44511082138201,
-      "grad_norm": 1.531510829925537,
-      "learning_rate": 5.1484422110552766e-05,
-      "loss": 5.5942,
-      "step": 48791
-    },
-    {
-      "epoch": 25.44563233376793,
-      "grad_norm": 1.593832015991211,
-      "learning_rate": 5.1483417085427144e-05,
-      "loss": 5.2321,
-      "step": 48792
-    },
-    {
-      "epoch": 25.446153846153845,
-      "grad_norm": 1.5372546911239624,
-      "learning_rate": 5.148241206030151e-05,
-      "loss": 4.9021,
-      "step": 48793
-    },
-    {
-      "epoch": 25.446675358539764,
-      "grad_norm": 1.5231353044509888,
-      "learning_rate": 5.148140703517588e-05,
-      "loss": 5.4177,
-      "step": 48794
-    },
-    {
-      "epoch": 25.447196870925683,
-      "grad_norm": 1.7047935724258423,
-      "learning_rate": 5.148040201005026e-05,
-      "loss": 5.1001,
-      "step": 48795
-    },
-    {
-      "epoch": 25.447718383311603,
-      "grad_norm": 1.6195977926254272,
-      "learning_rate": 5.147939698492462e-05,
-      "loss": 4.7889,
-      "step": 48796
-    },
-    {
-      "epoch": 25.448239895697522,
-      "grad_norm": 1.5864826440811157,
-      "learning_rate": 5.1478391959799e-05,
-      "loss": 5.2971,
-      "step": 48797
-    },
-    {
-      "epoch": 25.448761408083442,
-      "grad_norm": 1.6397333145141602,
-      "learning_rate": 5.1477386934673364e-05,
-      "loss": 5.5575,
-      "step": 48798
-    },
-    {
-      "epoch": 25.44928292046936,
-      "grad_norm": 1.5663326978683472,
-      "learning_rate": 5.147638190954774e-05,
-      "loss": 5.3391,
-      "step": 48799
-    },
-    {
-      "epoch": 25.44980443285528,
-      "grad_norm": 1.5343658924102783,
-      "learning_rate": 5.147537688442211e-05,
-      "loss": 5.3795,
-      "step": 48800
-    },
-    {
-      "epoch": 25.4503259452412,
-      "grad_norm": 1.7041765451431274,
-      "learning_rate": 5.147437185929649e-05,
-      "loss": 4.8936,
-      "step": 48801
-    },
-    {
-      "epoch": 25.45084745762712,
-      "grad_norm": 1.674311876296997,
-      "learning_rate": 5.1473366834170855e-05,
-      "loss": 5.0903,
-      "step": 48802
-    },
-    {
-      "epoch": 25.45136897001304,
-      "grad_norm": 1.5441570281982422,
-      "learning_rate": 5.147236180904523e-05,
-      "loss": 5.112,
-      "step": 48803
-    },
-    {
-      "epoch": 25.45189048239896,
-      "grad_norm": 1.6969025135040283,
-      "learning_rate": 5.14713567839196e-05,
-      "loss": 4.6042,
-      "step": 48804
-    },
-    {
-      "epoch": 25.452411994784875,
-      "grad_norm": 1.6273289918899536,
-      "learning_rate": 5.1470351758793975e-05,
-      "loss": 5.0513,
-      "step": 48805
-    },
-    {
-      "epoch": 25.452933507170794,
-      "grad_norm": 1.5250355005264282,
-      "learning_rate": 5.1469346733668346e-05,
-      "loss": 5.3076,
-      "step": 48806
-    },
-    {
-      "epoch": 25.453455019556714,
-      "grad_norm": 1.5515892505645752,
-      "learning_rate": 5.146834170854271e-05,
-      "loss": 5.4651,
-      "step": 48807
-    },
-    {
-      "epoch": 25.453976531942633,
-      "grad_norm": 1.5210728645324707,
-      "learning_rate": 5.146733668341709e-05,
-      "loss": 5.1738,
-      "step": 48808
-    },
-    {
-      "epoch": 25.454498044328552,
-      "grad_norm": 1.7820396423339844,
-      "learning_rate": 5.146633165829146e-05,
-      "loss": 5.1028,
-      "step": 48809
-    },
-    {
-      "epoch": 25.455019556714472,
-      "grad_norm": 1.7052854299545288,
-      "learning_rate": 5.146532663316584e-05,
-      "loss": 4.5929,
-      "step": 48810
-    },
-    {
-      "epoch": 25.45554106910039,
-      "grad_norm": 1.5416651964187622,
-      "learning_rate": 5.14643216080402e-05,
-      "loss": 4.1768,
-      "step": 48811
-    },
-    {
-      "epoch": 25.45606258148631,
-      "grad_norm": 1.703073501586914,
-      "learning_rate": 5.146331658291458e-05,
-      "loss": 5.0261,
-      "step": 48812
-    },
-    {
-      "epoch": 25.45658409387223,
-      "grad_norm": 1.718988060951233,
-      "learning_rate": 5.146231155778894e-05,
-      "loss": 5.4606,
-      "step": 48813
-    },
-    {
-      "epoch": 25.45710560625815,
-      "grad_norm": 1.723810076713562,
-      "learning_rate": 5.146130653266332e-05,
-      "loss": 5.2196,
-      "step": 48814
-    },
-    {
-      "epoch": 25.45762711864407,
-      "grad_norm": 1.5199912786483765,
-      "learning_rate": 5.146030150753769e-05,
-      "loss": 5.2068,
-      "step": 48815
-    },
-    {
-      "epoch": 25.45814863102999,
-      "grad_norm": 1.5404404401779175,
-      "learning_rate": 5.145929648241207e-05,
-      "loss": 5.2157,
-      "step": 48816
-    },
-    {
-      "epoch": 25.458670143415905,
-      "grad_norm": 1.4743837118148804,
-      "learning_rate": 5.1458291457286434e-05,
-      "loss": 5.4207,
-      "step": 48817
-    },
-    {
-      "epoch": 25.459191655801824,
-      "grad_norm": 1.652788519859314,
-      "learning_rate": 5.145728643216081e-05,
-      "loss": 4.7403,
-      "step": 48818
-    },
-    {
-      "epoch": 25.459713168187744,
-      "grad_norm": 1.6875842809677124,
-      "learning_rate": 5.1456281407035176e-05,
-      "loss": 4.6103,
-      "step": 48819
-    },
-    {
-      "epoch": 25.460234680573663,
-      "grad_norm": 1.47905695438385,
-      "learning_rate": 5.145527638190955e-05,
-      "loss": 5.6414,
-      "step": 48820
-    },
-    {
-      "epoch": 25.460756192959582,
-      "grad_norm": 1.5418161153793335,
-      "learning_rate": 5.1454271356783925e-05,
-      "loss": 5.4784,
-      "step": 48821
-    },
-    {
-      "epoch": 25.461277705345502,
-      "grad_norm": 1.518918514251709,
-      "learning_rate": 5.145326633165829e-05,
-      "loss": 5.5744,
-      "step": 48822
-    },
-    {
-      "epoch": 25.46179921773142,
-      "grad_norm": 1.6102043390274048,
-      "learning_rate": 5.145226130653267e-05,
-      "loss": 4.8869,
-      "step": 48823
-    },
-    {
-      "epoch": 25.46232073011734,
-      "grad_norm": 1.6451234817504883,
-      "learning_rate": 5.145125628140703e-05,
-      "loss": 5.0642,
-      "step": 48824
-    },
-    {
-      "epoch": 25.46284224250326,
-      "grad_norm": 1.584173560142517,
-      "learning_rate": 5.145025125628141e-05,
-      "loss": 5.187,
-      "step": 48825
-    },
-    {
-      "epoch": 25.46336375488918,
-      "grad_norm": 1.5857398509979248,
-      "learning_rate": 5.144924623115578e-05,
-      "loss": 5.4063,
-      "step": 48826
-    },
-    {
-      "epoch": 25.4638852672751,
-      "grad_norm": 1.4840774536132812,
-      "learning_rate": 5.144824120603016e-05,
-      "loss": 5.5332,
-      "step": 48827
-    },
-    {
-      "epoch": 25.46440677966102,
-      "grad_norm": 1.5834085941314697,
-      "learning_rate": 5.144723618090452e-05,
-      "loss": 5.0523,
-      "step": 48828
-    },
-    {
-      "epoch": 25.464928292046935,
-      "grad_norm": 1.6352564096450806,
-      "learning_rate": 5.14462311557789e-05,
-      "loss": 5.0238,
-      "step": 48829
-    },
-    {
-      "epoch": 25.465449804432854,
-      "grad_norm": 1.6033756732940674,
-      "learning_rate": 5.1445226130653265e-05,
-      "loss": 4.2107,
-      "step": 48830
-    },
-    {
-      "epoch": 25.465971316818774,
-      "grad_norm": 1.5955867767333984,
-      "learning_rate": 5.144422110552764e-05,
-      "loss": 4.9774,
-      "step": 48831
-    },
-    {
-      "epoch": 25.466492829204693,
-      "grad_norm": 1.5744930505752563,
-      "learning_rate": 5.1443216080402013e-05,
-      "loss": 5.0291,
-      "step": 48832
-    },
-    {
-      "epoch": 25.467014341590613,
-      "grad_norm": 1.5444729328155518,
-      "learning_rate": 5.144221105527638e-05,
-      "loss": 5.4331,
-      "step": 48833
-    },
-    {
-      "epoch": 25.467535853976532,
-      "grad_norm": 1.5675379037857056,
-      "learning_rate": 5.1441206030150756e-05,
-      "loss": 5.2658,
-      "step": 48834
-    },
-    {
-      "epoch": 25.46805736636245,
-      "grad_norm": 1.5810344219207764,
-      "learning_rate": 5.1440201005025127e-05,
-      "loss": 4.6379,
-      "step": 48835
-    },
-    {
-      "epoch": 25.46857887874837,
-      "grad_norm": 1.7068331241607666,
-      "learning_rate": 5.1439195979899504e-05,
-      "loss": 4.8198,
-      "step": 48836
-    },
-    {
-      "epoch": 25.46910039113429,
-      "grad_norm": 1.5384563207626343,
-      "learning_rate": 5.143819095477387e-05,
-      "loss": 5.3832,
-      "step": 48837
-    },
-    {
-      "epoch": 25.46962190352021,
-      "grad_norm": 1.6042972803115845,
-      "learning_rate": 5.1437185929648247e-05,
-      "loss": 5.042,
-      "step": 48838
-    },
-    {
-      "epoch": 25.47014341590613,
-      "grad_norm": 1.580981969833374,
-      "learning_rate": 5.143618090452261e-05,
-      "loss": 5.1976,
-      "step": 48839
-    },
-    {
-      "epoch": 25.470664928292045,
-      "grad_norm": 1.6592692136764526,
-      "learning_rate": 5.143517587939699e-05,
-      "loss": 5.0191,
-      "step": 48840
-    },
-    {
-      "epoch": 25.471186440677965,
-      "grad_norm": 1.6729260683059692,
-      "learning_rate": 5.143417085427136e-05,
-      "loss": 4.86,
-      "step": 48841
-    },
-    {
-      "epoch": 25.471707953063884,
-      "grad_norm": 1.6502329111099243,
-      "learning_rate": 5.143316582914574e-05,
-      "loss": 5.5569,
-      "step": 48842
-    },
-    {
-      "epoch": 25.472229465449804,
-      "grad_norm": 1.5887800455093384,
-      "learning_rate": 5.14321608040201e-05,
-      "loss": 5.4833,
-      "step": 48843
-    },
-    {
-      "epoch": 25.472750977835723,
-      "grad_norm": 1.618137240409851,
-      "learning_rate": 5.143115577889448e-05,
-      "loss": 5.0205,
-      "step": 48844
-    },
-    {
-      "epoch": 25.473272490221643,
-      "grad_norm": 1.6214656829833984,
-      "learning_rate": 5.1430150753768844e-05,
-      "loss": 4.6935,
-      "step": 48845
-    },
-    {
-      "epoch": 25.473794002607562,
-      "grad_norm": 1.5549969673156738,
-      "learning_rate": 5.1429145728643215e-05,
-      "loss": 5.1297,
-      "step": 48846
-    },
-    {
-      "epoch": 25.47431551499348,
-      "grad_norm": 1.525326132774353,
-      "learning_rate": 5.142814070351759e-05,
-      "loss": 5.1739,
-      "step": 48847
-    },
-    {
-      "epoch": 25.4748370273794,
-      "grad_norm": 1.5185112953186035,
-      "learning_rate": 5.142713567839196e-05,
-      "loss": 4.9045,
-      "step": 48848
-    },
-    {
-      "epoch": 25.47535853976532,
-      "grad_norm": 1.5881303548812866,
-      "learning_rate": 5.1426130653266335e-05,
-      "loss": 5.0601,
-      "step": 48849
-    },
-    {
-      "epoch": 25.47588005215124,
-      "grad_norm": 1.5866230726242065,
-      "learning_rate": 5.14251256281407e-05,
-      "loss": 4.9786,
-      "step": 48850
-    },
-    {
-      "epoch": 25.47640156453716,
-      "grad_norm": 1.5859715938568115,
-      "learning_rate": 5.142412060301508e-05,
-      "loss": 4.9209,
-      "step": 48851
-    },
-    {
-      "epoch": 25.476923076923075,
-      "grad_norm": 1.5441240072250366,
-      "learning_rate": 5.142311557788945e-05,
-      "loss": 4.9609,
-      "step": 48852
-    },
-    {
-      "epoch": 25.477444589308995,
-      "grad_norm": 1.5736846923828125,
-      "learning_rate": 5.1422110552763826e-05,
-      "loss": 5.4321,
-      "step": 48853
-    },
-    {
-      "epoch": 25.477966101694914,
-      "grad_norm": 1.6335018873214722,
-      "learning_rate": 5.142110552763819e-05,
-      "loss": 4.9046,
-      "step": 48854
-    },
-    {
-      "epoch": 25.478487614080834,
-      "grad_norm": 1.6391160488128662,
-      "learning_rate": 5.142010050251257e-05,
-      "loss": 4.9707,
-      "step": 48855
-    },
-    {
-      "epoch": 25.479009126466753,
-      "grad_norm": 1.573941707611084,
-      "learning_rate": 5.141909547738694e-05,
-      "loss": 5.1651,
-      "step": 48856
-    },
-    {
-      "epoch": 25.479530638852673,
-      "grad_norm": 1.5641443729400635,
-      "learning_rate": 5.141809045226132e-05,
-      "loss": 4.8327,
-      "step": 48857
-    },
-    {
-      "epoch": 25.480052151238592,
-      "grad_norm": 1.561559796333313,
-      "learning_rate": 5.141708542713568e-05,
-      "loss": 5.1975,
-      "step": 48858
-    },
-    {
-      "epoch": 25.48057366362451,
-      "grad_norm": 1.516135573387146,
-      "learning_rate": 5.1416080402010045e-05,
-      "loss": 5.1197,
-      "step": 48859
-    },
-    {
-      "epoch": 25.48109517601043,
-      "grad_norm": 1.566941261291504,
-      "learning_rate": 5.141507537688442e-05,
-      "loss": 5.0972,
-      "step": 48860
-    },
-    {
-      "epoch": 25.48161668839635,
-      "grad_norm": 1.4774816036224365,
-      "learning_rate": 5.1414070351758794e-05,
-      "loss": 5.2628,
-      "step": 48861
-    },
-    {
-      "epoch": 25.48213820078227,
-      "grad_norm": 1.5867723226547241,
-      "learning_rate": 5.141306532663317e-05,
-      "loss": 5.0021,
-      "step": 48862
-    },
-    {
-      "epoch": 25.48265971316819,
-      "grad_norm": 1.6100220680236816,
-      "learning_rate": 5.1412060301507536e-05,
-      "loss": 5.3488,
-      "step": 48863
-    },
-    {
-      "epoch": 25.483181225554105,
-      "grad_norm": 1.53816556930542,
-      "learning_rate": 5.1411055276381914e-05,
-      "loss": 5.724,
-      "step": 48864
-    },
-    {
-      "epoch": 25.483702737940025,
-      "grad_norm": 1.6329442262649536,
-      "learning_rate": 5.141005025125628e-05,
-      "loss": 4.896,
-      "step": 48865
-    },
-    {
-      "epoch": 25.484224250325944,
-      "grad_norm": 1.5224145650863647,
-      "learning_rate": 5.1409045226130656e-05,
-      "loss": 5.4182,
-      "step": 48866
-    },
-    {
-      "epoch": 25.484745762711864,
-      "grad_norm": 1.5675692558288574,
-      "learning_rate": 5.140804020100503e-05,
-      "loss": 5.4212,
-      "step": 48867
-    },
-    {
-      "epoch": 25.485267275097783,
-      "grad_norm": 1.588382601737976,
-      "learning_rate": 5.1407035175879405e-05,
-      "loss": 4.6441,
-      "step": 48868
-    },
-    {
-      "epoch": 25.485788787483703,
-      "grad_norm": 1.6388295888900757,
-      "learning_rate": 5.140603015075377e-05,
-      "loss": 5.2317,
-      "step": 48869
-    },
-    {
-      "epoch": 25.486310299869622,
-      "grad_norm": 1.5081381797790527,
-      "learning_rate": 5.140502512562815e-05,
-      "loss": 5.5675,
-      "step": 48870
-    },
-    {
-      "epoch": 25.48683181225554,
-      "grad_norm": 1.542486548423767,
-      "learning_rate": 5.140402010050251e-05,
-      "loss": 5.2376,
-      "step": 48871
-    },
-    {
-      "epoch": 25.48735332464146,
-      "grad_norm": 1.683075189590454,
-      "learning_rate": 5.140301507537689e-05,
-      "loss": 5.0428,
-      "step": 48872
-    },
-    {
-      "epoch": 25.48787483702738,
-      "grad_norm": 1.5263593196868896,
-      "learning_rate": 5.140201005025126e-05,
-      "loss": 4.9742,
-      "step": 48873
-    },
-    {
-      "epoch": 25.4883963494133,
-      "grad_norm": 1.5141197443008423,
-      "learning_rate": 5.1401005025125625e-05,
-      "loss": 4.8841,
-      "step": 48874
-    },
-    {
-      "epoch": 25.48891786179922,
-      "grad_norm": 1.7149105072021484,
-      "learning_rate": 5.14e-05,
-      "loss": 4.5746,
-      "step": 48875
-    },
-    {
-      "epoch": 25.489439374185135,
-      "grad_norm": 1.5876408815383911,
-      "learning_rate": 5.1398994974874374e-05,
-      "loss": 5.0164,
-      "step": 48876
-    },
-    {
-      "epoch": 25.489960886571055,
-      "grad_norm": 1.4524147510528564,
-      "learning_rate": 5.139798994974875e-05,
-      "loss": 5.1354,
-      "step": 48877
-    },
-    {
-      "epoch": 25.490482398956974,
-      "grad_norm": 1.5720962285995483,
-      "learning_rate": 5.1396984924623116e-05,
-      "loss": 5.2437,
-      "step": 48878
-    },
-    {
-      "epoch": 25.491003911342894,
-      "grad_norm": 1.5880882740020752,
-      "learning_rate": 5.1395979899497494e-05,
-      "loss": 5.0448,
-      "step": 48879
-    },
-    {
-      "epoch": 25.491525423728813,
-      "grad_norm": 1.5590746402740479,
-      "learning_rate": 5.139497487437186e-05,
-      "loss": 5.35,
-      "step": 48880
-    },
-    {
-      "epoch": 25.492046936114733,
-      "grad_norm": 1.5714714527130127,
-      "learning_rate": 5.1393969849246236e-05,
-      "loss": 5.4067,
-      "step": 48881
-    },
-    {
-      "epoch": 25.492568448500652,
-      "grad_norm": 1.4915602207183838,
-      "learning_rate": 5.139296482412061e-05,
-      "loss": 5.4174,
-      "step": 48882
-    },
-    {
-      "epoch": 25.49308996088657,
-      "grad_norm": 1.5792478322982788,
-      "learning_rate": 5.1391959798994985e-05,
-      "loss": 5.1946,
-      "step": 48883
-    },
-    {
-      "epoch": 25.49361147327249,
-      "grad_norm": 1.5718097686767578,
-      "learning_rate": 5.139095477386935e-05,
-      "loss": 5.1431,
-      "step": 48884
-    },
-    {
-      "epoch": 25.49413298565841,
-      "grad_norm": 1.6049352884292603,
-      "learning_rate": 5.138994974874373e-05,
-      "loss": 5.1031,
-      "step": 48885
-    },
-    {
-      "epoch": 25.49465449804433,
-      "grad_norm": 1.562516212463379,
-      "learning_rate": 5.138894472361809e-05,
-      "loss": 5.2154,
-      "step": 48886
-    },
-    {
-      "epoch": 25.49517601043025,
-      "grad_norm": 1.6091045141220093,
-      "learning_rate": 5.138793969849246e-05,
-      "loss": 4.7574,
-      "step": 48887
-    },
-    {
-      "epoch": 25.495697522816165,
-      "grad_norm": 1.6490730047225952,
-      "learning_rate": 5.138693467336684e-05,
-      "loss": 5.3149,
-      "step": 48888
-    },
-    {
-      "epoch": 25.496219035202085,
-      "grad_norm": 1.4985495805740356,
-      "learning_rate": 5.1385929648241204e-05,
-      "loss": 5.5153,
-      "step": 48889
-    },
-    {
-      "epoch": 25.496740547588004,
-      "grad_norm": 1.5908024311065674,
-      "learning_rate": 5.138492462311558e-05,
-      "loss": 5.3134,
-      "step": 48890
-    },
-    {
-      "epoch": 25.497262059973924,
-      "grad_norm": 1.5000015497207642,
-      "learning_rate": 5.1383919597989946e-05,
-      "loss": 4.9861,
-      "step": 48891
-    },
-    {
-      "epoch": 25.497783572359843,
-      "grad_norm": 1.577677607536316,
-      "learning_rate": 5.1382914572864324e-05,
-      "loss": 5.1195,
-      "step": 48892
-    },
-    {
-      "epoch": 25.498305084745763,
-      "grad_norm": 1.6747487783432007,
-      "learning_rate": 5.1381909547738695e-05,
-      "loss": 5.0455,
-      "step": 48893
-    },
-    {
-      "epoch": 25.498826597131682,
-      "grad_norm": 1.5834234952926636,
-      "learning_rate": 5.138090452261307e-05,
-      "loss": 5.1386,
-      "step": 48894
-    },
-    {
-      "epoch": 25.4993481095176,
-      "grad_norm": 1.6298543214797974,
-      "learning_rate": 5.137989949748744e-05,
-      "loss": 5.1563,
-      "step": 48895
-    },
-    {
-      "epoch": 25.49986962190352,
-      "grad_norm": 1.681318998336792,
-      "learning_rate": 5.1378894472361815e-05,
-      "loss": 5.2451,
-      "step": 48896
-    },
-    {
-      "epoch": 25.50039113428944,
-      "grad_norm": 1.511142373085022,
-      "learning_rate": 5.1377889447236186e-05,
-      "loss": 5.0406,
-      "step": 48897
-    },
-    {
-      "epoch": 25.50091264667536,
-      "grad_norm": 1.4955912828445435,
-      "learning_rate": 5.1376884422110564e-05,
-      "loss": 4.6995,
-      "step": 48898
-    },
-    {
-      "epoch": 25.50143415906128,
-      "grad_norm": 1.5270886421203613,
-      "learning_rate": 5.137587939698493e-05,
-      "loss": 5.1926,
-      "step": 48899
-    },
-    {
-      "epoch": 25.501955671447195,
-      "grad_norm": 1.5147688388824463,
-      "learning_rate": 5.137487437185929e-05,
-      "loss": 5.3912,
-      "step": 48900
-    },
-    {
-      "epoch": 25.502477183833115,
-      "grad_norm": 1.709395408630371,
-      "learning_rate": 5.137386934673367e-05,
-      "loss": 4.8436,
-      "step": 48901
-    },
-    {
-      "epoch": 25.502998696219034,
-      "grad_norm": 1.6178325414657593,
-      "learning_rate": 5.137286432160804e-05,
-      "loss": 5.4644,
-      "step": 48902
-    },
-    {
-      "epoch": 25.503520208604954,
-      "grad_norm": 1.632393717765808,
-      "learning_rate": 5.137185929648242e-05,
-      "loss": 5.1787,
-      "step": 48903
-    },
-    {
-      "epoch": 25.504041720990873,
-      "grad_norm": 1.6363539695739746,
-      "learning_rate": 5.137085427135678e-05,
-      "loss": 5.6988,
-      "step": 48904
-    },
-    {
-      "epoch": 25.504563233376793,
-      "grad_norm": 1.653057336807251,
-      "learning_rate": 5.136984924623116e-05,
-      "loss": 5.354,
-      "step": 48905
-    },
-    {
-      "epoch": 25.505084745762712,
-      "grad_norm": 1.5960164070129395,
-      "learning_rate": 5.1368844221105525e-05,
-      "loss": 5.2744,
-      "step": 48906
-    },
-    {
-      "epoch": 25.50560625814863,
-      "grad_norm": 1.5465582609176636,
-      "learning_rate": 5.13678391959799e-05,
-      "loss": 5.8128,
-      "step": 48907
-    },
-    {
-      "epoch": 25.50612777053455,
-      "grad_norm": 1.63881254196167,
-      "learning_rate": 5.1366834170854274e-05,
-      "loss": 4.86,
-      "step": 48908
-    },
-    {
-      "epoch": 25.50664928292047,
-      "grad_norm": 1.6568875312805176,
-      "learning_rate": 5.136582914572865e-05,
-      "loss": 4.8478,
-      "step": 48909
-    },
-    {
-      "epoch": 25.50717079530639,
-      "grad_norm": 1.5548490285873413,
-      "learning_rate": 5.1364824120603016e-05,
-      "loss": 5.1717,
-      "step": 48910
-    },
-    {
-      "epoch": 25.50769230769231,
-      "grad_norm": 1.6086336374282837,
-      "learning_rate": 5.1363819095477394e-05,
-      "loss": 4.8986,
-      "step": 48911
-    },
-    {
-      "epoch": 25.508213820078225,
-      "grad_norm": 1.631535291671753,
-      "learning_rate": 5.136281407035176e-05,
-      "loss": 5.3861,
-      "step": 48912
-    },
-    {
-      "epoch": 25.508735332464145,
-      "grad_norm": 1.6927114725112915,
-      "learning_rate": 5.136180904522613e-05,
-      "loss": 5.2206,
-      "step": 48913
-    },
-    {
-      "epoch": 25.509256844850064,
-      "grad_norm": 1.5992997884750366,
-      "learning_rate": 5.136080402010051e-05,
-      "loss": 5.4473,
-      "step": 48914
-    },
-    {
-      "epoch": 25.509778357235984,
-      "grad_norm": 1.5967971086502075,
-      "learning_rate": 5.135979899497487e-05,
-      "loss": 4.9,
-      "step": 48915
-    },
-    {
-      "epoch": 25.510299869621903,
-      "grad_norm": 1.7133796215057373,
-      "learning_rate": 5.135879396984925e-05,
-      "loss": 5.0104,
-      "step": 48916
-    },
-    {
-      "epoch": 25.510821382007823,
-      "grad_norm": 1.5564383268356323,
-      "learning_rate": 5.1357788944723614e-05,
-      "loss": 5.3011,
-      "step": 48917
-    },
-    {
-      "epoch": 25.511342894393742,
-      "grad_norm": 1.6722098588943481,
-      "learning_rate": 5.135678391959799e-05,
-      "loss": 4.6525,
-      "step": 48918
-    },
-    {
-      "epoch": 25.51186440677966,
-      "grad_norm": 1.6176867485046387,
-      "learning_rate": 5.135577889447236e-05,
-      "loss": 4.7227,
-      "step": 48919
-    },
-    {
-      "epoch": 25.51238591916558,
-      "grad_norm": 1.4190634489059448,
-      "learning_rate": 5.135477386934674e-05,
-      "loss": 5.121,
-      "step": 48920
-    },
-    {
-      "epoch": 25.5129074315515,
-      "grad_norm": 1.57921302318573,
-      "learning_rate": 5.1353768844221105e-05,
-      "loss": 4.9851,
-      "step": 48921
-    },
-    {
-      "epoch": 25.51342894393742,
-      "grad_norm": 1.644585371017456,
-      "learning_rate": 5.135276381909548e-05,
-      "loss": 5.2725,
-      "step": 48922
-    },
-    {
-      "epoch": 25.513950456323336,
-      "grad_norm": 1.5846506357192993,
-      "learning_rate": 5.1351758793969854e-05,
-      "loss": 5.2033,
-      "step": 48923
-    },
-    {
-      "epoch": 25.514471968709255,
-      "grad_norm": 1.4948312044143677,
-      "learning_rate": 5.135075376884423e-05,
-      "loss": 5.336,
-      "step": 48924
-    },
-    {
-      "epoch": 25.514993481095175,
-      "grad_norm": 1.679413914680481,
-      "learning_rate": 5.1349748743718596e-05,
-      "loss": 5.0964,
-      "step": 48925
-    },
-    {
-      "epoch": 25.515514993481094,
-      "grad_norm": 1.7198975086212158,
-      "learning_rate": 5.134874371859296e-05,
-      "loss": 5.0451,
-      "step": 48926
-    },
-    {
-      "epoch": 25.516036505867014,
-      "grad_norm": 1.4578702449798584,
-      "learning_rate": 5.134773869346734e-05,
-      "loss": 5.3018,
-      "step": 48927
-    },
-    {
-      "epoch": 25.516558018252933,
-      "grad_norm": 1.6058918237686157,
-      "learning_rate": 5.134673366834171e-05,
-      "loss": 5.5306,
-      "step": 48928
-    },
-    {
-      "epoch": 25.517079530638853,
-      "grad_norm": 1.4843854904174805,
-      "learning_rate": 5.134572864321609e-05,
-      "loss": 5.2767,
-      "step": 48929
-    },
-    {
-      "epoch": 25.517601043024772,
-      "grad_norm": 1.6361196041107178,
-      "learning_rate": 5.134472361809045e-05,
-      "loss": 4.9591,
-      "step": 48930
-    },
-    {
-      "epoch": 25.51812255541069,
-      "grad_norm": 1.5310431718826294,
-      "learning_rate": 5.134371859296483e-05,
-      "loss": 5.4363,
-      "step": 48931
-    },
-    {
-      "epoch": 25.51864406779661,
-      "grad_norm": 1.5858047008514404,
-      "learning_rate": 5.134271356783919e-05,
-      "loss": 5.3242,
-      "step": 48932
-    },
-    {
-      "epoch": 25.51916558018253,
-      "grad_norm": 1.4427858591079712,
-      "learning_rate": 5.134170854271357e-05,
-      "loss": 4.874,
-      "step": 48933
-    },
-    {
-      "epoch": 25.51968709256845,
-      "grad_norm": 1.6017920970916748,
-      "learning_rate": 5.134070351758794e-05,
-      "loss": 5.4163,
-      "step": 48934
-    },
-    {
-      "epoch": 25.52020860495437,
-      "grad_norm": 1.5858510732650757,
-      "learning_rate": 5.133969849246232e-05,
-      "loss": 5.0746,
-      "step": 48935
-    },
-    {
-      "epoch": 25.520730117340285,
-      "grad_norm": 1.6703598499298096,
-      "learning_rate": 5.1338693467336684e-05,
-      "loss": 5.26,
-      "step": 48936
-    },
-    {
-      "epoch": 25.521251629726205,
-      "grad_norm": 1.5491387844085693,
-      "learning_rate": 5.133768844221106e-05,
-      "loss": 4.7892,
-      "step": 48937
-    },
-    {
-      "epoch": 25.521773142112124,
-      "grad_norm": 1.5607072114944458,
-      "learning_rate": 5.1336683417085426e-05,
-      "loss": 5.6935,
-      "step": 48938
-    },
-    {
-      "epoch": 25.522294654498044,
-      "grad_norm": 1.698013424873352,
-      "learning_rate": 5.13356783919598e-05,
-      "loss": 5.3044,
-      "step": 48939
-    },
-    {
-      "epoch": 25.522816166883963,
-      "grad_norm": 1.5756579637527466,
-      "learning_rate": 5.1334673366834175e-05,
-      "loss": 5.1928,
-      "step": 48940
-    },
-    {
-      "epoch": 25.523337679269883,
-      "grad_norm": 1.5580360889434814,
-      "learning_rate": 5.133366834170854e-05,
-      "loss": 5.1655,
-      "step": 48941
-    },
-    {
-      "epoch": 25.523859191655802,
-      "grad_norm": 1.5623502731323242,
-      "learning_rate": 5.133266331658292e-05,
-      "loss": 5.0914,
-      "step": 48942
-    },
-    {
-      "epoch": 25.52438070404172,
-      "grad_norm": 1.6716681718826294,
-      "learning_rate": 5.133165829145729e-05,
-      "loss": 4.9984,
-      "step": 48943
-    },
-    {
-      "epoch": 25.52490221642764,
-      "grad_norm": 1.5713835954666138,
-      "learning_rate": 5.1330653266331666e-05,
-      "loss": 5.2237,
-      "step": 48944
-    },
-    {
-      "epoch": 25.52542372881356,
-      "grad_norm": 1.6077550649642944,
-      "learning_rate": 5.132964824120603e-05,
-      "loss": 5.1262,
-      "step": 48945
-    },
-    {
-      "epoch": 25.52594524119948,
-      "grad_norm": 1.474477767944336,
-      "learning_rate": 5.132864321608041e-05,
-      "loss": 5.0913,
-      "step": 48946
-    },
-    {
-      "epoch": 25.526466753585396,
-      "grad_norm": 1.5703102350234985,
-      "learning_rate": 5.132763819095477e-05,
-      "loss": 5.2446,
-      "step": 48947
-    },
-    {
-      "epoch": 25.526988265971315,
-      "grad_norm": 1.5775409936904907,
-      "learning_rate": 5.132663316582915e-05,
-      "loss": 5.0805,
-      "step": 48948
-    },
-    {
-      "epoch": 25.527509778357235,
-      "grad_norm": 1.5041606426239014,
-      "learning_rate": 5.132562814070352e-05,
-      "loss": 5.6556,
-      "step": 48949
-    },
-    {
-      "epoch": 25.528031290743154,
-      "grad_norm": 1.6099536418914795,
-      "learning_rate": 5.13246231155779e-05,
-      "loss": 5.1549,
-      "step": 48950
-    },
-    {
-      "epoch": 25.528552803129074,
-      "grad_norm": 1.5015314817428589,
-      "learning_rate": 5.1323618090452263e-05,
-      "loss": 5.4397,
-      "step": 48951
-    },
-    {
-      "epoch": 25.529074315514993,
-      "grad_norm": 1.7202929258346558,
-      "learning_rate": 5.132261306532663e-05,
-      "loss": 4.9111,
-      "step": 48952
-    },
-    {
-      "epoch": 25.529595827900913,
-      "grad_norm": 1.563065528869629,
-      "learning_rate": 5.1321608040201006e-05,
-      "loss": 5.0855,
-      "step": 48953
-    },
-    {
-      "epoch": 25.530117340286832,
-      "grad_norm": 1.7508015632629395,
-      "learning_rate": 5.132060301507538e-05,
-      "loss": 5.2483,
-      "step": 48954
-    },
-    {
-      "epoch": 25.53063885267275,
-      "grad_norm": 1.4619090557098389,
-      "learning_rate": 5.1319597989949754e-05,
-      "loss": 5.3557,
-      "step": 48955
-    },
-    {
-      "epoch": 25.53116036505867,
-      "grad_norm": 1.5642729997634888,
-      "learning_rate": 5.131859296482412e-05,
-      "loss": 5.4751,
-      "step": 48956
-    },
-    {
-      "epoch": 25.53168187744459,
-      "grad_norm": 1.4821125268936157,
-      "learning_rate": 5.1317587939698497e-05,
-      "loss": 5.8119,
-      "step": 48957
-    },
-    {
-      "epoch": 25.53220338983051,
-      "grad_norm": 2.3511264324188232,
-      "learning_rate": 5.131658291457286e-05,
-      "loss": 4.7796,
-      "step": 48958
-    },
-    {
-      "epoch": 25.532724902216426,
-      "grad_norm": 1.6256109476089478,
-      "learning_rate": 5.131557788944724e-05,
-      "loss": 5.5253,
-      "step": 48959
-    },
-    {
-      "epoch": 25.533246414602345,
-      "grad_norm": 1.6675409078598022,
-      "learning_rate": 5.131457286432161e-05,
-      "loss": 5.393,
-      "step": 48960
-    },
-    {
-      "epoch": 25.533767926988265,
-      "grad_norm": 1.6966451406478882,
-      "learning_rate": 5.131356783919599e-05,
-      "loss": 4.7059,
-      "step": 48961
-    },
-    {
-      "epoch": 25.534289439374184,
-      "grad_norm": 1.6025164127349854,
-      "learning_rate": 5.131256281407035e-05,
-      "loss": 4.961,
-      "step": 48962
-    },
-    {
-      "epoch": 25.534810951760104,
-      "grad_norm": 1.642952561378479,
-      "learning_rate": 5.131155778894473e-05,
-      "loss": 5.2535,
-      "step": 48963
-    },
-    {
-      "epoch": 25.535332464146023,
-      "grad_norm": 1.6073254346847534,
-      "learning_rate": 5.13105527638191e-05,
-      "loss": 4.9832,
-      "step": 48964
-    },
-    {
-      "epoch": 25.535853976531943,
-      "grad_norm": 1.545487403869629,
-      "learning_rate": 5.130954773869348e-05,
-      "loss": 5.4233,
-      "step": 48965
-    },
-    {
-      "epoch": 25.536375488917862,
-      "grad_norm": 1.62546968460083,
-      "learning_rate": 5.130854271356784e-05,
-      "loss": 5.2636,
-      "step": 48966
-    },
-    {
-      "epoch": 25.53689700130378,
-      "grad_norm": 1.645876169204712,
-      "learning_rate": 5.130753768844221e-05,
-      "loss": 4.544,
-      "step": 48967
-    },
-    {
-      "epoch": 25.5374185136897,
-      "grad_norm": 1.5837656259536743,
-      "learning_rate": 5.1306532663316585e-05,
-      "loss": 5.3983,
-      "step": 48968
-    },
-    {
-      "epoch": 25.53794002607562,
-      "grad_norm": 1.5915272235870361,
-      "learning_rate": 5.1305527638190956e-05,
-      "loss": 4.7639,
-      "step": 48969
-    },
-    {
-      "epoch": 25.53846153846154,
-      "grad_norm": 1.5823966264724731,
-      "learning_rate": 5.1304522613065334e-05,
-      "loss": 5.3679,
-      "step": 48970
-    },
-    {
-      "epoch": 25.538983050847456,
-      "grad_norm": 1.5275877714157104,
-      "learning_rate": 5.13035175879397e-05,
-      "loss": 5.5351,
-      "step": 48971
-    },
-    {
-      "epoch": 25.539504563233375,
-      "grad_norm": 1.6859833002090454,
-      "learning_rate": 5.1302512562814076e-05,
-      "loss": 5.3203,
-      "step": 48972
-    },
-    {
-      "epoch": 25.540026075619295,
-      "grad_norm": 1.5309557914733887,
-      "learning_rate": 5.130150753768844e-05,
-      "loss": 5.6007,
-      "step": 48973
-    },
-    {
-      "epoch": 25.540547588005214,
-      "grad_norm": 1.564135193824768,
-      "learning_rate": 5.130050251256282e-05,
-      "loss": 5.0097,
-      "step": 48974
-    },
-    {
-      "epoch": 25.541069100391134,
-      "grad_norm": 1.503042459487915,
-      "learning_rate": 5.129949748743719e-05,
-      "loss": 5.3741,
-      "step": 48975
-    },
-    {
-      "epoch": 25.541590612777053,
-      "grad_norm": 1.605412244796753,
-      "learning_rate": 5.129849246231157e-05,
-      "loss": 4.6807,
-      "step": 48976
-    },
-    {
-      "epoch": 25.542112125162973,
-      "grad_norm": 1.5730574131011963,
-      "learning_rate": 5.129748743718593e-05,
-      "loss": 4.9591,
-      "step": 48977
-    },
-    {
-      "epoch": 25.542633637548892,
-      "grad_norm": 1.6715433597564697,
-      "learning_rate": 5.129648241206031e-05,
-      "loss": 5.0996,
-      "step": 48978
-    },
-    {
-      "epoch": 25.54315514993481,
-      "grad_norm": 1.5096261501312256,
-      "learning_rate": 5.129547738693467e-05,
-      "loss": 4.9399,
-      "step": 48979
-    },
-    {
-      "epoch": 25.54367666232073,
-      "grad_norm": 1.5704644918441772,
-      "learning_rate": 5.1294472361809044e-05,
-      "loss": 5.0699,
-      "step": 48980
-    },
-    {
-      "epoch": 25.54419817470665,
-      "grad_norm": 1.5576554536819458,
-      "learning_rate": 5.129346733668342e-05,
-      "loss": 5.112,
-      "step": 48981
-    },
-    {
-      "epoch": 25.54471968709257,
-      "grad_norm": 1.5142595767974854,
-      "learning_rate": 5.1292462311557786e-05,
-      "loss": 5.5454,
-      "step": 48982
-    },
-    {
-      "epoch": 25.545241199478486,
-      "grad_norm": 1.6365838050842285,
-      "learning_rate": 5.1291457286432164e-05,
-      "loss": 4.8897,
-      "step": 48983
-    },
-    {
-      "epoch": 25.545762711864406,
-      "grad_norm": 1.6057697534561157,
-      "learning_rate": 5.1290452261306535e-05,
-      "loss": 5.2623,
-      "step": 48984
-    },
-    {
-      "epoch": 25.546284224250325,
-      "grad_norm": 1.733473300933838,
-      "learning_rate": 5.128944723618091e-05,
-      "loss": 5.2683,
-      "step": 48985
-    },
-    {
-      "epoch": 25.546805736636244,
-      "grad_norm": 1.6063380241394043,
-      "learning_rate": 5.128844221105528e-05,
-      "loss": 5.2876,
-      "step": 48986
-    },
-    {
-      "epoch": 25.547327249022164,
-      "grad_norm": 1.5926600694656372,
-      "learning_rate": 5.1287437185929655e-05,
-      "loss": 4.9494,
-      "step": 48987
-    },
-    {
-      "epoch": 25.547848761408083,
-      "grad_norm": 1.4451466798782349,
-      "learning_rate": 5.128643216080402e-05,
-      "loss": 5.3811,
-      "step": 48988
-    },
-    {
-      "epoch": 25.548370273794003,
-      "grad_norm": 1.4686737060546875,
-      "learning_rate": 5.12854271356784e-05,
-      "loss": 5.2443,
-      "step": 48989
-    },
-    {
-      "epoch": 25.548891786179922,
-      "grad_norm": 1.7466670274734497,
-      "learning_rate": 5.128442211055277e-05,
-      "loss": 4.7747,
-      "step": 48990
-    },
-    {
-      "epoch": 25.54941329856584,
-      "grad_norm": 1.5338099002838135,
-      "learning_rate": 5.1283417085427146e-05,
-      "loss": 5.4289,
-      "step": 48991
-    },
-    {
-      "epoch": 25.54993481095176,
-      "grad_norm": 1.6174393892288208,
-      "learning_rate": 5.128241206030151e-05,
-      "loss": 4.9862,
-      "step": 48992
-    },
-    {
-      "epoch": 25.55045632333768,
-      "grad_norm": 1.59501314163208,
-      "learning_rate": 5.1281407035175875e-05,
-      "loss": 5.3261,
-      "step": 48993
-    },
-    {
-      "epoch": 25.5509778357236,
-      "grad_norm": 1.598372220993042,
-      "learning_rate": 5.128040201005025e-05,
-      "loss": 5.0401,
-      "step": 48994
-    },
-    {
-      "epoch": 25.551499348109516,
-      "grad_norm": 1.6180822849273682,
-      "learning_rate": 5.1279396984924624e-05,
-      "loss": 5.3037,
-      "step": 48995
-    },
-    {
-      "epoch": 25.552020860495436,
-      "grad_norm": 1.5147316455841064,
-      "learning_rate": 5.1278391959799e-05,
-      "loss": 5.7119,
-      "step": 48996
-    },
-    {
-      "epoch": 25.552542372881355,
-      "grad_norm": 1.6093584299087524,
-      "learning_rate": 5.1277386934673366e-05,
-      "loss": 4.8476,
-      "step": 48997
-    },
-    {
-      "epoch": 25.553063885267274,
-      "grad_norm": 1.6142923831939697,
-      "learning_rate": 5.1276381909547744e-05,
-      "loss": 5.0304,
-      "step": 48998
-    },
-    {
-      "epoch": 25.553585397653194,
-      "grad_norm": 1.555130124092102,
-      "learning_rate": 5.127537688442211e-05,
-      "loss": 4.9458,
-      "step": 48999
-    },
-    {
-      "epoch": 25.554106910039113,
-      "grad_norm": 1.5466177463531494,
-      "learning_rate": 5.1274371859296486e-05,
-      "loss": 5.0993,
-      "step": 49000
-    },
-    {
-      "epoch": 25.554628422425033,
-      "grad_norm": 1.4153964519500732,
-      "learning_rate": 5.127336683417086e-05,
-      "loss": 5.7253,
-      "step": 49001
-    },
-    {
-      "epoch": 25.555149934810952,
-      "grad_norm": 1.541817307472229,
-      "learning_rate": 5.1272361809045235e-05,
-      "loss": 5.2991,
-      "step": 49002
-    },
-    {
-      "epoch": 25.555671447196872,
-      "grad_norm": 1.7319304943084717,
-      "learning_rate": 5.12713567839196e-05,
-      "loss": 4.9992,
-      "step": 49003
-    },
-    {
-      "epoch": 25.55619295958279,
-      "grad_norm": 1.7234410047531128,
-      "learning_rate": 5.127035175879398e-05,
-      "loss": 5.1853,
-      "step": 49004
-    },
-    {
-      "epoch": 25.55671447196871,
-      "grad_norm": 1.7207286357879639,
-      "learning_rate": 5.126934673366834e-05,
-      "loss": 5.0319,
-      "step": 49005
-    },
-    {
-      "epoch": 25.557235984354627,
-      "grad_norm": 1.4930751323699951,
-      "learning_rate": 5.126834170854271e-05,
-      "loss": 5.4419,
-      "step": 49006
-    },
-    {
-      "epoch": 25.557757496740546,
-      "grad_norm": 1.4706486463546753,
-      "learning_rate": 5.126733668341709e-05,
-      "loss": 5.2305,
-      "step": 49007
-    },
-    {
-      "epoch": 25.558279009126466,
-      "grad_norm": 1.5037871599197388,
-      "learning_rate": 5.1266331658291454e-05,
-      "loss": 5.3633,
-      "step": 49008
-    },
-    {
-      "epoch": 25.558800521512385,
-      "grad_norm": 1.6010459661483765,
-      "learning_rate": 5.126532663316583e-05,
-      "loss": 5.5258,
-      "step": 49009
-    },
-    {
-      "epoch": 25.559322033898304,
-      "grad_norm": 1.5585507154464722,
-      "learning_rate": 5.12643216080402e-05,
-      "loss": 4.9245,
-      "step": 49010
-    },
-    {
-      "epoch": 25.559843546284224,
-      "grad_norm": 1.5710432529449463,
-      "learning_rate": 5.126331658291458e-05,
-      "loss": 5.3101,
-      "step": 49011
-    },
-    {
-      "epoch": 25.560365058670143,
-      "grad_norm": 1.6948398351669312,
-      "learning_rate": 5.1262311557788945e-05,
-      "loss": 5.1777,
-      "step": 49012
-    },
-    {
-      "epoch": 25.560886571056063,
-      "grad_norm": 1.6445525884628296,
-      "learning_rate": 5.126130653266332e-05,
-      "loss": 5.2708,
-      "step": 49013
-    },
-    {
-      "epoch": 25.561408083441982,
-      "grad_norm": 1.7027512788772583,
-      "learning_rate": 5.126030150753769e-05,
-      "loss": 4.9081,
-      "step": 49014
-    },
-    {
-      "epoch": 25.561929595827902,
-      "grad_norm": 1.6997045278549194,
-      "learning_rate": 5.1259296482412065e-05,
-      "loss": 4.7336,
-      "step": 49015
-    },
-    {
-      "epoch": 25.56245110821382,
-      "grad_norm": 1.702027440071106,
-      "learning_rate": 5.1258291457286436e-05,
-      "loss": 5.2344,
-      "step": 49016
-    },
-    {
-      "epoch": 25.56297262059974,
-      "grad_norm": 1.5749287605285645,
-      "learning_rate": 5.1257286432160814e-05,
-      "loss": 5.3817,
-      "step": 49017
-    },
-    {
-      "epoch": 25.56349413298566,
-      "grad_norm": 1.6183366775512695,
-      "learning_rate": 5.125628140703518e-05,
-      "loss": 5.1276,
-      "step": 49018
-    },
-    {
-      "epoch": 25.564015645371576,
-      "grad_norm": 1.6305228471755981,
-      "learning_rate": 5.125527638190954e-05,
-      "loss": 5.3911,
-      "step": 49019
-    },
-    {
-      "epoch": 25.564537157757496,
-      "grad_norm": 1.6615885496139526,
-      "learning_rate": 5.125427135678392e-05,
-      "loss": 4.9578,
-      "step": 49020
-    },
-    {
-      "epoch": 25.565058670143415,
-      "grad_norm": 1.440714716911316,
-      "learning_rate": 5.125326633165829e-05,
-      "loss": 5.2327,
-      "step": 49021
-    },
-    {
-      "epoch": 25.565580182529335,
-      "grad_norm": 1.645450234413147,
-      "learning_rate": 5.125226130653267e-05,
-      "loss": 5.0039,
-      "step": 49022
-    },
-    {
-      "epoch": 25.566101694915254,
-      "grad_norm": 1.5840418338775635,
-      "learning_rate": 5.1251256281407033e-05,
-      "loss": 4.8198,
-      "step": 49023
-    },
-    {
-      "epoch": 25.566623207301173,
-      "grad_norm": 1.5111219882965088,
-      "learning_rate": 5.125025125628141e-05,
-      "loss": 5.809,
-      "step": 49024
-    },
-    {
-      "epoch": 25.567144719687093,
-      "grad_norm": 1.5315684080123901,
-      "learning_rate": 5.1249246231155775e-05,
-      "loss": 5.1731,
-      "step": 49025
-    },
-    {
-      "epoch": 25.567666232073012,
-      "grad_norm": 1.6320276260375977,
-      "learning_rate": 5.124824120603015e-05,
-      "loss": 5.4076,
-      "step": 49026
-    },
-    {
-      "epoch": 25.568187744458932,
-      "grad_norm": 1.5277915000915527,
-      "learning_rate": 5.1247236180904524e-05,
-      "loss": 5.2828,
-      "step": 49027
-    },
-    {
-      "epoch": 25.56870925684485,
-      "grad_norm": 1.5580967664718628,
-      "learning_rate": 5.12462311557789e-05,
-      "loss": 5.1776,
-      "step": 49028
-    },
-    {
-      "epoch": 25.56923076923077,
-      "grad_norm": 1.580306053161621,
-      "learning_rate": 5.1245226130653266e-05,
-      "loss": 5.1172,
-      "step": 49029
-    },
-    {
-      "epoch": 25.569752281616687,
-      "grad_norm": 1.7054685354232788,
-      "learning_rate": 5.1244221105527644e-05,
-      "loss": 4.6018,
-      "step": 49030
-    },
-    {
-      "epoch": 25.570273794002606,
-      "grad_norm": 1.6213661432266235,
-      "learning_rate": 5.1243216080402015e-05,
-      "loss": 5.0899,
-      "step": 49031
-    },
-    {
-      "epoch": 25.570795306388526,
-      "grad_norm": 1.5690258741378784,
-      "learning_rate": 5.124221105527638e-05,
-      "loss": 4.5694,
-      "step": 49032
-    },
-    {
-      "epoch": 25.571316818774445,
-      "grad_norm": 1.5594779253005981,
-      "learning_rate": 5.124120603015076e-05,
-      "loss": 5.0868,
-      "step": 49033
-    },
-    {
-      "epoch": 25.571838331160365,
-      "grad_norm": 1.7052491903305054,
-      "learning_rate": 5.124020100502512e-05,
-      "loss": 4.7867,
-      "step": 49034
-    },
-    {
-      "epoch": 25.572359843546284,
-      "grad_norm": 1.5800230503082275,
-      "learning_rate": 5.12391959798995e-05,
-      "loss": 5.4971,
-      "step": 49035
-    },
-    {
-      "epoch": 25.572881355932203,
-      "grad_norm": 1.5323662757873535,
-      "learning_rate": 5.123819095477387e-05,
-      "loss": 4.637,
-      "step": 49036
-    },
-    {
-      "epoch": 25.573402868318123,
-      "grad_norm": 1.5504729747772217,
-      "learning_rate": 5.123718592964825e-05,
-      "loss": 5.3233,
-      "step": 49037
-    },
-    {
-      "epoch": 25.573924380704042,
-      "grad_norm": 1.491260051727295,
-      "learning_rate": 5.123618090452261e-05,
-      "loss": 5.0176,
-      "step": 49038
-    },
-    {
-      "epoch": 25.574445893089962,
-      "grad_norm": 1.3582632541656494,
-      "learning_rate": 5.123517587939699e-05,
-      "loss": 4.8532,
-      "step": 49039
-    },
-    {
-      "epoch": 25.57496740547588,
-      "grad_norm": 1.5719590187072754,
-      "learning_rate": 5.1234170854271355e-05,
-      "loss": 4.8876,
-      "step": 49040
-    },
-    {
-      "epoch": 25.5754889178618,
-      "grad_norm": 1.6891262531280518,
-      "learning_rate": 5.123316582914573e-05,
-      "loss": 5.2117,
-      "step": 49041
-    },
-    {
-      "epoch": 25.576010430247717,
-      "grad_norm": 1.5597712993621826,
-      "learning_rate": 5.1232160804020104e-05,
-      "loss": 5.3418,
-      "step": 49042
-    },
-    {
-      "epoch": 25.576531942633636,
-      "grad_norm": 1.5350569486618042,
-      "learning_rate": 5.123115577889448e-05,
-      "loss": 5.3075,
-      "step": 49043
-    },
-    {
-      "epoch": 25.577053455019556,
-      "grad_norm": 1.5435817241668701,
-      "learning_rate": 5.1230150753768846e-05,
-      "loss": 5.0219,
-      "step": 49044
-    },
-    {
-      "epoch": 25.577574967405475,
-      "grad_norm": 1.5865895748138428,
-      "learning_rate": 5.122914572864321e-05,
-      "loss": 4.7047,
-      "step": 49045
-    },
-    {
-      "epoch": 25.578096479791395,
-      "grad_norm": 1.695312261581421,
-      "learning_rate": 5.122814070351759e-05,
-      "loss": 4.928,
-      "step": 49046
-    },
-    {
-      "epoch": 25.578617992177314,
-      "grad_norm": 1.5961538553237915,
-      "learning_rate": 5.122713567839196e-05,
-      "loss": 5.2791,
-      "step": 49047
-    },
-    {
-      "epoch": 25.579139504563233,
-      "grad_norm": 1.632523536682129,
-      "learning_rate": 5.122613065326634e-05,
-      "loss": 5.7174,
-      "step": 49048
-    },
-    {
-      "epoch": 25.579661016949153,
-      "grad_norm": 1.7493418455123901,
-      "learning_rate": 5.12251256281407e-05,
-      "loss": 4.821,
-      "step": 49049
-    },
-    {
-      "epoch": 25.580182529335072,
-      "grad_norm": 1.600051999092102,
-      "learning_rate": 5.122412060301508e-05,
-      "loss": 5.0546,
-      "step": 49050
-    },
-    {
-      "epoch": 25.580704041720992,
-      "grad_norm": 1.5523103475570679,
-      "learning_rate": 5.122311557788945e-05,
-      "loss": 5.2628,
-      "step": 49051
-    },
-    {
-      "epoch": 25.58122555410691,
-      "grad_norm": 1.5774461030960083,
-      "learning_rate": 5.122211055276383e-05,
-      "loss": 5.5472,
-      "step": 49052
-    },
-    {
-      "epoch": 25.58174706649283,
-      "grad_norm": 1.658156156539917,
-      "learning_rate": 5.122110552763819e-05,
-      "loss": 4.5366,
-      "step": 49053
-    },
-    {
-      "epoch": 25.582268578878747,
-      "grad_norm": 1.5598790645599365,
-      "learning_rate": 5.122010050251257e-05,
-      "loss": 5.0243,
-      "step": 49054
-    },
-    {
-      "epoch": 25.582790091264666,
-      "grad_norm": 1.5811610221862793,
-      "learning_rate": 5.1219095477386934e-05,
-      "loss": 5.2329,
-      "step": 49055
-    },
-    {
-      "epoch": 25.583311603650586,
-      "grad_norm": 1.669939398765564,
-      "learning_rate": 5.121809045226131e-05,
-      "loss": 5.4969,
-      "step": 49056
-    },
-    {
-      "epoch": 25.583833116036505,
-      "grad_norm": 1.7163331508636475,
-      "learning_rate": 5.121708542713568e-05,
-      "loss": 5.3208,
-      "step": 49057
-    },
-    {
-      "epoch": 25.584354628422425,
-      "grad_norm": 1.539265513420105,
-      "learning_rate": 5.121608040201006e-05,
-      "loss": 4.7531,
-      "step": 49058
-    },
-    {
-      "epoch": 25.584876140808344,
-      "grad_norm": 1.5464622974395752,
-      "learning_rate": 5.1215075376884425e-05,
-      "loss": 5.112,
-      "step": 49059
-    },
-    {
-      "epoch": 25.585397653194264,
-      "grad_norm": 1.5035585165023804,
-      "learning_rate": 5.121407035175879e-05,
-      "loss": 5.2482,
-      "step": 49060
-    },
-    {
-      "epoch": 25.585919165580183,
-      "grad_norm": 1.3709064722061157,
-      "learning_rate": 5.121306532663317e-05,
-      "loss": 4.9723,
-      "step": 49061
-    },
-    {
-      "epoch": 25.586440677966102,
-      "grad_norm": 1.5009701251983643,
-      "learning_rate": 5.121206030150754e-05,
-      "loss": 5.7504,
-      "step": 49062
-    },
-    {
-      "epoch": 25.586962190352022,
-      "grad_norm": 1.676430344581604,
-      "learning_rate": 5.1211055276381916e-05,
-      "loss": 5.1274,
-      "step": 49063
-    },
-    {
-      "epoch": 25.58748370273794,
-      "grad_norm": 1.5884846448898315,
-      "learning_rate": 5.121005025125628e-05,
-      "loss": 5.1782,
-      "step": 49064
-    },
-    {
-      "epoch": 25.58800521512386,
-      "grad_norm": 1.6321966648101807,
-      "learning_rate": 5.120904522613066e-05,
-      "loss": 5.4647,
-      "step": 49065
-    },
-    {
-      "epoch": 25.588526727509777,
-      "grad_norm": 1.6178611516952515,
-      "learning_rate": 5.120804020100502e-05,
-      "loss": 5.1295,
-      "step": 49066
-    },
-    {
-      "epoch": 25.589048239895696,
-      "grad_norm": 1.561342716217041,
-      "learning_rate": 5.12070351758794e-05,
-      "loss": 5.2864,
-      "step": 49067
-    },
-    {
-      "epoch": 25.589569752281616,
-      "grad_norm": 1.5370728969573975,
-      "learning_rate": 5.120603015075377e-05,
-      "loss": 5.5172,
-      "step": 49068
-    },
-    {
-      "epoch": 25.590091264667535,
-      "grad_norm": 1.52692711353302,
-      "learning_rate": 5.120502512562815e-05,
-      "loss": 5.5754,
-      "step": 49069
-    },
-    {
-      "epoch": 25.590612777053455,
-      "grad_norm": 1.580615520477295,
-      "learning_rate": 5.1204020100502513e-05,
-      "loss": 4.9255,
-      "step": 49070
-    },
-    {
-      "epoch": 25.591134289439374,
-      "grad_norm": 1.5649259090423584,
-      "learning_rate": 5.120301507537689e-05,
-      "loss": 5.3427,
-      "step": 49071
-    },
-    {
-      "epoch": 25.591655801825294,
-      "grad_norm": 1.5840065479278564,
-      "learning_rate": 5.1202010050251256e-05,
-      "loss": 5.4155,
-      "step": 49072
-    },
-    {
-      "epoch": 25.592177314211213,
-      "grad_norm": 1.5581624507904053,
-      "learning_rate": 5.120100502512563e-05,
-      "loss": 5.1219,
-      "step": 49073
-    },
-    {
-      "epoch": 25.592698826597132,
-      "grad_norm": 1.5416855812072754,
-      "learning_rate": 5.1200000000000004e-05,
-      "loss": 5.3425,
-      "step": 49074
-    },
-    {
-      "epoch": 25.593220338983052,
-      "grad_norm": 1.780268669128418,
-      "learning_rate": 5.119899497487437e-05,
-      "loss": 4.9414,
-      "step": 49075
-    },
-    {
-      "epoch": 25.59374185136897,
-      "grad_norm": 1.577286958694458,
-      "learning_rate": 5.1197989949748747e-05,
-      "loss": 5.0229,
-      "step": 49076
-    },
-    {
-      "epoch": 25.59426336375489,
-      "grad_norm": 1.6103581190109253,
-      "learning_rate": 5.119698492462312e-05,
-      "loss": 5.0205,
-      "step": 49077
-    },
-    {
-      "epoch": 25.594784876140807,
-      "grad_norm": 1.492685079574585,
-      "learning_rate": 5.1195979899497495e-05,
-      "loss": 4.9954,
-      "step": 49078
-    },
-    {
-      "epoch": 25.595306388526726,
-      "grad_norm": 1.5804518461227417,
-      "learning_rate": 5.119497487437186e-05,
-      "loss": 5.5262,
-      "step": 49079
-    },
-    {
-      "epoch": 25.595827900912646,
-      "grad_norm": 1.6918801069259644,
-      "learning_rate": 5.119396984924624e-05,
-      "loss": 4.8448,
-      "step": 49080
-    },
-    {
-      "epoch": 25.596349413298565,
-      "grad_norm": 1.5641834735870361,
-      "learning_rate": 5.11929648241206e-05,
-      "loss": 5.2087,
-      "step": 49081
-    },
-    {
-      "epoch": 25.596870925684485,
-      "grad_norm": 1.488663911819458,
-      "learning_rate": 5.119195979899498e-05,
-      "loss": 5.3457,
-      "step": 49082
-    },
-    {
-      "epoch": 25.597392438070404,
-      "grad_norm": 1.5597506761550903,
-      "learning_rate": 5.119095477386935e-05,
-      "loss": 5.1572,
-      "step": 49083
-    },
-    {
-      "epoch": 25.597913950456324,
-      "grad_norm": 1.6166141033172607,
-      "learning_rate": 5.118994974874373e-05,
-      "loss": 5.11,
-      "step": 49084
-    },
-    {
-      "epoch": 25.598435462842243,
-      "grad_norm": 1.5621421337127686,
-      "learning_rate": 5.118894472361809e-05,
-      "loss": 5.6174,
-      "step": 49085
-    },
-    {
-      "epoch": 25.598956975228162,
-      "grad_norm": 1.5743974447250366,
-      "learning_rate": 5.118793969849246e-05,
-      "loss": 5.1747,
-      "step": 49086
-    },
-    {
-      "epoch": 25.599478487614082,
-      "grad_norm": 1.5802388191223145,
-      "learning_rate": 5.1186934673366835e-05,
-      "loss": 5.5929,
-      "step": 49087
-    },
-    {
-      "epoch": 25.6,
-      "grad_norm": 1.668918251991272,
-      "learning_rate": 5.1185929648241206e-05,
-      "loss": 5.3279,
-      "step": 49088
-    },
-    {
-      "epoch": 25.60052151238592,
-      "grad_norm": 1.5416659116744995,
-      "learning_rate": 5.1184924623115584e-05,
-      "loss": 4.5311,
-      "step": 49089
-    },
-    {
-      "epoch": 25.601043024771837,
-      "grad_norm": 1.5511246919631958,
-      "learning_rate": 5.118391959798995e-05,
-      "loss": 5.4148,
-      "step": 49090
-    },
-    {
-      "epoch": 25.601564537157756,
-      "grad_norm": 1.540688157081604,
-      "learning_rate": 5.1182914572864326e-05,
-      "loss": 5.1984,
-      "step": 49091
-    },
-    {
-      "epoch": 25.602086049543676,
-      "grad_norm": 1.6104117631912231,
-      "learning_rate": 5.118190954773869e-05,
-      "loss": 5.1352,
-      "step": 49092
-    },
-    {
-      "epoch": 25.602607561929595,
-      "grad_norm": 1.432166337966919,
-      "learning_rate": 5.118090452261307e-05,
-      "loss": 5.6438,
-      "step": 49093
-    },
-    {
-      "epoch": 25.603129074315515,
-      "grad_norm": 1.6237026453018188,
-      "learning_rate": 5.117989949748744e-05,
-      "loss": 4.7466,
-      "step": 49094
-    },
-    {
-      "epoch": 25.603650586701434,
-      "grad_norm": 1.6028341054916382,
-      "learning_rate": 5.117889447236182e-05,
-      "loss": 5.0335,
-      "step": 49095
-    },
-    {
-      "epoch": 25.604172099087354,
-      "grad_norm": 1.744069218635559,
-      "learning_rate": 5.117788944723618e-05,
-      "loss": 5.1671,
-      "step": 49096
-    },
-    {
-      "epoch": 25.604693611473273,
-      "grad_norm": 1.6247950792312622,
-      "learning_rate": 5.117688442211056e-05,
-      "loss": 4.9478,
-      "step": 49097
-    },
-    {
-      "epoch": 25.605215123859193,
-      "grad_norm": 1.5563956499099731,
-      "learning_rate": 5.117587939698493e-05,
-      "loss": 5.121,
-      "step": 49098
-    },
-    {
-      "epoch": 25.605736636245112,
-      "grad_norm": 1.4922857284545898,
-      "learning_rate": 5.1174874371859294e-05,
-      "loss": 5.1336,
-      "step": 49099
-    },
-    {
-      "epoch": 25.60625814863103,
-      "grad_norm": 1.568471908569336,
-      "learning_rate": 5.117386934673367e-05,
-      "loss": 5.4451,
-      "step": 49100
-    },
-    {
-      "epoch": 25.60677966101695,
-      "grad_norm": 1.521238923072815,
-      "learning_rate": 5.1172864321608036e-05,
-      "loss": 5.2977,
-      "step": 49101
-    },
-    {
-      "epoch": 25.607301173402867,
-      "grad_norm": 1.558831810951233,
-      "learning_rate": 5.1171859296482414e-05,
-      "loss": 5.2558,
-      "step": 49102
-    },
-    {
-      "epoch": 25.607822685788786,
-      "grad_norm": 1.6216282844543457,
-      "learning_rate": 5.1170854271356785e-05,
-      "loss": 5.3554,
-      "step": 49103
-    },
-    {
-      "epoch": 25.608344198174706,
-      "grad_norm": 1.4886577129364014,
-      "learning_rate": 5.116984924623116e-05,
-      "loss": 5.3471,
-      "step": 49104
-    },
-    {
-      "epoch": 25.608865710560625,
-      "grad_norm": 1.645517110824585,
-      "learning_rate": 5.116884422110553e-05,
-      "loss": 5.3122,
-      "step": 49105
-    },
-    {
-      "epoch": 25.609387222946545,
-      "grad_norm": 1.727455735206604,
-      "learning_rate": 5.1167839195979905e-05,
-      "loss": 5.1367,
-      "step": 49106
-    },
-    {
-      "epoch": 25.609908735332464,
-      "grad_norm": 1.5247913599014282,
-      "learning_rate": 5.116683417085427e-05,
-      "loss": 5.4192,
-      "step": 49107
-    },
-    {
-      "epoch": 25.610430247718384,
-      "grad_norm": 1.5219420194625854,
-      "learning_rate": 5.116582914572865e-05,
-      "loss": 5.2749,
-      "step": 49108
-    },
-    {
-      "epoch": 25.610951760104303,
-      "grad_norm": 1.5105136632919312,
-      "learning_rate": 5.116482412060302e-05,
-      "loss": 5.6621,
-      "step": 49109
-    },
-    {
-      "epoch": 25.611473272490223,
-      "grad_norm": 1.5334784984588623,
-      "learning_rate": 5.1163819095477396e-05,
-      "loss": 5.0583,
-      "step": 49110
-    },
-    {
-      "epoch": 25.611994784876142,
-      "grad_norm": 1.5742101669311523,
-      "learning_rate": 5.116281407035176e-05,
-      "loss": 5.1638,
-      "step": 49111
-    },
-    {
-      "epoch": 25.61251629726206,
-      "grad_norm": 1.5286507606506348,
-      "learning_rate": 5.1161809045226125e-05,
-      "loss": 5.1399,
-      "step": 49112
-    },
-    {
-      "epoch": 25.613037809647977,
-      "grad_norm": 1.89841628074646,
-      "learning_rate": 5.11608040201005e-05,
-      "loss": 4.994,
-      "step": 49113
-    },
-    {
-      "epoch": 25.613559322033897,
-      "grad_norm": 1.4501591920852661,
-      "learning_rate": 5.1159798994974874e-05,
-      "loss": 5.6411,
-      "step": 49114
-    },
-    {
-      "epoch": 25.614080834419816,
-      "grad_norm": 1.6618890762329102,
-      "learning_rate": 5.115879396984925e-05,
-      "loss": 5.0869,
-      "step": 49115
-    },
-    {
-      "epoch": 25.614602346805736,
-      "grad_norm": 1.5400587320327759,
-      "learning_rate": 5.1157788944723616e-05,
-      "loss": 5.2713,
-      "step": 49116
-    },
-    {
-      "epoch": 25.615123859191655,
-      "grad_norm": 1.5222609043121338,
-      "learning_rate": 5.1156783919597994e-05,
-      "loss": 5.3277,
-      "step": 49117
-    },
-    {
-      "epoch": 25.615645371577575,
-      "grad_norm": 1.5732358694076538,
-      "learning_rate": 5.1155778894472365e-05,
-      "loss": 5.3679,
-      "step": 49118
-    },
-    {
-      "epoch": 25.616166883963494,
-      "grad_norm": 1.581425428390503,
-      "learning_rate": 5.115477386934674e-05,
-      "loss": 5.1489,
-      "step": 49119
-    },
-    {
-      "epoch": 25.616688396349414,
-      "grad_norm": 1.5363318920135498,
-      "learning_rate": 5.115376884422111e-05,
-      "loss": 5.1545,
-      "step": 49120
-    },
-    {
-      "epoch": 25.617209908735333,
-      "grad_norm": 1.531980037689209,
-      "learning_rate": 5.1152763819095485e-05,
-      "loss": 4.8717,
-      "step": 49121
-    },
-    {
-      "epoch": 25.617731421121253,
-      "grad_norm": 1.6081993579864502,
-      "learning_rate": 5.115175879396985e-05,
-      "loss": 5.2685,
-      "step": 49122
-    },
-    {
-      "epoch": 25.618252933507172,
-      "grad_norm": 1.6733735799789429,
-      "learning_rate": 5.115075376884423e-05,
-      "loss": 4.4109,
-      "step": 49123
-    },
-    {
-      "epoch": 25.61877444589309,
-      "grad_norm": 1.524556279182434,
-      "learning_rate": 5.11497487437186e-05,
-      "loss": 5.662,
-      "step": 49124
-    },
-    {
-      "epoch": 25.619295958279007,
-      "grad_norm": 1.5400904417037964,
-      "learning_rate": 5.114874371859296e-05,
-      "loss": 4.976,
-      "step": 49125
-    },
-    {
-      "epoch": 25.619817470664927,
-      "grad_norm": 1.5070056915283203,
-      "learning_rate": 5.114773869346734e-05,
-      "loss": 5.5631,
-      "step": 49126
-    },
-    {
-      "epoch": 25.620338983050846,
-      "grad_norm": 1.5775527954101562,
-      "learning_rate": 5.1146733668341704e-05,
-      "loss": 4.8431,
-      "step": 49127
-    },
-    {
-      "epoch": 25.620860495436766,
-      "grad_norm": 1.6206878423690796,
-      "learning_rate": 5.114572864321608e-05,
-      "loss": 4.9052,
-      "step": 49128
-    },
-    {
-      "epoch": 25.621382007822685,
-      "grad_norm": 1.5241345167160034,
-      "learning_rate": 5.114472361809045e-05,
-      "loss": 5.4531,
-      "step": 49129
-    },
-    {
-      "epoch": 25.621903520208605,
-      "grad_norm": 1.5432389974594116,
-      "learning_rate": 5.114371859296483e-05,
-      "loss": 4.9168,
-      "step": 49130
-    },
-    {
-      "epoch": 25.622425032594524,
-      "grad_norm": 1.493800401687622,
-      "learning_rate": 5.1142713567839195e-05,
-      "loss": 5.4643,
-      "step": 49131
-    },
-    {
-      "epoch": 25.622946544980444,
-      "grad_norm": 1.5379667282104492,
-      "learning_rate": 5.114170854271357e-05,
-      "loss": 5.679,
-      "step": 49132
-    },
-    {
-      "epoch": 25.623468057366363,
-      "grad_norm": 1.4993019104003906,
-      "learning_rate": 5.114070351758794e-05,
-      "loss": 5.4243,
-      "step": 49133
-    },
-    {
-      "epoch": 25.623989569752283,
-      "grad_norm": 1.5834985971450806,
-      "learning_rate": 5.1139698492462315e-05,
-      "loss": 4.7397,
-      "step": 49134
-    },
-    {
-      "epoch": 25.624511082138202,
-      "grad_norm": 1.5567588806152344,
-      "learning_rate": 5.1138693467336686e-05,
-      "loss": 5.179,
-      "step": 49135
-    },
-    {
-      "epoch": 25.62503259452412,
-      "grad_norm": 1.523097276687622,
-      "learning_rate": 5.1137688442211064e-05,
-      "loss": 4.7582,
-      "step": 49136
-    },
-    {
-      "epoch": 25.625554106910037,
-      "grad_norm": 1.6018332242965698,
-      "learning_rate": 5.113668341708543e-05,
-      "loss": 5.4773,
-      "step": 49137
-    },
-    {
-      "epoch": 25.626075619295957,
-      "grad_norm": 1.4554353952407837,
-      "learning_rate": 5.11356783919598e-05,
-      "loss": 5.6108,
-      "step": 49138
-    },
-    {
-      "epoch": 25.626597131681876,
-      "grad_norm": 1.4911459684371948,
-      "learning_rate": 5.113467336683418e-05,
-      "loss": 5.3634,
-      "step": 49139
-    },
-    {
-      "epoch": 25.627118644067796,
-      "grad_norm": 1.5685687065124512,
-      "learning_rate": 5.113366834170854e-05,
-      "loss": 5.3543,
-      "step": 49140
-    },
-    {
-      "epoch": 25.627640156453715,
-      "grad_norm": 1.5878912210464478,
-      "learning_rate": 5.113266331658292e-05,
-      "loss": 5.427,
-      "step": 49141
-    },
-    {
-      "epoch": 25.628161668839635,
-      "grad_norm": 1.536974310874939,
-      "learning_rate": 5.1131658291457283e-05,
-      "loss": 5.0973,
-      "step": 49142
-    },
-    {
-      "epoch": 25.628683181225554,
-      "grad_norm": 1.5840245485305786,
-      "learning_rate": 5.113065326633166e-05,
-      "loss": 5.2464,
-      "step": 49143
-    },
-    {
-      "epoch": 25.629204693611474,
-      "grad_norm": 1.6162400245666504,
-      "learning_rate": 5.112964824120603e-05,
-      "loss": 5.2495,
-      "step": 49144
-    },
-    {
-      "epoch": 25.629726205997393,
-      "grad_norm": 1.5294398069381714,
-      "learning_rate": 5.112864321608041e-05,
-      "loss": 5.3281,
-      "step": 49145
-    },
-    {
-      "epoch": 25.630247718383313,
-      "grad_norm": 1.6995820999145508,
-      "learning_rate": 5.1127638190954774e-05,
-      "loss": 5.0571,
-      "step": 49146
-    },
-    {
-      "epoch": 25.630769230769232,
-      "grad_norm": 1.5836718082427979,
-      "learning_rate": 5.112663316582915e-05,
-      "loss": 5.3949,
-      "step": 49147
-    },
-    {
-      "epoch": 25.63129074315515,
-      "grad_norm": 1.5894503593444824,
-      "learning_rate": 5.1125628140703516e-05,
-      "loss": 5.0718,
-      "step": 49148
-    },
-    {
-      "epoch": 25.631812255541067,
-      "grad_norm": 1.5298786163330078,
-      "learning_rate": 5.1124623115577894e-05,
-      "loss": 5.1053,
-      "step": 49149
-    },
-    {
-      "epoch": 25.632333767926987,
-      "grad_norm": 1.5406190156936646,
-      "learning_rate": 5.1123618090452265e-05,
-      "loss": 5.3833,
-      "step": 49150
-    },
-    {
-      "epoch": 25.632855280312906,
-      "grad_norm": 1.5842901468276978,
-      "learning_rate": 5.112261306532664e-05,
-      "loss": 5.3909,
-      "step": 49151
-    },
-    {
-      "epoch": 25.633376792698826,
-      "grad_norm": 1.5715540647506714,
-      "learning_rate": 5.112160804020101e-05,
-      "loss": 5.0765,
-      "step": 49152
-    },
-    {
-      "epoch": 25.633898305084745,
-      "grad_norm": 1.6718087196350098,
-      "learning_rate": 5.112060301507537e-05,
-      "loss": 5.5125,
-      "step": 49153
-    },
-    {
-      "epoch": 25.634419817470665,
-      "grad_norm": 1.6007423400878906,
-      "learning_rate": 5.111959798994975e-05,
-      "loss": 5.24,
-      "step": 49154
-    },
-    {
-      "epoch": 25.634941329856584,
-      "grad_norm": 1.5500690937042236,
-      "learning_rate": 5.111859296482412e-05,
-      "loss": 5.3468,
-      "step": 49155
-    },
-    {
-      "epoch": 25.635462842242504,
-      "grad_norm": 1.5863889455795288,
-      "learning_rate": 5.11175879396985e-05,
-      "loss": 5.0517,
-      "step": 49156
-    },
-    {
-      "epoch": 25.635984354628423,
-      "grad_norm": 1.5894200801849365,
-      "learning_rate": 5.111658291457286e-05,
-      "loss": 5.7581,
-      "step": 49157
-    },
-    {
-      "epoch": 25.636505867014343,
-      "grad_norm": 1.7194513082504272,
-      "learning_rate": 5.111557788944724e-05,
-      "loss": 4.7074,
-      "step": 49158
-    },
-    {
-      "epoch": 25.637027379400262,
-      "grad_norm": 1.5938149690628052,
-      "learning_rate": 5.1114572864321605e-05,
-      "loss": 5.3489,
-      "step": 49159
-    },
-    {
-      "epoch": 25.63754889178618,
-      "grad_norm": 1.579986333847046,
-      "learning_rate": 5.111356783919598e-05,
-      "loss": 4.9928,
-      "step": 49160
-    },
-    {
-      "epoch": 25.638070404172097,
-      "grad_norm": 1.594711422920227,
-      "learning_rate": 5.1112562814070354e-05,
-      "loss": 5.1172,
-      "step": 49161
-    },
-    {
-      "epoch": 25.638591916558017,
-      "grad_norm": 1.5814934968948364,
-      "learning_rate": 5.111155778894473e-05,
-      "loss": 5.2272,
-      "step": 49162
-    },
-    {
-      "epoch": 25.639113428943936,
-      "grad_norm": 1.6510212421417236,
-      "learning_rate": 5.1110552763819096e-05,
-      "loss": 5.0588,
-      "step": 49163
-    },
-    {
-      "epoch": 25.639634941329856,
-      "grad_norm": 1.510820984840393,
-      "learning_rate": 5.1109547738693474e-05,
-      "loss": 4.7249,
-      "step": 49164
-    },
-    {
-      "epoch": 25.640156453715775,
-      "grad_norm": 1.5110193490982056,
-      "learning_rate": 5.1108542713567845e-05,
-      "loss": 4.4376,
-      "step": 49165
-    },
-    {
-      "epoch": 25.640677966101695,
-      "grad_norm": 1.6780357360839844,
-      "learning_rate": 5.110753768844221e-05,
-      "loss": 4.9861,
-      "step": 49166
-    },
-    {
-      "epoch": 25.641199478487614,
-      "grad_norm": 1.5355644226074219,
-      "learning_rate": 5.110653266331659e-05,
-      "loss": 5.076,
-      "step": 49167
-    },
-    {
-      "epoch": 25.641720990873534,
-      "grad_norm": 1.5135924816131592,
-      "learning_rate": 5.110552763819095e-05,
-      "loss": 5.2787,
-      "step": 49168
-    },
-    {
-      "epoch": 25.642242503259453,
-      "grad_norm": 1.5608631372451782,
-      "learning_rate": 5.110452261306533e-05,
-      "loss": 5.1104,
-      "step": 49169
-    },
-    {
-      "epoch": 25.642764015645373,
-      "grad_norm": 1.4812742471694946,
-      "learning_rate": 5.11035175879397e-05,
-      "loss": 5.6825,
-      "step": 49170
-    },
-    {
-      "epoch": 25.643285528031292,
-      "grad_norm": 1.6270685195922852,
-      "learning_rate": 5.110251256281408e-05,
-      "loss": 4.8272,
-      "step": 49171
-    },
-    {
-      "epoch": 25.64380704041721,
-      "grad_norm": 1.569156289100647,
-      "learning_rate": 5.110150753768844e-05,
-      "loss": 5.1153,
-      "step": 49172
-    },
-    {
-      "epoch": 25.644328552803128,
-      "grad_norm": 1.4872324466705322,
-      "learning_rate": 5.110050251256282e-05,
-      "loss": 5.4806,
-      "step": 49173
-    },
-    {
-      "epoch": 25.644850065189047,
-      "grad_norm": 1.523077130317688,
-      "learning_rate": 5.1099497487437184e-05,
-      "loss": 5.1265,
-      "step": 49174
-    },
-    {
-      "epoch": 25.645371577574966,
-      "grad_norm": 1.5932987928390503,
-      "learning_rate": 5.109849246231156e-05,
-      "loss": 4.9285,
-      "step": 49175
-    },
-    {
-      "epoch": 25.645893089960886,
-      "grad_norm": 1.534177303314209,
-      "learning_rate": 5.109748743718593e-05,
-      "loss": 5.0235,
-      "step": 49176
-    },
-    {
-      "epoch": 25.646414602346805,
-      "grad_norm": 1.5331041812896729,
-      "learning_rate": 5.109648241206031e-05,
-      "loss": 5.4574,
-      "step": 49177
-    },
-    {
-      "epoch": 25.646936114732725,
-      "grad_norm": 1.614147663116455,
-      "learning_rate": 5.1095477386934675e-05,
-      "loss": 5.0047,
-      "step": 49178
-    },
-    {
-      "epoch": 25.647457627118644,
-      "grad_norm": 1.5699708461761475,
-      "learning_rate": 5.109447236180904e-05,
-      "loss": 5.3624,
-      "step": 49179
-    },
-    {
-      "epoch": 25.647979139504564,
-      "grad_norm": 1.574995994567871,
-      "learning_rate": 5.109346733668342e-05,
-      "loss": 5.2963,
-      "step": 49180
-    },
-    {
-      "epoch": 25.648500651890483,
-      "grad_norm": 1.6028552055358887,
-      "learning_rate": 5.109246231155779e-05,
-      "loss": 5.0963,
-      "step": 49181
-    },
-    {
-      "epoch": 25.649022164276403,
-      "grad_norm": 1.5327364206314087,
-      "learning_rate": 5.1091457286432166e-05,
-      "loss": 5.4057,
-      "step": 49182
-    },
-    {
-      "epoch": 25.649543676662322,
-      "grad_norm": 1.5549194812774658,
-      "learning_rate": 5.109045226130653e-05,
-      "loss": 5.1443,
-      "step": 49183
-    },
-    {
-      "epoch": 25.65006518904824,
-      "grad_norm": 1.5881567001342773,
-      "learning_rate": 5.108944723618091e-05,
-      "loss": 4.735,
-      "step": 49184
-    },
-    {
-      "epoch": 25.650586701434158,
-      "grad_norm": 1.5192891359329224,
-      "learning_rate": 5.108844221105528e-05,
-      "loss": 5.1489,
-      "step": 49185
-    },
-    {
-      "epoch": 25.651108213820077,
-      "grad_norm": 1.6969565153121948,
-      "learning_rate": 5.108743718592966e-05,
-      "loss": 5.359,
-      "step": 49186
-    },
-    {
-      "epoch": 25.651629726205996,
-      "grad_norm": 1.5812264680862427,
-      "learning_rate": 5.108643216080402e-05,
-      "loss": 5.2244,
-      "step": 49187
-    },
-    {
-      "epoch": 25.652151238591916,
-      "grad_norm": 1.5947718620300293,
-      "learning_rate": 5.10854271356784e-05,
-      "loss": 4.9289,
-      "step": 49188
-    },
-    {
-      "epoch": 25.652672750977835,
-      "grad_norm": 1.5899145603179932,
-      "learning_rate": 5.1084422110552763e-05,
-      "loss": 5.3391,
-      "step": 49189
-    },
-    {
-      "epoch": 25.653194263363755,
-      "grad_norm": 1.5251857042312622,
-      "learning_rate": 5.108341708542714e-05,
-      "loss": 5.6224,
-      "step": 49190
-    },
-    {
-      "epoch": 25.653715775749674,
-      "grad_norm": 1.5458552837371826,
-      "learning_rate": 5.108241206030151e-05,
-      "loss": 5.5962,
-      "step": 49191
-    },
-    {
-      "epoch": 25.654237288135594,
-      "grad_norm": 1.5834414958953857,
-      "learning_rate": 5.108140703517588e-05,
-      "loss": 5.2672,
-      "step": 49192
-    },
-    {
-      "epoch": 25.654758800521513,
-      "grad_norm": 1.6412686109542847,
-      "learning_rate": 5.1080402010050254e-05,
-      "loss": 5.5993,
-      "step": 49193
-    },
-    {
-      "epoch": 25.655280312907433,
-      "grad_norm": 1.5350762605667114,
-      "learning_rate": 5.107939698492462e-05,
-      "loss": 5.3073,
-      "step": 49194
-    },
-    {
-      "epoch": 25.655801825293352,
-      "grad_norm": 1.5967094898223877,
-      "learning_rate": 5.1078391959798997e-05,
-      "loss": 5.0799,
-      "step": 49195
-    },
-    {
-      "epoch": 25.656323337679268,
-      "grad_norm": 1.48600435256958,
-      "learning_rate": 5.107738693467337e-05,
-      "loss": 5.3525,
-      "step": 49196
-    },
-    {
-      "epoch": 25.656844850065188,
-      "grad_norm": 1.5038576126098633,
-      "learning_rate": 5.1076381909547745e-05,
-      "loss": 4.816,
-      "step": 49197
-    },
-    {
-      "epoch": 25.657366362451107,
-      "grad_norm": 1.6242690086364746,
-      "learning_rate": 5.107537688442211e-05,
-      "loss": 5.2313,
-      "step": 49198
-    },
-    {
-      "epoch": 25.657887874837026,
-      "grad_norm": 1.5811368227005005,
-      "learning_rate": 5.107437185929649e-05,
-      "loss": 5.3565,
-      "step": 49199
-    },
-    {
-      "epoch": 25.658409387222946,
-      "grad_norm": 1.5848184823989868,
-      "learning_rate": 5.107336683417085e-05,
-      "loss": 5.3558,
-      "step": 49200
-    },
-    {
-      "epoch": 25.658930899608865,
-      "grad_norm": 1.653534173965454,
-      "learning_rate": 5.107236180904523e-05,
-      "loss": 5.3303,
-      "step": 49201
-    },
-    {
-      "epoch": 25.659452411994785,
-      "grad_norm": 1.584797978401184,
-      "learning_rate": 5.10713567839196e-05,
-      "loss": 5.2902,
-      "step": 49202
-    },
-    {
-      "epoch": 25.659973924380704,
-      "grad_norm": 1.5770349502563477,
-      "learning_rate": 5.107035175879398e-05,
-      "loss": 5.2156,
-      "step": 49203
-    },
-    {
-      "epoch": 25.660495436766624,
-      "grad_norm": 1.6067333221435547,
-      "learning_rate": 5.106934673366834e-05,
-      "loss": 4.8372,
-      "step": 49204
-    },
-    {
-      "epoch": 25.661016949152543,
-      "grad_norm": 1.5497162342071533,
-      "learning_rate": 5.1068341708542714e-05,
-      "loss": 5.4877,
-      "step": 49205
-    },
-    {
-      "epoch": 25.661538461538463,
-      "grad_norm": 1.6151719093322754,
-      "learning_rate": 5.106733668341709e-05,
-      "loss": 5.2687,
-      "step": 49206
-    },
-    {
-      "epoch": 25.662059973924382,
-      "grad_norm": 1.49131178855896,
-      "learning_rate": 5.1066331658291456e-05,
-      "loss": 5.1805,
-      "step": 49207
-    },
-    {
-      "epoch": 25.6625814863103,
-      "grad_norm": 1.5304248332977295,
-      "learning_rate": 5.1065326633165834e-05,
-      "loss": 5.3188,
-      "step": 49208
-    },
-    {
-      "epoch": 25.663102998696218,
-      "grad_norm": 1.476938009262085,
-      "learning_rate": 5.10643216080402e-05,
-      "loss": 5.4825,
-      "step": 49209
-    },
-    {
-      "epoch": 25.663624511082137,
-      "grad_norm": 1.5041245222091675,
-      "learning_rate": 5.1063316582914576e-05,
-      "loss": 5.2277,
-      "step": 49210
-    },
-    {
-      "epoch": 25.664146023468057,
-      "grad_norm": 1.539077877998352,
-      "learning_rate": 5.106231155778895e-05,
-      "loss": 5.6029,
-      "step": 49211
-    },
-    {
-      "epoch": 25.664667535853976,
-      "grad_norm": 1.5351275205612183,
-      "learning_rate": 5.1061306532663325e-05,
-      "loss": 5.2239,
-      "step": 49212
-    },
-    {
-      "epoch": 25.665189048239895,
-      "grad_norm": 1.7225120067596436,
-      "learning_rate": 5.106030150753769e-05,
-      "loss": 4.6062,
-      "step": 49213
-    },
-    {
-      "epoch": 25.665710560625815,
-      "grad_norm": 1.564483880996704,
-      "learning_rate": 5.105929648241207e-05,
-      "loss": 4.9629,
-      "step": 49214
-    },
-    {
-      "epoch": 25.666232073011734,
-      "grad_norm": 1.5136295557022095,
-      "learning_rate": 5.105829145728643e-05,
-      "loss": 5.5324,
-      "step": 49215
-    },
-    {
-      "epoch": 25.666753585397654,
-      "grad_norm": 1.5076289176940918,
-      "learning_rate": 5.105728643216081e-05,
-      "loss": 5.2321,
-      "step": 49216
-    },
-    {
-      "epoch": 25.667275097783573,
-      "grad_norm": 1.5511009693145752,
-      "learning_rate": 5.105628140703518e-05,
-      "loss": 5.0414,
-      "step": 49217
-    },
-    {
-      "epoch": 25.667796610169493,
-      "grad_norm": 1.5190536975860596,
-      "learning_rate": 5.1055276381909544e-05,
-      "loss": 5.32,
-      "step": 49218
-    },
-    {
-      "epoch": 25.668318122555412,
-      "grad_norm": 1.8434364795684814,
-      "learning_rate": 5.105427135678392e-05,
-      "loss": 5.1263,
-      "step": 49219
-    },
-    {
-      "epoch": 25.668839634941328,
-      "grad_norm": 1.657619833946228,
-      "learning_rate": 5.1053266331658286e-05,
-      "loss": 5.1775,
-      "step": 49220
-    },
-    {
-      "epoch": 25.669361147327248,
-      "grad_norm": 1.6549525260925293,
-      "learning_rate": 5.1052261306532664e-05,
-      "loss": 5.4529,
-      "step": 49221
-    },
-    {
-      "epoch": 25.669882659713167,
-      "grad_norm": 1.5634465217590332,
-      "learning_rate": 5.1051256281407035e-05,
-      "loss": 5.4136,
-      "step": 49222
-    },
-    {
-      "epoch": 25.670404172099087,
-      "grad_norm": 1.6824818849563599,
-      "learning_rate": 5.105025125628141e-05,
-      "loss": 5.2232,
-      "step": 49223
-    },
-    {
-      "epoch": 25.670925684485006,
-      "grad_norm": 1.5213370323181152,
-      "learning_rate": 5.104924623115578e-05,
-      "loss": 5.1193,
-      "step": 49224
-    },
-    {
-      "epoch": 25.671447196870925,
-      "grad_norm": 1.5512503385543823,
-      "learning_rate": 5.1048241206030155e-05,
-      "loss": 4.8926,
-      "step": 49225
-    },
-    {
-      "epoch": 25.671968709256845,
-      "grad_norm": 1.5930274724960327,
-      "learning_rate": 5.1047236180904526e-05,
-      "loss": 5.1097,
-      "step": 49226
-    },
-    {
-      "epoch": 25.672490221642764,
-      "grad_norm": 1.6112669706344604,
-      "learning_rate": 5.1046231155778904e-05,
-      "loss": 5.1282,
-      "step": 49227
-    },
-    {
-      "epoch": 25.673011734028684,
-      "grad_norm": 1.6454752683639526,
-      "learning_rate": 5.104522613065327e-05,
-      "loss": 5.0288,
-      "step": 49228
-    },
-    {
-      "epoch": 25.673533246414603,
-      "grad_norm": 1.6691042184829712,
-      "learning_rate": 5.1044221105527646e-05,
-      "loss": 4.5209,
-      "step": 49229
-    },
-    {
-      "epoch": 25.674054758800523,
-      "grad_norm": 1.5245823860168457,
-      "learning_rate": 5.104321608040201e-05,
-      "loss": 5.488,
-      "step": 49230
-    },
-    {
-      "epoch": 25.674576271186442,
-      "grad_norm": 1.5637465715408325,
-      "learning_rate": 5.104221105527638e-05,
-      "loss": 5.309,
-      "step": 49231
-    },
-    {
-      "epoch": 25.675097783572358,
-      "grad_norm": 1.588206171989441,
-      "learning_rate": 5.104120603015076e-05,
-      "loss": 4.6802,
-      "step": 49232
-    },
-    {
-      "epoch": 25.675619295958278,
-      "grad_norm": 1.5431079864501953,
-      "learning_rate": 5.1040201005025124e-05,
-      "loss": 5.2082,
-      "step": 49233
-    },
-    {
-      "epoch": 25.676140808344197,
-      "grad_norm": 1.7240742444992065,
-      "learning_rate": 5.10391959798995e-05,
-      "loss": 4.9943,
-      "step": 49234
-    },
-    {
-      "epoch": 25.676662320730117,
-      "grad_norm": 1.576658844947815,
-      "learning_rate": 5.1038190954773866e-05,
-      "loss": 4.9839,
-      "step": 49235
-    },
-    {
-      "epoch": 25.677183833116036,
-      "grad_norm": 1.6010979413986206,
-      "learning_rate": 5.1037185929648244e-05,
-      "loss": 5.3563,
-      "step": 49236
-    },
-    {
-      "epoch": 25.677705345501955,
-      "grad_norm": 1.7214113473892212,
-      "learning_rate": 5.1036180904522615e-05,
-      "loss": 4.9439,
-      "step": 49237
-    },
-    {
-      "epoch": 25.678226857887875,
-      "grad_norm": 1.6144274473190308,
-      "learning_rate": 5.103517587939699e-05,
-      "loss": 5.1714,
-      "step": 49238
-    },
-    {
-      "epoch": 25.678748370273794,
-      "grad_norm": 1.608327031135559,
-      "learning_rate": 5.103417085427136e-05,
-      "loss": 5.3308,
-      "step": 49239
-    },
-    {
-      "epoch": 25.679269882659714,
-      "grad_norm": 1.7126433849334717,
-      "learning_rate": 5.1033165829145735e-05,
-      "loss": 5.3201,
-      "step": 49240
-    },
-    {
-      "epoch": 25.679791395045633,
-      "grad_norm": 1.636181116104126,
-      "learning_rate": 5.10321608040201e-05,
-      "loss": 5.397,
-      "step": 49241
-    },
-    {
-      "epoch": 25.680312907431553,
-      "grad_norm": 1.6679613590240479,
-      "learning_rate": 5.103115577889448e-05,
-      "loss": 5.5405,
-      "step": 49242
-    },
-    {
-      "epoch": 25.680834419817472,
-      "grad_norm": 1.5550286769866943,
-      "learning_rate": 5.103015075376885e-05,
-      "loss": 5.0277,
-      "step": 49243
-    },
-    {
-      "epoch": 25.681355932203388,
-      "grad_norm": 1.411195993423462,
-      "learning_rate": 5.1029145728643226e-05,
-      "loss": 5.2612,
-      "step": 49244
-    },
-    {
-      "epoch": 25.681877444589308,
-      "grad_norm": 1.5721213817596436,
-      "learning_rate": 5.102814070351759e-05,
-      "loss": 5.5007,
-      "step": 49245
-    },
-    {
-      "epoch": 25.682398956975227,
-      "grad_norm": 1.6363822221755981,
-      "learning_rate": 5.1027135678391954e-05,
-      "loss": 5.1863,
-      "step": 49246
-    },
-    {
-      "epoch": 25.682920469361147,
-      "grad_norm": 1.6044268608093262,
-      "learning_rate": 5.102613065326633e-05,
-      "loss": 5.3254,
-      "step": 49247
-    },
-    {
-      "epoch": 25.683441981747066,
-      "grad_norm": 1.667681097984314,
-      "learning_rate": 5.10251256281407e-05,
-      "loss": 5.1161,
-      "step": 49248
-    },
-    {
-      "epoch": 25.683963494132986,
-      "grad_norm": 1.6071640253067017,
-      "learning_rate": 5.102412060301508e-05,
-      "loss": 5.3012,
-      "step": 49249
-    },
-    {
-      "epoch": 25.684485006518905,
-      "grad_norm": 1.5835165977478027,
-      "learning_rate": 5.1023115577889445e-05,
-      "loss": 5.187,
-      "step": 49250
-    },
-    {
-      "epoch": 25.685006518904824,
-      "grad_norm": 1.6036936044692993,
-      "learning_rate": 5.102211055276382e-05,
-      "loss": 5.2875,
-      "step": 49251
-    },
-    {
-      "epoch": 25.685528031290744,
-      "grad_norm": 1.6675140857696533,
-      "learning_rate": 5.1021105527638194e-05,
-      "loss": 5.0991,
-      "step": 49252
-    },
-    {
-      "epoch": 25.686049543676663,
-      "grad_norm": 1.5969549417495728,
-      "learning_rate": 5.102010050251257e-05,
-      "loss": 5.628,
-      "step": 49253
-    },
-    {
-      "epoch": 25.686571056062583,
-      "grad_norm": 1.5670088529586792,
-      "learning_rate": 5.1019095477386936e-05,
-      "loss": 5.4903,
-      "step": 49254
-    },
-    {
-      "epoch": 25.687092568448502,
-      "grad_norm": 1.6088773012161255,
-      "learning_rate": 5.1018090452261314e-05,
-      "loss": 5.344,
-      "step": 49255
-    },
-    {
-      "epoch": 25.687614080834418,
-      "grad_norm": 1.523258924484253,
-      "learning_rate": 5.101708542713568e-05,
-      "loss": 4.9579,
-      "step": 49256
-    },
-    {
-      "epoch": 25.688135593220338,
-      "grad_norm": 1.677573323249817,
-      "learning_rate": 5.1016080402010056e-05,
-      "loss": 5.1265,
-      "step": 49257
-    },
-    {
-      "epoch": 25.688657105606257,
-      "grad_norm": 1.619982123374939,
-      "learning_rate": 5.101507537688443e-05,
-      "loss": 5.2179,
-      "step": 49258
-    },
-    {
-      "epoch": 25.689178617992177,
-      "grad_norm": 1.6546483039855957,
-      "learning_rate": 5.101407035175879e-05,
-      "loss": 4.5367,
-      "step": 49259
-    },
-    {
-      "epoch": 25.689700130378096,
-      "grad_norm": 1.5715004205703735,
-      "learning_rate": 5.101306532663317e-05,
-      "loss": 5.1798,
-      "step": 49260
-    },
-    {
-      "epoch": 25.690221642764016,
-      "grad_norm": 1.6097177267074585,
-      "learning_rate": 5.1012060301507533e-05,
-      "loss": 5.3305,
-      "step": 49261
-    },
-    {
-      "epoch": 25.690743155149935,
-      "grad_norm": 1.515420913696289,
-      "learning_rate": 5.101105527638191e-05,
-      "loss": 5.3069,
-      "step": 49262
-    },
-    {
-      "epoch": 25.691264667535854,
-      "grad_norm": 1.5719488859176636,
-      "learning_rate": 5.101005025125628e-05,
-      "loss": 5.5418,
-      "step": 49263
-    },
-    {
-      "epoch": 25.691786179921774,
-      "grad_norm": 1.5730029344558716,
-      "learning_rate": 5.100904522613066e-05,
-      "loss": 5.2989,
-      "step": 49264
-    },
-    {
-      "epoch": 25.692307692307693,
-      "grad_norm": 1.6662899255752563,
-      "learning_rate": 5.1008040201005024e-05,
-      "loss": 4.9136,
-      "step": 49265
-    },
-    {
-      "epoch": 25.692829204693613,
-      "grad_norm": 1.5487481355667114,
-      "learning_rate": 5.10070351758794e-05,
-      "loss": 5.2022,
-      "step": 49266
-    },
-    {
-      "epoch": 25.693350717079532,
-      "grad_norm": 1.5404555797576904,
-      "learning_rate": 5.1006030150753767e-05,
-      "loss": 4.4817,
-      "step": 49267
-    },
-    {
-      "epoch": 25.69387222946545,
-      "grad_norm": 1.6130956411361694,
-      "learning_rate": 5.1005025125628144e-05,
-      "loss": 5.3488,
-      "step": 49268
-    },
-    {
-      "epoch": 25.694393741851368,
-      "grad_norm": 1.6026216745376587,
-      "learning_rate": 5.1004020100502515e-05,
-      "loss": 4.6538,
-      "step": 49269
-    },
-    {
-      "epoch": 25.694915254237287,
-      "grad_norm": 1.5152256488800049,
-      "learning_rate": 5.100301507537689e-05,
-      "loss": 5.4396,
-      "step": 49270
-    },
-    {
-      "epoch": 25.695436766623207,
-      "grad_norm": 1.5680166482925415,
-      "learning_rate": 5.100201005025126e-05,
-      "loss": 5.2168,
-      "step": 49271
-    },
-    {
-      "epoch": 25.695958279009126,
-      "grad_norm": 1.552886962890625,
-      "learning_rate": 5.100100502512563e-05,
-      "loss": 5.3511,
-      "step": 49272
-    },
-    {
-      "epoch": 25.696479791395046,
-      "grad_norm": 1.5223517417907715,
-      "learning_rate": 5.1000000000000006e-05,
-      "loss": 5.2843,
-      "step": 49273
-    },
-    {
-      "epoch": 25.697001303780965,
-      "grad_norm": 1.5725103616714478,
-      "learning_rate": 5.099899497487437e-05,
-      "loss": 4.9953,
-      "step": 49274
-    },
-    {
-      "epoch": 25.697522816166884,
-      "grad_norm": 1.5718297958374023,
-      "learning_rate": 5.099798994974875e-05,
-      "loss": 4.644,
-      "step": 49275
-    },
-    {
-      "epoch": 25.698044328552804,
-      "grad_norm": 1.5226248502731323,
-      "learning_rate": 5.099698492462311e-05,
-      "loss": 5.3806,
-      "step": 49276
-    },
-    {
-      "epoch": 25.698565840938723,
-      "grad_norm": 1.4396837949752808,
-      "learning_rate": 5.099597989949749e-05,
-      "loss": 5.3608,
-      "step": 49277
-    },
-    {
-      "epoch": 25.699087353324643,
-      "grad_norm": 1.6610798835754395,
-      "learning_rate": 5.099497487437186e-05,
-      "loss": 4.7569,
-      "step": 49278
-    },
-    {
-      "epoch": 25.69960886571056,
-      "grad_norm": 1.5905038118362427,
-      "learning_rate": 5.099396984924624e-05,
-      "loss": 4.7241,
-      "step": 49279
-    },
-    {
-      "epoch": 25.70013037809648,
-      "grad_norm": 1.6128963232040405,
-      "learning_rate": 5.0992964824120604e-05,
-      "loss": 5.1978,
-      "step": 49280
-    },
-    {
-      "epoch": 25.700651890482398,
-      "grad_norm": 1.58334481716156,
-      "learning_rate": 5.099195979899498e-05,
-      "loss": 4.8412,
-      "step": 49281
-    },
-    {
-      "epoch": 25.701173402868317,
-      "grad_norm": 1.5157861709594727,
-      "learning_rate": 5.0990954773869346e-05,
-      "loss": 5.6653,
-      "step": 49282
-    },
-    {
-      "epoch": 25.701694915254237,
-      "grad_norm": 1.5605593919754028,
-      "learning_rate": 5.0989949748743724e-05,
-      "loss": 4.7357,
-      "step": 49283
-    },
-    {
-      "epoch": 25.702216427640156,
-      "grad_norm": 1.4497307538986206,
-      "learning_rate": 5.0988944723618095e-05,
-      "loss": 5.7749,
-      "step": 49284
-    },
-    {
-      "epoch": 25.702737940026076,
-      "grad_norm": 1.531130313873291,
-      "learning_rate": 5.098793969849246e-05,
-      "loss": 5.1743,
-      "step": 49285
-    },
-    {
-      "epoch": 25.703259452411995,
-      "grad_norm": 1.5210529565811157,
-      "learning_rate": 5.098693467336684e-05,
-      "loss": 4.9848,
-      "step": 49286
-    },
-    {
-      "epoch": 25.703780964797915,
-      "grad_norm": 1.5862457752227783,
-      "learning_rate": 5.09859296482412e-05,
-      "loss": 5.0022,
-      "step": 49287
-    },
-    {
-      "epoch": 25.704302477183834,
-      "grad_norm": 1.5462348461151123,
-      "learning_rate": 5.098492462311558e-05,
-      "loss": 5.583,
-      "step": 49288
-    },
-    {
-      "epoch": 25.704823989569753,
-      "grad_norm": 1.523478627204895,
-      "learning_rate": 5.098391959798995e-05,
-      "loss": 5.0313,
-      "step": 49289
-    },
-    {
-      "epoch": 25.705345501955673,
-      "grad_norm": 1.4469002485275269,
-      "learning_rate": 5.098291457286433e-05,
-      "loss": 5.3035,
-      "step": 49290
-    },
-    {
-      "epoch": 25.705867014341592,
-      "grad_norm": 1.4993956089019775,
-      "learning_rate": 5.098190954773869e-05,
-      "loss": 5.1864,
-      "step": 49291
-    },
-    {
-      "epoch": 25.70638852672751,
-      "grad_norm": 1.6559021472930908,
-      "learning_rate": 5.098090452261307e-05,
-      "loss": 5.181,
-      "step": 49292
-    },
-    {
-      "epoch": 25.706910039113428,
-      "grad_norm": 1.5400997400283813,
-      "learning_rate": 5.097989949748744e-05,
-      "loss": 5.3143,
-      "step": 49293
-    },
-    {
-      "epoch": 25.707431551499347,
-      "grad_norm": 1.551695704460144,
-      "learning_rate": 5.097889447236182e-05,
-      "loss": 5.2553,
-      "step": 49294
-    },
-    {
-      "epoch": 25.707953063885267,
-      "grad_norm": 1.7338887453079224,
-      "learning_rate": 5.097788944723618e-05,
-      "loss": 4.8879,
-      "step": 49295
-    },
-    {
-      "epoch": 25.708474576271186,
-      "grad_norm": 1.631761074066162,
-      "learning_rate": 5.097688442211056e-05,
-      "loss": 5.3042,
-      "step": 49296
-    },
-    {
-      "epoch": 25.708996088657106,
-      "grad_norm": 1.5082178115844727,
-      "learning_rate": 5.0975879396984925e-05,
-      "loss": 5.3269,
-      "step": 49297
-    },
-    {
-      "epoch": 25.709517601043025,
-      "grad_norm": 1.5709047317504883,
-      "learning_rate": 5.0974874371859296e-05,
-      "loss": 5.0901,
-      "step": 49298
-    },
-    {
-      "epoch": 25.710039113428945,
-      "grad_norm": 1.5274094343185425,
-      "learning_rate": 5.0973869346733674e-05,
-      "loss": 5.1424,
-      "step": 49299
-    },
-    {
-      "epoch": 25.710560625814864,
-      "grad_norm": 1.521639108657837,
-      "learning_rate": 5.097286432160804e-05,
-      "loss": 5.1964,
-      "step": 49300
-    },
-    {
-      "epoch": 25.711082138200783,
-      "grad_norm": 1.6231335401535034,
-      "learning_rate": 5.0971859296482416e-05,
-      "loss": 5.1485,
-      "step": 49301
-    },
-    {
-      "epoch": 25.711603650586703,
-      "grad_norm": 1.670088291168213,
-      "learning_rate": 5.097085427135678e-05,
-      "loss": 4.681,
-      "step": 49302
-    },
-    {
-      "epoch": 25.71212516297262,
-      "grad_norm": 1.6490533351898193,
-      "learning_rate": 5.096984924623116e-05,
-      "loss": 5.0684,
-      "step": 49303
-    },
-    {
-      "epoch": 25.71264667535854,
-      "grad_norm": 1.5283091068267822,
-      "learning_rate": 5.096884422110553e-05,
-      "loss": 4.6461,
-      "step": 49304
-    },
-    {
-      "epoch": 25.713168187744458,
-      "grad_norm": 1.5674279928207397,
-      "learning_rate": 5.096783919597991e-05,
-      "loss": 5.1307,
-      "step": 49305
-    },
-    {
-      "epoch": 25.713689700130377,
-      "grad_norm": 1.5542749166488647,
-      "learning_rate": 5.096683417085427e-05,
-      "loss": 5.5638,
-      "step": 49306
-    },
-    {
-      "epoch": 25.714211212516297,
-      "grad_norm": 1.4839057922363281,
-      "learning_rate": 5.096582914572865e-05,
-      "loss": 4.7605,
-      "step": 49307
-    },
-    {
-      "epoch": 25.714732724902216,
-      "grad_norm": 1.5267364978790283,
-      "learning_rate": 5.0964824120603014e-05,
-      "loss": 5.4467,
-      "step": 49308
-    },
-    {
-      "epoch": 25.715254237288136,
-      "grad_norm": 1.527262568473816,
-      "learning_rate": 5.096381909547739e-05,
-      "loss": 5.5758,
-      "step": 49309
-    },
-    {
-      "epoch": 25.715775749674055,
-      "grad_norm": 1.6077589988708496,
-      "learning_rate": 5.096281407035176e-05,
-      "loss": 5.3501,
-      "step": 49310
-    },
-    {
-      "epoch": 25.716297262059975,
-      "grad_norm": 1.5534435510635376,
-      "learning_rate": 5.096180904522613e-05,
-      "loss": 5.0382,
-      "step": 49311
-    },
-    {
-      "epoch": 25.716818774445894,
-      "grad_norm": 1.5131340026855469,
-      "learning_rate": 5.0960804020100504e-05,
-      "loss": 5.4865,
-      "step": 49312
-    },
-    {
-      "epoch": 25.717340286831814,
-      "grad_norm": 1.4838604927062988,
-      "learning_rate": 5.0959798994974876e-05,
-      "loss": 5.6405,
-      "step": 49313
-    },
-    {
-      "epoch": 25.717861799217733,
-      "grad_norm": 1.5507867336273193,
-      "learning_rate": 5.095879396984925e-05,
-      "loss": 5.3115,
-      "step": 49314
-    },
-    {
-      "epoch": 25.71838331160365,
-      "grad_norm": 1.541771411895752,
-      "learning_rate": 5.095778894472362e-05,
-      "loss": 5.7158,
-      "step": 49315
-    },
-    {
-      "epoch": 25.71890482398957,
-      "grad_norm": 1.6661646366119385,
-      "learning_rate": 5.0956783919597995e-05,
-      "loss": 4.4535,
-      "step": 49316
-    },
-    {
-      "epoch": 25.719426336375488,
-      "grad_norm": 1.5948138236999512,
-      "learning_rate": 5.095577889447236e-05,
-      "loss": 5.1337,
-      "step": 49317
-    },
-    {
-      "epoch": 25.719947848761407,
-      "grad_norm": 1.5485633611679077,
-      "learning_rate": 5.095477386934674e-05,
-      "loss": 4.8621,
-      "step": 49318
-    },
-    {
-      "epoch": 25.720469361147327,
-      "grad_norm": 1.4856048822402954,
-      "learning_rate": 5.095376884422111e-05,
-      "loss": 5.4332,
-      "step": 49319
-    },
-    {
-      "epoch": 25.720990873533246,
-      "grad_norm": 1.6084976196289062,
-      "learning_rate": 5.0952763819095486e-05,
-      "loss": 5.257,
-      "step": 49320
-    },
-    {
-      "epoch": 25.721512385919166,
-      "grad_norm": 1.6129834651947021,
-      "learning_rate": 5.095175879396985e-05,
-      "loss": 5.2827,
-      "step": 49321
-    },
-    {
-      "epoch": 25.722033898305085,
-      "grad_norm": 1.510401725769043,
-      "learning_rate": 5.095075376884423e-05,
-      "loss": 5.5475,
-      "step": 49322
-    },
-    {
-      "epoch": 25.722555410691005,
-      "grad_norm": 1.5679363012313843,
-      "learning_rate": 5.094974874371859e-05,
-      "loss": 5.511,
-      "step": 49323
-    },
-    {
-      "epoch": 25.723076923076924,
-      "grad_norm": 1.4712754487991333,
-      "learning_rate": 5.094874371859297e-05,
-      "loss": 5.1872,
-      "step": 49324
-    },
-    {
-      "epoch": 25.723598435462844,
-      "grad_norm": 1.581040859222412,
-      "learning_rate": 5.094773869346734e-05,
-      "loss": 5.4413,
-      "step": 49325
-    },
-    {
-      "epoch": 25.724119947848763,
-      "grad_norm": 1.5893595218658447,
-      "learning_rate": 5.0946733668341706e-05,
-      "loss": 5.3012,
-      "step": 49326
-    },
-    {
-      "epoch": 25.72464146023468,
-      "grad_norm": 1.5263031721115112,
-      "learning_rate": 5.0945728643216084e-05,
-      "loss": 5.4571,
-      "step": 49327
-    },
-    {
-      "epoch": 25.7251629726206,
-      "grad_norm": 1.4815030097961426,
-      "learning_rate": 5.094472361809045e-05,
-      "loss": 5.1848,
-      "step": 49328
-    },
-    {
-      "epoch": 25.725684485006518,
-      "grad_norm": 1.5153950452804565,
-      "learning_rate": 5.0943718592964826e-05,
-      "loss": 5.6504,
-      "step": 49329
-    },
-    {
-      "epoch": 25.726205997392437,
-      "grad_norm": 1.6310980319976807,
-      "learning_rate": 5.09427135678392e-05,
-      "loss": 5.3064,
-      "step": 49330
-    },
-    {
-      "epoch": 25.726727509778357,
-      "grad_norm": 1.5234180688858032,
-      "learning_rate": 5.0941708542713575e-05,
-      "loss": 5.4569,
-      "step": 49331
-    },
-    {
-      "epoch": 25.727249022164276,
-      "grad_norm": 1.5557876825332642,
-      "learning_rate": 5.094070351758794e-05,
-      "loss": 5.5143,
-      "step": 49332
-    },
-    {
-      "epoch": 25.727770534550196,
-      "grad_norm": 1.514122486114502,
-      "learning_rate": 5.093969849246232e-05,
-      "loss": 4.9582,
-      "step": 49333
-    },
-    {
-      "epoch": 25.728292046936115,
-      "grad_norm": 1.5339465141296387,
-      "learning_rate": 5.093869346733668e-05,
-      "loss": 5.192,
-      "step": 49334
-    },
-    {
-      "epoch": 25.728813559322035,
-      "grad_norm": 1.6598092317581177,
-      "learning_rate": 5.093768844221106e-05,
-      "loss": 4.3742,
-      "step": 49335
-    },
-    {
-      "epoch": 25.729335071707954,
-      "grad_norm": 1.528059959411621,
-      "learning_rate": 5.093668341708543e-05,
-      "loss": 5.2403,
-      "step": 49336
-    },
-    {
-      "epoch": 25.729856584093874,
-      "grad_norm": 1.552580714225769,
-      "learning_rate": 5.093567839195981e-05,
-      "loss": 5.5117,
-      "step": 49337
-    },
-    {
-      "epoch": 25.730378096479793,
-      "grad_norm": 1.5830333232879639,
-      "learning_rate": 5.093467336683417e-05,
-      "loss": 5.3754,
-      "step": 49338
-    },
-    {
-      "epoch": 25.73089960886571,
-      "grad_norm": 1.5579842329025269,
-      "learning_rate": 5.093366834170854e-05,
-      "loss": 5.4018,
-      "step": 49339
-    },
-    {
-      "epoch": 25.73142112125163,
-      "grad_norm": 1.5273287296295166,
-      "learning_rate": 5.093266331658292e-05,
-      "loss": 5.4341,
-      "step": 49340
-    },
-    {
-      "epoch": 25.731942633637548,
-      "grad_norm": 1.565125584602356,
-      "learning_rate": 5.0931658291457285e-05,
-      "loss": 4.9918,
-      "step": 49341
-    },
-    {
-      "epoch": 25.732464146023467,
-      "grad_norm": 1.5764261484146118,
-      "learning_rate": 5.093065326633166e-05,
-      "loss": 5.2748,
-      "step": 49342
-    },
-    {
-      "epoch": 25.732985658409387,
-      "grad_norm": 1.723837971687317,
-      "learning_rate": 5.092964824120603e-05,
-      "loss": 5.5267,
-      "step": 49343
-    },
-    {
-      "epoch": 25.733507170795306,
-      "grad_norm": 1.5930277109146118,
-      "learning_rate": 5.0928643216080405e-05,
-      "loss": 5.1642,
-      "step": 49344
-    },
-    {
-      "epoch": 25.734028683181226,
-      "grad_norm": 1.5656791925430298,
-      "learning_rate": 5.0927638190954776e-05,
-      "loss": 5.4041,
-      "step": 49345
-    },
-    {
-      "epoch": 25.734550195567145,
-      "grad_norm": 1.5205588340759277,
-      "learning_rate": 5.0926633165829154e-05,
-      "loss": 5.2768,
-      "step": 49346
-    },
-    {
-      "epoch": 25.735071707953065,
-      "grad_norm": 1.5250924825668335,
-      "learning_rate": 5.092562814070352e-05,
-      "loss": 5.5712,
-      "step": 49347
-    },
-    {
-      "epoch": 25.735593220338984,
-      "grad_norm": 1.518324613571167,
-      "learning_rate": 5.0924623115577896e-05,
-      "loss": 4.9638,
-      "step": 49348
-    },
-    {
-      "epoch": 25.736114732724904,
-      "grad_norm": 1.685738444328308,
-      "learning_rate": 5.092361809045226e-05,
-      "loss": 5.2608,
-      "step": 49349
-    },
-    {
-      "epoch": 25.736636245110823,
-      "grad_norm": 1.5388789176940918,
-      "learning_rate": 5.092261306532664e-05,
-      "loss": 5.3668,
-      "step": 49350
-    },
-    {
-      "epoch": 25.73715775749674,
-      "grad_norm": 1.4986523389816284,
-      "learning_rate": 5.092160804020101e-05,
-      "loss": 5.5643,
-      "step": 49351
-    },
-    {
-      "epoch": 25.73767926988266,
-      "grad_norm": 1.687217116355896,
-      "learning_rate": 5.0920603015075374e-05,
-      "loss": 5.1522,
-      "step": 49352
-    },
-    {
-      "epoch": 25.738200782268578,
-      "grad_norm": 1.6430494785308838,
-      "learning_rate": 5.091959798994975e-05,
-      "loss": 5.1228,
-      "step": 49353
-    },
-    {
-      "epoch": 25.738722294654497,
-      "grad_norm": 1.6552033424377441,
-      "learning_rate": 5.0918592964824116e-05,
-      "loss": 5.2693,
-      "step": 49354
-    },
-    {
-      "epoch": 25.739243807040417,
-      "grad_norm": 1.5962262153625488,
-      "learning_rate": 5.0917587939698494e-05,
-      "loss": 5.4971,
-      "step": 49355
-    },
-    {
-      "epoch": 25.739765319426336,
-      "grad_norm": 1.607325792312622,
-      "learning_rate": 5.0916582914572865e-05,
-      "loss": 5.137,
-      "step": 49356
-    },
-    {
-      "epoch": 25.740286831812256,
-      "grad_norm": 1.5405610799789429,
-      "learning_rate": 5.091557788944724e-05,
-      "loss": 5.4557,
-      "step": 49357
-    },
-    {
-      "epoch": 25.740808344198175,
-      "grad_norm": 1.6379045248031616,
-      "learning_rate": 5.091457286432161e-05,
-      "loss": 3.7919,
-      "step": 49358
-    },
-    {
-      "epoch": 25.741329856584095,
-      "grad_norm": 1.7428029775619507,
-      "learning_rate": 5.0913567839195985e-05,
-      "loss": 4.7714,
-      "step": 49359
-    },
-    {
-      "epoch": 25.741851368970014,
-      "grad_norm": 1.6489003896713257,
-      "learning_rate": 5.0912562814070356e-05,
-      "loss": 5.1461,
-      "step": 49360
-    },
-    {
-      "epoch": 25.742372881355934,
-      "grad_norm": 1.676252841949463,
-      "learning_rate": 5.0911557788944733e-05,
-      "loss": 4.8913,
-      "step": 49361
-    },
-    {
-      "epoch": 25.742894393741853,
-      "grad_norm": 1.6986085176467896,
-      "learning_rate": 5.09105527638191e-05,
-      "loss": 4.8659,
-      "step": 49362
-    },
-    {
-      "epoch": 25.74341590612777,
-      "grad_norm": 1.727142333984375,
-      "learning_rate": 5.0909547738693476e-05,
-      "loss": 5.1141,
-      "step": 49363
-    },
-    {
-      "epoch": 25.74393741851369,
-      "grad_norm": 1.5111390352249146,
-      "learning_rate": 5.090854271356784e-05,
-      "loss": 5.5886,
-      "step": 49364
-    },
-    {
-      "epoch": 25.744458930899608,
-      "grad_norm": 1.5994421243667603,
-      "learning_rate": 5.090753768844221e-05,
-      "loss": 5.2149,
-      "step": 49365
-    },
-    {
-      "epoch": 25.744980443285527,
-      "grad_norm": 1.60080885887146,
-      "learning_rate": 5.090653266331659e-05,
-      "loss": 4.8966,
-      "step": 49366
-    },
-    {
-      "epoch": 25.745501955671447,
-      "grad_norm": 1.5894638299942017,
-      "learning_rate": 5.090552763819095e-05,
-      "loss": 5.0134,
-      "step": 49367
-    },
-    {
-      "epoch": 25.746023468057366,
-      "grad_norm": 1.634663701057434,
-      "learning_rate": 5.090452261306533e-05,
-      "loss": 5.0327,
-      "step": 49368
-    },
-    {
-      "epoch": 25.746544980443286,
-      "grad_norm": 1.5257906913757324,
-      "learning_rate": 5.0903517587939695e-05,
-      "loss": 5.4684,
-      "step": 49369
-    },
-    {
-      "epoch": 25.747066492829205,
-      "grad_norm": 1.4024096727371216,
-      "learning_rate": 5.090251256281407e-05,
-      "loss": 5.702,
-      "step": 49370
-    },
-    {
-      "epoch": 25.747588005215125,
-      "grad_norm": 1.5542651414871216,
-      "learning_rate": 5.0901507537688444e-05,
-      "loss": 5.4471,
-      "step": 49371
-    },
-    {
-      "epoch": 25.748109517601044,
-      "grad_norm": 1.5993911027908325,
-      "learning_rate": 5.090050251256282e-05,
-      "loss": 5.2184,
-      "step": 49372
-    },
-    {
-      "epoch": 25.748631029986964,
-      "grad_norm": 1.7226794958114624,
-      "learning_rate": 5.0899497487437186e-05,
-      "loss": 4.8263,
-      "step": 49373
-    },
-    {
-      "epoch": 25.749152542372883,
-      "grad_norm": 1.4643621444702148,
-      "learning_rate": 5.0898492462311564e-05,
-      "loss": 4.6592,
-      "step": 49374
-    },
-    {
-      "epoch": 25.7496740547588,
-      "grad_norm": 1.573024868965149,
-      "learning_rate": 5.089748743718593e-05,
-      "loss": 5.2439,
-      "step": 49375
-    },
-    {
-      "epoch": 25.75019556714472,
-      "grad_norm": 1.5033069849014282,
-      "learning_rate": 5.0896482412060306e-05,
-      "loss": 5.2823,
-      "step": 49376
-    },
-    {
-      "epoch": 25.750717079530638,
-      "grad_norm": 1.782434105873108,
-      "learning_rate": 5.089547738693468e-05,
-      "loss": 5.0652,
-      "step": 49377
-    },
-    {
-      "epoch": 25.751238591916557,
-      "grad_norm": 1.5558956861495972,
-      "learning_rate": 5.089447236180904e-05,
-      "loss": 5.077,
-      "step": 49378
-    },
-    {
-      "epoch": 25.751760104302477,
-      "grad_norm": 1.5077533721923828,
-      "learning_rate": 5.089346733668342e-05,
-      "loss": 5.8387,
-      "step": 49379
-    },
-    {
-      "epoch": 25.752281616688396,
-      "grad_norm": 1.5379624366760254,
-      "learning_rate": 5.089246231155779e-05,
-      "loss": 5.4914,
-      "step": 49380
-    },
-    {
-      "epoch": 25.752803129074316,
-      "grad_norm": 1.5976415872573853,
-      "learning_rate": 5.089145728643217e-05,
-      "loss": 5.0191,
-      "step": 49381
-    },
-    {
-      "epoch": 25.753324641460235,
-      "grad_norm": 1.7827743291854858,
-      "learning_rate": 5.089045226130653e-05,
-      "loss": 4.6834,
-      "step": 49382
-    },
-    {
-      "epoch": 25.753846153846155,
-      "grad_norm": 1.5357930660247803,
-      "learning_rate": 5.088944723618091e-05,
-      "loss": 5.3996,
-      "step": 49383
-    },
-    {
-      "epoch": 25.754367666232074,
-      "grad_norm": 1.4840511083602905,
-      "learning_rate": 5.0888442211055274e-05,
-      "loss": 5.1559,
-      "step": 49384
-    },
-    {
-      "epoch": 25.754889178617994,
-      "grad_norm": 1.6122114658355713,
-      "learning_rate": 5.088743718592965e-05,
-      "loss": 5.2721,
-      "step": 49385
-    },
-    {
-      "epoch": 25.75541069100391,
-      "grad_norm": 1.7136075496673584,
-      "learning_rate": 5.088643216080402e-05,
-      "loss": 5.0306,
-      "step": 49386
-    },
-    {
-      "epoch": 25.75593220338983,
-      "grad_norm": 1.6198806762695312,
-      "learning_rate": 5.08854271356784e-05,
-      "loss": 4.4853,
-      "step": 49387
-    },
-    {
-      "epoch": 25.75645371577575,
-      "grad_norm": 1.5914064645767212,
-      "learning_rate": 5.0884422110552765e-05,
-      "loss": 5.349,
-      "step": 49388
-    },
-    {
-      "epoch": 25.756975228161668,
-      "grad_norm": 1.6038780212402344,
-      "learning_rate": 5.088341708542714e-05,
-      "loss": 5.6565,
-      "step": 49389
-    },
-    {
-      "epoch": 25.757496740547587,
-      "grad_norm": 1.5908387899398804,
-      "learning_rate": 5.088241206030151e-05,
-      "loss": 5.5372,
-      "step": 49390
-    },
-    {
-      "epoch": 25.758018252933507,
-      "grad_norm": 1.495862364768982,
-      "learning_rate": 5.088140703517588e-05,
-      "loss": 5.2625,
-      "step": 49391
-    },
-    {
-      "epoch": 25.758539765319426,
-      "grad_norm": 1.5612355470657349,
-      "learning_rate": 5.0880402010050256e-05,
-      "loss": 5.1864,
-      "step": 49392
-    },
-    {
-      "epoch": 25.759061277705346,
-      "grad_norm": 1.7252806425094604,
-      "learning_rate": 5.087939698492462e-05,
-      "loss": 5.2858,
-      "step": 49393
-    },
-    {
-      "epoch": 25.759582790091265,
-      "grad_norm": 1.6472508907318115,
-      "learning_rate": 5.0878391959799e-05,
-      "loss": 5.2819,
-      "step": 49394
-    },
-    {
-      "epoch": 25.760104302477185,
-      "grad_norm": 1.55177640914917,
-      "learning_rate": 5.087738693467336e-05,
-      "loss": 5.2607,
-      "step": 49395
-    },
-    {
-      "epoch": 25.760625814863104,
-      "grad_norm": 1.6208715438842773,
-      "learning_rate": 5.087638190954774e-05,
-      "loss": 5.5314,
-      "step": 49396
-    },
-    {
-      "epoch": 25.761147327249024,
-      "grad_norm": 1.532757043838501,
-      "learning_rate": 5.087537688442211e-05,
-      "loss": 5.5742,
-      "step": 49397
-    },
-    {
-      "epoch": 25.761668839634943,
-      "grad_norm": 1.6950056552886963,
-      "learning_rate": 5.087437185929649e-05,
-      "loss": 4.7683,
-      "step": 49398
-    },
-    {
-      "epoch": 25.76219035202086,
-      "grad_norm": 1.5801416635513306,
-      "learning_rate": 5.0873366834170854e-05,
-      "loss": 5.1743,
-      "step": 49399
-    },
-    {
-      "epoch": 25.76271186440678,
-      "grad_norm": 1.553388237953186,
-      "learning_rate": 5.087236180904523e-05,
-      "loss": 5.2229,
-      "step": 49400
-    },
-    {
-      "epoch": 25.763233376792698,
-      "grad_norm": 1.6794772148132324,
-      "learning_rate": 5.08713567839196e-05,
-      "loss": 4.9229,
-      "step": 49401
-    },
-    {
-      "epoch": 25.763754889178617,
-      "grad_norm": 1.532986044883728,
-      "learning_rate": 5.087035175879398e-05,
-      "loss": 5.2419,
-      "step": 49402
-    },
-    {
-      "epoch": 25.764276401564537,
-      "grad_norm": 1.5179522037506104,
-      "learning_rate": 5.0869346733668345e-05,
-      "loss": 5.4521,
-      "step": 49403
-    },
-    {
-      "epoch": 25.764797913950456,
-      "grad_norm": 1.5609605312347412,
-      "learning_rate": 5.086834170854271e-05,
-      "loss": 5.3162,
-      "step": 49404
-    },
-    {
-      "epoch": 25.765319426336376,
-      "grad_norm": 1.583979606628418,
-      "learning_rate": 5.086733668341709e-05,
-      "loss": 5.3165,
-      "step": 49405
-    },
-    {
-      "epoch": 25.765840938722295,
-      "grad_norm": 1.5300724506378174,
-      "learning_rate": 5.086633165829146e-05,
-      "loss": 5.0535,
-      "step": 49406
-    },
-    {
-      "epoch": 25.766362451108215,
-      "grad_norm": 1.5345032215118408,
-      "learning_rate": 5.0865326633165836e-05,
-      "loss": 5.3729,
-      "step": 49407
-    },
-    {
-      "epoch": 25.766883963494134,
-      "grad_norm": 1.46602201461792,
-      "learning_rate": 5.08643216080402e-05,
-      "loss": 5.4618,
-      "step": 49408
-    },
-    {
-      "epoch": 25.767405475880054,
-      "grad_norm": 1.5750476121902466,
-      "learning_rate": 5.086331658291458e-05,
-      "loss": 4.7539,
-      "step": 49409
-    },
-    {
-      "epoch": 25.76792698826597,
-      "grad_norm": 1.50304114818573,
-      "learning_rate": 5.086231155778894e-05,
-      "loss": 5.1944,
-      "step": 49410
-    },
-    {
-      "epoch": 25.76844850065189,
-      "grad_norm": 1.5158883333206177,
-      "learning_rate": 5.086130653266332e-05,
-      "loss": 5.4726,
-      "step": 49411
-    },
-    {
-      "epoch": 25.76897001303781,
-      "grad_norm": 1.5553834438323975,
-      "learning_rate": 5.086030150753769e-05,
-      "loss": 5.383,
-      "step": 49412
-    },
-    {
-      "epoch": 25.769491525423728,
-      "grad_norm": 1.523966908454895,
-      "learning_rate": 5.085929648241207e-05,
-      "loss": 5.3523,
-      "step": 49413
-    },
-    {
-      "epoch": 25.770013037809647,
-      "grad_norm": 1.6457067728042603,
-      "learning_rate": 5.085829145728643e-05,
-      "loss": 5.3854,
-      "step": 49414
-    },
-    {
-      "epoch": 25.770534550195567,
-      "grad_norm": 1.6466647386550903,
-      "learning_rate": 5.085728643216081e-05,
-      "loss": 5.3926,
-      "step": 49415
-    },
-    {
-      "epoch": 25.771056062581486,
-      "grad_norm": 1.518704891204834,
-      "learning_rate": 5.0856281407035175e-05,
-      "loss": 5.1084,
-      "step": 49416
-    },
-    {
-      "epoch": 25.771577574967406,
-      "grad_norm": 1.682518482208252,
-      "learning_rate": 5.085527638190955e-05,
-      "loss": 4.8017,
-      "step": 49417
-    },
-    {
-      "epoch": 25.772099087353325,
-      "grad_norm": 1.5554362535476685,
-      "learning_rate": 5.0854271356783924e-05,
-      "loss": 5.4385,
-      "step": 49418
-    },
-    {
-      "epoch": 25.772620599739245,
-      "grad_norm": 1.6435552835464478,
-      "learning_rate": 5.085326633165829e-05,
-      "loss": 4.4241,
-      "step": 49419
-    },
-    {
-      "epoch": 25.773142112125164,
-      "grad_norm": 1.587552547454834,
-      "learning_rate": 5.0852261306532666e-05,
-      "loss": 5.6095,
-      "step": 49420
-    },
-    {
-      "epoch": 25.773663624511084,
-      "grad_norm": 1.6835442781448364,
-      "learning_rate": 5.085125628140703e-05,
-      "loss": 5.2674,
-      "step": 49421
-    },
-    {
-      "epoch": 25.774185136897,
-      "grad_norm": 1.6397761106491089,
-      "learning_rate": 5.085025125628141e-05,
-      "loss": 5.3686,
-      "step": 49422
-    },
-    {
-      "epoch": 25.77470664928292,
-      "grad_norm": 1.5409375429153442,
-      "learning_rate": 5.084924623115578e-05,
-      "loss": 5.4077,
-      "step": 49423
-    },
-    {
-      "epoch": 25.77522816166884,
-      "grad_norm": 1.6750065088272095,
-      "learning_rate": 5.084824120603016e-05,
-      "loss": 5.0674,
-      "step": 49424
-    },
-    {
-      "epoch": 25.775749674054758,
-      "grad_norm": 1.5270531177520752,
-      "learning_rate": 5.084723618090452e-05,
-      "loss": 5.1853,
-      "step": 49425
-    },
-    {
-      "epoch": 25.776271186440677,
-      "grad_norm": 1.5520479679107666,
-      "learning_rate": 5.08462311557789e-05,
-      "loss": 5.2444,
-      "step": 49426
-    },
-    {
-      "epoch": 25.776792698826597,
-      "grad_norm": 1.4951086044311523,
-      "learning_rate": 5.084522613065327e-05,
-      "loss": 5.3754,
-      "step": 49427
-    },
-    {
-      "epoch": 25.777314211212516,
-      "grad_norm": 1.6303112506866455,
-      "learning_rate": 5.084422110552765e-05,
-      "loss": 5.1761,
-      "step": 49428
-    },
-    {
-      "epoch": 25.777835723598436,
-      "grad_norm": 1.4683022499084473,
-      "learning_rate": 5.084321608040201e-05,
-      "loss": 5.3622,
-      "step": 49429
-    },
-    {
-      "epoch": 25.778357235984355,
-      "grad_norm": 1.515688419342041,
-      "learning_rate": 5.084221105527639e-05,
-      "loss": 5.0638,
-      "step": 49430
-    },
-    {
-      "epoch": 25.778878748370275,
-      "grad_norm": 1.6373214721679688,
-      "learning_rate": 5.0841206030150755e-05,
-      "loss": 4.8379,
-      "step": 49431
-    },
-    {
-      "epoch": 25.779400260756194,
-      "grad_norm": 1.5292974710464478,
-      "learning_rate": 5.0840201005025126e-05,
-      "loss": 5.3852,
-      "step": 49432
-    },
-    {
-      "epoch": 25.779921773142114,
-      "grad_norm": 1.5218408107757568,
-      "learning_rate": 5.0839195979899503e-05,
-      "loss": 5.6581,
-      "step": 49433
-    },
-    {
-      "epoch": 25.78044328552803,
-      "grad_norm": 1.6062564849853516,
-      "learning_rate": 5.083819095477387e-05,
-      "loss": 5.5268,
-      "step": 49434
-    },
-    {
-      "epoch": 25.78096479791395,
-      "grad_norm": 1.570929765701294,
-      "learning_rate": 5.0837185929648245e-05,
-      "loss": 4.6978,
-      "step": 49435
-    },
-    {
-      "epoch": 25.78148631029987,
-      "grad_norm": 1.5691919326782227,
-      "learning_rate": 5.083618090452261e-05,
-      "loss": 4.9259,
-      "step": 49436
-    },
-    {
-      "epoch": 25.782007822685788,
-      "grad_norm": 1.4936182498931885,
-      "learning_rate": 5.083517587939699e-05,
-      "loss": 4.8499,
-      "step": 49437
-    },
-    {
-      "epoch": 25.782529335071708,
-      "grad_norm": 1.7058517932891846,
-      "learning_rate": 5.083417085427136e-05,
-      "loss": 4.7133,
-      "step": 49438
-    },
-    {
-      "epoch": 25.783050847457627,
-      "grad_norm": 1.6431388854980469,
-      "learning_rate": 5.0833165829145736e-05,
-      "loss": 4.9613,
-      "step": 49439
-    },
-    {
-      "epoch": 25.783572359843546,
-      "grad_norm": 1.534572720527649,
-      "learning_rate": 5.08321608040201e-05,
-      "loss": 5.5145,
-      "step": 49440
-    },
-    {
-      "epoch": 25.784093872229466,
-      "grad_norm": 1.6392868757247925,
-      "learning_rate": 5.083115577889448e-05,
-      "loss": 5.3342,
-      "step": 49441
-    },
-    {
-      "epoch": 25.784615384615385,
-      "grad_norm": 1.4516806602478027,
-      "learning_rate": 5.083015075376884e-05,
-      "loss": 5.1834,
-      "step": 49442
-    },
-    {
-      "epoch": 25.785136897001305,
-      "grad_norm": 1.6050617694854736,
-      "learning_rate": 5.082914572864322e-05,
-      "loss": 5.2445,
-      "step": 49443
-    },
-    {
-      "epoch": 25.785658409387224,
-      "grad_norm": 1.630660057067871,
-      "learning_rate": 5.082814070351759e-05,
-      "loss": 5.2186,
-      "step": 49444
-    },
-    {
-      "epoch": 25.786179921773144,
-      "grad_norm": 1.5883435010910034,
-      "learning_rate": 5.0827135678391956e-05,
-      "loss": 5.4551,
-      "step": 49445
-    },
-    {
-      "epoch": 25.78670143415906,
-      "grad_norm": 1.6089329719543457,
-      "learning_rate": 5.0826130653266334e-05,
-      "loss": 5.1317,
-      "step": 49446
-    },
-    {
-      "epoch": 25.78722294654498,
-      "grad_norm": 1.5228828191757202,
-      "learning_rate": 5.0825125628140705e-05,
-      "loss": 5.1683,
-      "step": 49447
-    },
-    {
-      "epoch": 25.7877444589309,
-      "grad_norm": 1.561089038848877,
-      "learning_rate": 5.082412060301508e-05,
-      "loss": 4.3452,
-      "step": 49448
-    },
-    {
-      "epoch": 25.788265971316818,
-      "grad_norm": 1.614431619644165,
-      "learning_rate": 5.082311557788945e-05,
-      "loss": 4.9129,
-      "step": 49449
-    },
-    {
-      "epoch": 25.788787483702738,
-      "grad_norm": 1.6357392072677612,
-      "learning_rate": 5.0822110552763825e-05,
-      "loss": 4.9095,
-      "step": 49450
-    },
-    {
-      "epoch": 25.789308996088657,
-      "grad_norm": 1.6077940464019775,
-      "learning_rate": 5.082110552763819e-05,
-      "loss": 4.8337,
-      "step": 49451
-    },
-    {
-      "epoch": 25.789830508474576,
-      "grad_norm": 1.560876727104187,
-      "learning_rate": 5.082010050251257e-05,
-      "loss": 5.3804,
-      "step": 49452
-    },
-    {
-      "epoch": 25.790352020860496,
-      "grad_norm": 1.5736517906188965,
-      "learning_rate": 5.081909547738694e-05,
-      "loss": 5.3139,
-      "step": 49453
-    },
-    {
-      "epoch": 25.790873533246415,
-      "grad_norm": 1.4973363876342773,
-      "learning_rate": 5.0818090452261316e-05,
-      "loss": 5.2046,
-      "step": 49454
-    },
-    {
-      "epoch": 25.791395045632335,
-      "grad_norm": 1.6135526895523071,
-      "learning_rate": 5.081708542713568e-05,
-      "loss": 5.061,
-      "step": 49455
-    },
-    {
-      "epoch": 25.791916558018254,
-      "grad_norm": 1.437915325164795,
-      "learning_rate": 5.081608040201006e-05,
-      "loss": 5.4908,
-      "step": 49456
-    },
-    {
-      "epoch": 25.792438070404174,
-      "grad_norm": 1.6477243900299072,
-      "learning_rate": 5.081507537688442e-05,
-      "loss": 5.3331,
-      "step": 49457
-    },
-    {
-      "epoch": 25.79295958279009,
-      "grad_norm": 1.5716592073440552,
-      "learning_rate": 5.081407035175879e-05,
-      "loss": 5.3689,
-      "step": 49458
-    },
-    {
-      "epoch": 25.79348109517601,
-      "grad_norm": 1.5989402532577515,
-      "learning_rate": 5.081306532663317e-05,
-      "loss": 5.1507,
-      "step": 49459
-    },
-    {
-      "epoch": 25.79400260756193,
-      "grad_norm": 1.685147762298584,
-      "learning_rate": 5.0812060301507535e-05,
-      "loss": 5.3705,
-      "step": 49460
-    },
-    {
-      "epoch": 25.794524119947848,
-      "grad_norm": 1.5592409372329712,
-      "learning_rate": 5.081105527638191e-05,
-      "loss": 5.3501,
-      "step": 49461
-    },
-    {
-      "epoch": 25.795045632333768,
-      "grad_norm": 1.6644999980926514,
-      "learning_rate": 5.081005025125628e-05,
-      "loss": 4.8858,
-      "step": 49462
-    },
-    {
-      "epoch": 25.795567144719687,
-      "grad_norm": 1.5945544242858887,
-      "learning_rate": 5.0809045226130655e-05,
-      "loss": 4.9049,
-      "step": 49463
-    },
-    {
-      "epoch": 25.796088657105607,
-      "grad_norm": 1.6035552024841309,
-      "learning_rate": 5.0808040201005026e-05,
-      "loss": 5.0885,
-      "step": 49464
-    },
-    {
-      "epoch": 25.796610169491526,
-      "grad_norm": 1.6711628437042236,
-      "learning_rate": 5.0807035175879404e-05,
-      "loss": 5.4894,
-      "step": 49465
-    },
-    {
-      "epoch": 25.797131681877445,
-      "grad_norm": 1.5728657245635986,
-      "learning_rate": 5.080603015075377e-05,
-      "loss": 5.2985,
-      "step": 49466
-    },
-    {
-      "epoch": 25.797653194263365,
-      "grad_norm": 1.6542669534683228,
-      "learning_rate": 5.0805025125628146e-05,
-      "loss": 4.7953,
-      "step": 49467
-    },
-    {
-      "epoch": 25.798174706649284,
-      "grad_norm": 1.5409718751907349,
-      "learning_rate": 5.080402010050252e-05,
-      "loss": 4.8169,
-      "step": 49468
-    },
-    {
-      "epoch": 25.7986962190352,
-      "grad_norm": 1.5437167882919312,
-      "learning_rate": 5.0803015075376895e-05,
-      "loss": 5.1603,
-      "step": 49469
-    },
-    {
-      "epoch": 25.79921773142112,
-      "grad_norm": 1.553315281867981,
-      "learning_rate": 5.080201005025126e-05,
-      "loss": 4.8439,
-      "step": 49470
-    },
-    {
-      "epoch": 25.79973924380704,
-      "grad_norm": 1.5841330289840698,
-      "learning_rate": 5.0801005025125624e-05,
-      "loss": 4.9966,
-      "step": 49471
-    },
-    {
-      "epoch": 25.80026075619296,
-      "grad_norm": 1.597845196723938,
-      "learning_rate": 5.08e-05,
-      "loss": 4.8417,
-      "step": 49472
-    },
-    {
-      "epoch": 25.800782268578878,
-      "grad_norm": 1.504225254058838,
-      "learning_rate": 5.079899497487437e-05,
-      "loss": 5.1945,
-      "step": 49473
-    },
-    {
-      "epoch": 25.801303780964798,
-      "grad_norm": 1.521648645401001,
-      "learning_rate": 5.079798994974875e-05,
-      "loss": 5.0905,
-      "step": 49474
-    },
-    {
-      "epoch": 25.801825293350717,
-      "grad_norm": 1.565415620803833,
-      "learning_rate": 5.0796984924623115e-05,
-      "loss": 5.4848,
-      "step": 49475
-    },
-    {
-      "epoch": 25.802346805736637,
-      "grad_norm": 1.434044361114502,
-      "learning_rate": 5.079597989949749e-05,
-      "loss": 5.308,
-      "step": 49476
-    },
-    {
-      "epoch": 25.802868318122556,
-      "grad_norm": 1.4445827007293701,
-      "learning_rate": 5.079497487437186e-05,
-      "loss": 5.3758,
-      "step": 49477
-    },
-    {
-      "epoch": 25.803389830508475,
-      "grad_norm": 1.7290375232696533,
-      "learning_rate": 5.0793969849246235e-05,
-      "loss": 4.9254,
-      "step": 49478
-    },
-    {
-      "epoch": 25.803911342894395,
-      "grad_norm": 1.5250073671340942,
-      "learning_rate": 5.0792964824120606e-05,
-      "loss": 4.9081,
-      "step": 49479
-    },
-    {
-      "epoch": 25.804432855280314,
-      "grad_norm": 1.5946053266525269,
-      "learning_rate": 5.0791959798994983e-05,
-      "loss": 5.5218,
-      "step": 49480
-    },
-    {
-      "epoch": 25.804954367666234,
-      "grad_norm": 1.5173187255859375,
-      "learning_rate": 5.079095477386935e-05,
-      "loss": 5.3331,
-      "step": 49481
-    },
-    {
-      "epoch": 25.80547588005215,
-      "grad_norm": 1.496277093887329,
-      "learning_rate": 5.0789949748743726e-05,
-      "loss": 5.4589,
-      "step": 49482
-    },
-    {
-      "epoch": 25.80599739243807,
-      "grad_norm": 1.5747895240783691,
-      "learning_rate": 5.078894472361809e-05,
-      "loss": 4.696,
-      "step": 49483
-    },
-    {
-      "epoch": 25.80651890482399,
-      "grad_norm": 1.5228025913238525,
-      "learning_rate": 5.078793969849246e-05,
-      "loss": 5.1654,
-      "step": 49484
-    },
-    {
-      "epoch": 25.807040417209908,
-      "grad_norm": 1.5916661024093628,
-      "learning_rate": 5.078693467336684e-05,
-      "loss": 4.9558,
-      "step": 49485
-    },
-    {
-      "epoch": 25.807561929595828,
-      "grad_norm": 1.5315215587615967,
-      "learning_rate": 5.07859296482412e-05,
-      "loss": 5.2467,
-      "step": 49486
-    },
-    {
-      "epoch": 25.808083441981747,
-      "grad_norm": 1.5918775796890259,
-      "learning_rate": 5.078492462311558e-05,
-      "loss": 5.3183,
-      "step": 49487
-    },
-    {
-      "epoch": 25.808604954367667,
-      "grad_norm": 1.5715746879577637,
-      "learning_rate": 5.0783919597989945e-05,
-      "loss": 5.0498,
-      "step": 49488
-    },
-    {
-      "epoch": 25.809126466753586,
-      "grad_norm": 1.623359203338623,
-      "learning_rate": 5.078291457286432e-05,
-      "loss": 5.3226,
-      "step": 49489
-    },
-    {
-      "epoch": 25.809647979139505,
-      "grad_norm": 1.5779606103897095,
-      "learning_rate": 5.0781909547738694e-05,
-      "loss": 5.3027,
-      "step": 49490
-    },
-    {
-      "epoch": 25.810169491525425,
-      "grad_norm": 1.5456421375274658,
-      "learning_rate": 5.078090452261307e-05,
-      "loss": 4.9652,
-      "step": 49491
-    },
-    {
-      "epoch": 25.810691003911344,
-      "grad_norm": 1.5808297395706177,
-      "learning_rate": 5.0779899497487436e-05,
-      "loss": 4.9155,
-      "step": 49492
-    },
-    {
-      "epoch": 25.81121251629726,
-      "grad_norm": 1.6394097805023193,
-      "learning_rate": 5.0778894472361814e-05,
-      "loss": 5.1293,
-      "step": 49493
-    },
-    {
-      "epoch": 25.81173402868318,
-      "grad_norm": 1.595434308052063,
-      "learning_rate": 5.0777889447236185e-05,
-      "loss": 5.1263,
-      "step": 49494
-    },
-    {
-      "epoch": 25.8122555410691,
-      "grad_norm": 1.592504620552063,
-      "learning_rate": 5.077688442211056e-05,
-      "loss": 5.6836,
-      "step": 49495
-    },
-    {
-      "epoch": 25.81277705345502,
-      "grad_norm": 1.6243627071380615,
-      "learning_rate": 5.077587939698493e-05,
-      "loss": 4.4058,
-      "step": 49496
-    },
-    {
-      "epoch": 25.813298565840938,
-      "grad_norm": 1.5522271394729614,
-      "learning_rate": 5.077487437185929e-05,
-      "loss": 5.4176,
-      "step": 49497
-    },
-    {
-      "epoch": 25.813820078226858,
-      "grad_norm": 1.5830094814300537,
-      "learning_rate": 5.077386934673367e-05,
-      "loss": 5.4552,
-      "step": 49498
-    },
-    {
-      "epoch": 25.814341590612777,
-      "grad_norm": 1.4778363704681396,
-      "learning_rate": 5.077286432160804e-05,
-      "loss": 5.4711,
-      "step": 49499
-    },
-    {
-      "epoch": 25.814863102998697,
-      "grad_norm": 1.703542709350586,
-      "learning_rate": 5.077185929648242e-05,
-      "loss": 4.8205,
-      "step": 49500
-    },
-    {
-      "epoch": 25.815384615384616,
-      "grad_norm": 1.4930484294891357,
-      "learning_rate": 5.077085427135678e-05,
-      "loss": 5.5047,
-      "step": 49501
-    },
-    {
-      "epoch": 25.815906127770536,
-      "grad_norm": 1.5209585428237915,
-      "learning_rate": 5.076984924623116e-05,
-      "loss": 4.7171,
-      "step": 49502
-    },
-    {
-      "epoch": 25.816427640156455,
-      "grad_norm": 1.6047848463058472,
-      "learning_rate": 5.0768844221105524e-05,
-      "loss": 5.2137,
-      "step": 49503
-    },
-    {
-      "epoch": 25.816949152542374,
-      "grad_norm": 1.4573417901992798,
-      "learning_rate": 5.07678391959799e-05,
-      "loss": 5.3798,
-      "step": 49504
-    },
-    {
-      "epoch": 25.81747066492829,
-      "grad_norm": 1.5448285341262817,
-      "learning_rate": 5.076683417085427e-05,
-      "loss": 5.1724,
-      "step": 49505
-    },
-    {
-      "epoch": 25.81799217731421,
-      "grad_norm": 1.590221881866455,
-      "learning_rate": 5.076582914572865e-05,
-      "loss": 5.6442,
-      "step": 49506
-    },
-    {
-      "epoch": 25.81851368970013,
-      "grad_norm": 1.5802807807922363,
-      "learning_rate": 5.0764824120603015e-05,
-      "loss": 5.4314,
-      "step": 49507
-    },
-    {
-      "epoch": 25.81903520208605,
-      "grad_norm": 1.6349058151245117,
-      "learning_rate": 5.076381909547739e-05,
-      "loss": 4.8415,
-      "step": 49508
-    },
-    {
-      "epoch": 25.819556714471968,
-      "grad_norm": 1.5726505517959595,
-      "learning_rate": 5.076281407035176e-05,
-      "loss": 5.3082,
-      "step": 49509
-    },
-    {
-      "epoch": 25.820078226857888,
-      "grad_norm": 1.5084272623062134,
-      "learning_rate": 5.0761809045226135e-05,
-      "loss": 5.1957,
-      "step": 49510
-    },
-    {
-      "epoch": 25.820599739243807,
-      "grad_norm": 1.6754173040390015,
-      "learning_rate": 5.0760804020100506e-05,
-      "loss": 5.366,
-      "step": 49511
-    },
-    {
-      "epoch": 25.821121251629727,
-      "grad_norm": 1.5665489435195923,
-      "learning_rate": 5.075979899497487e-05,
-      "loss": 5.4624,
-      "step": 49512
-    },
-    {
-      "epoch": 25.821642764015646,
-      "grad_norm": 1.5872730016708374,
-      "learning_rate": 5.075879396984925e-05,
-      "loss": 5.1742,
-      "step": 49513
-    },
-    {
-      "epoch": 25.822164276401566,
-      "grad_norm": 1.5471688508987427,
-      "learning_rate": 5.075778894472362e-05,
-      "loss": 5.532,
-      "step": 49514
-    },
-    {
-      "epoch": 25.822685788787485,
-      "grad_norm": 1.4969478845596313,
-      "learning_rate": 5.0756783919598e-05,
-      "loss": 5.3935,
-      "step": 49515
-    },
-    {
-      "epoch": 25.823207301173404,
-      "grad_norm": 1.6961020231246948,
-      "learning_rate": 5.075577889447236e-05,
-      "loss": 4.5413,
-      "step": 49516
-    },
-    {
-      "epoch": 25.82372881355932,
-      "grad_norm": 1.6403387784957886,
-      "learning_rate": 5.075477386934674e-05,
-      "loss": 5.3438,
-      "step": 49517
-    },
-    {
-      "epoch": 25.82425032594524,
-      "grad_norm": 1.5857869386672974,
-      "learning_rate": 5.0753768844221104e-05,
-      "loss": 4.8563,
-      "step": 49518
-    },
-    {
-      "epoch": 25.82477183833116,
-      "grad_norm": 1.6066441535949707,
-      "learning_rate": 5.075276381909548e-05,
-      "loss": 5.4583,
-      "step": 49519
-    },
-    {
-      "epoch": 25.82529335071708,
-      "grad_norm": 1.4666965007781982,
-      "learning_rate": 5.075175879396985e-05,
-      "loss": 4.2803,
-      "step": 49520
-    },
-    {
-      "epoch": 25.825814863103,
-      "grad_norm": 1.5224428176879883,
-      "learning_rate": 5.075075376884423e-05,
-      "loss": 4.6428,
-      "step": 49521
-    },
-    {
-      "epoch": 25.826336375488918,
-      "grad_norm": 1.567661166191101,
-      "learning_rate": 5.0749748743718595e-05,
-      "loss": 4.8531,
-      "step": 49522
-    },
-    {
-      "epoch": 25.826857887874837,
-      "grad_norm": 1.5749965906143188,
-      "learning_rate": 5.074874371859297e-05,
-      "loss": 5.6642,
-      "step": 49523
-    },
-    {
-      "epoch": 25.827379400260757,
-      "grad_norm": 1.5790988206863403,
-      "learning_rate": 5.074773869346734e-05,
-      "loss": 4.9828,
-      "step": 49524
-    },
-    {
-      "epoch": 25.827900912646676,
-      "grad_norm": 1.537879467010498,
-      "learning_rate": 5.074673366834171e-05,
-      "loss": 5.6543,
-      "step": 49525
-    },
-    {
-      "epoch": 25.828422425032596,
-      "grad_norm": 1.5758427381515503,
-      "learning_rate": 5.0745728643216086e-05,
-      "loss": 5.0475,
-      "step": 49526
-    },
-    {
-      "epoch": 25.828943937418515,
-      "grad_norm": 1.5924466848373413,
-      "learning_rate": 5.074472361809045e-05,
-      "loss": 5.1402,
-      "step": 49527
-    },
-    {
-      "epoch": 25.829465449804434,
-      "grad_norm": 1.5065189599990845,
-      "learning_rate": 5.074371859296483e-05,
-      "loss": 5.0869,
-      "step": 49528
-    },
-    {
-      "epoch": 25.82998696219035,
-      "grad_norm": 1.7503803968429565,
-      "learning_rate": 5.074271356783919e-05,
-      "loss": 4.9759,
-      "step": 49529
-    },
-    {
-      "epoch": 25.83050847457627,
-      "grad_norm": 1.5792039632797241,
-      "learning_rate": 5.074170854271357e-05,
-      "loss": 5.3077,
-      "step": 49530
-    },
-    {
-      "epoch": 25.83102998696219,
-      "grad_norm": 1.5053184032440186,
-      "learning_rate": 5.074070351758794e-05,
-      "loss": 5.3391,
-      "step": 49531
-    },
-    {
-      "epoch": 25.83155149934811,
-      "grad_norm": 1.6212868690490723,
-      "learning_rate": 5.073969849246232e-05,
-      "loss": 5.2022,
-      "step": 49532
-    },
-    {
-      "epoch": 25.83207301173403,
-      "grad_norm": 1.5017002820968628,
-      "learning_rate": 5.073869346733668e-05,
-      "loss": 5.2441,
-      "step": 49533
-    },
-    {
-      "epoch": 25.832594524119948,
-      "grad_norm": 1.5462223291397095,
-      "learning_rate": 5.073768844221106e-05,
-      "loss": 5.4444,
-      "step": 49534
-    },
-    {
-      "epoch": 25.833116036505867,
-      "grad_norm": 1.5560812950134277,
-      "learning_rate": 5.073668341708543e-05,
-      "loss": 5.6249,
-      "step": 49535
-    },
-    {
-      "epoch": 25.833637548891787,
-      "grad_norm": 1.5451829433441162,
-      "learning_rate": 5.073567839195981e-05,
-      "loss": 5.6649,
-      "step": 49536
-    },
-    {
-      "epoch": 25.834159061277706,
-      "grad_norm": 1.5800509452819824,
-      "learning_rate": 5.0734673366834174e-05,
-      "loss": 5.065,
-      "step": 49537
-    },
-    {
-      "epoch": 25.834680573663626,
-      "grad_norm": 1.554251790046692,
-      "learning_rate": 5.073366834170854e-05,
-      "loss": 5.3505,
-      "step": 49538
-    },
-    {
-      "epoch": 25.835202086049545,
-      "grad_norm": 1.6748102903366089,
-      "learning_rate": 5.0732663316582916e-05,
-      "loss": 4.7725,
-      "step": 49539
-    },
-    {
-      "epoch": 25.835723598435465,
-      "grad_norm": 1.510166883468628,
-      "learning_rate": 5.073165829145729e-05,
-      "loss": 5.3833,
-      "step": 49540
-    },
-    {
-      "epoch": 25.83624511082138,
-      "grad_norm": 1.521899938583374,
-      "learning_rate": 5.0730653266331665e-05,
-      "loss": 5.1791,
-      "step": 49541
-    },
-    {
-      "epoch": 25.8367666232073,
-      "grad_norm": 1.718515157699585,
-      "learning_rate": 5.072964824120603e-05,
-      "loss": 4.7595,
-      "step": 49542
-    },
-    {
-      "epoch": 25.83728813559322,
-      "grad_norm": 1.657780647277832,
-      "learning_rate": 5.072864321608041e-05,
-      "loss": 4.8947,
-      "step": 49543
-    },
-    {
-      "epoch": 25.83780964797914,
-      "grad_norm": 1.6749402284622192,
-      "learning_rate": 5.072763819095477e-05,
-      "loss": 4.6637,
-      "step": 49544
-    },
-    {
-      "epoch": 25.83833116036506,
-      "grad_norm": 1.5060797929763794,
-      "learning_rate": 5.072663316582915e-05,
-      "loss": 4.8567,
-      "step": 49545
-    },
-    {
-      "epoch": 25.838852672750978,
-      "grad_norm": 1.4830248355865479,
-      "learning_rate": 5.072562814070352e-05,
-      "loss": 5.3516,
-      "step": 49546
-    },
-    {
-      "epoch": 25.839374185136897,
-      "grad_norm": 1.5191891193389893,
-      "learning_rate": 5.07246231155779e-05,
-      "loss": 5.3817,
-      "step": 49547
-    },
-    {
-      "epoch": 25.839895697522817,
-      "grad_norm": 1.9443784952163696,
-      "learning_rate": 5.072361809045226e-05,
-      "loss": 4.8821,
-      "step": 49548
-    },
-    {
-      "epoch": 25.840417209908736,
-      "grad_norm": 1.5723484754562378,
-      "learning_rate": 5.072261306532664e-05,
-      "loss": 4.9877,
-      "step": 49549
-    },
-    {
-      "epoch": 25.840938722294656,
-      "grad_norm": 1.6439976692199707,
-      "learning_rate": 5.0721608040201005e-05,
-      "loss": 5.1718,
-      "step": 49550
-    },
-    {
-      "epoch": 25.841460234680575,
-      "grad_norm": 1.6379692554473877,
-      "learning_rate": 5.0720603015075376e-05,
-      "loss": 4.8394,
-      "step": 49551
-    },
-    {
-      "epoch": 25.841981747066495,
-      "grad_norm": 1.6673640012741089,
-      "learning_rate": 5.0719597989949753e-05,
-      "loss": 4.7225,
-      "step": 49552
-    },
-    {
-      "epoch": 25.84250325945241,
-      "grad_norm": 1.5845872163772583,
-      "learning_rate": 5.071859296482412e-05,
-      "loss": 5.0996,
-      "step": 49553
-    },
-    {
-      "epoch": 25.84302477183833,
-      "grad_norm": 1.6790130138397217,
-      "learning_rate": 5.0717587939698496e-05,
-      "loss": 5.0704,
-      "step": 49554
-    },
-    {
-      "epoch": 25.84354628422425,
-      "grad_norm": 1.574972152709961,
-      "learning_rate": 5.0716582914572867e-05,
-      "loss": 5.0735,
-      "step": 49555
-    },
-    {
-      "epoch": 25.84406779661017,
-      "grad_norm": 1.6345430612564087,
-      "learning_rate": 5.0715577889447244e-05,
-      "loss": 5.5593,
-      "step": 49556
-    },
-    {
-      "epoch": 25.84458930899609,
-      "grad_norm": 1.7090246677398682,
-      "learning_rate": 5.071457286432161e-05,
-      "loss": 4.7014,
-      "step": 49557
-    },
-    {
-      "epoch": 25.845110821382008,
-      "grad_norm": 1.5257865190505981,
-      "learning_rate": 5.0713567839195986e-05,
-      "loss": 5.5454,
-      "step": 49558
-    },
-    {
-      "epoch": 25.845632333767927,
-      "grad_norm": 1.5447801351547241,
-      "learning_rate": 5.071256281407035e-05,
-      "loss": 5.4407,
-      "step": 49559
-    },
-    {
-      "epoch": 25.846153846153847,
-      "grad_norm": 1.5845047235488892,
-      "learning_rate": 5.071155778894473e-05,
-      "loss": 4.9534,
-      "step": 49560
-    },
-    {
-      "epoch": 25.846675358539766,
-      "grad_norm": 1.511177897453308,
-      "learning_rate": 5.07105527638191e-05,
-      "loss": 5.4336,
-      "step": 49561
-    },
-    {
-      "epoch": 25.847196870925686,
-      "grad_norm": 1.6541502475738525,
-      "learning_rate": 5.070954773869348e-05,
-      "loss": 4.9978,
-      "step": 49562
-    },
-    {
-      "epoch": 25.847718383311605,
-      "grad_norm": 1.5122650861740112,
-      "learning_rate": 5.070854271356784e-05,
-      "loss": 5.0489,
-      "step": 49563
-    },
-    {
-      "epoch": 25.848239895697525,
-      "grad_norm": 1.5937185287475586,
-      "learning_rate": 5.0707537688442206e-05,
-      "loss": 5.2806,
-      "step": 49564
-    },
-    {
-      "epoch": 25.84876140808344,
-      "grad_norm": 1.6221301555633545,
-      "learning_rate": 5.0706532663316584e-05,
-      "loss": 5.0985,
-      "step": 49565
-    },
-    {
-      "epoch": 25.84928292046936,
-      "grad_norm": 1.639850378036499,
-      "learning_rate": 5.0705527638190955e-05,
-      "loss": 5.1789,
-      "step": 49566
-    },
-    {
-      "epoch": 25.84980443285528,
-      "grad_norm": 1.6068971157073975,
-      "learning_rate": 5.070452261306533e-05,
-      "loss": 5.0583,
-      "step": 49567
-    },
-    {
-      "epoch": 25.8503259452412,
-      "grad_norm": 1.5830589532852173,
-      "learning_rate": 5.07035175879397e-05,
-      "loss": 5.1855,
-      "step": 49568
-    },
-    {
-      "epoch": 25.85084745762712,
-      "grad_norm": 1.5380163192749023,
-      "learning_rate": 5.0702512562814075e-05,
-      "loss": 5.1835,
-      "step": 49569
-    },
-    {
-      "epoch": 25.851368970013038,
-      "grad_norm": 1.6589494943618774,
-      "learning_rate": 5.070150753768844e-05,
-      "loss": 5.4719,
-      "step": 49570
-    },
-    {
-      "epoch": 25.851890482398957,
-      "grad_norm": 1.7805697917938232,
-      "learning_rate": 5.070050251256282e-05,
-      "loss": 4.4424,
-      "step": 49571
-    },
-    {
-      "epoch": 25.852411994784877,
-      "grad_norm": 1.515160083770752,
-      "learning_rate": 5.069949748743719e-05,
-      "loss": 5.3538,
-      "step": 49572
-    },
-    {
-      "epoch": 25.852933507170796,
-      "grad_norm": 1.621850848197937,
-      "learning_rate": 5.0698492462311566e-05,
-      "loss": 5.3807,
-      "step": 49573
-    },
-    {
-      "epoch": 25.853455019556716,
-      "grad_norm": 1.5697494745254517,
-      "learning_rate": 5.069748743718593e-05,
-      "loss": 5.2471,
-      "step": 49574
-    },
-    {
-      "epoch": 25.853976531942635,
-      "grad_norm": 1.5473352670669556,
-      "learning_rate": 5.069648241206031e-05,
-      "loss": 5.519,
-      "step": 49575
-    },
-    {
-      "epoch": 25.85449804432855,
-      "grad_norm": 1.619220495223999,
-      "learning_rate": 5.069547738693467e-05,
-      "loss": 5.0582,
-      "step": 49576
-    },
-    {
-      "epoch": 25.85501955671447,
-      "grad_norm": 1.5892395973205566,
-      "learning_rate": 5.069447236180904e-05,
-      "loss": 5.558,
-      "step": 49577
-    },
-    {
-      "epoch": 25.85554106910039,
-      "grad_norm": 1.5641698837280273,
-      "learning_rate": 5.069346733668342e-05,
-      "loss": 4.6808,
-      "step": 49578
-    },
-    {
-      "epoch": 25.85606258148631,
-      "grad_norm": 1.5560139417648315,
-      "learning_rate": 5.0692462311557785e-05,
-      "loss": 5.5736,
-      "step": 49579
-    },
-    {
-      "epoch": 25.85658409387223,
-      "grad_norm": 1.6315935850143433,
-      "learning_rate": 5.069145728643216e-05,
-      "loss": 5.2588,
-      "step": 49580
-    },
-    {
-      "epoch": 25.85710560625815,
-      "grad_norm": 1.50831139087677,
-      "learning_rate": 5.0690452261306534e-05,
-      "loss": 4.7513,
-      "step": 49581
-    },
-    {
-      "epoch": 25.857627118644068,
-      "grad_norm": 1.5721049308776855,
-      "learning_rate": 5.068944723618091e-05,
-      "loss": 5.3292,
-      "step": 49582
-    },
-    {
-      "epoch": 25.858148631029987,
-      "grad_norm": 1.5985435247421265,
-      "learning_rate": 5.0688442211055276e-05,
-      "loss": 5.5423,
-      "step": 49583
-    },
-    {
-      "epoch": 25.858670143415907,
-      "grad_norm": 1.6277351379394531,
-      "learning_rate": 5.0687437185929654e-05,
-      "loss": 5.101,
-      "step": 49584
-    },
-    {
-      "epoch": 25.859191655801826,
-      "grad_norm": 1.6405960321426392,
-      "learning_rate": 5.068643216080402e-05,
-      "loss": 5.3736,
-      "step": 49585
-    },
-    {
-      "epoch": 25.859713168187746,
-      "grad_norm": 1.5931187868118286,
-      "learning_rate": 5.0685427135678396e-05,
-      "loss": 5.1179,
-      "step": 49586
-    },
-    {
-      "epoch": 25.860234680573665,
-      "grad_norm": 1.5884819030761719,
-      "learning_rate": 5.068442211055277e-05,
-      "loss": 4.4604,
-      "step": 49587
-    },
-    {
-      "epoch": 25.860756192959585,
-      "grad_norm": 1.5796113014221191,
-      "learning_rate": 5.0683417085427145e-05,
-      "loss": 5.3554,
-      "step": 49588
-    },
-    {
-      "epoch": 25.8612777053455,
-      "grad_norm": 1.6691526174545288,
-      "learning_rate": 5.068241206030151e-05,
-      "loss": 5.0553,
-      "step": 49589
-    },
-    {
-      "epoch": 25.86179921773142,
-      "grad_norm": 1.6295133829116821,
-      "learning_rate": 5.0681407035175874e-05,
-      "loss": 5.5442,
-      "step": 49590
-    },
-    {
-      "epoch": 25.86232073011734,
-      "grad_norm": 1.6541332006454468,
-      "learning_rate": 5.068040201005025e-05,
-      "loss": 5.1741,
-      "step": 49591
-    },
-    {
-      "epoch": 25.86284224250326,
-      "grad_norm": 1.626213550567627,
-      "learning_rate": 5.067939698492462e-05,
-      "loss": 5.3281,
-      "step": 49592
-    },
-    {
-      "epoch": 25.86336375488918,
-      "grad_norm": 1.5253450870513916,
-      "learning_rate": 5.0678391959799e-05,
-      "loss": 5.7736,
-      "step": 49593
-    },
-    {
-      "epoch": 25.863885267275098,
-      "grad_norm": 1.6202216148376465,
-      "learning_rate": 5.0677386934673365e-05,
-      "loss": 5.2085,
-      "step": 49594
-    },
-    {
-      "epoch": 25.864406779661017,
-      "grad_norm": 1.5051106214523315,
-      "learning_rate": 5.067638190954774e-05,
-      "loss": 5.3637,
-      "step": 49595
-    },
-    {
-      "epoch": 25.864928292046937,
-      "grad_norm": 1.5268385410308838,
-      "learning_rate": 5.067537688442211e-05,
-      "loss": 5.0723,
-      "step": 49596
-    },
-    {
-      "epoch": 25.865449804432856,
-      "grad_norm": 1.5474015474319458,
-      "learning_rate": 5.0674371859296485e-05,
-      "loss": 5.389,
-      "step": 49597
-    },
-    {
-      "epoch": 25.865971316818776,
-      "grad_norm": 1.705034613609314,
-      "learning_rate": 5.0673366834170856e-05,
-      "loss": 5.271,
-      "step": 49598
-    },
-    {
-      "epoch": 25.866492829204695,
-      "grad_norm": 1.6580655574798584,
-      "learning_rate": 5.0672361809045233e-05,
-      "loss": 5.2843,
-      "step": 49599
-    },
-    {
-      "epoch": 25.86701434159061,
-      "grad_norm": 1.6960854530334473,
-      "learning_rate": 5.06713567839196e-05,
-      "loss": 5.2925,
-      "step": 49600
-    },
-    {
-      "epoch": 25.86753585397653,
-      "grad_norm": 1.4683719873428345,
-      "learning_rate": 5.0670351758793976e-05,
-      "loss": 4.6469,
-      "step": 49601
-    },
-    {
-      "epoch": 25.86805736636245,
-      "grad_norm": 1.6177430152893066,
-      "learning_rate": 5.066934673366835e-05,
-      "loss": 5.3442,
-      "step": 49602
-    },
-    {
-      "epoch": 25.86857887874837,
-      "grad_norm": 1.542026400566101,
-      "learning_rate": 5.0668341708542724e-05,
-      "loss": 5.36,
-      "step": 49603
-    },
-    {
-      "epoch": 25.86910039113429,
-      "grad_norm": 1.4495600461959839,
-      "learning_rate": 5.066733668341709e-05,
-      "loss": 5.1327,
-      "step": 49604
-    },
-    {
-      "epoch": 25.86962190352021,
-      "grad_norm": 1.596571445465088,
-      "learning_rate": 5.066633165829145e-05,
-      "loss": 4.9321,
-      "step": 49605
-    },
-    {
-      "epoch": 25.870143415906128,
-      "grad_norm": 1.484727382659912,
-      "learning_rate": 5.066532663316583e-05,
-      "loss": 4.9196,
-      "step": 49606
-    },
-    {
-      "epoch": 25.870664928292047,
-      "grad_norm": 1.60414719581604,
-      "learning_rate": 5.06643216080402e-05,
-      "loss": 4.8008,
-      "step": 49607
-    },
-    {
-      "epoch": 25.871186440677967,
-      "grad_norm": 1.5871433019638062,
-      "learning_rate": 5.066331658291458e-05,
-      "loss": 5.045,
-      "step": 49608
-    },
-    {
-      "epoch": 25.871707953063886,
-      "grad_norm": 1.554621696472168,
-      "learning_rate": 5.0662311557788944e-05,
-      "loss": 4.867,
-      "step": 49609
-    },
-    {
-      "epoch": 25.872229465449806,
-      "grad_norm": 1.4861476421356201,
-      "learning_rate": 5.066130653266332e-05,
-      "loss": 5.5229,
-      "step": 49610
-    },
-    {
-      "epoch": 25.872750977835725,
-      "grad_norm": 1.4941426515579224,
-      "learning_rate": 5.0660301507537686e-05,
-      "loss": 5.2716,
-      "step": 49611
-    },
-    {
-      "epoch": 25.87327249022164,
-      "grad_norm": 1.6487493515014648,
-      "learning_rate": 5.0659296482412064e-05,
-      "loss": 4.7105,
-      "step": 49612
-    },
-    {
-      "epoch": 25.87379400260756,
-      "grad_norm": 1.5879939794540405,
-      "learning_rate": 5.0658291457286435e-05,
-      "loss": 4.8758,
-      "step": 49613
-    },
-    {
-      "epoch": 25.87431551499348,
-      "grad_norm": 1.650282382965088,
-      "learning_rate": 5.065728643216081e-05,
-      "loss": 5.5628,
-      "step": 49614
-    },
-    {
-      "epoch": 25.8748370273794,
-      "grad_norm": 1.5934847593307495,
-      "learning_rate": 5.065628140703518e-05,
-      "loss": 5.2864,
-      "step": 49615
-    },
-    {
-      "epoch": 25.87535853976532,
-      "grad_norm": 1.8107908964157104,
-      "learning_rate": 5.0655276381909555e-05,
-      "loss": 5.0097,
-      "step": 49616
-    },
-    {
-      "epoch": 25.87588005215124,
-      "grad_norm": 1.6516040563583374,
-      "learning_rate": 5.065427135678392e-05,
-      "loss": 5.2388,
-      "step": 49617
-    },
-    {
-      "epoch": 25.876401564537158,
-      "grad_norm": 1.605051875114441,
-      "learning_rate": 5.065326633165829e-05,
-      "loss": 5.4741,
-      "step": 49618
-    },
-    {
-      "epoch": 25.876923076923077,
-      "grad_norm": 1.514475703239441,
-      "learning_rate": 5.065226130653267e-05,
-      "loss": 4.9164,
-      "step": 49619
-    },
-    {
-      "epoch": 25.877444589308997,
-      "grad_norm": 1.6271687746047974,
-      "learning_rate": 5.065125628140703e-05,
-      "loss": 5.1857,
-      "step": 49620
-    },
-    {
-      "epoch": 25.877966101694916,
-      "grad_norm": 1.5166915655136108,
-      "learning_rate": 5.065025125628141e-05,
-      "loss": 5.5638,
-      "step": 49621
-    },
-    {
-      "epoch": 25.878487614080836,
-      "grad_norm": 1.6739660501480103,
-      "learning_rate": 5.064924623115578e-05,
-      "loss": 4.8131,
-      "step": 49622
-    },
-    {
-      "epoch": 25.879009126466755,
-      "grad_norm": 1.5725338459014893,
-      "learning_rate": 5.064824120603016e-05,
-      "loss": 5.4286,
-      "step": 49623
-    },
-    {
-      "epoch": 25.87953063885267,
-      "grad_norm": 1.4778566360473633,
-      "learning_rate": 5.064723618090452e-05,
-      "loss": 5.46,
-      "step": 49624
-    },
-    {
-      "epoch": 25.88005215123859,
-      "grad_norm": 1.5577316284179688,
-      "learning_rate": 5.06462311557789e-05,
-      "loss": 5.5285,
-      "step": 49625
-    },
-    {
-      "epoch": 25.88057366362451,
-      "grad_norm": 1.6236218214035034,
-      "learning_rate": 5.0645226130653265e-05,
-      "loss": 5.229,
-      "step": 49626
-    },
-    {
-      "epoch": 25.88109517601043,
-      "grad_norm": 1.7519948482513428,
-      "learning_rate": 5.064422110552764e-05,
-      "loss": 5.1624,
-      "step": 49627
-    },
-    {
-      "epoch": 25.88161668839635,
-      "grad_norm": 1.4752305746078491,
-      "learning_rate": 5.0643216080402014e-05,
-      "loss": 5.5079,
-      "step": 49628
-    },
-    {
-      "epoch": 25.88213820078227,
-      "grad_norm": 1.456191062927246,
-      "learning_rate": 5.064221105527639e-05,
-      "loss": 5.6835,
-      "step": 49629
-    },
-    {
-      "epoch": 25.882659713168188,
-      "grad_norm": 1.532139778137207,
-      "learning_rate": 5.0641206030150756e-05,
-      "loss": 5.3627,
-      "step": 49630
-    },
-    {
-      "epoch": 25.883181225554107,
-      "grad_norm": 1.4621633291244507,
-      "learning_rate": 5.064020100502512e-05,
-      "loss": 4.7263,
-      "step": 49631
-    },
-    {
-      "epoch": 25.883702737940027,
-      "grad_norm": 1.5841830968856812,
-      "learning_rate": 5.06391959798995e-05,
-      "loss": 5.3692,
-      "step": 49632
-    },
-    {
-      "epoch": 25.884224250325946,
-      "grad_norm": 1.4491742849349976,
-      "learning_rate": 5.063819095477387e-05,
-      "loss": 5.6551,
-      "step": 49633
-    },
-    {
-      "epoch": 25.884745762711866,
-      "grad_norm": 1.5217739343643188,
-      "learning_rate": 5.063718592964825e-05,
-      "loss": 5.4256,
-      "step": 49634
-    },
-    {
-      "epoch": 25.885267275097785,
-      "grad_norm": 1.5143941640853882,
-      "learning_rate": 5.063618090452261e-05,
-      "loss": 4.2643,
-      "step": 49635
-    },
-    {
-      "epoch": 25.8857887874837,
-      "grad_norm": 1.5677764415740967,
-      "learning_rate": 5.063517587939699e-05,
-      "loss": 4.901,
-      "step": 49636
-    },
-    {
-      "epoch": 25.88631029986962,
-      "grad_norm": 1.5307576656341553,
-      "learning_rate": 5.0634170854271354e-05,
-      "loss": 5.0591,
-      "step": 49637
-    },
-    {
-      "epoch": 25.88683181225554,
-      "grad_norm": 1.4900758266448975,
-      "learning_rate": 5.063316582914573e-05,
-      "loss": 5.6119,
-      "step": 49638
-    },
-    {
-      "epoch": 25.88735332464146,
-      "grad_norm": 1.6042336225509644,
-      "learning_rate": 5.06321608040201e-05,
-      "loss": 5.2681,
-      "step": 49639
-    },
-    {
-      "epoch": 25.88787483702738,
-      "grad_norm": 1.5299960374832153,
-      "learning_rate": 5.063115577889448e-05,
-      "loss": 5.3143,
-      "step": 49640
-    },
-    {
-      "epoch": 25.8883963494133,
-      "grad_norm": 1.6244276762008667,
-      "learning_rate": 5.0630150753768845e-05,
-      "loss": 5.3325,
-      "step": 49641
-    },
-    {
-      "epoch": 25.888917861799218,
-      "grad_norm": 1.699157476425171,
-      "learning_rate": 5.062914572864322e-05,
-      "loss": 4.7789,
-      "step": 49642
-    },
-    {
-      "epoch": 25.889439374185137,
-      "grad_norm": 1.4542511701583862,
-      "learning_rate": 5.0628140703517594e-05,
-      "loss": 5.3077,
-      "step": 49643
-    },
-    {
-      "epoch": 25.889960886571057,
-      "grad_norm": 1.4602171182632446,
-      "learning_rate": 5.062713567839196e-05,
-      "loss": 5.4686,
-      "step": 49644
-    },
-    {
-      "epoch": 25.890482398956976,
-      "grad_norm": 1.582260012626648,
-      "learning_rate": 5.0626130653266336e-05,
-      "loss": 5.0797,
-      "step": 49645
-    },
-    {
-      "epoch": 25.891003911342896,
-      "grad_norm": 1.5520819425582886,
-      "learning_rate": 5.06251256281407e-05,
-      "loss": 5.6351,
-      "step": 49646
-    },
-    {
-      "epoch": 25.891525423728815,
-      "grad_norm": 1.5767394304275513,
-      "learning_rate": 5.062412060301508e-05,
-      "loss": 5.147,
-      "step": 49647
-    },
-    {
-      "epoch": 25.89204693611473,
-      "grad_norm": 1.568686842918396,
-      "learning_rate": 5.062311557788945e-05,
-      "loss": 4.9463,
-      "step": 49648
-    },
-    {
-      "epoch": 25.89256844850065,
-      "grad_norm": 1.5151571035385132,
-      "learning_rate": 5.062211055276383e-05,
-      "loss": 5.5886,
-      "step": 49649
-    },
-    {
-      "epoch": 25.89308996088657,
-      "grad_norm": 1.6559059619903564,
-      "learning_rate": 5.062110552763819e-05,
-      "loss": 5.3048,
-      "step": 49650
-    },
-    {
-      "epoch": 25.89361147327249,
-      "grad_norm": 1.7158441543579102,
-      "learning_rate": 5.062010050251257e-05,
-      "loss": 5.0173,
-      "step": 49651
-    },
-    {
-      "epoch": 25.89413298565841,
-      "grad_norm": 1.6177597045898438,
-      "learning_rate": 5.061909547738693e-05,
-      "loss": 5.4282,
-      "step": 49652
-    },
-    {
-      "epoch": 25.89465449804433,
-      "grad_norm": 1.5130457878112793,
-      "learning_rate": 5.061809045226131e-05,
-      "loss": 5.3224,
-      "step": 49653
-    },
-    {
-      "epoch": 25.895176010430248,
-      "grad_norm": 1.5841517448425293,
-      "learning_rate": 5.061708542713568e-05,
-      "loss": 5.1208,
-      "step": 49654
-    },
-    {
-      "epoch": 25.895697522816167,
-      "grad_norm": 1.5145801305770874,
-      "learning_rate": 5.061608040201006e-05,
-      "loss": 5.3165,
-      "step": 49655
-    },
-    {
-      "epoch": 25.896219035202087,
-      "grad_norm": 1.5488249063491821,
-      "learning_rate": 5.0615075376884424e-05,
-      "loss": 5.2356,
-      "step": 49656
-    },
-    {
-      "epoch": 25.896740547588006,
-      "grad_norm": 1.5108458995819092,
-      "learning_rate": 5.061407035175879e-05,
-      "loss": 5.8774,
-      "step": 49657
-    },
-    {
-      "epoch": 25.897262059973926,
-      "grad_norm": 1.5750727653503418,
-      "learning_rate": 5.0613065326633166e-05,
-      "loss": 5.3238,
-      "step": 49658
-    },
-    {
-      "epoch": 25.89778357235984,
-      "grad_norm": 1.619541883468628,
-      "learning_rate": 5.061206030150754e-05,
-      "loss": 5.33,
-      "step": 49659
-    },
-    {
-      "epoch": 25.89830508474576,
-      "grad_norm": 1.6741149425506592,
-      "learning_rate": 5.0611055276381915e-05,
-      "loss": 4.9091,
-      "step": 49660
-    },
-    {
-      "epoch": 25.89882659713168,
-      "grad_norm": 1.6016485691070557,
-      "learning_rate": 5.061005025125628e-05,
-      "loss": 4.6525,
-      "step": 49661
-    },
-    {
-      "epoch": 25.8993481095176,
-      "grad_norm": 1.70486319065094,
-      "learning_rate": 5.060904522613066e-05,
-      "loss": 5.4695,
-      "step": 49662
-    },
-    {
-      "epoch": 25.89986962190352,
-      "grad_norm": 1.6090404987335205,
-      "learning_rate": 5.060804020100502e-05,
-      "loss": 4.7666,
-      "step": 49663
-    },
-    {
-      "epoch": 25.90039113428944,
-      "grad_norm": 1.538678526878357,
-      "learning_rate": 5.06070351758794e-05,
-      "loss": 4.9463,
-      "step": 49664
-    },
-    {
-      "epoch": 25.90091264667536,
-      "grad_norm": 1.5059574842453003,
-      "learning_rate": 5.060603015075377e-05,
-      "loss": 4.8128,
-      "step": 49665
-    },
-    {
-      "epoch": 25.901434159061278,
-      "grad_norm": 1.604464054107666,
-      "learning_rate": 5.060502512562815e-05,
-      "loss": 5.1057,
-      "step": 49666
-    },
-    {
-      "epoch": 25.901955671447197,
-      "grad_norm": 1.5495764017105103,
-      "learning_rate": 5.060402010050251e-05,
-      "loss": 5.0951,
-      "step": 49667
-    },
-    {
-      "epoch": 25.902477183833117,
-      "grad_norm": 1.6324249505996704,
-      "learning_rate": 5.060301507537689e-05,
-      "loss": 4.887,
-      "step": 49668
-    },
-    {
-      "epoch": 25.902998696219036,
-      "grad_norm": 1.4892849922180176,
-      "learning_rate": 5.060201005025126e-05,
-      "loss": 5.2126,
-      "step": 49669
-    },
-    {
-      "epoch": 25.903520208604956,
-      "grad_norm": 1.588434100151062,
-      "learning_rate": 5.0601005025125626e-05,
-      "loss": 5.2475,
-      "step": 49670
-    },
-    {
-      "epoch": 25.904041720990875,
-      "grad_norm": 1.5233299732208252,
-      "learning_rate": 5.0600000000000003e-05,
-      "loss": 5.771,
-      "step": 49671
-    },
-    {
-      "epoch": 25.90456323337679,
-      "grad_norm": 1.6560766696929932,
-      "learning_rate": 5.059899497487437e-05,
-      "loss": 5.0398,
-      "step": 49672
-    },
-    {
-      "epoch": 25.90508474576271,
-      "grad_norm": 1.53208589553833,
-      "learning_rate": 5.0597989949748746e-05,
-      "loss": 5.3525,
-      "step": 49673
-    },
-    {
-      "epoch": 25.90560625814863,
-      "grad_norm": 1.5583393573760986,
-      "learning_rate": 5.0596984924623117e-05,
-      "loss": 5.3723,
-      "step": 49674
-    },
-    {
-      "epoch": 25.90612777053455,
-      "grad_norm": 1.5406908988952637,
-      "learning_rate": 5.0595979899497494e-05,
-      "loss": 5.3343,
-      "step": 49675
-    },
-    {
-      "epoch": 25.90664928292047,
-      "grad_norm": 1.4924697875976562,
-      "learning_rate": 5.059497487437186e-05,
-      "loss": 5.134,
-      "step": 49676
-    },
-    {
-      "epoch": 25.90717079530639,
-      "grad_norm": 1.5244537591934204,
-      "learning_rate": 5.0593969849246237e-05,
-      "loss": 5.0286,
-      "step": 49677
-    },
-    {
-      "epoch": 25.907692307692308,
-      "grad_norm": 1.6429951190948486,
-      "learning_rate": 5.05929648241206e-05,
-      "loss": 5.1592,
-      "step": 49678
-    },
-    {
-      "epoch": 25.908213820078227,
-      "grad_norm": 1.4395991563796997,
-      "learning_rate": 5.059195979899498e-05,
-      "loss": 5.3489,
-      "step": 49679
-    },
-    {
-      "epoch": 25.908735332464147,
-      "grad_norm": 1.572118878364563,
-      "learning_rate": 5.059095477386935e-05,
-      "loss": 4.4116,
-      "step": 49680
-    },
-    {
-      "epoch": 25.909256844850066,
-      "grad_norm": 1.59121835231781,
-      "learning_rate": 5.058994974874373e-05,
-      "loss": 5.4703,
-      "step": 49681
-    },
-    {
-      "epoch": 25.909778357235986,
-      "grad_norm": 1.5811995267868042,
-      "learning_rate": 5.058894472361809e-05,
-      "loss": 5.1028,
-      "step": 49682
-    },
-    {
-      "epoch": 25.910299869621902,
-      "grad_norm": 1.6827082633972168,
-      "learning_rate": 5.058793969849247e-05,
-      "loss": 4.8138,
-      "step": 49683
-    },
-    {
-      "epoch": 25.91082138200782,
-      "grad_norm": 1.6263772249221802,
-      "learning_rate": 5.0586934673366834e-05,
-      "loss": 4.7933,
-      "step": 49684
-    },
-    {
-      "epoch": 25.91134289439374,
-      "grad_norm": 1.5300601720809937,
-      "learning_rate": 5.0585929648241205e-05,
-      "loss": 5.6238,
-      "step": 49685
-    },
-    {
-      "epoch": 25.91186440677966,
-      "grad_norm": 1.553573489189148,
-      "learning_rate": 5.058492462311558e-05,
-      "loss": 5.4116,
-      "step": 49686
-    },
-    {
-      "epoch": 25.91238591916558,
-      "grad_norm": 1.5374358892440796,
-      "learning_rate": 5.058391959798995e-05,
-      "loss": 5.2711,
-      "step": 49687
-    },
-    {
-      "epoch": 25.9129074315515,
-      "grad_norm": 1.6005572080612183,
-      "learning_rate": 5.0582914572864325e-05,
-      "loss": 4.8762,
-      "step": 49688
-    },
-    {
-      "epoch": 25.91342894393742,
-      "grad_norm": 1.5607010126113892,
-      "learning_rate": 5.0581909547738696e-05,
-      "loss": 5.2806,
-      "step": 49689
-    },
-    {
-      "epoch": 25.913950456323338,
-      "grad_norm": 1.5925556421279907,
-      "learning_rate": 5.0580904522613074e-05,
-      "loss": 5.0532,
-      "step": 49690
-    },
-    {
-      "epoch": 25.914471968709258,
-      "grad_norm": 1.4775947332382202,
-      "learning_rate": 5.057989949748744e-05,
-      "loss": 5.6018,
-      "step": 49691
-    },
-    {
-      "epoch": 25.914993481095177,
-      "grad_norm": 1.5478781461715698,
-      "learning_rate": 5.0578894472361816e-05,
-      "loss": 5.6898,
-      "step": 49692
-    },
-    {
-      "epoch": 25.915514993481096,
-      "grad_norm": 1.533357858657837,
-      "learning_rate": 5.057788944723618e-05,
-      "loss": 5.3855,
-      "step": 49693
-    },
-    {
-      "epoch": 25.916036505867016,
-      "grad_norm": 1.7544904947280884,
-      "learning_rate": 5.057688442211056e-05,
-      "loss": 5.0734,
-      "step": 49694
-    },
-    {
-      "epoch": 25.916558018252932,
-      "grad_norm": 1.5242489576339722,
-      "learning_rate": 5.057587939698493e-05,
-      "loss": 5.3787,
-      "step": 49695
-    },
-    {
-      "epoch": 25.91707953063885,
-      "grad_norm": 1.6481091976165771,
-      "learning_rate": 5.057487437185931e-05,
-      "loss": 5.131,
-      "step": 49696
-    },
-    {
-      "epoch": 25.91760104302477,
-      "grad_norm": 1.5271209478378296,
-      "learning_rate": 5.057386934673367e-05,
-      "loss": 5.122,
-      "step": 49697
-    },
-    {
-      "epoch": 25.91812255541069,
-      "grad_norm": 1.630136489868164,
-      "learning_rate": 5.0572864321608035e-05,
-      "loss": 4.5078,
-      "step": 49698
-    },
-    {
-      "epoch": 25.91864406779661,
-      "grad_norm": 1.5832562446594238,
-      "learning_rate": 5.057185929648241e-05,
-      "loss": 5.2866,
-      "step": 49699
-    },
-    {
-      "epoch": 25.91916558018253,
-      "grad_norm": 1.7158887386322021,
-      "learning_rate": 5.0570854271356784e-05,
-      "loss": 5.1356,
-      "step": 49700
-    },
-    {
-      "epoch": 25.91968709256845,
-      "grad_norm": 1.6296045780181885,
-      "learning_rate": 5.056984924623116e-05,
-      "loss": 5.0555,
-      "step": 49701
-    },
-    {
-      "epoch": 25.920208604954368,
-      "grad_norm": 1.5664219856262207,
-      "learning_rate": 5.0568844221105526e-05,
-      "loss": 5.17,
-      "step": 49702
-    },
-    {
-      "epoch": 25.920730117340288,
-      "grad_norm": 1.585598349571228,
-      "learning_rate": 5.0567839195979904e-05,
-      "loss": 5.295,
-      "step": 49703
-    },
-    {
-      "epoch": 25.921251629726207,
-      "grad_norm": 1.599936842918396,
-      "learning_rate": 5.056683417085427e-05,
-      "loss": 5.3111,
-      "step": 49704
-    },
-    {
-      "epoch": 25.921773142112126,
-      "grad_norm": 1.583478569984436,
-      "learning_rate": 5.0565829145728646e-05,
-      "loss": 5.1241,
-      "step": 49705
-    },
-    {
-      "epoch": 25.922294654498046,
-      "grad_norm": 1.6150143146514893,
-      "learning_rate": 5.056482412060302e-05,
-      "loss": 5.037,
-      "step": 49706
-    },
-    {
-      "epoch": 25.922816166883962,
-      "grad_norm": 1.567314863204956,
-      "learning_rate": 5.0563819095477395e-05,
-      "loss": 4.809,
-      "step": 49707
-    },
-    {
-      "epoch": 25.92333767926988,
-      "grad_norm": 1.5850931406021118,
-      "learning_rate": 5.056281407035176e-05,
-      "loss": 5.2791,
-      "step": 49708
-    },
-    {
-      "epoch": 25.9238591916558,
-      "grad_norm": 1.5912948846817017,
-      "learning_rate": 5.056180904522614e-05,
-      "loss": 5.1318,
-      "step": 49709
-    },
-    {
-      "epoch": 25.92438070404172,
-      "grad_norm": 1.5874093770980835,
-      "learning_rate": 5.056080402010051e-05,
-      "loss": 4.8861,
-      "step": 49710
-    },
-    {
-      "epoch": 25.92490221642764,
-      "grad_norm": 1.5220916271209717,
-      "learning_rate": 5.055979899497487e-05,
-      "loss": 5.2674,
-      "step": 49711
-    },
-    {
-      "epoch": 25.92542372881356,
-      "grad_norm": 1.5719192028045654,
-      "learning_rate": 5.055879396984925e-05,
-      "loss": 5.2808,
-      "step": 49712
-    },
-    {
-      "epoch": 25.92594524119948,
-      "grad_norm": 1.5864789485931396,
-      "learning_rate": 5.0557788944723615e-05,
-      "loss": 5.4356,
-      "step": 49713
-    },
-    {
-      "epoch": 25.926466753585398,
-      "grad_norm": 1.6272556781768799,
-      "learning_rate": 5.055678391959799e-05,
-      "loss": 5.2329,
-      "step": 49714
-    },
-    {
-      "epoch": 25.926988265971318,
-      "grad_norm": 1.5981441736221313,
-      "learning_rate": 5.0555778894472364e-05,
-      "loss": 4.7477,
-      "step": 49715
-    },
-    {
-      "epoch": 25.927509778357237,
-      "grad_norm": 1.6219923496246338,
-      "learning_rate": 5.055477386934674e-05,
-      "loss": 5.0067,
-      "step": 49716
-    },
-    {
-      "epoch": 25.928031290743156,
-      "grad_norm": 1.595299482345581,
-      "learning_rate": 5.0553768844221106e-05,
-      "loss": 5.4259,
-      "step": 49717
-    },
-    {
-      "epoch": 25.928552803129076,
-      "grad_norm": 1.6182856559753418,
-      "learning_rate": 5.0552763819095484e-05,
-      "loss": 5.199,
-      "step": 49718
-    },
-    {
-      "epoch": 25.929074315514992,
-      "grad_norm": 1.6206239461898804,
-      "learning_rate": 5.055175879396985e-05,
-      "loss": 5.3385,
-      "step": 49719
-    },
-    {
-      "epoch": 25.92959582790091,
-      "grad_norm": 1.5726951360702515,
-      "learning_rate": 5.0550753768844226e-05,
-      "loss": 5.7093,
-      "step": 49720
-    },
-    {
-      "epoch": 25.93011734028683,
-      "grad_norm": 1.5933152437210083,
-      "learning_rate": 5.05497487437186e-05,
-      "loss": 4.9724,
-      "step": 49721
-    },
-    {
-      "epoch": 25.93063885267275,
-      "grad_norm": 1.5899559259414673,
-      "learning_rate": 5.0548743718592974e-05,
-      "loss": 4.9525,
-      "step": 49722
-    },
-    {
-      "epoch": 25.93116036505867,
-      "grad_norm": 1.8017326593399048,
-      "learning_rate": 5.054773869346734e-05,
-      "loss": 4.62,
-      "step": 49723
-    },
-    {
-      "epoch": 25.93168187744459,
-      "grad_norm": 1.50344979763031,
-      "learning_rate": 5.05467336683417e-05,
-      "loss": 5.4385,
-      "step": 49724
-    },
-    {
-      "epoch": 25.93220338983051,
-      "grad_norm": 1.588431715965271,
-      "learning_rate": 5.054572864321608e-05,
-      "loss": 4.9762,
-      "step": 49725
-    },
-    {
-      "epoch": 25.932724902216428,
-      "grad_norm": 1.6521419286727905,
-      "learning_rate": 5.054472361809045e-05,
-      "loss": 5.5268,
-      "step": 49726
-    },
-    {
-      "epoch": 25.933246414602348,
-      "grad_norm": 1.5612716674804688,
-      "learning_rate": 5.054371859296483e-05,
-      "loss": 5.0163,
-      "step": 49727
-    },
-    {
-      "epoch": 25.933767926988267,
-      "grad_norm": 1.5705504417419434,
-      "learning_rate": 5.0542713567839194e-05,
-      "loss": 5.1319,
-      "step": 49728
-    },
-    {
-      "epoch": 25.934289439374187,
-      "grad_norm": 1.5617585182189941,
-      "learning_rate": 5.054170854271357e-05,
-      "loss": 5.2808,
-      "step": 49729
-    },
-    {
-      "epoch": 25.934810951760106,
-      "grad_norm": 1.469019889831543,
-      "learning_rate": 5.054070351758794e-05,
-      "loss": 5.7607,
-      "step": 49730
-    },
-    {
-      "epoch": 25.935332464146022,
-      "grad_norm": 1.6591640710830688,
-      "learning_rate": 5.053969849246232e-05,
-      "loss": 5.4767,
-      "step": 49731
-    },
-    {
-      "epoch": 25.93585397653194,
-      "grad_norm": 1.5485382080078125,
-      "learning_rate": 5.0538693467336685e-05,
-      "loss": 5.4298,
-      "step": 49732
-    },
-    {
-      "epoch": 25.93637548891786,
-      "grad_norm": 1.4884490966796875,
-      "learning_rate": 5.053768844221106e-05,
-      "loss": 5.3236,
-      "step": 49733
-    },
-    {
-      "epoch": 25.93689700130378,
-      "grad_norm": 1.6683859825134277,
-      "learning_rate": 5.053668341708543e-05,
-      "loss": 4.5523,
-      "step": 49734
-    },
-    {
-      "epoch": 25.9374185136897,
-      "grad_norm": 1.5376648902893066,
-      "learning_rate": 5.0535678391959805e-05,
-      "loss": 5.3399,
-      "step": 49735
-    },
-    {
-      "epoch": 25.93794002607562,
-      "grad_norm": 1.5781092643737793,
-      "learning_rate": 5.0534673366834176e-05,
-      "loss": 5.0177,
-      "step": 49736
-    },
-    {
-      "epoch": 25.93846153846154,
-      "grad_norm": 1.507241129875183,
-      "learning_rate": 5.053366834170854e-05,
-      "loss": 5.4279,
-      "step": 49737
-    },
-    {
-      "epoch": 25.938983050847458,
-      "grad_norm": 1.5954539775848389,
-      "learning_rate": 5.053266331658292e-05,
-      "loss": 5.5577,
-      "step": 49738
-    },
-    {
-      "epoch": 25.939504563233378,
-      "grad_norm": 1.5570346117019653,
-      "learning_rate": 5.053165829145728e-05,
-      "loss": 5.4836,
-      "step": 49739
-    },
-    {
-      "epoch": 25.940026075619297,
-      "grad_norm": 1.5413252115249634,
-      "learning_rate": 5.053065326633166e-05,
-      "loss": 5.2395,
-      "step": 49740
-    },
-    {
-      "epoch": 25.940547588005217,
-      "grad_norm": 1.6190102100372314,
-      "learning_rate": 5.052964824120603e-05,
-      "loss": 5.1257,
-      "step": 49741
-    },
-    {
-      "epoch": 25.941069100391136,
-      "grad_norm": 1.6561888456344604,
-      "learning_rate": 5.052864321608041e-05,
-      "loss": 4.9659,
-      "step": 49742
-    },
-    {
-      "epoch": 25.941590612777052,
-      "grad_norm": 1.714999794960022,
-      "learning_rate": 5.052763819095477e-05,
-      "loss": 4.6342,
-      "step": 49743
-    },
-    {
-      "epoch": 25.94211212516297,
-      "grad_norm": 1.5849323272705078,
-      "learning_rate": 5.052663316582915e-05,
-      "loss": 5.3327,
-      "step": 49744
-    },
-    {
-      "epoch": 25.94263363754889,
-      "grad_norm": 1.6554206609725952,
-      "learning_rate": 5.0525628140703515e-05,
-      "loss": 5.2266,
-      "step": 49745
-    },
-    {
-      "epoch": 25.94315514993481,
-      "grad_norm": 1.5512250661849976,
-      "learning_rate": 5.052462311557789e-05,
-      "loss": 4.9699,
-      "step": 49746
-    },
-    {
-      "epoch": 25.94367666232073,
-      "grad_norm": 1.5667290687561035,
-      "learning_rate": 5.0523618090452264e-05,
-      "loss": 5.4505,
-      "step": 49747
-    },
-    {
-      "epoch": 25.94419817470665,
-      "grad_norm": 1.5546510219573975,
-      "learning_rate": 5.052261306532664e-05,
-      "loss": 5.2626,
-      "step": 49748
-    },
-    {
-      "epoch": 25.94471968709257,
-      "grad_norm": 1.4921232461929321,
-      "learning_rate": 5.0521608040201006e-05,
-      "loss": 4.6246,
-      "step": 49749
-    },
-    {
-      "epoch": 25.945241199478488,
-      "grad_norm": 1.4779304265975952,
-      "learning_rate": 5.052060301507537e-05,
-      "loss": 5.5724,
-      "step": 49750
-    },
-    {
-      "epoch": 25.945762711864408,
-      "grad_norm": 1.5938775539398193,
-      "learning_rate": 5.051959798994975e-05,
-      "loss": 5.4145,
-      "step": 49751
-    },
-    {
-      "epoch": 25.946284224250327,
-      "grad_norm": 1.446546196937561,
-      "learning_rate": 5.051859296482412e-05,
-      "loss": 5.5248,
-      "step": 49752
-    },
-    {
-      "epoch": 25.946805736636247,
-      "grad_norm": 1.6441422700881958,
-      "learning_rate": 5.05175879396985e-05,
-      "loss": 5.0558,
-      "step": 49753
-    },
-    {
-      "epoch": 25.947327249022166,
-      "grad_norm": 1.4974286556243896,
-      "learning_rate": 5.051658291457286e-05,
-      "loss": 5.5311,
-      "step": 49754
-    },
-    {
-      "epoch": 25.947848761408082,
-      "grad_norm": 1.5930880308151245,
-      "learning_rate": 5.051557788944724e-05,
-      "loss": 5.0789,
-      "step": 49755
-    },
-    {
-      "epoch": 25.948370273794,
-      "grad_norm": 1.7435616254806519,
-      "learning_rate": 5.051457286432161e-05,
-      "loss": 4.4324,
-      "step": 49756
-    },
-    {
-      "epoch": 25.94889178617992,
-      "grad_norm": 1.5853358507156372,
-      "learning_rate": 5.051356783919599e-05,
-      "loss": 5.3403,
-      "step": 49757
-    },
-    {
-      "epoch": 25.94941329856584,
-      "grad_norm": 1.5644216537475586,
-      "learning_rate": 5.051256281407035e-05,
-      "loss": 5.7999,
-      "step": 49758
-    },
-    {
-      "epoch": 25.94993481095176,
-      "grad_norm": 1.669603705406189,
-      "learning_rate": 5.051155778894473e-05,
-      "loss": 5.3256,
-      "step": 49759
-    },
-    {
-      "epoch": 25.95045632333768,
-      "grad_norm": 1.6117825508117676,
-      "learning_rate": 5.0510552763819095e-05,
-      "loss": 4.8755,
-      "step": 49760
-    },
-    {
-      "epoch": 25.9509778357236,
-      "grad_norm": 1.4320485591888428,
-      "learning_rate": 5.050954773869347e-05,
-      "loss": 4.9666,
-      "step": 49761
-    },
-    {
-      "epoch": 25.951499348109518,
-      "grad_norm": 1.5288912057876587,
-      "learning_rate": 5.0508542713567844e-05,
-      "loss": 5.5408,
-      "step": 49762
-    },
-    {
-      "epoch": 25.952020860495438,
-      "grad_norm": 1.6417725086212158,
-      "learning_rate": 5.050753768844221e-05,
-      "loss": 5.0686,
-      "step": 49763
-    },
-    {
-      "epoch": 25.952542372881357,
-      "grad_norm": 1.555238127708435,
-      "learning_rate": 5.0506532663316586e-05,
-      "loss": 5.5991,
-      "step": 49764
-    },
-    {
-      "epoch": 25.953063885267277,
-      "grad_norm": 1.4612008333206177,
-      "learning_rate": 5.050552763819095e-05,
-      "loss": 5.5105,
-      "step": 49765
-    },
-    {
-      "epoch": 25.953585397653193,
-      "grad_norm": 1.6306742429733276,
-      "learning_rate": 5.050452261306533e-05,
-      "loss": 5.3811,
-      "step": 49766
-    },
-    {
-      "epoch": 25.954106910039112,
-      "grad_norm": 1.57932710647583,
-      "learning_rate": 5.05035175879397e-05,
-      "loss": 5.5904,
-      "step": 49767
-    },
-    {
-      "epoch": 25.95462842242503,
-      "grad_norm": 1.5359058380126953,
-      "learning_rate": 5.050251256281408e-05,
-      "loss": 5.6282,
-      "step": 49768
-    },
-    {
-      "epoch": 25.95514993481095,
-      "grad_norm": 1.7532477378845215,
-      "learning_rate": 5.050150753768844e-05,
-      "loss": 5.0519,
-      "step": 49769
-    },
-    {
-      "epoch": 25.95567144719687,
-      "grad_norm": 1.6247059106826782,
-      "learning_rate": 5.050050251256282e-05,
-      "loss": 5.3184,
-      "step": 49770
-    },
-    {
-      "epoch": 25.95619295958279,
-      "grad_norm": 1.581208348274231,
-      "learning_rate": 5.049949748743718e-05,
-      "loss": 5.1828,
-      "step": 49771
-    },
-    {
-      "epoch": 25.95671447196871,
-      "grad_norm": 1.633225440979004,
-      "learning_rate": 5.049849246231156e-05,
-      "loss": 4.7107,
-      "step": 49772
-    },
-    {
-      "epoch": 25.95723598435463,
-      "grad_norm": 2.315793991088867,
-      "learning_rate": 5.049748743718593e-05,
-      "loss": 4.9035,
-      "step": 49773
-    },
-    {
-      "epoch": 25.957757496740548,
-      "grad_norm": 1.6242378950119019,
-      "learning_rate": 5.049648241206031e-05,
-      "loss": 5.0585,
-      "step": 49774
-    },
-    {
-      "epoch": 25.958279009126468,
-      "grad_norm": 1.5733084678649902,
-      "learning_rate": 5.0495477386934674e-05,
-      "loss": 5.1664,
-      "step": 49775
-    },
-    {
-      "epoch": 25.958800521512387,
-      "grad_norm": 1.532844066619873,
-      "learning_rate": 5.049447236180905e-05,
-      "loss": 4.9759,
-      "step": 49776
-    },
-    {
-      "epoch": 25.959322033898307,
-      "grad_norm": 1.6117173433303833,
-      "learning_rate": 5.049346733668342e-05,
-      "loss": 5.1307,
-      "step": 49777
-    },
-    {
-      "epoch": 25.959843546284226,
-      "grad_norm": 1.569705843925476,
-      "learning_rate": 5.049246231155779e-05,
-      "loss": 5.0656,
-      "step": 49778
-    },
-    {
-      "epoch": 25.960365058670142,
-      "grad_norm": 1.6132324934005737,
-      "learning_rate": 5.0491457286432165e-05,
-      "loss": 4.7158,
-      "step": 49779
-    },
-    {
-      "epoch": 25.96088657105606,
-      "grad_norm": 1.5446442365646362,
-      "learning_rate": 5.049045226130653e-05,
-      "loss": 4.8318,
-      "step": 49780
-    },
-    {
-      "epoch": 25.96140808344198,
-      "grad_norm": 1.5413283109664917,
-      "learning_rate": 5.048944723618091e-05,
-      "loss": 5.5106,
-      "step": 49781
-    },
-    {
-      "epoch": 25.9619295958279,
-      "grad_norm": 1.4867891073226929,
-      "learning_rate": 5.048844221105528e-05,
-      "loss": 5.4121,
-      "step": 49782
-    },
-    {
-      "epoch": 25.96245110821382,
-      "grad_norm": 1.549892544746399,
-      "learning_rate": 5.0487437185929656e-05,
-      "loss": 5.2531,
-      "step": 49783
-    },
-    {
-      "epoch": 25.96297262059974,
-      "grad_norm": 1.5599087476730347,
-      "learning_rate": 5.048643216080402e-05,
-      "loss": 4.9882,
-      "step": 49784
-    },
-    {
-      "epoch": 25.96349413298566,
-      "grad_norm": 1.617982029914856,
-      "learning_rate": 5.04854271356784e-05,
-      "loss": 5.0741,
-      "step": 49785
-    },
-    {
-      "epoch": 25.96401564537158,
-      "grad_norm": 1.493056058883667,
-      "learning_rate": 5.048442211055276e-05,
-      "loss": 5.3912,
-      "step": 49786
-    },
-    {
-      "epoch": 25.964537157757498,
-      "grad_norm": 1.6201637983322144,
-      "learning_rate": 5.048341708542714e-05,
-      "loss": 4.8395,
-      "step": 49787
-    },
-    {
-      "epoch": 25.965058670143417,
-      "grad_norm": 1.5247822999954224,
-      "learning_rate": 5.048241206030151e-05,
-      "loss": 5.6184,
-      "step": 49788
-    },
-    {
-      "epoch": 25.965580182529337,
-      "grad_norm": 1.662615180015564,
-      "learning_rate": 5.048140703517589e-05,
-      "loss": 4.8433,
-      "step": 49789
-    },
-    {
-      "epoch": 25.966101694915253,
-      "grad_norm": 1.5597120523452759,
-      "learning_rate": 5.0480402010050253e-05,
-      "loss": 4.9448,
-      "step": 49790
-    },
-    {
-      "epoch": 25.966623207301172,
-      "grad_norm": 1.6713917255401611,
-      "learning_rate": 5.047939698492462e-05,
-      "loss": 4.8546,
-      "step": 49791
-    },
-    {
-      "epoch": 25.96714471968709,
-      "grad_norm": 1.5485515594482422,
-      "learning_rate": 5.0478391959798996e-05,
-      "loss": 5.513,
-      "step": 49792
-    },
-    {
-      "epoch": 25.96766623207301,
-      "grad_norm": 1.492374300956726,
-      "learning_rate": 5.0477386934673367e-05,
-      "loss": 5.5201,
-      "step": 49793
-    },
-    {
-      "epoch": 25.96818774445893,
-      "grad_norm": 1.6300476789474487,
-      "learning_rate": 5.0476381909547744e-05,
-      "loss": 4.9398,
-      "step": 49794
-    },
-    {
-      "epoch": 25.96870925684485,
-      "grad_norm": 1.5074901580810547,
-      "learning_rate": 5.047537688442211e-05,
-      "loss": 5.2326,
-      "step": 49795
-    },
-    {
-      "epoch": 25.96923076923077,
-      "grad_norm": 1.5828238725662231,
-      "learning_rate": 5.0474371859296487e-05,
-      "loss": 5.1619,
-      "step": 49796
-    },
-    {
-      "epoch": 25.96975228161669,
-      "grad_norm": 1.511972427368164,
-      "learning_rate": 5.047336683417086e-05,
-      "loss": 5.2499,
-      "step": 49797
-    },
-    {
-      "epoch": 25.97027379400261,
-      "grad_norm": 1.630491852760315,
-      "learning_rate": 5.0472361809045235e-05,
-      "loss": 5.0824,
-      "step": 49798
-    },
-    {
-      "epoch": 25.970795306388528,
-      "grad_norm": 1.5411858558654785,
-      "learning_rate": 5.04713567839196e-05,
-      "loss": 5.3288,
-      "step": 49799
-    },
-    {
-      "epoch": 25.971316818774447,
-      "grad_norm": 1.5976699590682983,
-      "learning_rate": 5.047035175879398e-05,
-      "loss": 5.0371,
-      "step": 49800
-    },
-    {
-      "epoch": 25.971838331160367,
-      "grad_norm": 1.6064398288726807,
-      "learning_rate": 5.046934673366834e-05,
-      "loss": 5.2998,
-      "step": 49801
-    },
-    {
-      "epoch": 25.972359843546283,
-      "grad_norm": 1.6003201007843018,
-      "learning_rate": 5.046834170854272e-05,
-      "loss": 5.0896,
-      "step": 49802
-    },
-    {
-      "epoch": 25.972881355932202,
-      "grad_norm": 1.499881386756897,
-      "learning_rate": 5.046733668341709e-05,
-      "loss": 5.2199,
-      "step": 49803
-    },
-    {
-      "epoch": 25.97340286831812,
-      "grad_norm": 1.4750078916549683,
-      "learning_rate": 5.0466331658291455e-05,
-      "loss": 5.2687,
-      "step": 49804
-    },
-    {
-      "epoch": 25.97392438070404,
-      "grad_norm": 1.4993540048599243,
-      "learning_rate": 5.046532663316583e-05,
-      "loss": 4.7765,
-      "step": 49805
-    },
-    {
-      "epoch": 25.97444589308996,
-      "grad_norm": 1.5658129453659058,
-      "learning_rate": 5.04643216080402e-05,
-      "loss": 4.8905,
-      "step": 49806
-    },
-    {
-      "epoch": 25.97496740547588,
-      "grad_norm": 1.5606919527053833,
-      "learning_rate": 5.0463316582914575e-05,
-      "loss": 5.5872,
-      "step": 49807
-    },
-    {
-      "epoch": 25.9754889178618,
-      "grad_norm": 1.5068895816802979,
-      "learning_rate": 5.0462311557788946e-05,
-      "loss": 5.4762,
-      "step": 49808
-    },
-    {
-      "epoch": 25.97601043024772,
-      "grad_norm": 1.6147211790084839,
-      "learning_rate": 5.0461306532663324e-05,
-      "loss": 5.2261,
-      "step": 49809
-    },
-    {
-      "epoch": 25.97653194263364,
-      "grad_norm": 1.5323469638824463,
-      "learning_rate": 5.046030150753769e-05,
-      "loss": 4.9664,
-      "step": 49810
-    },
-    {
-      "epoch": 25.977053455019558,
-      "grad_norm": 1.5205397605895996,
-      "learning_rate": 5.0459296482412066e-05,
-      "loss": 5.7251,
-      "step": 49811
-    },
-    {
-      "epoch": 25.977574967405477,
-      "grad_norm": 1.521767497062683,
-      "learning_rate": 5.045829145728643e-05,
-      "loss": 5.3711,
-      "step": 49812
-    },
-    {
-      "epoch": 25.978096479791397,
-      "grad_norm": 1.5131927728652954,
-      "learning_rate": 5.045728643216081e-05,
-      "loss": 5.24,
-      "step": 49813
-    },
-    {
-      "epoch": 25.978617992177313,
-      "grad_norm": 1.6066757440567017,
-      "learning_rate": 5.045628140703518e-05,
-      "loss": 5.4673,
-      "step": 49814
-    },
-    {
-      "epoch": 25.979139504563232,
-      "grad_norm": 1.592763900756836,
-      "learning_rate": 5.045527638190956e-05,
-      "loss": 4.8468,
-      "step": 49815
-    },
-    {
-      "epoch": 25.97966101694915,
-      "grad_norm": 1.5876563787460327,
-      "learning_rate": 5.045427135678392e-05,
-      "loss": 5.2206,
-      "step": 49816
-    },
-    {
-      "epoch": 25.98018252933507,
-      "grad_norm": 1.4939838647842407,
-      "learning_rate": 5.0453266331658285e-05,
-      "loss": 5.3974,
-      "step": 49817
-    },
-    {
-      "epoch": 25.98070404172099,
-      "grad_norm": 1.644418716430664,
-      "learning_rate": 5.045226130653266e-05,
-      "loss": 4.5166,
-      "step": 49818
-    },
-    {
-      "epoch": 25.98122555410691,
-      "grad_norm": 1.58175790309906,
-      "learning_rate": 5.0451256281407034e-05,
-      "loss": 5.2525,
-      "step": 49819
-    },
-    {
-      "epoch": 25.98174706649283,
-      "grad_norm": 1.5747029781341553,
-      "learning_rate": 5.045025125628141e-05,
-      "loss": 5.6392,
-      "step": 49820
-    },
-    {
-      "epoch": 25.98226857887875,
-      "grad_norm": 1.5908254384994507,
-      "learning_rate": 5.0449246231155776e-05,
-      "loss": 5.4578,
-      "step": 49821
-    },
-    {
-      "epoch": 25.98279009126467,
-      "grad_norm": 1.5626230239868164,
-      "learning_rate": 5.0448241206030154e-05,
-      "loss": 4.9507,
-      "step": 49822
-    },
-    {
-      "epoch": 25.983311603650588,
-      "grad_norm": 1.56340491771698,
-      "learning_rate": 5.0447236180904525e-05,
-      "loss": 5.2032,
-      "step": 49823
-    },
-    {
-      "epoch": 25.983833116036507,
-      "grad_norm": 1.4311556816101074,
-      "learning_rate": 5.04462311557789e-05,
-      "loss": 5.4195,
-      "step": 49824
-    },
-    {
-      "epoch": 25.984354628422427,
-      "grad_norm": 1.5117915868759155,
-      "learning_rate": 5.044522613065327e-05,
-      "loss": 5.13,
-      "step": 49825
-    },
-    {
-      "epoch": 25.984876140808343,
-      "grad_norm": 1.454552173614502,
-      "learning_rate": 5.0444221105527645e-05,
-      "loss": 5.6917,
-      "step": 49826
-    },
-    {
-      "epoch": 25.985397653194262,
-      "grad_norm": 1.4178123474121094,
-      "learning_rate": 5.044321608040201e-05,
-      "loss": 5.3776,
-      "step": 49827
-    },
-    {
-      "epoch": 25.98591916558018,
-      "grad_norm": 1.4790244102478027,
-      "learning_rate": 5.044221105527639e-05,
-      "loss": 5.2721,
-      "step": 49828
-    },
-    {
-      "epoch": 25.9864406779661,
-      "grad_norm": 1.6627033948898315,
-      "learning_rate": 5.044120603015076e-05,
-      "loss": 5.0514,
-      "step": 49829
-    },
-    {
-      "epoch": 25.98696219035202,
-      "grad_norm": 1.549363374710083,
-      "learning_rate": 5.044020100502512e-05,
-      "loss": 5.1919,
-      "step": 49830
-    },
-    {
-      "epoch": 25.98748370273794,
-      "grad_norm": 1.6241368055343628,
-      "learning_rate": 5.04391959798995e-05,
-      "loss": 4.5678,
-      "step": 49831
-    },
-    {
-      "epoch": 25.98800521512386,
-      "grad_norm": 1.5540589094161987,
-      "learning_rate": 5.0438190954773865e-05,
-      "loss": 4.867,
-      "step": 49832
-    },
-    {
-      "epoch": 25.98852672750978,
-      "grad_norm": 1.490044116973877,
-      "learning_rate": 5.043718592964824e-05,
-      "loss": 5.0171,
-      "step": 49833
-    },
-    {
-      "epoch": 25.9890482398957,
-      "grad_norm": 1.5490013360977173,
-      "learning_rate": 5.0436180904522614e-05,
-      "loss": 5.3056,
-      "step": 49834
-    },
-    {
-      "epoch": 25.989569752281618,
-      "grad_norm": 1.5932077169418335,
-      "learning_rate": 5.043517587939699e-05,
-      "loss": 4.9279,
-      "step": 49835
-    },
-    {
-      "epoch": 25.990091264667537,
-      "grad_norm": 1.5124351978302002,
-      "learning_rate": 5.0434170854271356e-05,
-      "loss": 5.3837,
-      "step": 49836
-    },
-    {
-      "epoch": 25.990612777053457,
-      "grad_norm": 1.550491213798523,
-      "learning_rate": 5.0433165829145734e-05,
-      "loss": 5.5344,
-      "step": 49837
-    },
-    {
-      "epoch": 25.991134289439373,
-      "grad_norm": 1.5916805267333984,
-      "learning_rate": 5.04321608040201e-05,
-      "loss": 5.3522,
-      "step": 49838
-    },
-    {
-      "epoch": 25.991655801825292,
-      "grad_norm": 1.6321598291397095,
-      "learning_rate": 5.0431155778894476e-05,
-      "loss": 5.6047,
-      "step": 49839
-    },
-    {
-      "epoch": 25.99217731421121,
-      "grad_norm": 1.5555574893951416,
-      "learning_rate": 5.043015075376885e-05,
-      "loss": 5.3387,
-      "step": 49840
-    },
-    {
-      "epoch": 25.99269882659713,
-      "grad_norm": 1.5286091566085815,
-      "learning_rate": 5.0429145728643225e-05,
-      "loss": 5.3712,
-      "step": 49841
-    },
-    {
-      "epoch": 25.99322033898305,
-      "grad_norm": 1.5893055200576782,
-      "learning_rate": 5.042814070351759e-05,
-      "loss": 4.9381,
-      "step": 49842
-    },
-    {
-      "epoch": 25.99374185136897,
-      "grad_norm": 1.4439986944198608,
-      "learning_rate": 5.042713567839196e-05,
-      "loss": 4.7925,
-      "step": 49843
-    },
-    {
-      "epoch": 25.99426336375489,
-      "grad_norm": 1.6688432693481445,
-      "learning_rate": 5.042613065326634e-05,
-      "loss": 4.8472,
-      "step": 49844
-    },
-    {
-      "epoch": 25.99478487614081,
-      "grad_norm": 1.6317939758300781,
-      "learning_rate": 5.04251256281407e-05,
-      "loss": 5.2285,
-      "step": 49845
-    },
-    {
-      "epoch": 25.99530638852673,
-      "grad_norm": 1.5751293897628784,
-      "learning_rate": 5.042412060301508e-05,
-      "loss": 5.4734,
-      "step": 49846
-    },
-    {
-      "epoch": 25.995827900912648,
-      "grad_norm": 1.6407735347747803,
-      "learning_rate": 5.0423115577889444e-05,
-      "loss": 5.2709,
-      "step": 49847
-    },
-    {
-      "epoch": 25.996349413298567,
-      "grad_norm": 1.6169683933258057,
-      "learning_rate": 5.042211055276382e-05,
-      "loss": 4.7377,
-      "step": 49848
-    },
-    {
-      "epoch": 25.996870925684483,
-      "grad_norm": 1.5979669094085693,
-      "learning_rate": 5.042110552763819e-05,
-      "loss": 4.9615,
-      "step": 49849
-    },
-    {
-      "epoch": 25.997392438070403,
-      "grad_norm": 1.5826510190963745,
-      "learning_rate": 5.042010050251257e-05,
-      "loss": 5.1735,
-      "step": 49850
-    },
-    {
-      "epoch": 25.997913950456322,
-      "grad_norm": 1.5629518032073975,
-      "learning_rate": 5.0419095477386935e-05,
-      "loss": 4.8396,
-      "step": 49851
-    },
-    {
-      "epoch": 25.99843546284224,
-      "grad_norm": 1.5789201259613037,
-      "learning_rate": 5.041809045226131e-05,
-      "loss": 4.9036,
-      "step": 49852
-    },
-    {
-      "epoch": 25.99895697522816,
-      "grad_norm": 1.6691685914993286,
-      "learning_rate": 5.041708542713568e-05,
-      "loss": 4.7955,
-      "step": 49853
-    },
-    {
-      "epoch": 25.99947848761408,
-      "grad_norm": 1.629495620727539,
-      "learning_rate": 5.0416080402010055e-05,
-      "loss": 5.1245,
-      "step": 49854
-    },
-    {
-      "epoch": 26.0,
-      "grad_norm": 1.7617546319961548,
-      "learning_rate": 5.0415075376884426e-05,
-      "loss": 5.7405,
-      "step": 49855
-    },
-    {
-      "epoch": 26.00052151238592,
-      "grad_norm": 1.630406141281128,
-      "learning_rate": 5.041407035175879e-05,
-      "loss": 5.0202,
-      "step": 49856
-    },
-    {
-      "epoch": 26.00104302477184,
-      "grad_norm": 1.5694475173950195,
-      "learning_rate": 5.041306532663317e-05,
-      "loss": 5.447,
-      "step": 49857
-    },
-    {
-      "epoch": 26.00156453715776,
-      "grad_norm": 1.4947025775909424,
-      "learning_rate": 5.041206030150753e-05,
-      "loss": 5.4293,
-      "step": 49858
-    },
-    {
-      "epoch": 26.002086049543678,
-      "grad_norm": 1.593772530555725,
-      "learning_rate": 5.041105527638191e-05,
-      "loss": 5.2391,
-      "step": 49859
-    },
-    {
-      "epoch": 26.002607561929597,
-      "grad_norm": 1.6247615814208984,
-      "learning_rate": 5.041005025125628e-05,
-      "loss": 5.3606,
-      "step": 49860
-    },
-    {
-      "epoch": 26.003129074315513,
-      "grad_norm": 1.6541873216629028,
-      "learning_rate": 5.040904522613066e-05,
-      "loss": 5.3258,
-      "step": 49861
-    },
-    {
-      "epoch": 26.003650586701433,
-      "grad_norm": 1.590934157371521,
-      "learning_rate": 5.040804020100502e-05,
-      "loss": 5.338,
-      "step": 49862
-    },
-    {
-      "epoch": 26.004172099087352,
-      "grad_norm": 1.5482243299484253,
-      "learning_rate": 5.04070351758794e-05,
-      "loss": 5.0853,
-      "step": 49863
-    },
-    {
-      "epoch": 26.00469361147327,
-      "grad_norm": 1.5913010835647583,
-      "learning_rate": 5.040603015075377e-05,
-      "loss": 5.1561,
-      "step": 49864
-    },
-    {
-      "epoch": 26.00521512385919,
-      "grad_norm": 1.5768563747406006,
-      "learning_rate": 5.040502512562815e-05,
-      "loss": 5.3283,
-      "step": 49865
-    },
-    {
-      "epoch": 26.00573663624511,
-      "grad_norm": 1.5579520463943481,
-      "learning_rate": 5.0404020100502514e-05,
-      "loss": 5.5463,
-      "step": 49866
-    },
-    {
-      "epoch": 26.00625814863103,
-      "grad_norm": 1.5443542003631592,
-      "learning_rate": 5.040301507537689e-05,
-      "loss": 5.646,
-      "step": 49867
-    },
-    {
-      "epoch": 26.00677966101695,
-      "grad_norm": 1.5288281440734863,
-      "learning_rate": 5.0402010050251256e-05,
-      "loss": 5.4475,
-      "step": 49868
-    },
-    {
-      "epoch": 26.00730117340287,
-      "grad_norm": 1.6905237436294556,
-      "learning_rate": 5.0401005025125634e-05,
-      "loss": 4.9491,
-      "step": 49869
-    },
-    {
-      "epoch": 26.00782268578879,
-      "grad_norm": 1.602318525314331,
-      "learning_rate": 5.0400000000000005e-05,
-      "loss": 5.0166,
-      "step": 49870
-    },
-    {
-      "epoch": 26.008344198174708,
-      "grad_norm": 1.5511987209320068,
-      "learning_rate": 5.039899497487437e-05,
-      "loss": 5.4206,
-      "step": 49871
-    },
-    {
-      "epoch": 26.008865710560627,
-      "grad_norm": 1.6969717741012573,
-      "learning_rate": 5.039798994974875e-05,
-      "loss": 5.086,
-      "step": 49872
-    },
-    {
-      "epoch": 26.009387222946543,
-      "grad_norm": 1.6044172048568726,
-      "learning_rate": 5.039698492462311e-05,
-      "loss": 5.1749,
-      "step": 49873
-    },
-    {
-      "epoch": 26.009908735332463,
-      "grad_norm": 1.5165537595748901,
-      "learning_rate": 5.039597989949749e-05,
-      "loss": 5.0529,
-      "step": 49874
-    },
-    {
-      "epoch": 26.010430247718382,
-      "grad_norm": 1.5432796478271484,
-      "learning_rate": 5.039497487437186e-05,
-      "loss": 5.6221,
-      "step": 49875
-    },
-    {
-      "epoch": 26.0109517601043,
-      "grad_norm": 1.7155026197433472,
-      "learning_rate": 5.039396984924624e-05,
-      "loss": 4.8293,
-      "step": 49876
-    },
-    {
-      "epoch": 26.01147327249022,
-      "grad_norm": 1.5145021677017212,
-      "learning_rate": 5.03929648241206e-05,
-      "loss": 5.574,
-      "step": 49877
-    },
-    {
-      "epoch": 26.01199478487614,
-      "grad_norm": 1.5249179601669312,
-      "learning_rate": 5.039195979899498e-05,
-      "loss": 5.472,
-      "step": 49878
-    },
-    {
-      "epoch": 26.01251629726206,
-      "grad_norm": 1.6135358810424805,
-      "learning_rate": 5.0390954773869345e-05,
-      "loss": 5.0177,
-      "step": 49879
-    },
-    {
-      "epoch": 26.01303780964798,
-      "grad_norm": 1.4863982200622559,
-      "learning_rate": 5.038994974874372e-05,
-      "loss": 5.1835,
-      "step": 49880
-    },
-    {
-      "epoch": 26.0135593220339,
-      "grad_norm": 1.5394470691680908,
-      "learning_rate": 5.0388944723618094e-05,
-      "loss": 5.284,
-      "step": 49881
-    },
-    {
-      "epoch": 26.01408083441982,
-      "grad_norm": 1.5694096088409424,
-      "learning_rate": 5.038793969849247e-05,
-      "loss": 5.092,
-      "step": 49882
-    },
-    {
-      "epoch": 26.014602346805738,
-      "grad_norm": 1.5273618698120117,
-      "learning_rate": 5.0386934673366836e-05,
-      "loss": 5.4781,
-      "step": 49883
-    },
-    {
-      "epoch": 26.015123859191657,
-      "grad_norm": 1.5585932731628418,
-      "learning_rate": 5.038592964824121e-05,
-      "loss": 5.2642,
-      "step": 49884
-    },
-    {
-      "epoch": 26.015645371577573,
-      "grad_norm": 1.6852264404296875,
-      "learning_rate": 5.0384924623115585e-05,
-      "loss": 5.2391,
-      "step": 49885
-    },
-    {
-      "epoch": 26.016166883963493,
-      "grad_norm": 1.617247223854065,
-      "learning_rate": 5.038391959798995e-05,
-      "loss": 4.9123,
-      "step": 49886
-    },
-    {
-      "epoch": 26.016688396349412,
-      "grad_norm": 1.5400272607803345,
-      "learning_rate": 5.038291457286433e-05,
-      "loss": 4.8747,
-      "step": 49887
-    },
-    {
-      "epoch": 26.01720990873533,
-      "grad_norm": 1.579487919807434,
-      "learning_rate": 5.038190954773869e-05,
-      "loss": 5.4667,
-      "step": 49888
-    },
-    {
-      "epoch": 26.01773142112125,
-      "grad_norm": 1.7204338312149048,
-      "learning_rate": 5.038090452261307e-05,
-      "loss": 4.4433,
-      "step": 49889
-    },
-    {
-      "epoch": 26.01825293350717,
-      "grad_norm": 1.4713044166564941,
-      "learning_rate": 5.037989949748744e-05,
-      "loss": 4.8028,
-      "step": 49890
-    },
-    {
-      "epoch": 26.01877444589309,
-      "grad_norm": 1.571448564529419,
-      "learning_rate": 5.037889447236182e-05,
-      "loss": 5.4677,
-      "step": 49891
-    },
-    {
-      "epoch": 26.01929595827901,
-      "grad_norm": 1.6490477323532104,
-      "learning_rate": 5.037788944723618e-05,
-      "loss": 5.4673,
-      "step": 49892
-    },
-    {
-      "epoch": 26.01981747066493,
-      "grad_norm": 1.478212594985962,
-      "learning_rate": 5.037688442211056e-05,
-      "loss": 5.3408,
-      "step": 49893
-    },
-    {
-      "epoch": 26.02033898305085,
-      "grad_norm": 1.6048938035964966,
-      "learning_rate": 5.0375879396984924e-05,
-      "loss": 5.084,
-      "step": 49894
-    },
-    {
-      "epoch": 26.020860495436768,
-      "grad_norm": 1.617571234703064,
-      "learning_rate": 5.03748743718593e-05,
-      "loss": 4.6362,
-      "step": 49895
-    },
-    {
-      "epoch": 26.021382007822687,
-      "grad_norm": 1.7693649530410767,
-      "learning_rate": 5.037386934673367e-05,
-      "loss": 4.2675,
-      "step": 49896
-    },
-    {
-      "epoch": 26.021903520208603,
-      "grad_norm": 1.514443278312683,
-      "learning_rate": 5.037286432160804e-05,
-      "loss": 5.3821,
-      "step": 49897
-    },
-    {
-      "epoch": 26.022425032594523,
-      "grad_norm": 1.5993214845657349,
-      "learning_rate": 5.0371859296482415e-05,
-      "loss": 4.6702,
-      "step": 49898
-    },
-    {
-      "epoch": 26.022946544980442,
-      "grad_norm": 1.5937093496322632,
-      "learning_rate": 5.037085427135678e-05,
-      "loss": 4.6762,
-      "step": 49899
-    },
-    {
-      "epoch": 26.02346805736636,
-      "grad_norm": 1.6407947540283203,
-      "learning_rate": 5.036984924623116e-05,
-      "loss": 4.9777,
-      "step": 49900
-    },
-    {
-      "epoch": 26.02398956975228,
-      "grad_norm": 1.5661416053771973,
-      "learning_rate": 5.036884422110553e-05,
-      "loss": 4.5352,
-      "step": 49901
-    },
-    {
-      "epoch": 26.0245110821382,
-      "grad_norm": 1.5704617500305176,
-      "learning_rate": 5.0367839195979906e-05,
-      "loss": 5.1742,
-      "step": 49902
-    },
-    {
-      "epoch": 26.02503259452412,
-      "grad_norm": 1.611840844154358,
-      "learning_rate": 5.036683417085427e-05,
-      "loss": 4.9322,
-      "step": 49903
-    },
-    {
-      "epoch": 26.02555410691004,
-      "grad_norm": 1.6492494344711304,
-      "learning_rate": 5.036582914572865e-05,
-      "loss": 5.1859,
-      "step": 49904
-    },
-    {
-      "epoch": 26.02607561929596,
-      "grad_norm": 1.602911114692688,
-      "learning_rate": 5.036482412060301e-05,
-      "loss": 4.9235,
-      "step": 49905
-    },
-    {
-      "epoch": 26.02659713168188,
-      "grad_norm": 1.6575928926467896,
-      "learning_rate": 5.036381909547739e-05,
-      "loss": 4.5271,
-      "step": 49906
-    },
-    {
-      "epoch": 26.027118644067798,
-      "grad_norm": 1.7428065538406372,
-      "learning_rate": 5.036281407035176e-05,
-      "loss": 4.9101,
-      "step": 49907
-    },
-    {
-      "epoch": 26.027640156453717,
-      "grad_norm": 1.5835801362991333,
-      "learning_rate": 5.036180904522614e-05,
-      "loss": 5.0906,
-      "step": 49908
-    },
-    {
-      "epoch": 26.028161668839633,
-      "grad_norm": 1.53526771068573,
-      "learning_rate": 5.0360804020100503e-05,
-      "loss": 5.176,
-      "step": 49909
-    },
-    {
-      "epoch": 26.028683181225553,
-      "grad_norm": 1.3856509923934937,
-      "learning_rate": 5.0359798994974875e-05,
-      "loss": 5.5657,
-      "step": 49910
-    },
-    {
-      "epoch": 26.029204693611472,
-      "grad_norm": 1.554999828338623,
-      "learning_rate": 5.035879396984925e-05,
-      "loss": 5.3563,
-      "step": 49911
-    },
-    {
-      "epoch": 26.02972620599739,
-      "grad_norm": 1.6727173328399658,
-      "learning_rate": 5.035778894472362e-05,
-      "loss": 4.6031,
-      "step": 49912
-    },
-    {
-      "epoch": 26.03024771838331,
-      "grad_norm": 1.5232245922088623,
-      "learning_rate": 5.0356783919597994e-05,
-      "loss": 5.3846,
-      "step": 49913
-    },
-    {
-      "epoch": 26.03076923076923,
-      "grad_norm": 1.6969258785247803,
-      "learning_rate": 5.035577889447236e-05,
-      "loss": 5.0519,
-      "step": 49914
-    },
-    {
-      "epoch": 26.03129074315515,
-      "grad_norm": 1.5561836957931519,
-      "learning_rate": 5.0354773869346737e-05,
-      "loss": 5.3353,
-      "step": 49915
-    },
-    {
-      "epoch": 26.03181225554107,
-      "grad_norm": 1.5368386507034302,
-      "learning_rate": 5.035376884422111e-05,
-      "loss": 5.6348,
-      "step": 49916
-    },
-    {
-      "epoch": 26.03233376792699,
-      "grad_norm": 1.5277912616729736,
-      "learning_rate": 5.0352763819095485e-05,
-      "loss": 5.4047,
-      "step": 49917
-    },
-    {
-      "epoch": 26.03285528031291,
-      "grad_norm": 1.5470176935195923,
-      "learning_rate": 5.035175879396985e-05,
-      "loss": 5.4136,
-      "step": 49918
-    },
-    {
-      "epoch": 26.033376792698828,
-      "grad_norm": 1.5946000814437866,
-      "learning_rate": 5.035075376884423e-05,
-      "loss": 5.3486,
-      "step": 49919
-    },
-    {
-      "epoch": 26.033898305084747,
-      "grad_norm": 1.4964848756790161,
-      "learning_rate": 5.034974874371859e-05,
-      "loss": 5.0878,
-      "step": 49920
-    },
-    {
-      "epoch": 26.034419817470663,
-      "grad_norm": 1.5625120401382446,
-      "learning_rate": 5.034874371859297e-05,
-      "loss": 5.255,
-      "step": 49921
-    },
-    {
-      "epoch": 26.034941329856583,
-      "grad_norm": 1.6137350797653198,
-      "learning_rate": 5.034773869346734e-05,
-      "loss": 5.1863,
-      "step": 49922
-    },
-    {
-      "epoch": 26.035462842242502,
-      "grad_norm": 1.4832732677459717,
-      "learning_rate": 5.0346733668341705e-05,
-      "loss": 5.6676,
-      "step": 49923
-    },
-    {
-      "epoch": 26.03598435462842,
-      "grad_norm": 1.523400902748108,
-      "learning_rate": 5.034572864321608e-05,
-      "loss": 5.3586,
-      "step": 49924
-    },
-    {
-      "epoch": 26.03650586701434,
-      "grad_norm": 1.5644515752792358,
-      "learning_rate": 5.034472361809045e-05,
-      "loss": 5.2654,
-      "step": 49925
-    },
-    {
-      "epoch": 26.03702737940026,
-      "grad_norm": 1.4728585481643677,
-      "learning_rate": 5.0343718592964825e-05,
-      "loss": 5.6056,
-      "step": 49926
-    },
-    {
-      "epoch": 26.03754889178618,
-      "grad_norm": 1.6028105020523071,
-      "learning_rate": 5.0342713567839196e-05,
-      "loss": 4.6443,
-      "step": 49927
-    },
-    {
-      "epoch": 26.0380704041721,
-      "grad_norm": 1.5580484867095947,
-      "learning_rate": 5.0341708542713574e-05,
-      "loss": 5.1627,
-      "step": 49928
-    },
-    {
-      "epoch": 26.03859191655802,
-      "grad_norm": 1.5718873739242554,
-      "learning_rate": 5.034070351758794e-05,
-      "loss": 5.3933,
-      "step": 49929
-    },
-    {
-      "epoch": 26.03911342894394,
-      "grad_norm": 1.5662686824798584,
-      "learning_rate": 5.0339698492462316e-05,
-      "loss": 5.1158,
-      "step": 49930
-    },
-    {
-      "epoch": 26.039634941329858,
-      "grad_norm": 1.5889496803283691,
-      "learning_rate": 5.033869346733669e-05,
-      "loss": 4.9535,
-      "step": 49931
-    },
-    {
-      "epoch": 26.040156453715777,
-      "grad_norm": 1.4810431003570557,
-      "learning_rate": 5.0337688442211065e-05,
-      "loss": 5.682,
-      "step": 49932
-    },
-    {
-      "epoch": 26.040677966101693,
-      "grad_norm": 1.5462135076522827,
-      "learning_rate": 5.033668341708543e-05,
-      "loss": 4.6778,
-      "step": 49933
-    },
-    {
-      "epoch": 26.041199478487613,
-      "grad_norm": 1.5868403911590576,
-      "learning_rate": 5.033567839195981e-05,
-      "loss": 4.976,
-      "step": 49934
-    },
-    {
-      "epoch": 26.041720990873532,
-      "grad_norm": 1.655432939529419,
-      "learning_rate": 5.033467336683417e-05,
-      "loss": 4.4422,
-      "step": 49935
-    },
-    {
-      "epoch": 26.042242503259452,
-      "grad_norm": 1.539458990097046,
-      "learning_rate": 5.033366834170854e-05,
-      "loss": 5.0,
-      "step": 49936
-    },
-    {
-      "epoch": 26.04276401564537,
-      "grad_norm": 1.5608280897140503,
-      "learning_rate": 5.033266331658292e-05,
-      "loss": 5.3505,
-      "step": 49937
-    },
-    {
-      "epoch": 26.04328552803129,
-      "grad_norm": 1.8141158819198608,
-      "learning_rate": 5.0331658291457284e-05,
-      "loss": 4.8851,
-      "step": 49938
-    },
-    {
-      "epoch": 26.04380704041721,
-      "grad_norm": 1.4827202558517456,
-      "learning_rate": 5.033065326633166e-05,
-      "loss": 5.0209,
-      "step": 49939
-    },
-    {
-      "epoch": 26.04432855280313,
-      "grad_norm": 1.5417380332946777,
-      "learning_rate": 5.0329648241206026e-05,
-      "loss": 5.4906,
-      "step": 49940
-    },
-    {
-      "epoch": 26.04485006518905,
-      "grad_norm": 1.6341389417648315,
-      "learning_rate": 5.0328643216080404e-05,
-      "loss": 5.3053,
-      "step": 49941
-    },
-    {
-      "epoch": 26.04537157757497,
-      "grad_norm": 1.5426335334777832,
-      "learning_rate": 5.0327638190954775e-05,
-      "loss": 5.4556,
-      "step": 49942
-    },
-    {
-      "epoch": 26.045893089960888,
-      "grad_norm": 1.5502915382385254,
-      "learning_rate": 5.032663316582915e-05,
-      "loss": 4.8803,
-      "step": 49943
-    },
-    {
-      "epoch": 26.046414602346807,
-      "grad_norm": 1.6502996683120728,
-      "learning_rate": 5.032562814070352e-05,
-      "loss": 4.9702,
-      "step": 49944
-    },
-    {
-      "epoch": 26.046936114732723,
-      "grad_norm": 1.5974037647247314,
-      "learning_rate": 5.0324623115577895e-05,
-      "loss": 5.3433,
-      "step": 49945
-    },
-    {
-      "epoch": 26.047457627118643,
-      "grad_norm": 1.6698493957519531,
-      "learning_rate": 5.032361809045226e-05,
-      "loss": 4.9863,
-      "step": 49946
-    },
-    {
-      "epoch": 26.047979139504562,
-      "grad_norm": 1.6427046060562134,
-      "learning_rate": 5.032261306532664e-05,
-      "loss": 5.4131,
-      "step": 49947
-    },
-    {
-      "epoch": 26.048500651890482,
-      "grad_norm": 1.656296968460083,
-      "learning_rate": 5.032160804020101e-05,
-      "loss": 5.1012,
-      "step": 49948
-    },
-    {
-      "epoch": 26.0490221642764,
-      "grad_norm": 1.7394770383834839,
-      "learning_rate": 5.032060301507537e-05,
-      "loss": 4.7086,
-      "step": 49949
-    },
-    {
-      "epoch": 26.04954367666232,
-      "grad_norm": 1.6759474277496338,
-      "learning_rate": 5.031959798994975e-05,
-      "loss": 5.0219,
-      "step": 49950
-    },
-    {
-      "epoch": 26.05006518904824,
-      "grad_norm": 1.6022017002105713,
-      "learning_rate": 5.031859296482412e-05,
-      "loss": 4.9726,
-      "step": 49951
-    },
-    {
-      "epoch": 26.05058670143416,
-      "grad_norm": 1.5920095443725586,
-      "learning_rate": 5.03175879396985e-05,
-      "loss": 5.0485,
-      "step": 49952
-    },
-    {
-      "epoch": 26.05110821382008,
-      "grad_norm": 1.5178791284561157,
-      "learning_rate": 5.0316582914572864e-05,
-      "loss": 5.3107,
-      "step": 49953
-    },
-    {
-      "epoch": 26.051629726206,
-      "grad_norm": 1.543992519378662,
-      "learning_rate": 5.031557788944724e-05,
-      "loss": 5.4051,
-      "step": 49954
-    },
-    {
-      "epoch": 26.052151238591918,
-      "grad_norm": 1.6477265357971191,
-      "learning_rate": 5.0314572864321606e-05,
-      "loss": 5.0291,
-      "step": 49955
-    },
-    {
-      "epoch": 26.052672750977834,
-      "grad_norm": 1.6444275379180908,
-      "learning_rate": 5.0313567839195984e-05,
-      "loss": 4.9704,
-      "step": 49956
-    },
-    {
-      "epoch": 26.053194263363753,
-      "grad_norm": 1.5968772172927856,
-      "learning_rate": 5.0312562814070355e-05,
-      "loss": 5.5423,
-      "step": 49957
-    },
-    {
-      "epoch": 26.053715775749673,
-      "grad_norm": 1.5367923974990845,
-      "learning_rate": 5.031155778894473e-05,
-      "loss": 4.5087,
-      "step": 49958
-    },
-    {
-      "epoch": 26.054237288135592,
-      "grad_norm": 1.595871925354004,
-      "learning_rate": 5.03105527638191e-05,
-      "loss": 5.184,
-      "step": 49959
-    },
-    {
-      "epoch": 26.054758800521512,
-      "grad_norm": 1.5525952577590942,
-      "learning_rate": 5.0309547738693475e-05,
-      "loss": 5.2862,
-      "step": 49960
-    },
-    {
-      "epoch": 26.05528031290743,
-      "grad_norm": 1.664811372756958,
-      "learning_rate": 5.030854271356784e-05,
-      "loss": 4.5795,
-      "step": 49961
-    },
-    {
-      "epoch": 26.05580182529335,
-      "grad_norm": 1.6122732162475586,
-      "learning_rate": 5.030753768844222e-05,
-      "loss": 5.1343,
-      "step": 49962
-    },
-    {
-      "epoch": 26.05632333767927,
-      "grad_norm": 1.5548722743988037,
-      "learning_rate": 5.030653266331659e-05,
-      "loss": 5.494,
-      "step": 49963
-    },
-    {
-      "epoch": 26.05684485006519,
-      "grad_norm": 1.527498722076416,
-      "learning_rate": 5.030552763819095e-05,
-      "loss": 5.5403,
-      "step": 49964
-    },
-    {
-      "epoch": 26.05736636245111,
-      "grad_norm": 1.6688705682754517,
-      "learning_rate": 5.030452261306533e-05,
-      "loss": 5.0659,
-      "step": 49965
-    },
-    {
-      "epoch": 26.05788787483703,
-      "grad_norm": 1.6740771532058716,
-      "learning_rate": 5.0303517587939694e-05,
-      "loss": 4.4977,
-      "step": 49966
-    },
-    {
-      "epoch": 26.058409387222948,
-      "grad_norm": 1.6127047538757324,
-      "learning_rate": 5.030251256281407e-05,
-      "loss": 5.3266,
-      "step": 49967
-    },
-    {
-      "epoch": 26.058930899608864,
-      "grad_norm": 1.5834952592849731,
-      "learning_rate": 5.030150753768844e-05,
-      "loss": 5.0688,
-      "step": 49968
-    },
-    {
-      "epoch": 26.059452411994783,
-      "grad_norm": 1.5595929622650146,
-      "learning_rate": 5.030050251256282e-05,
-      "loss": 5.6848,
-      "step": 49969
-    },
-    {
-      "epoch": 26.059973924380703,
-      "grad_norm": 1.527309775352478,
-      "learning_rate": 5.0299497487437185e-05,
-      "loss": 5.0097,
-      "step": 49970
-    },
-    {
-      "epoch": 26.060495436766622,
-      "grad_norm": 1.6456427574157715,
-      "learning_rate": 5.029849246231156e-05,
-      "loss": 4.9512,
-      "step": 49971
-    },
-    {
-      "epoch": 26.061016949152542,
-      "grad_norm": 1.631182312965393,
-      "learning_rate": 5.0297487437185934e-05,
-      "loss": 5.2295,
-      "step": 49972
-    },
-    {
-      "epoch": 26.06153846153846,
-      "grad_norm": 1.5534121990203857,
-      "learning_rate": 5.029648241206031e-05,
-      "loss": 5.4364,
-      "step": 49973
-    },
-    {
-      "epoch": 26.06205997392438,
-      "grad_norm": 1.657080888748169,
-      "learning_rate": 5.0295477386934676e-05,
-      "loss": 4.8284,
-      "step": 49974
-    },
-    {
-      "epoch": 26.0625814863103,
-      "grad_norm": 1.6267520189285278,
-      "learning_rate": 5.0294472361809054e-05,
-      "loss": 4.9535,
-      "step": 49975
-    },
-    {
-      "epoch": 26.06310299869622,
-      "grad_norm": 1.4828643798828125,
-      "learning_rate": 5.029346733668342e-05,
-      "loss": 5.6494,
-      "step": 49976
-    },
-    {
-      "epoch": 26.06362451108214,
-      "grad_norm": 1.5096355676651,
-      "learning_rate": 5.029246231155779e-05,
-      "loss": 5.5209,
-      "step": 49977
-    },
-    {
-      "epoch": 26.06414602346806,
-      "grad_norm": 1.4697860479354858,
-      "learning_rate": 5.029145728643217e-05,
-      "loss": 5.5019,
-      "step": 49978
-    },
-    {
-      "epoch": 26.064667535853978,
-      "grad_norm": 1.6198960542678833,
-      "learning_rate": 5.029045226130653e-05,
-      "loss": 5.5207,
-      "step": 49979
-    },
-    {
-      "epoch": 26.065189048239894,
-      "grad_norm": 1.6579374074935913,
-      "learning_rate": 5.028944723618091e-05,
-      "loss": 5.009,
-      "step": 49980
-    },
-    {
-      "epoch": 26.065710560625813,
-      "grad_norm": 1.5776405334472656,
-      "learning_rate": 5.0288442211055273e-05,
-      "loss": 5.0292,
-      "step": 49981
-    },
-    {
-      "epoch": 26.066232073011733,
-      "grad_norm": 1.5483795404434204,
-      "learning_rate": 5.028743718592965e-05,
-      "loss": 5.5258,
-      "step": 49982
-    },
-    {
-      "epoch": 26.066753585397652,
-      "grad_norm": 1.6215063333511353,
-      "learning_rate": 5.028643216080402e-05,
-      "loss": 4.8534,
-      "step": 49983
-    },
-    {
-      "epoch": 26.067275097783572,
-      "grad_norm": 1.5702906847000122,
-      "learning_rate": 5.02854271356784e-05,
-      "loss": 5.2473,
-      "step": 49984
-    },
-    {
-      "epoch": 26.06779661016949,
-      "grad_norm": 1.6264228820800781,
-      "learning_rate": 5.0284422110552764e-05,
-      "loss": 4.7399,
-      "step": 49985
-    },
-    {
-      "epoch": 26.06831812255541,
-      "grad_norm": 1.598844289779663,
-      "learning_rate": 5.028341708542714e-05,
-      "loss": 5.193,
-      "step": 49986
-    },
-    {
-      "epoch": 26.06883963494133,
-      "grad_norm": 1.5737974643707275,
-      "learning_rate": 5.0282412060301506e-05,
-      "loss": 4.6594,
-      "step": 49987
-    },
-    {
-      "epoch": 26.06936114732725,
-      "grad_norm": 1.5599888563156128,
-      "learning_rate": 5.0281407035175884e-05,
-      "loss": 5.1594,
-      "step": 49988
-    },
-    {
-      "epoch": 26.06988265971317,
-      "grad_norm": 1.6480402946472168,
-      "learning_rate": 5.0280402010050255e-05,
-      "loss": 5.381,
-      "step": 49989
-    },
-    {
-      "epoch": 26.07040417209909,
-      "grad_norm": 1.5630629062652588,
-      "learning_rate": 5.027939698492462e-05,
-      "loss": 5.5411,
-      "step": 49990
-    },
-    {
-      "epoch": 26.070925684485008,
-      "grad_norm": 1.4982770681381226,
-      "learning_rate": 5.0278391959799e-05,
-      "loss": 5.3476,
-      "step": 49991
-    },
-    {
-      "epoch": 26.071447196870924,
-      "grad_norm": 1.6861292123794556,
-      "learning_rate": 5.027738693467336e-05,
-      "loss": 5.1003,
-      "step": 49992
-    },
-    {
-      "epoch": 26.071968709256844,
-      "grad_norm": 1.559387445449829,
-      "learning_rate": 5.027638190954774e-05,
-      "loss": 5.1098,
-      "step": 49993
-    },
-    {
-      "epoch": 26.072490221642763,
-      "grad_norm": 1.601373314857483,
-      "learning_rate": 5.027537688442211e-05,
-      "loss": 5.2072,
-      "step": 49994
-    },
-    {
-      "epoch": 26.073011734028682,
-      "grad_norm": 1.5824064016342163,
-      "learning_rate": 5.027437185929649e-05,
-      "loss": 5.2241,
-      "step": 49995
-    },
-    {
-      "epoch": 26.073533246414602,
-      "grad_norm": 1.641309142112732,
-      "learning_rate": 5.027336683417085e-05,
-      "loss": 5.6711,
-      "step": 49996
-    },
-    {
-      "epoch": 26.07405475880052,
-      "grad_norm": 1.5635870695114136,
-      "learning_rate": 5.027236180904523e-05,
-      "loss": 5.247,
-      "step": 49997
-    },
-    {
-      "epoch": 26.07457627118644,
-      "grad_norm": 1.541756510734558,
-      "learning_rate": 5.02713567839196e-05,
-      "loss": 5.4775,
-      "step": 49998
-    },
-    {
-      "epoch": 26.07509778357236,
-      "grad_norm": 1.5502567291259766,
-      "learning_rate": 5.027035175879398e-05,
-      "loss": 5.4107,
-      "step": 49999
-    },
-    {
-      "epoch": 26.07561929595828,
-      "grad_norm": 1.6434838771820068,
-      "learning_rate": 5.0269346733668344e-05,
-      "loss": 4.7404,
-      "step": 50000
-    },
-    {
-      "epoch": 26.07561929595828,
-      "eval_loss": 5.325130939483643,
-      "eval_runtime": 42.6665,
-      "eval_samples_per_second": 28.734,
-      "eval_steps_per_second": 3.609,
-      "step": 50000
-    }
-  ],
-  "logging_steps": 1,
-  "max_steps": 100000,
-  "num_input_tokens_seen": 0,
-  "num_train_epochs": 53,
-  "save_steps": 5000,
-  "total_flos": 7.454840432225157e+17,
-  "train_batch_size": 4,
-  "trial_name": null,
-  "trial_params": null
-}